{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 24129,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.1443905673670685e-05,
      "grad_norm": 0.7630084753036499,
      "learning_rate": 5e-06,
      "loss": 1.073,
      "step": 1
    },
    {
      "epoch": 8.288781134734137e-05,
      "grad_norm": 0.7379353046417236,
      "learning_rate": 4.999792780471633e-06,
      "loss": 1.0439,
      "step": 2
    },
    {
      "epoch": 0.00012433171702101205,
      "grad_norm": 0.5726526975631714,
      "learning_rate": 4.999585560943264e-06,
      "loss": 1.0112,
      "step": 3
    },
    {
      "epoch": 0.00016577562269468274,
      "grad_norm": 0.6247206330299377,
      "learning_rate": 4.999378341414895e-06,
      "loss": 1.0464,
      "step": 4
    },
    {
      "epoch": 0.00020721952836835343,
      "grad_norm": 0.5651471614837646,
      "learning_rate": 4.999171121886527e-06,
      "loss": 0.9998,
      "step": 5
    },
    {
      "epoch": 0.0002486634340420241,
      "grad_norm": 0.556253969669342,
      "learning_rate": 4.998963902358159e-06,
      "loss": 0.96,
      "step": 6
    },
    {
      "epoch": 0.0002901073397156948,
      "grad_norm": 0.5390304327011108,
      "learning_rate": 4.99875668282979e-06,
      "loss": 0.9778,
      "step": 7
    },
    {
      "epoch": 0.0003315512453893655,
      "grad_norm": 0.6542398929595947,
      "learning_rate": 4.998549463301422e-06,
      "loss": 1.0061,
      "step": 8
    },
    {
      "epoch": 0.0003729951510630362,
      "grad_norm": 0.7193869948387146,
      "learning_rate": 4.998342243773053e-06,
      "loss": 1.0396,
      "step": 9
    },
    {
      "epoch": 0.00041443905673670687,
      "grad_norm": 0.6612483859062195,
      "learning_rate": 4.998135024244685e-06,
      "loss": 0.9978,
      "step": 10
    },
    {
      "epoch": 0.00045588296241037753,
      "grad_norm": 0.6465231776237488,
      "learning_rate": 4.997927804716317e-06,
      "loss": 1.0269,
      "step": 11
    },
    {
      "epoch": 0.0004973268680840482,
      "grad_norm": 0.6349920034408569,
      "learning_rate": 4.997720585187948e-06,
      "loss": 0.958,
      "step": 12
    },
    {
      "epoch": 0.000538770773757719,
      "grad_norm": 0.6287173628807068,
      "learning_rate": 4.9975133656595804e-06,
      "loss": 1.0046,
      "step": 13
    },
    {
      "epoch": 0.0005802146794313896,
      "grad_norm": 0.5536159873008728,
      "learning_rate": 4.997306146131212e-06,
      "loss": 0.9456,
      "step": 14
    },
    {
      "epoch": 0.0006216585851050603,
      "grad_norm": 0.5690981149673462,
      "learning_rate": 4.997098926602843e-06,
      "loss": 0.9475,
      "step": 15
    },
    {
      "epoch": 0.000663102490778731,
      "grad_norm": 0.512193500995636,
      "learning_rate": 4.996891707074475e-06,
      "loss": 0.9431,
      "step": 16
    },
    {
      "epoch": 0.0007045463964524016,
      "grad_norm": 0.5570981502532959,
      "learning_rate": 4.996684487546107e-06,
      "loss": 0.9458,
      "step": 17
    },
    {
      "epoch": 0.0007459903021260724,
      "grad_norm": 0.5056589245796204,
      "learning_rate": 4.996477268017738e-06,
      "loss": 0.925,
      "step": 18
    },
    {
      "epoch": 0.0007874342077997431,
      "grad_norm": 0.5428594350814819,
      "learning_rate": 4.9962700484893704e-06,
      "loss": 0.9546,
      "step": 19
    },
    {
      "epoch": 0.0008288781134734137,
      "grad_norm": 0.5670416355133057,
      "learning_rate": 4.996062828961002e-06,
      "loss": 0.9832,
      "step": 20
    },
    {
      "epoch": 0.0008703220191470844,
      "grad_norm": 0.5489663481712341,
      "learning_rate": 4.995855609432633e-06,
      "loss": 0.9514,
      "step": 21
    },
    {
      "epoch": 0.0009117659248207551,
      "grad_norm": 0.5362726449966431,
      "learning_rate": 4.9956483899042654e-06,
      "loss": 0.9153,
      "step": 22
    },
    {
      "epoch": 0.0009532098304944258,
      "grad_norm": 0.5166146755218506,
      "learning_rate": 4.995441170375897e-06,
      "loss": 0.9258,
      "step": 23
    },
    {
      "epoch": 0.0009946537361680964,
      "grad_norm": 0.5216695666313171,
      "learning_rate": 4.995233950847528e-06,
      "loss": 0.9104,
      "step": 24
    },
    {
      "epoch": 0.0010360976418417672,
      "grad_norm": 0.5722406506538391,
      "learning_rate": 4.99502673131916e-06,
      "loss": 0.9812,
      "step": 25
    },
    {
      "epoch": 0.001077541547515438,
      "grad_norm": 0.599118709564209,
      "learning_rate": 4.994819511790792e-06,
      "loss": 0.9851,
      "step": 26
    },
    {
      "epoch": 0.0011189854531891085,
      "grad_norm": 0.5487611293792725,
      "learning_rate": 4.994612292262423e-06,
      "loss": 0.9912,
      "step": 27
    },
    {
      "epoch": 0.0011604293588627793,
      "grad_norm": 0.5384929180145264,
      "learning_rate": 4.994405072734055e-06,
      "loss": 0.9438,
      "step": 28
    },
    {
      "epoch": 0.0012018732645364498,
      "grad_norm": 0.5511430501937866,
      "learning_rate": 4.994197853205687e-06,
      "loss": 0.9172,
      "step": 29
    },
    {
      "epoch": 0.0012433171702101206,
      "grad_norm": 0.49488845467567444,
      "learning_rate": 4.993990633677318e-06,
      "loss": 0.8768,
      "step": 30
    },
    {
      "epoch": 0.0012847610758837914,
      "grad_norm": 0.4959312975406647,
      "learning_rate": 4.99378341414895e-06,
      "loss": 0.9048,
      "step": 31
    },
    {
      "epoch": 0.001326204981557462,
      "grad_norm": 0.531954824924469,
      "learning_rate": 4.993576194620581e-06,
      "loss": 0.9719,
      "step": 32
    },
    {
      "epoch": 0.0013676488872311327,
      "grad_norm": 0.5291857719421387,
      "learning_rate": 4.993368975092213e-06,
      "loss": 0.8948,
      "step": 33
    },
    {
      "epoch": 0.0014090927929048033,
      "grad_norm": 0.4953014552593231,
      "learning_rate": 4.993161755563845e-06,
      "loss": 0.8916,
      "step": 34
    },
    {
      "epoch": 0.001450536698578474,
      "grad_norm": 0.5361866354942322,
      "learning_rate": 4.992954536035477e-06,
      "loss": 0.9556,
      "step": 35
    },
    {
      "epoch": 0.0014919806042521448,
      "grad_norm": 0.5809301733970642,
      "learning_rate": 4.992747316507108e-06,
      "loss": 0.9189,
      "step": 36
    },
    {
      "epoch": 0.0015334245099258154,
      "grad_norm": 0.5245487689971924,
      "learning_rate": 4.99254009697874e-06,
      "loss": 0.9507,
      "step": 37
    },
    {
      "epoch": 0.0015748684155994861,
      "grad_norm": 0.5433439612388611,
      "learning_rate": 4.992332877450372e-06,
      "loss": 0.9312,
      "step": 38
    },
    {
      "epoch": 0.0016163123212731567,
      "grad_norm": 0.49423933029174805,
      "learning_rate": 4.992125657922003e-06,
      "loss": 0.8755,
      "step": 39
    },
    {
      "epoch": 0.0016577562269468275,
      "grad_norm": 0.5284749865531921,
      "learning_rate": 4.991918438393635e-06,
      "loss": 0.9126,
      "step": 40
    },
    {
      "epoch": 0.0016992001326204982,
      "grad_norm": 0.5137740969657898,
      "learning_rate": 4.991711218865266e-06,
      "loss": 0.9541,
      "step": 41
    },
    {
      "epoch": 0.0017406440382941688,
      "grad_norm": 0.5341567397117615,
      "learning_rate": 4.991503999336898e-06,
      "loss": 0.9265,
      "step": 42
    },
    {
      "epoch": 0.0017820879439678396,
      "grad_norm": 0.5049672722816467,
      "learning_rate": 4.99129677980853e-06,
      "loss": 0.8508,
      "step": 43
    },
    {
      "epoch": 0.0018235318496415101,
      "grad_norm": 0.548191487789154,
      "learning_rate": 4.991089560280161e-06,
      "loss": 0.8953,
      "step": 44
    },
    {
      "epoch": 0.001864975755315181,
      "grad_norm": 0.7055763602256775,
      "learning_rate": 4.990882340751793e-06,
      "loss": 0.9331,
      "step": 45
    },
    {
      "epoch": 0.0019064196609888517,
      "grad_norm": 0.4973026514053345,
      "learning_rate": 4.990675121223425e-06,
      "loss": 0.9463,
      "step": 46
    },
    {
      "epoch": 0.0019478635666625222,
      "grad_norm": 0.45394378900527954,
      "learning_rate": 4.990467901695056e-06,
      "loss": 0.8792,
      "step": 47
    },
    {
      "epoch": 0.0019893074723361928,
      "grad_norm": 0.45260077714920044,
      "learning_rate": 4.990260682166687e-06,
      "loss": 0.8354,
      "step": 48
    },
    {
      "epoch": 0.0020307513780098638,
      "grad_norm": 0.502605140209198,
      "learning_rate": 4.99005346263832e-06,
      "loss": 0.8909,
      "step": 49
    },
    {
      "epoch": 0.0020721952836835343,
      "grad_norm": 0.46680155396461487,
      "learning_rate": 4.989846243109951e-06,
      "loss": 0.8728,
      "step": 50
    },
    {
      "epoch": 0.002113639189357205,
      "grad_norm": 0.4857476055622101,
      "learning_rate": 4.989639023581582e-06,
      "loss": 0.8711,
      "step": 51
    },
    {
      "epoch": 0.002155083095030876,
      "grad_norm": 0.49140453338623047,
      "learning_rate": 4.989431804053214e-06,
      "loss": 0.895,
      "step": 52
    },
    {
      "epoch": 0.0021965270007045464,
      "grad_norm": 0.4997093379497528,
      "learning_rate": 4.989224584524846e-06,
      "loss": 0.9133,
      "step": 53
    },
    {
      "epoch": 0.002237970906378217,
      "grad_norm": 0.5029782652854919,
      "learning_rate": 4.989017364996478e-06,
      "loss": 0.8987,
      "step": 54
    },
    {
      "epoch": 0.002279414812051888,
      "grad_norm": 0.5551053881645203,
      "learning_rate": 4.98881014546811e-06,
      "loss": 0.9167,
      "step": 55
    },
    {
      "epoch": 0.0023208587177255585,
      "grad_norm": 0.49238938093185425,
      "learning_rate": 4.988602925939741e-06,
      "loss": 0.865,
      "step": 56
    },
    {
      "epoch": 0.002362302623399229,
      "grad_norm": 0.49992960691452026,
      "learning_rate": 4.988395706411372e-06,
      "loss": 0.9646,
      "step": 57
    },
    {
      "epoch": 0.0024037465290728996,
      "grad_norm": 0.555106520652771,
      "learning_rate": 4.988188486883005e-06,
      "loss": 0.812,
      "step": 58
    },
    {
      "epoch": 0.0024451904347465706,
      "grad_norm": 0.5102022886276245,
      "learning_rate": 4.987981267354636e-06,
      "loss": 0.8608,
      "step": 59
    },
    {
      "epoch": 0.002486634340420241,
      "grad_norm": 0.5039694309234619,
      "learning_rate": 4.987774047826267e-06,
      "loss": 0.9617,
      "step": 60
    },
    {
      "epoch": 0.0025280782460939118,
      "grad_norm": 0.5221696496009827,
      "learning_rate": 4.987566828297899e-06,
      "loss": 0.8892,
      "step": 61
    },
    {
      "epoch": 0.0025695221517675827,
      "grad_norm": 0.5181254744529724,
      "learning_rate": 4.987359608769531e-06,
      "loss": 0.991,
      "step": 62
    },
    {
      "epoch": 0.0026109660574412533,
      "grad_norm": 0.5053512454032898,
      "learning_rate": 4.987152389241162e-06,
      "loss": 0.9097,
      "step": 63
    },
    {
      "epoch": 0.002652409963114924,
      "grad_norm": 0.49161139130592346,
      "learning_rate": 4.986945169712794e-06,
      "loss": 0.8901,
      "step": 64
    },
    {
      "epoch": 0.002693853868788595,
      "grad_norm": 0.4739881753921509,
      "learning_rate": 4.986737950184426e-06,
      "loss": 0.8699,
      "step": 65
    },
    {
      "epoch": 0.0027352977744622654,
      "grad_norm": 0.5461541414260864,
      "learning_rate": 4.986530730656057e-06,
      "loss": 0.8945,
      "step": 66
    },
    {
      "epoch": 0.002776741680135936,
      "grad_norm": 0.5793424844741821,
      "learning_rate": 4.986323511127689e-06,
      "loss": 0.936,
      "step": 67
    },
    {
      "epoch": 0.0028181855858096065,
      "grad_norm": 0.4656437039375305,
      "learning_rate": 4.98611629159932e-06,
      "loss": 0.8499,
      "step": 68
    },
    {
      "epoch": 0.0028596294914832775,
      "grad_norm": 0.5063554644584656,
      "learning_rate": 4.985909072070952e-06,
      "loss": 0.8713,
      "step": 69
    },
    {
      "epoch": 0.002901073397156948,
      "grad_norm": 0.545987606048584,
      "learning_rate": 4.985701852542584e-06,
      "loss": 0.957,
      "step": 70
    },
    {
      "epoch": 0.0029425173028306186,
      "grad_norm": 0.5510537028312683,
      "learning_rate": 4.985494633014216e-06,
      "loss": 0.8796,
      "step": 71
    },
    {
      "epoch": 0.0029839612085042896,
      "grad_norm": 0.5463219285011292,
      "learning_rate": 4.9852874134858474e-06,
      "loss": 0.9336,
      "step": 72
    },
    {
      "epoch": 0.00302540511417796,
      "grad_norm": 0.5715978145599365,
      "learning_rate": 4.985080193957479e-06,
      "loss": 0.9844,
      "step": 73
    },
    {
      "epoch": 0.0030668490198516307,
      "grad_norm": 0.5243675112724304,
      "learning_rate": 4.984872974429111e-06,
      "loss": 0.8474,
      "step": 74
    },
    {
      "epoch": 0.0031082929255253017,
      "grad_norm": 0.47537675499916077,
      "learning_rate": 4.9846657549007424e-06,
      "loss": 0.8713,
      "step": 75
    },
    {
      "epoch": 0.0031497368311989723,
      "grad_norm": 0.48445916175842285,
      "learning_rate": 4.984458535372374e-06,
      "loss": 0.9302,
      "step": 76
    },
    {
      "epoch": 0.003191180736872643,
      "grad_norm": 0.5203693509101868,
      "learning_rate": 4.984251315844005e-06,
      "loss": 0.894,
      "step": 77
    },
    {
      "epoch": 0.0032326246425463134,
      "grad_norm": 0.48706889152526855,
      "learning_rate": 4.9840440963156374e-06,
      "loss": 0.8809,
      "step": 78
    },
    {
      "epoch": 0.0032740685482199844,
      "grad_norm": 0.4953436553478241,
      "learning_rate": 4.983836876787269e-06,
      "loss": 0.8833,
      "step": 79
    },
    {
      "epoch": 0.003315512453893655,
      "grad_norm": 0.5255720615386963,
      "learning_rate": 4.9836296572589e-06,
      "loss": 0.9133,
      "step": 80
    },
    {
      "epoch": 0.0033569563595673255,
      "grad_norm": 0.5064576864242554,
      "learning_rate": 4.9834224377305324e-06,
      "loss": 0.8767,
      "step": 81
    },
    {
      "epoch": 0.0033984002652409965,
      "grad_norm": 0.4973648190498352,
      "learning_rate": 4.983215218202164e-06,
      "loss": 0.8743,
      "step": 82
    },
    {
      "epoch": 0.003439844170914667,
      "grad_norm": 0.5787176489830017,
      "learning_rate": 4.983007998673795e-06,
      "loss": 0.9216,
      "step": 83
    },
    {
      "epoch": 0.0034812880765883376,
      "grad_norm": 0.5432944893836975,
      "learning_rate": 4.982800779145427e-06,
      "loss": 0.8425,
      "step": 84
    },
    {
      "epoch": 0.0035227319822620086,
      "grad_norm": 0.5159140825271606,
      "learning_rate": 4.982593559617059e-06,
      "loss": 0.8672,
      "step": 85
    },
    {
      "epoch": 0.003564175887935679,
      "grad_norm": 0.4964894950389862,
      "learning_rate": 4.98238634008869e-06,
      "loss": 0.8904,
      "step": 86
    },
    {
      "epoch": 0.0036056197936093497,
      "grad_norm": 0.4951587915420532,
      "learning_rate": 4.9821791205603224e-06,
      "loss": 0.8574,
      "step": 87
    },
    {
      "epoch": 0.0036470636992830202,
      "grad_norm": 0.46900156140327454,
      "learning_rate": 4.981971901031953e-06,
      "loss": 0.8496,
      "step": 88
    },
    {
      "epoch": 0.0036885076049566912,
      "grad_norm": 0.5107403993606567,
      "learning_rate": 4.981764681503585e-06,
      "loss": 0.8232,
      "step": 89
    },
    {
      "epoch": 0.003729951510630362,
      "grad_norm": 0.4707816541194916,
      "learning_rate": 4.9815574619752174e-06,
      "loss": 0.8472,
      "step": 90
    },
    {
      "epoch": 0.0037713954163040323,
      "grad_norm": 0.5420551896095276,
      "learning_rate": 4.981350242446849e-06,
      "loss": 0.8201,
      "step": 91
    },
    {
      "epoch": 0.0038128393219777033,
      "grad_norm": 0.4953294098377228,
      "learning_rate": 4.98114302291848e-06,
      "loss": 0.8284,
      "step": 92
    },
    {
      "epoch": 0.003854283227651374,
      "grad_norm": 0.5386088490486145,
      "learning_rate": 4.980935803390112e-06,
      "loss": 0.9541,
      "step": 93
    },
    {
      "epoch": 0.0038957271333250445,
      "grad_norm": 0.5048432350158691,
      "learning_rate": 4.980728583861744e-06,
      "loss": 0.8738,
      "step": 94
    },
    {
      "epoch": 0.0039371710389987154,
      "grad_norm": 0.5135311484336853,
      "learning_rate": 4.980521364333375e-06,
      "loss": 0.8145,
      "step": 95
    },
    {
      "epoch": 0.0039786149446723856,
      "grad_norm": 0.5058515667915344,
      "learning_rate": 4.980314144805007e-06,
      "loss": 0.8142,
      "step": 96
    },
    {
      "epoch": 0.0040200588503460566,
      "grad_norm": 0.47229743003845215,
      "learning_rate": 4.980106925276639e-06,
      "loss": 0.8455,
      "step": 97
    },
    {
      "epoch": 0.0040615027560197275,
      "grad_norm": 0.485802561044693,
      "learning_rate": 4.97989970574827e-06,
      "loss": 0.8601,
      "step": 98
    },
    {
      "epoch": 0.004102946661693398,
      "grad_norm": 0.4353331923484802,
      "learning_rate": 4.979692486219902e-06,
      "loss": 0.8562,
      "step": 99
    },
    {
      "epoch": 0.004144390567367069,
      "grad_norm": 0.4869312644004822,
      "learning_rate": 4.979485266691533e-06,
      "loss": 0.8806,
      "step": 100
    },
    {
      "epoch": 0.00418583447304074,
      "grad_norm": 0.5096659064292908,
      "learning_rate": 4.979278047163165e-06,
      "loss": 0.8943,
      "step": 101
    },
    {
      "epoch": 0.00422727837871441,
      "grad_norm": 0.5356626510620117,
      "learning_rate": 4.979070827634797e-06,
      "loss": 0.8645,
      "step": 102
    },
    {
      "epoch": 0.004268722284388081,
      "grad_norm": 0.5393323302268982,
      "learning_rate": 4.978863608106429e-06,
      "loss": 0.8652,
      "step": 103
    },
    {
      "epoch": 0.004310166190061752,
      "grad_norm": 0.4678276777267456,
      "learning_rate": 4.978656388578059e-06,
      "loss": 0.8943,
      "step": 104
    },
    {
      "epoch": 0.004351610095735422,
      "grad_norm": 0.5040932893753052,
      "learning_rate": 4.978449169049692e-06,
      "loss": 0.8877,
      "step": 105
    },
    {
      "epoch": 0.004393054001409093,
      "grad_norm": 0.4834267199039459,
      "learning_rate": 4.978241949521324e-06,
      "loss": 0.8384,
      "step": 106
    },
    {
      "epoch": 0.004434497907082764,
      "grad_norm": 0.555371105670929,
      "learning_rate": 4.978034729992955e-06,
      "loss": 0.9224,
      "step": 107
    },
    {
      "epoch": 0.004475941812756434,
      "grad_norm": 0.5358298420906067,
      "learning_rate": 4.977827510464587e-06,
      "loss": 0.874,
      "step": 108
    },
    {
      "epoch": 0.004517385718430105,
      "grad_norm": 0.49929314851760864,
      "learning_rate": 4.977620290936218e-06,
      "loss": 0.8464,
      "step": 109
    },
    {
      "epoch": 0.004558829624103776,
      "grad_norm": 0.5089736580848694,
      "learning_rate": 4.97741307140785e-06,
      "loss": 0.866,
      "step": 110
    },
    {
      "epoch": 0.004600273529777446,
      "grad_norm": 0.506629228591919,
      "learning_rate": 4.977205851879482e-06,
      "loss": 0.8469,
      "step": 111
    },
    {
      "epoch": 0.004641717435451117,
      "grad_norm": 0.4904714524745941,
      "learning_rate": 4.976998632351113e-06,
      "loss": 0.8372,
      "step": 112
    },
    {
      "epoch": 0.004683161341124787,
      "grad_norm": 0.4875148832798004,
      "learning_rate": 4.976791412822744e-06,
      "loss": 0.915,
      "step": 113
    },
    {
      "epoch": 0.004724605246798458,
      "grad_norm": 0.5229685306549072,
      "learning_rate": 4.976584193294377e-06,
      "loss": 0.8477,
      "step": 114
    },
    {
      "epoch": 0.004766049152472129,
      "grad_norm": 0.5013989210128784,
      "learning_rate": 4.976376973766008e-06,
      "loss": 0.8608,
      "step": 115
    },
    {
      "epoch": 0.004807493058145799,
      "grad_norm": 0.5142388939857483,
      "learning_rate": 4.976169754237639e-06,
      "loss": 0.8477,
      "step": 116
    },
    {
      "epoch": 0.00484893696381947,
      "grad_norm": 0.5292401909828186,
      "learning_rate": 4.975962534709272e-06,
      "loss": 0.8997,
      "step": 117
    },
    {
      "epoch": 0.004890380869493141,
      "grad_norm": 0.5275249481201172,
      "learning_rate": 4.975755315180903e-06,
      "loss": 0.8364,
      "step": 118
    },
    {
      "epoch": 0.004931824775166811,
      "grad_norm": 0.5310258269309998,
      "learning_rate": 4.975548095652534e-06,
      "loss": 0.8779,
      "step": 119
    },
    {
      "epoch": 0.004973268680840482,
      "grad_norm": 0.4797632098197937,
      "learning_rate": 4.975340876124166e-06,
      "loss": 0.7776,
      "step": 120
    },
    {
      "epoch": 0.005014712586514153,
      "grad_norm": 0.5749223232269287,
      "learning_rate": 4.975133656595798e-06,
      "loss": 0.8945,
      "step": 121
    },
    {
      "epoch": 0.0050561564921878235,
      "grad_norm": 0.4948255717754364,
      "learning_rate": 4.974926437067429e-06,
      "loss": 0.8572,
      "step": 122
    },
    {
      "epoch": 0.0050976003978614945,
      "grad_norm": 0.4390929341316223,
      "learning_rate": 4.974719217539062e-06,
      "loss": 0.8337,
      "step": 123
    },
    {
      "epoch": 0.0051390443035351655,
      "grad_norm": 0.5416498184204102,
      "learning_rate": 4.974511998010693e-06,
      "loss": 0.9121,
      "step": 124
    },
    {
      "epoch": 0.005180488209208836,
      "grad_norm": 0.5238378643989563,
      "learning_rate": 4.974304778482324e-06,
      "loss": 0.8374,
      "step": 125
    },
    {
      "epoch": 0.005221932114882507,
      "grad_norm": 0.4952799677848816,
      "learning_rate": 4.974097558953957e-06,
      "loss": 0.8687,
      "step": 126
    },
    {
      "epoch": 0.005263376020556178,
      "grad_norm": 0.4925903379917145,
      "learning_rate": 4.973890339425588e-06,
      "loss": 0.8286,
      "step": 127
    },
    {
      "epoch": 0.005304819926229848,
      "grad_norm": 0.4815710783004761,
      "learning_rate": 4.973683119897219e-06,
      "loss": 0.8416,
      "step": 128
    },
    {
      "epoch": 0.005346263831903519,
      "grad_norm": 0.5659744739532471,
      "learning_rate": 4.973475900368851e-06,
      "loss": 0.8613,
      "step": 129
    },
    {
      "epoch": 0.00538770773757719,
      "grad_norm": 0.4779456853866577,
      "learning_rate": 4.973268680840483e-06,
      "loss": 0.8179,
      "step": 130
    },
    {
      "epoch": 0.00542915164325086,
      "grad_norm": 0.44191497564315796,
      "learning_rate": 4.9730614613121144e-06,
      "loss": 0.8591,
      "step": 131
    },
    {
      "epoch": 0.005470595548924531,
      "grad_norm": 0.4689691960811615,
      "learning_rate": 4.972854241783746e-06,
      "loss": 0.8972,
      "step": 132
    },
    {
      "epoch": 0.005512039454598201,
      "grad_norm": 0.4754927456378937,
      "learning_rate": 4.972647022255378e-06,
      "loss": 0.8386,
      "step": 133
    },
    {
      "epoch": 0.005553483360271872,
      "grad_norm": 0.5093626976013184,
      "learning_rate": 4.9724398027270094e-06,
      "loss": 0.9167,
      "step": 134
    },
    {
      "epoch": 0.005594927265945543,
      "grad_norm": 0.5663705468177795,
      "learning_rate": 4.972232583198641e-06,
      "loss": 0.9009,
      "step": 135
    },
    {
      "epoch": 0.005636371171619213,
      "grad_norm": 0.5459282398223877,
      "learning_rate": 4.972025363670272e-06,
      "loss": 0.9341,
      "step": 136
    },
    {
      "epoch": 0.005677815077292884,
      "grad_norm": 0.5546956658363342,
      "learning_rate": 4.9718181441419044e-06,
      "loss": 0.9043,
      "step": 137
    },
    {
      "epoch": 0.005719258982966555,
      "grad_norm": 0.5535392761230469,
      "learning_rate": 4.971610924613536e-06,
      "loss": 0.853,
      "step": 138
    },
    {
      "epoch": 0.005760702888640225,
      "grad_norm": 0.5206494927406311,
      "learning_rate": 4.971403705085168e-06,
      "loss": 0.9067,
      "step": 139
    },
    {
      "epoch": 0.005802146794313896,
      "grad_norm": 0.5393713712692261,
      "learning_rate": 4.971196485556799e-06,
      "loss": 0.8645,
      "step": 140
    },
    {
      "epoch": 0.005843590699987567,
      "grad_norm": 0.5049880146980286,
      "learning_rate": 4.970989266028431e-06,
      "loss": 0.8533,
      "step": 141
    },
    {
      "epoch": 0.005885034605661237,
      "grad_norm": 0.5119956731796265,
      "learning_rate": 4.970782046500063e-06,
      "loss": 0.7871,
      "step": 142
    },
    {
      "epoch": 0.005926478511334908,
      "grad_norm": 0.4828684329986572,
      "learning_rate": 4.9705748269716944e-06,
      "loss": 0.864,
      "step": 143
    },
    {
      "epoch": 0.005967922417008579,
      "grad_norm": 0.4998629689216614,
      "learning_rate": 4.970367607443326e-06,
      "loss": 0.9089,
      "step": 144
    },
    {
      "epoch": 0.006009366322682249,
      "grad_norm": 0.5004873871803284,
      "learning_rate": 4.970160387914957e-06,
      "loss": 0.8389,
      "step": 145
    },
    {
      "epoch": 0.00605081022835592,
      "grad_norm": 0.525513768196106,
      "learning_rate": 4.9699531683865894e-06,
      "loss": 0.9302,
      "step": 146
    },
    {
      "epoch": 0.006092254134029591,
      "grad_norm": 0.4276360273361206,
      "learning_rate": 4.969745948858221e-06,
      "loss": 0.7893,
      "step": 147
    },
    {
      "epoch": 0.0061336980397032614,
      "grad_norm": 0.5161129832267761,
      "learning_rate": 4.969538729329852e-06,
      "loss": 0.9001,
      "step": 148
    },
    {
      "epoch": 0.006175141945376932,
      "grad_norm": 0.5208853483200073,
      "learning_rate": 4.969331509801484e-06,
      "loss": 0.8379,
      "step": 149
    },
    {
      "epoch": 0.006216585851050603,
      "grad_norm": 0.4908878207206726,
      "learning_rate": 4.969124290273116e-06,
      "loss": 0.8953,
      "step": 150
    },
    {
      "epoch": 0.0062580297567242735,
      "grad_norm": 0.46791812777519226,
      "learning_rate": 4.968917070744747e-06,
      "loss": 0.8303,
      "step": 151
    },
    {
      "epoch": 0.0062994736623979445,
      "grad_norm": 0.4564230740070343,
      "learning_rate": 4.968709851216379e-06,
      "loss": 0.9202,
      "step": 152
    },
    {
      "epoch": 0.006340917568071615,
      "grad_norm": 0.4829869270324707,
      "learning_rate": 4.968502631688011e-06,
      "loss": 0.8589,
      "step": 153
    },
    {
      "epoch": 0.006382361473745286,
      "grad_norm": 0.5078172087669373,
      "learning_rate": 4.968295412159642e-06,
      "loss": 0.8279,
      "step": 154
    },
    {
      "epoch": 0.006423805379418957,
      "grad_norm": 0.4795092046260834,
      "learning_rate": 4.9680881926312745e-06,
      "loss": 0.8956,
      "step": 155
    },
    {
      "epoch": 0.006465249285092627,
      "grad_norm": 0.4985639452934265,
      "learning_rate": 4.967880973102905e-06,
      "loss": 0.8203,
      "step": 156
    },
    {
      "epoch": 0.006506693190766298,
      "grad_norm": 0.5322465300559998,
      "learning_rate": 4.967673753574537e-06,
      "loss": 0.9133,
      "step": 157
    },
    {
      "epoch": 0.006548137096439969,
      "grad_norm": 0.5147272348403931,
      "learning_rate": 4.9674665340461695e-06,
      "loss": 0.894,
      "step": 158
    },
    {
      "epoch": 0.006589581002113639,
      "grad_norm": 0.5311028361320496,
      "learning_rate": 4.967259314517801e-06,
      "loss": 0.9231,
      "step": 159
    },
    {
      "epoch": 0.00663102490778731,
      "grad_norm": 0.4999554753303528,
      "learning_rate": 4.967052094989432e-06,
      "loss": 0.8215,
      "step": 160
    },
    {
      "epoch": 0.006672468813460981,
      "grad_norm": 0.5298110246658325,
      "learning_rate": 4.966844875461064e-06,
      "loss": 0.8933,
      "step": 161
    },
    {
      "epoch": 0.006713912719134651,
      "grad_norm": 0.46060001850128174,
      "learning_rate": 4.966637655932696e-06,
      "loss": 0.7771,
      "step": 162
    },
    {
      "epoch": 0.006755356624808322,
      "grad_norm": 0.4887921214103699,
      "learning_rate": 4.966430436404327e-06,
      "loss": 0.8943,
      "step": 163
    },
    {
      "epoch": 0.006796800530481993,
      "grad_norm": 0.5011575222015381,
      "learning_rate": 4.966223216875959e-06,
      "loss": 0.9082,
      "step": 164
    },
    {
      "epoch": 0.006838244436155663,
      "grad_norm": 0.5304556488990784,
      "learning_rate": 4.96601599734759e-06,
      "loss": 0.8735,
      "step": 165
    },
    {
      "epoch": 0.006879688341829334,
      "grad_norm": 0.522753119468689,
      "learning_rate": 4.965808777819222e-06,
      "loss": 0.8738,
      "step": 166
    },
    {
      "epoch": 0.006921132247503005,
      "grad_norm": 0.4741937220096588,
      "learning_rate": 4.965601558290854e-06,
      "loss": 0.855,
      "step": 167
    },
    {
      "epoch": 0.006962576153176675,
      "grad_norm": 0.5090004205703735,
      "learning_rate": 4.965394338762485e-06,
      "loss": 0.8445,
      "step": 168
    },
    {
      "epoch": 0.007004020058850346,
      "grad_norm": 0.6183812618255615,
      "learning_rate": 4.965187119234117e-06,
      "loss": 0.9092,
      "step": 169
    },
    {
      "epoch": 0.007045463964524017,
      "grad_norm": 0.5028287768363953,
      "learning_rate": 4.964979899705749e-06,
      "loss": 0.8486,
      "step": 170
    },
    {
      "epoch": 0.007086907870197687,
      "grad_norm": 0.513696014881134,
      "learning_rate": 4.964772680177381e-06,
      "loss": 0.9048,
      "step": 171
    },
    {
      "epoch": 0.007128351775871358,
      "grad_norm": 0.44042113423347473,
      "learning_rate": 4.964565460649011e-06,
      "loss": 0.748,
      "step": 172
    },
    {
      "epoch": 0.007169795681545028,
      "grad_norm": 0.5216784477233887,
      "learning_rate": 4.964358241120644e-06,
      "loss": 0.8972,
      "step": 173
    },
    {
      "epoch": 0.007211239587218699,
      "grad_norm": 0.5251628160476685,
      "learning_rate": 4.964151021592275e-06,
      "loss": 0.8831,
      "step": 174
    },
    {
      "epoch": 0.00725268349289237,
      "grad_norm": 0.5277590751647949,
      "learning_rate": 4.963943802063907e-06,
      "loss": 0.9287,
      "step": 175
    },
    {
      "epoch": 0.0072941273985660405,
      "grad_norm": 0.4672757387161255,
      "learning_rate": 4.963736582535539e-06,
      "loss": 0.8489,
      "step": 176
    },
    {
      "epoch": 0.0073355713042397115,
      "grad_norm": 0.5269532203674316,
      "learning_rate": 4.96352936300717e-06,
      "loss": 0.8674,
      "step": 177
    },
    {
      "epoch": 0.0073770152099133825,
      "grad_norm": 0.5004445910453796,
      "learning_rate": 4.963322143478802e-06,
      "loss": 0.8079,
      "step": 178
    },
    {
      "epoch": 0.007418459115587053,
      "grad_norm": 0.5009570717811584,
      "learning_rate": 4.963114923950434e-06,
      "loss": 0.8538,
      "step": 179
    },
    {
      "epoch": 0.007459903021260724,
      "grad_norm": 0.473165899515152,
      "learning_rate": 4.962907704422065e-06,
      "loss": 0.801,
      "step": 180
    },
    {
      "epoch": 0.007501346926934395,
      "grad_norm": 0.4999670088291168,
      "learning_rate": 4.962700484893696e-06,
      "loss": 0.8811,
      "step": 181
    },
    {
      "epoch": 0.007542790832608065,
      "grad_norm": 0.4970436692237854,
      "learning_rate": 4.962493265365329e-06,
      "loss": 0.9043,
      "step": 182
    },
    {
      "epoch": 0.007584234738281736,
      "grad_norm": 0.5126577615737915,
      "learning_rate": 4.96228604583696e-06,
      "loss": 0.9436,
      "step": 183
    },
    {
      "epoch": 0.007625678643955407,
      "grad_norm": 0.48738300800323486,
      "learning_rate": 4.962078826308591e-06,
      "loss": 0.8391,
      "step": 184
    },
    {
      "epoch": 0.007667122549629077,
      "grad_norm": 0.5486866235733032,
      "learning_rate": 4.961871606780224e-06,
      "loss": 0.8625,
      "step": 185
    },
    {
      "epoch": 0.007708566455302748,
      "grad_norm": 0.507525622844696,
      "learning_rate": 4.961664387251855e-06,
      "loss": 0.856,
      "step": 186
    },
    {
      "epoch": 0.007750010360976419,
      "grad_norm": 0.5589158535003662,
      "learning_rate": 4.961457167723486e-06,
      "loss": 0.8716,
      "step": 187
    },
    {
      "epoch": 0.007791454266650089,
      "grad_norm": 0.4950938820838928,
      "learning_rate": 4.961249948195118e-06,
      "loss": 0.877,
      "step": 188
    },
    {
      "epoch": 0.007832898172323759,
      "grad_norm": 0.5249196887016296,
      "learning_rate": 4.96104272866675e-06,
      "loss": 0.9199,
      "step": 189
    },
    {
      "epoch": 0.007874342077997431,
      "grad_norm": 0.46564406156539917,
      "learning_rate": 4.9608355091383814e-06,
      "loss": 0.8347,
      "step": 190
    },
    {
      "epoch": 0.007915785983671101,
      "grad_norm": 0.5101157426834106,
      "learning_rate": 4.960628289610014e-06,
      "loss": 0.9353,
      "step": 191
    },
    {
      "epoch": 0.007957229889344771,
      "grad_norm": 0.4670466184616089,
      "learning_rate": 4.960421070081645e-06,
      "loss": 0.8596,
      "step": 192
    },
    {
      "epoch": 0.007998673795018443,
      "grad_norm": 0.45002996921539307,
      "learning_rate": 4.9602138505532764e-06,
      "loss": 0.8745,
      "step": 193
    },
    {
      "epoch": 0.008040117700692113,
      "grad_norm": 0.48973751068115234,
      "learning_rate": 4.960006631024909e-06,
      "loss": 0.8765,
      "step": 194
    },
    {
      "epoch": 0.008081561606365783,
      "grad_norm": 0.49519243836402893,
      "learning_rate": 4.95979941149654e-06,
      "loss": 0.8516,
      "step": 195
    },
    {
      "epoch": 0.008123005512039455,
      "grad_norm": 0.49807009100914,
      "learning_rate": 4.9595921919681714e-06,
      "loss": 0.8057,
      "step": 196
    },
    {
      "epoch": 0.008164449417713125,
      "grad_norm": 0.5044920444488525,
      "learning_rate": 4.959384972439803e-06,
      "loss": 0.8525,
      "step": 197
    },
    {
      "epoch": 0.008205893323386795,
      "grad_norm": 0.5080502033233643,
      "learning_rate": 4.959177752911435e-06,
      "loss": 0.8638,
      "step": 198
    },
    {
      "epoch": 0.008247337229060467,
      "grad_norm": 0.5046712160110474,
      "learning_rate": 4.9589705333830664e-06,
      "loss": 0.9036,
      "step": 199
    },
    {
      "epoch": 0.008288781134734137,
      "grad_norm": 0.4891860783100128,
      "learning_rate": 4.958763313854698e-06,
      "loss": 0.894,
      "step": 200
    },
    {
      "epoch": 0.008330225040407807,
      "grad_norm": 0.5205092430114746,
      "learning_rate": 4.958556094326329e-06,
      "loss": 0.812,
      "step": 201
    },
    {
      "epoch": 0.00837166894608148,
      "grad_norm": 0.4664292633533478,
      "learning_rate": 4.9583488747979614e-06,
      "loss": 0.8679,
      "step": 202
    },
    {
      "epoch": 0.00841311285175515,
      "grad_norm": 0.48823976516723633,
      "learning_rate": 4.958141655269593e-06,
      "loss": 0.8948,
      "step": 203
    },
    {
      "epoch": 0.00845455675742882,
      "grad_norm": 0.4761791527271271,
      "learning_rate": 4.957934435741224e-06,
      "loss": 0.8499,
      "step": 204
    },
    {
      "epoch": 0.008496000663102491,
      "grad_norm": 0.4740946888923645,
      "learning_rate": 4.9577272162128564e-06,
      "loss": 0.8782,
      "step": 205
    },
    {
      "epoch": 0.008537444568776162,
      "grad_norm": 0.49799829721450806,
      "learning_rate": 4.957519996684488e-06,
      "loss": 0.8657,
      "step": 206
    },
    {
      "epoch": 0.008578888474449832,
      "grad_norm": 0.44548892974853516,
      "learning_rate": 4.95731277715612e-06,
      "loss": 0.8411,
      "step": 207
    },
    {
      "epoch": 0.008620332380123504,
      "grad_norm": 0.48993632197380066,
      "learning_rate": 4.9571055576277514e-06,
      "loss": 0.8479,
      "step": 208
    },
    {
      "epoch": 0.008661776285797174,
      "grad_norm": 0.5193896889686584,
      "learning_rate": 4.956898338099383e-06,
      "loss": 0.8127,
      "step": 209
    },
    {
      "epoch": 0.008703220191470844,
      "grad_norm": 0.4925740957260132,
      "learning_rate": 4.956691118571015e-06,
      "loss": 0.8674,
      "step": 210
    },
    {
      "epoch": 0.008744664097144516,
      "grad_norm": 0.4636530578136444,
      "learning_rate": 4.9564838990426465e-06,
      "loss": 0.8237,
      "step": 211
    },
    {
      "epoch": 0.008786108002818186,
      "grad_norm": 0.4979214668273926,
      "learning_rate": 4.956276679514278e-06,
      "loss": 0.7773,
      "step": 212
    },
    {
      "epoch": 0.008827551908491856,
      "grad_norm": 0.5074848532676697,
      "learning_rate": 4.956069459985909e-06,
      "loss": 0.7693,
      "step": 213
    },
    {
      "epoch": 0.008868995814165528,
      "grad_norm": 0.5405175685882568,
      "learning_rate": 4.9558622404575415e-06,
      "loss": 0.8757,
      "step": 214
    },
    {
      "epoch": 0.008910439719839198,
      "grad_norm": 0.4779643416404724,
      "learning_rate": 4.955655020929173e-06,
      "loss": 0.8401,
      "step": 215
    },
    {
      "epoch": 0.008951883625512868,
      "grad_norm": 0.474722683429718,
      "learning_rate": 4.955447801400804e-06,
      "loss": 0.7988,
      "step": 216
    },
    {
      "epoch": 0.00899332753118654,
      "grad_norm": 0.4955325722694397,
      "learning_rate": 4.955240581872436e-06,
      "loss": 0.8997,
      "step": 217
    },
    {
      "epoch": 0.00903477143686021,
      "grad_norm": 0.5195847749710083,
      "learning_rate": 4.955033362344068e-06,
      "loss": 0.8855,
      "step": 218
    },
    {
      "epoch": 0.00907621534253388,
      "grad_norm": 0.4938107132911682,
      "learning_rate": 4.954826142815699e-06,
      "loss": 0.8835,
      "step": 219
    },
    {
      "epoch": 0.009117659248207552,
      "grad_norm": 0.5264998078346252,
      "learning_rate": 4.954618923287331e-06,
      "loss": 0.9331,
      "step": 220
    },
    {
      "epoch": 0.009159103153881222,
      "grad_norm": 0.5167268514633179,
      "learning_rate": 4.954411703758963e-06,
      "loss": 0.9111,
      "step": 221
    },
    {
      "epoch": 0.009200547059554892,
      "grad_norm": 0.4937087893486023,
      "learning_rate": 4.954204484230594e-06,
      "loss": 0.7659,
      "step": 222
    },
    {
      "epoch": 0.009241990965228562,
      "grad_norm": 0.5105080604553223,
      "learning_rate": 4.9539972647022265e-06,
      "loss": 0.8772,
      "step": 223
    },
    {
      "epoch": 0.009283434870902234,
      "grad_norm": 0.6784617900848389,
      "learning_rate": 4.953790045173857e-06,
      "loss": 0.9214,
      "step": 224
    },
    {
      "epoch": 0.009324878776575904,
      "grad_norm": 0.4829583168029785,
      "learning_rate": 4.953582825645489e-06,
      "loss": 0.9202,
      "step": 225
    },
    {
      "epoch": 0.009366322682249574,
      "grad_norm": 0.4833115339279175,
      "learning_rate": 4.953375606117121e-06,
      "loss": 0.8379,
      "step": 226
    },
    {
      "epoch": 0.009407766587923246,
      "grad_norm": 0.5154611468315125,
      "learning_rate": 4.953168386588753e-06,
      "loss": 0.9465,
      "step": 227
    },
    {
      "epoch": 0.009449210493596916,
      "grad_norm": 0.5287801623344421,
      "learning_rate": 4.952961167060384e-06,
      "loss": 0.8738,
      "step": 228
    },
    {
      "epoch": 0.009490654399270586,
      "grad_norm": 0.4534788131713867,
      "learning_rate": 4.952753947532016e-06,
      "loss": 0.8386,
      "step": 229
    },
    {
      "epoch": 0.009532098304944258,
      "grad_norm": 0.4852525293827057,
      "learning_rate": 4.952546728003648e-06,
      "loss": 0.8203,
      "step": 230
    },
    {
      "epoch": 0.009573542210617928,
      "grad_norm": 0.4988188147544861,
      "learning_rate": 4.952339508475279e-06,
      "loss": 0.8201,
      "step": 231
    },
    {
      "epoch": 0.009614986116291599,
      "grad_norm": 0.5310726165771484,
      "learning_rate": 4.952132288946911e-06,
      "loss": 0.8464,
      "step": 232
    },
    {
      "epoch": 0.00965643002196527,
      "grad_norm": 0.48887568712234497,
      "learning_rate": 4.951925069418542e-06,
      "loss": 0.8406,
      "step": 233
    },
    {
      "epoch": 0.00969787392763894,
      "grad_norm": 0.483868807554245,
      "learning_rate": 4.951717849890174e-06,
      "loss": 0.8789,
      "step": 234
    },
    {
      "epoch": 0.00973931783331261,
      "grad_norm": 0.4487422704696655,
      "learning_rate": 4.951510630361806e-06,
      "loss": 0.8269,
      "step": 235
    },
    {
      "epoch": 0.009780761738986283,
      "grad_norm": 0.49090155959129333,
      "learning_rate": 4.951303410833437e-06,
      "loss": 0.8433,
      "step": 236
    },
    {
      "epoch": 0.009822205644659953,
      "grad_norm": 0.4861697852611542,
      "learning_rate": 4.951096191305069e-06,
      "loss": 0.8247,
      "step": 237
    },
    {
      "epoch": 0.009863649550333623,
      "grad_norm": 0.4558032751083374,
      "learning_rate": 4.950888971776701e-06,
      "loss": 0.8132,
      "step": 238
    },
    {
      "epoch": 0.009905093456007295,
      "grad_norm": 0.44360578060150146,
      "learning_rate": 4.950681752248333e-06,
      "loss": 0.8379,
      "step": 239
    },
    {
      "epoch": 0.009946537361680965,
      "grad_norm": 0.5036263465881348,
      "learning_rate": 4.950474532719963e-06,
      "loss": 0.8787,
      "step": 240
    },
    {
      "epoch": 0.009987981267354635,
      "grad_norm": 0.5028334259986877,
      "learning_rate": 4.950267313191596e-06,
      "loss": 0.8867,
      "step": 241
    },
    {
      "epoch": 0.010029425173028307,
      "grad_norm": 0.4768572449684143,
      "learning_rate": 4.950060093663227e-06,
      "loss": 0.9009,
      "step": 242
    },
    {
      "epoch": 0.010070869078701977,
      "grad_norm": 0.4231841266155243,
      "learning_rate": 4.949852874134859e-06,
      "loss": 0.8049,
      "step": 243
    },
    {
      "epoch": 0.010112312984375647,
      "grad_norm": 0.4906524419784546,
      "learning_rate": 4.949645654606491e-06,
      "loss": 0.8005,
      "step": 244
    },
    {
      "epoch": 0.010153756890049319,
      "grad_norm": 0.5032555460929871,
      "learning_rate": 4.949438435078122e-06,
      "loss": 0.8647,
      "step": 245
    },
    {
      "epoch": 0.010195200795722989,
      "grad_norm": 0.49362102150917053,
      "learning_rate": 4.949231215549754e-06,
      "loss": 0.8459,
      "step": 246
    },
    {
      "epoch": 0.010236644701396659,
      "grad_norm": 0.48964157700538635,
      "learning_rate": 4.949023996021386e-06,
      "loss": 0.8621,
      "step": 247
    },
    {
      "epoch": 0.010278088607070331,
      "grad_norm": 0.5083989500999451,
      "learning_rate": 4.948816776493017e-06,
      "loss": 0.8601,
      "step": 248
    },
    {
      "epoch": 0.010319532512744001,
      "grad_norm": 0.5186017155647278,
      "learning_rate": 4.9486095569646484e-06,
      "loss": 0.8828,
      "step": 249
    },
    {
      "epoch": 0.010360976418417671,
      "grad_norm": 0.4820512533187866,
      "learning_rate": 4.948402337436281e-06,
      "loss": 0.8096,
      "step": 250
    },
    {
      "epoch": 0.010402420324091343,
      "grad_norm": 0.5429985523223877,
      "learning_rate": 4.948195117907912e-06,
      "loss": 0.9194,
      "step": 251
    },
    {
      "epoch": 0.010443864229765013,
      "grad_norm": 0.46637988090515137,
      "learning_rate": 4.9479878983795434e-06,
      "loss": 0.7842,
      "step": 252
    },
    {
      "epoch": 0.010485308135438683,
      "grad_norm": 0.5143275260925293,
      "learning_rate": 4.947780678851175e-06,
      "loss": 0.8152,
      "step": 253
    },
    {
      "epoch": 0.010526752041112355,
      "grad_norm": 0.4902860224246979,
      "learning_rate": 4.947573459322807e-06,
      "loss": 0.7815,
      "step": 254
    },
    {
      "epoch": 0.010568195946786025,
      "grad_norm": 0.49528950452804565,
      "learning_rate": 4.947366239794439e-06,
      "loss": 0.8345,
      "step": 255
    },
    {
      "epoch": 0.010609639852459695,
      "grad_norm": 0.4657568633556366,
      "learning_rate": 4.94715902026607e-06,
      "loss": 0.9146,
      "step": 256
    },
    {
      "epoch": 0.010651083758133367,
      "grad_norm": 0.49593663215637207,
      "learning_rate": 4.946951800737702e-06,
      "loss": 0.8413,
      "step": 257
    },
    {
      "epoch": 0.010692527663807037,
      "grad_norm": 0.47337037324905396,
      "learning_rate": 4.9467445812093334e-06,
      "loss": 0.7865,
      "step": 258
    },
    {
      "epoch": 0.010733971569480708,
      "grad_norm": 0.5408326387405396,
      "learning_rate": 4.946537361680966e-06,
      "loss": 0.8765,
      "step": 259
    },
    {
      "epoch": 0.01077541547515438,
      "grad_norm": 0.5662903785705566,
      "learning_rate": 4.946330142152597e-06,
      "loss": 0.9335,
      "step": 260
    },
    {
      "epoch": 0.01081685938082805,
      "grad_norm": 0.4936153292655945,
      "learning_rate": 4.9461229226242284e-06,
      "loss": 0.8359,
      "step": 261
    },
    {
      "epoch": 0.01085830328650172,
      "grad_norm": 0.49883532524108887,
      "learning_rate": 4.94591570309586e-06,
      "loss": 0.8175,
      "step": 262
    },
    {
      "epoch": 0.01089974719217539,
      "grad_norm": 0.49196162819862366,
      "learning_rate": 4.945708483567492e-06,
      "loss": 0.8367,
      "step": 263
    },
    {
      "epoch": 0.010941191097849062,
      "grad_norm": 0.49908918142318726,
      "learning_rate": 4.9455012640391234e-06,
      "loss": 0.8755,
      "step": 264
    },
    {
      "epoch": 0.010982635003522732,
      "grad_norm": 0.47234293818473816,
      "learning_rate": 4.945294044510755e-06,
      "loss": 0.8337,
      "step": 265
    },
    {
      "epoch": 0.011024078909196402,
      "grad_norm": 0.486908882856369,
      "learning_rate": 4.945086824982387e-06,
      "loss": 0.8013,
      "step": 266
    },
    {
      "epoch": 0.011065522814870074,
      "grad_norm": 0.5201804041862488,
      "learning_rate": 4.9448796054540184e-06,
      "loss": 0.7981,
      "step": 267
    },
    {
      "epoch": 0.011106966720543744,
      "grad_norm": 0.5242292284965515,
      "learning_rate": 4.94467238592565e-06,
      "loss": 0.8794,
      "step": 268
    },
    {
      "epoch": 0.011148410626217414,
      "grad_norm": 0.4720354676246643,
      "learning_rate": 4.944465166397281e-06,
      "loss": 0.8882,
      "step": 269
    },
    {
      "epoch": 0.011189854531891086,
      "grad_norm": 0.512622594833374,
      "learning_rate": 4.9442579468689135e-06,
      "loss": 0.8042,
      "step": 270
    },
    {
      "epoch": 0.011231298437564756,
      "grad_norm": 0.5217183828353882,
      "learning_rate": 4.944050727340545e-06,
      "loss": 0.854,
      "step": 271
    },
    {
      "epoch": 0.011272742343238426,
      "grad_norm": 0.4697608947753906,
      "learning_rate": 4.943843507812176e-06,
      "loss": 0.8735,
      "step": 272
    },
    {
      "epoch": 0.011314186248912098,
      "grad_norm": 0.4464910626411438,
      "learning_rate": 4.9436362882838085e-06,
      "loss": 0.8435,
      "step": 273
    },
    {
      "epoch": 0.011355630154585768,
      "grad_norm": 0.4901544749736786,
      "learning_rate": 4.94342906875544e-06,
      "loss": 0.8521,
      "step": 274
    },
    {
      "epoch": 0.011397074060259438,
      "grad_norm": 0.4696155786514282,
      "learning_rate": 4.943221849227072e-06,
      "loss": 0.8298,
      "step": 275
    },
    {
      "epoch": 0.01143851796593311,
      "grad_norm": 0.45748764276504517,
      "learning_rate": 4.9430146296987035e-06,
      "loss": 0.7368,
      "step": 276
    },
    {
      "epoch": 0.01147996187160678,
      "grad_norm": 0.48581135272979736,
      "learning_rate": 4.942807410170335e-06,
      "loss": 0.8708,
      "step": 277
    },
    {
      "epoch": 0.01152140577728045,
      "grad_norm": 0.473838746547699,
      "learning_rate": 4.942600190641966e-06,
      "loss": 0.8337,
      "step": 278
    },
    {
      "epoch": 0.011562849682954122,
      "grad_norm": 0.48033320903778076,
      "learning_rate": 4.9423929711135985e-06,
      "loss": 0.9041,
      "step": 279
    },
    {
      "epoch": 0.011604293588627792,
      "grad_norm": 0.4913444519042969,
      "learning_rate": 4.94218575158523e-06,
      "loss": 0.8762,
      "step": 280
    },
    {
      "epoch": 0.011645737494301462,
      "grad_norm": 0.584656298160553,
      "learning_rate": 4.941978532056861e-06,
      "loss": 0.9578,
      "step": 281
    },
    {
      "epoch": 0.011687181399975134,
      "grad_norm": 0.5384389162063599,
      "learning_rate": 4.9417713125284935e-06,
      "loss": 0.853,
      "step": 282
    },
    {
      "epoch": 0.011728625305648804,
      "grad_norm": 0.4899871349334717,
      "learning_rate": 4.941564093000125e-06,
      "loss": 0.8254,
      "step": 283
    },
    {
      "epoch": 0.011770069211322474,
      "grad_norm": 0.551979124546051,
      "learning_rate": 4.941356873471756e-06,
      "loss": 0.8796,
      "step": 284
    },
    {
      "epoch": 0.011811513116996146,
      "grad_norm": 0.5795146822929382,
      "learning_rate": 4.941149653943388e-06,
      "loss": 0.9009,
      "step": 285
    },
    {
      "epoch": 0.011852957022669816,
      "grad_norm": 0.47303688526153564,
      "learning_rate": 4.94094243441502e-06,
      "loss": 0.7842,
      "step": 286
    },
    {
      "epoch": 0.011894400928343487,
      "grad_norm": 0.4773743450641632,
      "learning_rate": 4.940735214886651e-06,
      "loss": 0.7971,
      "step": 287
    },
    {
      "epoch": 0.011935844834017158,
      "grad_norm": 0.5141909718513489,
      "learning_rate": 4.940527995358283e-06,
      "loss": 0.8618,
      "step": 288
    },
    {
      "epoch": 0.011977288739690829,
      "grad_norm": 0.4858902096748352,
      "learning_rate": 4.940320775829915e-06,
      "loss": 0.8523,
      "step": 289
    },
    {
      "epoch": 0.012018732645364499,
      "grad_norm": 0.4706079661846161,
      "learning_rate": 4.940113556301546e-06,
      "loss": 0.8262,
      "step": 290
    },
    {
      "epoch": 0.01206017655103817,
      "grad_norm": 0.5480523705482483,
      "learning_rate": 4.9399063367731785e-06,
      "loss": 0.8411,
      "step": 291
    },
    {
      "epoch": 0.01210162045671184,
      "grad_norm": 0.48437756299972534,
      "learning_rate": 4.939699117244809e-06,
      "loss": 0.8318,
      "step": 292
    },
    {
      "epoch": 0.01214306436238551,
      "grad_norm": 0.47752273082733154,
      "learning_rate": 4.939491897716441e-06,
      "loss": 0.8462,
      "step": 293
    },
    {
      "epoch": 0.012184508268059183,
      "grad_norm": 0.4477703869342804,
      "learning_rate": 4.939284678188073e-06,
      "loss": 0.7792,
      "step": 294
    },
    {
      "epoch": 0.012225952173732853,
      "grad_norm": 0.48229581117630005,
      "learning_rate": 4.939077458659705e-06,
      "loss": 0.8062,
      "step": 295
    },
    {
      "epoch": 0.012267396079406523,
      "grad_norm": 0.4598805606365204,
      "learning_rate": 4.938870239131336e-06,
      "loss": 0.8247,
      "step": 296
    },
    {
      "epoch": 0.012308839985080195,
      "grad_norm": 0.48081985116004944,
      "learning_rate": 4.938663019602968e-06,
      "loss": 0.8345,
      "step": 297
    },
    {
      "epoch": 0.012350283890753865,
      "grad_norm": 0.5329732298851013,
      "learning_rate": 4.9384558000746e-06,
      "loss": 0.8604,
      "step": 298
    },
    {
      "epoch": 0.012391727796427535,
      "grad_norm": 0.4876191020011902,
      "learning_rate": 4.938248580546231e-06,
      "loss": 0.8289,
      "step": 299
    },
    {
      "epoch": 0.012433171702101207,
      "grad_norm": 0.46362125873565674,
      "learning_rate": 4.938041361017863e-06,
      "loss": 0.8381,
      "step": 300
    },
    {
      "epoch": 0.012474615607774877,
      "grad_norm": 0.5167970061302185,
      "learning_rate": 4.937834141489494e-06,
      "loss": 0.8306,
      "step": 301
    },
    {
      "epoch": 0.012516059513448547,
      "grad_norm": 0.5373725891113281,
      "learning_rate": 4.937626921961126e-06,
      "loss": 0.8604,
      "step": 302
    },
    {
      "epoch": 0.012557503419122219,
      "grad_norm": 0.5176645517349243,
      "learning_rate": 4.937419702432758e-06,
      "loss": 0.8745,
      "step": 303
    },
    {
      "epoch": 0.012598947324795889,
      "grad_norm": 0.5150823593139648,
      "learning_rate": 4.937212482904389e-06,
      "loss": 0.8682,
      "step": 304
    },
    {
      "epoch": 0.01264039123046956,
      "grad_norm": 0.504682719707489,
      "learning_rate": 4.93700526337602e-06,
      "loss": 0.8591,
      "step": 305
    },
    {
      "epoch": 0.01268183513614323,
      "grad_norm": 0.49569401144981384,
      "learning_rate": 4.936798043847653e-06,
      "loss": 0.8188,
      "step": 306
    },
    {
      "epoch": 0.012723279041816901,
      "grad_norm": 0.4833688735961914,
      "learning_rate": 4.936590824319285e-06,
      "loss": 0.8728,
      "step": 307
    },
    {
      "epoch": 0.012764722947490571,
      "grad_norm": 0.4999264180660248,
      "learning_rate": 4.9363836047909154e-06,
      "loss": 0.9021,
      "step": 308
    },
    {
      "epoch": 0.012806166853164241,
      "grad_norm": 0.4826984703540802,
      "learning_rate": 4.936176385262548e-06,
      "loss": 0.8523,
      "step": 309
    },
    {
      "epoch": 0.012847610758837913,
      "grad_norm": 0.5844562649726868,
      "learning_rate": 4.935969165734179e-06,
      "loss": 0.8499,
      "step": 310
    },
    {
      "epoch": 0.012889054664511583,
      "grad_norm": 0.5145385265350342,
      "learning_rate": 4.935761946205811e-06,
      "loss": 0.8762,
      "step": 311
    },
    {
      "epoch": 0.012930498570185254,
      "grad_norm": 0.48962658643722534,
      "learning_rate": 4.935554726677443e-06,
      "loss": 0.8506,
      "step": 312
    },
    {
      "epoch": 0.012971942475858925,
      "grad_norm": 0.49469470977783203,
      "learning_rate": 4.935347507149074e-06,
      "loss": 0.8962,
      "step": 313
    },
    {
      "epoch": 0.013013386381532595,
      "grad_norm": 0.4592374265193939,
      "learning_rate": 4.9351402876207054e-06,
      "loss": 0.8123,
      "step": 314
    },
    {
      "epoch": 0.013054830287206266,
      "grad_norm": 0.47087806463241577,
      "learning_rate": 4.934933068092338e-06,
      "loss": 0.8069,
      "step": 315
    },
    {
      "epoch": 0.013096274192879937,
      "grad_norm": 0.49014976620674133,
      "learning_rate": 4.934725848563969e-06,
      "loss": 0.8447,
      "step": 316
    },
    {
      "epoch": 0.013137718098553608,
      "grad_norm": 0.511128842830658,
      "learning_rate": 4.9345186290356004e-06,
      "loss": 0.8772,
      "step": 317
    },
    {
      "epoch": 0.013179162004227278,
      "grad_norm": 0.5179100632667542,
      "learning_rate": 4.934311409507233e-06,
      "loss": 0.8306,
      "step": 318
    },
    {
      "epoch": 0.01322060590990095,
      "grad_norm": 0.48943230509757996,
      "learning_rate": 4.934104189978864e-06,
      "loss": 0.8708,
      "step": 319
    },
    {
      "epoch": 0.01326204981557462,
      "grad_norm": 0.4891606569290161,
      "learning_rate": 4.9338969704504954e-06,
      "loss": 0.7983,
      "step": 320
    },
    {
      "epoch": 0.01330349372124829,
      "grad_norm": 0.5088717341423035,
      "learning_rate": 4.933689750922127e-06,
      "loss": 0.8728,
      "step": 321
    },
    {
      "epoch": 0.013344937626921962,
      "grad_norm": 0.48573896288871765,
      "learning_rate": 4.933482531393759e-06,
      "loss": 0.8342,
      "step": 322
    },
    {
      "epoch": 0.013386381532595632,
      "grad_norm": 0.47867557406425476,
      "learning_rate": 4.9332753118653904e-06,
      "loss": 0.8232,
      "step": 323
    },
    {
      "epoch": 0.013427825438269302,
      "grad_norm": 0.4582088589668274,
      "learning_rate": 4.933068092337022e-06,
      "loss": 0.8135,
      "step": 324
    },
    {
      "epoch": 0.013469269343942974,
      "grad_norm": 0.46959298849105835,
      "learning_rate": 4.932860872808654e-06,
      "loss": 0.8037,
      "step": 325
    },
    {
      "epoch": 0.013510713249616644,
      "grad_norm": 0.47610822319984436,
      "learning_rate": 4.9326536532802854e-06,
      "loss": 0.8269,
      "step": 326
    },
    {
      "epoch": 0.013552157155290314,
      "grad_norm": 0.5315168499946594,
      "learning_rate": 4.932446433751918e-06,
      "loss": 0.96,
      "step": 327
    },
    {
      "epoch": 0.013593601060963986,
      "grad_norm": 0.4996245801448822,
      "learning_rate": 4.932239214223549e-06,
      "loss": 0.8848,
      "step": 328
    },
    {
      "epoch": 0.013635044966637656,
      "grad_norm": 0.4850527048110962,
      "learning_rate": 4.9320319946951805e-06,
      "loss": 0.8176,
      "step": 329
    },
    {
      "epoch": 0.013676488872311326,
      "grad_norm": 0.5390617847442627,
      "learning_rate": 4.931824775166812e-06,
      "loss": 0.8579,
      "step": 330
    },
    {
      "epoch": 0.013717932777984998,
      "grad_norm": 0.4859105348587036,
      "learning_rate": 4.931617555638444e-06,
      "loss": 0.7898,
      "step": 331
    },
    {
      "epoch": 0.013759376683658668,
      "grad_norm": 0.4938487708568573,
      "learning_rate": 4.9314103361100755e-06,
      "loss": 0.8411,
      "step": 332
    },
    {
      "epoch": 0.013800820589332338,
      "grad_norm": 0.4921363890171051,
      "learning_rate": 4.931203116581707e-06,
      "loss": 0.8577,
      "step": 333
    },
    {
      "epoch": 0.01384226449500601,
      "grad_norm": 0.44038423895835876,
      "learning_rate": 4.930995897053339e-06,
      "loss": 0.798,
      "step": 334
    },
    {
      "epoch": 0.01388370840067968,
      "grad_norm": 0.5288088917732239,
      "learning_rate": 4.9307886775249705e-06,
      "loss": 0.8899,
      "step": 335
    },
    {
      "epoch": 0.01392515230635335,
      "grad_norm": 0.4595298171043396,
      "learning_rate": 4.930581457996602e-06,
      "loss": 0.8213,
      "step": 336
    },
    {
      "epoch": 0.013966596212027022,
      "grad_norm": 0.47665566205978394,
      "learning_rate": 4.930374238468233e-06,
      "loss": 0.7621,
      "step": 337
    },
    {
      "epoch": 0.014008040117700692,
      "grad_norm": 0.4706921875476837,
      "learning_rate": 4.9301670189398655e-06,
      "loss": 0.7566,
      "step": 338
    },
    {
      "epoch": 0.014049484023374362,
      "grad_norm": 0.4992421567440033,
      "learning_rate": 4.929959799411497e-06,
      "loss": 0.7972,
      "step": 339
    },
    {
      "epoch": 0.014090927929048034,
      "grad_norm": 0.4747614562511444,
      "learning_rate": 4.929752579883128e-06,
      "loss": 0.811,
      "step": 340
    },
    {
      "epoch": 0.014132371834721704,
      "grad_norm": 0.4470059871673584,
      "learning_rate": 4.9295453603547605e-06,
      "loss": 0.7942,
      "step": 341
    },
    {
      "epoch": 0.014173815740395375,
      "grad_norm": 0.4530155658721924,
      "learning_rate": 4.929338140826392e-06,
      "loss": 0.8584,
      "step": 342
    },
    {
      "epoch": 0.014215259646069046,
      "grad_norm": 0.4985761344432831,
      "learning_rate": 4.929130921298024e-06,
      "loss": 0.8533,
      "step": 343
    },
    {
      "epoch": 0.014256703551742717,
      "grad_norm": 0.4647839367389679,
      "learning_rate": 4.9289237017696555e-06,
      "loss": 0.8103,
      "step": 344
    },
    {
      "epoch": 0.014298147457416387,
      "grad_norm": 0.5143153071403503,
      "learning_rate": 4.928716482241287e-06,
      "loss": 0.8511,
      "step": 345
    },
    {
      "epoch": 0.014339591363090057,
      "grad_norm": 0.45094966888427734,
      "learning_rate": 4.928509262712918e-06,
      "loss": 0.843,
      "step": 346
    },
    {
      "epoch": 0.014381035268763729,
      "grad_norm": 0.49729955196380615,
      "learning_rate": 4.9283020431845505e-06,
      "loss": 0.8755,
      "step": 347
    },
    {
      "epoch": 0.014422479174437399,
      "grad_norm": 0.6007020473480225,
      "learning_rate": 4.928094823656182e-06,
      "loss": 0.8201,
      "step": 348
    },
    {
      "epoch": 0.014463923080111069,
      "grad_norm": 0.4756455421447754,
      "learning_rate": 4.927887604127813e-06,
      "loss": 0.8464,
      "step": 349
    },
    {
      "epoch": 0.01450536698578474,
      "grad_norm": 0.4875801205635071,
      "learning_rate": 4.9276803845994455e-06,
      "loss": 0.8767,
      "step": 350
    },
    {
      "epoch": 0.01454681089145841,
      "grad_norm": 0.5090059638023376,
      "learning_rate": 4.927473165071077e-06,
      "loss": 0.8137,
      "step": 351
    },
    {
      "epoch": 0.014588254797132081,
      "grad_norm": 0.4570600688457489,
      "learning_rate": 4.927265945542708e-06,
      "loss": 0.8076,
      "step": 352
    },
    {
      "epoch": 0.014629698702805753,
      "grad_norm": 0.5911429524421692,
      "learning_rate": 4.92705872601434e-06,
      "loss": 0.9153,
      "step": 353
    },
    {
      "epoch": 0.014671142608479423,
      "grad_norm": 0.46505096554756165,
      "learning_rate": 4.926851506485972e-06,
      "loss": 0.76,
      "step": 354
    },
    {
      "epoch": 0.014712586514153093,
      "grad_norm": 0.5193902850151062,
      "learning_rate": 4.926644286957603e-06,
      "loss": 0.8336,
      "step": 355
    },
    {
      "epoch": 0.014754030419826765,
      "grad_norm": 0.5347269773483276,
      "learning_rate": 4.926437067429235e-06,
      "loss": 0.9011,
      "step": 356
    },
    {
      "epoch": 0.014795474325500435,
      "grad_norm": 0.5115196108818054,
      "learning_rate": 4.926229847900866e-06,
      "loss": 0.9023,
      "step": 357
    },
    {
      "epoch": 0.014836918231174105,
      "grad_norm": 0.4776272177696228,
      "learning_rate": 4.926022628372498e-06,
      "loss": 0.8584,
      "step": 358
    },
    {
      "epoch": 0.014878362136847777,
      "grad_norm": 0.48909175395965576,
      "learning_rate": 4.9258154088441305e-06,
      "loss": 0.8142,
      "step": 359
    },
    {
      "epoch": 0.014919806042521447,
      "grad_norm": 0.47103431820869446,
      "learning_rate": 4.925608189315762e-06,
      "loss": 0.798,
      "step": 360
    },
    {
      "epoch": 0.014961249948195117,
      "grad_norm": 0.4640336036682129,
      "learning_rate": 4.925400969787393e-06,
      "loss": 0.7783,
      "step": 361
    },
    {
      "epoch": 0.01500269385386879,
      "grad_norm": 0.45157426595687866,
      "learning_rate": 4.925193750259025e-06,
      "loss": 0.8083,
      "step": 362
    },
    {
      "epoch": 0.01504413775954246,
      "grad_norm": 0.48852115869522095,
      "learning_rate": 4.924986530730657e-06,
      "loss": 0.8718,
      "step": 363
    },
    {
      "epoch": 0.01508558166521613,
      "grad_norm": 0.5745199918746948,
      "learning_rate": 4.924779311202288e-06,
      "loss": 0.9255,
      "step": 364
    },
    {
      "epoch": 0.015127025570889801,
      "grad_norm": 0.47590112686157227,
      "learning_rate": 4.92457209167392e-06,
      "loss": 0.7534,
      "step": 365
    },
    {
      "epoch": 0.015168469476563471,
      "grad_norm": 0.5060416460037231,
      "learning_rate": 4.924364872145551e-06,
      "loss": 0.8748,
      "step": 366
    },
    {
      "epoch": 0.015209913382237141,
      "grad_norm": 0.47386276721954346,
      "learning_rate": 4.924157652617183e-06,
      "loss": 0.9031,
      "step": 367
    },
    {
      "epoch": 0.015251357287910813,
      "grad_norm": 0.5125477910041809,
      "learning_rate": 4.923950433088815e-06,
      "loss": 0.844,
      "step": 368
    },
    {
      "epoch": 0.015292801193584483,
      "grad_norm": 0.4907439351081848,
      "learning_rate": 4.923743213560446e-06,
      "loss": 0.856,
      "step": 369
    },
    {
      "epoch": 0.015334245099258154,
      "grad_norm": 0.4889748990535736,
      "learning_rate": 4.923535994032078e-06,
      "loss": 0.9004,
      "step": 370
    },
    {
      "epoch": 0.015375689004931825,
      "grad_norm": 0.479399710893631,
      "learning_rate": 4.92332877450371e-06,
      "loss": 0.7988,
      "step": 371
    },
    {
      "epoch": 0.015417132910605496,
      "grad_norm": 0.5161091685295105,
      "learning_rate": 4.923121554975341e-06,
      "loss": 0.8877,
      "step": 372
    },
    {
      "epoch": 0.015458576816279166,
      "grad_norm": 0.4821130931377411,
      "learning_rate": 4.9229143354469724e-06,
      "loss": 0.8564,
      "step": 373
    },
    {
      "epoch": 0.015500020721952838,
      "grad_norm": 0.47413793206214905,
      "learning_rate": 4.922707115918605e-06,
      "loss": 0.8455,
      "step": 374
    },
    {
      "epoch": 0.015541464627626508,
      "grad_norm": 0.46731749176979065,
      "learning_rate": 4.922499896390236e-06,
      "loss": 0.854,
      "step": 375
    },
    {
      "epoch": 0.015582908533300178,
      "grad_norm": 0.5367522239685059,
      "learning_rate": 4.9222926768618674e-06,
      "loss": 0.8652,
      "step": 376
    },
    {
      "epoch": 0.01562435243897385,
      "grad_norm": 0.49957993626594543,
      "learning_rate": 4.9220854573335e-06,
      "loss": 0.8335,
      "step": 377
    },
    {
      "epoch": 0.015665796344647518,
      "grad_norm": 0.5024069547653198,
      "learning_rate": 4.921878237805131e-06,
      "loss": 0.7959,
      "step": 378
    },
    {
      "epoch": 0.01570724025032119,
      "grad_norm": 0.46874380111694336,
      "learning_rate": 4.921671018276763e-06,
      "loss": 0.8059,
      "step": 379
    },
    {
      "epoch": 0.015748684155994862,
      "grad_norm": 0.4658451974391937,
      "learning_rate": 4.921463798748395e-06,
      "loss": 0.8418,
      "step": 380
    },
    {
      "epoch": 0.015790128061668532,
      "grad_norm": 0.5046740770339966,
      "learning_rate": 4.921256579220026e-06,
      "loss": 0.8706,
      "step": 381
    },
    {
      "epoch": 0.015831571967342202,
      "grad_norm": 0.4836690425872803,
      "learning_rate": 4.9210493596916574e-06,
      "loss": 0.8196,
      "step": 382
    },
    {
      "epoch": 0.015873015873015872,
      "grad_norm": 0.48251408338546753,
      "learning_rate": 4.92084214016329e-06,
      "loss": 0.8528,
      "step": 383
    },
    {
      "epoch": 0.015914459778689542,
      "grad_norm": 0.4738592803478241,
      "learning_rate": 4.920634920634921e-06,
      "loss": 0.8,
      "step": 384
    },
    {
      "epoch": 0.015955903684363216,
      "grad_norm": 0.5402851700782776,
      "learning_rate": 4.9204277011065524e-06,
      "loss": 0.8914,
      "step": 385
    },
    {
      "epoch": 0.015997347590036886,
      "grad_norm": 0.49203255772590637,
      "learning_rate": 4.920220481578185e-06,
      "loss": 0.8574,
      "step": 386
    },
    {
      "epoch": 0.016038791495710556,
      "grad_norm": 0.4661460816860199,
      "learning_rate": 4.920013262049816e-06,
      "loss": 0.8533,
      "step": 387
    },
    {
      "epoch": 0.016080235401384226,
      "grad_norm": 0.48006075620651245,
      "learning_rate": 4.9198060425214475e-06,
      "loss": 0.803,
      "step": 388
    },
    {
      "epoch": 0.016121679307057896,
      "grad_norm": 0.4801277220249176,
      "learning_rate": 4.919598822993079e-06,
      "loss": 0.8674,
      "step": 389
    },
    {
      "epoch": 0.016163123212731566,
      "grad_norm": 0.44423675537109375,
      "learning_rate": 4.919391603464711e-06,
      "loss": 0.7825,
      "step": 390
    },
    {
      "epoch": 0.01620456711840524,
      "grad_norm": 0.4668760895729065,
      "learning_rate": 4.9191843839363425e-06,
      "loss": 0.8147,
      "step": 391
    },
    {
      "epoch": 0.01624601102407891,
      "grad_norm": 0.5075100660324097,
      "learning_rate": 4.918977164407974e-06,
      "loss": 0.906,
      "step": 392
    },
    {
      "epoch": 0.01628745492975258,
      "grad_norm": 0.4797791540622711,
      "learning_rate": 4.918769944879606e-06,
      "loss": 0.8662,
      "step": 393
    },
    {
      "epoch": 0.01632889883542625,
      "grad_norm": 0.49664506316185,
      "learning_rate": 4.9185627253512375e-06,
      "loss": 0.7861,
      "step": 394
    },
    {
      "epoch": 0.01637034274109992,
      "grad_norm": 0.4621793329715729,
      "learning_rate": 4.91835550582287e-06,
      "loss": 0.8477,
      "step": 395
    },
    {
      "epoch": 0.01641178664677359,
      "grad_norm": 0.5162895917892456,
      "learning_rate": 4.918148286294501e-06,
      "loss": 0.8728,
      "step": 396
    },
    {
      "epoch": 0.016453230552447264,
      "grad_norm": 0.5187440514564514,
      "learning_rate": 4.9179410667661325e-06,
      "loss": 0.7927,
      "step": 397
    },
    {
      "epoch": 0.016494674458120934,
      "grad_norm": 0.48792213201522827,
      "learning_rate": 4.917733847237764e-06,
      "loss": 0.8474,
      "step": 398
    },
    {
      "epoch": 0.016536118363794605,
      "grad_norm": 0.4553443193435669,
      "learning_rate": 4.917526627709396e-06,
      "loss": 0.7991,
      "step": 399
    },
    {
      "epoch": 0.016577562269468275,
      "grad_norm": 0.4740474224090576,
      "learning_rate": 4.9173194081810275e-06,
      "loss": 0.8228,
      "step": 400
    },
    {
      "epoch": 0.016619006175141945,
      "grad_norm": 0.49421992897987366,
      "learning_rate": 4.917112188652659e-06,
      "loss": 0.8513,
      "step": 401
    },
    {
      "epoch": 0.016660450080815615,
      "grad_norm": 0.5225406885147095,
      "learning_rate": 4.916904969124291e-06,
      "loss": 0.9307,
      "step": 402
    },
    {
      "epoch": 0.01670189398648929,
      "grad_norm": 0.453609824180603,
      "learning_rate": 4.9166977495959225e-06,
      "loss": 0.7539,
      "step": 403
    },
    {
      "epoch": 0.01674333789216296,
      "grad_norm": 0.4644213914871216,
      "learning_rate": 4.916490530067554e-06,
      "loss": 0.8186,
      "step": 404
    },
    {
      "epoch": 0.01678478179783663,
      "grad_norm": 0.5403565168380737,
      "learning_rate": 4.916283310539185e-06,
      "loss": 0.842,
      "step": 405
    },
    {
      "epoch": 0.0168262257035103,
      "grad_norm": 0.4868334233760834,
      "learning_rate": 4.9160760910108175e-06,
      "loss": 0.7981,
      "step": 406
    },
    {
      "epoch": 0.01686766960918397,
      "grad_norm": 0.47065234184265137,
      "learning_rate": 4.915868871482449e-06,
      "loss": 0.8123,
      "step": 407
    },
    {
      "epoch": 0.01690911351485764,
      "grad_norm": 0.5169198513031006,
      "learning_rate": 4.91566165195408e-06,
      "loss": 0.8466,
      "step": 408
    },
    {
      "epoch": 0.01695055742053131,
      "grad_norm": 0.44264426827430725,
      "learning_rate": 4.915454432425712e-06,
      "loss": 0.7815,
      "step": 409
    },
    {
      "epoch": 0.016992001326204983,
      "grad_norm": 0.46051862835884094,
      "learning_rate": 4.915247212897344e-06,
      "loss": 0.8281,
      "step": 410
    },
    {
      "epoch": 0.017033445231878653,
      "grad_norm": 0.48056861758232117,
      "learning_rate": 4.915039993368976e-06,
      "loss": 0.7898,
      "step": 411
    },
    {
      "epoch": 0.017074889137552323,
      "grad_norm": 0.5130049586296082,
      "learning_rate": 4.9148327738406075e-06,
      "loss": 0.8306,
      "step": 412
    },
    {
      "epoch": 0.017116333043225993,
      "grad_norm": 0.45076465606689453,
      "learning_rate": 4.914625554312239e-06,
      "loss": 0.7935,
      "step": 413
    },
    {
      "epoch": 0.017157776948899663,
      "grad_norm": 0.46089985966682434,
      "learning_rate": 4.91441833478387e-06,
      "loss": 0.8059,
      "step": 414
    },
    {
      "epoch": 0.017199220854573333,
      "grad_norm": 0.45488303899765015,
      "learning_rate": 4.9142111152555025e-06,
      "loss": 0.7965,
      "step": 415
    },
    {
      "epoch": 0.017240664760247007,
      "grad_norm": 0.4720882773399353,
      "learning_rate": 4.914003895727134e-06,
      "loss": 0.8225,
      "step": 416
    },
    {
      "epoch": 0.017282108665920677,
      "grad_norm": 0.46068084239959717,
      "learning_rate": 4.913796676198765e-06,
      "loss": 0.804,
      "step": 417
    },
    {
      "epoch": 0.017323552571594347,
      "grad_norm": 0.5240599513053894,
      "learning_rate": 4.913589456670397e-06,
      "loss": 0.8711,
      "step": 418
    },
    {
      "epoch": 0.017364996477268017,
      "grad_norm": 0.6012427806854248,
      "learning_rate": 4.913382237142029e-06,
      "loss": 0.8191,
      "step": 419
    },
    {
      "epoch": 0.017406440382941687,
      "grad_norm": 0.5937158465385437,
      "learning_rate": 4.91317501761366e-06,
      "loss": 0.8796,
      "step": 420
    },
    {
      "epoch": 0.017447884288615358,
      "grad_norm": 0.48416364192962646,
      "learning_rate": 4.912967798085292e-06,
      "loss": 0.8274,
      "step": 421
    },
    {
      "epoch": 0.01748932819428903,
      "grad_norm": 0.44899284839630127,
      "learning_rate": 4.912760578556924e-06,
      "loss": 0.8484,
      "step": 422
    },
    {
      "epoch": 0.0175307720999627,
      "grad_norm": 0.4800012707710266,
      "learning_rate": 4.912553359028555e-06,
      "loss": 0.7979,
      "step": 423
    },
    {
      "epoch": 0.01757221600563637,
      "grad_norm": 0.4837320148944855,
      "learning_rate": 4.912346139500187e-06,
      "loss": 0.8186,
      "step": 424
    },
    {
      "epoch": 0.01761365991131004,
      "grad_norm": 0.4740530848503113,
      "learning_rate": 4.912138919971818e-06,
      "loss": 0.8291,
      "step": 425
    },
    {
      "epoch": 0.01765510381698371,
      "grad_norm": 0.4518044590950012,
      "learning_rate": 4.91193170044345e-06,
      "loss": 0.8049,
      "step": 426
    },
    {
      "epoch": 0.017696547722657382,
      "grad_norm": 0.4841671586036682,
      "learning_rate": 4.911724480915082e-06,
      "loss": 0.8516,
      "step": 427
    },
    {
      "epoch": 0.017737991628331055,
      "grad_norm": 0.4785279631614685,
      "learning_rate": 4.911517261386714e-06,
      "loss": 0.8169,
      "step": 428
    },
    {
      "epoch": 0.017779435534004726,
      "grad_norm": 0.4598132371902466,
      "learning_rate": 4.911310041858345e-06,
      "loss": 0.8069,
      "step": 429
    },
    {
      "epoch": 0.017820879439678396,
      "grad_norm": 0.5076266527175903,
      "learning_rate": 4.911102822329977e-06,
      "loss": 0.8569,
      "step": 430
    },
    {
      "epoch": 0.017862323345352066,
      "grad_norm": 0.45577993988990784,
      "learning_rate": 4.910895602801609e-06,
      "loss": 0.7512,
      "step": 431
    },
    {
      "epoch": 0.017903767251025736,
      "grad_norm": 0.5087863206863403,
      "learning_rate": 4.91068838327324e-06,
      "loss": 0.7761,
      "step": 432
    },
    {
      "epoch": 0.017945211156699406,
      "grad_norm": 0.4486319422721863,
      "learning_rate": 4.910481163744872e-06,
      "loss": 0.8982,
      "step": 433
    },
    {
      "epoch": 0.01798665506237308,
      "grad_norm": 0.5014278292655945,
      "learning_rate": 4.910273944216503e-06,
      "loss": 0.8162,
      "step": 434
    },
    {
      "epoch": 0.01802809896804675,
      "grad_norm": 0.46839311718940735,
      "learning_rate": 4.910066724688135e-06,
      "loss": 0.7803,
      "step": 435
    },
    {
      "epoch": 0.01806954287372042,
      "grad_norm": 0.49171018600463867,
      "learning_rate": 4.909859505159767e-06,
      "loss": 0.8779,
      "step": 436
    },
    {
      "epoch": 0.01811098677939409,
      "grad_norm": 0.4889352023601532,
      "learning_rate": 4.909652285631398e-06,
      "loss": 0.8862,
      "step": 437
    },
    {
      "epoch": 0.01815243068506776,
      "grad_norm": 0.4969484806060791,
      "learning_rate": 4.90944506610303e-06,
      "loss": 0.8142,
      "step": 438
    },
    {
      "epoch": 0.01819387459074143,
      "grad_norm": 0.45526519417762756,
      "learning_rate": 4.909237846574662e-06,
      "loss": 0.8193,
      "step": 439
    },
    {
      "epoch": 0.018235318496415104,
      "grad_norm": 0.4844783544540405,
      "learning_rate": 4.909030627046293e-06,
      "loss": 0.8396,
      "step": 440
    },
    {
      "epoch": 0.018276762402088774,
      "grad_norm": 0.5045813918113708,
      "learning_rate": 4.9088234075179244e-06,
      "loss": 0.8481,
      "step": 441
    },
    {
      "epoch": 0.018318206307762444,
      "grad_norm": 0.547040581703186,
      "learning_rate": 4.908616187989557e-06,
      "loss": 0.8477,
      "step": 442
    },
    {
      "epoch": 0.018359650213436114,
      "grad_norm": 0.4526495039463043,
      "learning_rate": 4.908408968461188e-06,
      "loss": 0.8064,
      "step": 443
    },
    {
      "epoch": 0.018401094119109784,
      "grad_norm": 0.5089373588562012,
      "learning_rate": 4.9082017489328195e-06,
      "loss": 0.843,
      "step": 444
    },
    {
      "epoch": 0.018442538024783454,
      "grad_norm": 0.4830186367034912,
      "learning_rate": 4.907994529404451e-06,
      "loss": 0.8269,
      "step": 445
    },
    {
      "epoch": 0.018483981930457125,
      "grad_norm": 0.539989173412323,
      "learning_rate": 4.907787309876083e-06,
      "loss": 0.8621,
      "step": 446
    },
    {
      "epoch": 0.018525425836130798,
      "grad_norm": 0.4397867023944855,
      "learning_rate": 4.907580090347715e-06,
      "loss": 0.7822,
      "step": 447
    },
    {
      "epoch": 0.01856686974180447,
      "grad_norm": 0.48654913902282715,
      "learning_rate": 4.907372870819347e-06,
      "loss": 0.8457,
      "step": 448
    },
    {
      "epoch": 0.01860831364747814,
      "grad_norm": 0.4765969514846802,
      "learning_rate": 4.907165651290978e-06,
      "loss": 0.7732,
      "step": 449
    },
    {
      "epoch": 0.01864975755315181,
      "grad_norm": 0.5010995268821716,
      "learning_rate": 4.9069584317626095e-06,
      "loss": 0.7888,
      "step": 450
    },
    {
      "epoch": 0.01869120145882548,
      "grad_norm": 0.45635828375816345,
      "learning_rate": 4.906751212234242e-06,
      "loss": 0.7874,
      "step": 451
    },
    {
      "epoch": 0.01873264536449915,
      "grad_norm": 0.5300329923629761,
      "learning_rate": 4.906543992705873e-06,
      "loss": 0.8276,
      "step": 452
    },
    {
      "epoch": 0.018774089270172822,
      "grad_norm": 0.4671211540699005,
      "learning_rate": 4.9063367731775045e-06,
      "loss": 0.8667,
      "step": 453
    },
    {
      "epoch": 0.018815533175846492,
      "grad_norm": 0.46540865302085876,
      "learning_rate": 4.906129553649137e-06,
      "loss": 0.8335,
      "step": 454
    },
    {
      "epoch": 0.018856977081520163,
      "grad_norm": 0.49816012382507324,
      "learning_rate": 4.905922334120768e-06,
      "loss": 0.8743,
      "step": 455
    },
    {
      "epoch": 0.018898420987193833,
      "grad_norm": 0.5244741439819336,
      "learning_rate": 4.9057151145923995e-06,
      "loss": 0.8552,
      "step": 456
    },
    {
      "epoch": 0.018939864892867503,
      "grad_norm": 0.4913051724433899,
      "learning_rate": 4.905507895064031e-06,
      "loss": 0.8618,
      "step": 457
    },
    {
      "epoch": 0.018981308798541173,
      "grad_norm": 0.5005473494529724,
      "learning_rate": 4.905300675535663e-06,
      "loss": 0.8523,
      "step": 458
    },
    {
      "epoch": 0.019022752704214847,
      "grad_norm": 0.4599267244338989,
      "learning_rate": 4.9050934560072945e-06,
      "loss": 0.7976,
      "step": 459
    },
    {
      "epoch": 0.019064196609888517,
      "grad_norm": 0.4852423369884491,
      "learning_rate": 4.904886236478926e-06,
      "loss": 0.8057,
      "step": 460
    },
    {
      "epoch": 0.019105640515562187,
      "grad_norm": 0.451443076133728,
      "learning_rate": 4.904679016950557e-06,
      "loss": 0.801,
      "step": 461
    },
    {
      "epoch": 0.019147084421235857,
      "grad_norm": 0.4907212555408478,
      "learning_rate": 4.9044717974221895e-06,
      "loss": 0.8125,
      "step": 462
    },
    {
      "epoch": 0.019188528326909527,
      "grad_norm": 0.46715471148490906,
      "learning_rate": 4.904264577893822e-06,
      "loss": 0.8167,
      "step": 463
    },
    {
      "epoch": 0.019229972232583197,
      "grad_norm": 0.47199976444244385,
      "learning_rate": 4.904057358365453e-06,
      "loss": 0.8157,
      "step": 464
    },
    {
      "epoch": 0.01927141613825687,
      "grad_norm": 0.46977463364601135,
      "learning_rate": 4.9038501388370845e-06,
      "loss": 0.8757,
      "step": 465
    },
    {
      "epoch": 0.01931286004393054,
      "grad_norm": 0.44438350200653076,
      "learning_rate": 4.903642919308716e-06,
      "loss": 0.76,
      "step": 466
    },
    {
      "epoch": 0.01935430394960421,
      "grad_norm": 0.5002657771110535,
      "learning_rate": 4.903435699780348e-06,
      "loss": 0.7944,
      "step": 467
    },
    {
      "epoch": 0.01939574785527788,
      "grad_norm": 0.5109201669692993,
      "learning_rate": 4.9032284802519795e-06,
      "loss": 0.8711,
      "step": 468
    },
    {
      "epoch": 0.01943719176095155,
      "grad_norm": 0.5137448310852051,
      "learning_rate": 4.903021260723611e-06,
      "loss": 0.752,
      "step": 469
    },
    {
      "epoch": 0.01947863566662522,
      "grad_norm": 0.5363143086433411,
      "learning_rate": 4.902814041195242e-06,
      "loss": 0.8176,
      "step": 470
    },
    {
      "epoch": 0.019520079572298895,
      "grad_norm": 0.4738991856575012,
      "learning_rate": 4.9026068216668745e-06,
      "loss": 0.7893,
      "step": 471
    },
    {
      "epoch": 0.019561523477972565,
      "grad_norm": 0.4510260224342346,
      "learning_rate": 4.902399602138506e-06,
      "loss": 0.7937,
      "step": 472
    },
    {
      "epoch": 0.019602967383646235,
      "grad_norm": 0.4918208420276642,
      "learning_rate": 4.902192382610137e-06,
      "loss": 0.7979,
      "step": 473
    },
    {
      "epoch": 0.019644411289319905,
      "grad_norm": 0.48828259110450745,
      "learning_rate": 4.9019851630817695e-06,
      "loss": 0.8721,
      "step": 474
    },
    {
      "epoch": 0.019685855194993575,
      "grad_norm": 0.43061384558677673,
      "learning_rate": 4.901777943553401e-06,
      "loss": 0.8055,
      "step": 475
    },
    {
      "epoch": 0.019727299100667246,
      "grad_norm": 0.4545750617980957,
      "learning_rate": 4.901570724025032e-06,
      "loss": 0.7976,
      "step": 476
    },
    {
      "epoch": 0.01976874300634092,
      "grad_norm": 0.5229188799858093,
      "learning_rate": 4.901363504496664e-06,
      "loss": 0.8301,
      "step": 477
    },
    {
      "epoch": 0.01981018691201459,
      "grad_norm": 0.44408050179481506,
      "learning_rate": 4.901156284968296e-06,
      "loss": 0.78,
      "step": 478
    },
    {
      "epoch": 0.01985163081768826,
      "grad_norm": 0.5063157677650452,
      "learning_rate": 4.900949065439927e-06,
      "loss": 0.8621,
      "step": 479
    },
    {
      "epoch": 0.01989307472336193,
      "grad_norm": 0.4920934736728668,
      "learning_rate": 4.9007418459115595e-06,
      "loss": 0.7847,
      "step": 480
    },
    {
      "epoch": 0.0199345186290356,
      "grad_norm": 0.5848267674446106,
      "learning_rate": 4.900534626383191e-06,
      "loss": 0.8501,
      "step": 481
    },
    {
      "epoch": 0.01997596253470927,
      "grad_norm": 0.48913031816482544,
      "learning_rate": 4.900327406854822e-06,
      "loss": 0.8008,
      "step": 482
    },
    {
      "epoch": 0.020017406440382943,
      "grad_norm": 0.5036863088607788,
      "learning_rate": 4.9001201873264545e-06,
      "loss": 0.8018,
      "step": 483
    },
    {
      "epoch": 0.020058850346056614,
      "grad_norm": 0.46786680817604065,
      "learning_rate": 4.899912967798086e-06,
      "loss": 0.8413,
      "step": 484
    },
    {
      "epoch": 0.020100294251730284,
      "grad_norm": 0.47618743777275085,
      "learning_rate": 4.899705748269717e-06,
      "loss": 0.7832,
      "step": 485
    },
    {
      "epoch": 0.020141738157403954,
      "grad_norm": 0.4857902228832245,
      "learning_rate": 4.899498528741349e-06,
      "loss": 0.8784,
      "step": 486
    },
    {
      "epoch": 0.020183182063077624,
      "grad_norm": 0.5140090584754944,
      "learning_rate": 4.899291309212981e-06,
      "loss": 0.8889,
      "step": 487
    },
    {
      "epoch": 0.020224625968751294,
      "grad_norm": 0.4780973792076111,
      "learning_rate": 4.899084089684612e-06,
      "loss": 0.7751,
      "step": 488
    },
    {
      "epoch": 0.020266069874424964,
      "grad_norm": 0.46658629179000854,
      "learning_rate": 4.898876870156244e-06,
      "loss": 0.8184,
      "step": 489
    },
    {
      "epoch": 0.020307513780098638,
      "grad_norm": 0.47505757212638855,
      "learning_rate": 4.898669650627876e-06,
      "loss": 0.7725,
      "step": 490
    },
    {
      "epoch": 0.020348957685772308,
      "grad_norm": 0.4759451150894165,
      "learning_rate": 4.898462431099507e-06,
      "loss": 0.8167,
      "step": 491
    },
    {
      "epoch": 0.020390401591445978,
      "grad_norm": 0.4769792854785919,
      "learning_rate": 4.898255211571139e-06,
      "loss": 0.8152,
      "step": 492
    },
    {
      "epoch": 0.020431845497119648,
      "grad_norm": 0.5034412145614624,
      "learning_rate": 4.89804799204277e-06,
      "loss": 0.8057,
      "step": 493
    },
    {
      "epoch": 0.020473289402793318,
      "grad_norm": 0.44975370168685913,
      "learning_rate": 4.897840772514402e-06,
      "loss": 0.7925,
      "step": 494
    },
    {
      "epoch": 0.02051473330846699,
      "grad_norm": 0.4678061306476593,
      "learning_rate": 4.897633552986034e-06,
      "loss": 0.804,
      "step": 495
    },
    {
      "epoch": 0.020556177214140662,
      "grad_norm": 0.4949014484882355,
      "learning_rate": 4.897426333457666e-06,
      "loss": 0.8298,
      "step": 496
    },
    {
      "epoch": 0.020597621119814332,
      "grad_norm": 0.5196720957756042,
      "learning_rate": 4.8972191139292964e-06,
      "loss": 0.8376,
      "step": 497
    },
    {
      "epoch": 0.020639065025488002,
      "grad_norm": 0.5207451581954956,
      "learning_rate": 4.897011894400929e-06,
      "loss": 0.8726,
      "step": 498
    },
    {
      "epoch": 0.020680508931161672,
      "grad_norm": 0.46578139066696167,
      "learning_rate": 4.896804674872561e-06,
      "loss": 0.7998,
      "step": 499
    },
    {
      "epoch": 0.020721952836835342,
      "grad_norm": 0.5228472948074341,
      "learning_rate": 4.896597455344192e-06,
      "loss": 0.8379,
      "step": 500
    },
    {
      "epoch": 0.020763396742509013,
      "grad_norm": 0.4439935088157654,
      "learning_rate": 4.896390235815824e-06,
      "loss": 0.7885,
      "step": 501
    },
    {
      "epoch": 0.020804840648182686,
      "grad_norm": 0.46296626329421997,
      "learning_rate": 4.896183016287455e-06,
      "loss": 0.7659,
      "step": 502
    },
    {
      "epoch": 0.020846284553856356,
      "grad_norm": 0.49447980523109436,
      "learning_rate": 4.895975796759087e-06,
      "loss": 0.8208,
      "step": 503
    },
    {
      "epoch": 0.020887728459530026,
      "grad_norm": 0.45511969923973083,
      "learning_rate": 4.895768577230719e-06,
      "loss": 0.7983,
      "step": 504
    },
    {
      "epoch": 0.020929172365203697,
      "grad_norm": 0.4478197991847992,
      "learning_rate": 4.89556135770235e-06,
      "loss": 0.7834,
      "step": 505
    },
    {
      "epoch": 0.020970616270877367,
      "grad_norm": 0.4402836263179779,
      "learning_rate": 4.8953541381739815e-06,
      "loss": 0.7676,
      "step": 506
    },
    {
      "epoch": 0.021012060176551037,
      "grad_norm": 0.44297662377357483,
      "learning_rate": 4.895146918645614e-06,
      "loss": 0.7883,
      "step": 507
    },
    {
      "epoch": 0.02105350408222471,
      "grad_norm": 0.5047253966331482,
      "learning_rate": 4.894939699117245e-06,
      "loss": 0.9138,
      "step": 508
    },
    {
      "epoch": 0.02109494798789838,
      "grad_norm": 0.5168031454086304,
      "learning_rate": 4.8947324795888765e-06,
      "loss": 0.8809,
      "step": 509
    },
    {
      "epoch": 0.02113639189357205,
      "grad_norm": 0.5152113437652588,
      "learning_rate": 4.894525260060509e-06,
      "loss": 0.7847,
      "step": 510
    },
    {
      "epoch": 0.02117783579924572,
      "grad_norm": 0.46421441435813904,
      "learning_rate": 4.89431804053214e-06,
      "loss": 0.7681,
      "step": 511
    },
    {
      "epoch": 0.02121927970491939,
      "grad_norm": 0.46760526299476624,
      "learning_rate": 4.8941108210037715e-06,
      "loss": 0.79,
      "step": 512
    },
    {
      "epoch": 0.02126072361059306,
      "grad_norm": 0.470059871673584,
      "learning_rate": 4.893903601475403e-06,
      "loss": 0.8193,
      "step": 513
    },
    {
      "epoch": 0.021302167516266735,
      "grad_norm": 0.4791758358478546,
      "learning_rate": 4.893696381947035e-06,
      "loss": 0.8418,
      "step": 514
    },
    {
      "epoch": 0.021343611421940405,
      "grad_norm": 0.46390190720558167,
      "learning_rate": 4.893489162418667e-06,
      "loss": 0.8459,
      "step": 515
    },
    {
      "epoch": 0.021385055327614075,
      "grad_norm": 0.43868038058280945,
      "learning_rate": 4.893281942890299e-06,
      "loss": 0.7942,
      "step": 516
    },
    {
      "epoch": 0.021426499233287745,
      "grad_norm": 0.47130662202835083,
      "learning_rate": 4.89307472336193e-06,
      "loss": 0.854,
      "step": 517
    },
    {
      "epoch": 0.021467943138961415,
      "grad_norm": 0.4984039068222046,
      "learning_rate": 4.8928675038335615e-06,
      "loss": 0.8887,
      "step": 518
    },
    {
      "epoch": 0.021509387044635085,
      "grad_norm": 0.4585687220096588,
      "learning_rate": 4.892660284305194e-06,
      "loss": 0.8198,
      "step": 519
    },
    {
      "epoch": 0.02155083095030876,
      "grad_norm": 0.496217280626297,
      "learning_rate": 4.892453064776825e-06,
      "loss": 0.8025,
      "step": 520
    },
    {
      "epoch": 0.02159227485598243,
      "grad_norm": 0.47160789370536804,
      "learning_rate": 4.8922458452484565e-06,
      "loss": 0.7446,
      "step": 521
    },
    {
      "epoch": 0.0216337187616561,
      "grad_norm": 0.5480387210845947,
      "learning_rate": 4.892038625720088e-06,
      "loss": 0.9136,
      "step": 522
    },
    {
      "epoch": 0.02167516266732977,
      "grad_norm": 0.44932806491851807,
      "learning_rate": 4.89183140619172e-06,
      "loss": 0.7328,
      "step": 523
    },
    {
      "epoch": 0.02171660657300344,
      "grad_norm": 0.4601422846317291,
      "learning_rate": 4.8916241866633515e-06,
      "loss": 0.7666,
      "step": 524
    },
    {
      "epoch": 0.02175805047867711,
      "grad_norm": 0.47914057970046997,
      "learning_rate": 4.891416967134983e-06,
      "loss": 0.799,
      "step": 525
    },
    {
      "epoch": 0.02179949438435078,
      "grad_norm": 0.47169601917266846,
      "learning_rate": 4.891209747606615e-06,
      "loss": 0.8933,
      "step": 526
    },
    {
      "epoch": 0.021840938290024453,
      "grad_norm": 0.4357249438762665,
      "learning_rate": 4.8910025280782465e-06,
      "loss": 0.7849,
      "step": 527
    },
    {
      "epoch": 0.021882382195698123,
      "grad_norm": 0.5225855708122253,
      "learning_rate": 4.890795308549878e-06,
      "loss": 0.793,
      "step": 528
    },
    {
      "epoch": 0.021923826101371793,
      "grad_norm": 0.5255203247070312,
      "learning_rate": 4.890588089021509e-06,
      "loss": 0.9143,
      "step": 529
    },
    {
      "epoch": 0.021965270007045463,
      "grad_norm": 0.47456568479537964,
      "learning_rate": 4.8903808694931415e-06,
      "loss": 0.8313,
      "step": 530
    },
    {
      "epoch": 0.022006713912719134,
      "grad_norm": 0.48161983489990234,
      "learning_rate": 4.890173649964773e-06,
      "loss": 0.8484,
      "step": 531
    },
    {
      "epoch": 0.022048157818392804,
      "grad_norm": 0.4773552417755127,
      "learning_rate": 4.889966430436405e-06,
      "loss": 0.8198,
      "step": 532
    },
    {
      "epoch": 0.022089601724066477,
      "grad_norm": 0.43980589509010315,
      "learning_rate": 4.8897592109080365e-06,
      "loss": 0.8496,
      "step": 533
    },
    {
      "epoch": 0.022131045629740147,
      "grad_norm": 0.4770891070365906,
      "learning_rate": 4.889551991379668e-06,
      "loss": 0.7791,
      "step": 534
    },
    {
      "epoch": 0.022172489535413818,
      "grad_norm": 0.4900587499141693,
      "learning_rate": 4.8893447718513e-06,
      "loss": 0.7479,
      "step": 535
    },
    {
      "epoch": 0.022213933441087488,
      "grad_norm": 0.48809048533439636,
      "learning_rate": 4.8891375523229315e-06,
      "loss": 0.8226,
      "step": 536
    },
    {
      "epoch": 0.022255377346761158,
      "grad_norm": 0.48735225200653076,
      "learning_rate": 4.888930332794563e-06,
      "loss": 0.853,
      "step": 537
    },
    {
      "epoch": 0.022296821252434828,
      "grad_norm": 0.48354893922805786,
      "learning_rate": 4.888723113266194e-06,
      "loss": 0.8655,
      "step": 538
    },
    {
      "epoch": 0.0223382651581085,
      "grad_norm": 0.4572780728340149,
      "learning_rate": 4.8885158937378265e-06,
      "loss": 0.7361,
      "step": 539
    },
    {
      "epoch": 0.02237970906378217,
      "grad_norm": 0.505377471446991,
      "learning_rate": 4.888308674209458e-06,
      "loss": 0.8479,
      "step": 540
    },
    {
      "epoch": 0.022421152969455842,
      "grad_norm": 0.4808878004550934,
      "learning_rate": 4.888101454681089e-06,
      "loss": 0.7456,
      "step": 541
    },
    {
      "epoch": 0.022462596875129512,
      "grad_norm": 0.48792564868927,
      "learning_rate": 4.8878942351527215e-06,
      "loss": 0.834,
      "step": 542
    },
    {
      "epoch": 0.022504040780803182,
      "grad_norm": 0.4690813422203064,
      "learning_rate": 4.887687015624353e-06,
      "loss": 0.7578,
      "step": 543
    },
    {
      "epoch": 0.022545484686476852,
      "grad_norm": 0.5344796180725098,
      "learning_rate": 4.887479796095984e-06,
      "loss": 0.8955,
      "step": 544
    },
    {
      "epoch": 0.022586928592150526,
      "grad_norm": 0.47094985842704773,
      "learning_rate": 4.887272576567616e-06,
      "loss": 0.7598,
      "step": 545
    },
    {
      "epoch": 0.022628372497824196,
      "grad_norm": 0.5210912227630615,
      "learning_rate": 4.887065357039248e-06,
      "loss": 0.9221,
      "step": 546
    },
    {
      "epoch": 0.022669816403497866,
      "grad_norm": 0.4628584384918213,
      "learning_rate": 4.886858137510879e-06,
      "loss": 0.7937,
      "step": 547
    },
    {
      "epoch": 0.022711260309171536,
      "grad_norm": 0.48059824109077454,
      "learning_rate": 4.8866509179825115e-06,
      "loss": 0.7708,
      "step": 548
    },
    {
      "epoch": 0.022752704214845206,
      "grad_norm": 0.4434678852558136,
      "learning_rate": 4.886443698454142e-06,
      "loss": 0.7505,
      "step": 549
    },
    {
      "epoch": 0.022794148120518876,
      "grad_norm": 0.4747712016105652,
      "learning_rate": 4.886236478925774e-06,
      "loss": 0.8174,
      "step": 550
    },
    {
      "epoch": 0.02283559202619255,
      "grad_norm": 0.4849909842014313,
      "learning_rate": 4.8860292593974065e-06,
      "loss": 0.8782,
      "step": 551
    },
    {
      "epoch": 0.02287703593186622,
      "grad_norm": 0.5049998760223389,
      "learning_rate": 4.885822039869038e-06,
      "loss": 0.8596,
      "step": 552
    },
    {
      "epoch": 0.02291847983753989,
      "grad_norm": 0.5053225755691528,
      "learning_rate": 4.885614820340669e-06,
      "loss": 0.7964,
      "step": 553
    },
    {
      "epoch": 0.02295992374321356,
      "grad_norm": 0.43197908997535706,
      "learning_rate": 4.885407600812301e-06,
      "loss": 0.8164,
      "step": 554
    },
    {
      "epoch": 0.02300136764888723,
      "grad_norm": 0.47206753492355347,
      "learning_rate": 4.885200381283933e-06,
      "loss": 0.8247,
      "step": 555
    },
    {
      "epoch": 0.0230428115545609,
      "grad_norm": 0.453747034072876,
      "learning_rate": 4.884993161755564e-06,
      "loss": 0.7861,
      "step": 556
    },
    {
      "epoch": 0.023084255460234574,
      "grad_norm": 0.4709733724594116,
      "learning_rate": 4.884785942227196e-06,
      "loss": 0.7666,
      "step": 557
    },
    {
      "epoch": 0.023125699365908244,
      "grad_norm": 0.4618700444698334,
      "learning_rate": 4.884578722698827e-06,
      "loss": 0.823,
      "step": 558
    },
    {
      "epoch": 0.023167143271581914,
      "grad_norm": 0.4612927734851837,
      "learning_rate": 4.884371503170459e-06,
      "loss": 0.8152,
      "step": 559
    },
    {
      "epoch": 0.023208587177255584,
      "grad_norm": 0.4989731013774872,
      "learning_rate": 4.884164283642091e-06,
      "loss": 0.833,
      "step": 560
    },
    {
      "epoch": 0.023250031082929255,
      "grad_norm": 0.4959282875061035,
      "learning_rate": 4.883957064113722e-06,
      "loss": 0.8604,
      "step": 561
    },
    {
      "epoch": 0.023291474988602925,
      "grad_norm": 0.46370023488998413,
      "learning_rate": 4.883749844585354e-06,
      "loss": 0.8252,
      "step": 562
    },
    {
      "epoch": 0.0233329188942766,
      "grad_norm": 0.46025994420051575,
      "learning_rate": 4.883542625056986e-06,
      "loss": 0.7876,
      "step": 563
    },
    {
      "epoch": 0.02337436279995027,
      "grad_norm": 0.4388412833213806,
      "learning_rate": 4.883335405528618e-06,
      "loss": 0.8208,
      "step": 564
    },
    {
      "epoch": 0.02341580670562394,
      "grad_norm": 0.4504384696483612,
      "learning_rate": 4.8831281860002485e-06,
      "loss": 0.7512,
      "step": 565
    },
    {
      "epoch": 0.02345725061129761,
      "grad_norm": 0.4930025637149811,
      "learning_rate": 4.882920966471881e-06,
      "loss": 0.803,
      "step": 566
    },
    {
      "epoch": 0.02349869451697128,
      "grad_norm": 0.4779817461967468,
      "learning_rate": 4.882713746943513e-06,
      "loss": 0.8284,
      "step": 567
    },
    {
      "epoch": 0.02354013842264495,
      "grad_norm": 0.48746493458747864,
      "learning_rate": 4.882506527415144e-06,
      "loss": 0.8496,
      "step": 568
    },
    {
      "epoch": 0.02358158232831862,
      "grad_norm": 0.4496462643146515,
      "learning_rate": 4.882299307886776e-06,
      "loss": 0.8191,
      "step": 569
    },
    {
      "epoch": 0.023623026233992293,
      "grad_norm": 0.4515238106250763,
      "learning_rate": 4.882092088358407e-06,
      "loss": 0.7618,
      "step": 570
    },
    {
      "epoch": 0.023664470139665963,
      "grad_norm": 0.49928152561187744,
      "learning_rate": 4.881884868830039e-06,
      "loss": 0.8364,
      "step": 571
    },
    {
      "epoch": 0.023705914045339633,
      "grad_norm": 0.4698212742805481,
      "learning_rate": 4.881677649301671e-06,
      "loss": 0.7612,
      "step": 572
    },
    {
      "epoch": 0.023747357951013303,
      "grad_norm": 0.4687686860561371,
      "learning_rate": 4.881470429773302e-06,
      "loss": 0.7778,
      "step": 573
    },
    {
      "epoch": 0.023788801856686973,
      "grad_norm": 0.5051504969596863,
      "learning_rate": 4.8812632102449335e-06,
      "loss": 0.845,
      "step": 574
    },
    {
      "epoch": 0.023830245762360643,
      "grad_norm": 0.4640558958053589,
      "learning_rate": 4.881055990716566e-06,
      "loss": 0.8052,
      "step": 575
    },
    {
      "epoch": 0.023871689668034317,
      "grad_norm": 0.47961655259132385,
      "learning_rate": 4.880848771188197e-06,
      "loss": 0.7686,
      "step": 576
    },
    {
      "epoch": 0.023913133573707987,
      "grad_norm": 0.4662036597728729,
      "learning_rate": 4.8806415516598285e-06,
      "loss": 0.8304,
      "step": 577
    },
    {
      "epoch": 0.023954577479381657,
      "grad_norm": 0.4470343291759491,
      "learning_rate": 4.880434332131461e-06,
      "loss": 0.822,
      "step": 578
    },
    {
      "epoch": 0.023996021385055327,
      "grad_norm": 0.4809962213039398,
      "learning_rate": 4.880227112603092e-06,
      "loss": 0.8489,
      "step": 579
    },
    {
      "epoch": 0.024037465290728997,
      "grad_norm": 0.49782678484916687,
      "learning_rate": 4.880019893074724e-06,
      "loss": 0.864,
      "step": 580
    },
    {
      "epoch": 0.024078909196402667,
      "grad_norm": 0.4834621846675873,
      "learning_rate": 4.879812673546355e-06,
      "loss": 0.7886,
      "step": 581
    },
    {
      "epoch": 0.02412035310207634,
      "grad_norm": 0.4877339005470276,
      "learning_rate": 4.879605454017987e-06,
      "loss": 0.7717,
      "step": 582
    },
    {
      "epoch": 0.02416179700775001,
      "grad_norm": 0.45702728629112244,
      "learning_rate": 4.8793982344896185e-06,
      "loss": 0.8247,
      "step": 583
    },
    {
      "epoch": 0.02420324091342368,
      "grad_norm": 0.4746725261211395,
      "learning_rate": 4.879191014961251e-06,
      "loss": 0.7629,
      "step": 584
    },
    {
      "epoch": 0.02424468481909735,
      "grad_norm": 0.5049235224723816,
      "learning_rate": 4.878983795432882e-06,
      "loss": 0.8403,
      "step": 585
    },
    {
      "epoch": 0.02428612872477102,
      "grad_norm": 0.4915831685066223,
      "learning_rate": 4.8787765759045135e-06,
      "loss": 0.7683,
      "step": 586
    },
    {
      "epoch": 0.02432757263044469,
      "grad_norm": 0.48449084162712097,
      "learning_rate": 4.878569356376146e-06,
      "loss": 0.7861,
      "step": 587
    },
    {
      "epoch": 0.024369016536118365,
      "grad_norm": 0.5011218786239624,
      "learning_rate": 4.878362136847777e-06,
      "loss": 0.7993,
      "step": 588
    },
    {
      "epoch": 0.024410460441792035,
      "grad_norm": 0.47988468408584595,
      "learning_rate": 4.8781549173194085e-06,
      "loss": 0.7902,
      "step": 589
    },
    {
      "epoch": 0.024451904347465706,
      "grad_norm": 0.46121472120285034,
      "learning_rate": 4.87794769779104e-06,
      "loss": 0.8059,
      "step": 590
    },
    {
      "epoch": 0.024493348253139376,
      "grad_norm": 0.5387938022613525,
      "learning_rate": 4.877740478262672e-06,
      "loss": 0.8933,
      "step": 591
    },
    {
      "epoch": 0.024534792158813046,
      "grad_norm": 0.4928279519081116,
      "learning_rate": 4.8775332587343035e-06,
      "loss": 0.7759,
      "step": 592
    },
    {
      "epoch": 0.024576236064486716,
      "grad_norm": 0.4332732558250427,
      "learning_rate": 4.877326039205935e-06,
      "loss": 0.7856,
      "step": 593
    },
    {
      "epoch": 0.02461767997016039,
      "grad_norm": 0.5063583254814148,
      "learning_rate": 4.877118819677567e-06,
      "loss": 0.8286,
      "step": 594
    },
    {
      "epoch": 0.02465912387583406,
      "grad_norm": 0.5039634108543396,
      "learning_rate": 4.8769116001491985e-06,
      "loss": 0.8147,
      "step": 595
    },
    {
      "epoch": 0.02470056778150773,
      "grad_norm": 0.48804840445518494,
      "learning_rate": 4.87670438062083e-06,
      "loss": 0.7988,
      "step": 596
    },
    {
      "epoch": 0.0247420116871814,
      "grad_norm": 0.48544737696647644,
      "learning_rate": 4.876497161092461e-06,
      "loss": 0.8467,
      "step": 597
    },
    {
      "epoch": 0.02478345559285507,
      "grad_norm": 0.49868589639663696,
      "learning_rate": 4.8762899415640935e-06,
      "loss": 0.8496,
      "step": 598
    },
    {
      "epoch": 0.02482489949852874,
      "grad_norm": 0.4890144169330597,
      "learning_rate": 4.876082722035725e-06,
      "loss": 0.8284,
      "step": 599
    },
    {
      "epoch": 0.024866343404202414,
      "grad_norm": 0.46914008259773254,
      "learning_rate": 4.875875502507357e-06,
      "loss": 0.7963,
      "step": 600
    },
    {
      "epoch": 0.024907787309876084,
      "grad_norm": 0.501280665397644,
      "learning_rate": 4.8756682829789885e-06,
      "loss": 0.814,
      "step": 601
    },
    {
      "epoch": 0.024949231215549754,
      "grad_norm": 0.487753689289093,
      "learning_rate": 4.87546106345062e-06,
      "loss": 0.791,
      "step": 602
    },
    {
      "epoch": 0.024990675121223424,
      "grad_norm": 0.47674545645713806,
      "learning_rate": 4.875253843922252e-06,
      "loss": 0.7776,
      "step": 603
    },
    {
      "epoch": 0.025032119026897094,
      "grad_norm": 0.491993248462677,
      "learning_rate": 4.8750466243938835e-06,
      "loss": 0.8093,
      "step": 604
    },
    {
      "epoch": 0.025073562932570764,
      "grad_norm": 0.5122238397598267,
      "learning_rate": 4.874839404865515e-06,
      "loss": 0.8975,
      "step": 605
    },
    {
      "epoch": 0.025115006838244438,
      "grad_norm": 0.5160649418830872,
      "learning_rate": 4.874632185337146e-06,
      "loss": 0.8159,
      "step": 606
    },
    {
      "epoch": 0.025156450743918108,
      "grad_norm": 0.49168211221694946,
      "learning_rate": 4.8744249658087785e-06,
      "loss": 0.8467,
      "step": 607
    },
    {
      "epoch": 0.025197894649591778,
      "grad_norm": 0.44403907656669617,
      "learning_rate": 4.87421774628041e-06,
      "loss": 0.823,
      "step": 608
    },
    {
      "epoch": 0.025239338555265448,
      "grad_norm": 0.4288485646247864,
      "learning_rate": 4.874010526752041e-06,
      "loss": 0.782,
      "step": 609
    },
    {
      "epoch": 0.02528078246093912,
      "grad_norm": 0.47649186849594116,
      "learning_rate": 4.873803307223673e-06,
      "loss": 0.8376,
      "step": 610
    },
    {
      "epoch": 0.02532222636661279,
      "grad_norm": 0.46790993213653564,
      "learning_rate": 4.873596087695305e-06,
      "loss": 0.8308,
      "step": 611
    },
    {
      "epoch": 0.02536367027228646,
      "grad_norm": 0.4332045912742615,
      "learning_rate": 4.873388868166936e-06,
      "loss": 0.7708,
      "step": 612
    },
    {
      "epoch": 0.025405114177960132,
      "grad_norm": 0.4512092173099518,
      "learning_rate": 4.873181648638568e-06,
      "loss": 0.7969,
      "step": 613
    },
    {
      "epoch": 0.025446558083633802,
      "grad_norm": 0.49965643882751465,
      "learning_rate": 4.8729744291102e-06,
      "loss": 0.8408,
      "step": 614
    },
    {
      "epoch": 0.025488001989307472,
      "grad_norm": 0.47372087836265564,
      "learning_rate": 4.872767209581831e-06,
      "loss": 0.8291,
      "step": 615
    },
    {
      "epoch": 0.025529445894981143,
      "grad_norm": 0.4483170807361603,
      "learning_rate": 4.8725599900534635e-06,
      "loss": 0.7595,
      "step": 616
    },
    {
      "epoch": 0.025570889800654813,
      "grad_norm": 0.46808502078056335,
      "learning_rate": 4.872352770525094e-06,
      "loss": 0.8196,
      "step": 617
    },
    {
      "epoch": 0.025612333706328483,
      "grad_norm": 0.44011443853378296,
      "learning_rate": 4.872145550996726e-06,
      "loss": 0.7788,
      "step": 618
    },
    {
      "epoch": 0.025653777612002156,
      "grad_norm": 0.5084178447723389,
      "learning_rate": 4.871938331468358e-06,
      "loss": 0.8479,
      "step": 619
    },
    {
      "epoch": 0.025695221517675827,
      "grad_norm": 0.4346061646938324,
      "learning_rate": 4.87173111193999e-06,
      "loss": 0.783,
      "step": 620
    },
    {
      "epoch": 0.025736665423349497,
      "grad_norm": 0.44072747230529785,
      "learning_rate": 4.871523892411621e-06,
      "loss": 0.7644,
      "step": 621
    },
    {
      "epoch": 0.025778109329023167,
      "grad_norm": 0.45584553480148315,
      "learning_rate": 4.871316672883253e-06,
      "loss": 0.801,
      "step": 622
    },
    {
      "epoch": 0.025819553234696837,
      "grad_norm": 0.5316863059997559,
      "learning_rate": 4.871109453354885e-06,
      "loss": 0.8276,
      "step": 623
    },
    {
      "epoch": 0.025860997140370507,
      "grad_norm": 0.45720532536506653,
      "learning_rate": 4.870902233826516e-06,
      "loss": 0.8145,
      "step": 624
    },
    {
      "epoch": 0.02590244104604418,
      "grad_norm": 0.4098816215991974,
      "learning_rate": 4.870695014298148e-06,
      "loss": 0.7986,
      "step": 625
    },
    {
      "epoch": 0.02594388495171785,
      "grad_norm": 0.4784835875034332,
      "learning_rate": 4.870487794769779e-06,
      "loss": 0.7578,
      "step": 626
    },
    {
      "epoch": 0.02598532885739152,
      "grad_norm": 0.45547032356262207,
      "learning_rate": 4.870280575241411e-06,
      "loss": 0.7825,
      "step": 627
    },
    {
      "epoch": 0.02602677276306519,
      "grad_norm": 0.4823001027107239,
      "learning_rate": 4.870073355713043e-06,
      "loss": 0.7881,
      "step": 628
    },
    {
      "epoch": 0.02606821666873886,
      "grad_norm": 0.415988951921463,
      "learning_rate": 4.869866136184674e-06,
      "loss": 0.7856,
      "step": 629
    },
    {
      "epoch": 0.02610966057441253,
      "grad_norm": 0.4871703088283539,
      "learning_rate": 4.869658916656306e-06,
      "loss": 0.7993,
      "step": 630
    },
    {
      "epoch": 0.026151104480086205,
      "grad_norm": 0.4808284640312195,
      "learning_rate": 4.869451697127938e-06,
      "loss": 0.7407,
      "step": 631
    },
    {
      "epoch": 0.026192548385759875,
      "grad_norm": 0.458878755569458,
      "learning_rate": 4.86924447759957e-06,
      "loss": 0.7998,
      "step": 632
    },
    {
      "epoch": 0.026233992291433545,
      "grad_norm": 0.5022748112678528,
      "learning_rate": 4.8690372580712005e-06,
      "loss": 0.792,
      "step": 633
    },
    {
      "epoch": 0.026275436197107215,
      "grad_norm": 0.5150101780891418,
      "learning_rate": 4.868830038542833e-06,
      "loss": 0.8372,
      "step": 634
    },
    {
      "epoch": 0.026316880102780885,
      "grad_norm": 0.4635588228702545,
      "learning_rate": 4.868622819014464e-06,
      "loss": 0.7617,
      "step": 635
    },
    {
      "epoch": 0.026358324008454555,
      "grad_norm": 0.5002948045730591,
      "learning_rate": 4.868415599486096e-06,
      "loss": 0.8218,
      "step": 636
    },
    {
      "epoch": 0.02639976791412823,
      "grad_norm": 0.5043407678604126,
      "learning_rate": 4.868208379957728e-06,
      "loss": 0.8483,
      "step": 637
    },
    {
      "epoch": 0.0264412118198019,
      "grad_norm": 0.4270593523979187,
      "learning_rate": 4.868001160429359e-06,
      "loss": 0.7446,
      "step": 638
    },
    {
      "epoch": 0.02648265572547557,
      "grad_norm": 0.5148329734802246,
      "learning_rate": 4.867793940900991e-06,
      "loss": 0.7773,
      "step": 639
    },
    {
      "epoch": 0.02652409963114924,
      "grad_norm": 0.514662504196167,
      "learning_rate": 4.867586721372623e-06,
      "loss": 0.8442,
      "step": 640
    },
    {
      "epoch": 0.02656554353682291,
      "grad_norm": 0.48593640327453613,
      "learning_rate": 4.867379501844254e-06,
      "loss": 0.8217,
      "step": 641
    },
    {
      "epoch": 0.02660698744249658,
      "grad_norm": 0.49880021810531616,
      "learning_rate": 4.8671722823158855e-06,
      "loss": 0.8069,
      "step": 642
    },
    {
      "epoch": 0.026648431348170253,
      "grad_norm": 0.4725196957588196,
      "learning_rate": 4.866965062787518e-06,
      "loss": 0.8081,
      "step": 643
    },
    {
      "epoch": 0.026689875253843923,
      "grad_norm": 0.4438782036304474,
      "learning_rate": 4.866757843259149e-06,
      "loss": 0.7932,
      "step": 644
    },
    {
      "epoch": 0.026731319159517593,
      "grad_norm": 0.47148048877716064,
      "learning_rate": 4.8665506237307805e-06,
      "loss": 0.8704,
      "step": 645
    },
    {
      "epoch": 0.026772763065191264,
      "grad_norm": 0.4726231098175049,
      "learning_rate": 4.866343404202413e-06,
      "loss": 0.79,
      "step": 646
    },
    {
      "epoch": 0.026814206970864934,
      "grad_norm": 0.4968494772911072,
      "learning_rate": 4.866136184674044e-06,
      "loss": 0.8047,
      "step": 647
    },
    {
      "epoch": 0.026855650876538604,
      "grad_norm": 0.43174630403518677,
      "learning_rate": 4.865928965145676e-06,
      "loss": 0.7822,
      "step": 648
    },
    {
      "epoch": 0.026897094782212274,
      "grad_norm": 0.43715959787368774,
      "learning_rate": 4.865721745617307e-06,
      "loss": 0.7703,
      "step": 649
    },
    {
      "epoch": 0.026938538687885948,
      "grad_norm": 0.4591558873653412,
      "learning_rate": 4.865514526088939e-06,
      "loss": 0.7905,
      "step": 650
    },
    {
      "epoch": 0.026979982593559618,
      "grad_norm": 0.5094489455223083,
      "learning_rate": 4.8653073065605705e-06,
      "loss": 0.8013,
      "step": 651
    },
    {
      "epoch": 0.027021426499233288,
      "grad_norm": 0.49252450466156006,
      "learning_rate": 4.865100087032203e-06,
      "loss": 0.9199,
      "step": 652
    },
    {
      "epoch": 0.027062870404906958,
      "grad_norm": 0.4657875597476959,
      "learning_rate": 4.864892867503834e-06,
      "loss": 0.7932,
      "step": 653
    },
    {
      "epoch": 0.027104314310580628,
      "grad_norm": 0.501944363117218,
      "learning_rate": 4.8646856479754655e-06,
      "loss": 0.7993,
      "step": 654
    },
    {
      "epoch": 0.027145758216254298,
      "grad_norm": 0.48011940717697144,
      "learning_rate": 4.864478428447098e-06,
      "loss": 0.7837,
      "step": 655
    },
    {
      "epoch": 0.027187202121927972,
      "grad_norm": 0.5366302728652954,
      "learning_rate": 4.864271208918729e-06,
      "loss": 0.8665,
      "step": 656
    },
    {
      "epoch": 0.027228646027601642,
      "grad_norm": 0.4874062240123749,
      "learning_rate": 4.8640639893903605e-06,
      "loss": 0.781,
      "step": 657
    },
    {
      "epoch": 0.027270089933275312,
      "grad_norm": 0.4771879315376282,
      "learning_rate": 4.863856769861992e-06,
      "loss": 0.7959,
      "step": 658
    },
    {
      "epoch": 0.027311533838948982,
      "grad_norm": 0.46157220005989075,
      "learning_rate": 4.863649550333624e-06,
      "loss": 0.8113,
      "step": 659
    },
    {
      "epoch": 0.027352977744622652,
      "grad_norm": 0.4611596465110779,
      "learning_rate": 4.8634423308052555e-06,
      "loss": 0.7826,
      "step": 660
    },
    {
      "epoch": 0.027394421650296322,
      "grad_norm": 0.46256789565086365,
      "learning_rate": 4.863235111276887e-06,
      "loss": 0.833,
      "step": 661
    },
    {
      "epoch": 0.027435865555969996,
      "grad_norm": 0.4736841320991516,
      "learning_rate": 4.863027891748518e-06,
      "loss": 0.7576,
      "step": 662
    },
    {
      "epoch": 0.027477309461643666,
      "grad_norm": 0.47134703397750854,
      "learning_rate": 4.8628206722201505e-06,
      "loss": 0.7837,
      "step": 663
    },
    {
      "epoch": 0.027518753367317336,
      "grad_norm": 0.46176448464393616,
      "learning_rate": 4.862613452691782e-06,
      "loss": 0.7808,
      "step": 664
    },
    {
      "epoch": 0.027560197272991006,
      "grad_norm": 0.4663459062576294,
      "learning_rate": 4.862406233163413e-06,
      "loss": 0.7507,
      "step": 665
    },
    {
      "epoch": 0.027601641178664676,
      "grad_norm": 0.5323469042778015,
      "learning_rate": 4.8621990136350455e-06,
      "loss": 0.811,
      "step": 666
    },
    {
      "epoch": 0.027643085084338347,
      "grad_norm": 0.4284200072288513,
      "learning_rate": 4.861991794106677e-06,
      "loss": 0.7959,
      "step": 667
    },
    {
      "epoch": 0.02768452899001202,
      "grad_norm": 0.43598589301109314,
      "learning_rate": 4.861784574578309e-06,
      "loss": 0.7664,
      "step": 668
    },
    {
      "epoch": 0.02772597289568569,
      "grad_norm": 0.5601446032524109,
      "learning_rate": 4.8615773550499405e-06,
      "loss": 0.8423,
      "step": 669
    },
    {
      "epoch": 0.02776741680135936,
      "grad_norm": 0.47742122411727905,
      "learning_rate": 4.861370135521572e-06,
      "loss": 0.7942,
      "step": 670
    },
    {
      "epoch": 0.02780886070703303,
      "grad_norm": 0.5415261387825012,
      "learning_rate": 4.861162915993203e-06,
      "loss": 0.7916,
      "step": 671
    },
    {
      "epoch": 0.0278503046127067,
      "grad_norm": 0.501384973526001,
      "learning_rate": 4.8609556964648355e-06,
      "loss": 0.7976,
      "step": 672
    },
    {
      "epoch": 0.02789174851838037,
      "grad_norm": 0.4537607431411743,
      "learning_rate": 4.860748476936467e-06,
      "loss": 0.7682,
      "step": 673
    },
    {
      "epoch": 0.027933192424054044,
      "grad_norm": 0.4841330051422119,
      "learning_rate": 4.860541257408098e-06,
      "loss": 0.7483,
      "step": 674
    },
    {
      "epoch": 0.027974636329727715,
      "grad_norm": 0.4894607961177826,
      "learning_rate": 4.8603340378797305e-06,
      "loss": 0.833,
      "step": 675
    },
    {
      "epoch": 0.028016080235401385,
      "grad_norm": 0.4641619920730591,
      "learning_rate": 4.860126818351362e-06,
      "loss": 0.74,
      "step": 676
    },
    {
      "epoch": 0.028057524141075055,
      "grad_norm": 0.4701530337333679,
      "learning_rate": 4.859919598822993e-06,
      "loss": 0.7727,
      "step": 677
    },
    {
      "epoch": 0.028098968046748725,
      "grad_norm": 0.46435436606407166,
      "learning_rate": 4.859712379294625e-06,
      "loss": 0.8499,
      "step": 678
    },
    {
      "epoch": 0.028140411952422395,
      "grad_norm": 0.4807320237159729,
      "learning_rate": 4.859505159766257e-06,
      "loss": 0.8267,
      "step": 679
    },
    {
      "epoch": 0.02818185585809607,
      "grad_norm": 0.46613964438438416,
      "learning_rate": 4.859297940237888e-06,
      "loss": 0.7937,
      "step": 680
    },
    {
      "epoch": 0.02822329976376974,
      "grad_norm": 0.4798445999622345,
      "learning_rate": 4.85909072070952e-06,
      "loss": 0.7693,
      "step": 681
    },
    {
      "epoch": 0.02826474366944341,
      "grad_norm": 0.4688875675201416,
      "learning_rate": 4.858883501181152e-06,
      "loss": 0.7822,
      "step": 682
    },
    {
      "epoch": 0.02830618757511708,
      "grad_norm": 0.5104736685752869,
      "learning_rate": 4.858676281652783e-06,
      "loss": 0.8098,
      "step": 683
    },
    {
      "epoch": 0.02834763148079075,
      "grad_norm": 0.44255852699279785,
      "learning_rate": 4.8584690621244155e-06,
      "loss": 0.7878,
      "step": 684
    },
    {
      "epoch": 0.02838907538646442,
      "grad_norm": 0.461228609085083,
      "learning_rate": 4.858261842596046e-06,
      "loss": 0.8057,
      "step": 685
    },
    {
      "epoch": 0.028430519292138093,
      "grad_norm": 0.4532942771911621,
      "learning_rate": 4.858054623067678e-06,
      "loss": 0.7722,
      "step": 686
    },
    {
      "epoch": 0.028471963197811763,
      "grad_norm": 0.4623205363750458,
      "learning_rate": 4.85784740353931e-06,
      "loss": 0.7737,
      "step": 687
    },
    {
      "epoch": 0.028513407103485433,
      "grad_norm": 0.4836650788784027,
      "learning_rate": 4.857640184010942e-06,
      "loss": 0.8596,
      "step": 688
    },
    {
      "epoch": 0.028554851009159103,
      "grad_norm": 0.5195596814155579,
      "learning_rate": 4.857432964482573e-06,
      "loss": 0.8511,
      "step": 689
    },
    {
      "epoch": 0.028596294914832773,
      "grad_norm": 0.4888651669025421,
      "learning_rate": 4.857225744954205e-06,
      "loss": 0.7889,
      "step": 690
    },
    {
      "epoch": 0.028637738820506443,
      "grad_norm": 0.4511382281780243,
      "learning_rate": 4.857018525425837e-06,
      "loss": 0.7546,
      "step": 691
    },
    {
      "epoch": 0.028679182726180114,
      "grad_norm": 0.42539161443710327,
      "learning_rate": 4.856811305897468e-06,
      "loss": 0.8123,
      "step": 692
    },
    {
      "epoch": 0.028720626631853787,
      "grad_norm": 0.44170787930488586,
      "learning_rate": 4.8566040863691e-06,
      "loss": 0.8096,
      "step": 693
    },
    {
      "epoch": 0.028762070537527457,
      "grad_norm": 0.46986204385757446,
      "learning_rate": 4.856396866840731e-06,
      "loss": 0.814,
      "step": 694
    },
    {
      "epoch": 0.028803514443201127,
      "grad_norm": 0.48899999260902405,
      "learning_rate": 4.856189647312363e-06,
      "loss": 0.854,
      "step": 695
    },
    {
      "epoch": 0.028844958348874798,
      "grad_norm": 0.48885342478752136,
      "learning_rate": 4.855982427783995e-06,
      "loss": 0.7871,
      "step": 696
    },
    {
      "epoch": 0.028886402254548468,
      "grad_norm": 0.4346855878829956,
      "learning_rate": 4.855775208255626e-06,
      "loss": 0.7156,
      "step": 697
    },
    {
      "epoch": 0.028927846160222138,
      "grad_norm": 0.4426824152469635,
      "learning_rate": 4.855567988727258e-06,
      "loss": 0.803,
      "step": 698
    },
    {
      "epoch": 0.02896929006589581,
      "grad_norm": 0.4890058636665344,
      "learning_rate": 4.85536076919889e-06,
      "loss": 0.8206,
      "step": 699
    },
    {
      "epoch": 0.02901073397156948,
      "grad_norm": 0.42204558849334717,
      "learning_rate": 4.855153549670522e-06,
      "loss": 0.7463,
      "step": 700
    },
    {
      "epoch": 0.02905217787724315,
      "grad_norm": 0.46971461176872253,
      "learning_rate": 4.8549463301421525e-06,
      "loss": 0.7766,
      "step": 701
    },
    {
      "epoch": 0.02909362178291682,
      "grad_norm": 0.5029205679893494,
      "learning_rate": 4.854739110613785e-06,
      "loss": 0.783,
      "step": 702
    },
    {
      "epoch": 0.029135065688590492,
      "grad_norm": 0.5121147632598877,
      "learning_rate": 4.854531891085416e-06,
      "loss": 0.8743,
      "step": 703
    },
    {
      "epoch": 0.029176509594264162,
      "grad_norm": 0.480033814907074,
      "learning_rate": 4.854324671557048e-06,
      "loss": 0.7969,
      "step": 704
    },
    {
      "epoch": 0.029217953499937836,
      "grad_norm": 0.47659072279930115,
      "learning_rate": 4.85411745202868e-06,
      "loss": 0.7544,
      "step": 705
    },
    {
      "epoch": 0.029259397405611506,
      "grad_norm": 0.4808111786842346,
      "learning_rate": 4.853910232500311e-06,
      "loss": 0.8296,
      "step": 706
    },
    {
      "epoch": 0.029300841311285176,
      "grad_norm": 0.45272013545036316,
      "learning_rate": 4.853703012971943e-06,
      "loss": 0.7883,
      "step": 707
    },
    {
      "epoch": 0.029342285216958846,
      "grad_norm": 0.478324830532074,
      "learning_rate": 4.853495793443575e-06,
      "loss": 0.813,
      "step": 708
    },
    {
      "epoch": 0.029383729122632516,
      "grad_norm": 0.4619821012020111,
      "learning_rate": 4.853288573915206e-06,
      "loss": 0.7922,
      "step": 709
    },
    {
      "epoch": 0.029425173028306186,
      "grad_norm": 0.44082069396972656,
      "learning_rate": 4.8530813543868375e-06,
      "loss": 0.7776,
      "step": 710
    },
    {
      "epoch": 0.02946661693397986,
      "grad_norm": 0.4187260866165161,
      "learning_rate": 4.85287413485847e-06,
      "loss": 0.7288,
      "step": 711
    },
    {
      "epoch": 0.02950806083965353,
      "grad_norm": 0.4479549527168274,
      "learning_rate": 4.852666915330101e-06,
      "loss": 0.7932,
      "step": 712
    },
    {
      "epoch": 0.0295495047453272,
      "grad_norm": 0.4630022644996643,
      "learning_rate": 4.8524596958017325e-06,
      "loss": 0.7803,
      "step": 713
    },
    {
      "epoch": 0.02959094865100087,
      "grad_norm": 0.4839380085468292,
      "learning_rate": 4.852252476273364e-06,
      "loss": 0.8027,
      "step": 714
    },
    {
      "epoch": 0.02963239255667454,
      "grad_norm": 0.52446448802948,
      "learning_rate": 4.852045256744996e-06,
      "loss": 0.8098,
      "step": 715
    },
    {
      "epoch": 0.02967383646234821,
      "grad_norm": 0.4584200978279114,
      "learning_rate": 4.851838037216628e-06,
      "loss": 0.8459,
      "step": 716
    },
    {
      "epoch": 0.029715280368021884,
      "grad_norm": 0.45347079634666443,
      "learning_rate": 4.851630817688259e-06,
      "loss": 0.8269,
      "step": 717
    },
    {
      "epoch": 0.029756724273695554,
      "grad_norm": 0.46492689847946167,
      "learning_rate": 4.851423598159891e-06,
      "loss": 0.7617,
      "step": 718
    },
    {
      "epoch": 0.029798168179369224,
      "grad_norm": 0.45526203513145447,
      "learning_rate": 4.8512163786315225e-06,
      "loss": 0.7146,
      "step": 719
    },
    {
      "epoch": 0.029839612085042894,
      "grad_norm": 0.4939683973789215,
      "learning_rate": 4.851009159103155e-06,
      "loss": 0.7861,
      "step": 720
    },
    {
      "epoch": 0.029881055990716564,
      "grad_norm": 0.4907040596008301,
      "learning_rate": 4.850801939574786e-06,
      "loss": 0.8308,
      "step": 721
    },
    {
      "epoch": 0.029922499896390235,
      "grad_norm": 0.4499310553073883,
      "learning_rate": 4.8505947200464175e-06,
      "loss": 0.8,
      "step": 722
    },
    {
      "epoch": 0.029963943802063908,
      "grad_norm": 0.4763675034046173,
      "learning_rate": 4.850387500518049e-06,
      "loss": 0.7598,
      "step": 723
    },
    {
      "epoch": 0.03000538770773758,
      "grad_norm": 0.4945710003376007,
      "learning_rate": 4.850180280989681e-06,
      "loss": 0.8596,
      "step": 724
    },
    {
      "epoch": 0.03004683161341125,
      "grad_norm": 0.4911648631095886,
      "learning_rate": 4.8499730614613125e-06,
      "loss": 0.8088,
      "step": 725
    },
    {
      "epoch": 0.03008827551908492,
      "grad_norm": 0.4712153673171997,
      "learning_rate": 4.849765841932944e-06,
      "loss": 0.792,
      "step": 726
    },
    {
      "epoch": 0.03012971942475859,
      "grad_norm": 0.45330849289894104,
      "learning_rate": 4.849558622404576e-06,
      "loss": 0.8354,
      "step": 727
    },
    {
      "epoch": 0.03017116333043226,
      "grad_norm": 0.4708772599697113,
      "learning_rate": 4.8493514028762075e-06,
      "loss": 0.8049,
      "step": 728
    },
    {
      "epoch": 0.03021260723610593,
      "grad_norm": 0.4524977505207062,
      "learning_rate": 4.849144183347839e-06,
      "loss": 0.7917,
      "step": 729
    },
    {
      "epoch": 0.030254051141779603,
      "grad_norm": 0.5035803318023682,
      "learning_rate": 4.84893696381947e-06,
      "loss": 0.7781,
      "step": 730
    },
    {
      "epoch": 0.030295495047453273,
      "grad_norm": 0.47172626852989197,
      "learning_rate": 4.8487297442911025e-06,
      "loss": 0.7856,
      "step": 731
    },
    {
      "epoch": 0.030336938953126943,
      "grad_norm": 0.4861360192298889,
      "learning_rate": 4.848522524762734e-06,
      "loss": 0.8303,
      "step": 732
    },
    {
      "epoch": 0.030378382858800613,
      "grad_norm": 0.44855186343193054,
      "learning_rate": 4.848315305234365e-06,
      "loss": 0.7377,
      "step": 733
    },
    {
      "epoch": 0.030419826764474283,
      "grad_norm": 0.477257639169693,
      "learning_rate": 4.8481080857059975e-06,
      "loss": 0.7854,
      "step": 734
    },
    {
      "epoch": 0.030461270670147953,
      "grad_norm": 0.5123553276062012,
      "learning_rate": 4.847900866177629e-06,
      "loss": 0.8237,
      "step": 735
    },
    {
      "epoch": 0.030502714575821627,
      "grad_norm": 0.49826470017433167,
      "learning_rate": 4.847693646649261e-06,
      "loss": 0.8208,
      "step": 736
    },
    {
      "epoch": 0.030544158481495297,
      "grad_norm": 0.4987088441848755,
      "learning_rate": 4.8474864271208925e-06,
      "loss": 0.7612,
      "step": 737
    },
    {
      "epoch": 0.030585602387168967,
      "grad_norm": 0.5046583414077759,
      "learning_rate": 4.847279207592524e-06,
      "loss": 0.8271,
      "step": 738
    },
    {
      "epoch": 0.030627046292842637,
      "grad_norm": 0.5096154808998108,
      "learning_rate": 4.847071988064155e-06,
      "loss": 0.8513,
      "step": 739
    },
    {
      "epoch": 0.030668490198516307,
      "grad_norm": 0.43660619854927063,
      "learning_rate": 4.8468647685357875e-06,
      "loss": 0.8267,
      "step": 740
    },
    {
      "epoch": 0.030709934104189977,
      "grad_norm": 0.4789735674858093,
      "learning_rate": 4.846657549007419e-06,
      "loss": 0.7532,
      "step": 741
    },
    {
      "epoch": 0.03075137800986365,
      "grad_norm": 0.43951645493507385,
      "learning_rate": 4.84645032947905e-06,
      "loss": 0.7708,
      "step": 742
    },
    {
      "epoch": 0.03079282191553732,
      "grad_norm": 0.46978306770324707,
      "learning_rate": 4.8462431099506825e-06,
      "loss": 0.8218,
      "step": 743
    },
    {
      "epoch": 0.03083426582121099,
      "grad_norm": 0.5086688995361328,
      "learning_rate": 4.846035890422314e-06,
      "loss": 0.8218,
      "step": 744
    },
    {
      "epoch": 0.03087570972688466,
      "grad_norm": 0.4742758870124817,
      "learning_rate": 4.845828670893945e-06,
      "loss": 0.7444,
      "step": 745
    },
    {
      "epoch": 0.03091715363255833,
      "grad_norm": 0.4543149769306183,
      "learning_rate": 4.845621451365577e-06,
      "loss": 0.7594,
      "step": 746
    },
    {
      "epoch": 0.030958597538232,
      "grad_norm": 0.4872720241546631,
      "learning_rate": 4.845414231837209e-06,
      "loss": 0.7544,
      "step": 747
    },
    {
      "epoch": 0.031000041443905675,
      "grad_norm": 0.4720541536808014,
      "learning_rate": 4.84520701230884e-06,
      "loss": 0.8198,
      "step": 748
    },
    {
      "epoch": 0.031041485349579345,
      "grad_norm": 0.448688805103302,
      "learning_rate": 4.844999792780472e-06,
      "loss": 0.7834,
      "step": 749
    },
    {
      "epoch": 0.031082929255253015,
      "grad_norm": 0.4824693202972412,
      "learning_rate": 4.844792573252104e-06,
      "loss": 0.8193,
      "step": 750
    },
    {
      "epoch": 0.031124373160926685,
      "grad_norm": 0.5053086280822754,
      "learning_rate": 4.844585353723735e-06,
      "loss": 0.7954,
      "step": 751
    },
    {
      "epoch": 0.031165817066600356,
      "grad_norm": 0.46447333693504333,
      "learning_rate": 4.8443781341953675e-06,
      "loss": 0.7952,
      "step": 752
    },
    {
      "epoch": 0.031207260972274026,
      "grad_norm": 0.5440570712089539,
      "learning_rate": 4.844170914666999e-06,
      "loss": 0.8206,
      "step": 753
    },
    {
      "epoch": 0.0312487048779477,
      "grad_norm": 0.46322107315063477,
      "learning_rate": 4.84396369513863e-06,
      "loss": 0.7988,
      "step": 754
    },
    {
      "epoch": 0.031290148783621366,
      "grad_norm": 0.5141944885253906,
      "learning_rate": 4.843756475610262e-06,
      "loss": 0.8601,
      "step": 755
    },
    {
      "epoch": 0.031331592689295036,
      "grad_norm": 0.47407636046409607,
      "learning_rate": 4.843549256081894e-06,
      "loss": 0.8118,
      "step": 756
    },
    {
      "epoch": 0.03137303659496871,
      "grad_norm": 0.502138614654541,
      "learning_rate": 4.843342036553525e-06,
      "loss": 0.8162,
      "step": 757
    },
    {
      "epoch": 0.03141448050064238,
      "grad_norm": 0.4771718382835388,
      "learning_rate": 4.843134817025157e-06,
      "loss": 0.8069,
      "step": 758
    },
    {
      "epoch": 0.03145592440631605,
      "grad_norm": 0.5140120387077332,
      "learning_rate": 4.842927597496789e-06,
      "loss": 0.781,
      "step": 759
    },
    {
      "epoch": 0.031497368311989724,
      "grad_norm": 0.4754839539527893,
      "learning_rate": 4.84272037796842e-06,
      "loss": 0.8174,
      "step": 760
    },
    {
      "epoch": 0.031538812217663394,
      "grad_norm": 0.4931987524032593,
      "learning_rate": 4.842513158440052e-06,
      "loss": 0.7605,
      "step": 761
    },
    {
      "epoch": 0.031580256123337064,
      "grad_norm": 0.49690335988998413,
      "learning_rate": 4.842305938911683e-06,
      "loss": 0.8198,
      "step": 762
    },
    {
      "epoch": 0.031621700029010734,
      "grad_norm": 0.4721209704875946,
      "learning_rate": 4.842098719383315e-06,
      "loss": 0.803,
      "step": 763
    },
    {
      "epoch": 0.031663143934684404,
      "grad_norm": 0.45114579796791077,
      "learning_rate": 4.841891499854947e-06,
      "loss": 0.8425,
      "step": 764
    },
    {
      "epoch": 0.031704587840358074,
      "grad_norm": 0.4510507583618164,
      "learning_rate": 4.841684280326578e-06,
      "loss": 0.7568,
      "step": 765
    },
    {
      "epoch": 0.031746031746031744,
      "grad_norm": 0.4513837993144989,
      "learning_rate": 4.8414770607982095e-06,
      "loss": 0.7849,
      "step": 766
    },
    {
      "epoch": 0.031787475651705414,
      "grad_norm": 0.510221004486084,
      "learning_rate": 4.841269841269842e-06,
      "loss": 0.7654,
      "step": 767
    },
    {
      "epoch": 0.031828919557379085,
      "grad_norm": 0.47325363755226135,
      "learning_rate": 4.841062621741474e-06,
      "loss": 0.803,
      "step": 768
    },
    {
      "epoch": 0.031870363463052755,
      "grad_norm": 0.46081799268722534,
      "learning_rate": 4.8408554022131045e-06,
      "loss": 0.7904,
      "step": 769
    },
    {
      "epoch": 0.03191180736872643,
      "grad_norm": 0.5060776472091675,
      "learning_rate": 4.840648182684737e-06,
      "loss": 0.7927,
      "step": 770
    },
    {
      "epoch": 0.0319532512744001,
      "grad_norm": 0.47735685110092163,
      "learning_rate": 4.840440963156368e-06,
      "loss": 0.8318,
      "step": 771
    },
    {
      "epoch": 0.03199469518007377,
      "grad_norm": 0.4513167142868042,
      "learning_rate": 4.840233743628e-06,
      "loss": 0.7734,
      "step": 772
    },
    {
      "epoch": 0.03203613908574744,
      "grad_norm": 0.5181498527526855,
      "learning_rate": 4.840026524099632e-06,
      "loss": 0.8201,
      "step": 773
    },
    {
      "epoch": 0.03207758299142111,
      "grad_norm": 0.48776254057884216,
      "learning_rate": 4.839819304571263e-06,
      "loss": 0.7979,
      "step": 774
    },
    {
      "epoch": 0.03211902689709478,
      "grad_norm": 0.45880892872810364,
      "learning_rate": 4.8396120850428945e-06,
      "loss": 0.8087,
      "step": 775
    },
    {
      "epoch": 0.03216047080276845,
      "grad_norm": 0.4686993658542633,
      "learning_rate": 4.839404865514527e-06,
      "loss": 0.8337,
      "step": 776
    },
    {
      "epoch": 0.03220191470844212,
      "grad_norm": 0.4397232234477997,
      "learning_rate": 4.839197645986158e-06,
      "loss": 0.7578,
      "step": 777
    },
    {
      "epoch": 0.03224335861411579,
      "grad_norm": 0.4607049226760864,
      "learning_rate": 4.8389904264577895e-06,
      "loss": 0.8018,
      "step": 778
    },
    {
      "epoch": 0.03228480251978946,
      "grad_norm": 0.4511394500732422,
      "learning_rate": 4.838783206929422e-06,
      "loss": 0.8093,
      "step": 779
    },
    {
      "epoch": 0.03232624642546313,
      "grad_norm": 0.4307085871696472,
      "learning_rate": 4.838575987401053e-06,
      "loss": 0.7722,
      "step": 780
    },
    {
      "epoch": 0.0323676903311368,
      "grad_norm": 0.4876425862312317,
      "learning_rate": 4.8383687678726845e-06,
      "loss": 0.7537,
      "step": 781
    },
    {
      "epoch": 0.03240913423681048,
      "grad_norm": 0.4792354702949524,
      "learning_rate": 4.838161548344316e-06,
      "loss": 0.8357,
      "step": 782
    },
    {
      "epoch": 0.03245057814248415,
      "grad_norm": 0.4683007299900055,
      "learning_rate": 4.837954328815948e-06,
      "loss": 0.7688,
      "step": 783
    },
    {
      "epoch": 0.03249202204815782,
      "grad_norm": 0.46011966466903687,
      "learning_rate": 4.8377471092875795e-06,
      "loss": 0.8152,
      "step": 784
    },
    {
      "epoch": 0.03253346595383149,
      "grad_norm": 0.48548996448516846,
      "learning_rate": 4.837539889759211e-06,
      "loss": 0.814,
      "step": 785
    },
    {
      "epoch": 0.03257490985950516,
      "grad_norm": 0.4813372492790222,
      "learning_rate": 4.837332670230843e-06,
      "loss": 0.8186,
      "step": 786
    },
    {
      "epoch": 0.03261635376517883,
      "grad_norm": 0.4809481203556061,
      "learning_rate": 4.8371254507024745e-06,
      "loss": 0.7795,
      "step": 787
    },
    {
      "epoch": 0.0326577976708525,
      "grad_norm": 0.47284194827079773,
      "learning_rate": 4.836918231174107e-06,
      "loss": 0.8435,
      "step": 788
    },
    {
      "epoch": 0.03269924157652617,
      "grad_norm": 0.48725366592407227,
      "learning_rate": 4.836711011645738e-06,
      "loss": 0.7795,
      "step": 789
    },
    {
      "epoch": 0.03274068548219984,
      "grad_norm": 0.5093293190002441,
      "learning_rate": 4.8365037921173695e-06,
      "loss": 0.8564,
      "step": 790
    },
    {
      "epoch": 0.03278212938787351,
      "grad_norm": 0.45030084252357483,
      "learning_rate": 4.836296572589001e-06,
      "loss": 0.7986,
      "step": 791
    },
    {
      "epoch": 0.03282357329354718,
      "grad_norm": 0.45919808745384216,
      "learning_rate": 4.836089353060633e-06,
      "loss": 0.8132,
      "step": 792
    },
    {
      "epoch": 0.03286501719922085,
      "grad_norm": 0.46075549721717834,
      "learning_rate": 4.8358821335322645e-06,
      "loss": 0.7678,
      "step": 793
    },
    {
      "epoch": 0.03290646110489453,
      "grad_norm": 0.5387317538261414,
      "learning_rate": 4.835674914003896e-06,
      "loss": 0.8026,
      "step": 794
    },
    {
      "epoch": 0.0329479050105682,
      "grad_norm": 0.446251779794693,
      "learning_rate": 4.835467694475528e-06,
      "loss": 0.7966,
      "step": 795
    },
    {
      "epoch": 0.03298934891624187,
      "grad_norm": 0.4801095426082611,
      "learning_rate": 4.8352604749471595e-06,
      "loss": 0.785,
      "step": 796
    },
    {
      "epoch": 0.03303079282191554,
      "grad_norm": 0.47974684834480286,
      "learning_rate": 4.835053255418791e-06,
      "loss": 0.7939,
      "step": 797
    },
    {
      "epoch": 0.03307223672758921,
      "grad_norm": 0.46524977684020996,
      "learning_rate": 4.834846035890422e-06,
      "loss": 0.8074,
      "step": 798
    },
    {
      "epoch": 0.03311368063326288,
      "grad_norm": 0.5188291072845459,
      "learning_rate": 4.8346388163620545e-06,
      "loss": 0.843,
      "step": 799
    },
    {
      "epoch": 0.03315512453893655,
      "grad_norm": 0.47121351957321167,
      "learning_rate": 4.834431596833686e-06,
      "loss": 0.843,
      "step": 800
    },
    {
      "epoch": 0.03319656844461022,
      "grad_norm": 0.4718429446220398,
      "learning_rate": 4.834224377305317e-06,
      "loss": 0.8169,
      "step": 801
    },
    {
      "epoch": 0.03323801235028389,
      "grad_norm": 0.6425647139549255,
      "learning_rate": 4.834017157776949e-06,
      "loss": 0.8171,
      "step": 802
    },
    {
      "epoch": 0.03327945625595756,
      "grad_norm": 0.5799872875213623,
      "learning_rate": 4.833809938248581e-06,
      "loss": 0.7239,
      "step": 803
    },
    {
      "epoch": 0.03332090016163123,
      "grad_norm": 0.46306636929512024,
      "learning_rate": 4.833602718720213e-06,
      "loss": 0.8054,
      "step": 804
    },
    {
      "epoch": 0.0333623440673049,
      "grad_norm": 0.4275192320346832,
      "learning_rate": 4.8333954991918445e-06,
      "loss": 0.7664,
      "step": 805
    },
    {
      "epoch": 0.03340378797297858,
      "grad_norm": 0.5045058131217957,
      "learning_rate": 4.833188279663476e-06,
      "loss": 0.7881,
      "step": 806
    },
    {
      "epoch": 0.03344523187865225,
      "grad_norm": 0.44459038972854614,
      "learning_rate": 4.832981060135107e-06,
      "loss": 0.853,
      "step": 807
    },
    {
      "epoch": 0.03348667578432592,
      "grad_norm": 0.40647628903388977,
      "learning_rate": 4.8327738406067395e-06,
      "loss": 0.7952,
      "step": 808
    },
    {
      "epoch": 0.03352811968999959,
      "grad_norm": 0.4865948557853699,
      "learning_rate": 4.832566621078371e-06,
      "loss": 0.8064,
      "step": 809
    },
    {
      "epoch": 0.03356956359567326,
      "grad_norm": 0.4718790650367737,
      "learning_rate": 4.832359401550002e-06,
      "loss": 0.8225,
      "step": 810
    },
    {
      "epoch": 0.03361100750134693,
      "grad_norm": 0.49259084463119507,
      "learning_rate": 4.8321521820216345e-06,
      "loss": 0.7886,
      "step": 811
    },
    {
      "epoch": 0.0336524514070206,
      "grad_norm": 0.48640403151512146,
      "learning_rate": 4.831944962493266e-06,
      "loss": 0.7993,
      "step": 812
    },
    {
      "epoch": 0.03369389531269427,
      "grad_norm": 0.43770548701286316,
      "learning_rate": 4.831737742964897e-06,
      "loss": 0.8164,
      "step": 813
    },
    {
      "epoch": 0.03373533921836794,
      "grad_norm": 0.49797001481056213,
      "learning_rate": 4.831530523436529e-06,
      "loss": 0.8281,
      "step": 814
    },
    {
      "epoch": 0.03377678312404161,
      "grad_norm": 0.4484914243221283,
      "learning_rate": 4.831323303908161e-06,
      "loss": 0.8237,
      "step": 815
    },
    {
      "epoch": 0.03381822702971528,
      "grad_norm": 0.514343798160553,
      "learning_rate": 4.831116084379792e-06,
      "loss": 0.7988,
      "step": 816
    },
    {
      "epoch": 0.03385967093538895,
      "grad_norm": 0.48368391394615173,
      "learning_rate": 4.830908864851424e-06,
      "loss": 0.8132,
      "step": 817
    },
    {
      "epoch": 0.03390111484106262,
      "grad_norm": 0.4572143256664276,
      "learning_rate": 4.830701645323055e-06,
      "loss": 0.7871,
      "step": 818
    },
    {
      "epoch": 0.033942558746736295,
      "grad_norm": 0.46438106894493103,
      "learning_rate": 4.830494425794687e-06,
      "loss": 0.7914,
      "step": 819
    },
    {
      "epoch": 0.033984002652409966,
      "grad_norm": 0.4726259708404541,
      "learning_rate": 4.8302872062663196e-06,
      "loss": 0.801,
      "step": 820
    },
    {
      "epoch": 0.034025446558083636,
      "grad_norm": 0.4818885624408722,
      "learning_rate": 4.830079986737951e-06,
      "loss": 0.8318,
      "step": 821
    },
    {
      "epoch": 0.034066890463757306,
      "grad_norm": 0.44731706380844116,
      "learning_rate": 4.829872767209582e-06,
      "loss": 0.7549,
      "step": 822
    },
    {
      "epoch": 0.034108334369430976,
      "grad_norm": 0.44706088304519653,
      "learning_rate": 4.829665547681214e-06,
      "loss": 0.7705,
      "step": 823
    },
    {
      "epoch": 0.034149778275104646,
      "grad_norm": 0.48286518454551697,
      "learning_rate": 4.829458328152846e-06,
      "loss": 0.8022,
      "step": 824
    },
    {
      "epoch": 0.034191222180778316,
      "grad_norm": 0.4611935615539551,
      "learning_rate": 4.829251108624477e-06,
      "loss": 0.8064,
      "step": 825
    },
    {
      "epoch": 0.034232666086451986,
      "grad_norm": 0.5443453192710876,
      "learning_rate": 4.829043889096109e-06,
      "loss": 0.8209,
      "step": 826
    },
    {
      "epoch": 0.034274109992125656,
      "grad_norm": 0.47493240237236023,
      "learning_rate": 4.82883666956774e-06,
      "loss": 0.7751,
      "step": 827
    },
    {
      "epoch": 0.03431555389779933,
      "grad_norm": 0.4742193818092346,
      "learning_rate": 4.828629450039372e-06,
      "loss": 0.7634,
      "step": 828
    },
    {
      "epoch": 0.034356997803473,
      "grad_norm": 0.44926828145980835,
      "learning_rate": 4.828422230511004e-06,
      "loss": 0.761,
      "step": 829
    },
    {
      "epoch": 0.03439844170914667,
      "grad_norm": 0.4580516219139099,
      "learning_rate": 4.828215010982635e-06,
      "loss": 0.7488,
      "step": 830
    },
    {
      "epoch": 0.034439885614820344,
      "grad_norm": 0.45133379101753235,
      "learning_rate": 4.828007791454267e-06,
      "loss": 0.749,
      "step": 831
    },
    {
      "epoch": 0.034481329520494014,
      "grad_norm": 0.43070828914642334,
      "learning_rate": 4.827800571925899e-06,
      "loss": 0.7842,
      "step": 832
    },
    {
      "epoch": 0.034522773426167684,
      "grad_norm": 0.463408499956131,
      "learning_rate": 4.82759335239753e-06,
      "loss": 0.804,
      "step": 833
    },
    {
      "epoch": 0.034564217331841354,
      "grad_norm": 0.424195796251297,
      "learning_rate": 4.8273861328691615e-06,
      "loss": 0.7493,
      "step": 834
    },
    {
      "epoch": 0.034605661237515024,
      "grad_norm": 0.4392853081226349,
      "learning_rate": 4.827178913340794e-06,
      "loss": 0.8049,
      "step": 835
    },
    {
      "epoch": 0.034647105143188694,
      "grad_norm": 0.5163056254386902,
      "learning_rate": 4.826971693812425e-06,
      "loss": 0.8274,
      "step": 836
    },
    {
      "epoch": 0.034688549048862365,
      "grad_norm": 0.46480703353881836,
      "learning_rate": 4.8267644742840565e-06,
      "loss": 0.8071,
      "step": 837
    },
    {
      "epoch": 0.034729992954536035,
      "grad_norm": 0.4739395081996918,
      "learning_rate": 4.826557254755689e-06,
      "loss": 0.7964,
      "step": 838
    },
    {
      "epoch": 0.034771436860209705,
      "grad_norm": 0.45199519395828247,
      "learning_rate": 4.82635003522732e-06,
      "loss": 0.803,
      "step": 839
    },
    {
      "epoch": 0.034812880765883375,
      "grad_norm": 0.44721484184265137,
      "learning_rate": 4.826142815698952e-06,
      "loss": 0.7593,
      "step": 840
    },
    {
      "epoch": 0.034854324671557045,
      "grad_norm": 0.4465482831001282,
      "learning_rate": 4.825935596170584e-06,
      "loss": 0.814,
      "step": 841
    },
    {
      "epoch": 0.034895768577230715,
      "grad_norm": 0.5203372836112976,
      "learning_rate": 4.825728376642215e-06,
      "loss": 0.8416,
      "step": 842
    },
    {
      "epoch": 0.03493721248290439,
      "grad_norm": 0.48009631037712097,
      "learning_rate": 4.8255211571138465e-06,
      "loss": 0.905,
      "step": 843
    },
    {
      "epoch": 0.03497865638857806,
      "grad_norm": 0.48944076895713806,
      "learning_rate": 4.825313937585479e-06,
      "loss": 0.813,
      "step": 844
    },
    {
      "epoch": 0.03502010029425173,
      "grad_norm": 0.5161928534507751,
      "learning_rate": 4.82510671805711e-06,
      "loss": 0.8391,
      "step": 845
    },
    {
      "epoch": 0.0350615441999254,
      "grad_norm": 0.49111801385879517,
      "learning_rate": 4.8248994985287415e-06,
      "loss": 0.7148,
      "step": 846
    },
    {
      "epoch": 0.03510298810559907,
      "grad_norm": 0.452984482049942,
      "learning_rate": 4.824692279000374e-06,
      "loss": 0.7896,
      "step": 847
    },
    {
      "epoch": 0.03514443201127274,
      "grad_norm": 0.44530120491981506,
      "learning_rate": 4.824485059472005e-06,
      "loss": 0.7717,
      "step": 848
    },
    {
      "epoch": 0.03518587591694641,
      "grad_norm": 0.4604248106479645,
      "learning_rate": 4.8242778399436365e-06,
      "loss": 0.8225,
      "step": 849
    },
    {
      "epoch": 0.03522731982262008,
      "grad_norm": 0.49463069438934326,
      "learning_rate": 4.824070620415268e-06,
      "loss": 0.7812,
      "step": 850
    },
    {
      "epoch": 0.03526876372829375,
      "grad_norm": 0.4273708760738373,
      "learning_rate": 4.8238634008869e-06,
      "loss": 0.7625,
      "step": 851
    },
    {
      "epoch": 0.03531020763396742,
      "grad_norm": 0.5574250221252441,
      "learning_rate": 4.8236561813585315e-06,
      "loss": 0.8408,
      "step": 852
    },
    {
      "epoch": 0.035351651539641094,
      "grad_norm": 0.49560749530792236,
      "learning_rate": 4.823448961830163e-06,
      "loss": 0.8496,
      "step": 853
    },
    {
      "epoch": 0.035393095445314764,
      "grad_norm": 0.5219417810440063,
      "learning_rate": 4.823241742301794e-06,
      "loss": 0.814,
      "step": 854
    },
    {
      "epoch": 0.035434539350988434,
      "grad_norm": 0.4465983510017395,
      "learning_rate": 4.8230345227734265e-06,
      "loss": 0.8118,
      "step": 855
    },
    {
      "epoch": 0.03547598325666211,
      "grad_norm": 0.4661894142627716,
      "learning_rate": 4.822827303245059e-06,
      "loss": 0.7755,
      "step": 856
    },
    {
      "epoch": 0.03551742716233578,
      "grad_norm": 0.45798560976982117,
      "learning_rate": 4.82262008371669e-06,
      "loss": 0.8206,
      "step": 857
    },
    {
      "epoch": 0.03555887106800945,
      "grad_norm": 0.447319358587265,
      "learning_rate": 4.8224128641883215e-06,
      "loss": 0.8215,
      "step": 858
    },
    {
      "epoch": 0.03560031497368312,
      "grad_norm": 0.47446373105049133,
      "learning_rate": 4.822205644659953e-06,
      "loss": 0.7686,
      "step": 859
    },
    {
      "epoch": 0.03564175887935679,
      "grad_norm": 0.5082666873931885,
      "learning_rate": 4.821998425131585e-06,
      "loss": 0.8584,
      "step": 860
    },
    {
      "epoch": 0.03568320278503046,
      "grad_norm": 0.4586051404476166,
      "learning_rate": 4.8217912056032165e-06,
      "loss": 0.7856,
      "step": 861
    },
    {
      "epoch": 0.03572464669070413,
      "grad_norm": 0.4524220824241638,
      "learning_rate": 4.821583986074848e-06,
      "loss": 0.7869,
      "step": 862
    },
    {
      "epoch": 0.0357660905963778,
      "grad_norm": 0.47431883215904236,
      "learning_rate": 4.821376766546479e-06,
      "loss": 0.8022,
      "step": 863
    },
    {
      "epoch": 0.03580753450205147,
      "grad_norm": 0.4569200277328491,
      "learning_rate": 4.8211695470181115e-06,
      "loss": 0.8242,
      "step": 864
    },
    {
      "epoch": 0.03584897840772514,
      "grad_norm": 0.4440664052963257,
      "learning_rate": 4.820962327489743e-06,
      "loss": 0.7637,
      "step": 865
    },
    {
      "epoch": 0.03589042231339881,
      "grad_norm": 0.4485827386379242,
      "learning_rate": 4.820755107961374e-06,
      "loss": 0.8101,
      "step": 866
    },
    {
      "epoch": 0.03593186621907248,
      "grad_norm": 0.46246907114982605,
      "learning_rate": 4.8205478884330065e-06,
      "loss": 0.7793,
      "step": 867
    },
    {
      "epoch": 0.03597331012474616,
      "grad_norm": 0.4496723413467407,
      "learning_rate": 4.820340668904638e-06,
      "loss": 0.7684,
      "step": 868
    },
    {
      "epoch": 0.03601475403041983,
      "grad_norm": 0.43587517738342285,
      "learning_rate": 4.820133449376269e-06,
      "loss": 0.7339,
      "step": 869
    },
    {
      "epoch": 0.0360561979360935,
      "grad_norm": 0.4498605728149414,
      "learning_rate": 4.819926229847901e-06,
      "loss": 0.8093,
      "step": 870
    },
    {
      "epoch": 0.03609764184176717,
      "grad_norm": 0.44379284977912903,
      "learning_rate": 4.819719010319533e-06,
      "loss": 0.7986,
      "step": 871
    },
    {
      "epoch": 0.03613908574744084,
      "grad_norm": 0.4766504466533661,
      "learning_rate": 4.819511790791165e-06,
      "loss": 0.7517,
      "step": 872
    },
    {
      "epoch": 0.03618052965311451,
      "grad_norm": 0.45311790704727173,
      "learning_rate": 4.8193045712627966e-06,
      "loss": 0.7595,
      "step": 873
    },
    {
      "epoch": 0.03622197355878818,
      "grad_norm": 0.4438081979751587,
      "learning_rate": 4.819097351734428e-06,
      "loss": 0.7576,
      "step": 874
    },
    {
      "epoch": 0.03626341746446185,
      "grad_norm": 0.4847230017185211,
      "learning_rate": 4.818890132206059e-06,
      "loss": 0.8088,
      "step": 875
    },
    {
      "epoch": 0.03630486137013552,
      "grad_norm": 0.43706944584846497,
      "learning_rate": 4.8186829126776916e-06,
      "loss": 0.7964,
      "step": 876
    },
    {
      "epoch": 0.03634630527580919,
      "grad_norm": 0.47320547699928284,
      "learning_rate": 4.818475693149323e-06,
      "loss": 0.783,
      "step": 877
    },
    {
      "epoch": 0.03638774918148286,
      "grad_norm": 0.4701194167137146,
      "learning_rate": 4.818268473620954e-06,
      "loss": 0.8491,
      "step": 878
    },
    {
      "epoch": 0.03642919308715653,
      "grad_norm": 0.44274404644966125,
      "learning_rate": 4.818061254092586e-06,
      "loss": 0.7834,
      "step": 879
    },
    {
      "epoch": 0.03647063699283021,
      "grad_norm": 0.4761819839477539,
      "learning_rate": 4.817854034564218e-06,
      "loss": 0.812,
      "step": 880
    },
    {
      "epoch": 0.03651208089850388,
      "grad_norm": 0.4779709577560425,
      "learning_rate": 4.817646815035849e-06,
      "loss": 0.781,
      "step": 881
    },
    {
      "epoch": 0.03655352480417755,
      "grad_norm": 0.42150968313217163,
      "learning_rate": 4.817439595507481e-06,
      "loss": 0.7827,
      "step": 882
    },
    {
      "epoch": 0.03659496870985122,
      "grad_norm": 0.46990767121315,
      "learning_rate": 4.817232375979113e-06,
      "loss": 0.8174,
      "step": 883
    },
    {
      "epoch": 0.03663641261552489,
      "grad_norm": 0.4755314588546753,
      "learning_rate": 4.817025156450744e-06,
      "loss": 0.7622,
      "step": 884
    },
    {
      "epoch": 0.03667785652119856,
      "grad_norm": 0.4565747082233429,
      "learning_rate": 4.816817936922376e-06,
      "loss": 0.7478,
      "step": 885
    },
    {
      "epoch": 0.03671930042687223,
      "grad_norm": 0.4576369822025299,
      "learning_rate": 4.816610717394007e-06,
      "loss": 0.8198,
      "step": 886
    },
    {
      "epoch": 0.0367607443325459,
      "grad_norm": 0.44579315185546875,
      "learning_rate": 4.816403497865639e-06,
      "loss": 0.8186,
      "step": 887
    },
    {
      "epoch": 0.03680218823821957,
      "grad_norm": 0.44514021277427673,
      "learning_rate": 4.816196278337271e-06,
      "loss": 0.8125,
      "step": 888
    },
    {
      "epoch": 0.03684363214389324,
      "grad_norm": 0.491281121969223,
      "learning_rate": 4.815989058808903e-06,
      "loss": 0.8308,
      "step": 889
    },
    {
      "epoch": 0.03688507604956691,
      "grad_norm": 0.4317649006843567,
      "learning_rate": 4.815781839280534e-06,
      "loss": 0.8181,
      "step": 890
    },
    {
      "epoch": 0.03692651995524058,
      "grad_norm": 0.4579673707485199,
      "learning_rate": 4.815574619752166e-06,
      "loss": 0.7024,
      "step": 891
    },
    {
      "epoch": 0.03696796386091425,
      "grad_norm": 0.4602883756160736,
      "learning_rate": 4.815367400223798e-06,
      "loss": 0.7803,
      "step": 892
    },
    {
      "epoch": 0.037009407766587926,
      "grad_norm": 0.517935037612915,
      "learning_rate": 4.815160180695429e-06,
      "loss": 0.8113,
      "step": 893
    },
    {
      "epoch": 0.037050851672261596,
      "grad_norm": 0.48947110772132874,
      "learning_rate": 4.814952961167061e-06,
      "loss": 0.8264,
      "step": 894
    },
    {
      "epoch": 0.037092295577935266,
      "grad_norm": 0.47166675329208374,
      "learning_rate": 4.814745741638692e-06,
      "loss": 0.7656,
      "step": 895
    },
    {
      "epoch": 0.03713373948360894,
      "grad_norm": 0.44415104389190674,
      "learning_rate": 4.814538522110324e-06,
      "loss": 0.7576,
      "step": 896
    },
    {
      "epoch": 0.03717518338928261,
      "grad_norm": 0.4769923686981201,
      "learning_rate": 4.814331302581956e-06,
      "loss": 0.7905,
      "step": 897
    },
    {
      "epoch": 0.03721662729495628,
      "grad_norm": 0.4704188406467438,
      "learning_rate": 4.814124083053587e-06,
      "loss": 0.7661,
      "step": 898
    },
    {
      "epoch": 0.03725807120062995,
      "grad_norm": 0.4227563142776489,
      "learning_rate": 4.813916863525219e-06,
      "loss": 0.7949,
      "step": 899
    },
    {
      "epoch": 0.03729951510630362,
      "grad_norm": 0.4472195506095886,
      "learning_rate": 4.813709643996851e-06,
      "loss": 0.7991,
      "step": 900
    },
    {
      "epoch": 0.03734095901197729,
      "grad_norm": 0.4472038447856903,
      "learning_rate": 4.813502424468482e-06,
      "loss": 0.7644,
      "step": 901
    },
    {
      "epoch": 0.03738240291765096,
      "grad_norm": 0.4717763364315033,
      "learning_rate": 4.8132952049401135e-06,
      "loss": 0.8101,
      "step": 902
    },
    {
      "epoch": 0.03742384682332463,
      "grad_norm": 0.5045163631439209,
      "learning_rate": 4.813087985411746e-06,
      "loss": 0.7876,
      "step": 903
    },
    {
      "epoch": 0.0374652907289983,
      "grad_norm": 0.439333975315094,
      "learning_rate": 4.812880765883377e-06,
      "loss": 0.7568,
      "step": 904
    },
    {
      "epoch": 0.037506734634671975,
      "grad_norm": 0.47774791717529297,
      "learning_rate": 4.812673546355009e-06,
      "loss": 0.8582,
      "step": 905
    },
    {
      "epoch": 0.037548178540345645,
      "grad_norm": 0.4379725456237793,
      "learning_rate": 4.81246632682664e-06,
      "loss": 0.7554,
      "step": 906
    },
    {
      "epoch": 0.037589622446019315,
      "grad_norm": 0.45828449726104736,
      "learning_rate": 4.812259107298272e-06,
      "loss": 0.8049,
      "step": 907
    },
    {
      "epoch": 0.037631066351692985,
      "grad_norm": 0.5110654234886169,
      "learning_rate": 4.812051887769904e-06,
      "loss": 0.782,
      "step": 908
    },
    {
      "epoch": 0.037672510257366655,
      "grad_norm": 0.436849981546402,
      "learning_rate": 4.811844668241536e-06,
      "loss": 0.7869,
      "step": 909
    },
    {
      "epoch": 0.037713954163040325,
      "grad_norm": 0.5075635313987732,
      "learning_rate": 4.811637448713167e-06,
      "loss": 0.8484,
      "step": 910
    },
    {
      "epoch": 0.037755398068713995,
      "grad_norm": 0.4612642228603363,
      "learning_rate": 4.8114302291847985e-06,
      "loss": 0.8284,
      "step": 911
    },
    {
      "epoch": 0.037796841974387665,
      "grad_norm": 0.4357260763645172,
      "learning_rate": 4.811223009656431e-06,
      "loss": 0.7539,
      "step": 912
    },
    {
      "epoch": 0.037838285880061336,
      "grad_norm": 0.41013216972351074,
      "learning_rate": 4.811015790128062e-06,
      "loss": 0.7336,
      "step": 913
    },
    {
      "epoch": 0.037879729785735006,
      "grad_norm": 0.5076503157615662,
      "learning_rate": 4.8108085705996935e-06,
      "loss": 0.8237,
      "step": 914
    },
    {
      "epoch": 0.037921173691408676,
      "grad_norm": 0.5303014516830444,
      "learning_rate": 4.810601351071325e-06,
      "loss": 0.8384,
      "step": 915
    },
    {
      "epoch": 0.037962617597082346,
      "grad_norm": 0.5182031393051147,
      "learning_rate": 4.810394131542957e-06,
      "loss": 0.8159,
      "step": 916
    },
    {
      "epoch": 0.03800406150275602,
      "grad_norm": 0.4546484053134918,
      "learning_rate": 4.8101869120145885e-06,
      "loss": 0.7715,
      "step": 917
    },
    {
      "epoch": 0.03804550540842969,
      "grad_norm": 0.4660073220729828,
      "learning_rate": 4.80997969248622e-06,
      "loss": 0.8127,
      "step": 918
    },
    {
      "epoch": 0.03808694931410336,
      "grad_norm": 0.5112962126731873,
      "learning_rate": 4.809772472957852e-06,
      "loss": 0.8281,
      "step": 919
    },
    {
      "epoch": 0.03812839321977703,
      "grad_norm": 0.467479407787323,
      "learning_rate": 4.8095652534294835e-06,
      "loss": 0.7578,
      "step": 920
    },
    {
      "epoch": 0.038169837125450704,
      "grad_norm": 0.4521731436252594,
      "learning_rate": 4.809358033901115e-06,
      "loss": 0.7488,
      "step": 921
    },
    {
      "epoch": 0.038211281031124374,
      "grad_norm": 0.42674127221107483,
      "learning_rate": 4.809150814372746e-06,
      "loss": 0.752,
      "step": 922
    },
    {
      "epoch": 0.038252724936798044,
      "grad_norm": 0.42088428139686584,
      "learning_rate": 4.8089435948443785e-06,
      "loss": 0.7764,
      "step": 923
    },
    {
      "epoch": 0.038294168842471714,
      "grad_norm": 0.47706201672554016,
      "learning_rate": 4.80873637531601e-06,
      "loss": 0.7947,
      "step": 924
    },
    {
      "epoch": 0.038335612748145384,
      "grad_norm": 0.4748612940311432,
      "learning_rate": 4.808529155787642e-06,
      "loss": 0.8188,
      "step": 925
    },
    {
      "epoch": 0.038377056653819054,
      "grad_norm": 0.45237982273101807,
      "learning_rate": 4.8083219362592735e-06,
      "loss": 0.7661,
      "step": 926
    },
    {
      "epoch": 0.038418500559492724,
      "grad_norm": 0.4897715747356415,
      "learning_rate": 4.808114716730905e-06,
      "loss": 0.7974,
      "step": 927
    },
    {
      "epoch": 0.038459944465166394,
      "grad_norm": 0.4391297996044159,
      "learning_rate": 4.807907497202537e-06,
      "loss": 0.7759,
      "step": 928
    },
    {
      "epoch": 0.03850138837084007,
      "grad_norm": 0.4632616341114044,
      "learning_rate": 4.8077002776741685e-06,
      "loss": 0.7887,
      "step": 929
    },
    {
      "epoch": 0.03854283227651374,
      "grad_norm": 0.4474062919616699,
      "learning_rate": 4.8074930581458e-06,
      "loss": 0.8311,
      "step": 930
    },
    {
      "epoch": 0.03858427618218741,
      "grad_norm": 0.48291000723838806,
      "learning_rate": 4.807285838617431e-06,
      "loss": 0.7312,
      "step": 931
    },
    {
      "epoch": 0.03862572008786108,
      "grad_norm": 0.4543563723564148,
      "learning_rate": 4.8070786190890636e-06,
      "loss": 0.8115,
      "step": 932
    },
    {
      "epoch": 0.03866716399353475,
      "grad_norm": 0.5148612260818481,
      "learning_rate": 4.806871399560695e-06,
      "loss": 0.8621,
      "step": 933
    },
    {
      "epoch": 0.03870860789920842,
      "grad_norm": 0.4792311489582062,
      "learning_rate": 4.806664180032326e-06,
      "loss": 0.8262,
      "step": 934
    },
    {
      "epoch": 0.03875005180488209,
      "grad_norm": 0.46342992782592773,
      "learning_rate": 4.8064569605039586e-06,
      "loss": 0.8003,
      "step": 935
    },
    {
      "epoch": 0.03879149571055576,
      "grad_norm": 0.45019346475601196,
      "learning_rate": 4.80624974097559e-06,
      "loss": 0.7874,
      "step": 936
    },
    {
      "epoch": 0.03883293961622943,
      "grad_norm": 0.5109577775001526,
      "learning_rate": 4.806042521447221e-06,
      "loss": 0.8142,
      "step": 937
    },
    {
      "epoch": 0.0388743835219031,
      "grad_norm": 0.44418174028396606,
      "learning_rate": 4.805835301918853e-06,
      "loss": 0.7892,
      "step": 938
    },
    {
      "epoch": 0.03891582742757677,
      "grad_norm": 0.4462260901927948,
      "learning_rate": 4.805628082390485e-06,
      "loss": 0.7908,
      "step": 939
    },
    {
      "epoch": 0.03895727133325044,
      "grad_norm": 0.4844035804271698,
      "learning_rate": 4.805420862862116e-06,
      "loss": 0.7781,
      "step": 940
    },
    {
      "epoch": 0.03899871523892411,
      "grad_norm": 0.4537885785102844,
      "learning_rate": 4.8052136433337486e-06,
      "loss": 0.7194,
      "step": 941
    },
    {
      "epoch": 0.03904015914459779,
      "grad_norm": 0.43595224618911743,
      "learning_rate": 4.80500642380538e-06,
      "loss": 0.7346,
      "step": 942
    },
    {
      "epoch": 0.03908160305027146,
      "grad_norm": 0.4676819443702698,
      "learning_rate": 4.804799204277011e-06,
      "loss": 0.7646,
      "step": 943
    },
    {
      "epoch": 0.03912304695594513,
      "grad_norm": 0.5424928665161133,
      "learning_rate": 4.8045919847486436e-06,
      "loss": 0.8176,
      "step": 944
    },
    {
      "epoch": 0.0391644908616188,
      "grad_norm": 0.5223216414451599,
      "learning_rate": 4.804384765220275e-06,
      "loss": 0.8591,
      "step": 945
    },
    {
      "epoch": 0.03920593476729247,
      "grad_norm": 0.4513929784297943,
      "learning_rate": 4.804177545691906e-06,
      "loss": 0.7903,
      "step": 946
    },
    {
      "epoch": 0.03924737867296614,
      "grad_norm": 0.4726882576942444,
      "learning_rate": 4.803970326163538e-06,
      "loss": 0.8093,
      "step": 947
    },
    {
      "epoch": 0.03928882257863981,
      "grad_norm": 0.4442755877971649,
      "learning_rate": 4.80376310663517e-06,
      "loss": 0.822,
      "step": 948
    },
    {
      "epoch": 0.03933026648431348,
      "grad_norm": 0.48815596103668213,
      "learning_rate": 4.803555887106801e-06,
      "loss": 0.7595,
      "step": 949
    },
    {
      "epoch": 0.03937171038998715,
      "grad_norm": 0.4548647999763489,
      "learning_rate": 4.803348667578433e-06,
      "loss": 0.7561,
      "step": 950
    },
    {
      "epoch": 0.03941315429566082,
      "grad_norm": 0.48345446586608887,
      "learning_rate": 4.803141448050065e-06,
      "loss": 0.7722,
      "step": 951
    },
    {
      "epoch": 0.03945459820133449,
      "grad_norm": 0.553205132484436,
      "learning_rate": 4.802934228521696e-06,
      "loss": 0.8356,
      "step": 952
    },
    {
      "epoch": 0.03949604210700816,
      "grad_norm": 0.506557822227478,
      "learning_rate": 4.802727008993328e-06,
      "loss": 0.8259,
      "step": 953
    },
    {
      "epoch": 0.03953748601268184,
      "grad_norm": 0.46744129061698914,
      "learning_rate": 4.802519789464959e-06,
      "loss": 0.7456,
      "step": 954
    },
    {
      "epoch": 0.03957892991835551,
