{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 5864,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00017053206002728513,
      "grad_norm": 7.650136947631836,
      "learning_rate": 0.0,
      "loss": 0.8342,
      "step": 1
    },
    {
      "epoch": 0.00034106412005457026,
      "grad_norm": 1.9354814291000366,
      "learning_rate": 1.1363636363636366e-08,
      "loss": 0.7446,
      "step": 2
    },
    {
      "epoch": 0.0005115961800818554,
      "grad_norm": 2.511084794998169,
      "learning_rate": 2.2727272727272732e-08,
      "loss": 0.7271,
      "step": 3
    },
    {
      "epoch": 0.0006821282401091405,
      "grad_norm": 1.2728736400604248,
      "learning_rate": 3.409090909090909e-08,
      "loss": 0.717,
      "step": 4
    },
    {
      "epoch": 0.0008526603001364257,
      "grad_norm": 1.1760114431381226,
      "learning_rate": 4.5454545454545464e-08,
      "loss": 0.7129,
      "step": 5
    },
    {
      "epoch": 0.0010231923601637107,
      "grad_norm": 1.9061273336410522,
      "learning_rate": 5.681818181818183e-08,
      "loss": 0.7003,
      "step": 6
    },
    {
      "epoch": 0.0011937244201909959,
      "grad_norm": 0.9682210087776184,
      "learning_rate": 6.818181818181819e-08,
      "loss": 0.6932,
      "step": 7
    },
    {
      "epoch": 0.001364256480218281,
      "grad_norm": 0.7187206149101257,
      "learning_rate": 7.954545454545456e-08,
      "loss": 0.6776,
      "step": 8
    },
    {
      "epoch": 0.0015347885402455662,
      "grad_norm": 0.595079243183136,
      "learning_rate": 9.090909090909093e-08,
      "loss": 0.6984,
      "step": 9
    },
    {
      "epoch": 0.0017053206002728514,
      "grad_norm": 0.6756548285484314,
      "learning_rate": 1.0227272727272729e-07,
      "loss": 0.6788,
      "step": 10
    },
    {
      "epoch": 0.0018758526603001365,
      "grad_norm": 0.5831514000892639,
      "learning_rate": 1.1363636363636366e-07,
      "loss": 0.6979,
      "step": 11
    },
    {
      "epoch": 0.0020463847203274215,
      "grad_norm": 0.4670233428478241,
      "learning_rate": 1.2500000000000002e-07,
      "loss": 0.6816,
      "step": 12
    },
    {
      "epoch": 0.002216916780354707,
      "grad_norm": 0.5446857213973999,
      "learning_rate": 1.3636363636363637e-07,
      "loss": 0.6633,
      "step": 13
    },
    {
      "epoch": 0.0023874488403819918,
      "grad_norm": 0.513152539730072,
      "learning_rate": 1.4772727272727275e-07,
      "loss": 0.6866,
      "step": 14
    },
    {
      "epoch": 0.002557980900409277,
      "grad_norm": 0.4256693720817566,
      "learning_rate": 1.5909090909090913e-07,
      "loss": 0.6829,
      "step": 15
    },
    {
      "epoch": 0.002728512960436562,
      "grad_norm": 0.5119574069976807,
      "learning_rate": 1.7045454545454548e-07,
      "loss": 0.6785,
      "step": 16
    },
    {
      "epoch": 0.002899045020463847,
      "grad_norm": 0.4757070541381836,
      "learning_rate": 1.8181818181818186e-07,
      "loss": 0.6841,
      "step": 17
    },
    {
      "epoch": 0.0030695770804911324,
      "grad_norm": 0.44998493790626526,
      "learning_rate": 1.931818181818182e-07,
      "loss": 0.6794,
      "step": 18
    },
    {
      "epoch": 0.0032401091405184173,
      "grad_norm": 0.49852901697158813,
      "learning_rate": 2.0454545454545458e-07,
      "loss": 0.6769,
      "step": 19
    },
    {
      "epoch": 0.0034106412005457027,
      "grad_norm": 0.4366150498390198,
      "learning_rate": 2.1590909090909094e-07,
      "loss": 0.6676,
      "step": 20
    },
    {
      "epoch": 0.0035811732605729877,
      "grad_norm": 0.4075460731983185,
      "learning_rate": 2.272727272727273e-07,
      "loss": 0.6667,
      "step": 21
    },
    {
      "epoch": 0.003751705320600273,
      "grad_norm": 0.43057799339294434,
      "learning_rate": 2.386363636363637e-07,
      "loss": 0.6619,
      "step": 22
    },
    {
      "epoch": 0.003922237380627558,
      "grad_norm": 0.45519155263900757,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.6564,
      "step": 23
    },
    {
      "epoch": 0.004092769440654843,
      "grad_norm": 0.3911649286746979,
      "learning_rate": 2.613636363636364e-07,
      "loss": 0.6528,
      "step": 24
    },
    {
      "epoch": 0.004263301500682128,
      "grad_norm": 0.44951483607292175,
      "learning_rate": 2.7272727272727274e-07,
      "loss": 0.6502,
      "step": 25
    },
    {
      "epoch": 0.004433833560709414,
      "grad_norm": 0.3782355487346649,
      "learning_rate": 2.8409090909090915e-07,
      "loss": 0.6738,
      "step": 26
    },
    {
      "epoch": 0.004604365620736698,
      "grad_norm": 0.4313599765300751,
      "learning_rate": 2.954545454545455e-07,
      "loss": 0.6676,
      "step": 27
    },
    {
      "epoch": 0.0047748976807639835,
      "grad_norm": 0.4656371474266052,
      "learning_rate": 3.0681818181818185e-07,
      "loss": 0.6689,
      "step": 28
    },
    {
      "epoch": 0.004945429740791269,
      "grad_norm": 0.4009706676006317,
      "learning_rate": 3.1818181818181825e-07,
      "loss": 0.6708,
      "step": 29
    },
    {
      "epoch": 0.005115961800818554,
      "grad_norm": 0.45048245787620544,
      "learning_rate": 3.295454545454546e-07,
      "loss": 0.6633,
      "step": 30
    },
    {
      "epoch": 0.005286493860845839,
      "grad_norm": 0.4189516305923462,
      "learning_rate": 3.4090909090909096e-07,
      "loss": 0.6435,
      "step": 31
    },
    {
      "epoch": 0.005457025920873124,
      "grad_norm": 0.4668561518192291,
      "learning_rate": 3.5227272727272736e-07,
      "loss": 0.6507,
      "step": 32
    },
    {
      "epoch": 0.0056275579809004096,
      "grad_norm": 0.4583468735218048,
      "learning_rate": 3.636363636363637e-07,
      "loss": 0.642,
      "step": 33
    },
    {
      "epoch": 0.005798090040927694,
      "grad_norm": 0.540885329246521,
      "learning_rate": 3.7500000000000006e-07,
      "loss": 0.6798,
      "step": 34
    },
    {
      "epoch": 0.0059686221009549794,
      "grad_norm": 0.49574148654937744,
      "learning_rate": 3.863636363636364e-07,
      "loss": 0.6524,
      "step": 35
    },
    {
      "epoch": 0.006139154160982265,
      "grad_norm": 0.4109654426574707,
      "learning_rate": 3.9772727272727276e-07,
      "loss": 0.6608,
      "step": 36
    },
    {
      "epoch": 0.00630968622100955,
      "grad_norm": 0.3942180573940277,
      "learning_rate": 4.0909090909090917e-07,
      "loss": 0.6657,
      "step": 37
    },
    {
      "epoch": 0.006480218281036835,
      "grad_norm": 0.401726096868515,
      "learning_rate": 4.204545454545455e-07,
      "loss": 0.6473,
      "step": 38
    },
    {
      "epoch": 0.00665075034106412,
      "grad_norm": 0.448159396648407,
      "learning_rate": 4.3181818181818187e-07,
      "loss": 0.6499,
      "step": 39
    },
    {
      "epoch": 0.0068212824010914054,
      "grad_norm": 0.41607365012168884,
      "learning_rate": 4.431818181818183e-07,
      "loss": 0.6484,
      "step": 40
    },
    {
      "epoch": 0.00699181446111869,
      "grad_norm": 0.40875402092933655,
      "learning_rate": 4.545454545454546e-07,
      "loss": 0.6613,
      "step": 41
    },
    {
      "epoch": 0.007162346521145975,
      "grad_norm": 0.46847590804100037,
      "learning_rate": 4.65909090909091e-07,
      "loss": 0.6506,
      "step": 42
    },
    {
      "epoch": 0.007332878581173261,
      "grad_norm": 0.473928838968277,
      "learning_rate": 4.772727272727274e-07,
      "loss": 0.6264,
      "step": 43
    },
    {
      "epoch": 0.007503410641200546,
      "grad_norm": 0.6333487033843994,
      "learning_rate": 4.886363636363637e-07,
      "loss": 0.6591,
      "step": 44
    },
    {
      "epoch": 0.007673942701227831,
      "grad_norm": 0.9283931255340576,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.6611,
      "step": 45
    },
    {
      "epoch": 0.007844474761255117,
      "grad_norm": 0.8683100938796997,
      "learning_rate": 5.113636363636364e-07,
      "loss": 0.6401,
      "step": 46
    },
    {
      "epoch": 0.008015006821282401,
      "grad_norm": 0.4764223098754883,
      "learning_rate": 5.227272727272728e-07,
      "loss": 0.6398,
      "step": 47
    },
    {
      "epoch": 0.008185538881309686,
      "grad_norm": 0.692018985748291,
      "learning_rate": 5.340909090909091e-07,
      "loss": 0.6331,
      "step": 48
    },
    {
      "epoch": 0.008356070941336972,
      "grad_norm": 0.760298490524292,
      "learning_rate": 5.454545454545455e-07,
      "loss": 0.6503,
      "step": 49
    },
    {
      "epoch": 0.008526603001364257,
      "grad_norm": 0.48740214109420776,
      "learning_rate": 5.568181818181818e-07,
      "loss": 0.6273,
      "step": 50
    },
    {
      "epoch": 0.008697135061391541,
      "grad_norm": 0.623894989490509,
      "learning_rate": 5.681818181818183e-07,
      "loss": 0.6529,
      "step": 51
    },
    {
      "epoch": 0.008867667121418827,
      "grad_norm": 0.45810043811798096,
      "learning_rate": 5.795454545454546e-07,
      "loss": 0.6339,
      "step": 52
    },
    {
      "epoch": 0.009038199181446112,
      "grad_norm": 0.5587273240089417,
      "learning_rate": 5.90909090909091e-07,
      "loss": 0.6411,
      "step": 53
    },
    {
      "epoch": 0.009208731241473396,
      "grad_norm": 0.6215199828147888,
      "learning_rate": 6.022727272727273e-07,
      "loss": 0.6568,
      "step": 54
    },
    {
      "epoch": 0.009379263301500683,
      "grad_norm": 0.5236825346946716,
      "learning_rate": 6.136363636363637e-07,
      "loss": 0.6385,
      "step": 55
    },
    {
      "epoch": 0.009549795361527967,
      "grad_norm": 0.5010786652565002,
      "learning_rate": 6.250000000000002e-07,
      "loss": 0.6522,
      "step": 56
    },
    {
      "epoch": 0.009720327421555252,
      "grad_norm": 0.5132748484611511,
      "learning_rate": 6.363636363636365e-07,
      "loss": 0.619,
      "step": 57
    },
    {
      "epoch": 0.009890859481582538,
      "grad_norm": 0.5371212959289551,
      "learning_rate": 6.477272727272729e-07,
      "loss": 0.6224,
      "step": 58
    },
    {
      "epoch": 0.010061391541609822,
      "grad_norm": 0.5243486166000366,
      "learning_rate": 6.590909090909092e-07,
      "loss": 0.6524,
      "step": 59
    },
    {
      "epoch": 0.010231923601637109,
      "grad_norm": 0.5441023707389832,
      "learning_rate": 6.704545454545456e-07,
      "loss": 0.6522,
      "step": 60
    },
    {
      "epoch": 0.010402455661664393,
      "grad_norm": 0.5234872698783875,
      "learning_rate": 6.818181818181819e-07,
      "loss": 0.6329,
      "step": 61
    },
    {
      "epoch": 0.010572987721691678,
      "grad_norm": 0.5267979502677917,
      "learning_rate": 6.931818181818184e-07,
      "loss": 0.6364,
      "step": 62
    },
    {
      "epoch": 0.010743519781718964,
      "grad_norm": 0.5362216234207153,
      "learning_rate": 7.045454545454547e-07,
      "loss": 0.6296,
      "step": 63
    },
    {
      "epoch": 0.010914051841746248,
      "grad_norm": 0.5315608978271484,
      "learning_rate": 7.159090909090911e-07,
      "loss": 0.6616,
      "step": 64
    },
    {
      "epoch": 0.011084583901773533,
      "grad_norm": 0.4777587652206421,
      "learning_rate": 7.272727272727274e-07,
      "loss": 0.6318,
      "step": 65
    },
    {
      "epoch": 0.011255115961800819,
      "grad_norm": 0.5749709606170654,
      "learning_rate": 7.386363636363638e-07,
      "loss": 0.6301,
      "step": 66
    },
    {
      "epoch": 0.011425648021828104,
      "grad_norm": 0.4934489130973816,
      "learning_rate": 7.500000000000001e-07,
      "loss": 0.629,
      "step": 67
    },
    {
      "epoch": 0.011596180081855388,
      "grad_norm": 0.4862441420555115,
      "learning_rate": 7.613636363636365e-07,
      "loss": 0.6349,
      "step": 68
    },
    {
      "epoch": 0.011766712141882674,
      "grad_norm": 0.5512414574623108,
      "learning_rate": 7.727272727272728e-07,
      "loss": 0.6253,
      "step": 69
    },
    {
      "epoch": 0.011937244201909959,
      "grad_norm": 0.4864642322063446,
      "learning_rate": 7.840909090909092e-07,
      "loss": 0.6349,
      "step": 70
    },
    {
      "epoch": 0.012107776261937243,
      "grad_norm": 0.5771092176437378,
      "learning_rate": 7.954545454545455e-07,
      "loss": 0.6605,
      "step": 71
    },
    {
      "epoch": 0.01227830832196453,
      "grad_norm": 0.6331750750541687,
      "learning_rate": 8.068181818181819e-07,
      "loss": 0.6405,
      "step": 72
    },
    {
      "epoch": 0.012448840381991814,
      "grad_norm": 0.5899202823638916,
      "learning_rate": 8.181818181818183e-07,
      "loss": 0.6187,
      "step": 73
    },
    {
      "epoch": 0.0126193724420191,
      "grad_norm": 0.7244130969047546,
      "learning_rate": 8.295454545454547e-07,
      "loss": 0.6297,
      "step": 74
    },
    {
      "epoch": 0.012789904502046385,
      "grad_norm": 0.5164639949798584,
      "learning_rate": 8.40909090909091e-07,
      "loss": 0.6563,
      "step": 75
    },
    {
      "epoch": 0.01296043656207367,
      "grad_norm": 0.6244091391563416,
      "learning_rate": 8.522727272727274e-07,
      "loss": 0.6456,
      "step": 76
    },
    {
      "epoch": 0.013130968622100956,
      "grad_norm": 0.8265087008476257,
      "learning_rate": 8.636363636363637e-07,
      "loss": 0.6299,
      "step": 77
    },
    {
      "epoch": 0.01330150068212824,
      "grad_norm": 0.8179631233215332,
      "learning_rate": 8.750000000000001e-07,
      "loss": 0.6127,
      "step": 78
    },
    {
      "epoch": 0.013472032742155525,
      "grad_norm": 0.6222761869430542,
      "learning_rate": 8.863636363636365e-07,
      "loss": 0.628,
      "step": 79
    },
    {
      "epoch": 0.013642564802182811,
      "grad_norm": 0.5206494331359863,
      "learning_rate": 8.977272727272729e-07,
      "loss": 0.6299,
      "step": 80
    },
    {
      "epoch": 0.013813096862210095,
      "grad_norm": 0.5181920528411865,
      "learning_rate": 9.090909090909093e-07,
      "loss": 0.6319,
      "step": 81
    },
    {
      "epoch": 0.01398362892223738,
      "grad_norm": 0.6216228604316711,
      "learning_rate": 9.204545454545456e-07,
      "loss": 0.6336,
      "step": 82
    },
    {
      "epoch": 0.014154160982264666,
      "grad_norm": 0.6102015972137451,
      "learning_rate": 9.31818181818182e-07,
      "loss": 0.6238,
      "step": 83
    },
    {
      "epoch": 0.01432469304229195,
      "grad_norm": 0.5613488554954529,
      "learning_rate": 9.431818181818184e-07,
      "loss": 0.6319,
      "step": 84
    },
    {
      "epoch": 0.014495225102319235,
      "grad_norm": 0.5151708722114563,
      "learning_rate": 9.545454545454548e-07,
      "loss": 0.6096,
      "step": 85
    },
    {
      "epoch": 0.014665757162346521,
      "grad_norm": 0.5965888500213623,
      "learning_rate": 9.659090909090911e-07,
      "loss": 0.6321,
      "step": 86
    },
    {
      "epoch": 0.014836289222373806,
      "grad_norm": 0.6112313270568848,
      "learning_rate": 9.772727272727275e-07,
      "loss": 0.646,
      "step": 87
    },
    {
      "epoch": 0.015006821282401092,
      "grad_norm": 0.6595168709754944,
      "learning_rate": 9.886363636363638e-07,
      "loss": 0.6235,
      "step": 88
    },
    {
      "epoch": 0.015177353342428377,
      "grad_norm": 0.5075546503067017,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.6289,
      "step": 89
    },
    {
      "epoch": 0.015347885402455661,
      "grad_norm": 0.6066994667053223,
      "learning_rate": 1.0113636363636367e-06,
      "loss": 0.6259,
      "step": 90
    },
    {
      "epoch": 0.015518417462482947,
      "grad_norm": 0.7509892582893372,
      "learning_rate": 1.0227272727272729e-06,
      "loss": 0.6194,
      "step": 91
    },
    {
      "epoch": 0.015688949522510234,
      "grad_norm": 0.7574506998062134,
      "learning_rate": 1.0340909090909094e-06,
      "loss": 0.6168,
      "step": 92
    },
    {
      "epoch": 0.015859481582537516,
      "grad_norm": 0.449613094329834,
      "learning_rate": 1.0454545454545456e-06,
      "loss": 0.6037,
      "step": 93
    },
    {
      "epoch": 0.016030013642564803,
      "grad_norm": 0.6445989608764648,
      "learning_rate": 1.0568181818181821e-06,
      "loss": 0.6481,
      "step": 94
    },
    {
      "epoch": 0.01620054570259209,
      "grad_norm": 0.5870500802993774,
      "learning_rate": 1.0681818181818183e-06,
      "loss": 0.6253,
      "step": 95
    },
    {
      "epoch": 0.01637107776261937,
      "grad_norm": 0.4731763005256653,
      "learning_rate": 1.0795454545454548e-06,
      "loss": 0.6132,
      "step": 96
    },
    {
      "epoch": 0.016541609822646658,
      "grad_norm": 0.5874089598655701,
      "learning_rate": 1.090909090909091e-06,
      "loss": 0.6051,
      "step": 97
    },
    {
      "epoch": 0.016712141882673944,
      "grad_norm": 0.5979234576225281,
      "learning_rate": 1.1022727272727275e-06,
      "loss": 0.6119,
      "step": 98
    },
    {
      "epoch": 0.016882673942701227,
      "grad_norm": 0.5582343935966492,
      "learning_rate": 1.1136363636363637e-06,
      "loss": 0.6189,
      "step": 99
    },
    {
      "epoch": 0.017053206002728513,
      "grad_norm": 0.44007357954978943,
      "learning_rate": 1.1250000000000002e-06,
      "loss": 0.6143,
      "step": 100
    },
    {
      "epoch": 0.0172237380627558,
      "grad_norm": 0.48567453026771545,
      "learning_rate": 1.1363636363636366e-06,
      "loss": 0.6127,
      "step": 101
    },
    {
      "epoch": 0.017394270122783082,
      "grad_norm": 0.5077864527702332,
      "learning_rate": 1.147727272727273e-06,
      "loss": 0.611,
      "step": 102
    },
    {
      "epoch": 0.01756480218281037,
      "grad_norm": 0.467916876077652,
      "learning_rate": 1.1590909090909093e-06,
      "loss": 0.6212,
      "step": 103
    },
    {
      "epoch": 0.017735334242837655,
      "grad_norm": 0.46231311559677124,
      "learning_rate": 1.1704545454545456e-06,
      "loss": 0.6244,
      "step": 104
    },
    {
      "epoch": 0.017905866302864937,
      "grad_norm": 0.48694437742233276,
      "learning_rate": 1.181818181818182e-06,
      "loss": 0.6145,
      "step": 105
    },
    {
      "epoch": 0.018076398362892224,
      "grad_norm": 0.6872959733009338,
      "learning_rate": 1.1931818181818183e-06,
      "loss": 0.627,
      "step": 106
    },
    {
      "epoch": 0.01824693042291951,
      "grad_norm": 0.43043017387390137,
      "learning_rate": 1.2045454545454547e-06,
      "loss": 0.6096,
      "step": 107
    },
    {
      "epoch": 0.018417462482946793,
      "grad_norm": 0.43259918689727783,
      "learning_rate": 1.215909090909091e-06,
      "loss": 0.6111,
      "step": 108
    },
    {
      "epoch": 0.01858799454297408,
      "grad_norm": 0.4670185446739197,
      "learning_rate": 1.2272727272727274e-06,
      "loss": 0.6122,
      "step": 109
    },
    {
      "epoch": 0.018758526603001365,
      "grad_norm": 0.4173937737941742,
      "learning_rate": 1.2386363636363638e-06,
      "loss": 0.6081,
      "step": 110
    },
    {
      "epoch": 0.018929058663028648,
      "grad_norm": 0.4710495173931122,
      "learning_rate": 1.2500000000000003e-06,
      "loss": 0.6028,
      "step": 111
    },
    {
      "epoch": 0.019099590723055934,
      "grad_norm": 0.5226242542266846,
      "learning_rate": 1.2613636363636367e-06,
      "loss": 0.6045,
      "step": 112
    },
    {
      "epoch": 0.01927012278308322,
      "grad_norm": 0.6086929440498352,
      "learning_rate": 1.272727272727273e-06,
      "loss": 0.6127,
      "step": 113
    },
    {
      "epoch": 0.019440654843110503,
      "grad_norm": 0.5822572112083435,
      "learning_rate": 1.2840909090909094e-06,
      "loss": 0.6039,
      "step": 114
    },
    {
      "epoch": 0.01961118690313779,
      "grad_norm": 0.49154001474380493,
      "learning_rate": 1.2954545454545457e-06,
      "loss": 0.621,
      "step": 115
    },
    {
      "epoch": 0.019781718963165076,
      "grad_norm": 0.4529736638069153,
      "learning_rate": 1.306818181818182e-06,
      "loss": 0.6174,
      "step": 116
    },
    {
      "epoch": 0.01995225102319236,
      "grad_norm": 0.5126652717590332,
      "learning_rate": 1.3181818181818184e-06,
      "loss": 0.5937,
      "step": 117
    },
    {
      "epoch": 0.020122783083219645,
      "grad_norm": 0.6636311411857605,
      "learning_rate": 1.3295454545454548e-06,
      "loss": 0.6058,
      "step": 118
    },
    {
      "epoch": 0.02029331514324693,
      "grad_norm": 0.7904946804046631,
      "learning_rate": 1.3409090909090911e-06,
      "loss": 0.6257,
      "step": 119
    },
    {
      "epoch": 0.020463847203274217,
      "grad_norm": 0.708732008934021,
      "learning_rate": 1.3522727272727275e-06,
      "loss": 0.6168,
      "step": 120
    },
    {
      "epoch": 0.0206343792633015,
      "grad_norm": 0.5224078297615051,
      "learning_rate": 1.3636363636363638e-06,
      "loss": 0.6344,
      "step": 121
    },
    {
      "epoch": 0.020804911323328786,
      "grad_norm": 0.4924236536026001,
      "learning_rate": 1.3750000000000002e-06,
      "loss": 0.5914,
      "step": 122
    },
    {
      "epoch": 0.020975443383356072,
      "grad_norm": 0.6102842688560486,
      "learning_rate": 1.3863636363636367e-06,
      "loss": 0.6104,
      "step": 123
    },
    {
      "epoch": 0.021145975443383355,
      "grad_norm": 0.45268720388412476,
      "learning_rate": 1.3977272727272729e-06,
      "loss": 0.6122,
      "step": 124
    },
    {
      "epoch": 0.02131650750341064,
      "grad_norm": 0.5215243697166443,
      "learning_rate": 1.4090909090909094e-06,
      "loss": 0.6402,
      "step": 125
    },
    {
      "epoch": 0.021487039563437928,
      "grad_norm": 0.4833904206752777,
      "learning_rate": 1.4204545454545456e-06,
      "loss": 0.6016,
      "step": 126
    },
    {
      "epoch": 0.02165757162346521,
      "grad_norm": 0.5406088829040527,
      "learning_rate": 1.4318181818181821e-06,
      "loss": 0.6133,
      "step": 127
    },
    {
      "epoch": 0.021828103683492497,
      "grad_norm": 0.5165327787399292,
      "learning_rate": 1.4431818181818183e-06,
      "loss": 0.6082,
      "step": 128
    },
    {
      "epoch": 0.021998635743519783,
      "grad_norm": 0.40911534428596497,
      "learning_rate": 1.4545454545454548e-06,
      "loss": 0.6027,
      "step": 129
    },
    {
      "epoch": 0.022169167803547066,
      "grad_norm": 0.47644490003585815,
      "learning_rate": 1.465909090909091e-06,
      "loss": 0.6196,
      "step": 130
    },
    {
      "epoch": 0.022339699863574352,
      "grad_norm": 0.4921034276485443,
      "learning_rate": 1.4772727272727275e-06,
      "loss": 0.6125,
      "step": 131
    },
    {
      "epoch": 0.022510231923601638,
      "grad_norm": 0.6039801239967346,
      "learning_rate": 1.4886363636363637e-06,
      "loss": 0.6069,
      "step": 132
    },
    {
      "epoch": 0.02268076398362892,
      "grad_norm": 0.4514678418636322,
      "learning_rate": 1.5000000000000002e-06,
      "loss": 0.6017,
      "step": 133
    },
    {
      "epoch": 0.022851296043656207,
      "grad_norm": 0.4566296637058258,
      "learning_rate": 1.5113636363636368e-06,
      "loss": 0.5971,
      "step": 134
    },
    {
      "epoch": 0.023021828103683493,
      "grad_norm": 0.5192421674728394,
      "learning_rate": 1.522727272727273e-06,
      "loss": 0.6104,
      "step": 135
    },
    {
      "epoch": 0.023192360163710776,
      "grad_norm": 0.5545705556869507,
      "learning_rate": 1.5340909090909095e-06,
      "loss": 0.6007,
      "step": 136
    },
    {
      "epoch": 0.023362892223738062,
      "grad_norm": 0.43825459480285645,
      "learning_rate": 1.5454545454545457e-06,
      "loss": 0.607,
      "step": 137
    },
    {
      "epoch": 0.02353342428376535,
      "grad_norm": 0.4573563039302826,
      "learning_rate": 1.5568181818181822e-06,
      "loss": 0.6144,
      "step": 138
    },
    {
      "epoch": 0.02370395634379263,
      "grad_norm": 0.5552775263786316,
      "learning_rate": 1.5681818181818184e-06,
      "loss": 0.6113,
      "step": 139
    },
    {
      "epoch": 0.023874488403819918,
      "grad_norm": 0.42399027943611145,
      "learning_rate": 1.579545454545455e-06,
      "loss": 0.6072,
      "step": 140
    },
    {
      "epoch": 0.024045020463847204,
      "grad_norm": 0.42997267842292786,
      "learning_rate": 1.590909090909091e-06,
      "loss": 0.6223,
      "step": 141
    },
    {
      "epoch": 0.024215552523874487,
      "grad_norm": 0.4232839047908783,
      "learning_rate": 1.6022727272727276e-06,
      "loss": 0.6205,
      "step": 142
    },
    {
      "epoch": 0.024386084583901773,
      "grad_norm": 0.43472787737846375,
      "learning_rate": 1.6136363636363638e-06,
      "loss": 0.608,
      "step": 143
    },
    {
      "epoch": 0.02455661664392906,
      "grad_norm": 0.4148360788822174,
      "learning_rate": 1.6250000000000003e-06,
      "loss": 0.6064,
      "step": 144
    },
    {
      "epoch": 0.024727148703956345,
      "grad_norm": 0.4622041583061218,
      "learning_rate": 1.6363636363636367e-06,
      "loss": 0.6148,
      "step": 145
    },
    {
      "epoch": 0.024897680763983628,
      "grad_norm": 0.4319433271884918,
      "learning_rate": 1.647727272727273e-06,
      "loss": 0.599,
      "step": 146
    },
    {
      "epoch": 0.025068212824010914,
      "grad_norm": 0.5556517243385315,
      "learning_rate": 1.6590909090909094e-06,
      "loss": 0.611,
      "step": 147
    },
    {
      "epoch": 0.0252387448840382,
      "grad_norm": 0.675335168838501,
      "learning_rate": 1.6704545454545457e-06,
      "loss": 0.6072,
      "step": 148
    },
    {
      "epoch": 0.025409276944065484,
      "grad_norm": 0.7137088775634766,
      "learning_rate": 1.681818181818182e-06,
      "loss": 0.5873,
      "step": 149
    },
    {
      "epoch": 0.02557980900409277,
      "grad_norm": 0.6885473728179932,
      "learning_rate": 1.6931818181818184e-06,
      "loss": 0.6084,
      "step": 150
    },
    {
      "epoch": 0.025750341064120056,
      "grad_norm": 0.430575966835022,
      "learning_rate": 1.7045454545454548e-06,
      "loss": 0.6157,
      "step": 151
    },
    {
      "epoch": 0.02592087312414734,
      "grad_norm": 0.539156973361969,
      "learning_rate": 1.7159090909090911e-06,
      "loss": 0.6024,
      "step": 152
    },
    {
      "epoch": 0.026091405184174625,
      "grad_norm": 0.7821847796440125,
      "learning_rate": 1.7272727272727275e-06,
      "loss": 0.6004,
      "step": 153
    },
    {
      "epoch": 0.02626193724420191,
      "grad_norm": 0.6745071411132812,
      "learning_rate": 1.7386363636363638e-06,
      "loss": 0.6099,
      "step": 154
    },
    {
      "epoch": 0.026432469304229194,
      "grad_norm": 0.44672396779060364,
      "learning_rate": 1.7500000000000002e-06,
      "loss": 0.615,
      "step": 155
    },
    {
      "epoch": 0.02660300136425648,
      "grad_norm": 0.5009127259254456,
      "learning_rate": 1.7613636363636367e-06,
      "loss": 0.6085,
      "step": 156
    },
    {
      "epoch": 0.026773533424283767,
      "grad_norm": 0.6595852375030518,
      "learning_rate": 1.772727272727273e-06,
      "loss": 0.5926,
      "step": 157
    },
    {
      "epoch": 0.02694406548431105,
      "grad_norm": 0.6197225451469421,
      "learning_rate": 1.7840909090909095e-06,
      "loss": 0.6137,
      "step": 158
    },
    {
      "epoch": 0.027114597544338336,
      "grad_norm": 0.39933961629867554,
      "learning_rate": 1.7954545454545458e-06,
      "loss": 0.6051,
      "step": 159
    },
    {
      "epoch": 0.027285129604365622,
      "grad_norm": 0.5831077098846436,
      "learning_rate": 1.8068181818181822e-06,
      "loss": 0.6239,
      "step": 160
    },
    {
      "epoch": 0.027455661664392905,
      "grad_norm": 0.4417240023612976,
      "learning_rate": 1.8181818181818185e-06,
      "loss": 0.5922,
      "step": 161
    },
    {
      "epoch": 0.02762619372442019,
      "grad_norm": 0.4070188105106354,
      "learning_rate": 1.8295454545454549e-06,
      "loss": 0.602,
      "step": 162
    },
    {
      "epoch": 0.027796725784447477,
      "grad_norm": 0.5691607594490051,
      "learning_rate": 1.8409090909090912e-06,
      "loss": 0.6004,
      "step": 163
    },
    {
      "epoch": 0.02796725784447476,
      "grad_norm": 0.6013835668563843,
      "learning_rate": 1.8522727272727276e-06,
      "loss": 0.598,
      "step": 164
    },
    {
      "epoch": 0.028137789904502046,
      "grad_norm": 0.6091790199279785,
      "learning_rate": 1.863636363636364e-06,
      "loss": 0.594,
      "step": 165
    },
    {
      "epoch": 0.028308321964529332,
      "grad_norm": 0.3511148691177368,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.5843,
      "step": 166
    },
    {
      "epoch": 0.028478854024556615,
      "grad_norm": 0.4900568127632141,
      "learning_rate": 1.8863636363636368e-06,
      "loss": 0.591,
      "step": 167
    },
    {
      "epoch": 0.0286493860845839,
      "grad_norm": 0.49398183822631836,
      "learning_rate": 1.897727272727273e-06,
      "loss": 0.5938,
      "step": 168
    },
    {
      "epoch": 0.028819918144611188,
      "grad_norm": 0.46477246284484863,
      "learning_rate": 1.9090909090909095e-06,
      "loss": 0.5878,
      "step": 169
    },
    {
      "epoch": 0.02899045020463847,
      "grad_norm": 0.4071483910083771,
      "learning_rate": 1.9204545454545457e-06,
      "loss": 0.6173,
      "step": 170
    },
    {
      "epoch": 0.029160982264665757,
      "grad_norm": 0.44497352838516235,
      "learning_rate": 1.9318181818181822e-06,
      "loss": 0.6006,
      "step": 171
    },
    {
      "epoch": 0.029331514324693043,
      "grad_norm": 0.5880485773086548,
      "learning_rate": 1.9431818181818184e-06,
      "loss": 0.5992,
      "step": 172
    },
    {
      "epoch": 0.02950204638472033,
      "grad_norm": 0.4656350314617157,
      "learning_rate": 1.954545454545455e-06,
      "loss": 0.6085,
      "step": 173
    },
    {
      "epoch": 0.029672578444747612,
      "grad_norm": 0.4730178117752075,
      "learning_rate": 1.965909090909091e-06,
      "loss": 0.5865,
      "step": 174
    },
    {
      "epoch": 0.029843110504774898,
      "grad_norm": 0.5038389563560486,
      "learning_rate": 1.9772727272727276e-06,
      "loss": 0.5919,
      "step": 175
    },
    {
      "epoch": 0.030013642564802184,
      "grad_norm": 0.43242865800857544,
      "learning_rate": 1.9886363636363638e-06,
      "loss": 0.6107,
      "step": 176
    },
    {
      "epoch": 0.030184174624829467,
      "grad_norm": 0.6300122737884521,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.6194,
      "step": 177
    },
    {
      "epoch": 0.030354706684856753,
      "grad_norm": 0.8253108263015747,
      "learning_rate": 1.9999998474716537e-06,
      "loss": 0.6009,
      "step": 178
    },
    {
      "epoch": 0.03052523874488404,
      "grad_norm": 0.7405084371566772,
      "learning_rate": 1.9999993898866594e-06,
      "loss": 0.6022,
      "step": 179
    },
    {
      "epoch": 0.030695770804911322,
      "grad_norm": 0.478266179561615,
      "learning_rate": 1.999998627245158e-06,
      "loss": 0.5899,
      "step": 180
    },
    {
      "epoch": 0.03086630286493861,
      "grad_norm": 0.3928435742855072,
      "learning_rate": 1.9999975595473815e-06,
      "loss": 0.594,
      "step": 181
    },
    {
      "epoch": 0.031036834924965895,
      "grad_norm": 0.5914251804351807,
      "learning_rate": 1.9999961867936557e-06,
      "loss": 0.6086,
      "step": 182
    },
    {
      "epoch": 0.031207366984993178,
      "grad_norm": 0.4950089454650879,
      "learning_rate": 1.9999945089843996e-06,
      "loss": 0.5991,
      "step": 183
    },
    {
      "epoch": 0.03137789904502047,
      "grad_norm": 0.40972012281417847,
      "learning_rate": 1.999992526120125e-06,
      "loss": 0.6021,
      "step": 184
    },
    {
      "epoch": 0.03154843110504775,
      "grad_norm": 0.7472282648086548,
      "learning_rate": 1.9999902382014366e-06,
      "loss": 0.5961,
      "step": 185
    },
    {
      "epoch": 0.03171896316507503,
      "grad_norm": 0.55669766664505,
      "learning_rate": 1.999987645229032e-06,
      "loss": 0.5975,
      "step": 186
    },
    {
      "epoch": 0.03188949522510232,
      "grad_norm": 0.3858289122581482,
      "learning_rate": 1.999984747203703e-06,
      "loss": 0.5918,
      "step": 187
    },
    {
      "epoch": 0.032060027285129605,
      "grad_norm": 0.5737813115119934,
      "learning_rate": 1.9999815441263336e-06,
      "loss": 0.5905,
      "step": 188
    },
    {
      "epoch": 0.03223055934515689,
      "grad_norm": 0.5033409595489502,
      "learning_rate": 1.9999780359979e-06,
      "loss": 0.6084,
      "step": 189
    },
    {
      "epoch": 0.03240109140518418,
      "grad_norm": 0.40473321080207825,
      "learning_rate": 1.9999742228194737e-06,
      "loss": 0.5977,
      "step": 190
    },
    {
      "epoch": 0.03257162346521146,
      "grad_norm": 0.4098617434501648,
      "learning_rate": 1.9999701045922165e-06,
      "loss": 0.6047,
      "step": 191
    },
    {
      "epoch": 0.03274215552523874,
      "grad_norm": 0.47473642230033875,
      "learning_rate": 1.999965681317386e-06,
      "loss": 0.5994,
      "step": 192
    },
    {
      "epoch": 0.03291268758526603,
      "grad_norm": 0.4402756094932556,
      "learning_rate": 1.9999609529963303e-06,
      "loss": 0.6056,
      "step": 193
    },
    {
      "epoch": 0.033083219645293316,
      "grad_norm": 0.3741694688796997,
      "learning_rate": 1.999955919630493e-06,
      "loss": 0.5947,
      "step": 194
    },
    {
      "epoch": 0.0332537517053206,
      "grad_norm": 0.3938404619693756,
      "learning_rate": 1.9999505812214086e-06,
      "loss": 0.6096,
      "step": 195
    },
    {
      "epoch": 0.03342428376534789,
      "grad_norm": 0.3473811149597168,
      "learning_rate": 1.9999449377707063e-06,
      "loss": 0.5899,
      "step": 196
    },
    {
      "epoch": 0.03359481582537517,
      "grad_norm": 0.3678300678730011,
      "learning_rate": 1.999938989280108e-06,
      "loss": 0.6118,
      "step": 197
    },
    {
      "epoch": 0.033765347885402454,
      "grad_norm": 0.34986555576324463,
      "learning_rate": 1.999932735751427e-06,
      "loss": 0.6361,
      "step": 198
    },
    {
      "epoch": 0.033935879945429744,
      "grad_norm": 0.39926373958587646,
      "learning_rate": 1.999926177186572e-06,
      "loss": 0.5901,
      "step": 199
    },
    {
      "epoch": 0.034106412005457026,
      "grad_norm": 0.3418705463409424,
      "learning_rate": 1.9999193135875433e-06,
      "loss": 0.5948,
      "step": 200
    },
    {
      "epoch": 0.03427694406548431,
      "grad_norm": 0.42050284147262573,
      "learning_rate": 1.999912144956435e-06,
      "loss": 0.5944,
      "step": 201
    },
    {
      "epoch": 0.0344474761255116,
      "grad_norm": 0.4259498417377472,
      "learning_rate": 1.9999046712954335e-06,
      "loss": 0.5853,
      "step": 202
    },
    {
      "epoch": 0.03461800818553888,
      "grad_norm": 0.4480075538158417,
      "learning_rate": 1.9998968926068193e-06,
      "loss": 0.6023,
      "step": 203
    },
    {
      "epoch": 0.034788540245566164,
      "grad_norm": 0.5295131802558899,
      "learning_rate": 1.9998888088929648e-06,
      "loss": 0.597,
      "step": 204
    },
    {
      "epoch": 0.034959072305593454,
      "grad_norm": 0.5217076539993286,
      "learning_rate": 1.9998804201563363e-06,
      "loss": 0.5916,
      "step": 205
    },
    {
      "epoch": 0.03512960436562074,
      "grad_norm": 0.42749714851379395,
      "learning_rate": 1.9998717263994928e-06,
      "loss": 0.582,
      "step": 206
    },
    {
      "epoch": 0.03530013642564802,
      "grad_norm": 0.48688915371894836,
      "learning_rate": 1.9998627276250863e-06,
      "loss": 0.618,
      "step": 207
    },
    {
      "epoch": 0.03547066848567531,
      "grad_norm": 0.7113652229309082,
      "learning_rate": 1.999853423835862e-06,
      "loss": 0.595,
      "step": 208
    },
    {
      "epoch": 0.03564120054570259,
      "grad_norm": 0.8222281336784363,
      "learning_rate": 1.9998438150346577e-06,
      "loss": 0.5974,
      "step": 209
    },
    {
      "epoch": 0.035811732605729875,
      "grad_norm": 0.5764082074165344,
      "learning_rate": 1.9998339012244052e-06,
      "loss": 0.5945,
      "step": 210
    },
    {
      "epoch": 0.035982264665757165,
      "grad_norm": 0.40010660886764526,
      "learning_rate": 1.9998236824081293e-06,
      "loss": 0.5976,
      "step": 211
    },
    {
      "epoch": 0.03615279672578445,
      "grad_norm": 0.6307498812675476,
      "learning_rate": 1.999813158588946e-06,
      "loss": 0.5949,
      "step": 212
    },
    {
      "epoch": 0.03632332878581173,
      "grad_norm": 0.5890904664993286,
      "learning_rate": 1.999802329770066e-06,
      "loss": 0.6107,
      "step": 213
    },
    {
      "epoch": 0.03649386084583902,
      "grad_norm": 0.45250949263572693,
      "learning_rate": 1.999791195954793e-06,
      "loss": 0.5806,
      "step": 214
    },
    {
      "epoch": 0.0366643929058663,
      "grad_norm": 0.45594438910484314,
      "learning_rate": 1.9997797571465236e-06,
      "loss": 0.5946,
      "step": 215
    },
    {
      "epoch": 0.036834924965893585,
      "grad_norm": 0.567471981048584,
      "learning_rate": 1.999768013348747e-06,
      "loss": 0.6035,
      "step": 216
    },
    {
      "epoch": 0.037005457025920875,
      "grad_norm": 0.4891415238380432,
      "learning_rate": 1.9997559645650458e-06,
      "loss": 0.602,
      "step": 217
    },
    {
      "epoch": 0.03717598908594816,
      "grad_norm": 0.4505879878997803,
      "learning_rate": 1.9997436107990957e-06,
      "loss": 0.5841,
      "step": 218
    },
    {
      "epoch": 0.03734652114597544,
      "grad_norm": 0.3941557705402374,
      "learning_rate": 1.9997309520546647e-06,
      "loss": 0.6117,
      "step": 219
    },
    {
      "epoch": 0.03751705320600273,
      "grad_norm": 0.421119749546051,
      "learning_rate": 1.9997179883356153e-06,
      "loss": 0.5985,
      "step": 220
    },
    {
      "epoch": 0.03768758526603001,
      "grad_norm": 0.4365604519844055,
      "learning_rate": 1.999704719645902e-06,
      "loss": 0.5842,
      "step": 221
    },
    {
      "epoch": 0.037858117326057296,
      "grad_norm": 0.4222920536994934,
      "learning_rate": 1.9996911459895717e-06,
      "loss": 0.5883,
      "step": 222
    },
    {
      "epoch": 0.038028649386084586,
      "grad_norm": 0.39965325593948364,
      "learning_rate": 1.999677267370766e-06,
      "loss": 0.6064,
      "step": 223
    },
    {
      "epoch": 0.03819918144611187,
      "grad_norm": 0.42957979440689087,
      "learning_rate": 1.9996630837937187e-06,
      "loss": 0.594,
      "step": 224
    },
    {
      "epoch": 0.03836971350613915,
      "grad_norm": 0.5226802825927734,
      "learning_rate": 1.9996485952627556e-06,
      "loss": 0.5942,
      "step": 225
    },
    {
      "epoch": 0.03854024556616644,
      "grad_norm": 0.42182719707489014,
      "learning_rate": 1.9996338017822976e-06,
      "loss": 0.5774,
      "step": 226
    },
    {
      "epoch": 0.038710777626193724,
      "grad_norm": 0.39415428042411804,
      "learning_rate": 1.9996187033568567e-06,
      "loss": 0.5861,
      "step": 227
    },
    {
      "epoch": 0.038881309686221006,
      "grad_norm": 0.37894678115844727,
      "learning_rate": 1.9996032999910397e-06,
      "loss": 0.5761,
      "step": 228
    },
    {
      "epoch": 0.039051841746248296,
      "grad_norm": 0.4413073658943176,
      "learning_rate": 1.999587591689545e-06,
      "loss": 0.5966,
      "step": 229
    },
    {
      "epoch": 0.03922237380627558,
      "grad_norm": 0.47882312536239624,
      "learning_rate": 1.9995715784571643e-06,
      "loss": 0.61,
      "step": 230
    },
    {
      "epoch": 0.03939290586630286,
      "grad_norm": 0.49211686849594116,
      "learning_rate": 1.999555260298783e-06,
      "loss": 0.5948,
      "step": 231
    },
    {
      "epoch": 0.03956343792633015,
      "grad_norm": 0.4007163941860199,
      "learning_rate": 1.999538637219378e-06,
      "loss": 0.5921,
      "step": 232
    },
    {
      "epoch": 0.039733969986357434,
      "grad_norm": 0.628340482711792,
      "learning_rate": 1.9995217092240224e-06,
      "loss": 0.593,
      "step": 233
    },
    {
      "epoch": 0.03990450204638472,
      "grad_norm": 0.5330571532249451,
      "learning_rate": 1.999504476317878e-06,
      "loss": 0.5964,
      "step": 234
    },
    {
      "epoch": 0.04007503410641201,
      "grad_norm": 0.6231827735900879,
      "learning_rate": 1.9994869385062033e-06,
      "loss": 0.5889,
      "step": 235
    },
    {
      "epoch": 0.04024556616643929,
      "grad_norm": 0.5359860062599182,
      "learning_rate": 1.9994690957943474e-06,
      "loss": 0.6076,
      "step": 236
    },
    {
      "epoch": 0.04041609822646658,
      "grad_norm": 0.3939756751060486,
      "learning_rate": 1.9994509481877538e-06,
      "loss": 0.5759,
      "step": 237
    },
    {
      "epoch": 0.04058663028649386,
      "grad_norm": 0.43472591042518616,
      "learning_rate": 1.9994324956919587e-06,
      "loss": 0.5948,
      "step": 238
    },
    {
      "epoch": 0.040757162346521145,
      "grad_norm": 0.5427122116088867,
      "learning_rate": 1.9994137383125902e-06,
      "loss": 0.5963,
      "step": 239
    },
    {
      "epoch": 0.040927694406548434,
      "grad_norm": 0.3906486928462982,
      "learning_rate": 1.999394676055372e-06,
      "loss": 0.5919,
      "step": 240
    },
    {
      "epoch": 0.04109822646657572,
      "grad_norm": 0.45824775099754333,
      "learning_rate": 1.9993753089261175e-06,
      "loss": 0.5932,
      "step": 241
    },
    {
      "epoch": 0.041268758526603,
      "grad_norm": 0.4277664124965668,
      "learning_rate": 1.9993556369307358e-06,
      "loss": 0.6218,
      "step": 242
    },
    {
      "epoch": 0.04143929058663029,
      "grad_norm": 0.3937847912311554,
      "learning_rate": 1.999335660075228e-06,
      "loss": 0.5984,
      "step": 243
    },
    {
      "epoch": 0.04160982264665757,
      "grad_norm": 0.5465782880783081,
      "learning_rate": 1.999315378365687e-06,
      "loss": 0.5884,
      "step": 244
    },
    {
      "epoch": 0.041780354706684855,
      "grad_norm": 0.4681834876537323,
      "learning_rate": 1.9992947918083014e-06,
      "loss": 0.5912,
      "step": 245
    },
    {
      "epoch": 0.041950886766712145,
      "grad_norm": 0.5907049179077148,
      "learning_rate": 1.9992739004093505e-06,
      "loss": 0.5876,
      "step": 246
    },
    {
      "epoch": 0.04212141882673943,
      "grad_norm": 0.5100119113922119,
      "learning_rate": 1.9992527041752075e-06,
      "loss": 0.5964,
      "step": 247
    },
    {
      "epoch": 0.04229195088676671,
      "grad_norm": 0.3652644157409668,
      "learning_rate": 1.9992312031123382e-06,
      "loss": 0.5868,
      "step": 248
    },
    {
      "epoch": 0.042462482946794,
      "grad_norm": 0.4750256836414337,
      "learning_rate": 1.999209397227302e-06,
      "loss": 0.5878,
      "step": 249
    },
    {
      "epoch": 0.04263301500682128,
      "grad_norm": 0.7315317392349243,
      "learning_rate": 1.999187286526751e-06,
      "loss": 0.5781,
      "step": 250
    },
    {
      "epoch": 0.042803547066848566,
      "grad_norm": 0.5293675065040588,
      "learning_rate": 1.99916487101743e-06,
      "loss": 0.5927,
      "step": 251
    },
    {
      "epoch": 0.042974079126875855,
      "grad_norm": 0.3959099054336548,
      "learning_rate": 1.9991421507061763e-06,
      "loss": 0.5798,
      "step": 252
    },
    {
      "epoch": 0.04314461118690314,
      "grad_norm": 0.6561183929443359,
      "learning_rate": 1.9991191255999224e-06,
      "loss": 0.5831,
      "step": 253
    },
    {
      "epoch": 0.04331514324693042,
      "grad_norm": 0.6775041818618774,
      "learning_rate": 1.9990957957056913e-06,
      "loss": 0.6045,
      "step": 254
    },
    {
      "epoch": 0.04348567530695771,
      "grad_norm": 0.47806668281555176,
      "learning_rate": 1.9990721610305997e-06,
      "loss": 0.5921,
      "step": 255
    },
    {
      "epoch": 0.04365620736698499,
      "grad_norm": 0.38560113310813904,
      "learning_rate": 1.9990482215818584e-06,
      "loss": 0.5793,
      "step": 256
    },
    {
      "epoch": 0.043826739427012276,
      "grad_norm": 0.49209868907928467,
      "learning_rate": 1.9990239773667694e-06,
      "loss": 0.5913,
      "step": 257
    },
    {
      "epoch": 0.043997271487039566,
      "grad_norm": 0.4512802064418793,
      "learning_rate": 1.998999428392729e-06,
      "loss": 0.5981,
      "step": 258
    },
    {
      "epoch": 0.04416780354706685,
      "grad_norm": 0.3730643391609192,
      "learning_rate": 1.9989745746672258e-06,
      "loss": 0.5803,
      "step": 259
    },
    {
      "epoch": 0.04433833560709413,
      "grad_norm": 0.40619736909866333,
      "learning_rate": 1.998949416197842e-06,
      "loss": 0.5955,
      "step": 260
    },
    {
      "epoch": 0.04450886766712142,
      "grad_norm": 0.5041584372520447,
      "learning_rate": 1.998923952992252e-06,
      "loss": 0.589,
      "step": 261
    },
    {
      "epoch": 0.044679399727148704,
      "grad_norm": 0.43416324257850647,
      "learning_rate": 1.9988981850582243e-06,
      "loss": 0.5942,
      "step": 262
    },
    {
      "epoch": 0.04484993178717599,
      "grad_norm": 0.39469459652900696,
      "learning_rate": 1.9988721124036185e-06,
      "loss": 0.5744,
      "step": 263
    },
    {
      "epoch": 0.045020463847203276,
      "grad_norm": 0.5740176439285278,
      "learning_rate": 1.9988457350363883e-06,
      "loss": 0.5682,
      "step": 264
    },
    {
      "epoch": 0.04519099590723056,
      "grad_norm": 0.5018396973609924,
      "learning_rate": 1.998819052964581e-06,
      "loss": 0.59,
      "step": 265
    },
    {
      "epoch": 0.04536152796725784,
      "grad_norm": 0.5312762260437012,
      "learning_rate": 1.9987920661963364e-06,
      "loss": 0.5769,
      "step": 266
    },
    {
      "epoch": 0.04553206002728513,
      "grad_norm": 0.43061333894729614,
      "learning_rate": 1.9987647747398855e-06,
      "loss": 0.5836,
      "step": 267
    },
    {
      "epoch": 0.045702592087312414,
      "grad_norm": 0.3743135333061218,
      "learning_rate": 1.998737178603555e-06,
      "loss": 0.5939,
      "step": 268
    },
    {
      "epoch": 0.0458731241473397,
      "grad_norm": 0.5071547627449036,
      "learning_rate": 1.998709277795763e-06,
      "loss": 0.5825,
      "step": 269
    },
    {
      "epoch": 0.04604365620736699,
      "grad_norm": 0.3874017000198364,
      "learning_rate": 1.998681072325021e-06,
      "loss": 0.5973,
      "step": 270
    },
    {
      "epoch": 0.04621418826739427,
      "grad_norm": 0.46630197763442993,
      "learning_rate": 1.9986525621999325e-06,
      "loss": 0.5684,
      "step": 271
    },
    {
      "epoch": 0.04638472032742155,
      "grad_norm": 0.36790570616722107,
      "learning_rate": 1.9986237474291955e-06,
      "loss": 0.574,
      "step": 272
    },
    {
      "epoch": 0.04655525238744884,
      "grad_norm": 0.5044417977333069,
      "learning_rate": 1.9985946280215996e-06,
      "loss": 0.6082,
      "step": 273
    },
    {
      "epoch": 0.046725784447476125,
      "grad_norm": 0.5328083634376526,
      "learning_rate": 1.9985652039860284e-06,
      "loss": 0.5826,
      "step": 274
    },
    {
      "epoch": 0.04689631650750341,
      "grad_norm": 0.4631241261959076,
      "learning_rate": 1.998535475331458e-06,
      "loss": 0.5792,
      "step": 275
    },
    {
      "epoch": 0.0470668485675307,
      "grad_norm": 0.4844154417514801,
      "learning_rate": 1.9985054420669563e-06,
      "loss": 0.5739,
      "step": 276
    },
    {
      "epoch": 0.04723738062755798,
      "grad_norm": 0.3497856855392456,
      "learning_rate": 1.998475104201686e-06,
      "loss": 0.6017,
      "step": 277
    },
    {
      "epoch": 0.04740791268758526,
      "grad_norm": 0.42583030462265015,
      "learning_rate": 1.9984444617449024e-06,
      "loss": 0.5789,
      "step": 278
    },
    {
      "epoch": 0.04757844474761255,
      "grad_norm": 0.44119739532470703,
      "learning_rate": 1.998413514705952e-06,
      "loss": 0.5819,
      "step": 279
    },
    {
      "epoch": 0.047748976807639835,
      "grad_norm": 0.45854702591896057,
      "learning_rate": 1.9983822630942755e-06,
      "loss": 0.5996,
      "step": 280
    },
    {
      "epoch": 0.04791950886766712,
      "grad_norm": 0.5364755392074585,
      "learning_rate": 1.998350706919407e-06,
      "loss": 0.5845,
      "step": 281
    },
    {
      "epoch": 0.04809004092769441,
      "grad_norm": 0.4300476312637329,
      "learning_rate": 1.998318846190973e-06,
      "loss": 0.5721,
      "step": 282
    },
    {
      "epoch": 0.04826057298772169,
      "grad_norm": 0.42480289936065674,
      "learning_rate": 1.9982866809186923e-06,
      "loss": 0.583,
      "step": 283
    },
    {
      "epoch": 0.048431105047748974,
      "grad_norm": 0.4817947447299957,
      "learning_rate": 1.9982542111123776e-06,
      "loss": 0.583,
      "step": 284
    },
    {
      "epoch": 0.04860163710777626,
      "grad_norm": 0.4101852774620056,
      "learning_rate": 1.9982214367819334e-06,
      "loss": 0.6063,
      "step": 285
    },
    {
      "epoch": 0.048772169167803546,
      "grad_norm": 0.48095643520355225,
      "learning_rate": 1.998188357937358e-06,
      "loss": 0.5855,
      "step": 286
    },
    {
      "epoch": 0.04894270122783083,
      "grad_norm": 0.41014596819877625,
      "learning_rate": 1.998154974588743e-06,
      "loss": 0.6027,
      "step": 287
    },
    {
      "epoch": 0.04911323328785812,
      "grad_norm": 0.44107893109321594,
      "learning_rate": 1.9981212867462715e-06,
      "loss": 0.5899,
      "step": 288
    },
    {
      "epoch": 0.0492837653478854,
      "grad_norm": 0.3553464710712433,
      "learning_rate": 1.9980872944202205e-06,
      "loss": 0.5827,
      "step": 289
    },
    {
      "epoch": 0.04945429740791269,
      "grad_norm": 0.4601629376411438,
      "learning_rate": 1.998052997620959e-06,
      "loss": 0.6081,
      "step": 290
    },
    {
      "epoch": 0.049624829467939974,
      "grad_norm": 0.4139860272407532,
      "learning_rate": 1.9980183963589505e-06,
      "loss": 0.5803,
      "step": 291
    },
    {
      "epoch": 0.049795361527967257,
      "grad_norm": 0.36177587509155273,
      "learning_rate": 1.9979834906447494e-06,
      "loss": 0.5828,
      "step": 292
    },
    {
      "epoch": 0.049965893587994546,
      "grad_norm": 0.8259939551353455,
      "learning_rate": 1.9979482804890046e-06,
      "loss": 0.5832,
      "step": 293
    },
    {
      "epoch": 0.05013642564802183,
      "grad_norm": 0.43055734038352966,
      "learning_rate": 1.9979127659024564e-06,
      "loss": 0.572,
      "step": 294
    },
    {
      "epoch": 0.05030695770804911,
      "grad_norm": 0.4655931890010834,
      "learning_rate": 1.9978769468959393e-06,
      "loss": 0.572,
      "step": 295
    },
    {
      "epoch": 0.0504774897680764,
      "grad_norm": 0.5628933906555176,
      "learning_rate": 1.997840823480381e-06,
      "loss": 0.589,
      "step": 296
    },
    {
      "epoch": 0.050648021828103684,
      "grad_norm": 0.4914795756340027,
      "learning_rate": 1.9978043956667993e-06,
      "loss": 0.5885,
      "step": 297
    },
    {
      "epoch": 0.05081855388813097,
      "grad_norm": 0.43875017762184143,
      "learning_rate": 1.997767663466308e-06,
      "loss": 0.5857,
      "step": 298
    },
    {
      "epoch": 0.05098908594815826,
      "grad_norm": 0.3887084126472473,
      "learning_rate": 1.997730626890112e-06,
      "loss": 0.5991,
      "step": 299
    },
    {
      "epoch": 0.05115961800818554,
      "grad_norm": 0.679341197013855,
      "learning_rate": 1.99769328594951e-06,
      "loss": 0.5751,
      "step": 300
    },
    {
      "epoch": 0.05133015006821282,
      "grad_norm": 0.4228670597076416,
      "learning_rate": 1.997655640655893e-06,
      "loss": 0.5882,
      "step": 301
    },
    {
      "epoch": 0.05150068212824011,
      "grad_norm": 0.4219878613948822,
      "learning_rate": 1.9976176910207447e-06,
      "loss": 0.5812,
      "step": 302
    },
    {
      "epoch": 0.051671214188267395,
      "grad_norm": 0.3721599876880646,
      "learning_rate": 1.997579437055642e-06,
      "loss": 0.5886,
      "step": 303
    },
    {
      "epoch": 0.05184174624829468,
      "grad_norm": 0.5150498747825623,
      "learning_rate": 1.9975408787722545e-06,
      "loss": 0.5813,
      "step": 304
    },
    {
      "epoch": 0.05201227830832197,
      "grad_norm": 0.4999360740184784,
      "learning_rate": 1.9975020161823446e-06,
      "loss": 0.5814,
      "step": 305
    },
    {
      "epoch": 0.05218281036834925,
      "grad_norm": 0.6294407844543457,
      "learning_rate": 1.9974628492977682e-06,
      "loss": 0.5735,
      "step": 306
    },
    {
      "epoch": 0.05235334242837653,
      "grad_norm": 0.7571012377738953,
      "learning_rate": 1.9974233781304727e-06,
      "loss": 0.5746,
      "step": 307
    },
    {
      "epoch": 0.05252387448840382,
      "grad_norm": 0.5309390425682068,
      "learning_rate": 1.997383602692499e-06,
      "loss": 0.5875,
      "step": 308
    },
    {
      "epoch": 0.052694406548431105,
      "grad_norm": 0.3978341221809387,
      "learning_rate": 1.997343522995981e-06,
      "loss": 0.5964,
      "step": 309
    },
    {
      "epoch": 0.05286493860845839,
      "grad_norm": 0.46522951126098633,
      "learning_rate": 1.9973031390531458e-06,
      "loss": 0.5808,
      "step": 310
    },
    {
      "epoch": 0.05303547066848568,
      "grad_norm": 0.41105544567108154,
      "learning_rate": 1.9972624508763126e-06,
      "loss": 0.5744,
      "step": 311
    },
    {
      "epoch": 0.05320600272851296,
      "grad_norm": 0.4005431532859802,
      "learning_rate": 1.997221458477893e-06,
      "loss": 0.5859,
      "step": 312
    },
    {
      "epoch": 0.05337653478854024,
      "grad_norm": 0.510310709476471,
      "learning_rate": 1.9971801618703925e-06,
      "loss": 0.5831,
      "step": 313
    },
    {
      "epoch": 0.05354706684856753,
      "grad_norm": 0.5165649056434631,
      "learning_rate": 1.997138561066409e-06,
      "loss": 0.5892,
      "step": 314
    },
    {
      "epoch": 0.053717598908594816,
      "grad_norm": 0.46517428755760193,
      "learning_rate": 1.9970966560786327e-06,
      "loss": 0.5689,
      "step": 315
    },
    {
      "epoch": 0.0538881309686221,
      "grad_norm": 0.47656869888305664,
      "learning_rate": 1.9970544469198473e-06,
      "loss": 0.584,
      "step": 316
    },
    {
      "epoch": 0.05405866302864939,
      "grad_norm": 0.5681532621383667,
      "learning_rate": 1.9970119336029287e-06,
      "loss": 0.5822,
      "step": 317
    },
    {
      "epoch": 0.05422919508867667,
      "grad_norm": 0.6467899084091187,
      "learning_rate": 1.9969691161408462e-06,
      "loss": 0.5704,
      "step": 318
    },
    {
      "epoch": 0.054399727148703954,
      "grad_norm": 0.4735259711742401,
      "learning_rate": 1.9969259945466615e-06,
      "loss": 0.5892,
      "step": 319
    },
    {
      "epoch": 0.054570259208731244,
      "grad_norm": 0.4352891743183136,
      "learning_rate": 1.9968825688335293e-06,
      "loss": 0.58,
      "step": 320
    },
    {
      "epoch": 0.054740791268758526,
      "grad_norm": 0.5583917498588562,
      "learning_rate": 1.996838839014696e-06,
      "loss": 0.5783,
      "step": 321
    },
    {
      "epoch": 0.05491132332878581,
      "grad_norm": 0.3809167742729187,
      "learning_rate": 1.996794805103503e-06,
      "loss": 0.5912,
      "step": 322
    },
    {
      "epoch": 0.0550818553888131,
      "grad_norm": 0.5830413699150085,
      "learning_rate": 1.9967504671133822e-06,
      "loss": 0.5998,
      "step": 323
    },
    {
      "epoch": 0.05525238744884038,
      "grad_norm": 0.6099497675895691,
      "learning_rate": 1.9967058250578593e-06,
      "loss": 0.5764,
      "step": 324
    },
    {
      "epoch": 0.055422919508867664,
      "grad_norm": 0.4644138514995575,
      "learning_rate": 1.996660878950553e-06,
      "loss": 0.5787,
      "step": 325
    },
    {
      "epoch": 0.055593451568894954,
      "grad_norm": 0.5825083255767822,
      "learning_rate": 1.9966156288051743e-06,
      "loss": 0.5764,
      "step": 326
    },
    {
      "epoch": 0.05576398362892224,
      "grad_norm": 0.6491715908050537,
      "learning_rate": 1.9965700746355273e-06,
      "loss": 0.572,
      "step": 327
    },
    {
      "epoch": 0.05593451568894952,
      "grad_norm": 0.5463743209838867,
      "learning_rate": 1.9965242164555077e-06,
      "loss": 0.5911,
      "step": 328
    },
    {
      "epoch": 0.05610504774897681,
      "grad_norm": 0.4275131821632385,
      "learning_rate": 1.996478054279106e-06,
      "loss": 0.571,
      "step": 329
    },
    {
      "epoch": 0.05627557980900409,
      "grad_norm": 0.4297434687614441,
      "learning_rate": 1.996431588120403e-06,
      "loss": 0.5778,
      "step": 330
    },
    {
      "epoch": 0.056446111869031375,
      "grad_norm": 0.5487849712371826,
      "learning_rate": 1.996384817993575e-06,
      "loss": 0.5717,
      "step": 331
    },
    {
      "epoch": 0.056616643929058665,
      "grad_norm": 0.4654536545276642,
      "learning_rate": 1.9963377439128886e-06,
      "loss": 0.5827,
      "step": 332
    },
    {
      "epoch": 0.05678717598908595,
      "grad_norm": 0.43091699481010437,
      "learning_rate": 1.996290365892704e-06,
      "loss": 0.5822,
      "step": 333
    },
    {
      "epoch": 0.05695770804911323,
      "grad_norm": 0.4967304468154907,
      "learning_rate": 1.9962426839474745e-06,
      "loss": 0.5858,
      "step": 334
    },
    {
      "epoch": 0.05712824010914052,
      "grad_norm": 0.39026960730552673,
      "learning_rate": 1.996194698091746e-06,
      "loss": 0.5835,
      "step": 335
    },
    {
      "epoch": 0.0572987721691678,
      "grad_norm": 0.4674072861671448,
      "learning_rate": 1.996146408340156e-06,
      "loss": 0.5748,
      "step": 336
    },
    {
      "epoch": 0.057469304229195085,
      "grad_norm": 0.38436296582221985,
      "learning_rate": 1.996097814707437e-06,
      "loss": 0.5794,
      "step": 337
    },
    {
      "epoch": 0.057639836289222375,
      "grad_norm": 0.33945173025131226,
      "learning_rate": 1.996048917208412e-06,
      "loss": 0.5787,
      "step": 338
    },
    {
      "epoch": 0.05781036834924966,
      "grad_norm": 0.45001330971717834,
      "learning_rate": 1.995999715857997e-06,
      "loss": 0.5771,
      "step": 339
    },
    {
      "epoch": 0.05798090040927694,
      "grad_norm": 0.37709715962409973,
      "learning_rate": 1.995950210671202e-06,
      "loss": 0.5987,
      "step": 340
    },
    {
      "epoch": 0.05815143246930423,
      "grad_norm": 0.48131653666496277,
      "learning_rate": 1.9959004016631293e-06,
      "loss": 0.5858,
      "step": 341
    },
    {
      "epoch": 0.05832196452933151,
      "grad_norm": 0.4368843734264374,
      "learning_rate": 1.995850288848972e-06,
      "loss": 0.5897,
      "step": 342
    },
    {
      "epoch": 0.0584924965893588,
      "grad_norm": 0.4104810655117035,
      "learning_rate": 1.9957998722440187e-06,
      "loss": 0.5781,
      "step": 343
    },
    {
      "epoch": 0.058663028649386086,
      "grad_norm": 0.5733034014701843,
      "learning_rate": 1.9957491518636482e-06,
      "loss": 0.5789,
      "step": 344
    },
    {
      "epoch": 0.05883356070941337,
      "grad_norm": 0.37760886549949646,
      "learning_rate": 1.9956981277233343e-06,
      "loss": 0.5799,
      "step": 345
    },
    {
      "epoch": 0.05900409276944066,
      "grad_norm": 0.4927324056625366,
      "learning_rate": 1.9956467998386414e-06,
      "loss": 0.5748,
      "step": 346
    },
    {
      "epoch": 0.05917462482946794,
      "grad_norm": 0.5612856149673462,
      "learning_rate": 1.9955951682252276e-06,
      "loss": 0.5835,
      "step": 347
    },
    {
      "epoch": 0.059345156889495224,
      "grad_norm": 0.7066859006881714,
      "learning_rate": 1.9955432328988437e-06,
      "loss": 0.5854,
      "step": 348
    },
    {
      "epoch": 0.05951568894952251,
      "grad_norm": 0.6565737128257751,
      "learning_rate": 1.9954909938753328e-06,
      "loss": 0.5813,
      "step": 349
    },
    {
      "epoch": 0.059686221009549796,
      "grad_norm": 0.41528597474098206,
      "learning_rate": 1.9954384511706304e-06,
      "loss": 0.5807,
      "step": 350
    },
    {
      "epoch": 0.05985675306957708,
      "grad_norm": 0.5107154250144958,
      "learning_rate": 1.9953856048007657e-06,
      "loss": 0.6013,
      "step": 351
    },
    {
      "epoch": 0.06002728512960437,
      "grad_norm": 0.5143330097198486,
      "learning_rate": 1.9953324547818592e-06,
      "loss": 0.5769,
      "step": 352
    },
    {
      "epoch": 0.06019781718963165,
      "grad_norm": 0.47564592957496643,
      "learning_rate": 1.995279001130125e-06,
      "loss": 0.5918,
      "step": 353
    },
    {
      "epoch": 0.060368349249658934,
      "grad_norm": 0.44760820269584656,
      "learning_rate": 1.9952252438618693e-06,
      "loss": 0.5785,
      "step": 354
    },
    {
      "epoch": 0.060538881309686224,
      "grad_norm": 0.37445268034935,
      "learning_rate": 1.9951711829934916e-06,
      "loss": 0.5714,
      "step": 355
    },
    {
      "epoch": 0.06070941336971351,
      "grad_norm": 0.7027251720428467,
      "learning_rate": 1.9951168185414827e-06,
      "loss": 0.5701,
      "step": 356
    },
    {
      "epoch": 0.06087994542974079,
      "grad_norm": 0.8148019909858704,
      "learning_rate": 1.9950621505224278e-06,
      "loss": 0.5913,
      "step": 357
    },
    {
      "epoch": 0.06105047748976808,
      "grad_norm": 0.47353798151016235,
      "learning_rate": 1.995007178953003e-06,
      "loss": 0.5964,
      "step": 358
    },
    {
      "epoch": 0.06122100954979536,
      "grad_norm": 0.5995876789093018,
      "learning_rate": 1.994951903849978e-06,
      "loss": 0.5922,
      "step": 359
    },
    {
      "epoch": 0.061391541609822645,
      "grad_norm": 0.8017838597297668,
      "learning_rate": 1.994896325230215e-06,
      "loss": 0.5747,
      "step": 360
    },
    {
      "epoch": 0.061562073669849934,
      "grad_norm": 0.5751956105232239,
      "learning_rate": 1.994840443110668e-06,
      "loss": 0.5922,
      "step": 361
    },
    {
      "epoch": 0.06173260572987722,
      "grad_norm": 0.4752778112888336,
      "learning_rate": 1.9947842575083853e-06,
      "loss": 0.6171,
      "step": 362
    },
    {
      "epoch": 0.0619031377899045,
      "grad_norm": 0.684117317199707,
      "learning_rate": 1.9947277684405055e-06,
      "loss": 0.5765,
      "step": 363
    },
    {
      "epoch": 0.06207366984993179,
      "grad_norm": 0.6569706797599792,
      "learning_rate": 1.994670975924262e-06,
      "loss": 0.575,
      "step": 364
    },
    {
      "epoch": 0.06224420190995907,
      "grad_norm": 0.47546330094337463,
      "learning_rate": 1.9946138799769792e-06,
      "loss": 0.5873,
      "step": 365
    },
    {
      "epoch": 0.062414733969986355,
      "grad_norm": 0.6601859331130981,
      "learning_rate": 1.9945564806160746e-06,
      "loss": 0.571,
      "step": 366
    },
    {
      "epoch": 0.06258526603001364,
      "grad_norm": 0.48071083426475525,
      "learning_rate": 1.994498777859058e-06,
      "loss": 0.5712,
      "step": 367
    },
    {
      "epoch": 0.06275579809004093,
      "grad_norm": 0.5320371985435486,
      "learning_rate": 1.994440771723533e-06,
      "loss": 0.6049,
      "step": 368
    },
    {
      "epoch": 0.06292633015006821,
      "grad_norm": 0.5746079087257385,
      "learning_rate": 1.9943824622271938e-06,
      "loss": 0.5853,
      "step": 369
    },
    {
      "epoch": 0.0630968622100955,
      "grad_norm": 0.42983976006507874,
      "learning_rate": 1.9943238493878286e-06,
      "loss": 0.5753,
      "step": 370
    },
    {
      "epoch": 0.06326739427012279,
      "grad_norm": 0.4530344307422638,
      "learning_rate": 1.994264933223317e-06,
      "loss": 0.5919,
      "step": 371
    },
    {
      "epoch": 0.06343792633015007,
      "grad_norm": 0.5168627500534058,
      "learning_rate": 1.9942057137516327e-06,
      "loss": 0.579,
      "step": 372
    },
    {
      "epoch": 0.06360845839017736,
      "grad_norm": 0.4011041820049286,
      "learning_rate": 1.9941461909908402e-06,
      "loss": 0.5559,
      "step": 373
    },
    {
      "epoch": 0.06377899045020465,
      "grad_norm": 0.44738543033599854,
      "learning_rate": 1.9940863649590977e-06,
      "loss": 0.5696,
      "step": 374
    },
    {
      "epoch": 0.06394952251023192,
      "grad_norm": 0.4560238718986511,
      "learning_rate": 1.9940262356746557e-06,
      "loss": 0.5892,
      "step": 375
    },
    {
      "epoch": 0.06412005457025921,
      "grad_norm": 0.4311150908470154,
      "learning_rate": 1.9939658031558566e-06,
      "loss": 0.6006,
      "step": 376
    },
    {
      "epoch": 0.0642905866302865,
      "grad_norm": 0.5050395131111145,
      "learning_rate": 1.993905067421136e-06,
      "loss": 0.5677,
      "step": 377
    },
    {
      "epoch": 0.06446111869031378,
      "grad_norm": 0.44102242588996887,
      "learning_rate": 1.9938440284890217e-06,
      "loss": 0.5881,
      "step": 378
    },
    {
      "epoch": 0.06463165075034107,
      "grad_norm": 0.3900633156299591,
      "learning_rate": 1.993782686378134e-06,
      "loss": 0.5681,
      "step": 379
    },
    {
      "epoch": 0.06480218281036836,
      "grad_norm": 0.4846239686012268,
      "learning_rate": 1.9937210411071857e-06,
      "loss": 0.5726,
      "step": 380
    },
    {
      "epoch": 0.06497271487039563,
      "grad_norm": 0.4322761297225952,
      "learning_rate": 1.9936590926949824e-06,
      "loss": 0.575,
      "step": 381
    },
    {
      "epoch": 0.06514324693042292,
      "grad_norm": 0.43640026450157166,
      "learning_rate": 1.9935968411604214e-06,
      "loss": 0.5758,
      "step": 382
    },
    {
      "epoch": 0.06531377899045021,
      "grad_norm": 0.44950467348098755,
      "learning_rate": 1.9935342865224937e-06,
      "loss": 0.5948,
      "step": 383
    },
    {
      "epoch": 0.06548431105047749,
      "grad_norm": 0.6357623934745789,
      "learning_rate": 1.993471428800281e-06,
      "loss": 0.5728,
      "step": 384
    },
    {
      "epoch": 0.06565484311050478,
      "grad_norm": 0.6120256781578064,
      "learning_rate": 1.993408268012959e-06,
      "loss": 0.5658,
      "step": 385
    },
    {
      "epoch": 0.06582537517053207,
      "grad_norm": 0.3851131796836853,
      "learning_rate": 1.993344804179795e-06,
      "loss": 0.5744,
      "step": 386
    },
    {
      "epoch": 0.06599590723055934,
      "grad_norm": 0.495846688747406,
      "learning_rate": 1.99328103732015e-06,
      "loss": 0.5777,
      "step": 387
    },
    {
      "epoch": 0.06616643929058663,
      "grad_norm": 0.824712872505188,
      "learning_rate": 1.9932169674534752e-06,
      "loss": 0.5611,
      "step": 388
    },
    {
      "epoch": 0.06633697135061392,
      "grad_norm": 0.8101972937583923,
      "learning_rate": 1.9931525945993165e-06,
      "loss": 0.5836,
      "step": 389
    },
    {
      "epoch": 0.0665075034106412,
      "grad_norm": 0.4071904420852661,
      "learning_rate": 1.993087918777311e-06,
      "loss": 0.575,
      "step": 390
    },
    {
      "epoch": 0.06667803547066849,
      "grad_norm": 0.5939446687698364,
      "learning_rate": 1.9930229400071888e-06,
      "loss": 0.6035,
      "step": 391
    },
    {
      "epoch": 0.06684856753069578,
      "grad_norm": 0.657426118850708,
      "learning_rate": 1.9929576583087716e-06,
      "loss": 0.579,
      "step": 392
    },
    {
      "epoch": 0.06701909959072305,
      "grad_norm": 0.4552478492259979,
      "learning_rate": 1.9928920737019734e-06,
      "loss": 0.5747,
      "step": 393
    },
    {
      "epoch": 0.06718963165075034,
      "grad_norm": 0.42880189418792725,
      "learning_rate": 1.992826186206803e-06,
      "loss": 0.5828,
      "step": 394
    },
    {
      "epoch": 0.06736016371077763,
      "grad_norm": 0.4603235423564911,
      "learning_rate": 1.9927599958433586e-06,
      "loss": 0.5822,
      "step": 395
    },
    {
      "epoch": 0.06753069577080491,
      "grad_norm": 0.5346198081970215,
      "learning_rate": 1.9926935026318323e-06,
      "loss": 0.5906,
      "step": 396
    },
    {
      "epoch": 0.0677012278308322,
      "grad_norm": 0.4686392843723297,
      "learning_rate": 1.992626706592508e-06,
      "loss": 0.5619,
      "step": 397
    },
    {
      "epoch": 0.06787175989085949,
      "grad_norm": 0.4141104817390442,
      "learning_rate": 1.9925596077457624e-06,
      "loss": 0.5676,
      "step": 398
    },
    {
      "epoch": 0.06804229195088676,
      "grad_norm": 0.42539089918136597,
      "learning_rate": 1.9924922061120644e-06,
      "loss": 0.571,
      "step": 399
    },
    {
      "epoch": 0.06821282401091405,
      "grad_norm": 0.44370517134666443,
      "learning_rate": 1.992424501711976e-06,
      "loss": 0.5882,
      "step": 400
    },
    {
      "epoch": 0.06838335607094134,
      "grad_norm": 0.43443042039871216,
      "learning_rate": 1.9923564945661506e-06,
      "loss": 0.5742,
      "step": 401
    },
    {
      "epoch": 0.06855388813096862,
      "grad_norm": 0.4402327835559845,
      "learning_rate": 1.9922881846953335e-06,
      "loss": 0.5759,
      "step": 402
    },
    {
      "epoch": 0.06872442019099591,
      "grad_norm": 0.35972359776496887,
      "learning_rate": 1.992219572120364e-06,
      "loss": 0.5733,
      "step": 403
    },
    {
      "epoch": 0.0688949522510232,
      "grad_norm": 0.4337236285209656,
      "learning_rate": 1.992150656862172e-06,
      "loss": 0.572,
      "step": 404
    },
    {
      "epoch": 0.06906548431105047,
      "grad_norm": 0.45187443494796753,
      "learning_rate": 1.9920814389417814e-06,
      "loss": 0.563,
      "step": 405
    },
    {
      "epoch": 0.06923601637107776,
      "grad_norm": 0.4660555422306061,
      "learning_rate": 1.9920119183803068e-06,
      "loss": 0.608,
      "step": 406
    },
    {
      "epoch": 0.06940654843110505,
      "grad_norm": 0.41995012760162354,
      "learning_rate": 1.9919420951989565e-06,
      "loss": 0.578,
      "step": 407
    },
    {
      "epoch": 0.06957708049113233,
      "grad_norm": 0.5486136078834534,
      "learning_rate": 1.99187196941903e-06,
      "loss": 0.5737,
      "step": 408
    },
    {
      "epoch": 0.06974761255115962,
      "grad_norm": 0.6215372085571289,
      "learning_rate": 1.9918015410619203e-06,
      "loss": 0.575,
      "step": 409
    },
    {
      "epoch": 0.06991814461118691,
      "grad_norm": 0.5288968086242676,
      "learning_rate": 1.9917308101491118e-06,
      "loss": 0.5749,
      "step": 410
    },
    {
      "epoch": 0.07008867667121418,
      "grad_norm": 0.5099157094955444,
      "learning_rate": 1.991659776702181e-06,
      "loss": 0.5823,
      "step": 411
    },
    {
      "epoch": 0.07025920873124147,
      "grad_norm": 0.46361568570137024,
      "learning_rate": 1.991588440742797e-06,
      "loss": 0.5711,
      "step": 412
    },
    {
      "epoch": 0.07042974079126876,
      "grad_norm": 0.47007569670677185,
      "learning_rate": 1.991516802292723e-06,
      "loss": 0.5746,
      "step": 413
    },
    {
      "epoch": 0.07060027285129604,
      "grad_norm": 0.5757049322128296,
      "learning_rate": 1.991444861373811e-06,
      "loss": 0.5821,
      "step": 414
    },
    {
      "epoch": 0.07077080491132333,
      "grad_norm": 0.4328621029853821,
      "learning_rate": 1.991372618008007e-06,
      "loss": 0.5736,
      "step": 415
    },
    {
      "epoch": 0.07094133697135062,
      "grad_norm": 0.501839816570282,
      "learning_rate": 1.991300072217351e-06,
      "loss": 0.5821,
      "step": 416
    },
    {
      "epoch": 0.0711118690313779,
      "grad_norm": 0.36504441499710083,
      "learning_rate": 1.9912272240239716e-06,
      "loss": 0.5882,
      "step": 417
    },
    {
      "epoch": 0.07128240109140518,
      "grad_norm": 0.4125383794307709,
      "learning_rate": 1.9911540734500932e-06,
      "loss": 0.5628,
      "step": 418
    },
    {
      "epoch": 0.07145293315143247,
      "grad_norm": 0.40303322672843933,
      "learning_rate": 1.99108062051803e-06,
      "loss": 0.6005,
      "step": 419
    },
    {
      "epoch": 0.07162346521145975,
      "grad_norm": 0.4227195978164673,
      "learning_rate": 1.9910068652501893e-06,
      "loss": 0.5529,
      "step": 420
    },
    {
      "epoch": 0.07179399727148704,
      "grad_norm": 0.37034130096435547,
      "learning_rate": 1.9909328076690713e-06,
      "loss": 0.5755,
      "step": 421
    },
    {
      "epoch": 0.07196452933151433,
      "grad_norm": 0.38170015811920166,
      "learning_rate": 1.990858447797267e-06,
      "loss": 0.5717,
      "step": 422
    },
    {
      "epoch": 0.0721350613915416,
      "grad_norm": 0.4700663387775421,
      "learning_rate": 1.990783785657461e-06,
      "loss": 0.582,
      "step": 423
    },
    {
      "epoch": 0.0723055934515689,
      "grad_norm": 0.42466485500335693,
      "learning_rate": 1.990708821272429e-06,
      "loss": 0.5798,
      "step": 424
    },
    {
      "epoch": 0.07247612551159618,
      "grad_norm": 0.44316986203193665,
      "learning_rate": 1.9906335546650395e-06,
      "loss": 0.555,
      "step": 425
    },
    {
      "epoch": 0.07264665757162346,
      "grad_norm": 0.36553263664245605,
      "learning_rate": 1.990557985858253e-06,
      "loss": 0.5596,
      "step": 426
    },
    {
      "epoch": 0.07281718963165075,
      "grad_norm": 0.4100362956523895,
      "learning_rate": 1.990482114875123e-06,
      "loss": 0.5537,
      "step": 427
    },
    {
      "epoch": 0.07298772169167804,
      "grad_norm": 0.4145418703556061,
      "learning_rate": 1.9904059417387935e-06,
      "loss": 0.5602,
      "step": 428
    },
    {
      "epoch": 0.07315825375170532,
      "grad_norm": 0.42742791771888733,
      "learning_rate": 1.9903294664725023e-06,
      "loss": 0.5833,
      "step": 429
    },
    {
      "epoch": 0.0733287858117326,
      "grad_norm": 0.33802473545074463,
      "learning_rate": 1.990252689099578e-06,
      "loss": 0.5578,
      "step": 430
    },
    {
      "epoch": 0.0734993178717599,
      "grad_norm": 0.4198910593986511,
      "learning_rate": 1.9901756096434426e-06,
      "loss": 0.5842,
      "step": 431
    },
    {
      "epoch": 0.07366984993178717,
      "grad_norm": 0.36407625675201416,
      "learning_rate": 1.99009822812761e-06,
      "loss": 0.5658,
      "step": 432
    },
    {
      "epoch": 0.07384038199181446,
      "grad_norm": 0.36379313468933105,
      "learning_rate": 1.9900205445756847e-06,
      "loss": 0.57,
      "step": 433
    },
    {
      "epoch": 0.07401091405184175,
      "grad_norm": 0.4023984372615814,
      "learning_rate": 1.989942559011366e-06,
      "loss": 0.577,
      "step": 434
    },
    {
      "epoch": 0.07418144611186903,
      "grad_norm": 0.3459740877151489,
      "learning_rate": 1.989864271458443e-06,
      "loss": 0.5719,
      "step": 435
    },
    {
      "epoch": 0.07435197817189632,
      "grad_norm": 0.4210052788257599,
      "learning_rate": 1.9897856819407983e-06,
      "loss": 0.5766,
      "step": 436
    },
    {
      "epoch": 0.0745225102319236,
      "grad_norm": 0.3438369333744049,
      "learning_rate": 1.989706790482406e-06,
      "loss": 0.5681,
      "step": 437
    },
    {
      "epoch": 0.07469304229195088,
      "grad_norm": 0.3840697407722473,
      "learning_rate": 1.9896275971073327e-06,
      "loss": 0.5962,
      "step": 438
    },
    {
      "epoch": 0.07486357435197817,
      "grad_norm": 0.3564974069595337,
      "learning_rate": 1.9895481018397363e-06,
      "loss": 0.5657,
      "step": 439
    },
    {
      "epoch": 0.07503410641200546,
      "grad_norm": 0.3731328547000885,
      "learning_rate": 1.989468304703868e-06,
      "loss": 0.569,
      "step": 440
    },
    {
      "epoch": 0.07520463847203274,
      "grad_norm": 0.42112016677856445,
      "learning_rate": 1.98938820572407e-06,
      "loss": 0.5673,
      "step": 441
    },
    {
      "epoch": 0.07537517053206003,
      "grad_norm": 0.429995596408844,
      "learning_rate": 1.989307804924777e-06,
      "loss": 0.5469,
      "step": 442
    },
    {
      "epoch": 0.07554570259208732,
      "grad_norm": 0.46121692657470703,
      "learning_rate": 1.9892271023305163e-06,
      "loss": 0.5784,
      "step": 443
    },
    {
      "epoch": 0.07571623465211459,
      "grad_norm": 0.5717109441757202,
      "learning_rate": 1.9891460979659064e-06,
      "loss": 0.5791,
      "step": 444
    },
    {
      "epoch": 0.07588676671214188,
      "grad_norm": 0.5568169355392456,
      "learning_rate": 1.9890647918556583e-06,
      "loss": 0.5677,
      "step": 445
    },
    {
      "epoch": 0.07605729877216917,
      "grad_norm": 0.39502644538879395,
      "learning_rate": 1.988983184024575e-06,
      "loss": 0.5864,
      "step": 446
    },
    {
      "epoch": 0.07622783083219645,
      "grad_norm": 0.4009121060371399,
      "learning_rate": 1.988901274497551e-06,
      "loss": 0.569,
      "step": 447
    },
    {
      "epoch": 0.07639836289222374,
      "grad_norm": 0.5456749796867371,
      "learning_rate": 1.988819063299574e-06,
      "loss": 0.5593,
      "step": 448
    },
    {
      "epoch": 0.07656889495225103,
      "grad_norm": 0.5741431713104248,
      "learning_rate": 1.9887365504557236e-06,
      "loss": 0.5749,
      "step": 449
    },
    {
      "epoch": 0.0767394270122783,
      "grad_norm": 0.530440628528595,
      "learning_rate": 1.9886537359911693e-06,
      "loss": 0.589,
      "step": 450
    },
    {
      "epoch": 0.07690995907230559,
      "grad_norm": 0.362716406583786,
      "learning_rate": 1.9885706199311757e-06,
      "loss": 0.5791,
      "step": 451
    },
    {
      "epoch": 0.07708049113233288,
      "grad_norm": 0.5184724926948547,
      "learning_rate": 1.9884872023010974e-06,
      "loss": 0.5861,
      "step": 452
    },
    {
      "epoch": 0.07725102319236016,
      "grad_norm": 0.4775357246398926,
      "learning_rate": 1.988403483126381e-06,
      "loss": 0.57,
      "step": 453
    },
    {
      "epoch": 0.07742155525238745,
      "grad_norm": 0.46023473143577576,
      "learning_rate": 1.988319462432566e-06,
      "loss": 0.5847,
      "step": 454
    },
    {
      "epoch": 0.07759208731241474,
      "grad_norm": 0.6719965934753418,
      "learning_rate": 1.9882351402452837e-06,
      "loss": 0.5629,
      "step": 455
    },
    {
      "epoch": 0.07776261937244201,
      "grad_norm": 0.7485231161117554,
      "learning_rate": 1.988150516590257e-06,
      "loss": 0.5688,
      "step": 456
    },
    {
      "epoch": 0.0779331514324693,
      "grad_norm": 0.5788260102272034,
      "learning_rate": 1.988065591493301e-06,
      "loss": 0.5696,
      "step": 457
    },
    {
      "epoch": 0.07810368349249659,
      "grad_norm": 0.5395895838737488,
      "learning_rate": 1.987980364980322e-06,
      "loss": 0.5785,
      "step": 458
    },
    {
      "epoch": 0.07827421555252387,
      "grad_norm": 0.43305692076683044,
      "learning_rate": 1.9878948370773196e-06,
      "loss": 0.5633,
      "step": 459
    },
    {
      "epoch": 0.07844474761255116,
      "grad_norm": 0.569288432598114,
      "learning_rate": 1.987809007810385e-06,
      "loss": 0.5796,
      "step": 460
    },
    {
      "epoch": 0.07861527967257845,
      "grad_norm": 0.49687209725379944,
      "learning_rate": 1.9877228772056993e-06,
      "loss": 0.5695,
      "step": 461
    },
    {
      "epoch": 0.07878581173260572,
      "grad_norm": 0.3611188530921936,
      "learning_rate": 1.9876364452895394e-06,
      "loss": 0.5752,
      "step": 462
    },
    {
      "epoch": 0.07895634379263301,
      "grad_norm": 0.5431200265884399,
      "learning_rate": 1.9875497120882705e-06,
      "loss": 0.5878,
      "step": 463
    },
    {
      "epoch": 0.0791268758526603,
      "grad_norm": 0.3528672158718109,
      "learning_rate": 1.987462677628352e-06,
      "loss": 0.5591,
      "step": 464
    },
    {
      "epoch": 0.07929740791268758,
      "grad_norm": 0.5321730971336365,
      "learning_rate": 1.9873753419363336e-06,
      "loss": 0.5747,
      "step": 465
    },
    {
      "epoch": 0.07946793997271487,
      "grad_norm": 0.4939395487308502,
      "learning_rate": 1.9872877050388583e-06,
      "loss": 0.5767,
      "step": 466
    },
    {
      "epoch": 0.07963847203274216,
      "grad_norm": 0.4068276882171631,
      "learning_rate": 1.9871997669626597e-06,
      "loss": 0.5797,
      "step": 467
    },
    {
      "epoch": 0.07980900409276943,
      "grad_norm": 0.4314722418785095,
      "learning_rate": 1.9871115277345643e-06,
      "loss": 0.5579,
      "step": 468
    },
    {
      "epoch": 0.07997953615279672,
      "grad_norm": 0.4068446457386017,
      "learning_rate": 1.98702298738149e-06,
      "loss": 0.5835,
      "step": 469
    },
    {
      "epoch": 0.08015006821282401,
      "grad_norm": 0.3883475661277771,
      "learning_rate": 1.9869341459304463e-06,
      "loss": 0.5777,
      "step": 470
    },
    {
      "epoch": 0.08032060027285129,
      "grad_norm": 0.37347447872161865,
      "learning_rate": 1.9868450034085355e-06,
      "loss": 0.5686,
      "step": 471
    },
    {
      "epoch": 0.08049113233287858,
      "grad_norm": 0.40475261211395264,
      "learning_rate": 1.9867555598429507e-06,
      "loss": 0.5955,
      "step": 472
    },
    {
      "epoch": 0.08066166439290587,
      "grad_norm": 0.44075220823287964,
      "learning_rate": 1.9866658152609774e-06,
      "loss": 0.5733,
      "step": 473
    },
    {
      "epoch": 0.08083219645293316,
      "grad_norm": 0.4294121563434601,
      "learning_rate": 1.9865757696899925e-06,
      "loss": 0.5787,
      "step": 474
    },
    {
      "epoch": 0.08100272851296043,
      "grad_norm": 0.4231039583683014,
      "learning_rate": 1.9864854231574652e-06,
      "loss": 0.5594,
      "step": 475
    },
    {
      "epoch": 0.08117326057298772,
      "grad_norm": 0.3832063674926758,
      "learning_rate": 1.9863947756909565e-06,
      "loss": 0.5884,
      "step": 476
    },
    {
      "epoch": 0.08134379263301501,
      "grad_norm": 0.4569307267665863,
      "learning_rate": 1.986303827318119e-06,
      "loss": 0.5579,
      "step": 477
    },
    {
      "epoch": 0.08151432469304229,
      "grad_norm": 0.5801775455474854,
      "learning_rate": 1.986212578066697e-06,
      "loss": 0.5844,
      "step": 478
    },
    {
      "epoch": 0.08168485675306958,
      "grad_norm": 0.4718495309352875,
      "learning_rate": 1.9861210279645262e-06,
      "loss": 0.5863,
      "step": 479
    },
    {
      "epoch": 0.08185538881309687,
      "grad_norm": 0.42160969972610474,
      "learning_rate": 1.9860291770395352e-06,
      "loss": 0.5624,
      "step": 480
    },
    {
      "epoch": 0.08202592087312414,
      "grad_norm": 0.430442214012146,
      "learning_rate": 1.9859370253197438e-06,
      "loss": 0.597,
      "step": 481
    },
    {
      "epoch": 0.08219645293315143,
      "grad_norm": 0.40239301323890686,
      "learning_rate": 1.985844572833263e-06,
      "loss": 0.564,
      "step": 482
    },
    {
      "epoch": 0.08236698499317872,
      "grad_norm": 0.4097753167152405,
      "learning_rate": 1.9857518196082966e-06,
      "loss": 0.556,
      "step": 483
    },
    {
      "epoch": 0.082537517053206,
      "grad_norm": 0.417348712682724,
      "learning_rate": 1.9856587656731393e-06,
      "loss": 0.5774,
      "step": 484
    },
    {
      "epoch": 0.08270804911323329,
      "grad_norm": 0.41074228286743164,
      "learning_rate": 1.985565411056178e-06,
      "loss": 0.5638,
      "step": 485
    },
    {
      "epoch": 0.08287858117326058,
      "grad_norm": 0.40855127573013306,
      "learning_rate": 1.9854717557858907e-06,
      "loss": 0.5903,
      "step": 486
    },
    {
      "epoch": 0.08304911323328786,
      "grad_norm": 0.4109269380569458,
      "learning_rate": 1.985377799890848e-06,
      "loss": 0.5752,
      "step": 487
    },
    {
      "epoch": 0.08321964529331514,
      "grad_norm": 0.5294615030288696,
      "learning_rate": 1.9852835433997114e-06,
      "loss": 0.5646,
      "step": 488
    },
    {
      "epoch": 0.08339017735334243,
      "grad_norm": 0.3962743282318115,
      "learning_rate": 1.9851889863412346e-06,
      "loss": 0.5778,
      "step": 489
    },
    {
      "epoch": 0.08356070941336971,
      "grad_norm": 0.39832717180252075,
      "learning_rate": 1.985094128744264e-06,
      "loss": 0.5866,
      "step": 490
    },
    {
      "epoch": 0.083731241473397,
      "grad_norm": 0.42551106214523315,
      "learning_rate": 1.9849989706377344e-06,
      "loss": 0.5749,
      "step": 491
    },
    {
      "epoch": 0.08390177353342429,
      "grad_norm": 0.3772289454936981,
      "learning_rate": 1.9849035120506757e-06,
      "loss": 0.575,
      "step": 492
    },
    {
      "epoch": 0.08407230559345157,
      "grad_norm": 0.3782070279121399,
      "learning_rate": 1.9848077530122084e-06,
      "loss": 0.5703,
      "step": 493
    },
    {
      "epoch": 0.08424283765347886,
      "grad_norm": 0.38931572437286377,
      "learning_rate": 1.984711693551544e-06,
      "loss": 0.5645,
      "step": 494
    },
    {
      "epoch": 0.08441336971350615,
      "grad_norm": 0.3773365020751953,
      "learning_rate": 1.9846153336979857e-06,
      "loss": 0.5983,
      "step": 495
    },
    {
      "epoch": 0.08458390177353342,
      "grad_norm": 0.47619354724884033,
      "learning_rate": 1.9845186734809297e-06,
      "loss": 0.5838,
      "step": 496
    },
    {
      "epoch": 0.08475443383356071,
      "grad_norm": 0.5386983752250671,
      "learning_rate": 1.9844217129298618e-06,
      "loss": 0.5664,
      "step": 497
    },
    {
      "epoch": 0.084924965893588,
      "grad_norm": 0.49994444847106934,
      "learning_rate": 1.9843244520743614e-06,
      "loss": 0.5888,
      "step": 498
    },
    {
      "epoch": 0.08509549795361528,
      "grad_norm": 0.4456343650817871,
      "learning_rate": 1.984226890944098e-06,
      "loss": 0.5736,
      "step": 499
    },
    {
      "epoch": 0.08526603001364257,
      "grad_norm": 0.4084942042827606,
      "learning_rate": 1.9841290295688334e-06,
      "loss": 0.5647,
      "step": 500
    },
    {
      "epoch": 0.08543656207366986,
      "grad_norm": 0.43860793113708496,
      "learning_rate": 1.984030867978421e-06,
      "loss": 0.5679,
      "step": 501
    },
    {
      "epoch": 0.08560709413369713,
      "grad_norm": 0.4625251293182373,
      "learning_rate": 1.9839324062028053e-06,
      "loss": 0.5576,
      "step": 502
    },
    {
      "epoch": 0.08577762619372442,
      "grad_norm": 0.4842986464500427,
      "learning_rate": 1.983833644272023e-06,
      "loss": 0.5773,
      "step": 503
    },
    {
      "epoch": 0.08594815825375171,
      "grad_norm": 0.4727341830730438,
      "learning_rate": 1.9837345822162022e-06,
      "loss": 0.576,
      "step": 504
    },
    {
      "epoch": 0.08611869031377899,
      "grad_norm": 0.40692001581192017,
      "learning_rate": 1.9836352200655623e-06,
      "loss": 0.5696,
      "step": 505
    },
    {
      "epoch": 0.08628922237380628,
      "grad_norm": 0.3653680086135864,
      "learning_rate": 1.9835355578504146e-06,
      "loss": 0.5759,
      "step": 506
    },
    {
      "epoch": 0.08645975443383357,
      "grad_norm": 0.4303702414035797,
      "learning_rate": 1.983435595601161e-06,
      "loss": 0.5641,
      "step": 507
    },
    {
      "epoch": 0.08663028649386084,
      "grad_norm": 0.5330203175544739,
      "learning_rate": 1.9833353333482964e-06,
      "loss": 0.5667,
      "step": 508
    },
    {
      "epoch": 0.08680081855388813,
      "grad_norm": 0.4900553226470947,
      "learning_rate": 1.983234771122406e-06,
      "loss": 0.5819,
      "step": 509
    },
    {
      "epoch": 0.08697135061391542,
      "grad_norm": 0.3540118932723999,
      "learning_rate": 1.9831339089541672e-06,
      "loss": 0.5652,
      "step": 510
    },
    {
      "epoch": 0.0871418826739427,
      "grad_norm": 0.484788715839386,
      "learning_rate": 1.9830327468743492e-06,
      "loss": 0.572,
      "step": 511
    },
    {
      "epoch": 0.08731241473396999,
      "grad_norm": 0.38806799054145813,
      "learning_rate": 1.9829312849138114e-06,
      "loss": 0.5646,
      "step": 512
    },
    {
      "epoch": 0.08748294679399728,
      "grad_norm": 0.4288354814052582,
      "learning_rate": 1.9828295231035057e-06,
      "loss": 0.5733,
      "step": 513
    },
    {
      "epoch": 0.08765347885402455,
      "grad_norm": 0.4415360391139984,
      "learning_rate": 1.982727461474475e-06,
      "loss": 0.5645,
      "step": 514
    },
    {
      "epoch": 0.08782401091405184,
      "grad_norm": 0.4523155093193054,
      "learning_rate": 1.9826251000578543e-06,
      "loss": 0.567,
      "step": 515
    },
    {
      "epoch": 0.08799454297407913,
      "grad_norm": 0.6626890897750854,
      "learning_rate": 1.9825224388848694e-06,
      "loss": 0.5566,
      "step": 516
    },
    {
      "epoch": 0.08816507503410641,
      "grad_norm": 0.7616189122200012,
      "learning_rate": 1.9824194779868375e-06,
      "loss": 0.5727,
      "step": 517
    },
    {
      "epoch": 0.0883356070941337,
      "grad_norm": 0.7304683327674866,
      "learning_rate": 1.9823162173951683e-06,
      "loss": 0.5767,
      "step": 518
    },
    {
      "epoch": 0.08850613915416099,
      "grad_norm": 0.5385451316833496,
      "learning_rate": 1.9822126571413617e-06,
      "loss": 0.5743,
      "step": 519
    },
    {
      "epoch": 0.08867667121418826,
      "grad_norm": 0.49225184321403503,
      "learning_rate": 1.9821087972570094e-06,
      "loss": 0.5807,
      "step": 520
    },
    {
      "epoch": 0.08884720327421555,
      "grad_norm": 0.4323744773864746,
      "learning_rate": 1.9820046377737943e-06,
      "loss": 0.5709,
      "step": 521
    },
    {
      "epoch": 0.08901773533424284,
      "grad_norm": 0.6151190996170044,
      "learning_rate": 1.9819001787234915e-06,
      "loss": 0.5605,
      "step": 522
    },
    {
      "epoch": 0.08918826739427012,
      "grad_norm": 0.6538416147232056,
      "learning_rate": 1.9817954201379666e-06,
      "loss": 0.5868,
      "step": 523
    },
    {
      "epoch": 0.08935879945429741,
      "grad_norm": 0.44516924023628235,
      "learning_rate": 1.9816903620491766e-06,
      "loss": 0.5631,
      "step": 524
    },
    {
      "epoch": 0.0895293315143247,
      "grad_norm": 0.40045231580734253,
      "learning_rate": 1.981585004489171e-06,
      "loss": 0.5787,
      "step": 525
    },
    {
      "epoch": 0.08969986357435197,
      "grad_norm": 0.4614306390285492,
      "learning_rate": 1.981479347490089e-06,
      "loss": 0.5786,
      "step": 526
    },
    {
      "epoch": 0.08987039563437926,
      "grad_norm": 0.3792482912540436,
      "learning_rate": 1.981373391084163e-06,
      "loss": 0.5671,
      "step": 527
    },
    {
      "epoch": 0.09004092769440655,
      "grad_norm": 0.41413459181785583,
      "learning_rate": 1.981267135303714e-06,
      "loss": 0.5785,
      "step": 528
    },
    {
      "epoch": 0.09021145975443383,
      "grad_norm": 0.35130906105041504,
      "learning_rate": 1.981160580181158e-06,
      "loss": 0.5761,
      "step": 529
    },
    {
      "epoch": 0.09038199181446112,
      "grad_norm": 0.442913293838501,
      "learning_rate": 1.9810537257489988e-06,
      "loss": 0.5761,
      "step": 530
    },
    {
      "epoch": 0.09055252387448841,
      "grad_norm": 0.39329108595848083,
      "learning_rate": 1.980946572039834e-06,
      "loss": 0.5773,
      "step": 531
    },
    {
      "epoch": 0.09072305593451568,
      "grad_norm": 0.37966036796569824,
      "learning_rate": 1.9808391190863515e-06,
      "loss": 0.5695,
      "step": 532
    },
    {
      "epoch": 0.09089358799454297,
      "grad_norm": 0.4225713610649109,
      "learning_rate": 1.98073136692133e-06,
      "loss": 0.5786,
      "step": 533
    },
    {
      "epoch": 0.09106412005457026,
      "grad_norm": 0.4584408700466156,
      "learning_rate": 1.98062331557764e-06,
      "loss": 0.5699,
      "step": 534
    },
    {
      "epoch": 0.09123465211459754,
      "grad_norm": 0.41836604475975037,
      "learning_rate": 1.9805149650882442e-06,
      "loss": 0.5715,
      "step": 535
    },
    {
      "epoch": 0.09140518417462483,
      "grad_norm": 0.4590608477592468,
      "learning_rate": 1.980406315486195e-06,
      "loss": 0.5785,
      "step": 536
    },
    {
      "epoch": 0.09157571623465212,
      "grad_norm": 0.49678942561149597,
      "learning_rate": 1.9802973668046367e-06,
      "loss": 0.561,
      "step": 537
    },
    {
      "epoch": 0.0917462482946794,
      "grad_norm": 0.3588040769100189,
      "learning_rate": 1.980188119076804e-06,
      "loss": 0.5658,
      "step": 538
    },
    {
      "epoch": 0.09191678035470668,
      "grad_norm": 0.4952399432659149,
      "learning_rate": 1.9800785723360257e-06,
      "loss": 0.5821,
      "step": 539
    },
    {
      "epoch": 0.09208731241473397,
      "grad_norm": 0.4507637023925781,
      "learning_rate": 1.979968726615718e-06,
      "loss": 0.5812,
      "step": 540
    },
    {
      "epoch": 0.09225784447476125,
      "grad_norm": 0.388642281293869,
      "learning_rate": 1.979858581949391e-06,
      "loss": 0.5736,
      "step": 541
    },
    {
      "epoch": 0.09242837653478854,
      "grad_norm": 0.47354656457901,
      "learning_rate": 1.979748138370644e-06,
      "loss": 0.5744,
      "step": 542
    },
    {
      "epoch": 0.09259890859481583,
      "grad_norm": 0.350930780172348,
      "learning_rate": 1.97963739591317e-06,
      "loss": 0.571,
      "step": 543
    },
    {
      "epoch": 0.0927694406548431,
      "grad_norm": 0.428281307220459,
      "learning_rate": 1.979526354610751e-06,
      "loss": 0.5591,
      "step": 544
    },
    {
      "epoch": 0.0929399727148704,
      "grad_norm": 0.4820069670677185,
      "learning_rate": 1.97941501449726e-06,
      "loss": 0.5862,
      "step": 545
    },
    {
      "epoch": 0.09311050477489768,
      "grad_norm": 0.5184879302978516,
      "learning_rate": 1.9793033756066635e-06,
      "loss": 0.5524,
      "step": 546
    },
    {
      "epoch": 0.09328103683492496,
      "grad_norm": 0.40888434648513794,
      "learning_rate": 1.9791914379730175e-06,
      "loss": 0.56,
      "step": 547
    },
    {
      "epoch": 0.09345156889495225,
      "grad_norm": 0.35360148549079895,
      "learning_rate": 1.9790792016304687e-06,
      "loss": 0.577,
      "step": 548
    },
    {
      "epoch": 0.09362210095497954,
      "grad_norm": 0.4426764249801636,
      "learning_rate": 1.9789666666132557e-06,
      "loss": 0.5913,
      "step": 549
    },
    {
      "epoch": 0.09379263301500682,
      "grad_norm": 0.38881126046180725,
      "learning_rate": 1.978853832955708e-06,
      "loss": 0.576,
      "step": 550
    },
    {
      "epoch": 0.0939631650750341,
      "grad_norm": 0.40196335315704346,
      "learning_rate": 1.9787407006922466e-06,
      "loss": 0.5899,
      "step": 551
    },
    {
      "epoch": 0.0941336971350614,
      "grad_norm": 0.37533479928970337,
      "learning_rate": 1.978627269857383e-06,
      "loss": 0.5804,
      "step": 552
    },
    {
      "epoch": 0.09430422919508867,
      "grad_norm": 0.34749701619148254,
      "learning_rate": 1.9785135404857202e-06,
      "loss": 0.5586,
      "step": 553
    },
    {
      "epoch": 0.09447476125511596,
      "grad_norm": 0.40509840846061707,
      "learning_rate": 1.978399512611952e-06,
      "loss": 0.5604,
      "step": 554
    },
    {
      "epoch": 0.09464529331514325,
      "grad_norm": 0.43824177980422974,
      "learning_rate": 1.9782851862708634e-06,
      "loss": 0.5616,
      "step": 555
    },
    {
      "epoch": 0.09481582537517053,
      "grad_norm": 0.5045179724693298,
      "learning_rate": 1.97817056149733e-06,
      "loss": 0.5608,
      "step": 556
    },
    {
      "epoch": 0.09498635743519782,
      "grad_norm": 0.4929228723049164,
      "learning_rate": 1.97805563832632e-06,
      "loss": 0.5768,
      "step": 557
    },
    {
      "epoch": 0.0951568894952251,
      "grad_norm": 0.3765542507171631,
      "learning_rate": 1.9779404167928904e-06,
      "loss": 0.5512,
      "step": 558
    },
    {
      "epoch": 0.09532742155525238,
      "grad_norm": 0.3653373420238495,
      "learning_rate": 1.9778248969321907e-06,
      "loss": 0.5653,
      "step": 559
    },
    {
      "epoch": 0.09549795361527967,
      "grad_norm": 0.4602164328098297,
      "learning_rate": 1.9777090787794607e-06,
      "loss": 0.5974,
      "step": 560
    },
    {
      "epoch": 0.09566848567530696,
      "grad_norm": 0.5155764818191528,
      "learning_rate": 1.977592962370032e-06,
      "loss": 0.5544,
      "step": 561
    },
    {
      "epoch": 0.09583901773533424,
      "grad_norm": 0.6241593360900879,
      "learning_rate": 1.9774765477393262e-06,
      "loss": 0.5837,
      "step": 562
    },
    {
      "epoch": 0.09600954979536153,
      "grad_norm": 0.6194249391555786,
      "learning_rate": 1.977359834922857e-06,
      "loss": 0.5662,
      "step": 563
    },
    {
      "epoch": 0.09618008185538882,
      "grad_norm": 0.5258529782295227,
      "learning_rate": 1.9772428239562277e-06,
      "loss": 0.5525,
      "step": 564
    },
    {
      "epoch": 0.09635061391541609,
      "grad_norm": 0.4712018370628357,
      "learning_rate": 1.9771255148751334e-06,
      "loss": 0.5716,
      "step": 565
    },
    {
      "epoch": 0.09652114597544338,
      "grad_norm": 0.38999003171920776,
      "learning_rate": 1.9770079077153607e-06,
      "loss": 0.5593,
      "step": 566
    },
    {
      "epoch": 0.09669167803547067,
      "grad_norm": 0.5134232640266418,
      "learning_rate": 1.9768900025127853e-06,
      "loss": 0.5628,
      "step": 567
    },
    {
      "epoch": 0.09686221009549795,
      "grad_norm": 0.4734049439430237,
      "learning_rate": 1.976771799303376e-06,
      "loss": 0.5643,
      "step": 568
    },
    {
      "epoch": 0.09703274215552524,
      "grad_norm": 0.42365288734436035,
      "learning_rate": 1.9766532981231915e-06,
      "loss": 0.568,
      "step": 569
    },
    {
      "epoch": 0.09720327421555253,
      "grad_norm": 0.7270985841751099,
      "learning_rate": 1.976534499008381e-06,
      "loss": 0.5636,
      "step": 570
    },
    {
      "epoch": 0.0973738062755798,
      "grad_norm": 0.8700385093688965,
      "learning_rate": 1.976415401995184e-06,
      "loss": 0.5673,
      "step": 571
    },
    {
      "epoch": 0.09754433833560709,
      "grad_norm": 0.6275728940963745,
      "learning_rate": 1.9762960071199336e-06,
      "loss": 0.5772,
      "step": 572
    },
    {
      "epoch": 0.09771487039563438,
      "grad_norm": 0.3480617105960846,
      "learning_rate": 1.9761763144190514e-06,
      "loss": 0.5675,
      "step": 573
    },
    {
      "epoch": 0.09788540245566166,
      "grad_norm": 0.6419890522956848,
      "learning_rate": 1.9760563239290495e-06,
      "loss": 0.5673,
      "step": 574
    },
    {
      "epoch": 0.09805593451568895,
      "grad_norm": 0.5626255869865417,
      "learning_rate": 1.9759360356865333e-06,
      "loss": 0.5794,
      "step": 575
    },
    {
      "epoch": 0.09822646657571624,
      "grad_norm": 0.3339516520500183,
      "learning_rate": 1.9758154497281966e-06,
      "loss": 0.5617,
      "step": 576
    },
    {
      "epoch": 0.09839699863574353,
      "grad_norm": 0.5528090000152588,
      "learning_rate": 1.975694566090825e-06,
      "loss": 0.5675,
      "step": 577
    },
    {
      "epoch": 0.0985675306957708,
      "grad_norm": 0.4296857714653015,
      "learning_rate": 1.975573384811295e-06,
      "loss": 0.5794,
      "step": 578
    },
    {
      "epoch": 0.09873806275579809,
      "grad_norm": 0.43378129601478577,
      "learning_rate": 1.975451905926574e-06,
      "loss": 0.5643,
      "step": 579
    },
    {
      "epoch": 0.09890859481582538,
      "grad_norm": 0.6495099663734436,
      "learning_rate": 1.9753301294737196e-06,
      "loss": 0.5582,
      "step": 580
    },
    {
      "epoch": 0.09907912687585266,
      "grad_norm": 0.42808154225349426,
      "learning_rate": 1.975208055489881e-06,
      "loss": 0.5682,
      "step": 581
    },
    {
      "epoch": 0.09924965893587995,
      "grad_norm": 0.4758428633213043,
      "learning_rate": 1.975085684012297e-06,
      "loss": 0.5626,
      "step": 582
    },
    {
      "epoch": 0.09942019099590724,
      "grad_norm": 0.6097862720489502,
      "learning_rate": 1.974963015078298e-06,
      "loss": 0.5737,
      "step": 583
    },
    {
      "epoch": 0.09959072305593451,
      "grad_norm": 0.3477315306663513,
      "learning_rate": 1.9748400487253056e-06,
      "loss": 0.5623,
      "step": 584
    },
    {
      "epoch": 0.0997612551159618,
      "grad_norm": 0.6643428206443787,
      "learning_rate": 1.9747167849908306e-06,
      "loss": 0.5662,
      "step": 585
    },
    {
      "epoch": 0.09993178717598909,
      "grad_norm": 0.5639585256576538,
      "learning_rate": 1.974593223912476e-06,
      "loss": 0.5592,
      "step": 586
    },
    {
      "epoch": 0.10010231923601637,
      "grad_norm": 0.3957652151584625,
      "learning_rate": 1.974469365527935e-06,
      "loss": 0.5758,
      "step": 587
    },
    {
      "epoch": 0.10027285129604366,
      "grad_norm": 0.7016974091529846,
      "learning_rate": 1.9743452098749913e-06,
      "loss": 0.5672,
      "step": 588
    },
    {
      "epoch": 0.10044338335607095,
      "grad_norm": 0.5436629056930542,
      "learning_rate": 1.9742207569915194e-06,
      "loss": 0.5701,
      "step": 589
    },
    {
      "epoch": 0.10061391541609822,
      "grad_norm": 0.4033794105052948,
      "learning_rate": 1.9740960069154842e-06,
      "loss": 0.5663,
      "step": 590
    },
    {
      "epoch": 0.10078444747612551,
      "grad_norm": 0.5005258321762085,
      "learning_rate": 1.973970959684942e-06,
      "loss": 0.5587,
      "step": 591
    },
    {
      "epoch": 0.1009549795361528,
      "grad_norm": 0.5139448642730713,
      "learning_rate": 1.973845615338039e-06,
      "loss": 0.5812,
      "step": 592
    },
    {
      "epoch": 0.10112551159618008,
      "grad_norm": 0.3810591995716095,
      "learning_rate": 1.9737199739130123e-06,
      "loss": 0.5699,
      "step": 593
    },
    {
      "epoch": 0.10129604365620737,
      "grad_norm": 0.4457865357398987,
      "learning_rate": 1.9735940354481902e-06,
      "loss": 0.5712,
      "step": 594
    },
    {
      "epoch": 0.10146657571623466,
      "grad_norm": 0.4659377336502075,
      "learning_rate": 1.9734677999819903e-06,
      "loss": 0.5545,
      "step": 595
    },
    {
      "epoch": 0.10163710777626193,
      "grad_norm": 0.3576979339122772,
      "learning_rate": 1.973341267552922e-06,
      "loss": 0.5772,
      "step": 596
    },
    {
      "epoch": 0.10180763983628922,
      "grad_norm": 0.4095207154750824,
      "learning_rate": 1.9732144381995848e-06,
      "loss": 0.5595,
      "step": 597
    },
    {
      "epoch": 0.10197817189631651,
      "grad_norm": 0.42436665296554565,
      "learning_rate": 1.973087311960669e-06,
      "loss": 0.5691,
      "step": 598
    },
    {
      "epoch": 0.10214870395634379,
      "grad_norm": 0.3475062847137451,
      "learning_rate": 1.972959888874955e-06,
      "loss": 0.5619,
      "step": 599
    },
    {
      "epoch": 0.10231923601637108,
      "grad_norm": 0.48956984281539917,
      "learning_rate": 1.972832168981314e-06,
      "loss": 0.5551,
      "step": 600
    },
    {
      "epoch": 0.10248976807639837,
      "grad_norm": 0.452088862657547,
      "learning_rate": 1.9727041523187084e-06,
      "loss": 0.5848,
      "step": 601
    },
    {
      "epoch": 0.10266030013642564,
      "grad_norm": 0.36038264632225037,
      "learning_rate": 1.97257583892619e-06,
      "loss": 0.5731,
      "step": 602
    },
    {
      "epoch": 0.10283083219645293,
      "grad_norm": 0.36114734411239624,
      "learning_rate": 1.972447228842902e-06,
      "loss": 0.5808,
      "step": 603
    },
    {
      "epoch": 0.10300136425648022,
      "grad_norm": 0.38365817070007324,
      "learning_rate": 1.9723183221080775e-06,
      "loss": 0.5543,
      "step": 604
    },
    {
      "epoch": 0.1031718963165075,
      "grad_norm": 0.34079596400260925,
      "learning_rate": 1.97218911876104e-06,
      "loss": 0.5809,
      "step": 605
    },
    {
      "epoch": 0.10334242837653479,
      "grad_norm": 0.36070719361305237,
      "learning_rate": 1.972059618841205e-06,
      "loss": 0.5574,
      "step": 606
    },
    {
      "epoch": 0.10351296043656208,
      "grad_norm": 0.3969060480594635,
      "learning_rate": 1.9719298223880763e-06,
      "loss": 0.5544,
      "step": 607
    },
    {
      "epoch": 0.10368349249658936,
      "grad_norm": 0.38917815685272217,
      "learning_rate": 1.971799729441249e-06,
      "loss": 0.5511,
      "step": 608
    },
    {
      "epoch": 0.10385402455661664,
      "grad_norm": 0.4314303696155548,
      "learning_rate": 1.97166934004041e-06,
      "loss": 0.5594,
      "step": 609
    },
    {
      "epoch": 0.10402455661664393,
      "grad_norm": 0.3728209137916565,
      "learning_rate": 1.971538654225335e-06,
      "loss": 0.5679,
      "step": 610
    },
    {
      "epoch": 0.10419508867667121,
      "grad_norm": 0.3441550135612488,
      "learning_rate": 1.97140767203589e-06,
      "loss": 0.5536,
      "step": 611
    },
    {
      "epoch": 0.1043656207366985,
      "grad_norm": 0.39468538761138916,
      "learning_rate": 1.971276393512032e-06,
      "loss": 0.5636,
      "step": 612
    },
    {
      "epoch": 0.10453615279672579,
      "grad_norm": 0.37954941391944885,
      "learning_rate": 1.971144818693809e-06,
      "loss": 0.5664,
      "step": 613
    },
    {
      "epoch": 0.10470668485675307,
      "grad_norm": 0.39011162519454956,
      "learning_rate": 1.9710129476213587e-06,
      "loss": 0.5618,
      "step": 614
    },
    {
      "epoch": 0.10487721691678036,
      "grad_norm": 0.3826182782649994,
      "learning_rate": 1.970880780334909e-06,
      "loss": 0.5677,
      "step": 615
    },
    {
      "epoch": 0.10504774897680765,
      "grad_norm": 0.48914775252342224,
      "learning_rate": 1.9707483168747783e-06,
      "loss": 0.5678,
      "step": 616
    },
    {
      "epoch": 0.10521828103683492,
      "grad_norm": 0.4400955140590668,
      "learning_rate": 1.970615557281376e-06,
      "loss": 0.5677,
      "step": 617
    },
    {
      "epoch": 0.10538881309686221,
      "grad_norm": 0.4150509238243103,
      "learning_rate": 1.9704825015952007e-06,
      "loss": 0.562,
      "step": 618
    },
    {
      "epoch": 0.1055593451568895,
      "grad_norm": 0.50996994972229,
      "learning_rate": 1.9703491498568423e-06,
      "loss": 0.5587,
      "step": 619
    },
    {
      "epoch": 0.10572987721691678,
      "grad_norm": 0.4692332148551941,
      "learning_rate": 1.97021550210698e-06,
      "loss": 0.5753,
      "step": 620
    },
    {
      "epoch": 0.10590040927694407,
      "grad_norm": 0.4658254384994507,
      "learning_rate": 1.9700815583863853e-06,
      "loss": 0.5565,
      "step": 621
    },
    {
      "epoch": 0.10607094133697136,
      "grad_norm": 0.4767058789730072,
      "learning_rate": 1.9699473187359174e-06,
      "loss": 0.57,
      "step": 622
    },
    {
      "epoch": 0.10624147339699863,
      "grad_norm": 0.34791260957717896,
      "learning_rate": 1.969812783196528e-06,
      "loss": 0.5713,
      "step": 623
    },
    {
      "epoch": 0.10641200545702592,
      "grad_norm": 0.43950381875038147,
      "learning_rate": 1.9696779518092567e-06,
      "loss": 0.5585,
      "step": 624
    },
    {
      "epoch": 0.10658253751705321,
      "grad_norm": 0.5135130882263184,
      "learning_rate": 1.969542824615235e-06,
      "loss": 0.5726,
      "step": 625
    },
    {
      "epoch": 0.10675306957708049,
      "grad_norm": 0.4924056828022003,
      "learning_rate": 1.969407401655686e-06,
      "loss": 0.5541,
      "step": 626
    },
    {
      "epoch": 0.10692360163710778,
      "grad_norm": 0.4262337386608124,
      "learning_rate": 1.96927168297192e-06,
      "loss": 0.6055,
      "step": 627
    },
    {
      "epoch": 0.10709413369713507,
      "grad_norm": 0.4578431248664856,
      "learning_rate": 1.9691356686053386e-06,
      "loss": 0.5653,
      "step": 628
    },
    {
      "epoch": 0.10726466575716234,
      "grad_norm": 0.6619389653205872,
      "learning_rate": 1.9689993585974345e-06,
      "loss": 0.5747,
      "step": 629
    },
    {
      "epoch": 0.10743519781718963,
      "grad_norm": 0.5923303365707397,
      "learning_rate": 1.96886275298979e-06,
      "loss": 0.5562,
      "step": 630
    },
    {
      "epoch": 0.10760572987721692,
      "grad_norm": 0.44426047801971436,
      "learning_rate": 1.9687258518240777e-06,
      "loss": 0.5355,
      "step": 631
    },
    {
      "epoch": 0.1077762619372442,
      "grad_norm": 0.5126156210899353,
      "learning_rate": 1.9685886551420594e-06,
      "loss": 0.5499,
      "step": 632
    },
    {
      "epoch": 0.10794679399727149,
      "grad_norm": 0.6300952434539795,
      "learning_rate": 1.9684511629855893e-06,
      "loss": 0.5624,
      "step": 633
    },
    {
      "epoch": 0.10811732605729878,
      "grad_norm": 0.5002975463867188,
      "learning_rate": 1.9683133753966086e-06,
      "loss": 0.5744,
      "step": 634
    },
    {
      "epoch": 0.10828785811732605,
      "grad_norm": 0.39933618903160095,
      "learning_rate": 1.9681752924171517e-06,
      "loss": 0.5805,
      "step": 635
    },
    {
      "epoch": 0.10845839017735334,
      "grad_norm": 0.6121488809585571,
      "learning_rate": 1.9680369140893407e-06,
      "loss": 0.5578,
      "step": 636
    },
    {
      "epoch": 0.10862892223738063,
      "grad_norm": 0.6016712188720703,
      "learning_rate": 1.9678982404553897e-06,
      "loss": 0.5535,
      "step": 637
    },
    {
      "epoch": 0.10879945429740791,
      "grad_norm": 0.4365065395832062,
      "learning_rate": 1.9677592715576016e-06,
      "loss": 0.5817,
      "step": 638
    },
    {
      "epoch": 0.1089699863574352,
      "grad_norm": 0.5382389426231384,
      "learning_rate": 1.9676200074383695e-06,
      "loss": 0.5595,
      "step": 639
    },
    {
      "epoch": 0.10914051841746249,
      "grad_norm": 0.552986204624176,
      "learning_rate": 1.9674804481401776e-06,
      "loss": 0.5728,
      "step": 640
    },
    {
      "epoch": 0.10931105047748976,
      "grad_norm": 0.36936983466148376,
      "learning_rate": 1.9673405937055987e-06,
      "loss": 0.5591,
      "step": 641
    },
    {
      "epoch": 0.10948158253751705,
      "grad_norm": 0.4702188968658447,
      "learning_rate": 1.9672004441772966e-06,
      "loss": 0.5633,
      "step": 642
    },
    {
      "epoch": 0.10965211459754434,
      "grad_norm": 0.5598739981651306,
      "learning_rate": 1.967059999598025e-06,
      "loss": 0.5663,
      "step": 643
    },
    {
      "epoch": 0.10982264665757162,
      "grad_norm": 0.4790910482406616,
      "learning_rate": 1.9669192600106275e-06,
      "loss": 0.5574,
      "step": 644
    },
    {
      "epoch": 0.10999317871759891,
      "grad_norm": 0.35663726925849915,
      "learning_rate": 1.9667782254580378e-06,
      "loss": 0.5501,
      "step": 645
    },
    {
      "epoch": 0.1101637107776262,
      "grad_norm": 0.43719905614852905,
      "learning_rate": 1.9666368959832782e-06,
      "loss": 0.5702,
      "step": 646
    },
    {
      "epoch": 0.11033424283765347,
      "grad_norm": 0.5694594383239746,
      "learning_rate": 1.9664952716294637e-06,
      "loss": 0.5461,
      "step": 647
    },
    {
      "epoch": 0.11050477489768076,
      "grad_norm": 0.5919437408447266,
      "learning_rate": 1.966353352439797e-06,
      "loss": 0.5492,
      "step": 648
    },
    {
      "epoch": 0.11067530695770805,
      "grad_norm": 0.5850440263748169,
      "learning_rate": 1.9662111384575717e-06,
      "loss": 0.543,
      "step": 649
    },
    {
      "epoch": 0.11084583901773533,
      "grad_norm": 0.41119086742401123,
      "learning_rate": 1.9660686297261712e-06,
      "loss": 0.5627,
      "step": 650
    },
    {
      "epoch": 0.11101637107776262,
      "grad_norm": 0.3939213156700134,
      "learning_rate": 1.9659258262890686e-06,
      "loss": 0.552,
      "step": 651
    },
    {
      "epoch": 0.11118690313778991,
      "grad_norm": 0.5054365396499634,
      "learning_rate": 1.965782728189827e-06,
      "loss": 0.564,
      "step": 652
    },
    {
      "epoch": 0.11135743519781718,
      "grad_norm": 0.5284506678581238,
      "learning_rate": 1.9656393354720997e-06,
      "loss": 0.5487,
      "step": 653
    },
    {
      "epoch": 0.11152796725784447,
      "grad_norm": 0.3425239324569702,
      "learning_rate": 1.9654956481796294e-06,
      "loss": 0.5675,
      "step": 654
    },
    {
      "epoch": 0.11169849931787176,
      "grad_norm": 0.46462586522102356,
      "learning_rate": 1.965351666356249e-06,
      "loss": 0.5544,
      "step": 655
    },
    {
      "epoch": 0.11186903137789904,
      "grad_norm": 0.4571056067943573,
      "learning_rate": 1.9652073900458806e-06,
      "loss": 0.5683,
      "step": 656
    },
    {
      "epoch": 0.11203956343792633,
      "grad_norm": 0.39550215005874634,
      "learning_rate": 1.965062819292537e-06,
      "loss": 0.5567,
      "step": 657
    },
    {
      "epoch": 0.11221009549795362,
      "grad_norm": 0.407055139541626,
      "learning_rate": 1.9649179541403213e-06,
      "loss": 0.5532,
      "step": 658
    },
    {
      "epoch": 0.1123806275579809,
      "grad_norm": 0.3835711181163788,
      "learning_rate": 1.9647727946334244e-06,
      "loss": 0.5417,
      "step": 659
    },
    {
      "epoch": 0.11255115961800818,
      "grad_norm": 0.38728925585746765,
      "learning_rate": 1.964627340816129e-06,
      "loss": 0.5688,
      "step": 660
    },
    {
      "epoch": 0.11272169167803547,
      "grad_norm": 0.33165115118026733,
      "learning_rate": 1.9644815927328056e-06,
      "loss": 0.5679,
      "step": 661
    },
    {
      "epoch": 0.11289222373806275,
      "grad_norm": 0.42033204436302185,
      "learning_rate": 1.9643355504279168e-06,
      "loss": 0.5632,
      "step": 662
    },
    {
      "epoch": 0.11306275579809004,
      "grad_norm": 0.3787094056606293,
      "learning_rate": 1.9641892139460133e-06,
      "loss": 0.5561,
      "step": 663
    },
    {
      "epoch": 0.11323328785811733,
      "grad_norm": 0.4706180989742279,
      "learning_rate": 1.9640425833317364e-06,
      "loss": 0.5674,
      "step": 664
    },
    {
      "epoch": 0.1134038199181446,
      "grad_norm": 0.4153285622596741,
      "learning_rate": 1.963895658629816e-06,
      "loss": 0.5623,
      "step": 665
    },
    {
      "epoch": 0.1135743519781719,
      "grad_norm": 0.4143642783164978,
      "learning_rate": 1.9637484398850736e-06,
      "loss": 0.5578,
      "step": 666
    },
    {
      "epoch": 0.11374488403819918,
      "grad_norm": 0.4237208962440491,
      "learning_rate": 1.963600927142418e-06,
      "loss": 0.5547,
      "step": 667
    },
    {
      "epoch": 0.11391541609822646,
      "grad_norm": 0.4191353917121887,
      "learning_rate": 1.96345312044685e-06,
      "loss": 0.5664,
      "step": 668
    },
    {
      "epoch": 0.11408594815825375,
      "grad_norm": 0.4121340811252594,
      "learning_rate": 1.963305019843458e-06,
      "loss": 0.5471,
      "step": 669
    },
    {
      "epoch": 0.11425648021828104,
      "grad_norm": 0.48723238706588745,
      "learning_rate": 1.963156625377422e-06,
      "loss": 0.5619,
      "step": 670
    },
    {
      "epoch": 0.11442701227830832,
      "grad_norm": 0.4804185628890991,
      "learning_rate": 1.96300793709401e-06,
      "loss": 0.563,
      "step": 671
    },
    {
      "epoch": 0.1145975443383356,
      "grad_norm": 0.413357138633728,
      "learning_rate": 1.9628589550385812e-06,
      "loss": 0.5604,
      "step": 672
    },
    {
      "epoch": 0.1147680763983629,
      "grad_norm": 0.7102183699607849,
      "learning_rate": 1.9627096792565827e-06,
      "loss": 0.5671,
      "step": 673
    },
    {
      "epoch": 0.11493860845839017,
      "grad_norm": 0.7849295139312744,
      "learning_rate": 1.9625601097935527e-06,
      "loss": 0.5438,
      "step": 674
    },
    {
      "epoch": 0.11510914051841746,
      "grad_norm": 0.49737444519996643,
      "learning_rate": 1.962410246695118e-06,
      "loss": 0.5545,
      "step": 675
    },
    {
      "epoch": 0.11527967257844475,
      "grad_norm": 0.46542850136756897,
      "learning_rate": 1.9622600900069958e-06,
      "loss": 0.5758,
      "step": 676
    },
    {
      "epoch": 0.11545020463847203,
      "grad_norm": 0.6955744624137878,
      "learning_rate": 1.9621096397749917e-06,
      "loss": 0.5381,
      "step": 677
    },
    {
      "epoch": 0.11562073669849932,
      "grad_norm": 0.5565853714942932,
      "learning_rate": 1.961958896045002e-06,
      "loss": 0.5796,
      "step": 678
    },
    {
      "epoch": 0.1157912687585266,
      "grad_norm": 0.39091721177101135,
      "learning_rate": 1.9618078588630126e-06,
      "loss": 0.5541,
      "step": 679
    },
    {
      "epoch": 0.11596180081855388,
      "grad_norm": 0.675932765007019,
      "learning_rate": 1.961656528275097e-06,
      "loss": 0.5747,
      "step": 680
    },
    {
      "epoch": 0.11613233287858117,
      "grad_norm": 0.7067851424217224,
      "learning_rate": 1.9615049043274208e-06,
      "loss": 0.5641,
      "step": 681
    },
    {
      "epoch": 0.11630286493860846,
      "grad_norm": 0.39402681589126587,
      "learning_rate": 1.9613529870662373e-06,
      "loss": 0.569,
      "step": 682
    },
    {
      "epoch": 0.11647339699863575,
      "grad_norm": 0.6251634359359741,
      "learning_rate": 1.9612007765378907e-06,
      "loss": 0.5594,
      "step": 683
    },
    {
      "epoch": 0.11664392905866303,
      "grad_norm": 0.6259270906448364,
      "learning_rate": 1.9610482727888127e-06,
      "loss": 0.5607,
      "step": 684
    },
    {
      "epoch": 0.11681446111869032,
      "grad_norm": 0.3737480342388153,
      "learning_rate": 1.960895475865526e-06,
      "loss": 0.5602,
      "step": 685
    },
    {
      "epoch": 0.1169849931787176,
      "grad_norm": 0.5192775130271912,
      "learning_rate": 1.960742385814643e-06,
      "loss": 0.5535,
      "step": 686
    },
    {
      "epoch": 0.11715552523874488,
      "grad_norm": 0.4264221489429474,
      "learning_rate": 1.9605890026828636e-06,
      "loss": 0.5531,
      "step": 687
    },
    {
      "epoch": 0.11732605729877217,
      "grad_norm": 0.46946707367897034,
      "learning_rate": 1.9604353265169797e-06,
      "loss": 0.56,
      "step": 688
    },
    {
      "epoch": 0.11749658935879946,
      "grad_norm": 0.46235525608062744,
      "learning_rate": 1.9602813573638707e-06,
      "loss": 0.5665,
      "step": 689
    },
    {
      "epoch": 0.11766712141882674,
      "grad_norm": 0.39178624749183655,
      "learning_rate": 1.9601270952705053e-06,
      "loss": 0.568,
      "step": 690
    },
    {
      "epoch": 0.11783765347885403,
      "grad_norm": 0.43903523683547974,
      "learning_rate": 1.959972540283943e-06,
      "loss": 0.5393,
      "step": 691
    },
    {
      "epoch": 0.11800818553888132,
      "grad_norm": 0.48046907782554626,
      "learning_rate": 1.959817692451331e-06,
      "loss": 0.5628,
      "step": 692
    },
    {
      "epoch": 0.11817871759890859,
      "grad_norm": 0.4377974569797516,
      "learning_rate": 1.959662551819908e-06,
      "loss": 0.56,
      "step": 693
    },
    {
      "epoch": 0.11834924965893588,
      "grad_norm": 0.37869778275489807,
      "learning_rate": 1.9595071184369994e-06,
      "loss": 0.5617,
      "step": 694
    },
    {
      "epoch": 0.11851978171896317,
      "grad_norm": 0.5731699466705322,
      "learning_rate": 1.959351392350022e-06,
      "loss": 0.573,
      "step": 695
    },
    {
      "epoch": 0.11869031377899045,
      "grad_norm": 0.3743929862976074,
      "learning_rate": 1.9591953736064805e-06,
      "loss": 0.5589,
      "step": 696
    },
    {
      "epoch": 0.11886084583901774,
      "grad_norm": 0.4147673547267914,
      "learning_rate": 1.9590390622539695e-06,
      "loss": 0.5485,
      "step": 697
    },
    {
      "epoch": 0.11903137789904503,
      "grad_norm": 0.4705844521522522,
      "learning_rate": 1.9588824583401736e-06,
      "loss": 0.5599,
      "step": 698
    },
    {
      "epoch": 0.1192019099590723,
      "grad_norm": 0.39228910207748413,
      "learning_rate": 1.958725561912865e-06,
      "loss": 0.5785,
      "step": 699
    },
    {
      "epoch": 0.11937244201909959,
      "grad_norm": 0.39754730463027954,
      "learning_rate": 1.9585683730199067e-06,
      "loss": 0.557,
      "step": 700
    },
    {
      "epoch": 0.11954297407912688,
      "grad_norm": 0.46609994769096375,
      "learning_rate": 1.9584108917092495e-06,
      "loss": 0.56,
      "step": 701
    },
    {
      "epoch": 0.11971350613915416,
      "grad_norm": 0.3678927421569824,
      "learning_rate": 1.9582531180289346e-06,
      "loss": 0.5619,
      "step": 702
    },
    {
      "epoch": 0.11988403819918145,
      "grad_norm": 0.42708367109298706,
      "learning_rate": 1.9580950520270917e-06,
      "loss": 0.5481,
      "step": 703
    },
    {
      "epoch": 0.12005457025920874,
      "grad_norm": 0.5537395477294922,
      "learning_rate": 1.9579366937519403e-06,
      "loss": 0.5645,
      "step": 704
    },
    {
      "epoch": 0.12022510231923601,
      "grad_norm": 0.47820043563842773,
      "learning_rate": 1.957778043251788e-06,
      "loss": 0.5698,
      "step": 705
    },
    {
      "epoch": 0.1203956343792633,
      "grad_norm": 0.43340158462524414,
      "learning_rate": 1.9576191005750324e-06,
      "loss": 0.5571,
      "step": 706
    },
    {
      "epoch": 0.12056616643929059,
      "grad_norm": 0.48589396476745605,
      "learning_rate": 1.9574598657701608e-06,
      "loss": 0.5693,
      "step": 707
    },
    {
      "epoch": 0.12073669849931787,
      "grad_norm": 0.3963133990764618,
      "learning_rate": 1.957300338885748e-06,
      "loss": 0.5527,
      "step": 708
    },
    {
      "epoch": 0.12090723055934516,
      "grad_norm": 0.5045079588890076,
      "learning_rate": 1.9571405199704583e-06,
      "loss": 0.5576,
      "step": 709
    },
    {
      "epoch": 0.12107776261937245,
      "grad_norm": 0.4563431441783905,
      "learning_rate": 1.956980409073047e-06,
      "loss": 0.55,
      "step": 710
    },
    {
      "epoch": 0.12124829467939972,
      "grad_norm": 0.44609710574150085,
      "learning_rate": 1.956820006242356e-06,
      "loss": 0.5525,
      "step": 711
    },
    {
      "epoch": 0.12141882673942701,
      "grad_norm": 0.4390665888786316,
      "learning_rate": 1.956659311527317e-06,
      "loss": 0.559,
      "step": 712
    },
    {
      "epoch": 0.1215893587994543,
      "grad_norm": 0.3507465422153473,
      "learning_rate": 1.956498324976952e-06,
      "loss": 0.5407,
      "step": 713
    },
    {
      "epoch": 0.12175989085948158,
      "grad_norm": 0.40906068682670593,
      "learning_rate": 1.95633704664037e-06,
      "loss": 0.5526,
      "step": 714
    },
    {
      "epoch": 0.12193042291950887,
      "grad_norm": 0.3908653259277344,
      "learning_rate": 1.956175476566771e-06,
      "loss": 0.5483,
      "step": 715
    },
    {
      "epoch": 0.12210095497953616,
      "grad_norm": 0.3964634835720062,
      "learning_rate": 1.9560136148054424e-06,
      "loss": 0.555,
      "step": 716
    },
    {
      "epoch": 0.12227148703956343,
      "grad_norm": 0.4320867359638214,
      "learning_rate": 1.955851461405761e-06,
      "loss": 0.5557,
      "step": 717
    },
    {
      "epoch": 0.12244201909959072,
      "grad_norm": 0.4480936825275421,
      "learning_rate": 1.9556890164171935e-06,
      "loss": 0.5432,
      "step": 718
    },
    {
      "epoch": 0.12261255115961801,
      "grad_norm": 0.3845686912536621,
      "learning_rate": 1.9555262798892943e-06,
      "loss": 0.5479,
      "step": 719
    },
    {
      "epoch": 0.12278308321964529,
      "grad_norm": 0.5671671032905579,
      "learning_rate": 1.9553632518717076e-06,
      "loss": 0.5558,
      "step": 720
    },
    {
      "epoch": 0.12295361527967258,
      "grad_norm": 0.7879170179367065,
      "learning_rate": 1.9551999324141658e-06,
      "loss": 0.5562,
      "step": 721
    },
    {
      "epoch": 0.12312414733969987,
      "grad_norm": 0.7197359204292297,
      "learning_rate": 1.955036321566491e-06,
      "loss": 0.5574,
      "step": 722
    },
    {
      "epoch": 0.12329467939972714,
      "grad_norm": 0.3807520270347595,
      "learning_rate": 1.9548724193785934e-06,
      "loss": 0.5548,
      "step": 723
    },
    {
      "epoch": 0.12346521145975443,
      "grad_norm": 0.5508958697319031,
      "learning_rate": 1.954708225900473e-06,
      "loss": 0.5645,
      "step": 724
    },
    {
      "epoch": 0.12363574351978172,
      "grad_norm": 0.7994507551193237,
      "learning_rate": 1.9545437411822174e-06,
      "loss": 0.5529,
      "step": 725
    },
    {
      "epoch": 0.123806275579809,
      "grad_norm": 0.6162607669830322,
      "learning_rate": 1.9543789652740043e-06,
      "loss": 0.5558,
      "step": 726
    },
    {
      "epoch": 0.12397680763983629,
      "grad_norm": 0.3924804925918579,
      "learning_rate": 1.9542138982260994e-06,
      "loss": 0.5507,
      "step": 727
    },
    {
      "epoch": 0.12414733969986358,
      "grad_norm": 0.712841272354126,
      "learning_rate": 1.9540485400888577e-06,
      "loss": 0.5667,
      "step": 728
    },
    {
      "epoch": 0.12431787175989086,
      "grad_norm": 0.5594262480735779,
      "learning_rate": 1.9538828909127233e-06,
      "loss": 0.5487,
      "step": 729
    },
    {
      "epoch": 0.12448840381991814,
      "grad_norm": 0.4557380974292755,
      "learning_rate": 1.9537169507482273e-06,
      "loss": 0.5477,
      "step": 730
    },
    {
      "epoch": 0.12465893587994543,
      "grad_norm": 0.5845224857330322,
      "learning_rate": 1.953550719645992e-06,
      "loss": 0.5531,
      "step": 731
    },
    {
      "epoch": 0.12482946793997271,
      "grad_norm": 0.5004369020462036,
      "learning_rate": 1.9533841976567266e-06,
      "loss": 0.546,
      "step": 732
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.4830988347530365,
      "learning_rate": 1.9532173848312304e-06,
      "loss": 0.5651,
      "step": 733
    },
    {
