{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9995676610462603,
  "eval_steps": 500,
  "global_step": 578,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0017293558149589277,
      "grad_norm": 14.538986412992182,
      "learning_rate": 0.0,
      "loss": 1.6313,
      "step": 1
    },
    {
      "epoch": 0.0034587116299178555,
      "grad_norm": 14.326584895058424,
      "learning_rate": 5.555555555555555e-07,
      "loss": 1.5796,
      "step": 2
    },
    {
      "epoch": 0.005188067444876783,
      "grad_norm": 14.303713845230615,
      "learning_rate": 1.111111111111111e-06,
      "loss": 1.6146,
      "step": 3
    },
    {
      "epoch": 0.006917423259835711,
      "grad_norm": 14.270354087741836,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.601,
      "step": 4
    },
    {
      "epoch": 0.00864677907479464,
      "grad_norm": 13.056520461824615,
      "learning_rate": 2.222222222222222e-06,
      "loss": 1.5589,
      "step": 5
    },
    {
      "epoch": 0.010376134889753566,
      "grad_norm": 10.770002519875254,
      "learning_rate": 2.7777777777777783e-06,
      "loss": 1.4667,
      "step": 6
    },
    {
      "epoch": 0.012105490704712495,
      "grad_norm": 7.95324664209596,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.2753,
      "step": 7
    },
    {
      "epoch": 0.013834846519671422,
      "grad_norm": 7.532832624580702,
      "learning_rate": 3.88888888888889e-06,
      "loss": 1.2647,
      "step": 8
    },
    {
      "epoch": 0.01556420233463035,
      "grad_norm": 5.2620835279629645,
      "learning_rate": 4.444444444444444e-06,
      "loss": 1.1498,
      "step": 9
    },
    {
      "epoch": 0.01729355814958928,
      "grad_norm": 5.262371985501062,
      "learning_rate": 5e-06,
      "loss": 1.0606,
      "step": 10
    },
    {
      "epoch": 0.019022913964548204,
      "grad_norm": 4.75552495213558,
      "learning_rate": 5.555555555555557e-06,
      "loss": 1.039,
      "step": 11
    },
    {
      "epoch": 0.020752269779507133,
      "grad_norm": 4.109629839647021,
      "learning_rate": 6.111111111111112e-06,
      "loss": 0.9838,
      "step": 12
    },
    {
      "epoch": 0.02248162559446606,
      "grad_norm": 3.2895830782411433,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.9485,
      "step": 13
    },
    {
      "epoch": 0.02421098140942499,
      "grad_norm": 3.4515341118841554,
      "learning_rate": 7.222222222222223e-06,
      "loss": 0.8929,
      "step": 14
    },
    {
      "epoch": 0.02594033722438392,
      "grad_norm": 3.5675009192020917,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.8761,
      "step": 15
    },
    {
      "epoch": 0.027669693039342844,
      "grad_norm": 3.1238221221969846,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.8579,
      "step": 16
    },
    {
      "epoch": 0.029399048854301772,
      "grad_norm": 2.730324810433143,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.825,
      "step": 17
    },
    {
      "epoch": 0.0311284046692607,
      "grad_norm": 2.142543702234471,
      "learning_rate": 9.444444444444445e-06,
      "loss": 0.8275,
      "step": 18
    },
    {
      "epoch": 0.03285776048421963,
      "grad_norm": 2.2407490739694618,
      "learning_rate": 1e-05,
      "loss": 0.798,
      "step": 19
    },
    {
      "epoch": 0.03458711629917856,
      "grad_norm": 2.084673650668219,
      "learning_rate": 9.999921320324328e-06,
      "loss": 0.7791,
      "step": 20
    },
    {
      "epoch": 0.03631647211413749,
      "grad_norm": 2.2050652721508817,
      "learning_rate": 9.999685283773504e-06,
      "loss": 0.801,
      "step": 21
    },
    {
      "epoch": 0.03804582792909641,
      "grad_norm": 1.9949942374964489,
      "learning_rate": 9.999291897776043e-06,
      "loss": 0.7673,
      "step": 22
    },
    {
      "epoch": 0.03977518374405534,
      "grad_norm": 1.9222200778544485,
      "learning_rate": 9.998741174712534e-06,
      "loss": 0.7546,
      "step": 23
    },
    {
      "epoch": 0.041504539559014265,
      "grad_norm": 2.085559689480793,
      "learning_rate": 9.998033131915266e-06,
      "loss": 0.7733,
      "step": 24
    },
    {
      "epoch": 0.043233895373973194,
      "grad_norm": 1.731301007495722,
      "learning_rate": 9.997167791667668e-06,
      "loss": 0.7406,
      "step": 25
    },
    {
      "epoch": 0.04496325118893212,
      "grad_norm": 1.9090014318422763,
      "learning_rate": 9.996145181203616e-06,
      "loss": 0.7478,
      "step": 26
    },
    {
      "epoch": 0.04669260700389105,
      "grad_norm": 1.6716462018458187,
      "learning_rate": 9.994965332706574e-06,
      "loss": 0.7293,
      "step": 27
    },
    {
      "epoch": 0.04842196281884998,
      "grad_norm": 1.641744923224004,
      "learning_rate": 9.993628283308582e-06,
      "loss": 0.7157,
      "step": 28
    },
    {
      "epoch": 0.05015131863380891,
      "grad_norm": 1.5639464733995954,
      "learning_rate": 9.992134075089085e-06,
      "loss": 0.7337,
      "step": 29
    },
    {
      "epoch": 0.05188067444876784,
      "grad_norm": 1.5670396520195686,
      "learning_rate": 9.990482755073607e-06,
      "loss": 0.6944,
      "step": 30
    },
    {
      "epoch": 0.05361003026372676,
      "grad_norm": 1.5501858401764228,
      "learning_rate": 9.98867437523228e-06,
      "loss": 0.7382,
      "step": 31
    },
    {
      "epoch": 0.05533938607868569,
      "grad_norm": 1.5475730980812552,
      "learning_rate": 9.986708992478202e-06,
      "loss": 0.6836,
      "step": 32
    },
    {
      "epoch": 0.057068741893644616,
      "grad_norm": 1.5368459860394041,
      "learning_rate": 9.984586668665641e-06,
      "loss": 0.7309,
      "step": 33
    },
    {
      "epoch": 0.058798097708603544,
      "grad_norm": 1.573756977286856,
      "learning_rate": 9.982307470588097e-06,
      "loss": 0.6921,
      "step": 34
    },
    {
      "epoch": 0.06052745352356247,
      "grad_norm": 1.4598653582851544,
      "learning_rate": 9.979871469976197e-06,
      "loss": 0.6824,
      "step": 35
    },
    {
      "epoch": 0.0622568093385214,
      "grad_norm": 1.3407147196823341,
      "learning_rate": 9.977278743495434e-06,
      "loss": 0.695,
      "step": 36
    },
    {
      "epoch": 0.06398616515348032,
      "grad_norm": 1.3427046109275345,
      "learning_rate": 9.974529372743762e-06,
      "loss": 0.6849,
      "step": 37
    },
    {
      "epoch": 0.06571552096843926,
      "grad_norm": 1.385016607008538,
      "learning_rate": 9.97162344424902e-06,
      "loss": 0.704,
      "step": 38
    },
    {
      "epoch": 0.06744487678339818,
      "grad_norm": 1.3624925480173673,
      "learning_rate": 9.968561049466214e-06,
      "loss": 0.6648,
      "step": 39
    },
    {
      "epoch": 0.06917423259835712,
      "grad_norm": 1.6070208118878249,
      "learning_rate": 9.965342284774633e-06,
      "loss": 0.7004,
      "step": 40
    },
    {
      "epoch": 0.07090358841331604,
      "grad_norm": 1.4348520811083416,
      "learning_rate": 9.961967251474823e-06,
      "loss": 0.6625,
      "step": 41
    },
    {
      "epoch": 0.07263294422827497,
      "grad_norm": 1.3921519261612385,
      "learning_rate": 9.958436055785391e-06,
      "loss": 0.6709,
      "step": 42
    },
    {
      "epoch": 0.0743623000432339,
      "grad_norm": 1.3214257024283271,
      "learning_rate": 9.954748808839675e-06,
      "loss": 0.6526,
      "step": 43
    },
    {
      "epoch": 0.07609165585819282,
      "grad_norm": 1.3703861516781,
      "learning_rate": 9.950905626682229e-06,
      "loss": 0.6485,
      "step": 44
    },
    {
      "epoch": 0.07782101167315175,
      "grad_norm": 1.3610898546796686,
      "learning_rate": 9.946906630265184e-06,
      "loss": 0.6509,
      "step": 45
    },
    {
      "epoch": 0.07955036748811067,
      "grad_norm": 1.4666639612285957,
      "learning_rate": 9.942751945444437e-06,
      "loss": 0.6575,
      "step": 46
    },
    {
      "epoch": 0.08127972330306961,
      "grad_norm": 1.2396616063563637,
      "learning_rate": 9.938441702975689e-06,
      "loss": 0.6395,
      "step": 47
    },
    {
      "epoch": 0.08300907911802853,
      "grad_norm": 1.4092903953844345,
      "learning_rate": 9.933976038510334e-06,
      "loss": 0.6554,
      "step": 48
    },
    {
      "epoch": 0.08473843493298747,
      "grad_norm": 1.7100974038421335,
      "learning_rate": 9.92935509259118e-06,
      "loss": 0.6528,
      "step": 49
    },
    {
      "epoch": 0.08646779074794639,
      "grad_norm": 1.4242700514926885,
      "learning_rate": 9.924579010648042e-06,
      "loss": 0.6486,
      "step": 50
    },
    {
      "epoch": 0.08819714656290532,
      "grad_norm": 1.5209362334431213,
      "learning_rate": 9.91964794299315e-06,
      "loss": 0.6773,
      "step": 51
    },
    {
      "epoch": 0.08992650237786425,
      "grad_norm": 1.4185285804137782,
      "learning_rate": 9.914562044816424e-06,
      "loss": 0.6488,
      "step": 52
    },
    {
      "epoch": 0.09165585819282317,
      "grad_norm": 1.3378741936637544,
      "learning_rate": 9.909321476180594e-06,
      "loss": 0.6517,
      "step": 53
    },
    {
      "epoch": 0.0933852140077821,
      "grad_norm": 1.3516713188326783,
      "learning_rate": 9.903926402016153e-06,
      "loss": 0.6396,
      "step": 54
    },
    {
      "epoch": 0.09511456982274102,
      "grad_norm": 1.5410717485635828,
      "learning_rate": 9.898376992116179e-06,
      "loss": 0.6447,
      "step": 55
    },
    {
      "epoch": 0.09684392563769996,
      "grad_norm": 1.5098400169635777,
      "learning_rate": 9.892673421130979e-06,
      "loss": 0.6342,
      "step": 56
    },
    {
      "epoch": 0.09857328145265888,
      "grad_norm": 1.4229655753415995,
      "learning_rate": 9.886815868562596e-06,
      "loss": 0.623,
      "step": 57
    },
    {
      "epoch": 0.10030263726761782,
      "grad_norm": 1.3977945588526572,
      "learning_rate": 9.88080451875917e-06,
      "loss": 0.6397,
      "step": 58
    },
    {
      "epoch": 0.10203199308257674,
      "grad_norm": 1.2877794451825768,
      "learning_rate": 9.874639560909118e-06,
      "loss": 0.6697,
      "step": 59
    },
    {
      "epoch": 0.10376134889753567,
      "grad_norm": 1.4091354007797192,
      "learning_rate": 9.868321189035196e-06,
      "loss": 0.6359,
      "step": 60
    },
    {
      "epoch": 0.1054907047124946,
      "grad_norm": 1.362969728293516,
      "learning_rate": 9.861849601988384e-06,
      "loss": 0.6418,
      "step": 61
    },
    {
      "epoch": 0.10722006052745352,
      "grad_norm": 1.453550759696413,
      "learning_rate": 9.855225003441629e-06,
      "loss": 0.6469,
      "step": 62
    },
    {
      "epoch": 0.10894941634241245,
      "grad_norm": 1.4051406418213799,
      "learning_rate": 9.848447601883436e-06,
      "loss": 0.6257,
      "step": 63
    },
    {
      "epoch": 0.11067877215737137,
      "grad_norm": 1.3211083898505493,
      "learning_rate": 9.841517610611309e-06,
      "loss": 0.6396,
      "step": 64
    },
    {
      "epoch": 0.11240812797233031,
      "grad_norm": 1.3740318481622646,
      "learning_rate": 9.834435247725032e-06,
      "loss": 0.6259,
      "step": 65
    },
    {
      "epoch": 0.11413748378728923,
      "grad_norm": 1.484320978899618,
      "learning_rate": 9.827200736119815e-06,
      "loss": 0.6219,
      "step": 66
    },
    {
      "epoch": 0.11586683960224817,
      "grad_norm": 1.3380643332618636,
      "learning_rate": 9.819814303479268e-06,
      "loss": 0.6192,
      "step": 67
    },
    {
      "epoch": 0.11759619541720709,
      "grad_norm": 1.4153726476019632,
      "learning_rate": 9.812276182268236e-06,
      "loss": 0.617,
      "step": 68
    },
    {
      "epoch": 0.11932555123216602,
      "grad_norm": 1.435579910730408,
      "learning_rate": 9.804586609725499e-06,
      "loss": 0.6082,
      "step": 69
    },
    {
      "epoch": 0.12105490704712495,
      "grad_norm": 1.3611927229844167,
      "learning_rate": 9.79674582785628e-06,
      "loss": 0.6234,
      "step": 70
    },
    {
      "epoch": 0.12278426286208387,
      "grad_norm": 1.4203295780177876,
      "learning_rate": 9.788754083424654e-06,
      "loss": 0.6333,
      "step": 71
    },
    {
      "epoch": 0.1245136186770428,
      "grad_norm": 1.29595870212252,
      "learning_rate": 9.78061162794576e-06,
      "loss": 0.6267,
      "step": 72
    },
    {
      "epoch": 0.12624297449200172,
      "grad_norm": 1.5861725748430446,