      "grad_norm": 0.47676709294319153,
      "learning_rate": 4.802312569936591e-06,
      "loss": 0.802,
      "step": 955
    },
    {
      "epoch": 0.03962037382402918,
      "grad_norm": 0.47377076745033264,
      "learning_rate": 4.802105350408223e-06,
      "loss": 0.8062,
      "step": 956
    },
    {
      "epoch": 0.03966181772970285,
      "grad_norm": 0.4776807427406311,
      "learning_rate": 4.801898130879855e-06,
      "loss": 0.7788,
      "step": 957
    },
    {
      "epoch": 0.03970326163537652,
      "grad_norm": 0.4366399049758911,
      "learning_rate": 4.8016909113514855e-06,
      "loss": 0.8247,
      "step": 958
    },
    {
      "epoch": 0.03974470554105019,
      "grad_norm": 0.4870122969150543,
      "learning_rate": 4.801483691823118e-06,
      "loss": 0.8557,
      "step": 959
    },
    {
      "epoch": 0.03978614944672386,
      "grad_norm": 0.4365279972553253,
      "learning_rate": 4.80127647229475e-06,
      "loss": 0.7952,
      "step": 960
    },
    {
      "epoch": 0.03982759335239753,
      "grad_norm": 0.4440559148788452,
      "learning_rate": 4.801069252766381e-06,
      "loss": 0.8191,
      "step": 961
    },
    {
      "epoch": 0.0398690372580712,
      "grad_norm": 0.45365670323371887,
      "learning_rate": 4.800862033238013e-06,
      "loss": 0.7848,
      "step": 962
    },
    {
      "epoch": 0.03991048116374487,
      "grad_norm": 0.43977826833724976,
      "learning_rate": 4.800654813709644e-06,
      "loss": 0.8115,
      "step": 963
    },
    {
      "epoch": 0.03995192506941854,
      "grad_norm": 0.4278276562690735,
      "learning_rate": 4.800447594181276e-06,
      "loss": 0.7417,
      "step": 964
    },
    {
      "epoch": 0.03999336897509221,
      "grad_norm": 0.45921963453292847,
      "learning_rate": 4.800240374652908e-06,
      "loss": 0.8319,
      "step": 965
    },
    {
      "epoch": 0.04003481288076589,
      "grad_norm": 0.4661577641963959,
      "learning_rate": 4.800033155124539e-06,
      "loss": 0.7759,
      "step": 966
    },
    {
      "epoch": 0.04007625678643956,
      "grad_norm": 0.4300996661186218,
      "learning_rate": 4.7998259355961705e-06,
      "loss": 0.7546,
      "step": 967
    },
    {
      "epoch": 0.04011770069211323,
      "grad_norm": 0.4361669719219208,
      "learning_rate": 4.799618716067803e-06,
      "loss": 0.7178,
      "step": 968
    },
    {
      "epoch": 0.0401591445977869,
      "grad_norm": 0.4488246738910675,
      "learning_rate": 4.799411496539434e-06,
      "loss": 0.7783,
      "step": 969
    },
    {
      "epoch": 0.04020058850346057,
      "grad_norm": 0.4444413483142853,
      "learning_rate": 4.7992042770110655e-06,
      "loss": 0.8276,
      "step": 970
    },
    {
      "epoch": 0.04024203240913424,
      "grad_norm": 0.4681646227836609,
      "learning_rate": 4.798997057482698e-06,
      "loss": 0.7952,
      "step": 971
    },
    {
      "epoch": 0.04028347631480791,
      "grad_norm": 0.4517576992511749,
      "learning_rate": 4.798789837954329e-06,
      "loss": 0.7209,
      "step": 972
    },
    {
      "epoch": 0.04032492022048158,
      "grad_norm": 0.4871593415737152,
      "learning_rate": 4.798582618425961e-06,
      "loss": 0.7661,
      "step": 973
    },
    {
      "epoch": 0.04036636412615525,
      "grad_norm": 0.46669653058052063,
      "learning_rate": 4.798375398897592e-06,
      "loss": 0.7603,
      "step": 974
    },
    {
      "epoch": 0.04040780803182892,
      "grad_norm": 0.4552239775657654,
      "learning_rate": 4.798168179369224e-06,
      "loss": 0.8652,
      "step": 975
    },
    {
      "epoch": 0.04044925193750259,
      "grad_norm": 0.44540682435035706,
      "learning_rate": 4.7979609598408555e-06,
      "loss": 0.7534,
      "step": 976
    },
    {
      "epoch": 0.04049069584317626,
      "grad_norm": 0.46907296776771545,
      "learning_rate": 4.797753740312488e-06,
      "loss": 0.7979,
      "step": 977
    },
    {
      "epoch": 0.04053213974884993,
      "grad_norm": 0.46654975414276123,
      "learning_rate": 4.797546520784119e-06,
      "loss": 0.7939,
      "step": 978
    },
    {
      "epoch": 0.040573583654523605,
      "grad_norm": 0.45213475823402405,
      "learning_rate": 4.7973393012557505e-06,
      "loss": 0.804,
      "step": 979
    },
    {
      "epoch": 0.040615027560197275,
      "grad_norm": 0.4843692183494568,
      "learning_rate": 4.797132081727383e-06,
      "loss": 0.8472,
      "step": 980
    },
    {
      "epoch": 0.040656471465870946,
      "grad_norm": 0.4351000189781189,
      "learning_rate": 4.796924862199014e-06,
      "loss": 0.7732,
      "step": 981
    },
    {
      "epoch": 0.040697915371544616,
      "grad_norm": 0.46643754839897156,
      "learning_rate": 4.7967176426706455e-06,
      "loss": 0.8257,
      "step": 982
    },
    {
      "epoch": 0.040739359277218286,
      "grad_norm": 0.47618910670280457,
      "learning_rate": 4.796510423142277e-06,
      "loss": 0.8284,
      "step": 983
    },
    {
      "epoch": 0.040780803182891956,
      "grad_norm": 0.4954012334346771,
      "learning_rate": 4.796303203613909e-06,
      "loss": 0.8013,
      "step": 984
    },
    {
      "epoch": 0.040822247088565626,
      "grad_norm": 0.5031293034553528,
      "learning_rate": 4.7960959840855405e-06,
      "loss": 0.8403,
      "step": 985
    },
    {
      "epoch": 0.040863690994239296,
      "grad_norm": 0.4458840787410736,
      "learning_rate": 4.795888764557172e-06,
      "loss": 0.7712,
      "step": 986
    },
    {
      "epoch": 0.040905134899912966,
      "grad_norm": 0.46724194288253784,
      "learning_rate": 4.795681545028804e-06,
      "loss": 0.7683,
      "step": 987
    },
    {
      "epoch": 0.040946578805586636,
      "grad_norm": 0.45755645632743835,
      "learning_rate": 4.7954743255004355e-06,
      "loss": 0.7847,
      "step": 988
    },
    {
      "epoch": 0.04098802271126031,
      "grad_norm": 0.4850478768348694,
      "learning_rate": 4.795267105972067e-06,
      "loss": 0.8101,
      "step": 989
    },
    {
      "epoch": 0.04102946661693398,
      "grad_norm": 0.48655596375465393,
      "learning_rate": 4.795059886443698e-06,
      "loss": 0.8138,
      "step": 990
    },
    {
      "epoch": 0.041070910522607654,
      "grad_norm": 0.4782755672931671,
      "learning_rate": 4.7948526669153306e-06,
      "loss": 0.7961,
      "step": 991
    },
    {
      "epoch": 0.041112354428281324,
      "grad_norm": 0.49097394943237305,
      "learning_rate": 4.794645447386962e-06,
      "loss": 0.7705,
      "step": 992
    },
    {
      "epoch": 0.041153798333954994,
      "grad_norm": 0.48887911438941956,
      "learning_rate": 4.794438227858594e-06,
      "loss": 0.8474,
      "step": 993
    },
    {
      "epoch": 0.041195242239628664,
      "grad_norm": 0.5012791156768799,
      "learning_rate": 4.7942310083302256e-06,
      "loss": 0.7839,
      "step": 994
    },
    {
      "epoch": 0.041236686145302334,
      "grad_norm": 0.4260593056678772,
      "learning_rate": 4.794023788801857e-06,
      "loss": 0.7913,
      "step": 995
    },
    {
      "epoch": 0.041278130050976004,
      "grad_norm": 0.47090914845466614,
      "learning_rate": 4.793816569273489e-06,
      "loss": 0.7939,
      "step": 996
    },
    {
      "epoch": 0.041319573956649674,
      "grad_norm": 0.43993470072746277,
      "learning_rate": 4.7936093497451206e-06,
      "loss": 0.7917,
      "step": 997
    },
    {
      "epoch": 0.041361017862323345,
      "grad_norm": 0.4615001082420349,
      "learning_rate": 4.793402130216752e-06,
      "loss": 0.7561,
      "step": 998
    },
    {
      "epoch": 0.041402461767997015,
      "grad_norm": 0.4298277199268341,
      "learning_rate": 4.793194910688383e-06,
      "loss": 0.8127,
      "step": 999
    },
    {
      "epoch": 0.041443905673670685,
      "grad_norm": 0.4404458999633789,
      "learning_rate": 4.7929876911600156e-06,
      "loss": 0.7771,
      "step": 1000
    },
    {
      "epoch": 0.041485349579344355,
      "grad_norm": 0.5292378664016724,
      "learning_rate": 4.792780471631647e-06,
      "loss": 0.8579,
      "step": 1001
    },
    {
      "epoch": 0.041526793485018025,
      "grad_norm": 0.4654186964035034,
      "learning_rate": 4.792573252103278e-06,
      "loss": 0.8076,
      "step": 1002
    },
    {
      "epoch": 0.0415682373906917,
      "grad_norm": 0.45650607347488403,
      "learning_rate": 4.7923660325749106e-06,
      "loss": 0.8188,
      "step": 1003
    },
    {
      "epoch": 0.04160968129636537,
      "grad_norm": 0.45740291476249695,
      "learning_rate": 4.792158813046542e-06,
      "loss": 0.8037,
      "step": 1004
    },
    {
      "epoch": 0.04165112520203904,
      "grad_norm": 0.44465282559394836,
      "learning_rate": 4.791951593518173e-06,
      "loss": 0.7495,
      "step": 1005
    },
    {
      "epoch": 0.04169256910771271,
      "grad_norm": 0.4487740099430084,
      "learning_rate": 4.791744373989805e-06,
      "loss": 0.7794,
      "step": 1006
    },
    {
      "epoch": 0.04173401301338638,
      "grad_norm": 0.4635974168777466,
      "learning_rate": 4.791537154461437e-06,
      "loss": 0.7729,
      "step": 1007
    },
    {
      "epoch": 0.04177545691906005,
      "grad_norm": 0.4493637979030609,
      "learning_rate": 4.791329934933068e-06,
      "loss": 0.7737,
      "step": 1008
    },
    {
      "epoch": 0.04181690082473372,
      "grad_norm": 0.4791603982448578,
      "learning_rate": 4.7911227154047006e-06,
      "loss": 0.7959,
      "step": 1009
    },
    {
      "epoch": 0.04185834473040739,
      "grad_norm": 0.4576243758201599,
      "learning_rate": 4.790915495876331e-06,
      "loss": 0.8398,
      "step": 1010
    },
    {
      "epoch": 0.04189978863608106,
      "grad_norm": 0.47298121452331543,
      "learning_rate": 4.790708276347963e-06,
      "loss": 0.7998,
      "step": 1011
    },
    {
      "epoch": 0.04194123254175473,
      "grad_norm": 0.48323437571525574,
      "learning_rate": 4.790501056819596e-06,
      "loss": 0.8804,
      "step": 1012
    },
    {
      "epoch": 0.0419826764474284,
      "grad_norm": 0.474826455116272,
      "learning_rate": 4.790293837291227e-06,
      "loss": 0.8137,
      "step": 1013
    },
    {
      "epoch": 0.042024120353102073,
      "grad_norm": 0.4563380777835846,
      "learning_rate": 4.790086617762858e-06,
      "loss": 0.7656,
      "step": 1014
    },
    {
      "epoch": 0.042065564258775744,
      "grad_norm": 0.47896111011505127,
      "learning_rate": 4.78987939823449e-06,
      "loss": 0.8835,
      "step": 1015
    },
    {
      "epoch": 0.04210700816444942,
      "grad_norm": 0.4392525851726532,
      "learning_rate": 4.789672178706122e-06,
      "loss": 0.7024,
      "step": 1016
    },
    {
      "epoch": 0.04214845207012309,
      "grad_norm": 0.5214576721191406,
      "learning_rate": 4.789464959177753e-06,
      "loss": 0.7979,
      "step": 1017
    },
    {
      "epoch": 0.04218989597579676,
      "grad_norm": 0.4837241768836975,
      "learning_rate": 4.789257739649385e-06,
      "loss": 0.7976,
      "step": 1018
    },
    {
      "epoch": 0.04223133988147043,
      "grad_norm": 0.4692282974720001,
      "learning_rate": 4.789050520121016e-06,
      "loss": 0.8574,
      "step": 1019
    },
    {
      "epoch": 0.0422727837871441,
      "grad_norm": 0.5051727890968323,
      "learning_rate": 4.788843300592648e-06,
      "loss": 0.764,
      "step": 1020
    },
    {
      "epoch": 0.04231422769281777,
      "grad_norm": 0.4650651216506958,
      "learning_rate": 4.78863608106428e-06,
      "loss": 0.803,
      "step": 1021
    },
    {
      "epoch": 0.04235567159849144,
      "grad_norm": 0.4585110545158386,
      "learning_rate": 4.788428861535911e-06,
      "loss": 0.8154,
      "step": 1022
    },
    {
      "epoch": 0.04239711550416511,
      "grad_norm": 0.46456143260002136,
      "learning_rate": 4.788221642007543e-06,
      "loss": 0.8394,
      "step": 1023
    },
    {
      "epoch": 0.04243855940983878,
      "grad_norm": 0.470569372177124,
      "learning_rate": 4.788014422479175e-06,
      "loss": 0.7769,
      "step": 1024
    },
    {
      "epoch": 0.04248000331551245,
      "grad_norm": 0.47860991954803467,
      "learning_rate": 4.787807202950807e-06,
      "loss": 0.8289,
      "step": 1025
    },
    {
      "epoch": 0.04252144722118612,
      "grad_norm": 0.4554358720779419,
      "learning_rate": 4.7875999834224375e-06,
      "loss": 0.7986,
      "step": 1026
    },
    {
      "epoch": 0.04256289112685979,
      "grad_norm": 0.45716437697410583,
      "learning_rate": 4.78739276389407e-06,
      "loss": 0.7822,
      "step": 1027
    },
    {
      "epoch": 0.04260433503253347,
      "grad_norm": 0.4630057215690613,
      "learning_rate": 4.787185544365701e-06,
      "loss": 0.7949,
      "step": 1028
    },
    {
      "epoch": 0.04264577893820714,
      "grad_norm": 0.4765913188457489,
      "learning_rate": 4.786978324837333e-06,
      "loss": 0.8516,
      "step": 1029
    },
    {
      "epoch": 0.04268722284388081,
      "grad_norm": 0.4592381417751312,
      "learning_rate": 4.786771105308965e-06,
      "loss": 0.7993,
      "step": 1030
    },
    {
      "epoch": 0.04272866674955448,
      "grad_norm": 0.5326821208000183,
      "learning_rate": 4.786563885780596e-06,
      "loss": 0.8147,
      "step": 1031
    },
    {
      "epoch": 0.04277011065522815,
      "grad_norm": 0.43418946862220764,
      "learning_rate": 4.786356666252228e-06,
      "loss": 0.7595,
      "step": 1032
    },
    {
      "epoch": 0.04281155456090182,
      "grad_norm": 0.4404500126838684,
      "learning_rate": 4.78614944672386e-06,
      "loss": 0.8386,
      "step": 1033
    },
    {
      "epoch": 0.04285299846657549,
      "grad_norm": 0.4906412363052368,
      "learning_rate": 4.785942227195491e-06,
      "loss": 0.7761,
      "step": 1034
    },
    {
      "epoch": 0.04289444237224916,
      "grad_norm": 0.4952963888645172,
      "learning_rate": 4.7857350076671225e-06,
      "loss": 0.7893,
      "step": 1035
    },
    {
      "epoch": 0.04293588627792283,
      "grad_norm": 0.436569482088089,
      "learning_rate": 4.785527788138755e-06,
      "loss": 0.7563,
      "step": 1036
    },
    {
      "epoch": 0.0429773301835965,
      "grad_norm": 0.5171957015991211,
      "learning_rate": 4.785320568610386e-06,
      "loss": 0.8477,
      "step": 1037
    },
    {
      "epoch": 0.04301877408927017,
      "grad_norm": 0.4241662621498108,
      "learning_rate": 4.7851133490820175e-06,
      "loss": 0.738,
      "step": 1038
    },
    {
      "epoch": 0.04306021799494384,
      "grad_norm": 0.4370529055595398,
      "learning_rate": 4.78490612955365e-06,
      "loss": 0.7903,
      "step": 1039
    },
    {
      "epoch": 0.04310166190061752,
      "grad_norm": 0.4420294761657715,
      "learning_rate": 4.784698910025281e-06,
      "loss": 0.7524,
      "step": 1040
    },
    {
      "epoch": 0.04314310580629119,
      "grad_norm": 0.46341052651405334,
      "learning_rate": 4.784491690496913e-06,
      "loss": 0.7883,
      "step": 1041
    },
    {
      "epoch": 0.04318454971196486,
      "grad_norm": 0.43772614002227783,
      "learning_rate": 4.784284470968544e-06,
      "loss": 0.7148,
      "step": 1042
    },
    {
      "epoch": 0.04322599361763853,
      "grad_norm": 0.4657709002494812,
      "learning_rate": 4.784077251440176e-06,
      "loss": 0.8296,
      "step": 1043
    },
    {
      "epoch": 0.0432674375233122,
      "grad_norm": 0.41048768162727356,
      "learning_rate": 4.7838700319118075e-06,
      "loss": 0.7358,
      "step": 1044
    },
    {
      "epoch": 0.04330888142898587,
      "grad_norm": 0.48912936449050903,
      "learning_rate": 4.78366281238344e-06,
      "loss": 0.8477,
      "step": 1045
    },
    {
      "epoch": 0.04335032533465954,
      "grad_norm": 0.515285313129425,
      "learning_rate": 4.783455592855071e-06,
      "loss": 0.7952,
      "step": 1046
    },
    {
      "epoch": 0.04339176924033321,
      "grad_norm": 0.4189876616001129,
      "learning_rate": 4.7832483733267025e-06,
      "loss": 0.7969,
      "step": 1047
    },
    {
      "epoch": 0.04343321314600688,
      "grad_norm": 0.4508560299873352,
      "learning_rate": 4.783041153798335e-06,
      "loss": 0.7974,
      "step": 1048
    },
    {
      "epoch": 0.04347465705168055,
      "grad_norm": 0.47887948155403137,
      "learning_rate": 4.782833934269966e-06,
      "loss": 0.7988,
      "step": 1049
    },
    {
      "epoch": 0.04351610095735422,
      "grad_norm": 0.44417762756347656,
      "learning_rate": 4.7826267147415976e-06,
      "loss": 0.7744,
      "step": 1050
    },
    {
      "epoch": 0.04355754486302789,
      "grad_norm": 0.4393226206302643,
      "learning_rate": 4.782419495213229e-06,
      "loss": 0.7338,
      "step": 1051
    },
    {
      "epoch": 0.04359898876870156,
      "grad_norm": 0.46549293398857117,
      "learning_rate": 4.782212275684861e-06,
      "loss": 0.8191,
      "step": 1052
    },
    {
      "epoch": 0.043640432674375236,
      "grad_norm": 0.45024704933166504,
      "learning_rate": 4.7820050561564926e-06,
      "loss": 0.7842,
      "step": 1053
    },
    {
      "epoch": 0.043681876580048906,
      "grad_norm": 0.45627233386039734,
      "learning_rate": 4.781797836628124e-06,
      "loss": 0.7979,
      "step": 1054
    },
    {
      "epoch": 0.043723320485722576,
      "grad_norm": 0.46247386932373047,
      "learning_rate": 4.781590617099756e-06,
      "loss": 0.8213,
      "step": 1055
    },
    {
      "epoch": 0.043764764391396246,
      "grad_norm": 0.5227290987968445,
      "learning_rate": 4.7813833975713876e-06,
      "loss": 0.8562,
      "step": 1056
    },
    {
      "epoch": 0.04380620829706992,
      "grad_norm": 0.4721840023994446,
      "learning_rate": 4.781176178043019e-06,
      "loss": 0.7395,
      "step": 1057
    },
    {
      "epoch": 0.04384765220274359,
      "grad_norm": 0.46023762226104736,
      "learning_rate": 4.78096895851465e-06,
      "loss": 0.8157,
      "step": 1058
    },
    {
      "epoch": 0.04388909610841726,
      "grad_norm": 0.4536685049533844,
      "learning_rate": 4.7807617389862826e-06,
      "loss": 0.7578,
      "step": 1059
    },
    {
      "epoch": 0.04393054001409093,
      "grad_norm": 0.43217524886131287,
      "learning_rate": 4.780554519457914e-06,
      "loss": 0.8279,
      "step": 1060
    },
    {
      "epoch": 0.0439719839197646,
      "grad_norm": 0.456112802028656,
      "learning_rate": 4.780347299929546e-06,
      "loss": 0.8474,
      "step": 1061
    },
    {
      "epoch": 0.04401342782543827,
      "grad_norm": 0.4474206864833832,
      "learning_rate": 4.7801400804011776e-06,
      "loss": 0.7642,
      "step": 1062
    },
    {
      "epoch": 0.04405487173111194,
      "grad_norm": 0.47980812191963196,
      "learning_rate": 4.779932860872809e-06,
      "loss": 0.8501,
      "step": 1063
    },
    {
      "epoch": 0.04409631563678561,
      "grad_norm": 0.4759003818035126,
      "learning_rate": 4.779725641344441e-06,
      "loss": 0.7739,
      "step": 1064
    },
    {
      "epoch": 0.044137759542459284,
      "grad_norm": 0.43127790093421936,
      "learning_rate": 4.7795184218160726e-06,
      "loss": 0.7468,
      "step": 1065
    },
    {
      "epoch": 0.044179203448132955,
      "grad_norm": 0.45588958263397217,
      "learning_rate": 4.779311202287704e-06,
      "loss": 0.7634,
      "step": 1066
    },
    {
      "epoch": 0.044220647353806625,
      "grad_norm": 0.44599223136901855,
      "learning_rate": 4.779103982759335e-06,
      "loss": 0.7515,
      "step": 1067
    },
    {
      "epoch": 0.044262091259480295,
      "grad_norm": 0.4741554260253906,
      "learning_rate": 4.7788967632309676e-06,
      "loss": 0.7725,
      "step": 1068
    },
    {
      "epoch": 0.044303535165153965,
      "grad_norm": 0.4776347577571869,
      "learning_rate": 4.778689543702599e-06,
      "loss": 0.7939,
      "step": 1069
    },
    {
      "epoch": 0.044344979070827635,
      "grad_norm": 0.4661857485771179,
      "learning_rate": 4.77848232417423e-06,
      "loss": 0.8621,
      "step": 1070
    },
    {
      "epoch": 0.044386422976501305,
      "grad_norm": 0.48425886034965515,
      "learning_rate": 4.778275104645862e-06,
      "loss": 0.7522,
      "step": 1071
    },
    {
      "epoch": 0.044427866882174975,
      "grad_norm": 0.48456665873527527,
      "learning_rate": 4.778067885117494e-06,
      "loss": 0.8384,
      "step": 1072
    },
    {
      "epoch": 0.044469310787848645,
      "grad_norm": 0.47244054079055786,
      "learning_rate": 4.777860665589125e-06,
      "loss": 0.7417,
      "step": 1073
    },
    {
      "epoch": 0.044510754693522316,
      "grad_norm": 0.42493709921836853,
      "learning_rate": 4.777653446060757e-06,
      "loss": 0.7424,
      "step": 1074
    },
    {
      "epoch": 0.044552198599195986,
      "grad_norm": 0.4758439064025879,
      "learning_rate": 4.777446226532389e-06,
      "loss": 0.8054,
      "step": 1075
    },
    {
      "epoch": 0.044593642504869656,
      "grad_norm": 0.489058256149292,
      "learning_rate": 4.77723900700402e-06,
      "loss": 0.8093,
      "step": 1076
    },
    {
      "epoch": 0.04463508641054333,
      "grad_norm": 0.4572688639163971,
      "learning_rate": 4.777031787475653e-06,
      "loss": 0.7811,
      "step": 1077
    },
    {
      "epoch": 0.044676530316217,
      "grad_norm": 0.4598316252231598,
      "learning_rate": 4.776824567947284e-06,
      "loss": 0.7764,
      "step": 1078
    },
    {
      "epoch": 0.04471797422189067,
      "grad_norm": 0.48176220059394836,
      "learning_rate": 4.776617348418915e-06,
      "loss": 0.8389,
      "step": 1079
    },
    {
      "epoch": 0.04475941812756434,
      "grad_norm": 0.4744527339935303,
      "learning_rate": 4.776410128890547e-06,
      "loss": 0.8042,
      "step": 1080
    },
    {
      "epoch": 0.04480086203323801,
      "grad_norm": 0.49460116028785706,
      "learning_rate": 4.776202909362179e-06,
      "loss": 0.8757,
      "step": 1081
    },
    {
      "epoch": 0.044842305938911683,
      "grad_norm": 0.44487905502319336,
      "learning_rate": 4.77599568983381e-06,
      "loss": 0.7737,
      "step": 1082
    },
    {
      "epoch": 0.044883749844585354,
      "grad_norm": 0.48597803711891174,
      "learning_rate": 4.775788470305442e-06,
      "loss": 0.7971,
      "step": 1083
    },
    {
      "epoch": 0.044925193750259024,
      "grad_norm": 0.4437675476074219,
      "learning_rate": 4.775581250777074e-06,
      "loss": 0.8245,
      "step": 1084
    },
    {
      "epoch": 0.044966637655932694,
      "grad_norm": 0.45028436183929443,
      "learning_rate": 4.775374031248705e-06,
      "loss": 0.7422,
      "step": 1085
    },
    {
      "epoch": 0.045008081561606364,
      "grad_norm": 0.46277710795402527,
      "learning_rate": 4.775166811720337e-06,
      "loss": 0.8352,
      "step": 1086
    },
    {
      "epoch": 0.045049525467280034,
      "grad_norm": 0.5086259841918945,
      "learning_rate": 4.774959592191968e-06,
      "loss": 0.8068,
      "step": 1087
    },
    {
      "epoch": 0.045090969372953704,
      "grad_norm": 0.5544159412384033,
      "learning_rate": 4.7747523726636e-06,
      "loss": 0.8794,
      "step": 1088
    },
    {
      "epoch": 0.04513241327862738,
      "grad_norm": 0.4555202126502991,
      "learning_rate": 4.774545153135232e-06,
      "loss": 0.8228,
      "step": 1089
    },
    {
      "epoch": 0.04517385718430105,
      "grad_norm": 0.4569913148880005,
      "learning_rate": 4.774337933606863e-06,
      "loss": 0.8098,
      "step": 1090
    },
    {
      "epoch": 0.04521530108997472,
      "grad_norm": 0.425127774477005,
      "learning_rate": 4.774130714078495e-06,
      "loss": 0.7759,
      "step": 1091
    },
    {
      "epoch": 0.04525674499564839,
      "grad_norm": 0.42579367756843567,
      "learning_rate": 4.773923494550127e-06,
      "loss": 0.7352,
      "step": 1092
    },
    {
      "epoch": 0.04529818890132206,
      "grad_norm": 0.4868384301662445,
      "learning_rate": 4.773716275021759e-06,
      "loss": 0.7891,
      "step": 1093
    },
    {
      "epoch": 0.04533963280699573,
      "grad_norm": 0.4713551104068756,
      "learning_rate": 4.7735090554933895e-06,
      "loss": 0.7776,
      "step": 1094
    },
    {
      "epoch": 0.0453810767126694,
      "grad_norm": 0.4447707533836365,
      "learning_rate": 4.773301835965022e-06,
      "loss": 0.7825,
      "step": 1095
    },
    {
      "epoch": 0.04542252061834307,
      "grad_norm": 0.4642535150051117,
      "learning_rate": 4.773094616436653e-06,
      "loss": 0.7607,
      "step": 1096
    },
    {
      "epoch": 0.04546396452401674,
      "grad_norm": 0.45153817534446716,
      "learning_rate": 4.772887396908285e-06,
      "loss": 0.8589,
      "step": 1097
    },
    {
      "epoch": 0.04550540842969041,
      "grad_norm": 0.4268827736377716,
      "learning_rate": 4.772680177379917e-06,
      "loss": 0.7378,
      "step": 1098
    },
    {
      "epoch": 0.04554685233536408,
      "grad_norm": 0.4873831868171692,
      "learning_rate": 4.772472957851548e-06,
      "loss": 0.8479,
      "step": 1099
    },
    {
      "epoch": 0.04558829624103775,
      "grad_norm": 0.4293820261955261,
      "learning_rate": 4.77226573832318e-06,
      "loss": 0.7791,
      "step": 1100
    },
    {
      "epoch": 0.04562974014671142,
      "grad_norm": 0.5006012320518494,
      "learning_rate": 4.772058518794812e-06,
      "loss": 0.7483,
      "step": 1101
    },
    {
      "epoch": 0.0456711840523851,
      "grad_norm": 0.4601713716983795,
      "learning_rate": 4.771851299266443e-06,
      "loss": 0.7852,
      "step": 1102
    },
    {
      "epoch": 0.04571262795805877,
      "grad_norm": 0.4780215322971344,
      "learning_rate": 4.7716440797380745e-06,
      "loss": 0.7512,
      "step": 1103
    },
    {
      "epoch": 0.04575407186373244,
      "grad_norm": 0.46182432770729065,
      "learning_rate": 4.771436860209707e-06,
      "loss": 0.8018,
      "step": 1104
    },
    {
      "epoch": 0.04579551576940611,
      "grad_norm": 0.5341752171516418,
      "learning_rate": 4.771229640681338e-06,
      "loss": 0.7598,
      "step": 1105
    },
    {
      "epoch": 0.04583695967507978,
      "grad_norm": 0.4713796079158783,
      "learning_rate": 4.7710224211529695e-06,
      "loss": 0.821,
      "step": 1106
    },
    {
      "epoch": 0.04587840358075345,
      "grad_norm": 0.5094240307807922,
      "learning_rate": 4.770815201624602e-06,
      "loss": 0.8196,
      "step": 1107
    },
    {
      "epoch": 0.04591984748642712,
      "grad_norm": 0.48558828234672546,
      "learning_rate": 4.770607982096233e-06,
      "loss": 0.7471,
      "step": 1108
    },
    {
      "epoch": 0.04596129139210079,
      "grad_norm": 0.4777097702026367,
      "learning_rate": 4.770400762567865e-06,
      "loss": 0.7449,
      "step": 1109
    },
    {
      "epoch": 0.04600273529777446,
      "grad_norm": 0.45582717657089233,
      "learning_rate": 4.770193543039496e-06,
      "loss": 0.7649,
      "step": 1110
    },
    {
      "epoch": 0.04604417920344813,
      "grad_norm": 0.4614855945110321,
      "learning_rate": 4.769986323511128e-06,
      "loss": 0.7795,
      "step": 1111
    },
    {
      "epoch": 0.0460856231091218,
      "grad_norm": 0.4478304386138916,
      "learning_rate": 4.7697791039827596e-06,
      "loss": 0.8113,
      "step": 1112
    },
    {
      "epoch": 0.04612706701479547,
      "grad_norm": 0.48237845301628113,
      "learning_rate": 4.769571884454392e-06,
      "loss": 0.8135,
      "step": 1113
    },
    {
      "epoch": 0.04616851092046915,
      "grad_norm": 0.5383651852607727,
      "learning_rate": 4.769364664926023e-06,
      "loss": 0.8545,
      "step": 1114
    },
    {
      "epoch": 0.04620995482614282,
      "grad_norm": 0.4325060248374939,
      "learning_rate": 4.7691574453976546e-06,
      "loss": 0.8008,
      "step": 1115
    },
    {
      "epoch": 0.04625139873181649,
      "grad_norm": 0.4721400737762451,
      "learning_rate": 4.768950225869287e-06,
      "loss": 0.8584,
      "step": 1116
    },
    {
      "epoch": 0.04629284263749016,
      "grad_norm": 0.4709514379501343,
      "learning_rate": 4.768743006340918e-06,
      "loss": 0.8298,
      "step": 1117
    },
    {
      "epoch": 0.04633428654316383,
      "grad_norm": 0.501682698726654,
      "learning_rate": 4.7685357868125496e-06,
      "loss": 0.8245,
      "step": 1118
    },
    {
      "epoch": 0.0463757304488375,
      "grad_norm": 0.4643377661705017,
      "learning_rate": 4.768328567284181e-06,
      "loss": 0.7642,
      "step": 1119
    },
    {
      "epoch": 0.04641717435451117,
      "grad_norm": 0.48200082778930664,
      "learning_rate": 4.768121347755813e-06,
      "loss": 0.7991,
      "step": 1120
    },
    {
      "epoch": 0.04645861826018484,
      "grad_norm": 0.4852246046066284,
      "learning_rate": 4.7679141282274446e-06,
      "loss": 0.8018,
      "step": 1121
    },
    {
      "epoch": 0.04650006216585851,
      "grad_norm": 0.45635098218917847,
      "learning_rate": 4.767706908699076e-06,
      "loss": 0.748,
      "step": 1122
    },
    {
      "epoch": 0.04654150607153218,
      "grad_norm": 0.4639163315296173,
      "learning_rate": 4.767499689170707e-06,
      "loss": 0.7729,
      "step": 1123
    },
    {
      "epoch": 0.04658294997720585,
      "grad_norm": 0.4940907657146454,
      "learning_rate": 4.7672924696423396e-06,
      "loss": 0.813,
      "step": 1124
    },
    {
      "epoch": 0.04662439388287952,
      "grad_norm": 0.4561811089515686,
      "learning_rate": 4.767085250113972e-06,
      "loss": 0.7661,
      "step": 1125
    },
    {
      "epoch": 0.0466658377885532,
      "grad_norm": 0.41685640811920166,
      "learning_rate": 4.766878030585602e-06,
      "loss": 0.7219,
      "step": 1126
    },
    {
      "epoch": 0.04670728169422687,
      "grad_norm": 0.4310116767883301,
      "learning_rate": 4.7666708110572346e-06,
      "loss": 0.7629,
      "step": 1127
    },
    {
      "epoch": 0.04674872559990054,
      "grad_norm": 0.4604714810848236,
      "learning_rate": 4.766463591528866e-06,
      "loss": 0.7888,
      "step": 1128
    },
    {
      "epoch": 0.04679016950557421,
      "grad_norm": 0.5103821754455566,
      "learning_rate": 4.766256372000498e-06,
      "loss": 0.7864,
      "step": 1129
    },
    {
      "epoch": 0.04683161341124788,
      "grad_norm": 0.5306897759437561,
      "learning_rate": 4.76604915247213e-06,
      "loss": 0.8103,
      "step": 1130
    },
    {
      "epoch": 0.04687305731692155,
      "grad_norm": 0.4589296579360962,
      "learning_rate": 4.765841932943761e-06,
      "loss": 0.7878,
      "step": 1131
    },
    {
      "epoch": 0.04691450122259522,
      "grad_norm": 0.4608319103717804,
      "learning_rate": 4.765634713415392e-06,
      "loss": 0.7869,
      "step": 1132
    },
    {
      "epoch": 0.04695594512826889,
      "grad_norm": 0.4795214533805847,
      "learning_rate": 4.765427493887025e-06,
      "loss": 0.7703,
      "step": 1133
    },
    {
      "epoch": 0.04699738903394256,
      "grad_norm": 0.4532718360424042,
      "learning_rate": 4.765220274358656e-06,
      "loss": 0.7974,
      "step": 1134
    },
    {
      "epoch": 0.04703883293961623,
      "grad_norm": 0.47184133529663086,
      "learning_rate": 4.765013054830287e-06,
      "loss": 0.7329,
      "step": 1135
    },
    {
      "epoch": 0.0470802768452899,
      "grad_norm": 0.4561813771724701,
      "learning_rate": 4.76480583530192e-06,
      "loss": 0.8391,
      "step": 1136
    },
    {
      "epoch": 0.04712172075096357,
      "grad_norm": 0.48163068294525146,
      "learning_rate": 4.764598615773551e-06,
      "loss": 0.8845,
      "step": 1137
    },
    {
      "epoch": 0.04716316465663724,
      "grad_norm": 0.5012065768241882,
      "learning_rate": 4.764391396245182e-06,
      "loss": 0.7657,
      "step": 1138
    },
    {
      "epoch": 0.047204608562310915,
      "grad_norm": 0.4675280749797821,
      "learning_rate": 4.764184176716814e-06,
      "loss": 0.7937,
      "step": 1139
    },
    {
      "epoch": 0.047246052467984585,
      "grad_norm": 0.4500986337661743,
      "learning_rate": 4.763976957188446e-06,
      "loss": 0.7605,
      "step": 1140
    },
    {
      "epoch": 0.047287496373658255,
      "grad_norm": 0.47017499804496765,
      "learning_rate": 4.763769737660077e-06,
      "loss": 0.7668,
      "step": 1141
    },
    {
      "epoch": 0.047328940279331926,
      "grad_norm": 0.4638185203075409,
      "learning_rate": 4.763562518131709e-06,
      "loss": 0.7778,
      "step": 1142
    },
    {
      "epoch": 0.047370384185005596,
      "grad_norm": 0.425403356552124,
      "learning_rate": 4.763355298603341e-06,
      "loss": 0.7761,
      "step": 1143
    },
    {
      "epoch": 0.047411828090679266,
      "grad_norm": 0.48976048827171326,
      "learning_rate": 4.763148079074972e-06,
      "loss": 0.7825,
      "step": 1144
    },
    {
      "epoch": 0.047453271996352936,
      "grad_norm": 0.4575006067752838,
      "learning_rate": 4.762940859546605e-06,
      "loss": 0.7527,
      "step": 1145
    },
    {
      "epoch": 0.047494715902026606,
      "grad_norm": 0.47396737337112427,
      "learning_rate": 4.762733640018236e-06,
      "loss": 0.7181,
      "step": 1146
    },
    {
      "epoch": 0.047536159807700276,
      "grad_norm": 0.4640709459781647,
      "learning_rate": 4.762526420489867e-06,
      "loss": 0.8,
      "step": 1147
    },
    {
      "epoch": 0.047577603713373946,
      "grad_norm": 0.4709440767765045,
      "learning_rate": 4.762319200961499e-06,
      "loss": 0.7522,
      "step": 1148
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 0.47328272461891174,
      "learning_rate": 4.762111981433131e-06,
      "loss": 0.7909,
      "step": 1149
    },
    {
      "epoch": 0.047660491524721287,
      "grad_norm": 0.4334677457809448,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.7891,
      "step": 1150
    },
    {
      "epoch": 0.047701935430394964,
      "grad_norm": 0.4913560748100281,
      "learning_rate": 4.761697542376394e-06,
      "loss": 0.8271,
      "step": 1151
    },
    {
      "epoch": 0.047743379336068634,
      "grad_norm": 0.49631425738334656,
      "learning_rate": 4.761490322848026e-06,
      "loss": 0.8167,
      "step": 1152
    },
    {
      "epoch": 0.047784823241742304,
      "grad_norm": 0.47891414165496826,
      "learning_rate": 4.761283103319657e-06,
      "loss": 0.8074,
      "step": 1153
    },
    {
      "epoch": 0.047826267147415974,
      "grad_norm": 0.5080652832984924,
      "learning_rate": 4.761075883791289e-06,
      "loss": 0.8113,
      "step": 1154
    },
    {
      "epoch": 0.047867711053089644,
      "grad_norm": 0.44767627120018005,
      "learning_rate": 4.76086866426292e-06,
      "loss": 0.7991,
      "step": 1155
    },
    {
      "epoch": 0.047909154958763314,
      "grad_norm": 0.4517739713191986,
      "learning_rate": 4.760661444734552e-06,
      "loss": 0.7751,
      "step": 1156
    },
    {
      "epoch": 0.047950598864436984,
      "grad_norm": 0.4694949686527252,
      "learning_rate": 4.760454225206184e-06,
      "loss": 0.7852,
      "step": 1157
    },
    {
      "epoch": 0.047992042770110654,
      "grad_norm": 0.46059003472328186,
      "learning_rate": 4.760247005677815e-06,
      "loss": 0.7761,
      "step": 1158
    },
    {
      "epoch": 0.048033486675784325,
      "grad_norm": 0.43597954511642456,
      "learning_rate": 4.7600397861494465e-06,
      "loss": 0.7859,
      "step": 1159
    },
    {
      "epoch": 0.048074930581457995,
      "grad_norm": 0.45111414790153503,
      "learning_rate": 4.759832566621079e-06,
      "loss": 0.7434,
      "step": 1160
    },
    {
      "epoch": 0.048116374487131665,
      "grad_norm": 0.4659668207168579,
      "learning_rate": 4.759625347092711e-06,
      "loss": 0.8137,
      "step": 1161
    },
    {
      "epoch": 0.048157818392805335,
      "grad_norm": 0.4754352271556854,
      "learning_rate": 4.7594181275643415e-06,
      "loss": 0.8413,
      "step": 1162
    },
    {
      "epoch": 0.04819926229847901,
      "grad_norm": 0.4473358392715454,
      "learning_rate": 4.759210908035974e-06,
      "loss": 0.8242,
      "step": 1163
    },
    {
      "epoch": 0.04824070620415268,
      "grad_norm": 0.4281471073627472,
      "learning_rate": 4.759003688507605e-06,
      "loss": 0.7441,
      "step": 1164
    },
    {
      "epoch": 0.04828215010982635,
      "grad_norm": 0.480766624212265,
      "learning_rate": 4.758796468979237e-06,
      "loss": 0.8167,
      "step": 1165
    },
    {
      "epoch": 0.04832359401550002,
      "grad_norm": 0.4934529960155487,
      "learning_rate": 4.758589249450869e-06,
      "loss": 0.7783,
      "step": 1166
    },
    {
      "epoch": 0.04836503792117369,
      "grad_norm": 0.5014682412147522,
      "learning_rate": 4.7583820299225e-06,
      "loss": 0.8022,
      "step": 1167
    },
    {
      "epoch": 0.04840648182684736,
      "grad_norm": 0.45076847076416016,
      "learning_rate": 4.758174810394132e-06,
      "loss": 0.7622,
      "step": 1168
    },
    {
      "epoch": 0.04844792573252103,
      "grad_norm": 0.486648827791214,
      "learning_rate": 4.757967590865764e-06,
      "loss": 0.7939,
      "step": 1169
    },
    {
      "epoch": 0.0484893696381947,
      "grad_norm": 0.446209192276001,
      "learning_rate": 4.757760371337395e-06,
      "loss": 0.7676,
      "step": 1170
    },
    {
      "epoch": 0.04853081354386837,
      "grad_norm": 0.4450940787792206,
      "learning_rate": 4.7575531518090266e-06,
      "loss": 0.7261,
      "step": 1171
    },
    {
      "epoch": 0.04857225744954204,
      "grad_norm": 0.4777759313583374,
      "learning_rate": 4.757345932280659e-06,
      "loss": 0.7751,
      "step": 1172
    },
    {
      "epoch": 0.04861370135521571,
      "grad_norm": 0.46364954113960266,
      "learning_rate": 4.75713871275229e-06,
      "loss": 0.8276,
      "step": 1173
    },
    {
      "epoch": 0.04865514526088938,
      "grad_norm": 0.4810357689857483,
      "learning_rate": 4.7569314932239216e-06,
      "loss": 0.8169,
      "step": 1174
    },
    {
      "epoch": 0.048696589166563053,
      "grad_norm": 0.4100513756275177,
      "learning_rate": 4.756724273695553e-06,
      "loss": 0.7263,
      "step": 1175
    },
    {
      "epoch": 0.04873803307223673,
      "grad_norm": 0.4895865321159363,
      "learning_rate": 4.756517054167185e-06,
      "loss": 0.7684,
      "step": 1176
    },
    {
      "epoch": 0.0487794769779104,
      "grad_norm": 0.43320727348327637,
      "learning_rate": 4.756309834638817e-06,
      "loss": 0.741,
      "step": 1177
    },
    {
      "epoch": 0.04882092088358407,
      "grad_norm": 0.4820486307144165,
      "learning_rate": 4.756102615110448e-06,
      "loss": 0.8364,
      "step": 1178
    },
    {
      "epoch": 0.04886236478925774,
      "grad_norm": 0.4768984317779541,
      "learning_rate": 4.75589539558208e-06,
      "loss": 0.7727,
      "step": 1179
    },
    {
      "epoch": 0.04890380869493141,
      "grad_norm": 0.4264664053916931,
      "learning_rate": 4.7556881760537116e-06,
      "loss": 0.8025,
      "step": 1180
    },
    {
      "epoch": 0.04894525260060508,
      "grad_norm": 0.43624719977378845,
      "learning_rate": 4.755480956525344e-06,
      "loss": 0.7637,
      "step": 1181
    },
    {
      "epoch": 0.04898669650627875,
      "grad_norm": 0.4732251763343811,
      "learning_rate": 4.755273736996975e-06,
      "loss": 0.8291,
      "step": 1182
    },
    {
      "epoch": 0.04902814041195242,
      "grad_norm": 0.485446035861969,
      "learning_rate": 4.7550665174686066e-06,
      "loss": 0.8235,
      "step": 1183
    },
    {
      "epoch": 0.04906958431762609,
      "grad_norm": 0.46888467669487,
      "learning_rate": 4.754859297940238e-06,
      "loss": 0.7854,
      "step": 1184
    },
    {
      "epoch": 0.04911102822329976,
      "grad_norm": 0.5527017116546631,
      "learning_rate": 4.75465207841187e-06,
      "loss": 0.8191,
      "step": 1185
    },
    {
      "epoch": 0.04915247212897343,
      "grad_norm": 0.4443214535713196,
      "learning_rate": 4.7544448588835016e-06,
      "loss": 0.7739,
      "step": 1186
    },
    {
      "epoch": 0.0491939160346471,
      "grad_norm": 0.505150556564331,
      "learning_rate": 4.754237639355133e-06,
      "loss": 0.7915,
      "step": 1187
    },
    {
      "epoch": 0.04923535994032078,
      "grad_norm": 0.4519720673561096,
      "learning_rate": 4.754030419826765e-06,
      "loss": 0.8313,
      "step": 1188
    },
    {
      "epoch": 0.04927680384599445,
      "grad_norm": 0.46611085534095764,
      "learning_rate": 4.753823200298397e-06,
      "loss": 0.7517,
      "step": 1189
    },
    {
      "epoch": 0.04931824775166812,
      "grad_norm": 0.45731350779533386,
      "learning_rate": 4.753615980770028e-06,
      "loss": 0.7847,
      "step": 1190
    },
    {
      "epoch": 0.04935969165734179,
      "grad_norm": 0.4220495820045471,
      "learning_rate": 4.753408761241659e-06,
      "loss": 0.8071,
      "step": 1191
    },
    {
      "epoch": 0.04940113556301546,
      "grad_norm": 0.44269758462905884,
      "learning_rate": 4.753201541713292e-06,
      "loss": 0.7178,
      "step": 1192
    },
    {
      "epoch": 0.04944257946868913,
      "grad_norm": 0.4705568552017212,
      "learning_rate": 4.752994322184923e-06,
      "loss": 0.7759,
      "step": 1193
    },
    {
      "epoch": 0.0494840233743628,
      "grad_norm": 0.4717589020729065,
      "learning_rate": 4.752787102656554e-06,
      "loss": 0.7947,
      "step": 1194
    },
    {
      "epoch": 0.04952546728003647,
      "grad_norm": 0.42914721369743347,
      "learning_rate": 4.752579883128187e-06,
      "loss": 0.7551,
      "step": 1195
    },
    {
      "epoch": 0.04956691118571014,
      "grad_norm": 0.45622196793556213,
      "learning_rate": 4.752372663599818e-06,
      "loss": 0.8096,
      "step": 1196
    },
    {
      "epoch": 0.04960835509138381,
      "grad_norm": 0.4357914924621582,
      "learning_rate": 4.75216544407145e-06,
      "loss": 0.7366,
      "step": 1197
    },
    {
      "epoch": 0.04964979899705748,
      "grad_norm": 0.45871320366859436,
      "learning_rate": 4.751958224543082e-06,
      "loss": 0.7644,
      "step": 1198
    },
    {
      "epoch": 0.04969124290273115,
      "grad_norm": 0.4517674446105957,
      "learning_rate": 4.751751005014713e-06,
      "loss": 0.7742,
      "step": 1199
    },
    {
      "epoch": 0.04973268680840483,
      "grad_norm": 0.4565979540348053,
      "learning_rate": 4.751543785486344e-06,
      "loss": 0.7725,
      "step": 1200
    },
    {
      "epoch": 0.0497741307140785,
      "grad_norm": 0.5017029643058777,
      "learning_rate": 4.751336565957977e-06,
      "loss": 0.7979,
      "step": 1201
    },
    {
      "epoch": 0.04981557461975217,
      "grad_norm": 0.4316957890987396,
      "learning_rate": 4.751129346429608e-06,
      "loss": 0.7595,
      "step": 1202
    },
    {
      "epoch": 0.04985701852542584,
      "grad_norm": 0.44598981738090515,
      "learning_rate": 4.750922126901239e-06,
      "loss": 0.8191,
      "step": 1203
    },
    {
      "epoch": 0.04989846243109951,
      "grad_norm": 0.4582918882369995,
      "learning_rate": 4.750714907372872e-06,
      "loss": 0.7502,
      "step": 1204
    },
    {
      "epoch": 0.04993990633677318,
      "grad_norm": 0.44995737075805664,
      "learning_rate": 4.750507687844503e-06,
      "loss": 0.8379,
      "step": 1205
    },
    {
      "epoch": 0.04998135024244685,
      "grad_norm": 0.4665464162826538,
      "learning_rate": 4.750300468316134e-06,
      "loss": 0.7756,
      "step": 1206
    },
    {
      "epoch": 0.05002279414812052,
      "grad_norm": 0.4967416226863861,
      "learning_rate": 4.750093248787766e-06,
      "loss": 0.7883,
      "step": 1207
    },
    {
      "epoch": 0.05006423805379419,
      "grad_norm": 0.4364146590232849,
      "learning_rate": 4.749886029259398e-06,
      "loss": 0.7576,
      "step": 1208
    },
    {
      "epoch": 0.05010568195946786,
      "grad_norm": 0.510164201259613,
      "learning_rate": 4.749678809731029e-06,
      "loss": 0.7432,
      "step": 1209
    },
    {
      "epoch": 0.05014712586514153,
      "grad_norm": 0.4519222676753998,
      "learning_rate": 4.749471590202661e-06,
      "loss": 0.7375,
      "step": 1210
    },
    {
      "epoch": 0.0501885697708152,
      "grad_norm": 0.49874672293663025,
      "learning_rate": 4.749264370674292e-06,
      "loss": 0.7998,
      "step": 1211
    },
    {
      "epoch": 0.050230013676488876,
      "grad_norm": 0.45865997672080994,
      "learning_rate": 4.749057151145924e-06,
      "loss": 0.7561,
      "step": 1212
    },
    {
      "epoch": 0.050271457582162546,
      "grad_norm": 0.4214809834957123,
      "learning_rate": 4.748849931617557e-06,
      "loss": 0.783,
      "step": 1213
    },
    {
      "epoch": 0.050312901487836216,
      "grad_norm": 0.4474555253982544,
      "learning_rate": 4.748642712089188e-06,
      "loss": 0.7888,
      "step": 1214
    },
    {
      "epoch": 0.050354345393509886,
      "grad_norm": 0.44800856709480286,
      "learning_rate": 4.748435492560819e-06,
      "loss": 0.7622,
      "step": 1215
    },
    {
      "epoch": 0.050395789299183556,
      "grad_norm": 0.4633291959762573,
      "learning_rate": 4.748228273032451e-06,
      "loss": 0.7732,
      "step": 1216
    },
    {
      "epoch": 0.050437233204857226,
      "grad_norm": 0.4754111170768738,
      "learning_rate": 4.748021053504083e-06,
      "loss": 0.7993,
      "step": 1217
    },
    {
      "epoch": 0.050478677110530897,
      "grad_norm": 0.4533904790878296,
      "learning_rate": 4.747813833975714e-06,
      "loss": 0.7434,
      "step": 1218
    },
    {
      "epoch": 0.05052012101620457,
      "grad_norm": 0.44537970423698425,
      "learning_rate": 4.747606614447346e-06,
      "loss": 0.7986,
      "step": 1219
    },
    {
      "epoch": 0.05056156492187824,
      "grad_norm": 0.48878997564315796,
      "learning_rate": 4.747399394918977e-06,
      "loss": 0.7981,
      "step": 1220
    },
    {
      "epoch": 0.05060300882755191,
      "grad_norm": 0.48571038246154785,
      "learning_rate": 4.747192175390609e-06,
      "loss": 0.8184,
      "step": 1221
    },
    {
      "epoch": 0.05064445273322558,
      "grad_norm": 0.47738444805145264,
      "learning_rate": 4.746984955862241e-06,
      "loss": 0.8713,
      "step": 1222
    },
    {
      "epoch": 0.05068589663889925,
      "grad_norm": 0.5319210290908813,
      "learning_rate": 4.746777736333872e-06,
      "loss": 0.8184,
      "step": 1223
    },
    {
      "epoch": 0.05072734054457292,
      "grad_norm": 0.4766693115234375,
      "learning_rate": 4.746570516805504e-06,
      "loss": 0.8274,
      "step": 1224
    },
    {
      "epoch": 0.050768784450246594,
      "grad_norm": 0.4183281660079956,
      "learning_rate": 4.746363297277136e-06,
      "loss": 0.7517,
      "step": 1225
    },
    {
      "epoch": 0.050810228355920264,
      "grad_norm": 0.4512433111667633,
      "learning_rate": 4.746156077748767e-06,
      "loss": 0.7671,
      "step": 1226
    },
    {
      "epoch": 0.050851672261593935,
      "grad_norm": 0.43581053614616394,
      "learning_rate": 4.7459488582203986e-06,
      "loss": 0.7961,
      "step": 1227
    },
    {
      "epoch": 0.050893116167267605,
      "grad_norm": 0.47661325335502625,
      "learning_rate": 4.745741638692031e-06,
      "loss": 0.8318,
      "step": 1228
    },
    {
      "epoch": 0.050934560072941275,
      "grad_norm": 0.4405752718448639,
      "learning_rate": 4.745534419163663e-06,
      "loss": 0.7595,
      "step": 1229
    },
    {
      "epoch": 0.050976003978614945,
      "grad_norm": 0.4849138855934143,
      "learning_rate": 4.7453271996352936e-06,
      "loss": 0.7538,
      "step": 1230
    },
    {
      "epoch": 0.051017447884288615,
      "grad_norm": 0.4781374931335449,
      "learning_rate": 4.745119980106926e-06,
      "loss": 0.7668,
      "step": 1231
    },
    {
      "epoch": 0.051058891789962285,
      "grad_norm": 0.452385812997818,
      "learning_rate": 4.744912760578557e-06,
      "loss": 0.8054,
      "step": 1232
    },
    {
      "epoch": 0.051100335695635955,
      "grad_norm": 0.4391435980796814,
      "learning_rate": 4.744705541050189e-06,
      "loss": 0.8394,
      "step": 1233
    },
    {
      "epoch": 0.051141779601309625,
      "grad_norm": 0.44852590560913086,
      "learning_rate": 4.744498321521821e-06,
      "loss": 0.793,
      "step": 1234
    },
    {
      "epoch": 0.051183223506983296,
      "grad_norm": 0.4417193830013275,
      "learning_rate": 4.744291101993452e-06,
      "loss": 0.7629,
      "step": 1235
    },
    {
      "epoch": 0.051224667412656966,
      "grad_norm": 0.46197283267974854,
      "learning_rate": 4.7440838824650836e-06,
      "loss": 0.7734,
      "step": 1236
    },
    {
      "epoch": 0.05126611131833064,
      "grad_norm": 0.4928566813468933,
      "learning_rate": 4.743876662936716e-06,
      "loss": 0.8452,
      "step": 1237
    },
    {
      "epoch": 0.05130755522400431,
      "grad_norm": 0.4651973247528076,
      "learning_rate": 4.743669443408347e-06,
      "loss": 0.8003,
      "step": 1238
    },
    {
      "epoch": 0.05134899912967798,
      "grad_norm": 0.4768562912940979,
      "learning_rate": 4.7434622238799786e-06,
      "loss": 0.7642,
      "step": 1239
    },
    {
      "epoch": 0.05139044303535165,
      "grad_norm": 0.4419132173061371,
      "learning_rate": 4.743255004351611e-06,
      "loss": 0.7263,
      "step": 1240
    },
    {
      "epoch": 0.05143188694102532,
      "grad_norm": 0.4663984179496765,
      "learning_rate": 4.743047784823242e-06,
      "loss": 0.7424,
      "step": 1241
    },
    {
      "epoch": 0.05147333084669899,
      "grad_norm": 0.43316197395324707,
      "learning_rate": 4.7428405652948736e-06,
      "loss": 0.7874,
      "step": 1242
    },
    {
      "epoch": 0.05151477475237266,
      "grad_norm": 0.469791442155838,
      "learning_rate": 4.742633345766505e-06,
      "loss": 0.8474,
      "step": 1243
    },
    {
      "epoch": 0.051556218658046334,
      "grad_norm": 0.4579676389694214,
      "learning_rate": 4.742426126238137e-06,
      "loss": 0.7749,
      "step": 1244
    },
    {
      "epoch": 0.051597662563720004,
      "grad_norm": 0.4559544026851654,
      "learning_rate": 4.7422189067097686e-06,
      "loss": 0.7773,
      "step": 1245
    },
    {
      "epoch": 0.051639106469393674,
      "grad_norm": 0.46836841106414795,
      "learning_rate": 4.7420116871814e-06,
      "loss": 0.7803,
      "step": 1246
    },
    {
      "epoch": 0.051680550375067344,
      "grad_norm": 0.45971202850341797,
      "learning_rate": 4.741804467653032e-06,
      "loss": 0.7688,
      "step": 1247
    },
    {
      "epoch": 0.051721994280741014,
      "grad_norm": 0.4328509569168091,
      "learning_rate": 4.741597248124664e-06,
      "loss": 0.6812,
      "step": 1248
    },
    {
      "epoch": 0.05176343818641469,
      "grad_norm": 0.45725953578948975,
      "learning_rate": 4.741390028596296e-06,
      "loss": 0.79,
      "step": 1249
    },
    {
      "epoch": 0.05180488209208836,
      "grad_norm": 0.4621913433074951,
      "learning_rate": 4.741182809067927e-06,
      "loss": 0.7808,
      "step": 1250
    },
    {
      "epoch": 0.05184632599776203,
      "grad_norm": 0.42693284153938293,
      "learning_rate": 4.740975589539559e-06,
      "loss": 0.7255,
      "step": 1251
    },
    {
      "epoch": 0.0518877699034357,
      "grad_norm": 0.4413866400718689,
      "learning_rate": 4.74076837001119e-06,
      "loss": 0.8127,
      "step": 1252
    },
    {
      "epoch": 0.05192921380910937,
      "grad_norm": 0.46110832691192627,
      "learning_rate": 4.740561150482822e-06,
      "loss": 0.7615,
      "step": 1253
    },
    {
      "epoch": 0.05197065771478304,
      "grad_norm": 0.46893593668937683,
      "learning_rate": 4.740353930954454e-06,
      "loss": 0.7832,
      "step": 1254
    },
    {
      "epoch": 0.05201210162045671,
      "grad_norm": 0.46650230884552,
      "learning_rate": 4.740146711426085e-06,
      "loss": 0.7595,
      "step": 1255
    },
    {
      "epoch": 0.05205354552613038,
      "grad_norm": 0.45120805501937866,
      "learning_rate": 4.739939491897717e-06,
      "loss": 0.7834,
      "step": 1256
    },
    {
      "epoch": 0.05209498943180405,
      "grad_norm": 0.4651395082473755,
      "learning_rate": 4.739732272369349e-06,
      "loss": 0.7977,
      "step": 1257
    },
    {
      "epoch": 0.05213643333747772,
      "grad_norm": 0.4553588330745697,
      "learning_rate": 4.73952505284098e-06,
      "loss": 0.7512,
      "step": 1258
    },
    {
      "epoch": 0.05217787724315139,
      "grad_norm": 0.42591726779937744,
      "learning_rate": 4.739317833312611e-06,
      "loss": 0.7185,
      "step": 1259
    },
    {
      "epoch": 0.05221932114882506,
      "grad_norm": 0.4449254274368286,
      "learning_rate": 4.739110613784244e-06,
      "loss": 0.8083,
      "step": 1260
    },
    {
      "epoch": 0.05226076505449873,
      "grad_norm": 0.48253604769706726,
      "learning_rate": 4.738903394255875e-06,
      "loss": 0.7739,
      "step": 1261
    },
    {
      "epoch": 0.05230220896017241,
      "grad_norm": 0.5065217614173889,
      "learning_rate": 4.738696174727506e-06,
      "loss": 0.8267,
      "step": 1262
    },
    {
      "epoch": 0.05234365286584608,
      "grad_norm": 0.4378811717033386,
      "learning_rate": 4.738488955199138e-06,
      "loss": 0.7341,
      "step": 1263
    },
    {
      "epoch": 0.05238509677151975,
      "grad_norm": 0.46514174342155457,
      "learning_rate": 4.73828173567077e-06,
      "loss": 0.8027,
      "step": 1264
    },
    {
      "epoch": 0.05242654067719342,
      "grad_norm": 0.47746387124061584,
      "learning_rate": 4.738074516142402e-06,
      "loss": 0.7625,
      "step": 1265
    },
    {
      "epoch": 0.05246798458286709,
      "grad_norm": 0.44600263237953186,
      "learning_rate": 4.737867296614034e-06,
      "loss": 0.7877,
      "step": 1266
    },
    {
      "epoch": 0.05250942848854076,
      "grad_norm": 0.5016936659812927,
      "learning_rate": 4.737660077085665e-06,
      "loss": 0.8289,
      "step": 1267
    },
    {
      "epoch": 0.05255087239421443,
      "grad_norm": 0.451377809047699,
      "learning_rate": 4.737452857557296e-06,
      "loss": 0.7678,
      "step": 1268
    },
    {
      "epoch": 0.0525923162998881,
      "grad_norm": 0.48705771565437317,
      "learning_rate": 4.737245638028929e-06,
      "loss": 0.7784,
      "step": 1269
    },
    {
      "epoch": 0.05263376020556177,
      "grad_norm": 0.4315616190433502,
      "learning_rate": 4.73703841850056e-06,
      "loss": 0.7045,
      "step": 1270
    },
    {
      "epoch": 0.05267520411123544,
      "grad_norm": 0.44772353768348694,
      "learning_rate": 4.736831198972191e-06,
      "loss": 0.7651,
      "step": 1271
    },
    {
      "epoch": 0.05271664801690911,
      "grad_norm": 0.4597644805908203,
      "learning_rate": 4.736623979443823e-06,
      "loss": 0.7969,
      "step": 1272
    },
    {
      "epoch": 0.05275809192258278,
      "grad_norm": 0.4657128155231476,
      "learning_rate": 4.736416759915455e-06,
      "loss": 0.7544,
      "step": 1273
    },
    {
      "epoch": 0.05279953582825646,
      "grad_norm": 0.4601643681526184,
      "learning_rate": 4.736209540387086e-06,
      "loss": 0.7812,
      "step": 1274
    },
    {
      "epoch": 0.05284097973393013,
      "grad_norm": 0.44550570845603943,
      "learning_rate": 4.736002320858718e-06,
      "loss": 0.8123,
      "step": 1275
    },
    {
      "epoch": 0.0528824236396038,
      "grad_norm": 0.47838711738586426,
      "learning_rate": 4.73579510133035e-06,
      "loss": 0.7943,
      "step": 1276
    },
    {
      "epoch": 0.05292386754527747,
      "grad_norm": 0.4126342833042145,
      "learning_rate": 4.735587881801981e-06,
      "loss": 0.7927,
      "step": 1277
    },
    {
      "epoch": 0.05296531145095114,
      "grad_norm": 0.452780157327652,
      "learning_rate": 4.735380662273613e-06,
      "loss": 0.8071,
      "step": 1278
    },
    {
      "epoch": 0.05300675535662481,
      "grad_norm": 0.4484234154224396,
      "learning_rate": 4.735173442745244e-06,
      "loss": 0.7646,
      "step": 1279
    },
    {
      "epoch": 0.05304819926229848,
      "grad_norm": 0.4484492838382721,
      "learning_rate": 4.734966223216876e-06,
      "loss": 0.8096,
      "step": 1280
    },
    {
      "epoch": 0.05308964316797215,
      "grad_norm": 0.4752931296825409,
      "learning_rate": 4.734759003688508e-06,
      "loss": 0.7748,
      "step": 1281
    },
    {
      "epoch": 0.05313108707364582,
      "grad_norm": 0.4608139991760254,
      "learning_rate": 4.73455178416014e-06,
      "loss": 0.7795,
      "step": 1282
    },
    {
      "epoch": 0.05317253097931949,
      "grad_norm": 0.46299976110458374,
      "learning_rate": 4.734344564631771e-06,
      "loss": 0.8064,
      "step": 1283
    },
    {
      "epoch": 0.05321397488499316,
      "grad_norm": 0.4294183552265167,
      "learning_rate": 4.734137345103403e-06,
      "loss": 0.7417,
      "step": 1284
    },
    {
      "epoch": 0.05325541879066683,
      "grad_norm": 0.4433351159095764,
      "learning_rate": 4.733930125575035e-06,
      "loss": 0.7664,
      "step": 1285
    },
    {
      "epoch": 0.053296862696340506,
      "grad_norm": 0.4322192072868347,
      "learning_rate": 4.733722906046666e-06,
      "loss": 0.7188,
      "step": 1286
    },
    {
      "epoch": 0.05333830660201418,
      "grad_norm": 0.47386589646339417,
      "learning_rate": 4.733515686518298e-06,
      "loss": 0.8521,
      "step": 1287
    },
    {
      "epoch": 0.05337975050768785,
      "grad_norm": 0.41507917642593384,
      "learning_rate": 4.733308466989929e-06,
      "loss": 0.7207,
      "step": 1288
    },
    {
      "epoch": 0.05342119441336152,
      "grad_norm": 0.4711005389690399,
      "learning_rate": 4.733101247461561e-06,
      "loss": 0.7949,
      "step": 1289
    },
    {
      "epoch": 0.05346263831903519,
      "grad_norm": 0.4546932578086853,
      "learning_rate": 4.732894027933193e-06,
      "loss": 0.7776,
      "step": 1290
    },
    {
      "epoch": 0.05350408222470886,
      "grad_norm": 0.44616714119911194,
      "learning_rate": 4.732686808404824e-06,
      "loss": 0.8046,
      "step": 1291
    },
    {
      "epoch": 0.05354552613038253,
      "grad_norm": 0.42783424258232117,
      "learning_rate": 4.732479588876456e-06,
      "loss": 0.8296,
      "step": 1292
    },
    {
      "epoch": 0.0535869700360562,
      "grad_norm": 0.5227195620536804,
      "learning_rate": 4.732272369348088e-06,
      "loss": 0.7788,
      "step": 1293
    },
    {
      "epoch": 0.05362841394172987,
      "grad_norm": 0.4596068859100342,
      "learning_rate": 4.732065149819719e-06,
      "loss": 0.7849,
      "step": 1294
    },
    {
      "epoch": 0.05366985784740354,
      "grad_norm": 0.4606890082359314,
      "learning_rate": 4.7318579302913506e-06,
      "loss": 0.8018,
      "step": 1295
    },
    {
      "epoch": 0.05371130175307721,
      "grad_norm": 0.4518755078315735,
      "learning_rate": 4.731650710762983e-06,
      "loss": 0.761,
      "step": 1296
    },
    {
      "epoch": 0.05375274565875088,
      "grad_norm": 0.42724910378456116,
      "learning_rate": 4.731443491234614e-06,
      "loss": 0.7271,
      "step": 1297
    },
    {
      "epoch": 0.05379418956442455,
      "grad_norm": 0.4897063076496124,
      "learning_rate": 4.731236271706246e-06,
      "loss": 0.8608,
      "step": 1298
    },
    {
      "epoch": 0.053835633470098225,
      "grad_norm": 0.46976977586746216,
      "learning_rate": 4.731029052177878e-06,
      "loss": 0.7422,
      "step": 1299
    },
    {
      "epoch": 0.053877077375771895,
      "grad_norm": 0.4982362687587738,
      "learning_rate": 4.730821832649509e-06,
      "loss": 0.8499,
      "step": 1300
    },
    {
      "epoch": 0.053918521281445565,
      "grad_norm": 0.4878818690776825,
      "learning_rate": 4.730614613121141e-06,
      "loss": 0.7991,
      "step": 1301
    },
    {
      "epoch": 0.053959965187119235,
      "grad_norm": 0.46012625098228455,
      "learning_rate": 4.730407393592773e-06,
      "loss": 0.7606,
      "step": 1302
    },
    {
      "epoch": 0.054001409092792906,
      "grad_norm": 0.4629114270210266,
      "learning_rate": 4.730200174064404e-06,
      "loss": 0.7686,
      "step": 1303
    },
    {
      "epoch": 0.054042852998466576,
      "grad_norm": 0.4677286446094513,
      "learning_rate": 4.7299929545360356e-06,
      "loss": 0.7405,
      "step": 1304
    },
    {
      "epoch": 0.054084296904140246,
      "grad_norm": 0.49047672748565674,
      "learning_rate": 4.729785735007668e-06,
      "loss": 0.8142,
      "step": 1305
    },
    {
      "epoch": 0.054125740809813916,
      "grad_norm": 0.42805495858192444,
      "learning_rate": 4.729578515479299e-06,
      "loss": 0.7424,
      "step": 1306
    },
    {
      "epoch": 0.054167184715487586,
      "grad_norm": 0.44661861658096313,
      "learning_rate": 4.729371295950931e-06,
      "loss": 0.7452,
      "step": 1307
    },
    {
      "epoch": 0.054208628621161256,
      "grad_norm": 0.4593750834465027,
      "learning_rate": 4.729164076422563e-06,
      "loss": 0.7402,
      "step": 1308
    },
    {
      "epoch": 0.054250072526834926,
      "grad_norm": 0.4440130293369293,
      "learning_rate": 4.728956856894194e-06,
      "loss": 0.8062,
      "step": 1309
    },
    {
      "epoch": 0.054291516432508596,
      "grad_norm": 0.44352737069129944,
      "learning_rate": 4.728749637365826e-06,
      "loss": 0.7673,
      "step": 1310
    },
    {
      "epoch": 0.05433296033818227,
      "grad_norm": 0.4515160322189331,
      "learning_rate": 4.728542417837457e-06,
      "loss": 0.7649,
      "step": 1311
    },
    {
      "epoch": 0.054374404243855944,
      "grad_norm": 0.4359804689884186,
      "learning_rate": 4.728335198309089e-06,
      "loss": 0.7666,
      "step": 1312
    },
    {
      "epoch": 0.054415848149529614,
      "grad_norm": 0.4292433559894562,
      "learning_rate": 4.728127978780721e-06,
      "loss": 0.7646,
      "step": 1313
    },
    {
      "epoch": 0.054457292055203284,
      "grad_norm": 0.44275203347206116,
      "learning_rate": 4.727920759252352e-06,
      "loss": 0.782,
      "step": 1314
    },
    {
      "epoch": 0.054498735960876954,
      "grad_norm": 0.45347097516059875,
      "learning_rate": 4.727713539723983e-06,
      "loss": 0.7654,
      "step": 1315
    },
    {
      "epoch": 0.054540179866550624,
      "grad_norm": 0.5133408904075623,
      "learning_rate": 4.727506320195616e-06,
      "loss": 0.874,
      "step": 1316
    },
    {
      "epoch": 0.054581623772224294,
      "grad_norm": 0.47484713792800903,
      "learning_rate": 4.727299100667248e-06,
      "loss": 0.8022,
      "step": 1317
    },
    {
      "epoch": 0.054623067677897964,
      "grad_norm": 0.4249754846096039,
      "learning_rate": 4.727091881138879e-06,
      "loss": 0.7523,
      "step": 1318
    },
    {
      "epoch": 0.054664511583571634,
      "grad_norm": 0.46269604563713074,
      "learning_rate": 4.726884661610511e-06,
      "loss": 0.822,
      "step": 1319
    },
    {
      "epoch": 0.054705955489245305,
      "grad_norm": 0.5005534887313843,
      "learning_rate": 4.726677442082142e-06,
      "loss": 0.7722,
      "step": 1320
    },
    {
      "epoch": 0.054747399394918975,
      "grad_norm": 0.45653414726257324,
      "learning_rate": 4.726470222553774e-06,
      "loss": 0.8069,
      "step": 1321
    },
    {
      "epoch": 0.054788843300592645,
      "grad_norm": 0.4670669436454773,
      "learning_rate": 4.726263003025406e-06,
      "loss": 0.8142,
      "step": 1322
    },
    {
      "epoch": 0.05483028720626632,
      "grad_norm": 0.41480040550231934,
      "learning_rate": 4.726055783497037e-06,
      "loss": 0.7656,
      "step": 1323
    },
    {
      "epoch": 0.05487173111193999,
      "grad_norm": 0.41693124175071716,
      "learning_rate": 4.725848563968668e-06,
      "loss": 0.7551,
      "step": 1324
    },
    {
      "epoch": 0.05491317501761366,
      "grad_norm": 0.4565029740333557,
      "learning_rate": 4.725641344440301e-06,
      "loss": 0.7603,
      "step": 1325
    },
    {
      "epoch": 0.05495461892328733,
      "grad_norm": 0.47091975808143616,
      "learning_rate": 4.725434124911932e-06,
      "loss": 0.7986,
      "step": 1326
    },
    {
      "epoch": 0.054996062828961,
      "grad_norm": 0.44375982880592346,
      "learning_rate": 4.725226905383563e-06,
      "loss": 0.7969,
      "step": 1327
    },
    {
      "epoch": 0.05503750673463467,
      "grad_norm": 0.4751598536968231,
      "learning_rate": 4.725019685855196e-06,
      "loss": 0.8193,
      "step": 1328
    },
    {
      "epoch": 0.05507895064030834,
      "grad_norm": 0.5229005813598633,
      "learning_rate": 4.724812466326827e-06,
      "loss": 0.8435,
      "step": 1329
    },
    {
      "epoch": 0.05512039454598201,
      "grad_norm": 0.48139140009880066,
      "learning_rate": 4.724605246798458e-06,
      "loss": 0.853,
      "step": 1330
    },
    {
      "epoch": 0.05516183845165568,
      "grad_norm": 0.43209391832351685,
      "learning_rate": 4.72439802727009e-06,
      "loss": 0.7439,
      "step": 1331
    },
    {
      "epoch": 0.05520328235732935,
      "grad_norm": 0.44706329703330994,
      "learning_rate": 4.724190807741722e-06,
      "loss": 0.7378,
      "step": 1332
    },
    {
      "epoch": 0.05524472626300302,
      "grad_norm": 0.43676483631134033,
      "learning_rate": 4.723983588213353e-06,
      "loss": 0.7466,
      "step": 1333
    },
    {
      "epoch": 0.05528617016867669,
      "grad_norm": 0.45234090089797974,
      "learning_rate": 4.723776368684986e-06,
      "loss": 0.7336,
      "step": 1334
    },
    {
      "epoch": 0.05532761407435036,
      "grad_norm": 0.47403037548065186,
      "learning_rate": 4.723569149156617e-06,
      "loss": 0.7788,
      "step": 1335
    },
    {
      "epoch": 0.05536905798002404,
      "grad_norm": 0.4357151389122009,
      "learning_rate": 4.723361929628248e-06,
      "loss": 0.7493,
      "step": 1336
    },
    {
      "epoch": 0.05541050188569771,
      "grad_norm": 0.47107189893722534,
      "learning_rate": 4.723154710099881e-06,
      "loss": 0.7499,
      "step": 1337
    },
    {
      "epoch": 0.05545194579137138,
      "grad_norm": 0.4392364025115967,
      "learning_rate": 4.722947490571512e-06,
      "loss": 0.7244,
      "step": 1338
    },
    {
      "epoch": 0.05549338969704505,
      "grad_norm": 0.43431559205055237,
      "learning_rate": 4.722740271043143e-06,
      "loss": 0.7168,
      "step": 1339
    },
    {
      "epoch": 0.05553483360271872,
      "grad_norm": 0.486935019493103,
      "learning_rate": 4.722533051514775e-06,
      "loss": 0.7734,
      "step": 1340
    },
    {
      "epoch": 0.05557627750839239,
      "grad_norm": 0.44165152311325073,
      "learning_rate": 4.722325831986407e-06,
      "loss": 0.7812,
      "step": 1341
    },
    {
      "epoch": 0.05561772141406606,
      "grad_norm": 0.4955763518810272,
      "learning_rate": 4.722118612458038e-06,
      "loss": 0.865,
      "step": 1342
    },
    {
      "epoch": 0.05565916531973973,
      "grad_norm": 0.47544750571250916,
      "learning_rate": 4.72191139292967e-06,
      "loss": 0.7437,
      "step": 1343
    },
    {
      "epoch": 0.0557006092254134,
      "grad_norm": 0.4376749098300934,
      "learning_rate": 4.721704173401302e-06,
      "loss": 0.7534,
      "step": 1344
    },
    {
      "epoch": 0.05574205313108707,
      "grad_norm": 0.43140000104904175,
      "learning_rate": 4.721496953872933e-06,
      "loss": 0.7615,
      "step": 1345
    },
    {
      "epoch": 0.05578349703676074,
      "grad_norm": 0.4696462154388428,
      "learning_rate": 4.721289734344565e-06,
      "loss": 0.729,
      "step": 1346
    },
    {
      "epoch": 0.05582494094243441,
      "grad_norm": 0.4473852515220642,
      "learning_rate": 4.721082514816196e-06,
      "loss": 0.802,
      "step": 1347
    },
    {
      "epoch": 0.05586638484810809,
      "grad_norm": 0.48599231243133545,
      "learning_rate": 4.720875295287828e-06,
      "loss": 0.783,
      "step": 1348
    },
    {
      "epoch": 0.05590782875378176,
      "grad_norm": 0.42560818791389465,
      "learning_rate": 4.72066807575946e-06,
      "loss": 0.7476,
      "step": 1349
    },
    {
      "epoch": 0.05594927265945543,
      "grad_norm": 0.46392273902893066,
      "learning_rate": 4.720460856231092e-06,
      "loss": 0.8218,
      "step": 1350
    },
    {
      "epoch": 0.0559907165651291,
      "grad_norm": 0.4391045868396759,
      "learning_rate": 4.720253636702723e-06,
      "loss": 0.7446,
      "step": 1351
    },
    {
      "epoch": 0.05603216047080277,
      "grad_norm": 0.45642098784446716,
      "learning_rate": 4.720046417174355e-06,
      "loss": 0.7983,
      "step": 1352
    },
    {
      "epoch": 0.05607360437647644,
      "grad_norm": 0.48248791694641113,
      "learning_rate": 4.719839197645987e-06,
      "loss": 0.8186,
      "step": 1353
    },
    {
      "epoch": 0.05611504828215011,
      "grad_norm": 0.4436752498149872,
      "learning_rate": 4.719631978117618e-06,
      "loss": 0.832,
      "step": 1354
    },
    {
      "epoch": 0.05615649218782378,
      "grad_norm": 0.4888905882835388,
      "learning_rate": 4.71942475858925e-06,
      "loss": 0.7773,
      "step": 1355
    },
    {
      "epoch": 0.05619793609349745,
      "grad_norm": 0.430061399936676,
      "learning_rate": 4.719217539060881e-06,
      "loss": 0.7517,
      "step": 1356
    },
    {
      "epoch": 0.05623937999917112,
      "grad_norm": 0.43851524591445923,
      "learning_rate": 4.719010319532513e-06,
      "loss": 0.7766,
      "step": 1357
    },
    {
      "epoch": 0.05628082390484479,
      "grad_norm": 0.4564257562160492,
      "learning_rate": 4.718803100004145e-06,
      "loss": 0.8306,
      "step": 1358
    },
    {
      "epoch": 0.05632226781051846,
      "grad_norm": 0.46868595480918884,
      "learning_rate": 4.718595880475776e-06,
      "loss": 0.7771,
      "step": 1359
    },
    {
      "epoch": 0.05636371171619214,
      "grad_norm": 0.4619618356227875,
      "learning_rate": 4.718388660947408e-06,
      "loss": 0.7883,
      "step": 1360
    },
    {
      "epoch": 0.05640515562186581,
      "grad_norm": 0.4858212471008301,
      "learning_rate": 4.71818144141904e-06,
      "loss": 0.8131,
      "step": 1361
    },
    {
      "epoch": 0.05644659952753948,
      "grad_norm": 0.46253591775894165,
      "learning_rate": 4.717974221890671e-06,
      "loss": 0.8906,
      "step": 1362
    },
    {
      "epoch": 0.05648804343321315,
      "grad_norm": 0.44040533900260925,
      "learning_rate": 4.7177670023623026e-06,
      "loss": 0.7595,
      "step": 1363
    },
    {
      "epoch": 0.05652948733888682,
      "grad_norm": 0.47029218077659607,
      "learning_rate": 4.717559782833935e-06,
      "loss": 0.8262,
      "step": 1364
    },
    {
      "epoch": 0.05657093124456049,
      "grad_norm": 0.4347769021987915,
      "learning_rate": 4.717352563305566e-06,
      "loss": 0.7627,
      "step": 1365
    },
    {
      "epoch": 0.05661237515023416,
      "grad_norm": 0.4234626293182373,
      "learning_rate": 4.7171453437771984e-06,
      "loss": 0.7991,
      "step": 1366
    },
    {
      "epoch": 0.05665381905590783,
      "grad_norm": 0.425837904214859,
      "learning_rate": 4.716938124248829e-06,
      "loss": 0.7546,
      "step": 1367
    },
    {
      "epoch": 0.0566952629615815,
      "grad_norm": 0.4850502908229828,
      "learning_rate": 4.716730904720461e-06,
      "loss": 0.8048,
      "step": 1368
    },
    {
      "epoch": 0.05673670686725517,
      "grad_norm": 0.4783625900745392,
      "learning_rate": 4.7165236851920934e-06,
      "loss": 0.7489,
      "step": 1369
    },
    {
      "epoch": 0.05677815077292884,
      "grad_norm": 0.4982050061225891,
      "learning_rate": 4.716316465663725e-06,
      "loss": 0.7686,
      "step": 1370
    },
    {
      "epoch": 0.05681959467860251,
      "grad_norm": 0.47850582003593445,
      "learning_rate": 4.716109246135356e-06,
      "loss": 0.8325,
      "step": 1371
    },
    {
      "epoch": 0.056861038584276186,
      "grad_norm": 0.4520390033721924,
      "learning_rate": 4.715902026606988e-06,
      "loss": 0.7834,
      "step": 1372
    },
    {
      "epoch": 0.056902482489949856,
      "grad_norm": 0.4395468235015869,
      "learning_rate": 4.71569480707862e-06,
      "loss": 0.7832,
      "step": 1373
    },
    {
      "epoch": 0.056943926395623526,
      "grad_norm": 0.49860715866088867,
      "learning_rate": 4.715487587550251e-06,
      "loss": 0.8125,
      "step": 1374
    },
    {
      "epoch": 0.056985370301297196,
      "grad_norm": 0.4712047576904297,
      "learning_rate": 4.715280368021883e-06,
      "loss": 0.7559,
      "step": 1375
    },
    {
      "epoch": 0.057026814206970866,
      "grad_norm": 0.43081262707710266,
      "learning_rate": 4.715073148493514e-06,
      "loss": 0.749,
      "step": 1376
    },
    {
      "epoch": 0.057068258112644536,
      "grad_norm": 0.49594810605049133,
      "learning_rate": 4.714865928965146e-06,
      "loss": 0.748,
      "step": 1377
    },
    {
      "epoch": 0.057109702018318206,
      "grad_norm": 0.4482758939266205,
      "learning_rate": 4.714658709436778e-06,
      "loss": 0.7896,
      "step": 1378
    },
    {
      "epoch": 0.057151145923991876,
      "grad_norm": 0.4533829391002655,
      "learning_rate": 4.714451489908409e-06,
      "loss": 0.7551,
      "step": 1379
    },
    {
      "epoch": 0.05719258982966555,
      "grad_norm": 0.5021113157272339,
      "learning_rate": 4.714244270380041e-06,
      "loss": 0.7808,
      "step": 1380
    },
    {
      "epoch": 0.05723403373533922,
      "grad_norm": 0.5018630623817444,
      "learning_rate": 4.714037050851673e-06,
      "loss": 0.8,
      "step": 1381
    },
    {
      "epoch": 0.05727547764101289,
      "grad_norm": 0.5723501443862915,
      "learning_rate": 4.713829831323304e-06,
      "loss": 0.8782,
      "step": 1382
    },
    {
      "epoch": 0.05731692154668656,
      "grad_norm": 0.4557441473007202,
      "learning_rate": 4.713622611794935e-06,
      "loss": 0.8135,
      "step": 1383
    },
    {
      "epoch": 0.05735836545236023,
      "grad_norm": 0.49008315801620483,
      "learning_rate": 4.713415392266568e-06,
      "loss": 0.728,
      "step": 1384
    },
    {
      "epoch": 0.057399809358033904,
      "grad_norm": 0.49280646443367004,
      "learning_rate": 4.713208172738199e-06,
      "loss": 0.7808,
      "step": 1385
    },
    {
      "epoch": 0.057441253263707574,
      "grad_norm": 0.47247087955474854,
      "learning_rate": 4.713000953209831e-06,
      "loss": 0.7986,
      "step": 1386
    },
    {
      "epoch": 0.057482697169381244,
      "grad_norm": 0.4705398678779602,
      "learning_rate": 4.712793733681463e-06,
      "loss": 0.7925,
      "step": 1387
    },
    {
      "epoch": 0.057524141075054915,
      "grad_norm": 0.44505465030670166,
      "learning_rate": 4.712586514153094e-06,
      "loss": 0.7905,
      "step": 1388
    },
    {
      "epoch": 0.057565584980728585,
      "grad_norm": 0.5128839015960693,
      "learning_rate": 4.712379294624726e-06,
      "loss": 0.7319,
      "step": 1389
    },
    {
      "epoch": 0.057607028886402255,
      "grad_norm": 0.4779432713985443,
      "learning_rate": 4.712172075096358e-06,
      "loss": 0.7871,
      "step": 1390
    },
    {
      "epoch": 0.057648472792075925,
      "grad_norm": 0.4249486029148102,
      "learning_rate": 4.711964855567989e-06,
      "loss": 0.7109,
      "step": 1391
    },
    {
      "epoch": 0.057689916697749595,
      "grad_norm": 0.43019023537635803,
      "learning_rate": 4.71175763603962e-06,
      "loss": 0.8232,
      "step": 1392
    },
    {
      "epoch": 0.057731360603423265,
      "grad_norm": 0.43577319383621216,
      "learning_rate": 4.711550416511253e-06,
      "loss": 0.7368,
      "step": 1393
    },
    {
      "epoch": 0.057772804509096935,
      "grad_norm": 0.44867873191833496,
      "learning_rate": 4.711343196982884e-06,
      "loss": 0.7561,
      "step": 1394
    },
    {
      "epoch": 0.057814248414770605,
      "grad_norm": 0.45300158858299255,
      "learning_rate": 4.711135977454515e-06,
      "loss": 0.845,
      "step": 1395
    },
    {
      "epoch": 0.057855692320444276,
      "grad_norm": 0.45936110615730286,
      "learning_rate": 4.710928757926148e-06,
      "loss": 0.8066,
      "step": 1396
    },
    {
      "epoch": 0.05789713622611795,
      "grad_norm": 0.4553302228450775,
      "learning_rate": 4.710721538397779e-06,
      "loss": 0.7469,
      "step": 1397
    },
    {
      "epoch": 0.05793858013179162,
      "grad_norm": 0.4707634747028351,
      "learning_rate": 4.71051431886941e-06,
      "loss": 0.786,
      "step": 1398
    },
    {
      "epoch": 0.05798002403746529,
      "grad_norm": 0.49710139632225037,
      "learning_rate": 4.710307099341042e-06,
      "loss": 0.8093,
      "step": 1399
    },
    {
      "epoch": 0.05802146794313896,
      "grad_norm": 0.4399222135543823,
      "learning_rate": 4.710099879812674e-06,
      "loss": 0.7859,
      "step": 1400
    },
    {
      "epoch": 0.05806291184881263,
      "grad_norm": 0.4463959336280823,
      "learning_rate": 4.709892660284305e-06,
      "loss": 0.7429,
      "step": 1401
    },
    {
      "epoch": 0.0581043557544863,
      "grad_norm": 0.48263630270957947,
      "learning_rate": 4.709685440755938e-06,
      "loss": 0.7974,
      "step": 1402
    },
    {
      "epoch": 0.05814579966015997,
      "grad_norm": 0.47611480951309204,
      "learning_rate": 4.709478221227569e-06,
      "loss": 0.7676,
      "step": 1403
    },
    {
      "epoch": 0.05818724356583364,
      "grad_norm": 0.44558659195899963,
      "learning_rate": 4.7092710016992e-06,
      "loss": 0.7546,
      "step": 1404
    },
    {
      "epoch": 0.058228687471507314,
      "grad_norm": 0.4709329605102539,
      "learning_rate": 4.709063782170833e-06,
      "loss": 0.8303,
      "step": 1405
    },
    {
      "epoch": 0.058270131377180984,
      "grad_norm": 0.4659077227115631,
      "learning_rate": 4.708856562642464e-06,
      "loss": 0.8159,
      "step": 1406
    },
    {
      "epoch": 0.058311575282854654,
      "grad_norm": 0.4674285650253296,
      "learning_rate": 4.708649343114095e-06,
      "loss": 0.7412,
      "step": 1407
    },
    {
      "epoch": 0.058353019188528324,
      "grad_norm": 0.46319887042045593,
      "learning_rate": 4.708442123585727e-06,
      "loss": 0.8301,
      "step": 1408
    },
    {
      "epoch": 0.058394463094202,
      "grad_norm": 0.4718414545059204,
      "learning_rate": 4.708234904057359e-06,
      "loss": 0.7737,
      "step": 1409
    },
    {
      "epoch": 0.05843590699987567,
      "grad_norm": 0.42893242835998535,
      "learning_rate": 4.70802768452899e-06,
      "loss": 0.73,
      "step": 1410
    },
    {
      "epoch": 0.05847735090554934,
      "grad_norm": 0.42724114656448364,
      "learning_rate": 4.707820465000622e-06,
      "loss": 0.7969,
      "step": 1411
    },
    {
      "epoch": 0.05851879481122301,
      "grad_norm": 0.4840550422668457,
      "learning_rate": 4.707613245472254e-06,
      "loss": 0.8049,
      "step": 1412
    },
    {
      "epoch": 0.05856023871689668,
      "grad_norm": 0.4904533922672272,
      "learning_rate": 4.707406025943885e-06,
      "loss": 0.7834,
      "step": 1413
    },
    {
      "epoch": 0.05860168262257035,
      "grad_norm": 0.4311313033103943,
      "learning_rate": 4.707198806415517e-06,
      "loss": 0.6936,
      "step": 1414
    },
    {
      "epoch": 0.05864312652824402,
      "grad_norm": 0.45090553164482117,
      "learning_rate": 4.706991586887148e-06,
      "loss": 0.8176,
      "step": 1415
    },
    {
      "epoch": 0.05868457043391769,
      "grad_norm": 0.39051318168640137,
      "learning_rate": 4.70678436735878e-06,
      "loss": 0.752,
      "step": 1416
    },
    {
      "epoch": 0.05872601433959136,
      "grad_norm": 0.4844968616962433,
      "learning_rate": 4.706577147830412e-06,
      "loss": 0.7537,
      "step": 1417
    },
    {
      "epoch": 0.05876745824526503,
      "grad_norm": 0.4283495843410492,
      "learning_rate": 4.706369928302044e-06,
      "loss": 0.7722,
      "step": 1418
    },
    {
      "epoch": 0.0588089021509387,
      "grad_norm": 0.47521117329597473,
      "learning_rate": 4.7061627087736746e-06,
      "loss": 0.7988,
      "step": 1419
    },
    {
      "epoch": 0.05885034605661237,
      "grad_norm": 0.46391305327415466,
      "learning_rate": 4.705955489245307e-06,
      "loss": 0.8076,
      "step": 1420
    },
    {
      "epoch": 0.05889178996228604,
      "grad_norm": 0.45172983407974243,
      "learning_rate": 4.705748269716939e-06,
      "loss": 0.8005,
      "step": 1421
    },
    {
      "epoch": 0.05893323386795972,
      "grad_norm": 0.4898605942726135,
      "learning_rate": 4.7055410501885704e-06,
      "loss": 0.8145,
      "step": 1422
    },
    {
      "epoch": 0.05897467777363339,
      "grad_norm": 0.43830570578575134,
      "learning_rate": 4.705333830660202e-06,
      "loss": 0.8438,
      "step": 1423
    },
    {
      "epoch": 0.05901612167930706,
      "grad_norm": 0.43851032853126526,
      "learning_rate": 4.705126611131833e-06,
      "loss": 0.7546,
      "step": 1424
    },
    {
      "epoch": 0.05905756558498073,
      "grad_norm": 0.44282862544059753,
      "learning_rate": 4.7049193916034654e-06,
      "loss": 0.7466,
      "step": 1425
    },
    {
      "epoch": 0.0590990094906544,
      "grad_norm": 0.5012243390083313,
      "learning_rate": 4.704712172075097e-06,
      "loss": 0.7932,
      "step": 1426
    },
    {
      "epoch": 0.05914045339632807,
      "grad_norm": 0.42376306653022766,
      "learning_rate": 4.704504952546728e-06,
      "loss": 0.7358,
      "step": 1427
    },
    {
      "epoch": 0.05918189730200174,
      "grad_norm": 0.46943458914756775,
      "learning_rate": 4.70429773301836e-06,
      "loss": 0.7771,
      "step": 1428
    },
    {
      "epoch": 0.05922334120767541,
      "grad_norm": 0.48581844568252563,
      "learning_rate": 4.704090513489992e-06,
      "loss": 0.7795,
      "step": 1429
    },
    {
      "epoch": 0.05926478511334908,
      "grad_norm": 0.5191372036933899,
      "learning_rate": 4.703883293961623e-06,
      "loss": 0.8181,
      "step": 1430
    },
    {
      "epoch": 0.05930622901902275,
      "grad_norm": 0.5243611931800842,
      "learning_rate": 4.703676074433255e-06,
      "loss": 0.8247,
      "step": 1431
    },
    {
      "epoch": 0.05934767292469642,
      "grad_norm": 0.5012795925140381,
      "learning_rate": 4.703468854904887e-06,
      "loss": 0.7383,
      "step": 1432
    },
    {
      "epoch": 0.05938911683037009,
      "grad_norm": 0.46006128191947937,
      "learning_rate": 4.703261635376518e-06,
      "loss": 0.823,
      "step": 1433
    },
    {
      "epoch": 0.05943056073604377,
      "grad_norm": 0.4853540062904358,
      "learning_rate": 4.7030544158481504e-06,
      "loss": 0.8584,
      "step": 1434
    },
    {
      "epoch": 0.05947200464171744,
      "grad_norm": 0.4474965035915375,
      "learning_rate": 4.702847196319781e-06,
      "loss": 0.7578,
      "step": 1435
    },
    {
      "epoch": 0.05951344854739111,
      "grad_norm": 0.45749303698539734,
      "learning_rate": 4.702639976791413e-06,
      "loss": 0.7471,
      "step": 1436
    },
    {
      "epoch": 0.05955489245306478,
      "grad_norm": 0.4620939791202545,
      "learning_rate": 4.702432757263045e-06,
      "loss": 0.7568,
      "step": 1437
    },
    {
      "epoch": 0.05959633635873845,
      "grad_norm": 0.4508247673511505,
      "learning_rate": 4.702225537734677e-06,
      "loss": 0.7805,
      "step": 1438
    },
    {
      "epoch": 0.05963778026441212,
      "grad_norm": 0.43038907647132874,
      "learning_rate": 4.702018318206308e-06,
      "loss": 0.7759,
      "step": 1439
    },
    {
      "epoch": 0.05967922417008579,
      "grad_norm": 0.4614555239677429,
      "learning_rate": 4.70181109867794e-06,
      "loss": 0.7096,
      "step": 1440
    },
    {
      "epoch": 0.05972066807575946,
      "grad_norm": 0.5100172162055969,
      "learning_rate": 4.701603879149572e-06,
      "loss": 0.8169,
      "step": 1441
    },
    {
      "epoch": 0.05976211198143313,
      "grad_norm": 0.42111605405807495,
      "learning_rate": 4.701396659621203e-06,
      "loss": 0.7976,
      "step": 1442
    },
    {
      "epoch": 0.0598035558871068,
      "grad_norm": 0.46577054262161255,
      "learning_rate": 4.701189440092835e-06,
      "loss": 0.7642,
      "step": 1443
    },
    {
      "epoch": 0.05984499979278047,
      "grad_norm": 0.4743770956993103,
      "learning_rate": 4.700982220564466e-06,
      "loss": 0.8225,
      "step": 1444
    },
    {
      "epoch": 0.05988644369845414,
      "grad_norm": 0.4821048378944397,
      "learning_rate": 4.700775001036098e-06,
      "loss": 0.7988,
      "step": 1445
    },
    {
      "epoch": 0.059927887604127816,
      "grad_norm": 0.4471295475959778,
      "learning_rate": 4.70056778150773e-06,
      "loss": 0.78,
      "step": 1446
    },
    {
      "epoch": 0.059969331509801486,
      "grad_norm": 0.5131109356880188,
      "learning_rate": 4.700360561979361e-06,
      "loss": 0.813,
      "step": 1447
    },
    {
      "epoch": 0.06001077541547516,
      "grad_norm": 0.4415600895881653,
      "learning_rate": 4.700153342450993e-06,
      "loss": 0.7524,
      "step": 1448
    },
    {
      "epoch": 0.06005221932114883,
      "grad_norm": 0.4578064978122711,
      "learning_rate": 4.699946122922625e-06,
      "loss": 0.7898,
      "step": 1449
    },
    {
      "epoch": 0.0600936632268225,
      "grad_norm": 0.6379995942115784,
      "learning_rate": 4.699738903394257e-06,
      "loss": 0.8171,
      "step": 1450
    },
    {
      "epoch": 0.06013510713249617,
      "grad_norm": 0.4599008858203888,
      "learning_rate": 4.699531683865887e-06,
      "loss": 0.7585,
      "step": 1451
    },
    {
      "epoch": 0.06017655103816984,
      "grad_norm": 0.49943751096725464,
      "learning_rate": 4.69932446433752e-06,
      "loss": 0.77,
      "step": 1452
    },
    {
      "epoch": 0.06021799494384351,
      "grad_norm": 0.45110073685646057,
      "learning_rate": 4.699117244809151e-06,
      "loss": 0.7832,
      "step": 1453
    },
    {
      "epoch": 0.06025943884951718,
      "grad_norm": 0.4260523021221161,
      "learning_rate": 4.698910025280783e-06,
      "loss": 0.761,
      "step": 1454
    },
    {
      "epoch": 0.06030088275519085,
      "grad_norm": 0.4733874499797821,
      "learning_rate": 4.698702805752415e-06,
      "loss": 0.7107,
      "step": 1455
    },
    {
      "epoch": 0.06034232666086452,
      "grad_norm": 0.46258240938186646,
      "learning_rate": 4.698495586224046e-06,
      "loss": 0.7787,
      "step": 1456
    },
    {
      "epoch": 0.06038377056653819,
      "grad_norm": 0.4139939546585083,
      "learning_rate": 4.698288366695678e-06,
      "loss": 0.7402,
      "step": 1457
    },
    {
      "epoch": 0.06042521447221186,
      "grad_norm": 0.45899829268455505,
      "learning_rate": 4.69808114716731e-06,
      "loss": 0.7964,
      "step": 1458
    },
    {
      "epoch": 0.060466658377885535,
      "grad_norm": 0.41487041115760803,
      "learning_rate": 4.697873927638941e-06,
      "loss": 0.7874,
      "step": 1459
    },
    {
      "epoch": 0.060508102283559205,
      "grad_norm": 0.44744035601615906,
      "learning_rate": 4.697666708110572e-06,
      "loss": 0.7849,
      "step": 1460
    },
    {
      "epoch": 0.060549546189232875,
      "grad_norm": 0.4636276066303253,
      "learning_rate": 4.697459488582205e-06,
      "loss": 0.7988,
      "step": 1461
    },
    {
      "epoch": 0.060590990094906545,
      "grad_norm": 0.4395328462123871,
      "learning_rate": 4.697252269053836e-06,
      "loss": 0.7308,
      "step": 1462
    },
    {
      "epoch": 0.060632434000580215,
      "grad_norm": 0.47965776920318604,
      "learning_rate": 4.697045049525467e-06,
      "loss": 0.8169,
      "step": 1463
    },
    {
      "epoch": 0.060673877906253885,
      "grad_norm": 0.4725888967514038,
      "learning_rate": 4.6968378299971e-06,
      "loss": 0.7573,
      "step": 1464
    },
    {
      "epoch": 0.060715321811927556,
      "grad_norm": 0.459410160779953,
      "learning_rate": 4.696630610468731e-06,
      "loss": 0.7124,
      "step": 1465
    },
    {
      "epoch": 0.060756765717601226,
      "grad_norm": 0.47637781500816345,
      "learning_rate": 4.696423390940362e-06,
      "loss": 0.7988,
      "step": 1466
    },
    {
      "epoch": 0.060798209623274896,
      "grad_norm": 0.463990181684494,
      "learning_rate": 4.696216171411994e-06,
      "loss": 0.7771,
      "step": 1467
    },
    {
      "epoch": 0.060839653528948566,
      "grad_norm": 0.4259353280067444,
      "learning_rate": 4.696008951883626e-06,
      "loss": 0.7422,
      "step": 1468
    },
    {
      "epoch": 0.060881097434622236,
      "grad_norm": 0.443270206451416,
      "learning_rate": 4.695801732355257e-06,
      "loss": 0.8042,
      "step": 1469
    },
    {
      "epoch": 0.060922541340295906,
      "grad_norm": 0.44445404410362244,
      "learning_rate": 4.69559451282689e-06,
      "loss": 0.7572,
      "step": 1470
    },
    {
      "epoch": 0.06096398524596958,
      "grad_norm": 0.48976948857307434,
      "learning_rate": 4.695387293298521e-06,
      "loss": 0.8477,
      "step": 1471
    },
    {
      "epoch": 0.06100542915164325,
      "grad_norm": 0.437501460313797,
      "learning_rate": 4.695180073770152e-06,
      "loss": 0.7725,
      "step": 1472
    },
    {
      "epoch": 0.061046873057316924,
      "grad_norm": 0.4239104092121124,
      "learning_rate": 4.694972854241785e-06,
      "loss": 0.7419,
      "step": 1473
    },
    {
      "epoch": 0.061088316962990594,
      "grad_norm": 0.42657554149627686,
      "learning_rate": 4.694765634713416e-06,
      "loss": 0.7693,
      "step": 1474
    },
    {
      "epoch": 0.061129760868664264,
      "grad_norm": 0.4616900682449341,
      "learning_rate": 4.694558415185047e-06,
      "loss": 0.8027,
      "step": 1475
    },
    {
      "epoch": 0.061171204774337934,
      "grad_norm": 0.4385174512863159,
      "learning_rate": 4.694351195656679e-06,
      "loss": 0.824,
      "step": 1476
    },
    {
      "epoch": 0.061212648680011604,
      "grad_norm": 0.5222707390785217,
      "learning_rate": 4.694143976128311e-06,
      "loss": 0.7588,
      "step": 1477
    },
    {
      "epoch": 0.061254092585685274,
      "grad_norm": 0.46972206234931946,
      "learning_rate": 4.693936756599942e-06,
      "loss": 0.7766,
      "step": 1478
    },
    {
      "epoch": 0.061295536491358944,
      "grad_norm": 0.47943025827407837,
      "learning_rate": 4.693729537071574e-06,
      "loss": 0.8193,
      "step": 1479
    },
    {
      "epoch": 0.061336980397032614,
      "grad_norm": 0.42880117893218994,
      "learning_rate": 4.693522317543205e-06,
      "loss": 0.7551,
      "step": 1480
    },
    {
      "epoch": 0.061378424302706285,
      "grad_norm": 0.4733676314353943,
      "learning_rate": 4.6933150980148374e-06,
      "loss": 0.7756,
      "step": 1481
    },
    {
      "epoch": 0.061419868208379955,
      "grad_norm": 0.47677120566368103,
      "learning_rate": 4.693107878486469e-06,
      "loss": 0.7781,
      "step": 1482
    },
    {
      "epoch": 0.06146131211405363,
      "grad_norm": 0.465616375207901,
      "learning_rate": 4.6929006589581e-06,
      "loss": 0.7251,
      "step": 1483
    },
    {
      "epoch": 0.0615027560197273,
      "grad_norm": 0.4498819410800934,
      "learning_rate": 4.6926934394297324e-06,
      "loss": 0.7742,
      "step": 1484
    },
    {
      "epoch": 0.06154419992540097,
      "grad_norm": 0.46503645181655884,
      "learning_rate": 4.692486219901364e-06,
      "loss": 0.7786,
      "step": 1485
    },
    {
      "epoch": 0.06158564383107464,
      "grad_norm": 0.4819374978542328,
      "learning_rate": 4.692279000372996e-06,
      "loss": 0.7644,
      "step": 1486
    },
    {
      "epoch": 0.06162708773674831,
      "grad_norm": 0.4457657039165497,
      "learning_rate": 4.692071780844627e-06,
      "loss": 0.7695,
      "step": 1487
    },
    {
      "epoch": 0.06166853164242198,
      "grad_norm": 0.44114214181900024,
      "learning_rate": 4.691864561316259e-06,
      "loss": 0.793,
      "step": 1488
    },
    {
      "epoch": 0.06170997554809565,
      "grad_norm": 0.4471187889575958,
      "learning_rate": 4.69165734178789e-06,
      "loss": 0.7827,
      "step": 1489
    },
    {
      "epoch": 0.06175141945376932,
      "grad_norm": 0.4443762004375458,
      "learning_rate": 4.6914501222595224e-06,
      "loss": 0.7593,
      "step": 1490
    },
    {
      "epoch": 0.06179286335944299,
      "grad_norm": 0.4737187922000885,
      "learning_rate": 4.691242902731154e-06,
      "loss": 0.8752,
      "step": 1491
    },
    {
      "epoch": 0.06183430726511666,
      "grad_norm": 0.4137257933616638,
      "learning_rate": 4.691035683202785e-06,
      "loss": 0.715,
      "step": 1492
    },
    {
      "epoch": 0.06187575117079033,
      "grad_norm": 0.4365899860858917,
      "learning_rate": 4.6908284636744174e-06,
      "loss": 0.7819,
      "step": 1493
    },
    {
      "epoch": 0.061917195076464,
      "grad_norm": 0.48087775707244873,
      "learning_rate": 4.690621244146049e-06,
      "loss": 0.7505,
      "step": 1494
    },
    {
      "epoch": 0.06195863898213768,
      "grad_norm": 0.4672144651412964,
      "learning_rate": 4.69041402461768e-06,
      "loss": 0.7654,
      "step": 1495
    },
    {
      "epoch": 0.06200008288781135,
      "grad_norm": 0.4776240885257721,
      "learning_rate": 4.690206805089312e-06,
      "loss": 0.8257,
      "step": 1496
    },
    {
      "epoch": 0.06204152679348502,
      "grad_norm": 0.4997063875198364,
      "learning_rate": 4.689999585560944e-06,
      "loss": 0.806,
      "step": 1497
    },
    {
      "epoch": 0.06208297069915869,
      "grad_norm": 0.502731204032898,
      "learning_rate": 4.689792366032575e-06,
      "loss": 0.8081,
      "step": 1498
    },
    {
      "epoch": 0.06212441460483236,
      "grad_norm": 0.4546010494232178,
      "learning_rate": 4.689585146504207e-06,
      "loss": 0.8054,
      "step": 1499
    },
    {
      "epoch": 0.06216585851050603,
      "grad_norm": 0.43346622586250305,
      "learning_rate": 4.689377926975839e-06,
      "loss": 0.8323,
      "step": 1500
    },
    {
      "epoch": 0.0622073024161797,
      "grad_norm": 0.410091757774353,
      "learning_rate": 4.68917070744747e-06,
      "loss": 0.7571,
      "step": 1501
    },
    {
      "epoch": 0.06224874632185337,
      "grad_norm": 0.43019333481788635,
      "learning_rate": 4.6889634879191025e-06,
      "loss": 0.783,
      "step": 1502
    },
    {
      "epoch": 0.06229019022752704,
      "grad_norm": 0.47395995259284973,
      "learning_rate": 4.688756268390733e-06,
      "loss": 0.8215,
      "step": 1503
    },
    {
      "epoch": 0.06233163413320071,
      "grad_norm": 0.4449435770511627,
      "learning_rate": 4.688549048862365e-06,
      "loss": 0.7222,
      "step": 1504
    },
    {
      "epoch": 0.06237307803887438,
      "grad_norm": 0.476831316947937,
      "learning_rate": 4.688341829333997e-06,
      "loss": 0.7849,
      "step": 1505
    },
    {
      "epoch": 0.06241452194454805,
      "grad_norm": 0.45852184295654297,
      "learning_rate": 4.688134609805629e-06,
      "loss": 0.8142,
      "step": 1506
    },
    {
      "epoch": 0.06245596585022172,
      "grad_norm": 0.4783462882041931,
      "learning_rate": 4.68792739027726e-06,
      "loss": 0.7932,
      "step": 1507
    },
    {
      "epoch": 0.0624974097558954,
      "grad_norm": 0.4608703553676605,
      "learning_rate": 4.687720170748892e-06,
      "loss": 0.8369,
      "step": 1508
    },
    {
      "epoch": 0.06253885366156907,
      "grad_norm": 0.5077956318855286,
      "learning_rate": 4.687512951220524e-06,
      "loss": 0.8015,
      "step": 1509
    },
    {
      "epoch": 0.06258029756724273,
      "grad_norm": 0.4172244071960449,
      "learning_rate": 4.687305731692155e-06,
      "loss": 0.7051,
      "step": 1510
    },
    {
      "epoch": 0.06262174147291641,
      "grad_norm": 0.4445269703865051,
      "learning_rate": 4.687098512163787e-06,
      "loss": 0.7339,
      "step": 1511
    },
    {
      "epoch": 0.06266318537859007,
      "grad_norm": 0.4446474313735962,
      "learning_rate": 4.686891292635418e-06,
      "loss": 0.7917,
      "step": 1512
    },
    {
      "epoch": 0.06270462928426375,
      "grad_norm": 0.4380851089954376,
      "learning_rate": 4.68668407310705e-06,
      "loss": 0.7114,
      "step": 1513
    },
    {
      "epoch": 0.06274607318993743,
      "grad_norm": 0.4798642694950104,
      "learning_rate": 4.686476853578682e-06,
      "loss": 0.8422,
      "step": 1514
    },
    {
      "epoch": 0.06278751709561109,
      "grad_norm": 0.43944793939590454,
      "learning_rate": 4.686269634050313e-06,
      "loss": 0.7322,
      "step": 1515
    },
    {
      "epoch": 0.06282896100128477,
      "grad_norm": 0.5210661292076111,
      "learning_rate": 4.686062414521944e-06,
      "loss": 0.7836,
      "step": 1516
    },
    {
      "epoch": 0.06287040490695843,
      "grad_norm": 0.4589276611804962,
      "learning_rate": 4.685855194993577e-06,
      "loss": 0.783,
      "step": 1517
    },
    {
      "epoch": 0.0629118488126321,
      "grad_norm": 0.4895331561565399,
      "learning_rate": 4.685647975465209e-06,
      "loss": 0.8013,
      "step": 1518
    },
    {
      "epoch": 0.06295329271830577,
      "grad_norm": 0.43782681226730347,
      "learning_rate": 4.685440755936839e-06,
      "loss": 0.8188,
      "step": 1519
    },
    {
      "epoch": 0.06299473662397945,
      "grad_norm": 0.4066360294818878,
      "learning_rate": 4.685233536408472e-06,
      "loss": 0.7422,
      "step": 1520
    },
    {
      "epoch": 0.06303618052965311,
      "grad_norm": 0.4670545160770416,
      "learning_rate": 4.685026316880103e-06,
      "loss": 0.7659,
      "step": 1521
    },
    {
      "epoch": 0.06307762443532679,
      "grad_norm": 0.47140517830848694,
      "learning_rate": 4.684819097351735e-06,
      "loss": 0.761,
      "step": 1522
    },
    {
      "epoch": 0.06311906834100045,
      "grad_norm": 0.4558713138103485,
      "learning_rate": 4.684611877823367e-06,
      "loss": 0.7257,
      "step": 1523
    },
    {
      "epoch": 0.06316051224667413,
      "grad_norm": 0.39512622356414795,
      "learning_rate": 4.684404658294998e-06,
      "loss": 0.7466,
      "step": 1524
    },
    {
      "epoch": 0.06320195615234779,
      "grad_norm": 0.43386945128440857,
      "learning_rate": 4.68419743876663e-06,
      "loss": 0.7727,
      "step": 1525
    },
    {
      "epoch": 0.06324340005802147,
      "grad_norm": 0.4203483760356903,
      "learning_rate": 4.683990219238262e-06,
      "loss": 0.7289,
      "step": 1526
    },
    {
      "epoch": 0.06328484396369514,
      "grad_norm": 0.4208204746246338,
      "learning_rate": 4.683782999709893e-06,
      "loss": 0.7422,
      "step": 1527
    },
    {
      "epoch": 0.06332628786936881,
      "grad_norm": 0.4615058898925781,
      "learning_rate": 4.683575780181524e-06,
      "loss": 0.7581,
      "step": 1528
    },
    {
      "epoch": 0.06336773177504249,
      "grad_norm": 0.48876938223838806,
      "learning_rate": 4.683368560653157e-06,
      "loss": 0.7644,
      "step": 1529
    },
    {
      "epoch": 0.06340917568071615,
      "grad_norm": 0.42914554476737976,
      "learning_rate": 4.683161341124788e-06,
      "loss": 0.8068,
      "step": 1530
    },
    {
      "epoch": 0.06345061958638983,
      "grad_norm": 0.47351449728012085,
      "learning_rate": 4.682954121596419e-06,
      "loss": 0.7629,
      "step": 1531
    },
    {
      "epoch": 0.06349206349206349,
      "grad_norm": 0.4491868019104004,
      "learning_rate": 4.682746902068051e-06,
      "loss": 0.8113,
      "step": 1532
    },
    {
      "epoch": 0.06353350739773717,
      "grad_norm": 0.4416409134864807,
      "learning_rate": 4.682539682539683e-06,
      "loss": 0.731,
      "step": 1533
    },
    {
      "epoch": 0.06357495130341083,
      "grad_norm": 0.4276033341884613,
      "learning_rate": 4.682332463011314e-06,
      "loss": 0.7473,
      "step": 1534
    },
    {
      "epoch": 0.0636163952090845,
      "grad_norm": 0.4465705454349518,
      "learning_rate": 4.682125243482946e-06,
      "loss": 0.7928,
      "step": 1535
    },
    {
      "epoch": 0.06365783911475817,
      "grad_norm": 0.4503258168697357,
      "learning_rate": 4.681918023954578e-06,
      "loss": 0.8105,
      "step": 1536
    },
    {
      "epoch": 0.06369928302043185,
      "grad_norm": 0.4535083472728729,
      "learning_rate": 4.681710804426209e-06,
      "loss": 0.7063,
      "step": 1537
    },
    {
      "epoch": 0.06374072692610551,
      "grad_norm": 0.441040575504303,
      "learning_rate": 4.681503584897842e-06,
      "loss": 0.7783,
      "step": 1538
    },
    {
      "epoch": 0.06378217083177919,
      "grad_norm": 0.4295855760574341,
      "learning_rate": 4.681296365369473e-06,
      "loss": 0.748,
      "step": 1539
    },
    {
      "epoch": 0.06382361473745286,
      "grad_norm": 0.42273619771003723,
      "learning_rate": 4.6810891458411044e-06,
      "loss": 0.7822,
      "step": 1540
    },
    {
      "epoch": 0.06386505864312653,
      "grad_norm": 0.4403887987136841,
      "learning_rate": 4.680881926312736e-06,
      "loss": 0.7822,
      "step": 1541
    },
    {
      "epoch": 0.0639065025488002,
      "grad_norm": 0.42560821771621704,
      "learning_rate": 4.680674706784368e-06,
      "loss": 0.7515,
      "step": 1542
    },
    {
      "epoch": 0.06394794645447387,
      "grad_norm": 0.4214054346084595,
      "learning_rate": 4.6804674872559994e-06,
      "loss": 0.7244,
      "step": 1543
    },
    {
      "epoch": 0.06398939036014754,
      "grad_norm": 0.4270678758621216,
      "learning_rate": 4.680260267727631e-06,
      "loss": 0.7324,
      "step": 1544
    },
    {
      "epoch": 0.06403083426582121,
      "grad_norm": 0.45748379826545715,
      "learning_rate": 4.680053048199263e-06,
      "loss": 0.7888,
      "step": 1545
    },
    {
      "epoch": 0.06407227817149488,
      "grad_norm": 0.45477867126464844,
      "learning_rate": 4.6798458286708944e-06,
      "loss": 0.7793,
      "step": 1546
    },
    {
      "epoch": 0.06411372207716855,
      "grad_norm": 0.42887723445892334,
      "learning_rate": 4.679638609142526e-06,
      "loss": 0.7788,
      "step": 1547
    },
    {
      "epoch": 0.06415516598284222,
      "grad_norm": 0.44434836506843567,
      "learning_rate": 4.679431389614157e-06,
      "loss": 0.7415,
      "step": 1548
    },
    {
      "epoch": 0.06419660988851589,
      "grad_norm": 0.46402639150619507,
      "learning_rate": 4.6792241700857894e-06,
      "loss": 0.7993,
      "step": 1549
    },
    {
      "epoch": 0.06423805379418956,
      "grad_norm": 0.45409196615219116,
      "learning_rate": 4.679016950557421e-06,
      "loss": 0.8044,
      "step": 1550
    },
    {
      "epoch": 0.06427949769986324,
      "grad_norm": 0.4705032408237457,
      "learning_rate": 4.678809731029052e-06,
      "loss": 0.8037,
      "step": 1551
    },
    {
      "epoch": 0.0643209416055369,
      "grad_norm": 0.4447742998600006,
      "learning_rate": 4.6786025115006844e-06,
      "loss": 0.8101,
      "step": 1552
    },
    {
      "epoch": 0.06436238551121058,
      "grad_norm": 0.4853380620479584,
      "learning_rate": 4.678395291972316e-06,
      "loss": 0.8069,
      "step": 1553
    },
    {
      "epoch": 0.06440382941688425,
      "grad_norm": 0.4272081255912781,
      "learning_rate": 4.678188072443948e-06,
      "loss": 0.7444,
      "step": 1554
    },
    {
      "epoch": 0.06444527332255792,
      "grad_norm": 0.48527634143829346,
      "learning_rate": 4.677980852915579e-06,
      "loss": 0.8231,
      "step": 1555
    },
    {
      "epoch": 0.06448671722823159,
      "grad_norm": 0.4676365852355957,
      "learning_rate": 4.677773633387211e-06,
      "loss": 0.7561,
      "step": 1556
    },
    {
      "epoch": 0.06452816113390526,
      "grad_norm": 0.47746527194976807,
      "learning_rate": 4.677566413858842e-06,
      "loss": 0.7898,
      "step": 1557
    },
    {
      "epoch": 0.06456960503957893,
      "grad_norm": 0.44547533988952637,
      "learning_rate": 4.6773591943304744e-06,
      "loss": 0.7888,
      "step": 1558
    },
    {
      "epoch": 0.0646110489452526,
      "grad_norm": 0.4056016504764557,
      "learning_rate": 4.677151974802106e-06,
      "loss": 0.6843,
      "step": 1559
    },
    {
      "epoch": 0.06465249285092627,
      "grad_norm": 0.44302716851234436,
      "learning_rate": 4.676944755273737e-06,
      "loss": 0.821,
      "step": 1560
    },
    {
      "epoch": 0.06469393675659994,
      "grad_norm": 0.4118632674217224,
      "learning_rate": 4.6767375357453695e-06,
      "loss": 0.7083,
      "step": 1561
    },
    {
      "epoch": 0.0647353806622736,
      "grad_norm": 0.46236279606819153,
      "learning_rate": 4.676530316217001e-06,
      "loss": 0.7646,
      "step": 1562
    },
    {
      "epoch": 0.06477682456794728,
      "grad_norm": 0.49624866247177124,
      "learning_rate": 4.676323096688632e-06,
      "loss": 0.7404,
      "step": 1563
    },
    {
      "epoch": 0.06481826847362096,
      "grad_norm": 0.4422208070755005,
      "learning_rate": 4.676115877160264e-06,
      "loss": 0.7632,
      "step": 1564
    },
    {
      "epoch": 0.06485971237929462,
      "grad_norm": 0.48854729533195496,
      "learning_rate": 4.675908657631896e-06,
      "loss": 0.7744,
      "step": 1565
    },
    {
      "epoch": 0.0649011562849683,
      "grad_norm": 0.44912952184677124,
      "learning_rate": 4.675701438103527e-06,
      "loss": 0.7324,
      "step": 1566
    },
    {
      "epoch": 0.06494260019064196,
      "grad_norm": 0.468426376581192,
      "learning_rate": 4.675494218575159e-06,
      "loss": 0.7429,
      "step": 1567
    },
    {
      "epoch": 0.06498404409631564,
      "grad_norm": 0.4519636332988739,
      "learning_rate": 4.67528699904679e-06,
      "loss": 0.8245,
      "step": 1568
    },
    {
      "epoch": 0.0650254880019893,
      "grad_norm": 0.49408769607543945,
      "learning_rate": 4.675079779518422e-06,
      "loss": 0.7715,
      "step": 1569
    },
    {
      "epoch": 0.06506693190766298,
      "grad_norm": 0.46566078066825867,
      "learning_rate": 4.6748725599900545e-06,
      "loss": 0.7288,
      "step": 1570
    },
    {
      "epoch": 0.06510837581333664,
      "grad_norm": 0.43638646602630615,
      "learning_rate": 4.674665340461685e-06,
      "loss": 0.75,
      "step": 1571
    },
    {
      "epoch": 0.06514981971901032,
      "grad_norm": 0.4693780541419983,
      "learning_rate": 4.674458120933317e-06,
      "loss": 0.748,
      "step": 1572
    },
    {
      "epoch": 0.06519126362468398,
      "grad_norm": 0.4606259763240814,
      "learning_rate": 4.674250901404949e-06,
      "loss": 0.7391,
      "step": 1573
    },
    {
      "epoch": 0.06523270753035766,
      "grad_norm": 0.45809102058410645,
      "learning_rate": 4.674043681876581e-06,
      "loss": 0.7793,
      "step": 1574
    },
    {
      "epoch": 0.06527415143603134,
      "grad_norm": 0.43349623680114746,
      "learning_rate": 4.673836462348212e-06,
      "loss": 0.7623,
      "step": 1575
    },
    {
      "epoch": 0.065315595341705,
      "grad_norm": 0.4302397668361664,
      "learning_rate": 4.673629242819844e-06,
      "loss": 0.7102,
      "step": 1576
    },
    {
      "epoch": 0.06535703924737868,
      "grad_norm": 0.4467158317565918,
      "learning_rate": 4.673422023291475e-06,
      "loss": 0.7842,
      "step": 1577
    },
    {
      "epoch": 0.06539848315305234,
      "grad_norm": 0.5204900503158569,
      "learning_rate": 4.673214803763107e-06,
      "loss": 0.7776,
      "step": 1578
    },
    {
      "epoch": 0.06543992705872602,
      "grad_norm": 0.4935821294784546,
      "learning_rate": 4.673007584234739e-06,
      "loss": 0.8323,
      "step": 1579
    },
    {
      "epoch": 0.06548137096439968,
      "grad_norm": 0.450651079416275,
      "learning_rate": 4.67280036470637e-06,
      "loss": 0.7993,
      "step": 1580
    },
    {
      "epoch": 0.06552281487007336,
      "grad_norm": 0.41631844639778137,
      "learning_rate": 4.672593145178002e-06,
      "loss": 0.7034,
      "step": 1581
    },
    {
      "epoch": 0.06556425877574702,
      "grad_norm": 0.45050907135009766,
      "learning_rate": 4.672385925649634e-06,
      "loss": 0.7585,
      "step": 1582
    },
    {
      "epoch": 0.0656057026814207,
      "grad_norm": 0.4495653808116913,
      "learning_rate": 4.672178706121265e-06,
      "loss": 0.7335,
      "step": 1583
    },
    {
      "epoch": 0.06564714658709436,
      "grad_norm": 0.42504242062568665,
      "learning_rate": 4.671971486592896e-06,
      "loss": 0.7749,
      "step": 1584
    },
    {
      "epoch": 0.06568859049276804,
      "grad_norm": 0.44387149810791016,
      "learning_rate": 4.671764267064529e-06,
      "loss": 0.7383,
      "step": 1585
    },
    {
      "epoch": 0.0657300343984417,
      "grad_norm": 0.45584166049957275,
      "learning_rate": 4.671557047536161e-06,
      "loss": 0.744,
      "step": 1586
    },
    {
      "epoch": 0.06577147830411538,
      "grad_norm": 0.4470459520816803,
      "learning_rate": 4.671349828007791e-06,
      "loss": 0.7817,
      "step": 1587
    },
    {
      "epoch": 0.06581292220978906,
      "grad_norm": 0.45685678720474243,
      "learning_rate": 4.671142608479424e-06,
      "loss": 0.7222,
      "step": 1588
    },
    {
      "epoch": 0.06585436611546272,
      "grad_norm": 0.4680252969264984,
      "learning_rate": 4.670935388951055e-06,
      "loss": 0.7336,
      "step": 1589
    },
    {
      "epoch": 0.0658958100211364,
      "grad_norm": 0.42901307344436646,
      "learning_rate": 4.670728169422687e-06,
      "loss": 0.7434,
      "step": 1590
    },
    {
      "epoch": 0.06593725392681006,
      "grad_norm": 0.45285874605178833,
      "learning_rate": 4.670520949894319e-06,
      "loss": 0.7166,
      "step": 1591
    },
    {
      "epoch": 0.06597869783248374,
      "grad_norm": 0.4365731477737427,
      "learning_rate": 4.67031373036595e-06,
      "loss": 0.7371,
      "step": 1592
    },
    {
      "epoch": 0.0660201417381574,
      "grad_norm": 0.44257187843322754,
      "learning_rate": 4.670106510837581e-06,
      "loss": 0.7249,
      "step": 1593
    },
    {
      "epoch": 0.06606158564383108,
      "grad_norm": 0.4507882297039032,
      "learning_rate": 4.669899291309214e-06,
      "loss": 0.762,
      "step": 1594
    },
    {
      "epoch": 0.06610302954950474,
      "grad_norm": 0.43328067660331726,
      "learning_rate": 4.669692071780845e-06,
      "loss": 0.7585,
      "step": 1595
    },
    {
      "epoch": 0.06614447345517842,
      "grad_norm": 0.44777682423591614,
      "learning_rate": 4.669484852252476e-06,
      "loss": 0.7573,
      "step": 1596
    },
    {
      "epoch": 0.06618591736085208,
      "grad_norm": 0.4552536606788635,
      "learning_rate": 4.669277632724109e-06,
      "loss": 0.7507,
      "step": 1597
    },
    {
      "epoch": 0.06622736126652576,
      "grad_norm": 0.48245152831077576,
      "learning_rate": 4.66907041319574e-06,
      "loss": 0.7755,
      "step": 1598
    },
    {
      "epoch": 0.06626880517219942,
      "grad_norm": 0.47234493494033813,
      "learning_rate": 4.6688631936673714e-06,
      "loss": 0.7712,
      "step": 1599
    },
    {
      "epoch": 0.0663102490778731,
      "grad_norm": 0.5068262815475464,
      "learning_rate": 4.668655974139003e-06,
      "loss": 0.7896,
      "step": 1600
    },
    {
      "epoch": 0.06635169298354678,
      "grad_norm": 0.45288634300231934,
      "learning_rate": 4.668448754610635e-06,
      "loss": 0.7129,
      "step": 1601
    },
    {
      "epoch": 0.06639313688922044,
      "grad_norm": 0.4196891486644745,
      "learning_rate": 4.6682415350822664e-06,
      "loss": 0.718,
      "step": 1602
    },
    {
      "epoch": 0.06643458079489412,
      "grad_norm": 0.473163902759552,
      "learning_rate": 4.668034315553898e-06,
      "loss": 0.7795,
      "step": 1603
    },
    {
      "epoch": 0.06647602470056778,
      "grad_norm": 0.4324408769607544,
      "learning_rate": 4.66782709602553e-06,
      "loss": 0.7424,
      "step": 1604
    },
    {
      "epoch": 0.06651746860624146,
      "grad_norm": 0.46043696999549866,
      "learning_rate": 4.6676198764971614e-06,
      "loss": 0.8027,
      "step": 1605
    },
    {
      "epoch": 0.06655891251191512,
      "grad_norm": 0.4144105315208435,
      "learning_rate": 4.667412656968794e-06,
      "loss": 0.7634,
      "step": 1606
    },
    {
      "epoch": 0.0666003564175888,
      "grad_norm": 0.4830254912376404,
      "learning_rate": 4.667205437440425e-06,
      "loss": 0.7607,
      "step": 1607
    },
    {
      "epoch": 0.06664180032326246,
      "grad_norm": 0.4808370769023895,
      "learning_rate": 4.6669982179120564e-06,
      "loss": 0.8357,
      "step": 1608
    },
    {
      "epoch": 0.06668324422893614,
      "grad_norm": 0.4250320792198181,
      "learning_rate": 4.666790998383688e-06,
      "loss": 0.7776,
      "step": 1609
    },
    {
      "epoch": 0.0667246881346098,
      "grad_norm": 0.4177250862121582,
      "learning_rate": 4.66658377885532e-06,
      "loss": 0.7646,
      "step": 1610
    },
    {
      "epoch": 0.06676613204028348,
      "grad_norm": 0.3989412486553192,
      "learning_rate": 4.6663765593269514e-06,
      "loss": 0.7439,
      "step": 1611
    },
    {
      "epoch": 0.06680757594595715,
      "grad_norm": 0.4994300305843353,
      "learning_rate": 4.666169339798583e-06,
      "loss": 0.8162,
      "step": 1612
    },
    {
      "epoch": 0.06684901985163082,
      "grad_norm": 0.422446072101593,
      "learning_rate": 4.665962120270215e-06,
      "loss": 0.743,
      "step": 1613
    },
    {
      "epoch": 0.0668904637573045,
      "grad_norm": 0.4232836067676544,
      "learning_rate": 4.6657549007418464e-06,
      "loss": 0.7207,
      "step": 1614
    },
    {
      "epoch": 0.06693190766297816,
      "grad_norm": 0.45700791478157043,
      "learning_rate": 4.665547681213478e-06,
      "loss": 0.7759,
      "step": 1615
    },
    {
      "epoch": 0.06697335156865183,
      "grad_norm": 0.4804055690765381,
      "learning_rate": 4.665340461685109e-06,
      "loss": 0.76,
      "step": 1616
    },
    {
      "epoch": 0.0670147954743255,
      "grad_norm": 0.45987018942832947,
      "learning_rate": 4.6651332421567414e-06,
      "loss": 0.7122,
      "step": 1617
    },
    {
      "epoch": 0.06705623937999917,
      "grad_norm": 0.477420836687088,
      "learning_rate": 4.664926022628373e-06,
      "loss": 0.8225,
      "step": 1618
    },
    {
      "epoch": 0.06709768328567284,
      "grad_norm": 0.4248233437538147,
      "learning_rate": 4.664718803100004e-06,
      "loss": 0.7201,
      "step": 1619
    },
    {
      "epoch": 0.06713912719134651,
      "grad_norm": 0.4365939497947693,
      "learning_rate": 4.664511583571636e-06,
      "loss": 0.7185,
      "step": 1620
    },
    {
      "epoch": 0.06718057109702018,
      "grad_norm": 0.4280262887477875,
      "learning_rate": 4.664304364043268e-06,
      "loss": 0.8062,
      "step": 1621
    },
    {
      "epoch": 0.06722201500269386,
      "grad_norm": 0.43703097105026245,
      "learning_rate": 4.6640971445149e-06,
      "loss": 0.7522,
      "step": 1622
    },
    {
      "epoch": 0.06726345890836752,
      "grad_norm": 0.4505535066127777,
      "learning_rate": 4.6638899249865315e-06,
      "loss": 0.7996,
      "step": 1623
    },
    {
      "epoch": 0.0673049028140412,
      "grad_norm": 0.4685426354408264,
      "learning_rate": 4.663682705458163e-06,
      "loss": 0.781,
      "step": 1624
    },
    {
      "epoch": 0.06734634671971487,
      "grad_norm": 0.43711674213409424,
      "learning_rate": 4.663475485929794e-06,
      "loss": 0.7649,
      "step": 1625
    },
    {
      "epoch": 0.06738779062538854,
      "grad_norm": 0.43566495180130005,
      "learning_rate": 4.6632682664014265e-06,
      "loss": 0.733,
      "step": 1626
    },
    {
      "epoch": 0.06742923453106221,
      "grad_norm": 0.4581296741962433,
      "learning_rate": 4.663061046873058e-06,
      "loss": 0.7876,
      "step": 1627
    },
    {
      "epoch": 0.06747067843673588,
      "grad_norm": 0.5490488409996033,
      "learning_rate": 4.662853827344689e-06,
      "loss": 0.9067,
      "step": 1628
    },
    {
      "epoch": 0.06751212234240955,
      "grad_norm": 0.46907538175582886,
      "learning_rate": 4.662646607816321e-06,
      "loss": 0.7646,
      "step": 1629
    },
    {
      "epoch": 0.06755356624808322,
      "grad_norm": 0.49457135796546936,
      "learning_rate": 4.662439388287953e-06,
      "loss": 0.7527,
      "step": 1630
    },
    {
      "epoch": 0.0675950101537569,
      "grad_norm": 0.4529842436313629,
      "learning_rate": 4.662232168759584e-06,
      "loss": 0.7402,
      "step": 1631
    },
    {
      "epoch": 0.06763645405943056,
      "grad_norm": 0.45160120725631714,
      "learning_rate": 4.662024949231216e-06,
      "loss": 0.7551,
      "step": 1632
    },
    {
      "epoch": 0.06767789796510423,
      "grad_norm": 0.4262199103832245,
      "learning_rate": 4.661817729702848e-06,
      "loss": 0.7288,
      "step": 1633
    },
    {
      "epoch": 0.0677193418707779,
      "grad_norm": 0.454715758562088,
      "learning_rate": 4.661610510174479e-06,
      "loss": 0.7764,
      "step": 1634
    },
    {
      "epoch": 0.06776078577645157,
      "grad_norm": 0.5298017263412476,
      "learning_rate": 4.661403290646111e-06,
      "loss": 0.8652,
      "step": 1635
    },
    {
      "epoch": 0.06780222968212524,
      "grad_norm": 0.4710478186607361,
      "learning_rate": 4.661196071117742e-06,
      "loss": 0.7661,
      "step": 1636
    },
    {
      "epoch": 0.06784367358779891,
      "grad_norm": 0.5291531085968018,
      "learning_rate": 4.660988851589374e-06,
      "loss": 0.7936,
      "step": 1637
    },
    {
      "epoch": 0.06788511749347259,
      "grad_norm": 0.4823746979236603,
      "learning_rate": 4.660781632061006e-06,
      "loss": 0.7834,
      "step": 1638
    },
    {
      "epoch": 0.06792656139914625,
      "grad_norm": 0.43640822172164917,
      "learning_rate": 4.660574412532637e-06,
      "loss": 0.8044,
      "step": 1639
    },
    {
      "epoch": 0.06796800530481993,
      "grad_norm": 0.48759734630584717,
      "learning_rate": 4.660367193004269e-06,
      "loss": 0.7856,
      "step": 1640
    },
    {
      "epoch": 0.0680094492104936,
      "grad_norm": 0.49422430992126465,
      "learning_rate": 4.660159973475901e-06,
      "loss": 0.8447,
      "step": 1641
    },
    {
      "epoch": 0.06805089311616727,
      "grad_norm": 0.48106130957603455,
      "learning_rate": 4.659952753947533e-06,
      "loss": 0.7454,
      "step": 1642
    },
    {
      "epoch": 0.06809233702184093,
      "grad_norm": 0.4398304224014282,
      "learning_rate": 4.659745534419164e-06,
      "loss": 0.7585,
      "step": 1643
    },
    {
      "epoch": 0.06813378092751461,
      "grad_norm": 0.448781818151474,
      "learning_rate": 4.659538314890796e-06,
      "loss": 0.7178,
      "step": 1644
    },
    {
      "epoch": 0.06817522483318827,
      "grad_norm": 0.43553951382637024,
      "learning_rate": 4.659331095362427e-06,
      "loss": 0.7378,
      "step": 1645
    },
    {
      "epoch": 0.06821666873886195,
      "grad_norm": 0.4915466904640198,
      "learning_rate": 4.659123875834059e-06,
      "loss": 0.8108,
      "step": 1646
    },
    {
      "epoch": 0.06825811264453562,
      "grad_norm": 0.43068745732307434,
      "learning_rate": 4.658916656305691e-06,
      "loss": 0.759,
      "step": 1647
    },
    {
      "epoch": 0.06829955655020929,
      "grad_norm": 0.48811671137809753,
      "learning_rate": 4.658709436777322e-06,
      "loss": 0.7252,
      "step": 1648
    },
    {
      "epoch": 0.06834100045588297,
      "grad_norm": 0.42943674325942993,
      "learning_rate": 4.658502217248954e-06,
      "loss": 0.7346,
      "step": 1649
    },
    {
      "epoch": 0.06838244436155663,
      "grad_norm": 0.45456504821777344,
      "learning_rate": 4.658294997720586e-06,
      "loss": 0.8611,
      "step": 1650
    },
    {
      "epoch": 0.06842388826723031,
      "grad_norm": 0.471125990152359,
      "learning_rate": 4.658087778192217e-06,
      "loss": 0.7593,
      "step": 1651
    },
    {
      "epoch": 0.06846533217290397,
      "grad_norm": 0.42458441853523254,
      "learning_rate": 4.657880558663848e-06,
      "loss": 0.7579,
      "step": 1652
    },
    {
      "epoch": 0.06850677607857765,
      "grad_norm": 0.48126089572906494,
      "learning_rate": 4.657673339135481e-06,
      "loss": 0.7642,
      "step": 1653
    },
    {
      "epoch": 0.06854821998425131,
      "grad_norm": 0.48993566632270813,
      "learning_rate": 4.657466119607112e-06,
      "loss": 0.8167,
      "step": 1654
    },
    {
      "epoch": 0.06858966388992499,
      "grad_norm": 0.4605776369571686,
      "learning_rate": 4.657258900078743e-06,
      "loss": 0.7513,
      "step": 1655
    },
    {
      "epoch": 0.06863110779559865,
      "grad_norm": 0.4563843905925751,
      "learning_rate": 4.657051680550376e-06,
      "loss": 0.7415,
      "step": 1656
    },
    {
      "epoch": 0.06867255170127233,
      "grad_norm": 0.47484487295150757,
      "learning_rate": 4.656844461022007e-06,
      "loss": 0.7703,
      "step": 1657
    },
    {
      "epoch": 0.068713995606946,
      "grad_norm": 0.41562286019325256,
      "learning_rate": 4.656637241493639e-06,
      "loss": 0.7227,
      "step": 1658
    },
    {
      "epoch": 0.06875543951261967,
      "grad_norm": 0.46570515632629395,
      "learning_rate": 4.656430021965271e-06,
      "loss": 0.772,
      "step": 1659
    },
    {
      "epoch": 0.06879688341829333,
      "grad_norm": 0.407827228307724,
      "learning_rate": 4.656222802436902e-06,
      "loss": 0.7377,
      "step": 1660
    },
    {
      "epoch": 0.06883832732396701,
      "grad_norm": 0.46641311049461365,
      "learning_rate": 4.6560155829085334e-06,
      "loss": 0.7974,
      "step": 1661
    },
    {
      "epoch": 0.06887977122964069,
      "grad_norm": 0.45002833008766174,
      "learning_rate": 4.655808363380166e-06,
      "loss": 0.7939,
      "step": 1662
    },
    {
      "epoch": 0.06892121513531435,
      "grad_norm": 0.4493491053581238,
      "learning_rate": 4.655601143851797e-06,
      "loss": 0.8389,
      "step": 1663
    },
    {
      "epoch": 0.06896265904098803,
      "grad_norm": 0.42593252658843994,
      "learning_rate": 4.6553939243234284e-06,
      "loss": 0.6902,
      "step": 1664
    },
    {
      "epoch": 0.06900410294666169,
      "grad_norm": 0.4662606120109558,
      "learning_rate": 4.655186704795061e-06,
      "loss": 0.7375,
      "step": 1665
    },
    {
      "epoch": 0.06904554685233537,
      "grad_norm": 0.4615080654621124,
      "learning_rate": 4.654979485266692e-06,
      "loss": 0.7437,
      "step": 1666
    },
    {
      "epoch": 0.06908699075800903,
      "grad_norm": 0.44211065769195557,
      "learning_rate": 4.6547722657383234e-06,
      "loss": 0.7905,
      "step": 1667
    },
    {
      "epoch": 0.06912843466368271,
      "grad_norm": 0.465715616941452,
      "learning_rate": 4.654565046209955e-06,
      "loss": 0.7839,
      "step": 1668
    },
    {
      "epoch": 0.06916987856935637,
      "grad_norm": 0.4777056872844696,
      "learning_rate": 4.654357826681587e-06,
      "loss": 0.7595,
      "step": 1669
    },
    {
      "epoch": 0.06921132247503005,
      "grad_norm": 0.4435918629169464,
      "learning_rate": 4.6541506071532184e-06,
      "loss": 0.8022,
      "step": 1670
    },
    {
      "epoch": 0.06925276638070371,
      "grad_norm": 0.4559552073478699,
      "learning_rate": 4.65394338762485e-06,
      "loss": 0.761,
      "step": 1671
    },
    {
      "epoch": 0.06929421028637739,
      "grad_norm": 0.444330632686615,
      "learning_rate": 4.653736168096481e-06,
      "loss": 0.7439,
      "step": 1672
    },
    {
      "epoch": 0.06933565419205105,
      "grad_norm": 0.4851124584674835,
      "learning_rate": 4.6535289485681134e-06,
      "loss": 0.8171,
      "step": 1673
    },
    {
      "epoch": 0.06937709809772473,
      "grad_norm": 0.43949106335639954,
      "learning_rate": 4.653321729039746e-06,
      "loss": 0.8054,
      "step": 1674
    },
    {
      "epoch": 0.0694185420033984,
      "grad_norm": 0.435026079416275,
      "learning_rate": 4.653114509511377e-06,
      "loss": 0.8062,
      "step": 1675
    },
    {
      "epoch": 0.06945998590907207,
      "grad_norm": 0.43054988980293274,
      "learning_rate": 4.6529072899830084e-06,
      "loss": 0.7554,
      "step": 1676
    },
    {
      "epoch": 0.06950142981474575,
      "grad_norm": 0.4451897144317627,
      "learning_rate": 4.65270007045464e-06,
      "loss": 0.7673,
      "step": 1677
    },
    {
      "epoch": 0.06954287372041941,
      "grad_norm": 0.4691118896007538,
      "learning_rate": 4.652492850926272e-06,
      "loss": 0.781,
      "step": 1678
    },
    {
      "epoch": 0.06958431762609309,
      "grad_norm": 0.4606602191925049,
      "learning_rate": 4.6522856313979035e-06,
      "loss": 0.8108,
      "step": 1679
    },
    {
      "epoch": 0.06962576153176675,
      "grad_norm": 0.47613826394081116,
      "learning_rate": 4.652078411869535e-06,
      "loss": 0.8213,
      "step": 1680
    },
    {
      "epoch": 0.06966720543744043,
      "grad_norm": 0.45324766635894775,
      "learning_rate": 4.651871192341166e-06,
      "loss": 0.8015,
      "step": 1681
    },
    {
      "epoch": 0.06970864934311409,
      "grad_norm": 0.48258739709854126,
      "learning_rate": 4.6516639728127985e-06,
      "loss": 0.8423,
      "step": 1682
    },
    {
      "epoch": 0.06975009324878777,
      "grad_norm": 0.40911394357681274,
      "learning_rate": 4.65145675328443e-06,
      "loss": 0.6725,
      "step": 1683
    },
    {
      "epoch": 0.06979153715446143,
      "grad_norm": 0.45600974559783936,
      "learning_rate": 4.651249533756061e-06,
      "loss": 0.7744,
      "step": 1684
    },
    {
      "epoch": 0.06983298106013511,
      "grad_norm": 0.4636862277984619,
      "learning_rate": 4.6510423142276935e-06,
      "loss": 0.7986,
      "step": 1685
    },
    {
      "epoch": 0.06987442496580878,
      "grad_norm": 0.44224169850349426,
      "learning_rate": 4.650835094699325e-06,
      "loss": 0.7793,
      "step": 1686
    },
    {
      "epoch": 0.06991586887148245,
      "grad_norm": 0.4245778024196625,
      "learning_rate": 4.650627875170956e-06,
      "loss": 0.7649,
      "step": 1687
    },
    {
      "epoch": 0.06995731277715612,
      "grad_norm": 0.4712643325328827,
      "learning_rate": 4.650420655642588e-06,
      "loss": 0.792,
      "step": 1688
    },
    {
      "epoch": 0.06999875668282979,
      "grad_norm": 0.41124558448791504,
      "learning_rate": 4.65021343611422e-06,
      "loss": 0.7478,
      "step": 1689
    },
    {
      "epoch": 0.07004020058850347,
      "grad_norm": 0.4189101457595825,
      "learning_rate": 4.650006216585851e-06,
      "loss": 0.7231,
      "step": 1690
    },
    {
      "epoch": 0.07008164449417713,
      "grad_norm": 0.44122588634490967,
      "learning_rate": 4.6497989970574835e-06,
      "loss": 0.7632,
      "step": 1691
    },
    {
      "epoch": 0.0701230883998508,
      "grad_norm": 0.44176021218299866,
      "learning_rate": 4.649591777529115e-06,
      "loss": 0.7637,
      "step": 1692
    },
    {
      "epoch": 0.07016453230552447,
      "grad_norm": 0.47611570358276367,
      "learning_rate": 4.649384558000746e-06,
      "loss": 0.79,
      "step": 1693
    },
    {
      "epoch": 0.07020597621119815,
      "grad_norm": 0.4398527145385742,
      "learning_rate": 4.6491773384723785e-06,
      "loss": 0.7966,
      "step": 1694
    },
    {
      "epoch": 0.07024742011687181,
      "grad_norm": 0.49485838413238525,
      "learning_rate": 4.64897011894401e-06,
      "loss": 0.7539,
      "step": 1695
    },
    {
      "epoch": 0.07028886402254549,
      "grad_norm": 0.46022871136665344,
      "learning_rate": 4.648762899415641e-06,
      "loss": 0.771,
      "step": 1696
    },
    {
      "epoch": 0.07033030792821915,
      "grad_norm": 0.4651918113231659,
      "learning_rate": 4.648555679887273e-06,
      "loss": 0.7654,
      "step": 1697
    },
    {
      "epoch": 0.07037175183389283,
      "grad_norm": 0.4170083701610565,
      "learning_rate": 4.648348460358905e-06,
      "loss": 0.73,
      "step": 1698
    },
    {
      "epoch": 0.0704131957395665,
      "grad_norm": 0.43438443541526794,
      "learning_rate": 4.648141240830536e-06,
      "loss": 0.7209,
      "step": 1699
    },
    {
      "epoch": 0.07045463964524017,
      "grad_norm": 0.4811383783817291,
      "learning_rate": 4.647934021302168e-06,
      "loss": 0.7734,
      "step": 1700
    },
    {
      "epoch": 0.07049608355091384,
      "grad_norm": 0.47572991251945496,
      "learning_rate": 4.6477268017738e-06,
      "loss": 0.7439,
      "step": 1701
    },
    {
      "epoch": 0.0705375274565875,
      "grad_norm": 0.46861550211906433,
      "learning_rate": 4.647519582245431e-06,
      "loss": 0.8271,
      "step": 1702
    },
    {
      "epoch": 0.07057897136226118,
      "grad_norm": 0.534471333026886,
      "learning_rate": 4.647312362717063e-06,
      "loss": 0.7556,
      "step": 1703
    },
    {
      "epoch": 0.07062041526793485,
      "grad_norm": 0.5152859091758728,
      "learning_rate": 4.647105143188694e-06,
      "loss": 0.7727,
      "step": 1704
    },
    {
      "epoch": 0.07066185917360852,
      "grad_norm": 0.4691943824291229,
      "learning_rate": 4.646897923660326e-06,
      "loss": 0.7405,
      "step": 1705
    },
    {
      "epoch": 0.07070330307928219,
      "grad_norm": 0.45241427421569824,
      "learning_rate": 4.646690704131958e-06,
      "loss": 0.7932,
      "step": 1706
    },
    {
      "epoch": 0.07074474698495586,
      "grad_norm": 0.4873049855232239,
      "learning_rate": 4.646483484603589e-06,
      "loss": 0.781,
      "step": 1707
    },
    {
      "epoch": 0.07078619089062953,
      "grad_norm": 0.4366340637207031,
      "learning_rate": 4.646276265075221e-06,
      "loss": 0.7665,
      "step": 1708
    },
    {
      "epoch": 0.0708276347963032,
      "grad_norm": 0.4557400047779083,
      "learning_rate": 4.646069045546853e-06,
      "loss": 0.7502,
      "step": 1709
    },
    {
      "epoch": 0.07086907870197687,
      "grad_norm": 0.44708120822906494,
      "learning_rate": 4.645861826018485e-06,
      "loss": 0.7664,
      "step": 1710
    },
    {
      "epoch": 0.07091052260765054,
      "grad_norm": 0.4516519606113434,
      "learning_rate": 4.645654606490116e-06,
      "loss": 0.7607,
      "step": 1711
    },
    {
      "epoch": 0.07095196651332422,
      "grad_norm": 0.4604353904724121,
      "learning_rate": 4.645447386961748e-06,
      "loss": 0.7756,
      "step": 1712
    },
    {
      "epoch": 0.07099341041899788,
      "grad_norm": 0.47678330540657043,
      "learning_rate": 4.645240167433379e-06,
      "loss": 0.7566,
      "step": 1713
    },
    {
      "epoch": 0.07103485432467156,
      "grad_norm": 0.4621085226535797,
      "learning_rate": 4.645032947905011e-06,
      "loss": 0.7588,
      "step": 1714
    },
    {
      "epoch": 0.07107629823034523,
      "grad_norm": 0.46358540654182434,
      "learning_rate": 4.644825728376643e-06,
      "loss": 0.7412,
      "step": 1715
    },
    {
      "epoch": 0.0711177421360189,
      "grad_norm": 0.4876124858856201,
      "learning_rate": 4.644618508848274e-06,
      "loss": 0.7544,
      "step": 1716
    },
    {
      "epoch": 0.07115918604169257,
      "grad_norm": 0.49796178936958313,
      "learning_rate": 4.644411289319906e-06,
      "loss": 0.8438,
      "step": 1717
    },
    {
      "epoch": 0.07120062994736624,
      "grad_norm": 0.44890812039375305,
      "learning_rate": 4.644204069791538e-06,
      "loss": 0.7877,
      "step": 1718
    },
    {
      "epoch": 0.0712420738530399,
      "grad_norm": 0.4732809066772461,
      "learning_rate": 4.643996850263169e-06,
      "loss": 0.748,
      "step": 1719
    },
    {
      "epoch": 0.07128351775871358,
      "grad_norm": 0.46018171310424805,
      "learning_rate": 4.6437896307348004e-06,
      "loss": 0.7466,
      "step": 1720
    },
    {
      "epoch": 0.07132496166438725,
      "grad_norm": 0.4570143222808838,
      "learning_rate": 4.643582411206433e-06,
      "loss": 0.7585,
      "step": 1721
    },
    {
      "epoch": 0.07136640557006092,
      "grad_norm": 0.4149571359157562,
      "learning_rate": 4.643375191678064e-06,
      "loss": 0.7573,
      "step": 1722
    },
    {
      "epoch": 0.0714078494757346,
      "grad_norm": 0.43422549962997437,
      "learning_rate": 4.6431679721496954e-06,
      "loss": 0.77,
      "step": 1723
    },
    {
      "epoch": 0.07144929338140826,
      "grad_norm": 0.49335604906082153,
      "learning_rate": 4.642960752621327e-06,
      "loss": 0.8105,
      "step": 1724
    },
    {
      "epoch": 0.07149073728708194,
      "grad_norm": 0.6594836115837097,
      "learning_rate": 4.642753533092959e-06,
      "loss": 0.7544,
      "step": 1725
    },
    {
      "epoch": 0.0715321811927556,
      "grad_norm": 0.44890618324279785,
      "learning_rate": 4.642546313564591e-06,
      "loss": 0.7498,
      "step": 1726
    },
    {
      "epoch": 0.07157362509842928,
      "grad_norm": 0.455856591463089,
      "learning_rate": 4.642339094036223e-06,
      "loss": 0.7451,
      "step": 1727
    },
    {
      "epoch": 0.07161506900410294,
      "grad_norm": 0.4311956465244293,
      "learning_rate": 4.642131874507854e-06,
      "loss": 0.7832,
      "step": 1728
    },
    {
      "epoch": 0.07165651290977662,
      "grad_norm": 0.44108057022094727,
      "learning_rate": 4.6419246549794854e-06,
      "loss": 0.7881,
      "step": 1729
    },
    {
      "epoch": 0.07169795681545028,
      "grad_norm": 0.4821377396583557,
      "learning_rate": 4.641717435451118e-06,
      "loss": 0.8066,
      "step": 1730
    },
    {
      "epoch": 0.07173940072112396,
      "grad_norm": 0.46589818596839905,
      "learning_rate": 4.641510215922749e-06,
      "loss": 0.7408,
      "step": 1731
    },
    {
      "epoch": 0.07178084462679762,
      "grad_norm": 0.4932117164134979,
      "learning_rate": 4.6413029963943804e-06,
      "loss": 0.8564,
      "step": 1732
    },
    {
      "epoch": 0.0718222885324713,
      "grad_norm": 0.4159393906593323,
      "learning_rate": 4.641095776866012e-06,
      "loss": 0.7754,
      "step": 1733
    },
    {
      "epoch": 0.07186373243814496,
      "grad_norm": 0.5250783562660217,
      "learning_rate": 4.640888557337644e-06,
      "loss": 0.8098,
      "step": 1734
    },
    {
      "epoch": 0.07190517634381864,
      "grad_norm": 0.43870437145233154,
      "learning_rate": 4.6406813378092754e-06,
      "loss": 0.8259,
      "step": 1735
    },
    {
      "epoch": 0.07194662024949232,
      "grad_norm": 0.4944676160812378,
      "learning_rate": 4.640474118280907e-06,
      "loss": 0.7622,
      "step": 1736
    },
    {
      "epoch": 0.07198806415516598,
      "grad_norm": 0.46410930156707764,
      "learning_rate": 4.640266898752539e-06,
      "loss": 0.7576,
      "step": 1737
    },
    {
      "epoch": 0.07202950806083966,
      "grad_norm": 0.46144595742225647,
      "learning_rate": 4.6400596792241705e-06,
      "loss": 0.79,
      "step": 1738
    },
    {
      "epoch": 0.07207095196651332,
      "grad_norm": 0.4611121416091919,
      "learning_rate": 4.639852459695802e-06,
      "loss": 0.7866,
      "step": 1739
    },
    {
      "epoch": 0.072112395872187,
      "grad_norm": 0.4492151439189911,
      "learning_rate": 4.639645240167433e-06,
      "loss": 0.781,
      "step": 1740
    },
    {
      "epoch": 0.07215383977786066,
      "grad_norm": 0.4567330479621887,
      "learning_rate": 4.6394380206390655e-06,
      "loss": 0.7595,
      "step": 1741
    },
    {
      "epoch": 0.07219528368353434,
      "grad_norm": 0.4769268333911896,
      "learning_rate": 4.639230801110697e-06,
      "loss": 0.8462,
      "step": 1742
    },
    {
      "epoch": 0.072236727589208,
      "grad_norm": 0.46909695863723755,
      "learning_rate": 4.639023581582329e-06,
      "loss": 0.7361,
      "step": 1743
    },
    {
      "epoch": 0.07227817149488168,
      "grad_norm": 0.4566887319087982,
      "learning_rate": 4.6388163620539605e-06,
      "loss": 0.7832,
      "step": 1744
    },
    {
      "epoch": 0.07231961540055534,
      "grad_norm": 0.4635995626449585,
      "learning_rate": 4.638609142525592e-06,
      "loss": 0.8091,
      "step": 1745
    },
    {
      "epoch": 0.07236105930622902,
      "grad_norm": 0.4519689083099365,
      "learning_rate": 4.638401922997224e-06,
      "loss": 0.7576,
      "step": 1746
    },
    {
      "epoch": 0.07240250321190268,
      "grad_norm": 0.476813405752182,
      "learning_rate": 4.6381947034688555e-06,
      "loss": 0.7925,
      "step": 1747
    },
    {
      "epoch": 0.07244394711757636,
      "grad_norm": 0.4610627293586731,
      "learning_rate": 4.637987483940487e-06,
      "loss": 0.7559,
      "step": 1748
    },
    {
      "epoch": 0.07248539102325004,
      "grad_norm": 0.4306154251098633,
      "learning_rate": 4.637780264412118e-06,
      "loss": 0.7178,
      "step": 1749
    },
    {
      "epoch": 0.0725268349289237,
      "grad_norm": 0.46092885732650757,
      "learning_rate": 4.6375730448837505e-06,
      "loss": 0.7737,
      "step": 1750
    },
    {
      "epoch": 0.07256827883459738,
      "grad_norm": 0.465017169713974,
      "learning_rate": 4.637365825355382e-06,
      "loss": 0.76,
      "step": 1751
    },
    {
      "epoch": 0.07260972274027104,
      "grad_norm": 0.4280245304107666,
      "learning_rate": 4.637158605827013e-06,
      "loss": 0.7446,
      "step": 1752
    },
    {
      "epoch": 0.07265116664594472,
      "grad_norm": 0.4563804268836975,
      "learning_rate": 4.6369513862986455e-06,
      "loss": 0.7339,
      "step": 1753
    },
    {
      "epoch": 0.07269261055161838,
      "grad_norm": 0.5036217570304871,
      "learning_rate": 4.636744166770277e-06,
      "loss": 0.7544,
      "step": 1754
    },
    {
      "epoch": 0.07273405445729206,
      "grad_norm": 0.4994726777076721,
      "learning_rate": 4.636536947241908e-06,
      "loss": 0.8147,
      "step": 1755
    },
    {
      "epoch": 0.07277549836296572,
      "grad_norm": 0.45011934638023376,
      "learning_rate": 4.63632972771354e-06,
      "loss": 0.8234,
      "step": 1756
    },
    {
      "epoch": 0.0728169422686394,
      "grad_norm": 0.44011151790618896,
      "learning_rate": 4.636122508185172e-06,
      "loss": 0.7568,
      "step": 1757
    },
    {
      "epoch": 0.07285838617431306,
      "grad_norm": 0.4454973340034485,
      "learning_rate": 4.635915288656803e-06,
      "loss": 0.8315,
      "step": 1758
    },
    {
      "epoch": 0.07289983007998674,
      "grad_norm": 0.45458829402923584,
      "learning_rate": 4.6357080691284355e-06,
      "loss": 0.7703,
      "step": 1759
    },
    {
      "epoch": 0.07294127398566042,
      "grad_norm": 0.43965891003608704,
      "learning_rate": 4.635500849600067e-06,
      "loss": 0.7583,
      "step": 1760
    },
    {
      "epoch": 0.07298271789133408,
      "grad_norm": 0.43473947048187256,
      "learning_rate": 4.635293630071698e-06,
      "loss": 0.7438,
      "step": 1761
    },
    {
      "epoch": 0.07302416179700776,
      "grad_norm": 0.46284231543540955,
      "learning_rate": 4.6350864105433305e-06,
      "loss": 0.7578,
      "step": 1762
    },
    {
      "epoch": 0.07306560570268142,
      "grad_norm": 0.4338874816894531,
      "learning_rate": 4.634879191014962e-06,
      "loss": 0.7261,
      "step": 1763
    },
    {
      "epoch": 0.0731070496083551,
      "grad_norm": 0.47353309392929077,
      "learning_rate": 4.634671971486593e-06,
      "loss": 0.8005,
      "step": 1764
    },
    {
      "epoch": 0.07314849351402876,
      "grad_norm": 0.4663343131542206,
      "learning_rate": 4.634464751958225e-06,
      "loss": 0.7659,
      "step": 1765
    },
    {
      "epoch": 0.07318993741970244,
      "grad_norm": 0.4709503948688507,
      "learning_rate": 4.634257532429857e-06,
      "loss": 0.8142,
      "step": 1766
    },
    {
      "epoch": 0.0732313813253761,
      "grad_norm": 0.458511084318161,
      "learning_rate": 4.634050312901488e-06,
      "loss": 0.7346,
      "step": 1767
    },
    {
      "epoch": 0.07327282523104978,
      "grad_norm": 0.44481754302978516,
      "learning_rate": 4.63384309337312e-06,
      "loss": 0.7502,
      "step": 1768
    },
    {
      "epoch": 0.07331426913672344,
      "grad_norm": 0.44879600405693054,
      "learning_rate": 4.633635873844752e-06,
      "loss": 0.7341,
      "step": 1769
    },
    {
      "epoch": 0.07335571304239712,
      "grad_norm": 0.4418463110923767,
      "learning_rate": 4.633428654316383e-06,
      "loss": 0.7878,
      "step": 1770
    },
    {
      "epoch": 0.07339715694807078,
      "grad_norm": 0.487883985042572,
      "learning_rate": 4.633221434788015e-06,
      "loss": 0.7795,
      "step": 1771
    },
    {
      "epoch": 0.07343860085374446,
      "grad_norm": 0.45064201951026917,
      "learning_rate": 4.633014215259646e-06,
      "loss": 0.7622,
      "step": 1772
    },
    {
      "epoch": 0.07348004475941813,
      "grad_norm": 0.415192574262619,
      "learning_rate": 4.632806995731278e-06,
      "loss": 0.7065,
      "step": 1773
    },
    {
      "epoch": 0.0735214886650918,
      "grad_norm": 0.4516944885253906,
      "learning_rate": 4.63259977620291e-06,
      "loss": 0.7307,
      "step": 1774
    },
    {
      "epoch": 0.07356293257076547,
      "grad_norm": 0.4642982482910156,
      "learning_rate": 4.632392556674542e-06,
      "loss": 0.791,
      "step": 1775
    },
    {
      "epoch": 0.07360437647643914,
      "grad_norm": 0.43663284182548523,
      "learning_rate": 4.6321853371461724e-06,
      "loss": 0.7566,
      "step": 1776
    },
    {
      "epoch": 0.07364582038211281,
      "grad_norm": 0.4271349608898163,
      "learning_rate": 4.631978117617805e-06,
      "loss": 0.7798,
      "step": 1777
    },
    {
      "epoch": 0.07368726428778648,
      "grad_norm": 0.43104761838912964,
      "learning_rate": 4.631770898089437e-06,
      "loss": 0.7344,
      "step": 1778
    },
    {
      "epoch": 0.07372870819346015,
      "grad_norm": 0.4554607570171356,
      "learning_rate": 4.631563678561068e-06,
      "loss": 0.7886,
      "step": 1779
    },
    {
      "epoch": 0.07377015209913382,
      "grad_norm": 0.45183441042900085,
      "learning_rate": 4.6313564590327e-06,
      "loss": 0.7346,
      "step": 1780
    },
    {
      "epoch": 0.0738115960048075,
      "grad_norm": 0.41788461804389954,
      "learning_rate": 4.631149239504331e-06,
      "loss": 0.7795,
      "step": 1781
    },
    {
      "epoch": 0.07385303991048116,
      "grad_norm": 0.4727994203567505,
      "learning_rate": 4.630942019975963e-06,
      "loss": 0.7151,
      "step": 1782
    },
    {
      "epoch": 0.07389448381615484,
      "grad_norm": 0.45103439688682556,
      "learning_rate": 4.630734800447595e-06,
      "loss": 0.7476,
      "step": 1783
    },
    {
      "epoch": 0.0739359277218285,
      "grad_norm": 0.4557344913482666,
      "learning_rate": 4.630527580919226e-06,
      "loss": 0.7764,
      "step": 1784
    },
    {
      "epoch": 0.07397737162750218,
      "grad_norm": 0.44310054183006287,
      "learning_rate": 4.6303203613908574e-06,
      "loss": 0.7083,
      "step": 1785
    },
    {
      "epoch": 0.07401881553317585,
      "grad_norm": 0.4647631347179413,
      "learning_rate": 4.63011314186249e-06,
      "loss": 0.7305,
      "step": 1786
    },
    {
      "epoch": 0.07406025943884952,
      "grad_norm": 0.44456273317337036,
      "learning_rate": 4.629905922334121e-06,
      "loss": 0.8049,
      "step": 1787
    },
    {
      "epoch": 0.07410170334452319,
      "grad_norm": 0.41680532693862915,
      "learning_rate": 4.6296987028057524e-06,
      "loss": 0.7422,
      "step": 1788
    },
    {
      "epoch": 0.07414314725019686,
      "grad_norm": 0.4570443332195282,
      "learning_rate": 4.629491483277385e-06,
      "loss": 0.7717,
      "step": 1789
    },
    {
      "epoch": 0.07418459115587053,
      "grad_norm": 0.4430422782897949,
      "learning_rate": 4.629284263749016e-06,
      "loss": 0.7542,
      "step": 1790
    },
    {
      "epoch": 0.0742260350615442,
      "grad_norm": 0.4563167691230774,
      "learning_rate": 4.6290770442206474e-06,
      "loss": 0.7268,
      "step": 1791
    },
    {
      "epoch": 0.07426747896721787,
      "grad_norm": 0.47300484776496887,
      "learning_rate": 4.628869824692279e-06,
      "loss": 0.7957,
      "step": 1792
    },
    {
      "epoch": 0.07430892287289154,
      "grad_norm": 0.4505418539047241,
      "learning_rate": 4.628662605163911e-06,
      "loss": 0.7656,
      "step": 1793
    },
    {
      "epoch": 0.07435036677856521,
      "grad_norm": 0.42699867486953735,
      "learning_rate": 4.6284553856355424e-06,
      "loss": 0.7562,
      "step": 1794
    },
    {
      "epoch": 0.07439181068423888,
      "grad_norm": 0.44827938079833984,
      "learning_rate": 4.628248166107175e-06,
      "loss": 0.72,
      "step": 1795
    },
    {
      "epoch": 0.07443325458991255,
      "grad_norm": 0.5012211799621582,
      "learning_rate": 4.628040946578806e-06,
      "loss": 0.7952,
      "step": 1796
    },
    {
      "epoch": 0.07447469849558623,
      "grad_norm": 0.45306214690208435,
      "learning_rate": 4.6278337270504375e-06,
      "loss": 0.7749,
      "step": 1797
    },
    {
      "epoch": 0.0745161424012599,
      "grad_norm": 0.45056119561195374,
      "learning_rate": 4.62762650752207e-06,
      "loss": 0.7694,
      "step": 1798
    },
    {
      "epoch": 0.07455758630693357,
      "grad_norm": 0.4454525411128998,
      "learning_rate": 4.627419287993701e-06,
      "loss": 0.7412,
      "step": 1799
    },
    {
      "epoch": 0.07459903021260723,
      "grad_norm": 0.45956140756607056,
      "learning_rate": 4.6272120684653325e-06,
      "loss": 0.7942,
      "step": 1800
    },
    {
      "epoch": 0.07464047411828091,
      "grad_norm": 0.4564327597618103,
      "learning_rate": 4.627004848936964e-06,
      "loss": 0.8018,
      "step": 1801
    },
    {
      "epoch": 0.07468191802395457,
      "grad_norm": 0.4510807693004608,
      "learning_rate": 4.626797629408596e-06,
      "loss": 0.8018,
      "step": 1802
    },
    {
      "epoch": 0.07472336192962825,
      "grad_norm": 0.5180459022521973,
      "learning_rate": 4.6265904098802275e-06,
      "loss": 0.8035,
      "step": 1803
    },
    {
      "epoch": 0.07476480583530191,
      "grad_norm": 0.44152164459228516,
      "learning_rate": 4.626383190351859e-06,
      "loss": 0.7864,
      "step": 1804
    },
    {
      "epoch": 0.07480624974097559,
      "grad_norm": 0.47599801421165466,
      "learning_rate": 4.626175970823491e-06,
      "loss": 0.8071,
      "step": 1805
    },
    {
      "epoch": 0.07484769364664925,
      "grad_norm": 0.4898378252983093,
      "learning_rate": 4.6259687512951225e-06,
      "loss": 0.7656,
      "step": 1806
    },
    {
      "epoch": 0.07488913755232293,
      "grad_norm": 0.4746813178062439,
      "learning_rate": 4.625761531766754e-06,
      "loss": 0.7423,
      "step": 1807
    },
    {
      "epoch": 0.0749305814579966,
      "grad_norm": 0.4731309115886688,
      "learning_rate": 4.625554312238385e-06,
      "loss": 0.7949,
      "step": 1808
    },
    {
      "epoch": 0.07497202536367027,
      "grad_norm": 0.43048739433288574,
      "learning_rate": 4.6253470927100175e-06,
      "loss": 0.7268,
      "step": 1809
    },
    {
      "epoch": 0.07501346926934395,
      "grad_norm": 0.4917040169239044,
      "learning_rate": 4.625139873181649e-06,
      "loss": 0.7804,
      "step": 1810
    },
    {
      "epoch": 0.07505491317501761,
      "grad_norm": 0.4528306722640991,
      "learning_rate": 4.624932653653281e-06,
      "loss": 0.7878,
      "step": 1811
    },
    {
      "epoch": 0.07509635708069129,
      "grad_norm": 0.4328532814979553,
      "learning_rate": 4.624725434124912e-06,
      "loss": 0.7419,
      "step": 1812
    },
    {
      "epoch": 0.07513780098636495,
      "grad_norm": 0.44404372572898865,
      "learning_rate": 4.624518214596544e-06,
      "loss": 0.7148,
      "step": 1813
    },
    {
      "epoch": 0.07517924489203863,
      "grad_norm": 0.43389609456062317,
      "learning_rate": 4.624310995068176e-06,
      "loss": 0.802,
      "step": 1814
    },
    {
      "epoch": 0.07522068879771229,
      "grad_norm": 0.42758652567863464,
      "learning_rate": 4.6241037755398075e-06,
      "loss": 0.7297,
      "step": 1815
    },
    {
      "epoch": 0.07526213270338597,
      "grad_norm": 0.42928996682167053,
      "learning_rate": 4.623896556011439e-06,
      "loss": 0.7554,
      "step": 1816
    },
    {
      "epoch": 0.07530357660905963,
      "grad_norm": 0.4639790654182434,
      "learning_rate": 4.62368933648307e-06,
      "loss": 0.7373,
      "step": 1817
    },
    {
      "epoch": 0.07534502051473331,
      "grad_norm": 0.4239185154438019,
      "learning_rate": 4.6234821169547025e-06,
      "loss": 0.7074,
      "step": 1818
    },
    {
      "epoch": 0.07538646442040697,
      "grad_norm": 0.4425605237483978,
      "learning_rate": 4.623274897426334e-06,
      "loss": 0.7534,
      "step": 1819
    },
    {
      "epoch": 0.07542790832608065,
      "grad_norm": 0.45999810099601746,
      "learning_rate": 4.623067677897965e-06,
      "loss": 0.8037,
      "step": 1820
    },
    {
      "epoch": 0.07546935223175431,
      "grad_norm": 0.4081891179084778,
      "learning_rate": 4.6228604583695975e-06,
      "loss": 0.7026,
      "step": 1821
    },
    {
      "epoch": 0.07551079613742799,
      "grad_norm": 0.510579526424408,
      "learning_rate": 4.622653238841229e-06,
      "loss": 0.8203,
      "step": 1822
    },
    {
      "epoch": 0.07555224004310167,
      "grad_norm": 0.49238812923431396,
      "learning_rate": 4.62244601931286e-06,
      "loss": 0.7388,
      "step": 1823
    },
    {
      "epoch": 0.07559368394877533,
      "grad_norm": 0.46058303117752075,
      "learning_rate": 4.622238799784492e-06,
      "loss": 0.7537,
      "step": 1824
    },
    {
      "epoch": 0.07563512785444901,
      "grad_norm": 0.4655688405036926,
      "learning_rate": 4.622031580256124e-06,
      "loss": 0.7637,
      "step": 1825
    },
    {
      "epoch": 0.07567657176012267,
      "grad_norm": 0.44930219650268555,
      "learning_rate": 4.621824360727755e-06,
      "loss": 0.8131,
      "step": 1826
    },
    {
      "epoch": 0.07571801566579635,
      "grad_norm": 0.4302092492580414,
      "learning_rate": 4.6216171411993875e-06,
      "loss": 0.7528,
      "step": 1827
    },
    {
      "epoch": 0.07575945957147001,
      "grad_norm": 0.423718124628067,
      "learning_rate": 4.621409921671018e-06,
      "loss": 0.7457,
      "step": 1828
    },
    {
      "epoch": 0.07580090347714369,
      "grad_norm": 0.46975329518318176,
      "learning_rate": 4.62120270214265e-06,
      "loss": 0.7742,
      "step": 1829
    },
    {
      "epoch": 0.07584234738281735,
      "grad_norm": 0.45333707332611084,
      "learning_rate": 4.6209954826142825e-06,
      "loss": 0.7683,
      "step": 1830
    },
    {
      "epoch": 0.07588379128849103,
      "grad_norm": 0.48876506090164185,
      "learning_rate": 4.620788263085914e-06,
      "loss": 0.8501,
      "step": 1831
    },
    {
      "epoch": 0.07592523519416469,
      "grad_norm": 0.46611636877059937,
      "learning_rate": 4.620581043557545e-06,
      "loss": 0.8022,
      "step": 1832
    },
    {
      "epoch": 0.07596667909983837,
      "grad_norm": 0.4619766175746918,
      "learning_rate": 4.620373824029177e-06,
      "loss": 0.7773,
      "step": 1833
    },
    {
      "epoch": 0.07600812300551205,
      "grad_norm": 0.46724531054496765,
      "learning_rate": 4.620166604500809e-06,
      "loss": 0.745,
      "step": 1834
    },
    {
      "epoch": 0.07604956691118571,
      "grad_norm": 0.46512022614479065,
      "learning_rate": 4.61995938497244e-06,
      "loss": 0.71,
      "step": 1835
    },
    {
      "epoch": 0.07609101081685939,
      "grad_norm": 0.4498942494392395,
      "learning_rate": 4.619752165444072e-06,
      "loss": 0.7638,
      "step": 1836
    },
    {
      "epoch": 0.07613245472253305,
      "grad_norm": 0.44740912318229675,
      "learning_rate": 4.619544945915703e-06,
      "loss": 0.7937,
      "step": 1837
    },
    {
      "epoch": 0.07617389862820673,
      "grad_norm": 0.4779653251171112,
      "learning_rate": 4.619337726387335e-06,
      "loss": 0.8081,
      "step": 1838
    },
    {
      "epoch": 0.07621534253388039,
      "grad_norm": 0.4262476861476898,
      "learning_rate": 4.619130506858967e-06,
      "loss": 0.7131,
      "step": 1839
    },
    {
      "epoch": 0.07625678643955407,
      "grad_norm": 0.4475859999656677,
      "learning_rate": 4.618923287330598e-06,
      "loss": 0.7463,
      "step": 1840
    },
    {
      "epoch": 0.07629823034522773,
      "grad_norm": 0.4014715850353241,
      "learning_rate": 4.61871606780223e-06,
      "loss": 0.6975,
      "step": 1841
    },
    {
      "epoch": 0.07633967425090141,
      "grad_norm": 0.4433557987213135,
      "learning_rate": 4.618508848273862e-06,
      "loss": 0.7993,
      "step": 1842
    },
    {
      "epoch": 0.07638111815657507,
      "grad_norm": 0.46803778409957886,
      "learning_rate": 4.618301628745494e-06,
      "loss": 0.7153,
      "step": 1843
    },
    {
      "epoch": 0.07642256206224875,
      "grad_norm": 0.4221131503582001,
      "learning_rate": 4.6180944092171244e-06,
      "loss": 0.709,
      "step": 1844
    },
    {
      "epoch": 0.07646400596792241,
      "grad_norm": 0.4289880394935608,
      "learning_rate": 4.617887189688757e-06,
      "loss": 0.8154,
      "step": 1845
    },
    {
      "epoch": 0.07650544987359609,
      "grad_norm": 0.47242870926856995,
      "learning_rate": 4.617679970160388e-06,
      "loss": 0.8127,
      "step": 1846
    },
    {
      "epoch": 0.07654689377926976,
      "grad_norm": 0.4612913727760315,
      "learning_rate": 4.61747275063202e-06,
      "loss": 0.791,
      "step": 1847
    },
    {
      "epoch": 0.07658833768494343,
      "grad_norm": 0.4600946009159088,
      "learning_rate": 4.617265531103652e-06,
      "loss": 0.7964,
      "step": 1848
    },
    {
      "epoch": 0.0766297815906171,
      "grad_norm": 0.5173287987709045,
      "learning_rate": 4.617058311575283e-06,
      "loss": 0.8289,
      "step": 1849
    },
    {
      "epoch": 0.07667122549629077,
      "grad_norm": 0.4896432161331177,
      "learning_rate": 4.616851092046915e-06,
      "loss": 0.8408,
      "step": 1850
    },
    {
      "epoch": 0.07671266940196445,
      "grad_norm": 0.44647738337516785,
      "learning_rate": 4.616643872518547e-06,
      "loss": 0.7458,
      "step": 1851
    },
    {
      "epoch": 0.07675411330763811,
      "grad_norm": 0.4450191259384155,
      "learning_rate": 4.616436652990178e-06,
      "loss": 0.7839,
      "step": 1852
    },
    {
      "epoch": 0.07679555721331179,
      "grad_norm": 0.4258873462677002,
      "learning_rate": 4.6162294334618094e-06,
      "loss": 0.7537,
      "step": 1853
    },
    {
      "epoch": 0.07683700111898545,
      "grad_norm": 0.4854743480682373,
      "learning_rate": 4.616022213933442e-06,
      "loss": 0.7612,
      "step": 1854
    },
    {
      "epoch": 0.07687844502465913,
      "grad_norm": 0.45001545548439026,
      "learning_rate": 4.615814994405073e-06,
      "loss": 0.7522,
      "step": 1855
    },
    {
      "epoch": 0.07691988893033279,
      "grad_norm": 0.48317083716392517,
      "learning_rate": 4.6156077748767045e-06,
      "loss": 0.7351,
      "step": 1856
    },
    {
      "epoch": 0.07696133283600647,
      "grad_norm": 0.44324567914009094,
      "learning_rate": 4.615400555348337e-06,
      "loss": 0.7673,
      "step": 1857
    },
    {
      "epoch": 0.07700277674168014,
      "grad_norm": 0.47571250796318054,
      "learning_rate": 4.615193335819968e-06,
      "loss": 0.772,
      "step": 1858
    },
    {
      "epoch": 0.0770442206473538,
      "grad_norm": 0.44750216603279114,
      "learning_rate": 4.6149861162915995e-06,
      "loss": 0.7661,
      "step": 1859
    },
    {
      "epoch": 0.07708566455302748,
      "grad_norm": 0.4616469442844391,
      "learning_rate": 4.614778896763231e-06,
      "loss": 0.7515,
      "step": 1860
    },
    {
      "epoch": 0.07712710845870115,
      "grad_norm": 0.44754815101623535,
      "learning_rate": 4.614571677234863e-06,
      "loss": 0.769,
      "step": 1861
    },
    {
      "epoch": 0.07716855236437482,
      "grad_norm": 0.4579787850379944,
      "learning_rate": 4.6143644577064945e-06,
      "loss": 0.7974,
      "step": 1862
    },
    {
      "epoch": 0.07720999627004849,
      "grad_norm": 0.4634620249271393,
      "learning_rate": 4.614157238178127e-06,
      "loss": 0.7599,
      "step": 1863
    },
    {
      "epoch": 0.07725144017572216,
      "grad_norm": 0.5181484222412109,
      "learning_rate": 4.613950018649758e-06,
      "loss": 0.8057,
      "step": 1864
    },
    {
      "epoch": 0.07729288408139583,
      "grad_norm": 0.4826219975948334,
      "learning_rate": 4.6137427991213895e-06,
      "loss": 0.7866,
      "step": 1865
    },
    {
      "epoch": 0.0773343279870695,
      "grad_norm": 0.4336509704589844,
      "learning_rate": 4.613535579593022e-06,
      "loss": 0.6987,
      "step": 1866
    },
    {
      "epoch": 0.07737577189274317,
      "grad_norm": 0.5099712014198303,
      "learning_rate": 4.613328360064653e-06,
      "loss": 0.7466,
      "step": 1867
    },
    {
      "epoch": 0.07741721579841684,
      "grad_norm": 0.44108128547668457,
      "learning_rate": 4.6131211405362845e-06,
      "loss": 0.7507,
      "step": 1868
    },
    {
      "epoch": 0.07745865970409051,
      "grad_norm": 0.46545690298080444,
      "learning_rate": 4.612913921007916e-06,
      "loss": 0.7551,
      "step": 1869
    },
    {
      "epoch": 0.07750010360976418,
      "grad_norm": 0.4285983443260193,
      "learning_rate": 4.612706701479548e-06,
      "loss": 0.7216,
      "step": 1870
    },
    {
      "epoch": 0.07754154751543786,
      "grad_norm": 0.45119625329971313,
      "learning_rate": 4.6124994819511795e-06,
      "loss": 0.7583,
      "step": 1871
    },
    {
      "epoch": 0.07758299142111152,
      "grad_norm": 0.4831332862377167,
      "learning_rate": 4.612292262422811e-06,
      "loss": 0.7666,
      "step": 1872
    },
    {
      "epoch": 0.0776244353267852,
      "grad_norm": 0.43292492628097534,
      "learning_rate": 4.612085042894442e-06,
      "loss": 0.7573,
      "step": 1873
    },
    {
      "epoch": 0.07766587923245886,
      "grad_norm": 0.4782642722129822,
      "learning_rate": 4.6118778233660745e-06,
      "loss": 0.8037,
      "step": 1874
    },
    {
      "epoch": 0.07770732313813254,
      "grad_norm": 0.5037537217140198,
      "learning_rate": 4.611670603837706e-06,
      "loss": 0.783,
      "step": 1875
    },
    {
      "epoch": 0.0777487670438062,
      "grad_norm": 0.43975114822387695,
      "learning_rate": 4.611463384309337e-06,
      "loss": 0.7878,
      "step": 1876
    },
    {
      "epoch": 0.07779021094947988,
      "grad_norm": 0.467578649520874,
      "learning_rate": 4.6112561647809695e-06,
      "loss": 0.79,
      "step": 1877
    },
    {
      "epoch": 0.07783165485515355,
      "grad_norm": 0.44706177711486816,
      "learning_rate": 4.611048945252601e-06,
      "loss": 0.8015,
      "step": 1878
    },
    {
      "epoch": 0.07787309876082722,
      "grad_norm": 0.45218080282211304,
      "learning_rate": 4.610841725724233e-06,
      "loss": 0.7659,
      "step": 1879
    },
    {
      "epoch": 0.07791454266650089,
      "grad_norm": 0.43183889985084534,
      "learning_rate": 4.610634506195864e-06,
      "loss": 0.761,
      "step": 1880
    },
    {
      "epoch": 0.07795598657217456,
      "grad_norm": 0.4780930280685425,
      "learning_rate": 4.610427286667496e-06,
      "loss": 0.7717,
      "step": 1881
    },
    {
      "epoch": 0.07799743047784823,
      "grad_norm": 0.47216352820396423,
      "learning_rate": 4.610220067139128e-06,
      "loss": 0.7263,
      "step": 1882
    },
    {
      "epoch": 0.0780388743835219,
      "grad_norm": 0.4856502413749695,
      "learning_rate": 4.6100128476107595e-06,
      "loss": 0.7703,
      "step": 1883
    },
    {
      "epoch": 0.07808031828919558,
      "grad_norm": 0.4615204334259033,
      "learning_rate": 4.609805628082391e-06,
      "loss": 0.7324,
      "step": 1884
    },
    {
      "epoch": 0.07812176219486924,
      "grad_norm": 0.4522905945777893,
      "learning_rate": 4.609598408554022e-06,
      "loss": 0.8352,
      "step": 1885
    },
    {
      "epoch": 0.07816320610054292,
      "grad_norm": 0.4266384243965149,
      "learning_rate": 4.6093911890256545e-06,
      "loss": 0.7415,
      "step": 1886
    },
    {
      "epoch": 0.07820465000621658,
      "grad_norm": 0.4685969054698944,
      "learning_rate": 4.609183969497286e-06,
      "loss": 0.7869,
      "step": 1887
    },
    {
      "epoch": 0.07824609391189026,
      "grad_norm": 0.47244465351104736,
      "learning_rate": 4.608976749968917e-06,
      "loss": 0.7898,
      "step": 1888
    },
    {
      "epoch": 0.07828753781756392,
      "grad_norm": 0.4805828332901001,
      "learning_rate": 4.608769530440549e-06,
      "loss": 0.7979,
      "step": 1889
    },
    {
      "epoch": 0.0783289817232376,
      "grad_norm": 0.45322510600090027,
      "learning_rate": 4.608562310912181e-06,
      "loss": 0.7227,
      "step": 1890
    },
    {
      "epoch": 0.07837042562891126,
      "grad_norm": 0.4402100443840027,
      "learning_rate": 4.608355091383812e-06,
      "loss": 0.7358,
      "step": 1891
    },
    {
      "epoch": 0.07841186953458494,
      "grad_norm": 0.463314414024353,
      "learning_rate": 4.608147871855444e-06,
      "loss": 0.844,
      "step": 1892
    },
    {
      "epoch": 0.0784533134402586,
      "grad_norm": 0.4943394958972931,
      "learning_rate": 4.607940652327076e-06,
      "loss": 0.8186,
      "step": 1893
    },
    {
      "epoch": 0.07849475734593228,
      "grad_norm": 0.4464211165904999,
      "learning_rate": 4.607733432798707e-06,
      "loss": 0.7534,
      "step": 1894
    },
    {
      "epoch": 0.07853620125160596,
      "grad_norm": 0.46243879199028015,
      "learning_rate": 4.6075262132703395e-06,
      "loss": 0.7307,
      "step": 1895
    },
    {
      "epoch": 0.07857764515727962,
      "grad_norm": 0.45432737469673157,
      "learning_rate": 4.60731899374197e-06,
      "loss": 0.7786,
      "step": 1896
    },
    {
      "epoch": 0.0786190890629533,
      "grad_norm": 0.45538121461868286,
      "learning_rate": 4.607111774213602e-06,
      "loss": 0.791,
      "step": 1897
    },
    {
      "epoch": 0.07866053296862696,
      "grad_norm": 0.4796019196510315,
      "learning_rate": 4.606904554685234e-06,
      "loss": 0.7864,
      "step": 1898
    },
    {
      "epoch": 0.07870197687430064,
      "grad_norm": 0.4167230427265167,
      "learning_rate": 4.606697335156866e-06,
      "loss": 0.7327,
      "step": 1899
    },
    {
      "epoch": 0.0787434207799743,
      "grad_norm": 0.4214766323566437,
      "learning_rate": 4.606490115628497e-06,
      "loss": 0.761,
      "step": 1900
    },
    {
      "epoch": 0.07878486468564798,
      "grad_norm": 0.4327312707901001,
      "learning_rate": 4.606282896100129e-06,
      "loss": 0.6953,
      "step": 1901
    },
    {
      "epoch": 0.07882630859132164,
      "grad_norm": 0.44663581252098083,
      "learning_rate": 4.606075676571761e-06,
      "loss": 0.7932,
      "step": 1902
    },
    {
      "epoch": 0.07886775249699532,
      "grad_norm": 0.4363744258880615,
      "learning_rate": 4.605868457043392e-06,
      "loss": 0.7649,
      "step": 1903
    },
    {
      "epoch": 0.07890919640266898,
      "grad_norm": 0.4099828004837036,
      "learning_rate": 4.605661237515024e-06,
      "loss": 0.8247,
      "step": 1904
    },
    {
      "epoch": 0.07895064030834266,
      "grad_norm": 0.4296228885650635,
      "learning_rate": 4.605454017986655e-06,
      "loss": 0.7888,
      "step": 1905
    },
    {
      "epoch": 0.07899208421401632,
      "grad_norm": 0.434773713350296,
      "learning_rate": 4.605246798458287e-06,
      "loss": 0.7224,
      "step": 1906
    },
    {
      "epoch": 0.07903352811969,
      "grad_norm": 0.4375283122062683,