      "epoch": 0.1251705320600273,
      "grad_norm": 0.4703359305858612,
      "learning_rate": 1.9530502812203903e-06,
      "loss": 0.5527,
      "step": 734
    },
    {
      "epoch": 0.12534106412005458,
      "grad_norm": 0.44919008016586304,
      "learning_rate": 1.952882886875182e-06,
      "loss": 0.565,
      "step": 735
    },
    {
      "epoch": 0.12551159618008187,
      "grad_norm": 0.6079019904136658,
      "learning_rate": 1.952715201846671e-06,
      "loss": 0.5582,
      "step": 736
    },
    {
      "epoch": 0.12568212824010913,
      "grad_norm": 0.4172450602054596,
      "learning_rate": 1.9525472261860113e-06,
      "loss": 0.5489,
      "step": 737
    },
    {
      "epoch": 0.12585266030013642,
      "grad_norm": 0.6292310357093811,
      "learning_rate": 1.952378959944443e-06,
      "loss": 0.5562,
      "step": 738
    },
    {
      "epoch": 0.1260231923601637,
      "grad_norm": 0.5903599262237549,
      "learning_rate": 1.952210403173299e-06,
      "loss": 0.5781,
      "step": 739
    },
    {
      "epoch": 0.126193724420191,
      "grad_norm": 0.3425973653793335,
      "learning_rate": 1.952041555923997e-06,
      "loss": 0.5507,
      "step": 740
    },
    {
      "epoch": 0.1263642564802183,
      "grad_norm": 0.6512162089347839,
      "learning_rate": 1.9518724182480464e-06,
      "loss": 0.578,
      "step": 741
    },
    {
      "epoch": 0.12653478854024558,
      "grad_norm": 0.51100754737854,
      "learning_rate": 1.9517029901970427e-06,
      "loss": 0.5726,
      "step": 742
    },
    {
      "epoch": 0.12670532060027284,
      "grad_norm": 0.41974154114723206,
      "learning_rate": 1.9515332718226714e-06,
      "loss": 0.5413,
      "step": 743
    },
    {
      "epoch": 0.12687585266030013,
      "grad_norm": 0.5773981213569641,
      "learning_rate": 1.9513632631767066e-06,
      "loss": 0.5574,
      "step": 744
    },
    {
      "epoch": 0.12704638472032742,
      "grad_norm": 0.534631073474884,
      "learning_rate": 1.9511929643110097e-06,
      "loss": 0.5651,
      "step": 745
    },
    {
      "epoch": 0.1272169167803547,
      "grad_norm": 0.42577818036079407,
      "learning_rate": 1.9510223752775325e-06,
      "loss": 0.5523,
      "step": 746
    },
    {
      "epoch": 0.127387448840382,
      "grad_norm": 0.5253404974937439,
      "learning_rate": 1.950851496128314e-06,
      "loss": 0.5495,
      "step": 747
    },
    {
      "epoch": 0.1275579809004093,
      "grad_norm": 0.43779006600379944,
      "learning_rate": 1.9506803269154815e-06,
      "loss": 0.5716,
      "step": 748
    },
    {
      "epoch": 0.12772851296043655,
      "grad_norm": 0.5063499808311462,
      "learning_rate": 1.9505088676912516e-06,
      "loss": 0.5485,
      "step": 749
    },
    {
      "epoch": 0.12789904502046384,
      "grad_norm": 0.5007257461547852,
      "learning_rate": 1.9503371185079296e-06,
      "loss": 0.5498,
      "step": 750
    },
    {
      "epoch": 0.12806957708049113,
      "grad_norm": 0.47316974401474,
      "learning_rate": 1.9501650794179083e-06,
      "loss": 0.553,
      "step": 751
    },
    {
      "epoch": 0.12824010914051842,
      "grad_norm": 0.4601980149745941,
      "learning_rate": 1.9499927504736694e-06,
      "loss": 0.5443,
      "step": 752
    },
    {
      "epoch": 0.1284106412005457,
      "grad_norm": 0.5112723112106323,
      "learning_rate": 1.9498201317277832e-06,
      "loss": 0.5439,
      "step": 753
    },
    {
      "epoch": 0.128581173260573,
      "grad_norm": 0.37005728483200073,
      "learning_rate": 1.9496472232329076e-06,
      "loss": 0.5589,
      "step": 754
    },
    {
      "epoch": 0.12875170532060026,
      "grad_norm": 0.4853660762310028,
      "learning_rate": 1.9494740250417903e-06,
      "loss": 0.5443,
      "step": 755
    },
    {
      "epoch": 0.12892223738062755,
      "grad_norm": 0.4041537642478943,
      "learning_rate": 1.949300537207266e-06,
      "loss": 0.5486,
      "step": 756
    },
    {
      "epoch": 0.12909276944065484,
      "grad_norm": 0.40221962332725525,
      "learning_rate": 1.949126759782258e-06,
      "loss": 0.5317,
      "step": 757
    },
    {
      "epoch": 0.12926330150068213,
      "grad_norm": 0.46186521649360657,
      "learning_rate": 1.9489526928197795e-06,
      "loss": 0.5705,
      "step": 758
    },
    {
      "epoch": 0.12943383356070942,
      "grad_norm": 0.47338148951530457,
      "learning_rate": 1.9487783363729296e-06,
      "loss": 0.552,
      "step": 759
    },
    {
      "epoch": 0.1296043656207367,
      "grad_norm": 0.44101738929748535,
      "learning_rate": 1.948603690494898e-06,
      "loss": 0.5483,
      "step": 760
    },
    {
      "epoch": 0.12977489768076397,
      "grad_norm": 0.41624701023101807,
      "learning_rate": 1.9484287552389604e-06,
      "loss": 0.5553,
      "step": 761
    },
    {
      "epoch": 0.12994542974079126,
      "grad_norm": 0.3483124077320099,
      "learning_rate": 1.948253530658483e-06,
      "loss": 0.5583,
      "step": 762
    },
    {
      "epoch": 0.13011596180081855,
      "grad_norm": 0.4810324013233185,
      "learning_rate": 1.9480780168069182e-06,
      "loss": 0.5474,
      "step": 763
    },
    {
      "epoch": 0.13028649386084584,
      "grad_norm": 0.4634459912776947,
      "learning_rate": 1.947902213737809e-06,
      "loss": 0.5442,
      "step": 764
    },
    {
      "epoch": 0.13045702592087313,
      "grad_norm": 0.33125039935112,
      "learning_rate": 1.9477261215047836e-06,
      "loss": 0.5611,
      "step": 765
    },
    {
      "epoch": 0.13062755798090042,
      "grad_norm": 0.4639657735824585,
      "learning_rate": 1.947549740161562e-06,
      "loss": 0.5572,
      "step": 766
    },
    {
      "epoch": 0.13079809004092768,
      "grad_norm": 0.52504962682724,
      "learning_rate": 1.9473730697619487e-06,
      "loss": 0.5623,
      "step": 767
    },
    {
      "epoch": 0.13096862210095497,
      "grad_norm": 0.41981661319732666,
      "learning_rate": 1.9471961103598393e-06,
      "loss": 0.5655,
      "step": 768
    },
    {
      "epoch": 0.13113915416098226,
      "grad_norm": 0.4140869081020355,
      "learning_rate": 1.9470188620092163e-06,
      "loss": 0.5506,
      "step": 769
    },
    {
      "epoch": 0.13130968622100955,
      "grad_norm": 0.5477293729782104,
      "learning_rate": 1.9468413247641505e-06,
      "loss": 0.5592,
      "step": 770
    },
    {
      "epoch": 0.13148021828103684,
      "grad_norm": 0.4434651732444763,
      "learning_rate": 1.9466634986788006e-06,
      "loss": 0.5561,
      "step": 771
    },
    {
      "epoch": 0.13165075034106413,
      "grad_norm": 0.5599554181098938,
      "learning_rate": 1.9464853838074138e-06,
      "loss": 0.5509,
      "step": 772
    },
    {
      "epoch": 0.1318212824010914,
      "grad_norm": 0.417349249124527,
      "learning_rate": 1.9463069802043253e-06,
      "loss": 0.5643,
      "step": 773
    },
    {
      "epoch": 0.13199181446111868,
      "grad_norm": 0.4504680633544922,
      "learning_rate": 1.946128287923958e-06,
      "loss": 0.5596,
      "step": 774
    },
    {
      "epoch": 0.13216234652114597,
      "grad_norm": 0.4446078836917877,
      "learning_rate": 1.945949307020824e-06,
      "loss": 0.5599,
      "step": 775
    },
    {
      "epoch": 0.13233287858117326,
      "grad_norm": 0.3625372648239136,
      "learning_rate": 1.9457700375495212e-06,
      "loss": 0.5671,
      "step": 776
    },
    {
      "epoch": 0.13250341064120055,
      "grad_norm": 0.48231276869773865,
      "learning_rate": 1.945590479564738e-06,
      "loss": 0.5482,
      "step": 777
    },
    {
      "epoch": 0.13267394270122784,
      "grad_norm": 0.4121912717819214,
      "learning_rate": 1.9454106331212498e-06,
      "loss": 0.5527,
      "step": 778
    },
    {
      "epoch": 0.1328444747612551,
      "grad_norm": 0.4654589891433716,
      "learning_rate": 1.9452304982739198e-06,
      "loss": 0.5591,
      "step": 779
    },
    {
      "epoch": 0.1330150068212824,
      "grad_norm": 0.5038689374923706,
      "learning_rate": 1.945050075077699e-06,
      "loss": 0.5535,
      "step": 780
    },
    {
      "epoch": 0.13318553888130968,
      "grad_norm": 0.39660993218421936,
      "learning_rate": 1.9448693635876267e-06,
      "loss": 0.56,
      "step": 781
    },
    {
      "epoch": 0.13335607094133697,
      "grad_norm": 0.47360673546791077,
      "learning_rate": 1.9446883638588304e-06,
      "loss": 0.5518,
      "step": 782
    },
    {
      "epoch": 0.13352660300136426,
      "grad_norm": 0.3952374756336212,
      "learning_rate": 1.9445070759465256e-06,
      "loss": 0.5724,
      "step": 783
    },
    {
      "epoch": 0.13369713506139155,
      "grad_norm": 0.5022841691970825,
      "learning_rate": 1.944325499906015e-06,
      "loss": 0.5549,
      "step": 784
    },
    {
      "epoch": 0.13386766712141882,
      "grad_norm": 0.5010923743247986,
      "learning_rate": 1.9441436357926896e-06,
      "loss": 0.5597,
      "step": 785
    },
    {
      "epoch": 0.1340381991814461,
      "grad_norm": 0.4214611053466797,
      "learning_rate": 1.943961483662028e-06,
      "loss": 0.5529,
      "step": 786
    },
    {
      "epoch": 0.1342087312414734,
      "grad_norm": 0.47380971908569336,
      "learning_rate": 1.9437790435695973e-06,
      "loss": 0.5471,
      "step": 787
    },
    {
      "epoch": 0.13437926330150068,
      "grad_norm": 0.515921950340271,
      "learning_rate": 1.943596315571052e-06,
      "loss": 0.5633,
      "step": 788
    },
    {
      "epoch": 0.13454979536152797,
      "grad_norm": 0.5466718077659607,
      "learning_rate": 1.943413299722135e-06,
      "loss": 0.5501,
      "step": 789
    },
    {
      "epoch": 0.13472032742155526,
      "grad_norm": 0.4565638303756714,
      "learning_rate": 1.9432299960786757e-06,
      "loss": 0.5543,
      "step": 790
    },
    {
      "epoch": 0.13489085948158253,
      "grad_norm": 0.3882017433643341,
      "learning_rate": 1.9430464046965923e-06,
      "loss": 0.5615,
      "step": 791
    },
    {
      "epoch": 0.13506139154160982,
      "grad_norm": 0.5496183633804321,
      "learning_rate": 1.9428625256318908e-06,
      "loss": 0.5627,
      "step": 792
    },
    {
      "epoch": 0.1352319236016371,
      "grad_norm": 0.4625122845172882,
      "learning_rate": 1.9426783589406645e-06,
      "loss": 0.5486,
      "step": 793
    },
    {
      "epoch": 0.1354024556616644,
      "grad_norm": 0.41997969150543213,
      "learning_rate": 1.942493904679095e-06,
      "loss": 0.5452,
      "step": 794
    },
    {
      "epoch": 0.13557298772169168,
      "grad_norm": 0.4320921003818512,
      "learning_rate": 1.942309162903451e-06,
      "loss": 0.5618,
      "step": 795
    },
    {
      "epoch": 0.13574351978171897,
      "grad_norm": 0.43192803859710693,
      "learning_rate": 1.9421241336700892e-06,
      "loss": 0.5618,
      "step": 796
    },
    {
      "epoch": 0.13591405184174624,
      "grad_norm": 0.3801654577255249,
      "learning_rate": 1.9419388170354548e-06,
      "loss": 0.5574,
      "step": 797
    },
    {
      "epoch": 0.13608458390177353,
      "grad_norm": 0.4831063449382782,
      "learning_rate": 1.9417532130560784e-06,
      "loss": 0.548,
      "step": 798
    },
    {
      "epoch": 0.13625511596180082,
      "grad_norm": 0.38888534903526306,
      "learning_rate": 1.941567321788581e-06,
      "loss": 0.5708,
      "step": 799
    },
    {
      "epoch": 0.1364256480218281,
      "grad_norm": 0.4589107632637024,
      "learning_rate": 1.9413811432896697e-06,
      "loss": 0.5742,
      "step": 800
    },
    {
      "epoch": 0.1365961800818554,
      "grad_norm": 0.4890725314617157,
      "learning_rate": 1.9411946776161388e-06,
      "loss": 0.5634,
      "step": 801
    },
    {
      "epoch": 0.13676671214188268,
      "grad_norm": 0.39768195152282715,
      "learning_rate": 1.9410079248248717e-06,
      "loss": 0.5542,
      "step": 802
    },
    {
      "epoch": 0.13693724420190995,
      "grad_norm": 0.48835986852645874,
      "learning_rate": 1.9408208849728386e-06,
      "loss": 0.5556,
      "step": 803
    },
    {
      "epoch": 0.13710777626193724,
      "grad_norm": 0.436010479927063,
      "learning_rate": 1.9406335581170965e-06,
      "loss": 0.5677,
      "step": 804
    },
    {
      "epoch": 0.13727830832196453,
      "grad_norm": 0.344992458820343,
      "learning_rate": 1.9404459443147916e-06,
      "loss": 0.5693,
      "step": 805
    },
    {
      "epoch": 0.13744884038199182,
      "grad_norm": 0.3935587406158447,
      "learning_rate": 1.940258043623156e-06,
      "loss": 0.5492,
      "step": 806
    },
    {
      "epoch": 0.1376193724420191,
      "grad_norm": 0.45187273621559143,
      "learning_rate": 1.9400698560995105e-06,
      "loss": 0.5611,
      "step": 807
    },
    {
      "epoch": 0.1377899045020464,
      "grad_norm": 0.3705230951309204,
      "learning_rate": 1.939881381801263e-06,
      "loss": 0.5736,
      "step": 808
    },
    {
      "epoch": 0.13796043656207366,
      "grad_norm": 0.39386627078056335,
      "learning_rate": 1.9396926207859087e-06,
      "loss": 0.5685,
      "step": 809
    },
    {
      "epoch": 0.13813096862210095,
      "grad_norm": 0.3702477812767029,
      "learning_rate": 1.9395035731110303e-06,
      "loss": 0.5601,
      "step": 810
    },
    {
      "epoch": 0.13830150068212824,
      "grad_norm": 0.3687109649181366,
      "learning_rate": 1.9393142388342982e-06,
      "loss": 0.5639,
      "step": 811
    },
    {
      "epoch": 0.13847203274215553,
      "grad_norm": 0.42839211225509644,
      "learning_rate": 1.9391246180134703e-06,
      "loss": 0.553,
      "step": 812
    },
    {
      "epoch": 0.13864256480218282,
      "grad_norm": 0.3752674162387848,
      "learning_rate": 1.9389347107063913e-06,
      "loss": 0.5514,
      "step": 813
    },
    {
      "epoch": 0.1388130968622101,
      "grad_norm": 0.38368934392929077,
      "learning_rate": 1.938744516970994e-06,
      "loss": 0.5716,
      "step": 814
    },
    {
      "epoch": 0.13898362892223737,
      "grad_norm": 0.3754592835903168,
      "learning_rate": 1.938554036865298e-06,
      "loss": 0.5458,
      "step": 815
    },
    {
      "epoch": 0.13915416098226466,
      "grad_norm": 0.3761025667190552,
      "learning_rate": 1.9383632704474107e-06,
      "loss": 0.563,
      "step": 816
    },
    {
      "epoch": 0.13932469304229195,
      "grad_norm": 0.36144253611564636,
      "learning_rate": 1.9381722177755267e-06,
      "loss": 0.5773,
      "step": 817
    },
    {
      "epoch": 0.13949522510231924,
      "grad_norm": 0.33363327383995056,
      "learning_rate": 1.9379808789079276e-06,
      "loss": 0.5337,
      "step": 818
    },
    {
      "epoch": 0.13966575716234653,
      "grad_norm": 0.4120798110961914,
      "learning_rate": 1.937789253902983e-06,
      "loss": 0.5612,
      "step": 819
    },
    {
      "epoch": 0.13983628922237382,
      "grad_norm": 0.4070011079311371,
      "learning_rate": 1.9375973428191492e-06,
      "loss": 0.5584,
      "step": 820
    },
    {
      "epoch": 0.14000682128240108,
      "grad_norm": 0.42036911845207214,
      "learning_rate": 1.93740514571497e-06,
      "loss": 0.5463,
      "step": 821
    },
    {
      "epoch": 0.14017735334242837,
      "grad_norm": 0.4309782087802887,
      "learning_rate": 1.9372126626490765e-06,
      "loss": 0.5846,
      "step": 822
    },
    {
      "epoch": 0.14034788540245566,
      "grad_norm": 0.3897610306739807,
      "learning_rate": 1.937019893680187e-06,
      "loss": 0.5504,
      "step": 823
    },
    {
      "epoch": 0.14051841746248295,
      "grad_norm": 0.3699517548084259,
      "learning_rate": 1.9368268388671068e-06,
      "loss": 0.5592,
      "step": 824
    },
    {
      "epoch": 0.14068894952251024,
      "grad_norm": 0.3632091283798218,
      "learning_rate": 1.9366334982687284e-06,
      "loss": 0.551,
      "step": 825
    },
    {
      "epoch": 0.14085948158253753,
      "grad_norm": 1.0206141471862793,
      "learning_rate": 1.9364398719440315e-06,
      "loss": 0.554,
      "step": 826
    },
    {
      "epoch": 0.1410300136425648,
      "grad_norm": 0.5427378416061401,
      "learning_rate": 1.936245959952084e-06,
      "loss": 0.5673,
      "step": 827
    },
    {
      "epoch": 0.14120054570259208,
      "grad_norm": 0.631756603717804,
      "learning_rate": 1.9360517623520393e-06,
      "loss": 0.5532,
      "step": 828
    },
    {
      "epoch": 0.14137107776261937,
      "grad_norm": 0.6074979305267334,
      "learning_rate": 1.9358572792031385e-06,
      "loss": 0.5628,
      "step": 829
    },
    {
      "epoch": 0.14154160982264666,
      "grad_norm": 0.48977214097976685,
      "learning_rate": 1.935662510564711e-06,
      "loss": 0.5575,
      "step": 830
    },
    {
      "epoch": 0.14171214188267395,
      "grad_norm": 0.4265683591365814,
      "learning_rate": 1.935467456496171e-06,
      "loss": 0.561,
      "step": 831
    },
    {
      "epoch": 0.14188267394270124,
      "grad_norm": 0.4576287865638733,
      "learning_rate": 1.935272117057022e-06,
      "loss": 0.556,
      "step": 832
    },
    {
      "epoch": 0.1420532060027285,
      "grad_norm": 0.45377016067504883,
      "learning_rate": 1.9350764923068535e-06,
      "loss": 0.5561,
      "step": 833
    },
    {
      "epoch": 0.1422237380627558,
      "grad_norm": 0.3955170214176178,
      "learning_rate": 1.9348805823053414e-06,
      "loss": 0.5769,
      "step": 834
    },
    {
      "epoch": 0.14239427012278308,
      "grad_norm": 0.4813169538974762,
      "learning_rate": 1.9346843871122506e-06,
      "loss": 0.5433,
      "step": 835
    },
    {
      "epoch": 0.14256480218281037,
      "grad_norm": 0.42308658361434937,
      "learning_rate": 1.93448790678743e-06,
      "loss": 0.5545,
      "step": 836
    },
    {
      "epoch": 0.14273533424283766,
      "grad_norm": 0.5245870351791382,
      "learning_rate": 1.934291141390819e-06,
      "loss": 0.5594,
      "step": 837
    },
    {
      "epoch": 0.14290586630286495,
      "grad_norm": 0.5994154214859009,
      "learning_rate": 1.9340940909824415e-06,
      "loss": 0.5606,
      "step": 838
    },
    {
      "epoch": 0.14307639836289224,
      "grad_norm": 0.43548011779785156,
      "learning_rate": 1.9338967556224087e-06,
      "loss": 0.5467,
      "step": 839
    },
    {
      "epoch": 0.1432469304229195,
      "grad_norm": 0.469954252243042,
      "learning_rate": 1.9336991353709197e-06,
      "loss": 0.5459,
      "step": 840
    },
    {
      "epoch": 0.1434174624829468,
      "grad_norm": 0.5122477412223816,
      "learning_rate": 1.933501230288259e-06,
      "loss": 0.5522,
      "step": 841
    },
    {
      "epoch": 0.14358799454297408,
      "grad_norm": 0.43415263295173645,
      "learning_rate": 1.9333030404348004e-06,
      "loss": 0.5567,
      "step": 842
    },
    {
      "epoch": 0.14375852660300137,
      "grad_norm": 0.42007696628570557,
      "learning_rate": 1.933104565871001e-06,
      "loss": 0.5413,
      "step": 843
    },
    {
      "epoch": 0.14392905866302866,
      "grad_norm": 0.4120525121688843,
      "learning_rate": 1.9329058066574088e-06,
      "loss": 0.553,
      "step": 844
    },
    {
      "epoch": 0.14409959072305595,
      "grad_norm": 0.34494391083717346,
      "learning_rate": 1.9327067628546553e-06,
      "loss": 0.5322,
      "step": 845
    },
    {
      "epoch": 0.1442701227830832,
      "grad_norm": 0.5468747615814209,
      "learning_rate": 1.9325074345234602e-06,
      "loss": 0.5515,
      "step": 846
    },
    {
      "epoch": 0.1444406548431105,
      "grad_norm": 0.6220876574516296,
      "learning_rate": 1.9323078217246308e-06,
      "loss": 0.5583,
      "step": 847
    },
    {
      "epoch": 0.1446111869031378,
      "grad_norm": 0.5272266268730164,
      "learning_rate": 1.9321079245190597e-06,
      "loss": 0.576,
      "step": 848
    },
    {
      "epoch": 0.14478171896316508,
      "grad_norm": 0.39103057980537415,
      "learning_rate": 1.931907742967727e-06,
      "loss": 0.5595,
      "step": 849
    },
    {
      "epoch": 0.14495225102319237,
      "grad_norm": 0.5823642611503601,
      "learning_rate": 1.931707277131699e-06,
      "loss": 0.5576,
      "step": 850
    },
    {
      "epoch": 0.14512278308321966,
      "grad_norm": 0.5365065932273865,
      "learning_rate": 1.93150652707213e-06,
      "loss": 0.5616,
      "step": 851
    },
    {
      "epoch": 0.14529331514324692,
      "grad_norm": 0.4258389174938202,
      "learning_rate": 1.9313054928502598e-06,
      "loss": 0.5504,
      "step": 852
    },
    {
      "epoch": 0.1454638472032742,
      "grad_norm": 0.5903677940368652,
      "learning_rate": 1.931104174527415e-06,
      "loss": 0.5575,
      "step": 853
    },
    {
      "epoch": 0.1456343792633015,
      "grad_norm": 0.48347654938697815,
      "learning_rate": 1.9309025721650092e-06,
      "loss": 0.5417,
      "step": 854
    },
    {
      "epoch": 0.1458049113233288,
      "grad_norm": 0.4118397533893585,
      "learning_rate": 1.9307006858245428e-06,
      "loss": 0.5493,
      "step": 855
    },
    {
      "epoch": 0.14597544338335608,
      "grad_norm": 0.4552883505821228,
      "learning_rate": 1.930498515567602e-06,
      "loss": 0.5398,
      "step": 856
    },
    {
      "epoch": 0.14614597544338337,
      "grad_norm": 0.3356091380119324,
      "learning_rate": 1.9302960614558604e-06,
      "loss": 0.5512,
      "step": 857
    },
    {
      "epoch": 0.14631650750341063,
      "grad_norm": 0.4241127073764801,
      "learning_rate": 1.9300933235510788e-06,
      "loss": 0.5621,
      "step": 858
    },
    {
      "epoch": 0.14648703956343792,
      "grad_norm": 0.39341604709625244,
      "learning_rate": 1.929890301915103e-06,
      "loss": 0.5598,
      "step": 859
    },
    {
      "epoch": 0.1466575716234652,
      "grad_norm": 0.3889201283454895,
      "learning_rate": 1.929686996609866e-06,
      "loss": 0.5463,
      "step": 860
    },
    {
      "epoch": 0.1468281036834925,
      "grad_norm": 0.40517398715019226,
      "learning_rate": 1.929483407697387e-06,
      "loss": 0.5603,
      "step": 861
    },
    {
      "epoch": 0.1469986357435198,
      "grad_norm": 0.4488038718700409,
      "learning_rate": 1.9292795352397737e-06,
      "loss": 0.5567,
      "step": 862
    },
    {
      "epoch": 0.14716916780354708,
      "grad_norm": 0.49497777223587036,
      "learning_rate": 1.929075379299218e-06,
      "loss": 0.5485,
      "step": 863
    },
    {
      "epoch": 0.14733969986357434,
      "grad_norm": 0.45604822039604187,
      "learning_rate": 1.9288709399379984e-06,
      "loss": 0.552,
      "step": 864
    },
    {
      "epoch": 0.14751023192360163,
      "grad_norm": 0.3964408040046692,
      "learning_rate": 1.928666217218481e-06,
      "loss": 0.5665,
      "step": 865
    },
    {
      "epoch": 0.14768076398362892,
      "grad_norm": 0.4897339940071106,
      "learning_rate": 1.928461211203118e-06,
      "loss": 0.5387,
      "step": 866
    },
    {
      "epoch": 0.1478512960436562,
      "grad_norm": 0.477505624294281,
      "learning_rate": 1.9282559219544476e-06,
      "loss": 0.5553,
      "step": 867
    },
    {
      "epoch": 0.1480218281036835,
      "grad_norm": 0.4992867410182953,
      "learning_rate": 1.9280503495350953e-06,
      "loss": 0.5678,
      "step": 868
    },
    {
      "epoch": 0.1481923601637108,
      "grad_norm": 0.542177140712738,
      "learning_rate": 1.927844494007771e-06,
      "loss": 0.5472,
      "step": 869
    },
    {
      "epoch": 0.14836289222373805,
      "grad_norm": 0.41191962361335754,
      "learning_rate": 1.9276383554352736e-06,
      "loss": 0.5753,
      "step": 870
    },
    {
      "epoch": 0.14853342428376534,
      "grad_norm": 0.643372118473053,
      "learning_rate": 1.9274319338804866e-06,
      "loss": 0.5607,
      "step": 871
    },
    {
      "epoch": 0.14870395634379263,
      "grad_norm": 0.757784903049469,
      "learning_rate": 1.92722522940638e-06,
      "loss": 0.5533,
      "step": 872
    },
    {
      "epoch": 0.14887448840381992,
      "grad_norm": 0.47195321321487427,
      "learning_rate": 1.9270182420760104e-06,
      "loss": 0.5521,
      "step": 873
    },
    {
      "epoch": 0.1490450204638472,
      "grad_norm": 0.5352654457092285,
      "learning_rate": 1.926810971952521e-06,
      "loss": 0.5766,
      "step": 874
    },
    {
      "epoch": 0.1492155525238745,
      "grad_norm": 0.6699765920639038,
      "learning_rate": 1.9266034190991415e-06,
      "loss": 0.574,
      "step": 875
    },
    {
      "epoch": 0.14938608458390176,
      "grad_norm": 0.44111526012420654,
      "learning_rate": 1.926395583579186e-06,
      "loss": 0.5707,
      "step": 876
    },
    {
      "epoch": 0.14955661664392905,
      "grad_norm": 0.5262956023216248,
      "learning_rate": 1.926187465456057e-06,
      "loss": 0.5551,
      "step": 877
    },
    {
      "epoch": 0.14972714870395634,
      "grad_norm": 0.5442524552345276,
      "learning_rate": 1.925979064793242e-06,
      "loss": 0.5482,
      "step": 878
    },
    {
      "epoch": 0.14989768076398363,
      "grad_norm": 0.46086442470550537,
      "learning_rate": 1.9257703816543146e-06,
      "loss": 0.5678,
      "step": 879
    },
    {
      "epoch": 0.15006821282401092,
      "grad_norm": 0.45078033208847046,
      "learning_rate": 1.925561416102936e-06,
      "loss": 0.5542,
      "step": 880
    },
    {
      "epoch": 0.1502387448840382,
      "grad_norm": 0.3949615955352783,
      "learning_rate": 1.9253521682028516e-06,
      "loss": 0.549,
      "step": 881
    },
    {
      "epoch": 0.15040927694406547,
      "grad_norm": 0.4516843855381012,
      "learning_rate": 1.9251426380178947e-06,
      "loss": 0.5714,
      "step": 882
    },
    {
      "epoch": 0.15057980900409276,
      "grad_norm": 0.3815528154373169,
      "learning_rate": 1.9249328256119834e-06,
      "loss": 0.5473,
      "step": 883
    },
    {
      "epoch": 0.15075034106412005,
      "grad_norm": 0.4536670446395874,
      "learning_rate": 1.9247227310491224e-06,
      "loss": 0.5674,
      "step": 884
    },
    {
      "epoch": 0.15092087312414734,
      "grad_norm": 0.3400889039039612,
      "learning_rate": 1.924512354393402e-06,
      "loss": 0.5777,
      "step": 885
    },
    {
      "epoch": 0.15109140518417463,
      "grad_norm": 0.485637903213501,
      "learning_rate": 1.9243016957090003e-06,
      "loss": 0.5513,
      "step": 886
    },
    {
      "epoch": 0.15126193724420192,
      "grad_norm": 0.4044179916381836,
      "learning_rate": 1.9240907550601787e-06,
      "loss": 0.5454,
      "step": 887
    },
    {
      "epoch": 0.15143246930422918,
      "grad_norm": 0.43215033411979675,
      "learning_rate": 1.923879532511287e-06,
      "loss": 0.555,
      "step": 888
    },
    {
      "epoch": 0.15160300136425647,
      "grad_norm": 0.46532517671585083,
      "learning_rate": 1.9236680281267597e-06,
      "loss": 0.5576,
      "step": 889
    },
    {
      "epoch": 0.15177353342428376,
      "grad_norm": 0.3941131830215454,
      "learning_rate": 1.923456241971118e-06,
      "loss": 0.5381,
      "step": 890
    },
    {
      "epoch": 0.15194406548431105,
      "grad_norm": 0.46986761689186096,
      "learning_rate": 1.923244174108968e-06,
      "loss": 0.5365,
      "step": 891
    },
    {
      "epoch": 0.15211459754433834,
      "grad_norm": 0.4742710292339325,
      "learning_rate": 1.923031824605003e-06,
      "loss": 0.5623,
      "step": 892
    },
    {
      "epoch": 0.15228512960436563,
      "grad_norm": 0.41251805424690247,
      "learning_rate": 1.922819193524001e-06,
      "loss": 0.5533,
      "step": 893
    },
    {
      "epoch": 0.1524556616643929,
      "grad_norm": 0.47409358620643616,
      "learning_rate": 1.9226062809308273e-06,
      "loss": 0.5622,
      "step": 894
    },
    {
      "epoch": 0.15262619372442018,
      "grad_norm": 0.4420965015888214,
      "learning_rate": 1.9223930868904317e-06,
      "loss": 0.5421,
      "step": 895
    },
    {
      "epoch": 0.15279672578444747,
      "grad_norm": 0.48168325424194336,
      "learning_rate": 1.922179611467851e-06,
      "loss": 0.5488,
      "step": 896
    },
    {
      "epoch": 0.15296725784447476,
      "grad_norm": 0.46085238456726074,
      "learning_rate": 1.921965854728207e-06,
      "loss": 0.5497,
      "step": 897
    },
    {
      "epoch": 0.15313778990450205,
      "grad_norm": 0.38156741857528687,
      "learning_rate": 1.921751816736708e-06,
      "loss": 0.5654,
      "step": 898
    },
    {
      "epoch": 0.15330832196452934,
      "grad_norm": 0.4924895167350769,
      "learning_rate": 1.9215374975586467e-06,
      "loss": 0.5607,
      "step": 899
    },
    {
      "epoch": 0.1534788540245566,
      "grad_norm": 0.44584453105926514,
      "learning_rate": 1.9213228972594035e-06,
      "loss": 0.545,
      "step": 900
    },
    {
      "epoch": 0.1536493860845839,
      "grad_norm": 0.3664495348930359,
      "learning_rate": 1.9211080159044437e-06,
      "loss": 0.5444,
      "step": 901
    },
    {
      "epoch": 0.15381991814461118,
      "grad_norm": 0.5687245726585388,
      "learning_rate": 1.9208928535593182e-06,
      "loss": 0.5544,
      "step": 902
    },
    {
      "epoch": 0.15399045020463847,
      "grad_norm": 0.5331858992576599,
      "learning_rate": 1.920677410289663e-06,
      "loss": 0.5549,
      "step": 903
    },
    {
      "epoch": 0.15416098226466576,
      "grad_norm": 0.468654602766037,
      "learning_rate": 1.9204616861612016e-06,
      "loss": 0.5313,
      "step": 904
    },
    {
      "epoch": 0.15433151432469305,
      "grad_norm": 0.40355348587036133,
      "learning_rate": 1.920245681239741e-06,
      "loss": 0.5492,
      "step": 905
    },
    {
      "epoch": 0.15450204638472032,
      "grad_norm": 0.5259311199188232,
      "learning_rate": 1.920029395591176e-06,
      "loss": 0.5557,
      "step": 906
    },
    {
      "epoch": 0.1546725784447476,
      "grad_norm": 0.48963090777397156,
      "learning_rate": 1.9198128292814852e-06,
      "loss": 0.552,
      "step": 907
    },
    {
      "epoch": 0.1548431105047749,
      "grad_norm": 0.36119452118873596,
      "learning_rate": 1.919595982376734e-06,
      "loss": 0.5656,
      "step": 908
    },
    {
      "epoch": 0.15501364256480218,
      "grad_norm": 0.456949383020401,
      "learning_rate": 1.919378854943073e-06,
      "loss": 0.571,
      "step": 909
    },
    {
      "epoch": 0.15518417462482947,
      "grad_norm": 0.45524856448173523,
      "learning_rate": 1.9191614470467382e-06,
      "loss": 0.562,
      "step": 910
    },
    {
      "epoch": 0.15535470668485676,
      "grad_norm": 0.5876911878585815,
      "learning_rate": 1.9189437587540513e-06,
      "loss": 0.5733,
      "step": 911
    },
    {
      "epoch": 0.15552523874488403,
      "grad_norm": 0.48358115553855896,
      "learning_rate": 1.9187257901314194e-06,
      "loss": 0.5519,
      "step": 912
    },
    {
      "epoch": 0.15569577080491132,
      "grad_norm": 0.39478081464767456,
      "learning_rate": 1.918507541245336e-06,
      "loss": 0.5618,
      "step": 913
    },
    {
      "epoch": 0.1558663028649386,
      "grad_norm": 0.4886495769023895,
      "learning_rate": 1.918289012162379e-06,
      "loss": 0.5627,
      "step": 914
    },
    {
      "epoch": 0.1560368349249659,
      "grad_norm": 0.5879368185997009,
      "learning_rate": 1.918070202949212e-06,
      "loss": 0.5482,
      "step": 915
    },
    {
      "epoch": 0.15620736698499318,
      "grad_norm": 0.7165773510932922,
      "learning_rate": 1.9178511136725843e-06,
      "loss": 0.5584,
      "step": 916
    },
    {
      "epoch": 0.15637789904502047,
      "grad_norm": 0.7150624990463257,
      "learning_rate": 1.9176317443993307e-06,
      "loss": 0.5639,
      "step": 917
    },
    {
      "epoch": 0.15654843110504774,
      "grad_norm": 0.45642855763435364,
      "learning_rate": 1.9174120951963706e-06,
      "loss": 0.5507,
      "step": 918
    },
    {
      "epoch": 0.15671896316507503,
      "grad_norm": 0.42844220995903015,
      "learning_rate": 1.9171921661307105e-06,
      "loss": 0.5409,
      "step": 919
    },
    {
      "epoch": 0.15688949522510232,
      "grad_norm": 0.5636053085327148,
      "learning_rate": 1.9169719572694406e-06,
      "loss": 0.5448,
      "step": 920
    },
    {
      "epoch": 0.1570600272851296,
      "grad_norm": 0.4291675388813019,
      "learning_rate": 1.9167514686797373e-06,
      "loss": 0.5603,
      "step": 921
    },
    {
      "epoch": 0.1572305593451569,
      "grad_norm": 0.4401952028274536,
      "learning_rate": 1.916530700428862e-06,
      "loss": 0.5476,
      "step": 922
    },
    {
      "epoch": 0.15740109140518418,
      "grad_norm": 0.5289968252182007,
      "learning_rate": 1.9163096525841614e-06,
      "loss": 0.55,
      "step": 923
    },
    {
      "epoch": 0.15757162346521145,
      "grad_norm": 0.38858067989349365,
      "learning_rate": 1.916088325213068e-06,
      "loss": 0.5564,
      "step": 924
    },
    {
      "epoch": 0.15774215552523874,
      "grad_norm": 0.477230429649353,
      "learning_rate": 1.915866718383099e-06,
      "loss": 0.5518,
      "step": 925
    },
    {
      "epoch": 0.15791268758526603,
      "grad_norm": 0.5338437557220459,
      "learning_rate": 1.915644832161857e-06,
      "loss": 0.5662,
      "step": 926
    },
    {
      "epoch": 0.15808321964529332,
      "grad_norm": 0.4159543812274933,
      "learning_rate": 1.91542266661703e-06,
      "loss": 0.5654,
      "step": 927
    },
    {
      "epoch": 0.1582537517053206,
      "grad_norm": 0.3964542746543884,
      "learning_rate": 1.9152002218163904e-06,
      "loss": 0.5541,
      "step": 928
    },
    {
      "epoch": 0.1584242837653479,
      "grad_norm": 0.5510343909263611,
      "learning_rate": 1.9149774978277976e-06,
      "loss": 0.5575,
      "step": 929
    },
    {
      "epoch": 0.15859481582537516,
      "grad_norm": 0.4642314910888672,
      "learning_rate": 1.9147544947191947e-06,
      "loss": 0.5476,
      "step": 930
    },
    {
      "epoch": 0.15876534788540245,
      "grad_norm": 0.4070025086402893,
      "learning_rate": 1.9145312125586095e-06,
      "loss": 0.5626,
      "step": 931
    },
    {
      "epoch": 0.15893587994542974,
      "grad_norm": 0.41872039437294006,
      "learning_rate": 1.914307651414157e-06,
      "loss": 0.5482,
      "step": 932
    },
    {
      "epoch": 0.15910641200545703,
      "grad_norm": 0.3984706401824951,
      "learning_rate": 1.914083811354035e-06,
      "loss": 0.5533,
      "step": 933
    },
    {
      "epoch": 0.15927694406548432,
      "grad_norm": 0.40920400619506836,
      "learning_rate": 1.9138596924465277e-06,
      "loss": 0.5505,
      "step": 934
    },
    {
      "epoch": 0.1594474761255116,
      "grad_norm": 0.3676402270793915,
      "learning_rate": 1.9136352947600047e-06,
      "loss": 0.5634,
      "step": 935
    },
    {
      "epoch": 0.15961800818553887,
      "grad_norm": 0.4323231875896454,
      "learning_rate": 1.9134106183629184e-06,
      "loss": 0.5611,
      "step": 936
    },
    {
      "epoch": 0.15978854024556616,
      "grad_norm": 0.39513805508613586,
      "learning_rate": 1.9131856633238096e-06,
      "loss": 0.5493,
      "step": 937
    },
    {
      "epoch": 0.15995907230559345,
      "grad_norm": 0.47802916169166565,
      "learning_rate": 1.9129604297113015e-06,
      "loss": 0.5516,
      "step": 938
    },
    {
      "epoch": 0.16012960436562074,
      "grad_norm": 0.394153892993927,
      "learning_rate": 1.9127349175941032e-06,
      "loss": 0.5467,
      "step": 939
    },
    {
      "epoch": 0.16030013642564803,
      "grad_norm": 0.4861249029636383,
      "learning_rate": 1.9125091270410088e-06,
      "loss": 0.5611,
      "step": 940
    },
    {
      "epoch": 0.16047066848567532,
      "grad_norm": 0.5115260481834412,
      "learning_rate": 1.912283058120897e-06,
      "loss": 0.5379,
      "step": 941
    },
    {
      "epoch": 0.16064120054570258,
      "grad_norm": 0.3449493646621704,
      "learning_rate": 1.9120567109027323e-06,
      "loss": 0.5678,
      "step": 942
    },
    {
      "epoch": 0.16081173260572987,
      "grad_norm": 0.4588490426540375,
      "learning_rate": 1.9118300854555625e-06,
      "loss": 0.5415,
      "step": 943
    },
    {
      "epoch": 0.16098226466575716,
      "grad_norm": 0.41280704736709595,
      "learning_rate": 1.9116031818485217e-06,
      "loss": 0.5542,
      "step": 944
    },
    {
      "epoch": 0.16115279672578445,
      "grad_norm": 0.3902633786201477,
      "learning_rate": 1.9113760001508284e-06,
      "loss": 0.5663,
      "step": 945
    },
    {
      "epoch": 0.16132332878581174,
      "grad_norm": 0.3669435977935791,
      "learning_rate": 1.9111485404317853e-06,
      "loss": 0.5393,
      "step": 946
    },
    {
      "epoch": 0.16149386084583903,
      "grad_norm": 0.3833954632282257,
      "learning_rate": 1.9109208027607818e-06,
      "loss": 0.5643,
      "step": 947
    },
    {
      "epoch": 0.16166439290586632,
      "grad_norm": 0.4436415135860443,
      "learning_rate": 1.9106927872072893e-06,
      "loss": 0.54,
      "step": 948
    },
    {
      "epoch": 0.16183492496589358,
      "grad_norm": 0.36787348985671997,
      "learning_rate": 1.9104644938408666e-06,
      "loss": 0.5533,
      "step": 949
    },
    {
      "epoch": 0.16200545702592087,
      "grad_norm": 0.5090466141700745,
      "learning_rate": 1.9102359227311555e-06,
      "loss": 0.5408,
      "step": 950
    },
    {
      "epoch": 0.16217598908594816,
      "grad_norm": 0.4917638599872589,
      "learning_rate": 1.9100070739478833e-06,
      "loss": 0.5659,
      "step": 951
    },
    {
      "epoch": 0.16234652114597545,
      "grad_norm": 0.3450438678264618,
      "learning_rate": 1.909777947560862e-06,
      "loss": 0.5501,
      "step": 952
    },
    {
      "epoch": 0.16251705320600274,
      "grad_norm": 0.47705134749412537,
      "learning_rate": 1.909548543639988e-06,
      "loss": 0.563,
      "step": 953
    },
    {
      "epoch": 0.16268758526603003,
      "grad_norm": 0.41855838894844055,
      "learning_rate": 1.9093188622552425e-06,
      "loss": 0.5501,
      "step": 954
    },
    {
      "epoch": 0.1628581173260573,
      "grad_norm": 0.48679107427597046,
      "learning_rate": 1.9090889034766917e-06,
      "loss": 0.5496,
      "step": 955
    },
    {
      "epoch": 0.16302864938608458,
      "grad_norm": 0.4727509319782257,
      "learning_rate": 1.9088586673744853e-06,
      "loss": 0.5451,
      "step": 956
    },
    {
      "epoch": 0.16319918144611187,
      "grad_norm": 0.482374906539917,
      "learning_rate": 1.9086281540188593e-06,
      "loss": 0.5506,
      "step": 957
    },
    {
      "epoch": 0.16336971350613916,
      "grad_norm": 0.47090649604797363,
      "learning_rate": 1.9083973634801322e-06,
      "loss": 0.5512,
      "step": 958
    },
    {
      "epoch": 0.16354024556616645,
      "grad_norm": 0.47305747866630554,
      "learning_rate": 1.908166295828709e-06,
      "loss": 0.5626,
      "step": 959
    },
    {
      "epoch": 0.16371077762619374,
      "grad_norm": 0.39664438366889954,
      "learning_rate": 1.9079349511350787e-06,
      "loss": 0.5689,
      "step": 960
    },
    {
      "epoch": 0.163881309686221,
      "grad_norm": 0.4571145474910736,
      "learning_rate": 1.9077033294698137e-06,
      "loss": 0.5583,
      "step": 961
    },
    {
      "epoch": 0.1640518417462483,
      "grad_norm": 0.3828239142894745,
      "learning_rate": 1.9074714309035723e-06,
      "loss": 0.5401,
      "step": 962
    },
    {
      "epoch": 0.16422237380627558,
      "grad_norm": 0.4175052046775818,
      "learning_rate": 1.9072392555070967e-06,
      "loss": 0.5396,
      "step": 963
    },
    {
      "epoch": 0.16439290586630287,
      "grad_norm": 0.41923534870147705,
      "learning_rate": 1.9070068033512134e-06,
      "loss": 0.5477,
      "step": 964
    },
    {
      "epoch": 0.16456343792633016,
      "grad_norm": 0.4210037589073181,
      "learning_rate": 1.9067740745068335e-06,
      "loss": 0.5478,
      "step": 965
    },
    {
      "epoch": 0.16473396998635745,
      "grad_norm": 0.5002460479736328,
      "learning_rate": 1.9065410690449526e-06,
      "loss": 0.5392,
      "step": 966
    },
    {
      "epoch": 0.1649045020463847,
      "grad_norm": 0.34181100130081177,
      "learning_rate": 1.9063077870366503e-06,
      "loss": 0.5337,
      "step": 967
    },
    {
      "epoch": 0.165075034106412,
      "grad_norm": 0.39679914712905884,
      "learning_rate": 1.906074228553091e-06,
      "loss": 0.5657,
      "step": 968
    },
    {
      "epoch": 0.1652455661664393,
      "grad_norm": 0.40103983879089355,
      "learning_rate": 1.9058403936655237e-06,
      "loss": 0.5326,
      "step": 969
    },
    {
      "epoch": 0.16541609822646658,
      "grad_norm": 0.4794570207595825,
      "learning_rate": 1.9056062824452805e-06,
      "loss": 0.5581,
      "step": 970
    },
    {
      "epoch": 0.16558663028649387,
      "grad_norm": 0.4267297089099884,
      "learning_rate": 1.9053718949637792e-06,
      "loss": 0.5458,
      "step": 971
    },
    {
      "epoch": 0.16575716234652116,
      "grad_norm": 0.43018782138824463,
      "learning_rate": 1.905137231292521e-06,
      "loss": 0.5417,
      "step": 972
    },
    {
      "epoch": 0.16592769440654842,
      "grad_norm": 0.47038307785987854,
      "learning_rate": 1.9049022915030915e-06,
      "loss": 0.5524,
      "step": 973
    },
    {
      "epoch": 0.1660982264665757,
      "grad_norm": 0.38833221793174744,
      "learning_rate": 1.904667075667161e-06,
      "loss": 0.5526,
      "step": 974
    },
    {
      "epoch": 0.166268758526603,
      "grad_norm": 0.4255011975765228,
      "learning_rate": 1.9044315838564838e-06,
      "loss": 0.5696,
      "step": 975
    },
    {
      "epoch": 0.1664392905866303,
      "grad_norm": 0.3458400368690491,
      "learning_rate": 1.9041958161428976e-06,
      "loss": 0.5473,
      "step": 976
    },
    {
      "epoch": 0.16660982264665758,
      "grad_norm": 0.41008731722831726,
      "learning_rate": 1.9039597725983255e-06,
      "loss": 0.5409,
      "step": 977
    },
    {
      "epoch": 0.16678035470668487,
      "grad_norm": 0.47080957889556885,
      "learning_rate": 1.903723453294774e-06,
      "loss": 0.5488,
      "step": 978
    },
    {
      "epoch": 0.16695088676671213,
      "grad_norm": 0.4224494397640228,
      "learning_rate": 1.9034868583043335e-06,
      "loss": 0.5601,
      "step": 979
    },
    {
      "epoch": 0.16712141882673942,
      "grad_norm": 0.38210219144821167,
      "learning_rate": 1.9032499876991793e-06,
      "loss": 0.5615,
      "step": 980
    },
    {
      "epoch": 0.1672919508867667,
      "grad_norm": 0.5055222511291504,
      "learning_rate": 1.9030128415515704e-06,
      "loss": 0.56,
      "step": 981
    },
    {
      "epoch": 0.167462482946794,
      "grad_norm": 0.41526558995246887,
      "learning_rate": 1.90277541993385e-06,
      "loss": 0.5478,
      "step": 982
    },
    {
      "epoch": 0.1676330150068213,
      "grad_norm": 0.38019880652427673,
      "learning_rate": 1.9025377229184442e-06,
      "loss": 0.5555,
      "step": 983
    },
    {
      "epoch": 0.16780354706684858,
      "grad_norm": 0.48236194252967834,
      "learning_rate": 1.9022997505778653e-06,
      "loss": 0.5391,
      "step": 984
    },
    {
      "epoch": 0.16797407912687584,
      "grad_norm": 0.450827419757843,
      "learning_rate": 1.9020615029847076e-06,
      "loss": 0.5604,
      "step": 985
    },
    {
      "epoch": 0.16814461118690313,
      "grad_norm": 0.4144051671028137,
      "learning_rate": 1.9018229802116501e-06,
      "loss": 0.5422,
      "step": 986
    },
    {
      "epoch": 0.16831514324693042,
      "grad_norm": 0.38686126470565796,
      "learning_rate": 1.9015841823314562e-06,
      "loss": 0.5461,
      "step": 987
    },
    {
      "epoch": 0.1684856753069577,
      "grad_norm": 0.4112605154514313,
      "learning_rate": 1.9013451094169725e-06,
      "loss": 0.5329,
      "step": 988
    },
    {
      "epoch": 0.168656207366985,
      "grad_norm": 0.4212754964828491,
      "learning_rate": 1.9011057615411299e-06,
      "loss": 0.5681,
      "step": 989
    },
    {
      "epoch": 0.1688267394270123,
      "grad_norm": 0.45845523476600647,
      "learning_rate": 1.900866138776943e-06,
      "loss": 0.5573,
      "step": 990
    },
    {
      "epoch": 0.16899727148703955,
      "grad_norm": 0.5667742490768433,
      "learning_rate": 1.9006262411975102e-06,
      "loss": 0.5599,
      "step": 991
    },
    {
      "epoch": 0.16916780354706684,
      "grad_norm": 0.5878269672393799,
      "learning_rate": 1.9003860688760143e-06,
      "loss": 0.5534,
      "step": 992
    },
    {
      "epoch": 0.16933833560709413,
      "grad_norm": 0.6614935994148254,
      "learning_rate": 1.900145621885721e-06,
      "loss": 0.5464,
      "step": 993
    },
    {
      "epoch": 0.16950886766712142,
      "grad_norm": 0.44332489371299744,
      "learning_rate": 1.8999049002999805e-06,
      "loss": 0.5511,
      "step": 994
    },
    {
      "epoch": 0.1696793997271487,
      "grad_norm": 0.3465108871459961,
      "learning_rate": 1.8996639041922266e-06,
      "loss": 0.5506,
      "step": 995
    },
    {
      "epoch": 0.169849931787176,
      "grad_norm": 0.4552614688873291,
      "learning_rate": 1.8994226336359765e-06,
      "loss": 0.5495,
      "step": 996
    },
    {
      "epoch": 0.17002046384720326,
      "grad_norm": 0.502493679523468,
      "learning_rate": 1.8991810887048315e-06,
      "loss": 0.5524,
      "step": 997
    },
    {
      "epoch": 0.17019099590723055,
      "grad_norm": 0.3657921254634857,
      "learning_rate": 1.8989392694724766e-06,
      "loss": 0.5616,
      "step": 998
    },
    {
      "epoch": 0.17036152796725784,
      "grad_norm": 0.49448162317276,
      "learning_rate": 1.8986971760126806e-06,
      "loss": 0.5446,
      "step": 999
    },
    {
      "epoch": 0.17053206002728513,
      "grad_norm": 0.6634671688079834,
      "learning_rate": 1.8984548083992952e-06,
      "loss": 0.5656,
      "step": 1000
    },
    {
      "epoch": 0.17070259208731242,
      "grad_norm": 0.7345843315124512,
      "learning_rate": 1.8982121667062567e-06,
      "loss": 0.5497,
      "step": 1001
    },
    {
      "epoch": 0.1708731241473397,
      "grad_norm": 0.45212310552597046,
      "learning_rate": 1.8979692510075842e-06,
      "loss": 0.5444,
      "step": 1002
    },
    {
      "epoch": 0.17104365620736697,
      "grad_norm": 0.3672301471233368,
      "learning_rate": 1.8977260613773812e-06,
      "loss": 0.5407,
      "step": 1003
    },
    {
      "epoch": 0.17121418826739426,
      "grad_norm": 0.5045791864395142,
      "learning_rate": 1.8974825978898339e-06,
      "loss": 0.5597,
      "step": 1004
    },
    {
      "epoch": 0.17138472032742155,
      "grad_norm": 0.4292773902416229,
      "learning_rate": 1.8972388606192126e-06,
      "loss": 0.5406,
      "step": 1005
    },
    {
      "epoch": 0.17155525238744884,
      "grad_norm": 0.3314538598060608,
      "learning_rate": 1.896994849639871e-06,
      "loss": 0.5534,
      "step": 1006
    },
    {
      "epoch": 0.17172578444747613,
      "grad_norm": 0.43244096636772156,
      "learning_rate": 1.8967505650262465e-06,
      "loss": 0.5411,
      "step": 1007
    },
    {
      "epoch": 0.17189631650750342,
      "grad_norm": 0.4487178921699524,
      "learning_rate": 1.8965060068528594e-06,
      "loss": 0.5489,
      "step": 1008
    },
    {
      "epoch": 0.17206684856753068,
      "grad_norm": 0.4456799328327179,
      "learning_rate": 1.8962611751943143e-06,
      "loss": 0.5577,
      "step": 1009
    },
    {
      "epoch": 0.17223738062755797,
      "grad_norm": 0.3741670846939087,
      "learning_rate": 1.8960160701252982e-06,
      "loss": 0.5506,
      "step": 1010
    },
    {
      "epoch": 0.17240791268758526,
      "grad_norm": 0.5865883231163025,
      "learning_rate": 1.8957706917205822e-06,
      "loss": 0.5448,
      "step": 1011
    },
    {
      "epoch": 0.17257844474761255,
      "grad_norm": 0.673882782459259,
      "learning_rate": 1.8955250400550208e-06,
      "loss": 0.5612,
      "step": 1012
    },
    {
      "epoch": 0.17274897680763984,
      "grad_norm": 0.5418551564216614,
      "learning_rate": 1.8952791152035516e-06,
      "loss": 0.5577,
      "step": 1013
    },
    {
      "epoch": 0.17291950886766713,
      "grad_norm": 0.3859636187553406,
      "learning_rate": 1.8950329172411955e-06,
      "loss": 0.5432,
      "step": 1014
    },
    {
      "epoch": 0.1730900409276944,
      "grad_norm": 0.3586163818836212,
      "learning_rate": 1.894786446243057e-06,
      "loss": 0.555,
      "step": 1015
    },
    {
      "epoch": 0.17326057298772168,
      "grad_norm": 0.4536883533000946,
      "learning_rate": 1.8945397022843236e-06,
      "loss": 0.5517,
      "step": 1016
    },
    {
      "epoch": 0.17343110504774897,
      "grad_norm": 0.49440905451774597,
      "learning_rate": 1.8942926854402662e-06,
      "loss": 0.5409,
      "step": 1017
    },
    {
      "epoch": 0.17360163710777626,
      "grad_norm": 0.3685409724712372,
      "learning_rate": 1.894045395786239e-06,
      "loss": 0.5721,
      "step": 1018
    },
    {
      "epoch": 0.17377216916780355,
      "grad_norm": 0.42508232593536377,
      "learning_rate": 1.8937978333976797e-06,
      "loss": 0.557,
      "step": 1019
    },
    {
      "epoch": 0.17394270122783084,
      "grad_norm": 0.4841102659702301,
      "learning_rate": 1.8935499983501082e-06,
      "loss": 0.5488,
      "step": 1020
    },
    {
      "epoch": 0.1741132332878581,
      "grad_norm": 0.43548738956451416,
      "learning_rate": 1.8933018907191287e-06,
      "loss": 0.5481,
      "step": 1021
    },
    {
      "epoch": 0.1742837653478854,
      "grad_norm": 0.39762040972709656,
      "learning_rate": 1.8930535105804277e-06,
      "loss": 0.5654,
      "step": 1022
    },
    {
      "epoch": 0.17445429740791268,
      "grad_norm": 0.4504348337650299,
      "learning_rate": 1.892804858009776e-06,
      "loss": 0.5444,
      "step": 1023
    },
    {
      "epoch": 0.17462482946793997,
      "grad_norm": 0.5455803275108337,
      "learning_rate": 1.8925559330830256e-06,
      "loss": 0.5553,
      "step": 1024
    },
    {
      "epoch": 0.17479536152796726,
      "grad_norm": 0.551396369934082,
      "learning_rate": 1.8923067358761138e-06,
      "loss": 0.56,
      "step": 1025
    },
    {
      "epoch": 0.17496589358799455,
      "grad_norm": 0.4443368911743164,
      "learning_rate": 1.892057266465059e-06,
      "loss": 0.5432,
      "step": 1026
    },
    {
      "epoch": 0.17513642564802182,
      "grad_norm": 0.3978983759880066,
      "learning_rate": 1.8918075249259642e-06,
      "loss": 0.5501,
      "step": 1027
    },
    {
      "epoch": 0.1753069577080491,
      "grad_norm": 0.4175165891647339,
      "learning_rate": 1.8915575113350144e-06,
      "loss": 0.5499,
      "step": 1028
    },
    {
      "epoch": 0.1754774897680764,
      "grad_norm": 0.381310373544693,
      "learning_rate": 1.891307225768478e-06,
      "loss": 0.5414,
      "step": 1029
    },
    {
      "epoch": 0.17564802182810368,
      "grad_norm": 0.44892942905426025,
      "learning_rate": 1.891056668302706e-06,
      "loss": 0.5506,
      "step": 1030
    },
    {
      "epoch": 0.17581855388813097,
      "grad_norm": 0.4433884024620056,
      "learning_rate": 1.890805839014133e-06,
      "loss": 0.5556,
      "step": 1031
    },
    {
      "epoch": 0.17598908594815826,
      "grad_norm": 0.36218923330307007,
      "learning_rate": 1.8905547379792762e-06,
      "loss": 0.5377,
      "step": 1032
    },
    {
      "epoch": 0.17615961800818553,
      "grad_norm": 0.49523016810417175,
      "learning_rate": 1.8903033652747355e-06,
      "loss": 0.5572,
      "step": 1033
    },
    {
      "epoch": 0.17633015006821282,
      "grad_norm": 0.5547558069229126,
      "learning_rate": 1.8900517209771934e-06,
      "loss": 0.5579,
      "step": 1034
    },
    {
      "epoch": 0.1765006821282401,
      "grad_norm": 0.45448794960975647,
      "learning_rate": 1.8897998051634167e-06,
      "loss": 0.5451,
      "step": 1035
    },
    {
      "epoch": 0.1766712141882674,
      "grad_norm": 0.45425528287887573,
      "learning_rate": 1.8895476179102529e-06,
      "loss": 0.5616,
      "step": 1036
    },
    {
      "epoch": 0.17684174624829468,
      "grad_norm": 0.5401942729949951,
      "learning_rate": 1.8892951592946345e-06,
      "loss": 0.5648,
      "step": 1037
    },
    {
      "epoch": 0.17701227830832197,
      "grad_norm": 0.48139169812202454,
      "learning_rate": 1.8890424293935744e-06,
      "loss": 0.5333,
      "step": 1038
    },
    {
      "epoch": 0.17718281036834924,
      "grad_norm": 0.5250762701034546,
      "learning_rate": 1.8887894282841707e-06,
      "loss": 0.5572,
      "step": 1039
    },
    {
      "epoch": 0.17735334242837653,
      "grad_norm": 0.49231913685798645,
      "learning_rate": 1.8885361560436024e-06,
      "loss": 0.5411,
      "step": 1040
    },
    {
      "epoch": 0.17752387448840382,
      "grad_norm": 0.44401469826698303,
      "learning_rate": 1.8882826127491323e-06,
      "loss": 0.5631,
      "step": 1041
    },
    {
      "epoch": 0.1776944065484311,
      "grad_norm": 0.5123889446258545,
      "learning_rate": 1.888028798478105e-06,
      "loss": 0.5647,
      "step": 1042
    },
    {
      "epoch": 0.1778649386084584,
      "grad_norm": 0.48957693576812744,
      "learning_rate": 1.8877747133079488e-06,
      "loss": 0.5417,
      "step": 1043
    },
    {
      "epoch": 0.17803547066848568,
      "grad_norm": 0.5831753611564636,
      "learning_rate": 1.8875203573161737e-06,
      "loss": 0.558,
      "step": 1044
    },
    {
      "epoch": 0.17820600272851295,
      "grad_norm": 0.6821199655532837,
      "learning_rate": 1.8872657305803731e-06,
      "loss": 0.5435,
      "step": 1045
    },
    {
      "epoch": 0.17837653478854024,
      "grad_norm": 0.3453556001186371,
      "learning_rate": 1.887010833178222e-06,
      "loss": 0.5501,
      "step": 1046
    },
    {
      "epoch": 0.17854706684856753,
      "grad_norm": 0.5605396032333374,
      "learning_rate": 1.8867556651874792e-06,
      "loss": 0.5386,
      "step": 1047
    },
    {
      "epoch": 0.17871759890859482,
      "grad_norm": 0.7059539556503296,
      "learning_rate": 1.8865002266859847e-06,
      "loss": 0.5408,
      "step": 1048
    },
    {
      "epoch": 0.1788881309686221,
      "grad_norm": 0.536009669303894,
      "learning_rate": 1.8862445177516624e-06,
      "loss": 0.5471,
      "step": 1049
    },
    {
      "epoch": 0.1790586630286494,
      "grad_norm": 0.4778744578361511,
      "learning_rate": 1.8859885384625174e-06,
      "loss": 0.5323,
      "step": 1050
    },
    {
      "epoch": 0.17922919508867668,
      "grad_norm": 0.6259710788726807,
      "learning_rate": 1.8857322888966384e-06,
      "loss": 0.5486,
      "step": 1051
    },
    {
      "epoch": 0.17939972714870395,
      "grad_norm": 0.5543518662452698,
      "learning_rate": 1.8854757691321959e-06,
      "loss": 0.5742,
      "step": 1052
    },
    {
      "epoch": 0.17957025920873124,
      "grad_norm": 0.45125091075897217,
      "learning_rate": 1.8852189792474428e-06,
      "loss": 0.5356,
      "step": 1053
    },
    {
      "epoch": 0.17974079126875853,
      "grad_norm": 0.7473435401916504,
      "learning_rate": 1.8849619193207146e-06,
      "loss": 0.5473,
      "step": 1054
    },
    {
      "epoch": 0.17991132332878582,
      "grad_norm": 0.5112062692642212,
      "learning_rate": 1.8847045894304297e-06,
      "loss": 0.5491,
      "step": 1055
    },
    {
      "epoch": 0.1800818553888131,
      "grad_norm": 0.4993838369846344,
      "learning_rate": 1.8844469896550877e-06,
      "loss": 0.5469,
      "step": 1056
    },
    {
      "epoch": 0.1802523874488404,
      "grad_norm": 0.5624857544898987,
      "learning_rate": 1.884189120073271e-06,
      "loss": 0.5346,
      "step": 1057
    },
    {
      "epoch": 0.18042291950886766,
      "grad_norm": 0.506741464138031,
      "learning_rate": 1.8839309807636451e-06,
      "loss": 0.5491,
      "step": 1058
    },
    {
      "epoch": 0.18059345156889495,
      "grad_norm": 0.4264898896217346,
      "learning_rate": 1.8836725718049564e-06,
      "loss": 0.5607,
      "step": 1059
    },
    {
      "epoch": 0.18076398362892224,
      "grad_norm": 0.4436824321746826,
      "learning_rate": 1.883413893276035e-06,
      "loss": 0.537,
      "step": 1060
    },
    {
      "epoch": 0.18093451568894953,
      "grad_norm": 0.5607333183288574,
      "learning_rate": 1.8831549452557918e-06,
      "loss": 0.5567,
      "step": 1061
    },
    {
      "epoch": 0.18110504774897682,
      "grad_norm": 0.3857938349246979,
      "learning_rate": 1.8828957278232217e-06,
      "loss": 0.5405,
      "step": 1062
    },
    {
      "epoch": 0.1812755798090041,
      "grad_norm": 0.3912268280982971,
      "learning_rate": 1.8826362410573992e-06,
      "loss": 0.5401,
      "step": 1063
    },
    {
      "epoch": 0.18144611186903137,
      "grad_norm": 0.43382149934768677,
      "learning_rate": 1.8823764850374836e-06,
      "loss": 0.5541,
      "step": 1064
    },
    {
      "epoch": 0.18161664392905866,
      "grad_norm": 0.47907930612564087,
      "learning_rate": 1.8821164598427148e-06,
      "loss": 0.5404,
      "step": 1065
    },
    {
      "epoch": 0.18178717598908595,
      "grad_norm": 0.40004122257232666,
      "learning_rate": 1.8818561655524157e-06,
      "loss": 0.5373,
      "step": 1066
    },
    {
      "epoch": 0.18195770804911324,
      "grad_norm": 0.3792261481285095,
      "learning_rate": 1.88159560224599e-06,
      "loss": 0.5402,
      "step": 1067
    },
    {
      "epoch": 0.18212824010914053,
      "grad_norm": 0.42916253209114075,
      "learning_rate": 1.8813347700029246e-06,
      "loss": 0.5357,
      "step": 1068
    },
    {
      "epoch": 0.18229877216916782,
      "grad_norm": 0.4503953456878662,
      "learning_rate": 1.8810736689027887e-06,
      "loss": 0.5514,
      "step": 1069
    },
    {
      "epoch": 0.18246930422919508,
      "grad_norm": 0.4629843533039093,
      "learning_rate": 1.8808122990252322e-06,
      "loss": 0.5415,
      "step": 1070
    },
    {
      "epoch": 0.18263983628922237,
      "grad_norm": 0.5524861216545105,
      "learning_rate": 1.8805506604499884e-06,
      "loss": 0.5379,
      "step": 1071
    },
    {
      "epoch": 0.18281036834924966,
      "grad_norm": 0.5018559694290161,
      "learning_rate": 1.880288753256871e-06,
      "loss": 0.559,
      "step": 1072
    },
    {
      "epoch": 0.18298090040927695,
      "grad_norm": 0.4084283709526062,
      "learning_rate": 1.8800265775257772e-06,
      "loss": 0.5417,
      "step": 1073
    },
    {
      "epoch": 0.18315143246930424,
      "grad_norm": 0.6081492304801941,
      "learning_rate": 1.8797641333366854e-06,
      "loss": 0.5493,
      "step": 1074
    },
    {
      "epoch": 0.18332196452933153,
      "grad_norm": 0.7344405651092529,
      "learning_rate": 1.8795014207696557e-06,
      "loss": 0.5664,
      "step": 1075
    },
    {
      "epoch": 0.1834924965893588,
      "grad_norm": 0.5947568416595459,
      "learning_rate": 1.8792384399048306e-06,
      "loss": 0.5554,
      "step": 1076
    },
    {
      "epoch": 0.18366302864938608,
      "grad_norm": 0.39973506331443787,
      "learning_rate": 1.878975190822434e-06,
      "loss": 0.547,
      "step": 1077
    },
    {
      "epoch": 0.18383356070941337,
      "grad_norm": 0.559581995010376,
      "learning_rate": 1.8787116736027722e-06,
      "loss": 0.5423,
      "step": 1078
    },
    {
      "epoch": 0.18400409276944066,
      "grad_norm": 0.521888256072998,
      "learning_rate": 1.878447888326232e-06,
      "loss": 0.5455,
      "step": 1079
    },
    {
      "epoch": 0.18417462482946795,
      "grad_norm": 0.3867820203304291,
      "learning_rate": 1.8781838350732836e-06,
      "loss": 0.5397,
      "step": 1080
    },
    {
      "epoch": 0.18434515688949524,
      "grad_norm": 0.5141144394874573,
      "learning_rate": 1.8779195139244782e-06,
      "loss": 0.5558,
      "step": 1081
    },
    {
      "epoch": 0.1845156889495225,
      "grad_norm": 0.685032308101654,
      "learning_rate": 1.8776549249604485e-06,
      "loss": 0.5759,
      "step": 1082
    },
    {
      "epoch": 0.1846862210095498,
      "grad_norm": 0.43259522318840027,
      "learning_rate": 1.8773900682619091e-06,
      "loss": 0.5405,
      "step": 1083
    },
    {
      "epoch": 0.18485675306957708,
      "grad_norm": 0.3838390111923218,
      "learning_rate": 1.8771249439096566e-06,
      "loss": 0.5571,
      "step": 1084
    },
    {
      "epoch": 0.18502728512960437,
      "grad_norm": 0.5635690093040466,
      "learning_rate": 1.8768595519845685e-06,
      "loss": 0.5538,
      "step": 1085
    },
    {
      "epoch": 0.18519781718963166,
      "grad_norm": 0.5552285313606262,
      "learning_rate": 1.8765938925676049e-06,
      "loss": 0.5296,
      "step": 1086
    },
    {
      "epoch": 0.18536834924965895,
      "grad_norm": 0.39209043979644775,
      "learning_rate": 1.8763279657398065e-06,
      "loss": 0.5453,
      "step": 1087
    },
    {
      "epoch": 0.1855388813096862,
      "grad_norm": 0.458196222782135,
      "learning_rate": 1.8760617715822967e-06,
      "loss": 0.5717,
      "step": 1088
    },
    {
      "epoch": 0.1857094133697135,
      "grad_norm": 0.40345677733421326,
      "learning_rate": 1.8757953101762788e-06,
      "loss": 0.5426,
      "step": 1089
    },
    {
      "epoch": 0.1858799454297408,
      "grad_norm": 0.3410954177379608,
      "learning_rate": 1.8755285816030397e-06,
      "loss": 0.537,
      "step": 1090
    },
    {
      "epoch": 0.18605047748976808,
      "grad_norm": 0.4490777850151062,
      "learning_rate": 1.8752615859439462e-06,
      "loss": 0.5424,
      "step": 1091
    },
    {
      "epoch": 0.18622100954979537,
      "grad_norm": 0.4328539967536926,
      "learning_rate": 1.874994323280447e-06,
      "loss": 0.5557,
      "step": 1092
    },
    {
      "epoch": 0.18639154160982266,
      "grad_norm": 0.3498595058917999,
      "learning_rate": 1.8747267936940725e-06,
      "loss": 0.5373,
      "step": 1093
    },
    {
      "epoch": 0.18656207366984992,
      "grad_norm": 0.3731926381587982,
      "learning_rate": 1.8744589972664349e-06,
      "loss": 0.5309,
      "step": 1094
    },
    {
      "epoch": 0.1867326057298772,
      "grad_norm": 0.40898609161376953,
      "learning_rate": 1.8741909340792263e-06,
      "loss": 0.5525,
      "step": 1095
    },
    {
      "epoch": 0.1869031377899045,
      "grad_norm": 0.43006113171577454,
      "learning_rate": 1.8739226042142218e-06,
      "loss": 0.5428,
      "step": 1096
    },
    {
      "epoch": 0.1870736698499318,
      "grad_norm": 0.42080700397491455,
      "learning_rate": 1.8736540077532775e-06,
      "loss": 0.5514,
      "step": 1097
    },
    {
      "epoch": 0.18724420190995908,
      "grad_norm": 0.461102694272995,
      "learning_rate": 1.87338514477833e-06,
      "loss": 0.5435,
      "step": 1098
    },
    {
      "epoch": 0.18741473396998637,
      "grad_norm": 0.430658221244812,
      "learning_rate": 1.8731160153713978e-06,
      "loss": 0.5485,
      "step": 1099
    },
    {
      "epoch": 0.18758526603001363,
      "grad_norm": 0.44055265188217163,
      "learning_rate": 1.8728466196145808e-06,
      "loss": 0.5456,
      "step": 1100
    },
    {
      "epoch": 0.18775579809004092,
      "grad_norm": 0.4936978816986084,
      "learning_rate": 1.8725769575900601e-06,
      "loss": 0.5392,
      "step": 1101
    },
    {
      "epoch": 0.1879263301500682,
      "grad_norm": 0.5400622487068176,
      "learning_rate": 1.8723070293800976e-06,
      "loss": 0.5406,
      "step": 1102
    },
    {
      "epoch": 0.1880968622100955,
      "grad_norm": 0.35937973856925964,
      "learning_rate": 1.8720368350670364e-06,
      "loss": 0.5331,
      "step": 1103
    },
    {
      "epoch": 0.1882673942701228,
      "grad_norm": 0.48796388506889343,
      "learning_rate": 1.871766374733302e-06,
      "loss": 0.5527,
      "step": 1104
    },
    {
      "epoch": 0.18843792633015008,
      "grad_norm": 0.6496586203575134,
      "learning_rate": 1.8714956484613997e-06,
      "loss": 0.5523,
      "step": 1105
    },
    {
      "epoch": 0.18860845839017734,
      "grad_norm": 0.6348623037338257,
      "learning_rate": 1.8712246563339161e-06,
      "loss": 0.5529,
      "step": 1106
    },
    {
      "epoch": 0.18877899045020463,
      "grad_norm": 0.5044506788253784,
      "learning_rate": 1.8709533984335194e-06,
      "loss": 0.5482,
      "step": 1107
    },
    {
      "epoch": 0.18894952251023192,
      "grad_norm": 0.5128569602966309,
      "learning_rate": 1.870681874842959e-06,
      "loss": 0.5724,
      "step": 1108
    },
    {
      "epoch": 0.1891200545702592,
      "grad_norm": 0.6813385486602783,
      "learning_rate": 1.8704100856450645e-06,
      "loss": 0.5376,
      "step": 1109
    },
    {
      "epoch": 0.1892905866302865,
      "grad_norm": 0.5284555554389954,
      "learning_rate": 1.8701380309227467e-06,
      "loss": 0.5366,
      "step": 1110
    },
    {
      "epoch": 0.1894611186903138,
      "grad_norm": 0.4766194522380829,
      "learning_rate": 1.8698657107589985e-06,
      "loss": 0.5472,
      "step": 1111
    },
    {
      "epoch": 0.18963165075034105,
      "grad_norm": 0.5370320081710815,
      "learning_rate": 1.8695931252368923e-06,
      "loss": 0.5468,
      "step": 1112
    },
    {
      "epoch": 0.18980218281036834,
      "grad_norm": 0.5143731832504272,
      "learning_rate": 1.869320274439583e-06,
      "loss": 0.5733,
      "step": 1113
    },
    {
      "epoch": 0.18997271487039563,
      "grad_norm": 0.38538476824760437,
      "learning_rate": 1.8690471584503047e-06,
      "loss": 0.5756,
      "step": 1114
    },
    {
      "epoch": 0.19014324693042292,
      "grad_norm": 0.556102454662323,
      "learning_rate": 1.8687737773523734e-06,
      "loss": 0.5558,
      "step": 1115
    },
    {
      "epoch": 0.1903137789904502,
      "grad_norm": 0.4762594699859619,
      "learning_rate": 1.8685001312291863e-06,
      "loss": 0.5392,
      "step": 1116
    },
    {
      "epoch": 0.1904843110504775,
      "grad_norm": 0.35018035769462585,
      "learning_rate": 1.8682262201642206e-06,
      "loss": 0.5344,
      "step": 1117
    },
    {
      "epoch": 0.19065484311050476,
      "grad_norm": 0.4512125551700592,
      "learning_rate": 1.867952044241035e-06,
      "loss": 0.5407,
      "step": 1118
    },
    {
      "epoch": 0.19082537517053205,
      "grad_norm": 0.39985963702201843,
      "learning_rate": 1.8676776035432682e-06,
      "loss": 0.5528,
      "step": 1119
    },
    {
      "epoch": 0.19099590723055934,
      "grad_norm": 0.538284182548523,
      "learning_rate": 1.8674028981546407e-06,
      "loss": 0.5543,
      "step": 1120
    },
    {
      "epoch": 0.19116643929058663,
      "grad_norm": 0.45395785570144653,
      "learning_rate": 1.8671279281589531e-06,
      "loss": 0.5509,
      "step": 1121
    },
    {
      "epoch": 0.19133697135061392,
      "grad_norm": 0.46653100848197937,
      "learning_rate": 1.8668526936400864e-06,
      "loss": 0.5499,
      "step": 1122
    },
    {
      "epoch": 0.1915075034106412,
      "grad_norm": 0.5032127499580383,
      "learning_rate": 1.8665771946820033e-06,
      "loss": 0.5548,
      "step": 1123
    },
    {
      "epoch": 0.19167803547066847,
      "grad_norm": 0.43430373072624207,
      "learning_rate": 1.8663014313687463e-06,
      "loss": 0.541,
      "step": 1124
    },
    {
      "epoch": 0.19184856753069576,
      "grad_norm": 0.6326541304588318,
      "learning_rate": 1.866025403784439e-06,
      "loss": 0.5392,
      "step": 1125
    },
    {
      "epoch": 0.19201909959072305,
      "grad_norm": 0.5785826444625854,
      "learning_rate": 1.8657491120132853e-06,
      "loss": 0.5416,
      "step": 1126
    },
    {
      "epoch": 0.19218963165075034,
      "grad_norm": 0.35987457633018494,
      "learning_rate": 1.8654725561395698e-06,
      "loss": 0.5389,
      "step": 1127
    },
    {
      "epoch": 0.19236016371077763,
      "grad_norm": 0.5511410236358643,
      "learning_rate": 1.8651957362476582e-06,
      "loss": 0.5232,
      "step": 1128
    },
    {
      "epoch": 0.19253069577080492,
      "grad_norm": 0.5044106245040894,
      "learning_rate": 1.8649186524219957e-06,
      "loss": 0.543,
      "step": 1129
    },
    {
      "epoch": 0.19270122783083218,
      "grad_norm": 0.442216157913208,
      "learning_rate": 1.8646413047471087e-06,
      "loss": 0.5511,
      "step": 1130
    },
    {
      "epoch": 0.19287175989085947,
      "grad_norm": 0.40444454550743103,
      "learning_rate": 1.864363693307604e-06,
      "loss": 0.5422,
      "step": 1131
    },
    {
      "epoch": 0.19304229195088676,
      "grad_norm": 0.4315738081932068,
      "learning_rate": 1.8640858181881693e-06,
      "loss": 0.5516,
      "step": 1132
    },
    {
      "epoch": 0.19321282401091405,
      "grad_norm": 0.380001425743103,
      "learning_rate": 1.8638076794735715e-06,
      "loss": 0.5345,
      "step": 1133
    },
    {
      "epoch": 0.19338335607094134,
      "grad_norm": 0.4576973617076874,
      "learning_rate": 1.8635292772486593e-06,
      "loss": 0.5418,
      "step": 1134
    },
    {
      "epoch": 0.19355388813096863,
      "grad_norm": 0.4385460317134857,
      "learning_rate": 1.8632506115983605e-06,
      "loss": 0.5425,
      "step": 1135
    },
    {
      "epoch": 0.1937244201909959,
      "grad_norm": 0.37570932507514954,
      "learning_rate": 1.8629716826076845e-06,
      "loss": 0.5611,
      "step": 1136
    },
    {
      "epoch": 0.19389495225102318,
      "grad_norm": 0.46132612228393555,
      "learning_rate": 1.8626924903617202e-06,
      "loss": 0.5384,
      "step": 1137
    },
    {
      "epoch": 0.19406548431105047,
      "grad_norm": 0.4091589152812958,
      "learning_rate": 1.8624130349456371e-06,
      "loss": 0.5387,
      "step": 1138
    },
    {
      "epoch": 0.19423601637107776,
      "grad_norm": 0.4387233257293701,
      "learning_rate": 1.862133316444685e-06,
      "loss": 0.5389,
      "step": 1139
    },
    {
      "epoch": 0.19440654843110505,
      "grad_norm": 0.3781164884567261,
      "learning_rate": 1.861853334944194e-06,
      "loss": 0.531,
      "step": 1140
    },
    {
      "epoch": 0.19457708049113234,
      "grad_norm": 0.4493245780467987,
      "learning_rate": 1.8615730905295742e-06,
      "loss": 0.5426,
      "step": 1141
    },
    {
      "epoch": 0.1947476125511596,
      "grad_norm": 0.5568568110466003,
      "learning_rate": 1.861292583286316e-06,
      "loss": 0.5501,
      "step": 1142
    },
    {
      "epoch": 0.1949181446111869,
      "grad_norm": 0.47942033410072327,
      "learning_rate": 1.8610118132999899e-06,
      "loss": 0.532,
      "step": 1143
    },
    {
      "epoch": 0.19508867667121418,
      "grad_norm": 0.40977925062179565,
      "learning_rate": 1.860730780656247e-06,
      "loss": 0.5472,
      "step": 1144
    },
    {
      "epoch": 0.19525920873124147,
      "grad_norm": 0.5302245616912842,
      "learning_rate": 1.8604494854408178e-06,
      "loss": 0.5472,
      "step": 1145
    },
    {
      "epoch": 0.19542974079126876,
      "grad_norm": 0.5703160762786865,
      "learning_rate": 1.8601679277395136e-06,
      "loss": 0.5508,
      "step": 1146
    },
    {
      "epoch": 0.19560027285129605,
      "grad_norm": 0.4289019703865051,
      "learning_rate": 1.8598861076382254e-06,
      "loss": 0.5321,
      "step": 1147
    },
    {
      "epoch": 0.19577080491132332,
      "grad_norm": 0.4585319757461548,
      "learning_rate": 1.8596040252229243e-06,
      "loss": 0.5484,
      "step": 1148
    },
    {
      "epoch": 0.1959413369713506,
      "grad_norm": 0.5592920184135437,
      "learning_rate": 1.8593216805796613e-06,
      "loss": 0.5462,
      "step": 1149
    },
    {
      "epoch": 0.1961118690313779,
      "grad_norm": 0.40145450830459595,
      "learning_rate": 1.8590390737945678e-06,
      "loss": 0.5377,
      "step": 1150
    },
    {
      "epoch": 0.19628240109140518,
      "grad_norm": 0.48376187682151794,
      "learning_rate": 1.8587562049538542e-06,
      "loss": 0.5303,
      "step": 1151
    },
    {
      "epoch": 0.19645293315143247,
      "grad_norm": 0.6458383202552795,
      "learning_rate": 1.8584730741438129e-06,
      "loss": 0.5525,
      "step": 1152
    },
    {
      "epoch": 0.19662346521145976,
      "grad_norm": 0.45522838830947876,
      "learning_rate": 1.8581896814508134e-06,
      "loss": 0.5406,
      "step": 1153
    },
    {
      "epoch": 0.19679399727148705,
      "grad_norm": 0.5150303840637207,
      "learning_rate": 1.857906026961307e-06,
      "loss": 0.5334,
      "step": 1154
    },
    {
      "epoch": 0.19696452933151432,
      "grad_norm": 0.43621817231178284,
      "learning_rate": 1.8576221107618246e-06,
      "loss": 0.5431,
      "step": 1155
    },
    {
      "epoch": 0.1971350613915416,
      "grad_norm": 0.48908352851867676,
      "learning_rate": 1.8573379329389768e-06,
      "loss": 0.5586,
      "step": 1156
    },
    {
      "epoch": 0.1973055934515689,
      "grad_norm": 0.5652856230735779,
      "learning_rate": 1.857053493579454e-06,
      "loss": 0.54,
      "step": 1157
    },
    {
      "epoch": 0.19747612551159618,
      "grad_norm": 0.5483607649803162,
      "learning_rate": 1.8567687927700256e-06,
      "loss": 0.5388,
      "step": 1158
    },
    {
      "epoch": 0.19764665757162347,
      "grad_norm": 0.47800490260124207,
      "learning_rate": 1.8564838305975425e-06,
      "loss": 0.5464,
      "step": 1159
    },
    {
      "epoch": 0.19781718963165076,
      "grad_norm": 0.5164735317230225,
      "learning_rate": 1.8561986071489336e-06,
      "loss": 0.5366,
      "step": 1160
    },
    {
      "epoch": 0.19798772169167803,
      "grad_norm": 0.712042510509491,
      "learning_rate": 1.8559131225112088e-06,
      "loss": 0.5509,
      "step": 1161
    },
    {
      "epoch": 0.19815825375170532,
      "grad_norm": 0.5521159768104553,
      "learning_rate": 1.8556273767714565e-06,
      "loss": 0.5549,
      "step": 1162
    },
    {
      "epoch": 0.1983287858117326,
      "grad_norm": 0.4541512727737427,
      "learning_rate": 1.8553413700168456e-06,
      "loss": 0.5376,
      "step": 1163
    },
    {
      "epoch": 0.1984993178717599,
      "grad_norm": 0.4656938314437866,
      "learning_rate": 1.8550551023346246e-06,
      "loss": 0.5364,
      "step": 1164
    },
    {
      "epoch": 0.19866984993178718,
      "grad_norm": 0.5346999764442444,
      "learning_rate": 1.8547685738121208e-06,
      "loss": 0.5527,
      "step": 1165
    },
    {
      "epoch": 0.19884038199181447,
      "grad_norm": 0.5368838310241699,
      "learning_rate": 1.8544817845367423e-06,
      "loss": 0.5471,
      "step": 1166
    },
    {
      "epoch": 0.19901091405184174,
      "grad_norm": 0.6363966464996338,
      "learning_rate": 1.8541947345959757e-06,
      "loss": 0.5509,
      "step": 1167
    },
    {
      "epoch": 0.19918144611186903,
      "grad_norm": 0.4643917679786682,
      "learning_rate": 1.8539074240773877e-06,
      "loss": 0.5342,
      "step": 1168
    },
    {
      "epoch": 0.19935197817189632,
      "grad_norm": 0.49029770493507385,
      "learning_rate": 1.8536198530686241e-06,
      "loss": 0.5407,
      "step": 1169
    },
    {
      "epoch": 0.1995225102319236,
      "grad_norm": 0.592406690120697,
      "learning_rate": 1.8533320216574103e-06,
      "loss": 0.5459,
      "step": 1170
    },
    {
      "epoch": 0.1996930422919509,
      "grad_norm": 0.397834837436676,
      "learning_rate": 1.8530439299315517e-06,
      "loss": 0.5466,
      "step": 1171
    },
    {
      "epoch": 0.19986357435197818,
      "grad_norm": 0.5638540983200073,
      "learning_rate": 1.8527555779789323e-06,
      "loss": 0.5371,
      "step": 1172
    },
    {
      "epoch": 0.20003410641200545,
      "grad_norm": 0.42020824551582336,
      "learning_rate": 1.8524669658875154e-06,
      "loss": 0.5408,
      "step": 1173
    },
    {
      "epoch": 0.20020463847203274,
      "grad_norm": 0.5150777101516724,
      "learning_rate": 1.8521780937453446e-06,
      "loss": 0.5405,
      "step": 1174
    },
    {
      "epoch": 0.20037517053206003,
      "grad_norm": 0.5404800176620483,
      "learning_rate": 1.851888961640542e-06,
      "loss": 0.5397,
      "step": 1175
    },
    {
      "epoch": 0.20054570259208732,
      "grad_norm": 0.4149342477321625,
      "learning_rate": 1.8515995696613098e-06,
      "loss": 0.5353,
      "step": 1176
    },
    {
      "epoch": 0.2007162346521146,
      "grad_norm": 0.521208643913269,
      "learning_rate": 1.8513099178959282e-06,
      "loss": 0.5495,
      "step": 1177
    },
    {
      "epoch": 0.2008867667121419,
      "grad_norm": 0.4055381417274475,
      "learning_rate": 1.8510200064327577e-06,
      "loss": 0.5464,
      "step": 1178
    },
    {
      "epoch": 0.20105729877216916,
      "grad_norm": 0.439575731754303,
      "learning_rate": 1.8507298353602376e-06,
      "loss": 0.5343,
      "step": 1179
    },
    {
      "epoch": 0.20122783083219645,
      "grad_norm": 0.456194669008255,
      "learning_rate": 1.8504394047668872e-06,
      "loss": 0.5445,
      "step": 1180
    },
    {
      "epoch": 0.20139836289222374,
      "grad_norm": 0.4100000262260437,
      "learning_rate": 1.8501487147413035e-06,
      "loss": 0.5419,
      "step": 1181
    },
    {
      "epoch": 0.20156889495225103,
      "grad_norm": 0.41677936911582947,
      "learning_rate": 1.849857765372164e-06,
      "loss": 0.5643,
      "step": 1182
    },
    {
      "epoch": 0.20173942701227832,
      "grad_norm": 0.404525488615036,
      "learning_rate": 1.849566556748224e-06,
      "loss": 0.5447,
      "step": 1183
    },
    {
      "epoch": 0.2019099590723056,
      "grad_norm": 0.40742960572242737,
      "learning_rate": 1.8492750889583194e-06,
      "loss": 0.5253,
      "step": 1184
    },
    {
      "epoch": 0.20208049113233287,
      "grad_norm": 0.4676227867603302,
      "learning_rate": 1.8489833620913644e-06,
      "loss": 0.5362,
      "step": 1185
    },
    {
      "epoch": 0.20225102319236016,
      "grad_norm": 0.3785782754421234,
      "learning_rate": 1.8486913762363517e-06,
      "loss": 0.5448,
      "step": 1186
    },
    {
      "epoch": 0.20242155525238745,
      "grad_norm": 0.3994957506656647,
      "learning_rate": 1.8483991314823536e-06,
      "loss": 0.5359,
      "step": 1187
    },
    {
      "epoch": 0.20259208731241474,
      "grad_norm": 0.4070039391517639,
      "learning_rate": 1.848106627918522e-06,
      "loss": 0.5461,
      "step": 1188
    },
    {
      "epoch": 0.20276261937244203,
      "grad_norm": 0.5237683057785034,
      "learning_rate": 1.8478138656340862e-06,
      "loss": 0.5494,
      "step": 1189
    },
    {
      "epoch": 0.20293315143246932,
      "grad_norm": 0.3871895670890808,
      "learning_rate": 1.8475208447183558e-06,
      "loss": 0.5684,
      "step": 1190
    },
    {
      "epoch": 0.20310368349249658,
      "grad_norm": 0.4152975082397461,
      "learning_rate": 1.8472275652607187e-06,
      "loss": 0.5337,
      "step": 1191
    },
    {
      "epoch": 0.20327421555252387,
      "grad_norm": 0.44173455238342285,
      "learning_rate": 1.846934027350642e-06,
      "loss": 0.5474,
      "step": 1192
    },
    {
      "epoch": 0.20344474761255116,
      "grad_norm": 0.4573276937007904,
      "learning_rate": 1.846640231077671e-06,
      "loss": 0.5465,
      "step": 1193
    },
    {
      "epoch": 0.20361527967257845,
      "grad_norm": 0.3855762779712677,
      "learning_rate": 1.8463461765314302e-06,
      "loss": 0.5494,
      "step": 1194
    },
    {
      "epoch": 0.20378581173260574,
      "grad_norm": 0.4517650306224823,
      "learning_rate": 1.846051863801623e-06,
      "loss": 0.5424,
      "step": 1195
    },
    {
      "epoch": 0.20395634379263303,
      "grad_norm": 0.5178718566894531,
      "learning_rate": 1.8457572929780317e-06,
      "loss": 0.5361,
      "step": 1196
    },
    {
      "epoch": 0.2041268758526603,
      "grad_norm": 0.3623640835285187,
      "learning_rate": 1.845462464150517e-06,
      "loss": 0.5471,
      "step": 1197
    },
    {
      "epoch": 0.20429740791268758,
      "grad_norm": 0.4305116832256317,
      "learning_rate": 1.8451673774090185e-06,
      "loss": 0.5439,
      "step": 1198
    },
    {
      "epoch": 0.20446793997271487,
      "grad_norm": 0.4251005947589874,
      "learning_rate": 1.844872032843554e-06,
      "loss": 0.5418,
      "step": 1199
    },
    {
      "epoch": 0.20463847203274216,
      "grad_norm": 0.40618592500686646,
      "learning_rate": 1.8445764305442205e-06,
      "loss": 0.5442,
      "step": 1200
    },
    {
      "epoch": 0.20480900409276945,
      "grad_norm": 0.4994547963142395,
      "learning_rate": 1.8442805706011938e-06,
      "loss": 0.5606,
      "step": 1201
    },
    {
      "epoch": 0.20497953615279674,
      "grad_norm": 0.38456836342811584,
      "learning_rate": 1.8439844531047277e-06,
      "loss": 0.5436,
      "step": 1202
    },
    {
      "epoch": 0.205150068212824,
      "grad_norm": 0.3920612335205078,
      "learning_rate": 1.8436880781451547e-06,
      "loss": 0.5288,
      "step": 1203
    },
    {
      "epoch": 0.2053206002728513,
      "grad_norm": 0.38321173191070557,
      "learning_rate": 1.843391445812886e-06,
      "loss": 0.5412,
      "step": 1204
    },
    {
      "epoch": 0.20549113233287858,
      "grad_norm": 0.4655633568763733,
      "learning_rate": 1.8430945561984115e-06,
      "loss": 0.5461,
      "step": 1205
    },
    {
      "epoch": 0.20566166439290587,
      "grad_norm": 0.404006689786911,
      "learning_rate": 1.842797409392299e-06,
      "loss": 0.5531,
      "step": 1206
    },
    {
      "epoch": 0.20583219645293316,
      "grad_norm": 0.4156908094882965,
      "learning_rate": 1.8425000054851956e-06,
      "loss": 0.5405,
      "step": 1207
    },
    {
      "epoch": 0.20600272851296045,
      "grad_norm": 0.4246964156627655,
      "learning_rate": 1.842202344567826e-06,
      "loss": 0.5354,
      "step": 1208
    },
    {
      "epoch": 0.2061732605729877,
      "grad_norm": 0.3740110993385315,
      "learning_rate": 1.841904426730994e-06,
      "loss": 0.5334,
      "step": 1209
    },
    {
      "epoch": 0.206343792633015,
      "grad_norm": 0.40257853269577026,
      "learning_rate": 1.8416062520655808e-06,
      "loss": 0.5448,
      "step": 1210
    },
    {
      "epoch": 0.2065143246930423,
      "grad_norm": 0.42841023206710815,
      "learning_rate": 1.841307820662547e-06,
      "loss": 0.5467,
      "step": 1211
    },
    {
      "epoch": 0.20668485675306958,
      "grad_norm": 0.4049186408519745,
      "learning_rate": 1.8410091326129314e-06,
      "loss": 0.5371,
      "step": 1212
    },
    {
      "epoch": 0.20685538881309687,
      "grad_norm": 0.44290080666542053,
      "learning_rate": 1.8407101880078502e-06,
      "loss": 0.5429,
      "step": 1213
    },
    {
      "epoch": 0.20702592087312416,
      "grad_norm": 0.3257935643196106,
      "learning_rate": 1.8404109869384988e-06,
      "loss": 0.5436,
      "step": 1214
    },
    {
      "epoch": 0.20719645293315142,
      "grad_norm": 0.410832941532135,
      "learning_rate": 1.8401115294961501e-06,
      "loss": 0.5453,
      "step": 1215
    },
    {
      "epoch": 0.2073669849931787,
      "grad_norm": 0.3964151442050934,
      "learning_rate": 1.839811815772156e-06,
      "loss": 0.5411,
      "step": 1216
    },
    {
      "epoch": 0.207537517053206,
      "grad_norm": 0.3891664147377014,
      "learning_rate": 1.8395118458579461e-06,
      "loss": 0.5293,
      "step": 1217
    },
    {
      "epoch": 0.2077080491132333,
      "grad_norm": 0.4442242681980133,
      "learning_rate": 1.8392116198450285e-06,
      "loss": 0.5433,
      "step": 1218
    },
    {
      "epoch": 0.20787858117326058,
      "grad_norm": 0.4433388411998749,
      "learning_rate": 1.8389111378249886e-06,
      "loss": 0.5403,
      "step": 1219
    },
    {
      "epoch": 0.20804911323328787,
      "grad_norm": 0.3714314103126526,
      "learning_rate": 1.8386103998894909e-06,
      "loss": 0.5255,
      "step": 1220
    },
    {
      "epoch": 0.20821964529331513,
      "grad_norm": 0.4090864062309265,
      "learning_rate": 1.8383094061302767e-06,
      "loss": 0.5419,
      "step": 1221
    },
    {
      "epoch": 0.20839017735334242,
      "grad_norm": 0.4515744745731354,
      "learning_rate": 1.8380081566391674e-06,
      "loss": 0.5469,
      "step": 1222
    },
    {
      "epoch": 0.2085607094133697,
      "grad_norm": 0.46093738079071045,
      "learning_rate": 1.8377066515080606e-06,
      "loss": 0.5332,
      "step": 1223
    },
    {
      "epoch": 0.208731241473397,
      "grad_norm": 0.4779576361179352,
      "learning_rate": 1.837404890828932e-06,
      "loss": 0.5405,
      "step": 1224
    },
    {
      "epoch": 0.2089017735334243,
      "grad_norm": 0.4057524502277374,
      "learning_rate": 1.837102874693836e-06,
      "loss": 0.5383,
      "step": 1225
    },
    {
      "epoch": 0.20907230559345158,
      "grad_norm": 0.4065392315387726,
      "learning_rate": 1.8368006031949054e-06,
      "loss": 0.5408,
      "step": 1226
    },
    {
      "epoch": 0.20924283765347884,
      "grad_norm": 0.4654337167739868,
      "learning_rate": 1.836498076424349e-06,
      "loss": 0.5321,
      "step": 1227
    },
    {
      "epoch": 0.20941336971350613,
      "grad_norm": 0.4152114689350128,
      "learning_rate": 1.8361952944744555e-06,
      "loss": 0.5488,
      "step": 1228
    },
    {
      "epoch": 0.20958390177353342,
      "grad_norm": 0.35753628611564636,
      "learning_rate": 1.83589225743759e-06,
      "loss": 0.5408,
      "step": 1229
    },
    {
      "epoch": 0.2097544338335607,
      "grad_norm": 0.4003051519393921,
      "learning_rate": 1.8355889654061963e-06,
      "loss": 0.5397,
      "step": 1230
    },
    {
      "epoch": 0.209924965893588,
      "grad_norm": 0.422453910112381,
      "learning_rate": 1.8352854184727954e-06,
      "loss": 0.5361,
      "step": 1231
    },
    {
      "epoch": 0.2100954979536153,
      "grad_norm": 0.4182709753513336,
      "learning_rate": 1.8349816167299866e-06,
      "loss": 0.5346,
      "step": 1232
    },
    {
      "epoch": 0.21026603001364255,
      "grad_norm": 0.42172756791114807,
      "learning_rate": 1.8346775602704465e-06,
      "loss": 0.54,
      "step": 1233
    },
    {
      "epoch": 0.21043656207366984,
      "grad_norm": 0.547315776348114,
      "learning_rate": 1.8343732491869294e-06,
      "loss": 0.5373,
      "step": 1234
    },
    {
      "epoch": 0.21060709413369713,
      "grad_norm": 0.48611629009246826,
      "learning_rate": 1.834068683572268e-06,
      "loss": 0.5476,
      "step": 1235
    },
    {
      "epoch": 0.21077762619372442,
      "grad_norm": 0.45138785243034363,
      "learning_rate": 1.8337638635193712e-06,
      "loss": 0.5395,
      "step": 1236
    },
    {
      "epoch": 0.2109481582537517,
      "grad_norm": 0.37378039956092834,
      "learning_rate": 1.8334587891212272e-06,
      "loss": 0.5449,
      "step": 1237
    },
    {
      "epoch": 0.211118690313779,
      "grad_norm": 0.41004690527915955,
      "learning_rate": 1.8331534604709002e-06,
      "loss": 0.54,
      "step": 1238
    },
    {
      "epoch": 0.21128922237380626,
      "grad_norm": 0.5197843313217163,
      "learning_rate": 1.8328478776615336e-06,
      "loss": 0.5211,
      "step": 1239
    },
    {
      "epoch": 0.21145975443383355,
      "grad_norm": 0.46441686153411865,
      "learning_rate": 1.8325420407863471e-06,
      "loss": 0.5462,
      "step": 1240
    },
    {
      "epoch": 0.21163028649386084,
      "grad_norm": 0.44501668214797974,
      "learning_rate": 1.8322359499386384e-06,
      "loss": 0.5453,
      "step": 1241
    },
    {
      "epoch": 0.21180081855388813,
      "grad_norm": 0.42854249477386475,
      "learning_rate": 1.8319296052117823e-06,
      "loss": 0.5192,
      "step": 1242
    },
    {
      "epoch": 0.21197135061391542,
      "grad_norm": 0.5542727708816528,
      "learning_rate": 1.8316230066992314e-06,
      "loss": 0.5319,
      "step": 1243
    },
    {
      "epoch": 0.2121418826739427,
      "grad_norm": 0.5215505957603455,
      "learning_rate": 1.831316154494516e-06,
      "loss": 0.5465,
      "step": 1244
    },
    {
      "epoch": 0.21231241473396997,
      "grad_norm": 0.4085174798965454,
      "learning_rate": 1.8310090486912425e-06,
      "loss": 0.5434,
      "step": 1245
    },
    {
      "epoch": 0.21248294679399726,
      "grad_norm": 0.477775514125824,
      "learning_rate": 1.8307016893830967e-06,
      "loss": 0.5421,
      "step": 1246
    },
    {
      "epoch": 0.21265347885402455,
      "grad_norm": 0.3507140576839447,
      "learning_rate": 1.8303940766638398e-06,
      "loss": 0.5407,
      "step": 1247
    },
    {
      "epoch": 0.21282401091405184,
      "grad_norm": 0.47276178002357483,
      "learning_rate": 1.8300862106273116e-06,
      "loss": 0.539,
      "step": 1248
    },
    {
      "epoch": 0.21299454297407913,
      "grad_norm": 0.3437632918357849,
      "learning_rate": 1.8297780913674288e-06,
      "loss": 0.5411,
      "step": 1249
    },
    {
      "epoch": 0.21316507503410642,
      "grad_norm": 0.38725805282592773,
      "learning_rate": 1.8294697189781844e-06,
      "loss": 0.5454,
      "step": 1250
    },
    {
      "epoch": 0.21333560709413368,
      "grad_norm": 0.4623619019985199,
      "learning_rate": 1.8291610935536504e-06,
      "loss": 0.5397,
      "step": 1251
    },
    {
      "epoch": 0.21350613915416097,
      "grad_norm": 0.4034901559352875,
      "learning_rate": 1.8288522151879744e-06,
      "loss": 0.5383,
      "step": 1252
    },
    {
      "epoch": 0.21367667121418826,
      "grad_norm": 0.4366649389266968,
      "learning_rate": 1.828543083975382e-06,
      "loss": 0.5511,
      "step": 1253
    },
    {
      "epoch": 0.21384720327421555,
      "grad_norm": 0.4827879071235657,
      "learning_rate": 1.8282337000101764e-06,
      "loss": 0.5471,
      "step": 1254
    },
    {
      "epoch": 0.21401773533424284,
      "grad_norm": 0.5037616491317749,
      "learning_rate": 1.827924063386736e-06,
      "loss": 0.5346,
      "step": 1255
    },
    {
      "epoch": 0.21418826739427013,
      "grad_norm": 0.401431679725647,
      "learning_rate": 1.8276141741995187e-06,
      "loss": 0.5498,
      "step": 1256
    },
    {
      "epoch": 0.21435879945429742,
      "grad_norm": 0.4014277756214142,
      "learning_rate": 1.8273040325430577e-06,
      "loss": 0.552,
      "step": 1257
    },
    {
      "epoch": 0.21452933151432468,
      "grad_norm": 0.39532434940338135,
      "learning_rate": 1.8269936385119635e-06,
      "loss": 0.5521,
      "step": 1258
    },
    {
      "epoch": 0.21469986357435197,
      "grad_norm": 0.3930286467075348,
      "learning_rate": 1.8266829922009248e-06,
      "loss": 0.5446,
      "step": 1259
    },
    {
      "epoch": 0.21487039563437926,
      "grad_norm": 0.414559543132782,
      "learning_rate": 1.8263720937047052e-06,
      "loss": 0.5623,
      "step": 1260
    },
    {
      "epoch": 0.21504092769440655,
      "grad_norm": 0.5679807662963867,
      "learning_rate": 1.8260609431181472e-06,
      "loss": 0.5411,
      "step": 1261
    },
    {
      "epoch": 0.21521145975443384,
      "grad_norm": 0.381727010011673,
      "learning_rate": 1.8257495405361694e-06,
      "loss": 0.5386,
      "step": 1262
    },
    {
      "epoch": 0.21538199181446113,
      "grad_norm": 0.5208612680435181,
      "learning_rate": 1.8254378860537666e-06,
      "loss": 0.5475,
      "step": 1263
    },
    {
      "epoch": 0.2155525238744884,
      "grad_norm": 0.44777414202690125,
      "learning_rate": 1.8251259797660115e-06,
      "loss": 0.53,
      "step": 1264
    },
    {
      "epoch": 0.21572305593451568,
      "grad_norm": 0.38557693362236023,
      "learning_rate": 1.8248138217680532e-06,
      "loss": 0.5426,
      "step": 1265
    },
    {
      "epoch": 0.21589358799454297,
      "grad_norm": 0.6629211902618408,
      "learning_rate": 1.8245014121551177e-06,
      "loss": 0.5521,
      "step": 1266
    },
    {
      "epoch": 0.21606412005457026,
      "grad_norm": 0.6797375082969666,
      "learning_rate": 1.8241887510225071e-06,
      "loss": 0.5662,
      "step": 1267
    },
    {
      "epoch": 0.21623465211459755,
      "grad_norm": 0.5131363868713379,
      "learning_rate": 1.8238758384656011e-06,
      "loss": 0.5426,
      "step": 1268
    },
    {
      "epoch": 0.21640518417462484,
      "grad_norm": 0.4549805223941803,
      "learning_rate": 1.8235626745798562e-06,
      "loss": 0.5389,
      "step": 1269
    },
    {
      "epoch": 0.2165757162346521,
      "grad_norm": 0.6096103191375732,
      "learning_rate": 1.8232492594608044e-06,
      "loss": 0.5359,
      "step": 1270
    },
    {
      "epoch": 0.2167462482946794,
      "grad_norm": 0.5440108776092529,
      "learning_rate": 1.8229355932040557e-06,
      "loss": 0.542,
      "step": 1271
    },
    {
      "epoch": 0.21691678035470668,
      "grad_norm": 0.38566699624061584,
      "learning_rate": 1.8226216759052958e-06,
      "loss": 0.5391,
      "step": 1272
    },
    {
      "epoch": 0.21708731241473397,
      "grad_norm": 0.503471314907074,
      "learning_rate": 1.8223075076602873e-06,
      "loss": 0.5427,
      "step": 1273
    },
    {
      "epoch": 0.21725784447476126,
      "grad_norm": 0.4962663948535919,
      "learning_rate": 1.8219930885648691e-06,
      "loss": 0.54,
      "step": 1274
    },
    {
      "epoch": 0.21742837653478855,
      "grad_norm": 0.417920857667923,
      "learning_rate": 1.8216784187149571e-06,
      "loss": 0.5368,
      "step": 1275
    },
    {
      "epoch": 0.21759890859481582,
      "grad_norm": 0.34477683901786804,
      "learning_rate": 1.8213634982065437e-06,
      "loss": 0.5392,
      "step": 1276
    },
    {
      "epoch": 0.2177694406548431,
      "grad_norm": 0.3951910436153412,
      "learning_rate": 1.8210483271356973e-06,
      "loss": 0.5498,
      "step": 1277
    },
    {
      "epoch": 0.2179399727148704,
      "grad_norm": 0.4015583097934723,
      "learning_rate": 1.8207329055985624e-06,
      "loss": 0.5371,
      "step": 1278
    },
    {
      "epoch": 0.21811050477489768,
      "grad_norm": 0.3596416115760803,
      "learning_rate": 1.8204172336913611e-06,
      "loss": 0.535,
      "step": 1279
    },
    {
      "epoch": 0.21828103683492497,
      "grad_norm": 0.3580999970436096,
      "learning_rate": 1.8201013115103913e-06,
      "loss": 0.547,
      "step": 1280
    },
    {
      "epoch": 0.21845156889495226,
      "grad_norm": 0.3621145486831665,
      "learning_rate": 1.8197851391520267e-06,
      "loss": 0.5491,
      "step": 1281
    },
    {
      "epoch": 0.21862210095497953,
      "grad_norm": 0.4546815752983093,
      "learning_rate": 1.819468716712718e-06,
      "loss": 0.5493,
      "step": 1282
    },
    {
      "epoch": 0.21879263301500682,
      "grad_norm": 0.37099504470825195,
      "learning_rate": 1.8191520442889922e-06,
      "loss": 0.5512,
      "step": 1283
    },
    {
      "epoch": 0.2189631650750341,
      "grad_norm": 0.444868803024292,
      "learning_rate": 1.818835121977452e-06,
      "loss": 0.5492,
      "step": 1284
    },
    {
      "epoch": 0.2191336971350614,
      "grad_norm": 0.46764564514160156,
      "learning_rate": 1.818517949874777e-06,
      "loss": 0.5284,
      "step": 1285
    },
    {
      "epoch": 0.21930422919508868,
      "grad_norm": 0.48628371953964233,
      "learning_rate": 1.818200528077722e-06,
      "loss": 0.5249,
      "step": 1286
    },
    {
      "epoch": 0.21947476125511597,
      "grad_norm": 0.5340228080749512,
      "learning_rate": 1.8178828566831194e-06,
      "loss": 0.5378,
      "step": 1287
    },
    {
      "epoch": 0.21964529331514324,
      "grad_norm": 0.4188104569911957,
      "learning_rate": 1.8175649357878767e-06,
      "loss": 0.5404,
      "step": 1288
    },
    {
      "epoch": 0.21981582537517053,
      "grad_norm": 0.5152527093887329,
      "learning_rate": 1.8172467654889782e-06,
      "loss": 0.538,
      "step": 1289
    },
    {
      "epoch": 0.21998635743519782,
      "grad_norm": 0.7427700161933899,
      "learning_rate": 1.8169283458834832e-06,
      "loss": 0.547,
      "step": 1290
    },
    {
      "epoch": 0.2201568894952251,
      "grad_norm": 0.5649572610855103,
      "learning_rate": 1.8166096770685279e-06,
      "loss": 0.5332,
      "step": 1291
    },
    {
      "epoch": 0.2203274215552524,
      "grad_norm": 0.46063774824142456,
      "learning_rate": 1.8162907591413248e-06,
      "loss": 0.5572,
      "step": 1292
    },
    {
      "epoch": 0.22049795361527968,
      "grad_norm": 0.5200372934341431,
      "learning_rate": 1.8159715921991613e-06,
      "loss": 0.5475,
      "step": 1293
    },
    {
      "epoch": 0.22066848567530695,
      "grad_norm": 0.5936065316200256,
      "learning_rate": 1.8156521763394019e-06,
      "loss": 0.5578,
      "step": 1294
    },
    {
      "epoch": 0.22083901773533424,
      "grad_norm": 0.430730402469635,
      "learning_rate": 1.8153325116594864e-06,
      "loss": 0.5383,
      "step": 1295
    },
    {
      "epoch": 0.22100954979536153,
      "grad_norm": 0.5382259488105774,
      "learning_rate": 1.8150125982569307e-06,
      "loss": 0.5297,
      "step": 1296
    },
    {
      "epoch": 0.22118008185538882,
      "grad_norm": 0.6202133297920227,
      "learning_rate": 1.8146924362293262e-06,
      "loss": 0.5409,
      "step": 1297
    },
    {
      "epoch": 0.2213506139154161,
      "grad_norm": 0.445716917514801,
      "learning_rate": 1.8143720256743411e-06,
      "loss": 0.5378,
      "step": 1298
    },
    {
      "epoch": 0.2215211459754434,
      "grad_norm": 0.46867361664772034,
      "learning_rate": 1.814051366689718e-06,
      "loss": 0.5441,
      "step": 1299
    },
    {
      "epoch": 0.22169167803547066,
      "grad_norm": 0.5404912829399109,
      "learning_rate": 1.813730459373277e-06,
      "loss": 0.5589,
      "step": 1300
    },
    {
      "epoch": 0.22186221009549795,
      "grad_norm": 0.4060252606868744,
      "learning_rate": 1.813409303822912e-06,
      "loss": 0.5414,
      "step": 1301
    },
    {
      "epoch": 0.22203274215552524,
      "grad_norm": 0.6208741664886475,
      "learning_rate": 1.8130879001365946e-06,
      "loss": 0.5467,
      "step": 1302
    },
    {
      "epoch": 0.22220327421555253,
      "grad_norm": 0.5801400542259216,
      "learning_rate": 1.8127662484123704e-06,
      "loss": 0.5449,
      "step": 1303
    },
    {
      "epoch": 0.22237380627557982,
      "grad_norm": 0.47542262077331543,
      "learning_rate": 1.812444348748362e-06,
      "loss": 0.5361,
      "step": 1304
    },
    {
      "epoch": 0.2225443383356071,
      "grad_norm": 0.5469897389411926,
      "learning_rate": 1.8121222012427668e-06,
      "loss": 0.5364,
      "step": 1305
    },
    {
      "epoch": 0.22271487039563437,
      "grad_norm": 0.477078914642334,
      "learning_rate": 1.8117998059938576e-06,
      "loss": 0.5367,
      "step": 1306
    },
    {
      "epoch": 0.22288540245566166,
      "grad_norm": 0.5777487754821777,
      "learning_rate": 1.811477163099984e-06,
      "loss": 0.5258,
      "step": 1307
    },
    {
      "epoch": 0.22305593451568895,
      "grad_norm": 0.5522773861885071,
      "learning_rate": 1.81115427265957e-06,
      "loss": 0.5483,
      "step": 1308
    },
    {
      "epoch": 0.22322646657571624,
      "grad_norm": 0.4026612341403961,
      "learning_rate": 1.8108311347711152e-06,
      "loss": 0.5373,
      "step": 1309
    },
    {
      "epoch": 0.22339699863574353,
      "grad_norm": 0.6066965460777283,
      "learning_rate": 1.8105077495331957e-06,
      "loss": 0.5359,
      "step": 1310
    },
    {
      "epoch": 0.22356753069577082,
      "grad_norm": 0.4083956480026245,
      "learning_rate": 1.8101841170444617e-06,
      "loss": 0.5222,
      "step": 1311
    },
    {
      "epoch": 0.22373806275579808,
      "grad_norm": 0.46423399448394775,
      "learning_rate": 1.8098602374036397e-06,
      "loss": 0.553,
      "step": 1312
    },
    {
      "epoch": 0.22390859481582537,
      "grad_norm": 0.5001481175422668,
      "learning_rate": 1.8095361107095315e-06,
      "loss": 0.5369,
      "step": 1313
    },
    {
      "epoch": 0.22407912687585266,
      "grad_norm": 0.351531058549881,
      "learning_rate": 1.8092117370610138e-06,
      "loss": 0.5366,
      "step": 1314
    },
    {
      "epoch": 0.22424965893587995,
      "grad_norm": 0.5711095929145813,
      "learning_rate": 1.8088871165570392e-06,
      "loss": 0.5398,
      "step": 1315
    },
    {
      "epoch": 0.22442019099590724,
      "grad_norm": 0.5152797698974609,
      "learning_rate": 1.808562249296635e-06,
      "loss": 0.5287,
      "step": 1316
    },
    {
      "epoch": 0.22459072305593453,
      "grad_norm": 0.4698987603187561,
      "learning_rate": 1.8082371353789048e-06,
      "loss": 0.5343,
      "step": 1317
    },
    {
      "epoch": 0.2247612551159618,
      "grad_norm": 0.5477645993232727,
      "learning_rate": 1.8079117749030259e-06,
      "loss": 0.5405,
      "step": 1318
    },
    {
      "epoch": 0.22493178717598908,
      "grad_norm": 0.5242092609405518,
      "learning_rate": 1.8075861679682526e-06,
      "loss": 0.5324,
      "step": 1319
    },
    {
      "epoch": 0.22510231923601637,
      "grad_norm": 0.45128878951072693,
      "learning_rate": 1.8072603146739128e-06,
      "loss": 0.5498,
      "step": 1320
    },
    {
      "epoch": 0.22527285129604366,
      "grad_norm": 0.5173705220222473,
      "learning_rate": 1.8069342151194106e-06,
      "loss": 0.5331,
      "step": 1321
    },
    {
      "epoch": 0.22544338335607095,
      "grad_norm": 0.5540408492088318,
      "learning_rate": 1.8066078694042245e-06,
      "loss": 0.5348,
      "step": 1322
    },
    {
      "epoch": 0.22561391541609824,
      "grad_norm": 0.38541969656944275,
      "learning_rate": 1.8062812776279088e-06,
      "loss": 0.5237,
      "step": 1323
    },
    {
      "epoch": 0.2257844474761255,
      "grad_norm": 0.5497429966926575,
      "learning_rate": 1.8059544398900925e-06,
      "loss": 0.5488,
      "step": 1324
    },
    {
      "epoch": 0.2259549795361528,
      "grad_norm": 0.5218777656555176,
      "learning_rate": 1.8056273562904792e-06,
      "loss": 0.5442,
      "step": 1325
    },
    {
      "epoch": 0.22612551159618008,
      "grad_norm": 0.37961360812187195,
      "learning_rate": 1.8053000269288487e-06,
      "loss": 0.5404,
      "step": 1326
    },
    {
      "epoch": 0.22629604365620737,
      "grad_norm": 0.5503019690513611,
      "learning_rate": 1.8049724519050538e-06,
      "loss": 0.5262,
      "step": 1327
    },
    {
      "epoch": 0.22646657571623466,
      "grad_norm": 0.510767936706543,
      "learning_rate": 1.804644631319025e-06,
      "loss": 0.5468,
      "step": 1328
    },
    {
      "epoch": 0.22663710777626195,
      "grad_norm": 0.4793156683444977,
      "learning_rate": 1.8043165652707652e-06,
      "loss": 0.5402,
      "step": 1329
    },
    {
      "epoch": 0.2268076398362892,
      "grad_norm": 0.40747585892677307,
      "learning_rate": 1.803988253860353e-06,
      "loss": 0.5397,
      "step": 1330
    },
    {
      "epoch": 0.2269781718963165,
      "grad_norm": 0.440603643655777,
      "learning_rate": 1.8036596971879426e-06,
      "loss": 0.5273,
      "step": 1331
    },
    {
      "epoch": 0.2271487039563438,
      "grad_norm": 0.5874107480049133,
      "learning_rate": 1.803330895353762e-06,
      "loss": 0.5544,
      "step": 1332
    },
    {
      "epoch": 0.22731923601637108,
      "grad_norm": 0.5229067802429199,
      "learning_rate": 1.8030018484581149e-06,
      "loss": 0.5341,
      "step": 1333
    },
    {
      "epoch": 0.22748976807639837,
      "grad_norm": 0.350911945104599,
      "learning_rate": 1.8026725566013785e-06,
      "loss": 0.5396,
      "step": 1334
    },
    {
      "epoch": 0.22766030013642566,
      "grad_norm": 0.5130938291549683,
      "learning_rate": 1.802343019884006e-06,
      "loss": 0.5373,
      "step": 1335
    },
    {
      "epoch": 0.22783083219645292,
      "grad_norm": 0.6306084990501404,
      "learning_rate": 1.8020132384065244e-06,
      "loss": 0.5329,
      "step": 1336
    },
    {
      "epoch": 0.2280013642564802,
      "grad_norm": 0.5357939600944519,
      "learning_rate": 1.8016832122695362e-06,
      "loss": 0.5496,
      "step": 1337
    },
    {
      "epoch": 0.2281718963165075,
      "grad_norm": 0.3664228022098541,
      "learning_rate": 1.801352941573718e-06,
      "loss": 0.531,
      "step": 1338
    },
    {
      "epoch": 0.2283424283765348,
      "grad_norm": 0.5649370551109314,
      "learning_rate": 1.8010224264198207e-06,
      "loss": 0.5354,
      "step": 1339
    },
    {
      "epoch": 0.22851296043656208,
      "grad_norm": 0.6709194779396057,
      "learning_rate": 1.800691666908671e-06,
      "loss": 0.5418,
      "step": 1340
    },
    {
      "epoch": 0.22868349249658937,
      "grad_norm": 0.441829115152359,
      "learning_rate": 1.800360663141168e-06,
      "loss": 0.5525,
      "step": 1341
    },
    {
      "epoch": 0.22885402455661663,
      "grad_norm": 0.38027212023735046,
      "learning_rate": 1.8000294152182878e-06,
      "loss": 0.5368,
      "step": 1342
    },
    {
      "epoch": 0.22902455661664392,
      "grad_norm": 0.39649662375450134,
      "learning_rate": 1.799697923241079e-06,
      "loss": 0.5308,
      "step": 1343
    },
    {
      "epoch": 0.2291950886766712,
      "grad_norm": 0.39250436425209045,
      "learning_rate": 1.7993661873106662e-06,
      "loss": 0.5328,
      "step": 1344
    },
    {
      "epoch": 0.2293656207366985,
      "grad_norm": 0.37690094113349915,
      "learning_rate": 1.7990342075282472e-06,
      "loss": 0.5258,
      "step": 1345
    },
    {
      "epoch": 0.2295361527967258,
      "grad_norm": 0.4618053436279297,
      "learning_rate": 1.7987019839950944e-06,
      "loss": 0.5412,
      "step": 1346
    },
    {
      "epoch": 0.22970668485675308,
      "grad_norm": 0.47363170981407166,
      "learning_rate": 1.7983695168125553e-06,
      "loss": 0.5292,
      "step": 1347
    },
    {
      "epoch": 0.22987721691678034,
      "grad_norm": 0.44554126262664795,
      "learning_rate": 1.7980368060820511e-06,
      "loss": 0.5278,
      "step": 1348
    },
    {
      "epoch": 0.23004774897680763,
      "grad_norm": 0.49328240752220154,
      "learning_rate": 1.7977038519050774e-06,
      "loss": 0.5424,
      "step": 1349
    },
    {
      "epoch": 0.23021828103683492,
      "grad_norm": 0.4327653646469116,
      "learning_rate": 1.797370654383204e-06,
      "loss": 0.543,
      "step": 1350
    },
    {
      "epoch": 0.2303888130968622,
      "grad_norm": 0.5031704902648926,
      "learning_rate": 1.7970372136180756e-06,
      "loss": 0.5332,
      "step": 1351
    },
    {
      "epoch": 0.2305593451568895,
      "grad_norm": 0.5221416354179382,
      "learning_rate": 1.7967035297114095e-06,
      "loss": 0.5275,
      "step": 1352
    },
    {
      "epoch": 0.2307298772169168,
      "grad_norm": 0.35783857107162476,
      "learning_rate": 1.796369602764999e-06,
      "loss": 0.5364,
      "step": 1353
    },
    {
      "epoch": 0.23090040927694405,
      "grad_norm": 0.4733849763870239,
      "learning_rate": 1.7960354328807106e-06,
      "loss": 0.5288,
      "step": 1354
    },
    {
      "epoch": 0.23107094133697134,
      "grad_norm": 0.3745383620262146,
      "learning_rate": 1.7957010201604847e-06,
      "loss": 0.5447,
      "step": 1355
    },
    {
      "epoch": 0.23124147339699863,
      "grad_norm": 0.4308125078678131,
      "learning_rate": 1.7953663647063365e-06,
      "loss": 0.5425,
      "step": 1356
    },
    {
      "epoch": 0.23141200545702592,
      "grad_norm": 0.4075262248516083,
      "learning_rate": 1.795031466620355e-06,
      "loss": 0.5344,
      "step": 1357
    },
    {
      "epoch": 0.2315825375170532,
      "grad_norm": 0.43857333064079285,
      "learning_rate": 1.794696326004703e-06,
      "loss": 0.5375,
      "step": 1358
    },
    {
      "epoch": 0.2317530695770805,
      "grad_norm": 0.5499160885810852,
      "learning_rate": 1.7943609429616174e-06,
      "loss": 0.5464,
      "step": 1359
    },
    {
      "epoch": 0.23192360163710776,
      "grad_norm": 0.5283775329589844,
      "learning_rate": 1.7940253175934086e-06,
      "loss": 0.5405,
      "step": 1360
    },
    {
      "epoch": 0.23209413369713505,
      "grad_norm": 0.558288037776947,
      "learning_rate": 1.7936894500024616e-06,
      "loss": 0.5373,
      "step": 1361
    },
    {
      "epoch": 0.23226466575716234,
      "grad_norm": 0.39163830876350403,
      "learning_rate": 1.7933533402912355e-06,
      "loss": 0.5412,
      "step": 1362
    },
    {
      "epoch": 0.23243519781718963,
      "grad_norm": 0.5775661468505859,
      "learning_rate": 1.7930169885622622e-06,
      "loss": 0.527,
      "step": 1363
    },
    {
      "epoch": 0.23260572987721692,
      "grad_norm": 0.5608419179916382,
      "learning_rate": 1.7926803949181486e-06,
      "loss": 0.5332,
      "step": 1364
    },
    {
      "epoch": 0.2327762619372442,
      "grad_norm": 0.4189216196537018,
      "learning_rate": 1.7923435594615744e-06,
      "loss": 0.5418,
      "step": 1365
    },
    {
      "epoch": 0.2329467939972715,
      "grad_norm": 0.5751291513442993,
      "learning_rate": 1.792006482295294e-06,
      "loss": 0.539,
      "step": 1366
    },
    {
      "epoch": 0.23311732605729876,
      "grad_norm": 0.4839267134666443,
      "learning_rate": 1.7916691635221346e-06,
      "loss": 0.5465,
      "step": 1367
    },
    {
      "epoch": 0.23328785811732605,
      "grad_norm": 0.4686916768550873,
      "learning_rate": 1.7913316032449976e-06,
      "loss": 0.5534,
      "step": 1368
    },
    {
      "epoch": 0.23345839017735334,
      "grad_norm": 0.4905444383621216,
      "learning_rate": 1.7909938015668581e-06,
      "loss": 0.5419,
      "step": 1369
    },
    {
      "epoch": 0.23362892223738063,
      "grad_norm": 0.5154764652252197,
      "learning_rate": 1.790655758590765e-06,
      "loss": 0.5481,
      "step": 1370
    },
    {
      "epoch": 0.23379945429740792,
      "grad_norm": 0.4774230122566223,
      "learning_rate": 1.79031747441984e-06,
      "loss": 0.5269,
      "step": 1371
    },
    {
      "epoch": 0.2339699863574352,
      "grad_norm": 0.3737574815750122,
      "learning_rate": 1.7899789491572794e-06,
      "loss": 0.5326,
      "step": 1372
    },
    {
      "epoch": 0.23414051841746247,
      "grad_norm": 0.41364020109176636,
      "learning_rate": 1.7896401829063524e-06,
      "loss": 0.5555,
      "step": 1373
    },
    {
      "epoch": 0.23431105047748976,
      "grad_norm": 0.41541293263435364,
      "learning_rate": 1.7893011757704024e-06,
      "loss": 0.5374,
      "step": 1374
    },
    {
      "epoch": 0.23448158253751705,
      "grad_norm": 0.4567691385746002,
      "learning_rate": 1.788961927852845e-06,
      "loss": 0.5438,
      "step": 1375
    },
    {
      "epoch": 0.23465211459754434,
      "grad_norm": 0.46262142062187195,
      "learning_rate": 1.7886224392571707e-06,
      "loss": 0.5419,
      "step": 1376
    },
    {
      "epoch": 0.23482264665757163,
      "grad_norm": 0.48895400762557983,
      "learning_rate": 1.7882827100869423e-06,
      "loss": 0.5562,
      "step": 1377
    },
    {
      "epoch": 0.23499317871759892,
      "grad_norm": 0.6779786348342896,
      "learning_rate": 1.7879427404457967e-06,
      "loss": 0.5455,
      "step": 1378
    },
    {
      "epoch": 0.23516371077762618,
      "grad_norm": 0.5912104249000549,
      "learning_rate": 1.7876025304374441e-06,
      "loss": 0.5356,
      "step": 1379
    },
    {
      "epoch": 0.23533424283765347,
      "grad_norm": 0.3976776599884033,
      "learning_rate": 1.7872620801656675e-06,
      "loss": 0.5336,
      "step": 1380
    },
    {
      "epoch": 0.23550477489768076,
      "grad_norm": 0.6555095314979553,
      "learning_rate": 1.7869213897343238e-06,
      "loss": 0.5451,
      "step": 1381
    },
    {
      "epoch": 0.23567530695770805,
      "grad_norm": 0.7012636661529541,
      "learning_rate": 1.7865804592473424e-06,
      "loss": 0.5406,
      "step": 1382
    },
    {
      "epoch": 0.23584583901773534,
      "grad_norm": 0.5283592343330383,
      "learning_rate": 1.786239288808727e-06,
      "loss": 0.5485,
      "step": 1383
    },
    {
      "epoch": 0.23601637107776263,
      "grad_norm": 0.4392828345298767,
      "learning_rate": 1.7858978785225537e-06,
      "loss": 0.5427,
      "step": 1384
    },
    {
      "epoch": 0.2361869031377899,
      "grad_norm": 0.5758363008499146,
      "learning_rate": 1.785556228492972e-06,
      "loss": 0.5465,
      "step": 1385
    },
    {
      "epoch": 0.23635743519781718,
      "grad_norm": 0.5296410322189331,
      "learning_rate": 1.7852143388242048e-06,
      "loss": 0.531,
      "step": 1386
    },
    {
      "epoch": 0.23652796725784447,
      "grad_norm": 0.3970211148262024,
      "learning_rate": 1.7848722096205473e-06,
      "loss": 0.5535,
      "step": 1387
    },
    {
      "epoch": 0.23669849931787176,
      "grad_norm": 0.6295623183250427,
      "learning_rate": 1.7845298409863684e-06,
      "loss": 0.5366,
      "step": 1388
    },
    {
      "epoch": 0.23686903137789905,
      "grad_norm": 0.45526471734046936,
      "learning_rate": 1.7841872330261104e-06,
      "loss": 0.5509,
      "step": 1389
    },
    {
      "epoch": 0.23703956343792634,
      "grad_norm": 0.5751361846923828,
      "learning_rate": 1.7838443858442876e-06,
      "loss": 0.5459,
      "step": 1390
    },
    {
      "epoch": 0.2372100954979536,
      "grad_norm": 0.5891697406768799,
      "learning_rate": 1.7835012995454885e-06,
      "loss": 0.5561,
      "step": 1391
    },
    {
      "epoch": 0.2373806275579809,
      "grad_norm": 0.5361023545265198,
      "learning_rate": 1.7831579742343731e-06,
      "loss": 0.537,
      "step": 1392
    },
    {
      "epoch": 0.23755115961800818,
      "grad_norm": 0.5258536338806152,
      "learning_rate": 1.7828144100156757e-06,
      "loss": 0.5172,
      "step": 1393
    },
    {
      "epoch": 0.23772169167803547,
      "grad_norm": 0.6109665632247925,
      "learning_rate": 1.7824706069942026e-06,
      "loss": 0.5425,
      "step": 1394
    },
    {
      "epoch": 0.23789222373806276,
      "grad_norm": 0.4555569291114807,
      "learning_rate": 1.782126565274833e-06,
      "loss": 0.5563,
      "step": 1395
    },
    {
      "epoch": 0.23806275579809005,
      "grad_norm": 0.4836210012435913,
      "learning_rate": 1.7817822849625195e-06,
      "loss": 0.5375,
      "step": 1396
    },
    {
      "epoch": 0.23823328785811732,
      "grad_norm": 0.5767562985420227,
      "learning_rate": 1.781437766162287e-06,
      "loss": 0.5467,
      "step": 1397
    },
    {
      "epoch": 0.2384038199181446,
      "grad_norm": 0.40231022238731384,
      "learning_rate": 1.781093008979233e-06,
      "loss": 0.5388,
      "step": 1398
    },
    {
      "epoch": 0.2385743519781719,
      "grad_norm": 0.5109986066818237,
      "learning_rate": 1.7807480135185285e-06,
      "loss": 0.5251,
      "step": 1399
    },
    {
      "epoch": 0.23874488403819918,
      "grad_norm": 0.484244167804718,
      "learning_rate": 1.7804027798854166e-06,
      "loss": 0.544,
      "step": 1400
    },
    {
      "epoch": 0.23891541609822647,
      "grad_norm": 0.515340268611908,
      "learning_rate": 1.7800573081852125e-06,
      "loss": 0.5219,
      "step": 1401
    },
    {
      "epoch": 0.23908594815825376,
      "grad_norm": 0.5590576529502869,
      "learning_rate": 1.7797115985233052e-06,
      "loss": 0.5427,
      "step": 1402
    },
    {
      "epoch": 0.23925648021828103,
      "grad_norm": 0.45182856917381287,
      "learning_rate": 1.7793656510051558e-06,
      "loss": 0.5387,
      "step": 1403
    },
    {
      "epoch": 0.23942701227830832,
      "grad_norm": 0.49502822756767273,
      "learning_rate": 1.7790194657362974e-06,
      "loss": 0.5385,
      "step": 1404
    },
    {
      "epoch": 0.2395975443383356,
      "grad_norm": 0.5761579871177673,
      "learning_rate": 1.7786730428223365e-06,
      "loss": 0.5401,
      "step": 1405
    },
    {
      "epoch": 0.2397680763983629,
      "grad_norm": 0.4353269934654236,
      "learning_rate": 1.7783263823689523e-06,
      "loss": 0.528,
      "step": 1406
    },
    {
      "epoch": 0.23993860845839018,
      "grad_norm": 0.3897313177585602,
      "learning_rate": 1.7779794844818947e-06,
      "loss": 0.5276,
      "step": 1407
    },
    {
      "epoch": 0.24010914051841747,
      "grad_norm": 0.4709970951080322,
      "learning_rate": 1.7776323492669878e-06,
      "loss": 0.5405,
      "step": 1408
    },
    {
      "epoch": 0.24027967257844474,
      "grad_norm": 0.5096486210823059,
      "learning_rate": 1.7772849768301277e-06,
      "loss": 0.5343,
      "step": 1409
    },
    {
      "epoch": 0.24045020463847203,
      "grad_norm": 0.38168245553970337,
      "learning_rate": 1.7769373672772824e-06,
      "loss": 0.5346,
      "step": 1410
    },
    {
      "epoch": 0.24062073669849932,
      "grad_norm": 0.34209996461868286,
      "learning_rate": 1.776589520714493e-06,
      "loss": 0.54,
      "step": 1411
    },
    {
      "epoch": 0.2407912687585266,
      "grad_norm": 0.44546452164649963,
      "learning_rate": 1.7762414372478718e-06,
      "loss": 0.536,
      "step": 1412
    },
    {
      "epoch": 0.2409618008185539,
      "grad_norm": 0.37246155738830566,
      "learning_rate": 1.7758931169836041e-06,
      "loss": 0.528,
      "step": 1413
    },
    {
      "epoch": 0.24113233287858118,
      "grad_norm": 0.3510386347770691,
      "learning_rate": 1.775544560027948e-06,
      "loss": 0.5315,
      "step": 1414
    },
    {
      "epoch": 0.24130286493860845,
      "grad_norm": 0.3805161714553833,
      "learning_rate": 1.7751957664872318e-06,
      "loss": 0.5311,
      "step": 1415
    },
    {
      "epoch": 0.24147339699863574,
      "grad_norm": 0.35687196254730225,
      "learning_rate": 1.7748467364678587e-06,
      "loss": 0.5518,
      "step": 1416
    },
    {
      "epoch": 0.24164392905866303,
      "grad_norm": 0.37188974022865295,
      "learning_rate": 1.774497470076302e-06,
      "loss": 0.5411,
      "step": 1417
    },
    {
      "epoch": 0.24181446111869032,
      "grad_norm": 0.3701164126396179,
      "learning_rate": 1.7741479674191078e-06,
      "loss": 0.5432,
      "step": 1418
    },
    {
      "epoch": 0.2419849931787176,
      "grad_norm": 0.4429377317428589,
      "learning_rate": 1.773798228602894e-06,
      "loss": 0.5277,
      "step": 1419
    },
    {
      "epoch": 0.2421555252387449,
      "grad_norm": 0.4325040876865387,
      "learning_rate": 1.7734482537343514e-06,
      "loss": 0.5358,
      "step": 1420
    },
    {
      "epoch": 0.24232605729877216,
      "grad_norm": 0.4019585847854614,
      "learning_rate": 1.7730980429202415e-06,
      "loss": 0.5328,
      "step": 1421
    },
    {
      "epoch": 0.24249658935879945,
      "grad_norm": 0.4692414104938507,
      "learning_rate": 1.7727475962673986e-06,
      "loss": 0.5395,
      "step": 1422
    },
    {
      "epoch": 0.24266712141882674,
      "grad_norm": 0.4960983395576477,
      "learning_rate": 1.7723969138827293e-06,
      "loss": 0.5354,
      "step": 1423
    },
    {
      "epoch": 0.24283765347885403,
      "grad_norm": 0.4993603527545929,
      "learning_rate": 1.7720459958732112e-06,
      "loss": 0.5208,
      "step": 1424
    },
    {
      "epoch": 0.24300818553888132,
      "grad_norm": 0.5154334306716919,
      "learning_rate": 1.771694842345894e-06,
      "loss": 0.5344,
      "step": 1425
    },
    {
      "epoch": 0.2431787175989086,
      "grad_norm": 0.4017626941204071,
      "learning_rate": 1.7713434534078998e-06,
      "loss": 0.5361,
      "step": 1426
    },
    {
      "epoch": 0.24334924965893587,
      "grad_norm": 0.49590057134628296,
      "learning_rate": 1.770991829166422e-06,
      "loss": 0.5545,
      "step": 1427
    },
    {
      "epoch": 0.24351978171896316,
      "grad_norm": 0.6026977896690369,
      "learning_rate": 1.770639969728726e-06,
      "loss": 0.5229,
      "step": 1428
    },
    {
      "epoch": 0.24369031377899045,
      "grad_norm": 0.4571629762649536,
      "learning_rate": 1.7702878752021488e-06,
      "loss": 0.5439,
      "step": 1429
    },
    {
      "epoch": 0.24386084583901774,
      "grad_norm": 0.3930206596851349,
      "learning_rate": 1.769935545694099e-06,
      "loss": 0.5346,
      "step": 1430
    },
    {
      "epoch": 0.24403137789904503,
      "grad_norm": 0.49052783846855164,
      "learning_rate": 1.7695829813120575e-06,
      "loss": 0.5547,
      "step": 1431
    },
    {
      "epoch": 0.24420190995907232,
      "grad_norm": 0.45432084798812866,
      "learning_rate": 1.7692301821635764e-06,
      "loss": 0.5354,
      "step": 1432
    },
    {
      "epoch": 0.24437244201909958,
      "grad_norm": 0.4165271520614624,
      "learning_rate": 1.768877148356279e-06,
      "loss": 0.536,
      "step": 1433
    },
    {
      "epoch": 0.24454297407912687,
      "grad_norm": 0.34783709049224854,
      "learning_rate": 1.7685238799978607e-06,
      "loss": 0.5441,
      "step": 1434
    },
    {
      "epoch": 0.24471350613915416,
      "grad_norm": 0.4623855948448181,
      "learning_rate": 1.7681703771960886e-06,
      "loss": 0.5444,
      "step": 1435
    },
    {
      "epoch": 0.24488403819918145,
      "grad_norm": 0.44057196378707886,
      "learning_rate": 1.7678166400588013e-06,
      "loss": 0.5268,
      "step": 1436
    },
    {
      "epoch": 0.24505457025920874,
      "grad_norm": 0.5099006295204163,
      "learning_rate": 1.7674626686939081e-06,
      "loss": 0.5346,
      "step": 1437
    },
    {
      "epoch": 0.24522510231923603,
      "grad_norm": 0.48367926478385925,
      "learning_rate": 1.767108463209391e-06,
      "loss": 0.5404,
      "step": 1438
    },
    {
      "epoch": 0.2453956343792633,
      "grad_norm": 0.4167617857456207,
      "learning_rate": 1.7667540237133025e-06,
      "loss": 0.5279,
      "step": 1439
    },
    {
      "epoch": 0.24556616643929058,
      "grad_norm": 0.37517234683036804,
      "learning_rate": 1.7663993503137661e-06,
      "loss": 0.5471,
      "step": 1440
    },
    {
      "epoch": 0.24573669849931787,
      "grad_norm": 0.4140487313270569,
      "learning_rate": 1.7660444431189781e-06,
      "loss": 0.548,
      "step": 1441
    },
    {
      "epoch": 0.24590723055934516,
      "grad_norm": 0.4195196032524109,
      "learning_rate": 1.7656893022372055e-06,
      "loss": 0.5442,
      "step": 1442
    },
    {
      "epoch": 0.24607776261937245,
      "grad_norm": 0.3868313133716583,
      "learning_rate": 1.7653339277767855e-06,
      "loss": 0.5393,
      "step": 1443
    },
    {
      "epoch": 0.24624829467939974,
      "grad_norm": 0.347720205783844,
      "learning_rate": 1.764978319846128e-06,
      "loss": 0.5369,
      "step": 1444
    },
    {
      "epoch": 0.246418826739427,
      "grad_norm": 0.41931644082069397,
      "learning_rate": 1.7646224785537137e-06,
      "loss": 0.5386,
      "step": 1445
    },
    {
      "epoch": 0.2465893587994543,
      "grad_norm": 0.36928820610046387,
      "learning_rate": 1.7642664040080941e-06,
      "loss": 0.5301,
      "step": 1446
    },
    {
      "epoch": 0.24675989085948158,
      "grad_norm": 0.4465715289115906,
      "learning_rate": 1.763910096317892e-06,
      "loss": 0.5563,
      "step": 1447
    },
    {
      "epoch": 0.24693042291950887,
      "grad_norm": 0.39555251598358154,
      "learning_rate": 1.763553555591802e-06,
      "loss": 0.5324,
      "step": 1448
    },
    {
      "epoch": 0.24710095497953616,
      "grad_norm": 0.3805316090583801,
      "learning_rate": 1.7631967819385885e-06,
      "loss": 0.539,
      "step": 1449
    },
    {
      "epoch": 0.24727148703956345,
      "grad_norm": 0.4752626121044159,
      "learning_rate": 1.7628397754670881e-06,
      "loss": 0.5375,
      "step": 1450
    },
    {
      "epoch": 0.2474420190995907,
      "grad_norm": 0.4983365833759308,
      "learning_rate": 1.7624825362862084e-06,
      "loss": 0.5298,
      "step": 1451
    },
    {
      "epoch": 0.247612551159618,
      "grad_norm": 0.4518239498138428,
      "learning_rate": 1.7621250645049268e-06,
      "loss": 0.5432,
      "step": 1452
    },
    {
      "epoch": 0.2477830832196453,
      "grad_norm": 0.4357149004936218,
      "learning_rate": 1.761767360232293e-06,
      "loss": 0.5442,
      "step": 1453
    },
    {
      "epoch": 0.24795361527967258,
      "grad_norm": 0.4708903431892395,
      "learning_rate": 1.761409423577427e-06,
      "loss": 0.5316,
      "step": 1454
    },
    {
      "epoch": 0.24812414733969987,
      "grad_norm": 0.5494210720062256,
      "learning_rate": 1.7610512546495196e-06,
      "loss": 0.5368,
      "step": 1455
    },
    {
      "epoch": 0.24829467939972716,
      "grad_norm": 0.4802795946598053,
      "learning_rate": 1.7606928535578328e-06,
      "loss": 0.5277,
      "step": 1456
    },
    {
      "epoch": 0.24846521145975442,
      "grad_norm": 0.4052109718322754,
      "learning_rate": 1.7603342204116992e-06,
      "loss": 0.5441,
      "step": 1457
    },
    {
      "epoch": 0.2486357435197817,
      "grad_norm": 0.5663003921508789,
      "learning_rate": 1.759975355320522e-06,
      "loss": 0.5519,
      "step": 1458
    },
    {
      "epoch": 0.248806275579809,
      "grad_norm": 0.4638127088546753,
      "learning_rate": 1.7596162583937758e-06,
      "loss": 0.5299,
      "step": 1459
    },
    {
      "epoch": 0.2489768076398363,
      "grad_norm": 0.4754791855812073,
      "learning_rate": 1.7592569297410056e-06,
      "loss": 0.5322,
      "step": 1460
    },
    {
      "epoch": 0.24914733969986358,
      "grad_norm": 0.43075990676879883,
      "learning_rate": 1.7588973694718265e-06,
      "loss": 0.5425,
      "step": 1461
    },
    {
      "epoch": 0.24931787175989087,
      "grad_norm": 0.567526638507843,
      "learning_rate": 1.7585375776959248e-06,
      "loss": 0.5426,
      "step": 1462
    },
    {
      "epoch": 0.24948840381991813,
      "grad_norm": 0.43639516830444336,
      "learning_rate": 1.758177554523058e-06,
      "loss": 0.5346,
      "step": 1463
    },
    {
      "epoch": 0.24965893587994542,
      "grad_norm": 0.36125215888023376,
      "learning_rate": 1.757817300063053e-06,
      "loss": 0.5348,
      "step": 1464
    },
    {
      "epoch": 0.2498294679399727,