      "learning_rate": 9.772318717677905e-06,
      "loss": 0.6008,
      "step": 73
    },
    {
      "epoch": 0.12797233030696065,
      "grad_norm": 1.3871875221331877,
      "learning_rate": 9.763875613614482e-06,
      "loss": 0.6099,
      "step": 74
    },
    {
      "epoch": 0.1297016861219196,
      "grad_norm": 1.3701024745992456,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.6165,
      "step": 75
    },
    {
      "epoch": 0.13143104193687852,
      "grad_norm": 1.3461257221248066,
      "learning_rate": 9.746539891700558e-06,
      "loss": 0.6329,
      "step": 76
    },
    {
      "epoch": 0.13316039775183744,
      "grad_norm": 1.3814830311022952,
      "learning_rate": 9.737647819437645e-06,
      "loss": 0.6238,
      "step": 77
    },
    {
      "epoch": 0.13488975356679636,
      "grad_norm": 1.3317024763543674,
      "learning_rate": 9.728606644537177e-06,
      "loss": 0.6436,
      "step": 78
    },
    {
      "epoch": 0.13661910938175528,
      "grad_norm": 1.3559610943853933,
      "learning_rate": 9.719416651541839e-06,
      "loss": 0.6093,
      "step": 79
    },
    {
      "epoch": 0.13834846519671423,
      "grad_norm": 1.3142606469258882,
      "learning_rate": 9.710078129677895e-06,
      "loss": 0.6265,
      "step": 80
    },
    {
      "epoch": 0.14007782101167315,
      "grad_norm": 1.4252162273974214,
      "learning_rate": 9.700591372846096e-06,
      "loss": 0.62,
      "step": 81
    },
    {
      "epoch": 0.14180717682663208,
      "grad_norm": 1.284483722976347,
      "learning_rate": 9.690956679612422e-06,
      "loss": 0.6387,
      "step": 82
    },
    {
      "epoch": 0.143536532641591,
      "grad_norm": 1.4287308609026501,
      "learning_rate": 9.681174353198687e-06,
      "loss": 0.6155,
      "step": 83
    },
    {
      "epoch": 0.14526588845654995,
      "grad_norm": 1.3659377539854851,
      "learning_rate": 9.671244701472999e-06,
      "loss": 0.5813,
      "step": 84
    },
    {
      "epoch": 0.14699524427150887,
      "grad_norm": 1.3254239612936913,
      "learning_rate": 9.661168036940071e-06,
      "loss": 0.607,
      "step": 85
    },
    {
      "epoch": 0.1487246000864678,
      "grad_norm": 1.3325242903071612,
      "learning_rate": 9.650944676731383e-06,
      "loss": 0.5811,
      "step": 86
    },
    {
      "epoch": 0.1504539559014267,
      "grad_norm": 1.3721199226780008,
      "learning_rate": 9.640574942595195e-06,
      "loss": 0.6213,
      "step": 87
    },
    {
      "epoch": 0.15218331171638563,
      "grad_norm": 1.3252154289940847,
      "learning_rate": 9.63005916088644e-06,
      "loss": 0.598,
      "step": 88
    },
    {
      "epoch": 0.15391266753134458,
      "grad_norm": 1.341549271858989,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.6159,
      "step": 89
    },
    {
      "epoch": 0.1556420233463035,
      "grad_norm": 1.306931095537237,
      "learning_rate": 9.608590783142471e-06,
      "loss": 0.5866,
      "step": 90
    },
    {
      "epoch": 0.15737137916126243,
      "grad_norm": 1.3395574538440298,
      "learning_rate": 9.597638862757255e-06,
      "loss": 0.6132,
      "step": 91
    },
    {
      "epoch": 0.15910073497622135,
      "grad_norm": 1.4247495939885102,
      "learning_rate": 9.586542246078203e-06,
      "loss": 0.6033,
      "step": 92
    },
    {
      "epoch": 0.1608300907911803,
      "grad_norm": 1.4103034043856166,
      "learning_rate": 9.5753012823366e-06,
      "loss": 0.5958,
      "step": 93
    },
    {
      "epoch": 0.16255944660613922,
      "grad_norm": 1.2779807076086858,
      "learning_rate": 9.563916325306595e-06,
      "loss": 0.5999,
      "step": 94
    },
    {
      "epoch": 0.16428880242109814,
      "grad_norm": 1.3678860405045183,
      "learning_rate": 9.552387733294081e-06,
      "loss": 0.6047,
      "step": 95
    },
    {
      "epoch": 0.16601815823605706,
      "grad_norm": 1.4177812307852564,
      "learning_rate": 9.540715869125407e-06,
      "loss": 0.6032,
      "step": 96
    },
    {
      "epoch": 0.16774751405101598,
      "grad_norm": 1.3929063773953412,
      "learning_rate": 9.528901100135971e-06,
      "loss": 0.5914,
      "step": 97
    },
    {
      "epoch": 0.16947686986597493,
      "grad_norm": 1.4452904668325175,
      "learning_rate": 9.51694379815865e-06,
      "loss": 0.591,
      "step": 98
    },
    {
      "epoch": 0.17120622568093385,
      "grad_norm": 1.384558049529939,
      "learning_rate": 9.504844339512096e-06,
      "loss": 0.5935,
      "step": 99
    },
    {
      "epoch": 0.17293558149589278,
      "grad_norm": 1.3744626820811583,
      "learning_rate": 9.492603104988907e-06,
      "loss": 0.6093,
      "step": 100
    },
    {
      "epoch": 0.1746649373108517,
      "grad_norm": 1.4015343739030506,
      "learning_rate": 9.480220479843627e-06,
      "loss": 0.5864,
      "step": 101
    },
    {
      "epoch": 0.17639429312581065,
      "grad_norm": 1.3434145487030242,
      "learning_rate": 9.467696853780625e-06,
      "loss": 0.5907,
      "step": 102
    },
    {
      "epoch": 0.17812364894076957,
      "grad_norm": 1.3106599461212003,
      "learning_rate": 9.45503262094184e-06,
      "loss": 0.6035,
      "step": 103
    },
    {
      "epoch": 0.1798530047557285,
      "grad_norm": 1.2960137141000876,
      "learning_rate": 9.442228179894362e-06,
      "loss": 0.58,
      "step": 104
    },
    {
      "epoch": 0.1815823605706874,
      "grad_norm": 1.339610530818961,
      "learning_rate": 9.4292839336179e-06,
      "loss": 0.6019,
      "step": 105
    },
    {
      "epoch": 0.18331171638564633,
      "grad_norm": 1.437718437019135,
      "learning_rate": 9.416200289492092e-06,
      "loss": 0.6098,
      "step": 106
    },
    {
      "epoch": 0.18504107220060528,
      "grad_norm": 1.4523767804936614,
      "learning_rate": 9.40297765928369e-06,
      "loss": 0.6018,
      "step": 107
    },
    {
      "epoch": 0.1867704280155642,
      "grad_norm": 1.2828185912784842,
      "learning_rate": 9.389616459133597e-06,
      "loss": 0.5764,
      "step": 108
    },
    {
      "epoch": 0.18849978383052313,
      "grad_norm": 1.4256098510891433,
      "learning_rate": 9.376117109543769e-06,
      "loss": 0.6159,
      "step": 109
    },
    {
      "epoch": 0.19022913964548205,
      "grad_norm": 1.3501440943806944,
      "learning_rate": 9.362480035363987e-06,
      "loss": 0.5793,
      "step": 110
    },
    {
      "epoch": 0.191958495460441,
      "grad_norm": 1.3361244720373389,
      "learning_rate": 9.348705665778479e-06,
      "loss": 0.5712,
      "step": 111
    },
    {
      "epoch": 0.19368785127539992,
      "grad_norm": 1.2782083486564007,
      "learning_rate": 9.334794434292416e-06,
      "loss": 0.5825,
      "step": 112
    },
    {
      "epoch": 0.19541720709035884,
      "grad_norm": 1.2823908295358897,
      "learning_rate": 9.320746778718274e-06,
      "loss": 0.6332,
      "step": 113
    },
    {
      "epoch": 0.19714656290531776,
      "grad_norm": 1.4254059120553437,
      "learning_rate": 9.306563141162046e-06,
      "loss": 0.6021,
      "step": 114
    },
    {
      "epoch": 0.19887591872027668,
      "grad_norm": 1.3615562993020114,
      "learning_rate": 9.292243968009332e-06,
      "loss": 0.5919,
      "step": 115
    },
    {
      "epoch": 0.20060527453523563,
      "grad_norm": 1.4003633053761828,
      "learning_rate": 9.27778970991129e-06,
      "loss": 0.6122,
      "step": 116
    },
    {
      "epoch": 0.20233463035019456,
      "grad_norm": 1.4139482327306883,
      "learning_rate": 9.263200821770462e-06,
      "loss": 0.5731,
      "step": 117
    },
    {
      "epoch": 0.20406398616515348,
      "grad_norm": 1.4647997165826316,
      "learning_rate": 9.248477762726438e-06,
      "loss": 0.5797,
      "step": 118
    },
    {
      "epoch": 0.2057933419801124,
      "grad_norm": 1.3922473806972735,
      "learning_rate": 9.233620996141421e-06,
      "loss": 0.5598,
      "step": 119
    },
    {
      "epoch": 0.20752269779507135,
      "grad_norm": 1.3498874982995888,
      "learning_rate": 9.218630989585647e-06,
      "loss": 0.5824,
      "step": 120
    },
    {
      "epoch": 0.20925205361003027,
      "grad_norm": 1.4059303390475026,
      "learning_rate": 9.203508214822652e-06,
      "loss": 0.6209,
      "step": 121
    },
    {
      "epoch": 0.2109814094249892,
      "grad_norm": 1.4908066800771167,
      "learning_rate": 9.188253147794443e-06,
      "loss": 0.5784,
      "step": 122
    },
    {
      "epoch": 0.2127107652399481,
      "grad_norm": 1.3718507338485877,
      "learning_rate": 9.172866268606514e-06,
      "loss": 0.5962,
      "step": 123
    },
    {
      "epoch": 0.21444012105490703,
      "grad_norm": 1.402469081391288,
      "learning_rate": 9.157348061512728e-06,
      "loss": 0.5929,
      "step": 124
    },
    {
      "epoch": 0.21616947686986598,
      "grad_norm": 1.3206199086660828,
      "learning_rate": 9.141699014900084e-06,
      "loss": 0.5903,
      "step": 125
    },
    {
      "epoch": 0.2178988326848249,
      "grad_norm": 1.3281498552638782,
      "learning_rate": 9.125919621273348e-06,
      "loss": 0.5946,
      "step": 126
    },
    {
      "epoch": 0.21962818849978383,
      "grad_norm": 1.3209410288937773,
      "learning_rate": 9.110010377239552e-06,
      "loss": 0.5626,
      "step": 127
    },
    {
      "epoch": 0.22135754431474275,
      "grad_norm": 1.3329308119803185,
      "learning_rate": 9.093971783492354e-06,
      "loss": 0.5918,
      "step": 128
    },
    {
      "epoch": 0.2230869001297017,
      "grad_norm": 1.347489334146743,
      "learning_rate": 9.077804344796302e-06,
      "loss": 0.5878,
      "step": 129
    },
    {
      "epoch": 0.22481625594466062,
      "grad_norm": 1.4072905998791443,
      "learning_rate": 9.061508569970926e-06,
      "loss": 0.5895,
      "step": 130
    },
    {
      "epoch": 0.22654561175961954,
      "grad_norm": 1.4623429864651023,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.5935,
      "step": 131
    },
    {
      "epoch": 0.22827496757457846,
      "grad_norm": 1.3789532216954916,
      "learning_rate": 9.028534067389087e-06,
      "loss": 0.5886,
      "step": 132
    },
    {
      "epoch": 0.23000432338953739,
      "grad_norm": 1.3644289851875127,
      "learning_rate": 9.011856377401891e-06,
      "loss": 0.5682,
      "step": 133
    },
    {
      "epoch": 0.23173367920449633,
      "grad_norm": 1.2442423854348275,
      "learning_rate": 8.995052426791247e-06,
      "loss": 0.5684,
      "step": 134
    },
    {
      "epoch": 0.23346303501945526,
      "grad_norm": 1.3245488152280875,
      "learning_rate": 8.978122744408905e-06,
      "loss": 0.5643,
      "step": 135
    },
    {
      "epoch": 0.23519239083441418,
      "grad_norm": 1.3934642875348537,
      "learning_rate": 8.961067863063638e-06,
      "loss": 0.5707,
      "step": 136
    },
    {
      "epoch": 0.2369217466493731,
      "grad_norm": 1.3586108466135645,
      "learning_rate": 8.943888319504456e-06,
      "loss": 0.5936,
      "step": 137
    },
    {
      "epoch": 0.23865110246433205,
      "grad_norm": 1.2787757325817652,
      "learning_rate": 8.926584654403725e-06,
      "loss": 0.5743,
      "step": 138
    },
    {
      "epoch": 0.24038045827929097,
      "grad_norm": 1.2401520618532709,
      "learning_rate": 8.90915741234015e-06,
      "loss": 0.5738,
      "step": 139
    },
    {
      "epoch": 0.2421098140942499,
      "grad_norm": 1.3715834375196405,
      "learning_rate": 8.89160714178163e-06,
      "loss": 0.594,
      "step": 140
    },
    {
      "epoch": 0.2438391699092088,
      "grad_norm": 1.3920239192472779,
      "learning_rate": 8.873934395068006e-06,
      "loss": 0.5891,
      "step": 141
    },
    {
      "epoch": 0.24556852572416774,
      "grad_norm": 1.280071307258292,
      "learning_rate": 8.856139728393667e-06,
      "loss": 0.586,
      "step": 142
    },
    {
      "epoch": 0.24729788153912668,
      "grad_norm": 1.4348874555698314,
      "learning_rate": 8.838223701790057e-06,
      "loss": 0.6031,
      "step": 143
    },
    {
      "epoch": 0.2490272373540856,
      "grad_norm": 1.2605512084319181,
      "learning_rate": 8.820186879108038e-06,
      "loss": 0.579,
      "step": 144
    },
    {
      "epoch": 0.25075659316904453,
      "grad_norm": 1.3303160437037416,
      "learning_rate": 8.802029828000157e-06,
      "loss": 0.5848,
      "step": 145
    },
    {