      "learning_rate": 4.605039578929919e-06,
      "loss": 0.7751,
      "step": 1907
    },
    {
      "epoch": 0.07907497202536368,
      "grad_norm": 0.4627740681171417,
      "learning_rate": 4.60483235940155e-06,
      "loss": 0.761,
      "step": 1908
    },
    {
      "epoch": 0.07911641593103734,
      "grad_norm": 0.42502644658088684,
      "learning_rate": 4.604625139873182e-06,
      "loss": 0.7346,
      "step": 1909
    },
    {
      "epoch": 0.07915785983671102,
      "grad_norm": 0.4524705111980438,
      "learning_rate": 4.604417920344814e-06,
      "loss": 0.7283,
      "step": 1910
    },
    {
      "epoch": 0.07919930374238468,
      "grad_norm": 0.4681034982204437,
      "learning_rate": 4.604210700816446e-06,
      "loss": 0.8347,
      "step": 1911
    },
    {
      "epoch": 0.07924074764805836,
      "grad_norm": 0.46250027418136597,
      "learning_rate": 4.6040034812880764e-06,
      "loss": 0.7784,
      "step": 1912
    },
    {
      "epoch": 0.07928219155373202,
      "grad_norm": 0.4581931531429291,
      "learning_rate": 4.603796261759709e-06,
      "loss": 0.8081,
      "step": 1913
    },
    {
      "epoch": 0.0793236354594057,
      "grad_norm": 0.4367046654224396,
      "learning_rate": 4.60358904223134e-06,
      "loss": 0.7017,
      "step": 1914
    },
    {
      "epoch": 0.07936507936507936,
      "grad_norm": 0.46020442247390747,
      "learning_rate": 4.603381822702972e-06,
      "loss": 0.7974,
      "step": 1915
    },
    {
      "epoch": 0.07940652327075304,
      "grad_norm": 0.41448089480400085,
      "learning_rate": 4.603174603174604e-06,
      "loss": 0.7317,
      "step": 1916
    },
    {
      "epoch": 0.0794479671764267,
      "grad_norm": 0.4349333345890045,
      "learning_rate": 4.602967383646235e-06,
      "loss": 0.6948,
      "step": 1917
    },
    {
      "epoch": 0.07948941108210038,
      "grad_norm": 0.4636598527431488,
      "learning_rate": 4.602760164117867e-06,
      "loss": 0.7952,
      "step": 1918
    },
    {
      "epoch": 0.07953085498777404,
      "grad_norm": 0.48504671454429626,
      "learning_rate": 4.602552944589499e-06,
      "loss": 0.825,
      "step": 1919
    },
    {
      "epoch": 0.07957229889344772,
      "grad_norm": 0.49243369698524475,
      "learning_rate": 4.60234572506113e-06,
      "loss": 0.803,
      "step": 1920
    },
    {
      "epoch": 0.0796137427991214,
      "grad_norm": 0.4513554871082306,
      "learning_rate": 4.6021385055327615e-06,
      "loss": 0.7253,
      "step": 1921
    },
    {
      "epoch": 0.07965518670479506,
      "grad_norm": 0.4651351571083069,
      "learning_rate": 4.601931286004394e-06,
      "loss": 0.7645,
      "step": 1922
    },
    {
      "epoch": 0.07969663061046874,
      "grad_norm": 0.4624966084957123,
      "learning_rate": 4.601724066476025e-06,
      "loss": 0.7452,
      "step": 1923
    },
    {
      "epoch": 0.0797380745161424,
      "grad_norm": 0.4130507707595825,
      "learning_rate": 4.6015168469476565e-06,
      "loss": 0.7126,
      "step": 1924
    },
    {
      "epoch": 0.07977951842181608,
      "grad_norm": 0.4527469277381897,
      "learning_rate": 4.601309627419288e-06,
      "loss": 0.7476,
      "step": 1925
    },
    {
      "epoch": 0.07982096232748974,
      "grad_norm": 0.4475564956665039,
      "learning_rate": 4.60110240789092e-06,
      "loss": 0.7803,
      "step": 1926
    },
    {
      "epoch": 0.07986240623316342,
      "grad_norm": 0.44627103209495544,
      "learning_rate": 4.6008951883625515e-06,
      "loss": 0.689,
      "step": 1927
    },
    {
      "epoch": 0.07990385013883708,
      "grad_norm": 0.4628615081310272,
      "learning_rate": 4.600687968834183e-06,
      "loss": 0.8145,
      "step": 1928
    },
    {
      "epoch": 0.07994529404451076,
      "grad_norm": 0.4401334822177887,
      "learning_rate": 4.600480749305815e-06,
      "loss": 0.7498,
      "step": 1929
    },
    {
      "epoch": 0.07998673795018442,
      "grad_norm": 0.4138008654117584,
      "learning_rate": 4.6002735297774465e-06,
      "loss": 0.6868,
      "step": 1930
    },
    {
      "epoch": 0.0800281818558581,
      "grad_norm": 0.4302521347999573,
      "learning_rate": 4.600066310249079e-06,
      "loss": 0.7698,
      "step": 1931
    },
    {
      "epoch": 0.08006962576153177,
      "grad_norm": 0.4618249535560608,
      "learning_rate": 4.59985909072071e-06,
      "loss": 0.8013,
      "step": 1932
    },
    {
      "epoch": 0.08011106966720544,
      "grad_norm": 0.441426157951355,
      "learning_rate": 4.5996518711923415e-06,
      "loss": 0.739,
      "step": 1933
    },
    {
      "epoch": 0.08015251357287911,
      "grad_norm": 0.4190951883792877,
      "learning_rate": 4.599444651663973e-06,
      "loss": 0.6686,
      "step": 1934
    },
    {
      "epoch": 0.08019395747855278,
      "grad_norm": 0.4135172367095947,
      "learning_rate": 4.599237432135605e-06,
      "loss": 0.7686,
      "step": 1935
    },
    {
      "epoch": 0.08023540138422645,
      "grad_norm": 0.44404852390289307,
      "learning_rate": 4.5990302126072365e-06,
      "loss": 0.6989,
      "step": 1936
    },
    {
      "epoch": 0.08027684528990012,
      "grad_norm": 0.4356656074523926,
      "learning_rate": 4.598822993078868e-06,
      "loss": 0.8096,
      "step": 1937
    },
    {
      "epoch": 0.0803182891955738,
      "grad_norm": 0.4899286925792694,
      "learning_rate": 4.5986157735505e-06,
      "loss": 0.8252,
      "step": 1938
    },
    {
      "epoch": 0.08035973310124746,
      "grad_norm": 0.445220023393631,
      "learning_rate": 4.5984085540221315e-06,
      "loss": 0.7866,
      "step": 1939
    },
    {
      "epoch": 0.08040117700692113,
      "grad_norm": 0.4810304641723633,
      "learning_rate": 4.598201334493763e-06,
      "loss": 0.772,
      "step": 1940
    },
    {
      "epoch": 0.0804426209125948,
      "grad_norm": 0.4239565134048462,
      "learning_rate": 4.597994114965394e-06,
      "loss": 0.7096,
      "step": 1941
    },
    {
      "epoch": 0.08048406481826847,
      "grad_norm": 0.456948846578598,
      "learning_rate": 4.5977868954370265e-06,
      "loss": 0.7961,
      "step": 1942
    },
    {
      "epoch": 0.08052550872394214,
      "grad_norm": 0.4533609449863434,
      "learning_rate": 4.597579675908658e-06,
      "loss": 0.7356,
      "step": 1943
    },
    {
      "epoch": 0.08056695262961582,
      "grad_norm": 0.4711451232433319,
      "learning_rate": 4.597372456380289e-06,
      "loss": 0.7578,
      "step": 1944
    },
    {
      "epoch": 0.08060839653528949,
      "grad_norm": 0.47080543637275696,
      "learning_rate": 4.5971652368519215e-06,
      "loss": 0.8152,
      "step": 1945
    },
    {
      "epoch": 0.08064984044096316,
      "grad_norm": 0.4283369183540344,
      "learning_rate": 4.596958017323553e-06,
      "loss": 0.7065,
      "step": 1946
    },
    {
      "epoch": 0.08069128434663683,
      "grad_norm": 0.4214217960834503,
      "learning_rate": 4.596750797795185e-06,
      "loss": 0.7466,
      "step": 1947
    },
    {
      "epoch": 0.0807327282523105,
      "grad_norm": 0.4439764618873596,
      "learning_rate": 4.5965435782668165e-06,
      "loss": 0.8103,
      "step": 1948
    },
    {
      "epoch": 0.08077417215798417,
      "grad_norm": 0.44330283999443054,
      "learning_rate": 4.596336358738448e-06,
      "loss": 0.8318,
      "step": 1949
    },
    {
      "epoch": 0.08081561606365784,
      "grad_norm": 0.41693663597106934,
      "learning_rate": 4.596129139210079e-06,
      "loss": 0.7219,
      "step": 1950
    },
    {
      "epoch": 0.08085705996933151,
      "grad_norm": 0.4680829346179962,
      "learning_rate": 4.5959219196817115e-06,
      "loss": 0.7461,
      "step": 1951
    },
    {
      "epoch": 0.08089850387500518,
      "grad_norm": 0.43037012219429016,
      "learning_rate": 4.595714700153343e-06,
      "loss": 0.7157,
      "step": 1952
    },
    {
      "epoch": 0.08093994778067885,
      "grad_norm": 0.49243882298469543,
      "learning_rate": 4.595507480624974e-06,
      "loss": 0.7725,
      "step": 1953
    },
    {
      "epoch": 0.08098139168635252,
      "grad_norm": 0.44856080412864685,
      "learning_rate": 4.5953002610966065e-06,
      "loss": 0.7278,
      "step": 1954
    },
    {
      "epoch": 0.0810228355920262,
      "grad_norm": 0.4417436420917511,
      "learning_rate": 4.595093041568238e-06,
      "loss": 0.7793,
      "step": 1955
    },
    {
      "epoch": 0.08106427949769986,
      "grad_norm": 0.40614381432533264,
      "learning_rate": 4.594885822039869e-06,
      "loss": 0.7368,
      "step": 1956
    },
    {
      "epoch": 0.08110572340337353,
      "grad_norm": 0.4587932527065277,
      "learning_rate": 4.594678602511501e-06,
      "loss": 0.7842,
      "step": 1957
    },
    {
      "epoch": 0.08114716730904721,
      "grad_norm": 0.4725556969642639,
      "learning_rate": 4.594471382983133e-06,
      "loss": 0.7366,
      "step": 1958
    },
    {
      "epoch": 0.08118861121472087,
      "grad_norm": 0.44160541892051697,
      "learning_rate": 4.594264163454764e-06,
      "loss": 0.7676,
      "step": 1959
    },
    {
      "epoch": 0.08123005512039455,
      "grad_norm": 0.43638449907302856,
      "learning_rate": 4.594056943926396e-06,
      "loss": 0.7231,
      "step": 1960
    },
    {
      "epoch": 0.08127149902606821,
      "grad_norm": 0.43149739503860474,
      "learning_rate": 4.593849724398028e-06,
      "loss": 0.6981,
      "step": 1961
    },
    {
      "epoch": 0.08131294293174189,
      "grad_norm": 0.47470149397850037,
      "learning_rate": 4.593642504869659e-06,
      "loss": 0.7316,
      "step": 1962
    },
    {
      "epoch": 0.08135438683741555,
      "grad_norm": 0.4233434796333313,
      "learning_rate": 4.5934352853412915e-06,
      "loss": 0.7559,
      "step": 1963
    },
    {
      "epoch": 0.08139583074308923,
      "grad_norm": 0.46559303998947144,
      "learning_rate": 4.593228065812922e-06,
      "loss": 0.8,
      "step": 1964
    },
    {
      "epoch": 0.0814372746487629,
      "grad_norm": 0.45025813579559326,
      "learning_rate": 4.593020846284554e-06,
      "loss": 0.7407,
      "step": 1965
    },
    {
      "epoch": 0.08147871855443657,
      "grad_norm": 0.4029502868652344,
      "learning_rate": 4.592813626756186e-06,
      "loss": 0.696,
      "step": 1966
    },
    {
      "epoch": 0.08152016246011023,
      "grad_norm": 0.4447357654571533,
      "learning_rate": 4.592606407227818e-06,
      "loss": 0.7544,
      "step": 1967
    },
    {
      "epoch": 0.08156160636578391,
      "grad_norm": 0.4972272217273712,
      "learning_rate": 4.592399187699449e-06,
      "loss": 0.7617,
      "step": 1968
    },
    {
      "epoch": 0.08160305027145759,
      "grad_norm": 0.4270951449871063,
      "learning_rate": 4.592191968171081e-06,
      "loss": 0.8064,
      "step": 1969
    },
    {
      "epoch": 0.08164449417713125,
      "grad_norm": 0.4376661777496338,
      "learning_rate": 4.591984748642713e-06,
      "loss": 0.6991,
      "step": 1970
    },
    {
      "epoch": 0.08168593808280493,
      "grad_norm": 0.4429546892642975,
      "learning_rate": 4.591777529114344e-06,
      "loss": 0.7368,
      "step": 1971
    },
    {
      "epoch": 0.08172738198847859,
      "grad_norm": 0.4448795020580292,
      "learning_rate": 4.591570309585976e-06,
      "loss": 0.7522,
      "step": 1972
    },
    {
      "epoch": 0.08176882589415227,
      "grad_norm": 0.43445849418640137,
      "learning_rate": 4.591363090057607e-06,
      "loss": 0.7544,
      "step": 1973
    },
    {
      "epoch": 0.08181026979982593,
      "grad_norm": 0.46531492471694946,
      "learning_rate": 4.591155870529239e-06,
      "loss": 0.7952,
      "step": 1974
    },
    {
      "epoch": 0.08185171370549961,
      "grad_norm": 0.44272223114967346,
      "learning_rate": 4.590948651000871e-06,
      "loss": 0.7378,
      "step": 1975
    },
    {
      "epoch": 0.08189315761117327,
      "grad_norm": 0.4426511824131012,
      "learning_rate": 4.590741431472502e-06,
      "loss": 0.7012,
      "step": 1976
    },
    {
      "epoch": 0.08193460151684695,
      "grad_norm": 0.4502531588077545,
      "learning_rate": 4.5905342119441335e-06,
      "loss": 0.7981,
      "step": 1977
    },
    {
      "epoch": 0.08197604542252061,
      "grad_norm": 0.4311940371990204,
      "learning_rate": 4.590326992415766e-06,
      "loss": 0.7292,
      "step": 1978
    },
    {
      "epoch": 0.08201748932819429,
      "grad_norm": 0.44279080629348755,
      "learning_rate": 4.590119772887398e-06,
      "loss": 0.7612,
      "step": 1979
    },
    {
      "epoch": 0.08205893323386795,
      "grad_norm": 0.4665888845920563,
      "learning_rate": 4.5899125533590285e-06,
      "loss": 0.7866,
      "step": 1980
    },
    {
      "epoch": 0.08210037713954163,
      "grad_norm": 0.4285626709461212,
      "learning_rate": 4.589705333830661e-06,
      "loss": 0.8071,
      "step": 1981
    },
    {
      "epoch": 0.08214182104521531,
      "grad_norm": 0.38651043176651,
      "learning_rate": 4.589498114302292e-06,
      "loss": 0.7646,
      "step": 1982
    },
    {
      "epoch": 0.08218326495088897,
      "grad_norm": 0.4539370834827423,
      "learning_rate": 4.589290894773924e-06,
      "loss": 0.7781,
      "step": 1983
    },
    {
      "epoch": 0.08222470885656265,
      "grad_norm": 0.45575448870658875,
      "learning_rate": 4.589083675245556e-06,
      "loss": 0.7468,
      "step": 1984
    },
    {
      "epoch": 0.08226615276223631,
      "grad_norm": 0.4629756808280945,
      "learning_rate": 4.588876455717187e-06,
      "loss": 0.7571,
      "step": 1985
    },
    {
      "epoch": 0.08230759666790999,
      "grad_norm": 0.46269774436950684,
      "learning_rate": 4.5886692361888185e-06,
      "loss": 0.7686,
      "step": 1986
    },
    {
      "epoch": 0.08234904057358365,
      "grad_norm": 0.4966188073158264,
      "learning_rate": 4.588462016660451e-06,
      "loss": 0.7507,
      "step": 1987
    },
    {
      "epoch": 0.08239048447925733,
      "grad_norm": 0.41765865683555603,
      "learning_rate": 4.588254797132082e-06,
      "loss": 0.7181,
      "step": 1988
    },
    {
      "epoch": 0.08243192838493099,
      "grad_norm": 0.4706251919269562,
      "learning_rate": 4.5880475776037135e-06,
      "loss": 0.7397,
      "step": 1989
    },
    {
      "epoch": 0.08247337229060467,
      "grad_norm": 0.4798257350921631,
      "learning_rate": 4.587840358075346e-06,
      "loss": 0.8026,
      "step": 1990
    },
    {
      "epoch": 0.08251481619627833,
      "grad_norm": 0.42262500524520874,
      "learning_rate": 4.587633138546977e-06,
      "loss": 0.7637,
      "step": 1991
    },
    {
      "epoch": 0.08255626010195201,
      "grad_norm": 0.4746358394622803,
      "learning_rate": 4.5874259190186085e-06,
      "loss": 0.7744,
      "step": 1992
    },
    {
      "epoch": 0.08259770400762567,
      "grad_norm": 0.4563295841217041,
      "learning_rate": 4.58721869949024e-06,
      "loss": 0.8027,
      "step": 1993
    },
    {
      "epoch": 0.08263914791329935,
      "grad_norm": 0.4500739276409149,
      "learning_rate": 4.587011479961872e-06,
      "loss": 0.7471,
      "step": 1994
    },
    {
      "epoch": 0.08268059181897303,
      "grad_norm": 0.47931191325187683,
      "learning_rate": 4.5868042604335035e-06,
      "loss": 0.8054,
      "step": 1995
    },
    {
      "epoch": 0.08272203572464669,
      "grad_norm": 0.44504377245903015,
      "learning_rate": 4.586597040905135e-06,
      "loss": 0.7583,
      "step": 1996
    },
    {
      "epoch": 0.08276347963032037,
      "grad_norm": 0.4895908832550049,
      "learning_rate": 4.586389821376767e-06,
      "loss": 0.7554,
      "step": 1997
    },
    {
      "epoch": 0.08280492353599403,
      "grad_norm": 0.43344560265541077,
      "learning_rate": 4.5861826018483985e-06,
      "loss": 0.7681,
      "step": 1998
    },
    {
      "epoch": 0.0828463674416677,
      "grad_norm": 0.43635421991348267,
      "learning_rate": 4.585975382320031e-06,
      "loss": 0.7266,
      "step": 1999
    },
    {
      "epoch": 0.08288781134734137,
      "grad_norm": 0.43765681982040405,
      "learning_rate": 4.585768162791662e-06,
      "loss": 0.7883,
      "step": 2000
    },
    {
      "epoch": 0.08292925525301505,
      "grad_norm": 0.47952449321746826,
      "learning_rate": 4.5855609432632935e-06,
      "loss": 0.7192,
      "step": 2001
    },
    {
      "epoch": 0.08297069915868871,
      "grad_norm": 0.47122427821159363,
      "learning_rate": 4.585353723734925e-06,
      "loss": 0.8179,
      "step": 2002
    },
    {
      "epoch": 0.08301214306436239,
      "grad_norm": 0.472246915102005,
      "learning_rate": 4.585146504206557e-06,
      "loss": 0.79,
      "step": 2003
    },
    {
      "epoch": 0.08305358697003605,
      "grad_norm": 0.4613513946533203,
      "learning_rate": 4.5849392846781885e-06,
      "loss": 0.7603,
      "step": 2004
    },
    {
      "epoch": 0.08309503087570973,
      "grad_norm": 0.43824225664138794,
      "learning_rate": 4.58473206514982e-06,
      "loss": 0.7854,
      "step": 2005
    },
    {
      "epoch": 0.0831364747813834,
      "grad_norm": 0.4677543640136719,
      "learning_rate": 4.584524845621452e-06,
      "loss": 0.7914,
      "step": 2006
    },
    {
      "epoch": 0.08317791868705707,
      "grad_norm": 0.45726025104522705,
      "learning_rate": 4.5843176260930835e-06,
      "loss": 0.7512,
      "step": 2007
    },
    {
      "epoch": 0.08321936259273074,
      "grad_norm": 0.46152862906455994,
      "learning_rate": 4.584110406564715e-06,
      "loss": 0.78,
      "step": 2008
    },
    {
      "epoch": 0.08326080649840441,
      "grad_norm": 0.42357802391052246,
      "learning_rate": 4.583903187036346e-06,
      "loss": 0.7854,
      "step": 2009
    },
    {
      "epoch": 0.08330225040407808,
      "grad_norm": 0.3974681794643402,
      "learning_rate": 4.5836959675079785e-06,
      "loss": 0.7004,
      "step": 2010
    },
    {
      "epoch": 0.08334369430975175,
      "grad_norm": 0.4321729838848114,
      "learning_rate": 4.58348874797961e-06,
      "loss": 0.8076,
      "step": 2011
    },
    {
      "epoch": 0.08338513821542543,
      "grad_norm": 0.4288319945335388,
      "learning_rate": 4.583281528451241e-06,
      "loss": 0.7783,
      "step": 2012
    },
    {
      "epoch": 0.08342658212109909,
      "grad_norm": 0.4062556326389313,
      "learning_rate": 4.5830743089228735e-06,
      "loss": 0.6647,
      "step": 2013
    },
    {
      "epoch": 0.08346802602677277,
      "grad_norm": 0.4751129448413849,
      "learning_rate": 4.582867089394505e-06,
      "loss": 0.7788,
      "step": 2014
    },
    {
      "epoch": 0.08350946993244643,
      "grad_norm": 0.47318270802497864,
      "learning_rate": 4.582659869866137e-06,
      "loss": 0.8162,
      "step": 2015
    },
    {
      "epoch": 0.0835509138381201,
      "grad_norm": 0.42475032806396484,
      "learning_rate": 4.5824526503377685e-06,
      "loss": 0.7732,
      "step": 2016
    },
    {
      "epoch": 0.08359235774379377,
      "grad_norm": 0.4487524926662445,
      "learning_rate": 4.5822454308094e-06,
      "loss": 0.7549,
      "step": 2017
    },
    {
      "epoch": 0.08363380164946745,
      "grad_norm": 0.44328489899635315,
      "learning_rate": 4.582038211281031e-06,
      "loss": 0.7993,
      "step": 2018
    },
    {
      "epoch": 0.08367524555514112,
      "grad_norm": 0.46115025877952576,
      "learning_rate": 4.5818309917526635e-06,
      "loss": 0.7607,
      "step": 2019
    },
    {
      "epoch": 0.08371668946081479,
      "grad_norm": 0.40650030970573425,
      "learning_rate": 4.581623772224295e-06,
      "loss": 0.6987,
      "step": 2020
    },
    {
      "epoch": 0.08375813336648846,
      "grad_norm": 0.41901496052742004,
      "learning_rate": 4.581416552695926e-06,
      "loss": 0.7,
      "step": 2021
    },
    {
      "epoch": 0.08379957727216213,
      "grad_norm": 0.446773886680603,
      "learning_rate": 4.5812093331675585e-06,
      "loss": 0.7585,
      "step": 2022
    },
    {
      "epoch": 0.0838410211778358,
      "grad_norm": 0.46564799547195435,
      "learning_rate": 4.58100211363919e-06,
      "loss": 0.7622,
      "step": 2023
    },
    {
      "epoch": 0.08388246508350947,
      "grad_norm": 0.49480322003364563,
      "learning_rate": 4.580794894110821e-06,
      "loss": 0.7839,
      "step": 2024
    },
    {
      "epoch": 0.08392390898918314,
      "grad_norm": 0.47816893458366394,
      "learning_rate": 4.580587674582453e-06,
      "loss": 0.7383,
      "step": 2025
    },
    {
      "epoch": 0.0839653528948568,
      "grad_norm": 0.4518311321735382,
      "learning_rate": 4.580380455054085e-06,
      "loss": 0.7544,
      "step": 2026
    },
    {
      "epoch": 0.08400679680053048,
      "grad_norm": 0.4600609242916107,
      "learning_rate": 4.580173235525716e-06,
      "loss": 0.8022,
      "step": 2027
    },
    {
      "epoch": 0.08404824070620415,
      "grad_norm": 0.43373215198516846,
      "learning_rate": 4.579966015997348e-06,
      "loss": 0.7205,
      "step": 2028
    },
    {
      "epoch": 0.08408968461187782,
      "grad_norm": 0.453763872385025,
      "learning_rate": 4.579758796468979e-06,
      "loss": 0.7444,
      "step": 2029
    },
    {
      "epoch": 0.08413112851755149,
      "grad_norm": 0.5405118465423584,
      "learning_rate": 4.579551576940611e-06,
      "loss": 0.7988,
      "step": 2030
    },
    {
      "epoch": 0.08417257242322516,
      "grad_norm": 0.45820483565330505,
      "learning_rate": 4.5793443574122435e-06,
      "loss": 0.8445,
      "step": 2031
    },
    {
      "epoch": 0.08421401632889884,
      "grad_norm": 0.4445345997810364,
      "learning_rate": 4.579137137883874e-06,
      "loss": 0.7205,
      "step": 2032
    },
    {
      "epoch": 0.0842554602345725,
      "grad_norm": 0.4242554306983948,
      "learning_rate": 4.578929918355506e-06,
      "loss": 0.743,
      "step": 2033
    },
    {
      "epoch": 0.08429690414024618,
      "grad_norm": 0.4658140540122986,
      "learning_rate": 4.578722698827138e-06,
      "loss": 0.769,
      "step": 2034
    },
    {
      "epoch": 0.08433834804591984,
      "grad_norm": 0.4437236785888672,
      "learning_rate": 4.57851547929877e-06,
      "loss": 0.7383,
      "step": 2035
    },
    {
      "epoch": 0.08437979195159352,
      "grad_norm": 0.44685378670692444,
      "learning_rate": 4.578308259770401e-06,
      "loss": 0.7769,
      "step": 2036
    },
    {
      "epoch": 0.08442123585726719,
      "grad_norm": 0.4076836109161377,
      "learning_rate": 4.578101040242033e-06,
      "loss": 0.6995,
      "step": 2037
    },
    {
      "epoch": 0.08446267976294086,
      "grad_norm": 0.44023066759109497,
      "learning_rate": 4.577893820713664e-06,
      "loss": 0.7913,
      "step": 2038
    },
    {
      "epoch": 0.08450412366861453,
      "grad_norm": 0.4099765121936798,
      "learning_rate": 4.577686601185296e-06,
      "loss": 0.7917,
      "step": 2039
    },
    {
      "epoch": 0.0845455675742882,
      "grad_norm": 0.42275750637054443,
      "learning_rate": 4.577479381656928e-06,
      "loss": 0.7505,
      "step": 2040
    },
    {
      "epoch": 0.08458701147996187,
      "grad_norm": 0.46205073595046997,
      "learning_rate": 4.577272162128559e-06,
      "loss": 0.7925,
      "step": 2041
    },
    {
      "epoch": 0.08462845538563554,
      "grad_norm": 0.4566892385482788,
      "learning_rate": 4.577064942600191e-06,
      "loss": 0.7844,
      "step": 2042
    },
    {
      "epoch": 0.08466989929130922,
      "grad_norm": 0.420128732919693,
      "learning_rate": 4.576857723071823e-06,
      "loss": 0.7554,
      "step": 2043
    },
    {
      "epoch": 0.08471134319698288,
      "grad_norm": 0.46599262952804565,
      "learning_rate": 4.576650503543454e-06,
      "loss": 0.77,
      "step": 2044
    },
    {
      "epoch": 0.08475278710265656,
      "grad_norm": 0.44940847158432007,
      "learning_rate": 4.5764432840150855e-06,
      "loss": 0.7651,
      "step": 2045
    },
    {
      "epoch": 0.08479423100833022,
      "grad_norm": 0.46521294116973877,
      "learning_rate": 4.576236064486718e-06,
      "loss": 0.7649,
      "step": 2046
    },
    {
      "epoch": 0.0848356749140039,
      "grad_norm": 0.47755691409111023,
      "learning_rate": 4.576028844958349e-06,
      "loss": 0.7881,
      "step": 2047
    },
    {
      "epoch": 0.08487711881967756,
      "grad_norm": 0.4413328766822815,
      "learning_rate": 4.5758216254299805e-06,
      "loss": 0.7373,
      "step": 2048
    },
    {
      "epoch": 0.08491856272535124,
      "grad_norm": 0.48594722151756287,
      "learning_rate": 4.575614405901613e-06,
      "loss": 0.7664,
      "step": 2049
    },
    {
      "epoch": 0.0849600066310249,
      "grad_norm": 0.4363350570201874,
      "learning_rate": 4.575407186373244e-06,
      "loss": 0.7007,
      "step": 2050
    },
    {
      "epoch": 0.08500145053669858,
      "grad_norm": 0.42702746391296387,
      "learning_rate": 4.575199966844876e-06,
      "loss": 0.7295,
      "step": 2051
    },
    {
      "epoch": 0.08504289444237224,
      "grad_norm": 0.45240429043769836,
      "learning_rate": 4.574992747316508e-06,
      "loss": 0.7747,
      "step": 2052
    },
    {
      "epoch": 0.08508433834804592,
      "grad_norm": 0.430084764957428,
      "learning_rate": 4.574785527788139e-06,
      "loss": 0.7402,
      "step": 2053
    },
    {
      "epoch": 0.08512578225371958,
      "grad_norm": 0.43184056878089905,
      "learning_rate": 4.5745783082597705e-06,
      "loss": 0.7717,
      "step": 2054
    },
    {
      "epoch": 0.08516722615939326,
      "grad_norm": 0.4326060116291046,
      "learning_rate": 4.574371088731403e-06,
      "loss": 0.7966,
      "step": 2055
    },
    {
      "epoch": 0.08520867006506694,
      "grad_norm": 0.41815802454948425,
      "learning_rate": 4.574163869203034e-06,
      "loss": 0.7358,
      "step": 2056
    },
    {
      "epoch": 0.0852501139707406,
      "grad_norm": 0.436719685792923,
      "learning_rate": 4.5739566496746655e-06,
      "loss": 0.7148,
      "step": 2057
    },
    {
      "epoch": 0.08529155787641428,
      "grad_norm": 0.4108443558216095,
      "learning_rate": 4.573749430146298e-06,
      "loss": 0.728,
      "step": 2058
    },
    {
      "epoch": 0.08533300178208794,
      "grad_norm": 0.43048134446144104,
      "learning_rate": 4.573542210617929e-06,
      "loss": 0.7246,
      "step": 2059
    },
    {
      "epoch": 0.08537444568776162,
      "grad_norm": 0.43298161029815674,
      "learning_rate": 4.5733349910895605e-06,
      "loss": 0.7622,
      "step": 2060
    },
    {
      "epoch": 0.08541588959343528,
      "grad_norm": 0.43782326579093933,
      "learning_rate": 4.573127771561192e-06,
      "loss": 0.7415,
      "step": 2061
    },
    {
      "epoch": 0.08545733349910896,
      "grad_norm": 0.4883464574813843,
      "learning_rate": 4.572920552032824e-06,
      "loss": 0.7957,
      "step": 2062
    },
    {
      "epoch": 0.08549877740478262,
      "grad_norm": 0.4369872808456421,
      "learning_rate": 4.5727133325044555e-06,
      "loss": 0.8206,
      "step": 2063
    },
    {
      "epoch": 0.0855402213104563,
      "grad_norm": 0.4636523127555847,
      "learning_rate": 4.572506112976087e-06,
      "loss": 0.8057,
      "step": 2064
    },
    {
      "epoch": 0.08558166521612996,
      "grad_norm": 0.4862779378890991,
      "learning_rate": 4.572298893447719e-06,
      "loss": 0.7648,
      "step": 2065
    },
    {
      "epoch": 0.08562310912180364,
      "grad_norm": 0.44743070006370544,
      "learning_rate": 4.5720916739193505e-06,
      "loss": 0.7405,
      "step": 2066
    },
    {
      "epoch": 0.0856645530274773,
      "grad_norm": 0.5206807851791382,
      "learning_rate": 4.571884454390983e-06,
      "loss": 0.8411,
      "step": 2067
    },
    {
      "epoch": 0.08570599693315098,
      "grad_norm": 0.46740373969078064,
      "learning_rate": 4.571677234862614e-06,
      "loss": 0.7673,
      "step": 2068
    },
    {
      "epoch": 0.08574744083882466,
      "grad_norm": 0.44362175464630127,
      "learning_rate": 4.5714700153342455e-06,
      "loss": 0.8147,
      "step": 2069
    },
    {
      "epoch": 0.08578888474449832,
      "grad_norm": 0.4234970510005951,
      "learning_rate": 4.571262795805877e-06,
      "loss": 0.7571,
      "step": 2070
    },
    {
      "epoch": 0.085830328650172,
      "grad_norm": 0.4279879629611969,
      "learning_rate": 4.571055576277509e-06,
      "loss": 0.7905,
      "step": 2071
    },
    {
      "epoch": 0.08587177255584566,
      "grad_norm": 0.42760854959487915,
      "learning_rate": 4.5708483567491405e-06,
      "loss": 0.7793,
      "step": 2072
    },
    {
      "epoch": 0.08591321646151934,
      "grad_norm": 0.47730764746665955,
      "learning_rate": 4.570641137220772e-06,
      "loss": 0.7266,
      "step": 2073
    },
    {
      "epoch": 0.085954660367193,
      "grad_norm": 0.4440353810787201,
      "learning_rate": 4.570433917692404e-06,
      "loss": 0.7228,
      "step": 2074
    },
    {
      "epoch": 0.08599610427286668,
      "grad_norm": 0.442749559879303,
      "learning_rate": 4.5702266981640355e-06,
      "loss": 0.7861,
      "step": 2075
    },
    {
      "epoch": 0.08603754817854034,
      "grad_norm": 0.47952306270599365,
      "learning_rate": 4.570019478635667e-06,
      "loss": 0.782,
      "step": 2076
    },
    {
      "epoch": 0.08607899208421402,
      "grad_norm": 0.4330749213695526,
      "learning_rate": 4.569812259107298e-06,
      "loss": 0.775,
      "step": 2077
    },
    {
      "epoch": 0.08612043598988768,
      "grad_norm": 0.4757136404514313,
      "learning_rate": 4.5696050395789305e-06,
      "loss": 0.8667,
      "step": 2078
    },
    {
      "epoch": 0.08616187989556136,
      "grad_norm": 0.4289725422859192,
      "learning_rate": 4.569397820050562e-06,
      "loss": 0.7312,
      "step": 2079
    },
    {
      "epoch": 0.08620332380123504,
      "grad_norm": 0.44398823380470276,
      "learning_rate": 4.569190600522193e-06,
      "loss": 0.7118,
      "step": 2080
    },
    {
      "epoch": 0.0862447677069087,
      "grad_norm": 0.4574585556983948,
      "learning_rate": 4.568983380993825e-06,
      "loss": 0.8228,
      "step": 2081
    },
    {
      "epoch": 0.08628621161258238,
      "grad_norm": 0.43872395157814026,
      "learning_rate": 4.568776161465457e-06,
      "loss": 0.7446,
      "step": 2082
    },
    {
      "epoch": 0.08632765551825604,
      "grad_norm": 0.48115092515945435,
      "learning_rate": 4.568568941937089e-06,
      "loss": 0.8088,
      "step": 2083
    },
    {
      "epoch": 0.08636909942392972,
      "grad_norm": 0.43788594007492065,
      "learning_rate": 4.5683617224087205e-06,
      "loss": 0.7666,
      "step": 2084
    },
    {
      "epoch": 0.08641054332960338,
      "grad_norm": 0.548052966594696,
      "learning_rate": 4.568154502880352e-06,
      "loss": 0.78,
      "step": 2085
    },
    {
      "epoch": 0.08645198723527706,
      "grad_norm": 0.45637160539627075,
      "learning_rate": 4.567947283351983e-06,
      "loss": 0.7542,
      "step": 2086
    },
    {
      "epoch": 0.08649343114095072,
      "grad_norm": 0.4704807698726654,
      "learning_rate": 4.5677400638236155e-06,
      "loss": 0.7786,
      "step": 2087
    },
    {
      "epoch": 0.0865348750466244,
      "grad_norm": 0.43890872597694397,
      "learning_rate": 4.567532844295247e-06,
      "loss": 0.7681,
      "step": 2088
    },
    {
      "epoch": 0.08657631895229806,
      "grad_norm": 0.47336286306381226,
      "learning_rate": 4.567325624766878e-06,
      "loss": 0.738,
      "step": 2089
    },
    {
      "epoch": 0.08661776285797174,
      "grad_norm": 0.5095197558403015,
      "learning_rate": 4.56711840523851e-06,
      "loss": 0.7847,
      "step": 2090
    },
    {
      "epoch": 0.0866592067636454,
      "grad_norm": 0.4290327727794647,
      "learning_rate": 4.566911185710142e-06,
      "loss": 0.7085,
      "step": 2091
    },
    {
      "epoch": 0.08670065066931908,
      "grad_norm": 0.44808709621429443,
      "learning_rate": 4.566703966181773e-06,
      "loss": 0.7822,
      "step": 2092
    },
    {
      "epoch": 0.08674209457499275,
      "grad_norm": 0.4921129047870636,
      "learning_rate": 4.566496746653405e-06,
      "loss": 0.7795,
      "step": 2093
    },
    {
      "epoch": 0.08678353848066642,
      "grad_norm": 0.3895568251609802,
      "learning_rate": 4.566289527125037e-06,
      "loss": 0.7161,
      "step": 2094
    },
    {
      "epoch": 0.0868249823863401,
      "grad_norm": 0.43971338868141174,
      "learning_rate": 4.566082307596668e-06,
      "loss": 0.7393,
      "step": 2095
    },
    {
      "epoch": 0.08686642629201376,
      "grad_norm": 0.44493353366851807,
      "learning_rate": 4.5658750880683e-06,
      "loss": 0.7795,
      "step": 2096
    },
    {
      "epoch": 0.08690787019768743,
      "grad_norm": 0.4715208411216736,
      "learning_rate": 4.565667868539931e-06,
      "loss": 0.7412,
      "step": 2097
    },
    {
      "epoch": 0.0869493141033611,
      "grad_norm": 0.4596356451511383,
      "learning_rate": 4.565460649011563e-06,
      "loss": 0.7715,
      "step": 2098
    },
    {
      "epoch": 0.08699075800903477,
      "grad_norm": 0.4211691617965698,
      "learning_rate": 4.565253429483195e-06,
      "loss": 0.7393,
      "step": 2099
    },
    {
      "epoch": 0.08703220191470844,
      "grad_norm": 0.46153560280799866,
      "learning_rate": 4.565046209954826e-06,
      "loss": 0.7485,
      "step": 2100
    },
    {
      "epoch": 0.08707364582038211,
      "grad_norm": 0.4615577161312103,
      "learning_rate": 4.564838990426458e-06,
      "loss": 0.793,
      "step": 2101
    },
    {
      "epoch": 0.08711508972605578,
      "grad_norm": 0.4837665557861328,
      "learning_rate": 4.56463177089809e-06,
      "loss": 0.7795,
      "step": 2102
    },
    {
      "epoch": 0.08715653363172945,
      "grad_norm": 0.4686914384365082,
      "learning_rate": 4.564424551369722e-06,
      "loss": 0.8149,
      "step": 2103
    },
    {
      "epoch": 0.08719797753740312,
      "grad_norm": 0.45950695872306824,
      "learning_rate": 4.564217331841353e-06,
      "loss": 0.7915,
      "step": 2104
    },
    {
      "epoch": 0.0872394214430768,
      "grad_norm": 0.4595943093299866,
      "learning_rate": 4.564010112312985e-06,
      "loss": 0.7483,
      "step": 2105
    },
    {
      "epoch": 0.08728086534875047,
      "grad_norm": 0.4625067710876465,
      "learning_rate": 4.563802892784616e-06,
      "loss": 0.8135,
      "step": 2106
    },
    {
      "epoch": 0.08732230925442414,
      "grad_norm": 0.42756184935569763,
      "learning_rate": 4.563595673256248e-06,
      "loss": 0.7593,
      "step": 2107
    },
    {
      "epoch": 0.08736375316009781,
      "grad_norm": 0.48455724120140076,
      "learning_rate": 4.56338845372788e-06,
      "loss": 0.8306,
      "step": 2108
    },
    {
      "epoch": 0.08740519706577148,
      "grad_norm": 0.41179582476615906,
      "learning_rate": 4.563181234199511e-06,
      "loss": 0.7173,
      "step": 2109
    },
    {
      "epoch": 0.08744664097144515,
      "grad_norm": 0.41084831953048706,
      "learning_rate": 4.562974014671143e-06,
      "loss": 0.6708,
      "step": 2110
    },
    {
      "epoch": 0.08748808487711882,
      "grad_norm": 0.4693790674209595,
      "learning_rate": 4.562766795142775e-06,
      "loss": 0.7507,
      "step": 2111
    },
    {
      "epoch": 0.08752952878279249,
      "grad_norm": 0.4720715582370758,
      "learning_rate": 4.562559575614406e-06,
      "loss": 0.8196,
      "step": 2112
    },
    {
      "epoch": 0.08757097268846616,
      "grad_norm": 0.4883628785610199,
      "learning_rate": 4.5623523560860375e-06,
      "loss": 0.7793,
      "step": 2113
    },
    {
      "epoch": 0.08761241659413983,
      "grad_norm": 0.4234831631183624,
      "learning_rate": 4.56214513655767e-06,
      "loss": 0.7026,
      "step": 2114
    },
    {
      "epoch": 0.0876538604998135,
      "grad_norm": 0.46147656440734863,
      "learning_rate": 4.561937917029301e-06,
      "loss": 0.7417,
      "step": 2115
    },
    {
      "epoch": 0.08769530440548717,
      "grad_norm": 0.4476986229419708,
      "learning_rate": 4.5617306975009325e-06,
      "loss": 0.7954,
      "step": 2116
    },
    {
      "epoch": 0.08773674831116085,
      "grad_norm": 0.42565757036209106,
      "learning_rate": 4.561523477972565e-06,
      "loss": 0.7241,
      "step": 2117
    },
    {
      "epoch": 0.08777819221683451,
      "grad_norm": 0.41643714904785156,
      "learning_rate": 4.561316258444196e-06,
      "loss": 0.761,
      "step": 2118
    },
    {
      "epoch": 0.08781963612250819,
      "grad_norm": 0.4193111062049866,
      "learning_rate": 4.561109038915828e-06,
      "loss": 0.7162,
      "step": 2119
    },
    {
      "epoch": 0.08786108002818185,
      "grad_norm": 0.48522114753723145,
      "learning_rate": 4.56090181938746e-06,
      "loss": 0.7461,
      "step": 2120
    },
    {
      "epoch": 0.08790252393385553,
      "grad_norm": 0.4176945686340332,
      "learning_rate": 4.560694599859091e-06,
      "loss": 0.7898,
      "step": 2121
    },
    {
      "epoch": 0.0879439678395292,
      "grad_norm": 0.4507688283920288,
      "learning_rate": 4.5604873803307225e-06,
      "loss": 0.8208,
      "step": 2122
    },
    {
      "epoch": 0.08798541174520287,
      "grad_norm": 0.4388856589794159,
      "learning_rate": 4.560280160802355e-06,
      "loss": 0.7109,
      "step": 2123
    },
    {
      "epoch": 0.08802685565087653,
      "grad_norm": 0.4523058235645294,
      "learning_rate": 4.560072941273986e-06,
      "loss": 0.7888,
      "step": 2124
    },
    {
      "epoch": 0.08806829955655021,
      "grad_norm": 0.4143514037132263,
      "learning_rate": 4.5598657217456175e-06,
      "loss": 0.7422,
      "step": 2125
    },
    {
      "epoch": 0.08810974346222387,
      "grad_norm": 0.43963322043418884,
      "learning_rate": 4.55965850221725e-06,
      "loss": 0.7083,
      "step": 2126
    },
    {
      "epoch": 0.08815118736789755,
      "grad_norm": 0.43000903725624084,
      "learning_rate": 4.559451282688881e-06,
      "loss": 0.7068,
      "step": 2127
    },
    {
      "epoch": 0.08819263127357121,
      "grad_norm": 0.4487351179122925,
      "learning_rate": 4.5592440631605125e-06,
      "loss": 0.7871,
      "step": 2128
    },
    {
      "epoch": 0.08823407517924489,
      "grad_norm": 0.4382845163345337,
      "learning_rate": 4.559036843632144e-06,
      "loss": 0.7869,
      "step": 2129
    },
    {
      "epoch": 0.08827551908491857,
      "grad_norm": 0.4721006155014038,
      "learning_rate": 4.558829624103776e-06,
      "loss": 0.6833,
      "step": 2130
    },
    {
      "epoch": 0.08831696299059223,
      "grad_norm": 0.4833557903766632,
      "learning_rate": 4.5586224045754075e-06,
      "loss": 0.7559,
      "step": 2131
    },
    {
      "epoch": 0.08835840689626591,
      "grad_norm": 0.4151139259338379,
      "learning_rate": 4.558415185047039e-06,
      "loss": 0.7314,
      "step": 2132
    },
    {
      "epoch": 0.08839985080193957,
      "grad_norm": 0.4292290210723877,
      "learning_rate": 4.55820796551867e-06,
      "loss": 0.7209,
      "step": 2133
    },
    {
      "epoch": 0.08844129470761325,
      "grad_norm": 0.392171710729599,
      "learning_rate": 4.5580007459903025e-06,
      "loss": 0.7556,
      "step": 2134
    },
    {
      "epoch": 0.08848273861328691,
      "grad_norm": 0.40045082569122314,
      "learning_rate": 4.557793526461935e-06,
      "loss": 0.7462,
      "step": 2135
    },
    {
      "epoch": 0.08852418251896059,
      "grad_norm": 0.46485257148742676,
      "learning_rate": 4.557586306933566e-06,
      "loss": 0.7937,
      "step": 2136
    },
    {
      "epoch": 0.08856562642463425,
      "grad_norm": 0.42961594462394714,
      "learning_rate": 4.5573790874051975e-06,
      "loss": 0.7849,
      "step": 2137
    },
    {
      "epoch": 0.08860707033030793,
      "grad_norm": 0.46037158370018005,
      "learning_rate": 4.557171867876829e-06,
      "loss": 0.7925,
      "step": 2138
    },
    {
      "epoch": 0.0886485142359816,
      "grad_norm": 0.4338686764240265,
      "learning_rate": 4.556964648348461e-06,
      "loss": 0.7146,
      "step": 2139
    },
    {
      "epoch": 0.08868995814165527,
      "grad_norm": 0.4479596018791199,
      "learning_rate": 4.5567574288200925e-06,
      "loss": 0.7203,
      "step": 2140
    },
    {
      "epoch": 0.08873140204732895,
      "grad_norm": 0.421063095331192,
      "learning_rate": 4.556550209291724e-06,
      "loss": 0.7537,
      "step": 2141
    },
    {
      "epoch": 0.08877284595300261,
      "grad_norm": 0.3934994339942932,
      "learning_rate": 4.556342989763355e-06,
      "loss": 0.7366,
      "step": 2142
    },
    {
      "epoch": 0.08881428985867629,
      "grad_norm": 0.4762936532497406,
      "learning_rate": 4.5561357702349875e-06,
      "loss": 0.7483,
      "step": 2143
    },
    {
      "epoch": 0.08885573376434995,
      "grad_norm": 0.4404817223548889,
      "learning_rate": 4.555928550706619e-06,
      "loss": 0.7859,
      "step": 2144
    },
    {
      "epoch": 0.08889717767002363,
      "grad_norm": 0.444450318813324,
      "learning_rate": 4.55572133117825e-06,
      "loss": 0.7773,
      "step": 2145
    },
    {
      "epoch": 0.08893862157569729,
      "grad_norm": 0.41657501459121704,
      "learning_rate": 4.5555141116498825e-06,
      "loss": 0.7781,
      "step": 2146
    },
    {
      "epoch": 0.08898006548137097,
      "grad_norm": 0.462635338306427,
      "learning_rate": 4.555306892121514e-06,
      "loss": 0.7537,
      "step": 2147
    },
    {
      "epoch": 0.08902150938704463,
      "grad_norm": 0.44731810688972473,
      "learning_rate": 4.555099672593145e-06,
      "loss": 0.7394,
      "step": 2148
    },
    {
      "epoch": 0.08906295329271831,
      "grad_norm": 0.4948531687259674,
      "learning_rate": 4.554892453064777e-06,
      "loss": 0.7473,
      "step": 2149
    },
    {
      "epoch": 0.08910439719839197,
      "grad_norm": 0.45920631289482117,
      "learning_rate": 4.554685233536409e-06,
      "loss": 0.7312,
      "step": 2150
    },
    {
      "epoch": 0.08914584110406565,
      "grad_norm": 0.4803909361362457,
      "learning_rate": 4.55447801400804e-06,
      "loss": 0.771,
      "step": 2151
    },
    {
      "epoch": 0.08918728500973931,
      "grad_norm": 0.4469764828681946,
      "learning_rate": 4.5542707944796725e-06,
      "loss": 0.8157,
      "step": 2152
    },
    {
      "epoch": 0.08922872891541299,
      "grad_norm": 0.44928717613220215,
      "learning_rate": 4.554063574951304e-06,
      "loss": 0.7478,
      "step": 2153
    },
    {
      "epoch": 0.08927017282108667,
      "grad_norm": 0.46665889024734497,
      "learning_rate": 4.553856355422935e-06,
      "loss": 0.7363,
      "step": 2154
    },
    {
      "epoch": 0.08931161672676033,
      "grad_norm": 0.4329316318035126,
      "learning_rate": 4.5536491358945675e-06,
      "loss": 0.7427,
      "step": 2155
    },
    {
      "epoch": 0.089353060632434,
      "grad_norm": 0.4584934711456299,
      "learning_rate": 4.553441916366199e-06,
      "loss": 0.7533,
      "step": 2156
    },
    {
      "epoch": 0.08939450453810767,
      "grad_norm": 0.41812360286712646,
      "learning_rate": 4.55323469683783e-06,
      "loss": 0.7803,
      "step": 2157
    },
    {
      "epoch": 0.08943594844378135,
      "grad_norm": 0.47068193554878235,
      "learning_rate": 4.553027477309462e-06,
      "loss": 0.7443,
      "step": 2158
    },
    {
      "epoch": 0.08947739234945501,
      "grad_norm": 0.44261103868484497,
      "learning_rate": 4.552820257781094e-06,
      "loss": 0.793,
      "step": 2159
    },
    {
      "epoch": 0.08951883625512869,
      "grad_norm": 0.4177280068397522,
      "learning_rate": 4.552613038252725e-06,
      "loss": 0.7266,
      "step": 2160
    },
    {
      "epoch": 0.08956028016080235,
      "grad_norm": 0.47246408462524414,
      "learning_rate": 4.552405818724357e-06,
      "loss": 0.7979,
      "step": 2161
    },
    {
      "epoch": 0.08960172406647603,
      "grad_norm": 0.4821568429470062,
      "learning_rate": 4.552198599195989e-06,
      "loss": 0.731,
      "step": 2162
    },
    {
      "epoch": 0.08964316797214969,
      "grad_norm": 0.45064979791641235,
      "learning_rate": 4.55199137966762e-06,
      "loss": 0.7869,
      "step": 2163
    },
    {
      "epoch": 0.08968461187782337,
      "grad_norm": 0.44254612922668457,
      "learning_rate": 4.551784160139252e-06,
      "loss": 0.729,
      "step": 2164
    },
    {
      "epoch": 0.08972605578349703,
      "grad_norm": 0.41059446334838867,
      "learning_rate": 4.551576940610883e-06,
      "loss": 0.752,
      "step": 2165
    },
    {
      "epoch": 0.08976749968917071,
      "grad_norm": 0.4164467751979828,
      "learning_rate": 4.551369721082515e-06,
      "loss": 0.7522,
      "step": 2166
    },
    {
      "epoch": 0.08980894359484438,
      "grad_norm": 0.4167279303073883,
      "learning_rate": 4.551162501554147e-06,
      "loss": 0.7561,
      "step": 2167
    },
    {
      "epoch": 0.08985038750051805,
      "grad_norm": 0.47502654790878296,
      "learning_rate": 4.550955282025779e-06,
      "loss": 0.8206,
      "step": 2168
    },
    {
      "epoch": 0.08989183140619172,
      "grad_norm": 0.4153778553009033,
      "learning_rate": 4.5507480624974095e-06,
      "loss": 0.708,
      "step": 2169
    },
    {
      "epoch": 0.08993327531186539,
      "grad_norm": 0.5410426259040833,
      "learning_rate": 4.550540842969042e-06,
      "loss": 0.8738,
      "step": 2170
    },
    {
      "epoch": 0.08997471921753906,
      "grad_norm": 0.40579575300216675,
      "learning_rate": 4.550333623440674e-06,
      "loss": 0.7051,
      "step": 2171
    },
    {
      "epoch": 0.09001616312321273,
      "grad_norm": 0.4710105359554291,
      "learning_rate": 4.550126403912305e-06,
      "loss": 0.7837,
      "step": 2172
    },
    {
      "epoch": 0.0900576070288864,
      "grad_norm": 0.6112576127052307,
      "learning_rate": 4.549919184383937e-06,
      "loss": 0.7905,
      "step": 2173
    },
    {
      "epoch": 0.09009905093456007,
      "grad_norm": 0.4224766194820404,
      "learning_rate": 4.549711964855568e-06,
      "loss": 0.7351,
      "step": 2174
    },
    {
      "epoch": 0.09014049484023375,
      "grad_norm": 0.4324423670768738,
      "learning_rate": 4.5495047453272e-06,
      "loss": 0.762,
      "step": 2175
    },
    {
      "epoch": 0.09018193874590741,
      "grad_norm": 0.4714941382408142,
      "learning_rate": 4.549297525798832e-06,
      "loss": 0.8257,
      "step": 2176
    },
    {
      "epoch": 0.09022338265158109,
      "grad_norm": 0.45183053612709045,
      "learning_rate": 4.549090306270463e-06,
      "loss": 0.7898,
      "step": 2177
    },
    {
      "epoch": 0.09026482655725476,
      "grad_norm": 0.4123820662498474,
      "learning_rate": 4.548883086742095e-06,
      "loss": 0.7273,
      "step": 2178
    },
    {
      "epoch": 0.09030627046292843,
      "grad_norm": 0.41791433095932007,
      "learning_rate": 4.548675867213727e-06,
      "loss": 0.7314,
      "step": 2179
    },
    {
      "epoch": 0.0903477143686021,
      "grad_norm": 0.4358455538749695,
      "learning_rate": 4.548468647685358e-06,
      "loss": 0.7366,
      "step": 2180
    },
    {
      "epoch": 0.09038915827427577,
      "grad_norm": 0.4254612624645233,
      "learning_rate": 4.5482614281569895e-06,
      "loss": 0.7024,
      "step": 2181
    },
    {
      "epoch": 0.09043060217994944,
      "grad_norm": 0.4269399642944336,
      "learning_rate": 4.548054208628622e-06,
      "loss": 0.7455,
      "step": 2182
    },
    {
      "epoch": 0.0904720460856231,
      "grad_norm": 0.443946897983551,
      "learning_rate": 4.547846989100253e-06,
      "loss": 0.75,
      "step": 2183
    },
    {
      "epoch": 0.09051348999129678,
      "grad_norm": 0.41850340366363525,
      "learning_rate": 4.5476397695718845e-06,
      "loss": 0.7671,
      "step": 2184
    },
    {
      "epoch": 0.09055493389697045,
      "grad_norm": 0.434459924697876,
      "learning_rate": 4.547432550043516e-06,
      "loss": 0.7163,
      "step": 2185
    },
    {
      "epoch": 0.09059637780264412,
      "grad_norm": 0.44378164410591125,
      "learning_rate": 4.547225330515148e-06,
      "loss": 0.7025,
      "step": 2186
    },
    {
      "epoch": 0.09063782170831779,
      "grad_norm": 0.45942020416259766,
      "learning_rate": 4.54701811098678e-06,
      "loss": 0.7581,
      "step": 2187
    },
    {
      "epoch": 0.09067926561399146,
      "grad_norm": 0.4831501245498657,
      "learning_rate": 4.546810891458412e-06,
      "loss": 0.79,
      "step": 2188
    },
    {
      "epoch": 0.09072070951966513,
      "grad_norm": 0.45938074588775635,
      "learning_rate": 4.546603671930043e-06,
      "loss": 0.782,
      "step": 2189
    },
    {
      "epoch": 0.0907621534253388,
      "grad_norm": 0.42186984419822693,
      "learning_rate": 4.5463964524016745e-06,
      "loss": 0.6826,
      "step": 2190
    },
    {
      "epoch": 0.09080359733101248,
      "grad_norm": 0.38702264428138733,
      "learning_rate": 4.546189232873307e-06,
      "loss": 0.7036,
      "step": 2191
    },
    {
      "epoch": 0.09084504123668614,
      "grad_norm": 0.4260973632335663,
      "learning_rate": 4.545982013344938e-06,
      "loss": 0.7407,
      "step": 2192
    },
    {
      "epoch": 0.09088648514235982,
      "grad_norm": 0.40211766958236694,
      "learning_rate": 4.5457747938165695e-06,
      "loss": 0.7639,
      "step": 2193
    },
    {
      "epoch": 0.09092792904803348,
      "grad_norm": 0.43616944551467896,
      "learning_rate": 4.545567574288201e-06,
      "loss": 0.7733,
      "step": 2194
    },
    {
      "epoch": 0.09096937295370716,
      "grad_norm": 0.47748178243637085,
      "learning_rate": 4.545360354759833e-06,
      "loss": 0.7722,
      "step": 2195
    },
    {
      "epoch": 0.09101081685938082,
      "grad_norm": 0.4525977373123169,
      "learning_rate": 4.5451531352314645e-06,
      "loss": 0.7145,
      "step": 2196
    },
    {
      "epoch": 0.0910522607650545,
      "grad_norm": 0.42960894107818604,
      "learning_rate": 4.544945915703096e-06,
      "loss": 0.7783,
      "step": 2197
    },
    {
      "epoch": 0.09109370467072817,
      "grad_norm": 0.43034225702285767,
      "learning_rate": 4.544738696174728e-06,
      "loss": 0.7727,
      "step": 2198
    },
    {
      "epoch": 0.09113514857640184,
      "grad_norm": 0.43229594826698303,
      "learning_rate": 4.5445314766463595e-06,
      "loss": 0.8049,
      "step": 2199
    },
    {
      "epoch": 0.0911765924820755,
      "grad_norm": 0.4749002456665039,
      "learning_rate": 4.544324257117991e-06,
      "loss": 0.8025,
      "step": 2200
    },
    {
      "epoch": 0.09121803638774918,
      "grad_norm": 0.4332364499568939,
      "learning_rate": 4.544117037589622e-06,
      "loss": 0.748,
      "step": 2201
    },
    {
      "epoch": 0.09125948029342285,
      "grad_norm": 0.4769018292427063,
      "learning_rate": 4.5439098180612545e-06,
      "loss": 0.7388,
      "step": 2202
    },
    {
      "epoch": 0.09130092419909652,
      "grad_norm": 0.4501079320907593,
      "learning_rate": 4.543702598532886e-06,
      "loss": 0.7856,
      "step": 2203
    },
    {
      "epoch": 0.0913423681047702,
      "grad_norm": 0.41608110070228577,
      "learning_rate": 4.543495379004518e-06,
      "loss": 0.7788,
      "step": 2204
    },
    {
      "epoch": 0.09138381201044386,
      "grad_norm": 0.4151354730129242,
      "learning_rate": 4.5432881594761495e-06,
      "loss": 0.7214,
      "step": 2205
    },
    {
      "epoch": 0.09142525591611754,
      "grad_norm": 0.45879191160202026,
      "learning_rate": 4.543080939947781e-06,
      "loss": 0.7197,
      "step": 2206
    },
    {
      "epoch": 0.0914666998217912,
      "grad_norm": 0.45620161294937134,
      "learning_rate": 4.542873720419413e-06,
      "loss": 0.7737,
      "step": 2207
    },
    {
      "epoch": 0.09150814372746488,
      "grad_norm": 0.43143945932388306,
      "learning_rate": 4.5426665008910445e-06,
      "loss": 0.7932,
      "step": 2208
    },
    {
      "epoch": 0.09154958763313854,
      "grad_norm": 0.44396746158599854,
      "learning_rate": 4.542459281362676e-06,
      "loss": 0.8022,
      "step": 2209
    },
    {
      "epoch": 0.09159103153881222,
      "grad_norm": 0.4247385263442993,
      "learning_rate": 4.542252061834307e-06,
      "loss": 0.7632,
      "step": 2210
    },
    {
      "epoch": 0.09163247544448588,
      "grad_norm": 0.44195446372032166,
      "learning_rate": 4.5420448423059395e-06,
      "loss": 0.7297,
      "step": 2211
    },
    {
      "epoch": 0.09167391935015956,
      "grad_norm": 0.4388570487499237,
      "learning_rate": 4.541837622777571e-06,
      "loss": 0.7844,
      "step": 2212
    },
    {
      "epoch": 0.09171536325583322,
      "grad_norm": 0.44064751267433167,
      "learning_rate": 4.541630403249202e-06,
      "loss": 0.7893,
      "step": 2213
    },
    {
      "epoch": 0.0917568071615069,
      "grad_norm": 0.4489535391330719,
      "learning_rate": 4.5414231837208345e-06,
      "loss": 0.7346,
      "step": 2214
    },
    {
      "epoch": 0.09179825106718058,
      "grad_norm": 0.43959829211235046,
      "learning_rate": 4.541215964192466e-06,
      "loss": 0.7675,
      "step": 2215
    },
    {
      "epoch": 0.09183969497285424,
      "grad_norm": 0.4443313181400299,
      "learning_rate": 4.541008744664097e-06,
      "loss": 0.7329,
      "step": 2216
    },
    {
      "epoch": 0.09188113887852792,
      "grad_norm": 0.4552377760410309,
      "learning_rate": 4.540801525135729e-06,
      "loss": 0.7402,
      "step": 2217
    },
    {
      "epoch": 0.09192258278420158,
      "grad_norm": 0.45374757051467896,
      "learning_rate": 4.540594305607361e-06,
      "loss": 0.7852,
      "step": 2218
    },
    {
      "epoch": 0.09196402668987526,
      "grad_norm": 0.40958738327026367,
      "learning_rate": 4.540387086078992e-06,
      "loss": 0.7463,
      "step": 2219
    },
    {
      "epoch": 0.09200547059554892,
      "grad_norm": 0.4430464208126068,
      "learning_rate": 4.5401798665506245e-06,
      "loss": 0.7616,
      "step": 2220
    },
    {
      "epoch": 0.0920469145012226,
      "grad_norm": 0.4105941355228424,
      "learning_rate": 4.539972647022255e-06,
      "loss": 0.7471,
      "step": 2221
    },
    {
      "epoch": 0.09208835840689626,
      "grad_norm": 0.4176451861858368,
      "learning_rate": 4.539765427493887e-06,
      "loss": 0.7048,
      "step": 2222
    },
    {
      "epoch": 0.09212980231256994,
      "grad_norm": 0.42156854271888733,
      "learning_rate": 4.5395582079655195e-06,
      "loss": 0.7032,
      "step": 2223
    },
    {
      "epoch": 0.0921712462182436,
      "grad_norm": 0.42863553762435913,
      "learning_rate": 4.539350988437151e-06,
      "loss": 0.7466,
      "step": 2224
    },
    {
      "epoch": 0.09221269012391728,
      "grad_norm": 0.4090249836444855,
      "learning_rate": 4.539143768908782e-06,
      "loss": 0.7502,
      "step": 2225
    },
    {
      "epoch": 0.09225413402959094,
      "grad_norm": 0.4316571354866028,
      "learning_rate": 4.538936549380414e-06,
      "loss": 0.7739,
      "step": 2226
    },
    {
      "epoch": 0.09229557793526462,
      "grad_norm": 0.48924705386161804,
      "learning_rate": 4.538729329852046e-06,
      "loss": 0.8042,
      "step": 2227
    },
    {
      "epoch": 0.0923370218409383,
      "grad_norm": 0.4519100785255432,
      "learning_rate": 4.538522110323677e-06,
      "loss": 0.7473,
      "step": 2228
    },
    {
      "epoch": 0.09237846574661196,
      "grad_norm": 0.417803019285202,
      "learning_rate": 4.538314890795309e-06,
      "loss": 0.7415,
      "step": 2229
    },
    {
      "epoch": 0.09241990965228564,
      "grad_norm": 0.4330748915672302,
      "learning_rate": 4.53810767126694e-06,
      "loss": 0.7053,
      "step": 2230
    },
    {
      "epoch": 0.0924613535579593,
      "grad_norm": 0.43731456995010376,
      "learning_rate": 4.537900451738572e-06,
      "loss": 0.7944,
      "step": 2231
    },
    {
      "epoch": 0.09250279746363298,
      "grad_norm": 0.4673445224761963,
      "learning_rate": 4.537693232210204e-06,
      "loss": 0.7341,
      "step": 2232
    },
    {
      "epoch": 0.09254424136930664,
      "grad_norm": 0.4800216853618622,
      "learning_rate": 4.537486012681835e-06,
      "loss": 0.7487,
      "step": 2233
    },
    {
      "epoch": 0.09258568527498032,
      "grad_norm": 0.46607205271720886,
      "learning_rate": 4.537278793153467e-06,
      "loss": 0.8008,
      "step": 2234
    },
    {
      "epoch": 0.09262712918065398,
      "grad_norm": 0.4351932108402252,
      "learning_rate": 4.537071573625099e-06,
      "loss": 0.7266,
      "step": 2235
    },
    {
      "epoch": 0.09266857308632766,
      "grad_norm": 0.4656636714935303,
      "learning_rate": 4.536864354096731e-06,
      "loss": 0.7661,
      "step": 2236
    },
    {
      "epoch": 0.09271001699200132,
      "grad_norm": 0.48204293847084045,
      "learning_rate": 4.5366571345683615e-06,
      "loss": 0.8252,
      "step": 2237
    },
    {
      "epoch": 0.092751460897675,
      "grad_norm": 0.4329359531402588,
      "learning_rate": 4.536449915039994e-06,
      "loss": 0.7908,
      "step": 2238
    },
    {
      "epoch": 0.09279290480334866,
      "grad_norm": 0.47167110443115234,
      "learning_rate": 4.536242695511626e-06,
      "loss": 0.7334,
      "step": 2239
    },
    {
      "epoch": 0.09283434870902234,
      "grad_norm": 0.43544137477874756,
      "learning_rate": 4.536035475983257e-06,
      "loss": 0.7808,
      "step": 2240
    },
    {
      "epoch": 0.09287579261469601,
      "grad_norm": 0.4410264790058136,
      "learning_rate": 4.535828256454889e-06,
      "loss": 0.748,
      "step": 2241
    },
    {
      "epoch": 0.09291723652036968,
      "grad_norm": 0.40531888604164124,
      "learning_rate": 4.53562103692652e-06,
      "loss": 0.7112,
      "step": 2242
    },
    {
      "epoch": 0.09295868042604336,
      "grad_norm": 0.45738229155540466,
      "learning_rate": 4.535413817398152e-06,
      "loss": 0.7966,
      "step": 2243
    },
    {
      "epoch": 0.09300012433171702,
      "grad_norm": 0.4197201728820801,
      "learning_rate": 4.535206597869784e-06,
      "loss": 0.7849,
      "step": 2244
    },
    {
      "epoch": 0.0930415682373907,
      "grad_norm": 0.4063815176486969,
      "learning_rate": 4.534999378341415e-06,
      "loss": 0.7271,
      "step": 2245
    },
    {
      "epoch": 0.09308301214306436,
      "grad_norm": 0.4056161642074585,
      "learning_rate": 4.5347921588130465e-06,
      "loss": 0.7253,
      "step": 2246
    },
    {
      "epoch": 0.09312445604873804,
      "grad_norm": 0.42396119236946106,
      "learning_rate": 4.534584939284679e-06,
      "loss": 0.7036,
      "step": 2247
    },
    {
      "epoch": 0.0931658999544117,
      "grad_norm": 0.448004812002182,
      "learning_rate": 4.53437771975631e-06,
      "loss": 0.761,
      "step": 2248
    },
    {
      "epoch": 0.09320734386008538,
      "grad_norm": 0.47135046124458313,
      "learning_rate": 4.5341705002279415e-06,
      "loss": 0.7759,
      "step": 2249
    },
    {
      "epoch": 0.09324878776575904,
      "grad_norm": 0.44456666707992554,
      "learning_rate": 4.533963280699574e-06,
      "loss": 0.7004,
      "step": 2250
    },
    {
      "epoch": 0.09329023167143272,
      "grad_norm": 0.4362632930278778,
      "learning_rate": 4.533756061171205e-06,
      "loss": 0.7052,
      "step": 2251
    },
    {
      "epoch": 0.0933316755771064,
      "grad_norm": 0.43177878856658936,
      "learning_rate": 4.5335488416428365e-06,
      "loss": 0.761,
      "step": 2252
    },
    {
      "epoch": 0.09337311948278006,
      "grad_norm": 0.4715634882450104,
      "learning_rate": 4.533341622114468e-06,
      "loss": 0.7847,
      "step": 2253
    },
    {
      "epoch": 0.09341456338845373,
      "grad_norm": 0.433210551738739,
      "learning_rate": 4.5331344025861e-06,
      "loss": 0.7864,
      "step": 2254
    },
    {
      "epoch": 0.0934560072941274,
      "grad_norm": 0.4599769711494446,
      "learning_rate": 4.5329271830577315e-06,
      "loss": 0.7391,
      "step": 2255
    },
    {
      "epoch": 0.09349745119980107,
      "grad_norm": 0.45310524106025696,
      "learning_rate": 4.532719963529364e-06,
      "loss": 0.6971,
      "step": 2256
    },
    {
      "epoch": 0.09353889510547474,
      "grad_norm": 0.43435582518577576,
      "learning_rate": 4.532512744000995e-06,
      "loss": 0.7406,
      "step": 2257
    },
    {
      "epoch": 0.09358033901114841,
      "grad_norm": 0.43927106261253357,
      "learning_rate": 4.5323055244726265e-06,
      "loss": 0.7043,
      "step": 2258
    },
    {
      "epoch": 0.09362178291682208,
      "grad_norm": 0.4449627697467804,
      "learning_rate": 4.532098304944259e-06,
      "loss": 0.6885,
      "step": 2259
    },
    {
      "epoch": 0.09366322682249575,
      "grad_norm": 0.45201462507247925,
      "learning_rate": 4.53189108541589e-06,
      "loss": 0.8042,
      "step": 2260
    },
    {
      "epoch": 0.09370467072816942,
      "grad_norm": 0.4368429481983185,
      "learning_rate": 4.5316838658875215e-06,
      "loss": 0.7289,
      "step": 2261
    },
    {
      "epoch": 0.0937461146338431,
      "grad_norm": 0.45706290006637573,
      "learning_rate": 4.531476646359153e-06,
      "loss": 0.769,
      "step": 2262
    },
    {
      "epoch": 0.09378755853951676,
      "grad_norm": 0.42083480954170227,
      "learning_rate": 4.531269426830785e-06,
      "loss": 0.6844,
      "step": 2263
    },
    {
      "epoch": 0.09382900244519043,
      "grad_norm": 0.44023415446281433,
      "learning_rate": 4.5310622073024165e-06,
      "loss": 0.7339,
      "step": 2264
    },
    {
      "epoch": 0.09387044635086411,
      "grad_norm": 0.42445361614227295,
      "learning_rate": 4.530854987774048e-06,
      "loss": 0.6924,
      "step": 2265
    },
    {
      "epoch": 0.09391189025653778,
      "grad_norm": 0.46486759185791016,
      "learning_rate": 4.53064776824568e-06,
      "loss": 0.734,
      "step": 2266
    },
    {
      "epoch": 0.09395333416221145,
      "grad_norm": 0.45771509408950806,
      "learning_rate": 4.5304405487173115e-06,
      "loss": 0.8447,
      "step": 2267
    },
    {
      "epoch": 0.09399477806788512,
      "grad_norm": 0.45528465509414673,
      "learning_rate": 4.530233329188943e-06,
      "loss": 0.7349,
      "step": 2268
    },
    {
      "epoch": 0.09403622197355879,
      "grad_norm": 0.44459179043769836,
      "learning_rate": 4.530026109660574e-06,
      "loss": 0.769,
      "step": 2269
    },
    {
      "epoch": 0.09407766587923246,
      "grad_norm": 0.4331132173538208,
      "learning_rate": 4.5298188901322065e-06,
      "loss": 0.6864,
      "step": 2270
    },
    {
      "epoch": 0.09411910978490613,
      "grad_norm": 0.4719201326370239,
      "learning_rate": 4.529611670603838e-06,
      "loss": 0.7505,
      "step": 2271
    },
    {
      "epoch": 0.0941605536905798,
      "grad_norm": 0.41887593269348145,
      "learning_rate": 4.52940445107547e-06,
      "loss": 0.7229,
      "step": 2272
    },
    {
      "epoch": 0.09420199759625347,
      "grad_norm": 0.4257392883300781,
      "learning_rate": 4.5291972315471015e-06,
      "loss": 0.7419,
      "step": 2273
    },
    {
      "epoch": 0.09424344150192714,
      "grad_norm": 0.4003089666366577,
      "learning_rate": 4.528990012018733e-06,
      "loss": 0.7562,
      "step": 2274
    },
    {
      "epoch": 0.09428488540760081,
      "grad_norm": 0.4327313303947449,
      "learning_rate": 4.528782792490365e-06,
      "loss": 0.7163,
      "step": 2275
    },
    {
      "epoch": 0.09432632931327448,
      "grad_norm": 0.4412200450897217,
      "learning_rate": 4.5285755729619965e-06,
      "loss": 0.7474,
      "step": 2276
    },
    {
      "epoch": 0.09436777321894815,
      "grad_norm": 0.48064202070236206,
      "learning_rate": 4.528368353433628e-06,
      "loss": 0.7634,
      "step": 2277
    },
    {
      "epoch": 0.09440921712462183,
      "grad_norm": 0.49322381615638733,
      "learning_rate": 4.528161133905259e-06,
      "loss": 0.7866,
      "step": 2278
    },
    {
      "epoch": 0.0944506610302955,
      "grad_norm": 0.4353357255458832,
      "learning_rate": 4.5279539143768915e-06,
      "loss": 0.7461,
      "step": 2279
    },
    {
      "epoch": 0.09449210493596917,
      "grad_norm": 0.4630735516548157,
      "learning_rate": 4.527746694848523e-06,
      "loss": 0.7156,
      "step": 2280
    },
    {
      "epoch": 0.09453354884164283,
      "grad_norm": 0.4633691906929016,
      "learning_rate": 4.527539475320154e-06,
      "loss": 0.7847,
      "step": 2281
    },
    {
      "epoch": 0.09457499274731651,
      "grad_norm": 0.4207591712474823,
      "learning_rate": 4.527332255791786e-06,
      "loss": 0.7368,
      "step": 2282
    },
    {
      "epoch": 0.09461643665299017,
      "grad_norm": 0.4572641849517822,
      "learning_rate": 4.527125036263418e-06,
      "loss": 0.8069,
      "step": 2283
    },
    {
      "epoch": 0.09465788055866385,
      "grad_norm": 0.41714513301849365,
      "learning_rate": 4.526917816735049e-06,
      "loss": 0.7258,
      "step": 2284
    },
    {
      "epoch": 0.09469932446433751,
      "grad_norm": 0.45576101541519165,
      "learning_rate": 4.526710597206681e-06,
      "loss": 0.7556,
      "step": 2285
    },
    {
      "epoch": 0.09474076837001119,
      "grad_norm": 0.4351746141910553,
      "learning_rate": 4.526503377678313e-06,
      "loss": 0.7434,
      "step": 2286
    },
    {
      "epoch": 0.09478221227568485,
      "grad_norm": 0.4505622386932373,
      "learning_rate": 4.526296158149944e-06,
      "loss": 0.7644,
      "step": 2287
    },
    {
      "epoch": 0.09482365618135853,
      "grad_norm": 0.4147385060787201,
      "learning_rate": 4.5260889386215766e-06,
      "loss": 0.7148,
      "step": 2288
    },
    {
      "epoch": 0.09486510008703221,
      "grad_norm": 0.4238247275352478,
      "learning_rate": 4.525881719093207e-06,
      "loss": 0.7292,
      "step": 2289
    },
    {
      "epoch": 0.09490654399270587,
      "grad_norm": 0.44142287969589233,
      "learning_rate": 4.525674499564839e-06,
      "loss": 0.7595,
      "step": 2290
    },
    {
      "epoch": 0.09494798789837955,
      "grad_norm": 0.45950353145599365,
      "learning_rate": 4.525467280036471e-06,
      "loss": 0.7754,
      "step": 2291
    },
    {
      "epoch": 0.09498943180405321,
      "grad_norm": 0.43218994140625,
      "learning_rate": 4.525260060508103e-06,
      "loss": 0.7825,
      "step": 2292
    },
    {
      "epoch": 0.09503087570972689,
      "grad_norm": 0.4497007131576538,
      "learning_rate": 4.525052840979734e-06,
      "loss": 0.7441,
      "step": 2293
    },
    {
      "epoch": 0.09507231961540055,
      "grad_norm": 0.43192440271377563,
      "learning_rate": 4.524845621451366e-06,
      "loss": 0.699,
      "step": 2294
    },
    {
      "epoch": 0.09511376352107423,
      "grad_norm": 0.5404394865036011,
      "learning_rate": 4.524638401922998e-06,
      "loss": 0.7585,
      "step": 2295
    },
    {
      "epoch": 0.09515520742674789,
      "grad_norm": 0.41696080565452576,
      "learning_rate": 4.524431182394629e-06,
      "loss": 0.7375,
      "step": 2296
    },
    {
      "epoch": 0.09519665133242157,
      "grad_norm": 0.43760812282562256,
      "learning_rate": 4.524223962866261e-06,
      "loss": 0.7527,
      "step": 2297
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.43164634704589844,
      "learning_rate": 4.524016743337892e-06,
      "loss": 0.7449,
      "step": 2298
    },
    {
      "epoch": 0.09527953914376891,
      "grad_norm": 0.42652374505996704,
      "learning_rate": 4.523809523809524e-06,
      "loss": 0.7286,
      "step": 2299
    },
    {
      "epoch": 0.09532098304944257,
      "grad_norm": 0.414614737033844,
      "learning_rate": 4.523602304281156e-06,
      "loss": 0.7874,
      "step": 2300
    },
    {
      "epoch": 0.09536242695511625,
      "grad_norm": 0.4192710518836975,
      "learning_rate": 4.523395084752787e-06,
      "loss": 0.8032,
      "step": 2301
    },
    {
      "epoch": 0.09540387086078993,
      "grad_norm": 0.4798049032688141,
      "learning_rate": 4.523187865224419e-06,
      "loss": 0.7407,
      "step": 2302
    },
    {
      "epoch": 0.09544531476646359,
      "grad_norm": 0.49399110674858093,
      "learning_rate": 4.522980645696051e-06,
      "loss": 0.7969,
      "step": 2303
    },
    {
      "epoch": 0.09548675867213727,
      "grad_norm": 0.4425923824310303,
      "learning_rate": 4.522773426167683e-06,
      "loss": 0.7661,
      "step": 2304
    },
    {
      "epoch": 0.09552820257781093,
      "grad_norm": 0.40333569049835205,
      "learning_rate": 4.5225662066393135e-06,
      "loss": 0.7089,
      "step": 2305
    },
    {
      "epoch": 0.09556964648348461,
      "grad_norm": 0.448172390460968,
      "learning_rate": 4.522358987110946e-06,
      "loss": 0.7136,
      "step": 2306
    },
    {
      "epoch": 0.09561109038915827,
      "grad_norm": 0.4313826858997345,
      "learning_rate": 4.522151767582577e-06,
      "loss": 0.738,
      "step": 2307
    },
    {
      "epoch": 0.09565253429483195,
      "grad_norm": 0.45727795362472534,
      "learning_rate": 4.521944548054209e-06,
      "loss": 0.7377,
      "step": 2308
    },
    {
      "epoch": 0.09569397820050561,
      "grad_norm": 0.4521796405315399,
      "learning_rate": 4.521737328525841e-06,
      "loss": 0.7378,
      "step": 2309
    },
    {
      "epoch": 0.09573542210617929,
      "grad_norm": 0.4456719756126404,
      "learning_rate": 4.521530108997472e-06,
      "loss": 0.7361,
      "step": 2310
    },
    {
      "epoch": 0.09577686601185295,
      "grad_norm": 0.4329984486103058,
      "learning_rate": 4.521322889469104e-06,
      "loss": 0.7621,
      "step": 2311
    },
    {
      "epoch": 0.09581830991752663,
      "grad_norm": 0.42228633165359497,
      "learning_rate": 4.521115669940736e-06,
      "loss": 0.7156,
      "step": 2312
    },
    {
      "epoch": 0.09585975382320029,
      "grad_norm": 0.4969192445278168,
      "learning_rate": 4.520908450412367e-06,
      "loss": 0.7617,
      "step": 2313
    },
    {
      "epoch": 0.09590119772887397,
      "grad_norm": 0.4804287254810333,
      "learning_rate": 4.5207012308839985e-06,
      "loss": 0.8372,
      "step": 2314
    },
    {
      "epoch": 0.09594264163454765,
      "grad_norm": 0.43762508034706116,
      "learning_rate": 4.520494011355631e-06,
      "loss": 0.7146,
      "step": 2315
    },
    {
      "epoch": 0.09598408554022131,
      "grad_norm": 0.5054106712341309,
      "learning_rate": 4.520286791827262e-06,
      "loss": 0.8176,
      "step": 2316
    },
    {
      "epoch": 0.09602552944589499,
      "grad_norm": 0.4765777885913849,
      "learning_rate": 4.5200795722988935e-06,
      "loss": 0.7622,
      "step": 2317
    },
    {
      "epoch": 0.09606697335156865,
      "grad_norm": 0.4310450255870819,
      "learning_rate": 4.519872352770526e-06,
      "loss": 0.7227,
      "step": 2318
    },
    {
      "epoch": 0.09610841725724233,
      "grad_norm": 0.47286340594291687,
      "learning_rate": 4.519665133242157e-06,
      "loss": 0.8103,
      "step": 2319
    },
    {
      "epoch": 0.09614986116291599,
      "grad_norm": 0.43081432580947876,
      "learning_rate": 4.519457913713789e-06,
      "loss": 0.7185,
      "step": 2320
    },
    {
      "epoch": 0.09619130506858967,
      "grad_norm": 0.4538341164588928,
      "learning_rate": 4.51925069418542e-06,
      "loss": 0.7561,
      "step": 2321
    },
    {
      "epoch": 0.09623274897426333,
      "grad_norm": 0.44100120663642883,
      "learning_rate": 4.519043474657052e-06,
      "loss": 0.7252,
      "step": 2322
    },
    {
      "epoch": 0.096274192879937,
      "grad_norm": 0.4621768891811371,
      "learning_rate": 4.5188362551286835e-06,
      "loss": 0.8247,
      "step": 2323
    },
    {
      "epoch": 0.09631563678561067,
      "grad_norm": 0.4036440849304199,
      "learning_rate": 4.518629035600316e-06,
      "loss": 0.6981,
      "step": 2324
    },
    {
      "epoch": 0.09635708069128435,
      "grad_norm": 0.4260554313659668,
      "learning_rate": 4.518421816071947e-06,
      "loss": 0.7581,
      "step": 2325
    },
    {
      "epoch": 0.09639852459695802,
      "grad_norm": 0.3968186378479004,
      "learning_rate": 4.5182145965435785e-06,
      "loss": 0.6887,
      "step": 2326
    },
    {
      "epoch": 0.09643996850263169,
      "grad_norm": 0.4328472912311554,
      "learning_rate": 4.518007377015211e-06,
      "loss": 0.7793,
      "step": 2327
    },
    {
      "epoch": 0.09648141240830536,
      "grad_norm": 0.40531015396118164,
      "learning_rate": 4.517800157486842e-06,
      "loss": 0.7085,
      "step": 2328
    },
    {
      "epoch": 0.09652285631397903,
      "grad_norm": 0.44980961084365845,
      "learning_rate": 4.5175929379584735e-06,
      "loss": 0.7505,
      "step": 2329
    },
    {
      "epoch": 0.0965643002196527,
      "grad_norm": 0.4177475571632385,
      "learning_rate": 4.517385718430105e-06,
      "loss": 0.7262,
      "step": 2330
    },
    {
      "epoch": 0.09660574412532637,
      "grad_norm": 0.4285491704940796,
      "learning_rate": 4.517178498901737e-06,
      "loss": 0.7157,
      "step": 2331
    },
    {
      "epoch": 0.09664718803100004,
      "grad_norm": 0.43489131331443787,
      "learning_rate": 4.5169712793733685e-06,
      "loss": 0.7301,
      "step": 2332
    },
    {
      "epoch": 0.09668863193667371,
      "grad_norm": 0.4530631899833679,
      "learning_rate": 4.516764059845e-06,
      "loss": 0.8218,
      "step": 2333
    },
    {
      "epoch": 0.09673007584234738,
      "grad_norm": 0.47835442423820496,
      "learning_rate": 4.516556840316631e-06,
      "loss": 0.7917,
      "step": 2334
    },
    {
      "epoch": 0.09677151974802105,
      "grad_norm": 0.4514983296394348,
      "learning_rate": 4.5163496207882635e-06,
      "loss": 0.7356,
      "step": 2335
    },
    {
      "epoch": 0.09681296365369473,
      "grad_norm": 0.45691269636154175,
      "learning_rate": 4.516142401259895e-06,
      "loss": 0.7798,
      "step": 2336
    },
    {
      "epoch": 0.09685440755936839,
      "grad_norm": 0.543431282043457,
      "learning_rate": 4.515935181731526e-06,
      "loss": 0.7854,
      "step": 2337
    },
    {
      "epoch": 0.09689585146504207,
      "grad_norm": 0.44547513127326965,
      "learning_rate": 4.5157279622031585e-06,
      "loss": 0.7864,
      "step": 2338
    },
    {
      "epoch": 0.09693729537071574,
      "grad_norm": 0.42431867122650146,
      "learning_rate": 4.51552074267479e-06,
      "loss": 0.7405,
      "step": 2339
    },
    {
      "epoch": 0.0969787392763894,
      "grad_norm": 0.4645279347896576,
      "learning_rate": 4.515313523146422e-06,
      "loss": 0.7289,
      "step": 2340
    },
    {
      "epoch": 0.09702018318206308,
      "grad_norm": 0.46230387687683105,
      "learning_rate": 4.5151063036180535e-06,
      "loss": 0.7703,
      "step": 2341
    },
    {
      "epoch": 0.09706162708773675,
      "grad_norm": 0.4439159333705902,
      "learning_rate": 4.514899084089685e-06,
      "loss": 0.7275,
      "step": 2342
    },
    {
      "epoch": 0.09710307099341042,
      "grad_norm": 0.43524986505508423,
      "learning_rate": 4.514691864561316e-06,
      "loss": 0.7283,
      "step": 2343
    },
    {
      "epoch": 0.09714451489908409,
      "grad_norm": 0.4739815890789032,
      "learning_rate": 4.5144846450329486e-06,
      "loss": 0.7825,
      "step": 2344
    },
    {
      "epoch": 0.09718595880475776,
      "grad_norm": 0.4382423460483551,
      "learning_rate": 4.51427742550458e-06,
      "loss": 0.7793,
      "step": 2345
    },
    {
      "epoch": 0.09722740271043143,
      "grad_norm": 0.43471187353134155,
      "learning_rate": 4.514070205976211e-06,
      "loss": 0.7449,
      "step": 2346
    },
    {
      "epoch": 0.0972688466161051,
      "grad_norm": 0.4425707161426544,
      "learning_rate": 4.5138629864478436e-06,
      "loss": 0.7762,
      "step": 2347
    },
    {
      "epoch": 0.09731029052177877,
      "grad_norm": 0.43949663639068604,
      "learning_rate": 4.513655766919475e-06,
      "loss": 0.7577,
      "step": 2348
    },
    {
      "epoch": 0.09735173442745244,
      "grad_norm": 0.42473387718200684,
      "learning_rate": 4.513448547391106e-06,
      "loss": 0.6975,
      "step": 2349
    },
    {
      "epoch": 0.09739317833312611,
      "grad_norm": 0.42673492431640625,
      "learning_rate": 4.513241327862738e-06,
      "loss": 0.7512,
      "step": 2350
    },
    {
      "epoch": 0.09743462223879978,
      "grad_norm": 0.43722769618034363,
      "learning_rate": 4.51303410833437e-06,
      "loss": 0.7368,
      "step": 2351
    },
    {
      "epoch": 0.09747606614447346,
      "grad_norm": 0.4997045397758484,
      "learning_rate": 4.512826888806001e-06,
      "loss": 0.8533,
      "step": 2352
    },
    {
      "epoch": 0.09751751005014712,
      "grad_norm": 0.45096832513809204,
      "learning_rate": 4.512619669277633e-06,
      "loss": 0.7732,
      "step": 2353
    },
    {
      "epoch": 0.0975589539558208,
      "grad_norm": 0.43369951844215393,
      "learning_rate": 4.512412449749265e-06,
      "loss": 0.6788,
      "step": 2354
    },
    {
      "epoch": 0.09760039786149446,
      "grad_norm": 0.42994019389152527,
      "learning_rate": 4.512205230220896e-06,
      "loss": 0.7388,
      "step": 2355
    },
    {
      "epoch": 0.09764184176716814,
      "grad_norm": 0.47742557525634766,
      "learning_rate": 4.5119980106925286e-06,
      "loss": 0.7717,
      "step": 2356
    },
    {
      "epoch": 0.0976832856728418,
      "grad_norm": 0.4480413794517517,
      "learning_rate": 4.511790791164159e-06,
      "loss": 0.8276,
      "step": 2357
    },
    {
      "epoch": 0.09772472957851548,
      "grad_norm": 0.46683523058891296,
      "learning_rate": 4.511583571635791e-06,
      "loss": 0.7778,
      "step": 2358
    },
    {
      "epoch": 0.09776617348418914,
      "grad_norm": 0.4650786221027374,
      "learning_rate": 4.511376352107423e-06,
      "loss": 0.7566,
      "step": 2359
    },
    {
      "epoch": 0.09780761738986282,
      "grad_norm": 0.4565683901309967,
      "learning_rate": 4.511169132579055e-06,
      "loss": 0.8088,
      "step": 2360
    },
    {
      "epoch": 0.09784906129553649,
      "grad_norm": 0.45051896572113037,
      "learning_rate": 4.510961913050686e-06,
      "loss": 0.7339,
      "step": 2361
    },
    {
      "epoch": 0.09789050520121016,
      "grad_norm": 0.47070810198783875,
      "learning_rate": 4.510754693522318e-06,
      "loss": 0.7905,
      "step": 2362
    },
    {
      "epoch": 0.09793194910688384,
      "grad_norm": 0.4410671293735504,
      "learning_rate": 4.51054747399395e-06,
      "loss": 0.7964,
      "step": 2363
    },
    {
      "epoch": 0.0979733930125575,
      "grad_norm": 0.47352078557014465,
      "learning_rate": 4.510340254465581e-06,
      "loss": 0.7673,
      "step": 2364
    },
    {
      "epoch": 0.09801483691823118,
      "grad_norm": 0.47699326276779175,
      "learning_rate": 4.510133034937213e-06,
      "loss": 0.8162,
      "step": 2365
    },
    {
      "epoch": 0.09805628082390484,
      "grad_norm": 0.475406289100647,
      "learning_rate": 4.509925815408844e-06,
      "loss": 0.7203,
      "step": 2366
    },
    {
      "epoch": 0.09809772472957852,
      "grad_norm": 0.46560657024383545,
      "learning_rate": 4.509718595880476e-06,
      "loss": 0.7744,
      "step": 2367
    },
    {
      "epoch": 0.09813916863525218,
      "grad_norm": 0.42518529295921326,
      "learning_rate": 4.509511376352108e-06,
      "loss": 0.791,
      "step": 2368
    },
    {
      "epoch": 0.09818061254092586,
      "grad_norm": 0.4530797302722931,
      "learning_rate": 4.509304156823739e-06,
      "loss": 0.8135,
      "step": 2369
    },
    {
      "epoch": 0.09822205644659952,
      "grad_norm": 0.4168483018875122,
      "learning_rate": 4.509096937295371e-06,
      "loss": 0.6866,
      "step": 2370
    },
    {
      "epoch": 0.0982635003522732,
      "grad_norm": 0.42781397700309753,
      "learning_rate": 4.508889717767003e-06,
      "loss": 0.7295,
      "step": 2371
    },
    {
      "epoch": 0.09830494425794686,
      "grad_norm": 0.4907296597957611,
      "learning_rate": 4.508682498238635e-06,
      "loss": 0.8088,
      "step": 2372
    },
    {
      "epoch": 0.09834638816362054,
      "grad_norm": 0.4282696545124054,
      "learning_rate": 4.5084752787102655e-06,
      "loss": 0.6992,
      "step": 2373
    },
    {
      "epoch": 0.0983878320692942,
      "grad_norm": 0.4264950752258301,
      "learning_rate": 4.508268059181898e-06,
      "loss": 0.801,
      "step": 2374
    },
    {
      "epoch": 0.09842927597496788,
      "grad_norm": 0.45369046926498413,
      "learning_rate": 4.508060839653529e-06,
      "loss": 0.7729,
      "step": 2375
    },
    {
      "epoch": 0.09847071988064156,
      "grad_norm": 0.42282843589782715,
      "learning_rate": 4.507853620125161e-06,
      "loss": 0.7598,
      "step": 2376
    },
    {
      "epoch": 0.09851216378631522,
      "grad_norm": 0.41421714425086975,
      "learning_rate": 4.507646400596793e-06,
      "loss": 0.7455,
      "step": 2377
    },
    {
      "epoch": 0.0985536076919889,
      "grad_norm": 0.425515353679657,
      "learning_rate": 4.507439181068424e-06,
      "loss": 0.7725,
      "step": 2378
    },
    {
      "epoch": 0.09859505159766256,
      "grad_norm": 0.4112476110458374,
      "learning_rate": 4.507231961540056e-06,
      "loss": 0.7505,
      "step": 2379
    },
    {
      "epoch": 0.09863649550333624,
      "grad_norm": 0.43414306640625,
      "learning_rate": 4.507024742011688e-06,
      "loss": 0.7224,
      "step": 2380
    },
    {
      "epoch": 0.0986779394090099,
      "grad_norm": 0.4688867926597595,
      "learning_rate": 4.506817522483319e-06,
      "loss": 0.7466,
      "step": 2381
    },
    {
      "epoch": 0.09871938331468358,
      "grad_norm": 0.4288109838962555,
      "learning_rate": 4.5066103029549505e-06,
      "loss": 0.7375,
      "step": 2382
    },
    {
      "epoch": 0.09876082722035724,
      "grad_norm": 0.44669681787490845,
      "learning_rate": 4.506403083426583e-06,
      "loss": 0.7356,
      "step": 2383
    },
    {
      "epoch": 0.09880227112603092,
      "grad_norm": 0.43237873911857605,
      "learning_rate": 4.506195863898214e-06,
      "loss": 0.75,
      "step": 2384
    },
    {
      "epoch": 0.09884371503170458,
      "grad_norm": 0.4640229046344757,
      "learning_rate": 4.5059886443698455e-06,
      "loss": 0.707,
      "step": 2385
    },
    {
      "epoch": 0.09888515893737826,
      "grad_norm": 0.4520900547504425,
      "learning_rate": 4.505781424841477e-06,
      "loss": 0.7847,
      "step": 2386
    },
    {
      "epoch": 0.09892660284305192,
      "grad_norm": 0.45799869298934937,
      "learning_rate": 4.505574205313109e-06,
      "loss": 0.8435,
      "step": 2387
    },
    {
      "epoch": 0.0989680467487256,
      "grad_norm": 0.4351831078529358,
      "learning_rate": 4.505366985784741e-06,
      "loss": 0.8125,
      "step": 2388
    },
    {
      "epoch": 0.09900949065439928,
      "grad_norm": 0.40404462814331055,
      "learning_rate": 4.505159766256372e-06,
      "loss": 0.7172,
      "step": 2389
    },
    {
      "epoch": 0.09905093456007294,
      "grad_norm": 0.4919203817844391,
      "learning_rate": 4.504952546728004e-06,
      "loss": 0.7803,
      "step": 2390
    },
    {
      "epoch": 0.09909237846574662,
      "grad_norm": 0.45305073261260986,
      "learning_rate": 4.5047453271996355e-06,
      "loss": 0.7959,
      "step": 2391
    },
    {
      "epoch": 0.09913382237142028,
      "grad_norm": 0.48433443903923035,
      "learning_rate": 4.504538107671268e-06,
      "loss": 0.7947,
      "step": 2392
    },
    {
      "epoch": 0.09917526627709396,
      "grad_norm": 0.44037485122680664,
      "learning_rate": 4.504330888142899e-06,
      "loss": 0.7468,
      "step": 2393
    },
    {
      "epoch": 0.09921671018276762,
      "grad_norm": 0.4270979166030884,
      "learning_rate": 4.5041236686145305e-06,
      "loss": 0.74,
      "step": 2394
    },
    {
      "epoch": 0.0992581540884413,
      "grad_norm": 0.4232253134250641,
      "learning_rate": 4.503916449086162e-06,
      "loss": 0.7063,
      "step": 2395
    },
    {
      "epoch": 0.09929959799411496,
      "grad_norm": 0.42492538690567017,
      "learning_rate": 4.503709229557794e-06,
      "loss": 0.7073,
      "step": 2396
    },
    {
      "epoch": 0.09934104189978864,
      "grad_norm": 0.4379681944847107,
      "learning_rate": 4.5035020100294255e-06,
      "loss": 0.7483,
      "step": 2397
    },
    {
      "epoch": 0.0993824858054623,
      "grad_norm": 0.42816707491874695,
      "learning_rate": 4.503294790501057e-06,
      "loss": 0.7349,
      "step": 2398
    },
    {
      "epoch": 0.09942392971113598,
      "grad_norm": 0.44663336873054504,
      "learning_rate": 4.503087570972689e-06,
      "loss": 0.7542,
      "step": 2399
    },
    {
      "epoch": 0.09946537361680965,
      "grad_norm": 0.4171334505081177,
      "learning_rate": 4.5028803514443205e-06,
      "loss": 0.7378,
      "step": 2400
    },
    {
      "epoch": 0.09950681752248332,
      "grad_norm": 0.4079301655292511,
      "learning_rate": 4.502673131915952e-06,
      "loss": 0.6808,
      "step": 2401
    },
    {
      "epoch": 0.099548261428157,
      "grad_norm": 0.4644434154033661,
      "learning_rate": 4.502465912387583e-06,
      "loss": 0.7812,
      "step": 2402
    },
    {
      "epoch": 0.09958970533383066,
      "grad_norm": 0.4629955589771271,
      "learning_rate": 4.5022586928592156e-06,
      "loss": 0.7646,
      "step": 2403
    },
    {
      "epoch": 0.09963114923950434,
      "grad_norm": 0.4141670763492584,
      "learning_rate": 4.502051473330847e-06,
      "loss": 0.7648,
      "step": 2404
    },
    {
      "epoch": 0.099672593145178,
      "grad_norm": 0.41274768114089966,
      "learning_rate": 4.501844253802478e-06,
      "loss": 0.7202,
      "step": 2405
    },
    {
      "epoch": 0.09971403705085168,
      "grad_norm": 0.4692912697792053,
      "learning_rate": 4.5016370342741106e-06,
      "loss": 0.752,
      "step": 2406
    },
    {
      "epoch": 0.09975548095652534,
      "grad_norm": 0.46297743916511536,
      "learning_rate": 4.501429814745742e-06,
      "loss": 0.7844,
      "step": 2407
    },
    {
      "epoch": 0.09979692486219902,
      "grad_norm": 0.4384287893772125,
      "learning_rate": 4.501222595217374e-06,
      "loss": 0.6854,
      "step": 2408
    },
    {
      "epoch": 0.09983836876787268,
      "grad_norm": 0.4322988986968994,
      "learning_rate": 4.5010153756890056e-06,
      "loss": 0.7126,
      "step": 2409
    },
    {
      "epoch": 0.09987981267354636,
      "grad_norm": 0.4194031059741974,
      "learning_rate": 4.500808156160637e-06,
      "loss": 0.7532,
      "step": 2410
    },
    {
      "epoch": 0.09992125657922002,
      "grad_norm": 0.4283793270587921,
      "learning_rate": 4.500600936632268e-06,
      "loss": 0.7795,
      "step": 2411
    },
    {
      "epoch": 0.0999627004848937,
      "grad_norm": 0.42768609523773193,
      "learning_rate": 4.5003937171039006e-06,
      "loss": 0.7563,
      "step": 2412
    },
    {
      "epoch": 0.10000414439056737,
      "grad_norm": 0.42699337005615234,
      "learning_rate": 4.500186497575532e-06,
      "loss": 0.717,
      "step": 2413
    },
    {
      "epoch": 0.10004558829624104,
      "grad_norm": 0.4601013660430908,
      "learning_rate": 4.499979278047163e-06,
      "loss": 0.8013,
      "step": 2414
    },
    {
      "epoch": 0.10008703220191471,
      "grad_norm": 0.4400237798690796,
      "learning_rate": 4.4997720585187956e-06,
      "loss": 0.7749,
      "step": 2415
    },
    {
      "epoch": 0.10012847610758838,
      "grad_norm": 0.4356423318386078,
      "learning_rate": 4.499564838990427e-06,
      "loss": 0.7533,
      "step": 2416
    },
    {
      "epoch": 0.10016992001326205,
      "grad_norm": 0.41153624653816223,
      "learning_rate": 4.499357619462058e-06,
      "loss": 0.6985,
      "step": 2417
    },
    {
      "epoch": 0.10021136391893572,
      "grad_norm": 0.4521646201610565,
      "learning_rate": 4.49915039993369e-06,
      "loss": 0.7161,
      "step": 2418
    },
    {
      "epoch": 0.1002528078246094,
      "grad_norm": 0.4561423659324646,
      "learning_rate": 4.498943180405322e-06,
      "loss": 0.7825,
      "step": 2419
    },
    {
      "epoch": 0.10029425173028306,
      "grad_norm": 0.4558728039264679,
      "learning_rate": 4.498735960876953e-06,
      "loss": 0.7307,
      "step": 2420
    },
    {
      "epoch": 0.10033569563595673,
      "grad_norm": 0.44516822695732117,
      "learning_rate": 4.498528741348585e-06,
      "loss": 0.802,
      "step": 2421
    },
    {
      "epoch": 0.1003771395416304,
      "grad_norm": 0.45842206478118896,
      "learning_rate": 4.498321521820217e-06,
      "loss": 0.7593,
      "step": 2422
    },
    {
      "epoch": 0.10041858344730407,
      "grad_norm": 0.42760998010635376,
      "learning_rate": 4.498114302291848e-06,
      "loss": 0.7222,
      "step": 2423
    },
    {
      "epoch": 0.10046002735297775,
      "grad_norm": 0.43931469321250916,
      "learning_rate": 4.497907082763481e-06,
      "loss": 0.7515,
      "step": 2424
    },
    {
      "epoch": 0.10050147125865141,
      "grad_norm": 0.4742116928100586,
      "learning_rate": 4.497699863235111e-06,
      "loss": 0.7638,
      "step": 2425
    },
    {
      "epoch": 0.10054291516432509,
      "grad_norm": 0.43348821997642517,
      "learning_rate": 4.497492643706743e-06,
      "loss": 0.7791,
      "step": 2426
    },
    {
      "epoch": 0.10058435906999875,
      "grad_norm": 0.5565977692604065,
      "learning_rate": 4.497285424178375e-06,
      "loss": 0.8474,
      "step": 2427
    },
    {
      "epoch": 0.10062580297567243,
      "grad_norm": 0.45578351616859436,
      "learning_rate": 4.497078204650007e-06,
      "loss": 0.7788,
      "step": 2428
    },
    {
      "epoch": 0.1006672468813461,
      "grad_norm": 0.4438045024871826,
      "learning_rate": 4.496870985121638e-06,
      "loss": 0.7173,
      "step": 2429
    },
    {
      "epoch": 0.10070869078701977,
      "grad_norm": 0.4236430525779724,
      "learning_rate": 4.49666376559327e-06,
      "loss": 0.749,
      "step": 2430
    },
    {
      "epoch": 0.10075013469269344,
      "grad_norm": 0.455037385225296,
      "learning_rate": 4.496456546064902e-06,
      "loss": 0.7686,
      "step": 2431
    },
    {
      "epoch": 0.10079157859836711,
      "grad_norm": 0.42024797201156616,
      "learning_rate": 4.496249326536533e-06,
      "loss": 0.6854,
      "step": 2432
    },
    {
      "epoch": 0.10083302250404078,
      "grad_norm": 0.42516160011291504,
      "learning_rate": 4.496042107008165e-06,
      "loss": 0.7798,
      "step": 2433
    },
    {
      "epoch": 0.10087446640971445,
      "grad_norm": 0.43512436747550964,
      "learning_rate": 4.495834887479796e-06,
      "loss": 0.7808,
      "step": 2434
    },
    {
      "epoch": 0.10091591031538812,
      "grad_norm": 0.43528231978416443,
      "learning_rate": 4.495627667951428e-06,
      "loss": 0.708,
      "step": 2435
    },
    {
      "epoch": 0.10095735422106179,
      "grad_norm": 0.4744385778903961,
      "learning_rate": 4.49542044842306e-06,
      "loss": 0.761,
      "step": 2436
    },
    {
      "epoch": 0.10099879812673547,
      "grad_norm": 0.430096298456192,
      "learning_rate": 4.495213228894691e-06,
      "loss": 0.7249,
      "step": 2437
    },
    {
      "epoch": 0.10104024203240913,
      "grad_norm": 0.4591529071331024,
      "learning_rate": 4.4950060093663225e-06,
      "loss": 0.7539,
      "step": 2438
    },
    {
      "epoch": 0.10108168593808281,
      "grad_norm": 0.431298166513443,
      "learning_rate": 4.494798789837955e-06,
      "loss": 0.7463,
      "step": 2439
    },
    {
      "epoch": 0.10112312984375647,
      "grad_norm": 0.4818033277988434,
      "learning_rate": 4.494591570309587e-06,
      "loss": 0.7974,
      "step": 2440
    },
    {
      "epoch": 0.10116457374943015,
      "grad_norm": 0.41909778118133545,
      "learning_rate": 4.4943843507812175e-06,
      "loss": 0.7341,
      "step": 2441
    },
    {
      "epoch": 0.10120601765510381,
      "grad_norm": 0.44373542070388794,
      "learning_rate": 4.49417713125285e-06,
      "loss": 0.8207,
      "step": 2442
    },
    {
      "epoch": 0.10124746156077749,
      "grad_norm": 0.4719321131706238,
      "learning_rate": 4.493969911724481e-06,
      "loss": 0.8121,
      "step": 2443
    },
    {
      "epoch": 0.10128890546645115,
      "grad_norm": 0.4662165939807892,
      "learning_rate": 4.493762692196113e-06,
      "loss": 0.7581,
      "step": 2444
    },
    {
      "epoch": 0.10133034937212483,
      "grad_norm": 0.44250357151031494,
      "learning_rate": 4.493555472667745e-06,
      "loss": 0.749,
      "step": 2445
    },
    {
      "epoch": 0.1013717932777985,
      "grad_norm": 0.4132310450077057,
      "learning_rate": 4.493348253139376e-06,
      "loss": 0.7676,
      "step": 2446
    },
    {
      "epoch": 0.10141323718347217,
      "grad_norm": 0.46850740909576416,
      "learning_rate": 4.4931410336110075e-06,
      "loss": 0.8042,
      "step": 2447
    },
    {
      "epoch": 0.10145468108914583,
      "grad_norm": 0.44116607308387756,
      "learning_rate": 4.49293381408264e-06,
      "loss": 0.7405,
      "step": 2448
    },
    {
      "epoch": 0.10149612499481951,
      "grad_norm": 0.46291738748550415,
      "learning_rate": 4.492726594554271e-06,
      "loss": 0.7917,
      "step": 2449
    },
    {
      "epoch": 0.10153756890049319,
      "grad_norm": 0.4250962436199188,
      "learning_rate": 4.4925193750259025e-06,
      "loss": 0.7881,
      "step": 2450
    },
    {
      "epoch": 0.10157901280616685,
      "grad_norm": 0.434390127658844,
      "learning_rate": 4.492312155497535e-06,
      "loss": 0.7727,
      "step": 2451
    },
    {
      "epoch": 0.10162045671184053,
      "grad_norm": 0.4137495458126068,
      "learning_rate": 4.492104935969166e-06,
      "loss": 0.7195,
      "step": 2452
    },
    {
      "epoch": 0.10166190061751419,
      "grad_norm": 0.42863282561302185,
      "learning_rate": 4.4918977164407975e-06,
      "loss": 0.7402,
      "step": 2453
    },
    {
      "epoch": 0.10170334452318787,
      "grad_norm": 0.4046284556388855,
      "learning_rate": 4.491690496912429e-06,
      "loss": 0.7312,
      "step": 2454
    },
    {
      "epoch": 0.10174478842886153,
      "grad_norm": 0.47594335675239563,
      "learning_rate": 4.491483277384061e-06,
      "loss": 0.8127,
      "step": 2455
    },
    {
      "epoch": 0.10178623233453521,
      "grad_norm": 0.43427473306655884,
      "learning_rate": 4.4912760578556925e-06,
      "loss": 0.7078,
      "step": 2456
    },
    {
      "epoch": 0.10182767624020887,
      "grad_norm": 0.4573884606361389,
      "learning_rate": 4.491068838327324e-06,
      "loss": 0.7356,
      "step": 2457
    },
    {
      "epoch": 0.10186912014588255,
      "grad_norm": 0.5104098916053772,
      "learning_rate": 4.490861618798956e-06,
      "loss": 0.876,
      "step": 2458
    },
    {
      "epoch": 0.10191056405155621,
      "grad_norm": 0.43663156032562256,
      "learning_rate": 4.4906543992705876e-06,
      "loss": 0.722,
      "step": 2459
    },
    {
      "epoch": 0.10195200795722989,
      "grad_norm": 0.47033756971359253,
      "learning_rate": 4.49044717974222e-06,
      "loss": 0.7706,
      "step": 2460
    },
    {
      "epoch": 0.10199345186290357,
      "grad_norm": 0.4738774299621582,
      "learning_rate": 4.490239960213851e-06,
      "loss": 0.7686,
      "step": 2461
    },
    {
      "epoch": 0.10203489576857723,
      "grad_norm": 0.454360693693161,
      "learning_rate": 4.4900327406854826e-06,
      "loss": 0.7281,
      "step": 2462
    },
    {
      "epoch": 0.10207633967425091,
      "grad_norm": 0.38072669506073,
      "learning_rate": 4.489825521157114e-06,
      "loss": 0.6763,
      "step": 2463
    },
    {
      "epoch": 0.10211778357992457,
      "grad_norm": 0.4466675817966461,
      "learning_rate": 4.489618301628746e-06,
      "loss": 0.7584,
      "step": 2464
    },
    {
      "epoch": 0.10215922748559825,
      "grad_norm": 0.40937522053718567,
      "learning_rate": 4.4894110821003776e-06,
      "loss": 0.7222,
      "step": 2465
    },
    {
      "epoch": 0.10220067139127191,
      "grad_norm": 0.4338492751121521,
      "learning_rate": 4.489203862572009e-06,
      "loss": 0.7441,
      "step": 2466
    },
    {
      "epoch": 0.10224211529694559,
      "grad_norm": 0.414019376039505,
      "learning_rate": 4.488996643043641e-06,
      "loss": 0.7166,
      "step": 2467
    },
    {
      "epoch": 0.10228355920261925,
      "grad_norm": 0.4475066065788269,
      "learning_rate": 4.4887894235152726e-06,
      "loss": 0.7474,
      "step": 2468
    },
    {
      "epoch": 0.10232500310829293,
      "grad_norm": 0.4825494885444641,
      "learning_rate": 4.488582203986904e-06,
      "loss": 0.8452,
      "step": 2469
    },
    {
      "epoch": 0.10236644701396659,
      "grad_norm": 0.4023198187351227,
      "learning_rate": 4.488374984458535e-06,
      "loss": 0.7152,
      "step": 2470
    },
    {
      "epoch": 0.10240789091964027,
      "grad_norm": 0.452576607465744,
      "learning_rate": 4.4881677649301676e-06,
      "loss": 0.7773,
      "step": 2471
    },
    {
      "epoch": 0.10244933482531393,
      "grad_norm": 0.45021331310272217,
      "learning_rate": 4.487960545401799e-06,
      "loss": 0.7952,
      "step": 2472
    },
    {
      "epoch": 0.10249077873098761,
      "grad_norm": 0.4413323998451233,
      "learning_rate": 4.48775332587343e-06,
      "loss": 0.7383,
      "step": 2473
    },
    {
      "epoch": 0.10253222263666129,
      "grad_norm": 0.5122082829475403,
      "learning_rate": 4.4875461063450626e-06,
      "loss": 0.783,
      "step": 2474
    },
    {
      "epoch": 0.10257366654233495,
      "grad_norm": 0.4322240650653839,
      "learning_rate": 4.487338886816694e-06,
      "loss": 0.6738,
      "step": 2475
    },
    {
      "epoch": 0.10261511044800863,
      "grad_norm": 0.44828489422798157,
      "learning_rate": 4.487131667288326e-06,
      "loss": 0.7644,
      "step": 2476
    },
    {
      "epoch": 0.10265655435368229,
      "grad_norm": 0.4050629138946533,
      "learning_rate": 4.4869244477599576e-06,
      "loss": 0.7172,
      "step": 2477
    },
    {
      "epoch": 0.10269799825935597,
      "grad_norm": 0.4359167814254761,
      "learning_rate": 4.486717228231589e-06,
      "loss": 0.7549,
      "step": 2478
    },
    {
      "epoch": 0.10273944216502963,
      "grad_norm": 0.43352094292640686,
      "learning_rate": 4.48651000870322e-06,
      "loss": 0.739,
      "step": 2479
    },
    {
      "epoch": 0.1027808860707033,
      "grad_norm": 0.45650553703308105,
      "learning_rate": 4.486302789174853e-06,
      "loss": 0.7534,
      "step": 2480
    },
    {
      "epoch": 0.10282232997637697,
      "grad_norm": 0.4786919951438904,
      "learning_rate": 4.486095569646484e-06,
      "loss": 0.7859,
      "step": 2481
    },
    {
      "epoch": 0.10286377388205065,
      "grad_norm": 0.409066766500473,
      "learning_rate": 4.485888350118115e-06,
      "loss": 0.7078,
      "step": 2482
    },
    {
      "epoch": 0.10290521778772431,
      "grad_norm": 0.49375882744789124,
      "learning_rate": 4.485681130589748e-06,
      "loss": 0.7885,
      "step": 2483
    },
    {
      "epoch": 0.10294666169339799,
      "grad_norm": 0.4763132929801941,
      "learning_rate": 4.485473911061379e-06,
      "loss": 0.7661,
      "step": 2484
    },
    {
      "epoch": 0.10298810559907165,
      "grad_norm": 0.48109549283981323,
      "learning_rate": 4.48526669153301e-06,
      "loss": 0.751,
      "step": 2485
    },
    {
      "epoch": 0.10302954950474533,
      "grad_norm": 0.45813482999801636,
      "learning_rate": 4.485059472004642e-06,
      "loss": 0.7551,
      "step": 2486
    },
    {
      "epoch": 0.103070993410419,
      "grad_norm": 0.44686657190322876,
      "learning_rate": 4.484852252476274e-06,
      "loss": 0.7463,
      "step": 2487
    },
    {
      "epoch": 0.10311243731609267,
      "grad_norm": 0.48090413212776184,
      "learning_rate": 4.484645032947905e-06,
      "loss": 0.802,
      "step": 2488
    },
    {
      "epoch": 0.10315388122176634,
      "grad_norm": 0.4238318204879761,
      "learning_rate": 4.484437813419537e-06,
      "loss": 0.7263,
      "step": 2489
    },
    {
      "epoch": 0.10319532512744001,
      "grad_norm": 0.42219048738479614,
      "learning_rate": 4.484230593891168e-06,
      "loss": 0.771,
      "step": 2490
    },
    {
      "epoch": 0.10323676903311368,
      "grad_norm": 0.43518730998039246,
      "learning_rate": 4.4840233743628e-06,
      "loss": 0.7622,
      "step": 2491
    },
    {
      "epoch": 0.10327821293878735,
      "grad_norm": 0.4431527256965637,
      "learning_rate": 4.483816154834433e-06,
      "loss": 0.7168,
      "step": 2492
    },
    {
      "epoch": 0.10331965684446102,
      "grad_norm": 0.439017653465271,
      "learning_rate": 4.483608935306064e-06,
      "loss": 0.7205,
      "step": 2493
    },
    {
      "epoch": 0.10336110075013469,
      "grad_norm": 0.430141806602478,
      "learning_rate": 4.483401715777695e-06,
      "loss": 0.6853,
      "step": 2494
    },
    {
      "epoch": 0.10340254465580836,
      "grad_norm": 0.49170151352882385,
      "learning_rate": 4.483194496249327e-06,
      "loss": 0.7952,
      "step": 2495
    },
    {
      "epoch": 0.10344398856148203,
      "grad_norm": 0.43707892298698425,
      "learning_rate": 4.482987276720959e-06,
      "loss": 0.7352,
      "step": 2496
    },
    {
      "epoch": 0.1034854324671557,
      "grad_norm": 0.4098874032497406,
      "learning_rate": 4.48278005719259e-06,
      "loss": 0.6731,
      "step": 2497
    },
    {
      "epoch": 0.10352687637282938,
      "grad_norm": 0.41782280802726746,
      "learning_rate": 4.482572837664222e-06,
      "loss": 0.7188,
      "step": 2498
    },
    {
      "epoch": 0.10356832027850305,
      "grad_norm": 0.5006802082061768,
      "learning_rate": 4.482365618135853e-06,
      "loss": 0.8198,
      "step": 2499
    },
    {
      "epoch": 0.10360976418417672,
      "grad_norm": 0.44261375069618225,
      "learning_rate": 4.482158398607485e-06,
      "loss": 0.7517,
      "step": 2500
    },
    {
      "epoch": 0.10365120808985039,
      "grad_norm": 0.3972148597240448,
      "learning_rate": 4.481951179079117e-06,
      "loss": 0.7292,
      "step": 2501
    },
    {
      "epoch": 0.10369265199552406,
      "grad_norm": 0.45424169301986694,
      "learning_rate": 4.481743959550748e-06,
      "loss": 0.7461,
      "step": 2502
    },
    {
      "epoch": 0.10373409590119773,
      "grad_norm": 0.420501708984375,
      "learning_rate": 4.48153674002238e-06,
      "loss": 0.7889,
      "step": 2503
    },
    {
      "epoch": 0.1037755398068714,
      "grad_norm": 0.42208561301231384,
      "learning_rate": 4.481329520494012e-06,
      "loss": 0.7844,
      "step": 2504
    },
    {
      "epoch": 0.10381698371254507,
      "grad_norm": 0.46895819902420044,
      "learning_rate": 4.481122300965643e-06,
      "loss": 0.8037,
      "step": 2505
    },
    {
      "epoch": 0.10385842761821874,
      "grad_norm": 0.39926281571388245,
      "learning_rate": 4.4809150814372745e-06,
      "loss": 0.6953,
      "step": 2506
    },
    {
      "epoch": 0.1038998715238924,
      "grad_norm": 0.45672452449798584,
      "learning_rate": 4.480707861908907e-06,
      "loss": 0.7646,
      "step": 2507
    },
    {
      "epoch": 0.10394131542956608,
      "grad_norm": 0.46605178713798523,
      "learning_rate": 4.480500642380538e-06,
      "loss": 0.8022,
      "step": 2508
    },
    {
      "epoch": 0.10398275933523975,
      "grad_norm": 0.4274391233921051,
      "learning_rate": 4.4802934228521695e-06,
      "loss": 0.7285,
      "step": 2509
    },
    {
      "epoch": 0.10402420324091342,
      "grad_norm": 0.4286547303199768,
      "learning_rate": 4.480086203323802e-06,
      "loss": 0.7651,
      "step": 2510
    },
    {
      "epoch": 0.1040656471465871,
      "grad_norm": 0.43266093730926514,
      "learning_rate": 4.479878983795433e-06,
      "loss": 0.752,
      "step": 2511
    },
    {
      "epoch": 0.10410709105226076,
      "grad_norm": 0.4616123139858246,
      "learning_rate": 4.479671764267065e-06,
      "loss": 0.7871,
      "step": 2512
    },
    {
      "epoch": 0.10414853495793444,
      "grad_norm": 0.4442767798900604,
      "learning_rate": 4.479464544738697e-06,
      "loss": 0.7927,
      "step": 2513
    },
    {
      "epoch": 0.1041899788636081,
      "grad_norm": 0.41791102290153503,
      "learning_rate": 4.479257325210328e-06,
      "loss": 0.7212,
      "step": 2514
    },
    {
      "epoch": 0.10423142276928178,
      "grad_norm": 0.44849756360054016,
      "learning_rate": 4.4790501056819595e-06,
      "loss": 0.7461,
      "step": 2515
    },
    {
      "epoch": 0.10427286667495544,
      "grad_norm": 0.4391424059867859,
      "learning_rate": 4.478842886153592e-06,
      "loss": 0.7451,
      "step": 2516
    },
    {
      "epoch": 0.10431431058062912,
      "grad_norm": 0.43719181418418884,
      "learning_rate": 4.478635666625223e-06,
      "loss": 0.7454,
      "step": 2517
    },
    {
      "epoch": 0.10435575448630278,
      "grad_norm": 0.44565802812576294,
      "learning_rate": 4.4784284470968546e-06,
      "loss": 0.7488,
      "step": 2518
    },
    {
      "epoch": 0.10439719839197646,
      "grad_norm": 0.4534648656845093,
      "learning_rate": 4.478221227568487e-06,
      "loss": 0.7,
      "step": 2519
    },
    {
      "epoch": 0.10443864229765012,
      "grad_norm": 0.5323036313056946,
      "learning_rate": 4.478014008040118e-06,
      "loss": 0.8271,
      "step": 2520
    },
    {
      "epoch": 0.1044800862033238,
      "grad_norm": 0.43365198373794556,
      "learning_rate": 4.4778067885117496e-06,
      "loss": 0.8071,
      "step": 2521
    },
    {
      "epoch": 0.10452153010899747,
      "grad_norm": 0.38153836131095886,
      "learning_rate": 4.477599568983381e-06,
      "loss": 0.6445,
      "step": 2522
    },
    {
      "epoch": 0.10456297401467114,
      "grad_norm": 0.44764432311058044,
      "learning_rate": 4.477392349455013e-06,
      "loss": 0.7786,
      "step": 2523
    },
    {
      "epoch": 0.10460441792034482,
      "grad_norm": 0.454918771982193,
      "learning_rate": 4.4771851299266446e-06,
      "loss": 0.7927,
      "step": 2524
    },
    {
      "epoch": 0.10464586182601848,
      "grad_norm": 0.44914036989212036,
      "learning_rate": 4.476977910398276e-06,
      "loss": 0.6968,
      "step": 2525
    },
    {
      "epoch": 0.10468730573169216,
      "grad_norm": 0.46323275566101074,
      "learning_rate": 4.476770690869907e-06,
      "loss": 0.7622,
      "step": 2526
    },
    {
      "epoch": 0.10472874963736582,
      "grad_norm": 0.42381998896598816,
      "learning_rate": 4.4765634713415396e-06,
      "loss": 0.6992,
      "step": 2527
    },
    {
      "epoch": 0.1047701935430395,
      "grad_norm": 0.47327014803886414,
      "learning_rate": 4.476356251813172e-06,
      "loss": 0.7434,
      "step": 2528
    },
    {
      "epoch": 0.10481163744871316,
      "grad_norm": 0.4108796715736389,
      "learning_rate": 4.476149032284803e-06,
      "loss": 0.7495,
      "step": 2529
    },
    {
      "epoch": 0.10485308135438684,
      "grad_norm": 0.45763590931892395,
      "learning_rate": 4.4759418127564346e-06,
      "loss": 0.7761,
      "step": 2530
    },
    {
      "epoch": 0.1048945252600605,
      "grad_norm": 0.43307146430015564,
      "learning_rate": 4.475734593228066e-06,
      "loss": 0.7324,
      "step": 2531
    },
    {
      "epoch": 0.10493596916573418,
      "grad_norm": 0.4290485382080078,
      "learning_rate": 4.475527373699698e-06,
      "loss": 0.7224,
      "step": 2532
    },
    {
      "epoch": 0.10497741307140784,
      "grad_norm": 0.4771115183830261,
      "learning_rate": 4.4753201541713296e-06,
      "loss": 0.7942,
      "step": 2533
    },
    {
      "epoch": 0.10501885697708152,
      "grad_norm": 0.4714314639568329,
      "learning_rate": 4.475112934642961e-06,
      "loss": 0.7698,
      "step": 2534
    },
    {
      "epoch": 0.1050603008827552,
      "grad_norm": 0.4226863980293274,
      "learning_rate": 4.474905715114593e-06,
      "loss": 0.7131,
      "step": 2535
    },
    {
      "epoch": 0.10510174478842886,
      "grad_norm": 0.4509071707725525,
      "learning_rate": 4.4746984955862246e-06,
      "loss": 0.8113,
      "step": 2536
    },
    {
      "epoch": 0.10514318869410254,
      "grad_norm": 0.43067076802253723,
      "learning_rate": 4.474491276057856e-06,
      "loss": 0.7808,
      "step": 2537
    },
    {
      "epoch": 0.1051846325997762,
      "grad_norm": 0.46068039536476135,
      "learning_rate": 4.474284056529487e-06,
      "loss": 0.8279,
      "step": 2538
    },
    {
      "epoch": 0.10522607650544988,
      "grad_norm": 0.43353769183158875,
      "learning_rate": 4.47407683700112e-06,
      "loss": 0.7732,
      "step": 2539
    },
    {
      "epoch": 0.10526752041112354,
      "grad_norm": 0.46565356850624084,
      "learning_rate": 4.473869617472751e-06,
      "loss": 0.7446,
      "step": 2540
    },
    {
      "epoch": 0.10530896431679722,
      "grad_norm": 0.46216118335723877,
      "learning_rate": 4.473662397944382e-06,
      "loss": 0.8176,
      "step": 2541
    },
    {
      "epoch": 0.10535040822247088,
      "grad_norm": 0.44926711916923523,
      "learning_rate": 4.473455178416014e-06,
      "loss": 0.7461,
      "step": 2542
    },
    {
      "epoch": 0.10539185212814456,
      "grad_norm": 0.4275146424770355,
      "learning_rate": 4.473247958887646e-06,
      "loss": 0.6875,
      "step": 2543
    },
    {
      "epoch": 0.10543329603381822,
      "grad_norm": 0.4907032549381256,
      "learning_rate": 4.473040739359278e-06,
      "loss": 0.7419,
      "step": 2544
    },
    {
      "epoch": 0.1054747399394919,
      "grad_norm": 0.4054960012435913,
      "learning_rate": 4.47283351983091e-06,
      "loss": 0.7429,
      "step": 2545
    },
    {
      "epoch": 0.10551618384516556,
      "grad_norm": 0.42188334465026855,
      "learning_rate": 4.472626300302541e-06,
      "loss": 0.7524,
      "step": 2546
    },
    {
      "epoch": 0.10555762775083924,
      "grad_norm": 0.40305376052856445,
      "learning_rate": 4.472419080774172e-06,
      "loss": 0.7075,
      "step": 2547
    },
    {
      "epoch": 0.10559907165651292,
      "grad_norm": 0.4404488801956177,
      "learning_rate": 4.472211861245805e-06,
      "loss": 0.7319,
      "step": 2548
    },
    {
      "epoch": 0.10564051556218658,
      "grad_norm": 0.45241448283195496,
      "learning_rate": 4.472004641717436e-06,
      "loss": 0.7332,
      "step": 2549
    },
    {
      "epoch": 0.10568195946786026,
      "grad_norm": 0.43186691403388977,
      "learning_rate": 4.471797422189067e-06,
      "loss": 0.7551,
      "step": 2550
    },
    {
      "epoch": 0.10572340337353392,
      "grad_norm": 0.45685335993766785,
      "learning_rate": 4.471590202660699e-06,
      "loss": 0.782,
      "step": 2551
    },
    {
      "epoch": 0.1057648472792076,
      "grad_norm": 0.4209917485713959,
      "learning_rate": 4.471382983132331e-06,
      "loss": 0.6509,
      "step": 2552
    },
    {
      "epoch": 0.10580629118488126,
      "grad_norm": 0.4786608815193176,
      "learning_rate": 4.471175763603962e-06,
      "loss": 0.7712,
      "step": 2553
    },
    {
      "epoch": 0.10584773509055494,
      "grad_norm": 0.44296300411224365,
      "learning_rate": 4.470968544075594e-06,
      "loss": 0.7753,
      "step": 2554
    },
    {
      "epoch": 0.1058891789962286,
      "grad_norm": 0.4304429292678833,
      "learning_rate": 4.470761324547226e-06,
      "loss": 0.7332,
      "step": 2555
    },
    {
      "epoch": 0.10593062290190228,
      "grad_norm": 0.47287848591804504,
      "learning_rate": 4.470554105018857e-06,
      "loss": 0.8389,
      "step": 2556
    },
    {
      "epoch": 0.10597206680757594,
      "grad_norm": 0.4448120594024658,
      "learning_rate": 4.470346885490489e-06,
      "loss": 0.7271,
      "step": 2557
    },
    {
      "epoch": 0.10601351071324962,
      "grad_norm": 0.41209185123443604,
      "learning_rate": 4.47013966596212e-06,
      "loss": 0.7312,
      "step": 2558
    },
    {
      "epoch": 0.10605495461892328,
      "grad_norm": 0.42535993456840515,
      "learning_rate": 4.469932446433752e-06,
      "loss": 0.79,
      "step": 2559
    },
    {
      "epoch": 0.10609639852459696,
      "grad_norm": 0.41931992769241333,
      "learning_rate": 4.469725226905384e-06,
      "loss": 0.7361,
      "step": 2560
    },
    {
      "epoch": 0.10613784243027063,
      "grad_norm": 0.4432069659233093,
      "learning_rate": 4.469518007377016e-06,
      "loss": 0.7776,
      "step": 2561
    },
    {
      "epoch": 0.1061792863359443,
      "grad_norm": 0.41144877672195435,
      "learning_rate": 4.469310787848647e-06,
      "loss": 0.7363,
      "step": 2562
    },
    {
      "epoch": 0.10622073024161797,
      "grad_norm": 0.3849984407424927,
      "learning_rate": 4.469103568320279e-06,
      "loss": 0.6924,
      "step": 2563
    },
    {
      "epoch": 0.10626217414729164,
      "grad_norm": 0.4466310143470764,
      "learning_rate": 4.468896348791911e-06,
      "loss": 0.7712,
      "step": 2564
    },
    {
      "epoch": 0.10630361805296532,
      "grad_norm": 0.4497566223144531,
      "learning_rate": 4.468689129263542e-06,
      "loss": 0.8071,
      "step": 2565
    },
    {
      "epoch": 0.10634506195863898,
      "grad_norm": 0.4391622245311737,
      "learning_rate": 4.468481909735174e-06,
      "loss": 0.7344,
      "step": 2566
    },
    {
      "epoch": 0.10638650586431266,
      "grad_norm": 0.4689805507659912,
      "learning_rate": 4.468274690206805e-06,
      "loss": 0.7126,
      "step": 2567
    },
    {
      "epoch": 0.10642794976998632,
      "grad_norm": 0.42441266775131226,
      "learning_rate": 4.468067470678437e-06,
      "loss": 0.7561,
      "step": 2568
    },
    {
      "epoch": 0.10646939367566,
      "grad_norm": 0.4246194064617157,
      "learning_rate": 4.467860251150069e-06,
      "loss": 0.7844,
      "step": 2569
    },
    {
      "epoch": 0.10651083758133366,
      "grad_norm": 0.4340292811393738,
      "learning_rate": 4.4676530316217e-06,
      "loss": 0.7267,
      "step": 2570
    },
    {
      "epoch": 0.10655228148700734,
      "grad_norm": 0.4364875853061676,
      "learning_rate": 4.467445812093332e-06,
      "loss": 0.8528,
      "step": 2571
    },
    {
      "epoch": 0.10659372539268101,
      "grad_norm": 0.4468110203742981,
      "learning_rate": 4.467238592564964e-06,
      "loss": 0.7371,
      "step": 2572
    },
    {
      "epoch": 0.10663516929835468,
      "grad_norm": 0.42062655091285706,
      "learning_rate": 4.467031373036595e-06,
      "loss": 0.7671,
      "step": 2573
    },
    {
      "epoch": 0.10667661320402835,
      "grad_norm": 0.42927059531211853,
      "learning_rate": 4.4668241535082265e-06,
      "loss": 0.7045,
      "step": 2574
    },
    {
      "epoch": 0.10671805710970202,
      "grad_norm": 0.4020428955554962,
      "learning_rate": 4.466616933979859e-06,
      "loss": 0.7146,
      "step": 2575
    },
    {
      "epoch": 0.1067595010153757,
      "grad_norm": 0.4750811457633972,
      "learning_rate": 4.46640971445149e-06,
      "loss": 0.7996,
      "step": 2576
    },
    {
      "epoch": 0.10680094492104936,
      "grad_norm": 0.40622594952583313,
      "learning_rate": 4.4662024949231216e-06,
      "loss": 0.7219,
      "step": 2577
    },
    {
      "epoch": 0.10684238882672303,
      "grad_norm": 0.4146714210510254,
      "learning_rate": 4.465995275394753e-06,
      "loss": 0.7397,
      "step": 2578
    },
    {
      "epoch": 0.1068838327323967,
      "grad_norm": 0.42968955636024475,
      "learning_rate": 4.465788055866385e-06,
      "loss": 0.6786,
      "step": 2579
    },
    {
      "epoch": 0.10692527663807037,
      "grad_norm": 0.4560728073120117,
      "learning_rate": 4.465580836338017e-06,
      "loss": 0.7311,
      "step": 2580
    },
    {
      "epoch": 0.10696672054374404,
      "grad_norm": 0.44176554679870605,
      "learning_rate": 4.465373616809649e-06,
      "loss": 0.7356,
      "step": 2581
    },
    {
      "epoch": 0.10700816444941771,
      "grad_norm": 0.43821981549263,
      "learning_rate": 4.46516639728128e-06,
      "loss": 0.7476,
      "step": 2582
    },
    {
      "epoch": 0.10704960835509138,
      "grad_norm": 0.4196327328681946,
      "learning_rate": 4.4649591777529116e-06,
      "loss": 0.7139,
      "step": 2583
    },
    {
      "epoch": 0.10709105226076505,
      "grad_norm": 0.4328017830848694,
      "learning_rate": 4.464751958224544e-06,
      "loss": 0.7554,
      "step": 2584
    },
    {
      "epoch": 0.10713249616643873,
      "grad_norm": 0.48729100823402405,
      "learning_rate": 4.464544738696175e-06,
      "loss": 0.8037,
      "step": 2585
    },
    {
      "epoch": 0.1071739400721124,
      "grad_norm": 0.45932716131210327,
      "learning_rate": 4.4643375191678066e-06,
      "loss": 0.7622,
      "step": 2586
    },
    {
      "epoch": 0.10721538397778607,
      "grad_norm": 0.41230276226997375,
      "learning_rate": 4.464130299639438e-06,
      "loss": 0.7251,
      "step": 2587
    },
    {
      "epoch": 0.10725682788345973,
      "grad_norm": 0.409277081489563,
      "learning_rate": 4.46392308011107e-06,
      "loss": 0.7498,
      "step": 2588
    },
    {
      "epoch": 0.10729827178913341,
      "grad_norm": 0.43532902002334595,
      "learning_rate": 4.4637158605827016e-06,
      "loss": 0.74,
      "step": 2589
    },
    {
      "epoch": 0.10733971569480708,
      "grad_norm": 0.4510020613670349,
      "learning_rate": 4.463508641054333e-06,
      "loss": 0.734,
      "step": 2590
    },
    {
      "epoch": 0.10738115960048075,
      "grad_norm": 0.45036837458610535,
      "learning_rate": 4.463301421525965e-06,
      "loss": 0.6987,
      "step": 2591
    },
    {
      "epoch": 0.10742260350615442,
      "grad_norm": 0.41393253207206726,
      "learning_rate": 4.4630942019975966e-06,
      "loss": 0.752,
      "step": 2592
    },
    {
      "epoch": 0.10746404741182809,
      "grad_norm": 0.467176228761673,
      "learning_rate": 4.462886982469228e-06,
      "loss": 0.8203,
      "step": 2593
    },
    {
      "epoch": 0.10750549131750176,
      "grad_norm": 0.4415111541748047,
      "learning_rate": 4.462679762940859e-06,
      "loss": 0.7368,
      "step": 2594
    },
    {
      "epoch": 0.10754693522317543,
      "grad_norm": 0.43596959114074707,
      "learning_rate": 4.4624725434124916e-06,
      "loss": 0.75,
      "step": 2595
    },
    {
      "epoch": 0.1075883791288491,
      "grad_norm": 0.42121055722236633,
      "learning_rate": 4.462265323884124e-06,
      "loss": 0.7039,
      "step": 2596
    },
    {
      "epoch": 0.10762982303452277,
      "grad_norm": 0.4978175163269043,
      "learning_rate": 4.462058104355755e-06,
      "loss": 0.8113,
      "step": 2597
    },
    {
      "epoch": 0.10767126694019645,
      "grad_norm": 0.44776424765586853,
      "learning_rate": 4.461850884827387e-06,
      "loss": 0.731,
      "step": 2598
    },
    {
      "epoch": 0.10771271084587011,
      "grad_norm": 0.46529704332351685,
      "learning_rate": 4.461643665299018e-06,
      "loss": 0.7786,
      "step": 2599
    },
    {
      "epoch": 0.10775415475154379,
      "grad_norm": 0.4415530264377594,
      "learning_rate": 4.46143644577065e-06,
      "loss": 0.7673,
      "step": 2600
    },
    {
      "epoch": 0.10779559865721745,
      "grad_norm": 0.41878318786621094,
      "learning_rate": 4.461229226242282e-06,
      "loss": 0.7866,
      "step": 2601
    },
    {
      "epoch": 0.10783704256289113,
      "grad_norm": 0.4278019964694977,
      "learning_rate": 4.461022006713913e-06,
      "loss": 0.7476,
      "step": 2602
    },
    {
      "epoch": 0.1078784864685648,
      "grad_norm": 0.47304239869117737,
      "learning_rate": 4.460814787185544e-06,
      "loss": 0.7734,
      "step": 2603
    },
    {
      "epoch": 0.10791993037423847,
      "grad_norm": 0.4639924466609955,
      "learning_rate": 4.460607567657177e-06,
      "loss": 0.7527,
      "step": 2604
    },
    {
      "epoch": 0.10796137427991213,
      "grad_norm": 0.4383813142776489,
      "learning_rate": 4.460400348128808e-06,
      "loss": 0.729,
      "step": 2605
    },
    {
      "epoch": 0.10800281818558581,
      "grad_norm": 0.4432336986064911,
      "learning_rate": 4.460193128600439e-06,
      "loss": 0.8047,
      "step": 2606
    },
    {
      "epoch": 0.10804426209125947,
      "grad_norm": 0.4126124978065491,
      "learning_rate": 4.459985909072072e-06,
      "loss": 0.8184,
      "step": 2607
    },
    {
      "epoch": 0.10808570599693315,
      "grad_norm": 0.40564653277397156,
      "learning_rate": 4.459778689543703e-06,
      "loss": 0.7267,
      "step": 2608
    },
    {
      "epoch": 0.10812714990260683,
      "grad_norm": 0.4342767894268036,
      "learning_rate": 4.459571470015334e-06,
      "loss": 0.7693,
      "step": 2609
    },
    {
      "epoch": 0.10816859380828049,
      "grad_norm": 0.4603235125541687,
      "learning_rate": 4.459364250486966e-06,
      "loss": 0.7416,
      "step": 2610
    },
    {
      "epoch": 0.10821003771395417,
      "grad_norm": 0.4464719295501709,
      "learning_rate": 4.459157030958598e-06,
      "loss": 0.7688,
      "step": 2611
    },
    {
      "epoch": 0.10825148161962783,
      "grad_norm": 0.44505593180656433,
      "learning_rate": 4.458949811430229e-06,
      "loss": 0.7732,
      "step": 2612
    },
    {
      "epoch": 0.10829292552530151,
      "grad_norm": 0.4624794125556946,
      "learning_rate": 4.458742591901862e-06,
      "loss": 0.7346,
      "step": 2613
    },
    {
      "epoch": 0.10833436943097517,
      "grad_norm": 0.44707027077674866,
      "learning_rate": 4.458535372373493e-06,
      "loss": 0.7065,
      "step": 2614
    },
    {
      "epoch": 0.10837581333664885,
      "grad_norm": 0.43216633796691895,
      "learning_rate": 4.458328152845124e-06,
      "loss": 0.7554,
      "step": 2615
    },
    {
      "epoch": 0.10841725724232251,
      "grad_norm": 0.44442763924598694,
      "learning_rate": 4.458120933316757e-06,
      "loss": 0.7781,
      "step": 2616
    },
    {
      "epoch": 0.10845870114799619,
      "grad_norm": 0.4309834837913513,
      "learning_rate": 4.457913713788388e-06,
      "loss": 0.7676,
      "step": 2617
    },
    {
      "epoch": 0.10850014505366985,
      "grad_norm": 0.4000563621520996,
      "learning_rate": 4.457706494260019e-06,
      "loss": 0.7036,
      "step": 2618
    },
    {
      "epoch": 0.10854158895934353,
      "grad_norm": 0.4388279318809509,
      "learning_rate": 4.457499274731651e-06,
      "loss": 0.783,
      "step": 2619
    },
    {
      "epoch": 0.10858303286501719,
      "grad_norm": 0.4665571451187134,
      "learning_rate": 4.457292055203283e-06,
      "loss": 0.782,
      "step": 2620
    },
    {
      "epoch": 0.10862447677069087,
      "grad_norm": 0.42799240350723267,
      "learning_rate": 4.457084835674914e-06,
      "loss": 0.7573,
      "step": 2621
    },
    {
      "epoch": 0.10866592067636455,
      "grad_norm": 0.4220544695854187,
      "learning_rate": 4.456877616146546e-06,
      "loss": 0.71,
      "step": 2622
    },
    {
      "epoch": 0.10870736458203821,
      "grad_norm": 0.4573543071746826,
      "learning_rate": 4.456670396618178e-06,
      "loss": 0.8184,
      "step": 2623
    },
    {
      "epoch": 0.10874880848771189,
      "grad_norm": 0.45465657114982605,
      "learning_rate": 4.456463177089809e-06,
      "loss": 0.7831,
      "step": 2624
    },
    {
      "epoch": 0.10879025239338555,
      "grad_norm": 0.4359979033470154,
      "learning_rate": 4.456255957561441e-06,
      "loss": 0.7766,
      "step": 2625
    },
    {
      "epoch": 0.10883169629905923,
      "grad_norm": 0.4265727996826172,
      "learning_rate": 4.456048738033072e-06,
      "loss": 0.7344,
      "step": 2626
    },
    {
      "epoch": 0.10887314020473289,
      "grad_norm": 0.46706801652908325,
      "learning_rate": 4.455841518504704e-06,
      "loss": 0.8069,
      "step": 2627
    },
    {
      "epoch": 0.10891458411040657,
      "grad_norm": 0.4411037564277649,
      "learning_rate": 4.455634298976336e-06,
      "loss": 0.7567,
      "step": 2628
    },
    {
      "epoch": 0.10895602801608023,
      "grad_norm": 0.4615686237812042,
      "learning_rate": 4.455427079447968e-06,
      "loss": 0.7776,
      "step": 2629
    },
    {
      "epoch": 0.10899747192175391,
      "grad_norm": 0.44318488240242004,
      "learning_rate": 4.4552198599195985e-06,
      "loss": 0.7323,
      "step": 2630
    },
    {
      "epoch": 0.10903891582742757,
      "grad_norm": 0.43315544724464417,
      "learning_rate": 4.455012640391231e-06,
      "loss": 0.7627,
      "step": 2631
    },
    {
      "epoch": 0.10908035973310125,
      "grad_norm": 0.45497411489486694,
      "learning_rate": 4.454805420862863e-06,
      "loss": 0.7639,
      "step": 2632
    },
    {
      "epoch": 0.10912180363877491,
      "grad_norm": 0.4300709068775177,
      "learning_rate": 4.454598201334494e-06,
      "loss": 0.7358,
      "step": 2633
    },
    {
      "epoch": 0.10916324754444859,
      "grad_norm": 0.44174784421920776,
      "learning_rate": 4.454390981806126e-06,
      "loss": 0.7284,
      "step": 2634
    },
    {
      "epoch": 0.10920469145012227,
      "grad_norm": 0.4037221670150757,
      "learning_rate": 4.454183762277757e-06,
      "loss": 0.7273,
      "step": 2635
    },
    {
      "epoch": 0.10924613535579593,
      "grad_norm": 0.4601919949054718,
      "learning_rate": 4.453976542749389e-06,
      "loss": 0.8163,
      "step": 2636
    },
    {
      "epoch": 0.1092875792614696,
      "grad_norm": 0.4445796310901642,
      "learning_rate": 4.453769323221021e-06,
      "loss": 0.7301,
      "step": 2637
    },
    {
      "epoch": 0.10932902316714327,
      "grad_norm": 0.4262236952781677,
      "learning_rate": 4.453562103692652e-06,
      "loss": 0.7471,
      "step": 2638
    },
    {
      "epoch": 0.10937046707281695,
      "grad_norm": 0.4741750657558441,
      "learning_rate": 4.4533548841642836e-06,
      "loss": 0.7175,
      "step": 2639
    },
    {
      "epoch": 0.10941191097849061,
      "grad_norm": 0.45271116495132446,
      "learning_rate": 4.453147664635916e-06,
      "loss": 0.7458,
      "step": 2640
    },
    {
      "epoch": 0.10945335488416429,
      "grad_norm": 0.46410730481147766,
      "learning_rate": 4.452940445107547e-06,
      "loss": 0.792,
      "step": 2641
    },
    {
      "epoch": 0.10949479878983795,
      "grad_norm": 0.47136595845222473,
      "learning_rate": 4.4527332255791786e-06,
      "loss": 0.7114,
      "step": 2642
    },
    {
      "epoch": 0.10953624269551163,
      "grad_norm": 0.4221239984035492,
      "learning_rate": 4.452526006050811e-06,
      "loss": 0.7295,
      "step": 2643
    },
    {
      "epoch": 0.10957768660118529,
      "grad_norm": 0.44076746702194214,
      "learning_rate": 4.452318786522442e-06,
      "loss": 0.7869,
      "step": 2644
    },
    {
      "epoch": 0.10961913050685897,
      "grad_norm": 0.45560532808303833,
      "learning_rate": 4.4521115669940736e-06,
      "loss": 0.6993,
      "step": 2645
    },
    {
      "epoch": 0.10966057441253264,
      "grad_norm": 0.44277799129486084,
      "learning_rate": 4.451904347465705e-06,
      "loss": 0.7393,
      "step": 2646
    },
    {
      "epoch": 0.1097020183182063,
      "grad_norm": 0.42285966873168945,
      "learning_rate": 4.451697127937337e-06,
      "loss": 0.7262,
      "step": 2647
    },
    {
      "epoch": 0.10974346222387998,
      "grad_norm": 0.461492121219635,
      "learning_rate": 4.4514899084089686e-06,
      "loss": 0.7725,
      "step": 2648
    },
    {
      "epoch": 0.10978490612955365,
      "grad_norm": 0.44727009534835815,
      "learning_rate": 4.451282688880601e-06,
      "loss": 0.7903,
      "step": 2649
    },
    {
      "epoch": 0.10982635003522732,
      "grad_norm": 0.4347189962863922,
      "learning_rate": 4.451075469352232e-06,
      "loss": 0.7607,
      "step": 2650
    },
    {
      "epoch": 0.10986779394090099,
      "grad_norm": 0.47462618350982666,
      "learning_rate": 4.4508682498238636e-06,
      "loss": 0.7822,
      "step": 2651
    },
    {
      "epoch": 0.10990923784657466,
      "grad_norm": 0.46686244010925293,
      "learning_rate": 4.450661030295496e-06,
      "loss": 0.791,
      "step": 2652
    },
    {
      "epoch": 0.10995068175224833,
      "grad_norm": 0.42717838287353516,
      "learning_rate": 4.450453810767127e-06,
      "loss": 0.759,
      "step": 2653
    },
    {
      "epoch": 0.109992125657922,
      "grad_norm": 0.48696795105934143,
      "learning_rate": 4.4502465912387586e-06,
      "loss": 0.7581,
      "step": 2654
    },
    {
      "epoch": 0.11003356956359567,
      "grad_norm": 0.44663336873054504,
      "learning_rate": 4.45003937171039e-06,
      "loss": 0.7292,
      "step": 2655
    },
    {
      "epoch": 0.11007501346926934,
      "grad_norm": 0.46645960211753845,
      "learning_rate": 4.449832152182022e-06,
      "loss": 0.7485,
      "step": 2656
    },
    {
      "epoch": 0.11011645737494301,
      "grad_norm": 0.49968335032463074,
      "learning_rate": 4.449624932653654e-06,
      "loss": 0.7942,
      "step": 2657
    },
    {
      "epoch": 0.11015790128061669,
      "grad_norm": 0.43673422932624817,
      "learning_rate": 4.449417713125285e-06,
      "loss": 0.7932,
      "step": 2658
    },
    {
      "epoch": 0.11019934518629036,
      "grad_norm": 0.4820801615715027,
      "learning_rate": 4.449210493596917e-06,
      "loss": 0.7268,
      "step": 2659
    },
    {
      "epoch": 0.11024078909196403,
      "grad_norm": 0.445431113243103,
      "learning_rate": 4.449003274068549e-06,
      "loss": 0.7314,
      "step": 2660
    },
    {
      "epoch": 0.1102822329976377,
      "grad_norm": 0.44820666313171387,
      "learning_rate": 4.44879605454018e-06,
      "loss": 0.7737,
      "step": 2661
    },
    {
      "epoch": 0.11032367690331137,
      "grad_norm": 0.4587949514389038,
      "learning_rate": 4.448588835011811e-06,
      "loss": 0.7268,
      "step": 2662
    },
    {
      "epoch": 0.11036512080898504,
      "grad_norm": 0.4527418613433838,
      "learning_rate": 4.448381615483444e-06,
      "loss": 0.7917,
      "step": 2663
    },
    {
      "epoch": 0.1104065647146587,
      "grad_norm": 0.4519367516040802,
      "learning_rate": 4.448174395955075e-06,
      "loss": 0.7375,
      "step": 2664
    },
    {
      "epoch": 0.11044800862033238,
      "grad_norm": 0.4452589154243469,
      "learning_rate": 4.447967176426707e-06,
      "loss": 0.7505,
      "step": 2665
    },
    {
      "epoch": 0.11048945252600605,
      "grad_norm": 0.42448121309280396,
      "learning_rate": 4.447759956898339e-06,
      "loss": 0.7324,
      "step": 2666
    },
    {
      "epoch": 0.11053089643167972,
      "grad_norm": 0.4940919578075409,
      "learning_rate": 4.44755273736997e-06,
      "loss": 0.7766,
      "step": 2667
    },
    {
      "epoch": 0.11057234033735339,
      "grad_norm": 0.4402981698513031,
      "learning_rate": 4.447345517841602e-06,
      "loss": 0.7637,
      "step": 2668
    },
    {
      "epoch": 0.11061378424302706,
      "grad_norm": 0.4562647342681885,
      "learning_rate": 4.447138298313234e-06,
      "loss": 0.7572,
      "step": 2669
    },
    {
      "epoch": 0.11065522814870073,
      "grad_norm": 0.41179582476615906,
      "learning_rate": 4.446931078784865e-06,
      "loss": 0.7583,
      "step": 2670
    },
    {
      "epoch": 0.1106966720543744,
      "grad_norm": 0.4757821261882782,
      "learning_rate": 4.446723859256496e-06,
      "loss": 0.7209,
      "step": 2671
    },
    {
      "epoch": 0.11073811596004808,
      "grad_norm": 0.41268208622932434,
      "learning_rate": 4.446516639728129e-06,
      "loss": 0.7495,
      "step": 2672
    },
    {
      "epoch": 0.11077955986572174,
      "grad_norm": 0.4295004606246948,
      "learning_rate": 4.44630942019976e-06,
      "loss": 0.7671,
      "step": 2673
    },
    {
      "epoch": 0.11082100377139542,
      "grad_norm": 0.45691704750061035,
      "learning_rate": 4.446102200671391e-06,
      "loss": 0.7629,
      "step": 2674
    },
    {
      "epoch": 0.11086244767706908,
      "grad_norm": 0.4245308041572571,
      "learning_rate": 4.445894981143024e-06,
      "loss": 0.7201,
      "step": 2675
    },
    {
      "epoch": 0.11090389158274276,
      "grad_norm": 0.42640846967697144,
      "learning_rate": 4.445687761614655e-06,
      "loss": 0.7725,
      "step": 2676
    },
    {
      "epoch": 0.11094533548841642,
      "grad_norm": 0.4370523989200592,
      "learning_rate": 4.445480542086286e-06,
      "loss": 0.731,
      "step": 2677
    },
    {
      "epoch": 0.1109867793940901,
      "grad_norm": 0.4884931147098541,
      "learning_rate": 4.445273322557918e-06,
      "loss": 0.8086,
      "step": 2678
    },
    {
      "epoch": 0.11102822329976376,
      "grad_norm": 0.4532877206802368,
      "learning_rate": 4.44506610302955e-06,
      "loss": 0.692,
      "step": 2679
    },
    {
      "epoch": 0.11106966720543744,
      "grad_norm": 0.452131986618042,
      "learning_rate": 4.444858883501181e-06,
      "loss": 0.7786,
      "step": 2680
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 0.4116024971008301,
      "learning_rate": 4.444651663972814e-06,
      "loss": 0.7418,
      "step": 2681
    },
    {
      "epoch": 0.11115255501678478,
      "grad_norm": 0.46495407819747925,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.7256,
      "step": 2682
    },
    {
      "epoch": 0.11119399892245846,
      "grad_norm": 0.403834730386734,
      "learning_rate": 4.444237224916076e-06,
      "loss": 0.6836,
      "step": 2683
    },
    {
      "epoch": 0.11123544282813212,
      "grad_norm": 0.47125089168548584,
      "learning_rate": 4.444030005387709e-06,
      "loss": 0.7793,
      "step": 2684
    },
    {
      "epoch": 0.1112768867338058,
      "grad_norm": 0.48455461859703064,
      "learning_rate": 4.44382278585934e-06,
      "loss": 0.8442,
      "step": 2685
    },
    {
      "epoch": 0.11131833063947946,
      "grad_norm": 0.42586570978164673,
      "learning_rate": 4.443615566330971e-06,
      "loss": 0.7323,
      "step": 2686
    },
    {
      "epoch": 0.11135977454515314,
      "grad_norm": 0.42276620864868164,
      "learning_rate": 4.443408346802603e-06,
      "loss": 0.6871,
      "step": 2687
    },
    {
      "epoch": 0.1114012184508268,
      "grad_norm": 0.45395201444625854,
      "learning_rate": 4.443201127274235e-06,
      "loss": 0.7664,
      "step": 2688
    },
    {
      "epoch": 0.11144266235650048,
      "grad_norm": 0.4246496856212616,
      "learning_rate": 4.442993907745866e-06,
      "loss": 0.748,
      "step": 2689
    },
    {
      "epoch": 0.11148410626217414,
      "grad_norm": 0.5120562314987183,
      "learning_rate": 4.442786688217498e-06,
      "loss": 0.7644,
      "step": 2690
    },
    {
      "epoch": 0.11152555016784782,
      "grad_norm": 0.41809457540512085,
      "learning_rate": 4.442579468689129e-06,
      "loss": 0.7563,
      "step": 2691
    },
    {
      "epoch": 0.11156699407352148,
      "grad_norm": 0.4900083839893341,
      "learning_rate": 4.442372249160761e-06,
      "loss": 0.7598,
      "step": 2692
    },
    {
      "epoch": 0.11160843797919516,
      "grad_norm": 0.4255141019821167,
      "learning_rate": 4.442165029632393e-06,
      "loss": 0.7058,
      "step": 2693
    },
    {
      "epoch": 0.11164988188486882,
      "grad_norm": 0.4236755073070526,
      "learning_rate": 4.441957810104024e-06,
      "loss": 0.7273,
      "step": 2694
    },
    {
      "epoch": 0.1116913257905425,
      "grad_norm": 0.45340660214424133,
      "learning_rate": 4.441750590575656e-06,
      "loss": 0.7192,
      "step": 2695
    },
    {
      "epoch": 0.11173276969621618,
      "grad_norm": 0.4422585666179657,
      "learning_rate": 4.441543371047288e-06,
      "loss": 0.7549,
      "step": 2696
    },
    {
      "epoch": 0.11177421360188984,
      "grad_norm": 0.4164329469203949,
      "learning_rate": 4.44133615151892e-06,
      "loss": 0.6973,
      "step": 2697
    },
    {
      "epoch": 0.11181565750756352,
      "grad_norm": 0.41723304986953735,
      "learning_rate": 4.4411289319905506e-06,
      "loss": 0.7424,
      "step": 2698
    },
    {
      "epoch": 0.11185710141323718,
      "grad_norm": 0.4329625368118286,
      "learning_rate": 4.440921712462183e-06,
      "loss": 0.7218,
      "step": 2699
    },
    {
      "epoch": 0.11189854531891086,
      "grad_norm": 0.44698718190193176,
      "learning_rate": 4.440714492933814e-06,
      "loss": 0.7212,
      "step": 2700
    },
    {
      "epoch": 0.11193998922458452,
      "grad_norm": 0.44116076827049255,
      "learning_rate": 4.440507273405446e-06,
      "loss": 0.7168,
      "step": 2701
    },
    {
      "epoch": 0.1119814331302582,
      "grad_norm": 0.4057162404060364,
      "learning_rate": 4.440300053877078e-06,
      "loss": 0.6975,
      "step": 2702
    },
    {
      "epoch": 0.11202287703593186,
      "grad_norm": 0.44595563411712646,
      "learning_rate": 4.440092834348709e-06,
      "loss": 0.7754,
      "step": 2703
    },
    {
      "epoch": 0.11206432094160554,
      "grad_norm": 0.43207693099975586,
      "learning_rate": 4.439885614820341e-06,
      "loss": 0.7507,
      "step": 2704
    },
    {
      "epoch": 0.1121057648472792,
      "grad_norm": 0.44918304681777954,
      "learning_rate": 4.439678395291973e-06,
      "loss": 0.7043,
      "step": 2705
    },
    {
      "epoch": 0.11214720875295288,
      "grad_norm": 0.4036732316017151,
      "learning_rate": 4.439471175763604e-06,
      "loss": 0.6873,
      "step": 2706
    },
    {
      "epoch": 0.11218865265862656,
      "grad_norm": 0.5025752186775208,
      "learning_rate": 4.4392639562352356e-06,
      "loss": 0.8108,
      "step": 2707
    },
    {
      "epoch": 0.11223009656430022,
      "grad_norm": 0.44829800724983215,
      "learning_rate": 4.439056736706868e-06,
      "loss": 0.7283,
      "step": 2708
    },
    {
      "epoch": 0.1122715404699739,
      "grad_norm": 0.42198845744132996,
      "learning_rate": 4.438849517178499e-06,
      "loss": 0.7668,
      "step": 2709
    },
    {
      "epoch": 0.11231298437564756,
      "grad_norm": 0.4700067937374115,
      "learning_rate": 4.4386422976501306e-06,
      "loss": 0.7439,
      "step": 2710
    },
    {
      "epoch": 0.11235442828132124,
      "grad_norm": 0.4713732600212097,
      "learning_rate": 4.438435078121763e-06,
      "loss": 0.7996,
      "step": 2711
    },
    {
      "epoch": 0.1123958721869949,
      "grad_norm": 0.4232773780822754,
      "learning_rate": 4.438227858593394e-06,
      "loss": 0.6786,
      "step": 2712
    },
    {
      "epoch": 0.11243731609266858,
      "grad_norm": 0.42234620451927185,
      "learning_rate": 4.438020639065026e-06,
      "loss": 0.746,
      "step": 2713
    },
    {
      "epoch": 0.11247875999834224,
      "grad_norm": 0.4128069281578064,
      "learning_rate": 4.437813419536657e-06,
      "loss": 0.7758,
      "step": 2714
    },
    {
      "epoch": 0.11252020390401592,
      "grad_norm": 0.4280022382736206,
      "learning_rate": 4.437606200008289e-06,
      "loss": 0.7405,
      "step": 2715
    },
    {
      "epoch": 0.11256164780968958,
      "grad_norm": 0.44889023900032043,
      "learning_rate": 4.437398980479921e-06,
      "loss": 0.7471,
      "step": 2716
    },
    {
      "epoch": 0.11260309171536326,
      "grad_norm": 0.4560686945915222,
      "learning_rate": 4.437191760951553e-06,
      "loss": 0.7874,
      "step": 2717
    },
    {
      "epoch": 0.11264453562103692,
      "grad_norm": 0.42480576038360596,
      "learning_rate": 4.436984541423184e-06,
      "loss": 0.781,
      "step": 2718
    },
    {
      "epoch": 0.1126859795267106,
      "grad_norm": 0.45407724380493164,
      "learning_rate": 4.436777321894816e-06,
      "loss": 0.7649,
      "step": 2719
    },
    {
      "epoch": 0.11272742343238427,
      "grad_norm": 0.4099579453468323,
      "learning_rate": 4.436570102366448e-06,
      "loss": 0.7183,
      "step": 2720
    },
    {
      "epoch": 0.11276886733805794,
      "grad_norm": 0.4756443500518799,
      "learning_rate": 4.436362882838079e-06,
      "loss": 0.7568,
      "step": 2721
    },
    {
      "epoch": 0.11281031124373161,
      "grad_norm": 0.4558103084564209,
      "learning_rate": 4.436155663309711e-06,
      "loss": 0.7421,
      "step": 2722
    },
    {
      "epoch": 0.11285175514940528,
      "grad_norm": 0.4526768624782562,
      "learning_rate": 4.435948443781342e-06,
      "loss": 0.7441,
      "step": 2723
    },
    {
      "epoch": 0.11289319905507895,
      "grad_norm": 0.43101441860198975,
      "learning_rate": 4.435741224252974e-06,
      "loss": 0.7542,
      "step": 2724
    },
    {
      "epoch": 0.11293464296075262,
      "grad_norm": 0.45606544613838196,
      "learning_rate": 4.435534004724606e-06,
      "loss": 0.647,
      "step": 2725
    },
    {
      "epoch": 0.1129760868664263,
      "grad_norm": 0.4541940689086914,
      "learning_rate": 4.435326785196237e-06,
      "loss": 0.7512,
      "step": 2726
    },
    {
      "epoch": 0.11301753077209996,
      "grad_norm": 0.42886826395988464,
      "learning_rate": 4.435119565667869e-06,
      "loss": 0.7468,
      "step": 2727
    },
    {
      "epoch": 0.11305897467777364,
      "grad_norm": 0.43372687697410583,
      "learning_rate": 4.434912346139501e-06,
      "loss": 0.6733,
      "step": 2728
    },
    {
      "epoch": 0.1131004185834473,
      "grad_norm": 0.5088160634040833,
      "learning_rate": 4.434705126611132e-06,
      "loss": 0.8206,
      "step": 2729
    },
    {
      "epoch": 0.11314186248912098,
      "grad_norm": 0.4149557054042816,
      "learning_rate": 4.434497907082763e-06,
      "loss": 0.6901,
      "step": 2730
    },
    {
      "epoch": 0.11318330639479464,
      "grad_norm": 0.42920351028442383,
      "learning_rate": 4.434290687554396e-06,
      "loss": 0.7567,
      "step": 2731
    },
    {
      "epoch": 0.11322475030046832,
      "grad_norm": 0.440023809671402,
      "learning_rate": 4.434083468026027e-06,
      "loss": 0.7563,
      "step": 2732
    },
    {
      "epoch": 0.11326619420614199,
      "grad_norm": 0.43239036202430725,
      "learning_rate": 4.433876248497659e-06,
      "loss": 0.759,
      "step": 2733
    },
    {
      "epoch": 0.11330763811181566,
      "grad_norm": 0.4365358054637909,
      "learning_rate": 4.433669028969291e-06,
      "loss": 0.699,
      "step": 2734
    },
    {
      "epoch": 0.11334908201748933,
      "grad_norm": 0.42414575815200806,
      "learning_rate": 4.433461809440922e-06,
      "loss": 0.7576,
      "step": 2735
    },
    {
      "epoch": 0.113390525923163,
      "grad_norm": 0.4178504943847656,
      "learning_rate": 4.433254589912554e-06,
      "loss": 0.7292,
      "step": 2736
    },
    {
      "epoch": 0.11343196982883667,
      "grad_norm": 0.4506518542766571,
      "learning_rate": 4.433047370384186e-06,
      "loss": 0.7974,
      "step": 2737
    },
    {
      "epoch": 0.11347341373451034,
      "grad_norm": 0.41514337062835693,
      "learning_rate": 4.432840150855817e-06,
      "loss": 0.7666,
      "step": 2738
    },
    {
      "epoch": 0.11351485764018401,
      "grad_norm": 0.440890371799469,
      "learning_rate": 4.432632931327448e-06,
      "loss": 0.7747,
      "step": 2739
    },
    {
      "epoch": 0.11355630154585768,
      "grad_norm": 0.4055033028125763,
      "learning_rate": 4.432425711799081e-06,
      "loss": 0.7517,
      "step": 2740
    },
    {
      "epoch": 0.11359774545153135,
      "grad_norm": 0.4848836362361908,
      "learning_rate": 4.432218492270712e-06,
      "loss": 0.8174,
      "step": 2741
    },
    {
      "epoch": 0.11363918935720502,
      "grad_norm": 0.4597133696079254,
      "learning_rate": 4.432011272742343e-06,
      "loss": 0.7229,
      "step": 2742
    },
    {
      "epoch": 0.1136806332628787,
      "grad_norm": 0.4338879883289337,
      "learning_rate": 4.431804053213975e-06,
      "loss": 0.6809,
      "step": 2743
    },
    {
      "epoch": 0.11372207716855237,
      "grad_norm": 0.4469551146030426,
      "learning_rate": 4.431596833685607e-06,
      "loss": 0.8262,
      "step": 2744
    },
    {
      "epoch": 0.11376352107422603,
      "grad_norm": 0.42504581809043884,
      "learning_rate": 4.431389614157238e-06,
      "loss": 0.7744,
      "step": 2745
    },
    {
      "epoch": 0.11380496497989971,
      "grad_norm": 0.44543859362602234,
      "learning_rate": 4.43118239462887e-06,
      "loss": 0.7966,
      "step": 2746
    },
    {
      "epoch": 0.11384640888557337,
      "grad_norm": 0.428748220205307,
      "learning_rate": 4.430975175100502e-06,
      "loss": 0.7635,
      "step": 2747
    },
    {
      "epoch": 0.11388785279124705,
      "grad_norm": 0.41466960310935974,
      "learning_rate": 4.430767955572133e-06,
      "loss": 0.7266,
      "step": 2748
    },
    {
      "epoch": 0.11392929669692071,
      "grad_norm": 0.4365457594394684,
      "learning_rate": 4.430560736043766e-06,
      "loss": 0.7126,
      "step": 2749
    },
    {
      "epoch": 0.11397074060259439,
      "grad_norm": 0.41085758805274963,
      "learning_rate": 4.430353516515396e-06,
      "loss": 0.7722,
      "step": 2750
    },
    {
      "epoch": 0.11401218450826806,
      "grad_norm": 0.43684977293014526,
      "learning_rate": 4.430146296987028e-06,
      "loss": 0.74,
      "step": 2751
    },
    {
      "epoch": 0.11405362841394173,
      "grad_norm": 0.46030476689338684,
      "learning_rate": 4.42993907745866e-06,
      "loss": 0.7358,
      "step": 2752
    },
    {
      "epoch": 0.1140950723196154,
      "grad_norm": 0.45212092995643616,
      "learning_rate": 4.429731857930292e-06,
      "loss": 0.7379,
      "step": 2753
    },
    {
      "epoch": 0.11413651622528907,
      "grad_norm": 0.4332735538482666,
      "learning_rate": 4.429524638401923e-06,
      "loss": 0.7803,
      "step": 2754
    },
    {
      "epoch": 0.11417796013096274,
      "grad_norm": 0.5195847153663635,
      "learning_rate": 4.429317418873555e-06,
      "loss": 0.7593,
      "step": 2755
    },
    {
      "epoch": 0.11421940403663641,
      "grad_norm": 0.5018402934074402,
      "learning_rate": 4.429110199345187e-06,
      "loss": 0.7976,
      "step": 2756
    },
    {
      "epoch": 0.11426084794231009,
      "grad_norm": 0.4255725145339966,
      "learning_rate": 4.428902979816818e-06,
      "loss": 0.7102,
      "step": 2757
    },
    {
      "epoch": 0.11430229184798375,
      "grad_norm": 0.44767361879348755,
      "learning_rate": 4.42869576028845e-06,
      "loss": 0.7186,
      "step": 2758
    },
    {
      "epoch": 0.11434373575365743,
      "grad_norm": 0.44474491477012634,
      "learning_rate": 4.428488540760081e-06,
      "loss": 0.7628,
      "step": 2759
    },
    {
      "epoch": 0.1143851796593311,
      "grad_norm": 0.47939980030059814,
      "learning_rate": 4.428281321231713e-06,
      "loss": 0.7556,
      "step": 2760
    },
    {
      "epoch": 0.11442662356500477,
      "grad_norm": 0.4553447961807251,
      "learning_rate": 4.428074101703345e-06,
      "loss": 0.7471,
      "step": 2761
    },
    {
      "epoch": 0.11446806747067843,
      "grad_norm": 0.4238758385181427,
      "learning_rate": 4.427866882174976e-06,
      "loss": 0.7615,
      "step": 2762
    },
    {
      "epoch": 0.11450951137635211,
      "grad_norm": 0.4461573660373688,
      "learning_rate": 4.427659662646608e-06,
      "loss": 0.7463,
      "step": 2763
    },
    {
      "epoch": 0.11455095528202577,
      "grad_norm": 0.46187373995780945,
      "learning_rate": 4.42745244311824e-06,
      "loss": 0.7378,
      "step": 2764
    },
    {
      "epoch": 0.11459239918769945,
      "grad_norm": 0.4527425765991211,
      "learning_rate": 4.427245223589872e-06,
      "loss": 0.7766,
      "step": 2765
    },
    {
      "epoch": 0.11463384309337311,
      "grad_norm": 0.43725085258483887,
      "learning_rate": 4.4270380040615026e-06,
      "loss": 0.7151,
      "step": 2766
    },
    {
      "epoch": 0.11467528699904679,
      "grad_norm": 0.42665162682533264,
      "learning_rate": 4.426830784533135e-06,
      "loss": 0.677,
      "step": 2767
    },
    {
      "epoch": 0.11471673090472045,
      "grad_norm": 0.4208908677101135,
      "learning_rate": 4.426623565004766e-06,
      "loss": 0.7637,
      "step": 2768
    },
    {
      "epoch": 0.11475817481039413,
      "grad_norm": 0.46116581559181213,
      "learning_rate": 4.426416345476398e-06,
      "loss": 0.8081,
      "step": 2769
    },
    {
      "epoch": 0.11479961871606781,
      "grad_norm": 0.4412277638912201,
      "learning_rate": 4.42620912594803e-06,
      "loss": 0.7505,
      "step": 2770
    },
    {
      "epoch": 0.11484106262174147,
      "grad_norm": 0.39626502990722656,
      "learning_rate": 4.426001906419661e-06,
      "loss": 0.7441,
      "step": 2771
    },
    {
      "epoch": 0.11488250652741515,
      "grad_norm": 0.45137420296669006,
      "learning_rate": 4.425794686891293e-06,
      "loss": 0.7388,
      "step": 2772
    },
    {
      "epoch": 0.11492395043308881,
      "grad_norm": 0.45052120089530945,
      "learning_rate": 4.425587467362925e-06,
      "loss": 0.7671,
      "step": 2773
    },
    {
      "epoch": 0.11496539433876249,
      "grad_norm": 0.4616880416870117,
      "learning_rate": 4.425380247834556e-06,
      "loss": 0.7346,
      "step": 2774
    },
    {
      "epoch": 0.11500683824443615,
      "grad_norm": 0.4316191077232361,
      "learning_rate": 4.425173028306188e-06,
      "loss": 0.7233,
      "step": 2775
    },
    {
      "epoch": 0.11504828215010983,
      "grad_norm": 0.4336789846420288,
      "learning_rate": 4.42496580877782e-06,
      "loss": 0.7964,
      "step": 2776
    },
    {
      "epoch": 0.11508972605578349,
      "grad_norm": 0.7159467339515686,
      "learning_rate": 4.424758589249451e-06,
      "loss": 0.7932,
      "step": 2777
    },
    {
      "epoch": 0.11513116996145717,
      "grad_norm": 0.4159564971923828,
      "learning_rate": 4.424551369721083e-06,
      "loss": 0.7306,
      "step": 2778
    },
    {
      "epoch": 0.11517261386713083,
      "grad_norm": 0.43652257323265076,
      "learning_rate": 4.424344150192715e-06,
      "loss": 0.73,
      "step": 2779
    },
    {
      "epoch": 0.11521405777280451,
      "grad_norm": 0.43522194027900696,
      "learning_rate": 4.424136930664346e-06,
      "loss": 0.7019,
      "step": 2780
    },
    {
      "epoch": 0.11525550167847819,
      "grad_norm": 0.45654237270355225,
      "learning_rate": 4.4239297111359784e-06,
      "loss": 0.7343,
      "step": 2781
    },
    {
      "epoch": 0.11529694558415185,
      "grad_norm": 0.4355565309524536,
      "learning_rate": 4.423722491607609e-06,
      "loss": 0.7112,
      "step": 2782
    },
    {
      "epoch": 0.11533838948982553,
      "grad_norm": 0.44342443346977234,
      "learning_rate": 4.423515272079241e-06,
      "loss": 0.7437,
      "step": 2783
    },
    {
      "epoch": 0.11537983339549919,
      "grad_norm": 0.45237478613853455,
      "learning_rate": 4.423308052550873e-06,
      "loss": 0.7783,
      "step": 2784
    },
    {
      "epoch": 0.11542127730117287,
      "grad_norm": 0.4410804510116577,
      "learning_rate": 4.423100833022505e-06,
      "loss": 0.793,
      "step": 2785
    },
    {
      "epoch": 0.11546272120684653,
      "grad_norm": 0.40193304419517517,
      "learning_rate": 4.422893613494136e-06,
      "loss": 0.7175,
      "step": 2786
    },
    {
      "epoch": 0.11550416511252021,
      "grad_norm": 0.46401405334472656,
      "learning_rate": 4.422686393965768e-06,
      "loss": 0.7769,
      "step": 2787
    },
    {
      "epoch": 0.11554560901819387,
      "grad_norm": 0.4480639100074768,
      "learning_rate": 4.4224791744374e-06,
      "loss": 0.7383,
      "step": 2788
    },
    {
      "epoch": 0.11558705292386755,
      "grad_norm": 0.4836352467536926,
      "learning_rate": 4.422271954909031e-06,
      "loss": 0.8516,
      "step": 2789
    },
    {
      "epoch": 0.11562849682954121,
      "grad_norm": 0.4390290677547455,
      "learning_rate": 4.422064735380663e-06,
      "loss": 0.7268,
      "step": 2790
    },
    {
      "epoch": 0.11566994073521489,
      "grad_norm": 0.40692535042762756,
      "learning_rate": 4.421857515852294e-06,
      "loss": 0.7292,
      "step": 2791
    },
    {
      "epoch": 0.11571138464088855,
      "grad_norm": 0.41716626286506653,
      "learning_rate": 4.421650296323926e-06,
      "loss": 0.7427,
      "step": 2792
    },
    {
      "epoch": 0.11575282854656223,
      "grad_norm": 0.4251644015312195,
      "learning_rate": 4.421443076795558e-06,
      "loss": 0.7302,
      "step": 2793
    },
    {
      "epoch": 0.1157942724522359,
      "grad_norm": 0.45191988348960876,
      "learning_rate": 4.421235857267189e-06,
      "loss": 0.7212,
      "step": 2794
    },
    {
      "epoch": 0.11583571635790957,
      "grad_norm": 0.5190073251724243,
      "learning_rate": 4.42102863773882e-06,
      "loss": 0.7698,
      "step": 2795
    },
    {
      "epoch": 0.11587716026358325,
      "grad_norm": 0.44582998752593994,
      "learning_rate": 4.420821418210453e-06,
      "loss": 0.7231,
      "step": 2796
    },
    {
      "epoch": 0.11591860416925691,
      "grad_norm": 0.4789949059486389,
      "learning_rate": 4.420614198682084e-06,
      "loss": 0.802,
      "step": 2797
    },
    {
      "epoch": 0.11596004807493059,
      "grad_norm": 0.42170462012290955,
      "learning_rate": 4.420406979153715e-06,
      "loss": 0.7805,
      "step": 2798
    },
    {
      "epoch": 0.11600149198060425,
      "grad_norm": 0.46203818917274475,
      "learning_rate": 4.420199759625348e-06,
      "loss": 0.7177,
      "step": 2799
    },
    {
      "epoch": 0.11604293588627793,
      "grad_norm": 0.45809707045555115,
      "learning_rate": 4.419992540096979e-06,
      "loss": 0.8408,
      "step": 2800
    },
    {
      "epoch": 0.11608437979195159,
      "grad_norm": 0.4528385102748871,
      "learning_rate": 4.419785320568611e-06,
      "loss": 0.7649,
      "step": 2801
    },
    {
      "epoch": 0.11612582369762527,
      "grad_norm": 0.45841744542121887,
      "learning_rate": 4.419578101040243e-06,
      "loss": 0.749,
      "step": 2802
    },
    {
      "epoch": 0.11616726760329893,
      "grad_norm": 0.43136778473854065,
      "learning_rate": 4.419370881511874e-06,
      "loss": 0.7336,
      "step": 2803
    },
    {
      "epoch": 0.1162087115089726,
      "grad_norm": 0.4518454968929291,
      "learning_rate": 4.419163661983505e-06,
      "loss": 0.8184,
      "step": 2804
    },
    {
      "epoch": 0.11625015541464627,
      "grad_norm": 0.43808814883232117,
      "learning_rate": 4.418956442455138e-06,
      "loss": 0.7864,
      "step": 2805
    },
    {
      "epoch": 0.11629159932031995,
      "grad_norm": 0.4551939070224762,
      "learning_rate": 4.418749222926769e-06,
      "loss": 0.792,
      "step": 2806
    },
    {
      "epoch": 0.11633304322599362,
      "grad_norm": 0.4534491002559662,
      "learning_rate": 4.4185420033984e-06,
      "loss": 0.7656,
      "step": 2807
    },
    {
      "epoch": 0.11637448713166729,
      "grad_norm": 0.47964048385620117,
      "learning_rate": 4.418334783870033e-06,
      "loss": 0.7192,
      "step": 2808
    },
    {
      "epoch": 0.11641593103734096,
      "grad_norm": 0.602651834487915,
      "learning_rate": 4.418127564341664e-06,
      "loss": 0.811,
      "step": 2809
    },
    {
      "epoch": 0.11645737494301463,
      "grad_norm": 0.4457892179489136,
      "learning_rate": 4.417920344813295e-06,
      "loss": 0.7021,
      "step": 2810
    },
    {
      "epoch": 0.1164988188486883,
      "grad_norm": 0.44804519414901733,
      "learning_rate": 4.417713125284927e-06,
      "loss": 0.833,
      "step": 2811
    },
    {
      "epoch": 0.11654026275436197,
      "grad_norm": 0.42318880558013916,
      "learning_rate": 4.417505905756559e-06,
      "loss": 0.7114,
      "step": 2812
    },
    {
      "epoch": 0.11658170666003564,
      "grad_norm": 0.42301371693611145,
      "learning_rate": 4.41729868622819e-06,
      "loss": 0.6998,
      "step": 2813
    },
    {
      "epoch": 0.11662315056570931,
      "grad_norm": 0.44101518392562866,
      "learning_rate": 4.417091466699822e-06,
      "loss": 0.7778,
      "step": 2814
    },
    {
      "epoch": 0.11666459447138298,
      "grad_norm": 0.4298291802406311,
      "learning_rate": 4.416884247171454e-06,
      "loss": 0.7009,
      "step": 2815
    },
    {
      "epoch": 0.11670603837705665,
      "grad_norm": 0.41529765725135803,
      "learning_rate": 4.416677027643085e-06,
      "loss": 0.7078,
      "step": 2816
    },
    {
      "epoch": 0.11674748228273032,
      "grad_norm": 0.4735003709793091,
      "learning_rate": 4.416469808114718e-06,
      "loss": 0.7148,
      "step": 2817
    },
    {
      "epoch": 0.116788926188404,
      "grad_norm": 0.47490695118904114,
      "learning_rate": 4.416262588586349e-06,
      "loss": 0.7284,
      "step": 2818
    },
    {
      "epoch": 0.11683037009407767,
      "grad_norm": 0.4374504089355469,
      "learning_rate": 4.41605536905798e-06,
      "loss": 0.7554,
      "step": 2819
    },
    {
      "epoch": 0.11687181399975134,
      "grad_norm": 0.4816664755344391,
      "learning_rate": 4.415848149529612e-06,
      "loss": 0.7041,
      "step": 2820
    },
    {
      "epoch": 0.116913257905425,
      "grad_norm": 0.4486456513404846,
      "learning_rate": 4.415640930001244e-06,
      "loss": 0.6736,
      "step": 2821
    },
    {
      "epoch": 0.11695470181109868,
      "grad_norm": 0.4763176739215851,
      "learning_rate": 4.415433710472875e-06,
      "loss": 0.8169,
      "step": 2822
    },
    {
      "epoch": 0.11699614571677235,
      "grad_norm": 0.4386517107486725,
      "learning_rate": 4.415226490944507e-06,
      "loss": 0.7712,
      "step": 2823
    },
    {
      "epoch": 0.11703758962244602,
      "grad_norm": 0.43346861004829407,
      "learning_rate": 4.415019271416139e-06,
      "loss": 0.7861,
      "step": 2824
    },
    {
      "epoch": 0.11707903352811969,
      "grad_norm": 0.44658470153808594,
      "learning_rate": 4.41481205188777e-06,
      "loss": 0.8118,
      "step": 2825
    },
    {
      "epoch": 0.11712047743379336,
      "grad_norm": 0.40515998005867004,
      "learning_rate": 4.414604832359402e-06,
      "loss": 0.6902,
      "step": 2826
    },
    {
      "epoch": 0.11716192133946703,
      "grad_norm": 0.4851025342941284,
      "learning_rate": 4.414397612831033e-06,
      "loss": 0.7261,
      "step": 2827
    },
    {
      "epoch": 0.1172033652451407,
      "grad_norm": 0.42829954624176025,
      "learning_rate": 4.414190393302665e-06,
      "loss": 0.7395,
      "step": 2828
    },
    {
      "epoch": 0.11724480915081437,
      "grad_norm": 0.4407811164855957,
      "learning_rate": 4.413983173774297e-06,
      "loss": 0.6882,
      "step": 2829
    },
    {
      "epoch": 0.11728625305648804,
      "grad_norm": 0.4513356685638428,
      "learning_rate": 4.413775954245928e-06,
      "loss": 0.7505,
      "step": 2830
    },
    {
      "epoch": 0.11732769696216172,
      "grad_norm": 0.4447530210018158,
      "learning_rate": 4.41356873471756e-06,
      "loss": 0.7021,
      "step": 2831
    },
    {
      "epoch": 0.11736914086783538,
      "grad_norm": 0.4741172194480896,
      "learning_rate": 4.413361515189192e-06,
      "loss": 0.8032,
      "step": 2832
    },
    {
      "epoch": 0.11741058477350906,
      "grad_norm": 0.42143911123275757,
      "learning_rate": 4.413154295660824e-06,
      "loss": 0.7732,
      "step": 2833
    },
    {
      "epoch": 0.11745202867918272,
      "grad_norm": 0.4959280788898468,
      "learning_rate": 4.412947076132455e-06,
      "loss": 0.7542,
      "step": 2834
    },
    {
      "epoch": 0.1174934725848564,
      "grad_norm": 0.46807384490966797,
      "learning_rate": 4.412739856604087e-06,
      "loss": 0.7886,
      "step": 2835
    },
    {
      "epoch": 0.11753491649053006,
      "grad_norm": 0.44510769844055176,
      "learning_rate": 4.412532637075718e-06,
      "loss": 0.6924,
      "step": 2836
    },
    {
      "epoch": 0.11757636039620374,
      "grad_norm": 0.4515511393547058,
      "learning_rate": 4.4123254175473504e-06,
      "loss": 0.8188,
      "step": 2837
    },
    {
      "epoch": 0.1176178043018774,
      "grad_norm": 0.40653982758522034,
      "learning_rate": 4.412118198018982e-06,
      "loss": 0.762,
      "step": 2838
    },
    {
      "epoch": 0.11765924820755108,
      "grad_norm": 0.4183030128479004,
      "learning_rate": 4.411910978490613e-06,
      "loss": 0.7261,
      "step": 2839
    },
    {
      "epoch": 0.11770069211322474,
      "grad_norm": 0.4268110692501068,
      "learning_rate": 4.4117037589622454e-06,
      "loss": 0.8242,
      "step": 2840
    },
    {
      "epoch": 0.11774213601889842,
      "grad_norm": 0.44316720962524414,
      "learning_rate": 4.411496539433877e-06,
      "loss": 0.7043,
      "step": 2841
    },
    {
      "epoch": 0.11778357992457208,
      "grad_norm": 0.4742127060890198,
      "learning_rate": 4.411289319905508e-06,
      "loss": 0.8293,
      "step": 2842
    },
    {
      "epoch": 0.11782502383024576,
      "grad_norm": 0.48345068097114563,
      "learning_rate": 4.41108210037714e-06,
      "loss": 0.7239,
      "step": 2843
    },
    {
      "epoch": 0.11786646773591944,
      "grad_norm": 0.44981148838996887,
      "learning_rate": 4.410874880848772e-06,
      "loss": 0.7478,
      "step": 2844
    },
    {
      "epoch": 0.1179079116415931,
      "grad_norm": 0.45307838916778564,
      "learning_rate": 4.410667661320403e-06,
      "loss": 0.7642,
      "step": 2845
    },
    {
      "epoch": 0.11794935554726678,
      "grad_norm": 0.42163315415382385,
      "learning_rate": 4.410460441792035e-06,
      "loss": 0.7073,
      "step": 2846
    },
    {
      "epoch": 0.11799079945294044,
      "grad_norm": 0.4323239028453827,
      "learning_rate": 4.410253222263666e-06,
      "loss": 0.7734,
      "step": 2847
    },
    {
      "epoch": 0.11803224335861412,
      "grad_norm": 0.45530304312705994,
      "learning_rate": 4.410046002735298e-06,
      "loss": 0.7803,
      "step": 2848
    },
    {
      "epoch": 0.11807368726428778,
      "grad_norm": 0.44923776388168335,
      "learning_rate": 4.4098387832069304e-06,
      "loss": 0.7141,
      "step": 2849
    },
    {
      "epoch": 0.11811513116996146,
      "grad_norm": 0.40777868032455444,
      "learning_rate": 4.409631563678561e-06,
      "loss": 0.7539,
      "step": 2850
    },
    {
      "epoch": 0.11815657507563512,
      "grad_norm": 0.44848841428756714,
      "learning_rate": 4.409424344150193e-06,
      "loss": 0.7744,
      "step": 2851
    },
    {
      "epoch": 0.1181980189813088,
      "grad_norm": 0.42930397391319275,
      "learning_rate": 4.409217124621825e-06,
      "loss": 0.7891,
      "step": 2852
    },
    {
      "epoch": 0.11823946288698246,
      "grad_norm": 0.42152366042137146,
      "learning_rate": 4.409009905093457e-06,
      "loss": 0.7692,
      "step": 2853
    },
    {
      "epoch": 0.11828090679265614,
      "grad_norm": 0.4435049295425415,
      "learning_rate": 4.408802685565088e-06,
      "loss": 0.7336,
      "step": 2854
    },
    {
      "epoch": 0.11832235069832982,
      "grad_norm": 0.4527696669101715,
      "learning_rate": 4.40859546603672e-06,
      "loss": 0.7751,
      "step": 2855
    },
    {
      "epoch": 0.11836379460400348,
      "grad_norm": 0.4073115885257721,
      "learning_rate": 4.408388246508351e-06,
      "loss": 0.7212,
      "step": 2856
    },
    {
      "epoch": 0.11840523850967716,
      "grad_norm": 0.4140501916408539,
      "learning_rate": 4.408181026979983e-06,
      "loss": 0.8445,
      "step": 2857
    },
    {
      "epoch": 0.11844668241535082,
      "grad_norm": 0.4351992607116699,
      "learning_rate": 4.407973807451615e-06,
      "loss": 0.7981,
      "step": 2858
    },
    {
      "epoch": 0.1184881263210245,
      "grad_norm": 0.4774436354637146,
      "learning_rate": 4.407766587923246e-06,
      "loss": 0.6777,
      "step": 2859
    },
    {
      "epoch": 0.11852957022669816,