      "grad_norm": 0.4189085066318512,
      "learning_rate": 1.7574568144258079e-06,
      "loss": 0.5315,
      "step": 1465
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39513495564460754,
      "learning_rate": 1.7570960977212915e-06,
      "loss": 0.5408,
      "step": 1466
    },
    {
      "epoch": 0.25017053206002726,
      "grad_norm": 0.3659518361091614,
      "learning_rate": 1.756735150059543e-06,
      "loss": 0.5476,
      "step": 1467
    },
    {
      "epoch": 0.2503410641200546,
      "grad_norm": 0.401780366897583,
      "learning_rate": 1.7563739715506713e-06,
      "loss": 0.5473,
      "step": 1468
    },
    {
      "epoch": 0.25051159618008184,
      "grad_norm": 0.4212344288825989,
      "learning_rate": 1.756012562304857e-06,
      "loss": 0.5269,
      "step": 1469
    },
    {
      "epoch": 0.25068212824010916,
      "grad_norm": 0.378131628036499,
      "learning_rate": 1.7556509224323496e-06,
      "loss": 0.5233,
      "step": 1470
    },
    {
      "epoch": 0.2508526603001364,
      "grad_norm": 0.3757961392402649,
      "learning_rate": 1.7552890520434706e-06,
      "loss": 0.5456,
      "step": 1471
    },
    {
      "epoch": 0.25102319236016374,
      "grad_norm": 0.47206035256385803,
      "learning_rate": 1.75492695124861e-06,
      "loss": 0.55,
      "step": 1472
    },
    {
      "epoch": 0.251193724420191,
      "grad_norm": 0.34729981422424316,
      "learning_rate": 1.7545646201582305e-06,
      "loss": 0.5406,
      "step": 1473
    },
    {
      "epoch": 0.25136425648021826,
      "grad_norm": 0.47744041681289673,
      "learning_rate": 1.7542020588828621e-06,
      "loss": 0.5344,
      "step": 1474
    },
    {
      "epoch": 0.2515347885402456,
      "grad_norm": 0.4247567653656006,
      "learning_rate": 1.7538392675331076e-06,
      "loss": 0.5597,
      "step": 1475
    },
    {
      "epoch": 0.25170532060027284,
      "grad_norm": 0.5592153072357178,
      "learning_rate": 1.7534762462196389e-06,
      "loss": 0.5418,
      "step": 1476
    },
    {
      "epoch": 0.25187585266030016,
      "grad_norm": 0.5871926546096802,
      "learning_rate": 1.7531129950531973e-06,
      "loss": 0.525,
      "step": 1477
    },
    {
      "epoch": 0.2520463847203274,
      "grad_norm": 0.3953045606613159,
      "learning_rate": 1.7527495141445957e-06,
      "loss": 0.5435,
      "step": 1478
    },
    {
      "epoch": 0.2522169167803547,
      "grad_norm": 0.48583656549453735,
      "learning_rate": 1.7523858036047159e-06,
      "loss": 0.557,
      "step": 1479
    },
    {
      "epoch": 0.252387448840382,
      "grad_norm": 0.3983100354671478,
      "learning_rate": 1.7520218635445108e-06,
      "loss": 0.5399,
      "step": 1480
    },
    {
      "epoch": 0.25255798090040926,
      "grad_norm": 0.38920262455940247,
      "learning_rate": 1.751657694075002e-06,
      "loss": 0.5276,
      "step": 1481
    },
    {
      "epoch": 0.2527285129604366,
      "grad_norm": 0.4583071172237396,
      "learning_rate": 1.7512932953072826e-06,
      "loss": 0.5263,
      "step": 1482
    },
    {
      "epoch": 0.25289904502046384,
      "grad_norm": 0.40535056591033936,
      "learning_rate": 1.7509286673525145e-06,
      "loss": 0.5359,
      "step": 1483
    },
    {
      "epoch": 0.25306957708049116,
      "grad_norm": 0.4464225172996521,
      "learning_rate": 1.75056381032193e-06,
      "loss": 0.5446,
      "step": 1484
    },
    {
      "epoch": 0.2532401091405184,
      "grad_norm": 0.5687764286994934,
      "learning_rate": 1.7501987243268309e-06,
      "loss": 0.5421,
      "step": 1485
    },
    {
      "epoch": 0.2534106412005457,
      "grad_norm": 0.4144216775894165,
      "learning_rate": 1.7498334094785896e-06,
      "loss": 0.5363,
      "step": 1486
    },
    {
      "epoch": 0.253581173260573,
      "grad_norm": 0.5526368021965027,
      "learning_rate": 1.7494678658886474e-06,
      "loss": 0.5316,
      "step": 1487
    },
    {
      "epoch": 0.25375170532060026,
      "grad_norm": 0.4659339189529419,
      "learning_rate": 1.749102093668516e-06,
      "loss": 0.5334,
      "step": 1488
    },
    {
      "epoch": 0.2539222373806276,
      "grad_norm": 0.39899182319641113,
      "learning_rate": 1.7487360929297768e-06,
      "loss": 0.5356,
      "step": 1489
    },
    {
      "epoch": 0.25409276944065484,
      "grad_norm": 0.509774386882782,
      "learning_rate": 1.7483698637840805e-06,
      "loss": 0.539,
      "step": 1490
    },
    {
      "epoch": 0.2542633015006821,
      "grad_norm": 0.5092775225639343,
      "learning_rate": 1.748003406343148e-06,
      "loss": 0.5271,
      "step": 1491
    },
    {
      "epoch": 0.2544338335607094,
      "grad_norm": 0.42660048604011536,
      "learning_rate": 1.7476367207187697e-06,
      "loss": 0.5413,
      "step": 1492
    },
    {
      "epoch": 0.2546043656207367,
      "grad_norm": 0.35999831557273865,
      "learning_rate": 1.7472698070228051e-06,
      "loss": 0.5335,
      "step": 1493
    },
    {
      "epoch": 0.254774897680764,
      "grad_norm": 0.3916492164134979,
      "learning_rate": 1.746902665367184e-06,
      "loss": 0.5439,
      "step": 1494
    },
    {
      "epoch": 0.25494542974079126,
      "grad_norm": 0.36775171756744385,
      "learning_rate": 1.7465352958639052e-06,
      "loss": 0.5386,
      "step": 1495
    },
    {
      "epoch": 0.2551159618008186,
      "grad_norm": 0.4069823622703552,
      "learning_rate": 1.7461676986250373e-06,
      "loss": 0.5463,
      "step": 1496
    },
    {
      "epoch": 0.25528649386084584,
      "grad_norm": 0.3523212671279907,
      "learning_rate": 1.7457998737627184e-06,
      "loss": 0.54,
      "step": 1497
    },
    {
      "epoch": 0.2554570259208731,
      "grad_norm": 0.4446667432785034,
      "learning_rate": 1.7454318213891557e-06,
      "loss": 0.5282,
      "step": 1498
    },
    {
      "epoch": 0.2556275579809004,
      "grad_norm": 0.398403137922287,
      "learning_rate": 1.7450635416166262e-06,
      "loss": 0.528,
      "step": 1499
    },
    {
      "epoch": 0.2557980900409277,
      "grad_norm": 0.4375839829444885,
      "learning_rate": 1.7446950345574764e-06,
      "loss": 0.5429,
      "step": 1500
    },
    {
      "epoch": 0.255968622100955,
      "grad_norm": 0.41334497928619385,
      "learning_rate": 1.7443263003241213e-06,
      "loss": 0.5182,
      "step": 1501
    },
    {
      "epoch": 0.25613915416098226,
      "grad_norm": 0.4163067936897278,
      "learning_rate": 1.743957339029046e-06,
      "loss": 0.5474,
      "step": 1502
    },
    {
      "epoch": 0.2563096862210095,
      "grad_norm": 0.44295674562454224,
      "learning_rate": 1.743588150784805e-06,
      "loss": 0.5352,
      "step": 1503
    },
    {
      "epoch": 0.25648021828103684,
      "grad_norm": 0.3465801775455475,
      "learning_rate": 1.7432187357040207e-06,
      "loss": 0.5285,
      "step": 1504
    },
    {
      "epoch": 0.2566507503410641,
      "grad_norm": 0.39809703826904297,
      "learning_rate": 1.7428490938993865e-06,
      "loss": 0.5229,
      "step": 1505
    },
    {
      "epoch": 0.2568212824010914,
      "grad_norm": 0.42837026715278625,
      "learning_rate": 1.7424792254836638e-06,
      "loss": 0.527,
      "step": 1506
    },
    {
      "epoch": 0.2569918144611187,
      "grad_norm": 0.3859080672264099,
      "learning_rate": 1.7421091305696835e-06,
      "loss": 0.5412,
      "step": 1507
    },
    {
      "epoch": 0.257162346521146,
      "grad_norm": 0.33818215131759644,
      "learning_rate": 1.7417388092703454e-06,
      "loss": 0.5227,
      "step": 1508
    },
    {
      "epoch": 0.25733287858117326,
      "grad_norm": 0.428977906703949,
      "learning_rate": 1.7413682616986187e-06,
      "loss": 0.5369,
      "step": 1509
    },
    {
      "epoch": 0.2575034106412005,
      "grad_norm": 0.3944161832332611,
      "learning_rate": 1.740997487967541e-06,
      "loss": 0.5301,
      "step": 1510
    },
    {
      "epoch": 0.25767394270122784,
      "grad_norm": 0.39399105310440063,
      "learning_rate": 1.7406264881902197e-06,
      "loss": 0.5211,
      "step": 1511
    },
    {
      "epoch": 0.2578444747612551,
      "grad_norm": 0.47329261898994446,
      "learning_rate": 1.7402552624798306e-06,
      "loss": 0.5299,
      "step": 1512
    },
    {
      "epoch": 0.2580150068212824,
      "grad_norm": 0.6779657602310181,
      "learning_rate": 1.7398838109496188e-06,
      "loss": 0.5208,
      "step": 1513
    },
    {
      "epoch": 0.2581855388813097,
      "grad_norm": 0.7127777934074402,
      "learning_rate": 1.739512133712898e-06,
      "loss": 0.5359,
      "step": 1514
    },
    {
      "epoch": 0.25835607094133695,
      "grad_norm": 0.47813621163368225,
      "learning_rate": 1.7391402308830507e-06,
      "loss": 0.5411,
      "step": 1515
    },
    {
      "epoch": 0.25852660300136426,
      "grad_norm": 0.4114779829978943,
      "learning_rate": 1.738768102573528e-06,
      "loss": 0.533,
      "step": 1516
    },
    {
      "epoch": 0.2586971350613915,
      "grad_norm": 0.572684645652771,
      "learning_rate": 1.738395748897851e-06,
      "loss": 0.5256,
      "step": 1517
    },
    {
      "epoch": 0.25886766712141884,
      "grad_norm": 0.40316465497016907,
      "learning_rate": 1.7380231699696081e-06,
      "loss": 0.5448,
      "step": 1518
    },
    {
      "epoch": 0.2590381991814461,
      "grad_norm": 0.41706883907318115,
      "learning_rate": 1.737650365902457e-06,
      "loss": 0.5454,
      "step": 1519
    },
    {
      "epoch": 0.2592087312414734,
      "grad_norm": 0.4692871868610382,
      "learning_rate": 1.7372773368101244e-06,
      "loss": 0.5417,
      "step": 1520
    },
    {
      "epoch": 0.2593792633015007,
      "grad_norm": 0.5374259352684021,
      "learning_rate": 1.7369040828064048e-06,
      "loss": 0.5369,
      "step": 1521
    },
    {
      "epoch": 0.25954979536152795,
      "grad_norm": 0.5214207172393799,
      "learning_rate": 1.7365306040051623e-06,
      "loss": 0.5342,
      "step": 1522
    },
    {
      "epoch": 0.25972032742155526,
      "grad_norm": 0.4177664518356323,
      "learning_rate": 1.736156900520329e-06,
      "loss": 0.5433,
      "step": 1523
    },
    {
      "epoch": 0.2598908594815825,
      "grad_norm": 0.42407530546188354,
      "learning_rate": 1.7357829724659056e-06,
      "loss": 0.5325,
      "step": 1524
    },
    {
      "epoch": 0.26006139154160984,
      "grad_norm": 0.3311309218406677,
      "learning_rate": 1.7354088199559612e-06,
      "loss": 0.5361,
      "step": 1525
    },
    {
      "epoch": 0.2602319236016371,
      "grad_norm": 0.5420066118240356,
      "learning_rate": 1.735034443104634e-06,
      "loss": 0.5271,
      "step": 1526
    },
    {
      "epoch": 0.26040245566166437,
      "grad_norm": 0.497405469417572,
      "learning_rate": 1.7346598420261298e-06,
      "loss": 0.5335,
      "step": 1527
    },
    {
      "epoch": 0.2605729877216917,
      "grad_norm": 0.4364621639251709,
      "learning_rate": 1.734285016834723e-06,
      "loss": 0.5524,
      "step": 1528
    },
    {
      "epoch": 0.26074351978171895,
      "grad_norm": 0.5040315985679626,
      "learning_rate": 1.7339099676447567e-06,
      "loss": 0.5319,
      "step": 1529
    },
    {
      "epoch": 0.26091405184174626,
      "grad_norm": 0.4158141613006592,
      "learning_rate": 1.7335346945706421e-06,
      "loss": 0.5485,
      "step": 1530
    },
    {
      "epoch": 0.2610845839017735,
      "grad_norm": 0.4395507276058197,
      "learning_rate": 1.7331591977268594e-06,
      "loss": 0.5188,
      "step": 1531
    },
    {
      "epoch": 0.26125511596180084,
      "grad_norm": 0.47268790006637573,
      "learning_rate": 1.7327834772279552e-06,
      "loss": 0.5307,
      "step": 1532
    },
    {
      "epoch": 0.2614256480218281,
      "grad_norm": 0.4334191381931305,
      "learning_rate": 1.7324075331885468e-06,
      "loss": 0.5362,
      "step": 1533
    },
    {
      "epoch": 0.26159618008185537,
      "grad_norm": 0.6413484811782837,
      "learning_rate": 1.7320313657233177e-06,
      "loss": 0.529,
      "step": 1534
    },
    {
      "epoch": 0.2617667121418827,
      "grad_norm": 0.3648528754711151,
      "learning_rate": 1.7316549749470205e-06,
      "loss": 0.5225,
      "step": 1535
    },
    {
      "epoch": 0.26193724420190995,
      "grad_norm": 0.4816478490829468,
      "learning_rate": 1.7312783609744755e-06,
      "loss": 0.5444,
      "step": 1536
    },
    {
      "epoch": 0.26210777626193726,
      "grad_norm": 0.470768004655838,
      "learning_rate": 1.7309015239205719e-06,
      "loss": 0.5327,
      "step": 1537
    },
    {
      "epoch": 0.2622783083219645,
      "grad_norm": 0.5041337013244629,
      "learning_rate": 1.730524463900266e-06,
      "loss": 0.539,
      "step": 1538
    },
    {
      "epoch": 0.2624488403819918,
      "grad_norm": 0.46768811345100403,
      "learning_rate": 1.7301471810285824e-06,
      "loss": 0.5355,
      "step": 1539
    },
    {
      "epoch": 0.2626193724420191,
      "grad_norm": 0.4774124026298523,
      "learning_rate": 1.7297696754206138e-06,
      "loss": 0.5454,
      "step": 1540
    },
    {
      "epoch": 0.26278990450204637,
      "grad_norm": 0.4992062449455261,
      "learning_rate": 1.7293919471915207e-06,
      "loss": 0.5319,
      "step": 1541
    },
    {
      "epoch": 0.2629604365620737,
      "grad_norm": 0.40920349955558777,
      "learning_rate": 1.7290139964565322e-06,
      "loss": 0.5432,
      "step": 1542
    },
    {
      "epoch": 0.26313096862210095,
      "grad_norm": 0.39942824840545654,
      "learning_rate": 1.7286358233309441e-06,
      "loss": 0.5403,
      "step": 1543
    },
    {
      "epoch": 0.26330150068212826,
      "grad_norm": 0.40174806118011475,
      "learning_rate": 1.7282574279301206e-06,
      "loss": 0.5395,
      "step": 1544
    },
    {
      "epoch": 0.2634720327421555,
      "grad_norm": 0.5135958790779114,
      "learning_rate": 1.7278788103694944e-06,
      "loss": 0.5331,
      "step": 1545
    },
    {
      "epoch": 0.2636425648021828,
      "grad_norm": 0.5391552448272705,
      "learning_rate": 1.7274999707645646e-06,
      "loss": 0.5385,
      "step": 1546
    },
    {
      "epoch": 0.2638130968622101,
      "grad_norm": 0.37942448258399963,
      "learning_rate": 1.7271209092308994e-06,
      "loss": 0.5223,
      "step": 1547
    },
    {
      "epoch": 0.26398362892223737,
      "grad_norm": 0.4890262484550476,
      "learning_rate": 1.7267416258841332e-06,
      "loss": 0.5287,
      "step": 1548
    },
    {
      "epoch": 0.2641541609822647,
      "grad_norm": 0.43532058596611023,
      "learning_rate": 1.7263621208399699e-06,
      "loss": 0.5188,
      "step": 1549
    },
    {
      "epoch": 0.26432469304229195,
      "grad_norm": 0.5018872618675232,
      "learning_rate": 1.7259823942141794e-06,
      "loss": 0.5317,
      "step": 1550
    },
    {
      "epoch": 0.2644952251023192,
      "grad_norm": 0.5198800563812256,
      "learning_rate": 1.7256024461226004e-06,
      "loss": 0.5359,
      "step": 1551
    },
    {
      "epoch": 0.2646657571623465,
      "grad_norm": 0.5587589740753174,
      "learning_rate": 1.7252222766811379e-06,
      "loss": 0.551,
      "step": 1552
    },
    {
      "epoch": 0.2648362892223738,
      "grad_norm": 0.5261849164962769,
      "learning_rate": 1.7248418860057656e-06,
      "loss": 0.54,
      "step": 1553
    },
    {
      "epoch": 0.2650068212824011,
      "grad_norm": 0.45196717977523804,
      "learning_rate": 1.7244612742125238e-06,
      "loss": 0.5316,
      "step": 1554
    },
    {
      "epoch": 0.26517735334242837,
      "grad_norm": 0.6772377490997314,
      "learning_rate": 1.7240804414175214e-06,
      "loss": 0.5185,
      "step": 1555
    },
    {
      "epoch": 0.2653478854024557,
      "grad_norm": 0.4614553451538086,
      "learning_rate": 1.7236993877369334e-06,
      "loss": 0.5206,
      "step": 1556
    },
    {
      "epoch": 0.26551841746248295,
      "grad_norm": 0.4851749837398529,
      "learning_rate": 1.7233181132870032e-06,
      "loss": 0.5461,
      "step": 1557
    },
    {
      "epoch": 0.2656889495225102,
      "grad_norm": 0.5044141411781311,
      "learning_rate": 1.7229366181840405e-06,
      "loss": 0.5424,
      "step": 1558
    },
    {
      "epoch": 0.2658594815825375,
      "grad_norm": 0.4354725480079651,
      "learning_rate": 1.7225549025444234e-06,
      "loss": 0.5245,
      "step": 1559
    },
    {
      "epoch": 0.2660300136425648,
      "grad_norm": 0.5553825497627258,
      "learning_rate": 1.7221729664845968e-06,
      "loss": 0.5199,
      "step": 1560
    },
    {
      "epoch": 0.2662005457025921,
      "grad_norm": 0.4498416781425476,
      "learning_rate": 1.7217908101210727e-06,
      "loss": 0.5315,
      "step": 1561
    },
    {
      "epoch": 0.26637107776261937,
      "grad_norm": 0.5130297541618347,
      "learning_rate": 1.7214084335704309e-06,
      "loss": 0.5289,
      "step": 1562
    },
    {
      "epoch": 0.26654160982264663,
      "grad_norm": 0.43296122550964355,
      "learning_rate": 1.7210258369493173e-06,
      "loss": 0.5363,
      "step": 1563
    },
    {
      "epoch": 0.26671214188267395,
      "grad_norm": 0.45668303966522217,
      "learning_rate": 1.7206430203744456e-06,
      "loss": 0.5387,
      "step": 1564
    },
    {
      "epoch": 0.2668826739427012,
      "grad_norm": 0.549055278301239,
      "learning_rate": 1.720259983962597e-06,
      "loss": 0.5297,
      "step": 1565
    },
    {
      "epoch": 0.2670532060027285,
      "grad_norm": 0.3640718460083008,
      "learning_rate": 1.7198767278306191e-06,
      "loss": 0.5298,
      "step": 1566
    },
    {
      "epoch": 0.2672237380627558,
      "grad_norm": 0.5561659932136536,
      "learning_rate": 1.7194932520954268e-06,
      "loss": 0.5207,
      "step": 1567
    },
    {
      "epoch": 0.2673942701227831,
      "grad_norm": 0.5278323888778687,
      "learning_rate": 1.7191095568740018e-06,
      "loss": 0.5263,
      "step": 1568
    },
    {
      "epoch": 0.26756480218281037,
      "grad_norm": 0.41605237126350403,
      "learning_rate": 1.718725642283393e-06,
      "loss": 0.5318,
      "step": 1569
    },
    {
      "epoch": 0.26773533424283763,
      "grad_norm": 0.3954503536224365,
      "learning_rate": 1.7183415084407163e-06,
      "loss": 0.5242,
      "step": 1570
    },
    {
      "epoch": 0.26790586630286495,
      "grad_norm": 0.44009891152381897,
      "learning_rate": 1.7179571554631536e-06,
      "loss": 0.5363,
      "step": 1571
    },
    {
      "epoch": 0.2680763983628922,
      "grad_norm": 0.580356240272522,
      "learning_rate": 1.7175725834679553e-06,
      "loss": 0.5163,
      "step": 1572
    },
    {
      "epoch": 0.2682469304229195,
      "grad_norm": 0.4962151348590851,
      "learning_rate": 1.717187792572437e-06,
      "loss": 0.5246,
      "step": 1573
    },
    {
      "epoch": 0.2684174624829468,
      "grad_norm": 0.39382004737854004,
      "learning_rate": 1.716802782893982e-06,
      "loss": 0.5356,
      "step": 1574
    },
    {
      "epoch": 0.2685879945429741,
      "grad_norm": 0.4842481017112732,
      "learning_rate": 1.7164175545500403e-06,
      "loss": 0.523,
      "step": 1575
    },
    {
      "epoch": 0.26875852660300137,
      "grad_norm": 0.5253110527992249,
      "learning_rate": 1.7160321076581277e-06,
      "loss": 0.5354,
      "step": 1576
    },
    {
      "epoch": 0.26892905866302863,
      "grad_norm": 0.39040082693099976,
      "learning_rate": 1.7156464423358279e-06,
      "loss": 0.5269,
      "step": 1577
    },
    {
      "epoch": 0.26909959072305595,
      "grad_norm": 0.39658838510513306,
      "learning_rate": 1.7152605587007906e-06,
      "loss": 0.5282,
      "step": 1578
    },
    {
      "epoch": 0.2692701227830832,
      "grad_norm": 0.5355466604232788,
      "learning_rate": 1.7148744568707323e-06,
      "loss": 0.5382,
      "step": 1579
    },
    {
      "epoch": 0.2694406548431105,
      "grad_norm": 0.38271233439445496,
      "learning_rate": 1.7144881369634354e-06,
      "loss": 0.5288,
      "step": 1580
    },
    {
      "epoch": 0.2696111869031378,
      "grad_norm": 0.39254850149154663,
      "learning_rate": 1.7141015990967502e-06,
      "loss": 0.5249,
      "step": 1581
    },
    {
      "epoch": 0.26978171896316505,
      "grad_norm": 0.5330004096031189,
      "learning_rate": 1.7137148433885918e-06,
      "loss": 0.5234,
      "step": 1582
    },
    {
      "epoch": 0.26995225102319237,
      "grad_norm": 0.5613977909088135,
      "learning_rate": 1.7133278699569433e-06,
      "loss": 0.5135,
      "step": 1583
    },
    {
      "epoch": 0.27012278308321963,
      "grad_norm": 0.36673110723495483,
      "learning_rate": 1.7129406789198531e-06,
      "loss": 0.5292,
      "step": 1584
    },
    {
      "epoch": 0.27029331514324695,
      "grad_norm": 0.405582994222641,
      "learning_rate": 1.7125532703954367e-06,
      "loss": 0.5271,
      "step": 1585
    },
    {
      "epoch": 0.2704638472032742,
      "grad_norm": 0.49316197633743286,
      "learning_rate": 1.712165644501876e-06,
      "loss": 0.5293,
      "step": 1586
    },
    {
      "epoch": 0.2706343792633015,
      "grad_norm": 0.405775785446167,
      "learning_rate": 1.7117778013574177e-06,
      "loss": 0.5323,
      "step": 1587
    },
    {
      "epoch": 0.2708049113233288,
      "grad_norm": 0.4072016775608063,
      "learning_rate": 1.7113897410803768e-06,
      "loss": 0.5351,
      "step": 1588
    },
    {
      "epoch": 0.27097544338335605,
      "grad_norm": 0.5345101952552795,
      "learning_rate": 1.7110014637891341e-06,
      "loss": 0.5429,
      "step": 1589
    },
    {
      "epoch": 0.27114597544338337,
      "grad_norm": 0.4184858798980713,
      "learning_rate": 1.7106129696021353e-06,
      "loss": 0.5286,
      "step": 1590
    },
    {
      "epoch": 0.27131650750341063,
      "grad_norm": 0.40997418761253357,
      "learning_rate": 1.7102242586378936e-06,
      "loss": 0.5256,
      "step": 1591
    },
    {
      "epoch": 0.27148703956343795,
      "grad_norm": 0.5290420055389404,
      "learning_rate": 1.7098353310149877e-06,
      "loss": 0.5406,
      "step": 1592
    },
    {
      "epoch": 0.2716575716234652,
      "grad_norm": 0.3918258845806122,
      "learning_rate": 1.7094461868520626e-06,
      "loss": 0.5471,
      "step": 1593
    },
    {
      "epoch": 0.2718281036834925,
      "grad_norm": 0.41981926560401917,
      "learning_rate": 1.7090568262678294e-06,
      "loss": 0.5435,
      "step": 1594
    },
    {
      "epoch": 0.2719986357435198,
      "grad_norm": 0.5090052485466003,
      "learning_rate": 1.7086672493810654e-06,
      "loss": 0.5339,
      "step": 1595
    },
    {
      "epoch": 0.27216916780354705,
      "grad_norm": 0.401683509349823,
      "learning_rate": 1.7082774563106129e-06,
      "loss": 0.5197,
      "step": 1596
    },
    {
      "epoch": 0.27233969986357437,
      "grad_norm": 0.3491285443305969,
      "learning_rate": 1.7078874471753819e-06,
      "loss": 0.5388,
      "step": 1597
    },
    {
      "epoch": 0.27251023192360163,
      "grad_norm": 0.4440529942512512,
      "learning_rate": 1.7074972220943466e-06,
      "loss": 0.5343,
      "step": 1598
    },
    {
      "epoch": 0.27268076398362895,
      "grad_norm": 0.4229428172111511,
      "learning_rate": 1.7071067811865479e-06,
      "loss": 0.5364,
      "step": 1599
    },
    {
      "epoch": 0.2728512960436562,
      "grad_norm": 0.4363120496273041,
      "learning_rate": 1.7067161245710924e-06,
      "loss": 0.5367,
      "step": 1600
    },
    {
      "epoch": 0.2730218281036835,
      "grad_norm": 0.4940813183784485,
      "learning_rate": 1.7063252523671523e-06,
      "loss": 0.5412,
      "step": 1601
    },
    {
      "epoch": 0.2731923601637108,
      "grad_norm": 0.4062732458114624,
      "learning_rate": 1.7059341646939666e-06,
      "loss": 0.5332,
      "step": 1602
    },
    {
      "epoch": 0.27336289222373805,
      "grad_norm": 0.45208001136779785,
      "learning_rate": 1.7055428616708384e-06,
      "loss": 0.5463,
      "step": 1603
    },
    {
      "epoch": 0.27353342428376537,
      "grad_norm": 0.44971469044685364,
      "learning_rate": 1.7051513434171375e-06,
      "loss": 0.5347,
      "step": 1604
    },
    {
      "epoch": 0.27370395634379263,
      "grad_norm": 0.4267110228538513,
      "learning_rate": 1.7047596100522994e-06,
      "loss": 0.5314,
      "step": 1605
    },
    {
      "epoch": 0.2738744884038199,
      "grad_norm": 0.42136669158935547,
      "learning_rate": 1.7043676616958247e-06,
      "loss": 0.5323,
      "step": 1606
    },
    {
      "epoch": 0.2740450204638472,
      "grad_norm": 0.53708815574646,
      "learning_rate": 1.7039754984672799e-06,
      "loss": 0.5391,
      "step": 1607
    },
    {
      "epoch": 0.2742155525238745,
      "grad_norm": 0.3569575846195221,
      "learning_rate": 1.7035831204862973e-06,
      "loss": 0.5404,
      "step": 1608
    },
    {
      "epoch": 0.2743860845839018,
      "grad_norm": 0.45229849219322205,
      "learning_rate": 1.703190527872574e-06,
      "loss": 0.5329,
      "step": 1609
    },
    {
      "epoch": 0.27455661664392905,
      "grad_norm": 0.5555593371391296,
      "learning_rate": 1.7027977207458733e-06,
      "loss": 0.52,
      "step": 1610
    },
    {
      "epoch": 0.27472714870395637,
      "grad_norm": 0.45333659648895264,
      "learning_rate": 1.702404699226024e-06,
      "loss": 0.5447,
      "step": 1611
    },
    {
      "epoch": 0.27489768076398363,
      "grad_norm": 0.37671902775764465,
      "learning_rate": 1.7020114634329191e-06,
      "loss": 0.5257,
      "step": 1612
    },
    {
      "epoch": 0.2750682128240109,
      "grad_norm": 0.5203356146812439,
      "learning_rate": 1.7016180134865183e-06,
      "loss": 0.5417,
      "step": 1613
    },
    {
      "epoch": 0.2752387448840382,
      "grad_norm": 0.44975242018699646,
      "learning_rate": 1.7012243495068463e-06,
      "loss": 0.5354,
      "step": 1614
    },
    {
      "epoch": 0.2754092769440655,
      "grad_norm": 0.4513874053955078,
      "learning_rate": 1.7008304716139924e-06,
      "loss": 0.5252,
      "step": 1615
    },
    {
      "epoch": 0.2755798090040928,
      "grad_norm": 0.3808591067790985,
      "learning_rate": 1.7004363799281122e-06,
      "loss": 0.5302,
      "step": 1616
    },
    {
      "epoch": 0.27575034106412005,
      "grad_norm": 0.4400796890258789,
      "learning_rate": 1.7000420745694257e-06,
      "loss": 0.5325,
      "step": 1617
    },
    {
      "epoch": 0.2759208731241473,
      "grad_norm": 0.4776574373245239,
      "learning_rate": 1.6996475556582185e-06,
      "loss": 0.5363,
      "step": 1618
    },
    {
      "epoch": 0.27609140518417463,
      "grad_norm": 0.3435958921909332,
      "learning_rate": 1.6992528233148416e-06,
      "loss": 0.5414,
      "step": 1619
    },
    {
      "epoch": 0.2762619372442019,
      "grad_norm": 0.41727060079574585,
      "learning_rate": 1.6988578776597096e-06,
      "loss": 0.5581,
      "step": 1620
    },
    {
      "epoch": 0.2764324693042292,
      "grad_norm": 0.37507522106170654,
      "learning_rate": 1.6984627188133047e-06,
      "loss": 0.5399,
      "step": 1621
    },
    {
      "epoch": 0.2766030013642565,
      "grad_norm": 0.4046799838542938,
      "learning_rate": 1.6980673468961719e-06,
      "loss": 0.5316,
      "step": 1622
    },
    {
      "epoch": 0.2767735334242838,
      "grad_norm": 0.569532573223114,
      "learning_rate": 1.6976717620289223e-06,
      "loss": 0.5307,
      "step": 1623
    },
    {
      "epoch": 0.27694406548431105,
      "grad_norm": 0.42097124457359314,
      "learning_rate": 1.6972759643322316e-06,
      "loss": 0.5297,
      "step": 1624
    },
    {
      "epoch": 0.2771145975443383,
      "grad_norm": 0.5242999792098999,
      "learning_rate": 1.6968799539268408e-06,
      "loss": 0.5553,
      "step": 1625
    },
    {
      "epoch": 0.27728512960436563,
      "grad_norm": 0.41940924525260925,
      "learning_rate": 1.6964837309335555e-06,
      "loss": 0.5299,
      "step": 1626
    },
    {
      "epoch": 0.2774556616643929,
      "grad_norm": 0.48386019468307495,
      "learning_rate": 1.696087295473246e-06,
      "loss": 0.5394,
      "step": 1627
    },
    {
      "epoch": 0.2776261937244202,
      "grad_norm": 0.601997971534729,
      "learning_rate": 1.6956906476668474e-06,
      "loss": 0.5269,
      "step": 1628
    },
    {
      "epoch": 0.2777967257844475,
      "grad_norm": 0.35025733709335327,
      "learning_rate": 1.6952937876353597e-06,
      "loss": 0.5362,
      "step": 1629
    },
    {
      "epoch": 0.27796725784447474,
      "grad_norm": 0.4714794456958771,
      "learning_rate": 1.6948967154998487e-06,
      "loss": 0.533,
      "step": 1630
    },
    {
      "epoch": 0.27813778990450205,
      "grad_norm": 0.35530272126197815,
      "learning_rate": 1.6944994313814429e-06,
      "loss": 0.5377,
      "step": 1631
    },
    {
      "epoch": 0.2783083219645293,
      "grad_norm": 0.5158945918083191,
      "learning_rate": 1.694101935401337e-06,
      "loss": 0.5358,
      "step": 1632
    },
    {
      "epoch": 0.27847885402455663,
      "grad_norm": 0.36836668848991394,
      "learning_rate": 1.6937042276807894e-06,
      "loss": 0.5408,
      "step": 1633
    },
    {
      "epoch": 0.2786493860845839,
      "grad_norm": 0.5216159224510193,
      "learning_rate": 1.6933063083411236e-06,
      "loss": 0.5266,
      "step": 1634
    },
    {
      "epoch": 0.2788199181446112,
      "grad_norm": 0.746419370174408,
      "learning_rate": 1.6929081775037278e-06,
      "loss": 0.5368,
      "step": 1635
    },
    {
      "epoch": 0.2789904502046385,
      "grad_norm": 0.4515080153942108,
      "learning_rate": 1.6925098352900545e-06,
      "loss": 0.5255,
      "step": 1636
    },
    {
      "epoch": 0.27916098226466574,
      "grad_norm": 0.5471440553665161,
      "learning_rate": 1.6921112818216203e-06,
      "loss": 0.5347,
      "step": 1637
    },
    {
      "epoch": 0.27933151432469305,
      "grad_norm": 0.4010627865791321,
      "learning_rate": 1.6917125172200067e-06,
      "loss": 0.545,
      "step": 1638
    },
    {
      "epoch": 0.2795020463847203,
      "grad_norm": 0.5758441686630249,
      "learning_rate": 1.6913135416068598e-06,
      "loss": 0.543,
      "step": 1639
    },
    {
      "epoch": 0.27967257844474763,
      "grad_norm": 0.39412951469421387,
      "learning_rate": 1.6909143551038897e-06,
      "loss": 0.5276,
      "step": 1640
    },
    {
      "epoch": 0.2798431105047749,
      "grad_norm": 0.6113039255142212,
      "learning_rate": 1.6905149578328704e-06,
      "loss": 0.5461,
      "step": 1641
    },
    {
      "epoch": 0.28001364256480216,
      "grad_norm": 0.5534077286720276,
      "learning_rate": 1.6901153499156416e-06,
      "loss": 0.5343,
      "step": 1642
    },
    {
      "epoch": 0.2801841746248295,
      "grad_norm": 0.3831448256969452,
      "learning_rate": 1.689715531474106e-06,
      "loss": 0.542,
      "step": 1643
    },
    {
      "epoch": 0.28035470668485674,
      "grad_norm": 0.5919682383537292,
      "learning_rate": 1.6893155026302302e-06,
      "loss": 0.526,
      "step": 1644
    },
    {
      "epoch": 0.28052523874488405,
      "grad_norm": 0.3408511281013489,
      "learning_rate": 1.6889152635060467e-06,
      "loss": 0.5294,
      "step": 1645
    },
    {
      "epoch": 0.2806957708049113,
      "grad_norm": 0.49439018964767456,
      "learning_rate": 1.6885148142236505e-06,
      "loss": 0.5378,
      "step": 1646
    },
    {
      "epoch": 0.28086630286493863,
      "grad_norm": 0.4036789834499359,
      "learning_rate": 1.6881141549052015e-06,
      "loss": 0.5322,
      "step": 1647
    },
    {
      "epoch": 0.2810368349249659,
      "grad_norm": 0.4894874095916748,
      "learning_rate": 1.6877132856729238e-06,
      "loss": 0.5385,
      "step": 1648
    },
    {
      "epoch": 0.28120736698499316,
      "grad_norm": 0.713665783405304,
      "learning_rate": 1.6873122066491048e-06,
      "loss": 0.5356,
      "step": 1649
    },
    {
      "epoch": 0.2813778990450205,
      "grad_norm": 0.4714016020298004,
      "learning_rate": 1.6869109179560965e-06,
      "loss": 0.5326,
      "step": 1650
    },
    {
      "epoch": 0.28154843110504774,
      "grad_norm": 0.5672765374183655,
      "learning_rate": 1.6865094197163148e-06,
      "loss": 0.5237,
      "step": 1651
    },
    {
      "epoch": 0.28171896316507505,
      "grad_norm": 0.4787973463535309,
      "learning_rate": 1.6861077120522391e-06,
      "loss": 0.5304,
      "step": 1652
    },
    {
      "epoch": 0.2818894952251023,
      "grad_norm": 0.5834455490112305,
      "learning_rate": 1.6857057950864134e-06,
      "loss": 0.5387,
      "step": 1653
    },
    {
      "epoch": 0.2820600272851296,
      "grad_norm": 0.44850626587867737,
      "learning_rate": 1.6853036689414449e-06,
      "loss": 0.534,
      "step": 1654
    },
    {
      "epoch": 0.2822305593451569,
      "grad_norm": 0.4338832497596741,
      "learning_rate": 1.6849013337400053e-06,
      "loss": 0.5214,
      "step": 1655
    },
    {
      "epoch": 0.28240109140518416,
      "grad_norm": 0.4489261507987976,
      "learning_rate": 1.684498789604829e-06,
      "loss": 0.5397,
      "step": 1656
    },
    {
      "epoch": 0.2825716234652115,
      "grad_norm": 0.47586989402770996,
      "learning_rate": 1.6840960366587152e-06,
      "loss": 0.5275,
      "step": 1657
    },
    {
      "epoch": 0.28274215552523874,
      "grad_norm": 0.4325336515903473,
      "learning_rate": 1.6836930750245265e-06,
      "loss": 0.5262,
      "step": 1658
    },
    {
      "epoch": 0.28291268758526605,
      "grad_norm": 0.39559781551361084,
      "learning_rate": 1.6832899048251887e-06,
      "loss": 0.5303,
      "step": 1659
    },
    {
      "epoch": 0.2830832196452933,
      "grad_norm": 0.44162189960479736,
      "learning_rate": 1.6828865261836919e-06,
      "loss": 0.5194,
      "step": 1660
    },
    {
      "epoch": 0.2832537517053206,
      "grad_norm": 0.4153518080711365,
      "learning_rate": 1.6824829392230888e-06,
      "loss": 0.524,
      "step": 1661
    },
    {
      "epoch": 0.2834242837653479,
      "grad_norm": 0.4147271513938904,
      "learning_rate": 1.682079144066497e-06,
      "loss": 0.5323,
      "step": 1662
    },
    {
      "epoch": 0.28359481582537516,
      "grad_norm": 0.5544766783714294,
      "learning_rate": 1.6816751408370968e-06,
      "loss": 0.5238,
      "step": 1663
    },
    {
      "epoch": 0.2837653478854025,
      "grad_norm": 0.4279305338859558,
      "learning_rate": 1.6812709296581317e-06,
      "loss": 0.5344,
      "step": 1664
    },
    {
      "epoch": 0.28393587994542974,
      "grad_norm": 0.3990345001220703,
      "learning_rate": 1.6808665106529094e-06,
      "loss": 0.5256,
      "step": 1665
    },
    {
      "epoch": 0.284106412005457,
      "grad_norm": 0.4134671092033386,
      "learning_rate": 1.680461883944801e-06,
      "loss": 0.5281,
      "step": 1666
    },
    {
      "epoch": 0.2842769440654843,
      "grad_norm": 0.43525567650794983,
      "learning_rate": 1.6800570496572397e-06,
      "loss": 0.5351,
      "step": 1667
    },
    {
      "epoch": 0.2844474761255116,
      "grad_norm": 0.4455080032348633,
      "learning_rate": 1.6796520079137233e-06,
      "loss": 0.5217,
      "step": 1668
    },
    {
      "epoch": 0.2846180081855389,
      "grad_norm": 0.3639788031578064,
      "learning_rate": 1.6792467588378125e-06,
      "loss": 0.5241,
      "step": 1669
    },
    {
      "epoch": 0.28478854024556616,
      "grad_norm": 0.38478967547416687,
      "learning_rate": 1.6788413025531312e-06,
      "loss": 0.5272,
      "step": 1670
    },
    {
      "epoch": 0.2849590723055935,
      "grad_norm": 0.339579701423645,
      "learning_rate": 1.6784356391833666e-06,
      "loss": 0.527,
      "step": 1671
    },
    {
      "epoch": 0.28512960436562074,
      "grad_norm": 0.4318312406539917,
      "learning_rate": 1.6780297688522691e-06,
      "loss": 0.5237,
      "step": 1672
    },
    {
      "epoch": 0.285300136425648,
      "grad_norm": 0.4525909125804901,
      "learning_rate": 1.677623691683652e-06,
      "loss": 0.5274,
      "step": 1673
    },
    {
      "epoch": 0.2854706684856753,
      "grad_norm": 0.36183860898017883,
      "learning_rate": 1.6772174078013922e-06,
      "loss": 0.5281,
      "step": 1674
    },
    {
      "epoch": 0.2856412005457026,
      "grad_norm": 0.4849749505519867,
      "learning_rate": 1.676810917329429e-06,
      "loss": 0.5271,
      "step": 1675
    },
    {
      "epoch": 0.2858117326057299,
      "grad_norm": 0.7375707626342773,
      "learning_rate": 1.676404220391765e-06,
      "loss": 0.5312,
      "step": 1676
    },
    {
      "epoch": 0.28598226466575716,
      "grad_norm": 0.6052662134170532,
      "learning_rate": 1.675997317112466e-06,
      "loss": 0.5216,
      "step": 1677
    },
    {
      "epoch": 0.2861527967257845,
      "grad_norm": 0.36949390172958374,
      "learning_rate": 1.6755902076156607e-06,
      "loss": 0.5353,
      "step": 1678
    },
    {
      "epoch": 0.28632332878581174,
      "grad_norm": 0.4570222795009613,
      "learning_rate": 1.6751828920255398e-06,
      "loss": 0.5299,
      "step": 1679
    },
    {
      "epoch": 0.286493860845839,
      "grad_norm": 0.5875330567359924,
      "learning_rate": 1.6747753704663586e-06,
      "loss": 0.5316,
      "step": 1680
    },
    {
      "epoch": 0.2866643929058663,
      "grad_norm": 0.40268129110336304,
      "learning_rate": 1.674367643062434e-06,
      "loss": 0.5277,
      "step": 1681
    },
    {
      "epoch": 0.2868349249658936,
      "grad_norm": 0.39442431926727295,
      "learning_rate": 1.6739597099381458e-06,
      "loss": 0.5238,
      "step": 1682
    },
    {
      "epoch": 0.2870054570259209,
      "grad_norm": 0.4880406856536865,
      "learning_rate": 1.673551571217937e-06,
      "loss": 0.5259,
      "step": 1683
    },
    {
      "epoch": 0.28717598908594816,
      "grad_norm": 0.4166980981826782,
      "learning_rate": 1.6731432270263128e-06,
      "loss": 0.5368,
      "step": 1684
    },
    {
      "epoch": 0.2873465211459754,
      "grad_norm": 0.33825594186782837,
      "learning_rate": 1.672734677487841e-06,
      "loss": 0.5261,
      "step": 1685
    },
    {
      "epoch": 0.28751705320600274,
      "grad_norm": 0.5011537671089172,
      "learning_rate": 1.672325922727153e-06,
      "loss": 0.5179,
      "step": 1686
    },
    {
      "epoch": 0.28768758526603,
      "grad_norm": 0.4549483358860016,
      "learning_rate": 1.6719169628689422e-06,
      "loss": 0.5306,
      "step": 1687
    },
    {
      "epoch": 0.2878581173260573,
      "grad_norm": 0.35790887475013733,
      "learning_rate": 1.671507798037964e-06,
      "loss": 0.5369,
      "step": 1688
    },
    {
      "epoch": 0.2880286493860846,
      "grad_norm": 0.5589377284049988,
      "learning_rate": 1.6710984283590371e-06,
      "loss": 0.5297,
      "step": 1689
    },
    {
      "epoch": 0.2881991814461119,
      "grad_norm": 0.5511754751205444,
      "learning_rate": 1.6706888539570427e-06,
      "loss": 0.5398,
      "step": 1690
    },
    {
      "epoch": 0.28836971350613916,
      "grad_norm": 0.4057330787181854,
      "learning_rate": 1.6702790749569239e-06,
      "loss": 0.5349,
      "step": 1691
    },
    {
      "epoch": 0.2885402455661664,
      "grad_norm": 0.4141501784324646,
      "learning_rate": 1.6698690914836864e-06,
      "loss": 0.5238,
      "step": 1692
    },
    {
      "epoch": 0.28871077762619374,
      "grad_norm": 0.44292518496513367,
      "learning_rate": 1.669458903662399e-06,
      "loss": 0.5206,
      "step": 1693
    },
    {
      "epoch": 0.288881309686221,
      "grad_norm": 0.4296228587627411,
      "learning_rate": 1.6690485116181917e-06,
      "loss": 0.5321,
      "step": 1694
    },
    {
      "epoch": 0.2890518417462483,
      "grad_norm": 0.38201814889907837,
      "learning_rate": 1.6686379154762576e-06,
      "loss": 0.5326,
      "step": 1695
    },
    {
      "epoch": 0.2892223738062756,
      "grad_norm": 0.5104760527610779,
      "learning_rate": 1.6682271153618518e-06,
      "loss": 0.5337,
      "step": 1696
    },
    {
      "epoch": 0.28939290586630284,
      "grad_norm": 0.4310562014579773,
      "learning_rate": 1.6678161114002917e-06,
      "loss": 0.5315,
      "step": 1697
    },
    {
      "epoch": 0.28956343792633016,
      "grad_norm": 0.3841532766819,
      "learning_rate": 1.6674049037169564e-06,
      "loss": 0.5177,
      "step": 1698
    },
    {
      "epoch": 0.2897339699863574,
      "grad_norm": 0.4063761532306671,
      "learning_rate": 1.6669934924372882e-06,
      "loss": 0.5239,
      "step": 1699
    },
    {
      "epoch": 0.28990450204638474,
      "grad_norm": 0.40579575300216675,
      "learning_rate": 1.66658187768679e-06,
      "loss": 0.5307,
      "step": 1700
    },
    {
      "epoch": 0.290075034106412,
      "grad_norm": 0.3862999379634857,
      "learning_rate": 1.6661700595910286e-06,
      "loss": 0.5293,
      "step": 1701
    },
    {
      "epoch": 0.2902455661664393,
      "grad_norm": 0.3648364841938019,
      "learning_rate": 1.6657580382756307e-06,
      "loss": 0.5191,
      "step": 1702
    },
    {
      "epoch": 0.2904160982264666,
      "grad_norm": 0.48320475220680237,
      "learning_rate": 1.6653458138662873e-06,
      "loss": 0.5388,
      "step": 1703
    },
    {
      "epoch": 0.29058663028649384,
      "grad_norm": 0.39256301522254944,
      "learning_rate": 1.66493338648875e-06,
      "loss": 0.5405,
      "step": 1704
    },
    {
      "epoch": 0.29075716234652116,
      "grad_norm": 0.39358004927635193,
      "learning_rate": 1.6645207562688322e-06,
      "loss": 0.5257,
      "step": 1705
    },
    {
      "epoch": 0.2909276944065484,
      "grad_norm": 0.42495185136795044,
      "learning_rate": 1.6641079233324093e-06,
      "loss": 0.5398,
      "step": 1706
    },
    {
      "epoch": 0.29109822646657574,
      "grad_norm": 0.5367429256439209,
      "learning_rate": 1.6636948878054193e-06,
      "loss": 0.5191,
      "step": 1707
    },
    {
      "epoch": 0.291268758526603,
      "grad_norm": 0.5471576452255249,
      "learning_rate": 1.6632816498138613e-06,
      "loss": 0.5229,
      "step": 1708
    },
    {
      "epoch": 0.29143929058663026,
      "grad_norm": 0.43130508065223694,
      "learning_rate": 1.662868209483796e-06,
      "loss": 0.5301,
      "step": 1709
    },
    {
      "epoch": 0.2916098226466576,
      "grad_norm": 0.3890998959541321,
      "learning_rate": 1.6624545669413462e-06,
      "loss": 0.5242,
      "step": 1710
    },
    {
      "epoch": 0.29178035470668484,
      "grad_norm": 0.3883855640888214,
      "learning_rate": 1.6620407223126969e-06,
      "loss": 0.5204,
      "step": 1711
    },
    {
      "epoch": 0.29195088676671216,
      "grad_norm": 0.3944658637046814,
      "learning_rate": 1.6616266757240935e-06,
      "loss": 0.5347,
      "step": 1712
    },
    {
      "epoch": 0.2921214188267394,
      "grad_norm": 0.45826256275177,
      "learning_rate": 1.6612124273018442e-06,
      "loss": 0.5346,
      "step": 1713
    },
    {
      "epoch": 0.29229195088676674,
      "grad_norm": 0.3530188798904419,
      "learning_rate": 1.6607979771723176e-06,
      "loss": 0.5333,
      "step": 1714
    },
    {
      "epoch": 0.292462482946794,
      "grad_norm": 0.4453744888305664,
      "learning_rate": 1.660383325461945e-06,
      "loss": 0.5352,
      "step": 1715
    },
    {
      "epoch": 0.29263301500682126,
      "grad_norm": 0.4675208628177643,
      "learning_rate": 1.659968472297219e-06,
      "loss": 0.5327,
      "step": 1716
    },
    {
      "epoch": 0.2928035470668486,
      "grad_norm": 0.5276213884353638,
      "learning_rate": 1.6595534178046928e-06,
      "loss": 0.5325,
      "step": 1717
    },
    {
      "epoch": 0.29297407912687584,
      "grad_norm": 0.3921293020248413,
      "learning_rate": 1.659138162110981e-06,
      "loss": 0.5468,
      "step": 1718
    },
    {
      "epoch": 0.29314461118690316,
      "grad_norm": 0.40789127349853516,
      "learning_rate": 1.6587227053427614e-06,
      "loss": 0.5253,
      "step": 1719
    },
    {
      "epoch": 0.2933151432469304,
      "grad_norm": 0.5174045562744141,
      "learning_rate": 1.6583070476267713e-06,
      "loss": 0.5344,
      "step": 1720
    },
    {
      "epoch": 0.2934856753069577,
      "grad_norm": 0.44151756167411804,
      "learning_rate": 1.6578911890898096e-06,
      "loss": 0.5242,
      "step": 1721
    },
    {
      "epoch": 0.293656207366985,
      "grad_norm": 0.38198962807655334,
      "learning_rate": 1.6574751298587371e-06,
      "loss": 0.5176,
      "step": 1722
    },
    {
      "epoch": 0.29382673942701226,
      "grad_norm": 0.4616943299770355,
      "learning_rate": 1.6570588700604753e-06,
      "loss": 0.519,
      "step": 1723
    },
    {
      "epoch": 0.2939972714870396,
      "grad_norm": 0.5245501399040222,
      "learning_rate": 1.6566424098220065e-06,
      "loss": 0.5216,
      "step": 1724
    },
    {
      "epoch": 0.29416780354706684,
      "grad_norm": 0.4258287847042084,
      "learning_rate": 1.656225749270376e-06,
      "loss": 0.5366,
      "step": 1725
    },
    {
      "epoch": 0.29433833560709416,
      "grad_norm": 0.49658289551734924,
      "learning_rate": 1.6558088885326877e-06,
      "loss": 0.5326,
      "step": 1726
    },
    {
      "epoch": 0.2945088676671214,
      "grad_norm": 0.5772252082824707,
      "learning_rate": 1.6553918277361082e-06,
      "loss": 0.5313,
      "step": 1727
    },
    {
      "epoch": 0.2946793997271487,
      "grad_norm": 0.47091925144195557,
      "learning_rate": 1.654974567007865e-06,
      "loss": 0.5276,
      "step": 1728
    },
    {
      "epoch": 0.294849931787176,
      "grad_norm": 0.4931386709213257,
      "learning_rate": 1.6545571064752455e-06,
      "loss": 0.5436,
      "step": 1729
    },
    {
      "epoch": 0.29502046384720326,
      "grad_norm": 0.5427204370498657,
      "learning_rate": 1.6541394462655996e-06,
      "loss": 0.5366,
      "step": 1730
    },
    {
      "epoch": 0.2951909959072306,
      "grad_norm": 0.6340658664703369,
      "learning_rate": 1.6537215865063367e-06,
      "loss": 0.518,
      "step": 1731
    },
    {
      "epoch": 0.29536152796725784,
      "grad_norm": 0.43148940801620483,
      "learning_rate": 1.653303527324928e-06,
      "loss": 0.5288,
      "step": 1732
    },
    {
      "epoch": 0.2955320600272851,
      "grad_norm": 0.47688204050064087,
      "learning_rate": 1.6528852688489055e-06,
      "loss": 0.5278,
      "step": 1733
    },
    {
      "epoch": 0.2957025920873124,
      "grad_norm": 0.6697880029678345,
      "learning_rate": 1.6524668112058619e-06,
      "loss": 0.5333,
      "step": 1734
    },
    {
      "epoch": 0.2958731241473397,
      "grad_norm": 0.5715264081954956,
      "learning_rate": 1.6520481545234496e-06,
      "loss": 0.5413,
      "step": 1735
    },
    {
      "epoch": 0.296043656207367,
      "grad_norm": 0.3737260401248932,
      "learning_rate": 1.6516292989293836e-06,
      "loss": 0.5244,
      "step": 1736
    },
    {
      "epoch": 0.29621418826739426,
      "grad_norm": 0.40294066071510315,
      "learning_rate": 1.6512102445514376e-06,
      "loss": 0.5339,
      "step": 1737
    },
    {
      "epoch": 0.2963847203274216,
      "grad_norm": 0.3857491910457611,
      "learning_rate": 1.6507909915174477e-06,
      "loss": 0.5286,
      "step": 1738
    },
    {
      "epoch": 0.29655525238744884,
      "grad_norm": 0.3942945599555969,
      "learning_rate": 1.65037153995531e-06,
      "loss": 0.5307,
      "step": 1739
    },
    {
      "epoch": 0.2967257844474761,
      "grad_norm": 0.429793119430542,
      "learning_rate": 1.6499518899929803e-06,
      "loss": 0.5377,
      "step": 1740
    },
    {
      "epoch": 0.2968963165075034,
      "grad_norm": 0.427202969789505,
      "learning_rate": 1.6495320417584758e-06,
      "loss": 0.5402,
      "step": 1741
    },
    {
      "epoch": 0.2970668485675307,
      "grad_norm": 0.3815003037452698,
      "learning_rate": 1.6491119953798748e-06,
      "loss": 0.5378,
      "step": 1742
    },
    {
      "epoch": 0.297237380627558,
      "grad_norm": 0.4911755919456482,
      "learning_rate": 1.6486917509853142e-06,
      "loss": 0.5348,
      "step": 1743
    },
    {
      "epoch": 0.29740791268758526,
      "grad_norm": 0.535512387752533,
      "learning_rate": 1.648271308702993e-06,
      "loss": 0.5341,
      "step": 1744
    },
    {
      "epoch": 0.2975784447476125,
      "grad_norm": 0.4352928698062897,
      "learning_rate": 1.64785066866117e-06,
      "loss": 0.528,
      "step": 1745
    },
    {
      "epoch": 0.29774897680763984,
      "grad_norm": 0.4337961673736572,
      "learning_rate": 1.6474298309881636e-06,
      "loss": 0.5352,
      "step": 1746
    },
    {
      "epoch": 0.2979195088676671,
      "grad_norm": 0.5219410061836243,
      "learning_rate": 1.6470087958123536e-06,
      "loss": 0.5308,
      "step": 1747
    },
    {
      "epoch": 0.2980900409276944,
      "grad_norm": 0.45884940028190613,
      "learning_rate": 1.6465875632621797e-06,
      "loss": 0.5304,
      "step": 1748
    },
    {
      "epoch": 0.2982605729877217,
      "grad_norm": 0.4260612726211548,
      "learning_rate": 1.6461661334661416e-06,
      "loss": 0.5395,
      "step": 1749
    },
    {
      "epoch": 0.298431105047749,
      "grad_norm": 0.4612407088279724,
      "learning_rate": 1.645744506552799e-06,
      "loss": 0.5343,
      "step": 1750
    },
    {
      "epoch": 0.29860163710777626,
      "grad_norm": 0.5481771230697632,
      "learning_rate": 1.6453226826507723e-06,
      "loss": 0.5383,
      "step": 1751
    },
    {
      "epoch": 0.2987721691678035,
      "grad_norm": 0.4502001702785492,
      "learning_rate": 1.6449006618887422e-06,
      "loss": 0.534,
      "step": 1752
    },
    {
      "epoch": 0.29894270122783084,
      "grad_norm": 0.48182806372642517,
      "learning_rate": 1.6444784443954482e-06,
      "loss": 0.5359,
      "step": 1753
    },
    {
      "epoch": 0.2991132332878581,
      "grad_norm": 0.4174640476703644,
      "learning_rate": 1.6440560302996907e-06,
      "loss": 0.5341,
      "step": 1754
    },
    {
      "epoch": 0.2992837653478854,
      "grad_norm": 0.5465035438537598,
      "learning_rate": 1.6436334197303298e-06,
      "loss": 0.5379,
      "step": 1755
    },
    {
      "epoch": 0.2994542974079127,
      "grad_norm": 0.41326817870140076,
      "learning_rate": 1.6432106128162862e-06,
      "loss": 0.5295,
      "step": 1756
    },
    {
      "epoch": 0.29962482946793995,
      "grad_norm": 0.3652679920196533,
      "learning_rate": 1.6427876096865395e-06,
      "loss": 0.5207,
      "step": 1757
    },
    {
      "epoch": 0.29979536152796726,
      "grad_norm": 0.3553360104560852,
      "learning_rate": 1.6423644104701302e-06,
      "loss": 0.537,
      "step": 1758
    },
    {
      "epoch": 0.2999658935879945,
      "grad_norm": 0.3630985915660858,
      "learning_rate": 1.6419410152961574e-06,
      "loss": 0.537,
      "step": 1759
    },
    {
      "epoch": 0.30013642564802184,
      "grad_norm": 0.41660600900650024,
      "learning_rate": 1.641517424293781e-06,
      "loss": 0.5434,
      "step": 1760
    },
    {
      "epoch": 0.3003069577080491,
      "grad_norm": 0.4603380858898163,
      "learning_rate": 1.6410936375922204e-06,
      "loss": 0.544,
      "step": 1761
    },
    {
      "epoch": 0.3004774897680764,
      "grad_norm": 0.35790368914604187,
      "learning_rate": 1.6406696553207544e-06,
      "loss": 0.5298,
      "step": 1762
    },
    {
      "epoch": 0.3006480218281037,
      "grad_norm": 0.35639557242393494,
      "learning_rate": 1.6402454776087214e-06,
      "loss": 0.534,
      "step": 1763
    },
    {
      "epoch": 0.30081855388813095,
      "grad_norm": 0.4377691149711609,
      "learning_rate": 1.6398211045855198e-06,
      "loss": 0.525,
      "step": 1764
    },
    {
      "epoch": 0.30098908594815826,
      "grad_norm": 0.33262577652931213,
      "learning_rate": 1.6393965363806075e-06,
      "loss": 0.5294,
      "step": 1765
    },
    {
      "epoch": 0.3011596180081855,
      "grad_norm": 0.4677214026451111,
      "learning_rate": 1.6389717731235022e-06,
      "loss": 0.5274,
      "step": 1766
    },
    {
      "epoch": 0.30133015006821284,
      "grad_norm": 0.5282312035560608,
      "learning_rate": 1.63854681494378e-06,
      "loss": 0.5314,
      "step": 1767
    },
    {
      "epoch": 0.3015006821282401,
      "grad_norm": 0.3520287573337555,
      "learning_rate": 1.6381216619710781e-06,
      "loss": 0.5272,
      "step": 1768
    },
    {
      "epoch": 0.30167121418826737,
      "grad_norm": 0.5878857374191284,
      "learning_rate": 1.6376963143350915e-06,
      "loss": 0.5206,
      "step": 1769
    },
    {
      "epoch": 0.3018417462482947,
      "grad_norm": 0.47125834226608276,
      "learning_rate": 1.6372707721655757e-06,
      "loss": 0.5354,
      "step": 1770
    },
    {
      "epoch": 0.30201227830832195,
      "grad_norm": 0.4633807837963104,
      "learning_rate": 1.6368450355923453e-06,
      "loss": 0.5275,
      "step": 1771
    },
    {
      "epoch": 0.30218281036834926,
      "grad_norm": 0.5724704265594482,
      "learning_rate": 1.6364191047452738e-06,
      "loss": 0.532,
      "step": 1772
    },
    {
      "epoch": 0.3023533424283765,
      "grad_norm": 0.5451125502586365,
      "learning_rate": 1.6359929797542946e-06,
      "loss": 0.534,
      "step": 1773
    },
    {
      "epoch": 0.30252387448840384,
      "grad_norm": 0.5832334756851196,
      "learning_rate": 1.6355666607493995e-06,
      "loss": 0.535,
      "step": 1774
    },
    {
      "epoch": 0.3026944065484311,
      "grad_norm": 0.3607366979122162,
      "learning_rate": 1.6351401478606406e-06,
      "loss": 0.5342,
      "step": 1775
    },
    {
      "epoch": 0.30286493860845837,
      "grad_norm": 0.5791069269180298,
      "learning_rate": 1.634713441218128e-06,
      "loss": 0.5453,
      "step": 1776
    },
    {
      "epoch": 0.3030354706684857,
      "grad_norm": 0.5250142812728882,
      "learning_rate": 1.6342865409520318e-06,
      "loss": 0.5248,
      "step": 1777
    },
    {
      "epoch": 0.30320600272851295,
      "grad_norm": 0.34585243463516235,
      "learning_rate": 1.6338594471925802e-06,
      "loss": 0.5264,
      "step": 1778
    },
    {
      "epoch": 0.30337653478854026,
      "grad_norm": 0.5033993124961853,
      "learning_rate": 1.6334321600700616e-06,
      "loss": 0.5311,
      "step": 1779
    },
    {
      "epoch": 0.3035470668485675,
      "grad_norm": 0.40762490034103394,
      "learning_rate": 1.6330046797148225e-06,
      "loss": 0.5348,
      "step": 1780
    },
    {
      "epoch": 0.30371759890859484,
      "grad_norm": 0.52232426404953,
      "learning_rate": 1.6325770062572688e-06,
      "loss": 0.528,
      "step": 1781
    },
    {
      "epoch": 0.3038881309686221,
      "grad_norm": 0.4686141312122345,
      "learning_rate": 1.6321491398278647e-06,
      "loss": 0.5323,
      "step": 1782
    },
    {
      "epoch": 0.30405866302864937,
      "grad_norm": 0.5295236110687256,
      "learning_rate": 1.6317210805571349e-06,
      "loss": 0.5195,
      "step": 1783
    },
    {
      "epoch": 0.3042291950886767,
      "grad_norm": 0.7101245522499084,
      "learning_rate": 1.63129282857566e-06,
      "loss": 0.5341,
      "step": 1784
    },
    {
      "epoch": 0.30439972714870395,
      "grad_norm": 0.5113368034362793,
      "learning_rate": 1.6308643840140829e-06,
      "loss": 0.5356,
      "step": 1785
    },
    {
      "epoch": 0.30457025920873126,
      "grad_norm": 0.4097968339920044,
      "learning_rate": 1.6304357470031021e-06,
      "loss": 0.5405,
      "step": 1786
    },
    {
      "epoch": 0.3047407912687585,
      "grad_norm": 0.45384204387664795,
      "learning_rate": 1.6300069176734768e-06,
      "loss": 0.549,
      "step": 1787
    },
    {
      "epoch": 0.3049113233287858,
      "grad_norm": 0.5580360889434814,
      "learning_rate": 1.6295778961560244e-06,
      "loss": 0.539,
      "step": 1788
    },
    {
      "epoch": 0.3050818553888131,
      "grad_norm": 0.3851800262928009,
      "learning_rate": 1.6291486825816207e-06,
      "loss": 0.5301,
      "step": 1789
    },
    {
      "epoch": 0.30525238744884037,
      "grad_norm": 0.45921897888183594,
      "learning_rate": 1.6287192770811997e-06,
      "loss": 0.5577,
      "step": 1790
    },
    {
      "epoch": 0.3054229195088677,
      "grad_norm": 0.469555139541626,
      "learning_rate": 1.6282896797857549e-06,
      "loss": 0.5355,
      "step": 1791
    },
    {
      "epoch": 0.30559345156889495,
      "grad_norm": 0.46167778968811035,
      "learning_rate": 1.6278598908263377e-06,
      "loss": 0.5261,
      "step": 1792
    },
    {
      "epoch": 0.30576398362892226,
      "grad_norm": 0.43485838174819946,
      "learning_rate": 1.627429910334058e-06,
      "loss": 0.5255,
      "step": 1793
    },
    {
      "epoch": 0.3059345156889495,
      "grad_norm": 0.47659534215927124,
      "learning_rate": 1.6269997384400846e-06,
      "loss": 0.5459,
      "step": 1794
    },
    {
      "epoch": 0.3061050477489768,
      "grad_norm": 0.5137026906013489,
      "learning_rate": 1.626569375275644e-06,
      "loss": 0.5347,
      "step": 1795
    },
    {
      "epoch": 0.3062755798090041,
      "grad_norm": 0.4063645899295807,
      "learning_rate": 1.6261388209720213e-06,
      "loss": 0.5282,
      "step": 1796
    },
    {
      "epoch": 0.30644611186903137,
      "grad_norm": 0.4230020344257355,
      "learning_rate": 1.6257080756605603e-06,
      "loss": 0.528,
      "step": 1797
    },
    {
      "epoch": 0.3066166439290587,
      "grad_norm": 0.4565592110157013,
      "learning_rate": 1.6252771394726623e-06,
      "loss": 0.5346,
      "step": 1798
    },
    {
      "epoch": 0.30678717598908595,
      "grad_norm": 0.46513283252716064,
      "learning_rate": 1.6248460125397876e-06,
      "loss": 0.5364,
      "step": 1799
    },
    {
      "epoch": 0.3069577080491132,
      "grad_norm": 0.3886967897415161,
      "learning_rate": 1.6244146949934542e-06,
      "loss": 0.529,
      "step": 1800
    },
    {
      "epoch": 0.3071282401091405,
      "grad_norm": 0.4297249913215637,
      "learning_rate": 1.6239831869652385e-06,
      "loss": 0.5566,
      "step": 1801
    },
    {
      "epoch": 0.3072987721691678,
      "grad_norm": 0.4115973114967346,
      "learning_rate": 1.6235514885867749e-06,
      "loss": 0.533,
      "step": 1802
    },
    {
      "epoch": 0.3074693042291951,
      "grad_norm": 0.5175999402999878,
      "learning_rate": 1.623119599989756e-06,
      "loss": 0.5493,
      "step": 1803
    },
    {
      "epoch": 0.30763983628922237,
      "grad_norm": 0.6544990539550781,
      "learning_rate": 1.6226875213059317e-06,
      "loss": 0.5272,
      "step": 1804
    },
    {
      "epoch": 0.3078103683492497,
      "grad_norm": 0.4819190502166748,
      "learning_rate": 1.622255252667111e-06,
      "loss": 0.522,
      "step": 1805
    },
    {
      "epoch": 0.30798090040927695,
      "grad_norm": 0.4651631712913513,
      "learning_rate": 1.6218227942051607e-06,
      "loss": 0.5298,
      "step": 1806
    },
    {
      "epoch": 0.3081514324693042,
      "grad_norm": 1.1320414543151855,
      "learning_rate": 1.6213901460520047e-06,
      "loss": 0.5272,
      "step": 1807
    },
    {
      "epoch": 0.3083219645293315,
      "grad_norm": 0.351107120513916,
      "learning_rate": 1.620957308339625e-06,
      "loss": 0.5476,
      "step": 1808
    },
    {
      "epoch": 0.3084924965893588,
      "grad_norm": 0.4684494733810425,
      "learning_rate": 1.620524281200062e-06,
      "loss": 0.529,
      "step": 1809
    },
    {
      "epoch": 0.3086630286493861,
      "grad_norm": 0.3853071331977844,
      "learning_rate": 1.6200910647654134e-06,
      "loss": 0.5453,
      "step": 1810
    },
    {
      "epoch": 0.30883356070941337,
      "grad_norm": 0.6147592067718506,
      "learning_rate": 1.6196576591678349e-06,
      "loss": 0.5268,
      "step": 1811
    },
    {
      "epoch": 0.30900409276944063,
      "grad_norm": 0.4014926254749298,
      "learning_rate": 1.6192240645395398e-06,
      "loss": 0.5349,
      "step": 1812
    },
    {
      "epoch": 0.30917462482946795,
      "grad_norm": 0.40421199798583984,
      "learning_rate": 1.6187902810127983e-06,
      "loss": 0.5522,
      "step": 1813
    },
    {
      "epoch": 0.3093451568894952,
      "grad_norm": 0.6183474063873291,
      "learning_rate": 1.6183563087199405e-06,
      "loss": 0.5163,
      "step": 1814
    },
    {
      "epoch": 0.3095156889495225,
      "grad_norm": 0.4411429464817047,
      "learning_rate": 1.617922147793351e-06,
      "loss": 0.5455,
      "step": 1815
    },
    {
      "epoch": 0.3096862210095498,
      "grad_norm": 0.4115869402885437,
      "learning_rate": 1.6174877983654744e-06,
      "loss": 0.5285,
      "step": 1816
    },
    {
      "epoch": 0.3098567530695771,
      "grad_norm": 0.39461439847946167,
      "learning_rate": 1.6170532605688114e-06,
      "loss": 0.534,
      "step": 1817
    },
    {
      "epoch": 0.31002728512960437,
      "grad_norm": 0.42679721117019653,
      "learning_rate": 1.616618534535921e-06,
      "loss": 0.5293,
      "step": 1818
    },
    {
      "epoch": 0.31019781718963163,
      "grad_norm": 0.5123752951622009,
      "learning_rate": 1.6161836203994194e-06,
      "loss": 0.5351,
      "step": 1819
    },
    {
      "epoch": 0.31036834924965895,
      "grad_norm": 0.6214609146118164,
      "learning_rate": 1.6157485182919797e-06,
      "loss": 0.5224,
      "step": 1820
    },
    {
      "epoch": 0.3105388813096862,
      "grad_norm": 0.5766216516494751,
      "learning_rate": 1.6153132283463329e-06,
      "loss": 0.5086,
      "step": 1821
    },
    {
      "epoch": 0.3107094133697135,
      "grad_norm": 0.4433499574661255,
      "learning_rate": 1.6148777506952673e-06,
      "loss": 0.5397,
      "step": 1822
    },
    {
      "epoch": 0.3108799454297408,
      "grad_norm": 0.3571315109729767,
      "learning_rate": 1.6144420854716278e-06,
      "loss": 0.5385,
      "step": 1823
    },
    {
      "epoch": 0.31105047748976805,
      "grad_norm": 0.4480735659599304,
      "learning_rate": 1.6140062328083173e-06,
      "loss": 0.5445,
      "step": 1824
    },
    {
      "epoch": 0.31122100954979537,
      "grad_norm": 0.4067867696285248,
      "learning_rate": 1.6135701928382953e-06,
      "loss": 0.5189,
      "step": 1825
    },
    {
      "epoch": 0.31139154160982263,
      "grad_norm": 0.38373708724975586,
      "learning_rate": 1.6131339656945792e-06,
      "loss": 0.5512,
      "step": 1826
    },
    {
      "epoch": 0.31156207366984995,
      "grad_norm": 0.4451771676540375,
      "learning_rate": 1.6126975515102424e-06,
      "loss": 0.5351,
      "step": 1827
    },
    {
      "epoch": 0.3117326057298772,
      "grad_norm": 0.3590299189090729,
      "learning_rate": 1.6122609504184167e-06,
      "loss": 0.5366,
      "step": 1828
    },
    {
      "epoch": 0.3119031377899045,
      "grad_norm": 0.4824784994125366,
      "learning_rate": 1.6118241625522898e-06,
      "loss": 0.5485,
      "step": 1829
    },
    {
      "epoch": 0.3120736698499318,
      "grad_norm": 0.4515202045440674,
      "learning_rate": 1.6113871880451065e-06,
      "loss": 0.5277,
      "step": 1830
    },
    {
      "epoch": 0.31224420190995905,
      "grad_norm": 0.4325205981731415,
      "learning_rate": 1.6109500270301694e-06,
      "loss": 0.5326,
      "step": 1831
    },
    {
      "epoch": 0.31241473396998637,
      "grad_norm": 0.4184912443161011,
      "learning_rate": 1.6105126796408368e-06,
      "loss": 0.5357,
      "step": 1832
    },
    {
      "epoch": 0.31258526603001363,
      "grad_norm": 0.46820399165153503,
      "learning_rate": 1.6100751460105245e-06,
      "loss": 0.5348,
      "step": 1833
    },
    {
      "epoch": 0.31275579809004095,
      "grad_norm": 0.40271228551864624,
      "learning_rate": 1.6096374262727057e-06,
      "loss": 0.5425,
      "step": 1834
    },
    {
      "epoch": 0.3129263301500682,
      "grad_norm": 0.38214290142059326,
      "learning_rate": 1.6091995205609092e-06,
      "loss": 0.5198,
      "step": 1835
    },
    {
      "epoch": 0.3130968622100955,
      "grad_norm": 0.5242385864257812,
      "learning_rate": 1.608761429008721e-06,
      "loss": 0.5422,
      "step": 1836
    },
    {
      "epoch": 0.3132673942701228,
      "grad_norm": 0.4800110459327698,
      "learning_rate": 1.608323151749784e-06,
      "loss": 0.5247,
      "step": 1837
    },
    {
      "epoch": 0.31343792633015005,
      "grad_norm": 0.3811364471912384,
      "learning_rate": 1.607884688917798e-06,
      "loss": 0.5411,
      "step": 1838
    },
    {
      "epoch": 0.31360845839017737,
      "grad_norm": 0.40190911293029785,
      "learning_rate": 1.6074460406465182e-06,
      "loss": 0.5282,
      "step": 1839
    },
    {
      "epoch": 0.31377899045020463,
      "grad_norm": 0.5054764747619629,
      "learning_rate": 1.6070072070697578e-06,
      "loss": 0.5547,
      "step": 1840
    },
    {
      "epoch": 0.31394952251023195,
      "grad_norm": 0.3729870915412903,
      "learning_rate": 1.6065681883213857e-06,
      "loss": 0.5306,
      "step": 1841
    },
    {
      "epoch": 0.3141200545702592,
      "grad_norm": 0.3761366605758667,
      "learning_rate": 1.6061289845353279e-06,
      "loss": 0.5319,
      "step": 1842
    },
    {
      "epoch": 0.3142905866302865,
      "grad_norm": 0.4457892179489136,
      "learning_rate": 1.6056895958455657e-06,
      "loss": 0.5406,
      "step": 1843
    },
    {
      "epoch": 0.3144611186903138,
      "grad_norm": 0.3641684949398041,
      "learning_rate": 1.6052500223861383e-06,
      "loss": 0.5264,
      "step": 1844
    },
    {
      "epoch": 0.31463165075034105,
      "grad_norm": 0.4447363615036011,
      "learning_rate": 1.6048102642911399e-06,
      "loss": 0.5265,
      "step": 1845
    },
    {
      "epoch": 0.31480218281036837,
      "grad_norm": 0.4733704626560211,
      "learning_rate": 1.6043703216947225e-06,
      "loss": 0.5254,
      "step": 1846
    },
    {
      "epoch": 0.31497271487039563,
      "grad_norm": 0.3722693920135498,
      "learning_rate": 1.6039301947310924e-06,
      "loss": 0.5337,
      "step": 1847
    },
    {
      "epoch": 0.3151432469304229,
      "grad_norm": 0.4494919180870056,
      "learning_rate": 1.6034898835345144e-06,
      "loss": 0.533,
      "step": 1848
    },
    {
      "epoch": 0.3153137789904502,
      "grad_norm": 0.42897871136665344,
      "learning_rate": 1.6030493882393075e-06,
      "loss": 0.5329,
      "step": 1849
    },
    {
      "epoch": 0.3154843110504775,
      "grad_norm": 0.45090794563293457,
      "learning_rate": 1.6026087089798483e-06,
      "loss": 0.5352,
      "step": 1850
    },
    {
      "epoch": 0.3156548431105048,
      "grad_norm": 0.37061816453933716,
      "learning_rate": 1.6021678458905686e-06,
      "loss": 0.5278,
      "step": 1851
    },
    {
      "epoch": 0.31582537517053205,
      "grad_norm": 0.4528961181640625,
      "learning_rate": 1.601726799105957e-06,
      "loss": 0.5247,
      "step": 1852
    },
    {
      "epoch": 0.31599590723055937,
      "grad_norm": 0.4259662628173828,
      "learning_rate": 1.6012855687605574e-06,
      "loss": 0.5376,
      "step": 1853
    },
    {
      "epoch": 0.31616643929058663,
      "grad_norm": 0.43850255012512207,
      "learning_rate": 1.6008441549889703e-06,
      "loss": 0.5398,
      "step": 1854
    },
    {
      "epoch": 0.3163369713506139,
      "grad_norm": 0.40549373626708984,
      "learning_rate": 1.600402557925852e-06,
      "loss": 0.5229,
      "step": 1855
    },
    {
      "epoch": 0.3165075034106412,
      "grad_norm": 0.43591824173927307,
      "learning_rate": 1.5999607777059141e-06,
      "loss": 0.5385,
      "step": 1856
    },
    {
      "epoch": 0.3166780354706685,
      "grad_norm": 0.4598686695098877,
      "learning_rate": 1.5995188144639252e-06,
      "loss": 0.5373,
      "step": 1857
    },
    {
      "epoch": 0.3168485675306958,
      "grad_norm": 0.34796208143234253,
      "learning_rate": 1.599076668334709e-06,
      "loss": 0.5257,
      "step": 1858
    },
    {
      "epoch": 0.31701909959072305,
      "grad_norm": 0.46574661135673523,
      "learning_rate": 1.598634339453145e-06,
      "loss": 0.5219,
      "step": 1859
    },
    {
      "epoch": 0.3171896316507503,
      "grad_norm": 0.4256969094276428,
      "learning_rate": 1.598191827954169e-06,
      "loss": 0.5286,
      "step": 1860
    },
    {
      "epoch": 0.31736016371077763,
      "grad_norm": 0.37936362624168396,
      "learning_rate": 1.5977491339727716e-06,
      "loss": 0.5332,
      "step": 1861
    },
    {
      "epoch": 0.3175306957708049,
      "grad_norm": 0.4368015229701996,
      "learning_rate": 1.5973062576439999e-06,
      "loss": 0.521,
      "step": 1862
    },
    {
      "epoch": 0.3177012278308322,
      "grad_norm": 0.364070862531662,
      "learning_rate": 1.5968631991029557e-06,
      "loss": 0.5412,
      "step": 1863
    },
    {
      "epoch": 0.3178717598908595,
      "grad_norm": 0.4169939160346985,
      "learning_rate": 1.5964199584847977e-06,
      "loss": 0.5329,
      "step": 1864
    },
    {
      "epoch": 0.3180422919508868,
      "grad_norm": 0.3581223487854004,
      "learning_rate": 1.595976535924739e-06,
      "loss": 0.527,
      "step": 1865
    },
    {
      "epoch": 0.31821282401091405,
      "grad_norm": 0.47882622480392456,
      "learning_rate": 1.595532931558049e-06,
      "loss": 0.5297,
      "step": 1866
    },
    {
      "epoch": 0.3183833560709413,
      "grad_norm": 0.43346402049064636,
      "learning_rate": 1.5950891455200516e-06,
      "loss": 0.5233,
      "step": 1867
    },
    {
      "epoch": 0.31855388813096863,
      "grad_norm": 0.458211749792099,
      "learning_rate": 1.5946451779461271e-06,
      "loss": 0.5312,
      "step": 1868
    },
    {
      "epoch": 0.3187244201909959,
      "grad_norm": 0.5317389965057373,
      "learning_rate": 1.5942010289717107e-06,
      "loss": 0.5225,
      "step": 1869
    },
    {
      "epoch": 0.3188949522510232,
      "grad_norm": 0.4372017979621887,
      "learning_rate": 1.5937566987322932e-06,
      "loss": 0.5266,
      "step": 1870
    },
    {
      "epoch": 0.3190654843110505,
      "grad_norm": 0.4571208655834198,
      "learning_rate": 1.5933121873634202e-06,
      "loss": 0.5217,
      "step": 1871
    },
    {
      "epoch": 0.31923601637107774,
      "grad_norm": 0.4793291389942169,
      "learning_rate": 1.592867495000693e-06,
      "loss": 0.5137,
      "step": 1872
    },
    {
      "epoch": 0.31940654843110505,
      "grad_norm": 0.56477290391922,
      "learning_rate": 1.592422621779768e-06,
      "loss": 0.5182,
      "step": 1873
    },
    {
      "epoch": 0.3195770804911323,
      "grad_norm": 0.497898131608963,
      "learning_rate": 1.5919775678363565e-06,
      "loss": 0.5215,
      "step": 1874
    },
    {
      "epoch": 0.31974761255115963,
      "grad_norm": 0.43701592087745667,
      "learning_rate": 1.5915323333062257e-06,
      "loss": 0.526,
      "step": 1875
    },
    {
      "epoch": 0.3199181446111869,
      "grad_norm": 0.6302632689476013,
      "learning_rate": 1.591086918325197e-06,
      "loss": 0.5271,
      "step": 1876
    },
    {
      "epoch": 0.3200886766712142,
      "grad_norm": 0.5440297722816467,
      "learning_rate": 1.5906413230291474e-06,
      "loss": 0.5266,
      "step": 1877
    },
    {
      "epoch": 0.3202592087312415,
      "grad_norm": 0.6256676912307739,
      "learning_rate": 1.5901955475540085e-06,
      "loss": 0.5331,
      "step": 1878
    },
    {
      "epoch": 0.32042974079126874,
      "grad_norm": 0.7379283308982849,
      "learning_rate": 1.5897495920357676e-06,
      "loss": 0.5251,
      "step": 1879
    },
    {
      "epoch": 0.32060027285129605,
      "grad_norm": 0.5450626015663147,
      "learning_rate": 1.5893034566104658e-06,
      "loss": 0.5261,
      "step": 1880
    },
    {
      "epoch": 0.3207708049113233,
      "grad_norm": 0.4530295431613922,
      "learning_rate": 1.5888571414141998e-06,
      "loss": 0.5315,
      "step": 1881
    },
    {
      "epoch": 0.32094133697135063,
      "grad_norm": 0.47940734028816223,
      "learning_rate": 1.5884106465831212e-06,
      "loss": 0.5267,
      "step": 1882
    },
    {
      "epoch": 0.3211118690313779,
      "grad_norm": 0.49188658595085144,
      "learning_rate": 1.5879639722534365e-06,
      "loss": 0.5228,
      "step": 1883
    },
    {
      "epoch": 0.32128240109140516,
      "grad_norm": 0.46712371706962585,
      "learning_rate": 1.5875171185614063e-06,
      "loss": 0.5258,
      "step": 1884
    },
    {
      "epoch": 0.3214529331514325,
      "grad_norm": 0.5874091982841492,
      "learning_rate": 1.5870700856433465e-06,
      "loss": 0.526,
      "step": 1885
    },
    {
      "epoch": 0.32162346521145974,
      "grad_norm": 0.41451871395111084,
      "learning_rate": 1.5866228736356271e-06,
      "loss": 0.5298,
      "step": 1886
    },
    {
      "epoch": 0.32179399727148705,
      "grad_norm": 0.5600169897079468,
      "learning_rate": 1.5861754826746738e-06,
      "loss": 0.5244,
      "step": 1887
    },
    {
      "epoch": 0.3219645293315143,
      "grad_norm": 0.6498214602470398,
      "learning_rate": 1.5857279128969657e-06,
      "loss": 0.5294,
      "step": 1888
    },
    {
      "epoch": 0.32213506139154163,
      "grad_norm": 0.6391829252243042,
      "learning_rate": 1.5852801644390368e-06,
      "loss": 0.5334,
      "step": 1889
    },
    {
      "epoch": 0.3223055934515689,
      "grad_norm": 0.46014702320098877,
      "learning_rate": 1.5848322374374766e-06,
      "loss": 0.5404,
      "step": 1890
    },
    {
      "epoch": 0.32247612551159616,
      "grad_norm": 0.4720998704433441,
      "learning_rate": 1.5843841320289273e-06,
      "loss": 0.5324,
      "step": 1891
    },
    {
      "epoch": 0.3226466575716235,
      "grad_norm": 0.5924428701400757,
      "learning_rate": 1.5839358483500872e-06,
      "loss": 0.5467,
      "step": 1892
    },
    {
      "epoch": 0.32281718963165074,
      "grad_norm": 0.4617968797683716,
      "learning_rate": 1.5834873865377077e-06,
      "loss": 0.5461,
      "step": 1893
    },
    {
      "epoch": 0.32298772169167805,
      "grad_norm": 0.4647984504699707,
      "learning_rate": 1.5830387467285952e-06,
      "loss": 0.5327,
      "step": 1894
    },
    {
      "epoch": 0.3231582537517053,
      "grad_norm": 0.47549039125442505,
      "learning_rate": 1.5825899290596103e-06,
      "loss": 0.525,
      "step": 1895
    },
    {
      "epoch": 0.32332878581173263,
      "grad_norm": 0.41526469588279724,
      "learning_rate": 1.5821409336676678e-06,
      "loss": 0.5148,
      "step": 1896
    },
    {
      "epoch": 0.3234993178717599,
      "grad_norm": 0.4488104581832886,
      "learning_rate": 1.581691760689737e-06,
      "loss": 0.5427,
      "step": 1897
    },
    {
      "epoch": 0.32366984993178716,
      "grad_norm": 0.3801409900188446,
      "learning_rate": 1.581242410262841e-06,
      "loss": 0.5222,
      "step": 1898
    },
    {
      "epoch": 0.3238403819918145,
      "grad_norm": 0.4768373370170593,
      "learning_rate": 1.5807928825240567e-06,
      "loss": 0.5282,
      "step": 1899
    },
    {
      "epoch": 0.32401091405184174,
      "grad_norm": 0.45949891209602356,
      "learning_rate": 1.5803431776105163e-06,
      "loss": 0.5228,
      "step": 1900
    },
    {
      "epoch": 0.32418144611186905,
      "grad_norm": 0.41923388838768005,
      "learning_rate": 1.579893295659405e-06,
      "loss": 0.529,
      "step": 1901
    },
    {
      "epoch": 0.3243519781718963,
      "grad_norm": 0.40380582213401794,
      "learning_rate": 1.5794432368079619e-06,
      "loss": 0.5313,
      "step": 1902
    },
    {
      "epoch": 0.3245225102319236,
      "grad_norm": 0.5308855772018433,
      "learning_rate": 1.578993001193481e-06,
      "loss": 0.526,
      "step": 1903
    },
    {
      "epoch": 0.3246930422919509,
      "grad_norm": 0.5592904686927795,
      "learning_rate": 1.5785425889533097e-06,
      "loss": 0.5313,
      "step": 1904
    },
    {
      "epoch": 0.32486357435197816,
      "grad_norm": 0.4044639766216278,
      "learning_rate": 1.5780920002248487e-06,
      "loss": 0.5402,
      "step": 1905
    },
    {
      "epoch": 0.3250341064120055,
      "grad_norm": 0.40651005506515503,
      "learning_rate": 1.5776412351455536e-06,
      "loss": 0.5383,
      "step": 1906
    },
    {
      "epoch": 0.32520463847203274,
      "grad_norm": 0.5054802894592285,
      "learning_rate": 1.5771902938529333e-06,
      "loss": 0.528,
      "step": 1907
    },
    {
      "epoch": 0.32537517053206005,
      "grad_norm": 0.5022133588790894,
      "learning_rate": 1.5767391764845502e-06,
      "loss": 0.5277,
      "step": 1908
    },
    {
      "epoch": 0.3255457025920873,
      "grad_norm": 0.36169329285621643,
      "learning_rate": 1.5762878831780207e-06,
      "loss": 0.5249,
      "step": 1909
    },
    {
      "epoch": 0.3257162346521146,
      "grad_norm": 0.41136419773101807,
      "learning_rate": 1.575836414071015e-06,
      "loss": 0.5265,
      "step": 1910
    },
    {
      "epoch": 0.3258867667121419,
      "grad_norm": 0.40905439853668213,
      "learning_rate": 1.575384769301257e-06,
      "loss": 0.516,
      "step": 1911
    },
    {
      "epoch": 0.32605729877216916,
      "grad_norm": 0.35940247774124146,
      "learning_rate": 1.5749329490065234e-06,
      "loss": 0.5278,
      "step": 1912
    },
    {
      "epoch": 0.3262278308321965,
      "grad_norm": 0.3744727373123169,
      "learning_rate": 1.5744809533246452e-06,
      "loss": 0.5467,
      "step": 1913
    },
    {
      "epoch": 0.32639836289222374,
      "grad_norm": 0.3644757866859436,
      "learning_rate": 1.5740287823935068e-06,
      "loss": 0.5221,
      "step": 1914
    },
    {
      "epoch": 0.326568894952251,
      "grad_norm": 0.3962186872959137,
      "learning_rate": 1.5735764363510464e-06,
      "loss": 0.5266,
      "step": 1915
    },
    {
      "epoch": 0.3267394270122783,
      "grad_norm": 0.3878701329231262,
      "learning_rate": 1.5731239153352546e-06,
      "loss": 0.5331,
      "step": 1916
    },
    {
      "epoch": 0.3269099590723056,
      "grad_norm": 0.39507779479026794,
      "learning_rate": 1.572671219484176e-06,
      "loss": 0.5265,
      "step": 1917
    },
    {
      "epoch": 0.3270804911323329,
      "grad_norm": 0.3882591128349304,
      "learning_rate": 1.5722183489359083e-06,
      "loss": 0.5404,
      "step": 1918
    },
    {
      "epoch": 0.32725102319236016,
      "grad_norm": 0.3593181371688843,
      "learning_rate": 1.5717653038286036e-06,
      "loss": 0.5336,
      "step": 1919
    },
    {
      "epoch": 0.3274215552523875,
      "grad_norm": 0.3732438087463379,
      "learning_rate": 1.5713120843004653e-06,
      "loss": 0.5382,
      "step": 1920
    },
    {
      "epoch": 0.32759208731241474,
      "grad_norm": 0.39789363741874695,
      "learning_rate": 1.5708586904897516e-06,
      "loss": 0.5301,
      "step": 1921
    },
    {
      "epoch": 0.327762619372442,
      "grad_norm": 0.3832828104496002,
      "learning_rate": 1.5704051225347734e-06,
      "loss": 0.5175,
      "step": 1922
    },
    {
      "epoch": 0.3279331514324693,
      "grad_norm": 0.37655749917030334,
      "learning_rate": 1.5699513805738943e-06,
      "loss": 0.5201,
      "step": 1923
    },
    {
      "epoch": 0.3281036834924966,
      "grad_norm": 0.4358007311820984,
      "learning_rate": 1.5694974647455318e-06,
      "loss": 0.5203,
      "step": 1924
    },
    {
      "epoch": 0.3282742155525239,
      "grad_norm": 0.4240171015262604,
      "learning_rate": 1.5690433751881553e-06,
      "loss": 0.5205,
      "step": 1925
    },
    {
      "epoch": 0.32844474761255116,
      "grad_norm": 0.5076821446418762,
      "learning_rate": 1.5685891120402883e-06,
      "loss": 0.5246,
      "step": 1926
    },
    {
      "epoch": 0.3286152796725784,
      "grad_norm": 0.41013339161872864,
      "learning_rate": 1.5681346754405066e-06,
      "loss": 0.5288,
      "step": 1927
    },
    {
      "epoch": 0.32878581173260574,
      "grad_norm": 0.42070189118385315,
      "learning_rate": 1.5676800655274394e-06,
      "loss": 0.5262,
      "step": 1928
    },
    {
      "epoch": 0.328956343792633,
      "grad_norm": 0.3666054308414459,
      "learning_rate": 1.5672252824397683e-06,
      "loss": 0.5382,
      "step": 1929
    },
    {
      "epoch": 0.3291268758526603,
      "grad_norm": 0.49871453642845154,
      "learning_rate": 1.5667703263162278e-06,
      "loss": 0.5142,
      "step": 1930
    },
    {
      "epoch": 0.3292974079126876,
      "grad_norm": 0.4860789179801941,
      "learning_rate": 1.5663151972956057e-06,
      "loss": 0.518,
      "step": 1931
    },
    {
      "epoch": 0.3294679399727149,
      "grad_norm": 0.4729858338832855,
      "learning_rate": 1.565859895516742e-06,
      "loss": 0.5248,
      "step": 1932
    },
    {
      "epoch": 0.32963847203274216,
      "grad_norm": 0.4114147424697876,
      "learning_rate": 1.5654044211185293e-06,
      "loss": 0.5318,
      "step": 1933
    },
    {
      "epoch": 0.3298090040927694,
      "grad_norm": 0.3970237672328949,
      "learning_rate": 1.5649487742399133e-06,
      "loss": 0.5412,
      "step": 1934
    },
    {
      "epoch": 0.32997953615279674,
      "grad_norm": 0.5709453821182251,
      "learning_rate": 1.5644929550198921e-06,
      "loss": 0.5333,
      "step": 1935
    },
    {
      "epoch": 0.330150068212824,
      "grad_norm": 0.4466193616390228,
      "learning_rate": 1.5640369635975164e-06,
      "loss": 0.5275,
      "step": 1936
    },
    {
      "epoch": 0.3303206002728513,
      "grad_norm": 0.46157199144363403,
      "learning_rate": 1.5635808001118898e-06,
      "loss": 0.5309,
      "step": 1937
    },
    {
      "epoch": 0.3304911323328786,
      "grad_norm": 0.4336494505405426,
      "learning_rate": 1.5631244647021674e-06,
      "loss": 0.5237,
      "step": 1938
    },
    {
      "epoch": 0.33066166439290584,
      "grad_norm": 0.4006889760494232,
      "learning_rate": 1.5626679575075578e-06,
      "loss": 0.5389,
      "step": 1939
    },
    {
      "epoch": 0.33083219645293316,
      "grad_norm": 0.5048114657402039,
      "learning_rate": 1.5622112786673214e-06,
      "loss": 0.5337,
      "step": 1940
    },
    {
      "epoch": 0.3310027285129604,
      "grad_norm": 0.6756312251091003,
      "learning_rate": 1.5617544283207713e-06,
      "loss": 0.5331,
      "step": 1941
    },
    {
      "epoch": 0.33117326057298774,
      "grad_norm": 0.6390357613563538,
      "learning_rate": 1.5612974066072723e-06,
      "loss": 0.5336,
      "step": 1942
    },
    {
      "epoch": 0.331343792633015,
      "grad_norm": 0.43792104721069336,
      "learning_rate": 1.5608402136662425e-06,
      "loss": 0.5467,
      "step": 1943
    },
    {
      "epoch": 0.3315143246930423,
      "grad_norm": 0.4027770757675171,
      "learning_rate": 1.5603828496371514e-06,
      "loss": 0.519,
      "step": 1944
    },
    {
      "epoch": 0.3316848567530696,
      "grad_norm": 0.6208730936050415,
      "learning_rate": 1.5599253146595213e-06,
      "loss": 0.5205,
      "step": 1945
    },
    {
      "epoch": 0.33185538881309684,
      "grad_norm": 0.5827608108520508,
      "learning_rate": 1.5594676088729253e-06,
      "loss": 0.5209,
      "step": 1946
    },
    {
      "epoch": 0.33202592087312416,
      "grad_norm": 0.3326941728591919,
      "learning_rate": 1.559009732416991e-06,
      "loss": 0.5336,
      "step": 1947
    },
    {
      "epoch": 0.3321964529331514,
      "grad_norm": 0.5395388603210449,
      "learning_rate": 1.5585516854313957e-06,
      "loss": 0.5229,
      "step": 1948
    },
    {
      "epoch": 0.33236698499317874,
      "grad_norm": 0.5696554780006409,
      "learning_rate": 1.5580934680558698e-06,
      "loss": 0.5303,
      "step": 1949
    },
    {
      "epoch": 0.332537517053206,
      "grad_norm": 0.5093621015548706,
      "learning_rate": 1.557635080430196e-06,
      "loss": 0.5302,
      "step": 1950
    },
    {
      "epoch": 0.33270804911323326,
      "grad_norm": 0.4694986343383789,
      "learning_rate": 1.5571765226942084e-06,
      "loss": 0.5333,
      "step": 1951
    },
    {
      "epoch": 0.3328785811732606,
      "grad_norm": 0.5953072309494019,
      "learning_rate": 1.5567177949877927e-06,
      "loss": 0.5287,
      "step": 1952
    },
    {
      "epoch": 0.33304911323328784,
      "grad_norm": 0.42887023091316223,
      "learning_rate": 1.5562588974508874e-06,
      "loss": 0.5212,
      "step": 1953
    },
    {
      "epoch": 0.33321964529331516,
      "grad_norm": 0.45148569345474243,
      "learning_rate": 1.5557998302234816e-06,
      "loss": 0.5289,
      "step": 1954
    },
    {
      "epoch": 0.3333901773533424,
      "grad_norm": 0.5752291679382324,
      "learning_rate": 1.555340593445618e-06,
      "loss": 0.539,
      "step": 1955
    },
    {
      "epoch": 0.33356070941336974,
      "grad_norm": 0.37518638372421265,
      "learning_rate": 1.5548811872573885e-06,
      "loss": 0.5322,
      "step": 1956
    },
    {
      "epoch": 0.333731241473397,
      "grad_norm": 0.47894930839538574,
      "learning_rate": 1.5544216117989389e-06,
      "loss": 0.5256,
      "step": 1957
    },
    {
      "epoch": 0.33390177353342426,
      "grad_norm": 0.5407999753952026,
      "learning_rate": 1.5539618672104652e-06,
      "loss": 0.516,
      "step": 1958
    },
    {
      "epoch": 0.3340723055934516,
      "grad_norm": 0.3796008825302124,
      "learning_rate": 1.5535019536322159e-06,
      "loss": 0.5129,
      "step": 1959
    },
    {
      "epoch": 0.33424283765347884,
      "grad_norm": 0.4745221436023712,
      "learning_rate": 1.5530418712044909e-06,
      "loss": 0.5347,
      "step": 1960
    },
    {
      "epoch": 0.33441336971350616,
      "grad_norm": 0.5195663571357727,
      "learning_rate": 1.552581620067641e-06,
      "loss": 0.5291,
      "step": 1961
    },
    {
      "epoch": 0.3345839017735334,
      "grad_norm": 0.4340425133705139,
      "learning_rate": 1.5521212003620693e-06,
      "loss": 0.5173,
      "step": 1962
    },
    {
      "epoch": 0.3347544338335607,
      "grad_norm": 0.43442365527153015,
      "learning_rate": 1.5516606122282298e-06,
      "loss": 0.5323,
      "step": 1963
    },
    {
      "epoch": 0.334924965893588,
      "grad_norm": 0.4027803838253021,
      "learning_rate": 1.5511998558066276e-06,
      "loss": 0.5329,
      "step": 1964
    },
    {
      "epoch": 0.33509549795361526,
      "grad_norm": 0.384951114654541,
      "learning_rate": 1.5507389312378199e-06,
      "loss": 0.5183,
      "step": 1965
    },
    {
      "epoch": 0.3352660300136426,
      "grad_norm": 0.43811824917793274,
      "learning_rate": 1.5502778386624148e-06,
      "loss": 0.5307,
      "step": 1966
    },
    {
      "epoch": 0.33543656207366984,
      "grad_norm": 0.3942059874534607,
      "learning_rate": 1.5498165782210715e-06,
      "loss": 0.5291,
      "step": 1967
    },
    {
      "epoch": 0.33560709413369716,
      "grad_norm": 0.5205844044685364,
      "learning_rate": 1.549355150054501e-06,
      "loss": 0.5273,
      "step": 1968
    },
    {
      "epoch": 0.3357776261937244,
      "grad_norm": 0.415243536233902,
      "learning_rate": 1.5488935543034646e-06,
      "loss": 0.5248,
      "step": 1969
    },
    {
      "epoch": 0.3359481582537517,
      "grad_norm": 0.47817909717559814,
      "learning_rate": 1.5484317911087752e-06,
      "loss": 0.5311,
      "step": 1970
    },
    {
      "epoch": 0.336118690313779,
      "grad_norm": 0.48120763897895813,
      "learning_rate": 1.5479698606112973e-06,
      "loss": 0.5255,
      "step": 1971
    },
    {
      "epoch": 0.33628922237380626,
      "grad_norm": 0.42829394340515137,
      "learning_rate": 1.5475077629519452e-06,
      "loss": 0.515,
      "step": 1972
    },
    {
      "epoch": 0.3364597544338336,
      "grad_norm": 0.48222029209136963,
      "learning_rate": 1.5470454982716852e-06,
      "loss": 0.5163,
      "step": 1973
    },
    {
      "epoch": 0.33663028649386084,
      "grad_norm": 0.5472458004951477,
      "learning_rate": 1.546583066711534e-06,
      "loss": 0.535,
      "step": 1974
    },
    {
      "epoch": 0.3368008185538881,
      "grad_norm": 0.6080836057662964,
      "learning_rate": 1.54612046841256e-06,
      "loss": 0.5256,
      "step": 1975
    },
    {
      "epoch": 0.3369713506139154,
      "grad_norm": 0.5023837685585022,
      "learning_rate": 1.5456577035158815e-06,
      "loss": 0.5301,
      "step": 1976
    },
    {
      "epoch": 0.3371418826739427,
      "grad_norm": 0.41623133420944214,
      "learning_rate": 1.5451947721626677e-06,
      "loss": 0.5393,
      "step": 1977
    },
    {
      "epoch": 0.33731241473397,
      "grad_norm": 0.6211003065109253,
      "learning_rate": 1.5447316744941399e-06,
      "loss": 0.5327,
      "step": 1978
    },
    {
      "epoch": 0.33748294679399726,
      "grad_norm": 0.5653514862060547,
      "learning_rate": 1.5442684106515684e-06,
      "loss": 0.5275,
      "step": 1979
    },
    {
      "epoch": 0.3376534788540246,
      "grad_norm": 0.4624597430229187,
      "learning_rate": 1.5438049807762746e-06,
      "loss": 0.5332,
      "step": 1980
    },
    {
      "epoch": 0.33782401091405184,
      "grad_norm": 0.39047080278396606,
      "learning_rate": 1.5433413850096316e-06,
      "loss": 0.5188,
      "step": 1981
    },
    {
      "epoch": 0.3379945429740791,
      "grad_norm": 0.4766196608543396,
      "learning_rate": 1.5428776234930622e-06,
      "loss": 0.5251,
      "step": 1982
    },
    {
      "epoch": 0.3381650750341064,
      "grad_norm": 0.45972636342048645,
      "learning_rate": 1.54241369636804e-06,
      "loss": 0.5222,
      "step": 1983
    },
    {
      "epoch": 0.3383356070941337,
      "grad_norm": 0.455423504114151,
      "learning_rate": 1.5419496037760887e-06,
      "loss": 0.5189,
      "step": 1984
    },
    {
      "epoch": 0.338506139154161,
      "grad_norm": 0.4104258120059967,
      "learning_rate": 1.5414853458587835e-06,
      "loss": 0.5182,
      "step": 1985
    },
    {
      "epoch": 0.33867667121418826,
      "grad_norm": 0.4785384237766266,
      "learning_rate": 1.5410209227577487e-06,
      "loss": 0.5242,
      "step": 1986
    },
    {
      "epoch": 0.3388472032742155,
      "grad_norm": 0.5701748132705688,
      "learning_rate": 1.5405563346146604e-06,
      "loss": 0.5248,
      "step": 1987
    },
    {
      "epoch": 0.33901773533424284,
      "grad_norm": 0.5212503671646118,
      "learning_rate": 1.5400915815712434e-06,
      "loss": 0.5267,
      "step": 1988
    },
    {
      "epoch": 0.3391882673942701,
      "grad_norm": 0.3825041353702545,
      "learning_rate": 1.5396266637692746e-06,
      "loss": 0.5258,
      "step": 1989
    },
    {
      "epoch": 0.3393587994542974,
      "grad_norm": 0.5028911232948303,
      "learning_rate": 1.5391615813505794e-06,
      "loss": 0.5435,
      "step": 1990
    },
    {
      "epoch": 0.3395293315143247,
      "grad_norm": 0.46265122294425964,
      "learning_rate": 1.5386963344570355e-06,
      "loss": 0.5167,
      "step": 1991
    },
    {
      "epoch": 0.339699863574352,
      "grad_norm": 0.511637806892395,
      "learning_rate": 1.5382309232305687e-06,
      "loss": 0.5324,
      "step": 1992
    },
    {
      "epoch": 0.33987039563437926,
      "grad_norm": 0.5030692219734192,
      "learning_rate": 1.5377653478131558e-06,
      "loss": 0.5254,
      "step": 1993
    },
    {
      "epoch": 0.3400409276944065,
      "grad_norm": 0.4326249659061432,
      "learning_rate": 1.5372996083468244e-06,
      "loss": 0.5222,
      "step": 1994
    },
    {
      "epoch": 0.34021145975443384,
      "grad_norm": 0.6164526343345642,
      "learning_rate": 1.5368337049736505e-06,
      "loss": 0.536,
      "step": 1995
    },
    {
      "epoch": 0.3403819918144611,
      "grad_norm": 0.6666930317878723,
      "learning_rate": 1.5363676378357615e-06,
      "loss": 0.5299,
      "step": 1996
    },
    {
      "epoch": 0.3405525238744884,
      "grad_norm": 0.40014538168907166,
      "learning_rate": 1.5359014070753346e-06,
      "loss": 0.5318,
      "step": 1997
    },
    {
      "epoch": 0.3407230559345157,
      "grad_norm": 0.5744419693946838,
      "learning_rate": 1.535435012834596e-06,
      "loss": 0.5231,
      "step": 1998
    },
    {
      "epoch": 0.340893587994543,
      "grad_norm": 0.5867252349853516,
      "learning_rate": 1.5349684552558227e-06,
      "loss": 0.5269,
      "step": 1999
    },
    {
      "epoch": 0.34106412005457026,
      "grad_norm": 0.4233283996582031,
      "learning_rate": 1.5345017344813415e-06,
      "loss": 0.5262,
      "step": 2000
    },
    {
      "epoch": 0.3412346521145975,
      "grad_norm": 0.5572194457054138,
      "learning_rate": 1.5340348506535285e-06,
      "loss": 0.5439,
      "step": 2001
    },
    {
      "epoch": 0.34140518417462484,
      "grad_norm": 0.46090197563171387,
      "learning_rate": 1.5335678039148094e-06,
      "loss": 0.5302,
      "step": 2002
    },
    {
      "epoch": 0.3415757162346521,
      "grad_norm": 0.3734171688556671,
      "learning_rate": 1.5331005944076602e-06,
      "loss": 0.5215,
      "step": 2003
    },
    {
      "epoch": 0.3417462482946794,
      "grad_norm": 0.6022562384605408,
      "learning_rate": 1.5326332222746061e-06,
      "loss": 0.5221,
      "step": 2004
    },
    {
      "epoch": 0.3419167803547067,
      "grad_norm": 0.4694681465625763,
      "learning_rate": 1.5321656876582226e-06,
      "loss": 0.5396,
      "step": 2005
    },
    {
      "epoch": 0.34208731241473395,
      "grad_norm": 0.41527536511421204,
      "learning_rate": 1.5316979907011336e-06,
      "loss": 0.5219,
      "step": 2006
    },
    {
      "epoch": 0.34225784447476126,
      "grad_norm": 0.6797448992729187,
      "learning_rate": 1.531230131546014e-06,
      "loss": 0.5231,
      "step": 2007
    },
    {
      "epoch": 0.3424283765347885,
      "grad_norm": 0.4598335027694702,
      "learning_rate": 1.5307621103355864e-06,
      "loss": 0.5207,
      "step": 2008
    },
    {
      "epoch": 0.34259890859481584,
      "grad_norm": 0.47826674580574036,
      "learning_rate": 1.5302939272126246e-06,
      "loss": 0.5482,
      "step": 2009
    },
    {
      "epoch": 0.3427694406548431,
      "grad_norm": 0.46293771266937256,
      "learning_rate": 1.5298255823199506e-06,
      "loss": 0.5179,
      "step": 2010
    },
    {
      "epoch": 0.3429399727148704,
      "grad_norm": 0.4489620327949524,
      "learning_rate": 1.5293570758004363e-06,
      "loss": 0.5285,
      "step": 2011
    },
    {
      "epoch": 0.3431105047748977,
      "grad_norm": 0.5135905742645264,
      "learning_rate": 1.5288884077970024e-06,
      "loss": 0.5119,
      "step": 2012
    },
    {
      "epoch": 0.34328103683492495,
      "grad_norm": 0.3585302531719208,
      "learning_rate": 1.5284195784526198e-06,
      "loss": 0.5128,
      "step": 2013
    },
    {
      "epoch": 0.34345156889495226,
      "grad_norm": 0.5710442662239075,
      "learning_rate": 1.5279505879103074e-06,
      "loss": 0.5284,
      "step": 2014
    },
    {
      "epoch": 0.3436221009549795,
      "grad_norm": 0.5375210046768188,
      "learning_rate": 1.5274814363131347e-06,
      "loss": 0.5112,
      "step": 2015
    },
    {
      "epoch": 0.34379263301500684,
      "grad_norm": 0.4037111699581146,
      "learning_rate": 1.5270121238042188e-06,
      "loss": 0.5386,
      "step": 2016
    },
    {
      "epoch": 0.3439631650750341,
      "grad_norm": 1.513051152229309,
      "learning_rate": 1.5265426505267266e-06,
      "loss": 0.52,
      "step": 2017
    },
    {
      "epoch": 0.34413369713506137,
      "grad_norm": 0.44357699155807495,
      "learning_rate": 1.5260730166238747e-06,
      "loss": 0.5279,
      "step": 2018
    },
    {
      "epoch": 0.3443042291950887,
      "grad_norm": 0.44351810216903687,
      "learning_rate": 1.5256032222389278e-06,
      "loss": 0.5325,
      "step": 2019
    },
    {
      "epoch": 0.34447476125511595,
      "grad_norm": 0.41944336891174316,
      "learning_rate": 1.5251332675151994e-06,
      "loss": 0.5376,
      "step": 2020
    },
    {
      "epoch": 0.34464529331514326,
      "grad_norm": 0.3723226487636566,
      "learning_rate": 1.5246631525960528e-06,
      "loss": 0.5357,
      "step": 2021
    },
    {
      "epoch": 0.3448158253751705,
      "grad_norm": 0.45165905356407166,
      "learning_rate": 1.5241928776248995e-06,
      "loss": 0.5277,
      "step": 2022
    },
    {
      "epoch": 0.34498635743519784,
      "grad_norm": 0.3919413685798645,
      "learning_rate": 1.5237224427452003e-06,
      "loss": 0.5169,
      "step": 2023
    },
    {
      "epoch": 0.3451568894952251,
      "grad_norm": 0.41514766216278076,
      "learning_rate": 1.5232518481004642e-06,
      "loss": 0.5273,
      "step": 2024
    },
    {
      "epoch": 0.34532742155525237,
      "grad_norm": 0.385572612285614,
      "learning_rate": 1.5227810938342495e-06,
      "loss": 0.5279,
      "step": 2025
    },
    {
      "epoch": 0.3454979536152797,
      "grad_norm": 0.42853543162345886,
      "learning_rate": 1.5223101800901627e-06,
      "loss": 0.5399,
      "step": 2026
    },
    {
      "epoch": 0.34566848567530695,
      "grad_norm": 0.39967915415763855,
      "learning_rate": 1.5218391070118592e-06,
      "loss": 0.5278,
      "step": 2027
    },
    {
      "epoch": 0.34583901773533426,
      "grad_norm": 0.44363322854042053,
      "learning_rate": 1.521367874743043e-06,
      "loss": 0.5257,
      "step": 2028
    },
    {
      "epoch": 0.3460095497953615,
      "grad_norm": 0.40447431802749634,
      "learning_rate": 1.520896483427467e-06,
      "loss": 0.5288,
      "step": 2029
    },
    {
      "epoch": 0.3461800818553888,
      "grad_norm": 0.4379490613937378,
      "learning_rate": 1.520424933208932e-06,
      "loss": 0.5274,
      "step": 2030
    },
    {
      "epoch": 0.3463506139154161,
      "grad_norm": 0.375988632440567,
      "learning_rate": 1.5199532242312872e-06,
      "loss": 0.5148,
      "step": 2031
    },
    {
      "epoch": 0.34652114597544337,
      "grad_norm": 0.38080793619155884,
      "learning_rate": 1.5194813566384313e-06,
      "loss": 0.53,
      "step": 2032
    },
    {
      "epoch": 0.3466916780354707,
      "grad_norm": 0.48653528094291687,
      "learning_rate": 1.5190093305743105e-06,
      "loss": 0.5278,
      "step": 2033
    },
    {
      "epoch": 0.34686221009549795,
      "grad_norm": 0.36069509387016296,
      "learning_rate": 1.5185371461829187e-06,
      "loss": 0.5204,
      "step": 2034
    },
    {
      "epoch": 0.34703274215552526,
      "grad_norm": 0.4373205602169037,
      "learning_rate": 1.5180648036083e-06,
      "loss": 0.531,
      "step": 2035
    },
    {
      "epoch": 0.3472032742155525,
      "grad_norm": 0.40521979331970215,
      "learning_rate": 1.517592302994545e-06,
      "loss": 0.5415,
      "step": 2036
    },
    {
      "epoch": 0.3473738062755798,
      "grad_norm": 0.37990278005599976,
      "learning_rate": 1.5171196444857935e-06,
      "loss": 0.5285,
      "step": 2037
    },
    {
      "epoch": 0.3475443383356071,
      "grad_norm": 0.4962809383869171,
      "learning_rate": 1.5166468282262328e-06,
      "loss": 0.5419,
      "step": 2038
    },
    {
      "epoch": 0.34771487039563437,
      "grad_norm": 0.37079575657844543,
      "learning_rate": 1.5161738543600991e-06,
      "loss": 0.5143,
      "step": 2039
    },
    {
      "epoch": 0.3478854024556617,
      "grad_norm": 0.4124794602394104,
      "learning_rate": 1.515700723031676e-06,
      "loss": 0.5289,
      "step": 2040
    },
    {
      "epoch": 0.34805593451568895,
      "grad_norm": 0.42842522263526917,
      "learning_rate": 1.5152274343852955e-06,
      "loss": 0.5228,
      "step": 2041
    },
    {
      "epoch": 0.3482264665757162,
      "grad_norm": 0.4419716000556946,
      "learning_rate": 1.5147539885653372e-06,
      "loss": 0.5255,
      "step": 2042
    },
    {
      "epoch": 0.3483969986357435,
      "grad_norm": 0.37058091163635254,
      "learning_rate": 1.514280385716229e-06,
      "loss": 0.5273,
      "step": 2043
    },
    {
      "epoch": 0.3485675306957708,
      "grad_norm": 0.4142124652862549,
      "learning_rate": 1.513806625982447e-06,
      "loss": 0.5241,
      "step": 2044
    },
    {
      "epoch": 0.3487380627557981,
      "grad_norm": 0.3658922016620636,
      "learning_rate": 1.513332709508514e-06,
      "loss": 0.5156,
      "step": 2045
    },
    {
      "epoch": 0.34890859481582537,
      "grad_norm": 0.41257330775260925,
      "learning_rate": 1.5128586364390022e-06,
      "loss": 0.5337,
      "step": 2046
    },
    {
      "epoch": 0.3490791268758527,
      "grad_norm": 0.40474775433540344,
      "learning_rate": 1.5123844069185307e-06,
      "loss": 0.5188,
      "step": 2047
    },
    {
      "epoch": 0.34924965893587995,
      "grad_norm": 0.4597495198249817,
      "learning_rate": 1.5119100210917658e-06,
      "loss": 0.5319,
      "step": 2048
    },
    {
      "epoch": 0.3494201909959072,
      "grad_norm": 0.4251413643360138,
      "learning_rate": 1.5114354791034226e-06,
      "loss": 0.5208,
      "step": 2049
    },
    {
      "epoch": 0.3495907230559345,
      "grad_norm": 0.37406793236732483,
      "learning_rate": 1.5109607810982628e-06,
      "loss": 0.5184,
      "step": 2050
    },
    {
      "epoch": 0.3497612551159618,
      "grad_norm": 0.47679436206817627,
      "learning_rate": 1.5104859272210966e-06,
      "loss": 0.5132,
      "step": 2051
    },
    {
      "epoch": 0.3499317871759891,
      "grad_norm": 0.45877325534820557,
      "learning_rate": 1.5100109176167815e-06,
      "loss": 0.5237,
      "step": 2052
    },
    {
      "epoch": 0.35010231923601637,
      "grad_norm": 0.4083709716796875,
      "learning_rate": 1.509535752430222e-06,
      "loss": 0.5328,
      "step": 2053
    },
    {
      "epoch": 0.35027285129604363,
      "grad_norm": 0.3754033148288727,
      "learning_rate": 1.5090604318063705e-06,
      "loss": 0.5276,
      "step": 2054
    },
    {
      "epoch": 0.35044338335607095,
      "grad_norm": 0.4184139370918274,
      "learning_rate": 1.5085849558902268e-06,
      "loss": 0.5178,
      "step": 2055
    },
    {
      "epoch": 0.3506139154160982,
      "grad_norm": 0.35778215527534485,
      "learning_rate": 1.5081093248268384e-06,
      "loss": 0.5261,
      "step": 2056
    },
    {
      "epoch": 0.3507844474761255,
      "grad_norm": 0.4060825705528259,
      "learning_rate": 1.507633538761299e-06,
      "loss": 0.5356,
      "step": 2057
    },
    {
      "epoch": 0.3509549795361528,
      "grad_norm": 0.35478487610816956,
      "learning_rate": 1.5071575978387506e-06,
      "loss": 0.5267,
      "step": 2058
    },
    {
      "epoch": 0.3511255115961801,
      "grad_norm": 0.42642742395401,
      "learning_rate": 1.506681502204382e-06,
      "loss": 0.5162,
      "step": 2059
    },
    {
      "epoch": 0.35129604365620737,
      "grad_norm": 0.4379814565181732,
      "learning_rate": 1.50620525200343e-06,
      "loss": 0.5341,
      "step": 2060
    },
    {
      "epoch": 0.35146657571623463,
      "grad_norm": 0.40005528926849365,
      "learning_rate": 1.5057288473811775e-06,
      "loss": 0.5218,
      "step": 2061
    },
    {
      "epoch": 0.35163710777626195,
      "grad_norm": 0.5353875756263733,
      "learning_rate": 1.5052522884829544e-06,
      "loss": 0.5281,
      "step": 2062
    },
    {
      "epoch": 0.3518076398362892,
      "grad_norm": 0.6442404389381409,
      "learning_rate": 1.5047755754541393e-06,
      "loss": 0.5289,
      "step": 2063
    },
    {
      "epoch": 0.3519781718963165,
      "grad_norm": 0.4252280294895172,
      "learning_rate": 1.5042987084401561e-06,
      "loss": 0.5277,
      "step": 2064
    },
    {
      "epoch": 0.3521487039563438,
      "grad_norm": 0.527929961681366,
      "learning_rate": 1.5038216875864758e-06,
      "loss": 0.5145,
      "step": 2065
    },
    {
      "epoch": 0.35231923601637105,
      "grad_norm": 0.41897308826446533,
      "learning_rate": 1.5033445130386176e-06,
      "loss": 0.5266,
      "step": 2066
    },
    {
      "epoch": 0.35248976807639837,
      "grad_norm": 0.5344181060791016,
      "learning_rate": 1.5028671849421465e-06,
      "loss": 0.5192,
      "step": 2067
    },
    {
      "epoch": 0.35266030013642563,
      "grad_norm": 0.47897011041641235,
      "learning_rate": 1.5023897034426743e-06,
      "loss": 0.5155,
      "step": 2068
    },
    {
      "epoch": 0.35283083219645295,
      "grad_norm": 0.4926980435848236,
      "learning_rate": 1.5019120686858605e-06,
      "loss": 0.5302,
      "step": 2069
    },
    {
      "epoch": 0.3530013642564802,
      "grad_norm": 0.6491872072219849,
      "learning_rate": 1.5014342808174105e-06,
      "loss": 0.5222,
      "step": 2070
    },
    {
      "epoch": 0.3531718963165075,
      "grad_norm": 0.4459693431854248,
      "learning_rate": 1.5009563399830766e-06,
      "loss": 0.5275,
      "step": 2071
    },
    {
      "epoch": 0.3533424283765348,
      "grad_norm": 0.44589418172836304,
      "learning_rate": 1.5004782463286582e-06,
      "loss": 0.5262,
      "step": 2072
    },
    {
      "epoch": 0.35351296043656205,
      "grad_norm": 0.38912469148635864,
      "learning_rate": 1.5000000000000002e-06,
      "loss": 0.5324,
      "step": 2073
    },
    {
      "epoch": 0.35368349249658937,
      "grad_norm": 0.4303382933139801,
      "learning_rate": 1.4995216011429958e-06,
      "loss": 0.5037,
      "step": 2074
    },
    {
      "epoch": 0.35385402455661663,
      "grad_norm": 0.52245032787323,
      "learning_rate": 1.4990430499035831e-06,
      "loss": 0.5237,
      "step": 2075
    },
    {
      "epoch": 0.35402455661664395,
      "grad_norm": 0.3739396333694458,
      "learning_rate": 1.4985643464277476e-06,
      "loss": 0.5173,
      "step": 2076
    },
    {
      "epoch": 0.3541950886766712,
      "grad_norm": 0.47938236594200134,
      "learning_rate": 1.4980854908615212e-06,
      "loss": 0.5119,
      "step": 2077
    },
    {
      "epoch": 0.3543656207366985,
      "grad_norm": 0.5574185252189636,
      "learning_rate": 1.4976064833509816e-06,
      "loss": 0.5169,
      "step": 2078
    },
    {
      "epoch": 0.3545361527967258,
      "grad_norm": 0.7085524797439575,
      "learning_rate": 1.4971273240422536e-06,
      "loss": 0.5241,
      "step": 2079
    },
    {
      "epoch": 0.35470668485675305,
      "grad_norm": 0.5693588256835938,
      "learning_rate": 1.4966480130815076e-06,
      "loss": 0.5168,
      "step": 2080
    },
    {
      "epoch": 0.35487721691678037,
      "grad_norm": 0.39197826385498047,
      "learning_rate": 1.496168550614961e-06,
      "loss": 0.5253,
      "step": 2081
    },
    {
      "epoch": 0.35504774897680763,
      "grad_norm": 0.43353649973869324,
      "learning_rate": 1.4956889367888765e-06,
      "loss": 0.5288,
      "step": 2082
    },
    {
      "epoch": 0.35521828103683495,
      "grad_norm": 0.41520053148269653,
      "learning_rate": 1.4952091717495642e-06,
      "loss": 0.5193,
      "step": 2083
    },
    {
      "epoch": 0.3553888130968622,
      "grad_norm": 0.5020126104354858,
      "learning_rate": 1.4947292556433791e-06,
      "loss": 0.5348,
      "step": 2084
    },
    {
      "epoch": 0.3555593451568895,
      "grad_norm": 0.3375995457172394,
      "learning_rate": 1.4942491886167231e-06,
      "loss": 0.5184,
      "step": 2085
    },
    {
      "epoch": 0.3557298772169168,
      "grad_norm": 0.5305654406547546,
      "learning_rate": 1.4937689708160435e-06,
      "loss": 0.5225,
      "step": 2086
    },
    {
      "epoch": 0.35590040927694405,
      "grad_norm": 0.3381396234035492,
      "learning_rate": 1.4932886023878347e-06,
      "loss": 0.516,
      "step": 2087
    },
    {
      "epoch": 0.35607094133697137,
      "grad_norm": 0.5799662470817566,
      "learning_rate": 1.4928080834786358e-06,
      "loss": 0.5087,
      "step": 2088
    },
    {
      "epoch": 0.35624147339699863,
      "grad_norm": 0.582531750202179,
      "learning_rate": 1.492327414235032e-06,
      "loss": 0.5208,
      "step": 2089
    },
    {
      "epoch": 0.3564120054570259,
      "grad_norm": 0.36300772428512573,
      "learning_rate": 1.491846594803655e-06,
      "loss": 0.5227,
      "step": 2090
    },
    {
      "epoch": 0.3565825375170532,
      "grad_norm": 0.5540021657943726,
      "learning_rate": 1.4913656253311822e-06,
      "loss": 0.5191,
      "step": 2091
    },
    {
      "epoch": 0.3567530695770805,
      "grad_norm": 0.4059838652610779,
      "learning_rate": 1.490884505964336e-06,
      "loss": 0.5174,
      "step": 2092
    },
    {
      "epoch": 0.3569236016371078,
      "grad_norm": 0.47956252098083496,
      "learning_rate": 1.4904032368498858e-06,
      "loss": 0.5095,
      "step": 2093
    },
    {
      "epoch": 0.35709413369713505,
      "grad_norm": 0.4530734717845917,
      "learning_rate": 1.4899218181346455e-06,
      "loss": 0.5362,
      "step": 2094
    },
    {
      "epoch": 0.35726466575716237,
      "grad_norm": 0.5519397854804993,
      "learning_rate": 1.4894402499654751e-06,
      "loss": 0.5253,
      "step": 2095
    },
    {
      "epoch": 0.35743519781718963,
      "grad_norm": 0.39683836698532104,
      "learning_rate": 1.4889585324892805e-06,
      "loss": 0.515,
      "step": 2096
    },
    {
      "epoch": 0.3576057298772169,
      "grad_norm": 0.605188250541687,
      "learning_rate": 1.4884766658530128e-06,
      "loss": 0.5167,
      "step": 2097
    },
    {
      "epoch": 0.3577762619372442,
      "grad_norm": 0.4661652445793152,
      "learning_rate": 1.4879946502036679e-06,
      "loss": 0.5241,
      "step": 2098
    },
    {
      "epoch": 0.3579467939972715,
      "grad_norm": 0.5223652720451355,
      "learning_rate": 1.4875124856882886e-06,
      "loss": 0.5177,
      "step": 2099
    },
    {
      "epoch": 0.3581173260572988,
      "grad_norm": 0.6159395575523376,
      "learning_rate": 1.4870301724539627e-06,
      "loss": 0.5001,
      "step": 2100
    },
    {
      "epoch": 0.35828785811732605,
      "grad_norm": 0.4143725633621216,
      "learning_rate": 1.4865477106478222e-06,
      "loss": 0.5057,
      "step": 2101
    },
    {
      "epoch": 0.35845839017735337,
      "grad_norm": 0.6344066262245178,
      "learning_rate": 1.4860651004170463e-06,
      "loss": 0.5356,
      "step": 2102
    },
    {
      "epoch": 0.35862892223738063,
      "grad_norm": 0.4035511910915375,
      "learning_rate": 1.4855823419088577e-06,
      "loss": 0.5212,
      "step": 2103
    },
    {
      "epoch": 0.3587994542974079,
      "grad_norm": 0.5761251449584961,
      "learning_rate": 1.4850994352705251e-06,
      "loss": 0.5194,
      "step": 2104
    },
    {
      "epoch": 0.3589699863574352,
      "grad_norm": 0.38429519534111023,
      "learning_rate": 1.4846163806493628e-06,
      "loss": 0.5078,
      "step": 2105
    },
    {
      "epoch": 0.3591405184174625,
      "grad_norm": 0.5760840177536011,
      "learning_rate": 1.4841331781927298e-06,
      "loss": 0.5188,
      "step": 2106
    },
    {
      "epoch": 0.3593110504774898,
      "grad_norm": 0.4178948700428009,
      "learning_rate": 1.4836498280480301e-06,
      "loss": 0.5174,
      "step": 2107
    },
    {
      "epoch": 0.35948158253751705,
      "grad_norm": 0.397052526473999,
      "learning_rate": 1.483166330362713e-06,
      "loss": 0.5159,
      "step": 2108
    },
    {
      "epoch": 0.3596521145975443,
      "grad_norm": 0.43821024894714355,
      "learning_rate": 1.4826826852842727e-06,
      "loss": 0.5144,
      "step": 2109
    },
    {
      "epoch": 0.35982264665757163,
      "grad_norm": 0.5855873227119446,
      "learning_rate": 1.4821988929602482e-06,
      "loss": 0.5122,
      "step": 2110
    },
    {
      "epoch": 0.3599931787175989,
      "grad_norm": 0.5009761452674866,
      "learning_rate": 1.4817149535382239e-06,
      "loss": 0.513,
      "step": 2111
    },
    {
      "epoch": 0.3601637107776262,
      "grad_norm": 0.5092044472694397,
      "learning_rate": 1.4812308671658285e-06,
      "loss": 0.5117,
      "step": 2112
    },
    {
      "epoch": 0.3603342428376535,
      "grad_norm": 0.3675675690174103,
      "learning_rate": 1.4807466339907358e-06,
      "loss": 0.5035,
      "step": 2113
    },
    {
      "epoch": 0.3605047748976808,
      "grad_norm": 0.42256852984428406,
      "learning_rate": 1.4802622541606645e-06,
      "loss": 0.5109,
      "step": 2114
    },
    {
      "epoch": 0.36067530695770805,
      "grad_norm": 0.48379117250442505,
      "learning_rate": 1.4797777278233779e-06,
      "loss": 0.5182,
      "step": 2115
    },
    {
      "epoch": 0.3608458390177353,
      "grad_norm": 0.4444460868835449,
      "learning_rate": 1.4792930551266836e-06,
      "loss": 0.522,
      "step": 2116
    },
    {
      "epoch": 0.36101637107776263,
      "grad_norm": 0.4920353889465332,
      "learning_rate": 1.478808236218435e-06,
      "loss": 0.5078,
      "step": 2117
    },
    {
      "epoch": 0.3611869031377899,
      "grad_norm": 0.45749807357788086,
      "learning_rate": 1.4783232712465289e-06,
      "loss": 0.5082,
      "step": 2118
    },
    {
      "epoch": 0.3613574351978172,
      "grad_norm": 0.48693403601646423,
      "learning_rate": 1.4778381603589071e-06,
      "loss": 0.508,
      "step": 2119
    },
    {
      "epoch": 0.3615279672578445,
      "grad_norm": 0.35000139474868774,
      "learning_rate": 1.477352903703556e-06,
      "loss": 0.5108,
      "step": 2120
    },
    {
      "epoch": 0.36169849931787174,
      "grad_norm": 0.39079880714416504,
      "learning_rate": 1.4768675014285065e-06,
      "loss": 0.5207,
      "step": 2121
    },
    {
      "epoch": 0.36186903137789905,
      "grad_norm": 0.48962411284446716,
      "learning_rate": 1.4763819536818332e-06,
      "loss": 0.5146,
      "step": 2122
    },
    {
      "epoch": 0.3620395634379263,
      "grad_norm": 0.423258900642395,
      "learning_rate": 1.4758962606116566e-06,
      "loss": 0.5147,
      "step": 2123
    },
    {
      "epoch": 0.36221009549795363,
      "grad_norm": 0.414676308631897,
      "learning_rate": 1.4754104223661402e-06,
      "loss": 0.5241,
      "step": 2124
    },
    {
      "epoch": 0.3623806275579809,
      "grad_norm": 0.4666195809841156,
      "learning_rate": 1.4749244390934922e-06,
      "loss": 0.5182,
      "step": 2125
    },
    {
      "epoch": 0.3625511596180082,
      "grad_norm": 0.4646506905555725,
      "learning_rate": 1.4744383109419649e-06,
      "loss": 0.5143,
      "step": 2126
    },
    {
      "epoch": 0.3627216916780355,
      "grad_norm": 0.48451265692710876,
      "learning_rate": 1.4739520380598552e-06,
      "loss": 0.5187,
      "step": 2127
    },
    {
      "epoch": 0.36289222373806274,
      "grad_norm": 0.4569034278392792,
      "learning_rate": 1.473465620595504e-06,
      "loss": 0.5135,
      "step": 2128
    },
    {
      "epoch": 0.36306275579809005,
      "grad_norm": 0.4225984215736389,
      "learning_rate": 1.4729790586972957e-06,
      "loss": 0.5251,
      "step": 2129
    },
    {
      "epoch": 0.3632332878581173,
      "grad_norm": 0.6046484112739563,
      "learning_rate": 1.4724923525136597e-06,
      "loss": 0.515,
      "step": 2130
    },
    {
      "epoch": 0.36340381991814463,
      "grad_norm": 0.6546425223350525,
      "learning_rate": 1.4720055021930688e-06,
      "loss": 0.5179,
      "step": 2131
    },
    {
      "epoch": 0.3635743519781719,
      "grad_norm": 0.4017772972583771,
      "learning_rate": 1.4715185078840402e-06,
      "loss": 0.5184,
      "step": 2132
    },
    {
      "epoch": 0.36374488403819916,
      "grad_norm": 0.4456183612346649,
      "learning_rate": 1.4710313697351343e-06,
      "loss": 0.5152,
      "step": 2133
    },
    {
      "epoch": 0.3639154160982265,
      "grad_norm": 0.4506753981113434,
      "learning_rate": 1.470544087894956e-06,
      "loss": 0.5225,
      "step": 2134
    },
    {
      "epoch": 0.36408594815825374,
      "grad_norm": 0.5804704427719116,
      "learning_rate": 1.4700566625121545e-06,
      "loss": 0.5212,
      "step": 2135
    },
    {
      "epoch": 0.36425648021828105,
      "grad_norm": 0.564107358455658,
      "learning_rate": 1.4695690937354214e-06,
      "loss": 0.5107,
      "step": 2136
    },
    {
      "epoch": 0.3644270122783083,
      "grad_norm": 0.3576640784740448,
      "learning_rate": 1.469081381713493e-06,
      "loss": 0.5176,
      "step": 2137
    },
    {
      "epoch": 0.36459754433833563,
      "grad_norm": 0.5520999431610107,
      "learning_rate": 1.4685935265951495e-06,
      "loss": 0.5277,
      "step": 2138
    },
    {
      "epoch": 0.3647680763983629,
      "grad_norm": 0.5637651085853577,
      "learning_rate": 1.468105528529214e-06,
      "loss": 0.5197,
      "step": 2139
    },
    {
      "epoch": 0.36493860845839016,
      "grad_norm": 0.4603327512741089,
      "learning_rate": 1.4676173876645537e-06,
      "loss": 0.5225,
      "step": 2140
    },
    {
      "epoch": 0.3651091405184175,
      "grad_norm": 0.6356070637702942,
      "learning_rate": 1.4671291041500792e-06,
      "loss": 0.5365,
      "step": 2141
    },
    {
      "epoch": 0.36527967257844474,
      "grad_norm": 0.3971726596355438,
      "learning_rate": 1.4666406781347446e-06,
      "loss": 0.5232,
      "step": 2142
    },
    {
      "epoch": 0.36545020463847205,
      "grad_norm": 0.5633137226104736,
      "learning_rate": 1.4661521097675479e-06,
      "loss": 0.5296,
      "step": 2143
    },
    {
      "epoch": 0.3656207366984993,
      "grad_norm": 0.49675676226615906,
      "learning_rate": 1.4656633991975295e-06,
      "loss": 0.5449,
      "step": 2144
    },
    {
      "epoch": 0.3657912687585266,
      "grad_norm": 0.48119738698005676,
      "learning_rate": 1.465174546573774e-06,
      "loss": 0.5168,
      "step": 2145
    },
    {
      "epoch": 0.3659618008185539,
      "grad_norm": 0.5705353617668152,
      "learning_rate": 1.4646855520454097e-06,
      "loss": 0.5262,
      "step": 2146
    },
    {
      "epoch": 0.36613233287858116,
      "grad_norm": 0.48952609300613403,
      "learning_rate": 1.464196415761607e-06,
      "loss": 0.5217,
      "step": 2147
    },
    {
      "epoch": 0.3663028649386085,
      "grad_norm": 0.5046576261520386,
      "learning_rate": 1.4637071378715806e-06,
      "loss": 0.5298,
      "step": 2148
    },
    {
      "epoch": 0.36647339699863574,
      "grad_norm": 0.4642289876937866,
      "learning_rate": 1.463217718524588e-06,
      "loss": 0.5328,
      "step": 2149
    },
    {
      "epoch": 0.36664392905866305,
      "grad_norm": 0.5076698064804077,
      "learning_rate": 1.46272815786993e-06,
      "loss": 0.5161,
      "step": 2150
    },
    {
      "epoch": 0.3668144611186903,
      "grad_norm": 0.4147416949272156,
      "learning_rate": 1.4622384560569495e-06,
      "loss": 0.523,
      "step": 2151
    },
    {
      "epoch": 0.3669849931787176,
      "grad_norm": 0.5995153188705444,
      "learning_rate": 1.4617486132350345e-06,
      "loss": 0.5182,
      "step": 2152
    },
    {
      "epoch": 0.3671555252387449,
      "grad_norm": 0.5449314713478088,
      "learning_rate": 1.4612586295536136e-06,
      "loss": 0.5188,
      "step": 2153
    },
    {
      "epoch": 0.36732605729877216,
      "grad_norm": 0.6206721067428589,
      "learning_rate": 1.4607685051621605e-06,
      "loss": 0.5346,
      "step": 2154
    },
    {
      "epoch": 0.3674965893587995,
      "grad_norm": 0.3753843307495117,
      "learning_rate": 1.4602782402101909e-06,
      "loss": 0.5253,
      "step": 2155
    },
    {
      "epoch": 0.36766712141882674,
      "grad_norm": 0.5091269016265869,
      "learning_rate": 1.4597878348472626e-06,
      "loss": 0.5293,
      "step": 2156
    },
    {
      "epoch": 0.367837653478854,
      "grad_norm": 0.46207624673843384,
      "learning_rate": 1.4592972892229782e-06,
      "loss": 0.5218,
      "step": 2157
    },
    {
      "epoch": 0.3680081855388813,
      "grad_norm": 0.47035977244377136,
      "learning_rate": 1.458806603486981e-06,
      "loss": 0.5433,
      "step": 2158
    },
    {
      "epoch": 0.3681787175989086,
      "grad_norm": 0.5492236614227295,
      "learning_rate": 1.4583157777889587e-06,
      "loss": 0.5431,
      "step": 2159
    },
    {
      "epoch": 0.3683492496589359,
      "grad_norm": 0.4612725079059601,
      "learning_rate": 1.4578248122786402e-06,
      "loss": 0.5256,
      "step": 2160
    },
    {
      "epoch": 0.36851978171896316,
      "grad_norm": 0.4277294874191284,
      "learning_rate": 1.4573337071057982e-06,
      "loss": 0.5245,
      "step": 2161
    },
    {
      "epoch": 0.3686903137789905,
      "grad_norm": 0.6318309307098389,
      "learning_rate": 1.4568424624202476e-06,
      "loss": 0.5223,
      "step": 2162
    },
    {
      "epoch": 0.36886084583901774,
      "grad_norm": 0.5150088667869568,
      "learning_rate": 1.456351078371846e-06,
      "loss": 0.5376,
      "step": 2163
    },
    {
      "epoch": 0.369031377899045,
      "grad_norm": 0.35254305601119995,
      "learning_rate": 1.4558595551104932e-06,
      "loss": 0.5078,
      "step": 2164
    },
    {
      "epoch": 0.3692019099590723,
      "grad_norm": 0.47404444217681885,
      "learning_rate": 1.4553678927861316e-06,
      "loss": 0.5224,
      "step": 2165
    },
    {
      "epoch": 0.3693724420190996,
      "grad_norm": 0.3993200957775116,
      "learning_rate": 1.4548760915487465e-06,
      "loss": 0.5354,
      "step": 2166
    },
    {
      "epoch": 0.3695429740791269,
      "grad_norm": 0.47942835092544556,
      "learning_rate": 1.4543841515483647e-06,
      "loss": 0.5307,
      "step": 2167
    },
    {
      "epoch": 0.36971350613915416,
      "grad_norm": 0.49483659863471985,
      "learning_rate": 1.453892072935056e-06,
      "loss": 0.5184,
      "step": 2168
    },
    {
      "epoch": 0.3698840381991814,
      "grad_norm": 0.35012009739875793,
      "learning_rate": 1.453399855858932e-06,
      "loss": 0.5275,
      "step": 2169
    },
    {
      "epoch": 0.37005457025920874,
      "grad_norm": 0.5223888158798218,
      "learning_rate": 1.4529075004701475e-06,
      "loss": 0.544,
      "step": 2170
    },
    {
      "epoch": 0.370225102319236,
      "grad_norm": 0.49425366520881653,
      "learning_rate": 1.452415006918898e-06,
      "loss": 0.5196,
      "step": 2171
    },
    {
      "epoch": 0.3703956343792633,
      "grad_norm": 0.4764786660671234,
      "learning_rate": 1.4519223753554224e-06,
      "loss": 0.5229,
      "step": 2172
    },
    {
      "epoch": 0.3705661664392906,
      "grad_norm": 0.4898337125778198,
      "learning_rate": 1.4514296059300016e-06,
      "loss": 0.5238,
      "step": 2173
    },
    {
      "epoch": 0.3707366984993179,
      "grad_norm": 0.41198065876960754,
      "learning_rate": 1.4509366987929576e-06,
      "loss": 0.5399,
      "step": 2174
    },
    {
      "epoch": 0.37090723055934516,
      "grad_norm": 0.5006463527679443,
      "learning_rate": 1.4504436540946547e-06,
      "loss": 0.5251,
      "step": 2175
    },
    {
      "epoch": 0.3710777626193724,
      "grad_norm": 0.3775595426559448,
      "learning_rate": 1.4499504719855003e-06,
      "loss": 0.5187,
      "step": 2176
    },
    {
      "epoch": 0.37124829467939974,
      "grad_norm": 0.4422166645526886,
      "learning_rate": 1.4494571526159427e-06,
      "loss": 0.5182,
      "step": 2177
    },
    {
      "epoch": 0.371418826739427,
      "grad_norm": 0.5421523451805115,
      "learning_rate": 1.448963696136472e-06,
      "loss": 0.5196,
      "step": 2178
    },
    {
      "epoch": 0.3715893587994543,
      "grad_norm": 0.43927666544914246,
      "learning_rate": 1.4484701026976207e-06,
      "loss": 0.5293,
      "step": 2179
    },
    {
      "epoch": 0.3717598908594816,
      "grad_norm": 0.48039036989212036,
      "learning_rate": 1.4479763724499627e-06,
      "loss": 0.5324,
      "step": 2180
    },
    {
      "epoch": 0.37193042291950884,
      "grad_norm": 0.46991389989852905,
      "learning_rate": 1.4474825055441137e-06,
      "loss": 0.5313,
      "step": 2181
    },
    {
      "epoch": 0.37210095497953616,
      "grad_norm": 0.45620325207710266,
      "learning_rate": 1.446988502130731e-06,
      "loss": 0.5234,
      "step": 2182
    },
    {
      "epoch": 0.3722714870395634,
      "grad_norm": 0.514532208442688,
      "learning_rate": 1.4464943623605139e-06,
      "loss": 0.5377,
      "step": 2183
    },
    {
      "epoch": 0.37244201909959074,
      "grad_norm": 0.44968950748443604,
      "learning_rate": 1.4460000863842024e-06,
      "loss": 0.5345,
      "step": 2184
    },
    {
      "epoch": 0.372612551159618,
      "grad_norm": 0.4595887362957001,
      "learning_rate": 1.4455056743525794e-06,
      "loss": 0.5111,
      "step": 2185
    },
    {
      "epoch": 0.3727830832196453,
      "grad_norm": 0.5584328174591064,
      "learning_rate": 1.4450111264164684e-06,
      "loss": 0.5284,
      "step": 2186
    },
    {
      "epoch": 0.3729536152796726,
      "grad_norm": 0.394270658493042,
      "learning_rate": 1.4445164427267346e-06,
      "loss": 0.5194,
      "step": 2187
    },
    {
      "epoch": 0.37312414733969984,
      "grad_norm": 0.4683585464954376,
      "learning_rate": 1.4440216234342842e-06,
      "loss": 0.5328,
      "step": 2188
    },
    {
      "epoch": 0.37329467939972716,
      "grad_norm": 0.4643068313598633,
      "learning_rate": 1.443526668690066e-06,
      "loss": 0.5389,
      "step": 2189
    },
    {
      "epoch": 0.3734652114597544,
      "grad_norm": 0.38309845328330994,
      "learning_rate": 1.4430315786450685e-06,
      "loss": 0.5294,
      "step": 2190
    },
    {
      "epoch": 0.37363574351978174,
      "grad_norm": 0.46844732761383057,
      "learning_rate": 1.442536353450322e-06,
      "loss": 0.5072,
      "step": 2191
    },
    {
      "epoch": 0.373806275579809,
      "grad_norm": 0.48929598927497864,
      "learning_rate": 1.4420409932568992e-06,
      "loss": 0.5241,
      "step": 2192
    },
    {
      "epoch": 0.37397680763983626,
      "grad_norm": 0.5985404849052429,
      "learning_rate": 1.4415454982159123e-06,
      "loss": 0.5243,
      "step": 2193
    },
    {
      "epoch": 0.3741473396998636,
      "grad_norm": 0.4248568117618561,
      "learning_rate": 1.4410498684785153e-06,
      "loss": 0.5336,
      "step": 2194
    },
    {
      "epoch": 0.37431787175989084,
      "grad_norm": 0.4180203676223755,
      "learning_rate": 1.440554104195904e-06,
      "loss": 0.5286,