      "epoch": 0.25248594898400345,
      "grad_norm": 1.4582040782143946,
      "learning_rate": 8.783753119902766e-06,
      "loss": 0.5758,
      "step": 146
    },
    {
      "epoch": 0.25421530479896237,
      "grad_norm": 1.322898639608662,
      "learning_rate": 8.765357330018056e-06,
      "loss": 0.5869,
      "step": 147
    },
    {
      "epoch": 0.2559446606139213,
      "grad_norm": 1.384126692032235,
      "learning_rate": 8.746843037295936e-06,
      "loss": 0.5588,
      "step": 148
    },
    {
      "epoch": 0.2576740164288802,
      "grad_norm": 1.3286178990420618,
      "learning_rate": 8.728210824415829e-06,
      "loss": 0.5688,
      "step": 149
    },
    {
      "epoch": 0.2594033722438392,
      "grad_norm": 1.255847642552832,
      "learning_rate": 8.70946127776832e-06,
      "loss": 0.5436,
      "step": 150
    },
    {
      "epoch": 0.2611327280587981,
      "grad_norm": 1.5446363806642398,
      "learning_rate": 8.690594987436705e-06,
      "loss": 0.5669,
      "step": 151
    },
    {
      "epoch": 0.26286208387375704,
      "grad_norm": 1.3185596223777625,
      "learning_rate": 8.671612547178428e-06,
      "loss": 0.5652,
      "step": 152
    },
    {
      "epoch": 0.26459143968871596,
      "grad_norm": 1.3558175344108856,
      "learning_rate": 8.652514554406388e-06,
      "loss": 0.5872,
      "step": 153
    },
    {
      "epoch": 0.2663207955036749,
      "grad_norm": 1.4068740291750244,
      "learning_rate": 8.633301610170136e-06,
      "loss": 0.5683,
      "step": 154
    },
    {
      "epoch": 0.2680501513186338,
      "grad_norm": 1.3133616699361563,
      "learning_rate": 8.613974319136959e-06,
      "loss": 0.5693,
      "step": 155
    },
    {
      "epoch": 0.2697795071335927,
      "grad_norm": 1.2872738406840478,
      "learning_rate": 8.594533289572852e-06,
      "loss": 0.5787,
      "step": 156
    },
    {
      "epoch": 0.27150886294855164,
      "grad_norm": 1.3034688709492834,
      "learning_rate": 8.574979133323378e-06,
      "loss": 0.5644,
      "step": 157
    },
    {
      "epoch": 0.27323821876351057,
      "grad_norm": 1.2605674494333527,
      "learning_rate": 8.555312465794402e-06,
      "loss": 0.5758,
      "step": 158
    },
    {
      "epoch": 0.27496757457846954,
      "grad_norm": 1.3396122302856128,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.6039,
      "step": 159
    },
    {
      "epoch": 0.27669693039342846,
      "grad_norm": 1.2966490137544022,
      "learning_rate": 8.515644076206652e-06,
      "loss": 0.5645,
      "step": 160
    },
    {
      "epoch": 0.2784262862083874,
      "grad_norm": 1.4311476322477343,
      "learning_rate": 8.495643602586287e-06,
      "loss": 0.5705,
      "step": 161
    },
    {
      "epoch": 0.2801556420233463,
      "grad_norm": 1.3422910724323573,
      "learning_rate": 8.475533114523954e-06,
      "loss": 0.5656,
      "step": 162
    },
    {
      "epoch": 0.28188499783830523,
      "grad_norm": 1.2772032094100403,
      "learning_rate": 8.455313244934324e-06,
      "loss": 0.5907,
      "step": 163
    },
    {
      "epoch": 0.28361435365326415,
      "grad_norm": 1.3072283691035707,
      "learning_rate": 8.43498463017451e-06,
      "loss": 0.5696,
      "step": 164
    },
    {
      "epoch": 0.2853437094682231,
      "grad_norm": 1.3046935804589632,
      "learning_rate": 8.414547910024035e-06,
      "loss": 0.5775,
      "step": 165
    },
    {
      "epoch": 0.287073065283182,
      "grad_norm": 1.2976841761251101,
      "learning_rate": 8.39400372766471e-06,
      "loss": 0.5714,
      "step": 166
    },
    {
      "epoch": 0.2888024210981409,
      "grad_norm": 1.3247225670276452,
      "learning_rate": 8.373352729660373e-06,
      "loss": 0.5527,
      "step": 167
    },
    {
      "epoch": 0.2905317769130999,
      "grad_norm": 1.3304518084464831,
      "learning_rate": 8.352595565936554e-06,
      "loss": 0.5772,
      "step": 168
    },
    {
      "epoch": 0.2922611327280588,
      "grad_norm": 1.3567530464099045,
      "learning_rate": 8.331732889760021e-06,
      "loss": 0.5724,
      "step": 169
    },
    {
      "epoch": 0.29399048854301774,
      "grad_norm": 1.3864092503101992,
      "learning_rate": 8.310765357718207e-06,
      "loss": 0.5785,
      "step": 170
    },
    {
      "epoch": 0.29571984435797666,
      "grad_norm": 1.3431891176626012,
      "learning_rate": 8.289693629698564e-06,
      "loss": 0.5907,
      "step": 171
    },
    {
      "epoch": 0.2974492001729356,
      "grad_norm": 1.3269529964125137,
      "learning_rate": 8.268518368867781e-06,
      "loss": 0.5794,
      "step": 172
    },
    {
      "epoch": 0.2991785559878945,
      "grad_norm": 1.3677645716695777,
      "learning_rate": 8.247240241650918e-06,
      "loss": 0.5889,
      "step": 173
    },
    {
      "epoch": 0.3009079118028534,
      "grad_norm": 1.4344791877726926,
      "learning_rate": 8.22585991771044e-06,
      "loss": 0.5501,
      "step": 174
    },
    {
      "epoch": 0.30263726761781234,
      "grad_norm": 1.3773781206385498,
      "learning_rate": 8.204378069925121e-06,
      "loss": 0.5435,
      "step": 175
    },
    {
      "epoch": 0.30436662343277127,
      "grad_norm": 1.3415343046809804,
      "learning_rate": 8.182795374368893e-06,
      "loss": 0.5658,
      "step": 176
    },
    {
      "epoch": 0.30609597924773024,
      "grad_norm": 1.3607519825023242,
      "learning_rate": 8.16111251028955e-06,
      "loss": 0.5726,
      "step": 177
    },
    {
      "epoch": 0.30782533506268916,
      "grad_norm": 1.3372886670110973,
      "learning_rate": 8.139330160087374e-06,
      "loss": 0.5801,
      "step": 178
    },
    {
      "epoch": 0.3095546908776481,
      "grad_norm": 1.45444571694518,
      "learning_rate": 8.117449009293668e-06,
      "loss": 0.5675,
      "step": 179
    },
    {
      "epoch": 0.311284046692607,
      "grad_norm": 1.4232508965452562,
      "learning_rate": 8.095469746549172e-06,
      "loss": 0.5502,
      "step": 180
    },
    {
      "epoch": 0.31301340250756593,
      "grad_norm": 1.3513827085112067,
      "learning_rate": 8.073393063582386e-06,
      "loss": 0.5855,
      "step": 181
    },
    {
      "epoch": 0.31474275832252485,
      "grad_norm": 1.3052608382865691,
      "learning_rate": 8.051219655187818e-06,
      "loss": 0.5618,
      "step": 182
    },
    {
      "epoch": 0.3164721141374838,
      "grad_norm": 1.306514474737617,
      "learning_rate": 8.0289502192041e-06,
      "loss": 0.5617,
      "step": 183
    },
    {
      "epoch": 0.3182014699524427,
      "grad_norm": 1.2967196879064988,
      "learning_rate": 8.00658545649203e-06,
      "loss": 0.5976,
      "step": 184
    },
    {
      "epoch": 0.3199308257674016,
      "grad_norm": 1.4576437100632567,
      "learning_rate": 7.984126070912519e-06,
      "loss": 0.5688,
      "step": 185
    },
    {
      "epoch": 0.3216601815823606,
      "grad_norm": 1.3704399401870944,
      "learning_rate": 7.961572769304437e-06,
      "loss": 0.5631,
      "step": 186
    },
    {
      "epoch": 0.3233895373973195,
      "grad_norm": 1.319375380678081,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.5498,
      "step": 187
    },
    {
      "epoch": 0.32511889321227844,
      "grad_norm": 1.3529344720904868,
      "learning_rate": 7.916187260114264e-06,
      "loss": 0.5423,
      "step": 188
    },
    {
      "epoch": 0.32684824902723736,
      "grad_norm": 1.3027982855324163,
      "learning_rate": 7.89335648089903e-06,
      "loss": 0.582,
      "step": 189
    },
    {
      "epoch": 0.3285776048421963,
      "grad_norm": 1.4050936006586736,
      "learning_rate": 7.870434642343984e-06,
      "loss": 0.5636,
      "step": 190
    },
    {
      "epoch": 0.3303069606571552,
      "grad_norm": 1.2865924125836552,
      "learning_rate": 7.84742246584226e-06,
      "loss": 0.5494,
      "step": 191
    },
    {
      "epoch": 0.3320363164721141,
      "grad_norm": 1.3715611689044482,
      "learning_rate": 7.82432067563009e-06,
      "loss": 0.566,
      "step": 192
    },
    {
      "epoch": 0.33376567228707305,
      "grad_norm": 1.3718978175645358,
      "learning_rate": 7.801129998764014e-06,
      "loss": 0.5657,
      "step": 193
    },
    {
      "epoch": 0.33549502810203197,
      "grad_norm": 1.2502241262211298,
      "learning_rate": 7.777851165098012e-06,
      "loss": 0.5656,
      "step": 194
    },
    {
      "epoch": 0.33722438391699094,
      "grad_norm": 1.3852373692132978,
      "learning_rate": 7.754484907260513e-06,
      "loss": 0.5658,
      "step": 195
    },
    {
      "epoch": 0.33895373973194987,
      "grad_norm": 1.3703136522991974,
      "learning_rate": 7.731031960631354e-06,
      "loss": 0.5716,
      "step": 196
    },
    {
      "epoch": 0.3406830955469088,
      "grad_norm": 1.2486585135623867,
      "learning_rate": 7.70749306331863e-06,
      "loss": 0.5651,
      "step": 197
    },
    {
      "epoch": 0.3424124513618677,
      "grad_norm": 1.2601495096730775,
      "learning_rate": 7.68386895613546e-06,
      "loss": 0.5546,
      "step": 198
    },
    {
      "epoch": 0.34414180717682663,
      "grad_norm": 1.2502364746593353,
      "learning_rate": 7.660160382576683e-06,
      "loss": 0.5483,
      "step": 199
    },
    {
      "epoch": 0.34587116299178555,
      "grad_norm": 1.3477149756990934,
      "learning_rate": 7.636368088795451e-06,
      "loss": 0.5472,
      "step": 200
    },
    {
      "epoch": 0.3476005188067445,
      "grad_norm": 1.2901176761284996,
      "learning_rate": 7.612492823579744e-06,
      "loss": 0.5421,
      "step": 201
    },
    {
      "epoch": 0.3493298746217034,
      "grad_norm": 1.4658846157690446,
      "learning_rate": 7.588535338328816e-06,
      "loss": 0.5906,
      "step": 202
    },
    {
      "epoch": 0.3510592304366623,
      "grad_norm": 1.2964558327514957,
      "learning_rate": 7.564496387029532e-06,
      "loss": 0.5772,
      "step": 203
    },
    {
      "epoch": 0.3527885862516213,
      "grad_norm": 1.2170162880503597,
      "learning_rate": 7.540376726232648e-06,
      "loss": 0.5808,
      "step": 204
    },
    {
      "epoch": 0.3545179420665802,
      "grad_norm": 1.3354147357012909,
      "learning_rate": 7.516177115029002e-06,
      "loss": 0.5496,
      "step": 205
    },
    {
      "epoch": 0.35624729788153914,
      "grad_norm": 1.3770402354419473,
      "learning_rate": 7.491898315025615e-06,
      "loss": 0.5718,
      "step": 206
    },
    {
      "epoch": 0.35797665369649806,
      "grad_norm": 1.4004748528087516,
      "learning_rate": 7.467541090321735e-06,
      "loss": 0.5696,
      "step": 207
    },
    {
      "epoch": 0.359706009511457,
      "grad_norm": 1.3103570302853653,
      "learning_rate": 7.443106207484776e-06,
      "loss": 0.5817,
      "step": 208
    },
    {
      "epoch": 0.3614353653264159,
      "grad_norm": 1.4334519797119796,
      "learning_rate": 7.4185944355261996e-06,
      "loss": 0.5549,
      "step": 209
    },
    {
      "epoch": 0.3631647211413748,
      "grad_norm": 1.397522237730983,
      "learning_rate": 7.3940065458773146e-06,
      "loss": 0.5436,
      "step": 210
    },
    {
      "epoch": 0.36489407695633375,
      "grad_norm": 1.2965013000965513,
      "learning_rate": 7.369343312364994e-06,
      "loss": 0.5648,
      "step": 211
    },
    {
      "epoch": 0.36662343277129267,
      "grad_norm": 1.3115268564578664,
      "learning_rate": 7.344605511187322e-06,
      "loss": 0.5663,
      "step": 212
    },
    {
      "epoch": 0.36835278858625164,
      "grad_norm": 1.2857253578787209,
      "learning_rate": 7.319793920889171e-06,
      "loss": 0.5465,
      "step": 213
    },
    {
      "epoch": 0.37008214440121057,
      "grad_norm": 1.2254130773211296,
      "learning_rate": 7.294909322337689e-06,
      "loss": 0.5332,
      "step": 214
    },
    {
      "epoch": 0.3718115002161695,
      "grad_norm": 1.2505477215695469,
      "learning_rate": 7.269952498697734e-06,
      "loss": 0.5519,
      "step": 215
    },
    {
      "epoch": 0.3735408560311284,
      "grad_norm": 1.2309151575793142,
      "learning_rate": 7.244924235407224e-06,
      "loss": 0.556,
      "step": 216
    },
    {
      "epoch": 0.37527021184608733,
      "grad_norm": 1.2736490892998513,
      "learning_rate": 7.219825320152411e-06,
      "loss": 0.5753,
      "step": 217
    },
    {
      "epoch": 0.37699956766104625,
      "grad_norm": 1.188067487294118,