      "grad_norm": 0.4216829538345337,
      "learning_rate": 4.407559368394878e-06,
      "loss": 0.7344,
      "step": 2860
    },
    {
      "epoch": 0.11857101413237184,
      "grad_norm": 0.4740227162837982,
      "learning_rate": 4.40735214886651e-06,
      "loss": 0.7681,
      "step": 2861
    },
    {
      "epoch": 0.1186124580380455,
      "grad_norm": 0.45561811327934265,
      "learning_rate": 4.407144929338141e-06,
      "loss": 0.7295,
      "step": 2862
    },
    {
      "epoch": 0.11865390194371918,
      "grad_norm": 0.42300140857696533,
      "learning_rate": 4.406937709809772e-06,
      "loss": 0.7798,
      "step": 2863
    },
    {
      "epoch": 0.11869534584939284,
      "grad_norm": 0.448377400636673,
      "learning_rate": 4.406730490281405e-06,
      "loss": 0.7944,
      "step": 2864
    },
    {
      "epoch": 0.11873678975506652,
      "grad_norm": 0.4524001479148865,
      "learning_rate": 4.406523270753036e-06,
      "loss": 0.761,
      "step": 2865
    },
    {
      "epoch": 0.11877823366074018,
      "grad_norm": 0.4489184021949768,
      "learning_rate": 4.406316051224667e-06,
      "loss": 0.7761,
      "step": 2866
    },
    {
      "epoch": 0.11881967756641386,
      "grad_norm": 0.39889711141586304,
      "learning_rate": 4.4061088316963e-06,
      "loss": 0.7598,
      "step": 2867
    },
    {
      "epoch": 0.11886112147208754,
      "grad_norm": 0.5122438073158264,
      "learning_rate": 4.405901612167931e-06,
      "loss": 0.7888,
      "step": 2868
    },
    {
      "epoch": 0.1189025653777612,
      "grad_norm": 0.4299689531326294,
      "learning_rate": 4.405694392639563e-06,
      "loss": 0.761,
      "step": 2869
    },
    {
      "epoch": 0.11894400928343488,
      "grad_norm": 0.4325951039791107,
      "learning_rate": 4.405487173111195e-06,
      "loss": 0.7532,
      "step": 2870
    },
    {
      "epoch": 0.11898545318910854,
      "grad_norm": 0.43123406171798706,
      "learning_rate": 4.405279953582826e-06,
      "loss": 0.7488,
      "step": 2871
    },
    {
      "epoch": 0.11902689709478222,
      "grad_norm": 0.46627750992774963,
      "learning_rate": 4.405072734054457e-06,
      "loss": 0.703,
      "step": 2872
    },
    {
      "epoch": 0.11906834100045588,
      "grad_norm": 0.44895997643470764,
      "learning_rate": 4.40486551452609e-06,
      "loss": 0.7954,
      "step": 2873
    },
    {
      "epoch": 0.11910978490612956,
      "grad_norm": 0.4298035800457001,
      "learning_rate": 4.404658294997721e-06,
      "loss": 0.7396,
      "step": 2874
    },
    {
      "epoch": 0.11915122881180322,
      "grad_norm": 0.4431382715702057,
      "learning_rate": 4.404451075469352e-06,
      "loss": 0.7332,
      "step": 2875
    },
    {
      "epoch": 0.1191926727174769,
      "grad_norm": 0.4455166459083557,
      "learning_rate": 4.404243855940985e-06,
      "loss": 0.7067,
      "step": 2876
    },
    {
      "epoch": 0.11923411662315056,
      "grad_norm": 0.5093527436256409,
      "learning_rate": 4.404036636412616e-06,
      "loss": 0.7581,
      "step": 2877
    },
    {
      "epoch": 0.11927556052882424,
      "grad_norm": 0.43303173780441284,
      "learning_rate": 4.403829416884247e-06,
      "loss": 0.7734,
      "step": 2878
    },
    {
      "epoch": 0.1193170044344979,
      "grad_norm": 0.4100220203399658,
      "learning_rate": 4.403622197355879e-06,
      "loss": 0.7384,
      "step": 2879
    },
    {
      "epoch": 0.11935844834017158,
      "grad_norm": 0.436322957277298,
      "learning_rate": 4.403414977827511e-06,
      "loss": 0.7747,
      "step": 2880
    },
    {
      "epoch": 0.11939989224584525,
      "grad_norm": 0.4316721260547638,
      "learning_rate": 4.403207758299142e-06,
      "loss": 0.6942,
      "step": 2881
    },
    {
      "epoch": 0.11944133615151892,
      "grad_norm": 0.4586374759674072,
      "learning_rate": 4.403000538770774e-06,
      "loss": 0.886,
      "step": 2882
    },
    {
      "epoch": 0.1194827800571926,
      "grad_norm": 0.46106719970703125,
      "learning_rate": 4.402793319242405e-06,
      "loss": 0.7458,
      "step": 2883
    },
    {
      "epoch": 0.11952422396286626,
      "grad_norm": 0.46807432174682617,
      "learning_rate": 4.402586099714037e-06,
      "loss": 0.7629,
      "step": 2884
    },
    {
      "epoch": 0.11956566786853993,
      "grad_norm": 0.4275927245616913,
      "learning_rate": 4.40237888018567e-06,
      "loss": 0.705,
      "step": 2885
    },
    {
      "epoch": 0.1196071117742136,
      "grad_norm": 0.4402529299259186,
      "learning_rate": 4.402171660657301e-06,
      "loss": 0.7563,
      "step": 2886
    },
    {
      "epoch": 0.11964855567988728,
      "grad_norm": 0.43256986141204834,
      "learning_rate": 4.401964441128932e-06,
      "loss": 0.7546,
      "step": 2887
    },
    {
      "epoch": 0.11968999958556094,
      "grad_norm": 0.42073407769203186,
      "learning_rate": 4.401757221600564e-06,
      "loss": 0.7241,
      "step": 2888
    },
    {
      "epoch": 0.11973144349123462,
      "grad_norm": 0.4477077126502991,
      "learning_rate": 4.401550002072196e-06,
      "loss": 0.7366,
      "step": 2889
    },
    {
      "epoch": 0.11977288739690828,
      "grad_norm": 0.4193621873855591,
      "learning_rate": 4.401342782543827e-06,
      "loss": 0.6843,
      "step": 2890
    },
    {
      "epoch": 0.11981433130258196,
      "grad_norm": 0.4283328652381897,
      "learning_rate": 4.401135563015459e-06,
      "loss": 0.6659,
      "step": 2891
    },
    {
      "epoch": 0.11985577520825563,
      "grad_norm": 0.4206558167934418,
      "learning_rate": 4.400928343487091e-06,
      "loss": 0.7437,
      "step": 2892
    },
    {
      "epoch": 0.1198972191139293,
      "grad_norm": 0.4411073625087738,
      "learning_rate": 4.4007211239587224e-06,
      "loss": 0.7627,
      "step": 2893
    },
    {
      "epoch": 0.11993866301960297,
      "grad_norm": 0.44352224469184875,
      "learning_rate": 4.400513904430354e-06,
      "loss": 0.7817,
      "step": 2894
    },
    {
      "epoch": 0.11998010692527664,
      "grad_norm": 0.4222686290740967,
      "learning_rate": 4.400306684901985e-06,
      "loss": 0.7134,
      "step": 2895
    },
    {
      "epoch": 0.12002155083095031,
      "grad_norm": 0.45008957386016846,
      "learning_rate": 4.4000994653736174e-06,
      "loss": 0.71,
      "step": 2896
    },
    {
      "epoch": 0.12006299473662398,
      "grad_norm": 0.44572871923446655,
      "learning_rate": 4.399892245845249e-06,
      "loss": 0.73,
      "step": 2897
    },
    {
      "epoch": 0.12010443864229765,
      "grad_norm": 0.40540584921836853,
      "learning_rate": 4.39968502631688e-06,
      "loss": 0.7375,
      "step": 2898
    },
    {
      "epoch": 0.12014588254797132,
      "grad_norm": 0.4343511760234833,
      "learning_rate": 4.399477806788512e-06,
      "loss": 0.771,
      "step": 2899
    },
    {
      "epoch": 0.120187326453645,
      "grad_norm": 0.45251742005348206,
      "learning_rate": 4.399270587260144e-06,
      "loss": 0.7266,
      "step": 2900
    },
    {
      "epoch": 0.12022877035931866,
      "grad_norm": 0.4736974239349365,
      "learning_rate": 4.399063367731776e-06,
      "loss": 0.7695,
      "step": 2901
    },
    {
      "epoch": 0.12027021426499233,
      "grad_norm": 0.4555623233318329,
      "learning_rate": 4.398856148203407e-06,
      "loss": 0.7463,
      "step": 2902
    },
    {
      "epoch": 0.120311658170666,
      "grad_norm": 0.4468677043914795,
      "learning_rate": 4.398648928675039e-06,
      "loss": 0.7612,
      "step": 2903
    },
    {
      "epoch": 0.12035310207633967,
      "grad_norm": 0.3920592665672302,
      "learning_rate": 4.39844170914667e-06,
      "loss": 0.6816,
      "step": 2904
    },
    {
      "epoch": 0.12039454598201335,
      "grad_norm": 0.44974249601364136,
      "learning_rate": 4.3982344896183024e-06,
      "loss": 0.7266,
      "step": 2905
    },
    {
      "epoch": 0.12043598988768701,
      "grad_norm": 0.47260984778404236,
      "learning_rate": 4.398027270089934e-06,
      "loss": 0.7517,
      "step": 2906
    },
    {
      "epoch": 0.12047743379336069,
      "grad_norm": 0.40345868468284607,
      "learning_rate": 4.397820050561565e-06,
      "loss": 0.713,
      "step": 2907
    },
    {
      "epoch": 0.12051887769903435,
      "grad_norm": 0.449064165353775,
      "learning_rate": 4.397612831033197e-06,
      "loss": 0.7798,
      "step": 2908
    },
    {
      "epoch": 0.12056032160470803,
      "grad_norm": 0.44487282633781433,
      "learning_rate": 4.397405611504829e-06,
      "loss": 0.7124,
      "step": 2909
    },
    {
      "epoch": 0.1206017655103817,
      "grad_norm": 0.41603511571884155,
      "learning_rate": 4.39719839197646e-06,
      "loss": 0.748,
      "step": 2910
    },
    {
      "epoch": 0.12064320941605537,
      "grad_norm": 0.4312427043914795,
      "learning_rate": 4.396991172448092e-06,
      "loss": 0.6926,
      "step": 2911
    },
    {
      "epoch": 0.12068465332172904,
      "grad_norm": 0.4089362323284149,
      "learning_rate": 4.396783952919724e-06,
      "loss": 0.707,
      "step": 2912
    },
    {
      "epoch": 0.12072609722740271,
      "grad_norm": 0.44800814986228943,
      "learning_rate": 4.396576733391355e-06,
      "loss": 0.7507,
      "step": 2913
    },
    {
      "epoch": 0.12076754113307638,
      "grad_norm": 0.40577617287635803,
      "learning_rate": 4.396369513862987e-06,
      "loss": 0.7172,
      "step": 2914
    },
    {
      "epoch": 0.12080898503875005,
      "grad_norm": 0.4601323902606964,
      "learning_rate": 4.396162294334618e-06,
      "loss": 0.7559,
      "step": 2915
    },
    {
      "epoch": 0.12085042894442372,
      "grad_norm": 0.4524902105331421,
      "learning_rate": 4.39595507480625e-06,
      "loss": 0.7922,
      "step": 2916
    },
    {
      "epoch": 0.12089187285009739,
      "grad_norm": 0.4456711411476135,
      "learning_rate": 4.395747855277882e-06,
      "loss": 0.6899,
      "step": 2917
    },
    {
      "epoch": 0.12093331675577107,
      "grad_norm": 0.42603957653045654,
      "learning_rate": 4.395540635749513e-06,
      "loss": 0.7205,
      "step": 2918
    },
    {
      "epoch": 0.12097476066144473,
      "grad_norm": 0.45018836855888367,
      "learning_rate": 4.395333416221145e-06,
      "loss": 0.8123,
      "step": 2919
    },
    {
      "epoch": 0.12101620456711841,
      "grad_norm": 0.4259987473487854,
      "learning_rate": 4.395126196692777e-06,
      "loss": 0.7389,
      "step": 2920
    },
    {
      "epoch": 0.12105764847279207,
      "grad_norm": 0.42970335483551025,
      "learning_rate": 4.394918977164409e-06,
      "loss": 0.7197,
      "step": 2921
    },
    {
      "epoch": 0.12109909237846575,
      "grad_norm": 0.47988903522491455,
      "learning_rate": 4.39471175763604e-06,
      "loss": 0.7737,
      "step": 2922
    },
    {
      "epoch": 0.12114053628413941,
      "grad_norm": 0.4404818117618561,
      "learning_rate": 4.394504538107672e-06,
      "loss": 0.7515,
      "step": 2923
    },
    {
      "epoch": 0.12118198018981309,
      "grad_norm": 0.3940800428390503,
      "learning_rate": 4.394297318579303e-06,
      "loss": 0.75,
      "step": 2924
    },
    {
      "epoch": 0.12122342409548675,
      "grad_norm": 0.45118245482444763,
      "learning_rate": 4.394090099050935e-06,
      "loss": 0.7585,
      "step": 2925
    },
    {
      "epoch": 0.12126486800116043,
      "grad_norm": 0.4794446527957916,
      "learning_rate": 4.393882879522567e-06,
      "loss": 0.8066,
      "step": 2926
    },
    {
      "epoch": 0.1213063119068341,
      "grad_norm": 0.46741825342178345,
      "learning_rate": 4.393675659994198e-06,
      "loss": 0.7278,
      "step": 2927
    },
    {
      "epoch": 0.12134775581250777,
      "grad_norm": 0.444358766078949,
      "learning_rate": 4.39346844046583e-06,
      "loss": 0.7727,
      "step": 2928
    },
    {
      "epoch": 0.12138919971818145,
      "grad_norm": 0.45364388823509216,
      "learning_rate": 4.393261220937462e-06,
      "loss": 0.7749,
      "step": 2929
    },
    {
      "epoch": 0.12143064362385511,
      "grad_norm": 0.45744767785072327,
      "learning_rate": 4.393054001409093e-06,
      "loss": 0.771,
      "step": 2930
    },
    {
      "epoch": 0.12147208752952879,
      "grad_norm": 0.45419538021087646,
      "learning_rate": 4.392846781880724e-06,
      "loss": 0.7424,
      "step": 2931
    },
    {
      "epoch": 0.12151353143520245,
      "grad_norm": 0.449069619178772,
      "learning_rate": 4.392639562352357e-06,
      "loss": 0.7705,
      "step": 2932
    },
    {
      "epoch": 0.12155497534087613,
      "grad_norm": 0.41920971870422363,
      "learning_rate": 4.392432342823988e-06,
      "loss": 0.7261,
      "step": 2933
    },
    {
      "epoch": 0.12159641924654979,
      "grad_norm": 0.4316352307796478,
      "learning_rate": 4.392225123295619e-06,
      "loss": 0.7712,
      "step": 2934
    },
    {
      "epoch": 0.12163786315222347,
      "grad_norm": 0.5340210199356079,
      "learning_rate": 4.392017903767251e-06,
      "loss": 0.8035,
      "step": 2935
    },
    {
      "epoch": 0.12167930705789713,
      "grad_norm": 0.43614745140075684,
      "learning_rate": 4.391810684238883e-06,
      "loss": 0.7351,
      "step": 2936
    },
    {
      "epoch": 0.12172075096357081,
      "grad_norm": 0.4065016210079193,
      "learning_rate": 4.391603464710515e-06,
      "loss": 0.7637,
      "step": 2937
    },
    {
      "epoch": 0.12176219486924447,
      "grad_norm": 0.4912908971309662,
      "learning_rate": 4.391396245182147e-06,
      "loss": 0.7786,
      "step": 2938
    },
    {
      "epoch": 0.12180363877491815,
      "grad_norm": 0.4159591495990753,
      "learning_rate": 4.391189025653778e-06,
      "loss": 0.719,
      "step": 2939
    },
    {
      "epoch": 0.12184508268059181,
      "grad_norm": 0.47345712780952454,
      "learning_rate": 4.390981806125409e-06,
      "loss": 0.7397,
      "step": 2940
    },
    {
      "epoch": 0.12188652658626549,
      "grad_norm": 0.4322415292263031,
      "learning_rate": 4.390774586597042e-06,
      "loss": 0.74,
      "step": 2941
    },
    {
      "epoch": 0.12192797049193917,
      "grad_norm": 0.46097829937934875,
      "learning_rate": 4.390567367068673e-06,
      "loss": 0.8188,
      "step": 2942
    },
    {
      "epoch": 0.12196941439761283,
      "grad_norm": 0.4559336304664612,
      "learning_rate": 4.390360147540304e-06,
      "loss": 0.7959,
      "step": 2943
    },
    {
      "epoch": 0.1220108583032865,
      "grad_norm": 0.44493523240089417,
      "learning_rate": 4.390152928011936e-06,
      "loss": 0.7668,
      "step": 2944
    },
    {
      "epoch": 0.12205230220896017,
      "grad_norm": 0.4219712018966675,
      "learning_rate": 4.389945708483568e-06,
      "loss": 0.7839,
      "step": 2945
    },
    {
      "epoch": 0.12209374611463385,
      "grad_norm": 0.43804484605789185,
      "learning_rate": 4.389738488955199e-06,
      "loss": 0.7244,
      "step": 2946
    },
    {
      "epoch": 0.12213519002030751,
      "grad_norm": 0.4084610641002655,
      "learning_rate": 4.389531269426831e-06,
      "loss": 0.7383,
      "step": 2947
    },
    {
      "epoch": 0.12217663392598119,
      "grad_norm": 0.41283106803894043,
      "learning_rate": 4.389324049898463e-06,
      "loss": 0.731,
      "step": 2948
    },
    {
      "epoch": 0.12221807783165485,
      "grad_norm": 0.4139554798603058,
      "learning_rate": 4.389116830370094e-06,
      "loss": 0.6837,
      "step": 2949
    },
    {
      "epoch": 0.12225952173732853,
      "grad_norm": 0.4571232199668884,
      "learning_rate": 4.388909610841726e-06,
      "loss": 0.7466,
      "step": 2950
    },
    {
      "epoch": 0.12230096564300219,
      "grad_norm": 0.4528792202472687,
      "learning_rate": 4.388702391313357e-06,
      "loss": 0.8035,
      "step": 2951
    },
    {
      "epoch": 0.12234240954867587,
      "grad_norm": 0.4224989414215088,
      "learning_rate": 4.3884951717849894e-06,
      "loss": 0.7268,
      "step": 2952
    },
    {
      "epoch": 0.12238385345434953,
      "grad_norm": 0.41462570428848267,
      "learning_rate": 4.388287952256622e-06,
      "loss": 0.7725,
      "step": 2953
    },
    {
      "epoch": 0.12242529736002321,
      "grad_norm": 0.4372042119503021,
      "learning_rate": 4.388080732728253e-06,
      "loss": 0.7341,
      "step": 2954
    },
    {
      "epoch": 0.12246674126569689,
      "grad_norm": 0.4109656810760498,
      "learning_rate": 4.3878735131998844e-06,
      "loss": 0.6855,
      "step": 2955
    },
    {
      "epoch": 0.12250818517137055,
      "grad_norm": 0.4509662985801697,
      "learning_rate": 4.387666293671516e-06,
      "loss": 0.8091,
      "step": 2956
    },
    {
      "epoch": 0.12254962907704423,
      "grad_norm": 0.4340035021305084,
      "learning_rate": 4.387459074143148e-06,
      "loss": 0.7505,
      "step": 2957
    },
    {
      "epoch": 0.12259107298271789,
      "grad_norm": 0.43569880723953247,
      "learning_rate": 4.3872518546147794e-06,
      "loss": 0.7517,
      "step": 2958
    },
    {
      "epoch": 0.12263251688839157,
      "grad_norm": 0.4499976933002472,
      "learning_rate": 4.387044635086411e-06,
      "loss": 0.811,
      "step": 2959
    },
    {
      "epoch": 0.12267396079406523,
      "grad_norm": 0.43169182538986206,
      "learning_rate": 4.386837415558042e-06,
      "loss": 0.74,
      "step": 2960
    },
    {
      "epoch": 0.1227154046997389,
      "grad_norm": 0.4279778301715851,
      "learning_rate": 4.3866301960296744e-06,
      "loss": 0.7776,
      "step": 2961
    },
    {
      "epoch": 0.12275684860541257,
      "grad_norm": 0.41094255447387695,
      "learning_rate": 4.386422976501306e-06,
      "loss": 0.7539,
      "step": 2962
    },
    {
      "epoch": 0.12279829251108625,
      "grad_norm": 0.4356067180633545,
      "learning_rate": 4.386215756972937e-06,
      "loss": 0.7899,
      "step": 2963
    },
    {
      "epoch": 0.12283973641675991,
      "grad_norm": 0.41189250349998474,
      "learning_rate": 4.3860085374445694e-06,
      "loss": 0.7063,
      "step": 2964
    },
    {
      "epoch": 0.12288118032243359,
      "grad_norm": 0.44539976119995117,
      "learning_rate": 4.385801317916201e-06,
      "loss": 0.7643,
      "step": 2965
    },
    {
      "epoch": 0.12292262422810726,
      "grad_norm": 0.4432009160518646,
      "learning_rate": 4.385594098387832e-06,
      "loss": 0.772,
      "step": 2966
    },
    {
      "epoch": 0.12296406813378093,
      "grad_norm": 0.4932974576950073,
      "learning_rate": 4.385386878859464e-06,
      "loss": 0.7695,
      "step": 2967
    },
    {
      "epoch": 0.1230055120394546,
      "grad_norm": 0.42760759592056274,
      "learning_rate": 4.385179659331096e-06,
      "loss": 0.67,
      "step": 2968
    },
    {
      "epoch": 0.12304695594512827,
      "grad_norm": 0.4448728859424591,
      "learning_rate": 4.384972439802727e-06,
      "loss": 0.7644,
      "step": 2969
    },
    {
      "epoch": 0.12308839985080194,
      "grad_norm": 0.4384644329547882,
      "learning_rate": 4.384765220274359e-06,
      "loss": 0.7247,
      "step": 2970
    },
    {
      "epoch": 0.12312984375647561,
      "grad_norm": 0.4520401358604431,
      "learning_rate": 4.384558000745991e-06,
      "loss": 0.7673,
      "step": 2971
    },
    {
      "epoch": 0.12317128766214928,
      "grad_norm": 0.45528051257133484,
      "learning_rate": 4.384350781217622e-06,
      "loss": 0.7649,
      "step": 2972
    },
    {
      "epoch": 0.12321273156782295,
      "grad_norm": 0.49318763613700867,
      "learning_rate": 4.3841435616892545e-06,
      "loss": 0.761,
      "step": 2973
    },
    {
      "epoch": 0.12325417547349662,
      "grad_norm": 0.4304576516151428,
      "learning_rate": 4.383936342160886e-06,
      "loss": 0.752,
      "step": 2974
    },
    {
      "epoch": 0.12329561937917029,
      "grad_norm": 0.4300905466079712,
      "learning_rate": 4.383729122632517e-06,
      "loss": 0.7407,
      "step": 2975
    },
    {
      "epoch": 0.12333706328484396,
      "grad_norm": 0.4694576561450958,
      "learning_rate": 4.383521903104149e-06,
      "loss": 0.7551,
      "step": 2976
    },
    {
      "epoch": 0.12337850719051763,
      "grad_norm": 0.42495089769363403,
      "learning_rate": 4.383314683575781e-06,
      "loss": 0.783,
      "step": 2977
    },
    {
      "epoch": 0.1234199510961913,
      "grad_norm": 0.4327869415283203,
      "learning_rate": 4.383107464047412e-06,
      "loss": 0.7305,
      "step": 2978
    },
    {
      "epoch": 0.12346139500186498,
      "grad_norm": 0.4472733438014984,
      "learning_rate": 4.382900244519044e-06,
      "loss": 0.7358,
      "step": 2979
    },
    {
      "epoch": 0.12350283890753865,
      "grad_norm": 0.48325395584106445,
      "learning_rate": 4.382693024990676e-06,
      "loss": 0.7482,
      "step": 2980
    },
    {
      "epoch": 0.12354428281321232,
      "grad_norm": 0.5064083933830261,
      "learning_rate": 4.382485805462307e-06,
      "loss": 0.7468,
      "step": 2981
    },
    {
      "epoch": 0.12358572671888599,
      "grad_norm": 0.4461282193660736,
      "learning_rate": 4.382278585933939e-06,
      "loss": 0.7247,
      "step": 2982
    },
    {
      "epoch": 0.12362717062455966,
      "grad_norm": 0.4487856924533844,
      "learning_rate": 4.38207136640557e-06,
      "loss": 0.7207,
      "step": 2983
    },
    {
      "epoch": 0.12366861453023333,
      "grad_norm": 0.41898149251937866,
      "learning_rate": 4.381864146877202e-06,
      "loss": 0.7168,
      "step": 2984
    },
    {
      "epoch": 0.123710058435907,
      "grad_norm": 0.5153146982192993,
      "learning_rate": 4.381656927348834e-06,
      "loss": 0.8486,
      "step": 2985
    },
    {
      "epoch": 0.12375150234158067,
      "grad_norm": 0.4400433599948883,
      "learning_rate": 4.381449707820465e-06,
      "loss": 0.7878,
      "step": 2986
    },
    {
      "epoch": 0.12379294624725434,
      "grad_norm": 0.4535733461380005,
      "learning_rate": 4.381242488292096e-06,
      "loss": 0.7603,
      "step": 2987
    },
    {
      "epoch": 0.123834390152928,
      "grad_norm": 0.44199758768081665,
      "learning_rate": 4.381035268763729e-06,
      "loss": 0.7864,
      "step": 2988
    },
    {
      "epoch": 0.12387583405860168,
      "grad_norm": 0.43000081181526184,
      "learning_rate": 4.380828049235361e-06,
      "loss": 0.7812,
      "step": 2989
    },
    {
      "epoch": 0.12391727796427536,
      "grad_norm": 0.434529572725296,
      "learning_rate": 4.380620829706992e-06,
      "loss": 0.7559,
      "step": 2990
    },
    {
      "epoch": 0.12395872186994902,
      "grad_norm": 0.39960378408432007,
      "learning_rate": 4.380413610178624e-06,
      "loss": 0.7512,
      "step": 2991
    },
    {
      "epoch": 0.1240001657756227,
      "grad_norm": 0.46752679347991943,
      "learning_rate": 4.380206390650255e-06,
      "loss": 0.7847,
      "step": 2992
    },
    {
      "epoch": 0.12404160968129636,
      "grad_norm": 0.44536927342414856,
      "learning_rate": 4.379999171121887e-06,
      "loss": 0.7485,
      "step": 2993
    },
    {
      "epoch": 0.12408305358697004,
      "grad_norm": 0.46931543946266174,
      "learning_rate": 4.379791951593519e-06,
      "loss": 0.7788,
      "step": 2994
    },
    {
      "epoch": 0.1241244974926437,
      "grad_norm": 0.450585275888443,
      "learning_rate": 4.37958473206515e-06,
      "loss": 0.7852,
      "step": 2995
    },
    {
      "epoch": 0.12416594139831738,
      "grad_norm": 0.4099261164665222,
      "learning_rate": 4.379377512536781e-06,
      "loss": 0.7432,
      "step": 2996
    },
    {
      "epoch": 0.12420738530399104,
      "grad_norm": 0.42129647731781006,
      "learning_rate": 4.379170293008414e-06,
      "loss": 0.7612,
      "step": 2997
    },
    {
      "epoch": 0.12424882920966472,
      "grad_norm": 0.4596455991268158,
      "learning_rate": 4.378963073480045e-06,
      "loss": 0.7452,
      "step": 2998
    },
    {
      "epoch": 0.12429027311533838,
      "grad_norm": 0.4385070502758026,
      "learning_rate": 4.378755853951676e-06,
      "loss": 0.6667,
      "step": 2999
    },
    {
      "epoch": 0.12433171702101206,
      "grad_norm": 0.4641198515892029,
      "learning_rate": 4.378548634423309e-06,
      "loss": 0.7365,
      "step": 3000
    },
    {
      "epoch": 0.12437316092668572,
      "grad_norm": 0.4399277865886688,
      "learning_rate": 4.37834141489494e-06,
      "loss": 0.738,
      "step": 3001
    },
    {
      "epoch": 0.1244146048323594,
      "grad_norm": 0.44207197427749634,
      "learning_rate": 4.378134195366571e-06,
      "loss": 0.8035,
      "step": 3002
    },
    {
      "epoch": 0.12445604873803308,
      "grad_norm": 0.4337393641471863,
      "learning_rate": 4.377926975838203e-06,
      "loss": 0.7451,
      "step": 3003
    },
    {
      "epoch": 0.12449749264370674,
      "grad_norm": 0.4571448564529419,
      "learning_rate": 4.377719756309835e-06,
      "loss": 0.7678,
      "step": 3004
    },
    {
      "epoch": 0.12453893654938042,
      "grad_norm": 0.49137502908706665,
      "learning_rate": 4.377512536781466e-06,
      "loss": 0.8188,
      "step": 3005
    },
    {
      "epoch": 0.12458038045505408,
      "grad_norm": 0.44722428917884827,
      "learning_rate": 4.377305317253099e-06,
      "loss": 0.7583,
      "step": 3006
    },
    {
      "epoch": 0.12462182436072776,
      "grad_norm": 0.4372878968715668,
      "learning_rate": 4.37709809772473e-06,
      "loss": 0.7061,
      "step": 3007
    },
    {
      "epoch": 0.12466326826640142,
      "grad_norm": 0.3905053436756134,
      "learning_rate": 4.376890878196361e-06,
      "loss": 0.6547,
      "step": 3008
    },
    {
      "epoch": 0.1247047121720751,
      "grad_norm": 0.4260025918483734,
      "learning_rate": 4.376683658667994e-06,
      "loss": 0.7659,
      "step": 3009
    },
    {
      "epoch": 0.12474615607774876,
      "grad_norm": 0.4143962562084198,
      "learning_rate": 4.376476439139625e-06,
      "loss": 0.7529,
      "step": 3010
    },
    {
      "epoch": 0.12478759998342244,
      "grad_norm": 0.4323735535144806,
      "learning_rate": 4.3762692196112564e-06,
      "loss": 0.7195,
      "step": 3011
    },
    {
      "epoch": 0.1248290438890961,
      "grad_norm": 0.42522549629211426,
      "learning_rate": 4.376062000082888e-06,
      "loss": 0.7368,
      "step": 3012
    },
    {
      "epoch": 0.12487048779476978,
      "grad_norm": 0.4337555170059204,
      "learning_rate": 4.37585478055452e-06,
      "loss": 0.7361,
      "step": 3013
    },
    {
      "epoch": 0.12491193170044344,
      "grad_norm": 0.4288124740123749,
      "learning_rate": 4.3756475610261514e-06,
      "loss": 0.7161,
      "step": 3014
    },
    {
      "epoch": 0.12495337560611712,
      "grad_norm": 0.4593937397003174,
      "learning_rate": 4.375440341497783e-06,
      "loss": 0.7607,
      "step": 3015
    },
    {
      "epoch": 0.1249948195117908,
      "grad_norm": 0.4478203058242798,
      "learning_rate": 4.375233121969415e-06,
      "loss": 0.7432,
      "step": 3016
    },
    {
      "epoch": 0.12503626341746446,
      "grad_norm": 0.43422773480415344,
      "learning_rate": 4.3750259024410464e-06,
      "loss": 0.7393,
      "step": 3017
    },
    {
      "epoch": 0.12507770732313814,
      "grad_norm": 0.4143618047237396,
      "learning_rate": 4.374818682912678e-06,
      "loss": 0.7397,
      "step": 3018
    },
    {
      "epoch": 0.12511915122881181,
      "grad_norm": 0.4719943404197693,
      "learning_rate": 4.374611463384309e-06,
      "loss": 0.7642,
      "step": 3019
    },
    {
      "epoch": 0.12516059513448546,
      "grad_norm": 0.4293724000453949,
      "learning_rate": 4.3744042438559414e-06,
      "loss": 0.78,
      "step": 3020
    },
    {
      "epoch": 0.12520203904015914,
      "grad_norm": 0.4163784980773926,
      "learning_rate": 4.374197024327573e-06,
      "loss": 0.7156,
      "step": 3021
    },
    {
      "epoch": 0.12524348294583282,
      "grad_norm": 0.4988276958465576,
      "learning_rate": 4.373989804799205e-06,
      "loss": 0.7073,
      "step": 3022
    },
    {
      "epoch": 0.1252849268515065,
      "grad_norm": 0.4419735074043274,
      "learning_rate": 4.3737825852708364e-06,
      "loss": 0.795,
      "step": 3023
    },
    {
      "epoch": 0.12532637075718014,
      "grad_norm": 0.4776926338672638,
      "learning_rate": 4.373575365742468e-06,
      "loss": 0.7587,
      "step": 3024
    },
    {
      "epoch": 0.12536781466285382,
      "grad_norm": 0.4149540364742279,
      "learning_rate": 4.3733681462141e-06,
      "loss": 0.7532,
      "step": 3025
    },
    {
      "epoch": 0.1254092585685275,
      "grad_norm": 0.3856103718280792,
      "learning_rate": 4.3731609266857314e-06,
      "loss": 0.7129,
      "step": 3026
    },
    {
      "epoch": 0.12545070247420118,
      "grad_norm": 0.4375776946544647,
      "learning_rate": 4.372953707157363e-06,
      "loss": 0.7985,
      "step": 3027
    },
    {
      "epoch": 0.12549214637987485,
      "grad_norm": 0.4199005365371704,
      "learning_rate": 4.372746487628994e-06,
      "loss": 0.7688,
      "step": 3028
    },
    {
      "epoch": 0.1255335902855485,
      "grad_norm": 0.44858938455581665,
      "learning_rate": 4.3725392681006265e-06,
      "loss": 0.7104,
      "step": 3029
    },
    {
      "epoch": 0.12557503419122218,
      "grad_norm": 0.44958579540252686,
      "learning_rate": 4.372332048572258e-06,
      "loss": 0.7324,
      "step": 3030
    },
    {
      "epoch": 0.12561647809689586,
      "grad_norm": 0.45155689120292664,
      "learning_rate": 4.372124829043889e-06,
      "loss": 0.7001,
      "step": 3031
    },
    {
      "epoch": 0.12565792200256953,
      "grad_norm": 0.490447461605072,
      "learning_rate": 4.3719176095155215e-06,
      "loss": 0.8486,
      "step": 3032
    },
    {
      "epoch": 0.12569936590824318,
      "grad_norm": 0.4496222734451294,
      "learning_rate": 4.371710389987153e-06,
      "loss": 0.7847,
      "step": 3033
    },
    {
      "epoch": 0.12574080981391686,
      "grad_norm": 0.4434768855571747,
      "learning_rate": 4.371503170458784e-06,
      "loss": 0.7484,
      "step": 3034
    },
    {
      "epoch": 0.12578225371959054,
      "grad_norm": 0.39802369475364685,
      "learning_rate": 4.371295950930416e-06,
      "loss": 0.7393,
      "step": 3035
    },
    {
      "epoch": 0.1258236976252642,
      "grad_norm": 0.4425632357597351,
      "learning_rate": 4.371088731402048e-06,
      "loss": 0.7412,
      "step": 3036
    },
    {
      "epoch": 0.12586514153093786,
      "grad_norm": 0.41171756386756897,
      "learning_rate": 4.370881511873679e-06,
      "loss": 0.718,
      "step": 3037
    },
    {
      "epoch": 0.12590658543661154,
      "grad_norm": 0.43169206380844116,
      "learning_rate": 4.3706742923453115e-06,
      "loss": 0.7114,
      "step": 3038
    },
    {
      "epoch": 0.12594802934228522,
      "grad_norm": 0.4423796534538269,
      "learning_rate": 4.370467072816942e-06,
      "loss": 0.8245,
      "step": 3039
    },
    {
      "epoch": 0.1259894732479589,
      "grad_norm": 0.45997801423072815,
      "learning_rate": 4.370259853288574e-06,
      "loss": 0.8053,
      "step": 3040
    },
    {
      "epoch": 0.12603091715363257,
      "grad_norm": 0.41697803139686584,
      "learning_rate": 4.3700526337602065e-06,
      "loss": 0.7527,
      "step": 3041
    },
    {
      "epoch": 0.12607236105930622,
      "grad_norm": 0.41332074999809265,
      "learning_rate": 4.369845414231838e-06,
      "loss": 0.7273,
      "step": 3042
    },
    {
      "epoch": 0.1261138049649799,
      "grad_norm": 0.44044753909111023,
      "learning_rate": 4.369638194703469e-06,
      "loss": 0.7289,
      "step": 3043
    },
    {
      "epoch": 0.12615524887065357,
      "grad_norm": 0.43636149168014526,
      "learning_rate": 4.369430975175101e-06,
      "loss": 0.7742,
      "step": 3044
    },
    {
      "epoch": 0.12619669277632725,
      "grad_norm": 0.4411197602748871,
      "learning_rate": 4.369223755646733e-06,
      "loss": 0.748,
      "step": 3045
    },
    {
      "epoch": 0.1262381366820009,
      "grad_norm": 0.4383072853088379,
      "learning_rate": 4.369016536118364e-06,
      "loss": 0.7751,
      "step": 3046
    },
    {
      "epoch": 0.12627958058767458,
      "grad_norm": 0.4440144896507263,
      "learning_rate": 4.368809316589996e-06,
      "loss": 0.7419,
      "step": 3047
    },
    {
      "epoch": 0.12632102449334826,
      "grad_norm": 0.4993993937969208,
      "learning_rate": 4.368602097061627e-06,
      "loss": 0.7664,
      "step": 3048
    },
    {
      "epoch": 0.12636246839902193,
      "grad_norm": 0.42858147621154785,
      "learning_rate": 4.368394877533259e-06,
      "loss": 0.7562,
      "step": 3049
    },
    {
      "epoch": 0.12640391230469558,
      "grad_norm": 0.4735046327114105,
      "learning_rate": 4.368187658004891e-06,
      "loss": 0.772,
      "step": 3050
    },
    {
      "epoch": 0.12644535621036926,
      "grad_norm": 0.48770636320114136,
      "learning_rate": 4.367980438476522e-06,
      "loss": 0.7937,
      "step": 3051
    },
    {
      "epoch": 0.12648680011604294,
      "grad_norm": 0.3792746365070343,
      "learning_rate": 4.367773218948154e-06,
      "loss": 0.6726,
      "step": 3052
    },
    {
      "epoch": 0.1265282440217166,
      "grad_norm": 0.46097612380981445,
      "learning_rate": 4.367565999419786e-06,
      "loss": 0.7332,
      "step": 3053
    },
    {
      "epoch": 0.1265696879273903,
      "grad_norm": 0.40957921743392944,
      "learning_rate": 4.367358779891417e-06,
      "loss": 0.6927,
      "step": 3054
    },
    {
      "epoch": 0.12661113183306394,
      "grad_norm": 0.45638734102249146,
      "learning_rate": 4.367151560363048e-06,
      "loss": 0.7284,
      "step": 3055
    },
    {
      "epoch": 0.12665257573873762,
      "grad_norm": 0.5011829733848572,
      "learning_rate": 4.366944340834681e-06,
      "loss": 0.7761,
      "step": 3056
    },
    {
      "epoch": 0.1266940196444113,
      "grad_norm": 0.387362539768219,
      "learning_rate": 4.366737121306312e-06,
      "loss": 0.7012,
      "step": 3057
    },
    {
      "epoch": 0.12673546355008497,
      "grad_norm": 0.4381357729434967,
      "learning_rate": 4.366529901777944e-06,
      "loss": 0.7415,
      "step": 3058
    },
    {
      "epoch": 0.12677690745575862,
      "grad_norm": 0.44601863622665405,
      "learning_rate": 4.366322682249576e-06,
      "loss": 0.7715,
      "step": 3059
    },
    {
      "epoch": 0.1268183513614323,
      "grad_norm": 0.43791520595550537,
      "learning_rate": 4.366115462721207e-06,
      "loss": 0.7739,
      "step": 3060
    },
    {
      "epoch": 0.12685979526710597,
      "grad_norm": 0.4443438947200775,
      "learning_rate": 4.365908243192839e-06,
      "loss": 0.771,
      "step": 3061
    },
    {
      "epoch": 0.12690123917277965,
      "grad_norm": 0.4737672507762909,
      "learning_rate": 4.365701023664471e-06,
      "loss": 0.7205,
      "step": 3062
    },
    {
      "epoch": 0.1269426830784533,
      "grad_norm": 0.42101678252220154,
      "learning_rate": 4.365493804136102e-06,
      "loss": 0.7632,
      "step": 3063
    },
    {
      "epoch": 0.12698412698412698,
      "grad_norm": 0.43539363145828247,
      "learning_rate": 4.365286584607733e-06,
      "loss": 0.7179,
      "step": 3064
    },
    {
      "epoch": 0.12702557088980065,
      "grad_norm": 0.43787074089050293,
      "learning_rate": 4.365079365079366e-06,
      "loss": 0.7397,
      "step": 3065
    },
    {
      "epoch": 0.12706701479547433,
      "grad_norm": 0.4471369683742523,
      "learning_rate": 4.364872145550997e-06,
      "loss": 0.7703,
      "step": 3066
    },
    {
      "epoch": 0.127108458701148,
      "grad_norm": 0.43062305450439453,
      "learning_rate": 4.364664926022628e-06,
      "loss": 0.7574,
      "step": 3067
    },
    {
      "epoch": 0.12714990260682166,
      "grad_norm": 0.44166362285614014,
      "learning_rate": 4.364457706494261e-06,
      "loss": 0.7693,
      "step": 3068
    },
    {
      "epoch": 0.12719134651249533,
      "grad_norm": 0.43321236968040466,
      "learning_rate": 4.364250486965892e-06,
      "loss": 0.7594,
      "step": 3069
    },
    {
      "epoch": 0.127232790418169,
      "grad_norm": 0.4487900733947754,
      "learning_rate": 4.3640432674375234e-06,
      "loss": 0.7559,
      "step": 3070
    },
    {
      "epoch": 0.1272742343238427,
      "grad_norm": 0.4515455365180969,
      "learning_rate": 4.363836047909155e-06,
      "loss": 0.7415,
      "step": 3071
    },
    {
      "epoch": 0.12731567822951634,
      "grad_norm": 0.4024031460285187,
      "learning_rate": 4.363628828380787e-06,
      "loss": 0.7407,
      "step": 3072
    },
    {
      "epoch": 0.12735712213519002,
      "grad_norm": 0.528317391872406,
      "learning_rate": 4.3634216088524184e-06,
      "loss": 0.7402,
      "step": 3073
    },
    {
      "epoch": 0.1273985660408637,
      "grad_norm": 0.44311344623565674,
      "learning_rate": 4.363214389324051e-06,
      "loss": 0.781,
      "step": 3074
    },
    {
      "epoch": 0.12744000994653737,
      "grad_norm": 0.4543335735797882,
      "learning_rate": 4.363007169795682e-06,
      "loss": 0.8591,
      "step": 3075
    },
    {
      "epoch": 0.12748145385221102,
      "grad_norm": 0.39681488275527954,
      "learning_rate": 4.3627999502673134e-06,
      "loss": 0.7146,
      "step": 3076
    },
    {
      "epoch": 0.1275228977578847,
      "grad_norm": 0.4308423101902008,
      "learning_rate": 4.362592730738946e-06,
      "loss": 0.7194,
      "step": 3077
    },
    {
      "epoch": 0.12756434166355837,
      "grad_norm": 0.3952022194862366,
      "learning_rate": 4.362385511210577e-06,
      "loss": 0.7471,
      "step": 3078
    },
    {
      "epoch": 0.12760578556923205,
      "grad_norm": 0.442682147026062,
      "learning_rate": 4.3621782916822084e-06,
      "loss": 0.717,
      "step": 3079
    },
    {
      "epoch": 0.12764722947490573,
      "grad_norm": 0.4454176127910614,
      "learning_rate": 4.36197107215384e-06,
      "loss": 0.7273,
      "step": 3080
    },
    {
      "epoch": 0.12768867338057938,
      "grad_norm": 0.39107218384742737,
      "learning_rate": 4.361763852625472e-06,
      "loss": 0.707,
      "step": 3081
    },
    {
      "epoch": 0.12773011728625305,
      "grad_norm": 0.44460615515708923,
      "learning_rate": 4.3615566330971034e-06,
      "loss": 0.7937,
      "step": 3082
    },
    {
      "epoch": 0.12777156119192673,
      "grad_norm": 0.44838109612464905,
      "learning_rate": 4.361349413568735e-06,
      "loss": 0.7693,
      "step": 3083
    },
    {
      "epoch": 0.1278130050976004,
      "grad_norm": 0.4448307156562805,
      "learning_rate": 4.361142194040367e-06,
      "loss": 0.801,
      "step": 3084
    },
    {
      "epoch": 0.12785444900327406,
      "grad_norm": 0.4295814633369446,
      "learning_rate": 4.3609349745119984e-06,
      "loss": 0.7084,
      "step": 3085
    },
    {
      "epoch": 0.12789589290894773,
      "grad_norm": 0.45000559091567993,
      "learning_rate": 4.36072775498363e-06,
      "loss": 0.7268,
      "step": 3086
    },
    {
      "epoch": 0.1279373368146214,
      "grad_norm": 0.41829347610473633,
      "learning_rate": 4.360520535455261e-06,
      "loss": 0.762,
      "step": 3087
    },
    {
      "epoch": 0.1279787807202951,
      "grad_norm": 0.4617316722869873,
      "learning_rate": 4.3603133159268935e-06,
      "loss": 0.7268,
      "step": 3088
    },
    {
      "epoch": 0.12802022462596876,
      "grad_norm": 0.4162595868110657,
      "learning_rate": 4.360106096398525e-06,
      "loss": 0.7449,
      "step": 3089
    },
    {
      "epoch": 0.12806166853164241,
      "grad_norm": 0.4488827884197235,
      "learning_rate": 4.359898876870157e-06,
      "loss": 0.7729,
      "step": 3090
    },
    {
      "epoch": 0.1281031124373161,
      "grad_norm": 0.4416542053222656,
      "learning_rate": 4.359691657341788e-06,
      "loss": 0.7539,
      "step": 3091
    },
    {
      "epoch": 0.12814455634298977,
      "grad_norm": 0.40357670187950134,
      "learning_rate": 4.35948443781342e-06,
      "loss": 0.6993,
      "step": 3092
    },
    {
      "epoch": 0.12818600024866345,
      "grad_norm": 0.39695340394973755,
      "learning_rate": 4.359277218285052e-06,
      "loss": 0.6995,
      "step": 3093
    },
    {
      "epoch": 0.1282274441543371,
      "grad_norm": 0.40370526909828186,
      "learning_rate": 4.3590699987566835e-06,
      "loss": 0.7346,
      "step": 3094
    },
    {
      "epoch": 0.12826888806001077,
      "grad_norm": 0.44125548005104065,
      "learning_rate": 4.358862779228315e-06,
      "loss": 0.6948,
      "step": 3095
    },
    {
      "epoch": 0.12831033196568445,
      "grad_norm": 0.4347740411758423,
      "learning_rate": 4.358655559699946e-06,
      "loss": 0.7532,
      "step": 3096
    },
    {
      "epoch": 0.12835177587135813,
      "grad_norm": 0.4402252733707428,
      "learning_rate": 4.3584483401715785e-06,
      "loss": 0.7091,
      "step": 3097
    },
    {
      "epoch": 0.12839321977703178,
      "grad_norm": 0.4315100312232971,
      "learning_rate": 4.35824112064321e-06,
      "loss": 0.7283,
      "step": 3098
    },
    {
      "epoch": 0.12843466368270545,
      "grad_norm": 0.4290374517440796,
      "learning_rate": 4.358033901114841e-06,
      "loss": 0.734,
      "step": 3099
    },
    {
      "epoch": 0.12847610758837913,
      "grad_norm": 0.48330551385879517,
      "learning_rate": 4.357826681586473e-06,
      "loss": 0.7529,
      "step": 3100
    },
    {
      "epoch": 0.1285175514940528,
      "grad_norm": 0.41826382279396057,
      "learning_rate": 4.357619462058105e-06,
      "loss": 0.7124,
      "step": 3101
    },
    {
      "epoch": 0.12855899539972648,
      "grad_norm": 0.46399909257888794,
      "learning_rate": 4.357412242529736e-06,
      "loss": 0.7466,
      "step": 3102
    },
    {
      "epoch": 0.12860043930540013,
      "grad_norm": 0.4514673054218292,
      "learning_rate": 4.357205023001368e-06,
      "loss": 0.6809,
      "step": 3103
    },
    {
      "epoch": 0.1286418832110738,
      "grad_norm": 0.45701003074645996,
      "learning_rate": 4.356997803473e-06,
      "loss": 0.8027,
      "step": 3104
    },
    {
      "epoch": 0.1286833271167475,
      "grad_norm": 0.3961890637874603,
      "learning_rate": 4.356790583944631e-06,
      "loss": 0.7341,
      "step": 3105
    },
    {
      "epoch": 0.12872477102242116,
      "grad_norm": 0.4148775041103363,
      "learning_rate": 4.3565833644162635e-06,
      "loss": 0.7786,
      "step": 3106
    },
    {
      "epoch": 0.1287662149280948,
      "grad_norm": 0.445646196603775,
      "learning_rate": 4.356376144887894e-06,
      "loss": 0.7068,
      "step": 3107
    },
    {
      "epoch": 0.1288076588337685,
      "grad_norm": 0.42004355788230896,
      "learning_rate": 4.356168925359526e-06,
      "loss": 0.7711,
      "step": 3108
    },
    {
      "epoch": 0.12884910273944217,
      "grad_norm": 0.41880160570144653,
      "learning_rate": 4.355961705831158e-06,
      "loss": 0.6787,
      "step": 3109
    },
    {
      "epoch": 0.12889054664511584,
      "grad_norm": 0.46185821294784546,
      "learning_rate": 4.35575448630279e-06,
      "loss": 0.7668,
      "step": 3110
    },
    {
      "epoch": 0.1289319905507895,
      "grad_norm": 0.43727779388427734,
      "learning_rate": 4.355547266774421e-06,
      "loss": 0.7405,
      "step": 3111
    },
    {
      "epoch": 0.12897343445646317,
      "grad_norm": 0.40684273838996887,
      "learning_rate": 4.355340047246053e-06,
      "loss": 0.731,
      "step": 3112
    },
    {
      "epoch": 0.12901487836213685,
      "grad_norm": 0.44705119729042053,
      "learning_rate": 4.355132827717685e-06,
      "loss": 0.6917,
      "step": 3113
    },
    {
      "epoch": 0.12905632226781052,
      "grad_norm": 0.4360480010509491,
      "learning_rate": 4.354925608189316e-06,
      "loss": 0.781,
      "step": 3114
    },
    {
      "epoch": 0.1290977661734842,
      "grad_norm": 0.4306013584136963,
      "learning_rate": 4.354718388660948e-06,
      "loss": 0.7815,
      "step": 3115
    },
    {
      "epoch": 0.12913921007915785,
      "grad_norm": 0.4243769347667694,
      "learning_rate": 4.354511169132579e-06,
      "loss": 0.7509,
      "step": 3116
    },
    {
      "epoch": 0.12918065398483153,
      "grad_norm": 0.4535301625728607,
      "learning_rate": 4.354303949604211e-06,
      "loss": 0.6791,
      "step": 3117
    },
    {
      "epoch": 0.1292220978905052,
      "grad_norm": 0.4242030680179596,
      "learning_rate": 4.354096730075843e-06,
      "loss": 0.7239,
      "step": 3118
    },
    {
      "epoch": 0.12926354179617888,
      "grad_norm": 0.4205408990383148,
      "learning_rate": 4.353889510547474e-06,
      "loss": 0.7393,
      "step": 3119
    },
    {
      "epoch": 0.12930498570185253,
      "grad_norm": 0.4527781307697296,
      "learning_rate": 4.353682291019106e-06,
      "loss": 0.7505,
      "step": 3120
    },
    {
      "epoch": 0.1293464296075262,
      "grad_norm": 0.4405979514122009,
      "learning_rate": 4.353475071490738e-06,
      "loss": 0.74,
      "step": 3121
    },
    {
      "epoch": 0.12938787351319989,
      "grad_norm": 0.42699339985847473,
      "learning_rate": 4.353267851962369e-06,
      "loss": 0.7524,
      "step": 3122
    },
    {
      "epoch": 0.12942931741887356,
      "grad_norm": 0.43927860260009766,
      "learning_rate": 4.353060632434e-06,
      "loss": 0.6995,
      "step": 3123
    },
    {
      "epoch": 0.1294707613245472,
      "grad_norm": 0.4340886175632477,
      "learning_rate": 4.352853412905633e-06,
      "loss": 0.7976,
      "step": 3124
    },
    {
      "epoch": 0.1295122052302209,
      "grad_norm": 0.4235353171825409,
      "learning_rate": 4.352646193377264e-06,
      "loss": 0.7427,
      "step": 3125
    },
    {
      "epoch": 0.12955364913589457,
      "grad_norm": 0.4078538417816162,
      "learning_rate": 4.352438973848896e-06,
      "loss": 0.7251,
      "step": 3126
    },
    {
      "epoch": 0.12959509304156824,
      "grad_norm": 0.417101114988327,
      "learning_rate": 4.352231754320528e-06,
      "loss": 0.7047,
      "step": 3127
    },
    {
      "epoch": 0.12963653694724192,
      "grad_norm": 0.4423764944076538,
      "learning_rate": 4.352024534792159e-06,
      "loss": 0.7605,
      "step": 3128
    },
    {
      "epoch": 0.12967798085291557,
      "grad_norm": 0.41999706625938416,
      "learning_rate": 4.351817315263791e-06,
      "loss": 0.748,
      "step": 3129
    },
    {
      "epoch": 0.12971942475858925,
      "grad_norm": 0.3999955952167511,
      "learning_rate": 4.351610095735423e-06,
      "loss": 0.7222,
      "step": 3130
    },
    {
      "epoch": 0.12976086866426292,
      "grad_norm": 0.42091724276542664,
      "learning_rate": 4.351402876207054e-06,
      "loss": 0.7526,
      "step": 3131
    },
    {
      "epoch": 0.1298023125699366,
      "grad_norm": 0.44426479935646057,
      "learning_rate": 4.3511956566786854e-06,
      "loss": 0.7349,
      "step": 3132
    },
    {
      "epoch": 0.12984375647561025,
      "grad_norm": 0.45102280378341675,
      "learning_rate": 4.350988437150318e-06,
      "loss": 0.7507,
      "step": 3133
    },
    {
      "epoch": 0.12988520038128393,
      "grad_norm": 0.4707501232624054,
      "learning_rate": 4.350781217621949e-06,
      "loss": 0.7944,
      "step": 3134
    },
    {
      "epoch": 0.1299266442869576,
      "grad_norm": 0.4298677444458008,
      "learning_rate": 4.3505739980935804e-06,
      "loss": 0.7773,
      "step": 3135
    },
    {
      "epoch": 0.12996808819263128,
      "grad_norm": 0.47007015347480774,
      "learning_rate": 4.350366778565213e-06,
      "loss": 0.8071,
      "step": 3136
    },
    {
      "epoch": 0.13000953209830493,
      "grad_norm": 0.43346676230430603,
      "learning_rate": 4.350159559036844e-06,
      "loss": 0.771,
      "step": 3137
    },
    {
      "epoch": 0.1300509760039786,
      "grad_norm": 0.43266457319259644,
      "learning_rate": 4.3499523395084754e-06,
      "loss": 0.7354,
      "step": 3138
    },
    {
      "epoch": 0.13009241990965228,
      "grad_norm": 0.4574453830718994,
      "learning_rate": 4.349745119980107e-06,
      "loss": 0.8176,
      "step": 3139
    },
    {
      "epoch": 0.13013386381532596,
      "grad_norm": 0.4442731440067291,
      "learning_rate": 4.349537900451739e-06,
      "loss": 0.7595,
      "step": 3140
    },
    {
      "epoch": 0.13017530772099964,
      "grad_norm": 0.4240463376045227,
      "learning_rate": 4.3493306809233704e-06,
      "loss": 0.7268,
      "step": 3141
    },
    {
      "epoch": 0.1302167516266733,
      "grad_norm": 0.4719716012477875,
      "learning_rate": 4.349123461395003e-06,
      "loss": 0.7417,
      "step": 3142
    },
    {
      "epoch": 0.13025819553234697,
      "grad_norm": 0.43604332208633423,
      "learning_rate": 4.348916241866633e-06,
      "loss": 0.6914,
      "step": 3143
    },
    {
      "epoch": 0.13029963943802064,
      "grad_norm": 0.4241204857826233,
      "learning_rate": 4.3487090223382654e-06,
      "loss": 0.7007,
      "step": 3144
    },
    {
      "epoch": 0.13034108334369432,
      "grad_norm": 0.44318723678588867,
      "learning_rate": 4.348501802809898e-06,
      "loss": 0.7031,
      "step": 3145
    },
    {
      "epoch": 0.13038252724936797,
      "grad_norm": 0.40496712923049927,
      "learning_rate": 4.348294583281529e-06,
      "loss": 0.7048,
      "step": 3146
    },
    {
      "epoch": 0.13042397115504165,
      "grad_norm": 0.4204508662223816,
      "learning_rate": 4.3480873637531605e-06,
      "loss": 0.7161,
      "step": 3147
    },
    {
      "epoch": 0.13046541506071532,
      "grad_norm": 0.4179397225379944,
      "learning_rate": 4.347880144224792e-06,
      "loss": 0.7515,
      "step": 3148
    },
    {
      "epoch": 0.130506858966389,
      "grad_norm": 0.43040668964385986,
      "learning_rate": 4.347672924696424e-06,
      "loss": 0.6654,
      "step": 3149
    },
    {
      "epoch": 0.13054830287206268,
      "grad_norm": 0.4331599175930023,
      "learning_rate": 4.3474657051680555e-06,
      "loss": 0.8103,
      "step": 3150
    },
    {
      "epoch": 0.13058974677773633,
      "grad_norm": 0.4416988790035248,
      "learning_rate": 4.347258485639687e-06,
      "loss": 0.7239,
      "step": 3151
    },
    {
      "epoch": 0.13063119068341,
      "grad_norm": 0.4782413840293884,
      "learning_rate": 4.347051266111318e-06,
      "loss": 0.8135,
      "step": 3152
    },
    {
      "epoch": 0.13067263458908368,
      "grad_norm": 0.4335290193557739,
      "learning_rate": 4.3468440465829505e-06,
      "loss": 0.7173,
      "step": 3153
    },
    {
      "epoch": 0.13071407849475736,
      "grad_norm": 0.448612242937088,
      "learning_rate": 4.346636827054582e-06,
      "loss": 0.7466,
      "step": 3154
    },
    {
      "epoch": 0.130755522400431,
      "grad_norm": 0.43644189834594727,
      "learning_rate": 4.346429607526213e-06,
      "loss": 0.7297,
      "step": 3155
    },
    {
      "epoch": 0.13079696630610468,
      "grad_norm": 0.4221780300140381,
      "learning_rate": 4.3462223879978455e-06,
      "loss": 0.7546,
      "step": 3156
    },
    {
      "epoch": 0.13083841021177836,
      "grad_norm": 0.40728509426116943,
      "learning_rate": 4.346015168469477e-06,
      "loss": 0.7498,
      "step": 3157
    },
    {
      "epoch": 0.13087985411745204,
      "grad_norm": 0.42668360471725464,
      "learning_rate": 4.345807948941109e-06,
      "loss": 0.7466,
      "step": 3158
    },
    {
      "epoch": 0.1309212980231257,
      "grad_norm": 0.4262279272079468,
      "learning_rate": 4.34560072941274e-06,
      "loss": 0.7385,
      "step": 3159
    },
    {
      "epoch": 0.13096274192879936,
      "grad_norm": 0.4606529772281647,
      "learning_rate": 4.345393509884372e-06,
      "loss": 0.7588,
      "step": 3160
    },
    {
      "epoch": 0.13100418583447304,
      "grad_norm": 0.4493151605129242,
      "learning_rate": 4.345186290356003e-06,
      "loss": 0.7566,
      "step": 3161
    },
    {
      "epoch": 0.13104562974014672,
      "grad_norm": 0.4094814956188202,
      "learning_rate": 4.3449790708276355e-06,
      "loss": 0.7405,
      "step": 3162
    },
    {
      "epoch": 0.1310870736458204,
      "grad_norm": 0.4557473361492157,
      "learning_rate": 4.344771851299267e-06,
      "loss": 0.7952,
      "step": 3163
    },
    {
      "epoch": 0.13112851755149404,
      "grad_norm": 0.45078229904174805,
      "learning_rate": 4.344564631770898e-06,
      "loss": 0.7612,
      "step": 3164
    },
    {
      "epoch": 0.13116996145716772,
      "grad_norm": 0.4167359471321106,
      "learning_rate": 4.3443574122425305e-06,
      "loss": 0.7173,
      "step": 3165
    },
    {
      "epoch": 0.1312114053628414,
      "grad_norm": 0.4306659996509552,
      "learning_rate": 4.344150192714162e-06,
      "loss": 0.7551,
      "step": 3166
    },
    {
      "epoch": 0.13125284926851508,
      "grad_norm": 0.4610150456428528,
      "learning_rate": 4.343942973185793e-06,
      "loss": 0.7682,
      "step": 3167
    },
    {
      "epoch": 0.13129429317418873,
      "grad_norm": 0.4115774929523468,
      "learning_rate": 4.343735753657425e-06,
      "loss": 0.7168,
      "step": 3168
    },
    {
      "epoch": 0.1313357370798624,
      "grad_norm": 0.43952569365501404,
      "learning_rate": 4.343528534129057e-06,
      "loss": 0.7122,
      "step": 3169
    },
    {
      "epoch": 0.13137718098553608,
      "grad_norm": 0.4395902454853058,
      "learning_rate": 4.343321314600688e-06,
      "loss": 0.7059,
      "step": 3170
    },
    {
      "epoch": 0.13141862489120976,
      "grad_norm": 0.444606214761734,
      "learning_rate": 4.34311409507232e-06,
      "loss": 0.7233,
      "step": 3171
    },
    {
      "epoch": 0.1314600687968834,
      "grad_norm": 0.48068445920944214,
      "learning_rate": 4.342906875543952e-06,
      "loss": 0.771,
      "step": 3172
    },
    {
      "epoch": 0.13150151270255708,
      "grad_norm": 0.39579591155052185,
      "learning_rate": 4.342699656015583e-06,
      "loss": 0.6877,
      "step": 3173
    },
    {
      "epoch": 0.13154295660823076,
      "grad_norm": 0.4405721426010132,
      "learning_rate": 4.3424924364872155e-06,
      "loss": 0.7402,
      "step": 3174
    },
    {
      "epoch": 0.13158440051390444,
      "grad_norm": 0.4287813901901245,
      "learning_rate": 4.342285216958846e-06,
      "loss": 0.7473,
      "step": 3175
    },
    {
      "epoch": 0.13162584441957811,
      "grad_norm": 0.444559782743454,
      "learning_rate": 4.342077997430478e-06,
      "loss": 0.7805,
      "step": 3176
    },
    {
      "epoch": 0.13166728832525176,
      "grad_norm": 0.40739330649375916,
      "learning_rate": 4.34187077790211e-06,
      "loss": 0.7112,
      "step": 3177
    },
    {
      "epoch": 0.13170873223092544,
      "grad_norm": 0.43769571185112,
      "learning_rate": 4.341663558373742e-06,
      "loss": 0.7341,
      "step": 3178
    },
    {
      "epoch": 0.13175017613659912,
      "grad_norm": 0.4883594810962677,
      "learning_rate": 4.341456338845373e-06,
      "loss": 0.7534,
      "step": 3179
    },
    {
      "epoch": 0.1317916200422728,
      "grad_norm": 0.4317252039909363,
      "learning_rate": 4.341249119317005e-06,
      "loss": 0.7488,
      "step": 3180
    },
    {
      "epoch": 0.13183306394794644,
      "grad_norm": 0.4231094419956207,
      "learning_rate": 4.341041899788637e-06,
      "loss": 0.6772,
      "step": 3181
    },
    {
      "epoch": 0.13187450785362012,
      "grad_norm": 0.4293449819087982,
      "learning_rate": 4.340834680260268e-06,
      "loss": 0.7444,
      "step": 3182
    },
    {
      "epoch": 0.1319159517592938,
      "grad_norm": 0.42454656958580017,
      "learning_rate": 4.3406274607319e-06,
      "loss": 0.7776,
      "step": 3183
    },
    {
      "epoch": 0.13195739566496748,
      "grad_norm": 0.3830515742301941,
      "learning_rate": 4.340420241203531e-06,
      "loss": 0.7133,
      "step": 3184
    },
    {
      "epoch": 0.13199883957064112,
      "grad_norm": 0.4094764292240143,
      "learning_rate": 4.340213021675163e-06,
      "loss": 0.7407,
      "step": 3185
    },
    {
      "epoch": 0.1320402834763148,
      "grad_norm": 0.4515026807785034,
      "learning_rate": 4.340005802146795e-06,
      "loss": 0.7629,
      "step": 3186
    },
    {
      "epoch": 0.13208172738198848,
      "grad_norm": 0.40350958704948425,
      "learning_rate": 4.339798582618426e-06,
      "loss": 0.7378,
      "step": 3187
    },
    {
      "epoch": 0.13212317128766216,
      "grad_norm": 0.4185269773006439,
      "learning_rate": 4.3395913630900574e-06,
      "loss": 0.7222,
      "step": 3188
    },
    {
      "epoch": 0.13216461519333583,
      "grad_norm": 0.3971675634384155,
      "learning_rate": 4.33938414356169e-06,
      "loss": 0.7391,
      "step": 3189
    },
    {
      "epoch": 0.13220605909900948,
      "grad_norm": 0.4705488085746765,
      "learning_rate": 4.339176924033322e-06,
      "loss": 0.7688,
      "step": 3190
    },
    {
      "epoch": 0.13224750300468316,
      "grad_norm": 0.44235652685165405,
      "learning_rate": 4.3389697045049524e-06,
      "loss": 0.7261,
      "step": 3191
    },
    {
      "epoch": 0.13228894691035684,
      "grad_norm": 0.46159955859184265,
      "learning_rate": 4.338762484976585e-06,
      "loss": 0.7402,
      "step": 3192
    },
    {
      "epoch": 0.1323303908160305,
      "grad_norm": 0.40473857522010803,
      "learning_rate": 4.338555265448216e-06,
      "loss": 0.6825,
      "step": 3193
    },
    {
      "epoch": 0.13237183472170416,
      "grad_norm": 0.41825172305107117,
      "learning_rate": 4.338348045919848e-06,
      "loss": 0.7617,
      "step": 3194
    },
    {
      "epoch": 0.13241327862737784,
      "grad_norm": 0.41895949840545654,
      "learning_rate": 4.33814082639148e-06,
      "loss": 0.7319,
      "step": 3195
    },
    {
      "epoch": 0.13245472253305152,
      "grad_norm": 0.4375893771648407,
      "learning_rate": 4.337933606863111e-06,
      "loss": 0.7463,
      "step": 3196
    },
    {
      "epoch": 0.1324961664387252,
      "grad_norm": 0.47569039463996887,
      "learning_rate": 4.337726387334743e-06,
      "loss": 0.8157,
      "step": 3197
    },
    {
      "epoch": 0.13253761034439884,
      "grad_norm": 0.42760708928108215,
      "learning_rate": 4.337519167806375e-06,
      "loss": 0.6865,
      "step": 3198
    },
    {
      "epoch": 0.13257905425007252,
      "grad_norm": 0.4418492019176483,
      "learning_rate": 4.337311948278006e-06,
      "loss": 0.7866,
      "step": 3199
    },
    {
      "epoch": 0.1326204981557462,
      "grad_norm": 0.42479759454727173,
      "learning_rate": 4.3371047287496374e-06,
      "loss": 0.7052,
      "step": 3200
    },
    {
      "epoch": 0.13266194206141987,
      "grad_norm": 0.4728982448577881,
      "learning_rate": 4.33689750922127e-06,
      "loss": 0.7617,
      "step": 3201
    },
    {
      "epoch": 0.13270338596709355,
      "grad_norm": 0.44464796781539917,
      "learning_rate": 4.336690289692901e-06,
      "loss": 0.7169,
      "step": 3202
    },
    {
      "epoch": 0.1327448298727672,
      "grad_norm": 0.4362186789512634,
      "learning_rate": 4.3364830701645324e-06,
      "loss": 0.7261,
      "step": 3203
    },
    {
      "epoch": 0.13278627377844088,
      "grad_norm": 0.43370163440704346,
      "learning_rate": 4.336275850636164e-06,
      "loss": 0.74,
      "step": 3204
    },
    {
      "epoch": 0.13282771768411455,
      "grad_norm": 0.4373166859149933,
      "learning_rate": 4.336068631107796e-06,
      "loss": 0.7583,
      "step": 3205
    },
    {
      "epoch": 0.13286916158978823,
      "grad_norm": 0.43720272183418274,
      "learning_rate": 4.3358614115794275e-06,
      "loss": 0.7885,
      "step": 3206
    },
    {
      "epoch": 0.13291060549546188,
      "grad_norm": 0.44951850175857544,
      "learning_rate": 4.335654192051059e-06,
      "loss": 0.7463,
      "step": 3207
    },
    {
      "epoch": 0.13295204940113556,
      "grad_norm": 0.4723678231239319,
      "learning_rate": 4.335446972522691e-06,
      "loss": 0.7705,
      "step": 3208
    },
    {
      "epoch": 0.13299349330680924,
      "grad_norm": 0.44698601961135864,
      "learning_rate": 4.3352397529943225e-06,
      "loss": 0.7474,
      "step": 3209
    },
    {
      "epoch": 0.1330349372124829,
      "grad_norm": 0.45008814334869385,
      "learning_rate": 4.335032533465955e-06,
      "loss": 0.7432,
      "step": 3210
    },
    {
      "epoch": 0.13307638111815656,
      "grad_norm": 0.409915953874588,
      "learning_rate": 4.334825313937586e-06,
      "loss": 0.7231,
      "step": 3211
    },
    {
      "epoch": 0.13311782502383024,
      "grad_norm": 0.43293067812919617,
      "learning_rate": 4.3346180944092175e-06,
      "loss": 0.7415,
      "step": 3212
    },
    {
      "epoch": 0.13315926892950392,
      "grad_norm": 0.466900497674942,
      "learning_rate": 4.334410874880849e-06,
      "loss": 0.7241,
      "step": 3213
    },
    {
      "epoch": 0.1332007128351776,
      "grad_norm": 0.39762866497039795,
      "learning_rate": 4.334203655352481e-06,
      "loss": 0.7322,
      "step": 3214
    },
    {
      "epoch": 0.13324215674085127,
      "grad_norm": 0.45224279165267944,
      "learning_rate": 4.3339964358241125e-06,
      "loss": 0.7708,
      "step": 3215
    },
    {
      "epoch": 0.13328360064652492,
      "grad_norm": 0.45189782977104187,
      "learning_rate": 4.333789216295744e-06,
      "loss": 0.7986,
      "step": 3216
    },
    {
      "epoch": 0.1333250445521986,
      "grad_norm": 0.4780016243457794,
      "learning_rate": 4.333581996767376e-06,
      "loss": 0.7485,
      "step": 3217
    },
    {
      "epoch": 0.13336648845787227,
      "grad_norm": 0.4144827723503113,
      "learning_rate": 4.3333747772390075e-06,
      "loss": 0.772,
      "step": 3218
    },
    {
      "epoch": 0.13340793236354595,
      "grad_norm": 0.4675489366054535,
      "learning_rate": 4.333167557710639e-06,
      "loss": 0.7981,
      "step": 3219
    },
    {
      "epoch": 0.1334493762692196,
      "grad_norm": 0.43756312131881714,
      "learning_rate": 4.33296033818227e-06,
      "loss": 0.728,
      "step": 3220
    },
    {
      "epoch": 0.13349082017489328,
      "grad_norm": 0.4187451899051666,
      "learning_rate": 4.3327531186539025e-06,
      "loss": 0.6693,
      "step": 3221
    },
    {
      "epoch": 0.13353226408056695,
      "grad_norm": 0.44672414660453796,
      "learning_rate": 4.332545899125534e-06,
      "loss": 0.7822,
      "step": 3222
    },
    {
      "epoch": 0.13357370798624063,
      "grad_norm": 0.41613438725471497,
      "learning_rate": 4.332338679597165e-06,
      "loss": 0.7236,
      "step": 3223
    },
    {
      "epoch": 0.1336151518919143,
      "grad_norm": 0.44749435782432556,
      "learning_rate": 4.3321314600687975e-06,
      "loss": 0.7554,
      "step": 3224
    },
    {
      "epoch": 0.13365659579758796,
      "grad_norm": 0.4225656986236572,
      "learning_rate": 4.331924240540429e-06,
      "loss": 0.744,
      "step": 3225
    },
    {
      "epoch": 0.13369803970326163,
      "grad_norm": 0.45646581053733826,
      "learning_rate": 4.331717021012061e-06,
      "loss": 0.772,
      "step": 3226
    },
    {
      "epoch": 0.1337394836089353,
      "grad_norm": 0.4501378834247589,
      "learning_rate": 4.331509801483692e-06,
      "loss": 0.725,
      "step": 3227
    },
    {
      "epoch": 0.133780927514609,
      "grad_norm": 0.46129289269447327,
      "learning_rate": 4.331302581955324e-06,
      "loss": 0.7576,
      "step": 3228
    },
    {
      "epoch": 0.13382237142028264,
      "grad_norm": 0.44970420002937317,
      "learning_rate": 4.331095362426955e-06,
      "loss": 0.748,
      "step": 3229
    },
    {
      "epoch": 0.13386381532595631,
      "grad_norm": 0.44980454444885254,
      "learning_rate": 4.3308881428985875e-06,
      "loss": 0.7805,
      "step": 3230
    },
    {
      "epoch": 0.13390525923163,
      "grad_norm": 0.4057115316390991,
      "learning_rate": 4.330680923370219e-06,
      "loss": 0.7798,
      "step": 3231
    },
    {
      "epoch": 0.13394670313730367,
      "grad_norm": 0.4357262849807739,
      "learning_rate": 4.33047370384185e-06,
      "loss": 0.8035,
      "step": 3232
    },
    {
      "epoch": 0.13398814704297732,
      "grad_norm": 0.473522424697876,
      "learning_rate": 4.3302664843134825e-06,
      "loss": 0.788,
      "step": 3233
    },
    {
      "epoch": 0.134029590948651,
      "grad_norm": 0.40912574529647827,
      "learning_rate": 4.330059264785114e-06,
      "loss": 0.6713,
      "step": 3234
    },
    {
      "epoch": 0.13407103485432467,
      "grad_norm": 0.41532158851623535,
      "learning_rate": 4.329852045256745e-06,
      "loss": 0.7729,
      "step": 3235
    },
    {
      "epoch": 0.13411247875999835,
      "grad_norm": 0.48318198323249817,
      "learning_rate": 4.329644825728377e-06,
      "loss": 0.7312,
      "step": 3236
    },
    {
      "epoch": 0.13415392266567203,
      "grad_norm": 0.5060474872589111,
      "learning_rate": 4.329437606200009e-06,
      "loss": 0.7581,
      "step": 3237
    },
    {
      "epoch": 0.13419536657134568,
      "grad_norm": 0.41201525926589966,
      "learning_rate": 4.32923038667164e-06,
      "loss": 0.749,
      "step": 3238
    },
    {
      "epoch": 0.13423681047701935,
      "grad_norm": 0.4343048930168152,
      "learning_rate": 4.329023167143272e-06,
      "loss": 0.7188,
      "step": 3239
    },
    {
      "epoch": 0.13427825438269303,
      "grad_norm": 0.428295373916626,
      "learning_rate": 4.328815947614903e-06,
      "loss": 0.73,
      "step": 3240
    },
    {
      "epoch": 0.1343196982883667,
      "grad_norm": 0.4618155062198639,
      "learning_rate": 4.328608728086535e-06,
      "loss": 0.7122,
      "step": 3241
    },
    {
      "epoch": 0.13436114219404036,
      "grad_norm": 0.4237076938152313,
      "learning_rate": 4.3284015085581675e-06,
      "loss": 0.6992,
      "step": 3242
    },
    {
      "epoch": 0.13440258609971403,
      "grad_norm": 0.4964372515678406,
      "learning_rate": 4.328194289029798e-06,
      "loss": 0.7593,
      "step": 3243
    },
    {
      "epoch": 0.1344440300053877,
      "grad_norm": 0.43819156289100647,
      "learning_rate": 4.32798706950143e-06,
      "loss": 0.697,
      "step": 3244
    },
    {
      "epoch": 0.1344854739110614,
      "grad_norm": 0.4704984426498413,
      "learning_rate": 4.327779849973062e-06,
      "loss": 0.7153,
      "step": 3245
    },
    {
      "epoch": 0.13452691781673504,
      "grad_norm": 0.5024917721748352,
      "learning_rate": 4.327572630444694e-06,
      "loss": 0.8037,
      "step": 3246
    },
    {
      "epoch": 0.1345683617224087,
      "grad_norm": 0.4122340679168701,
      "learning_rate": 4.327365410916325e-06,
      "loss": 0.7778,
      "step": 3247
    },
    {
      "epoch": 0.1346098056280824,
      "grad_norm": 0.4296201765537262,
      "learning_rate": 4.327158191387957e-06,
      "loss": 0.7239,
      "step": 3248
    },
    {
      "epoch": 0.13465124953375607,
      "grad_norm": 0.41790303587913513,
      "learning_rate": 4.326950971859589e-06,
      "loss": 0.7373,
      "step": 3249
    },
    {
      "epoch": 0.13469269343942974,
      "grad_norm": 0.43771839141845703,
      "learning_rate": 4.32674375233122e-06,
      "loss": 0.7749,
      "step": 3250
    },
    {
      "epoch": 0.1347341373451034,
      "grad_norm": 0.43307211995124817,
      "learning_rate": 4.326536532802852e-06,
      "loss": 0.7705,
      "step": 3251
    },
    {
      "epoch": 0.13477558125077707,
      "grad_norm": 0.45304903388023376,
      "learning_rate": 4.326329313274483e-06,
      "loss": 0.719,
      "step": 3252
    },
    {
      "epoch": 0.13481702515645075,
      "grad_norm": 0.42284977436065674,
      "learning_rate": 4.326122093746115e-06,
      "loss": 0.6877,
      "step": 3253
    },
    {
      "epoch": 0.13485846906212443,
      "grad_norm": 0.41939017176628113,
      "learning_rate": 4.325914874217747e-06,
      "loss": 0.7179,
      "step": 3254
    },
    {
      "epoch": 0.13489991296779807,
      "grad_norm": 0.45863446593284607,
      "learning_rate": 4.325707654689378e-06,
      "loss": 0.7891,
      "step": 3255
    },
    {
      "epoch": 0.13494135687347175,
      "grad_norm": 0.45165812969207764,
      "learning_rate": 4.3255004351610094e-06,
      "loss": 0.7942,
      "step": 3256
    },
    {
      "epoch": 0.13498280077914543,
      "grad_norm": 0.40971502661705017,
      "learning_rate": 4.325293215632642e-06,
      "loss": 0.7324,
      "step": 3257
    },
    {
      "epoch": 0.1350242446848191,
      "grad_norm": 0.4014582335948944,
      "learning_rate": 4.325085996104274e-06,
      "loss": 0.7385,
      "step": 3258
    },
    {
      "epoch": 0.13506568859049276,
      "grad_norm": 0.41865915060043335,
      "learning_rate": 4.3248787765759044e-06,
      "loss": 0.7656,
      "step": 3259
    },
    {
      "epoch": 0.13510713249616643,
      "grad_norm": 0.4344530999660492,
      "learning_rate": 4.324671557047537e-06,
      "loss": 0.7493,
      "step": 3260
    },
    {
      "epoch": 0.1351485764018401,
      "grad_norm": 0.41098886728286743,
      "learning_rate": 4.324464337519168e-06,
      "loss": 0.7461,
      "step": 3261
    },
    {
      "epoch": 0.1351900203075138,
      "grad_norm": 0.4427797198295593,
      "learning_rate": 4.3242571179908e-06,
      "loss": 0.7266,
      "step": 3262
    },
    {
      "epoch": 0.13523146421318746,
      "grad_norm": 0.5158562660217285,
      "learning_rate": 4.324049898462432e-06,
      "loss": 0.7805,
      "step": 3263
    },
    {
      "epoch": 0.1352729081188611,
      "grad_norm": 0.4522770643234253,
      "learning_rate": 4.323842678934063e-06,
      "loss": 0.7476,
      "step": 3264
    },
    {
      "epoch": 0.1353143520245348,
      "grad_norm": 0.4183885455131531,
      "learning_rate": 4.3236354594056945e-06,
      "loss": 0.7151,
      "step": 3265
    },
    {
      "epoch": 0.13535579593020847,
      "grad_norm": 0.42074689269065857,
      "learning_rate": 4.323428239877327e-06,
      "loss": 0.7244,
      "step": 3266
    },
    {
      "epoch": 0.13539723983588214,
      "grad_norm": 0.42470109462738037,
      "learning_rate": 4.323221020348958e-06,
      "loss": 0.735,
      "step": 3267
    },
    {
      "epoch": 0.1354386837415558,
      "grad_norm": 0.48983752727508545,
      "learning_rate": 4.3230138008205895e-06,
      "loss": 0.7788,
      "step": 3268
    },
    {
      "epoch": 0.13548012764722947,
      "grad_norm": 0.4303596615791321,
      "learning_rate": 4.322806581292222e-06,
      "loss": 0.6938,
      "step": 3269
    },
    {
      "epoch": 0.13552157155290315,
      "grad_norm": 0.4334631860256195,
      "learning_rate": 4.322599361763853e-06,
      "loss": 0.7323,
      "step": 3270
    },
    {
      "epoch": 0.13556301545857682,
      "grad_norm": 0.4621518552303314,
      "learning_rate": 4.3223921422354845e-06,
      "loss": 0.7722,
      "step": 3271
    },
    {
      "epoch": 0.13560445936425047,
      "grad_norm": 0.45699045062065125,
      "learning_rate": 4.322184922707116e-06,
      "loss": 0.7908,
      "step": 3272
    },
    {
      "epoch": 0.13564590326992415,
      "grad_norm": 0.4064429700374603,
      "learning_rate": 4.321977703178748e-06,
      "loss": 0.7195,
      "step": 3273
    },
    {
      "epoch": 0.13568734717559783,
      "grad_norm": 0.4567306935787201,
      "learning_rate": 4.3217704836503795e-06,
      "loss": 0.813,
      "step": 3274
    },
    {
      "epoch": 0.1357287910812715,
      "grad_norm": 0.42542627453804016,
      "learning_rate": 4.321563264122011e-06,
      "loss": 0.7336,
      "step": 3275
    },
    {
      "epoch": 0.13577023498694518,
      "grad_norm": 0.4275822937488556,
      "learning_rate": 4.321356044593643e-06,
      "loss": 0.7197,
      "step": 3276
    },
    {
      "epoch": 0.13581167889261883,
      "grad_norm": 0.4292237162590027,
      "learning_rate": 4.3211488250652745e-06,
      "loss": 0.7317,
      "step": 3277
    },
    {
      "epoch": 0.1358531227982925,
      "grad_norm": 0.4524635672569275,
      "learning_rate": 4.320941605536907e-06,
      "loss": 0.7356,
      "step": 3278
    },
    {
      "epoch": 0.13589456670396619,
      "grad_norm": 0.43820255994796753,
      "learning_rate": 4.320734386008538e-06,
      "loss": 0.7183,
      "step": 3279
    },
    {
      "epoch": 0.13593601060963986,
      "grad_norm": 0.4431719481945038,
      "learning_rate": 4.3205271664801695e-06,
      "loss": 0.7849,
      "step": 3280
    },
    {
      "epoch": 0.1359774545153135,
      "grad_norm": 0.4130253195762634,
      "learning_rate": 4.320319946951801e-06,
      "loss": 0.7122,
      "step": 3281
    },
    {
      "epoch": 0.1360188984209872,
      "grad_norm": 0.4279428720474243,
      "learning_rate": 4.320112727423433e-06,
      "loss": 0.7747,
      "step": 3282
    },
    {
      "epoch": 0.13606034232666087,
      "grad_norm": 0.4873661696910858,
      "learning_rate": 4.3199055078950645e-06,
      "loss": 0.7856,
      "step": 3283
    },
    {
      "epoch": 0.13610178623233454,
      "grad_norm": 0.42564108967781067,
      "learning_rate": 4.319698288366696e-06,
      "loss": 0.7549,
      "step": 3284
    },
    {
      "epoch": 0.1361432301380082,
      "grad_norm": 0.4233907461166382,
      "learning_rate": 4.319491068838328e-06,
      "loss": 0.7986,
      "step": 3285
    },
    {
      "epoch": 0.13618467404368187,
      "grad_norm": 0.46017396450042725,
      "learning_rate": 4.3192838493099595e-06,
      "loss": 0.7651,
      "step": 3286
    },
    {
      "epoch": 0.13622611794935555,
      "grad_norm": 0.4139768183231354,
      "learning_rate": 4.319076629781591e-06,
      "loss": 0.7264,
      "step": 3287
    },
    {
      "epoch": 0.13626756185502922,
      "grad_norm": 0.4284610450267792,
      "learning_rate": 4.318869410253222e-06,
      "loss": 0.731,
      "step": 3288
    },
    {
      "epoch": 0.1363090057607029,
      "grad_norm": 0.3987548053264618,
      "learning_rate": 4.3186621907248545e-06,
      "loss": 0.7677,
      "step": 3289
    },
    {
      "epoch": 0.13635044966637655,
      "grad_norm": 0.4312137961387634,
      "learning_rate": 4.318454971196486e-06,
      "loss": 0.7651,
      "step": 3290
    },
    {
      "epoch": 0.13639189357205023,
      "grad_norm": 0.6239796280860901,
      "learning_rate": 4.318247751668117e-06,
      "loss": 0.7858,
      "step": 3291
    },
    {
      "epoch": 0.1364333374777239,
      "grad_norm": 0.4032297432422638,
      "learning_rate": 4.318040532139749e-06,
      "loss": 0.7095,
      "step": 3292
    },
    {
      "epoch": 0.13647478138339758,
      "grad_norm": 0.45367351174354553,
      "learning_rate": 4.317833312611381e-06,
      "loss": 0.7629,
      "step": 3293
    },
    {
      "epoch": 0.13651622528907123,
      "grad_norm": 0.47958412766456604,
      "learning_rate": 4.317626093083013e-06,
      "loss": 0.8145,
      "step": 3294
    },
    {
      "epoch": 0.1365576691947449,
      "grad_norm": 0.42471709847450256,
      "learning_rate": 4.317418873554644e-06,
      "loss": 0.7554,
      "step": 3295
    },
    {
      "epoch": 0.13659911310041858,
      "grad_norm": 0.465071439743042,
      "learning_rate": 4.317211654026276e-06,
      "loss": 0.7935,
      "step": 3296
    },
    {
      "epoch": 0.13664055700609226,
      "grad_norm": 0.40812942385673523,
      "learning_rate": 4.317004434497907e-06,
      "loss": 0.7136,
      "step": 3297
    },
    {
      "epoch": 0.13668200091176594,
      "grad_norm": 0.4135816693305969,
      "learning_rate": 4.3167972149695395e-06,
      "loss": 0.7083,
      "step": 3298
    },
    {
      "epoch": 0.1367234448174396,
      "grad_norm": 0.40938976407051086,
      "learning_rate": 4.316589995441171e-06,
      "loss": 0.7122,
      "step": 3299
    },
    {
      "epoch": 0.13676488872311326,
      "grad_norm": 0.4250430166721344,
      "learning_rate": 4.316382775912802e-06,
      "loss": 0.7507,
      "step": 3300
    },
    {
      "epoch": 0.13680633262878694,
      "grad_norm": 0.4539840817451477,
      "learning_rate": 4.316175556384434e-06,
      "loss": 0.7373,
      "step": 3301
    },
    {
      "epoch": 0.13684777653446062,
      "grad_norm": 0.45435410737991333,
      "learning_rate": 4.315968336856066e-06,
      "loss": 0.7261,
      "step": 3302
    },
    {
      "epoch": 0.13688922044013427,
      "grad_norm": 0.49231502413749695,
      "learning_rate": 4.315761117327697e-06,
      "loss": 0.729,
      "step": 3303
    },
    {
      "epoch": 0.13693066434580795,
      "grad_norm": 0.433064341545105,
      "learning_rate": 4.315553897799329e-06,
      "loss": 0.7062,
      "step": 3304
    },
    {
      "epoch": 0.13697210825148162,
      "grad_norm": 0.4862290620803833,
      "learning_rate": 4.315346678270961e-06,
      "loss": 0.7634,
      "step": 3305
    },
    {
      "epoch": 0.1370135521571553,
      "grad_norm": 0.43565258383750916,
      "learning_rate": 4.315139458742592e-06,
      "loss": 0.7664,
      "step": 3306
    },
    {
      "epoch": 0.13705499606282895,
      "grad_norm": 0.42442816495895386,
      "learning_rate": 4.314932239214224e-06,
      "loss": 0.7771,
      "step": 3307
    },
    {
      "epoch": 0.13709643996850263,
      "grad_norm": 0.4279419183731079,
      "learning_rate": 4.314725019685855e-06,
      "loss": 0.7104,
      "step": 3308
    },
    {
      "epoch": 0.1371378838741763,
      "grad_norm": 0.4139658212661743,
      "learning_rate": 4.314517800157487e-06,
      "loss": 0.7595,
      "step": 3309
    },
    {
      "epoch": 0.13717932777984998,
      "grad_norm": 0.4316956102848053,
      "learning_rate": 4.3143105806291195e-06,
      "loss": 0.7417,
      "step": 3310
    },
    {
      "epoch": 0.13722077168552366,
      "grad_norm": 0.5083264112472534,
      "learning_rate": 4.31410336110075e-06,
      "loss": 0.7698,
      "step": 3311
    },
    {
      "epoch": 0.1372622155911973,
      "grad_norm": 0.42437848448753357,
      "learning_rate": 4.313896141572382e-06,
      "loss": 0.7766,
      "step": 3312
    },
    {
      "epoch": 0.13730365949687098,
      "grad_norm": 0.42439892888069153,
      "learning_rate": 4.313688922044014e-06,
      "loss": 0.7124,
      "step": 3313
    },
    {
      "epoch": 0.13734510340254466,
      "grad_norm": 0.4494580924510956,
      "learning_rate": 4.313481702515646e-06,
      "loss": 0.7375,
      "step": 3314
    },
    {
      "epoch": 0.13738654730821834,
      "grad_norm": 0.41579461097717285,
      "learning_rate": 4.313274482987277e-06,
      "loss": 0.7512,
      "step": 3315
    },
    {
      "epoch": 0.137427991213892,
      "grad_norm": 0.46532508730888367,
      "learning_rate": 4.313067263458909e-06,
      "loss": 0.8091,
      "step": 3316
    },
    {
      "epoch": 0.13746943511956566,
      "grad_norm": 0.4343154728412628,
      "learning_rate": 4.31286004393054e-06,
      "loss": 0.7671,
      "step": 3317
    },
    {
      "epoch": 0.13751087902523934,
      "grad_norm": 0.46249207854270935,
      "learning_rate": 4.312652824402172e-06,
      "loss": 0.8242,
      "step": 3318
    },
    {
      "epoch": 0.13755232293091302,
      "grad_norm": 0.4311218559741974,
      "learning_rate": 4.312445604873804e-06,
      "loss": 0.7026,
      "step": 3319
    },
    {
      "epoch": 0.13759376683658667,
      "grad_norm": 0.4132808744907379,
      "learning_rate": 4.312238385345435e-06,
      "loss": 0.728,
      "step": 3320
    },
    {
      "epoch": 0.13763521074226034,
      "grad_norm": 0.4490644335746765,
      "learning_rate": 4.312031165817067e-06,
      "loss": 0.7307,
      "step": 3321
    },
    {
      "epoch": 0.13767665464793402,
      "grad_norm": 0.47734448313713074,
      "learning_rate": 4.311823946288699e-06,
      "loss": 0.7844,
      "step": 3322
    },
    {
      "epoch": 0.1377180985536077,
      "grad_norm": 1.2305147647857666,
      "learning_rate": 4.31161672676033e-06,
      "loss": 0.7303,
      "step": 3323
    },
    {
      "epoch": 0.13775954245928138,
      "grad_norm": 0.40227973461151123,
      "learning_rate": 4.3114095072319615e-06,
      "loss": 0.6914,
      "step": 3324
    },
    {
      "epoch": 0.13780098636495502,
      "grad_norm": 0.44827958941459656,
      "learning_rate": 4.311202287703594e-06,
      "loss": 0.7725,
      "step": 3325
    },
    {
      "epoch": 0.1378424302706287,
      "grad_norm": 0.4013510048389435,
      "learning_rate": 4.310995068175225e-06,
      "loss": 0.7131,
      "step": 3326
    },
    {
      "epoch": 0.13788387417630238,
      "grad_norm": 0.4368317425251007,
      "learning_rate": 4.3107878486468565e-06,
      "loss": 0.7012,
      "step": 3327
    },
    {
      "epoch": 0.13792531808197606,
      "grad_norm": 0.4506840109825134,
      "learning_rate": 4.310580629118489e-06,
      "loss": 0.7493,
      "step": 3328
    },
    {
      "epoch": 0.1379667619876497,
      "grad_norm": 0.40915173292160034,
      "learning_rate": 4.31037340959012e-06,
      "loss": 0.7351,
      "step": 3329
    },
    {
      "epoch": 0.13800820589332338,
      "grad_norm": 0.44817087054252625,
      "learning_rate": 4.310166190061752e-06,
      "loss": 0.7555,
      "step": 3330
    },
    {
      "epoch": 0.13804964979899706,
      "grad_norm": 0.4021657407283783,
      "learning_rate": 4.309958970533384e-06,
      "loss": 0.7273,
      "step": 3331
    },
    {
      "epoch": 0.13809109370467074,
      "grad_norm": 0.40061238408088684,
      "learning_rate": 4.309751751005015e-06,
      "loss": 0.7651,
      "step": 3332
    },
    {
      "epoch": 0.13813253761034439,
      "grad_norm": 0.44201990962028503,
      "learning_rate": 4.3095445314766465e-06,
      "loss": 0.7546,
      "step": 3333
    },
    {
      "epoch": 0.13817398151601806,
      "grad_norm": 0.4271180331707001,
      "learning_rate": 4.309337311948279e-06,
      "loss": 0.7583,
      "step": 3334
    },
    {
      "epoch": 0.13821542542169174,
      "grad_norm": 0.43941429257392883,
      "learning_rate": 4.30913009241991e-06,
      "loss": 0.7485,
      "step": 3335
    },
    {
      "epoch": 0.13825686932736542,
      "grad_norm": 0.4266482889652252,
      "learning_rate": 4.3089228728915415e-06,
      "loss": 0.7192,
      "step": 3336
    },
    {
      "epoch": 0.1382983132330391,
      "grad_norm": 0.4823482930660248,
      "learning_rate": 4.308715653363174e-06,
      "loss": 0.7314,
      "step": 3337
    },
    {
      "epoch": 0.13833975713871274,
      "grad_norm": 0.4269290268421173,
      "learning_rate": 4.308508433834805e-06,
      "loss": 0.7869,
      "step": 3338
    },
    {
      "epoch": 0.13838120104438642,
      "grad_norm": 0.42131197452545166,
      "learning_rate": 4.3083012143064365e-06,
      "loss": 0.7495,
      "step": 3339
    },
    {
      "epoch": 0.1384226449500601,
      "grad_norm": 0.45400270819664,
      "learning_rate": 4.308093994778068e-06,
      "loss": 0.7749,
      "step": 3340
    },
    {
      "epoch": 0.13846408885573377,
      "grad_norm": 0.4411889314651489,
      "learning_rate": 4.3078867752497e-06,
      "loss": 0.7817,
      "step": 3341
    },
    {
      "epoch": 0.13850553276140742,
      "grad_norm": 0.4161721169948578,
      "learning_rate": 4.3076795557213315e-06,
      "loss": 0.7742,
      "step": 3342
    },
    {
      "epoch": 0.1385469766670811,
      "grad_norm": 0.45293962955474854,
      "learning_rate": 4.307472336192963e-06,
      "loss": 0.7197,
      "step": 3343
    },
    {
      "epoch": 0.13858842057275478,
      "grad_norm": 0.46139729022979736,
      "learning_rate": 4.307265116664594e-06,
      "loss": 0.7737,
      "step": 3344
    },
    {
      "epoch": 0.13862986447842846,
      "grad_norm": 0.41932469606399536,
      "learning_rate": 4.3070578971362265e-06,
      "loss": 0.7228,
      "step": 3345
    },
    {
      "epoch": 0.1386713083841021,
      "grad_norm": 0.4702572226524353,
      "learning_rate": 4.306850677607859e-06,
      "loss": 0.7961,
      "step": 3346
    },
    {
      "epoch": 0.13871275228977578,
      "grad_norm": 0.4401925206184387,
      "learning_rate": 4.30664345807949e-06,
      "loss": 0.7523,
      "step": 3347
    },
    {
      "epoch": 0.13875419619544946,
      "grad_norm": 0.45952650904655457,
      "learning_rate": 4.3064362385511215e-06,
      "loss": 0.7715,
      "step": 3348
    },
    {
      "epoch": 0.13879564010112314,
      "grad_norm": 0.44228479266166687,
      "learning_rate": 4.306229019022753e-06,
      "loss": 0.7395,
      "step": 3349
    },
    {
      "epoch": 0.1388370840067968,
      "grad_norm": 0.4234450161457062,
      "learning_rate": 4.306021799494385e-06,
      "loss": 0.7694,
      "step": 3350
    },
    {
      "epoch": 0.13887852791247046,
      "grad_norm": 0.42631199955940247,
      "learning_rate": 4.3058145799660165e-06,
      "loss": 0.7473,
      "step": 3351
    },
    {
      "epoch": 0.13891997181814414,
      "grad_norm": 0.43943336606025696,
      "learning_rate": 4.305607360437648e-06,
      "loss": 0.7341,
      "step": 3352
    },
    {
      "epoch": 0.13896141572381782,
      "grad_norm": 0.3992968499660492,
      "learning_rate": 4.305400140909279e-06,
      "loss": 0.7251,
      "step": 3353
    },
    {
      "epoch": 0.1390028596294915,
      "grad_norm": 0.46236473321914673,
      "learning_rate": 4.3051929213809115e-06,
      "loss": 0.7545,
      "step": 3354
    },
    {
      "epoch": 0.13904430353516514,
      "grad_norm": 0.40973255038261414,
      "learning_rate": 4.304985701852543e-06,
      "loss": 0.7483,
      "step": 3355
    },
    {
      "epoch": 0.13908574744083882,
      "grad_norm": 0.4474811255931854,
      "learning_rate": 4.304778482324174e-06,
      "loss": 0.6654,
      "step": 3356
    },
    {
      "epoch": 0.1391271913465125,
      "grad_norm": 0.42747166752815247,
      "learning_rate": 4.3045712627958065e-06,
      "loss": 0.7522,
      "step": 3357
    },
    {
      "epoch": 0.13916863525218617,
      "grad_norm": 0.44262632727622986,
      "learning_rate": 4.304364043267438e-06,
      "loss": 0.7087,
      "step": 3358
    },
    {
      "epoch": 0.13921007915785982,
      "grad_norm": 0.39586323499679565,
      "learning_rate": 4.304156823739069e-06,
      "loss": 0.6575,
      "step": 3359
    },
    {
      "epoch": 0.1392515230635335,
      "grad_norm": 0.4459928572177887,
      "learning_rate": 4.303949604210701e-06,
      "loss": 0.7324,
      "step": 3360
    },
    {
      "epoch": 0.13929296696920718,
      "grad_norm": 0.4681191146373749,
      "learning_rate": 4.303742384682333e-06,
      "loss": 0.7527,
      "step": 3361
    },
    {
      "epoch": 0.13933441087488085,
      "grad_norm": 0.4402056932449341,
      "learning_rate": 4.303535165153964e-06,
      "loss": 0.7529,
      "step": 3362
    },
    {
      "epoch": 0.13937585478055453,
      "grad_norm": 0.4247872233390808,
      "learning_rate": 4.3033279456255965e-06,
      "loss": 0.7188,
      "step": 3363
    },
    {
      "epoch": 0.13941729868622818,
      "grad_norm": 0.4483569860458374,
      "learning_rate": 4.303120726097228e-06,
      "loss": 0.7046,
      "step": 3364
    },
    {
      "epoch": 0.13945874259190186,
      "grad_norm": 0.40512576699256897,
      "learning_rate": 4.302913506568859e-06,
      "loss": 0.7341,
      "step": 3365
    },
    {
      "epoch": 0.13950018649757553,
      "grad_norm": 0.4109475016593933,
      "learning_rate": 4.3027062870404915e-06,
      "loss": 0.7424,
      "step": 3366
    },
    {
      "epoch": 0.1395416304032492,
      "grad_norm": 0.47797730565071106,
      "learning_rate": 4.302499067512123e-06,
      "loss": 0.8013,
      "step": 3367
    },
    {
      "epoch": 0.13958307430892286,
      "grad_norm": 0.4332086145877838,
      "learning_rate": 4.302291847983754e-06,
      "loss": 0.6888,
      "step": 3368
    },
    {
      "epoch": 0.13962451821459654,
      "grad_norm": 0.42073312401771545,
      "learning_rate": 4.302084628455386e-06,
      "loss": 0.7136,
      "step": 3369
    },
    {
      "epoch": 0.13966596212027022,
      "grad_norm": 0.4236204922199249,
      "learning_rate": 4.301877408927018e-06,
      "loss": 0.7434,
      "step": 3370
    },
    {
      "epoch": 0.1397074060259439,
      "grad_norm": 0.41219261288642883,
      "learning_rate": 4.301670189398649e-06,
      "loss": 0.6863,
      "step": 3371
    },
    {
      "epoch": 0.13974884993161757,
      "grad_norm": 0.4015968441963196,
      "learning_rate": 4.301462969870281e-06,
      "loss": 0.7253,
      "step": 3372
    },
    {
      "epoch": 0.13979029383729122,
      "grad_norm": 0.43729519844055176,
      "learning_rate": 4.301255750341913e-06,
      "loss": 0.8142,
      "step": 3373
    },
    {
      "epoch": 0.1398317377429649,
      "grad_norm": 0.4236052334308624,
      "learning_rate": 4.301048530813544e-06,
      "loss": 0.761,
      "step": 3374
    },
    {
      "epoch": 0.13987318164863857,
      "grad_norm": 0.3898746371269226,
      "learning_rate": 4.300841311285176e-06,
      "loss": 0.6331,
      "step": 3375
    },
    {
      "epoch": 0.13991462555431225,
      "grad_norm": 0.4334551990032196,
      "learning_rate": 4.300634091756807e-06,
      "loss": 0.7014,
      "step": 3376
    },
    {
      "epoch": 0.1399560694599859,
      "grad_norm": 0.40040749311447144,
      "learning_rate": 4.300426872228439e-06,
      "loss": 0.6868,
      "step": 3377
    },
    {
      "epoch": 0.13999751336565958,
      "grad_norm": 0.4336819350719452,
      "learning_rate": 4.300219652700071e-06,
      "loss": 0.7603,
      "step": 3378
    },
    {
      "epoch": 0.14003895727133325,
      "grad_norm": 0.4108978509902954,
      "learning_rate": 4.300012433171702e-06,
      "loss": 0.7292,
      "step": 3379
    },
    {
      "epoch": 0.14008040117700693,
      "grad_norm": 0.4317830204963684,
      "learning_rate": 4.299805213643334e-06,
      "loss": 0.7084,
      "step": 3380
    },
    {
      "epoch": 0.14012184508268058,
      "grad_norm": 0.4439917802810669,
      "learning_rate": 4.299597994114966e-06,
      "loss": 0.734,
      "step": 3381
    },
    {
      "epoch": 0.14016328898835426,
      "grad_norm": 0.4173215627670288,
      "learning_rate": 4.299390774586598e-06,
      "loss": 0.731,
      "step": 3382
    },
    {
      "epoch": 0.14020473289402793,
      "grad_norm": 0.460292786359787,
      "learning_rate": 4.299183555058229e-06,
      "loss": 0.7661,
      "step": 3383
    },
    {
      "epoch": 0.1402461767997016,
      "grad_norm": 0.4207804799079895,
      "learning_rate": 4.298976335529861e-06,
      "loss": 0.719,
      "step": 3384
    },
    {
      "epoch": 0.1402876207053753,
      "grad_norm": 0.428084135055542,
      "learning_rate": 4.298769116001492e-06,
      "loss": 0.6895,
      "step": 3385
    },
    {
      "epoch": 0.14032906461104894,
      "grad_norm": 0.447248637676239,
      "learning_rate": 4.298561896473124e-06,
      "loss": 0.7157,
      "step": 3386
    },
    {
      "epoch": 0.14037050851672261,
      "grad_norm": 0.44536519050598145,
      "learning_rate": 4.298354676944756e-06,
      "loss": 0.7671,
      "step": 3387
    },
    {
      "epoch": 0.1404119524223963,
      "grad_norm": 0.41328883171081543,
      "learning_rate": 4.298147457416387e-06,
      "loss": 0.7095,
      "step": 3388
    },
    {
      "epoch": 0.14045339632806997,
      "grad_norm": 0.43724849820137024,
      "learning_rate": 4.297940237888019e-06,
      "loss": 0.7336,
      "step": 3389
    },
    {
      "epoch": 0.14049484023374362,
      "grad_norm": 0.46767181158065796,
      "learning_rate": 4.297733018359651e-06,
      "loss": 0.783,
      "step": 3390
    },
    {
      "epoch": 0.1405362841394173,
      "grad_norm": 0.4089604616165161,
      "learning_rate": 4.297525798831282e-06,
      "loss": 0.7812,
      "step": 3391
    },
    {
      "epoch": 0.14057772804509097,
      "grad_norm": 0.4268758296966553,
      "learning_rate": 4.2973185793029135e-06,
      "loss": 0.7498,
      "step": 3392
    },
    {
      "epoch": 0.14061917195076465,
      "grad_norm": 0.4097006320953369,
      "learning_rate": 4.297111359774546e-06,
      "loss": 0.7845,
      "step": 3393
    },
    {
      "epoch": 0.1406606158564383,
      "grad_norm": 0.4642042815685272,
      "learning_rate": 4.296904140246177e-06,
      "loss": 0.7285,
      "step": 3394
    },
    {
      "epoch": 0.14070205976211198,
      "grad_norm": 0.46710604429244995,
      "learning_rate": 4.2966969207178085e-06,
      "loss": 0.7606,
      "step": 3395
    },
    {
      "epoch": 0.14074350366778565,
      "grad_norm": 0.4127083122730255,
      "learning_rate": 4.29648970118944e-06,
      "loss": 0.6692,
      "step": 3396
    },
    {
      "epoch": 0.14078494757345933,
      "grad_norm": 0.4331100583076477,
      "learning_rate": 4.296282481661072e-06,
      "loss": 0.7859,
      "step": 3397
    },
    {
      "epoch": 0.140826391479133,
      "grad_norm": 0.43853187561035156,
      "learning_rate": 4.296075262132704e-06,
      "loss": 0.7822,
      "step": 3398
    },
    {
      "epoch": 0.14086783538480666,
      "grad_norm": 0.41086718440055847,
      "learning_rate": 4.295868042604336e-06,
      "loss": 0.7275,
      "step": 3399
    },
    {
      "epoch": 0.14090927929048033,
      "grad_norm": 0.41194015741348267,
      "learning_rate": 4.295660823075967e-06,
      "loss": 0.7148,
      "step": 3400
    },
    {
      "epoch": 0.140950723196154,
      "grad_norm": 0.4312911033630371,
      "learning_rate": 4.2954536035475985e-06,
      "loss": 0.7366,
      "step": 3401
    },
    {
      "epoch": 0.1409921671018277,
      "grad_norm": 0.38340893387794495,
      "learning_rate": 4.295246384019231e-06,
      "loss": 0.6613,
      "step": 3402
    },
    {
      "epoch": 0.14103361100750134,
      "grad_norm": 0.4471784830093384,
      "learning_rate": 4.295039164490862e-06,
      "loss": 0.7183,
      "step": 3403
    },
    {
      "epoch": 0.141075054913175,
      "grad_norm": 0.4324750602245331,
      "learning_rate": 4.2948319449624935e-06,
      "loss": 0.7515,
      "step": 3404
    },
    {
      "epoch": 0.1411164988188487,
      "grad_norm": 0.4392637610435486,
      "learning_rate": 4.294624725434125e-06,
      "loss": 0.7554,
      "step": 3405
    },
    {
      "epoch": 0.14115794272452237,
      "grad_norm": 0.4784635901451111,
      "learning_rate": 4.294417505905757e-06,
      "loss": 0.7439,
      "step": 3406
    },
    {
      "epoch": 0.14119938663019602,
      "grad_norm": 0.39625808596611023,
      "learning_rate": 4.2942102863773885e-06,
      "loss": 0.6875,
      "step": 3407
    },
    {
      "epoch": 0.1412408305358697,
      "grad_norm": 0.43132859468460083,
      "learning_rate": 4.29400306684902e-06,
      "loss": 0.78,
      "step": 3408
    },
    {
      "epoch": 0.14128227444154337,
      "grad_norm": 0.435649573802948,
      "learning_rate": 4.293795847320652e-06,
      "loss": 0.6616,
      "step": 3409
    },
    {
      "epoch": 0.14132371834721705,
      "grad_norm": 0.4050649404525757,
      "learning_rate": 4.2935886277922835e-06,
      "loss": 0.7496,
      "step": 3410
    },
    {
      "epoch": 0.14136516225289072,
      "grad_norm": 0.4385804831981659,
      "learning_rate": 4.293381408263915e-06,
      "loss": 0.7773,
      "step": 3411
    },
    {
      "epoch": 0.14140660615856437,
      "grad_norm": 0.42110100388526917,
      "learning_rate": 4.293174188735546e-06,
      "loss": 0.7549,
      "step": 3412
    },
    {
      "epoch": 0.14144805006423805,
      "grad_norm": 0.4377802610397339,
      "learning_rate": 4.2929669692071785e-06,
      "loss": 0.7051,
      "step": 3413
    },
    {
      "epoch": 0.14148949396991173,
      "grad_norm": 0.43398773670196533,
      "learning_rate": 4.29275974967881e-06,
      "loss": 0.7561,
      "step": 3414
    },
    {
      "epoch": 0.1415309378755854,
      "grad_norm": 0.4242144525051117,
      "learning_rate": 4.292552530150442e-06,
      "loss": 0.7158,
      "step": 3415
    },
    {
      "epoch": 0.14157238178125905,
      "grad_norm": 0.49774014949798584,
      "learning_rate": 4.2923453106220735e-06,
      "loss": 0.7537,
      "step": 3416
    },
    {
      "epoch": 0.14161382568693273,
      "grad_norm": 0.4257213771343231,
      "learning_rate": 4.292138091093705e-06,
      "loss": 0.7463,
      "step": 3417
    },
    {
      "epoch": 0.1416552695926064,
      "grad_norm": 0.4529542028903961,
      "learning_rate": 4.291930871565337e-06,
      "loss": 0.7406,
      "step": 3418
    },
    {
      "epoch": 0.14169671349828009,
      "grad_norm": 0.40651723742485046,
      "learning_rate": 4.2917236520369685e-06,
      "loss": 0.7239,
      "step": 3419
    },
    {
      "epoch": 0.14173815740395374,
      "grad_norm": 0.4605000615119934,
      "learning_rate": 4.2915164325086e-06,
      "loss": 0.7917,
      "step": 3420
    },
    {
      "epoch": 0.1417796013096274,
      "grad_norm": 0.46843579411506653,
      "learning_rate": 4.291309212980231e-06,
      "loss": 0.7854,
      "step": 3421
    },
    {
      "epoch": 0.1418210452153011,
      "grad_norm": 0.42019957304000854,
      "learning_rate": 4.2911019934518635e-06,
      "loss": 0.7336,
      "step": 3422
    },
    {
      "epoch": 0.14186248912097477,
      "grad_norm": 0.45614132285118103,
      "learning_rate": 4.290894773923495e-06,
      "loss": 0.7771,
      "step": 3423
    },
    {
      "epoch": 0.14190393302664844,
      "grad_norm": 0.4390544593334198,
      "learning_rate": 4.290687554395126e-06,
      "loss": 0.7145,
      "step": 3424
    },
    {
      "epoch": 0.1419453769323221,
      "grad_norm": 0.46028438210487366,
      "learning_rate": 4.2904803348667585e-06,
      "loss": 0.7913,
      "step": 3425
    },
    {
      "epoch": 0.14198682083799577,
      "grad_norm": 0.4318721294403076,
      "learning_rate": 4.29027311533839e-06,
      "loss": 0.7617,
      "step": 3426
    },
    {
      "epoch": 0.14202826474366945,
      "grad_norm": 0.44680488109588623,
      "learning_rate": 4.290065895810021e-06,
      "loss": 0.8103,
      "step": 3427
    },
    {
      "epoch": 0.14206970864934312,
      "grad_norm": 0.44753387570381165,
      "learning_rate": 4.289858676281653e-06,
      "loss": 0.749,
      "step": 3428
    },
    {
      "epoch": 0.14211115255501677,
      "grad_norm": 0.41088226437568665,
      "learning_rate": 4.289651456753285e-06,
      "loss": 0.7596,
      "step": 3429
    },
    {
      "epoch": 0.14215259646069045,
      "grad_norm": 0.4437929689884186,
      "learning_rate": 4.289444237224916e-06,
      "loss": 0.71,
      "step": 3430
    },
    {
      "epoch": 0.14219404036636413,
      "grad_norm": 0.43385204672813416,
      "learning_rate": 4.2892370176965485e-06,
      "loss": 0.735,
      "step": 3431
    },
    {
      "epoch": 0.1422354842720378,
      "grad_norm": 0.442533940076828,
      "learning_rate": 4.28902979816818e-06,
      "loss": 0.7471,
      "step": 3432
    },
    {
      "epoch": 0.14227692817771148,
      "grad_norm": 0.4379854202270508,
      "learning_rate": 4.288822578639811e-06,
      "loss": 0.7002,
      "step": 3433
    },
    {
      "epoch": 0.14231837208338513,
      "grad_norm": 0.4227604269981384,
      "learning_rate": 4.2886153591114435e-06,
      "loss": 0.7241,
      "step": 3434
    },
    {
      "epoch": 0.1423598159890588,
      "grad_norm": 0.4352363049983978,
      "learning_rate": 4.288408139583075e-06,
      "loss": 0.7261,
      "step": 3435
    },
    {
      "epoch": 0.14240125989473248,
      "grad_norm": 0.4293971359729767,
      "learning_rate": 4.288200920054706e-06,
      "loss": 0.6913,
      "step": 3436
    },
    {
      "epoch": 0.14244270380040616,
      "grad_norm": 0.54514080286026,
      "learning_rate": 4.287993700526338e-06,
      "loss": 0.7251,
      "step": 3437
    },
    {
      "epoch": 0.1424841477060798,
      "grad_norm": 0.42512738704681396,
      "learning_rate": 4.28778648099797e-06,
      "loss": 0.7371,
      "step": 3438
    },
    {
      "epoch": 0.1425255916117535,
      "grad_norm": 0.45656123757362366,
      "learning_rate": 4.287579261469601e-06,
      "loss": 0.772,
      "step": 3439
    },
    {
      "epoch": 0.14256703551742717,
      "grad_norm": 0.43020185828208923,
      "learning_rate": 4.287372041941233e-06,
      "loss": 0.7285,
      "step": 3440
    },
    {
      "epoch": 0.14260847942310084,
      "grad_norm": 0.43705663084983826,
      "learning_rate": 4.287164822412865e-06,
      "loss": 0.6764,
      "step": 3441
    },
    {
      "epoch": 0.1426499233287745,
      "grad_norm": 0.45319026708602905,
      "learning_rate": 4.286957602884496e-06,
      "loss": 0.7238,
      "step": 3442
    },
    {
      "epoch": 0.14269136723444817,
      "grad_norm": 0.40755826234817505,
      "learning_rate": 4.286750383356128e-06,
      "loss": 0.6689,
      "step": 3443
    },
    {
      "epoch": 0.14273281114012185,
      "grad_norm": 0.4272681176662445,
      "learning_rate": 4.286543163827759e-06,
      "loss": 0.7615,
      "step": 3444
    },
    {
      "epoch": 0.14277425504579552,
      "grad_norm": 0.4659278094768524,
      "learning_rate": 4.286335944299391e-06,
      "loss": 0.7437,
      "step": 3445
    },
    {
      "epoch": 0.1428156989514692,
      "grad_norm": 0.4307320713996887,
      "learning_rate": 4.286128724771023e-06,
      "loss": 0.744,
      "step": 3446
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.45658209919929504,
      "learning_rate": 4.285921505242654e-06,
      "loss": 0.7683,
      "step": 3447
    },
    {
      "epoch": 0.14289858676281653,
      "grad_norm": 0.4416615068912506,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.7361,
      "step": 3448
    },
    {
      "epoch": 0.1429400306684902,
      "grad_norm": 0.4493853747844696,
      "learning_rate": 4.285507066185918e-06,
      "loss": 0.7104,
      "step": 3449
    },
    {
      "epoch": 0.14298147457416388,
      "grad_norm": 0.38244369626045227,
      "learning_rate": 4.28529984665755e-06,
      "loss": 0.7253,
      "step": 3450
    },
    {
      "epoch": 0.14302291847983753,
      "grad_norm": 0.4370732307434082,
      "learning_rate": 4.285092627129181e-06,
      "loss": 0.7568,
      "step": 3451
    },
    {
      "epoch": 0.1430643623855112,
      "grad_norm": 0.4673200845718384,
      "learning_rate": 4.284885407600813e-06,
      "loss": 0.7708,
      "step": 3452
    },
    {
      "epoch": 0.14310580629118488,
      "grad_norm": 0.461574524641037,
      "learning_rate": 4.284678188072444e-06,
      "loss": 0.7571,
      "step": 3453
    },
    {
      "epoch": 0.14314725019685856,
      "grad_norm": 0.4427507519721985,
      "learning_rate": 4.284470968544076e-06,
      "loss": 0.7747,
      "step": 3454
    },
    {
      "epoch": 0.1431886941025322,
      "grad_norm": 0.42441776394844055,
      "learning_rate": 4.284263749015708e-06,
      "loss": 0.7646,
      "step": 3455
    },
    {
      "epoch": 0.1432301380082059,
      "grad_norm": 0.4430643320083618,
      "learning_rate": 4.284056529487339e-06,
      "loss": 0.7299,
      "step": 3456
    },
    {
      "epoch": 0.14327158191387956,
      "grad_norm": 0.4265633225440979,
      "learning_rate": 4.2838493099589705e-06,
      "loss": 0.6846,
      "step": 3457
    },
    {
      "epoch": 0.14331302581955324,
      "grad_norm": 0.4507368505001068,
      "learning_rate": 4.283642090430603e-06,
      "loss": 0.7822,
      "step": 3458
    },
    {
      "epoch": 0.14335446972522692,
      "grad_norm": 0.40229934453964233,
      "learning_rate": 4.283434870902234e-06,
      "loss": 0.7261,
      "step": 3459
    },
    {
      "epoch": 0.14339591363090057,
      "grad_norm": 0.4339674413204193,
      "learning_rate": 4.2832276513738655e-06,
      "loss": 0.6611,
      "step": 3460
    },
    {
      "epoch": 0.14343735753657424,
      "grad_norm": 0.424968421459198,
      "learning_rate": 4.283020431845498e-06,
      "loss": 0.7578,
      "step": 3461
    },
    {
      "epoch": 0.14347880144224792,
      "grad_norm": 0.4391248822212219,
      "learning_rate": 4.282813212317129e-06,
      "loss": 0.7832,
      "step": 3462
    },
    {
      "epoch": 0.1435202453479216,
      "grad_norm": 0.47979751229286194,
      "learning_rate": 4.2826059927887605e-06,
      "loss": 0.7716,
      "step": 3463
    },
    {
      "epoch": 0.14356168925359525,
      "grad_norm": 0.44123777747154236,
      "learning_rate": 4.282398773260392e-06,
      "loss": 0.7141,
      "step": 3464
    },
    {
      "epoch": 0.14360313315926893,
      "grad_norm": 0.46980735659599304,
      "learning_rate": 4.282191553732024e-06,
      "loss": 0.7388,
      "step": 3465
    },
    {
      "epoch": 0.1436445770649426,
      "grad_norm": 0.40594521164894104,
      "learning_rate": 4.2819843342036555e-06,
      "loss": 0.696,
      "step": 3466
    },
    {
      "epoch": 0.14368602097061628,
      "grad_norm": 0.4325354993343353,
      "learning_rate": 4.281777114675288e-06,
      "loss": 0.7344,
      "step": 3467
    },
    {
      "epoch": 0.14372746487628993,
      "grad_norm": 0.42500731348991394,
      "learning_rate": 4.281569895146919e-06,
      "loss": 0.6895,
      "step": 3468
    },
    {
      "epoch": 0.1437689087819636,
      "grad_norm": 0.43492645025253296,
      "learning_rate": 4.2813626756185505e-06,
      "loss": 0.6851,
      "step": 3469
    },
    {
      "epoch": 0.14381035268763728,
      "grad_norm": 0.42430299520492554,
      "learning_rate": 4.281155456090183e-06,
      "loss": 0.6852,
      "step": 3470
    },
    {
      "epoch": 0.14385179659331096,
      "grad_norm": 0.4382709860801697,
      "learning_rate": 4.280948236561814e-06,
      "loss": 0.7198,
      "step": 3471
    },
    {
      "epoch": 0.14389324049898464,
      "grad_norm": 0.4364079535007477,
      "learning_rate": 4.2807410170334455e-06,
      "loss": 0.7444,
      "step": 3472
    },
    {
      "epoch": 0.1439346844046583,
      "grad_norm": 0.456495463848114,
      "learning_rate": 4.280533797505077e-06,
      "loss": 0.7461,
      "step": 3473
    },
    {
      "epoch": 0.14397612831033196,
      "grad_norm": 0.44718798995018005,
      "learning_rate": 4.280326577976709e-06,
      "loss": 0.7671,
      "step": 3474
    },
    {
      "epoch": 0.14401757221600564,
      "grad_norm": 0.4819701313972473,
      "learning_rate": 4.2801193584483405e-06,
      "loss": 0.7366,
      "step": 3475
    },
    {
      "epoch": 0.14405901612167932,
      "grad_norm": 0.4162538945674896,
      "learning_rate": 4.279912138919972e-06,
      "loss": 0.7332,
      "step": 3476
    },
    {
      "epoch": 0.14410046002735297,
      "grad_norm": 0.5408386588096619,
      "learning_rate": 4.279704919391604e-06,
      "loss": 0.7814,
      "step": 3477
    },
    {
      "epoch": 0.14414190393302664,
      "grad_norm": 0.43330806493759155,
      "learning_rate": 4.2794976998632355e-06,
      "loss": 0.7573,
      "step": 3478
    },
    {
      "epoch": 0.14418334783870032,
      "grad_norm": 0.4250098764896393,
      "learning_rate": 4.279290480334867e-06,
      "loss": 0.7593,
      "step": 3479
    },
    {
      "epoch": 0.144224791744374,
      "grad_norm": 0.4699741303920746,
      "learning_rate": 4.279083260806498e-06,
      "loss": 0.7786,
      "step": 3480
    },
    {
      "epoch": 0.14426623565004765,
      "grad_norm": 0.442393958568573,
      "learning_rate": 4.2788760412781305e-06,
      "loss": 0.7375,
      "step": 3481
    },
    {
      "epoch": 0.14430767955572132,
      "grad_norm": 0.4113013446331024,
      "learning_rate": 4.278668821749762e-06,
      "loss": 0.7412,
      "step": 3482
    },
    {
      "epoch": 0.144349123461395,
      "grad_norm": 0.46772781014442444,
      "learning_rate": 4.278461602221394e-06,
      "loss": 0.7527,
      "step": 3483
    },
    {
      "epoch": 0.14439056736706868,
      "grad_norm": 0.41785454750061035,
      "learning_rate": 4.278254382693025e-06,
      "loss": 0.7554,
      "step": 3484
    },
    {
      "epoch": 0.14443201127274236,
      "grad_norm": 0.4000510275363922,
      "learning_rate": 4.278047163164657e-06,
      "loss": 0.7734,
      "step": 3485
    },
    {
      "epoch": 0.144473455178416,
      "grad_norm": 0.4850148856639862,
      "learning_rate": 4.277839943636289e-06,
      "loss": 0.7705,
      "step": 3486
    },
    {
      "epoch": 0.14451489908408968,
      "grad_norm": 0.4162720739841461,
      "learning_rate": 4.2776327241079205e-06,
      "loss": 0.738,
      "step": 3487
    },
    {
      "epoch": 0.14455634298976336,
      "grad_norm": 0.4242589771747589,
      "learning_rate": 4.277425504579552e-06,
      "loss": 0.7737,
      "step": 3488
    },
    {
      "epoch": 0.14459778689543704,
      "grad_norm": 0.41584208607673645,
      "learning_rate": 4.277218285051183e-06,
      "loss": 0.7625,
      "step": 3489
    },
    {
      "epoch": 0.14463923080111069,
      "grad_norm": 0.4112046957015991,
      "learning_rate": 4.2770110655228155e-06,
      "loss": 0.7417,
      "step": 3490
    },
    {
      "epoch": 0.14468067470678436,
      "grad_norm": 0.3992477059364319,
      "learning_rate": 4.276803845994447e-06,
      "loss": 0.6951,
      "step": 3491
    },
    {
      "epoch": 0.14472211861245804,
      "grad_norm": 0.418252557516098,
      "learning_rate": 4.276596626466078e-06,
      "loss": 0.7751,
      "step": 3492
    },
    {
      "epoch": 0.14476356251813172,
      "grad_norm": 0.4534588158130646,
      "learning_rate": 4.2763894069377105e-06,
      "loss": 0.7522,
      "step": 3493
    },
    {
      "epoch": 0.14480500642380537,
      "grad_norm": 0.43014875054359436,
      "learning_rate": 4.276182187409342e-06,
      "loss": 0.7549,
      "step": 3494
    },
    {
      "epoch": 0.14484645032947904,
      "grad_norm": 0.437710702419281,
      "learning_rate": 4.275974967880973e-06,
      "loss": 0.7339,
      "step": 3495
    },
    {
      "epoch": 0.14488789423515272,
      "grad_norm": 0.39932766556739807,
      "learning_rate": 4.275767748352605e-06,
      "loss": 0.7107,
      "step": 3496
    },
    {
      "epoch": 0.1449293381408264,
      "grad_norm": 0.4366492033004761,
      "learning_rate": 4.275560528824237e-06,
      "loss": 0.803,
      "step": 3497
    },
    {
      "epoch": 0.14497078204650007,
      "grad_norm": 0.45354920625686646,
      "learning_rate": 4.275353309295868e-06,
      "loss": 0.7188,
      "step": 3498
    },
    {
      "epoch": 0.14501222595217372,
      "grad_norm": 0.4322458803653717,
      "learning_rate": 4.2751460897675005e-06,
      "loss": 0.7305,
      "step": 3499
    },
    {
      "epoch": 0.1450536698578474,
      "grad_norm": 0.3932023048400879,
      "learning_rate": 4.274938870239131e-06,
      "loss": 0.7068,
      "step": 3500
    },
    {
      "epoch": 0.14509511376352108,
      "grad_norm": 0.4517843723297119,
      "learning_rate": 4.274731650710763e-06,
      "loss": 0.7703,
      "step": 3501
    },
    {
      "epoch": 0.14513655766919475,
      "grad_norm": 0.4348665475845337,
      "learning_rate": 4.2745244311823955e-06,
      "loss": 0.7231,
      "step": 3502
    },
    {
      "epoch": 0.1451780015748684,
      "grad_norm": 0.4205609858036041,
      "learning_rate": 4.274317211654027e-06,
      "loss": 0.7456,
      "step": 3503
    },
    {
      "epoch": 0.14521944548054208,
      "grad_norm": 0.42239752411842346,
      "learning_rate": 4.274109992125658e-06,
      "loss": 0.7769,
      "step": 3504
    },
    {
      "epoch": 0.14526088938621576,
      "grad_norm": 0.41269490122795105,
      "learning_rate": 4.27390277259729e-06,
      "loss": 0.7185,
      "step": 3505
    },
    {
      "epoch": 0.14530233329188944,
      "grad_norm": 0.4312281906604767,
      "learning_rate": 4.273695553068922e-06,
      "loss": 0.7313,
      "step": 3506
    },
    {
      "epoch": 0.1453437771975631,
      "grad_norm": 0.4436001777648926,
      "learning_rate": 4.273488333540553e-06,
      "loss": 0.7188,
      "step": 3507
    },
    {
      "epoch": 0.14538522110323676,
      "grad_norm": 0.46300846338272095,
      "learning_rate": 4.273281114012185e-06,
      "loss": 0.7646,
      "step": 3508
    },
    {
      "epoch": 0.14542666500891044,
      "grad_norm": 0.42279136180877686,
      "learning_rate": 4.273073894483816e-06,
      "loss": 0.7061,
      "step": 3509
    },
    {
      "epoch": 0.14546810891458412,
      "grad_norm": 0.4688092768192291,
      "learning_rate": 4.272866674955448e-06,
      "loss": 0.7526,
      "step": 3510
    },
    {
      "epoch": 0.1455095528202578,
      "grad_norm": 0.6708021759986877,
      "learning_rate": 4.27265945542708e-06,
      "loss": 0.7153,
      "step": 3511
    },
    {
      "epoch": 0.14555099672593144,
      "grad_norm": 0.4083500802516937,
      "learning_rate": 4.272452235898711e-06,
      "loss": 0.7288,
      "step": 3512
    },
    {
      "epoch": 0.14559244063160512,
      "grad_norm": 0.43831339478492737,
      "learning_rate": 4.272245016370343e-06,
      "loss": 0.7676,
      "step": 3513
    },
    {
      "epoch": 0.1456338845372788,
      "grad_norm": 0.4630539119243622,
      "learning_rate": 4.272037796841975e-06,
      "loss": 0.7388,
      "step": 3514
    },
    {
      "epoch": 0.14567532844295247,
      "grad_norm": 0.4154176414012909,
      "learning_rate": 4.271830577313606e-06,
      "loss": 0.6853,
      "step": 3515
    },
    {
      "epoch": 0.14571677234862612,
      "grad_norm": 0.4656500816345215,
      "learning_rate": 4.2716233577852375e-06,
      "loss": 0.7324,
      "step": 3516
    },
    {
      "epoch": 0.1457582162542998,
      "grad_norm": 0.40697988867759705,
      "learning_rate": 4.27141613825687e-06,
      "loss": 0.7178,
      "step": 3517
    },
    {
      "epoch": 0.14579966015997348,
      "grad_norm": 0.44282448291778564,
      "learning_rate": 4.271208918728501e-06,
      "loss": 0.6907,
      "step": 3518
    },
    {
      "epoch": 0.14584110406564715,
      "grad_norm": 0.40109995007514954,
      "learning_rate": 4.271001699200133e-06,
      "loss": 0.6815,
      "step": 3519
    },
    {
      "epoch": 0.14588254797132083,
      "grad_norm": 0.4698004722595215,
      "learning_rate": 4.270794479671765e-06,
      "loss": 0.7817,
      "step": 3520
    },
    {
      "epoch": 0.14592399187699448,
      "grad_norm": 0.4324088990688324,
      "learning_rate": 4.270587260143396e-06,
      "loss": 0.7324,
      "step": 3521
    },
    {
      "epoch": 0.14596543578266816,
      "grad_norm": 0.4603823125362396,
      "learning_rate": 4.270380040615028e-06,
      "loss": 0.741,
      "step": 3522
    },
    {
      "epoch": 0.14600687968834183,
      "grad_norm": 0.41372841596603394,
      "learning_rate": 4.27017282108666e-06,
      "loss": 0.726,
      "step": 3523
    },
    {
      "epoch": 0.1460483235940155,
      "grad_norm": 0.42065271735191345,
      "learning_rate": 4.269965601558291e-06,
      "loss": 0.7506,
      "step": 3524
    },
    {
      "epoch": 0.14608976749968916,
      "grad_norm": 0.4407220482826233,
      "learning_rate": 4.2697583820299225e-06,
      "loss": 0.7522,
      "step": 3525
    },
    {
      "epoch": 0.14613121140536284,
      "grad_norm": 0.46513113379478455,
      "learning_rate": 4.269551162501555e-06,
      "loss": 0.7695,
      "step": 3526
    },
    {
      "epoch": 0.14617265531103651,
      "grad_norm": 0.4022018313407898,
      "learning_rate": 4.269343942973186e-06,
      "loss": 0.7001,
      "step": 3527
    },
    {
      "epoch": 0.1462140992167102,
      "grad_norm": 0.4551241397857666,
      "learning_rate": 4.2691367234448175e-06,
      "loss": 0.7739,
      "step": 3528
    },
    {
      "epoch": 0.14625554312238384,
      "grad_norm": 0.4327596127986908,
      "learning_rate": 4.26892950391645e-06,
      "loss": 0.723,
      "step": 3529
    },
    {
      "epoch": 0.14629698702805752,
      "grad_norm": 0.41593167185783386,
      "learning_rate": 4.268722284388081e-06,
      "loss": 0.7646,
      "step": 3530
    },
    {
      "epoch": 0.1463384309337312,
      "grad_norm": 0.4278425872325897,
      "learning_rate": 4.2685150648597125e-06,
      "loss": 0.6973,
      "step": 3531
    },
    {
      "epoch": 0.14637987483940487,
      "grad_norm": 0.43814510107040405,
      "learning_rate": 4.268307845331344e-06,
      "loss": 0.7639,
      "step": 3532
    },
    {
      "epoch": 0.14642131874507855,
      "grad_norm": 0.43373367190361023,
      "learning_rate": 4.268100625802976e-06,
      "loss": 0.7402,
      "step": 3533
    },
    {
      "epoch": 0.1464627626507522,
      "grad_norm": 0.44189921021461487,
      "learning_rate": 4.2678934062746075e-06,
      "loss": 0.73,
      "step": 3534
    },
    {
      "epoch": 0.14650420655642588,
      "grad_norm": 0.4363210201263428,
      "learning_rate": 4.26768618674624e-06,
      "loss": 0.7893,
      "step": 3535
    },
    {
      "epoch": 0.14654565046209955,
      "grad_norm": 0.4836121201515198,
      "learning_rate": 4.267478967217871e-06,
      "loss": 0.8115,
      "step": 3536
    },
    {
      "epoch": 0.14658709436777323,
      "grad_norm": 0.45794326066970825,
      "learning_rate": 4.2672717476895025e-06,
      "loss": 0.7864,
      "step": 3537
    },
    {
      "epoch": 0.14662853827344688,
      "grad_norm": 0.4145175814628601,
      "learning_rate": 4.267064528161135e-06,
      "loss": 0.7749,
      "step": 3538
    },
    {
      "epoch": 0.14666998217912056,
      "grad_norm": 0.4149456024169922,
      "learning_rate": 4.266857308632766e-06,
      "loss": 0.6948,
      "step": 3539
    },
    {
      "epoch": 0.14671142608479423,
      "grad_norm": 0.4114832282066345,
      "learning_rate": 4.2666500891043975e-06,
      "loss": 0.7629,
      "step": 3540
    },
    {
      "epoch": 0.1467528699904679,
      "grad_norm": 0.4136492609977722,
      "learning_rate": 4.266442869576029e-06,
      "loss": 0.7,
      "step": 3541
    },
    {
      "epoch": 0.14679431389614156,
      "grad_norm": 0.447996586561203,
      "learning_rate": 4.266235650047661e-06,
      "loss": 0.7268,
      "step": 3542
    },
    {
      "epoch": 0.14683575780181524,
      "grad_norm": 0.4522111117839813,
      "learning_rate": 4.2660284305192925e-06,
      "loss": 0.7478,
      "step": 3543
    },
    {
      "epoch": 0.1468772017074889,
      "grad_norm": 0.44529223442077637,
      "learning_rate": 4.265821210990924e-06,
      "loss": 0.7354,
      "step": 3544
    },
    {
      "epoch": 0.1469186456131626,
      "grad_norm": 0.45901861786842346,
      "learning_rate": 4.265613991462555e-06,
      "loss": 0.7432,
      "step": 3545
    },
    {
      "epoch": 0.14696008951883627,
      "grad_norm": 0.4476220905780792,
      "learning_rate": 4.2654067719341875e-06,
      "loss": 0.7363,
      "step": 3546
    },
    {
      "epoch": 0.14700153342450992,
      "grad_norm": 0.4382518231868744,
      "learning_rate": 4.265199552405819e-06,
      "loss": 0.7825,
      "step": 3547
    },
    {
      "epoch": 0.1470429773301836,
      "grad_norm": 0.42589089274406433,
      "learning_rate": 4.26499233287745e-06,
      "loss": 0.7344,
      "step": 3548
    },
    {
      "epoch": 0.14708442123585727,
      "grad_norm": 0.46473634243011475,
      "learning_rate": 4.2647851133490825e-06,
      "loss": 0.7041,
      "step": 3549
    },
    {
      "epoch": 0.14712586514153095,
      "grad_norm": 0.4355294704437256,
      "learning_rate": 4.264577893820714e-06,
      "loss": 0.7573,
      "step": 3550
    },
    {
      "epoch": 0.1471673090472046,
      "grad_norm": 0.429289847612381,
      "learning_rate": 4.264370674292346e-06,
      "loss": 0.7227,
      "step": 3551
    },
    {
      "epoch": 0.14720875295287827,
      "grad_norm": 0.43157416582107544,
      "learning_rate": 4.264163454763977e-06,
      "loss": 0.7251,
      "step": 3552
    },
    {
      "epoch": 0.14725019685855195,
      "grad_norm": 0.4183496832847595,
      "learning_rate": 4.263956235235609e-06,
      "loss": 0.7416,
      "step": 3553
    },
    {
      "epoch": 0.14729164076422563,
      "grad_norm": 0.44607776403427124,
      "learning_rate": 4.263749015707241e-06,
      "loss": 0.7109,
      "step": 3554
    },
    {
      "epoch": 0.14733308466989928,
      "grad_norm": 0.4660865366458893,
      "learning_rate": 4.2635417961788725e-06,
      "loss": 0.803,
      "step": 3555
    },
    {
      "epoch": 0.14737452857557296,
      "grad_norm": 0.42630720138549805,
      "learning_rate": 4.263334576650504e-06,
      "loss": 0.7761,
      "step": 3556
    },
    {
      "epoch": 0.14741597248124663,
      "grad_norm": 0.41773396730422974,
      "learning_rate": 4.263127357122135e-06,
      "loss": 0.7043,
      "step": 3557
    },
    {
      "epoch": 0.1474574163869203,
      "grad_norm": 0.4292723536491394,
      "learning_rate": 4.2629201375937675e-06,
      "loss": 0.7739,
      "step": 3558
    },
    {
      "epoch": 0.147498860292594,
      "grad_norm": 0.4077301323413849,
      "learning_rate": 4.262712918065399e-06,
      "loss": 0.7283,
      "step": 3559
    },
    {
      "epoch": 0.14754030419826764,
      "grad_norm": 0.4134046137332916,
      "learning_rate": 4.26250569853703e-06,
      "loss": 0.7029,
      "step": 3560
    },
    {
      "epoch": 0.1475817481039413,
      "grad_norm": 0.438385546207428,
      "learning_rate": 4.262298479008662e-06,
      "loss": 0.7144,
      "step": 3561
    },
    {
      "epoch": 0.147623192009615,
      "grad_norm": 0.47375229001045227,
      "learning_rate": 4.262091259480294e-06,
      "loss": 0.7732,
      "step": 3562
    },
    {
      "epoch": 0.14766463591528867,
      "grad_norm": 0.4498146176338196,
      "learning_rate": 4.261884039951925e-06,
      "loss": 0.762,
      "step": 3563
    },
    {
      "epoch": 0.14770607982096232,
      "grad_norm": 0.4412408173084259,
      "learning_rate": 4.261676820423557e-06,
      "loss": 0.7686,
      "step": 3564
    },
    {
      "epoch": 0.147747523726636,
      "grad_norm": 0.39256995916366577,
      "learning_rate": 4.261469600895189e-06,
      "loss": 0.6788,
      "step": 3565
    },
    {
      "epoch": 0.14778896763230967,
      "grad_norm": 0.43233808875083923,
      "learning_rate": 4.26126238136682e-06,
      "loss": 0.7018,
      "step": 3566
    },
    {
      "epoch": 0.14783041153798335,
      "grad_norm": 0.41340142488479614,
      "learning_rate": 4.2610551618384525e-06,
      "loss": 0.7051,
      "step": 3567
    },
    {
      "epoch": 0.147871855443657,
      "grad_norm": 0.451794296503067,
      "learning_rate": 4.260847942310083e-06,
      "loss": 0.7329,
      "step": 3568
    },
    {
      "epoch": 0.14791329934933067,
      "grad_norm": 0.4397566318511963,
      "learning_rate": 4.260640722781715e-06,
      "loss": 0.6978,
      "step": 3569
    },
    {
      "epoch": 0.14795474325500435,
      "grad_norm": 0.41024884581565857,
      "learning_rate": 4.260433503253347e-06,
      "loss": 0.6921,
      "step": 3570
    },
    {
      "epoch": 0.14799618716067803,
      "grad_norm": 0.4070987403392792,
      "learning_rate": 4.260226283724979e-06,
      "loss": 0.7483,
      "step": 3571
    },
    {
      "epoch": 0.1480376310663517,
      "grad_norm": 0.4577427804470062,
      "learning_rate": 4.26001906419661e-06,
      "loss": 0.6964,
      "step": 3572
    },
    {
      "epoch": 0.14807907497202535,
      "grad_norm": 0.44319581985473633,
      "learning_rate": 4.259811844668242e-06,
      "loss": 0.7606,
      "step": 3573
    },
    {
      "epoch": 0.14812051887769903,
      "grad_norm": 0.4242476522922516,
      "learning_rate": 4.259604625139874e-06,
      "loss": 0.7629,
      "step": 3574
    },
    {
      "epoch": 0.1481619627833727,
      "grad_norm": 0.43759962916374207,
      "learning_rate": 4.259397405611505e-06,
      "loss": 0.7307,
      "step": 3575
    },
    {
      "epoch": 0.14820340668904639,
      "grad_norm": 0.41502565145492554,
      "learning_rate": 4.259190186083137e-06,
      "loss": 0.751,
      "step": 3576
    },
    {
      "epoch": 0.14824485059472003,
      "grad_norm": 0.4376647472381592,
      "learning_rate": 4.258982966554768e-06,
      "loss": 0.7952,
      "step": 3577
    },
    {
      "epoch": 0.1482862945003937,
      "grad_norm": 0.423888623714447,
      "learning_rate": 4.2587757470264e-06,
      "loss": 0.724,
      "step": 3578
    },
    {
      "epoch": 0.1483277384060674,
      "grad_norm": 0.4060913622379303,
      "learning_rate": 4.258568527498032e-06,
      "loss": 0.7397,
      "step": 3579
    },
    {
      "epoch": 0.14836918231174107,
      "grad_norm": 0.41397637128829956,
      "learning_rate": 4.258361307969663e-06,
      "loss": 0.7385,
      "step": 3580
    },
    {
      "epoch": 0.14841062621741474,
      "grad_norm": 0.42415210604667664,
      "learning_rate": 4.258154088441295e-06,
      "loss": 0.7397,
      "step": 3581
    },
    {
      "epoch": 0.1484520701230884,
      "grad_norm": 0.42841020226478577,
      "learning_rate": 4.257946868912927e-06,
      "loss": 0.7101,
      "step": 3582
    },
    {
      "epoch": 0.14849351402876207,
      "grad_norm": 0.3950115144252777,
      "learning_rate": 4.257739649384559e-06,
      "loss": 0.7822,
      "step": 3583
    },
    {
      "epoch": 0.14853495793443575,
      "grad_norm": 0.41325604915618896,
      "learning_rate": 4.2575324298561895e-06,
      "loss": 0.7448,
      "step": 3584
    },
    {
      "epoch": 0.14857640184010942,
      "grad_norm": 0.4179607331752777,
      "learning_rate": 4.257325210327822e-06,
      "loss": 0.7587,
      "step": 3585
    },
    {
      "epoch": 0.14861784574578307,
      "grad_norm": 0.44397249817848206,
      "learning_rate": 4.257117990799453e-06,
      "loss": 0.7317,
      "step": 3586
    },
    {
      "epoch": 0.14865928965145675,
      "grad_norm": 0.423122763633728,
      "learning_rate": 4.256910771271085e-06,
      "loss": 0.6818,
      "step": 3587
    },
    {
      "epoch": 0.14870073355713043,
      "grad_norm": 0.43470704555511475,
      "learning_rate": 4.256703551742717e-06,
      "loss": 0.702,
      "step": 3588
    },
    {
      "epoch": 0.1487421774628041,
      "grad_norm": 0.4402065873146057,
      "learning_rate": 4.256496332214348e-06,
      "loss": 0.7617,
      "step": 3589
    },
    {
      "epoch": 0.14878362136847775,
      "grad_norm": 0.44065746665000916,
      "learning_rate": 4.25628911268598e-06,
      "loss": 0.8018,
      "step": 3590
    },
    {
      "epoch": 0.14882506527415143,
      "grad_norm": 0.4382827579975128,
      "learning_rate": 4.256081893157612e-06,
      "loss": 0.6772,
      "step": 3591
    },
    {
      "epoch": 0.1488665091798251,
      "grad_norm": 0.4135058522224426,
      "learning_rate": 4.255874673629243e-06,
      "loss": 0.7045,
      "step": 3592
    },
    {
      "epoch": 0.14890795308549878,
      "grad_norm": 0.40508443117141724,
      "learning_rate": 4.2556674541008745e-06,
      "loss": 0.7466,
      "step": 3593
    },
    {
      "epoch": 0.14894939699117246,
      "grad_norm": 0.3963937759399414,
      "learning_rate": 4.255460234572507e-06,
      "loss": 0.7258,
      "step": 3594
    },
    {
      "epoch": 0.1489908408968461,
      "grad_norm": 0.4512398838996887,
      "learning_rate": 4.255253015044138e-06,
      "loss": 0.811,
      "step": 3595
    },
    {
      "epoch": 0.1490322848025198,
      "grad_norm": 0.3936716318130493,
      "learning_rate": 4.2550457955157695e-06,
      "loss": 0.7217,
      "step": 3596
    },
    {
      "epoch": 0.14907372870819346,
      "grad_norm": 0.43926966190338135,
      "learning_rate": 4.254838575987401e-06,
      "loss": 0.77,
      "step": 3597
    },
    {
      "epoch": 0.14911517261386714,
      "grad_norm": 0.46051591634750366,
      "learning_rate": 4.254631356459033e-06,
      "loss": 0.7808,
      "step": 3598
    },
    {
      "epoch": 0.1491566165195408,
      "grad_norm": 0.40303388237953186,
      "learning_rate": 4.2544241369306645e-06,
      "loss": 0.7302,
      "step": 3599
    },
    {
      "epoch": 0.14919806042521447,
      "grad_norm": 0.4292672872543335,
      "learning_rate": 4.254216917402296e-06,
      "loss": 0.7432,
      "step": 3600
    },
    {
      "epoch": 0.14923950433088815,
      "grad_norm": 0.439406156539917,
      "learning_rate": 4.254009697873928e-06,
      "loss": 0.76,
      "step": 3601
    },
    {
      "epoch": 0.14928094823656182,
      "grad_norm": 0.44368961453437805,
      "learning_rate": 4.2538024783455595e-06,
      "loss": 0.7229,
      "step": 3602
    },
    {
      "epoch": 0.14932239214223547,
      "grad_norm": 0.3987877666950226,
      "learning_rate": 4.253595258817192e-06,
      "loss": 0.7163,
      "step": 3603
    },
    {
      "epoch": 0.14936383604790915,
      "grad_norm": 0.430250346660614,
      "learning_rate": 4.253388039288823e-06,
      "loss": 0.7388,
      "step": 3604
    },
    {
      "epoch": 0.14940527995358283,
      "grad_norm": 0.46438106894493103,
      "learning_rate": 4.2531808197604545e-06,
      "loss": 0.7856,
      "step": 3605
    },
    {
      "epoch": 0.1494467238592565,
      "grad_norm": 0.46950626373291016,
      "learning_rate": 4.252973600232087e-06,
      "loss": 0.6853,
      "step": 3606
    },
    {
      "epoch": 0.14948816776493018,
      "grad_norm": 0.4444265365600586,
      "learning_rate": 4.252766380703718e-06,
      "loss": 0.7473,
      "step": 3607
    },
    {
      "epoch": 0.14952961167060383,
      "grad_norm": 0.40145230293273926,
      "learning_rate": 4.2525591611753495e-06,
      "loss": 0.7161,
      "step": 3608
    },
    {
      "epoch": 0.1495710555762775,
      "grad_norm": 0.4515431821346283,
      "learning_rate": 4.252351941646981e-06,
      "loss": 0.73,
      "step": 3609
    },
    {
      "epoch": 0.14961249948195118,
      "grad_norm": 0.4358542263507843,
      "learning_rate": 4.252144722118613e-06,
      "loss": 0.6992,
      "step": 3610
    },
    {
      "epoch": 0.14965394338762486,
      "grad_norm": 0.4030960500240326,
      "learning_rate": 4.2519375025902445e-06,
      "loss": 0.6978,
      "step": 3611
    },
    {
      "epoch": 0.1496953872932985,
      "grad_norm": 0.4064294099807739,
      "learning_rate": 4.251730283061876e-06,
      "loss": 0.7324,
      "step": 3612
    },
    {
      "epoch": 0.1497368311989722,
      "grad_norm": 0.39400628209114075,
      "learning_rate": 4.251523063533507e-06,
      "loss": 0.7573,
      "step": 3613
    },
    {
      "epoch": 0.14977827510464586,
      "grad_norm": 0.399495005607605,
      "learning_rate": 4.2513158440051395e-06,
      "loss": 0.7009,
      "step": 3614
    },
    {
      "epoch": 0.14981971901031954,
      "grad_norm": 0.4368586242198944,
      "learning_rate": 4.251108624476771e-06,
      "loss": 0.7644,
      "step": 3615
    },
    {
      "epoch": 0.1498611629159932,
      "grad_norm": 0.3966556489467621,
      "learning_rate": 4.250901404948402e-06,
      "loss": 0.6858,
      "step": 3616
    },
    {
      "epoch": 0.14990260682166687,
      "grad_norm": 0.43613311648368835,
      "learning_rate": 4.2506941854200345e-06,
      "loss": 0.7029,
      "step": 3617
    },
    {
      "epoch": 0.14994405072734054,
      "grad_norm": 0.43202292919158936,
      "learning_rate": 4.250486965891666e-06,
      "loss": 0.7192,
      "step": 3618
    },
    {
      "epoch": 0.14998549463301422,
      "grad_norm": 0.44344326853752136,
      "learning_rate": 4.250279746363298e-06,
      "loss": 0.7344,
      "step": 3619
    },
    {
      "epoch": 0.1500269385386879,
      "grad_norm": 0.39629971981048584,
      "learning_rate": 4.250072526834929e-06,
      "loss": 0.6735,
      "step": 3620
    },
    {
      "epoch": 0.15006838244436155,
      "grad_norm": 0.40750643610954285,
      "learning_rate": 4.249865307306561e-06,
      "loss": 0.7263,
      "step": 3621
    },
    {
      "epoch": 0.15010982635003522,
      "grad_norm": 0.48804667592048645,
      "learning_rate": 4.249658087778192e-06,
      "loss": 0.7789,
      "step": 3622
    },
    {
      "epoch": 0.1501512702557089,
      "grad_norm": 0.5001088976860046,
      "learning_rate": 4.2494508682498245e-06,
      "loss": 0.7737,
      "step": 3623
    },
    {
      "epoch": 0.15019271416138258,
      "grad_norm": 0.39449793100357056,
      "learning_rate": 4.249243648721456e-06,
      "loss": 0.7244,
      "step": 3624
    },
    {
      "epoch": 0.15023415806705623,
      "grad_norm": 0.4193180501461029,
      "learning_rate": 4.249036429193087e-06,
      "loss": 0.7214,
      "step": 3625
    },
    {
      "epoch": 0.1502756019727299,
      "grad_norm": 0.3914564549922943,
      "learning_rate": 4.2488292096647195e-06,
      "loss": 0.6865,
      "step": 3626
    },
    {
      "epoch": 0.15031704587840358,
      "grad_norm": 0.3871206045150757,
      "learning_rate": 4.248621990136351e-06,
      "loss": 0.6923,
      "step": 3627
    },
    {
      "epoch": 0.15035848978407726,
      "grad_norm": 0.4146389663219452,
      "learning_rate": 4.248414770607982e-06,
      "loss": 0.7463,
      "step": 3628
    },
    {
      "epoch": 0.1503999336897509,
      "grad_norm": 0.38069748878479004,
      "learning_rate": 4.248207551079614e-06,
      "loss": 0.6357,
      "step": 3629
    },
    {
      "epoch": 0.15044137759542459,
      "grad_norm": 0.4219701886177063,
      "learning_rate": 4.248000331551246e-06,
      "loss": 0.7177,
      "step": 3630
    },
    {
      "epoch": 0.15048282150109826,
      "grad_norm": 0.4784638583660126,
      "learning_rate": 4.247793112022877e-06,
      "loss": 0.7349,
      "step": 3631
    },
    {
      "epoch": 0.15052426540677194,
      "grad_norm": 0.44536247849464417,
      "learning_rate": 4.247585892494509e-06,
      "loss": 0.7607,
      "step": 3632
    },
    {
      "epoch": 0.15056570931244562,
      "grad_norm": 0.4355635941028595,
      "learning_rate": 4.247378672966141e-06,
      "loss": 0.7251,
      "step": 3633
    },
    {
      "epoch": 0.15060715321811927,
      "grad_norm": 0.44101041555404663,
      "learning_rate": 4.247171453437772e-06,
      "loss": 0.7249,
      "step": 3634
    },
    {
      "epoch": 0.15064859712379294,
      "grad_norm": 0.40255945920944214,
      "learning_rate": 4.2469642339094046e-06,
      "loss": 0.7219,
      "step": 3635
    },
    {
      "epoch": 0.15069004102946662,
      "grad_norm": 0.4074898362159729,
      "learning_rate": 4.246757014381035e-06,
      "loss": 0.7041,
      "step": 3636
    },
    {
      "epoch": 0.1507314849351403,
      "grad_norm": 0.4288147985935211,
      "learning_rate": 4.246549794852667e-06,
      "loss": 0.7053,
      "step": 3637
    },
    {
      "epoch": 0.15077292884081395,
      "grad_norm": 0.4557374119758606,
      "learning_rate": 4.246342575324299e-06,
      "loss": 0.7749,
      "step": 3638
    },
    {
      "epoch": 0.15081437274648762,
      "grad_norm": 0.43249452114105225,
      "learning_rate": 4.246135355795931e-06,
      "loss": 0.7347,
      "step": 3639
    },
    {
      "epoch": 0.1508558166521613,
      "grad_norm": 0.4268703758716583,
      "learning_rate": 4.245928136267562e-06,
      "loss": 0.7239,
      "step": 3640
    },
    {
      "epoch": 0.15089726055783498,
      "grad_norm": 0.44752857089042664,
      "learning_rate": 4.245720916739194e-06,
      "loss": 0.7336,
      "step": 3641
    },
    {
      "epoch": 0.15093870446350863,
      "grad_norm": 0.4321291148662567,
      "learning_rate": 4.245513697210826e-06,
      "loss": 0.7771,
      "step": 3642
    },
    {
      "epoch": 0.1509801483691823,
      "grad_norm": 0.4716063141822815,
      "learning_rate": 4.245306477682457e-06,
      "loss": 0.7715,
      "step": 3643
    },
    {
      "epoch": 0.15102159227485598,
      "grad_norm": 0.458738774061203,
      "learning_rate": 4.245099258154089e-06,
      "loss": 0.7217,
      "step": 3644
    },
    {
      "epoch": 0.15106303618052966,
      "grad_norm": 0.4909980297088623,
      "learning_rate": 4.24489203862572e-06,
      "loss": 0.7834,
      "step": 3645
    },
    {
      "epoch": 0.15110448008620334,
      "grad_norm": 0.43763333559036255,
      "learning_rate": 4.244684819097352e-06,
      "loss": 0.7527,
      "step": 3646
    },
    {
      "epoch": 0.15114592399187698,
      "grad_norm": 0.42772412300109863,
      "learning_rate": 4.244477599568984e-06,
      "loss": 0.7329,
      "step": 3647
    },
    {
      "epoch": 0.15118736789755066,
      "grad_norm": 0.40519919991493225,
      "learning_rate": 4.244270380040615e-06,
      "loss": 0.6915,
      "step": 3648
    },
    {
      "epoch": 0.15122881180322434,
      "grad_norm": 0.4585369825363159,
      "learning_rate": 4.2440631605122465e-06,
      "loss": 0.7924,
      "step": 3649
    },
    {
      "epoch": 0.15127025570889802,
      "grad_norm": 0.43139317631721497,
      "learning_rate": 4.243855940983879e-06,
      "loss": 0.7683,
      "step": 3650
    },
    {
      "epoch": 0.15131169961457167,
      "grad_norm": 0.41319629549980164,
      "learning_rate": 4.243648721455511e-06,
      "loss": 0.7083,
      "step": 3651
    },
    {
      "epoch": 0.15135314352024534,
      "grad_norm": 0.3986436128616333,
      "learning_rate": 4.2434415019271415e-06,
      "loss": 0.6721,
      "step": 3652
    },
    {
      "epoch": 0.15139458742591902,
      "grad_norm": 0.47634583711624146,
      "learning_rate": 4.243234282398774e-06,
      "loss": 0.7754,
      "step": 3653
    },
    {
      "epoch": 0.1514360313315927,
      "grad_norm": 0.46902167797088623,
      "learning_rate": 4.243027062870405e-06,
      "loss": 0.7502,
      "step": 3654
    },
    {
      "epoch": 0.15147747523726637,
      "grad_norm": 0.4037449061870575,
      "learning_rate": 4.242819843342037e-06,
      "loss": 0.7,
      "step": 3655
    },
    {
      "epoch": 0.15151891914294002,
      "grad_norm": 0.43108341097831726,
      "learning_rate": 4.242612623813669e-06,
      "loss": 0.7468,
      "step": 3656
    },
    {
      "epoch": 0.1515603630486137,
      "grad_norm": 0.45734941959381104,
      "learning_rate": 4.2424054042853e-06,
      "loss": 0.7212,
      "step": 3657
    },
    {
      "epoch": 0.15160180695428738,
      "grad_norm": 0.47673550248146057,
      "learning_rate": 4.2421981847569315e-06,
      "loss": 0.7717,
      "step": 3658
    },
    {
      "epoch": 0.15164325085996105,
      "grad_norm": 0.4326879382133484,
      "learning_rate": 4.241990965228564e-06,
      "loss": 0.7538,
      "step": 3659
    },
    {
      "epoch": 0.1516846947656347,
      "grad_norm": 0.4235236942768097,
      "learning_rate": 4.241783745700195e-06,
      "loss": 0.7549,
      "step": 3660
    },
    {
      "epoch": 0.15172613867130838,
      "grad_norm": 0.41055363416671753,
      "learning_rate": 4.2415765261718265e-06,
      "loss": 0.7318,
      "step": 3661
    },
    {
      "epoch": 0.15176758257698206,
      "grad_norm": 0.46266499161720276,
      "learning_rate": 4.241369306643459e-06,
      "loss": 0.7629,
      "step": 3662
    },
    {
      "epoch": 0.15180902648265573,
      "grad_norm": 0.42659202218055725,
      "learning_rate": 4.24116208711509e-06,
      "loss": 0.7251,
      "step": 3663
    },
    {
      "epoch": 0.15185047038832938,
      "grad_norm": 0.427820086479187,
      "learning_rate": 4.2409548675867215e-06,
      "loss": 0.7002,
      "step": 3664
    },
    {
      "epoch": 0.15189191429400306,
      "grad_norm": 0.4459049105644226,
      "learning_rate": 4.240747648058353e-06,
      "loss": 0.7202,
      "step": 3665
    },
    {
      "epoch": 0.15193335819967674,
      "grad_norm": 0.440362811088562,
      "learning_rate": 4.240540428529985e-06,
      "loss": 0.7164,
      "step": 3666
    },
    {
      "epoch": 0.15197480210535042,
      "grad_norm": 0.4040077030658722,
      "learning_rate": 4.2403332090016165e-06,
      "loss": 0.7285,
      "step": 3667
    },
    {
      "epoch": 0.1520162460110241,
      "grad_norm": 0.4361995458602905,
      "learning_rate": 4.240125989473248e-06,
      "loss": 0.7588,
      "step": 3668
    },
    {
      "epoch": 0.15205768991669774,
      "grad_norm": 0.41343316435813904,
      "learning_rate": 4.23991876994488e-06,
      "loss": 0.7262,
      "step": 3669
    },
    {
      "epoch": 0.15209913382237142,
      "grad_norm": 0.4115023612976074,
      "learning_rate": 4.2397115504165115e-06,
      "loss": 0.6897,
      "step": 3670
    },
    {
      "epoch": 0.1521405777280451,
      "grad_norm": 0.42587971687316895,
      "learning_rate": 4.239504330888144e-06,
      "loss": 0.7446,
      "step": 3671
    },
    {
      "epoch": 0.15218202163371877,
      "grad_norm": 0.44701018929481506,
      "learning_rate": 4.239297111359775e-06,
      "loss": 0.7031,
      "step": 3672
    },
    {
      "epoch": 0.15222346553939242,
      "grad_norm": 0.44883471727371216,
      "learning_rate": 4.2390898918314065e-06,
      "loss": 0.6941,
      "step": 3673
    },
    {
      "epoch": 0.1522649094450661,
      "grad_norm": 0.44280728697776794,
      "learning_rate": 4.238882672303038e-06,
      "loss": 0.7102,
      "step": 3674
    },
    {
      "epoch": 0.15230635335073978,
      "grad_norm": 0.4050421416759491,
      "learning_rate": 4.23867545277467e-06,
      "loss": 0.7335,
      "step": 3675
    },
    {
      "epoch": 0.15234779725641345,
      "grad_norm": 0.40624454617500305,
      "learning_rate": 4.2384682332463015e-06,
      "loss": 0.7367,
      "step": 3676
    },
    {
      "epoch": 0.1523892411620871,
      "grad_norm": 0.4146789312362671,
      "learning_rate": 4.238261013717933e-06,
      "loss": 0.707,
      "step": 3677
    },
    {
      "epoch": 0.15243068506776078,
      "grad_norm": 0.44670915603637695,
      "learning_rate": 4.238053794189565e-06,
      "loss": 0.7158,
      "step": 3678
    },
    {
      "epoch": 0.15247212897343446,
      "grad_norm": 0.46492907404899597,
      "learning_rate": 4.2378465746611965e-06,
      "loss": 0.7979,
      "step": 3679
    },
    {
      "epoch": 0.15251357287910813,
      "grad_norm": 0.40393638610839844,
      "learning_rate": 4.237639355132828e-06,
      "loss": 0.6559,
      "step": 3680
    },
    {
      "epoch": 0.1525550167847818,
      "grad_norm": 0.46904534101486206,
      "learning_rate": 4.237432135604459e-06,
      "loss": 0.7677,
      "step": 3681
    },
    {
      "epoch": 0.15259646069045546,
      "grad_norm": 0.41780877113342285,
      "learning_rate": 4.2372249160760915e-06,
      "loss": 0.7354,
      "step": 3682
    },
    {
      "epoch": 0.15263790459612914,
      "grad_norm": 0.42746788263320923,
      "learning_rate": 4.237017696547723e-06,
      "loss": 0.7036,
      "step": 3683
    },
    {
      "epoch": 0.15267934850180281,
      "grad_norm": 0.45559629797935486,
      "learning_rate": 4.236810477019354e-06,
      "loss": 0.7527,
      "step": 3684
    },
    {
      "epoch": 0.1527207924074765,
      "grad_norm": 0.5866517424583435,
      "learning_rate": 4.2366032574909865e-06,
      "loss": 0.7939,
      "step": 3685
    },
    {
      "epoch": 0.15276223631315014,
      "grad_norm": 0.42547816038131714,
      "learning_rate": 4.236396037962618e-06,
      "loss": 0.7096,
      "step": 3686
    },
    {
      "epoch": 0.15280368021882382,
      "grad_norm": 0.5292379260063171,
      "learning_rate": 4.23618881843425e-06,
      "loss": 0.8098,
      "step": 3687
    },
    {
      "epoch": 0.1528451241244975,
      "grad_norm": 0.45485520362854004,
      "learning_rate": 4.235981598905881e-06,
      "loss": 0.7356,
      "step": 3688
    },
    {
      "epoch": 0.15288656803017117,
      "grad_norm": 0.4807022213935852,
      "learning_rate": 4.235774379377513e-06,
      "loss": 0.728,
      "step": 3689
    },
    {
      "epoch": 0.15292801193584482,
      "grad_norm": 0.44598907232284546,
      "learning_rate": 4.235567159849144e-06,
      "loss": 0.8257,
      "step": 3690
    },
    {
      "epoch": 0.1529694558415185,
      "grad_norm": 0.43914106488227844,
      "learning_rate": 4.2353599403207765e-06,
      "loss": 0.7654,
      "step": 3691
    },
    {
      "epoch": 0.15301089974719218,
      "grad_norm": 0.4122999906539917,
      "learning_rate": 4.235152720792408e-06,
      "loss": 0.7502,
      "step": 3692
    },
    {
      "epoch": 0.15305234365286585,
      "grad_norm": 0.4320218563079834,
      "learning_rate": 4.234945501264039e-06,
      "loss": 0.7546,
      "step": 3693
    },
    {
      "epoch": 0.15309378755853953,
      "grad_norm": 0.4195154309272766,
      "learning_rate": 4.2347382817356716e-06,
      "loss": 0.7177,
      "step": 3694
    },
    {
      "epoch": 0.15313523146421318,
      "grad_norm": 0.4758714735507965,
      "learning_rate": 4.234531062207303e-06,
      "loss": 0.7284,
      "step": 3695
    },
    {
      "epoch": 0.15317667536988686,
      "grad_norm": 0.4443349242210388,
      "learning_rate": 4.234323842678934e-06,
      "loss": 0.7363,
      "step": 3696
    },
    {
      "epoch": 0.15321811927556053,
      "grad_norm": 0.4219074249267578,
      "learning_rate": 4.234116623150566e-06,
      "loss": 0.7231,
      "step": 3697
    },
    {
      "epoch": 0.1532595631812342,
      "grad_norm": 0.4744405746459961,
      "learning_rate": 4.233909403622198e-06,
      "loss": 0.8752,
      "step": 3698
    },
    {
      "epoch": 0.15330100708690786,
      "grad_norm": 0.45451921224594116,
      "learning_rate": 4.233702184093829e-06,
      "loss": 0.811,
      "step": 3699
    },
    {
      "epoch": 0.15334245099258154,
      "grad_norm": 0.4308885633945465,
      "learning_rate": 4.233494964565461e-06,
      "loss": 0.7266,
      "step": 3700
    },
    {
      "epoch": 0.1533838948982552,
      "grad_norm": 0.4299415647983551,
      "learning_rate": 4.233287745037092e-06,
      "loss": 0.7097,
      "step": 3701
    },
    {
      "epoch": 0.1534253388039289,
      "grad_norm": 0.43409693241119385,
      "learning_rate": 4.233080525508724e-06,
      "loss": 0.707,
      "step": 3702
    },
    {
      "epoch": 0.15346678270960254,
      "grad_norm": 0.4016571640968323,
      "learning_rate": 4.2328733059803566e-06,
      "loss": 0.7205,
      "step": 3703
    },
    {
      "epoch": 0.15350822661527622,
      "grad_norm": 0.43497949838638306,
      "learning_rate": 4.232666086451987e-06,
      "loss": 0.7119,
      "step": 3704
    },
    {
      "epoch": 0.1535496705209499,
      "grad_norm": 0.45771169662475586,
      "learning_rate": 4.232458866923619e-06,
      "loss": 0.7278,
      "step": 3705
    },
    {
      "epoch": 0.15359111442662357,
      "grad_norm": 0.4203149080276489,
      "learning_rate": 4.232251647395251e-06,
      "loss": 0.668,
      "step": 3706
    },
    {
      "epoch": 0.15363255833229725,
      "grad_norm": 0.4154300093650818,
      "learning_rate": 4.232044427866883e-06,
      "loss": 0.741,
      "step": 3707
    },
    {
      "epoch": 0.1536740022379709,
      "grad_norm": 0.4568198025226593,
      "learning_rate": 4.231837208338514e-06,
      "loss": 0.7621,
      "step": 3708
    },
    {
      "epoch": 0.15371544614364457,
      "grad_norm": 0.4174901247024536,
      "learning_rate": 4.231629988810146e-06,
      "loss": 0.7434,
      "step": 3709
    },
    {
      "epoch": 0.15375689004931825,
      "grad_norm": 0.40741077065467834,
      "learning_rate": 4.231422769281777e-06,
      "loss": 0.715,
      "step": 3710
    },
    {
      "epoch": 0.15379833395499193,
      "grad_norm": 0.43890616297721863,
      "learning_rate": 4.231215549753409e-06,
      "loss": 0.7468,
      "step": 3711
    },
    {
      "epoch": 0.15383977786066558,
      "grad_norm": 0.5174891352653503,
      "learning_rate": 4.231008330225041e-06,
      "loss": 0.7142,
      "step": 3712
    },
    {
      "epoch": 0.15388122176633925,
      "grad_norm": 0.41212040185928345,
      "learning_rate": 4.230801110696672e-06,
      "loss": 0.6943,
      "step": 3713
    },
    {
      "epoch": 0.15392266567201293,
      "grad_norm": 0.43226858973503113,
      "learning_rate": 4.230593891168304e-06,
      "loss": 0.7426,
      "step": 3714
    },
    {
      "epoch": 0.1539641095776866,
      "grad_norm": 0.4193629324436188,
      "learning_rate": 4.230386671639936e-06,
      "loss": 0.6771,
      "step": 3715
    },
    {
      "epoch": 0.15400555348336029,
      "grad_norm": 0.4190411865711212,
      "learning_rate": 4.230179452111567e-06,
      "loss": 0.6957,
      "step": 3716
    },
    {
      "epoch": 0.15404699738903394,
      "grad_norm": 0.41139307618141174,
      "learning_rate": 4.2299722325831985e-06,
      "loss": 0.7343,
      "step": 3717
    },
    {
      "epoch": 0.1540884412947076,
      "grad_norm": 0.42168131470680237,
      "learning_rate": 4.229765013054831e-06,
      "loss": 0.7883,
      "step": 3718
    },
    {
      "epoch": 0.1541298852003813,
      "grad_norm": 0.4320341646671295,
      "learning_rate": 4.229557793526462e-06,
      "loss": 0.7852,
      "step": 3719
    },
    {
      "epoch": 0.15417132910605497,
      "grad_norm": 0.42037323117256165,
      "learning_rate": 4.2293505739980935e-06,
      "loss": 0.7559,
      "step": 3720
    },
    {
      "epoch": 0.15421277301172862,
      "grad_norm": 0.42893674969673157,
      "learning_rate": 4.229143354469726e-06,
      "loss": 0.7651,
      "step": 3721
    },
    {
      "epoch": 0.1542542169174023,
      "grad_norm": 0.45412829518318176,
      "learning_rate": 4.228936134941357e-06,
      "loss": 0.7717,
      "step": 3722
    },
    {
      "epoch": 0.15429566082307597,
      "grad_norm": 0.43226689100265503,
      "learning_rate": 4.228728915412989e-06,
      "loss": 0.7659,
      "step": 3723
    },
    {
      "epoch": 0.15433710472874965,
      "grad_norm": 0.4245906472206116,
      "learning_rate": 4.228521695884621e-06,
      "loss": 0.7463,
      "step": 3724
    },
    {
      "epoch": 0.1543785486344233,
      "grad_norm": 0.43555760383605957,
      "learning_rate": 4.228314476356252e-06,
      "loss": 0.7593,
      "step": 3725
    },
    {
      "epoch": 0.15441999254009697,
      "grad_norm": 0.42883971333503723,
      "learning_rate": 4.2281072568278835e-06,
      "loss": 0.6471,
      "step": 3726
    },
    {
      "epoch": 0.15446143644577065,
      "grad_norm": 0.4438154697418213,
      "learning_rate": 4.227900037299516e-06,
      "loss": 0.7188,
      "step": 3727
    },
    {
      "epoch": 0.15450288035144433,
      "grad_norm": 0.4603244960308075,
      "learning_rate": 4.227692817771147e-06,
      "loss": 0.7139,
      "step": 3728
    },
    {
      "epoch": 0.154544324257118,
      "grad_norm": 0.4458857476711273,
      "learning_rate": 4.2274855982427785e-06,
      "loss": 0.7078,
      "step": 3729
    },
    {
      "epoch": 0.15458576816279165,
      "grad_norm": 0.4062007665634155,
      "learning_rate": 4.227278378714411e-06,
      "loss": 0.6869,
      "step": 3730
    },
    {
      "epoch": 0.15462721206846533,
      "grad_norm": 0.47022193670272827,
      "learning_rate": 4.227071159186042e-06,
      "loss": 0.7849,
      "step": 3731
    },
    {
      "epoch": 0.154668655974139,
      "grad_norm": 0.4200233221054077,
      "learning_rate": 4.2268639396576735e-06,
      "loss": 0.739,
      "step": 3732
    },
    {
      "epoch": 0.15471009987981268,
      "grad_norm": 0.4448118805885315,
      "learning_rate": 4.226656720129305e-06,
      "loss": 0.7271,
      "step": 3733
    },
    {
      "epoch": 0.15475154378548633,
      "grad_norm": 0.41340065002441406,
      "learning_rate": 4.226449500600937e-06,
      "loss": 0.7073,
      "step": 3734
    },
    {
      "epoch": 0.15479298769116,
      "grad_norm": 0.4488571882247925,
      "learning_rate": 4.2262422810725685e-06,
      "loss": 0.748,
      "step": 3735
    },
    {
      "epoch": 0.1548344315968337,
      "grad_norm": 0.4039192199707031,
      "learning_rate": 4.2260350615442e-06,
      "loss": 0.6909,
      "step": 3736
    },
    {
      "epoch": 0.15487587550250737,
      "grad_norm": 0.4350754916667938,
      "learning_rate": 4.225827842015832e-06,
      "loss": 0.7705,
      "step": 3737
    },
    {
      "epoch": 0.15491731940818101,
      "grad_norm": 0.44145551323890686,
      "learning_rate": 4.2256206224874635e-06,
      "loss": 0.7205,
      "step": 3738
    },
    {
      "epoch": 0.1549587633138547,
      "grad_norm": 0.39144453406333923,
      "learning_rate": 4.225413402959096e-06,
      "loss": 0.731,
      "step": 3739
    },
    {
      "epoch": 0.15500020721952837,
      "grad_norm": 0.43198150396347046,
      "learning_rate": 4.225206183430727e-06,
      "loss": 0.7961,
      "step": 3740
    },
    {
      "epoch": 0.15504165112520205,
      "grad_norm": 0.4303108751773834,
      "learning_rate": 4.2249989639023585e-06,
      "loss": 0.7178,
      "step": 3741
    },
    {
      "epoch": 0.15508309503087572,
      "grad_norm": 0.42614153027534485,
      "learning_rate": 4.22479174437399e-06,
      "loss": 0.7305,
      "step": 3742
    },
    {
      "epoch": 0.15512453893654937,
      "grad_norm": 0.4599325656890869,
      "learning_rate": 4.224584524845622e-06,
      "loss": 0.728,
      "step": 3743
    },
    {
      "epoch": 0.15516598284222305,
      "grad_norm": 0.4178580343723297,
      "learning_rate": 4.2243773053172535e-06,
      "loss": 0.7483,
      "step": 3744
    },
    {
      "epoch": 0.15520742674789673,
      "grad_norm": 0.43739956617355347,
      "learning_rate": 4.224170085788885e-06,
      "loss": 0.731,
      "step": 3745
    },
    {
      "epoch": 0.1552488706535704,
      "grad_norm": 0.424846887588501,
      "learning_rate": 4.223962866260517e-06,
      "loss": 0.7227,
      "step": 3746
    },
    {
      "epoch": 0.15529031455924405,
      "grad_norm": 0.42290574312210083,
      "learning_rate": 4.2237556467321485e-06,
      "loss": 0.7251,
      "step": 3747
    },
    {
      "epoch": 0.15533175846491773,
      "grad_norm": 0.4364088177680969,
      "learning_rate": 4.22354842720378e-06,
      "loss": 0.7498,
      "step": 3748
    },
    {
      "epoch": 0.1553732023705914,
      "grad_norm": 0.39693793654441833,
      "learning_rate": 4.223341207675411e-06,
      "loss": 0.6687,
      "step": 3749
    },
    {
      "epoch": 0.15541464627626508,
      "grad_norm": 0.46199992299079895,
      "learning_rate": 4.2231339881470435e-06,
      "loss": 0.761,
      "step": 3750
    },
    {
      "epoch": 0.15545609018193873,
      "grad_norm": 0.43053698539733887,
      "learning_rate": 4.222926768618675e-06,
      "loss": 0.7136,
      "step": 3751
    },
    {
      "epoch": 0.1554975340876124,
      "grad_norm": 0.4421553909778595,
      "learning_rate": 4.222719549090306e-06,
      "loss": 0.7683,
      "step": 3752
    },
    {
      "epoch": 0.1555389779932861,
      "grad_norm": 0.45947518944740295,
      "learning_rate": 4.222512329561938e-06,
      "loss": 0.7603,
      "step": 3753
    },
    {
      "epoch": 0.15558042189895976,
      "grad_norm": 0.4375973343849182,
      "learning_rate": 4.22230511003357e-06,
      "loss": 0.756,
      "step": 3754
    },
    {
      "epoch": 0.15562186580463344,
      "grad_norm": 0.4328485131263733,
      "learning_rate": 4.222097890505202e-06,
      "loss": 0.7197,
      "step": 3755
    },
    {
      "epoch": 0.1556633097103071,
      "grad_norm": 0.4104398787021637,
      "learning_rate": 4.2218906709768336e-06,
      "loss": 0.707,
      "step": 3756
    },
    {
      "epoch": 0.15570475361598077,
      "grad_norm": 0.439996600151062,
      "learning_rate": 4.221683451448465e-06,
      "loss": 0.7365,
      "step": 3757
    },
    {
      "epoch": 0.15574619752165444,
      "grad_norm": 0.42545443773269653,
      "learning_rate": 4.221476231920096e-06,
      "loss": 0.7856,
      "step": 3758
    },
    {
      "epoch": 0.15578764142732812,
      "grad_norm": 0.43424174189567566,
      "learning_rate": 4.2212690123917286e-06,
      "loss": 0.7655,
      "step": 3759
    },
    {
      "epoch": 0.15582908533300177,
      "grad_norm": 0.4118076264858246,
      "learning_rate": 4.22106179286336e-06,
      "loss": 0.7544,
      "step": 3760
    },
    {
      "epoch": 0.15587052923867545,
      "grad_norm": 0.42083072662353516,
      "learning_rate": 4.220854573334991e-06,
      "loss": 0.7202,
      "step": 3761
    },
    {
      "epoch": 0.15591197314434913,
      "grad_norm": 0.44824084639549255,
      "learning_rate": 4.220647353806623e-06,
      "loss": 0.6895,
      "step": 3762
    },
    {
      "epoch": 0.1559534170500228,
      "grad_norm": 0.4726134240627289,
      "learning_rate": 4.220440134278255e-06,
      "loss": 0.7805,
      "step": 3763
    },
    {
      "epoch": 0.15599486095569645,
      "grad_norm": 0.43774133920669556,
      "learning_rate": 4.220232914749886e-06,
      "loss": 0.7761,
      "step": 3764
    },
    {
      "epoch": 0.15603630486137013,
      "grad_norm": 0.46876615285873413,
      "learning_rate": 4.220025695221518e-06,
      "loss": 0.7654,
      "step": 3765
    },
    {
      "epoch": 0.1560777487670438,
      "grad_norm": 0.4385366141796112,
      "learning_rate": 4.21981847569315e-06,
      "loss": 0.6725,
      "step": 3766
    },
    {
      "epoch": 0.15611919267271748,
      "grad_norm": 0.45882877707481384,
      "learning_rate": 4.219611256164781e-06,
      "loss": 0.7383,
      "step": 3767
    },
    {
      "epoch": 0.15616063657839116,
      "grad_norm": 0.4636717438697815,
      "learning_rate": 4.219404036636413e-06,
      "loss": 0.7627,
      "step": 3768
    },
    {
      "epoch": 0.1562020804840648,
      "grad_norm": 0.41499027609825134,
      "learning_rate": 4.219196817108044e-06,
      "loss": 0.7092,
      "step": 3769
    },
    {
      "epoch": 0.1562435243897385,
      "grad_norm": 0.4373124837875366,
      "learning_rate": 4.218989597579676e-06,
      "loss": 0.7122,
      "step": 3770
    },
    {
      "epoch": 0.15628496829541216,
      "grad_norm": 0.4432714879512787,
      "learning_rate": 4.218782378051308e-06,
      "loss": 0.7781,
      "step": 3771
    },
    {
      "epoch": 0.15632641220108584,
      "grad_norm": 0.4590091407299042,
      "learning_rate": 4.218575158522939e-06,
      "loss": 0.7834,
      "step": 3772
    },
    {
      "epoch": 0.1563678561067595,
      "grad_norm": 0.43703097105026245,
      "learning_rate": 4.218367938994571e-06,
      "loss": 0.7715,
      "step": 3773
    },
    {
      "epoch": 0.15640930001243317,
      "grad_norm": 0.45799267292022705,
      "learning_rate": 4.218160719466203e-06,
      "loss": 0.7039,
      "step": 3774
    },
    {
      "epoch": 0.15645074391810684,
      "grad_norm": 0.40787288546562195,
      "learning_rate": 4.217953499937835e-06,
      "loss": 0.7441,
      "step": 3775
    },
    {
      "epoch": 0.15649218782378052,
      "grad_norm": 0.4136776924133301,
      "learning_rate": 4.217746280409466e-06,
      "loss": 0.7788,
      "step": 3776
    },
    {
      "epoch": 0.15653363172945417,
      "grad_norm": 0.4273936450481415,
      "learning_rate": 4.217539060881098e-06,
      "loss": 0.7148,
      "step": 3777
    },
    {
      "epoch": 0.15657507563512785,
      "grad_norm": 0.4586282968521118,
      "learning_rate": 4.217331841352729e-06,
      "loss": 0.8235,
      "step": 3778
    },
    {
      "epoch": 0.15661651954080152,
      "grad_norm": 0.40617039799690247,
      "learning_rate": 4.217124621824361e-06,
      "loss": 0.7317,
      "step": 3779
    },
    {
      "epoch": 0.1566579634464752,
      "grad_norm": 0.45033735036849976,
      "learning_rate": 4.216917402295993e-06,
      "loss": 0.7651,
      "step": 3780
    },
    {
      "epoch": 0.15669940735214888,
      "grad_norm": 0.42508837580680847,
      "learning_rate": 4.216710182767624e-06,
      "loss": 0.7024,
      "step": 3781
    },
    {
      "epoch": 0.15674085125782253,
      "grad_norm": 0.41419878602027893,
      "learning_rate": 4.216502963239256e-06,
      "loss": 0.7351,
      "step": 3782
    },
    {
      "epoch": 0.1567822951634962,
      "grad_norm": 0.46937480568885803,
      "learning_rate": 4.216295743710888e-06,
      "loss": 0.741,
      "step": 3783
    },
    {
      "epoch": 0.15682373906916988,
      "grad_norm": 0.5076144933700562,
      "learning_rate": 4.216088524182519e-06,
      "loss": 0.8005,
      "step": 3784
    },
    {
      "epoch": 0.15686518297484356,
      "grad_norm": 0.43180370330810547,
      "learning_rate": 4.2158813046541505e-06,
      "loss": 0.6908,
      "step": 3785
    },
    {
      "epoch": 0.1569066268805172,
      "grad_norm": 0.43027907609939575,
      "learning_rate": 4.215674085125783e-06,
      "loss": 0.7168,
      "step": 3786
    },
    {
      "epoch": 0.15694807078619089,
      "grad_norm": 0.4288286864757538,
      "learning_rate": 4.215466865597414e-06,
      "loss": 0.756,
      "step": 3787
    },
    {
      "epoch": 0.15698951469186456,
      "grad_norm": 0.43886905908584595,
      "learning_rate": 4.2152596460690455e-06,
      "loss": 0.7681,
      "step": 3788
    },
    {
      "epoch": 0.15703095859753824,
      "grad_norm": 0.45099616050720215,
      "learning_rate": 4.215052426540678e-06,
      "loss": 0.7822,
      "step": 3789
    },
    {
      "epoch": 0.15707240250321192,
      "grad_norm": 0.4101656675338745,
      "learning_rate": 4.214845207012309e-06,
      "loss": 0.7307,
      "step": 3790
    },
    {
      "epoch": 0.15711384640888557,
      "grad_norm": 0.460316926240921,
      "learning_rate": 4.214637987483941e-06,
      "loss": 0.7715,
      "step": 3791
    },
    {
      "epoch": 0.15715529031455924,
      "grad_norm": 0.42900216579437256,
      "learning_rate": 4.214430767955573e-06,
      "loss": 0.7971,
      "step": 3792
    },
    {
      "epoch": 0.15719673422023292,
      "grad_norm": 0.4359113574028015,
      "learning_rate": 4.214223548427204e-06,
      "loss": 0.7041,
      "step": 3793
    },
    {
      "epoch": 0.1572381781259066,
      "grad_norm": 0.4067431092262268,
      "learning_rate": 4.2140163288988355e-06,
      "loss": 0.738,
      "step": 3794
    },
    {
      "epoch": 0.15727962203158025,
      "grad_norm": 0.4477531611919403,
      "learning_rate": 4.213809109370468e-06,
      "loss": 0.7571,
      "step": 3795
    },
    {
      "epoch": 0.15732106593725392,
      "grad_norm": 0.39705902338027954,
      "learning_rate": 4.213601889842099e-06,
      "loss": 0.7002,
      "step": 3796
    },
    {
      "epoch": 0.1573625098429276,
      "grad_norm": 0.47838208079338074,
      "learning_rate": 4.2133946703137305e-06,
      "loss": 0.7654,
      "step": 3797
    },
    {
      "epoch": 0.15740395374860128,
      "grad_norm": 0.41748538613319397,
      "learning_rate": 4.213187450785363e-06,
      "loss": 0.72,
      "step": 3798
    },
    {
      "epoch": 0.15744539765427493,
      "grad_norm": 0.40485039353370667,
      "learning_rate": 4.212980231256994e-06,
      "loss": 0.7607,
      "step": 3799
    },
    {
      "epoch": 0.1574868415599486,
      "grad_norm": 0.4300915002822876,
      "learning_rate": 4.2127730117286255e-06,
      "loss": 0.7488,
      "step": 3800
    },
    {
      "epoch": 0.15752828546562228,
      "grad_norm": 0.38883349299430847,
      "learning_rate": 4.212565792200257e-06,
      "loss": 0.714,
      "step": 3801
    },
    {
      "epoch": 0.15756972937129596,
      "grad_norm": 0.406230092048645,
      "learning_rate": 4.212358572671889e-06,
      "loss": 0.7583,
      "step": 3802
    },
    {
      "epoch": 0.15761117327696963,
      "grad_norm": 0.4403274655342102,
      "learning_rate": 4.2121513531435205e-06,
      "loss": 0.7842,
      "step": 3803
    },
    {
      "epoch": 0.15765261718264328,
      "grad_norm": 0.42502692341804504,
      "learning_rate": 4.211944133615152e-06,
      "loss": 0.7688,
      "step": 3804
    },
    {
      "epoch": 0.15769406108831696,
      "grad_norm": 0.43706291913986206,
      "learning_rate": 4.211736914086783e-06,
      "loss": 0.8064,
      "step": 3805
    },
    {
      "epoch": 0.15773550499399064,
      "grad_norm": 0.4406483769416809,
      "learning_rate": 4.2115296945584155e-06,
      "loss": 0.7446,
      "step": 3806
    },
    {
      "epoch": 0.15777694889966432,
      "grad_norm": 0.4406915605068207,
      "learning_rate": 4.211322475030048e-06,
      "loss": 0.7148,
      "step": 3807
    },
    {
      "epoch": 0.15781839280533796,
      "grad_norm": 0.40346699953079224,
      "learning_rate": 4.211115255501679e-06,
      "loss": 0.7522,
      "step": 3808
    },
    {
      "epoch": 0.15785983671101164,
      "grad_norm": 0.4076678454875946,
      "learning_rate": 4.2109080359733105e-06,
      "loss": 0.7419,
      "step": 3809
    },
    {
      "epoch": 0.15790128061668532,
      "grad_norm": 0.4552779495716095,
      "learning_rate": 4.210700816444942e-06,
      "loss": 0.7433,
      "step": 3810
    },
    {
      "epoch": 0.157942724522359,
      "grad_norm": 0.4086398184299469,
      "learning_rate": 4.210493596916574e-06,
      "loss": 0.7224,
      "step": 3811
    },
    {
      "epoch": 0.15798416842803265,
      "grad_norm": 0.43320825695991516,
      "learning_rate": 4.2102863773882056e-06,
      "loss": 0.7432,
      "step": 3812
    },
    {
      "epoch": 0.15802561233370632,
      "grad_norm": 0.4551692008972168,
      "learning_rate": 4.210079157859837e-06,
      "loss": 0.7246,