      "step": 2195
    },
    {
      "epoch": 0.37448840381991816,
      "grad_norm": 0.4288659393787384,
      "learning_rate": 1.4400582055193143e-06,
      "loss": 0.5372,
      "step": 2196
    },
    {
      "epoch": 0.3746589358799454,
      "grad_norm": 0.4723179340362549,
      "learning_rate": 1.4395621726000235e-06,
      "loss": 0.5226,
      "step": 2197
    },
    {
      "epoch": 0.37482946793997274,
      "grad_norm": 0.48728570342063904,
      "learning_rate": 1.4390660055893496e-06,
      "loss": 0.5294,
      "step": 2198
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.4278232157230377,
      "learning_rate": 1.4385697046386515e-06,
      "loss": 0.52,
      "step": 2199
    },
    {
      "epoch": 0.37517053206002726,
      "grad_norm": 0.37193500995635986,
      "learning_rate": 1.4380732698993294e-06,
      "loss": 0.5189,
      "step": 2200
    },
    {
      "epoch": 0.3753410641200546,
      "grad_norm": 0.5276831388473511,
      "learning_rate": 1.437576701522824e-06,
      "loss": 0.543,
      "step": 2201
    },
    {
      "epoch": 0.37551159618008184,
      "grad_norm": 0.485494464635849,
      "learning_rate": 1.437079999660617e-06,
      "loss": 0.5206,
      "step": 2202
    },
    {
      "epoch": 0.37568212824010916,
      "grad_norm": 0.4124785363674164,
      "learning_rate": 1.4365831644642302e-06,
      "loss": 0.5146,
      "step": 2203
    },
    {
      "epoch": 0.3758526603001364,
      "grad_norm": 0.4600485563278198,
      "learning_rate": 1.436086196085227e-06,
      "loss": 0.5115,
      "step": 2204
    },
    {
      "epoch": 0.37602319236016374,
      "grad_norm": 0.4196353256702423,
      "learning_rate": 1.4355890946752103e-06,
      "loss": 0.5269,
      "step": 2205
    },
    {
      "epoch": 0.376193724420191,
      "grad_norm": 0.4327806532382965,
      "learning_rate": 1.435091860385825e-06,
      "loss": 0.5141,
      "step": 2206
    },
    {
      "epoch": 0.37636425648021826,
      "grad_norm": 0.3661247491836548,
      "learning_rate": 1.434594493368755e-06,
      "loss": 0.5096,
      "step": 2207
    },
    {
      "epoch": 0.3765347885402456,
      "grad_norm": 0.45045968890190125,
      "learning_rate": 1.4340969937757259e-06,
      "loss": 0.5304,
      "step": 2208
    },
    {
      "epoch": 0.37670532060027284,
      "grad_norm": 0.4632346034049988,
      "learning_rate": 1.433599361758503e-06,
      "loss": 0.5364,
      "step": 2209
    },
    {
      "epoch": 0.37687585266030016,
      "grad_norm": 0.44703951478004456,
      "learning_rate": 1.4331015974688925e-06,
      "loss": 0.5356,
      "step": 2210
    },
    {
      "epoch": 0.3770463847203274,
      "grad_norm": 0.5135074257850647,
      "learning_rate": 1.4326037010587407e-06,
      "loss": 0.5191,
      "step": 2211
    },
    {
      "epoch": 0.3772169167803547,
      "grad_norm": 0.407526433467865,
      "learning_rate": 1.432105672679934e-06,
      "loss": 0.5211,
      "step": 2212
    },
    {
      "epoch": 0.377387448840382,
      "grad_norm": 0.4435472786426544,
      "learning_rate": 1.4316075124843996e-06,
      "loss": 0.5214,
      "step": 2213
    },
    {
      "epoch": 0.37755798090040926,
      "grad_norm": 0.5043038725852966,
      "learning_rate": 1.4311092206241048e-06,
      "loss": 0.5227,
      "step": 2214
    },
    {
      "epoch": 0.3777285129604366,
      "grad_norm": 0.3871316611766815,
      "learning_rate": 1.430610797251056e-06,
      "loss": 0.5277,
      "step": 2215
    },
    {
      "epoch": 0.37789904502046384,
      "grad_norm": 0.48304349184036255,
      "learning_rate": 1.4301122425173015e-06,
      "loss": 0.5209,
      "step": 2216
    },
    {
      "epoch": 0.37806957708049116,
      "grad_norm": 0.5158309936523438,
      "learning_rate": 1.4296135565749281e-06,
      "loss": 0.5238,
      "step": 2217
    },
    {
      "epoch": 0.3782401091405184,
      "grad_norm": 0.3965489864349365,
      "learning_rate": 1.4291147395760637e-06,
      "loss": 0.5383,
      "step": 2218
    },
    {
      "epoch": 0.3784106412005457,
      "grad_norm": 0.47193005681037903,
      "learning_rate": 1.4286157916728757e-06,
      "loss": 0.5146,
      "step": 2219
    },
    {
      "epoch": 0.378581173260573,
      "grad_norm": 0.4689446985721588,
      "learning_rate": 1.4281167130175714e-06,
      "loss": 0.5243,
      "step": 2220
    },
    {
      "epoch": 0.37875170532060026,
      "grad_norm": 0.43518558144569397,
      "learning_rate": 1.4276175037623982e-06,
      "loss": 0.521,
      "step": 2221
    },
    {
      "epoch": 0.3789222373806276,
      "grad_norm": 0.34477195143699646,
      "learning_rate": 1.4271181640596429e-06,
      "loss": 0.5194,
      "step": 2222
    },
    {
      "epoch": 0.37909276944065484,
      "grad_norm": 0.538991391658783,
      "learning_rate": 1.426618694061633e-06,
      "loss": 0.5217,
      "step": 2223
    },
    {
      "epoch": 0.3792633015006821,
      "grad_norm": 0.5617110729217529,
      "learning_rate": 1.4261190939207345e-06,
      "loss": 0.5058,
      "step": 2224
    },
    {
      "epoch": 0.3794338335607094,
      "grad_norm": 0.6033090353012085,
      "learning_rate": 1.4256193637893542e-06,
      "loss": 0.5331,
      "step": 2225
    },
    {
      "epoch": 0.3796043656207367,
      "grad_norm": 0.5407723188400269,
      "learning_rate": 1.425119503819938e-06,
      "loss": 0.5353,
      "step": 2226
    },
    {
      "epoch": 0.379774897680764,
      "grad_norm": 0.5833804607391357,
      "learning_rate": 1.4246195141649717e-06,
      "loss": 0.5204,
      "step": 2227
    },
    {
      "epoch": 0.37994542974079126,
      "grad_norm": 0.727768063545227,
      "learning_rate": 1.4241193949769802e-06,
      "loss": 0.5283,
      "step": 2228
    },
    {
      "epoch": 0.3801159618008186,
      "grad_norm": 0.4322068989276886,
      "learning_rate": 1.4236191464085287e-06,
      "loss": 0.5353,
      "step": 2229
    },
    {
      "epoch": 0.38028649386084584,
      "grad_norm": 0.4498087167739868,
      "learning_rate": 1.4231187686122205e-06,
      "loss": 0.5273,
      "step": 2230
    },
    {
      "epoch": 0.3804570259208731,
      "grad_norm": 0.6119716763496399,
      "learning_rate": 1.4226182617406998e-06,
      "loss": 0.5331,
      "step": 2231
    },
    {
      "epoch": 0.3806275579809004,
      "grad_norm": 0.3565307855606079,
      "learning_rate": 1.4221176259466495e-06,
      "loss": 0.5248,
      "step": 2232
    },
    {
      "epoch": 0.3807980900409277,
      "grad_norm": 0.5091411471366882,
      "learning_rate": 1.421616861382792e-06,
      "loss": 0.5385,
      "step": 2233
    },
    {
      "epoch": 0.380968622100955,
      "grad_norm": 0.5057311058044434,
      "learning_rate": 1.4211159682018886e-06,
      "loss": 0.5155,
      "step": 2234
    },
    {
      "epoch": 0.38113915416098226,
      "grad_norm": 0.37856343388557434,
      "learning_rate": 1.4206149465567404e-06,
      "loss": 0.5205,
      "step": 2235
    },
    {
      "epoch": 0.3813096862210095,
      "grad_norm": 0.6066394448280334,
      "learning_rate": 1.4201137966001875e-06,
      "loss": 0.5244,
      "step": 2236
    },
    {
      "epoch": 0.38148021828103684,
      "grad_norm": 0.4703941345214844,
      "learning_rate": 1.4196125184851085e-06,
      "loss": 0.5485,
      "step": 2237
    },
    {
      "epoch": 0.3816507503410641,
      "grad_norm": 0.402211993932724,
      "learning_rate": 1.4191111123644223e-06,
      "loss": 0.526,
      "step": 2238
    },
    {
      "epoch": 0.3818212824010914,
      "grad_norm": 0.5045869946479797,
      "learning_rate": 1.4186095783910856e-06,
      "loss": 0.5225,
      "step": 2239
    },
    {
      "epoch": 0.3819918144611187,
      "grad_norm": 0.5181368589401245,
      "learning_rate": 1.418107916718095e-06,
      "loss": 0.529,
      "step": 2240
    },
    {
      "epoch": 0.382162346521146,
      "grad_norm": 0.4280795753002167,
      "learning_rate": 1.417606127498486e-06,
      "loss": 0.5219,
      "step": 2241
    },
    {
      "epoch": 0.38233287858117326,
      "grad_norm": 0.526380717754364,
      "learning_rate": 1.417104210885332e-06,
      "loss": 0.5159,
      "step": 2242
    },
    {
      "epoch": 0.3825034106412005,
      "grad_norm": 0.5254361033439636,
      "learning_rate": 1.4166021670317466e-06,
      "loss": 0.522,
      "step": 2243
    },
    {
      "epoch": 0.38267394270122784,
      "grad_norm": 0.626717209815979,
      "learning_rate": 1.416099996090882e-06,
      "loss": 0.5169,
      "step": 2244
    },
    {
      "epoch": 0.3828444747612551,
      "grad_norm": 0.546101450920105,
      "learning_rate": 1.4155976982159281e-06,
      "loss": 0.5239,
      "step": 2245
    },
    {
      "epoch": 0.3830150068212824,
      "grad_norm": 0.48548194766044617,
      "learning_rate": 1.4150952735601144e-06,
      "loss": 0.5138,
      "step": 2246
    },
    {
      "epoch": 0.3831855388813097,
      "grad_norm": 0.43900060653686523,
      "learning_rate": 1.4145927222767092e-06,
      "loss": 0.5182,
      "step": 2247
    },
    {
      "epoch": 0.38335607094133695,
      "grad_norm": 0.44476473331451416,
      "learning_rate": 1.4140900445190188e-06,
      "loss": 0.5223,
      "step": 2248
    },
    {
      "epoch": 0.38352660300136426,
      "grad_norm": 0.4795719385147095,
      "learning_rate": 1.4135872404403883e-06,
      "loss": 0.5306,
      "step": 2249
    },
    {
      "epoch": 0.3836971350613915,
      "grad_norm": 0.39385172724723816,
      "learning_rate": 1.413084310194202e-06,
      "loss": 0.5283,
      "step": 2250
    },
    {
      "epoch": 0.38386766712141884,
      "grad_norm": 0.5722368955612183,
      "learning_rate": 1.4125812539338817e-06,
      "loss": 0.5322,
      "step": 2251
    },
    {
      "epoch": 0.3840381991814461,
      "grad_norm": 0.42854925990104675,
      "learning_rate": 1.412078071812888e-06,
      "loss": 0.5328,
      "step": 2252
    },
    {
      "epoch": 0.3842087312414734,
      "grad_norm": 0.4990580677986145,
      "learning_rate": 1.4115747639847205e-06,
      "loss": 0.529,
      "step": 2253
    },
    {
      "epoch": 0.3843792633015007,
      "grad_norm": 0.38250014185905457,
      "learning_rate": 1.411071330602916e-06,
      "loss": 0.5127,
      "step": 2254
    },
    {
      "epoch": 0.38454979536152795,
      "grad_norm": 0.5111120939254761,
      "learning_rate": 1.4105677718210508e-06,
      "loss": 0.5226,
      "step": 2255
    },
    {
      "epoch": 0.38472032742155526,
      "grad_norm": 0.3982873260974884,
      "learning_rate": 1.4100640877927385e-06,
      "loss": 0.5284,
      "step": 2256
    },
    {
      "epoch": 0.3848908594815825,
      "grad_norm": 0.44653573632240295,
      "learning_rate": 1.4095602786716312e-06,
      "loss": 0.5231,
      "step": 2257
    },
    {
      "epoch": 0.38506139154160984,
      "grad_norm": 0.4471384286880493,
      "learning_rate": 1.4090563446114197e-06,
      "loss": 0.5169,
      "step": 2258
    },
    {
      "epoch": 0.3852319236016371,
      "grad_norm": 0.48839473724365234,
      "learning_rate": 1.4085522857658321e-06,
      "loss": 0.5165,
      "step": 2259
    },
    {
      "epoch": 0.38540245566166437,
      "grad_norm": 0.5165295004844666,
      "learning_rate": 1.408048102288635e-06,
      "loss": 0.5246,
      "step": 2260
    },
    {
      "epoch": 0.3855729877216917,
      "grad_norm": 0.45982906222343445,
      "learning_rate": 1.4075437943336328e-06,
      "loss": 0.5046,
      "step": 2261
    },
    {
      "epoch": 0.38574351978171895,
      "grad_norm": 0.5657344460487366,
      "learning_rate": 1.4070393620546685e-06,
      "loss": 0.5252,
      "step": 2262
    },
    {
      "epoch": 0.38591405184174626,
      "grad_norm": 0.40879642963409424,
      "learning_rate": 1.406534805605622e-06,
      "loss": 0.514,
      "step": 2263
    },
    {
      "epoch": 0.3860845839017735,
      "grad_norm": 0.48264119029045105,
      "learning_rate": 1.4060301251404117e-06,
      "loss": 0.5148,
      "step": 2264
    },
    {
      "epoch": 0.38625511596180084,
      "grad_norm": 0.47972968220710754,
      "learning_rate": 1.4055253208129942e-06,
      "loss": 0.5274,
      "step": 2265
    },
    {
      "epoch": 0.3864256480218281,
      "grad_norm": 0.4307866394519806,
      "learning_rate": 1.4050203927773628e-06,
      "loss": 0.5125,
      "step": 2266
    },
    {
      "epoch": 0.38659618008185537,
      "grad_norm": 0.5022274255752563,
      "learning_rate": 1.4045153411875497e-06,
      "loss": 0.5162,
      "step": 2267
    },
    {
      "epoch": 0.3867667121418827,
      "grad_norm": 0.413129597902298,
      "learning_rate": 1.404010166197624e-06,
      "loss": 0.5293,
      "step": 2268
    },
    {
      "epoch": 0.38693724420190995,
      "grad_norm": 0.4377501606941223,
      "learning_rate": 1.403504867961693e-06,
      "loss": 0.5411,
      "step": 2269
    },
    {
      "epoch": 0.38710777626193726,
      "grad_norm": 0.501990020275116,
      "learning_rate": 1.4029994466339009e-06,
      "loss": 0.5214,
      "step": 2270
    },
    {
      "epoch": 0.3872783083219645,
      "grad_norm": 0.49853450059890747,
      "learning_rate": 1.40249390236843e-06,
      "loss": 0.5194,
      "step": 2271
    },
    {
      "epoch": 0.3874488403819918,
      "grad_norm": 0.46902474761009216,
      "learning_rate": 1.4019882353195003e-06,
      "loss": 0.5256,
      "step": 2272
    },
    {
      "epoch": 0.3876193724420191,
      "grad_norm": 0.46112629771232605,
      "learning_rate": 1.401482445641368e-06,
      "loss": 0.5245,
      "step": 2273
    },
    {
      "epoch": 0.38778990450204637,
      "grad_norm": 0.39838969707489014,
      "learning_rate": 1.4009765334883288e-06,
      "loss": 0.5142,
      "step": 2274
    },
    {
      "epoch": 0.3879604365620737,
      "grad_norm": 0.39665982127189636,
      "learning_rate": 1.4004704990147142e-06,
      "loss": 0.5362,
      "step": 2275
    },
    {
      "epoch": 0.38813096862210095,
      "grad_norm": 0.4456673860549927,
      "learning_rate": 1.3999643423748932e-06,
      "loss": 0.524,
      "step": 2276
    },
    {
      "epoch": 0.38830150068212826,
      "grad_norm": 0.48728182911872864,
      "learning_rate": 1.3994580637232718e-06,
      "loss": 0.5105,
      "step": 2277
    },
    {
      "epoch": 0.3884720327421555,
      "grad_norm": 0.4686523973941803,
      "learning_rate": 1.3989516632142946e-06,
      "loss": 0.529,
      "step": 2278
    },
    {
      "epoch": 0.3886425648021828,
      "grad_norm": 0.41218599677085876,
      "learning_rate": 1.3984451410024422e-06,
      "loss": 0.5275,
      "step": 2279
    },
    {
      "epoch": 0.3888130968622101,
      "grad_norm": 0.6312074661254883,
      "learning_rate": 1.397938497242232e-06,
      "loss": 0.5303,
      "step": 2280
    },
    {
      "epoch": 0.38898362892223737,
      "grad_norm": 0.541376531124115,
      "learning_rate": 1.3974317320882203e-06,
      "loss": 0.5208,
      "step": 2281
    },
    {
      "epoch": 0.3891541609822647,
      "grad_norm": 0.3768637776374817,
      "learning_rate": 1.3969248456949977e-06,
      "loss": 0.538,
      "step": 2282
    },
    {
      "epoch": 0.38932469304229195,
      "grad_norm": 0.5026829838752747,
      "learning_rate": 1.3964178382171943e-06,
      "loss": 0.5389,
      "step": 2283
    },
    {
      "epoch": 0.3894952251023192,
      "grad_norm": 0.4570540487766266,
      "learning_rate": 1.3959107098094758e-06,
      "loss": 0.5123,
      "step": 2284
    },
    {
      "epoch": 0.3896657571623465,
      "grad_norm": 0.4274302124977112,
      "learning_rate": 1.3954034606265453e-06,
      "loss": 0.5098,
      "step": 2285
    },
    {
      "epoch": 0.3898362892223738,
      "grad_norm": 0.4339488744735718,
      "learning_rate": 1.3948960908231424e-06,
      "loss": 0.5349,
      "step": 2286
    },
    {
      "epoch": 0.3900068212824011,
      "grad_norm": 0.3499389588832855,
      "learning_rate": 1.3943886005540437e-06,
      "loss": 0.5277,
      "step": 2287
    },
    {
      "epoch": 0.39017735334242837,
      "grad_norm": 0.3941383361816406,
      "learning_rate": 1.3938809899740623e-06,
      "loss": 0.5294,
      "step": 2288
    },
    {
      "epoch": 0.3903478854024557,
      "grad_norm": 0.39110586047172546,
      "learning_rate": 1.3933732592380486e-06,
      "loss": 0.5139,
      "step": 2289
    },
    {
      "epoch": 0.39051841746248295,
      "grad_norm": 0.3773321211338043,
      "learning_rate": 1.3928654085008888e-06,
      "loss": 0.5284,
      "step": 2290
    },
    {
      "epoch": 0.3906889495225102,
      "grad_norm": 0.4338504374027252,
      "learning_rate": 1.392357437917507e-06,
      "loss": 0.5233,
      "step": 2291
    },
    {
      "epoch": 0.3908594815825375,
      "grad_norm": 0.4193192720413208,
      "learning_rate": 1.391849347642862e-06,
      "loss": 0.5075,
      "step": 2292
    },
    {
      "epoch": 0.3910300136425648,
      "grad_norm": 0.49013859033584595,
      "learning_rate": 1.3913411378319504e-06,
      "loss": 0.5208,
      "step": 2293
    },
    {
      "epoch": 0.3912005457025921,
      "grad_norm": 0.5191242098808289,
      "learning_rate": 1.3908328086398056e-06,
      "loss": 0.5271,
      "step": 2294
    },
    {
      "epoch": 0.39137107776261937,
      "grad_norm": 0.43833374977111816,
      "learning_rate": 1.390324360221496e-06,
      "loss": 0.52,
      "step": 2295
    },
    {
      "epoch": 0.39154160982264663,
      "grad_norm": 0.35007816553115845,
      "learning_rate": 1.389815792732128e-06,
      "loss": 0.5254,
      "step": 2296
    },
    {
      "epoch": 0.39171214188267395,
      "grad_norm": 0.5012977123260498,
      "learning_rate": 1.3893071063268433e-06,
      "loss": 0.5052,
      "step": 2297
    },
    {
      "epoch": 0.3918826739427012,
      "grad_norm": 0.4835526943206787,
      "learning_rate": 1.3887983011608194e-06,
      "loss": 0.5206,
      "step": 2298
    },
    {
      "epoch": 0.3920532060027285,
      "grad_norm": 0.5067761540412903,
      "learning_rate": 1.3882893773892716e-06,
      "loss": 0.5223,
      "step": 2299
    },
    {
      "epoch": 0.3922237380627558,
      "grad_norm": 0.4298695921897888,
      "learning_rate": 1.3877803351674497e-06,
      "loss": 0.515,
      "step": 2300
    },
    {
      "epoch": 0.3923942701227831,
      "grad_norm": 0.6273713707923889,
      "learning_rate": 1.3872711746506415e-06,
      "loss": 0.5162,
      "step": 2301
    },
    {
      "epoch": 0.39256480218281037,
      "grad_norm": 0.6779302954673767,
      "learning_rate": 1.3867618959941687e-06,
      "loss": 0.5213,
      "step": 2302
    },
    {
      "epoch": 0.39273533424283763,
      "grad_norm": 0.3425716757774353,
      "learning_rate": 1.386252499353391e-06,
      "loss": 0.5204,
      "step": 2303
    },
    {
      "epoch": 0.39290586630286495,
      "grad_norm": 0.5272434949874878,
      "learning_rate": 1.3857429848837025e-06,
      "loss": 0.5177,
      "step": 2304
    },
    {
      "epoch": 0.3930763983628922,
      "grad_norm": 0.363675594329834,
      "learning_rate": 1.3852333527405348e-06,
      "loss": 0.5113,
      "step": 2305
    },
    {
      "epoch": 0.3932469304229195,
      "grad_norm": 0.5462455153465271,
      "learning_rate": 1.3847236030793539e-06,
      "loss": 0.5254,
      "step": 2306
    },
    {
      "epoch": 0.3934174624829468,
      "grad_norm": 0.3669333755970001,
      "learning_rate": 1.384213736055663e-06,
      "loss": 0.5248,
      "step": 2307
    },
    {
      "epoch": 0.3935879945429741,
      "grad_norm": 0.46187612414360046,
      "learning_rate": 1.3837037518249999e-06,
      "loss": 0.525,
      "step": 2308
    },
    {
      "epoch": 0.39375852660300137,
      "grad_norm": 0.40820184350013733,
      "learning_rate": 1.3831936505429392e-06,
      "loss": 0.5158,
      "step": 2309
    },
    {
      "epoch": 0.39392905866302863,
      "grad_norm": 0.3934749662876129,
      "learning_rate": 1.38268343236509e-06,
      "loss": 0.5195,
      "step": 2310
    },
    {
      "epoch": 0.39409959072305595,
      "grad_norm": 0.5167784094810486,
      "learning_rate": 1.3821730974470983e-06,
      "loss": 0.5304,
      "step": 2311
    },
    {
      "epoch": 0.3942701227830832,
      "grad_norm": 0.3810725808143616,
      "learning_rate": 1.3816626459446452e-06,
      "loss": 0.5276,
      "step": 2312
    },
    {
      "epoch": 0.3944406548431105,
      "grad_norm": 0.49950870871543884,
      "learning_rate": 1.3811520780134473e-06,
      "loss": 0.5093,
      "step": 2313
    },
    {
      "epoch": 0.3946111869031378,
      "grad_norm": 0.4600581228733063,
      "learning_rate": 1.3806413938092568e-06,
      "loss": 0.5229,
      "step": 2314
    },
    {
      "epoch": 0.39478171896316505,
      "grad_norm": 0.3954922556877136,
      "learning_rate": 1.380130593487861e-06,
      "loss": 0.5319,
      "step": 2315
    },
    {
      "epoch": 0.39495225102319237,
      "grad_norm": 0.43245062232017517,
      "learning_rate": 1.379619677205083e-06,
      "loss": 0.5283,
      "step": 2316
    },
    {
      "epoch": 0.39512278308321963,
      "grad_norm": 0.43941789865493774,
      "learning_rate": 1.3791086451167815e-06,
      "loss": 0.5174,
      "step": 2317
    },
    {
      "epoch": 0.39529331514324695,
      "grad_norm": 0.4048894941806793,
      "learning_rate": 1.3785974973788507e-06,
      "loss": 0.507,
      "step": 2318
    },
    {
      "epoch": 0.3954638472032742,
      "grad_norm": 0.36088645458221436,
      "learning_rate": 1.3780862341472184e-06,
      "loss": 0.5296,
      "step": 2319
    },
    {
      "epoch": 0.3956343792633015,
      "grad_norm": 0.5440859198570251,
      "learning_rate": 1.3775748555778498e-06,
      "loss": 0.5291,
      "step": 2320
    },
    {
      "epoch": 0.3958049113233288,
      "grad_norm": 0.4294169843196869,
      "learning_rate": 1.3770633618267445e-06,
      "loss": 0.5162,
      "step": 2321
    },
    {
      "epoch": 0.39597544338335605,
      "grad_norm": 0.4461040496826172,
      "learning_rate": 1.3765517530499365e-06,
      "loss": 0.5187,
      "step": 2322
    },
    {
      "epoch": 0.39614597544338337,
      "grad_norm": 0.4923355281352997,
      "learning_rate": 1.3760400294034957e-06,
      "loss": 0.5143,
      "step": 2323
    },
    {
      "epoch": 0.39631650750341063,
      "grad_norm": 0.3550982177257538,
      "learning_rate": 1.3755281910435268e-06,
      "loss": 0.5266,
      "step": 2324
    },
    {
      "epoch": 0.39648703956343795,
      "grad_norm": 0.44544970989227295,
      "learning_rate": 1.3750162381261694e-06,
      "loss": 0.5328,
      "step": 2325
    },
    {
      "epoch": 0.3966575716234652,
      "grad_norm": 0.47301942110061646,
      "learning_rate": 1.3745041708075985e-06,
      "loss": 0.5281,
      "step": 2326
    },
    {
      "epoch": 0.3968281036834925,
      "grad_norm": 0.4111325740814209,
      "learning_rate": 1.3739919892440236e-06,
      "loss": 0.5214,
      "step": 2327
    },
    {
      "epoch": 0.3969986357435198,
      "grad_norm": 0.3858391344547272,
      "learning_rate": 1.373479693591689e-06,
      "loss": 0.5268,
      "step": 2328
    },
    {
      "epoch": 0.39716916780354705,
      "grad_norm": 0.5584406852722168,
      "learning_rate": 1.3729672840068734e-06,
      "loss": 0.5264,
      "step": 2329
    },
    {
      "epoch": 0.39733969986357437,
      "grad_norm": 0.5933360457420349,
      "learning_rate": 1.3724547606458918e-06,
      "loss": 0.5241,
      "step": 2330
    },
    {
      "epoch": 0.39751023192360163,
      "grad_norm": 0.6099199056625366,
      "learning_rate": 1.3719421236650923e-06,
      "loss": 0.5174,
      "step": 2331
    },
    {
      "epoch": 0.39768076398362895,
      "grad_norm": 0.48414215445518494,
      "learning_rate": 1.3714293732208584e-06,
      "loss": 0.5181,
      "step": 2332
    },
    {
      "epoch": 0.3978512960436562,
      "grad_norm": 0.5974043011665344,
      "learning_rate": 1.3709165094696077e-06,
      "loss": 0.5216,
      "step": 2333
    },
    {
      "epoch": 0.3980218281036835,
      "grad_norm": 0.4411584734916687,
      "learning_rate": 1.3704035325677932e-06,
      "loss": 0.5177,
      "step": 2334
    },
    {
      "epoch": 0.3981923601637108,
      "grad_norm": 0.49411454796791077,
      "learning_rate": 1.3698904426719015e-06,
      "loss": 0.5123,
      "step": 2335
    },
    {
      "epoch": 0.39836289222373805,
      "grad_norm": 0.3845353126525879,
      "learning_rate": 1.3693772399384548e-06,
      "loss": 0.5159,
      "step": 2336
    },
    {
      "epoch": 0.39853342428376537,
      "grad_norm": 0.47084566950798035,
      "learning_rate": 1.368863924524008e-06,
      "loss": 0.5309,
      "step": 2337
    },
    {
      "epoch": 0.39870395634379263,
      "grad_norm": 0.39838436245918274,
      "learning_rate": 1.3683504965851523e-06,
      "loss": 0.5172,
      "step": 2338
    },
    {
      "epoch": 0.3988744884038199,
      "grad_norm": 0.39336255192756653,
      "learning_rate": 1.367836956278512e-06,
      "loss": 0.5219,
      "step": 2339
    },
    {
      "epoch": 0.3990450204638472,
      "grad_norm": 0.4916491210460663,
      "learning_rate": 1.3673233037607459e-06,
      "loss": 0.5018,
      "step": 2340
    },
    {
      "epoch": 0.3992155525238745,
      "grad_norm": 0.3593173325061798,
      "learning_rate": 1.3668095391885474e-06,
      "loss": 0.5197,
      "step": 2341
    },
    {
      "epoch": 0.3993860845839018,
      "grad_norm": 0.3894543945789337,
      "learning_rate": 1.3662956627186434e-06,
      "loss": 0.5207,
      "step": 2342
    },
    {
      "epoch": 0.39955661664392905,
      "grad_norm": 0.40112966299057007,
      "learning_rate": 1.3657816745077958e-06,
      "loss": 0.5118,
      "step": 2343
    },
    {
      "epoch": 0.39972714870395637,
      "grad_norm": 0.41793084144592285,
      "learning_rate": 1.3652675747127996e-06,
      "loss": 0.5173,
      "step": 2344
    },
    {
      "epoch": 0.39989768076398363,
      "grad_norm": 0.4209323227405548,
      "learning_rate": 1.3647533634904852e-06,
      "loss": 0.5166,
      "step": 2345
    },
    {
      "epoch": 0.4000682128240109,
      "grad_norm": 0.3716372549533844,
      "learning_rate": 1.3642390409977156e-06,
      "loss": 0.5062,
      "step": 2346
    },
    {
      "epoch": 0.4002387448840382,
      "grad_norm": 0.3547699749469757,
      "learning_rate": 1.3637246073913885e-06,
      "loss": 0.5066,
      "step": 2347
    },
    {
      "epoch": 0.4004092769440655,
      "grad_norm": 0.3524262607097626,
      "learning_rate": 1.363210062828435e-06,
      "loss": 0.5227,
      "step": 2348
    },
    {
      "epoch": 0.4005798090040928,
      "grad_norm": 0.3743390440940857,
      "learning_rate": 1.3626954074658212e-06,
      "loss": 0.5013,
      "step": 2349
    },
    {
      "epoch": 0.40075034106412005,
      "grad_norm": 0.41511663794517517,
      "learning_rate": 1.3621806414605448e-06,
      "loss": 0.5232,
      "step": 2350
    },
    {
      "epoch": 0.4009208731241473,
      "grad_norm": 0.4181976616382599,
      "learning_rate": 1.3616657649696398e-06,
      "loss": 0.513,
      "step": 2351
    },
    {
      "epoch": 0.40109140518417463,
      "grad_norm": 0.44195061922073364,
      "learning_rate": 1.3611507781501723e-06,
      "loss": 0.5193,
      "step": 2352
    },
    {
      "epoch": 0.4012619372442019,
      "grad_norm": 0.40733179450035095,
      "learning_rate": 1.360635681159243e-06,
      "loss": 0.5096,
      "step": 2353
    },
    {
      "epoch": 0.4014324693042292,
      "grad_norm": 0.49988704919815063,
      "learning_rate": 1.3601204741539843e-06,
      "loss": 0.511,
      "step": 2354
    },
    {
      "epoch": 0.4016030013642565,
      "grad_norm": 0.43027475476264954,
      "learning_rate": 1.359605157291565e-06,
      "loss": 0.5181,
      "step": 2355
    },
    {
      "epoch": 0.4017735334242838,
      "grad_norm": 0.4969724714756012,
      "learning_rate": 1.359089730729185e-06,
      "loss": 0.5138,
      "step": 2356
    },
    {
      "epoch": 0.40194406548431105,
      "grad_norm": 0.7880139946937561,
      "learning_rate": 1.3585741946240793e-06,
      "loss": 0.5093,
      "step": 2357
    },
    {
      "epoch": 0.4021145975443383,
      "grad_norm": 0.47539666295051575,
      "learning_rate": 1.3580585491335153e-06,
      "loss": 0.5032,
      "step": 2358
    },
    {
      "epoch": 0.40228512960436563,
      "grad_norm": 0.4887872040271759,
      "learning_rate": 1.3575427944147939e-06,
      "loss": 0.5026,
      "step": 2359
    },
    {
      "epoch": 0.4024556616643929,
      "grad_norm": 0.4302110970020294,
      "learning_rate": 1.35702693062525e-06,
      "loss": 0.5176,
      "step": 2360
    },
    {
      "epoch": 0.4026261937244202,
      "grad_norm": 0.44501402974128723,
      "learning_rate": 1.3565109579222513e-06,
      "loss": 0.5192,
      "step": 2361
    },
    {
      "epoch": 0.4027967257844475,
      "grad_norm": 0.5135705471038818,
      "learning_rate": 1.3559948764631983e-06,
      "loss": 0.53,
      "step": 2362
    },
    {
      "epoch": 0.40296725784447474,
      "grad_norm": 0.35561659932136536,
      "learning_rate": 1.3554786864055252e-06,
      "loss": 0.5132,
      "step": 2363
    },
    {
      "epoch": 0.40313778990450205,
      "grad_norm": 0.532179594039917,
      "learning_rate": 1.3549623879066996e-06,
      "loss": 0.5234,
      "step": 2364
    },
    {
      "epoch": 0.4033083219645293,
      "grad_norm": 0.5692497491836548,
      "learning_rate": 1.3544459811242213e-06,
      "loss": 0.5352,
      "step": 2365
    },
    {
      "epoch": 0.40347885402455663,
      "grad_norm": 0.3840283453464508,
      "learning_rate": 1.3539294662156241e-06,
      "loss": 0.5233,
      "step": 2366
    },
    {
      "epoch": 0.4036493860845839,
      "grad_norm": 0.4098012447357178,
      "learning_rate": 1.3534128433384743e-06,
      "loss": 0.5163,
      "step": 2367
    },
    {
      "epoch": 0.4038199181446112,
      "grad_norm": 0.4464457035064697,
      "learning_rate": 1.3528961126503704e-06,
      "loss": 0.5119,
      "step": 2368
    },
    {
      "epoch": 0.4039904502046385,
      "grad_norm": 0.4604474902153015,
      "learning_rate": 1.3523792743089457e-06,
      "loss": 0.544,
      "step": 2369
    },
    {
      "epoch": 0.40416098226466574,
      "grad_norm": 0.36774614453315735,
      "learning_rate": 1.3518623284718645e-06,
      "loss": 0.5302,
      "step": 2370
    },
    {
      "epoch": 0.40433151432469305,
      "grad_norm": 0.4567425549030304,
      "learning_rate": 1.351345275296825e-06,
      "loss": 0.5336,
      "step": 2371
    },
    {
      "epoch": 0.4045020463847203,
      "grad_norm": 0.42046618461608887,
      "learning_rate": 1.3508281149415571e-06,
      "loss": 0.5257,
      "step": 2372
    },
    {
      "epoch": 0.40467257844474763,
      "grad_norm": 0.5821329951286316,
      "learning_rate": 1.3503108475638246e-06,
      "loss": 0.5263,
      "step": 2373
    },
    {
      "epoch": 0.4048431105047749,
      "grad_norm": 0.40758076310157776,
      "learning_rate": 1.3497934733214233e-06,
      "loss": 0.5071,
      "step": 2374
    },
    {
      "epoch": 0.40501364256480216,
      "grad_norm": 0.39630794525146484,
      "learning_rate": 1.349275992372181e-06,
      "loss": 0.5252,
      "step": 2375
    },
    {
      "epoch": 0.4051841746248295,
      "grad_norm": 0.4020107388496399,
      "learning_rate": 1.34875840487396e-06,
      "loss": 0.5133,
      "step": 2376
    },
    {
      "epoch": 0.40535470668485674,
      "grad_norm": 0.39645326137542725,
      "learning_rate": 1.3482407109846529e-06,
      "loss": 0.52,
      "step": 2377
    },
    {
      "epoch": 0.40552523874488405,
      "grad_norm": 0.4198255240917206,
      "learning_rate": 1.3477229108621856e-06,
      "loss": 0.5201,
      "step": 2378
    },
    {
      "epoch": 0.4056957708049113,
      "grad_norm": 0.3982833921909332,
      "learning_rate": 1.3472050046645168e-06,
      "loss": 0.5325,
      "step": 2379
    },
    {
      "epoch": 0.40586630286493863,
      "grad_norm": 0.43688592314720154,
      "learning_rate": 1.3466869925496373e-06,
      "loss": 0.4985,
      "step": 2380
    },
    {
      "epoch": 0.4060368349249659,
      "grad_norm": 0.4188794791698456,
      "learning_rate": 1.3461688746755705e-06,
      "loss": 0.5165,
      "step": 2381
    },
    {
      "epoch": 0.40620736698499316,
      "grad_norm": 0.40829628705978394,
      "learning_rate": 1.3456506512003708e-06,
      "loss": 0.5237,
      "step": 2382
    },
    {
      "epoch": 0.4063778990450205,
      "grad_norm": 0.3171619176864624,
      "learning_rate": 1.3451323222821268e-06,
      "loss": 0.5105,
      "step": 2383
    },
    {
      "epoch": 0.40654843110504774,
      "grad_norm": 0.4364809989929199,
      "learning_rate": 1.3446138880789572e-06,
      "loss": 0.5233,
      "step": 2384
    },
    {
      "epoch": 0.40671896316507505,
      "grad_norm": 0.3572435677051544,
      "learning_rate": 1.3440953487490145e-06,
      "loss": 0.518,
      "step": 2385
    },
    {
      "epoch": 0.4068894952251023,
      "grad_norm": 0.4391252100467682,
      "learning_rate": 1.3435767044504827e-06,
      "loss": 0.5117,
      "step": 2386
    },
    {
      "epoch": 0.4070600272851296,
      "grad_norm": 0.4334906339645386,
      "learning_rate": 1.3430579553415772e-06,
      "loss": 0.5136,
      "step": 2387
    },
    {
      "epoch": 0.4072305593451569,
      "grad_norm": 0.3785998821258545,
      "learning_rate": 1.342539101580546e-06,
      "loss": 0.5311,
      "step": 2388
    },
    {
      "epoch": 0.40740109140518416,
      "grad_norm": 0.40511175990104675,
      "learning_rate": 1.342020143325669e-06,
      "loss": 0.5063,
      "step": 2389
    },
    {
      "epoch": 0.4075716234652115,
      "grad_norm": 0.42973968386650085,
      "learning_rate": 1.341501080735258e-06,
      "loss": 0.5109,
      "step": 2390
    },
    {
      "epoch": 0.40774215552523874,
      "grad_norm": 0.38002124428749084,
      "learning_rate": 1.3409819139676562e-06,
      "loss": 0.5143,
      "step": 2391
    },
    {
      "epoch": 0.40791268758526605,
      "grad_norm": 0.4136904776096344,
      "learning_rate": 1.3404626431812395e-06,
      "loss": 0.5093,
      "step": 2392
    },
    {
      "epoch": 0.4080832196452933,
      "grad_norm": 0.3670330345630646,
      "learning_rate": 1.3399432685344144e-06,
      "loss": 0.5153,
      "step": 2393
    },
    {
      "epoch": 0.4082537517053206,
      "grad_norm": 0.3782718777656555,
      "learning_rate": 1.3394237901856196e-06,
      "loss": 0.5128,
      "step": 2394
    },
    {
      "epoch": 0.4084242837653479,
      "grad_norm": 0.39012637734413147,
      "learning_rate": 1.3389042082933258e-06,
      "loss": 0.5225,
      "step": 2395
    },
    {
      "epoch": 0.40859481582537516,
      "grad_norm": 0.38971012830734253,
      "learning_rate": 1.3383845230160348e-06,
      "loss": 0.5148,
      "step": 2396
    },
    {
      "epoch": 0.4087653478854025,
      "grad_norm": 0.3932477831840515,
      "learning_rate": 1.3378647345122797e-06,
      "loss": 0.5271,
      "step": 2397
    },
    {
      "epoch": 0.40893587994542974,
      "grad_norm": 0.37541669607162476,
      "learning_rate": 1.337344842940626e-06,
      "loss": 0.5174,
      "step": 2398
    },
    {
      "epoch": 0.409106412005457,
      "grad_norm": 0.43585965037345886,
      "learning_rate": 1.33682484845967e-06,
      "loss": 0.5162,
      "step": 2399
    },
    {
      "epoch": 0.4092769440654843,
      "grad_norm": 0.6322606205940247,
      "learning_rate": 1.3363047512280393e-06,
      "loss": 0.5239,
      "step": 2400
    },
    {
      "epoch": 0.4094474761255116,
      "grad_norm": 0.636260986328125,
      "learning_rate": 1.335784551404393e-06,
      "loss": 0.5231,
      "step": 2401
    },
    {
      "epoch": 0.4096180081855389,
      "grad_norm": 0.34062281250953674,
      "learning_rate": 1.3352642491474216e-06,
      "loss": 0.5235,
      "step": 2402
    },
    {
      "epoch": 0.40978854024556616,
      "grad_norm": 0.5656194090843201,
      "learning_rate": 1.3347438446158469e-06,
      "loss": 0.5166,
      "step": 2403
    },
    {
      "epoch": 0.4099590723055935,
      "grad_norm": 0.758741557598114,
      "learning_rate": 1.3342233379684218e-06,
      "loss": 0.5187,
      "step": 2404
    },
    {
      "epoch": 0.41012960436562074,
      "grad_norm": 0.4431105852127075,
      "learning_rate": 1.3337027293639303e-06,
      "loss": 0.5174,
      "step": 2405
    },
    {
      "epoch": 0.410300136425648,
      "grad_norm": 0.5212686061859131,
      "learning_rate": 1.3331820189611874e-06,
      "loss": 0.5238,
      "step": 2406
    },
    {
      "epoch": 0.4104706684856753,
      "grad_norm": 0.6115703582763672,
      "learning_rate": 1.3326612069190395e-06,
      "loss": 0.5142,
      "step": 2407
    },
    {
      "epoch": 0.4106412005457026,
      "grad_norm": 0.5647085309028625,
      "learning_rate": 1.3321402933963638e-06,
      "loss": 0.5204,
      "step": 2408
    },
    {
      "epoch": 0.4108117326057299,
      "grad_norm": 0.5248672962188721,
      "learning_rate": 1.3316192785520683e-06,
      "loss": 0.5131,
      "step": 2409
    },
    {
      "epoch": 0.41098226466575716,
      "grad_norm": 0.41708463430404663,
      "learning_rate": 1.3310981625450917e-06,
      "loss": 0.5285,
      "step": 2410
    },
    {
      "epoch": 0.4111527967257845,
      "grad_norm": 0.5683760643005371,
      "learning_rate": 1.330576945534405e-06,
      "loss": 0.5241,
      "step": 2411
    },
    {
      "epoch": 0.41132332878581174,
      "grad_norm": 0.38032862544059753,
      "learning_rate": 1.330055627679008e-06,
      "loss": 0.5105,
      "step": 2412
    },
    {
      "epoch": 0.411493860845839,
      "grad_norm": 0.42650604248046875,
      "learning_rate": 1.3295342091379324e-06,
      "loss": 0.5221,
      "step": 2413
    },
    {
      "epoch": 0.4116643929058663,
      "grad_norm": 0.5264064073562622,
      "learning_rate": 1.3290126900702404e-06,
      "loss": 0.5186,
      "step": 2414
    },
    {
      "epoch": 0.4118349249658936,
      "grad_norm": 0.42799603939056396,
      "learning_rate": 1.3284910706350251e-06,
      "loss": 0.5217,
      "step": 2415
    },
    {
      "epoch": 0.4120054570259209,
      "grad_norm": 0.4704800248146057,
      "learning_rate": 1.3279693509914102e-06,
      "loss": 0.5118,
      "step": 2416
    },
    {
      "epoch": 0.41217598908594816,
      "grad_norm": 0.41373327374458313,
      "learning_rate": 1.3274475312985492e-06,
      "loss": 0.5147,
      "step": 2417
    },
    {
      "epoch": 0.4123465211459754,
      "grad_norm": 0.44518589973449707,
      "learning_rate": 1.326925611715627e-06,
      "loss": 0.5247,
      "step": 2418
    },
    {
      "epoch": 0.41251705320600274,
      "grad_norm": 0.6880121231079102,
      "learning_rate": 1.3264035924018585e-06,
      "loss": 0.5348,
      "step": 2419
    },
    {
      "epoch": 0.41268758526603,
      "grad_norm": 0.5007564425468445,
      "learning_rate": 1.3258814735164893e-06,
      "loss": 0.5354,
      "step": 2420
    },
    {
      "epoch": 0.4128581173260573,
      "grad_norm": 0.481301486492157,
      "learning_rate": 1.3253592552187951e-06,
      "loss": 0.5053,
      "step": 2421
    },
    {
      "epoch": 0.4130286493860846,
      "grad_norm": 0.46144989132881165,
      "learning_rate": 1.3248369376680825e-06,
      "loss": 0.5065,
      "step": 2422
    },
    {
      "epoch": 0.4131991814461119,
      "grad_norm": 0.5711754560470581,
      "learning_rate": 1.3243145210236877e-06,
      "loss": 0.5348,
      "step": 2423
    },
    {
      "epoch": 0.41336971350613916,
      "grad_norm": 0.6206766963005066,
      "learning_rate": 1.3237920054449776e-06,
      "loss": 0.5256,
      "step": 2424
    },
    {
      "epoch": 0.4135402455661664,
      "grad_norm": 0.43933790922164917,
      "learning_rate": 1.3232693910913486e-06,
      "loss": 0.522,
      "step": 2425
    },
    {
      "epoch": 0.41371077762619374,
      "grad_norm": 0.4370189905166626,
      "learning_rate": 1.322746678122228e-06,
      "loss": 0.5255,
      "step": 2426
    },
    {
      "epoch": 0.413881309686221,
      "grad_norm": 0.5346240401268005,
      "learning_rate": 1.322223866697073e-06,
      "loss": 0.5282,
      "step": 2427
    },
    {
      "epoch": 0.4140518417462483,
      "grad_norm": 0.46231934428215027,
      "learning_rate": 1.3217009569753704e-06,
      "loss": 0.5201,
      "step": 2428
    },
    {
      "epoch": 0.4142223738062756,
      "grad_norm": 0.40670233964920044,
      "learning_rate": 1.3211779491166376e-06,
      "loss": 0.5192,
      "step": 2429
    },
    {
      "epoch": 0.41439290586630284,
      "grad_norm": 0.42665842175483704,
      "learning_rate": 1.3206548432804216e-06,
      "loss": 0.5177,
      "step": 2430
    },
    {
      "epoch": 0.41456343792633016,
      "grad_norm": 0.5745131373405457,
      "learning_rate": 1.3201316396262993e-06,
      "loss": 0.5196,
      "step": 2431
    },
    {
      "epoch": 0.4147339699863574,
      "grad_norm": 0.48258480429649353,
      "learning_rate": 1.3196083383138776e-06,
      "loss": 0.5214,
      "step": 2432
    },
    {
      "epoch": 0.41490450204638474,
      "grad_norm": 0.4157755672931671,
      "learning_rate": 1.3190849395027927e-06,
      "loss": 0.5276,
      "step": 2433
    },
    {
      "epoch": 0.415075034106412,
      "grad_norm": 0.5874139070510864,
      "learning_rate": 1.318561443352711e-06,
      "loss": 0.5085,
      "step": 2434
    },
    {
      "epoch": 0.4152455661664393,
      "grad_norm": 0.44920822978019714,
      "learning_rate": 1.3180378500233292e-06,
      "loss": 0.5122,
      "step": 2435
    },
    {
      "epoch": 0.4154160982264666,
      "grad_norm": 0.42331570386886597,
      "learning_rate": 1.317514159674372e-06,
      "loss": 0.5165,
      "step": 2436
    },
    {
      "epoch": 0.41558663028649384,
      "grad_norm": 0.5718460083007812,
      "learning_rate": 1.3169903724655952e-06,
      "loss": 0.523,
      "step": 2437
    },
    {
      "epoch": 0.41575716234652116,
      "grad_norm": 0.4036332964897156,
      "learning_rate": 1.3164664885567833e-06,
      "loss": 0.5158,
      "step": 2438
    },
    {
      "epoch": 0.4159276944065484,
      "grad_norm": 0.5522937178611755,
      "learning_rate": 1.315942508107751e-06,
      "loss": 0.5006,
      "step": 2439
    },
    {
      "epoch": 0.41609822646657574,
      "grad_norm": 0.46217483282089233,
      "learning_rate": 1.3154184312783418e-06,
      "loss": 0.5188,
      "step": 2440
    },
    {
      "epoch": 0.416268758526603,
      "grad_norm": 0.3863103985786438,
      "learning_rate": 1.3148942582284287e-06,
      "loss": 0.5265,
      "step": 2441
    },
    {
      "epoch": 0.41643929058663026,
      "grad_norm": 0.45400503277778625,
      "learning_rate": 1.3143699891179141e-06,
      "loss": 0.5169,
      "step": 2442
    },
    {
      "epoch": 0.4166098226466576,
      "grad_norm": 0.3837527632713318,
      "learning_rate": 1.3138456241067303e-06,
      "loss": 0.5172,
      "step": 2443
    },
    {
      "epoch": 0.41678035470668484,
      "grad_norm": 0.45465466380119324,
      "learning_rate": 1.313321163354838e-06,
      "loss": 0.5099,
      "step": 2444
    },
    {
      "epoch": 0.41695088676671216,
      "grad_norm": 0.3569267988204956,
      "learning_rate": 1.3127966070222274e-06,
      "loss": 0.5144,
      "step": 2445
    },
    {
      "epoch": 0.4171214188267394,
      "grad_norm": 0.40313395857810974,
      "learning_rate": 1.312271955268918e-06,
      "loss": 0.5311,
      "step": 2446
    },
    {
      "epoch": 0.41729195088676674,
      "grad_norm": 0.4062875509262085,
      "learning_rate": 1.3117472082549588e-06,
      "loss": 0.508,
      "step": 2447
    },
    {
      "epoch": 0.417462482946794,
      "grad_norm": 0.4219088852405548,
      "learning_rate": 1.311222366140427e-06,
      "loss": 0.5047,
      "step": 2448
    },
    {
      "epoch": 0.41763301500682126,
      "grad_norm": 0.3890495002269745,
      "learning_rate": 1.3106974290854287e-06,
      "loss": 0.5218,
      "step": 2449
    },
    {
      "epoch": 0.4178035470668486,
      "grad_norm": 0.36203303933143616,
      "learning_rate": 1.3101723972501003e-06,
      "loss": 0.516,
      "step": 2450
    },
    {
      "epoch": 0.41797407912687584,
      "grad_norm": 0.343011736869812,
      "learning_rate": 1.309647270794606e-06,
      "loss": 0.5083,
      "step": 2451
    },
    {
      "epoch": 0.41814461118690316,
      "grad_norm": 0.3661821782588959,
      "learning_rate": 1.3091220498791388e-06,
      "loss": 0.5202,
      "step": 2452
    },
    {
      "epoch": 0.4183151432469304,
      "grad_norm": 0.37110432982444763,
      "learning_rate": 1.3085967346639214e-06,
      "loss": 0.522,
      "step": 2453
    },
    {
      "epoch": 0.4184856753069577,
      "grad_norm": 0.4287030100822449,
      "learning_rate": 1.308071325309204e-06,
      "loss": 0.5193,
      "step": 2454
    },
    {
      "epoch": 0.418656207366985,
      "grad_norm": 0.4393904507160187,
      "learning_rate": 1.3075458219752672e-06,
      "loss": 0.5035,
      "step": 2455
    },
    {
      "epoch": 0.41882673942701226,
      "grad_norm": 0.33759546279907227,
      "learning_rate": 1.3070202248224182e-06,
      "loss": 0.5124,
      "step": 2456
    },
    {
      "epoch": 0.4189972714870396,
      "grad_norm": 0.4863134026527405,
      "learning_rate": 1.306494534010995e-06,
      "loss": 0.5139,
      "step": 2457
    },
    {
      "epoch": 0.41916780354706684,
      "grad_norm": 0.43133077025413513,
      "learning_rate": 1.3059687497013627e-06,
      "loss": 0.5197,
      "step": 2458
    },
    {
      "epoch": 0.41933833560709416,
      "grad_norm": 0.3589963912963867,
      "learning_rate": 1.3054428720539148e-06,
      "loss": 0.5283,
      "step": 2459
    },
    {
      "epoch": 0.4195088676671214,
      "grad_norm": 0.3672996461391449,
      "learning_rate": 1.3049169012290748e-06,
      "loss": 0.5174,
      "step": 2460
    },
    {
      "epoch": 0.4196793997271487,
      "grad_norm": 0.39421847462654114,
      "learning_rate": 1.3043908373872923e-06,
      "loss": 0.5183,
      "step": 2461
    },
    {
      "epoch": 0.419849931787176,
      "grad_norm": 0.38718149065971375,
      "learning_rate": 1.3038646806890479e-06,
      "loss": 0.5291,
      "step": 2462
    },
    {
      "epoch": 0.42002046384720326,
      "grad_norm": 0.3574490249156952,
      "learning_rate": 1.3033384312948488e-06,
      "loss": 0.5189,
      "step": 2463
    },
    {
      "epoch": 0.4201909959072306,
      "grad_norm": 0.38356050848960876,
      "learning_rate": 1.3028120893652306e-06,
      "loss": 0.51,
      "step": 2464
    },
    {
      "epoch": 0.42036152796725784,
      "grad_norm": 0.3885401487350464,
      "learning_rate": 1.3022856550607574e-06,
      "loss": 0.5261,
      "step": 2465
    },
    {
      "epoch": 0.4205320600272851,
      "grad_norm": 0.37062355875968933,
      "learning_rate": 1.301759128542022e-06,
      "loss": 0.5205,
      "step": 2466
    },
    {
      "epoch": 0.4207025920873124,
      "grad_norm": 0.4637162685394287,
      "learning_rate": 1.3012325099696444e-06,
      "loss": 0.5163,
      "step": 2467
    },
    {
      "epoch": 0.4208731241473397,
      "grad_norm": 0.4647408127784729,
      "learning_rate": 1.3007057995042733e-06,
      "loss": 0.5172,
      "step": 2468
    },
    {
      "epoch": 0.421043656207367,
      "grad_norm": 0.3651617765426636,
      "learning_rate": 1.3001789973065853e-06,
      "loss": 0.5168,
      "step": 2469
    },
    {
      "epoch": 0.42121418826739426,
      "grad_norm": 0.3944820165634155,
      "learning_rate": 1.2996521035372847e-06,
      "loss": 0.5259,
      "step": 2470
    },
    {
      "epoch": 0.4213847203274216,
      "grad_norm": 0.5131377577781677,
      "learning_rate": 1.2991251183571045e-06,
      "loss": 0.5259,
      "step": 2471
    },
    {
      "epoch": 0.42155525238744884,
      "grad_norm": 0.6908408999443054,
      "learning_rate": 1.2985980419268045e-06,
      "loss": 0.5264,
      "step": 2472
    },
    {
      "epoch": 0.4217257844474761,
      "grad_norm": 0.5145502090454102,
      "learning_rate": 1.298070874407173e-06,
      "loss": 0.5227,
      "step": 2473
    },
    {
      "epoch": 0.4218963165075034,
      "grad_norm": 0.42013034224510193,
      "learning_rate": 1.2975436159590262e-06,
      "loss": 0.5315,
      "step": 2474
    },
    {
      "epoch": 0.4220668485675307,
      "grad_norm": 0.6404533982276917,
      "learning_rate": 1.2970162667432076e-06,
      "loss": 0.5218,
      "step": 2475
    },
    {
      "epoch": 0.422237380627558,
      "grad_norm": 0.7172034382820129,
      "learning_rate": 1.2964888269205888e-06,
      "loss": 0.5202,
      "step": 2476
    },
    {
      "epoch": 0.42240791268758526,
      "grad_norm": 0.4771793484687805,
      "learning_rate": 1.2959612966520684e-06,
      "loss": 0.5105,
      "step": 2477
    },
    {
      "epoch": 0.4225784447476125,
      "grad_norm": 0.39662736654281616,
      "learning_rate": 1.2954336760985738e-06,
      "loss": 0.5223,
      "step": 2478
    },
    {
      "epoch": 0.42274897680763984,
      "grad_norm": 0.6462126970291138,
      "learning_rate": 1.2949059654210588e-06,
      "loss": 0.5211,
      "step": 2479
    },
    {
      "epoch": 0.4229195088676671,
      "grad_norm": 0.4810219705104828,
      "learning_rate": 1.294378164780505e-06,
      "loss": 0.5328,
      "step": 2480
    },
    {
      "epoch": 0.4230900409276944,
      "grad_norm": 0.5060651302337646,
      "learning_rate": 1.2938502743379213e-06,
      "loss": 0.5133,
      "step": 2481
    },
    {
      "epoch": 0.4232605729877217,
      "grad_norm": 0.6474683880805969,
      "learning_rate": 1.2933222942543445e-06,
      "loss": 0.516,
      "step": 2482
    },
    {
      "epoch": 0.423431105047749,
      "grad_norm": 0.380156546831131,
      "learning_rate": 1.2927942246908383e-06,
      "loss": 0.5237,
      "step": 2483
    },
    {
      "epoch": 0.42360163710777626,
      "grad_norm": 0.5847602486610413,
      "learning_rate": 1.2922660658084946e-06,
      "loss": 0.5206,
      "step": 2484
    },
    {
      "epoch": 0.4237721691678035,
      "grad_norm": 0.5852933526039124,
      "learning_rate": 1.2917378177684305e-06,
      "loss": 0.5333,
      "step": 2485
    },
    {
      "epoch": 0.42394270122783084,
      "grad_norm": 0.3461321294307709,
      "learning_rate": 1.291209480731793e-06,
      "loss": 0.4982,
      "step": 2486
    },
    {
      "epoch": 0.4241132332878581,
      "grad_norm": 0.5730729103088379,
      "learning_rate": 1.2906810548597533e-06,
      "loss": 0.5151,
      "step": 2487
    },
    {
      "epoch": 0.4242837653478854,
      "grad_norm": 0.4181745946407318,
      "learning_rate": 1.2901525403135127e-06,
      "loss": 0.5277,
      "step": 2488
    },
    {
      "epoch": 0.4244542974079127,
      "grad_norm": 0.4739188849925995,
      "learning_rate": 1.2896239372542973e-06,
      "loss": 0.5342,
      "step": 2489
    },
    {
      "epoch": 0.42462482946793995,
      "grad_norm": 0.5515470504760742,
      "learning_rate": 1.2890952458433611e-06,
      "loss": 0.5276,
      "step": 2490
    },
    {
      "epoch": 0.42479536152796726,
      "grad_norm": 0.3359604775905609,
      "learning_rate": 1.2885664662419852e-06,
      "loss": 0.5377,
      "step": 2491
    },
    {
      "epoch": 0.4249658935879945,
      "grad_norm": 0.5798554420471191,
      "learning_rate": 1.2880375986114771e-06,
      "loss": 0.5217,
      "step": 2492
    },
    {
      "epoch": 0.42513642564802184,
      "grad_norm": 0.4647768437862396,
      "learning_rate": 1.2875086431131717e-06,
      "loss": 0.5191,
      "step": 2493
    },
    {
      "epoch": 0.4253069577080491,
      "grad_norm": 0.38144662976264954,
      "learning_rate": 1.28697959990843e-06,
      "loss": 0.5183,
      "step": 2494
    },
    {
      "epoch": 0.4254774897680764,
      "grad_norm": 0.504442572593689,
      "learning_rate": 1.2864504691586405e-06,
      "loss": 0.5129,
      "step": 2495
    },
    {
      "epoch": 0.4256480218281037,
      "grad_norm": 0.3849034011363983,
      "learning_rate": 1.2859212510252178e-06,
      "loss": 0.5212,
      "step": 2496
    },
    {
      "epoch": 0.42581855388813095,
      "grad_norm": 0.3967511057853699,
      "learning_rate": 1.2853919456696038e-06,
      "loss": 0.5092,
      "step": 2497
    },
    {
      "epoch": 0.42598908594815826,
      "grad_norm": 0.4667202830314636,
      "learning_rate": 1.2848625532532664e-06,
      "loss": 0.5361,
      "step": 2498
    },
    {
      "epoch": 0.4261596180081855,
      "grad_norm": 0.4540242850780487,
      "learning_rate": 1.2843330739377004e-06,
      "loss": 0.5316,
      "step": 2499
    },
    {
      "epoch": 0.42633015006821284,
      "grad_norm": 0.4949699342250824,
      "learning_rate": 1.283803507884427e-06,
      "loss": 0.5095,
      "step": 2500
    },
    {
      "epoch": 0.4265006821282401,
      "grad_norm": 0.4083558917045593,
      "learning_rate": 1.2832738552549935e-06,
      "loss": 0.522,
      "step": 2501
    },
    {
      "epoch": 0.42667121418826737,
      "grad_norm": 0.44709694385528564,
      "learning_rate": 1.282744116210975e-06,
      "loss": 0.5333,
      "step": 2502
    },
    {
      "epoch": 0.4268417462482947,
      "grad_norm": 0.4692583680152893,
      "learning_rate": 1.282214290913971e-06,
      "loss": 0.5329,
      "step": 2503
    },
    {
      "epoch": 0.42701227830832195,
      "grad_norm": 0.48816362023353577,
      "learning_rate": 1.281684379525608e-06,
      "loss": 0.5374,
      "step": 2504
    },
    {
      "epoch": 0.42718281036834926,
      "grad_norm": 0.42191752791404724,
      "learning_rate": 1.2811543822075398e-06,
      "loss": 0.5149,
      "step": 2505
    },
    {
      "epoch": 0.4273533424283765,
      "grad_norm": 0.5076993107795715,
      "learning_rate": 1.2806242991214455e-06,
      "loss": 0.5166,
      "step": 2506
    },
    {
      "epoch": 0.42752387448840384,
      "grad_norm": 0.49512800574302673,
      "learning_rate": 1.2800941304290302e-06,
      "loss": 0.5253,
      "step": 2507
    },
    {
      "epoch": 0.4276944065484311,
      "grad_norm": 0.3681822419166565,
      "learning_rate": 1.2795638762920254e-06,
      "loss": 0.5305,
      "step": 2508
    },
    {
      "epoch": 0.42786493860845837,
      "grad_norm": 0.4675699770450592,
      "learning_rate": 1.2790335368721892e-06,
      "loss": 0.529,
      "step": 2509
    },
    {
      "epoch": 0.4280354706684857,
      "grad_norm": 0.4286367893218994,
      "learning_rate": 1.2785031123313046e-06,
      "loss": 0.5161,
      "step": 2510
    },
    {
      "epoch": 0.42820600272851295,
      "grad_norm": 0.35007959604263306,
      "learning_rate": 1.2779726028311814e-06,
      "loss": 0.5233,
      "step": 2511
    },
    {
      "epoch": 0.42837653478854026,
      "grad_norm": 0.4429203271865845,
      "learning_rate": 1.277442008533655e-06,
      "loss": 0.5355,
      "step": 2512
    },
    {
      "epoch": 0.4285470668485675,
      "grad_norm": 0.34723106026649475,
      "learning_rate": 1.276911329600587e-06,
      "loss": 0.5098,
      "step": 2513
    },
    {
      "epoch": 0.42871759890859484,
      "grad_norm": 0.39496925473213196,
      "learning_rate": 1.276380566193864e-06,
      "loss": 0.517,
      "step": 2514
    },
    {
      "epoch": 0.4288881309686221,
      "grad_norm": 0.42757344245910645,
      "learning_rate": 1.2758497184753996e-06,
      "loss": 0.5079,
      "step": 2515
    },
    {
      "epoch": 0.42905866302864937,
      "grad_norm": 0.36008015275001526,
      "learning_rate": 1.2753187866071319e-06,
      "loss": 0.5226,
      "step": 2516
    },
    {
      "epoch": 0.4292291950886767,
      "grad_norm": 0.4112105369567871,
      "learning_rate": 1.2747877707510253e-06,
      "loss": 0.5095,
      "step": 2517
    },
    {
      "epoch": 0.42939972714870395,
      "grad_norm": 0.354653924703598,
      "learning_rate": 1.2742566710690703e-06,
      "loss": 0.5151,
      "step": 2518
    },
    {
      "epoch": 0.42957025920873126,
      "grad_norm": 0.41790133714675903,
      "learning_rate": 1.273725487723282e-06,
      "loss": 0.5117,
      "step": 2519
    },
    {
      "epoch": 0.4297407912687585,
      "grad_norm": 0.34194713830947876,
      "learning_rate": 1.2731942208757011e-06,
      "loss": 0.5109,
      "step": 2520
    },
    {
      "epoch": 0.4299113233287858,
      "grad_norm": 0.435222327709198,
      "learning_rate": 1.2726628706883948e-06,
      "loss": 0.5324,
      "step": 2521
    },
    {
      "epoch": 0.4300818553888131,
      "grad_norm": 0.4597698450088501,
      "learning_rate": 1.2721314373234544e-06,
      "loss": 0.5238,
      "step": 2522
    },
    {
      "epoch": 0.43025238744884037,
      "grad_norm": 0.43926915526390076,
      "learning_rate": 1.2715999209429975e-06,
      "loss": 0.5098,
      "step": 2523
    },
    {
      "epoch": 0.4304229195088677,
      "grad_norm": 0.600695788860321,
      "learning_rate": 1.2710683217091669e-06,
      "loss": 0.5158,
      "step": 2524
    },
    {
      "epoch": 0.43059345156889495,
      "grad_norm": 0.549557626247406,
      "learning_rate": 1.2705366397841304e-06,
      "loss": 0.5343,
      "step": 2525
    },
    {
      "epoch": 0.43076398362892226,
      "grad_norm": 0.43904390931129456,
      "learning_rate": 1.2700048753300808e-06,
      "loss": 0.5275,
      "step": 2526
    },
    {
      "epoch": 0.4309345156889495,
      "grad_norm": 0.446802020072937,
      "learning_rate": 1.2694730285092365e-06,
      "loss": 0.5184,
      "step": 2527
    },
    {
      "epoch": 0.4311050477489768,
      "grad_norm": 0.4266647696495056,
      "learning_rate": 1.2689410994838415e-06,
      "loss": 0.5134,
      "step": 2528
    },
    {
      "epoch": 0.4312755798090041,
      "grad_norm": 0.3999023735523224,
      "learning_rate": 1.2684090884161635e-06,
      "loss": 0.5227,
      "step": 2529
    },
    {
      "epoch": 0.43144611186903137,
      "grad_norm": 0.4615280330181122,
      "learning_rate": 1.2678769954684967e-06,
      "loss": 0.5178,
      "step": 2530
    },
    {
      "epoch": 0.4316166439290587,
      "grad_norm": 0.37461206316947937,
      "learning_rate": 1.2673448208031593e-06,
      "loss": 0.5245,
      "step": 2531
    },
    {
      "epoch": 0.43178717598908595,
      "grad_norm": 0.6492916345596313,
      "learning_rate": 1.2668125645824945e-06,
      "loss": 0.5193,
      "step": 2532
    },
    {
      "epoch": 0.4319577080491132,
      "grad_norm": 0.3368079662322998,
      "learning_rate": 1.2662802269688715e-06,
      "loss": 0.5156,
      "step": 2533
    },
    {
      "epoch": 0.4321282401091405,
      "grad_norm": 0.37977778911590576,
      "learning_rate": 1.2657478081246826e-06,
      "loss": 0.5198,
      "step": 2534
    },
    {
      "epoch": 0.4322987721691678,
      "grad_norm": 0.39551234245300293,
      "learning_rate": 1.2652153082123459e-06,
      "loss": 0.5198,
      "step": 2535
    },
    {
      "epoch": 0.4324693042291951,
      "grad_norm": 0.4476735591888428,
      "learning_rate": 1.2646827273943043e-06,
      "loss": 0.5185,
      "step": 2536
    },
    {
      "epoch": 0.43263983628922237,
      "grad_norm": 0.45589369535446167,
      "learning_rate": 1.2641500658330248e-06,
      "loss": 0.5189,
      "step": 2537
    },
    {
      "epoch": 0.4328103683492497,
      "grad_norm": 0.4167502224445343,
      "learning_rate": 1.2636173236909996e-06,
      "loss": 0.5158,
      "step": 2538
    },
    {
      "epoch": 0.43298090040927695,
      "grad_norm": 0.42350873351097107,
      "learning_rate": 1.2630845011307453e-06,
      "loss": 0.5093,
      "step": 2539
    },
    {
      "epoch": 0.4331514324693042,
      "grad_norm": 0.4852334260940552,
      "learning_rate": 1.262551598314803e-06,
      "loss": 0.5202,
      "step": 2540
    },
    {
      "epoch": 0.4333219645293315,
      "grad_norm": 0.3522724509239197,
      "learning_rate": 1.2620186154057383e-06,
      "loss": 0.515,
      "step": 2541
    },
    {
      "epoch": 0.4334924965893588,
      "grad_norm": 0.4680606424808502,
      "learning_rate": 1.261485552566141e-06,
      "loss": 0.5082,
      "step": 2542
    },
    {
      "epoch": 0.4336630286493861,
      "grad_norm": 0.46284425258636475,
      "learning_rate": 1.2609524099586256e-06,
      "loss": 0.5163,
      "step": 2543
    },
    {
      "epoch": 0.43383356070941337,
      "grad_norm": 0.35307639837265015,
      "learning_rate": 1.2604191877458308e-06,
      "loss": 0.5184,
      "step": 2544
    },
    {
      "epoch": 0.43400409276944063,
      "grad_norm": 0.4327385723590851,
      "learning_rate": 1.2598858860904193e-06,
      "loss": 0.5291,
      "step": 2545
    },
    {
      "epoch": 0.43417462482946795,
      "grad_norm": 0.4670164883136749,
      "learning_rate": 1.2593525051550792e-06,
      "loss": 0.5163,
      "step": 2546
    },
    {
      "epoch": 0.4343451568894952,
      "grad_norm": 0.39771389961242676,
      "learning_rate": 1.258819045102521e-06,
      "loss": 0.5164,
      "step": 2547
    },
    {
      "epoch": 0.4345156889495225,
      "grad_norm": 0.4475443363189697,
      "learning_rate": 1.2582855060954808e-06,
      "loss": 0.5067,
      "step": 2548
    },
    {
      "epoch": 0.4346862210095498,
      "grad_norm": 0.4804600179195404,
      "learning_rate": 1.2577518882967182e-06,
      "loss": 0.5181,
      "step": 2549
    },
    {
      "epoch": 0.4348567530695771,
      "grad_norm": 0.35155707597732544,
      "learning_rate": 1.2572181918690164e-06,
      "loss": 0.5177,
      "step": 2550
    },
    {
      "epoch": 0.43502728512960437,
      "grad_norm": 0.41589272022247314,
      "learning_rate": 1.2566844169751837e-06,
      "loss": 0.5078,
      "step": 2551
    },
    {
      "epoch": 0.43519781718963163,
      "grad_norm": 0.43547019362449646,
      "learning_rate": 1.2561505637780515e-06,
      "loss": 0.5179,
      "step": 2552
    },
    {
      "epoch": 0.43536834924965895,
      "grad_norm": 0.35825809836387634,
      "learning_rate": 1.255616632440475e-06,
      "loss": 0.5254,
      "step": 2553
    },
    {
      "epoch": 0.4355388813096862,
      "grad_norm": 0.3896865248680115,
      "learning_rate": 1.2550826231253336e-06,
      "loss": 0.5141,
      "step": 2554
    },
    {
      "epoch": 0.4357094133697135,
      "grad_norm": 0.3662707209587097,
      "learning_rate": 1.254548535995531e-06,
      "loss": 0.5197,
      "step": 2555
    },
    {
      "epoch": 0.4358799454297408,
      "grad_norm": 0.4130113422870636,
      "learning_rate": 1.2540143712139935e-06,
      "loss": 0.5106,
      "step": 2556
    },
    {
      "epoch": 0.43605047748976805,
      "grad_norm": 0.4342820942401886,
      "learning_rate": 1.2534801289436716e-06,
      "loss": 0.5146,
      "step": 2557
    },
    {
      "epoch": 0.43622100954979537,
      "grad_norm": 0.4437672793865204,
      "learning_rate": 1.25294580934754e-06,
      "loss": 0.5218,
      "step": 2558
    },
    {
      "epoch": 0.43639154160982263,
      "grad_norm": 0.43240147829055786,
      "learning_rate": 1.252411412588596e-06,
      "loss": 0.5192,
      "step": 2559
    },
    {
      "epoch": 0.43656207366984995,
      "grad_norm": 0.4971958100795746,
      "learning_rate": 1.2518769388298606e-06,
      "loss": 0.5165,
      "step": 2560
    },
    {
      "epoch": 0.4367326057298772,
      "grad_norm": 0.4937379062175751,
      "learning_rate": 1.2513423882343797e-06,
      "loss": 0.5088,
      "step": 2561
    },
    {
      "epoch": 0.4369031377899045,
      "grad_norm": 0.44213834404945374,
      "learning_rate": 1.2508077609652204e-06,
      "loss": 0.5179,
      "step": 2562
    },
    {
      "epoch": 0.4370736698499318,
      "grad_norm": 0.4229736924171448,
      "learning_rate": 1.2502730571854747e-06,
      "loss": 0.5228,
      "step": 2563
    },
    {
      "epoch": 0.43724420190995905,
      "grad_norm": 0.5028852224349976,
      "learning_rate": 1.2497382770582582e-06,
      "loss": 0.5113,
      "step": 2564
    },
    {
      "epoch": 0.43741473396998637,
      "grad_norm": 0.4949003756046295,
      "learning_rate": 1.2492034207467082e-06,
      "loss": 0.5146,
      "step": 2565
    },
    {
      "epoch": 0.43758526603001363,
      "grad_norm": 0.43527311086654663,
      "learning_rate": 1.2486684884139868e-06,
      "loss": 0.522,
      "step": 2566
    },
    {
      "epoch": 0.43775579809004095,
      "grad_norm": 0.4403958022594452,
      "learning_rate": 1.2481334802232782e-06,
      "loss": 0.5138,
      "step": 2567
    },
    {
      "epoch": 0.4379263301500682,
      "grad_norm": 0.5198670029640198,
      "learning_rate": 1.2475983963377908e-06,
      "loss": 0.5127,
      "step": 2568
    },
    {
      "epoch": 0.4380968622100955,
      "grad_norm": 0.4271129369735718,
      "learning_rate": 1.247063236920755e-06,
      "loss": 0.5112,
      "step": 2569
    },
    {
      "epoch": 0.4382673942701228,
      "grad_norm": 0.4282945692539215,
      "learning_rate": 1.2465280021354255e-06,
      "loss": 0.5124,
      "step": 2570
    },
    {
      "epoch": 0.43843792633015005,
      "grad_norm": 0.5314650535583496,
      "learning_rate": 1.245992692145078e-06,
      "loss": 0.5135,
      "step": 2571
    },
    {
      "epoch": 0.43860845839017737,
      "grad_norm": 0.3869926631450653,
      "learning_rate": 1.245457307113014e-06,
      "loss": 0.5148,
      "step": 2572
    },
    {
      "epoch": 0.43877899045020463,
      "grad_norm": 0.3884088397026062,
      "learning_rate": 1.2449218472025549e-06,
      "loss": 0.5271,
      "step": 2573
    },
    {
      "epoch": 0.43894952251023195,
      "grad_norm": 0.48845013976097107,
      "learning_rate": 1.2443863125770473e-06,
      "loss": 0.5215,
      "step": 2574
    },
    {
      "epoch": 0.4391200545702592,
      "grad_norm": 0.44390156865119934,
      "learning_rate": 1.243850703399859e-06,
      "loss": 0.5128,
      "step": 2575
    },
    {
      "epoch": 0.4392905866302865,
      "grad_norm": 0.37384307384490967,
      "learning_rate": 1.243315019834381e-06,
      "loss": 0.5167,
      "step": 2576
    },
    {
      "epoch": 0.4394611186903138,
      "grad_norm": 0.45294705033302307,
      "learning_rate": 1.242779262044028e-06,
      "loss": 0.5217,
      "step": 2577
    },
    {
      "epoch": 0.43963165075034105,
      "grad_norm": 0.5091094970703125,
      "learning_rate": 1.2422434301922358e-06,
      "loss": 0.5254,
      "step": 2578
    },
    {
      "epoch": 0.43980218281036837,
      "grad_norm": 0.364459365606308,
      "learning_rate": 1.2417075244424638e-06,
      "loss": 0.5058,
      "step": 2579
    },
    {
      "epoch": 0.43997271487039563,
      "grad_norm": 0.4908876419067383,
      "learning_rate": 1.241171544958194e-06,
      "loss": 0.4999,
      "step": 2580
    },
    {
      "epoch": 0.4401432469304229,
      "grad_norm": 0.6493481993675232,
      "learning_rate": 1.2406354919029294e-06,
      "loss": 0.5307,
      "step": 2581
    },
    {
      "epoch": 0.4403137789904502,
      "grad_norm": 0.5132988095283508,
      "learning_rate": 1.2400993654401974e-06,
      "loss": 0.5156,
      "step": 2582
    },
    {
      "epoch": 0.4404843110504775,
      "grad_norm": 0.4408731758594513,
      "learning_rate": 1.2395631657335471e-06,
      "loss": 0.5192,
      "step": 2583
    },
    {
      "epoch": 0.4406548431105048,
      "grad_norm": 0.524540901184082,
      "learning_rate": 1.2390268929465494e-06,
      "loss": 0.5092,
      "step": 2584
    },
    {
      "epoch": 0.44082537517053205,
      "grad_norm": 0.5012420415878296,
      "learning_rate": 1.2384905472427975e-06,
      "loss": 0.5114,
      "step": 2585
    },
    {
      "epoch": 0.44099590723055937,
      "grad_norm": 0.5423673987388611,
      "learning_rate": 1.2379541287859086e-06,
      "loss": 0.5128,
      "step": 2586
    },
    {
      "epoch": 0.44116643929058663,
      "grad_norm": 0.48322269320487976,
      "learning_rate": 1.2374176377395196e-06,
      "loss": 0.522,
      "step": 2587
    },
    {
      "epoch": 0.4413369713506139,
      "grad_norm": 0.5741340517997742,
      "learning_rate": 1.2368810742672911e-06,
      "loss": 0.5064,
      "step": 2588
    },
    {
      "epoch": 0.4415075034106412,
      "grad_norm": 0.4697413444519043,
      "learning_rate": 1.2363444385329052e-06,
      "loss": 0.5216,
      "step": 2589
    },
    {
      "epoch": 0.4416780354706685,
      "grad_norm": 0.4826829135417938,
      "learning_rate": 1.2358077307000663e-06,
      "loss": 0.5178,
      "step": 2590
    },
    {
      "epoch": 0.4418485675306958,
      "grad_norm": 0.45279660820961,
      "learning_rate": 1.2352709509325006e-06,
      "loss": 0.5115,
      "step": 2591
    },
    {
      "epoch": 0.44201909959072305,
      "grad_norm": 0.38878384232521057,
      "learning_rate": 1.2347340993939569e-06,
      "loss": 0.5173,
      "step": 2592
    },
    {
      "epoch": 0.4421896316507503,
      "grad_norm": 0.5535585880279541,
      "learning_rate": 1.2341971762482045e-06,
      "loss": 0.5106,
      "step": 2593
    },
    {
      "epoch": 0.44236016371077763,
      "grad_norm": 0.4223726689815521,
      "learning_rate": 1.233660181659036e-06,
      "loss": 0.5172,
      "step": 2594
    },
    {
      "epoch": 0.4425306957708049,
      "grad_norm": 0.42020705342292786,
      "learning_rate": 1.233123115790265e-06,
      "loss": 0.539,
      "step": 2595
    },
    {
      "epoch": 0.4427012278308322,
      "grad_norm": 0.43967166543006897,
      "learning_rate": 1.2325859788057274e-06,
      "loss": 0.5054,
      "step": 2596
    },
    {
      "epoch": 0.4428717598908595,
      "grad_norm": 0.3231273889541626,
      "learning_rate": 1.2320487708692797e-06,
      "loss": 0.5142,
      "step": 2597
    },
    {
      "epoch": 0.4430422919508868,
      "grad_norm": 0.4471146762371063,
      "learning_rate": 1.2315114921448013e-06,
      "loss": 0.5186,
      "step": 2598
    },
    {
      "epoch": 0.44321282401091405,
      "grad_norm": 0.40211814641952515,
      "learning_rate": 1.2309741427961927e-06,
      "loss": 0.5255,
      "step": 2599
    },
    {
      "epoch": 0.4433833560709413,
      "grad_norm": 0.42954233288764954,
      "learning_rate": 1.2304367229873756e-06,
      "loss": 0.5116,
      "step": 2600
    },
    {
      "epoch": 0.44355388813096863,
      "grad_norm": 0.6420926451683044,
      "learning_rate": 1.2298992328822937e-06,
      "loss": 0.5109,
      "step": 2601
    },
    {
      "epoch": 0.4437244201909959,
      "grad_norm": 0.5584779381752014,
      "learning_rate": 1.229361672644912e-06,
      "loss": 0.5143,
      "step": 2602
    },
    {
      "epoch": 0.4438949522510232,
      "grad_norm": 0.3985927700996399,
      "learning_rate": 1.2288240424392168e-06,
      "loss": 0.5024,
      "step": 2603
    },
    {
      "epoch": 0.4440654843110505,
      "grad_norm": 0.44909727573394775,
      "learning_rate": 1.2282863424292157e-06,
      "loss": 0.5189,
      "step": 2604
    },
    {
      "epoch": 0.44423601637107774,
      "grad_norm": 0.5488520860671997,
      "learning_rate": 1.227748572778938e-06,
      "loss": 0.5198,
      "step": 2605
    },
    {
      "epoch": 0.44440654843110505,
      "grad_norm": 0.45999622344970703,
      "learning_rate": 1.2272107336524334e-06,
      "loss": 0.5198,
      "step": 2606
    },
    {
      "epoch": 0.4445770804911323,
      "grad_norm": 0.5359815955162048,
      "learning_rate": 1.2266728252137736e-06,
      "loss": 0.5169,
      "step": 2607
    },
    {
      "epoch": 0.44474761255115963,
      "grad_norm": 0.5039993524551392,
      "learning_rate": 1.2261348476270514e-06,
      "loss": 0.5265,
      "step": 2608
    },
    {
      "epoch": 0.4449181446111869,
      "grad_norm": 0.46197831630706787,
      "learning_rate": 1.2255968010563801e-06,
      "loss": 0.5165,
      "step": 2609
    },
    {
      "epoch": 0.4450886766712142,
      "grad_norm": 0.44239184260368347,
      "learning_rate": 1.2250586856658946e-06,
      "loss": 0.5149,
      "step": 2610
    },
    {
      "epoch": 0.4452592087312415,
      "grad_norm": 0.5294963717460632,
      "learning_rate": 1.224520501619751e-06,
      "loss": 0.5126,
      "step": 2611
    },
    {
      "epoch": 0.44542974079126874,
      "grad_norm": 0.40589872002601624,
      "learning_rate": 1.2239822490821249e-06,
      "loss": 0.5245,
      "step": 2612
    },
    {
      "epoch": 0.44560027285129605,
      "grad_norm": 0.4539930522441864,
      "learning_rate": 1.2234439282172145e-06,
      "loss": 0.5163,
      "step": 2613
    },
    {
      "epoch": 0.4457708049113233,
      "grad_norm": 0.44857555627822876,
      "learning_rate": 1.2229055391892384e-06,
      "loss": 0.5104,
      "step": 2614
    },
    {
      "epoch": 0.44594133697135063,
      "grad_norm": 0.5156710147857666,
      "learning_rate": 1.2223670821624355e-06,
      "loss": 0.5081,
      "step": 2615
    },
    {
      "epoch": 0.4461118690313779,
      "grad_norm": 0.34579339623451233,
      "learning_rate": 1.2218285573010654e-06,
      "loss": 0.5181,
      "step": 2616
    },
    {
      "epoch": 0.44628240109140516,
      "grad_norm": 0.5912550687789917,
      "learning_rate": 1.2212899647694094e-06,
      "loss": 0.5162,
      "step": 2617
    },
    {
      "epoch": 0.4464529331514325,
      "grad_norm": 0.4505324959754944,
      "learning_rate": 1.2207513047317685e-06,
      "loss": 0.5228,
      "step": 2618
    },
    {
      "epoch": 0.44662346521145974,
      "grad_norm": 0.42937779426574707,
      "learning_rate": 1.2202125773524641e-06,
      "loss": 0.5093,
      "step": 2619
    },
    {
      "epoch": 0.44679399727148705,
      "grad_norm": 0.4112337529659271,
      "learning_rate": 1.2196737827958393e-06,
      "loss": 0.5122,
      "step": 2620
    },
    {
      "epoch": 0.4469645293315143,
      "grad_norm": 0.3574349880218506,
      "learning_rate": 1.2191349212262563e-06,
      "loss": 0.5197,
      "step": 2621
    },
    {
      "epoch": 0.44713506139154163,
      "grad_norm": 0.48161396384239197,
      "learning_rate": 1.218595992808099e-06,
      "loss": 0.5111,
      "step": 2622
    },
    {
      "epoch": 0.4473055934515689,
      "grad_norm": 0.6249895095825195,
      "learning_rate": 1.2180569977057707e-06,
      "loss": 0.5032,
      "step": 2623
    },
    {
      "epoch": 0.44747612551159616,
      "grad_norm": 0.45782434940338135,
      "learning_rate": 1.2175179360836956e-06,
      "loss": 0.531,
      "step": 2624
    },
    {
      "epoch": 0.4476466575716235,
      "grad_norm": 0.3763248026371002,
      "learning_rate": 1.2169788081063183e-06,
      "loss": 0.5087,
      "step": 2625
    },
    {
      "epoch": 0.44781718963165074,
      "grad_norm": 0.5124301910400391,
      "learning_rate": 1.216439613938103e-06,
      "loss": 0.5101,
      "step": 2626
    },
    {
      "epoch": 0.44798772169167805,
      "grad_norm": 0.41541916131973267,
      "learning_rate": 1.215900353743535e-06,
      "loss": 0.5108,
      "step": 2627
    },
    {
      "epoch": 0.4481582537517053,
      "grad_norm": 0.37283164262771606,
      "learning_rate": 1.2153610276871187e-06,
      "loss": 0.5183,
      "step": 2628
    },
    {
      "epoch": 0.44832878581173263,
      "grad_norm": 0.5663114786148071,
      "learning_rate": 1.2148216359333795e-06,
      "loss": 0.5059,
      "step": 2629
    },
    {
      "epoch": 0.4484993178717599,
      "grad_norm": 0.5109935998916626,
      "learning_rate": 1.214282178646862e-06,
      "loss": 0.4996,
      "step": 2630
    },
    {
      "epoch": 0.44866984993178716,
      "grad_norm": 0.45246654748916626,
      "learning_rate": 1.2137426559921318e-06,
      "loss": 0.5014,
      "step": 2631
    },
    {
      "epoch": 0.4488403819918145,
      "grad_norm": 0.498762309551239,
      "learning_rate": 1.2132030681337738e-06,
      "loss": 0.5135,
      "step": 2632
    },
    {
      "epoch": 0.44901091405184174,
      "grad_norm": 0.6093807220458984,
      "learning_rate": 1.2126634152363925e-06,
      "loss": 0.5157,
      "step": 2633
    },
    {
      "epoch": 0.44918144611186905,
      "grad_norm": 0.4820914566516876,
      "learning_rate": 1.2121236974646127e-06,
      "loss": 0.5281,
      "step": 2634
    },
    {
      "epoch": 0.4493519781718963,
      "grad_norm": 0.40519341826438904,
      "learning_rate": 1.2115839149830798e-06,
      "loss": 0.5134,
      "step": 2635
    },
    {
      "epoch": 0.4495225102319236,
      "grad_norm": 0.5242198705673218,
      "learning_rate": 1.2110440679564567e-06,
      "loss": 0.5173,
      "step": 2636
    },
    {
      "epoch": 0.4496930422919509,
      "grad_norm": 0.379910945892334,
      "learning_rate": 1.210504156549428e-06,
      "loss": 0.5074,
      "step": 2637
    },
    {
      "epoch": 0.44986357435197816,
      "grad_norm": 0.4137760102748871,
      "learning_rate": 1.2099641809266978e-06,
      "loss": 0.5283,
      "step": 2638
    },
    {
      "epoch": 0.4500341064120055,
      "grad_norm": 0.4397674798965454,
      "learning_rate": 1.2094241412529884e-06,
      "loss": 0.5227,
      "step": 2639
    },
    {
      "epoch": 0.45020463847203274,
      "grad_norm": 0.42571043968200684,
      "learning_rate": 1.208884037693043e-06,
      "loss": 0.5006,
      "step": 2640
    },
    {
      "epoch": 0.45037517053206005,
      "grad_norm": 0.4621511995792389,
      "learning_rate": 1.2083438704116235e-06,
      "loss": 0.5283,
      "step": 2641
    },
    {
      "epoch": 0.4505457025920873,
      "grad_norm": 0.4140154719352722,
      "learning_rate": 1.207803639573512e-06,
      "loss": 0.5192,
      "step": 2642
    },
    {
      "epoch": 0.4507162346521146,
      "grad_norm": 0.4587588906288147,
      "learning_rate": 1.2072633453435093e-06,
      "loss": 0.5205,
      "step": 2643
    },
    {
      "epoch": 0.4508867667121419,
      "grad_norm": 0.3948979377746582,
      "learning_rate": 1.2067229878864355e-06,
      "loss": 0.5086,
      "step": 2644
    },
    {
      "epoch": 0.45105729877216916,
      "grad_norm": 0.4351895749568939,
      "learning_rate": 1.206182567367131e-06,
      "loss": 0.5336,
      "step": 2645
    },
    {
      "epoch": 0.4512278308321965,
      "grad_norm": 0.4272501766681671,
      "learning_rate": 1.2056420839504536e-06,
      "loss": 0.5174,
      "step": 2646
    },
    {
      "epoch": 0.45139836289222374,
      "grad_norm": 0.3314935564994812,
      "learning_rate": 1.2051015378012824e-06,
      "loss": 0.5191,
      "step": 2647
    },
    {
      "epoch": 0.451568894952251,
      "grad_norm": 0.3617965579032898,
      "learning_rate": 1.2045609290845139e-06,
      "loss": 0.5259,
      "step": 2648
    },
    {
      "epoch": 0.4517394270122783,
      "grad_norm": 0.3362520635128021,
      "learning_rate": 1.204020257965065e-06,
      "loss": 0.5148,
      "step": 2649
    },
    {
      "epoch": 0.4519099590723056,
      "grad_norm": 0.389750212430954,
      "learning_rate": 1.2034795246078707e-06,
      "loss": 0.5092,
      "step": 2650
    },
    {
      "epoch": 0.4520804911323329,
      "grad_norm": 0.4418506622314453,
      "learning_rate": 1.2029387291778855e-06,
      "loss": 0.5287,
      "step": 2651
    },
    {
      "epoch": 0.45225102319236016,
      "grad_norm": 0.349391371011734,
      "learning_rate": 1.2023978718400821e-06,
      "loss": 0.526,
      "step": 2652
    },
    {
      "epoch": 0.4524215552523875,
      "grad_norm": 0.533689558506012,
      "learning_rate": 1.2018569527594533e-06,
      "loss": 0.5168,
      "step": 2653
    },
    {
      "epoch": 0.45259208731241474,
      "grad_norm": 0.7081760168075562,
      "learning_rate": 1.2013159721010103e-06,
      "loss": 0.5263,
      "step": 2654
    },
    {
      "epoch": 0.452762619372442,
      "grad_norm": 0.5865058302879333,
      "learning_rate": 1.2007749300297818e-06,
      "loss": 0.5188,
      "step": 2655
    },
    {
      "epoch": 0.4529331514324693,
      "grad_norm": 0.39208078384399414,
      "learning_rate": 1.2002338267108172e-06,
      "loss": 0.5193,
      "step": 2656
    },
    {
      "epoch": 0.4531036834924966,
      "grad_norm": 0.5570546984672546,
      "learning_rate": 1.199692662309184e-06,
      "loss": 0.5234,
      "step": 2657
    },
    {
      "epoch": 0.4532742155525239,
      "grad_norm": 0.5408545136451721,
      "learning_rate": 1.199151436989967e-06,
      "loss": 0.5138,
      "step": 2658
    },
    {
      "epoch": 0.45344474761255116,
      "grad_norm": 0.33498716354370117,
      "learning_rate": 1.1986101509182712e-06,
      "loss": 0.5095,
      "step": 2659
    },
    {
      "epoch": 0.4536152796725784,
      "grad_norm": 0.46534663438796997,
      "learning_rate": 1.1980688042592193e-06,
      "loss": 0.5018,
      "step": 2660
    },
    {
      "epoch": 0.45378581173260574,
      "grad_norm": 0.4153357446193695,
      "learning_rate": 1.1975273971779529e-06,
      "loss": 0.5273,
      "step": 2661
    },
    {
      "epoch": 0.453956343792633,
      "grad_norm": 0.43084970116615295,
      "learning_rate": 1.1969859298396318e-06,
      "loss": 0.515,
      "step": 2662
    },
    {
      "epoch": 0.4541268758526603,
      "grad_norm": 0.5067896842956543,
      "learning_rate": 1.1964444024094346e-06,
      "loss": 0.4999,
      "step": 2663
    },
    {
      "epoch": 0.4542974079126876,
      "grad_norm": 0.41986194252967834,
      "learning_rate": 1.195902815052557e-06,
      "loss": 0.5285,
      "step": 2664
    },
    {
      "epoch": 0.4544679399727149,
      "grad_norm": 0.4705030918121338,
      "learning_rate": 1.1953611679342143e-06,
      "loss": 0.5474,
      "step": 2665
    },
    {
      "epoch": 0.45463847203274216,
      "grad_norm": 0.4286784529685974,
      "learning_rate": 1.1948194612196403e-06,
      "loss": 0.5154,
      "step": 2666
    },
    {
      "epoch": 0.4548090040927694,
      "grad_norm": 0.3951047360897064,
      "learning_rate": 1.194277695074085e-06,
      "loss": 0.5075,
      "step": 2667
    },
    {
      "epoch": 0.45497953615279674,
      "grad_norm": 0.44156450033187866,
      "learning_rate": 1.1937358696628186e-06,
      "loss": 0.5234,
      "step": 2668
    },
    {
      "epoch": 0.455150068212824,
      "grad_norm": 0.45110470056533813,
      "learning_rate": 1.1931939851511284e-06,
      "loss": 0.531,
      "step": 2669
    },
    {
      "epoch": 0.4553206002728513,
      "grad_norm": 0.3986232578754425,
      "learning_rate": 1.1926520417043196e-06,
      "loss": 0.5176,
      "step": 2670
    },
    {
      "epoch": 0.4554911323328786,
      "grad_norm": 0.38686272501945496,
      "learning_rate": 1.192110039487716e-06,
      "loss": 0.5148,
      "step": 2671
    },
    {
      "epoch": 0.45566166439290584,
      "grad_norm": 0.486208975315094,
      "learning_rate": 1.1915679786666592e-06,
      "loss": 0.5293,
      "step": 2672
    },
    {
      "epoch": 0.45583219645293316,
      "grad_norm": 0.5004016160964966,
      "learning_rate": 1.1910258594065078e-06,
      "loss": 0.5232,
      "step": 2673
    },
    {
      "epoch": 0.4560027285129604,
      "grad_norm": 0.3850324749946594,
      "learning_rate": 1.1904836818726399e-06,
      "loss": 0.5284,
      "step": 2674
    },
    {
      "epoch": 0.45617326057298774,
      "grad_norm": 0.36427977681159973,
      "learning_rate": 1.1899414462304495e-06,
      "loss": 0.5296,
      "step": 2675
    },
    {
      "epoch": 0.456343792633015,
      "grad_norm": 0.42257794737815857,
      "learning_rate": 1.1893991526453494e-06,
      "loss": 0.5167,
      "step": 2676
    },
    {
      "epoch": 0.4565143246930423,
      "grad_norm": 0.3928245007991791,
      "learning_rate": 1.18885680128277e-06,
      "loss": 0.5142,
      "step": 2677
    },
    {
      "epoch": 0.4566848567530696,
      "grad_norm": 0.39062950015068054,
      "learning_rate": 1.1883143923081593e-06,
      "loss": 0.5094,
      "step": 2678
    },
    {
      "epoch": 0.45685538881309684,
      "grad_norm": 0.4366256594657898,
      "learning_rate": 1.1877719258869828e-06,
      "loss": 0.5274,
      "step": 2679
    },
    {
      "epoch": 0.45702592087312416,
      "grad_norm": 0.48190537095069885,
      "learning_rate": 1.1872294021847232e-06,
      "loss": 0.5197,
      "step": 2680
    },
    {
      "epoch": 0.4571964529331514,
      "grad_norm": 0.34900742769241333,
      "learning_rate": 1.1866868213668815e-06,
      "loss": 0.5189,
      "step": 2681
    },
    {
      "epoch": 0.45736698499317874,
      "grad_norm": 0.45446139574050903,
      "learning_rate": 1.1861441835989754e-06,
      "loss": 0.5291,
      "step": 2682
    },
    {
      "epoch": 0.457537517053206,
      "grad_norm": 0.47677281498908997,
      "learning_rate": 1.1856014890465397e-06,
      "loss": 0.5234,
      "step": 2683
    },
    {
      "epoch": 0.45770804911323326,
      "grad_norm": 0.38069313764572144,
      "learning_rate": 1.1850587378751276e-06,
      "loss": 0.5248,
      "step": 2684
    },
    {
      "epoch": 0.4578785811732606,
      "grad_norm": 0.36604219675064087,
      "learning_rate": 1.1845159302503086e-06,
      "loss": 0.5036,
      "step": 2685
    },
    {
      "epoch": 0.45804911323328784,
      "grad_norm": 0.42271777987480164,
      "learning_rate": 1.1839730663376703e-06,
      "loss": 0.5144,
      "step": 2686
    },
    {
      "epoch": 0.45821964529331516,
      "grad_norm": 0.42497578263282776,
      "learning_rate": 1.1834301463028167e-06,
      "loss": 0.5205,
      "step": 2687
    },
    {
      "epoch": 0.4583901773533424,
      "grad_norm": 0.4204833507537842,
      "learning_rate": 1.1828871703113689e-06,
      "loss": 0.5284,
      "step": 2688
    },
    {
      "epoch": 0.45856070941336974,
      "grad_norm": 0.4158252477645874,
      "learning_rate": 1.1823441385289656e-06,
      "loss": 0.5157,
      "step": 2689
    },
    {
      "epoch": 0.458731241473397,
      "grad_norm": 0.46866708993911743,
      "learning_rate": 1.1818010511212625e-06,
      "loss": 0.5224,
      "step": 2690
    },
    {
      "epoch": 0.45890177353342426,
      "grad_norm": 0.4711799621582031,
      "learning_rate": 1.1812579082539318e-06,
      "loss": 0.513,
      "step": 2691
    },
    {
      "epoch": 0.4590723055934516,
      "grad_norm": 0.6736330389976501,
      "learning_rate": 1.1807147100926626e-06,
      "loss": 0.5061,
      "step": 2692
    },
    {
      "epoch": 0.45924283765347884,
      "grad_norm": 0.6674357056617737,
      "learning_rate": 1.1801714568031617e-06,
      "loss": 0.5326,
      "step": 2693
    },
    {
      "epoch": 0.45941336971350616,
      "grad_norm": 0.6352670192718506,
      "learning_rate": 1.179628148551152e-06,
      "loss": 0.5358,
      "step": 2694
    },
    {
      "epoch": 0.4595839017735334,
      "grad_norm": 0.5269036889076233,
      "learning_rate": 1.1790847855023727e-06,
      "loss": 0.5132,
      "step": 2695
    },
    {
      "epoch": 0.4597544338335607,
      "grad_norm": 0.46895119547843933,
      "learning_rate": 1.178541367822581e-06,
      "loss": 0.5118,
      "step": 2696
    },
    {
      "epoch": 0.459924965893588,
      "grad_norm": 0.6236005425453186,
      "learning_rate": 1.1779978956775506e-06,
      "loss": 0.5222,
      "step": 2697
    },
    {
      "epoch": 0.46009549795361526,
      "grad_norm": 0.48662877082824707,
      "learning_rate": 1.1774543692330701e-06,
      "loss": 0.5203,
      "step": 2698
    },
    {
      "epoch": 0.4602660300136426,
      "grad_norm": 0.4667477309703827,
      "learning_rate": 1.1769107886549465e-06,
      "loss": 0.5168,
      "step": 2699
    },
    {
      "epoch": 0.46043656207366984,
      "grad_norm": 0.4196608066558838,
      "learning_rate": 1.1763671541090027e-06,
      "loss": 0.5126,
      "step": 2700
    },
    {
      "epoch": 0.46060709413369716,
      "grad_norm": 0.5549945831298828,
      "learning_rate": 1.1758234657610779e-06,
      "loss": 0.528,
      "step": 2701
    },
    {
      "epoch": 0.4607776261937244,
      "grad_norm": 0.3792808949947357,
      "learning_rate": 1.175279723777028e-06,
      "loss": 0.5097,
      "step": 2702
    },
    {
      "epoch": 0.4609481582537517,
      "grad_norm": 0.5246736407279968,
      "learning_rate": 1.1747359283227251e-06,
      "loss": 0.5034,
      "step": 2703
    },
    {
      "epoch": 0.461118690313779,
      "grad_norm": 0.5328319668769836,
      "learning_rate": 1.174192079564058e-06,
      "loss": 0.5206,
      "step": 2704
    },
    {
      "epoch": 0.46128922237380626,
      "grad_norm": 0.42339950799942017,
      "learning_rate": 1.1736481776669307e-06,
      "loss": 0.5107,
      "step": 2705
    },
    {
      "epoch": 0.4614597544338336,
      "grad_norm": 0.4282023310661316,
      "learning_rate": 1.1731042227972646e-06,
      "loss": 0.5278,
      "step": 2706
    },
    {
      "epoch": 0.46163028649386084,
      "grad_norm": 0.42935076355934143,
      "learning_rate": 1.1725602151209963e-06,
      "loss": 0.524,
      "step": 2707
    },
    {
      "epoch": 0.4618008185538881,
      "grad_norm": 0.5306663513183594,
      "learning_rate": 1.1720161548040794e-06,
      "loss": 0.5255,
      "step": 2708
    },
    {
      "epoch": 0.4619713506139154,
      "grad_norm": 0.43583446741104126,
      "learning_rate": 1.1714720420124833e-06,
      "loss": 0.5233,
      "step": 2709
    },
    {
      "epoch": 0.4621418826739427,
      "grad_norm": 0.4610784947872162,
      "learning_rate": 1.1709278769121929e-06,
      "loss": 0.5121,
      "step": 2710
    },
    {
      "epoch": 0.46231241473397,
      "grad_norm": 0.414095014333725,
      "learning_rate": 1.1703836596692094e-06,
      "loss": 0.5149,
      "step": 2711
    },
    {
      "epoch": 0.46248294679399726,
      "grad_norm": 0.4780326783657074,
      "learning_rate": 1.16983939044955e-06,
      "loss": 0.5058,
      "step": 2712
    },
    {
      "epoch": 0.4626534788540246,
      "grad_norm": 0.5305455923080444,
      "learning_rate": 1.169295069419248e-06,
      "loss": 0.5088,
      "step": 2713
    },
    {
      "epoch": 0.46282401091405184,
      "grad_norm": 0.36969780921936035,
      "learning_rate": 1.1687506967443516e-06,
      "loss": 0.5167,
      "step": 2714
    },
    {
      "epoch": 0.4629945429740791,
      "grad_norm": 0.47688403725624084,
      "learning_rate": 1.168206272590926e-06,
      "loss": 0.5223,
      "step": 2715
    },
    {
      "epoch": 0.4631650750341064,
      "grad_norm": 0.43216776847839355,
      "learning_rate": 1.1676617971250505e-06,
      "loss": 0.5096,
      "step": 2716
    },
    {
      "epoch": 0.4633356070941337,
      "grad_norm": 0.49492067098617554,
      "learning_rate": 1.1671172705128217e-06,
      "loss": 0.513,
      "step": 2717
    },
    {
      "epoch": 0.463506139154161,
      "grad_norm": 0.4933083951473236,
      "learning_rate": 1.166572692920351e-06,
      "loss": 0.508,
      "step": 2718
    },
    {
      "epoch": 0.46367667121418826,
      "grad_norm": 0.4092305302619934,
      "learning_rate": 1.166028064513765e-06,
      "loss": 0.5214,
      "step": 2719
    },
    {
      "epoch": 0.4638472032742155,
      "grad_norm": 0.5383063554763794,
      "learning_rate": 1.1654833854592067e-06,
      "loss": 0.5273,
      "step": 2720
    },
    {
      "epoch": 0.46401773533424284,
      "grad_norm": 0.3836551606655121,
      "learning_rate": 1.1649386559228343e-06,
      "loss": 0.5189,
      "step": 2721
    },
    {
      "epoch": 0.4641882673942701,
      "grad_norm": 0.5678832530975342,
      "learning_rate": 1.1643938760708207e-06,
      "loss": 0.5089,
      "step": 2722
    },
    {
      "epoch": 0.4643587994542974,
      "grad_norm": 0.4550630748271942,
      "learning_rate": 1.1638490460693547e-06,
      "loss": 0.5319,
      "step": 2723
    },
    {
      "epoch": 0.4645293315143247,
      "grad_norm": 0.431937575340271,
      "learning_rate": 1.1633041660846406e-06,
      "loss": 0.5343,
      "step": 2724
    },
    {
      "epoch": 0.464699863574352,
      "grad_norm": 0.47891420125961304,
      "learning_rate": 1.1627592362828974e-06,
      "loss": 0.5196,
      "step": 2725
    },
    {
      "epoch": 0.46487039563437926,
      "grad_norm": 0.4315849244594574,
      "learning_rate": 1.1622142568303596e-06,
      "loss": 0.5291,
      "step": 2726
    },
    {
      "epoch": 0.4650409276944065,
      "grad_norm": 0.5645219087600708,
      "learning_rate": 1.1616692278932775e-06,
      "loss": 0.5213,
      "step": 2727
    },
    {
      "epoch": 0.46521145975443384,
      "grad_norm": 0.41248124837875366,
      "learning_rate": 1.1611241496379145e-06,
      "loss": 0.5216,
      "step": 2728
    },
    {
      "epoch": 0.4653819918144611,
      "grad_norm": 0.5293885469436646,
      "learning_rate": 1.1605790222305517e-06,
      "loss": 0.5101,
      "step": 2729
    },
    {
      "epoch": 0.4655525238744884,
      "grad_norm": 0.5205586552619934,
      "learning_rate": 1.160033845837483e-06,
      "loss": 0.5221,
      "step": 2730
    },
    {
      "epoch": 0.4657230559345157,
      "grad_norm": 0.573829710483551,
      "learning_rate": 1.1594886206250187e-06,
      "loss": 0.5098,
      "step": 2731
    },
    {
      "epoch": 0.465893587994543,
      "grad_norm": 0.5038020610809326,
      "learning_rate": 1.1589433467594827e-06,
      "loss": 0.5069,
      "step": 2732
    },
    {
      "epoch": 0.46606412005457026,
      "grad_norm": 0.5627439022064209,
      "learning_rate": 1.1583980244072151e-06,
      "loss": 0.5098,
      "step": 2733
    },
    {
      "epoch": 0.4662346521145975,
      "grad_norm": 0.45539039373397827,
      "learning_rate": 1.1578526537345699e-06,
      "loss": 0.5081,
      "step": 2734
    },
    {
      "epoch": 0.46640518417462484,
      "grad_norm": 0.5840868949890137,
      "learning_rate": 1.157307234907916e-06,
      "loss": 0.5194,
      "step": 2735
    },
    {
      "epoch": 0.4665757162346521,
      "grad_norm": 0.4483746290206909,
      "learning_rate": 1.1567617680936373e-06,
      "loss": 0.5116,
      "step": 2736
    },
    {
      "epoch": 0.4667462482946794,
      "grad_norm": 0.4605935215950012,
      "learning_rate": 1.156216253458132e-06,
      "loss": 0.5073,
      "step": 2737
    },
    {
      "epoch": 0.4669167803547067,
      "grad_norm": 0.3922935426235199,
      "learning_rate": 1.1556706911678126e-06,
      "loss": 0.5125,
      "step": 2738
    },
    {
      "epoch": 0.46708731241473395,
      "grad_norm": 0.4241859018802643,
      "learning_rate": 1.1551250813891068e-06,
      "loss": 0.5119,
      "step": 2739
    },
    {
      "epoch": 0.46725784447476126,
      "grad_norm": 0.4191175103187561,
      "learning_rate": 1.1545794242884569e-06,
      "loss": 0.5128,
      "step": 2740
    },
    {
      "epoch": 0.4674283765347885,
      "grad_norm": 0.40228796005249023,
      "learning_rate": 1.1540337200323188e-06,
      "loss": 0.51,
      "step": 2741
    },
    {
      "epoch": 0.46759890859481584,
      "grad_norm": 0.45950260758399963,
      "learning_rate": 1.153487968787163e-06,
      "loss": 0.5123,
      "step": 2742
    },
    {
      "epoch": 0.4677694406548431,
      "grad_norm": 0.4424385726451874,
      "learning_rate": 1.1529421707194754e-06,
      "loss": 0.5178,
      "step": 2743
    },
    {
      "epoch": 0.4679399727148704,
      "grad_norm": 0.4888211190700531,
      "learning_rate": 1.1523963259957549e-06,
      "loss": 0.5218,
      "step": 2744
    },
    {
      "epoch": 0.4681105047748977,
      "grad_norm": 0.4428039789199829,
      "learning_rate": 1.1518504347825146e-06,
      "loss": 0.5114,
      "step": 2745
    },
    {
      "epoch": 0.46828103683492495,
      "grad_norm": 0.4225022494792938,
      "learning_rate": 1.151304497246283e-06,
      "loss": 0.5083,
      "step": 2746
    },
    {
      "epoch": 0.46845156889495226,
      "grad_norm": 0.46187952160835266,
      "learning_rate": 1.1507585135536017e-06,
      "loss": 0.5126,
      "step": 2747
    },
    {
      "epoch": 0.4686221009549795,
      "grad_norm": 0.4135882556438446,
      "learning_rate": 1.1502124838710267e-06,
      "loss": 0.5216,
      "step": 2748
    },
    {
      "epoch": 0.46879263301500684,
      "grad_norm": 0.45884090662002563,
      "learning_rate": 1.1496664083651283e-06,
      "loss": 0.5152,
      "step": 2749
    },
    {
      "epoch": 0.4689631650750341,
      "grad_norm": 0.40598103404045105,
      "learning_rate": 1.1491202872024897e-06,
      "loss": 0.5108,
      "step": 2750
    },
    {
      "epoch": 0.46913369713506137,
      "grad_norm": 0.36626097559928894,
      "learning_rate": 1.1485741205497096e-06,
      "loss": 0.5047,
      "step": 2751
    },
    {
      "epoch": 0.4693042291950887,
      "grad_norm": 0.43675321340560913,
      "learning_rate": 1.1480279085733997e-06,
      "loss": 0.5075,
      "step": 2752
    },
    {
      "epoch": 0.46947476125511595,
      "grad_norm": 0.415915310382843,
      "learning_rate": 1.1474816514401853e-06,
      "loss": 0.5203,
      "step": 2753
    },
    {
      "epoch": 0.46964529331514326,
      "grad_norm": 0.4475071430206299,
      "learning_rate": 1.1469353493167058e-06,
      "loss": 0.5092,
      "step": 2754
    },
    {
      "epoch": 0.4698158253751705,
      "grad_norm": 0.43330270051956177,
      "learning_rate": 1.1463890023696145e-06,
      "loss": 0.509,
      "step": 2755
    },
    {
      "epoch": 0.46998635743519784,
      "grad_norm": 0.4516720771789551,
      "learning_rate": 1.1458426107655781e-06,
      "loss": 0.5084,
      "step": 2756
    },
    {
      "epoch": 0.4701568894952251,
      "grad_norm": 0.5397230982780457,
      "learning_rate": 1.145296174671277e-06,
      "loss": 0.5206,
      "step": 2757
    },
    {
      "epoch": 0.47032742155525237,
      "grad_norm": 0.5378764867782593,
      "learning_rate": 1.1447496942534057e-06,
      "loss": 0.5257,
      "step": 2758
    },
    {
      "epoch": 0.4704979536152797,
      "grad_norm": 0.5214230418205261,
      "learning_rate": 1.1442031696786707e-06,
      "loss": 0.5277,
      "step": 2759
    },
    {
      "epoch": 0.47066848567530695,
      "grad_norm": 0.4218697249889374,
      "learning_rate": 1.143656601113794e-06,
      "loss": 0.5091,
      "step": 2760
    },
    {
      "epoch": 0.47083901773533426,
      "grad_norm": 0.46463119983673096,
      "learning_rate": 1.143109988725509e-06,
      "loss": 0.5135,
      "step": 2761
    },
    {
      "epoch": 0.4710095497953615,
      "grad_norm": 0.4969916343688965,
      "learning_rate": 1.1425633326805646e-06,
      "loss": 0.5159,
      "step": 2762
    },
    {
      "epoch": 0.4711800818553888,
      "grad_norm": 0.4337003827095032,
      "learning_rate": 1.142016633145721e-06,
      "loss": 0.5173,
      "step": 2763
    },
    {
      "epoch": 0.4713506139154161,
      "grad_norm": 0.500644326210022,
      "learning_rate": 1.1414698902877528e-06,
      "loss": 0.5197,
      "step": 2764
    },
    {
      "epoch": 0.47152114597544337,
      "grad_norm": 0.553901731967926,
      "learning_rate": 1.1409231042734478e-06,
      "loss": 0.5137,
      "step": 2765
    },
    {
      "epoch": 0.4716916780354707,
      "grad_norm": 0.47816386818885803,
      "learning_rate": 1.1403762752696063e-06,
      "loss": 0.5124,
      "step": 2766
    },
    {
      "epoch": 0.47186221009549795,
      "grad_norm": 0.3969438970088959,
      "learning_rate": 1.1398294034430424e-06,
      "loss": 0.5041,
      "step": 2767
    },
    {
      "epoch": 0.47203274215552526,
      "grad_norm": 0.4518330991268158,
      "learning_rate": 1.1392824889605835e-06,
      "loss": 0.5071,
      "step": 2768
    },
    {
      "epoch": 0.4722032742155525,
      "grad_norm": 0.63393634557724,
      "learning_rate": 1.1387355319890685e-06,
      "loss": 0.5224,
      "step": 2769
    },
    {
      "epoch": 0.4723738062755798,
      "grad_norm": 0.5123917460441589,
      "learning_rate": 1.1381885326953508e-06,
      "loss": 0.5103,
      "step": 2770
    },
    {
      "epoch": 0.4725443383356071,
      "grad_norm": 0.3764648139476776,
      "learning_rate": 1.1376414912462966e-06,
      "loss": 0.5102,
      "step": 2771
    },
    {
      "epoch": 0.47271487039563437,
      "grad_norm": 0.523368239402771,
      "learning_rate": 1.137094407808784e-06,
      "loss": 0.5276,
      "step": 2772
    },
    {
      "epoch": 0.4728854024556617,
      "grad_norm": 0.3376198709011078,
      "learning_rate": 1.1365472825497043e-06,
      "loss": 0.5134,
      "step": 2773
    },
    {
      "epoch": 0.47305593451568895,
      "grad_norm": 0.5027055144309998,
      "learning_rate": 1.1360001156359624e-06,
      "loss": 0.5068,
      "step": 2774
    },
    {
      "epoch": 0.4732264665757162,
      "grad_norm": 0.45107799768447876,
      "learning_rate": 1.135452907234475e-06,
      "loss": 0.5024,
      "step": 2775
    },
    {
      "epoch": 0.4733969986357435,
      "grad_norm": 0.47594237327575684,
      "learning_rate": 1.1349056575121714e-06,
      "loss": 0.5217,
      "step": 2776
    },
    {
      "epoch": 0.4735675306957708,
      "grad_norm": 0.495340496301651,
      "learning_rate": 1.134358366635994e-06,
      "loss": 0.5101,
      "step": 2777
    },
    {
      "epoch": 0.4737380627557981,
      "grad_norm": 0.4050142765045166,
      "learning_rate": 1.1338110347728973e-06,
      "loss": 0.5053,
      "step": 2778
    },
    {
      "epoch": 0.47390859481582537,
      "grad_norm": 0.4666402041912079,
      "learning_rate": 1.1332636620898487e-06,
      "loss": 0.5138,
      "step": 2779
    },
    {
      "epoch": 0.4740791268758527,
      "grad_norm": 0.403601735830307,
      "learning_rate": 1.1327162487538284e-06,
      "loss": 0.5143,
      "step": 2780
    },
    {
      "epoch": 0.47424965893587995,
      "grad_norm": 0.4783070981502533,
      "learning_rate": 1.1321687949318277e-06,
      "loss": 0.5007,
      "step": 2781
    },
    {
      "epoch": 0.4744201909959072,
      "grad_norm": 0.6048984527587891,
      "learning_rate": 1.1316213007908514e-06,
      "loss": 0.5092,
      "step": 2782
    },
    {
      "epoch": 0.4745907230559345,
      "grad_norm": 0.39854034781455994,
      "learning_rate": 1.1310737664979169e-06,
      "loss": 0.5126,
      "step": 2783
    },
    {
      "epoch": 0.4747612551159618,
      "grad_norm": 0.5207147002220154,
      "learning_rate": 1.130526192220052e-06,
      "loss": 0.514,
      "step": 2784
    },
    {
      "epoch": 0.4749317871759891,
      "grad_norm": 0.6135899424552917,
      "learning_rate": 1.1299785781242983e-06,
      "loss": 0.5014,
      "step": 2785
    },
    {
      "epoch": 0.47510231923601637,
      "grad_norm": 0.5457688570022583,
      "learning_rate": 1.1294309243777097e-06,
      "loss": 0.5098,
      "step": 2786
    },
    {
      "epoch": 0.47527285129604363,
      "grad_norm": 0.44323474168777466,
      "learning_rate": 1.1288832311473509e-06,
      "loss": 0.5142,
      "step": 2787
    },
    {
      "epoch": 0.47544338335607095,
      "grad_norm": 0.40029415488243103,
      "learning_rate": 1.1283354986002997e-06,
      "loss": 0.5173,
      "step": 2788
    },
    {
      "epoch": 0.4756139154160982,
      "grad_norm": 0.4944052994251251,
      "learning_rate": 1.1277877269036463e-06,
      "loss": 0.5206,
      "step": 2789
    },
    {
      "epoch": 0.4757844474761255,
      "grad_norm": 0.37801826000213623,
      "learning_rate": 1.1272399162244908e-06,
      "loss": 0.5143,
      "step": 2790
    },
    {
      "epoch": 0.4759549795361528,
      "grad_norm": 0.4556177258491516,
      "learning_rate": 1.1266920667299473e-06,
      "loss": 0.5117,
      "step": 2791
    },
    {
      "epoch": 0.4761255115961801,
      "grad_norm": 0.4916275441646576,
      "learning_rate": 1.126144178587141e-06,
      "loss": 0.5183,
      "step": 2792
    },
    {
      "epoch": 0.47629604365620737,
      "grad_norm": 0.3971802294254303,
      "learning_rate": 1.1255962519632083e-06,
      "loss": 0.5098,
      "step": 2793
    },
    {
      "epoch": 0.47646657571623463,
      "grad_norm": 0.5458121299743652,
      "learning_rate": 1.1250482870252985e-06,
      "loss": 0.5121,
      "step": 2794
    },
    {
      "epoch": 0.47663710777626195,
      "grad_norm": 0.7432170510292053,
      "learning_rate": 1.1245002839405715e-06,
      "loss": 0.5216,
      "step": 2795
    },
    {
      "epoch": 0.4768076398362892,
      "grad_norm": 0.5885338187217712,
      "learning_rate": 1.1239522428761996e-06,
      "loss": 0.52,
      "step": 2796
    },
    {
      "epoch": 0.4769781718963165,
      "grad_norm": 0.36348477005958557,
      "learning_rate": 1.1234041639993664e-06,
      "loss": 0.5196,
      "step": 2797
    },
    {
      "epoch": 0.4771487039563438,
      "grad_norm": 0.6516478061676025,
      "learning_rate": 1.122856047477267e-06,
      "loss": 0.5212,
      "step": 2798
    },
    {
      "epoch": 0.47731923601637105,
      "grad_norm": 0.5506260395050049,
      "learning_rate": 1.122307893477108e-06,
      "loss": 0.5197,
      "step": 2799
    },
    {
      "epoch": 0.47748976807639837,
      "grad_norm": 0.34980344772338867,
      "learning_rate": 1.1217597021661071e-06,
      "loss": 0.5132,
      "step": 2800
    },
    {
      "epoch": 0.47766030013642563,
      "grad_norm": 0.5539247989654541,
      "learning_rate": 1.121211473711494e-06,
      "loss": 0.5019,
      "step": 2801
    },
    {
      "epoch": 0.47783083219645295,
      "grad_norm": 0.40467381477355957,
      "learning_rate": 1.1206632082805099e-06,
      "loss": 0.5137,
      "step": 2802
    },
    {
      "epoch": 0.4780013642564802,
      "grad_norm": 0.3608384132385254,
      "learning_rate": 1.120114906040406e-06,
      "loss": 0.5236,
      "step": 2803
    },
    {
      "epoch": 0.4781718963165075,
      "grad_norm": 0.41437652707099915,
      "learning_rate": 1.1195665671584464e-06,
      "loss": 0.5126,
      "step": 2804
    },
    {
      "epoch": 0.4783424283765348,
      "grad_norm": 0.35578078031539917,
      "learning_rate": 1.119018191801905e-06,
      "loss": 0.518,
      "step": 2805
    },
    {
      "epoch": 0.47851296043656205,
      "grad_norm": 0.43420663475990295,
      "learning_rate": 1.1184697801380675e-06,
      "loss": 0.5169,
      "step": 2806
    },
    {
      "epoch": 0.47868349249658937,
      "grad_norm": 0.4014913737773895,
      "learning_rate": 1.1179213323342305e-06,
      "loss": 0.5162,
      "step": 2807
    },
    {
      "epoch": 0.47885402455661663,
      "grad_norm": 0.4695773124694824,
      "learning_rate": 1.117372848557702e-06,
      "loss": 0.5227,
      "step": 2808
    },
    {
      "epoch": 0.47902455661664395,
      "grad_norm": 0.5101913213729858,
      "learning_rate": 1.1168243289758002e-06,
      "loss": 0.5246,
      "step": 2809
    },
    {
      "epoch": 0.4791950886766712,
      "grad_norm": 0.46013885736465454,
      "learning_rate": 1.1162757737558546e-06,
      "loss": 0.5141,
      "step": 2810
    },
    {
      "epoch": 0.4793656207366985,
      "grad_norm": 0.46474412083625793,
      "learning_rate": 1.1157271830652065e-06,
      "loss": 0.5162,
      "step": 2811
    },
    {
      "epoch": 0.4795361527967258,
      "grad_norm": 0.38749903440475464,
      "learning_rate": 1.115178557071206e-06,
      "loss": 0.5156,
      "step": 2812
    },
    {
      "epoch": 0.47970668485675305,
      "grad_norm": 0.43065133690834045,
      "learning_rate": 1.114629895941216e-06,
      "loss": 0.5211,
      "step": 2813
    },
    {
      "epoch": 0.47987721691678037,
      "grad_norm": 0.4182431399822235,
      "learning_rate": 1.114081199842609e-06,
      "loss": 0.531,
      "step": 2814
    },
    {
      "epoch": 0.48004774897680763,
      "grad_norm": 0.49661538004875183,
      "learning_rate": 1.113532468942768e-06,
      "loss": 0.509,
      "step": 2815
    },
    {
      "epoch": 0.48021828103683495,
      "grad_norm": 0.5234668254852295,
      "learning_rate": 1.1129837034090877e-06,
      "loss": 0.5221,
      "step": 2816
    },
    {
      "epoch": 0.4803888130968622,
      "grad_norm": 0.6002328991889954,
      "learning_rate": 1.1124349034089725e-06,
      "loss": 0.5253,
      "step": 2817
    },
    {
      "epoch": 0.4805593451568895,
      "grad_norm": 0.3524320721626282,
      "learning_rate": 1.1118860691098372e-06,
      "loss": 0.5361,
      "step": 2818
    },
    {
      "epoch": 0.4807298772169168,
      "grad_norm": 0.5052375793457031,
      "learning_rate": 1.1113372006791075e-06,
      "loss": 0.5061,
      "step": 2819
    },
    {
      "epoch": 0.48090040927694405,
      "grad_norm": 0.4445066452026367,
      "learning_rate": 1.1107882982842197e-06,
      "loss": 0.525,
      "step": 2820
    },
    {
      "epoch": 0.48107094133697137,
      "grad_norm": 0.3796360492706299,
      "learning_rate": 1.1102393620926197e-06,
      "loss": 0.5083,
      "step": 2821
    },
    {
      "epoch": 0.48124147339699863,
      "grad_norm": 0.4080047905445099,
      "learning_rate": 1.1096903922717646e-06,
      "loss": 0.5205,
      "step": 2822
    },
    {
      "epoch": 0.4814120054570259,
      "grad_norm": 0.4210074245929718,
      "learning_rate": 1.1091413889891213e-06,
      "loss": 0.523,
      "step": 2823
    },
    {
      "epoch": 0.4815825375170532,
      "grad_norm": 0.44352221488952637,
      "learning_rate": 1.1085923524121663e-06,
      "loss": 0.5274,
      "step": 2824
    },
    {
      "epoch": 0.4817530695770805,
      "grad_norm": 0.46211662888526917,
      "learning_rate": 1.1080432827083875e-06,
      "loss": 0.5199,
      "step": 2825
    },
    {
      "epoch": 0.4819236016371078,
      "grad_norm": 0.4778159260749817,
      "learning_rate": 1.107494180045282e-06,
      "loss": 0.5188,
      "step": 2826
    },
    {
      "epoch": 0.48209413369713505,
      "grad_norm": 0.47870731353759766,
      "learning_rate": 1.1069450445903575e-06,
      "loss": 0.5114,
      "step": 2827
    },
    {
      "epoch": 0.48226466575716237,
      "grad_norm": 0.44184383749961853,
      "learning_rate": 1.106395876511131e-06,
      "loss": 0.5096,
      "step": 2828
    },
    {
      "epoch": 0.48243519781718963,
      "grad_norm": 0.5257557034492493,
      "learning_rate": 1.1058466759751305e-06,
      "loss": 0.5197,
      "step": 2829
    },
    {
      "epoch": 0.4826057298772169,
      "grad_norm": 0.4767327308654785,
      "learning_rate": 1.105297443149893e-06,
      "loss": 0.5233,
      "step": 2830
    },
    {
      "epoch": 0.4827762619372442,
      "grad_norm": 0.5272330045700073,
      "learning_rate": 1.1047481782029655e-06,
      "loss": 0.527,
      "step": 2831
    },
    {
      "epoch": 0.4829467939972715,
      "grad_norm": 0.5258830785751343,
      "learning_rate": 1.1041988813019052e-06,
      "loss": 0.5189,
      "step": 2832
    },
    {
      "epoch": 0.4831173260572988,
      "grad_norm": 0.4522515833377838,
      "learning_rate": 1.1036495526142783e-06,
      "loss": 0.5173,
      "step": 2833
    },
    {
      "epoch": 0.48328785811732605,
      "grad_norm": 0.6141322255134583,
      "learning_rate": 1.103100192307662e-06,
      "loss": 0.541,
      "step": 2834
    },
    {
      "epoch": 0.48345839017735337,
      "grad_norm": 0.37719348073005676,
      "learning_rate": 1.1025508005496419e-06,
      "loss": 0.5002,
      "step": 2835
    },
    {
      "epoch": 0.48362892223738063,
      "grad_norm": 0.6246870160102844,
      "learning_rate": 1.1020013775078134e-06,
      "loss": 0.519,
      "step": 2836
    },
    {
      "epoch": 0.4837994542974079,
      "grad_norm": 0.5842008590698242,
      "learning_rate": 1.1014519233497822e-06,
      "loss": 0.5316,
      "step": 2837
    },
    {
      "epoch": 0.4839699863574352,
      "grad_norm": 0.46794211864471436,
      "learning_rate": 1.1009024382431625e-06,
      "loss": 0.5244,
      "step": 2838
    },
    {
      "epoch": 0.4841405184174625,
      "grad_norm": 0.4208543598651886,
      "learning_rate": 1.1003529223555787e-06,
      "loss": 0.5145,
      "step": 2839
    },
    {
      "epoch": 0.4843110504774898,
      "grad_norm": 0.429081529378891,
      "learning_rate": 1.0998033758546642e-06,
      "loss": 0.5112,
      "step": 2840
    },
    {
      "epoch": 0.48448158253751705,
      "grad_norm": 0.44880661368370056,
      "learning_rate": 1.099253798908062e-06,
      "loss": 0.5245,
      "step": 2841
    },
    {
      "epoch": 0.4846521145975443,
      "grad_norm": 0.4802000820636749,
      "learning_rate": 1.0987041916834237e-06,
      "loss": 0.5206,
      "step": 2842
    },
    {
      "epoch": 0.48482264665757163,
      "grad_norm": 0.5355414152145386,
      "learning_rate": 1.0981545543484111e-06,
      "loss": 0.5095,
      "step": 2843
    },
    {
      "epoch": 0.4849931787175989,
      "grad_norm": 0.434264212846756,
      "learning_rate": 1.0976048870706952e-06,
      "loss": 0.5206,
      "step": 2844
    },
    {
      "epoch": 0.4851637107776262,
      "grad_norm": 0.5525639653205872,
      "learning_rate": 1.0970551900179545e-06,
      "loss": 0.5338,
      "step": 2845
    },
    {
      "epoch": 0.4853342428376535,
      "grad_norm": 0.4568883180618286,
      "learning_rate": 1.0965054633578789e-06,
      "loss": 0.5131,
      "step": 2846
    },
    {
      "epoch": 0.4855047748976808,
      "grad_norm": 0.5602957606315613,
      "learning_rate": 1.0959557072581654e-06,
      "loss": 0.4977,
      "step": 2847
    },
    {
      "epoch": 0.48567530695770805,
      "grad_norm": 0.552498459815979,
      "learning_rate": 1.0954059218865213e-06,
      "loss": 0.5246,
      "step": 2848
    },
    {
      "epoch": 0.4858458390177353,
      "grad_norm": 0.5079182982444763,
      "learning_rate": 1.094856107410662e-06,
      "loss": 0.5097,
      "step": 2849
    },
    {
      "epoch": 0.48601637107776263,
      "grad_norm": 0.4334324300289154,
      "learning_rate": 1.094306263998312e-06,
      "loss": 0.5152,
      "step": 2850
    },
    {
      "epoch": 0.4861869031377899,
      "grad_norm": 0.5188226103782654,
      "learning_rate": 1.0937563918172052e-06,
      "loss": 0.5159,
      "step": 2851
    },
    {
      "epoch": 0.4863574351978172,
      "grad_norm": 0.4685499370098114,
      "learning_rate": 1.0932064910350836e-06,
      "loss": 0.5064,
      "step": 2852
    },
    {
      "epoch": 0.4865279672578445,
      "grad_norm": 0.4312712252140045,
      "learning_rate": 1.092656561819698e-06,
      "loss": 0.5146,
      "step": 2853
    },
    {
      "epoch": 0.48669849931787174,
      "grad_norm": 0.4952917993068695,
      "learning_rate": 1.0921066043388078e-06,
      "loss": 0.5091,
      "step": 2854
    },
    {
      "epoch": 0.48686903137789905,
      "grad_norm": 0.36304834485054016,
      "learning_rate": 1.0915566187601814e-06,
      "loss": 0.5082,
      "step": 2855
    },
    {
      "epoch": 0.4870395634379263,
      "grad_norm": 0.46254342794418335,
      "learning_rate": 1.0910066052515956e-06,
      "loss": 0.519,
      "step": 2856
    },
    {
      "epoch": 0.48721009549795363,
      "grad_norm": 0.3810098171234131,
      "learning_rate": 1.0904565639808358e-06,
      "loss": 0.4963,
      "step": 2857
    },
    {
      "epoch": 0.4873806275579809,
      "grad_norm": 0.42867311835289,
      "learning_rate": 1.0899064951156957e-06,
      "loss": 0.4969,
      "step": 2858
    },
    {
      "epoch": 0.4875511596180082,
      "grad_norm": 0.3933941721916199,
      "learning_rate": 1.0893563988239773e-06,
      "loss": 0.5132,
      "step": 2859
    },
    {
      "epoch": 0.4877216916780355,
      "grad_norm": 0.3573196828365326,
      "learning_rate": 1.0888062752734915e-06,
      "loss": 0.521,
      "step": 2860
    },
    {
      "epoch": 0.48789222373806274,
      "grad_norm": 0.3725636601448059,
      "learning_rate": 1.0882561246320572e-06,
      "loss": 0.52,
      "step": 2861
    },
    {
      "epoch": 0.48806275579809005,
      "grad_norm": 0.39118441939353943,
      "learning_rate": 1.0877059470675007e-06,
      "loss": 0.5227,
      "step": 2862
    },
    {
      "epoch": 0.4882332878581173,
      "grad_norm": 0.42825067043304443,
      "learning_rate": 1.0871557427476585e-06,
      "loss": 0.5184,
      "step": 2863
    },
    {
      "epoch": 0.48840381991814463,
      "grad_norm": 0.4308590590953827,
      "learning_rate": 1.0866055118403735e-06,
      "loss": 0.5167,
      "step": 2864
    },
    {
      "epoch": 0.4885743519781719,
      "grad_norm": 0.39540690183639526,
      "learning_rate": 1.0860552545134972e-06,
      "loss": 0.5036,
      "step": 2865
    },
    {
      "epoch": 0.48874488403819916,
      "grad_norm": 0.4011569917201996,
      "learning_rate": 1.0855049709348896e-06,
      "loss": 0.515,
      "step": 2866
    },
    {
      "epoch": 0.4889154160982265,
      "grad_norm": 0.4647868275642395,
      "learning_rate": 1.084954661272418e-06,
      "loss": 0.5212,
      "step": 2867
    },
    {
      "epoch": 0.48908594815825374,
      "grad_norm": 0.39145034551620483,
      "learning_rate": 1.0844043256939585e-06,
      "loss": 0.5219,
      "step": 2868
    },
    {
      "epoch": 0.48925648021828105,
      "grad_norm": 0.40879708528518677,
      "learning_rate": 1.083853964367395e-06,
      "loss": 0.52,
      "step": 2869
    },
    {
      "epoch": 0.4894270122783083,
      "grad_norm": 0.425218403339386,
      "learning_rate": 1.0833035774606176e-06,
      "loss": 0.5027,
      "step": 2870
    },
    {
      "epoch": 0.48959754433833563,
      "grad_norm": 0.36446821689605713,
      "learning_rate": 1.0827531651415268e-06,
      "loss": 0.5184,
      "step": 2871
    },
    {
      "epoch": 0.4897680763983629,
      "grad_norm": 0.43742719292640686,
      "learning_rate": 1.0822027275780289e-06,
      "loss": 0.5025,
      "step": 2872
    },
    {
      "epoch": 0.48993860845839016,
      "grad_norm": 0.4060668647289276,
      "learning_rate": 1.0816522649380384e-06,
      "loss": 0.5188,
      "step": 2873
    },
    {
      "epoch": 0.4901091405184175,
      "grad_norm": 0.4716636836528778,
      "learning_rate": 1.0811017773894781e-06,
      "loss": 0.4979,
      "step": 2874
    },
    {
      "epoch": 0.49027967257844474,
      "grad_norm": 0.4533891975879669,
      "learning_rate": 1.080551265100278e-06,
      "loss": 0.5021,
      "step": 2875
    },
    {
      "epoch": 0.49045020463847205,
      "grad_norm": 0.35617199540138245,
      "learning_rate": 1.080000728238375e-06,
      "loss": 0.5142,
      "step": 2876
    },
    {
      "epoch": 0.4906207366984993,
      "grad_norm": 0.3525446951389313,
      "learning_rate": 1.0794501669717146e-06,
      "loss": 0.5241,
      "step": 2877
    },
    {
      "epoch": 0.4907912687585266,
      "grad_norm": 0.38873711228370667,
      "learning_rate": 1.0788995814682487e-06,
      "loss": 0.5047,
      "step": 2878
    },
    {
      "epoch": 0.4909618008185539,
      "grad_norm": 0.3586026430130005,
      "learning_rate": 1.0783489718959379e-06,
      "loss": 0.5123,
      "step": 2879
    },
    {
      "epoch": 0.49113233287858116,
      "grad_norm": 0.4532160460948944,
      "learning_rate": 1.077798338422748e-06,
      "loss": 0.5145,
      "step": 2880
    },
    {
      "epoch": 0.4913028649386085,
      "grad_norm": 0.4291527569293976,
      "learning_rate": 1.0772476812166546e-06,
      "loss": 0.5198,
      "step": 2881
    },
    {
      "epoch": 0.49147339699863574,
      "grad_norm": 0.41730403900146484,
      "learning_rate": 1.076697000445639e-06,
      "loss": 0.5257,
      "step": 2882
    },
    {
      "epoch": 0.49164392905866305,
      "grad_norm": 0.49823060631752014,
      "learning_rate": 1.0761462962776898e-06,
      "loss": 0.5069,
      "step": 2883
    },
    {
      "epoch": 0.4918144611186903,
      "grad_norm": 0.635140061378479,
      "learning_rate": 1.0755955688808037e-06,
      "loss": 0.5228,
      "step": 2884
    },
    {
      "epoch": 0.4919849931787176,
      "grad_norm": 0.5624722242355347,
      "learning_rate": 1.075044818422983e-06,
      "loss": 0.526,
      "step": 2885
    },
    {
      "epoch": 0.4921555252387449,
      "grad_norm": 0.6729494333267212,
      "learning_rate": 1.0744940450722379e-06,
      "loss": 0.5223,
      "step": 2886
    },
    {
      "epoch": 0.49232605729877216,
      "grad_norm": 0.6167219877243042,
      "learning_rate": 1.073943248996586e-06,
      "loss": 0.5068,
      "step": 2887
    },
    {
      "epoch": 0.4924965893587995,
      "grad_norm": 0.45822858810424805,
      "learning_rate": 1.073392430364051e-06,
      "loss": 0.5097,
      "step": 2888
    },
    {
      "epoch": 0.49266712141882674,
      "grad_norm": 0.7486544847488403,
      "learning_rate": 1.0728415893426637e-06,
      "loss": 0.502,
      "step": 2889
    },
    {
      "epoch": 0.492837653478854,
      "grad_norm": 0.4141033887863159,
      "learning_rate": 1.0722907261004618e-06,
      "loss": 0.5138,
      "step": 2890
    },
    {
      "epoch": 0.4930081855388813,
      "grad_norm": 0.6413578987121582,
      "learning_rate": 1.0717398408054906e-06,
      "loss": 0.5168,
      "step": 2891
    },
    {
      "epoch": 0.4931787175989086,
      "grad_norm": 0.49279406666755676,
      "learning_rate": 1.0711889336258004e-06,
      "loss": 0.5076,
      "step": 2892
    },
    {
      "epoch": 0.4933492496589359,
      "grad_norm": 0.4992629587650299,
      "learning_rate": 1.0706380047294497e-06,
      "loss": 0.5015,
      "step": 2893
    },
    {
      "epoch": 0.49351978171896316,
      "grad_norm": 0.506038248538971,
      "learning_rate": 1.0700870542845027e-06,
      "loss": 0.5209,
      "step": 2894
    },
    {
      "epoch": 0.4936903137789905,
      "grad_norm": 0.48432669043540955,
      "learning_rate": 1.0695360824590306e-06,
      "loss": 0.5105,
      "step": 2895
    },
    {
      "epoch": 0.49386084583901774,
      "grad_norm": 0.6199078559875488,
      "learning_rate": 1.068985089421111e-06,
      "loss": 0.5154,
      "step": 2896
    },
    {
      "epoch": 0.494031377899045,
      "grad_norm": 0.4056002199649811,
      "learning_rate": 1.0684340753388283e-06,
      "loss": 0.5193,
      "step": 2897
    },
    {
      "epoch": 0.4942019099590723,
      "grad_norm": 0.5460501313209534,
      "learning_rate": 1.0678830403802728e-06,
      "loss": 0.509,
      "step": 2898
    },
    {
      "epoch": 0.4943724420190996,
      "grad_norm": 0.4023360013961792,
      "learning_rate": 1.067331984713541e-06,
      "loss": 0.5092,
      "step": 2899
    },
    {
      "epoch": 0.4945429740791269,
      "grad_norm": 0.42605215311050415,
      "learning_rate": 1.0667809085067376e-06,
      "loss": 0.523,
      "step": 2900
    },
    {
      "epoch": 0.49471350613915416,
      "grad_norm": 0.4854925572872162,
      "learning_rate": 1.0662298119279702e-06,
      "loss": 0.515,
      "step": 2901
    },
    {
      "epoch": 0.4948840381991814,
      "grad_norm": 0.3809516131877899,
      "learning_rate": 1.0656786951453556e-06,
      "loss": 0.5149,
      "step": 2902
    },
    {
      "epoch": 0.49505457025920874,
      "grad_norm": 0.4749661087989807,
      "learning_rate": 1.0651275583270155e-06,
      "loss": 0.4991,
      "step": 2903
    },
    {
      "epoch": 0.495225102319236,
      "grad_norm": 0.35364648699760437,
      "learning_rate": 1.0645764016410777e-06,
      "loss": 0.5192,
      "step": 2904
    },
    {
      "epoch": 0.4953956343792633,
      "grad_norm": 0.51173996925354,
      "learning_rate": 1.064025225255676e-06,
      "loss": 0.5171,
      "step": 2905
    },
    {
      "epoch": 0.4955661664392906,
      "grad_norm": 0.47044792771339417,
      "learning_rate": 1.0634740293389512e-06,
      "loss": 0.5209,
      "step": 2906
    },
    {
      "epoch": 0.4957366984993179,
      "grad_norm": 0.3833619952201843,
      "learning_rate": 1.0629228140590488e-06,
      "loss": 0.509,
      "step": 2907
    },
    {
      "epoch": 0.49590723055934516,
      "grad_norm": 0.49732497334480286,
      "learning_rate": 1.0623715795841208e-06,
      "loss": 0.5132,
      "step": 2908
    },
    {
      "epoch": 0.4960777626193724,
      "grad_norm": 0.3693459630012512,
      "learning_rate": 1.061820326082325e-06,
      "loss": 0.5199,
      "step": 2909
    },
    {
      "epoch": 0.49624829467939974,
      "grad_norm": 0.4614149034023285,
      "learning_rate": 1.0612690537218247e-06,
      "loss": 0.5199,
      "step": 2910
    },
    {
      "epoch": 0.496418826739427,
      "grad_norm": 0.4367601275444031,
      "learning_rate": 1.0607177626707894e-06,
      "loss": 0.5109,
      "step": 2911
    },
    {
      "epoch": 0.4965893587994543,
      "grad_norm": 0.4477463960647583,
      "learning_rate": 1.0601664530973946e-06,
      "loss": 0.5001,
      "step": 2912
    },
    {
      "epoch": 0.4967598908594816,
      "grad_norm": 0.44696980714797974,
      "learning_rate": 1.05961512516982e-06,
      "loss": 0.5143,
      "step": 2913
    },
    {
      "epoch": 0.49693042291950884,
      "grad_norm": 0.3601173460483551,
      "learning_rate": 1.0590637790562527e-06,
      "loss": 0.5097,
      "step": 2914
    },
    {
      "epoch": 0.49710095497953616,
      "grad_norm": 0.49771782755851746,
      "learning_rate": 1.0585124149248842e-06,
      "loss": 0.5046,
      "step": 2915
    },
    {
      "epoch": 0.4972714870395634,
      "grad_norm": 0.4949286878108978,
      "learning_rate": 1.057961032943912e-06,
      "loss": 0.528,
      "step": 2916
    },
    {
      "epoch": 0.49744201909959074,
      "grad_norm": 0.37806761264801025,
      "learning_rate": 1.0574096332815389e-06,
      "loss": 0.5117,
      "step": 2917
    },
    {
      "epoch": 0.497612551159618,
      "grad_norm": 0.574327290058136,
      "learning_rate": 1.0568582161059725e-06,
      "loss": 0.5078,
      "step": 2918
    },
    {
      "epoch": 0.4977830832196453,
      "grad_norm": 0.38559508323669434,
      "learning_rate": 1.0563067815854267e-06,
      "loss": 0.5031,
      "step": 2919
    },
    {
      "epoch": 0.4979536152796726,
      "grad_norm": 0.5216695666313171,
      "learning_rate": 1.0557553298881204e-06,
      "loss": 0.5249,
      "step": 2920
    },
    {
      "epoch": 0.49812414733969984,
      "grad_norm": 0.5521080493927002,
      "learning_rate": 1.0552038611822775e-06,
      "loss": 0.5216,
      "step": 2921
    },
    {
      "epoch": 0.49829467939972716,
      "grad_norm": 0.4828493893146515,
      "learning_rate": 1.0546523756361272e-06,
      "loss": 0.5181,
      "step": 2922
    },
    {
      "epoch": 0.4984652114597544,
      "grad_norm": 0.5427225828170776,
      "learning_rate": 1.054100873417904e-06,
      "loss": 0.5269,
      "step": 2923
    },
    {
      "epoch": 0.49863574351978174,
      "grad_norm": 0.48165130615234375,
      "learning_rate": 1.0535493546958471e-06,
      "loss": 0.523,
      "step": 2924
    },
    {
      "epoch": 0.498806275579809,
      "grad_norm": 0.41212841868400574,
      "learning_rate": 1.0529978196382012e-06,
      "loss": 0.5248,
      "step": 2925
    },
    {
      "epoch": 0.49897680763983626,
      "grad_norm": 0.5029284954071045,