      "learning_rate": 7.194656542843103e-06,
      "loss": 0.5457,
      "step": 218
    },
    {
      "epoch": 0.3787289234760052,
      "grad_norm": 1.3573649976256121,
      "learning_rate": 7.169418695587791e-06,
      "loss": 0.5317,
      "step": 219
    },
    {
      "epoch": 0.3804582792909641,
      "grad_norm": 1.3104765075971507,
      "learning_rate": 7.1441125726687336e-06,
      "loss": 0.5416,
      "step": 220
    },
    {
      "epoch": 0.382187635105923,
      "grad_norm": 1.2524746976217112,
      "learning_rate": 7.118738970516944e-06,
      "loss": 0.5231,
      "step": 221
    },
    {
      "epoch": 0.383916990920882,
      "grad_norm": 1.2912855193020514,
      "learning_rate": 7.093298687687141e-06,
      "loss": 0.577,
      "step": 222
    },
    {
      "epoch": 0.3856463467358409,
      "grad_norm": 1.277812278463467,
      "learning_rate": 7.067792524832604e-06,
      "loss": 0.5627,
      "step": 223
    },
    {
      "epoch": 0.38737570255079984,
      "grad_norm": 1.3047580618945316,
      "learning_rate": 7.042221284679982e-06,
      "loss": 0.5385,
      "step": 224
    },
    {
      "epoch": 0.38910505836575876,
      "grad_norm": 1.1904223056651926,
      "learning_rate": 7.016585772004026e-06,
      "loss": 0.5504,
      "step": 225
    },
    {
      "epoch": 0.3908344141807177,
      "grad_norm": 1.2398301118878279,
      "learning_rate": 6.990886793602268e-06,
      "loss": 0.5463,
      "step": 226
    },
    {
      "epoch": 0.3925637699956766,
      "grad_norm": 1.2249280934558917,
      "learning_rate": 6.965125158269619e-06,
      "loss": 0.5524,
      "step": 227
    },
    {
      "epoch": 0.3942931258106355,
      "grad_norm": 1.3151361816315885,
      "learning_rate": 6.939301676772927e-06,
      "loss": 0.5643,
      "step": 228
    },
    {
      "epoch": 0.39602248162559445,
      "grad_norm": 1.4155270180747608,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.5396,
      "step": 229
    },
    {
      "epoch": 0.39775183744055337,
      "grad_norm": 1.3122619817083747,
      "learning_rate": 6.887472428061285e-06,
      "loss": 0.5533,
      "step": 230
    },
    {
      "epoch": 0.39948119325551235,
      "grad_norm": 1.2806770091938722,
      "learning_rate": 6.8614682920097265e-06,
      "loss": 0.5403,
      "step": 231
    },
    {
      "epoch": 0.40121054907047127,
      "grad_norm": 1.3470173683287574,
      "learning_rate": 6.835405572069572e-06,
      "loss": 0.5395,
      "step": 232
    },
    {
      "epoch": 0.4029399048854302,
      "grad_norm": 1.3345925151371414,
      "learning_rate": 6.809285088483361e-06,
      "loss": 0.5488,
      "step": 233
    },
    {
      "epoch": 0.4046692607003891,
      "grad_norm": 1.364191950346444,
      "learning_rate": 6.783107663311566e-06,
      "loss": 0.5597,
      "step": 234
    },
    {
      "epoch": 0.40639861651534803,
      "grad_norm": 1.229870751967945,
      "learning_rate": 6.7568741204067145e-06,
      "loss": 0.5532,
      "step": 235
    },
    {
      "epoch": 0.40812797233030695,
      "grad_norm": 1.2511626652001453,
      "learning_rate": 6.730585285387465e-06,
      "loss": 0.5599,
      "step": 236
    },
    {
      "epoch": 0.4098573281452659,
      "grad_norm": 1.1963629945054264,
      "learning_rate": 6.704241985612625e-06,
      "loss": 0.5662,
      "step": 237
    },
    {
      "epoch": 0.4115866839602248,
      "grad_norm": 1.28804770356803,
      "learning_rate": 6.6778450501551065e-06,
      "loss": 0.5801,
      "step": 238
    },
    {
      "epoch": 0.4133160397751837,
      "grad_norm": 1.2590013900847816,
      "learning_rate": 6.651395309775837e-06,
      "loss": 0.5644,
      "step": 239
    },
    {
      "epoch": 0.4150453955901427,
      "grad_norm": 1.2551864064194094,
      "learning_rate": 6.6248935968976135e-06,
      "loss": 0.5394,
      "step": 240
    },
    {
      "epoch": 0.4167747514051016,
      "grad_norm": 1.324809116558963,
      "learning_rate": 6.598340745578908e-06,
      "loss": 0.5698,
      "step": 241
    },
    {
      "epoch": 0.41850410722006054,
      "grad_norm": 1.3179110819475957,
      "learning_rate": 6.571737591487611e-06,
      "loss": 0.5438,
      "step": 242
    },
    {
      "epoch": 0.42023346303501946,
      "grad_norm": 1.2492716164423538,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.5534,
      "step": 243
    },
    {
      "epoch": 0.4219628188499784,
      "grad_norm": 1.3793013860663719,
      "learning_rate": 6.518383725548074e-06,
      "loss": 0.5507,
      "step": 244
    },
    {
      "epoch": 0.4236921746649373,
      "grad_norm": 1.2552437595758494,
      "learning_rate": 6.491634692845781e-06,
      "loss": 0.5283,
      "step": 245
    },
    {
      "epoch": 0.4254215304798962,
      "grad_norm": 1.320285724366206,
      "learning_rate": 6.464838715609945e-06,
      "loss": 0.5381,
      "step": 246
    },
    {
      "epoch": 0.42715088629485515,
      "grad_norm": 1.321244687463225,
      "learning_rate": 6.437996637160086e-06,
      "loss": 0.5632,
      "step": 247
    },
    {
      "epoch": 0.42888024210981407,
      "grad_norm": 1.298650328971777,
      "learning_rate": 6.411109302266616e-06,
      "loss": 0.5436,
      "step": 248
    },
    {
      "epoch": 0.43060959792477305,
      "grad_norm": 1.2466524194529058,
      "learning_rate": 6.384177557124247e-06,
      "loss": 0.536,
      "step": 249
    },
    {
      "epoch": 0.43233895373973197,
      "grad_norm": 1.2659973914064022,
      "learning_rate": 6.3572022493253715e-06,
      "loss": 0.5666,
      "step": 250
    },
    {
      "epoch": 0.4340683095546909,
      "grad_norm": 1.3125673782338234,
      "learning_rate": 6.330184227833376e-06,
      "loss": 0.5269,
      "step": 251
    },
    {
      "epoch": 0.4357976653696498,
      "grad_norm": 1.2471832027915164,
      "learning_rate": 6.303124342955928e-06,
      "loss": 0.5277,
      "step": 252
    },
    {
      "epoch": 0.43752702118460873,
      "grad_norm": 1.3130769407311655,
      "learning_rate": 6.276023446318214e-06,
      "loss": 0.5628,
      "step": 253
    },
    {
      "epoch": 0.43925637699956765,
      "grad_norm": 1.2407127444236052,
      "learning_rate": 6.248882390836135e-06,
      "loss": 0.5583,
      "step": 254
    },
    {
      "epoch": 0.4409857328145266,
      "grad_norm": 1.374714499529323,
      "learning_rate": 6.2217020306894705e-06,
      "loss": 0.5298,
      "step": 255
    },
    {
      "epoch": 0.4427150886294855,
      "grad_norm": 1.4800838058998909,
      "learning_rate": 6.194483221294989e-06,
      "loss": 0.5469,
      "step": 256
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 1.1817922709588438,
      "learning_rate": 6.1672268192795285e-06,
      "loss": 0.5425,
      "step": 257
    },
    {
      "epoch": 0.4461738002594034,
      "grad_norm": 1.2263700782728304,
      "learning_rate": 6.139933682453035e-06,
      "loss": 0.5598,
      "step": 258
    },
    {
      "epoch": 0.4479031560743623,
      "grad_norm": 1.3081432662421466,
      "learning_rate": 6.112604669781572e-06,
      "loss": 0.5451,
      "step": 259
    },
    {
      "epoch": 0.44963251188932124,
      "grad_norm": 1.3868536914950849,
      "learning_rate": 6.085240641360281e-06,
      "loss": 0.5424,
      "step": 260
    },
    {
      "epoch": 0.45136186770428016,
      "grad_norm": 1.2274166600226197,
      "learning_rate": 6.057842458386315e-06,
      "loss": 0.537,
      "step": 261
    },
    {
      "epoch": 0.4530912235192391,
      "grad_norm": 1.272644249773423,
      "learning_rate": 6.030410983131733e-06,
      "loss": 0.56,
      "step": 262
    },
    {
      "epoch": 0.454820579334198,
      "grad_norm": 1.2893096518593115,
      "learning_rate": 6.002947078916365e-06,
      "loss": 0.541,
      "step": 263
    },
    {
      "epoch": 0.4565499351491569,
      "grad_norm": 1.2447414895886646,
      "learning_rate": 5.975451610080643e-06,
      "loss": 0.5439,
      "step": 264
    },
    {
      "epoch": 0.45827929096411585,
      "grad_norm": 1.217222966677454,
      "learning_rate": 5.947925441958393e-06,
      "loss": 0.542,
      "step": 265
    },
    {
      "epoch": 0.46000864677907477,
      "grad_norm": 1.2886825234554569,
      "learning_rate": 5.920369440849609e-06,
      "loss": 0.5543,
      "step": 266
    },
    {
      "epoch": 0.46173800259403375,
      "grad_norm": 1.3251969210640597,
      "learning_rate": 5.892784473993184e-06,
      "loss": 0.5399,
      "step": 267
    },
    {
      "epoch": 0.46346735840899267,
      "grad_norm": 1.2268752980398028,
      "learning_rate": 5.865171409539614e-06,
      "loss": 0.5467,
      "step": 268
    },
    {
      "epoch": 0.4651967142239516,
      "grad_norm": 1.3068892898030797,
      "learning_rate": 5.837531116523683e-06,
      "loss": 0.5401,
      "step": 269
    },
    {
      "epoch": 0.4669260700389105,
      "grad_norm": 1.2594383194818888,
      "learning_rate": 5.809864464837105e-06,
      "loss": 0.5584,
      "step": 270
    },
    {
      "epoch": 0.46865542585386943,
      "grad_norm": 1.2235669424052997,
      "learning_rate": 5.782172325201155e-06,
      "loss": 0.5311,
      "step": 271
    },
    {
      "epoch": 0.47038478166882836,
      "grad_norm": 1.3219449360910378,
      "learning_rate": 5.754455569139258e-06,
      "loss": 0.5299,
      "step": 272
    },
    {
      "epoch": 0.4721141374837873,
      "grad_norm": 1.2887549822531428,
      "learning_rate": 5.726715068949564e-06,
      "loss": 0.5451,
      "step": 273
    },
    {
      "epoch": 0.4738434932987462,
      "grad_norm": 1.2358733083826365,
      "learning_rate": 5.698951697677498e-06,
      "loss": 0.5379,
      "step": 274
    },
    {
      "epoch": 0.4755728491137051,
      "grad_norm": 1.3090051655251154,
      "learning_rate": 5.671166329088278e-06,
      "loss": 0.5213,
      "step": 275
    },
    {
      "epoch": 0.4773022049286641,
      "grad_norm": 1.3042917976396255,
      "learning_rate": 5.643359837639419e-06,
      "loss": 0.5235,
      "step": 276
    },
    {
      "epoch": 0.479031560743623,
      "grad_norm": 1.298269280525944,
      "learning_rate": 5.615533098453215e-06,
      "loss": 0.5545,
      "step": 277
    },
    {
      "epoch": 0.48076091655858194,
      "grad_norm": 1.2012345750779942,
      "learning_rate": 5.587686987289189e-06,
      "loss": 0.5544,
      "step": 278
    },
    {
      "epoch": 0.48249027237354086,
      "grad_norm": 1.3007588033367934,
      "learning_rate": 5.559822380516539e-06,
      "loss": 0.5271,
      "step": 279
    },
    {
      "epoch": 0.4842196281884998,
      "grad_norm": 1.3534538377292569,
      "learning_rate": 5.531940155086557e-06,
      "loss": 0.5688,
      "step": 280
    },
    {
      "epoch": 0.4859489840034587,
      "grad_norm": 1.243334314478317,
      "learning_rate": 5.504041188505022e-06,
      "loss": 0.5473,
      "step": 281
    },
    {
      "epoch": 0.4876783398184176,
      "grad_norm": 1.306645781202671,
      "learning_rate": 5.476126358804594e-06,
      "loss": 0.5422,
      "step": 282
    },
    {
      "epoch": 0.48940769563337655,
      "grad_norm": 1.280897747736695,
      "learning_rate": 5.448196544517168e-06,
      "loss": 0.5516,
      "step": 283
    },
    {
      "epoch": 0.49113705144833547,
      "grad_norm": 1.4064375573811547,
      "learning_rate": 5.420252624646238e-06,
      "loss": 0.5467,
      "step": 284
    },
    {
      "epoch": 0.49286640726329445,
      "grad_norm": 1.4119469280433727,
      "learning_rate": 5.392295478639226e-06,
      "loss": 0.5712,
      "step": 285
    },
    {
      "epoch": 0.49459576307825337,
      "grad_norm": 1.3832811952775954,
      "learning_rate": 5.3643259863598015e-06,
      "loss": 0.5311,
      "step": 286
    },
    {
      "epoch": 0.4963251188932123,
      "grad_norm": 1.2793380120458782,
      "learning_rate": 5.336345028060199e-06,
      "loss": 0.5464,
      "step": 287
    },
    {
      "epoch": 0.4980544747081712,
      "grad_norm": 1.325075817941503,
      "learning_rate": 5.308353484353508e-06,
      "loss": 0.5472,
      "step": 288
    },
    {
      "epoch": 0.49978383052313013,
      "grad_norm": 1.323776420208779,
      "learning_rate": 5.2803522361859596e-06,
      "loss": 0.5162,
      "step": 289
    },
    {
      "epoch": 0.5015131863380891,
      "grad_norm": 1.2724751142456727,
      "learning_rate": 5.252342164809204e-06,
      "loss": 0.5419,
      "step": 290
    },
    {
      "epoch": 0.503242542153048,