      "step": 3813
    },
    {
      "epoch": 0.15806705623938,
      "grad_norm": 0.4776467978954315,
      "learning_rate": 4.209871938331468e-06,
      "loss": 0.7891,
      "step": 3814
    },
    {
      "epoch": 0.15810850014505368,
      "grad_norm": 0.41093844175338745,
      "learning_rate": 4.2096647188031006e-06,
      "loss": 0.7476,
      "step": 3815
    },
    {
      "epoch": 0.15814994405072735,
      "grad_norm": 0.4701308608055115,
      "learning_rate": 4.209457499274732e-06,
      "loss": 0.7942,
      "step": 3816
    },
    {
      "epoch": 0.158191387956401,
      "grad_norm": 0.44050124287605286,
      "learning_rate": 4.209250279746363e-06,
      "loss": 0.7574,
      "step": 3817
    },
    {
      "epoch": 0.15823283186207468,
      "grad_norm": 0.45372331142425537,
      "learning_rate": 4.2090430602179956e-06,
      "loss": 0.7195,
      "step": 3818
    },
    {
      "epoch": 0.15827427576774836,
      "grad_norm": 0.4233514368534088,
      "learning_rate": 4.208835840689627e-06,
      "loss": 0.6733,
      "step": 3819
    },
    {
      "epoch": 0.15831571967342203,
      "grad_norm": 0.47585606575012207,
      "learning_rate": 4.208628621161258e-06,
      "loss": 0.7422,
      "step": 3820
    },
    {
      "epoch": 0.15835716357909568,
      "grad_norm": 0.4351378083229065,
      "learning_rate": 4.20842140163289e-06,
      "loss": 0.7527,
      "step": 3821
    },
    {
      "epoch": 0.15839860748476936,
      "grad_norm": 0.44203707575798035,
      "learning_rate": 4.208214182104522e-06,
      "loss": 0.7351,
      "step": 3822
    },
    {
      "epoch": 0.15844005139044304,
      "grad_norm": 0.4211277365684509,
      "learning_rate": 4.208006962576153e-06,
      "loss": 0.7217,
      "step": 3823
    },
    {
      "epoch": 0.15848149529611671,
      "grad_norm": 0.3870598077774048,
      "learning_rate": 4.2077997430477856e-06,
      "loss": 0.7488,
      "step": 3824
    },
    {
      "epoch": 0.15852293920179036,
      "grad_norm": 0.5050339698791504,
      "learning_rate": 4.207592523519417e-06,
      "loss": 0.8223,
      "step": 3825
    },
    {
      "epoch": 0.15856438310746404,
      "grad_norm": 0.4232695400714874,
      "learning_rate": 4.207385303991048e-06,
      "loss": 0.7443,
      "step": 3826
    },
    {
      "epoch": 0.15860582701313772,
      "grad_norm": 0.429105281829834,
      "learning_rate": 4.2071780844626806e-06,
      "loss": 0.7402,
      "step": 3827
    },
    {
      "epoch": 0.1586472709188114,
      "grad_norm": 0.4061126708984375,
      "learning_rate": 4.206970864934312e-06,
      "loss": 0.7383,
      "step": 3828
    },
    {
      "epoch": 0.15868871482448507,
      "grad_norm": 0.41521525382995605,
      "learning_rate": 4.206763645405943e-06,
      "loss": 0.792,
      "step": 3829
    },
    {
      "epoch": 0.15873015873015872,
      "grad_norm": 0.422641396522522,
      "learning_rate": 4.206556425877575e-06,
      "loss": 0.715,
      "step": 3830
    },
    {
      "epoch": 0.1587716026358324,
      "grad_norm": 0.4782736897468567,
      "learning_rate": 4.206349206349207e-06,
      "loss": 0.761,
      "step": 3831
    },
    {
      "epoch": 0.15881304654150608,
      "grad_norm": 0.42144641280174255,
      "learning_rate": 4.206141986820838e-06,
      "loss": 0.7354,
      "step": 3832
    },
    {
      "epoch": 0.15885449044717975,
      "grad_norm": 0.41282618045806885,
      "learning_rate": 4.20593476729247e-06,
      "loss": 0.7241,
      "step": 3833
    },
    {
      "epoch": 0.1588959343528534,
      "grad_norm": 0.41399601101875305,
      "learning_rate": 4.205727547764102e-06,
      "loss": 0.7102,
      "step": 3834
    },
    {
      "epoch": 0.15893737825852708,
      "grad_norm": 0.4068455696105957,
      "learning_rate": 4.205520328235733e-06,
      "loss": 0.7114,
      "step": 3835
    },
    {
      "epoch": 0.15897882216420076,
      "grad_norm": 0.4130920171737671,
      "learning_rate": 4.205313108707365e-06,
      "loss": 0.7224,
      "step": 3836
    },
    {
      "epoch": 0.15902026606987443,
      "grad_norm": 0.4225328862667084,
      "learning_rate": 4.205105889178996e-06,
      "loss": 0.6973,
      "step": 3837
    },
    {
      "epoch": 0.15906170997554808,
      "grad_norm": 0.47452792525291443,
      "learning_rate": 4.204898669650628e-06,
      "loss": 0.7434,
      "step": 3838
    },
    {
      "epoch": 0.15910315388122176,
      "grad_norm": 0.42832037806510925,
      "learning_rate": 4.20469145012226e-06,
      "loss": 0.8097,
      "step": 3839
    },
    {
      "epoch": 0.15914459778689544,
      "grad_norm": 0.6010823845863342,
      "learning_rate": 4.204484230593891e-06,
      "loss": 0.7976,
      "step": 3840
    },
    {
      "epoch": 0.1591860416925691,
      "grad_norm": 0.40898996591567993,
      "learning_rate": 4.2042770110655225e-06,
      "loss": 0.7407,
      "step": 3841
    },
    {
      "epoch": 0.1592274855982428,
      "grad_norm": 0.4171966016292572,
      "learning_rate": 4.204069791537155e-06,
      "loss": 0.6851,
      "step": 3842
    },
    {
      "epoch": 0.15926892950391644,
      "grad_norm": 0.4605291187763214,
      "learning_rate": 4.203862572008787e-06,
      "loss": 0.7437,
      "step": 3843
    },
    {
      "epoch": 0.15931037340959012,
      "grad_norm": 0.4282947778701782,
      "learning_rate": 4.203655352480418e-06,
      "loss": 0.6973,
      "step": 3844
    },
    {
      "epoch": 0.1593518173152638,
      "grad_norm": 0.4084917902946472,
      "learning_rate": 4.20344813295205e-06,
      "loss": 0.6914,
      "step": 3845
    },
    {
      "epoch": 0.15939326122093747,
      "grad_norm": 0.42546719312667847,
      "learning_rate": 4.203240913423681e-06,
      "loss": 0.7482,
      "step": 3846
    },
    {
      "epoch": 0.15943470512661112,
      "grad_norm": 0.4307466745376587,
      "learning_rate": 4.203033693895313e-06,
      "loss": 0.7617,
      "step": 3847
    },
    {
      "epoch": 0.1594761490322848,
      "grad_norm": 0.46401798725128174,
      "learning_rate": 4.202826474366945e-06,
      "loss": 0.7565,
      "step": 3848
    },
    {
      "epoch": 0.15951759293795847,
      "grad_norm": 0.4168911874294281,
      "learning_rate": 4.202619254838576e-06,
      "loss": 0.7178,
      "step": 3849
    },
    {
      "epoch": 0.15955903684363215,
      "grad_norm": 0.4438225030899048,
      "learning_rate": 4.202412035310208e-06,
      "loss": 0.7593,
      "step": 3850
    },
    {
      "epoch": 0.1596004807493058,
      "grad_norm": 0.3990876078605652,
      "learning_rate": 4.20220481578184e-06,
      "loss": 0.7274,
      "step": 3851
    },
    {
      "epoch": 0.15964192465497948,
      "grad_norm": 0.40294134616851807,
      "learning_rate": 4.201997596253471e-06,
      "loss": 0.7137,
      "step": 3852
    },
    {
      "epoch": 0.15968336856065315,
      "grad_norm": 0.3944995105266571,
      "learning_rate": 4.2017903767251025e-06,
      "loss": 0.698,
      "step": 3853
    },
    {
      "epoch": 0.15972481246632683,
      "grad_norm": 0.40305161476135254,
      "learning_rate": 4.201583157196735e-06,
      "loss": 0.7211,
      "step": 3854
    },
    {
      "epoch": 0.1597662563720005,
      "grad_norm": 0.45723584294319153,
      "learning_rate": 4.201375937668366e-06,
      "loss": 0.7339,
      "step": 3855
    },
    {
      "epoch": 0.15980770027767416,
      "grad_norm": 0.4147433042526245,
      "learning_rate": 4.2011687181399975e-06,
      "loss": 0.7229,
      "step": 3856
    },
    {
      "epoch": 0.15984914418334784,
      "grad_norm": 0.4272827208042145,
      "learning_rate": 4.200961498611629e-06,
      "loss": 0.7533,
      "step": 3857
    },
    {
      "epoch": 0.1598905880890215,
      "grad_norm": 0.4190048575401306,
      "learning_rate": 4.200754279083261e-06,
      "loss": 0.6923,
      "step": 3858
    },
    {
      "epoch": 0.1599320319946952,
      "grad_norm": 0.41579949855804443,
      "learning_rate": 4.200547059554893e-06,
      "loss": 0.7451,
      "step": 3859
    },
    {
      "epoch": 0.15997347590036884,
      "grad_norm": 0.48038020730018616,
      "learning_rate": 4.200339840026525e-06,
      "loss": 0.8193,
      "step": 3860
    },
    {
      "epoch": 0.16001491980604252,
      "grad_norm": 0.409093976020813,
      "learning_rate": 4.200132620498156e-06,
      "loss": 0.7903,
      "step": 3861
    },
    {
      "epoch": 0.1600563637117162,
      "grad_norm": 0.4334113597869873,
      "learning_rate": 4.1999254009697875e-06,
      "loss": 0.7181,
      "step": 3862
    },
    {
      "epoch": 0.16009780761738987,
      "grad_norm": 0.4502509534358978,
      "learning_rate": 4.19971818144142e-06,
      "loss": 0.7964,
      "step": 3863
    },
    {
      "epoch": 0.16013925152306355,
      "grad_norm": 0.484048992395401,
      "learning_rate": 4.199510961913051e-06,
      "loss": 0.7456,
      "step": 3864
    },
    {
      "epoch": 0.1601806954287372,
      "grad_norm": 0.4655064642429352,
      "learning_rate": 4.1993037423846825e-06,
      "loss": 0.8149,
      "step": 3865
    },
    {
      "epoch": 0.16022213933441087,
      "grad_norm": 0.39345523715019226,
      "learning_rate": 4.199096522856314e-06,
      "loss": 0.7698,
      "step": 3866
    },
    {
      "epoch": 0.16026358324008455,
      "grad_norm": 0.4604015350341797,
      "learning_rate": 4.198889303327946e-06,
      "loss": 0.7786,
      "step": 3867
    },
    {
      "epoch": 0.16030502714575823,
      "grad_norm": 0.42139995098114014,
      "learning_rate": 4.1986820837995775e-06,
      "loss": 0.7173,
      "step": 3868
    },
    {
      "epoch": 0.16034647105143188,
      "grad_norm": 0.4157603681087494,
      "learning_rate": 4.198474864271209e-06,
      "loss": 0.7422,
      "step": 3869
    },
    {
      "epoch": 0.16038791495710555,
      "grad_norm": 0.43068990111351013,
      "learning_rate": 4.198267644742841e-06,
      "loss": 0.7275,
      "step": 3870
    },
    {
      "epoch": 0.16042935886277923,
      "grad_norm": 0.4326135516166687,
      "learning_rate": 4.1980604252144726e-06,
      "loss": 0.7882,
      "step": 3871
    },
    {
      "epoch": 0.1604708027684529,
      "grad_norm": 0.47131362557411194,
      "learning_rate": 4.197853205686104e-06,
      "loss": 0.7666,
      "step": 3872
    },
    {
      "epoch": 0.16051224667412656,
      "grad_norm": 0.40668225288391113,
      "learning_rate": 4.197645986157735e-06,
      "loss": 0.6924,
      "step": 3873
    },
    {
      "epoch": 0.16055369057980023,
      "grad_norm": 0.42884138226509094,
      "learning_rate": 4.1974387666293676e-06,
      "loss": 0.7778,
      "step": 3874
    },
    {
      "epoch": 0.1605951344854739,
      "grad_norm": 0.4121173024177551,
      "learning_rate": 4.197231547100999e-06,
      "loss": 0.7334,
      "step": 3875
    },
    {
      "epoch": 0.1606365783911476,
      "grad_norm": 0.4066050052642822,
      "learning_rate": 4.197024327572631e-06,
      "loss": 0.7073,
      "step": 3876
    },
    {
      "epoch": 0.16067802229682127,
      "grad_norm": 0.4591585099697113,
      "learning_rate": 4.1968171080442626e-06,
      "loss": 0.7766,
      "step": 3877
    },
    {
      "epoch": 0.16071946620249491,
      "grad_norm": 0.3993017077445984,
      "learning_rate": 4.196609888515894e-06,
      "loss": 0.7126,
      "step": 3878
    },
    {
      "epoch": 0.1607609101081686,
      "grad_norm": 0.49140071868896484,
      "learning_rate": 4.196402668987526e-06,
      "loss": 0.7837,
      "step": 3879
    },
    {
      "epoch": 0.16080235401384227,
      "grad_norm": 0.44633105397224426,
      "learning_rate": 4.1961954494591576e-06,
      "loss": 0.7556,
      "step": 3880
    },
    {
      "epoch": 0.16084379791951595,
      "grad_norm": 0.41756877303123474,
      "learning_rate": 4.195988229930789e-06,
      "loss": 0.6919,
      "step": 3881
    },
    {
      "epoch": 0.1608852418251896,
      "grad_norm": 0.4424387812614441,
      "learning_rate": 4.19578101040242e-06,
      "loss": 0.6985,
      "step": 3882
    },
    {
      "epoch": 0.16092668573086327,
      "grad_norm": 0.4155738055706024,
      "learning_rate": 4.1955737908740526e-06,
      "loss": 0.7263,
      "step": 3883
    },
    {
      "epoch": 0.16096812963653695,
      "grad_norm": 0.44334840774536133,
      "learning_rate": 4.195366571345684e-06,
      "loss": 0.7375,
      "step": 3884
    },
    {
      "epoch": 0.16100957354221063,
      "grad_norm": 0.45792505145072937,
      "learning_rate": 4.195159351817315e-06,
      "loss": 0.7279,
      "step": 3885
    },
    {
      "epoch": 0.16105101744788428,
      "grad_norm": 0.40332263708114624,
      "learning_rate": 4.1949521322889476e-06,
      "loss": 0.7096,
      "step": 3886
    },
    {
      "epoch": 0.16109246135355795,
      "grad_norm": 0.41727304458618164,
      "learning_rate": 4.194744912760579e-06,
      "loss": 0.7327,
      "step": 3887
    },
    {
      "epoch": 0.16113390525923163,
      "grad_norm": 0.4250168204307556,
      "learning_rate": 4.19453769323221e-06,
      "loss": 0.7893,
      "step": 3888
    },
    {
      "epoch": 0.1611753491649053,
      "grad_norm": 0.44715920090675354,
      "learning_rate": 4.194330473703842e-06,
      "loss": 0.7297,
      "step": 3889
    },
    {
      "epoch": 0.16121679307057898,
      "grad_norm": 0.4113903045654297,
      "learning_rate": 4.194123254175474e-06,
      "loss": 0.7454,
      "step": 3890
    },
    {
      "epoch": 0.16125823697625263,
      "grad_norm": 0.39251089096069336,
      "learning_rate": 4.193916034647105e-06,
      "loss": 0.6851,
      "step": 3891
    },
    {
      "epoch": 0.1612996808819263,
      "grad_norm": 0.426839679479599,
      "learning_rate": 4.193708815118738e-06,
      "loss": 0.7734,
      "step": 3892
    },
    {
      "epoch": 0.1613411247876,
      "grad_norm": 0.45610150694847107,
      "learning_rate": 4.193501595590368e-06,
      "loss": 0.7053,
      "step": 3893
    },
    {
      "epoch": 0.16138256869327366,
      "grad_norm": 0.4198380410671234,
      "learning_rate": 4.193294376062e-06,
      "loss": 0.7378,
      "step": 3894
    },
    {
      "epoch": 0.16142401259894731,
      "grad_norm": 0.4052564799785614,
      "learning_rate": 4.193087156533633e-06,
      "loss": 0.7124,
      "step": 3895
    },
    {
      "epoch": 0.161465456504621,
      "grad_norm": 0.4669317305088043,
      "learning_rate": 4.192879937005264e-06,
      "loss": 0.7631,
      "step": 3896
    },
    {
      "epoch": 0.16150690041029467,
      "grad_norm": 0.43615469336509705,
      "learning_rate": 4.192672717476895e-06,
      "loss": 0.73,
      "step": 3897
    },
    {
      "epoch": 0.16154834431596835,
      "grad_norm": 0.43209102749824524,
      "learning_rate": 4.192465497948527e-06,
      "loss": 0.702,
      "step": 3898
    },
    {
      "epoch": 0.161589788221642,
      "grad_norm": 0.43262773752212524,
      "learning_rate": 4.192258278420159e-06,
      "loss": 0.7737,
      "step": 3899
    },
    {
      "epoch": 0.16163123212731567,
      "grad_norm": 0.4481184780597687,
      "learning_rate": 4.19205105889179e-06,
      "loss": 0.7319,
      "step": 3900
    },
    {
      "epoch": 0.16167267603298935,
      "grad_norm": 0.43672117590904236,
      "learning_rate": 4.191843839363422e-06,
      "loss": 0.7717,
      "step": 3901
    },
    {
      "epoch": 0.16171411993866303,
      "grad_norm": 0.41126543283462524,
      "learning_rate": 4.191636619835053e-06,
      "loss": 0.7668,
      "step": 3902
    },
    {
      "epoch": 0.1617555638443367,
      "grad_norm": 0.4169687032699585,
      "learning_rate": 4.191429400306685e-06,
      "loss": 0.6929,
      "step": 3903
    },
    {
      "epoch": 0.16179700775001035,
      "grad_norm": 0.450638085603714,
      "learning_rate": 4.191222180778317e-06,
      "loss": 0.752,
      "step": 3904
    },
    {
      "epoch": 0.16183845165568403,
      "grad_norm": 0.43379783630371094,
      "learning_rate": 4.191014961249948e-06,
      "loss": 0.7185,
      "step": 3905
    },
    {
      "epoch": 0.1618798955613577,
      "grad_norm": 0.42769724130630493,
      "learning_rate": 4.19080774172158e-06,
      "loss": 0.6768,
      "step": 3906
    },
    {
      "epoch": 0.16192133946703138,
      "grad_norm": 0.45519348978996277,
      "learning_rate": 4.190600522193212e-06,
      "loss": 0.7336,
      "step": 3907
    },
    {
      "epoch": 0.16196278337270503,
      "grad_norm": 0.40760353207588196,
      "learning_rate": 4.190393302664844e-06,
      "loss": 0.6626,
      "step": 3908
    },
    {
      "epoch": 0.1620042272783787,
      "grad_norm": 0.40230047702789307,
      "learning_rate": 4.1901860831364745e-06,
      "loss": 0.7207,
      "step": 3909
    },
    {
      "epoch": 0.1620456711840524,
      "grad_norm": 0.4131404161453247,
      "learning_rate": 4.189978863608107e-06,
      "loss": 0.717,
      "step": 3910
    },
    {
      "epoch": 0.16208711508972606,
      "grad_norm": 0.5082923173904419,
      "learning_rate": 4.189771644079739e-06,
      "loss": 0.8196,
      "step": 3911
    },
    {
      "epoch": 0.1621285589953997,
      "grad_norm": 0.40603211522102356,
      "learning_rate": 4.18956442455137e-06,
      "loss": 0.7483,
      "step": 3912
    },
    {
      "epoch": 0.1621700029010734,
      "grad_norm": 0.43264731764793396,
      "learning_rate": 4.189357205023002e-06,
      "loss": 0.7549,
      "step": 3913
    },
    {
      "epoch": 0.16221144680674707,
      "grad_norm": 0.4516741931438446,
      "learning_rate": 4.189149985494633e-06,
      "loss": 0.7024,
      "step": 3914
    },
    {
      "epoch": 0.16225289071242074,
      "grad_norm": 0.41795071959495544,
      "learning_rate": 4.188942765966265e-06,
      "loss": 0.7427,
      "step": 3915
    },
    {
      "epoch": 0.16229433461809442,
      "grad_norm": 0.4450035095214844,
      "learning_rate": 4.188735546437897e-06,
      "loss": 0.8152,
      "step": 3916
    },
    {
      "epoch": 0.16233577852376807,
      "grad_norm": 0.4148249924182892,
      "learning_rate": 4.188528326909528e-06,
      "loss": 0.7222,
      "step": 3917
    },
    {
      "epoch": 0.16237722242944175,
      "grad_norm": 0.4537728726863861,
      "learning_rate": 4.1883211073811595e-06,
      "loss": 0.7715,
      "step": 3918
    },
    {
      "epoch": 0.16241866633511542,
      "grad_norm": 0.43246373534202576,
      "learning_rate": 4.188113887852792e-06,
      "loss": 0.7498,
      "step": 3919
    },
    {
      "epoch": 0.1624601102407891,
      "grad_norm": 0.41484004259109497,
      "learning_rate": 4.187906668324423e-06,
      "loss": 0.6858,
      "step": 3920
    },
    {
      "epoch": 0.16250155414646275,
      "grad_norm": 0.4074244797229767,
      "learning_rate": 4.1876994487960545e-06,
      "loss": 0.7598,
      "step": 3921
    },
    {
      "epoch": 0.16254299805213643,
      "grad_norm": 0.42424190044403076,
      "learning_rate": 4.187492229267687e-06,
      "loss": 0.7505,
      "step": 3922
    },
    {
      "epoch": 0.1625844419578101,
      "grad_norm": 0.4053076207637787,
      "learning_rate": 4.187285009739318e-06,
      "loss": 0.7412,
      "step": 3923
    },
    {
      "epoch": 0.16262588586348378,
      "grad_norm": 0.48449522256851196,
      "learning_rate": 4.1870777902109495e-06,
      "loss": 0.7935,
      "step": 3924
    },
    {
      "epoch": 0.16266732976915743,
      "grad_norm": 0.41274046897888184,
      "learning_rate": 4.186870570682581e-06,
      "loss": 0.7275,
      "step": 3925
    },
    {
      "epoch": 0.1627087736748311,
      "grad_norm": 0.43400174379348755,
      "learning_rate": 4.186663351154213e-06,
      "loss": 0.77,
      "step": 3926
    },
    {
      "epoch": 0.16275021758050479,
      "grad_norm": 0.43948426842689514,
      "learning_rate": 4.1864561316258445e-06,
      "loss": 0.7354,
      "step": 3927
    },
    {
      "epoch": 0.16279166148617846,
      "grad_norm": 0.4341945946216583,
      "learning_rate": 4.186248912097477e-06,
      "loss": 0.7493,
      "step": 3928
    },
    {
      "epoch": 0.16283310539185214,
      "grad_norm": 0.4131866693496704,
      "learning_rate": 4.186041692569108e-06,
      "loss": 0.7366,
      "step": 3929
    },
    {
      "epoch": 0.1628745492975258,
      "grad_norm": 0.43092653155326843,
      "learning_rate": 4.1858344730407396e-06,
      "loss": 0.6958,
      "step": 3930
    },
    {
      "epoch": 0.16291599320319947,
      "grad_norm": 0.43224841356277466,
      "learning_rate": 4.185627253512372e-06,
      "loss": 0.6899,
      "step": 3931
    },
    {
      "epoch": 0.16295743710887314,
      "grad_norm": 0.43330833315849304,
      "learning_rate": 4.185420033984003e-06,
      "loss": 0.7175,
      "step": 3932
    },
    {
      "epoch": 0.16299888101454682,
      "grad_norm": 0.4147273302078247,
      "learning_rate": 4.1852128144556346e-06,
      "loss": 0.7378,
      "step": 3933
    },
    {
      "epoch": 0.16304032492022047,
      "grad_norm": 0.5594637393951416,
      "learning_rate": 4.185005594927266e-06,
      "loss": 0.8049,
      "step": 3934
    },
    {
      "epoch": 0.16308176882589415,
      "grad_norm": 0.448483943939209,
      "learning_rate": 4.184798375398898e-06,
      "loss": 0.7856,
      "step": 3935
    },
    {
      "epoch": 0.16312321273156782,
      "grad_norm": 0.40614721179008484,
      "learning_rate": 4.1845911558705296e-06,
      "loss": 0.6211,
      "step": 3936
    },
    {
      "epoch": 0.1631646566372415,
      "grad_norm": 0.4241173565387726,
      "learning_rate": 4.184383936342161e-06,
      "loss": 0.7198,
      "step": 3937
    },
    {
      "epoch": 0.16320610054291518,
      "grad_norm": 0.42847543954849243,
      "learning_rate": 4.184176716813793e-06,
      "loss": 0.7295,
      "step": 3938
    },
    {
      "epoch": 0.16324754444858883,
      "grad_norm": 0.41963788866996765,
      "learning_rate": 4.1839694972854246e-06,
      "loss": 0.7625,
      "step": 3939
    },
    {
      "epoch": 0.1632889883542625,
      "grad_norm": 0.4731791019439697,
      "learning_rate": 4.183762277757056e-06,
      "loss": 0.7205,
      "step": 3940
    },
    {
      "epoch": 0.16333043225993618,
      "grad_norm": 0.4068329632282257,
      "learning_rate": 4.183555058228687e-06,
      "loss": 0.6885,
      "step": 3941
    },
    {
      "epoch": 0.16337187616560986,
      "grad_norm": 0.4217120110988617,
      "learning_rate": 4.1833478387003196e-06,
      "loss": 0.7158,
      "step": 3942
    },
    {
      "epoch": 0.1634133200712835,
      "grad_norm": 0.42139700055122375,
      "learning_rate": 4.183140619171951e-06,
      "loss": 0.7356,
      "step": 3943
    },
    {
      "epoch": 0.16345476397695718,
      "grad_norm": 0.4465329051017761,
      "learning_rate": 4.182933399643583e-06,
      "loss": 0.7734,
      "step": 3944
    },
    {
      "epoch": 0.16349620788263086,
      "grad_norm": 0.4422605335712433,
      "learning_rate": 4.182726180115214e-06,
      "loss": 0.7888,
      "step": 3945
    },
    {
      "epoch": 0.16353765178830454,
      "grad_norm": 0.4786761999130249,
      "learning_rate": 4.182518960586846e-06,
      "loss": 0.7808,
      "step": 3946
    },
    {
      "epoch": 0.1635790956939782,
      "grad_norm": 0.41335394978523254,
      "learning_rate": 4.182311741058478e-06,
      "loss": 0.7201,
      "step": 3947
    },
    {
      "epoch": 0.16362053959965187,
      "grad_norm": 0.42506518959999084,
      "learning_rate": 4.18210452153011e-06,
      "loss": 0.7446,
      "step": 3948
    },
    {
      "epoch": 0.16366198350532554,
      "grad_norm": 0.423186331987381,
      "learning_rate": 4.181897302001741e-06,
      "loss": 0.7307,
      "step": 3949
    },
    {
      "epoch": 0.16370342741099922,
      "grad_norm": 0.4302513599395752,
      "learning_rate": 4.181690082473372e-06,
      "loss": 0.666,
      "step": 3950
    },
    {
      "epoch": 0.1637448713166729,
      "grad_norm": 0.387676477432251,
      "learning_rate": 4.181482862945005e-06,
      "loss": 0.7036,
      "step": 3951
    },
    {
      "epoch": 0.16378631522234655,
      "grad_norm": 0.4358347952365875,
      "learning_rate": 4.181275643416636e-06,
      "loss": 0.7473,
      "step": 3952
    },
    {
      "epoch": 0.16382775912802022,
      "grad_norm": 0.5031672716140747,
      "learning_rate": 4.181068423888267e-06,
      "loss": 0.8708,
      "step": 3953
    },
    {
      "epoch": 0.1638692030336939,
      "grad_norm": 0.3876796066761017,
      "learning_rate": 4.180861204359899e-06,
      "loss": 0.6792,
      "step": 3954
    },
    {
      "epoch": 0.16391064693936758,
      "grad_norm": 0.4248257577419281,
      "learning_rate": 4.180653984831531e-06,
      "loss": 0.7629,
      "step": 3955
    },
    {
      "epoch": 0.16395209084504123,
      "grad_norm": 0.45140397548675537,
      "learning_rate": 4.180446765303162e-06,
      "loss": 0.7388,
      "step": 3956
    },
    {
      "epoch": 0.1639935347507149,
      "grad_norm": 0.4637148678302765,
      "learning_rate": 4.180239545774794e-06,
      "loss": 0.6992,
      "step": 3957
    },
    {
      "epoch": 0.16403497865638858,
      "grad_norm": 0.4246406555175781,
      "learning_rate": 4.180032326246426e-06,
      "loss": 0.7332,
      "step": 3958
    },
    {
      "epoch": 0.16407642256206226,
      "grad_norm": 0.39203453063964844,
      "learning_rate": 4.179825106718057e-06,
      "loss": 0.7146,
      "step": 3959
    },
    {
      "epoch": 0.1641178664677359,
      "grad_norm": 0.41353005170822144,
      "learning_rate": 4.17961788718969e-06,
      "loss": 0.7155,
      "step": 3960
    },
    {
      "epoch": 0.16415931037340958,
      "grad_norm": 0.41028448939323425,
      "learning_rate": 4.17941066766132e-06,
      "loss": 0.6794,
      "step": 3961
    },
    {
      "epoch": 0.16420075427908326,
      "grad_norm": 0.44359683990478516,
      "learning_rate": 4.179203448132952e-06,
      "loss": 0.7217,
      "step": 3962
    },
    {
      "epoch": 0.16424219818475694,
      "grad_norm": 0.452467679977417,
      "learning_rate": 4.178996228604585e-06,
      "loss": 0.7759,
      "step": 3963
    },
    {
      "epoch": 0.16428364209043061,
      "grad_norm": 0.4533520042896271,
      "learning_rate": 4.178789009076216e-06,
      "loss": 0.6893,
      "step": 3964
    },
    {
      "epoch": 0.16432508599610426,
      "grad_norm": 0.4402236342430115,
      "learning_rate": 4.178581789547847e-06,
      "loss": 0.6709,
      "step": 3965
    },
    {
      "epoch": 0.16436652990177794,
      "grad_norm": 0.432903915643692,
      "learning_rate": 4.178374570019479e-06,
      "loss": 0.8,
      "step": 3966
    },
    {
      "epoch": 0.16440797380745162,
      "grad_norm": 0.4020920693874359,
      "learning_rate": 4.178167350491111e-06,
      "loss": 0.6924,
      "step": 3967
    },
    {
      "epoch": 0.1644494177131253,
      "grad_norm": 0.42549023032188416,
      "learning_rate": 4.177960130962742e-06,
      "loss": 0.7205,
      "step": 3968
    },
    {
      "epoch": 0.16449086161879894,
      "grad_norm": 0.4136500954627991,
      "learning_rate": 4.177752911434374e-06,
      "loss": 0.7224,
      "step": 3969
    },
    {
      "epoch": 0.16453230552447262,
      "grad_norm": 0.43125662207603455,
      "learning_rate": 4.177545691906005e-06,
      "loss": 0.6881,
      "step": 3970
    },
    {
      "epoch": 0.1645737494301463,
      "grad_norm": 0.4146919548511505,
      "learning_rate": 4.177338472377637e-06,
      "loss": 0.7551,
      "step": 3971
    },
    {
      "epoch": 0.16461519333581998,
      "grad_norm": 0.4323260486125946,
      "learning_rate": 4.177131252849269e-06,
      "loss": 0.7119,
      "step": 3972
    },
    {
      "epoch": 0.16465663724149363,
      "grad_norm": 0.41846591234207153,
      "learning_rate": 4.1769240333209e-06,
      "loss": 0.7036,
      "step": 3973
    },
    {
      "epoch": 0.1646980811471673,
      "grad_norm": 0.40757814049720764,
      "learning_rate": 4.176716813792532e-06,
      "loss": 0.7043,
      "step": 3974
    },
    {
      "epoch": 0.16473952505284098,
      "grad_norm": 0.4579413831233978,
      "learning_rate": 4.176509594264164e-06,
      "loss": 0.8242,
      "step": 3975
    },
    {
      "epoch": 0.16478096895851466,
      "grad_norm": 0.38648974895477295,
      "learning_rate": 4.176302374735796e-06,
      "loss": 0.7195,
      "step": 3976
    },
    {
      "epoch": 0.16482241286418833,
      "grad_norm": 0.4380299150943756,
      "learning_rate": 4.1760951552074265e-06,
      "loss": 0.7676,
      "step": 3977
    },
    {
      "epoch": 0.16486385676986198,
      "grad_norm": 0.4183562099933624,
      "learning_rate": 4.175887935679059e-06,
      "loss": 0.686,
      "step": 3978
    },
    {
      "epoch": 0.16490530067553566,
      "grad_norm": 0.4199469983577728,
      "learning_rate": 4.17568071615069e-06,
      "loss": 0.6858,
      "step": 3979
    },
    {
      "epoch": 0.16494674458120934,
      "grad_norm": 0.4656279981136322,
      "learning_rate": 4.175473496622322e-06,
      "loss": 0.7596,
      "step": 3980
    },
    {
      "epoch": 0.16498818848688301,
      "grad_norm": 0.41313594579696655,
      "learning_rate": 4.175266277093954e-06,
      "loss": 0.7543,
      "step": 3981
    },
    {
      "epoch": 0.16502963239255666,
      "grad_norm": 0.42932602763175964,
      "learning_rate": 4.175059057565585e-06,
      "loss": 0.7078,
      "step": 3982
    },
    {
      "epoch": 0.16507107629823034,
      "grad_norm": 0.44365110993385315,
      "learning_rate": 4.174851838037217e-06,
      "loss": 0.7307,
      "step": 3983
    },
    {
      "epoch": 0.16511252020390402,
      "grad_norm": 0.4406120479106903,
      "learning_rate": 4.174644618508849e-06,
      "loss": 0.6836,
      "step": 3984
    },
    {
      "epoch": 0.1651539641095777,
      "grad_norm": 0.4153088927268982,
      "learning_rate": 4.17443739898048e-06,
      "loss": 0.6853,
      "step": 3985
    },
    {
      "epoch": 0.16519540801525134,
      "grad_norm": 0.42502832412719727,
      "learning_rate": 4.1742301794521115e-06,
      "loss": 0.7,
      "step": 3986
    },
    {
      "epoch": 0.16523685192092502,
      "grad_norm": 0.3973308801651001,
      "learning_rate": 4.174022959923744e-06,
      "loss": 0.7063,
      "step": 3987
    },
    {
      "epoch": 0.1652782958265987,
      "grad_norm": 0.448110431432724,
      "learning_rate": 4.173815740395375e-06,
      "loss": 0.7849,
      "step": 3988
    },
    {
      "epoch": 0.16531973973227237,
      "grad_norm": 0.40824928879737854,
      "learning_rate": 4.1736085208670066e-06,
      "loss": 0.7045,
      "step": 3989
    },
    {
      "epoch": 0.16536118363794605,
      "grad_norm": 0.42675846815109253,
      "learning_rate": 4.173401301338639e-06,
      "loss": 0.7175,
      "step": 3990
    },
    {
      "epoch": 0.1654026275436197,
      "grad_norm": 0.44289180636405945,
      "learning_rate": 4.17319408181027e-06,
      "loss": 0.6904,
      "step": 3991
    },
    {
      "epoch": 0.16544407144929338,
      "grad_norm": 0.45638832449913025,
      "learning_rate": 4.1729868622819016e-06,
      "loss": 0.7351,
      "step": 3992
    },
    {
      "epoch": 0.16548551535496706,
      "grad_norm": 0.4534793198108673,
      "learning_rate": 4.172779642753533e-06,
      "loss": 0.7827,
      "step": 3993
    },
    {
      "epoch": 0.16552695926064073,
      "grad_norm": 0.4408583641052246,
      "learning_rate": 4.172572423225165e-06,
      "loss": 0.7581,
      "step": 3994
    },
    {
      "epoch": 0.16556840316631438,
      "grad_norm": 0.3889797031879425,
      "learning_rate": 4.1723652036967966e-06,
      "loss": 0.6903,
      "step": 3995
    },
    {
      "epoch": 0.16560984707198806,
      "grad_norm": 0.4430372714996338,
      "learning_rate": 4.172157984168429e-06,
      "loss": 0.7732,
      "step": 3996
    },
    {
      "epoch": 0.16565129097766174,
      "grad_norm": 0.44308802485466003,
      "learning_rate": 4.17195076464006e-06,
      "loss": 0.7124,
      "step": 3997
    },
    {
      "epoch": 0.1656927348833354,
      "grad_norm": 0.41470906138420105,
      "learning_rate": 4.1717435451116916e-06,
      "loss": 0.7473,
      "step": 3998
    },
    {
      "epoch": 0.1657341787890091,
      "grad_norm": 0.44298768043518066,
      "learning_rate": 4.171536325583324e-06,
      "loss": 0.7802,
      "step": 3999
    },
    {
      "epoch": 0.16577562269468274,
      "grad_norm": 0.4316277801990509,
      "learning_rate": 4.171329106054955e-06,
      "loss": 0.6786,
      "step": 4000
    },
    {
      "epoch": 0.16581706660035642,
      "grad_norm": 0.42960840463638306,
      "learning_rate": 4.1711218865265866e-06,
      "loss": 0.7012,
      "step": 4001
    },
    {
      "epoch": 0.1658585105060301,
      "grad_norm": 0.40676987171173096,
      "learning_rate": 4.170914666998218e-06,
      "loss": 0.7371,
      "step": 4002
    },
    {
      "epoch": 0.16589995441170377,
      "grad_norm": 0.4225432872772217,
      "learning_rate": 4.17070744746985e-06,
      "loss": 0.7908,
      "step": 4003
    },
    {
      "epoch": 0.16594139831737742,
      "grad_norm": 0.4528430700302124,
      "learning_rate": 4.1705002279414816e-06,
      "loss": 0.7756,
      "step": 4004
    },
    {
      "epoch": 0.1659828422230511,
      "grad_norm": 0.43588078022003174,
      "learning_rate": 4.170293008413113e-06,
      "loss": 0.738,
      "step": 4005
    },
    {
      "epoch": 0.16602428612872477,
      "grad_norm": 0.4559016525745392,
      "learning_rate": 4.170085788884744e-06,
      "loss": 0.7429,
      "step": 4006
    },
    {
      "epoch": 0.16606573003439845,
      "grad_norm": 0.43830540776252747,
      "learning_rate": 4.169878569356377e-06,
      "loss": 0.7094,
      "step": 4007
    },
    {
      "epoch": 0.1661071739400721,
      "grad_norm": 0.41823479533195496,
      "learning_rate": 4.169671349828008e-06,
      "loss": 0.7025,
      "step": 4008
    },
    {
      "epoch": 0.16614861784574578,
      "grad_norm": 0.4519365131855011,
      "learning_rate": 4.169464130299639e-06,
      "loss": 0.7263,
      "step": 4009
    },
    {
      "epoch": 0.16619006175141945,
      "grad_norm": 0.45159727334976196,
      "learning_rate": 4.169256910771272e-06,
      "loss": 0.7444,
      "step": 4010
    },
    {
      "epoch": 0.16623150565709313,
      "grad_norm": 0.422893226146698,
      "learning_rate": 4.169049691242903e-06,
      "loss": 0.6951,
      "step": 4011
    },
    {
      "epoch": 0.1662729495627668,
      "grad_norm": 0.516808032989502,
      "learning_rate": 4.168842471714535e-06,
      "loss": 0.7462,
      "step": 4012
    },
    {
      "epoch": 0.16631439346844046,
      "grad_norm": 0.40142881870269775,
      "learning_rate": 4.168635252186166e-06,
      "loss": 0.7177,
      "step": 4013
    },
    {
      "epoch": 0.16635583737411413,
      "grad_norm": 0.45885249972343445,
      "learning_rate": 4.168428032657798e-06,
      "loss": 0.6953,
      "step": 4014
    },
    {
      "epoch": 0.1663972812797878,
      "grad_norm": 0.4287058711051941,
      "learning_rate": 4.168220813129429e-06,
      "loss": 0.7703,
      "step": 4015
    },
    {
      "epoch": 0.1664387251854615,
      "grad_norm": 0.4489665627479553,
      "learning_rate": 4.168013593601062e-06,
      "loss": 0.7551,
      "step": 4016
    },
    {
      "epoch": 0.16648016909113514,
      "grad_norm": 0.4240235686302185,
      "learning_rate": 4.167806374072693e-06,
      "loss": 0.6899,
      "step": 4017
    },
    {
      "epoch": 0.16652161299680882,
      "grad_norm": 0.41933944821357727,
      "learning_rate": 4.167599154544324e-06,
      "loss": 0.7435,
      "step": 4018
    },
    {
      "epoch": 0.1665630569024825,
      "grad_norm": 0.42885735630989075,
      "learning_rate": 4.167391935015957e-06,
      "loss": 0.7434,
      "step": 4019
    },
    {
      "epoch": 0.16660450080815617,
      "grad_norm": 0.40203186869621277,
      "learning_rate": 4.167184715487588e-06,
      "loss": 0.7095,
      "step": 4020
    },
    {
      "epoch": 0.16664594471382982,
      "grad_norm": 0.41601675748825073,
      "learning_rate": 4.166977495959219e-06,
      "loss": 0.7109,
      "step": 4021
    },
    {
      "epoch": 0.1666873886195035,
      "grad_norm": 0.40761032700538635,
      "learning_rate": 4.166770276430851e-06,
      "loss": 0.7393,
      "step": 4022
    },
    {
      "epoch": 0.16672883252517717,
      "grad_norm": 0.42859432101249695,
      "learning_rate": 4.166563056902483e-06,
      "loss": 0.7266,
      "step": 4023
    },
    {
      "epoch": 0.16677027643085085,
      "grad_norm": 0.4399731159210205,
      "learning_rate": 4.166355837374114e-06,
      "loss": 0.7351,
      "step": 4024
    },
    {
      "epoch": 0.16681172033652453,
      "grad_norm": 0.44113704562187195,
      "learning_rate": 4.166148617845746e-06,
      "loss": 0.7808,
      "step": 4025
    },
    {
      "epoch": 0.16685316424219818,
      "grad_norm": 0.42984268069267273,
      "learning_rate": 4.165941398317378e-06,
      "loss": 0.7239,
      "step": 4026
    },
    {
      "epoch": 0.16689460814787185,
      "grad_norm": 0.42799112200737,
      "learning_rate": 4.165734178789009e-06,
      "loss": 0.7688,
      "step": 4027
    },
    {
      "epoch": 0.16693605205354553,
      "grad_norm": 0.40136244893074036,
      "learning_rate": 4.165526959260642e-06,
      "loss": 0.7172,
      "step": 4028
    },
    {
      "epoch": 0.1669774959592192,
      "grad_norm": 0.42733749747276306,
      "learning_rate": 4.165319739732272e-06,
      "loss": 0.6819,
      "step": 4029
    },
    {
      "epoch": 0.16701893986489286,
      "grad_norm": 0.441489577293396,
      "learning_rate": 4.165112520203904e-06,
      "loss": 0.7354,
      "step": 4030
    },
    {
      "epoch": 0.16706038377056653,
      "grad_norm": 0.3815517723560333,
      "learning_rate": 4.164905300675536e-06,
      "loss": 0.6633,
      "step": 4031
    },
    {
      "epoch": 0.1671018276762402,
      "grad_norm": 0.4398477375507355,
      "learning_rate": 4.164698081147168e-06,
      "loss": 0.8042,
      "step": 4032
    },
    {
      "epoch": 0.1671432715819139,
      "grad_norm": 0.43279215693473816,
      "learning_rate": 4.164490861618799e-06,
      "loss": 0.7329,
      "step": 4033
    },
    {
      "epoch": 0.16718471548758754,
      "grad_norm": 0.4133869707584381,
      "learning_rate": 4.164283642090431e-06,
      "loss": 0.6512,
      "step": 4034
    },
    {
      "epoch": 0.16722615939326121,
      "grad_norm": 0.41181620955467224,
      "learning_rate": 4.164076422562063e-06,
      "loss": 0.6775,
      "step": 4035
    },
    {
      "epoch": 0.1672676032989349,
      "grad_norm": 0.48185110092163086,
      "learning_rate": 4.163869203033694e-06,
      "loss": 0.7527,
      "step": 4036
    },
    {
      "epoch": 0.16730904720460857,
      "grad_norm": 0.44024205207824707,
      "learning_rate": 4.163661983505326e-06,
      "loss": 0.76,
      "step": 4037
    },
    {
      "epoch": 0.16735049111028225,
      "grad_norm": 0.4023451507091522,
      "learning_rate": 4.163454763976957e-06,
      "loss": 0.7004,
      "step": 4038
    },
    {
      "epoch": 0.1673919350159559,
      "grad_norm": 0.44601884484291077,
      "learning_rate": 4.163247544448589e-06,
      "loss": 0.7803,
      "step": 4039
    },
    {
      "epoch": 0.16743337892162957,
      "grad_norm": 0.4677041471004486,
      "learning_rate": 4.163040324920221e-06,
      "loss": 0.7546,
      "step": 4040
    },
    {
      "epoch": 0.16747482282730325,
      "grad_norm": 0.44302669167518616,
      "learning_rate": 4.162833105391852e-06,
      "loss": 0.7373,
      "step": 4041
    },
    {
      "epoch": 0.16751626673297693,
      "grad_norm": 0.42496371269226074,
      "learning_rate": 4.162625885863484e-06,
      "loss": 0.7747,
      "step": 4042
    },
    {
      "epoch": 0.16755771063865058,
      "grad_norm": 0.446289986371994,
      "learning_rate": 4.162418666335116e-06,
      "loss": 0.822,
      "step": 4043
    },
    {
      "epoch": 0.16759915454432425,
      "grad_norm": 0.40851324796676636,
      "learning_rate": 4.162211446806748e-06,
      "loss": 0.6687,
      "step": 4044
    },
    {
      "epoch": 0.16764059844999793,
      "grad_norm": 0.4266902506351471,
      "learning_rate": 4.1620042272783785e-06,
      "loss": 0.6819,
      "step": 4045
    },
    {
      "epoch": 0.1676820423556716,
      "grad_norm": 0.39563530683517456,
      "learning_rate": 4.161797007750011e-06,
      "loss": 0.71,
      "step": 4046
    },
    {
      "epoch": 0.16772348626134526,
      "grad_norm": 0.43665483593940735,
      "learning_rate": 4.161589788221642e-06,
      "loss": 0.7131,
      "step": 4047
    },
    {
      "epoch": 0.16776493016701893,
      "grad_norm": 0.4138428568840027,
      "learning_rate": 4.161382568693274e-06,
      "loss": 0.7512,
      "step": 4048
    },
    {
      "epoch": 0.1678063740726926,
      "grad_norm": 0.4195769429206848,
      "learning_rate": 4.161175349164906e-06,
      "loss": 0.79,
      "step": 4049
    },
    {
      "epoch": 0.1678478179783663,
      "grad_norm": 0.40827223658561707,
      "learning_rate": 4.160968129636537e-06,
      "loss": 0.7583,
      "step": 4050
    },
    {
      "epoch": 0.16788926188403996,
      "grad_norm": 0.44803479313850403,
      "learning_rate": 4.160760910108169e-06,
      "loss": 0.6868,
      "step": 4051
    },
    {
      "epoch": 0.1679307057897136,
      "grad_norm": 0.44756048917770386,
      "learning_rate": 4.160553690579801e-06,
      "loss": 0.7944,
      "step": 4052
    },
    {
      "epoch": 0.1679721496953873,
      "grad_norm": 0.4231491684913635,
      "learning_rate": 4.160346471051432e-06,
      "loss": 0.676,
      "step": 4053
    },
    {
      "epoch": 0.16801359360106097,
      "grad_norm": 0.4277089536190033,
      "learning_rate": 4.1601392515230636e-06,
      "loss": 0.7444,
      "step": 4054
    },
    {
      "epoch": 0.16805503750673464,
      "grad_norm": 0.4218698740005493,
      "learning_rate": 4.159932031994696e-06,
      "loss": 0.7844,
      "step": 4055
    },
    {
      "epoch": 0.1680964814124083,
      "grad_norm": 0.43698549270629883,
      "learning_rate": 4.159724812466327e-06,
      "loss": 0.6996,
      "step": 4056
    },
    {
      "epoch": 0.16813792531808197,
      "grad_norm": 0.46601346135139465,
      "learning_rate": 4.1595175929379586e-06,
      "loss": 0.8347,
      "step": 4057
    },
    {
      "epoch": 0.16817936922375565,
      "grad_norm": 0.4093257486820221,
      "learning_rate": 4.15931037340959e-06,
      "loss": 0.7715,
      "step": 4058
    },
    {
      "epoch": 0.16822081312942933,
      "grad_norm": 0.4418031573295593,
      "learning_rate": 4.159103153881222e-06,
      "loss": 0.7483,
      "step": 4059
    },
    {
      "epoch": 0.16826225703510297,
      "grad_norm": 0.4325525164604187,
      "learning_rate": 4.1588959343528536e-06,
      "loss": 0.7891,
      "step": 4060
    },
    {
      "epoch": 0.16830370094077665,
      "grad_norm": 0.4223649799823761,
      "learning_rate": 4.158688714824485e-06,
      "loss": 0.754,
      "step": 4061
    },
    {
      "epoch": 0.16834514484645033,
      "grad_norm": 0.46665868163108826,
      "learning_rate": 4.158481495296117e-06,
      "loss": 0.7937,
      "step": 4062
    },
    {
      "epoch": 0.168386588752124,
      "grad_norm": 0.42945510149002075,
      "learning_rate": 4.1582742757677486e-06,
      "loss": 0.783,
      "step": 4063
    },
    {
      "epoch": 0.16842803265779768,
      "grad_norm": 0.4173888862133026,
      "learning_rate": 4.158067056239381e-06,
      "loss": 0.8145,
      "step": 4064
    },
    {
      "epoch": 0.16846947656347133,
      "grad_norm": 0.45253029465675354,
      "learning_rate": 4.157859836711012e-06,
      "loss": 0.7291,
      "step": 4065
    },
    {
      "epoch": 0.168510920469145,
      "grad_norm": 0.4167127013206482,
      "learning_rate": 4.157652617182644e-06,
      "loss": 0.7217,
      "step": 4066
    },
    {
      "epoch": 0.1685523643748187,
      "grad_norm": 0.43157029151916504,
      "learning_rate": 4.157445397654275e-06,
      "loss": 0.6862,
      "step": 4067
    },
    {
      "epoch": 0.16859380828049236,
      "grad_norm": 0.4422221779823303,
      "learning_rate": 4.157238178125907e-06,
      "loss": 0.7424,
      "step": 4068
    },
    {
      "epoch": 0.168635252186166,
      "grad_norm": 0.42759624123573303,
      "learning_rate": 4.157030958597539e-06,
      "loss": 0.7703,
      "step": 4069
    },
    {
      "epoch": 0.1686766960918397,
      "grad_norm": 0.4230738878250122,
      "learning_rate": 4.15682373906917e-06,
      "loss": 0.7111,
      "step": 4070
    },
    {
      "epoch": 0.16871813999751337,
      "grad_norm": 0.39961332082748413,
      "learning_rate": 4.156616519540802e-06,
      "loss": 0.7043,
      "step": 4071
    },
    {
      "epoch": 0.16875958390318704,
      "grad_norm": 0.4062801003456116,
      "learning_rate": 4.156409300012434e-06,
      "loss": 0.7009,
      "step": 4072
    },
    {
      "epoch": 0.16880102780886072,
      "grad_norm": 0.44138258695602417,
      "learning_rate": 4.156202080484065e-06,
      "loss": 0.7405,
      "step": 4073
    },
    {
      "epoch": 0.16884247171453437,
      "grad_norm": 0.44288939237594604,
      "learning_rate": 4.155994860955696e-06,
      "loss": 0.693,
      "step": 4074
    },
    {
      "epoch": 0.16888391562020805,
      "grad_norm": 0.4357079267501831,
      "learning_rate": 4.155787641427329e-06,
      "loss": 0.7783,
      "step": 4075
    },
    {
      "epoch": 0.16892535952588172,
      "grad_norm": 0.4279991388320923,
      "learning_rate": 4.15558042189896e-06,
      "loss": 0.688,
      "step": 4076
    },
    {
      "epoch": 0.1689668034315554,
      "grad_norm": 0.4231995642185211,
      "learning_rate": 4.155373202370591e-06,
      "loss": 0.6996,
      "step": 4077
    },
    {
      "epoch": 0.16900824733722905,
      "grad_norm": 0.4232848584651947,
      "learning_rate": 4.155165982842224e-06,
      "loss": 0.7244,
      "step": 4078
    },
    {
      "epoch": 0.16904969124290273,
      "grad_norm": 0.4019148647785187,
      "learning_rate": 4.154958763313855e-06,
      "loss": 0.73,
      "step": 4079
    },
    {
      "epoch": 0.1690911351485764,
      "grad_norm": 0.42240452766418457,
      "learning_rate": 4.154751543785487e-06,
      "loss": 0.7595,
      "step": 4080
    },
    {
      "epoch": 0.16913257905425008,
      "grad_norm": 0.4243178367614746,
      "learning_rate": 4.154544324257119e-06,
      "loss": 0.7229,
      "step": 4081
    },
    {
      "epoch": 0.16917402295992373,
      "grad_norm": 0.41425445675849915,
      "learning_rate": 4.15433710472875e-06,
      "loss": 0.7991,
      "step": 4082
    },
    {
      "epoch": 0.1692154668655974,
      "grad_norm": 0.41091328859329224,
      "learning_rate": 4.154129885200381e-06,
      "loss": 0.6925,
      "step": 4083
    },
    {
      "epoch": 0.16925691077127109,
      "grad_norm": 0.4051581025123596,
      "learning_rate": 4.153922665672014e-06,
      "loss": 0.7988,
      "step": 4084
    },
    {
      "epoch": 0.16929835467694476,
      "grad_norm": 0.4197920262813568,
      "learning_rate": 4.153715446143645e-06,
      "loss": 0.72,
      "step": 4085
    },
    {
      "epoch": 0.16933979858261844,
      "grad_norm": 0.41297194361686707,
      "learning_rate": 4.153508226615276e-06,
      "loss": 0.7186,
      "step": 4086
    },
    {
      "epoch": 0.1693812424882921,
      "grad_norm": 0.4065501391887665,
      "learning_rate": 4.153301007086909e-06,
      "loss": 0.7136,
      "step": 4087
    },
    {
      "epoch": 0.16942268639396577,
      "grad_norm": 0.4138219356536865,
      "learning_rate": 4.15309378755854e-06,
      "loss": 0.6766,
      "step": 4088
    },
    {
      "epoch": 0.16946413029963944,
      "grad_norm": 0.439310759305954,
      "learning_rate": 4.152886568030171e-06,
      "loss": 0.7302,
      "step": 4089
    },
    {
      "epoch": 0.16950557420531312,
      "grad_norm": 0.46391627192497253,
      "learning_rate": 4.152679348501803e-06,
      "loss": 0.7849,
      "step": 4090
    },
    {
      "epoch": 0.16954701811098677,
      "grad_norm": 0.4477780759334564,
      "learning_rate": 4.152472128973435e-06,
      "loss": 0.7234,
      "step": 4091
    },
    {
      "epoch": 0.16958846201666045,
      "grad_norm": 0.4495815634727478,
      "learning_rate": 4.152264909445066e-06,
      "loss": 0.7729,
      "step": 4092
    },
    {
      "epoch": 0.16962990592233412,
      "grad_norm": 0.48693814873695374,
      "learning_rate": 4.152057689916698e-06,
      "loss": 0.8271,
      "step": 4093
    },
    {
      "epoch": 0.1696713498280078,
      "grad_norm": 0.446256548166275,
      "learning_rate": 4.15185047038833e-06,
      "loss": 0.761,
      "step": 4094
    },
    {
      "epoch": 0.16971279373368145,
      "grad_norm": 0.42985719442367554,
      "learning_rate": 4.151643250859961e-06,
      "loss": 0.7383,
      "step": 4095
    },
    {
      "epoch": 0.16975423763935513,
      "grad_norm": 0.4384533166885376,
      "learning_rate": 4.151436031331594e-06,
      "loss": 0.7612,
      "step": 4096
    },
    {
      "epoch": 0.1697956815450288,
      "grad_norm": 0.42223790287971497,
      "learning_rate": 4.151228811803224e-06,
      "loss": 0.7559,
      "step": 4097
    },
    {
      "epoch": 0.16983712545070248,
      "grad_norm": 0.4279736280441284,
      "learning_rate": 4.151021592274856e-06,
      "loss": 0.7524,
      "step": 4098
    },
    {
      "epoch": 0.16987856935637616,
      "grad_norm": 0.44225189089775085,
      "learning_rate": 4.150814372746488e-06,
      "loss": 0.7634,
      "step": 4099
    },
    {
      "epoch": 0.1699200132620498,
      "grad_norm": 0.4234447777271271,
      "learning_rate": 4.15060715321812e-06,
      "loss": 0.7036,
      "step": 4100
    },
    {
      "epoch": 0.16996145716772348,
      "grad_norm": 0.4547146260738373,
      "learning_rate": 4.150399933689751e-06,
      "loss": 0.7412,
      "step": 4101
    },
    {
      "epoch": 0.17000290107339716,
      "grad_norm": 0.45162782073020935,
      "learning_rate": 4.150192714161383e-06,
      "loss": 0.7188,
      "step": 4102
    },
    {
      "epoch": 0.17004434497907084,
      "grad_norm": 0.4342907965183258,
      "learning_rate": 4.149985494633015e-06,
      "loss": 0.777,
      "step": 4103
    },
    {
      "epoch": 0.1700857888847445,
      "grad_norm": 0.4333324134349823,
      "learning_rate": 4.149778275104646e-06,
      "loss": 0.7083,
      "step": 4104
    },
    {
      "epoch": 0.17012723279041816,
      "grad_norm": 0.4206410050392151,
      "learning_rate": 4.149571055576278e-06,
      "loss": 0.7418,
      "step": 4105
    },
    {
      "epoch": 0.17016867669609184,
      "grad_norm": 0.3920663297176361,
      "learning_rate": 4.149363836047909e-06,
      "loss": 0.7363,
      "step": 4106
    },
    {
      "epoch": 0.17021012060176552,
      "grad_norm": 0.4368460178375244,
      "learning_rate": 4.149156616519541e-06,
      "loss": 0.7361,
      "step": 4107
    },
    {
      "epoch": 0.17025156450743917,
      "grad_norm": 0.4273522198200226,
      "learning_rate": 4.148949396991173e-06,
      "loss": 0.7595,
      "step": 4108
    },
    {
      "epoch": 0.17029300841311285,
      "grad_norm": 0.4365537762641907,
      "learning_rate": 4.148742177462804e-06,
      "loss": 0.7723,
      "step": 4109
    },
    {
      "epoch": 0.17033445231878652,
      "grad_norm": 0.4658549726009369,
      "learning_rate": 4.1485349579344356e-06,
      "loss": 0.7766,
      "step": 4110
    },
    {
      "epoch": 0.1703758962244602,
      "grad_norm": 0.44773876667022705,
      "learning_rate": 4.148327738406068e-06,
      "loss": 0.7271,
      "step": 4111
    },
    {
      "epoch": 0.17041734013013388,
      "grad_norm": 0.4418567717075348,
      "learning_rate": 4.1481205188777e-06,
      "loss": 0.7227,
      "step": 4112
    },
    {
      "epoch": 0.17045878403580753,
      "grad_norm": 0.4278064966201782,
      "learning_rate": 4.1479132993493306e-06,
      "loss": 0.7418,
      "step": 4113
    },
    {
      "epoch": 0.1705002279414812,
      "grad_norm": 0.41931578516960144,
      "learning_rate": 4.147706079820963e-06,
      "loss": 0.7349,
      "step": 4114
    },
    {
      "epoch": 0.17054167184715488,
      "grad_norm": 0.4626260995864868,
      "learning_rate": 4.147498860292594e-06,
      "loss": 0.7537,
      "step": 4115
    },
    {
      "epoch": 0.17058311575282856,
      "grad_norm": 0.4503745138645172,
      "learning_rate": 4.147291640764226e-06,
      "loss": 0.7825,
      "step": 4116
    },
    {
      "epoch": 0.1706245596585022,
      "grad_norm": 0.48199525475502014,
      "learning_rate": 4.147084421235858e-06,
      "loss": 0.814,
      "step": 4117
    },
    {
      "epoch": 0.17066600356417588,
      "grad_norm": 0.4059840440750122,
      "learning_rate": 4.146877201707489e-06,
      "loss": 0.7341,
      "step": 4118
    },
    {
      "epoch": 0.17070744746984956,
      "grad_norm": 0.4386116564273834,
      "learning_rate": 4.1466699821791206e-06,
      "loss": 0.7388,
      "step": 4119
    },
    {
      "epoch": 0.17074889137552324,
      "grad_norm": 0.48894020915031433,
      "learning_rate": 4.146462762650753e-06,
      "loss": 0.8145,
      "step": 4120
    },
    {
      "epoch": 0.1707903352811969,
      "grad_norm": 0.45389366149902344,
      "learning_rate": 4.146255543122384e-06,
      "loss": 0.72,
      "step": 4121
    },
    {
      "epoch": 0.17083177918687056,
      "grad_norm": 0.4295516312122345,
      "learning_rate": 4.1460483235940156e-06,
      "loss": 0.7039,
      "step": 4122
    },
    {
      "epoch": 0.17087322309254424,
      "grad_norm": 0.44105562567710876,
      "learning_rate": 4.145841104065648e-06,
      "loss": 0.6743,
      "step": 4123
    },
    {
      "epoch": 0.17091466699821792,
      "grad_norm": 0.44636979699134827,
      "learning_rate": 4.145633884537279e-06,
      "loss": 0.7712,
      "step": 4124
    },
    {
      "epoch": 0.1709561109038916,
      "grad_norm": 0.47286689281463623,
      "learning_rate": 4.145426665008911e-06,
      "loss": 0.7695,
      "step": 4125
    },
    {
      "epoch": 0.17099755480956524,
      "grad_norm": 0.4634702205657959,
      "learning_rate": 4.145219445480542e-06,
      "loss": 0.7666,
      "step": 4126
    },
    {
      "epoch": 0.17103899871523892,
      "grad_norm": 0.4256576597690582,
      "learning_rate": 4.145012225952174e-06,
      "loss": 0.7419,
      "step": 4127
    },
    {
      "epoch": 0.1710804426209126,
      "grad_norm": 0.4284919202327728,
      "learning_rate": 4.144805006423806e-06,
      "loss": 0.7633,
      "step": 4128
    },
    {
      "epoch": 0.17112188652658628,
      "grad_norm": 0.47939515113830566,
      "learning_rate": 4.144597786895437e-06,
      "loss": 0.7878,
      "step": 4129
    },
    {
      "epoch": 0.17116333043225992,
      "grad_norm": 0.40596550703048706,
      "learning_rate": 4.144390567367069e-06,
      "loss": 0.7238,
      "step": 4130
    },
    {
      "epoch": 0.1712047743379336,
      "grad_norm": 0.439281165599823,
      "learning_rate": 4.144183347838701e-06,
      "loss": 0.7467,
      "step": 4131
    },
    {
      "epoch": 0.17124621824360728,
      "grad_norm": 0.4507501721382141,
      "learning_rate": 4.143976128310333e-06,
      "loss": 0.7004,
      "step": 4132
    },
    {
      "epoch": 0.17128766214928096,
      "grad_norm": 0.43110886216163635,
      "learning_rate": 4.143768908781964e-06,
      "loss": 0.7115,
      "step": 4133
    },
    {
      "epoch": 0.1713291060549546,
      "grad_norm": 0.40868330001831055,
      "learning_rate": 4.143561689253596e-06,
      "loss": 0.7634,
      "step": 4134
    },
    {
      "epoch": 0.17137054996062828,
      "grad_norm": 0.4266672730445862,
      "learning_rate": 4.143354469725227e-06,
      "loss": 0.7334,
      "step": 4135
    },
    {
      "epoch": 0.17141199386630196,
      "grad_norm": 0.39296823740005493,
      "learning_rate": 4.143147250196859e-06,
      "loss": 0.7358,
      "step": 4136
    },
    {
      "epoch": 0.17145343777197564,
      "grad_norm": 0.45111462473869324,
      "learning_rate": 4.142940030668491e-06,
      "loss": 0.7788,
      "step": 4137
    },
    {
      "epoch": 0.1714948816776493,
      "grad_norm": 0.3998972475528717,
      "learning_rate": 4.142732811140122e-06,
      "loss": 0.7017,
      "step": 4138
    },
    {
      "epoch": 0.17153632558332296,
      "grad_norm": 0.41973164677619934,
      "learning_rate": 4.142525591611754e-06,
      "loss": 0.7468,
      "step": 4139
    },
    {
      "epoch": 0.17157776948899664,
      "grad_norm": 0.404287189245224,
      "learning_rate": 4.142318372083386e-06,
      "loss": 0.6727,
      "step": 4140
    },
    {
      "epoch": 0.17161921339467032,
      "grad_norm": 0.4475458264350891,
      "learning_rate": 4.142111152555017e-06,
      "loss": 0.8062,
      "step": 4141
    },
    {
      "epoch": 0.171660657300344,
      "grad_norm": 0.3942428231239319,
      "learning_rate": 4.141903933026648e-06,
      "loss": 0.7148,
      "step": 4142
    },
    {
      "epoch": 0.17170210120601764,
      "grad_norm": 0.4355577528476715,
      "learning_rate": 4.141696713498281e-06,
      "loss": 0.7273,
      "step": 4143
    },
    {
      "epoch": 0.17174354511169132,
      "grad_norm": 0.40482786297798157,
      "learning_rate": 4.141489493969912e-06,
      "loss": 0.7554,
      "step": 4144
    },
    {
      "epoch": 0.171784989017365,
      "grad_norm": 0.4200670123100281,
      "learning_rate": 4.141282274441543e-06,
      "loss": 0.6975,
      "step": 4145
    },
    {
      "epoch": 0.17182643292303867,
      "grad_norm": 0.3948560953140259,
      "learning_rate": 4.141075054913176e-06,
      "loss": 0.7036,
      "step": 4146
    },
    {
      "epoch": 0.17186787682871235,
      "grad_norm": 0.43616798520088196,
      "learning_rate": 4.140867835384807e-06,
      "loss": 0.741,
      "step": 4147
    },
    {
      "epoch": 0.171909320734386,
      "grad_norm": 0.39573100209236145,
      "learning_rate": 4.140660615856439e-06,
      "loss": 0.7046,
      "step": 4148
    },
    {
      "epoch": 0.17195076464005968,
      "grad_norm": 0.8043258190155029,
      "learning_rate": 4.140453396328071e-06,
      "loss": 0.7205,
      "step": 4149
    },
    {
      "epoch": 0.17199220854573335,
      "grad_norm": 0.44154685735702515,
      "learning_rate": 4.140246176799702e-06,
      "loss": 0.804,
      "step": 4150
    },
    {
      "epoch": 0.17203365245140703,
      "grad_norm": 0.4319092035293579,
      "learning_rate": 4.140038957271333e-06,
      "loss": 0.7446,
      "step": 4151
    },
    {
      "epoch": 0.17207509635708068,
      "grad_norm": 0.4138278663158417,
      "learning_rate": 4.139831737742966e-06,
      "loss": 0.718,
      "step": 4152
    },
    {
      "epoch": 0.17211654026275436,
      "grad_norm": 0.4278290271759033,
      "learning_rate": 4.139624518214597e-06,
      "loss": 0.7494,
      "step": 4153
    },
    {
      "epoch": 0.17215798416842804,
      "grad_norm": 0.452711284160614,
      "learning_rate": 4.139417298686228e-06,
      "loss": 0.6721,
      "step": 4154
    },
    {
      "epoch": 0.1721994280741017,
      "grad_norm": 0.41718626022338867,
      "learning_rate": 4.139210079157861e-06,
      "loss": 0.7144,
      "step": 4155
    },
    {
      "epoch": 0.17224087197977536,
      "grad_norm": 0.4729371964931488,
      "learning_rate": 4.139002859629492e-06,
      "loss": 0.8325,
      "step": 4156
    },
    {
      "epoch": 0.17228231588544904,
      "grad_norm": 0.4808465242385864,
      "learning_rate": 4.138795640101123e-06,
      "loss": 0.7349,
      "step": 4157
    },
    {
      "epoch": 0.17232375979112272,
      "grad_norm": 0.4174172580242157,
      "learning_rate": 4.138588420572755e-06,
      "loss": 0.6959,
      "step": 4158
    },
    {
      "epoch": 0.1723652036967964,
      "grad_norm": 0.45987755060195923,
      "learning_rate": 4.138381201044387e-06,
      "loss": 0.7039,
      "step": 4159
    },
    {
      "epoch": 0.17240664760247007,
      "grad_norm": 0.4068875312805176,
      "learning_rate": 4.138173981516018e-06,
      "loss": 0.7463,
      "step": 4160
    },
    {
      "epoch": 0.17244809150814372,
      "grad_norm": 0.4682177007198334,
      "learning_rate": 4.13796676198765e-06,
      "loss": 0.7634,
      "step": 4161
    },
    {
      "epoch": 0.1724895354138174,
      "grad_norm": 0.4065361022949219,
      "learning_rate": 4.137759542459281e-06,
      "loss": 0.6597,
      "step": 4162
    },
    {
      "epoch": 0.17253097931949107,
      "grad_norm": 0.41984304785728455,
      "learning_rate": 4.137552322930913e-06,
      "loss": 0.7788,
      "step": 4163
    },
    {
      "epoch": 0.17257242322516475,
      "grad_norm": 0.4202154874801636,
      "learning_rate": 4.137345103402546e-06,
      "loss": 0.719,
      "step": 4164
    },
    {
      "epoch": 0.1726138671308384,
      "grad_norm": 0.4359605014324188,
      "learning_rate": 4.137137883874176e-06,
      "loss": 0.6997,
      "step": 4165
    },
    {
      "epoch": 0.17265531103651208,
      "grad_norm": 0.45010313391685486,
      "learning_rate": 4.136930664345808e-06,
      "loss": 0.7617,
      "step": 4166
    },
    {
      "epoch": 0.17269675494218575,
      "grad_norm": 0.433231383562088,
      "learning_rate": 4.13672344481744e-06,
      "loss": 0.7437,
      "step": 4167
    },
    {
      "epoch": 0.17273819884785943,
      "grad_norm": 0.42225828766822815,
      "learning_rate": 4.136516225289072e-06,
      "loss": 0.6831,
      "step": 4168
    },
    {
      "epoch": 0.17277964275353308,
      "grad_norm": 0.4486711621284485,
      "learning_rate": 4.136309005760703e-06,
      "loss": 0.7192,
      "step": 4169
    },
    {
      "epoch": 0.17282108665920676,
      "grad_norm": 0.39400067925453186,
      "learning_rate": 4.136101786232335e-06,
      "loss": 0.676,
      "step": 4170
    },
    {
      "epoch": 0.17286253056488043,
      "grad_norm": 0.43347644805908203,
      "learning_rate": 4.135894566703966e-06,
      "loss": 0.7278,
      "step": 4171
    },
    {
      "epoch": 0.1729039744705541,
      "grad_norm": 0.4101540744304657,
      "learning_rate": 4.135687347175598e-06,
      "loss": 0.7332,
      "step": 4172
    },
    {
      "epoch": 0.1729454183762278,
      "grad_norm": 0.4142490327358246,
      "learning_rate": 4.13548012764723e-06,
      "loss": 0.6735,
      "step": 4173
    },
    {
      "epoch": 0.17298686228190144,
      "grad_norm": 0.4112522304058075,
      "learning_rate": 4.135272908118861e-06,
      "loss": 0.7722,
      "step": 4174
    },
    {
      "epoch": 0.17302830618757511,
      "grad_norm": 0.43039146065711975,
      "learning_rate": 4.135065688590493e-06,
      "loss": 0.7041,
      "step": 4175
    },
    {
      "epoch": 0.1730697500932488,
      "grad_norm": 0.44352975487709045,
      "learning_rate": 4.134858469062125e-06,
      "loss": 0.7698,
      "step": 4176
    },
    {
      "epoch": 0.17311119399892247,
      "grad_norm": 0.39544183015823364,
      "learning_rate": 4.134651249533756e-06,
      "loss": 0.6682,
      "step": 4177
    },
    {
      "epoch": 0.17315263790459612,
      "grad_norm": 0.47532305121421814,
      "learning_rate": 4.1344440300053876e-06,
      "loss": 0.7827,
      "step": 4178
    },
    {
      "epoch": 0.1731940818102698,
      "grad_norm": 0.4396321773529053,
      "learning_rate": 4.13423681047702e-06,
      "loss": 0.7385,
      "step": 4179
    },
    {
      "epoch": 0.17323552571594347,
      "grad_norm": 0.43444523215293884,
      "learning_rate": 4.134029590948651e-06,
      "loss": 0.7607,
      "step": 4180
    },
    {
      "epoch": 0.17327696962161715,
      "grad_norm": 0.42582258582115173,
      "learning_rate": 4.1338223714202826e-06,
      "loss": 0.7576,
      "step": 4181
    },
    {
      "epoch": 0.1733184135272908,
      "grad_norm": 0.4471827745437622,
      "learning_rate": 4.133615151891915e-06,
      "loss": 0.7507,
      "step": 4182
    },
    {
      "epoch": 0.17335985743296448,
      "grad_norm": 0.47482526302337646,
      "learning_rate": 4.133407932363546e-06,
      "loss": 0.7964,
      "step": 4183
    },
    {
      "epoch": 0.17340130133863815,
      "grad_norm": 0.4040158987045288,
      "learning_rate": 4.1332007128351784e-06,
      "loss": 0.6931,
      "step": 4184
    },
    {
      "epoch": 0.17344274524431183,
      "grad_norm": 0.4479696452617645,
      "learning_rate": 4.13299349330681e-06,
      "loss": 0.7115,
      "step": 4185
    },
    {
      "epoch": 0.1734841891499855,
      "grad_norm": 0.41967323422431946,
      "learning_rate": 4.132786273778441e-06,
      "loss": 0.7207,
      "step": 4186
    },
    {
      "epoch": 0.17352563305565916,
      "grad_norm": 0.4370481073856354,
      "learning_rate": 4.132579054250073e-06,
      "loss": 0.7883,
      "step": 4187
    },
    {
      "epoch": 0.17356707696133283,
      "grad_norm": 0.42650651931762695,
      "learning_rate": 4.132371834721705e-06,
      "loss": 0.7749,
      "step": 4188
    },
    {
      "epoch": 0.1736085208670065,
      "grad_norm": 0.4119715094566345,
      "learning_rate": 4.132164615193336e-06,
      "loss": 0.7407,
      "step": 4189
    },
    {
      "epoch": 0.1736499647726802,
      "grad_norm": 0.4390302002429962,
      "learning_rate": 4.131957395664968e-06,
      "loss": 0.701,
      "step": 4190
    },
    {
      "epoch": 0.17369140867835384,
      "grad_norm": 0.40100395679473877,
      "learning_rate": 4.1317501761366e-06,
      "loss": 0.7158,
      "step": 4191
    },
    {
      "epoch": 0.17373285258402751,
      "grad_norm": 0.39124998450279236,
      "learning_rate": 4.131542956608231e-06,
      "loss": 0.7228,
      "step": 4192
    },
    {
      "epoch": 0.1737742964897012,
      "grad_norm": 0.4262610673904419,
      "learning_rate": 4.131335737079863e-06,
      "loss": 0.7166,
      "step": 4193
    },
    {
      "epoch": 0.17381574039537487,
      "grad_norm": 0.43274012207984924,
      "learning_rate": 4.131128517551494e-06,
      "loss": 0.7268,
      "step": 4194
    },
    {
      "epoch": 0.17385718430104852,
      "grad_norm": 0.38935357332229614,
      "learning_rate": 4.130921298023126e-06,
      "loss": 0.7111,
      "step": 4195
    },
    {
      "epoch": 0.1738986282067222,
      "grad_norm": 0.41760388016700745,
      "learning_rate": 4.130714078494758e-06,
      "loss": 0.6892,
      "step": 4196
    },
    {
      "epoch": 0.17394007211239587,
      "grad_norm": 0.41746559739112854,
      "learning_rate": 4.130506858966389e-06,
      "loss": 0.7036,
      "step": 4197
    },
    {
      "epoch": 0.17398151601806955,
      "grad_norm": 0.4128584861755371,
      "learning_rate": 4.13029963943802e-06,
      "loss": 0.7644,
      "step": 4198
    },
    {
      "epoch": 0.17402295992374323,
      "grad_norm": 0.3830883204936981,
      "learning_rate": 4.130092419909653e-06,
      "loss": 0.6658,
      "step": 4199
    },
    {
      "epoch": 0.17406440382941687,
      "grad_norm": 0.4201417863368988,
      "learning_rate": 4.129885200381285e-06,
      "loss": 0.7802,
      "step": 4200
    },
    {
      "epoch": 0.17410584773509055,
      "grad_norm": 0.4346417784690857,
      "learning_rate": 4.129677980852916e-06,
      "loss": 0.7305,
      "step": 4201
    },
    {
      "epoch": 0.17414729164076423,
      "grad_norm": 0.4209913909435272,
      "learning_rate": 4.129470761324548e-06,
      "loss": 0.708,
      "step": 4202
    },
    {
      "epoch": 0.1741887355464379,
      "grad_norm": 0.42489689588546753,
      "learning_rate": 4.129263541796179e-06,
      "loss": 0.7178,
      "step": 4203
    },
    {
      "epoch": 0.17423017945211156,
      "grad_norm": 0.44613325595855713,
      "learning_rate": 4.129056322267811e-06,
      "loss": 0.7522,
      "step": 4204
    },
    {
      "epoch": 0.17427162335778523,
      "grad_norm": 0.4278450608253479,
      "learning_rate": 4.128849102739443e-06,
      "loss": 0.7473,
      "step": 4205
    },
    {
      "epoch": 0.1743130672634589,
      "grad_norm": 0.40253785252571106,
      "learning_rate": 4.128641883211074e-06,
      "loss": 0.772,
      "step": 4206
    },
    {
      "epoch": 0.1743545111691326,
      "grad_norm": 0.41442564129829407,
      "learning_rate": 4.128434663682706e-06,
      "loss": 0.7129,
      "step": 4207
    },
    {
      "epoch": 0.17439595507480624,
      "grad_norm": 0.4364762306213379,
      "learning_rate": 4.128227444154338e-06,
      "loss": 0.6649,
      "step": 4208
    },
    {
      "epoch": 0.1744373989804799,
      "grad_norm": 0.44691237807273865,
      "learning_rate": 4.128020224625969e-06,
      "loss": 0.7659,
      "step": 4209
    },
    {
      "epoch": 0.1744788428861536,
      "grad_norm": 0.4337712228298187,
      "learning_rate": 4.1278130050976e-06,
      "loss": 0.7446,
      "step": 4210
    },
    {
      "epoch": 0.17452028679182727,
      "grad_norm": 0.5062041878700256,
      "learning_rate": 4.127605785569233e-06,
      "loss": 0.7197,
      "step": 4211
    },
    {
      "epoch": 0.17456173069750094,
      "grad_norm": 0.44443896412849426,
      "learning_rate": 4.127398566040864e-06,
      "loss": 0.7415,
      "step": 4212
    },
    {
      "epoch": 0.1746031746031746,
      "grad_norm": 0.417375773191452,
      "learning_rate": 4.127191346512495e-06,
      "loss": 0.7375,
      "step": 4213
    },
    {
      "epoch": 0.17464461850884827,
      "grad_norm": 0.4277600944042206,
      "learning_rate": 4.126984126984127e-06,
      "loss": 0.7286,
      "step": 4214
    },
    {
      "epoch": 0.17468606241452195,
      "grad_norm": 0.4206206500530243,
      "learning_rate": 4.126776907455759e-06,
      "loss": 0.6997,
      "step": 4215
    },
    {
      "epoch": 0.17472750632019562,
      "grad_norm": 0.427682489156723,
      "learning_rate": 4.126569687927391e-06,
      "loss": 0.6738,
      "step": 4216
    },
    {
      "epoch": 0.17476895022586927,
      "grad_norm": 0.39393875002861023,
      "learning_rate": 4.126362468399023e-06,
      "loss": 0.6687,
      "step": 4217
    },
    {
      "epoch": 0.17481039413154295,
      "grad_norm": 0.47567206621170044,
      "learning_rate": 4.126155248870654e-06,
      "loss": 0.7629,
      "step": 4218
    },
    {
      "epoch": 0.17485183803721663,
      "grad_norm": 0.4112507998943329,
      "learning_rate": 4.125948029342285e-06,
      "loss": 0.7551,
      "step": 4219
    },
    {
      "epoch": 0.1748932819428903,
      "grad_norm": 0.49653053283691406,
      "learning_rate": 4.125740809813918e-06,
      "loss": 0.6866,
      "step": 4220
    },
    {
      "epoch": 0.17493472584856398,
      "grad_norm": 0.4260661005973816,
      "learning_rate": 4.125533590285549e-06,
      "loss": 0.7341,
      "step": 4221
    },
    {
      "epoch": 0.17497616975423763,
      "grad_norm": 0.4743230938911438,
      "learning_rate": 4.12532637075718e-06,
      "loss": 0.8005,
      "step": 4222
    },
    {
      "epoch": 0.1750176136599113,
      "grad_norm": 0.49983835220336914,
      "learning_rate": 4.125119151228812e-06,
      "loss": 0.7412,
      "step": 4223
    },
    {
      "epoch": 0.17505905756558499,
      "grad_norm": 0.42489340901374817,
      "learning_rate": 4.124911931700444e-06,
      "loss": 0.7681,
      "step": 4224
    },
    {
      "epoch": 0.17510050147125866,
      "grad_norm": 0.43316906690597534,
      "learning_rate": 4.124704712172075e-06,
      "loss": 0.7222,
      "step": 4225
    },
    {
      "epoch": 0.1751419453769323,
      "grad_norm": 0.42767786979675293,
      "learning_rate": 4.124497492643707e-06,
      "loss": 0.6965,
      "step": 4226
    },
    {
      "epoch": 0.175183389282606,
      "grad_norm": 0.4111917316913605,
      "learning_rate": 4.124290273115339e-06,
      "loss": 0.6895,
      "step": 4227
    },
    {
      "epoch": 0.17522483318827967,
      "grad_norm": 0.4314042329788208,
      "learning_rate": 4.12408305358697e-06,
      "loss": 0.739,
      "step": 4228
    },
    {
      "epoch": 0.17526627709395334,
      "grad_norm": 0.4350607097148895,
      "learning_rate": 4.123875834058602e-06,
      "loss": 0.7375,
      "step": 4229
    },
    {
      "epoch": 0.175307720999627,
      "grad_norm": 0.45220619440078735,
      "learning_rate": 4.123668614530233e-06,
      "loss": 0.7595,
      "step": 4230
    },
    {
      "epoch": 0.17534916490530067,
      "grad_norm": 0.45952433347702026,
      "learning_rate": 4.123461395001865e-06,
      "loss": 0.7957,
      "step": 4231
    },
    {
      "epoch": 0.17539060881097435,
      "grad_norm": 0.4205242991447449,
      "learning_rate": 4.123254175473497e-06,
      "loss": 0.7061,
      "step": 4232
    },
    {
      "epoch": 0.17543205271664802,
      "grad_norm": 0.45333150029182434,
      "learning_rate": 4.123046955945129e-06,
      "loss": 0.6853,
      "step": 4233
    },
    {
      "epoch": 0.1754734966223217,
      "grad_norm": 0.42033901810646057,
      "learning_rate": 4.12283973641676e-06,
      "loss": 0.672,
      "step": 4234
    },
    {
      "epoch": 0.17551494052799535,
      "grad_norm": 0.46275144815444946,
      "learning_rate": 4.122632516888392e-06,
      "loss": 0.7386,
      "step": 4235
    },
    {
      "epoch": 0.17555638443366903,
      "grad_norm": 0.45021653175354004,
      "learning_rate": 4.122425297360024e-06,
      "loss": 0.7385,
      "step": 4236
    },
    {
      "epoch": 0.1755978283393427,
      "grad_norm": 0.4110069274902344,
      "learning_rate": 4.122218077831655e-06,
      "loss": 0.6906,
      "step": 4237
    },
    {
      "epoch": 0.17563927224501638,
      "grad_norm": 0.4427078068256378,
      "learning_rate": 4.122010858303287e-06,
      "loss": 0.7339,
      "step": 4238
    },
    {
      "epoch": 0.17568071615069003,
      "grad_norm": 0.46116575598716736,
      "learning_rate": 4.121803638774918e-06,
      "loss": 0.7412,
      "step": 4239
    },
    {
      "epoch": 0.1757221600563637,
      "grad_norm": 0.39947497844696045,
      "learning_rate": 4.12159641924655e-06,
      "loss": 0.7489,
      "step": 4240
    },
    {
      "epoch": 0.17576360396203738,
      "grad_norm": 0.43869084119796753,
      "learning_rate": 4.121389199718182e-06,
      "loss": 0.7412,
      "step": 4241
    },
    {
      "epoch": 0.17580504786771106,
      "grad_norm": 0.42985743284225464,
      "learning_rate": 4.121181980189813e-06,
      "loss": 0.7114,
      "step": 4242
    },
    {
      "epoch": 0.1758464917733847,
      "grad_norm": 0.41117119789123535,
      "learning_rate": 4.1209747606614454e-06,
      "loss": 0.7871,
      "step": 4243
    },
    {
      "epoch": 0.1758879356790584,
      "grad_norm": 0.41140472888946533,
      "learning_rate": 4.120767541133077e-06,
      "loss": 0.7231,
      "step": 4244
    },
    {
      "epoch": 0.17592937958473207,
      "grad_norm": 0.457061231136322,
      "learning_rate": 4.120560321604708e-06,
      "loss": 0.75,
      "step": 4245
    },
    {
      "epoch": 0.17597082349040574,
      "grad_norm": 0.46602439880371094,
      "learning_rate": 4.12035310207634e-06,
      "loss": 0.7336,
      "step": 4246
    },
    {
      "epoch": 0.17601226739607942,
      "grad_norm": 0.4160914421081543,
      "learning_rate": 4.120145882547972e-06,
      "loss": 0.689,
      "step": 4247
    },
    {
      "epoch": 0.17605371130175307,
      "grad_norm": 0.43183350563049316,
      "learning_rate": 4.119938663019603e-06,
      "loss": 0.7625,
      "step": 4248
    },
    {
      "epoch": 0.17609515520742675,
      "grad_norm": 0.3859313130378723,
      "learning_rate": 4.119731443491235e-06,
      "loss": 0.77,
      "step": 4249
    },
    {
      "epoch": 0.17613659911310042,
      "grad_norm": 0.492988646030426,
      "learning_rate": 4.119524223962866e-06,
      "loss": 0.7579,
      "step": 4250
    },
    {
      "epoch": 0.1761780430187741,
      "grad_norm": 0.39864224195480347,
      "learning_rate": 4.119317004434498e-06,
      "loss": 0.7153,
      "step": 4251
    },
    {
      "epoch": 0.17621948692444775,
      "grad_norm": 0.40799373388290405,
      "learning_rate": 4.1191097849061304e-06,
      "loss": 0.7589,
      "step": 4252
    },
    {
      "epoch": 0.17626093083012143,
      "grad_norm": 0.4360237121582031,
      "learning_rate": 4.118902565377762e-06,
      "loss": 0.729,
      "step": 4253
    },
    {
      "epoch": 0.1763023747357951,
      "grad_norm": 0.41178247332572937,
      "learning_rate": 4.118695345849393e-06,
      "loss": 0.7241,
      "step": 4254
    },
    {
      "epoch": 0.17634381864146878,
      "grad_norm": 0.42210111021995544,
      "learning_rate": 4.118488126321025e-06,
      "loss": 0.7601,
      "step": 4255
    },
    {
      "epoch": 0.17638526254714243,
      "grad_norm": 0.4160026013851166,
      "learning_rate": 4.118280906792657e-06,
      "loss": 0.6973,
      "step": 4256
    },
    {
      "epoch": 0.1764267064528161,
      "grad_norm": 0.4246573746204376,
      "learning_rate": 4.118073687264288e-06,
      "loss": 0.7444,
      "step": 4257
    },
    {
      "epoch": 0.17646815035848978,
      "grad_norm": 0.41488221287727356,
      "learning_rate": 4.11786646773592e-06,
      "loss": 0.7316,
      "step": 4258
    },
    {
      "epoch": 0.17650959426416346,
      "grad_norm": 0.4662647247314453,
      "learning_rate": 4.117659248207551e-06,
      "loss": 0.7405,
      "step": 4259
    },
    {
      "epoch": 0.17655103816983714,
      "grad_norm": 0.3996306359767914,
      "learning_rate": 4.117452028679183e-06,
      "loss": 0.6599,
      "step": 4260
    },
    {
      "epoch": 0.1765924820755108,
      "grad_norm": 0.42388278245925903,
      "learning_rate": 4.117244809150815e-06,
      "loss": 0.6536,
      "step": 4261
    },
    {
      "epoch": 0.17663392598118446,
      "grad_norm": 0.4344223141670227,
      "learning_rate": 4.117037589622446e-06,
      "loss": 0.7434,
      "step": 4262
    },
    {
      "epoch": 0.17667536988685814,
      "grad_norm": 0.3850594758987427,
      "learning_rate": 4.116830370094078e-06,
      "loss": 0.6919,
      "step": 4263
    },
    {
      "epoch": 0.17671681379253182,
      "grad_norm": 0.45665624737739563,
      "learning_rate": 4.11662315056571e-06,
      "loss": 0.7454,
      "step": 4264
    },
    {
      "epoch": 0.17675825769820547,
      "grad_norm": 0.44247543811798096,
      "learning_rate": 4.116415931037341e-06,
      "loss": 0.7869,
      "step": 4265
    },
    {
      "epoch": 0.17679970160387914,
      "grad_norm": 0.4546608328819275,
      "learning_rate": 4.116208711508972e-06,
      "loss": 0.7522,
      "step": 4266
    },
    {
      "epoch": 0.17684114550955282,
      "grad_norm": 0.41410019993782043,
      "learning_rate": 4.116001491980605e-06,
      "loss": 0.7443,
      "step": 4267
    },
    {
      "epoch": 0.1768825894152265,
      "grad_norm": 0.4559858739376068,
      "learning_rate": 4.115794272452237e-06,
      "loss": 0.741,
      "step": 4268
    },
    {
      "epoch": 0.17692403332090015,
      "grad_norm": 0.42087429761886597,
      "learning_rate": 4.115587052923868e-06,
      "loss": 0.7236,
      "step": 4269
    },
    {
      "epoch": 0.17696547722657383,
      "grad_norm": 0.4106200635433197,
      "learning_rate": 4.1153798333955e-06,
      "loss": 0.6926,
      "step": 4270
    },
    {
      "epoch": 0.1770069211322475,
      "grad_norm": 0.45233482122421265,
      "learning_rate": 4.115172613867131e-06,
      "loss": 0.7683,
      "step": 4271
    },
    {
      "epoch": 0.17704836503792118,
      "grad_norm": 0.43223050236701965,
      "learning_rate": 4.114965394338763e-06,
      "loss": 0.6833,
      "step": 4272
    },
    {
      "epoch": 0.17708980894359486,
      "grad_norm": 0.4224109351634979,
      "learning_rate": 4.114758174810395e-06,
      "loss": 0.712,
      "step": 4273
    },
    {
      "epoch": 0.1771312528492685,
      "grad_norm": 0.4079419672489166,
      "learning_rate": 4.114550955282026e-06,
      "loss": 0.6975,
      "step": 4274
    },
    {
      "epoch": 0.17717269675494218,
      "grad_norm": 0.41453585028648376,
      "learning_rate": 4.114343735753657e-06,
      "loss": 0.6721,
      "step": 4275
    },
    {
      "epoch": 0.17721414066061586,
      "grad_norm": 0.46978724002838135,
      "learning_rate": 4.11413651622529e-06,
      "loss": 0.7283,
      "step": 4276
    },
    {
      "epoch": 0.17725558456628954,
      "grad_norm": 0.4823302924633026,
      "learning_rate": 4.113929296696921e-06,
      "loss": 0.7764,
      "step": 4277
    },
    {
      "epoch": 0.1772970284719632,
      "grad_norm": 0.4234614968299866,
      "learning_rate": 4.113722077168552e-06,
      "loss": 0.7303,
      "step": 4278
    },
    {
      "epoch": 0.17733847237763686,
      "grad_norm": 0.44675666093826294,
      "learning_rate": 4.113514857640185e-06,
      "loss": 0.7837,
      "step": 4279
    },
    {
      "epoch": 0.17737991628331054,
      "grad_norm": 0.47405174374580383,
      "learning_rate": 4.113307638111816e-06,
      "loss": 0.7217,
      "step": 4280
    },
    {
      "epoch": 0.17742136018898422,
      "grad_norm": 0.4084571599960327,
      "learning_rate": 4.113100418583447e-06,
      "loss": 0.7075,
      "step": 4281
    },
    {
      "epoch": 0.1774628040946579,
      "grad_norm": 0.4558860957622528,
      "learning_rate": 4.112893199055079e-06,
      "loss": 0.7074,
      "step": 4282
    },
    {
      "epoch": 0.17750424800033154,
      "grad_norm": 0.42903947830200195,
      "learning_rate": 4.112685979526711e-06,
      "loss": 0.7527,
      "step": 4283
    },
    {
      "epoch": 0.17754569190600522,
      "grad_norm": 0.40029436349868774,
      "learning_rate": 4.112478759998342e-06,
      "loss": 0.6987,
      "step": 4284
    },
    {
      "epoch": 0.1775871358116789,
      "grad_norm": 0.42030277848243713,
      "learning_rate": 4.112271540469975e-06,
      "loss": 0.6965,
      "step": 4285
    },
    {
      "epoch": 0.17762857971735257,
      "grad_norm": 0.40933334827423096,
      "learning_rate": 4.112064320941606e-06,
      "loss": 0.7434,
      "step": 4286
    },
    {
      "epoch": 0.17767002362302622,
      "grad_norm": 0.4166834056377411,
      "learning_rate": 4.111857101413237e-06,
      "loss": 0.7035,
      "step": 4287
    },
    {
      "epoch": 0.1777114675286999,
      "grad_norm": 0.43815746903419495,
      "learning_rate": 4.11164988188487e-06,
      "loss": 0.731,
      "step": 4288
    },
    {
      "epoch": 0.17775291143437358,
      "grad_norm": 0.4044334292411804,
      "learning_rate": 4.111442662356501e-06,
      "loss": 0.7351,
      "step": 4289
    },
    {
      "epoch": 0.17779435534004726,
      "grad_norm": 0.4471001923084259,
      "learning_rate": 4.111235442828132e-06,
      "loss": 0.7148,
      "step": 4290
    },
    {
      "epoch": 0.1778357992457209,
      "grad_norm": 0.4518488347530365,
      "learning_rate": 4.111028223299764e-06,
      "loss": 0.7686,
      "step": 4291
    },
    {
      "epoch": 0.17787724315139458,
      "grad_norm": 0.4025476276874542,
      "learning_rate": 4.110821003771396e-06,
      "loss": 0.6982,
      "step": 4292
    },
    {
      "epoch": 0.17791868705706826,
      "grad_norm": 0.4361026883125305,
      "learning_rate": 4.110613784243027e-06,
      "loss": 0.8044,
      "step": 4293
    },
    {
      "epoch": 0.17796013096274194,
      "grad_norm": 0.4315468966960907,
      "learning_rate": 4.110406564714659e-06,
      "loss": 0.762,
      "step": 4294
    },
    {
      "epoch": 0.1780015748684156,
      "grad_norm": 0.4462752342224121,
      "learning_rate": 4.110199345186291e-06,
      "loss": 0.7991,
      "step": 4295
    },
    {
      "epoch": 0.17804301877408926,
      "grad_norm": 0.41034141182899475,
      "learning_rate": 4.109992125657922e-06,
      "loss": 0.7325,
      "step": 4296
    },
    {
      "epoch": 0.17808446267976294,
      "grad_norm": 0.4040284752845764,
      "learning_rate": 4.109784906129554e-06,
      "loss": 0.679,
      "step": 4297
    },
    {
      "epoch": 0.17812590658543662,
      "grad_norm": 0.46674495935440063,
      "learning_rate": 4.109577686601185e-06,
      "loss": 0.7847,
      "step": 4298
    },
    {
      "epoch": 0.1781673504911103,
      "grad_norm": 0.40785902738571167,
      "learning_rate": 4.109370467072817e-06,
      "loss": 0.7009,
      "step": 4299
    },
    {
      "epoch": 0.17820879439678394,
      "grad_norm": 0.4218190908432007,
      "learning_rate": 4.109163247544449e-06,
      "loss": 0.7023,
      "step": 4300
    },
    {
      "epoch": 0.17825023830245762,
      "grad_norm": 0.4493870437145233,
      "learning_rate": 4.108956028016081e-06,
      "loss": 0.6936,
      "step": 4301
    },
    {
      "epoch": 0.1782916822081313,
      "grad_norm": 0.5042098760604858,
      "learning_rate": 4.108748808487712e-06,
      "loss": 0.7284,
      "step": 4302
    },
    {
      "epoch": 0.17833312611380497,
      "grad_norm": 0.416984498500824,
      "learning_rate": 4.108541588959344e-06,
      "loss": 0.7144,
      "step": 4303
    },
    {
      "epoch": 0.17837457001947862,
      "grad_norm": 0.4273073077201843,
      "learning_rate": 4.108334369430976e-06,
      "loss": 0.7437,
      "step": 4304
    },
    {
      "epoch": 0.1784160139251523,
      "grad_norm": 0.41094937920570374,
      "learning_rate": 4.1081271499026074e-06,
      "loss": 0.7384,
      "step": 4305
    },
    {
      "epoch": 0.17845745783082598,
      "grad_norm": 0.5612376928329468,
      "learning_rate": 4.107919930374239e-06,
      "loss": 0.8157,
      "step": 4306
    },
    {
      "epoch": 0.17849890173649965,
      "grad_norm": 0.46674543619155884,
      "learning_rate": 4.10771271084587e-06,
      "loss": 0.7106,
      "step": 4307
    },
    {
      "epoch": 0.17854034564217333,
      "grad_norm": 0.4352622330188751,
      "learning_rate": 4.1075054913175024e-06,
      "loss": 0.7216,
      "step": 4308
    },
    {
      "epoch": 0.17858178954784698,
      "grad_norm": 0.4645962715148926,
      "learning_rate": 4.107298271789134e-06,
      "loss": 0.7854,
      "step": 4309
    },
    {
      "epoch": 0.17862323345352066,
      "grad_norm": 0.4313540756702423,
      "learning_rate": 4.107091052260765e-06,
      "loss": 0.7395,
      "step": 4310
    },
    {
      "epoch": 0.17866467735919433,
      "grad_norm": 0.42626938223838806,
      "learning_rate": 4.106883832732397e-06,
      "loss": 0.7283,
      "step": 4311
    },
    {
      "epoch": 0.178706121264868,
      "grad_norm": 0.4383326768875122,
      "learning_rate": 4.106676613204029e-06,
      "loss": 0.7449,
      "step": 4312
    },
    {
      "epoch": 0.17874756517054166,
      "grad_norm": 0.39216348528862,
      "learning_rate": 4.10646939367566e-06,
      "loss": 0.6851,
      "step": 4313
    },
    {
      "epoch": 0.17878900907621534,
      "grad_norm": 0.42828142642974854,
      "learning_rate": 4.106262174147292e-06,
      "loss": 0.7074,
      "step": 4314
    },
    {
      "epoch": 0.17883045298188902,
      "grad_norm": 0.42427563667297363,
      "learning_rate": 4.106054954618924e-06,
      "loss": 0.6914,
      "step": 4315
    },
    {
      "epoch": 0.1788718968875627,
      "grad_norm": 0.42758116126060486,
      "learning_rate": 4.105847735090555e-06,
      "loss": 0.752,
      "step": 4316
    },
    {
      "epoch": 0.17891334079323634,
      "grad_norm": 0.41149428486824036,
      "learning_rate": 4.105640515562187e-06,
      "loss": 0.7261,
      "step": 4317
    },
    {
      "epoch": 0.17895478469891002,
      "grad_norm": 0.4862017035484314,
      "learning_rate": 4.105433296033818e-06,
      "loss": 0.7585,
      "step": 4318
    },
    {
      "epoch": 0.1789962286045837,
      "grad_norm": 0.4164459705352783,
      "learning_rate": 4.10522607650545e-06,
      "loss": 0.7014,
      "step": 4319
    },
    {
      "epoch": 0.17903767251025737,
      "grad_norm": 0.4301208257675171,
      "learning_rate": 4.105018856977082e-06,
      "loss": 0.749,
      "step": 4320
    },
    {
      "epoch": 0.17907911641593105,
      "grad_norm": 0.44435566663742065,
      "learning_rate": 4.104811637448714e-06,
      "loss": 0.7686,
      "step": 4321
    },
    {
      "epoch": 0.1791205603216047,
      "grad_norm": 0.4533887207508087,
      "learning_rate": 4.104604417920345e-06,
      "loss": 0.752,
      "step": 4322
    },
    {
      "epoch": 0.17916200422727838,
      "grad_norm": 0.4279111623764038,
      "learning_rate": 4.104397198391977e-06,
      "loss": 0.7498,
      "step": 4323
    },
    {
      "epoch": 0.17920344813295205,
      "grad_norm": 0.43398571014404297,
      "learning_rate": 4.104189978863609e-06,
      "loss": 0.7939,
      "step": 4324
    },
    {
      "epoch": 0.17924489203862573,
      "grad_norm": 0.4085802733898163,
      "learning_rate": 4.10398275933524e-06,
      "loss": 0.718,
      "step": 4325
    },
    {
      "epoch": 0.17928633594429938,
      "grad_norm": 0.4387429356575012,
      "learning_rate": 4.103775539806872e-06,
      "loss": 0.7324,
      "step": 4326
    },
    {
      "epoch": 0.17932777984997306,
      "grad_norm": 0.5018410086631775,
      "learning_rate": 4.103568320278503e-06,
      "loss": 0.7593,
      "step": 4327
    },
    {
      "epoch": 0.17936922375564673,
      "grad_norm": 0.4165845513343811,
      "learning_rate": 4.103361100750135e-06,
      "loss": 0.7273,
      "step": 4328
    },
    {
      "epoch": 0.1794106676613204,
      "grad_norm": 0.4184613525867462,
      "learning_rate": 4.103153881221767e-06,
      "loss": 0.7676,
      "step": 4329
    },
    {
      "epoch": 0.17945211156699406,
      "grad_norm": 0.4521818161010742,
      "learning_rate": 4.102946661693398e-06,
      "loss": 0.7544,
      "step": 4330
    },
    {
      "epoch": 0.17949355547266774,
      "grad_norm": 0.41748806834220886,
      "learning_rate": 4.10273944216503e-06,
      "loss": 0.7218,
      "step": 4331
    },
    {
      "epoch": 0.17953499937834141,
      "grad_norm": 0.4290112853050232,
      "learning_rate": 4.102532222636662e-06,
      "loss": 0.701,
      "step": 4332
    },
    {
      "epoch": 0.1795764432840151,
      "grad_norm": 0.43585166335105896,
      "learning_rate": 4.102325003108293e-06,
      "loss": 0.6853,
      "step": 4333
    },
    {
      "epoch": 0.17961788718968877,
      "grad_norm": 0.41494202613830566,
      "learning_rate": 4.102117783579924e-06,
      "loss": 0.6816,
      "step": 4334
    },
    {
      "epoch": 0.17965933109536242,
      "grad_norm": 0.4312579333782196,
      "learning_rate": 4.101910564051557e-06,
      "loss": 0.7366,
      "step": 4335
    },
    {
      "epoch": 0.1797007750010361,
      "grad_norm": 0.3995453715324402,
      "learning_rate": 4.101703344523188e-06,
      "loss": 0.6433,
      "step": 4336
    },
    {
      "epoch": 0.17974221890670977,
      "grad_norm": 0.5056896805763245,
      "learning_rate": 4.10149612499482e-06,
      "loss": 0.7495,
      "step": 4337
    },
    {
      "epoch": 0.17978366281238345,
      "grad_norm": 0.4615245461463928,
      "learning_rate": 4.101288905466452e-06,
      "loss": 0.7788,
      "step": 4338
    },
    {
      "epoch": 0.1798251067180571,
      "grad_norm": 0.46029555797576904,
      "learning_rate": 4.101081685938083e-06,
      "loss": 0.7499,
      "step": 4339
    },
    {
      "epoch": 0.17986655062373078,
      "grad_norm": 0.4137091040611267,
      "learning_rate": 4.100874466409715e-06,
      "loss": 0.7512,
      "step": 4340
    },
    {
      "epoch": 0.17990799452940445,
      "grad_norm": 0.41946813464164734,
      "learning_rate": 4.100667246881347e-06,
      "loss": 0.7297,
      "step": 4341
    },
    {
      "epoch": 0.17994943843507813,
      "grad_norm": 0.4331010580062866,
      "learning_rate": 4.100460027352978e-06,
      "loss": 0.7334,
      "step": 4342
    },
    {
      "epoch": 0.17999088234075178,
      "grad_norm": 0.435464471578598,
      "learning_rate": 4.100252807824609e-06,
      "loss": 0.7371,
      "step": 4343
    },
    {
      "epoch": 0.18003232624642546,
      "grad_norm": 0.44494694471359253,
      "learning_rate": 4.100045588296242e-06,
      "loss": 0.7443,
      "step": 4344
    },
    {
      "epoch": 0.18007377015209913,
      "grad_norm": 0.42720088362693787,
      "learning_rate": 4.099838368767873e-06,
      "loss": 0.7146,
      "step": 4345
    },
    {
      "epoch": 0.1801152140577728,
      "grad_norm": 0.45642855763435364,
      "learning_rate": 4.099631149239504e-06,
      "loss": 0.729,
      "step": 4346
    },
    {
      "epoch": 0.1801566579634465,
      "grad_norm": 0.44834575057029724,
      "learning_rate": 4.099423929711137e-06,
      "loss": 0.7793,
      "step": 4347
    },
    {
      "epoch": 0.18019810186912014,
      "grad_norm": 0.45838481187820435,
      "learning_rate": 4.099216710182768e-06,
      "loss": 0.7378,
      "step": 4348
    },
    {
      "epoch": 0.1802395457747938,
      "grad_norm": 0.40973523259162903,
      "learning_rate": 4.099009490654399e-06,
      "loss": 0.7196,
      "step": 4349
    },
    {
      "epoch": 0.1802809896804675,
      "grad_norm": 0.4404100775718689,
      "learning_rate": 4.098802271126031e-06,
      "loss": 0.7258,
      "step": 4350
    },
    {
      "epoch": 0.18032243358614117,
      "grad_norm": 0.4101592004299164,
      "learning_rate": 4.098595051597663e-06,
      "loss": 0.7319,
      "step": 4351
    },
    {
      "epoch": 0.18036387749181482,
      "grad_norm": 0.4168979525566101,
      "learning_rate": 4.098387832069294e-06,
      "loss": 0.7074,
      "step": 4352
    },
    {
      "epoch": 0.1804053213974885,
      "grad_norm": 0.41221973299980164,
      "learning_rate": 4.098180612540927e-06,
      "loss": 0.7434,
      "step": 4353
    },
    {
      "epoch": 0.18044676530316217,
      "grad_norm": 0.41986793279647827,
      "learning_rate": 4.097973393012557e-06,
      "loss": 0.7124,
      "step": 4354
    },
    {
      "epoch": 0.18048820920883585,
      "grad_norm": 0.476686954498291,
      "learning_rate": 4.097766173484189e-06,
      "loss": 0.7437,
      "step": 4355
    },
    {
      "epoch": 0.18052965311450953,
      "grad_norm": 0.4245510399341583,
      "learning_rate": 4.097558953955822e-06,
      "loss": 0.739,
      "step": 4356
    },
    {
      "epoch": 0.18057109702018317,
      "grad_norm": 0.3986087739467621,
      "learning_rate": 4.097351734427453e-06,
      "loss": 0.7214,
      "step": 4357
    },
    {
      "epoch": 0.18061254092585685,
      "grad_norm": 0.40530362725257874,
      "learning_rate": 4.097144514899084e-06,
      "loss": 0.6965,
      "step": 4358
    },
    {
      "epoch": 0.18065398483153053,
      "grad_norm": 0.425922691822052,
      "learning_rate": 4.096937295370716e-06,
      "loss": 0.7234,
      "step": 4359
    },
    {
      "epoch": 0.1806954287372042,
      "grad_norm": 0.4088808000087738,
      "learning_rate": 4.096730075842348e-06,
      "loss": 0.7268,
      "step": 4360
    },
    {
      "epoch": 0.18073687264287785,
      "grad_norm": 0.4442114233970642,
      "learning_rate": 4.0965228563139794e-06,
      "loss": 0.7271,
      "step": 4361
    },
    {
      "epoch": 0.18077831654855153,
      "grad_norm": 0.4285662770271301,
      "learning_rate": 4.096315636785611e-06,
      "loss": 0.751,
      "step": 4362
    },
    {
      "epoch": 0.1808197604542252,
      "grad_norm": 0.39213985204696655,
      "learning_rate": 4.096108417257242e-06,
      "loss": 0.7224,
      "step": 4363
    },
    {
      "epoch": 0.18086120435989889,
      "grad_norm": 0.4457820653915405,
      "learning_rate": 4.0959011977288744e-06,
      "loss": 0.7184,
      "step": 4364
    },
    {
      "epoch": 0.18090264826557254,
      "grad_norm": 0.4317987263202667,
      "learning_rate": 4.095693978200506e-06,
      "loss": 0.6943,
      "step": 4365
    },
    {
      "epoch": 0.1809440921712462,
      "grad_norm": 0.45262524485588074,
      "learning_rate": 4.095486758672137e-06,
      "loss": 0.7598,
      "step": 4366
    },
    {
      "epoch": 0.1809855360769199,
      "grad_norm": 0.4363974928855896,
      "learning_rate": 4.0952795391437694e-06,
      "loss": 0.7041,
      "step": 4367
    },
    {
      "epoch": 0.18102697998259357,
      "grad_norm": 0.40521758794784546,
      "learning_rate": 4.095072319615401e-06,
      "loss": 0.7156,
      "step": 4368
    },
    {
      "epoch": 0.18106842388826724,
      "grad_norm": 0.4111277163028717,
      "learning_rate": 4.094865100087033e-06,
      "loss": 0.7568,
      "step": 4369
    },
    {
      "epoch": 0.1811098677939409,
      "grad_norm": 0.4023473560810089,
      "learning_rate": 4.094657880558664e-06,
      "loss": 0.6863,
      "step": 4370
    },
    {
      "epoch": 0.18115131169961457,
      "grad_norm": 0.4092337191104889,
      "learning_rate": 4.094450661030296e-06,
      "loss": 0.6956,
      "step": 4371
    },
    {
      "epoch": 0.18119275560528825,
      "grad_norm": 0.40245145559310913,
      "learning_rate": 4.094243441501927e-06,
      "loss": 0.6514,
      "step": 4372
    },
    {
      "epoch": 0.18123419951096192,
      "grad_norm": 0.4169035255908966,
      "learning_rate": 4.0940362219735594e-06,
      "loss": 0.7634,
      "step": 4373
    },
    {
      "epoch": 0.18127564341663557,
      "grad_norm": 0.42160505056381226,
      "learning_rate": 4.093829002445191e-06,
      "loss": 0.6617,
      "step": 4374
    },
    {
      "epoch": 0.18131708732230925,
      "grad_norm": 0.4221780300140381,
      "learning_rate": 4.093621782916822e-06,
      "loss": 0.6989,
      "step": 4375
    },
    {
      "epoch": 0.18135853122798293,
      "grad_norm": 0.4273332357406616,
      "learning_rate": 4.0934145633884544e-06,
      "loss": 0.7268,
      "step": 4376
    },
    {
      "epoch": 0.1813999751336566,
      "grad_norm": 0.42516419291496277,
      "learning_rate": 4.093207343860086e-06,
      "loss": 0.7375,
      "step": 4377
    },
    {
      "epoch": 0.18144141903933025,
      "grad_norm": 0.4265301823616028,
      "learning_rate": 4.093000124331717e-06,
      "loss": 0.7668,
      "step": 4378
    },
    {
      "epoch": 0.18148286294500393,
      "grad_norm": 0.4175337851047516,
      "learning_rate": 4.092792904803349e-06,
      "loss": 0.7224,
      "step": 4379
    },
    {
      "epoch": 0.1815243068506776,
      "grad_norm": 0.4374377727508545,
      "learning_rate": 4.092585685274981e-06,
      "loss": 0.7087,
      "step": 4380
    },
    {
      "epoch": 0.18156575075635129,
      "grad_norm": 0.4657685160636902,
      "learning_rate": 4.092378465746612e-06,
      "loss": 0.7585,
      "step": 4381
    },
    {
      "epoch": 0.18160719466202496,
      "grad_norm": 0.42439979314804077,
      "learning_rate": 4.092171246218244e-06,
      "loss": 0.7021,
      "step": 4382
    },
    {
      "epoch": 0.1816486385676986,
      "grad_norm": 0.4083669185638428,
      "learning_rate": 4.091964026689876e-06,
      "loss": 0.7361,
      "step": 4383
    },
    {
      "epoch": 0.1816900824733723,
      "grad_norm": 0.425096720457077,
      "learning_rate": 4.091756807161507e-06,
      "loss": 0.703,
      "step": 4384
    },
    {
      "epoch": 0.18173152637904597,
      "grad_norm": 0.43057724833488464,
      "learning_rate": 4.091549587633139e-06,
      "loss": 0.6553,
      "step": 4385
    },
    {
      "epoch": 0.18177297028471964,
      "grad_norm": 0.4205147624015808,
      "learning_rate": 4.09134236810477e-06,
      "loss": 0.6995,
      "step": 4386
    },
    {
      "epoch": 0.1818144141903933,
      "grad_norm": 0.43575456738471985,
      "learning_rate": 4.091135148576402e-06,
      "loss": 0.7371,
      "step": 4387
    },
    {
      "epoch": 0.18185585809606697,
      "grad_norm": 0.41343721747398376,
      "learning_rate": 4.090927929048034e-06,
      "loss": 0.6698,
      "step": 4388
    },
    {
      "epoch": 0.18189730200174065,
      "grad_norm": 0.45800450444221497,
      "learning_rate": 4.090720709519666e-06,
      "loss": 0.7058,
      "step": 4389
    },
    {
      "epoch": 0.18193874590741432,
      "grad_norm": 0.4350242614746094,
      "learning_rate": 4.090513489991297e-06,
      "loss": 0.7446,
      "step": 4390
    },
    {
      "epoch": 0.18198018981308797,
      "grad_norm": 0.4259223937988281,
      "learning_rate": 4.090306270462929e-06,
      "loss": 0.7288,
      "step": 4391
    },
    {
      "epoch": 0.18202163371876165,
      "grad_norm": 0.40063032507896423,
      "learning_rate": 4.090099050934561e-06,
      "loss": 0.7249,
      "step": 4392
    },
    {
      "epoch": 0.18206307762443533,
      "grad_norm": 0.40788915753364563,
      "learning_rate": 4.089891831406192e-06,
      "loss": 0.7175,
      "step": 4393
    },
    {
      "epoch": 0.182104521530109,
      "grad_norm": 0.43516749143600464,
      "learning_rate": 4.089684611877824e-06,
      "loss": 0.729,
      "step": 4394
    },
    {
      "epoch": 0.18214596543578268,
      "grad_norm": 0.42761242389678955,
      "learning_rate": 4.089477392349455e-06,
      "loss": 0.7712,
      "step": 4395
    },
    {
      "epoch": 0.18218740934145633,
      "grad_norm": 0.4414919912815094,
      "learning_rate": 4.089270172821087e-06,
      "loss": 0.7437,
      "step": 4396
    },
    {
      "epoch": 0.18222885324713,
      "grad_norm": 0.4128112196922302,
      "learning_rate": 4.089062953292719e-06,
      "loss": 0.7532,
      "step": 4397
    },
    {
      "epoch": 0.18227029715280368,
      "grad_norm": 0.48242437839508057,
      "learning_rate": 4.08885573376435e-06,
      "loss": 0.7078,
      "step": 4398
    },
    {
      "epoch": 0.18231174105847736,
      "grad_norm": 0.5164169073104858,
      "learning_rate": 4.088648514235982e-06,
      "loss": 0.811,
      "step": 4399
    },
    {
      "epoch": 0.182353184964151,
      "grad_norm": 0.4361169934272766,
      "learning_rate": 4.088441294707614e-06,
      "loss": 0.7205,
      "step": 4400
    },
    {
      "epoch": 0.1823946288698247,
      "grad_norm": 0.41624781489372253,
      "learning_rate": 4.088234075179245e-06,
      "loss": 0.6914,
      "step": 4401
    },
    {
      "epoch": 0.18243607277549836,
      "grad_norm": 0.4548285901546478,
      "learning_rate": 4.088026855650876e-06,
      "loss": 0.7029,
      "step": 4402
    },
    {
      "epoch": 0.18247751668117204,
      "grad_norm": 0.4837510585784912,
      "learning_rate": 4.087819636122509e-06,
      "loss": 0.7388,
      "step": 4403
    },
    {
      "epoch": 0.1825189605868457,
      "grad_norm": 0.43088245391845703,
      "learning_rate": 4.08761241659414e-06,
      "loss": 0.7451,
      "step": 4404
    },
    {
      "epoch": 0.18256040449251937,
      "grad_norm": 0.40221038460731506,
      "learning_rate": 4.087405197065772e-06,
      "loss": 0.6858,
      "step": 4405
    },
    {
      "epoch": 0.18260184839819305,
      "grad_norm": 0.38958480954170227,
      "learning_rate": 4.087197977537404e-06,
      "loss": 0.7008,
      "step": 4406
    },
    {
      "epoch": 0.18264329230386672,
      "grad_norm": 0.43535566329956055,
      "learning_rate": 4.086990758009035e-06,
      "loss": 0.7241,
      "step": 4407
    },
    {
      "epoch": 0.1826847362095404,
      "grad_norm": 0.4632434546947479,
      "learning_rate": 4.086783538480667e-06,
      "loss": 0.75,
      "step": 4408
    },
    {
      "epoch": 0.18272618011521405,
      "grad_norm": 0.390055775642395,
      "learning_rate": 4.086576318952299e-06,
      "loss": 0.6946,
      "step": 4409
    },
    {
      "epoch": 0.18276762402088773,
      "grad_norm": 0.38322508335113525,
      "learning_rate": 4.08636909942393e-06,
      "loss": 0.715,
      "step": 4410
    },
    {
      "epoch": 0.1828090679265614,
      "grad_norm": 0.41152116656303406,
      "learning_rate": 4.086161879895561e-06,
      "loss": 0.6768,
      "step": 4411
    },
    {
      "epoch": 0.18285051183223508,
      "grad_norm": 0.4003947973251343,
      "learning_rate": 4.085954660367194e-06,
      "loss": 0.7375,
      "step": 4412
    },
    {
      "epoch": 0.18289195573790873,
      "grad_norm": 0.4073386490345001,
      "learning_rate": 4.085747440838825e-06,
      "loss": 0.697,
      "step": 4413
    },
    {
      "epoch": 0.1829333996435824,
      "grad_norm": 0.43427029252052307,
      "learning_rate": 4.085540221310456e-06,
      "loss": 0.7385,
      "step": 4414
    },
    {
      "epoch": 0.18297484354925608,
      "grad_norm": 0.4059550166130066,
      "learning_rate": 4.085333001782088e-06,
      "loss": 0.7085,
      "step": 4415
    },
    {
      "epoch": 0.18301628745492976,
      "grad_norm": 0.40467342734336853,
      "learning_rate": 4.08512578225372e-06,
      "loss": 0.7189,
      "step": 4416
    },
    {
      "epoch": 0.1830577313606034,
      "grad_norm": 0.48845353722572327,
      "learning_rate": 4.084918562725351e-06,
      "loss": 0.7295,
      "step": 4417
    },
    {
      "epoch": 0.1830991752662771,
      "grad_norm": 0.46005767583847046,
      "learning_rate": 4.084711343196983e-06,
      "loss": 0.7432,
      "step": 4418
    },
    {
      "epoch": 0.18314061917195076,
      "grad_norm": 0.41732802987098694,
      "learning_rate": 4.084504123668615e-06,
      "loss": 0.7646,
      "step": 4419
    },
    {
      "epoch": 0.18318206307762444,
      "grad_norm": 0.45906782150268555,
      "learning_rate": 4.0842969041402464e-06,
      "loss": 0.7654,
      "step": 4420
    },
    {
      "epoch": 0.18322350698329812,
      "grad_norm": 0.42317450046539307,
      "learning_rate": 4.084089684611879e-06,
      "loss": 0.7029,
      "step": 4421
    },
    {
      "epoch": 0.18326495088897177,
      "grad_norm": 0.4046235978603363,
      "learning_rate": 4.083882465083509e-06,
      "loss": 0.6926,
      "step": 4422
    },
    {
      "epoch": 0.18330639479464544,
      "grad_norm": 0.4445026218891144,
      "learning_rate": 4.0836752455551414e-06,
      "loss": 0.7227,
      "step": 4423
    },
    {
      "epoch": 0.18334783870031912,
      "grad_norm": 0.4224022924900055,
      "learning_rate": 4.083468026026773e-06,
      "loss": 0.6873,
      "step": 4424
    },
    {
      "epoch": 0.1833892826059928,
      "grad_norm": 0.4156215488910675,
      "learning_rate": 4.083260806498405e-06,
      "loss": 0.7388,
      "step": 4425
    },
    {
      "epoch": 0.18343072651166645,
      "grad_norm": 0.400493860244751,
      "learning_rate": 4.0830535869700364e-06,
      "loss": 0.7314,
      "step": 4426
    },
    {
      "epoch": 0.18347217041734012,
      "grad_norm": 0.4409414231777191,
      "learning_rate": 4.082846367441668e-06,
      "loss": 0.7916,
      "step": 4427
    },
    {
      "epoch": 0.1835136143230138,
      "grad_norm": 0.4662216305732727,
      "learning_rate": 4.0826391479133e-06,
      "loss": 0.7144,
      "step": 4428
    },
    {
      "epoch": 0.18355505822868748,
      "grad_norm": 0.40235865116119385,
      "learning_rate": 4.0824319283849314e-06,
      "loss": 0.6978,
      "step": 4429
    },
    {
      "epoch": 0.18359650213436116,
      "grad_norm": 0.4039965569972992,
      "learning_rate": 4.082224708856563e-06,
      "loss": 0.6951,
      "step": 4430
    },
    {
      "epoch": 0.1836379460400348,
      "grad_norm": 0.416631817817688,
      "learning_rate": 4.082017489328194e-06,
      "loss": 0.6821,
      "step": 4431
    },
    {
      "epoch": 0.18367938994570848,
      "grad_norm": 0.43682849407196045,
      "learning_rate": 4.0818102697998264e-06,
      "loss": 0.7312,
      "step": 4432
    },
    {
      "epoch": 0.18372083385138216,
      "grad_norm": 0.4431561827659607,
      "learning_rate": 4.081603050271458e-06,
      "loss": 0.7601,
      "step": 4433
    },
    {
      "epoch": 0.18376227775705584,
      "grad_norm": 0.42798924446105957,
      "learning_rate": 4.081395830743089e-06,
      "loss": 0.719,
      "step": 4434
    },
    {
      "epoch": 0.18380372166272949,
      "grad_norm": 0.48576799035072327,
      "learning_rate": 4.0811886112147214e-06,
      "loss": 0.8704,
      "step": 4435
    },
    {
      "epoch": 0.18384516556840316,
      "grad_norm": 0.4031914174556732,
      "learning_rate": 4.080981391686353e-06,
      "loss": 0.7087,
      "step": 4436
    },
    {
      "epoch": 0.18388660947407684,
      "grad_norm": 0.46489468216896057,
      "learning_rate": 4.080774172157985e-06,
      "loss": 0.7272,
      "step": 4437
    },
    {
      "epoch": 0.18392805337975052,
      "grad_norm": 0.39807406067848206,
      "learning_rate": 4.080566952629616e-06,
      "loss": 0.7043,
      "step": 4438
    },
    {
      "epoch": 0.18396949728542417,
      "grad_norm": 0.44383782148361206,
      "learning_rate": 4.080359733101248e-06,
      "loss": 0.6855,
      "step": 4439
    },
    {
      "epoch": 0.18401094119109784,
      "grad_norm": 0.4440809190273285,
      "learning_rate": 4.080152513572879e-06,
      "loss": 0.8196,
      "step": 4440
    },
    {
      "epoch": 0.18405238509677152,
      "grad_norm": 0.4540383815765381,
      "learning_rate": 4.0799452940445115e-06,
      "loss": 0.739,
      "step": 4441
    },
    {
      "epoch": 0.1840938290024452,
      "grad_norm": 0.44314637780189514,
      "learning_rate": 4.079738074516143e-06,
      "loss": 0.7979,
      "step": 4442
    },
    {
      "epoch": 0.18413527290811887,
      "grad_norm": 0.434135764837265,
      "learning_rate": 4.079530854987774e-06,
      "loss": 0.7267,
      "step": 4443
    },
    {
      "epoch": 0.18417671681379252,
      "grad_norm": 0.4147430956363678,
      "learning_rate": 4.0793236354594065e-06,
      "loss": 0.7275,
      "step": 4444
    },
    {
      "epoch": 0.1842181607194662,
      "grad_norm": 0.47405385971069336,
      "learning_rate": 4.079116415931038e-06,
      "loss": 0.7388,
      "step": 4445
    },
    {
      "epoch": 0.18425960462513988,
      "grad_norm": 0.4174647927284241,
      "learning_rate": 4.078909196402669e-06,
      "loss": 0.7234,
      "step": 4446
    },
    {
      "epoch": 0.18430104853081355,
      "grad_norm": 0.4337713420391083,
      "learning_rate": 4.078701976874301e-06,
      "loss": 0.7205,
      "step": 4447
    },
    {
      "epoch": 0.1843424924364872,
      "grad_norm": 0.44968995451927185,
      "learning_rate": 4.078494757345933e-06,
      "loss": 0.7454,
      "step": 4448
    },
    {
      "epoch": 0.18438393634216088,
      "grad_norm": 0.47263431549072266,
      "learning_rate": 4.078287537817564e-06,
      "loss": 0.7214,
      "step": 4449
    },
    {
      "epoch": 0.18442538024783456,
      "grad_norm": 0.4364013373851776,
      "learning_rate": 4.078080318289196e-06,
      "loss": 0.7451,
      "step": 4450
    },
    {
      "epoch": 0.18446682415350824,
      "grad_norm": 0.42869269847869873,
      "learning_rate": 4.077873098760828e-06,
      "loss": 0.7695,
      "step": 4451
    },
    {
      "epoch": 0.18450826805918188,
      "grad_norm": 0.4293075203895569,
      "learning_rate": 4.077665879232459e-06,
      "loss": 0.7273,
      "step": 4452
    },
    {
      "epoch": 0.18454971196485556,
      "grad_norm": 0.4507058560848236,
      "learning_rate": 4.0774586597040915e-06,
      "loss": 0.7593,
      "step": 4453
    },
    {
      "epoch": 0.18459115587052924,
      "grad_norm": 0.4253017008304596,
      "learning_rate": 4.077251440175722e-06,
      "loss": 0.7192,
      "step": 4454
    },
    {
      "epoch": 0.18463259977620292,
      "grad_norm": 0.41147664189338684,
      "learning_rate": 4.077044220647354e-06,
      "loss": 0.7231,
      "step": 4455
    },
    {
      "epoch": 0.1846740436818766,
      "grad_norm": 0.4458121061325073,
      "learning_rate": 4.076837001118986e-06,
      "loss": 0.7437,
      "step": 4456
    },
    {
      "epoch": 0.18471548758755024,
      "grad_norm": 0.46288084983825684,
      "learning_rate": 4.076629781590618e-06,
      "loss": 0.7725,
      "step": 4457
    },
    {
      "epoch": 0.18475693149322392,
      "grad_norm": 0.39144647121429443,
      "learning_rate": 4.076422562062249e-06,
      "loss": 0.6869,
      "step": 4458
    },
    {
      "epoch": 0.1847983753988976,
      "grad_norm": 0.4227301776409149,
      "learning_rate": 4.076215342533881e-06,
      "loss": 0.7253,
      "step": 4459
    },
    {
      "epoch": 0.18483981930457127,
      "grad_norm": 0.45118051767349243,
      "learning_rate": 4.076008123005513e-06,
      "loss": 0.7749,
      "step": 4460
    },
    {
      "epoch": 0.18488126321024492,
      "grad_norm": 0.43180012702941895,
      "learning_rate": 4.075800903477144e-06,
      "loss": 0.7208,
      "step": 4461
    },
    {
      "epoch": 0.1849227071159186,
      "grad_norm": 0.4133701026439667,
      "learning_rate": 4.075593683948776e-06,
      "loss": 0.7098,
      "step": 4462
    },
    {
      "epoch": 0.18496415102159228,
      "grad_norm": 0.3951307535171509,
      "learning_rate": 4.075386464420407e-06,
      "loss": 0.6824,
      "step": 4463
    },
    {
      "epoch": 0.18500559492726595,
      "grad_norm": 0.4145299196243286,
      "learning_rate": 4.075179244892039e-06,
      "loss": 0.7207,
      "step": 4464
    },
    {
      "epoch": 0.1850470388329396,
      "grad_norm": 0.43935656547546387,
      "learning_rate": 4.074972025363671e-06,
      "loss": 0.7515,
      "step": 4465
    },
    {
      "epoch": 0.18508848273861328,
      "grad_norm": 0.3960407078266144,
      "learning_rate": 4.074764805835302e-06,
      "loss": 0.7119,
      "step": 4466
    },
    {
      "epoch": 0.18512992664428696,
      "grad_norm": 0.44292742013931274,
      "learning_rate": 4.074557586306933e-06,
      "loss": 0.7581,
      "step": 4467
    },
    {
      "epoch": 0.18517137054996063,
      "grad_norm": 0.4133780598640442,
      "learning_rate": 4.074350366778566e-06,
      "loss": 0.728,
      "step": 4468
    },
    {
      "epoch": 0.1852128144556343,
      "grad_norm": 0.4245140254497528,
      "learning_rate": 4.074143147250197e-06,
      "loss": 0.6985,
      "step": 4469
    },
    {
      "epoch": 0.18525425836130796,
      "grad_norm": 0.37724199891090393,
      "learning_rate": 4.073935927721828e-06,
      "loss": 0.7209,
      "step": 4470
    },
    {
      "epoch": 0.18529570226698164,
      "grad_norm": 0.42624831199645996,
      "learning_rate": 4.073728708193461e-06,
      "loss": 0.71,
      "step": 4471
    },
    {
      "epoch": 0.18533714617265531,
      "grad_norm": 0.43911442160606384,
      "learning_rate": 4.073521488665092e-06,
      "loss": 0.7061,
      "step": 4472
    },
    {
      "epoch": 0.185378590078329,
      "grad_norm": 0.4500071406364441,
      "learning_rate": 4.073314269136724e-06,
      "loss": 0.7402,
      "step": 4473
    },
    {
      "epoch": 0.18542003398400264,
      "grad_norm": 0.4119059443473816,
      "learning_rate": 4.073107049608356e-06,
      "loss": 0.7153,
      "step": 4474
    },
    {
      "epoch": 0.18546147788967632,
      "grad_norm": 0.3883725106716156,
      "learning_rate": 4.072899830079987e-06,
      "loss": 0.6755,
      "step": 4475
    },
    {
      "epoch": 0.18550292179535,
      "grad_norm": 0.44351139664649963,
      "learning_rate": 4.072692610551618e-06,
      "loss": 0.6979,
      "step": 4476
    },
    {
      "epoch": 0.18554436570102367,
      "grad_norm": 0.41512972116470337,
      "learning_rate": 4.072485391023251e-06,
      "loss": 0.7676,
      "step": 4477
    },
    {
      "epoch": 0.18558580960669732,
      "grad_norm": 0.4742549657821655,
      "learning_rate": 4.072278171494882e-06,
      "loss": 0.751,
      "step": 4478
    },
    {
      "epoch": 0.185627253512371,
      "grad_norm": 0.4293517768383026,
      "learning_rate": 4.0720709519665134e-06,
      "loss": 0.7659,
      "step": 4479
    },
    {
      "epoch": 0.18566869741804468,
      "grad_norm": 0.4198879599571228,
      "learning_rate": 4.071863732438146e-06,
      "loss": 0.7866,
      "step": 4480
    },
    {
      "epoch": 0.18571014132371835,
      "grad_norm": 0.4431789219379425,
      "learning_rate": 4.071656512909777e-06,
      "loss": 0.7429,
      "step": 4481
    },
    {
      "epoch": 0.18575158522939203,
      "grad_norm": 0.3953503966331482,
      "learning_rate": 4.0714492933814084e-06,
      "loss": 0.6848,
      "step": 4482
    },
    {
      "epoch": 0.18579302913506568,
      "grad_norm": 0.40179479122161865,
      "learning_rate": 4.07124207385304e-06,
      "loss": 0.7356,
      "step": 4483
    },
    {
      "epoch": 0.18583447304073936,
      "grad_norm": 0.4447646141052246,
      "learning_rate": 4.071034854324672e-06,
      "loss": 0.7722,
      "step": 4484
    },
    {
      "epoch": 0.18587591694641303,
      "grad_norm": 0.4205664396286011,
      "learning_rate": 4.0708276347963034e-06,
      "loss": 0.7019,
      "step": 4485
    },
    {
      "epoch": 0.1859173608520867,
      "grad_norm": 0.3871537148952484,
      "learning_rate": 4.070620415267935e-06,
      "loss": 0.6826,
      "step": 4486
    },
    {
      "epoch": 0.18595880475776036,
      "grad_norm": 0.42107969522476196,
      "learning_rate": 4.070413195739567e-06,
      "loss": 0.6982,
      "step": 4487
    },
    {
      "epoch": 0.18600024866343404,
      "grad_norm": 0.394418865442276,
      "learning_rate": 4.0702059762111984e-06,
      "loss": 0.6499,
      "step": 4488
    },
    {
      "epoch": 0.18604169256910771,
      "grad_norm": 0.4322035014629364,
      "learning_rate": 4.069998756682831e-06,
      "loss": 0.6816,
      "step": 4489
    },
    {
      "epoch": 0.1860831364747814,
      "grad_norm": 0.41145429015159607,
      "learning_rate": 4.069791537154461e-06,
      "loss": 0.7397,
      "step": 4490
    },
    {
      "epoch": 0.18612458038045504,
      "grad_norm": 0.45068588852882385,
      "learning_rate": 4.0695843176260934e-06,
      "loss": 0.7671,
      "step": 4491
    },
    {
      "epoch": 0.18616602428612872,
      "grad_norm": 0.4302988648414612,
      "learning_rate": 4.069377098097725e-06,
      "loss": 0.7092,
      "step": 4492
    },
    {
      "epoch": 0.1862074681918024,
      "grad_norm": 0.43004539608955383,
      "learning_rate": 4.069169878569357e-06,
      "loss": 0.7545,
      "step": 4493
    },
    {
      "epoch": 0.18624891209747607,
      "grad_norm": 0.41236555576324463,
      "learning_rate": 4.0689626590409884e-06,
      "loss": 0.7395,
      "step": 4494
    },
    {
      "epoch": 0.18629035600314975,
      "grad_norm": 0.4554157555103302,
      "learning_rate": 4.06875543951262e-06,
      "loss": 0.7711,
      "step": 4495
    },
    {
      "epoch": 0.1863317999088234,
      "grad_norm": 0.39537763595581055,
      "learning_rate": 4.068548219984252e-06,
      "loss": 0.7258,
      "step": 4496
    },
    {
      "epoch": 0.18637324381449707,
      "grad_norm": 0.4299403131008148,
      "learning_rate": 4.0683410004558834e-06,
      "loss": 0.7246,
      "step": 4497
    },
    {
      "epoch": 0.18641468772017075,
      "grad_norm": 0.4268067181110382,
      "learning_rate": 4.068133780927515e-06,
      "loss": 0.7306,
      "step": 4498
    },
    {
      "epoch": 0.18645613162584443,
      "grad_norm": 0.3723105788230896,
      "learning_rate": 4.067926561399146e-06,
      "loss": 0.6775,
      "step": 4499
    },
    {
      "epoch": 0.18649757553151808,
      "grad_norm": 0.4244597256183624,
      "learning_rate": 4.0677193418707785e-06,
      "loss": 0.7826,
      "step": 4500
    },
    {
      "epoch": 0.18653901943719176,
      "grad_norm": 0.38987529277801514,
      "learning_rate": 4.06751212234241e-06,
      "loss": 0.7034,
      "step": 4501
    },
    {
      "epoch": 0.18658046334286543,
      "grad_norm": 0.4161781668663025,
      "learning_rate": 4.067304902814041e-06,
      "loss": 0.7283,
      "step": 4502
    },
    {
      "epoch": 0.1866219072485391,
      "grad_norm": 0.39756909012794495,
      "learning_rate": 4.0670976832856735e-06,
      "loss": 0.7031,
      "step": 4503
    },
    {
      "epoch": 0.1866633511542128,
      "grad_norm": 0.43187960982322693,
      "learning_rate": 4.066890463757305e-06,
      "loss": 0.7335,
      "step": 4504
    },
    {
      "epoch": 0.18670479505988644,
      "grad_norm": 0.44504618644714355,
      "learning_rate": 4.066683244228937e-06,
      "loss": 0.7786,
      "step": 4505
    },
    {
      "epoch": 0.1867462389655601,
      "grad_norm": 0.3973124325275421,
      "learning_rate": 4.066476024700568e-06,
      "loss": 0.743,
      "step": 4506
    },
    {
      "epoch": 0.1867876828712338,
      "grad_norm": 0.40122586488723755,
      "learning_rate": 4.0662688051722e-06,
      "loss": 0.7004,
      "step": 4507
    },
    {
      "epoch": 0.18682912677690747,
      "grad_norm": 0.4111984968185425,
      "learning_rate": 4.066061585643831e-06,
      "loss": 0.7212,
      "step": 4508
    },
    {
      "epoch": 0.18687057068258112,
      "grad_norm": 0.3946007490158081,
      "learning_rate": 4.0658543661154635e-06,
      "loss": 0.6963,
      "step": 4509
    },
    {
      "epoch": 0.1869120145882548,
      "grad_norm": 0.4154515862464905,
      "learning_rate": 4.065647146587095e-06,
      "loss": 0.7365,
      "step": 4510
    },
    {
      "epoch": 0.18695345849392847,
      "grad_norm": 0.44412219524383545,
      "learning_rate": 4.065439927058726e-06,
      "loss": 0.7659,
      "step": 4511
    },
    {
      "epoch": 0.18699490239960215,
      "grad_norm": 0.4414145350456238,
      "learning_rate": 4.0652327075303585e-06,
      "loss": 0.7441,
      "step": 4512
    },
    {
      "epoch": 0.1870363463052758,
      "grad_norm": 0.4308534264564514,
      "learning_rate": 4.06502548800199e-06,
      "loss": 0.7639,
      "step": 4513
    },
    {
      "epoch": 0.18707779021094947,
      "grad_norm": 0.42125049233436584,
      "learning_rate": 4.064818268473621e-06,
      "loss": 0.6951,
      "step": 4514
    },
    {
      "epoch": 0.18711923411662315,
      "grad_norm": 0.461679607629776,
      "learning_rate": 4.064611048945253e-06,
      "loss": 0.7249,
      "step": 4515
    },
    {
      "epoch": 0.18716067802229683,
      "grad_norm": 0.4059103727340698,
      "learning_rate": 4.064403829416885e-06,
      "loss": 0.7175,
      "step": 4516
    },
    {
      "epoch": 0.1872021219279705,
      "grad_norm": 0.45587003231048584,
      "learning_rate": 4.064196609888516e-06,
      "loss": 0.7754,
      "step": 4517
    },
    {
      "epoch": 0.18724356583364415,
      "grad_norm": 0.4460197389125824,
      "learning_rate": 4.063989390360148e-06,
      "loss": 0.7114,
      "step": 4518
    },
    {
      "epoch": 0.18728500973931783,
      "grad_norm": 0.44447067379951477,
      "learning_rate": 4.063782170831779e-06,
      "loss": 0.7465,
      "step": 4519
    },
    {
      "epoch": 0.1873264536449915,
      "grad_norm": 0.4353641867637634,
      "learning_rate": 4.063574951303411e-06,
      "loss": 0.7134,
      "step": 4520
    },
    {
      "epoch": 0.18736789755066519,
      "grad_norm": 0.4644107222557068,
      "learning_rate": 4.0633677317750435e-06,
      "loss": 0.7518,
      "step": 4521
    },
    {
      "epoch": 0.18740934145633883,
      "grad_norm": 0.4185137450695038,
      "learning_rate": 4.063160512246674e-06,
      "loss": 0.7175,
      "step": 4522
    },
    {
      "epoch": 0.1874507853620125,
      "grad_norm": 0.4142124652862549,
      "learning_rate": 4.062953292718306e-06,
      "loss": 0.7031,
      "step": 4523
    },
    {
      "epoch": 0.1874922292676862,
      "grad_norm": 0.4611794352531433,
      "learning_rate": 4.062746073189938e-06,
      "loss": 0.728,
      "step": 4524
    },
    {
      "epoch": 0.18753367317335987,
      "grad_norm": 0.44782915711402893,
      "learning_rate": 4.06253885366157e-06,
      "loss": 0.7666,
      "step": 4525
    },
    {
      "epoch": 0.18757511707903352,
      "grad_norm": 0.39586141705513,
      "learning_rate": 4.062331634133201e-06,
      "loss": 0.6879,
      "step": 4526
    },
    {
      "epoch": 0.1876165609847072,
      "grad_norm": 0.415486603975296,
      "learning_rate": 4.062124414604833e-06,
      "loss": 0.7319,
      "step": 4527
    },
    {
      "epoch": 0.18765800489038087,
      "grad_norm": 0.46490809321403503,
      "learning_rate": 4.061917195076464e-06,
      "loss": 0.7817,
      "step": 4528
    },
    {
      "epoch": 0.18769944879605455,
      "grad_norm": 0.4135245978832245,
      "learning_rate": 4.061709975548096e-06,
      "loss": 0.73,
      "step": 4529
    },
    {
      "epoch": 0.18774089270172822,
      "grad_norm": 0.421370267868042,
      "learning_rate": 4.061502756019728e-06,
      "loss": 0.7412,
      "step": 4530
    },
    {
      "epoch": 0.18778233660740187,
      "grad_norm": 0.39247825741767883,
      "learning_rate": 4.061295536491359e-06,
      "loss": 0.6741,
      "step": 4531
    },
    {
      "epoch": 0.18782378051307555,
      "grad_norm": 0.41833725571632385,
      "learning_rate": 4.061088316962991e-06,
      "loss": 0.7236,
      "step": 4532
    },
    {
      "epoch": 0.18786522441874923,
      "grad_norm": 0.45110923051834106,
      "learning_rate": 4.060881097434623e-06,
      "loss": 0.7419,
      "step": 4533
    },
    {
      "epoch": 0.1879066683244229,
      "grad_norm": 0.45057857036590576,
      "learning_rate": 4.060673877906254e-06,
      "loss": 0.7288,
      "step": 4534
    },
    {
      "epoch": 0.18794811223009655,
      "grad_norm": 0.4229746162891388,
      "learning_rate": 4.060466658377885e-06,
      "loss": 0.7893,
      "step": 4535
    },
    {
      "epoch": 0.18798955613577023,
      "grad_norm": 0.4103749692440033,
      "learning_rate": 4.060259438849518e-06,
      "loss": 0.7485,
      "step": 4536
    },
    {
      "epoch": 0.1880310000414439,
      "grad_norm": 0.4778692424297333,
      "learning_rate": 4.060052219321149e-06,
      "loss": 0.7432,
      "step": 4537
    },
    {
      "epoch": 0.18807244394711758,
      "grad_norm": 0.4249318540096283,
      "learning_rate": 4.0598449997927804e-06,
      "loss": 0.7505,
      "step": 4538
    },
    {
      "epoch": 0.18811388785279123,
      "grad_norm": 0.39245545864105225,
      "learning_rate": 4.059637780264413e-06,
      "loss": 0.7092,
      "step": 4539
    },
    {
      "epoch": 0.1881553317584649,
      "grad_norm": 0.42722639441490173,
      "learning_rate": 4.059430560736044e-06,
      "loss": 0.7949,
      "step": 4540
    },
    {
      "epoch": 0.1881967756641386,
      "grad_norm": 0.42837244272232056,
      "learning_rate": 4.059223341207676e-06,
      "loss": 0.71,
      "step": 4541
    },
    {
      "epoch": 0.18823821956981227,
      "grad_norm": 0.3988135755062103,
      "learning_rate": 4.059016121679308e-06,
      "loss": 0.6587,
      "step": 4542
    },
    {
      "epoch": 0.18827966347548594,
      "grad_norm": 0.43268293142318726,
      "learning_rate": 4.058808902150939e-06,
      "loss": 0.7151,
      "step": 4543
    },
    {
      "epoch": 0.1883211073811596,
      "grad_norm": 0.39718854427337646,
      "learning_rate": 4.0586016826225704e-06,
      "loss": 0.6899,
      "step": 4544
    },
    {
      "epoch": 0.18836255128683327,
      "grad_norm": 0.38562101125717163,
      "learning_rate": 4.058394463094203e-06,
      "loss": 0.6707,
      "step": 4545
    },
    {
      "epoch": 0.18840399519250695,
      "grad_norm": 0.4162629246711731,
      "learning_rate": 4.058187243565834e-06,
      "loss": 0.7649,
      "step": 4546
    },
    {
      "epoch": 0.18844543909818062,
      "grad_norm": 0.42889317870140076,
      "learning_rate": 4.0579800240374654e-06,
      "loss": 0.6963,
      "step": 4547
    },
    {
      "epoch": 0.18848688300385427,
      "grad_norm": 0.39304643869400024,
      "learning_rate": 4.057772804509098e-06,
      "loss": 0.6926,
      "step": 4548
    },
    {
      "epoch": 0.18852832690952795,
      "grad_norm": 0.4116537272930145,
      "learning_rate": 4.057565584980729e-06,
      "loss": 0.7185,
      "step": 4549
    },
    {
      "epoch": 0.18856977081520163,
      "grad_norm": 0.43976545333862305,
      "learning_rate": 4.0573583654523604e-06,
      "loss": 0.7808,
      "step": 4550
    },
    {
      "epoch": 0.1886112147208753,
      "grad_norm": 0.4314555823802948,
      "learning_rate": 4.057151145923992e-06,
      "loss": 0.7341,
      "step": 4551
    },
    {
      "epoch": 0.18865265862654895,
      "grad_norm": 0.4224686622619629,
      "learning_rate": 4.056943926395624e-06,
      "loss": 0.7239,
      "step": 4552
    },
    {
      "epoch": 0.18869410253222263,
      "grad_norm": 0.4143499732017517,
      "learning_rate": 4.0567367068672554e-06,
      "loss": 0.731,
      "step": 4553
    },
    {
      "epoch": 0.1887355464378963,
      "grad_norm": 0.46924149990081787,
      "learning_rate": 4.056529487338887e-06,
      "loss": 0.752,
      "step": 4554
    },
    {
      "epoch": 0.18877699034356998,
      "grad_norm": 0.43394437432289124,
      "learning_rate": 4.056322267810518e-06,
      "loss": 0.6842,
      "step": 4555
    },
    {
      "epoch": 0.18881843424924366,
      "grad_norm": 0.44339719414711,
      "learning_rate": 4.0561150482821504e-06,
      "loss": 0.7294,
      "step": 4556
    },
    {
      "epoch": 0.1888598781549173,
      "grad_norm": 0.44915178418159485,
      "learning_rate": 4.055907828753783e-06,
      "loss": 0.7786,
      "step": 4557
    },
    {
      "epoch": 0.188901322060591,
      "grad_norm": 0.43381553888320923,
      "learning_rate": 4.055700609225413e-06,
      "loss": 0.7251,
      "step": 4558
    },
    {
      "epoch": 0.18894276596626466,
      "grad_norm": 0.42380794882774353,
      "learning_rate": 4.0554933896970455e-06,
      "loss": 0.6536,
      "step": 4559
    },
    {
      "epoch": 0.18898420987193834,
      "grad_norm": 0.4210614562034607,
      "learning_rate": 4.055286170168677e-06,
      "loss": 0.7212,
      "step": 4560
    },
    {
      "epoch": 0.189025653777612,
      "grad_norm": 0.43417122960090637,
      "learning_rate": 4.055078950640309e-06,
      "loss": 0.7566,
      "step": 4561
    },
    {
      "epoch": 0.18906709768328567,
      "grad_norm": 0.4526064991950989,
      "learning_rate": 4.0548717311119405e-06,
      "loss": 0.7253,
      "step": 4562
    },
    {
      "epoch": 0.18910854158895934,
      "grad_norm": 0.39087745547294617,
      "learning_rate": 4.054664511583572e-06,
      "loss": 0.7168,
      "step": 4563
    },
    {
      "epoch": 0.18914998549463302,
      "grad_norm": 0.44654592871665955,
      "learning_rate": 4.054457292055204e-06,
      "loss": 0.7649,
      "step": 4564
    },
    {
      "epoch": 0.1891914294003067,
      "grad_norm": 0.4870911240577698,
      "learning_rate": 4.0542500725268355e-06,
      "loss": 0.7771,
      "step": 4565
    },
    {
      "epoch": 0.18923287330598035,
      "grad_norm": 0.3836575448513031,
      "learning_rate": 4.054042852998467e-06,
      "loss": 0.6902,
      "step": 4566
    },
    {
      "epoch": 0.18927431721165403,
      "grad_norm": 0.42108556628227234,
      "learning_rate": 4.053835633470098e-06,
      "loss": 0.7312,
      "step": 4567
    },
    {
      "epoch": 0.1893157611173277,
      "grad_norm": 0.4212074875831604,
      "learning_rate": 4.0536284139417305e-06,
      "loss": 0.7681,
      "step": 4568
    },
    {
      "epoch": 0.18935720502300138,
      "grad_norm": 0.4120270311832428,
      "learning_rate": 4.053421194413362e-06,
      "loss": 0.7211,
      "step": 4569
    },
    {
      "epoch": 0.18939864892867503,
      "grad_norm": 0.4326016306877136,
      "learning_rate": 4.053213974884993e-06,
      "loss": 0.7134,
      "step": 4570
    },
    {
      "epoch": 0.1894400928343487,
      "grad_norm": 0.4769447147846222,
      "learning_rate": 4.053006755356625e-06,
      "loss": 0.7971,
      "step": 4571
    },
    {
      "epoch": 0.18948153674002238,
      "grad_norm": 0.4197107255458832,
      "learning_rate": 4.052799535828257e-06,
      "loss": 0.7734,
      "step": 4572
    },
    {
      "epoch": 0.18952298064569606,
      "grad_norm": 0.4613288640975952,
      "learning_rate": 4.052592316299889e-06,
      "loss": 0.7039,
      "step": 4573
    },
    {
      "epoch": 0.1895644245513697,
      "grad_norm": 0.46716392040252686,
      "learning_rate": 4.05238509677152e-06,
      "loss": 0.7341,
      "step": 4574
    },
    {
      "epoch": 0.18960586845704339,
      "grad_norm": 0.4393904507160187,
      "learning_rate": 4.052177877243152e-06,
      "loss": 0.7639,
      "step": 4575
    },
    {
      "epoch": 0.18964731236271706,
      "grad_norm": 0.4309116005897522,
      "learning_rate": 4.051970657714783e-06,
      "loss": 0.7261,
      "step": 4576
    },
    {
      "epoch": 0.18968875626839074,
      "grad_norm": 0.4547387659549713,
      "learning_rate": 4.0517634381864155e-06,
      "loss": 0.7209,
      "step": 4577
    },
    {
      "epoch": 0.18973020017406442,
      "grad_norm": 0.4115113615989685,
      "learning_rate": 4.051556218658047e-06,
      "loss": 0.7495,
      "step": 4578
    },
    {
      "epoch": 0.18977164407973807,
      "grad_norm": 0.4658704996109009,
      "learning_rate": 4.051348999129678e-06,
      "loss": 0.7639,
      "step": 4579
    },
    {
      "epoch": 0.18981308798541174,
      "grad_norm": 0.4295717477798462,
      "learning_rate": 4.05114177960131e-06,
      "loss": 0.7456,
      "step": 4580
    },
    {
      "epoch": 0.18985453189108542,
      "grad_norm": 0.4501297175884247,
      "learning_rate": 4.050934560072942e-06,
      "loss": 0.7111,
      "step": 4581
    },
    {
      "epoch": 0.1898959757967591,
      "grad_norm": 0.4124656617641449,
      "learning_rate": 4.050727340544573e-06,
      "loss": 0.7607,
      "step": 4582
    },
    {
      "epoch": 0.18993741970243275,
      "grad_norm": 0.3898340165615082,
      "learning_rate": 4.050520121016205e-06,
      "loss": 0.7148,
      "step": 4583
    },
    {
      "epoch": 0.18997886360810642,
      "grad_norm": 0.39090073108673096,
      "learning_rate": 4.050312901487837e-06,
      "loss": 0.7058,
      "step": 4584
    },
    {
      "epoch": 0.1900203075137801,
      "grad_norm": 0.46643778681755066,
      "learning_rate": 4.050105681959468e-06,
      "loss": 0.7339,
      "step": 4585
    },
    {
      "epoch": 0.19006175141945378,
      "grad_norm": 0.39459338784217834,
      "learning_rate": 4.0498984624311e-06,
      "loss": 0.7043,
      "step": 4586
    },
    {
      "epoch": 0.19010319532512743,
      "grad_norm": 0.43162456154823303,
      "learning_rate": 4.049691242902731e-06,
      "loss": 0.77,
      "step": 4587
    },
    {
      "epoch": 0.1901446392308011,
      "grad_norm": 0.40933850407600403,
      "learning_rate": 4.049484023374363e-06,
      "loss": 0.6736,
      "step": 4588
    },
    {
      "epoch": 0.19018608313647478,
      "grad_norm": 0.40110868215560913,
      "learning_rate": 4.049276803845995e-06,
      "loss": 0.6628,
      "step": 4589
    },
    {
      "epoch": 0.19022752704214846,
      "grad_norm": 0.43051692843437195,
      "learning_rate": 4.049069584317626e-06,
      "loss": 0.7307,
      "step": 4590
    },
    {
      "epoch": 0.19026897094782214,
      "grad_norm": 0.4257526993751526,
      "learning_rate": 4.048862364789258e-06,
      "loss": 0.687,
      "step": 4591
    },
    {
      "epoch": 0.19031041485349579,
      "grad_norm": 0.423849880695343,
      "learning_rate": 4.04865514526089e-06,
      "loss": 0.7463,
      "step": 4592
    },
    {
      "epoch": 0.19035185875916946,
      "grad_norm": 0.44114941358566284,
      "learning_rate": 4.048447925732522e-06,
      "loss": 0.749,
      "step": 4593
    },
    {
      "epoch": 0.19039330266484314,
      "grad_norm": 0.4252367913722992,
      "learning_rate": 4.048240706204153e-06,
      "loss": 0.7301,
      "step": 4594
    },
    {
      "epoch": 0.19043474657051682,
      "grad_norm": 0.40352678298950195,
      "learning_rate": 4.048033486675785e-06,
      "loss": 0.6644,
      "step": 4595
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.41463467478752136,
      "learning_rate": 4.047826267147416e-06,
      "loss": 0.7327,
      "step": 4596
    },
    {
      "epoch": 0.19051763438186414,
      "grad_norm": 0.4293050169944763,
      "learning_rate": 4.047619047619048e-06,
      "loss": 0.6785,
      "step": 4597
    },
    {
      "epoch": 0.19055907828753782,
      "grad_norm": 0.3792731463909149,
      "learning_rate": 4.04741182809068e-06,
      "loss": 0.7295,
      "step": 4598
    },
    {
      "epoch": 0.1906005221932115,
      "grad_norm": 0.4735502600669861,
      "learning_rate": 4.047204608562311e-06,
      "loss": 0.7788,
      "step": 4599
    },
    {
      "epoch": 0.19064196609888515,
      "grad_norm": 0.4165715277194977,
      "learning_rate": 4.046997389033943e-06,
      "loss": 0.6671,
      "step": 4600
    },
    {
      "epoch": 0.19068341000455882,
      "grad_norm": 0.461496502161026,
      "learning_rate": 4.046790169505575e-06,
      "loss": 0.6561,
      "step": 4601
    },
    {
      "epoch": 0.1907248539102325,
      "grad_norm": 0.411302387714386,
      "learning_rate": 4.046582949977206e-06,
      "loss": 0.6951,
      "step": 4602
    },
    {
      "epoch": 0.19076629781590618,
      "grad_norm": 0.4746898412704468,
      "learning_rate": 4.0463757304488374e-06,
      "loss": 0.7471,
      "step": 4603
    },
    {
      "epoch": 0.19080774172157985,
      "grad_norm": 0.44204774498939514,
      "learning_rate": 4.04616851092047e-06,
      "loss": 0.7278,
      "step": 4604
    },
    {
      "epoch": 0.1908491856272535,
      "grad_norm": 0.42630013823509216,
      "learning_rate": 4.045961291392101e-06,
      "loss": 0.707,
      "step": 4605
    },
    {
      "epoch": 0.19089062953292718,
      "grad_norm": 0.40937915444374084,
      "learning_rate": 4.0457540718637324e-06,
      "loss": 0.6929,
      "step": 4606
    },
    {
      "epoch": 0.19093207343860086,
      "grad_norm": 0.455097496509552,
      "learning_rate": 4.045546852335364e-06,
      "loss": 0.7732,
      "step": 4607
    },
    {
      "epoch": 0.19097351734427453,
      "grad_norm": 0.4573996067047119,
      "learning_rate": 4.045339632806996e-06,
      "loss": 0.7803,
      "step": 4608
    },
    {
      "epoch": 0.19101496124994818,
      "grad_norm": 0.41557538509368896,
      "learning_rate": 4.045132413278628e-06,
      "loss": 0.6805,
      "step": 4609
    },
    {
      "epoch": 0.19105640515562186,
      "grad_norm": 0.4120057225227356,
      "learning_rate": 4.04492519375026e-06,
      "loss": 0.6747,
      "step": 4610
    },
    {
      "epoch": 0.19109784906129554,
      "grad_norm": 0.420397013425827,
      "learning_rate": 4.044717974221891e-06,
      "loss": 0.7373,
      "step": 4611
    },
    {
      "epoch": 0.19113929296696922,
      "grad_norm": 0.45325493812561035,
      "learning_rate": 4.0445107546935224e-06,
      "loss": 0.7023,
      "step": 4612
    },
    {
      "epoch": 0.19118073687264286,
      "grad_norm": 0.5066434741020203,
      "learning_rate": 4.044303535165155e-06,
      "loss": 0.767,
      "step": 4613
    },
    {
      "epoch": 0.19122218077831654,
      "grad_norm": 0.4181893765926361,
      "learning_rate": 4.044096315636786e-06,
      "loss": 0.7271,
      "step": 4614
    },
    {
      "epoch": 0.19126362468399022,
      "grad_norm": 0.39929765462875366,
      "learning_rate": 4.0438890961084174e-06,
      "loss": 0.6302,
      "step": 4615
    },
    {
      "epoch": 0.1913050685896639,
      "grad_norm": 0.45301181077957153,
      "learning_rate": 4.043681876580049e-06,
      "loss": 0.698,
      "step": 4616
    },
    {
      "epoch": 0.19134651249533757,
      "grad_norm": 0.4343165457248688,
      "learning_rate": 4.043474657051681e-06,
      "loss": 0.7102,
      "step": 4617
    },
    {
      "epoch": 0.19138795640101122,
      "grad_norm": 0.42214369773864746,
      "learning_rate": 4.0432674375233125e-06,
      "loss": 0.7515,
      "step": 4618
    },
    {
      "epoch": 0.1914294003066849,
      "grad_norm": 0.409945011138916,
      "learning_rate": 4.043060217994944e-06,
      "loss": 0.7046,
      "step": 4619
    },
    {
      "epoch": 0.19147084421235858,
      "grad_norm": 0.41377487778663635,
      "learning_rate": 4.042852998466576e-06,
      "loss": 0.7241,
      "step": 4620
    },
    {
      "epoch": 0.19151228811803225,
      "grad_norm": 0.5082868933677673,
      "learning_rate": 4.0426457789382075e-06,
      "loss": 0.7485,
      "step": 4621
    },
    {
      "epoch": 0.1915537320237059,
      "grad_norm": 0.4290807843208313,
      "learning_rate": 4.042438559409839e-06,
      "loss": 0.6943,
      "step": 4622
    },
    {
      "epoch": 0.19159517592937958,
      "grad_norm": 0.4289205074310303,
      "learning_rate": 4.04223133988147e-06,
      "loss": 0.7024,
      "step": 4623
    },
    {
      "epoch": 0.19163661983505326,
      "grad_norm": 0.4343908727169037,
      "learning_rate": 4.0420241203531025e-06,
      "loss": 0.7393,
      "step": 4624
    },
    {
      "epoch": 0.19167806374072693,
      "grad_norm": 0.41442152857780457,
      "learning_rate": 4.041816900824735e-06,
      "loss": 0.7085,
      "step": 4625
    },
    {
      "epoch": 0.19171950764640058,
      "grad_norm": 0.47668692469596863,
      "learning_rate": 4.041609681296366e-06,
      "loss": 0.7617,
      "step": 4626
    },
    {
      "epoch": 0.19176095155207426,
      "grad_norm": 0.4013442099094391,
      "learning_rate": 4.0414024617679975e-06,
      "loss": 0.6982,
      "step": 4627
    },
    {
      "epoch": 0.19180239545774794,
      "grad_norm": 0.38636621832847595,
      "learning_rate": 4.041195242239629e-06,
      "loss": 0.6858,
      "step": 4628
    },
    {
      "epoch": 0.19184383936342161,
      "grad_norm": 0.42310792207717896,
      "learning_rate": 4.040988022711261e-06,
      "loss": 0.7351,
      "step": 4629
    },
    {
      "epoch": 0.1918852832690953,
      "grad_norm": 0.4351435899734497,
      "learning_rate": 4.0407808031828925e-06,
      "loss": 0.7478,
      "step": 4630
    },
    {
      "epoch": 0.19192672717476894,
      "grad_norm": 0.42125818133354187,
      "learning_rate": 4.040573583654524e-06,
      "loss": 0.7399,
      "step": 4631
    },
    {
      "epoch": 0.19196817108044262,
      "grad_norm": 0.4257926344871521,
      "learning_rate": 4.040366364126155e-06,
      "loss": 0.7258,
      "step": 4632
    },
    {
      "epoch": 0.1920096149861163,
      "grad_norm": 0.4051901400089264,
      "learning_rate": 4.0401591445977875e-06,
      "loss": 0.6868,
      "step": 4633
    },
    {
      "epoch": 0.19205105889178997,
      "grad_norm": 0.43173763155937195,
      "learning_rate": 4.039951925069419e-06,
      "loss": 0.7466,
      "step": 4634
    },
    {
      "epoch": 0.19209250279746362,
      "grad_norm": 0.379215806722641,
      "learning_rate": 4.03974470554105e-06,
      "loss": 0.6306,
      "step": 4635
    },
    {
      "epoch": 0.1921339467031373,
      "grad_norm": 0.4459201693534851,
      "learning_rate": 4.0395374860126825e-06,
      "loss": 0.814,
      "step": 4636
    },
    {
      "epoch": 0.19217539060881098,
      "grad_norm": 0.39524680376052856,
      "learning_rate": 4.039330266484314e-06,
      "loss": 0.704,
      "step": 4637
    },
    {
      "epoch": 0.19221683451448465,
      "grad_norm": 0.4332173466682434,
      "learning_rate": 4.039123046955945e-06,
      "loss": 0.7601,
      "step": 4638
    },
    {
      "epoch": 0.19225827842015833,
      "grad_norm": 0.4003659188747406,
      "learning_rate": 4.038915827427577e-06,
      "loss": 0.7054,
      "step": 4639
    },
    {
      "epoch": 0.19229972232583198,
      "grad_norm": 0.3922554850578308,
      "learning_rate": 4.038708607899209e-06,
      "loss": 0.6956,
      "step": 4640
    },
    {
      "epoch": 0.19234116623150566,
      "grad_norm": 0.41231662034988403,
      "learning_rate": 4.03850138837084e-06,
      "loss": 0.708,
      "step": 4641
    },
    {
      "epoch": 0.19238261013717933,
      "grad_norm": 0.4207746088504791,
      "learning_rate": 4.038294168842472e-06,
      "loss": 0.751,
      "step": 4642
    },
    {
      "epoch": 0.192424054042853,
      "grad_norm": 0.43675875663757324,
      "learning_rate": 4.038086949314104e-06,
      "loss": 0.7655,
      "step": 4643
    },
    {
      "epoch": 0.19246549794852666,
      "grad_norm": 0.44737759232521057,
      "learning_rate": 4.037879729785735e-06,
      "loss": 0.7456,
      "step": 4644
    },
    {
      "epoch": 0.19250694185420034,
      "grad_norm": 0.41867637634277344,
      "learning_rate": 4.0376725102573675e-06,
      "loss": 0.6938,
      "step": 4645
    },
    {
      "epoch": 0.192548385759874,
      "grad_norm": 0.4108772575855255,
      "learning_rate": 4.037465290728999e-06,
      "loss": 0.6995,
      "step": 4646
    },
    {
      "epoch": 0.1925898296655477,
      "grad_norm": 0.4237038195133209,
      "learning_rate": 4.03725807120063e-06,
      "loss": 0.688,
      "step": 4647
    },
    {
      "epoch": 0.19263127357122134,
      "grad_norm": 0.4350278675556183,
      "learning_rate": 4.037050851672262e-06,
      "loss": 0.7729,
      "step": 4648
    },
    {
      "epoch": 0.19267271747689502,
      "grad_norm": 0.46770337224006653,
      "learning_rate": 4.036843632143894e-06,
      "loss": 0.7792,
      "step": 4649
    },
    {
      "epoch": 0.1927141613825687,
      "grad_norm": 0.4315145015716553,
      "learning_rate": 4.036636412615525e-06,
      "loss": 0.7316,
      "step": 4650
    },
    {
      "epoch": 0.19275560528824237,
      "grad_norm": 0.4256931245326996,
      "learning_rate": 4.036429193087157e-06,
      "loss": 0.698,
      "step": 4651
    },
    {
      "epoch": 0.19279704919391605,
      "grad_norm": 0.44189783930778503,
      "learning_rate": 4.036221973558789e-06,
      "loss": 0.7883,
      "step": 4652
    },
    {
      "epoch": 0.1928384930995897,
      "grad_norm": 0.4143000543117523,
      "learning_rate": 4.03601475403042e-06,
      "loss": 0.749,
      "step": 4653
    },
    {
      "epoch": 0.19287993700526337,
      "grad_norm": 0.42647290229797363,
      "learning_rate": 4.035807534502052e-06,
      "loss": 0.7305,
      "step": 4654
    },
    {
      "epoch": 0.19292138091093705,
      "grad_norm": 0.3948799967765808,
      "learning_rate": 4.035600314973683e-06,
      "loss": 0.7607,
      "step": 4655
    },
    {
      "epoch": 0.19296282481661073,
      "grad_norm": 0.4015995264053345,
      "learning_rate": 4.035393095445315e-06,
      "loss": 0.6929,
      "step": 4656
    },
    {
      "epoch": 0.19300426872228438,
      "grad_norm": 0.4328002333641052,
      "learning_rate": 4.035185875916947e-06,
      "loss": 0.7261,
      "step": 4657
    },
    {
      "epoch": 0.19304571262795805,
      "grad_norm": 0.4168425500392914,
      "learning_rate": 4.034978656388578e-06,
      "loss": 0.6941,
      "step": 4658
    },
    {
      "epoch": 0.19308715653363173,
      "grad_norm": 0.4482492208480835,
      "learning_rate": 4.0347714368602094e-06,
      "loss": 0.7507,
      "step": 4659
    },
    {
      "epoch": 0.1931286004393054,
      "grad_norm": 0.4020955264568329,
      "learning_rate": 4.034564217331842e-06,
      "loss": 0.7126,
      "step": 4660
    },
    {
      "epoch": 0.19317004434497906,
      "grad_norm": 0.42129167914390564,
      "learning_rate": 4.034356997803474e-06,
      "loss": 0.698,
      "step": 4661
    },
    {
      "epoch": 0.19321148825065274,
      "grad_norm": 0.4337470531463623,
      "learning_rate": 4.034149778275105e-06,
      "loss": 0.7654,
      "step": 4662
    },
    {
      "epoch": 0.1932529321563264,
      "grad_norm": 0.4724188446998596,
      "learning_rate": 4.033942558746737e-06,
      "loss": 0.7325,
      "step": 4663
    },
    {
      "epoch": 0.1932943760620001,
      "grad_norm": 0.4335235357284546,
      "learning_rate": 4.033735339218368e-06,
      "loss": 0.7449,
      "step": 4664
    },
    {
      "epoch": 0.19333581996767377,
      "grad_norm": 0.3979817032814026,
      "learning_rate": 4.03352811969e-06,
      "loss": 0.6597,
      "step": 4665
    },
    {
      "epoch": 0.19337726387334742,
      "grad_norm": 0.4097374379634857,
      "learning_rate": 4.033320900161632e-06,
      "loss": 0.7461,
      "step": 4666
    },
    {
      "epoch": 0.1934187077790211,
      "grad_norm": 0.4207673668861389,
      "learning_rate": 4.033113680633263e-06,
      "loss": 0.739,
      "step": 4667
    },
    {
      "epoch": 0.19346015168469477,
      "grad_norm": 0.4265035092830658,
      "learning_rate": 4.0329064611048944e-06,
      "loss": 0.7495,
      "step": 4668
    },
    {
      "epoch": 0.19350159559036845,
      "grad_norm": 0.4405616819858551,
      "learning_rate": 4.032699241576527e-06,
      "loss": 0.7598,
      "step": 4669
    },
    {
      "epoch": 0.1935430394960421,
      "grad_norm": 0.4505399167537689,
      "learning_rate": 4.032492022048158e-06,
      "loss": 0.7483,
      "step": 4670
    },
    {
      "epoch": 0.19358448340171577,
      "grad_norm": 0.46396902203559875,
      "learning_rate": 4.0322848025197894e-06,
      "loss": 0.7576,
      "step": 4671
    },
    {
      "epoch": 0.19362592730738945,
      "grad_norm": 0.4049902558326721,
      "learning_rate": 4.032077582991422e-06,
      "loss": 0.7776,
      "step": 4672
    },
    {
      "epoch": 0.19366737121306313,
      "grad_norm": 0.4100184440612793,
      "learning_rate": 4.031870363463053e-06,
      "loss": 0.7155,
      "step": 4673
    },
    {
      "epoch": 0.19370881511873678,
      "grad_norm": 0.46287715435028076,
      "learning_rate": 4.0316631439346844e-06,
      "loss": 0.7488,
      "step": 4674
    },
    {
      "epoch": 0.19375025902441045,
      "grad_norm": 0.42693856358528137,
      "learning_rate": 4.031455924406316e-06,
      "loss": 0.7178,
      "step": 4675
    },
    {
      "epoch": 0.19379170293008413,
      "grad_norm": 0.4630483090877533,
      "learning_rate": 4.031248704877948e-06,
      "loss": 0.7424,
      "step": 4676
    },
    {
      "epoch": 0.1938331468357578,
      "grad_norm": 0.44017019867897034,
      "learning_rate": 4.0310414853495795e-06,
      "loss": 0.7343,
      "step": 4677
    },
    {
      "epoch": 0.19387459074143149,
      "grad_norm": 0.4545411765575409,
      "learning_rate": 4.030834265821212e-06,
      "loss": 0.7278,
      "step": 4678
    },
    {
      "epoch": 0.19391603464710513,
      "grad_norm": 0.4146547019481659,
      "learning_rate": 4.030627046292843e-06,
      "loss": 0.7363,
      "step": 4679
    },
    {
      "epoch": 0.1939574785527788,
      "grad_norm": 0.4015137851238251,
      "learning_rate": 4.0304198267644745e-06,
      "loss": 0.709,
      "step": 4680
    },
    {
      "epoch": 0.1939989224584525,
      "grad_norm": 0.45078110694885254,
      "learning_rate": 4.030212607236107e-06,
      "loss": 0.7137,
      "step": 4681
    },
    {
      "epoch": 0.19404036636412617,
      "grad_norm": 0.4350622892379761,
      "learning_rate": 4.030005387707738e-06,
      "loss": 0.7303,
      "step": 4682
    },
    {
      "epoch": 0.19408181026979981,
      "grad_norm": 0.4460679292678833,
      "learning_rate": 4.0297981681793695e-06,
      "loss": 0.718,
      "step": 4683
    },
    {
      "epoch": 0.1941232541754735,
      "grad_norm": 0.45762166380882263,
      "learning_rate": 4.029590948651001e-06,
      "loss": 0.7769,
      "step": 4684
    },
    {
      "epoch": 0.19416469808114717,
      "grad_norm": 0.41030752658843994,
      "learning_rate": 4.029383729122633e-06,
      "loss": 0.7402,
      "step": 4685
    },
    {
      "epoch": 0.19420614198682085,
      "grad_norm": 0.41783419251441956,
      "learning_rate": 4.0291765095942645e-06,
      "loss": 0.7539,
      "step": 4686
    },
    {
      "epoch": 0.1942475858924945,
      "grad_norm": 0.42504847049713135,
      "learning_rate": 4.028969290065896e-06,
      "loss": 0.7483,
      "step": 4687
    },
    {
      "epoch": 0.19428902979816817,
      "grad_norm": 0.46193593740463257,
      "learning_rate": 4.028762070537528e-06,
      "loss": 0.7795,
      "step": 4688
    },
    {
      "epoch": 0.19433047370384185,
      "grad_norm": 0.4177502989768982,
      "learning_rate": 4.0285548510091595e-06,
      "loss": 0.7109,
      "step": 4689
    },
    {
      "epoch": 0.19437191760951553,
      "grad_norm": 0.43712031841278076,
      "learning_rate": 4.028347631480791e-06,
      "loss": 0.7317,
      "step": 4690
    },
    {
      "epoch": 0.1944133615151892,
      "grad_norm": 0.40311703085899353,
      "learning_rate": 4.028140411952422e-06,
      "loss": 0.6644,
      "step": 4691
    },
    {
      "epoch": 0.19445480542086285,
      "grad_norm": 0.42702972888946533,
      "learning_rate": 4.0279331924240545e-06,
      "loss": 0.7402,
      "step": 4692
    },
    {
      "epoch": 0.19449624932653653,
      "grad_norm": 0.42521578073501587,
      "learning_rate": 4.027725972895686e-06,
      "loss": 0.7621,
      "step": 4693
    },
    {
      "epoch": 0.1945376932322102,
      "grad_norm": 0.4639471173286438,
      "learning_rate": 4.027518753367318e-06,
      "loss": 0.7231,
      "step": 4694
    },
    {
      "epoch": 0.19457913713788388,
      "grad_norm": 0.7300617098808289,
      "learning_rate": 4.0273115338389495e-06,
      "loss": 0.8115,
      "step": 4695
    },
    {
      "epoch": 0.19462058104355753,
      "grad_norm": 0.42894530296325684,
      "learning_rate": 4.027104314310581e-06,
      "loss": 0.7463,
      "step": 4696
    },
    {
      "epoch": 0.1946620249492312,
      "grad_norm": 0.4382060766220093,
      "learning_rate": 4.026897094782213e-06,
      "loss": 0.7388,
      "step": 4697
    },
    {
      "epoch": 0.1947034688549049,
      "grad_norm": 0.4475901126861572,
      "learning_rate": 4.0266898752538445e-06,
      "loss": 0.7185,
      "step": 4698
    },
    {
      "epoch": 0.19474491276057856,
      "grad_norm": 0.4147927165031433,
      "learning_rate": 4.026482655725476e-06,
      "loss": 0.7539,
      "step": 4699
    },
    {
      "epoch": 0.19478635666625221,
      "grad_norm": 0.4532407522201538,
      "learning_rate": 4.026275436197107e-06,
      "loss": 0.7031,
      "step": 4700
    },
    {
      "epoch": 0.1948278005719259,
      "grad_norm": 0.4391915500164032,
      "learning_rate": 4.0260682166687395e-06,
      "loss": 0.7428,
      "step": 4701
    },
    {
      "epoch": 0.19486924447759957,
      "grad_norm": 0.4146566689014435,
      "learning_rate": 4.025860997140371e-06,
      "loss": 0.7,
      "step": 4702
    },
    {
      "epoch": 0.19491068838327325,
      "grad_norm": 0.3942202031612396,
      "learning_rate": 4.025653777612002e-06,
      "loss": 0.7435,
      "step": 4703
    },
    {
      "epoch": 0.19495213228894692,
      "grad_norm": 0.4221835732460022,
      "learning_rate": 4.0254465580836345e-06,
      "loss": 0.7161,
      "step": 4704
    },
    {
      "epoch": 0.19499357619462057,
      "grad_norm": 0.4317571222782135,
      "learning_rate": 4.025239338555266e-06,
      "loss": 0.7295,
      "step": 4705
    },
    {
      "epoch": 0.19503502010029425,
      "grad_norm": 0.4274677634239197,
      "learning_rate": 4.025032119026897e-06,
      "loss": 0.6987,
      "step": 4706
    },
    {
      "epoch": 0.19507646400596793,
      "grad_norm": 0.4098416268825531,
      "learning_rate": 4.024824899498529e-06,
      "loss": 0.6824,
      "step": 4707
    },
    {
      "epoch": 0.1951179079116416,
      "grad_norm": 0.4584222733974457,
      "learning_rate": 4.024617679970161e-06,
      "loss": 0.802,
      "step": 4708
    },
    {
      "epoch": 0.19515935181731525,
      "grad_norm": 0.44911476969718933,
      "learning_rate": 4.024410460441792e-06,
      "loss": 0.6997,
      "step": 4709
    },
    {
      "epoch": 0.19520079572298893,
      "grad_norm": 0.41559773683547974,
      "learning_rate": 4.024203240913424e-06,
      "loss": 0.6874,
      "step": 4710
    },
    {
      "epoch": 0.1952422396286626,
      "grad_norm": 0.4233519434928894,
      "learning_rate": 4.023996021385055e-06,
      "loss": 0.7418,
      "step": 4711
    },
    {
      "epoch": 0.19528368353433628,
      "grad_norm": 0.4086543619632721,
      "learning_rate": 4.023788801856687e-06,
      "loss": 0.739,
      "step": 4712
    },
    {
      "epoch": 0.19532512744000996,
      "grad_norm": 0.41571271419525146,
      "learning_rate": 4.0235815823283195e-06,
      "loss": 0.74,
      "step": 4713
    },
    {
      "epoch": 0.1953665713456836,
      "grad_norm": 0.416396826505661,
      "learning_rate": 4.023374362799951e-06,
      "loss": 0.7745,
      "step": 4714
    },
    {
      "epoch": 0.1954080152513573,
      "grad_norm": 0.394703209400177,
      "learning_rate": 4.023167143271582e-06,
      "loss": 0.7347,
      "step": 4715
    },
    {
      "epoch": 0.19544945915703096,
      "grad_norm": 0.4165135622024536,
      "learning_rate": 4.022959923743214e-06,
      "loss": 0.6516,
      "step": 4716
    },
    {
      "epoch": 0.19549090306270464,
      "grad_norm": 0.4322117865085602,
      "learning_rate": 4.022752704214846e-06,
      "loss": 0.7524,
      "step": 4717
    },
    {
      "epoch": 0.1955323469683783,
      "grad_norm": 0.4325159788131714,
      "learning_rate": 4.022545484686477e-06,
      "loss": 0.7544,
      "step": 4718
    },
    {
      "epoch": 0.19557379087405197,
      "grad_norm": 0.4852556586265564,
      "learning_rate": 4.022338265158109e-06,
      "loss": 0.7518,
      "step": 4719
    },
    {
      "epoch": 0.19561523477972564,
      "grad_norm": 0.4567812979221344,
      "learning_rate": 4.02213104562974e-06,
      "loss": 0.7661,
      "step": 4720
    },
    {
      "epoch": 0.19565667868539932,
      "grad_norm": 0.43893447518348694,
      "learning_rate": 4.021923826101372e-06,
      "loss": 0.7236,
      "step": 4721
    },
    {
      "epoch": 0.19569812259107297,
      "grad_norm": 0.4195935130119324,
      "learning_rate": 4.021716606573004e-06,
      "loss": 0.7136,
      "step": 4722
    },
    {
      "epoch": 0.19573956649674665,
      "grad_norm": 0.4108867645263672,
      "learning_rate": 4.021509387044635e-06,
      "loss": 0.7021,
      "step": 4723
    },
    {
      "epoch": 0.19578101040242032,
      "grad_norm": 0.4351408779621124,
      "learning_rate": 4.021302167516267e-06,
      "loss": 0.7207,
      "step": 4724
    },
    {
      "epoch": 0.195822454308094,
      "grad_norm": 0.4044736325740814,
      "learning_rate": 4.021094947987899e-06,
      "loss": 0.7556,
      "step": 4725
    },
    {
      "epoch": 0.19586389821376768,
      "grad_norm": 0.40096017718315125,
      "learning_rate": 4.02088772845953e-06,
      "loss": 0.7661,
      "step": 4726
    },
    {
      "epoch": 0.19590534211944133,
      "grad_norm": 0.4301908016204834,
      "learning_rate": 4.0206805089311614e-06,
      "loss": 0.7642,
      "step": 4727
    },
    {
      "epoch": 0.195946786025115,
      "grad_norm": 0.4242943227291107,
      "learning_rate": 4.020473289402794e-06,
      "loss": 0.7063,
      "step": 4728
    },
    {
      "epoch": 0.19598822993078868,
      "grad_norm": 0.4021899402141571,
      "learning_rate": 4.020266069874425e-06,
      "loss": 0.7346,
      "step": 4729
    },
    {
      "epoch": 0.19602967383646236,
      "grad_norm": 0.39232689142227173,
      "learning_rate": 4.020058850346057e-06,
      "loss": 0.708,
      "step": 4730
    },
    {
      "epoch": 0.196071117742136,
      "grad_norm": 0.41039562225341797,
      "learning_rate": 4.019851630817689e-06,
      "loss": 0.705,
      "step": 4731
    },
    {
      "epoch": 0.19611256164780969,
      "grad_norm": 0.42049160599708557,
      "learning_rate": 4.01964441128932e-06,
      "loss": 0.7432,
      "step": 4732
    },
    {
      "epoch": 0.19615400555348336,
      "grad_norm": 0.45206204056739807,
      "learning_rate": 4.019437191760952e-06,
      "loss": 0.769,
      "step": 4733
    },
    {
      "epoch": 0.19619544945915704,
      "grad_norm": 0.4173351228237152,
      "learning_rate": 4.019229972232584e-06,
      "loss": 0.7271,
      "step": 4734
    },
    {
      "epoch": 0.1962368933648307,
      "grad_norm": 0.45424026250839233,
      "learning_rate": 4.019022752704215e-06,
      "loss": 0.7788,
      "step": 4735
    },
    {
      "epoch": 0.19627833727050437,
      "grad_norm": 0.4203905165195465,
      "learning_rate": 4.0188155331758465e-06,
      "loss": 0.7263,
      "step": 4736
    },
    {
      "epoch": 0.19631978117617804,
      "grad_norm": 0.42331433296203613,
      "learning_rate": 4.018608313647479e-06,
      "loss": 0.7061,
      "step": 4737
    },
    {
      "epoch": 0.19636122508185172,
      "grad_norm": 0.43471938371658325,
      "learning_rate": 4.01840109411911e-06,
      "loss": 0.7249,
      "step": 4738
    },
    {
      "epoch": 0.1964026689875254,
      "grad_norm": 0.407564252614975,
      "learning_rate": 4.0181938745907415e-06,
      "loss": 0.7015,
      "step": 4739
    },
    {
      "epoch": 0.19644411289319905,
      "grad_norm": 0.4292104244232178,
      "learning_rate": 4.017986655062374e-06,
      "loss": 0.7625,
      "step": 4740
    },
    {
      "epoch": 0.19648555679887272,
      "grad_norm": 0.40588265657424927,
      "learning_rate": 4.017779435534005e-06,
      "loss": 0.7732,
      "step": 4741
    },
    {
      "epoch": 0.1965270007045464,
      "grad_norm": 0.4564080834388733,
      "learning_rate": 4.0175722160056365e-06,
      "loss": 0.6998,
      "step": 4742
    },
    {
      "epoch": 0.19656844461022008,
      "grad_norm": 0.41946300864219666,
      "learning_rate": 4.017364996477268e-06,
      "loss": 0.7659,
      "step": 4743
    },
    {
      "epoch": 0.19660988851589373,
      "grad_norm": 0.40659552812576294,
      "learning_rate": 4.0171577769489e-06,
      "loss": 0.6931,
      "step": 4744
    },
    {
      "epoch": 0.1966513324215674,
      "grad_norm": 0.4200018346309662,
      "learning_rate": 4.0169505574205315e-06,
      "loss": 0.719,
      "step": 4745
    },
    {
      "epoch": 0.19669277632724108,
      "grad_norm": 0.39950311183929443,
      "learning_rate": 4.016743337892164e-06,
      "loss": 0.7173,
      "step": 4746
    },
    {
      "epoch": 0.19673422023291476,
      "grad_norm": 0.465248703956604,
      "learning_rate": 4.016536118363795e-06,
      "loss": 0.7844,
      "step": 4747
    },
    {
      "epoch": 0.1967756641385884,
      "grad_norm": 0.42291945219039917,
      "learning_rate": 4.0163288988354265e-06,
      "loss": 0.7068,
      "step": 4748
    },
    {
      "epoch": 0.19681710804426208,
      "grad_norm": 0.46166113018989563,
      "learning_rate": 4.016121679307059e-06,
      "loss": 0.7686,
      "step": 4749
    },
    {
      "epoch": 0.19685855194993576,
      "grad_norm": 0.4403153955936432,
      "learning_rate": 4.01591445977869e-06,
      "loss": 0.71,
      "step": 4750
    },
    {
      "epoch": 0.19689999585560944,
      "grad_norm": 0.42193281650543213,
      "learning_rate": 4.0157072402503215e-06,
      "loss": 0.7202,
      "step": 4751
    },
    {
      "epoch": 0.19694143976128312,
      "grad_norm": 0.41574758291244507,
      "learning_rate": 4.015500020721953e-06,
      "loss": 0.7223,
      "step": 4752
    },
    {
      "epoch": 0.19698288366695677,
      "grad_norm": 0.42333516478538513,
      "learning_rate": 4.015292801193585e-06,
      "loss": 0.7599,
      "step": 4753
    },
    {
      "epoch": 0.19702432757263044,
      "grad_norm": 0.4132175147533417,
      "learning_rate": 4.0150855816652165e-06,
      "loss": 0.7576,
      "step": 4754
    },
    {
      "epoch": 0.19706577147830412,
      "grad_norm": 0.4266379475593567,
      "learning_rate": 4.014878362136848e-06,
      "loss": 0.7629,
      "step": 4755
    },
    {
      "epoch": 0.1971072153839778,
      "grad_norm": 0.4091418385505676,
      "learning_rate": 4.01467114260848e-06,
      "loss": 0.7119,
      "step": 4756
    },
    {
      "epoch": 0.19714865928965145,
      "grad_norm": 0.4157925844192505,
      "learning_rate": 4.0144639230801115e-06,
      "loss": 0.709,
      "step": 4757
    },
    {
      "epoch": 0.19719010319532512,
      "grad_norm": 0.46335315704345703,
      "learning_rate": 4.014256703551743e-06,
      "loss": 0.7345,
      "step": 4758
    },
    {
      "epoch": 0.1972315471009988,
      "grad_norm": 0.43674609065055847,
      "learning_rate": 4.014049484023374e-06,
      "loss": 0.7651,
      "step": 4759
    },
    {
      "epoch": 0.19727299100667248,
      "grad_norm": 0.4438663721084595,
      "learning_rate": 4.0138422644950065e-06,
      "loss": 0.866,
      "step": 4760
    },
    {
      "epoch": 0.19731443491234613,
      "grad_norm": 0.4615185558795929,
      "learning_rate": 4.013635044966638e-06,
      "loss": 0.7827,
      "step": 4761
    },
    {
      "epoch": 0.1973558788180198,
      "grad_norm": 0.41016241908073425,
      "learning_rate": 4.01342782543827e-06,
      "loss": 0.7964,
      "step": 4762
    },
    {
      "epoch": 0.19739732272369348,
      "grad_norm": 0.39544183015823364,
      "learning_rate": 4.013220605909901e-06,
      "loss": 0.6904,
      "step": 4763
    },
    {
      "epoch": 0.19743876662936716,
      "grad_norm": 0.4233435392379761,
      "learning_rate": 4.013013386381533e-06,
      "loss": 0.7327,
      "step": 4764
    },
    {
      "epoch": 0.19748021053504083,
      "grad_norm": 0.44317662715911865,
      "learning_rate": 4.012806166853165e-06,
      "loss": 0.7527,
      "step": 4765
    },
    {
      "epoch": 0.19752165444071448,
      "grad_norm": 0.4219374656677246,
      "learning_rate": 4.0125989473247965e-06,
      "loss": 0.7488,
      "step": 4766
    },
    {
      "epoch": 0.19756309834638816,
      "grad_norm": 0.43344035744667053,
      "learning_rate": 4.012391727796428e-06,