      "learning_rate": 1.0524462684132157e-06,
      "loss": 0.5132,
      "step": 2926
    },
    {
      "epoch": 0.4991473396998636,
      "grad_norm": 0.4131450355052948,
      "learning_rate": 1.0518947011891448e-06,
      "loss": 0.5092,
      "step": 2927
    },
    {
      "epoch": 0.49931787175989084,
      "grad_norm": 0.44756412506103516,
      "learning_rate": 1.051343118134248e-06,
      "loss": 0.5216,
      "step": 2928
    },
    {
      "epoch": 0.49948840381991816,
      "grad_norm": 0.4106735289096832,
      "learning_rate": 1.0507915194167893e-06,
      "loss": 0.5198,
      "step": 2929
    },
    {
      "epoch": 0.4996589358799454,
      "grad_norm": 0.5048739314079285,
      "learning_rate": 1.0502399052050378e-06,
      "loss": 0.5186,
      "step": 2930
    },
    {
      "epoch": 0.49982946793997274,
      "grad_norm": 0.547248125076294,
      "learning_rate": 1.0496882756672667e-06,
      "loss": 0.5242,
      "step": 2931
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41109588742256165,
      "learning_rate": 1.0491366309717549e-06,
      "loss": 0.5322,
      "step": 2932
    },
    {
      "epoch": 0.5001705320600273,
      "grad_norm": 0.4631933271884918,
      "learning_rate": 1.0485849712867844e-06,
      "loss": 0.5025,
      "step": 2933
    },
    {
      "epoch": 0.5003410641200545,
      "grad_norm": 0.3782828748226166,
      "learning_rate": 1.0480332967806436e-06,
      "loss": 0.5161,
      "step": 2934
    },
    {
      "epoch": 0.5005115961800819,
      "grad_norm": 0.43556368350982666,
      "learning_rate": 1.0474816076216237e-06,
      "loss": 0.53,
      "step": 2935
    },
    {
      "epoch": 0.5006821282401092,
      "grad_norm": 0.4012020230293274,
      "learning_rate": 1.0469299039780218e-06,
      "loss": 0.5137,
      "step": 2936
    },
    {
      "epoch": 0.5008526603001364,
      "grad_norm": 0.4638964831829071,
      "learning_rate": 1.0463781860181387e-06,
      "loss": 0.5277,
      "step": 2937
    },
    {
      "epoch": 0.5010231923601637,
      "grad_norm": 0.4636898636817932,
      "learning_rate": 1.0458264539102791e-06,
      "loss": 0.5314,
      "step": 2938
    },
    {
      "epoch": 0.501193724420191,
      "grad_norm": 0.363491415977478,
      "learning_rate": 1.0452747078227536e-06,
      "loss": 0.5171,
      "step": 2939
    },
    {
      "epoch": 0.5013642564802183,
      "grad_norm": 0.4326697885990143,
      "learning_rate": 1.044722947923875e-06,
      "loss": 0.5218,
      "step": 2940
    },
    {
      "epoch": 0.5015347885402456,
      "grad_norm": 0.5075788497924805,
      "learning_rate": 1.0441711743819618e-06,
      "loss": 0.5256,
      "step": 2941
    },
    {
      "epoch": 0.5017053206002728,
      "grad_norm": 0.6462679505348206,
      "learning_rate": 1.0436193873653363e-06,
      "loss": 0.534,
      "step": 2942
    },
    {
      "epoch": 0.5018758526603001,
      "grad_norm": 0.5000764727592468,
      "learning_rate": 1.0430675870423246e-06,
      "loss": 0.528,
      "step": 2943
    },
    {
      "epoch": 0.5020463847203275,
      "grad_norm": 0.45578262209892273,
      "learning_rate": 1.0425157735812572e-06,
      "loss": 0.5266,
      "step": 2944
    },
    {
      "epoch": 0.5022169167803547,
      "grad_norm": 0.6543422341346741,
      "learning_rate": 1.0419639471504683e-06,
      "loss": 0.5154,
      "step": 2945
    },
    {
      "epoch": 0.502387448840382,
      "grad_norm": 0.6309098601341248,
      "learning_rate": 1.0414121079182969e-06,
      "loss": 0.5236,
      "step": 2946
    },
    {
      "epoch": 0.5025579809004093,
      "grad_norm": 0.5663894414901733,
      "learning_rate": 1.0408602560530845e-06,
      "loss": 0.523,
      "step": 2947
    },
    {
      "epoch": 0.5027285129604365,
      "grad_norm": 0.454391211271286,
      "learning_rate": 1.0403083917231775e-06,
      "loss": 0.527,
      "step": 2948
    },
    {
      "epoch": 0.5028990450204639,
      "grad_norm": 0.4640673100948334,
      "learning_rate": 1.039756515096926e-06,
      "loss": 0.5185,
      "step": 2949
    },
    {
      "epoch": 0.5030695770804912,
      "grad_norm": 0.3818853795528412,
      "learning_rate": 1.0392046263426833e-06,
      "loss": 0.5175,
      "step": 2950
    },
    {
      "epoch": 0.5032401091405184,
      "grad_norm": 0.43370091915130615,
      "learning_rate": 1.0386527256288066e-06,
      "loss": 0.51,
      "step": 2951
    },
    {
      "epoch": 0.5034106412005457,
      "grad_norm": 0.49242711067199707,
      "learning_rate": 1.0381008131236576e-06,
      "loss": 0.5152,
      "step": 2952
    },
    {
      "epoch": 0.503581173260573,
      "grad_norm": 0.41316694021224976,
      "learning_rate": 1.0375488889956003e-06,
      "loss": 0.5147,
      "step": 2953
    },
    {
      "epoch": 0.5037517053206003,
      "grad_norm": 0.4294221103191376,
      "learning_rate": 1.0369969534130031e-06,
      "loss": 0.5175,
      "step": 2954
    },
    {
      "epoch": 0.5039222373806276,
      "grad_norm": 0.3391348421573639,
      "learning_rate": 1.0364450065442379e-06,
      "loss": 0.5031,
      "step": 2955
    },
    {
      "epoch": 0.5040927694406548,
      "grad_norm": 0.6170563697814941,
      "learning_rate": 1.0358930485576794e-06,
      "loss": 0.5033,
      "step": 2956
    },
    {
      "epoch": 0.5042633015006821,
      "grad_norm": 0.5732797980308533,
      "learning_rate": 1.035341079621706e-06,
      "loss": 0.5173,
      "step": 2957
    },
    {
      "epoch": 0.5044338335607094,
      "grad_norm": 0.3765585422515869,
      "learning_rate": 1.0347890999047e-06,
      "loss": 0.5338,
      "step": 2958
    },
    {
      "epoch": 0.5046043656207367,
      "grad_norm": 0.5756282806396484,
      "learning_rate": 1.0342371095750458e-06,
      "loss": 0.5016,
      "step": 2959
    },
    {
      "epoch": 0.504774897680764,
      "grad_norm": 0.5575646758079529,
      "learning_rate": 1.0336851088011322e-06,
      "loss": 0.5021,
      "step": 2960
    },
    {
      "epoch": 0.5049454297407913,
      "grad_norm": 0.4088723957538605,
      "learning_rate": 1.033133097751351e-06,
      "loss": 0.5094,
      "step": 2961
    },
    {
      "epoch": 0.5051159618008185,
      "grad_norm": 0.6527062058448792,
      "learning_rate": 1.0325810765940962e-06,
      "loss": 0.5093,
      "step": 2962
    },
    {
      "epoch": 0.5052864938608458,
      "grad_norm": 0.591107964515686,
      "learning_rate": 1.0320290454977662e-06,
      "loss": 0.5152,
      "step": 2963
    },
    {
      "epoch": 0.5054570259208732,
      "grad_norm": 0.5586540102958679,
      "learning_rate": 1.0314770046307613e-06,
      "loss": 0.5035,
      "step": 2964
    },
    {
      "epoch": 0.5056275579809004,
      "grad_norm": 0.49104535579681396,
      "learning_rate": 1.0309249541614854e-06,
      "loss": 0.5141,
      "step": 2965
    },
    {
      "epoch": 0.5057980900409277,
      "grad_norm": 0.6136547327041626,
      "learning_rate": 1.0303728942583453e-06,
      "loss": 0.5035,
      "step": 2966
    },
    {
      "epoch": 0.505968622100955,
      "grad_norm": 0.5435359477996826,
      "learning_rate": 1.0298208250897505e-06,
      "loss": 0.5019,
      "step": 2967
    },
    {
      "epoch": 0.5061391541609823,
      "grad_norm": 0.6272081732749939,
      "learning_rate": 1.0292687468241137e-06,
      "loss": 0.5139,
      "step": 2968
    },
    {
      "epoch": 0.5063096862210096,
      "grad_norm": 0.5421731472015381,
      "learning_rate": 1.0287166596298492e-06,
      "loss": 0.5028,
      "step": 2969
    },
    {
      "epoch": 0.5064802182810368,
      "grad_norm": 0.5001494884490967,
      "learning_rate": 1.028164563675376e-06,
      "loss": 0.5092,
      "step": 2970
    },
    {
      "epoch": 0.5066507503410641,
      "grad_norm": 0.6292235255241394,
      "learning_rate": 1.027612459129114e-06,
      "loss": 0.5036,
      "step": 2971
    },
    {
      "epoch": 0.5068212824010914,
      "grad_norm": 0.37526097893714905,
      "learning_rate": 1.0270603461594862e-06,
      "loss": 0.516,
      "step": 2972
    },
    {
      "epoch": 0.5069918144611187,
      "grad_norm": 0.6769919395446777,
      "learning_rate": 1.0265082249349188e-06,
      "loss": 0.5046,
      "step": 2973
    },
    {
      "epoch": 0.507162346521146,
      "grad_norm": 0.39580002427101135,
      "learning_rate": 1.02595609562384e-06,
      "loss": 0.5182,
      "step": 2974
    },
    {
      "epoch": 0.5073328785811733,
      "grad_norm": 0.6619483232498169,
      "learning_rate": 1.0254039583946804e-06,
      "loss": 0.5132,
      "step": 2975
    },
    {
      "epoch": 0.5075034106412005,
      "grad_norm": 0.5659598708152771,
      "learning_rate": 1.0248518134158732e-06,
      "loss": 0.503,
      "step": 2976
    },
    {
      "epoch": 0.5076739427012278,
      "grad_norm": 0.5417669415473938,
      "learning_rate": 1.0242996608558543e-06,
      "loss": 0.5119,
      "step": 2977
    },
    {
      "epoch": 0.5078444747612552,
      "grad_norm": 0.5509529709815979,
      "learning_rate": 1.0237475008830608e-06,
      "loss": 0.5094,
      "step": 2978
    },
    {
      "epoch": 0.5080150068212824,
      "grad_norm": 0.4370351731777191,
      "learning_rate": 1.0231953336659336e-06,
      "loss": 0.5269,
      "step": 2979
    },
    {
      "epoch": 0.5081855388813097,
      "grad_norm": 0.5920512080192566,
      "learning_rate": 1.0226431593729146e-06,
      "loss": 0.5026,
      "step": 2980
    },
    {
      "epoch": 0.508356070941337,
      "grad_norm": 0.4498980939388275,
      "learning_rate": 1.0220909781724479e-06,
      "loss": 0.5119,
      "step": 2981
    },
    {
      "epoch": 0.5085266030013642,
      "grad_norm": 0.654874861240387,
      "learning_rate": 1.0215387902329806e-06,
      "loss": 0.5116,
      "step": 2982
    },
    {
      "epoch": 0.5086971350613916,
      "grad_norm": 0.4357649087905884,
      "learning_rate": 1.0209865957229613e-06,
      "loss": 0.5143,
      "step": 2983
    },
    {
      "epoch": 0.5088676671214188,
      "grad_norm": 0.5545060038566589,
      "learning_rate": 1.0204343948108403e-06,
      "loss": 0.5138,
      "step": 2984
    },
    {
      "epoch": 0.5090381991814461,
      "grad_norm": 0.552129328250885,
      "learning_rate": 1.0198821876650702e-06,
      "loss": 0.5121,
      "step": 2985
    },
    {
      "epoch": 0.5092087312414734,
      "grad_norm": 0.6152744889259338,
      "learning_rate": 1.0193299744541065e-06,
      "loss": 0.5171,
      "step": 2986
    },
    {
      "epoch": 0.5093792633015006,
      "grad_norm": 0.42170122265815735,
      "learning_rate": 1.0187777553464038e-06,
      "loss": 0.5061,
      "step": 2987
    },
    {
      "epoch": 0.509549795361528,
      "grad_norm": 0.5506285429000854,
      "learning_rate": 1.0182255305104215e-06,
      "loss": 0.5162,
      "step": 2988
    },
    {
      "epoch": 0.5097203274215553,
      "grad_norm": 0.5527845025062561,
      "learning_rate": 1.0176733001146192e-06,
      "loss": 0.511,
      "step": 2989
    },
    {
      "epoch": 0.5098908594815825,
      "grad_norm": 0.43512266874313354,
      "learning_rate": 1.0171210643274582e-06,
      "loss": 0.4988,
      "step": 2990
    },
    {
      "epoch": 0.5100613915416098,
      "grad_norm": 0.4943109452724457,
      "learning_rate": 1.016568823317402e-06,
      "loss": 0.5137,
      "step": 2991
    },
    {
      "epoch": 0.5102319236016372,
      "grad_norm": 0.3899883031845093,
      "learning_rate": 1.0160165772529151e-06,
      "loss": 0.5275,
      "step": 2992
    },
    {
      "epoch": 0.5104024556616644,
      "grad_norm": 0.45650139451026917,
      "learning_rate": 1.0154643263024645e-06,
      "loss": 0.5107,
      "step": 2993
    },
    {
      "epoch": 0.5105729877216917,
      "grad_norm": 0.38634830713272095,
      "learning_rate": 1.0149120706345176e-06,
      "loss": 0.5121,
      "step": 2994
    },
    {
      "epoch": 0.510743519781719,
      "grad_norm": 0.46154528856277466,
      "learning_rate": 1.0143598104175435e-06,
      "loss": 0.5104,
      "step": 2995
    },
    {
      "epoch": 0.5109140518417462,
      "grad_norm": 0.5653091073036194,
      "learning_rate": 1.0138075458200133e-06,
      "loss": 0.5148,
      "step": 2996
    },
    {
      "epoch": 0.5110845839017736,
      "grad_norm": 0.47336509823799133,
      "learning_rate": 1.0132552770103988e-06,
      "loss": 0.5131,
      "step": 2997
    },
    {
      "epoch": 0.5112551159618008,
      "grad_norm": 0.4555085599422455,
      "learning_rate": 1.0127030041571733e-06,
      "loss": 0.5052,
      "step": 2998
    },
    {
      "epoch": 0.5114256480218281,
      "grad_norm": 0.395914226770401,
      "learning_rate": 1.0121507274288112e-06,
      "loss": 0.5144,
      "step": 2999
    },
    {
      "epoch": 0.5115961800818554,
      "grad_norm": 0.466209352016449,
      "learning_rate": 1.0115984469937885e-06,
      "loss": 0.5233,
      "step": 3000
    },
    {
      "epoch": 0.5117667121418826,
      "grad_norm": 0.4672727882862091,
      "learning_rate": 1.011046163020582e-06,
      "loss": 0.5094,
      "step": 3001
    },
    {
      "epoch": 0.51193724420191,
      "grad_norm": 0.38614600896835327,
      "learning_rate": 1.0104938756776696e-06,
      "loss": 0.5152,
      "step": 3002
    },
    {
      "epoch": 0.5121077762619373,
      "grad_norm": 0.4924212396144867,
      "learning_rate": 1.00994158513353e-06,
      "loss": 0.5141,
      "step": 3003
    },
    {
      "epoch": 0.5122783083219645,
      "grad_norm": 0.4234980642795563,
      "learning_rate": 1.009389291556643e-06,
      "loss": 0.5138,
      "step": 3004
    },
    {
      "epoch": 0.5124488403819918,
      "grad_norm": 0.5439965128898621,
      "learning_rate": 1.0088369951154904e-06,
      "loss": 0.5159,
      "step": 3005
    },
    {
      "epoch": 0.512619372442019,
      "grad_norm": 0.4107781648635864,
      "learning_rate": 1.0082846959785528e-06,
      "loss": 0.5081,
      "step": 3006
    },
    {
      "epoch": 0.5127899045020464,
      "grad_norm": 0.4752189517021179,
      "learning_rate": 1.0077323943143135e-06,
      "loss": 0.5137,
      "step": 3007
    },
    {
      "epoch": 0.5129604365620737,
      "grad_norm": 0.43996524810791016,
      "learning_rate": 1.0071800902912555e-06,
      "loss": 0.5139,
      "step": 3008
    },
    {
      "epoch": 0.513130968622101,
      "grad_norm": 0.34770911931991577,
      "learning_rate": 1.0066277840778627e-06,
      "loss": 0.515,
      "step": 3009
    },
    {
      "epoch": 0.5133015006821282,
      "grad_norm": 0.5615414977073669,
      "learning_rate": 1.0060754758426204e-06,
      "loss": 0.5093,
      "step": 3010
    },
    {
      "epoch": 0.5134720327421555,
      "grad_norm": 0.5643600225448608,
      "learning_rate": 1.0055231657540133e-06,
      "loss": 0.5072,
      "step": 3011
    },
    {
      "epoch": 0.5136425648021828,
      "grad_norm": 0.4654306173324585,
      "learning_rate": 1.0049708539805274e-06,
      "loss": 0.5197,
      "step": 3012
    },
    {
      "epoch": 0.5138130968622101,
      "grad_norm": 0.4522657096385956,
      "learning_rate": 1.0044185406906493e-06,
      "loss": 0.5174,
      "step": 3013
    },
    {
      "epoch": 0.5139836289222374,
      "grad_norm": 0.4825378656387329,
      "learning_rate": 1.003866226052866e-06,
      "loss": 0.5215,
      "step": 3014
    },
    {
      "epoch": 0.5141541609822646,
      "grad_norm": 0.40646103024482727,
      "learning_rate": 1.0033139102356643e-06,
      "loss": 0.5123,
      "step": 3015
    },
    {
      "epoch": 0.514324693042292,
      "grad_norm": 0.35144925117492676,
      "learning_rate": 1.0027615934075322e-06,
      "loss": 0.5184,
      "step": 3016
    },
    {
      "epoch": 0.5144952251023193,
      "grad_norm": 0.46346837282180786,
      "learning_rate": 1.0022092757369576e-06,
      "loss": 0.5117,
      "step": 3017
    },
    {
      "epoch": 0.5146657571623465,
      "grad_norm": 0.42966172099113464,
      "learning_rate": 1.0016569573924288e-06,
      "loss": 0.5138,
      "step": 3018
    },
    {
      "epoch": 0.5148362892223738,
      "grad_norm": 0.37160927057266235,
      "learning_rate": 1.0011046385424335e-06,
      "loss": 0.5076,
      "step": 3019
    },
    {
      "epoch": 0.515006821282401,
      "grad_norm": 0.5897586941719055,
      "learning_rate": 1.0005523193554615e-06,
      "loss": 0.5164,
      "step": 3020
    },
    {
      "epoch": 0.5151773533424284,
      "grad_norm": 0.45426005125045776,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.5091,
      "step": 3021
    },
    {
      "epoch": 0.5153478854024557,
      "grad_norm": 0.415068656206131,
      "learning_rate": 9.994476806445394e-07,
      "loss": 0.5107,
      "step": 3022
    },
    {
      "epoch": 0.515518417462483,
      "grad_norm": 0.5580230951309204,
      "learning_rate": 9.988953614575669e-07,
      "loss": 0.5052,
      "step": 3023
    },
    {
      "epoch": 0.5156889495225102,
      "grad_norm": 0.4167662560939789,
      "learning_rate": 9.98343042607572e-07,
      "loss": 0.5057,
      "step": 3024
    },
    {
      "epoch": 0.5158594815825375,
      "grad_norm": 0.41590461134910583,
      "learning_rate": 9.977907242630427e-07,
      "loss": 0.5214,
      "step": 3025
    },
    {
      "epoch": 0.5160300136425648,
      "grad_norm": 0.4774099588394165,
      "learning_rate": 9.972384065924684e-07,
      "loss": 0.5142,
      "step": 3026
    },
    {
      "epoch": 0.5162005457025921,
      "grad_norm": 0.34568390250205994,
      "learning_rate": 9.96686089764336e-07,
      "loss": 0.508,
      "step": 3027
    },
    {
      "epoch": 0.5163710777626194,
      "grad_norm": 0.4375649094581604,
      "learning_rate": 9.961337739471345e-07,
      "loss": 0.5079,
      "step": 3028
    },
    {
      "epoch": 0.5165416098226466,
      "grad_norm": 0.35460013151168823,
      "learning_rate": 9.95581459309351e-07,
      "loss": 0.524,
      "step": 3029
    },
    {
      "epoch": 0.5167121418826739,
      "grad_norm": 0.42628592252731323,
      "learning_rate": 9.95029146019473e-07,
      "loss": 0.5116,
      "step": 3030
    },
    {
      "epoch": 0.5168826739427013,
      "grad_norm": 0.40556424856185913,
      "learning_rate": 9.944768342459873e-07,
      "loss": 0.5251,
      "step": 3031
    },
    {
      "epoch": 0.5170532060027285,
      "grad_norm": 0.36979591846466064,
      "learning_rate": 9.939245241573801e-07,
      "loss": 0.5207,
      "step": 3032
    },
    {
      "epoch": 0.5172237380627558,
      "grad_norm": 0.4147460162639618,
      "learning_rate": 9.933722159221377e-07,
      "loss": 0.5219,
      "step": 3033
    },
    {
      "epoch": 0.517394270122783,
      "grad_norm": 0.40708544850349426,
      "learning_rate": 9.928199097087448e-07,
      "loss": 0.5118,
      "step": 3034
    },
    {
      "epoch": 0.5175648021828103,
      "grad_norm": 0.37231746315956116,
      "learning_rate": 9.92267605685687e-07,
      "loss": 0.511,
      "step": 3035
    },
    {
      "epoch": 0.5177353342428377,
      "grad_norm": 0.3565663695335388,
      "learning_rate": 9.917153040214475e-07,
      "loss": 0.5111,
      "step": 3036
    },
    {
      "epoch": 0.517905866302865,
      "grad_norm": 0.36910760402679443,
      "learning_rate": 9.911630048845101e-07,
      "loss": 0.5102,
      "step": 3037
    },
    {
      "epoch": 0.5180763983628922,
      "grad_norm": 0.3910926580429077,
      "learning_rate": 9.906107084433573e-07,
      "loss": 0.4988,
      "step": 3038
    },
    {
      "epoch": 0.5182469304229195,
      "grad_norm": 0.47788098454475403,
      "learning_rate": 9.900584148664706e-07,
      "loss": 0.5046,
      "step": 3039
    },
    {
      "epoch": 0.5184174624829468,
      "grad_norm": 0.4352160096168518,
      "learning_rate": 9.895061243223312e-07,
      "loss": 0.5179,
      "step": 3040
    },
    {
      "epoch": 0.5185879945429741,
      "grad_norm": 0.4005039632320404,
      "learning_rate": 9.889538369794185e-07,
      "loss": 0.5084,
      "step": 3041
    },
    {
      "epoch": 0.5187585266030014,
      "grad_norm": 0.3612534999847412,
      "learning_rate": 9.884015530062119e-07,
      "loss": 0.5088,
      "step": 3042
    },
    {
      "epoch": 0.5189290586630286,
      "grad_norm": 0.49642056226730347,
      "learning_rate": 9.87849272571189e-07,
      "loss": 0.5052,
      "step": 3043
    },
    {
      "epoch": 0.5190995907230559,
      "grad_norm": 0.3578031063079834,
      "learning_rate": 9.87296995842827e-07,
      "loss": 0.5041,
      "step": 3044
    },
    {
      "epoch": 0.5192701227830833,
      "grad_norm": 0.4707255959510803,
      "learning_rate": 9.86744722989602e-07,
      "loss": 0.5136,
      "step": 3045
    },
    {
      "epoch": 0.5194406548431105,
      "grad_norm": 0.46732622385025024,
      "learning_rate": 9.86192454179987e-07,
      "loss": 0.4998,
      "step": 3046
    },
    {
      "epoch": 0.5196111869031378,
      "grad_norm": 0.36986300349235535,
      "learning_rate": 9.85640189582457e-07,
      "loss": 0.5026,
      "step": 3047
    },
    {
      "epoch": 0.519781718963165,
      "grad_norm": 0.39670634269714355,
      "learning_rate": 9.85087929365483e-07,
      "loss": 0.5172,
      "step": 3048
    },
    {
      "epoch": 0.5199522510231923,
      "grad_norm": 0.46700161695480347,
      "learning_rate": 9.84535673697536e-07,
      "loss": 0.5174,
      "step": 3049
    },
    {
      "epoch": 0.5201227830832197,
      "grad_norm": 0.5552010536193848,
      "learning_rate": 9.839834227470852e-07,
      "loss": 0.5088,
      "step": 3050
    },
    {
      "epoch": 0.520293315143247,
      "grad_norm": 0.4279073476791382,
      "learning_rate": 9.834311766825986e-07,
      "loss": 0.5052,
      "step": 3051
    },
    {
      "epoch": 0.5204638472032742,
      "grad_norm": 0.3493303656578064,
      "learning_rate": 9.828789356725421e-07,
      "loss": 0.5204,
      "step": 3052
    },
    {
      "epoch": 0.5206343792633015,
      "grad_norm": 0.4930592477321625,
      "learning_rate": 9.823266998853814e-07,
      "loss": 0.5065,
      "step": 3053
    },
    {
      "epoch": 0.5208049113233287,
      "grad_norm": 0.46885567903518677,
      "learning_rate": 9.817744694895792e-07,
      "loss": 0.5048,
      "step": 3054
    },
    {
      "epoch": 0.5209754433833561,
      "grad_norm": 0.42892614006996155,
      "learning_rate": 9.812222446535965e-07,
      "loss": 0.5076,
      "step": 3055
    },
    {
      "epoch": 0.5211459754433834,
      "grad_norm": 0.4651287794113159,
      "learning_rate": 9.806700255458943e-07,
      "loss": 0.5122,
      "step": 3056
    },
    {
      "epoch": 0.5213165075034106,
      "grad_norm": 0.4076344668865204,
      "learning_rate": 9.8011781233493e-07,
      "loss": 0.5085,
      "step": 3057
    },
    {
      "epoch": 0.5214870395634379,
      "grad_norm": 0.6352848410606384,
      "learning_rate": 9.795656051891602e-07,
      "loss": 0.5072,
      "step": 3058
    },
    {
      "epoch": 0.5216575716234653,
      "grad_norm": 0.4328773319721222,
      "learning_rate": 9.790134042770393e-07,
      "loss": 0.5162,
      "step": 3059
    },
    {
      "epoch": 0.5218281036834925,
      "grad_norm": 0.3829416036605835,
      "learning_rate": 9.7846120976702e-07,
      "loss": 0.5197,
      "step": 3060
    },
    {
      "epoch": 0.5219986357435198,
      "grad_norm": 0.3622390925884247,
      "learning_rate": 9.779090218275527e-07,
      "loss": 0.5038,
      "step": 3061
    },
    {
      "epoch": 0.522169167803547,
      "grad_norm": 0.38840627670288086,
      "learning_rate": 9.77356840627086e-07,
      "loss": 0.5145,
      "step": 3062
    },
    {
      "epoch": 0.5223396998635743,
      "grad_norm": 0.41121238470077515,
      "learning_rate": 9.768046663340671e-07,
      "loss": 0.5084,
      "step": 3063
    },
    {
      "epoch": 0.5225102319236017,
      "grad_norm": 0.3751465082168579,
      "learning_rate": 9.762524991169393e-07,
      "loss": 0.5109,
      "step": 3064
    },
    {
      "epoch": 0.522680763983629,
      "grad_norm": 0.49738723039627075,
      "learning_rate": 9.757003391441463e-07,
      "loss": 0.4974,
      "step": 3065
    },
    {
      "epoch": 0.5228512960436562,
      "grad_norm": 0.43950966000556946,
      "learning_rate": 9.75148186584127e-07,
      "loss": 0.5168,
      "step": 3066
    },
    {
      "epoch": 0.5230218281036835,
      "grad_norm": 0.4372667372226715,
      "learning_rate": 9.745960416053201e-07,
      "loss": 0.5041,
      "step": 3067
    },
    {
      "epoch": 0.5231923601637107,
      "grad_norm": 0.46111032366752625,
      "learning_rate": 9.740439043761605e-07,
      "loss": 0.5105,
      "step": 3068
    },
    {
      "epoch": 0.5233628922237381,
      "grad_norm": 0.5779291391372681,
      "learning_rate": 9.734917750650818e-07,
      "loss": 0.5112,
      "step": 3069
    },
    {
      "epoch": 0.5235334242837654,
      "grad_norm": 0.5116357803344727,
      "learning_rate": 9.729396538405144e-07,
      "loss": 0.5141,
      "step": 3070
    },
    {
      "epoch": 0.5237039563437926,
      "grad_norm": 0.44367215037345886,
      "learning_rate": 9.723875408708866e-07,
      "loss": 0.5067,
      "step": 3071
    },
    {
      "epoch": 0.5238744884038199,
      "grad_norm": 0.5452920198440552,
      "learning_rate": 9.718354363246245e-07,
      "loss": 0.5052,
      "step": 3072
    },
    {
      "epoch": 0.5240450204638472,
      "grad_norm": 0.5436156988143921,
      "learning_rate": 9.712833403701511e-07,
      "loss": 0.5013,
      "step": 3073
    },
    {
      "epoch": 0.5242155525238745,
      "grad_norm": 0.5032275915145874,
      "learning_rate": 9.707312531758869e-07,
      "loss": 0.5187,
      "step": 3074
    },
    {
      "epoch": 0.5243860845839018,
      "grad_norm": 0.4169192314147949,
      "learning_rate": 9.701791749102497e-07,
      "loss": 0.5139,
      "step": 3075
    },
    {
      "epoch": 0.524556616643929,
      "grad_norm": 0.4356226623058319,
      "learning_rate": 9.69627105741655e-07,
      "loss": 0.5155,
      "step": 3076
    },
    {
      "epoch": 0.5247271487039563,
      "grad_norm": 0.497597873210907,
      "learning_rate": 9.690750458385151e-07,
      "loss": 0.5078,
      "step": 3077
    },
    {
      "epoch": 0.5248976807639836,
      "grad_norm": 0.360411673784256,
      "learning_rate": 9.685229953692392e-07,
      "loss": 0.5152,
      "step": 3078
    },
    {
      "epoch": 0.525068212824011,
      "grad_norm": 0.5131139755249023,
      "learning_rate": 9.679709545022344e-07,
      "loss": 0.5061,
      "step": 3079
    },
    {
      "epoch": 0.5252387448840382,
      "grad_norm": 0.44146037101745605,
      "learning_rate": 9.67418923405904e-07,
      "loss": 0.4981,
      "step": 3080
    },
    {
      "epoch": 0.5254092769440655,
      "grad_norm": 0.3964713513851166,
      "learning_rate": 9.668669022486495e-07,
      "loss": 0.5057,
      "step": 3081
    },
    {
      "epoch": 0.5255798090040927,
      "grad_norm": 0.42470380663871765,
      "learning_rate": 9.66314891198868e-07,
      "loss": 0.5104,
      "step": 3082
    },
    {
      "epoch": 0.5257503410641201,
      "grad_norm": 0.488445520401001,
      "learning_rate": 9.657628904249547e-07,
      "loss": 0.5013,
      "step": 3083
    },
    {
      "epoch": 0.5259208731241474,
      "grad_norm": 0.5390302538871765,
      "learning_rate": 9.652109000953008e-07,
      "loss": 0.5103,
      "step": 3084
    },
    {
      "epoch": 0.5260914051841746,
      "grad_norm": 0.37049973011016846,
      "learning_rate": 9.646589203782943e-07,
      "loss": 0.5156,
      "step": 3085
    },
    {
      "epoch": 0.5262619372442019,
      "grad_norm": 0.4628862142562866,
      "learning_rate": 9.641069514423214e-07,
      "loss": 0.5112,
      "step": 3086
    },
    {
      "epoch": 0.5264324693042292,
      "grad_norm": 0.5830252766609192,
      "learning_rate": 9.635549934557627e-07,
      "loss": 0.5076,
      "step": 3087
    },
    {
      "epoch": 0.5266030013642565,
      "grad_norm": 0.46300166845321655,
      "learning_rate": 9.630030465869974e-07,
      "loss": 0.5254,
      "step": 3088
    },
    {
      "epoch": 0.5267735334242838,
      "grad_norm": 0.44175946712493896,
      "learning_rate": 9.624511110043998e-07,
      "loss": 0.523,
      "step": 3089
    },
    {
      "epoch": 0.526944065484311,
      "grad_norm": 0.41492757201194763,
      "learning_rate": 9.61899186876343e-07,
      "loss": 0.5051,
      "step": 3090
    },
    {
      "epoch": 0.5271145975443383,
      "grad_norm": 0.4990520477294922,
      "learning_rate": 9.613472743711937e-07,
      "loss": 0.502,
      "step": 3091
    },
    {
      "epoch": 0.5272851296043656,
      "grad_norm": 0.425994336605072,
      "learning_rate": 9.607953736573173e-07,
      "loss": 0.5174,
      "step": 3092
    },
    {
      "epoch": 0.527455661664393,
      "grad_norm": 0.4710950255393982,
      "learning_rate": 9.602434849030747e-07,
      "loss": 0.5128,
      "step": 3093
    },
    {
      "epoch": 0.5276261937244202,
      "grad_norm": 0.42161130905151367,
      "learning_rate": 9.59691608276823e-07,
      "loss": 0.5136,
      "step": 3094
    },
    {
      "epoch": 0.5277967257844475,
      "grad_norm": 0.36253276467323303,
      "learning_rate": 9.59139743946916e-07,
      "loss": 0.5055,
      "step": 3095
    },
    {
      "epoch": 0.5279672578444747,
      "grad_norm": 0.5080360174179077,
      "learning_rate": 9.585878920817032e-07,
      "loss": 0.5085,
      "step": 3096
    },
    {
      "epoch": 0.528137789904502,
      "grad_norm": 0.48823845386505127,
      "learning_rate": 9.58036052849532e-07,
      "loss": 0.5119,
      "step": 3097
    },
    {
      "epoch": 0.5283083219645294,
      "grad_norm": 0.38697031140327454,
      "learning_rate": 9.574842264187431e-07,
      "loss": 0.5214,
      "step": 3098
    },
    {
      "epoch": 0.5284788540245566,
      "grad_norm": 0.5833395719528198,
      "learning_rate": 9.569324129576757e-07,
      "loss": 0.5383,
      "step": 3099
    },
    {
      "epoch": 0.5286493860845839,
      "grad_norm": 0.5464732646942139,
      "learning_rate": 9.563806126346643e-07,
      "loss": 0.5107,
      "step": 3100
    },
    {
      "epoch": 0.5288199181446112,
      "grad_norm": 0.44349968433380127,
      "learning_rate": 9.558288256180387e-07,
      "loss": 0.5251,
      "step": 3101
    },
    {
      "epoch": 0.5289904502046384,
      "grad_norm": 0.4714204668998718,
      "learning_rate": 9.552770520761256e-07,
      "loss": 0.5167,
      "step": 3102
    },
    {
      "epoch": 0.5291609822646658,
      "grad_norm": 0.4778265953063965,
      "learning_rate": 9.54725292177247e-07,
      "loss": 0.5194,
      "step": 3103
    },
    {
      "epoch": 0.529331514324693,
      "grad_norm": 0.39685267210006714,
      "learning_rate": 9.541735460897212e-07,
      "loss": 0.5108,
      "step": 3104
    },
    {
      "epoch": 0.5295020463847203,
      "grad_norm": 0.47264039516448975,
      "learning_rate": 9.536218139818615e-07,
      "loss": 0.5057,
      "step": 3105
    },
    {
      "epoch": 0.5296725784447476,
      "grad_norm": 0.39579394459724426,
      "learning_rate": 9.530700960219787e-07,
      "loss": 0.5103,
      "step": 3106
    },
    {
      "epoch": 0.529843110504775,
      "grad_norm": 0.4496646523475647,
      "learning_rate": 9.525183923783769e-07,
      "loss": 0.5122,
      "step": 3107
    },
    {
      "epoch": 0.5300136425648022,
      "grad_norm": 0.5286594033241272,
      "learning_rate": 9.519667032193568e-07,
      "loss": 0.5137,
      "step": 3108
    },
    {
      "epoch": 0.5301841746248295,
      "grad_norm": 0.38828226923942566,
      "learning_rate": 9.514150287132161e-07,
      "loss": 0.5097,
      "step": 3109
    },
    {
      "epoch": 0.5303547066848567,
      "grad_norm": 0.39556238055229187,
      "learning_rate": 9.508633690282457e-07,
      "loss": 0.5174,
      "step": 3110
    },
    {
      "epoch": 0.530525238744884,
      "grad_norm": 0.3622030019760132,
      "learning_rate": 9.503117243327337e-07,
      "loss": 0.5115,
      "step": 3111
    },
    {
      "epoch": 0.5306957708049114,
      "grad_norm": 0.3833901584148407,
      "learning_rate": 9.497600947949626e-07,
      "loss": 0.5142,
      "step": 3112
    },
    {
      "epoch": 0.5308663028649386,
      "grad_norm": 0.38286980986595154,
      "learning_rate": 9.492084805832111e-07,
      "loss": 0.5074,
      "step": 3113
    },
    {
      "epoch": 0.5310368349249659,
      "grad_norm": 0.3525489270687103,
      "learning_rate": 9.486568818657522e-07,
      "loss": 0.5191,
      "step": 3114
    },
    {
      "epoch": 0.5312073669849932,
      "grad_norm": 0.39206886291503906,
      "learning_rate": 9.481052988108558e-07,
      "loss": 0.5167,
      "step": 3115
    },
    {
      "epoch": 0.5313778990450204,
      "grad_norm": 0.3706628978252411,
      "learning_rate": 9.475537315867851e-07,
      "loss": 0.5194,
      "step": 3116
    },
    {
      "epoch": 0.5315484311050478,
      "grad_norm": 0.4675542116165161,
      "learning_rate": 9.470021803617991e-07,
      "loss": 0.5163,
      "step": 3117
    },
    {
      "epoch": 0.531718963165075,
      "grad_norm": 0.4427764117717743,
      "learning_rate": 9.464506453041533e-07,
      "loss": 0.5143,
      "step": 3118
    },
    {
      "epoch": 0.5318894952251023,
      "grad_norm": 0.3776489794254303,
      "learning_rate": 9.458991265820963e-07,
      "loss": 0.5236,
      "step": 3119
    },
    {
      "epoch": 0.5320600272851296,
      "grad_norm": 0.40652182698249817,
      "learning_rate": 9.453476243638731e-07,
      "loss": 0.5171,
      "step": 3120
    },
    {
      "epoch": 0.5322305593451568,
      "grad_norm": 0.41259488463401794,
      "learning_rate": 9.447961388177228e-07,
      "loss": 0.5131,
      "step": 3121
    },
    {
      "epoch": 0.5324010914051842,
      "grad_norm": 0.4117606282234192,
      "learning_rate": 9.4424467011188e-07,
      "loss": 0.5285,
      "step": 3122
    },
    {
      "epoch": 0.5325716234652115,
      "grad_norm": 0.4624188542366028,
      "learning_rate": 9.436932184145738e-07,
      "loss": 0.5124,
      "step": 3123
    },
    {
      "epoch": 0.5327421555252387,
      "grad_norm": 0.3732365369796753,
      "learning_rate": 9.431417838940281e-07,
      "loss": 0.5117,
      "step": 3124
    },
    {
      "epoch": 0.532912687585266,
      "grad_norm": 0.42948514223098755,
      "learning_rate": 9.425903667184619e-07,
      "loss": 0.5089,
      "step": 3125
    },
    {
      "epoch": 0.5330832196452933,
      "grad_norm": 0.5732792615890503,
      "learning_rate": 9.420389670560881e-07,
      "loss": 0.5171,
      "step": 3126
    },
    {
      "epoch": 0.5332537517053206,
      "grad_norm": 0.701641857624054,
      "learning_rate": 9.41487585075116e-07,
      "loss": 0.5106,
      "step": 3127
    },
    {
      "epoch": 0.5334242837653479,
      "grad_norm": 0.6061239838600159,
      "learning_rate": 9.409362209437475e-07,
      "loss": 0.519,
      "step": 3128
    },
    {
      "epoch": 0.5335948158253752,
      "grad_norm": 0.40224212408065796,
      "learning_rate": 9.403848748301804e-07,
      "loss": 0.5081,
      "step": 3129
    },
    {
      "epoch": 0.5337653478854024,
      "grad_norm": 0.5239095687866211,
      "learning_rate": 9.39833546902606e-07,
      "loss": 0.5132,
      "step": 3130
    },
    {
      "epoch": 0.5339358799454298,
      "grad_norm": 0.6446024775505066,
      "learning_rate": 9.39282237329211e-07,
      "loss": 0.5107,
      "step": 3131
    },
    {
      "epoch": 0.534106412005457,
      "grad_norm": 0.43921467661857605,
      "learning_rate": 9.387309462781759e-07,
      "loss": 0.509,
      "step": 3132
    },
    {
      "epoch": 0.5342769440654843,
      "grad_norm": 0.45637884736061096,
      "learning_rate": 9.381796739176754e-07,
      "loss": 0.5099,
      "step": 3133
    },
    {
      "epoch": 0.5344474761255116,
      "grad_norm": 0.5327367782592773,
      "learning_rate": 9.376284204158799e-07,
      "loss": 0.505,
      "step": 3134
    },
    {
      "epoch": 0.5346180081855388,
      "grad_norm": 0.34699317812919617,
      "learning_rate": 9.370771859409515e-07,
      "loss": 0.5003,
      "step": 3135
    },
    {
      "epoch": 0.5347885402455662,
      "grad_norm": 0.4476794898509979,
      "learning_rate": 9.365259706610492e-07,
      "loss": 0.5071,
      "step": 3136
    },
    {
      "epoch": 0.5349590723055935,
      "grad_norm": 0.5096878409385681,
      "learning_rate": 9.359747747443242e-07,
      "loss": 0.5055,
      "step": 3137
    },
    {
      "epoch": 0.5351296043656207,
      "grad_norm": 0.33670952916145325,
      "learning_rate": 9.354235983589229e-07,
      "loss": 0.5079,
      "step": 3138
    },
    {
      "epoch": 0.535300136425648,
      "grad_norm": 0.41531407833099365,
      "learning_rate": 9.348724416729853e-07,
      "loss": 0.516,
      "step": 3139
    },
    {
      "epoch": 0.5354706684856753,
      "grad_norm": 0.41998550295829773,
      "learning_rate": 9.34321304854645e-07,
      "loss": 0.5047,
      "step": 3140
    },
    {
      "epoch": 0.5356412005457026,
      "grad_norm": 0.3687032163143158,
      "learning_rate": 9.337701880720304e-07,
      "loss": 0.4934,
      "step": 3141
    },
    {
      "epoch": 0.5358117326057299,
      "grad_norm": 0.36565807461738586,
      "learning_rate": 9.332190914932629e-07,
      "loss": 0.4982,
      "step": 3142
    },
    {
      "epoch": 0.5359822646657572,
      "grad_norm": 0.389960914850235,
      "learning_rate": 9.326680152864592e-07,
      "loss": 0.5048,
      "step": 3143
    },
    {
      "epoch": 0.5361527967257844,
      "grad_norm": 0.44277599453926086,
      "learning_rate": 9.321169596197276e-07,
      "loss": 0.5068,
      "step": 3144
    },
    {
      "epoch": 0.5363233287858117,
      "grad_norm": 0.4231550991535187,
      "learning_rate": 9.315659246611721e-07,
      "loss": 0.4989,
      "step": 3145
    },
    {
      "epoch": 0.536493860845839,
      "grad_norm": 0.4637570083141327,
      "learning_rate": 9.310149105788893e-07,
      "loss": 0.5062,
      "step": 3146
    },
    {
      "epoch": 0.5366643929058663,
      "grad_norm": 0.34833332896232605,
      "learning_rate": 9.3046391754097e-07,
      "loss": 0.5156,
      "step": 3147
    },
    {
      "epoch": 0.5368349249658936,
      "grad_norm": 0.41452687978744507,
      "learning_rate": 9.299129457154979e-07,
      "loss": 0.5069,
      "step": 3148
    },
    {
      "epoch": 0.5370054570259208,
      "grad_norm": 0.42182138562202454,
      "learning_rate": 9.293619952705509e-07,
      "loss": 0.4999,
      "step": 3149
    },
    {
      "epoch": 0.5371759890859482,
      "grad_norm": 0.36753180623054504,
      "learning_rate": 9.288110663742001e-07,
      "loss": 0.4943,
      "step": 3150
    },
    {
      "epoch": 0.5373465211459755,
      "grad_norm": 0.417085200548172,
      "learning_rate": 9.282601591945097e-07,
      "loss": 0.513,
      "step": 3151
    },
    {
      "epoch": 0.5375170532060027,
      "grad_norm": 0.5050751566886902,
      "learning_rate": 9.277092738995385e-07,
      "loss": 0.4978,
      "step": 3152
    },
    {
      "epoch": 0.53768758526603,
      "grad_norm": 0.3634929656982422,
      "learning_rate": 9.271584106573365e-07,
      "loss": 0.5179,
      "step": 3153
    },
    {
      "epoch": 0.5378581173260573,
      "grad_norm": 0.4662638306617737,
      "learning_rate": 9.266075696359496e-07,
      "loss": 0.5064,
      "step": 3154
    },
    {
      "epoch": 0.5380286493860846,
      "grad_norm": 0.47530871629714966,
      "learning_rate": 9.260567510034146e-07,
      "loss": 0.5048,
      "step": 3155
    },
    {
      "epoch": 0.5381991814461119,
      "grad_norm": 0.3750818073749542,
      "learning_rate": 9.255059549277625e-07,
      "loss": 0.4968,
      "step": 3156
    },
    {
      "epoch": 0.5383697135061392,
      "grad_norm": 0.4817650318145752,
      "learning_rate": 9.249551815770178e-07,
      "loss": 0.5161,
      "step": 3157
    },
    {
      "epoch": 0.5385402455661664,
      "grad_norm": 0.385273277759552,
      "learning_rate": 9.24404431119197e-07,
      "loss": 0.5048,
      "step": 3158
    },
    {
      "epoch": 0.5387107776261937,
      "grad_norm": 0.39786481857299805,
      "learning_rate": 9.238537037223106e-07,
      "loss": 0.5026,
      "step": 3159
    },
    {
      "epoch": 0.538881309686221,
      "grad_norm": 0.38584035634994507,
      "learning_rate": 9.233029995543612e-07,
      "loss": 0.5089,
      "step": 3160
    },
    {
      "epoch": 0.5390518417462483,
      "grad_norm": 0.36149051785469055,
      "learning_rate": 9.227523187833456e-07,
      "loss": 0.5134,
      "step": 3161
    },
    {
      "epoch": 0.5392223738062756,
      "grad_norm": 0.3750452697277069,
      "learning_rate": 9.222016615772527e-07,
      "loss": 0.5117,
      "step": 3162
    },
    {
      "epoch": 0.5393929058663028,
      "grad_norm": 0.33727264404296875,
      "learning_rate": 9.216510281040628e-07,
      "loss": 0.5042,
      "step": 3163
    },
    {
      "epoch": 0.5395634379263301,
      "grad_norm": 0.39926877617836,
      "learning_rate": 9.211004185317517e-07,
      "loss": 0.5012,
      "step": 3164
    },
    {
      "epoch": 0.5397339699863575,
      "grad_norm": 0.4951111078262329,
      "learning_rate": 9.205498330282858e-07,
      "loss": 0.5006,
      "step": 3165
    },
    {
      "epoch": 0.5399045020463847,
      "grad_norm": 0.3622974157333374,
      "learning_rate": 9.199992717616253e-07,
      "loss": 0.5041,
      "step": 3166
    },
    {
      "epoch": 0.540075034106412,
      "grad_norm": 0.40538978576660156,
      "learning_rate": 9.194487348997224e-07,
      "loss": 0.4917,
      "step": 3167
    },
    {
      "epoch": 0.5402455661664393,
      "grad_norm": 0.470736563205719,
      "learning_rate": 9.188982226105223e-07,
      "loss": 0.5079,
      "step": 3168
    },
    {
      "epoch": 0.5404160982264665,
      "grad_norm": 0.33788180351257324,
      "learning_rate": 9.183477350619617e-07,
      "loss": 0.5116,
      "step": 3169
    },
    {
      "epoch": 0.5405866302864939,
      "grad_norm": 0.4198651611804962,
      "learning_rate": 9.177972724219717e-07,
      "loss": 0.5108,
      "step": 3170
    },
    {
      "epoch": 0.5407571623465212,
      "grad_norm": 0.42386049032211304,
      "learning_rate": 9.172468348584741e-07,
      "loss": 0.5058,
      "step": 3171
    },
    {
      "epoch": 0.5409276944065484,
      "grad_norm": 0.4214495122432709,
      "learning_rate": 9.166964225393827e-07,
      "loss": 0.508,
      "step": 3172
    },
    {
      "epoch": 0.5410982264665757,
      "grad_norm": 0.3338247239589691,
      "learning_rate": 9.161460356326058e-07,
      "loss": 0.5041,
      "step": 3173
    },
    {
      "epoch": 0.541268758526603,
      "grad_norm": 0.37674450874328613,
      "learning_rate": 9.155956743060417e-07,
      "loss": 0.504,
      "step": 3174
    },
    {
      "epoch": 0.5414392905866303,
      "grad_norm": 0.3888840675354004,
      "learning_rate": 9.150453387275823e-07,
      "loss": 0.5033,
      "step": 3175
    },
    {
      "epoch": 0.5416098226466576,
      "grad_norm": 0.3640449047088623,
      "learning_rate": 9.14495029065111e-07,
      "loss": 0.5128,
      "step": 3176
    },
    {
      "epoch": 0.5417803547066848,
      "grad_norm": 0.40192922949790955,
      "learning_rate": 9.139447454865035e-07,
      "loss": 0.5085,
      "step": 3177
    },
    {
      "epoch": 0.5419508867667121,
      "grad_norm": 0.3610800802707672,
      "learning_rate": 9.133944881596273e-07,
      "loss": 0.5013,
      "step": 3178
    },
    {
      "epoch": 0.5421214188267395,
      "grad_norm": 0.5778458118438721,
      "learning_rate": 9.128442572523419e-07,
      "loss": 0.51,
      "step": 3179
    },
    {
      "epoch": 0.5422919508867667,
      "grad_norm": 0.6906993985176086,
      "learning_rate": 9.122940529324998e-07,
      "loss": 0.5077,
      "step": 3180
    },
    {
      "epoch": 0.542462482946794,
      "grad_norm": 0.5404031276702881,
      "learning_rate": 9.117438753679434e-07,
      "loss": 0.506,
      "step": 3181
    },
    {
      "epoch": 0.5426330150068213,
      "grad_norm": 0.41175949573516846,
      "learning_rate": 9.111937247265089e-07,
      "loss": 0.5104,
      "step": 3182
    },
    {
      "epoch": 0.5428035470668485,
      "grad_norm": 0.37495842576026917,
      "learning_rate": 9.10643601176023e-07,
      "loss": 0.4939,
      "step": 3183
    },
    {
      "epoch": 0.5429740791268759,
      "grad_norm": 0.5369127988815308,
      "learning_rate": 9.100935048843048e-07,
      "loss": 0.5115,
      "step": 3184
    },
    {
      "epoch": 0.5431446111869032,
      "grad_norm": 0.45206519961357117,
      "learning_rate": 9.095434360191643e-07,
      "loss": 0.5149,
      "step": 3185
    },
    {
      "epoch": 0.5433151432469304,
      "grad_norm": 0.3371720612049103,
      "learning_rate": 9.089933947484048e-07,
      "loss": 0.5028,
      "step": 3186
    },
    {
      "epoch": 0.5434856753069577,
      "grad_norm": 0.4170704483985901,
      "learning_rate": 9.084433812398192e-07,
      "loss": 0.4902,
      "step": 3187
    },
    {
      "epoch": 0.543656207366985,
      "grad_norm": 0.40767478942871094,
      "learning_rate": 9.078933956611925e-07,
      "loss": 0.505,
      "step": 3188
    },
    {
      "epoch": 0.5438267394270123,
      "grad_norm": 0.4488968253135681,
      "learning_rate": 9.073434381803025e-07,
      "loss": 0.4982,
      "step": 3189
    },
    {
      "epoch": 0.5439972714870396,
      "grad_norm": 0.3908664584159851,
      "learning_rate": 9.067935089649167e-07,
      "loss": 0.5007,
      "step": 3190
    },
    {
      "epoch": 0.5441678035470668,
      "grad_norm": 0.3626483678817749,
      "learning_rate": 9.062436081827951e-07,
      "loss": 0.5117,
      "step": 3191
    },
    {
      "epoch": 0.5443383356070941,
      "grad_norm": 0.4268817901611328,
      "learning_rate": 9.056937360016881e-07,
      "loss": 0.5233,
      "step": 3192
    },
    {
      "epoch": 0.5445088676671214,
      "grad_norm": 0.35921770334243774,
      "learning_rate": 9.051438925893385e-07,
      "loss": 0.51,
      "step": 3193
    },
    {
      "epoch": 0.5446793997271487,
      "grad_norm": 0.4715425670146942,
      "learning_rate": 9.045940781134794e-07,
      "loss": 0.5065,
      "step": 3194
    },
    {
      "epoch": 0.544849931787176,
      "grad_norm": 0.4177314341068268,
      "learning_rate": 9.040442927418352e-07,
      "loss": 0.5079,
      "step": 3195
    },
    {
      "epoch": 0.5450204638472033,
      "grad_norm": 0.4322715401649475,
      "learning_rate": 9.034945366421219e-07,
      "loss": 0.5081,
      "step": 3196
    },
    {
      "epoch": 0.5451909959072305,
      "grad_norm": 0.4056400656700134,
      "learning_rate": 9.029448099820457e-07,
      "loss": 0.4995,
      "step": 3197
    },
    {
      "epoch": 0.5453615279672579,
      "grad_norm": 0.4988594949245453,
      "learning_rate": 9.023951129293055e-07,
      "loss": 0.5139,
      "step": 3198
    },
    {
      "epoch": 0.5455320600272852,
      "grad_norm": 0.5313045382499695,
      "learning_rate": 9.01845445651589e-07,
      "loss": 0.5142,
      "step": 3199
    },
    {
      "epoch": 0.5457025920873124,
      "grad_norm": 0.44585803151130676,
      "learning_rate": 9.012958083165766e-07,
      "loss": 0.5125,
      "step": 3200
    },
    {
      "epoch": 0.5458731241473397,
      "grad_norm": 0.37890028953552246,
      "learning_rate": 9.007462010919388e-07,
      "loss": 0.5197,
      "step": 3201
    },
    {
      "epoch": 0.546043656207367,
      "grad_norm": 0.41216591000556946,
      "learning_rate": 9.001966241453363e-07,
      "loss": 0.5216,
      "step": 3202
    },
    {
      "epoch": 0.5462141882673943,
      "grad_norm": 0.3926418125629425,
      "learning_rate": 8.996470776444218e-07,
      "loss": 0.509,
      "step": 3203
    },
    {
      "epoch": 0.5463847203274216,
      "grad_norm": 0.4098254442214966,
      "learning_rate": 8.990975617568379e-07,
      "loss": 0.5067,
      "step": 3204
    },
    {
      "epoch": 0.5465552523874488,
      "grad_norm": 0.4524468779563904,
      "learning_rate": 8.985480766502185e-07,
      "loss": 0.4996,
      "step": 3205
    },
    {
      "epoch": 0.5467257844474761,
      "grad_norm": 0.3585079610347748,
      "learning_rate": 8.979986224921868e-07,
      "loss": 0.528,
      "step": 3206
    },
    {
      "epoch": 0.5468963165075034,
      "grad_norm": 0.44177594780921936,
      "learning_rate": 8.974491994503586e-07,
      "loss": 0.5174,
      "step": 3207
    },
    {
      "epoch": 0.5470668485675307,
      "grad_norm": 0.5257750153541565,
      "learning_rate": 8.968998076923382e-07,
      "loss": 0.5132,
      "step": 3208
    },
    {
      "epoch": 0.547237380627558,
      "grad_norm": 0.5164710879325867,
      "learning_rate": 8.963504473857219e-07,
      "loss": 0.5038,
      "step": 3209
    },
    {
      "epoch": 0.5474079126875853,
      "grad_norm": 0.47033828496932983,
      "learning_rate": 8.958011186980954e-07,
      "loss": 0.5092,
      "step": 3210
    },
    {
      "epoch": 0.5475784447476125,
      "grad_norm": 0.34907910227775574,
      "learning_rate": 8.95251821797035e-07,
      "loss": 0.5168,
      "step": 3211
    },
    {
      "epoch": 0.5477489768076398,
      "grad_norm": 0.5167688727378845,
      "learning_rate": 8.947025568501076e-07,
      "loss": 0.5157,
      "step": 3212
    },
    {
      "epoch": 0.5479195088676672,
      "grad_norm": 0.4435700476169586,
      "learning_rate": 8.9415332402487e-07,
      "loss": 0.5127,
      "step": 3213
    },
    {
      "epoch": 0.5480900409276944,
      "grad_norm": 0.38165926933288574,
      "learning_rate": 8.936041234888694e-07,
      "loss": 0.5167,
      "step": 3214
    },
    {
      "epoch": 0.5482605729877217,
      "grad_norm": 0.4699343740940094,
      "learning_rate": 8.930549554096429e-07,
      "loss": 0.5038,
      "step": 3215
    },
    {
      "epoch": 0.548431105047749,
      "grad_norm": 0.3915722370147705,
      "learning_rate": 8.925058199547183e-07,
      "loss": 0.5098,
      "step": 3216
    },
    {
      "epoch": 0.5486016371077762,
      "grad_norm": 0.3906923532485962,
      "learning_rate": 8.919567172916131e-07,
      "loss": 0.5075,
      "step": 3217
    },
    {
      "epoch": 0.5487721691678036,
      "grad_norm": 0.45521485805511475,
      "learning_rate": 8.914076475878342e-07,
      "loss": 0.4987,
      "step": 3218
    },
    {
      "epoch": 0.5489427012278308,
      "grad_norm": 0.41876670718193054,
      "learning_rate": 8.908586110108795e-07,
      "loss": 0.5032,
      "step": 3219
    },
    {
      "epoch": 0.5491132332878581,
      "grad_norm": 0.5294049978256226,
      "learning_rate": 8.903096077282358e-07,
      "loss": 0.5062,
      "step": 3220
    },
    {
      "epoch": 0.5492837653478854,
      "grad_norm": 0.450884610414505,
      "learning_rate": 8.897606379073807e-07,
      "loss": 0.5055,
      "step": 3221
    },
    {
      "epoch": 0.5494542974079127,
      "grad_norm": 0.5044273138046265,
      "learning_rate": 8.892117017157806e-07,
      "loss": 0.5038,
      "step": 3222
    },
    {
      "epoch": 0.54962482946794,
      "grad_norm": 0.42872220277786255,
      "learning_rate": 8.88662799320893e-07,
      "loss": 0.5085,
      "step": 3223
    },
    {
      "epoch": 0.5497953615279673,
      "grad_norm": 0.42034220695495605,
      "learning_rate": 8.881139308901632e-07,
      "loss": 0.5109,
      "step": 3224
    },
    {
      "epoch": 0.5499658935879945,
      "grad_norm": 0.42412039637565613,
      "learning_rate": 8.87565096591028e-07,
      "loss": 0.5038,
      "step": 3225
    },
    {
      "epoch": 0.5501364256480218,
      "grad_norm": 0.4287491738796234,
      "learning_rate": 8.870162965909128e-07,
      "loss": 0.5076,
      "step": 3226
    },
    {
      "epoch": 0.5503069577080492,
      "grad_norm": 0.39096325635910034,
      "learning_rate": 8.864675310572324e-07,
      "loss": 0.52,
      "step": 3227
    },
    {
      "epoch": 0.5504774897680764,
      "grad_norm": 0.4827026426792145,
      "learning_rate": 8.859188001573917e-07,
      "loss": 0.5066,
      "step": 3228
    },
    {
      "epoch": 0.5506480218281037,
      "grad_norm": 0.41472142934799194,
      "learning_rate": 8.853701040587844e-07,
      "loss": 0.5058,
      "step": 3229
    },
    {
      "epoch": 0.550818553888131,
      "grad_norm": 0.37448644638061523,
      "learning_rate": 8.848214429287945e-07,
      "loss": 0.5129,
      "step": 3230
    },
    {
      "epoch": 0.5509890859481582,
      "grad_norm": 0.4399203658103943,
      "learning_rate": 8.842728169347939e-07,
      "loss": 0.5163,
      "step": 3231
    },
    {
      "epoch": 0.5511596180081856,
      "grad_norm": 0.3696085810661316,
      "learning_rate": 8.837242262441457e-07,
      "loss": 0.5135,
      "step": 3232
    },
    {
      "epoch": 0.5513301500682128,
      "grad_norm": 0.44104957580566406,
      "learning_rate": 8.831756710242006e-07,
      "loss": 0.4954,
      "step": 3233
    },
    {
      "epoch": 0.5515006821282401,
      "grad_norm": 0.368233323097229,
      "learning_rate": 8.826271514422984e-07,
      "loss": 0.514,
      "step": 3234
    },
    {
      "epoch": 0.5516712141882674,
      "grad_norm": 0.4140407145023346,
      "learning_rate": 8.820786676657698e-07,
      "loss": 0.5028,
      "step": 3235
    },
    {
      "epoch": 0.5518417462482946,
      "grad_norm": 0.4295588433742523,
      "learning_rate": 8.815302198619329e-07,
      "loss": 0.4929,
      "step": 3236
    },
    {
      "epoch": 0.552012278308322,
      "grad_norm": 0.4363129436969757,
      "learning_rate": 8.809818081980955e-07,
      "loss": 0.5101,
      "step": 3237
    },
    {
      "epoch": 0.5521828103683493,
      "grad_norm": 0.45414286851882935,
      "learning_rate": 8.80433432841554e-07,
      "loss": 0.5178,
      "step": 3238
    },
    {
      "epoch": 0.5523533424283765,
      "grad_norm": 0.5862538814544678,
      "learning_rate": 8.798850939595943e-07,
      "loss": 0.5096,
      "step": 3239
    },
    {
      "epoch": 0.5525238744884038,
      "grad_norm": 0.5525916218757629,
      "learning_rate": 8.793367917194908e-07,
      "loss": 0.5001,
      "step": 3240
    },
    {
      "epoch": 0.552694406548431,
      "grad_norm": 0.5342692136764526,
      "learning_rate": 8.787885262885064e-07,
      "loss": 0.5042,
      "step": 3241
    },
    {
      "epoch": 0.5528649386084584,
      "grad_norm": 0.4479496479034424,
      "learning_rate": 8.782402978338936e-07,
      "loss": 0.5152,
      "step": 3242
    },
    {
      "epoch": 0.5530354706684857,
      "grad_norm": 0.46658965945243835,
      "learning_rate": 8.776921065228924e-07,
      "loss": 0.5169,
      "step": 3243
    },
    {
      "epoch": 0.553206002728513,
      "grad_norm": 0.49106746912002563,
      "learning_rate": 8.771439525227335e-07,
      "loss": 0.5107,
      "step": 3244
    },
    {
      "epoch": 0.5533765347885402,
      "grad_norm": 0.5296341776847839,
      "learning_rate": 8.765958360006338e-07,
      "loss": 0.5091,
      "step": 3245
    },
    {
      "epoch": 0.5535470668485676,
      "grad_norm": 0.5267268419265747,
      "learning_rate": 8.760477571238007e-07,
      "loss": 0.5118,
      "step": 3246
    },
    {
      "epoch": 0.5537175989085948,
      "grad_norm": 0.41154804825782776,
      "learning_rate": 8.754997160594288e-07,
      "loss": 0.4995,
      "step": 3247
    },
    {
      "epoch": 0.5538881309686221,
      "grad_norm": 0.38790273666381836,
      "learning_rate": 8.749517129747021e-07,
      "loss": 0.4999,
      "step": 3248
    },
    {
      "epoch": 0.5540586630286494,
      "grad_norm": 0.442150741815567,
      "learning_rate": 8.744037480367922e-07,
      "loss": 0.5054,
      "step": 3249
    },
    {
      "epoch": 0.5542291950886766,
      "grad_norm": 0.36079126596450806,
      "learning_rate": 8.738558214128595e-07,
      "loss": 0.5083,
      "step": 3250
    },
    {
      "epoch": 0.554399727148704,
      "grad_norm": 0.6051453948020935,
      "learning_rate": 8.733079332700533e-07,
      "loss": 0.505,
      "step": 3251
    },
    {
      "epoch": 0.5545702592087313,
      "grad_norm": 0.533747673034668,
      "learning_rate": 8.727600837755096e-07,
      "loss": 0.4941,
      "step": 3252
    },
    {
      "epoch": 0.5547407912687585,
      "grad_norm": 0.4203532934188843,
      "learning_rate": 8.722122730963544e-07,
      "loss": 0.5084,
      "step": 3253
    },
    {
      "epoch": 0.5549113233287858,
      "grad_norm": 0.5379196405410767,
      "learning_rate": 8.716645013997004e-07,
      "loss": 0.5148,
      "step": 3254
    },
    {
      "epoch": 0.555081855388813,
      "grad_norm": 0.4198959469795227,
      "learning_rate": 8.711167688526495e-07,
      "loss": 0.5177,
      "step": 3255
    },
    {
      "epoch": 0.5552523874488404,
      "grad_norm": 0.47099459171295166,
      "learning_rate": 8.70569075622291e-07,
      "loss": 0.5112,
      "step": 3256
    },
    {
      "epoch": 0.5554229195088677,
      "grad_norm": 0.5614515542984009,
      "learning_rate": 8.700214218757021e-07,
      "loss": 0.5151,
      "step": 3257
    },
    {
      "epoch": 0.555593451568895,
      "grad_norm": 0.3508898913860321,
      "learning_rate": 8.694738077799487e-07,
      "loss": 0.5176,
      "step": 3258
    },
    {
      "epoch": 0.5557639836289222,
      "grad_norm": 0.5202228426933289,
      "learning_rate": 8.689262335020836e-07,
      "loss": 0.5096,
      "step": 3259
    },
    {
      "epoch": 0.5559345156889495,
      "grad_norm": 0.4458746910095215,
      "learning_rate": 8.683786992091489e-07,
      "loss": 0.5188,
      "step": 3260
    },
    {
      "epoch": 0.5561050477489768,
      "grad_norm": 0.3903205692768097,
      "learning_rate": 8.678312050681725e-07,
      "loss": 0.5101,
      "step": 3261
    },
    {
      "epoch": 0.5562755798090041,
      "grad_norm": 0.40033742785453796,
      "learning_rate": 8.67283751246172e-07,
      "loss": 0.5097,
      "step": 3262
    },
    {
      "epoch": 0.5564461118690314,
      "grad_norm": 0.4283085763454437,
      "learning_rate": 8.667363379101514e-07,
      "loss": 0.5046,
      "step": 3263
    },
    {
      "epoch": 0.5566166439290586,
      "grad_norm": 0.5091626644134521,
      "learning_rate": 8.661889652271031e-07,
      "loss": 0.5058,
      "step": 3264
    },
    {
      "epoch": 0.556787175989086,
      "grad_norm": 0.4430004358291626,
      "learning_rate": 8.656416333640068e-07,
      "loss": 0.517,
      "step": 3265
    },
    {
      "epoch": 0.5569577080491133,
      "grad_norm": 0.41451820731163025,
      "learning_rate": 8.650943424878291e-07,
      "loss": 0.5153,
      "step": 3266
    },
    {
      "epoch": 0.5571282401091405,
      "grad_norm": 0.43322592973709106,
      "learning_rate": 8.645470927655256e-07,
      "loss": 0.5051,
      "step": 3267
    },
    {
      "epoch": 0.5572987721691678,
      "grad_norm": 0.535474419593811,
      "learning_rate": 8.639998843640377e-07,
      "loss": 0.5031,
      "step": 3268
    },
    {
      "epoch": 0.557469304229195,
      "grad_norm": 0.46225863695144653,
      "learning_rate": 8.634527174502961e-07,
      "loss": 0.5047,
      "step": 3269
    },
    {
      "epoch": 0.5576398362892224,
      "grad_norm": 0.44309374690055847,
      "learning_rate": 8.629055921912165e-07,
      "loss": 0.5109,
      "step": 3270
    },
    {
      "epoch": 0.5578103683492497,
      "grad_norm": 0.5729724764823914,
      "learning_rate": 8.623585087537039e-07,
      "loss": 0.5114,
      "step": 3271
    },
    {
      "epoch": 0.557980900409277,
      "grad_norm": 0.4734056890010834,
      "learning_rate": 8.618114673046495e-07,
      "loss": 0.5318,
      "step": 3272
    },
    {
      "epoch": 0.5581514324693042,
      "grad_norm": 0.41645896434783936,
      "learning_rate": 8.61264468010932e-07,
      "loss": 0.515,
      "step": 3273
    },
    {
      "epoch": 0.5583219645293315,
      "grad_norm": 0.6118432283401489,
      "learning_rate": 8.607175110394172e-07,
      "loss": 0.5029,
      "step": 3274
    },
    {
      "epoch": 0.5584924965893588,
      "grad_norm": 0.5564768314361572,
      "learning_rate": 8.601705965569578e-07,
      "loss": 0.5058,
      "step": 3275
    },
    {
      "epoch": 0.5586630286493861,
      "grad_norm": 0.41912588477134705,
      "learning_rate": 8.596237247303942e-07,
      "loss": 0.5096,
      "step": 3276
    },
    {
      "epoch": 0.5588335607094134,
      "grad_norm": 0.5847219228744507,
      "learning_rate": 8.590768957265526e-07,
      "loss": 0.5122,
      "step": 3277
    },
    {
      "epoch": 0.5590040927694406,
      "grad_norm": 0.5370835661888123,
      "learning_rate": 8.585301097122474e-07,
      "loss": 0.5121,
      "step": 3278
    },
    {
      "epoch": 0.5591746248294679,
      "grad_norm": 0.4499013125896454,
      "learning_rate": 8.579833668542797e-07,
      "loss": 0.5173,
      "step": 3279
    },
    {
      "epoch": 0.5593451568894953,
      "grad_norm": 0.5028849244117737,
      "learning_rate": 8.574366673194358e-07,
      "loss": 0.5184,
      "step": 3280
    },
    {
      "epoch": 0.5595156889495225,
      "grad_norm": 0.6504407525062561,
      "learning_rate": 8.568900112744913e-07,
      "loss": 0.5077,
      "step": 3281
    },
    {
      "epoch": 0.5596862210095498,
      "grad_norm": 0.3862699270248413,
      "learning_rate": 8.563433988862065e-07,
      "loss": 0.5235,
      "step": 3282
    },
    {
      "epoch": 0.559856753069577,
      "grad_norm": 0.5601797699928284,
      "learning_rate": 8.557968303213297e-07,
      "loss": 0.508,
      "step": 3283
    },
    {
      "epoch": 0.5600272851296043,
      "grad_norm": 0.5876139998435974,
      "learning_rate": 8.552503057465949e-07,
      "loss": 0.5148,
      "step": 3284
    },
    {
      "epoch": 0.5601978171896317,
      "grad_norm": 0.36474472284317017,
      "learning_rate": 8.547038253287234e-07,
      "loss": 0.5152,
      "step": 3285
    },
    {
      "epoch": 0.560368349249659,
      "grad_norm": 0.5364213585853577,
      "learning_rate": 8.541573892344221e-07,
      "loss": 0.5174,
      "step": 3286
    },
    {
      "epoch": 0.5605388813096862,
      "grad_norm": 0.46430879831314087,
      "learning_rate": 8.536109976303859e-07,
      "loss": 0.5031,
      "step": 3287
    },
    {
      "epoch": 0.5607094133697135,
      "grad_norm": 0.4579286575317383,
      "learning_rate": 8.530646506832949e-07,
      "loss": 0.5067,
      "step": 3288
    },
    {
      "epoch": 0.5608799454297408,
      "grad_norm": 0.4855574071407318,
      "learning_rate": 8.525183485598152e-07,
      "loss": 0.5078,
      "step": 3289
    },
    {
      "epoch": 0.5610504774897681,
      "grad_norm": 0.40624135732650757,
      "learning_rate": 8.519720914266009e-07,
      "loss": 0.5156,
      "step": 3290
    },
    {
      "epoch": 0.5612210095497954,
      "grad_norm": 0.49204781651496887,
      "learning_rate": 8.514258794502906e-07,
      "loss": 0.5055,
      "step": 3291
    },
    {
      "epoch": 0.5613915416098226,
      "grad_norm": 0.4669218063354492,
      "learning_rate": 8.508797127975106e-07,
      "loss": 0.509,
      "step": 3292
    },
    {
      "epoch": 0.5615620736698499,
      "grad_norm": 0.5329267978668213,
      "learning_rate": 8.503335916348723e-07,
      "loss": 0.5175,
      "step": 3293
    },
    {
      "epoch": 0.5617326057298773,
      "grad_norm": 0.44461092352867126,
      "learning_rate": 8.497875161289738e-07,
      "loss": 0.5104,
      "step": 3294
    },
    {
      "epoch": 0.5619031377899045,
      "grad_norm": 0.47285303473472595,
      "learning_rate": 8.492414864463988e-07,
      "loss": 0.5246,
      "step": 3295
    },
    {
      "epoch": 0.5620736698499318,
      "grad_norm": 0.43100637197494507,
      "learning_rate": 8.486955027537172e-07,
      "loss": 0.5081,
      "step": 3296
    },
    {
      "epoch": 0.562244201909959,
      "grad_norm": 0.535839319229126,
      "learning_rate": 8.48149565217486e-07,
      "loss": 0.5141,
      "step": 3297
    },
    {
      "epoch": 0.5624147339699863,
      "grad_norm": 0.4091265797615051,
      "learning_rate": 8.476036740042457e-07,
      "loss": 0.5224,
      "step": 3298
    },
    {
      "epoch": 0.5625852660300137,
      "grad_norm": 0.4831198751926422,
      "learning_rate": 8.470578292805251e-07,
      "loss": 0.5016,
      "step": 3299
    },
    {
      "epoch": 0.562755798090041,
      "grad_norm": 0.49579057097435,
      "learning_rate": 8.465120312128371e-07,
      "loss": 0.5192,
      "step": 3300
    },
    {
      "epoch": 0.5629263301500682,
      "grad_norm": 0.5047213435173035,
      "learning_rate": 8.459662799676819e-07,
      "loss": 0.525,
      "step": 3301
    },
    {
      "epoch": 0.5630968622100955,
      "grad_norm": 0.44361287355422974,
      "learning_rate": 8.454205757115435e-07,
      "loss": 0.5213,
      "step": 3302
    },
    {
      "epoch": 0.5632673942701227,
      "grad_norm": 0.4202507734298706,
      "learning_rate": 8.448749186108936e-07,
      "loss": 0.5083,
      "step": 3303
    },
    {
      "epoch": 0.5634379263301501,
      "grad_norm": 0.5279480814933777,
      "learning_rate": 8.443293088321881e-07,
      "loss": 0.5062,
      "step": 3304
    },
    {
      "epoch": 0.5636084583901774,
      "grad_norm": 0.5536814332008362,
      "learning_rate": 8.437837465418686e-07,
      "loss": 0.5097,
      "step": 3305
    },
    {
      "epoch": 0.5637789904502046,
      "grad_norm": 0.5440229177474976,
      "learning_rate": 8.432382319063631e-07,
      "loss": 0.5128,
      "step": 3306
    },
    {
      "epoch": 0.5639495225102319,
      "grad_norm": 0.3901594579219818,
      "learning_rate": 8.426927650920841e-07,
      "loss": 0.5143,
      "step": 3307
    },
    {
      "epoch": 0.5641200545702592,
      "grad_norm": 0.49661266803741455,
      "learning_rate": 8.421473462654304e-07,
      "loss": 0.5159,
      "step": 3308
    },
    {
      "epoch": 0.5642905866302865,
      "grad_norm": 0.5394790768623352,
      "learning_rate": 8.416019755927852e-07,
      "loss": 0.5162,
      "step": 3309
    },
    {
      "epoch": 0.5644611186903138,
      "grad_norm": 0.4680979549884796,
      "learning_rate": 8.410566532405176e-07,
      "loss": 0.519,
      "step": 3310
    },
    {
      "epoch": 0.564631650750341,
      "grad_norm": 0.4116993248462677,
      "learning_rate": 8.40511379374982e-07,
      "loss": 0.5165,
      "step": 3311
    },
    {
      "epoch": 0.5648021828103683,
      "grad_norm": 0.5312186479568481,
      "learning_rate": 8.399661541625174e-07,
      "loss": 0.5268,
      "step": 3312
    },
    {
      "epoch": 0.5649727148703957,
      "grad_norm": 0.5763075947761536,
      "learning_rate": 8.394209777694489e-07,
      "loss": 0.5081,
      "step": 3313
    },
    {
      "epoch": 0.565143246930423,
      "grad_norm": 0.47177785634994507,
      "learning_rate": 8.388758503620856e-07,
      "loss": 0.512,
      "step": 3314
    },
    {
      "epoch": 0.5653137789904502,
      "grad_norm": 0.39427313208580017,
      "learning_rate": 8.383307721067232e-07,
      "loss": 0.51,
      "step": 3315
    },
    {
      "epoch": 0.5654843110504775,
      "grad_norm": 0.4980703890323639,
      "learning_rate": 8.377857431696405e-07,
      "loss": 0.5093,
      "step": 3316
    },
    {
      "epoch": 0.5656548431105047,
      "grad_norm": 0.3834083676338196,
      "learning_rate": 8.37240763717103e-07,
      "loss": 0.4996,
      "step": 3317
    },
    {
      "epoch": 0.5658253751705321,
      "grad_norm": 0.518470287322998,
      "learning_rate": 8.3669583391536e-07,
      "loss": 0.5063,
      "step": 3318
    },
    {
      "epoch": 0.5659959072305594,
      "grad_norm": 0.50276780128479,
      "learning_rate": 8.361509539306457e-07,
      "loss": 0.5073,
      "step": 3319
    },
    {
      "epoch": 0.5661664392905866,
      "grad_norm": 0.4729515612125397,
      "learning_rate": 8.356061239291799e-07,
      "loss": 0.5109,
      "step": 3320
    },
    {
      "epoch": 0.5663369713506139,
      "grad_norm": 0.46731480956077576,
      "learning_rate": 8.350613440771662e-07,
      "loss": 0.5067,
      "step": 3321
    },
    {
      "epoch": 0.5665075034106412,
      "grad_norm": 0.4468371272087097,
      "learning_rate": 8.345166145407936e-07,
      "loss": 0.5031,
      "step": 3322
    },
    {
      "epoch": 0.5666780354706685,
      "grad_norm": 0.424031525850296,
      "learning_rate": 8.339719354862352e-07,
      "loss": 0.5042,
      "step": 3323
    },
    {
      "epoch": 0.5668485675306958,
      "grad_norm": 0.3681753873825073,
      "learning_rate": 8.334273070796496e-07,
      "loss": 0.4957,
      "step": 3324
    },
    {
      "epoch": 0.567019099590723,
      "grad_norm": 0.519213080406189,
      "learning_rate": 8.328827294871787e-07,
      "loss": 0.5096,
      "step": 3325
    },
    {
      "epoch": 0.5671896316507503,
      "grad_norm": 0.4725252091884613,
      "learning_rate": 8.3233820287495e-07,
      "loss": 0.5221,
      "step": 3326
    },
    {
      "epoch": 0.5673601637107776,
      "grad_norm": 0.378584086894989,
      "learning_rate": 8.317937274090748e-07,
      "loss": 0.5168,
      "step": 3327
    },
    {
      "epoch": 0.567530695770805,
      "grad_norm": 0.46946248412132263,
      "learning_rate": 8.312493032556488e-07,
      "loss": 0.5068,
      "step": 3328
    },
    {
      "epoch": 0.5677012278308322,
      "grad_norm": 0.4181266129016876,
      "learning_rate": 8.307049305807526e-07,
      "loss": 0.5022,
      "step": 3329
    },
    {
      "epoch": 0.5678717598908595,
      "grad_norm": 0.46339544653892517,
      "learning_rate": 8.301606095504502e-07,
      "loss": 0.5052,
      "step": 3330
    },
    {
      "epoch": 0.5680422919508867,
      "grad_norm": 0.5504735708236694,
      "learning_rate": 8.296163403307912e-07,
      "loss": 0.4989,
      "step": 3331
    },
    {
      "epoch": 0.568212824010914,
      "grad_norm": 0.40414270758628845,
      "learning_rate": 8.290721230878075e-07,
      "loss": 0.5136,
      "step": 3332
    },
    {
      "epoch": 0.5683833560709414,
      "grad_norm": 0.4675517678260803,
      "learning_rate": 8.285279579875171e-07,
      "loss": 0.4997,
      "step": 3333
    },
    {
      "epoch": 0.5685538881309686,
      "grad_norm": 0.40143638849258423,
      "learning_rate": 8.27983845195921e-07,
      "loss": 0.5199,
      "step": 3334
    },
    {
      "epoch": 0.5687244201909959,
      "grad_norm": 0.43805834650993347,
      "learning_rate": 8.274397848790042e-07,
      "loss": 0.515,
      "step": 3335
    },
    {
      "epoch": 0.5688949522510232,
      "grad_norm": 0.48092415928840637,
      "learning_rate": 8.268957772027362e-07,
      "loss": 0.4987,
      "step": 3336
    },
    {
      "epoch": 0.5690654843110505,
      "grad_norm": 0.3663955330848694,
      "learning_rate": 8.263518223330698e-07,
      "loss": 0.5049,
      "step": 3337
    },
    {
      "epoch": 0.5692360163710778,
      "grad_norm": 0.5174563527107239,
      "learning_rate": 8.258079204359427e-07,
      "loss": 0.5111,
      "step": 3338
    },
    {
      "epoch": 0.569406548431105,
      "grad_norm": 0.4845089912414551,
      "learning_rate": 8.252640716772749e-07,
      "loss": 0.5083,
      "step": 3339
    },
    {
      "epoch": 0.5695770804911323,
      "grad_norm": 0.48515674471855164,
      "learning_rate": 8.247202762229724e-07,
      "loss": 0.5043,
      "step": 3340
    },
    {
      "epoch": 0.5697476125511596,
      "grad_norm": 0.4666123688220978,
      "learning_rate": 8.241765342389223e-07,
      "loss": 0.5016,
      "step": 3341
    },
    {
      "epoch": 0.569918144611187,
      "grad_norm": 0.3873993754386902,
      "learning_rate": 8.236328458909976e-07,
      "loss": 0.5087,
      "step": 3342
    },
    {
      "epoch": 0.5700886766712142,
      "grad_norm": 0.3890726864337921,
      "learning_rate": 8.23089211345054e-07,
      "loss": 0.5068,
      "step": 3343
    },
    {
      "epoch": 0.5702592087312415,
      "grad_norm": 0.3836226165294647,
      "learning_rate": 8.225456307669302e-07,
      "loss": 0.5125,
      "step": 3344
    },
    {
      "epoch": 0.5704297407912687,
      "grad_norm": 0.45001837611198425,
      "learning_rate": 8.2200210432245e-07,
      "loss": 0.5034,
      "step": 3345
    },
    {
      "epoch": 0.570600272851296,
      "grad_norm": 0.43816789984703064,
      "learning_rate": 8.214586321774191e-07,
      "loss": 0.5026,
      "step": 3346
    },
    {
      "epoch": 0.5707708049113234,
      "grad_norm": 0.4553149342536926,
      "learning_rate": 8.209152144976276e-07,
      "loss": 0.513,
      "step": 3347
    },
    {
      "epoch": 0.5709413369713506,
      "grad_norm": 0.38841933012008667,
      "learning_rate": 8.203718514488484e-07,
      "loss": 0.5038,
      "step": 3348
    },
    {
      "epoch": 0.5711118690313779,
      "grad_norm": 0.479274719953537,
      "learning_rate": 8.198285431968389e-07,
      "loss": 0.4954,
      "step": 3349
    },
    {
      "epoch": 0.5712824010914052,
      "grad_norm": 0.5238628387451172,
      "learning_rate": 8.192852899073381e-07,
      "loss": 0.5164,
      "step": 3350
    },
    {
      "epoch": 0.5714529331514324,
      "grad_norm": 0.38953498005867004,
      "learning_rate": 8.187420917460687e-07,
      "loss": 0.5083,
      "step": 3351
    },
    {
      "epoch": 0.5716234652114598,
      "grad_norm": 0.407196044921875,
      "learning_rate": 8.181989488787382e-07,
      "loss": 0.5073,
      "step": 3352
    },
    {
      "epoch": 0.571793997271487,
      "grad_norm": 0.43075326085090637,
      "learning_rate": 8.176558614710348e-07,
      "loss": 0.5148,
      "step": 3353
    },
    {
      "epoch": 0.5719645293315143,
      "grad_norm": 0.46021196246147156,
      "learning_rate": 8.171128296886317e-07,
      "loss": 0.5198,
      "step": 3354
    },
    {
      "epoch": 0.5721350613915416,
      "grad_norm": 0.46036338806152344,
      "learning_rate": 8.165698536971839e-07,
      "loss": 0.4978,
      "step": 3355
    },
    {
      "epoch": 0.572305593451569,
      "grad_norm": 0.4110967516899109,
      "learning_rate": 8.160269336623302e-07,
      "loss": 0.503,
      "step": 3356
    },
    {
      "epoch": 0.5724761255115962,
      "grad_norm": 0.43071249127388,
      "learning_rate": 8.154840697496918e-07,
      "loss": 0.5049,
      "step": 3357
    },
    {
      "epoch": 0.5726466575716235,
      "grad_norm": 0.4496152698993683,
      "learning_rate": 8.14941262124873e-07,
      "loss": 0.4995,
      "step": 3358
    },
    {
      "epoch": 0.5728171896316507,
      "grad_norm": 0.4697989225387573,
      "learning_rate": 8.143985109534609e-07,
      "loss": 0.5092,
      "step": 3359
    },
    {
      "epoch": 0.572987721691678,
      "grad_norm": 0.4309699833393097,
      "learning_rate": 8.138558164010252e-07,
      "loss": 0.4953,
      "step": 3360
    },
    {
      "epoch": 0.5731582537517054,
      "grad_norm": 0.38893890380859375,
      "learning_rate": 8.13313178633119e-07,
      "loss": 0.4957,
      "step": 3361
    },
    {
      "epoch": 0.5733287858117326,
      "grad_norm": 0.497585654258728,
      "learning_rate": 8.12770597815277e-07,
      "loss": 0.5056,
      "step": 3362
    },
    {
      "epoch": 0.5734993178717599,
      "grad_norm": 0.43927890062332153,
      "learning_rate": 8.122280741130177e-07,
      "loss": 0.5109,
      "step": 3363
    },
    {
      "epoch": 0.5736698499317872,
      "grad_norm": 0.48195308446884155,
      "learning_rate": 8.116856076918409e-07,
      "loss": 0.5158,
      "step": 3364
    },
    {
      "epoch": 0.5738403819918144,
      "grad_norm": 0.5446553230285645,
      "learning_rate": 8.111431987172304e-07,
      "loss": 0.5075,
      "step": 3365
    },
    {
      "epoch": 0.5740109140518418,
      "grad_norm": 0.39556846022605896,
      "learning_rate": 8.106008473546512e-07,
      "loss": 0.5109,
      "step": 3366
    },
    {
      "epoch": 0.574181446111869,
      "grad_norm": 0.5156859755516052,
      "learning_rate": 8.10058553769551e-07,
      "loss": 0.5011,
      "step": 3367
    },
    {
      "epoch": 0.5743519781718963,
      "grad_norm": 0.5255439877510071,
      "learning_rate": 8.095163181273609e-07,
      "loss": 0.5039,
      "step": 3368
    },
    {
      "epoch": 0.5745225102319236,
      "grad_norm": 0.5791894793510437,
      "learning_rate": 8.089741405934923e-07,
      "loss": 0.511,
      "step": 3369
    },
    {
      "epoch": 0.5746930422919508,
      "grad_norm": 0.4920649826526642,
      "learning_rate": 8.084320213333413e-07,
      "loss": 0.5049,
      "step": 3370
    },
    {
      "epoch": 0.5748635743519782,
      "grad_norm": 0.4737277328968048,
      "learning_rate": 8.078899605122842e-07,
      "loss": 0.4991,
      "step": 3371
    },
    {
      "epoch": 0.5750341064120055,
      "grad_norm": 0.47556519508361816,
      "learning_rate": 8.073479582956808e-07,
      "loss": 0.5036,
      "step": 3372
    },
    {
      "epoch": 0.5752046384720327,
      "grad_norm": 0.5832779407501221,
      "learning_rate": 8.068060148488723e-07,
      "loss": 0.5038,
      "step": 3373
    },
    {
      "epoch": 0.57537517053206,
      "grad_norm": 0.6054734587669373,
      "learning_rate": 8.062641303371819e-07,
      "loss": 0.5144,
      "step": 3374
    },
    {
      "epoch": 0.5755457025920873,
      "grad_norm": 0.4068281054496765,
      "learning_rate": 8.057223049259156e-07,
      "loss": 0.5004,
      "step": 3375
    },
    {
      "epoch": 0.5757162346521146,
      "grad_norm": 0.4555928111076355,
      "learning_rate": 8.051805387803603e-07,
      "loss": 0.5097,
      "step": 3376
    },
    {
      "epoch": 0.5758867667121419,
      "grad_norm": 0.5093322396278381,
      "learning_rate": 8.046388320657861e-07,
      "loss": 0.5004,
      "step": 3377
    },
    {
      "epoch": 0.5760572987721692,
      "grad_norm": 0.38450372219085693,
      "learning_rate": 8.040971849474433e-07,
      "loss": 0.5005,
      "step": 3378
    },
    {
      "epoch": 0.5762278308321964,
      "grad_norm": 0.5079644322395325,
      "learning_rate": 8.035555975905661e-07,
      "loss": 0.5154,
      "step": 3379
    },
    {
      "epoch": 0.5763983628922238,
      "grad_norm": 0.37952208518981934,
      "learning_rate": 8.030140701603684e-07,
      "loss": 0.5094,
      "step": 3380
    },
    {
      "epoch": 0.576568894952251,
      "grad_norm": 0.49966394901275635,
      "learning_rate": 8.024726028220475e-07,
      "loss": 0.4944,
      "step": 3381
    },
    {
      "epoch": 0.5767394270122783,
      "grad_norm": 0.44472551345825195,
      "learning_rate": 8.019311957407812e-07,
      "loss": 0.5114,
      "step": 3382
    },
    {
      "epoch": 0.5769099590723056,
      "grad_norm": 0.42356055974960327,
      "learning_rate": 8.013898490817295e-07,
      "loss": 0.4964,
      "step": 3383
    },
    {
      "epoch": 0.5770804911323328,
      "grad_norm": 0.49199751019477844,
      "learning_rate": 8.008485630100336e-07,
      "loss": 0.5038,
      "step": 3384
    },
    {
      "epoch": 0.5772510231923602,
      "grad_norm": 0.3819704055786133,
      "learning_rate": 8.003073376908165e-07,
      "loss": 0.5036,
      "step": 3385
    },
    {
      "epoch": 0.5774215552523875,
      "grad_norm": 0.47768130898475647,
      "learning_rate": 7.997661732891831e-07,
      "loss": 0.5118,
      "step": 3386
    },
    {
      "epoch": 0.5775920873124147,
      "grad_norm": 0.42203018069267273,
      "learning_rate": 7.992250699702184e-07,
      "loss": 0.4919,
      "step": 3387
    },
    {
      "epoch": 0.577762619372442,
      "grad_norm": 0.41574037075042725,
      "learning_rate": 7.986840278989903e-07,
      "loss": 0.5117,
      "step": 3388
    },
    {
      "epoch": 0.5779331514324693,
      "grad_norm": 0.4440769553184509,
      "learning_rate": 7.981430472405471e-07,
      "loss": 0.5062,
      "step": 3389
    },
    {
      "epoch": 0.5781036834924966,
      "grad_norm": 0.377749502658844,
      "learning_rate": 7.976021281599183e-07,
      "loss": 0.5134,
      "step": 3390
    },
    {
      "epoch": 0.5782742155525239,
      "grad_norm": 0.375430166721344,
      "learning_rate": 7.970612708221153e-07,
      "loss": 0.4975,
      "step": 3391
    },
    {
      "epoch": 0.5784447476125512,
      "grad_norm": 0.43011754751205444,
      "learning_rate": 7.9652047539213e-07,
      "loss": 0.5051,
      "step": 3392
    },
    {
      "epoch": 0.5786152796725784,
      "grad_norm": 0.3825616240501404,
      "learning_rate": 7.959797420349356e-07,
      "loss": 0.4956,
      "step": 3393
    },
    {
      "epoch": 0.5787858117326057,
      "grad_norm": 0.38164180517196655,
      "learning_rate": 7.954390709154863e-07,
      "loss": 0.5092,
      "step": 3394
    },
    {
      "epoch": 0.578956343792633,
      "grad_norm": 0.35513877868652344,
      "learning_rate": 7.948984621987181e-07,
      "loss": 0.5056,
      "step": 3395
    },
    {
      "epoch": 0.5791268758526603,
      "grad_norm": 0.41221168637275696,
      "learning_rate": 7.943579160495466e-07,
      "loss": 0.5131,
      "step": 3396
    },
    {
      "epoch": 0.5792974079126876,
      "grad_norm": 0.37688201665878296,
      "learning_rate": 7.938174326328696e-07,
      "loss": 0.5038,
      "step": 3397
    },
    {
      "epoch": 0.5794679399727148,
      "grad_norm": 0.4648706912994385,
      "learning_rate": 7.932770121135648e-07,
      "loss": 0.5067,
      "step": 3398
    },
    {
      "epoch": 0.5796384720327421,
      "grad_norm": 0.49630671739578247,
      "learning_rate": 7.927366546564911e-07,
      "loss": 0.5002,
      "step": 3399
    },
    {
      "epoch": 0.5798090040927695,
      "grad_norm": 0.4030620753765106,
      "learning_rate": 7.921963604264884e-07,
      "loss": 0.5017,
      "step": 3400
    },
    {
      "epoch": 0.5799795361527967,
      "grad_norm": 0.4664936065673828,
      "learning_rate": 7.916561295883768e-07,
      "loss": 0.5088,
      "step": 3401
    },
    {
      "epoch": 0.580150068212824,
      "grad_norm": 0.49963656067848206,
      "learning_rate": 7.911159623069576e-07,
      "loss": 0.4964,
      "step": 3402
    },
    {
      "epoch": 0.5803206002728513,
      "grad_norm": 0.4847971796989441,
      "learning_rate": 7.905758587470119e-07,
      "loss": 0.501,
      "step": 3403
    },
    {
      "epoch": 0.5804911323328786,
      "grad_norm": 0.48904651403427124,
      "learning_rate": 7.900358190733027e-07,
      "loss": 0.4932,
      "step": 3404
    },
    {
      "epoch": 0.5806616643929059,
      "grad_norm": 0.39514946937561035,
      "learning_rate": 7.894958434505726e-07,
      "loss": 0.5011,
      "step": 3405
    },
    {
      "epoch": 0.5808321964529332,
      "grad_norm": 0.5680859088897705,
      "learning_rate": 7.889559320435437e-07,
      "loss": 0.5034,
      "step": 3406
    },
    {
      "epoch": 0.5810027285129604,
      "grad_norm": 0.5151458978652954,
      "learning_rate": 7.88416085016921e-07,
      "loss": 0.4995,
      "step": 3407
    },
    {
      "epoch": 0.5811732605729877,
      "grad_norm": 0.4129141569137573,
      "learning_rate": 7.878763025353875e-07,
      "loss": 0.5064,
      "step": 3408
    },
    {
      "epoch": 0.581343792633015,
      "grad_norm": 0.5038077235221863,
      "learning_rate": 7.873365847636081e-07,
      "loss": 0.4989,
      "step": 3409
    },
    {
      "epoch": 0.5815143246930423,
      "grad_norm": 0.42516499757766724,
      "learning_rate": 7.867969318662268e-07,
      "loss": 0.5146,
      "step": 3410
    },
    {
      "epoch": 0.5816848567530696,
      "grad_norm": 0.5718380808830261,
      "learning_rate": 7.862573440078688e-07,
      "loss": 0.5106,
      "step": 3411
    },
    {
      "epoch": 0.5818553888130968,
      "grad_norm": 0.43516722321510315,
      "learning_rate": 7.857178213531386e-07,
      "loss": 0.5091,
      "step": 3412
    },
    {
      "epoch": 0.5820259208731241,
      "grad_norm": 0.5345479249954224,
      "learning_rate": 7.851783640666209e-07,
      "loss": 0.5034,
      "step": 3413
    },
    {
      "epoch": 0.5821964529331515,
      "grad_norm": 0.47487872838974,
      "learning_rate": 7.84638972312882e-07,
      "loss": 0.5073,
      "step": 3414
    },
    {
      "epoch": 0.5823669849931787,
      "grad_norm": 0.42536306381225586,
      "learning_rate": 7.840996462564655e-07,
      "loss": 0.5094,
      "step": 3415
    },
    {
      "epoch": 0.582537517053206,
      "grad_norm": 0.5384988784790039,
      "learning_rate": 7.835603860618972e-07,
      "loss": 0.4991,
      "step": 3416
    },
    {
      "epoch": 0.5827080491132333,
      "grad_norm": 0.4918755888938904,
      "learning_rate": 7.830211918936821e-07,
      "loss": 0.5217,
      "step": 3417
    },
    {
      "epoch": 0.5828785811732605,
      "grad_norm": 0.4351602792739868,
      "learning_rate": 7.824820639163049e-07,
      "loss": 0.5083,
      "step": 3418
    },
    {
      "epoch": 0.5830491132332879,
      "grad_norm": 0.4212780296802521,
      "learning_rate": 7.819430022942298e-07,
      "loss": 0.5074,
      "step": 3419
    },
    {
      "epoch": 0.5832196452933152,
      "grad_norm": 0.439581960439682,
      "learning_rate": 7.814040071919016e-07,
      "loss": 0.5052,
      "step": 3420
    },
    {
      "epoch": 0.5833901773533424,
      "grad_norm": 0.4218754470348358,
      "learning_rate": 7.808650787737443e-07,
      "loss": 0.4963,
      "step": 3421
    },
    {
      "epoch": 0.5835607094133697,
      "grad_norm": 0.441415935754776,
      "learning_rate": 7.803262172041611e-07,
      "loss": 0.517,
      "step": 3422
    },
    {
      "epoch": 0.583731241473397,
      "grad_norm": 0.5194700360298157,
      "learning_rate": 7.797874226475362e-07,
      "loss": 0.5146,
      "step": 3423
    },
    {
      "epoch": 0.5839017735334243,
      "grad_norm": 0.3435308039188385,
      "learning_rate": 7.79248695268232e-07,
      "loss": 0.5244,
      "step": 3424
    },
    {
      "epoch": 0.5840723055934516,
      "grad_norm": 0.44093021750450134,
      "learning_rate": 7.787100352305909e-07,
      "loss": 0.5132,
      "step": 3425
    },
    {
      "epoch": 0.5842428376534788,
      "grad_norm": 0.44838279485702515,
      "learning_rate": 7.781714426989347e-07,
      "loss": 0.508,
      "step": 3426
    },
    {
      "epoch": 0.5844133697135061,
      "grad_norm": 1.4488683938980103,
      "learning_rate": 7.776329178375651e-07,
      "loss": 0.5336,
      "step": 3427
    },
    {
      "epoch": 0.5845839017735335,
      "grad_norm": 0.48720383644104004,
      "learning_rate": 7.770944608107622e-07,
      "loss": 0.5148,
      "step": 3428
    },
    {
      "epoch": 0.5847544338335607,
      "grad_norm": 0.392868310213089,
      "learning_rate": 7.765560717827858e-07,
      "loss": 0.5061,
      "step": 3429
    },
    {
      "epoch": 0.584924965893588,
      "grad_norm": 0.3861992359161377,
      "learning_rate": 7.760177509178758e-07,
      "loss": 0.5166,
      "step": 3430
    },
    {
      "epoch": 0.5850954979536153,
      "grad_norm": 0.5674988031387329,
      "learning_rate": 7.754794983802496e-07,
      "loss": 0.5126,
      "step": 3431
    },
    {
      "epoch": 0.5852660300136425,
      "grad_norm": 0.4710623025894165,
      "learning_rate": 7.749413143341056e-07,
      "loss": 0.5124,
      "step": 3432
    },
    {
      "epoch": 0.5854365620736699,
      "grad_norm": 0.3838975131511688,
      "learning_rate": 7.744031989436201e-07,
      "loss": 0.5087,
      "step": 3433
    },
    {
      "epoch": 0.5856070941336972,
      "grad_norm": 0.45589226484298706,
      "learning_rate": 7.738651523729491e-07,
      "loss": 0.512,
      "step": 3434
    },
    {
      "epoch": 0.5857776261937244,
      "grad_norm": 0.3960130214691162,
      "learning_rate": 7.733271747862266e-07,
      "loss": 0.5034,
      "step": 3435
    },
    {
      "epoch": 0.5859481582537517,
      "grad_norm": 0.4465065002441406,
      "learning_rate": 7.727892663475672e-07,
      "loss": 0.5236,
      "step": 3436
    },
    {
      "epoch": 0.586118690313779,
      "grad_norm": 0.47854310274124146,
      "learning_rate": 7.722514272210628e-07,
      "loss": 0.5036,
      "step": 3437
    },
    {
      "epoch": 0.5862892223738063,
      "grad_norm": 0.38538694381713867,
      "learning_rate": 7.717136575707847e-07,
      "loss": 0.5217,
      "step": 3438
    },
    {
      "epoch": 0.5864597544338336,
      "grad_norm": 0.38084253668785095,
      "learning_rate": 7.711759575607838e-07,
      "loss": 0.53,
      "step": 3439
    },
    {
      "epoch": 0.5866302864938608,
      "grad_norm": 0.4159857928752899,
      "learning_rate": 7.706383273550883e-07,
      "loss": 0.5142,
      "step": 3440
    },
    {
      "epoch": 0.5868008185538881,
      "grad_norm": 0.40294331312179565,
      "learning_rate": 7.701007671177067e-07,
      "loss": 0.512,
      "step": 3441
    },
    {
      "epoch": 0.5869713506139154,
      "grad_norm": 0.39000409841537476,
      "learning_rate": 7.695632770126247e-07,
      "loss": 0.5046,
      "step": 3442
    },
    {
      "epoch": 0.5871418826739427,
      "grad_norm": 0.41626620292663574,
      "learning_rate": 7.690258572038078e-07,
      "loss": 0.5216,
      "step": 3443
    },
    {
      "epoch": 0.58731241473397,
      "grad_norm": 0.3855079114437103,
      "learning_rate": 7.684885078551991e-07,
      "loss": 0.5061,
      "step": 3444
    },
    {
      "epoch": 0.5874829467939973,
      "grad_norm": 0.38733789324760437,
      "learning_rate": 7.679512291307208e-07,
      "loss": 0.5116,
      "step": 3445
    },
    {
      "epoch": 0.5876534788540245,
      "grad_norm": 0.4045097827911377,
      "learning_rate": 7.674140211942732e-07,
      "loss": 0.5064,
      "step": 3446
    },
    {
      "epoch": 0.5878240109140518,
      "grad_norm": 0.40880081057548523,
      "learning_rate": 7.668768842097353e-07,
      "loss": 0.5273,
      "step": 3447
    },
    {
      "epoch": 0.5879945429740792,
      "grad_norm": 0.45862430334091187,
      "learning_rate": 7.663398183409645e-07,
      "loss": 0.5319,
      "step": 3448
    },
    {
      "epoch": 0.5881650750341064,
      "grad_norm": 0.35381215810775757,
      "learning_rate": 7.658028237517958e-07,
      "loss": 0.5178,
      "step": 3449
    },
    {
      "epoch": 0.5883356070941337,
      "grad_norm": 0.4584713280200958,
      "learning_rate": 7.652659006060436e-07,
      "loss": 0.5144,
      "step": 3450
    },
    {
      "epoch": 0.588506139154161,
      "grad_norm": 0.47925877571105957,
      "learning_rate": 7.647290490674997e-07,
      "loss": 0.5002,
      "step": 3451
    },
    {
      "epoch": 0.5886766712141883,
      "grad_norm": 0.36030012369155884,
      "learning_rate": 7.641922692999341e-07,
      "loss": 0.5165,
      "step": 3452
    },
    {
      "epoch": 0.5888472032742156,
      "grad_norm": 0.4329701066017151,
      "learning_rate": 7.636555614670954e-07,
      "loss": 0.5127,
      "step": 3453
    },
    {
      "epoch": 0.5890177353342428,
      "grad_norm": 0.4515969157218933,
      "learning_rate": 7.631189257327093e-07,
      "loss": 0.5046,
      "step": 3454
    },
    {
      "epoch": 0.5891882673942701,
      "grad_norm": 0.504779577255249,
      "learning_rate": 7.625823622604809e-07,
      "loss": 0.5058,
      "step": 3455
    },
    {
      "epoch": 0.5893587994542974,
      "grad_norm": 0.5348877310752869,
      "learning_rate": 7.620458712140916e-07,
      "loss": 0.5176,
      "step": 3456
    },
    {
      "epoch": 0.5895293315143247,
      "grad_norm": 0.4561511278152466,
      "learning_rate": 7.615094527572027e-07,
      "loss": 0.5042,
      "step": 3457
    },
    {
      "epoch": 0.589699863574352,
      "grad_norm": 0.5060758590698242,
      "learning_rate": 7.60973107053451e-07,
      "loss": 0.5286,
      "step": 3458
    },
    {
      "epoch": 0.5898703956343793,
      "grad_norm": 0.4909929037094116,
      "learning_rate": 7.604368342664534e-07,
      "loss": 0.5105,
      "step": 3459
    },
    {
      "epoch": 0.5900409276944065,
      "grad_norm": 0.4966579079627991,
      "learning_rate": 7.599006345598031e-07,
      "loss": 0.5055,
      "step": 3460
    },
    {
      "epoch": 0.5902114597544338,
      "grad_norm": 0.6382631063461304,
      "learning_rate": 7.59364508097071e-07,
      "loss": 0.5035,
      "step": 3461
    },
    {
      "epoch": 0.5903819918144612,
      "grad_norm": 0.575469970703125,
      "learning_rate": 7.588284550418068e-07,
      "loss": 0.5102,
      "step": 3462
    },
    {
      "epoch": 0.5905525238744884,
      "grad_norm": 0.5283661484718323,
      "learning_rate": 7.582924755575366e-07,
      "loss": 0.5117,
      "step": 3463
    },
    {
      "epoch": 0.5907230559345157,
      "grad_norm": 0.5372909903526306,
      "learning_rate": 7.577565698077647e-07,
      "loss": 0.5043,
      "step": 3464
    },
    {
      "epoch": 0.590893587994543,
      "grad_norm": 0.5969436764717102,
      "learning_rate": 7.572207379559723e-07,
      "loss": 0.5133,
      "step": 3465
    },
    {
      "epoch": 0.5910641200545702,
      "grad_norm": 0.5289788842201233,
      "learning_rate": 7.566849801656193e-07,
      "loss": 0.508,
      "step": 3466
    },
    {
      "epoch": 0.5912346521145976,
      "grad_norm": 0.4017581343650818,
      "learning_rate": 7.561492966001418e-07,
      "loss": 0.5134,
      "step": 3467
    },
    {
      "epoch": 0.5914051841746248,
      "grad_norm": 0.6081069707870483,
      "learning_rate": 7.556136874229532e-07,
      "loss": 0.5159,
      "step": 3468
    },
    {
      "epoch": 0.5915757162346521,
      "grad_norm": 0.5566324591636658,
      "learning_rate": 7.550781527974455e-07,
      "loss": 0.5205,
      "step": 3469
    },
    {
      "epoch": 0.5917462482946794,
      "grad_norm": 0.5150846838951111,
      "learning_rate": 7.545426928869864e-07,
      "loss": 0.5025,
      "step": 3470
    },
    {
      "epoch": 0.5919167803547067,
      "grad_norm": 0.4910479784011841,
      "learning_rate": 7.540073078549223e-07,
      "loss": 0.5104,
      "step": 3471
    },
    {
      "epoch": 0.592087312414734,
      "grad_norm": 0.4645948112010956,
      "learning_rate": 7.534719978645751e-07,
      "loss": 0.5118,
      "step": 3472
    },
    {
      "epoch": 0.5922578444747613,
      "grad_norm": 0.4650384485721588,
      "learning_rate": 7.529367630792454e-07,
      "loss": 0.5212,
      "step": 3473
    },
    {
      "epoch": 0.5924283765347885,
      "grad_norm": 0.4093935787677765,
      "learning_rate": 7.524016036622095e-07,
      "loss": 0.5123,
      "step": 3474
    },
    {
      "epoch": 0.5925989085948158,
      "grad_norm": 0.5681697130203247,
      "learning_rate": 7.518665197767224e-07,
      "loss": 0.5034,
      "step": 3475
    },
    {
      "epoch": 0.5927694406548432,
      "grad_norm": 0.5197209715843201,
      "learning_rate": 7.513315115860141e-07,
      "loss": 0.5123,
      "step": 3476
    },
    {
      "epoch": 0.5929399727148704,
      "grad_norm": 0.6382885575294495,
      "learning_rate": 7.507965792532922e-07,
      "loss": 0.5143,
      "step": 3477
    },
    {
      "epoch": 0.5931105047748977,
      "grad_norm": 0.6706441044807434,
      "learning_rate": 7.502617229417424e-07,
      "loss": 0.5073,
      "step": 3478
    },
    {
      "epoch": 0.593281036834925,
      "grad_norm": 0.48597392439842224,
      "learning_rate": 7.497269428145256e-07,
      "loss": 0.5143,
      "step": 3479
    },
    {
      "epoch": 0.5934515688949522,
      "grad_norm": 0.4961046576499939,
      "learning_rate": 7.491922390347802e-07,
      "loss": 0.5279,
      "step": 3480
    },
    {
      "epoch": 0.5936221009549796,
      "grad_norm": 0.578697919845581,
      "learning_rate": 7.486576117656209e-07,
      "loss": 0.5149,
      "step": 3481
    },
    {
      "epoch": 0.5937926330150068,
      "grad_norm": 0.6151126623153687,
      "learning_rate": 7.481230611701397e-07,
      "loss": 0.5216,
      "step": 3482
    },
    {
      "epoch": 0.5939631650750341,
      "grad_norm": 0.37831974029541016,
      "learning_rate": 7.475885874114048e-07,
      "loss": 0.5036,
      "step": 3483
    },
    {
      "epoch": 0.5941336971350614,
      "grad_norm": 0.507587194442749,
      "learning_rate": 7.470541906524603e-07,
      "loss": 0.5156,
      "step": 3484
    },
    {
      "epoch": 0.5943042291950886,
      "grad_norm": 0.5136541724205017,
      "learning_rate": 7.465198710563289e-07,
      "loss": 0.5046,
      "step": 3485
    },
    {
      "epoch": 0.594474761255116,
      "grad_norm": 0.44345173239707947,
      "learning_rate": 7.459856287860069e-07,
      "loss": 0.4937,
      "step": 3486
    },
    {
      "epoch": 0.5946452933151433,
      "grad_norm": 0.462137907743454,
      "learning_rate": 7.454514640044694e-07,
      "loss": 0.5011,
      "step": 3487
    },
    {
      "epoch": 0.5948158253751705,
      "grad_norm": 0.40550386905670166,
      "learning_rate": 7.449173768746665e-07,
      "loss": 0.5083,
      "step": 3488
    },
    {
      "epoch": 0.5949863574351978,
      "grad_norm": 0.4405902028083801,
      "learning_rate": 7.443833675595255e-07,
      "loss": 0.5087,
      "step": 3489
    },
    {
      "epoch": 0.595156889495225,
      "grad_norm": 0.4645247757434845,
      "learning_rate": 7.438494362219493e-07,
      "loss": 0.52,
      "step": 3490
    },
    {
      "epoch": 0.5953274215552524,
      "grad_norm": 0.423967570066452,
      "learning_rate": 7.433155830248168e-07,
      "loss": 0.5037,
      "step": 3491
    },
    {
      "epoch": 0.5954979536152797,
      "grad_norm": 0.4495370090007782,
      "learning_rate": 7.427818081309842e-07,
      "loss": 0.5106,
      "step": 3492
    },
    {
      "epoch": 0.595668485675307,
      "grad_norm": 0.38820216059684753,
      "learning_rate": 7.422481117032823e-07,
      "loss": 0.512,
      "step": 3493
    },
    {
      "epoch": 0.5958390177353342,
      "grad_norm": 0.4023274779319763,
      "learning_rate": 7.417144939045198e-07,
      "loss": 0.5024,
      "step": 3494
    },
    {
      "epoch": 0.5960095497953616,
      "grad_norm": 0.4757567346096039,
      "learning_rate": 7.411809548974793e-07,
      "loss": 0.5166,
      "step": 3495
    },
    {
      "epoch": 0.5961800818553888,
      "grad_norm": 0.48980042338371277,
      "learning_rate": 7.406474948449214e-07,
      "loss": 0.5129,
      "step": 3496
    },
    {
      "epoch": 0.5963506139154161,
      "grad_norm": 0.4435463845729828,
      "learning_rate": 7.401141139095809e-07,
      "loss": 0.5018,
      "step": 3497
    },
    {
      "epoch": 0.5965211459754434,
      "grad_norm": 0.4933076500892639,
      "learning_rate": 7.395808122541697e-07,
      "loss": 0.5087,
      "step": 3498
    },
    {
      "epoch": 0.5966916780354706,
      "grad_norm": 0.5443232655525208,
      "learning_rate": 7.39047590041375e-07,
      "loss": 0.5108,
      "step": 3499
    },
    {
      "epoch": 0.596862210095498,
      "grad_norm": 0.487313836812973,
      "learning_rate": 7.385144474338595e-07,
      "loss": 0.517,
      "step": 3500
    },
    {
      "epoch": 0.5970327421555253,
      "grad_norm": 0.5534808039665222,
      "learning_rate": 7.379813845942623e-07,
      "loss": 0.5103,
      "step": 3501
    },
    {
      "epoch": 0.5972032742155525,
      "grad_norm": 0.5094048380851746,
      "learning_rate": 7.374484016851971e-07,
      "loss": 0.5048,
      "step": 3502
    },
    {
      "epoch": 0.5973738062755798,
      "grad_norm": 0.4158516526222229,
      "learning_rate": 7.369154988692552e-07,
      "loss": 0.5028,
      "step": 3503
    },
    {
      "epoch": 0.597544338335607,
      "grad_norm": 0.6141080856323242,
      "learning_rate": 7.363826763090006e-07,
      "loss": 0.5068,
      "step": 3504
    },
    {
      "epoch": 0.5977148703956344,
      "grad_norm": 0.4984683096408844,
      "learning_rate": 7.358499341669757e-07,
      "loss": 0.5051,
      "step": 3505
    },
    {
      "epoch": 0.5978854024556617,
      "grad_norm": 0.5119192004203796,
      "learning_rate": 7.353172726056964e-07,
      "loss": 0.5099,
      "step": 3506
    },
    {
      "epoch": 0.598055934515689,
      "grad_norm": 0.5512797832489014,
      "learning_rate": 7.347846917876546e-07,
      "loss": 0.4988,
      "step": 3507
    },
    {
      "epoch": 0.5982264665757162,
      "grad_norm": 0.4553689658641815,
      "learning_rate": 7.342521918753181e-07,
      "loss": 0.5025,
      "step": 3508
    },
    {
      "epoch": 0.5983969986357435,
      "grad_norm": 0.6097546219825745,
      "learning_rate": 7.33719773031129e-07,
      "loss": 0.5177,
      "step": 3509
    },
    {
      "epoch": 0.5985675306957708,
      "grad_norm": 0.41999533772468567,
      "learning_rate": 7.331874354175058e-07,
      "loss": 0.5155,
      "step": 3510
    },
    {
      "epoch": 0.5987380627557981,
      "grad_norm": 0.5839874744415283,
      "learning_rate": 7.326551791968409e-07,
      "loss": 0.5068,
      "step": 3511
    },
    {
      "epoch": 0.5989085948158254,
      "grad_norm": 0.4493328630924225,
      "learning_rate": 7.321230045315036e-07,
      "loss": 0.5115,
      "step": 3512
    },
    {
      "epoch": 0.5990791268758526,
      "grad_norm": 0.3619157373905182,
      "learning_rate": 7.315909115838367e-07,
      "loss": 0.5035,
      "step": 3513
    },
    {
      "epoch": 0.5992496589358799,
      "grad_norm": 0.508014440536499,
      "learning_rate": 7.31058900516159e-07,
      "loss": 0.5265,
      "step": 3514
    },
    {
      "epoch": 0.5994201909959073,
      "grad_norm": 0.43933480978012085,
      "learning_rate": 7.305269714907639e-07,
      "loss": 0.5091,
      "step": 3515
    },
    {
      "epoch": 0.5995907230559345,
      "grad_norm": 0.3917537033557892,
      "learning_rate": 7.299951246699198e-07,
      "loss": 0.5156,
      "step": 3516
    },
    {
      "epoch": 0.5997612551159618,
      "grad_norm": 0.4158930480480194,
      "learning_rate": 7.294633602158703e-07,
      "loss": 0.4951,
      "step": 3517
    },
    {
      "epoch": 0.599931787175989,
      "grad_norm": 0.4387792944908142,
      "learning_rate": 7.289316782908335e-07,
      "loss": 0.501,
      "step": 3518
    },
    {
      "epoch": 0.6001023192360164,
      "grad_norm": 0.42189764976501465,
      "learning_rate": 7.28400079057003e-07,
      "loss": 0.5104,
      "step": 3519
    },
    {
      "epoch": 0.6002728512960437,
      "grad_norm": 0.3482825756072998,
      "learning_rate": 7.278685626765459e-07,
      "loss": 0.4977,
      "step": 3520
    },
    {
      "epoch": 0.600443383356071,
      "grad_norm": 0.4253080487251282,
      "learning_rate": 7.273371293116056e-07,
      "loss": 0.4884,
      "step": 3521
    },
    {
      "epoch": 0.6006139154160982,
      "grad_norm": 0.42421674728393555,
      "learning_rate": 7.268057791242994e-07,
      "loss": 0.5102,
      "step": 3522
    },
    {
      "epoch": 0.6007844474761255,
      "grad_norm": 0.4532250761985779,
      "learning_rate": 7.262745122767184e-07,
      "loss": 0.5064,
      "step": 3523
    },
    {
      "epoch": 0.6009549795361528,
      "grad_norm": 0.49927690625190735,
      "learning_rate": 7.257433289309301e-07,
      "loss": 0.5113,
      "step": 3524
    },
    {
      "epoch": 0.6011255115961801,
      "grad_norm": 0.45842334628105164,
      "learning_rate": 7.252122292489748e-07,
      "loss": 0.5072,
      "step": 3525
    },
    {
      "epoch": 0.6012960436562074,
      "grad_norm": 0.5091456174850464,
      "learning_rate": 7.246812133928686e-07,
      "loss": 0.5005,
      "step": 3526
    },
    {
      "epoch": 0.6014665757162346,
      "grad_norm": 0.5681532025337219,
      "learning_rate": 7.241502815246009e-07,
      "loss": 0.5031,
      "step": 3527
    },
    {
      "epoch": 0.6016371077762619,
      "grad_norm": 0.37529727816581726,
      "learning_rate": 7.236194338061365e-07,
      "loss": 0.5053,
      "step": 3528
    },
    {
      "epoch": 0.6018076398362893,
      "grad_norm": 0.6965998411178589,
      "learning_rate": 7.230886703994137e-07,
      "loss": 0.4997,
      "step": 3529
    },
    {
      "epoch": 0.6019781718963165,
      "grad_norm": 0.5670928359031677,
      "learning_rate": 7.225579914663453e-07,
      "loss": 0.4948,
      "step": 3530
    },
    {
      "epoch": 0.6021487039563438,
      "grad_norm": 0.5892690420150757,
      "learning_rate": 7.220273971688193e-07,
      "loss": 0.5054,
      "step": 3531
    },
    {
      "epoch": 0.602319236016371,
      "grad_norm": 0.45653554797172546,
      "learning_rate": 7.214968876686957e-07,
      "loss": 0.5055,
      "step": 3532
    },
    {
      "epoch": 0.6024897680763983,
      "grad_norm": 0.44902780652046204,
      "learning_rate": 7.209664631278113e-07,
      "loss": 0.5134,
      "step": 3533
    },
    {
      "epoch": 0.6026603001364257,
      "grad_norm": 0.4176679849624634,
      "learning_rate": 7.204361237079747e-07,
      "loss": 0.4995,
      "step": 3534
    },
    {
      "epoch": 0.602830832196453,
      "grad_norm": 0.39181068539619446,
      "learning_rate": 7.199058695709703e-07,
      "loss": 0.4956,
      "step": 3535
    },
    {
      "epoch": 0.6030013642564802,
      "grad_norm": 0.40236714482307434,
      "learning_rate": 7.19375700878555e-07,
      "loss": 0.5061,
      "step": 3536
    },
    {
      "epoch": 0.6031718963165075,
      "grad_norm": 0.4550640881061554,
      "learning_rate": 7.188456177924606e-07,
      "loss": 0.4971,
      "step": 3537
    },
    {
      "epoch": 0.6033424283765347,
      "grad_norm": 0.5484567880630493,
      "learning_rate": 7.183156204743926e-07,
      "loss": 0.5072,
      "step": 3538
    },
    {
      "epoch": 0.6035129604365621,
      "grad_norm": 0.4102098047733307,
      "learning_rate": 7.177857090860297e-07,
      "loss": 0.4999,
      "step": 3539
    },
    {
      "epoch": 0.6036834924965894,
      "grad_norm": 0.45663487911224365,
      "learning_rate": 7.172558837890256e-07,
      "loss": 0.4946,
      "step": 3540
    },
    {
      "epoch": 0.6038540245566166,
      "grad_norm": 0.43537938594818115,
      "learning_rate": 7.167261447450065e-07,
      "loss": 0.4949,
      "step": 3541
    },
    {
      "epoch": 0.6040245566166439,
      "grad_norm": 0.4498659670352936,
      "learning_rate": 7.161964921155733e-07,
      "loss": 0.5055,
      "step": 3542
    },
    {
      "epoch": 0.6041950886766713,
      "grad_norm": 0.538729727268219,
      "learning_rate": 7.156669260622998e-07,
      "loss": 0.5177,
      "step": 3543
    },
    {
      "epoch": 0.6043656207366985,
      "grad_norm": 0.3828926384449005,
      "learning_rate": 7.151374467467339e-07,
      "loss": 0.5022,
      "step": 3544
    },
    {
      "epoch": 0.6045361527967258,
      "grad_norm": 0.5743893384933472,
      "learning_rate": 7.146080543303966e-07,
      "loss": 0.496,
      "step": 3545
    },
    {
      "epoch": 0.604706684856753,
      "grad_norm": 0.46721726655960083,
      "learning_rate": 7.140787489747825e-07,
      "loss": 0.503,
      "step": 3546
    },
    {
      "epoch": 0.6048772169167803,
      "grad_norm": 0.4808953106403351,
      "learning_rate": 7.135495308413601e-07,
      "loss": 0.5032,
      "step": 3547
    },
    {
      "epoch": 0.6050477489768077,
      "grad_norm": 0.5239036083221436,
      "learning_rate": 7.130204000915702e-07,
      "loss": 0.4965,
      "step": 3548
    },
    {
      "epoch": 0.605218281036835,
      "grad_norm": 0.3505937457084656,
      "learning_rate": 7.124913568868288e-07,
      "loss": 0.5052,
      "step": 3549
    },
    {
      "epoch": 0.6053888130968622,
      "grad_norm": 0.5914868712425232,
      "learning_rate": 7.119624013885232e-07,
      "loss": 0.4989,
      "step": 3550
    },
    {
      "epoch": 0.6055593451568895,
      "grad_norm": 0.5071284770965576,
      "learning_rate": 7.114335337580153e-07,
      "loss": 0.498,
      "step": 3551
    },
    {
      "epoch": 0.6057298772169167,
      "grad_norm": 0.574246883392334,
      "learning_rate": 7.109047541566392e-07,
      "loss": 0.4953,
      "step": 3552
    },
    {
      "epoch": 0.6059004092769441,
      "grad_norm": 0.4426918923854828,
      "learning_rate": 7.103760627457031e-07,
      "loss": 0.4996,
      "step": 3553
    },
    {
      "epoch": 0.6060709413369714,
      "grad_norm": 0.5500320196151733,
      "learning_rate": 7.098474596864879e-07,
      "loss": 0.5078,
      "step": 3554
    },
    {
      "epoch": 0.6062414733969986,
      "grad_norm": 0.6294930577278137,
      "learning_rate": 7.09318945140247e-07,
      "loss": 0.5169,
      "step": 3555
    },
    {
      "epoch": 0.6064120054570259,
      "grad_norm": 0.4358702600002289,
      "learning_rate": 7.087905192682079e-07,
      "loss": 0.5118,
      "step": 3556
    },
    {
      "epoch": 0.6065825375170532,
      "grad_norm": 0.5665766000747681,
      "learning_rate": 7.082621822315697e-07,
      "loss": 0.4931,
      "step": 3557
    },
    {
      "epoch": 0.6067530695770805,
      "grad_norm": 0.5058866739273071,
      "learning_rate": 7.07733934191506e-07,
      "loss": 0.5063,
      "step": 3558
    },
    {
      "epoch": 0.6069236016371078,
      "grad_norm": 0.40108829736709595,
      "learning_rate": 7.072057753091617e-07,
      "loss": 0.52,
      "step": 3559
    },
    {
      "epoch": 0.607094133697135,
      "grad_norm": 0.5396955609321594,
      "learning_rate": 7.066777057456561e-07,
      "loss": 0.5025,
      "step": 3560
    },
    {
      "epoch": 0.6072646657571623,
      "grad_norm": 0.4852167069911957,
      "learning_rate": 7.061497256620793e-07,
      "loss": 0.5017,
      "step": 3561
    },
    {
      "epoch": 0.6074351978171897,
      "grad_norm": 0.39662599563598633,
      "learning_rate": 7.056218352194958e-07,
      "loss": 0.5023,
      "step": 3562
    },
    {
      "epoch": 0.607605729877217,
      "grad_norm": 0.4817587733268738,
      "learning_rate": 7.050940345789419e-07,
      "loss": 0.5086,
      "step": 3563
    },
    {
      "epoch": 0.6077762619372442,
      "grad_norm": 0.42608875036239624,
      "learning_rate": 7.045663239014266e-07,
      "loss": 0.4994,
      "step": 3564
    },
    {
      "epoch": 0.6079467939972715,
      "grad_norm": 0.36719074845314026,
      "learning_rate": 7.04038703347932e-07,
      "loss": 0.5057,
      "step": 3565
    },
    {
      "epoch": 0.6081173260572987,
      "grad_norm": 0.4195724129676819,
      "learning_rate": 7.035111730794116e-07,
      "loss": 0.4947,
      "step": 3566
    },
    {
      "epoch": 0.6082878581173261,
      "grad_norm": 0.40635064244270325,
      "learning_rate": 7.029837332567929e-07,
      "loss": 0.5008,
      "step": 3567
    },
    {
      "epoch": 0.6084583901773534,
      "grad_norm": 0.45059990882873535,
      "learning_rate": 7.024563840409746e-07,
      "loss": 0.4961,
      "step": 3568
    },
    {
      "epoch": 0.6086289222373806,
      "grad_norm": 0.42194345593452454,
      "learning_rate": 7.019291255928274e-07,
      "loss": 0.4958,
      "step": 3569
    },
    {
      "epoch": 0.6087994542974079,
      "grad_norm": 0.39710208773612976,
      "learning_rate": 7.01401958073196e-07,
      "loss": 0.5048,
      "step": 3570
    },
    {
      "epoch": 0.6089699863574352,
      "grad_norm": 0.4051033556461334,
      "learning_rate": 7.008748816428959e-07,
      "loss": 0.4988,
      "step": 3571
    },
    {
      "epoch": 0.6091405184174625,
      "grad_norm": 0.34922194480895996,
      "learning_rate": 7.003478964627157e-07,
      "loss": 0.502,
      "step": 3572
    },
    {
      "epoch": 0.6093110504774898,
      "grad_norm": 0.44243019819259644,
      "learning_rate": 6.998210026934149e-07,
      "loss": 0.5043,
      "step": 3573
    },
    {
      "epoch": 0.609481582537517,
      "grad_norm": 0.4132860600948334,
      "learning_rate": 6.992942004957271e-07,
      "loss": 0.4976,
      "step": 3574
    },
    {
      "epoch": 0.6096521145975443,
      "grad_norm": 0.49820494651794434,
      "learning_rate": 6.987674900303559e-07,
      "loss": 0.5086,
      "step": 3575
    },
    {
      "epoch": 0.6098226466575716,
      "grad_norm": 0.42352399230003357,
      "learning_rate": 6.982408714579783e-07,
      "loss": 0.5066,
      "step": 3576
    },
    {
      "epoch": 0.609993178717599,
      "grad_norm": 0.3854990303516388,
      "learning_rate": 6.97714344939243e-07,
      "loss": 0.5026,
      "step": 3577
    },
    {
      "epoch": 0.6101637107776262,
      "grad_norm": 0.39532220363616943,
      "learning_rate": 6.971879106347699e-07,
      "loss": 0.5048,
      "step": 3578
    },
    {
      "epoch": 0.6103342428376535,
      "grad_norm": 0.4156137704849243,
      "learning_rate": 6.966615687051517e-07,
      "loss": 0.499,
      "step": 3579
    },
    {
      "epoch": 0.6105047748976807,
      "grad_norm": 0.4135255813598633,
      "learning_rate": 6.961353193109524e-07,
      "loss": 0.5068,
      "step": 3580
    },
    {
      "epoch": 0.610675306957708,
      "grad_norm": 0.4143788516521454,
      "learning_rate": 6.95609162612708e-07,
      "loss": 0.4992,
      "step": 3581
    },
    {
      "epoch": 0.6108458390177354,
      "grad_norm": 0.42542439699172974,
      "learning_rate": 6.950830987709257e-07,
      "loss": 0.4977,
      "step": 3582
    },
    {
      "epoch": 0.6110163710777626,
      "grad_norm": 0.42292311787605286,
      "learning_rate": 6.945571279460857e-07,
      "loss": 0.495,
      "step": 3583
    },
    {
      "epoch": 0.6111869031377899,
      "grad_norm": 0.46934792399406433,
      "learning_rate": 6.940312502986381e-07,
      "loss": 0.5119,
      "step": 3584
    },
    {
      "epoch": 0.6113574351978172,
      "grad_norm": 0.4840346574783325,
      "learning_rate": 6.935054659890053e-07,
      "loss": 0.4886,
      "step": 3585
    },
    {
      "epoch": 0.6115279672578445,
      "grad_norm": 0.3723541498184204,
      "learning_rate": 6.92979775177582e-07,
      "loss": 0.5021,
      "step": 3586
    },
    {
      "epoch": 0.6116984993178718,
      "grad_norm": 0.4317921996116638,
      "learning_rate": 6.924541780247334e-07,
      "loss": 0.4999,
      "step": 3587
    },
    {
      "epoch": 0.611869031377899,
      "grad_norm": 0.5158261656761169,
      "learning_rate": 6.919286746907964e-07,
      "loss": 0.4978,
      "step": 3588
    },
    {
      "epoch": 0.6120395634379263,
      "grad_norm": 0.5118577480316162,
      "learning_rate": 6.914032653360792e-07,
      "loss": 0.5101,
      "step": 3589
    },
    {
      "epoch": 0.6122100954979536,
      "grad_norm": 0.4414294362068176,
      "learning_rate": 6.908779501208617e-07,
      "loss": 0.4923,
      "step": 3590
    },
    {
      "epoch": 0.612380627557981,
      "grad_norm": 0.36989259719848633,
      "learning_rate": 6.903527292053943e-07,
      "loss": 0.5025,
      "step": 3591
    },
    {
      "epoch": 0.6125511596180082,
      "grad_norm": 0.47652095556259155,
      "learning_rate": 6.898276027499003e-07,
      "loss": 0.5042,
      "step": 3592
    },
    {
      "epoch": 0.6127216916780355,
      "grad_norm": 0.4678318202495575,
      "learning_rate": 6.893025709145718e-07,
      "loss": 0.506,
      "step": 3593
    },
    {
      "epoch": 0.6128922237380627,
      "grad_norm": 0.33233606815338135,
      "learning_rate": 6.887776338595736e-07,
      "loss": 0.4991,
      "step": 3594
    },
    {
      "epoch": 0.61306275579809,
      "grad_norm": 0.41297683119773865,
      "learning_rate": 6.882527917450416e-07,
      "loss": 0.509,
      "step": 3595
    },
    {
      "epoch": 0.6132332878581174,
      "grad_norm": 0.374800443649292,
      "learning_rate": 6.877280447310822e-07,
      "loss": 0.4957,
      "step": 3596
    },
    {
      "epoch": 0.6134038199181446,
      "grad_norm": 0.40728360414505005,
      "learning_rate": 6.872033929777731e-07,
      "loss": 0.511,
      "step": 3597
    },
    {
      "epoch": 0.6135743519781719,
      "grad_norm": 0.4163961112499237,
      "learning_rate": 6.866788366451625e-07,
      "loss": 0.5038,
      "step": 3598
    },
    {
      "epoch": 0.6137448840381992,
      "grad_norm": 0.39934197068214417,
      "learning_rate": 6.861543758932703e-07,
      "loss": 0.4998,
      "step": 3599
    },
    {
      "epoch": 0.6139154160982264,
      "grad_norm": 0.41028591990470886,
      "learning_rate": 6.856300108820865e-07,
      "loss": 0.5018,
      "step": 3600
    },
    {
      "epoch": 0.6140859481582538,
      "grad_norm": 0.5780900120735168,
      "learning_rate": 6.851057417715719e-07,
      "loss": 0.5085,
      "step": 3601
    },
    {
      "epoch": 0.614256480218281,
      "grad_norm": 0.39337801933288574,
      "learning_rate": 6.845815687216591e-07,
      "loss": 0.5039,
      "step": 3602
    },
    {
      "epoch": 0.6144270122783083,
      "grad_norm": 0.42160749435424805,
      "learning_rate": 6.840574918922494e-07,
      "loss": 0.5064,
      "step": 3603
    },
    {
      "epoch": 0.6145975443383356,
      "grad_norm": 0.3935078978538513,
      "learning_rate": 6.83533511443217e-07,
      "loss": 0.5063,
      "step": 3604
    },
    {
      "epoch": 0.6147680763983628,
      "grad_norm": 0.4175839424133301,
      "learning_rate": 6.830096275344052e-07,
      "loss": 0.5025,
      "step": 3605
    },
    {
      "epoch": 0.6149386084583902,
      "grad_norm": 0.4644913673400879,
      "learning_rate": 6.824858403256284e-07,
      "loss": 0.5059,
      "step": 3606
    },
    {
      "epoch": 0.6151091405184175,
      "grad_norm": 0.5089910626411438,
      "learning_rate": 6.819621499766714e-07,
      "loss": 0.5097,
      "step": 3607
    },
    {
      "epoch": 0.6152796725784447,
      "grad_norm": 0.39027276635169983,
      "learning_rate": 6.814385566472893e-07,
      "loss": 0.5003,
      "step": 3608
    },
    {
      "epoch": 0.615450204638472,
      "grad_norm": 0.5722939372062683,
      "learning_rate": 6.809150604972079e-07,
      "loss": 0.506,
      "step": 3609
    },
    {
      "epoch": 0.6156207366984994,
      "grad_norm": 0.8283845782279968,
      "learning_rate": 6.803916616861228e-07,
      "loss": 0.5132,
      "step": 3610
    },
    {
      "epoch": 0.6157912687585266,
      "grad_norm": 0.5132255554199219,
      "learning_rate": 6.798683603737013e-07,
      "loss": 0.5057,
      "step": 3611
    },
    {
      "epoch": 0.6159618008185539,
      "grad_norm": 0.46312084794044495,
      "learning_rate": 6.793451567195786e-07,
      "loss": 0.5008,
      "step": 3612
    },
    {
      "epoch": 0.6161323328785812,
      "grad_norm": 0.6205770373344421,
      "learning_rate": 6.788220508833628e-07,
      "loss": 0.498,
      "step": 3613
    },
    {
      "epoch": 0.6163028649386084,
      "grad_norm": 0.4256286025047302,
      "learning_rate": 6.7829904302463e-07,
      "loss": 0.5111,
      "step": 3614
    },
    {
      "epoch": 0.6164733969986358,
      "grad_norm": 0.5326489806175232,
      "learning_rate": 6.777761333029275e-07,
      "loss": 0.5034,
      "step": 3615
    },
    {
      "epoch": 0.616643929058663,
      "grad_norm": 0.4669535458087921,
      "learning_rate": 6.772533218777725e-07,
      "loss": 0.502,
      "step": 3616
    },
    {
      "epoch": 0.6168144611186903,
      "grad_norm": 0.4774930775165558,
      "learning_rate": 6.767306089086519e-07,
      "loss": 0.5081,
      "step": 3617
    },
    {
      "epoch": 0.6169849931787176,
      "grad_norm": 0.6294277906417847,
      "learning_rate": 6.762079945550231e-07,
      "loss": 0.5053,
      "step": 3618
    },
    {
      "epoch": 0.6171555252387448,
      "grad_norm": 0.4758334755897522,
      "learning_rate": 6.756854789763124e-07,
      "loss": 0.5065,
      "step": 3619
    },
    {
      "epoch": 0.6173260572987722,
      "grad_norm": 0.545806884765625,
      "learning_rate": 6.75163062331918e-07,
      "loss": 0.4929,
      "step": 3620
    },
    {
      "epoch": 0.6174965893587995,
      "grad_norm": 0.539008617401123,
      "learning_rate": 6.74640744781205e-07,
      "loss": 0.4986,
      "step": 3621
    },
    {
      "epoch": 0.6176671214188267,
      "grad_norm": 0.47142142057418823,
      "learning_rate": 6.741185264835111e-07,
      "loss": 0.4932,
      "step": 3622
    },
    {
      "epoch": 0.617837653478854,
      "grad_norm": 0.7239853739738464,
      "learning_rate": 6.735964075981421e-07,
      "loss": 0.512,
      "step": 3623
    },
    {
      "epoch": 0.6180081855388813,
      "grad_norm": 0.5947273373603821,
      "learning_rate": 6.730743882843736e-07,
      "loss": 0.5102,
      "step": 3624
    },
    {
      "epoch": 0.6181787175989086,
      "grad_norm": 0.4315265715122223,
      "learning_rate": 6.725524687014515e-07,
      "loss": 0.4982,
      "step": 3625
    },
    {
      "epoch": 0.6183492496589359,
      "grad_norm": 0.6664741039276123,
      "learning_rate": 6.720306490085905e-07,
      "loss": 0.5013,
      "step": 3626
    },
    {
      "epoch": 0.6185197817189632,
      "grad_norm": 0.42937299609184265,
      "learning_rate": 6.715089293649753e-07,
      "loss": 0.5004,
      "step": 3627
    },
    {
      "epoch": 0.6186903137789904,
      "grad_norm": 0.5188443064689636,
      "learning_rate": 6.709873099297597e-07,
      "loss": 0.492,
      "step": 3628
    },
    {
      "epoch": 0.6188608458390177,
      "grad_norm": 0.6457641124725342,
      "learning_rate": 6.704657908620681e-07,
      "loss": 0.4971,
      "step": 3629
    },
    {
      "epoch": 0.619031377899045,
      "grad_norm": 0.39259135723114014,
      "learning_rate": 6.699443723209925e-07,
      "loss": 0.5043,
      "step": 3630
    },
    {
      "epoch": 0.6192019099590723,
      "grad_norm": 0.5514686703681946,
      "learning_rate": 6.694230544655956e-07,
      "loss": 0.504,
      "step": 3631
    },
    {
      "epoch": 0.6193724420190996,
      "grad_norm": 0.4360816180706024,
      "learning_rate": 6.689018374549087e-07,
      "loss": 0.5099,
      "step": 3632
    },
    {
      "epoch": 0.6195429740791268,
      "grad_norm": 0.576634407043457,
      "learning_rate": 6.683807214479324e-07,
      "loss": 0.4997,
      "step": 3633
    },
    {
      "epoch": 0.6197135061391542,
      "grad_norm": 0.6791767477989197,
      "learning_rate": 6.678597066036369e-07,
      "loss": 0.4935,
      "step": 3634
    },
    {
      "epoch": 0.6198840381991815,
      "grad_norm": 0.4800211787223816,
      "learning_rate": 6.67338793080961e-07,
      "loss": 0.4976,
      "step": 3635
    },
    {
      "epoch": 0.6200545702592087,
      "grad_norm": 0.5554141998291016,
      "learning_rate": 6.668179810388132e-07,
      "loss": 0.5091,
      "step": 3636
    },
    {
      "epoch": 0.620225102319236,
      "grad_norm": 0.4043811559677124,
      "learning_rate": 6.6629727063607e-07,
      "loss": 0.4953,
      "step": 3637
    },
    {
      "epoch": 0.6203956343792633,
      "grad_norm": 0.5499111413955688,
      "learning_rate": 6.657766620315784e-07,
      "loss": 0.5081,
      "step": 3638
    },
    {
      "epoch": 0.6205661664392906,
      "grad_norm": 0.6283267736434937,
      "learning_rate": 6.652561553841537e-07,
      "loss": 0.5007,
      "step": 3639
    },
    {
      "epoch": 0.6207366984993179,
      "grad_norm": 0.427605539560318,
      "learning_rate": 6.647357508525787e-07,
      "loss": 0.5107,
      "step": 3640
    },
    {
      "epoch": 0.6209072305593452,
      "grad_norm": 0.44812512397766113,
      "learning_rate": 6.642154485956077e-07,
      "loss": 0.5022,
      "step": 3641
    },
    {
      "epoch": 0.6210777626193724,
      "grad_norm": 0.535152792930603,
      "learning_rate": 6.636952487719613e-07,
      "loss": 0.5043,
      "step": 3642
    },
    {
      "epoch": 0.6212482946793997,
      "grad_norm": 0.4807773232460022,
      "learning_rate": 6.631751515403307e-07,
      "loss": 0.4924,
      "step": 3643
    },
    {
      "epoch": 0.621418826739427,
      "grad_norm": 0.4906376004219055,
      "learning_rate": 6.626551570593744e-07,
      "loss": 0.4986,
      "step": 3644
    },
    {
      "epoch": 0.6215893587994543,
      "grad_norm": 0.47907671332359314,
      "learning_rate": 6.621352654877207e-07,
      "loss": 0.4909,
      "step": 3645
    },
    {
      "epoch": 0.6217598908594816,
      "grad_norm": 0.4234660267829895,
      "learning_rate": 6.616154769839656e-07,
      "loss": 0.5065,
      "step": 3646
    },
    {
      "epoch": 0.6219304229195088,
      "grad_norm": 0.4363880753517151,
      "learning_rate": 6.610957917066745e-07,
      "loss": 0.5032,
      "step": 3647
    },
    {
      "epoch": 0.6221009549795361,
      "grad_norm": 0.4534679055213928,
      "learning_rate": 6.605762098143811e-07,
      "loss": 0.4969,
      "step": 3648
    },
    {
      "epoch": 0.6222714870395635,
      "grad_norm": 0.38198530673980713,
      "learning_rate": 6.60056731465586e-07,
      "loss": 0.5005,
      "step": 3649
    },
    {
      "epoch": 0.6224420190995907,
      "grad_norm": 0.3780840039253235,
      "learning_rate": 6.59537356818761e-07,
      "loss": 0.5052,
      "step": 3650
    },
    {
      "epoch": 0.622612551159618,
      "grad_norm": 0.42172762751579285,
      "learning_rate": 6.59018086032344e-07,
      "loss": 0.5041,
      "step": 3651
    },
    {
      "epoch": 0.6227830832196453,
      "grad_norm": 0.44250547885894775,
      "learning_rate": 6.584989192647425e-07,
      "loss": 0.4965,
      "step": 3652
    },
    {
      "epoch": 0.6229536152796725,
      "grad_norm": 0.4835226833820343,
      "learning_rate": 6.579798566743315e-07,
      "loss": 0.5082,
      "step": 3653
    },
    {
      "epoch": 0.6231241473396999,
      "grad_norm": 0.38454940915107727,
      "learning_rate": 6.574608984194546e-07,
      "loss": 0.5045,
      "step": 3654
    },
    {
      "epoch": 0.6232946793997272,
      "grad_norm": 0.5638027787208557,
      "learning_rate": 6.569420446584235e-07,
      "loss": 0.5024,
      "step": 3655
    },
    {
      "epoch": 0.6234652114597544,
      "grad_norm": 0.3625916838645935,
      "learning_rate": 6.564232955495176e-07,
      "loss": 0.5046,
      "step": 3656
    },
    {
      "epoch": 0.6236357435197817,
      "grad_norm": 0.4334798753261566,
      "learning_rate": 6.55904651250986e-07,
      "loss": 0.5027,
      "step": 3657
    },
    {
      "epoch": 0.623806275579809,
      "grad_norm": 0.40530920028686523,
      "learning_rate": 6.55386111921043e-07,
      "loss": 0.5066,
      "step": 3658
    },
    {
      "epoch": 0.6239768076398363,
      "grad_norm": 0.47581782937049866,
      "learning_rate": 6.548676777178738e-07,
      "loss": 0.5055,
      "step": 3659
    },
    {
      "epoch": 0.6241473396998636,
      "grad_norm": 0.42550137639045715,
      "learning_rate": 6.543493487996294e-07,
      "loss": 0.4991,
      "step": 3660
    },
    {
      "epoch": 0.6243178717598908,
      "grad_norm": 0.41812145709991455,
      "learning_rate": 6.538311253244301e-07,
      "loss": 0.5032,
      "step": 3661
    },
    {
      "epoch": 0.6244884038199181,
      "grad_norm": 0.5709978938102722,
      "learning_rate": 6.533130074503631e-07,