      "grad_norm": 1.3155061940618022,
      "learning_rate": 5.224324151752575e-06,
      "loss": 0.5108,
      "step": 291
    },
    {
      "epoch": 0.5049718979680069,
      "grad_norm": 1.325463563646962,
      "learning_rate": 5.1962990787953436e-06,
      "loss": 0.556,
      "step": 292
    },
    {
      "epoch": 0.5067012537829658,
      "grad_norm": 1.3557284931566147,
      "learning_rate": 5.168267827938971e-06,
      "loss": 0.5205,
      "step": 293
    },
    {
      "epoch": 0.5084306095979247,
      "grad_norm": 1.4690183640639547,
      "learning_rate": 5.140231281379345e-06,
      "loss": 0.5314,
      "step": 294
    },
    {
      "epoch": 0.5101599654128837,
      "grad_norm": 1.3889196264518369,
      "learning_rate": 5.112190321479026e-06,
      "loss": 0.5421,
      "step": 295
    },
    {
      "epoch": 0.5118893212278426,
      "grad_norm": 1.235942882014356,
      "learning_rate": 5.084145830739462e-06,
      "loss": 0.5311,
      "step": 296
    },
    {
      "epoch": 0.5136186770428015,
      "grad_norm": 1.306512327093981,
      "learning_rate": 5.05609869177323e-06,
      "loss": 0.5273,
      "step": 297
    },
    {
      "epoch": 0.5153480328577604,
      "grad_norm": 1.5917366093172325,
      "learning_rate": 5.028049787276249e-06,
      "loss": 0.5129,
      "step": 298
    },
    {
      "epoch": 0.5170773886727195,
      "grad_norm": 1.40594572028163,
      "learning_rate": 5e-06,
      "loss": 0.5203,
      "step": 299
    },
    {
      "epoch": 0.5188067444876784,
      "grad_norm": 1.262588520269215,
      "learning_rate": 4.971950212723753e-06,
      "loss": 0.5055,
      "step": 300
    },
    {
      "epoch": 0.5205361003026373,
      "grad_norm": 1.4092188833304773,
      "learning_rate": 4.943901308226771e-06,
      "loss": 0.5434,
      "step": 301
    },
    {
      "epoch": 0.5222654561175962,
      "grad_norm": 1.382690963218948,
      "learning_rate": 4.91585416926054e-06,
      "loss": 0.5449,
      "step": 302
    },
    {
      "epoch": 0.5239948119325551,
      "grad_norm": 1.3495688952134626,
      "learning_rate": 4.887809678520976e-06,
      "loss": 0.5277,
      "step": 303
    },
    {
      "epoch": 0.5257241677475141,
      "grad_norm": 1.2421088006172942,
      "learning_rate": 4.859768718620656e-06,
      "loss": 0.5393,
      "step": 304
    },
    {
      "epoch": 0.527453523562473,
      "grad_norm": 1.2488445572106857,
      "learning_rate": 4.831732172061032e-06,
      "loss": 0.5225,
      "step": 305
    },
    {
      "epoch": 0.5291828793774319,
      "grad_norm": 1.3844825536152099,
      "learning_rate": 4.803700921204659e-06,
      "loss": 0.5448,
      "step": 306
    },
    {
      "epoch": 0.5309122351923908,
      "grad_norm": 1.2550069336336056,
      "learning_rate": 4.775675848247427e-06,
      "loss": 0.5326,
      "step": 307
    },
    {
      "epoch": 0.5326415910073498,
      "grad_norm": 1.3042371941510982,
      "learning_rate": 4.747657835190796e-06,
      "loss": 0.5485,
      "step": 308
    },
    {
      "epoch": 0.5343709468223087,
      "grad_norm": 1.439469833400696,
      "learning_rate": 4.719647763814041e-06,
      "loss": 0.5235,
      "step": 309
    },
    {
      "epoch": 0.5361003026372676,
      "grad_norm": 1.258005734371576,
      "learning_rate": 4.691646515646492e-06,
      "loss": 0.5049,
      "step": 310
    },
    {
      "epoch": 0.5378296584522265,
      "grad_norm": 1.2574136262278794,
      "learning_rate": 4.663654971939802e-06,
      "loss": 0.5306,
      "step": 311
    },
    {
      "epoch": 0.5395590142671854,
      "grad_norm": 1.211366769644713,
      "learning_rate": 4.6356740136402e-06,
      "loss": 0.5158,
      "step": 312
    },
    {
      "epoch": 0.5412883700821444,
      "grad_norm": 1.404490080536933,
      "learning_rate": 4.6077045213607765e-06,
      "loss": 0.5106,
      "step": 313
    },
    {
      "epoch": 0.5430177258971033,
      "grad_norm": 1.3565090769351222,
      "learning_rate": 4.579747375353763e-06,
      "loss": 0.5115,
      "step": 314
    },
    {
      "epoch": 0.5447470817120622,
      "grad_norm": 1.3745499639927738,
      "learning_rate": 4.551803455482833e-06,
      "loss": 0.5352,
      "step": 315
    },
    {
      "epoch": 0.5464764375270211,
      "grad_norm": 1.3375963306559595,
      "learning_rate": 4.5238736411954075e-06,
      "loss": 0.5149,
      "step": 316
    },
    {
      "epoch": 0.5482057933419802,
      "grad_norm": 1.2416690865187434,
      "learning_rate": 4.4959588114949785e-06,
      "loss": 0.5183,
      "step": 317
    },
    {
      "epoch": 0.5499351491569391,
      "grad_norm": 1.2969100557148994,
      "learning_rate": 4.468059844913444e-06,
      "loss": 0.5305,
      "step": 318
    },
    {
      "epoch": 0.551664504971898,
      "grad_norm": 1.3028745121039649,
      "learning_rate": 4.4401776194834615e-06,
      "loss": 0.53,
      "step": 319
    },
    {
      "epoch": 0.5533938607868569,
      "grad_norm": 1.2529196452246425,
      "learning_rate": 4.4123130127108125e-06,
      "loss": 0.5631,
      "step": 320
    },
    {
      "epoch": 0.5551232166018158,
      "grad_norm": 1.2914281020735483,
      "learning_rate": 4.384466901546786e-06,
      "loss": 0.5029,
      "step": 321
    },
    {
      "epoch": 0.5568525724167748,
      "grad_norm": 1.3030244428041313,
      "learning_rate": 4.356640162360582e-06,
      "loss": 0.5547,
      "step": 322
    },
    {
      "epoch": 0.5585819282317337,
      "grad_norm": 1.310975612444778,
      "learning_rate": 4.3288336709117246e-06,
      "loss": 0.5476,
      "step": 323
    },
    {
      "epoch": 0.5603112840466926,
      "grad_norm": 1.3585233059927981,
      "learning_rate": 4.3010483023225045e-06,
      "loss": 0.5096,
      "step": 324
    },
    {
      "epoch": 0.5620406398616515,
      "grad_norm": 1.3858533161552185,
      "learning_rate": 4.273284931050438e-06,
      "loss": 0.5432,
      "step": 325
    },
    {
      "epoch": 0.5637699956766105,
      "grad_norm": 1.3005876771208573,
      "learning_rate": 4.245544430860743e-06,
      "loss": 0.5339,
      "step": 326
    },
    {
      "epoch": 0.5654993514915694,
      "grad_norm": 1.3987689407306876,
      "learning_rate": 4.217827674798845e-06,
      "loss": 0.5096,
      "step": 327
    },
    {
      "epoch": 0.5672287073065283,
      "grad_norm": 1.309864053992572,
      "learning_rate": 4.190135535162894e-06,
      "loss": 0.5074,
      "step": 328
    },
    {
      "epoch": 0.5689580631214872,
      "grad_norm": 1.3268629259565734,
      "learning_rate": 4.162468883476319e-06,
      "loss": 0.5386,
      "step": 329
    },
    {
      "epoch": 0.5706874189364461,
      "grad_norm": 1.3349966799384403,
      "learning_rate": 4.1348285904603865e-06,
      "loss": 0.5208,
      "step": 330
    },
    {
      "epoch": 0.5724167747514051,
      "grad_norm": 1.2661602913195409,
      "learning_rate": 4.107215526006818e-06,
      "loss": 0.5107,
      "step": 331
    },
    {
      "epoch": 0.574146130566364,
      "grad_norm": 1.2797920680965094,
      "learning_rate": 4.079630559150392e-06,
      "loss": 0.513,
      "step": 332
    },
    {
      "epoch": 0.5758754863813229,
      "grad_norm": 1.3077982663455536,
      "learning_rate": 4.052074558041608e-06,
      "loss": 0.5154,
      "step": 333
    },
    {
      "epoch": 0.5776048421962818,
      "grad_norm": 1.3760657710228763,
      "learning_rate": 4.02454838991936e-06,
      "loss": 0.5294,
      "step": 334
    },
    {
      "epoch": 0.5793341980112409,
      "grad_norm": 1.184811772801038,
      "learning_rate": 3.997052921083637e-06,
      "loss": 0.4896,
      "step": 335
    },
    {
      "epoch": 0.5810635538261998,
      "grad_norm": 1.4503676384261956,
      "learning_rate": 3.969589016868269e-06,
      "loss": 0.5131,
      "step": 336
    },
    {
      "epoch": 0.5827929096411587,
      "grad_norm": 1.2572496861000189,
      "learning_rate": 3.9421575416136866e-06,
      "loss": 0.5356,
      "step": 337
    },
    {
      "epoch": 0.5845222654561176,
      "grad_norm": 1.2168932626954965,
      "learning_rate": 3.9147593586397195e-06,
      "loss": 0.5012,
      "step": 338
    },
    {
      "epoch": 0.5862516212710766,
      "grad_norm": 1.335579085108439,
      "learning_rate": 3.887395330218429e-06,
      "loss": 0.5073,
      "step": 339
    },
    {
      "epoch": 0.5879809770860355,
      "grad_norm": 1.2148641410862266,
      "learning_rate": 3.8600663175469675e-06,
      "loss": 0.5216,
      "step": 340
    },
    {
      "epoch": 0.5897103329009944,
      "grad_norm": 1.2477985045133986,
      "learning_rate": 3.832773180720475e-06,
      "loss": 0.509,
      "step": 341
    },
    {
      "epoch": 0.5914396887159533,
      "grad_norm": 1.3416766428515066,
      "learning_rate": 3.8055167787050134e-06,
      "loss": 0.5163,
      "step": 342
    },
    {
      "epoch": 0.5931690445309122,
      "grad_norm": 1.2682767103868227,
      "learning_rate": 3.778297969310529e-06,
      "loss": 0.527,
      "step": 343
    },
    {
      "epoch": 0.5948984003458712,
      "grad_norm": 1.2414041985446462,
      "learning_rate": 3.751117609163865e-06,
      "loss": 0.5001,
      "step": 344
    },
    {
      "epoch": 0.5966277561608301,
      "grad_norm": 1.375054008335257,
      "learning_rate": 3.723976553681787e-06,
      "loss": 0.5342,
      "step": 345
    },
    {
      "epoch": 0.598357111975789,
      "grad_norm": 1.2480963236384548,
      "learning_rate": 3.6968756570440735e-06,
      "loss": 0.5028,
      "step": 346
    },
    {
      "epoch": 0.6000864677907479,
      "grad_norm": 1.2159599196486677,
      "learning_rate": 3.669815772166625e-06,
      "loss": 0.5125,
      "step": 347
    },
    {
      "epoch": 0.6018158236057068,
      "grad_norm": 1.4104148060749142,
      "learning_rate": 3.6427977506746293e-06,
      "loss": 0.5207,
      "step": 348
    },
    {
      "epoch": 0.6035451794206658,
      "grad_norm": 1.1950066056646578,
      "learning_rate": 3.6158224428757538e-06,
      "loss": 0.4943,
      "step": 349
    },
    {
      "epoch": 0.6052745352356247,
      "grad_norm": 1.147453355609145,
      "learning_rate": 3.5888906977333858e-06,
      "loss": 0.4992,
      "step": 350
    },
    {
      "epoch": 0.6070038910505836,
      "grad_norm": 1.4493578720570028,
      "learning_rate": 3.562003362839914e-06,
      "loss": 0.5213,
      "step": 351
    },
    {
      "epoch": 0.6087332468655425,
      "grad_norm": 1.2658733112800329,
      "learning_rate": 3.5351612843900555e-06,
      "loss": 0.5056,
      "step": 352
    },
    {
      "epoch": 0.6104626026805016,
      "grad_norm": 1.1757360378271826,
      "learning_rate": 3.50836530715422e-06,
      "loss": 0.4856,
      "step": 353
    },
    {
      "epoch": 0.6121919584954605,
      "grad_norm": 1.3055141846929021,
      "learning_rate": 3.4816162744519266e-06,
      "loss": 0.5317,
      "step": 354
    },
    {
      "epoch": 0.6139213143104194,
      "grad_norm": 1.37730807773082,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.5159,
      "step": 355
    },
    {
      "epoch": 0.6156506701253783,
      "grad_norm": 1.1947534769598933,
      "learning_rate": 3.42826240851239e-06,
      "loss": 0.5214,
      "step": 356
    },
    {
      "epoch": 0.6173800259403373,
      "grad_norm": 1.3226134189067087,
      "learning_rate": 3.4016592544210937e-06,
      "loss": 0.5264,
      "step": 357
    },
    {
      "epoch": 0.6191093817552962,
      "grad_norm": 1.41629808960968,
      "learning_rate": 3.375106403102389e-06,
      "loss": 0.5379,
      "step": 358
    },
    {
      "epoch": 0.6208387375702551,
      "grad_norm": 1.3266049262461792,
      "learning_rate": 3.3486046902241663e-06,
      "loss": 0.5017,
      "step": 359
    },
    {
      "epoch": 0.622568093385214,
      "grad_norm": 1.3117139861674612,
      "learning_rate": 3.322154949844897e-06,
      "loss": 0.5066,
      "step": 360
    },
    {
      "epoch": 0.6242974492001729,
      "grad_norm": 1.275223937661878,
      "learning_rate": 3.295758014387375e-06,
      "loss": 0.5107,
      "step": 361
    },
    {
      "epoch": 0.6260268050151319,
      "grad_norm": 1.3609897446280972,
      "learning_rate": 3.269414714612534e-06,
      "loss": 0.5211,
      "step": 362
    },
    {
      "epoch": 0.6277561608300908,
      "grad_norm": 1.4136951954746158,
      "learning_rate": 3.2431258795932863e-06,
      "loss": 0.5299,