      "loss": 0.7246,
      "step": 4767
    },
    {
      "epoch": 0.19760454225206184,
      "grad_norm": 0.4006750285625458,
      "learning_rate": 4.012184508268059e-06,
      "loss": 0.7444,
      "step": 4768
    },
    {
      "epoch": 0.19764598615773551,
      "grad_norm": 0.4036215543746948,
      "learning_rate": 4.0119772887396915e-06,
      "loss": 0.6097,
      "step": 4769
    },
    {
      "epoch": 0.19768743006340916,
      "grad_norm": 0.44010189175605774,
      "learning_rate": 4.011770069211323e-06,
      "loss": 0.7466,
      "step": 4770
    },
    {
      "epoch": 0.19772887396908284,
      "grad_norm": 0.4092191755771637,
      "learning_rate": 4.011562849682954e-06,
      "loss": 0.6782,
      "step": 4771
    },
    {
      "epoch": 0.19777031787475652,
      "grad_norm": 0.41537895798683167,
      "learning_rate": 4.011355630154586e-06,
      "loss": 0.6919,
      "step": 4772
    },
    {
      "epoch": 0.1978117617804302,
      "grad_norm": 0.4215163588523865,
      "learning_rate": 4.011148410626218e-06,
      "loss": 0.7454,
      "step": 4773
    },
    {
      "epoch": 0.19785320568610384,
      "grad_norm": 0.4559418857097626,
      "learning_rate": 4.010941191097849e-06,
      "loss": 0.8035,
      "step": 4774
    },
    {
      "epoch": 0.19789464959177752,
      "grad_norm": 0.40059658885002136,
      "learning_rate": 4.010733971569481e-06,
      "loss": 0.74,
      "step": 4775
    },
    {
      "epoch": 0.1979360934974512,
      "grad_norm": 0.4364556074142456,
      "learning_rate": 4.010526752041113e-06,
      "loss": 0.7571,
      "step": 4776
    },
    {
      "epoch": 0.19797753740312488,
      "grad_norm": 0.43708986043930054,
      "learning_rate": 4.010319532512744e-06,
      "loss": 0.7646,
      "step": 4777
    },
    {
      "epoch": 0.19801898130879855,
      "grad_norm": 0.40500515699386597,
      "learning_rate": 4.0101123129843765e-06,
      "loss": 0.7162,
      "step": 4778
    },
    {
      "epoch": 0.1980604252144722,
      "grad_norm": 0.4781336486339569,
      "learning_rate": 4.009905093456007e-06,
      "loss": 0.7686,
      "step": 4779
    },
    {
      "epoch": 0.19810186912014588,
      "grad_norm": 0.40493327379226685,
      "learning_rate": 4.009697873927639e-06,
      "loss": 0.7681,
      "step": 4780
    },
    {
      "epoch": 0.19814331302581956,
      "grad_norm": 0.41445615887641907,
      "learning_rate": 4.009490654399271e-06,
      "loss": 0.7227,
      "step": 4781
    },
    {
      "epoch": 0.19818475693149323,
      "grad_norm": 0.4168505370616913,
      "learning_rate": 4.009283434870903e-06,
      "loss": 0.7578,
      "step": 4782
    },
    {
      "epoch": 0.19822620083716688,
      "grad_norm": 0.4146203398704529,
      "learning_rate": 4.009076215342534e-06,
      "loss": 0.7356,
      "step": 4783
    },
    {
      "epoch": 0.19826764474284056,
      "grad_norm": 0.4667985737323761,
      "learning_rate": 4.008868995814166e-06,
      "loss": 0.7854,
      "step": 4784
    },
    {
      "epoch": 0.19830908864851424,
      "grad_norm": 0.4219963550567627,
      "learning_rate": 4.008661776285798e-06,
      "loss": 0.729,
      "step": 4785
    },
    {
      "epoch": 0.1983505325541879,
      "grad_norm": 0.4178311824798584,
      "learning_rate": 4.008454556757429e-06,
      "loss": 0.6351,
      "step": 4786
    },
    {
      "epoch": 0.1983919764598616,
      "grad_norm": 0.41092249751091003,
      "learning_rate": 4.008247337229061e-06,
      "loss": 0.7224,
      "step": 4787
    },
    {
      "epoch": 0.19843342036553524,
      "grad_norm": 0.40301552414894104,
      "learning_rate": 4.008040117700692e-06,
      "loss": 0.6832,
      "step": 4788
    },
    {
      "epoch": 0.19847486427120892,
      "grad_norm": 0.38286882638931274,
      "learning_rate": 4.007832898172324e-06,
      "loss": 0.6846,
      "step": 4789
    },
    {
      "epoch": 0.1985163081768826,
      "grad_norm": 0.4484480917453766,
      "learning_rate": 4.007625678643956e-06,
      "loss": 0.7231,
      "step": 4790
    },
    {
      "epoch": 0.19855775208255627,
      "grad_norm": 0.3929670453071594,
      "learning_rate": 4.007418459115587e-06,
      "loss": 0.7084,
      "step": 4791
    },
    {
      "epoch": 0.19859919598822992,
      "grad_norm": 0.4283328950405121,
      "learning_rate": 4.007211239587219e-06,
      "loss": 0.7183,
      "step": 4792
    },
    {
      "epoch": 0.1986406398939036,
      "grad_norm": 0.4432357847690582,
      "learning_rate": 4.007004020058851e-06,
      "loss": 0.7385,
      "step": 4793
    },
    {
      "epoch": 0.19868208379957727,
      "grad_norm": 0.42369091510772705,
      "learning_rate": 4.006796800530482e-06,
      "loss": 0.7517,
      "step": 4794
    },
    {
      "epoch": 0.19872352770525095,
      "grad_norm": 0.3779566287994385,
      "learning_rate": 4.0065895810021135e-06,
      "loss": 0.7324,
      "step": 4795
    },
    {
      "epoch": 0.1987649716109246,
      "grad_norm": 0.4781123697757721,
      "learning_rate": 4.006382361473746e-06,
      "loss": 0.7375,
      "step": 4796
    },
    {
      "epoch": 0.19880641551659828,
      "grad_norm": 0.39827635884284973,
      "learning_rate": 4.006175141945377e-06,
      "loss": 0.6938,
      "step": 4797
    },
    {
      "epoch": 0.19884785942227196,
      "grad_norm": 0.4233674705028534,
      "learning_rate": 4.005967922417009e-06,
      "loss": 0.7234,
      "step": 4798
    },
    {
      "epoch": 0.19888930332794563,
      "grad_norm": 0.424721360206604,
      "learning_rate": 4.005760702888641e-06,
      "loss": 0.7705,
      "step": 4799
    },
    {
      "epoch": 0.1989307472336193,
      "grad_norm": 0.4109826982021332,
      "learning_rate": 4.005553483360272e-06,
      "loss": 0.744,
      "step": 4800
    },
    {
      "epoch": 0.19897219113929296,
      "grad_norm": 0.4068213701248169,
      "learning_rate": 4.005346263831904e-06,
      "loss": 0.7263,
      "step": 4801
    },
    {
      "epoch": 0.19901363504496664,
      "grad_norm": 0.44508999586105347,
      "learning_rate": 4.005139044303536e-06,
      "loss": 0.7422,
      "step": 4802
    },
    {
      "epoch": 0.1990550789506403,
      "grad_norm": 0.45127061009407043,
      "learning_rate": 4.004931824775167e-06,
      "loss": 0.7576,
      "step": 4803
    },
    {
      "epoch": 0.199096522856314,
      "grad_norm": 0.4296184778213501,
      "learning_rate": 4.0047246052467985e-06,
      "loss": 0.7273,
      "step": 4804
    },
    {
      "epoch": 0.19913796676198764,
      "grad_norm": 0.41938453912734985,
      "learning_rate": 4.004517385718431e-06,
      "loss": 0.6823,
      "step": 4805
    },
    {
      "epoch": 0.19917941066766132,
      "grad_norm": 0.4013782739639282,
      "learning_rate": 4.004310166190062e-06,
      "loss": 0.7251,
      "step": 4806
    },
    {
      "epoch": 0.199220854573335,
      "grad_norm": 0.45763158798217773,
      "learning_rate": 4.0041029466616935e-06,
      "loss": 0.702,
      "step": 4807
    },
    {
      "epoch": 0.19926229847900867,
      "grad_norm": 0.4125504195690155,
      "learning_rate": 4.003895727133326e-06,
      "loss": 0.7383,
      "step": 4808
    },
    {
      "epoch": 0.19930374238468232,
      "grad_norm": 0.40301206707954407,
      "learning_rate": 4.003688507604957e-06,
      "loss": 0.7234,
      "step": 4809
    },
    {
      "epoch": 0.199345186290356,
      "grad_norm": 0.4162966012954712,
      "learning_rate": 4.0034812880765885e-06,
      "loss": 0.7351,
      "step": 4810
    },
    {
      "epoch": 0.19938663019602967,
      "grad_norm": 0.42247599363327026,
      "learning_rate": 4.00327406854822e-06,
      "loss": 0.7363,
      "step": 4811
    },
    {
      "epoch": 0.19942807410170335,
      "grad_norm": 0.40589475631713867,
      "learning_rate": 4.003066849019852e-06,
      "loss": 0.7085,
      "step": 4812
    },
    {
      "epoch": 0.19946951800737703,
      "grad_norm": 0.43162259459495544,
      "learning_rate": 4.0028596294914835e-06,
      "loss": 0.759,
      "step": 4813
    },
    {
      "epoch": 0.19951096191305068,
      "grad_norm": 0.4187929630279541,
      "learning_rate": 4.002652409963116e-06,
      "loss": 0.682,
      "step": 4814
    },
    {
      "epoch": 0.19955240581872435,
      "grad_norm": 0.4097151756286621,
      "learning_rate": 4.002445190434746e-06,
      "loss": 0.6958,
      "step": 4815
    },
    {
      "epoch": 0.19959384972439803,
      "grad_norm": 0.4315282702445984,
      "learning_rate": 4.0022379709063785e-06,
      "loss": 0.6919,
      "step": 4816
    },
    {
      "epoch": 0.1996352936300717,
      "grad_norm": 0.4187025725841522,
      "learning_rate": 4.002030751378011e-06,
      "loss": 0.7581,
      "step": 4817
    },
    {
      "epoch": 0.19967673753574536,
      "grad_norm": 0.40198373794555664,
      "learning_rate": 4.001823531849642e-06,
      "loss": 0.7001,
      "step": 4818
    },
    {
      "epoch": 0.19971818144141903,
      "grad_norm": 0.42783963680267334,
      "learning_rate": 4.0016163123212735e-06,
      "loss": 0.7449,
      "step": 4819
    },
    {
      "epoch": 0.1997596253470927,
      "grad_norm": 0.4018191397190094,
      "learning_rate": 4.001409092792905e-06,
      "loss": 0.7181,
      "step": 4820
    },
    {
      "epoch": 0.1998010692527664,
      "grad_norm": 0.42642685770988464,
      "learning_rate": 4.001201873264537e-06,
      "loss": 0.6744,
      "step": 4821
    },
    {
      "epoch": 0.19984251315844004,
      "grad_norm": 0.40416115522384644,
      "learning_rate": 4.0009946537361685e-06,
      "loss": 0.7126,
      "step": 4822
    },
    {
      "epoch": 0.19988395706411372,
      "grad_norm": 0.4481668174266815,
      "learning_rate": 4.0007874342078e-06,
      "loss": 0.7468,
      "step": 4823
    },
    {
      "epoch": 0.1999254009697874,
      "grad_norm": 0.4191342890262604,
      "learning_rate": 4.000580214679431e-06,
      "loss": 0.6472,
      "step": 4824
    },
    {
      "epoch": 0.19996684487546107,
      "grad_norm": 0.3984641432762146,
      "learning_rate": 4.0003729951510635e-06,
      "loss": 0.6853,
      "step": 4825
    },
    {
      "epoch": 0.20000828878113475,
      "grad_norm": 0.4206525981426239,
      "learning_rate": 4.000165775622695e-06,
      "loss": 0.6992,
      "step": 4826
    },
    {
      "epoch": 0.2000497326868084,
      "grad_norm": 0.4518857002258301,
      "learning_rate": 3.999958556094326e-06,
      "loss": 0.7422,
      "step": 4827
    },
    {
      "epoch": 0.20009117659248207,
      "grad_norm": 0.4590130150318146,
      "learning_rate": 3.9997513365659585e-06,
      "loss": 0.7649,
      "step": 4828
    },
    {
      "epoch": 0.20013262049815575,
      "grad_norm": 0.42221808433532715,
      "learning_rate": 3.99954411703759e-06,
      "loss": 0.7515,
      "step": 4829
    },
    {
      "epoch": 0.20017406440382943,
      "grad_norm": 0.4404647946357727,
      "learning_rate": 3.999336897509222e-06,
      "loss": 0.7876,
      "step": 4830
    },
    {
      "epoch": 0.20021550830950308,
      "grad_norm": 0.46112871170043945,
      "learning_rate": 3.999129677980853e-06,
      "loss": 0.7463,
      "step": 4831
    },
    {
      "epoch": 0.20025695221517675,
      "grad_norm": 0.42525625228881836,
      "learning_rate": 3.998922458452485e-06,
      "loss": 0.7258,
      "step": 4832
    },
    {
      "epoch": 0.20029839612085043,
      "grad_norm": 0.4054553508758545,
      "learning_rate": 3.998715238924116e-06,
      "loss": 0.6709,
      "step": 4833
    },
    {
      "epoch": 0.2003398400265241,
      "grad_norm": 0.4212915897369385,
      "learning_rate": 3.9985080193957485e-06,
      "loss": 0.7405,
      "step": 4834
    },
    {
      "epoch": 0.20038128393219776,
      "grad_norm": 0.4358137249946594,
      "learning_rate": 3.99830079986738e-06,
      "loss": 0.7234,
      "step": 4835
    },
    {
      "epoch": 0.20042272783787143,
      "grad_norm": 0.42056816816329956,
      "learning_rate": 3.998093580339011e-06,
      "loss": 0.7671,
      "step": 4836
    },
    {
      "epoch": 0.2004641717435451,
      "grad_norm": 0.41150349378585815,
      "learning_rate": 3.9978863608106435e-06,
      "loss": 0.7502,
      "step": 4837
    },
    {
      "epoch": 0.2005056156492188,
      "grad_norm": 0.4460347592830658,
      "learning_rate": 3.997679141282275e-06,
      "loss": 0.6982,
      "step": 4838
    },
    {
      "epoch": 0.20054705955489247,
      "grad_norm": 0.4188644289970398,
      "learning_rate": 3.997471921753906e-06,
      "loss": 0.6716,
      "step": 4839
    },
    {
      "epoch": 0.20058850346056611,
      "grad_norm": 0.40209364891052246,
      "learning_rate": 3.997264702225538e-06,
      "loss": 0.7612,
      "step": 4840
    },
    {
      "epoch": 0.2006299473662398,
      "grad_norm": 0.4218508005142212,
      "learning_rate": 3.99705748269717e-06,
      "loss": 0.7032,
      "step": 4841
    },
    {
      "epoch": 0.20067139127191347,
      "grad_norm": 0.4284106194972992,
      "learning_rate": 3.996850263168801e-06,
      "loss": 0.6982,
      "step": 4842
    },
    {
      "epoch": 0.20071283517758715,
      "grad_norm": 0.3973173499107361,
      "learning_rate": 3.996643043640433e-06,
      "loss": 0.6504,
      "step": 4843
    },
    {
      "epoch": 0.2007542790832608,
      "grad_norm": 0.3873210847377777,
      "learning_rate": 3.996435824112065e-06,
      "loss": 0.6855,
      "step": 4844
    },
    {
      "epoch": 0.20079572298893447,
      "grad_norm": 0.41400542855262756,
      "learning_rate": 3.996228604583696e-06,
      "loss": 0.7173,
      "step": 4845
    },
    {
      "epoch": 0.20083716689460815,
      "grad_norm": 0.38667428493499756,
      "learning_rate": 3.9960213850553285e-06,
      "loss": 0.6799,
      "step": 4846
    },
    {
      "epoch": 0.20087861080028183,
      "grad_norm": 0.422187864780426,
      "learning_rate": 3.995814165526959e-06,
      "loss": 0.7368,
      "step": 4847
    },
    {
      "epoch": 0.2009200547059555,
      "grad_norm": 0.4072923958301544,
      "learning_rate": 3.995606945998591e-06,
      "loss": 0.7122,
      "step": 4848
    },
    {
      "epoch": 0.20096149861162915,
      "grad_norm": 0.43887796998023987,
      "learning_rate": 3.995399726470223e-06,
      "loss": 0.7402,
      "step": 4849
    },
    {
      "epoch": 0.20100294251730283,
      "grad_norm": 0.3875519335269928,
      "learning_rate": 3.995192506941855e-06,
      "loss": 0.6552,
      "step": 4850
    },
    {
      "epoch": 0.2010443864229765,
      "grad_norm": 0.4179123640060425,
      "learning_rate": 3.994985287413486e-06,
      "loss": 0.7043,
      "step": 4851
    },
    {
      "epoch": 0.20108583032865018,
      "grad_norm": 0.46962517499923706,
      "learning_rate": 3.994778067885118e-06,
      "loss": 0.7632,
      "step": 4852
    },
    {
      "epoch": 0.20112727423432383,
      "grad_norm": 0.4453068971633911,
      "learning_rate": 3.99457084835675e-06,
      "loss": 0.6798,
      "step": 4853
    },
    {
      "epoch": 0.2011687181399975,
      "grad_norm": 0.4193930923938751,
      "learning_rate": 3.994363628828381e-06,
      "loss": 0.6865,
      "step": 4854
    },
    {
      "epoch": 0.2012101620456712,
      "grad_norm": 0.4669552445411682,
      "learning_rate": 3.994156409300013e-06,
      "loss": 0.7146,
      "step": 4855
    },
    {
      "epoch": 0.20125160595134486,
      "grad_norm": 0.4321751594543457,
      "learning_rate": 3.993949189771644e-06,
      "loss": 0.7151,
      "step": 4856
    },
    {
      "epoch": 0.2012930498570185,
      "grad_norm": 0.44281330704689026,
      "learning_rate": 3.993741970243276e-06,
      "loss": 0.7314,
      "step": 4857
    },
    {
      "epoch": 0.2013344937626922,
      "grad_norm": 0.47185227274894714,
      "learning_rate": 3.993534750714908e-06,
      "loss": 0.7406,
      "step": 4858
    },
    {
      "epoch": 0.20137593766836587,
      "grad_norm": 0.48666802048683167,
      "learning_rate": 3.993327531186539e-06,
      "loss": 0.7275,
      "step": 4859
    },
    {
      "epoch": 0.20141738157403954,
      "grad_norm": 0.39160531759262085,
      "learning_rate": 3.993120311658171e-06,
      "loss": 0.7263,
      "step": 4860
    },
    {
      "epoch": 0.20145882547971322,
      "grad_norm": 0.43907567858695984,
      "learning_rate": 3.992913092129803e-06,
      "loss": 0.7865,
      "step": 4861
    },
    {
      "epoch": 0.20150026938538687,
      "grad_norm": 0.4136238992214203,
      "learning_rate": 3.992705872601434e-06,
      "loss": 0.6722,
      "step": 4862
    },
    {
      "epoch": 0.20154171329106055,
      "grad_norm": 0.4332010746002197,
      "learning_rate": 3.9924986530730655e-06,
      "loss": 0.719,
      "step": 4863
    },
    {
      "epoch": 0.20158315719673423,
      "grad_norm": 0.43985840678215027,
      "learning_rate": 3.992291433544698e-06,
      "loss": 0.7703,
      "step": 4864
    },
    {
      "epoch": 0.2016246011024079,
      "grad_norm": 0.401887983083725,
      "learning_rate": 3.992084214016329e-06,
      "loss": 0.7217,
      "step": 4865
    },
    {
      "epoch": 0.20166604500808155,
      "grad_norm": 0.4438116252422333,
      "learning_rate": 3.991876994487961e-06,
      "loss": 0.7651,
      "step": 4866
    },
    {
      "epoch": 0.20170748891375523,
      "grad_norm": 0.4155856966972351,
      "learning_rate": 3.991669774959593e-06,
      "loss": 0.7063,
      "step": 4867
    },
    {
      "epoch": 0.2017489328194289,
      "grad_norm": 0.42813897132873535,
      "learning_rate": 3.991462555431224e-06,
      "loss": 0.6705,
      "step": 4868
    },
    {
      "epoch": 0.20179037672510258,
      "grad_norm": 0.4280458986759186,
      "learning_rate": 3.991255335902856e-06,
      "loss": 0.7317,
      "step": 4869
    },
    {
      "epoch": 0.20183182063077623,
      "grad_norm": 0.4266197383403778,
      "learning_rate": 3.991048116374488e-06,
      "loss": 0.7061,
      "step": 4870
    },
    {
      "epoch": 0.2018732645364499,
      "grad_norm": 0.4213167726993561,
      "learning_rate": 3.990840896846119e-06,
      "loss": 0.7058,
      "step": 4871
    },
    {
      "epoch": 0.20191470844212359,
      "grad_norm": 0.4270082712173462,
      "learning_rate": 3.9906336773177505e-06,
      "loss": 0.7188,
      "step": 4872
    },
    {
      "epoch": 0.20195615234779726,
      "grad_norm": 0.4213247001171112,
      "learning_rate": 3.990426457789383e-06,
      "loss": 0.7118,
      "step": 4873
    },
    {
      "epoch": 0.20199759625347094,
      "grad_norm": 0.4341457784175873,
      "learning_rate": 3.990219238261014e-06,
      "loss": 0.7256,
      "step": 4874
    },
    {
      "epoch": 0.2020390401591446,
      "grad_norm": 0.40950748324394226,
      "learning_rate": 3.9900120187326455e-06,
      "loss": 0.7239,
      "step": 4875
    },
    {
      "epoch": 0.20208048406481827,
      "grad_norm": 0.4047373831272125,
      "learning_rate": 3.989804799204277e-06,
      "loss": 0.7344,
      "step": 4876
    },
    {
      "epoch": 0.20212192797049194,
      "grad_norm": 0.4294915497303009,
      "learning_rate": 3.989597579675909e-06,
      "loss": 0.7649,
      "step": 4877
    },
    {
      "epoch": 0.20216337187616562,
      "grad_norm": 0.4314100444316864,
      "learning_rate": 3.9893903601475405e-06,
      "loss": 0.7839,
      "step": 4878
    },
    {
      "epoch": 0.20220481578183927,
      "grad_norm": 0.41535383462905884,
      "learning_rate": 3.989183140619172e-06,
      "loss": 0.7385,
      "step": 4879
    },
    {
      "epoch": 0.20224625968751295,
      "grad_norm": 0.4173435866832733,
      "learning_rate": 3.988975921090804e-06,
      "loss": 0.7635,
      "step": 4880
    },
    {
      "epoch": 0.20228770359318662,
      "grad_norm": 0.4282142221927643,
      "learning_rate": 3.9887687015624355e-06,
      "loss": 0.7429,
      "step": 4881
    },
    {
      "epoch": 0.2023291474988603,
      "grad_norm": 0.45594847202301025,
      "learning_rate": 3.988561482034068e-06,
      "loss": 0.6826,
      "step": 4882
    },
    {
      "epoch": 0.20237059140453395,
      "grad_norm": 0.395620733499527,
      "learning_rate": 3.988354262505698e-06,
      "loss": 0.6853,
      "step": 4883
    },
    {
      "epoch": 0.20241203531020763,
      "grad_norm": 0.4690057039260864,
      "learning_rate": 3.9881470429773305e-06,
      "loss": 0.7943,
      "step": 4884
    },
    {
      "epoch": 0.2024534792158813,
      "grad_norm": 0.463143527507782,
      "learning_rate": 3.987939823448962e-06,
      "loss": 0.7324,
      "step": 4885
    },
    {
      "epoch": 0.20249492312155498,
      "grad_norm": 0.43740883469581604,
      "learning_rate": 3.987732603920594e-06,
      "loss": 0.6914,
      "step": 4886
    },
    {
      "epoch": 0.20253636702722866,
      "grad_norm": 0.4480329751968384,
      "learning_rate": 3.9875253843922255e-06,
      "loss": 0.7478,
      "step": 4887
    },
    {
      "epoch": 0.2025778109329023,
      "grad_norm": 0.4253040850162506,
      "learning_rate": 3.987318164863857e-06,
      "loss": 0.7322,
      "step": 4888
    },
    {
      "epoch": 0.20261925483857599,
      "grad_norm": 0.42684316635131836,
      "learning_rate": 3.987110945335489e-06,
      "loss": 0.7192,
      "step": 4889
    },
    {
      "epoch": 0.20266069874424966,
      "grad_norm": 0.450831800699234,
      "learning_rate": 3.9869037258071205e-06,
      "loss": 0.7317,
      "step": 4890
    },
    {
      "epoch": 0.20270214264992334,
      "grad_norm": 0.44185078144073486,
      "learning_rate": 3.986696506278752e-06,
      "loss": 0.7506,
      "step": 4891
    },
    {
      "epoch": 0.202743586555597,
      "grad_norm": 0.40829259157180786,
      "learning_rate": 3.986489286750383e-06,
      "loss": 0.7305,
      "step": 4892
    },
    {
      "epoch": 0.20278503046127067,
      "grad_norm": 0.4085010886192322,
      "learning_rate": 3.9862820672220155e-06,
      "loss": 0.7202,
      "step": 4893
    },
    {
      "epoch": 0.20282647436694434,
      "grad_norm": 0.4650880992412567,
      "learning_rate": 3.986074847693647e-06,
      "loss": 0.8198,
      "step": 4894
    },
    {
      "epoch": 0.20286791827261802,
      "grad_norm": 0.40678155422210693,
      "learning_rate": 3.985867628165278e-06,
      "loss": 0.6752,
      "step": 4895
    },
    {
      "epoch": 0.20290936217829167,
      "grad_norm": 0.4065786898136139,
      "learning_rate": 3.9856604086369105e-06,
      "loss": 0.6946,
      "step": 4896
    },
    {
      "epoch": 0.20295080608396535,
      "grad_norm": 0.4177252948284149,
      "learning_rate": 3.985453189108542e-06,
      "loss": 0.7271,
      "step": 4897
    },
    {
      "epoch": 0.20299224998963902,
      "grad_norm": 0.39959901571273804,
      "learning_rate": 3.985245969580174e-06,
      "loss": 0.6813,
      "step": 4898
    },
    {
      "epoch": 0.2030336938953127,
      "grad_norm": 0.466714084148407,
      "learning_rate": 3.985038750051805e-06,
      "loss": 0.7603,
      "step": 4899
    },
    {
      "epoch": 0.20307513780098638,
      "grad_norm": 0.40733465552330017,
      "learning_rate": 3.984831530523437e-06,
      "loss": 0.7189,
      "step": 4900
    },
    {
      "epoch": 0.20311658170666003,
      "grad_norm": 0.43598905205726624,
      "learning_rate": 3.984624310995068e-06,
      "loss": 0.7666,
      "step": 4901
    },
    {
      "epoch": 0.2031580256123337,
      "grad_norm": 0.4384497106075287,
      "learning_rate": 3.9844170914667005e-06,
      "loss": 0.7441,
      "step": 4902
    },
    {
      "epoch": 0.20319946951800738,
      "grad_norm": 0.39899253845214844,
      "learning_rate": 3.984209871938332e-06,
      "loss": 0.7247,
      "step": 4903
    },
    {
      "epoch": 0.20324091342368106,
      "grad_norm": 0.429982990026474,
      "learning_rate": 3.984002652409963e-06,
      "loss": 0.6615,
      "step": 4904
    },
    {
      "epoch": 0.2032823573293547,
      "grad_norm": 0.41308340430259705,
      "learning_rate": 3.9837954328815955e-06,
      "loss": 0.7949,
      "step": 4905
    },
    {
      "epoch": 0.20332380123502838,
      "grad_norm": 0.4428352415561676,
      "learning_rate": 3.983588213353227e-06,
      "loss": 0.7732,
      "step": 4906
    },
    {
      "epoch": 0.20336524514070206,
      "grad_norm": 0.45400720834732056,
      "learning_rate": 3.983380993824858e-06,
      "loss": 0.7271,
      "step": 4907
    },
    {
      "epoch": 0.20340668904637574,
      "grad_norm": 0.41853067278862,
      "learning_rate": 3.98317377429649e-06,
      "loss": 0.7192,
      "step": 4908
    },
    {
      "epoch": 0.2034481329520494,
      "grad_norm": 0.42906585335731506,
      "learning_rate": 3.982966554768122e-06,
      "loss": 0.761,
      "step": 4909
    },
    {
      "epoch": 0.20348957685772306,
      "grad_norm": 0.40976741909980774,
      "learning_rate": 3.982759335239753e-06,
      "loss": 0.7561,
      "step": 4910
    },
    {
      "epoch": 0.20353102076339674,
      "grad_norm": 0.4233849346637726,
      "learning_rate": 3.982552115711385e-06,
      "loss": 0.7461,
      "step": 4911
    },
    {
      "epoch": 0.20357246466907042,
      "grad_norm": 0.4079173505306244,
      "learning_rate": 3.982344896183016e-06,
      "loss": 0.6841,
      "step": 4912
    },
    {
      "epoch": 0.2036139085747441,
      "grad_norm": 0.3959016799926758,
      "learning_rate": 3.982137676654648e-06,
      "loss": 0.7144,
      "step": 4913
    },
    {
      "epoch": 0.20365535248041775,
      "grad_norm": 0.41608691215515137,
      "learning_rate": 3.9819304571262805e-06,
      "loss": 0.7065,
      "step": 4914
    },
    {
      "epoch": 0.20369679638609142,
      "grad_norm": 0.42271432280540466,
      "learning_rate": 3.981723237597911e-06,
      "loss": 0.7539,
      "step": 4915
    },
    {
      "epoch": 0.2037382402917651,
      "grad_norm": 0.43617281317710876,
      "learning_rate": 3.981516018069543e-06,
      "loss": 0.7297,
      "step": 4916
    },
    {
      "epoch": 0.20377968419743878,
      "grad_norm": 0.4270938038825989,
      "learning_rate": 3.981308798541175e-06,
      "loss": 0.7177,
      "step": 4917
    },
    {
      "epoch": 0.20382112810311243,
      "grad_norm": 0.407021701335907,
      "learning_rate": 3.981101579012807e-06,
      "loss": 0.7134,
      "step": 4918
    },
    {
      "epoch": 0.2038625720087861,
      "grad_norm": 0.41759172081947327,
      "learning_rate": 3.980894359484438e-06,
      "loss": 0.7058,
      "step": 4919
    },
    {
      "epoch": 0.20390401591445978,
      "grad_norm": 0.43455517292022705,
      "learning_rate": 3.98068713995607e-06,
      "loss": 0.7368,
      "step": 4920
    },
    {
      "epoch": 0.20394545982013346,
      "grad_norm": 0.4410875141620636,
      "learning_rate": 3.980479920427702e-06,
      "loss": 0.7384,
      "step": 4921
    },
    {
      "epoch": 0.20398690372580713,
      "grad_norm": 0.4648277163505554,
      "learning_rate": 3.980272700899333e-06,
      "loss": 0.6777,
      "step": 4922
    },
    {
      "epoch": 0.20402834763148078,
      "grad_norm": 0.43786296248435974,
      "learning_rate": 3.980065481370965e-06,
      "loss": 0.7219,
      "step": 4923
    },
    {
      "epoch": 0.20406979153715446,
      "grad_norm": 0.42219990491867065,
      "learning_rate": 3.979858261842596e-06,
      "loss": 0.7212,
      "step": 4924
    },
    {
      "epoch": 0.20411123544282814,
      "grad_norm": 0.42502540349960327,
      "learning_rate": 3.979651042314228e-06,
      "loss": 0.7229,
      "step": 4925
    },
    {
      "epoch": 0.20415267934850181,
      "grad_norm": 0.4341442286968231,
      "learning_rate": 3.97944382278586e-06,
      "loss": 0.7133,
      "step": 4926
    },
    {
      "epoch": 0.20419412325417546,
      "grad_norm": 0.43252792954444885,
      "learning_rate": 3.979236603257491e-06,
      "loss": 0.7561,
      "step": 4927
    },
    {
      "epoch": 0.20423556715984914,
      "grad_norm": 0.41907423734664917,
      "learning_rate": 3.9790293837291225e-06,
      "loss": 0.6711,
      "step": 4928
    },
    {
      "epoch": 0.20427701106552282,
      "grad_norm": 0.4227280914783478,
      "learning_rate": 3.978822164200755e-06,
      "loss": 0.6853,
      "step": 4929
    },
    {
      "epoch": 0.2043184549711965,
      "grad_norm": 0.4302346110343933,
      "learning_rate": 3.978614944672386e-06,
      "loss": 0.6901,
      "step": 4930
    },
    {
      "epoch": 0.20435989887687014,
      "grad_norm": 0.42137134075164795,
      "learning_rate": 3.9784077251440175e-06,
      "loss": 0.6948,
      "step": 4931
    },
    {
      "epoch": 0.20440134278254382,
      "grad_norm": 0.4396488666534424,
      "learning_rate": 3.97820050561565e-06,
      "loss": 0.7751,
      "step": 4932
    },
    {
      "epoch": 0.2044427866882175,
      "grad_norm": 0.4327951967716217,
      "learning_rate": 3.977993286087281e-06,
      "loss": 0.8035,
      "step": 4933
    },
    {
      "epoch": 0.20448423059389118,
      "grad_norm": 0.42528650164604187,
      "learning_rate": 3.977786066558913e-06,
      "loss": 0.7483,
      "step": 4934
    },
    {
      "epoch": 0.20452567449956485,
      "grad_norm": 0.39143651723861694,
      "learning_rate": 3.977578847030545e-06,
      "loss": 0.6656,
      "step": 4935
    },
    {
      "epoch": 0.2045671184052385,
      "grad_norm": 0.46127450466156006,
      "learning_rate": 3.977371627502176e-06,
      "loss": 0.7507,
      "step": 4936
    },
    {
      "epoch": 0.20460856231091218,
      "grad_norm": 0.4289734363555908,
      "learning_rate": 3.9771644079738075e-06,
      "loss": 0.7341,
      "step": 4937
    },
    {
      "epoch": 0.20465000621658586,
      "grad_norm": 0.45152151584625244,
      "learning_rate": 3.97695718844544e-06,
      "loss": 0.7502,
      "step": 4938
    },
    {
      "epoch": 0.20469145012225953,
      "grad_norm": 0.421102374792099,
      "learning_rate": 3.976749968917071e-06,
      "loss": 0.7339,
      "step": 4939
    },
    {
      "epoch": 0.20473289402793318,
      "grad_norm": 0.4111858308315277,
      "learning_rate": 3.9765427493887025e-06,
      "loss": 0.6887,
      "step": 4940
    },
    {
      "epoch": 0.20477433793360686,
      "grad_norm": 0.4935225248336792,
      "learning_rate": 3.976335529860335e-06,
      "loss": 0.7375,
      "step": 4941
    },
    {
      "epoch": 0.20481578183928054,
      "grad_norm": 0.4377748370170593,
      "learning_rate": 3.976128310331966e-06,
      "loss": 0.762,
      "step": 4942
    },
    {
      "epoch": 0.2048572257449542,
      "grad_norm": 0.3995705842971802,
      "learning_rate": 3.9759210908035975e-06,
      "loss": 0.7666,
      "step": 4943
    },
    {
      "epoch": 0.20489866965062786,
      "grad_norm": 0.3820174038410187,
      "learning_rate": 3.975713871275229e-06,
      "loss": 0.7216,
      "step": 4944
    },
    {
      "epoch": 0.20494011355630154,
      "grad_norm": 0.43117663264274597,
      "learning_rate": 3.975506651746861e-06,
      "loss": 0.6843,
      "step": 4945
    },
    {
      "epoch": 0.20498155746197522,
      "grad_norm": 0.445446640253067,
      "learning_rate": 3.9752994322184925e-06,
      "loss": 0.7305,
      "step": 4946
    },
    {
      "epoch": 0.2050230013676489,
      "grad_norm": 0.47389575839042664,
      "learning_rate": 3.975092212690124e-06,
      "loss": 0.7622,
      "step": 4947
    },
    {
      "epoch": 0.20506444527332257,
      "grad_norm": 0.42647477984428406,
      "learning_rate": 3.974884993161756e-06,
      "loss": 0.7051,
      "step": 4948
    },
    {
      "epoch": 0.20510588917899622,
      "grad_norm": 0.4340187907218933,
      "learning_rate": 3.9746777736333875e-06,
      "loss": 0.6813,
      "step": 4949
    },
    {
      "epoch": 0.2051473330846699,
      "grad_norm": 0.3840288817882538,
      "learning_rate": 3.97447055410502e-06,
      "loss": 0.7229,
      "step": 4950
    },
    {
      "epoch": 0.20518877699034357,
      "grad_norm": 0.4030836820602417,
      "learning_rate": 3.974263334576651e-06,
      "loss": 0.7122,
      "step": 4951
    },
    {
      "epoch": 0.20523022089601725,
      "grad_norm": 0.4452937841415405,
      "learning_rate": 3.9740561150482825e-06,
      "loss": 0.74,
      "step": 4952
    },
    {
      "epoch": 0.2052716648016909,
      "grad_norm": 0.448479026556015,
      "learning_rate": 3.973848895519914e-06,
      "loss": 0.813,
      "step": 4953
    },
    {
      "epoch": 0.20531310870736458,
      "grad_norm": 0.4458146095275879,
      "learning_rate": 3.973641675991546e-06,
      "loss": 0.7532,
      "step": 4954
    },
    {
      "epoch": 0.20535455261303825,
      "grad_norm": 0.4272207021713257,
      "learning_rate": 3.9734344564631775e-06,
      "loss": 0.7322,
      "step": 4955
    },
    {
      "epoch": 0.20539599651871193,
      "grad_norm": 0.4285518527030945,
      "learning_rate": 3.973227236934809e-06,
      "loss": 0.7532,
      "step": 4956
    },
    {
      "epoch": 0.20543744042438558,
      "grad_norm": 0.40414348244667053,
      "learning_rate": 3.973020017406441e-06,
      "loss": 0.7859,
      "step": 4957
    },
    {
      "epoch": 0.20547888433005926,
      "grad_norm": 0.4786165654659271,
      "learning_rate": 3.9728127978780725e-06,
      "loss": 0.7493,
      "step": 4958
    },
    {
      "epoch": 0.20552032823573294,
      "grad_norm": 0.42078104615211487,
      "learning_rate": 3.972605578349704e-06,
      "loss": 0.7366,
      "step": 4959
    },
    {
      "epoch": 0.2055617721414066,
      "grad_norm": 0.43938061594963074,
      "learning_rate": 3.972398358821335e-06,
      "loss": 0.7202,
      "step": 4960
    },
    {
      "epoch": 0.2056032160470803,
      "grad_norm": 0.45234769582748413,
      "learning_rate": 3.9721911392929675e-06,
      "loss": 0.7113,
      "step": 4961
    },
    {
      "epoch": 0.20564465995275394,
      "grad_norm": 0.4003274738788605,
      "learning_rate": 3.971983919764599e-06,
      "loss": 0.6785,
      "step": 4962
    },
    {
      "epoch": 0.20568610385842762,
      "grad_norm": 0.4167240858078003,
      "learning_rate": 3.97177670023623e-06,
      "loss": 0.762,
      "step": 4963
    },
    {
      "epoch": 0.2057275477641013,
      "grad_norm": 0.40504613518714905,
      "learning_rate": 3.971569480707862e-06,
      "loss": 0.6776,
      "step": 4964
    },
    {
      "epoch": 0.20576899166977497,
      "grad_norm": 0.43072494864463806,
      "learning_rate": 3.971362261179494e-06,
      "loss": 0.7222,
      "step": 4965
    },
    {
      "epoch": 0.20581043557544862,
      "grad_norm": 0.44922390580177307,
      "learning_rate": 3.971155041651126e-06,
      "loss": 0.7678,
      "step": 4966
    },
    {
      "epoch": 0.2058518794811223,
      "grad_norm": 0.4145321846008301,
      "learning_rate": 3.970947822122757e-06,
      "loss": 0.7422,
      "step": 4967
    },
    {
      "epoch": 0.20589332338679597,
      "grad_norm": 0.3864027261734009,
      "learning_rate": 3.970740602594389e-06,
      "loss": 0.6833,
      "step": 4968
    },
    {
      "epoch": 0.20593476729246965,
      "grad_norm": 0.40929099917411804,
      "learning_rate": 3.97053338306602e-06,
      "loss": 0.7029,
      "step": 4969
    },
    {
      "epoch": 0.2059762111981433,
      "grad_norm": 0.44474220275878906,
      "learning_rate": 3.9703261635376525e-06,
      "loss": 0.729,
      "step": 4970
    },
    {
      "epoch": 0.20601765510381698,
      "grad_norm": 0.41776105761528015,
      "learning_rate": 3.970118944009284e-06,
      "loss": 0.7463,
      "step": 4971
    },
    {
      "epoch": 0.20605909900949065,
      "grad_norm": 0.443928986787796,
      "learning_rate": 3.969911724480915e-06,
      "loss": 0.7371,
      "step": 4972
    },
    {
      "epoch": 0.20610054291516433,
      "grad_norm": 0.42484357953071594,
      "learning_rate": 3.969704504952547e-06,
      "loss": 0.6926,
      "step": 4973
    },
    {
      "epoch": 0.206141986820838,
      "grad_norm": 0.4135028123855591,
      "learning_rate": 3.969497285424179e-06,
      "loss": 0.6887,
      "step": 4974
    },
    {
      "epoch": 0.20618343072651166,
      "grad_norm": 0.4275168478488922,
      "learning_rate": 3.96929006589581e-06,
      "loss": 0.7393,
      "step": 4975
    },
    {
      "epoch": 0.20622487463218533,
      "grad_norm": 0.423520028591156,
      "learning_rate": 3.969082846367442e-06,
      "loss": 0.7344,
      "step": 4976
    },
    {
      "epoch": 0.206266318537859,
      "grad_norm": 0.3947014808654785,
      "learning_rate": 3.968875626839074e-06,
      "loss": 0.8062,
      "step": 4977
    },
    {
      "epoch": 0.2063077624435327,
      "grad_norm": 0.4072122871875763,
      "learning_rate": 3.968668407310705e-06,
      "loss": 0.7012,
      "step": 4978
    },
    {
      "epoch": 0.20634920634920634,
      "grad_norm": 0.4207821786403656,
      "learning_rate": 3.968461187782337e-06,
      "loss": 0.7273,
      "step": 4979
    },
    {
      "epoch": 0.20639065025488001,
      "grad_norm": 0.43246567249298096,
      "learning_rate": 3.968253968253968e-06,
      "loss": 0.7186,
      "step": 4980
    },
    {
      "epoch": 0.2064320941605537,
      "grad_norm": 0.4550793766975403,
      "learning_rate": 3.9680467487256e-06,
      "loss": 0.7119,
      "step": 4981
    },
    {
      "epoch": 0.20647353806622737,
      "grad_norm": 0.4283657670021057,
      "learning_rate": 3.9678395291972325e-06,
      "loss": 0.7336,
      "step": 4982
    },
    {
      "epoch": 0.20651498197190102,
      "grad_norm": 0.4309161901473999,
      "learning_rate": 3.967632309668863e-06,
      "loss": 0.7582,
      "step": 4983
    },
    {
      "epoch": 0.2065564258775747,
      "grad_norm": 0.4237719178199768,
      "learning_rate": 3.967425090140495e-06,
      "loss": 0.6893,
      "step": 4984
    },
    {
      "epoch": 0.20659786978324837,
      "grad_norm": 0.4184347689151764,
      "learning_rate": 3.967217870612127e-06,
      "loss": 0.6606,
      "step": 4985
    },
    {
      "epoch": 0.20663931368892205,
      "grad_norm": 0.4322426915168762,
      "learning_rate": 3.967010651083759e-06,
      "loss": 0.7224,
      "step": 4986
    },
    {
      "epoch": 0.20668075759459573,
      "grad_norm": 0.42716529965400696,
      "learning_rate": 3.96680343155539e-06,
      "loss": 0.7131,
      "step": 4987
    },
    {
      "epoch": 0.20672220150026938,
      "grad_norm": 0.4076707363128662,
      "learning_rate": 3.966596212027022e-06,
      "loss": 0.7673,
      "step": 4988
    },
    {
      "epoch": 0.20676364540594305,
      "grad_norm": 0.40031224489212036,
      "learning_rate": 3.966388992498653e-06,
      "loss": 0.7062,
      "step": 4989
    },
    {
      "epoch": 0.20680508931161673,
      "grad_norm": 0.41259315609931946,
      "learning_rate": 3.966181772970285e-06,
      "loss": 0.655,
      "step": 4990
    },
    {
      "epoch": 0.2068465332172904,
      "grad_norm": 0.40943002700805664,
      "learning_rate": 3.965974553441917e-06,
      "loss": 0.7766,
      "step": 4991
    },
    {
      "epoch": 0.20688797712296406,
      "grad_norm": 0.38084378838539124,
      "learning_rate": 3.965767333913548e-06,
      "loss": 0.7024,
      "step": 4992
    },
    {
      "epoch": 0.20692942102863773,
      "grad_norm": 0.45761755108833313,
      "learning_rate": 3.96556011438518e-06,
      "loss": 0.7097,
      "step": 4993
    },
    {
      "epoch": 0.2069708649343114,
      "grad_norm": 0.4409826993942261,
      "learning_rate": 3.965352894856812e-06,
      "loss": 0.7443,
      "step": 4994
    },
    {
      "epoch": 0.2070123088399851,
      "grad_norm": 0.41890180110931396,
      "learning_rate": 3.965145675328443e-06,
      "loss": 0.6526,
      "step": 4995
    },
    {
      "epoch": 0.20705375274565876,
      "grad_norm": 0.4082462787628174,
      "learning_rate": 3.9649384558000745e-06,
      "loss": 0.7402,
      "step": 4996
    },
    {
      "epoch": 0.2070951966513324,
      "grad_norm": 0.4276891052722931,
      "learning_rate": 3.964731236271707e-06,
      "loss": 0.7399,
      "step": 4997
    },
    {
      "epoch": 0.2071366405570061,
      "grad_norm": 0.420146107673645,
      "learning_rate": 3.964524016743338e-06,
      "loss": 0.6967,
      "step": 4998
    },
    {
      "epoch": 0.20717808446267977,
      "grad_norm": 0.47646698355674744,
      "learning_rate": 3.9643167972149695e-06,
      "loss": 0.7947,
      "step": 4999
    },
    {
      "epoch": 0.20721952836835345,
      "grad_norm": 0.4139586389064789,
      "learning_rate": 3.964109577686602e-06,
      "loss": 0.6824,
      "step": 5000
    },
    {
      "epoch": 0.2072609722740271,
      "grad_norm": 0.3895176947116852,
      "learning_rate": 3.963902358158233e-06,
      "loss": 0.6851,
      "step": 5001
    },
    {
      "epoch": 0.20730241617970077,
      "grad_norm": 0.4015359878540039,
      "learning_rate": 3.963695138629865e-06,
      "loss": 0.7354,
      "step": 5002
    },
    {
      "epoch": 0.20734386008537445,
      "grad_norm": 0.3724995255470276,
      "learning_rate": 3.963487919101497e-06,
      "loss": 0.672,
      "step": 5003
    },
    {
      "epoch": 0.20738530399104813,
      "grad_norm": 0.4264490306377411,
      "learning_rate": 3.963280699573128e-06,
      "loss": 0.7178,
      "step": 5004
    },
    {
      "epoch": 0.20742674789672177,
      "grad_norm": 0.4242999255657196,
      "learning_rate": 3.9630734800447595e-06,
      "loss": 0.686,
      "step": 5005
    },
    {
      "epoch": 0.20746819180239545,
      "grad_norm": 0.4459003806114197,
      "learning_rate": 3.962866260516392e-06,
      "loss": 0.7822,
      "step": 5006
    },
    {
      "epoch": 0.20750963570806913,
      "grad_norm": 0.4635668992996216,
      "learning_rate": 3.962659040988023e-06,
      "loss": 0.6753,
      "step": 5007
    },
    {
      "epoch": 0.2075510796137428,
      "grad_norm": 0.3952350616455078,
      "learning_rate": 3.9624518214596545e-06,
      "loss": 0.6838,
      "step": 5008
    },
    {
      "epoch": 0.20759252351941648,
      "grad_norm": 0.44736242294311523,
      "learning_rate": 3.962244601931287e-06,
      "loss": 0.7402,
      "step": 5009
    },
    {
      "epoch": 0.20763396742509013,
      "grad_norm": 0.45444953441619873,
      "learning_rate": 3.962037382402918e-06,
      "loss": 0.7052,
      "step": 5010
    },
    {
      "epoch": 0.2076754113307638,
      "grad_norm": 0.41435879468917847,
      "learning_rate": 3.9618301628745495e-06,
      "loss": 0.7153,
      "step": 5011
    },
    {
      "epoch": 0.2077168552364375,
      "grad_norm": 0.3875054121017456,
      "learning_rate": 3.961622943346181e-06,
      "loss": 0.6581,
      "step": 5012
    },
    {
      "epoch": 0.20775829914211116,
      "grad_norm": 0.45671743154525757,
      "learning_rate": 3.961415723817813e-06,
      "loss": 0.7356,
      "step": 5013
    },
    {
      "epoch": 0.2077997430477848,
      "grad_norm": 0.4382782578468323,
      "learning_rate": 3.9612085042894445e-06,
      "loss": 0.6953,
      "step": 5014
    },
    {
      "epoch": 0.2078411869534585,
      "grad_norm": 0.4314959943294525,
      "learning_rate": 3.961001284761076e-06,
      "loss": 0.6853,
      "step": 5015
    },
    {
      "epoch": 0.20788263085913217,
      "grad_norm": 0.40663015842437744,
      "learning_rate": 3.960794065232707e-06,
      "loss": 0.7003,
      "step": 5016
    },
    {
      "epoch": 0.20792407476480584,
      "grad_norm": 0.4372977614402771,
      "learning_rate": 3.9605868457043395e-06,
      "loss": 0.7933,
      "step": 5017
    },
    {
      "epoch": 0.2079655186704795,
      "grad_norm": 0.42495372891426086,
      "learning_rate": 3.960379626175972e-06,
      "loss": 0.6821,
      "step": 5018
    },
    {
      "epoch": 0.20800696257615317,
      "grad_norm": 0.4128490388393402,
      "learning_rate": 3.960172406647603e-06,
      "loss": 0.7156,
      "step": 5019
    },
    {
      "epoch": 0.20804840648182685,
      "grad_norm": 0.43258345127105713,
      "learning_rate": 3.9599651871192345e-06,
      "loss": 0.6819,
      "step": 5020
    },
    {
      "epoch": 0.20808985038750052,
      "grad_norm": 0.43986499309539795,
      "learning_rate": 3.959757967590866e-06,
      "loss": 0.7612,
      "step": 5021
    },
    {
      "epoch": 0.2081312942931742,
      "grad_norm": 0.4230652451515198,
      "learning_rate": 3.959550748062498e-06,
      "loss": 0.7427,
      "step": 5022
    },
    {
      "epoch": 0.20817273819884785,
      "grad_norm": 0.420604944229126,
      "learning_rate": 3.9593435285341295e-06,
      "loss": 0.7271,
      "step": 5023
    },
    {
      "epoch": 0.20821418210452153,
      "grad_norm": 0.4139954149723053,
      "learning_rate": 3.959136309005761e-06,
      "loss": 0.6941,
      "step": 5024
    },
    {
      "epoch": 0.2082556260101952,
      "grad_norm": 0.42601391673088074,
      "learning_rate": 3.958929089477392e-06,
      "loss": 0.7332,
      "step": 5025
    },
    {
      "epoch": 0.20829706991586888,
      "grad_norm": 0.44510379433631897,
      "learning_rate": 3.9587218699490245e-06,
      "loss": 0.7406,
      "step": 5026
    },
    {
      "epoch": 0.20833851382154253,
      "grad_norm": 0.4313295781612396,
      "learning_rate": 3.958514650420656e-06,
      "loss": 0.6915,
      "step": 5027
    },
    {
      "epoch": 0.2083799577272162,
      "grad_norm": 0.4395928680896759,
      "learning_rate": 3.958307430892287e-06,
      "loss": 0.7025,
      "step": 5028
    },
    {
      "epoch": 0.20842140163288989,
      "grad_norm": 0.40598583221435547,
      "learning_rate": 3.9581002113639195e-06,
      "loss": 0.7339,
      "step": 5029
    },
    {
      "epoch": 0.20846284553856356,
      "grad_norm": 0.404128760099411,
      "learning_rate": 3.957892991835551e-06,
      "loss": 0.6766,
      "step": 5030
    },
    {
      "epoch": 0.2085042894442372,
      "grad_norm": 0.3969388008117676,
      "learning_rate": 3.957685772307182e-06,
      "loss": 0.6925,
      "step": 5031
    },
    {
      "epoch": 0.2085457333499109,
      "grad_norm": 0.40514540672302246,
      "learning_rate": 3.957478552778814e-06,
      "loss": 0.6978,
      "step": 5032
    },
    {
      "epoch": 0.20858717725558457,
      "grad_norm": 0.38450825214385986,
      "learning_rate": 3.957271333250446e-06,
      "loss": 0.7083,
      "step": 5033
    },
    {
      "epoch": 0.20862862116125824,
      "grad_norm": 0.4431573152542114,
      "learning_rate": 3.957064113722077e-06,
      "loss": 0.73,
      "step": 5034
    },
    {
      "epoch": 0.20867006506693192,
      "grad_norm": 0.44064074754714966,
      "learning_rate": 3.956856894193709e-06,
      "loss": 0.7795,
      "step": 5035
    },
    {
      "epoch": 0.20871150897260557,
      "grad_norm": 0.4039342403411865,
      "learning_rate": 3.956649674665341e-06,
      "loss": 0.728,
      "step": 5036
    },
    {
      "epoch": 0.20875295287827925,
      "grad_norm": 0.3996719717979431,
      "learning_rate": 3.956442455136972e-06,
      "loss": 0.7205,
      "step": 5037
    },
    {
      "epoch": 0.20879439678395292,
      "grad_norm": 0.44030120968818665,
      "learning_rate": 3.9562352356086045e-06,
      "loss": 0.728,
      "step": 5038
    },
    {
      "epoch": 0.2088358406896266,
      "grad_norm": 0.4313356280326843,
      "learning_rate": 3.956028016080236e-06,
      "loss": 0.7285,
      "step": 5039
    },
    {
      "epoch": 0.20887728459530025,
      "grad_norm": 0.435509592294693,
      "learning_rate": 3.955820796551867e-06,
      "loss": 0.7515,
      "step": 5040
    },
    {
      "epoch": 0.20891872850097393,
      "grad_norm": 0.4092146158218384,
      "learning_rate": 3.955613577023499e-06,
      "loss": 0.725,
      "step": 5041
    },
    {
      "epoch": 0.2089601724066476,
      "grad_norm": 0.43713656067848206,
      "learning_rate": 3.955406357495131e-06,
      "loss": 0.7039,
      "step": 5042
    },
    {
      "epoch": 0.20900161631232128,
      "grad_norm": 0.4088458716869354,
      "learning_rate": 3.955199137966762e-06,
      "loss": 0.691,
      "step": 5043
    },
    {
      "epoch": 0.20904306021799493,
      "grad_norm": 0.455867201089859,
      "learning_rate": 3.954991918438394e-06,
      "loss": 0.7429,
      "step": 5044
    },
    {
      "epoch": 0.2090845041236686,
      "grad_norm": 0.4370984435081482,
      "learning_rate": 3.954784698910026e-06,
      "loss": 0.7611,
      "step": 5045
    },
    {
      "epoch": 0.20912594802934228,
      "grad_norm": 0.4162183105945587,
      "learning_rate": 3.954577479381657e-06,
      "loss": 0.7268,
      "step": 5046
    },
    {
      "epoch": 0.20916739193501596,
      "grad_norm": 0.43249839544296265,
      "learning_rate": 3.954370259853289e-06,
      "loss": 0.7042,
      "step": 5047
    },
    {
      "epoch": 0.20920883584068964,
      "grad_norm": 0.3995538353919983,
      "learning_rate": 3.95416304032492e-06,
      "loss": 0.6987,
      "step": 5048
    },
    {
      "epoch": 0.2092502797463633,
      "grad_norm": 0.40230026841163635,
      "learning_rate": 3.953955820796552e-06,
      "loss": 0.6929,
      "step": 5049
    },
    {
      "epoch": 0.20929172365203697,
      "grad_norm": 0.41531893610954285,
      "learning_rate": 3.953748601268184e-06,
      "loss": 0.7115,
      "step": 5050
    },
    {
      "epoch": 0.20933316755771064,
      "grad_norm": 0.3995685279369354,
      "learning_rate": 3.953541381739815e-06,
      "loss": 0.7405,
      "step": 5051
    },
    {
      "epoch": 0.20937461146338432,
      "grad_norm": 0.4462067484855652,
      "learning_rate": 3.953334162211447e-06,
      "loss": 0.7222,
      "step": 5052
    },
    {
      "epoch": 0.20941605536905797,
      "grad_norm": 0.4951309859752655,
      "learning_rate": 3.953126942683079e-06,
      "loss": 0.8076,
      "step": 5053
    },
    {
      "epoch": 0.20945749927473165,
      "grad_norm": 0.45819154381752014,
      "learning_rate": 3.952919723154711e-06,
      "loss": 0.7302,
      "step": 5054
    },
    {
      "epoch": 0.20949894318040532,
      "grad_norm": 0.4239514470100403,
      "learning_rate": 3.952712503626342e-06,
      "loss": 0.7209,
      "step": 5055
    },
    {
      "epoch": 0.209540387086079,
      "grad_norm": 0.41919928789138794,
      "learning_rate": 3.952505284097974e-06,
      "loss": 0.739,
      "step": 5056
    },
    {
      "epoch": 0.20958183099175265,
      "grad_norm": 0.4425763487815857,
      "learning_rate": 3.952298064569605e-06,
      "loss": 0.7305,
      "step": 5057
    },
    {
      "epoch": 0.20962327489742633,
      "grad_norm": 0.42079198360443115,
      "learning_rate": 3.952090845041237e-06,
      "loss": 0.7084,
      "step": 5058
    },
    {
      "epoch": 0.2096647188031,
      "grad_norm": 0.4223988652229309,
      "learning_rate": 3.951883625512869e-06,
      "loss": 0.7566,
      "step": 5059
    },
    {
      "epoch": 0.20970616270877368,
      "grad_norm": 0.42822688817977905,
      "learning_rate": 3.9516764059845e-06,
      "loss": 0.7522,
      "step": 5060
    },
    {
      "epoch": 0.20974760661444736,
      "grad_norm": 0.41007164120674133,
      "learning_rate": 3.951469186456132e-06,
      "loss": 0.7178,
      "step": 5061
    },
    {
      "epoch": 0.209789050520121,
      "grad_norm": 0.39228421449661255,
      "learning_rate": 3.951261966927764e-06,
      "loss": 0.7,
      "step": 5062
    },
    {
      "epoch": 0.20983049442579468,
      "grad_norm": 0.4414163827896118,
      "learning_rate": 3.951054747399395e-06,
      "loss": 0.7214,
      "step": 5063
    },
    {
      "epoch": 0.20987193833146836,
      "grad_norm": 0.4180108308792114,
      "learning_rate": 3.9508475278710265e-06,
      "loss": 0.7729,
      "step": 5064
    },
    {
      "epoch": 0.20991338223714204,
      "grad_norm": 0.4564184546470642,
      "learning_rate": 3.950640308342659e-06,
      "loss": 0.7627,
      "step": 5065
    },
    {
      "epoch": 0.2099548261428157,
      "grad_norm": 0.4167024493217468,
      "learning_rate": 3.95043308881429e-06,
      "loss": 0.7095,
      "step": 5066
    },
    {
      "epoch": 0.20999627004848936,
      "grad_norm": 0.44754770398139954,
      "learning_rate": 3.9502258692859215e-06,
      "loss": 0.7255,
      "step": 5067
    },
    {
      "epoch": 0.21003771395416304,
      "grad_norm": 0.41250523924827576,
      "learning_rate": 3.950018649757553e-06,
      "loss": 0.6854,
      "step": 5068
    },
    {
      "epoch": 0.21007915785983672,
      "grad_norm": 0.43757668137550354,
      "learning_rate": 3.949811430229185e-06,
      "loss": 0.7227,
      "step": 5069
    },
    {
      "epoch": 0.2101206017655104,
      "grad_norm": 0.408383309841156,
      "learning_rate": 3.949604210700817e-06,
      "loss": 0.7249,
      "step": 5070
    },
    {
      "epoch": 0.21016204567118404,
      "grad_norm": 0.40836358070373535,
      "learning_rate": 3.949396991172449e-06,
      "loss": 0.7515,
      "step": 5071
    },
    {
      "epoch": 0.21020348957685772,
      "grad_norm": 0.44014766812324524,
      "learning_rate": 3.94918977164408e-06,
      "loss": 0.7651,
      "step": 5072
    },
    {
      "epoch": 0.2102449334825314,
      "grad_norm": 0.40353161096572876,
      "learning_rate": 3.9489825521157115e-06,
      "loss": 0.7284,
      "step": 5073
    },
    {
      "epoch": 0.21028637738820508,
      "grad_norm": 0.44718509912490845,
      "learning_rate": 3.948775332587344e-06,
      "loss": 0.731,
      "step": 5074
    },
    {
      "epoch": 0.21032782129387873,
      "grad_norm": 0.45299291610717773,
      "learning_rate": 3.948568113058975e-06,
      "loss": 0.7078,
      "step": 5075
    },
    {
      "epoch": 0.2103692651995524,
      "grad_norm": 0.446534126996994,
      "learning_rate": 3.9483608935306065e-06,
      "loss": 0.7344,
      "step": 5076
    },
    {
      "epoch": 0.21041070910522608,
      "grad_norm": 0.3768323063850403,
      "learning_rate": 3.948153674002238e-06,
      "loss": 0.689,
      "step": 5077
    },
    {
      "epoch": 0.21045215301089976,
      "grad_norm": 0.4404606223106384,
      "learning_rate": 3.94794645447387e-06,
      "loss": 0.7244,
      "step": 5078
    },
    {
      "epoch": 0.2104935969165734,
      "grad_norm": 0.4412422478199005,
      "learning_rate": 3.9477392349455015e-06,
      "loss": 0.6986,
      "step": 5079
    },
    {
      "epoch": 0.21053504082224708,
      "grad_norm": 0.42820578813552856,
      "learning_rate": 3.947532015417133e-06,
      "loss": 0.7676,
      "step": 5080
    },
    {
      "epoch": 0.21057648472792076,
      "grad_norm": 0.44106268882751465,
      "learning_rate": 3.947324795888765e-06,
      "loss": 0.7633,
      "step": 5081
    },
    {
      "epoch": 0.21061792863359444,
      "grad_norm": 0.43526729941368103,
      "learning_rate": 3.9471175763603965e-06,
      "loss": 0.7362,
      "step": 5082
    },
    {
      "epoch": 0.2106593725392681,
      "grad_norm": 0.4474290609359741,
      "learning_rate": 3.946910356832028e-06,
      "loss": 0.688,
      "step": 5083
    },
    {
      "epoch": 0.21070081644494176,
      "grad_norm": 0.45499250292778015,
      "learning_rate": 3.946703137303659e-06,
      "loss": 0.7366,
      "step": 5084
    },
    {
      "epoch": 0.21074226035061544,
      "grad_norm": 0.4522298574447632,
      "learning_rate": 3.9464959177752915e-06,
      "loss": 0.7322,
      "step": 5085
    },
    {
      "epoch": 0.21078370425628912,
      "grad_norm": 0.43330124020576477,
      "learning_rate": 3.946288698246923e-06,
      "loss": 0.7166,
      "step": 5086
    },
    {
      "epoch": 0.2108251481619628,
      "grad_norm": 0.3907892107963562,
      "learning_rate": 3.946081478718555e-06,
      "loss": 0.713,
      "step": 5087
    },
    {
      "epoch": 0.21086659206763644,
      "grad_norm": 0.4124689996242523,
      "learning_rate": 3.9458742591901865e-06,
      "loss": 0.709,
      "step": 5088
    },
    {
      "epoch": 0.21090803597331012,
      "grad_norm": 0.4523718059062958,
      "learning_rate": 3.945667039661818e-06,
      "loss": 0.6847,
      "step": 5089
    },
    {
      "epoch": 0.2109494798789838,
      "grad_norm": 0.4406546652317047,
      "learning_rate": 3.94545982013345e-06,
      "loss": 0.7771,
      "step": 5090
    },
    {
      "epoch": 0.21099092378465747,
      "grad_norm": 0.4222888648509979,
      "learning_rate": 3.9452526006050815e-06,
      "loss": 0.6774,
      "step": 5091
    },
    {
      "epoch": 0.21103236769033112,
      "grad_norm": 0.4384972155094147,
      "learning_rate": 3.945045381076713e-06,
      "loss": 0.6938,
      "step": 5092
    },
    {
      "epoch": 0.2110738115960048,
      "grad_norm": 0.4407550096511841,
      "learning_rate": 3.944838161548344e-06,
      "loss": 0.7139,
      "step": 5093
    },
    {
      "epoch": 0.21111525550167848,
      "grad_norm": 0.42434924840927124,
      "learning_rate": 3.9446309420199765e-06,
      "loss": 0.7142,
      "step": 5094
    },
    {
      "epoch": 0.21115669940735216,
      "grad_norm": 0.4632549583911896,
      "learning_rate": 3.944423722491608e-06,
      "loss": 0.7534,
      "step": 5095
    },
    {
      "epoch": 0.21119814331302583,
      "grad_norm": 0.395626962184906,
      "learning_rate": 3.944216502963239e-06,
      "loss": 0.6675,
      "step": 5096
    },
    {
      "epoch": 0.21123958721869948,
      "grad_norm": 0.38264575600624084,
      "learning_rate": 3.9440092834348715e-06,
      "loss": 0.698,
      "step": 5097
    },
    {
      "epoch": 0.21128103112437316,
      "grad_norm": 0.4366777241230011,
      "learning_rate": 3.943802063906503e-06,
      "loss": 0.7549,
      "step": 5098
    },
    {
      "epoch": 0.21132247503004684,
      "grad_norm": 0.399330198764801,
      "learning_rate": 3.943594844378134e-06,
      "loss": 0.6646,
      "step": 5099
    },
    {
      "epoch": 0.2113639189357205,
      "grad_norm": 0.44805723428726196,
      "learning_rate": 3.943387624849766e-06,
      "loss": 0.7253,
      "step": 5100
    },
    {
      "epoch": 0.21140536284139416,
      "grad_norm": 0.3996202349662781,
      "learning_rate": 3.943180405321398e-06,
      "loss": 0.7134,
      "step": 5101
    },
    {
      "epoch": 0.21144680674706784,
      "grad_norm": 0.4322737753391266,
      "learning_rate": 3.942973185793029e-06,
      "loss": 0.7317,
      "step": 5102
    },
    {
      "epoch": 0.21148825065274152,
      "grad_norm": 0.38949868083000183,
      "learning_rate": 3.942765966264661e-06,
      "loss": 0.6892,
      "step": 5103
    },
    {
      "epoch": 0.2115296945584152,
      "grad_norm": 0.48289552330970764,
      "learning_rate": 3.942558746736293e-06,
      "loss": 0.7013,
      "step": 5104
    },
    {
      "epoch": 0.21157113846408884,
      "grad_norm": 0.4455186426639557,
      "learning_rate": 3.942351527207924e-06,
      "loss": 0.6898,
      "step": 5105
    },
    {
      "epoch": 0.21161258236976252,
      "grad_norm": 0.4251181185245514,
      "learning_rate": 3.9421443076795566e-06,
      "loss": 0.7366,
      "step": 5106
    },
    {
      "epoch": 0.2116540262754362,
      "grad_norm": 0.41208216547966003,
      "learning_rate": 3.941937088151188e-06,
      "loss": 0.7183,
      "step": 5107
    },
    {
      "epoch": 0.21169547018110987,
      "grad_norm": 0.41952913999557495,
      "learning_rate": 3.941729868622819e-06,
      "loss": 0.7275,
      "step": 5108
    },
    {
      "epoch": 0.21173691408678355,
      "grad_norm": 0.40845420956611633,
      "learning_rate": 3.941522649094451e-06,
      "loss": 0.7336,
      "step": 5109
    },
    {
      "epoch": 0.2117783579924572,
      "grad_norm": 0.41922375559806824,
      "learning_rate": 3.941315429566083e-06,
      "loss": 0.717,
      "step": 5110
    },
    {
      "epoch": 0.21181980189813088,
      "grad_norm": 0.45428892970085144,
      "learning_rate": 3.941108210037714e-06,
      "loss": 0.7583,
      "step": 5111
    },
    {
      "epoch": 0.21186124580380455,
      "grad_norm": 0.43839678168296814,
      "learning_rate": 3.940900990509346e-06,
      "loss": 0.7095,
      "step": 5112
    },
    {
      "epoch": 0.21190268970947823,
      "grad_norm": 0.45987460017204285,
      "learning_rate": 3.940693770980978e-06,
      "loss": 0.749,
      "step": 5113
    },
    {
      "epoch": 0.21194413361515188,
      "grad_norm": 0.3921250104904175,
      "learning_rate": 3.940486551452609e-06,
      "loss": 0.7686,
      "step": 5114
    },
    {
      "epoch": 0.21198557752082556,
      "grad_norm": 0.44409358501434326,
      "learning_rate": 3.940279331924241e-06,
      "loss": 0.6809,
      "step": 5115
    },
    {
      "epoch": 0.21202702142649923,
      "grad_norm": 0.44515255093574524,
      "learning_rate": 3.940072112395872e-06,
      "loss": 0.7356,
      "step": 5116
    },
    {
      "epoch": 0.2120684653321729,
      "grad_norm": 0.41947421431541443,
      "learning_rate": 3.939864892867504e-06,
      "loss": 0.7096,
      "step": 5117
    },
    {
      "epoch": 0.21210990923784656,
      "grad_norm": 0.3819780647754669,
      "learning_rate": 3.939657673339136e-06,
      "loss": 0.6553,
      "step": 5118
    },
    {
      "epoch": 0.21215135314352024,
      "grad_norm": 0.4444463551044464,
      "learning_rate": 3.939450453810767e-06,
      "loss": 0.7119,
      "step": 5119
    },
    {
      "epoch": 0.21219279704919392,
      "grad_norm": 0.42417439818382263,
      "learning_rate": 3.9392432342823985e-06,
      "loss": 0.6917,
      "step": 5120
    },
    {
      "epoch": 0.2122342409548676,
      "grad_norm": 0.442605584859848,
      "learning_rate": 3.939036014754031e-06,
      "loss": 0.7161,
      "step": 5121
    },
    {
      "epoch": 0.21227568486054127,
      "grad_norm": 0.3828595280647278,
      "learning_rate": 3.938828795225663e-06,
      "loss": 0.6624,
      "step": 5122
    },
    {
      "epoch": 0.21231712876621492,
      "grad_norm": 0.40411755442619324,
      "learning_rate": 3.938621575697294e-06,
      "loss": 0.6982,
      "step": 5123
    },
    {
      "epoch": 0.2123585726718886,
      "grad_norm": 0.4094061255455017,
      "learning_rate": 3.938414356168926e-06,
      "loss": 0.7162,
      "step": 5124
    },
    {
      "epoch": 0.21240001657756227,
      "grad_norm": 0.40393975377082825,
      "learning_rate": 3.938207136640557e-06,
      "loss": 0.7522,
      "step": 5125
    },
    {
      "epoch": 0.21244146048323595,
      "grad_norm": 0.4101594090461731,
      "learning_rate": 3.937999917112189e-06,
      "loss": 0.73,
      "step": 5126
    },
    {
      "epoch": 0.2124829043889096,
      "grad_norm": 0.48825618624687195,
      "learning_rate": 3.937792697583821e-06,
      "loss": 0.7175,
      "step": 5127
    },
    {
      "epoch": 0.21252434829458328,
      "grad_norm": 0.4631562829017639,
      "learning_rate": 3.937585478055452e-06,
      "loss": 0.7239,
      "step": 5128
    },
    {
      "epoch": 0.21256579220025695,
      "grad_norm": 0.41530945897102356,
      "learning_rate": 3.9373782585270835e-06,
      "loss": 0.6995,
      "step": 5129
    },
    {
      "epoch": 0.21260723610593063,
      "grad_norm": 0.42156508564949036,
      "learning_rate": 3.937171038998716e-06,
      "loss": 0.7169,
      "step": 5130
    },
    {
      "epoch": 0.2126486800116043,
      "grad_norm": 0.4403069019317627,
      "learning_rate": 3.936963819470347e-06,
      "loss": 0.7493,
      "step": 5131
    },
    {
      "epoch": 0.21269012391727796,
      "grad_norm": 0.44364216923713684,
      "learning_rate": 3.9367565999419785e-06,
      "loss": 0.7576,
      "step": 5132
    },
    {
      "epoch": 0.21273156782295163,
      "grad_norm": 0.4258248507976532,
      "learning_rate": 3.936549380413611e-06,
      "loss": 0.702,
      "step": 5133
    },
    {
      "epoch": 0.2127730117286253,
      "grad_norm": 0.4212954640388489,
      "learning_rate": 3.936342160885242e-06,
      "loss": 0.7583,
      "step": 5134
    },
    {
      "epoch": 0.212814455634299,
      "grad_norm": 0.43298667669296265,
      "learning_rate": 3.9361349413568735e-06,
      "loss": 0.7878,
      "step": 5135
    },
    {
      "epoch": 0.21285589953997264,
      "grad_norm": 0.43907710909843445,
      "learning_rate": 3.935927721828505e-06,
      "loss": 0.7444,
      "step": 5136
    },
    {
      "epoch": 0.21289734344564631,
      "grad_norm": 0.4509316682815552,
      "learning_rate": 3.935720502300137e-06,
      "loss": 0.72,
      "step": 5137
    },
    {
      "epoch": 0.21293878735132,
      "grad_norm": 0.4226623773574829,
      "learning_rate": 3.9355132827717685e-06,
      "loss": 0.74,
      "step": 5138
    },
    {
      "epoch": 0.21298023125699367,
      "grad_norm": 0.4085758924484253,
      "learning_rate": 3.935306063243401e-06,
      "loss": 0.6952,
      "step": 5139
    },
    {
      "epoch": 0.21302167516266732,
      "grad_norm": 0.4268183410167694,
      "learning_rate": 3.935098843715032e-06,
      "loss": 0.718,
      "step": 5140
    },
    {
      "epoch": 0.213063119068341,
      "grad_norm": 0.43286436796188354,
      "learning_rate": 3.9348916241866635e-06,
      "loss": 0.6699,
      "step": 5141
    },
    {
      "epoch": 0.21310456297401467,
      "grad_norm": 0.4486483335494995,
      "learning_rate": 3.934684404658296e-06,
      "loss": 0.7191,
      "step": 5142
    },
    {
      "epoch": 0.21314600687968835,
      "grad_norm": 0.4399454593658447,
      "learning_rate": 3.934477185129927e-06,
      "loss": 0.7476,
      "step": 5143
    },
    {
      "epoch": 0.21318745078536203,
      "grad_norm": 0.45652467012405396,
      "learning_rate": 3.9342699656015585e-06,
      "loss": 0.7393,
      "step": 5144
    },
    {
      "epoch": 0.21322889469103568,
      "grad_norm": 0.3898095488548279,
      "learning_rate": 3.93406274607319e-06,
      "loss": 0.6449,
      "step": 5145
    },
    {
      "epoch": 0.21327033859670935,
      "grad_norm": 0.4764745235443115,
      "learning_rate": 3.933855526544822e-06,
      "loss": 0.7598,
      "step": 5146
    },
    {
      "epoch": 0.21331178250238303,
      "grad_norm": 0.40731653571128845,
      "learning_rate": 3.9336483070164535e-06,
      "loss": 0.7012,
      "step": 5147
    },
    {
      "epoch": 0.2133532264080567,
      "grad_norm": 0.4394112527370453,
      "learning_rate": 3.933441087488085e-06,
      "loss": 0.7021,
      "step": 5148
    },
    {
      "epoch": 0.21339467031373036,
      "grad_norm": 0.4351750612258911,
      "learning_rate": 3.933233867959717e-06,
      "loss": 0.7197,
      "step": 5149
    },
    {
      "epoch": 0.21343611421940403,
      "grad_norm": 0.4212436378002167,
      "learning_rate": 3.9330266484313485e-06,
      "loss": 0.6809,
      "step": 5150
    },
    {
      "epoch": 0.2134775581250777,
      "grad_norm": 0.40793144702911377,
      "learning_rate": 3.93281942890298e-06,
      "loss": 0.6821,
      "step": 5151
    },
    {
      "epoch": 0.2135190020307514,
      "grad_norm": 0.40093353390693665,
      "learning_rate": 3.932612209374611e-06,
      "loss": 0.6926,
      "step": 5152
    },
    {
      "epoch": 0.21356044593642504,
      "grad_norm": 0.4245559275150299,
      "learning_rate": 3.9324049898462435e-06,
      "loss": 0.7231,
      "step": 5153
    },
    {
      "epoch": 0.2136018898420987,
      "grad_norm": 0.42561954259872437,
      "learning_rate": 3.932197770317875e-06,
      "loss": 0.7131,
      "step": 5154
    },
    {
      "epoch": 0.2136433337477724,
      "grad_norm": 0.4299514889717102,
      "learning_rate": 3.931990550789507e-06,
      "loss": 0.738,
      "step": 5155
    },
    {
      "epoch": 0.21368477765344607,
      "grad_norm": 0.444667249917984,
      "learning_rate": 3.9317833312611385e-06,
      "loss": 0.7251,
      "step": 5156
    },
    {
      "epoch": 0.21372622155911974,
      "grad_norm": 0.4218398630619049,
      "learning_rate": 3.93157611173277e-06,
      "loss": 0.7035,
      "step": 5157
    },
    {
      "epoch": 0.2137676654647934,
      "grad_norm": 0.41172507405281067,
      "learning_rate": 3.931368892204402e-06,
      "loss": 0.7197,
      "step": 5158
    },
    {
      "epoch": 0.21380910937046707,
      "grad_norm": 0.38863325119018555,
      "learning_rate": 3.9311616726760335e-06,
      "loss": 0.71,
      "step": 5159
    },
    {
      "epoch": 0.21385055327614075,
      "grad_norm": 0.424308180809021,
      "learning_rate": 3.930954453147665e-06,
      "loss": 0.6819,
      "step": 5160
    },
    {
      "epoch": 0.21389199718181442,
      "grad_norm": 0.4315939247608185,
      "learning_rate": 3.930747233619296e-06,
      "loss": 0.7378,
      "step": 5161
    },
    {
      "epoch": 0.21393344108748807,
      "grad_norm": 0.4201556146144867,
      "learning_rate": 3.9305400140909286e-06,
      "loss": 0.749,
      "step": 5162
    },
    {
      "epoch": 0.21397488499316175,
      "grad_norm": 0.41834041476249695,
      "learning_rate": 3.93033279456256e-06,
      "loss": 0.731,
      "step": 5163
    },
    {
      "epoch": 0.21401632889883543,
      "grad_norm": 0.426853746175766,
      "learning_rate": 3.930125575034191e-06,
      "loss": 0.7478,
      "step": 5164
    },
    {
      "epoch": 0.2140577728045091,
      "grad_norm": 0.42516541481018066,
      "learning_rate": 3.9299183555058236e-06,
      "loss": 0.7206,
      "step": 5165
    },
    {
      "epoch": 0.21409921671018275,
      "grad_norm": 0.39548060297966003,
      "learning_rate": 3.929711135977455e-06,
      "loss": 0.7258,
      "step": 5166
    },
    {
      "epoch": 0.21414066061585643,
      "grad_norm": 0.39803415536880493,
      "learning_rate": 3.929503916449086e-06,
      "loss": 0.6753,
      "step": 5167
    },
    {
      "epoch": 0.2141821045215301,
      "grad_norm": 0.445502907037735,
      "learning_rate": 3.929296696920718e-06,
      "loss": 0.7654,
      "step": 5168
    },
    {
      "epoch": 0.21422354842720379,
      "grad_norm": 0.4254343509674072,
      "learning_rate": 3.92908947739235e-06,
      "loss": 0.7148,
      "step": 5169
    },
    {
      "epoch": 0.21426499233287746,
      "grad_norm": 0.41869983077049255,
      "learning_rate": 3.928882257863981e-06,
      "loss": 0.739,
      "step": 5170
    },
    {
      "epoch": 0.2143064362385511,
      "grad_norm": 0.412995308637619,
      "learning_rate": 3.9286750383356136e-06,
      "loss": 0.6746,
      "step": 5171
    },
    {
      "epoch": 0.2143478801442248,
      "grad_norm": 0.4200811982154846,
      "learning_rate": 3.928467818807244e-06,
      "loss": 0.7649,
      "step": 5172
    },
    {
      "epoch": 0.21438932404989847,
      "grad_norm": 0.43147769570350647,
      "learning_rate": 3.928260599278876e-06,
      "loss": 0.7271,
      "step": 5173
    },
    {
      "epoch": 0.21443076795557214,
      "grad_norm": 0.39612942934036255,
      "learning_rate": 3.9280533797505086e-06,
      "loss": 0.6746,
      "step": 5174
    },
    {
      "epoch": 0.2144722118612458,
      "grad_norm": 0.4001898467540741,
      "learning_rate": 3.92784616022214e-06,
      "loss": 0.7356,
      "step": 5175
    },
    {
      "epoch": 0.21451365576691947,
      "grad_norm": 0.43184083700180054,
      "learning_rate": 3.927638940693771e-06,
      "loss": 0.7747,
      "step": 5176
    },
    {
      "epoch": 0.21455509967259315,
      "grad_norm": 0.4445628523826599,
      "learning_rate": 3.927431721165403e-06,
      "loss": 0.7024,
      "step": 5177
    },
    {
      "epoch": 0.21459654357826682,
      "grad_norm": 0.4417478144168854,
      "learning_rate": 3.927224501637035e-06,
      "loss": 0.7415,
      "step": 5178
    },
    {
      "epoch": 0.21463798748394047,
      "grad_norm": 0.4569386839866638,
      "learning_rate": 3.927017282108666e-06,
      "loss": 0.7385,
      "step": 5179
    },
    {
      "epoch": 0.21467943138961415,
      "grad_norm": 0.4119894802570343,
      "learning_rate": 3.926810062580298e-06,
      "loss": 0.7297,
      "step": 5180
    },
    {
      "epoch": 0.21472087529528783,
      "grad_norm": 0.43150511384010315,
      "learning_rate": 3.926602843051929e-06,
      "loss": 0.677,
      "step": 5181
    },
    {
      "epoch": 0.2147623192009615,
      "grad_norm": 0.41464290022850037,
      "learning_rate": 3.926395623523561e-06,
      "loss": 0.7251,
      "step": 5182
    },
    {
      "epoch": 0.21480376310663518,
      "grad_norm": 0.4310956299304962,
      "learning_rate": 3.926188403995193e-06,
      "loss": 0.7439,
      "step": 5183
    },
    {
      "epoch": 0.21484520701230883,
      "grad_norm": 0.4162323772907257,
      "learning_rate": 3.925981184466824e-06,
      "loss": 0.6949,
      "step": 5184
    },
    {
      "epoch": 0.2148866509179825,
      "grad_norm": 0.43507465720176697,
      "learning_rate": 3.925773964938456e-06,
      "loss": 0.6674,
      "step": 5185
    },
    {
      "epoch": 0.21492809482365619,
      "grad_norm": 0.3878183364868164,
      "learning_rate": 3.925566745410088e-06,
      "loss": 0.6829,
      "step": 5186
    },
    {
      "epoch": 0.21496953872932986,
      "grad_norm": 0.40035155415534973,
      "learning_rate": 3.925359525881719e-06,
      "loss": 0.7157,
      "step": 5187
    },
    {
      "epoch": 0.2150109826350035,
      "grad_norm": 0.42390206456184387,
      "learning_rate": 3.9251523063533505e-06,
      "loss": 0.6993,
      "step": 5188
    },
    {
      "epoch": 0.2150524265406772,
      "grad_norm": 0.4005449414253235,
      "learning_rate": 3.924945086824983e-06,
      "loss": 0.6475,
      "step": 5189
    },
    {
      "epoch": 0.21509387044635087,
      "grad_norm": 0.4318193197250366,
      "learning_rate": 3.924737867296614e-06,
      "loss": 0.731,
      "step": 5190
    },
    {
      "epoch": 0.21513531435202454,
      "grad_norm": 0.40991804003715515,
      "learning_rate": 3.924530647768246e-06,
      "loss": 0.708,
      "step": 5191
    },
    {
      "epoch": 0.2151767582576982,
      "grad_norm": 0.4509611129760742,
      "learning_rate": 3.924323428239878e-06,
      "loss": 0.7427,
      "step": 5192
    },
    {
      "epoch": 0.21521820216337187,
      "grad_norm": 0.4485681653022766,
      "learning_rate": 3.924116208711509e-06,
      "loss": 0.7134,
      "step": 5193
    },
    {
      "epoch": 0.21525964606904555,
      "grad_norm": 0.44018790125846863,
      "learning_rate": 3.923908989183141e-06,
      "loss": 0.7583,
      "step": 5194
    },
    {
      "epoch": 0.21530108997471922,
      "grad_norm": 0.41066861152648926,
      "learning_rate": 3.923701769654773e-06,
      "loss": 0.6833,
      "step": 5195
    },
    {
      "epoch": 0.2153425338803929,
      "grad_norm": 0.4888826906681061,
      "learning_rate": 3.923494550126404e-06,
      "loss": 0.7869,
      "step": 5196
    },
    {
      "epoch": 0.21538397778606655,
      "grad_norm": 0.44207435846328735,
      "learning_rate": 3.9232873305980355e-06,
      "loss": 0.7661,
      "step": 5197
    },
    {
      "epoch": 0.21542542169174023,
      "grad_norm": 0.5160230994224548,
      "learning_rate": 3.923080111069668e-06,
      "loss": 0.7111,
      "step": 5198
    },
    {
      "epoch": 0.2154668655974139,
      "grad_norm": 0.4555935561656952,
      "learning_rate": 3.922872891541299e-06,
      "loss": 0.7375,
      "step": 5199
    },
    {
      "epoch": 0.21550830950308758,
      "grad_norm": 0.41654545068740845,
      "learning_rate": 3.9226656720129305e-06,
      "loss": 0.7556,
      "step": 5200
    },
    {
      "epoch": 0.21554975340876123,
      "grad_norm": 0.43797725439071655,
      "learning_rate": 3.922458452484563e-06,
      "loss": 0.7522,
      "step": 5201
    },
    {
      "epoch": 0.2155911973144349,
      "grad_norm": 0.4216921925544739,
      "learning_rate": 3.922251232956194e-06,
      "loss": 0.6951,
      "step": 5202
    },
    {
      "epoch": 0.21563264122010858,
      "grad_norm": 0.43047675490379333,
      "learning_rate": 3.9220440134278255e-06,
      "loss": 0.6967,
      "step": 5203
    },
    {
      "epoch": 0.21567408512578226,
      "grad_norm": 0.40262022614479065,
      "learning_rate": 3.921836793899457e-06,
      "loss": 0.7256,
      "step": 5204
    },
    {
      "epoch": 0.21571552903145594,
      "grad_norm": 0.4223463237285614,
      "learning_rate": 3.921629574371089e-06,
      "loss": 0.7355,
      "step": 5205
    },
    {
      "epoch": 0.2157569729371296,
      "grad_norm": 0.4099085032939911,
      "learning_rate": 3.9214223548427205e-06,
      "loss": 0.7021,
      "step": 5206
    },
    {
      "epoch": 0.21579841684280326,
      "grad_norm": 0.4082989990711212,
      "learning_rate": 3.921215135314353e-06,
      "loss": 0.7234,
      "step": 5207
    },
    {
      "epoch": 0.21583986074847694,
      "grad_norm": 0.44395506381988525,
      "learning_rate": 3.921007915785983e-06,
      "loss": 0.7214,
      "step": 5208
    },
    {
      "epoch": 0.21588130465415062,
      "grad_norm": 0.4146428406238556,
      "learning_rate": 3.9208006962576155e-06,
      "loss": 0.6987,
      "step": 5209
    },
    {
      "epoch": 0.21592274855982427,
      "grad_norm": 0.43167275190353394,
      "learning_rate": 3.920593476729248e-06,
      "loss": 0.6279,
      "step": 5210
    },
    {
      "epoch": 0.21596419246549795,
      "grad_norm": 0.4153953492641449,
      "learning_rate": 3.920386257200879e-06,
      "loss": 0.71,
      "step": 5211
    },
    {
      "epoch": 0.21600563637117162,
      "grad_norm": 0.4432026743888855,
      "learning_rate": 3.9201790376725105e-06,
      "loss": 0.6833,
      "step": 5212
    },
    {
      "epoch": 0.2160470802768453,
      "grad_norm": 0.4030596613883972,
      "learning_rate": 3.919971818144142e-06,
      "loss": 0.7466,
      "step": 5213
    },
    {
      "epoch": 0.21608852418251895,
      "grad_norm": 0.4269682466983795,
      "learning_rate": 3.919764598615774e-06,
      "loss": 0.7031,
      "step": 5214
    },
    {
      "epoch": 0.21612996808819263,
      "grad_norm": 0.460428386926651,
      "learning_rate": 3.9195573790874055e-06,
      "loss": 0.8051,
      "step": 5215
    },
    {
      "epoch": 0.2161714119938663,
      "grad_norm": 0.43371573090553284,
      "learning_rate": 3.919350159559037e-06,
      "loss": 0.7063,
      "step": 5216
    },
    {
      "epoch": 0.21621285589953998,
      "grad_norm": 0.39581677317619324,
      "learning_rate": 3.919142940030669e-06,
      "loss": 0.6775,
      "step": 5217
    },
    {
      "epoch": 0.21625429980521366,
      "grad_norm": 0.4207910895347595,
      "learning_rate": 3.9189357205023005e-06,
      "loss": 0.6982,
      "step": 5218
    },
    {
      "epoch": 0.2162957437108873,
      "grad_norm": 0.4456447660923004,
      "learning_rate": 3.918728500973932e-06,
      "loss": 0.748,
      "step": 5219
    },
    {
      "epoch": 0.21633718761656098,
      "grad_norm": 0.4551488757133484,
      "learning_rate": 3.918521281445563e-06,
      "loss": 0.7275,
      "step": 5220
    },
    {
      "epoch": 0.21637863152223466,
      "grad_norm": 0.463957279920578,
      "learning_rate": 3.9183140619171956e-06,
      "loss": 0.7722,
      "step": 5221
    },
    {
      "epoch": 0.21642007542790834,
      "grad_norm": 0.419675350189209,
      "learning_rate": 3.918106842388827e-06,
      "loss": 0.7603,
      "step": 5222
    },
    {
      "epoch": 0.216461519333582,
      "grad_norm": 0.4407963752746582,
      "learning_rate": 3.917899622860459e-06,
      "loss": 0.7513,
      "step": 5223
    },
    {
      "epoch": 0.21650296323925566,
      "grad_norm": 0.39439186453819275,
      "learning_rate": 3.91769240333209e-06,
      "loss": 0.6663,
      "step": 5224
    },
    {
      "epoch": 0.21654440714492934,
      "grad_norm": 0.41965436935424805,
      "learning_rate": 3.917485183803722e-06,
      "loss": 0.761,
      "step": 5225
    },
    {
      "epoch": 0.21658585105060302,
      "grad_norm": 0.43240153789520264,
      "learning_rate": 3.917277964275354e-06,
      "loss": 0.7869,
      "step": 5226
    },
    {
      "epoch": 0.21662729495627667,
      "grad_norm": 0.4187496602535248,
      "learning_rate": 3.9170707447469856e-06,
      "loss": 0.7039,
      "step": 5227
    },
    {
      "epoch": 0.21666873886195034,
      "grad_norm": 0.4194868505001068,
      "learning_rate": 3.916863525218617e-06,
      "loss": 0.7087,
      "step": 5228
    },
    {
      "epoch": 0.21671018276762402,
      "grad_norm": 0.44700080156326294,
      "learning_rate": 3.916656305690248e-06,
      "loss": 0.7776,
      "step": 5229
    },
    {
      "epoch": 0.2167516266732977,
      "grad_norm": 0.43503013253211975,
      "learning_rate": 3.9164490861618806e-06,
      "loss": 0.7288,
      "step": 5230
    },
    {
      "epoch": 0.21679307057897138,
      "grad_norm": 0.4542178511619568,
      "learning_rate": 3.916241866633512e-06,
      "loss": 0.7188,
      "step": 5231
    },
    {
      "epoch": 0.21683451448464502,
      "grad_norm": 0.40519118309020996,
      "learning_rate": 3.916034647105143e-06,
      "loss": 0.7,
      "step": 5232
    },
    {
      "epoch": 0.2168759583903187,
      "grad_norm": 0.44479313492774963,
      "learning_rate": 3.915827427576775e-06,
      "loss": 0.7708,
      "step": 5233
    },
    {
      "epoch": 0.21691740229599238,
      "grad_norm": 0.4119117856025696,
      "learning_rate": 3.915620208048407e-06,
      "loss": 0.7158,
      "step": 5234
    },
    {
      "epoch": 0.21695884620166606,
      "grad_norm": 0.4093351662158966,
      "learning_rate": 3.915412988520038e-06,
      "loss": 0.689,
      "step": 5235
    },
    {
      "epoch": 0.2170002901073397,
      "grad_norm": 0.4301641285419464,
      "learning_rate": 3.91520576899167e-06,
      "loss": 0.7321,
      "step": 5236
    },
    {
      "epoch": 0.21704173401301338,
      "grad_norm": 0.4339520037174225,
      "learning_rate": 3.914998549463302e-06,
      "loss": 0.7112,
      "step": 5237
    },
    {
      "epoch": 0.21708317791868706,
      "grad_norm": 0.45186248421669006,
      "learning_rate": 3.914791329934933e-06,
      "loss": 0.7366,
      "step": 5238
    },
    {
      "epoch": 0.21712462182436074,
      "grad_norm": 0.431069016456604,
      "learning_rate": 3.9145841104065656e-06,
      "loss": 0.741,
      "step": 5239
    },
    {
      "epoch": 0.21716606573003439,
      "grad_norm": 0.4019047021865845,
      "learning_rate": 3.914376890878196e-06,
      "loss": 0.7255,
      "step": 5240
    },
    {
      "epoch": 0.21720750963570806,
      "grad_norm": 0.405895471572876,
      "learning_rate": 3.914169671349828e-06,
      "loss": 0.7096,
      "step": 5241
    },
    {
      "epoch": 0.21724895354138174,
      "grad_norm": 0.40362638235092163,
      "learning_rate": 3.91396245182146e-06,
      "loss": 0.6694,
      "step": 5242
    },
    {
      "epoch": 0.21729039744705542,
      "grad_norm": 0.40513262152671814,
      "learning_rate": 3.913755232293092e-06,
      "loss": 0.6615,
      "step": 5243
    },
    {
      "epoch": 0.2173318413527291,
      "grad_norm": 0.43223991990089417,
      "learning_rate": 3.913548012764723e-06,
      "loss": 0.72,
      "step": 5244
    },
    {
      "epoch": 0.21737328525840274,
      "grad_norm": 0.4431764781475067,
      "learning_rate": 3.913340793236355e-06,
      "loss": 0.696,
      "step": 5245
    },
    {
      "epoch": 0.21741472916407642,
      "grad_norm": 0.4312267303466797,
      "learning_rate": 3.913133573707987e-06,
      "loss": 0.7383,
      "step": 5246
    },
    {
      "epoch": 0.2174561730697501,
      "grad_norm": 0.4163609445095062,
      "learning_rate": 3.912926354179618e-06,
      "loss": 0.6965,
      "step": 5247
    },
    {
      "epoch": 0.21749761697542377,
      "grad_norm": 0.41311073303222656,
      "learning_rate": 3.91271913465125e-06,
      "loss": 0.7644,
      "step": 5248
    },
    {
      "epoch": 0.21753906088109742,
      "grad_norm": 0.41359418630599976,
      "learning_rate": 3.912511915122881e-06,
      "loss": 0.7251,
      "step": 5249
    },
    {
      "epoch": 0.2175805047867711,
      "grad_norm": 0.45741909742355347,
      "learning_rate": 3.912304695594513e-06,
      "loss": 0.7701,
      "step": 5250
    },
    {
      "epoch": 0.21762194869244478,
      "grad_norm": 0.44259271025657654,
      "learning_rate": 3.912097476066145e-06,
      "loss": 0.7131,
      "step": 5251
    },
    {
      "epoch": 0.21766339259811845,
      "grad_norm": 0.4387892186641693,
      "learning_rate": 3.911890256537776e-06,
      "loss": 0.7439,
      "step": 5252
    },
    {
      "epoch": 0.2177048365037921,
      "grad_norm": 0.4638464152812958,
      "learning_rate": 3.911683037009408e-06,
      "loss": 0.7534,
      "step": 5253
    },
    {
      "epoch": 0.21774628040946578,
      "grad_norm": 0.3849570155143738,
      "learning_rate": 3.91147581748104e-06,
      "loss": 0.7159,
      "step": 5254
    },
    {
      "epoch": 0.21778772431513946,
      "grad_norm": 0.4461582601070404,
      "learning_rate": 3.911268597952671e-06,
      "loss": 0.7268,
      "step": 5255
    },
    {
      "epoch": 0.21782916822081314,
      "grad_norm": 0.39677193760871887,
      "learning_rate": 3.9110613784243025e-06,
      "loss": 0.6819,
      "step": 5256
    },
    {
      "epoch": 0.2178706121264868,
      "grad_norm": 0.4425094723701477,
      "learning_rate": 3.910854158895935e-06,
      "loss": 0.781,
      "step": 5257
    },
    {
      "epoch": 0.21791205603216046,
      "grad_norm": 0.4975070059299469,
      "learning_rate": 3.910646939367566e-06,
      "loss": 0.7612,
      "step": 5258
    },
    {
      "epoch": 0.21795349993783414,
      "grad_norm": 0.39364317059516907,
      "learning_rate": 3.910439719839198e-06,
      "loss": 0.719,
      "step": 5259
    },
    {
      "epoch": 0.21799494384350782,
      "grad_norm": 0.3889557719230652,
      "learning_rate": 3.91023250031083e-06,
      "loss": 0.6968,
      "step": 5260
    },
    {
      "epoch": 0.2180363877491815,
      "grad_norm": 0.4194556474685669,
      "learning_rate": 3.910025280782461e-06,
      "loss": 0.7183,
      "step": 5261
    },
    {
      "epoch": 0.21807783165485514,
      "grad_norm": 0.4292806386947632,
      "learning_rate": 3.909818061254093e-06,
      "loss": 0.7314,
      "step": 5262
    },
    {
      "epoch": 0.21811927556052882,
      "grad_norm": 0.43367791175842285,
      "learning_rate": 3.909610841725725e-06,
      "loss": 0.7319,
      "step": 5263
    },
    {
      "epoch": 0.2181607194662025,
      "grad_norm": 0.43757525086402893,
      "learning_rate": 3.909403622197356e-06,
      "loss": 0.7578,
      "step": 5264
    },
    {
      "epoch": 0.21820216337187617,
      "grad_norm": 0.4455266296863556,
      "learning_rate": 3.9091964026689875e-06,
      "loss": 0.74,
      "step": 5265
    },
    {
      "epoch": 0.21824360727754982,
      "grad_norm": 0.42043888568878174,
      "learning_rate": 3.90898918314062e-06,
      "loss": 0.7893,
      "step": 5266
    },
    {
      "epoch": 0.2182850511832235,
      "grad_norm": 0.4178803861141205,
      "learning_rate": 3.908781963612251e-06,
      "loss": 0.686,
      "step": 5267
    },
    {
      "epoch": 0.21832649508889718,
      "grad_norm": 0.4013496935367584,
      "learning_rate": 3.9085747440838825e-06,
      "loss": 0.7212,
      "step": 5268
    },
    {
      "epoch": 0.21836793899457085,
      "grad_norm": 0.429502010345459,
      "learning_rate": 3.908367524555514e-06,
      "loss": 0.689,
      "step": 5269
    },
    {
      "epoch": 0.21840938290024453,
      "grad_norm": 0.4205033779144287,
      "learning_rate": 3.908160305027146e-06,
      "loss": 0.6893,
      "step": 5270
    },
    {
      "epoch": 0.21845082680591818,
      "grad_norm": 0.4210493266582489,
      "learning_rate": 3.9079530854987775e-06,
      "loss": 0.7507,
      "step": 5271
    },
    {
      "epoch": 0.21849227071159186,
      "grad_norm": 0.4226735532283783,
      "learning_rate": 3.907745865970409e-06,
      "loss": 0.7174,
      "step": 5272
    },
    {
      "epoch": 0.21853371461726553,
      "grad_norm": 0.4503912925720215,
      "learning_rate": 3.907538646442041e-06,
      "loss": 0.7654,
      "step": 5273
    },
    {
      "epoch": 0.2185751585229392,
      "grad_norm": 0.4206906259059906,
      "learning_rate": 3.9073314269136725e-06,
      "loss": 0.7302,
      "step": 5274
    },
    {
      "epoch": 0.21861660242861286,
      "grad_norm": 0.39331772923469543,
      "learning_rate": 3.907124207385305e-06,
      "loss": 0.6672,
      "step": 5275
    },
    {
      "epoch": 0.21865804633428654,
      "grad_norm": 0.4659024477005005,
      "learning_rate": 3.906916987856936e-06,
      "loss": 0.7498,
      "step": 5276
    },
    {
      "epoch": 0.21869949023996021,
      "grad_norm": 0.4138321280479431,
      "learning_rate": 3.9067097683285675e-06,
      "loss": 0.7126,
      "step": 5277
    },
    {
      "epoch": 0.2187409341456339,
      "grad_norm": 0.42757049202919006,
      "learning_rate": 3.9065025488002e-06,
      "loss": 0.735,
      "step": 5278
    },
    {
      "epoch": 0.21878237805130757,
      "grad_norm": 0.40373218059539795,
      "learning_rate": 3.906295329271831e-06,
      "loss": 0.6997,
      "step": 5279
    },
    {
      "epoch": 0.21882382195698122,
      "grad_norm": 0.464275985956192,
      "learning_rate": 3.9060881097434626e-06,
      "loss": 0.7349,
      "step": 5280
    },
    {
      "epoch": 0.2188652658626549,
      "grad_norm": 0.3973308205604553,
      "learning_rate": 3.905880890215094e-06,
      "loss": 0.6904,
      "step": 5281
    },
    {
      "epoch": 0.21890670976832857,
      "grad_norm": 0.41097933053970337,
      "learning_rate": 3.905673670686726e-06,
      "loss": 0.7002,
      "step": 5282
    },
    {
      "epoch": 0.21894815367400225,
      "grad_norm": 0.4299159646034241,
      "learning_rate": 3.9054664511583576e-06,
      "loss": 0.7346,
      "step": 5283
    },
    {
      "epoch": 0.2189895975796759,
      "grad_norm": 0.4582728147506714,
      "learning_rate": 3.905259231629989e-06,
      "loss": 0.6819,
      "step": 5284
    },
    {
      "epoch": 0.21903104148534958,
      "grad_norm": 0.41970616579055786,
      "learning_rate": 3.90505201210162e-06,
      "loss": 0.7382,
      "step": 5285
    },
    {
      "epoch": 0.21907248539102325,
      "grad_norm": 0.43556076288223267,
      "learning_rate": 3.9048447925732526e-06,
      "loss": 0.7551,
      "step": 5286
    },
    {
      "epoch": 0.21911392929669693,
      "grad_norm": 0.42559918761253357,
      "learning_rate": 3.904637573044884e-06,
      "loss": 0.6956,
      "step": 5287
    },
    {
      "epoch": 0.21915537320237058,
      "grad_norm": 0.4457615315914154,
      "learning_rate": 3.904430353516515e-06,
      "loss": 0.7593,
      "step": 5288
    },
    {
      "epoch": 0.21919681710804426,
      "grad_norm": 0.3909178674221039,
      "learning_rate": 3.9042231339881476e-06,
      "loss": 0.6887,
      "step": 5289
    },
    {
      "epoch": 0.21923826101371793,
      "grad_norm": 0.44917765259742737,
      "learning_rate": 3.904015914459779e-06,
      "loss": 0.7339,
      "step": 5290
    },
    {
      "epoch": 0.2192797049193916,
      "grad_norm": 0.4159770607948303,
      "learning_rate": 3.903808694931411e-06,
      "loss": 0.7441,
      "step": 5291
    },
    {
      "epoch": 0.2193211488250653,
      "grad_norm": 0.4075734615325928,
      "learning_rate": 3.903601475403042e-06,
      "loss": 0.6919,
      "step": 5292
    },
    {
      "epoch": 0.21936259273073894,
      "grad_norm": 0.4347516596317291,
      "learning_rate": 3.903394255874674e-06,
      "loss": 0.762,
      "step": 5293
    },
    {
      "epoch": 0.2194040366364126,
      "grad_norm": 0.42917874455451965,
      "learning_rate": 3.903187036346305e-06,
      "loss": 0.7072,
      "step": 5294
    },
    {
      "epoch": 0.2194454805420863,
      "grad_norm": 0.407070517539978,
      "learning_rate": 3.9029798168179376e-06,
      "loss": 0.7308,
      "step": 5295
    },
    {
      "epoch": 0.21948692444775997,
      "grad_norm": 0.462415486574173,
      "learning_rate": 3.902772597289569e-06,
      "loss": 0.7507,
      "step": 5296
    },
    {
      "epoch": 0.21952836835343362,
      "grad_norm": 0.40350303053855896,
      "learning_rate": 3.9025653777612e-06,
      "loss": 0.6743,
      "step": 5297
    },
    {
      "epoch": 0.2195698122591073,
      "grad_norm": 0.4233470857143402,
      "learning_rate": 3.9023581582328326e-06,
      "loss": 0.6796,
      "step": 5298
    },
    {
      "epoch": 0.21961125616478097,
      "grad_norm": 0.4355199933052063,
      "learning_rate": 3.902150938704464e-06,
      "loss": 0.7158,
      "step": 5299
    },
    {
      "epoch": 0.21965270007045465,
      "grad_norm": 0.4205434322357178,
      "learning_rate": 3.901943719176095e-06,
      "loss": 0.6758,
      "step": 5300
    },
    {
      "epoch": 0.2196941439761283,
      "grad_norm": 0.4295797646045685,
      "learning_rate": 3.901736499647727e-06,
      "loss": 0.7261,
      "step": 5301
    },
    {
      "epoch": 0.21973558788180197,
      "grad_norm": 0.4505961537361145,
      "learning_rate": 3.901529280119359e-06,
      "loss": 0.7158,
      "step": 5302
    },
    {
      "epoch": 0.21977703178747565,
      "grad_norm": 0.3953051269054413,
      "learning_rate": 3.90132206059099e-06,
      "loss": 0.7018,
      "step": 5303
    },
    {
      "epoch": 0.21981847569314933,
      "grad_norm": 0.4183301627635956,
      "learning_rate": 3.901114841062622e-06,
      "loss": 0.7075,
      "step": 5304
    },
    {
      "epoch": 0.219859919598823,
      "grad_norm": 0.4219602942466736,
      "learning_rate": 3.900907621534254e-06,
      "loss": 0.6714,
      "step": 5305
    },
    {
      "epoch": 0.21990136350449666,
      "grad_norm": 0.40538957715034485,
      "learning_rate": 3.900700402005885e-06,
      "loss": 0.7947,
      "step": 5306
    },
    {
      "epoch": 0.21994280741017033,
      "grad_norm": 0.4250020682811737,
      "learning_rate": 3.900493182477518e-06,
      "loss": 0.717,
      "step": 5307
    },
    {
      "epoch": 0.219984251315844,
      "grad_norm": 0.4219813644886017,
      "learning_rate": 3.900285962949148e-06,
      "loss": 0.6973,
      "step": 5308
    },
    {
      "epoch": 0.2200256952215177,
      "grad_norm": 0.42195188999176025,
      "learning_rate": 3.90007874342078e-06,
      "loss": 0.6969,
      "step": 5309
    },
    {
      "epoch": 0.22006713912719134,
      "grad_norm": 0.44586315751075745,
      "learning_rate": 3.899871523892412e-06,
      "loss": 0.7317,
      "step": 5310
    },
    {
      "epoch": 0.220108583032865,
      "grad_norm": 0.4271865785121918,
      "learning_rate": 3.899664304364044e-06,
      "loss": 0.6866,
      "step": 5311
    },
    {
      "epoch": 0.2201500269385387,
      "grad_norm": 0.4247032701969147,
      "learning_rate": 3.899457084835675e-06,
      "loss": 0.7034,
      "step": 5312
    },
    {
      "epoch": 0.22019147084421237,
      "grad_norm": 0.38316693902015686,
      "learning_rate": 3.899249865307307e-06,
      "loss": 0.6882,
      "step": 5313
    },
    {
      "epoch": 0.22023291474988602,
      "grad_norm": 0.3932279348373413,
      "learning_rate": 3.899042645778939e-06,
      "loss": 0.6691,
      "step": 5314
    },
    {
      "epoch": 0.2202743586555597,
      "grad_norm": 0.4380730092525482,
      "learning_rate": 3.89883542625057e-06,
      "loss": 0.7395,
      "step": 5315
    },
    {
      "epoch": 0.22031580256123337,
      "grad_norm": 0.41117554903030396,
      "learning_rate": 3.898628206722202e-06,
      "loss": 0.7031,
      "step": 5316
    },
    {
      "epoch": 0.22035724646690705,
      "grad_norm": 0.4566909670829773,
      "learning_rate": 3.898420987193833e-06,
      "loss": 0.7732,
      "step": 5317
    },
    {
      "epoch": 0.22039869037258072,
      "grad_norm": 0.43347689509391785,
      "learning_rate": 3.898213767665465e-06,
      "loss": 0.7449,
      "step": 5318
    },
    {
      "epoch": 0.22044013427825437,
      "grad_norm": 0.41429945826530457,
      "learning_rate": 3.898006548137097e-06,
      "loss": 0.719,
      "step": 5319
    },
    {
      "epoch": 0.22048157818392805,
      "grad_norm": 0.41373735666275024,
      "learning_rate": 3.897799328608728e-06,
      "loss": 0.7329,
      "step": 5320
    },
    {
      "epoch": 0.22052302208960173,
      "grad_norm": 0.46192625164985657,
      "learning_rate": 3.8975921090803595e-06,
      "loss": 0.7244,
      "step": 5321
    },
    {
      "epoch": 0.2205644659952754,
      "grad_norm": 0.4624522626399994,
      "learning_rate": 3.897384889551992e-06,
      "loss": 0.748,
      "step": 5322
    },
    {
      "epoch": 0.22060590990094905,
      "grad_norm": 0.43876153230667114,
      "learning_rate": 3.897177670023624e-06,
      "loss": 0.7549,
      "step": 5323
    },
    {
      "epoch": 0.22064735380662273,
      "grad_norm": 0.45151281356811523,
      "learning_rate": 3.8969704504952545e-06,
      "loss": 0.7551,
      "step": 5324
    },
    {
      "epoch": 0.2206887977122964,
      "grad_norm": 0.43969374895095825,
      "learning_rate": 3.896763230966887e-06,
      "loss": 0.7244,
      "step": 5325
    },
    {
      "epoch": 0.22073024161797009,
      "grad_norm": 0.4198043942451477,
      "learning_rate": 3.896556011438518e-06,
      "loss": 0.7671,
      "step": 5326
    },
    {
      "epoch": 0.22077168552364373,
      "grad_norm": 0.43629544973373413,
      "learning_rate": 3.89634879191015e-06,
      "loss": 0.7296,
      "step": 5327
    },
    {
      "epoch": 0.2208131294293174,
      "grad_norm": 0.39687860012054443,
      "learning_rate": 3.896141572381782e-06,
      "loss": 0.7036,
      "step": 5328
    },
    {
      "epoch": 0.2208545733349911,
      "grad_norm": 0.427781343460083,
      "learning_rate": 3.895934352853413e-06,
      "loss": 0.7545,
      "step": 5329
    },
    {
      "epoch": 0.22089601724066477,
      "grad_norm": 0.4018625020980835,
      "learning_rate": 3.8957271333250445e-06,
      "loss": 0.7361,
      "step": 5330
    },
    {
      "epoch": 0.22093746114633844,
      "grad_norm": 0.4462703764438629,
      "learning_rate": 3.895519913796677e-06,
      "loss": 0.7405,
      "step": 5331
    },
    {
      "epoch": 0.2209789050520121,
      "grad_norm": 0.4512161612510681,
      "learning_rate": 3.895312694268308e-06,
      "loss": 0.7722,
      "step": 5332
    },
    {
      "epoch": 0.22102034895768577,
      "grad_norm": 0.3935202658176422,
      "learning_rate": 3.8951054747399395e-06,
      "loss": 0.7146,
      "step": 5333
    },
    {
      "epoch": 0.22106179286335945,
      "grad_norm": 0.4229908883571625,
      "learning_rate": 3.894898255211572e-06,
      "loss": 0.7461,
      "step": 5334
    },
    {
      "epoch": 0.22110323676903312,
      "grad_norm": 0.4277470111846924,
      "learning_rate": 3.894691035683203e-06,
      "loss": 0.7026,
      "step": 5335
    },
    {
      "epoch": 0.22114468067470677,
      "grad_norm": 0.45363834500312805,
      "learning_rate": 3.8944838161548345e-06,
      "loss": 0.6951,
      "step": 5336
    },
    {
      "epoch": 0.22118612458038045,
      "grad_norm": 0.4042293429374695,
      "learning_rate": 3.894276596626466e-06,
      "loss": 0.7598,
      "step": 5337
    },
    {
      "epoch": 0.22122756848605413,
      "grad_norm": 0.43923282623291016,
      "learning_rate": 3.894069377098098e-06,
      "loss": 0.7407,
      "step": 5338
    },
    {
      "epoch": 0.2212690123917278,
      "grad_norm": 0.44493502378463745,
      "learning_rate": 3.8938621575697296e-06,
      "loss": 0.7014,
      "step": 5339
    },
    {
      "epoch": 0.22131045629740145,
      "grad_norm": 0.41313377022743225,
      "learning_rate": 3.893654938041361e-06,
      "loss": 0.7126,
      "step": 5340
    },
    {
      "epoch": 0.22135190020307513,
      "grad_norm": 0.42617571353912354,
      "learning_rate": 3.893447718512993e-06,
      "loss": 0.7417,
      "step": 5341
    },
    {
      "epoch": 0.2213933441087488,
      "grad_norm": 0.4842333495616913,
      "learning_rate": 3.8932404989846246e-06,
      "loss": 0.7268,
      "step": 5342
    },
    {
      "epoch": 0.22143478801442248,
      "grad_norm": 0.4101858139038086,
      "learning_rate": 3.893033279456257e-06,
      "loss": 0.6903,
      "step": 5343
    },
    {
      "epoch": 0.22147623192009616,
      "grad_norm": 0.41965484619140625,
      "learning_rate": 3.892826059927888e-06,
      "loss": 0.7155,
      "step": 5344
    },
    {
      "epoch": 0.2215176758257698,
      "grad_norm": 0.40693676471710205,
      "learning_rate": 3.8926188403995196e-06,
      "loss": 0.7305,
      "step": 5345
    },
    {
      "epoch": 0.2215591197314435,
      "grad_norm": 0.39963477849960327,
      "learning_rate": 3.892411620871151e-06,
      "loss": 0.7334,
      "step": 5346
    },
    {
      "epoch": 0.22160056363711716,
      "grad_norm": 0.43732550740242004,
      "learning_rate": 3.892204401342783e-06,
      "loss": 0.769,
      "step": 5347
    },
    {
      "epoch": 0.22164200754279084,
      "grad_norm": 0.4208558201789856,
      "learning_rate": 3.8919971818144146e-06,
      "loss": 0.7488,
      "step": 5348
    },
    {
      "epoch": 0.2216834514484645,
      "grad_norm": 0.443558007478714,
      "learning_rate": 3.891789962286046e-06,
      "loss": 0.6896,
      "step": 5349
    },
    {
      "epoch": 0.22172489535413817,
      "grad_norm": 0.4308137595653534,
      "learning_rate": 3.891582742757678e-06,
      "loss": 0.745,
      "step": 5350
    },
    {
      "epoch": 0.22176633925981185,
      "grad_norm": 0.4351682662963867,
      "learning_rate": 3.8913755232293096e-06,
      "loss": 0.7507,
      "step": 5351
    },
    {
      "epoch": 0.22180778316548552,
      "grad_norm": 0.42666029930114746,
      "learning_rate": 3.891168303700941e-06,
      "loss": 0.67,
      "step": 5352
    },
    {
      "epoch": 0.2218492270711592,
      "grad_norm": 0.3992883265018463,
      "learning_rate": 3.890961084172572e-06,
      "loss": 0.6685,
      "step": 5353
    },
    {
      "epoch": 0.22189067097683285,
      "grad_norm": 0.4222527742385864,
      "learning_rate": 3.8907538646442046e-06,
      "loss": 0.771,
      "step": 5354
    },
    {
      "epoch": 0.22193211488250653,
      "grad_norm": 0.38464218378067017,
      "learning_rate": 3.890546645115836e-06,
      "loss": 0.7026,
      "step": 5355
    },
    {
      "epoch": 0.2219735587881802,
      "grad_norm": 0.44592520594596863,
      "learning_rate": 3.890339425587467e-06,
      "loss": 0.7336,
      "step": 5356
    },
    {
      "epoch": 0.22201500269385388,
      "grad_norm": 0.46838390827178955,
      "learning_rate": 3.8901322060590996e-06,
      "loss": 0.7388,
      "step": 5357
    },
    {
      "epoch": 0.22205644659952753,
      "grad_norm": 0.41290557384490967,
      "learning_rate": 3.889924986530731e-06,
      "loss": 0.7246,
      "step": 5358
    },
    {
      "epoch": 0.2220978905052012,
      "grad_norm": 0.4245964288711548,
      "learning_rate": 3.889717767002363e-06,
      "loss": 0.6963,
      "step": 5359
    },
    {
      "epoch": 0.22213933441087488,
      "grad_norm": 0.43396252393722534,
      "learning_rate": 3.889510547473994e-06,
      "loss": 0.7844,
      "step": 5360
    },
    {
      "epoch": 0.22218077831654856,
      "grad_norm": 0.3974953591823578,
      "learning_rate": 3.889303327945626e-06,
      "loss": 0.6871,
      "step": 5361
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.4315508306026459,
      "learning_rate": 3.889096108417257e-06,
      "loss": 0.7207,
      "step": 5362
    },
    {
      "epoch": 0.2222636661278959,
      "grad_norm": 0.42072367668151855,
      "learning_rate": 3.88888888888889e-06,
      "loss": 0.6936,
      "step": 5363
    },
    {
      "epoch": 0.22230511003356956,
      "grad_norm": 0.46010154485702515,
      "learning_rate": 3.888681669360521e-06,
      "loss": 0.7389,
      "step": 5364
    },
    {
      "epoch": 0.22234655393924324,
      "grad_norm": 0.4481925666332245,
      "learning_rate": 3.888474449832152e-06,
      "loss": 0.7495,
      "step": 5365
    },
    {
      "epoch": 0.22238799784491692,
      "grad_norm": 0.4129621982574463,
      "learning_rate": 3.888267230303785e-06,
      "loss": 0.7251,
      "step": 5366
    },
    {
      "epoch": 0.22242944175059057,
      "grad_norm": 0.3947811424732208,
      "learning_rate": 3.888060010775416e-06,
      "loss": 0.6549,
      "step": 5367
    },
    {
      "epoch": 0.22247088565626424,
      "grad_norm": 0.4491768777370453,
      "learning_rate": 3.887852791247047e-06,
      "loss": 0.7793,
      "step": 5368
    },
    {
      "epoch": 0.22251232956193792,
      "grad_norm": 0.42949792742729187,
      "learning_rate": 3.887645571718679e-06,
      "loss": 0.7639,
      "step": 5369
    },
    {
      "epoch": 0.2225537734676116,
      "grad_norm": 0.39363694190979004,
      "learning_rate": 3.887438352190311e-06,
      "loss": 0.7085,
      "step": 5370
    },
    {
      "epoch": 0.22259521737328525,
      "grad_norm": 0.3977954685688019,
      "learning_rate": 3.887231132661942e-06,
      "loss": 0.7415,
      "step": 5371
    },
    {
      "epoch": 0.22263666127895892,
      "grad_norm": 0.40267276763916016,
      "learning_rate": 3.887023913133574e-06,
      "loss": 0.7073,
      "step": 5372
    },
    {
      "epoch": 0.2226781051846326,
      "grad_norm": 0.4195236265659332,
      "learning_rate": 3.886816693605205e-06,
      "loss": 0.6488,
      "step": 5373
    },
    {
      "epoch": 0.22271954909030628,
      "grad_norm": 0.39649805426597595,
      "learning_rate": 3.886609474076837e-06,
      "loss": 0.6953,
      "step": 5374
    },
    {
      "epoch": 0.22276099299597993,
      "grad_norm": 0.459310382604599,
      "learning_rate": 3.88640225454847e-06,
      "loss": 0.7209,
      "step": 5375
    },
    {
      "epoch": 0.2228024369016536,
      "grad_norm": 0.3890974521636963,
      "learning_rate": 3.8861950350201e-06,
      "loss": 0.7395,
      "step": 5376
    },
    {
      "epoch": 0.22284388080732728,
      "grad_norm": 0.444986492395401,
      "learning_rate": 3.885987815491732e-06,
      "loss": 0.7712,
      "step": 5377
    },
    {
      "epoch": 0.22288532471300096,
      "grad_norm": 0.4290887117385864,
      "learning_rate": 3.885780595963364e-06,
      "loss": 0.7334,
      "step": 5378
    },
    {
      "epoch": 0.22292676861867464,
      "grad_norm": 0.411353200674057,
      "learning_rate": 3.885573376434996e-06,
      "loss": 0.7111,
      "step": 5379
    },
    {
      "epoch": 0.22296821252434829,
      "grad_norm": 0.40639862418174744,
      "learning_rate": 3.885366156906627e-06,
      "loss": 0.6843,
      "step": 5380
    },
    {
      "epoch": 0.22300965643002196,
      "grad_norm": 0.42232513427734375,
      "learning_rate": 3.885158937378259e-06,
      "loss": 0.7419,
      "step": 5381
    },
    {
      "epoch": 0.22305110033569564,
      "grad_norm": 0.4078530967235565,
      "learning_rate": 3.88495171784989e-06,
      "loss": 0.6921,
      "step": 5382
    },
    {
      "epoch": 0.22309254424136932,
      "grad_norm": 0.42062216997146606,
      "learning_rate": 3.884744498321522e-06,
      "loss": 0.7333,
      "step": 5383
    },
    {
      "epoch": 0.22313398814704297,
      "grad_norm": 0.38735029101371765,
      "learning_rate": 3.884537278793154e-06,
      "loss": 0.7354,
      "step": 5384
    },
    {
      "epoch": 0.22317543205271664,
      "grad_norm": 0.4239612817764282,
      "learning_rate": 3.884330059264785e-06,
      "loss": 0.6442,
      "step": 5385
    },
    {
      "epoch": 0.22321687595839032,
      "grad_norm": 0.4027920961380005,
      "learning_rate": 3.884122839736417e-06,
      "loss": 0.6699,
      "step": 5386
    },
    {
      "epoch": 0.223258319864064,
      "grad_norm": 0.41494491696357727,
      "learning_rate": 3.883915620208049e-06,
      "loss": 0.7114,
      "step": 5387
    },
    {
      "epoch": 0.22329976376973765,
      "grad_norm": 0.4129941761493683,
      "learning_rate": 3.88370840067968e-06,
      "loss": 0.7188,
      "step": 5388
    },
    {
      "epoch": 0.22334120767541132,
      "grad_norm": 0.39488479495048523,
      "learning_rate": 3.8835011811513115e-06,
      "loss": 0.708,
      "step": 5389
    },
    {
      "epoch": 0.223382651581085,
      "grad_norm": 0.3991526663303375,
      "learning_rate": 3.883293961622944e-06,
      "loss": 0.7112,
      "step": 5390
    },
    {
      "epoch": 0.22342409548675868,
      "grad_norm": 0.4231930077075958,
      "learning_rate": 3.883086742094575e-06,
      "loss": 0.7133,
      "step": 5391
    },
    {
      "epoch": 0.22346553939243236,
      "grad_norm": 0.4067850112915039,
      "learning_rate": 3.8828795225662065e-06,
      "loss": 0.6809,
      "step": 5392
    },
    {
      "epoch": 0.223506983298106,
      "grad_norm": 0.47294390201568604,
      "learning_rate": 3.882672303037839e-06,
      "loss": 0.7595,
      "step": 5393
    },
    {
      "epoch": 0.22354842720377968,
      "grad_norm": 0.38852059841156006,
      "learning_rate": 3.88246508350947e-06,
      "loss": 0.7078,
      "step": 5394
    },
    {
      "epoch": 0.22358987110945336,
      "grad_norm": 0.4436073899269104,
      "learning_rate": 3.882257863981102e-06,
      "loss": 0.7249,
      "step": 5395
    },
    {
      "epoch": 0.22363131501512704,
      "grad_norm": 0.43385428190231323,
      "learning_rate": 3.882050644452734e-06,
      "loss": 0.7432,
      "step": 5396
    },
    {
      "epoch": 0.22367275892080069,
      "grad_norm": 0.4161283075809479,
      "learning_rate": 3.881843424924365e-06,
      "loss": 0.6836,
      "step": 5397
    },
    {
      "epoch": 0.22371420282647436,
      "grad_norm": 0.4175952672958374,
      "learning_rate": 3.8816362053959966e-06,
      "loss": 0.6556,
      "step": 5398
    },
    {
      "epoch": 0.22375564673214804,
      "grad_norm": 0.4315580129623413,
      "learning_rate": 3.881428985867629e-06,
      "loss": 0.7202,
      "step": 5399
    },
    {
      "epoch": 0.22379709063782172,
      "grad_norm": 0.4115661382675171,
      "learning_rate": 3.88122176633926e-06,
      "loss": 0.7024,
      "step": 5400
    },
    {
      "epoch": 0.22383853454349537,
      "grad_norm": 0.46493178606033325,
      "learning_rate": 3.8810145468108916e-06,
      "loss": 0.7329,
      "step": 5401
    },
    {
      "epoch": 0.22387997844916904,
      "grad_norm": 0.42146313190460205,
      "learning_rate": 3.880807327282524e-06,
      "loss": 0.7344,
      "step": 5402
    },
    {
      "epoch": 0.22392142235484272,
      "grad_norm": 0.40090060234069824,
      "learning_rate": 3.880600107754155e-06,
      "loss": 0.7373,
      "step": 5403
    },
    {
      "epoch": 0.2239628662605164,
      "grad_norm": 0.43721842765808105,
      "learning_rate": 3.8803928882257866e-06,
      "loss": 0.7349,
      "step": 5404
    },
    {
      "epoch": 0.22400431016619007,
      "grad_norm": 0.4064861536026001,
      "learning_rate": 3.880185668697418e-06,
      "loss": 0.6781,
      "step": 5405
    },
    {
      "epoch": 0.22404575407186372,
      "grad_norm": 0.459197074174881,
      "learning_rate": 3.87997844916905e-06,
      "loss": 0.7908,
      "step": 5406
    },
    {
      "epoch": 0.2240871979775374,
      "grad_norm": 0.4414818584918976,
      "learning_rate": 3.8797712296406816e-06,
      "loss": 0.7632,
      "step": 5407
    },
    {
      "epoch": 0.22412864188321108,
      "grad_norm": 0.4600870609283447,
      "learning_rate": 3.879564010112313e-06,
      "loss": 0.6802,
      "step": 5408
    },
    {
      "epoch": 0.22417008578888475,
      "grad_norm": 0.4118979275226593,
      "learning_rate": 3.879356790583945e-06,
      "loss": 0.7485,
      "step": 5409
    },
    {
      "epoch": 0.2242115296945584,
      "grad_norm": 0.43717989325523376,
      "learning_rate": 3.8791495710555766e-06,
      "loss": 0.6829,
      "step": 5410
    },
    {
      "epoch": 0.22425297360023208,
      "grad_norm": 0.41334715485572815,
      "learning_rate": 3.878942351527209e-06,
      "loss": 0.6887,
      "step": 5411
    },
    {
      "epoch": 0.22429441750590576,
      "grad_norm": 0.46741175651550293,
      "learning_rate": 3.87873513199884e-06,
      "loss": 0.8096,
      "step": 5412
    },
    {
      "epoch": 0.22433586141157943,
      "grad_norm": 0.38367754220962524,
      "learning_rate": 3.8785279124704716e-06,
      "loss": 0.738,
      "step": 5413
    },
    {
      "epoch": 0.2243773053172531,
      "grad_norm": 0.41923055052757263,
      "learning_rate": 3.878320692942103e-06,
      "loss": 0.7153,
      "step": 5414
    },
    {
      "epoch": 0.22441874922292676,
      "grad_norm": 0.40232938528060913,
      "learning_rate": 3.878113473413735e-06,
      "loss": 0.6848,
      "step": 5415
    },
    {
      "epoch": 0.22446019312860044,
      "grad_norm": 0.43083250522613525,
      "learning_rate": 3.8779062538853666e-06,
      "loss": 0.7502,
      "step": 5416
    },
    {
      "epoch": 0.22450163703427412,
      "grad_norm": 0.40873226523399353,
      "learning_rate": 3.877699034356998e-06,
      "loss": 0.7146,
      "step": 5417
    },
    {
      "epoch": 0.2245430809399478,
      "grad_norm": 0.4067993760108948,
      "learning_rate": 3.87749181482863e-06,
      "loss": 0.6741,
      "step": 5418
    },
    {
      "epoch": 0.22458452484562144,
      "grad_norm": 0.40270358324050903,
      "learning_rate": 3.877284595300262e-06,
      "loss": 0.7146,
      "step": 5419
    },
    {
      "epoch": 0.22462596875129512,
      "grad_norm": 0.4092886447906494,
      "learning_rate": 3.877077375771893e-06,
      "loss": 0.6782,
      "step": 5420
    },
    {
      "epoch": 0.2246674126569688,
      "grad_norm": 0.37673670053482056,
      "learning_rate": 3.876870156243524e-06,
      "loss": 0.7356,
      "step": 5421
    },
    {
      "epoch": 0.22470885656264247,
      "grad_norm": 0.4131222665309906,
      "learning_rate": 3.876662936715157e-06,
      "loss": 0.7356,
      "step": 5422
    },
    {
      "epoch": 0.22475030046831612,
      "grad_norm": 0.36227917671203613,
      "learning_rate": 3.876455717186788e-06,
      "loss": 0.6914,
      "step": 5423
    },
    {
      "epoch": 0.2247917443739898,
      "grad_norm": 0.5417972207069397,
      "learning_rate": 3.876248497658419e-06,
      "loss": 0.7161,
      "step": 5424
    },
    {
      "epoch": 0.22483318827966348,
      "grad_norm": 0.4109330177307129,
      "learning_rate": 3.876041278130051e-06,
      "loss": 0.7234,
      "step": 5425
    },
    {
      "epoch": 0.22487463218533715,
      "grad_norm": 0.4306797981262207,
      "learning_rate": 3.875834058601683e-06,
      "loss": 0.7214,
      "step": 5426
    },
    {
      "epoch": 0.22491607609101083,
      "grad_norm": 0.4222170412540436,
      "learning_rate": 3.875626839073315e-06,
      "loss": 0.761,
      "step": 5427
    },
    {
      "epoch": 0.22495751999668448,
      "grad_norm": 0.3889314532279968,
      "learning_rate": 3.875419619544946e-06,
      "loss": 0.6997,
      "step": 5428
    },
    {
      "epoch": 0.22499896390235816,
      "grad_norm": 0.43970316648483276,
      "learning_rate": 3.875212400016578e-06,
      "loss": 0.6753,
      "step": 5429
    },
    {
      "epoch": 0.22504040780803183,
      "grad_norm": 0.39412930607795715,
      "learning_rate": 3.875005180488209e-06,
      "loss": 0.6882,
      "step": 5430
    },
    {
      "epoch": 0.2250818517137055,
      "grad_norm": 0.4443175792694092,
      "learning_rate": 3.874797960959842e-06,
      "loss": 0.7722,
      "step": 5431
    },
    {
      "epoch": 0.22512329561937916,
      "grad_norm": 0.387222021818161,
      "learning_rate": 3.874590741431473e-06,
      "loss": 0.678,
      "step": 5432
    },
    {
      "epoch": 0.22516473952505284,
      "grad_norm": 0.41984060406684875,
      "learning_rate": 3.874383521903104e-06,
      "loss": 0.7026,
      "step": 5433
    },
    {
      "epoch": 0.22520618343072651,
      "grad_norm": 0.4326939284801483,
      "learning_rate": 3.874176302374736e-06,
      "loss": 0.7444,
      "step": 5434
    },
    {
      "epoch": 0.2252476273364002,
      "grad_norm": 0.41396525502204895,
      "learning_rate": 3.873969082846368e-06,
      "loss": 0.6438,
      "step": 5435
    },
    {
      "epoch": 0.22528907124207384,
      "grad_norm": 0.44119128584861755,
      "learning_rate": 3.873761863317999e-06,
      "loss": 0.7363,
      "step": 5436
    },
    {
      "epoch": 0.22533051514774752,
      "grad_norm": 0.3955668807029724,
      "learning_rate": 3.873554643789631e-06,
      "loss": 0.7057,
      "step": 5437
    },
    {
      "epoch": 0.2253719590534212,
      "grad_norm": 0.45037102699279785,
      "learning_rate": 3.873347424261263e-06,
      "loss": 0.7327,
      "step": 5438
    },
    {
      "epoch": 0.22541340295909487,
      "grad_norm": 0.4175308644771576,
      "learning_rate": 3.873140204732894e-06,
      "loss": 0.7026,
      "step": 5439
    },
    {
      "epoch": 0.22545484686476855,
      "grad_norm": 0.4326030910015106,
      "learning_rate": 3.872932985204526e-06,
      "loss": 0.7246,
      "step": 5440
    },
    {
      "epoch": 0.2254962907704422,
      "grad_norm": 0.41220661997795105,
      "learning_rate": 3.872725765676157e-06,
      "loss": 0.7131,
      "step": 5441
    },
    {
      "epoch": 0.22553773467611588,
      "grad_norm": 0.4504355192184448,
      "learning_rate": 3.872518546147789e-06,
      "loss": 0.7356,
      "step": 5442
    },
    {
      "epoch": 0.22557917858178955,
      "grad_norm": 0.47715362906455994,
      "learning_rate": 3.872311326619421e-06,
      "loss": 0.7252,
      "step": 5443
    },
    {
      "epoch": 0.22562062248746323,
      "grad_norm": 0.41742831468582153,
      "learning_rate": 3.872104107091052e-06,
      "loss": 0.7037,
      "step": 5444
    },
    {
      "epoch": 0.22566206639313688,
      "grad_norm": 0.43994754552841187,
      "learning_rate": 3.871896887562684e-06,
      "loss": 0.7026,
      "step": 5445
    },
    {
      "epoch": 0.22570351029881056,
      "grad_norm": 0.4159255623817444,
      "learning_rate": 3.871689668034316e-06,
      "loss": 0.7087,
      "step": 5446
    },
    {
      "epoch": 0.22574495420448423,
      "grad_norm": 0.43156740069389343,
      "learning_rate": 3.871482448505948e-06,
      "loss": 0.7041,
      "step": 5447
    },
    {
      "epoch": 0.2257863981101579,
      "grad_norm": 0.4280935227870941,
      "learning_rate": 3.871275228977579e-06,
      "loss": 0.6992,
      "step": 5448
    },
    {
      "epoch": 0.22582784201583156,
      "grad_norm": 0.4054677188396454,
      "learning_rate": 3.871068009449211e-06,
      "loss": 0.6553,
      "step": 5449
    },
    {
      "epoch": 0.22586928592150524,
      "grad_norm": 0.4261319637298584,
      "learning_rate": 3.870860789920842e-06,
      "loss": 0.728,
      "step": 5450
    },
    {
      "epoch": 0.2259107298271789,
      "grad_norm": 0.4054911434650421,
      "learning_rate": 3.870653570392474e-06,
      "loss": 0.7097,
      "step": 5451
    },
    {
      "epoch": 0.2259521737328526,
      "grad_norm": 0.42237380146980286,
      "learning_rate": 3.870446350864106e-06,
      "loss": 0.7277,
      "step": 5452
    },
    {
      "epoch": 0.22599361763852627,
      "grad_norm": 0.4473530352115631,
      "learning_rate": 3.870239131335737e-06,
      "loss": 0.7302,
      "step": 5453
    },
    {
      "epoch": 0.22603506154419992,
      "grad_norm": 0.40123751759529114,
      "learning_rate": 3.870031911807369e-06,
      "loss": 0.6848,
      "step": 5454
    },
    {
      "epoch": 0.2260765054498736,
      "grad_norm": 0.4291873574256897,
      "learning_rate": 3.869824692279001e-06,
      "loss": 0.7205,
      "step": 5455
    },
    {
      "epoch": 0.22611794935554727,
      "grad_norm": 0.40258029103279114,
      "learning_rate": 3.869617472750632e-06,
      "loss": 0.6764,
      "step": 5456
    },
    {
      "epoch": 0.22615939326122095,
      "grad_norm": 0.42286384105682373,
      "learning_rate": 3.8694102532222636e-06,
      "loss": 0.7451,
      "step": 5457
    },
    {
      "epoch": 0.2262008371668946,
      "grad_norm": 0.39434486627578735,
      "learning_rate": 3.869203033693896e-06,
      "loss": 0.7441,
      "step": 5458
    },
    {
      "epoch": 0.22624228107256827,
      "grad_norm": 0.4507620930671692,
      "learning_rate": 3.868995814165527e-06,
      "loss": 0.7102,
      "step": 5459
    },
    {
      "epoch": 0.22628372497824195,
      "grad_norm": 0.43697381019592285,
      "learning_rate": 3.8687885946371586e-06,
      "loss": 0.7292,
      "step": 5460
    },
    {
      "epoch": 0.22632516888391563,
      "grad_norm": 0.4523143470287323,
      "learning_rate": 3.868581375108791e-06,
      "loss": 0.783,
      "step": 5461
    },
    {
      "epoch": 0.22636661278958928,
      "grad_norm": 0.44967952370643616,
      "learning_rate": 3.868374155580422e-06,
      "loss": 0.7114,
      "step": 5462
    },
    {
      "epoch": 0.22640805669526295,
      "grad_norm": 0.4223373830318451,
      "learning_rate": 3.868166936052054e-06,
      "loss": 0.7112,
      "step": 5463
    },
    {
      "epoch": 0.22644950060093663,
      "grad_norm": 0.4045344889163971,
      "learning_rate": 3.867959716523686e-06,
      "loss": 0.6956,
      "step": 5464
    },
    {
      "epoch": 0.2264909445066103,
      "grad_norm": 0.4269278049468994,
      "learning_rate": 3.867752496995317e-06,
      "loss": 0.7085,
      "step": 5465
    },
    {
      "epoch": 0.22653238841228399,
      "grad_norm": 0.4227130711078644,
      "learning_rate": 3.8675452774669486e-06,
      "loss": 0.7356,
      "step": 5466
    },
    {
      "epoch": 0.22657383231795764,
      "grad_norm": 0.4446948170661926,
      "learning_rate": 3.867338057938581e-06,
      "loss": 0.7344,
      "step": 5467
    },
    {
      "epoch": 0.2266152762236313,
      "grad_norm": 0.4144286811351776,
      "learning_rate": 3.867130838410212e-06,
      "loss": 0.6869,
      "step": 5468
    },
    {
      "epoch": 0.226656720129305,
      "grad_norm": 0.4508930742740631,
      "learning_rate": 3.8669236188818436e-06,
      "loss": 0.7537,
      "step": 5469
    },
    {
      "epoch": 0.22669816403497867,
      "grad_norm": 0.41933727264404297,
      "learning_rate": 3.866716399353476e-06,
      "loss": 0.7793,
      "step": 5470
    },
    {
      "epoch": 0.22673960794065232,
      "grad_norm": 0.4680531919002533,
      "learning_rate": 3.866509179825107e-06,
      "loss": 0.78,
      "step": 5471
    },
    {
      "epoch": 0.226781051846326,
      "grad_norm": 0.4368168115615845,
      "learning_rate": 3.8663019602967386e-06,
      "loss": 0.7441,
      "step": 5472
    },
    {
      "epoch": 0.22682249575199967,
      "grad_norm": 0.44176942110061646,
      "learning_rate": 3.86609474076837e-06,
      "loss": 0.7323,
      "step": 5473
    },
    {
      "epoch": 0.22686393965767335,
      "grad_norm": 0.408707857131958,
      "learning_rate": 3.865887521240002e-06,
      "loss": 0.7119,
      "step": 5474
    },
    {
      "epoch": 0.226905383563347,
      "grad_norm": 0.4001304507255554,
      "learning_rate": 3.8656803017116336e-06,
      "loss": 0.6742,
      "step": 5475
    },
    {
      "epoch": 0.22694682746902067,
      "grad_norm": 0.42633527517318726,
      "learning_rate": 3.865473082183265e-06,
      "loss": 0.7061,
      "step": 5476
    },
    {
      "epoch": 0.22698827137469435,
      "grad_norm": 0.4415960907936096,
      "learning_rate": 3.865265862654896e-06,
      "loss": 0.719,
      "step": 5477
    },
    {
      "epoch": 0.22702971528036803,
      "grad_norm": 0.4212667644023895,
      "learning_rate": 3.865058643126529e-06,
      "loss": 0.6997,
      "step": 5478
    },
    {
      "epoch": 0.2270711591860417,
      "grad_norm": 0.4382275640964508,
      "learning_rate": 3.864851423598161e-06,
      "loss": 0.7509,
      "step": 5479
    },
    {
      "epoch": 0.22711260309171535,
      "grad_norm": 0.40567347407341003,
      "learning_rate": 3.864644204069792e-06,
      "loss": 0.7084,
      "step": 5480
    },
    {
      "epoch": 0.22715404699738903,
      "grad_norm": 0.417024701833725,
      "learning_rate": 3.864436984541424e-06,
      "loss": 0.7395,
      "step": 5481
    },
    {
      "epoch": 0.2271954909030627,
      "grad_norm": 0.4204524755477905,
      "learning_rate": 3.864229765013055e-06,
      "loss": 0.7161,
      "step": 5482
    },
    {
      "epoch": 0.22723693480873638,
      "grad_norm": 0.43334466218948364,
      "learning_rate": 3.864022545484687e-06,
      "loss": 0.7349,
      "step": 5483
    },
    {
      "epoch": 0.22727837871441003,
      "grad_norm": 0.428235799074173,
      "learning_rate": 3.863815325956319e-06,
      "loss": 0.7878,
      "step": 5484
    },
    {
      "epoch": 0.2273198226200837,
      "grad_norm": 0.46112340688705444,
      "learning_rate": 3.86360810642795e-06,
      "loss": 0.7576,
      "step": 5485
    },
    {
      "epoch": 0.2273612665257574,
      "grad_norm": 0.3978957235813141,
      "learning_rate": 3.863400886899581e-06,
      "loss": 0.6677,
      "step": 5486
    },
    {
      "epoch": 0.22740271043143107,
      "grad_norm": 0.41339531540870667,
      "learning_rate": 3.863193667371214e-06,
      "loss": 0.703,
      "step": 5487
    },
    {
      "epoch": 0.22744415433710474,
      "grad_norm": 0.4287998080253601,
      "learning_rate": 3.862986447842845e-06,
      "loss": 0.6953,
      "step": 5488
    },
    {
      "epoch": 0.2274855982427784,
      "grad_norm": 0.3878413140773773,
      "learning_rate": 3.862779228314476e-06,
      "loss": 0.7263,
      "step": 5489
    },
    {
      "epoch": 0.22752704214845207,
      "grad_norm": 0.40765807032585144,
      "learning_rate": 3.862572008786109e-06,
      "loss": 0.7437,
      "step": 5490
    },
    {
      "epoch": 0.22756848605412575,
      "grad_norm": 0.42820507287979126,
      "learning_rate": 3.86236478925774e-06,
      "loss": 0.6677,
      "step": 5491
    },
    {
      "epoch": 0.22760992995979942,
      "grad_norm": 0.3935137689113617,
      "learning_rate": 3.862157569729371e-06,
      "loss": 0.7051,
      "step": 5492
    },
    {
      "epoch": 0.22765137386547307,
      "grad_norm": 0.4106156826019287,
      "learning_rate": 3.861950350201003e-06,
      "loss": 0.6794,
      "step": 5493
    },
    {
      "epoch": 0.22769281777114675,
      "grad_norm": 0.4227731227874756,
      "learning_rate": 3.861743130672635e-06,
      "loss": 0.7446,
      "step": 5494
    },
    {
      "epoch": 0.22773426167682043,
      "grad_norm": 0.4191523790359497,
      "learning_rate": 3.861535911144266e-06,
      "loss": 0.7449,
      "step": 5495
    },
    {
      "epoch": 0.2277757055824941,
      "grad_norm": 0.42453816533088684,
      "learning_rate": 3.861328691615899e-06,
      "loss": 0.7128,
      "step": 5496
    },
    {
      "epoch": 0.22781714948816775,
      "grad_norm": 0.4352606534957886,
      "learning_rate": 3.86112147208753e-06,
      "loss": 0.6963,
      "step": 5497
    },
    {
      "epoch": 0.22785859339384143,
      "grad_norm": 0.4386427402496338,
      "learning_rate": 3.860914252559161e-06,
      "loss": 0.7229,
      "step": 5498
    },
    {
      "epoch": 0.2279000372995151,
      "grad_norm": 0.4371580481529236,
      "learning_rate": 3.860707033030794e-06,
      "loss": 0.7444,
      "step": 5499
    },
    {
      "epoch": 0.22794148120518878,
      "grad_norm": 0.4059861898422241,
      "learning_rate": 3.860499813502425e-06,
      "loss": 0.7134,
      "step": 5500
    },
    {
      "epoch": 0.22798292511086246,
      "grad_norm": 0.4379953145980835,
      "learning_rate": 3.860292593974056e-06,
      "loss": 0.7385,
      "step": 5501
    },
    {
      "epoch": 0.2280243690165361,
      "grad_norm": 0.40456753969192505,
      "learning_rate": 3.860085374445688e-06,
      "loss": 0.6488,
      "step": 5502
    },
    {
      "epoch": 0.2280658129222098,
      "grad_norm": 0.38490694761276245,
      "learning_rate": 3.85987815491732e-06,
      "loss": 0.7454,
      "step": 5503
    },
    {
      "epoch": 0.22810725682788346,
      "grad_norm": 0.4183681607246399,
      "learning_rate": 3.859670935388951e-06,
      "loss": 0.7253,
      "step": 5504
    },
    {
      "epoch": 0.22814870073355714,
      "grad_norm": 0.40126121044158936,
      "learning_rate": 3.859463715860583e-06,
      "loss": 0.6815,
      "step": 5505
    },
    {
      "epoch": 0.2281901446392308,
      "grad_norm": 0.42413732409477234,
      "learning_rate": 3.859256496332215e-06,
      "loss": 0.6853,
      "step": 5506
    },
    {
      "epoch": 0.22823158854490447,
      "grad_norm": 0.43988099694252014,
      "learning_rate": 3.859049276803846e-06,
      "loss": 0.7333,
      "step": 5507
    },
    {
      "epoch": 0.22827303245057814,
      "grad_norm": 0.4104718565940857,
      "learning_rate": 3.858842057275478e-06,
      "loss": 0.7072,
      "step": 5508
    },
    {
      "epoch": 0.22831447635625182,
      "grad_norm": 0.43918296694755554,
      "learning_rate": 3.858634837747109e-06,
      "loss": 0.7798,
      "step": 5509
    },
    {
      "epoch": 0.22835592026192547,
      "grad_norm": 0.4656141996383667,
      "learning_rate": 3.858427618218741e-06,
      "loss": 0.7103,
      "step": 5510
    },
    {
      "epoch": 0.22839736416759915,
      "grad_norm": 0.42860352993011475,
      "learning_rate": 3.858220398690373e-06,
      "loss": 0.7073,
      "step": 5511
    },
    {
      "epoch": 0.22843880807327283,
      "grad_norm": 0.40856799483299255,
      "learning_rate": 3.858013179162004e-06,
      "loss": 0.7092,
      "step": 5512
    },
    {
      "epoch": 0.2284802519789465,
      "grad_norm": 0.43651264905929565,
      "learning_rate": 3.857805959633636e-06,
      "loss": 0.7434,
      "step": 5513
    },
    {
      "epoch": 0.22852169588462018,
      "grad_norm": 0.3919404447078705,
      "learning_rate": 3.857598740105268e-06,
      "loss": 0.6553,
      "step": 5514
    },
    {
      "epoch": 0.22856313979029383,
      "grad_norm": 0.439002126455307,
      "learning_rate": 3.8573915205769e-06,
      "loss": 0.7007,
      "step": 5515
    },
    {
      "epoch": 0.2286045836959675,
      "grad_norm": 0.46591916680336,
      "learning_rate": 3.857184301048531e-06,
      "loss": 0.7039,
      "step": 5516
    },
    {
      "epoch": 0.22864602760164118,
      "grad_norm": 0.41391071677207947,
      "learning_rate": 3.856977081520163e-06,
      "loss": 0.6882,
      "step": 5517
    },
    {
      "epoch": 0.22868747150731486,
      "grad_norm": 0.483742892742157,
      "learning_rate": 3.856769861991794e-06,
      "loss": 0.7131,
      "step": 5518
    },
    {
      "epoch": 0.2287289154129885,
      "grad_norm": 0.4225747585296631,
      "learning_rate": 3.856562642463426e-06,
      "loss": 0.7288,
      "step": 5519
    },
    {
      "epoch": 0.2287703593186622,
      "grad_norm": 0.40649762749671936,
      "learning_rate": 3.856355422935058e-06,
      "loss": 0.7061,
      "step": 5520
    },
    {
      "epoch": 0.22881180322433586,
      "grad_norm": 0.42282813787460327,
      "learning_rate": 3.856148203406689e-06,
      "loss": 0.7305,
      "step": 5521
    },
    {
      "epoch": 0.22885324713000954,
      "grad_norm": 0.43030551075935364,
      "learning_rate": 3.855940983878321e-06,
      "loss": 0.6943,
      "step": 5522
    },
    {
      "epoch": 0.2288946910356832,
      "grad_norm": 0.403641939163208,
      "learning_rate": 3.855733764349953e-06,
      "loss": 0.6803,
      "step": 5523
    },
    {
      "epoch": 0.22893613494135687,
      "grad_norm": 0.42390596866607666,
      "learning_rate": 3.855526544821584e-06,
      "loss": 0.7913,
      "step": 5524
    },
    {
      "epoch": 0.22897757884703054,
      "grad_norm": 0.40827974677085876,
      "learning_rate": 3.8553193252932156e-06,
      "loss": 0.6919,
      "step": 5525
    },
    {
      "epoch": 0.22901902275270422,
      "grad_norm": 0.4869561195373535,
      "learning_rate": 3.855112105764848e-06,
      "loss": 0.791,
      "step": 5526
    },
    {
      "epoch": 0.2290604666583779,
      "grad_norm": 0.4569469690322876,
      "learning_rate": 3.854904886236479e-06,
      "loss": 0.7087,
      "step": 5527
    },
    {
      "epoch": 0.22910191056405155,
      "grad_norm": 0.40679633617401123,
      "learning_rate": 3.8546976667081106e-06,
      "loss": 0.7145,
      "step": 5528
    },
    {
      "epoch": 0.22914335446972522,
      "grad_norm": 0.41108059883117676,
      "learning_rate": 3.854490447179742e-06,
      "loss": 0.7341,
      "step": 5529
    },
    {
      "epoch": 0.2291847983753989,
      "grad_norm": 0.42208361625671387,
      "learning_rate": 3.854283227651374e-06,
      "loss": 0.752,
      "step": 5530
    },
    {
      "epoch": 0.22922624228107258,
      "grad_norm": 0.4092867076396942,
      "learning_rate": 3.854076008123006e-06,
      "loss": 0.7585,
      "step": 5531
    },
    {
      "epoch": 0.22926768618674623,
      "grad_norm": 0.43245115876197815,
      "learning_rate": 3.853868788594638e-06,
      "loss": 0.6981,
      "step": 5532
    },
    {
      "epoch": 0.2293091300924199,
      "grad_norm": 0.4605277180671692,
      "learning_rate": 3.853661569066269e-06,
      "loss": 0.8114,
      "step": 5533
    },
    {
      "epoch": 0.22935057399809358,
      "grad_norm": 0.4263192415237427,
      "learning_rate": 3.8534543495379006e-06,
      "loss": 0.6931,
      "step": 5534
    },
    {
      "epoch": 0.22939201790376726,
      "grad_norm": 0.42190173268318176,
      "learning_rate": 3.853247130009533e-06,
      "loss": 0.771,
      "step": 5535
    },
    {
      "epoch": 0.2294334618094409,
      "grad_norm": 0.3998081386089325,
      "learning_rate": 3.853039910481164e-06,
      "loss": 0.6938,
      "step": 5536
    },
    {
      "epoch": 0.22947490571511459,
      "grad_norm": 0.41955485939979553,
      "learning_rate": 3.852832690952796e-06,
      "loss": 0.7522,
      "step": 5537
    },
    {
      "epoch": 0.22951634962078826,
      "grad_norm": 0.4481874406337738,
      "learning_rate": 3.852625471424427e-06,
      "loss": 0.6958,
      "step": 5538
    },
    {
      "epoch": 0.22955779352646194,
      "grad_norm": 0.4355217218399048,
      "learning_rate": 3.852418251896059e-06,
      "loss": 0.7839,
      "step": 5539
    },
    {
      "epoch": 0.22959923743213562,
      "grad_norm": 0.45812365412712097,
      "learning_rate": 3.852211032367691e-06,
      "loss": 0.7026,
      "step": 5540
    },
    {
      "epoch": 0.22964068133780927,
      "grad_norm": 0.47549113631248474,
      "learning_rate": 3.852003812839322e-06,
      "loss": 0.7703,
      "step": 5541
    },
    {
      "epoch": 0.22968212524348294,
      "grad_norm": 0.460114985704422,
      "learning_rate": 3.851796593310954e-06,
      "loss": 0.7961,
      "step": 5542
    },
    {
      "epoch": 0.22972356914915662,
      "grad_norm": 0.3793071210384369,
      "learning_rate": 3.851589373782586e-06,
      "loss": 0.6823,
      "step": 5543
    },
    {
      "epoch": 0.2297650130548303,
      "grad_norm": 0.4346812665462494,
      "learning_rate": 3.851382154254217e-06,
      "loss": 0.697,
      "step": 5544
    },
    {
      "epoch": 0.22980645696050395,
      "grad_norm": 0.40741777420043945,
      "learning_rate": 3.851174934725848e-06,
      "loss": 0.7086,
      "step": 5545
    },
    {
      "epoch": 0.22984790086617762,
      "grad_norm": 0.4668141007423401,
      "learning_rate": 3.850967715197481e-06,
      "loss": 0.7402,
      "step": 5546
    },
    {
      "epoch": 0.2298893447718513,
      "grad_norm": 0.4044327139854431,
      "learning_rate": 3.850760495669112e-06,
      "loss": 0.7505,
      "step": 5547
    },
    {
      "epoch": 0.22993078867752498,
      "grad_norm": 0.44644707441329956,
      "learning_rate": 3.850553276140744e-06,
      "loss": 0.7373,
      "step": 5548
    },
    {
      "epoch": 0.22997223258319863,
      "grad_norm": 0.41388368606567383,
      "learning_rate": 3.850346056612376e-06,
      "loss": 0.7742,
      "step": 5549
    },
    {
      "epoch": 0.2300136764888723,
      "grad_norm": 0.42548173666000366,
      "learning_rate": 3.850138837084007e-06,
      "loss": 0.6958,
      "step": 5550
    },
    {
      "epoch": 0.23005512039454598,
      "grad_norm": 0.3944973647594452,
      "learning_rate": 3.849931617555639e-06,
      "loss": 0.7595,
      "step": 5551
    },
    {
      "epoch": 0.23009656430021966,
      "grad_norm": 0.4284839332103729,
      "learning_rate": 3.849724398027271e-06,
      "loss": 0.6643,
      "step": 5552
    },
    {
      "epoch": 0.23013800820589334,
      "grad_norm": 0.4222176969051361,
      "learning_rate": 3.849517178498902e-06,
      "loss": 0.7008,
      "step": 5553
    },
    {
      "epoch": 0.23017945211156698,
      "grad_norm": 0.4128127992153168,
      "learning_rate": 3.849309958970533e-06,
      "loss": 0.7179,
      "step": 5554
    },
    {
      "epoch": 0.23022089601724066,
      "grad_norm": 0.407051146030426,
      "learning_rate": 3.849102739442166e-06,
      "loss": 0.7373,
      "step": 5555
    },
    {
      "epoch": 0.23026233992291434,
      "grad_norm": 0.43033885955810547,
      "learning_rate": 3.848895519913797e-06,
      "loss": 0.7322,
      "step": 5556
    },
    {
      "epoch": 0.23030378382858802,
      "grad_norm": 0.39870449900627136,
      "learning_rate": 3.848688300385428e-06,
      "loss": 0.689,
      "step": 5557
    },
    {
      "epoch": 0.23034522773426166,
      "grad_norm": 0.4337127208709717,
      "learning_rate": 3.848481080857061e-06,
      "loss": 0.6991,
      "step": 5558
    },
    {
      "epoch": 0.23038667163993534,
      "grad_norm": 0.4392353892326355,
      "learning_rate": 3.848273861328692e-06,
      "loss": 0.7349,
      "step": 5559
    },
    {
      "epoch": 0.23042811554560902,
      "grad_norm": 0.4600617587566376,
      "learning_rate": 3.848066641800323e-06,
      "loss": 0.7358,
      "step": 5560
    },
    {
      "epoch": 0.2304695594512827,
      "grad_norm": 0.43855032324790955,
      "learning_rate": 3.847859422271955e-06,
      "loss": 0.7451,
      "step": 5561
    },
    {
      "epoch": 0.23051100335695637,
      "grad_norm": 0.42295798659324646,
      "learning_rate": 3.847652202743587e-06,
      "loss": 0.7455,
      "step": 5562
    },
    {
      "epoch": 0.23055244726263002,
      "grad_norm": 0.4431600868701935,
      "learning_rate": 3.847444983215218e-06,
      "loss": 0.8052,
      "step": 5563
    },
    {
      "epoch": 0.2305938911683037,
      "grad_norm": 0.3967609703540802,
      "learning_rate": 3.847237763686851e-06,
      "loss": 0.6902,
      "step": 5564
    },
    {
      "epoch": 0.23063533507397738,
      "grad_norm": 0.4342823922634125,
      "learning_rate": 3.847030544158481e-06,
      "loss": 0.7316,
      "step": 5565
    },
    {
      "epoch": 0.23067677897965105,
      "grad_norm": 0.44450509548187256,
      "learning_rate": 3.846823324630113e-06,
      "loss": 0.7703,
      "step": 5566
    },
    {
      "epoch": 0.2307182228853247,
      "grad_norm": 0.444663941860199,
      "learning_rate": 3.846616105101746e-06,
      "loss": 0.7219,
      "step": 5567
    },
    {
      "epoch": 0.23075966679099838,
      "grad_norm": 0.4459175169467926,
      "learning_rate": 3.846408885573377e-06,
      "loss": 0.7515,
      "step": 5568
    },
    {
      "epoch": 0.23080111069667206,
      "grad_norm": 0.4098242223262787,
      "learning_rate": 3.846201666045008e-06,
      "loss": 0.6958,
      "step": 5569
    },
    {
      "epoch": 0.23084255460234573,
      "grad_norm": 0.4208475649356842,
      "learning_rate": 3.84599444651664e-06,
      "loss": 0.7028,
      "step": 5570
    },
    {
      "epoch": 0.23088399850801938,
      "grad_norm": 0.4160793423652649,
      "learning_rate": 3.845787226988272e-06,
      "loss": 0.7197,
      "step": 5571
    },
    {
      "epoch": 0.23092544241369306,
      "grad_norm": 0.44672784209251404,
      "learning_rate": 3.845580007459903e-06,
      "loss": 0.7805,
      "step": 5572
    },
    {
      "epoch": 0.23096688631936674,
      "grad_norm": 0.4576667845249176,
      "learning_rate": 3.845372787931535e-06,
      "loss": 0.7522,
      "step": 5573
    },
    {
      "epoch": 0.23100833022504041,
      "grad_norm": 0.42562589049339294,
      "learning_rate": 3.845165568403167e-06,
      "loss": 0.7189,
      "step": 5574
    },
    {
      "epoch": 0.2310497741307141,
      "grad_norm": 0.4158920645713806,
      "learning_rate": 3.844958348874798e-06,
      "loss": 0.7389,
      "step": 5575
    },
    {
      "epoch": 0.23109121803638774,
      "grad_norm": 0.40517657995224,
      "learning_rate": 3.84475112934643e-06,
      "loss": 0.658,
      "step": 5576
    },
    {
      "epoch": 0.23113266194206142,
      "grad_norm": 0.43797409534454346,
      "learning_rate": 3.844543909818061e-06,
      "loss": 0.7194,
      "step": 5577
    },
    {
      "epoch": 0.2311741058477351,
      "grad_norm": 0.4056747555732727,
      "learning_rate": 3.844336690289693e-06,
      "loss": 0.7466,
      "step": 5578
    },
    {
      "epoch": 0.23121554975340877,
      "grad_norm": 0.4311405122280121,
      "learning_rate": 3.844129470761325e-06,
      "loss": 0.7051,
      "step": 5579
    },
    {
      "epoch": 0.23125699365908242,
      "grad_norm": 0.40698176622390747,
      "learning_rate": 3.843922251232956e-06,
      "loss": 0.7419,
      "step": 5580
    },
    {
      "epoch": 0.2312984375647561,
      "grad_norm": 0.39689794182777405,
      "learning_rate": 3.8437150317045876e-06,
      "loss": 0.7014,
      "step": 5581
    },
    {
      "epoch": 0.23133988147042978,
      "grad_norm": 0.3983305096626282,
      "learning_rate": 3.84350781217622e-06,
      "loss": 0.6691,
      "step": 5582
    },
    {
      "epoch": 0.23138132537610345,
      "grad_norm": 0.399709552526474,
      "learning_rate": 3.843300592647852e-06,
      "loss": 0.6902,
      "step": 5583
    },
    {
      "epoch": 0.2314227692817771,
      "grad_norm": 0.3940114676952362,
      "learning_rate": 3.843093373119483e-06,
      "loss": 0.7532,
      "step": 5584
    },
    {
      "epoch": 0.23146421318745078,
      "grad_norm": 0.42690834403038025,
      "learning_rate": 3.842886153591115e-06,
      "loss": 0.7048,
      "step": 5585
    },
    {
      "epoch": 0.23150565709312446,
      "grad_norm": 0.40040233731269836,
      "learning_rate": 3.842678934062746e-06,
      "loss": 0.7151,
      "step": 5586
    },
    {
      "epoch": 0.23154710099879813,
      "grad_norm": 0.41486865282058716,
      "learning_rate": 3.842471714534378e-06,
      "loss": 0.7367,
      "step": 5587
    },
    {
      "epoch": 0.2315885449044718,
      "grad_norm": 0.40725836157798767,
      "learning_rate": 3.84226449500601e-06,
      "loss": 0.6962,
      "step": 5588
    },
    {
      "epoch": 0.23162998881014546,
      "grad_norm": 0.4552771747112274,
      "learning_rate": 3.842057275477641e-06,
      "loss": 0.7098,
      "step": 5589
    },
    {
      "epoch": 0.23167143271581914,
      "grad_norm": 0.4065553843975067,
      "learning_rate": 3.8418500559492726e-06,
      "loss": 0.6941,
      "step": 5590
    },
    {
      "epoch": 0.2317128766214928,
      "grad_norm": 0.3973666727542877,
      "learning_rate": 3.841642836420905e-06,
      "loss": 0.7322,
      "step": 5591
    },
    {
      "epoch": 0.2317543205271665,
      "grad_norm": 0.4085070490837097,
      "learning_rate": 3.841435616892536e-06,
      "loss": 0.7295,
      "step": 5592
    },
    {
      "epoch": 0.23179576443284014,
      "grad_norm": 0.4124259352684021,
      "learning_rate": 3.8412283973641676e-06,
      "loss": 0.6962,
      "step": 5593
    },
    {
      "epoch": 0.23183720833851382,
      "grad_norm": 0.4161011278629303,
      "learning_rate": 3.8410211778358e-06,
      "loss": 0.6648,
      "step": 5594
    },
    {
      "epoch": 0.2318786522441875,
      "grad_norm": 0.43627315759658813,
      "learning_rate": 3.840813958307431e-06,
      "loss": 0.7158,
      "step": 5595
    },
    {
      "epoch": 0.23192009614986117,
      "grad_norm": 0.40295183658599854,
      "learning_rate": 3.840606738779063e-06,
      "loss": 0.7008,
      "step": 5596
    },
    {
      "epoch": 0.23196154005553482,
      "grad_norm": 0.43996402621269226,
      "learning_rate": 3.840399519250694e-06,
      "loss": 0.7021,
      "step": 5597
    },
    {
      "epoch": 0.2320029839612085,
      "grad_norm": 0.4422784447669983,
      "learning_rate": 3.840192299722326e-06,
      "loss": 0.7378,
      "step": 5598
    },
    {
      "epoch": 0.23204442786688217,
      "grad_norm": 0.4222504496574402,
      "learning_rate": 3.839985080193958e-06,
      "loss": 0.7156,
      "step": 5599
    },
    {
      "epoch": 0.23208587177255585,
      "grad_norm": 0.40697091817855835,
      "learning_rate": 3.83977786066559e-06,
      "loss": 0.6831,
      "step": 5600
    },
    {
      "epoch": 0.23212731567822953,
      "grad_norm": 0.41009506583213806,
      "learning_rate": 3.839570641137221e-06,
      "loss": 0.7036,
      "step": 5601
    },
    {
      "epoch": 0.23216875958390318,
      "grad_norm": 0.40400463342666626,
      "learning_rate": 3.839363421608853e-06,
      "loss": 0.7324,
      "step": 5602
    },
    {
      "epoch": 0.23221020348957686,
      "grad_norm": 0.4255528748035431,
      "learning_rate": 3.839156202080485e-06,
      "loss": 0.726,
      "step": 5603
    },
    {
      "epoch": 0.23225164739525053,
      "grad_norm": 0.4111621379852295,
      "learning_rate": 3.838948982552116e-06,
      "loss": 0.6974,
      "step": 5604
    },
    {
      "epoch": 0.2322930913009242,
      "grad_norm": 0.45280954241752625,
      "learning_rate": 3.838741763023748e-06,
      "loss": 0.688,
      "step": 5605
    },
    {
      "epoch": 0.23233453520659786,
      "grad_norm": 0.4439070522785187,
      "learning_rate": 3.838534543495379e-06,
      "loss": 0.7565,
      "step": 5606
    },
    {
      "epoch": 0.23237597911227154,
      "grad_norm": 0.38374313712120056,
      "learning_rate": 3.838327323967011e-06,
      "loss": 0.7056,
      "step": 5607
    },
    {
      "epoch": 0.2324174230179452,
      "grad_norm": 0.4170466661453247,
      "learning_rate": 3.838120104438643e-06,
      "loss": 0.6824,
      "step": 5608
    },
    {
      "epoch": 0.2324588669236189,
      "grad_norm": 0.407936155796051,
      "learning_rate": 3.837912884910274e-06,
      "loss": 0.7627,
      "step": 5609
    },
    {
      "epoch": 0.23250031082929254,
      "grad_norm": 0.4167529046535492,
      "learning_rate": 3.837705665381906e-06,
      "loss": 0.6926,
      "step": 5610
    },
    {
      "epoch": 0.23254175473496622,
      "grad_norm": 0.39114439487457275,
      "learning_rate": 3.837498445853538e-06,
      "loss": 0.6617,
      "step": 5611
    },
    {
      "epoch": 0.2325831986406399,
      "grad_norm": 0.45526212453842163,
      "learning_rate": 3.837291226325169e-06,
      "loss": 0.7552,
      "step": 5612
    },
    {
      "epoch": 0.23262464254631357,
      "grad_norm": 0.4180341958999634,
      "learning_rate": 3.8370840067968e-06,
      "loss": 0.7297,
      "step": 5613
    },
    {
      "epoch": 0.23266608645198725,
      "grad_norm": 0.41361042857170105,
      "learning_rate": 3.836876787268433e-06,
      "loss": 0.6951,
      "step": 5614
    },
    {
      "epoch": 0.2327075303576609,
      "grad_norm": 0.45174503326416016,
      "learning_rate": 3.836669567740064e-06,
      "loss": 0.7603,
      "step": 5615
    },
    {
      "epoch": 0.23274897426333457,
      "grad_norm": 0.4098542630672455,
      "learning_rate": 3.836462348211696e-06,
      "loss": 0.6952,
      "step": 5616
    },
    {
      "epoch": 0.23279041816900825,
      "grad_norm": 0.4468437731266022,
      "learning_rate": 3.836255128683327e-06,
      "loss": 0.8069,
      "step": 5617
    },
    {
      "epoch": 0.23283186207468193,
      "grad_norm": 0.4475123882293701,
      "learning_rate": 3.836047909154959e-06,
      "loss": 0.7653,
      "step": 5618
    },
    {
      "epoch": 0.23287330598035558,
      "grad_norm": 0.41617894172668457,
      "learning_rate": 3.835840689626591e-06,
      "loss": 0.7305,
      "step": 5619
    },
    {
      "epoch": 0.23291474988602925,
      "grad_norm": 0.43787604570388794,
      "learning_rate": 3.835633470098223e-06,
      "loss": 0.7214,
      "step": 5620
    },
    {
      "epoch": 0.23295619379170293,
      "grad_norm": 0.3733513355255127,
      "learning_rate": 3.835426250569854e-06,
      "loss": 0.7084,
      "step": 5621
    },
    {
      "epoch": 0.2329976376973766,
      "grad_norm": 0.38421791791915894,
      "learning_rate": 3.835219031041485e-06,
      "loss": 0.6968,
      "step": 5622
    },
    {
      "epoch": 0.23303908160305026,
      "grad_norm": 0.4115132987499237,
      "learning_rate": 3.835011811513118e-06,
      "loss": 0.7371,
      "step": 5623
    },
    {
      "epoch": 0.23308052550872393,
      "grad_norm": 0.41062331199645996,
      "learning_rate": 3.834804591984749e-06,
      "loss": 0.7639,
      "step": 5624
    },
    {
      "epoch": 0.2331219694143976,
      "grad_norm": 0.4384192228317261,
      "learning_rate": 3.83459737245638e-06,
      "loss": 0.75,
      "step": 5625
    },
    {
      "epoch": 0.2331634133200713,
      "grad_norm": 0.43209701776504517,
      "learning_rate": 3.834390152928012e-06,
      "loss": 0.7104,
      "step": 5626
    },
    {
      "epoch": 0.23320485722574497,
      "grad_norm": 0.4395718276500702,
      "learning_rate": 3.834182933399644e-06,
      "loss": 0.7358,
      "step": 5627
    },
    {
      "epoch": 0.23324630113141862,
      "grad_norm": 0.3931125998497009,
      "learning_rate": 3.833975713871275e-06,
      "loss": 0.675,
      "step": 5628
    },
    {
      "epoch": 0.2332877450370923,
      "grad_norm": 0.3964489996433258,
      "learning_rate": 3.833768494342907e-06,
      "loss": 0.7073,
      "step": 5629
    },
    {
      "epoch": 0.23332918894276597,
      "grad_norm": 0.42992570996284485,
      "learning_rate": 3.833561274814539e-06,
      "loss": 0.6967,
      "step": 5630
    },
    {
      "epoch": 0.23337063284843965,
      "grad_norm": 0.45284217596054077,
      "learning_rate": 3.83335405528617e-06,
      "loss": 0.7554,
      "step": 5631
    },
    {
      "epoch": 0.2334120767541133,
      "grad_norm": 0.4270950257778168,
      "learning_rate": 3.833146835757803e-06,
      "loss": 0.7035,
      "step": 5632
    },
    {
      "epoch": 0.23345352065978697,
      "grad_norm": 0.4388374984264374,
      "learning_rate": 3.832939616229433e-06,
      "loss": 0.7377,
      "step": 5633
    },
    {
      "epoch": 0.23349496456546065,
      "grad_norm": 0.4012737572193146,
      "learning_rate": 3.832732396701065e-06,
      "loss": 0.6885,
      "step": 5634
    },
    {
      "epoch": 0.23353640847113433,
      "grad_norm": 0.4167604148387909,
      "learning_rate": 3.832525177172698e-06,
      "loss": 0.7905,
      "step": 5635
    },
    {
      "epoch": 0.233577852376808,
      "grad_norm": 0.4313300848007202,
      "learning_rate": 3.832317957644329e-06,
      "loss": 0.7479,
      "step": 5636
    },
    {
      "epoch": 0.23361929628248165,
      "grad_norm": 0.3826161324977875,
      "learning_rate": 3.83211073811596e-06,
      "loss": 0.6891,
      "step": 5637
    },
    {
      "epoch": 0.23366074018815533,
      "grad_norm": 0.4370412528514862,
      "learning_rate": 3.831903518587592e-06,
      "loss": 0.7036,
      "step": 5638
    },
    {
      "epoch": 0.233702184093829,
      "grad_norm": 0.4497278332710266,
      "learning_rate": 3.831696299059224e-06,
      "loss": 0.738,
      "step": 5639
    },
    {
      "epoch": 0.23374362799950268,
      "grad_norm": 0.4276197552680969,
      "learning_rate": 3.831489079530855e-06,
      "loss": 0.6699,
      "step": 5640
    },
    {
      "epoch": 0.23378507190517633,
      "grad_norm": 0.4678500294685364,
      "learning_rate": 3.831281860002487e-06,
      "loss": 0.7798,
      "step": 5641
    },
    {
      "epoch": 0.23382651581085,
      "grad_norm": 0.39157119393348694,
      "learning_rate": 3.831074640474118e-06,
      "loss": 0.7185,
      "step": 5642
    },
    {
      "epoch": 0.2338679597165237,
      "grad_norm": 0.451463907957077,
      "learning_rate": 3.83086742094575e-06,
      "loss": 0.7444,
      "step": 5643
    },
    {
      "epoch": 0.23390940362219736,
      "grad_norm": 0.40383991599082947,
      "learning_rate": 3.830660201417382e-06,
      "loss": 0.7285,
      "step": 5644
    },
    {
      "epoch": 0.23395084752787101,
      "grad_norm": 0.42612454295158386,
      "learning_rate": 3.830452981889013e-06,
      "loss": 0.7727,
      "step": 5645
    },
    {
      "epoch": 0.2339922914335447,
      "grad_norm": 0.4258905053138733,
      "learning_rate": 3.830245762360645e-06,
      "loss": 0.7695,
      "step": 5646
    },
    {
      "epoch": 0.23403373533921837,
      "grad_norm": 0.39262568950653076,
      "learning_rate": 3.830038542832277e-06,
      "loss": 0.6816,
      "step": 5647
    },
    {
      "epoch": 0.23407517924489205,
      "grad_norm": 0.421977162361145,
      "learning_rate": 3.829831323303909e-06,
      "loss": 0.7939,
      "step": 5648
    },
    {
      "epoch": 0.23411662315056572,
      "grad_norm": 0.43979400396347046,
      "learning_rate": 3.8296241037755396e-06,
      "loss": 0.7303,
      "step": 5649
    },
    {
      "epoch": 0.23415806705623937,
      "grad_norm": 0.4167293310165405,
      "learning_rate": 3.829416884247172e-06,
      "loss": 0.6978,
      "step": 5650
    },
    {
      "epoch": 0.23419951096191305,
      "grad_norm": 0.4585278332233429,
      "learning_rate": 3.829209664718803e-06,
      "loss": 0.7708,
      "step": 5651
    },
    {
      "epoch": 0.23424095486758673,
      "grad_norm": 0.41107773780822754,
      "learning_rate": 3.829002445190435e-06,
      "loss": 0.7233,
      "step": 5652
    },
    {
      "epoch": 0.2342823987732604,
      "grad_norm": 0.43208828568458557,
      "learning_rate": 3.828795225662067e-06,
      "loss": 0.6813,
      "step": 5653
    },
    {
      "epoch": 0.23432384267893405,
      "grad_norm": 0.4073016345500946,
      "learning_rate": 3.828588006133698e-06,
      "loss": 0.6816,
      "step": 5654
    },
    {
      "epoch": 0.23436528658460773,
      "grad_norm": 0.4553734362125397,
      "learning_rate": 3.8283807866053304e-06,
      "loss": 0.6709,
      "step": 5655
    },
    {
      "epoch": 0.2344067304902814,
      "grad_norm": 0.4247308671474457,
      "learning_rate": 3.828173567076962e-06,
      "loss": 0.7336,
      "step": 5656
    },
    {
      "epoch": 0.23444817439595508,
      "grad_norm": 0.37698736786842346,
      "learning_rate": 3.827966347548593e-06,
      "loss": 0.6797,
      "step": 5657
    },
    {
      "epoch": 0.23448961830162873,
      "grad_norm": 0.4262935221195221,
      "learning_rate": 3.827759128020225e-06,
      "loss": 0.7212,
      "step": 5658
    },
    {
      "epoch": 0.2345310622073024,
      "grad_norm": 0.4136400818824768,
      "learning_rate": 3.827551908491857e-06,
      "loss": 0.7256,
      "step": 5659
    },
    {
      "epoch": 0.2345725061129761,
      "grad_norm": 0.4240547716617584,
      "learning_rate": 3.827344688963488e-06,
      "loss": 0.6829,
      "step": 5660
    },
    {
      "epoch": 0.23461395001864976,
      "grad_norm": 0.4264974892139435,
      "learning_rate": 3.82713746943512e-06,
      "loss": 0.7017,
      "step": 5661
    },
    {
      "epoch": 0.23465539392432344,
      "grad_norm": 0.44271957874298096,
      "learning_rate": 3.826930249906752e-06,
      "loss": 0.7358,
      "step": 5662
    },
    {
      "epoch": 0.2346968378299971,
      "grad_norm": 0.3843737840652466,
      "learning_rate": 3.826723030378383e-06,
      "loss": 0.6865,
      "step": 5663
    },
    {
      "epoch": 0.23473828173567077,
      "grad_norm": 0.454171359539032,
      "learning_rate": 3.826515810850015e-06,
      "loss": 0.7719,
      "step": 5664
    },
    {
      "epoch": 0.23477972564134444,
      "grad_norm": 0.4293948709964752,
      "learning_rate": 3.826308591321646e-06,
      "loss": 0.7341,
      "step": 5665
    },
    {
      "epoch": 0.23482116954701812,
      "grad_norm": 0.4440838098526001,
      "learning_rate": 3.826101371793278e-06,
      "loss": 0.7117,
      "step": 5666
    },
    {
      "epoch": 0.23486261345269177,
      "grad_norm": 0.4099890887737274,
      "learning_rate": 3.82589415226491e-06,
      "loss": 0.7352,
      "step": 5667
    },
    {
      "epoch": 0.23490405735836545,
      "grad_norm": 0.4061611294746399,
      "learning_rate": 3.825686932736542e-06,
      "loss": 0.7332,
      "step": 5668
    },
    {
      "epoch": 0.23494550126403912,
      "grad_norm": 0.4340148866176605,
      "learning_rate": 3.825479713208173e-06,
      "loss": 0.6975,
      "step": 5669
    },
    {
      "epoch": 0.2349869451697128,
      "grad_norm": 0.4196752905845642,
      "learning_rate": 3.825272493679805e-06,
      "loss": 0.6703,
      "step": 5670
    },
    {
      "epoch": 0.23502838907538645,
      "grad_norm": 0.455152302980423,
      "learning_rate": 3.825065274151437e-06,
      "loss": 0.8018,
      "step": 5671
    },
    {
      "epoch": 0.23506983298106013,
      "grad_norm": 0.4546992778778076,
      "learning_rate": 3.824858054623068e-06,
      "loss": 0.75,
      "step": 5672
    },
    {
      "epoch": 0.2351112768867338,
      "grad_norm": 0.42143386602401733,
      "learning_rate": 3.8246508350947e-06,
      "loss": 0.7273,
      "step": 5673
    },
    {
      "epoch": 0.23515272079240748,
      "grad_norm": 0.39995497465133667,
      "learning_rate": 3.824443615566331e-06,
      "loss": 0.7046,
      "step": 5674
    },
    {
      "epoch": 0.23519416469808116,
      "grad_norm": 0.429457426071167,
      "learning_rate": 3.824236396037963e-06,
      "loss": 0.7219,
      "step": 5675
    },
    {
      "epoch": 0.2352356086037548,
      "grad_norm": 0.41710224747657776,
      "learning_rate": 3.824029176509595e-06,
      "loss": 0.7292,
      "step": 5676
    },
    {
      "epoch": 0.23527705250942849,
      "grad_norm": 0.4525255858898163,
      "learning_rate": 3.823821956981226e-06,
      "loss": 0.692,
      "step": 5677
    },
    {
      "epoch": 0.23531849641510216,
      "grad_norm": 0.4212068021297455,
      "learning_rate": 3.823614737452857e-06,
      "loss": 0.7673,
      "step": 5678
    },
    {
      "epoch": 0.23535994032077584,
      "grad_norm": 0.43606922030448914,
      "learning_rate": 3.82340751792449e-06,
      "loss": 0.7627,
      "step": 5679
    },
    {
      "epoch": 0.2354013842264495,
      "grad_norm": 0.5065741539001465,
      "learning_rate": 3.823200298396121e-06,
      "loss": 0.7336,
      "step": 5680
    },
    {
      "epoch": 0.23544282813212317,
      "grad_norm": 0.44878125190734863,
      "learning_rate": 3.822993078867752e-06,
      "loss": 0.7256,
      "step": 5681
    },
    {
      "epoch": 0.23548427203779684,
      "grad_norm": 0.4345020651817322,
      "learning_rate": 3.822785859339385e-06,
      "loss": 0.7488,
      "step": 5682
    },
    {
      "epoch": 0.23552571594347052,
      "grad_norm": 0.42891213297843933,
      "learning_rate": 3.822578639811016e-06,
      "loss": 0.6921,
      "step": 5683
    },
    {
      "epoch": 0.23556715984914417,
      "grad_norm": 0.4383144676685333,
      "learning_rate": 3.822371420282648e-06,
      "loss": 0.7041,
      "step": 5684
    },
    {
      "epoch": 0.23560860375481785,
      "grad_norm": 0.3908795118331909,
      "learning_rate": 3.822164200754279e-06,
      "loss": 0.6926,
      "step": 5685
    },
    {
      "epoch": 0.23565004766049152,
      "grad_norm": 0.3841250240802765,
      "learning_rate": 3.821956981225911e-06,
      "loss": 0.738,
      "step": 5686
    },
    {
      "epoch": 0.2356914915661652,
      "grad_norm": 0.41454118490219116,
      "learning_rate": 3.821749761697542e-06,
      "loss": 0.6569,
      "step": 5687
    },
    {
      "epoch": 0.23573293547183888,
      "grad_norm": 0.43832072615623474,
      "learning_rate": 3.821542542169175e-06,
      "loss": 0.8132,
      "step": 5688
    },
    {
      "epoch": 0.23577437937751253,
      "grad_norm": 0.3900536894798279,
      "learning_rate": 3.821335322640806e-06,
      "loss": 0.6915,
      "step": 5689
    },
    {
      "epoch": 0.2358158232831862,
      "grad_norm": 0.3901899456977844,
      "learning_rate": 3.821128103112437e-06,
      "loss": 0.7581,
      "step": 5690
    },
    {
      "epoch": 0.23585726718885988,
      "grad_norm": 0.39633095264434814,
      "learning_rate": 3.82092088358407e-06,
      "loss": 0.6641,
      "step": 5691
    },
    {
      "epoch": 0.23589871109453356,
      "grad_norm": 0.42954111099243164,
      "learning_rate": 3.820713664055701e-06,
      "loss": 0.6895,
      "step": 5692
    },
    {
      "epoch": 0.2359401550002072,
      "grad_norm": 0.414821982383728,
      "learning_rate": 3.820506444527332e-06,
      "loss": 0.6901,
      "step": 5693
    },
    {
      "epoch": 0.23598159890588088,
      "grad_norm": 0.3972018361091614,
      "learning_rate": 3.820299224998964e-06,
      "loss": 0.769,
      "step": 5694
    },
    {
      "epoch": 0.23602304281155456,
      "grad_norm": 0.4027329683303833,
      "learning_rate": 3.820092005470596e-06,
      "loss": 0.7014,
      "step": 5695
    },
    {
      "epoch": 0.23606448671722824,
      "grad_norm": 0.4301128089427948,
      "learning_rate": 3.819884785942227e-06,
      "loss": 0.7363,
      "step": 5696
    },
    {
      "epoch": 0.23610593062290192,
      "grad_norm": 0.41944462060928345,
      "learning_rate": 3.819677566413859e-06,
      "loss": 0.7151,
      "step": 5697
    },
    {
      "epoch": 0.23614737452857557,
      "grad_norm": 0.4161551296710968,
      "learning_rate": 3.819470346885491e-06,
      "loss": 0.7068,
      "step": 5698
    },
    {
      "epoch": 0.23618881843424924,
      "grad_norm": 0.44620856642723083,
      "learning_rate": 3.819263127357122e-06,
      "loss": 0.7441,
      "step": 5699
    },
    {
      "epoch": 0.23623026233992292,
      "grad_norm": 0.3926694393157959,
      "learning_rate": 3.819055907828755e-06,
      "loss": 0.707,
      "step": 5700
    },
    {
      "epoch": 0.2362717062455966,
      "grad_norm": 0.44288885593414307,
      "learning_rate": 3.818848688300385e-06,
      "loss": 0.7017,
      "step": 5701
    },
    {
      "epoch": 0.23631315015127025,
      "grad_norm": 0.41688084602355957,
      "learning_rate": 3.818641468772017e-06,
      "loss": 0.7346,
      "step": 5702
    },
    {
      "epoch": 0.23635459405694392,
      "grad_norm": 0.4357593059539795,
      "learning_rate": 3.818434249243649e-06,
      "loss": 0.75,
      "step": 5703
    },
    {
      "epoch": 0.2363960379626176,
      "grad_norm": 0.4038029611110687,
      "learning_rate": 3.818227029715281e-06,
      "loss": 0.7051,
      "step": 5704
    },
    {
      "epoch": 0.23643748186829128,
      "grad_norm": 0.41408681869506836,
      "learning_rate": 3.818019810186912e-06,
      "loss": 0.6863,
      "step": 5705
    },
    {
      "epoch": 0.23647892577396493,
      "grad_norm": 0.42879465222358704,
      "learning_rate": 3.817812590658544e-06,
      "loss": 0.7566,
      "step": 5706
    },
    {
      "epoch": 0.2365203696796386,
      "grad_norm": 0.4013672173023224,
      "learning_rate": 3.817605371130176e-06,
      "loss": 0.6781,
      "step": 5707
    },
    {
      "epoch": 0.23656181358531228,
      "grad_norm": 0.4084109961986542,
      "learning_rate": 3.817398151601807e-06,
      "loss": 0.7285,
      "step": 5708
    },
    {
      "epoch": 0.23660325749098596,
      "grad_norm": 0.3864114582538605,
      "learning_rate": 3.817190932073439e-06,
      "loss": 0.6968,
      "step": 5709
    },
    {
      "epoch": 0.23664470139665963,
      "grad_norm": 0.4046151340007782,
      "learning_rate": 3.81698371254507e-06,
      "loss": 0.7205,
      "step": 5710
    },
    {
      "epoch": 0.23668614530233328,
      "grad_norm": 0.4075384736061096,
      "learning_rate": 3.8167764930167024e-06,
      "loss": 0.7157,
      "step": 5711
    },
    {
      "epoch": 0.23672758920800696,
      "grad_norm": 0.49778109788894653,
      "learning_rate": 3.816569273488334e-06,
      "loss": 0.7288,
      "step": 5712
    },
    {
      "epoch": 0.23676903311368064,
      "grad_norm": 0.422605961561203,
      "learning_rate": 3.816362053959965e-06,
      "loss": 0.686,
      "step": 5713
    },
    {
      "epoch": 0.23681047701935432,
      "grad_norm": 0.4236895740032196,
      "learning_rate": 3.8161548344315974e-06,
      "loss": 0.7017,
      "step": 5714
    },
    {
      "epoch": 0.23685192092502796,
      "grad_norm": 0.42503538727760315,
      "learning_rate": 3.815947614903229e-06,
      "loss": 0.6646,
      "step": 5715
    },
    {
      "epoch": 0.23689336483070164,
      "grad_norm": 0.40171271562576294,
      "learning_rate": 3.815740395374861e-06,
      "loss": 0.7039,
      "step": 5716
    },
    {
      "epoch": 0.23693480873637532,
      "grad_norm": 0.46380874514579773,
      "learning_rate": 3.815533175846492e-06,
      "loss": 0.7178,
      "step": 5717
    },
    {
      "epoch": 0.236976252642049,
      "grad_norm": 0.4031495153903961,
      "learning_rate": 3.815325956318124e-06,
      "loss": 0.72,
      "step": 5718
    },
    {
      "epoch": 0.23701769654772264,
      "grad_norm": 0.40494564175605774,
      "learning_rate": 3.815118736789755e-06,
      "loss": 0.6987,
      "step": 5719
    },
    {
      "epoch": 0.23705914045339632,
      "grad_norm": 0.45907774567604065,
      "learning_rate": 3.8149115172613874e-06,
      "loss": 0.7384,
      "step": 5720
    },
    {
      "epoch": 0.23710058435907,
      "grad_norm": 0.40058204531669617,
      "learning_rate": 3.814704297733019e-06,