      "loss": 0.5026,
      "step": 3662
    },
    {
      "epoch": 0.6246589358799455,
      "grad_norm": 0.5457199215888977,
      "learning_rate": 6.527949953354836e-07,
      "loss": 0.503,
      "step": 3663
    },
    {
      "epoch": 0.6248294679399727,
      "grad_norm": 0.44440513849258423,
      "learning_rate": 6.522770891378151e-07,
      "loss": 0.5018,
      "step": 3664
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.46256470680236816,
      "learning_rate": 6.517592890153478e-07,
      "loss": 0.5143,
      "step": 3665
    },
    {
      "epoch": 0.6251705320600273,
      "grad_norm": 0.39488309621810913,
      "learning_rate": 6.512415951260406e-07,
      "loss": 0.4868,
      "step": 3666
    },
    {
      "epoch": 0.6253410641200545,
      "grad_norm": 0.6106573939323425,
      "learning_rate": 6.507240076278193e-07,
      "loss": 0.4977,
      "step": 3667
    },
    {
      "epoch": 0.6255115961800819,
      "grad_norm": 0.37441346049308777,
      "learning_rate": 6.502065266785774e-07,
      "loss": 0.508,
      "step": 3668
    },
    {
      "epoch": 0.6256821282401092,
      "grad_norm": 0.5005650520324707,
      "learning_rate": 6.496891524361757e-07,
      "loss": 0.4986,
      "step": 3669
    },
    {
      "epoch": 0.6258526603001364,
      "grad_norm": 0.43681997060775757,
      "learning_rate": 6.491718850584434e-07,
      "loss": 0.5016,
      "step": 3670
    },
    {
      "epoch": 0.6260231923601637,
      "grad_norm": 0.5549169182777405,
      "learning_rate": 6.486547247031756e-07,
      "loss": 0.509,
      "step": 3671
    },
    {
      "epoch": 0.626193724420191,
      "grad_norm": 0.5520680546760559,
      "learning_rate": 6.481376715281358e-07,
      "loss": 0.5001,
      "step": 3672
    },
    {
      "epoch": 0.6263642564802183,
      "grad_norm": 0.545689046382904,
      "learning_rate": 6.476207256910547e-07,
      "loss": 0.5075,
      "step": 3673
    },
    {
      "epoch": 0.6265347885402456,
      "grad_norm": 0.4276546239852905,
      "learning_rate": 6.471038873496296e-07,
      "loss": 0.511,
      "step": 3674
    },
    {
      "epoch": 0.6267053206002728,
      "grad_norm": 0.3931770920753479,
      "learning_rate": 6.465871566615264e-07,
      "loss": 0.4926,
      "step": 3675
    },
    {
      "epoch": 0.6268758526603001,
      "grad_norm": 0.436468243598938,
      "learning_rate": 6.460705337843762e-07,
      "loss": 0.4976,
      "step": 3676
    },
    {
      "epoch": 0.6270463847203275,
      "grad_norm": 0.41146066784858704,
      "learning_rate": 6.45554018875779e-07,
      "loss": 0.5098,
      "step": 3677
    },
    {
      "epoch": 0.6272169167803547,
      "grad_norm": 0.433461457490921,
      "learning_rate": 6.450376120933009e-07,
      "loss": 0.5072,
      "step": 3678
    },
    {
      "epoch": 0.627387448840382,
      "grad_norm": 0.42208781838417053,
      "learning_rate": 6.445213135944752e-07,
      "loss": 0.5074,
      "step": 3679
    },
    {
      "epoch": 0.6275579809004093,
      "grad_norm": 0.3760968744754791,
      "learning_rate": 6.440051235368023e-07,
      "loss": 0.4948,
      "step": 3680
    },
    {
      "epoch": 0.6277285129604365,
      "grad_norm": 0.4705601632595062,
      "learning_rate": 6.434890420777492e-07,
      "loss": 0.4915,
      "step": 3681
    },
    {
      "epoch": 0.6278990450204639,
      "grad_norm": 0.5115656852722168,
      "learning_rate": 6.429730693747504e-07,
      "loss": 0.5061,
      "step": 3682
    },
    {
      "epoch": 0.6280695770804912,
      "grad_norm": 0.4798704981803894,
      "learning_rate": 6.424572055852062e-07,
      "loss": 0.4926,
      "step": 3683
    },
    {
      "epoch": 0.6282401091405184,
      "grad_norm": 0.46192893385887146,
      "learning_rate": 6.419414508664851e-07,
      "loss": 0.5057,
      "step": 3684
    },
    {
      "epoch": 0.6284106412005457,
      "grad_norm": 0.46332821249961853,
      "learning_rate": 6.414258053759211e-07,
      "loss": 0.5008,
      "step": 3685
    },
    {
      "epoch": 0.628581173260573,
      "grad_norm": 0.4708835780620575,
      "learning_rate": 6.409102692708154e-07,
      "loss": 0.5044,
      "step": 3686
    },
    {
      "epoch": 0.6287517053206003,
      "grad_norm": 0.4883918762207031,
      "learning_rate": 6.403948427084357e-07,
      "loss": 0.5019,
      "step": 3687
    },
    {
      "epoch": 0.6289222373806276,
      "grad_norm": 0.5046584010124207,
      "learning_rate": 6.39879525846016e-07,
      "loss": 0.4981,
      "step": 3688
    },
    {
      "epoch": 0.6290927694406548,
      "grad_norm": 0.5189499258995056,
      "learning_rate": 6.393643188407579e-07,
      "loss": 0.4936,
      "step": 3689
    },
    {
      "epoch": 0.6292633015006821,
      "grad_norm": 0.4646134376525879,
      "learning_rate": 6.388492218498277e-07,
      "loss": 0.5089,
      "step": 3690
    },
    {
      "epoch": 0.6294338335607094,
      "grad_norm": 0.4980888366699219,
      "learning_rate": 6.383342350303606e-07,
      "loss": 0.5057,
      "step": 3691
    },
    {
      "epoch": 0.6296043656207367,
      "grad_norm": 0.36041122674942017,
      "learning_rate": 6.378193585394553e-07,
      "loss": 0.4913,
      "step": 3692
    },
    {
      "epoch": 0.629774897680764,
      "grad_norm": 0.44873934984207153,
      "learning_rate": 6.373045925341795e-07,
      "loss": 0.5044,
      "step": 3693
    },
    {
      "epoch": 0.6299454297407913,
      "grad_norm": 0.4433237910270691,
      "learning_rate": 6.367899371715654e-07,
      "loss": 0.5004,
      "step": 3694
    },
    {
      "epoch": 0.6301159618008185,
      "grad_norm": 0.383121132850647,
      "learning_rate": 6.362753926086119e-07,
      "loss": 0.4989,
      "step": 3695
    },
    {
      "epoch": 0.6302864938608458,
      "grad_norm": 0.43507349491119385,
      "learning_rate": 6.357609590022849e-07,
      "loss": 0.4991,
      "step": 3696
    },
    {
      "epoch": 0.6304570259208732,
      "grad_norm": 0.34954577684402466,
      "learning_rate": 6.352466365095151e-07,
      "loss": 0.5079,
      "step": 3697
    },
    {
      "epoch": 0.6306275579809004,
      "grad_norm": 0.36680907011032104,
      "learning_rate": 6.347324252872007e-07,
      "loss": 0.5076,
      "step": 3698
    },
    {
      "epoch": 0.6307980900409277,
      "grad_norm": 0.35321205854415894,
      "learning_rate": 6.342183254922046e-07,
      "loss": 0.5053,
      "step": 3699
    },
    {
      "epoch": 0.630968622100955,
      "grad_norm": 0.41826412081718445,
      "learning_rate": 6.337043372813573e-07,
      "loss": 0.4961,
      "step": 3700
    },
    {
      "epoch": 0.6311391541609823,
      "grad_norm": 0.3832581341266632,
      "learning_rate": 6.331904608114535e-07,
      "loss": 0.5133,
      "step": 3701
    },
    {
      "epoch": 0.6313096862210096,
      "grad_norm": 0.34321901202201843,
      "learning_rate": 6.326766962392545e-07,
      "loss": 0.496,
      "step": 3702
    },
    {
      "epoch": 0.6314802182810368,
      "grad_norm": 0.40280675888061523,
      "learning_rate": 6.321630437214885e-07,
      "loss": 0.5077,
      "step": 3703
    },
    {
      "epoch": 0.6316507503410641,
      "grad_norm": 0.4523974657058716,
      "learning_rate": 6.316495034148481e-07,
      "loss": 0.5223,
      "step": 3704
    },
    {
      "epoch": 0.6318212824010914,
      "grad_norm": 0.4422217011451721,
      "learning_rate": 6.311360754759924e-07,
      "loss": 0.5198,
      "step": 3705
    },
    {
      "epoch": 0.6319918144611187,
      "grad_norm": 0.4404323399066925,
      "learning_rate": 6.306227600615459e-07,
      "loss": 0.5003,
      "step": 3706
    },
    {
      "epoch": 0.632162346521146,
      "grad_norm": 0.4646601378917694,
      "learning_rate": 6.301095573280989e-07,
      "loss": 0.507,
      "step": 3707
    },
    {
      "epoch": 0.6323328785811733,
      "grad_norm": 0.5132860541343689,
      "learning_rate": 6.29596467432207e-07,
      "loss": 0.5073,
      "step": 3708
    },
    {
      "epoch": 0.6325034106412005,
      "grad_norm": 0.5008794069290161,
      "learning_rate": 6.290834905303927e-07,
      "loss": 0.488,
      "step": 3709
    },
    {
      "epoch": 0.6326739427012278,
      "grad_norm": 0.4249769449234009,
      "learning_rate": 6.285706267791424e-07,
      "loss": 0.4868,
      "step": 3710
    },
    {
      "epoch": 0.6328444747612552,
      "grad_norm": 0.4569952189922333,
      "learning_rate": 6.28057876334908e-07,
      "loss": 0.518,
      "step": 3711
    },
    {
      "epoch": 0.6330150068212824,
      "grad_norm": 0.45532703399658203,
      "learning_rate": 6.275452393541085e-07,
      "loss": 0.5047,
      "step": 3712
    },
    {
      "epoch": 0.6331855388813097,
      "grad_norm": 0.448268860578537,
      "learning_rate": 6.270327159931267e-07,
      "loss": 0.494,
      "step": 3713
    },
    {
      "epoch": 0.633356070941337,
      "grad_norm": 0.4130880534648895,
      "learning_rate": 6.265203064083117e-07,
      "loss": 0.4947,
      "step": 3714
    },
    {
      "epoch": 0.6335266030013642,
      "grad_norm": 0.455447793006897,
      "learning_rate": 6.260080107559768e-07,
      "loss": 0.5032,
      "step": 3715
    },
    {
      "epoch": 0.6336971350613916,
      "grad_norm": 0.3961569368839264,
      "learning_rate": 6.254958291924018e-07,
      "loss": 0.5032,
      "step": 3716
    },
    {
      "epoch": 0.6338676671214188,
      "grad_norm": 0.42894527316093445,
      "learning_rate": 6.249837618738311e-07,
      "loss": 0.5112,
      "step": 3717
    },
    {
      "epoch": 0.6340381991814461,
      "grad_norm": 0.47216030955314636,
      "learning_rate": 6.244718089564735e-07,
      "loss": 0.4972,
      "step": 3718
    },
    {
      "epoch": 0.6342087312414734,
      "grad_norm": 0.49208471179008484,
      "learning_rate": 6.23959970596505e-07,
      "loss": 0.4937,
      "step": 3719
    },
    {
      "epoch": 0.6343792633015006,
      "grad_norm": 0.4241389334201813,
      "learning_rate": 6.23448246950064e-07,
      "loss": 0.5003,
      "step": 3720
    },
    {
      "epoch": 0.634549795361528,
      "grad_norm": 0.3937288522720337,
      "learning_rate": 6.229366381732561e-07,
      "loss": 0.5108,
      "step": 3721
    },
    {
      "epoch": 0.6347203274215553,
      "grad_norm": 0.3928370177745819,
      "learning_rate": 6.224251444221503e-07,
      "loss": 0.5171,
      "step": 3722
    },
    {
      "epoch": 0.6348908594815825,
      "grad_norm": 0.42944595217704773,
      "learning_rate": 6.219137658527819e-07,
      "loss": 0.5072,
      "step": 3723
    },
    {
      "epoch": 0.6350613915416098,
      "grad_norm": 0.5143080949783325,
      "learning_rate": 6.2140250262115e-07,
      "loss": 0.4947,
      "step": 3724
    },
    {
      "epoch": 0.6352319236016372,
      "grad_norm": 0.339793860912323,
      "learning_rate": 6.208913548832188e-07,
      "loss": 0.5139,
      "step": 3725
    },
    {
      "epoch": 0.6354024556616644,
      "grad_norm": 0.4634465277194977,
      "learning_rate": 6.203803227949174e-07,
      "loss": 0.5131,
      "step": 3726
    },
    {
      "epoch": 0.6355729877216917,
      "grad_norm": 0.5924431085586548,
      "learning_rate": 6.198694065121394e-07,
      "loss": 0.4931,
      "step": 3727
    },
    {
      "epoch": 0.635743519781719,
      "grad_norm": 0.4708423912525177,
      "learning_rate": 6.19358606190744e-07,
      "loss": 0.4975,
      "step": 3728
    },
    {
      "epoch": 0.6359140518417462,
      "grad_norm": 0.400854229927063,
      "learning_rate": 6.188479219865529e-07,
      "loss": 0.5068,
      "step": 3729
    },
    {
      "epoch": 0.6360845839017736,
      "grad_norm": 0.49634453654289246,
      "learning_rate": 6.18337354055355e-07,
      "loss": 0.5124,
      "step": 3730
    },
    {
      "epoch": 0.6362551159618008,
      "grad_norm": 0.41015246510505676,
      "learning_rate": 6.178269025529018e-07,
      "loss": 0.4981,
      "step": 3731
    },
    {
      "epoch": 0.6364256480218281,
      "grad_norm": 0.42830201983451843,
      "learning_rate": 6.173165676349103e-07,
      "loss": 0.5096,
      "step": 3732
    },
    {
      "epoch": 0.6365961800818554,
      "grad_norm": 0.5014932751655579,
      "learning_rate": 6.168063494570615e-07,
      "loss": 0.4874,
      "step": 3733
    },
    {
      "epoch": 0.6367667121418826,
      "grad_norm": 0.5043054819107056,
      "learning_rate": 6.162962481750005e-07,
      "loss": 0.5184,
      "step": 3734
    },
    {
      "epoch": 0.63693724420191,
      "grad_norm": 0.4094924330711365,
      "learning_rate": 6.157862639443376e-07,
      "loss": 0.5177,
      "step": 3735
    },
    {
      "epoch": 0.6371077762619373,
      "grad_norm": 0.43844693899154663,
      "learning_rate": 6.152763969206461e-07,
      "loss": 0.5061,
      "step": 3736
    },
    {
      "epoch": 0.6372783083219645,
      "grad_norm": 0.48379039764404297,
      "learning_rate": 6.147666472594658e-07,
      "loss": 0.4945,
      "step": 3737
    },
    {
      "epoch": 0.6374488403819918,
      "grad_norm": 0.4411104619503021,
      "learning_rate": 6.142570151162978e-07,
      "loss": 0.4976,
      "step": 3738
    },
    {
      "epoch": 0.637619372442019,
      "grad_norm": 0.45324018597602844,
      "learning_rate": 6.137475006466097e-07,
      "loss": 0.5014,
      "step": 3739
    },
    {
      "epoch": 0.6377899045020464,
      "grad_norm": 0.4081723093986511,
      "learning_rate": 6.132381040058318e-07,
      "loss": 0.5051,
      "step": 3740
    },
    {
      "epoch": 0.6379604365620737,
      "grad_norm": 0.5187481045722961,
      "learning_rate": 6.127288253493591e-07,
      "loss": 0.5075,
      "step": 3741
    },
    {
      "epoch": 0.638130968622101,
      "grad_norm": 0.4198801815509796,
      "learning_rate": 6.122196648325507e-07,
      "loss": 0.501,
      "step": 3742
    },
    {
      "epoch": 0.6383015006821282,
      "grad_norm": 0.40012413263320923,
      "learning_rate": 6.117106226107291e-07,
      "loss": 0.4998,
      "step": 3743
    },
    {
      "epoch": 0.6384720327421555,
      "grad_norm": 0.45775869488716125,
      "learning_rate": 6.112016988391811e-07,
      "loss": 0.5314,
      "step": 3744
    },
    {
      "epoch": 0.6386425648021828,
      "grad_norm": 0.4114890694618225,
      "learning_rate": 6.106928936731572e-07,
      "loss": 0.5002,
      "step": 3745
    },
    {
      "epoch": 0.6388130968622101,
      "grad_norm": 0.44415518641471863,
      "learning_rate": 6.101842072678725e-07,
      "loss": 0.5067,
      "step": 3746
    },
    {
      "epoch": 0.6389836289222374,
      "grad_norm": 0.43056219816207886,
      "learning_rate": 6.096756397785041e-07,
      "loss": 0.5057,
      "step": 3747
    },
    {
      "epoch": 0.6391541609822646,
      "grad_norm": 0.3749973475933075,
      "learning_rate": 6.091671913601949e-07,
      "loss": 0.5136,
      "step": 3748
    },
    {
      "epoch": 0.639324693042292,
      "grad_norm": 0.46487438678741455,
      "learning_rate": 6.086588621680499e-07,
      "loss": 0.5059,
      "step": 3749
    },
    {
      "epoch": 0.6394952251023193,
      "grad_norm": 0.39657068252563477,
      "learning_rate": 6.081506523571385e-07,
      "loss": 0.5162,
      "step": 3750
    },
    {
      "epoch": 0.6396657571623465,
      "grad_norm": 0.45455026626586914,
      "learning_rate": 6.076425620824936e-07,
      "loss": 0.4889,
      "step": 3751
    },
    {
      "epoch": 0.6398362892223738,
      "grad_norm": 0.46647676825523376,
      "learning_rate": 6.071345914991114e-07,
      "loss": 0.5121,
      "step": 3752
    },
    {
      "epoch": 0.640006821282401,
      "grad_norm": 0.43015286326408386,
      "learning_rate": 6.06626740761952e-07,
      "loss": 0.5095,
      "step": 3753
    },
    {
      "epoch": 0.6401773533424284,
      "grad_norm": 0.4729073643684387,
      "learning_rate": 6.061190100259378e-07,
      "loss": 0.498,
      "step": 3754
    },
    {
      "epoch": 0.6403478854024557,
      "grad_norm": 0.44330206513404846,
      "learning_rate": 6.056113994459567e-07,
      "loss": 0.5053,
      "step": 3755
    },
    {
      "epoch": 0.640518417462483,
      "grad_norm": 0.4980153441429138,
      "learning_rate": 6.051039091768583e-07,
      "loss": 0.5196,
      "step": 3756
    },
    {
      "epoch": 0.6406889495225102,
      "grad_norm": 0.5406501889228821,
      "learning_rate": 6.045965393734551e-07,
      "loss": 0.5112,
      "step": 3757
    },
    {
      "epoch": 0.6408594815825375,
      "grad_norm": 0.4644082486629486,
      "learning_rate": 6.040892901905247e-07,
      "loss": 0.512,
      "step": 3758
    },
    {
      "epoch": 0.6410300136425648,
      "grad_norm": 0.5462424159049988,
      "learning_rate": 6.03582161782806e-07,
      "loss": 0.4952,
      "step": 3759
    },
    {
      "epoch": 0.6412005457025921,
      "grad_norm": 0.35903412103652954,
      "learning_rate": 6.030751543050029e-07,
      "loss": 0.5066,
      "step": 3760
    },
    {
      "epoch": 0.6413710777626194,
      "grad_norm": 0.39143240451812744,
      "learning_rate": 6.025682679117804e-07,
      "loss": 0.5019,
      "step": 3761
    },
    {
      "epoch": 0.6415416098226466,
      "grad_norm": 0.44252634048461914,
      "learning_rate": 6.020615027577684e-07,
      "loss": 0.5188,
      "step": 3762
    },
    {
      "epoch": 0.6417121418826739,
      "grad_norm": 0.4334603250026703,
      "learning_rate": 6.015548589975582e-07,
      "loss": 0.5007,
      "step": 3763
    },
    {
      "epoch": 0.6418826739427013,
      "grad_norm": 0.5139437913894653,
      "learning_rate": 6.010483367857057e-07,
      "loss": 0.4994,
      "step": 3764
    },
    {
      "epoch": 0.6420532060027285,
      "grad_norm": 0.5231024622917175,
      "learning_rate": 6.005419362767287e-07,
      "loss": 0.4997,
      "step": 3765
    },
    {
      "epoch": 0.6422237380627558,
      "grad_norm": 0.5182217359542847,
      "learning_rate": 6.000356576251075e-07,
      "loss": 0.5038,
      "step": 3766
    },
    {
      "epoch": 0.642394270122783,
      "grad_norm": 0.5182235240936279,
      "learning_rate": 5.995295009852862e-07,
      "loss": 0.5166,
      "step": 3767
    },
    {
      "epoch": 0.6425648021828103,
      "grad_norm": 0.45456960797309875,
      "learning_rate": 5.990234665116713e-07,
      "loss": 0.513,
      "step": 3768
    },
    {
      "epoch": 0.6427353342428377,
      "grad_norm": 0.5825948119163513,
      "learning_rate": 5.985175543586322e-07,
      "loss": 0.5027,
      "step": 3769
    },
    {
      "epoch": 0.642905866302865,
      "grad_norm": 0.4596474766731262,
      "learning_rate": 5.980117646805002e-07,
      "loss": 0.501,
      "step": 3770
    },
    {
      "epoch": 0.6430763983628922,
      "grad_norm": 0.5294000506401062,
      "learning_rate": 5.975060976315703e-07,
      "loss": 0.5101,
      "step": 3771
    },
    {
      "epoch": 0.6432469304229195,
      "grad_norm": 0.4442268908023834,
      "learning_rate": 5.970005533660997e-07,
      "loss": 0.5208,
      "step": 3772
    },
    {
      "epoch": 0.6434174624829468,
      "grad_norm": 0.510656476020813,
      "learning_rate": 5.964951320383072e-07,
      "loss": 0.5096,
      "step": 3773
    },
    {
      "epoch": 0.6435879945429741,
      "grad_norm": 0.5467284321784973,
      "learning_rate": 5.959898338023765e-07,
      "loss": 0.5131,
      "step": 3774
    },
    {
      "epoch": 0.6437585266030014,
      "grad_norm": 0.5428006052970886,
      "learning_rate": 5.954846588124506e-07,
      "loss": 0.5052,
      "step": 3775
    },
    {
      "epoch": 0.6439290586630286,
      "grad_norm": 0.5226187109947205,
      "learning_rate": 5.949796072226375e-07,
      "loss": 0.5006,
      "step": 3776
    },
    {
      "epoch": 0.6440995907230559,
      "grad_norm": 0.45977094769477844,
      "learning_rate": 5.944746791870063e-07,
      "loss": 0.5037,
      "step": 3777
    },
    {
      "epoch": 0.6442701227830833,
      "grad_norm": 0.43148553371429443,
      "learning_rate": 5.939698748595887e-07,
      "loss": 0.5105,
      "step": 3778
    },
    {
      "epoch": 0.6444406548431105,
      "grad_norm": 0.44732269644737244,
      "learning_rate": 5.934651943943786e-07,
      "loss": 0.4943,
      "step": 3779
    },
    {
      "epoch": 0.6446111869031378,
      "grad_norm": 0.4386376142501831,
      "learning_rate": 5.92960637945332e-07,
      "loss": 0.5176,
      "step": 3780
    },
    {
      "epoch": 0.644781718963165,
      "grad_norm": 0.4881812036037445,
      "learning_rate": 5.924562056663677e-07,
      "loss": 0.5126,
      "step": 3781
    },
    {
      "epoch": 0.6449522510231923,
      "grad_norm": 0.42850565910339355,
      "learning_rate": 5.919518977113653e-07,
      "loss": 0.4969,
      "step": 3782
    },
    {
      "epoch": 0.6451227830832197,
      "grad_norm": 0.5664390325546265,
      "learning_rate": 5.914477142341683e-07,
      "loss": 0.4965,
      "step": 3783
    },
    {
      "epoch": 0.645293315143247,
      "grad_norm": 0.45635077357292175,
      "learning_rate": 5.909436553885806e-07,
      "loss": 0.5197,
      "step": 3784
    },
    {
      "epoch": 0.6454638472032742,
      "grad_norm": 0.5621646046638489,
      "learning_rate": 5.90439721328369e-07,
      "loss": 0.4968,
      "step": 3785
    },
    {
      "epoch": 0.6456343792633015,
      "grad_norm": 0.4447271227836609,
      "learning_rate": 5.899359122072619e-07,
      "loss": 0.5149,
      "step": 3786
    },
    {
      "epoch": 0.6458049113233287,
      "grad_norm": 0.5080307722091675,
      "learning_rate": 5.894322281789495e-07,
      "loss": 0.5152,
      "step": 3787
    },
    {
      "epoch": 0.6459754433833561,
      "grad_norm": 0.5570019483566284,
      "learning_rate": 5.889286693970844e-07,
      "loss": 0.5046,
      "step": 3788
    },
    {
      "epoch": 0.6461459754433834,
      "grad_norm": 0.394611120223999,
      "learning_rate": 5.884252360152801e-07,
      "loss": 0.5162,
      "step": 3789
    },
    {
      "epoch": 0.6463165075034106,
      "grad_norm": 0.5731703639030457,
      "learning_rate": 5.879219281871125e-07,
      "loss": 0.5103,
      "step": 3790
    },
    {
      "epoch": 0.6464870395634379,
      "grad_norm": 0.48665788769721985,
      "learning_rate": 5.874187460661188e-07,
      "loss": 0.4975,
      "step": 3791
    },
    {
      "epoch": 0.6466575716234653,
      "grad_norm": 0.503473162651062,
      "learning_rate": 5.869156898057984e-07,
      "loss": 0.5027,
      "step": 3792
    },
    {
      "epoch": 0.6468281036834925,
      "grad_norm": 0.5446816682815552,
      "learning_rate": 5.864127595596119e-07,
      "loss": 0.5025,
      "step": 3793
    },
    {
      "epoch": 0.6469986357435198,
      "grad_norm": 0.4279031753540039,
      "learning_rate": 5.859099554809818e-07,
      "loss": 0.5001,
      "step": 3794
    },
    {
      "epoch": 0.647169167803547,
      "grad_norm": 0.6133118271827698,
      "learning_rate": 5.854072777232915e-07,
      "loss": 0.5042,
      "step": 3795
    },
    {
      "epoch": 0.6473396998635743,
      "grad_norm": 0.5323469638824463,
      "learning_rate": 5.84904726439886e-07,
      "loss": 0.505,
      "step": 3796
    },
    {
      "epoch": 0.6475102319236017,
      "grad_norm": 0.3823695778846741,
      "learning_rate": 5.844023017840725e-07,
      "loss": 0.4983,
      "step": 3797
    },
    {
      "epoch": 0.647680763983629,
      "grad_norm": 0.5067130923271179,
      "learning_rate": 5.839000039091185e-07,
      "loss": 0.5117,
      "step": 3798
    },
    {
      "epoch": 0.6478512960436562,
      "grad_norm": 0.4036625027656555,
      "learning_rate": 5.833978329682535e-07,
      "loss": 0.4973,
      "step": 3799
    },
    {
      "epoch": 0.6480218281036835,
      "grad_norm": 0.5435891151428223,
      "learning_rate": 5.828957891146682e-07,
      "loss": 0.5048,
      "step": 3800
    },
    {
      "epoch": 0.6481923601637107,
      "grad_norm": 0.3958360254764557,
      "learning_rate": 5.823938725015149e-07,
      "loss": 0.5119,
      "step": 3801
    },
    {
      "epoch": 0.6483628922237381,
      "grad_norm": 0.533546507358551,
      "learning_rate": 5.818920832819051e-07,
      "loss": 0.5019,
      "step": 3802
    },
    {
      "epoch": 0.6485334242837654,
      "grad_norm": 0.43487948179244995,
      "learning_rate": 5.813904216089149e-07,
      "loss": 0.5022,
      "step": 3803
    },
    {
      "epoch": 0.6487039563437926,
      "grad_norm": 0.4595380425453186,
      "learning_rate": 5.808888876355785e-07,
      "loss": 0.5076,
      "step": 3804
    },
    {
      "epoch": 0.6488744884038199,
      "grad_norm": 0.45527392625808716,
      "learning_rate": 5.803874815148917e-07,
      "loss": 0.503,
      "step": 3805
    },
    {
      "epoch": 0.6490450204638472,
      "grad_norm": 0.36405813694000244,
      "learning_rate": 5.798862033998132e-07,
      "loss": 0.5049,
      "step": 3806
    },
    {
      "epoch": 0.6492155525238745,
      "grad_norm": 0.5357425212860107,
      "learning_rate": 5.7938505344326e-07,
      "loss": 0.5053,
      "step": 3807
    },
    {
      "epoch": 0.6493860845839018,
      "grad_norm": 0.44985252618789673,
      "learning_rate": 5.788840317981118e-07,
      "loss": 0.5127,
      "step": 3808
    },
    {
      "epoch": 0.649556616643929,
      "grad_norm": 0.4584481418132782,
      "learning_rate": 5.783831386172083e-07,
      "loss": 0.5033,
      "step": 3809
    },
    {
      "epoch": 0.6497271487039563,
      "grad_norm": 0.6204929351806641,
      "learning_rate": 5.77882374053351e-07,
      "loss": 0.503,
      "step": 3810
    },
    {
      "epoch": 0.6498976807639836,
      "grad_norm": 0.42705973982810974,
      "learning_rate": 5.773817382593008e-07,
      "loss": 0.5033,
      "step": 3811
    },
    {
      "epoch": 0.650068212824011,
      "grad_norm": 0.3911517858505249,
      "learning_rate": 5.768812313877801e-07,
      "loss": 0.5098,
      "step": 3812
    },
    {
      "epoch": 0.6502387448840382,
      "grad_norm": 0.4088066816329956,
      "learning_rate": 5.763808535914724e-07,
      "loss": 0.494,
      "step": 3813
    },
    {
      "epoch": 0.6504092769440655,
      "grad_norm": 0.4033955931663513,
      "learning_rate": 5.758806050230199e-07,
      "loss": 0.5019,
      "step": 3814
    },
    {
      "epoch": 0.6505798090040927,
      "grad_norm": 0.4054994285106659,
      "learning_rate": 5.753804858350289e-07,
      "loss": 0.5012,
      "step": 3815
    },
    {
      "epoch": 0.6507503410641201,
      "grad_norm": 0.4501291811466217,
      "learning_rate": 5.748804961800623e-07,
      "loss": 0.507,
      "step": 3816
    },
    {
      "epoch": 0.6509208731241474,
      "grad_norm": 0.40860098600387573,
      "learning_rate": 5.743806362106461e-07,
      "loss": 0.5033,
      "step": 3817
    },
    {
      "epoch": 0.6510914051841746,
      "grad_norm": 0.4299365282058716,
      "learning_rate": 5.738809060792661e-07,
      "loss": 0.5023,
      "step": 3818
    },
    {
      "epoch": 0.6512619372442019,
      "grad_norm": 0.44453543424606323,
      "learning_rate": 5.733813059383677e-07,
      "loss": 0.5102,
      "step": 3819
    },
    {
      "epoch": 0.6514324693042292,
      "grad_norm": 0.40756890177726746,
      "learning_rate": 5.728818359403575e-07,
      "loss": 0.5032,
      "step": 3820
    },
    {
      "epoch": 0.6516030013642565,
      "grad_norm": 0.42828136682510376,
      "learning_rate": 5.723824962376022e-07,
      "loss": 0.5113,
      "step": 3821
    },
    {
      "epoch": 0.6517735334242838,
      "grad_norm": 0.4323231279850006,
      "learning_rate": 5.718832869824292e-07,
      "loss": 0.4994,
      "step": 3822
    },
    {
      "epoch": 0.651944065484311,
      "grad_norm": 0.49335092306137085,
      "learning_rate": 5.713842083271244e-07,
      "loss": 0.5107,
      "step": 3823
    },
    {
      "epoch": 0.6521145975443383,
      "grad_norm": 0.37755998969078064,
      "learning_rate": 5.708852604239367e-07,
      "loss": 0.5079,
      "step": 3824
    },
    {
      "epoch": 0.6522851296043656,
      "grad_norm": 0.43253105878829956,
      "learning_rate": 5.703864434250722e-07,
      "loss": 0.5041,
      "step": 3825
    },
    {
      "epoch": 0.652455661664393,
      "grad_norm": 0.49581003189086914,
      "learning_rate": 5.698877574826989e-07,
      "loss": 0.5075,
      "step": 3826
    },
    {
      "epoch": 0.6526261937244202,
      "grad_norm": 0.401373028755188,
      "learning_rate": 5.693892027489445e-07,
      "loss": 0.4968,
      "step": 3827
    },
    {
      "epoch": 0.6527967257844475,
      "grad_norm": 0.45689138770103455,
      "learning_rate": 5.688907793758959e-07,
      "loss": 0.509,
      "step": 3828
    },
    {
      "epoch": 0.6529672578444747,
      "grad_norm": 0.44455838203430176,
      "learning_rate": 5.683924875156007e-07,
      "loss": 0.5013,
      "step": 3829
    },
    {
      "epoch": 0.653137789904502,
      "grad_norm": 0.48018378019332886,
      "learning_rate": 5.678943273200662e-07,
      "loss": 0.5237,
      "step": 3830
    },
    {
      "epoch": 0.6533083219645294,
      "grad_norm": 0.43504878878593445,
      "learning_rate": 5.673962989412599e-07,
      "loss": 0.5118,
      "step": 3831
    },
    {
      "epoch": 0.6534788540245566,
      "grad_norm": 0.3849477767944336,
      "learning_rate": 5.668984025311077e-07,
      "loss": 0.5087,
      "step": 3832
    },
    {
      "epoch": 0.6536493860845839,
      "grad_norm": 0.45680201053619385,
      "learning_rate": 5.664006382414975e-07,
      "loss": 0.519,
      "step": 3833
    },
    {
      "epoch": 0.6538199181446112,
      "grad_norm": 0.4410555064678192,
      "learning_rate": 5.659030062242746e-07,
      "loss": 0.5174,
      "step": 3834
    },
    {
      "epoch": 0.6539904502046384,
      "grad_norm": 0.5044737458229065,
      "learning_rate": 5.654055066312455e-07,
      "loss": 0.5181,
      "step": 3835
    },
    {
      "epoch": 0.6541609822646658,
      "grad_norm": 0.4555228650569916,
      "learning_rate": 5.649081396141758e-07,
      "loss": 0.5098,
      "step": 3836
    },
    {
      "epoch": 0.654331514324693,
      "grad_norm": 0.4320026934146881,
      "learning_rate": 5.644109053247902e-07,
      "loss": 0.5083,
      "step": 3837
    },
    {
      "epoch": 0.6545020463847203,
      "grad_norm": 0.6127005815505981,
      "learning_rate": 5.639138039147737e-07,
      "loss": 0.5072,
      "step": 3838
    },
    {
      "epoch": 0.6546725784447476,
      "grad_norm": 0.5029910206794739,
      "learning_rate": 5.634168355357701e-07,
      "loss": 0.5104,
      "step": 3839
    },
    {
      "epoch": 0.654843110504775,
      "grad_norm": 0.4032624363899231,
      "learning_rate": 5.629200003393838e-07,
      "loss": 0.52,
      "step": 3840
    },
    {
      "epoch": 0.6550136425648022,
      "grad_norm": 0.4314282536506653,
      "learning_rate": 5.62423298477176e-07,
      "loss": 0.5172,
      "step": 3841
    },
    {
      "epoch": 0.6551841746248295,
      "grad_norm": 0.37860020995140076,
      "learning_rate": 5.619267301006711e-07,
      "loss": 0.5046,
      "step": 3842
    },
    {
      "epoch": 0.6553547066848567,
      "grad_norm": 0.4099591076374054,
      "learning_rate": 5.61430295361349e-07,
      "loss": 0.5122,
      "step": 3843
    },
    {
      "epoch": 0.655525238744884,
      "grad_norm": 0.38307932019233704,
      "learning_rate": 5.60933994410651e-07,
      "loss": 0.5057,
      "step": 3844
    },
    {
      "epoch": 0.6556957708049114,
      "grad_norm": 0.44933193922042847,
      "learning_rate": 5.604378273999773e-07,
      "loss": 0.5192,
      "step": 3845
    },
    {
      "epoch": 0.6558663028649386,
      "grad_norm": 0.3983853757381439,
      "learning_rate": 5.599417944806862e-07,
      "loss": 0.51,
      "step": 3846
    },
    {
      "epoch": 0.6560368349249659,
      "grad_norm": 0.43134453892707825,
      "learning_rate": 5.594458958040962e-07,
      "loss": 0.5024,
      "step": 3847
    },
    {
      "epoch": 0.6562073669849932,
      "grad_norm": 0.4693518877029419,
      "learning_rate": 5.58950131521485e-07,
      "loss": 0.5059,
      "step": 3848
    },
    {
      "epoch": 0.6563778990450204,
      "grad_norm": 0.48148661851882935,
      "learning_rate": 5.584545017840886e-07,
      "loss": 0.5057,
      "step": 3849
    },
    {
      "epoch": 0.6565484311050478,
      "grad_norm": 0.41671866178512573,
      "learning_rate": 5.579590067431016e-07,
      "loss": 0.5003,
      "step": 3850
    },
    {
      "epoch": 0.656718963165075,
      "grad_norm": 0.5516775846481323,
      "learning_rate": 5.574636465496783e-07,
      "loss": 0.5102,
      "step": 3851
    },
    {
      "epoch": 0.6568894952251023,
      "grad_norm": 0.3486853837966919,
      "learning_rate": 5.56968421354932e-07,
      "loss": 0.5124,
      "step": 3852
    },
    {
      "epoch": 0.6570600272851296,
      "grad_norm": 0.5256639719009399,
      "learning_rate": 5.564733313099344e-07,
      "loss": 0.5155,
      "step": 3853
    },
    {
      "epoch": 0.6572305593451568,
      "grad_norm": 0.45007023215293884,
      "learning_rate": 5.559783765657161e-07,
      "loss": 0.5015,
      "step": 3854
    },
    {
      "epoch": 0.6574010914051842,
      "grad_norm": 0.41359055042266846,
      "learning_rate": 5.554835572732658e-07,
      "loss": 0.5019,
      "step": 3855
    },
    {
      "epoch": 0.6575716234652115,
      "grad_norm": 0.4674064815044403,
      "learning_rate": 5.549888735835319e-07,
      "loss": 0.5111,
      "step": 3856
    },
    {
      "epoch": 0.6577421555252387,
      "grad_norm": 0.41691353917121887,
      "learning_rate": 5.544943256474208e-07,
      "loss": 0.4975,
      "step": 3857
    },
    {
      "epoch": 0.657912687585266,
      "grad_norm": 0.522437334060669,
      "learning_rate": 5.539999136157978e-07,
      "loss": 0.5235,
      "step": 3858
    },
    {
      "epoch": 0.6580832196452933,
      "grad_norm": 0.4822806119918823,
      "learning_rate": 5.535056376394869e-07,
      "loss": 0.5114,
      "step": 3859
    },
    {
      "epoch": 0.6582537517053206,
      "grad_norm": 0.4718954265117645,
      "learning_rate": 5.530114978692695e-07,
      "loss": 0.5035,
      "step": 3860
    },
    {
      "epoch": 0.6584242837653479,
      "grad_norm": 0.5045837759971619,
      "learning_rate": 5.525174944558867e-07,
      "loss": 0.5044,
      "step": 3861
    },
    {
      "epoch": 0.6585948158253752,
      "grad_norm": 0.4522147476673126,
      "learning_rate": 5.520236275500375e-07,
      "loss": 0.5004,
      "step": 3862
    },
    {
      "epoch": 0.6587653478854024,
      "grad_norm": 0.45062634348869324,
      "learning_rate": 5.515298973023797e-07,
      "loss": 0.5086,
      "step": 3863
    },
    {
      "epoch": 0.6589358799454298,
      "grad_norm": 0.5331596732139587,
      "learning_rate": 5.510363038635283e-07,
      "loss": 0.5131,
      "step": 3864
    },
    {
      "epoch": 0.659106412005457,
      "grad_norm": 0.6558175086975098,
      "learning_rate": 5.505428473840576e-07,
      "loss": 0.5163,
      "step": 3865
    },
    {
      "epoch": 0.6592769440654843,
      "grad_norm": 0.5517013669013977,
      "learning_rate": 5.500495280145003e-07,
      "loss": 0.5153,
      "step": 3866
    },
    {
      "epoch": 0.6594474761255116,
      "grad_norm": 0.5525789856910706,
      "learning_rate": 5.495563459053455e-07,
      "loss": 0.5121,
      "step": 3867
    },
    {
      "epoch": 0.6596180081855388,
      "grad_norm": 0.6826182007789612,
      "learning_rate": 5.490633012070433e-07,
      "loss": 0.5121,
      "step": 3868
    },
    {
      "epoch": 0.6597885402455662,
      "grad_norm": 0.393273264169693,
      "learning_rate": 5.485703940699989e-07,
      "loss": 0.4985,
      "step": 3869
    },
    {
      "epoch": 0.6599590723055935,
      "grad_norm": 0.5517546534538269,
      "learning_rate": 5.480776246445776e-07,
      "loss": 0.4929,
      "step": 3870
    },
    {
      "epoch": 0.6601296043656207,
      "grad_norm": 0.494389146566391,
      "learning_rate": 5.475849930811022e-07,
      "loss": 0.5149,
      "step": 3871
    },
    {
      "epoch": 0.660300136425648,
      "grad_norm": 0.4777924418449402,
      "learning_rate": 5.47092499529853e-07,
      "loss": 0.4982,
      "step": 3872
    },
    {
      "epoch": 0.6604706684856753,
      "grad_norm": 0.44324955344200134,
      "learning_rate": 5.466001441410682e-07,
      "loss": 0.506,
      "step": 3873
    },
    {
      "epoch": 0.6606412005457026,
      "grad_norm": 0.4134279191493988,
      "learning_rate": 5.461079270649445e-07,
      "loss": 0.4977,
      "step": 3874
    },
    {
      "epoch": 0.6608117326057299,
      "grad_norm": 0.4815727174282074,
      "learning_rate": 5.456158484516361e-07,
      "loss": 0.4951,
      "step": 3875
    },
    {
      "epoch": 0.6609822646657572,
      "grad_norm": 0.4558385908603668,
      "learning_rate": 5.451239084512538e-07,
      "loss": 0.4937,
      "step": 3876
    },
    {
      "epoch": 0.6611527967257844,
      "grad_norm": 0.43415579199790955,
      "learning_rate": 5.446321072138689e-07,
      "loss": 0.503,
      "step": 3877
    },
    {
      "epoch": 0.6613233287858117,
      "grad_norm": 0.4558171033859253,
      "learning_rate": 5.441404448895074e-07,
      "loss": 0.5016,
      "step": 3878
    },
    {
      "epoch": 0.661493860845839,
      "grad_norm": 0.4061969220638275,
      "learning_rate": 5.436489216281544e-07,
      "loss": 0.4943,
      "step": 3879
    },
    {
      "epoch": 0.6616643929058663,
      "grad_norm": 0.547091007232666,
      "learning_rate": 5.431575375797527e-07,
      "loss": 0.4971,
      "step": 3880
    },
    {
      "epoch": 0.6618349249658936,
      "grad_norm": 0.6131861209869385,
      "learning_rate": 5.426662928942024e-07,
      "loss": 0.5102,
      "step": 3881
    },
    {
      "epoch": 0.6620054570259208,
      "grad_norm": 0.4211820363998413,
      "learning_rate": 5.421751877213604e-07,
      "loss": 0.5098,
      "step": 3882
    },
    {
      "epoch": 0.6621759890859482,
      "grad_norm": 0.5724109411239624,
      "learning_rate": 5.416842222110419e-07,
      "loss": 0.5023,
      "step": 3883
    },
    {
      "epoch": 0.6623465211459755,
      "grad_norm": 0.4997083842754364,
      "learning_rate": 5.411933965130197e-07,
      "loss": 0.4978,
      "step": 3884
    },
    {
      "epoch": 0.6625170532060027,
      "grad_norm": 0.4442778527736664,
      "learning_rate": 5.40702710777022e-07,
      "loss": 0.5093,
      "step": 3885
    },
    {
      "epoch": 0.66268758526603,
      "grad_norm": 0.531562864780426,
      "learning_rate": 5.402121651527377e-07,
      "loss": 0.5012,
      "step": 3886
    },
    {
      "epoch": 0.6628581173260573,
      "grad_norm": 0.623119056224823,
      "learning_rate": 5.397217597898097e-07,
      "loss": 0.5025,
      "step": 3887
    },
    {
      "epoch": 0.6630286493860846,
      "grad_norm": 0.6248323321342468,
      "learning_rate": 5.392314948378397e-07,
      "loss": 0.4945,
      "step": 3888
    },
    {
      "epoch": 0.6631991814461119,
      "grad_norm": 0.40609654784202576,
      "learning_rate": 5.387413704463871e-07,
      "loss": 0.5072,
      "step": 3889
    },
    {
      "epoch": 0.6633697135061392,
      "grad_norm": 0.398261696100235,
      "learning_rate": 5.382513867649664e-07,
      "loss": 0.5041,
      "step": 3890
    },
    {
      "epoch": 0.6635402455661664,
      "grad_norm": 0.5225828289985657,
      "learning_rate": 5.377615439430508e-07,
      "loss": 0.5133,
      "step": 3891
    },
    {
      "epoch": 0.6637107776261937,
      "grad_norm": 0.3824782073497772,
      "learning_rate": 5.372718421300704e-07,
      "loss": 0.5098,
      "step": 3892
    },
    {
      "epoch": 0.663881309686221,
      "grad_norm": 0.5820587873458862,
      "learning_rate": 5.367822814754125e-07,
      "loss": 0.5074,
      "step": 3893
    },
    {
      "epoch": 0.6640518417462483,
      "grad_norm": 0.5361365079879761,
      "learning_rate": 5.362928621284194e-07,
      "loss": 0.5059,
      "step": 3894
    },
    {
      "epoch": 0.6642223738062756,
      "grad_norm": 0.49799999594688416,
      "learning_rate": 5.358035842383933e-07,
      "loss": 0.5044,
      "step": 3895
    },
    {
      "epoch": 0.6643929058663028,
      "grad_norm": 0.6610023975372314,
      "learning_rate": 5.353144479545907e-07,
      "loss": 0.5098,
      "step": 3896
    },
    {
      "epoch": 0.6645634379263301,
      "grad_norm": 0.4974028766155243,
      "learning_rate": 5.348254534262263e-07,
      "loss": 0.4858,
      "step": 3897
    },
    {
      "epoch": 0.6647339699863575,
      "grad_norm": 0.5856749415397644,
      "learning_rate": 5.343366008024713e-07,
      "loss": 0.4981,
      "step": 3898
    },
    {
      "epoch": 0.6649045020463847,
      "grad_norm": 0.5684167742729187,
      "learning_rate": 5.338478902324529e-07,
      "loss": 0.5152,
      "step": 3899
    },
    {
      "epoch": 0.665075034106412,
      "grad_norm": 0.5611979365348816,
      "learning_rate": 5.333593218652558e-07,
      "loss": 0.5063,
      "step": 3900
    },
    {
      "epoch": 0.6652455661664393,
      "grad_norm": 0.4313353896141052,
      "learning_rate": 5.328708958499212e-07,
      "loss": 0.5143,
      "step": 3901
    },
    {
      "epoch": 0.6654160982264665,
      "grad_norm": 0.4924471378326416,
      "learning_rate": 5.323826123354469e-07,
      "loss": 0.4946,
      "step": 3902
    },
    {
      "epoch": 0.6655866302864939,
      "grad_norm": 0.4859728217124939,
      "learning_rate": 5.318944714707861e-07,
      "loss": 0.5012,
      "step": 3903
    },
    {
      "epoch": 0.6657571623465212,
      "grad_norm": 0.40626779198646545,
      "learning_rate": 5.314064734048509e-07,
      "loss": 0.4989,
      "step": 3904
    },
    {
      "epoch": 0.6659276944065484,
      "grad_norm": 0.5151337385177612,
      "learning_rate": 5.309186182865077e-07,
      "loss": 0.5104,
      "step": 3905
    },
    {
      "epoch": 0.6660982264665757,
      "grad_norm": 0.3631163537502289,
      "learning_rate": 5.304309062645789e-07,
      "loss": 0.4974,
      "step": 3906
    },
    {
      "epoch": 0.666268758526603,
      "grad_norm": 0.4548703730106354,
      "learning_rate": 5.299433374878462e-07,
      "loss": 0.5019,
      "step": 3907
    },
    {
      "epoch": 0.6664392905866303,
      "grad_norm": 0.4133833944797516,
      "learning_rate": 5.294559121050443e-07,
      "loss": 0.5155,
      "step": 3908
    },
    {
      "epoch": 0.6666098226466576,
      "grad_norm": 0.41636455059051514,
      "learning_rate": 5.289686302648662e-07,
      "loss": 0.5054,
      "step": 3909
    },
    {
      "epoch": 0.6667803547066848,
      "grad_norm": 0.5298854112625122,
      "learning_rate": 5.284814921159603e-07,
      "loss": 0.5002,
      "step": 3910
    },
    {
      "epoch": 0.6669508867667121,
      "grad_norm": 0.481473833322525,
      "learning_rate": 5.279944978069319e-07,
      "loss": 0.491,
      "step": 3911
    },
    {
      "epoch": 0.6671214188267395,
      "grad_norm": 0.4647562801837921,
      "learning_rate": 5.275076474863409e-07,
      "loss": 0.522,
      "step": 3912
    },
    {
      "epoch": 0.6672919508867667,
      "grad_norm": 0.4734087586402893,
      "learning_rate": 5.270209413027048e-07,
      "loss": 0.5105,
      "step": 3913
    },
    {
      "epoch": 0.667462482946794,
      "grad_norm": 0.4895179271697998,
      "learning_rate": 5.265343794044965e-07,
      "loss": 0.5043,
      "step": 3914
    },
    {
      "epoch": 0.6676330150068213,
      "grad_norm": 0.5346952080726624,
      "learning_rate": 5.260479619401451e-07,
      "loss": 0.4984,
      "step": 3915
    },
    {
      "epoch": 0.6678035470668485,
      "grad_norm": 0.47937634587287903,
      "learning_rate": 5.255616890580357e-07,
      "loss": 0.4965,
      "step": 3916
    },
    {
      "epoch": 0.6679740791268759,
      "grad_norm": 0.46289533376693726,
      "learning_rate": 5.250755609065083e-07,
      "loss": 0.5058,
      "step": 3917
    },
    {
      "epoch": 0.6681446111869032,
      "grad_norm": 0.6581227779388428,
      "learning_rate": 5.245895776338602e-07,
      "loss": 0.4989,
      "step": 3918
    },
    {
      "epoch": 0.6683151432469304,
      "grad_norm": 0.573941171169281,
      "learning_rate": 5.241037393883436e-07,
      "loss": 0.4946,
      "step": 3919
    },
    {
      "epoch": 0.6684856753069577,
      "grad_norm": 0.5698941946029663,
      "learning_rate": 5.236180463181673e-07,
      "loss": 0.4941,
      "step": 3920
    },
    {
      "epoch": 0.668656207366985,
      "grad_norm": 0.47878286242485046,
      "learning_rate": 5.231324985714942e-07,
      "loss": 0.511,
      "step": 3921
    },
    {
      "epoch": 0.6688267394270123,
      "grad_norm": 0.4517112672328949,
      "learning_rate": 5.226470962964445e-07,
      "loss": 0.4999,
      "step": 3922
    },
    {
      "epoch": 0.6689972714870396,
      "grad_norm": 0.45299017429351807,
      "learning_rate": 5.221618396410933e-07,
      "loss": 0.4928,
      "step": 3923
    },
    {
      "epoch": 0.6691678035470668,
      "grad_norm": 0.5403193831443787,
      "learning_rate": 5.216767287534714e-07,
      "loss": 0.5172,
      "step": 3924
    },
    {
      "epoch": 0.6693383356070941,
      "grad_norm": 0.46739518642425537,
      "learning_rate": 5.211917637815655e-07,
      "loss": 0.5099,
      "step": 3925
    },
    {
      "epoch": 0.6695088676671214,
      "grad_norm": 0.38346195220947266,
      "learning_rate": 5.207069448733166e-07,
      "loss": 0.5008,
      "step": 3926
    },
    {
      "epoch": 0.6696793997271487,
      "grad_norm": 0.5461332201957703,
      "learning_rate": 5.202222721766226e-07,
      "loss": 0.5074,
      "step": 3927
    },
    {
      "epoch": 0.669849931787176,
      "grad_norm": 0.3497057855129242,
      "learning_rate": 5.197377458393363e-07,
      "loss": 0.5023,
      "step": 3928
    },
    {
      "epoch": 0.6700204638472033,
      "grad_norm": 0.48500290513038635,
      "learning_rate": 5.192533660092647e-07,
      "loss": 0.5051,
      "step": 3929
    },
    {
      "epoch": 0.6701909959072305,
      "grad_norm": 0.4321812689304352,
      "learning_rate": 5.18769132834172e-07,
      "loss": 0.5069,
      "step": 3930
    },
    {
      "epoch": 0.6703615279672579,
      "grad_norm": 0.380292683839798,
      "learning_rate": 5.182850464617766e-07,
      "loss": 0.5117,
      "step": 3931
    },
    {
      "epoch": 0.6705320600272852,
      "grad_norm": 0.535532534122467,
      "learning_rate": 5.17801107039752e-07,
      "loss": 0.5078,
      "step": 3932
    },
    {
      "epoch": 0.6707025920873124,
      "grad_norm": 0.5220469832420349,
      "learning_rate": 5.173173147157276e-07,
      "loss": 0.5089,
      "step": 3933
    },
    {
      "epoch": 0.6708731241473397,
      "grad_norm": 0.42341408133506775,
      "learning_rate": 5.168336696372876e-07,
      "loss": 0.4918,
      "step": 3934
    },
    {
      "epoch": 0.671043656207367,
      "grad_norm": 0.37881559133529663,
      "learning_rate": 5.163501719519704e-07,
      "loss": 0.4889,
      "step": 3935
    },
    {
      "epoch": 0.6712141882673943,
      "grad_norm": 0.42093196511268616,
      "learning_rate": 5.158668218072706e-07,
      "loss": 0.5068,
      "step": 3936
    },
    {
      "epoch": 0.6713847203274216,
      "grad_norm": 0.41691750288009644,
      "learning_rate": 5.153836193506379e-07,
      "loss": 0.504,
      "step": 3937
    },
    {
      "epoch": 0.6715552523874488,
      "grad_norm": 0.4368269741535187,
      "learning_rate": 5.149005647294755e-07,
      "loss": 0.5031,
      "step": 3938
    },
    {
      "epoch": 0.6717257844474761,
      "grad_norm": 0.385061115026474,
      "learning_rate": 5.144176580911431e-07,
      "loss": 0.5033,
      "step": 3939
    },
    {
      "epoch": 0.6718963165075034,
      "grad_norm": 0.40677857398986816,
      "learning_rate": 5.139348995829544e-07,
      "loss": 0.5092,
      "step": 3940
    },
    {
      "epoch": 0.6720668485675307,
      "grad_norm": 0.3767397701740265,
      "learning_rate": 5.13452289352178e-07,
      "loss": 0.515,
      "step": 3941
    },
    {
      "epoch": 0.672237380627558,
      "grad_norm": 0.36541303992271423,
      "learning_rate": 5.129698275460376e-07,
      "loss": 0.5199,
      "step": 3942
    },
    {
      "epoch": 0.6724079126875853,
      "grad_norm": 0.35933759808540344,
      "learning_rate": 5.124875143117117e-07,
      "loss": 0.5014,
      "step": 3943
    },
    {
      "epoch": 0.6725784447476125,
      "grad_norm": 0.5467775464057922,
      "learning_rate": 5.120053497963325e-07,
      "loss": 0.5118,
      "step": 3944
    },
    {
      "epoch": 0.6727489768076398,
      "grad_norm": 0.45338207483291626,
      "learning_rate": 5.115233341469878e-07,
      "loss": 0.5086,
      "step": 3945
    },
    {
      "epoch": 0.6729195088676672,
      "grad_norm": 0.42042842507362366,
      "learning_rate": 5.1104146751072e-07,
      "loss": 0.5013,
      "step": 3946
    },
    {
      "epoch": 0.6730900409276944,
      "grad_norm": 0.5868545770645142,
      "learning_rate": 5.105597500345249e-07,
      "loss": 0.5007,
      "step": 3947
    },
    {
      "epoch": 0.6732605729877217,
      "grad_norm": 0.41611066460609436,
      "learning_rate": 5.100781818653548e-07,
      "loss": 0.5186,
      "step": 3948
    },
    {
      "epoch": 0.673431105047749,
      "grad_norm": 0.5460180640220642,
      "learning_rate": 5.095967631501144e-07,
      "loss": 0.5149,
      "step": 3949
    },
    {
      "epoch": 0.6736016371077762,
      "grad_norm": 0.5588499307632446,
      "learning_rate": 5.09115494035664e-07,
      "loss": 0.5001,
      "step": 3950
    },
    {
      "epoch": 0.6737721691678036,
      "grad_norm": 0.3679949641227722,
      "learning_rate": 5.086343746688184e-07,
      "loss": 0.5091,
      "step": 3951
    },
    {
      "epoch": 0.6739427012278308,
      "grad_norm": 0.4741192162036896,
      "learning_rate": 5.081534051963455e-07,
      "loss": 0.5053,
      "step": 3952
    },
    {
      "epoch": 0.6741132332878581,
      "grad_norm": 0.4042551517486572,
      "learning_rate": 5.076725857649685e-07,
      "loss": 0.4951,
      "step": 3953
    },
    {
      "epoch": 0.6742837653478854,
      "grad_norm": 0.5772173404693604,
      "learning_rate": 5.071919165213648e-07,
      "loss": 0.5071,
      "step": 3954
    },
    {
      "epoch": 0.6744542974079127,
      "grad_norm": 0.5814014077186584,
      "learning_rate": 5.067113976121659e-07,
      "loss": 0.5097,
      "step": 3955
    },
    {
      "epoch": 0.67462482946794,
      "grad_norm": 0.4528033137321472,
      "learning_rate": 5.062310291839563e-07,
      "loss": 0.4971,
      "step": 3956
    },
    {
      "epoch": 0.6747953615279673,
      "grad_norm": 0.5089178085327148,
      "learning_rate": 5.057508113832773e-07,
      "loss": 0.5093,
      "step": 3957
    },
    {
      "epoch": 0.6749658935879945,
      "grad_norm": 0.591621994972229,
      "learning_rate": 5.052707443566212e-07,
      "loss": 0.5052,
      "step": 3958
    },
    {
      "epoch": 0.6751364256480218,
      "grad_norm": 0.6256417036056519,
      "learning_rate": 5.047908282504361e-07,
      "loss": 0.5047,
      "step": 3959
    },
    {
      "epoch": 0.6753069577080492,
      "grad_norm": 0.48071393370628357,
      "learning_rate": 5.043110632111239e-07,
      "loss": 0.5114,
      "step": 3960
    },
    {
      "epoch": 0.6754774897680764,
      "grad_norm": 0.4122591018676758,
      "learning_rate": 5.038314493850397e-07,
      "loss": 0.5142,
      "step": 3961
    },
    {
      "epoch": 0.6756480218281037,
      "grad_norm": 0.5708420276641846,
      "learning_rate": 5.033519869184928e-07,
      "loss": 0.4986,
      "step": 3962
    },
    {
      "epoch": 0.675818553888131,
      "grad_norm": 0.5030431747436523,
      "learning_rate": 5.028726759577467e-07,
      "loss": 0.4956,
      "step": 3963
    },
    {
      "epoch": 0.6759890859481582,
      "grad_norm": 0.46762701869010925,
      "learning_rate": 5.023935166490189e-07,
      "loss": 0.5037,
      "step": 3964
    },
    {
      "epoch": 0.6761596180081856,
      "grad_norm": 0.46051761507987976,
      "learning_rate": 5.01914509138479e-07,
      "loss": 0.5174,
      "step": 3965
    },
    {
      "epoch": 0.6763301500682128,
      "grad_norm": 0.4886033535003662,
      "learning_rate": 5.014356535722527e-07,
      "loss": 0.5029,
      "step": 3966
    },
    {
      "epoch": 0.6765006821282401,
      "grad_norm": 0.440492182970047,
      "learning_rate": 5.009569500964176e-07,
      "loss": 0.4995,
      "step": 3967
    },
    {
      "epoch": 0.6766712141882674,
      "grad_norm": 0.31643933057785034,
      "learning_rate": 5.004783988570046e-07,
      "loss": 0.5057,
      "step": 3968
    },
    {
      "epoch": 0.6768417462482946,
      "grad_norm": 0.4844537377357483,
      "learning_rate": 5.000000000000003e-07,
      "loss": 0.5024,
      "step": 3969
    },
    {
      "epoch": 0.677012278308322,
      "grad_norm": 0.45205482840538025,
      "learning_rate": 4.995217536713425e-07,
      "loss": 0.4922,
      "step": 3970
    },
    {
      "epoch": 0.6771828103683493,
      "grad_norm": 0.40974223613739014,
      "learning_rate": 4.990436600169238e-07,
      "loss": 0.4809,
      "step": 3971
    },
    {
      "epoch": 0.6773533424283765,
      "grad_norm": 0.45243552327156067,
      "learning_rate": 4.985657191825898e-07,
      "loss": 0.4977,
      "step": 3972
    },
    {
      "epoch": 0.6775238744884038,
      "grad_norm": 0.4544006288051605,
      "learning_rate": 4.980879313141401e-07,
      "loss": 0.5197,
      "step": 3973
    },
    {
      "epoch": 0.677694406548431,
      "grad_norm": 0.375340074300766,
      "learning_rate": 4.976102965573257e-07,
      "loss": 0.5101,
      "step": 3974
    },
    {
      "epoch": 0.6778649386084584,
      "grad_norm": 0.5214848518371582,
      "learning_rate": 4.97132815057854e-07,
      "loss": 0.5032,
      "step": 3975
    },
    {
      "epoch": 0.6780354706684857,
      "grad_norm": 0.501172661781311,
      "learning_rate": 4.966554869613832e-07,
      "loss": 0.5026,
      "step": 3976
    },
    {
      "epoch": 0.678206002728513,
      "grad_norm": 0.5273376703262329,
      "learning_rate": 4.961783124135245e-07,
      "loss": 0.5062,
      "step": 3977
    },
    {
      "epoch": 0.6783765347885402,
      "grad_norm": 0.5391064286231995,
      "learning_rate": 4.957012915598446e-07,
      "loss": 0.5048,
      "step": 3978
    },
    {
      "epoch": 0.6785470668485676,
      "grad_norm": 0.4569733440876007,
      "learning_rate": 4.952244245458612e-07,
      "loss": 0.4924,
      "step": 3979
    },
    {
      "epoch": 0.6787175989085948,
      "grad_norm": 0.4716772139072418,
      "learning_rate": 4.947477115170457e-07,
      "loss": 0.5129,
      "step": 3980
    },
    {
      "epoch": 0.6788881309686221,
      "grad_norm": 0.5155799388885498,
      "learning_rate": 4.94271152618823e-07,
      "loss": 0.4999,
      "step": 3981
    },
    {
      "epoch": 0.6790586630286494,
      "grad_norm": 0.44789519906044006,
      "learning_rate": 4.937947479965706e-07,
      "loss": 0.4979,
      "step": 3982
    },
    {
      "epoch": 0.6792291950886766,
      "grad_norm": 0.5172362327575684,
      "learning_rate": 4.933184977956184e-07,
      "loss": 0.4973,
      "step": 3983
    },
    {
      "epoch": 0.679399727148704,
      "grad_norm": 0.6166170835494995,
      "learning_rate": 4.928424021612499e-07,
      "loss": 0.5038,
      "step": 3984
    },
    {
      "epoch": 0.6795702592087313,
      "grad_norm": 0.40173619985580444,
      "learning_rate": 4.92366461238702e-07,
      "loss": 0.5035,
      "step": 3985
    },
    {
      "epoch": 0.6797407912687585,
      "grad_norm": 0.4869058430194855,
      "learning_rate": 4.918906751731621e-07,
      "loss": 0.5071,
      "step": 3986
    },
    {
      "epoch": 0.6799113233287858,
      "grad_norm": 0.5133165717124939,
      "learning_rate": 4.914150441097737e-07,
      "loss": 0.5077,
      "step": 3987
    },
    {
      "epoch": 0.680081855388813,
      "grad_norm": 0.5123385787010193,
      "learning_rate": 4.909395681936298e-07,
      "loss": 0.5151,
      "step": 3988
    },
    {
      "epoch": 0.6802523874488404,
      "grad_norm": 0.4512058198451996,
      "learning_rate": 4.904642475697783e-07,
      "loss": 0.5105,
      "step": 3989
    },
    {
      "epoch": 0.6804229195088677,
      "grad_norm": 0.47699469327926636,
      "learning_rate": 4.899890823832192e-07,
      "loss": 0.503,
      "step": 3990
    },
    {
      "epoch": 0.680593451568895,
      "grad_norm": 0.5350492596626282,
      "learning_rate": 4.895140727789038e-07,
      "loss": 0.4987,
      "step": 3991
    },
    {
      "epoch": 0.6807639836289222,
      "grad_norm": 0.3770885765552521,
      "learning_rate": 4.890392189017376e-07,
      "loss": 0.5131,
      "step": 3992
    },
    {
      "epoch": 0.6809345156889495,
      "grad_norm": 0.5495342016220093,
      "learning_rate": 4.885645208965779e-07,
      "loss": 0.4944,
      "step": 3993
    },
    {
      "epoch": 0.6811050477489768,
      "grad_norm": 0.5257266759872437,
      "learning_rate": 4.880899789082349e-07,
      "loss": 0.5086,
      "step": 3994
    },
    {
      "epoch": 0.6812755798090041,
      "grad_norm": 0.41709569096565247,
      "learning_rate": 4.876155930814696e-07,
      "loss": 0.5007,
      "step": 3995
    },
    {
      "epoch": 0.6814461118690314,
      "grad_norm": 0.5326833724975586,
      "learning_rate": 4.87141363560998e-07,
      "loss": 0.4997,
      "step": 3996
    },
    {
      "epoch": 0.6816166439290586,
      "grad_norm": 0.4406546354293823,
      "learning_rate": 4.866672904914863e-07,
      "loss": 0.5021,
      "step": 3997
    },
    {
      "epoch": 0.681787175989086,
      "grad_norm": 0.4062040150165558,
      "learning_rate": 4.861933740175535e-07,
      "loss": 0.5065,
      "step": 3998
    },
    {
      "epoch": 0.6819577080491133,
      "grad_norm": 0.46950456500053406,
      "learning_rate": 4.857196142837716e-07,
      "loss": 0.4975,
      "step": 3999
    },
    {
      "epoch": 0.6821282401091405,
      "grad_norm": 0.3387594223022461,
      "learning_rate": 4.852460114346635e-07,
      "loss": 0.5033,
      "step": 4000
    },
    {
      "epoch": 0.6822987721691678,
      "grad_norm": 0.44981011748313904,
      "learning_rate": 4.847725656147052e-07,
      "loss": 0.5117,
      "step": 4001
    },
    {
      "epoch": 0.682469304229195,
      "grad_norm": 0.3584601581096649,
      "learning_rate": 4.842992769683243e-07,
      "loss": 0.5091,
      "step": 4002
    },
    {
      "epoch": 0.6826398362892224,
      "grad_norm": 0.38688042759895325,
      "learning_rate": 4.838261456399012e-07,
      "loss": 0.5131,
      "step": 4003
    },
    {
      "epoch": 0.6828103683492497,
      "grad_norm": 0.43309056758880615,
      "learning_rate": 4.833531717737673e-07,
      "loss": 0.5038,
      "step": 4004
    },
    {
      "epoch": 0.682980900409277,
      "grad_norm": 0.3956446647644043,
      "learning_rate": 4.82880355514207e-07,
      "loss": 0.5068,
      "step": 4005
    },
    {
      "epoch": 0.6831514324693042,
      "grad_norm": 0.3466695547103882,
      "learning_rate": 4.824076970054554e-07,
      "loss": 0.4962,
      "step": 4006
    },
    {
      "epoch": 0.6833219645293315,
      "grad_norm": 0.34976479411125183,
      "learning_rate": 4.819351963917004e-07,
      "loss": 0.4899,
      "step": 4007
    },
    {
      "epoch": 0.6834924965893588,
      "grad_norm": 0.4025235176086426,
      "learning_rate": 4.814628538170819e-07,
      "loss": 0.506,
      "step": 4008
    },
    {
      "epoch": 0.6836630286493861,
      "grad_norm": 0.36366453766822815,
      "learning_rate": 4.809906694256903e-07,
      "loss": 0.5015,
      "step": 4009
    },
    {
      "epoch": 0.6838335607094134,
      "grad_norm": 0.36603251099586487,
      "learning_rate": 4.805186433615691e-07,
      "loss": 0.5039,
      "step": 4010
    },
    {
      "epoch": 0.6840040927694406,
      "grad_norm": 0.4938323199748993,
      "learning_rate": 4.800467757687131e-07,
      "loss": 0.4977,
      "step": 4011
    },
    {
      "epoch": 0.6841746248294679,
      "grad_norm": 0.5089431405067444,
      "learning_rate": 4.795750667910684e-07,
      "loss": 0.4969,
      "step": 4012
    },
    {
      "epoch": 0.6843451568894953,
      "grad_norm": 0.3810231685638428,
      "learning_rate": 4.791035165725332e-07,
      "loss": 0.5231,
      "step": 4013
    },
    {
      "epoch": 0.6845156889495225,
      "grad_norm": 0.3925723433494568,
      "learning_rate": 4.786321252569574e-07,
      "loss": 0.5087,
      "step": 4014
    },
    {
      "epoch": 0.6846862210095498,
      "grad_norm": 0.4524107873439789,
      "learning_rate": 4.781608929881412e-07,
      "loss": 0.4986,
      "step": 4015
    },
    {
      "epoch": 0.684856753069577,
      "grad_norm": 0.3928070664405823,
      "learning_rate": 4.776898199098377e-07,
      "loss": 0.5033,
      "step": 4016
    },
    {
      "epoch": 0.6850272851296043,
      "grad_norm": 0.47441133856773376,
      "learning_rate": 4.772189061657511e-07,
      "loss": 0.5035,
      "step": 4017
    },
    {
      "epoch": 0.6851978171896317,
      "grad_norm": 0.5430171489715576,
      "learning_rate": 4.767481518995362e-07,
      "loss": 0.5076,
      "step": 4018
    },
    {
      "epoch": 0.685368349249659,
      "grad_norm": 0.4817226231098175,
      "learning_rate": 4.762775572548e-07,
      "loss": 0.5049,
      "step": 4019
    },
    {
      "epoch": 0.6855388813096862,
      "grad_norm": 0.5431926846504211,
      "learning_rate": 4.7580712237510067e-07,
      "loss": 0.4965,
      "step": 4020
    },
    {
      "epoch": 0.6857094133697135,
      "grad_norm": 0.47494205832481384,
      "learning_rate": 4.753368474039474e-07,
      "loss": 0.5144,
      "step": 4021
    },
    {
      "epoch": 0.6858799454297408,
      "grad_norm": 0.3338436782360077,
      "learning_rate": 4.7486673248480113e-07,
      "loss": 0.503,
      "step": 4022
    },
    {
      "epoch": 0.6860504774897681,
      "grad_norm": 0.5947521328926086,
      "learning_rate": 4.743967777610728e-07,
      "loss": 0.5073,
      "step": 4023
    },
    {
      "epoch": 0.6862210095497954,
      "grad_norm": 0.5065069794654846,
      "learning_rate": 4.739269833761257e-07,
      "loss": 0.4901,
      "step": 4024
    },
    {
      "epoch": 0.6863915416098226,
      "grad_norm": 0.594238817691803,
      "learning_rate": 4.7345734947327354e-07,
      "loss": 0.4983,
      "step": 4025
    },
    {
      "epoch": 0.6865620736698499,
      "grad_norm": 0.4637637734413147,
      "learning_rate": 4.7298787619578195e-07,
      "loss": 0.5051,
      "step": 4026
    },
    {
      "epoch": 0.6867326057298773,
      "grad_norm": 0.4788523316383362,
      "learning_rate": 4.725185636868659e-07,
      "loss": 0.5207,
      "step": 4027
    },
    {
      "epoch": 0.6869031377899045,
      "grad_norm": 1.1171875,
      "learning_rate": 4.7204941208969283e-07,
      "loss": 0.5016,
      "step": 4028
    },
    {
      "epoch": 0.6870736698499318,
      "grad_norm": 0.4261132776737213,
      "learning_rate": 4.7158042154738094e-07,
      "loss": 0.4909,
      "step": 4029
    },
    {
      "epoch": 0.687244201909959,
      "grad_norm": 0.5414057970046997,
      "learning_rate": 4.711115922029978e-07,
      "loss": 0.507,
      "step": 4030
    },
    {
      "epoch": 0.6874147339699863,
      "grad_norm": 0.5545263290405273,
      "learning_rate": 4.706429241995643e-07,
      "loss": 0.507,
      "step": 4031
    },
    {
      "epoch": 0.6875852660300137,
      "grad_norm": 0.41530779004096985,
      "learning_rate": 4.7017441768004984e-07,
      "loss": 0.4946,
      "step": 4032
    },
    {
      "epoch": 0.687755798090041,
      "grad_norm": 0.4105913043022156,
      "learning_rate": 4.697060727873757e-07,
      "loss": 0.5042,
      "step": 4033
    },
    {
      "epoch": 0.6879263301500682,
      "grad_norm": 0.4197026491165161,
      "learning_rate": 4.692378896644138e-07,
      "loss": 0.5025,
      "step": 4034
    },
    {
      "epoch": 0.6880968622100955,
      "grad_norm": 0.418058842420578,
      "learning_rate": 4.6876986845398666e-07,
      "loss": 0.4983,
      "step": 4035
    },
    {
      "epoch": 0.6882673942701227,
      "grad_norm": 0.5100137591362,
      "learning_rate": 4.683020092988663e-07,
      "loss": 0.5035,
      "step": 4036
    },
    {
      "epoch": 0.6884379263301501,
      "grad_norm": 0.44037339091300964,
      "learning_rate": 4.6783431234177785e-07,
      "loss": 0.4955,
      "step": 4037
    },
    {
      "epoch": 0.6886084583901774,
      "grad_norm": 0.42577651143074036,
      "learning_rate": 4.6736677772539445e-07,
      "loss": 0.5116,
      "step": 4038
    },
    {
      "epoch": 0.6887789904502046,
      "grad_norm": 0.39582791924476624,
      "learning_rate": 4.668994055923401e-07,
      "loss": 0.4927,
      "step": 4039
    },
    {
      "epoch": 0.6889495225102319,
      "grad_norm": 0.5501611828804016,
      "learning_rate": 4.6643219608519124e-07,
      "loss": 0.4979,
      "step": 4040
    },
    {
      "epoch": 0.6891200545702592,
      "grad_norm": 0.5625162124633789,
      "learning_rate": 4.659651493464721e-07,
      "loss": 0.499,
      "step": 4041
    },
    {
      "epoch": 0.6892905866302865,
      "grad_norm": 0.42893165349960327,
      "learning_rate": 4.6549826551865875e-07,
      "loss": 0.4943,
      "step": 4042
    },
    {
      "epoch": 0.6894611186903138,
      "grad_norm": 0.47238093614578247,
      "learning_rate": 4.650315447441774e-07,
      "loss": 0.4977,
      "step": 4043
    },
    {
      "epoch": 0.689631650750341,
      "grad_norm": 0.38662639260292053,
      "learning_rate": 4.645649871654046e-07,
      "loss": 0.5065,
      "step": 4044
    },
    {
      "epoch": 0.6898021828103683,
      "grad_norm": 0.4396331012248993,
      "learning_rate": 4.64098592924666e-07,
      "loss": 0.4983,
      "step": 4045
    },
    {
      "epoch": 0.6899727148703957,
      "grad_norm": 0.4275635778903961,
      "learning_rate": 4.636323621642389e-07,
      "loss": 0.5036,
      "step": 4046
    },
    {
      "epoch": 0.690143246930423,
      "grad_norm": 0.4336490333080292,
      "learning_rate": 4.631662950263503e-07,
      "loss": 0.4962,
      "step": 4047
    },
    {
      "epoch": 0.6903137789904502,
      "grad_norm": 0.40953293442726135,
      "learning_rate": 4.627003916531761e-07,
      "loss": 0.5083,
      "step": 4048
    },
    {
      "epoch": 0.6904843110504775,
      "grad_norm": 0.44607603549957275,
      "learning_rate": 4.6223465218684463e-07,
      "loss": 0.5046,
      "step": 4049
    },
    {
      "epoch": 0.6906548431105047,
      "grad_norm": 0.3914839029312134,
      "learning_rate": 4.617690767694319e-07,
      "loss": 0.4991,
      "step": 4050
    },
    {
      "epoch": 0.6908253751705321,
      "grad_norm": 0.35671597719192505,
      "learning_rate": 4.6130366554296494e-07,
      "loss": 0.5078,
      "step": 4051
    },
    {
      "epoch": 0.6909959072305594,
      "grad_norm": 0.5149934887886047,
      "learning_rate": 4.608384186494206e-07,
      "loss": 0.4957,
      "step": 4052
    },
    {
      "epoch": 0.6911664392905866,
      "grad_norm": 0.3881840109825134,
      "learning_rate": 4.6037333623072613e-07,
      "loss": 0.4949,
      "step": 4053
    },
    {
      "epoch": 0.6913369713506139,
      "grad_norm": 0.4083363115787506,
      "learning_rate": 4.5990841842875714e-07,
      "loss": 0.4928,
      "step": 4054
    },
    {
      "epoch": 0.6915075034106412,
      "grad_norm": 0.4419262707233429,
      "learning_rate": 4.5944366538534027e-07,
      "loss": 0.4883,
      "step": 4055
    },
    {
      "epoch": 0.6916780354706685,
      "grad_norm": 0.43516024947166443,
      "learning_rate": 4.5897907724225194e-07,
      "loss": 0.4854,
      "step": 4056
    },
    {
      "epoch": 0.6918485675306958,
      "grad_norm": 0.6483615636825562,
      "learning_rate": 4.585146541412168e-07,
      "loss": 0.5063,
      "step": 4057
    },
    {
      "epoch": 0.692019099590723,
      "grad_norm": 0.49583396315574646,
      "learning_rate": 4.5805039622391174e-07,
      "loss": 0.4937,
      "step": 4058
    },
    {
      "epoch": 0.6921896316507503,
      "grad_norm": 0.4556744694709778,
      "learning_rate": 4.575863036319605e-07,
      "loss": 0.5005,
      "step": 4059
    },
    {
      "epoch": 0.6923601637107776,
      "grad_norm": 0.5007831454277039,
      "learning_rate": 4.5712237650693804e-07,
      "loss": 0.5054,
      "step": 4060
    },
    {
      "epoch": 0.692530695770805,
      "grad_norm": 0.41077107191085815,
      "learning_rate": 4.5665861499036894e-07,
      "loss": 0.5005,
      "step": 4061
    },
    {
      "epoch": 0.6927012278308322,
      "grad_norm": 0.48699361085891724,
      "learning_rate": 4.561950192237259e-07,
      "loss": 0.4917,
      "step": 4062
    },
    {
      "epoch": 0.6928717598908595,
      "grad_norm": 0.5592864751815796,
      "learning_rate": 4.5573158934843236e-07,
      "loss": 0.5045,
      "step": 4063
    },
    {
      "epoch": 0.6930422919508867,
      "grad_norm": 0.5952345132827759,
      "learning_rate": 4.5526832550586056e-07,
      "loss": 0.5055,
      "step": 4064
    },
    {
      "epoch": 0.693212824010914,
      "grad_norm": 0.5739447474479675,
      "learning_rate": 4.5480522783733276e-07,
      "loss": 0.4986,
      "step": 4065
    },
    {
      "epoch": 0.6933833560709414,
      "grad_norm": 0.4101806581020355,
      "learning_rate": 4.543422964841188e-07,
      "loss": 0.506,
      "step": 4066
    },
    {
      "epoch": 0.6935538881309686,
      "grad_norm": 0.561511754989624,
      "learning_rate": 4.538795315874405e-07,
      "loss": 0.4895,
      "step": 4067
    },
    {
      "epoch": 0.6937244201909959,
      "grad_norm": 0.539454996585846,
      "learning_rate": 4.5341693328846627e-07,
      "loss": 0.5154,
      "step": 4068
    },
    {
      "epoch": 0.6938949522510232,
      "grad_norm": 0.4128101170063019,
      "learning_rate": 4.5295450172831515e-07,
      "loss": 0.502,
      "step": 4069
    },
    {
      "epoch": 0.6940654843110505,
      "grad_norm": 0.4938734173774719,
      "learning_rate": 4.524922370480555e-07,
      "loss": 0.4923,
      "step": 4070
    },
    {
      "epoch": 0.6942360163710778,
      "grad_norm": 0.4720302224159241,
      "learning_rate": 4.5203013938870324e-07,
      "loss": 0.5023,
      "step": 4071
    },
    {
      "epoch": 0.694406548431105,
      "grad_norm": 0.42051199078559875,
      "learning_rate": 4.51568208891225e-07,
      "loss": 0.5067,
      "step": 4072
    },
    {
      "epoch": 0.6945770804911323,
      "grad_norm": 0.5297553539276123,
      "learning_rate": 4.511064456965358e-07,
      "loss": 0.4957,
      "step": 4073
    },
    {
      "epoch": 0.6947476125511596,
      "grad_norm": 0.5808680653572083,
      "learning_rate": 4.5064484994549966e-07,
      "loss": 0.5094,
      "step": 4074
    },
    {
      "epoch": 0.694918144611187,
      "grad_norm": 0.5174522399902344,
      "learning_rate": 4.5018342177892853e-07,
      "loss": 0.5117,
      "step": 4075
    },
    {
      "epoch": 0.6950886766712142,
      "grad_norm": 0.41268858313560486,
      "learning_rate": 4.4972216133758565e-07,
      "loss": 0.5047,
      "step": 4076
    },
    {
      "epoch": 0.6952592087312415,
      "grad_norm": 0.5427170395851135,
      "learning_rate": 4.4926106876218047e-07,
      "loss": 0.5143,
      "step": 4077
    },
    {
      "epoch": 0.6954297407912687,
      "grad_norm": 0.545264720916748,
      "learning_rate": 4.4880014419337274e-07,
      "loss": 0.5045,
      "step": 4078
    },
    {
      "epoch": 0.695600272851296,
      "grad_norm": 0.4440051317214966,
      "learning_rate": 4.483393877717709e-07,
      "loss": 0.4965,
      "step": 4079
    },
    {
      "epoch": 0.6957708049113234,
      "grad_norm": 0.6596631407737732,
      "learning_rate": 4.4787879963793116e-07,
      "loss": 0.5055,
      "step": 4080
    },
    {
      "epoch": 0.6959413369713506,
      "grad_norm": 0.6201470494270325,
      "learning_rate": 4.474183799323593e-07,
      "loss": 0.5144,
      "step": 4081
    },
    {
      "epoch": 0.6961118690313779,
      "grad_norm": 0.5427371263504028,
      "learning_rate": 4.4695812879550943e-07,
      "loss": 0.5019,
      "step": 4082
    },
    {
      "epoch": 0.6962824010914052,
      "grad_norm": 0.6451879739761353,
      "learning_rate": 4.4649804636778466e-07,
      "loss": 0.4977,
      "step": 4083
    },
    {
      "epoch": 0.6964529331514324,
      "grad_norm": 0.47948503494262695,
      "learning_rate": 4.460381327895354e-07,
      "loss": 0.5071,
      "step": 4084
    },
    {
      "epoch": 0.6966234652114598,
      "grad_norm": 0.7414005398750305,
      "learning_rate": 4.455783882010617e-07,
      "loss": 0.4989,
      "step": 4085
    },
    {
      "epoch": 0.696793997271487,
      "grad_norm": 0.4640277326107025,
      "learning_rate": 4.45118812742612e-07,
      "loss": 0.4982,
      "step": 4086
    },
    {
      "epoch": 0.6969645293315143,
      "grad_norm": 0.6825625896453857,
      "learning_rate": 4.4465940655438246e-07,
      "loss": 0.52,
      "step": 4087
    },
    {
      "epoch": 0.6971350613915416,
      "grad_norm": 0.7004467248916626,
      "learning_rate": 4.442001697765187e-07,
      "loss": 0.4982,
      "step": 4088
    },
    {
      "epoch": 0.697305593451569,
      "grad_norm": 0.40038001537323,
      "learning_rate": 4.437411025491131e-07,
      "loss": 0.498,
      "step": 4089
    },
    {
      "epoch": 0.6974761255115962,
      "grad_norm": 0.5949633121490479,
      "learning_rate": 4.432822050122077e-07,
      "loss": 0.4982,
      "step": 4090
    },
    {
      "epoch": 0.6976466575716235,
      "grad_norm": 0.4295502007007599,
      "learning_rate": 4.4282347730579203e-07,
      "loss": 0.51,
      "step": 4091
    },
    {
      "epoch": 0.6978171896316507,
      "grad_norm": 0.44073379039764404,
      "learning_rate": 4.4236491956980425e-07,
      "loss": 0.5196,
      "step": 4092
    },
    {
      "epoch": 0.697987721691678,
      "grad_norm": 0.5889855623245239,
      "learning_rate": 4.419065319441306e-07,
      "loss": 0.496,
      "step": 4093
    },
    {
      "epoch": 0.6981582537517054,
      "grad_norm": 0.39880111813545227,
      "learning_rate": 4.4144831456860485e-07,
      "loss": 0.5035,
      "step": 4094
    },
    {
      "epoch": 0.6983287858117326,
      "grad_norm": 0.5149140954017639,
      "learning_rate": 4.409902675830095e-07,
      "loss": 0.5002,
      "step": 4095
    },
    {
      "epoch": 0.6984993178717599,
      "grad_norm": 0.41448941826820374,
      "learning_rate": 4.4053239112707483e-07,
      "loss": 0.5047,
      "step": 4096
    },
    {
      "epoch": 0.6986698499317872,
      "grad_norm": 0.5223191976547241,
      "learning_rate": 4.400746853404795e-07,
      "loss": 0.504,
      "step": 4097
    },
    {
      "epoch": 0.6988403819918144,
      "grad_norm": 0.3906937837600708,
      "learning_rate": 4.3961715036284906e-07,
      "loss": 0.4993,
      "step": 4098
    },
    {
      "epoch": 0.6990109140518418,
      "grad_norm": 0.5009825825691223,
      "learning_rate": 4.391597863337579e-07,
      "loss": 0.5064,
      "step": 4099
    },
    {
      "epoch": 0.699181446111869,
      "grad_norm": 0.4940502643585205,
      "learning_rate": 4.3870259339272827e-07,
      "loss": 0.4939,
      "step": 4100
    },
    {
      "epoch": 0.6993519781718963,
      "grad_norm": 0.3717692792415619,
      "learning_rate": 4.3824557167922913e-07,
      "loss": 0.4929,
      "step": 4101
    },
    {
      "epoch": 0.6995225102319236,
      "grad_norm": 0.4262048900127411,
      "learning_rate": 4.3778872133267915e-07,
      "loss": 0.4985,
      "step": 4102
    },
    {
      "epoch": 0.6996930422919508,
      "grad_norm": 0.3647048771381378,
      "learning_rate": 4.373320424924427e-07,
      "loss": 0.4954,
      "step": 4103
    },
    {
      "epoch": 0.6998635743519782,
      "grad_norm": 0.5506351590156555,
      "learning_rate": 4.368755352978329e-07,
      "loss": 0.5134,
      "step": 4104
    },
    {
      "epoch": 0.7000341064120055,
      "grad_norm": 0.4715368449687958,
      "learning_rate": 4.364191998881105e-07,
      "loss": 0.5025,
      "step": 4105
    },
    {
      "epoch": 0.7002046384720327,
      "grad_norm": 0.46583038568496704,
      "learning_rate": 4.3596303640248395e-07,
      "loss": 0.5079,
      "step": 4106
    },
    {
      "epoch": 0.70037517053206,
      "grad_norm": 0.475370854139328,
      "learning_rate": 4.3550704498010837e-07,
      "loss": 0.4936,
      "step": 4107
    },
    {
      "epoch": 0.7005457025920873,
      "grad_norm": 0.46047744154930115,
      "learning_rate": 4.350512257600871e-07,
      "loss": 0.493,
      "step": 4108
    },
    {
      "epoch": 0.7007162346521146,
      "grad_norm": 0.5731782913208008,
      "learning_rate": 4.3459557888147144e-07,
      "loss": 0.5037,
      "step": 4109
    },
    {
      "epoch": 0.7008867667121419,
      "grad_norm": 0.4613938331604004,
      "learning_rate": 4.341401044832583e-07,
      "loss": 0.4878,
      "step": 4110
    },
    {
      "epoch": 0.7010572987721692,
      "grad_norm": 0.43121498823165894,
      "learning_rate": 4.336848027043948e-07,
      "loss": 0.5007,
      "step": 4111
    },
    {
      "epoch": 0.7012278308321964,
      "grad_norm": 0.44641441106796265,
      "learning_rate": 4.332296736837725e-07,
      "loss": 0.4976,
      "step": 4112
    },
    {
      "epoch": 0.7013983628922238,
      "grad_norm": 0.39351579546928406,
      "learning_rate": 4.327747175602321e-07,
      "loss": 0.4968,
      "step": 4113
    },
    {
      "epoch": 0.701568894952251,
      "grad_norm": 0.4338296949863434,
      "learning_rate": 4.3231993447256097e-07,
      "loss": 0.5026,
      "step": 4114
    },
    {
      "epoch": 0.7017394270122783,
      "grad_norm": 0.47230708599090576,
      "learning_rate": 4.3186532455949395e-07,
      "loss": 0.5041,
      "step": 4115
    },
    {
      "epoch": 0.7019099590723056,
      "grad_norm": 0.49445465207099915,
      "learning_rate": 4.314108879597122e-07,
      "loss": 0.4906,
      "step": 4116
    },
    {
      "epoch": 0.7020804911323328,
      "grad_norm": 0.36397585272789,
      "learning_rate": 4.309566248118451e-07,
      "loss": 0.495,
      "step": 4117
    },
    {
      "epoch": 0.7022510231923602,
      "grad_norm": 0.4672987759113312,
      "learning_rate": 4.30502535254469e-07,
      "loss": 0.4991,
      "step": 4118
    },
    {
      "epoch": 0.7024215552523875,
      "grad_norm": 0.4790498614311218,
      "learning_rate": 4.300486194261058e-07,
      "loss": 0.5088,
      "step": 4119
    },
    {
      "epoch": 0.7025920873124147,
      "grad_norm": 0.49927717447280884,
      "learning_rate": 4.29594877465227e-07,
      "loss": 0.4977,
      "step": 4120
    },
    {
      "epoch": 0.702762619372442,
      "grad_norm": 0.5749465823173523,
      "learning_rate": 4.291413095102487e-07,
      "loss": 0.4968,
      "step": 4121
    },
    {
      "epoch": 0.7029331514324693,
      "grad_norm": 0.43878257274627686,
      "learning_rate": 4.28687915699535e-07,
      "loss": 0.4952,
      "step": 4122
    },
    {
      "epoch": 0.7031036834924966,
      "grad_norm": 0.5072469711303711,
      "learning_rate": 4.282346961713972e-07,
      "loss": 0.5085,
      "step": 4123
    },
    {
      "epoch": 0.7032742155525239,
      "grad_norm": 0.5195170044898987,
      "learning_rate": 4.27781651064092e-07,
      "loss": 0.505,
      "step": 4124
    },
    {
      "epoch": 0.7034447476125512,
      "grad_norm": 0.42486515641212463,
      "learning_rate": 4.273287805158245e-07,
      "loss": 0.51,
      "step": 4125
    },
    {
      "epoch": 0.7036152796725784,
      "grad_norm": 0.4187718331813812,
      "learning_rate": 4.2687608466474585e-07,
      "loss": 0.5034,
      "step": 4126
    },
    {
      "epoch": 0.7037858117326057,
      "grad_norm": 0.48323482275009155,
      "learning_rate": 4.264235636489542e-07,
      "loss": 0.4961,
      "step": 4127
    },
    {
      "epoch": 0.703956343792633,
      "grad_norm": 0.5653338432312012,
      "learning_rate": 4.2597121760649313e-07,
      "loss": 0.5055,
      "step": 4128
    },
    {
      "epoch": 0.7041268758526603,
      "grad_norm": 0.4599945545196533,
      "learning_rate": 4.255190466753552e-07,
      "loss": 0.5071,
      "step": 4129
    },
    {
      "epoch": 0.7042974079126876,
      "grad_norm": 0.5306625366210938,
      "learning_rate": 4.2506705099347703e-07,
      "loss": 0.5123,
      "step": 4130
    },
    {
      "epoch": 0.7044679399727148,
      "grad_norm": 0.5877465605735779,
      "learning_rate": 4.246152306987435e-07,
      "loss": 0.5015,
      "step": 4131
    },
    {
      "epoch": 0.7046384720327421,
      "grad_norm": 0.5958652496337891,
      "learning_rate": 4.241635859289855e-07,
      "loss": 0.4952,
      "step": 4132
    },
    {
      "epoch": 0.7048090040927695,
      "grad_norm": 0.5003058910369873,
      "learning_rate": 4.237121168219798e-07,
      "loss": 0.5076,
      "step": 4133
    },
    {
      "epoch": 0.7049795361527967,
      "grad_norm": 0.45541825890541077,
      "learning_rate": 4.232608235154503e-07,
      "loss": 0.4911,
      "step": 4134
    },
    {
      "epoch": 0.705150068212824,
      "grad_norm": 0.5043857097625732,
      "learning_rate": 4.228097061470672e-07,
      "loss": 0.4892,
      "step": 4135
    },
    {
      "epoch": 0.7053206002728513,
      "grad_norm": 0.4577189087867737,
      "learning_rate": 4.223587648544471e-07,
      "loss": 0.4944,
      "step": 4136
    },
    {
      "epoch": 0.7054911323328786,
      "grad_norm": 0.4259335398674011,
      "learning_rate": 4.2190799977515155e-07,
      "loss": 0.4903,
      "step": 4137
    },
    {
      "epoch": 0.7056616643929059,
      "grad_norm": 0.5069748759269714,
      "learning_rate": 4.2145741104669094e-07,
      "loss": 0.4957,
      "step": 4138
    },
    {
      "epoch": 0.7058321964529332,
      "grad_norm": 0.47444620728492737,
      "learning_rate": 4.210069988065197e-07,
      "loss": 0.5005,
      "step": 4139
    },
    {
      "epoch": 0.7060027285129604,
      "grad_norm": 0.5006582140922546,
      "learning_rate": 4.2055676319203837e-07,
      "loss": 0.5068,
      "step": 4140
    },
    {
      "epoch": 0.7061732605729877,
      "grad_norm": 0.5743463039398193,
      "learning_rate": 4.201067043405957e-07,
      "loss": 0.5007,
      "step": 4141
    },
    {
      "epoch": 0.706343792633015,
      "grad_norm": 0.591344952583313,
      "learning_rate": 4.1965682238948417e-07,
      "loss": 0.4859,
      "step": 4142
    },
    {
      "epoch": 0.7065143246930423,
      "grad_norm": 0.43809980154037476,
      "learning_rate": 4.192071174759436e-07,
      "loss": 0.4964,
      "step": 4143
    },
    {
      "epoch": 0.7066848567530696,
      "grad_norm": 0.6915422081947327,
      "learning_rate": 4.1875758973715955e-07,
      "loss": 0.4972,
      "step": 4144
    },
    {
      "epoch": 0.7068553888130968,
      "grad_norm": 0.33378657698631287,
      "learning_rate": 4.183082393102637e-07,
      "loss": 0.4904,
      "step": 4145
    },
    {
      "epoch": 0.7070259208731241,
      "grad_norm": 0.7669427990913391,
      "learning_rate": 4.1785906633233233e-07,
      "loss": 0.5102,
      "step": 4146
    },
    {
      "epoch": 0.7071964529331515,
      "grad_norm": 0.6579416990280151,
      "learning_rate": 4.1741007094039024e-07,
      "loss": 0.4931,
      "step": 4147
    },
    {
      "epoch": 0.7073669849931787,
      "grad_norm": 0.5717936754226685,
      "learning_rate": 4.169612532714053e-07,
      "loss": 0.5032,
      "step": 4148
    },
    {
      "epoch": 0.707537517053206,
      "grad_norm": 0.6794208884239197,
      "learning_rate": 4.165126134622927e-07,
      "loss": 0.5005,
      "step": 4149
    },
    {
      "epoch": 0.7077080491132333,
      "grad_norm": 0.40618714690208435,
      "learning_rate": 4.1606415164991346e-07,
      "loss": 0.5082,
      "step": 4150
    },
    {
      "epoch": 0.7078785811732605,
      "grad_norm": 0.5979039072990417,
      "learning_rate": 4.1561586797107304e-07,
      "loss": 0.5037,
      "step": 4151
    },
    {
      "epoch": 0.7080491132332879,
      "grad_norm": 0.47896888852119446,
      "learning_rate": 4.151677625625238e-07,
      "loss": 0.5001,
      "step": 4152
    },
    {
      "epoch": 0.7082196452933152,
      "grad_norm": 0.44115006923675537,
      "learning_rate": 4.147198355609632e-07,
      "loss": 0.5108,
      "step": 4153
    },
    {
      "epoch": 0.7083901773533424,
      "grad_norm": 0.48424452543258667,
      "learning_rate": 4.1427208710303496e-07,
      "loss": 0.5033,
      "step": 4154
    },
    {
      "epoch": 0.7085607094133697,
      "grad_norm": 0.5310095548629761,
      "learning_rate": 4.1382451732532673e-07,
      "loss": 0.4922,
      "step": 4155
    },
    {
      "epoch": 0.708731241473397,
      "grad_norm": 0.44579505920410156,
      "learning_rate": 4.1337712636437326e-07,
      "loss": 0.4978,
      "step": 4156
    },
    {
      "epoch": 0.7089017735334243,
      "grad_norm": 0.6123139262199402,
      "learning_rate": 4.129299143566539e-07,
      "loss": 0.5046,
      "step": 4157
    },
    {
      "epoch": 0.7090723055934516,
      "grad_norm": 0.436917781829834,
      "learning_rate": 4.12482881438594e-07,
      "loss": 0.5029,
      "step": 4158
    },
    {
      "epoch": 0.7092428376534788,
      "grad_norm": 0.46517184376716614,
      "learning_rate": 4.1203602774656406e-07,
      "loss": 0.5099,
      "step": 4159
    },
    {
      "epoch": 0.7094133697135061,
      "grad_norm": 0.5581578016281128,
      "learning_rate": 4.1158935341687915e-07,
      "loss": 0.5083,
      "step": 4160
    },
    {
      "epoch": 0.7095839017735335,
      "grad_norm": 0.5356834530830383,
      "learning_rate": 4.1114285858580053e-07,
      "loss": 0.4991,
      "step": 4161
    },
    {
      "epoch": 0.7097544338335607,
      "grad_norm": 0.439238578081131,
      "learning_rate": 4.1069654338953494e-07,
      "loss": 0.5099,
      "step": 4162
    },
    {
      "epoch": 0.709924965893588,
      "grad_norm": 0.5053290128707886,
      "learning_rate": 4.1025040796423304e-07,
      "loss": 0.5099,
      "step": 4163
    },
    {
      "epoch": 0.7100954979536153,
      "grad_norm": 0.5811809301376343,
      "learning_rate": 4.098044524459918e-07,
      "loss": 0.5103,
      "step": 4164
    },
    {
      "epoch": 0.7102660300136425,
      "grad_norm": 0.47019311785697937,
      "learning_rate": 4.093586769708529e-07,
      "loss": 0.4947,
      "step": 4165
    },
    {
      "epoch": 0.7104365620736699,
      "grad_norm": 0.4350194036960602,
      "learning_rate": 4.089130816748032e-07,
      "loss": 0.5095,
      "step": 4166
    },
    {
      "epoch": 0.7106070941336972,
      "grad_norm": 0.5497310161590576,
      "learning_rate": 4.084676666937745e-07,
      "loss": 0.4981,
      "step": 4167
    },
    {
      "epoch": 0.7107776261937244,
      "grad_norm": 0.47565320134162903,
      "learning_rate": 4.080224321636438e-07,
      "loss": 0.5051,
      "step": 4168
    },
    {
      "epoch": 0.7109481582537517,
      "grad_norm": 0.43900391459465027,
      "learning_rate": 4.075773782202325e-07,
      "loss": 0.4887,
      "step": 4169
    },
    {
      "epoch": 0.711118690313779,
      "grad_norm": 0.44005173444747925,
      "learning_rate": 4.071325049993074e-07,
      "loss": 0.5046,
      "step": 4170
    },
    {
      "epoch": 0.7112892223738063,
      "grad_norm": 0.5328361392021179,
      "learning_rate": 4.0668781263658047e-07,
      "loss": 0.5037,
      "step": 4171
    },
    {
      "epoch": 0.7114597544338336,
      "grad_norm": 0.5043607354164124,
      "learning_rate": 4.0624330126770744e-07,
      "loss": 0.4974,
      "step": 4172
    },
    {
      "epoch": 0.7116302864938608,
      "grad_norm": 0.46430233120918274,
      "learning_rate": 4.057989710282897e-07,
      "loss": 0.5009,
      "step": 4173
    },
    {
      "epoch": 0.7118008185538881,
      "grad_norm": 0.4336589276790619,
      "learning_rate": 4.0535482205387324e-07,
      "loss": 0.4889,
      "step": 4174
    },
    {
      "epoch": 0.7119713506139154,
      "grad_norm": 0.5993257164955139,
      "learning_rate": 4.0491085447994874e-07,
      "loss": 0.4962,
      "step": 4175
    },
    {
      "epoch": 0.7121418826739427,
      "grad_norm": 0.6065536737442017,
      "learning_rate": 4.044670684419513e-07,
      "loss": 0.5099,
      "step": 4176
    },
    {
      "epoch": 0.71231241473397,
      "grad_norm": 0.42699581384658813,
      "learning_rate": 4.040234640752614e-07,
      "loss": 0.5028,
      "step": 4177
    },
    {
      "epoch": 0.7124829467939973,
      "grad_norm": 0.4277159869670868,
      "learning_rate": 4.035800415152027e-07,
      "loss": 0.5008,
      "step": 4178
    },
    {
      "epoch": 0.7126534788540245,
      "grad_norm": 0.4397408366203308,
      "learning_rate": 4.0313680089704456e-07,
      "loss": 0.502,
      "step": 4179
    },
    {
      "epoch": 0.7128240109140518,
      "grad_norm": 0.35132086277008057,
      "learning_rate": 4.0269374235600084e-07,
      "loss": 0.507,
      "step": 4180
    },
    {
      "epoch": 0.7129945429740792,
      "grad_norm": 0.44684553146362305,
      "learning_rate": 4.0225086602722857e-07,
      "loss": 0.4998,
      "step": 4181
    },
    {
      "epoch": 0.7131650750341064,
      "grad_norm": 0.39760348200798035,
      "learning_rate": 4.018081720458314e-07,
      "loss": 0.4939,
      "step": 4182
    },
    {
      "epoch": 0.7133356070941337,
      "grad_norm": 0.400193989276886,
      "learning_rate": 4.0136566054685514e-07,
      "loss": 0.5054,
      "step": 4183
    },
    {
      "epoch": 0.713506139154161,
      "grad_norm": 0.3968668580055237,
      "learning_rate": 4.009233316652911e-07,
      "loss": 0.4991,
      "step": 4184
    },
    {
      "epoch": 0.7136766712141883,
      "grad_norm": 0.3965734839439392,
      "learning_rate": 4.004811855360749e-07,
      "loss": 0.5012,
      "step": 4185
    },
    {
      "epoch": 0.7138472032742156,
      "grad_norm": 0.44471368193626404,
      "learning_rate": 4.0003922229408636e-07,
      "loss": 0.4973,
      "step": 4186
    },
    {
      "epoch": 0.7140177353342428,
      "grad_norm": 0.39058804512023926,
      "learning_rate": 3.9959744207414857e-07,
      "loss": 0.4928,
      "step": 4187
    },
    {
      "epoch": 0.7141882673942701,
      "grad_norm": 0.39435192942619324,
      "learning_rate": 3.9915584501103e-07,
      "loss": 0.5077,
      "step": 4188
    },
    {
      "epoch": 0.7143587994542974,
      "grad_norm": 0.49913209676742554,
      "learning_rate": 3.987144312394432e-07,
      "loss": 0.5066,
      "step": 4189
    },
    {
      "epoch": 0.7145293315143247,
      "grad_norm": 0.40681523084640503,
      "learning_rate": 3.982732008940432e-07,
      "loss": 0.5044,
      "step": 4190
    },
    {
      "epoch": 0.714699863574352,
      "grad_norm": 0.46381300687789917,
      "learning_rate": 3.9783215410943183e-07,
      "loss": 0.5074,
      "step": 4191
    },
    {
      "epoch": 0.7148703956343793,
      "grad_norm": 0.4717521071434021,
      "learning_rate": 3.973912910201521e-07,
      "loss": 0.5003,
      "step": 4192
    },
    {
      "epoch": 0.7150409276944065,
      "grad_norm": 0.5403792262077332,
      "learning_rate": 3.9695061176069277e-07,
      "loss": 0.5082,
      "step": 4193
    },
    {
      "epoch": 0.7152114597544338,
      "grad_norm": 0.3403872549533844,
      "learning_rate": 3.9651011646548626e-07,
      "loss": 0.4907,
      "step": 4194
    },
    {
      "epoch": 0.7153819918144612,
      "grad_norm": 0.5781459212303162,
      "learning_rate": 3.96069805268908e-07,
      "loss": 0.5053,
      "step": 4195
    },
    {
      "epoch": 0.7155525238744884,
      "grad_norm": 0.4355068802833557,
      "learning_rate": 3.9562967830527803e-07,
      "loss": 0.4954,
      "step": 4196
    },
    {
      "epoch": 0.7157230559345157,
      "grad_norm": 0.47292977571487427,
      "learning_rate": 3.9518973570886025e-07,
      "loss": 0.5048,
      "step": 4197
    },
    {
      "epoch": 0.715893587994543,
      "grad_norm": 0.6361278891563416,
      "learning_rate": 3.947499776138624e-07,
      "loss": 0.5122,
      "step": 4198
    },
    {
      "epoch": 0.7160641200545702,
      "grad_norm": 0.5250843167304993,
      "learning_rate": 3.9431040415443446e-07,
      "loss": 0.5035,
      "step": 4199
    },
    {
      "epoch": 0.7162346521145976,
      "grad_norm": 0.5073175430297852,
      "learning_rate": 3.938710154646727e-07,
      "loss": 0.5028,
      "step": 4200
    },
    {
      "epoch": 0.7164051841746248,
      "grad_norm": 0.5830045342445374,
      "learning_rate": 3.934318116786149e-07,
      "loss": 0.5096,
      "step": 4201
    },
    {
      "epoch": 0.7165757162346521,
      "grad_norm": 0.6108426451683044,
      "learning_rate": 3.9299279293024254e-07,
      "loss": 0.5094,
      "step": 4202
    },
    {
      "epoch": 0.7167462482946794,
      "grad_norm": 0.3694036900997162,
      "learning_rate": 3.9255395935348243e-07,
      "loss": 0.4977,
      "step": 4203
    },
    {
      "epoch": 0.7169167803547067,
      "grad_norm": 0.5815227031707764,
      "learning_rate": 3.921153110822027e-07,
      "loss": 0.5081,
      "step": 4204
    },
    {
      "epoch": 0.717087312414734,
      "grad_norm": 0.43358278274536133,
      "learning_rate": 3.9167684825021636e-07,
      "loss": 0.5,
      "step": 4205
    },
    {
      "epoch": 0.7172578444747613,
      "grad_norm": 0.38220107555389404,
      "learning_rate": 3.9123857099127943e-07,
      "loss": 0.4998,
      "step": 4206
    },
    {
      "epoch": 0.7174283765347885,
      "grad_norm": 0.4499671757221222,
      "learning_rate": 3.908004794390915e-07,
      "loss": 0.5066,
      "step": 4207
    },
    {
      "epoch": 0.7175989085948158,
      "grad_norm": 0.4196017384529114,
      "learning_rate": 3.903625737272945e-07,
      "loss": 0.4987,
      "step": 4208
    },
    {
      "epoch": 0.7177694406548432,
      "grad_norm": 0.4220138490200043,
      "learning_rate": 3.899248539894757e-07,
      "loss": 0.5036,
      "step": 4209
    },
    {
      "epoch": 0.7179399727148704,
      "grad_norm": 0.44207295775413513,
      "learning_rate": 3.8948732035916393e-07,
      "loss": 0.5113,
      "step": 4210
    },
    {
      "epoch": 0.7181105047748977,
      "grad_norm": 0.3891289234161377,
      "learning_rate": 3.89049972969831e-07,
      "loss": 0.5048,
      "step": 4211
    },
    {
      "epoch": 0.718281036834925,
      "grad_norm": 0.41989684104919434,
      "learning_rate": 3.886128119548939e-07,
      "loss": 0.5011,
      "step": 4212
    },
    {
      "epoch": 0.7184515688949522,
      "grad_norm": 0.466969758272171,
      "learning_rate": 3.881758374477106e-07,
      "loss": 0.51,
      "step": 4213
    },
    {
      "epoch": 0.7186221009549796,
      "grad_norm": 0.39666444063186646,
      "learning_rate": 3.877390495815835e-07,
      "loss": 0.5074,
      "step": 4214
    },
    {
      "epoch": 0.7187926330150068,
      "grad_norm": 0.3840790390968323,
      "learning_rate": 3.8730244848975765e-07,
      "loss": 0.502,
      "step": 4215
    },
    {
      "epoch": 0.7189631650750341,
      "grad_norm": 0.36528825759887695,
      "learning_rate": 3.868660343054214e-07,
      "loss": 0.511,
      "step": 4216
    },
    {
      "epoch": 0.7191336971350614,
      "grad_norm": 0.4255644381046295,
      "learning_rate": 3.8642980716170514e-07,
      "loss": 0.5096,
      "step": 4217
    },
    {
      "epoch": 0.7193042291950886,
      "grad_norm": 0.36999639868736267,
      "learning_rate": 3.859937671916833e-07,
      "loss": 0.5065,
      "step": 4218
    },
    {
      "epoch": 0.719474761255116,
      "grad_norm": 0.4373064935207367,
      "learning_rate": 3.8555791452837303e-07,
      "loss": 0.5037,
      "step": 4219
    },
    {
      "epoch": 0.7196452933151433,
      "grad_norm": 0.34947025775909424,
      "learning_rate": 3.8512224930473313e-07,
      "loss": 0.5005,
      "step": 4220
    },
    {
      "epoch": 0.7198158253751705,
      "grad_norm": 0.40553170442581177,
      "learning_rate": 3.846867716536676e-07,
      "loss": 0.5087,
      "step": 4221
    },
    {
      "epoch": 0.7199863574351978,
      "grad_norm": 0.45594432950019836,
      "learning_rate": 3.8425148170802064e-07,
      "loss": 0.5041,
      "step": 4222
    },
    {
      "epoch": 0.720156889495225,
      "grad_norm": 0.4455185532569885,
      "learning_rate": 3.83816379600581e-07,
      "loss": 0.4992,
      "step": 4223
    },
    {
      "epoch": 0.7203274215552524,
      "grad_norm": 0.4536997377872467,
      "learning_rate": 3.833814654640791e-07,
      "loss": 0.4979,
      "step": 4224
    },
    {
      "epoch": 0.7204979536152797,
      "grad_norm": 0.42399755120277405,
      "learning_rate": 3.82946739431189e-07,
      "loss": 0.4926,
      "step": 4225
    },
    {
      "epoch": 0.720668485675307,
      "grad_norm": 0.4298648238182068,
      "learning_rate": 3.825122016345261e-07,
      "loss": 0.5038,
      "step": 4226
    },
    {
      "epoch": 0.7208390177353342,
      "grad_norm": 0.416885107755661,
      "learning_rate": 3.8207785220664944e-07,
      "loss": 0.4946,
      "step": 4227
    },
    {
      "epoch": 0.7210095497953616,
      "grad_norm": 0.4566577076911926,
      "learning_rate": 3.8164369128006034e-07,
      "loss": 0.4859,
      "step": 4228
    },
    {
      "epoch": 0.7211800818553888,
      "grad_norm": 0.4706067144870758,
      "learning_rate": 3.8120971898720165e-07,
      "loss": 0.4922,
      "step": 4229
    },
    {
      "epoch": 0.7213506139154161,
      "grad_norm": 0.3943629562854767,
      "learning_rate": 3.807759354604609e-07,
      "loss": 0.506,
      "step": 4230
    },
    {
      "epoch": 0.7215211459754434,
      "grad_norm": 0.4761364459991455,
      "learning_rate": 3.803423408321655e-07,
      "loss": 0.4932,
      "step": 4231
    },
    {
      "epoch": 0.7216916780354706,
      "grad_norm": 0.37824156880378723,
      "learning_rate": 3.7990893523458685e-07,
      "loss": 0.5017,
      "step": 4232
    },
    {
      "epoch": 0.721862210095498,
      "grad_norm": 0.5847135782241821,
      "learning_rate": 3.7947571879993864e-07,
      "loss": 0.4983,
      "step": 4233
    },
    {
      "epoch": 0.7220327421555253,
      "grad_norm": 0.4698280394077301,
      "learning_rate": 3.790426916603756e-07,
      "loss": 0.4898,
      "step": 4234
    },
    {
      "epoch": 0.7222032742155525,
      "grad_norm": 0.5401884913444519,
      "learning_rate": 3.786098539479959e-07,
      "loss": 0.5024,
      "step": 4235
    },
    {
      "epoch": 0.7223738062755798,
      "grad_norm": 0.5750718116760254,
      "learning_rate": 3.7817720579483967e-07,
      "loss": 0.5022,
      "step": 4236
    },
    {
      "epoch": 0.722544338335607,
      "grad_norm": 0.4136122167110443,
      "learning_rate": 3.7774474733288904e-07,
      "loss": 0.5019,
      "step": 4237
    },
    {
      "epoch": 0.7227148703956344,
      "grad_norm": 0.605913519859314,
      "learning_rate": 3.7731247869406856e-07,
      "loss": 0.4899,
      "step": 4238
    },
    {
      "epoch": 0.7228854024556617,
      "grad_norm": 0.32366135716438293,
      "learning_rate": 3.7688040001024475e-07,
      "loss": 0.491,
      "step": 4239
    },
    {
      "epoch": 0.723055934515689,
      "grad_norm": 0.6166736483573914,
      "learning_rate": 3.764485114132256e-07,
      "loss": 0.4998,
      "step": 4240
    },
    {
      "epoch": 0.7232264665757162,
      "grad_norm": 0.3734305799007416,
      "learning_rate": 3.7601681303476195e-07,
      "loss": 0.4989,
      "step": 4241
    },
    {
      "epoch": 0.7233969986357435,
      "grad_norm": 0.5646785497665405,
      "learning_rate": 3.7558530500654646e-07,
      "loss": 0.4991,
      "step": 4242
    },
    {
      "epoch": 0.7235675306957708,
      "grad_norm": 0.5683301687240601,
      "learning_rate": 3.7515398746021294e-07,
      "loss": 0.5114,
      "step": 4243
    },
    {
      "epoch": 0.7237380627557981,
      "grad_norm": 0.5351408123970032,
      "learning_rate": 3.747228605273381e-07,
      "loss": 0.496,
      "step": 4244
    },
    {
      "epoch": 0.7239085948158254,
      "grad_norm": 0.4870230555534363,
      "learning_rate": 3.742919243394402e-07,
      "loss": 0.4964,
      "step": 4245
    },
    {
      "epoch": 0.7240791268758526,
      "grad_norm": 0.4613642990589142,
      "learning_rate": 3.7386117902797896e-07,
      "loss": 0.4969,
      "step": 4246
    },
    {
      "epoch": 0.7242496589358799,
      "grad_norm": 0.48472538590431213,
      "learning_rate": 3.7343062472435623e-07,
      "loss": 0.509,
      "step": 4247
    },
    {
      "epoch": 0.7244201909959073,
      "grad_norm": 0.45103564858436584,
      "learning_rate": 3.730002615599158e-07,
      "loss": 0.4988,
      "step": 4248
    },
    {
      "epoch": 0.7245907230559345,
      "grad_norm": 0.49282583594322205,
      "learning_rate": 3.7257008966594224e-07,
      "loss": 0.4984,
      "step": 4249
    },
    {
      "epoch": 0.7247612551159618,
      "grad_norm": 0.39775902032852173,
      "learning_rate": 3.7214010917366263e-07,
      "loss": 0.501,
      "step": 4250
    },
    {
      "epoch": 0.724931787175989,
      "grad_norm": 0.47374874353408813,
      "learning_rate": 3.7171032021424574e-07,
      "loss": 0.5019,
      "step": 4251
    },
    {
      "epoch": 0.7251023192360164,
      "grad_norm": 0.53701251745224,
      "learning_rate": 3.7128072291880083e-07,
      "loss": 0.4915,
      "step": 4252
    },
    {
      "epoch": 0.7252728512960437,
      "grad_norm": 0.5349600911140442,
      "learning_rate": 3.708513174183799e-07,
      "loss": 0.5009,
      "step": 4253
    },
    {
      "epoch": 0.725443383356071,
      "grad_norm": 0.44183728098869324,
      "learning_rate": 3.7042210384397596e-07,
      "loss": 0.4969,
      "step": 4254
    },
    {
      "epoch": 0.7256139154160982,
      "grad_norm": 0.4504479169845581,
      "learning_rate": 3.6999308232652337e-07,
      "loss": 0.499,
      "step": 4255
    },
    {
      "epoch": 0.7257844474761255,
      "grad_norm": 0.5018069744110107,
      "learning_rate": 3.695642529968985e-07,
      "loss": 0.5044,
      "step": 4256
    },
    {
      "epoch": 0.7259549795361528,
      "grad_norm": 0.3822757601737976,
      "learning_rate": 3.691356159859178e-07,
      "loss": 0.508,
      "step": 4257
    },
    {
      "epoch": 0.7261255115961801,
      "grad_norm": 0.4669684171676636,
      "learning_rate": 3.6870717142434025e-07,
      "loss": 0.4972,
      "step": 4258
    },
    {
      "epoch": 0.7262960436562074,
      "grad_norm": 0.4292471706867218,
      "learning_rate": 3.682789194428657e-07,
      "loss": 0.5001,
      "step": 4259
    },
    {
      "epoch": 0.7264665757162346,
      "grad_norm": 0.3445581793785095,
      "learning_rate": 3.6785086017213557e-07,
      "loss": 0.4917,
      "step": 4260
    },
    {
      "epoch": 0.7266371077762619,
      "grad_norm": 0.5252216458320618,
      "learning_rate": 3.6742299374273174e-07,
      "loss": 0.4981,
      "step": 4261
    },
    {
      "epoch": 0.7268076398362893,
      "grad_norm": 0.4755789041519165,
      "learning_rate": 3.6699532028517795e-07,
      "loss": 0.4996,
      "step": 4262
    },
    {
      "epoch": 0.7269781718963165,
      "grad_norm": 0.4374355673789978,
      "learning_rate": 3.6656783992993887e-07,
      "loss": 0.4956,
      "step": 4263
    },
    {
      "epoch": 0.7271487039563438,
      "grad_norm": 0.4354765713214874,
      "learning_rate": 3.6614055280742014e-07,
      "loss": 0.4954,
      "step": 4264
    },
    {
      "epoch": 0.727319236016371,
      "grad_norm": 0.43414583802223206,
      "learning_rate": 3.6571345904796897e-07,
      "loss": 0.5044,
      "step": 4265
    },
    {
      "epoch": 0.7274897680763983,
      "grad_norm": 0.46638140082359314,
      "learning_rate": 3.652865587818725e-07,
      "loss": 0.5041,
      "step": 4266
    },
    {
      "epoch": 0.7276603001364257,
      "grad_norm": 0.38081035017967224,
      "learning_rate": 3.6485985213935983e-07,
      "loss": 0.505,
      "step": 4267
    },
    {
      "epoch": 0.727830832196453,
      "grad_norm": 0.45600566267967224,
      "learning_rate": 3.6443333925060066e-07,
      "loss": 0.4945,
      "step": 4268
    },
    {
      "epoch": 0.7280013642564802,
      "grad_norm": 0.518722653388977,
      "learning_rate": 3.64007020245706e-07,
      "loss": 0.4937,
      "step": 4269
    },
    {
      "epoch": 0.7281718963165075,
      "grad_norm": 0.33963751792907715,
      "learning_rate": 3.635808952547263e-07,
      "loss": 0.4961,
      "step": 4270
    },
    {
      "epoch": 0.7283424283765347,
      "grad_norm": 0.446537584066391,
      "learning_rate": 3.631549644076552e-07,
      "loss": 0.4942,
      "step": 4271
    },
    {
      "epoch": 0.7285129604365621,
      "grad_norm": 0.4185526967048645,
      "learning_rate": 3.6272922783442504e-07,
      "loss": 0.501,
      "step": 4272
    },
    {
      "epoch": 0.7286834924965894,
      "grad_norm": 0.3797702193260193,
      "learning_rate": 3.6230368566490883e-07,
      "loss": 0.5059,
      "step": 4273
    },
    {
      "epoch": 0.7288540245566166,
      "grad_norm": 0.43924352526664734,
      "learning_rate": 3.618783380289226e-07,
      "loss": 0.4932,
      "step": 4274
    },
    {
      "epoch": 0.7290245566166439,
      "grad_norm": 0.348660945892334,
      "learning_rate": 3.6145318505622034e-07,
      "loss": 0.5097,
      "step": 4275
    },
    {
      "epoch": 0.7291950886766713,
      "grad_norm": 0.5583161115646362,
      "learning_rate": 3.6102822687649833e-07,
      "loss": 0.5093,
      "step": 4276
    },
    {
      "epoch": 0.7293656207366985,
      "grad_norm": 0.4120829105377197,
      "learning_rate": 3.6060346361939263e-07,
      "loss": 0.5087,
      "step": 4277
    },
    {
      "epoch": 0.7295361527967258,
      "grad_norm": 0.45520544052124023,
      "learning_rate": 3.6017889541448075e-07,
      "loss": 0.4949,
      "step": 4278
    },
    {
      "epoch": 0.729706684856753,
      "grad_norm": 0.525784969329834,
      "learning_rate": 3.597545223912792e-07,
      "loss": 0.495,
      "step": 4279
    },
    {
      "epoch": 0.7298772169167803,
      "grad_norm": 0.3586786091327667,
      "learning_rate": 3.5933034467924616e-07,
      "loss": 0.5182,
      "step": 4280
    },
    {
      "epoch": 0.7300477489768077,
      "grad_norm": 0.5064526796340942,
      "learning_rate": 3.5890636240778026e-07,
      "loss": 0.4949,
      "step": 4281
    },
    {
      "epoch": 0.730218281036835,
      "grad_norm": 0.4644537568092346,
      "learning_rate": 3.5848257570621905e-07,
      "loss": 0.5062,
      "step": 4282
    },
    {
      "epoch": 0.7303888130968622,
      "grad_norm": 0.34113574028015137,
      "learning_rate": 3.5805898470384307e-07,
      "loss": 0.5025,
      "step": 4283
    },
    {
      "epoch": 0.7305593451568895,
      "grad_norm": 0.5026168823242188,
      "learning_rate": 3.5763558952987025e-07,
      "loss": 0.5077,
      "step": 4284
    },
    {
      "epoch": 0.7307298772169167,
      "grad_norm": 0.459414005279541,
      "learning_rate": 3.572123903134607e-07,
      "loss": 0.5073,
      "step": 4285
    },
    {
      "epoch": 0.7309004092769441,
      "grad_norm": 0.44578710198402405,
      "learning_rate": 3.5678938718371415e-07,
      "loss": 0.512,
      "step": 4286
    },
    {
      "epoch": 0.7310709413369714,
      "grad_norm": 0.38436904549598694,
      "learning_rate": 3.5636658026967077e-07,
      "loss": 0.516,
      "step": 4287
    },
    {
      "epoch": 0.7312414733969986,
      "grad_norm": 0.47648823261260986,
      "learning_rate": 3.5594396970031e-07,
      "loss": 0.5065,
      "step": 4288
    },
    {
      "epoch": 0.7314120054570259,
      "grad_norm": 0.4478094279766083,
      "learning_rate": 3.555215556045524e-07,
      "loss": 0.5094,
      "step": 4289
    },
    {
      "epoch": 0.7315825375170532,
      "grad_norm": 0.4391409754753113,
      "learning_rate": 3.5509933811125855e-07,
      "loss": 0.5072,
      "step": 4290
    },
    {
      "epoch": 0.7317530695770805,
      "grad_norm": 0.5401695966720581,
      "learning_rate": 3.5467731734922767e-07,
      "loss": 0.5016,
      "step": 4291
    },
    {
      "epoch": 0.7319236016371078,
      "grad_norm": 0.39789271354675293,
      "learning_rate": 3.542554934472014e-07,
      "loss": 0.5166,
      "step": 4292
    },
    {
      "epoch": 0.732094133697135,
      "grad_norm": 0.7003616094589233,
      "learning_rate": 3.538338665338589e-07,
      "loss": 0.5065,
      "step": 4293
    },
    {
      "epoch": 0.7322646657571623,
      "grad_norm": 0.4756523668766022,
      "learning_rate": 3.534124367378207e-07,
      "loss": 0.5016,
      "step": 4294
    },
    {
      "epoch": 0.7324351978171897,
      "grad_norm": 0.3915746212005615,
      "learning_rate": 3.5299120418764703e-07,
      "loss": 0.4972,
      "step": 4295
    },
    {
      "epoch": 0.732605729877217,
      "grad_norm": 0.4228286147117615,
      "learning_rate": 3.525701690118371e-07,
      "loss": 0.5064,
      "step": 4296
    },
    {
      "epoch": 0.7327762619372442,
      "grad_norm": 0.45134690403938293,
      "learning_rate": 3.521493313388307e-07,
      "loss": 0.5139,
      "step": 4297
    },
    {
      "epoch": 0.7329467939972715,
      "grad_norm": 0.5544208288192749,
      "learning_rate": 3.517286912970074e-07,
      "loss": 0.4994,
      "step": 4298
    },
    {
      "epoch": 0.7331173260572987,
      "grad_norm": 0.5639966130256653,
      "learning_rate": 3.5130824901468643e-07,
      "loss": 0.4939,
      "step": 4299
    },
    {
      "epoch": 0.7332878581173261,
      "grad_norm": 0.5455499887466431,
      "learning_rate": 3.508880046201255e-07,
      "loss": 0.5041,
      "step": 4300
    },
    {
      "epoch": 0.7334583901773534,
      "grad_norm": 0.583670437335968,
      "learning_rate": 3.504679582415245e-07,
      "loss": 0.5016,
      "step": 4301
    },
    {
      "epoch": 0.7336289222373806,
      "grad_norm": 0.45605432987213135,
      "learning_rate": 3.500481100070201e-07,
      "loss": 0.5119,
      "step": 4302
    },
    {
      "epoch": 0.7337994542974079,
      "grad_norm": 0.7099131345748901,
      "learning_rate": 3.496284600446904e-07,
      "loss": 0.4941,
      "step": 4303
    },
    {
      "epoch": 0.7339699863574352,
      "grad_norm": 0.6572691202163696,
      "learning_rate": 3.492090084825528e-07,
      "loss": 0.4841,
      "step": 4304
    },
    {
      "epoch": 0.7341405184174625,
      "grad_norm": 0.570122480392456,
      "learning_rate": 3.4878975544856287e-07,
      "loss": 0.506,
      "step": 4305
    },
    {
      "epoch": 0.7343110504774898,
      "grad_norm": 0.5914926528930664,
      "learning_rate": 3.4837070107061713e-07,
      "loss": 0.4924,
      "step": 4306
    },
    {
      "epoch": 0.734481582537517,
      "grad_norm": 0.6704850792884827,
      "learning_rate": 3.4795184547655075e-07,
      "loss": 0.5031,
      "step": 4307
    },
    {
      "epoch": 0.7346521145975443,
      "grad_norm": 0.5630759596824646,
      "learning_rate": 3.475331887941388e-07,
      "loss": 0.5025,
      "step": 4308
    },
    {
      "epoch": 0.7348226466575716,
      "grad_norm": 0.49667057394981384,
      "learning_rate": 3.471147311510944e-07,
      "loss": 0.4918,
      "step": 4309
    },
    {
      "epoch": 0.734993178717599,
      "grad_norm": 0.6504514813423157,
      "learning_rate": 3.4669647267507225e-07,
      "loss": 0.4995,
      "step": 4310
    },
    {
      "epoch": 0.7351637107776262,
      "grad_norm": 0.5181241035461426,
      "learning_rate": 3.462784134936637e-07,
      "loss": 0.503,
      "step": 4311
    },
    {
      "epoch": 0.7353342428376535,
      "grad_norm": 0.5790863633155823,
      "learning_rate": 3.458605537344008e-07,
      "loss": 0.491,
      "step": 4312
    },
    {
      "epoch": 0.7355047748976807,
      "grad_norm": 0.5233603715896606,
      "learning_rate": 3.4544289352475503e-07,
      "loss": 0.499,
      "step": 4313
    },
    {
      "epoch": 0.735675306957708,
      "grad_norm": 0.5413016080856323,
      "learning_rate": 3.450254329921357e-07,
      "loss": 0.5026,
      "step": 4314
    },
    {
      "epoch": 0.7358458390177354,
      "grad_norm": 0.6021640300750732,
      "learning_rate": 3.446081722638921e-07,
      "loss": 0.5074,
      "step": 4315
    },
    {
      "epoch": 0.7360163710777626,
      "grad_norm": 0.5931169986724854,
      "learning_rate": 3.441911114673126e-07,
      "loss": 0.4925,
      "step": 4316
    },
    {
      "epoch": 0.7361869031377899,
      "grad_norm": 0.46309608221054077,
      "learning_rate": 3.437742507296247e-07,
      "loss": 0.5053,
      "step": 4317
    },
    {
      "epoch": 0.7363574351978172,
      "grad_norm": 0.5755923986434937,
      "learning_rate": 3.433575901779938e-07,
      "loss": 0.4917,
      "step": 4318
    },
    {
      "epoch": 0.7365279672578445,
      "grad_norm": 0.4366355538368225,
      "learning_rate": 3.4294112993952534e-07,
      "loss": 0.5065,
      "step": 4319
    },
    {
      "epoch": 0.7366984993178718,
      "grad_norm": 0.589140772819519,
      "learning_rate": 3.4252487014126335e-07,
      "loss": 0.4995,
      "step": 4320
    },
    {
      "epoch": 0.736869031377899,
      "grad_norm": 0.5909987092018127,
      "learning_rate": 3.421088109101907e-07,
      "loss": 0.4967,
      "step": 4321
    },
    {
      "epoch": 0.7370395634379263,
      "grad_norm": 0.46883001923561096,
      "learning_rate": 3.4169295237322934e-07,
      "loss": 0.498,
      "step": 4322
    },
    {
      "epoch": 0.7372100954979536,
      "grad_norm": 0.41802477836608887,
      "learning_rate": 3.4127729465723894e-07,
      "loss": 0.4955,
      "step": 4323
    },
    {
      "epoch": 0.737380627557981,
      "grad_norm": 0.46541598439216614,
      "learning_rate": 3.408618378890192e-07,
      "loss": 0.5041,
      "step": 4324
    },
    {
      "epoch": 0.7375511596180082,
      "grad_norm": 0.534308135509491,
      "learning_rate": 3.4044658219530785e-07,
      "loss": 0.5052,
      "step": 4325
    },
    {
      "epoch": 0.7377216916780355,
      "grad_norm": 0.4054575562477112,
      "learning_rate": 3.4003152770278127e-07,
      "loss": 0.498,
      "step": 4326
    },
    {
      "epoch": 0.7378922237380627,
      "grad_norm": 0.5337191820144653,
      "learning_rate": 3.3961667453805526e-07,
      "loss": 0.4885,
      "step": 4327
    },
    {
      "epoch": 0.73806275579809,
      "grad_norm": 0.37807995080947876,
      "learning_rate": 3.392020228276828e-07,
      "loss": 0.4994,
      "step": 4328
    },
    {
      "epoch": 0.7382332878581174,
      "grad_norm": 0.5176721811294556,
      "learning_rate": 3.3878757269815633e-07,
      "loss": 0.502,
      "step": 4329
    },
    {
      "epoch": 0.7384038199181446,
      "grad_norm": 0.4743328094482422,
      "learning_rate": 3.3837332427590676e-07,
      "loss": 0.4955,
      "step": 4330
    },
    {
      "epoch": 0.7385743519781719,
      "grad_norm": 0.423225998878479,
      "learning_rate": 3.379592776873038e-07,
      "loss": 0.4904,
      "step": 4331
    },
    {
      "epoch": 0.7387448840381992,
      "grad_norm": 0.5088225603103638,
      "learning_rate": 3.375454330586542e-07,
      "loss": 0.4896,
      "step": 4332
    },
    {
      "epoch": 0.7389154160982264,
      "grad_norm": 0.3379969298839569,
      "learning_rate": 3.371317905162046e-07,
      "loss": 0.5009,
      "step": 4333
    },
    {
      "epoch": 0.7390859481582538,
      "grad_norm": 0.445017009973526,
      "learning_rate": 3.367183501861395e-07,
      "loss": 0.4962,
      "step": 4334
    },
    {
      "epoch": 0.739256480218281,
      "grad_norm": 0.31372275948524475,
      "learning_rate": 3.3630511219458097e-07,
      "loss": 0.5007,
      "step": 4335
    },
    {
      "epoch": 0.7394270122783083,
      "grad_norm": 0.4493860900402069,
      "learning_rate": 3.3589207666759117e-07,
      "loss": 0.5125,
      "step": 4336
    },
    {
      "epoch": 0.7395975443383356,
      "grad_norm": 0.4906957149505615,
      "learning_rate": 3.354792437311684e-07,
      "loss": 0.5034,
      "step": 4337
    },
    {
      "epoch": 0.7397680763983628,
      "grad_norm": 0.4305751621723175,
      "learning_rate": 3.350666135112504e-07,
      "loss": 0.5023,
      "step": 4338
    },
    {
      "epoch": 0.7399386084583902,
      "grad_norm": 0.43462157249450684,
      "learning_rate": 3.346541861337127e-07,
      "loss": 0.5047,
      "step": 4339
    },
    {
      "epoch": 0.7401091405184175,
      "grad_norm": 0.4501499831676483,
      "learning_rate": 3.3424196172436953e-07,
      "loss": 0.5042,
      "step": 4340
    },
    {
      "epoch": 0.7402796725784447,
      "grad_norm": 0.4740625321865082,
      "learning_rate": 3.3382994040897203e-07,
      "loss": 0.5118,
      "step": 4341
    },
    {
      "epoch": 0.740450204638472,
      "grad_norm": 0.4675706923007965,
      "learning_rate": 3.334181223132103e-07,
      "loss": 0.5029,
      "step": 4342
    },
    {
      "epoch": 0.7406207366984994,
      "grad_norm": 0.3829689025878906,
      "learning_rate": 3.330065075627126e-07,
      "loss": 0.5014,
      "step": 4343
    },
    {
      "epoch": 0.7407912687585266,
      "grad_norm": 0.4442577362060547,
      "learning_rate": 3.325950962830437e-07,
      "loss": 0.4958,
      "step": 4344
    },
    {
      "epoch": 0.7409618008185539,
      "grad_norm": 0.47646093368530273,
      "learning_rate": 3.3218388859970877e-07,
      "loss": 0.5022,
      "step": 4345
    },
    {
      "epoch": 0.7411323328785812,
      "grad_norm": 0.3984088599681854,
      "learning_rate": 3.317728846381484e-07,
      "loss": 0.4956,
      "step": 4346
    },
    {
      "epoch": 0.7413028649386084,
      "grad_norm": 0.5109058022499084,
      "learning_rate": 3.3136208452374256e-07,
      "loss": 0.497,
      "step": 4347
    },
    {
      "epoch": 0.7414733969986358,
      "grad_norm": 0.41433700919151306,
      "learning_rate": 3.309514883818084e-07,
      "loss": 0.5066,
      "step": 4348
    },
    {
      "epoch": 0.741643929058663,
      "grad_norm": 0.4546804428100586,
      "learning_rate": 3.305410963376015e-07,
      "loss": 0.5003,
      "step": 4349
    },
    {
      "epoch": 0.7418144611186903,
      "grad_norm": 0.43137767910957336,
      "learning_rate": 3.301309085163139e-07,
      "loss": 0.4969,
      "step": 4350
    },
    {
      "epoch": 0.7419849931787176,
      "grad_norm": 0.44788989424705505,
      "learning_rate": 3.2972092504307664e-07,
      "loss": 0.5019,
      "step": 4351
    },
    {
      "epoch": 0.7421555252387448,
      "grad_norm": 0.5021728277206421,
      "learning_rate": 3.29311146042958e-07,
      "loss": 0.4981,
      "step": 4352
    },
    {
      "epoch": 0.7423260572987722,
      "grad_norm": 0.3996676504611969,
      "learning_rate": 3.289015716409632e-07,
      "loss": 0.4967,
      "step": 4353
    },
    {
      "epoch": 0.7424965893587995,
      "grad_norm": 0.5152726769447327,
      "learning_rate": 3.284922019620366e-07,
      "loss": 0.496,
      "step": 4354
    },
    {
      "epoch": 0.7426671214188267,
      "grad_norm": 0.5459908246994019,
      "learning_rate": 3.280830371310583e-07,
      "loss": 0.5029,
      "step": 4355
    },
    {
      "epoch": 0.742837653478854,
      "grad_norm": 0.3832405209541321,
      "learning_rate": 3.276740772728472e-07,
      "loss": 0.492,
      "step": 4356
    },
    {
      "epoch": 0.7430081855388813,
      "grad_norm": 0.44288480281829834,
      "learning_rate": 3.272653225121592e-07,
      "loss": 0.4985,
      "step": 4357
    },
    {
      "epoch": 0.7431787175989086,
      "grad_norm": 0.491085022687912,
      "learning_rate": 3.268567729736879e-07,
      "loss": 0.5004,
      "step": 4358
    },
    {
      "epoch": 0.7433492496589359,
      "grad_norm": 0.4536385238170624,
      "learning_rate": 3.2644842878206343e-07,
      "loss": 0.4981,
      "step": 4359
    },
    {
      "epoch": 0.7435197817189632,
      "grad_norm": 0.4160607159137726,
      "learning_rate": 3.260402900618544e-07,
      "loss": 0.4974,
      "step": 4360
    },
    {
      "epoch": 0.7436903137789904,
      "grad_norm": 0.4935556948184967,
      "learning_rate": 3.256323569375665e-07,
      "loss": 0.5009,
      "step": 4361
    },
    {
      "epoch": 0.7438608458390177,
      "grad_norm": 0.46699386835098267,
      "learning_rate": 3.2522462953364135e-07,
      "loss": 0.4939,
      "step": 4362
    },
    {
      "epoch": 0.744031377899045,
      "grad_norm": 0.45156747102737427,
      "learning_rate": 3.2481710797446043e-07,
      "loss": 0.4978,
      "step": 4363
    },
    {
      "epoch": 0.7442019099590723,
      "grad_norm": 0.455791711807251,
      "learning_rate": 3.244097923843398e-07,
      "loss": 0.4988,
      "step": 4364
    },
    {
      "epoch": 0.7443724420190996,
      "grad_norm": 0.4738144278526306,
      "learning_rate": 3.2400268288753423e-07,
      "loss": 0.5138,
      "step": 4365
    },
    {
      "epoch": 0.7445429740791268,
      "grad_norm": 0.4773343801498413,
      "learning_rate": 3.2359577960823557e-07,
      "loss": 0.5,
      "step": 4366
    },
    {
      "epoch": 0.7447135061391542,
      "grad_norm": 0.4243999421596527,
      "learning_rate": 3.231890826705715e-07,
      "loss": 0.4993,
      "step": 4367
    },
    {
      "epoch": 0.7448840381991815,
      "grad_norm": 0.3990984559059143,
      "learning_rate": 3.227825921986082e-07,
      "loss": 0.4981,
      "step": 4368
    },
    {
      "epoch": 0.7450545702592087,
      "grad_norm": 0.48051950335502625,
      "learning_rate": 3.223763083163482e-07,
      "loss": 0.4986,
      "step": 4369
    },
    {
      "epoch": 0.745225102319236,
      "grad_norm": 0.4449363350868225,
      "learning_rate": 3.2197023114773145e-07,
      "loss": 0.5012,
      "step": 4370
    },
    {
      "epoch": 0.7453956343792633,
      "grad_norm": 0.4994031488895416,
      "learning_rate": 3.215643608166336e-07,
      "loss": 0.4919,
      "step": 4371
    },
    {
      "epoch": 0.7455661664392906,
      "grad_norm": 0.37948834896087646,
      "learning_rate": 3.2115869744686937e-07,
      "loss": 0.489,
      "step": 4372
    },
    {
      "epoch": 0.7457366984993179,
      "grad_norm": 0.45996761322021484,
      "learning_rate": 3.2075324116218835e-07,
      "loss": 0.5035,
      "step": 4373
    },
    {
      "epoch": 0.7459072305593452,
      "grad_norm": 0.4702564775943756,
      "learning_rate": 3.2034799208627724e-07,
      "loss": 0.4908,
      "step": 4374
    },
    {
      "epoch": 0.7460777626193724,
      "grad_norm": 0.5308040976524353,
      "learning_rate": 3.1994295034276104e-07,
      "loss": 0.5055,
      "step": 4375
    },
    {
      "epoch": 0.7462482946793997,
      "grad_norm": 0.4410310387611389,
      "learning_rate": 3.195381160551997e-07,
      "loss": 0.4966,
      "step": 4376
    },
    {
      "epoch": 0.746418826739427,
      "grad_norm": 0.3720689117908478,
      "learning_rate": 3.191334893470908e-07,
      "loss": 0.5018,
      "step": 4377
    },
    {
      "epoch": 0.7465893587994543,
      "grad_norm": 0.5346623063087463,
      "learning_rate": 3.1872907034186854e-07,
      "loss": 0.4962,
      "step": 4378
    },
    {
      "epoch": 0.7467598908594816,
      "grad_norm": 0.6136433482170105,
      "learning_rate": 3.183248591629038e-07,
      "loss": 0.4885,
      "step": 4379
    },
    {
      "epoch": 0.7469304229195088,
      "grad_norm": 0.3642314672470093,
      "learning_rate": 3.179208559335031e-07,
      "loss": 0.5014,
      "step": 4380
    },
    {
      "epoch": 0.7471009549795361,
      "grad_norm": 0.5996096730232239,
      "learning_rate": 3.175170607769116e-07,
      "loss": 0.5006,
      "step": 4381
    },
    {
      "epoch": 0.7472714870395635,
      "grad_norm": 0.5387561321258545,
      "learning_rate": 3.17113473816309e-07,
      "loss": 0.5005,
      "step": 4382
    },
    {
      "epoch": 0.7474420190995907,
      "grad_norm": 0.4318586587905884,
      "learning_rate": 3.1671009517481157e-07,
      "loss": 0.5033,
      "step": 4383
    },
    {
      "epoch": 0.747612551159618,
      "grad_norm": 0.5770013332366943,
      "learning_rate": 3.16306924975474e-07,
      "loss": 0.4947,
      "step": 4384
    },
    {
      "epoch": 0.7477830832196453,
      "grad_norm": 0.43154647946357727,
      "learning_rate": 3.159039633412851e-07,
      "loss": 0.4985,
      "step": 4385
    },
    {
      "epoch": 0.7479536152796725,
      "grad_norm": 0.4739813208580017,
      "learning_rate": 3.1550121039517124e-07,
      "loss": 0.4918,
      "step": 4386
    },
    {
      "epoch": 0.7481241473396999,
      "grad_norm": 0.40933310985565186,
      "learning_rate": 3.15098666259995e-07,
      "loss": 0.498,
      "step": 4387
    },
    {
      "epoch": 0.7482946793997272,
      "grad_norm": 0.477927565574646,
      "learning_rate": 3.146963310585555e-07,
      "loss": 0.4966,
      "step": 4388
    },
    {
      "epoch": 0.7484652114597544,
      "grad_norm": 0.44899311661720276,
      "learning_rate": 3.14294204913587e-07,
      "loss": 0.5076,
      "step": 4389
    },
    {
      "epoch": 0.7486357435197817,
      "grad_norm": 0.5270994305610657,
      "learning_rate": 3.138922879477613e-07,
      "loss": 0.5012,
      "step": 4390
    },
    {
      "epoch": 0.748806275579809,
      "grad_norm": 0.5749548077583313,
      "learning_rate": 3.1349058028368565e-07,
      "loss": 0.4931,
      "step": 4391
    },
    {
      "epoch": 0.7489768076398363,
      "grad_norm": 0.5705610513687134,
      "learning_rate": 3.130890820439038e-07,
      "loss": 0.4904,
      "step": 4392
    },
    {
      "epoch": 0.7491473396998636,
      "grad_norm": 0.40459010004997253,
      "learning_rate": 3.1268779335089575e-07,
      "loss": 0.4945,
      "step": 4393
    },
    {
      "epoch": 0.7493178717598908,
      "grad_norm": 0.5784091949462891,
      "learning_rate": 3.1228671432707663e-07,
      "loss": 0.5037,
      "step": 4394
    },
    {
      "epoch": 0.7494884038199181,
      "grad_norm": 0.5292043685913086,
      "learning_rate": 3.118858450947987e-07,
      "loss": 0.497,
      "step": 4395
    },
    {
      "epoch": 0.7496589358799455,
      "grad_norm": 0.47194743156433105,
      "learning_rate": 3.114851857763497e-07,
      "loss": 0.5048,
      "step": 4396
    },
    {
      "epoch": 0.7498294679399727,
      "grad_norm": 0.4495631456375122,
      "learning_rate": 3.110847364939538e-07,
      "loss": 0.498,
      "step": 4397
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5370325446128845,