      "step": 363
    },
    {
      "epoch": 0.6294855166450497,
      "grad_norm": 1.3164541206983948,
      "learning_rate": 3.216892336688435e-06,
      "loss": 0.5251,
      "step": 364
    },
    {
      "epoch": 0.6312148724600086,
      "grad_norm": 1.2729230339815254,
      "learning_rate": 3.1907149115166403e-06,
      "loss": 0.5052,
      "step": 365
    },
    {
      "epoch": 0.6329442282749675,
      "grad_norm": 1.3720577760394352,
      "learning_rate": 3.1645944279304296e-06,
      "loss": 0.5142,
      "step": 366
    },
    {
      "epoch": 0.6346735840899265,
      "grad_norm": 1.3838881292780676,
      "learning_rate": 3.1385317079902743e-06,
      "loss": 0.5301,
      "step": 367
    },
    {
      "epoch": 0.6364029399048854,
      "grad_norm": 1.1823914759517473,
      "learning_rate": 3.112527571938717e-06,
      "loss": 0.5038,
      "step": 368
    },
    {
      "epoch": 0.6381322957198443,
      "grad_norm": 1.4852396956399265,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.5383,
      "step": 369
    },
    {
      "epoch": 0.6398616515348032,
      "grad_norm": 1.3537270664670102,
      "learning_rate": 3.060698323227075e-06,
      "loss": 0.507,
      "step": 370
    },
    {
      "epoch": 0.6415910073497623,
      "grad_norm": 1.313357691607948,
      "learning_rate": 3.0348748417303826e-06,
      "loss": 0.5076,
      "step": 371
    },
    {
      "epoch": 0.6433203631647212,
      "grad_norm": 1.3130162160688268,
      "learning_rate": 3.0091132063977344e-06,
      "loss": 0.5267,
      "step": 372
    },
    {
      "epoch": 0.6450497189796801,
      "grad_norm": 1.1920662735224292,
      "learning_rate": 2.9834142279959754e-06,
      "loss": 0.5155,
      "step": 373
    },
    {
      "epoch": 0.646779074794639,
      "grad_norm": 1.2025632883759985,
      "learning_rate": 2.95777871532002e-06,
      "loss": 0.5096,
      "step": 374
    },
    {
      "epoch": 0.648508430609598,
      "grad_norm": 1.3113109709173287,
      "learning_rate": 2.932207475167398e-06,
      "loss": 0.5002,
      "step": 375
    },
    {
      "epoch": 0.6502377864245569,
      "grad_norm": 1.2461184315574925,
      "learning_rate": 2.906701312312861e-06,
      "loss": 0.5169,
      "step": 376
    },
    {
      "epoch": 0.6519671422395158,
      "grad_norm": 1.2608068635952254,
      "learning_rate": 2.8812610294830568e-06,
      "loss": 0.5043,
      "step": 377
    },
    {
      "epoch": 0.6536964980544747,
      "grad_norm": 1.322071977690957,
      "learning_rate": 2.8558874273312677e-06,
      "loss": 0.5039,
      "step": 378
    },
    {
      "epoch": 0.6554258538694336,
      "grad_norm": 1.2333609191041428,
      "learning_rate": 2.83058130441221e-06,
      "loss": 0.5066,
      "step": 379
    },
    {
      "epoch": 0.6571552096843926,
      "grad_norm": 1.2541631476872825,
      "learning_rate": 2.805343457156898e-06,
      "loss": 0.5191,
      "step": 380
    },
    {
      "epoch": 0.6588845654993515,
      "grad_norm": 1.279925954820817,
      "learning_rate": 2.7801746798475905e-06,
      "loss": 0.4903,
      "step": 381
    },
    {
      "epoch": 0.6606139213143104,
      "grad_norm": 1.2393630411242251,
      "learning_rate": 2.755075764592777e-06,
      "loss": 0.5007,
      "step": 382
    },
    {
      "epoch": 0.6623432771292693,
      "grad_norm": 1.2113955627282111,
      "learning_rate": 2.7300475013022666e-06,
      "loss": 0.5187,
      "step": 383
    },
    {
      "epoch": 0.6640726329442282,
      "grad_norm": 1.332875296702971,
      "learning_rate": 2.705090677662311e-06,
      "loss": 0.5116,
      "step": 384
    },
    {
      "epoch": 0.6658019887591872,
      "grad_norm": 1.1776968966832972,
      "learning_rate": 2.6802060791108304e-06,
      "loss": 0.4994,
      "step": 385
    },
    {
      "epoch": 0.6675313445741461,
      "grad_norm": 1.2110870121335493,
      "learning_rate": 2.6553944888126772e-06,
      "loss": 0.5184,
      "step": 386
    },
    {
      "epoch": 0.669260700389105,
      "grad_norm": 1.3343623955630721,
      "learning_rate": 2.6306566876350072e-06,
      "loss": 0.5315,
      "step": 387
    },
    {
      "epoch": 0.6709900562040639,
      "grad_norm": 1.238593700529844,
      "learning_rate": 2.605993454122687e-06,
      "loss": 0.5211,
      "step": 388
    },
    {
      "epoch": 0.672719412019023,
      "grad_norm": 1.2623050551820758,
      "learning_rate": 2.5814055644738013e-06,
      "loss": 0.4985,
      "step": 389
    },
    {
      "epoch": 0.6744487678339819,
      "grad_norm": 1.313293854036044,
      "learning_rate": 2.5568937925152272e-06,
      "loss": 0.524,
      "step": 390
    },
    {
      "epoch": 0.6761781236489408,
      "grad_norm": 1.2517836743413833,
      "learning_rate": 2.532458909678266e-06,
      "loss": 0.5098,
      "step": 391
    },
    {
      "epoch": 0.6779074794638997,
      "grad_norm": 1.1520446930274513,
      "learning_rate": 2.508101684974387e-06,
      "loss": 0.5015,
      "step": 392
    },
    {
      "epoch": 0.6796368352788587,
      "grad_norm": 1.3005504514213648,
      "learning_rate": 2.483822884971e-06,
      "loss": 0.4803,
      "step": 393
    },
    {
      "epoch": 0.6813661910938176,
      "grad_norm": 1.234117732958065,
      "learning_rate": 2.4596232737673544e-06,
      "loss": 0.5196,
      "step": 394
    },
    {
      "epoch": 0.6830955469087765,
      "grad_norm": 1.1981906278111187,
      "learning_rate": 2.43550361297047e-06,
      "loss": 0.5059,
      "step": 395
    },
    {
      "epoch": 0.6848249027237354,
      "grad_norm": 1.2997430500346672,
      "learning_rate": 2.4114646616711844e-06,
      "loss": 0.5008,
      "step": 396
    },
    {
      "epoch": 0.6865542585386943,
      "grad_norm": 1.2614162559191038,
      "learning_rate": 2.387507176420256e-06,
      "loss": 0.502,
      "step": 397
    },
    {
      "epoch": 0.6882836143536533,
      "grad_norm": 1.188230979407375,
      "learning_rate": 2.3636319112045495e-06,
      "loss": 0.5164,
      "step": 398
    },
    {
      "epoch": 0.6900129701686122,
      "grad_norm": 1.334435091915391,
      "learning_rate": 2.339839617423318e-06,
      "loss": 0.5069,
      "step": 399
    },
    {
      "epoch": 0.6917423259835711,
      "grad_norm": 1.2428049634796243,
      "learning_rate": 2.31613104386454e-06,
      "loss": 0.4934,
      "step": 400
    },
    {
      "epoch": 0.69347168179853,
      "grad_norm": 1.283895066008161,
      "learning_rate": 2.2925069366813718e-06,
      "loss": 0.5075,
      "step": 401
    },
    {
      "epoch": 0.695201037613489,
      "grad_norm": 1.276327425480895,
      "learning_rate": 2.2689680393686457e-06,
      "loss": 0.5138,
      "step": 402
    },
    {
      "epoch": 0.6969303934284479,
      "grad_norm": 1.2630101628882442,
      "learning_rate": 2.245515092739488e-06,
      "loss": 0.5116,
      "step": 403
    },
    {
      "epoch": 0.6986597492434068,
      "grad_norm": 1.1913216900412702,
      "learning_rate": 2.2221488349019903e-06,
      "loss": 0.5139,
      "step": 404
    },
    {
      "epoch": 0.7003891050583657,
      "grad_norm": 1.3575890195381133,
      "learning_rate": 2.1988700012359865e-06,
      "loss": 0.4975,
      "step": 405
    },
    {
      "epoch": 0.7021184608733246,
      "grad_norm": 1.205731465743301,
      "learning_rate": 2.175679324369913e-06,
      "loss": 0.5196,
      "step": 406
    },
    {
      "epoch": 0.7038478166882837,
      "grad_norm": 1.2794680777002916,
      "learning_rate": 2.1525775341577404e-06,
      "loss": 0.4966,
      "step": 407
    },
    {
      "epoch": 0.7055771725032426,
      "grad_norm": 1.2296036184996322,
      "learning_rate": 2.1295653576560168e-06,
      "loss": 0.4951,
      "step": 408
    },
    {
      "epoch": 0.7073065283182015,
      "grad_norm": 1.2044790286403084,
      "learning_rate": 2.1066435191009717e-06,
      "loss": 0.4853,
      "step": 409
    },
    {
      "epoch": 0.7090358841331604,
      "grad_norm": 1.2218236581464923,
      "learning_rate": 2.083812739885738e-06,
      "loss": 0.4934,
      "step": 410
    },
    {
      "epoch": 0.7107652399481194,
      "grad_norm": 1.4009155350987372,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.4946,
      "step": 411
    },
    {
      "epoch": 0.7124945957630783,
      "grad_norm": 1.1812811747354344,
      "learning_rate": 2.038427230695565e-06,
      "loss": 0.4951,
      "step": 412
    },
    {
      "epoch": 0.7142239515780372,
      "grad_norm": 1.2191387291107492,
      "learning_rate": 2.0158739290874822e-06,
      "loss": 0.4986,
      "step": 413
    },
    {
      "epoch": 0.7159533073929961,
      "grad_norm": 1.329295095070758,
      "learning_rate": 1.9934145435079705e-06,
      "loss": 0.519,
      "step": 414
    },
    {
      "epoch": 0.717682663207955,
      "grad_norm": 1.2281812463093098,
      "learning_rate": 1.971049780795901e-06,
      "loss": 0.5497,
      "step": 415
    },
    {
      "epoch": 0.719412019022914,
      "grad_norm": 1.179567921331548,
      "learning_rate": 1.948780344812181e-06,
      "loss": 0.4875,
      "step": 416
    },
    {
      "epoch": 0.7211413748378729,
      "grad_norm": 1.3229803278267123,
      "learning_rate": 1.9266069364176144e-06,
      "loss": 0.513,
      "step": 417
    },
    {
      "epoch": 0.7228707306528318,
      "grad_norm": 1.207231556517107,
      "learning_rate": 1.9045302534508298e-06,
      "loss": 0.5077,
      "step": 418
    },
    {
      "epoch": 0.7246000864677907,
      "grad_norm": 1.239432725326481,
      "learning_rate": 1.8825509907063328e-06,
      "loss": 0.5081,
      "step": 419
    },
    {
      "epoch": 0.7263294422827496,
      "grad_norm": 1.2768017670747227,
      "learning_rate": 1.8606698399126261e-06,
      "loss": 0.5047,
      "step": 420
    },
    {
      "epoch": 0.7280587980977086,
      "grad_norm": 1.2443081672127343,
      "learning_rate": 1.838887489710452e-06,
      "loss": 0.5093,
      "step": 421
    },
    {
      "epoch": 0.7297881539126675,
      "grad_norm": 1.3248575565441225,
      "learning_rate": 1.8172046256311087e-06,
      "loss": 0.5096,
      "step": 422
    },
    {
      "epoch": 0.7315175097276264,
      "grad_norm": 1.2312772110980421,
      "learning_rate": 1.7956219300748796e-06,
      "loss": 0.4949,
      "step": 423
    },
    {
      "epoch": 0.7332468655425853,
      "grad_norm": 1.2725397947766917,
      "learning_rate": 1.7741400822895633e-06,
      "loss": 0.4913,
      "step": 424
    },
    {
      "epoch": 0.7349762213575444,
      "grad_norm": 1.2701112344747274,
      "learning_rate": 1.7527597583490825e-06,
      "loss": 0.5107,
      "step": 425
    },
    {
      "epoch": 0.7367055771725033,
      "grad_norm": 1.22988252225655,
      "learning_rate": 1.7314816311322219e-06,
      "loss": 0.487,
      "step": 426
    },
    {
      "epoch": 0.7384349329874622,
      "grad_norm": 1.23287252087033,
      "learning_rate": 1.7103063703014372e-06,
      "loss": 0.533,
      "step": 427
    },
    {
      "epoch": 0.7401642888024211,
      "grad_norm": 1.184456040905248,
      "learning_rate": 1.6892346422817945e-06,
      "loss": 0.4993,
      "step": 428
    },
    {
      "epoch": 0.74189364461738,
      "grad_norm": 1.220805415373837,
      "learning_rate": 1.6682671102399806e-06,
      "loss": 0.4865,
      "step": 429
    },
    {
      "epoch": 0.743623000432339,
      "grad_norm": 1.1994409780411663,
      "learning_rate": 1.647404434063447e-06,
      "loss": 0.4992,
      "step": 430
    },
    {
      "epoch": 0.7453523562472979,
      "grad_norm": 1.2631533307258702,
      "learning_rate": 1.6266472703396286e-06,
      "loss": 0.5236,
      "step": 431
    },
    {
      "epoch": 0.7470817120622568,
      "grad_norm": 1.1929507809675688,
      "learning_rate": 1.6059962723352912e-06,
      "loss": 0.4921,
      "step": 432
    },
    {
      "epoch": 0.7488110678772157,
      "grad_norm": 1.3115162066864332,
      "learning_rate": 1.5854520899759656e-06,
      "loss": 0.5123,
      "step": 433
    },
    {
      "epoch": 0.7505404236921747,
      "grad_norm": 1.204903938948009,
      "learning_rate": 1.5650153698254916e-06,
      "loss": 0.4761,
      "step": 434
    },
    {
      "epoch": 0.7522697795071336,
      "grad_norm": 1.3927511118217388,
      "learning_rate": 1.544686755065677e-06,
      "loss": 0.5209,
      "step": 435
    },
    {
      "epoch": 0.7539991353220925,
      "grad_norm": 1.1622881435387544,