      "loss": 0.7114,
      "step": 5721
    },
    {
      "epoch": 0.23714202826474368,
      "grad_norm": 0.4196268618106842,
      "learning_rate": 3.81449707820465e-06,
      "loss": 0.7091,
      "step": 5722
    },
    {
      "epoch": 0.23718347217041735,
      "grad_norm": 0.4563756287097931,
      "learning_rate": 3.814289858676282e-06,
      "loss": 0.7146,
      "step": 5723
    },
    {
      "epoch": 0.237224916076091,
      "grad_norm": 0.43273094296455383,
      "learning_rate": 3.8140826391479134e-06,
      "loss": 0.7554,
      "step": 5724
    },
    {
      "epoch": 0.23726635998176468,
      "grad_norm": 0.4640210270881653,
      "learning_rate": 3.8138754196195456e-06,
      "loss": 0.7822,
      "step": 5725
    },
    {
      "epoch": 0.23730780388743836,
      "grad_norm": 0.4169848561286926,
      "learning_rate": 3.8136682000911766e-06,
      "loss": 0.7122,
      "step": 5726
    },
    {
      "epoch": 0.23734924779311203,
      "grad_norm": 0.40893521904945374,
      "learning_rate": 3.813460980562809e-06,
      "loss": 0.6919,
      "step": 5727
    },
    {
      "epoch": 0.23739069169878568,
      "grad_norm": 0.40852832794189453,
      "learning_rate": 3.8132537610344402e-06,
      "loss": 0.7639,
      "step": 5728
    },
    {
      "epoch": 0.23743213560445936,
      "grad_norm": 0.3834085464477539,
      "learning_rate": 3.813046541506072e-06,
      "loss": 0.6548,
      "step": 5729
    },
    {
      "epoch": 0.23747357951013304,
      "grad_norm": 0.38973578810691833,
      "learning_rate": 3.8128393219777034e-06,
      "loss": 0.6938,
      "step": 5730
    },
    {
      "epoch": 0.23751502341580671,
      "grad_norm": 0.4480464458465576,
      "learning_rate": 3.8126321024493352e-06,
      "loss": 0.7878,
      "step": 5731
    },
    {
      "epoch": 0.23755646732148036,
      "grad_norm": 0.4180966019630432,
      "learning_rate": 3.812424882920967e-06,
      "loss": 0.6931,
      "step": 5732
    },
    {
      "epoch": 0.23759791122715404,
      "grad_norm": 0.3828777074813843,
      "learning_rate": 3.8122176633925984e-06,
      "loss": 0.6162,
      "step": 5733
    },
    {
      "epoch": 0.23763935513282772,
      "grad_norm": 0.4276162385940552,
      "learning_rate": 3.8120104438642302e-06,
      "loss": 0.7266,
      "step": 5734
    },
    {
      "epoch": 0.2376807990385014,
      "grad_norm": 0.47124335169792175,
      "learning_rate": 3.8118032243358616e-06,
      "loss": 0.7483,
      "step": 5735
    },
    {
      "epoch": 0.23772224294417507,
      "grad_norm": 0.44580909609794617,
      "learning_rate": 3.8115960048074934e-06,
      "loss": 0.7059,
      "step": 5736
    },
    {
      "epoch": 0.23776368684984872,
      "grad_norm": 0.40236401557922363,
      "learning_rate": 3.811388785279125e-06,
      "loss": 0.7024,
      "step": 5737
    },
    {
      "epoch": 0.2378051307555224,
      "grad_norm": 0.41843754053115845,
      "learning_rate": 3.8111815657507566e-06,
      "loss": 0.6826,
      "step": 5738
    },
    {
      "epoch": 0.23784657466119608,
      "grad_norm": 0.40612679719924927,
      "learning_rate": 3.810974346222388e-06,
      "loss": 0.7069,
      "step": 5739
    },
    {
      "epoch": 0.23788801856686975,
      "grad_norm": 0.44002223014831543,
      "learning_rate": 3.81076712669402e-06,
      "loss": 0.7502,
      "step": 5740
    },
    {
      "epoch": 0.2379294624725434,
      "grad_norm": 0.4455436170101166,
      "learning_rate": 3.8105599071656516e-06,
      "loss": 0.7419,
      "step": 5741
    },
    {
      "epoch": 0.23797090637821708,
      "grad_norm": 0.40342798829078674,
      "learning_rate": 3.810352687637283e-06,
      "loss": 0.7393,
      "step": 5742
    },
    {
      "epoch": 0.23801235028389076,
      "grad_norm": 0.4313731789588928,
      "learning_rate": 3.8101454681089152e-06,
      "loss": 0.7395,
      "step": 5743
    },
    {
      "epoch": 0.23805379418956443,
      "grad_norm": 0.43516477942466736,
      "learning_rate": 3.809938248580546e-06,
      "loss": 0.7454,
      "step": 5744
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 0.4345818758010864,
      "learning_rate": 3.8097310290521784e-06,
      "loss": 0.7108,
      "step": 5745
    },
    {
      "epoch": 0.23813668200091176,
      "grad_norm": 0.41741958260536194,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.7057,
      "step": 5746
    },
    {
      "epoch": 0.23817812590658544,
      "grad_norm": 0.4487571120262146,
      "learning_rate": 3.8093165899954416e-06,
      "loss": 0.7417,
      "step": 5747
    },
    {
      "epoch": 0.2382195698122591,
      "grad_norm": 0.39634135365486145,
      "learning_rate": 3.809109370467073e-06,
      "loss": 0.7078,
      "step": 5748
    },
    {
      "epoch": 0.2382610137179328,
      "grad_norm": 0.41375893354415894,
      "learning_rate": 3.808902150938705e-06,
      "loss": 0.7096,
      "step": 5749
    },
    {
      "epoch": 0.23830245762360644,
      "grad_norm": 0.42929109930992126,
      "learning_rate": 3.8086949314103366e-06,
      "loss": 0.751,
      "step": 5750
    },
    {
      "epoch": 0.23834390152928012,
      "grad_norm": 0.4596971273422241,
      "learning_rate": 3.808487711881968e-06,
      "loss": 0.7849,
      "step": 5751
    },
    {
      "epoch": 0.2383853454349538,
      "grad_norm": 0.41589468717575073,
      "learning_rate": 3.8082804923536e-06,
      "loss": 0.72,
      "step": 5752
    },
    {
      "epoch": 0.23842678934062747,
      "grad_norm": 0.4394682049751282,
      "learning_rate": 3.808073272825231e-06,
      "loss": 0.7524,
      "step": 5753
    },
    {
      "epoch": 0.23846823324630112,
      "grad_norm": 0.4704545736312866,
      "learning_rate": 3.807866053296863e-06,
      "loss": 0.6914,
      "step": 5754
    },
    {
      "epoch": 0.2385096771519748,
      "grad_norm": 0.4443422853946686,
      "learning_rate": 3.8076588337684944e-06,
      "loss": 0.7292,
      "step": 5755
    },
    {
      "epoch": 0.23855112105764847,
      "grad_norm": 0.43635162711143494,
      "learning_rate": 3.8074516142401262e-06,
      "loss": 0.7065,
      "step": 5756
    },
    {
      "epoch": 0.23859256496332215,
      "grad_norm": 0.43623578548431396,
      "learning_rate": 3.807244394711758e-06,
      "loss": 0.6941,
      "step": 5757
    },
    {
      "epoch": 0.2386340088689958,
      "grad_norm": 0.41745543479919434,
      "learning_rate": 3.8070371751833894e-06,
      "loss": 0.7542,
      "step": 5758
    },
    {
      "epoch": 0.23867545277466948,
      "grad_norm": 0.44767022132873535,
      "learning_rate": 3.8068299556550216e-06,
      "loss": 0.7456,
      "step": 5759
    },
    {
      "epoch": 0.23871689668034315,
      "grad_norm": 0.3962811529636383,
      "learning_rate": 3.8066227361266526e-06,
      "loss": 0.7224,
      "step": 5760
    },
    {
      "epoch": 0.23875834058601683,
      "grad_norm": 0.4212293326854706,
      "learning_rate": 3.806415516598285e-06,
      "loss": 0.7104,
      "step": 5761
    },
    {
      "epoch": 0.2387997844916905,
      "grad_norm": 0.414909690618515,
      "learning_rate": 3.8062082970699162e-06,
      "loss": 0.7109,
      "step": 5762
    },
    {
      "epoch": 0.23884122839736416,
      "grad_norm": 0.4372805953025818,
      "learning_rate": 3.806001077541548e-06,
      "loss": 0.7208,
      "step": 5763
    },
    {
      "epoch": 0.23888267230303784,
      "grad_norm": 0.4553775489330292,
      "learning_rate": 3.8057938580131794e-06,
      "loss": 0.6899,
      "step": 5764
    },
    {
      "epoch": 0.2389241162087115,
      "grad_norm": 0.45307597517967224,
      "learning_rate": 3.8055866384848112e-06,
      "loss": 0.7515,
      "step": 5765
    },
    {
      "epoch": 0.2389655601143852,
      "grad_norm": 0.41489067673683167,
      "learning_rate": 3.805379418956443e-06,
      "loss": 0.7096,
      "step": 5766
    },
    {
      "epoch": 0.23900700402005884,
      "grad_norm": 0.4089105725288391,
      "learning_rate": 3.8051721994280744e-06,
      "loss": 0.7048,
      "step": 5767
    },
    {
      "epoch": 0.23904844792573252,
      "grad_norm": 0.4157990515232086,
      "learning_rate": 3.8049649798997062e-06,
      "loss": 0.6785,
      "step": 5768
    },
    {
      "epoch": 0.2390898918314062,
      "grad_norm": 0.4679083228111267,
      "learning_rate": 3.8047577603713376e-06,
      "loss": 0.7137,
      "step": 5769
    },
    {
      "epoch": 0.23913133573707987,
      "grad_norm": 0.42091599106788635,
      "learning_rate": 3.8045505408429694e-06,
      "loss": 0.7498,
      "step": 5770
    },
    {
      "epoch": 0.23917277964275355,
      "grad_norm": 0.4160477817058563,
      "learning_rate": 3.804343321314601e-06,
      "loss": 0.7266,
      "step": 5771
    },
    {
      "epoch": 0.2392142235484272,
      "grad_norm": 0.38867446780204773,
      "learning_rate": 3.8041361017862326e-06,
      "loss": 0.6802,
      "step": 5772
    },
    {
      "epoch": 0.23925566745410087,
      "grad_norm": 0.4366922676563263,
      "learning_rate": 3.803928882257864e-06,
      "loss": 0.7422,
      "step": 5773
    },
    {
      "epoch": 0.23929711135977455,
      "grad_norm": 0.4187376797199249,
      "learning_rate": 3.803721662729496e-06,
      "loss": 0.73,
      "step": 5774
    },
    {
      "epoch": 0.23933855526544823,
      "grad_norm": 0.3953743577003479,
      "learning_rate": 3.803514443201128e-06,
      "loss": 0.7271,
      "step": 5775
    },
    {
      "epoch": 0.23937999917112188,
      "grad_norm": 0.4389801323413849,
      "learning_rate": 3.803307223672759e-06,
      "loss": 0.6765,
      "step": 5776
    },
    {
      "epoch": 0.23942144307679555,
      "grad_norm": 0.4331945776939392,
      "learning_rate": 3.8031000041443912e-06,
      "loss": 0.6956,
      "step": 5777
    },
    {
      "epoch": 0.23946288698246923,
      "grad_norm": 0.4251064658164978,
      "learning_rate": 3.802892784616022e-06,
      "loss": 0.7072,
      "step": 5778
    },
    {
      "epoch": 0.2395043308881429,
      "grad_norm": 0.41855165362358093,
      "learning_rate": 3.8026855650876544e-06,
      "loss": 0.7515,
      "step": 5779
    },
    {
      "epoch": 0.23954577479381656,
      "grad_norm": 0.40960487723350525,
      "learning_rate": 3.802478345559286e-06,
      "loss": 0.7339,
      "step": 5780
    },
    {
      "epoch": 0.23958721869949023,
      "grad_norm": 0.3710502088069916,
      "learning_rate": 3.8022711260309176e-06,
      "loss": 0.7004,
      "step": 5781
    },
    {
      "epoch": 0.2396286626051639,
      "grad_norm": 0.4348476827144623,
      "learning_rate": 3.802063906502549e-06,
      "loss": 0.7471,
      "step": 5782
    },
    {
      "epoch": 0.2396701065108376,
      "grad_norm": 0.40413665771484375,
      "learning_rate": 3.801856686974181e-06,
      "loss": 0.6738,
      "step": 5783
    },
    {
      "epoch": 0.23971155041651127,
      "grad_norm": 0.4238663911819458,
      "learning_rate": 3.8016494674458126e-06,
      "loss": 0.7302,
      "step": 5784
    },
    {
      "epoch": 0.23975299432218491,
      "grad_norm": 0.39110061526298523,
      "learning_rate": 3.801442247917444e-06,
      "loss": 0.7007,
      "step": 5785
    },
    {
      "epoch": 0.2397944382278586,
      "grad_norm": 0.4134226441383362,
      "learning_rate": 3.801235028389076e-06,
      "loss": 0.7375,
      "step": 5786
    },
    {
      "epoch": 0.23983588213353227,
      "grad_norm": 0.3887648284435272,
      "learning_rate": 3.8010278088607072e-06,
      "loss": 0.6868,
      "step": 5787
    },
    {
      "epoch": 0.23987732603920595,
      "grad_norm": 0.39514827728271484,
      "learning_rate": 3.800820589332339e-06,
      "loss": 0.7046,
      "step": 5788
    },
    {
      "epoch": 0.2399187699448796,
      "grad_norm": 0.4230422079563141,
      "learning_rate": 3.8006133698039704e-06,
      "loss": 0.7041,
      "step": 5789
    },
    {
      "epoch": 0.23996021385055327,
      "grad_norm": 0.4029048681259155,
      "learning_rate": 3.8004061502756022e-06,
      "loss": 0.7571,
      "step": 5790
    },
    {
      "epoch": 0.24000165775622695,
      "grad_norm": 0.39887550473213196,
      "learning_rate": 3.8001989307472336e-06,
      "loss": 0.6906,
      "step": 5791
    },
    {
      "epoch": 0.24004310166190063,
      "grad_norm": 0.41522154211997986,
      "learning_rate": 3.7999917112188654e-06,
      "loss": 0.6934,
      "step": 5792
    },
    {
      "epoch": 0.24008454556757428,
      "grad_norm": 0.3844720721244812,
      "learning_rate": 3.7997844916904976e-06,
      "loss": 0.6481,
      "step": 5793
    },
    {
      "epoch": 0.24012598947324795,
      "grad_norm": 0.4491621255874634,
      "learning_rate": 3.7995772721621286e-06,
      "loss": 0.7603,
      "step": 5794
    },
    {
      "epoch": 0.24016743337892163,
      "grad_norm": 0.4107879400253296,
      "learning_rate": 3.799370052633761e-06,
      "loss": 0.7009,
      "step": 5795
    },
    {
      "epoch": 0.2402088772845953,
      "grad_norm": 0.4060125946998596,
      "learning_rate": 3.7991628331053922e-06,
      "loss": 0.7256,
      "step": 5796
    },
    {
      "epoch": 0.24025032119026898,
      "grad_norm": 0.44591009616851807,
      "learning_rate": 3.798955613577024e-06,
      "loss": 0.7134,
      "step": 5797
    },
    {
      "epoch": 0.24029176509594263,
      "grad_norm": 0.42751947045326233,
      "learning_rate": 3.7987483940486554e-06,
      "loss": 0.7222,
      "step": 5798
    },
    {
      "epoch": 0.2403332090016163,
      "grad_norm": 0.4028674066066742,
      "learning_rate": 3.7985411745202872e-06,
      "loss": 0.6912,
      "step": 5799
    },
    {
      "epoch": 0.24037465290729,
      "grad_norm": 0.4358348250389099,
      "learning_rate": 3.7983339549919186e-06,
      "loss": 0.7185,
      "step": 5800
    },
    {
      "epoch": 0.24041609681296366,
      "grad_norm": 0.44847407937049866,
      "learning_rate": 3.7981267354635504e-06,
      "loss": 0.8047,
      "step": 5801
    },
    {
      "epoch": 0.2404575407186373,
      "grad_norm": 0.44925713539123535,
      "learning_rate": 3.7979195159351822e-06,
      "loss": 0.7249,
      "step": 5802
    },
    {
      "epoch": 0.240498984624311,
      "grad_norm": 0.4048302471637726,
      "learning_rate": 3.7977122964068136e-06,
      "loss": 0.7048,
      "step": 5803
    },
    {
      "epoch": 0.24054042852998467,
      "grad_norm": 0.42482465505599976,
      "learning_rate": 3.7975050768784454e-06,
      "loss": 0.665,
      "step": 5804
    },
    {
      "epoch": 0.24058187243565834,
      "grad_norm": 0.44000980257987976,
      "learning_rate": 3.797297857350077e-06,
      "loss": 0.7388,
      "step": 5805
    },
    {
      "epoch": 0.240623316341332,
      "grad_norm": 0.4325239360332489,
      "learning_rate": 3.7970906378217086e-06,
      "loss": 0.7317,
      "step": 5806
    },
    {
      "epoch": 0.24066476024700567,
      "grad_norm": 0.4055592715740204,
      "learning_rate": 3.79688341829334e-06,
      "loss": 0.6676,
      "step": 5807
    },
    {
      "epoch": 0.24070620415267935,
      "grad_norm": 0.45745575428009033,
      "learning_rate": 3.796676198764972e-06,
      "loss": 0.7197,
      "step": 5808
    },
    {
      "epoch": 0.24074764805835303,
      "grad_norm": 0.42545661330223083,
      "learning_rate": 3.796468979236603e-06,
      "loss": 0.7007,
      "step": 5809
    },
    {
      "epoch": 0.2407890919640267,
      "grad_norm": 0.39527949690818787,
      "learning_rate": 3.796261759708235e-06,
      "loss": 0.7102,
      "step": 5810
    },
    {
      "epoch": 0.24083053586970035,
      "grad_norm": 0.4333530366420746,
      "learning_rate": 3.7960545401798672e-06,
      "loss": 0.7373,
      "step": 5811
    },
    {
      "epoch": 0.24087197977537403,
      "grad_norm": 0.44869107007980347,
      "learning_rate": 3.795847320651498e-06,
      "loss": 0.7017,
      "step": 5812
    },
    {
      "epoch": 0.2409134236810477,
      "grad_norm": 0.40590861439704895,
      "learning_rate": 3.7956401011231304e-06,
      "loss": 0.6685,
      "step": 5813
    },
    {
      "epoch": 0.24095486758672138,
      "grad_norm": 0.43441474437713623,
      "learning_rate": 3.795432881594762e-06,
      "loss": 0.7646,
      "step": 5814
    },
    {
      "epoch": 0.24099631149239503,
      "grad_norm": 0.40566182136535645,
      "learning_rate": 3.7952256620663936e-06,
      "loss": 0.6803,
      "step": 5815
    },
    {
      "epoch": 0.2410377553980687,
      "grad_norm": 0.4118754267692566,
      "learning_rate": 3.795018442538025e-06,
      "loss": 0.7126,
      "step": 5816
    },
    {
      "epoch": 0.2410791993037424,
      "grad_norm": 0.4022071957588196,
      "learning_rate": 3.794811223009657e-06,
      "loss": 0.6919,
      "step": 5817
    },
    {
      "epoch": 0.24112064320941606,
      "grad_norm": 0.4032742977142334,
      "learning_rate": 3.7946040034812886e-06,
      "loss": 0.6829,
      "step": 5818
    },
    {
      "epoch": 0.2411620871150897,
      "grad_norm": 0.40466296672821045,
      "learning_rate": 3.79439678395292e-06,
      "loss": 0.6671,
      "step": 5819
    },
    {
      "epoch": 0.2412035310207634,
      "grad_norm": 0.43412697315216064,
      "learning_rate": 3.794189564424552e-06,
      "loss": 0.741,
      "step": 5820
    },
    {
      "epoch": 0.24124497492643707,
      "grad_norm": 0.4570331275463104,
      "learning_rate": 3.7939823448961832e-06,
      "loss": 0.7419,
      "step": 5821
    },
    {
      "epoch": 0.24128641883211074,
      "grad_norm": 0.4352063834667206,
      "learning_rate": 3.793775125367815e-06,
      "loss": 0.7368,
      "step": 5822
    },
    {
      "epoch": 0.24132786273778442,
      "grad_norm": 0.40237486362457275,
      "learning_rate": 3.7935679058394464e-06,
      "loss": 0.7034,
      "step": 5823
    },
    {
      "epoch": 0.24136930664345807,
      "grad_norm": 0.4045490026473999,
      "learning_rate": 3.7933606863110782e-06,
      "loss": 0.7112,
      "step": 5824
    },
    {
      "epoch": 0.24141075054913175,
      "grad_norm": 0.41520121693611145,
      "learning_rate": 3.7931534667827096e-06,
      "loss": 0.7114,
      "step": 5825
    },
    {
      "epoch": 0.24145219445480542,
      "grad_norm": 0.4231642484664917,
      "learning_rate": 3.7929462472543414e-06,
      "loss": 0.7332,
      "step": 5826
    },
    {
      "epoch": 0.2414936383604791,
      "grad_norm": 0.46858036518096924,
      "learning_rate": 3.7927390277259737e-06,
      "loss": 0.7434,
      "step": 5827
    },
    {
      "epoch": 0.24153508226615275,
      "grad_norm": 0.4160548150539398,
      "learning_rate": 3.7925318081976046e-06,
      "loss": 0.729,
      "step": 5828
    },
    {
      "epoch": 0.24157652617182643,
      "grad_norm": 0.4762648642063141,
      "learning_rate": 3.792324588669237e-06,
      "loss": 0.7217,
      "step": 5829
    },
    {
      "epoch": 0.2416179700775001,
      "grad_norm": 0.41812923550605774,
      "learning_rate": 3.7921173691408682e-06,
      "loss": 0.7141,
      "step": 5830
    },
    {
      "epoch": 0.24165941398317378,
      "grad_norm": 0.41029757261276245,
      "learning_rate": 3.7919101496125e-06,
      "loss": 0.7533,
      "step": 5831
    },
    {
      "epoch": 0.24170085788884743,
      "grad_norm": 0.4139564335346222,
      "learning_rate": 3.7917029300841314e-06,
      "loss": 0.6831,
      "step": 5832
    },
    {
      "epoch": 0.2417423017945211,
      "grad_norm": 0.4286637306213379,
      "learning_rate": 3.7914957105557632e-06,
      "loss": 0.7441,
      "step": 5833
    },
    {
      "epoch": 0.24178374570019479,
      "grad_norm": 0.41436415910720825,
      "learning_rate": 3.7912884910273946e-06,
      "loss": 0.7361,
      "step": 5834
    },
    {
      "epoch": 0.24182518960586846,
      "grad_norm": 0.432330459356308,
      "learning_rate": 3.7910812714990264e-06,
      "loss": 0.7693,
      "step": 5835
    },
    {
      "epoch": 0.24186663351154214,
      "grad_norm": 0.47075730562210083,
      "learning_rate": 3.7908740519706582e-06,
      "loss": 0.7302,
      "step": 5836
    },
    {
      "epoch": 0.2419080774172158,
      "grad_norm": 0.4582519233226776,
      "learning_rate": 3.7906668324422896e-06,
      "loss": 0.7432,
      "step": 5837
    },
    {
      "epoch": 0.24194952132288947,
      "grad_norm": 0.4032561182975769,
      "learning_rate": 3.7904596129139214e-06,
      "loss": 0.6575,
      "step": 5838
    },
    {
      "epoch": 0.24199096522856314,
      "grad_norm": 0.4440910220146179,
      "learning_rate": 3.790252393385553e-06,
      "loss": 0.667,
      "step": 5839
    },
    {
      "epoch": 0.24203240913423682,
      "grad_norm": 0.41109806299209595,
      "learning_rate": 3.7900451738571846e-06,
      "loss": 0.7581,
      "step": 5840
    },
    {
      "epoch": 0.24207385303991047,
      "grad_norm": 0.40728530287742615,
      "learning_rate": 3.789837954328816e-06,
      "loss": 0.7285,
      "step": 5841
    },
    {
      "epoch": 0.24211529694558415,
      "grad_norm": 0.4134881794452667,
      "learning_rate": 3.789630734800448e-06,
      "loss": 0.7385,
      "step": 5842
    },
    {
      "epoch": 0.24215674085125782,
      "grad_norm": 0.4340599477291107,
      "learning_rate": 3.789423515272079e-06,
      "loss": 0.7107,
      "step": 5843
    },
    {
      "epoch": 0.2421981847569315,
      "grad_norm": 0.4159688353538513,
      "learning_rate": 3.789216295743711e-06,
      "loss": 0.7017,
      "step": 5844
    },
    {
      "epoch": 0.24223962866260518,
      "grad_norm": 0.4615623950958252,
      "learning_rate": 3.7890090762153433e-06,
      "loss": 0.7747,
      "step": 5845
    },
    {
      "epoch": 0.24228107256827883,
      "grad_norm": 0.3711353540420532,
      "learning_rate": 3.7888018566869742e-06,
      "loss": 0.6298,
      "step": 5846
    },
    {
      "epoch": 0.2423225164739525,
      "grad_norm": 0.41842350363731384,
      "learning_rate": 3.7885946371586064e-06,
      "loss": 0.6559,
      "step": 5847
    },
    {
      "epoch": 0.24236396037962618,
      "grad_norm": 0.43508896231651306,
      "learning_rate": 3.788387417630238e-06,
      "loss": 0.701,
      "step": 5848
    },
    {
      "epoch": 0.24240540428529986,
      "grad_norm": 0.39882227778434753,
      "learning_rate": 3.7881801981018696e-06,
      "loss": 0.7327,
      "step": 5849
    },
    {
      "epoch": 0.2424468481909735,
      "grad_norm": 0.4302796721458435,
      "learning_rate": 3.787972978573501e-06,
      "loss": 0.72,
      "step": 5850
    },
    {
      "epoch": 0.24248829209664718,
      "grad_norm": 0.4106886088848114,
      "learning_rate": 3.787765759045133e-06,
      "loss": 0.6692,
      "step": 5851
    },
    {
      "epoch": 0.24252973600232086,
      "grad_norm": 0.44034191966056824,
      "learning_rate": 3.7875585395167642e-06,
      "loss": 0.6552,
      "step": 5852
    },
    {
      "epoch": 0.24257117990799454,
      "grad_norm": 0.44143223762512207,
      "learning_rate": 3.787351319988396e-06,
      "loss": 0.7223,
      "step": 5853
    },
    {
      "epoch": 0.2426126238136682,
      "grad_norm": 0.4074268341064453,
      "learning_rate": 3.787144100460028e-06,
      "loss": 0.7234,
      "step": 5854
    },
    {
      "epoch": 0.24265406771934186,
      "grad_norm": 0.4701578617095947,
      "learning_rate": 3.7869368809316592e-06,
      "loss": 0.761,
      "step": 5855
    },
    {
      "epoch": 0.24269551162501554,
      "grad_norm": 0.4345472753047943,
      "learning_rate": 3.786729661403291e-06,
      "loss": 0.7097,
      "step": 5856
    },
    {
      "epoch": 0.24273695553068922,
      "grad_norm": 0.4199623763561249,
      "learning_rate": 3.7865224418749224e-06,
      "loss": 0.7659,
      "step": 5857
    },
    {
      "epoch": 0.2427783994363629,
      "grad_norm": 0.44015181064605713,
      "learning_rate": 3.7863152223465542e-06,
      "loss": 0.7378,
      "step": 5858
    },
    {
      "epoch": 0.24281984334203655,
      "grad_norm": 0.38976845145225525,
      "learning_rate": 3.7861080028181856e-06,
      "loss": 0.6832,
      "step": 5859
    },
    {
      "epoch": 0.24286128724771022,
      "grad_norm": 0.4354557991027832,
      "learning_rate": 3.7859007832898174e-06,
      "loss": 0.7,
      "step": 5860
    },
    {
      "epoch": 0.2429027311533839,
      "grad_norm": 0.436708003282547,
      "learning_rate": 3.785693563761449e-06,
      "loss": 0.731,
      "step": 5861
    },
    {
      "epoch": 0.24294417505905758,
      "grad_norm": 0.4433440864086151,
      "learning_rate": 3.7854863442330806e-06,
      "loss": 0.7194,
      "step": 5862
    },
    {
      "epoch": 0.24298561896473123,
      "grad_norm": 0.40738534927368164,
      "learning_rate": 3.785279124704713e-06,
      "loss": 0.6716,
      "step": 5863
    },
    {
      "epoch": 0.2430270628704049,
      "grad_norm": 0.42629101872444153,
      "learning_rate": 3.7850719051763442e-06,
      "loss": 0.681,
      "step": 5864
    },
    {
      "epoch": 0.24306850677607858,
      "grad_norm": 0.41795089840888977,
      "learning_rate": 3.784864685647976e-06,
      "loss": 0.6449,
      "step": 5865
    },
    {
      "epoch": 0.24310995068175226,
      "grad_norm": 0.44087672233581543,
      "learning_rate": 3.7846574661196074e-06,
      "loss": 0.6858,
      "step": 5866
    },
    {
      "epoch": 0.2431513945874259,
      "grad_norm": 0.4304245114326477,
      "learning_rate": 3.7844502465912392e-06,
      "loss": 0.7803,
      "step": 5867
    },
    {
      "epoch": 0.24319283849309958,
      "grad_norm": 0.4216367304325104,
      "learning_rate": 3.7842430270628706e-06,
      "loss": 0.7319,
      "step": 5868
    },
    {
      "epoch": 0.24323428239877326,
      "grad_norm": 0.4494384825229645,
      "learning_rate": 3.7840358075345024e-06,
      "loss": 0.7861,
      "step": 5869
    },
    {
      "epoch": 0.24327572630444694,
      "grad_norm": 0.4380645751953125,
      "learning_rate": 3.7838285880061342e-06,
      "loss": 0.7434,
      "step": 5870
    },
    {
      "epoch": 0.24331717021012061,
      "grad_norm": 0.4153081476688385,
      "learning_rate": 3.7836213684777656e-06,
      "loss": 0.7573,
      "step": 5871
    },
    {
      "epoch": 0.24335861411579426,
      "grad_norm": 0.43208009004592896,
      "learning_rate": 3.7834141489493974e-06,
      "loss": 0.7185,
      "step": 5872
    },
    {
      "epoch": 0.24340005802146794,
      "grad_norm": 0.41795045137405396,
      "learning_rate": 3.783206929421029e-06,
      "loss": 0.6566,
      "step": 5873
    },
    {
      "epoch": 0.24344150192714162,
      "grad_norm": 0.46154290437698364,
      "learning_rate": 3.7829997098926606e-06,
      "loss": 0.7493,
      "step": 5874
    },
    {
      "epoch": 0.2434829458328153,
      "grad_norm": 0.405913382768631,
      "learning_rate": 3.782792490364292e-06,
      "loss": 0.6934,
      "step": 5875
    },
    {
      "epoch": 0.24352438973848894,
      "grad_norm": 0.44273924827575684,
      "learning_rate": 3.782585270835924e-06,
      "loss": 0.7294,
      "step": 5876
    },
    {
      "epoch": 0.24356583364416262,
      "grad_norm": 0.39289596676826477,
      "learning_rate": 3.7823780513075552e-06,
      "loss": 0.655,
      "step": 5877
    },
    {
      "epoch": 0.2436072775498363,
      "grad_norm": 0.37577953934669495,
      "learning_rate": 3.782170831779187e-06,
      "loss": 0.6777,
      "step": 5878
    },
    {
      "epoch": 0.24364872145550998,
      "grad_norm": 0.44401857256889343,
      "learning_rate": 3.7819636122508193e-06,
      "loss": 0.7168,
      "step": 5879
    },
    {
      "epoch": 0.24369016536118362,
      "grad_norm": 0.44974014163017273,
      "learning_rate": 3.7817563927224502e-06,
      "loss": 0.7056,
      "step": 5880
    },
    {
      "epoch": 0.2437316092668573,
      "grad_norm": 0.5078209638595581,
      "learning_rate": 3.7815491731940825e-06,
      "loss": 0.7122,
      "step": 5881
    },
    {
      "epoch": 0.24377305317253098,
      "grad_norm": 0.41901805996894836,
      "learning_rate": 3.781341953665714e-06,
      "loss": 0.7091,
      "step": 5882
    },
    {
      "epoch": 0.24381449707820466,
      "grad_norm": 0.4154927432537079,
      "learning_rate": 3.7811347341373456e-06,
      "loss": 0.7084,
      "step": 5883
    },
    {
      "epoch": 0.24385594098387833,
      "grad_norm": 0.3937971293926239,
      "learning_rate": 3.780927514608977e-06,
      "loss": 0.6868,
      "step": 5884
    },
    {
      "epoch": 0.24389738488955198,
      "grad_norm": 0.4110066592693329,
      "learning_rate": 3.780720295080609e-06,
      "loss": 0.7197,
      "step": 5885
    },
    {
      "epoch": 0.24393882879522566,
      "grad_norm": 0.45111915469169617,
      "learning_rate": 3.7805130755522402e-06,
      "loss": 0.7966,
      "step": 5886
    },
    {
      "epoch": 0.24398027270089934,
      "grad_norm": 0.3754710257053375,
      "learning_rate": 3.780305856023872e-06,
      "loss": 0.7288,
      "step": 5887
    },
    {
      "epoch": 0.244021716606573,
      "grad_norm": 0.4667139947414398,
      "learning_rate": 3.780098636495504e-06,
      "loss": 0.7539,
      "step": 5888
    },
    {
      "epoch": 0.24406316051224666,
      "grad_norm": 0.45286795496940613,
      "learning_rate": 3.7798914169671352e-06,
      "loss": 0.804,
      "step": 5889
    },
    {
      "epoch": 0.24410460441792034,
      "grad_norm": 0.4492534399032593,
      "learning_rate": 3.779684197438767e-06,
      "loss": 0.7129,
      "step": 5890
    },
    {
      "epoch": 0.24414604832359402,
      "grad_norm": 0.42626526951789856,
      "learning_rate": 3.7794769779103984e-06,
      "loss": 0.7632,
      "step": 5891
    },
    {
      "epoch": 0.2441874922292677,
      "grad_norm": 0.42425793409347534,
      "learning_rate": 3.7792697583820302e-06,
      "loss": 0.7212,
      "step": 5892
    },
    {
      "epoch": 0.24422893613494134,
      "grad_norm": 0.3978935778141022,
      "learning_rate": 3.7790625388536616e-06,
      "loss": 0.6788,
      "step": 5893
    },
    {
      "epoch": 0.24427038004061502,
      "grad_norm": 0.4289407730102539,
      "learning_rate": 3.7788553193252934e-06,
      "loss": 0.6829,
      "step": 5894
    },
    {
      "epoch": 0.2443118239462887,
      "grad_norm": 0.4366792142391205,
      "learning_rate": 3.778648099796925e-06,
      "loss": 0.7639,
      "step": 5895
    },
    {
      "epoch": 0.24435326785196237,
      "grad_norm": 0.41078776121139526,
      "learning_rate": 3.7784408802685566e-06,
      "loss": 0.7191,
      "step": 5896
    },
    {
      "epoch": 0.24439471175763605,
      "grad_norm": 0.3971204161643982,
      "learning_rate": 3.778233660740189e-06,
      "loss": 0.6943,
      "step": 5897
    },
    {
      "epoch": 0.2444361556633097,
      "grad_norm": 0.4188547134399414,
      "learning_rate": 3.7780264412118202e-06,
      "loss": 0.7273,
      "step": 5898
    },
    {
      "epoch": 0.24447759956898338,
      "grad_norm": 0.4125039875507355,
      "learning_rate": 3.777819221683452e-06,
      "loss": 0.7036,
      "step": 5899
    },
    {
      "epoch": 0.24451904347465706,
      "grad_norm": 0.42868760228157043,
      "learning_rate": 3.7776120021550834e-06,
      "loss": 0.7161,
      "step": 5900
    },
    {
      "epoch": 0.24456048738033073,
      "grad_norm": 0.4040088355541229,
      "learning_rate": 3.7774047826267152e-06,
      "loss": 0.7164,
      "step": 5901
    },
    {
      "epoch": 0.24460193128600438,
      "grad_norm": 0.4231020212173462,
      "learning_rate": 3.7771975630983466e-06,
      "loss": 0.6956,
      "step": 5902
    },
    {
      "epoch": 0.24464337519167806,
      "grad_norm": 0.40728503465652466,
      "learning_rate": 3.7769903435699784e-06,
      "loss": 0.7153,
      "step": 5903
    },
    {
      "epoch": 0.24468481909735174,
      "grad_norm": 0.4164527952671051,
      "learning_rate": 3.77678312404161e-06,
      "loss": 0.7344,
      "step": 5904
    },
    {
      "epoch": 0.2447262630030254,
      "grad_norm": 0.3739519715309143,
      "learning_rate": 3.7765759045132416e-06,
      "loss": 0.6434,
      "step": 5905
    },
    {
      "epoch": 0.24476770690869906,
      "grad_norm": 0.4463658034801483,
      "learning_rate": 3.7763686849848734e-06,
      "loss": 0.7329,
      "step": 5906
    },
    {
      "epoch": 0.24480915081437274,
      "grad_norm": 0.40119510889053345,
      "learning_rate": 3.776161465456505e-06,
      "loss": 0.667,
      "step": 5907
    },
    {
      "epoch": 0.24485059472004642,
      "grad_norm": 0.43785572052001953,
      "learning_rate": 3.7759542459281366e-06,
      "loss": 0.724,
      "step": 5908
    },
    {
      "epoch": 0.2448920386257201,
      "grad_norm": 0.4276801645755768,
      "learning_rate": 3.775747026399768e-06,
      "loss": 0.687,
      "step": 5909
    },
    {
      "epoch": 0.24493348253139377,
      "grad_norm": 0.41110527515411377,
      "learning_rate": 3.7755398068714e-06,
      "loss": 0.7173,
      "step": 5910
    },
    {
      "epoch": 0.24497492643706742,
      "grad_norm": 0.4754437804222107,
      "learning_rate": 3.7753325873430312e-06,
      "loss": 0.7399,
      "step": 5911
    },
    {
      "epoch": 0.2450163703427411,
      "grad_norm": 0.4362882077693939,
      "learning_rate": 3.775125367814663e-06,
      "loss": 0.6903,
      "step": 5912
    },
    {
      "epoch": 0.24505781424841477,
      "grad_norm": 0.3995305001735687,
      "learning_rate": 3.7749181482862944e-06,
      "loss": 0.6829,
      "step": 5913
    },
    {
      "epoch": 0.24509925815408845,
      "grad_norm": 0.4270041584968567,
      "learning_rate": 3.7747109287579267e-06,
      "loss": 0.7314,
      "step": 5914
    },
    {
      "epoch": 0.2451407020597621,
      "grad_norm": 0.41086867451667786,
      "learning_rate": 3.7745037092295585e-06,
      "loss": 0.7163,
      "step": 5915
    },
    {
      "epoch": 0.24518214596543578,
      "grad_norm": 0.4274124801158905,
      "learning_rate": 3.77429648970119e-06,
      "loss": 0.707,
      "step": 5916
    },
    {
      "epoch": 0.24522358987110945,
      "grad_norm": 0.38177773356437683,
      "learning_rate": 3.7740892701728217e-06,
      "loss": 0.6783,
      "step": 5917
    },
    {
      "epoch": 0.24526503377678313,
      "grad_norm": 0.4351160526275635,
      "learning_rate": 3.773882050644453e-06,
      "loss": 0.7225,
      "step": 5918
    },
    {
      "epoch": 0.2453064776824568,
      "grad_norm": 0.467419296503067,
      "learning_rate": 3.773674831116085e-06,
      "loss": 0.7566,
      "step": 5919
    },
    {
      "epoch": 0.24534792158813046,
      "grad_norm": 0.4119054675102234,
      "learning_rate": 3.7734676115877162e-06,
      "loss": 0.6718,
      "step": 5920
    },
    {
      "epoch": 0.24538936549380413,
      "grad_norm": 0.4179508686065674,
      "learning_rate": 3.773260392059348e-06,
      "loss": 0.6887,
      "step": 5921
    },
    {
      "epoch": 0.2454308093994778,
      "grad_norm": 0.4185778796672821,
      "learning_rate": 3.7730531725309794e-06,
      "loss": 0.7217,
      "step": 5922
    },
    {
      "epoch": 0.2454722533051515,
      "grad_norm": 0.42639216780662537,
      "learning_rate": 3.7728459530026112e-06,
      "loss": 0.6978,
      "step": 5923
    },
    {
      "epoch": 0.24551369721082514,
      "grad_norm": 0.40835005044937134,
      "learning_rate": 3.772638733474243e-06,
      "loss": 0.666,
      "step": 5924
    },
    {
      "epoch": 0.24555514111649882,
      "grad_norm": 0.433062881231308,
      "learning_rate": 3.7724315139458744e-06,
      "loss": 0.6959,
      "step": 5925
    },
    {
      "epoch": 0.2455965850221725,
      "grad_norm": 0.43335452675819397,
      "learning_rate": 3.7722242944175062e-06,
      "loss": 0.7346,
      "step": 5926
    },
    {
      "epoch": 0.24563802892784617,
      "grad_norm": 0.413692831993103,
      "learning_rate": 3.7720170748891376e-06,
      "loss": 0.7021,
      "step": 5927
    },
    {
      "epoch": 0.24567947283351982,
      "grad_norm": 0.40733984112739563,
      "learning_rate": 3.7718098553607694e-06,
      "loss": 0.7085,
      "step": 5928
    },
    {
      "epoch": 0.2457209167391935,
      "grad_norm": 0.4565483033657074,
      "learning_rate": 3.771602635832401e-06,
      "loss": 0.7649,
      "step": 5929
    },
    {
      "epoch": 0.24576236064486717,
      "grad_norm": 0.4183756411075592,
      "learning_rate": 3.7713954163040326e-06,
      "loss": 0.6854,
      "step": 5930
    },
    {
      "epoch": 0.24580380455054085,
      "grad_norm": 0.4421718120574951,
      "learning_rate": 3.771188196775665e-06,
      "loss": 0.7328,
      "step": 5931
    },
    {
      "epoch": 0.24584524845621453,
      "grad_norm": 0.44818130135536194,
      "learning_rate": 3.7709809772472963e-06,
      "loss": 0.7666,
      "step": 5932
    },
    {
      "epoch": 0.24588669236188818,
      "grad_norm": 0.46970635652542114,
      "learning_rate": 3.770773757718928e-06,
      "loss": 0.7161,
      "step": 5933
    },
    {
      "epoch": 0.24592813626756185,
      "grad_norm": 0.4007396697998047,
      "learning_rate": 3.7705665381905594e-06,
      "loss": 0.7092,
      "step": 5934
    },
    {
      "epoch": 0.24596958017323553,
      "grad_norm": 0.41251465678215027,
      "learning_rate": 3.7703593186621913e-06,
      "loss": 0.7422,
      "step": 5935
    },
    {
      "epoch": 0.2460110240789092,
      "grad_norm": 0.41428035497665405,
      "learning_rate": 3.7701520991338226e-06,
      "loss": 0.6802,
      "step": 5936
    },
    {
      "epoch": 0.24605246798458286,
      "grad_norm": 0.4202180504798889,
      "learning_rate": 3.7699448796054545e-06,
      "loss": 0.6816,
      "step": 5937
    },
    {
      "epoch": 0.24609391189025653,
      "grad_norm": 0.3958451449871063,
      "learning_rate": 3.769737660077086e-06,
      "loss": 0.7147,
      "step": 5938
    },
    {
      "epoch": 0.2461353557959302,
      "grad_norm": 0.41043153405189514,
      "learning_rate": 3.7695304405487176e-06,
      "loss": 0.7301,
      "step": 5939
    },
    {
      "epoch": 0.2461767997016039,
      "grad_norm": 0.42235997319221497,
      "learning_rate": 3.7693232210203495e-06,
      "loss": 0.7114,
      "step": 5940
    },
    {
      "epoch": 0.24621824360727754,
      "grad_norm": 0.3827250301837921,
      "learning_rate": 3.769116001491981e-06,
      "loss": 0.7188,
      "step": 5941
    },
    {
      "epoch": 0.24625968751295121,
      "grad_norm": 0.3816659152507782,
      "learning_rate": 3.7689087819636126e-06,
      "loss": 0.741,
      "step": 5942
    },
    {
      "epoch": 0.2463011314186249,
      "grad_norm": 0.4334506690502167,
      "learning_rate": 3.768701562435244e-06,
      "loss": 0.72,
      "step": 5943
    },
    {
      "epoch": 0.24634257532429857,
      "grad_norm": 0.4258240759372711,
      "learning_rate": 3.768494342906876e-06,
      "loss": 0.7051,
      "step": 5944
    },
    {
      "epoch": 0.24638401922997225,
      "grad_norm": 0.41385704278945923,
      "learning_rate": 3.7682871233785072e-06,
      "loss": 0.6339,
      "step": 5945
    },
    {
      "epoch": 0.2464254631356459,
      "grad_norm": 0.4213573932647705,
      "learning_rate": 3.768079903850139e-06,
      "loss": 0.7401,
      "step": 5946
    },
    {
      "epoch": 0.24646690704131957,
      "grad_norm": 0.37417614459991455,
      "learning_rate": 3.7678726843217704e-06,
      "loss": 0.7351,
      "step": 5947
    },
    {
      "epoch": 0.24650835094699325,
      "grad_norm": 0.4234601855278015,
      "learning_rate": 3.7676654647934027e-06,
      "loss": 0.7502,
      "step": 5948
    },
    {
      "epoch": 0.24654979485266693,
      "grad_norm": 0.42342886328697205,
      "learning_rate": 3.7674582452650345e-06,
      "loss": 0.6682,
      "step": 5949
    },
    {
      "epoch": 0.24659123875834058,
      "grad_norm": 0.4131332039833069,
      "learning_rate": 3.767251025736666e-06,
      "loss": 0.7109,
      "step": 5950
    },
    {
      "epoch": 0.24663268266401425,
      "grad_norm": 0.3708457052707672,
      "learning_rate": 3.7670438062082977e-06,
      "loss": 0.6621,
      "step": 5951
    },
    {
      "epoch": 0.24667412656968793,
      "grad_norm": 0.45090943574905396,
      "learning_rate": 3.766836586679929e-06,
      "loss": 0.7432,
      "step": 5952
    },
    {
      "epoch": 0.2467155704753616,
      "grad_norm": 0.4185495972633362,
      "learning_rate": 3.766629367151561e-06,
      "loss": 0.7048,
      "step": 5953
    },
    {
      "epoch": 0.24675701438103526,
      "grad_norm": 0.44534870982170105,
      "learning_rate": 3.7664221476231922e-06,
      "loss": 0.7883,
      "step": 5954
    },
    {
      "epoch": 0.24679845828670893,
      "grad_norm": 0.4371650815010071,
      "learning_rate": 3.766214928094824e-06,
      "loss": 0.7205,
      "step": 5955
    },
    {
      "epoch": 0.2468399021923826,
      "grad_norm": 0.4230118989944458,
      "learning_rate": 3.7660077085664554e-06,
      "loss": 0.7163,
      "step": 5956
    },
    {
      "epoch": 0.2468813460980563,
      "grad_norm": 0.4045245051383972,
      "learning_rate": 3.7658004890380872e-06,
      "loss": 0.7007,
      "step": 5957
    },
    {
      "epoch": 0.24692279000372996,
      "grad_norm": 0.4006343483924866,
      "learning_rate": 3.765593269509719e-06,
      "loss": 0.726,
      "step": 5958
    },
    {
      "epoch": 0.2469642339094036,
      "grad_norm": 0.4007108509540558,
      "learning_rate": 3.7653860499813504e-06,
      "loss": 0.6753,
      "step": 5959
    },
    {
      "epoch": 0.2470056778150773,
      "grad_norm": 0.477323055267334,
      "learning_rate": 3.7651788304529822e-06,
      "loss": 0.7179,
      "step": 5960
    },
    {
      "epoch": 0.24704712172075097,
      "grad_norm": 0.3884926438331604,
      "learning_rate": 3.7649716109246136e-06,
      "loss": 0.6909,
      "step": 5961
    },
    {
      "epoch": 0.24708856562642464,
      "grad_norm": 0.4601760804653168,
      "learning_rate": 3.7647643913962454e-06,
      "loss": 0.772,
      "step": 5962
    },
    {
      "epoch": 0.2471300095320983,
      "grad_norm": 0.46465587615966797,
      "learning_rate": 3.764557171867877e-06,
      "loss": 0.7383,
      "step": 5963
    },
    {
      "epoch": 0.24717145343777197,
      "grad_norm": 0.3981333076953888,
      "learning_rate": 3.7643499523395086e-06,
      "loss": 0.708,
      "step": 5964
    },
    {
      "epoch": 0.24721289734344565,
      "grad_norm": 0.41850435733795166,
      "learning_rate": 3.76414273281114e-06,
      "loss": 0.6968,
      "step": 5965
    },
    {
      "epoch": 0.24725434124911932,
      "grad_norm": 0.41070640087127686,
      "learning_rate": 3.7639355132827723e-06,
      "loss": 0.7349,
      "step": 5966
    },
    {
      "epoch": 0.24729578515479297,
      "grad_norm": 0.3784337639808655,
      "learning_rate": 3.763728293754404e-06,
      "loss": 0.6481,
      "step": 5967
    },
    {
      "epoch": 0.24733722906046665,
      "grad_norm": 0.43608611822128296,
      "learning_rate": 3.7635210742260355e-06,
      "loss": 0.7302,
      "step": 5968
    },
    {
      "epoch": 0.24737867296614033,
      "grad_norm": 0.45348963141441345,
      "learning_rate": 3.7633138546976673e-06,
      "loss": 0.7368,
      "step": 5969
    },
    {
      "epoch": 0.247420116871814,
      "grad_norm": 0.4023630917072296,
      "learning_rate": 3.7631066351692986e-06,
      "loss": 0.6952,
      "step": 5970
    },
    {
      "epoch": 0.24746156077748768,
      "grad_norm": 0.4500887095928192,
      "learning_rate": 3.7628994156409305e-06,
      "loss": 0.7659,
      "step": 5971
    },
    {
      "epoch": 0.24750300468316133,
      "grad_norm": 0.4325196146965027,
      "learning_rate": 3.762692196112562e-06,
      "loss": 0.7417,
      "step": 5972
    },
    {
      "epoch": 0.247544448588835,
      "grad_norm": 0.3835587203502655,
      "learning_rate": 3.7624849765841937e-06,
      "loss": 0.6768,
      "step": 5973
    },
    {
      "epoch": 0.24758589249450869,
      "grad_norm": 0.42258793115615845,
      "learning_rate": 3.762277757055825e-06,
      "loss": 0.7471,
      "step": 5974
    },
    {
      "epoch": 0.24762733640018236,
      "grad_norm": 0.4325827658176422,
      "learning_rate": 3.762070537527457e-06,
      "loss": 0.7296,
      "step": 5975
    },
    {
      "epoch": 0.247668780305856,
      "grad_norm": 0.44545701146125793,
      "learning_rate": 3.7618633179990887e-06,
      "loss": 0.7595,
      "step": 5976
    },
    {
      "epoch": 0.2477102242115297,
      "grad_norm": 0.42231130599975586,
      "learning_rate": 3.76165609847072e-06,
      "loss": 0.677,
      "step": 5977
    },
    {
      "epoch": 0.24775166811720337,
      "grad_norm": 0.40663230419158936,
      "learning_rate": 3.761448878942352e-06,
      "loss": 0.6736,
      "step": 5978
    },
    {
      "epoch": 0.24779311202287704,
      "grad_norm": 0.44123294949531555,
      "learning_rate": 3.7612416594139832e-06,
      "loss": 0.7395,
      "step": 5979
    },
    {
      "epoch": 0.24783455592855072,
      "grad_norm": 0.4406220316886902,
      "learning_rate": 3.761034439885615e-06,
      "loss": 0.7223,
      "step": 5980
    },
    {
      "epoch": 0.24787599983422437,
      "grad_norm": 0.42522433400154114,
      "learning_rate": 3.7608272203572464e-06,
      "loss": 0.7531,
      "step": 5981
    },
    {
      "epoch": 0.24791744373989805,
      "grad_norm": 0.4258928894996643,
      "learning_rate": 3.7606200008288787e-06,
      "loss": 0.7332,
      "step": 5982
    },
    {
      "epoch": 0.24795888764557172,
      "grad_norm": 0.45766928791999817,
      "learning_rate": 3.7604127813005096e-06,
      "loss": 0.7712,
      "step": 5983
    },
    {
      "epoch": 0.2480003315512454,
      "grad_norm": 0.41440561413764954,
      "learning_rate": 3.760205561772142e-06,
      "loss": 0.7668,
      "step": 5984
    },
    {
      "epoch": 0.24804177545691905,
      "grad_norm": 0.3836019039154053,
      "learning_rate": 3.7599983422437737e-06,
      "loss": 0.676,
      "step": 5985
    },
    {
      "epoch": 0.24808321936259273,
      "grad_norm": 0.4378919005393982,
      "learning_rate": 3.759791122715405e-06,
      "loss": 0.7791,
      "step": 5986
    },
    {
      "epoch": 0.2481246632682664,
      "grad_norm": 0.443541944026947,
      "learning_rate": 3.759583903187037e-06,
      "loss": 0.754,
      "step": 5987
    },
    {
      "epoch": 0.24816610717394008,
      "grad_norm": 0.4530448317527771,
      "learning_rate": 3.7593766836586682e-06,
      "loss": 0.7437,
      "step": 5988
    },
    {
      "epoch": 0.24820755107961373,
      "grad_norm": 0.41682401299476624,
      "learning_rate": 3.7591694641303e-06,
      "loss": 0.7129,
      "step": 5989
    },
    {
      "epoch": 0.2482489949852874,
      "grad_norm": 0.4373258650302887,
      "learning_rate": 3.7589622446019314e-06,
      "loss": 0.738,
      "step": 5990
    },
    {
      "epoch": 0.24829043889096108,
      "grad_norm": 0.4031035602092743,
      "learning_rate": 3.7587550250735633e-06,
      "loss": 0.7046,
      "step": 5991
    },
    {
      "epoch": 0.24833188279663476,
      "grad_norm": 0.40147364139556885,
      "learning_rate": 3.758547805545195e-06,
      "loss": 0.7228,
      "step": 5992
    },
    {
      "epoch": 0.24837332670230844,
      "grad_norm": 0.4440365731716156,
      "learning_rate": 3.7583405860168264e-06,
      "loss": 0.7961,
      "step": 5993
    },
    {
      "epoch": 0.2484147706079821,
      "grad_norm": 0.44299235939979553,
      "learning_rate": 3.7581333664884583e-06,
      "loss": 0.7114,
      "step": 5994
    },
    {
      "epoch": 0.24845621451365577,
      "grad_norm": 0.4473583996295929,
      "learning_rate": 3.7579261469600896e-06,
      "loss": 0.72,
      "step": 5995
    },
    {
      "epoch": 0.24849765841932944,
      "grad_norm": 0.41851088404655457,
      "learning_rate": 3.7577189274317215e-06,
      "loss": 0.6755,
      "step": 5996
    },
    {
      "epoch": 0.24853910232500312,
      "grad_norm": 0.4060959815979004,
      "learning_rate": 3.757511707903353e-06,
      "loss": 0.7395,
      "step": 5997
    },
    {
      "epoch": 0.24858054623067677,
      "grad_norm": 0.3918766677379608,
      "learning_rate": 3.7573044883749846e-06,
      "loss": 0.6843,
      "step": 5998
    },
    {
      "epoch": 0.24862199013635045,
      "grad_norm": 0.45000261068344116,
      "learning_rate": 3.757097268846616e-06,
      "loss": 0.7712,
      "step": 5999
    },
    {
      "epoch": 0.24866343404202412,
      "grad_norm": 0.4027317762374878,
      "learning_rate": 3.7568900493182483e-06,
      "loss": 0.7156,
      "step": 6000
    },
    {
      "epoch": 0.2487048779476978,
      "grad_norm": 0.3953678011894226,
      "learning_rate": 3.75668282978988e-06,
      "loss": 0.6929,
      "step": 6001
    },
    {
      "epoch": 0.24874632185337145,
      "grad_norm": 0.4020349383354187,
      "learning_rate": 3.7564756102615115e-06,
      "loss": 0.7173,
      "step": 6002
    },
    {
      "epoch": 0.24878776575904513,
      "grad_norm": 0.4084712564945221,
      "learning_rate": 3.7562683907331433e-06,
      "loss": 0.7018,
      "step": 6003
    },
    {
      "epoch": 0.2488292096647188,
      "grad_norm": 0.4320645332336426,
      "learning_rate": 3.7560611712047747e-06,
      "loss": 0.694,
      "step": 6004
    },
    {
      "epoch": 0.24887065357039248,
      "grad_norm": 0.4257984459400177,
      "learning_rate": 3.7558539516764065e-06,
      "loss": 0.6912,
      "step": 6005
    },
    {
      "epoch": 0.24891209747606616,
      "grad_norm": 0.437749445438385,
      "learning_rate": 3.755646732148038e-06,
      "loss": 0.7456,
      "step": 6006
    },
    {
      "epoch": 0.2489535413817398,
      "grad_norm": 0.4243311285972595,
      "learning_rate": 3.7554395126196697e-06,
      "loss": 0.7141,
      "step": 6007
    },
    {
      "epoch": 0.24899498528741348,
      "grad_norm": 0.4369007349014282,
      "learning_rate": 3.755232293091301e-06,
      "loss": 0.7524,
      "step": 6008
    },
    {
      "epoch": 0.24903642919308716,
      "grad_norm": 0.4096597731113434,
      "learning_rate": 3.755025073562933e-06,
      "loss": 0.7622,
      "step": 6009
    },
    {
      "epoch": 0.24907787309876084,
      "grad_norm": 0.41250327229499817,
      "learning_rate": 3.7548178540345647e-06,
      "loss": 0.718,
      "step": 6010
    },
    {
      "epoch": 0.2491193170044345,
      "grad_norm": 0.411704957485199,
      "learning_rate": 3.754610634506196e-06,
      "loss": 0.6879,
      "step": 6011
    },
    {
      "epoch": 0.24916076091010816,
      "grad_norm": 0.4320193827152252,
      "learning_rate": 3.754403414977828e-06,
      "loss": 0.6921,
      "step": 6012
    },
    {
      "epoch": 0.24920220481578184,
      "grad_norm": 0.3827728033065796,
      "learning_rate": 3.7541961954494592e-06,
      "loss": 0.6578,
      "step": 6013
    },
    {
      "epoch": 0.24924364872145552,
      "grad_norm": 0.42136314511299133,
      "learning_rate": 3.753988975921091e-06,
      "loss": 0.7383,
      "step": 6014
    },
    {
      "epoch": 0.24928509262712917,
      "grad_norm": 0.4044662117958069,
      "learning_rate": 3.7537817563927224e-06,
      "loss": 0.668,
      "step": 6015
    },
    {
      "epoch": 0.24932653653280284,
      "grad_norm": 0.41759005188941956,
      "learning_rate": 3.7535745368643547e-06,
      "loss": 0.6996,
      "step": 6016
    },
    {
      "epoch": 0.24936798043847652,
      "grad_norm": 0.4243796169757843,
      "learning_rate": 3.7533673173359856e-06,
      "loss": 0.7639,
      "step": 6017
    },
    {
      "epoch": 0.2494094243441502,
      "grad_norm": 0.42483577132225037,
      "learning_rate": 3.753160097807618e-06,
      "loss": 0.715,
      "step": 6018
    },
    {
      "epoch": 0.24945086824982388,
      "grad_norm": 0.39412689208984375,
      "learning_rate": 3.7529528782792497e-06,
      "loss": 0.6802,
      "step": 6019
    },
    {
      "epoch": 0.24949231215549753,
      "grad_norm": 0.4239161014556885,
      "learning_rate": 3.752745658750881e-06,
      "loss": 0.7122,
      "step": 6020
    },
    {
      "epoch": 0.2495337560611712,
      "grad_norm": 0.4078959822654724,
      "learning_rate": 3.752538439222513e-06,
      "loss": 0.6809,
      "step": 6021
    },
    {
      "epoch": 0.24957519996684488,
      "grad_norm": 0.42784857749938965,
      "learning_rate": 3.7523312196941443e-06,
      "loss": 0.7351,
      "step": 6022
    },
    {
      "epoch": 0.24961664387251856,
      "grad_norm": 0.42581865191459656,
      "learning_rate": 3.752124000165776e-06,
      "loss": 0.7458,
      "step": 6023
    },
    {
      "epoch": 0.2496580877781922,
      "grad_norm": 0.4087164103984833,
      "learning_rate": 3.7519167806374074e-06,
      "loss": 0.7249,
      "step": 6024
    },
    {
      "epoch": 0.24969953168386588,
      "grad_norm": 0.42763808369636536,
      "learning_rate": 3.7517095611090393e-06,
      "loss": 0.7197,
      "step": 6025
    },
    {
      "epoch": 0.24974097558953956,
      "grad_norm": 0.43708306550979614,
      "learning_rate": 3.7515023415806706e-06,
      "loss": 0.6758,
      "step": 6026
    },
    {
      "epoch": 0.24978241949521324,
      "grad_norm": 0.3713022470474243,
      "learning_rate": 3.7512951220523025e-06,
      "loss": 0.692,
      "step": 6027
    },
    {
      "epoch": 0.2498238634008869,
      "grad_norm": 0.37605151534080505,
      "learning_rate": 3.7510879025239343e-06,
      "loss": 0.6919,
      "step": 6028
    },
    {
      "epoch": 0.24986530730656056,
      "grad_norm": 0.38880184292793274,
      "learning_rate": 3.7508806829955656e-06,
      "loss": 0.7205,
      "step": 6029
    },
    {
      "epoch": 0.24990675121223424,
      "grad_norm": 0.4560619294643402,
      "learning_rate": 3.7506734634671975e-06,
      "loss": 0.6831,
      "step": 6030
    },
    {
      "epoch": 0.24994819511790792,
      "grad_norm": 0.447371244430542,
      "learning_rate": 3.750466243938829e-06,
      "loss": 0.7466,
      "step": 6031
    },
    {
      "epoch": 0.2499896390235816,
      "grad_norm": 0.4342488944530487,
      "learning_rate": 3.7502590244104607e-06,
      "loss": 0.7263,
      "step": 6032
    },
    {
      "epoch": 0.25003108292925524,
      "grad_norm": 0.4200589656829834,
      "learning_rate": 3.750051804882092e-06,
      "loss": 0.7305,
      "step": 6033
    },
    {
      "epoch": 0.2500725268349289,
      "grad_norm": 0.4387052059173584,
      "learning_rate": 3.7498445853537243e-06,
      "loss": 0.7075,
      "step": 6034
    },
    {
      "epoch": 0.2501139707406026,
      "grad_norm": 0.467589408159256,
      "learning_rate": 3.7496373658253552e-06,
      "loss": 0.7893,
      "step": 6035
    },
    {
      "epoch": 0.2501554146462763,
      "grad_norm": 0.3905572295188904,
      "learning_rate": 3.7494301462969875e-06,
      "loss": 0.6697,
      "step": 6036
    },
    {
      "epoch": 0.25019685855194995,
      "grad_norm": 0.4368402361869812,
      "learning_rate": 3.7492229267686193e-06,
      "loss": 0.7454,
      "step": 6037
    },
    {
      "epoch": 0.25023830245762363,
      "grad_norm": 0.46885940432548523,
      "learning_rate": 3.7490157072402507e-06,
      "loss": 0.7844,
      "step": 6038
    },
    {
      "epoch": 0.25027974636329725,
      "grad_norm": 0.40522655844688416,
      "learning_rate": 3.7488084877118825e-06,
      "loss": 0.7056,
      "step": 6039
    },
    {
      "epoch": 0.25032119026897093,
      "grad_norm": 0.4127856492996216,
      "learning_rate": 3.748601268183514e-06,
      "loss": 0.7169,
      "step": 6040
    },
    {
      "epoch": 0.2503626341746446,
      "grad_norm": 0.4249977767467499,
      "learning_rate": 3.7483940486551457e-06,
      "loss": 0.7151,
      "step": 6041
    },
    {
      "epoch": 0.2504040780803183,
      "grad_norm": 0.44575536251068115,
      "learning_rate": 3.748186829126777e-06,
      "loss": 0.709,
      "step": 6042
    },
    {
      "epoch": 0.25044552198599196,
      "grad_norm": 0.45445626974105835,
      "learning_rate": 3.747979609598409e-06,
      "loss": 0.7119,
      "step": 6043
    },
    {
      "epoch": 0.25048696589166564,
      "grad_norm": 0.3987843096256256,
      "learning_rate": 3.7477723900700402e-06,
      "loss": 0.6816,
      "step": 6044
    },
    {
      "epoch": 0.2505284097973393,
      "grad_norm": 0.45341265201568604,
      "learning_rate": 3.747565170541672e-06,
      "loss": 0.7505,
      "step": 6045
    },
    {
      "epoch": 0.250569853703013,
      "grad_norm": 0.4135962724685669,
      "learning_rate": 3.747357951013304e-06,
      "loss": 0.7034,
      "step": 6046
    },
    {
      "epoch": 0.25061129760868667,
      "grad_norm": 0.4489751160144806,
      "learning_rate": 3.7471507314849352e-06,
      "loss": 0.752,
      "step": 6047
    },
    {
      "epoch": 0.2506527415143603,
      "grad_norm": 0.4258221685886383,
      "learning_rate": 3.746943511956567e-06,
      "loss": 0.729,
      "step": 6048
    },
    {
      "epoch": 0.25069418542003397,
      "grad_norm": 0.44324541091918945,
      "learning_rate": 3.7467362924281984e-06,
      "loss": 0.7683,
      "step": 6049
    },
    {
      "epoch": 0.25073562932570764,
      "grad_norm": 0.41910234093666077,
      "learning_rate": 3.7465290728998307e-06,
      "loss": 0.7483,
      "step": 6050
    },
    {
      "epoch": 0.2507770732313813,
      "grad_norm": 0.4094471335411072,
      "learning_rate": 3.7463218533714616e-06,
      "loss": 0.6517,
      "step": 6051
    },
    {
      "epoch": 0.250818517137055,
      "grad_norm": 0.4462414085865021,
      "learning_rate": 3.746114633843094e-06,
      "loss": 0.7058,
      "step": 6052
    },
    {
      "epoch": 0.2508599610427287,
      "grad_norm": 0.434004545211792,
      "learning_rate": 3.7459074143147257e-06,
      "loss": 0.7698,
      "step": 6053
    },
    {
      "epoch": 0.25090140494840235,
      "grad_norm": 0.3918675184249878,
      "learning_rate": 3.745700194786357e-06,
      "loss": 0.6931,
      "step": 6054
    },
    {
      "epoch": 0.25094284885407603,
      "grad_norm": 0.3794451653957367,
      "learning_rate": 3.745492975257989e-06,
      "loss": 0.7218,
      "step": 6055
    },
    {
      "epoch": 0.2509842927597497,
      "grad_norm": 0.4094288647174835,
      "learning_rate": 3.7452857557296203e-06,
      "loss": 0.7207,
      "step": 6056
    },
    {
      "epoch": 0.2510257366654233,
      "grad_norm": 0.46224015951156616,
      "learning_rate": 3.745078536201252e-06,
      "loss": 0.7512,
      "step": 6057
    },
    {
      "epoch": 0.251067180571097,
      "grad_norm": 0.4420699179172516,
      "learning_rate": 3.7448713166728835e-06,
      "loss": 0.7397,
      "step": 6058
    },
    {
      "epoch": 0.2511086244767707,
      "grad_norm": 0.39591434597969055,
      "learning_rate": 3.7446640971445153e-06,
      "loss": 0.6831,
      "step": 6059
    },
    {
      "epoch": 0.25115006838244436,
      "grad_norm": 0.43759599328041077,
      "learning_rate": 3.7444568776161466e-06,
      "loss": 0.7205,
      "step": 6060
    },
    {
      "epoch": 0.25119151228811804,
      "grad_norm": 0.43256527185440063,
      "learning_rate": 3.7442496580877785e-06,
      "loss": 0.7148,
      "step": 6061
    },
    {
      "epoch": 0.2512329561937917,
      "grad_norm": 0.41243574023246765,
      "learning_rate": 3.7440424385594103e-06,
      "loss": 0.7302,
      "step": 6062
    },
    {
      "epoch": 0.2512744000994654,
      "grad_norm": 0.44729259610176086,
      "learning_rate": 3.7438352190310417e-06,
      "loss": 0.7402,
      "step": 6063
    },
    {
      "epoch": 0.25131584400513907,
      "grad_norm": 0.4826046824455261,
      "learning_rate": 3.7436279995026735e-06,
      "loss": 0.7119,
      "step": 6064
    },
    {
      "epoch": 0.2513572879108127,
      "grad_norm": 0.464213490486145,
      "learning_rate": 3.743420779974305e-06,
      "loss": 0.7991,
      "step": 6065
    },
    {
      "epoch": 0.25139873181648636,
      "grad_norm": 0.4252159595489502,
      "learning_rate": 3.7432135604459367e-06,
      "loss": 0.7197,
      "step": 6066
    },
    {
      "epoch": 0.25144017572216004,
      "grad_norm": 0.4489288926124573,
      "learning_rate": 3.743006340917568e-06,
      "loss": 0.7103,
      "step": 6067
    },
    {
      "epoch": 0.2514816196278337,
      "grad_norm": 0.3889477252960205,
      "learning_rate": 3.7427991213892003e-06,
      "loss": 0.7043,
      "step": 6068
    },
    {
      "epoch": 0.2515230635335074,
      "grad_norm": 0.3897220194339752,
      "learning_rate": 3.7425919018608312e-06,
      "loss": 0.6821,
      "step": 6069
    },
    {
      "epoch": 0.2515645074391811,
      "grad_norm": 0.45167189836502075,
      "learning_rate": 3.7423846823324635e-06,
      "loss": 0.7058,
      "step": 6070
    },
    {
      "epoch": 0.25160595134485475,
      "grad_norm": 0.4118085205554962,
      "learning_rate": 3.7421774628040953e-06,
      "loss": 0.7422,
      "step": 6071
    },
    {
      "epoch": 0.2516473952505284,
      "grad_norm": 0.41563406586647034,
      "learning_rate": 3.7419702432757267e-06,
      "loss": 0.6758,
      "step": 6072
    },
    {
      "epoch": 0.2516888391562021,
      "grad_norm": 0.3947673738002777,
      "learning_rate": 3.7417630237473585e-06,
      "loss": 0.7109,
      "step": 6073
    },
    {
      "epoch": 0.2517302830618757,
      "grad_norm": 0.4155599772930145,
      "learning_rate": 3.74155580421899e-06,
      "loss": 0.7319,
      "step": 6074
    },
    {
      "epoch": 0.2517717269675494,
      "grad_norm": 0.4301125109195709,
      "learning_rate": 3.7413485846906217e-06,
      "loss": 0.7153,
      "step": 6075
    },
    {
      "epoch": 0.2518131708732231,
      "grad_norm": 0.4532957971096039,
      "learning_rate": 3.741141365162253e-06,
      "loss": 0.7383,
      "step": 6076
    },
    {
      "epoch": 0.25185461477889676,
      "grad_norm": 0.41179510951042175,
      "learning_rate": 3.740934145633885e-06,
      "loss": 0.6641,
      "step": 6077
    },
    {
      "epoch": 0.25189605868457043,
      "grad_norm": 0.4385316073894501,
      "learning_rate": 3.7407269261055162e-06,
      "loss": 0.7422,
      "step": 6078
    },
    {
      "epoch": 0.2519375025902441,
      "grad_norm": 0.39749807119369507,
      "learning_rate": 3.740519706577148e-06,
      "loss": 0.7372,
      "step": 6079
    },
    {
      "epoch": 0.2519789464959178,
      "grad_norm": 0.4171895980834961,
      "learning_rate": 3.74031248704878e-06,
      "loss": 0.7681,
      "step": 6080
    },
    {
      "epoch": 0.25202039040159147,
      "grad_norm": 0.46200546622276306,
      "learning_rate": 3.7401052675204113e-06,
      "loss": 0.7395,
      "step": 6081
    },
    {
      "epoch": 0.25206183430726514,
      "grad_norm": 0.4416584074497223,
      "learning_rate": 3.739898047992043e-06,
      "loss": 0.7295,
      "step": 6082
    },
    {
      "epoch": 0.25210327821293876,
      "grad_norm": 0.4310278594493866,
      "learning_rate": 3.7396908284636744e-06,
      "loss": 0.7485,
      "step": 6083
    },
    {
      "epoch": 0.25214472211861244,
      "grad_norm": 0.4204119145870209,
      "learning_rate": 3.7394836089353067e-06,
      "loss": 0.7156,
      "step": 6084
    },
    {
      "epoch": 0.2521861660242861,
      "grad_norm": 0.4342232048511505,
      "learning_rate": 3.7392763894069376e-06,
      "loss": 0.7324,
      "step": 6085
    },
    {
      "epoch": 0.2522276099299598,
      "grad_norm": 0.395269513130188,
      "learning_rate": 3.73906916987857e-06,
      "loss": 0.7253,
      "step": 6086
    },
    {
      "epoch": 0.25226905383563347,
      "grad_norm": 0.39830729365348816,
      "learning_rate": 3.7388619503502013e-06,
      "loss": 0.6663,
      "step": 6087
    },
    {
      "epoch": 0.25231049774130715,
      "grad_norm": 0.432003915309906,
      "learning_rate": 3.738654730821833e-06,
      "loss": 0.7439,
      "step": 6088
    },
    {
      "epoch": 0.2523519416469808,
      "grad_norm": 0.40886783599853516,
      "learning_rate": 3.738447511293465e-06,
      "loss": 0.6995,
      "step": 6089
    },
    {
      "epoch": 0.2523933855526545,
      "grad_norm": 0.44154804944992065,
      "learning_rate": 3.7382402917650963e-06,
      "loss": 0.7522,
      "step": 6090
    },
    {
      "epoch": 0.2524348294583282,
      "grad_norm": 0.4480719566345215,
      "learning_rate": 3.738033072236728e-06,
      "loss": 0.7439,
      "step": 6091
    },
    {
      "epoch": 0.2524762733640018,
      "grad_norm": 0.4115225672721863,
      "learning_rate": 3.7378258527083595e-06,
      "loss": 0.6628,
      "step": 6092
    },
    {
      "epoch": 0.2525177172696755,
      "grad_norm": 0.3956930339336395,
      "learning_rate": 3.7376186331799913e-06,
      "loss": 0.6376,
      "step": 6093
    },
    {
      "epoch": 0.25255916117534916,
      "grad_norm": 0.43507394194602966,
      "learning_rate": 3.7374114136516227e-06,
      "loss": 0.6757,
      "step": 6094
    },
    {
      "epoch": 0.25260060508102283,
      "grad_norm": 0.5457044243812561,
      "learning_rate": 3.7372041941232545e-06,
      "loss": 0.8149,
      "step": 6095
    },
    {
      "epoch": 0.2526420489866965,
      "grad_norm": 0.397545725107193,
      "learning_rate": 3.736996974594886e-06,
      "loss": 0.7065,
      "step": 6096
    },
    {
      "epoch": 0.2526834928923702,
      "grad_norm": 0.4298243522644043,
      "learning_rate": 3.7367897550665177e-06,
      "loss": 0.7568,
      "step": 6097
    },
    {
      "epoch": 0.25272493679804386,
      "grad_norm": 0.39709949493408203,
      "learning_rate": 3.7365825355381495e-06,
      "loss": 0.7032,
      "step": 6098
    },
    {
      "epoch": 0.25276638070371754,
      "grad_norm": 0.4216078221797943,
      "learning_rate": 3.736375316009781e-06,
      "loss": 0.7374,
      "step": 6099
    },
    {
      "epoch": 0.25280782460939116,
      "grad_norm": 0.41981711983680725,
      "learning_rate": 3.736168096481413e-06,
      "loss": 0.6865,
      "step": 6100
    },
    {
      "epoch": 0.25284926851506484,
      "grad_norm": 0.4466809928417206,
      "learning_rate": 3.735960876953044e-06,
      "loss": 0.7103,
      "step": 6101
    },
    {
      "epoch": 0.2528907124207385,
      "grad_norm": 0.43563538789749146,
      "learning_rate": 3.7357536574246763e-06,
      "loss": 0.6917,
      "step": 6102
    },
    {
      "epoch": 0.2529321563264122,
      "grad_norm": 0.39142948389053345,
      "learning_rate": 3.7355464378963072e-06,
      "loss": 0.6704,
      "step": 6103
    },
    {
      "epoch": 0.25297360023208587,
      "grad_norm": 0.41214078664779663,
      "learning_rate": 3.7353392183679395e-06,
      "loss": 0.6733,
      "step": 6104
    },
    {
      "epoch": 0.25301504413775955,
      "grad_norm": 0.4458875358104706,
      "learning_rate": 3.735131998839571e-06,
      "loss": 0.7428,
      "step": 6105
    },
    {
      "epoch": 0.2530564880434332,
      "grad_norm": 0.41436514258384705,
      "learning_rate": 3.7349247793112027e-06,
      "loss": 0.734,
      "step": 6106
    },
    {
      "epoch": 0.2530979319491069,
      "grad_norm": 0.4259885549545288,
      "learning_rate": 3.7347175597828345e-06,
      "loss": 0.7358,
      "step": 6107
    },
    {
      "epoch": 0.2531393758547806,
      "grad_norm": 0.46017834544181824,
      "learning_rate": 3.734510340254466e-06,
      "loss": 0.7795,
      "step": 6108
    },
    {
      "epoch": 0.2531808197604542,
      "grad_norm": 0.42012926936149597,
      "learning_rate": 3.7343031207260977e-06,
      "loss": 0.7268,
      "step": 6109
    },
    {
      "epoch": 0.2532222636661279,
      "grad_norm": 0.4547656774520874,
      "learning_rate": 3.734095901197729e-06,
      "loss": 0.7327,
      "step": 6110
    },
    {
      "epoch": 0.25326370757180156,
      "grad_norm": 0.4279201924800873,
      "learning_rate": 3.733888681669361e-06,
      "loss": 0.7534,
      "step": 6111
    },
    {
      "epoch": 0.25330515147747523,
      "grad_norm": 0.41167742013931274,
      "learning_rate": 3.7336814621409923e-06,
      "loss": 0.7205,
      "step": 6112
    },
    {
      "epoch": 0.2533465953831489,
      "grad_norm": 0.4240303635597229,
      "learning_rate": 3.733474242612624e-06,
      "loss": 0.6852,
      "step": 6113
    },
    {
      "epoch": 0.2533880392888226,
      "grad_norm": 0.4501185119152069,
      "learning_rate": 3.733267023084256e-06,
      "loss": 0.7407,
      "step": 6114
    },
    {
      "epoch": 0.25342948319449626,
      "grad_norm": 0.42156773805618286,
      "learning_rate": 3.7330598035558873e-06,
      "loss": 0.7356,
      "step": 6115
    },
    {
      "epoch": 0.25347092710016994,
      "grad_norm": 0.3955972194671631,
      "learning_rate": 3.732852584027519e-06,
      "loss": 0.676,
      "step": 6116
    },
    {
      "epoch": 0.2535123710058436,
      "grad_norm": 0.428305059671402,
      "learning_rate": 3.7326453644991505e-06,
      "loss": 0.707,
      "step": 6117
    },
    {
      "epoch": 0.25355381491151724,
      "grad_norm": 0.41422170400619507,
      "learning_rate": 3.7324381449707827e-06,
      "loss": 0.7705,
      "step": 6118
    },
    {
      "epoch": 0.2535952588171909,
      "grad_norm": 0.41554686427116394,
      "learning_rate": 3.7322309254424136e-06,
      "loss": 0.7056,
      "step": 6119
    },
    {
      "epoch": 0.2536367027228646,
      "grad_norm": 0.39503031969070435,
      "learning_rate": 3.732023705914046e-06,
      "loss": 0.6763,
      "step": 6120
    },
    {
      "epoch": 0.25367814662853827,
      "grad_norm": 0.394479364156723,
      "learning_rate": 3.7318164863856773e-06,
      "loss": 0.6619,
      "step": 6121
    },
    {
      "epoch": 0.25371959053421195,
      "grad_norm": 0.3825823664665222,
      "learning_rate": 3.731609266857309e-06,
      "loss": 0.6973,
      "step": 6122
    },
    {
      "epoch": 0.2537610344398856,
      "grad_norm": 0.43291887640953064,
      "learning_rate": 3.731402047328941e-06,
      "loss": 0.7185,
      "step": 6123
    },
    {
      "epoch": 0.2538024783455593,
      "grad_norm": 0.407031387090683,
      "learning_rate": 3.7311948278005723e-06,
      "loss": 0.6604,
      "step": 6124
    },
    {
      "epoch": 0.253843922251233,
      "grad_norm": 0.45129281282424927,
      "learning_rate": 3.730987608272204e-06,
      "loss": 0.7351,
      "step": 6125
    },
    {
      "epoch": 0.2538853661569066,
      "grad_norm": 0.43731164932250977,
      "learning_rate": 3.7307803887438355e-06,
      "loss": 0.6721,
      "step": 6126
    },
    {
      "epoch": 0.2539268100625803,
      "grad_norm": 0.4282631278038025,
      "learning_rate": 3.7305731692154673e-06,
      "loss": 0.7319,
      "step": 6127
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 0.44185078144073486,
      "learning_rate": 3.7303659496870987e-06,
      "loss": 0.7065,
      "step": 6128
    },
    {
      "epoch": 0.25400969787392763,
      "grad_norm": 0.43341001868247986,
      "learning_rate": 3.7301587301587305e-06,
      "loss": 0.738,
      "step": 6129
    },
    {
      "epoch": 0.2540511417796013,
      "grad_norm": 0.38501355051994324,
      "learning_rate": 3.729951510630362e-06,
      "loss": 0.6528,
      "step": 6130
    },
    {
      "epoch": 0.254092585685275,
      "grad_norm": 0.407642126083374,
      "learning_rate": 3.7297442911019937e-06,
      "loss": 0.74,
      "step": 6131
    },
    {
      "epoch": 0.25413402959094866,
      "grad_norm": 0.4139276444911957,
      "learning_rate": 3.7295370715736255e-06,
      "loss": 0.6926,
      "step": 6132
    },
    {
      "epoch": 0.25417547349662234,
      "grad_norm": 0.43417662382125854,
      "learning_rate": 3.729329852045257e-06,
      "loss": 0.7432,
      "step": 6133
    },
    {
      "epoch": 0.254216917402296,
      "grad_norm": 0.4174732267856598,
      "learning_rate": 3.729122632516889e-06,
      "loss": 0.7297,
      "step": 6134
    },
    {
      "epoch": 0.25425836130796964,
      "grad_norm": 0.41027596592903137,
      "learning_rate": 3.72891541298852e-06,
      "loss": 0.7185,
      "step": 6135
    },
    {
      "epoch": 0.2542998052136433,
      "grad_norm": 0.43580448627471924,
      "learning_rate": 3.7287081934601523e-06,
      "loss": 0.752,
      "step": 6136
    },
    {
      "epoch": 0.254341249119317,
      "grad_norm": 0.4059290587902069,
      "learning_rate": 3.7285009739317832e-06,
      "loss": 0.7268,
      "step": 6137
    },
    {
      "epoch": 0.25438269302499067,
      "grad_norm": 0.39432117342948914,
      "learning_rate": 3.7282937544034155e-06,
      "loss": 0.7283,
      "step": 6138
    },
    {
      "epoch": 0.25442413693066435,
      "grad_norm": 0.4232621490955353,
      "learning_rate": 3.728086534875047e-06,
      "loss": 0.7236,
      "step": 6139
    },
    {
      "epoch": 0.254465580836338,
      "grad_norm": 0.4162512421607971,
      "learning_rate": 3.7278793153466787e-06,
      "loss": 0.7546,
      "step": 6140
    },
    {
      "epoch": 0.2545070247420117,
      "grad_norm": 0.4374372959136963,
      "learning_rate": 3.7276720958183105e-06,
      "loss": 0.7205,
      "step": 6141
    },
    {
      "epoch": 0.2545484686476854,
      "grad_norm": 0.4262143075466156,
      "learning_rate": 3.727464876289942e-06,
      "loss": 0.7872,
      "step": 6142
    },
    {
      "epoch": 0.25458991255335905,
      "grad_norm": 0.4533160924911499,
      "learning_rate": 3.7272576567615737e-06,
      "loss": 0.7571,
      "step": 6143
    },
    {
      "epoch": 0.2546313564590327,
      "grad_norm": 0.39063358306884766,
      "learning_rate": 3.727050437233205e-06,
      "loss": 0.728,
      "step": 6144
    },
    {
      "epoch": 0.25467280036470635,
      "grad_norm": 0.39919576048851013,
      "learning_rate": 3.726843217704837e-06,
      "loss": 0.7412,
      "step": 6145
    },
    {
      "epoch": 0.25471424427038003,
      "grad_norm": 0.44663751125335693,
      "learning_rate": 3.7266359981764683e-06,
      "loss": 0.7273,
      "step": 6146
    },
    {
      "epoch": 0.2547556881760537,
      "grad_norm": 0.4171029329299927,
      "learning_rate": 3.7264287786481e-06,
      "loss": 0.7157,
      "step": 6147
    },
    {
      "epoch": 0.2547971320817274,
      "grad_norm": 0.40051400661468506,
      "learning_rate": 3.7262215591197315e-06,
      "loss": 0.6776,
      "step": 6148
    },
    {
      "epoch": 0.25483857598740106,
      "grad_norm": 0.42343100905418396,
      "learning_rate": 3.7260143395913633e-06,
      "loss": 0.7246,
      "step": 6149
    },
    {
      "epoch": 0.25488001989307474,
      "grad_norm": 0.42666757106781006,
      "learning_rate": 3.725807120062995e-06,
      "loss": 0.7214,
      "step": 6150
    },
    {
      "epoch": 0.2549214637987484,
      "grad_norm": 0.3996857702732086,
      "learning_rate": 3.7255999005346265e-06,
      "loss": 0.6819,
      "step": 6151
    },
    {
      "epoch": 0.25496290770442204,
      "grad_norm": 0.4599875807762146,
      "learning_rate": 3.7253926810062587e-06,
      "loss": 0.6921,
      "step": 6152
    },
    {
      "epoch": 0.2550043516100957,
      "grad_norm": 0.39232611656188965,
      "learning_rate": 3.7251854614778897e-06,
      "loss": 0.7358,
      "step": 6153
    },
    {
      "epoch": 0.2550457955157694,
      "grad_norm": 0.3988505005836487,
      "learning_rate": 3.724978241949522e-06,
      "loss": 0.7046,
      "step": 6154
    },
    {
      "epoch": 0.25508723942144307,
      "grad_norm": 0.41904374957084656,
      "learning_rate": 3.7247710224211533e-06,
      "loss": 0.7188,
      "step": 6155
    },
    {
      "epoch": 0.25512868332711675,
      "grad_norm": 0.40041491389274597,
      "learning_rate": 3.724563802892785e-06,
      "loss": 0.6318,
      "step": 6156
    },
    {
      "epoch": 0.2551701272327904,
      "grad_norm": 0.4093829393386841,
      "learning_rate": 3.7243565833644165e-06,
      "loss": 0.6777,
      "step": 6157
    },
    {
      "epoch": 0.2552115711384641,
      "grad_norm": 0.400246262550354,
      "learning_rate": 3.7241493638360483e-06,
      "loss": 0.7234,
      "step": 6158
    },
    {
      "epoch": 0.2552530150441378,
      "grad_norm": 0.38729923963546753,
      "learning_rate": 3.72394214430768e-06,
      "loss": 0.7227,
      "step": 6159
    },
    {
      "epoch": 0.25529445894981145,
      "grad_norm": 0.41513344645500183,
      "learning_rate": 3.7237349247793115e-06,
      "loss": 0.7075,
      "step": 6160
    },
    {
      "epoch": 0.2553359028554851,
      "grad_norm": 0.41064581274986267,
      "learning_rate": 3.7235277052509433e-06,
      "loss": 0.6682,
      "step": 6161
    },
    {
      "epoch": 0.25537734676115875,
      "grad_norm": 0.42259982228279114,
      "learning_rate": 3.7233204857225747e-06,
      "loss": 0.7343,
      "step": 6162
    },
    {
      "epoch": 0.25541879066683243,
      "grad_norm": 0.44401171803474426,
      "learning_rate": 3.7231132661942065e-06,
      "loss": 0.7236,
      "step": 6163
    },
    {
      "epoch": 0.2554602345725061,
      "grad_norm": 0.4470435380935669,
      "learning_rate": 3.722906046665838e-06,
      "loss": 0.7195,
      "step": 6164
    },
    {
      "epoch": 0.2555016784781798,
      "grad_norm": 0.4186209738254547,
      "learning_rate": 3.7226988271374697e-06,
      "loss": 0.7407,
      "step": 6165
    },
    {
      "epoch": 0.25554312238385346,
      "grad_norm": 0.364193856716156,
      "learning_rate": 3.722491607609101e-06,
      "loss": 0.6737,
      "step": 6166
    },
    {
      "epoch": 0.25558456628952714,
      "grad_norm": 0.4577482044696808,
      "learning_rate": 3.722284388080733e-06,
      "loss": 0.7218,
      "step": 6167
    },
    {
      "epoch": 0.2556260101952008,
      "grad_norm": 0.4418712258338928,
      "learning_rate": 3.722077168552365e-06,
      "loss": 0.6958,
      "step": 6168
    },
    {
      "epoch": 0.2556674541008745,
      "grad_norm": 0.3981810510158539,
      "learning_rate": 3.721869949023996e-06,
      "loss": 0.7332,
      "step": 6169
    },
    {
      "epoch": 0.2557088980065481,
      "grad_norm": 0.43710967898368835,
      "learning_rate": 3.7216627294956283e-06,
      "loss": 0.7478,
      "step": 6170
    },
    {
      "epoch": 0.2557503419122218,
      "grad_norm": 0.42081883549690247,
      "learning_rate": 3.7214555099672593e-06,
      "loss": 0.6965,
      "step": 6171
    },
    {
      "epoch": 0.25579178581789547,
      "grad_norm": 0.39096295833587646,
      "learning_rate": 3.7212482904388915e-06,
      "loss": 0.6886,
      "step": 6172
    },
    {
      "epoch": 0.25583322972356914,
      "grad_norm": 0.3917675018310547,
      "learning_rate": 3.721041070910523e-06,
      "loss": 0.627,
      "step": 6173
    },
    {
      "epoch": 0.2558746736292428,
      "grad_norm": 0.3938373327255249,
      "learning_rate": 3.7208338513821547e-06,
      "loss": 0.6943,
      "step": 6174
    },
    {
      "epoch": 0.2559161175349165,
      "grad_norm": 0.4196326434612274,
      "learning_rate": 3.7206266318537865e-06,
      "loss": 0.7389,
      "step": 6175
    },
    {
      "epoch": 0.2559575614405902,
      "grad_norm": 0.43998026847839355,
      "learning_rate": 3.720419412325418e-06,
      "loss": 0.7117,
      "step": 6176
    },
    {
      "epoch": 0.25599900534626385,
      "grad_norm": 0.43819454312324524,
      "learning_rate": 3.7202121927970497e-06,
      "loss": 0.7494,
      "step": 6177
    },
    {
      "epoch": 0.25604044925193753,
      "grad_norm": 0.45608949661254883,
      "learning_rate": 3.720004973268681e-06,
      "loss": 0.7378,
      "step": 6178
    },
    {
      "epoch": 0.25608189315761115,
      "grad_norm": 0.41671475768089294,
      "learning_rate": 3.719797753740313e-06,
      "loss": 0.7065,
      "step": 6179
    },
    {
      "epoch": 0.25612333706328483,
      "grad_norm": 0.4319727420806885,
      "learning_rate": 3.7195905342119443e-06,
      "loss": 0.717,
      "step": 6180
    },
    {
      "epoch": 0.2561647809689585,
      "grad_norm": 0.4710858464241028,
      "learning_rate": 3.719383314683576e-06,
      "loss": 0.7307,
      "step": 6181
    },
    {
      "epoch": 0.2562062248746322,
      "grad_norm": 0.40955880284309387,
      "learning_rate": 3.7191760951552075e-06,
      "loss": 0.6978,
      "step": 6182
    },
    {
      "epoch": 0.25624766878030586,
      "grad_norm": 0.41667839884757996,
      "learning_rate": 3.7189688756268393e-06,
      "loss": 0.6904,
      "step": 6183
    },
    {
      "epoch": 0.25628911268597954,
      "grad_norm": 0.42708152532577515,
      "learning_rate": 3.718761656098471e-06,
      "loss": 0.6689,
      "step": 6184
    },
    {
      "epoch": 0.2563305565916532,
      "grad_norm": 0.42344146966934204,
      "learning_rate": 3.7185544365701025e-06,
      "loss": 0.7117,
      "step": 6185
    },
    {
      "epoch": 0.2563720004973269,
      "grad_norm": 0.4450227618217468,
      "learning_rate": 3.7183472170417347e-06,
      "loss": 0.7114,
      "step": 6186
    },
    {
      "epoch": 0.2564134444030005,
      "grad_norm": 0.4261973798274994,
      "learning_rate": 3.7181399975133657e-06,
      "loss": 0.7317,
      "step": 6187
    },
    {
      "epoch": 0.2564548883086742,
      "grad_norm": 0.40273383259773254,
      "learning_rate": 3.717932777984998e-06,
      "loss": 0.6539,
      "step": 6188
    },
    {
      "epoch": 0.25649633221434787,
      "grad_norm": 0.40947920083999634,
      "learning_rate": 3.7177255584566293e-06,
      "loss": 0.6924,
      "step": 6189
    },
    {
      "epoch": 0.25653777612002154,
      "grad_norm": 0.4407038986682892,
      "learning_rate": 3.717518338928261e-06,
      "loss": 0.7716,
      "step": 6190
    },
    {
      "epoch": 0.2565792200256952,
      "grad_norm": 0.4632432162761688,
      "learning_rate": 3.7173111193998925e-06,
      "loss": 0.6904,
      "step": 6191
    },
    {
      "epoch": 0.2566206639313689,
      "grad_norm": 0.43985557556152344,
      "learning_rate": 3.7171038998715243e-06,
      "loss": 0.7166,
      "step": 6192
    },
    {
      "epoch": 0.2566621078370426,
      "grad_norm": 0.4427506923675537,
      "learning_rate": 3.716896680343156e-06,
      "loss": 0.6951,
      "step": 6193
    },
    {
      "epoch": 0.25670355174271625,
      "grad_norm": 0.4563213884830475,
      "learning_rate": 3.7166894608147875e-06,
      "loss": 0.7432,
      "step": 6194
    },
    {
      "epoch": 0.25674499564838993,
      "grad_norm": 0.3936905860900879,
      "learning_rate": 3.7164822412864193e-06,
      "loss": 0.6648,
      "step": 6195
    },
    {
      "epoch": 0.25678643955406355,
      "grad_norm": 0.43082064390182495,
      "learning_rate": 3.7162750217580507e-06,
      "loss": 0.7024,
      "step": 6196
    },
    {
      "epoch": 0.2568278834597372,
      "grad_norm": 0.3919983208179474,
      "learning_rate": 3.7160678022296825e-06,
      "loss": 0.6605,
      "step": 6197
    },
    {
      "epoch": 0.2568693273654109,
      "grad_norm": 0.3916562497615814,
      "learning_rate": 3.715860582701314e-06,
      "loss": 0.667,
      "step": 6198
    },
    {
      "epoch": 0.2569107712710846,
      "grad_norm": 0.3910791277885437,
      "learning_rate": 3.7156533631729457e-06,
      "loss": 0.6899,
      "step": 6199
    },
    {
      "epoch": 0.25695221517675826,
      "grad_norm": 0.4630964398384094,
      "learning_rate": 3.715446143644577e-06,
      "loss": 0.7576,
      "step": 6200
    },
    {
      "epoch": 0.25699365908243194,
      "grad_norm": 0.4109917879104614,
      "learning_rate": 3.715238924116209e-06,
      "loss": 0.6863,
      "step": 6201
    },
    {
      "epoch": 0.2570351029881056,
      "grad_norm": 0.5220915675163269,
      "learning_rate": 3.715031704587841e-06,
      "loss": 0.7163,
      "step": 6202
    },
    {
      "epoch": 0.2570765468937793,
      "grad_norm": 0.4042677879333496,
      "learning_rate": 3.714824485059472e-06,
      "loss": 0.733,
      "step": 6203
    },
    {
      "epoch": 0.25711799079945297,
      "grad_norm": 0.39604419469833374,
      "learning_rate": 3.7146172655311043e-06,
      "loss": 0.7205,
      "step": 6204
    },
    {
      "epoch": 0.2571594347051266,
      "grad_norm": 0.38982924818992615,
      "learning_rate": 3.7144100460027353e-06,
      "loss": 0.7111,
      "step": 6205
    },
    {
      "epoch": 0.25720087861080027,
      "grad_norm": 0.41776251792907715,
      "learning_rate": 3.7142028264743675e-06,
      "loss": 0.6851,
      "step": 6206
    },
    {
      "epoch": 0.25724232251647394,
      "grad_norm": 0.3976080119609833,
      "learning_rate": 3.713995606945999e-06,
      "loss": 0.7209,
      "step": 6207
    },
    {
      "epoch": 0.2572837664221476,
      "grad_norm": 0.45019590854644775,
      "learning_rate": 3.7137883874176307e-06,
      "loss": 0.7611,
      "step": 6208
    },
    {
      "epoch": 0.2573252103278213,
      "grad_norm": 0.42690035700798035,
      "learning_rate": 3.713581167889262e-06,
      "loss": 0.7322,
      "step": 6209
    },
    {
      "epoch": 0.257366654233495,
      "grad_norm": 0.43929794430732727,
      "learning_rate": 3.713373948360894e-06,
      "loss": 0.6943,
      "step": 6210
    },
    {
      "epoch": 0.25740809813916865,
      "grad_norm": 0.44643911719322205,
      "learning_rate": 3.7131667288325257e-06,
      "loss": 0.7761,
      "step": 6211
    },
    {
      "epoch": 0.2574495420448423,
      "grad_norm": 0.45257800817489624,
      "learning_rate": 3.712959509304157e-06,
      "loss": 0.7457,
      "step": 6212
    },
    {
      "epoch": 0.25749098595051595,
      "grad_norm": 0.43421316146850586,
      "learning_rate": 3.712752289775789e-06,
      "loss": 0.6964,
      "step": 6213
    },
    {
      "epoch": 0.2575324298561896,
      "grad_norm": 0.41608262062072754,
      "learning_rate": 3.7125450702474203e-06,
      "loss": 0.6831,
      "step": 6214
    },
    {
      "epoch": 0.2575738737618633,
      "grad_norm": 0.4203833341598511,
      "learning_rate": 3.712337850719052e-06,
      "loss": 0.6833,
      "step": 6215
    },
    {
      "epoch": 0.257615317667537,
      "grad_norm": 0.41846519708633423,
      "learning_rate": 3.7121306311906835e-06,
      "loss": 0.6899,
      "step": 6216
    },
    {
      "epoch": 0.25765676157321066,
      "grad_norm": 0.4179471731185913,
      "learning_rate": 3.7119234116623153e-06,
      "loss": 0.7336,
      "step": 6217
    },
    {
      "epoch": 0.25769820547888433,
      "grad_norm": 0.43322238326072693,
      "learning_rate": 3.7117161921339467e-06,
      "loss": 0.7617,
      "step": 6218
    },
    {
      "epoch": 0.257739649384558,
      "grad_norm": 0.41138553619384766,
      "learning_rate": 3.7115089726055785e-06,
      "loss": 0.6938,
      "step": 6219
    },
    {
      "epoch": 0.2577810932902317,
      "grad_norm": 0.4735654890537262,
      "learning_rate": 3.7113017530772107e-06,
      "loss": 0.7441,
      "step": 6220
    },
    {
      "epoch": 0.25782253719590537,
      "grad_norm": 0.4648279547691345,
      "learning_rate": 3.7110945335488417e-06,
      "loss": 0.7776,
      "step": 6221
    },
    {
      "epoch": 0.257863981101579,
      "grad_norm": 0.40383368730545044,
      "learning_rate": 3.710887314020474e-06,
      "loss": 0.6841,
      "step": 6222
    },
    {
      "epoch": 0.25790542500725266,
      "grad_norm": 0.3935619592666626,
      "learning_rate": 3.7106800944921053e-06,
      "loss": 0.7061,
      "step": 6223
    },
    {
      "epoch": 0.25794686891292634,
      "grad_norm": 0.41908755898475647,
      "learning_rate": 3.710472874963737e-06,
      "loss": 0.6965,
      "step": 6224
    },
    {
      "epoch": 0.2579883128186,
      "grad_norm": 0.4447205662727356,
      "learning_rate": 3.7102656554353685e-06,
      "loss": 0.7329,
      "step": 6225
    },
    {
      "epoch": 0.2580297567242737,
      "grad_norm": 0.4485672116279602,
      "learning_rate": 3.7100584359070003e-06,
      "loss": 0.748,
      "step": 6226
    },
    {
      "epoch": 0.2580712006299474,
      "grad_norm": 0.4435478448867798,
      "learning_rate": 3.709851216378632e-06,
      "loss": 0.6873,
      "step": 6227
    },
    {
      "epoch": 0.25811264453562105,
      "grad_norm": 0.4243341088294983,
      "learning_rate": 3.7096439968502635e-06,
      "loss": 0.697,
      "step": 6228
    },
    {
      "epoch": 0.2581540884412947,
      "grad_norm": 0.41485750675201416,
      "learning_rate": 3.7094367773218953e-06,
      "loss": 0.7192,
      "step": 6229
    },
    {
      "epoch": 0.2581955323469684,
      "grad_norm": 0.4284738302230835,
      "learning_rate": 3.7092295577935267e-06,
      "loss": 0.7148,
      "step": 6230
    },
    {
      "epoch": 0.258236976252642,
      "grad_norm": 0.43482398986816406,
      "learning_rate": 3.7090223382651585e-06,
      "loss": 0.6978,
      "step": 6231
    },
    {
      "epoch": 0.2582784201583157,
      "grad_norm": 0.40069153904914856,
      "learning_rate": 3.70881511873679e-06,
      "loss": 0.6829,
      "step": 6232
    },
    {
      "epoch": 0.2583198640639894,
      "grad_norm": 0.40957170724868774,
      "learning_rate": 3.7086078992084217e-06,
      "loss": 0.7054,
      "step": 6233
    },
    {
      "epoch": 0.25836130796966306,
      "grad_norm": 0.42304345965385437,
      "learning_rate": 3.708400679680053e-06,
      "loss": 0.7288,
      "step": 6234
    },
    {
      "epoch": 0.25840275187533673,
      "grad_norm": 0.4186275005340576,
      "learning_rate": 3.708193460151685e-06,
      "loss": 0.687,
      "step": 6235
    },
    {
      "epoch": 0.2584441957810104,
      "grad_norm": 0.47155824303627014,
      "learning_rate": 3.707986240623317e-06,
      "loss": 0.782,
      "step": 6236
    },
    {
      "epoch": 0.2584856396866841,
      "grad_norm": 0.45221614837646484,
      "learning_rate": 3.707779021094948e-06,
      "loss": 0.688,
      "step": 6237
    },
    {
      "epoch": 0.25852708359235776,
      "grad_norm": 0.4429072141647339,
      "learning_rate": 3.7075718015665803e-06,
      "loss": 0.7297,
      "step": 6238
    },
    {
      "epoch": 0.25856852749803144,
      "grad_norm": 0.39871254563331604,
      "learning_rate": 3.7073645820382113e-06,
      "loss": 0.7063,
      "step": 6239
    },
    {
      "epoch": 0.25860997140370506,
      "grad_norm": 0.40279820561408997,
      "learning_rate": 3.7071573625098435e-06,
      "loss": 0.704,
      "step": 6240
    },
    {
      "epoch": 0.25865141530937874,
      "grad_norm": 0.40577587485313416,
      "learning_rate": 3.706950142981475e-06,
      "loss": 0.7048,
      "step": 6241
    },
    {
      "epoch": 0.2586928592150524,
      "grad_norm": 0.46221402287483215,
      "learning_rate": 3.7067429234531067e-06,
      "loss": 0.7419,
      "step": 6242
    },
    {
      "epoch": 0.2587343031207261,
      "grad_norm": 0.38397467136383057,
      "learning_rate": 3.706535703924738e-06,
      "loss": 0.7422,
      "step": 6243
    },
    {
      "epoch": 0.25877574702639977,
      "grad_norm": 0.4703851044178009,
      "learning_rate": 3.70632848439637e-06,
      "loss": 0.7024,
      "step": 6244
    },
    {
      "epoch": 0.25881719093207345,
      "grad_norm": 0.4419141113758087,
      "learning_rate": 3.7061212648680017e-06,
      "loss": 0.741,
      "step": 6245
    },
    {
      "epoch": 0.2588586348377471,
      "grad_norm": 0.4188176691532135,
      "learning_rate": 3.705914045339633e-06,
      "loss": 0.7383,
      "step": 6246
    },
    {
      "epoch": 0.2589000787434208,
      "grad_norm": 0.4133175015449524,
      "learning_rate": 3.705706825811265e-06,
      "loss": 0.6956,
      "step": 6247
    },
    {
      "epoch": 0.2589415226490944,
      "grad_norm": 0.43742525577545166,
      "learning_rate": 3.7054996062828963e-06,
      "loss": 0.7661,
      "step": 6248
    },
    {
      "epoch": 0.2589829665547681,
      "grad_norm": 0.4420631527900696,
      "learning_rate": 3.705292386754528e-06,
      "loss": 0.7517,
      "step": 6249
    },
    {
      "epoch": 0.2590244104604418,
      "grad_norm": 0.41893723607063293,
      "learning_rate": 3.7050851672261595e-06,
      "loss": 0.7678,
      "step": 6250
    },
    {
      "epoch": 0.25906585436611546,
      "grad_norm": 0.410117506980896,
      "learning_rate": 3.7048779476977913e-06,
      "loss": 0.6654,
      "step": 6251
    },
    {
      "epoch": 0.25910729827178913,
      "grad_norm": 0.41841989755630493,
      "learning_rate": 3.7046707281694227e-06,
      "loss": 0.6917,
      "step": 6252
    },
    {
      "epoch": 0.2591487421774628,
      "grad_norm": 0.40545976161956787,
      "learning_rate": 3.7044635086410545e-06,
      "loss": 0.7024,
      "step": 6253
    },
    {
      "epoch": 0.2591901860831365,
      "grad_norm": 0.4304497539997101,
      "learning_rate": 3.7042562891126867e-06,
      "loss": 0.7253,
      "step": 6254
    },
    {
      "epoch": 0.25923162998881016,
      "grad_norm": 0.41317278146743774,
      "learning_rate": 3.7040490695843177e-06,
      "loss": 0.676,
      "step": 6255
    },
    {
      "epoch": 0.25927307389448384,
      "grad_norm": 0.39634430408477783,
      "learning_rate": 3.70384185005595e-06,
      "loss": 0.6338,
      "step": 6256
    },
    {
      "epoch": 0.25931451780015746,
      "grad_norm": 0.4154151678085327,
      "learning_rate": 3.7036346305275813e-06,
      "loss": 0.6873,
      "step": 6257
    },
    {
      "epoch": 0.25935596170583114,
      "grad_norm": 0.40515556931495667,
      "learning_rate": 3.703427410999213e-06,
      "loss": 0.7227,
      "step": 6258
    },
    {
      "epoch": 0.2593974056115048,
      "grad_norm": 0.38794848322868347,
      "learning_rate": 3.7032201914708445e-06,
      "loss": 0.7146,
      "step": 6259
    },
    {
      "epoch": 0.2594388495171785,
      "grad_norm": 0.40703046321868896,
      "learning_rate": 3.7030129719424763e-06,
      "loss": 0.7219,
      "step": 6260
    },
    {
      "epoch": 0.25948029342285217,
      "grad_norm": 0.4242015480995178,
      "learning_rate": 3.7028057524141077e-06,
      "loss": 0.7512,
      "step": 6261
    },
    {
      "epoch": 0.25952173732852585,
      "grad_norm": 0.42097732424736023,
      "learning_rate": 3.7025985328857395e-06,
      "loss": 0.7018,
      "step": 6262
    },
    {
      "epoch": 0.2595631812341995,
      "grad_norm": 0.4927029311656952,
      "learning_rate": 3.7023913133573713e-06,
      "loss": 0.7856,
      "step": 6263
    },
    {
      "epoch": 0.2596046251398732,
      "grad_norm": 0.4092736542224884,
      "learning_rate": 3.7021840938290027e-06,
      "loss": 0.7104,
      "step": 6264
    },
    {
      "epoch": 0.2596460690455469,
      "grad_norm": 0.42849522829055786,
      "learning_rate": 3.7019768743006345e-06,
      "loss": 0.7214,
      "step": 6265
    },
    {
      "epoch": 0.2596875129512205,
      "grad_norm": 0.4125194847583771,
      "learning_rate": 3.701769654772266e-06,
      "loss": 0.7129,
      "step": 6266
    },
    {
      "epoch": 0.2597289568568942,
      "grad_norm": 0.42058929800987244,
      "learning_rate": 3.7015624352438977e-06,
      "loss": 0.7307,
      "step": 6267
    },
    {
      "epoch": 0.25977040076256785,
      "grad_norm": 0.4356994330883026,
      "learning_rate": 3.701355215715529e-06,
      "loss": 0.7751,
      "step": 6268
    },
    {
      "epoch": 0.25981184466824153,
      "grad_norm": 0.4404783248901367,
      "learning_rate": 3.701147996187161e-06,
      "loss": 0.7881,
      "step": 6269
    },
    {
      "epoch": 0.2598532885739152,
      "grad_norm": 0.4452175796031952,
      "learning_rate": 3.7009407766587923e-06,
      "loss": 0.7576,
      "step": 6270
    },
    {
      "epoch": 0.2598947324795889,
      "grad_norm": 0.42158225178718567,
      "learning_rate": 3.700733557130424e-06,
      "loss": 0.6813,
      "step": 6271
    },
    {
      "epoch": 0.25993617638526256,
      "grad_norm": 0.4066883325576782,
      "learning_rate": 3.7005263376020563e-06,
      "loss": 0.7653,
      "step": 6272
    },
    {
      "epoch": 0.25997762029093624,
      "grad_norm": 0.46558114886283875,
      "learning_rate": 3.7003191180736877e-06,
      "loss": 0.8022,
      "step": 6273
    },
    {
      "epoch": 0.26001906419660986,
      "grad_norm": 0.4042254090309143,
      "learning_rate": 3.7001118985453195e-06,
      "loss": 0.6973,
      "step": 6274
    },
    {
      "epoch": 0.26006050810228354,
      "grad_norm": 0.3820721209049225,
      "learning_rate": 3.699904679016951e-06,
      "loss": 0.6603,
      "step": 6275
    },
    {
      "epoch": 0.2601019520079572,
      "grad_norm": 0.4718785285949707,
      "learning_rate": 3.6996974594885827e-06,
      "loss": 0.7434,
      "step": 6276
    },
    {
      "epoch": 0.2601433959136309,
      "grad_norm": 0.4389110207557678,
      "learning_rate": 3.699490239960214e-06,
      "loss": 0.7476,
      "step": 6277
    },
    {
      "epoch": 0.26018483981930457,
      "grad_norm": 0.40768882632255554,
      "learning_rate": 3.699283020431846e-06,
      "loss": 0.6687,
      "step": 6278
    },
    {
      "epoch": 0.26022628372497825,
      "grad_norm": 0.4219058156013489,
      "learning_rate": 3.6990758009034773e-06,
      "loss": 0.7437,
      "step": 6279
    },
    {
      "epoch": 0.2602677276306519,
      "grad_norm": 0.43802395462989807,
      "learning_rate": 3.698868581375109e-06,
      "loss": 0.7358,
      "step": 6280
    },
    {
      "epoch": 0.2603091715363256,
      "grad_norm": 0.4397730827331543,
      "learning_rate": 3.698661361846741e-06,
      "loss": 0.7252,
      "step": 6281
    },
    {
      "epoch": 0.2603506154419993,
      "grad_norm": 0.4375289976596832,
      "learning_rate": 3.6984541423183723e-06,
      "loss": 0.6919,
      "step": 6282
    },
    {
      "epoch": 0.2603920593476729,
      "grad_norm": 0.41898784041404724,
      "learning_rate": 3.698246922790004e-06,
      "loss": 0.6738,
      "step": 6283
    },
    {
      "epoch": 0.2604335032533466,
      "grad_norm": 0.3902156352996826,
      "learning_rate": 3.6980397032616355e-06,
      "loss": 0.7195,
      "step": 6284
    },
    {
      "epoch": 0.26047494715902025,
      "grad_norm": 0.43567222356796265,
      "learning_rate": 3.6978324837332673e-06,
      "loss": 0.75,
      "step": 6285
    },
    {
      "epoch": 0.26051639106469393,
      "grad_norm": 0.4213079512119293,
      "learning_rate": 3.6976252642048987e-06,
      "loss": 0.6906,
      "step": 6286
    },
    {
      "epoch": 0.2605578349703676,
      "grad_norm": 0.44823721051216125,
      "learning_rate": 3.6974180446765305e-06,
      "loss": 0.7073,
      "step": 6287
    },
    {
      "epoch": 0.2605992788760413,
      "grad_norm": 0.4250064194202423,
      "learning_rate": 3.6972108251481627e-06,
      "loss": 0.7151,
      "step": 6288
    },
    {
      "epoch": 0.26064072278171496,
      "grad_norm": 0.4127296805381775,
      "learning_rate": 3.6970036056197937e-06,
      "loss": 0.7314,
      "step": 6289
    },
    {
      "epoch": 0.26068216668738864,
      "grad_norm": 0.37382060289382935,
      "learning_rate": 3.696796386091426e-06,
      "loss": 0.6868,
      "step": 6290
    },
    {
      "epoch": 0.2607236105930623,
      "grad_norm": 0.40120387077331543,
      "learning_rate": 3.6965891665630573e-06,
      "loss": 0.7058,
      "step": 6291
    },
    {
      "epoch": 0.26076505449873594,
      "grad_norm": 0.4066137969493866,
      "learning_rate": 3.696381947034689e-06,
      "loss": 0.6912,
      "step": 6292
    },
    {
      "epoch": 0.2608064984044096,
      "grad_norm": 0.38882172107696533,
      "learning_rate": 3.6961747275063205e-06,
      "loss": 0.7222,
      "step": 6293
    },
    {
      "epoch": 0.2608479423100833,
      "grad_norm": 0.4207928478717804,
      "learning_rate": 3.6959675079779523e-06,
      "loss": 0.7476,
      "step": 6294
    },
    {
      "epoch": 0.26088938621575697,
      "grad_norm": 0.42194119095802307,
      "learning_rate": 3.6957602884495837e-06,
      "loss": 0.7114,
      "step": 6295
    },
    {
      "epoch": 0.26093083012143065,
      "grad_norm": 0.39714813232421875,
      "learning_rate": 3.6955530689212155e-06,
      "loss": 0.6753,
      "step": 6296
    },
    {
      "epoch": 0.2609722740271043,
      "grad_norm": 0.44001829624176025,
      "learning_rate": 3.6953458493928473e-06,
      "loss": 0.7146,
      "step": 6297
    },
    {
      "epoch": 0.261013717932778,
      "grad_norm": 0.41741615533828735,
      "learning_rate": 3.6951386298644787e-06,
      "loss": 0.7354,
      "step": 6298
    },
    {
      "epoch": 0.2610551618384517,
      "grad_norm": 0.4478265345096588,
      "learning_rate": 3.6949314103361105e-06,
      "loss": 0.7429,
      "step": 6299
    },
    {
      "epoch": 0.26109660574412535,
      "grad_norm": 0.42756351828575134,
      "learning_rate": 3.694724190807742e-06,
      "loss": 0.7056,
      "step": 6300
    },
    {
      "epoch": 0.261138049649799,
      "grad_norm": 0.4073184132575989,
      "learning_rate": 3.6945169712793737e-06,
      "loss": 0.7039,
      "step": 6301
    },
    {
      "epoch": 0.26117949355547265,
      "grad_norm": 0.40883225202560425,
      "learning_rate": 3.694309751751005e-06,
      "loss": 0.7604,
      "step": 6302
    },
    {
      "epoch": 0.26122093746114633,
      "grad_norm": 0.42545434832572937,
      "learning_rate": 3.694102532222637e-06,
      "loss": 0.7839,
      "step": 6303
    },
    {
      "epoch": 0.26126238136682,
      "grad_norm": 0.44375982880592346,
      "learning_rate": 3.6938953126942683e-06,
      "loss": 0.7034,
      "step": 6304
    },
    {
      "epoch": 0.2613038252724937,
      "grad_norm": 0.41876929998397827,
      "learning_rate": 3.6936880931659e-06,
      "loss": 0.7417,
      "step": 6305
    },
    {
      "epoch": 0.26134526917816736,
      "grad_norm": 0.43917280435562134,
      "learning_rate": 3.6934808736375323e-06,
      "loss": 0.7205,
      "step": 6306
    },
    {
      "epoch": 0.26138671308384104,
      "grad_norm": 0.44418275356292725,
      "learning_rate": 3.6932736541091637e-06,
      "loss": 0.7347,
      "step": 6307
    },
    {
      "epoch": 0.2614281569895147,
      "grad_norm": 0.4193587005138397,
      "learning_rate": 3.6930664345807955e-06,
      "loss": 0.678,
      "step": 6308
    },
    {
      "epoch": 0.26146960089518834,
      "grad_norm": 0.4175031781196594,
      "learning_rate": 3.692859215052427e-06,
      "loss": 0.7229,
      "step": 6309
    },
    {
      "epoch": 0.261511044800862,
      "grad_norm": 0.4174012839794159,
      "learning_rate": 3.6926519955240587e-06,
      "loss": 0.6697,
      "step": 6310
    },
    {
      "epoch": 0.2615524887065357,
      "grad_norm": 0.4294160306453705,
      "learning_rate": 3.69244477599569e-06,
      "loss": 0.762,
      "step": 6311
    },
    {
      "epoch": 0.26159393261220937,
      "grad_norm": 0.4562845528125763,
      "learning_rate": 3.692237556467322e-06,
      "loss": 0.7053,
      "step": 6312
    },
    {
      "epoch": 0.26163537651788304,
      "grad_norm": 0.4365430176258087,
      "learning_rate": 3.6920303369389533e-06,
      "loss": 0.6494,
      "step": 6313
    },
    {
      "epoch": 0.2616768204235567,
      "grad_norm": 0.4535042345523834,
      "learning_rate": 3.691823117410585e-06,
      "loss": 0.7522,
      "step": 6314
    },
    {
      "epoch": 0.2617182643292304,
      "grad_norm": 0.4188600480556488,
      "learning_rate": 3.691615897882217e-06,
      "loss": 0.7795,
      "step": 6315
    },
    {
      "epoch": 0.2617597082349041,
      "grad_norm": 0.40750449895858765,
      "learning_rate": 3.6914086783538483e-06,
      "loss": 0.7018,
      "step": 6316
    },
    {
      "epoch": 0.26180115214057775,
      "grad_norm": 0.42078661918640137,
      "learning_rate": 3.69120145882548e-06,
      "loss": 0.6948,
      "step": 6317
    },
    {
      "epoch": 0.2618425960462514,
      "grad_norm": 0.40222805738449097,
      "learning_rate": 3.6909942392971115e-06,
      "loss": 0.6687,
      "step": 6318
    },
    {
      "epoch": 0.26188403995192505,
      "grad_norm": 0.4088183045387268,
      "learning_rate": 3.6907870197687433e-06,
      "loss": 0.7329,
      "step": 6319
    },
    {
      "epoch": 0.26192548385759873,
      "grad_norm": 0.3934989869594574,
      "learning_rate": 3.6905798002403747e-06,
      "loss": 0.6893,
      "step": 6320
    },
    {
      "epoch": 0.2619669277632724,
      "grad_norm": 0.3926834464073181,
      "learning_rate": 3.6903725807120065e-06,
      "loss": 0.72,
      "step": 6321
    },
    {
      "epoch": 0.2620083716689461,
      "grad_norm": 0.435479998588562,
      "learning_rate": 3.690165361183638e-06,
      "loss": 0.7522,
      "step": 6322
    },
    {
      "epoch": 0.26204981557461976,
      "grad_norm": 0.4318770468235016,
      "learning_rate": 3.6899581416552697e-06,
      "loss": 0.7498,
      "step": 6323
    },
    {
      "epoch": 0.26209125948029344,
      "grad_norm": 0.4274018704891205,
      "learning_rate": 3.689750922126902e-06,
      "loss": 0.7146,
      "step": 6324
    },
    {
      "epoch": 0.2621327033859671,
      "grad_norm": 0.4307129383087158,
      "learning_rate": 3.6895437025985333e-06,
      "loss": 0.6587,
      "step": 6325
    },
    {
      "epoch": 0.2621741472916408,
      "grad_norm": 0.42974191904067993,
      "learning_rate": 3.689336483070165e-06,
      "loss": 0.7151,
      "step": 6326
    },
    {
      "epoch": 0.2622155911973144,
      "grad_norm": 0.4023016691207886,
      "learning_rate": 3.6891292635417965e-06,
      "loss": 0.6248,
      "step": 6327
    },
    {
      "epoch": 0.2622570351029881,
      "grad_norm": 0.38814854621887207,
      "learning_rate": 3.6889220440134283e-06,
      "loss": 0.728,
      "step": 6328
    },
    {
      "epoch": 0.26229847900866177,
      "grad_norm": 0.3987759053707123,
      "learning_rate": 3.6887148244850597e-06,
      "loss": 0.7273,
      "step": 6329
    },
    {
      "epoch": 0.26233992291433544,
      "grad_norm": 0.420589804649353,
      "learning_rate": 3.6885076049566915e-06,
      "loss": 0.7642,
      "step": 6330
    },
    {
      "epoch": 0.2623813668200091,
      "grad_norm": 0.4020451307296753,
      "learning_rate": 3.688300385428323e-06,
      "loss": 0.6565,
      "step": 6331
    },
    {
      "epoch": 0.2624228107256828,
      "grad_norm": 0.41302037239074707,
      "learning_rate": 3.6880931658999547e-06,
      "loss": 0.7073,
      "step": 6332
    },
    {
      "epoch": 0.2624642546313565,
      "grad_norm": 0.4590699076652527,
      "learning_rate": 3.6878859463715865e-06,
      "loss": 0.674,
      "step": 6333
    },
    {
      "epoch": 0.26250569853703015,
      "grad_norm": 0.43627530336380005,
      "learning_rate": 3.687678726843218e-06,
      "loss": 0.7462,
      "step": 6334
    },
    {
      "epoch": 0.2625471424427038,
      "grad_norm": 0.41981321573257446,
      "learning_rate": 3.6874715073148497e-06,
      "loss": 0.7091,
      "step": 6335
    },
    {
      "epoch": 0.26258858634837745,
      "grad_norm": 0.40838518738746643,
      "learning_rate": 3.687264287786481e-06,
      "loss": 0.6886,
      "step": 6336
    },
    {
      "epoch": 0.26263003025405113,
      "grad_norm": 0.44184985756874084,
      "learning_rate": 3.687057068258113e-06,
      "loss": 0.6758,
      "step": 6337
    },
    {
      "epoch": 0.2626714741597248,
      "grad_norm": 0.38131433725357056,
      "learning_rate": 3.6868498487297443e-06,
      "loss": 0.6975,
      "step": 6338
    },
    {
      "epoch": 0.2627129180653985,
      "grad_norm": 0.38963258266448975,
      "learning_rate": 3.686642629201376e-06,
      "loss": 0.6221,
      "step": 6339
    },
    {
      "epoch": 0.26275436197107216,
      "grad_norm": 0.4219125509262085,
      "learning_rate": 3.6864354096730075e-06,
      "loss": 0.6963,
      "step": 6340
    },
    {
      "epoch": 0.26279580587674584,
      "grad_norm": 0.4498273432254791,
      "learning_rate": 3.6862281901446397e-06,
      "loss": 0.7437,
      "step": 6341
    },
    {
      "epoch": 0.2628372497824195,
      "grad_norm": 0.4100957214832306,
      "learning_rate": 3.6860209706162715e-06,
      "loss": 0.6973,
      "step": 6342
    },
    {
      "epoch": 0.2628786936880932,
      "grad_norm": 0.46511566638946533,
      "learning_rate": 3.685813751087903e-06,
      "loss": 0.7279,
      "step": 6343
    },
    {
      "epoch": 0.2629201375937668,
      "grad_norm": 0.43290814757347107,
      "learning_rate": 3.6856065315595347e-06,
      "loss": 0.7004,
      "step": 6344
    },
    {
      "epoch": 0.2629615814994405,
      "grad_norm": 0.4244616627693176,
      "learning_rate": 3.685399312031166e-06,
      "loss": 0.7302,
      "step": 6345
    },
    {
      "epoch": 0.26300302540511417,
      "grad_norm": 0.41384050250053406,
      "learning_rate": 3.685192092502798e-06,
      "loss": 0.7363,
      "step": 6346
    },
    {
      "epoch": 0.26304446931078784,
      "grad_norm": 0.4220217168331146,
      "learning_rate": 3.6849848729744293e-06,
      "loss": 0.6934,
      "step": 6347
    },
    {
      "epoch": 0.2630859132164615,
      "grad_norm": 0.46823200583457947,
      "learning_rate": 3.684777653446061e-06,
      "loss": 0.7351,
      "step": 6348
    },
    {
      "epoch": 0.2631273571221352,
      "grad_norm": 0.4548875391483307,
      "learning_rate": 3.684570433917693e-06,
      "loss": 0.7567,
      "step": 6349
    },
    {
      "epoch": 0.2631688010278089,
      "grad_norm": 0.43491724133491516,
      "learning_rate": 3.6843632143893243e-06,
      "loss": 0.6829,
      "step": 6350
    },
    {
      "epoch": 0.26321024493348255,
      "grad_norm": 0.40387147665023804,
      "learning_rate": 3.684155994860956e-06,
      "loss": 0.6733,
      "step": 6351
    },
    {
      "epoch": 0.26325168883915623,
      "grad_norm": 0.4099508225917816,
      "learning_rate": 3.6839487753325875e-06,
      "loss": 0.7078,
      "step": 6352
    },
    {
      "epoch": 0.26329313274482985,
      "grad_norm": 0.42148905992507935,
      "learning_rate": 3.6837415558042193e-06,
      "loss": 0.7048,
      "step": 6353
    },
    {
      "epoch": 0.2633345766505035,
      "grad_norm": 0.450489342212677,
      "learning_rate": 3.6835343362758507e-06,
      "loss": 0.7576,
      "step": 6354
    },
    {
      "epoch": 0.2633760205561772,
      "grad_norm": 0.40780889987945557,
      "learning_rate": 3.6833271167474825e-06,
      "loss": 0.6941,
      "step": 6355
    },
    {
      "epoch": 0.2634174644618509,
      "grad_norm": 0.45322784781455994,
      "learning_rate": 3.683119897219114e-06,
      "loss": 0.7654,
      "step": 6356
    },
    {
      "epoch": 0.26345890836752456,
      "grad_norm": 0.39057061076164246,
      "learning_rate": 3.6829126776907457e-06,
      "loss": 0.7151,
      "step": 6357
    },
    {
      "epoch": 0.26350035227319824,
      "grad_norm": 0.39810916781425476,
      "learning_rate": 3.682705458162378e-06,
      "loss": 0.6925,
      "step": 6358
    },
    {
      "epoch": 0.2635417961788719,
      "grad_norm": 0.42373794317245483,
      "learning_rate": 3.6824982386340093e-06,
      "loss": 0.7617,
      "step": 6359
    },
    {
      "epoch": 0.2635832400845456,
      "grad_norm": 0.44631630182266235,
      "learning_rate": 3.682291019105641e-06,
      "loss": 0.7673,
      "step": 6360
    },
    {
      "epoch": 0.2636246839902192,
      "grad_norm": 0.3960021734237671,
      "learning_rate": 3.6820837995772725e-06,
      "loss": 0.7035,
      "step": 6361
    },
    {
      "epoch": 0.2636661278958929,
      "grad_norm": 0.41545552015304565,
      "learning_rate": 3.6818765800489043e-06,
      "loss": 0.7275,
      "step": 6362
    },
    {
      "epoch": 0.26370757180156656,
      "grad_norm": 0.4479779303073883,
      "learning_rate": 3.6816693605205357e-06,
      "loss": 0.7133,
      "step": 6363
    },
    {
      "epoch": 0.26374901570724024,
      "grad_norm": 0.41297903656959534,
      "learning_rate": 3.6814621409921675e-06,
      "loss": 0.7292,
      "step": 6364
    },
    {
      "epoch": 0.2637904596129139,
      "grad_norm": 0.4408339858055115,
      "learning_rate": 3.681254921463799e-06,
      "loss": 0.7407,
      "step": 6365
    },
    {
      "epoch": 0.2638319035185876,
      "grad_norm": 0.42763832211494446,
      "learning_rate": 3.6810477019354307e-06,
      "loss": 0.7297,
      "step": 6366
    },
    {
      "epoch": 0.2638733474242613,
      "grad_norm": 0.46042579412460327,
      "learning_rate": 3.6808404824070625e-06,
      "loss": 0.7437,
      "step": 6367
    },
    {
      "epoch": 0.26391479132993495,
      "grad_norm": 0.44522392749786377,
      "learning_rate": 3.680633262878694e-06,
      "loss": 0.7048,
      "step": 6368
    },
    {
      "epoch": 0.2639562352356086,
      "grad_norm": 0.42436856031417847,
      "learning_rate": 3.6804260433503257e-06,
      "loss": 0.7113,
      "step": 6369
    },
    {
      "epoch": 0.26399767914128225,
      "grad_norm": 0.44359612464904785,
      "learning_rate": 3.680218823821957e-06,
      "loss": 0.7742,
      "step": 6370
    },
    {
      "epoch": 0.2640391230469559,
      "grad_norm": 0.41687852144241333,
      "learning_rate": 3.680011604293589e-06,
      "loss": 0.6907,
      "step": 6371
    },
    {
      "epoch": 0.2640805669526296,
      "grad_norm": 0.4312986731529236,
      "learning_rate": 3.6798043847652203e-06,
      "loss": 0.7466,
      "step": 6372
    },
    {
      "epoch": 0.2641220108583033,
      "grad_norm": 0.4330451488494873,
      "learning_rate": 3.679597165236852e-06,
      "loss": 0.6919,
      "step": 6373
    },
    {
      "epoch": 0.26416345476397696,
      "grad_norm": 0.43585044145584106,
      "learning_rate": 3.6793899457084835e-06,
      "loss": 0.7639,
      "step": 6374
    },
    {
      "epoch": 0.26420489866965063,
      "grad_norm": 0.442184716463089,
      "learning_rate": 3.6791827261801157e-06,
      "loss": 0.7219,
      "step": 6375
    },
    {
      "epoch": 0.2642463425753243,
      "grad_norm": 0.41909539699554443,
      "learning_rate": 3.6789755066517475e-06,
      "loss": 0.7004,
      "step": 6376
    },
    {
      "epoch": 0.264287786480998,
      "grad_norm": 0.4056285619735718,
      "learning_rate": 3.678768287123379e-06,
      "loss": 0.6863,
      "step": 6377
    },
    {
      "epoch": 0.26432923038667167,
      "grad_norm": 0.4126576781272888,
      "learning_rate": 3.6785610675950107e-06,
      "loss": 0.7012,
      "step": 6378
    },
    {
      "epoch": 0.2643706742923453,
      "grad_norm": 0.4524768888950348,
      "learning_rate": 3.678353848066642e-06,
      "loss": 0.7181,
      "step": 6379
    },
    {
      "epoch": 0.26441211819801896,
      "grad_norm": 0.4047909677028656,
      "learning_rate": 3.678146628538274e-06,
      "loss": 0.748,
      "step": 6380
    },
    {
      "epoch": 0.26445356210369264,
      "grad_norm": 0.4494902789592743,
      "learning_rate": 3.6779394090099053e-06,
      "loss": 0.7419,
      "step": 6381
    },
    {
      "epoch": 0.2644950060093663,
      "grad_norm": 0.4608725309371948,
      "learning_rate": 3.677732189481537e-06,
      "loss": 0.762,
      "step": 6382
    },
    {
      "epoch": 0.26453644991504,
      "grad_norm": 0.4417451024055481,
      "learning_rate": 3.6775249699531685e-06,
      "loss": 0.7363,
      "step": 6383
    },
    {
      "epoch": 0.26457789382071367,
      "grad_norm": 0.40939566493034363,
      "learning_rate": 3.6773177504248003e-06,
      "loss": 0.7197,
      "step": 6384
    },
    {
      "epoch": 0.26461933772638735,
      "grad_norm": 0.41147902607917786,
      "learning_rate": 3.677110530896432e-06,
      "loss": 0.7529,
      "step": 6385
    },
    {
      "epoch": 0.264660781632061,
      "grad_norm": 0.41552749276161194,
      "learning_rate": 3.6769033113680635e-06,
      "loss": 0.6954,
      "step": 6386
    },
    {
      "epoch": 0.2647022255377347,
      "grad_norm": 0.4078802168369293,
      "learning_rate": 3.6766960918396953e-06,
      "loss": 0.6814,
      "step": 6387
    },
    {
      "epoch": 0.2647436694434083,
      "grad_norm": 0.4642527103424072,
      "learning_rate": 3.6764888723113267e-06,
      "loss": 0.739,
      "step": 6388
    },
    {
      "epoch": 0.264785113349082,
      "grad_norm": 0.42460986971855164,
      "learning_rate": 3.6762816527829585e-06,
      "loss": 0.705,
      "step": 6389
    },
    {
      "epoch": 0.2648265572547557,
      "grad_norm": 0.42757096886634827,
      "learning_rate": 3.67607443325459e-06,
      "loss": 0.7211,
      "step": 6390
    },
    {
      "epoch": 0.26486800116042936,
      "grad_norm": 0.42514774203300476,
      "learning_rate": 3.6758672137262217e-06,
      "loss": 0.722,
      "step": 6391
    },
    {
      "epoch": 0.26490944506610303,
      "grad_norm": 0.4477381408214569,
      "learning_rate": 3.675659994197853e-06,
      "loss": 0.7505,
      "step": 6392
    },
    {
      "epoch": 0.2649508889717767,
      "grad_norm": 0.4089907705783844,
      "learning_rate": 3.6754527746694853e-06,
      "loss": 0.6439,
      "step": 6393
    },
    {
      "epoch": 0.2649923328774504,
      "grad_norm": 0.3891829252243042,
      "learning_rate": 3.675245555141117e-06,
      "loss": 0.7164,
      "step": 6394
    },
    {
      "epoch": 0.26503377678312406,
      "grad_norm": 0.43315836787223816,
      "learning_rate": 3.6750383356127485e-06,
      "loss": 0.6956,
      "step": 6395
    },
    {
      "epoch": 0.2650752206887977,
      "grad_norm": 0.4007335603237152,
      "learning_rate": 3.6748311160843803e-06,
      "loss": 0.6899,
      "step": 6396
    },
    {
      "epoch": 0.26511666459447136,
      "grad_norm": 0.4487016201019287,
      "learning_rate": 3.6746238965560117e-06,
      "loss": 0.7708,
      "step": 6397
    },
    {
      "epoch": 0.26515810850014504,
      "grad_norm": 0.44069451093673706,
      "learning_rate": 3.6744166770276435e-06,
      "loss": 0.6975,
      "step": 6398
    },
    {
      "epoch": 0.2651995524058187,
      "grad_norm": 0.46565812826156616,
      "learning_rate": 3.674209457499275e-06,
      "loss": 0.7306,
      "step": 6399
    },
    {
      "epoch": 0.2652409963114924,
      "grad_norm": 0.40225380659103394,
      "learning_rate": 3.6740022379709067e-06,
      "loss": 0.6874,
      "step": 6400
    },
    {
      "epoch": 0.26528244021716607,
      "grad_norm": 0.4467470645904541,
      "learning_rate": 3.673795018442538e-06,
      "loss": 0.7396,
      "step": 6401
    },
    {
      "epoch": 0.26532388412283975,
      "grad_norm": 0.406599223613739,
      "learning_rate": 3.67358779891417e-06,
      "loss": 0.6985,
      "step": 6402
    },
    {
      "epoch": 0.2653653280285134,
      "grad_norm": 0.42188259959220886,
      "learning_rate": 3.6733805793858017e-06,
      "loss": 0.7122,
      "step": 6403
    },
    {
      "epoch": 0.2654067719341871,
      "grad_norm": 0.4489630162715912,
      "learning_rate": 3.673173359857433e-06,
      "loss": 0.6533,
      "step": 6404
    },
    {
      "epoch": 0.2654482158398607,
      "grad_norm": 0.4150291681289673,
      "learning_rate": 3.672966140329065e-06,
      "loss": 0.7026,
      "step": 6405
    },
    {
      "epoch": 0.2654896597455344,
      "grad_norm": 0.4480176270008087,
      "learning_rate": 3.6727589208006963e-06,
      "loss": 0.6926,
      "step": 6406
    },
    {
      "epoch": 0.2655311036512081,
      "grad_norm": 0.4263271689414978,
      "learning_rate": 3.672551701272328e-06,
      "loss": 0.6873,
      "step": 6407
    },
    {
      "epoch": 0.26557254755688176,
      "grad_norm": 0.3974429666996002,
      "learning_rate": 3.6723444817439595e-06,
      "loss": 0.6924,
      "step": 6408
    },
    {
      "epoch": 0.26561399146255543,
      "grad_norm": 0.447875440120697,
      "learning_rate": 3.6721372622155917e-06,
      "loss": 0.7415,
      "step": 6409
    },
    {
      "epoch": 0.2656554353682291,
      "grad_norm": 0.4001208245754242,
      "learning_rate": 3.6719300426872235e-06,
      "loss": 0.7051,
      "step": 6410
    },
    {
      "epoch": 0.2656968792739028,
      "grad_norm": 0.4122028648853302,
      "learning_rate": 3.671722823158855e-06,
      "loss": 0.7593,
      "step": 6411
    },
    {
      "epoch": 0.26573832317957646,
      "grad_norm": 0.4092101752758026,
      "learning_rate": 3.6715156036304867e-06,
      "loss": 0.7118,
      "step": 6412
    },
    {
      "epoch": 0.26577976708525014,
      "grad_norm": 0.4194504916667938,
      "learning_rate": 3.671308384102118e-06,
      "loss": 0.7118,
      "step": 6413
    },
    {
      "epoch": 0.26582121099092376,
      "grad_norm": 0.4180678427219391,
      "learning_rate": 3.67110116457375e-06,
      "loss": 0.7251,
      "step": 6414
    },
    {
      "epoch": 0.26586265489659744,
      "grad_norm": 0.43582284450531006,
      "learning_rate": 3.6708939450453813e-06,
      "loss": 0.7334,
      "step": 6415
    },
    {
      "epoch": 0.2659040988022711,
      "grad_norm": 0.4737831652164459,
      "learning_rate": 3.670686725517013e-06,
      "loss": 0.7511,
      "step": 6416
    },
    {
      "epoch": 0.2659455427079448,
      "grad_norm": 0.42427587509155273,
      "learning_rate": 3.6704795059886445e-06,
      "loss": 0.6963,
      "step": 6417
    },
    {
      "epoch": 0.26598698661361847,
      "grad_norm": 0.436360239982605,
      "learning_rate": 3.6702722864602763e-06,
      "loss": 0.7324,
      "step": 6418
    },
    {
      "epoch": 0.26602843051929215,
      "grad_norm": 0.4255404770374298,
      "learning_rate": 3.670065066931908e-06,
      "loss": 0.7314,
      "step": 6419
    },
    {
      "epoch": 0.2660698744249658,
      "grad_norm": 0.4210447669029236,
      "learning_rate": 3.6698578474035395e-06,
      "loss": 0.7466,
      "step": 6420
    },
    {
      "epoch": 0.2661113183306395,
      "grad_norm": 0.4220668375492096,
      "learning_rate": 3.6696506278751713e-06,
      "loss": 0.7144,
      "step": 6421
    },
    {
      "epoch": 0.2661527622363131,
      "grad_norm": 0.42769214510917664,
      "learning_rate": 3.6694434083468027e-06,
      "loss": 0.7019,
      "step": 6422
    },
    {
      "epoch": 0.2661942061419868,
      "grad_norm": 0.4240381121635437,
      "learning_rate": 3.6692361888184345e-06,
      "loss": 0.702,
      "step": 6423
    },
    {
      "epoch": 0.2662356500476605,
      "grad_norm": 0.4884326159954071,
      "learning_rate": 3.669028969290066e-06,
      "loss": 0.7151,
      "step": 6424
    },
    {
      "epoch": 0.26627709395333415,
      "grad_norm": 0.43229031562805176,
      "learning_rate": 3.6688217497616977e-06,
      "loss": 0.7173,
      "step": 6425
    },
    {
      "epoch": 0.26631853785900783,
      "grad_norm": 0.4293804168701172,
      "learning_rate": 3.668614530233329e-06,
      "loss": 0.6862,
      "step": 6426
    },
    {
      "epoch": 0.2663599817646815,
      "grad_norm": 0.3997860252857208,
      "learning_rate": 3.6684073107049613e-06,
      "loss": 0.6982,
      "step": 6427
    },
    {
      "epoch": 0.2664014256703552,
      "grad_norm": 0.462211936712265,
      "learning_rate": 3.668200091176593e-06,
      "loss": 0.739,
      "step": 6428
    },
    {
      "epoch": 0.26644286957602886,
      "grad_norm": 0.3971467614173889,
      "learning_rate": 3.6679928716482245e-06,
      "loss": 0.6676,
      "step": 6429
    },
    {
      "epoch": 0.26648431348170254,
      "grad_norm": 0.4137811064720154,
      "learning_rate": 3.6677856521198563e-06,
      "loss": 0.7159,
      "step": 6430
    },
    {
      "epoch": 0.26652575738737616,
      "grad_norm": 0.4237821102142334,
      "learning_rate": 3.6675784325914877e-06,
      "loss": 0.7866,
      "step": 6431
    },
    {
      "epoch": 0.26656720129304984,
      "grad_norm": 0.41758179664611816,
      "learning_rate": 3.6673712130631195e-06,
      "loss": 0.7096,
      "step": 6432
    },
    {
      "epoch": 0.2666086451987235,
      "grad_norm": 0.45064622163772583,
      "learning_rate": 3.667163993534751e-06,
      "loss": 0.6929,
      "step": 6433
    },
    {
      "epoch": 0.2666500891043972,
      "grad_norm": 0.4277627170085907,
      "learning_rate": 3.6669567740063827e-06,
      "loss": 0.761,
      "step": 6434
    },
    {
      "epoch": 0.26669153301007087,
      "grad_norm": 0.44265028834342957,
      "learning_rate": 3.666749554478014e-06,
      "loss": 0.7246,
      "step": 6435
    },
    {
      "epoch": 0.26673297691574455,
      "grad_norm": 0.4414396286010742,
      "learning_rate": 3.666542334949646e-06,
      "loss": 0.6768,
      "step": 6436
    },
    {
      "epoch": 0.2667744208214182,
      "grad_norm": 0.39849019050598145,
      "learning_rate": 3.6663351154212777e-06,
      "loss": 0.6882,
      "step": 6437
    },
    {
      "epoch": 0.2668158647270919,
      "grad_norm": 0.4140229821205139,
      "learning_rate": 3.666127895892909e-06,
      "loss": 0.7466,
      "step": 6438
    },
    {
      "epoch": 0.2668573086327656,
      "grad_norm": 0.431058406829834,
      "learning_rate": 3.665920676364541e-06,
      "loss": 0.7037,
      "step": 6439
    },
    {
      "epoch": 0.2668987525384392,
      "grad_norm": 0.3874143660068512,
      "learning_rate": 3.6657134568361723e-06,
      "loss": 0.7214,
      "step": 6440
    },
    {
      "epoch": 0.2669401964441129,
      "grad_norm": 0.4047558605670929,
      "learning_rate": 3.665506237307804e-06,
      "loss": 0.6979,
      "step": 6441
    },
    {
      "epoch": 0.26698164034978655,
      "grad_norm": 0.4260207414627075,
      "learning_rate": 3.6652990177794355e-06,
      "loss": 0.7369,
      "step": 6442
    },
    {
      "epoch": 0.26702308425546023,
      "grad_norm": 0.4203431308269501,
      "learning_rate": 3.6650917982510677e-06,
      "loss": 0.7595,
      "step": 6443
    },
    {
      "epoch": 0.2670645281611339,
      "grad_norm": 0.44023802876472473,
      "learning_rate": 3.6648845787226987e-06,
      "loss": 0.6681,
      "step": 6444
    },
    {
      "epoch": 0.2671059720668076,
      "grad_norm": 0.4268922209739685,
      "learning_rate": 3.664677359194331e-06,
      "loss": 0.7581,
      "step": 6445
    },
    {
      "epoch": 0.26714741597248126,
      "grad_norm": 0.4216556251049042,
      "learning_rate": 3.6644701396659627e-06,
      "loss": 0.7388,
      "step": 6446
    },
    {
      "epoch": 0.26718885987815494,
      "grad_norm": 0.4321242570877075,
      "learning_rate": 3.664262920137594e-06,
      "loss": 0.7217,
      "step": 6447
    },
    {
      "epoch": 0.2672303037838286,
      "grad_norm": 0.4266868531703949,
      "learning_rate": 3.664055700609226e-06,
      "loss": 0.7427,
      "step": 6448
    },
    {
      "epoch": 0.26727174768950224,
      "grad_norm": 0.41267094016075134,
      "learning_rate": 3.6638484810808573e-06,
      "loss": 0.6991,
      "step": 6449
    },
    {
      "epoch": 0.2673131915951759,
      "grad_norm": 0.39347052574157715,
      "learning_rate": 3.663641261552489e-06,
      "loss": 0.7659,
      "step": 6450
    },
    {
      "epoch": 0.2673546355008496,
      "grad_norm": 0.425607830286026,
      "learning_rate": 3.6634340420241205e-06,
      "loss": 0.6931,
      "step": 6451
    },
    {
      "epoch": 0.26739607940652327,
      "grad_norm": 0.3932659924030304,
      "learning_rate": 3.6632268224957523e-06,
      "loss": 0.6582,
      "step": 6452
    },
    {
      "epoch": 0.26743752331219695,
      "grad_norm": 0.38975101709365845,
      "learning_rate": 3.6630196029673837e-06,
      "loss": 0.7083,
      "step": 6453
    },
    {
      "epoch": 0.2674789672178706,
      "grad_norm": 0.4292771518230438,
      "learning_rate": 3.6628123834390155e-06,
      "loss": 0.7197,
      "step": 6454
    },
    {
      "epoch": 0.2675204111235443,
      "grad_norm": 0.40929049253463745,
      "learning_rate": 3.6626051639106473e-06,
      "loss": 0.7073,
      "step": 6455
    },
    {
      "epoch": 0.267561855029218,
      "grad_norm": 0.41723617911338806,
      "learning_rate": 3.6623979443822787e-06,
      "loss": 0.6682,
      "step": 6456
    },
    {
      "epoch": 0.2676032989348916,
      "grad_norm": 0.4117160141468048,
      "learning_rate": 3.6621907248539105e-06,
      "loss": 0.7393,
      "step": 6457
    },
    {
      "epoch": 0.2676447428405653,
      "grad_norm": 0.4089653491973877,
      "learning_rate": 3.661983505325542e-06,
      "loss": 0.7305,
      "step": 6458
    },
    {
      "epoch": 0.26768618674623895,
      "grad_norm": 0.43648481369018555,
      "learning_rate": 3.661776285797174e-06,
      "loss": 0.6902,
      "step": 6459
    },
    {
      "epoch": 0.26772763065191263,
      "grad_norm": 0.40313035249710083,
      "learning_rate": 3.661569066268805e-06,
      "loss": 0.7368,
      "step": 6460
    },
    {
      "epoch": 0.2677690745575863,
      "grad_norm": 0.44201716780662537,
      "learning_rate": 3.6613618467404373e-06,
      "loss": 0.741,
      "step": 6461
    },
    {
      "epoch": 0.26781051846326,
      "grad_norm": 0.4262681007385254,
      "learning_rate": 3.6611546272120683e-06,
      "loss": 0.6985,
      "step": 6462
    },
    {
      "epoch": 0.26785196236893366,
      "grad_norm": 0.42730364203453064,
      "learning_rate": 3.6609474076837005e-06,
      "loss": 0.7054,
      "step": 6463
    },
    {
      "epoch": 0.26789340627460734,
      "grad_norm": 0.41901320219039917,
      "learning_rate": 3.6607401881553323e-06,
      "loss": 0.7102,
      "step": 6464
    },
    {
      "epoch": 0.267934850180281,
      "grad_norm": 0.43879425525665283,
      "learning_rate": 3.6605329686269637e-06,
      "loss": 0.7192,
      "step": 6465
    },
    {
      "epoch": 0.26797629408595464,
      "grad_norm": 0.43288737535476685,
      "learning_rate": 3.6603257490985955e-06,
      "loss": 0.7283,
      "step": 6466
    },
    {
      "epoch": 0.2680177379916283,
      "grad_norm": 0.42803654074668884,
      "learning_rate": 3.660118529570227e-06,
      "loss": 0.6982,
      "step": 6467
    },
    {
      "epoch": 0.268059181897302,
      "grad_norm": 0.433649480342865,
      "learning_rate": 3.6599113100418587e-06,
      "loss": 0.6531,
      "step": 6468
    },
    {
      "epoch": 0.26810062580297567,
      "grad_norm": 0.4290996789932251,
      "learning_rate": 3.65970409051349e-06,
      "loss": 0.7219,
      "step": 6469
    },
    {
      "epoch": 0.26814206970864934,
      "grad_norm": 0.42780062556266785,
      "learning_rate": 3.659496870985122e-06,
      "loss": 0.7267,
      "step": 6470
    },
    {
      "epoch": 0.268183513614323,
      "grad_norm": 0.4274875521659851,
      "learning_rate": 3.6592896514567537e-06,
      "loss": 0.7134,
      "step": 6471
    },
    {
      "epoch": 0.2682249575199967,
      "grad_norm": 0.4348224401473999,
      "learning_rate": 3.659082431928385e-06,
      "loss": 0.7258,
      "step": 6472
    },
    {
      "epoch": 0.2682664014256704,
      "grad_norm": 0.39090046286582947,
      "learning_rate": 3.658875212400017e-06,
      "loss": 0.6897,
      "step": 6473
    },
    {
      "epoch": 0.26830784533134405,
      "grad_norm": 0.41403937339782715,
      "learning_rate": 3.6586679928716483e-06,
      "loss": 0.7417,
      "step": 6474
    },
    {
      "epoch": 0.2683492892370177,
      "grad_norm": 0.39958029985427856,
      "learning_rate": 3.65846077334328e-06,
      "loss": 0.6487,
      "step": 6475
    },
    {
      "epoch": 0.26839073314269135,
      "grad_norm": 0.41907018423080444,
      "learning_rate": 3.6582535538149115e-06,
      "loss": 0.7037,
      "step": 6476
    },
    {
      "epoch": 0.26843217704836503,
      "grad_norm": 0.48555099964141846,
      "learning_rate": 3.6580463342865437e-06,
      "loss": 0.8104,
      "step": 6477
    },
    {
      "epoch": 0.2684736209540387,
      "grad_norm": 0.45586496591567993,
      "learning_rate": 3.6578391147581747e-06,
      "loss": 0.7585,
      "step": 6478
    },
    {
      "epoch": 0.2685150648597124,
      "grad_norm": 0.4613741636276245,
      "learning_rate": 3.657631895229807e-06,
      "loss": 0.7429,
      "step": 6479
    },
    {
      "epoch": 0.26855650876538606,
      "grad_norm": 0.42168542742729187,
      "learning_rate": 3.6574246757014387e-06,
      "loss": 0.709,
      "step": 6480
    },
    {
      "epoch": 0.26859795267105974,
      "grad_norm": 0.409851998090744,
      "learning_rate": 3.65721745617307e-06,
      "loss": 0.6951,
      "step": 6481
    },
    {
      "epoch": 0.2686393965767334,
      "grad_norm": 0.4115036129951477,
      "learning_rate": 3.657010236644702e-06,
      "loss": 0.7083,
      "step": 6482
    },
    {
      "epoch": 0.26868084048240704,
      "grad_norm": 0.4216599464416504,
      "learning_rate": 3.6568030171163333e-06,
      "loss": 0.7297,
      "step": 6483
    },
    {
      "epoch": 0.2687222843880807,
      "grad_norm": 0.41426679491996765,
      "learning_rate": 3.656595797587965e-06,
      "loss": 0.7537,
      "step": 6484
    },
    {
      "epoch": 0.2687637282937544,
      "grad_norm": 0.40590569376945496,
      "learning_rate": 3.6563885780595965e-06,
      "loss": 0.6902,
      "step": 6485
    },
    {
      "epoch": 0.26880517219942807,
      "grad_norm": 0.4273844361305237,
      "learning_rate": 3.6561813585312283e-06,
      "loss": 0.6809,
      "step": 6486
    },
    {
      "epoch": 0.26884661610510174,
      "grad_norm": 0.40438729524612427,
      "learning_rate": 3.6559741390028597e-06,
      "loss": 0.75,
      "step": 6487
    },
    {
      "epoch": 0.2688880600107754,
      "grad_norm": 0.4229685962200165,
      "learning_rate": 3.6557669194744915e-06,
      "loss": 0.7189,
      "step": 6488
    },
    {
      "epoch": 0.2689295039164491,
      "grad_norm": 0.3948942720890045,
      "learning_rate": 3.6555596999461233e-06,
      "loss": 0.7068,
      "step": 6489
    },
    {
      "epoch": 0.2689709478221228,
      "grad_norm": 0.4670775830745697,
      "learning_rate": 3.6553524804177547e-06,
      "loss": 0.7797,
      "step": 6490
    },
    {
      "epoch": 0.26901239172779645,
      "grad_norm": 0.4109322428703308,
      "learning_rate": 3.6551452608893865e-06,
      "loss": 0.6995,
      "step": 6491
    },
    {
      "epoch": 0.2690538356334701,
      "grad_norm": 0.41405197978019714,
      "learning_rate": 3.654938041361018e-06,
      "loss": 0.7074,
      "step": 6492
    },
    {
      "epoch": 0.26909527953914375,
      "grad_norm": 0.40317216515541077,
      "learning_rate": 3.65473082183265e-06,
      "loss": 0.6707,
      "step": 6493
    },
    {
      "epoch": 0.2691367234448174,
      "grad_norm": 0.41362228989601135,
      "learning_rate": 3.654523602304281e-06,
      "loss": 0.7036,
      "step": 6494
    },
    {
      "epoch": 0.2691781673504911,
      "grad_norm": 0.4566660523414612,
      "learning_rate": 3.6543163827759133e-06,
      "loss": 0.7386,
      "step": 6495
    },
    {
      "epoch": 0.2692196112561648,
      "grad_norm": 0.411600798368454,
      "learning_rate": 3.6541091632475443e-06,
      "loss": 0.7126,
      "step": 6496
    },
    {
      "epoch": 0.26926105516183846,
      "grad_norm": 0.41716650128364563,
      "learning_rate": 3.6539019437191765e-06,
      "loss": 0.7178,
      "step": 6497
    },
    {
      "epoch": 0.26930249906751214,
      "grad_norm": 0.394574373960495,
      "learning_rate": 3.6536947241908083e-06,
      "loss": 0.731,
      "step": 6498
    },
    {
      "epoch": 0.2693439429731858,
      "grad_norm": 0.39140281081199646,
      "learning_rate": 3.6534875046624397e-06,
      "loss": 0.7139,
      "step": 6499
    },
    {
      "epoch": 0.2693853868788595,
      "grad_norm": 0.4075028598308563,
      "learning_rate": 3.6532802851340715e-06,
      "loss": 0.7336,
      "step": 6500
    },
    {
      "epoch": 0.2694268307845331,
      "grad_norm": 0.41810670495033264,
      "learning_rate": 3.653073065605703e-06,
      "loss": 0.7471,
      "step": 6501
    },
    {
      "epoch": 0.2694682746902068,
      "grad_norm": 0.4515833854675293,
      "learning_rate": 3.6528658460773347e-06,
      "loss": 0.7764,
      "step": 6502
    },
    {
      "epoch": 0.26950971859588047,
      "grad_norm": 0.4228217601776123,
      "learning_rate": 3.652658626548966e-06,
      "loss": 0.8015,
      "step": 6503
    },
    {
      "epoch": 0.26955116250155414,
      "grad_norm": 0.41787222027778625,
      "learning_rate": 3.652451407020598e-06,
      "loss": 0.6761,
      "step": 6504
    },
    {
      "epoch": 0.2695926064072278,
      "grad_norm": 0.45666253566741943,
      "learning_rate": 3.6522441874922293e-06,
      "loss": 0.7216,
      "step": 6505
    },
    {
      "epoch": 0.2696340503129015,
      "grad_norm": 0.4375092089176178,
      "learning_rate": 3.652036967963861e-06,
      "loss": 0.7053,
      "step": 6506
    },
    {
      "epoch": 0.2696754942185752,
      "grad_norm": 0.44457709789276123,
      "learning_rate": 3.651829748435493e-06,
      "loss": 0.7341,
      "step": 6507
    },
    {
      "epoch": 0.26971693812424885,
      "grad_norm": 0.42417269945144653,
      "learning_rate": 3.6516225289071243e-06,
      "loss": 0.7122,
      "step": 6508
    },
    {
      "epoch": 0.26975838202992247,
      "grad_norm": 0.418937087059021,
      "learning_rate": 3.651415309378756e-06,
      "loss": 0.7085,
      "step": 6509
    },
    {
      "epoch": 0.26979982593559615,
      "grad_norm": 0.40689074993133545,
      "learning_rate": 3.6512080898503875e-06,
      "loss": 0.6692,
      "step": 6510
    },
    {
      "epoch": 0.2698412698412698,
      "grad_norm": 0.42415565252304077,
      "learning_rate": 3.6510008703220197e-06,
      "loss": 0.7255,
      "step": 6511
    },
    {
      "epoch": 0.2698827137469435,
      "grad_norm": 0.3947333097457886,
      "learning_rate": 3.6507936507936507e-06,
      "loss": 0.7051,
      "step": 6512
    },
    {
      "epoch": 0.2699241576526172,
      "grad_norm": 0.375267893075943,
      "learning_rate": 3.650586431265283e-06,
      "loss": 0.6907,
      "step": 6513
    },
    {
      "epoch": 0.26996560155829086,
      "grad_norm": 0.4119984805583954,
      "learning_rate": 3.6503792117369143e-06,
      "loss": 0.7532,
      "step": 6514
    },
    {
      "epoch": 0.27000704546396453,
      "grad_norm": 0.4136695861816406,
      "learning_rate": 3.650171992208546e-06,
      "loss": 0.7468,
      "step": 6515
    },
    {
      "epoch": 0.2700484893696382,
      "grad_norm": 0.41449809074401855,
      "learning_rate": 3.649964772680178e-06,
      "loss": 0.7236,
      "step": 6516
    },
    {
      "epoch": 0.2700899332753119,
      "grad_norm": 0.41555920243263245,
      "learning_rate": 3.6497575531518093e-06,
      "loss": 0.7454,
      "step": 6517
    },
    {
      "epoch": 0.2701313771809855,
      "grad_norm": 0.43922826647758484,
      "learning_rate": 3.649550333623441e-06,
      "loss": 0.7012,
      "step": 6518
    },
    {
      "epoch": 0.2701728210866592,
      "grad_norm": 0.3831327259540558,
      "learning_rate": 3.6493431140950725e-06,
      "loss": 0.6974,
      "step": 6519
    },
    {
      "epoch": 0.27021426499233286,
      "grad_norm": 0.4077354967594147,
      "learning_rate": 3.6491358945667043e-06,
      "loss": 0.647,
      "step": 6520
    },
    {
      "epoch": 0.27025570889800654,
      "grad_norm": 0.4273907244205475,
      "learning_rate": 3.6489286750383357e-06,
      "loss": 0.7275,
      "step": 6521
    },
    {
      "epoch": 0.2702971528036802,
      "grad_norm": 0.4297737777233124,
      "learning_rate": 3.6487214555099675e-06,
      "loss": 0.7585,
      "step": 6522
    },
    {
      "epoch": 0.2703385967093539,
      "grad_norm": 0.4168475866317749,
      "learning_rate": 3.648514235981599e-06,
      "loss": 0.7,
      "step": 6523
    },
    {
      "epoch": 0.2703800406150276,
      "grad_norm": 0.4295629560947418,
      "learning_rate": 3.6483070164532307e-06,
      "loss": 0.7021,
      "step": 6524
    },
    {
      "epoch": 0.27042148452070125,
      "grad_norm": 0.41213661432266235,
      "learning_rate": 3.6480997969248625e-06,
      "loss": 0.741,
      "step": 6525
    },
    {
      "epoch": 0.2704629284263749,
      "grad_norm": 0.435043603181839,
      "learning_rate": 3.647892577396494e-06,
      "loss": 0.6699,
      "step": 6526
    },
    {
      "epoch": 0.27050437233204855,
      "grad_norm": 0.3998495638370514,
      "learning_rate": 3.647685357868126e-06,
      "loss": 0.7452,
      "step": 6527
    },
    {
      "epoch": 0.2705458162377222,
      "grad_norm": 0.3891395032405853,
      "learning_rate": 3.647478138339757e-06,
      "loss": 0.696,
      "step": 6528
    },
    {
      "epoch": 0.2705872601433959,
      "grad_norm": 0.40130409598350525,
      "learning_rate": 3.6472709188113893e-06,
      "loss": 0.6704,
      "step": 6529
    },
    {
      "epoch": 0.2706287040490696,
      "grad_norm": 0.4047229588031769,
      "learning_rate": 3.6470636992830203e-06,
      "loss": 0.713,
      "step": 6530
    },
    {
      "epoch": 0.27067014795474326,
      "grad_norm": 0.43221205472946167,
      "learning_rate": 3.6468564797546525e-06,
      "loss": 0.7435,
      "step": 6531
    },
    {
      "epoch": 0.27071159186041693,
      "grad_norm": 0.4334634244441986,
      "learning_rate": 3.6466492602262843e-06,
      "loss": 0.7073,
      "step": 6532
    },
    {
      "epoch": 0.2707530357660906,
      "grad_norm": 0.41183388233184814,
      "learning_rate": 3.6464420406979157e-06,
      "loss": 0.6488,
      "step": 6533
    },
    {
      "epoch": 0.2707944796717643,
      "grad_norm": 0.42317378520965576,
      "learning_rate": 3.6462348211695475e-06,
      "loss": 0.696,
      "step": 6534
    },
    {
      "epoch": 0.27083592357743796,
      "grad_norm": 0.47166845202445984,
      "learning_rate": 3.646027601641179e-06,
      "loss": 0.7548,
      "step": 6535
    },
    {
      "epoch": 0.2708773674831116,
      "grad_norm": 0.40081077814102173,
      "learning_rate": 3.6458203821128107e-06,
      "loss": 0.7388,
      "step": 6536
    },
    {
      "epoch": 0.27091881138878526,
      "grad_norm": 0.4226260185241699,
      "learning_rate": 3.645613162584442e-06,
      "loss": 0.714,
      "step": 6537
    },
    {
      "epoch": 0.27096025529445894,
      "grad_norm": 0.4098697304725647,
      "learning_rate": 3.645405943056074e-06,
      "loss": 0.6904,
      "step": 6538
    },
    {
      "epoch": 0.2710016992001326,
      "grad_norm": 0.401472806930542,
      "learning_rate": 3.6451987235277053e-06,
      "loss": 0.7168,
      "step": 6539
    },
    {
      "epoch": 0.2710431431058063,
      "grad_norm": 0.4249964654445648,
      "learning_rate": 3.644991503999337e-06,
      "loss": 0.6854,
      "step": 6540
    },
    {
      "epoch": 0.27108458701147997,
      "grad_norm": 0.41084668040275574,
      "learning_rate": 3.644784284470969e-06,
      "loss": 0.6895,
      "step": 6541
    },
    {
      "epoch": 0.27112603091715365,
      "grad_norm": 0.42162275314331055,
      "learning_rate": 3.6445770649426003e-06,
      "loss": 0.7118,
      "step": 6542
    },
    {
      "epoch": 0.2711674748228273,
      "grad_norm": 0.40232178568840027,
      "learning_rate": 3.644369845414232e-06,
      "loss": 0.7129,
      "step": 6543
    },
    {
      "epoch": 0.27120891872850095,
      "grad_norm": 0.42122000455856323,
      "learning_rate": 3.6441626258858635e-06,
      "loss": 0.6743,
      "step": 6544
    },
    {
      "epoch": 0.2712503626341746,
      "grad_norm": 0.42902329564094543,
      "learning_rate": 3.6439554063574957e-06,
      "loss": 0.7581,
      "step": 6545
    },
    {
      "epoch": 0.2712918065398483,
      "grad_norm": 0.4251077473163605,
      "learning_rate": 3.6437481868291267e-06,
      "loss": 0.6899,
      "step": 6546
    },
    {
      "epoch": 0.271333250445522,
      "grad_norm": 0.44206947088241577,
      "learning_rate": 3.643540967300759e-06,
      "loss": 0.7224,
      "step": 6547
    },
    {
      "epoch": 0.27137469435119566,
      "grad_norm": 0.409263014793396,
      "learning_rate": 3.6433337477723903e-06,
      "loss": 0.7388,
      "step": 6548
    },
    {
      "epoch": 0.27141613825686933,
      "grad_norm": 0.40656447410583496,
      "learning_rate": 3.643126528244022e-06,
      "loss": 0.6951,
      "step": 6549
    },
    {
      "epoch": 0.271457582162543,
      "grad_norm": 0.4202533960342407,
      "learning_rate": 3.642919308715654e-06,
      "loss": 0.7252,
      "step": 6550
    },
    {
      "epoch": 0.2714990260682167,
      "grad_norm": 0.39591488242149353,
      "learning_rate": 3.6427120891872853e-06,
      "loss": 0.7678,
      "step": 6551
    },
    {
      "epoch": 0.27154046997389036,
      "grad_norm": 0.4272144138813019,
      "learning_rate": 3.642504869658917e-06,
      "loss": 0.7366,
      "step": 6552
    },
    {
      "epoch": 0.271581913879564,
      "grad_norm": 0.38607990741729736,
      "learning_rate": 3.6422976501305485e-06,
      "loss": 0.6488,
      "step": 6553
    },
    {
      "epoch": 0.27162335778523766,
      "grad_norm": 0.43631815910339355,
      "learning_rate": 3.6420904306021803e-06,
      "loss": 0.6951,
      "step": 6554
    },
    {
      "epoch": 0.27166480169091134,
      "grad_norm": 0.4450920522212982,
      "learning_rate": 3.6418832110738117e-06,
      "loss": 0.6943,
      "step": 6555
    },
    {
      "epoch": 0.271706245596585,
      "grad_norm": 0.44615814089775085,
      "learning_rate": 3.6416759915454435e-06,
      "loss": 0.7791,
      "step": 6556
    },
    {
      "epoch": 0.2717476895022587,
      "grad_norm": 0.4156721532344818,
      "learning_rate": 3.641468772017075e-06,
      "loss": 0.7034,
      "step": 6557
    },
    {
      "epoch": 0.27178913340793237,
      "grad_norm": 0.41198232769966125,
      "learning_rate": 3.6412615524887067e-06,
      "loss": 0.7192,
      "step": 6558
    },
    {
      "epoch": 0.27183057731360605,
      "grad_norm": 0.412671834230423,
      "learning_rate": 3.6410543329603385e-06,
      "loss": 0.6755,
      "step": 6559
    },
    {
      "epoch": 0.2718720212192797,
      "grad_norm": 0.4309196174144745,
      "learning_rate": 3.64084711343197e-06,
      "loss": 0.6851,
      "step": 6560
    },
    {
      "epoch": 0.2719134651249534,
      "grad_norm": 0.3754064440727234,
      "learning_rate": 3.640639893903602e-06,
      "loss": 0.6581,
      "step": 6561
    },
    {
      "epoch": 0.271954909030627,
      "grad_norm": 0.40977543592453003,
      "learning_rate": 3.640432674375233e-06,
      "loss": 0.6667,
      "step": 6562
    },
    {
      "epoch": 0.2719963529363007,
      "grad_norm": 0.43328800797462463,
      "learning_rate": 3.6402254548468653e-06,
      "loss": 0.6909,
      "step": 6563
    },
    {
      "epoch": 0.2720377968419744,
      "grad_norm": 0.4237446188926697,
      "learning_rate": 3.6400182353184963e-06,
      "loss": 0.7322,
      "step": 6564
    },
    {
      "epoch": 0.27207924074764805,
      "grad_norm": 0.3954063057899475,
      "learning_rate": 3.6398110157901285e-06,
      "loss": 0.6956,
      "step": 6565
    },
    {
      "epoch": 0.27212068465332173,
      "grad_norm": 0.4595375657081604,
      "learning_rate": 3.63960379626176e-06,
      "loss": 0.6622,
      "step": 6566
    },
    {
      "epoch": 0.2721621285589954,
      "grad_norm": 0.4188733696937561,
      "learning_rate": 3.6393965767333917e-06,
      "loss": 0.6924,
      "step": 6567
    },
    {
      "epoch": 0.2722035724646691,
      "grad_norm": 0.40549731254577637,
      "learning_rate": 3.6391893572050235e-06,
      "loss": 0.6572,
      "step": 6568
    },
    {
      "epoch": 0.27224501637034276,
      "grad_norm": 0.4261831045150757,
      "learning_rate": 3.638982137676655e-06,
      "loss": 0.7046,
      "step": 6569
    },
    {
      "epoch": 0.2722864602760164,
      "grad_norm": 0.4417162239551544,
      "learning_rate": 3.6387749181482867e-06,
      "loss": 0.688,
      "step": 6570
    },
    {
      "epoch": 0.27232790418169006,
      "grad_norm": 0.4300711452960968,
      "learning_rate": 3.638567698619918e-06,
      "loss": 0.7188,
      "step": 6571
    },
    {
      "epoch": 0.27236934808736374,
      "grad_norm": 0.40618598461151123,
      "learning_rate": 3.63836047909155e-06,
      "loss": 0.718,
      "step": 6572
    },
    {
      "epoch": 0.2724107919930374,
      "grad_norm": 0.41585955023765564,
      "learning_rate": 3.6381532595631813e-06,
      "loss": 0.7012,
      "step": 6573
    },
    {
      "epoch": 0.2724522358987111,
      "grad_norm": 0.39511069655418396,
      "learning_rate": 3.637946040034813e-06,
      "loss": 0.7192,
      "step": 6574
    },
    {
      "epoch": 0.27249367980438477,
      "grad_norm": 0.4753815829753876,
      "learning_rate": 3.6377388205064445e-06,
      "loss": 0.718,
      "step": 6575
    },
    {
      "epoch": 0.27253512371005845,
      "grad_norm": 0.3823046088218689,
      "learning_rate": 3.6375316009780763e-06,
      "loss": 0.6702,
      "step": 6576
    },
    {
      "epoch": 0.2725765676157321,
      "grad_norm": 0.4095743000507355,
      "learning_rate": 3.637324381449708e-06,
      "loss": 0.7222,
      "step": 6577
    },
    {
      "epoch": 0.2726180115214058,
      "grad_norm": 0.4372406303882599,
      "learning_rate": 3.6371171619213395e-06,
      "loss": 0.8042,
      "step": 6578
    },
    {
      "epoch": 0.2726594554270794,
      "grad_norm": 0.4187566339969635,
      "learning_rate": 3.6369099423929718e-06,
      "loss": 0.7112,
      "step": 6579
    },
    {
      "epoch": 0.2727008993327531,
      "grad_norm": 0.433600515127182,
      "learning_rate": 3.6367027228646027e-06,
      "loss": 0.7097,
      "step": 6580
    },
    {
      "epoch": 0.2727423432384268,
      "grad_norm": 0.4430500864982605,
      "learning_rate": 3.636495503336235e-06,
      "loss": 0.6813,
      "step": 6581
    },
    {
      "epoch": 0.27278378714410045,
      "grad_norm": 0.42461520433425903,
      "learning_rate": 3.6362882838078663e-06,
      "loss": 0.7238,
      "step": 6582
    },
    {
      "epoch": 0.27282523104977413,
      "grad_norm": 0.3987359404563904,
      "learning_rate": 3.636081064279498e-06,
      "loss": 0.6707,
      "step": 6583
    },
    {
      "epoch": 0.2728666749554478,
      "grad_norm": 0.4060910940170288,
      "learning_rate": 3.6358738447511295e-06,
      "loss": 0.7251,
      "step": 6584
    },
    {
      "epoch": 0.2729081188611215,
      "grad_norm": 0.415551096200943,
      "learning_rate": 3.6356666252227613e-06,
      "loss": 0.7188,
      "step": 6585
    },
    {
      "epoch": 0.27294956276679516,
      "grad_norm": 0.3749302625656128,
      "learning_rate": 3.635459405694393e-06,
      "loss": 0.6985,
      "step": 6586
    },
    {
      "epoch": 0.27299100667246884,
      "grad_norm": 0.4277219772338867,
      "learning_rate": 3.6352521861660245e-06,
      "loss": 0.7454,
      "step": 6587
    },
    {
      "epoch": 0.27303245057814246,
      "grad_norm": 0.3968074917793274,
      "learning_rate": 3.6350449666376563e-06,
      "loss": 0.7117,
      "step": 6588
    },
    {
      "epoch": 0.27307389448381614,
      "grad_norm": 0.43151921033859253,
      "learning_rate": 3.6348377471092877e-06,
      "loss": 0.7217,
      "step": 6589
    },
    {
      "epoch": 0.2731153383894898,
      "grad_norm": 0.4035494327545166,
      "learning_rate": 3.6346305275809195e-06,
      "loss": 0.6826,
      "step": 6590
    },
    {
      "epoch": 0.2731567822951635,
      "grad_norm": 0.4098740518093109,
      "learning_rate": 3.634423308052551e-06,
      "loss": 0.6539,
      "step": 6591
    },
    {
      "epoch": 0.27319822620083717,
      "grad_norm": 0.39176657795906067,
      "learning_rate": 3.6342160885241827e-06,
      "loss": 0.6987,
      "step": 6592
    },
    {
      "epoch": 0.27323967010651085,
      "grad_norm": 0.46942436695098877,
      "learning_rate": 3.6340088689958145e-06,
      "loss": 0.7546,
      "step": 6593
    },
    {
      "epoch": 0.2732811140121845,
      "grad_norm": 0.43935540318489075,
      "learning_rate": 3.633801649467446e-06,
      "loss": 0.7399,
      "step": 6594
    },
    {
      "epoch": 0.2733225579178582,
      "grad_norm": 0.4395289123058319,
      "learning_rate": 3.633594429939078e-06,
      "loss": 0.7263,
      "step": 6595
    },
    {
      "epoch": 0.2733640018235319,
      "grad_norm": 0.4925175607204437,
      "learning_rate": 3.633387210410709e-06,
      "loss": 0.7742,
      "step": 6596
    },
    {
      "epoch": 0.2734054457292055,
      "grad_norm": 0.433120459318161,
      "learning_rate": 3.6331799908823414e-06,
      "loss": 0.7405,
      "step": 6597
    },
    {
      "epoch": 0.2734468896348792,
      "grad_norm": 0.42321160435676575,
      "learning_rate": 3.6329727713539727e-06,
      "loss": 0.7244,
      "step": 6598
    },
    {
      "epoch": 0.27348833354055285,
      "grad_norm": 0.41757932305336,
      "learning_rate": 3.6327655518256045e-06,
      "loss": 0.692,
      "step": 6599
    },
    {
      "epoch": 0.27352977744622653,
      "grad_norm": 0.41712823510169983,
      "learning_rate": 3.632558332297236e-06,
      "loss": 0.6827,
      "step": 6600
    },
    {
      "epoch": 0.2735712213519002,
      "grad_norm": 0.4401552677154541,
      "learning_rate": 3.6323511127688677e-06,
      "loss": 0.7451,
      "step": 6601
    },
    {
      "epoch": 0.2736126652575739,
      "grad_norm": 0.37812578678131104,
      "learning_rate": 3.6321438932404996e-06,
      "loss": 0.6548,
      "step": 6602
    },
    {
      "epoch": 0.27365410916324756,
      "grad_norm": 0.40258845686912537,
      "learning_rate": 3.631936673712131e-06,
      "loss": 0.7045,
      "step": 6603
    },
    {
      "epoch": 0.27369555306892124,
      "grad_norm": 0.4452533721923828,
      "learning_rate": 3.6317294541837627e-06,
      "loss": 0.7125,
      "step": 6604
    },
    {
      "epoch": 0.27373699697459486,
      "grad_norm": 0.4513268768787384,
      "learning_rate": 3.631522234655394e-06,
      "loss": 0.7529,
      "step": 6605
    },
    {
      "epoch": 0.27377844088026854,
      "grad_norm": 0.4171219766139984,
      "learning_rate": 3.631315015127026e-06,
      "loss": 0.7478,
      "step": 6606
    },
    {
      "epoch": 0.2738198847859422,
      "grad_norm": 0.4141135811805725,
      "learning_rate": 3.6311077955986573e-06,
      "loss": 0.6663,
      "step": 6607
    },
    {
      "epoch": 0.2738613286916159,
      "grad_norm": 0.39465829730033875,
      "learning_rate": 3.630900576070289e-06,
      "loss": 0.6835,
      "step": 6608
    },
    {
      "epoch": 0.27390277259728957,
      "grad_norm": 0.42372745275497437,
      "learning_rate": 3.6306933565419205e-06,
      "loss": 0.6804,
      "step": 6609
    },
    {
      "epoch": 0.27394421650296324,
      "grad_norm": 0.40289783477783203,
      "learning_rate": 3.6304861370135523e-06,
      "loss": 0.6603,
      "step": 6610
    },
    {
      "epoch": 0.2739856604086369,
      "grad_norm": 0.43755820393562317,
      "learning_rate": 3.630278917485184e-06,
      "loss": 0.7361,
      "step": 6611
    },
    {
      "epoch": 0.2740271043143106,
      "grad_norm": 0.4324570894241333,
      "learning_rate": 3.6300716979568155e-06,
      "loss": 0.7397,
      "step": 6612
    },
    {
      "epoch": 0.2740685482199843,
      "grad_norm": 0.4048880934715271,
      "learning_rate": 3.6298644784284478e-06,
      "loss": 0.7668,
      "step": 6613
    },
    {
      "epoch": 0.2741099921256579,
      "grad_norm": 0.41328638792037964,
      "learning_rate": 3.6296572589000787e-06,
      "loss": 0.645,
      "step": 6614
    },
    {
      "epoch": 0.2741514360313316,
      "grad_norm": 0.4019462466239929,
      "learning_rate": 3.629450039371711e-06,
      "loss": 0.7576,
      "step": 6615
    },
    {
      "epoch": 0.27419287993700525,
      "grad_norm": 0.39426904916763306,
      "learning_rate": 3.6292428198433423e-06,
      "loss": 0.6985,
      "step": 6616
    },
    {
      "epoch": 0.27423432384267893,
      "grad_norm": 0.40611618757247925,
      "learning_rate": 3.629035600314974e-06,
      "loss": 0.7344,
      "step": 6617
    },
    {
      "epoch": 0.2742757677483526,
      "grad_norm": 0.4193512499332428,
      "learning_rate": 3.6288283807866055e-06,
      "loss": 0.7065,
      "step": 6618
    },
    {
      "epoch": 0.2743172116540263,
      "grad_norm": 0.3867458403110504,
      "learning_rate": 3.6286211612582373e-06,
      "loss": 0.6637,
      "step": 6619
    },
    {
      "epoch": 0.27435865555969996,
      "grad_norm": 0.441089004278183,
      "learning_rate": 3.628413941729869e-06,
      "loss": 0.7155,
      "step": 6620
    },
    {
      "epoch": 0.27440009946537364,
      "grad_norm": 0.417917400598526,
      "learning_rate": 3.6282067222015005e-06,
      "loss": 0.6486,
      "step": 6621
    },
    {
      "epoch": 0.2744415433710473,
      "grad_norm": 0.45735424757003784,
      "learning_rate": 3.6279995026731323e-06,
      "loss": 0.7744,
      "step": 6622
    },
    {
      "epoch": 0.27448298727672094,
      "grad_norm": 0.3985811173915863,
      "learning_rate": 3.6277922831447637e-06,
      "loss": 0.7168,
      "step": 6623
    },
    {
      "epoch": 0.2745244311823946,
      "grad_norm": 0.4091930389404297,
      "learning_rate": 3.6275850636163955e-06,
      "loss": 0.7261,
      "step": 6624
    },
    {
      "epoch": 0.2745658750880683,
      "grad_norm": 0.39693740010261536,
      "learning_rate": 3.627377844088027e-06,
      "loss": 0.7568,
      "step": 6625
    },
    {
      "epoch": 0.27460731899374197,
      "grad_norm": 0.38955074548721313,
      "learning_rate": 3.6271706245596587e-06,
      "loss": 0.6794,
      "step": 6626
    },
    {
      "epoch": 0.27464876289941564,
      "grad_norm": 0.4055379331111908,
      "learning_rate": 3.62696340503129e-06,
      "loss": 0.7133,
      "step": 6627
    },
    {
      "epoch": 0.2746902068050893,
      "grad_norm": 0.47656673192977905,
      "learning_rate": 3.626756185502922e-06,
      "loss": 0.7742,
      "step": 6628
    },
    {
      "epoch": 0.274731650710763,
      "grad_norm": 0.3985002040863037,
      "learning_rate": 3.626548965974554e-06,
      "loss": 0.7214,
      "step": 6629
    },
    {
      "epoch": 0.2747730946164367,
      "grad_norm": 0.3947330415248871,
      "learning_rate": 3.626341746446185e-06,
      "loss": 0.7168,
      "step": 6630
    },
    {
      "epoch": 0.2748145385221103,
      "grad_norm": 0.4155828356742859,
      "learning_rate": 3.6261345269178174e-06,
      "loss": 0.6669,
      "step": 6631
    },
    {
      "epoch": 0.274855982427784,
      "grad_norm": 0.4428500533103943,
      "learning_rate": 3.6259273073894487e-06,
      "loss": 0.7148,
      "step": 6632
    },
    {
      "epoch": 0.27489742633345765,
      "grad_norm": 0.3943454921245575,
      "learning_rate": 3.6257200878610806e-06,
      "loss": 0.6746,
      "step": 6633
    },
    {
      "epoch": 0.2749388702391313,
      "grad_norm": 0.454054594039917,
      "learning_rate": 3.625512868332712e-06,
      "loss": 0.7258,
      "step": 6634
    },
    {
      "epoch": 0.274980314144805,
      "grad_norm": 0.41878741979599,
      "learning_rate": 3.6253056488043437e-06,
      "loss": 0.7151,
      "step": 6635
    },
    {
      "epoch": 0.2750217580504787,
      "grad_norm": 0.4162358045578003,
      "learning_rate": 3.625098429275975e-06,
      "loss": 0.7238,
      "step": 6636
    },
    {
      "epoch": 0.27506320195615236,
      "grad_norm": 0.46346625685691833,
      "learning_rate": 3.624891209747607e-06,
      "loss": 0.7388,
      "step": 6637
    },
    {
      "epoch": 0.27510464586182604,
      "grad_norm": 0.4319714307785034,
      "learning_rate": 3.6246839902192388e-06,
      "loss": 0.7329,
      "step": 6638
    },
    {
      "epoch": 0.2751460897674997,
      "grad_norm": 0.41148892045021057,
      "learning_rate": 3.62447677069087e-06,
      "loss": 0.7036,
      "step": 6639
    },
    {
      "epoch": 0.27518753367317333,
      "grad_norm": 0.4208109676837921,
      "learning_rate": 3.624269551162502e-06,
      "loss": 0.6866,
      "step": 6640
    },
    {
      "epoch": 0.275228977578847,
      "grad_norm": 0.4079609513282776,
      "learning_rate": 3.6240623316341333e-06,
      "loss": 0.6829,
      "step": 6641
    },
    {
      "epoch": 0.2752704214845207,
      "grad_norm": 0.39932799339294434,
      "learning_rate": 3.623855112105765e-06,
      "loss": 0.6521,
      "step": 6642
    },
    {
      "epoch": 0.27531186539019437,
      "grad_norm": 0.4090936481952667,
      "learning_rate": 3.6236478925773965e-06,
      "loss": 0.7455,
      "step": 6643
    },
    {
      "epoch": 0.27535330929586804,
      "grad_norm": 0.3901226222515106,
      "learning_rate": 3.6234406730490283e-06,
      "loss": 0.7371,
      "step": 6644
    },
    {
      "epoch": 0.2753947532015417,
      "grad_norm": 0.4074028432369232,
      "learning_rate": 3.6232334535206606e-06,
      "loss": 0.7798,
      "step": 6645
    },
    {
      "epoch": 0.2754361971072154,
      "grad_norm": 0.45360350608825684,
      "learning_rate": 3.6230262339922915e-06,
      "loss": 0.7544,
      "step": 6646
    },
    {
      "epoch": 0.2754776410128891,
      "grad_norm": 0.45333215594291687,
      "learning_rate": 3.6228190144639238e-06,
      "loss": 0.7432,
      "step": 6647
    },
    {
      "epoch": 0.27551908491856275,
      "grad_norm": 0.42594078183174133,
      "learning_rate": 3.6226117949355547e-06,
      "loss": 0.66,
      "step": 6648
    },
    {
      "epoch": 0.2755605288242364,
      "grad_norm": 0.41911768913269043,
      "learning_rate": 3.622404575407187e-06,
      "loss": 0.7388,
      "step": 6649
    },
    {
      "epoch": 0.27560197272991005,
      "grad_norm": 0.4502609670162201,
      "learning_rate": 3.6221973558788183e-06,
      "loss": 0.6809,
      "step": 6650
    },
    {
      "epoch": 0.2756434166355837,
      "grad_norm": 0.4637426733970642,
      "learning_rate": 3.62199013635045e-06,
      "loss": 0.6959,
      "step": 6651
    },
    {
      "epoch": 0.2756848605412574,
      "grad_norm": 0.4150581359863281,
      "learning_rate": 3.6217829168220815e-06,
      "loss": 0.7444,
      "step": 6652
    },
    {
      "epoch": 0.2757263044469311,
      "grad_norm": 0.432466059923172,
      "learning_rate": 3.6215756972937133e-06,
      "loss": 0.7079,
      "step": 6653
    },
    {
      "epoch": 0.27576774835260476,
      "grad_norm": 0.38473695516586304,
      "learning_rate": 3.621368477765345e-06,
      "loss": 0.6846,
      "step": 6654
    },
    {
      "epoch": 0.27580919225827843,
      "grad_norm": 0.4081288278102875,
      "learning_rate": 3.6211612582369765e-06,
      "loss": 0.6968,
      "step": 6655
    },
    {
      "epoch": 0.2758506361639521,
      "grad_norm": 0.43915748596191406,
      "learning_rate": 3.6209540387086084e-06,
      "loss": 0.7622,
      "step": 6656
    },
    {
      "epoch": 0.2758920800696258,
      "grad_norm": 0.42115843296051025,
      "learning_rate": 3.6207468191802397e-06,
      "loss": 0.749,
      "step": 6657
    },
    {
      "epoch": 0.2759335239752994,
      "grad_norm": 0.4176490008831024,
      "learning_rate": 3.6205395996518715e-06,
      "loss": 0.7004,
      "step": 6658
    },
    {
      "epoch": 0.2759749678809731,
      "grad_norm": 0.4029833972454071,
      "learning_rate": 3.620332380123503e-06,
      "loss": 0.636,
      "step": 6659
    },
    {
      "epoch": 0.27601641178664676,
      "grad_norm": 0.42111310362815857,
      "learning_rate": 3.6201251605951347e-06,
      "loss": 0.7336,
      "step": 6660
    },
    {
      "epoch": 0.27605785569232044,
      "grad_norm": 0.432941734790802,
      "learning_rate": 3.619917941066766e-06,
      "loss": 0.7196,
      "step": 6661
    },
    {
      "epoch": 0.2760992995979941,
      "grad_norm": 0.4619463086128235,
      "learning_rate": 3.619710721538398e-06,
      "loss": 0.7312,
      "step": 6662
    },
    {
      "epoch": 0.2761407435036678,
      "grad_norm": 0.41966474056243896,
      "learning_rate": 3.61950350201003e-06,
      "loss": 0.7485,
      "step": 6663
    },
    {
      "epoch": 0.2761821874093415,
      "grad_norm": 0.4593070149421692,
      "learning_rate": 3.619296282481661e-06,
      "loss": 0.707,
      "step": 6664
    },
    {
      "epoch": 0.27622363131501515,
      "grad_norm": 0.4630683660507202,
      "learning_rate": 3.6190890629532934e-06,
      "loss": 0.7207,
      "step": 6665
    },
    {
      "epoch": 0.27626507522068877,
      "grad_norm": 0.5042412281036377,
      "learning_rate": 3.6188818434249248e-06,
      "loss": 0.7375,
      "step": 6666
    },
    {
      "epoch": 0.27630651912636245,
      "grad_norm": 0.39843347668647766,
      "learning_rate": 3.6186746238965566e-06,
      "loss": 0.6777,
      "step": 6667
    },
    {
      "epoch": 0.2763479630320361,
      "grad_norm": 0.4366994798183441,
      "learning_rate": 3.618467404368188e-06,
      "loss": 0.7119,
      "step": 6668
    },
    {
      "epoch": 0.2763894069377098,
      "grad_norm": 0.4382764399051666,
      "learning_rate": 3.6182601848398198e-06,
      "loss": 0.7378,
      "step": 6669
    },
    {
      "epoch": 0.2764308508433835,
      "grad_norm": 0.4502507746219635,
      "learning_rate": 3.618052965311451e-06,
      "loss": 0.7451,
      "step": 6670
    },
    {
      "epoch": 0.27647229474905716,
      "grad_norm": 0.42975178360939026,
      "learning_rate": 3.617845745783083e-06,
      "loss": 0.7288,
      "step": 6671
    },
    {
      "epoch": 0.27651373865473083,
      "grad_norm": 0.4352637827396393,
      "learning_rate": 3.6176385262547148e-06,
      "loss": 0.7229,
      "step": 6672
    },
    {
      "epoch": 0.2765551825604045,
      "grad_norm": 0.4746386706829071,
      "learning_rate": 3.617431306726346e-06,
      "loss": 0.7759,
      "step": 6673