      "learning_rate": 3.1068449736977015e-07,
      "loss": 0.4877,
      "step": 4398
    },
    {
      "epoch": 0.7501705320600273,
      "grad_norm": 0.6217759847640991,
      "learning_rate": 3.102844685258946e-07,
      "loss": 0.4871,
      "step": 4399
    },
    {
      "epoch": 0.7503410641200545,
      "grad_norm": 0.40713444352149963,
      "learning_rate": 3.0988465008435855e-07,
      "loss": 0.5077,
      "step": 4400
    },
    {
      "epoch": 0.7505115961800819,
      "grad_norm": 0.548819363117218,
      "learning_rate": 3.094850421671295e-07,
      "loss": 0.4979,
      "step": 4401
    },
    {
      "epoch": 0.7506821282401092,
      "grad_norm": 0.4234558939933777,
      "learning_rate": 3.0908564489611083e-07,
      "loss": 0.4944,
      "step": 4402
    },
    {
      "epoch": 0.7508526603001364,
      "grad_norm": 0.49774685502052307,
      "learning_rate": 3.086864583931405e-07,
      "loss": 0.4984,
      "step": 4403
    },
    {
      "epoch": 0.7510231923601637,
      "grad_norm": 0.5342044830322266,
      "learning_rate": 3.0828748277999356e-07,
      "loss": 0.5056,
      "step": 4404
    },
    {
      "epoch": 0.751193724420191,
      "grad_norm": 0.4261223077774048,
      "learning_rate": 3.078887181783803e-07,
      "loss": 0.493,
      "step": 4405
    },
    {
      "epoch": 0.7513642564802183,
      "grad_norm": 0.5373156070709229,
      "learning_rate": 3.074901647099461e-07,
      "loss": 0.4956,
      "step": 4406
    },
    {
      "epoch": 0.7515347885402456,
      "grad_norm": 0.45177051424980164,
      "learning_rate": 3.070918224962726e-07,
      "loss": 0.5044,
      "step": 4407
    },
    {
      "epoch": 0.7517053206002728,
      "grad_norm": 0.5984803438186646,
      "learning_rate": 3.0669369165887675e-07,
      "loss": 0.5015,
      "step": 4408
    },
    {
      "epoch": 0.7518758526603001,
      "grad_norm": 0.4983421862125397,
      "learning_rate": 3.062957723192111e-07,
      "loss": 0.5058,
      "step": 4409
    },
    {
      "epoch": 0.7520463847203275,
      "grad_norm": 0.49384385347366333,
      "learning_rate": 3.058980645986634e-07,
      "loss": 0.5034,
      "step": 4410
    },
    {
      "epoch": 0.7522169167803547,
      "grad_norm": 0.6117486357688904,
      "learning_rate": 3.0550056861855764e-07,
      "loss": 0.4932,
      "step": 4411
    },
    {
      "epoch": 0.752387448840382,
      "grad_norm": 0.4372576177120209,
      "learning_rate": 3.0510328450015173e-07,
      "loss": 0.4963,
      "step": 4412
    },
    {
      "epoch": 0.7525579809004093,
      "grad_norm": 0.4864978492259979,
      "learning_rate": 3.047062123646404e-07,
      "loss": 0.4874,
      "step": 4413
    },
    {
      "epoch": 0.7527285129604365,
      "grad_norm": 0.45403429865837097,
      "learning_rate": 3.043093523331533e-07,
      "loss": 0.508,
      "step": 4414
    },
    {
      "epoch": 0.7528990450204639,
      "grad_norm": 0.47508782148361206,
      "learning_rate": 3.0391270452675437e-07,
      "loss": 0.4959,
      "step": 4415
    },
    {
      "epoch": 0.7530695770804912,
      "grad_norm": 0.40006983280181885,
      "learning_rate": 3.0351626906644505e-07,
      "loss": 0.493,
      "step": 4416
    },
    {
      "epoch": 0.7532401091405184,
      "grad_norm": 0.5417851209640503,
      "learning_rate": 3.031200460731595e-07,
      "loss": 0.4966,
      "step": 4417
    },
    {
      "epoch": 0.7534106412005457,
      "grad_norm": 0.507758617401123,
      "learning_rate": 3.0272403566776856e-07,
      "loss": 0.4955,
      "step": 4418
    },
    {
      "epoch": 0.753581173260573,
      "grad_norm": 0.3795713186264038,
      "learning_rate": 3.0232823797107795e-07,
      "loss": 0.5046,
      "step": 4419
    },
    {
      "epoch": 0.7537517053206003,
      "grad_norm": 0.4745026230812073,
      "learning_rate": 3.019326531038287e-07,
      "loss": 0.5004,
      "step": 4420
    },
    {
      "epoch": 0.7539222373806276,
      "grad_norm": 0.485241174697876,
      "learning_rate": 3.015372811866958e-07,
      "loss": 0.4875,
      "step": 4421
    },
    {
      "epoch": 0.7540927694406548,
      "grad_norm": 0.6154580116271973,
      "learning_rate": 3.0114212234029065e-07,
      "loss": 0.4991,
      "step": 4422
    },
    {
      "epoch": 0.7542633015006821,
      "grad_norm": 1.3015596866607666,
      "learning_rate": 3.007471766851593e-07,
      "loss": 0.4945,
      "step": 4423
    },
    {
      "epoch": 0.7544338335607094,
      "grad_norm": 0.4312533438205719,
      "learning_rate": 3.003524443417817e-07,
      "loss": 0.4958,
      "step": 4424
    },
    {
      "epoch": 0.7546043656207367,
      "grad_norm": 0.5360495448112488,
      "learning_rate": 2.999579254305748e-07,
      "loss": 0.4995,
      "step": 4425
    },
    {
      "epoch": 0.754774897680764,
      "grad_norm": 0.382801353931427,
      "learning_rate": 2.9956362007188824e-07,
      "loss": 0.4952,
      "step": 4426
    },
    {
      "epoch": 0.7549454297407913,
      "grad_norm": 0.4729914665222168,
      "learning_rate": 2.9916952838600793e-07,
      "loss": 0.5043,
      "step": 4427
    },
    {
      "epoch": 0.7551159618008185,
      "grad_norm": 0.4124811887741089,
      "learning_rate": 2.987756504931543e-07,
      "loss": 0.4922,
      "step": 4428
    },
    {
      "epoch": 0.7552864938608458,
      "grad_norm": 0.4720993936061859,
      "learning_rate": 2.9838198651348207e-07,
      "loss": 0.5099,
      "step": 4429
    },
    {
      "epoch": 0.7554570259208732,
      "grad_norm": 0.48451411724090576,
      "learning_rate": 2.9798853656708127e-07,
      "loss": 0.4908,
      "step": 4430
    },
    {
      "epoch": 0.7556275579809004,
      "grad_norm": 0.6371461153030396,
      "learning_rate": 2.9759530077397637e-07,
      "loss": 0.5009,
      "step": 4431
    },
    {
      "epoch": 0.7557980900409277,
      "grad_norm": 0.48880821466445923,
      "learning_rate": 2.9720227925412693e-07,
      "loss": 0.498,
      "step": 4432
    },
    {
      "epoch": 0.755968622100955,
      "grad_norm": 0.4787331819534302,
      "learning_rate": 2.9680947212742596e-07,
      "loss": 0.5109,
      "step": 4433
    },
    {
      "epoch": 0.7561391541609823,
      "grad_norm": 0.5188673734664917,
      "learning_rate": 2.9641687951370305e-07,
      "loss": 0.4982,
      "step": 4434
    },
    {
      "epoch": 0.7563096862210096,
      "grad_norm": 0.411820650100708,
      "learning_rate": 2.9602450153272035e-07,
      "loss": 0.4994,
      "step": 4435
    },
    {
      "epoch": 0.7564802182810368,
      "grad_norm": 0.5018237233161926,
      "learning_rate": 2.9563233830417564e-07,
      "loss": 0.5034,
      "step": 4436
    },
    {
      "epoch": 0.7566507503410641,
      "grad_norm": 0.42866119742393494,
      "learning_rate": 2.9524038994770114e-07,
      "loss": 0.5043,
      "step": 4437
    },
    {
      "epoch": 0.7568212824010914,
      "grad_norm": 0.44395798444747925,
      "learning_rate": 2.948486565828629e-07,
      "loss": 0.5068,
      "step": 4438
    },
    {
      "epoch": 0.7569918144611187,
      "grad_norm": 0.4456336498260498,
      "learning_rate": 2.9445713832916203e-07,
      "loss": 0.5021,
      "step": 4439
    },
    {
      "epoch": 0.757162346521146,
      "grad_norm": 0.3952493369579315,
      "learning_rate": 2.9406583530603376e-07,
      "loss": 0.5005,
      "step": 4440
    },
    {
      "epoch": 0.7573328785811733,
      "grad_norm": 0.5085216760635376,
      "learning_rate": 2.93674747632848e-07,
      "loss": 0.5109,
      "step": 4441
    },
    {
      "epoch": 0.7575034106412005,
      "grad_norm": 0.38747841119766235,
      "learning_rate": 2.932838754289078e-07,
      "loss": 0.4886,
      "step": 4442
    },
    {
      "epoch": 0.7576739427012278,
      "grad_norm": 0.43506336212158203,
      "learning_rate": 2.928932188134526e-07,
      "loss": 0.5031,
      "step": 4443
    },
    {
      "epoch": 0.7578444747612552,
      "grad_norm": 0.38857099413871765,
      "learning_rate": 2.925027779056541e-07,
      "loss": 0.4937,
      "step": 4444
    },
    {
      "epoch": 0.7580150068212824,
      "grad_norm": 0.5130706429481506,
      "learning_rate": 2.9211255282461837e-07,
      "loss": 0.4845,
      "step": 4445
    },
    {
      "epoch": 0.7581855388813097,
      "grad_norm": 0.5911456942558289,
      "learning_rate": 2.917225436893874e-07,
      "loss": 0.4997,
      "step": 4446
    },
    {
      "epoch": 0.758356070941337,
      "grad_norm": 0.4976506531238556,
      "learning_rate": 2.913327506189351e-07,
      "loss": 0.4941,
      "step": 4447
    },
    {
      "epoch": 0.7585266030013642,
      "grad_norm": 0.4999774992465973,
      "learning_rate": 2.909431737321709e-07,
      "loss": 0.5136,
      "step": 4448
    },
    {
      "epoch": 0.7586971350613916,
      "grad_norm": 0.38214829564094543,
      "learning_rate": 2.9055381314793767e-07,
      "loss": 0.4967,
      "step": 4449
    },
    {
      "epoch": 0.7588676671214188,
      "grad_norm": 0.6079405546188354,
      "learning_rate": 2.9016466898501286e-07,
      "loss": 0.4927,
      "step": 4450
    },
    {
      "epoch": 0.7590381991814461,
      "grad_norm": 0.4319635331630707,
      "learning_rate": 2.8977574136210696e-07,
      "loss": 0.4958,
      "step": 4451
    },
    {
      "epoch": 0.7592087312414734,
      "grad_norm": 0.5009601712226868,
      "learning_rate": 2.893870303978651e-07,
      "loss": 0.4968,
      "step": 4452
    },
    {
      "epoch": 0.7593792633015006,
      "grad_norm": 0.5611957311630249,
      "learning_rate": 2.889985362108666e-07,
      "loss": 0.4899,
      "step": 4453
    },
    {
      "epoch": 0.759549795361528,
      "grad_norm": 0.391587495803833,
      "learning_rate": 2.8861025891962317e-07,
      "loss": 0.4914,
      "step": 4454
    },
    {
      "epoch": 0.7597203274215553,
      "grad_norm": 0.5809473991394043,
      "learning_rate": 2.8822219864258275e-07,
      "loss": 0.4972,
      "step": 4455
    },
    {
      "epoch": 0.7598908594815825,
      "grad_norm": 0.5915910601615906,
      "learning_rate": 2.878343554981247e-07,
      "loss": 0.5051,
      "step": 4456
    },
    {
      "epoch": 0.7600613915416098,
      "grad_norm": 0.47726011276245117,
      "learning_rate": 2.8744672960456346e-07,
      "loss": 0.4919,
      "step": 4457
    },
    {
      "epoch": 0.7602319236016372,
      "grad_norm": 0.4725464880466461,
      "learning_rate": 2.87059321080147e-07,
      "loss": 0.4915,
      "step": 4458
    },
    {
      "epoch": 0.7604024556616644,
      "grad_norm": 0.508224606513977,
      "learning_rate": 2.866721300430572e-07,
      "loss": 0.498,
      "step": 4459
    },
    {
      "epoch": 0.7605729877216917,
      "grad_norm": 0.4273339509963989,
      "learning_rate": 2.8628515661140856e-07,
      "loss": 0.4995,
      "step": 4460
    },
    {
      "epoch": 0.760743519781719,
      "grad_norm": 0.3837960660457611,
      "learning_rate": 2.858984009032503e-07,
      "loss": 0.4937,
      "step": 4461
    },
    {
      "epoch": 0.7609140518417462,
      "grad_norm": 0.44394344091415405,
      "learning_rate": 2.855118630365652e-07,
      "loss": 0.4942,
      "step": 4462
    },
    {
      "epoch": 0.7610845839017736,
      "grad_norm": 0.5421721935272217,
      "learning_rate": 2.851255431292681e-07,
      "loss": 0.4907,
      "step": 4463
    },
    {
      "epoch": 0.7612551159618008,
      "grad_norm": 0.5385862588882446,
      "learning_rate": 2.8473944129920987e-07,
      "loss": 0.4976,
      "step": 4464
    },
    {
      "epoch": 0.7614256480218281,
      "grad_norm": 0.4513562023639679,
      "learning_rate": 2.843535576641725e-07,
      "loss": 0.4957,
      "step": 4465
    },
    {
      "epoch": 0.7615961800818554,
      "grad_norm": 0.6170771718025208,
      "learning_rate": 2.839678923418726e-07,
      "loss": 0.498,
      "step": 4466
    },
    {
      "epoch": 0.7617667121418826,
      "grad_norm": 0.3476124703884125,
      "learning_rate": 2.835824454499604e-07,
      "loss": 0.4915,
      "step": 4467
    },
    {
      "epoch": 0.76193724420191,
      "grad_norm": 0.6450405716896057,
      "learning_rate": 2.8319721710601847e-07,
      "loss": 0.4961,
      "step": 4468
    },
    {
      "epoch": 0.7621077762619373,
      "grad_norm": 0.4782792925834656,
      "learning_rate": 2.8281220742756337e-07,
      "loss": 0.4848,
      "step": 4469
    },
    {
      "epoch": 0.7622783083219645,
      "grad_norm": 0.49631768465042114,
      "learning_rate": 2.8242741653204513e-07,
      "loss": 0.4936,
      "step": 4470
    },
    {
      "epoch": 0.7624488403819918,
      "grad_norm": 0.4693368375301361,
      "learning_rate": 2.8204284453684683e-07,
      "loss": 0.5045,
      "step": 4471
    },
    {
      "epoch": 0.762619372442019,
      "grad_norm": 0.4708783030509949,
      "learning_rate": 2.816584915592842e-07,
      "loss": 0.4899,
      "step": 4472
    },
    {
      "epoch": 0.7627899045020464,
      "grad_norm": 0.5898051261901855,
      "learning_rate": 2.812743577166075e-07,
      "loss": 0.5111,
      "step": 4473
    },
    {
      "epoch": 0.7629604365620737,
      "grad_norm": 0.4492618143558502,
      "learning_rate": 2.808904431259986e-07,
      "loss": 0.4966,
      "step": 4474
    },
    {
      "epoch": 0.763130968622101,
      "grad_norm": 0.5589696764945984,
      "learning_rate": 2.8050674790457356e-07,
      "loss": 0.5017,
      "step": 4475
    },
    {
      "epoch": 0.7633015006821282,
      "grad_norm": 0.5096504092216492,
      "learning_rate": 2.801232721693814e-07,
      "loss": 0.5044,
      "step": 4476
    },
    {
      "epoch": 0.7634720327421555,
      "grad_norm": 0.4462493658065796,
      "learning_rate": 2.7974001603740343e-07,
      "loss": 0.5001,
      "step": 4477
    },
    {
      "epoch": 0.7636425648021828,
      "grad_norm": 0.45079052448272705,
      "learning_rate": 2.793569796255548e-07,
      "loss": 0.5068,
      "step": 4478
    },
    {
      "epoch": 0.7638130968622101,
      "grad_norm": 0.44240808486938477,
      "learning_rate": 2.789741630506833e-07,
      "loss": 0.4968,
      "step": 4479
    },
    {
      "epoch": 0.7639836289222374,
      "grad_norm": 0.4910196363925934,
      "learning_rate": 2.7859156642956955e-07,
      "loss": 0.4889,
      "step": 4480
    },
    {
      "epoch": 0.7641541609822646,
      "grad_norm": 0.332023024559021,
      "learning_rate": 2.7820918987892747e-07,
      "loss": 0.4937,
      "step": 4481
    },
    {
      "epoch": 0.764324693042292,
      "grad_norm": 0.5296370387077332,
      "learning_rate": 2.7782703351540364e-07,
      "loss": 0.4893,
      "step": 4482
    },
    {
      "epoch": 0.7644952251023193,
      "grad_norm": 0.4386923313140869,
      "learning_rate": 2.774450974555769e-07,
      "loss": 0.502,
      "step": 4483
    },
    {
      "epoch": 0.7646657571623465,
      "grad_norm": 0.430686354637146,
      "learning_rate": 2.7706338181595986e-07,
      "loss": 0.5086,
      "step": 4484
    },
    {
      "epoch": 0.7648362892223738,
      "grad_norm": 0.5587331652641296,
      "learning_rate": 2.766818867129976e-07,
      "loss": 0.5095,
      "step": 4485
    },
    {
      "epoch": 0.765006821282401,
      "grad_norm": 0.5019792914390564,
      "learning_rate": 2.7630061226306715e-07,
      "loss": 0.4989,
      "step": 4486
    },
    {
      "epoch": 0.7651773533424284,
      "grad_norm": 0.4718862771987915,
      "learning_rate": 2.7591955858247913e-07,
      "loss": 0.4937,
      "step": 4487
    },
    {
      "epoch": 0.7653478854024557,
      "grad_norm": 0.5303227305412292,
      "learning_rate": 2.7553872578747644e-07,
      "loss": 0.4935,
      "step": 4488
    },
    {
      "epoch": 0.765518417462483,
      "grad_norm": 0.5563722252845764,
      "learning_rate": 2.751581139942349e-07,
      "loss": 0.5052,
      "step": 4489
    },
    {
      "epoch": 0.7656889495225102,
      "grad_norm": 0.44941800832748413,
      "learning_rate": 2.747777233188628e-07,
      "loss": 0.4982,
      "step": 4490
    },
    {
      "epoch": 0.7658594815825375,
      "grad_norm": 0.4935671091079712,
      "learning_rate": 2.7439755387740023e-07,
      "loss": 0.4948,
      "step": 4491
    },
    {
      "epoch": 0.7660300136425648,
      "grad_norm": 0.3984651565551758,
      "learning_rate": 2.7401760578582085e-07,
      "loss": 0.497,
      "step": 4492
    },
    {
      "epoch": 0.7662005457025921,
      "grad_norm": 0.6637492775917053,
      "learning_rate": 2.736378791600304e-07,
      "loss": 0.5156,
      "step": 4493
    },
    {
      "epoch": 0.7663710777626194,
      "grad_norm": 0.6302986145019531,
      "learning_rate": 2.732583741158671e-07,
      "loss": 0.4999,
      "step": 4494
    },
    {
      "epoch": 0.7665416098226466,
      "grad_norm": 0.3748490512371063,
      "learning_rate": 2.728790907691012e-07,
      "loss": 0.5024,
      "step": 4495
    },
    {
      "epoch": 0.7667121418826739,
      "grad_norm": 0.6701140999794006,
      "learning_rate": 2.7250002923543576e-07,
      "loss": 0.4969,
      "step": 4496
    },
    {
      "epoch": 0.7668826739427013,
      "grad_norm": 0.6671623587608337,
      "learning_rate": 2.7212118963050597e-07,
      "loss": 0.4965,
      "step": 4497
    },
    {
      "epoch": 0.7670532060027285,
      "grad_norm": 0.38242241740226746,
      "learning_rate": 2.7174257206987946e-07,
      "loss": 0.5043,
      "step": 4498
    },
    {
      "epoch": 0.7672237380627558,
      "grad_norm": 0.6067281365394592,
      "learning_rate": 2.7136417666905647e-07,
      "loss": 0.4997,
      "step": 4499
    },
    {
      "epoch": 0.767394270122783,
      "grad_norm": 0.5838785171508789,
      "learning_rate": 2.709860035434683e-07,
      "loss": 0.5098,
      "step": 4500
    },
    {
      "epoch": 0.7675648021828103,
      "grad_norm": 0.4934817850589752,
      "learning_rate": 2.706080528084795e-07,
      "loss": 0.5003,
      "step": 4501
    },
    {
      "epoch": 0.7677353342428377,
      "grad_norm": 0.4918252229690552,
      "learning_rate": 2.702303245793865e-07,
      "loss": 0.4968,
      "step": 4502
    },
    {
      "epoch": 0.767905866302865,
      "grad_norm": 0.490156352519989,
      "learning_rate": 2.6985281897141815e-07,
      "loss": 0.4829,
      "step": 4503
    },
    {
      "epoch": 0.7680763983628922,
      "grad_norm": 0.5496761798858643,
      "learning_rate": 2.694755360997342e-07,
      "loss": 0.4857,
      "step": 4504
    },
    {
      "epoch": 0.7682469304229195,
      "grad_norm": 0.4437698423862457,
      "learning_rate": 2.690984760794284e-07,
      "loss": 0.5017,
      "step": 4505
    },
    {
      "epoch": 0.7684174624829468,
      "grad_norm": 0.5472577214241028,
      "learning_rate": 2.6872163902552497e-07,
      "loss": 0.4911,
      "step": 4506
    },
    {
      "epoch": 0.7685879945429741,
      "grad_norm": 0.38096439838409424,
      "learning_rate": 2.683450250529798e-07,
      "loss": 0.4991,
      "step": 4507
    },
    {
      "epoch": 0.7687585266030014,
      "grad_norm": 0.5786945819854736,
      "learning_rate": 2.679686342766829e-07,
      "loss": 0.5001,
      "step": 4508
    },
    {
      "epoch": 0.7689290586630286,
      "grad_norm": 0.38383108377456665,
      "learning_rate": 2.675924668114537e-07,
      "loss": 0.5044,
      "step": 4509
    },
    {
      "epoch": 0.7690995907230559,
      "grad_norm": 0.5291785001754761,
      "learning_rate": 2.6721652277204503e-07,
      "loss": 0.5038,
      "step": 4510
    },
    {
      "epoch": 0.7692701227830833,
      "grad_norm": 0.4897950291633606,
      "learning_rate": 2.6684080227314115e-07,
      "loss": 0.4946,
      "step": 4511
    },
    {
      "epoch": 0.7694406548431105,
      "grad_norm": 0.45776981115341187,
      "learning_rate": 2.6646530542935835e-07,
      "loss": 0.4994,
      "step": 4512
    },
    {
      "epoch": 0.7696111869031378,
      "grad_norm": 0.41584646701812744,
      "learning_rate": 2.660900323552436e-07,
      "loss": 0.5043,
      "step": 4513
    },
    {
      "epoch": 0.769781718963165,
      "grad_norm": 0.38639041781425476,
      "learning_rate": 2.6571498316527755e-07,
      "loss": 0.4968,
      "step": 4514
    },
    {
      "epoch": 0.7699522510231923,
      "grad_norm": 0.4452049434185028,
      "learning_rate": 2.6534015797387105e-07,
      "loss": 0.5143,
      "step": 4515
    },
    {
      "epoch": 0.7701227830832197,
      "grad_norm": 0.4047050178050995,
      "learning_rate": 2.6496555689536634e-07,
      "loss": 0.4849,
      "step": 4516
    },
    {
      "epoch": 0.770293315143247,
      "grad_norm": 0.46245694160461426,
      "learning_rate": 2.645911800440392e-07,
      "loss": 0.5078,
      "step": 4517
    },
    {
      "epoch": 0.7704638472032742,
      "grad_norm": 0.49868836998939514,
      "learning_rate": 2.6421702753409475e-07,
      "loss": 0.4908,
      "step": 4518
    },
    {
      "epoch": 0.7706343792633015,
      "grad_norm": 0.44043809175491333,
      "learning_rate": 2.6384309947967124e-07,
      "loss": 0.498,
      "step": 4519
    },
    {
      "epoch": 0.7708049113233287,
      "grad_norm": 0.5068042874336243,
      "learning_rate": 2.634693959948378e-07,
      "loss": 0.5017,
      "step": 4520
    },
    {
      "epoch": 0.7709754433833561,
      "grad_norm": 0.514275848865509,
      "learning_rate": 2.6309591719359566e-07,
      "loss": 0.4937,
      "step": 4521
    },
    {
      "epoch": 0.7711459754433834,
      "grad_norm": 0.36614111065864563,
      "learning_rate": 2.627226631898761e-07,
      "loss": 0.5046,
      "step": 4522
    },
    {
      "epoch": 0.7713165075034106,
      "grad_norm": 0.4957779347896576,
      "learning_rate": 2.623496340975434e-07,
      "loss": 0.4925,
      "step": 4523
    },
    {
      "epoch": 0.7714870395634379,
      "grad_norm": 0.45070260763168335,
      "learning_rate": 2.6197683003039256e-07,
      "loss": 0.4907,
      "step": 4524
    },
    {
      "epoch": 0.7716575716234653,
      "grad_norm": 0.4274146854877472,
      "learning_rate": 2.616042511021492e-07,
      "loss": 0.4984,
      "step": 4525
    },
    {
      "epoch": 0.7718281036834925,
      "grad_norm": 0.3768433630466461,
      "learning_rate": 2.6123189742647226e-07,
      "loss": 0.5014,
      "step": 4526
    },
    {
      "epoch": 0.7719986357435198,
      "grad_norm": 0.37700000405311584,
      "learning_rate": 2.608597691169499e-07,
      "loss": 0.505,
      "step": 4527
    },
    {
      "epoch": 0.772169167803547,
      "grad_norm": 0.36473509669303894,
      "learning_rate": 2.6048786628710247e-07,
      "loss": 0.49,
      "step": 4528
    },
    {
      "epoch": 0.7723396998635743,
      "grad_norm": 0.4933687448501587,
      "learning_rate": 2.6011618905038175e-07,
      "loss": 0.4993,
      "step": 4529
    },
    {
      "epoch": 0.7725102319236017,
      "grad_norm": 0.46770721673965454,
      "learning_rate": 2.5974473752016987e-07,
      "loss": 0.4912,
      "step": 4530
    },
    {
      "epoch": 0.772680763983629,
      "grad_norm": 0.3749657869338989,
      "learning_rate": 2.593735118097808e-07,
      "loss": 0.4921,
      "step": 4531
    },
    {
      "epoch": 0.7728512960436562,
      "grad_norm": 0.4584238827228546,
      "learning_rate": 2.5900251203245945e-07,
      "loss": 0.5103,
      "step": 4532
    },
    {
      "epoch": 0.7730218281036835,
      "grad_norm": 0.4406527578830719,
      "learning_rate": 2.5863173830138214e-07,
      "loss": 0.4925,
      "step": 4533
    },
    {
      "epoch": 0.7731923601637107,
      "grad_norm": 0.43562811613082886,
      "learning_rate": 2.5826119072965495e-07,
      "loss": 0.5035,
      "step": 4534
    },
    {
      "epoch": 0.7733628922237381,
      "grad_norm": 0.5225963592529297,
      "learning_rate": 2.5789086943031705e-07,
      "loss": 0.4919,
      "step": 4535
    },
    {
      "epoch": 0.7735334242837654,
      "grad_norm": 0.4101422131061554,
      "learning_rate": 2.575207745163366e-07,
      "loss": 0.493,
      "step": 4536
    },
    {
      "epoch": 0.7737039563437926,
      "grad_norm": 0.43135133385658264,
      "learning_rate": 2.571509061006138e-07,
      "loss": 0.5007,
      "step": 4537
    },
    {
      "epoch": 0.7738744884038199,
      "grad_norm": 0.4580155909061432,
      "learning_rate": 2.5678126429597987e-07,
      "loss": 0.4936,
      "step": 4538
    },
    {
      "epoch": 0.7740450204638472,
      "grad_norm": 0.4478892683982849,
      "learning_rate": 2.564118492151957e-07,
      "loss": 0.4903,
      "step": 4539
    },
    {
      "epoch": 0.7742155525238745,
      "grad_norm": 0.4683667719364166,
      "learning_rate": 2.560426609709543e-07,
      "loss": 0.5054,
      "step": 4540
    },
    {
      "epoch": 0.7743860845839018,
      "grad_norm": 0.5699395537376404,
      "learning_rate": 2.5567369967587907e-07,
      "loss": 0.4867,
      "step": 4541
    },
    {
      "epoch": 0.774556616643929,
      "grad_norm": 0.4768561124801636,
      "learning_rate": 2.553049654425243e-07,
      "loss": 0.5034,
      "step": 4542
    },
    {
      "epoch": 0.7747271487039563,
      "grad_norm": 0.40361812710762024,
      "learning_rate": 2.549364583833739e-07,
      "loss": 0.4825,
      "step": 4543
    },
    {
      "epoch": 0.7748976807639836,
      "grad_norm": 0.45786723494529724,
      "learning_rate": 2.5456817861084474e-07,
      "loss": 0.5049,
      "step": 4544
    },
    {
      "epoch": 0.775068212824011,
      "grad_norm": 0.39945563673973083,
      "learning_rate": 2.542001262372821e-07,
      "loss": 0.5072,
      "step": 4545
    },
    {
      "epoch": 0.7752387448840382,
      "grad_norm": 0.4828221797943115,
      "learning_rate": 2.538323013749631e-07,
      "loss": 0.4982,
      "step": 4546
    },
    {
      "epoch": 0.7754092769440655,
      "grad_norm": 0.41438600420951843,
      "learning_rate": 2.534647041360955e-07,
      "loss": 0.4921,
      "step": 4547
    },
    {
      "epoch": 0.7755798090040927,
      "grad_norm": 0.376027911901474,
      "learning_rate": 2.5309733463281665e-07,
      "loss": 0.4951,
      "step": 4548
    },
    {
      "epoch": 0.7757503410641201,
      "grad_norm": 0.4809345602989197,
      "learning_rate": 2.5273019297719525e-07,
      "loss": 0.5037,
      "step": 4549
    },
    {
      "epoch": 0.7759208731241474,
      "grad_norm": 0.3895203173160553,
      "learning_rate": 2.523632792812306e-07,
      "loss": 0.4927,
      "step": 4550
    },
    {
      "epoch": 0.7760914051841746,
      "grad_norm": 0.42986854910850525,
      "learning_rate": 2.519965936568524e-07,
      "loss": 0.4939,
      "step": 4551
    },
    {
      "epoch": 0.7762619372442019,
      "grad_norm": 0.4543752670288086,
      "learning_rate": 2.516301362159195e-07,
      "loss": 0.4909,
      "step": 4552
    },
    {
      "epoch": 0.7764324693042292,
      "grad_norm": 0.49290981888771057,
      "learning_rate": 2.5126390707022356e-07,
      "loss": 0.4931,
      "step": 4553
    },
    {
      "epoch": 0.7766030013642565,
      "grad_norm": 0.47044581174850464,
      "learning_rate": 2.5089790633148426e-07,
      "loss": 0.4873,
      "step": 4554
    },
    {
      "epoch": 0.7767735334242838,
      "grad_norm": 0.40951424837112427,
      "learning_rate": 2.5053213411135293e-07,
      "loss": 0.4887,
      "step": 4555
    },
    {
      "epoch": 0.776944065484311,
      "grad_norm": 0.45414459705352783,
      "learning_rate": 2.50166590521411e-07,
      "loss": 0.4901,
      "step": 4556
    },
    {
      "epoch": 0.7771145975443383,
      "grad_norm": 0.4336784780025482,
      "learning_rate": 2.498012756731695e-07,
      "loss": 0.4989,
      "step": 4557
    },
    {
      "epoch": 0.7772851296043656,
      "grad_norm": 0.5325224995613098,
      "learning_rate": 2.494361896780705e-07,
      "loss": 0.4842,
      "step": 4558
    },
    {
      "epoch": 0.777455661664393,
      "grad_norm": 0.5549018979072571,
      "learning_rate": 2.4907133264748586e-07,
      "loss": 0.4966,
      "step": 4559
    },
    {
      "epoch": 0.7776261937244202,
      "grad_norm": 0.40138453245162964,
      "learning_rate": 2.4870670469271785e-07,
      "loss": 0.4968,
      "step": 4560
    },
    {
      "epoch": 0.7777967257844475,
      "grad_norm": 0.4988255202770233,
      "learning_rate": 2.483423059249983e-07,
      "loss": 0.4895,
      "step": 4561
    },
    {
      "epoch": 0.7779672578444747,
      "grad_norm": 0.4912653863430023,
      "learning_rate": 2.4797813645548973e-07,
      "loss": 0.4998,
      "step": 4562
    },
    {
      "epoch": 0.778137789904502,
      "grad_norm": 0.40639805793762207,
      "learning_rate": 2.476141963952844e-07,
      "loss": 0.4951,
      "step": 4563
    },
    {
      "epoch": 0.7783083219645294,
      "grad_norm": 0.5302715301513672,
      "learning_rate": 2.472504858554046e-07,
      "loss": 0.4897,
      "step": 4564
    },
    {
      "epoch": 0.7784788540245566,
      "grad_norm": 0.5033547282218933,
      "learning_rate": 2.468870049468032e-07,
      "loss": 0.5007,
      "step": 4565
    },
    {
      "epoch": 0.7786493860845839,
      "grad_norm": 0.4595785439014435,
      "learning_rate": 2.465237537803617e-07,
      "loss": 0.4971,
      "step": 4566
    },
    {
      "epoch": 0.7788199181446112,
      "grad_norm": 0.506290853023529,
      "learning_rate": 2.4616073246689254e-07,
      "loss": 0.4946,
      "step": 4567
    },
    {
      "epoch": 0.7789904502046384,
      "grad_norm": 0.5239617228507996,
      "learning_rate": 2.4579794111713797e-07,
      "loss": 0.4873,
      "step": 4568
    },
    {
      "epoch": 0.7791609822646658,
      "grad_norm": 0.3967903256416321,
      "learning_rate": 2.4543537984176983e-07,
      "loss": 0.5091,
      "step": 4569
    },
    {
      "epoch": 0.779331514324693,
      "grad_norm": 0.4829061031341553,
      "learning_rate": 2.4507304875139017e-07,
      "loss": 0.5017,
      "step": 4570
    },
    {
      "epoch": 0.7795020463847203,
      "grad_norm": 0.5171961784362793,
      "learning_rate": 2.4471094795653e-07,
      "loss": 0.4972,
      "step": 4571
    },
    {
      "epoch": 0.7796725784447476,
      "grad_norm": 0.4271591305732727,
      "learning_rate": 2.443490775676507e-07,
      "loss": 0.5031,
      "step": 4572
    },
    {
      "epoch": 0.779843110504775,
      "grad_norm": 0.4533305764198303,
      "learning_rate": 2.4398743769514345e-07,
      "loss": 0.5033,
      "step": 4573
    },
    {
      "epoch": 0.7800136425648022,
      "grad_norm": 0.45902299880981445,
      "learning_rate": 2.436260284493292e-07,
      "loss": 0.5008,
      "step": 4574
    },
    {
      "epoch": 0.7801841746248295,
      "grad_norm": 0.4595572054386139,
      "learning_rate": 2.4326484994045754e-07,
      "loss": 0.489,
      "step": 4575
    },
    {
      "epoch": 0.7803547066848567,
      "grad_norm": 0.5287113785743713,
      "learning_rate": 2.429039022787088e-07,
      "loss": 0.4937,
      "step": 4576
    },
    {
      "epoch": 0.780525238744884,
      "grad_norm": 0.48877352476119995,
      "learning_rate": 2.425431855741926e-07,
      "loss": 0.4846,
      "step": 4577
    },
    {
      "epoch": 0.7806957708049114,
      "grad_norm": 0.5545276403427124,
      "learning_rate": 2.4218269993694736e-07,
      "loss": 0.5099,
      "step": 4578
    },
    {
      "epoch": 0.7808663028649386,
      "grad_norm": 0.5091724395751953,
      "learning_rate": 2.4182244547694255e-07,
      "loss": 0.4957,
      "step": 4579
    },
    {
      "epoch": 0.7810368349249659,
      "grad_norm": 0.4598775804042816,
      "learning_rate": 2.414624223040756e-07,
      "loss": 0.4944,
      "step": 4580
    },
    {
      "epoch": 0.7812073669849932,
      "grad_norm": 0.5534014701843262,
      "learning_rate": 2.41102630528174e-07,
      "loss": 0.5014,
      "step": 4581
    },
    {
      "epoch": 0.7813778990450204,
      "grad_norm": 0.44594284892082214,
      "learning_rate": 2.407430702589948e-07,
      "loss": 0.5019,
      "step": 4582
    },
    {
      "epoch": 0.7815484311050478,
      "grad_norm": 0.461640328168869,
      "learning_rate": 2.403837416062246e-07,
      "loss": 0.4904,
      "step": 4583
    },
    {
      "epoch": 0.781718963165075,
      "grad_norm": 0.5966527462005615,
      "learning_rate": 2.400246446794783e-07,
      "loss": 0.5012,
      "step": 4584
    },
    {
      "epoch": 0.7818894952251023,
      "grad_norm": 0.3989943563938141,
      "learning_rate": 2.396657795883013e-07,
      "loss": 0.49,
      "step": 4585
    },
    {
      "epoch": 0.7820600272851296,
      "grad_norm": 0.548056423664093,
      "learning_rate": 2.3930714644216784e-07,
      "loss": 0.4977,
      "step": 4586
    },
    {
      "epoch": 0.7822305593451568,
      "grad_norm": 0.47558221220970154,
      "learning_rate": 2.3894874535048064e-07,
      "loss": 0.4935,
      "step": 4587
    },
    {
      "epoch": 0.7824010914051842,
      "grad_norm": 0.4634142220020294,
      "learning_rate": 2.3859057642257345e-07,
      "loss": 0.4945,
      "step": 4588
    },
    {
      "epoch": 0.7825716234652115,
      "grad_norm": 0.5071300864219666,
      "learning_rate": 2.382326397677073e-07,
      "loss": 0.4945,
      "step": 4589
    },
    {
      "epoch": 0.7827421555252387,
      "grad_norm": 0.3968707025051117,
      "learning_rate": 2.3787493549507348e-07,
      "loss": 0.4949,
      "step": 4590
    },
    {
      "epoch": 0.782912687585266,
      "grad_norm": 0.5555402636528015,
      "learning_rate": 2.3751746371379195e-07,
      "loss": 0.5007,
      "step": 4591
    },
    {
      "epoch": 0.7830832196452933,
      "grad_norm": 0.4424824118614197,
      "learning_rate": 2.371602245329121e-07,
      "loss": 0.48,
      "step": 4592
    },
    {
      "epoch": 0.7832537517053206,
      "grad_norm": 0.5057586431503296,
      "learning_rate": 2.3680321806141185e-07,
      "loss": 0.5071,
      "step": 4593
    },
    {
      "epoch": 0.7834242837653479,
      "grad_norm": 0.44602543115615845,
      "learning_rate": 2.3644644440819843e-07,
      "loss": 0.4994,
      "step": 4594
    },
    {
      "epoch": 0.7835948158253752,
      "grad_norm": 0.4586191475391388,
      "learning_rate": 2.3608990368210852e-07,
      "loss": 0.5036,
      "step": 4595
    },
    {
      "epoch": 0.7837653478854024,
      "grad_norm": 0.5694007873535156,
      "learning_rate": 2.3573359599190618e-07,
      "loss": 0.4943,
      "step": 4596
    },
    {
      "epoch": 0.7839358799454298,
      "grad_norm": 0.4275916516780853,
      "learning_rate": 2.353775214462868e-07,
      "loss": 0.504,
      "step": 4597
    },
    {
      "epoch": 0.784106412005457,
      "grad_norm": 0.46295440196990967,
      "learning_rate": 2.350216801538723e-07,
      "loss": 0.5034,
      "step": 4598
    },
    {
      "epoch": 0.7842769440654843,
      "grad_norm": 0.49615952372550964,
      "learning_rate": 2.3466607222321482e-07,
      "loss": 0.5159,
      "step": 4599
    },
    {
      "epoch": 0.7844474761255116,
      "grad_norm": 0.48973459005355835,
      "learning_rate": 2.3431069776279515e-07,
      "loss": 0.5121,
      "step": 4600
    },
    {
      "epoch": 0.7846180081855388,
      "grad_norm": 0.47051751613616943,
      "learning_rate": 2.3395555688102214e-07,
      "loss": 0.5045,
      "step": 4601
    },
    {
      "epoch": 0.7847885402455662,
      "grad_norm": 0.4109767973423004,
      "learning_rate": 2.3360064968623422e-07,
      "loss": 0.5067,
      "step": 4602
    },
    {
      "epoch": 0.7849590723055935,
      "grad_norm": 0.4106697738170624,
      "learning_rate": 2.33245976286698e-07,
      "loss": 0.5038,
      "step": 4603
    },
    {
      "epoch": 0.7851296043656207,
      "grad_norm": 0.4463314116001129,
      "learning_rate": 2.3289153679060952e-07,
      "loss": 0.507,
      "step": 4604
    },
    {
      "epoch": 0.785300136425648,
      "grad_norm": 0.5195944905281067,
      "learning_rate": 2.3253733130609192e-07,
      "loss": 0.5078,
      "step": 4605
    },
    {
      "epoch": 0.7854706684856753,
      "grad_norm": 0.4397755563259125,
      "learning_rate": 2.3218335994119912e-07,
      "loss": 0.5056,
      "step": 4606
    },
    {
      "epoch": 0.7856412005457026,
      "grad_norm": 0.409441739320755,
      "learning_rate": 2.3182962280391165e-07,
      "loss": 0.5062,
      "step": 4607
    },
    {
      "epoch": 0.7858117326057299,
      "grad_norm": 0.5190573930740356,
      "learning_rate": 2.3147612000213962e-07,
      "loss": 0.5072,
      "step": 4608
    },
    {
      "epoch": 0.7859822646657572,
      "grad_norm": 0.5061396360397339,
      "learning_rate": 2.3112285164372173e-07,
      "loss": 0.4996,
      "step": 4609
    },
    {
      "epoch": 0.7861527967257844,
      "grad_norm": 0.43007734417915344,
      "learning_rate": 2.307698178364243e-07,
      "loss": 0.5052,
      "step": 4610
    },
    {
      "epoch": 0.7863233287858117,
      "grad_norm": 0.5201791524887085,
      "learning_rate": 2.3041701868794288e-07,
      "loss": 0.499,
      "step": 4611
    },
    {
      "epoch": 0.786493860845839,
      "grad_norm": 0.4248791038990021,
      "learning_rate": 2.300644543059012e-07,
      "loss": 0.502,
      "step": 4612
    },
    {
      "epoch": 0.7866643929058663,
      "grad_norm": 0.40481671690940857,
      "learning_rate": 2.297121247978518e-07,
      "loss": 0.4925,
      "step": 4613
    },
    {
      "epoch": 0.7868349249658936,
      "grad_norm": 0.4447843134403229,
      "learning_rate": 2.293600302712742e-07,
      "loss": 0.5085,
      "step": 4614
    },
    {
      "epoch": 0.7870054570259208,
      "grad_norm": 0.6058522462844849,
      "learning_rate": 2.2900817083357832e-07,
      "loss": 0.4921,
      "step": 4615
    },
    {
      "epoch": 0.7871759890859482,
      "grad_norm": 0.46241334080696106,
      "learning_rate": 2.286565465921007e-07,
      "loss": 0.4988,
      "step": 4616
    },
    {
      "epoch": 0.7873465211459755,
      "grad_norm": 0.5033152103424072,
      "learning_rate": 2.2830515765410623e-07,
      "loss": 0.4892,
      "step": 4617
    },
    {
      "epoch": 0.7875170532060027,
      "grad_norm": 0.4851181209087372,
      "learning_rate": 2.2795400412678928e-07,
      "loss": 0.5072,
      "step": 4618
    },
    {
      "epoch": 0.78768758526603,
      "grad_norm": 0.5065212249755859,
      "learning_rate": 2.2760308611727102e-07,
      "loss": 0.4882,
      "step": 4619
    },
    {
      "epoch": 0.7878581173260573,
      "grad_norm": 0.6156156659126282,
      "learning_rate": 2.272524037326015e-07,
      "loss": 0.5022,
      "step": 4620
    },
    {
      "epoch": 0.7880286493860846,
      "grad_norm": 0.4571601152420044,
      "learning_rate": 2.2690195707975882e-07,
      "loss": 0.4996,
      "step": 4621
    },
    {
      "epoch": 0.7881991814461119,
      "grad_norm": 0.47723454236984253,
      "learning_rate": 2.2655174626564915e-07,
      "loss": 0.5001,
      "step": 4622
    },
    {
      "epoch": 0.7883697135061392,
      "grad_norm": 0.5316779613494873,
      "learning_rate": 2.2620177139710632e-07,
      "loss": 0.503,
      "step": 4623
    },
    {
      "epoch": 0.7885402455661664,
      "grad_norm": 0.5502619743347168,
      "learning_rate": 2.2585203258089264e-07,
      "loss": 0.4914,
      "step": 4624
    },
    {
      "epoch": 0.7887107776261937,
      "grad_norm": 0.5683772563934326,
      "learning_rate": 2.255025299236984e-07,
      "loss": 0.5019,
      "step": 4625
    },
    {
      "epoch": 0.788881309686221,
      "grad_norm": 0.45791706442832947,
      "learning_rate": 2.2515326353214152e-07,
      "loss": 0.4842,
      "step": 4626
    },
    {
      "epoch": 0.7890518417462483,
      "grad_norm": 0.5102027654647827,
      "learning_rate": 2.248042335127685e-07,
      "loss": 0.4976,
      "step": 4627
    },
    {
      "epoch": 0.7892223738062756,
      "grad_norm": 0.40297356247901917,
      "learning_rate": 2.244554399720528e-07,
      "loss": 0.4915,
      "step": 4628
    },
    {
      "epoch": 0.7893929058663028,
      "grad_norm": 0.4550130367279053,
      "learning_rate": 2.2410688301639618e-07,
      "loss": 0.4972,
      "step": 4629
    },
    {
      "epoch": 0.7895634379263301,
      "grad_norm": 0.5347708463668823,
      "learning_rate": 2.237585627521286e-07,
      "loss": 0.4883,
      "step": 4630
    },
    {
      "epoch": 0.7897339699863575,
      "grad_norm": 0.42419931292533875,
      "learning_rate": 2.2341047928550765e-07,
      "loss": 0.5032,
      "step": 4631
    },
    {
      "epoch": 0.7899045020463847,
      "grad_norm": 0.5129116177558899,
      "learning_rate": 2.230626327227179e-07,
      "loss": 0.5012,
      "step": 4632
    },
    {
      "epoch": 0.790075034106412,
      "grad_norm": 0.5253115296363831,
      "learning_rate": 2.2271502316987272e-07,
      "loss": 0.4891,
      "step": 4633
    },
    {
      "epoch": 0.7902455661664393,
      "grad_norm": 0.38880401849746704,
      "learning_rate": 2.223676507330125e-07,
      "loss": 0.4922,
      "step": 4634
    },
    {
      "epoch": 0.7904160982264665,
      "grad_norm": 0.5668837428092957,
      "learning_rate": 2.220205155181057e-07,
      "loss": 0.4936,
      "step": 4635
    },
    {
      "epoch": 0.7905866302864939,
      "grad_norm": 0.41947200894355774,
      "learning_rate": 2.2167361763104846e-07,
      "loss": 0.4942,
      "step": 4636
    },
    {
      "epoch": 0.7907571623465212,
      "grad_norm": 0.5962218642234802,
      "learning_rate": 2.2132695717766374e-07,
      "loss": 0.4875,
      "step": 4637
    },
    {
      "epoch": 0.7909276944065484,
      "grad_norm": 0.4495481550693512,
      "learning_rate": 2.2098053426370294e-07,
      "loss": 0.4971,
      "step": 4638
    },
    {
      "epoch": 0.7910982264665757,
      "grad_norm": 0.5138792395591736,
      "learning_rate": 2.2063434899484494e-07,
      "loss": 0.5022,
      "step": 4639
    },
    {
      "epoch": 0.791268758526603,
      "grad_norm": 0.5427955389022827,
      "learning_rate": 2.2028840147669529e-07,
      "loss": 0.503,
      "step": 4640
    },
    {
      "epoch": 0.7914392905866303,
      "grad_norm": 0.4466486871242523,
      "learning_rate": 2.19942691814788e-07,
      "loss": 0.4928,
      "step": 4641
    },
    {
      "epoch": 0.7916098226466576,
      "grad_norm": 0.6267170906066895,
      "learning_rate": 2.1959722011458398e-07,
      "loss": 0.4923,
      "step": 4642
    },
    {
      "epoch": 0.7917803547066848,
      "grad_norm": 0.4920784533023834,
      "learning_rate": 2.1925198648147175e-07,
      "loss": 0.4912,
      "step": 4643
    },
    {
      "epoch": 0.7919508867667121,
      "grad_norm": 0.5684638619422913,
      "learning_rate": 2.189069910207671e-07,
      "loss": 0.4943,
      "step": 4644
    },
    {
      "epoch": 0.7921214188267395,
      "grad_norm": 0.5882368683815002,
      "learning_rate": 2.185622338377135e-07,
      "loss": 0.4884,
      "step": 4645
    },
    {
      "epoch": 0.7922919508867667,
      "grad_norm": 0.5498755574226379,
      "learning_rate": 2.1821771503748092e-07,
      "loss": 0.5037,
      "step": 4646
    },
    {
      "epoch": 0.792462482946794,
      "grad_norm": 0.5423489212989807,
      "learning_rate": 2.1787343472516732e-07,
      "loss": 0.4927,
      "step": 4647
    },
    {
      "epoch": 0.7926330150068213,
      "grad_norm": 0.6042127013206482,
      "learning_rate": 2.1752939300579803e-07,
      "loss": 0.4992,
      "step": 4648
    },
    {
      "epoch": 0.7928035470668485,
      "grad_norm": 0.6443652510643005,
      "learning_rate": 2.1718558998432483e-07,
      "loss": 0.4856,
      "step": 4649
    },
    {
      "epoch": 0.7929740791268759,
      "grad_norm": 0.39260923862457275,
      "learning_rate": 2.168420257656272e-07,
      "loss": 0.5008,
      "step": 4650
    },
    {
      "epoch": 0.7931446111869032,
      "grad_norm": 0.5619345903396606,
      "learning_rate": 2.1649870045451185e-07,
      "loss": 0.4923,
      "step": 4651
    },
    {
      "epoch": 0.7933151432469304,
      "grad_norm": 0.3600766360759735,
      "learning_rate": 2.1615561415571252e-07,
      "loss": 0.4897,
      "step": 4652
    },
    {
      "epoch": 0.7934856753069577,
      "grad_norm": 0.4772145450115204,
      "learning_rate": 2.1581276697388976e-07,
      "loss": 0.4913,
      "step": 4653
    },
    {
      "epoch": 0.793656207366985,
      "grad_norm": 0.5267781019210815,
      "learning_rate": 2.1547015901363191e-07,
      "loss": 0.4922,
      "step": 4654
    },
    {
      "epoch": 0.7938267394270123,
      "grad_norm": 0.38906392455101013,
      "learning_rate": 2.151277903794532e-07,
      "loss": 0.5007,
      "step": 4655
    },
    {
      "epoch": 0.7939972714870396,
      "grad_norm": 0.4681243300437927,
      "learning_rate": 2.147856611757957e-07,
      "loss": 0.4967,
      "step": 4656
    },
    {
      "epoch": 0.7941678035470668,
      "grad_norm": 0.44217199087142944,
      "learning_rate": 2.1444377150702847e-07,
      "loss": 0.5008,
      "step": 4657
    },
    {
      "epoch": 0.7943383356070941,
      "grad_norm": 0.43402767181396484,
      "learning_rate": 2.1410212147744638e-07,
      "loss": 0.5007,
      "step": 4658
    },
    {
      "epoch": 0.7945088676671214,
      "grad_norm": 0.43505313992500305,
      "learning_rate": 2.1376071119127342e-07,
      "loss": 0.5071,
      "step": 4659
    },
    {
      "epoch": 0.7946793997271487,
      "grad_norm": 0.5335257053375244,
      "learning_rate": 2.1341954075265793e-07,
      "loss": 0.4924,
      "step": 4660
    },
    {
      "epoch": 0.794849931787176,
      "grad_norm": 0.5619838833808899,
      "learning_rate": 2.130786102656767e-07,
      "loss": 0.5025,
      "step": 4661
    },
    {
      "epoch": 0.7950204638472033,
      "grad_norm": 0.4406771659851074,
      "learning_rate": 2.1273791983433302e-07,
      "loss": 0.4848,
      "step": 4662
    },
    {
      "epoch": 0.7951909959072305,
      "grad_norm": 0.609472393989563,
      "learning_rate": 2.1239746956255638e-07,
      "loss": 0.5002,
      "step": 4663
    },
    {
      "epoch": 0.7953615279672579,
      "grad_norm": 0.4150250554084778,
      "learning_rate": 2.1205725955420354e-07,
      "loss": 0.49,
      "step": 4664
    },
    {
      "epoch": 0.7955320600272852,
      "grad_norm": 0.5046654343605042,
      "learning_rate": 2.1171728991305798e-07,
      "loss": 0.4919,
      "step": 4665
    },
    {
      "epoch": 0.7957025920873124,
      "grad_norm": 0.4161365032196045,
      "learning_rate": 2.1137756074282992e-07,
      "loss": 0.5001,
      "step": 4666
    },
    {
      "epoch": 0.7958731241473397,
      "grad_norm": 0.47114861011505127,
      "learning_rate": 2.1103807214715512e-07,
      "loss": 0.4944,
      "step": 4667
    },
    {
      "epoch": 0.796043656207367,
      "grad_norm": 0.5760286450386047,
      "learning_rate": 2.106988242295981e-07,
      "loss": 0.4911,
      "step": 4668
    },
    {
      "epoch": 0.7962141882673943,
      "grad_norm": 0.43809956312179565,
      "learning_rate": 2.1035981709364784e-07,
      "loss": 0.4979,
      "step": 4669
    },
    {
      "epoch": 0.7963847203274216,
      "grad_norm": 0.5407879948616028,
      "learning_rate": 2.1002105084272086e-07,
      "loss": 0.4978,
      "step": 4670
    },
    {
      "epoch": 0.7965552523874488,
      "grad_norm": 0.40855517983436584,
      "learning_rate": 2.0968252558016056e-07,
      "loss": 0.4943,
      "step": 4671
    },
    {
      "epoch": 0.7967257844474761,
      "grad_norm": 0.5361022353172302,
      "learning_rate": 2.093442414092356e-07,
      "loss": 0.507,
      "step": 4672
    },
    {
      "epoch": 0.7968963165075034,
      "grad_norm": 0.46625766158103943,
      "learning_rate": 2.090061984331423e-07,
      "loss": 0.498,
      "step": 4673
    },
    {
      "epoch": 0.7970668485675307,
      "grad_norm": 0.44276198744773865,
      "learning_rate": 2.0866839675500274e-07,
      "loss": 0.5048,
      "step": 4674
    },
    {
      "epoch": 0.797237380627558,
      "grad_norm": 0.5449864268302917,
      "learning_rate": 2.0833083647786603e-07,
      "loss": 0.4888,
      "step": 4675
    },
    {
      "epoch": 0.7974079126875853,
      "grad_norm": 0.41112643480300903,
      "learning_rate": 2.0799351770470613e-07,
      "loss": 0.4944,
      "step": 4676
    },
    {
      "epoch": 0.7975784447476125,
      "grad_norm": 0.5226261019706726,
      "learning_rate": 2.0765644053842583e-07,
      "loss": 0.4907,
      "step": 4677
    },
    {
      "epoch": 0.7977489768076398,
      "grad_norm": 0.4252602159976959,
      "learning_rate": 2.0731960508185192e-07,
      "loss": 0.4966,
      "step": 4678
    },
    {
      "epoch": 0.7979195088676672,
      "grad_norm": 0.5162569284439087,
      "learning_rate": 2.0698301143773793e-07,
      "loss": 0.5034,
      "step": 4679
    },
    {
      "epoch": 0.7980900409276944,
      "grad_norm": 0.48317229747772217,
      "learning_rate": 2.0664665970876497e-07,
      "loss": 0.4978,
      "step": 4680
    },
    {
      "epoch": 0.7982605729877217,
      "grad_norm": 0.38275066018104553,
      "learning_rate": 2.0631054999753877e-07,
      "loss": 0.5002,
      "step": 4681
    },
    {
      "epoch": 0.798431105047749,
      "grad_norm": 0.5556075572967529,
      "learning_rate": 2.0597468240659194e-07,
      "loss": 0.4898,
      "step": 4682
    },
    {
      "epoch": 0.7986016371077762,
      "grad_norm": 0.47922417521476746,
      "learning_rate": 2.0563905703838317e-07,
      "loss": 0.5095,
      "step": 4683
    },
    {
      "epoch": 0.7987721691678036,
      "grad_norm": 0.5476397275924683,
      "learning_rate": 2.0530367399529762e-07,
      "loss": 0.5018,
      "step": 4684
    },
    {
      "epoch": 0.7989427012278308,
      "grad_norm": 0.43253302574157715,
      "learning_rate": 2.0496853337964508e-07,
      "loss": 0.4987,
      "step": 4685
    },
    {
      "epoch": 0.7991132332878581,
      "grad_norm": 0.5218864679336548,
      "learning_rate": 2.0463363529366379e-07,
      "loss": 0.4994,
      "step": 4686
    },
    {
      "epoch": 0.7992837653478854,
      "grad_norm": 0.5182597637176514,
      "learning_rate": 2.0429897983951587e-07,
      "loss": 0.5008,
      "step": 4687
    },
    {
      "epoch": 0.7994542974079127,
      "grad_norm": 0.3908626139163971,
      "learning_rate": 2.0396456711928983e-07,
      "loss": 0.4998,
      "step": 4688
    },
    {
      "epoch": 0.79962482946794,
      "grad_norm": 0.4519367814064026,
      "learning_rate": 2.0363039723500158e-07,
      "loss": 0.4982,
      "step": 4689
    },
    {
      "epoch": 0.7997953615279673,
      "grad_norm": 0.3876539468765259,
      "learning_rate": 2.0329647028859101e-07,
      "loss": 0.5014,
      "step": 4690
    },
    {
      "epoch": 0.7999658935879945,
      "grad_norm": 0.501013457775116,
      "learning_rate": 2.0296278638192502e-07,
      "loss": 0.5081,
      "step": 4691
    },
    {
      "epoch": 0.8001364256480218,
      "grad_norm": 0.44867897033691406,
      "learning_rate": 2.0262934561679617e-07,
      "loss": 0.4989,
      "step": 4692
    },
    {
      "epoch": 0.8003069577080492,
      "grad_norm": 0.3659909963607788,
      "learning_rate": 2.0229614809492305e-07,
      "loss": 0.5088,
      "step": 4693
    },
    {
      "epoch": 0.8004774897680764,
      "grad_norm": 0.4918709993362427,
      "learning_rate": 2.019631939179493e-07,
      "loss": 0.5039,
      "step": 4694
    },
    {
      "epoch": 0.8006480218281037,
      "grad_norm": 0.40458929538726807,
      "learning_rate": 2.0163048318744494e-07,
      "loss": 0.5017,
      "step": 4695
    },
    {
      "epoch": 0.800818553888131,
      "grad_norm": 0.4595339894294739,
      "learning_rate": 2.0129801600490608e-07,
      "loss": 0.5111,
      "step": 4696
    },
    {
      "epoch": 0.8009890859481582,
      "grad_norm": 0.5194839835166931,
      "learning_rate": 2.0096579247175321e-07,
      "loss": 0.5088,
      "step": 4697
    },
    {
      "epoch": 0.8011596180081856,
      "grad_norm": 0.3969508111476898,
      "learning_rate": 2.0063381268933426e-07,
      "loss": 0.4982,
      "step": 4698
    },
    {
      "epoch": 0.8013301500682128,
      "grad_norm": 0.42648452520370483,
      "learning_rate": 2.0030207675892125e-07,
      "loss": 0.5058,
      "step": 4699
    },
    {
      "epoch": 0.8015006821282401,
      "grad_norm": 0.40289372205734253,
      "learning_rate": 1.9997058478171264e-07,
      "loss": 0.4778,
      "step": 4700
    },
    {
      "epoch": 0.8016712141882674,
      "grad_norm": 0.5085612535476685,
      "learning_rate": 1.9963933685883255e-07,
      "loss": 0.4984,
      "step": 4701
    },
    {
      "epoch": 0.8018417462482946,
      "grad_norm": 0.47089236974716187,
      "learning_rate": 1.993083330913298e-07,
      "loss": 0.4953,
      "step": 4702
    },
    {
      "epoch": 0.802012278308322,
      "grad_norm": 0.370245099067688,
      "learning_rate": 1.9897757358017957e-07,
      "loss": 0.5118,
      "step": 4703
    },
    {
      "epoch": 0.8021828103683493,
      "grad_norm": 0.3889061510562897,
      "learning_rate": 1.9864705842628242e-07,
      "loss": 0.497,
      "step": 4704
    },
    {
      "epoch": 0.8023533424283765,
      "grad_norm": 0.3743768036365509,
      "learning_rate": 1.9831678773046428e-07,
      "loss": 0.4986,
      "step": 4705
    },
    {
      "epoch": 0.8025238744884038,
      "grad_norm": 0.4428059160709381,
      "learning_rate": 1.9798676159347577e-07,
      "loss": 0.4849,
      "step": 4706
    },
    {
      "epoch": 0.802694406548431,
      "grad_norm": 0.3861129581928253,
      "learning_rate": 1.9765698011599467e-07,
      "loss": 0.4995,
      "step": 4707
    },
    {
      "epoch": 0.8028649386084584,
      "grad_norm": 0.4313037395477295,
      "learning_rate": 1.9732744339862199e-07,
      "loss": 0.4981,
      "step": 4708
    },
    {
      "epoch": 0.8030354706684857,
      "grad_norm": 0.48329731822013855,
      "learning_rate": 1.9699815154188564e-07,
      "loss": 0.49,
      "step": 4709
    },
    {
      "epoch": 0.803206002728513,
      "grad_norm": 0.3620572090148926,
      "learning_rate": 1.9666910464623847e-07,
      "loss": 0.5051,
      "step": 4710
    },
    {
      "epoch": 0.8033765347885402,
      "grad_norm": 0.4337708652019501,
      "learning_rate": 1.9634030281205785e-07,
      "loss": 0.5011,
      "step": 4711
    },
    {
      "epoch": 0.8035470668485676,
      "grad_norm": 0.4114370346069336,
      "learning_rate": 1.9601174613964725e-07,
      "loss": 0.4985,
      "step": 4712
    },
    {
      "epoch": 0.8037175989085948,
      "grad_norm": 0.4311154782772064,
      "learning_rate": 1.9568343472923526e-07,
      "loss": 0.5064,
      "step": 4713
    },
    {
      "epoch": 0.8038881309686221,
      "grad_norm": 0.3959866464138031,
      "learning_rate": 1.953553686809753e-07,
      "loss": 0.5048,
      "step": 4714
    },
    {
      "epoch": 0.8040586630286494,
      "grad_norm": 0.5132696628570557,
      "learning_rate": 1.9502754809494609e-07,
      "loss": 0.5014,
      "step": 4715
    },
    {
      "epoch": 0.8042291950886766,
      "grad_norm": 0.42080479860305786,
      "learning_rate": 1.9469997307115188e-07,
      "loss": 0.5027,
      "step": 4716
    },
    {
      "epoch": 0.804399727148704,
      "grad_norm": 0.40302586555480957,
      "learning_rate": 1.9437264370952103e-07,
      "loss": 0.4899,
      "step": 4717
    },
    {
      "epoch": 0.8045702592087313,
      "grad_norm": 0.4599449634552002,
      "learning_rate": 1.9404556010990796e-07,
      "loss": 0.4973,
      "step": 4718
    },
    {
      "epoch": 0.8047407912687585,
      "grad_norm": 0.49274709820747375,
      "learning_rate": 1.937187223720917e-07,
      "loss": 0.5,
      "step": 4719
    },
    {
      "epoch": 0.8049113233287858,
      "grad_norm": 0.38500869274139404,
      "learning_rate": 1.933921305957759e-07,
      "loss": 0.4905,
      "step": 4720
    },
    {
      "epoch": 0.805081855388813,
      "grad_norm": 0.4719330668449402,
      "learning_rate": 1.9306578488058993e-07,
      "loss": 0.4908,
      "step": 4721
    },
    {
      "epoch": 0.8052523874488404,
      "grad_norm": 0.4110085666179657,
      "learning_rate": 1.9273968532608758e-07,
      "loss": 0.4877,
      "step": 4722
    },
    {
      "epoch": 0.8054229195088677,
      "grad_norm": 0.5232803821563721,
      "learning_rate": 1.9241383203174769e-07,
      "loss": 0.5018,
      "step": 4723
    },
    {
      "epoch": 0.805593451568895,
      "grad_norm": 0.5426048636436462,
      "learning_rate": 1.9208822509697425e-07,
      "loss": 0.4889,
      "step": 4724
    },
    {
      "epoch": 0.8057639836289222,
      "grad_norm": 0.4066981077194214,
      "learning_rate": 1.9176286462109573e-07,
      "loss": 0.4943,
      "step": 4725
    },
    {
      "epoch": 0.8059345156889495,
      "grad_norm": 0.6040549874305725,
      "learning_rate": 1.9143775070336524e-07,
      "loss": 0.4991,
      "step": 4726
    },
    {
      "epoch": 0.8061050477489768,
      "grad_norm": 0.6267606616020203,
      "learning_rate": 1.911128834429612e-07,
      "loss": 0.5015,
      "step": 4727
    },
    {
      "epoch": 0.8062755798090041,
      "grad_norm": 0.41096246242523193,
      "learning_rate": 1.907882629389868e-07,
      "loss": 0.4926,
      "step": 4728
    },
    {
      "epoch": 0.8064461118690314,
      "grad_norm": 0.5135898590087891,
      "learning_rate": 1.90463889290469e-07,
      "loss": 0.5014,
      "step": 4729
    },
    {
      "epoch": 0.8066166439290586,
      "grad_norm": 0.5312707424163818,
      "learning_rate": 1.901397625963606e-07,
      "loss": 0.4955,
      "step": 4730
    },
    {
      "epoch": 0.806787175989086,
      "grad_norm": 0.41822531819343567,
      "learning_rate": 1.8981588295553856e-07,
      "loss": 0.4996,
      "step": 4731
    },
    {
      "epoch": 0.8069577080491133,
      "grad_norm": 0.37353163957595825,
      "learning_rate": 1.8949225046680464e-07,
      "loss": 0.4973,
      "step": 4732
    },
    {
      "epoch": 0.8071282401091405,
      "grad_norm": 0.40797603130340576,
      "learning_rate": 1.8916886522888512e-07,
      "loss": 0.4994,
      "step": 4733
    },
    {
      "epoch": 0.8072987721691678,
      "grad_norm": 0.3444114923477173,
      "learning_rate": 1.8884572734043052e-07,
      "loss": 0.4912,
      "step": 4734
    },
    {
      "epoch": 0.807469304229195,
      "grad_norm": 0.4250591993331909,
      "learning_rate": 1.885228369000164e-07,
      "loss": 0.4878,
      "step": 4735
    },
    {
      "epoch": 0.8076398362892224,
      "grad_norm": 0.4311605393886566,
      "learning_rate": 1.8820019400614265e-07,
      "loss": 0.4929,
      "step": 4736
    },
    {
      "epoch": 0.8078103683492497,
      "grad_norm": 0.41141265630722046,
      "learning_rate": 1.878777987572339e-07,
      "loss": 0.5018,
      "step": 4737
    },
    {
      "epoch": 0.807980900409277,
      "grad_norm": 0.3856441378593445,
      "learning_rate": 1.8755565125163845e-07,
      "loss": 0.484,
      "step": 4738
    },
    {
      "epoch": 0.8081514324693042,
      "grad_norm": 0.4327915906906128,
      "learning_rate": 1.872337515876298e-07,
      "loss": 0.4971,
      "step": 4739
    },
    {
      "epoch": 0.8083219645293315,
      "grad_norm": 0.5755069851875305,
      "learning_rate": 1.8691209986340598e-07,
      "loss": 0.4959,
      "step": 4740
    },
    {
      "epoch": 0.8084924965893588,
      "grad_norm": 0.37612563371658325,
      "learning_rate": 1.8659069617708806e-07,
      "loss": 0.4853,
      "step": 4741
    },
    {
      "epoch": 0.8086630286493861,
      "grad_norm": 0.3841988444328308,
      "learning_rate": 1.8626954062672355e-07,
      "loss": 0.4953,
      "step": 4742
    },
    {
      "epoch": 0.8088335607094134,
      "grad_norm": 0.37716707587242126,
      "learning_rate": 1.8594863331028226e-07,
      "loss": 0.4875,
      "step": 4743
    },
    {
      "epoch": 0.8090040927694406,
      "grad_norm": 0.426252156496048,
      "learning_rate": 1.8562797432565927e-07,
      "loss": 0.4973,
      "step": 4744
    },
    {
      "epoch": 0.8091746248294679,
      "grad_norm": 0.4737507104873657,
      "learning_rate": 1.8530756377067394e-07,
      "loss": 0.4858,
      "step": 4745
    },
    {
      "epoch": 0.8093451568894953,
      "grad_norm": 0.4766058325767517,
      "learning_rate": 1.8498740174306986e-07,
      "loss": 0.4879,
      "step": 4746
    },
    {
      "epoch": 0.8095156889495225,
      "grad_norm": 0.3758303225040436,
      "learning_rate": 1.8466748834051375e-07,
      "loss": 0.4847,
      "step": 4747
    },
    {
      "epoch": 0.8096862210095498,
      "grad_norm": 0.48403874039649963,
      "learning_rate": 1.8434782366059845e-07,
      "loss": 0.488,
      "step": 4748
    },
    {
      "epoch": 0.809856753069577,
      "grad_norm": 0.4095841646194458,
      "learning_rate": 1.8402840780083932e-07,
      "loss": 0.4857,
      "step": 4749
    },
    {
      "epoch": 0.8100272851296043,
      "grad_norm": 0.53790283203125,
      "learning_rate": 1.837092408586756e-07,
      "loss": 0.4896,
      "step": 4750
    },
    {
      "epoch": 0.8101978171896317,
      "grad_norm": 0.5169858932495117,
      "learning_rate": 1.8339032293147253e-07,
      "loss": 0.4875,
      "step": 4751
    },
    {
      "epoch": 0.810368349249659,
      "grad_norm": 0.42378848791122437,
      "learning_rate": 1.830716541165173e-07,
      "loss": 0.4898,
      "step": 4752
    },
    {
      "epoch": 0.8105388813096862,
      "grad_norm": 0.49144625663757324,
      "learning_rate": 1.8275323451102225e-07,
      "loss": 0.5023,
      "step": 4753
    },
    {
      "epoch": 0.8107094133697135,
      "grad_norm": 0.41334912180900574,
      "learning_rate": 1.8243506421212333e-07,
      "loss": 0.485,
      "step": 4754
    },
    {
      "epoch": 0.8108799454297408,
      "grad_norm": 0.5738430023193359,
      "learning_rate": 1.8211714331688093e-07,
      "loss": 0.4949,
      "step": 4755
    },
    {
      "epoch": 0.8110504774897681,
      "grad_norm": 0.47940924763679504,
      "learning_rate": 1.8179947192227834e-07,
      "loss": 0.4962,
      "step": 4756
    },
    {
      "epoch": 0.8112210095497954,
      "grad_norm": 0.4434712827205658,
      "learning_rate": 1.8148205012522358e-07,
      "loss": 0.4871,
      "step": 4757
    },
    {
      "epoch": 0.8113915416098226,
      "grad_norm": 0.5584090948104858,
      "learning_rate": 1.8116487802254868e-07,
      "loss": 0.4895,
      "step": 4758
    },
    {
      "epoch": 0.8115620736698499,
      "grad_norm": 0.587884783744812,
      "learning_rate": 1.808479557110081e-07,
      "loss": 0.4979,
      "step": 4759
    },
    {
      "epoch": 0.8117326057298773,
      "grad_norm": 0.6662174463272095,
      "learning_rate": 1.8053128328728243e-07,
      "loss": 0.4949,
      "step": 4760
    },
    {
      "epoch": 0.8119031377899045,
      "grad_norm": 0.3959568440914154,
      "learning_rate": 1.802148608479737e-07,
      "loss": 0.4866,
      "step": 4761
    },
    {
      "epoch": 0.8120736698499318,
      "grad_norm": 0.5144326686859131,
      "learning_rate": 1.7989868848960913e-07,
      "loss": 0.5032,
      "step": 4762
    },
    {
      "epoch": 0.812244201909959,
      "grad_norm": 0.7496875524520874,
      "learning_rate": 1.7958276630863903e-07,
      "loss": 0.4984,
      "step": 4763
    },
    {
      "epoch": 0.8124147339699863,
      "grad_norm": 0.4983857572078705,
      "learning_rate": 1.79267094401438e-07,
      "loss": 0.4983,
      "step": 4764
    },
    {
      "epoch": 0.8125852660300137,
      "grad_norm": 0.46375688910484314,
      "learning_rate": 1.789516728643033e-07,
      "loss": 0.4959,
      "step": 4765
    },
    {
      "epoch": 0.812755798090041,
      "grad_norm": 0.4899948239326477,
      "learning_rate": 1.786365017934566e-07,
      "loss": 0.5009,
      "step": 4766
    },
    {
      "epoch": 0.8129263301500682,
      "grad_norm": 0.5301331281661987,
      "learning_rate": 1.783215812850433e-07,
      "loss": 0.4961,
      "step": 4767
    },
    {
      "epoch": 0.8130968622100955,
      "grad_norm": 0.48252782225608826,
      "learning_rate": 1.7800691143513105e-07,
      "loss": 0.5049,
      "step": 4768
    },
    {
      "epoch": 0.8132673942701227,
      "grad_norm": 0.4890320301055908,
      "learning_rate": 1.7769249233971322e-07,
      "loss": 0.4909,
      "step": 4769
    },
    {
      "epoch": 0.8134379263301501,
      "grad_norm": 0.4261488616466522,
      "learning_rate": 1.7737832409470458e-07,
      "loss": 0.4958,
      "step": 4770
    },
    {
      "epoch": 0.8136084583901774,
      "grad_norm": 0.5246241688728333,
      "learning_rate": 1.7706440679594463e-07,
      "loss": 0.5065,
      "step": 4771
    },
    {
      "epoch": 0.8137789904502046,
      "grad_norm": 0.43197351694107056,
      "learning_rate": 1.7675074053919606e-07,
      "loss": 0.494,
      "step": 4772
    },
    {
      "epoch": 0.8139495225102319,
      "grad_norm": 0.49455928802490234,
      "learning_rate": 1.7643732542014438e-07,
      "loss": 0.4949,
      "step": 4773
    },
    {
      "epoch": 0.8141200545702592,
      "grad_norm": 0.4988599419593811,
      "learning_rate": 1.7612416153439917e-07,
      "loss": 0.4993,
      "step": 4774
    },
    {
      "epoch": 0.8142905866302865,
      "grad_norm": 0.5352221131324768,
      "learning_rate": 1.7581124897749337e-07,
      "loss": 0.5002,
      "step": 4775
    },
    {
      "epoch": 0.8144611186903138,
      "grad_norm": 0.47420138120651245,
      "learning_rate": 1.7549858784488316e-07,
      "loss": 0.5019,
      "step": 4776
    },
    {
      "epoch": 0.814631650750341,
      "grad_norm": 0.5199934840202332,
      "learning_rate": 1.7518617823194708e-07,
      "loss": 0.5024,
      "step": 4777
    },
    {
      "epoch": 0.8148021828103683,
      "grad_norm": 0.4888570010662079,
      "learning_rate": 1.7487402023398894e-07,
      "loss": 0.5032,
      "step": 4778
    },
    {
      "epoch": 0.8149727148703957,
      "grad_norm": 0.6146841049194336,
      "learning_rate": 1.745621139462338e-07,
      "loss": 0.5072,
      "step": 4779
    },
    {
      "epoch": 0.815143246930423,
      "grad_norm": 0.6330520510673523,
      "learning_rate": 1.7425045946383102e-07,
      "loss": 0.4926,
      "step": 4780
    },
    {
      "epoch": 0.8153137789904502,
      "grad_norm": 0.5201252102851868,
      "learning_rate": 1.739390568818532e-07,
      "loss": 0.4999,
      "step": 4781
    },
    {
      "epoch": 0.8154843110504775,
      "grad_norm": 0.501669704914093,
      "learning_rate": 1.7362790629529514e-07,
      "loss": 0.4924,
      "step": 4782
    },
    {
      "epoch": 0.8156548431105047,
      "grad_norm": 0.5077195167541504,
      "learning_rate": 1.733170077990758e-07,
      "loss": 0.4992,
      "step": 4783
    },
    {
      "epoch": 0.8158253751705321,
      "grad_norm": 0.533851683139801,
      "learning_rate": 1.7300636148803674e-07,
      "loss": 0.483,
      "step": 4784
    },
    {
      "epoch": 0.8159959072305594,
      "grad_norm": 0.5307408571243286,
      "learning_rate": 1.7269596745694297e-07,
      "loss": 0.5017,
      "step": 4785
    },
    {
      "epoch": 0.8161664392905866,
      "grad_norm": 0.43948373198509216,
      "learning_rate": 1.723858258004816e-07,
      "loss": 0.4951,
      "step": 4786
    },
    {
      "epoch": 0.8163369713506139,
      "grad_norm": 0.5590949058532715,
      "learning_rate": 1.7207593661326429e-07,
      "loss": 0.5046,
      "step": 4787
    },
    {
      "epoch": 0.8165075034106412,
      "grad_norm": 0.425890177488327,
      "learning_rate": 1.717662999898241e-07,
      "loss": 0.4891,
      "step": 4788
    },
    {
      "epoch": 0.8166780354706685,
      "grad_norm": 0.6012468338012695,
      "learning_rate": 1.7145691602461812e-07,
      "loss": 0.497,
      "step": 4789
    },
    {
      "epoch": 0.8168485675306958,
      "grad_norm": 0.5476680397987366,
      "learning_rate": 1.711477848120261e-07,
      "loss": 0.5022,
      "step": 4790
    },
    {
      "epoch": 0.817019099590723,
      "grad_norm": 0.48487743735313416,
      "learning_rate": 1.7083890644635017e-07,
      "loss": 0.4901,
      "step": 4791
    },
    {
      "epoch": 0.8171896316507503,
      "grad_norm": 0.4986053705215454,
      "learning_rate": 1.7053028102181601e-07,
      "loss": 0.4852,
      "step": 4792
    },
    {
      "epoch": 0.8173601637107776,
      "grad_norm": 0.4170510470867157,
      "learning_rate": 1.7022190863257184e-07,
      "loss": 0.4859,
      "step": 4793
    },
    {
      "epoch": 0.817530695770805,
      "grad_norm": 0.45927199721336365,
      "learning_rate": 1.6991378937268888e-07,
      "loss": 0.4956,
      "step": 4794
    },
    {
      "epoch": 0.8177012278308322,
      "grad_norm": 0.44013354182243347,
      "learning_rate": 1.6960592333616056e-07,
      "loss": 0.4939,
      "step": 4795
    },
    {
      "epoch": 0.8178717598908595,
      "grad_norm": 0.5605259537696838,
      "learning_rate": 1.6929831061690373e-07,
      "loss": 0.4967,
      "step": 4796
    },
    {
      "epoch": 0.8180422919508867,
      "grad_norm": 0.46457862854003906,
      "learning_rate": 1.6899095130875777e-07,
      "loss": 0.4988,
      "step": 4797
    },
    {
      "epoch": 0.818212824010914,
      "grad_norm": 0.41826432943344116,
      "learning_rate": 1.6868384550548456e-07,
      "loss": 0.4995,
      "step": 4798
    },
    {
      "epoch": 0.8183833560709414,
      "grad_norm": 0.48303788900375366,
      "learning_rate": 1.683769933007691e-07,
      "loss": 0.4926,
      "step": 4799
    },
    {
      "epoch": 0.8185538881309686,
      "grad_norm": 0.4445941746234894,
      "learning_rate": 1.680703947882182e-07,
      "loss": 0.4831,
      "step": 4800
    },
    {
      "epoch": 0.8187244201909959,
      "grad_norm": 0.446552038192749,
      "learning_rate": 1.67764050061362e-07,
      "loss": 0.4966,
      "step": 4801
    },
    {
      "epoch": 0.8188949522510232,
      "grad_norm": 0.5134671330451965,
      "learning_rate": 1.6745795921365316e-07,
      "loss": 0.489,
      "step": 4802
    },
    {
      "epoch": 0.8190654843110505,
      "grad_norm": 0.4526304602622986,
      "learning_rate": 1.6715212233846656e-07,
      "loss": 0.4958,
      "step": 4803
    },
    {
      "epoch": 0.8192360163710778,
      "grad_norm": 0.45974987745285034,
      "learning_rate": 1.6684653952910014e-07,
      "loss": 0.4956,
      "step": 4804
    },
    {
      "epoch": 0.819406548431105,
      "grad_norm": 0.49569687247276306,
      "learning_rate": 1.6654121087877342e-07,
      "loss": 0.5015,
      "step": 4805
    },
    {
      "epoch": 0.8195770804911323,
      "grad_norm": 0.34860217571258545,
      "learning_rate": 1.662361364806292e-07,
      "loss": 0.4901,
      "step": 4806
    },
    {
      "epoch": 0.8197476125511596,
      "grad_norm": 0.4773261845111847,
      "learning_rate": 1.6593131642773254e-07,
      "loss": 0.5003,
      "step": 4807
    },
    {
      "epoch": 0.819918144611187,
      "grad_norm": 0.37786346673965454,
      "learning_rate": 1.6562675081307103e-07,
      "loss": 0.504,
      "step": 4808
    },
    {
      "epoch": 0.8200886766712142,
      "grad_norm": 0.41945719718933105,
      "learning_rate": 1.6532243972955398e-07,
      "loss": 0.5095,
      "step": 4809
    },
    {
      "epoch": 0.8202592087312415,
      "grad_norm": 0.5154603123664856,
      "learning_rate": 1.6501838327001376e-07,
      "loss": 0.4932,
      "step": 4810
    },
    {
      "epoch": 0.8204297407912687,
      "grad_norm": 0.4044354259967804,
      "learning_rate": 1.647145815272051e-07,
      "loss": 0.491,
      "step": 4811
    },
    {
      "epoch": 0.820600272851296,
      "grad_norm": 0.5035597681999207,
      "learning_rate": 1.6441103459380394e-07,
      "loss": 0.4882,
      "step": 4812
    },
    {
      "epoch": 0.8207708049113234,
      "grad_norm": 0.4555085301399231,
      "learning_rate": 1.6410774256241036e-07,
      "loss": 0.5002,
      "step": 4813
    },
    {
      "epoch": 0.8209413369713506,
      "grad_norm": 0.3546191453933716,
      "learning_rate": 1.6380470552554484e-07,
      "loss": 0.495,
      "step": 4814
    },
    {
      "epoch": 0.8211118690313779,
      "grad_norm": 0.4193180501461029,
      "learning_rate": 1.635019235756511e-07,
      "loss": 0.4985,
      "step": 4815
    },
    {
      "epoch": 0.8212824010914052,
      "grad_norm": 0.3919258713722229,
      "learning_rate": 1.631993968050949e-07,
      "loss": 0.4978,
      "step": 4816
    },
    {
      "epoch": 0.8214529331514324,
      "grad_norm": 0.43044885993003845,
      "learning_rate": 1.628971253061642e-07,
      "loss": 0.5079,
      "step": 4817
    },
    {
      "epoch": 0.8216234652114598,
      "grad_norm": 0.3775131404399872,
      "learning_rate": 1.6259510917106847e-07,
      "loss": 0.4984,
      "step": 4818
    },
    {
      "epoch": 0.821793997271487,
      "grad_norm": 0.4048754572868347,
      "learning_rate": 1.6229334849194e-07,
      "loss": 0.4976,
      "step": 4819
    },
    {
      "epoch": 0.8219645293315143,
      "grad_norm": 0.40105101466178894,
      "learning_rate": 1.619918433608332e-07,
      "loss": 0.4888,
      "step": 4820
    },
    {
      "epoch": 0.8221350613915416,
      "grad_norm": 0.42789819836616516,
      "learning_rate": 1.6169059386972342e-07,
      "loss": 0.4954,
      "step": 4821
    },
    {
      "epoch": 0.822305593451569,
      "grad_norm": 0.43447840213775635,
      "learning_rate": 1.6138960011050976e-07,
      "loss": 0.4977,
      "step": 4822
    },
    {
      "epoch": 0.8224761255115962,
      "grad_norm": 0.5571722388267517,
      "learning_rate": 1.6108886217501185e-07,
      "loss": 0.5014,
      "step": 4823
    },
    {
      "epoch": 0.8226466575716235,
      "grad_norm": 0.3704339563846588,
      "learning_rate": 1.607883801549719e-07,
      "loss": 0.5003,
      "step": 4824
    },
    {
      "epoch": 0.8228171896316507,
      "grad_norm": 0.3826393485069275,
      "learning_rate": 1.60488154142054e-07,
      "loss": 0.4887,
      "step": 4825
    },
    {
      "epoch": 0.822987721691678,
      "grad_norm": 0.44714003801345825,
      "learning_rate": 1.6018818422784433e-07,
      "loss": 0.502,
      "step": 4826
    },
    {
      "epoch": 0.8231582537517054,
      "grad_norm": 0.4311603307723999,
      "learning_rate": 1.598884705038504e-07,
      "loss": 0.4865,
      "step": 4827
    },
    {
      "epoch": 0.8233287858117326,
      "grad_norm": 0.40986600518226624,
      "learning_rate": 1.5958901306150177e-07,
      "loss": 0.4857,
      "step": 4828
    },
    {
      "epoch": 0.8234993178717599,
      "grad_norm": 0.37676551938056946,
      "learning_rate": 1.5928981199215044e-07,
      "loss": 0.4954,
      "step": 4829
    },
    {
      "epoch": 0.8236698499317872,
      "grad_norm": 0.3872268497943878,
      "learning_rate": 1.5899086738706906e-07,
      "loss": 0.4921,
      "step": 4830
    },
    {
      "epoch": 0.8238403819918144,
      "grad_norm": 0.3704686462879181,
      "learning_rate": 1.5869217933745342e-07,
      "loss": 0.4853,
      "step": 4831
    },
    {
      "epoch": 0.8240109140518418,
      "grad_norm": 0.4022778272628784,
      "learning_rate": 1.5839374793441966e-07,
      "loss": 0.4984,
      "step": 4832
    },
    {
      "epoch": 0.824181446111869,
      "grad_norm": 0.35553985834121704,
      "learning_rate": 1.580955732690065e-07,
      "loss": 0.4937,
      "step": 4833
    },
    {
      "epoch": 0.8243519781718963,
      "grad_norm": 0.40467149019241333,
      "learning_rate": 1.5779765543217451e-07,
      "loss": 0.4971,
      "step": 4834
    },
    {
      "epoch": 0.8245225102319236,
      "grad_norm": 0.3793901205062866,
      "learning_rate": 1.5749999451480482e-07,
      "loss": 0.4892,
      "step": 4835
    },
    {
      "epoch": 0.8246930422919508,
      "grad_norm": 0.4459507167339325,
      "learning_rate": 1.5720259060770128e-07,
      "loss": 0.4964,
      "step": 4836
    },
    {
      "epoch": 0.8248635743519782,
      "grad_norm": 0.4005849361419678,
      "learning_rate": 1.5690544380158892e-07,
      "loss": 0.4913,
      "step": 4837
    },
    {
      "epoch": 0.8250341064120055,
      "grad_norm": 0.4723176658153534,
      "learning_rate": 1.5660855418711452e-07,
      "loss": 0.4925,
      "step": 4838
    },
    {
      "epoch": 0.8252046384720327,
      "grad_norm": 0.5478523969650269,
      "learning_rate": 1.5631192185484556e-07,
      "loss": 0.494,
      "step": 4839
    },
    {
      "epoch": 0.82537517053206,
      "grad_norm": 0.4246988594532013,
      "learning_rate": 1.5601554689527278e-07,
      "loss": 0.4973,
      "step": 4840
    },
    {
      "epoch": 0.8255457025920873,
      "grad_norm": 0.48962587118148804,
      "learning_rate": 1.5571942939880647e-07,
      "loss": 0.4982,
      "step": 4841
    },
    {
      "epoch": 0.8257162346521146,
      "grad_norm": 0.5545381903648376,
      "learning_rate": 1.5542356945577964e-07,
      "loss": 0.4835,
      "step": 4842
    },
    {
      "epoch": 0.8258867667121419,
      "grad_norm": 0.524614155292511,
      "learning_rate": 1.5512796715644643e-07,
      "loss": 0.4924,
      "step": 4843
    },
    {
      "epoch": 0.8260572987721692,
      "grad_norm": 0.540654182434082,
      "learning_rate": 1.5483262259098198e-07,
      "loss": 0.4818,
      "step": 4844
    },
    {
      "epoch": 0.8262278308321964,
      "grad_norm": 0.4969771206378937,
      "learning_rate": 1.545375358494833e-07,
      "loss": 0.4965,
      "step": 4845
    },
    {
      "epoch": 0.8263983628922238,
      "grad_norm": 0.4746538996696472,
      "learning_rate": 1.5424270702196844e-07,
      "loss": 0.4932,
      "step": 4846
    },
    {
      "epoch": 0.826568894952251,
      "grad_norm": 0.6185543537139893,
      "learning_rate": 1.5394813619837735e-07,
      "loss": 0.491,
      "step": 4847
    },
    {
      "epoch": 0.8267394270122783,
      "grad_norm": 0.44103044271469116,
      "learning_rate": 1.5365382346857008e-07,
      "loss": 0.4888,
      "step": 4848
    },
    {
      "epoch": 0.8269099590723056,
      "grad_norm": 0.40380629897117615,
      "learning_rate": 1.5335976892232951e-07,
      "loss": 0.5005,
      "step": 4849
    },
    {
      "epoch": 0.8270804911323328,
      "grad_norm": 0.4558597505092621,
      "learning_rate": 1.5306597264935865e-07,
      "loss": 0.4799,
      "step": 4850
    },
    {
      "epoch": 0.8272510231923602,
      "grad_norm": 0.4192386567592621,
      "learning_rate": 1.5277243473928151e-07,
      "loss": 0.4889,
      "step": 4851
    },
    {
      "epoch": 0.8274215552523875,
      "grad_norm": 0.420249879360199,
      "learning_rate": 1.524791552816446e-07,
      "loss": 0.4884,
      "step": 4852
    },
    {
      "epoch": 0.8275920873124147,
      "grad_norm": 0.6081094741821289,
      "learning_rate": 1.5218613436591423e-07,
      "loss": 0.4998,
      "step": 4853
    },
    {
      "epoch": 0.827762619372442,
      "grad_norm": 0.3973240852355957,
      "learning_rate": 1.5189337208147848e-07,
      "loss": 0.4903,
      "step": 4854
    },
    {
      "epoch": 0.8279331514324693,
      "grad_norm": 0.5748963356018066,
      "learning_rate": 1.5160086851764667e-07,
      "loss": 0.4918,
      "step": 4855
    },
    {
      "epoch": 0.8281036834924966,
      "grad_norm": 0.656119167804718,
      "learning_rate": 1.51308623763649e-07,
      "loss": 0.4913,
      "step": 4856
    },
    {
      "epoch": 0.8282742155525239,
      "grad_norm": 0.3806758224964142,
      "learning_rate": 1.5101663790863597e-07,
      "loss": 0.5002,
      "step": 4857
    },
    {
      "epoch": 0.8284447476125512,
      "grad_norm": 0.5381845235824585,
      "learning_rate": 1.5072491104168097e-07,
      "loss": 0.496,
      "step": 4858
    },
    {
      "epoch": 0.8286152796725784,
      "grad_norm": 0.34899216890335083,
      "learning_rate": 1.5043344325177633e-07,
      "loss": 0.489,
      "step": 4859
    },
    {
      "epoch": 0.8287858117326057,
      "grad_norm": 0.6310024261474609,
      "learning_rate": 1.5014223462783651e-07,
      "loss": 0.4919,
      "step": 4860
    },
    {
      "epoch": 0.828956343792633,
      "grad_norm": 0.5121638774871826,
      "learning_rate": 1.4985128525869698e-07,
      "loss": 0.4954,
      "step": 4861
    },
    {
      "epoch": 0.8291268758526603,
      "grad_norm": 0.48706522583961487,
      "learning_rate": 1.495605952331133e-07,
      "loss": 0.4951,
      "step": 4862
    },
    {
      "epoch": 0.8292974079126876,
      "grad_norm": 0.5409660339355469,
      "learning_rate": 1.4927016463976263e-07,
      "loss": 0.4893,
      "step": 4863
    },
    {
      "epoch": 0.8294679399727148,
      "grad_norm": 0.5660123825073242,
      "learning_rate": 1.489799935672427e-07,
      "loss": 0.5016,
      "step": 4864
    },
    {
      "epoch": 0.8296384720327421,
      "grad_norm": 0.6146024465560913,
      "learning_rate": 1.4869008210407245e-07,
      "loss": 0.482,
      "step": 4865
    },
    {
      "epoch": 0.8298090040927695,
      "grad_norm": 0.4160356819629669,
      "learning_rate": 1.4840043033869079e-07,
      "loss": 0.4966,
      "step": 4866
    },
    {
      "epoch": 0.8299795361527967,
      "grad_norm": 0.600299060344696,
      "learning_rate": 1.4811103835945818e-07,
      "loss": 0.4952,
      "step": 4867
    },
    {
      "epoch": 0.830150068212824,
      "grad_norm": 0.3432973623275757,
      "learning_rate": 1.4782190625465565e-07,
      "loss": 0.4979,
      "step": 4868
    },
    {
      "epoch": 0.8303206002728513,
      "grad_norm": 0.5366365313529968,
      "learning_rate": 1.4753303411248478e-07,
      "loss": 0.4836,
      "step": 4869
    },
    {
      "epoch": 0.8304911323328786,
      "grad_norm": 0.4105454683303833,
      "learning_rate": 1.4724442202106826e-07,
      "loss": 0.5035,
      "step": 4870
    },
    {
      "epoch": 0.8306616643929059,
      "grad_norm": 0.5173560976982117,
      "learning_rate": 1.469560700684486e-07,
      "loss": 0.4944,
      "step": 4871
    },
    {
      "epoch": 0.8308321964529332,
      "grad_norm": 0.5018391013145447,
      "learning_rate": 1.4666797834258983e-07,
      "loss": 0.5,
      "step": 4872
    },
    {
      "epoch": 0.8310027285129604,
      "grad_norm": 0.5189551711082458,
      "learning_rate": 1.4638014693137638e-07,
      "loss": 0.4897,
      "step": 4873
    },
    {
      "epoch": 0.8311732605729877,
      "grad_norm": 0.54872727394104,
      "learning_rate": 1.4609257592261275e-07,
      "loss": 0.4917,
      "step": 4874
    },
    {
      "epoch": 0.831343792633015,
      "grad_norm": 0.4522056579589844,
      "learning_rate": 1.4580526540402464e-07,
      "loss": 0.5072,
      "step": 4875
    },
    {
      "epoch": 0.8315143246930423,
      "grad_norm": 0.4563232362270355,
      "learning_rate": 1.4551821546325807e-07,
      "loss": 0.4927,
      "step": 4876
    },
    {
      "epoch": 0.8316848567530696,
      "grad_norm": 0.4634324908256531,
      "learning_rate": 1.4523142618787945e-07,
      "loss": 0.4882,
      "step": 4877
    },
    {
      "epoch": 0.8318553888130968,
      "grad_norm": 0.5055513381958008,
      "learning_rate": 1.449448976653758e-07,
      "loss": 0.4947,
      "step": 4878
    },
    {
      "epoch": 0.8320259208731241,
      "grad_norm": 0.6747018098831177,
      "learning_rate": 1.4465862998315485e-07,
      "loss": 0.5035,
      "step": 4879
    },
    {
      "epoch": 0.8321964529331515,
      "grad_norm": 0.4617733657360077,
      "learning_rate": 1.4437262322854397e-07,
      "loss": 0.4965,
      "step": 4880
    },
    {
      "epoch": 0.8323669849931787,
      "grad_norm": 0.4791738986968994,
      "learning_rate": 1.440868774887916e-07,
      "loss": 0.5042,
      "step": 4881
    },
    {
      "epoch": 0.832537517053206,
      "grad_norm": 0.43876227736473083,
      "learning_rate": 1.4380139285106678e-07,
      "loss": 0.5,
      "step": 4882
    },
    {
      "epoch": 0.8327080491132333,
      "grad_norm": 0.4775140881538391,
      "learning_rate": 1.4351616940245794e-07,
      "loss": 0.4883,
      "step": 4883
    },
    {
      "epoch": 0.8328785811732605,
      "grad_norm": 0.538390576839447,
      "learning_rate": 1.4323120722997464e-07,
      "loss": 0.4958,
      "step": 4884
    },
    {
      "epoch": 0.8330491132332879,
      "grad_norm": 0.44281888008117676,
      "learning_rate": 1.4294650642054654e-07,
      "loss": 0.4876,
      "step": 4885
    },
    {
      "epoch": 0.8332196452933152,
      "grad_norm": 0.5450493693351746,
      "learning_rate": 1.426620670610234e-07,
      "loss": 0.5038,
      "step": 4886
    },
    {
      "epoch": 0.8333901773533424,
      "grad_norm": 0.4807684123516083,
      "learning_rate": 1.4237788923817553e-07,
      "loss": 0.4868,
      "step": 4887
    },
    {
      "epoch": 0.8335607094133697,
      "grad_norm": 0.5991215705871582,
      "learning_rate": 1.4209397303869345e-07,
      "loss": 0.5017,
      "step": 4888
    },
    {
      "epoch": 0.833731241473397,
      "grad_norm": 0.4470359981060028,
      "learning_rate": 1.4181031854918716e-07,
      "loss": 0.4803,
      "step": 4889
    },
    {
      "epoch": 0.8339017735334243,
      "grad_norm": 0.6047620177268982,
      "learning_rate": 1.4152692585618767e-07,
      "loss": 0.4971,
      "step": 4890
    },
    {
      "epoch": 0.8340723055934516,
      "grad_norm": 0.575122594833374,
      "learning_rate": 1.4124379504614605e-07,
      "loss": 0.5003,
      "step": 4891
    },
    {
      "epoch": 0.8342428376534788,
      "grad_norm": 0.38666024804115295,
      "learning_rate": 1.4096092620543258e-07,
      "loss": 0.5019,
      "step": 4892
    },
    {
      "epoch": 0.8344133697135061,
      "grad_norm": 0.6803966164588928,
      "learning_rate": 1.4067831942033907e-07,
      "loss": 0.5025,
      "step": 4893
    },
    {
      "epoch": 0.8345839017735335,
      "grad_norm": 0.5980350971221924,
      "learning_rate": 1.4039597477707601e-07,
      "loss": 0.491,
      "step": 4894
    },
    {
      "epoch": 0.8347544338335607,
      "grad_norm": 0.44976601004600525,
      "learning_rate": 1.4011389236177477e-07,
      "loss": 0.5025,
      "step": 4895
    },
    {
      "epoch": 0.834924965893588,
      "grad_norm": 0.5966007709503174,
      "learning_rate": 1.3983207226048656e-07,
      "loss": 0.4937,
      "step": 4896
    },
    {
      "epoch": 0.8350954979536153,
      "grad_norm": 0.3893631398677826,
      "learning_rate": 1.3955051455918258e-07,
      "loss": 0.4935,
      "step": 4897
    },
    {
      "epoch": 0.8352660300136425,
      "grad_norm": 0.5327594876289368,
      "learning_rate": 1.392692193437534e-07,
      "loss": 0.5043,
      "step": 4898
    },
    {
      "epoch": 0.8354365620736699,
      "grad_norm": 0.506881833076477,
      "learning_rate": 1.3898818670001035e-07,
      "loss": 0.4896,
      "step": 4899
    },
    {
      "epoch": 0.8356070941336972,
      "grad_norm": 0.4647798538208008,
      "learning_rate": 1.387074167136846e-07,
      "loss": 0.4965,
      "step": 4900
    },
    {
      "epoch": 0.8357776261937244,
      "grad_norm": 0.4127386212348938,
      "learning_rate": 1.3842690947042605e-07,
      "loss": 0.5101,
      "step": 4901
    },
    {
      "epoch": 0.8359481582537517,
      "grad_norm": 0.4460466802120209,
      "learning_rate": 1.3814666505580632e-07,
      "loss": 0.4976,
      "step": 4902
    },
    {
      "epoch": 0.836118690313779,
      "grad_norm": 0.38933101296424866,
      "learning_rate": 1.3786668355531519e-07,
      "loss": 0.4986,
      "step": 4903
    },
    {
      "epoch": 0.8362892223738063,
      "grad_norm": 0.4646453559398651,
      "learning_rate": 1.375869650543631e-07,
      "loss": 0.4998,
      "step": 4904
    },
    {
      "epoch": 0.8364597544338336,
      "grad_norm": 0.4589424729347229,
      "learning_rate": 1.3730750963828035e-07,
      "loss": 0.4937,
      "step": 4905
    },
    {
      "epoch": 0.8366302864938608,
      "grad_norm": 0.3910622298717499,
      "learning_rate": 1.3702831739231595e-07,
      "loss": 0.4952,
      "step": 4906
    },
    {
      "epoch": 0.8368008185538881,
      "grad_norm": 0.6233266592025757,
      "learning_rate": 1.3674938840163987e-07,
      "loss": 0.4961,
      "step": 4907
    },
    {
      "epoch": 0.8369713506139154,
      "grad_norm": 0.46709907054901123,
      "learning_rate": 1.3647072275134123e-07,
      "loss": 0.4956,
      "step": 4908
    },
    {
      "epoch": 0.8371418826739427,
      "grad_norm": 0.4565236270427704,
      "learning_rate": 1.3619232052642907e-07,
      "loss": 0.4948,
      "step": 4909
    },
    {
      "epoch": 0.83731241473397,
      "grad_norm": 0.42009490728378296,
      "learning_rate": 1.3591418181183103e-07,
      "loss": 0.4913,
      "step": 4910
    },
    {
      "epoch": 0.8374829467939973,
      "grad_norm": 0.5678643584251404,
      "learning_rate": 1.3563630669239627e-07,
      "loss": 0.4978,
      "step": 4911
    },
    {
      "epoch": 0.8376534788540245,
      "grad_norm": 0.49767374992370605,
      "learning_rate": 1.3535869525289182e-07,
      "loss": 0.4909,
      "step": 4912
    },
    {
      "epoch": 0.8378240109140518,
      "grad_norm": 0.46664154529571533,
      "learning_rate": 1.3508134757800462e-07,
      "loss": 0.4893,
      "step": 4913
    },
    {
      "epoch": 0.8379945429740792,
      "grad_norm": 0.4974210560321808,
      "learning_rate": 1.3480426375234236e-07,
      "loss": 0.5035,
      "step": 4914
    },
    {
      "epoch": 0.8381650750341064,
      "grad_norm": 0.4552690386772156,
      "learning_rate": 1.3452744386043048e-07,
      "loss": 0.4987,
      "step": 4915
    },
    {
      "epoch": 0.8383356070941337,
      "grad_norm": 0.39272937178611755,
      "learning_rate": 1.3425088798671505e-07,
      "loss": 0.5001,
      "step": 4916
    },
    {
      "epoch": 0.838506139154161,
      "grad_norm": 0.5397828817367554,
      "learning_rate": 1.339745962155613e-07,
      "loss": 0.4963,
      "step": 4917
    },
    {
      "epoch": 0.8386766712141883,
      "grad_norm": 0.42156437039375305,
      "learning_rate": 1.336985686312542e-07,
      "loss": 0.5069,
      "step": 4918
    },
    {
      "epoch": 0.8388472032742156,
      "grad_norm": 0.42294806241989136,
      "learning_rate": 1.3342280531799692e-07,
      "loss": 0.4903,
      "step": 4919
    },
    {
      "epoch": 0.8390177353342428,
      "grad_norm": 0.3916904926300049,
      "learning_rate": 1.3314730635991393e-07,
      "loss": 0.4952,
      "step": 4920
    },
    {
      "epoch": 0.8391882673942701,
      "grad_norm": 0.536283552646637,
      "learning_rate": 1.3287207184104757e-07,
      "loss": 0.5039,
      "step": 4921
    },
    {
      "epoch": 0.8393587994542974,
      "grad_norm": 0.386033296585083,
      "learning_rate": 1.3259710184535958e-07,
      "loss": 0.4926,
      "step": 4922
    },
    {
      "epoch": 0.8395293315143247,
      "grad_norm": 0.5002760887145996,
      "learning_rate": 1.3232239645673217e-07,
      "loss": 0.4967,
      "step": 4923
    },
    {
      "epoch": 0.839699863574352,
      "grad_norm": 0.43488916754722595,
      "learning_rate": 1.3204795575896546e-07,
      "loss": 0.5035,
      "step": 4924
    },
    {
      "epoch": 0.8398703956343793,
      "grad_norm": 0.45359697937965393,
      "learning_rate": 1.3177377983577966e-07,
      "loss": 0.4934,
      "step": 4925
    },
    {
      "epoch": 0.8400409276944065,
      "grad_norm": 0.3645251989364624,
      "learning_rate": 1.3149986877081401e-07,
      "loss": 0.4916,
      "step": 4926
    },
    {
      "epoch": 0.8402114597544338,
      "grad_norm": 0.3643120229244232,
      "learning_rate": 1.31226222647627e-07,
      "loss": 0.5052,
      "step": 4927
    },
    {
      "epoch": 0.8403819918144612,
      "grad_norm": 0.47196370363235474,
      "learning_rate": 1.309528415496958e-07,
      "loss": 0.4885,
      "step": 4928
    },
    {
      "epoch": 0.8405525238744884,
      "grad_norm": 0.6321125626564026,
      "learning_rate": 1.3067972556041752e-07,
      "loss": 0.4935,
      "step": 4929
    },
    {
      "epoch": 0.8407230559345157,
      "grad_norm": 0.5271437764167786,
      "learning_rate": 1.304068747631081e-07,
      "loss": 0.5041,
      "step": 4930
    },
    {
      "epoch": 0.840893587994543,
      "grad_norm": 0.4212459921836853,
      "learning_rate": 1.3013428924100182e-07,
      "loss": 0.4928,
      "step": 4931
    },
    {
      "epoch": 0.8410641200545702,
      "grad_norm": 0.649498462677002,
      "learning_rate": 1.298619690772537e-07,
      "loss": 0.4894,
      "step": 4932
    },
    {
      "epoch": 0.8412346521145976,
      "grad_norm": 0.44042500853538513,
      "learning_rate": 1.2958991435493618e-07,
      "loss": 0.4901,
      "step": 4933
    },
    {
      "epoch": 0.8414051841746248,
      "grad_norm": 0.5599384903907776,
      "learning_rate": 1.2931812515704147e-07,
      "loss": 0.5038,
      "step": 4934
    },
    {
      "epoch": 0.8415757162346521,
      "grad_norm": 0.644747793674469,
      "learning_rate": 1.2904660156648075e-07,
      "loss": 0.4948,
      "step": 4935
    },
    {
      "epoch": 0.8417462482946794,
      "grad_norm": 0.3929704427719116,
      "learning_rate": 1.2877534366608425e-07,
      "loss": 0.4891,
      "step": 4936
    },
    {
      "epoch": 0.8419167803547067,
      "grad_norm": 0.556605875492096,
      "learning_rate": 1.2850435153860075e-07,
      "loss": 0.4917,
      "step": 4937
    },
    {
      "epoch": 0.842087312414734,
      "grad_norm": 0.6048465371131897,
      "learning_rate": 1.2823362526669825e-07,
      "loss": 0.4906,
      "step": 4938
    },
    {
      "epoch": 0.8422578444747613,
      "grad_norm": 0.44814008474349976,
      "learning_rate": 1.2796316493296402e-07,
      "loss": 0.5011,
      "step": 4939
    },
    {
      "epoch": 0.8424283765347885,
      "grad_norm": 0.49208498001098633,
      "learning_rate": 1.2769297061990294e-07,
      "loss": 0.4888,
      "step": 4940
    },
    {
      "epoch": 0.8425989085948158,
      "grad_norm": 0.529579222202301,
      "learning_rate": 1.2742304240994053e-07,
      "loss": 0.5029,
      "step": 4941
    },
    {
      "epoch": 0.8427694406548432,
      "grad_norm": 0.3871929943561554,
      "learning_rate": 1.271533803854197e-07,
      "loss": 0.4873,
      "step": 4942
    },
    {
      "epoch": 0.8429399727148704,
      "grad_norm": 0.5285574793815613,
      "learning_rate": 1.2688398462860264e-07,
      "loss": 0.5017,
      "step": 4943
    },
    {
      "epoch": 0.8431105047748977,
      "grad_norm": 0.3910968601703644,
      "learning_rate": 1.2661485522167062e-07,
      "loss": 0.4916,
      "step": 4944
    },
    {
      "epoch": 0.843281036834925,
      "grad_norm": 0.43763166666030884,
      "learning_rate": 1.2634599224672297e-07,
      "loss": 0.4985,
      "step": 4945
    },
    {
      "epoch": 0.8434515688949522,
      "grad_norm": 0.4757555425167084,
      "learning_rate": 1.2607739578577838e-07,
      "loss": 0.4939,
      "step": 4946
    },
    {
      "epoch": 0.8436221009549796,
      "grad_norm": 0.3990534842014313,
      "learning_rate": 1.2580906592077403e-07,
      "loss": 0.5001,
      "step": 4947
    },
    {
      "epoch": 0.8437926330150068,
      "grad_norm": 0.5799641609191895,
      "learning_rate": 1.2554100273356553e-07,
      "loss": 0.494,
      "step": 4948
    },
    {
      "epoch": 0.8439631650750341,
      "grad_norm": 0.4546438455581665,
      "learning_rate": 1.2527320630592767e-07,
      "loss": 0.4943,
      "step": 4949
    },
    {
      "epoch": 0.8441336971350614,
      "grad_norm": 0.451763778924942,
      "learning_rate": 1.2500567671955345e-07,
      "loss": 0.491,
      "step": 4950
    },
    {
      "epoch": 0.8443042291950886,
      "grad_norm": 0.45567867159843445,
      "learning_rate": 1.2473841405605427e-07,
      "loss": 0.4855,
      "step": 4951
    },
    {
      "epoch": 0.844474761255116,
      "grad_norm": 0.4821739196777344,
      "learning_rate": 1.2447141839696065e-07,
      "loss": 0.4906,
      "step": 4952
    },
    {
      "epoch": 0.8446452933151433,
      "grad_norm": 0.4332144558429718,
      "learning_rate": 1.242046898237216e-07,
      "loss": 0.4902,
      "step": 4953
    },
    {
      "epoch": 0.8448158253751705,
      "grad_norm": 0.3891392946243286,
      "learning_rate": 1.23938228417704e-07,
      "loss": 0.4948,
      "step": 4954
    },
    {
      "epoch": 0.8449863574351978,
      "grad_norm": 0.4128452241420746,
      "learning_rate": 1.2367203426019382e-07,
      "loss": 0.4895,
      "step": 4955
    },
    {
      "epoch": 0.845156889495225,
      "grad_norm": 0.5201817750930786,
      "learning_rate": 1.2340610743239545e-07,
      "loss": 0.4834,
      "step": 4956
    },
    {
      "epoch": 0.8453274215552524,
      "grad_norm": 0.43655940890312195,
      "learning_rate": 1.2314044801543168e-07,
      "loss": 0.4901,
      "step": 4957
    },
    {
      "epoch": 0.8454979536152797,
      "grad_norm": 0.39609473943710327,
      "learning_rate": 1.2287505609034367e-07,
      "loss": 0.4887,
      "step": 4958
    },
    {
      "epoch": 0.845668485675307,
      "grad_norm": 0.43933528661727905,
      "learning_rate": 1.2260993173809122e-07,
      "loss": 0.4963,
      "step": 4959
    },
    {
      "epoch": 0.8458390177353342,
      "grad_norm": 0.32647085189819336,
      "learning_rate": 1.2234507503955184e-07,
      "loss": 0.4835,
      "step": 4960
    },
    {
      "epoch": 0.8460095497953616,
      "grad_norm": 0.40228471159935,
      "learning_rate": 1.2208048607552215e-07,
      "loss": 0.4915,
      "step": 4961
    },
    {
      "epoch": 0.8461800818553888,
      "grad_norm": 0.5278027057647705,
      "learning_rate": 1.2181616492671677e-07,
      "loss": 0.4981,
      "step": 4962
    },
    {
      "epoch": 0.8463506139154161,
      "grad_norm": 0.4137642979621887,
      "learning_rate": 1.2155211167376837e-07,
      "loss": 0.5049,
      "step": 4963
    },
    {
      "epoch": 0.8465211459754434,
      "grad_norm": 0.5432256460189819,
      "learning_rate": 1.2128832639722845e-07,
      "loss": 0.4998,
      "step": 4964
    },
    {
      "epoch": 0.8466916780354706,
      "grad_norm": 0.39955243468284607,
      "learning_rate": 1.2102480917756633e-07,
      "loss": 0.497,
      "step": 4965
    },
    {
      "epoch": 0.846862210095498,
      "grad_norm": 0.4583352506160736,
      "learning_rate": 1.207615600951697e-07,
      "loss": 0.4955,
      "step": 4966
    },
    {
      "epoch": 0.8470327421555253,
      "grad_norm": 0.43716612458229065,
      "learning_rate": 1.2049857923034471e-07,
      "loss": 0.5037,
      "step": 4967
    },
    {
      "epoch": 0.8472032742155525,
      "grad_norm": 0.4870067536830902,
      "learning_rate": 1.202358666633151e-07,
      "loss": 0.5039,
      "step": 4968
    },
    {
      "epoch": 0.8473738062755798,
      "grad_norm": 0.46189653873443604,
      "learning_rate": 1.1997342247422318e-07,
      "loss": 0.4964,
      "step": 4969
    },
    {
      "epoch": 0.847544338335607,
      "grad_norm": 0.45632949471473694,
      "learning_rate": 1.197112467431294e-07,
      "loss": 0.4977,
      "step": 4970
    },
    {
      "epoch": 0.8477148703956344,
      "grad_norm": 0.4396877884864807,
      "learning_rate": 1.1944933955001225e-07,
      "loss": 0.4983,
      "step": 4971
    },
    {
      "epoch": 0.8478854024556617,
      "grad_norm": 0.4021702706813812,
      "learning_rate": 1.1918770097476818e-07,
      "loss": 0.4906,
      "step": 4972
    },
    {
      "epoch": 0.848055934515689,
      "grad_norm": 0.47800371050834656,
      "learning_rate": 1.1892633109721164e-07,
      "loss": 0.5015,
      "step": 4973
    },
    {
      "epoch": 0.8482264665757162,
      "grad_norm": 0.5028043389320374,
      "learning_rate": 1.1866522999707554e-07,
      "loss": 0.4934,
      "step": 4974
    },
    {
      "epoch": 0.8483969986357435,
      "grad_norm": 0.4028972089290619,
      "learning_rate": 1.1840439775401026e-07,
      "loss": 0.5008,
      "step": 4975
    },
    {
      "epoch": 0.8485675306957708,
      "grad_norm": 0.5216171145439148,
      "learning_rate": 1.1814383444758494e-07,
      "loss": 0.4799,
      "step": 4976
    },
    {
      "epoch": 0.8487380627557981,
      "grad_norm": 0.48240017890930176,
      "learning_rate": 1.1788354015728544e-07,
      "loss": 0.4938,
      "step": 4977
    },
    {
      "epoch": 0.8489085948158254,
      "grad_norm": 0.5118283033370972,
      "learning_rate": 1.1762351496251668e-07,
      "loss": 0.495,
      "step": 4978
    },
    {
      "epoch": 0.8490791268758526,
      "grad_norm": 0.4451368451118469,
      "learning_rate": 1.1736375894260111e-07,
      "loss": 0.4988,
      "step": 4979
    },
    {
      "epoch": 0.8492496589358799,
      "grad_norm": 0.5162202715873718,
      "learning_rate": 1.1710427217677912e-07,
      "loss": 0.4815,
      "step": 4980
    },
    {
      "epoch": 0.8494201909959073,
      "grad_norm": 0.5497837662696838,
      "learning_rate": 1.1684505474420828e-07,
      "loss": 0.4941,
      "step": 4981
    },
    {
      "epoch": 0.8495907230559345,
      "grad_norm": 0.44187721610069275,
      "learning_rate": 1.1658610672396544e-07,
      "loss": 0.4935,
      "step": 4982
    },
    {
      "epoch": 0.8497612551159618,
      "grad_norm": 0.5414014458656311,
      "learning_rate": 1.1632742819504406e-07,
      "loss": 0.4931,
      "step": 4983
    },
    {
      "epoch": 0.849931787175989,
      "grad_norm": 0.5177868604660034,
      "learning_rate": 1.1606901923635529e-07,
      "loss": 0.4961,
      "step": 4984
    },
    {
      "epoch": 0.8501023192360164,
      "grad_norm": 0.5770236849784851,
      "learning_rate": 1.1581087992672935e-07,
      "loss": 0.4944,
      "step": 4985
    },
    {
      "epoch": 0.8502728512960437,
      "grad_norm": 0.6375470757484436,
      "learning_rate": 1.1555301034491285e-07,
      "loss": 0.4998,
      "step": 4986
    },
    {
      "epoch": 0.850443383356071,
      "grad_norm": 0.3887059986591339,
      "learning_rate": 1.1529541056957073e-07,
      "loss": 0.4911,
      "step": 4987
    },
    {
      "epoch": 0.8506139154160982,
      "grad_norm": 0.565872848033905,
      "learning_rate": 1.1503808067928558e-07,
      "loss": 0.4981,
      "step": 4988
    },
    {
      "epoch": 0.8507844474761255,
      "grad_norm": 0.3640064001083374,
      "learning_rate": 1.1478102075255772e-07,
      "loss": 0.4947,
      "step": 4989
    },
    {
      "epoch": 0.8509549795361528,
      "grad_norm": 0.5178152918815613,
      "learning_rate": 1.1452423086780463e-07,
      "loss": 0.4861,
      "step": 4990
    },
    {
      "epoch": 0.8511255115961801,
      "grad_norm": 0.5512751936912537,
      "learning_rate": 1.14267711103362e-07,
      "loss": 0.5017,
      "step": 4991
    },
    {
      "epoch": 0.8512960436562074,
      "grad_norm": 0.4500180184841156,
      "learning_rate": 1.1401146153748311e-07,
      "loss": 0.4899,
      "step": 4992
    },
    {
      "epoch": 0.8514665757162346,
      "grad_norm": 0.4403330385684967,
      "learning_rate": 1.1375548224833811e-07,
      "loss": 0.5108,
      "step": 4993
    },
    {
      "epoch": 0.8516371077762619,
      "grad_norm": 0.36600425839424133,
      "learning_rate": 1.1349977331401575e-07,
      "loss": 0.5064,
      "step": 4994
    },
    {
      "epoch": 0.8518076398362893,
      "grad_norm": 0.3654356002807617,
      "learning_rate": 1.1324433481252141e-07,
      "loss": 0.5049,
      "step": 4995
    },
    {
      "epoch": 0.8519781718963165,
      "grad_norm": 0.36991676688194275,
      "learning_rate": 1.129891668217783e-07,
      "loss": 0.4946,
      "step": 4996
    },
    {
      "epoch": 0.8521487039563438,
      "grad_norm": 0.40469640493392944,
      "learning_rate": 1.127342694196273e-07,
      "loss": 0.4958,
      "step": 4997
    },
    {
      "epoch": 0.852319236016371,
      "grad_norm": 0.4282923936843872,
      "learning_rate": 1.1247964268382661e-07,
      "loss": 0.4859,
      "step": 4998
    },
    {
      "epoch": 0.8524897680763983,
      "grad_norm": 0.46982812881469727,
      "learning_rate": 1.1222528669205159e-07,
      "loss": 0.5041,
      "step": 4999
    },
    {
      "epoch": 0.8526603001364257,
      "grad_norm": 0.36918675899505615,
      "learning_rate": 1.1197120152189523e-07,
      "loss": 0.4915,
      "step": 5000
    },
    {
      "epoch": 0.852830832196453,
      "grad_norm": 0.49565747380256653,
      "learning_rate": 1.1171738725086835e-07,
      "loss": 0.492,
      "step": 5001
    },
    {
      "epoch": 0.8530013642564802,
      "grad_norm": 0.41457846760749817,
      "learning_rate": 1.1146384395639788e-07,
      "loss": 0.4898,
      "step": 5002
    },
    {
      "epoch": 0.8531718963165075,
      "grad_norm": 0.45303431153297424,
      "learning_rate": 1.1121057171582984e-07,
      "loss": 0.495,
      "step": 5003
    },
    {
      "epoch": 0.8533424283765347,
      "grad_norm": 0.46331140398979187,
      "learning_rate": 1.1095757060642588e-07,
      "loss": 0.5022,
      "step": 5004
    },
    {
      "epoch": 0.8535129604365621,
      "grad_norm": 0.518911600112915,
      "learning_rate": 1.1070484070536614e-07,
      "loss": 0.4841,
      "step": 5005
    },
    {
      "epoch": 0.8536834924965894,
      "grad_norm": 0.5267069339752197,
      "learning_rate": 1.1045238208974741e-07,
      "loss": 0.4957,
      "step": 5006
    },
    {
      "epoch": 0.8538540245566166,
      "grad_norm": 0.46358224749565125,
      "learning_rate": 1.1020019483658385e-07,
      "loss": 0.4886,
      "step": 5007
    },
    {
      "epoch": 0.8540245566166439,
      "grad_norm": 0.45485424995422363,
      "learning_rate": 1.0994827902280682e-07,
      "loss": 0.4955,
      "step": 5008
    },
    {
      "epoch": 0.8541950886766713,
      "grad_norm": 0.5479266047477722,
      "learning_rate": 1.0969663472526494e-07,
      "loss": 0.498,
      "step": 5009
    },
    {
      "epoch": 0.8543656207366985,
      "grad_norm": 0.5160921216011047,
      "learning_rate": 1.0944526202072426e-07,
      "loss": 0.4912,
      "step": 5010
    },
    {
      "epoch": 0.8545361527967258,
      "grad_norm": 0.39966824650764465,
      "learning_rate": 1.0919416098586712e-07,
      "loss": 0.491,
      "step": 5011
    },
    {
      "epoch": 0.854706684856753,
      "grad_norm": 0.45781365036964417,
      "learning_rate": 1.0894333169729433e-07,
      "loss": 0.4895,
      "step": 5012
    },
    {
      "epoch": 0.8548772169167803,
      "grad_norm": 0.36039382219314575,
      "learning_rate": 1.0869277423152247e-07,
      "loss": 0.4888,
      "step": 5013
    },
    {
      "epoch": 0.8550477489768077,
      "grad_norm": 0.5843361616134644,
      "learning_rate": 1.0844248866498598e-07,
      "loss": 0.49,
      "step": 5014
    },
    {
      "epoch": 0.855218281036835,
      "grad_norm": 0.5285351276397705,
      "learning_rate": 1.081924750740363e-07,
      "loss": 0.4888,
      "step": 5015
    },
    {
      "epoch": 0.8553888130968622,
      "grad_norm": 0.44772446155548096,
      "learning_rate": 1.0794273353494146e-07,
      "loss": 0.4812,
      "step": 5016
    },
    {
      "epoch": 0.8555593451568895,
      "grad_norm": 0.6695955991744995,
      "learning_rate": 1.0769326412388682e-07,
      "loss": 0.4976,
      "step": 5017
    },
    {
      "epoch": 0.8557298772169167,
      "grad_norm": 0.539442777633667,
      "learning_rate": 1.0744406691697474e-07,
      "loss": 0.4828,
      "step": 5018
    },
    {
      "epoch": 0.8559004092769441,
      "grad_norm": 0.5010582208633423,
      "learning_rate": 1.0719514199022475e-07,
      "loss": 0.4885,
      "step": 5019
    },
    {
      "epoch": 0.8560709413369714,
      "grad_norm": 0.4613291621208191,
      "learning_rate": 1.0694648941957243e-07,
      "loss": 0.4891,
      "step": 5020
    },
    {
      "epoch": 0.8562414733969986,
      "grad_norm": 0.4389791786670685,
      "learning_rate": 1.0669810928087178e-07,
      "loss": 0.4909,
      "step": 5021
    },
    {
      "epoch": 0.8564120054570259,
      "grad_norm": 0.5779827833175659,
      "learning_rate": 1.0645000164989218e-07,
      "loss": 0.4858,
      "step": 5022
    },
    {
      "epoch": 0.8565825375170532,
      "grad_norm": 0.46866342425346375,
      "learning_rate": 1.0620216660232075e-07,
      "loss": 0.4909,
      "step": 5023
    },
    {
      "epoch": 0.8567530695770805,
      "grad_norm": 0.4902592897415161,
      "learning_rate": 1.0595460421376126e-07,
      "loss": 0.4887,
      "step": 5024
    },
    {
      "epoch": 0.8569236016371078,
      "grad_norm": 0.5032135248184204,
      "learning_rate": 1.0570731455973416e-07,
      "loss": 0.501,
      "step": 5025
    },
    {
      "epoch": 0.857094133697135,
      "grad_norm": 0.4387756884098053,
      "learning_rate": 1.054602977156768e-07,
      "loss": 0.4862,
      "step": 5026
    },
    {
      "epoch": 0.8572646657571623,
      "grad_norm": 0.40801355242729187,
      "learning_rate": 1.0521355375694332e-07,
      "loss": 0.5032,
      "step": 5027
    },
    {
      "epoch": 0.8574351978171897,
      "grad_norm": 0.4306230843067169,
      "learning_rate": 1.0496708275880497e-07,
      "loss": 0.4903,
      "step": 5028
    },
    {
      "epoch": 0.857605729877217,
      "grad_norm": 0.3730475604534149,
      "learning_rate": 1.0472088479644883e-07,
      "loss": 0.4958,
      "step": 5029
    },
    {
      "epoch": 0.8577762619372442,
      "grad_norm": 0.42054077982902527,
      "learning_rate": 1.0447495994497959e-07,
      "loss": 0.4945,
      "step": 5030
    },
    {
      "epoch": 0.8579467939972715,
      "grad_norm": 0.4239777624607086,
      "learning_rate": 1.0422930827941803e-07,
      "loss": 0.4994,
      "step": 5031
    },
    {
      "epoch": 0.8581173260572987,
      "grad_norm": 0.39427489042282104,
      "learning_rate": 1.0398392987470208e-07,
      "loss": 0.5098,
      "step": 5032
    },
    {
      "epoch": 0.8582878581173261,
      "grad_norm": 0.43317997455596924,
      "learning_rate": 1.0373882480568621e-07,
      "loss": 0.5067,
      "step": 5033
    },
    {
      "epoch": 0.8584583901773534,
      "grad_norm": 0.4470864534378052,
      "learning_rate": 1.0349399314714082e-07,
      "loss": 0.4955,
      "step": 5034
    },
    {
      "epoch": 0.8586289222373806,
      "grad_norm": 0.3927958905696869,
      "learning_rate": 1.0324943497375384e-07,
      "loss": 0.4984,
      "step": 5035
    },
    {
      "epoch": 0.8587994542974079,
      "grad_norm": 0.42295411229133606,
      "learning_rate": 1.0300515036012927e-07,
      "loss": 0.5007,
      "step": 5036
    },
    {
      "epoch": 0.8589699863574352,
      "grad_norm": 0.5283581614494324,
      "learning_rate": 1.027611393807877e-07,
      "loss": 0.4972,
      "step": 5037
    },
    {
      "epoch": 0.8591405184174625,
      "grad_norm": 0.48814842104911804,
      "learning_rate": 1.025174021101666e-07,
      "loss": 0.4959,
      "step": 5038
    },
    {
      "epoch": 0.8593110504774898,
      "grad_norm": 0.4385189414024353,
      "learning_rate": 1.0227393862261937e-07,
      "loss": 0.5083,
      "step": 5039
    },
    {
      "epoch": 0.859481582537517,
      "grad_norm": 0.5434243083000183,
      "learning_rate": 1.0203074899241617e-07,
      "loss": 0.5133,
      "step": 5040
    },
    {
      "epoch": 0.8596521145975443,
      "grad_norm": 0.48706600069999695,
      "learning_rate": 1.0178783329374366e-07,
      "loss": 0.5069,
      "step": 5041
    },
    {
      "epoch": 0.8598226466575716,
      "grad_norm": 0.5134983658790588,
      "learning_rate": 1.0154519160070521e-07,
      "loss": 0.4947,
      "step": 5042
    },
    {
      "epoch": 0.859993178717599,
      "grad_norm": 0.5847577452659607,
      "learning_rate": 1.0130282398731984e-07,
      "loss": 0.4919,
      "step": 5043
    },
    {
      "epoch": 0.8601637107776262,
      "grad_norm": 0.5677647590637207,
      "learning_rate": 1.0106073052752364e-07,
      "loss": 0.5007,
      "step": 5044
    },
    {
      "epoch": 0.8603342428376535,
      "grad_norm": 0.4906907081604004,
      "learning_rate": 1.0081891129516898e-07,
      "loss": 0.4939,
      "step": 5045
    },
    {
      "epoch": 0.8605047748976807,
      "grad_norm": 0.4433136284351349,
      "learning_rate": 1.0057736636402382e-07,
      "loss": 0.503,
      "step": 5046
    },
    {
      "epoch": 0.860675306957708,
      "grad_norm": 0.550963819026947,
      "learning_rate": 1.0033609580777393e-07,
      "loss": 0.5154,
      "step": 5047
    },
    {
      "epoch": 0.8608458390177354,
      "grad_norm": 0.5232732892036438,
      "learning_rate": 1.0009509970001986e-07,
      "loss": 0.5014,
      "step": 5048
    },
    {
      "epoch": 0.8610163710777626,
      "grad_norm": 0.4386729896068573,
      "learning_rate": 9.985437811427935e-08,
      "loss": 0.4983,
      "step": 5049
    },
    {
      "epoch": 0.8611869031377899,
      "grad_norm": 0.5767355561256409,
      "learning_rate": 9.961393112398599e-08,
      "loss": 0.5067,
      "step": 5050
    },
    {
      "epoch": 0.8613574351978172,
      "grad_norm": 0.41584306955337524,
      "learning_rate": 9.937375880249013e-08,
      "loss": 0.5022,
      "step": 5051
    },
    {
      "epoch": 0.8615279672578445,
      "grad_norm": 0.5685270428657532,
      "learning_rate": 9.913386122305746e-08,
      "loss": 0.4941,
      "step": 5052
    },
    {
      "epoch": 0.8616984993178718,
      "grad_norm": 0.5781306624412537,
      "learning_rate": 9.889423845887048e-08,
      "loss": 0.4879,
      "step": 5053
    },
    {
      "epoch": 0.861869031377899,
      "grad_norm": 0.4216209948062897,
      "learning_rate": 9.865489058302791e-08,
      "loss": 0.5053,
      "step": 5054
    },
    {
      "epoch": 0.8620395634379263,
      "grad_norm": 0.6022716760635376,
      "learning_rate": 9.841581766854403e-08,
      "loss": 0.4927,
      "step": 5055
    },
    {
      "epoch": 0.8622100954979536,
      "grad_norm": 0.47441956400871277,
      "learning_rate": 9.817701978835026e-08,
      "loss": 0.4861,
      "step": 5056
    },
    {
      "epoch": 0.862380627557981,
      "grad_norm": 0.4886791408061981,
      "learning_rate": 9.79384970152928e-08,
      "loss": 0.5001,
      "step": 5057
    },
    {
      "epoch": 0.8625511596180082,
      "grad_norm": 0.5147485733032227,
      "learning_rate": 9.770024942213508e-08,
      "loss": 0.4978,
      "step": 5058
    },
    {
      "epoch": 0.8627216916780355,
      "grad_norm": 0.37711355090141296,
      "learning_rate": 9.746227708155588e-08,
      "loss": 0.4812,
      "step": 5059
    },
    {
      "epoch": 0.8628922237380627,
      "grad_norm": 0.6241587400436401,
      "learning_rate": 9.722458006615066e-08,
      "loss": 0.4887,
      "step": 5060
    },
    {
      "epoch": 0.86306275579809,
      "grad_norm": 0.43906447291374207,
      "learning_rate": 9.698715844842988e-08,
      "loss": 0.4887,
      "step": 5061
    },
    {
      "epoch": 0.8632332878581174,
      "grad_norm": 0.456028014421463,
      "learning_rate": 9.675001230082094e-08,
      "loss": 0.4856,
      "step": 5062
    },
    {
      "epoch": 0.8634038199181446,
      "grad_norm": 0.5518314838409424,
      "learning_rate": 9.651314169566695e-08,
      "loss": 0.4823,
      "step": 5063
    },
    {
      "epoch": 0.8635743519781719,
      "grad_norm": 0.4024985134601593,
      "learning_rate": 9.627654670522647e-08,
      "loss": 0.4892,
      "step": 5064
    },
    {
      "epoch": 0.8637448840381992,
      "grad_norm": 0.575171709060669,
      "learning_rate": 9.604022740167497e-08,
      "loss": 0.5066,
      "step": 5065
    },
    {
      "epoch": 0.8639154160982264,
      "grad_norm": 0.491485595703125,
      "learning_rate": 9.580418385710278e-08,
      "loss": 0.4864,
      "step": 5066
    },
    {
      "epoch": 0.8640859481582538,
      "grad_norm": 0.3869340419769287,
      "learning_rate": 9.556841614351664e-08,
      "loss": 0.4914,
      "step": 5067
    },
    {
      "epoch": 0.864256480218281,
      "grad_norm": 0.5501559376716614,
      "learning_rate": 9.533292433283915e-08,
      "loss": 0.4869,
      "step": 5068
    },
    {
      "epoch": 0.8644270122783083,
      "grad_norm": 0.3556576669216156,
      "learning_rate": 9.509770849690882e-08,
      "loss": 0.494,
      "step": 5069
    },
    {
      "epoch": 0.8645975443383356,
      "grad_norm": 0.6158584356307983,
      "learning_rate": 9.486276870747946e-08,
      "loss": 0.4884,
      "step": 5070
    },
    {
      "epoch": 0.8647680763983628,
      "grad_norm": 0.41104334592819214,
      "learning_rate": 9.462810503622122e-08,
      "loss": 0.4886,
      "step": 5071
    },
    {
      "epoch": 0.8649386084583902,
      "grad_norm": 0.4305412173271179,
      "learning_rate": 9.439371755472005e-08,
      "loss": 0.4923,
      "step": 5072
    },
    {
      "epoch": 0.8651091405184175,
      "grad_norm": 0.3920281231403351,
      "learning_rate": 9.415960633447674e-08,
      "loss": 0.4819,
      "step": 5073
    },
    {
      "epoch": 0.8652796725784447,
      "grad_norm": 0.4484941065311432,
      "learning_rate": 9.39257714469093e-08,
      "loss": 0.4918,
      "step": 5074
    },
    {
      "epoch": 0.865450204638472,
      "grad_norm": 0.41160324215888977,
      "learning_rate": 9.369221296335008e-08,
      "loss": 0.4872,
      "step": 5075
    },
    {
      "epoch": 0.8656207366984994,
      "grad_norm": 0.3788316249847412,
      "learning_rate": 9.345893095504778e-08,
      "loss": 0.4862,
      "step": 5076
    },
    {
      "epoch": 0.8657912687585266,
      "grad_norm": 0.46384909749031067,
      "learning_rate": 9.322592549316701e-08,
      "loss": 0.4842,
      "step": 5077
    },
    {
      "epoch": 0.8659618008185539,
      "grad_norm": 0.47224605083465576,
      "learning_rate": 9.299319664878704e-08,
      "loss": 0.4951,
      "step": 5078
    },
    {
      "epoch": 0.8661323328785812,
      "grad_norm": 0.3998444676399231,
      "learning_rate": 9.276074449290363e-08,
      "loss": 0.4919,
      "step": 5079
    },
    {
      "epoch": 0.8663028649386084,
      "grad_norm": 0.37172403931617737,
      "learning_rate": 9.252856909642794e-08,
      "loss": 0.4787,
      "step": 5080
    },
    {
      "epoch": 0.8664733969986358,
      "grad_norm": 0.4312663972377777,
      "learning_rate": 9.229667053018666e-08,
      "loss": 0.4894,
      "step": 5081
    },
    {
      "epoch": 0.866643929058663,
      "grad_norm": 0.41867002844810486,
      "learning_rate": 9.206504886492163e-08,
      "loss": 0.4868,
      "step": 5082
    },
    {
      "epoch": 0.8668144611186903,
      "grad_norm": 0.5481159090995789,
      "learning_rate": 9.183370417129124e-08,
      "loss": 0.4927,
      "step": 5083
    },
    {
      "epoch": 0.8669849931787176,
      "grad_norm": 0.46602848172187805,
      "learning_rate": 9.160263651986834e-08,
      "loss": 0.4848,
      "step": 5084
    },
    {
      "epoch": 0.8671555252387448,
      "grad_norm": 0.5236701369285583,
      "learning_rate": 9.137184598114136e-08,
      "loss": 0.4913,
      "step": 5085
    },
    {
      "epoch": 0.8673260572987722,
      "grad_norm": 0.6084462404251099,
      "learning_rate": 9.114133262551518e-08,
      "loss": 0.4849,
      "step": 5086
    },
    {
      "epoch": 0.8674965893587995,
      "grad_norm": 0.4827587604522705,
      "learning_rate": 9.091109652330888e-08,
      "loss": 0.49,
      "step": 5087
    },
    {
      "epoch": 0.8676671214188267,
      "grad_norm": 0.4759979248046875,
      "learning_rate": 9.068113774475785e-08,
      "loss": 0.4912,
      "step": 5088
    },
    {
      "epoch": 0.867837653478854,
      "grad_norm": 0.6387996077537537,
      "learning_rate": 9.04514563600123e-08,
      "loss": 0.4889,
      "step": 5089
    },
    {
      "epoch": 0.8680081855388813,
      "grad_norm": 0.4683969020843506,
      "learning_rate": 9.022205243913852e-08,
      "loss": 0.4834,
      "step": 5090
    },
    {
      "epoch": 0.8681787175989086,
      "grad_norm": 0.4864799678325653,
      "learning_rate": 8.999292605211697e-08,
      "loss": 0.4772,
      "step": 5091
    },
    {
      "epoch": 0.8683492496589359,
      "grad_norm": 0.40657493472099304,
      "learning_rate": 8.976407726884496e-08,
      "loss": 0.4879,
      "step": 5092
    },
    {
      "epoch": 0.8685197817189632,
      "grad_norm": 0.5255517363548279,
      "learning_rate": 8.953550615913398e-08,
      "loss": 0.4851,
      "step": 5093
    },
    {
      "epoch": 0.8686903137789904,
      "grad_norm": 0.48127636313438416,
      "learning_rate": 8.930721279271093e-08,
      "loss": 0.4762,
      "step": 5094
    },
    {
      "epoch": 0.8688608458390177,
      "grad_norm": 0.3790895938873291,
      "learning_rate": 8.907919723921889e-08,
      "loss": 0.4927,
      "step": 5095
    },
    {
      "epoch": 0.869031377899045,
      "grad_norm": 0.5422013998031616,
      "learning_rate": 8.885145956821496e-08,
      "loss": 0.4902,
      "step": 5096
    },
    {
      "epoch": 0.8692019099590723,
      "grad_norm": 0.4123353362083435,
      "learning_rate": 8.862399984917214e-08,
      "loss": 0.4844,
      "step": 5097
    },
    {
      "epoch": 0.8693724420190996,
      "grad_norm": 0.43875542283058167,
      "learning_rate": 8.839681815147872e-08,
      "loss": 0.4767,
      "step": 5098
    },
    {
      "epoch": 0.8695429740791268,
      "grad_norm": 0.5209375023841858,
      "learning_rate": 8.816991454443803e-08,
      "loss": 0.4841,
      "step": 5099
    },
    {
      "epoch": 0.8697135061391542,
      "grad_norm": 0.5397298336029053,
      "learning_rate": 8.794328909726825e-08,
      "loss": 0.4743,
      "step": 5100
    },
    {
      "epoch": 0.8698840381991815,
      "grad_norm": 0.4797486662864685,
      "learning_rate": 8.771694187910307e-08,
      "loss": 0.4876,
      "step": 5101
    },
    {
      "epoch": 0.8700545702592087,
      "grad_norm": 0.40398508310317993,
      "learning_rate": 8.74908729589914e-08,
      "loss": 0.4942,
      "step": 5102
    },
    {
      "epoch": 0.870225102319236,
      "grad_norm": 0.527980625629425,
      "learning_rate": 8.726508240589693e-08,
      "loss": 0.4855,
      "step": 5103
    },
    {
      "epoch": 0.8703956343792633,
      "grad_norm": 0.5567640662193298,
      "learning_rate": 8.703957028869881e-08,
      "loss": 0.4873,
      "step": 5104
    },
    {
      "epoch": 0.8705661664392906,
      "grad_norm": 0.4136599004268646,
      "learning_rate": 8.681433667619066e-08,
      "loss": 0.4807,
      "step": 5105
    },
    {
      "epoch": 0.8707366984993179,
      "grad_norm": 0.5601804852485657,
      "learning_rate": 8.658938163708175e-08,
      "loss": 0.4908,
      "step": 5106
    },
    {
      "epoch": 0.8709072305593452,
      "grad_norm": 0.35288894176483154,
      "learning_rate": 8.636470523999597e-08,
      "loss": 0.4844,
      "step": 5107
    },
    {
      "epoch": 0.8710777626193724,
      "grad_norm": 0.5323822498321533,
      "learning_rate": 8.614030755347269e-08,
      "loss": 0.4885,
      "step": 5108
    },
    {
      "epoch": 0.8712482946793997,
      "grad_norm": 0.4504302144050598,
      "learning_rate": 8.591618864596541e-08,
      "loss": 0.4865,
      "step": 5109
    },
    {
      "epoch": 0.871418826739427,
      "grad_norm": 0.46580883860588074,
      "learning_rate": 8.569234858584342e-08,
      "loss": 0.4929,
      "step": 5110
    },
    {
      "epoch": 0.8715893587994543,
      "grad_norm": 0.6324315667152405,
      "learning_rate": 8.546878744139065e-08,
      "loss": 0.4887,
      "step": 5111
    },
    {
      "epoch": 0.8717598908594816,
      "grad_norm": 0.4285943806171417,
      "learning_rate": 8.524550528080578e-08,
      "loss": 0.4835,
      "step": 5112
    },
    {
      "epoch": 0.8719304229195088,
      "grad_norm": 0.5929010510444641,
      "learning_rate": 8.502250217220276e-08,
      "loss": 0.493,
      "step": 5113
    },
    {
      "epoch": 0.8721009549795361,
      "grad_norm": 0.5795062184333801,
      "learning_rate": 8.479977818360986e-08,
      "loss": 0.4873,
      "step": 5114
    },
    {
      "epoch": 0.8722714870395635,
      "grad_norm": 0.3946116268634796,
      "learning_rate": 8.457733338297071e-08,
      "loss": 0.4955,
      "step": 5115
    },
    {
      "epoch": 0.8724420190995907,
      "grad_norm": 0.5786437392234802,
      "learning_rate": 8.43551678381437e-08,
      "loss": 0.4814,
      "step": 5116
    },
    {
      "epoch": 0.872612551159618,
      "grad_norm": 0.3585083484649658,
      "learning_rate": 8.41332816169016e-08,
      "loss": 0.4901,
      "step": 5117
    },
    {
      "epoch": 0.8727830832196453,
      "grad_norm": 0.5978417992591858,
      "learning_rate": 8.391167478693241e-08,
      "loss": 0.4959,
      "step": 5118
    },
    {
      "epoch": 0.8729536152796725,
      "grad_norm": 0.373609721660614,
      "learning_rate": 8.369034741583893e-08,
      "loss": 0.4852,
      "step": 5119
    },
    {
      "epoch": 0.8731241473396999,
      "grad_norm": 0.5995633006095886,
      "learning_rate": 8.346929957113848e-08,
      "loss": 0.4938,
      "step": 5120
    },
    {
      "epoch": 0.8732946793997272,
      "grad_norm": 0.4977155029773712,
      "learning_rate": 8.324853132026301e-08,
      "loss": 0.4851,
      "step": 5121
    },
    {
      "epoch": 0.8734652114597544,
      "grad_norm": 0.5150020122528076,
      "learning_rate": 8.302804273055976e-08,
      "loss": 0.4951,
      "step": 5122
    },
    {
      "epoch": 0.8736357435197817,
      "grad_norm": 0.6290849447250366,
      "learning_rate": 8.280783386928982e-08,
      "loss": 0.4773,
      "step": 5123
    },
    {
      "epoch": 0.873806275579809,
      "grad_norm": 0.4567163288593292,
      "learning_rate": 8.258790480362956e-08,
      "loss": 0.4962,
      "step": 5124
    },
    {
      "epoch": 0.8739768076398363,
      "grad_norm": 0.44084906578063965,
      "learning_rate": 8.236825560066997e-08,
      "loss": 0.4961,
      "step": 5125
    },
    {
      "epoch": 0.8741473396998636,
      "grad_norm": 0.5285959839820862,
      "learning_rate": 8.214888632741603e-08,
      "loss": 0.4841,
      "step": 5126
    },
    {
      "epoch": 0.8743178717598908,
      "grad_norm": 0.4977264106273651,
      "learning_rate": 8.192979705078853e-08,
      "loss": 0.4911,
      "step": 5127
    },
    {
      "epoch": 0.8744884038199181,
      "grad_norm": 0.5132419466972351,
      "learning_rate": 8.171098783762133e-08,
      "loss": 0.4838,
      "step": 5128
    },
    {
      "epoch": 0.8746589358799455,
      "grad_norm": 0.6812865734100342,
      "learning_rate": 8.149245875466417e-08,
      "loss": 0.4882,
      "step": 5129
    },
    {
      "epoch": 0.8748294679399727,
      "grad_norm": 0.5702094435691833,
      "learning_rate": 8.12742098685807e-08,
      "loss": 0.4923,
      "step": 5130
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.7046767473220825,
      "learning_rate": 8.105624124594925e-08,
      "loss": 0.4935,
      "step": 5131
    },
    {
      "epoch": 0.8751705320600273,