      "learning_rate": 1.5244668854760459e-06,
      "loss": 0.5054,
      "step": 436
    },
    {
      "epoch": 0.7557284911370514,
      "grad_norm": 1.2474163957236213,
      "learning_rate": 1.5043563974137132e-06,
      "loss": 0.5037,
      "step": 437
    },
    {
      "epoch": 0.7574578469520103,
      "grad_norm": 1.2604380979516712,
      "learning_rate": 1.4843559237933474e-06,
      "loss": 0.4891,
      "step": 438
    },
    {
      "epoch": 0.7591872027669693,
      "grad_norm": 1.2530851990162881,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.4903,
      "step": 439
    },
    {
      "epoch": 0.7609165585819282,
      "grad_norm": 1.2300568194662433,
      "learning_rate": 1.4446875342055988e-06,
      "loss": 0.5059,
      "step": 440
    },
    {
      "epoch": 0.7626459143968871,
      "grad_norm": 1.2621446251787094,
      "learning_rate": 1.4250208666766235e-06,
      "loss": 0.5084,
      "step": 441
    },
    {
      "epoch": 0.764375270211846,
      "grad_norm": 1.1714965845789171,
      "learning_rate": 1.4054667104271497e-06,
      "loss": 0.4939,
      "step": 442
    },
    {
      "epoch": 0.7661046260268051,
      "grad_norm": 1.150326828517619,
      "learning_rate": 1.3860256808630429e-06,
      "loss": 0.4998,
      "step": 443
    },
    {
      "epoch": 0.767833981841764,
      "grad_norm": 1.4388728760063987,
      "learning_rate": 1.3666983898298659e-06,
      "loss": 0.506,
      "step": 444
    },
    {
      "epoch": 0.7695633376567229,
      "grad_norm": 1.211792558367054,
      "learning_rate": 1.3474854455936126e-06,
      "loss": 0.4908,
      "step": 445
    },
    {
      "epoch": 0.7712926934716818,
      "grad_norm": 1.1773193686251473,
      "learning_rate": 1.3283874528215735e-06,
      "loss": 0.4838,
      "step": 446
    },
    {
      "epoch": 0.7730220492866408,
      "grad_norm": 1.2910457142537168,
      "learning_rate": 1.3094050125632973e-06,
      "loss": 0.5221,
      "step": 447
    },
    {
      "epoch": 0.7747514051015997,
      "grad_norm": 1.2455298854301133,
      "learning_rate": 1.2905387222316824e-06,
      "loss": 0.5098,
      "step": 448
    },
    {
      "epoch": 0.7764807609165586,
      "grad_norm": 1.2221925176741943,
      "learning_rate": 1.2717891755841722e-06,
      "loss": 0.4968,
      "step": 449
    },
    {
      "epoch": 0.7782101167315175,
      "grad_norm": 1.2690867828669088,
      "learning_rate": 1.2531569627040636e-06,
      "loss": 0.497,
      "step": 450
    },
    {
      "epoch": 0.7799394725464764,
      "grad_norm": 1.1942118517525946,
      "learning_rate": 1.234642669981946e-06,
      "loss": 0.524,
      "step": 451
    },
    {
      "epoch": 0.7816688283614354,
      "grad_norm": 1.188395242204491,
      "learning_rate": 1.2162468800972344e-06,
      "loss": 0.4797,
      "step": 452
    },
    {
      "epoch": 0.7833981841763943,
      "grad_norm": 1.239090182793657,
      "learning_rate": 1.1979701719998454e-06,
      "loss": 0.4852,
      "step": 453
    },
    {
      "epoch": 0.7851275399913532,
      "grad_norm": 1.2443125589753226,
      "learning_rate": 1.1798131208919628e-06,
      "loss": 0.4869,
      "step": 454
    },
    {
      "epoch": 0.7868568958063121,
      "grad_norm": 1.2094210133113774,
      "learning_rate": 1.1617762982099446e-06,
      "loss": 0.4977,
      "step": 455
    },
    {
      "epoch": 0.788586251621271,
      "grad_norm": 1.2353201091140928,
      "learning_rate": 1.143860271606333e-06,
      "loss": 0.4867,
      "step": 456
    },
    {
      "epoch": 0.79031560743623,
      "grad_norm": 1.2334222843362082,
      "learning_rate": 1.1260656049319957e-06,
      "loss": 0.4934,
      "step": 457
    },
    {
      "epoch": 0.7920449632511889,
      "grad_norm": 0.9697687189599682,
      "learning_rate": 1.1083928582183712e-06,
      "loss": 0.5128,
      "step": 458
    },
    {
      "epoch": 0.7937743190661478,
      "grad_norm": 0.9095178347207882,
      "learning_rate": 1.0908425876598512e-06,
      "loss": 0.5031,
      "step": 459
    },
    {
      "epoch": 0.7955036748811067,
      "grad_norm": 0.9310196051126621,
      "learning_rate": 1.0734153455962765e-06,
      "loss": 0.5117,
      "step": 460
    },
    {
      "epoch": 0.7972330306960658,
      "grad_norm": 0.9730539710942893,
      "learning_rate": 1.0561116804955451e-06,
      "loss": 0.5113,
      "step": 461
    },
    {
      "epoch": 0.7989623865110247,
      "grad_norm": 0.894786526061026,
      "learning_rate": 1.0389321369363636e-06,
      "loss": 0.5018,
      "step": 462
    },
    {
      "epoch": 0.8006917423259836,
      "grad_norm": 0.9674498638331304,
      "learning_rate": 1.0218772555910955e-06,
      "loss": 0.5042,
      "step": 463
    },
    {
      "epoch": 0.8024210981409425,
      "grad_norm": 0.9438765948616291,
      "learning_rate": 1.004947573208756e-06,
      "loss": 0.5198,
      "step": 464
    },
    {
      "epoch": 0.8041504539559015,
      "grad_norm": 0.9763518960839491,
      "learning_rate": 9.881436225981107e-07,
      "loss": 0.5058,
      "step": 465
    },
    {
      "epoch": 0.8058798097708604,
      "grad_norm": 1.0177200696682724,
      "learning_rate": 9.714659326109138e-07,
      "loss": 0.5146,
      "step": 466
    },
    {
      "epoch": 0.8076091655858193,
      "grad_norm": 0.9855108276847996,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.5047,
      "step": 467
    },
    {
      "epoch": 0.8093385214007782,
      "grad_norm": 0.9820568110711639,
      "learning_rate": 9.384914300290749e-07,
      "loss": 0.5105,
      "step": 468
    },
    {
      "epoch": 0.8110678772157371,
      "grad_norm": 0.9543734782446319,
      "learning_rate": 9.221956552036992e-07,
      "loss": 0.51,
      "step": 469
    },
    {
      "epoch": 0.8127972330306961,
      "grad_norm": 1.0542711582531745,
      "learning_rate": 9.060282165076462e-07,
      "loss": 0.5025,
      "step": 470
    },
    {
      "epoch": 0.814526588845655,
      "grad_norm": 0.8575237177652391,
      "learning_rate": 8.899896227604509e-07,
      "loss": 0.4907,
      "step": 471
    },
    {
      "epoch": 0.8162559446606139,
      "grad_norm": 0.9537736768650427,
      "learning_rate": 8.740803787266522e-07,
      "loss": 0.502,
      "step": 472
    },
    {
      "epoch": 0.8179853004755728,
      "grad_norm": 0.940840984310344,
      "learning_rate": 8.58300985099918e-07,
      "loss": 0.5079,
      "step": 473
    },
    {
      "epoch": 0.8197146562905318,
      "grad_norm": 0.9855222060260419,
      "learning_rate": 8.426519384872733e-07,
      "loss": 0.5101,
      "step": 474
    },
    {
      "epoch": 0.8214440121054907,
      "grad_norm": 1.0112701653651899,
      "learning_rate": 8.271337313934869e-07,
      "loss": 0.5122,
      "step": 475
    },
    {
      "epoch": 0.8231733679204496,
      "grad_norm": 0.9202094573581726,
      "learning_rate": 8.117468522055578e-07,
      "loss": 0.5083,
      "step": 476
    },
    {
      "epoch": 0.8249027237354085,
      "grad_norm": 0.9312213063787499,
      "learning_rate": 7.964917851773496e-07,
      "loss": 0.5058,
      "step": 477
    },
    {
      "epoch": 0.8266320795503674,
      "grad_norm": 0.871339607306761,
      "learning_rate": 7.813690104143557e-07,
      "loss": 0.4966,
      "step": 478
    },
    {
      "epoch": 0.8283614353653265,
      "grad_norm": 0.8812181772889855,
      "learning_rate": 7.663790038585794e-07,
      "loss": 0.5021,
      "step": 479
    },
    {
      "epoch": 0.8300907911802854,
      "grad_norm": 0.9210803401855143,
      "learning_rate": 7.515222372735648e-07,
      "loss": 0.5084,
      "step": 480
    },
    {
      "epoch": 0.8318201469952443,
      "grad_norm": 0.9105295776120536,
      "learning_rate": 7.367991782295392e-07,
      "loss": 0.5066,
      "step": 481
    },
    {
      "epoch": 0.8335495028102032,
      "grad_norm": 0.9553285353198891,
      "learning_rate": 7.222102900887102e-07,
      "loss": 0.5137,
      "step": 482
    },
    {
      "epoch": 0.8352788586251622,
      "grad_norm": 0.8726021151068502,
      "learning_rate": 7.077560319906696e-07,
      "loss": 0.5172,
      "step": 483
    },
    {
      "epoch": 0.8370082144401211,
      "grad_norm": 0.907090839860992,
      "learning_rate": 6.934368588379553e-07,
      "loss": 0.507,
      "step": 484
    },
    {
      "epoch": 0.83873757025508,
      "grad_norm": 0.9601616142220535,
      "learning_rate": 6.792532212817271e-07,
      "loss": 0.4932,
      "step": 485
    },
    {
      "epoch": 0.8404669260700389,
      "grad_norm": 0.9038178937915357,
      "learning_rate": 6.652055657075845e-07,
      "loss": 0.5158,
      "step": 486
    },
    {
      "epoch": 0.8421962818849978,
      "grad_norm": 0.8811166725350369,
      "learning_rate": 6.512943342215234e-07,
      "loss": 0.4991,
      "step": 487
    },
    {
      "epoch": 0.8439256376999568,
      "grad_norm": 0.910468026382921,
      "learning_rate": 6.375199646360142e-07,
      "loss": 0.4942,
      "step": 488
    },
    {
      "epoch": 0.8456549935149157,
      "grad_norm": 0.9415463030818443,
      "learning_rate": 6.238828904562316e-07,
      "loss": 0.4988,
      "step": 489
    },
    {
      "epoch": 0.8473843493298746,
      "grad_norm": 0.9306562448394409,
      "learning_rate": 6.103835408664032e-07,
      "loss": 0.5218,
      "step": 490
    },
    {
      "epoch": 0.8491137051448335,
      "grad_norm": 0.9306072629397343,
      "learning_rate": 5.9702234071631e-07,
      "loss": 0.5018,
      "step": 491
    },
    {
      "epoch": 0.8508430609597925,
      "grad_norm": 0.9978545888527431,
      "learning_rate": 5.83799710507909e-07,
      "loss": 0.5015,
      "step": 492
    },
    {
      "epoch": 0.8525724167747514,
      "grad_norm": 0.9281144531309704,
      "learning_rate": 5.707160663821009e-07,
      "loss": 0.5154,
      "step": 493
    },
    {
      "epoch": 0.8543017725897103,
      "grad_norm": 0.9648920267544575,
      "learning_rate": 5.577718201056392e-07,
      "loss": 0.5024,
      "step": 494
    },
    {
      "epoch": 0.8560311284046692,
      "grad_norm": 0.9518063416993736,
      "learning_rate": 5.449673790581611e-07,
      "loss": 0.5026,
      "step": 495
    },
    {
      "epoch": 0.8577604842196281,
      "grad_norm": 0.9090864949437945,
      "learning_rate": 5.323031462193757e-07,
      "loss": 0.4975,
      "step": 496
    },
    {
      "epoch": 0.8594898400345872,
      "grad_norm": 0.9027482960362793,
      "learning_rate": 5.197795201563744e-07,
      "loss": 0.5077,
      "step": 497
    },
    {
      "epoch": 0.8612191958495461,
      "grad_norm": 0.9825856982352978,
      "learning_rate": 5.073968950110941e-07,
      "loss": 0.5175,
      "step": 498
    },
    {
      "epoch": 0.862948551664505,
      "grad_norm": 1.0033645469552896,
      "learning_rate": 4.951556604879049e-07,
      "loss": 0.5103,
      "step": 499
    },
    {
      "epoch": 0.8646779074794639,
      "grad_norm": 1.11450811759059,
      "learning_rate": 4.830562018413532e-07,
      "loss": 0.5135,
      "step": 500
    },
    {
      "epoch": 0.8664072632944229,
      "grad_norm": 0.9286119478176571,
      "learning_rate": 4.710988998640298e-07,
      "loss": 0.4853,
      "step": 501
    },
    {
      "epoch": 0.8681366191093818,
      "grad_norm": 0.9352342724589618,
      "learning_rate": 4.5928413087459325e-07,
      "loss": 0.5027,
      "step": 502
    },
    {
      "epoch": 0.8698659749243407,
      "grad_norm": 0.9742580013974624,
      "learning_rate": 4.4761226670592074e-07,
      "loss": 0.5007,
      "step": 503
    },
    {
      "epoch": 0.8715953307392996,
      "grad_norm": 0.9536093124243782,
      "learning_rate": 4.3608367469340553e-07,
      "loss": 0.4894,
      "step": 504
    },
    {
      "epoch": 0.8733246865542585,
      "grad_norm": 0.9739686798504419,
      "learning_rate": 4.2469871766340096e-07,
      "loss": 0.5024,
      "step": 505
    },
    {
      "epoch": 0.8750540423692175,
      "grad_norm": 0.9379743999219975,
      "learning_rate": 4.134577539217965e-07,
      "loss": 0.4924,
      "step": 506
    },
    {
      "epoch": 0.8767833981841764,
      "grad_norm": 1.003766963981924,
      "learning_rate": 4.0236113724274716e-07,
      "loss": 0.4878,
      "step": 507
    },
    {
      "epoch": 0.8785127539991353,
      "grad_norm": 0.9524223981180386,
      "learning_rate": 3.9140921685753065e-07,
      "loss": 0.5068,
      "step": 508
    },
    {
      "epoch": 0.8802421098140942,
      "grad_norm": 0.8535490006491876,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.4838,
      "step": 509
    },
    {
      "epoch": 0.8819714656290532,
      "grad_norm": 0.9082342591220833,
      "learning_rate": 3.6994083911356114e-07,
      "loss": 0.5176,
      "step": 510
    },
    {
      "epoch": 0.8837008214440121,
      "grad_norm": 0.9478217951483088,
      "learning_rate": 3.5942505740480583e-07,
      "loss": 0.4883,
      "step": 511
    },
    {
      "epoch": 0.885430177258971,
      "grad_norm": 0.9157515033348981,
      "learning_rate": 3.4905532326861946e-07,
      "loss": 0.4856,
      "step": 512
    },
    {
      "epoch": 0.8871595330739299,
      "grad_norm": 0.8748310499093214,
      "learning_rate": 3.3883196305992906e-07,
      "loss": 0.501,
      "step": 513
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.9579844131883266,
      "learning_rate": 3.287552985270015e-07,
      "loss": 0.5008,
      "step": 514
    },
    {
      "epoch": 0.8906182447038478,
      "grad_norm": 0.9866794087631354,
      "learning_rate": 3.18825646801314e-07,
      "loss": 0.5056,
      "step": 515
    },
    {
      "epoch": 0.8923476005188068,
      "grad_norm": 0.9637431191512927,
      "learning_rate": 3.0904332038757977e-07,
      "loss": 0.5165,
      "step": 516
    },
    {
      "epoch": 0.8940769563337657,
      "grad_norm": 0.9332323606841268,
      "learning_rate": 2.9940862715390483e-07,
      "loss": 0.5105,
      "step": 517
    },
    {
      "epoch": 0.8958063121487246,
      "grad_norm": 1.009400230071864,
      "learning_rate": 2.899218703221052e-07,
      "loss": 0.5128,
      "step": 518
    },
    {
      "epoch": 0.8975356679636836,
      "grad_norm": 0.8996826766855536,
      "learning_rate": 2.8058334845816214e-07,
      "loss": 0.5001,
      "step": 519
    },
    {
      "epoch": 0.8992650237786425,
      "grad_norm": 0.8529312049842479,
      "learning_rate": 2.7139335546282287e-07,
      "loss": 0.4989,
      "step": 520
    },
    {
      "epoch": 0.9009943795936014,
      "grad_norm": 0.9524051149908734,
      "learning_rate": 2.6235218056235633e-07,
      "loss": 0.5085,
      "step": 521
    },
    {
      "epoch": 0.9027237354085603,
      "grad_norm": 0.9450098346365634,
      "learning_rate": 2.534601082994437e-07,
      "loss": 0.5041,
      "step": 522
    },
    {
      "epoch": 0.9044530912235192,
      "grad_norm": 0.9688024639636753,
      "learning_rate": 2.447174185242324e-07,
      "loss": 0.5139,
      "step": 523
    },
    {
      "epoch": 0.9061824470384782,
      "grad_norm": 0.9362455552180009,
      "learning_rate": 2.3612438638551837e-07,
      "loss": 0.4956,
      "step": 524
    },
    {
      "epoch": 0.9079118028534371,
      "grad_norm": 0.8598614937743414,
      "learning_rate": 2.276812823220964e-07,
      "loss": 0.4875,
      "step": 525
    },
    {
      "epoch": 0.909641158668396,
      "grad_norm": 0.909540077460528,
      "learning_rate": 2.1938837205424002e-07,
      "loss": 0.5117,
      "step": 526
    },
    {
      "epoch": 0.9113705144833549,
      "grad_norm": 0.9096506960895155,
      "learning_rate": 2.1124591657534776e-07,
      "loss": 0.4955,
      "step": 527
    },
    {
      "epoch": 0.9130998702983139,
      "grad_norm": 1.0142474014859533,
      "learning_rate": 2.032541721437209e-07,
      "loss": 0.507,
      "step": 528
    },
    {
      "epoch": 0.9148292261132728,
      "grad_norm": 0.9131442997723532,
      "learning_rate": 1.9541339027450256e-07,
      "loss": 0.4895,
      "step": 529
    },
    {
      "epoch": 0.9165585819282317,
      "grad_norm": 0.9853142920778307,
      "learning_rate": 1.8772381773176417e-07,
      "loss": 0.5211,
      "step": 530
    },
    {
      "epoch": 0.9182879377431906,
      "grad_norm": 0.8974474768112111,
      "learning_rate": 1.801856965207338e-07,
      "loss": 0.4976,
      "step": 531
    },
    {
      "epoch": 0.9200172935581495,
      "grad_norm": 0.9550109239119172,
      "learning_rate": 1.7279926388018564e-07,
      "loss": 0.5151,
      "step": 532
    },
    {
      "epoch": 0.9217466493731085,
      "grad_norm": 0.9072672003349709,
      "learning_rate": 1.6556475227496816e-07,
      "loss": 0.516,
      "step": 533
    },
    {
      "epoch": 0.9234760051880675,
      "grad_norm": 0.9399754149330305,
      "learning_rate": 1.5848238938869332e-07,
      "loss": 0.4909,
      "step": 534
    },
    {
      "epoch": 0.9252053610030264,
      "grad_norm": 0.9437594133784243,
      "learning_rate": 1.5155239811656562e-07,
      "loss": 0.498,
      "step": 535
    },
    {
      "epoch": 0.9269347168179853,
      "grad_norm": 0.9357434836371697,
      "learning_rate": 1.4477499655837278e-07,
      "loss": 0.4969,
      "step": 536
    },
    {
      "epoch": 0.9286640726329443,
      "grad_norm": 0.8936101786239388,
      "learning_rate": 1.3815039801161723e-07,
      "loss": 0.4913,
      "step": 537
    },
    {
      "epoch": 0.9303934284479032,
      "grad_norm": 0.9484054108177385,
      "learning_rate": 1.3167881096480372e-07,
      "loss": 0.4988,
      "step": 538
    },
    {
      "epoch": 0.9321227842628621,
      "grad_norm": 0.9563796786443849,
      "learning_rate": 1.253604390908819e-07,
      "loss": 0.4925,
      "step": 539
    },
    {
      "epoch": 0.933852140077821,
      "grad_norm": 0.8950613019917822,
      "learning_rate": 1.191954812408308e-07,
      "loss": 0.5052,
      "step": 540
    },
    {
      "epoch": 0.93558149589278,
      "grad_norm": 0.9125279893925492,
      "learning_rate": 1.1318413143740436e-07,
      "loss": 0.5041,
      "step": 541
    },
    {
      "epoch": 0.9373108517077389,
      "grad_norm": 0.9141300635124516,
      "learning_rate": 1.0732657886902308e-07,
      "loss": 0.5096,
      "step": 542
    },
    {
      "epoch": 0.9390402075226978,
      "grad_norm": 0.9096167507969538,
      "learning_rate": 1.0162300788382263e-07,
      "loss": 0.4882,
      "step": 543
    },
    {
      "epoch": 0.9407695633376567,
      "grad_norm": 0.9753044785021144,
      "learning_rate": 9.607359798384785e-08,
      "loss": 0.5103,
      "step": 544
    },
    {
      "epoch": 0.9424989191526156,
      "grad_norm": 0.8747544695243777,
      "learning_rate": 9.0678523819408e-08,
      "loss": 0.504,
      "step": 545
    },
    {
      "epoch": 0.9442282749675746,
      "grad_norm": 0.8448600329999801,
      "learning_rate": 8.543795518357767e-08,
      "loss": 0.5049,
      "step": 546
    },
    {
      "epoch": 0.9459576307825335,
      "grad_norm": 0.9445491960031619,
      "learning_rate": 8.035205700685167e-08,
      "loss": 0.5033,
      "step": 547
    },
    {
      "epoch": 0.9476869865974924,
      "grad_norm": 0.9108140075113013,
      "learning_rate": 7.542098935195918e-08,
      "loss": 0.5138,
      "step": 548
    },
    {
      "epoch": 0.9494163424124513,
      "grad_norm": 0.9014111637252965,
      "learning_rate": 7.064490740882057e-08,
      "loss": 0.505,
      "step": 549
    },
    {
      "epoch": 0.9511456982274102,
      "grad_norm": 0.8618772410756893,
      "learning_rate": 6.602396148966795e-08,
      "loss": 0.4939,
      "step": 550
    },
    {
      "epoch": 0.9528750540423692,
      "grad_norm": 0.9034604742193665,
      "learning_rate": 6.15582970243117e-08,
      "loss": 0.4856,
      "step": 551
    },
    {
      "epoch": 0.9546044098573282,
      "grad_norm": 0.8858115334722476,
      "learning_rate": 5.7248054555563704e-08,
      "loss": 0.4943,
      "step": 552
    },
    {
      "epoch": 0.9563337656722871,
      "grad_norm": 0.9413567454097705,
      "learning_rate": 5.3093369734816824e-08,
      "loss": 0.4983,
      "step": 553
    },
    {
      "epoch": 0.958063121487246,
      "grad_norm": 0.8631416544061653,
      "learning_rate": 4.909437331777178e-08,
      "loss": 0.4991,
      "step": 554
    },
    {
      "epoch": 0.959792477302205,
      "grad_norm": 0.8972216946376435,
      "learning_rate": 4.52511911603265e-08,
      "loss": 0.4979,
      "step": 555
    },
    {
      "epoch": 0.9615218331171639,
      "grad_norm": 0.9017046495091536,
      "learning_rate": 4.15639442146093e-08,
      "loss": 0.4968,
      "step": 556
    },
    {
      "epoch": 0.9632511889321228,
      "grad_norm": 0.8883293939758925,
      "learning_rate": 3.8032748525179684e-08,
      "loss": 0.4959,
      "step": 557
    },
    {
      "epoch": 0.9649805447470817,
      "grad_norm": 0.9234481550882554,
      "learning_rate": 3.465771522536854e-08,
      "loss": 0.5218,
      "step": 558
    },
    {
      "epoch": 0.9667099005620406,
      "grad_norm": 0.914125086466903,
      "learning_rate": 3.143895053378698e-08,
      "loss": 0.5087,
      "step": 559
    },
    {
      "epoch": 0.9684392563769996,
      "grad_norm": 0.8985893082923397,
      "learning_rate": 2.837655575097964e-08,
      "loss": 0.5053,
      "step": 560
    },
    {
      "epoch": 0.9701686121919585,
      "grad_norm": 0.9385196676766113,
      "learning_rate": 2.547062725623828e-08,
      "loss": 0.5116,
      "step": 561
    },
    {
      "epoch": 0.9718979680069174,
      "grad_norm": 0.9646239444487571,
      "learning_rate": 2.2721256504567026e-08,
      "loss": 0.5072,
      "step": 562
    },
    {
      "epoch": 0.9736273238218763,
      "grad_norm": 0.8747663830607033,
      "learning_rate": 2.012853002380466e-08,
      "loss": 0.5022,
      "step": 563
    },
    {
      "epoch": 0.9753566796368353,
      "grad_norm": 0.9334336705435727,
      "learning_rate": 1.769252941190458e-08,
      "loss": 0.5105,
      "step": 564
    },
    {
      "epoch": 0.9770860354517942,
      "grad_norm": 0.9194588921972738,
      "learning_rate": 1.541333133436018e-08,
      "loss": 0.4985,
      "step": 565
    },
    {
      "epoch": 0.9788153912667531,
      "grad_norm": 0.9520800525639294,
      "learning_rate": 1.3291007521799015e-08,
      "loss": 0.4977,
      "step": 566
    },
    {
      "epoch": 0.980544747081712,
      "grad_norm": 0.9058630469142649,
      "learning_rate": 1.132562476771959e-08,
      "loss": 0.4738,
      "step": 567
    },
    {
      "epoch": 0.9822741028966709,
      "grad_norm": 0.8879281633731727,
      "learning_rate": 9.51724492639361e-09,
      "loss": 0.5048,
      "step": 568
    },
    {
      "epoch": 0.9840034587116299,
      "grad_norm": 0.8984029908537556,
      "learning_rate": 7.865924910916977e-09,
      "loss": 0.5024,
      "step": 569
    },
    {
      "epoch": 0.9857328145265889,
      "grad_norm": 0.9121742433892018,
      "learning_rate": 6.371716691419005e-09,
      "loss": 0.5082,
      "step": 570
    },
    {
      "epoch": 0.9874621703415478,
      "grad_norm": 0.9237443595386216,
      "learning_rate": 5.034667293427053e-09,
      "loss": 0.5097,
      "step": 571
    },
    {
      "epoch": 0.9891915261565067,
      "grad_norm": 0.9117600237724628,
      "learning_rate": 3.854818796385495e-09,
      "loss": 0.5129,
      "step": 572
    },
    {
      "epoch": 0.9909208819714657,
      "grad_norm": 0.9527043115093743,
      "learning_rate": 2.8322083323334417e-09,
      "loss": 0.5024,
      "step": 573
    },
    {
      "epoch": 0.9926502377864246,
      "grad_norm": 0.9724377615202221,
      "learning_rate": 1.9668680847356735e-09,
      "loss": 0.4966,
      "step": 574
    },
    {
      "epoch": 0.9943795936013835,
      "grad_norm": 0.9737523393148931,
      "learning_rate": 1.2588252874673469e-09,
      "loss": 0.4976,
      "step": 575
    },
    {
      "epoch": 0.9961089494163424,
      "grad_norm": 0.8474552307580481,
      "learning_rate": 7.081022239591173e-10,
      "loss": 0.4907,
      "step": 576
    },
    {
      "epoch": 0.9978383052313013,
      "grad_norm": 0.8918148859611189,
      "learning_rate": 3.147162264971471e-10,
      "loss": 0.4957,
      "step": 577
    },
    {
      "epoch": 0.9995676610462603,
      "grad_norm": 0.8738364123206687,
      "learning_rate": 7.867967567354306e-11,
      "loss": 0.4892,
      "step": 578
    },
    {
      "epoch": 0.9995676610462603,
      "step": 578,
      "total_flos": 6.685487298542305e+17,
      "train_loss": 0.5686832449324816,
      "train_runtime": 23812.8313,
      "train_samples_per_second": 7.141,
      "train_steps_per_second": 0.024
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 578,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.685487298542305e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}