      "grad_norm": 0.7282161712646484,
      "learning_rate": 8.083855295326238e-08,
      "loss": 0.4899,
      "step": 5132
    },
    {
      "epoch": 0.8753410641200545,
      "grad_norm": 0.39676764607429504,
      "learning_rate": 8.062114505692743e-08,
      "loss": 0.4838,
      "step": 5133
    },
    {
      "epoch": 0.8755115961800819,
      "grad_norm": 0.5209677815437317,
      "learning_rate": 8.040401762326655e-08,
      "loss": 0.4847,
      "step": 5134
    },
    {
      "epoch": 0.8756821282401092,
      "grad_norm": 0.475814551115036,
      "learning_rate": 8.018717071851512e-08,
      "loss": 0.4919,
      "step": 5135
    },
    {
      "epoch": 0.8758526603001364,
      "grad_norm": 0.5101640820503235,
      "learning_rate": 7.997060440882454e-08,
      "loss": 0.4844,
      "step": 5136
    },
    {
      "epoch": 0.8760231923601637,
      "grad_norm": 0.610802948474884,
      "learning_rate": 7.975431876025931e-08,
      "loss": 0.4884,
      "step": 5137
    },
    {
      "epoch": 0.876193724420191,
      "grad_norm": 0.5693475604057312,
      "learning_rate": 7.953831383879896e-08,
      "loss": 0.4847,
      "step": 5138
    },
    {
      "epoch": 0.8763642564802183,
      "grad_norm": 0.505389392375946,
      "learning_rate": 7.932258971033747e-08,
      "loss": 0.4864,
      "step": 5139
    },
    {
      "epoch": 0.8765347885402456,
      "grad_norm": 0.39394593238830566,
      "learning_rate": 7.910714644068252e-08,
      "loss": 0.4977,
      "step": 5140
    },
    {
      "epoch": 0.8767053206002728,
      "grad_norm": 0.5716357231140137,
      "learning_rate": 7.889198409555666e-08,
      "loss": 0.4942,
      "step": 5141
    },
    {
      "epoch": 0.8768758526603001,
      "grad_norm": 0.4407377541065216,
      "learning_rate": 7.86771027405967e-08,
      "loss": 0.4931,
      "step": 5142
    },
    {
      "epoch": 0.8770463847203275,
      "grad_norm": 0.528765857219696,
      "learning_rate": 7.84625024413538e-08,
      "loss": 0.4771,
      "step": 5143
    },
    {
      "epoch": 0.8772169167803547,
      "grad_norm": 0.4909074902534485,
      "learning_rate": 7.824818326329265e-08,
      "loss": 0.494,
      "step": 5144
    },
    {
      "epoch": 0.877387448840382,
      "grad_norm": 0.4644756317138672,
      "learning_rate": 7.803414527179344e-08,
      "loss": 0.4946,
      "step": 5145
    },
    {
      "epoch": 0.8775579809004093,
      "grad_norm": 0.4189339280128479,
      "learning_rate": 7.782038853214936e-08,
      "loss": 0.4944,
      "step": 5146
    },
    {
      "epoch": 0.8777285129604365,
      "grad_norm": 0.47703874111175537,
      "learning_rate": 7.760691310956848e-08,
      "loss": 0.4969,
      "step": 5147
    },
    {
      "epoch": 0.8778990450204639,
      "grad_norm": 0.5297402739524841,
      "learning_rate": 7.739371906917326e-08,
      "loss": 0.498,
      "step": 5148
    },
    {
      "epoch": 0.8780695770804912,
      "grad_norm": 0.4273715317249298,
      "learning_rate": 7.718080647599946e-08,
      "loss": 0.4918,
      "step": 5149
    },
    {
      "epoch": 0.8782401091405184,
      "grad_norm": 0.4729132056236267,
      "learning_rate": 7.696817539499769e-08,
      "loss": 0.4946,
      "step": 5150
    },
    {
      "epoch": 0.8784106412005457,
      "grad_norm": 0.44749221205711365,
      "learning_rate": 7.675582589103249e-08,
      "loss": 0.4997,
      "step": 5151
    },
    {
      "epoch": 0.878581173260573,
      "grad_norm": 0.531842827796936,
      "learning_rate": 7.654375802888271e-08,
      "loss": 0.4927,
      "step": 5152
    },
    {
      "epoch": 0.8787517053206003,
      "grad_norm": 0.5416556596755981,
      "learning_rate": 7.633197187324049e-08,
      "loss": 0.4951,
      "step": 5153
    },
    {
      "epoch": 0.8789222373806276,
      "grad_norm": 0.43462786078453064,
      "learning_rate": 7.612046748871327e-08,
      "loss": 0.4975,
      "step": 5154
    },
    {
      "epoch": 0.8790927694406548,
      "grad_norm": 0.535729706287384,
      "learning_rate": 7.590924493982167e-08,
      "loss": 0.4922,
      "step": 5155
    },
    {
      "epoch": 0.8792633015006821,
      "grad_norm": 0.4104560613632202,
      "learning_rate": 7.569830429100013e-08,
      "loss": 0.5074,
      "step": 5156
    },
    {
      "epoch": 0.8794338335607094,
      "grad_norm": 0.5912339687347412,
      "learning_rate": 7.548764560659816e-08,
      "loss": 0.5,
      "step": 5157
    },
    {
      "epoch": 0.8796043656207367,
      "grad_norm": 0.48691239953041077,
      "learning_rate": 7.527726895087818e-08,
      "loss": 0.5054,
      "step": 5158
    },
    {
      "epoch": 0.879774897680764,
      "grad_norm": 0.3652518093585968,
      "learning_rate": 7.506717438801703e-08,
      "loss": 0.4919,
      "step": 5159
    },
    {
      "epoch": 0.8799454297407913,
      "grad_norm": 0.6726717948913574,
      "learning_rate": 7.485736198210559e-08,
      "loss": 0.4965,
      "step": 5160
    },
    {
      "epoch": 0.8801159618008185,
      "grad_norm": 0.34279900789260864,
      "learning_rate": 7.464783179714864e-08,
      "loss": 0.4835,
      "step": 5161
    },
    {
      "epoch": 0.8802864938608458,
      "grad_norm": 0.5112735629081726,
      "learning_rate": 7.443858389706449e-08,
      "loss": 0.4858,
      "step": 5162
    },
    {
      "epoch": 0.8804570259208732,
      "grad_norm": 0.4501814842224121,
      "learning_rate": 7.422961834568566e-08,
      "loss": 0.4892,
      "step": 5163
    },
    {
      "epoch": 0.8806275579809004,
      "grad_norm": 0.44029852747917175,
      "learning_rate": 7.40209352067588e-08,
      "loss": 0.4844,
      "step": 5164
    },
    {
      "epoch": 0.8807980900409277,
      "grad_norm": 0.5076018571853638,
      "learning_rate": 7.381253454394344e-08,
      "loss": 0.4811,
      "step": 5165
    },
    {
      "epoch": 0.880968622100955,
      "grad_norm": 0.42398062348365784,
      "learning_rate": 7.360441642081451e-08,
      "loss": 0.4827,
      "step": 5166
    },
    {
      "epoch": 0.8811391541609823,
      "grad_norm": 0.515562117099762,
      "learning_rate": 7.3396580900859e-08,
      "loss": 0.4884,
      "step": 5167
    },
    {
      "epoch": 0.8813096862210096,
      "grad_norm": 0.47236382961273193,
      "learning_rate": 7.318902804747897e-08,
      "loss": 0.4856,
      "step": 5168
    },
    {
      "epoch": 0.8814802182810368,
      "grad_norm": 0.4556432366371155,
      "learning_rate": 7.298175792398977e-08,
      "loss": 0.4945,
      "step": 5169
    },
    {
      "epoch": 0.8816507503410641,
      "grad_norm": 0.5327446460723877,
      "learning_rate": 7.27747705936206e-08,
      "loss": 0.482,
      "step": 5170
    },
    {
      "epoch": 0.8818212824010914,
      "grad_norm": 0.3763869106769562,
      "learning_rate": 7.256806611951396e-08,
      "loss": 0.4998,
      "step": 5171
    },
    {
      "epoch": 0.8819918144611187,
      "grad_norm": 0.5114072561264038,
      "learning_rate": 7.236164456472673e-08,
      "loss": 0.4874,
      "step": 5172
    },
    {
      "epoch": 0.882162346521146,
      "grad_norm": 0.4112900197505951,
      "learning_rate": 7.215550599222932e-08,
      "loss": 0.5014,
      "step": 5173
    },
    {
      "epoch": 0.8823328785811733,
      "grad_norm": 0.43765711784362793,
      "learning_rate": 7.194965046490521e-08,
      "loss": 0.4978,
      "step": 5174
    },
    {
      "epoch": 0.8825034106412005,
      "grad_norm": 0.6492908596992493,
      "learning_rate": 7.174407804555261e-08,
      "loss": 0.5022,
      "step": 5175
    },
    {
      "epoch": 0.8826739427012278,
      "grad_norm": 0.4558906555175781,
      "learning_rate": 7.153878879688226e-08,
      "loss": 0.4933,
      "step": 5176
    },
    {
      "epoch": 0.8828444747612552,
      "grad_norm": 0.6890559196472168,
      "learning_rate": 7.133378278151926e-08,
      "loss": 0.5027,
      "step": 5177
    },
    {
      "epoch": 0.8830150068212824,
      "grad_norm": 0.48971065878868103,
      "learning_rate": 7.112906006200216e-08,
      "loss": 0.4882,
      "step": 5178
    },
    {
      "epoch": 0.8831855388813097,
      "grad_norm": 0.6021869778633118,
      "learning_rate": 7.092462070078266e-08,
      "loss": 0.4962,
      "step": 5179
    },
    {
      "epoch": 0.883356070941337,
      "grad_norm": 0.6577697396278381,
      "learning_rate": 7.072046476022654e-08,
      "loss": 0.4882,
      "step": 5180
    },
    {
      "epoch": 0.8835266030013642,
      "grad_norm": 0.4322252571582794,
      "learning_rate": 7.051659230261299e-08,
      "loss": 0.5037,
      "step": 5181
    },
    {
      "epoch": 0.8836971350613916,
      "grad_norm": 0.6661120057106018,
      "learning_rate": 7.031300339013481e-08,
      "loss": 0.4843,
      "step": 5182
    },
    {
      "epoch": 0.8838676671214188,
      "grad_norm": 0.6816942095756531,
      "learning_rate": 7.010969808489754e-08,
      "loss": 0.5055,
      "step": 5183
    },
    {
      "epoch": 0.8840381991814461,
      "grad_norm": 0.6046971678733826,
      "learning_rate": 6.990667644892169e-08,
      "loss": 0.488,
      "step": 5184
    },
    {
      "epoch": 0.8842087312414734,
      "grad_norm": 0.6977609395980835,
      "learning_rate": 6.970393854413971e-08,
      "loss": 0.5025,
      "step": 5185
    },
    {
      "epoch": 0.8843792633015006,
      "grad_norm": 0.5585513710975647,
      "learning_rate": 6.950148443239846e-08,
      "loss": 0.4875,
      "step": 5186
    },
    {
      "epoch": 0.884549795361528,
      "grad_norm": 0.41966915130615234,
      "learning_rate": 6.929931417545788e-08,
      "loss": 0.5154,
      "step": 5187
    },
    {
      "epoch": 0.8847203274215553,
      "grad_norm": 0.5780752897262573,
      "learning_rate": 6.90974278349913e-08,
      "loss": 0.5023,
      "step": 5188
    },
    {
      "epoch": 0.8848908594815825,
      "grad_norm": 0.49180328845977783,
      "learning_rate": 6.889582547258545e-08,
      "loss": 0.4864,
      "step": 5189
    },
    {
      "epoch": 0.8850613915416098,
      "grad_norm": 0.5427616834640503,
      "learning_rate": 6.869450714974058e-08,
      "loss": 0.5007,
      "step": 5190
    },
    {
      "epoch": 0.8852319236016372,
      "grad_norm": 0.4825659692287445,
      "learning_rate": 6.849347292787013e-08,
      "loss": 0.4923,
      "step": 5191
    },
    {
      "epoch": 0.8854024556616644,
      "grad_norm": 0.47590872645378113,
      "learning_rate": 6.829272286830103e-08,
      "loss": 0.4864,
      "step": 5192
    },
    {
      "epoch": 0.8855729877216917,
      "grad_norm": 0.43711164593696594,
      "learning_rate": 6.809225703227352e-08,
      "loss": 0.488,
      "step": 5193
    },
    {
      "epoch": 0.885743519781719,
      "grad_norm": 0.5710449814796448,
      "learning_rate": 6.789207548094068e-08,
      "loss": 0.5006,
      "step": 5194
    },
    {
      "epoch": 0.8859140518417462,
      "grad_norm": 0.5110660195350647,
      "learning_rate": 6.769217827536957e-08,
      "loss": 0.4982,
      "step": 5195
    },
    {
      "epoch": 0.8860845839017736,
      "grad_norm": 0.3535400629043579,
      "learning_rate": 6.749256547654004e-08,
      "loss": 0.4929,
      "step": 5196
    },
    {
      "epoch": 0.8862551159618008,
      "grad_norm": 0.46066713333129883,
      "learning_rate": 6.729323714534531e-08,
      "loss": 0.4912,
      "step": 5197
    },
    {
      "epoch": 0.8864256480218281,
      "grad_norm": 0.3608464002609253,
      "learning_rate": 6.709419334259172e-08,
      "loss": 0.5013,
      "step": 5198
    },
    {
      "epoch": 0.8865961800818554,
      "grad_norm": 0.48876771330833435,
      "learning_rate": 6.689543412899914e-08,
      "loss": 0.5096,
      "step": 5199
    },
    {
      "epoch": 0.8867667121418826,
      "grad_norm": 0.466252863407135,
      "learning_rate": 6.669695956520018e-08,
      "loss": 0.496,
      "step": 5200
    },
    {
      "epoch": 0.88693724420191,
      "grad_norm": 0.45235931873321533,
      "learning_rate": 6.649876971174119e-08,
      "loss": 0.4862,
      "step": 5201
    },
    {
      "epoch": 0.8871077762619373,
      "grad_norm": 0.4634920656681061,
      "learning_rate": 6.630086462908081e-08,
      "loss": 0.5018,
      "step": 5202
    },
    {
      "epoch": 0.8872783083219645,
      "grad_norm": 0.4214763045310974,
      "learning_rate": 6.610324437759153e-08,
      "loss": 0.5075,
      "step": 5203
    },
    {
      "epoch": 0.8874488403819918,
      "grad_norm": 0.5170018672943115,
      "learning_rate": 6.59059090175589e-08,
      "loss": 0.4944,
      "step": 5204
    },
    {
      "epoch": 0.887619372442019,
      "grad_norm": 0.4873010516166687,
      "learning_rate": 6.57088586091813e-08,
      "loss": 0.4892,
      "step": 5205
    },
    {
      "epoch": 0.8877899045020464,
      "grad_norm": 0.5427590608596802,
      "learning_rate": 6.551209321257014e-08,
      "loss": 0.502,
      "step": 5206
    },
    {
      "epoch": 0.8879604365620737,
      "grad_norm": 0.4779849648475647,
      "learning_rate": 6.531561288775002e-08,
      "loss": 0.5112,
      "step": 5207
    },
    {
      "epoch": 0.888130968622101,
      "grad_norm": 0.5059120655059814,
      "learning_rate": 6.51194176946588e-08,
      "loss": 0.4968,
      "step": 5208
    },
    {
      "epoch": 0.8883015006821282,
      "grad_norm": 0.48147422075271606,
      "learning_rate": 6.492350769314693e-08,
      "loss": 0.5051,
      "step": 5209
    },
    {
      "epoch": 0.8884720327421555,
      "grad_norm": 0.36521458625793457,
      "learning_rate": 6.472788294297828e-08,
      "loss": 0.4892,
      "step": 5210
    },
    {
      "epoch": 0.8886425648021828,
      "grad_norm": 0.5547593832015991,
      "learning_rate": 6.453254350382921e-08,
      "loss": 0.5029,
      "step": 5211
    },
    {
      "epoch": 0.8888130968622101,
      "grad_norm": 0.48255598545074463,
      "learning_rate": 6.43374894352895e-08,
      "loss": 0.4751,
      "step": 5212
    },
    {
      "epoch": 0.8889836289222374,
      "grad_norm": 0.49428290128707886,
      "learning_rate": 6.414272079686157e-08,
      "loss": 0.4943,
      "step": 5213
    },
    {
      "epoch": 0.8891541609822646,
      "grad_norm": 0.46080899238586426,
      "learning_rate": 6.394823764796121e-08,
      "loss": 0.4904,
      "step": 5214
    },
    {
      "epoch": 0.889324693042292,
      "grad_norm": 0.4548596739768982,
      "learning_rate": 6.37540400479163e-08,
      "loss": 0.4857,
      "step": 5215
    },
    {
      "epoch": 0.8894952251023193,
      "grad_norm": 0.5355672836303711,
      "learning_rate": 6.356012805596868e-08,
      "loss": 0.4906,
      "step": 5216
    },
    {
      "epoch": 0.8896657571623465,
      "grad_norm": 0.38427338004112244,
      "learning_rate": 6.336650173127224e-08,
      "loss": 0.5018,
      "step": 5217
    },
    {
      "epoch": 0.8898362892223738,
      "grad_norm": 0.5887435078620911,
      "learning_rate": 6.317316113289375e-08,
      "loss": 0.4988,
      "step": 5218
    },
    {
      "epoch": 0.890006821282401,
      "grad_norm": 0.4572054445743561,
      "learning_rate": 6.298010631981345e-08,
      "loss": 0.4782,
      "step": 5219
    },
    {
      "epoch": 0.8901773533424284,
      "grad_norm": 0.5091980695724487,
      "learning_rate": 6.27873373509238e-08,
      "loss": 0.489,
      "step": 5220
    },
    {
      "epoch": 0.8903478854024557,
      "grad_norm": 0.6113678812980652,
      "learning_rate": 6.259485428503021e-08,
      "loss": 0.4883,
      "step": 5221
    },
    {
      "epoch": 0.890518417462483,
      "grad_norm": 0.4003571569919586,
      "learning_rate": 6.240265718085104e-08,
      "loss": 0.4926,
      "step": 5222
    },
    {
      "epoch": 0.8906889495225102,
      "grad_norm": 0.5379875898361206,
      "learning_rate": 6.221074609701739e-08,
      "loss": 0.4876,
      "step": 5223
    },
    {
      "epoch": 0.8908594815825375,
      "grad_norm": 0.4984959363937378,
      "learning_rate": 6.201912109207265e-08,
      "loss": 0.4949,
      "step": 5224
    },
    {
      "epoch": 0.8910300136425648,
      "grad_norm": 0.4263933598995209,
      "learning_rate": 6.182778222447384e-08,
      "loss": 0.494,
      "step": 5225
    },
    {
      "epoch": 0.8912005457025921,
      "grad_norm": 0.510601282119751,
      "learning_rate": 6.163672955258982e-08,
      "loss": 0.489,
      "step": 5226
    },
    {
      "epoch": 0.8913710777626194,
      "grad_norm": 0.4412578046321869,
      "learning_rate": 6.144596313470242e-08,
      "loss": 0.4941,
      "step": 5227
    },
    {
      "epoch": 0.8915416098226466,
      "grad_norm": 0.4267738461494446,
      "learning_rate": 6.125548302900653e-08,
      "loss": 0.4844,
      "step": 5228
    },
    {
      "epoch": 0.8917121418826739,
      "grad_norm": 0.516291618347168,
      "learning_rate": 6.106528929360911e-08,
      "loss": 0.4975,
      "step": 5229
    },
    {
      "epoch": 0.8918826739427013,
      "grad_norm": 0.5590835213661194,
      "learning_rate": 6.08753819865301e-08,
      "loss": 0.4855,
      "step": 5230
    },
    {
      "epoch": 0.8920532060027285,
      "grad_norm": 0.4871748983860016,
      "learning_rate": 6.068576116570192e-08,
      "loss": 0.484,
      "step": 5231
    },
    {
      "epoch": 0.8922237380627558,
      "grad_norm": 0.4421897530555725,
      "learning_rate": 6.049642688897006e-08,
      "loss": 0.4999,
      "step": 5232
    },
    {
      "epoch": 0.892394270122783,
      "grad_norm": 0.5166879296302795,
      "learning_rate": 6.030737921409169e-08,
      "loss": 0.5015,
      "step": 5233
    },
    {
      "epoch": 0.8925648021828103,
      "grad_norm": 0.408371239900589,
      "learning_rate": 6.011861819873733e-08,
      "loss": 0.4876,
      "step": 5234
    },
    {
      "epoch": 0.8927353342428377,
      "grad_norm": 0.6659005284309387,
      "learning_rate": 5.993014390048991e-08,
      "loss": 0.4929,
      "step": 5235
    },
    {
      "epoch": 0.892905866302865,
      "grad_norm": 0.4018940329551697,
      "learning_rate": 5.974195637684422e-08,
      "loss": 0.4928,
      "step": 5236
    },
    {
      "epoch": 0.8930763983628922,
      "grad_norm": 0.5677812695503235,
      "learning_rate": 5.95540556852089e-08,
      "loss": 0.4862,
      "step": 5237
    },
    {
      "epoch": 0.8932469304229195,
      "grad_norm": 0.466183066368103,
      "learning_rate": 5.936644188290386e-08,
      "loss": 0.4923,
      "step": 5238
    },
    {
      "epoch": 0.8934174624829468,
      "grad_norm": 0.573516309261322,
      "learning_rate": 5.9179115027161895e-08,
      "loss": 0.4921,
      "step": 5239
    },
    {
      "epoch": 0.8935879945429741,
      "grad_norm": 0.6025112867355347,
      "learning_rate": 5.899207517512873e-08,
      "loss": 0.4986,
      "step": 5240
    },
    {
      "epoch": 0.8937585266030014,
      "grad_norm": 0.46804147958755493,
      "learning_rate": 5.880532238386161e-08,
      "loss": 0.4945,
      "step": 5241
    },
    {
      "epoch": 0.8939290586630286,
      "grad_norm": 0.5839747190475464,
      "learning_rate": 5.8618856710330966e-08,
      "loss": 0.4805,
      "step": 5242
    },
    {
      "epoch": 0.8940995907230559,
      "grad_norm": 0.4837527573108673,
      "learning_rate": 5.8432678211419405e-08,
      "loss": 0.48,
      "step": 5243
    },
    {
      "epoch": 0.8942701227830833,
      "grad_norm": 0.5325782895088196,
      "learning_rate": 5.8246786943921946e-08,
      "loss": 0.4922,
      "step": 5244
    },
    {
      "epoch": 0.8944406548431105,
      "grad_norm": 0.41280415654182434,
      "learning_rate": 5.8061182964545777e-08,
      "loss": 0.4801,
      "step": 5245
    },
    {
      "epoch": 0.8946111869031378,
      "grad_norm": 0.47790399193763733,
      "learning_rate": 5.787586632991105e-08,
      "loss": 0.4817,
      "step": 5246
    },
    {
      "epoch": 0.894781718963165,
      "grad_norm": 0.4562673568725586,
      "learning_rate": 5.769083709654933e-08,
      "loss": 0.491,
      "step": 5247
    },
    {
      "epoch": 0.8949522510231923,
      "grad_norm": 0.3849164843559265,
      "learning_rate": 5.750609532090535e-08,
      "loss": 0.4871,
      "step": 5248
    },
    {
      "epoch": 0.8951227830832197,
      "grad_norm": 0.46881672739982605,
      "learning_rate": 5.732164105933591e-08,
      "loss": 0.4902,
      "step": 5249
    },
    {
      "epoch": 0.895293315143247,
      "grad_norm": 0.3761109709739685,
      "learning_rate": 5.713747436810968e-08,
      "loss": 0.4891,
      "step": 5250
    },
    {
      "epoch": 0.8954638472032742,
      "grad_norm": 0.4707009792327881,
      "learning_rate": 5.695359530340816e-08,
      "loss": 0.4927,
      "step": 5251
    },
    {
      "epoch": 0.8956343792633015,
      "grad_norm": 0.4072950482368469,
      "learning_rate": 5.677000392132482e-08,
      "loss": 0.5016,
      "step": 5252
    },
    {
      "epoch": 0.8958049113233287,
      "grad_norm": 0.4647950530052185,
      "learning_rate": 5.658670027786562e-08,
      "loss": 0.4899,
      "step": 5253
    },
    {
      "epoch": 0.8959754433833561,
      "grad_norm": 0.4397040903568268,
      "learning_rate": 5.6403684428948064e-08,
      "loss": 0.4834,
      "step": 5254
    },
    {
      "epoch": 0.8961459754433834,
      "grad_norm": 0.5452993512153625,
      "learning_rate": 5.622095643040305e-08,
      "loss": 0.4924,
      "step": 5255
    },
    {
      "epoch": 0.8963165075034106,
      "grad_norm": 0.44161882996559143,
      "learning_rate": 5.603851633797242e-08,
      "loss": 0.4938,
      "step": 5256
    },
    {
      "epoch": 0.8964870395634379,
      "grad_norm": 0.3644738793373108,
      "learning_rate": 5.5856364207311e-08,
      "loss": 0.4908,
      "step": 5257
    },
    {
      "epoch": 0.8966575716234653,
      "grad_norm": 0.47635167837142944,
      "learning_rate": 5.567450009398568e-08,
      "loss": 0.492,
      "step": 5258
    },
    {
      "epoch": 0.8968281036834925,
      "grad_norm": 0.3558651804924011,
      "learning_rate": 5.549292405347495e-08,
      "loss": 0.4875,
      "step": 5259
    },
    {
      "epoch": 0.8969986357435198,
      "grad_norm": 0.4246607720851898,
      "learning_rate": 5.531163614116986e-08,
      "loss": 0.5027,
      "step": 5260
    },
    {
      "epoch": 0.897169167803547,
      "grad_norm": 0.6894022822380066,
      "learning_rate": 5.513063641237371e-08,
      "loss": 0.4962,
      "step": 5261
    },
    {
      "epoch": 0.8973396998635743,
      "grad_norm": 0.39976760745048523,
      "learning_rate": 5.494992492230167e-08,
      "loss": 0.4825,
      "step": 5262
    },
    {
      "epoch": 0.8975102319236017,
      "grad_norm": 0.3872816264629364,
      "learning_rate": 5.476950172608065e-08,
      "loss": 0.4938,
      "step": 5263
    },
    {
      "epoch": 0.897680763983629,
      "grad_norm": 0.46357059478759766,
      "learning_rate": 5.45893668787505e-08,
      "loss": 0.5052,
      "step": 5264
    },
    {
      "epoch": 0.8978512960436562,
      "grad_norm": 0.5831755995750427,
      "learning_rate": 5.4409520435262155e-08,
      "loss": 0.4941,
      "step": 5265
    },
    {
      "epoch": 0.8980218281036835,
      "grad_norm": 0.43296298384666443,
      "learning_rate": 5.422996245047896e-08,
      "loss": 0.4876,
      "step": 5266
    },
    {
      "epoch": 0.8981923601637107,
      "grad_norm": 0.5589202046394348,
      "learning_rate": 5.405069297917665e-08,
      "loss": 0.4982,
      "step": 5267
    },
    {
      "epoch": 0.8983628922237381,
      "grad_norm": 0.4898163378238678,
      "learning_rate": 5.387171207604214e-08,
      "loss": 0.4842,
      "step": 5268
    },
    {
      "epoch": 0.8985334242837654,
      "grad_norm": 0.4723389446735382,
      "learning_rate": 5.369301979567499e-08,
      "loss": 0.5015,
      "step": 5269
    },
    {
      "epoch": 0.8987039563437926,
      "grad_norm": 0.4845086634159088,
      "learning_rate": 5.351461619258636e-08,
      "loss": 0.4801,
      "step": 5270
    },
    {
      "epoch": 0.8988744884038199,
      "grad_norm": 0.6618307828903198,
      "learning_rate": 5.3336501321199724e-08,
      "loss": 0.4971,
      "step": 5271
    },
    {
      "epoch": 0.8990450204638472,
      "grad_norm": 0.5553351640701294,
      "learning_rate": 5.315867523584984e-08,
      "loss": 0.5004,
      "step": 5272
    },
    {
      "epoch": 0.8992155525238745,
      "grad_norm": 0.41072601079940796,
      "learning_rate": 5.298113799078398e-08,
      "loss": 0.4875,
      "step": 5273
    },
    {
      "epoch": 0.8993860845839018,
      "grad_norm": 0.6561560034751892,
      "learning_rate": 5.280388964016093e-08,
      "loss": 0.4925,
      "step": 5274
    },
    {
      "epoch": 0.899556616643929,
      "grad_norm": 0.3833219110965729,
      "learning_rate": 5.262693023805166e-08,
      "loss": 0.4921,
      "step": 5275
    },
    {
      "epoch": 0.8997271487039563,
      "grad_norm": 0.662147045135498,
      "learning_rate": 5.245025983843888e-08,
      "loss": 0.4981,
      "step": 5276
    },
    {
      "epoch": 0.8998976807639836,
      "grad_norm": 0.45154041051864624,
      "learning_rate": 5.2273878495216704e-08,
      "loss": 0.4806,
      "step": 5277
    },
    {
      "epoch": 0.900068212824011,
      "grad_norm": 0.592162549495697,
      "learning_rate": 5.2097786262191755e-08,
      "loss": 0.4834,
      "step": 5278
    },
    {
      "epoch": 0.9002387448840382,
      "grad_norm": 0.5438944101333618,
      "learning_rate": 5.1921983193082065e-08,
      "loss": 0.4987,
      "step": 5279
    },
    {
      "epoch": 0.9004092769440655,
      "grad_norm": 0.43703946471214294,
      "learning_rate": 5.174646934151751e-08,
      "loss": 0.4925,
      "step": 5280
    },
    {
      "epoch": 0.9005798090040927,
      "grad_norm": 0.5259987711906433,
      "learning_rate": 5.157124476103992e-08,
      "loss": 0.4815,
      "step": 5281
    },
    {
      "epoch": 0.9007503410641201,
      "grad_norm": 0.4846348166465759,
      "learning_rate": 5.1396309505102444e-08,
      "loss": 0.4932,
      "step": 5282
    },
    {
      "epoch": 0.9009208731241474,
      "grad_norm": 0.5075209140777588,
      "learning_rate": 5.122166362707049e-08,
      "loss": 0.4941,
      "step": 5283
    },
    {
      "epoch": 0.9010914051841746,
      "grad_norm": 0.4234059154987335,
      "learning_rate": 5.104730718022078e-08,
      "loss": 0.4882,
      "step": 5284
    },
    {
      "epoch": 0.9012619372442019,
      "grad_norm": 0.49569621682167053,
      "learning_rate": 5.08732402177421e-08,
      "loss": 0.4891,
      "step": 5285
    },
    {
      "epoch": 0.9014324693042292,
      "grad_norm": 0.43381068110466003,
      "learning_rate": 5.069946279273464e-08,
      "loss": 0.4928,
      "step": 5286
    },
    {
      "epoch": 0.9016030013642565,
      "grad_norm": 0.4606774151325226,
      "learning_rate": 5.0525974958210326e-08,
      "loss": 0.4774,
      "step": 5287
    },
    {
      "epoch": 0.9017735334242838,
      "grad_norm": 0.5518508553504944,
      "learning_rate": 5.035277676709294e-08,
      "loss": 0.4845,
      "step": 5288
    },
    {
      "epoch": 0.901944065484311,
      "grad_norm": 0.5774950981140137,
      "learning_rate": 5.017986827221733e-08,
      "loss": 0.483,
      "step": 5289
    },
    {
      "epoch": 0.9021145975443383,
      "grad_norm": 0.5312044024467468,
      "learning_rate": 5.000724952633096e-08,
      "loss": 0.4885,
      "step": 5290
    },
    {
      "epoch": 0.9022851296043656,
      "grad_norm": 0.5509528517723083,
      "learning_rate": 4.9834920582091944e-08,
      "loss": 0.4931,
      "step": 5291
    },
    {
      "epoch": 0.902455661664393,
      "grad_norm": 0.6059635877609253,
      "learning_rate": 4.9662881492070554e-08,
      "loss": 0.4882,
      "step": 5292
    },
    {
      "epoch": 0.9026261937244202,
      "grad_norm": 0.4796038568019867,
      "learning_rate": 4.949113230874847e-08,
      "loss": 0.4827,
      "step": 5293
    },
    {
      "epoch": 0.9027967257844475,
      "grad_norm": 0.5281833410263062,
      "learning_rate": 4.9319673084518905e-08,
      "loss": 0.4876,
      "step": 5294
    },
    {
      "epoch": 0.9029672578444747,
      "grad_norm": 0.4335166811943054,
      "learning_rate": 4.914850387168657e-08,
      "loss": 0.4947,
      "step": 5295
    },
    {
      "epoch": 0.903137789904502,
      "grad_norm": 0.5843914151191711,
      "learning_rate": 4.8977624722467815e-08,
      "loss": 0.4753,
      "step": 5296
    },
    {
      "epoch": 0.9033083219645294,
      "grad_norm": 0.525813102722168,
      "learning_rate": 4.8807035688990485e-08,
      "loss": 0.4935,
      "step": 5297
    },
    {
      "epoch": 0.9034788540245566,
      "grad_norm": 0.45676755905151367,
      "learning_rate": 4.863673682329373e-08,
      "loss": 0.4844,
      "step": 5298
    },
    {
      "epoch": 0.9036493860845839,
      "grad_norm": 0.5244921445846558,
      "learning_rate": 4.8466728177328894e-08,
      "loss": 0.481,
      "step": 5299
    },
    {
      "epoch": 0.9038199181446112,
      "grad_norm": 0.5705152153968811,
      "learning_rate": 4.82970098029576e-08,
      "loss": 0.4914,
      "step": 5300
    },
    {
      "epoch": 0.9039904502046384,
      "grad_norm": 0.5858019590377808,
      "learning_rate": 4.8127581751953983e-08,
      "loss": 0.4892,
      "step": 5301
    },
    {
      "epoch": 0.9041609822646658,
      "grad_norm": 0.7156437039375305,
      "learning_rate": 4.7958444076002924e-08,
      "loss": 0.4949,
      "step": 5302
    },
    {
      "epoch": 0.904331514324693,
      "grad_norm": 0.5970281362533569,
      "learning_rate": 4.778959682670148e-08,
      "loss": 0.4838,
      "step": 5303
    },
    {
      "epoch": 0.9045020463847203,
      "grad_norm": 0.6081550717353821,
      "learning_rate": 4.7621040055557016e-08,
      "loss": 0.4877,
      "step": 5304
    },
    {
      "epoch": 0.9046725784447476,
      "grad_norm": 0.504256546497345,
      "learning_rate": 4.745277381398939e-08,
      "loss": 0.4919,
      "step": 5305
    },
    {
      "epoch": 0.904843110504775,
      "grad_norm": 0.5745243430137634,
      "learning_rate": 4.7284798153329095e-08,
      "loss": 0.4852,
      "step": 5306
    },
    {
      "epoch": 0.9050136425648022,
      "grad_norm": 0.44068363308906555,
      "learning_rate": 4.711711312481815e-08,
      "loss": 0.4892,
      "step": 5307
    },
    {
      "epoch": 0.9051841746248295,
      "grad_norm": 0.6510539054870605,
      "learning_rate": 4.694971877961041e-08,
      "loss": 0.4994,
      "step": 5308
    },
    {
      "epoch": 0.9053547066848567,
      "grad_norm": 0.43108850717544556,
      "learning_rate": 4.6782615168770146e-08,
      "loss": 0.4963,
      "step": 5309
    },
    {
      "epoch": 0.905525238744884,
      "grad_norm": 0.4596211910247803,
      "learning_rate": 4.661580234327379e-08,
      "loss": 0.4858,
      "step": 5310
    },
    {
      "epoch": 0.9056957708049114,
      "grad_norm": 0.4737480878829956,
      "learning_rate": 4.6449280354008643e-08,
      "loss": 0.4991,
      "step": 5311
    },
    {
      "epoch": 0.9058663028649386,
      "grad_norm": 0.43607211112976074,
      "learning_rate": 4.628304925177319e-08,
      "loss": 0.4853,
      "step": 5312
    },
    {
      "epoch": 0.9060368349249659,
      "grad_norm": 0.4824850559234619,
      "learning_rate": 4.61171090872774e-08,
      "loss": 0.4836,
      "step": 5313
    },
    {
      "epoch": 0.9062073669849932,
      "grad_norm": 0.5185177326202393,
      "learning_rate": 4.595145991114247e-08,
      "loss": 0.48,
      "step": 5314
    },
    {
      "epoch": 0.9063778990450204,
      "grad_norm": 0.49160775542259216,
      "learning_rate": 4.5786101773900863e-08,
      "loss": 0.4958,
      "step": 5315
    },
    {
      "epoch": 0.9065484311050478,
      "grad_norm": 0.5173426866531372,
      "learning_rate": 4.5621034725995995e-08,
      "loss": 0.504,
      "step": 5316
    },
    {
      "epoch": 0.906718963165075,
      "grad_norm": 0.44077467918395996,
      "learning_rate": 4.545625881778293e-08,
      "loss": 0.4898,
      "step": 5317
    },
    {
      "epoch": 0.9068894952251023,
      "grad_norm": 0.4675387442111969,
      "learning_rate": 4.529177409952746e-08,
      "loss": 0.4906,
      "step": 5318
    },
    {
      "epoch": 0.9070600272851296,
      "grad_norm": 0.6204301118850708,
      "learning_rate": 4.512758062140688e-08,
      "loss": 0.4969,
      "step": 5319
    },
    {
      "epoch": 0.9072305593451568,
      "grad_norm": 0.37933647632598877,
      "learning_rate": 4.4963678433509465e-08,
      "loss": 0.4906,
      "step": 5320
    },
    {
      "epoch": 0.9074010914051842,
      "grad_norm": 0.466796338558197,
      "learning_rate": 4.480006758583455e-08,
      "loss": 0.4983,
      "step": 5321
    },
    {
      "epoch": 0.9075716234652115,
      "grad_norm": 0.38458162546157837,
      "learning_rate": 4.463674812829278e-08,
      "loss": 0.5064,
      "step": 5322
    },
    {
      "epoch": 0.9077421555252387,
      "grad_norm": 0.4396178424358368,
      "learning_rate": 4.4473720110705944e-08,
      "loss": 0.4964,
      "step": 5323
    },
    {
      "epoch": 0.907912687585266,
      "grad_norm": 0.386715829372406,
      "learning_rate": 4.4310983582806945e-08,
      "loss": 0.4997,
      "step": 5324
    },
    {
      "epoch": 0.9080832196452933,
      "grad_norm": 0.39901241660118103,
      "learning_rate": 4.414853859423918e-08,
      "loss": 0.4905,
      "step": 5325
    },
    {
      "epoch": 0.9082537517053206,
      "grad_norm": 0.45329898595809937,
      "learning_rate": 4.3986385194558116e-08,
      "loss": 0.491,
      "step": 5326
    },
    {
      "epoch": 0.9084242837653479,
      "grad_norm": 0.44330501556396484,
      "learning_rate": 4.382452343322952e-08,
      "loss": 0.4949,
      "step": 5327
    },
    {
      "epoch": 0.9085948158253752,
      "grad_norm": 0.3553907573223114,
      "learning_rate": 4.366295335963011e-08,
      "loss": 0.5047,
      "step": 5328
    },
    {
      "epoch": 0.9087653478854024,
      "grad_norm": 0.5292105078697205,
      "learning_rate": 4.3501675023048475e-08,
      "loss": 0.4955,
      "step": 5329
    },
    {
      "epoch": 0.9089358799454298,
      "grad_norm": 0.4915977120399475,
      "learning_rate": 4.3340688472683246e-08,
      "loss": 0.51,
      "step": 5330
    },
    {
      "epoch": 0.909106412005457,
      "grad_norm": 0.3689645528793335,
      "learning_rate": 4.3179993757644595e-08,
      "loss": 0.4959,
      "step": 5331
    },
    {
      "epoch": 0.9092769440654843,
      "grad_norm": 0.4896616041660309,
      "learning_rate": 4.3019590926953403e-08,
      "loss": 0.5031,
      "step": 5332
    },
    {
      "epoch": 0.9094474761255116,
      "grad_norm": 0.44271454215049744,
      "learning_rate": 4.2859480029541996e-08,
      "loss": 0.4881,
      "step": 5333
    },
    {
      "epoch": 0.9096180081855388,
      "grad_norm": 0.44713589549064636,
      "learning_rate": 4.269966111425273e-08,
      "loss": 0.496,
      "step": 5334
    },
    {
      "epoch": 0.9097885402455662,
      "grad_norm": 0.4862429201602936,
      "learning_rate": 4.2540134229839716e-08,
      "loss": 0.4892,
      "step": 5335
    },
    {
      "epoch": 0.9099590723055935,
      "grad_norm": 0.5278210043907166,
      "learning_rate": 4.2380899424967817e-08,
      "loss": 0.4856,
      "step": 5336
    },
    {
      "epoch": 0.9101296043656207,
      "grad_norm": 0.475392609834671,
      "learning_rate": 4.222195674821239e-08,
      "loss": 0.4995,
      "step": 5337
    },
    {
      "epoch": 0.910300136425648,
      "grad_norm": 0.5768274664878845,
      "learning_rate": 4.2063306248060324e-08,
      "loss": 0.4927,
      "step": 5338
    },
    {
      "epoch": 0.9104706684856753,
      "grad_norm": 0.6023024916648865,
      "learning_rate": 4.190494797290868e-08,
      "loss": 0.4996,
      "step": 5339
    },
    {
      "epoch": 0.9106412005457026,
      "grad_norm": 0.47170257568359375,
      "learning_rate": 4.17468819710658e-08,
      "loss": 0.4957,
      "step": 5340
    },
    {
      "epoch": 0.9108117326057299,
      "grad_norm": 0.5494353771209717,
      "learning_rate": 4.158910829075092e-08,
      "loss": 0.4954,
      "step": 5341
    },
    {
      "epoch": 0.9109822646657572,
      "grad_norm": 0.44590601325035095,
      "learning_rate": 4.1431626980093846e-08,
      "loss": 0.485,
      "step": 5342
    },
    {
      "epoch": 0.9111527967257844,
      "grad_norm": 0.5497511625289917,
      "learning_rate": 4.1274438087135275e-08,
      "loss": 0.4951,
      "step": 5343
    },
    {
      "epoch": 0.9113233287858117,
      "grad_norm": 0.44941020011901855,
      "learning_rate": 4.1117541659826735e-08,
      "loss": 0.4925,
      "step": 5344
    },
    {
      "epoch": 0.911493860845839,
      "grad_norm": 0.4923752248287201,
      "learning_rate": 4.096093774603061e-08,
      "loss": 0.4912,
      "step": 5345
    },
    {
      "epoch": 0.9116643929058663,
      "grad_norm": 0.5564268231391907,
      "learning_rate": 4.08046263935199e-08,
      "loss": 0.4865,
      "step": 5346
    },
    {
      "epoch": 0.9118349249658936,
      "grad_norm": 0.5061978101730347,
      "learning_rate": 4.064860764997858e-08,
      "loss": 0.4947,
      "step": 5347
    },
    {
      "epoch": 0.9120054570259208,
      "grad_norm": 0.5099136829376221,
      "learning_rate": 4.0492881563001024e-08,
      "loss": 0.4826,
      "step": 5348
    },
    {
      "epoch": 0.9121759890859482,
      "grad_norm": 0.4287917912006378,
      "learning_rate": 4.033744818009244e-08,
      "loss": 0.4931,
      "step": 5349
    },
    {
      "epoch": 0.9123465211459755,
      "grad_norm": 0.590749979019165,
      "learning_rate": 4.0182307548669255e-08,
      "loss": 0.4881,
      "step": 5350
    },
    {
      "epoch": 0.9125170532060027,
      "grad_norm": 0.5917319059371948,
      "learning_rate": 4.0027459716057595e-08,
      "loss": 0.4967,
      "step": 5351
    },
    {
      "epoch": 0.91268758526603,
      "grad_norm": 0.6138724684715271,
      "learning_rate": 3.9872904729495127e-08,
      "loss": 0.5035,
      "step": 5352
    },
    {
      "epoch": 0.9128581173260573,
      "grad_norm": 0.4331624507904053,
      "learning_rate": 3.97186426361299e-08,
      "loss": 0.4889,
      "step": 5353
    },
    {
      "epoch": 0.9130286493860846,
      "grad_norm": 0.45480212569236755,
      "learning_rate": 3.9564673483020626e-08,
      "loss": 0.4847,
      "step": 5354
    },
    {
      "epoch": 0.9131991814461119,
      "grad_norm": 0.4731946587562561,
      "learning_rate": 3.9410997317136385e-08,
      "loss": 0.4912,
      "step": 5355
    },
    {
      "epoch": 0.9133697135061392,
      "grad_norm": 0.4428921341896057,
      "learning_rate": 3.9257614185357566e-08,
      "loss": 0.497,
      "step": 5356
    },
    {
      "epoch": 0.9135402455661664,
      "grad_norm": 0.45451268553733826,
      "learning_rate": 3.910452413447419e-08,
      "loss": 0.4849,
      "step": 5357
    },
    {
      "epoch": 0.9137107776261937,
      "grad_norm": 0.37964358925819397,
      "learning_rate": 3.8951727211187786e-08,
      "loss": 0.4889,
      "step": 5358
    },
    {
      "epoch": 0.913881309686221,
      "grad_norm": 0.46263939142227173,
      "learning_rate": 3.879922346210996e-08,
      "loss": 0.4952,
      "step": 5359
    },
    {
      "epoch": 0.9140518417462483,
      "grad_norm": 0.40812090039253235,
      "learning_rate": 3.8647012933762946e-08,
      "loss": 0.4991,
      "step": 5360
    },
    {
      "epoch": 0.9142223738062756,
      "grad_norm": 0.4596121311187744,
      "learning_rate": 3.849509567257959e-08,
      "loss": 0.484,
      "step": 5361
    },
    {
      "epoch": 0.9143929058663028,
      "grad_norm": 0.4200727939605713,
      "learning_rate": 3.834347172490327e-08,
      "loss": 0.4897,
      "step": 5362
    },
    {
      "epoch": 0.9145634379263301,
      "grad_norm": 0.46064093708992004,
      "learning_rate": 3.8192141136987996e-08,
      "loss": 0.4864,
      "step": 5363
    },
    {
      "epoch": 0.9147339699863575,
      "grad_norm": 0.402382493019104,
      "learning_rate": 3.804110395499805e-08,
      "loss": 0.4951,
      "step": 5364
    },
    {
      "epoch": 0.9149045020463847,
      "grad_norm": 0.4919821619987488,
      "learning_rate": 3.7890360225008584e-08,
      "loss": 0.4926,
      "step": 5365
    },
    {
      "epoch": 0.915075034106412,
      "grad_norm": 0.4360262453556061,
      "learning_rate": 3.773990999300459e-08,
      "loss": 0.4935,
      "step": 5366
    },
    {
      "epoch": 0.9152455661664393,
      "grad_norm": 0.44577082991600037,
      "learning_rate": 3.758975330488213e-08,
      "loss": 0.4864,
      "step": 5367
    },
    {
      "epoch": 0.9154160982264665,
      "grad_norm": 0.4504692852497101,
      "learning_rate": 3.743989020644767e-08,
      "loss": 0.493,
      "step": 5368
    },
    {
      "epoch": 0.9155866302864939,
      "grad_norm": 0.4033011496067047,
      "learning_rate": 3.729032074341743e-08,
      "loss": 0.4782,
      "step": 5369
    },
    {
      "epoch": 0.9157571623465212,
      "grad_norm": 0.5430464148521423,
      "learning_rate": 3.7141044961419243e-08,
      "loss": 0.4903,
      "step": 5370
    },
    {
      "epoch": 0.9159276944065484,
      "grad_norm": 0.5598674416542053,
      "learning_rate": 3.6992062905990223e-08,
      "loss": 0.4853,
      "step": 5371
    },
    {
      "epoch": 0.9160982264665757,
      "grad_norm": 0.37662777304649353,
      "learning_rate": 3.684337462257848e-08,
      "loss": 0.4944,
      "step": 5372
    },
    {
      "epoch": 0.916268758526603,
      "grad_norm": 0.40339040756225586,
      "learning_rate": 3.6694980156542494e-08,
      "loss": 0.4894,
      "step": 5373
    },
    {
      "epoch": 0.9164392905866303,
      "grad_norm": 0.35441353917121887,
      "learning_rate": 3.654687955315073e-08,
      "loss": 0.4928,
      "step": 5374
    },
    {
      "epoch": 0.9166098226466576,
      "grad_norm": 0.3989948630332947,
      "learning_rate": 3.639907285758249e-08,
      "loss": 0.4968,
      "step": 5375
    },
    {
      "epoch": 0.9167803547066848,
      "grad_norm": 0.4780336022377014,
      "learning_rate": 3.6251560114927036e-08,
      "loss": 0.478,
      "step": 5376
    },
    {
      "epoch": 0.9169508867667121,
      "grad_norm": 0.5224922895431519,
      "learning_rate": 3.6104341370184265e-08,
      "loss": 0.4922,
      "step": 5377
    },
    {
      "epoch": 0.9171214188267395,
      "grad_norm": 0.7130451798439026,
      "learning_rate": 3.5957416668264026e-08,
      "loss": 0.5005,
      "step": 5378
    },
    {
      "epoch": 0.9172919508867667,
      "grad_norm": 0.5847972631454468,
      "learning_rate": 3.581078605398703e-08,
      "loss": 0.4771,
      "step": 5379
    },
    {
      "epoch": 0.917462482946794,
      "grad_norm": 0.4442712366580963,
      "learning_rate": 3.566444957208359e-08,
      "loss": 0.4856,
      "step": 5380
    },
    {
      "epoch": 0.9176330150068213,
      "grad_norm": 0.4108210802078247,
      "learning_rate": 3.5518407267194804e-08,
      "loss": 0.4919,
      "step": 5381
    },
    {
      "epoch": 0.9178035470668485,
      "grad_norm": 0.49916204810142517,
      "learning_rate": 3.5372659183871896e-08,
      "loss": 0.485,
      "step": 5382
    },
    {
      "epoch": 0.9179740791268759,
      "grad_norm": 0.44206488132476807,
      "learning_rate": 3.522720536657609e-08,
      "loss": 0.4894,
      "step": 5383
    },
    {
      "epoch": 0.9181446111869032,
      "grad_norm": 0.43081364035606384,
      "learning_rate": 3.5082045859679117e-08,
      "loss": 0.4866,
      "step": 5384
    },
    {
      "epoch": 0.9183151432469304,
      "grad_norm": 0.5318077802658081,
      "learning_rate": 3.4937180707462996e-08,
      "loss": 0.484,
      "step": 5385
    },
    {
      "epoch": 0.9184856753069577,
      "grad_norm": 0.40832993388175964,
      "learning_rate": 3.479260995411982e-08,
      "loss": 0.4851,
      "step": 5386
    },
    {
      "epoch": 0.918656207366985,
      "grad_norm": 0.4995136260986328,
      "learning_rate": 3.4648333643751536e-08,
      "loss": 0.4958,
      "step": 5387
    },
    {
      "epoch": 0.9188267394270123,
      "grad_norm": 0.39905282855033875,
      "learning_rate": 3.450435182037104e-08,
      "loss": 0.4831,
      "step": 5388
    },
    {
      "epoch": 0.9189972714870396,
      "grad_norm": 0.46222537755966187,
      "learning_rate": 3.436066452790077e-08,
      "loss": 0.493,
      "step": 5389
    },
    {
      "epoch": 0.9191678035470668,
      "grad_norm": 0.4872012138366699,
      "learning_rate": 3.421727181017321e-08,
      "loss": 0.4909,
      "step": 5390
    },
    {
      "epoch": 0.9193383356070941,
      "grad_norm": 0.4322300851345062,
      "learning_rate": 3.40741737109318e-08,
      "loss": 0.4999,
      "step": 5391
    },
    {
      "epoch": 0.9195088676671214,
      "grad_norm": 0.4919182062149048,
      "learning_rate": 3.393137027382909e-08,
      "loss": 0.4779,
      "step": 5392
    },
    {
      "epoch": 0.9196793997271487,
      "grad_norm": 0.4974134862422943,
      "learning_rate": 3.3788861542428555e-08,
      "loss": 0.5065,
      "step": 5393
    },
    {
      "epoch": 0.919849931787176,
      "grad_norm": 0.48541364073753357,
      "learning_rate": 3.3646647560203305e-08,
      "loss": 0.4852,
      "step": 5394
    },
    {
      "epoch": 0.9200204638472033,
      "grad_norm": 0.5268510580062866,
      "learning_rate": 3.350472837053676e-08,
      "loss": 0.4939,
      "step": 5395
    },
    {
      "epoch": 0.9201909959072305,
      "grad_norm": 0.36918506026268005,
      "learning_rate": 3.336310401672194e-08,
      "loss": 0.474,
      "step": 5396
    },
    {
      "epoch": 0.9203615279672579,
      "grad_norm": 0.5224368572235107,
      "learning_rate": 3.322177454196285e-08,
      "loss": 0.4965,
      "step": 5397
    },
    {
      "epoch": 0.9205320600272852,
      "grad_norm": 0.35501906275749207,
      "learning_rate": 3.308073998937278e-08,
      "loss": 0.4875,
      "step": 5398
    },
    {
      "epoch": 0.9207025920873124,
      "grad_norm": 0.5201204419136047,
      "learning_rate": 3.294000040197509e-08,
      "loss": 0.5005,
      "step": 5399
    },
    {
      "epoch": 0.9208731241473397,
      "grad_norm": 0.4988919496536255,
      "learning_rate": 3.279955582270356e-08,
      "loss": 0.4832,
      "step": 5400
    },
    {
      "epoch": 0.921043656207367,
      "grad_norm": 0.44986864924430847,
      "learning_rate": 3.2659406294401566e-08,
      "loss": 0.4951,
      "step": 5401
    },
    {
      "epoch": 0.9212141882673943,
      "grad_norm": 0.4283299148082733,
      "learning_rate": 3.251955185982281e-08,
      "loss": 0.4831,
      "step": 5402
    },
    {
      "epoch": 0.9213847203274216,
      "grad_norm": 0.4074150621891022,
      "learning_rate": 3.2379992561630715e-08,
      "loss": 0.4884,
      "step": 5403
    },
    {
      "epoch": 0.9215552523874488,
      "grad_norm": 0.43735846877098083,
      "learning_rate": 3.2240728442399005e-08,
      "loss": 0.4847,
      "step": 5404
    },
    {
      "epoch": 0.9217257844474761,
      "grad_norm": 0.5105264186859131,
      "learning_rate": 3.2101759544610816e-08,
      "loss": 0.4921,
      "step": 5405
    },
    {
      "epoch": 0.9218963165075034,
      "grad_norm": 0.44802650809288025,
      "learning_rate": 3.1963085910659664e-08,
      "loss": 0.4778,
      "step": 5406
    },
    {
      "epoch": 0.9220668485675307,
      "grad_norm": 0.4990009367465973,
      "learning_rate": 3.182470758284895e-08,
      "loss": 0.4913,
      "step": 5407
    },
    {
      "epoch": 0.922237380627558,
      "grad_norm": 0.46665534377098083,
      "learning_rate": 3.168662460339167e-08,
      "loss": 0.4913,
      "step": 5408
    },
    {
      "epoch": 0.9224079126875853,
      "grad_norm": 0.403708279132843,
      "learning_rate": 3.1548837014411363e-08,
      "loss": 0.5045,
      "step": 5409
    },
    {
      "epoch": 0.9225784447476125,
      "grad_norm": 0.47914642095565796,
      "learning_rate": 3.141134485794073e-08,
      "loss": 0.5021,
      "step": 5410
    },
    {
      "epoch": 0.9227489768076398,
      "grad_norm": 0.6098238229751587,
      "learning_rate": 3.127414817592278e-08,
      "loss": 0.4976,
      "step": 5411
    },
    {
      "epoch": 0.9229195088676672,
      "grad_norm": 0.4115535318851471,
      "learning_rate": 3.113724701021026e-08,
      "loss": 0.4945,
      "step": 5412
    },
    {
      "epoch": 0.9230900409276944,
      "grad_norm": 0.3769288659095764,
      "learning_rate": 3.100064140256587e-08,
      "loss": 0.481,
      "step": 5413
    },
    {
      "epoch": 0.9232605729877217,
      "grad_norm": 0.33954548835754395,
      "learning_rate": 3.0864331394661834e-08,
      "loss": 0.49,
      "step": 5414
    },
    {
      "epoch": 0.923431105047749,
      "grad_norm": 0.48619434237480164,
      "learning_rate": 3.072831702808066e-08,
      "loss": 0.49,
      "step": 5415
    },
    {
      "epoch": 0.9236016371077762,
      "grad_norm": 0.5546624660491943,
      "learning_rate": 3.05925983443145e-08,
      "loss": 0.4811,
      "step": 5416
    },
    {
      "epoch": 0.9237721691678036,
      "grad_norm": 0.3814515769481659,
      "learning_rate": 3.0457175384764894e-08,
      "loss": 0.4973,
      "step": 5417
    },
    {
      "epoch": 0.9239427012278308,
      "grad_norm": 0.35709336400032043,
      "learning_rate": 3.0322048190744025e-08,
      "loss": 0.4823,
      "step": 5418
    },
    {
      "epoch": 0.9241132332878581,
      "grad_norm": 0.3805638551712036,
      "learning_rate": 3.018721680347281e-08,
      "loss": 0.4933,
      "step": 5419
    },
    {
      "epoch": 0.9242837653478854,
      "grad_norm": 0.3947684168815613,
      "learning_rate": 3.0052681264082895e-08,
      "loss": 0.4966,
      "step": 5420
    },
    {
      "epoch": 0.9244542974079127,
      "grad_norm": 0.44564321637153625,
      "learning_rate": 2.991844161361513e-08,
      "loss": 0.4768,
      "step": 5421
    },
    {
      "epoch": 0.92462482946794,
      "grad_norm": 0.3913787305355072,
      "learning_rate": 2.978449789301996e-08,
      "loss": 0.4991,
      "step": 5422
    },
    {
      "epoch": 0.9247953615279673,
      "grad_norm": 0.38056641817092896,
      "learning_rate": 2.9650850143158164e-08,
      "loss": 0.4913,
      "step": 5423
    },
    {
      "epoch": 0.9249658935879945,
      "grad_norm": 0.32223859429359436,
      "learning_rate": 2.9517498404799675e-08,
      "loss": 0.4829,
      "step": 5424
    },
    {
      "epoch": 0.9251364256480218,
      "grad_norm": 0.4325050711631775,
      "learning_rate": 2.93844427186244e-08,
      "loss": 0.4866,
      "step": 5425
    },
    {
      "epoch": 0.9253069577080492,
      "grad_norm": 0.3830873966217041,
      "learning_rate": 2.9251683125221865e-08,
      "loss": 0.4943,
      "step": 5426
    },
    {
      "epoch": 0.9254774897680764,
      "grad_norm": 0.35291770100593567,
      "learning_rate": 2.911921966509135e-08,
      "loss": 0.4878,
      "step": 5427
    },
    {
      "epoch": 0.9256480218281037,
      "grad_norm": 0.3755296766757965,
      "learning_rate": 2.8987052378641525e-08,
      "loss": 0.4909,
      "step": 5428
    },
    {
      "epoch": 0.925818553888131,
      "grad_norm": 0.4500507116317749,
      "learning_rate": 2.8855181306191142e-08,
      "loss": 0.487,
      "step": 5429
    },
    {
      "epoch": 0.9259890859481582,
      "grad_norm": 0.47844499349594116,
      "learning_rate": 2.872360648796824e-08,
      "loss": 0.5013,
      "step": 5430
    },
    {
      "epoch": 0.9261596180081856,
      "grad_norm": 0.4715981185436249,
      "learning_rate": 2.859232796411049e-08,
      "loss": 0.4878,
      "step": 5431
    },
    {
      "epoch": 0.9263301500682128,
      "grad_norm": 0.3902131915092468,
      "learning_rate": 2.846134577466542e-08,
      "loss": 0.4883,
      "step": 5432
    },
    {
      "epoch": 0.9265006821282401,
      "grad_norm": 0.5492172241210938,
      "learning_rate": 2.833065995958995e-08,
      "loss": 0.4916,
      "step": 5433
    },
    {
      "epoch": 0.9266712141882674,
      "grad_norm": 0.418430894613266,
      "learning_rate": 2.8200270558750744e-08,
      "loss": 0.4951,
      "step": 5434
    },
    {
      "epoch": 0.9268417462482946,
      "grad_norm": 0.46424612402915955,
      "learning_rate": 2.807017761192399e-08,
      "loss": 0.4952,
      "step": 5435
    },
    {
      "epoch": 0.927012278308322,
      "grad_norm": 0.4414074122905731,
      "learning_rate": 2.7940381158795384e-08,
      "loss": 0.4913,
      "step": 5436
    },
    {
      "epoch": 0.9271828103683493,
      "grad_norm": 0.4416884183883667,
      "learning_rate": 2.7810881238960033e-08,
      "loss": 0.4761,
      "step": 5437
    },
    {
      "epoch": 0.9273533424283765,
      "grad_norm": 0.5253694653511047,
      "learning_rate": 2.768167789192289e-08,
      "loss": 0.5034,
      "step": 5438
    },
    {
      "epoch": 0.9275238744884038,
      "grad_norm": 0.5157954692840576,
      "learning_rate": 2.7552771157098425e-08,
      "loss": 0.5028,
      "step": 5439
    },
    {
      "epoch": 0.927694406548431,
      "grad_norm": 0.5215798616409302,
      "learning_rate": 2.7424161073810186e-08,
      "loss": 0.4945,
      "step": 5440
    },
    {
      "epoch": 0.9278649386084584,
      "grad_norm": 0.44901686906814575,
      "learning_rate": 2.7295847681291902e-08,
      "loss": 0.4958,
      "step": 5441
    },
    {
      "epoch": 0.9280354706684857,
      "grad_norm": 0.5716381072998047,
      "learning_rate": 2.7167831018686144e-08,
      "loss": 0.4907,
      "step": 5442
    },
    {
      "epoch": 0.928206002728513,
      "grad_norm": 0.54997318983078,
      "learning_rate": 2.7040111125045345e-08,
      "loss": 0.4853,
      "step": 5443
    },
    {
      "epoch": 0.9283765347885402,
      "grad_norm": 0.5040403008460999,
      "learning_rate": 2.691268803933145e-08,
      "loss": 0.5013,
      "step": 5444
    },
    {
      "epoch": 0.9285470668485676,
      "grad_norm": 0.648098886013031,
      "learning_rate": 2.6785561800415475e-08,
      "loss": 0.5016,
      "step": 5445
    },
    {
      "epoch": 0.9287175989085948,
      "grad_norm": 0.4393656253814697,
      "learning_rate": 2.6658732447078285e-08,
      "loss": 0.5034,
      "step": 5446
    },
    {
      "epoch": 0.9288881309686221,
      "grad_norm": 0.5123600959777832,
      "learning_rate": 2.6532200018010048e-08,
      "loss": 0.4998,
      "step": 5447
    },
    {
      "epoch": 0.9290586630286494,
      "grad_norm": 0.5143542289733887,
      "learning_rate": 2.6405964551810327e-08,
      "loss": 0.5152,
      "step": 5448
    },
    {
      "epoch": 0.9292291950886766,
      "grad_norm": 0.5461297631263733,
      "learning_rate": 2.628002608698799e-08,
      "loss": 0.4902,
      "step": 5449
    },
    {
      "epoch": 0.929399727148704,
      "grad_norm": 0.588545560836792,
      "learning_rate": 2.615438466196141e-08,
      "loss": 0.4989,
      "step": 5450
    },
    {
      "epoch": 0.9295702592087313,
      "grad_norm": 0.4249182641506195,
      "learning_rate": 2.6029040315058488e-08,
      "loss": 0.4883,
      "step": 5451
    },
    {
      "epoch": 0.9297407912687585,
      "grad_norm": 0.5955637693405151,
      "learning_rate": 2.590399308451608e-08,
      "loss": 0.4951,
      "step": 5452
    },
    {
      "epoch": 0.9299113233287858,
      "grad_norm": 0.4577358365058899,
      "learning_rate": 2.5779243008481116e-08,
      "loss": 0.4933,
      "step": 5453
    },
    {
      "epoch": 0.930081855388813,
      "grad_norm": 0.435455322265625,
      "learning_rate": 2.5654790125009155e-08,
      "loss": 0.5086,
      "step": 5454
    },
    {
      "epoch": 0.9302523874488404,
      "grad_norm": 0.5645721554756165,
      "learning_rate": 2.5530634472065277e-08,
      "loss": 0.4977,
      "step": 5455
    },
    {
      "epoch": 0.9304229195088677,
      "grad_norm": 0.39590975642204285,
      "learning_rate": 2.540677608752429e-08,
      "loss": 0.4892,
      "step": 5456
    },
    {
      "epoch": 0.930593451568895,
      "grad_norm": 0.48490989208221436,
      "learning_rate": 2.528321500916986e-08,
      "loss": 0.4997,
      "step": 5457
    },
    {
      "epoch": 0.9307639836289222,
      "grad_norm": 0.4811006188392639,
      "learning_rate": 2.515995127469495e-08,
      "loss": 0.4881,
      "step": 5458
    },
    {
      "epoch": 0.9309345156889495,
      "grad_norm": 0.43111667037010193,
      "learning_rate": 2.503698492170237e-08,
      "loss": 0.5023,
      "step": 5459
    },
    {
      "epoch": 0.9311050477489768,
      "grad_norm": 0.535325288772583,
      "learning_rate": 2.4914315987703664e-08,
      "loss": 0.4918,
      "step": 5460
    },
    {
      "epoch": 0.9312755798090041,
      "grad_norm": 0.376253604888916,
      "learning_rate": 2.479194451011957e-08,
      "loss": 0.4995,
      "step": 5461
    },
    {
      "epoch": 0.9314461118690314,
      "grad_norm": 0.3538033962249756,
      "learning_rate": 2.4669870526280784e-08,
      "loss": 0.4944,
      "step": 5462
    },
    {
      "epoch": 0.9316166439290586,
      "grad_norm": 0.41819334030151367,
      "learning_rate": 2.45480940734264e-08,
      "loss": 0.508,
      "step": 5463
    },
    {
      "epoch": 0.931787175989086,
      "grad_norm": 0.4020731449127197,
      "learning_rate": 2.4426615188705263e-08,
      "loss": 0.5062,
      "step": 5464
    },
    {
      "epoch": 0.9319577080491133,
      "grad_norm": 0.5282530784606934,
      "learning_rate": 2.4305433909175394e-08,
      "loss": 0.501,
      "step": 5465
    },
    {
      "epoch": 0.9321282401091405,
      "grad_norm": 0.5189552307128906,
      "learning_rate": 2.418455027180389e-08,
      "loss": 0.4845,
      "step": 5466
    },
    {
      "epoch": 0.9322987721691678,
      "grad_norm": 0.40900495648384094,
      "learning_rate": 2.406396431346714e-08,
      "loss": 0.4929,
      "step": 5467
    },
    {
      "epoch": 0.932469304229195,
      "grad_norm": 0.43187782168388367,
      "learning_rate": 2.3943676070950608e-08,
      "loss": 0.4953,
      "step": 5468
    },
    {
      "epoch": 0.9326398362892224,
      "grad_norm": 0.43702763319015503,
      "learning_rate": 2.3823685580949277e-08,
      "loss": 0.5043,
      "step": 5469
    },
    {
      "epoch": 0.9328103683492497,
      "grad_norm": 0.5212153792381287,
      "learning_rate": 2.3703992880066642e-08,
      "loss": 0.4942,
      "step": 5470
    },
    {
      "epoch": 0.932980900409277,
      "grad_norm": 0.3882196247577667,
      "learning_rate": 2.3584598004816163e-08,
      "loss": 0.4867,
      "step": 5471
    },
    {
      "epoch": 0.9331514324693042,
      "grad_norm": 0.48326700925827026,
      "learning_rate": 2.346550099161982e-08,
      "loss": 0.4998,
      "step": 5472
    },
    {
      "epoch": 0.9333219645293315,
      "grad_norm": 0.46829846501350403,
      "learning_rate": 2.334670187680899e-08,
      "loss": 0.4851,
      "step": 5473
    },
    {
      "epoch": 0.9334924965893588,
      "grad_norm": 0.4301968216896057,
      "learning_rate": 2.3228200696624128e-08,
      "loss": 0.4866,
      "step": 5474
    },
    {
      "epoch": 0.9336630286493861,
      "grad_norm": 0.41791969537734985,
      "learning_rate": 2.3109997487214988e-08,
      "loss": 0.4993,
      "step": 5475
    },
    {
      "epoch": 0.9338335607094134,
      "grad_norm": 0.3805132210254669,
      "learning_rate": 2.2992092284639947e-08,
      "loss": 0.5012,
      "step": 5476
    },
    {
      "epoch": 0.9340040927694406,
      "grad_norm": 0.41024842858314514,
      "learning_rate": 2.287448512486701e-08,
      "loss": 0.4874,
      "step": 5477
    },
    {
      "epoch": 0.9341746248294679,
      "grad_norm": 0.49618643522262573,
      "learning_rate": 2.2757176043772922e-08,
      "loss": 0.5013,
      "step": 5478
    },
    {
      "epoch": 0.9343451568894953,
      "grad_norm": 0.5039137601852417,
      "learning_rate": 2.2640165077143618e-08,
      "loss": 0.4814,
      "step": 5479
    },
    {
      "epoch": 0.9345156889495225,
      "grad_norm": 0.3791121542453766,
      "learning_rate": 2.252345226067421e-08,
      "loss": 0.5065,
      "step": 5480
    },
    {
      "epoch": 0.9346862210095498,
      "grad_norm": 0.45290419459342957,
      "learning_rate": 2.240703762996843e-08,
      "loss": 0.4925,
      "step": 5481
    },
    {
      "epoch": 0.934856753069577,
      "grad_norm": 0.6099880337715149,
      "learning_rate": 2.2290921220539664e-08,
      "loss": 0.5044,
      "step": 5482
    },
    {
      "epoch": 0.9350272851296043,
      "grad_norm": 0.48982471227645874,
      "learning_rate": 2.217510306780979e-08,
      "loss": 0.4776,
      "step": 5483
    },
    {
      "epoch": 0.9351978171896317,
      "grad_norm": 0.37271150946617126,
      "learning_rate": 2.2059583207109994e-08,
      "loss": 0.5078,
      "step": 5484
    },
    {
      "epoch": 0.935368349249659,
      "grad_norm": 0.5908680558204651,
      "learning_rate": 2.1944361673680306e-08,
      "loss": 0.4831,
      "step": 5485
    },
    {
      "epoch": 0.9355388813096862,
      "grad_norm": 0.5035811066627502,
      "learning_rate": 2.1829438502669943e-08,
      "loss": 0.5011,
      "step": 5486
    },
    {
      "epoch": 0.9357094133697135,
      "grad_norm": 0.4935630261898041,
      "learning_rate": 2.1714813729136975e-08,
      "loss": 0.4832,
      "step": 5487
    },
    {
      "epoch": 0.9358799454297408,
      "grad_norm": 0.5124121904373169,
      "learning_rate": 2.160048738804821e-08,
      "loss": 0.4844,
      "step": 5488
    },
    {
      "epoch": 0.9360504774897681,
      "grad_norm": 0.4626336991786957,
      "learning_rate": 2.148645951428008e-08,
      "loss": 0.4827,
      "step": 5489
    },
    {
      "epoch": 0.9362210095497954,
      "grad_norm": 0.5183104872703552,
      "learning_rate": 2.1372730142617206e-08,
      "loss": 0.4942,
      "step": 5490
    },
    {
      "epoch": 0.9363915416098226,
      "grad_norm": 0.39121490716934204,
      "learning_rate": 2.1259299307753723e-08,
      "loss": 0.4876,
      "step": 5491
    },
    {
      "epoch": 0.9365620736698499,
      "grad_norm": 0.5651541948318481,
      "learning_rate": 2.11461670442924e-08,
      "loss": 0.4866,
      "step": 5492
    },
    {
      "epoch": 0.9367326057298773,
      "grad_norm": 0.46623361110687256,
      "learning_rate": 2.103333338674485e-08,
      "loss": 0.488,
      "step": 5493
    },
    {
      "epoch": 0.9369031377899045,
      "grad_norm": 0.4703865945339203,
      "learning_rate": 2.092079836953176e-08,
      "loss": 0.4953,
      "step": 5494
    },
    {
      "epoch": 0.9370736698499318,
      "grad_norm": 0.7352892160415649,
      "learning_rate": 2.0808562026982894e-08,
      "loss": 0.4873,
      "step": 5495
    },
    {
      "epoch": 0.937244201909959,
      "grad_norm": 0.4713911712169647,
      "learning_rate": 2.0696624393336643e-08,
      "loss": 0.4798,
      "step": 5496
    },
    {
      "epoch": 0.9374147339699863,
      "grad_norm": 0.5269001722335815,
      "learning_rate": 2.0584985502740018e-08,
      "loss": 0.483,
      "step": 5497
    },
    {
      "epoch": 0.9375852660300137,
      "grad_norm": 0.41727718710899353,
      "learning_rate": 2.0473645389249678e-08,
      "loss": 0.5097,
      "step": 5498
    },
    {
      "epoch": 0.937755798090041,
      "grad_norm": 0.5754560232162476,
      "learning_rate": 2.036260408683033e-08,
      "loss": 0.4885,
      "step": 5499
    },
    {
      "epoch": 0.9379263301500682,
      "grad_norm": 0.39966753125190735,
      "learning_rate": 2.0251861629356108e-08,
      "loss": 0.493,
      "step": 5500
    },
    {
      "epoch": 0.9380968622100955,
      "grad_norm": 0.45334547758102417,
      "learning_rate": 2.0141418050609543e-08,
      "loss": 0.5074,
      "step": 5501
    },
    {
      "epoch": 0.9382673942701227,
      "grad_norm": 0.49507802724838257,
      "learning_rate": 2.0031273384282348e-08,
      "loss": 0.4989,
      "step": 5502
    },
    {
      "epoch": 0.9384379263301501,
      "grad_norm": 0.4019278585910797,
      "learning_rate": 1.992142766397465e-08,
      "loss": 0.4993,
      "step": 5503
    },
    {
      "epoch": 0.9386084583901774,
      "grad_norm": 0.4670289158821106,
      "learning_rate": 1.981188092319586e-08,
      "loss": 0.4845,
      "step": 5504
    },
    {
      "epoch": 0.9387789904502046,
      "grad_norm": 0.4508693516254425,
      "learning_rate": 1.970263319536392e-08,
      "loss": 0.4897,
      "step": 5505
    },
    {
      "epoch": 0.9389495225102319,
      "grad_norm": 0.48310673236846924,
      "learning_rate": 1.9593684513805502e-08,
      "loss": 0.4843,
      "step": 5506
    },
    {
      "epoch": 0.9391200545702592,
      "grad_norm": 0.40250974893569946,
      "learning_rate": 1.9485034911756018e-08,
      "loss": 0.4924,
      "step": 5507
    },
    {
      "epoch": 0.9392905866302865,
      "grad_norm": 0.45123982429504395,
      "learning_rate": 1.937668442235996e-08,
      "loss": 0.4964,
      "step": 5508
    },
    {
      "epoch": 0.9394611186903138,
      "grad_norm": 0.40856313705444336,
      "learning_rate": 1.926863307867033e-08,
      "loss": 0.4964,
      "step": 5509
    },
    {
      "epoch": 0.939631650750341,
      "grad_norm": 0.49330079555511475,
      "learning_rate": 1.9160880913648985e-08,
      "loss": 0.4835,
      "step": 5510
    },
    {
      "epoch": 0.9398021828103683,
      "grad_norm": 0.5441690683364868,
      "learning_rate": 1.9053427960166184e-08,
      "loss": 0.4791,
      "step": 5511
    },
    {
      "epoch": 0.9399727148703957,
      "grad_norm": 0.5150030255317688,
      "learning_rate": 1.8946274251001374e-08,
      "loss": 0.49,
      "step": 5512
    },
    {
      "epoch": 0.940143246930423,
      "grad_norm": 0.5744228363037109,
      "learning_rate": 1.883941981884241e-08,
      "loss": 0.4852,
      "step": 5513
    },
    {
      "epoch": 0.9403137789904502,
      "grad_norm": 0.5330885648727417,
      "learning_rate": 1.8732864696286103e-08,
      "loss": 0.4864,
      "step": 5514
    },
    {
      "epoch": 0.9404843110504775,
      "grad_norm": 0.42424067854881287,
      "learning_rate": 1.8626608915837675e-08,
      "loss": 0.4934,
      "step": 5515
    },
    {
      "epoch": 0.9406548431105047,
      "grad_norm": 0.49726104736328125,
      "learning_rate": 1.85206525099112e-08,
      "loss": 0.4923,
      "step": 5516
    },
    {
      "epoch": 0.9408253751705321,
      "grad_norm": 0.4287932515144348,
      "learning_rate": 1.8414995510829372e-08,
      "loss": 0.4911,
      "step": 5517
    },
    {
      "epoch": 0.9409959072305594,
      "grad_norm": 0.5417968034744263,
      "learning_rate": 1.8309637950823636e-08,
      "loss": 0.4865,
      "step": 5518
    },
    {
      "epoch": 0.9411664392905866,
      "grad_norm": 0.6594215631484985,
      "learning_rate": 1.8204579862033945e-08,
      "loss": 0.4968,
      "step": 5519
    },
    {
      "epoch": 0.9413369713506139,
      "grad_norm": 0.4392363727092743,
      "learning_rate": 1.809982127650889e-08,
      "loss": 0.494,
      "step": 5520
    },
    {
      "epoch": 0.9415075034106412,
      "grad_norm": 0.4899759888648987,
      "learning_rate": 1.7995362226206015e-08,
      "loss": 0.4819,
      "step": 5521
    },
    {
      "epoch": 0.9416780354706685,
      "grad_norm": 0.5993614792823792,
      "learning_rate": 1.7891202742991165e-08,
      "loss": 0.4918,
      "step": 5522
    },
    {
      "epoch": 0.9418485675306958,
      "grad_norm": 0.4386999309062958,
      "learning_rate": 1.778734285863859e-08,
      "loss": 0.487,
      "step": 5523
    },
    {
      "epoch": 0.942019099590723,
      "grad_norm": 0.5626959204673767,
      "learning_rate": 1.7683782604831947e-08,
      "loss": 0.5061,
      "step": 5524
    },
    {
      "epoch": 0.9421896316507503,
      "grad_norm": 0.479331910610199,
      "learning_rate": 1.7580522013162628e-08,
      "loss": 0.485,
      "step": 5525
    },
    {
      "epoch": 0.9423601637107776,
      "grad_norm": 0.4703061282634735,
      "learning_rate": 1.7477561115130993e-08,
      "loss": 0.5045,
      "step": 5526
    },
    {
      "epoch": 0.942530695770805,
      "grad_norm": 0.38216519355773926,
      "learning_rate": 1.737489994214614e-08,
      "loss": 0.4831,
      "step": 5527
    },
    {
      "epoch": 0.9427012278308322,
      "grad_norm": 0.4150906503200531,
      "learning_rate": 1.7272538525525462e-08,
      "loss": 0.4905,
      "step": 5528
    },
    {
      "epoch": 0.9428717598908595,
      "grad_norm": 0.49322372674942017,
      "learning_rate": 1.717047689649487e-08,
      "loss": 0.4852,
      "step": 5529
    },
    {
      "epoch": 0.9430422919508867,
      "grad_norm": 0.4253634810447693,
      "learning_rate": 1.7068715086189127e-08,
      "loss": 0.489,
      "step": 5530
    },
    {
      "epoch": 0.943212824010914,
      "grad_norm": 0.5283239483833313,
      "learning_rate": 1.6967253125651286e-08,
      "loss": 0.488,
      "step": 5531
    },
    {
      "epoch": 0.9433833560709414,
      "grad_norm": 0.4244921803474426,
      "learning_rate": 1.6866091045832923e-08,
      "loss": 0.4779,
      "step": 5532
    },
    {
      "epoch": 0.9435538881309686,
      "grad_norm": 0.4265737235546112,
      "learning_rate": 1.6765228877594354e-08,
      "loss": 0.4911,
      "step": 5533
    },
    {
      "epoch": 0.9437244201909959,
      "grad_norm": 0.5310192704200745,
      "learning_rate": 1.6664666651704076e-08,
      "loss": 0.489,
      "step": 5534
    },
    {
      "epoch": 0.9438949522510232,
      "grad_norm": 0.5723373293876648,
      "learning_rate": 1.656440439883944e-08,
      "loss": 0.4993,
      "step": 5535
    },
    {
      "epoch": 0.9440654843110505,
      "grad_norm": 0.5719337463378906,
      "learning_rate": 1.6464442149585982e-08,
      "loss": 0.493,
      "step": 5536
    },
    {
      "epoch": 0.9442360163710778,
      "grad_norm": 0.44443824887275696,
      "learning_rate": 1.6364779934438082e-08,
      "loss": 0.4906,
      "step": 5537
    },
    {
      "epoch": 0.944406548431105,
      "grad_norm": 0.5066612958908081,
      "learning_rate": 1.6265417783798087e-08,
      "loss": 0.4847,
      "step": 5538
    },
    {
      "epoch": 0.9445770804911323,
      "grad_norm": 0.6740800142288208,
      "learning_rate": 1.6166355727977204e-08,
      "loss": 0.4975,
      "step": 5539
    },
    {
      "epoch": 0.9447476125511596,
      "grad_norm": 0.525609016418457,
      "learning_rate": 1.6067593797195025e-08,
      "loss": 0.4924,
      "step": 5540
    },
    {
      "epoch": 0.944918144611187,
      "grad_norm": 0.4890042841434479,
      "learning_rate": 1.596913202157935e-08,
      "loss": 0.4958,
      "step": 5541
    },
    {
      "epoch": 0.9450886766712142,
      "grad_norm": 0.4876374900341034,
      "learning_rate": 1.5870970431166813e-08,
      "loss": 0.4896,
      "step": 5542
    },
    {
      "epoch": 0.9452592087312415,
      "grad_norm": 0.5408183932304382,
      "learning_rate": 1.577310905590224e-08,
      "loss": 0.4887,
      "step": 5543
    },
    {
      "epoch": 0.9454297407912687,
      "grad_norm": 0.4260612428188324,
      "learning_rate": 1.5675547925638858e-08,
      "loss": 0.4791,
      "step": 5544
    },
    {
      "epoch": 0.945600272851296,
      "grad_norm": 0.5564858317375183,
      "learning_rate": 1.557828707013831e-08,
      "loss": 0.4874,
      "step": 5545
    },
    {
      "epoch": 0.9457708049113234,
      "grad_norm": 0.4741833508014679,
      "learning_rate": 1.548132651907064e-08,
      "loss": 0.4874,
      "step": 5546
    },
    {
      "epoch": 0.9459413369713506,
      "grad_norm": 0.43672487139701843,
      "learning_rate": 1.5384666302014406e-08,
      "loss": 0.497,
      "step": 5547
    },
    {
      "epoch": 0.9461118690313779,
      "grad_norm": 0.6418878436088562,
      "learning_rate": 1.528830644845647e-08,
      "loss": 0.4819,
      "step": 5548
    },
    {
      "epoch": 0.9462824010914052,
      "grad_norm": 0.49941152334213257,
      "learning_rate": 1.5192246987791983e-08,
      "loss": 0.4807,
      "step": 5549
    },
    {
      "epoch": 0.9464529331514324,
      "grad_norm": 0.5414541959762573,
      "learning_rate": 1.5096487949324502e-08,
      "loss": 0.4909,
      "step": 5550
    },
    {
      "epoch": 0.9466234652114598,
      "grad_norm": 0.5298616886138916,
      "learning_rate": 1.5001029362265995e-08,
      "loss": 0.4868,
      "step": 5551
    },
    {
      "epoch": 0.946793997271487,
      "grad_norm": 0.4212076663970947,
      "learning_rate": 1.490587125573684e-08,
      "loss": 0.4856,
      "step": 5552
    },
    {
      "epoch": 0.9469645293315143,
      "grad_norm": 0.567182719707489,
      "learning_rate": 1.4811013658765474e-08,
      "loss": 0.4972,
      "step": 5553
    },
    {
      "epoch": 0.9471350613915416,
      "grad_norm": 0.43124932050704956,
      "learning_rate": 1.4716456600288976e-08,
      "loss": 0.4999,
      "step": 5554
    },
    {
      "epoch": 0.947305593451569,
      "grad_norm": 0.47852280735969543,
      "learning_rate": 1.4622200109152496e-08,
      "loss": 0.4828,
      "step": 5555
    },
    {
      "epoch": 0.9474761255115962,
      "grad_norm": 0.4701859652996063,
      "learning_rate": 1.4528244214109812e-08,
      "loss": 0.4817,
      "step": 5556
    },
    {
      "epoch": 0.9476466575716235,
      "grad_norm": 0.3365345597267151,
      "learning_rate": 1.4434588943822548e-08,
      "loss": 0.4907,
      "step": 5557
    },
    {
      "epoch": 0.9478171896316507,
      "grad_norm": 0.6257134079933167,
      "learning_rate": 1.4341234326861077e-08,
      "loss": 0.488,
      "step": 5558
    },
    {
      "epoch": 0.947987721691678,
      "grad_norm": 0.47926685214042664,
      "learning_rate": 1.4248180391703616e-08,
      "loss": 0.4877,
      "step": 5559
    },
    {
      "epoch": 0.9481582537517054,
      "grad_norm": 0.4044884443283081,
      "learning_rate": 1.4155427166737124e-08,
      "loss": 0.4856,
      "step": 5560
    },
    {
      "epoch": 0.9483287858117326,
      "grad_norm": 0.47886356711387634,
      "learning_rate": 1.4062974680256637e-08,
      "loss": 0.4903,
      "step": 5561
    },
    {
      "epoch": 0.9484993178717599,
      "grad_norm": 0.5004807710647583,
      "learning_rate": 1.3970822960465036e-08,
      "loss": 0.4896,
      "step": 5562
    },
    {
      "epoch": 0.9486698499317872,
      "grad_norm": 0.5693298578262329,
      "learning_rate": 1.387897203547417e-08,
      "loss": 0.4882,
      "step": 5563
    },
    {
      "epoch": 0.9488403819918144,
      "grad_norm": 0.4116656482219696,
      "learning_rate": 1.3787421933303736e-08,
      "loss": 0.5013,
      "step": 5564
    },
    {
      "epoch": 0.9490109140518418,
      "grad_norm": 0.5718091130256653,
      "learning_rate": 1.3696172681881505e-08,
      "loss": 0.4925,
      "step": 5565
    },
    {
      "epoch": 0.949181446111869,
      "grad_norm": 0.45327019691467285,
      "learning_rate": 1.3605224309043875e-08,
      "loss": 0.4907,
      "step": 5566
    },
    {
      "epoch": 0.9493519781718963,
      "grad_norm": 0.4574928283691406,
      "learning_rate": 1.3514576842535212e-08,
      "loss": 0.4959,
      "step": 5567
    },
    {
      "epoch": 0.9495225102319236,
      "grad_norm": 0.5410921573638916,
      "learning_rate": 1.3424230310007948e-08,
      "loss": 0.4839,
      "step": 5568
    },
    {
      "epoch": 0.9496930422919508,
      "grad_norm": 0.4273688495159149,
      "learning_rate": 1.333418473902315e-08,
      "loss": 0.4821,
      "step": 5569
    },
    {
      "epoch": 0.9498635743519782,
      "grad_norm": 0.4793260097503662,
      "learning_rate": 1.3244440157049733e-08,
      "loss": 0.4956,
      "step": 5570
    },
    {
      "epoch": 0.9500341064120055,
      "grad_norm": 0.3881284296512604,
      "learning_rate": 1.3154996591464909e-08,
      "loss": 0.4829,
      "step": 5571
    },
    {
      "epoch": 0.9502046384720327,
      "grad_norm": 0.5769071578979492,
      "learning_rate": 1.3065854069553964e-08,
      "loss": 0.5037,
      "step": 5572
    },
    {
      "epoch": 0.95037517053206,
      "grad_norm": 0.3778176009654999,
      "learning_rate": 1.2977012618510477e-08,
      "loss": 0.491,
      "step": 5573
    },
    {
      "epoch": 0.9505457025920873,
      "grad_norm": 0.5648781061172485,
      "learning_rate": 1.2888472265436104e-08,
      "loss": 0.4947,
      "step": 5574
    },
    {
      "epoch": 0.9507162346521146,
      "grad_norm": 0.46096527576446533,
      "learning_rate": 1.2800233037340571e-08,
      "loss": 0.4882,
      "step": 5575
    },
    {
      "epoch": 0.9508867667121419,
      "grad_norm": 0.4529668390750885,
      "learning_rate": 1.2712294961142125e-08,
      "loss": 0.4867,
      "step": 5576
    },
    {
      "epoch": 0.9510572987721692,
      "grad_norm": 0.5278765559196472,
      "learning_rate": 1.262465806366664e-08,
      "loss": 0.4915,
      "step": 5577
    },
    {
      "epoch": 0.9512278308321964,
      "grad_norm": 0.4194834232330322,
      "learning_rate": 1.2537322371648287e-08,
      "loss": 0.497,
      "step": 5578
    },
    {
      "epoch": 0.9513983628922238,
      "grad_norm": 0.49713462591171265,
      "learning_rate": 1.2450287911729643e-08,
      "loss": 0.4888,
      "step": 5579
    },
    {
      "epoch": 0.951568894952251,
      "grad_norm": 0.45388472080230713,
      "learning_rate": 1.2363554710460913e-08,
      "loss": 0.4957,
      "step": 5580
    },
    {
      "epoch": 0.9517394270122783,
      "grad_norm": 0.4333135485649109,
      "learning_rate": 1.2277122794300822e-08,
      "loss": 0.4949,
      "step": 5581
    },
    {
      "epoch": 0.9519099590723056,
      "grad_norm": 0.47753965854644775,
      "learning_rate": 1.2190992189615837e-08,
      "loss": 0.4972,
      "step": 5582
    },
    {
      "epoch": 0.9520804911323328,
      "grad_norm": 0.519422173500061,
      "learning_rate": 1.2105162922680825e-08,
      "loss": 0.4914,
      "step": 5583
    },
    {
      "epoch": 0.9522510231923602,
      "grad_norm": 0.5419857501983643,
      "learning_rate": 1.2019635019678512e-08,
      "loss": 0.4928,
      "step": 5584
    },
    {
      "epoch": 0.9524215552523875,
      "grad_norm": 0.4927580952644348,
      "learning_rate": 1.1934408506699803e-08,
      "loss": 0.5068,
      "step": 5585
    },
    {
      "epoch": 0.9525920873124147,
      "grad_norm": 0.422260046005249,
      "learning_rate": 1.1849483409743457e-08,
      "loss": 0.4972,
      "step": 5586
    },
    {
      "epoch": 0.952762619372442,
      "grad_norm": 0.6205607652664185,
      "learning_rate": 1.176485975471664e-08,
      "loss": 0.4918,
      "step": 5587
    },
    {
      "epoch": 0.9529331514324693,
      "grad_norm": 0.40875399112701416,
      "learning_rate": 1.1680537567434259e-08,
      "loss": 0.491,
      "step": 5588
    },
    {
      "epoch": 0.9531036834924966,
      "grad_norm": 0.5891082882881165,
      "learning_rate": 1.1596516873619401e-08,
      "loss": 0.5017,
      "step": 5589
    },
    {
      "epoch": 0.9532742155525239,
      "grad_norm": 0.42861342430114746,
      "learning_rate": 1.1512797698903122e-08,
      "loss": 0.4878,
      "step": 5590
    },
    {
      "epoch": 0.9534447476125512,
      "grad_norm": 0.5009891390800476,
      "learning_rate": 1.1429380068824548e-08,
      "loss": 0.5049,
      "step": 5591
    },
    {
      "epoch": 0.9536152796725784,
      "grad_norm": 0.5420736074447632,
      "learning_rate": 1.1346264008830766e-08,
      "loss": 0.5004,
      "step": 5592
    },
    {
      "epoch": 0.9537858117326057,
      "grad_norm": 0.48001834750175476,
      "learning_rate": 1.1263449544276939e-08,
      "loss": 0.4903,
      "step": 5593
    },
    {
      "epoch": 0.953956343792633,
      "grad_norm": 0.5677177309989929,
      "learning_rate": 1.1180936700425972e-08,
      "loss": 0.4994,
      "step": 5594
    },
    {
      "epoch": 0.9541268758526603,
      "grad_norm": 0.49556687474250793,
      "learning_rate": 1.1098725502449172e-08,
      "loss": 0.5101,
      "step": 5595
    },
    {
      "epoch": 0.9542974079126876,
      "grad_norm": 0.6599041819572449,
      "learning_rate": 1.1016815975425589e-08,
      "loss": 0.4784,
      "step": 5596
    },
    {
      "epoch": 0.9544679399727148,
      "grad_norm": 0.5260385870933533,
      "learning_rate": 1.0935208144342125e-08,
      "loss": 0.5008,
      "step": 5597
    },
    {
      "epoch": 0.9546384720327421,
      "grad_norm": 0.4529414176940918,
      "learning_rate": 1.0853902034093979e-08,
      "loss": 0.4881,
      "step": 5598
    },
    {
      "epoch": 0.9548090040927695,
      "grad_norm": 0.49681493639945984,
      "learning_rate": 1.0772897669484085e-08,
      "loss": 0.4971,
      "step": 5599
    },
    {
      "epoch": 0.9549795361527967,
      "grad_norm": 0.39829450845718384,
      "learning_rate": 1.0692195075223234e-08,
      "loss": 0.4947,
      "step": 5600
    },
    {
      "epoch": 0.955150068212824,
      "grad_norm": 0.48838427662849426,
      "learning_rate": 1.0611794275930399e-08,
      "loss": 0.4948,
      "step": 5601
    },
    {
      "epoch": 0.9553206002728513,
      "grad_norm": 0.46820709109306335,
      "learning_rate": 1.0531695296132517e-08,
      "loss": 0.5054,
      "step": 5602
    },
    {
      "epoch": 0.9554911323328786,
      "grad_norm": 0.5562225580215454,
      "learning_rate": 1.045189816026393e-08,
      "loss": 0.4957,
      "step": 5603
    },
    {
      "epoch": 0.9556616643929059,
      "grad_norm": 0.45520704984664917,
      "learning_rate": 1.0372402892667722e-08,
      "loss": 0.5046,
      "step": 5604
    },
    {
      "epoch": 0.9558321964529332,
      "grad_norm": 0.4145662784576416,
      "learning_rate": 1.0293209517594162e-08,
      "loss": 0.4999,
      "step": 5605
    },
    {
      "epoch": 0.9560027285129604,
      "grad_norm": 0.4943574070930481,
      "learning_rate": 1.0214318059201922e-08,
      "loss": 0.4992,
      "step": 5606
    },
    {
      "epoch": 0.9561732605729877,
      "grad_norm": 0.5607316493988037,
      "learning_rate": 1.01357285415572e-08,
      "loss": 0.4953,
      "step": 5607
    },
    {
      "epoch": 0.956343792633015,
      "grad_norm": 0.40864089131355286,
      "learning_rate": 1.0057440988634372e-08,
      "loss": 0.4924,
      "step": 5608
    },
    {
      "epoch": 0.9565143246930423,
      "grad_norm": 0.43824154138565063,
      "learning_rate": 9.979455424315443e-09,
      "loss": 0.5078,
      "step": 5609
    },
    {
      "epoch": 0.9566848567530696,
      "grad_norm": 0.489147812128067,
      "learning_rate": 9.901771872390498e-09,
      "loss": 0.4924,
      "step": 5610
    },
    {
      "epoch": 0.9568553888130968,
      "grad_norm": 0.4240574836730957,
      "learning_rate": 9.824390356557579e-09,
      "loss": 0.508,
      "step": 5611
    },
    {
      "epoch": 0.9570259208731241,
      "grad_norm": 0.5207276344299316,
      "learning_rate": 9.747310900422136e-09,
      "loss": 0.5032,
      "step": 5612
    },
    {
      "epoch": 0.9571964529331515,
      "grad_norm": 0.4289999008178711,
      "learning_rate": 9.670533527498138e-09,
      "loss": 0.4841,
      "step": 5613
    },
    {
      "epoch": 0.9573669849931787,
      "grad_norm": 0.43601903319358826,
      "learning_rate": 9.594058261206741e-09,
      "loss": 0.4991,
      "step": 5614
    },
    {
      "epoch": 0.957537517053206,
      "grad_norm": 0.6186116337776184,
      "learning_rate": 9.517885124877281e-09,
      "loss": 0.4986,
      "step": 5615
    },
    {
      "epoch": 0.9577080491132333,
      "grad_norm": 0.43979814648628235,
      "learning_rate": 9.442014141747058e-09,
      "loss": 0.5052,
      "step": 5616
    },
    {
      "epoch": 0.9578785811732605,
      "grad_norm": 0.523296058177948,
      "learning_rate": 9.366445334960784e-09,
      "loss": 0.487,
      "step": 5617
    },
    {
      "epoch": 0.9580491132332879,
      "grad_norm": 0.4721669554710388,
      "learning_rate": 9.29117872757146e-09,
      "loss": 0.5055,
      "step": 5618
    },
    {
      "epoch": 0.9582196452933152,
      "grad_norm": 0.5010939836502075,
      "learning_rate": 9.216214342539388e-09,
      "loss": 0.4958,
      "step": 5619
    },
    {
      "epoch": 0.9583901773533424,
      "grad_norm": 0.5684243440628052,
      "learning_rate": 9.141552202733274e-09,
      "loss": 0.4935,
      "step": 5620
    },
    {
      "epoch": 0.9585607094133697,
      "grad_norm": 0.4389380216598511,
      "learning_rate": 9.067192330929009e-09,
      "loss": 0.5028,
      "step": 5621
    },
    {
      "epoch": 0.958731241473397,
      "grad_norm": 0.5720691680908203,
      "learning_rate": 8.993134749810896e-09,
      "loss": 0.497,
      "step": 5622
    },
    {
      "epoch": 0.9589017735334243,
      "grad_norm": 0.5201283693313599,
      "learning_rate": 8.919379481970415e-09,
      "loss": 0.4966,
      "step": 5623
    },
    {
      "epoch": 0.9590723055934516,
      "grad_norm": 0.44963565468788147,
      "learning_rate": 8.845926549907125e-09,
      "loss": 0.498,
      "step": 5624
    },
    {
      "epoch": 0.9592428376534788,
      "grad_norm": 0.4307955801486969,
      "learning_rate": 8.772775976028547e-09,
      "loss": 0.4891,
      "step": 5625
    },
    {
      "epoch": 0.9594133697135061,
      "grad_norm": 0.5067347288131714,
      "learning_rate": 8.699927782649609e-09,
      "loss": 0.4942,
      "step": 5626
    },
    {
      "epoch": 0.9595839017735335,
      "grad_norm": 0.49882620573043823,
      "learning_rate": 8.62738199199309e-09,
      "loss": 0.495,
      "step": 5627
    },
    {
      "epoch": 0.9597544338335607,
      "grad_norm": 0.38423335552215576,
      "learning_rate": 8.555138626189619e-09,
      "loss": 0.4995,
      "step": 5628
    },
    {
      "epoch": 0.959924965893588,
      "grad_norm": 0.6150292754173279,
      "learning_rate": 8.483197707277569e-09,
      "loss": 0.4817,
      "step": 5629
    },
    {
      "epoch": 0.9600954979536153,
      "grad_norm": 0.4514417052268982,
      "learning_rate": 8.41155925720283e-09,
      "loss": 0.4958,
      "step": 5630
    },
    {
      "epoch": 0.9602660300136425,
      "grad_norm": 0.6224324107170105,
      "learning_rate": 8.340223297819473e-09,
      "loss": 0.493,
      "step": 5631
    },
    {
      "epoch": 0.9604365620736699,
      "grad_norm": 0.45860835909843445,
      "learning_rate": 8.269189850888871e-09,
      "loss": 0.5014,
      "step": 5632
    },
    {
      "epoch": 0.9606070941336972,
      "grad_norm": 0.501257061958313,
      "learning_rate": 8.198458938080023e-09,
      "loss": 0.4912,
      "step": 5633
    },
    {
      "epoch": 0.9607776261937244,
      "grad_norm": 0.48550844192504883,
      "learning_rate": 8.128030580970338e-09,
      "loss": 0.496,
      "step": 5634
    },
    {
      "epoch": 0.9609481582537517,
      "grad_norm": 0.4650779664516449,
      "learning_rate": 8.05790480104407e-09,
      "loss": 0.4942,
      "step": 5635
    },
    {
      "epoch": 0.961118690313779,
      "grad_norm": 0.48873841762542725,
      "learning_rate": 7.988081619693666e-09,
      "loss": 0.4983,
      "step": 5636
    },
    {
      "epoch": 0.9612892223738063,
      "grad_norm": 0.40933525562286377,
      "learning_rate": 7.918561058219199e-09,
      "loss": 0.5015,
      "step": 5637
    },
    {
      "epoch": 0.9614597544338336,
      "grad_norm": 0.5009204149246216,
      "learning_rate": 7.84934313782837e-09,
      "loss": 0.4954,
      "step": 5638
    },
    {
      "epoch": 0.9616302864938608,
      "grad_norm": 0.4214461147785187,
      "learning_rate": 7.780427879636621e-09,
      "loss": 0.4884,
      "step": 5639
    },
    {
      "epoch": 0.9618008185538881,
      "grad_norm": 0.5430739521980286,
      "learning_rate": 7.711815304666805e-09,
      "loss": 0.4928,
      "step": 5640
    },
    {
      "epoch": 0.9619713506139154,
      "grad_norm": 0.45583394169807434,
      "learning_rate": 7.643505433849953e-09,
      "loss": 0.4893,
      "step": 5641
    },
    {
      "epoch": 0.9621418826739427,
      "grad_norm": 0.4132305383682251,
      "learning_rate": 7.575498288024175e-09,
      "loss": 0.5066,
      "step": 5642
    },
    {
      "epoch": 0.96231241473397,
      "grad_norm": 0.5108238458633423,
      "learning_rate": 7.507793887935656e-09,
      "loss": 0.4903,
      "step": 5643
    },
    {
      "epoch": 0.9624829467939973,
      "grad_norm": 0.47365233302116394,
      "learning_rate": 7.440392254237983e-09,
      "loss": 0.5016,
      "step": 5644
    },
    {
      "epoch": 0.9626534788540245,
      "grad_norm": 0.5538928508758545,
      "learning_rate": 7.3732934074926e-09,
      "loss": 0.4934,
      "step": 5645
    },
    {
      "epoch": 0.9628240109140518,
      "grad_norm": 0.401213675737381,
      "learning_rate": 7.306497368168353e-09,
      "loss": 0.4908,
      "step": 5646
    },
    {
      "epoch": 0.9629945429740792,
      "grad_norm": 0.48938992619514465,
      "learning_rate": 7.240004156641945e-09,
      "loss": 0.5084,
      "step": 5647
    },
    {
      "epoch": 0.9631650750341064,
      "grad_norm": 0.5778220295906067,
      "learning_rate": 7.173813793197371e-09,
      "loss": 0.4898,
      "step": 5648
    },
    {
      "epoch": 0.9633356070941337,
      "grad_norm": 0.44132354855537415,
      "learning_rate": 7.1079262980265914e-09,
      "loss": 0.4949,
      "step": 5649
    },
    {
      "epoch": 0.963506139154161,
      "grad_norm": 0.6237069964408875,
      "learning_rate": 7.042341691229083e-09,
      "loss": 0.4888,
      "step": 5650
    },
    {
      "epoch": 0.9636766712141883,
      "grad_norm": 0.7165541648864746,
      "learning_rate": 6.97705999281162e-09,
      "loss": 0.4915,
      "step": 5651
    },
    {
      "epoch": 0.9638472032742156,
      "grad_norm": 0.48227575421333313,
      "learning_rate": 6.91208122268916e-09,
      "loss": 0.4855,
      "step": 5652
    },
    {
      "epoch": 0.9640177353342428,
      "grad_norm": 0.5969128608703613,
      "learning_rate": 6.847405400683627e-09,
      "loss": 0.493,
      "step": 5653
    },
    {
      "epoch": 0.9641882673942701,
      "grad_norm": 0.4721926748752594,
      "learning_rate": 6.783032546524904e-09,
      "loss": 0.4832,
      "step": 5654
    },
    {
      "epoch": 0.9643587994542974,
      "grad_norm": 0.634792685508728,
      "learning_rate": 6.718962679850505e-09,
      "loss": 0.4922,
      "step": 5655
    },
    {
      "epoch": 0.9645293315143247,
      "grad_norm": 0.4622402489185333,
      "learning_rate": 6.65519582020513e-09,
      "loss": 0.4808,
      "step": 5656
    },
    {
      "epoch": 0.964699863574352,
      "grad_norm": 0.6484578847885132,
      "learning_rate": 6.591731987041439e-09,
      "loss": 0.4934,
      "step": 5657
    },
    {
      "epoch": 0.9648703956343793,
      "grad_norm": 0.539602518081665,
      "learning_rate": 6.528571199719503e-09,
      "loss": 0.5055,
      "step": 5658
    },
    {
      "epoch": 0.9650409276944065,
      "grad_norm": 0.36174818873405457,
      "learning_rate": 6.465713477506908e-09,
      "loss": 0.4803,
      "step": 5659
    },
    {
      "epoch": 0.9652114597544338,
      "grad_norm": 0.5808553695678711,
      "learning_rate": 6.403158839578761e-09,
      "loss": 0.5017,
      "step": 5660
    },
    {
      "epoch": 0.9653819918144612,
      "grad_norm": 0.44459226727485657,
      "learning_rate": 6.340907305017907e-09,
      "loss": 0.4927,
      "step": 5661
    },
    {
      "epoch": 0.9655525238744884,
      "grad_norm": 0.5725284814834595,
      "learning_rate": 6.27895889281438e-09,
      "loss": 0.4887,
      "step": 5662
    },
    {
      "epoch": 0.9657230559345157,
      "grad_norm": 0.5165860056877136,
      "learning_rate": 6.217313621866284e-09,
      "loss": 0.4948,
      "step": 5663
    },
    {
      "epoch": 0.965893587994543,
      "grad_norm": 0.4258648157119751,
      "learning_rate": 6.155971510978576e-09,
      "loss": 0.4918,
      "step": 5664
    },
    {
      "epoch": 0.9660641200545702,
      "grad_norm": 0.4650655686855316,
      "learning_rate": 6.094932578864288e-09,
      "loss": 0.4911,
      "step": 5665
    },
    {
      "epoch": 0.9662346521145976,
      "grad_norm": 0.5605490803718567,
      "learning_rate": 6.034196844143748e-09,
      "loss": 0.5022,
      "step": 5666
    },
    {
      "epoch": 0.9664051841746248,
      "grad_norm": 0.5298847556114197,
      "learning_rate": 5.9737643253446895e-09,
      "loss": 0.4785,
      "step": 5667
    },
    {
      "epoch": 0.9665757162346521,
      "grad_norm": 0.36176204681396484,
      "learning_rate": 5.913635040902477e-09,
      "loss": 0.4832,
      "step": 5668
    },
    {
      "epoch": 0.9667462482946794,
      "grad_norm": 0.5243964791297913,
      "learning_rate": 5.853809009160106e-09,
      "loss": 0.4855,
      "step": 5669
    },
    {
      "epoch": 0.9669167803547067,
      "grad_norm": 0.42310968041419983,
      "learning_rate": 5.794286248367753e-09,
      "loss": 0.4961,
      "step": 5670
    },
    {
      "epoch": 0.967087312414734,
      "grad_norm": 0.45878174901008606,
      "learning_rate": 5.73506677668323e-09,
      "loss": 0.4881,
      "step": 5671
    },
    {
      "epoch": 0.9672578444747613,
      "grad_norm": 0.481548547744751,
      "learning_rate": 5.676150612171972e-09,
      "loss": 0.4867,
      "step": 5672
    },
    {
      "epoch": 0.9674283765347885,
      "grad_norm": 0.3911304175853729,
      "learning_rate": 5.617537772806603e-09,
      "loss": 0.4869,
      "step": 5673
    },
    {
      "epoch": 0.9675989085948158,
      "grad_norm": 0.4614737331867218,
      "learning_rate": 5.559228276467377e-09,
      "loss": 0.5035,
      "step": 5674
    },
    {
      "epoch": 0.9677694406548432,
      "grad_norm": 0.4647872745990753,
      "learning_rate": 5.501222140942064e-09,
      "loss": 0.4828,
      "step": 5675
    },
    {
      "epoch": 0.9679399727148704,
      "grad_norm": 0.4832058250904083,
      "learning_rate": 5.443519383925845e-09,
      "loss": 0.4924,
      "step": 5676
    },
    {
      "epoch": 0.9681105047748977,
      "grad_norm": 0.37751418352127075,
      "learning_rate": 5.386120023021191e-09,
      "loss": 0.4899,
      "step": 5677
    },
    {
      "epoch": 0.968281036834925,
      "grad_norm": 0.4673527479171753,
      "learning_rate": 5.329024075738321e-09,
      "loss": 0.4967,
      "step": 5678
    },
    {
      "epoch": 0.9684515688949522,
      "grad_norm": 0.4402042031288147,
      "learning_rate": 5.272231559494634e-09,
      "loss": 0.4943,
      "step": 5679
    },
    {
      "epoch": 0.9686221009549796,
      "grad_norm": 0.4784039556980133,
      "learning_rate": 5.215742491615161e-09,
      "loss": 0.4939,
      "step": 5680
    },
    {
      "epoch": 0.9687926330150068,
      "grad_norm": 0.4998716413974762,
      "learning_rate": 5.1595568893321186e-09,
      "loss": 0.4975,
      "step": 5681
    },
    {
      "epoch": 0.9689631650750341,
      "grad_norm": 0.44880411028862,
      "learning_rate": 5.103674769785461e-09,
      "loss": 0.4754,
      "step": 5682
    },
    {
      "epoch": 0.9691336971350614,
      "grad_norm": 0.5895913243293762,
      "learning_rate": 5.0480961500223306e-09,
      "loss": 0.4812,
      "step": 5683
    },
    {
      "epoch": 0.9693042291950886,
      "grad_norm": 0.384233295917511,
      "learning_rate": 4.992821046997388e-09,
      "loss": 0.4783,
      "step": 5684
    },
    {
      "epoch": 0.969474761255116,
      "grad_norm": 0.4828539788722992,
      "learning_rate": 4.937849477572588e-09,
      "loss": 0.4828,
      "step": 5685
    },
    {
      "epoch": 0.9696452933151433,
      "grad_norm": 0.5714468955993652,
      "learning_rate": 4.883181458517406e-09,
      "loss": 0.4972,
      "step": 5686
    },
    {
      "epoch": 0.9698158253751705,
      "grad_norm": 0.4602378308773041,
      "learning_rate": 4.8288170065088354e-09,
      "loss": 0.4891,
      "step": 5687
    },
    {
      "epoch": 0.9699863574351978,
      "grad_norm": 0.5014626979827881,
      "learning_rate": 4.774756138130832e-09,
      "loss": 0.4879,
      "step": 5688
    },
    {
      "epoch": 0.970156889495225,
      "grad_norm": 0.5613033771514893,
      "learning_rate": 4.720998869875316e-09,
      "loss": 0.4889,
      "step": 5689
    },
    {
      "epoch": 0.9703274215552524,
      "grad_norm": 0.5589002370834351,
      "learning_rate": 4.667545218141057e-09,
      "loss": 0.4767,
      "step": 5690
    },
    {
      "epoch": 0.9704979536152797,
      "grad_norm": 0.5097793936729431,
      "learning_rate": 4.614395199234679e-09,
      "loss": 0.4862,
      "step": 5691
    },
    {
      "epoch": 0.970668485675307,
      "grad_norm": 0.5597699880599976,
      "learning_rate": 4.561548829369766e-09,
      "loss": 0.4955,
      "step": 5692
    },
    {
      "epoch": 0.9708390177353342,
      "grad_norm": 0.5028160214424133,
      "learning_rate": 4.509006124667648e-09,
      "loss": 0.4849,
      "step": 5693
    },
    {
      "epoch": 0.9710095497953616,
      "grad_norm": 0.5753107070922852,
      "learning_rate": 4.456767101156723e-09,
      "loss": 0.4922,
      "step": 5694
    },
    {
      "epoch": 0.9711800818553888,
      "grad_norm": 0.6566035151481628,
      "learning_rate": 4.404831774772689e-09,
      "loss": 0.4899,
      "step": 5695
    },
    {
      "epoch": 0.9713506139154161,
      "grad_norm": 0.48711341619491577,
      "learning_rate": 4.353200161358984e-09,
      "loss": 0.4811,
      "step": 5696
    },
    {
      "epoch": 0.9715211459754434,
      "grad_norm": 0.7718820571899414,
      "learning_rate": 4.301872276666119e-09,
      "loss": 0.4848,
      "step": 5697
    },
    {
      "epoch": 0.9716916780354706,
      "grad_norm": 0.6576699018478394,
      "learning_rate": 4.250848136352015e-09,
      "loss": 0.4974,
      "step": 5698
    },
    {
      "epoch": 0.971862210095498,
      "grad_norm": 0.588111400604248,
      "learning_rate": 4.200127755981887e-09,
      "loss": 0.4956,
      "step": 5699
    },
    {
      "epoch": 0.9720327421555253,
      "grad_norm": 0.7305927872657776,
      "learning_rate": 4.149711151028357e-09,
      "loss": 0.5016,
      "step": 5700
    },
    {
      "epoch": 0.9722032742155525,
      "grad_norm": 0.45234251022338867,
      "learning_rate": 4.099598336871347e-09,
      "loss": 0.4784,
      "step": 5701
    },
    {
      "epoch": 0.9723738062755798,
      "grad_norm": 0.6457910537719727,
      "learning_rate": 4.049789328798071e-09,
      "loss": 0.5076,
      "step": 5702
    },
    {
      "epoch": 0.972544338335607,
      "grad_norm": 0.4973987340927124,
      "learning_rate": 4.0002841420032635e-09,
      "loss": 0.4925,
      "step": 5703
    },
    {
      "epoch": 0.9727148703956344,
      "grad_norm": 0.608955442905426,
      "learning_rate": 3.951082791588512e-09,
      "loss": 0.4946,
      "step": 5704
    },
    {
      "epoch": 0.9728854024556617,
      "grad_norm": 0.48888248205184937,
      "learning_rate": 3.902185292563365e-09,
      "loss": 0.4911,
      "step": 5705
    },
    {
      "epoch": 0.973055934515689,
      "grad_norm": 0.5876244306564331,
      "learning_rate": 3.853591659844003e-09,
      "loss": 0.4878,
      "step": 5706
    },
    {
      "epoch": 0.9732264665757162,
      "grad_norm": 0.5642983317375183,
      "learning_rate": 3.8053019082544555e-09,
      "loss": 0.4797,
      "step": 5707
    },
    {
      "epoch": 0.9733969986357435,
      "grad_norm": 0.5118812322616577,
      "learning_rate": 3.757316052525828e-09,
      "loss": 0.4955,
      "step": 5708
    },
    {
      "epoch": 0.9735675306957708,
      "grad_norm": 0.6304372549057007,
      "learning_rate": 3.709634107296412e-09,
      "loss": 0.5018,
      "step": 5709
    },
    {
      "epoch": 0.9737380627557981,
      "grad_norm": 0.44215402007102966,
      "learning_rate": 3.662256087111904e-09,
      "loss": 0.4891,
      "step": 5710
    },
    {
      "epoch": 0.9739085948158254,
      "grad_norm": 0.6735357642173767,
      "learning_rate": 3.615182006425411e-09,
      "loss": 0.4917,
      "step": 5711
    },
    {
      "epoch": 0.9740791268758526,
      "grad_norm": 0.603859543800354,
      "learning_rate": 3.5684118795971115e-09,
      "loss": 0.496,
      "step": 5712
    },
    {
      "epoch": 0.9742496589358799,
      "grad_norm": 0.570151150226593,
      "learning_rate": 3.521945720894593e-09,
      "loss": 0.4985,
      "step": 5713
    },
    {
      "epoch": 0.9744201909959073,
      "grad_norm": 0.7017441391944885,
      "learning_rate": 3.4757835444926282e-09,
      "loss": 0.4774,
      "step": 5714
    },
    {
      "epoch": 0.9745907230559345,
      "grad_norm": 0.48164358735084534,
      "learning_rate": 3.4299253644732856e-09,
      "loss": 0.4926,
      "step": 5715
    },
    {
      "epoch": 0.9747612551159618,
      "grad_norm": 0.5758684873580933,
      "learning_rate": 3.384371194825931e-09,
      "loss": 0.487,
      "step": 5716
    },
    {
      "epoch": 0.974931787175989,
      "grad_norm": 0.38753560185432434,
      "learning_rate": 3.3391210494472253e-09,
      "loss": 0.502,
      "step": 5717
    },
    {
      "epoch": 0.9751023192360164,
      "grad_norm": 0.600566029548645,
      "learning_rate": 3.2941749421409044e-09,
      "loss": 0.4966,
      "step": 5718
    },
    {
      "epoch": 0.9752728512960437,
      "grad_norm": 0.5640249252319336,
      "learning_rate": 3.2495328866182233e-09,
      "loss": 0.4949,
      "step": 5719
    },
    {
      "epoch": 0.975443383356071,
      "grad_norm": 0.450433611869812,
      "learning_rate": 3.2051948964973993e-09,
      "loss": 0.4962,
      "step": 5720
    },
    {
      "epoch": 0.9756139154160982,
      "grad_norm": 0.5050912499427795,
      "learning_rate": 3.1611609853041684e-09,
      "loss": 0.4887,
      "step": 5721
    },
    {
      "epoch": 0.9757844474761255,
      "grad_norm": 0.43759477138519287,
      "learning_rate": 3.1174311664712302e-09,
      "loss": 0.4776,
      "step": 5722
    },
    {
      "epoch": 0.9759549795361528,
      "grad_norm": 0.5738310217857361,
      "learning_rate": 3.074005453338802e-09,
      "loss": 0.4947,
      "step": 5723
    },
    {
      "epoch": 0.9761255115961801,
      "grad_norm": 0.41161057353019714,
      "learning_rate": 3.030883859154066e-09,
      "loss": 0.5027,
      "step": 5724
    },
    {
      "epoch": 0.9762960436562074,
      "grad_norm": 0.4722995460033417,
      "learning_rate": 2.98806639707161e-09,
      "loss": 0.4945,
      "step": 5725
    },
    {
      "epoch": 0.9764665757162346,
      "grad_norm": 0.5092844367027283,
      "learning_rate": 2.9455530801532076e-09,
      "loss": 0.4894,
      "step": 5726
    },
    {
      "epoch": 0.9766371077762619,
      "grad_norm": 0.49190133810043335,
      "learning_rate": 2.9033439213677077e-09,
      "loss": 0.4892,
      "step": 5727
    },
    {
      "epoch": 0.9768076398362893,
      "grad_norm": 0.7835355997085571,
      "learning_rate": 2.8614389335914763e-09,
      "loss": 0.4967,
      "step": 5728
    },
    {
      "epoch": 0.9769781718963165,
      "grad_norm": 0.5099844336509705,
      "learning_rate": 2.819838129607844e-09,
      "loss": 0.4944,
      "step": 5729
    },
    {
      "epoch": 0.9771487039563438,
      "grad_norm": 0.5194500684738159,
      "learning_rate": 2.7785415221073257e-09,
      "loss": 0.4908,
      "step": 5730
    },
    {
      "epoch": 0.977319236016371,
      "grad_norm": 0.5863021016120911,
      "learning_rate": 2.737549123687844e-09,
      "loss": 0.5031,
      "step": 5731
    },
    {
      "epoch": 0.9774897680763983,
      "grad_norm": 0.49079203605651855,
      "learning_rate": 2.6968609468543964e-09,
      "loss": 0.4804,
      "step": 5732
    },
    {
      "epoch": 0.9776603001364257,
      "grad_norm": 0.5268674492835999,
      "learning_rate": 2.6564770040190537e-09,
      "loss": 0.5013,
      "step": 5733
    },
    {
      "epoch": 0.977830832196453,
      "grad_norm": 0.3950529396533966,
      "learning_rate": 2.616397307501295e-09,
      "loss": 0.491,
      "step": 5734
    },
    {
      "epoch": 0.9780013642564802,
      "grad_norm": 0.5526773929595947,
      "learning_rate": 2.5766218695277845e-09,
      "loss": 0.4882,
      "step": 5735
    },
    {
      "epoch": 0.9781718963165075,
      "grad_norm": 0.6158766150474548,
      "learning_rate": 2.537150702232261e-09,
      "loss": 0.4936,
      "step": 5736
    },
    {
      "epoch": 0.9783424283765347,
      "grad_norm": 0.46131011843681335,
      "learning_rate": 2.497983817655536e-09,
      "loss": 0.4892,
      "step": 5737
    },
    {
      "epoch": 0.9785129604365621,
      "grad_norm": 0.5086356401443481,
      "learning_rate": 2.4591212277457202e-09,
      "loss": 0.4899,
      "step": 5738
    },
    {
      "epoch": 0.9786834924965894,
      "grad_norm": 0.5378406643867493,
      "learning_rate": 2.420562944358329e-09,
      "loss": 0.4934,
      "step": 5739
    },
    {
      "epoch": 0.9788540245566166,
      "grad_norm": 0.4971339702606201,
      "learning_rate": 2.3823089792556212e-09,
      "loss": 0.5008,
      "step": 5740
    },
    {
      "epoch": 0.9790245566166439,
      "grad_norm": 0.5195417404174805,
      "learning_rate": 2.3443593441073723e-09,
      "loss": 0.4954,
      "step": 5741
    },
    {
      "epoch": 0.9791950886766713,
      "grad_norm": 0.43583396077156067,
      "learning_rate": 2.306714050490211e-09,
      "loss": 0.4881,
      "step": 5742
    },
    {
      "epoch": 0.9793656207366985,
      "grad_norm": 0.39053624868392944,
      "learning_rate": 2.2693731098881734e-09,
      "loss": 0.4844,
      "step": 5743
    },
    {
      "epoch": 0.9795361527967258,
      "grad_norm": 0.4212055802345276,
      "learning_rate": 2.2323365336924807e-09,
      "loss": 0.4957,
      "step": 5744
    },
    {
      "epoch": 0.979706684856753,
      "grad_norm": 0.4515783488750458,
      "learning_rate": 2.1956043332010957e-09,
      "loss": 0.4839,
      "step": 5745
    },
    {
      "epoch": 0.9798772169167803,
      "grad_norm": 0.41012388467788696,
      "learning_rate": 2.1591765196197213e-09,
      "loss": 0.4913,
      "step": 5746
    },
    {
      "epoch": 0.9800477489768077,
      "grad_norm": 0.48675966262817383,
      "learning_rate": 2.1230531040606906e-09,
      "loss": 0.498,
      "step": 5747
    },
    {
      "epoch": 0.980218281036835,
      "grad_norm": 0.644092857837677,
      "learning_rate": 2.0872340975438556e-09,
      "loss": 0.4929,
      "step": 5748
    },
    {
      "epoch": 0.9803888130968622,
      "grad_norm": 0.5100440979003906,
      "learning_rate": 2.0517195109959198e-09,
      "loss": 0.4862,
      "step": 5749
    },
    {
      "epoch": 0.9805593451568895,
      "grad_norm": 0.43121039867401123,
      "learning_rate": 2.016509355250884e-09,
      "loss": 0.4912,
      "step": 5750
    },
    {
      "epoch": 0.9807298772169167,
      "grad_norm": 0.5692516565322876,
      "learning_rate": 1.981603641049934e-09,
      "loss": 0.4878,
      "step": 5751
    },
    {
      "epoch": 0.9809004092769441,
      "grad_norm": 0.4968946874141693,
      "learning_rate": 1.9470023790411077e-09,
      "loss": 0.4923,
      "step": 5752
    },
    {
      "epoch": 0.9810709413369714,
      "grad_norm": 0.4381271302700043,
      "learning_rate": 1.912705579779961e-09,
      "loss": 0.4882,
      "step": 5753
    },
    {
      "epoch": 0.9812414733969986,
      "grad_norm": 0.5335595011711121,
      "learning_rate": 1.8787132537287923e-09,
      "loss": 0.5084,
      "step": 5754
    },
    {
      "epoch": 0.9814120054570259,
      "grad_norm": 0.40526238083839417,
      "learning_rate": 1.8450254112571954e-09,
      "loss": 0.4812,
      "step": 5755
    },
    {
      "epoch": 0.9815825375170532,
      "grad_norm": 0.5060843229293823,
      "learning_rate": 1.8116420626420606e-09,
      "loss": 0.4897,
      "step": 5756
    },
    {
      "epoch": 0.9817530695770805,
      "grad_norm": 0.5042193531990051,
      "learning_rate": 1.77856321806702e-09,
      "loss": 0.4972,
      "step": 5757
    },
    {
      "epoch": 0.9819236016371078,
      "grad_norm": 0.44039374589920044,
      "learning_rate": 1.7457888876230012e-09,
      "loss": 0.4933,
      "step": 5758
    },
    {
      "epoch": 0.982094133697135,
      "grad_norm": 0.6233049035072327,
      "learning_rate": 1.7133190813081182e-09,
      "loss": 0.4858,
      "step": 5759
    },
    {
      "epoch": 0.9822646657571623,
      "grad_norm": 0.39624595642089844,
      "learning_rate": 1.6811538090274473e-09,
      "loss": 0.4862,
      "step": 5760
    },
    {
      "epoch": 0.9824351978171897,
      "grad_norm": 0.437248557806015,
      "learning_rate": 1.6492930805932507e-09,
      "loss": 0.507,
      "step": 5761
    },
    {
      "epoch": 0.982605729877217,
      "grad_norm": 0.5529453754425049,
      "learning_rate": 1.617736905724754e-09,
      "loss": 0.4903,
      "step": 5762
    },
    {
      "epoch": 0.9827762619372442,
      "grad_norm": 0.412514865398407,
      "learning_rate": 1.5864852940485899e-09,
      "loss": 0.4861,
      "step": 5763
    },
    {
      "epoch": 0.9829467939972715,
      "grad_norm": 0.40067386627197266,
      "learning_rate": 1.5555382550981325e-09,
      "loss": 0.5007,
      "step": 5764
    },
    {
      "epoch": 0.9831173260572987,
      "grad_norm": 0.6422335505485535,
      "learning_rate": 1.5248957983139413e-09,
      "loss": 0.4917,
      "step": 5765
    },
    {
      "epoch": 0.9832878581173261,
      "grad_norm": 0.5342501997947693,
      "learning_rate": 1.4945579330437611e-09,
      "loss": 0.4907,
      "step": 5766
    },
    {
      "epoch": 0.9834583901773534,
      "grad_norm": 0.48356303572654724,
      "learning_rate": 1.464524668542411e-09,
      "loss": 0.4908,
      "step": 5767
    },
    {
      "epoch": 0.9836289222373806,
      "grad_norm": 0.5423564910888672,
      "learning_rate": 1.4347960139716734e-09,
      "loss": 0.491,
      "step": 5768
    },
    {
      "epoch": 0.9837994542974079,
      "grad_norm": 0.39374396204948425,
      "learning_rate": 1.405371978400516e-09,
      "loss": 0.493,
      "step": 5769
    },
    {
      "epoch": 0.9839699863574352,
      "grad_norm": 0.5703930854797363,
      "learning_rate": 1.3762525708048703e-09,
      "loss": 0.4961,
      "step": 5770
    },
    {
      "epoch": 0.9841405184174625,
      "grad_norm": 0.5441506505012512,
      "learning_rate": 1.3474378000678524e-09,
      "loss": 0.4895,
      "step": 5771
    },
    {
      "epoch": 0.9843110504774898,
      "grad_norm": 0.498138964176178,
      "learning_rate": 1.3189276749795422e-09,
      "loss": 0.4976,
      "step": 5772
    },
    {
      "epoch": 0.984481582537517,
      "grad_norm": 0.43553170561790466,
      "learning_rate": 1.290722204237316e-09,
      "loss": 0.4865,
      "step": 5773
    },
    {
      "epoch": 0.9846521145975443,
      "grad_norm": 0.45070740580558777,
      "learning_rate": 1.2628213964452908e-09,
      "loss": 0.4881,
      "step": 5774
    },
    {
      "epoch": 0.9848226466575716,
      "grad_norm": 0.7565078735351562,
      "learning_rate": 1.2352252601147697e-09,
      "loss": 0.4945,
      "step": 5775
    },
    {
      "epoch": 0.984993178717599,
      "grad_norm": 0.4107791781425476,
      "learning_rate": 1.207933803664241e-09,
      "loss": 0.4875,
      "step": 5776
    },
    {
      "epoch": 0.9851637107776262,
      "grad_norm": 0.5069323182106018,
      "learning_rate": 1.1809470354190446e-09,
      "loss": 0.5019,
      "step": 5777
    },
    {
      "epoch": 0.9853342428376535,
      "grad_norm": 0.49351727962493896,
      "learning_rate": 1.1542649636118176e-09,
      "loss": 0.4886,
      "step": 5778
    },
    {
      "epoch": 0.9855047748976807,
      "grad_norm": 0.463966429233551,
      "learning_rate": 1.127887596381938e-09,
      "loss": 0.4981,
      "step": 5779
    },
    {
      "epoch": 0.985675306957708,
      "grad_norm": 0.371059775352478,
      "learning_rate": 1.1018149417761915e-09,
      "loss": 0.4928,
      "step": 5780
    },
    {
      "epoch": 0.9858458390177354,
      "grad_norm": 0.4316602349281311,
      "learning_rate": 1.0760470077479935e-09,
      "loss": 0.4868,
      "step": 5781
    },
    {
      "epoch": 0.9860163710777626,
      "grad_norm": 0.42240771651268005,
      "learning_rate": 1.0505838021581672e-09,
      "loss": 0.4803,
      "step": 5782
    },
    {
      "epoch": 0.9861869031377899,
      "grad_norm": 0.3554317057132721,
      "learning_rate": 1.025425332774388e-09,
      "loss": 0.5001,
      "step": 5783
    },
    {
      "epoch": 0.9863574351978172,
      "grad_norm": 0.5055869817733765,
      "learning_rate": 1.0005716072714057e-09,
      "loss": 0.4855,
      "step": 5784
    },
    {
      "epoch": 0.9865279672578445,
      "grad_norm": 0.38027331233024597,
      "learning_rate": 9.76022633231155e-10,
      "loss": 0.4928,
      "step": 5785
    },
    {
      "epoch": 0.9866984993178718,
      "grad_norm": 0.5682066082954407,
      "learning_rate": 9.51778418142202e-10,
      "loss": 0.4864,
      "step": 5786
    },
    {
      "epoch": 0.986869031377899,
      "grad_norm": 0.5194091200828552,
      "learning_rate": 9.278389694006296e-10,
      "loss": 0.4938,
      "step": 5787
    },
    {
      "epoch": 0.9870395634379263,
      "grad_norm": 0.59715735912323,
      "learning_rate": 9.042042943092634e-10,
      "loss": 0.4848,
      "step": 5788
    },
    {
      "epoch": 0.9872100954979536,
      "grad_norm": 0.5079013109207153,
      "learning_rate": 8.808744000780023e-10,
      "loss": 0.4793,
      "step": 5789
    },
    {
      "epoch": 0.987380627557981,
      "grad_norm": 0.5454519391059875,
      "learning_rate": 8.578492938238204e-10,
      "loss": 0.4825,
      "step": 5790
    },
    {
      "epoch": 0.9875511596180082,
      "grad_norm": 0.6010236740112305,
      "learning_rate": 8.351289825705434e-10,
      "loss": 0.49,
      "step": 5791
    },
    {
      "epoch": 0.9877216916780355,
      "grad_norm": 0.4637555480003357,
      "learning_rate": 8.127134732494047e-10,
      "loss": 0.4798,
      "step": 5792
    },
    {
      "epoch": 0.9878922237380627,
      "grad_norm": 0.7311561107635498,
      "learning_rate": 7.906027726981569e-10,
      "loss": 0.4778,
      "step": 5793
    },
    {
      "epoch": 0.98806275579809,
      "grad_norm": 0.5522876977920532,
      "learning_rate": 7.687968876619601e-10,
      "loss": 0.4859,
      "step": 5794
    },
    {
      "epoch": 0.9882332878581174,
      "grad_norm": 0.4935724437236786,
      "learning_rate": 7.472958247928264e-10,
      "loss": 0.5005,
      "step": 5795
    },
    {
      "epoch": 0.9884038199181446,
      "grad_norm": 0.544166088104248,
      "learning_rate": 7.260995906497315e-10,
      "loss": 0.4781,
      "step": 5796
    },
    {
      "epoch": 0.9885743519781719,
      "grad_norm": 0.44626158475875854,
      "learning_rate": 7.052081916988363e-10,
      "loss": 0.4846,
      "step": 5797
    },
    {
      "epoch": 0.9887448840381992,
      "grad_norm": 0.5568285584449768,
      "learning_rate": 6.84621634313043e-10,
      "loss": 0.4868,
      "step": 5798
    },
    {
      "epoch": 0.9889154160982264,
      "grad_norm": 0.4396195113658905,
      "learning_rate": 6.643399247725502e-10,
      "loss": 0.489,
      "step": 5799
    },
    {
      "epoch": 0.9890859481582538,
      "grad_norm": 0.4277687668800354,
      "learning_rate": 6.443630692645198e-10,
      "loss": 0.475,
      "step": 5800
    },
    {
      "epoch": 0.989256480218281,
      "grad_norm": 0.491654634475708,
      "learning_rate": 6.246910738827439e-10,
      "loss": 0.4924,
      "step": 5801
    },
    {
      "epoch": 0.9894270122783083,
      "grad_norm": 0.4532473683357239,
      "learning_rate": 6.053239446285331e-10,
      "loss": 0.4878,
      "step": 5802
    },
    {
      "epoch": 0.9895975443383356,
      "grad_norm": 0.3788377642631531,
      "learning_rate": 5.862616874099392e-10,
      "loss": 0.4857,
      "step": 5803
    },
    {
      "epoch": 0.9897680763983628,
      "grad_norm": 0.4752003848552704,
      "learning_rate": 5.675043080418663e-10,
      "loss": 0.4864,
      "step": 5804
    },
    {
      "epoch": 0.9899386084583902,
      "grad_norm": 0.4157637655735016,
      "learning_rate": 5.49051812246515e-10,
      "loss": 0.483,
      "step": 5805
    },
    {
      "epoch": 0.9901091405184175,
      "grad_norm": 0.3814486563205719,
      "learning_rate": 5.309042056529379e-10,
      "loss": 0.477,
      "step": 5806
    },
    {
      "epoch": 0.9902796725784447,
      "grad_norm": 0.407847136259079,
      "learning_rate": 5.130614937971512e-10,
      "loss": 0.4838,
      "step": 5807
    },
    {
      "epoch": 0.990450204638472,
      "grad_norm": 0.3975071310997009,
      "learning_rate": 4.955236821222454e-10,
      "loss": 0.4962,
      "step": 5808
    },
    {
      "epoch": 0.9906207366984994,
      "grad_norm": 0.4644838571548462,
      "learning_rate": 4.78290775978163e-10,
      "loss": 0.4848,
      "step": 5809
    },
    {
      "epoch": 0.9907912687585266,
      "grad_norm": 0.4575129747390747,
      "learning_rate": 4.6136278062203225e-10,
      "loss": 0.4894,
      "step": 5810
    },
    {
      "epoch": 0.9909618008185539,
      "grad_norm": 0.5090674161911011,
      "learning_rate": 4.447397012177224e-10,
      "loss": 0.4721,
      "step": 5811
    },
    {
      "epoch": 0.9911323328785812,
      "grad_norm": 0.5588932037353516,
      "learning_rate": 4.284215428361771e-10,
      "loss": 0.4961,
      "step": 5812
    },
    {
      "epoch": 0.9913028649386084,
      "grad_norm": 0.39936065673828125,
      "learning_rate": 4.1240831045552546e-10,
      "loss": 0.4954,
      "step": 5813
    },
    {
      "epoch": 0.9914733969986358,
      "grad_norm": 0.5699079036712646,
      "learning_rate": 3.967000089607487e-10,
      "loss": 0.4795,
      "step": 5814
    },
    {
      "epoch": 0.991643929058663,
      "grad_norm": 0.4913797378540039,
      "learning_rate": 3.812966431435694e-10,
      "loss": 0.483,
      "step": 5815
    },
    {
      "epoch": 0.9918144611186903,
      "grad_norm": 0.5006852149963379,
      "learning_rate": 3.6619821770289557e-10,
      "loss": 0.4835,
      "step": 5816
    },
    {
      "epoch": 0.9919849931787176,
      "grad_norm": 0.6642242670059204,
      "learning_rate": 3.514047372448204e-10,
      "loss": 0.4936,
      "step": 5817
    },
    {
      "epoch": 0.9921555252387448,
      "grad_norm": 0.4854762554168701,
      "learning_rate": 3.3691620628195646e-10,
      "loss": 0.481,
      "step": 5818
    },
    {
      "epoch": 0.9923260572987722,
      "grad_norm": 0.6153271794319153,
      "learning_rate": 3.227326292343236e-10,
      "loss": 0.4844,
      "step": 5819
    },
    {
      "epoch": 0.9924965893587995,
      "grad_norm": 0.4837683439254761,
      "learning_rate": 3.08854010428572e-10,
      "loss": 0.4844,
      "step": 5820
    },
    {
      "epoch": 0.9926671214188267,
      "grad_norm": 0.45445963740348816,
      "learning_rate": 2.952803540985372e-10,
      "loss": 0.4774,
      "step": 5821
    },
    {
      "epoch": 0.992837653478854,
      "grad_norm": 0.614289402961731,
      "learning_rate": 2.820116643849069e-10,
      "loss": 0.4912,
      "step": 5822
    },
    {
      "epoch": 0.9930081855388813,
      "grad_norm": 0.4718392491340637,
      "learning_rate": 2.6904794533544336e-10,
      "loss": 0.489,
      "step": 5823
    },
    {
      "epoch": 0.9931787175989086,
      "grad_norm": 0.5621843934059143,
      "learning_rate": 2.5638920090476083e-10,
      "loss": 0.4929,
      "step": 5824
    },
    {
      "epoch": 0.9933492496589359,
      "grad_norm": 0.4721529185771942,
      "learning_rate": 2.440354349545482e-10,
      "loss": 0.4777,
      "step": 5825
    },
    {
      "epoch": 0.9935197817189632,
      "grad_norm": 0.39175862073898315,
      "learning_rate": 2.3198665125334643e-10,
      "loss": 0.4988,
      "step": 5826
    },
    {
      "epoch": 0.9936903137789904,
      "grad_norm": 0.48671218752861023,
      "learning_rate": 2.202428534767709e-10,
      "loss": 0.4873,
      "step": 5827
    },
    {
      "epoch": 0.9938608458390177,
      "grad_norm": 0.49351176619529724,
      "learning_rate": 2.0880404520728926e-10,
      "loss": 0.491,
      "step": 5828
    },
    {
      "epoch": 0.994031377899045,
      "grad_norm": 0.4758182168006897,
      "learning_rate": 1.9767022993444352e-10,
      "loss": 0.4853,
      "step": 5829
    },
    {
      "epoch": 0.9942019099590723,
      "grad_norm": 0.3767922520637512,
      "learning_rate": 1.8684141105462796e-10,
      "loss": 0.4894,
      "step": 5830
    },
    {
      "epoch": 0.9943724420190996,
      "grad_norm": 0.4667741060256958,
      "learning_rate": 1.7631759187131117e-10,
      "loss": 0.4783,
      "step": 5831
    },
    {
      "epoch": 0.9945429740791268,
      "grad_norm": 0.4266509711742401,
      "learning_rate": 1.6609877559481405e-10,
      "loss": 0.4867,
      "step": 5832
    },
    {
      "epoch": 0.9947135061391542,
      "grad_norm": 0.4854908287525177,
      "learning_rate": 1.5618496534242079e-10,
      "loss": 0.486,
      "step": 5833
    },
    {
      "epoch": 0.9948840381991815,
      "grad_norm": 0.5938842296600342,
      "learning_rate": 1.4657616413837895e-10,
      "loss": 0.4876,
      "step": 5834
    },
    {
      "epoch": 0.9950545702592087,
      "grad_norm": 0.4664316475391388,
      "learning_rate": 1.372723749141214e-10,
      "loss": 0.4848,
      "step": 5835
    },
    {
      "epoch": 0.995225102319236,
      "grad_norm": 0.44046300649642944,
      "learning_rate": 1.2827360050760018e-10,
      "loss": 0.4853,
      "step": 5836
    },
    {
      "epoch": 0.9953956343792633,
      "grad_norm": 0.5642649531364441,
      "learning_rate": 1.1957984366406384e-10,
      "loss": 0.4956,
      "step": 5837
    },
    {
      "epoch": 0.9955661664392906,
      "grad_norm": 0.7079784870147705,
      "learning_rate": 1.111911070356131e-10,
      "loss": 0.484,
      "step": 5838
    },
    {
      "epoch": 0.9957366984993179,
      "grad_norm": 0.4249776303768158,
      "learning_rate": 1.03107393181201e-10,
      "loss": 0.481,
      "step": 5839
    },
    {
      "epoch": 0.9959072305593452,
      "grad_norm": 0.5063623785972595,
      "learning_rate": 9.532870456685495e-11,
      "loss": 0.4966,
      "step": 5840
    },
    {
      "epoch": 0.9960777626193724,
      "grad_norm": 0.5348143577575684,
      "learning_rate": 8.785504356556564e-11,
      "loss": 0.4853,
      "step": 5841
    },
    {
      "epoch": 0.9962482946793997,
      "grad_norm": 0.5114048719406128,
      "learning_rate": 8.068641245717602e-11,
      "loss": 0.4763,
      "step": 5842
    },
    {
      "epoch": 0.996418826739427,
      "grad_norm": 0.5542433261871338,
      "learning_rate": 7.38228134284924e-11,
      "loss": 0.4853,
      "step": 5843
    },
    {
      "epoch": 0.9965893587994543,
      "grad_norm": 0.4604029357433319,
      "learning_rate": 6.726424857339543e-11,
      "loss": 0.479,
      "step": 5844
    },
    {
      "epoch": 0.9967598908594816,
      "grad_norm": 0.585250735282898,
      "learning_rate": 6.101071989261798e-11,
      "loss": 0.4891,
      "step": 5845
    },
    {
      "epoch": 0.9969304229195088,
      "grad_norm": 0.46360349655151367,
      "learning_rate": 5.506222929374528e-11,
      "loss": 0.469,
      "step": 5846
    },
    {
      "epoch": 0.9971009549795361,
      "grad_norm": 0.4242927134037018,
      "learning_rate": 4.941877859143685e-11,
      "loss": 0.491,
      "step": 5847
    },
    {
      "epoch": 0.9972714870395635,
      "grad_norm": 0.5808072090148926,
      "learning_rate": 4.408036950731554e-11,
      "loss": 0.4865,
      "step": 5848
    },
    {
      "epoch": 0.9974420190995907,
      "grad_norm": 0.4071265161037445,
      "learning_rate": 3.904700366985648e-11,
      "loss": 0.4936,
      "step": 5849
    },
    {
      "epoch": 0.997612551159618,
      "grad_norm": 0.5109954476356506,
      "learning_rate": 3.431868261449812e-11,
      "loss": 0.4887,
      "step": 5850
    },
    {
      "epoch": 0.9977830832196453,
      "grad_norm": 0.5592871904373169,
      "learning_rate": 2.9895407783753216e-11,
      "loss": 0.4916,
      "step": 5851
    },
    {
      "epoch": 0.9979536152796725,
      "grad_norm": 0.43258005380630493,
      "learning_rate": 2.5777180526764813e-11,
      "loss": 0.4855,
      "step": 5852
    },
    {
      "epoch": 0.9981241473396999,
      "grad_norm": 0.5342608094215393,
      "learning_rate": 2.1964002100083317e-11,
      "loss": 0.4842,
      "step": 5853
    },
    {
      "epoch": 0.9982946793997272,
      "grad_norm": 0.4164401888847351,
      "learning_rate": 1.8455873666778368e-11,
      "loss": 0.4969,
      "step": 5854
    },
    {
      "epoch": 0.9984652114597544,
      "grad_norm": 0.43266117572784424,
      "learning_rate": 1.5252796296993946e-11,
      "loss": 0.4852,
      "step": 5855
    },
    {
      "epoch": 0.9986357435197817,
      "grad_norm": 0.5233867764472961,
      "learning_rate": 1.2354770968059371e-11,
      "loss": 0.4971,
      "step": 5856
    },
    {
      "epoch": 0.998806275579809,
      "grad_norm": 0.42787232995033264,
      "learning_rate": 9.761798563823201e-12,
      "loss": 0.4788,
      "step": 5857
    },
    {
      "epoch": 0.9989768076398363,
      "grad_norm": 0.5615535974502563,
      "learning_rate": 7.473879875319335e-12,
      "loss": 0.4991,
      "step": 5858
    },
    {
      "epoch": 0.9991473396998636,
      "grad_norm": 0.47109538316726685,
      "learning_rate": 5.491015600656014e-12,
      "loss": 0.4858,
      "step": 5859
    },
    {
      "epoch": 0.9993178717598908,
      "grad_norm": 0.5316367149353027,
      "learning_rate": 3.813206344571719e-12,
      "loss": 0.479,
      "step": 5860
    },
    {
      "epoch": 0.9994884038199181,
      "grad_norm": 0.5030397176742554,
      "learning_rate": 2.4404526188792634e-12,
      "loss": 0.4877,
      "step": 5861
    },
    {
      "epoch": 0.9996589358799455,
      "grad_norm": 0.398891419172287,
      "learning_rate": 1.3727548423547732e-12,
      "loss": 0.4851,
      "step": 5862
    },
    {
      "epoch": 0.9998294679399727,
      "grad_norm": 0.41367173194885254,
      "learning_rate": 6.101133408487059e-13,
      "loss": 0.4926,
      "step": 5863
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4404045343399048,
      "learning_rate": 1.5252834684176266e-13,
      "loss": 0.483,
      "step": 5864
    },
    {
      "epoch": 1.0,
      "step": 5864,
      "total_flos": 5.461479216500315e+20,
      "train_loss": 0.09150631395949999,
      "train_runtime": 9502.0046,
      "train_samples_per_second": 17.525,
      "train_steps_per_second": 0.617
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 5864,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.461479216500315e+20,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}