    },
    {
      "epoch": 0.2765966264660782,
      "grad_norm": 0.4371624290943146,
      "learning_rate": 3.617224087197978e-06,
      "loss": 0.7455,
      "step": 6674
    },
    {
      "epoch": 0.2766380703717518,
      "grad_norm": 0.4262458086013794,
      "learning_rate": 3.6170168676696093e-06,
      "loss": 0.6592,
      "step": 6675
    },
    {
      "epoch": 0.2766795142774255,
      "grad_norm": 0.44028881192207336,
      "learning_rate": 3.616809648141241e-06,
      "loss": 0.7246,
      "step": 6676
    },
    {
      "epoch": 0.27672095818309916,
      "grad_norm": 0.4979361891746521,
      "learning_rate": 3.6166024286128725e-06,
      "loss": 0.7272,
      "step": 6677
    },
    {
      "epoch": 0.27676240208877284,
      "grad_norm": 0.4421960711479187,
      "learning_rate": 3.6163952090845043e-06,
      "loss": 0.7173,
      "step": 6678
    },
    {
      "epoch": 0.2768038459944465,
      "grad_norm": 0.4271281063556671,
      "learning_rate": 3.6161879895561357e-06,
      "loss": 0.7402,
      "step": 6679
    },
    {
      "epoch": 0.2768452899001202,
      "grad_norm": 0.4092029631137848,
      "learning_rate": 3.6159807700277675e-06,
      "loss": 0.6873,
      "step": 6680
    },
    {
      "epoch": 0.27688673380579387,
      "grad_norm": 0.3896249830722809,
      "learning_rate": 3.6157735504993998e-06,
      "loss": 0.6714,
      "step": 6681
    },
    {
      "epoch": 0.27692817771146755,
      "grad_norm": 0.429233580827713,
      "learning_rate": 3.6155663309710307e-06,
      "loss": 0.7041,
      "step": 6682
    },
    {
      "epoch": 0.2769696216171412,
      "grad_norm": 0.41778188943862915,
      "learning_rate": 3.615359111442663e-06,
      "loss": 0.713,
      "step": 6683
    },
    {
      "epoch": 0.27701106552281485,
      "grad_norm": 0.4221159517765045,
      "learning_rate": 3.6151518919142944e-06,
      "loss": 0.686,
      "step": 6684
    },
    {
      "epoch": 0.2770525094284885,
      "grad_norm": 0.41951772570610046,
      "learning_rate": 3.614944672385926e-06,
      "loss": 0.7151,
      "step": 6685
    },
    {
      "epoch": 0.2770939533341622,
      "grad_norm": 0.40577229857444763,
      "learning_rate": 3.6147374528575575e-06,
      "loss": 0.72,
      "step": 6686
    },
    {
      "epoch": 0.2771353972398359,
      "grad_norm": 0.41571900248527527,
      "learning_rate": 3.6145302333291894e-06,
      "loss": 0.7344,
      "step": 6687
    },
    {
      "epoch": 0.27717684114550956,
      "grad_norm": 0.4272485673427582,
      "learning_rate": 3.6143230138008207e-06,
      "loss": 0.733,
      "step": 6688
    },
    {
      "epoch": 0.27721828505118323,
      "grad_norm": 0.41691747307777405,
      "learning_rate": 3.6141157942724525e-06,
      "loss": 0.7324,
      "step": 6689
    },
    {
      "epoch": 0.2772597289568569,
      "grad_norm": 0.42767179012298584,
      "learning_rate": 3.6139085747440844e-06,
      "loss": 0.72,
      "step": 6690
    },
    {
      "epoch": 0.2773011728625306,
      "grad_norm": 0.40414050221443176,
      "learning_rate": 3.6137013552157157e-06,
      "loss": 0.7092,
      "step": 6691
    },
    {
      "epoch": 0.2773426167682042,
      "grad_norm": 0.4483755826950073,
      "learning_rate": 3.6134941356873476e-06,
      "loss": 0.7424,
      "step": 6692
    },
    {
      "epoch": 0.2773840606738779,
      "grad_norm": 0.38455766439437866,
      "learning_rate": 3.613286916158979e-06,
      "loss": 0.7007,
      "step": 6693
    },
    {
      "epoch": 0.27742550457955156,
      "grad_norm": 0.39429235458374023,
      "learning_rate": 3.6130796966306107e-06,
      "loss": 0.6873,
      "step": 6694
    },
    {
      "epoch": 0.27746694848522524,
      "grad_norm": 0.46908038854599,
      "learning_rate": 3.612872477102242e-06,
      "loss": 0.6982,
      "step": 6695
    },
    {
      "epoch": 0.2775083923908989,
      "grad_norm": 0.3853103816509247,
      "learning_rate": 3.612665257573874e-06,
      "loss": 0.6516,
      "step": 6696
    },
    {
      "epoch": 0.2775498362965726,
      "grad_norm": 0.3900322914123535,
      "learning_rate": 3.6124580380455053e-06,
      "loss": 0.6677,
      "step": 6697
    },
    {
      "epoch": 0.27759128020224627,
      "grad_norm": 0.41675224900245667,
      "learning_rate": 3.612250818517137e-06,
      "loss": 0.719,
      "step": 6698
    },
    {
      "epoch": 0.27763272410791995,
      "grad_norm": 0.4015946388244629,
      "learning_rate": 3.6120435989887694e-06,
      "loss": 0.6934,
      "step": 6699
    },
    {
      "epoch": 0.2776741680135936,
      "grad_norm": 0.40435823798179626,
      "learning_rate": 3.6118363794604008e-06,
      "loss": 0.6809,
      "step": 6700
    },
    {
      "epoch": 0.27771561191926725,
      "grad_norm": 0.3971484303474426,
      "learning_rate": 3.6116291599320326e-06,
      "loss": 0.7065,
      "step": 6701
    },
    {
      "epoch": 0.2777570558249409,
      "grad_norm": 0.4255174994468689,
      "learning_rate": 3.611421940403664e-06,
      "loss": 0.7009,
      "step": 6702
    },
    {
      "epoch": 0.2777984997306146,
      "grad_norm": 0.41539624333381653,
      "learning_rate": 3.6112147208752958e-06,
      "loss": 0.7688,
      "step": 6703
    },
    {
      "epoch": 0.2778399436362883,
      "grad_norm": 0.4160275161266327,
      "learning_rate": 3.611007501346927e-06,
      "loss": 0.718,
      "step": 6704
    },
    {
      "epoch": 0.27788138754196196,
      "grad_norm": 0.4690508246421814,
      "learning_rate": 3.610800281818559e-06,
      "loss": 0.7031,
      "step": 6705
    },
    {
      "epoch": 0.27792283144763563,
      "grad_norm": 0.40456023812294006,
      "learning_rate": 3.6105930622901908e-06,
      "loss": 0.6951,
      "step": 6706
    },
    {
      "epoch": 0.2779642753533093,
      "grad_norm": 0.39568477869033813,
      "learning_rate": 3.610385842761822e-06,
      "loss": 0.7195,
      "step": 6707
    },
    {
      "epoch": 0.278005719258983,
      "grad_norm": 0.41585689783096313,
      "learning_rate": 3.610178623233454e-06,
      "loss": 0.7043,
      "step": 6708
    },
    {
      "epoch": 0.27804716316465666,
      "grad_norm": 0.38726481795310974,
      "learning_rate": 3.6099714037050853e-06,
      "loss": 0.6836,
      "step": 6709
    },
    {
      "epoch": 0.2780886070703303,
      "grad_norm": 0.41844072937965393,
      "learning_rate": 3.609764184176717e-06,
      "loss": 0.7561,
      "step": 6710
    },
    {
      "epoch": 0.27813005097600396,
      "grad_norm": 0.41628044843673706,
      "learning_rate": 3.6095569646483485e-06,
      "loss": 0.6868,
      "step": 6711
    },
    {
      "epoch": 0.27817149488167764,
      "grad_norm": 0.408419132232666,
      "learning_rate": 3.6093497451199803e-06,
      "loss": 0.7351,
      "step": 6712
    },
    {
      "epoch": 0.2782129387873513,
      "grad_norm": 0.465655118227005,
      "learning_rate": 3.6091425255916117e-06,
      "loss": 0.7009,
      "step": 6713
    },
    {
      "epoch": 0.278254382693025,
      "grad_norm": 0.42494451999664307,
      "learning_rate": 3.6089353060632435e-06,
      "loss": 0.7006,
      "step": 6714
    },
    {
      "epoch": 0.27829582659869867,
      "grad_norm": 0.4616956114768982,
      "learning_rate": 3.6087280865348758e-06,
      "loss": 0.7642,
      "step": 6715
    },
    {
      "epoch": 0.27833727050437235,
      "grad_norm": 0.4089677631855011,
      "learning_rate": 3.6085208670065067e-06,
      "loss": 0.698,
      "step": 6716
    },
    {
      "epoch": 0.278378714410046,
      "grad_norm": 0.4243232011795044,
      "learning_rate": 3.608313647478139e-06,
      "loss": 0.7103,
      "step": 6717
    },
    {
      "epoch": 0.27842015831571965,
      "grad_norm": 0.39270323514938354,
      "learning_rate": 3.6081064279497704e-06,
      "loss": 0.677,
      "step": 6718
    },
    {
      "epoch": 0.2784616022213933,
      "grad_norm": 0.4386405646800995,
      "learning_rate": 3.607899208421402e-06,
      "loss": 0.693,
      "step": 6719
    },
    {
      "epoch": 0.278503046127067,
      "grad_norm": 0.4516928791999817,
      "learning_rate": 3.6076919888930336e-06,
      "loss": 0.7161,
      "step": 6720
    },
    {
      "epoch": 0.2785444900327407,
      "grad_norm": 0.43242600560188293,
      "learning_rate": 3.6074847693646654e-06,
      "loss": 0.7317,
      "step": 6721
    },
    {
      "epoch": 0.27858593393841435,
      "grad_norm": 0.4511552155017853,
      "learning_rate": 3.6072775498362967e-06,
      "loss": 0.7212,
      "step": 6722
    },
    {
      "epoch": 0.27862737784408803,
      "grad_norm": 0.4081852436065674,
      "learning_rate": 3.6070703303079286e-06,
      "loss": 0.6965,
      "step": 6723
    },
    {
      "epoch": 0.2786688217497617,
      "grad_norm": 0.41443973779678345,
      "learning_rate": 3.6068631107795604e-06,
      "loss": 0.7186,
      "step": 6724
    },
    {
      "epoch": 0.2787102656554354,
      "grad_norm": 0.39324405789375305,
      "learning_rate": 3.6066558912511918e-06,
      "loss": 0.663,
      "step": 6725
    },
    {
      "epoch": 0.27875170956110906,
      "grad_norm": 0.43067094683647156,
      "learning_rate": 3.6064486717228236e-06,
      "loss": 0.7019,
      "step": 6726
    },
    {
      "epoch": 0.2787931534667827,
      "grad_norm": 0.43750375509262085,
      "learning_rate": 3.606241452194455e-06,
      "loss": 0.7188,
      "step": 6727
    },
    {
      "epoch": 0.27883459737245636,
      "grad_norm": 0.38726162910461426,
      "learning_rate": 3.6060342326660868e-06,
      "loss": 0.6537,
      "step": 6728
    },
    {
      "epoch": 0.27887604127813004,
      "grad_norm": 0.3972233533859253,
      "learning_rate": 3.605827013137718e-06,
      "loss": 0.7028,
      "step": 6729
    },
    {
      "epoch": 0.2789174851838037,
      "grad_norm": 0.4139542579650879,
      "learning_rate": 3.60561979360935e-06,
      "loss": 0.7017,
      "step": 6730
    },
    {
      "epoch": 0.2789589290894774,
      "grad_norm": 0.4411020874977112,
      "learning_rate": 3.6054125740809813e-06,
      "loss": 0.6698,
      "step": 6731
    },
    {
      "epoch": 0.27900037299515107,
      "grad_norm": 0.40187785029411316,
      "learning_rate": 3.605205354552613e-06,
      "loss": 0.6788,
      "step": 6732
    },
    {
      "epoch": 0.27904181690082475,
      "grad_norm": 0.41578540205955505,
      "learning_rate": 3.6049981350242454e-06,
      "loss": 0.681,
      "step": 6733
    },
    {
      "epoch": 0.2790832608064984,
      "grad_norm": 0.41805529594421387,
      "learning_rate": 3.6047909154958768e-06,
      "loss": 0.7115,
      "step": 6734
    },
    {
      "epoch": 0.2791247047121721,
      "grad_norm": 0.41423499584198,
      "learning_rate": 3.6045836959675086e-06,
      "loss": 0.6995,
      "step": 6735
    },
    {
      "epoch": 0.2791661486178457,
      "grad_norm": 0.43814992904663086,
      "learning_rate": 3.60437647643914e-06,
      "loss": 0.7383,
      "step": 6736
    },
    {
      "epoch": 0.2792075925235194,
      "grad_norm": 0.4223230481147766,
      "learning_rate": 3.6041692569107718e-06,
      "loss": 0.7043,
      "step": 6737
    },
    {
      "epoch": 0.2792490364291931,
      "grad_norm": 0.47840362787246704,
      "learning_rate": 3.603962037382403e-06,
      "loss": 0.7167,
      "step": 6738
    },
    {
      "epoch": 0.27929048033486675,
      "grad_norm": 0.4244794249534607,
      "learning_rate": 3.603754817854035e-06,
      "loss": 0.646,
      "step": 6739
    },
    {
      "epoch": 0.27933192424054043,
      "grad_norm": 0.4094058871269226,
      "learning_rate": 3.6035475983256663e-06,
      "loss": 0.6912,
      "step": 6740
    },
    {
      "epoch": 0.2793733681462141,
      "grad_norm": 0.39633408188819885,
      "learning_rate": 3.603340378797298e-06,
      "loss": 0.7281,
      "step": 6741
    },
    {
      "epoch": 0.2794148120518878,
      "grad_norm": 0.4078458845615387,
      "learning_rate": 3.60313315926893e-06,
      "loss": 0.7327,
      "step": 6742
    },
    {
      "epoch": 0.27945625595756146,
      "grad_norm": 0.46389541029930115,
      "learning_rate": 3.6029259397405614e-06,
      "loss": 0.7939,
      "step": 6743
    },
    {
      "epoch": 0.27949769986323514,
      "grad_norm": 0.41833218932151794,
      "learning_rate": 3.602718720212193e-06,
      "loss": 0.739,
      "step": 6744
    },
    {
      "epoch": 0.27953914376890876,
      "grad_norm": 0.4221803545951843,
      "learning_rate": 3.6025115006838245e-06,
      "loss": 0.749,
      "step": 6745
    },
    {
      "epoch": 0.27958058767458244,
      "grad_norm": 0.3903099596500397,
      "learning_rate": 3.6023042811554564e-06,
      "loss": 0.7471,
      "step": 6746
    },
    {
      "epoch": 0.2796220315802561,
      "grad_norm": 0.4545293152332306,
      "learning_rate": 3.6020970616270877e-06,
      "loss": 0.6843,
      "step": 6747
    },
    {
      "epoch": 0.2796634754859298,
      "grad_norm": 0.4799537658691406,
      "learning_rate": 3.6018898420987196e-06,
      "loss": 0.7012,
      "step": 6748
    },
    {
      "epoch": 0.27970491939160347,
      "grad_norm": 0.4690294861793518,
      "learning_rate": 3.601682622570351e-06,
      "loss": 0.7593,
      "step": 6749
    },
    {
      "epoch": 0.27974636329727715,
      "grad_norm": 0.422396183013916,
      "learning_rate": 3.6014754030419827e-06,
      "loss": 0.7318,
      "step": 6750
    },
    {
      "epoch": 0.2797878072029508,
      "grad_norm": 0.4174599051475525,
      "learning_rate": 3.601268183513615e-06,
      "loss": 0.6791,
      "step": 6751
    },
    {
      "epoch": 0.2798292511086245,
      "grad_norm": 0.4698418974876404,
      "learning_rate": 3.6010609639852464e-06,
      "loss": 0.7712,
      "step": 6752
    },
    {
      "epoch": 0.2798706950142981,
      "grad_norm": 0.42749327421188354,
      "learning_rate": 3.600853744456878e-06,
      "loss": 0.7096,
      "step": 6753
    },
    {
      "epoch": 0.2799121389199718,
      "grad_norm": 0.43996813893318176,
      "learning_rate": 3.6006465249285096e-06,
      "loss": 0.663,
      "step": 6754
    },
    {
      "epoch": 0.2799535828256455,
      "grad_norm": 0.4503937363624573,
      "learning_rate": 3.6004393054001414e-06,
      "loss": 0.7441,
      "step": 6755
    },
    {
      "epoch": 0.27999502673131915,
      "grad_norm": 0.4242357909679413,
      "learning_rate": 3.6002320858717728e-06,
      "loss": 0.7095,
      "step": 6756
    },
    {
      "epoch": 0.28003647063699283,
      "grad_norm": 0.3889995217323303,
      "learning_rate": 3.6000248663434046e-06,
      "loss": 0.6693,
      "step": 6757
    },
    {
      "epoch": 0.2800779145426665,
      "grad_norm": 0.4084670841693878,
      "learning_rate": 3.599817646815036e-06,
      "loss": 0.6809,
      "step": 6758
    },
    {
      "epoch": 0.2801193584483402,
      "grad_norm": 0.42046937346458435,
      "learning_rate": 3.5996104272866678e-06,
      "loss": 0.7136,
      "step": 6759
    },
    {
      "epoch": 0.28016080235401386,
      "grad_norm": 0.41826221346855164,
      "learning_rate": 3.5994032077582996e-06,
      "loss": 0.6919,
      "step": 6760
    },
    {
      "epoch": 0.28020224625968754,
      "grad_norm": 0.46046873927116394,
      "learning_rate": 3.599195988229931e-06,
      "loss": 0.7378,
      "step": 6761
    },
    {
      "epoch": 0.28024369016536116,
      "grad_norm": 0.40140989422798157,
      "learning_rate": 3.5989887687015628e-06,
      "loss": 0.687,
      "step": 6762
    },
    {
      "epoch": 0.28028513407103484,
      "grad_norm": 0.48575130105018616,
      "learning_rate": 3.598781549173194e-06,
      "loss": 0.7227,
      "step": 6763
    },
    {
      "epoch": 0.2803265779767085,
      "grad_norm": 0.41963785886764526,
      "learning_rate": 3.598574329644826e-06,
      "loss": 0.7074,
      "step": 6764
    },
    {
      "epoch": 0.2803680218823822,
      "grad_norm": 0.47318413853645325,
      "learning_rate": 3.5983671101164573e-06,
      "loss": 0.7209,
      "step": 6765
    },
    {
      "epoch": 0.28040946578805587,
      "grad_norm": 0.41672495007514954,
      "learning_rate": 3.598159890588089e-06,
      "loss": 0.7214,
      "step": 6766
    },
    {
      "epoch": 0.28045090969372954,
      "grad_norm": 0.38916632533073425,
      "learning_rate": 3.5979526710597214e-06,
      "loss": 0.6873,
      "step": 6767
    },
    {
      "epoch": 0.2804923535994032,
      "grad_norm": 0.4564535319805145,
      "learning_rate": 3.5977454515313528e-06,
      "loss": 0.7455,
      "step": 6768
    },
    {
      "epoch": 0.2805337975050769,
      "grad_norm": 0.4205094873905182,
      "learning_rate": 3.5975382320029846e-06,
      "loss": 0.7141,
      "step": 6769
    },
    {
      "epoch": 0.2805752414107506,
      "grad_norm": 0.40840858221054077,
      "learning_rate": 3.597331012474616e-06,
      "loss": 0.7411,
      "step": 6770
    },
    {
      "epoch": 0.2806166853164242,
      "grad_norm": 0.4047994613647461,
      "learning_rate": 3.5971237929462478e-06,
      "loss": 0.7,
      "step": 6771
    },
    {
      "epoch": 0.2806581292220979,
      "grad_norm": 0.42194804549217224,
      "learning_rate": 3.596916573417879e-06,
      "loss": 0.6367,
      "step": 6772
    },
    {
      "epoch": 0.28069957312777155,
      "grad_norm": 0.4416385293006897,
      "learning_rate": 3.596709353889511e-06,
      "loss": 0.7307,
      "step": 6773
    },
    {
      "epoch": 0.28074101703344523,
      "grad_norm": 0.43495622277259827,
      "learning_rate": 3.5965021343611424e-06,
      "loss": 0.6785,
      "step": 6774
    },
    {
      "epoch": 0.2807824609391189,
      "grad_norm": 0.4550446569919586,
      "learning_rate": 3.596294914832774e-06,
      "loss": 0.7905,
      "step": 6775
    },
    {
      "epoch": 0.2808239048447926,
      "grad_norm": 0.3862614929676056,
      "learning_rate": 3.596087695304406e-06,
      "loss": 0.7422,
      "step": 6776
    },
    {
      "epoch": 0.28086534875046626,
      "grad_norm": 0.3790684938430786,
      "learning_rate": 3.5958804757760374e-06,
      "loss": 0.7302,
      "step": 6777
    },
    {
      "epoch": 0.28090679265613994,
      "grad_norm": 0.4111957848072052,
      "learning_rate": 3.595673256247669e-06,
      "loss": 0.7013,
      "step": 6778
    },
    {
      "epoch": 0.28094823656181356,
      "grad_norm": 0.37922751903533936,
      "learning_rate": 3.5954660367193006e-06,
      "loss": 0.682,
      "step": 6779
    },
    {
      "epoch": 0.28098968046748724,
      "grad_norm": 0.4079330563545227,
      "learning_rate": 3.5952588171909324e-06,
      "loss": 0.7131,
      "step": 6780
    },
    {
      "epoch": 0.2810311243731609,
      "grad_norm": 0.39573752880096436,
      "learning_rate": 3.5950515976625637e-06,
      "loss": 0.738,
      "step": 6781
    },
    {
      "epoch": 0.2810725682788346,
      "grad_norm": 0.3898983597755432,
      "learning_rate": 3.5948443781341956e-06,
      "loss": 0.6562,
      "step": 6782
    },
    {
      "epoch": 0.28111401218450827,
      "grad_norm": 0.4119100272655487,
      "learning_rate": 3.594637158605827e-06,
      "loss": 0.7031,
      "step": 6783
    },
    {
      "epoch": 0.28115545609018194,
      "grad_norm": 0.41005808115005493,
      "learning_rate": 3.594429939077459e-06,
      "loss": 0.6927,
      "step": 6784
    },
    {
      "epoch": 0.2811968999958556,
      "grad_norm": 0.417418509721756,
      "learning_rate": 3.594222719549091e-06,
      "loss": 0.6768,
      "step": 6785
    },
    {
      "epoch": 0.2812383439015293,
      "grad_norm": 0.4223876893520355,
      "learning_rate": 3.5940155000207224e-06,
      "loss": 0.6718,
      "step": 6786
    },
    {
      "epoch": 0.281279787807203,
      "grad_norm": 0.39712852239608765,
      "learning_rate": 3.593808280492354e-06,
      "loss": 0.6742,
      "step": 6787
    },
    {
      "epoch": 0.2813212317128766,
      "grad_norm": 0.38664698600769043,
      "learning_rate": 3.5936010609639856e-06,
      "loss": 0.708,
      "step": 6788
    },
    {
      "epoch": 0.2813626756185503,
      "grad_norm": 0.41835319995880127,
      "learning_rate": 3.5933938414356174e-06,
      "loss": 0.6664,
      "step": 6789
    },
    {
      "epoch": 0.28140411952422395,
      "grad_norm": 0.4258338510990143,
      "learning_rate": 3.5931866219072488e-06,
      "loss": 0.7495,
      "step": 6790
    },
    {
      "epoch": 0.2814455634298976,
      "grad_norm": 0.41461727023124695,
      "learning_rate": 3.5929794023788806e-06,
      "loss": 0.7327,
      "step": 6791
    },
    {
      "epoch": 0.2814870073355713,
      "grad_norm": 0.4301237463951111,
      "learning_rate": 3.592772182850512e-06,
      "loss": 0.7136,
      "step": 6792
    },
    {
      "epoch": 0.281528451241245,
      "grad_norm": 0.45910102128982544,
      "learning_rate": 3.5925649633221438e-06,
      "loss": 0.6887,
      "step": 6793
    },
    {
      "epoch": 0.28156989514691866,
      "grad_norm": 0.40189725160598755,
      "learning_rate": 3.5923577437937756e-06,
      "loss": 0.6957,
      "step": 6794
    },
    {
      "epoch": 0.28161133905259234,
      "grad_norm": 0.39384201169013977,
      "learning_rate": 3.592150524265407e-06,
      "loss": 0.7297,
      "step": 6795
    },
    {
      "epoch": 0.281652782958266,
      "grad_norm": 0.43027767539024353,
      "learning_rate": 3.5919433047370388e-06,
      "loss": 0.6921,
      "step": 6796
    },
    {
      "epoch": 0.28169422686393963,
      "grad_norm": 0.4064258337020874,
      "learning_rate": 3.59173608520867e-06,
      "loss": 0.7012,
      "step": 6797
    },
    {
      "epoch": 0.2817356707696133,
      "grad_norm": 0.47119030356407166,
      "learning_rate": 3.591528865680302e-06,
      "loss": 0.7407,
      "step": 6798
    },
    {
      "epoch": 0.281777114675287,
      "grad_norm": 0.422590047121048,
      "learning_rate": 3.5913216461519333e-06,
      "loss": 0.7405,
      "step": 6799
    },
    {
      "epoch": 0.28181855858096067,
      "grad_norm": 0.42778483033180237,
      "learning_rate": 3.591114426623565e-06,
      "loss": 0.7584,
      "step": 6800
    },
    {
      "epoch": 0.28186000248663434,
      "grad_norm": 0.49652472138404846,
      "learning_rate": 3.5909072070951965e-06,
      "loss": 0.7433,
      "step": 6801
    },
    {
      "epoch": 0.281901446392308,
      "grad_norm": 0.43851438164711,
      "learning_rate": 3.5906999875668288e-06,
      "loss": 0.7107,
      "step": 6802
    },
    {
      "epoch": 0.2819428902979817,
      "grad_norm": 0.4116474390029907,
      "learning_rate": 3.5904927680384606e-06,
      "loss": 0.7483,
      "step": 6803
    },
    {
      "epoch": 0.2819843342036554,
      "grad_norm": 0.4168466031551361,
      "learning_rate": 3.590285548510092e-06,
      "loss": 0.6995,
      "step": 6804
    },
    {
      "epoch": 0.28202577810932905,
      "grad_norm": 0.4503168761730194,
      "learning_rate": 3.5900783289817238e-06,
      "loss": 0.7612,
      "step": 6805
    },
    {
      "epoch": 0.28206722201500267,
      "grad_norm": 0.5053637623786926,
      "learning_rate": 3.589871109453355e-06,
      "loss": 0.7769,
      "step": 6806
    },
    {
      "epoch": 0.28210866592067635,
      "grad_norm": 0.3958262801170349,
      "learning_rate": 3.589663889924987e-06,
      "loss": 0.7206,
      "step": 6807
    },
    {
      "epoch": 0.28215010982635,
      "grad_norm": 0.42177364230155945,
      "learning_rate": 3.5894566703966184e-06,
      "loss": 0.7002,
      "step": 6808
    },
    {
      "epoch": 0.2821915537320237,
      "grad_norm": 0.41398656368255615,
      "learning_rate": 3.58924945086825e-06,
      "loss": 0.6926,
      "step": 6809
    },
    {
      "epoch": 0.2822329976376974,
      "grad_norm": 0.4351273775100708,
      "learning_rate": 3.5890422313398816e-06,
      "loss": 0.7874,
      "step": 6810
    },
    {
      "epoch": 0.28227444154337106,
      "grad_norm": 0.42515799403190613,
      "learning_rate": 3.5888350118115134e-06,
      "loss": 0.7341,
      "step": 6811
    },
    {
      "epoch": 0.28231588544904473,
      "grad_norm": 0.4272284507751465,
      "learning_rate": 3.588627792283145e-06,
      "loss": 0.6775,
      "step": 6812
    },
    {
      "epoch": 0.2823573293547184,
      "grad_norm": 0.4340786337852478,
      "learning_rate": 3.5884205727547766e-06,
      "loss": 0.6982,
      "step": 6813
    },
    {
      "epoch": 0.28239877326039203,
      "grad_norm": 0.42572149634361267,
      "learning_rate": 3.5882133532264084e-06,
      "loss": 0.7102,
      "step": 6814
    },
    {
      "epoch": 0.2824402171660657,
      "grad_norm": 0.3968062996864319,
      "learning_rate": 3.5880061336980398e-06,
      "loss": 0.7163,
      "step": 6815
    },
    {
      "epoch": 0.2824816610717394,
      "grad_norm": 0.4265420436859131,
      "learning_rate": 3.5877989141696716e-06,
      "loss": 0.6886,
      "step": 6816
    },
    {
      "epoch": 0.28252310497741306,
      "grad_norm": 0.43440908193588257,
      "learning_rate": 3.587591694641303e-06,
      "loss": 0.6893,
      "step": 6817
    },
    {
      "epoch": 0.28256454888308674,
      "grad_norm": 0.4182482063770294,
      "learning_rate": 3.587384475112935e-06,
      "loss": 0.7516,
      "step": 6818
    },
    {
      "epoch": 0.2826059927887604,
      "grad_norm": 0.47019872069358826,
      "learning_rate": 3.587177255584566e-06,
      "loss": 0.7336,
      "step": 6819
    },
    {
      "epoch": 0.2826474366944341,
      "grad_norm": 0.43315190076828003,
      "learning_rate": 3.5869700360561984e-06,
      "loss": 0.7214,
      "step": 6820
    },
    {
      "epoch": 0.2826888806001078,
      "grad_norm": 0.41775181889533997,
      "learning_rate": 3.58676281652783e-06,
      "loss": 0.7312,
      "step": 6821
    },
    {
      "epoch": 0.28273032450578145,
      "grad_norm": 0.4487468898296356,
      "learning_rate": 3.5865555969994616e-06,
      "loss": 0.7766,
      "step": 6822
    },
    {
      "epoch": 0.28277176841145507,
      "grad_norm": 0.4046151041984558,
      "learning_rate": 3.5863483774710934e-06,
      "loss": 0.6927,
      "step": 6823
    },
    {
      "epoch": 0.28281321231712875,
      "grad_norm": 0.3830421566963196,
      "learning_rate": 3.5861411579427248e-06,
      "loss": 0.6884,
      "step": 6824
    },
    {
      "epoch": 0.2828546562228024,
      "grad_norm": 0.4627036154270172,
      "learning_rate": 3.5859339384143566e-06,
      "loss": 0.6952,
      "step": 6825
    },
    {
      "epoch": 0.2828961001284761,
      "grad_norm": 0.4159393310546875,
      "learning_rate": 3.585726718885988e-06,
      "loss": 0.7522,
      "step": 6826
    },
    {
      "epoch": 0.2829375440341498,
      "grad_norm": 0.439156711101532,
      "learning_rate": 3.5855194993576198e-06,
      "loss": 0.7345,
      "step": 6827
    },
    {
      "epoch": 0.28297898793982346,
      "grad_norm": 0.42447394132614136,
      "learning_rate": 3.5853122798292516e-06,
      "loss": 0.7217,
      "step": 6828
    },
    {
      "epoch": 0.28302043184549713,
      "grad_norm": 0.4101831018924713,
      "learning_rate": 3.585105060300883e-06,
      "loss": 0.7296,
      "step": 6829
    },
    {
      "epoch": 0.2830618757511708,
      "grad_norm": 0.40400081872940063,
      "learning_rate": 3.5848978407725148e-06,
      "loss": 0.7402,
      "step": 6830
    },
    {
      "epoch": 0.2831033196568445,
      "grad_norm": 0.4275926649570465,
      "learning_rate": 3.584690621244146e-06,
      "loss": 0.7375,
      "step": 6831
    },
    {
      "epoch": 0.2831447635625181,
      "grad_norm": 0.44269418716430664,
      "learning_rate": 3.584483401715778e-06,
      "loss": 0.702,
      "step": 6832
    },
    {
      "epoch": 0.2831862074681918,
      "grad_norm": 0.4247515797615051,
      "learning_rate": 3.5842761821874094e-06,
      "loss": 0.6948,
      "step": 6833
    },
    {
      "epoch": 0.28322765137386546,
      "grad_norm": 0.39553365111351013,
      "learning_rate": 3.584068962659041e-06,
      "loss": 0.6577,
      "step": 6834
    },
    {
      "epoch": 0.28326909527953914,
      "grad_norm": 0.45518970489501953,
      "learning_rate": 3.5838617431306725e-06,
      "loss": 0.7421,
      "step": 6835
    },
    {
      "epoch": 0.2833105391852128,
      "grad_norm": 0.407031774520874,
      "learning_rate": 3.5836545236023048e-06,
      "loss": 0.6809,
      "step": 6836
    },
    {
      "epoch": 0.2833519830908865,
      "grad_norm": 0.4166657328605652,
      "learning_rate": 3.5834473040739366e-06,
      "loss": 0.693,
      "step": 6837
    },
    {
      "epoch": 0.28339342699656017,
      "grad_norm": 0.5113471150398254,
      "learning_rate": 3.583240084545568e-06,
      "loss": 0.729,
      "step": 6838
    },
    {
      "epoch": 0.28343487090223385,
      "grad_norm": 0.4281917214393616,
      "learning_rate": 3.5830328650171998e-06,
      "loss": 0.7344,
      "step": 6839
    },
    {
      "epoch": 0.28347631480790747,
      "grad_norm": 0.40506628155708313,
      "learning_rate": 3.582825645488831e-06,
      "loss": 0.7354,
      "step": 6840
    },
    {
      "epoch": 0.28351775871358115,
      "grad_norm": 0.44882896542549133,
      "learning_rate": 3.582618425960463e-06,
      "loss": 0.7577,
      "step": 6841
    },
    {
      "epoch": 0.2835592026192548,
      "grad_norm": 0.4255295693874359,
      "learning_rate": 3.5824112064320944e-06,
      "loss": 0.6532,
      "step": 6842
    },
    {
      "epoch": 0.2836006465249285,
      "grad_norm": 0.38957056403160095,
      "learning_rate": 3.582203986903726e-06,
      "loss": 0.6298,
      "step": 6843
    },
    {
      "epoch": 0.2836420904306022,
      "grad_norm": 0.4765869081020355,
      "learning_rate": 3.5819967673753576e-06,
      "loss": 0.7773,
      "step": 6844
    },
    {
      "epoch": 0.28368353433627586,
      "grad_norm": 0.47083091735839844,
      "learning_rate": 3.5817895478469894e-06,
      "loss": 0.7004,
      "step": 6845
    },
    {
      "epoch": 0.28372497824194953,
      "grad_norm": 0.3907012939453125,
      "learning_rate": 3.581582328318621e-06,
      "loss": 0.6726,
      "step": 6846
    },
    {
      "epoch": 0.2837664221476232,
      "grad_norm": 0.39952942728996277,
      "learning_rate": 3.5813751087902526e-06,
      "loss": 0.667,
      "step": 6847
    },
    {
      "epoch": 0.2838078660532969,
      "grad_norm": 0.41507667303085327,
      "learning_rate": 3.5811678892618844e-06,
      "loss": 0.72,
      "step": 6848
    },
    {
      "epoch": 0.2838493099589705,
      "grad_norm": 0.39529845118522644,
      "learning_rate": 3.5809606697335158e-06,
      "loss": 0.6989,
      "step": 6849
    },
    {
      "epoch": 0.2838907538646442,
      "grad_norm": 0.4155879318714142,
      "learning_rate": 3.5807534502051476e-06,
      "loss": 0.7163,
      "step": 6850
    },
    {
      "epoch": 0.28393219777031786,
      "grad_norm": 0.42982950806617737,
      "learning_rate": 3.580546230676779e-06,
      "loss": 0.6882,
      "step": 6851
    },
    {
      "epoch": 0.28397364167599154,
      "grad_norm": 0.44255757331848145,
      "learning_rate": 3.580339011148411e-06,
      "loss": 0.7227,
      "step": 6852
    },
    {
      "epoch": 0.2840150855816652,
      "grad_norm": 0.4067123532295227,
      "learning_rate": 3.580131791620042e-06,
      "loss": 0.739,
      "step": 6853
    },
    {
      "epoch": 0.2840565294873389,
      "grad_norm": 0.440277099609375,
      "learning_rate": 3.5799245720916744e-06,
      "loss": 0.6924,
      "step": 6854
    },
    {
      "epoch": 0.28409797339301257,
      "grad_norm": 0.38350799679756165,
      "learning_rate": 3.579717352563306e-06,
      "loss": 0.6819,
      "step": 6855
    },
    {
      "epoch": 0.28413941729868625,
      "grad_norm": 0.4033219516277313,
      "learning_rate": 3.5795101330349376e-06,
      "loss": 0.713,
      "step": 6856
    },
    {
      "epoch": 0.2841808612043599,
      "grad_norm": 0.4154597520828247,
      "learning_rate": 3.5793029135065694e-06,
      "loss": 0.7117,
      "step": 6857
    },
    {
      "epoch": 0.28422230511003355,
      "grad_norm": 0.40948623418807983,
      "learning_rate": 3.5790956939782008e-06,
      "loss": 0.6868,
      "step": 6858
    },
    {
      "epoch": 0.2842637490157072,
      "grad_norm": 0.4153733253479004,
      "learning_rate": 3.5788884744498326e-06,
      "loss": 0.6741,
      "step": 6859
    },
    {
      "epoch": 0.2843051929213809,
      "grad_norm": 0.4063993990421295,
      "learning_rate": 3.578681254921464e-06,
      "loss": 0.7068,
      "step": 6860
    },
    {
      "epoch": 0.2843466368270546,
      "grad_norm": 0.4236097037792206,
      "learning_rate": 3.5784740353930958e-06,
      "loss": 0.686,
      "step": 6861
    },
    {
      "epoch": 0.28438808073272825,
      "grad_norm": 0.4095933735370636,
      "learning_rate": 3.578266815864727e-06,
      "loss": 0.7007,
      "step": 6862
    },
    {
      "epoch": 0.28442952463840193,
      "grad_norm": 0.44229742884635925,
      "learning_rate": 3.578059596336359e-06,
      "loss": 0.6803,
      "step": 6863
    },
    {
      "epoch": 0.2844709685440756,
      "grad_norm": 0.3964507579803467,
      "learning_rate": 3.5778523768079908e-06,
      "loss": 0.6943,
      "step": 6864
    },
    {
      "epoch": 0.2845124124497493,
      "grad_norm": 0.4179035425186157,
      "learning_rate": 3.577645157279622e-06,
      "loss": 0.7395,
      "step": 6865
    },
    {
      "epoch": 0.28455385635542296,
      "grad_norm": 0.4197254180908203,
      "learning_rate": 3.577437937751254e-06,
      "loss": 0.6934,
      "step": 6866
    },
    {
      "epoch": 0.2845953002610966,
      "grad_norm": 0.39869406819343567,
      "learning_rate": 3.5772307182228854e-06,
      "loss": 0.6782,
      "step": 6867
    },
    {
      "epoch": 0.28463674416677026,
      "grad_norm": 0.424480140209198,
      "learning_rate": 3.577023498694517e-06,
      "loss": 0.7069,
      "step": 6868
    },
    {
      "epoch": 0.28467818807244394,
      "grad_norm": 0.41718241572380066,
      "learning_rate": 3.5768162791661486e-06,
      "loss": 0.7478,
      "step": 6869
    },
    {
      "epoch": 0.2847196319781176,
      "grad_norm": 0.406290739774704,
      "learning_rate": 3.5766090596377808e-06,
      "loss": 0.6545,
      "step": 6870
    },
    {
      "epoch": 0.2847610758837913,
      "grad_norm": 0.4433014690876007,
      "learning_rate": 3.5764018401094117e-06,
      "loss": 0.7258,
      "step": 6871
    },
    {
      "epoch": 0.28480251978946497,
      "grad_norm": 0.4183829128742218,
      "learning_rate": 3.576194620581044e-06,
      "loss": 0.7249,
      "step": 6872
    },
    {
      "epoch": 0.28484396369513865,
      "grad_norm": 0.4561851918697357,
      "learning_rate": 3.575987401052676e-06,
      "loss": 0.7566,
      "step": 6873
    },
    {
      "epoch": 0.2848854076008123,
      "grad_norm": 0.384236216545105,
      "learning_rate": 3.575780181524307e-06,
      "loss": 0.6927,
      "step": 6874
    },
    {
      "epoch": 0.28492685150648595,
      "grad_norm": 0.44704267382621765,
      "learning_rate": 3.575572961995939e-06,
      "loss": 0.7361,
      "step": 6875
    },
    {
      "epoch": 0.2849682954121596,
      "grad_norm": 0.40363720059394836,
      "learning_rate": 3.5753657424675704e-06,
      "loss": 0.7205,
      "step": 6876
    },
    {
      "epoch": 0.2850097393178333,
      "grad_norm": 0.4272850453853607,
      "learning_rate": 3.575158522939202e-06,
      "loss": 0.7063,
      "step": 6877
    },
    {
      "epoch": 0.285051183223507,
      "grad_norm": 0.47055304050445557,
      "learning_rate": 3.5749513034108336e-06,
      "loss": 0.6899,
      "step": 6878
    },
    {
      "epoch": 0.28509262712918065,
      "grad_norm": 0.43875035643577576,
      "learning_rate": 3.5747440838824654e-06,
      "loss": 0.7505,
      "step": 6879
    },
    {
      "epoch": 0.28513407103485433,
      "grad_norm": 0.4094933271408081,
      "learning_rate": 3.5745368643540968e-06,
      "loss": 0.6255,
      "step": 6880
    },
    {
      "epoch": 0.285175514940528,
      "grad_norm": 0.45186007022857666,
      "learning_rate": 3.5743296448257286e-06,
      "loss": 0.7068,
      "step": 6881
    },
    {
      "epoch": 0.2852169588462017,
      "grad_norm": 0.4162149727344513,
      "learning_rate": 3.5741224252973604e-06,
      "loss": 0.7461,
      "step": 6882
    },
    {
      "epoch": 0.28525840275187536,
      "grad_norm": 0.38705718517303467,
      "learning_rate": 3.5739152057689918e-06,
      "loss": 0.6702,
      "step": 6883
    },
    {
      "epoch": 0.285299846657549,
      "grad_norm": 0.41094326972961426,
      "learning_rate": 3.5737079862406236e-06,
      "loss": 0.7177,
      "step": 6884
    },
    {
      "epoch": 0.28534129056322266,
      "grad_norm": 0.4257148802280426,
      "learning_rate": 3.573500766712255e-06,
      "loss": 0.719,
      "step": 6885
    },
    {
      "epoch": 0.28538273446889634,
      "grad_norm": 0.39430439472198486,
      "learning_rate": 3.573293547183887e-06,
      "loss": 0.694,
      "step": 6886
    },
    {
      "epoch": 0.28542417837457,
      "grad_norm": 0.40021035075187683,
      "learning_rate": 3.573086327655518e-06,
      "loss": 0.6809,
      "step": 6887
    },
    {
      "epoch": 0.2854656222802437,
      "grad_norm": 0.42008695006370544,
      "learning_rate": 3.5728791081271504e-06,
      "loss": 0.6736,
      "step": 6888
    },
    {
      "epoch": 0.28550706618591737,
      "grad_norm": 0.4037550091743469,
      "learning_rate": 3.572671888598782e-06,
      "loss": 0.6841,
      "step": 6889
    },
    {
      "epoch": 0.28554851009159105,
      "grad_norm": 0.48485174775123596,
      "learning_rate": 3.5724646690704136e-06,
      "loss": 0.7343,
      "step": 6890
    },
    {
      "epoch": 0.2855899539972647,
      "grad_norm": 0.43097740411758423,
      "learning_rate": 3.5722574495420454e-06,
      "loss": 0.705,
      "step": 6891
    },
    {
      "epoch": 0.2856313979029384,
      "grad_norm": 0.39484497904777527,
      "learning_rate": 3.5720502300136768e-06,
      "loss": 0.7366,
      "step": 6892
    },
    {
      "epoch": 0.285672841808612,
      "grad_norm": 0.45459839701652527,
      "learning_rate": 3.5718430104853086e-06,
      "loss": 0.7476,
      "step": 6893
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.46201279759407043,
      "learning_rate": 3.57163579095694e-06,
      "loss": 0.7417,
      "step": 6894
    },
    {
      "epoch": 0.2857557296199594,
      "grad_norm": 0.41701561212539673,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 0.708,
      "step": 6895
    },
    {
      "epoch": 0.28579717352563305,
      "grad_norm": 0.39320895075798035,
      "learning_rate": 3.571221351900203e-06,
      "loss": 0.7002,
      "step": 6896
    },
    {
      "epoch": 0.28583861743130673,
      "grad_norm": 0.41212591528892517,
      "learning_rate": 3.571014132371835e-06,
      "loss": 0.7267,
      "step": 6897
    },
    {
      "epoch": 0.2858800613369804,
      "grad_norm": 0.44438445568084717,
      "learning_rate": 3.5708069128434668e-06,
      "loss": 0.7385,
      "step": 6898
    },
    {
      "epoch": 0.2859215052426541,
      "grad_norm": 0.4290604889392853,
      "learning_rate": 3.570599693315098e-06,
      "loss": 0.6769,
      "step": 6899
    },
    {
      "epoch": 0.28596294914832776,
      "grad_norm": 0.40066683292388916,
      "learning_rate": 3.57039247378673e-06,
      "loss": 0.6979,
      "step": 6900
    },
    {
      "epoch": 0.2860043930540014,
      "grad_norm": 0.4036765992641449,
      "learning_rate": 3.5701852542583614e-06,
      "loss": 0.7294,
      "step": 6901
    },
    {
      "epoch": 0.28604583695967506,
      "grad_norm": 0.4126715660095215,
      "learning_rate": 3.569978034729993e-06,
      "loss": 0.6887,
      "step": 6902
    },
    {
      "epoch": 0.28608728086534874,
      "grad_norm": 0.39421460032463074,
      "learning_rate": 3.5697708152016246e-06,
      "loss": 0.7019,
      "step": 6903
    },
    {
      "epoch": 0.2861287247710224,
      "grad_norm": 0.423803448677063,
      "learning_rate": 3.569563595673257e-06,
      "loss": 0.7495,
      "step": 6904
    },
    {
      "epoch": 0.2861701686766961,
      "grad_norm": 0.41758954524993896,
      "learning_rate": 3.5693563761448878e-06,
      "loss": 0.7222,
      "step": 6905
    },
    {
      "epoch": 0.28621161258236977,
      "grad_norm": 0.4164711833000183,
      "learning_rate": 3.56914915661652e-06,
      "loss": 0.6617,
      "step": 6906
    },
    {
      "epoch": 0.28625305648804344,
      "grad_norm": 0.44317251443862915,
      "learning_rate": 3.568941937088152e-06,
      "loss": 0.7107,
      "step": 6907
    },
    {
      "epoch": 0.2862945003937171,
      "grad_norm": 0.4078119993209839,
      "learning_rate": 3.568734717559783e-06,
      "loss": 0.6855,
      "step": 6908
    },
    {
      "epoch": 0.2863359442993908,
      "grad_norm": 0.4446542263031006,
      "learning_rate": 3.568527498031415e-06,
      "loss": 0.7556,
      "step": 6909
    },
    {
      "epoch": 0.2863773882050644,
      "grad_norm": 0.4360417425632477,
      "learning_rate": 3.5683202785030464e-06,
      "loss": 0.7273,
      "step": 6910
    },
    {
      "epoch": 0.2864188321107381,
      "grad_norm": 0.44036340713500977,
      "learning_rate": 3.568113058974678e-06,
      "loss": 0.7649,
      "step": 6911
    },
    {
      "epoch": 0.2864602760164118,
      "grad_norm": 0.4655351936817169,
      "learning_rate": 3.5679058394463096e-06,
      "loss": 0.7517,
      "step": 6912
    },
    {
      "epoch": 0.28650171992208545,
      "grad_norm": 0.4113084375858307,
      "learning_rate": 3.5676986199179414e-06,
      "loss": 0.7412,
      "step": 6913
    },
    {
      "epoch": 0.28654316382775913,
      "grad_norm": 0.3831551671028137,
      "learning_rate": 3.5674914003895728e-06,
      "loss": 0.7151,
      "step": 6914
    },
    {
      "epoch": 0.2865846077334328,
      "grad_norm": 0.4323503375053406,
      "learning_rate": 3.5672841808612046e-06,
      "loss": 0.6804,
      "step": 6915
    },
    {
      "epoch": 0.2866260516391065,
      "grad_norm": 0.4357014298439026,
      "learning_rate": 3.5670769613328364e-06,
      "loss": 0.6738,
      "step": 6916
    },
    {
      "epoch": 0.28666749554478016,
      "grad_norm": 0.4165201783180237,
      "learning_rate": 3.5668697418044678e-06,
      "loss": 0.6687,
      "step": 6917
    },
    {
      "epoch": 0.28670893945045384,
      "grad_norm": 0.40601587295532227,
      "learning_rate": 3.5666625222760996e-06,
      "loss": 0.7288,
      "step": 6918
    },
    {
      "epoch": 0.28675038335612746,
      "grad_norm": 0.42080700397491455,
      "learning_rate": 3.566455302747731e-06,
      "loss": 0.6406,
      "step": 6919
    },
    {
      "epoch": 0.28679182726180114,
      "grad_norm": 0.4565685987472534,
      "learning_rate": 3.566248083219363e-06,
      "loss": 0.7178,
      "step": 6920
    },
    {
      "epoch": 0.2868332711674748,
      "grad_norm": 0.4114568829536438,
      "learning_rate": 3.566040863690994e-06,
      "loss": 0.6576,
      "step": 6921
    },
    {
      "epoch": 0.2868747150731485,
      "grad_norm": 0.437321275472641,
      "learning_rate": 3.5658336441626264e-06,
      "loss": 0.7432,
      "step": 6922
    },
    {
      "epoch": 0.28691615897882217,
      "grad_norm": 0.4223303496837616,
      "learning_rate": 3.5656264246342574e-06,
      "loss": 0.7227,
      "step": 6923
    },
    {
      "epoch": 0.28695760288449584,
      "grad_norm": 0.42915213108062744,
      "learning_rate": 3.5654192051058896e-06,
      "loss": 0.7006,
      "step": 6924
    },
    {
      "epoch": 0.2869990467901695,
      "grad_norm": 0.4071757197380066,
      "learning_rate": 3.5652119855775214e-06,
      "loss": 0.6992,
      "step": 6925
    },
    {
      "epoch": 0.2870404906958432,
      "grad_norm": 0.4418925344944,
      "learning_rate": 3.5650047660491528e-06,
      "loss": 0.6819,
      "step": 6926
    },
    {
      "epoch": 0.2870819346015168,
      "grad_norm": 0.4335483908653259,
      "learning_rate": 3.5647975465207846e-06,
      "loss": 0.6946,
      "step": 6927
    },
    {
      "epoch": 0.2871233785071905,
      "grad_norm": 0.3746641278266907,
      "learning_rate": 3.564590326992416e-06,
      "loss": 0.7258,
      "step": 6928
    },
    {
      "epoch": 0.2871648224128642,
      "grad_norm": 0.3801139295101166,
      "learning_rate": 3.5643831074640478e-06,
      "loss": 0.6484,
      "step": 6929
    },
    {
      "epoch": 0.28720626631853785,
      "grad_norm": 0.380740225315094,
      "learning_rate": 3.564175887935679e-06,
      "loss": 0.6312,
      "step": 6930
    },
    {
      "epoch": 0.2872477102242115,
      "grad_norm": 0.4376373291015625,
      "learning_rate": 3.563968668407311e-06,
      "loss": 0.6968,
      "step": 6931
    },
    {
      "epoch": 0.2872891541298852,
      "grad_norm": 0.38756129145622253,
      "learning_rate": 3.5637614488789424e-06,
      "loss": 0.6603,
      "step": 6932
    },
    {
      "epoch": 0.2873305980355589,
      "grad_norm": 0.409757137298584,
      "learning_rate": 3.563554229350574e-06,
      "loss": 0.7363,
      "step": 6933
    },
    {
      "epoch": 0.28737204194123256,
      "grad_norm": 0.42089688777923584,
      "learning_rate": 3.563347009822206e-06,
      "loss": 0.6438,
      "step": 6934
    },
    {
      "epoch": 0.28741348584690624,
      "grad_norm": 0.4528174102306366,
      "learning_rate": 3.5631397902938374e-06,
      "loss": 0.7427,
      "step": 6935
    },
    {
      "epoch": 0.28745492975257986,
      "grad_norm": 0.4734955132007599,
      "learning_rate": 3.562932570765469e-06,
      "loss": 0.7549,
      "step": 6936
    },
    {
      "epoch": 0.28749637365825353,
      "grad_norm": 0.4542578160762787,
      "learning_rate": 3.5627253512371006e-06,
      "loss": 0.7195,
      "step": 6937
    },
    {
      "epoch": 0.2875378175639272,
      "grad_norm": 0.41403213143348694,
      "learning_rate": 3.562518131708733e-06,
      "loss": 0.6973,
      "step": 6938
    },
    {
      "epoch": 0.2875792614696009,
      "grad_norm": 0.4291495382785797,
      "learning_rate": 3.5623109121803638e-06,
      "loss": 0.762,
      "step": 6939
    },
    {
      "epoch": 0.28762070537527457,
      "grad_norm": 0.45406877994537354,
      "learning_rate": 3.562103692651996e-06,
      "loss": 0.7214,
      "step": 6940
    },
    {
      "epoch": 0.28766214928094824,
      "grad_norm": 0.41754409670829773,
      "learning_rate": 3.5618964731236274e-06,
      "loss": 0.688,
      "step": 6941
    },
    {
      "epoch": 0.2877035931866219,
      "grad_norm": 0.4389185905456543,
      "learning_rate": 3.561689253595259e-06,
      "loss": 0.6749,
      "step": 6942
    },
    {
      "epoch": 0.2877450370922956,
      "grad_norm": 0.4193671941757202,
      "learning_rate": 3.561482034066891e-06,
      "loss": 0.6975,
      "step": 6943
    },
    {
      "epoch": 0.2877864809979693,
      "grad_norm": 0.4176039695739746,
      "learning_rate": 3.5612748145385224e-06,
      "loss": 0.6968,
      "step": 6944
    },
    {
      "epoch": 0.2878279249036429,
      "grad_norm": 0.4057181477546692,
      "learning_rate": 3.561067595010154e-06,
      "loss": 0.7524,
      "step": 6945
    },
    {
      "epoch": 0.2878693688093166,
      "grad_norm": 0.42737317085266113,
      "learning_rate": 3.5608603754817856e-06,
      "loss": 0.7729,
      "step": 6946
    },
    {
      "epoch": 0.28791081271499025,
      "grad_norm": 0.4156981408596039,
      "learning_rate": 3.5606531559534174e-06,
      "loss": 0.6925,
      "step": 6947
    },
    {
      "epoch": 0.2879522566206639,
      "grad_norm": 0.40141257643699646,
      "learning_rate": 3.5604459364250488e-06,
      "loss": 0.7211,
      "step": 6948
    },
    {
      "epoch": 0.2879937005263376,
      "grad_norm": 0.38518083095550537,
      "learning_rate": 3.5602387168966806e-06,
      "loss": 0.6863,
      "step": 6949
    },
    {
      "epoch": 0.2880351444320113,
      "grad_norm": 0.4188710153102875,
      "learning_rate": 3.5600314973683124e-06,
      "loss": 0.6549,
      "step": 6950
    },
    {
      "epoch": 0.28807658833768496,
      "grad_norm": 0.3895302414894104,
      "learning_rate": 3.5598242778399438e-06,
      "loss": 0.6522,
      "step": 6951
    },
    {
      "epoch": 0.28811803224335863,
      "grad_norm": 0.4401589632034302,
      "learning_rate": 3.5596170583115756e-06,
      "loss": 0.7512,
      "step": 6952
    },
    {
      "epoch": 0.2881594761490323,
      "grad_norm": 0.40924349427223206,
      "learning_rate": 3.559409838783207e-06,
      "loss": 0.6821,
      "step": 6953
    },
    {
      "epoch": 0.28820092005470593,
      "grad_norm": 0.464637815952301,
      "learning_rate": 3.559202619254839e-06,
      "loss": 0.7324,
      "step": 6954
    },
    {
      "epoch": 0.2882423639603796,
      "grad_norm": 0.41231903433799744,
      "learning_rate": 3.55899539972647e-06,
      "loss": 0.7085,
      "step": 6955
    },
    {
      "epoch": 0.2882838078660533,
      "grad_norm": 0.43170279264450073,
      "learning_rate": 3.5587881801981024e-06,
      "loss": 0.7134,
      "step": 6956
    },
    {
      "epoch": 0.28832525177172696,
      "grad_norm": 0.39256665110588074,
      "learning_rate": 3.5585809606697338e-06,
      "loss": 0.7053,
      "step": 6957
    },
    {
      "epoch": 0.28836669567740064,
      "grad_norm": 0.42892512679100037,
      "learning_rate": 3.5583737411413656e-06,
      "loss": 0.6918,
      "step": 6958
    },
    {
      "epoch": 0.2884081395830743,
      "grad_norm": 0.43007582426071167,
      "learning_rate": 3.5581665216129974e-06,
      "loss": 0.6985,
      "step": 6959
    },
    {
      "epoch": 0.288449583488748,
      "grad_norm": 0.38796335458755493,
      "learning_rate": 3.557959302084629e-06,
      "loss": 0.7002,
      "step": 6960
    },
    {
      "epoch": 0.2884910273944217,
      "grad_norm": 0.42302998900413513,
      "learning_rate": 3.5577520825562606e-06,
      "loss": 0.7256,
      "step": 6961
    },
    {
      "epoch": 0.2885324713000953,
      "grad_norm": 0.42087680101394653,
      "learning_rate": 3.557544863027892e-06,
      "loss": 0.6429,
      "step": 6962
    },
    {
      "epoch": 0.28857391520576897,
      "grad_norm": 0.4813157618045807,
      "learning_rate": 3.557337643499524e-06,
      "loss": 0.7273,
      "step": 6963
    },
    {
      "epoch": 0.28861535911144265,
      "grad_norm": 0.4117436110973358,
      "learning_rate": 3.557130423971155e-06,
      "loss": 0.7234,
      "step": 6964
    },
    {
      "epoch": 0.2886568030171163,
      "grad_norm": 0.404603511095047,
      "learning_rate": 3.556923204442787e-06,
      "loss": 0.7378,
      "step": 6965
    },
    {
      "epoch": 0.28869824692279,
      "grad_norm": 0.46006280183792114,
      "learning_rate": 3.5567159849144184e-06,
      "loss": 0.7607,
      "step": 6966
    },
    {
      "epoch": 0.2887396908284637,
      "grad_norm": 0.43420031666755676,
      "learning_rate": 3.55650876538605e-06,
      "loss": 0.7297,
      "step": 6967
    },
    {
      "epoch": 0.28878113473413736,
      "grad_norm": 0.4212886095046997,
      "learning_rate": 3.556301545857682e-06,
      "loss": 0.7119,
      "step": 6968
    },
    {
      "epoch": 0.28882257863981103,
      "grad_norm": 0.4272632598876953,
      "learning_rate": 3.5560943263293134e-06,
      "loss": 0.6956,
      "step": 6969
    },
    {
      "epoch": 0.2888640225454847,
      "grad_norm": 0.4428102672100067,
      "learning_rate": 3.555887106800945e-06,
      "loss": 0.7521,
      "step": 6970
    },
    {
      "epoch": 0.28890546645115833,
      "grad_norm": 0.4249163866043091,
      "learning_rate": 3.5556798872725766e-06,
      "loss": 0.6909,
      "step": 6971
    },
    {
      "epoch": 0.288946910356832,
      "grad_norm": 0.40772005915641785,
      "learning_rate": 3.555472667744209e-06,
      "loss": 0.7166,
      "step": 6972
    },
    {
      "epoch": 0.2889883542625057,
      "grad_norm": 0.4057495594024658,
      "learning_rate": 3.5552654482158398e-06,
      "loss": 0.6938,
      "step": 6973
    },
    {
      "epoch": 0.28902979816817936,
      "grad_norm": 0.3959791958332062,
      "learning_rate": 3.555058228687472e-06,
      "loss": 0.6771,
      "step": 6974
    },
    {
      "epoch": 0.28907124207385304,
      "grad_norm": 0.41590359807014465,
      "learning_rate": 3.5548510091591034e-06,
      "loss": 0.6859,
      "step": 6975
    },
    {
      "epoch": 0.2891126859795267,
      "grad_norm": 0.43379271030426025,
      "learning_rate": 3.554643789630735e-06,
      "loss": 0.7222,
      "step": 6976
    },
    {
      "epoch": 0.2891541298852004,
      "grad_norm": 0.40103286504745483,
      "learning_rate": 3.554436570102367e-06,
      "loss": 0.6851,
      "step": 6977
    },
    {
      "epoch": 0.28919557379087407,
      "grad_norm": 0.4490341544151306,
      "learning_rate": 3.5542293505739984e-06,
      "loss": 0.7628,
      "step": 6978
    },
    {
      "epoch": 0.28923701769654775,
      "grad_norm": 0.4162808060646057,
      "learning_rate": 3.55402213104563e-06,
      "loss": 0.7213,
      "step": 6979
    },
    {
      "epoch": 0.28927846160222137,
      "grad_norm": 0.43383705615997314,
      "learning_rate": 3.5538149115172616e-06,
      "loss": 0.7554,
      "step": 6980
    },
    {
      "epoch": 0.28931990550789505,
      "grad_norm": 0.42794209718704224,
      "learning_rate": 3.5536076919888934e-06,
      "loss": 0.7288,
      "step": 6981
    },
    {
      "epoch": 0.2893613494135687,
      "grad_norm": 0.39905351400375366,
      "learning_rate": 3.5534004724605248e-06,
      "loss": 0.7463,
      "step": 6982
    },
    {
      "epoch": 0.2894027933192424,
      "grad_norm": 0.4054827094078064,
      "learning_rate": 3.5531932529321566e-06,
      "loss": 0.698,
      "step": 6983
    },
    {
      "epoch": 0.2894442372249161,
      "grad_norm": 0.4243659973144531,
      "learning_rate": 3.552986033403788e-06,
      "loss": 0.7156,
      "step": 6984
    },
    {
      "epoch": 0.28948568113058976,
      "grad_norm": 0.4119908809661865,
      "learning_rate": 3.5527788138754198e-06,
      "loss": 0.7126,
      "step": 6985
    },
    {
      "epoch": 0.28952712503626343,
      "grad_norm": 0.482086718082428,
      "learning_rate": 3.5525715943470516e-06,
      "loss": 0.7266,
      "step": 6986
    },
    {
      "epoch": 0.2895685689419371,
      "grad_norm": 0.40206649899482727,
      "learning_rate": 3.552364374818683e-06,
      "loss": 0.6976,
      "step": 6987
    },
    {
      "epoch": 0.28961001284761073,
      "grad_norm": 0.43516209721565247,
      "learning_rate": 3.552157155290315e-06,
      "loss": 0.7194,
      "step": 6988
    },
    {
      "epoch": 0.2896514567532844,
      "grad_norm": 0.41434088349342346,
      "learning_rate": 3.551949935761946e-06,
      "loss": 0.7152,
      "step": 6989
    },
    {
      "epoch": 0.2896929006589581,
      "grad_norm": 0.40670010447502136,
      "learning_rate": 3.5517427162335784e-06,
      "loss": 0.7114,
      "step": 6990
    },
    {
      "epoch": 0.28973434456463176,
      "grad_norm": 0.44849127531051636,
      "learning_rate": 3.55153549670521e-06,
      "loss": 0.6891,
      "step": 6991
    },
    {
      "epoch": 0.28977578847030544,
      "grad_norm": 0.42762190103530884,
      "learning_rate": 3.5513282771768416e-06,
      "loss": 0.7119,
      "step": 6992
    },
    {
      "epoch": 0.2898172323759791,
      "grad_norm": 0.3914492726325989,
      "learning_rate": 3.551121057648473e-06,
      "loss": 0.6821,
      "step": 6993
    },
    {
      "epoch": 0.2898586762816528,
      "grad_norm": 0.4504014551639557,
      "learning_rate": 3.550913838120105e-06,
      "loss": 0.7123,
      "step": 6994
    },
    {
      "epoch": 0.28990012018732647,
      "grad_norm": 0.4167000651359558,
      "learning_rate": 3.5507066185917366e-06,
      "loss": 0.6953,
      "step": 6995
    },
    {
      "epoch": 0.28994156409300015,
      "grad_norm": 0.4186878800392151,
      "learning_rate": 3.550499399063368e-06,
      "loss": 0.7194,
      "step": 6996
    },
    {
      "epoch": 0.28998300799867377,
      "grad_norm": 0.440375953912735,
      "learning_rate": 3.550292179535e-06,
      "loss": 0.6802,
      "step": 6997
    },
    {
      "epoch": 0.29002445190434745,
      "grad_norm": 0.42046085000038147,
      "learning_rate": 3.550084960006631e-06,
      "loss": 0.7114,
      "step": 6998
    },
    {
      "epoch": 0.2900658958100211,
      "grad_norm": 0.43134018778800964,
      "learning_rate": 3.549877740478263e-06,
      "loss": 0.7512,
      "step": 6999
    },
    {
      "epoch": 0.2901073397156948,
      "grad_norm": 0.4230429530143738,
      "learning_rate": 3.5496705209498944e-06,
      "loss": 0.6779,
      "step": 7000
    },
    {
      "epoch": 0.2901487836213685,
      "grad_norm": 0.43294405937194824,
      "learning_rate": 3.549463301421526e-06,
      "loss": 0.7444,
      "step": 7001
    },
    {
      "epoch": 0.29019022752704215,
      "grad_norm": 0.4135044515132904,
      "learning_rate": 3.549256081893158e-06,
      "loss": 0.7222,
      "step": 7002
    },
    {
      "epoch": 0.29023167143271583,
      "grad_norm": 0.42034912109375,
      "learning_rate": 3.5490488623647894e-06,
      "loss": 0.7415,
      "step": 7003
    },
    {
      "epoch": 0.2902731153383895,
      "grad_norm": 0.4094456732273102,
      "learning_rate": 3.5488416428364216e-06,
      "loss": 0.7119,
      "step": 7004
    },
    {
      "epoch": 0.2903145592440632,
      "grad_norm": 0.42484256625175476,
      "learning_rate": 3.5486344233080526e-06,
      "loss": 0.7241,
      "step": 7005
    },
    {
      "epoch": 0.2903560031497368,
      "grad_norm": 0.40012502670288086,
      "learning_rate": 3.548427203779685e-06,
      "loss": 0.6891,
      "step": 7006
    },
    {
      "epoch": 0.2903974470554105,
      "grad_norm": 0.41363972425460815,
      "learning_rate": 3.5482199842513158e-06,
      "loss": 0.7004,
      "step": 7007
    },
    {
      "epoch": 0.29043889096108416,
      "grad_norm": 0.4156229794025421,
      "learning_rate": 3.548012764722948e-06,
      "loss": 0.7446,
      "step": 7008
    },
    {
      "epoch": 0.29048033486675784,
      "grad_norm": 0.4352450966835022,
      "learning_rate": 3.5478055451945794e-06,
      "loss": 0.7249,
      "step": 7009
    },
    {
      "epoch": 0.2905217787724315,
      "grad_norm": 0.44006669521331787,
      "learning_rate": 3.547598325666211e-06,
      "loss": 0.7742,
      "step": 7010
    },
    {
      "epoch": 0.2905632226781052,
      "grad_norm": 0.40405523777008057,
      "learning_rate": 3.547391106137843e-06,
      "loss": 0.7136,
      "step": 7011
    },
    {
      "epoch": 0.29060466658377887,
      "grad_norm": 0.4279027581214905,
      "learning_rate": 3.5471838866094744e-06,
      "loss": 0.6897,
      "step": 7012
    },
    {
      "epoch": 0.29064611048945255,
      "grad_norm": 0.3949183523654938,
      "learning_rate": 3.546976667081106e-06,
      "loss": 0.6681,
      "step": 7013
    },
    {
      "epoch": 0.2906875543951262,
      "grad_norm": 0.3945504426956177,
      "learning_rate": 3.5467694475527376e-06,
      "loss": 0.6997,
      "step": 7014
    },
    {
      "epoch": 0.29072899830079985,
      "grad_norm": 0.4481295347213745,
      "learning_rate": 3.5465622280243694e-06,
      "loss": 0.6914,
      "step": 7015
    },
    {
      "epoch": 0.2907704422064735,
      "grad_norm": 0.4236062169075012,
      "learning_rate": 3.5463550084960008e-06,
      "loss": 0.7092,
      "step": 7016
    },
    {
      "epoch": 0.2908118861121472,
      "grad_norm": 0.4002746343612671,
      "learning_rate": 3.5461477889676326e-06,
      "loss": 0.6602,
      "step": 7017
    },
    {
      "epoch": 0.2908533300178209,
      "grad_norm": 0.40735870599746704,
      "learning_rate": 3.545940569439264e-06,
      "loss": 0.7043,
      "step": 7018
    },
    {
      "epoch": 0.29089477392349455,
      "grad_norm": 0.4325539767742157,
      "learning_rate": 3.545733349910896e-06,
      "loss": 0.6782,
      "step": 7019
    },
    {
      "epoch": 0.29093621782916823,
      "grad_norm": 0.41509175300598145,
      "learning_rate": 3.5455261303825276e-06,
      "loss": 0.7207,
      "step": 7020
    },
    {
      "epoch": 0.2909776617348419,
      "grad_norm": 0.48029980063438416,
      "learning_rate": 3.545318910854159e-06,
      "loss": 0.745,
      "step": 7021
    },
    {
      "epoch": 0.2910191056405156,
      "grad_norm": 0.49317941069602966,
      "learning_rate": 3.5451116913257912e-06,
      "loss": 0.7328,
      "step": 7022
    },
    {
      "epoch": 0.2910605495461892,
      "grad_norm": 0.362125039100647,
      "learning_rate": 3.544904471797422e-06,
      "loss": 0.6892,
      "step": 7023
    },
    {
      "epoch": 0.2911019934518629,
      "grad_norm": 0.45957720279693604,
      "learning_rate": 3.5446972522690544e-06,
      "loss": 0.7144,
      "step": 7024
    },
    {
      "epoch": 0.29114343735753656,
      "grad_norm": 0.410138338804245,
      "learning_rate": 3.544490032740686e-06,
      "loss": 0.7266,
      "step": 7025
    },
    {
      "epoch": 0.29118488126321024,
      "grad_norm": 0.4007231593132019,
      "learning_rate": 3.5442828132123176e-06,
      "loss": 0.6771,
      "step": 7026
    },
    {
      "epoch": 0.2912263251688839,
      "grad_norm": 0.41032305359840393,
      "learning_rate": 3.544075593683949e-06,
      "loss": 0.7388,
      "step": 7027
    },
    {
      "epoch": 0.2912677690745576,
      "grad_norm": 0.4154440760612488,
      "learning_rate": 3.543868374155581e-06,
      "loss": 0.7224,
      "step": 7028
    },
    {
      "epoch": 0.29130921298023127,
      "grad_norm": 0.4457628130912781,
      "learning_rate": 3.5436611546272126e-06,
      "loss": 0.7827,
      "step": 7029
    },
    {
      "epoch": 0.29135065688590495,
      "grad_norm": 0.4295470714569092,
      "learning_rate": 3.543453935098844e-06,
      "loss": 0.7007,
      "step": 7030
    },
    {
      "epoch": 0.2913921007915786,
      "grad_norm": 0.43864497542381287,
      "learning_rate": 3.543246715570476e-06,
      "loss": 0.7307,
      "step": 7031
    },
    {
      "epoch": 0.29143354469725224,
      "grad_norm": 0.4205676019191742,
      "learning_rate": 3.543039496042107e-06,
      "loss": 0.7283,
      "step": 7032
    },
    {
      "epoch": 0.2914749886029259,
      "grad_norm": 0.4613238275051117,
      "learning_rate": 3.542832276513739e-06,
      "loss": 0.7434,
      "step": 7033
    },
    {
      "epoch": 0.2915164325085996,
      "grad_norm": 0.41564393043518066,
      "learning_rate": 3.5426250569853704e-06,
      "loss": 0.7327,
      "step": 7034
    },
    {
      "epoch": 0.2915578764142733,
      "grad_norm": 0.41695675253868103,
      "learning_rate": 3.542417837457002e-06,
      "loss": 0.7327,
      "step": 7035
    },
    {
      "epoch": 0.29159932031994695,
      "grad_norm": 0.4341282248497009,
      "learning_rate": 3.5422106179286336e-06,
      "loss": 0.7039,
      "step": 7036
    },
    {
      "epoch": 0.29164076422562063,
      "grad_norm": 0.4031048119068146,
      "learning_rate": 3.5420033984002654e-06,
      "loss": 0.717,
      "step": 7037
    },
    {
      "epoch": 0.2916822081312943,
      "grad_norm": 0.4074817895889282,
      "learning_rate": 3.5417961788718976e-06,
      "loss": 0.6619,
      "step": 7038
    },
    {
      "epoch": 0.291723652036968,
      "grad_norm": 0.4168263375759125,
      "learning_rate": 3.5415889593435286e-06,
      "loss": 0.6832,
      "step": 7039
    },
    {
      "epoch": 0.29176509594264166,
      "grad_norm": 0.3894229829311371,
      "learning_rate": 3.541381739815161e-06,
      "loss": 0.717,
      "step": 7040
    },
    {
      "epoch": 0.2918065398483153,
      "grad_norm": 0.45080697536468506,
      "learning_rate": 3.5411745202867918e-06,
      "loss": 0.7322,
      "step": 7041
    },
    {
      "epoch": 0.29184798375398896,
      "grad_norm": 0.4704478681087494,
      "learning_rate": 3.540967300758424e-06,
      "loss": 0.7319,
      "step": 7042
    },
    {
      "epoch": 0.29188942765966264,
      "grad_norm": 0.45911282300949097,
      "learning_rate": 3.5407600812300554e-06,
      "loss": 0.7681,
      "step": 7043
    },
    {
      "epoch": 0.2919308715653363,
      "grad_norm": 0.4031442701816559,
      "learning_rate": 3.540552861701687e-06,
      "loss": 0.7283,
      "step": 7044
    },
    {
      "epoch": 0.29197231547101,
      "grad_norm": 0.41739749908447266,
      "learning_rate": 3.5403456421733186e-06,
      "loss": 0.6736,
      "step": 7045
    },
    {
      "epoch": 0.29201375937668367,
      "grad_norm": 0.41814297437667847,
      "learning_rate": 3.5401384226449504e-06,
      "loss": 0.6582,
      "step": 7046
    },
    {
      "epoch": 0.29205520328235735,
      "grad_norm": 0.46693477034568787,
      "learning_rate": 3.539931203116582e-06,
      "loss": 0.705,
      "step": 7047
    },
    {
      "epoch": 0.292096647188031,
      "grad_norm": 0.43323516845703125,
      "learning_rate": 3.5397239835882136e-06,
      "loss": 0.7058,
      "step": 7048
    },
    {
      "epoch": 0.29213809109370464,
      "grad_norm": 0.4054799973964691,
      "learning_rate": 3.5395167640598454e-06,
      "loss": 0.6619,
      "step": 7049
    },
    {
      "epoch": 0.2921795349993783,
      "grad_norm": 0.39226123690605164,
      "learning_rate": 3.539309544531477e-06,
      "loss": 0.741,
      "step": 7050
    },
    {
      "epoch": 0.292220978905052,
      "grad_norm": 0.39724302291870117,
      "learning_rate": 3.5391023250031086e-06,
      "loss": 0.6929,
      "step": 7051
    },
    {
      "epoch": 0.2922624228107257,
      "grad_norm": 0.41185876727104187,
      "learning_rate": 3.53889510547474e-06,
      "loss": 0.6763,
      "step": 7052
    },
    {
      "epoch": 0.29230386671639935,
      "grad_norm": 0.4077242314815521,
      "learning_rate": 3.538687885946372e-06,
      "loss": 0.6719,
      "step": 7053
    },
    {
      "epoch": 0.29234531062207303,
      "grad_norm": 0.4618271291255951,
      "learning_rate": 3.538480666418003e-06,
      "loss": 0.7434,
      "step": 7054
    },
    {
      "epoch": 0.2923867545277467,
      "grad_norm": 0.4621109068393707,
      "learning_rate": 3.538273446889635e-06,
      "loss": 0.7534,
      "step": 7055
    },
    {
      "epoch": 0.2924281984334204,
      "grad_norm": 0.45911550521850586,
      "learning_rate": 3.5380662273612672e-06,
      "loss": 0.6941,
      "step": 7056
    },
    {
      "epoch": 0.29246964233909406,
      "grad_norm": 0.4212634265422821,
      "learning_rate": 3.537859007832898e-06,
      "loss": 0.6725,
      "step": 7057
    },
    {
      "epoch": 0.2925110862447677,
      "grad_norm": 0.4052969813346863,
      "learning_rate": 3.5376517883045304e-06,
      "loss": 0.6921,
      "step": 7058
    },
    {
      "epoch": 0.29255253015044136,
      "grad_norm": 0.40122491121292114,
      "learning_rate": 3.537444568776162e-06,
      "loss": 0.7219,
      "step": 7059
    },
    {
      "epoch": 0.29259397405611504,
      "grad_norm": 0.40673404932022095,
      "learning_rate": 3.5372373492477936e-06,
      "loss": 0.6597,
      "step": 7060
    },
    {
      "epoch": 0.2926354179617887,
      "grad_norm": 0.43931958079338074,
      "learning_rate": 3.537030129719425e-06,
      "loss": 0.6665,
      "step": 7061
    },
    {
      "epoch": 0.2926768618674624,
      "grad_norm": 0.4202190935611725,
      "learning_rate": 3.536822910191057e-06,
      "loss": 0.6526,
      "step": 7062
    },
    {
      "epoch": 0.29271830577313607,
      "grad_norm": 0.4012390375137329,
      "learning_rate": 3.5366156906626886e-06,
      "loss": 0.6934,
      "step": 7063
    },
    {
      "epoch": 0.29275974967880974,
      "grad_norm": 0.4491875469684601,
      "learning_rate": 3.53640847113432e-06,
      "loss": 0.7014,
      "step": 7064
    },
    {
      "epoch": 0.2928011935844834,
      "grad_norm": 0.44229698181152344,
      "learning_rate": 3.536201251605952e-06,
      "loss": 0.7367,
      "step": 7065
    },
    {
      "epoch": 0.2928426374901571,
      "grad_norm": 0.4277951717376709,
      "learning_rate": 3.535994032077583e-06,
      "loss": 0.6821,
      "step": 7066
    },
    {
      "epoch": 0.2928840813958307,
      "grad_norm": 0.40657714009284973,
      "learning_rate": 3.535786812549215e-06,
      "loss": 0.6873,
      "step": 7067
    },
    {
      "epoch": 0.2929255253015044,
      "grad_norm": 0.42000019550323486,
      "learning_rate": 3.5355795930208464e-06,
      "loss": 0.7175,
      "step": 7068
    },
    {
      "epoch": 0.2929669692071781,
      "grad_norm": 0.42083606123924255,
      "learning_rate": 3.535372373492478e-06,
      "loss": 0.7001,
      "step": 7069
    },
    {
      "epoch": 0.29300841311285175,
      "grad_norm": 0.41385912895202637,
      "learning_rate": 3.5351651539641096e-06,
      "loss": 0.6873,
      "step": 7070
    },
    {
      "epoch": 0.29304985701852543,
      "grad_norm": 0.440414160490036,
      "learning_rate": 3.5349579344357414e-06,
      "loss": 0.7217,
      "step": 7071
    },
    {
      "epoch": 0.2930913009241991,
      "grad_norm": 0.3917098343372345,
      "learning_rate": 3.5347507149073736e-06,
      "loss": 0.7383,
      "step": 7072
    },
    {
      "epoch": 0.2931327448298728,
      "grad_norm": 0.4158962666988373,
      "learning_rate": 3.5345434953790046e-06,
      "loss": 0.7181,
      "step": 7073
    },
    {
      "epoch": 0.29317418873554646,
      "grad_norm": 0.4064818024635315,
      "learning_rate": 3.534336275850637e-06,
      "loss": 0.7064,
      "step": 7074
    },
    {
      "epoch": 0.2932156326412201,
      "grad_norm": 0.38659265637397766,
      "learning_rate": 3.5341290563222678e-06,
      "loss": 0.6716,
      "step": 7075
    },
    {
      "epoch": 0.29325707654689376,
      "grad_norm": 0.42484250664711,
      "learning_rate": 3.5339218367939e-06,
      "loss": 0.6854,
      "step": 7076
    },
    {
      "epoch": 0.29329852045256743,
      "grad_norm": 0.40913817286491394,
      "learning_rate": 3.5337146172655314e-06,
      "loss": 0.6979,
      "step": 7077
    },
    {
      "epoch": 0.2933399643582411,
      "grad_norm": 0.42239293456077576,
      "learning_rate": 3.5335073977371632e-06,
      "loss": 0.7402,
      "step": 7078
    },
    {
      "epoch": 0.2933814082639148,
      "grad_norm": 0.44987887144088745,
      "learning_rate": 3.5333001782087946e-06,
      "loss": 0.7437,
      "step": 7079
    },
    {
      "epoch": 0.29342285216958847,
      "grad_norm": 0.4089981019496918,
      "learning_rate": 3.5330929586804264e-06,
      "loss": 0.7224,
      "step": 7080
    },
    {
      "epoch": 0.29346429607526214,
      "grad_norm": 0.4088749885559082,
      "learning_rate": 3.5328857391520582e-06,
      "loss": 0.6968,
      "step": 7081
    },
    {
      "epoch": 0.2935057399809358,
      "grad_norm": 0.4087159335613251,
      "learning_rate": 3.5326785196236896e-06,
      "loss": 0.7311,
      "step": 7082
    },
    {
      "epoch": 0.2935471838866095,
      "grad_norm": 0.44354018568992615,
      "learning_rate": 3.5324713000953214e-06,
      "loss": 0.7124,
      "step": 7083
    },
    {
      "epoch": 0.2935886277922831,
      "grad_norm": 0.4237080514431,
      "learning_rate": 3.532264080566953e-06,
      "loss": 0.666,
      "step": 7084
    },
    {
      "epoch": 0.2936300716979568,
      "grad_norm": 0.4014838635921478,
      "learning_rate": 3.5320568610385846e-06,
      "loss": 0.6754,
      "step": 7085
    },
    {
      "epoch": 0.2936715156036305,
      "grad_norm": 0.4048781096935272,
      "learning_rate": 3.531849641510216e-06,
      "loss": 0.6864,
      "step": 7086
    },
    {
      "epoch": 0.29371295950930415,
      "grad_norm": 0.44869160652160645,
      "learning_rate": 3.531642421981848e-06,
      "loss": 0.6904,
      "step": 7087
    },
    {
      "epoch": 0.2937544034149778,
      "grad_norm": 0.44338735938072205,
      "learning_rate": 3.531435202453479e-06,
      "loss": 0.7227,
      "step": 7088
    },
    {
      "epoch": 0.2937958473206515,
      "grad_norm": 0.44682776927948,
      "learning_rate": 3.531227982925111e-06,
      "loss": 0.7411,
      "step": 7089
    },
    {
      "epoch": 0.2938372912263252,
      "grad_norm": 0.42456531524658203,
      "learning_rate": 3.5310207633967432e-06,
      "loss": 0.7163,
      "step": 7090
    },
    {
      "epoch": 0.29387873513199886,
      "grad_norm": 0.44708770513534546,
      "learning_rate": 3.530813543868374e-06,
      "loss": 0.6919,
      "step": 7091
    },
    {
      "epoch": 0.29392017903767254,
      "grad_norm": 0.41836851835250854,
      "learning_rate": 3.5306063243400064e-06,
      "loss": 0.7162,
      "step": 7092
    },
    {
      "epoch": 0.29396162294334616,
      "grad_norm": 0.4414134621620178,
      "learning_rate": 3.530399104811638e-06,
      "loss": 0.7686,
      "step": 7093
    },
    {
      "epoch": 0.29400306684901983,
      "grad_norm": 0.4520927965641022,
      "learning_rate": 3.5301918852832696e-06,
      "loss": 0.7322,
      "step": 7094
    },
    {
      "epoch": 0.2940445107546935,
      "grad_norm": 0.43770256638526917,
      "learning_rate": 3.529984665754901e-06,
      "loss": 0.7314,
      "step": 7095
    },
    {
      "epoch": 0.2940859546603672,
      "grad_norm": 0.44659924507141113,
      "learning_rate": 3.529777446226533e-06,
      "loss": 0.6968,
      "step": 7096
    },
    {
      "epoch": 0.29412739856604087,
      "grad_norm": 0.46751853823661804,
      "learning_rate": 3.529570226698164e-06,
      "loss": 0.7615,
      "step": 7097
    },
    {
      "epoch": 0.29416884247171454,
      "grad_norm": 0.4149044156074524,
      "learning_rate": 3.529363007169796e-06,
      "loss": 0.73,
      "step": 7098
    },
    {
      "epoch": 0.2942102863773882,
      "grad_norm": 0.4128616750240326,
      "learning_rate": 3.529155787641428e-06,
      "loss": 0.7405,
      "step": 7099
    },
    {
      "epoch": 0.2942517302830619,
      "grad_norm": 0.4283026456832886,
      "learning_rate": 3.528948568113059e-06,
      "loss": 0.6963,
      "step": 7100
    },
    {
      "epoch": 0.2942931741887356,
      "grad_norm": 0.4424697160720825,
      "learning_rate": 3.528741348584691e-06,
      "loss": 0.7268,
      "step": 7101
    },
    {
      "epoch": 0.2943346180944092,
      "grad_norm": 0.46526622772216797,
      "learning_rate": 3.5285341290563224e-06,
      "loss": 0.793,
      "step": 7102
    },
    {
      "epoch": 0.29437606200008287,
      "grad_norm": 0.40357786417007446,
      "learning_rate": 3.528326909527954e-06,
      "loss": 0.6758,
      "step": 7103
    },
    {
      "epoch": 0.29441750590575655,
      "grad_norm": 0.391897052526474,
      "learning_rate": 3.5281196899995856e-06,
      "loss": 0.729,
      "step": 7104
    },
    {
      "epoch": 0.2944589498114302,
      "grad_norm": 0.39721566438674927,
      "learning_rate": 3.5279124704712174e-06,
      "loss": 0.7028,
      "step": 7105
    },
    {
      "epoch": 0.2945003937171039,
      "grad_norm": 0.41112610697746277,
      "learning_rate": 3.5277052509428488e-06,
      "loss": 0.7286,
      "step": 7106
    },
    {
      "epoch": 0.2945418376227776,
      "grad_norm": 0.385834664106369,
      "learning_rate": 3.5274980314144806e-06,
      "loss": 0.6736,
      "step": 7107
    },
    {
      "epoch": 0.29458328152845126,
      "grad_norm": 0.45646584033966064,
      "learning_rate": 3.527290811886113e-06,
      "loss": 0.7043,
      "step": 7108
    },
    {
      "epoch": 0.29462472543412493,
      "grad_norm": 0.3908303678035736,
      "learning_rate": 3.527083592357744e-06,
      "loss": 0.7007,
      "step": 7109
    },
    {
      "epoch": 0.29466616933979856,
      "grad_norm": 0.4678192734718323,
      "learning_rate": 3.526876372829376e-06,
      "loss": 0.7385,
      "step": 7110
    },
    {
      "epoch": 0.29470761324547223,
      "grad_norm": 0.4291054308414459,
      "learning_rate": 3.5266691533010074e-06,
      "loss": 0.7261,
      "step": 7111
    },
    {
      "epoch": 0.2947490571511459,
      "grad_norm": 0.4021909236907959,
      "learning_rate": 3.5264619337726392e-06,
      "loss": 0.6973,
      "step": 7112
    },
    {
      "epoch": 0.2947905010568196,
      "grad_norm": 0.4399035573005676,
      "learning_rate": 3.5262547142442706e-06,
      "loss": 0.7483,
      "step": 7113
    },
    {
      "epoch": 0.29483194496249326,
      "grad_norm": 0.4211803674697876,
      "learning_rate": 3.5260474947159024e-06,
      "loss": 0.6836,
      "step": 7114
    },
    {
      "epoch": 0.29487338886816694,
      "grad_norm": 0.40034130215644836,
      "learning_rate": 3.525840275187534e-06,
      "loss": 0.6674,
      "step": 7115
    },
    {
      "epoch": 0.2949148327738406,
      "grad_norm": 0.44668257236480713,
      "learning_rate": 3.5256330556591656e-06,
      "loss": 0.7388,
      "step": 7116
    },
    {
      "epoch": 0.2949562766795143,
      "grad_norm": 0.463304340839386,
      "learning_rate": 3.5254258361307974e-06,
      "loss": 0.7444,
      "step": 7117
    },
    {
      "epoch": 0.294997720585188,
      "grad_norm": 0.3869861364364624,
      "learning_rate": 3.525218616602429e-06,
      "loss": 0.75,
      "step": 7118
    },
    {
      "epoch": 0.2950391644908616,
      "grad_norm": 0.44453781843185425,
      "learning_rate": 3.5250113970740606e-06,
      "loss": 0.6742,
      "step": 7119
    },
    {
      "epoch": 0.29508060839653527,
      "grad_norm": 0.42569708824157715,
      "learning_rate": 3.524804177545692e-06,
      "loss": 0.6909,
      "step": 7120
    },
    {
      "epoch": 0.29512205230220895,
      "grad_norm": 0.40034160017967224,
      "learning_rate": 3.524596958017324e-06,
      "loss": 0.7253,
      "step": 7121
    },
    {
      "epoch": 0.2951634962078826,
      "grad_norm": 0.4612511098384857,
      "learning_rate": 3.524389738488955e-06,
      "loss": 0.7583,
      "step": 7122
    },
    {
      "epoch": 0.2952049401135563,
      "grad_norm": 0.45482492446899414,
      "learning_rate": 3.524182518960587e-06,
      "loss": 0.7405,
      "step": 7123
    },
    {
      "epoch": 0.29524638401923,
      "grad_norm": 0.42838335037231445,
      "learning_rate": 3.5239752994322192e-06,
      "loss": 0.739,
      "step": 7124
    },
    {
      "epoch": 0.29528782792490366,
      "grad_norm": 0.40979689359664917,
      "learning_rate": 3.52376807990385e-06,
      "loss": 0.7202,
      "step": 7125
    },
    {
      "epoch": 0.29532927183057733,
      "grad_norm": 0.44307535886764526,
      "learning_rate": 3.5235608603754824e-06,
      "loss": 0.6858,
      "step": 7126
    },
    {
      "epoch": 0.295370715736251,
      "grad_norm": 0.3891412913799286,
      "learning_rate": 3.523353640847114e-06,
      "loss": 0.6938,
      "step": 7127
    },
    {
      "epoch": 0.29541215964192463,
      "grad_norm": 0.4264249801635742,
      "learning_rate": 3.5231464213187456e-06,
      "loss": 0.709,
      "step": 7128
    },
    {
      "epoch": 0.2954536035475983,
      "grad_norm": 0.3946525454521179,
      "learning_rate": 3.522939201790377e-06,
      "loss": 0.7429,
      "step": 7129
    },
    {
      "epoch": 0.295495047453272,
      "grad_norm": 0.4395090341567993,
      "learning_rate": 3.522731982262009e-06,
      "loss": 0.7383,
      "step": 7130
    },
    {
      "epoch": 0.29553649135894566,
      "grad_norm": 0.4664108157157898,
      "learning_rate": 3.52252476273364e-06,
      "loss": 0.7225,
      "step": 7131
    },
    {
      "epoch": 0.29557793526461934,
      "grad_norm": 0.4124675393104553,
      "learning_rate": 3.522317543205272e-06,
      "loss": 0.7067,
      "step": 7132
    },
    {
      "epoch": 0.295619379170293,
      "grad_norm": 0.4005729854106903,
      "learning_rate": 3.522110323676904e-06,
      "loss": 0.6813,
      "step": 7133
    },
    {
      "epoch": 0.2956608230759667,
      "grad_norm": 0.4342549443244934,
      "learning_rate": 3.521903104148535e-06,
      "loss": 0.7051,
      "step": 7134
    },
    {
      "epoch": 0.29570226698164037,
      "grad_norm": 0.4355202317237854,
      "learning_rate": 3.521695884620167e-06,
      "loss": 0.7397,
      "step": 7135
    },
    {
      "epoch": 0.295743710887314,
      "grad_norm": 0.41203951835632324,
      "learning_rate": 3.5214886650917984e-06,
      "loss": 0.6936,
      "step": 7136
    },
    {
      "epoch": 0.29578515479298767,
      "grad_norm": 0.41456541419029236,
      "learning_rate": 3.5212814455634302e-06,
      "loss": 0.7223,
      "step": 7137
    },
    {
      "epoch": 0.29582659869866135,
      "grad_norm": 0.40653249621391296,
      "learning_rate": 3.5210742260350616e-06,
      "loss": 0.6565,
      "step": 7138
    },
    {
      "epoch": 0.295868042604335,
      "grad_norm": 0.44600874185562134,
      "learning_rate": 3.5208670065066934e-06,
      "loss": 0.7428,
      "step": 7139
    },
    {
      "epoch": 0.2959094865100087,
      "grad_norm": 0.39629924297332764,
      "learning_rate": 3.520659786978325e-06,
      "loss": 0.7058,
      "step": 7140
    },
    {
      "epoch": 0.2959509304156824,
      "grad_norm": 0.4294497072696686,
      "learning_rate": 3.5204525674499566e-06,
      "loss": 0.6694,
      "step": 7141
    },
    {
      "epoch": 0.29599237432135606,
      "grad_norm": 0.41492316126823425,
      "learning_rate": 3.520245347921589e-06,
      "loss": 0.7421,
      "step": 7142
    },
    {
      "epoch": 0.29603381822702973,
      "grad_norm": 0.43981289863586426,
      "learning_rate": 3.5200381283932202e-06,
      "loss": 0.7219,
      "step": 7143
    },
    {
      "epoch": 0.2960752621327034,
      "grad_norm": 0.4086593985557556,
      "learning_rate": 3.519830908864852e-06,
      "loss": 0.7058,
      "step": 7144
    },
    {
      "epoch": 0.29611670603837703,
      "grad_norm": 0.4153338074684143,
      "learning_rate": 3.5196236893364834e-06,
      "loss": 0.7452,
      "step": 7145
    },
    {
      "epoch": 0.2961581499440507,
      "grad_norm": 0.46243083477020264,
      "learning_rate": 3.5194164698081152e-06,
      "loss": 0.7588,
      "step": 7146
    },
    {
      "epoch": 0.2961995938497244,
      "grad_norm": 0.40634676814079285,
      "learning_rate": 3.5192092502797466e-06,
      "loss": 0.6588,
      "step": 7147
    },
    {
      "epoch": 0.29624103775539806,
      "grad_norm": 0.42810526490211487,
      "learning_rate": 3.5190020307513784e-06,
      "loss": 0.7185,
      "step": 7148
    },
    {
      "epoch": 0.29628248166107174,
      "grad_norm": 0.43551889061927795,
      "learning_rate": 3.51879481122301e-06,
      "loss": 0.7178,
      "step": 7149
    },
    {
      "epoch": 0.2963239255667454,
      "grad_norm": 0.42054837942123413,
      "learning_rate": 3.5185875916946416e-06,
      "loss": 0.7146,
      "step": 7150
    },
    {
      "epoch": 0.2963653694724191,
      "grad_norm": 0.4027163088321686,
      "learning_rate": 3.5183803721662734e-06,
      "loss": 0.6924,
      "step": 7151
    },
    {
      "epoch": 0.29640681337809277,
      "grad_norm": 0.3875315189361572,
      "learning_rate": 3.518173152637905e-06,
      "loss": 0.6847,
      "step": 7152
    },
    {
      "epoch": 0.29644825728376645,
      "grad_norm": 0.4663865566253662,
      "learning_rate": 3.5179659331095366e-06,
      "loss": 0.7527,
      "step": 7153
    },
    {
      "epoch": 0.29648970118944007,
      "grad_norm": 0.39562007784843445,
      "learning_rate": 3.517758713581168e-06,
      "loss": 0.7437,
      "step": 7154
    },
    {
      "epoch": 0.29653114509511375,
      "grad_norm": 0.46124398708343506,
      "learning_rate": 3.5175514940528e-06,
      "loss": 0.7937,
      "step": 7155
    },
    {
      "epoch": 0.2965725890007874,
      "grad_norm": 0.4048207998275757,
      "learning_rate": 3.517344274524431e-06,
      "loss": 0.623,
      "step": 7156
    },
    {
      "epoch": 0.2966140329064611,
      "grad_norm": 0.43110814690589905,
      "learning_rate": 3.517137054996063e-06,
      "loss": 0.7168,
      "step": 7157
    },
    {
      "epoch": 0.2966554768121348,
      "grad_norm": 0.3786778151988983,
      "learning_rate": 3.5169298354676944e-06,
      "loss": 0.6995,
      "step": 7158
    },
    {
      "epoch": 0.29669692071780845,
      "grad_norm": 0.38718733191490173,
      "learning_rate": 3.516722615939326e-06,
      "loss": 0.6295,
      "step": 7159
    },
    {
      "epoch": 0.29673836462348213,
      "grad_norm": 0.41272595524787903,
      "learning_rate": 3.5165153964109584e-06,
      "loss": 0.7036,
      "step": 7160
    },
    {
      "epoch": 0.2967798085291558,
      "grad_norm": 0.4314512312412262,
      "learning_rate": 3.51630817688259e-06,
      "loss": 0.7661,
      "step": 7161
    },
    {
      "epoch": 0.2968212524348295,
      "grad_norm": 0.41737595200538635,
      "learning_rate": 3.5161009573542216e-06,
      "loss": 0.7366,
      "step": 7162
    },
    {
      "epoch": 0.2968626963405031,
      "grad_norm": 0.4425581693649292,
      "learning_rate": 3.515893737825853e-06,
      "loss": 0.772,
      "step": 7163
    },
    {
      "epoch": 0.2969041402461768,
      "grad_norm": 0.4189567267894745,
      "learning_rate": 3.515686518297485e-06,
      "loss": 0.7268,
      "step": 7164
    },
    {
      "epoch": 0.29694558415185046,
      "grad_norm": 0.4525361955165863,
      "learning_rate": 3.515479298769116e-06,
      "loss": 0.7756,
      "step": 7165
    },
    {
      "epoch": 0.29698702805752414,
      "grad_norm": 0.4020131528377533,
      "learning_rate": 3.515272079240748e-06,
      "loss": 0.6965,
      "step": 7166
    },
    {
      "epoch": 0.2970284719631978,
      "grad_norm": 0.4463595449924469,
      "learning_rate": 3.5150648597123794e-06,
      "loss": 0.6985,
      "step": 7167
    },
    {
      "epoch": 0.2970699158688715,
      "grad_norm": 0.4018239974975586,
      "learning_rate": 3.5148576401840112e-06,
      "loss": 0.7004,
      "step": 7168
    },
    {
      "epoch": 0.29711135977454517,
      "grad_norm": 0.40508419275283813,
      "learning_rate": 3.514650420655643e-06,
      "loss": 0.7307,
      "step": 7169
    },
    {
      "epoch": 0.29715280368021885,
      "grad_norm": 0.38697174191474915,
      "learning_rate": 3.5144432011272744e-06,
      "loss": 0.6548,
      "step": 7170
    },
    {
      "epoch": 0.29719424758589247,
      "grad_norm": 0.43206679821014404,
      "learning_rate": 3.5142359815989062e-06,
      "loss": 0.6857,
      "step": 7171
    },
    {
      "epoch": 0.29723569149156615,
      "grad_norm": 0.4318749010562897,
      "learning_rate": 3.5140287620705376e-06,
      "loss": 0.7341,
      "step": 7172
    },
    {
      "epoch": 0.2972771353972398,
      "grad_norm": 0.42490702867507935,
      "learning_rate": 3.5138215425421694e-06,
      "loss": 0.7399,
      "step": 7173
    },
    {
      "epoch": 0.2973185793029135,
      "grad_norm": 0.3991842567920685,
      "learning_rate": 3.513614323013801e-06,
      "loss": 0.666,
      "step": 7174
    },
    {
      "epoch": 0.2973600232085872,
      "grad_norm": 0.3963738977909088,
      "learning_rate": 3.5134071034854326e-06,
      "loss": 0.6902,
      "step": 7175
    },
    {
      "epoch": 0.29740146711426085,
      "grad_norm": 0.4453830420970917,
      "learning_rate": 3.513199883957064e-06,
      "loss": 0.7308,
      "step": 7176
    },
    {
      "epoch": 0.29744291101993453,
      "grad_norm": 0.4318297803401947,
      "learning_rate": 3.5129926644286962e-06,
      "loss": 0.7273,
      "step": 7177
    },
    {
      "epoch": 0.2974843549256082,
      "grad_norm": 0.4356853663921356,
      "learning_rate": 3.512785444900328e-06,
      "loss": 0.7351,
      "step": 7178
    },
    {
      "epoch": 0.2975257988312819,
      "grad_norm": 0.4319220185279846,
      "learning_rate": 3.5125782253719594e-06,
      "loss": 0.7283,
      "step": 7179
    },
    {
      "epoch": 0.2975672427369555,
      "grad_norm": 0.39121752977371216,
      "learning_rate": 3.5123710058435912e-06,
      "loss": 0.6591,
      "step": 7180
    },
    {
      "epoch": 0.2976086866426292,
      "grad_norm": 0.4612372815608978,
      "learning_rate": 3.5121637863152226e-06,
      "loss": 0.7107,
      "step": 7181
    },
    {
      "epoch": 0.29765013054830286,
      "grad_norm": 0.4082043766975403,
      "learning_rate": 3.5119565667868544e-06,
      "loss": 0.7466,
      "step": 7182
    },
    {
      "epoch": 0.29769157445397654,
      "grad_norm": 0.41979148983955383,
      "learning_rate": 3.511749347258486e-06,
      "loss": 0.6819,
      "step": 7183
    },
    {
      "epoch": 0.2977330183596502,
      "grad_norm": 0.38272106647491455,
      "learning_rate": 3.5115421277301176e-06,
      "loss": 0.6716,
      "step": 7184
    },
    {
      "epoch": 0.2977744622653239,
      "grad_norm": 0.42935314774513245,
      "learning_rate": 3.5113349082017494e-06,
      "loss": 0.7081,
      "step": 7185
    },
    {
      "epoch": 0.29781590617099757,
      "grad_norm": 0.39754393696784973,
      "learning_rate": 3.511127688673381e-06,
      "loss": 0.7283,
      "step": 7186
    },
    {
      "epoch": 0.29785735007667125,
      "grad_norm": 0.43284961581230164,
      "learning_rate": 3.5109204691450126e-06,
      "loss": 0.7068,
      "step": 7187
    },
    {
      "epoch": 0.2978987939823449,
      "grad_norm": 0.42626217007637024,
      "learning_rate": 3.510713249616644e-06,
      "loss": 0.6765,
      "step": 7188
    },
    {
      "epoch": 0.29794023788801854,
      "grad_norm": 0.3982051908969879,
      "learning_rate": 3.510506030088276e-06,
      "loss": 0.6838,
      "step": 7189
    },
    {
      "epoch": 0.2979816817936922,
      "grad_norm": 0.4121301472187042,
      "learning_rate": 3.510298810559907e-06,
      "loss": 0.7,
      "step": 7190
    },
    {
      "epoch": 0.2980231256993659,
      "grad_norm": 0.4281240999698639,
      "learning_rate": 3.510091591031539e-06,
      "loss": 0.7727,
      "step": 7191
    },
    {
      "epoch": 0.2980645696050396,
      "grad_norm": 0.44324633479118347,
      "learning_rate": 3.5098843715031704e-06,
      "loss": 0.7478,
      "step": 7192
    },
    {
      "epoch": 0.29810601351071325,
      "grad_norm": 0.4586889445781708,
      "learning_rate": 3.509677151974802e-06,
      "loss": 0.7642,
      "step": 7193
    },
    {
      "epoch": 0.29814745741638693,
      "grad_norm": 0.45172274112701416,
      "learning_rate": 3.5094699324464344e-06,
      "loss": 0.7288,
      "step": 7194
    },
    {
      "epoch": 0.2981889013220606,
      "grad_norm": 0.45836082100868225,
      "learning_rate": 3.509262712918066e-06,
      "loss": 0.7678,
      "step": 7195
    },
    {
      "epoch": 0.2982303452277343,
      "grad_norm": 0.4493940472602844,
      "learning_rate": 3.5090554933896976e-06,
      "loss": 0.7429,
      "step": 7196
    },
    {
      "epoch": 0.2982717891334079,
      "grad_norm": 0.4017223119735718,
      "learning_rate": 3.508848273861329e-06,
      "loss": 0.7212,
      "step": 7197
    },
    {
      "epoch": 0.2983132330390816,
      "grad_norm": 0.4682629704475403,
      "learning_rate": 3.508641054332961e-06,
      "loss": 0.7434,
      "step": 7198
    },
    {
      "epoch": 0.29835467694475526,
      "grad_norm": 0.4443550109863281,
      "learning_rate": 3.5084338348045922e-06,
      "loss": 0.6919,
      "step": 7199
    },
    {
      "epoch": 0.29839612085042894,
      "grad_norm": 0.43907901644706726,
      "learning_rate": 3.508226615276224e-06,
      "loss": 0.7657,
      "step": 7200
    },
    {
      "epoch": 0.2984375647561026,
      "grad_norm": 0.4336368143558502,
      "learning_rate": 3.5080193957478554e-06,
      "loss": 0.7363,
      "step": 7201
    },
    {
      "epoch": 0.2984790086617763,
      "grad_norm": 0.4099287688732147,
      "learning_rate": 3.5078121762194872e-06,
      "loss": 0.7312,
      "step": 7202
    },
    {
      "epoch": 0.29852045256744997,
      "grad_norm": 0.4032919704914093,
      "learning_rate": 3.507604956691119e-06,
      "loss": 0.6934,
      "step": 7203
    },
    {
      "epoch": 0.29856189647312364,
      "grad_norm": 0.39714422821998596,
      "learning_rate": 3.5073977371627504e-06,
      "loss": 0.6987,
      "step": 7204
    },
    {
      "epoch": 0.2986033403787973,
      "grad_norm": 0.3657318353652954,
      "learning_rate": 3.5071905176343822e-06,
      "loss": 0.6366,
      "step": 7205
    },
    {
      "epoch": 0.29864478428447094,
      "grad_norm": 0.4507036805152893,
      "learning_rate": 3.5069832981060136e-06,
      "loss": 0.7117,
      "step": 7206
    },
    {
      "epoch": 0.2986862281901446,
      "grad_norm": 0.40027666091918945,
      "learning_rate": 3.5067760785776454e-06,
      "loss": 0.7222,
      "step": 7207
    },
    {
      "epoch": 0.2987276720958183,
      "grad_norm": 0.39342984557151794,
      "learning_rate": 3.506568859049277e-06,
      "loss": 0.6865,
      "step": 7208
    },
    {
      "epoch": 0.298769116001492,
      "grad_norm": 0.40302762389183044,
      "learning_rate": 3.5063616395209086e-06,
      "loss": 0.6825,
      "step": 7209
    },
    {
      "epoch": 0.29881055990716565,
      "grad_norm": 0.390133261680603,
      "learning_rate": 3.50615441999254e-06,
      "loss": 0.6792,
      "step": 7210
    },
    {
      "epoch": 0.29885200381283933,
      "grad_norm": 0.44662928581237793,
      "learning_rate": 3.5059472004641722e-06,
      "loss": 0.698,
      "step": 7211
    },
    {
      "epoch": 0.298893447718513,
      "grad_norm": 0.4587511122226715,
      "learning_rate": 3.505739980935804e-06,
      "loss": 0.7292,
      "step": 7212
    },
    {
      "epoch": 0.2989348916241867,
      "grad_norm": 0.41628575325012207,
      "learning_rate": 3.5055327614074354e-06,
      "loss": 0.7014,
      "step": 7213
    },
    {
      "epoch": 0.29897633552986036,
      "grad_norm": 0.4412250518798828,
      "learning_rate": 3.5053255418790672e-06,
      "loss": 0.7402,
      "step": 7214
    },
    {
      "epoch": 0.299017779435534,
      "grad_norm": 0.4092228412628174,
      "learning_rate": 3.5051183223506986e-06,
      "loss": 0.7089,
      "step": 7215
    },
    {
      "epoch": 0.29905922334120766,
      "grad_norm": 0.4190283715724945,
      "learning_rate": 3.5049111028223304e-06,
      "loss": 0.7393,
      "step": 7216
    },
    {
      "epoch": 0.29910066724688134,
      "grad_norm": 0.4276978373527527,
      "learning_rate": 3.504703883293962e-06,
      "loss": 0.7036,
      "step": 7217
    },
    {
      "epoch": 0.299142111152555,
      "grad_norm": 0.4398876428604126,
      "learning_rate": 3.5044966637655936e-06,
      "loss": 0.6979,
      "step": 7218
    },
    {
      "epoch": 0.2991835550582287,
      "grad_norm": 0.4226187467575073,
      "learning_rate": 3.504289444237225e-06,
      "loss": 0.6531,
      "step": 7219
    },
    {
      "epoch": 0.29922499896390237,
      "grad_norm": 0.42949700355529785,
      "learning_rate": 3.504082224708857e-06,
      "loss": 0.6934,
      "step": 7220
    },
    {
      "epoch": 0.29926644286957604,
      "grad_norm": 0.4279370605945587,
      "learning_rate": 3.5038750051804886e-06,
      "loss": 0.7605,
      "step": 7221
    },
    {
      "epoch": 0.2993078867752497,
      "grad_norm": 0.41513577103614807,
      "learning_rate": 3.50366778565212e-06,
      "loss": 0.7103,
      "step": 7222
    },
    {
      "epoch": 0.2993493306809234,
      "grad_norm": 0.40499696135520935,
      "learning_rate": 3.503460566123752e-06,
      "loss": 0.6652,
      "step": 7223
    },
    {
      "epoch": 0.299390774586597,
      "grad_norm": 0.3943236768245697,
      "learning_rate": 3.503253346595383e-06,
      "loss": 0.6729,
      "step": 7224
    },
    {
      "epoch": 0.2994322184922707,
      "grad_norm": 0.4183778762817383,
      "learning_rate": 3.503046127067015e-06,
      "loss": 0.6957,
      "step": 7225
    },
    {
      "epoch": 0.2994736623979444,
      "grad_norm": 0.3920329809188843,
      "learning_rate": 3.5028389075386464e-06,
      "loss": 0.6736,
      "step": 7226
    },
    {
      "epoch": 0.29951510630361805,
      "grad_norm": 0.41843655705451965,
      "learning_rate": 3.5026316880102782e-06,
      "loss": 0.7083,
      "step": 7227
    },
    {
      "epoch": 0.2995565502092917,
      "grad_norm": 0.40758731961250305,
      "learning_rate": 3.5024244684819096e-06,
      "loss": 0.6938,
      "step": 7228
    },
    {
      "epoch": 0.2995979941149654,
      "grad_norm": 0.4295382499694824,
      "learning_rate": 3.502217248953542e-06,
      "loss": 0.6652,
      "step": 7229
    },
    {
      "epoch": 0.2996394380206391,
      "grad_norm": 0.41953715682029724,
      "learning_rate": 3.5020100294251736e-06,
      "loss": 0.7487,
      "step": 7230
    },
    {
      "epoch": 0.29968088192631276,
      "grad_norm": 0.40718942880630493,
      "learning_rate": 3.501802809896805e-06,
      "loss": 0.7048,
      "step": 7231
    },
    {
      "epoch": 0.2997223258319864,
      "grad_norm": 0.43632203340530396,
      "learning_rate": 3.501595590368437e-06,
      "loss": 0.7059,
      "step": 7232
    },
    {
      "epoch": 0.29976376973766006,
      "grad_norm": 0.434617817401886,
      "learning_rate": 3.5013883708400682e-06,
      "loss": 0.6945,
      "step": 7233
    },
    {
      "epoch": 0.29980521364333373,
      "grad_norm": 0.4579848647117615,
      "learning_rate": 3.5011811513117e-06,
      "loss": 0.7041,
      "step": 7234
    },
    {
      "epoch": 0.2998466575490074,
      "grad_norm": 0.41240590810775757,
      "learning_rate": 3.5009739317833314e-06,
      "loss": 0.726,
      "step": 7235
    },
    {
      "epoch": 0.2998881014546811,
      "grad_norm": 0.4259759187698364,
      "learning_rate": 3.5007667122549632e-06,
      "loss": 0.7434,
      "step": 7236
    },
    {
      "epoch": 0.29992954536035477,
      "grad_norm": 0.4138964116573334,
      "learning_rate": 3.5005594927265946e-06,
      "loss": 0.6958,
      "step": 7237
    },
    {
      "epoch": 0.29997098926602844,
      "grad_norm": 0.4226977229118347,
      "learning_rate": 3.5003522731982264e-06,
      "loss": 0.7195,
      "step": 7238
    },
    {
      "epoch": 0.3000124331717021,
      "grad_norm": 0.3952260911464691,
      "learning_rate": 3.5001450536698582e-06,
      "loss": 0.6283,
      "step": 7239
    },
    {
      "epoch": 0.3000538770773758,
      "grad_norm": 0.415253609418869,
      "learning_rate": 3.4999378341414896e-06,
      "loss": 0.7053,
      "step": 7240
    },
    {
      "epoch": 0.3000953209830494,
      "grad_norm": 0.5198934078216553,
      "learning_rate": 3.4997306146131214e-06,
      "loss": 0.781,
      "step": 7241
    },
    {
      "epoch": 0.3001367648887231,
      "grad_norm": 0.4069458544254303,
      "learning_rate": 3.499523395084753e-06,
      "loss": 0.7332,
      "step": 7242
    },
    {
      "epoch": 0.3001782087943968,
      "grad_norm": 0.43869447708129883,
      "learning_rate": 3.4993161755563846e-06,
      "loss": 0.7126,
      "step": 7243
    },
    {
      "epoch": 0.30021965270007045,
      "grad_norm": 0.4011303186416626,
      "learning_rate": 3.499108956028016e-06,
      "loss": 0.6982,
      "step": 7244
    },
    {
      "epoch": 0.3002610966057441,
      "grad_norm": 0.3755727708339691,
      "learning_rate": 3.4989017364996482e-06,
      "loss": 0.6907,
      "step": 7245
    },
    {
      "epoch": 0.3003025405114178,
      "grad_norm": 0.507211446762085,
      "learning_rate": 3.49869451697128e-06,
      "loss": 0.707,
      "step": 7246
    },
    {
      "epoch": 0.3003439844170915,
      "grad_norm": 0.4334816038608551,
      "learning_rate": 3.4984872974429114e-06,
      "loss": 0.7046,
      "step": 7247
    },
    {
      "epoch": 0.30038542832276516,
      "grad_norm": 0.4517782926559448,
      "learning_rate": 3.4982800779145432e-06,
      "loss": 0.7109,
      "step": 7248
    },
    {
      "epoch": 0.30042687222843883,
      "grad_norm": 0.45117029547691345,
      "learning_rate": 3.4980728583861746e-06,
      "loss": 0.7703,
      "step": 7249
    },
    {
      "epoch": 0.30046831613411246,
      "grad_norm": 0.4400414526462555,
      "learning_rate": 3.4978656388578064e-06,
      "loss": 0.76,
      "step": 7250
    },
    {
      "epoch": 0.30050976003978613,
      "grad_norm": 0.4033151865005493,
      "learning_rate": 3.497658419329438e-06,
      "loss": 0.7351,
      "step": 7251
    },
    {
      "epoch": 0.3005512039454598,
      "grad_norm": 0.3934030830860138,
      "learning_rate": 3.4974511998010696e-06,
      "loss": 0.6821,
      "step": 7252
    },
    {
      "epoch": 0.3005926478511335,
      "grad_norm": 0.3835507929325104,
      "learning_rate": 3.497243980272701e-06,
      "loss": 0.6487,
      "step": 7253
    },
    {
      "epoch": 0.30063409175680716,
      "grad_norm": 0.5462005138397217,
      "learning_rate": 3.497036760744333e-06,
      "loss": 0.7053,
      "step": 7254
    },
    {
      "epoch": 0.30067553566248084,
      "grad_norm": 0.41436243057250977,
      "learning_rate": 3.4968295412159646e-06,
      "loss": 0.6543,
      "step": 7255
    },
    {
      "epoch": 0.3007169795681545,
      "grad_norm": 0.41642627120018005,
      "learning_rate": 3.496622321687596e-06,
      "loss": 0.6919,
      "step": 7256
    },
    {
      "epoch": 0.3007584234738282,
      "grad_norm": 0.44936898350715637,
      "learning_rate": 3.496415102159228e-06,
      "loss": 0.7178,
      "step": 7257
    },
    {
      "epoch": 0.3007998673795018,
      "grad_norm": 0.43330711126327515,
      "learning_rate": 3.4962078826308592e-06,
      "loss": 0.7036,
      "step": 7258
    },
    {
      "epoch": 0.3008413112851755,
      "grad_norm": 0.4651854634284973,
      "learning_rate": 3.496000663102491e-06,
      "loss": 0.707,
      "step": 7259
    },
    {
      "epoch": 0.30088275519084917,
      "grad_norm": 0.43786242604255676,
      "learning_rate": 3.4957934435741224e-06,
      "loss": 0.7412,
      "step": 7260
    },
    {
      "epoch": 0.30092419909652285,
      "grad_norm": 0.42168518900871277,
      "learning_rate": 3.4955862240457542e-06,
      "loss": 0.7844,
      "step": 7261
    },
    {
      "epoch": 0.3009656430021965,
      "grad_norm": 0.4050391912460327,
      "learning_rate": 3.4953790045173856e-06,
      "loss": 0.688,
      "step": 7262
    },
    {
      "epoch": 0.3010070869078702,
      "grad_norm": 0.45023709535598755,
      "learning_rate": 3.495171784989018e-06,
      "loss": 0.7639,
      "step": 7263
    },
    {
      "epoch": 0.3010485308135439,
      "grad_norm": 0.4256753623485565,
      "learning_rate": 3.4949645654606496e-06,
      "loss": 0.7042,
      "step": 7264
    },
    {
      "epoch": 0.30108997471921756,
      "grad_norm": 0.4467214345932007,
      "learning_rate": 3.494757345932281e-06,
      "loss": 0.7354,
      "step": 7265
    },
    {
      "epoch": 0.30113141862489123,
      "grad_norm": 0.4159295856952667,
      "learning_rate": 3.494550126403913e-06,
      "loss": 0.6807,
      "step": 7266
    },
    {
      "epoch": 0.30117286253056486,
      "grad_norm": 0.40448349714279175,
      "learning_rate": 3.4943429068755442e-06,
      "loss": 0.707,
      "step": 7267
    },
    {
      "epoch": 0.30121430643623853,
      "grad_norm": 0.4183114767074585,
      "learning_rate": 3.494135687347176e-06,
      "loss": 0.6997,
      "step": 7268
    },
    {
      "epoch": 0.3012557503419122,
      "grad_norm": 0.3931894898414612,
      "learning_rate": 3.4939284678188074e-06,
      "loss": 0.6808,
      "step": 7269
    },
    {
      "epoch": 0.3012971942475859,
      "grad_norm": 0.4223577380180359,
      "learning_rate": 3.4937212482904392e-06,
      "loss": 0.6349,
      "step": 7270
    },
    {
      "epoch": 0.30133863815325956,
      "grad_norm": 0.4641520380973816,
      "learning_rate": 3.4935140287620706e-06,
      "loss": 0.6782,
      "step": 7271
    },
    {
      "epoch": 0.30138008205893324,
      "grad_norm": 0.38913285732269287,
      "learning_rate": 3.4933068092337024e-06,
      "loss": 0.7092,
      "step": 7272
    },
    {
      "epoch": 0.3014215259646069,
      "grad_norm": 0.422799676656723,
      "learning_rate": 3.4930995897053342e-06,
      "loss": 0.6954,
      "step": 7273
    },
    {
      "epoch": 0.3014629698702806,
      "grad_norm": 0.4143790602684021,
      "learning_rate": 3.4928923701769656e-06,
      "loss": 0.7059,
      "step": 7274
    },
    {
      "epoch": 0.30150441377595427,
      "grad_norm": 0.43554621934890747,
      "learning_rate": 3.4926851506485974e-06,
      "loss": 0.7664,
      "step": 7275
    },
    {
      "epoch": 0.3015458576816279,
      "grad_norm": 0.41336363554000854,
      "learning_rate": 3.492477931120229e-06,
      "loss": 0.7356,
      "step": 7276
    },
    {
      "epoch": 0.30158730158730157,
      "grad_norm": 0.444933146238327,
      "learning_rate": 3.4922707115918606e-06,
      "loss": 0.7192,
      "step": 7277
    },
    {
      "epoch": 0.30162874549297525,
      "grad_norm": 0.4312666952610016,
      "learning_rate": 3.492063492063492e-06,
      "loss": 0.7051,
      "step": 7278
    },
    {
      "epoch": 0.3016701893986489,
      "grad_norm": 0.4071757197380066,
      "learning_rate": 3.4918562725351242e-06,
      "loss": 0.6497,
      "step": 7279
    },
    {
      "epoch": 0.3017116333043226,
      "grad_norm": 0.41675958037376404,
      "learning_rate": 3.491649053006755e-06,
      "loss": 0.6895,
      "step": 7280
    },
    {
      "epoch": 0.3017530772099963,
      "grad_norm": 0.4068516194820404,
      "learning_rate": 3.4914418334783874e-06,
      "loss": 0.7742,
      "step": 7281
    },
    {
      "epoch": 0.30179452111566996,
      "grad_norm": 0.4245772063732147,
      "learning_rate": 3.4912346139500192e-06,
      "loss": 0.7061,
      "step": 7282
    },
    {
      "epoch": 0.30183596502134363,
      "grad_norm": 0.4134874641895294,
      "learning_rate": 3.4910273944216506e-06,
      "loss": 0.708,
      "step": 7283
    },
    {
      "epoch": 0.30187740892701725,
      "grad_norm": 0.40304699540138245,
      "learning_rate": 3.4908201748932824e-06,
      "loss": 0.6945,
      "step": 7284
    },
    {
      "epoch": 0.30191885283269093,
      "grad_norm": 0.4178912043571472,
      "learning_rate": 3.490612955364914e-06,
      "loss": 0.7441,
      "step": 7285
    },
    {
      "epoch": 0.3019602967383646,
      "grad_norm": 0.41995078325271606,
      "learning_rate": 3.4904057358365456e-06,
      "loss": 0.6744,
      "step": 7286
    },
    {
      "epoch": 0.3020017406440383,
      "grad_norm": 0.4135635197162628,
      "learning_rate": 3.490198516308177e-06,
      "loss": 0.6534,
      "step": 7287
    },
    {
      "epoch": 0.30204318454971196,
      "grad_norm": 0.4651942253112793,
      "learning_rate": 3.489991296779809e-06,
      "loss": 0.7122,
      "step": 7288
    },
    {
      "epoch": 0.30208462845538564,
      "grad_norm": 0.41350629925727844,
      "learning_rate": 3.4897840772514402e-06,
      "loss": 0.7224,
      "step": 7289
    },
    {
      "epoch": 0.3021260723610593,
      "grad_norm": 0.4290896952152252,
      "learning_rate": 3.489576857723072e-06,
      "loss": 0.7522,
      "step": 7290
    },
    {
      "epoch": 0.302167516266733,
      "grad_norm": 0.41058680415153503,
      "learning_rate": 3.489369638194704e-06,
      "loss": 0.6831,
      "step": 7291
    },
    {
      "epoch": 0.30220896017240667,
      "grad_norm": 0.41430723667144775,
      "learning_rate": 3.4891624186663352e-06,
      "loss": 0.7046,
      "step": 7292
    },
    {
      "epoch": 0.3022504040780803,
      "grad_norm": 0.3930942714214325,
      "learning_rate": 3.488955199137967e-06,
      "loss": 0.7231,
      "step": 7293
    },
    {
      "epoch": 0.30229184798375397,
      "grad_norm": 0.44576120376586914,
      "learning_rate": 3.4887479796095984e-06,
      "loss": 0.6772,
      "step": 7294
    },
    {
      "epoch": 0.30233329188942765,
      "grad_norm": 0.4149821698665619,
      "learning_rate": 3.4885407600812302e-06,
      "loss": 0.699,
      "step": 7295
    },
    {
      "epoch": 0.3023747357951013,
      "grad_norm": 0.40366923809051514,
      "learning_rate": 3.4883335405528616e-06,
      "loss": 0.6826,
      "step": 7296
    },
    {
      "epoch": 0.302416179700775,
      "grad_norm": 0.44799765944480896,
      "learning_rate": 3.488126321024494e-06,
      "loss": 0.7537,
      "step": 7297
    },
    {
      "epoch": 0.3024576236064487,
      "grad_norm": 0.424443781375885,
      "learning_rate": 3.487919101496125e-06,
      "loss": 0.6985,
      "step": 7298
    },
    {
      "epoch": 0.30249906751212235,
      "grad_norm": 0.4113139510154724,
      "learning_rate": 3.487711881967757e-06,
      "loss": 0.7191,
      "step": 7299
    },
    {
      "epoch": 0.30254051141779603,
      "grad_norm": 0.43825313448905945,
      "learning_rate": 3.487504662439389e-06,
      "loss": 0.717,
      "step": 7300
    },
    {
      "epoch": 0.3025819553234697,
      "grad_norm": 0.42495718598365784,
      "learning_rate": 3.4872974429110202e-06,
      "loss": 0.7325,
      "step": 7301
    },
    {
      "epoch": 0.30262339922914333,
      "grad_norm": 0.42719680070877075,
      "learning_rate": 3.487090223382652e-06,
      "loss": 0.7278,
      "step": 7302
    },
    {
      "epoch": 0.302664843134817,
      "grad_norm": 0.41216665506362915,
      "learning_rate": 3.4868830038542834e-06,
      "loss": 0.7305,
      "step": 7303
    },
    {
      "epoch": 0.3027062870404907,
      "grad_norm": 0.4190988540649414,
      "learning_rate": 3.4866757843259152e-06,
      "loss": 0.6643,
      "step": 7304
    },
    {
      "epoch": 0.30274773094616436,
      "grad_norm": 0.4264727830886841,
      "learning_rate": 3.4864685647975466e-06,
      "loss": 0.7004,
      "step": 7305
    },
    {
      "epoch": 0.30278917485183804,
      "grad_norm": 0.40712451934814453,
      "learning_rate": 3.4862613452691784e-06,
      "loss": 0.7065,
      "step": 7306
    },
    {
      "epoch": 0.3028306187575117,
      "grad_norm": 0.4183228015899658,
      "learning_rate": 3.4860541257408102e-06,
      "loss": 0.6849,
      "step": 7307
    },
    {
      "epoch": 0.3028720626631854,
      "grad_norm": 0.4363938271999359,
      "learning_rate": 3.4858469062124416e-06,
      "loss": 0.7085,
      "step": 7308
    },
    {
      "epoch": 0.30291350656885907,
      "grad_norm": 0.3858165144920349,
      "learning_rate": 3.4856396866840734e-06,
      "loss": 0.6718,
      "step": 7309
    },
    {
      "epoch": 0.30295495047453275,
      "grad_norm": 0.49259158968925476,
      "learning_rate": 3.485432467155705e-06,
      "loss": 0.7852,
      "step": 7310
    },
    {
      "epoch": 0.30299639438020637,
      "grad_norm": 0.3951805830001831,
      "learning_rate": 3.4852252476273366e-06,
      "loss": 0.6914,
      "step": 7311
    },
    {
      "epoch": 0.30303783828588005,
      "grad_norm": 0.40133509039878845,
      "learning_rate": 3.485018028098968e-06,
      "loss": 0.6826,
      "step": 7312
    },
    {
      "epoch": 0.3030792821915537,
      "grad_norm": 0.36809074878692627,
      "learning_rate": 3.4848108085706003e-06,
      "loss": 0.6423,
      "step": 7313
    },
    {
      "epoch": 0.3031207260972274,
      "grad_norm": 0.4135667383670807,
      "learning_rate": 3.4846035890422312e-06,
      "loss": 0.7102,
      "step": 7314
    },
    {
      "epoch": 0.3031621700029011,
      "grad_norm": 0.40837374329566956,
      "learning_rate": 3.4843963695138634e-06,
      "loss": 0.7173,
      "step": 7315
    },
    {
      "epoch": 0.30320361390857475,
      "grad_norm": 0.4230419993400574,
      "learning_rate": 3.4841891499854953e-06,
      "loss": 0.656,
      "step": 7316
    },
    {
      "epoch": 0.30324505781424843,
      "grad_norm": 0.4288572669029236,
      "learning_rate": 3.4839819304571266e-06,
      "loss": 0.7461,
      "step": 7317
    },
    {
      "epoch": 0.3032865017199221,
      "grad_norm": 0.41343367099761963,
      "learning_rate": 3.4837747109287585e-06,
      "loss": 0.6979,
      "step": 7318
    },
    {
      "epoch": 0.30332794562559573,
      "grad_norm": 0.3919431269168854,
      "learning_rate": 3.48356749140039e-06,
      "loss": 0.6714,
      "step": 7319
    },
    {
      "epoch": 0.3033693895312694,
      "grad_norm": 0.4088151752948761,
      "learning_rate": 3.4833602718720216e-06,
      "loss": 0.7322,
      "step": 7320
    },
    {
      "epoch": 0.3034108334369431,
      "grad_norm": 0.40012434124946594,
      "learning_rate": 3.483153052343653e-06,
      "loss": 0.6758,
      "step": 7321
    },
    {
      "epoch": 0.30345227734261676,
      "grad_norm": 0.39157670736312866,
      "learning_rate": 3.482945832815285e-06,
      "loss": 0.6531,
      "step": 7322
    },
    {
      "epoch": 0.30349372124829044,
      "grad_norm": 0.49350011348724365,
      "learning_rate": 3.4827386132869162e-06,
      "loss": 0.7073,
      "step": 7323
    },
    {
      "epoch": 0.3035351651539641,
      "grad_norm": 0.41279998421669006,
      "learning_rate": 3.482531393758548e-06,
      "loss": 0.7061,
      "step": 7324
    },
    {
      "epoch": 0.3035766090596378,
      "grad_norm": 0.3914352357387543,
      "learning_rate": 3.48232417423018e-06,
      "loss": 0.6938,
      "step": 7325
    },
    {
      "epoch": 0.30361805296531147,
      "grad_norm": 0.41931289434432983,
      "learning_rate": 3.4821169547018112e-06,
      "loss": 0.7234,
      "step": 7326
    },
    {
      "epoch": 0.30365949687098515,
      "grad_norm": 0.4103632867336273,
      "learning_rate": 3.481909735173443e-06,
      "loss": 0.7018,
      "step": 7327
    },
    {
      "epoch": 0.30370094077665877,
      "grad_norm": 0.39839738607406616,
      "learning_rate": 3.4817025156450744e-06,
      "loss": 0.7063,
      "step": 7328
    },
    {
      "epoch": 0.30374238468233244,
      "grad_norm": 0.4106575846672058,
      "learning_rate": 3.4814952961167067e-06,
      "loss": 0.7324,
      "step": 7329
    },
    {
      "epoch": 0.3037838285880061,
      "grad_norm": 0.431655615568161,
      "learning_rate": 3.4812880765883376e-06,
      "loss": 0.6924,
      "step": 7330
    },
    {
      "epoch": 0.3038252724936798,
      "grad_norm": 0.40878739953041077,
      "learning_rate": 3.48108085705997e-06,
      "loss": 0.7122,
      "step": 7331
    },
    {
      "epoch": 0.3038667163993535,
      "grad_norm": 0.4227137565612793,
      "learning_rate": 3.480873637531601e-06,
      "loss": 0.7437,
      "step": 7332
    },
    {
      "epoch": 0.30390816030502715,
      "grad_norm": 0.4016418159008026,
      "learning_rate": 3.480666418003233e-06,
      "loss": 0.6528,
      "step": 7333
    },
    {
      "epoch": 0.30394960421070083,
      "grad_norm": 0.4477592706680298,
      "learning_rate": 3.480459198474865e-06,
      "loss": 0.6904,
      "step": 7334
    },
    {
      "epoch": 0.3039910481163745,
      "grad_norm": 0.42992234230041504,
      "learning_rate": 3.4802519789464962e-06,
      "loss": 0.7183,
      "step": 7335
    },
    {
      "epoch": 0.3040324920220482,
      "grad_norm": 0.4285798966884613,
      "learning_rate": 3.480044759418128e-06,
      "loss": 0.7688,
      "step": 7336
    },
    {
      "epoch": 0.3040739359277218,
      "grad_norm": 0.4552157521247864,
      "learning_rate": 3.4798375398897594e-06,
      "loss": 0.7432,
      "step": 7337
    },
    {
      "epoch": 0.3041153798333955,
      "grad_norm": 0.41803181171417236,
      "learning_rate": 3.4796303203613912e-06,
      "loss": 0.6731,
      "step": 7338
    },
    {
      "epoch": 0.30415682373906916,
      "grad_norm": 0.40412840247154236,
      "learning_rate": 3.4794231008330226e-06,
      "loss": 0.714,
      "step": 7339
    },
    {
      "epoch": 0.30419826764474284,
      "grad_norm": 0.43529561161994934,
      "learning_rate": 3.4792158813046544e-06,
      "loss": 0.7042,
      "step": 7340
    },
    {
      "epoch": 0.3042397115504165,
      "grad_norm": 0.39643725752830505,
      "learning_rate": 3.479008661776286e-06,
      "loss": 0.6719,
      "step": 7341
    },
    {
      "epoch": 0.3042811554560902,
      "grad_norm": 0.4261438548564911,
      "learning_rate": 3.4788014422479176e-06,
      "loss": 0.7139,
      "step": 7342
    },
    {
      "epoch": 0.30432259936176387,
      "grad_norm": 0.40189990401268005,
      "learning_rate": 3.4785942227195494e-06,
      "loss": 0.7135,
      "step": 7343
    },
    {
      "epoch": 0.30436404326743755,
      "grad_norm": 0.4367200434207916,
      "learning_rate": 3.478387003191181e-06,
      "loss": 0.748,
      "step": 7344
    },
    {
      "epoch": 0.30440548717311117,
      "grad_norm": 0.38791799545288086,
      "learning_rate": 3.4781797836628126e-06,
      "loss": 0.6963,
      "step": 7345
    },
    {
      "epoch": 0.30444693107878484,
      "grad_norm": 0.44051676988601685,
      "learning_rate": 3.477972564134444e-06,
      "loss": 0.7262,
      "step": 7346
    },
    {
      "epoch": 0.3044883749844585,
      "grad_norm": 0.43658021092414856,
      "learning_rate": 3.4777653446060763e-06,
      "loss": 0.6653,
      "step": 7347
    },
    {
      "epoch": 0.3045298188901322,
      "grad_norm": 0.48692843317985535,
      "learning_rate": 3.4775581250777072e-06,
      "loss": 0.7354,
      "step": 7348
    },
    {
      "epoch": 0.3045712627958059,
      "grad_norm": 0.41655632853507996,
      "learning_rate": 3.4773509055493395e-06,
      "loss": 0.7336,
      "step": 7349
    },
    {
      "epoch": 0.30461270670147955,
      "grad_norm": 0.38513699173927307,
      "learning_rate": 3.477143686020971e-06,
      "loss": 0.71,
      "step": 7350
    },
    {
      "epoch": 0.30465415060715323,
      "grad_norm": 0.39964449405670166,
      "learning_rate": 3.4769364664926026e-06,
      "loss": 0.6849,
      "step": 7351
    },
    {
      "epoch": 0.3046955945128269,
      "grad_norm": 0.38176125288009644,
      "learning_rate": 3.4767292469642345e-06,
      "loss": 0.6635,
      "step": 7352
    },
    {
      "epoch": 0.3047370384185006,
      "grad_norm": 0.44331446290016174,
      "learning_rate": 3.476522027435866e-06,
      "loss": 0.6855,
      "step": 7353
    },
    {
      "epoch": 0.3047784823241742,
      "grad_norm": 0.44695717096328735,
      "learning_rate": 3.4763148079074977e-06,
      "loss": 0.7458,
      "step": 7354
    },
    {
      "epoch": 0.3048199262298479,
      "grad_norm": 0.40700972080230713,
      "learning_rate": 3.476107588379129e-06,
      "loss": 0.7051,
      "step": 7355
    },
    {
      "epoch": 0.30486137013552156,
      "grad_norm": 0.39625978469848633,
      "learning_rate": 3.475900368850761e-06,
      "loss": 0.6316,
      "step": 7356
    },
    {
      "epoch": 0.30490281404119524,
      "grad_norm": 0.43081000447273254,
      "learning_rate": 3.4756931493223922e-06,
      "loss": 0.7179,
      "step": 7357
    },
    {
      "epoch": 0.3049442579468689,
      "grad_norm": 0.426923006772995,
      "learning_rate": 3.475485929794024e-06,
      "loss": 0.7241,
      "step": 7358
    },
    {
      "epoch": 0.3049857018525426,
      "grad_norm": 0.41886407136917114,
      "learning_rate": 3.475278710265656e-06,
      "loss": 0.6332,
      "step": 7359
    },
    {
      "epoch": 0.30502714575821627,
      "grad_norm": 0.4363551735877991,
      "learning_rate": 3.4750714907372872e-06,
      "loss": 0.7268,
      "step": 7360
    },
    {
      "epoch": 0.30506858966388994,
      "grad_norm": 0.4262523353099823,
      "learning_rate": 3.474864271208919e-06,
      "loss": 0.6624,
      "step": 7361
    },
    {
      "epoch": 0.3051100335695636,
      "grad_norm": 0.4243740737438202,
      "learning_rate": 3.4746570516805504e-06,
      "loss": 0.739,
      "step": 7362
    },
    {
      "epoch": 0.30515147747523724,
      "grad_norm": 0.4773908257484436,
      "learning_rate": 3.4744498321521827e-06,
      "loss": 0.7805,
      "step": 7363
    },
    {
      "epoch": 0.3051929213809109,
      "grad_norm": 0.37355175614356995,
      "learning_rate": 3.4742426126238136e-06,
      "loss": 0.6504,
      "step": 7364
    },
    {
      "epoch": 0.3052343652865846,
      "grad_norm": 0.4607018530368805,
      "learning_rate": 3.474035393095446e-06,
      "loss": 0.7507,
      "step": 7365
    },
    {
      "epoch": 0.3052758091922583,
      "grad_norm": 0.4046276807785034,
      "learning_rate": 3.473828173567077e-06,
      "loss": 0.6639,
      "step": 7366
    },
    {
      "epoch": 0.30531725309793195,
      "grad_norm": 0.4054524600505829,
      "learning_rate": 3.473620954038709e-06,
      "loss": 0.7041,
      "step": 7367
    },
    {
      "epoch": 0.30535869700360563,
      "grad_norm": 0.42565783858299255,
      "learning_rate": 3.473413734510341e-06,
      "loss": 0.7205,
      "step": 7368
    },
    {
      "epoch": 0.3054001409092793,
      "grad_norm": 0.40289342403411865,
      "learning_rate": 3.4732065149819722e-06,
      "loss": 0.6892,
      "step": 7369
    },
    {
      "epoch": 0.305441584814953,
      "grad_norm": 0.4245118498802185,
      "learning_rate": 3.472999295453604e-06,
      "loss": 0.7468,
      "step": 7370
    },
    {
      "epoch": 0.30548302872062666,
      "grad_norm": 0.4195047914981842,
      "learning_rate": 3.4727920759252354e-06,
      "loss": 0.7258,
      "step": 7371
    },
    {
      "epoch": 0.3055244726263003,
      "grad_norm": 0.4067900776863098,
      "learning_rate": 3.4725848563968673e-06,
      "loss": 0.689,
      "step": 7372
    },
    {
      "epoch": 0.30556591653197396,
      "grad_norm": 0.41553857922554016,
      "learning_rate": 3.4723776368684986e-06,
      "loss": 0.6711,
      "step": 7373
    },
    {
      "epoch": 0.30560736043764763,
      "grad_norm": 0.4308988153934479,
      "learning_rate": 3.4721704173401304e-06,
      "loss": 0.7137,
      "step": 7374
    },
    {
      "epoch": 0.3056488043433213,
      "grad_norm": 0.4417855441570282,
      "learning_rate": 3.471963197811762e-06,
      "loss": 0.7781,
      "step": 7375
    },
    {
      "epoch": 0.305690248248995,
      "grad_norm": 0.38502851128578186,
      "learning_rate": 3.4717559782833936e-06,
      "loss": 0.7012,
      "step": 7376
    },
    {
      "epoch": 0.30573169215466867,
      "grad_norm": 0.4085834324359894,
      "learning_rate": 3.4715487587550255e-06,
      "loss": 0.677,
      "step": 7377
    },
    {
      "epoch": 0.30577313606034234,
      "grad_norm": 0.4353998601436615,
      "learning_rate": 3.471341539226657e-06,
      "loss": 0.7164,
      "step": 7378
    },
    {
      "epoch": 0.305814579966016,
      "grad_norm": 0.41055506467819214,
      "learning_rate": 3.4711343196982886e-06,
      "loss": 0.6957,
      "step": 7379
    },
    {
      "epoch": 0.30585602387168964,
      "grad_norm": 0.4064086377620697,
      "learning_rate": 3.47092710016992e-06,
      "loss": 0.7722,
      "step": 7380
    },
    {
      "epoch": 0.3058974677773633,
      "grad_norm": 0.4350390136241913,
      "learning_rate": 3.4707198806415523e-06,
      "loss": 0.6581,
      "step": 7381
    },
    {
      "epoch": 0.305938911683037,
      "grad_norm": 0.4039255380630493,
      "learning_rate": 3.4705126611131832e-06,
      "loss": 0.6798,
      "step": 7382
    },
    {
      "epoch": 0.3059803555887107,
      "grad_norm": 0.41461119055747986,
      "learning_rate": 3.4703054415848155e-06,
      "loss": 0.6838,
      "step": 7383
    },
    {
      "epoch": 0.30602179949438435,
      "grad_norm": 0.41529521346092224,
      "learning_rate": 3.470098222056447e-06,
      "loss": 0.7444,
      "step": 7384
    },
    {
      "epoch": 0.306063243400058,
      "grad_norm": 0.4189828634262085,
      "learning_rate": 3.4698910025280787e-06,
      "loss": 0.7542,
      "step": 7385
    },
    {
      "epoch": 0.3061046873057317,
      "grad_norm": 0.4064129889011383,
      "learning_rate": 3.4696837829997105e-06,
      "loss": 0.7058,
      "step": 7386
    },
    {
      "epoch": 0.3061461312114054,
      "grad_norm": 0.41242772340774536,
      "learning_rate": 3.469476563471342e-06,
      "loss": 0.7073,
      "step": 7387
    },
    {
      "epoch": 0.30618757511707906,
      "grad_norm": 0.4247068762779236,
      "learning_rate": 3.4692693439429737e-06,
      "loss": 0.7131,
      "step": 7388
    },
    {
      "epoch": 0.3062290190227527,
      "grad_norm": 0.4286353886127472,
      "learning_rate": 3.469062124414605e-06,
      "loss": 0.7241,
      "step": 7389
    },
    {
      "epoch": 0.30627046292842636,
      "grad_norm": 0.4292788505554199,
      "learning_rate": 3.468854904886237e-06,
      "loss": 0.6774,
      "step": 7390
    },
    {
      "epoch": 0.30631190683410003,
      "grad_norm": 0.4236053228378296,
      "learning_rate": 3.4686476853578682e-06,
      "loss": 0.7146,
      "step": 7391
    },
    {
      "epoch": 0.3063533507397737,
      "grad_norm": 0.40659981966018677,
      "learning_rate": 3.4684404658295e-06,
      "loss": 0.6735,
      "step": 7392
    },
    {
      "epoch": 0.3063947946454474,
      "grad_norm": 0.4336153566837311,
      "learning_rate": 3.4682332463011314e-06,
      "loss": 0.6946,
      "step": 7393
    },
    {
      "epoch": 0.30643623855112107,
      "grad_norm": 0.46234774589538574,
      "learning_rate": 3.4680260267727632e-06,
      "loss": 0.7273,
      "step": 7394
    },
    {
      "epoch": 0.30647768245679474,
      "grad_norm": 0.41391947865486145,
      "learning_rate": 3.467818807244395e-06,
      "loss": 0.7048,
      "step": 7395
    },
    {
      "epoch": 0.3065191263624684,
      "grad_norm": 0.4224640429019928,
      "learning_rate": 3.4676115877160264e-06,
      "loss": 0.7168,
      "step": 7396
    },
    {
      "epoch": 0.3065605702681421,
      "grad_norm": 0.44375935196876526,
      "learning_rate": 3.4674043681876587e-06,
      "loss": 0.7197,
      "step": 7397
    },
    {
      "epoch": 0.3066020141738157,
      "grad_norm": 0.39048343896865845,
      "learning_rate": 3.4671971486592896e-06,
      "loss": 0.6904,
      "step": 7398
    },
    {
      "epoch": 0.3066434580794894,
      "grad_norm": 0.4068012237548828,
      "learning_rate": 3.466989929130922e-06,
      "loss": 0.709,
      "step": 7399
    },
    {
      "epoch": 0.30668490198516307,
      "grad_norm": 0.4248158037662506,
      "learning_rate": 3.466782709602553e-06,
      "loss": 0.7651,
      "step": 7400
    },
    {
      "epoch": 0.30672634589083675,
      "grad_norm": 0.38290929794311523,
      "learning_rate": 3.466575490074185e-06,
      "loss": 0.6323,
      "step": 7401
    },
    {
      "epoch": 0.3067677897965104,
      "grad_norm": 0.39681991934776306,
      "learning_rate": 3.4663682705458164e-06,
      "loss": 0.6722,
      "step": 7402
    },
    {
      "epoch": 0.3068092337021841,
      "grad_norm": 0.48744985461235046,
      "learning_rate": 3.4661610510174483e-06,
      "loss": 0.7,
      "step": 7403
    },
    {
      "epoch": 0.3068506776078578,
      "grad_norm": 0.4756367802619934,
      "learning_rate": 3.46595383148908e-06,
      "loss": 0.7634,
      "step": 7404
    },
    {
      "epoch": 0.30689212151353146,
      "grad_norm": 0.3931790292263031,
      "learning_rate": 3.4657466119607114e-06,
      "loss": 0.6831,
      "step": 7405
    },
    {
      "epoch": 0.3069335654192051,
      "grad_norm": 0.40859824419021606,
      "learning_rate": 3.4655393924323433e-06,
      "loss": 0.6821,
      "step": 7406
    },
    {
      "epoch": 0.30697500932487876,
      "grad_norm": 0.4261351227760315,
      "learning_rate": 3.4653321729039746e-06,
      "loss": 0.6842,
      "step": 7407
    },
    {
      "epoch": 0.30701645323055243,
      "grad_norm": 0.4002237021923065,
      "learning_rate": 3.4651249533756065e-06,
      "loss": 0.7214,
      "step": 7408
    },
    {
      "epoch": 0.3070578971362261,
      "grad_norm": 0.40172311663627625,
      "learning_rate": 3.464917733847238e-06,
      "loss": 0.6337,
      "step": 7409
    },
    {
      "epoch": 0.3070993410418998,
      "grad_norm": 0.44855910539627075,
      "learning_rate": 3.4647105143188696e-06,
      "loss": 0.7794,
      "step": 7410
    },
    {
      "epoch": 0.30714078494757346,
      "grad_norm": 0.42929014563560486,
      "learning_rate": 3.464503294790501e-06,
      "loss": 0.7444,
      "step": 7411
    },
    {
      "epoch": 0.30718222885324714,
      "grad_norm": 0.38817235827445984,
      "learning_rate": 3.464296075262133e-06,
      "loss": 0.6581,
      "step": 7412
    },
    {
      "epoch": 0.3072236727589208,
      "grad_norm": 0.40362638235092163,
      "learning_rate": 3.4640888557337647e-06,
      "loss": 0.7319,
      "step": 7413
    },
    {
      "epoch": 0.3072651166645945,
      "grad_norm": 0.39868125319480896,
      "learning_rate": 3.463881636205396e-06,
      "loss": 0.645,
      "step": 7414
    },
    {
      "epoch": 0.3073065605702681,
      "grad_norm": 0.3945135176181793,
      "learning_rate": 3.4636744166770283e-06,
      "loss": 0.6547,
      "step": 7415
    },
    {
      "epoch": 0.3073480044759418,
      "grad_norm": 0.43538111448287964,
      "learning_rate": 3.4634671971486592e-06,
      "loss": 0.7075,
      "step": 7416
    },
    {
      "epoch": 0.30738944838161547,
      "grad_norm": 0.4041697084903717,
      "learning_rate": 3.4632599776202915e-06,
      "loss": 0.6976,
      "step": 7417
    },
    {
      "epoch": 0.30743089228728915,
      "grad_norm": 0.4234633147716522,
      "learning_rate": 3.463052758091923e-06,
      "loss": 0.6483,
      "step": 7418
    },
    {
      "epoch": 0.3074723361929628,
      "grad_norm": 0.40138617157936096,
      "learning_rate": 3.4628455385635547e-06,
      "loss": 0.6821,
      "step": 7419
    },
    {
      "epoch": 0.3075137800986365,
      "grad_norm": 0.3986906409263611,
      "learning_rate": 3.4626383190351865e-06,
      "loss": 0.7122,
      "step": 7420
    },
    {
      "epoch": 0.3075552240043102,
      "grad_norm": 0.40693172812461853,
      "learning_rate": 3.462431099506818e-06,
      "loss": 0.7253,
      "step": 7421
    },
    {
      "epoch": 0.30759666790998386,
      "grad_norm": 0.40595442056655884,
      "learning_rate": 3.4622238799784497e-06,
      "loss": 0.6517,
      "step": 7422
    },
    {
      "epoch": 0.30763811181565753,
      "grad_norm": 0.4506520926952362,
      "learning_rate": 3.462016660450081e-06,
      "loss": 0.6719,
      "step": 7423
    },
    {
      "epoch": 0.30767955572133115,
      "grad_norm": 0.4135625660419464,
      "learning_rate": 3.461809440921713e-06,
      "loss": 0.736,
      "step": 7424
    },
    {
      "epoch": 0.30772099962700483,
      "grad_norm": 0.4359142780303955,
      "learning_rate": 3.4616022213933442e-06,
      "loss": 0.6891,
      "step": 7425
    },
    {
      "epoch": 0.3077624435326785,
      "grad_norm": 0.41457247734069824,
      "learning_rate": 3.461395001864976e-06,
      "loss": 0.6521,
      "step": 7426
    },
    {
      "epoch": 0.3078038874383522,
      "grad_norm": 0.3983946144580841,
      "learning_rate": 3.4611877823366074e-06,
      "loss": 0.686,
      "step": 7427
    },
    {
      "epoch": 0.30784533134402586,
      "grad_norm": 0.41855570673942566,
      "learning_rate": 3.4609805628082392e-06,
      "loss": 0.7656,
      "step": 7428
    },
    {
      "epoch": 0.30788677524969954,
      "grad_norm": 0.42434927821159363,
      "learning_rate": 3.460773343279871e-06,
      "loss": 0.7097,
      "step": 7429
    },
    {
      "epoch": 0.3079282191553732,
      "grad_norm": 0.3782005310058594,
      "learning_rate": 3.4605661237515024e-06,
      "loss": 0.6736,
      "step": 7430
    },
    {
      "epoch": 0.3079696630610469,
      "grad_norm": 0.4046684503555298,
      "learning_rate": 3.4603589042231347e-06,
      "loss": 0.6934,
      "step": 7431
    },
    {
      "epoch": 0.30801110696672057,
      "grad_norm": 0.39992770552635193,
      "learning_rate": 3.4601516846947656e-06,
      "loss": 0.6533,
      "step": 7432
    },
    {
      "epoch": 0.3080525508723942,
      "grad_norm": 0.3961031138896942,
      "learning_rate": 3.459944465166398e-06,
      "loss": 0.7212,
      "step": 7433
    },
    {
      "epoch": 0.30809399477806787,
      "grad_norm": 0.4134383201599121,
      "learning_rate": 3.459737245638029e-06,
      "loss": 0.679,
      "step": 7434
    },
    {
      "epoch": 0.30813543868374155,
      "grad_norm": 0.3673158884048462,
      "learning_rate": 3.459530026109661e-06,
      "loss": 0.6405,
      "step": 7435
    },
    {
      "epoch": 0.3081768825894152,
      "grad_norm": 0.41948065161705017,
      "learning_rate": 3.4593228065812925e-06,
      "loss": 0.7307,
      "step": 7436
    },
    {
      "epoch": 0.3082183264950889,
      "grad_norm": 0.4199807643890381,
      "learning_rate": 3.4591155870529243e-06,
      "loss": 0.6355,
      "step": 7437
    },
    {
      "epoch": 0.3082597704007626,
      "grad_norm": 0.4437977075576782,
      "learning_rate": 3.458908367524556e-06,
      "loss": 0.7123,
      "step": 7438
    },
    {
      "epoch": 0.30830121430643626,
      "grad_norm": 0.42537516355514526,
      "learning_rate": 3.4587011479961875e-06,
      "loss": 0.7021,
      "step": 7439
    },
    {
      "epoch": 0.30834265821210993,
      "grad_norm": 0.462179034948349,
      "learning_rate": 3.4584939284678193e-06,
      "loss": 0.7058,
      "step": 7440
    },
    {
      "epoch": 0.30838410211778355,
      "grad_norm": 0.5247779488563538,
      "learning_rate": 3.4582867089394506e-06,
      "loss": 0.7861,
      "step": 7441
    },
    {
      "epoch": 0.30842554602345723,
      "grad_norm": 0.4208361804485321,
      "learning_rate": 3.4580794894110825e-06,
      "loss": 0.6921,
      "step": 7442
    },
    {
      "epoch": 0.3084669899291309,
      "grad_norm": 0.41391265392303467,
      "learning_rate": 3.457872269882714e-06,
      "loss": 0.6554,
      "step": 7443
    },
    {
      "epoch": 0.3085084338348046,
      "grad_norm": 0.4514371454715729,
      "learning_rate": 3.4576650503543457e-06,
      "loss": 0.7461,
      "step": 7444
    },
    {
      "epoch": 0.30854987774047826,
      "grad_norm": 0.43196672201156616,
      "learning_rate": 3.457457830825977e-06,
      "loss": 0.6942,
      "step": 7445
    },
    {
      "epoch": 0.30859132164615194,
      "grad_norm": 0.41849032044410706,
      "learning_rate": 3.457250611297609e-06,
      "loss": 0.7034,
      "step": 7446
    },
    {
      "epoch": 0.3086327655518256,
      "grad_norm": 0.38648521900177,
      "learning_rate": 3.4570433917692407e-06,
      "loss": 0.666,
      "step": 7447
    },
    {
      "epoch": 0.3086742094574993,
      "grad_norm": 0.40261563658714294,
      "learning_rate": 3.456836172240872e-06,
      "loss": 0.6942,
      "step": 7448
    },
    {
      "epoch": 0.30871565336317297,
      "grad_norm": 0.39538857340812683,
      "learning_rate": 3.4566289527125043e-06,
      "loss": 0.6466,
      "step": 7449
    },
    {
      "epoch": 0.3087570972688466,
      "grad_norm": 0.432583212852478,
      "learning_rate": 3.4564217331841352e-06,
      "loss": 0.7451,
      "step": 7450
    },
    {
      "epoch": 0.30879854117452027,
      "grad_norm": 0.4139600098133087,
      "learning_rate": 3.4562145136557675e-06,
      "loss": 0.6829,
      "step": 7451
    },
    {
      "epoch": 0.30883998508019395,
      "grad_norm": 0.3965872824192047,
      "learning_rate": 3.456007294127399e-06,
      "loss": 0.7104,
      "step": 7452
    },
    {
      "epoch": 0.3088814289858676,
      "grad_norm": 0.4589243531227112,
      "learning_rate": 3.4558000745990307e-06,
      "loss": 0.7174,
      "step": 7453
    },
    {
      "epoch": 0.3089228728915413,
      "grad_norm": 0.41212204098701477,
      "learning_rate": 3.455592855070662e-06,
      "loss": 0.7429,
      "step": 7454
    },
    {
      "epoch": 0.308964316797215,
      "grad_norm": 0.41620928049087524,
      "learning_rate": 3.455385635542294e-06,
      "loss": 0.7114,
      "step": 7455
    },
    {
      "epoch": 0.30900576070288865,
      "grad_norm": 0.4345521330833435,
      "learning_rate": 3.4551784160139257e-06,
      "loss": 0.748,
      "step": 7456
    },
    {
      "epoch": 0.30904720460856233,
      "grad_norm": 0.40595024824142456,
      "learning_rate": 3.454971196485557e-06,
      "loss": 0.6959,
      "step": 7457
    },
    {
      "epoch": 0.309088648514236,
      "grad_norm": 0.4046982526779175,
      "learning_rate": 3.454763976957189e-06,
      "loss": 0.7261,
      "step": 7458
    },
    {
      "epoch": 0.30913009241990963,
      "grad_norm": 0.41742590069770813,
      "learning_rate": 3.4545567574288203e-06,
      "loss": 0.7285,
      "step": 7459
    },
    {
      "epoch": 0.3091715363255833,
      "grad_norm": 0.4170762300491333,
      "learning_rate": 3.454349537900452e-06,
      "loss": 0.667,
      "step": 7460
    },
    {
      "epoch": 0.309212980231257,
      "grad_norm": 0.40838125348091125,
      "learning_rate": 3.4541423183720834e-06,
      "loss": 0.6522,
      "step": 7461
    },
    {
      "epoch": 0.30925442413693066,
      "grad_norm": 0.4063316285610199,
      "learning_rate": 3.4539350988437153e-06,
      "loss": 0.7156,
      "step": 7462
    },
    {
      "epoch": 0.30929586804260434,
      "grad_norm": 0.3804436922073364,
      "learning_rate": 3.4537278793153466e-06,
      "loss": 0.6378,
      "step": 7463
    },
    {
      "epoch": 0.309337311948278,
      "grad_norm": 0.40576303005218506,
      "learning_rate": 3.4535206597869784e-06,
      "loss": 0.7069,
      "step": 7464
    },
    {
      "epoch": 0.3093787558539517,
      "grad_norm": 0.4195157289505005,
      "learning_rate": 3.4533134402586107e-06,
      "loss": 0.7415,
      "step": 7465
    },
    {
      "epoch": 0.30942019975962537,
      "grad_norm": 0.3904954195022583,
      "learning_rate": 3.4531062207302416e-06,
      "loss": 0.6456,
      "step": 7466
    },
    {
      "epoch": 0.309461643665299,
      "grad_norm": 0.421353816986084,
      "learning_rate": 3.452899001201874e-06,
      "loss": 0.7053,
      "step": 7467
    },
    {
      "epoch": 0.30950308757097267,
      "grad_norm": 0.41370323300361633,
      "learning_rate": 3.452691781673505e-06,
      "loss": 0.6846,
      "step": 7468
    },
    {
      "epoch": 0.30954453147664635,
      "grad_norm": 0.4078402817249298,
      "learning_rate": 3.452484562145137e-06,
      "loss": 0.7607,
      "step": 7469
    },
    {
      "epoch": 0.30958597538232,
      "grad_norm": 0.4107280671596527,
      "learning_rate": 3.4522773426167685e-06,
      "loss": 0.7,
      "step": 7470
    },
    {
      "epoch": 0.3096274192879937,
      "grad_norm": 0.3857530951499939,
      "learning_rate": 3.4520701230884003e-06,
      "loss": 0.6826,
      "step": 7471
    },
    {
      "epoch": 0.3096688631936674,
      "grad_norm": 0.4017072319984436,
      "learning_rate": 3.4518629035600317e-06,
      "loss": 0.6921,
      "step": 7472
    },
    {
      "epoch": 0.30971030709934105,
      "grad_norm": 0.46790146827697754,
      "learning_rate": 3.4516556840316635e-06,
      "loss": 0.7645,
      "step": 7473
    },
    {
      "epoch": 0.30975175100501473,
      "grad_norm": 0.44236353039741516,
      "learning_rate": 3.4514484645032953e-06,
      "loss": 0.7515,
      "step": 7474
    },
    {
      "epoch": 0.3097931949106884,
      "grad_norm": 0.4386170208454132,
      "learning_rate": 3.4512412449749267e-06,
      "loss": 0.729,
      "step": 7475
    },
    {
      "epoch": 0.30983463881636203,
      "grad_norm": 0.4028629958629608,
      "learning_rate": 3.4510340254465585e-06,
      "loss": 0.6648,
      "step": 7476
    },
    {
      "epoch": 0.3098760827220357,
      "grad_norm": 0.445488303899765,
      "learning_rate": 3.45082680591819e-06,
      "loss": 0.7395,
      "step": 7477
    },
    {
      "epoch": 0.3099175266277094,
      "grad_norm": 0.3786115348339081,
      "learning_rate": 3.4506195863898217e-06,
      "loss": 0.6415,
      "step": 7478
    },
    {
      "epoch": 0.30995897053338306,
      "grad_norm": 0.4242904484272003,
      "learning_rate": 3.450412366861453e-06,
      "loss": 0.7103,
      "step": 7479
    },
    {
      "epoch": 0.31000041443905674,
      "grad_norm": 0.43441250920295715,
      "learning_rate": 3.450205147333085e-06,
      "loss": 0.7158,
      "step": 7480
    },
    {
      "epoch": 0.3100418583447304,
      "grad_norm": 0.40602102875709534,
      "learning_rate": 3.4499979278047167e-06,
      "loss": 0.6924,
      "step": 7481
    },
    {
      "epoch": 0.3100833022504041,
      "grad_norm": 0.4194507598876953,
      "learning_rate": 3.449790708276348e-06,
      "loss": 0.7148,
      "step": 7482
    },
    {
      "epoch": 0.31012474615607777,
      "grad_norm": 0.44745177030563354,
      "learning_rate": 3.4495834887479803e-06,
      "loss": 0.7522,
      "step": 7483
    },
    {
      "epoch": 0.31016619006175145,
      "grad_norm": 0.3976851999759674,
      "learning_rate": 3.4493762692196112e-06,
      "loss": 0.6609,
      "step": 7484
    },
    {
      "epoch": 0.31020763396742507,
      "grad_norm": 0.3988339602947235,
      "learning_rate": 3.4491690496912435e-06,
      "loss": 0.7292,
      "step": 7485
    },
    {
      "epoch": 0.31024907787309874,
      "grad_norm": 0.39881035685539246,
      "learning_rate": 3.448961830162875e-06,
      "loss": 0.7434,
      "step": 7486
    },
    {
      "epoch": 0.3102905217787724,
      "grad_norm": 0.39574187994003296,
      "learning_rate": 3.4487546106345067e-06,
      "loss": 0.6707,
      "step": 7487
    },
    {
      "epoch": 0.3103319656844461,
      "grad_norm": 0.440719872713089,
      "learning_rate": 3.448547391106138e-06,
      "loss": 0.7405,
      "step": 7488
    },
    {
      "epoch": 0.3103734095901198,
      "grad_norm": 0.41781434416770935,
      "learning_rate": 3.44834017157777e-06,
      "loss": 0.6636,
      "step": 7489
    },
    {
      "epoch": 0.31041485349579345,
      "grad_norm": 0.38664522767066956,
      "learning_rate": 3.4481329520494017e-06,
      "loss": 0.7581,
      "step": 7490
    },
    {
      "epoch": 0.31045629740146713,
      "grad_norm": 0.4182446300983429,
      "learning_rate": 3.447925732521033e-06,
      "loss": 0.7524,
      "step": 7491
    },
    {
      "epoch": 0.3104977413071408,
      "grad_norm": 0.43499240279197693,
      "learning_rate": 3.447718512992665e-06,
      "loss": 0.7274,
      "step": 7492
    },
    {
      "epoch": 0.31053918521281443,
      "grad_norm": 0.4100394546985626,
      "learning_rate": 3.4475112934642963e-06,
      "loss": 0.7344,
      "step": 7493
    },
    {
      "epoch": 0.3105806291184881,
      "grad_norm": 0.4300372898578644,
      "learning_rate": 3.447304073935928e-06,
      "loss": 0.7051,
      "step": 7494
    },
    {
      "epoch": 0.3106220730241618,
      "grad_norm": 0.40647703409194946,
      "learning_rate": 3.4470968544075595e-06,
      "loss": 0.7255,
      "step": 7495
    },
    {
      "epoch": 0.31066351692983546,
      "grad_norm": 0.3884662985801697,
      "learning_rate": 3.4468896348791913e-06,
      "loss": 0.6992,
      "step": 7496
    },
    {
      "epoch": 0.31070496083550914,
      "grad_norm": 0.39692962169647217,
      "learning_rate": 3.4466824153508226e-06,
      "loss": 0.7073,
      "step": 7497
    },
    {
      "epoch": 0.3107464047411828,
      "grad_norm": 0.3854995369911194,
      "learning_rate": 3.4464751958224545e-06,
      "loss": 0.6799,
      "step": 7498
    },
    {
      "epoch": 0.3107878486468565,
      "grad_norm": 0.42676448822021484,
      "learning_rate": 3.4462679762940867e-06,
      "loss": 0.7096,
      "step": 7499
    },
    {
      "epoch": 0.31082929255253017,
      "grad_norm": 0.46065542101860046,
      "learning_rate": 3.4460607567657176e-06,
      "loss": 0.6981,
      "step": 7500
    },
    {
      "epoch": 0.31087073645820384,
      "grad_norm": 0.41962409019470215,
      "learning_rate": 3.44585353723735e-06,
      "loss": 0.7021,
      "step": 7501
    },
    {
      "epoch": 0.31091218036387747,
      "grad_norm": 0.3986532986164093,
      "learning_rate": 3.4456463177089813e-06,
      "loss": 0.7167,
      "step": 7502
    },
    {
      "epoch": 0.31095362426955114,
      "grad_norm": 0.4155408442020416,
      "learning_rate": 3.445439098180613e-06,
      "loss": 0.7473,
      "step": 7503
    },
    {
      "epoch": 0.3109950681752248,
      "grad_norm": 0.4450230300426483,
      "learning_rate": 3.4452318786522445e-06,
      "loss": 0.6719,
      "step": 7504
    },
    {
      "epoch": 0.3110365120808985,
      "grad_norm": 0.4236619472503662,
      "learning_rate": 3.4450246591238763e-06,
      "loss": 0.7209,
      "step": 7505
    },
    {
      "epoch": 0.3110779559865722,
      "grad_norm": 0.4422362744808197,
      "learning_rate": 3.4448174395955077e-06,
      "loss": 0.7454,
      "step": 7506
    },
    {
      "epoch": 0.31111939989224585,
      "grad_norm": 0.39725545048713684,
      "learning_rate": 3.4446102200671395e-06,
      "loss": 0.6923,
      "step": 7507
    },
    {
      "epoch": 0.31116084379791953,
      "grad_norm": 0.4980255961418152,
      "learning_rate": 3.4444030005387713e-06,
      "loss": 0.7163,
      "step": 7508
    },
    {
      "epoch": 0.3112022877035932,
      "grad_norm": 0.42044591903686523,
      "learning_rate": 3.4441957810104027e-06,
      "loss": 0.6964,
      "step": 7509
    },
    {
      "epoch": 0.3112437316092669,
      "grad_norm": 0.40436574816703796,
      "learning_rate": 3.4439885614820345e-06,
      "loss": 0.7174,
      "step": 7510
    },
    {
      "epoch": 0.3112851755149405,
      "grad_norm": 0.39719322323799133,
      "learning_rate": 3.443781341953666e-06,
      "loss": 0.7346,
      "step": 7511
    },
    {
      "epoch": 0.3113266194206142,
      "grad_norm": 0.39710134267807007,
      "learning_rate": 3.4435741224252977e-06,
      "loss": 0.728,
      "step": 7512
    },
    {
      "epoch": 0.31136806332628786,
      "grad_norm": 0.40367019176483154,
      "learning_rate": 3.443366902896929e-06,
      "loss": 0.6814,
      "step": 7513
    },
    {
      "epoch": 0.31140950723196154,
      "grad_norm": 0.4167819917201996,
      "learning_rate": 3.443159683368561e-06,
      "loss": 0.704,
      "step": 7514
    },
    {
      "epoch": 0.3114509511376352,
      "grad_norm": 0.4126543402671814,
      "learning_rate": 3.4429524638401922e-06,
      "loss": 0.7334,
      "step": 7515
    },
    {
      "epoch": 0.3114923950433089,
      "grad_norm": 0.41189107298851013,
      "learning_rate": 3.442745244311824e-06,
      "loss": 0.736,
      "step": 7516
    },
    {
      "epoch": 0.31153383894898257,
      "grad_norm": 0.4293951690196991,
      "learning_rate": 3.4425380247834563e-06,
      "loss": 0.6775,
      "step": 7517
    },
    {
      "epoch": 0.31157528285465624,
      "grad_norm": 0.4264795780181885,
      "learning_rate": 3.4423308052550873e-06,
      "loss": 0.7024,
      "step": 7518
    },
    {
      "epoch": 0.3116167267603299,
      "grad_norm": 0.41424986720085144,
      "learning_rate": 3.4421235857267195e-06,
      "loss": 0.6647,
      "step": 7519
    },
    {
      "epoch": 0.31165817066600354,
      "grad_norm": 0.3883225619792938,
      "learning_rate": 3.441916366198351e-06,
      "loss": 0.6848,
      "step": 7520
    },
    {
      "epoch": 0.3116996145716772,
      "grad_norm": 0.40998461842536926,
      "learning_rate": 3.4417091466699827e-06,
      "loss": 0.7324,
      "step": 7521
    },
    {
      "epoch": 0.3117410584773509,
      "grad_norm": 0.4122975766658783,
      "learning_rate": 3.441501927141614e-06,
      "loss": 0.6619,
      "step": 7522
    },
    {
      "epoch": 0.3117825023830246,
      "grad_norm": 0.3984352946281433,
      "learning_rate": 3.441294707613246e-06,
      "loss": 0.7292,
      "step": 7523
    },
    {
      "epoch": 0.31182394628869825,
      "grad_norm": 0.4168083965778351,
      "learning_rate": 3.4410874880848773e-06,
      "loss": 0.7017,
      "step": 7524
    },
    {
      "epoch": 0.3118653901943719,
      "grad_norm": 0.45293116569519043,
      "learning_rate": 3.440880268556509e-06,
      "loss": 0.709,
      "step": 7525
    },
    {
      "epoch": 0.3119068341000456,
      "grad_norm": 0.4320991635322571,
      "learning_rate": 3.440673049028141e-06,
      "loss": 0.7201,
      "step": 7526
    },
    {
      "epoch": 0.3119482780057193,
      "grad_norm": 0.40693241357803345,
      "learning_rate": 3.4404658294997723e-06,
      "loss": 0.663,
      "step": 7527
    },
    {
      "epoch": 0.3119897219113929,
      "grad_norm": 0.3822488784790039,
      "learning_rate": 3.440258609971404e-06,
      "loss": 0.6321,
      "step": 7528
    },
    {
      "epoch": 0.3120311658170666,
      "grad_norm": 0.42916402220726013,
      "learning_rate": 3.4400513904430355e-06,
      "loss": 0.7495,
      "step": 7529
    },
    {
      "epoch": 0.31207260972274026,
      "grad_norm": 0.39259234070777893,
      "learning_rate": 3.4398441709146673e-06,
      "loss": 0.7396,
      "step": 7530
    },
    {
      "epoch": 0.31211405362841393,
      "grad_norm": 0.41370290517807007,
      "learning_rate": 3.4396369513862987e-06,
      "loss": 0.7477,
      "step": 7531
    },
    {
      "epoch": 0.3121554975340876,
      "grad_norm": 0.4350496232509613,
      "learning_rate": 3.4394297318579305e-06,
      "loss": 0.7299,
      "step": 7532
    },
    {
      "epoch": 0.3121969414397613,
      "grad_norm": 0.4280523955821991,
      "learning_rate": 3.439222512329562e-06,
      "loss": 0.6707,
      "step": 7533
    },
    {
      "epoch": 0.31223838534543497,
      "grad_norm": 0.4487139582633972,
      "learning_rate": 3.4390152928011937e-06,
      "loss": 0.7424,
      "step": 7534
    },
    {
      "epoch": 0.31227982925110864,
      "grad_norm": 0.37774088978767395,
      "learning_rate": 3.438808073272826e-06,
      "loss": 0.6804,
      "step": 7535
    },
    {
      "epoch": 0.3123212731567823,
      "grad_norm": 0.4136647880077362,
      "learning_rate": 3.4386008537444573e-06,
      "loss": 0.7026,
      "step": 7536
    },
    {
      "epoch": 0.31236271706245594,
      "grad_norm": 0.4362996220588684,
      "learning_rate": 3.438393634216089e-06,
      "loss": 0.6705,
      "step": 7537
    },
    {
      "epoch": 0.3124041609681296,
      "grad_norm": 0.455662339925766,
      "learning_rate": 3.4381864146877205e-06,
      "loss": 0.7585,
      "step": 7538
    },
    {
      "epoch": 0.3124456048738033,
      "grad_norm": 0.407144159078598,
      "learning_rate": 3.4379791951593523e-06,
      "loss": 0.7109,
      "step": 7539
    },
    {
      "epoch": 0.312487048779477,
      "grad_norm": 0.42984285950660706,
      "learning_rate": 3.4377719756309837e-06,
      "loss": 0.6954,
      "step": 7540
    },
    {
      "epoch": 0.31252849268515065,
      "grad_norm": 0.3828597962856293,
      "learning_rate": 3.4375647561026155e-06,
      "loss": 0.6338,
      "step": 7541
    },
    {
      "epoch": 0.3125699365908243,
      "grad_norm": 0.4313989281654358,
      "learning_rate": 3.4373575365742473e-06,
      "loss": 0.6616,
      "step": 7542
    },
    {
      "epoch": 0.312611380496498,
      "grad_norm": 0.4203551411628723,
      "learning_rate": 3.4371503170458787e-06,
      "loss": 0.7487,
      "step": 7543
    },
    {
      "epoch": 0.3126528244021717,
      "grad_norm": 0.42185747623443604,
      "learning_rate": 3.4369430975175105e-06,
      "loss": 0.689,
      "step": 7544
    },
    {
      "epoch": 0.31269426830784536,
      "grad_norm": 0.40505892038345337,
      "learning_rate": 3.436735877989142e-06,
      "loss": 0.693,
      "step": 7545
    },
    {
      "epoch": 0.312735712213519,
      "grad_norm": 0.43437132239341736,
      "learning_rate": 3.4365286584607737e-06,
      "loss": 0.7076,
      "step": 7546
    },
    {
      "epoch": 0.31277715611919266,
      "grad_norm": 0.4146994650363922,
      "learning_rate": 3.436321438932405e-06,
      "loss": 0.7051,
      "step": 7547
    },
    {
      "epoch": 0.31281860002486633,
      "grad_norm": 0.41935038566589355,
      "learning_rate": 3.436114219404037e-06,
      "loss": 0.7714,
      "step": 7548
    },
    {
      "epoch": 0.31286004393054,
      "grad_norm": 0.4182562232017517,
      "learning_rate": 3.4359069998756683e-06,
      "loss": 0.7354,
      "step": 7549
    },
    {
      "epoch": 0.3129014878362137,
      "grad_norm": 0.4311809837818146,
      "learning_rate": 3.4356997803473e-06,
      "loss": 0.6664,
      "step": 7550
    },
    {
      "epoch": 0.31294293174188736,
      "grad_norm": 0.41274723410606384,
      "learning_rate": 3.4354925608189323e-06,
      "loss": 0.7095,
      "step": 7551
    },
    {
      "epoch": 0.31298437564756104,
      "grad_norm": 0.4006761908531189,
      "learning_rate": 3.4352853412905633e-06,
      "loss": 0.72,
      "step": 7552
    },
    {
      "epoch": 0.3130258195532347,
      "grad_norm": 0.40339601039886475,
      "learning_rate": 3.4350781217621955e-06,
      "loss": 0.6765,
      "step": 7553
    },
    {
      "epoch": 0.31306726345890834,
      "grad_norm": 0.41138267517089844,
      "learning_rate": 3.434870902233827e-06,
      "loss": 0.676,
      "step": 7554
    },
    {
      "epoch": 0.313108707364582,
      "grad_norm": 0.4002581238746643,
      "learning_rate": 3.4346636827054587e-06,
      "loss": 0.6326,
      "step": 7555
    },
    {
      "epoch": 0.3131501512702557,
      "grad_norm": 0.43722182512283325,
      "learning_rate": 3.43445646317709e-06,
      "loss": 0.7369,
      "step": 7556
    },
    {
      "epoch": 0.31319159517592937,
      "grad_norm": 0.44610583782196045,
      "learning_rate": 3.434249243648722e-06,
      "loss": 0.7502,
      "step": 7557
    },
    {
      "epoch": 0.31323303908160305,
      "grad_norm": 0.40255096554756165,
      "learning_rate": 3.4340420241203533e-06,
      "loss": 0.693,
      "step": 7558
    },
    {
      "epoch": 0.3132744829872767,
      "grad_norm": 0.3938567042350769,
      "learning_rate": 3.433834804591985e-06,
      "loss": 0.7058,
      "step": 7559
    },
    {
      "epoch": 0.3133159268929504,
      "grad_norm": 0.4168805181980133,
      "learning_rate": 3.433627585063617e-06,
      "loss": 0.6844,
      "step": 7560
    },
    {
      "epoch": 0.3133573707986241,
      "grad_norm": 0.38529545068740845,
      "learning_rate": 3.4334203655352483e-06,
      "loss": 0.6611,
      "step": 7561
    },
    {
      "epoch": 0.31339881470429776,
      "grad_norm": 0.3977561891078949,
      "learning_rate": 3.43321314600688e-06,
      "loss": 0.7312,
      "step": 7562
    },
    {
      "epoch": 0.3134402586099714,
      "grad_norm": 0.4147842824459076,
      "learning_rate": 3.4330059264785115e-06,
      "loss": 0.7148,
      "step": 7563
    },
    {
      "epoch": 0.31348170251564506,
      "grad_norm": 0.3993547856807709,
      "learning_rate": 3.4327987069501433e-06,
      "loss": 0.7006,
      "step": 7564
    },
    {
      "epoch": 0.31352314642131873,
      "grad_norm": 0.4317115843296051,
      "learning_rate": 3.4325914874217747e-06,
      "loss": 0.7109,
      "step": 7565
    },
    {
      "epoch": 0.3135645903269924,
      "grad_norm": 0.4191565215587616,
      "learning_rate": 3.4323842678934065e-06,
      "loss": 0.6929,
      "step": 7566
    },
    {
      "epoch": 0.3136060342326661,
      "grad_norm": 0.4135379493236542,
      "learning_rate": 3.432177048365038e-06,
      "loss": 0.7278,
      "step": 7567
    },
    {
      "epoch": 0.31364747813833976,
      "grad_norm": 0.4381425976753235,
      "learning_rate": 3.4319698288366697e-06,
      "loss": 0.6702,
      "step": 7568
    },
    {
      "epoch": 0.31368892204401344,
      "grad_norm": 0.42175841331481934,
      "learning_rate": 3.431762609308302e-06,
      "loss": 0.7126,
      "step": 7569
    },
    {
      "epoch": 0.3137303659496871,
      "grad_norm": 0.44989556074142456,
      "learning_rate": 3.4315553897799333e-06,
      "loss": 0.7339,
      "step": 7570
    },
    {
      "epoch": 0.3137718098553608,
      "grad_norm": 0.4091661870479584,
      "learning_rate": 3.431348170251565e-06,
      "loss": 0.7267,
      "step": 7571
    },
    {
      "epoch": 0.3138132537610344,
      "grad_norm": 0.4093273878097534,
      "learning_rate": 3.4311409507231965e-06,
      "loss": 0.7229,
      "step": 7572
    },
    {
      "epoch": 0.3138546976667081,
      "grad_norm": 0.4181423783302307,
      "learning_rate": 3.4309337311948283e-06,
      "loss": 0.7021,
      "step": 7573
    },
    {
      "epoch": 0.31389614157238177,
      "grad_norm": 0.48012322187423706,
      "learning_rate": 3.4307265116664597e-06,
      "loss": 0.7498,
      "step": 7574
    },
    {
      "epoch": 0.31393758547805545,
      "grad_norm": 0.4160819351673126,
      "learning_rate": 3.4305192921380915e-06,
      "loss": 0.7339,
      "step": 7575
    },
    {
      "epoch": 0.3139790293837291,
      "grad_norm": 0.40453293919563293,
      "learning_rate": 3.430312072609723e-06,
      "loss": 0.686,
      "step": 7576
    },
    {
      "epoch": 0.3140204732894028,
      "grad_norm": 0.4028432369232178,
      "learning_rate": 3.4301048530813547e-06,
      "loss": 0.6974,
      "step": 7577
    },
    {
      "epoch": 0.3140619171950765,
      "grad_norm": 0.41426563262939453,
      "learning_rate": 3.4298976335529865e-06,
      "loss": 0.7039,
      "step": 7578
    },
    {
      "epoch": 0.31410336110075016,
      "grad_norm": 0.39181146025657654,
      "learning_rate": 3.429690414024618e-06,
      "loss": 0.7236,
      "step": 7579
    },
    {
      "epoch": 0.31414480500642383,
      "grad_norm": 0.4060283303260803,
      "learning_rate": 3.4294831944962497e-06,
      "loss": 0.6846,
      "step": 7580
    },
    {
      "epoch": 0.31418624891209745,
      "grad_norm": 0.42515116930007935,
      "learning_rate": 3.429275974967881e-06,
      "loss": 0.6862,
      "step": 7581
    },
    {
      "epoch": 0.31422769281777113,
      "grad_norm": 0.417490154504776,
      "learning_rate": 3.429068755439513e-06,
      "loss": 0.6772,
      "step": 7582
    },
    {
      "epoch": 0.3142691367234448,
      "grad_norm": 0.3897988200187683,
      "learning_rate": 3.4288615359111443e-06,
      "loss": 0.6956,
      "step": 7583
    },
    {
      "epoch": 0.3143105806291185,
      "grad_norm": 0.41970372200012207,
      "learning_rate": 3.428654316382776e-06,
      "loss": 0.7405,
      "step": 7584
    },
    {
      "epoch": 0.31435202453479216,
      "grad_norm": 0.41606375575065613,
      "learning_rate": 3.4284470968544075e-06,
      "loss": 0.6985,
      "step": 7585
    },
    {
      "epoch": 0.31439346844046584,
      "grad_norm": 0.4188237190246582,
      "learning_rate": 3.4282398773260393e-06,
      "loss": 0.7332,
      "step": 7586
    },
    {
      "epoch": 0.3144349123461395,
      "grad_norm": 0.4534447491168976,
      "learning_rate": 3.4280326577976715e-06,
      "loss": 0.7744,
      "step": 7587
    },
    {
      "epoch": 0.3144763562518132,
      "grad_norm": 0.38659927248954773,
      "learning_rate": 3.427825438269303e-06,
      "loss": 0.7052,
      "step": 7588
    },
    {
      "epoch": 0.3145178001574868,
      "grad_norm": 0.41620782017707825,
      "learning_rate": 3.4276182187409347e-06,
      "loss": 0.6715,
      "step": 7589
    },
    {
      "epoch": 0.3145592440631605,
      "grad_norm": 0.39819517731666565,
      "learning_rate": 3.427410999212566e-06,
      "loss": 0.6671,
      "step": 7590
    },
    {
      "epoch": 0.31460068796883417,
      "grad_norm": 0.4335789382457733,
      "learning_rate": 3.427203779684198e-06,
      "loss": 0.76,
      "step": 7591
    },
    {
      "epoch": 0.31464213187450785,
      "grad_norm": 0.42374077439308167,
      "learning_rate": 3.4269965601558293e-06,
      "loss": 0.6885,
      "step": 7592
    },
    {
      "epoch": 0.3146835757801815,
      "grad_norm": 0.4187897741794586,
      "learning_rate": 3.426789340627461e-06,
      "loss": 0.7407,
      "step": 7593
    },
    {
      "epoch": 0.3147250196858552,
      "grad_norm": 0.39622336626052856,
      "learning_rate": 3.4265821210990925e-06,
      "loss": 0.7104,
      "step": 7594
    },
    {
      "epoch": 0.3147664635915289,
      "grad_norm": 0.44573697447776794,
      "learning_rate": 3.4263749015707243e-06,
      "loss": 0.682,
      "step": 7595
    },
    {
      "epoch": 0.31480790749720255,
      "grad_norm": 0.4246913194656372,
      "learning_rate": 3.426167682042356e-06,
      "loss": 0.7302,
      "step": 7596
    },
    {
      "epoch": 0.31484935140287623,
      "grad_norm": 0.4266548752784729,
      "learning_rate": 3.4259604625139875e-06,
      "loss": 0.7114,
      "step": 7597
    },
    {
      "epoch": 0.31489079530854985,
      "grad_norm": 0.4314194619655609,
      "learning_rate": 3.4257532429856193e-06,
      "loss": 0.6843,
      "step": 7598
    },
    {
      "epoch": 0.31493223921422353,
      "grad_norm": 0.41192060708999634,
      "learning_rate": 3.4255460234572507e-06,
      "loss": 0.7031,
      "step": 7599
    },
    {
      "epoch": 0.3149736831198972,
      "grad_norm": 0.42709439992904663,
      "learning_rate": 3.4253388039288825e-06,
      "loss": 0.7706,
      "step": 7600
    },
    {
      "epoch": 0.3150151270255709,
      "grad_norm": 0.41548824310302734,
      "learning_rate": 3.425131584400514e-06,
      "loss": 0.6832,
      "step": 7601
    },
    {
      "epoch": 0.31505657093124456,
      "grad_norm": 0.4305908679962158,
      "learning_rate": 3.4249243648721457e-06,
      "loss": 0.7131,
      "step": 7602
    },
    {
      "epoch": 0.31509801483691824,
      "grad_norm": 0.4105145335197449,
      "learning_rate": 3.424717145343778e-06,
      "loss": 0.7451,
      "step": 7603
    },
    {
      "epoch": 0.3151394587425919,
      "grad_norm": 0.441808819770813,
      "learning_rate": 3.4245099258154093e-06,
      "loss": 0.7212,
      "step": 7604
    },
    {
      "epoch": 0.3151809026482656,
      "grad_norm": 0.4478551149368286,
      "learning_rate": 3.424302706287041e-06,
      "loss": 0.697,
      "step": 7605
    },
    {
      "epoch": 0.31522234655393927,
      "grad_norm": 0.39708220958709717,
      "learning_rate": 3.4240954867586725e-06,
      "loss": 0.6709,
      "step": 7606
    },
    {
      "epoch": 0.3152637904596129,
      "grad_norm": 0.4461628794670105,
      "learning_rate": 3.4238882672303043e-06,
      "loss": 0.7662,
      "step": 7607
    },
    {
      "epoch": 0.31530523436528657,
      "grad_norm": 0.42614296078681946,
      "learning_rate": 3.4236810477019357e-06,
      "loss": 0.7605,
      "step": 7608
    },
    {
      "epoch": 0.31534667827096025,
      "grad_norm": 0.4535657465457916,
      "learning_rate": 3.4234738281735675e-06,
      "loss": 0.6689,
      "step": 7609
    },
    {
      "epoch": 0.3153881221766339,
      "grad_norm": 0.4463537931442261,
      "learning_rate": 3.423266608645199e-06,
      "loss": 0.7075,
      "step": 7610
    },
    {
      "epoch": 0.3154295660823076,
      "grad_norm": 0.4315204620361328,
      "learning_rate": 3.4230593891168307e-06,
      "loss": 0.698,
      "step": 7611
    },
    {
      "epoch": 0.3154710099879813,
      "grad_norm": 0.38476037979125977,
      "learning_rate": 3.4228521695884625e-06,
      "loss": 0.7104,
      "step": 7612
    },
    {
      "epoch": 0.31551245389365495,
      "grad_norm": 0.3910635709762573,
      "learning_rate": 3.422644950060094e-06,
      "loss": 0.7151,
      "step": 7613
    },
    {
      "epoch": 0.31555389779932863,
      "grad_norm": 0.43930676579475403,
      "learning_rate": 3.4224377305317257e-06,
      "loss": 0.7388,
      "step": 7614
    },
    {
      "epoch": 0.31559534170500225,
      "grad_norm": 0.3941642642021179,
      "learning_rate": 3.422230511003357e-06,
      "loss": 0.7253,
      "step": 7615
    },
    {
      "epoch": 0.31563678561067593,
      "grad_norm": 0.4005005657672882,
      "learning_rate": 3.422023291474989e-06,
      "loss": 0.7275,
      "step": 7616
    },
    {
      "epoch": 0.3156782295163496,
      "grad_norm": 0.4259610176086426,
      "learning_rate": 3.4218160719466203e-06,
      "loss": 0.7018,
      "step": 7617
    },
    {
      "epoch": 0.3157196734220233,
      "grad_norm": 0.43708252906799316,
      "learning_rate": 3.421608852418252e-06,
      "loss": 0.7396,
      "step": 7618
    },
    {
      "epoch": 0.31576111732769696,
      "grad_norm": 0.4148063063621521,
      "learning_rate": 3.4214016328898835e-06,
      "loss": 0.7219,
      "step": 7619
    },
    {
      "epoch": 0.31580256123337064,
      "grad_norm": 0.38472887873649597,
      "learning_rate": 3.4211944133615153e-06,
      "loss": 0.7014,
      "step": 7620
    },
    {
      "epoch": 0.3158440051390443,
      "grad_norm": 0.4026165306568146,
      "learning_rate": 3.4209871938331475e-06,
      "loss": 0.6952,
      "step": 7621
    },
    {
      "epoch": 0.315885449044718,
      "grad_norm": 0.4526183605194092,
      "learning_rate": 3.420779974304779e-06,
      "loss": 0.7534,
      "step": 7622
    },
    {
      "epoch": 0.31592689295039167,
      "grad_norm": 0.4179258346557617,
      "learning_rate": 3.4205727547764107e-06,
      "loss": 0.7466,
      "step": 7623
    },
    {
      "epoch": 0.3159683368560653,
      "grad_norm": 0.3909228444099426,
      "learning_rate": 3.420365535248042e-06,
      "loss": 0.6909,
      "step": 7624
    },
    {
      "epoch": 0.31600978076173897,
      "grad_norm": 0.42192474007606506,
      "learning_rate": 3.420158315719674e-06,
      "loss": 0.7324,
      "step": 7625
    },
    {
      "epoch": 0.31605122466741264,
      "grad_norm": 0.3940717875957489,
      "learning_rate": 3.4199510961913053e-06,
      "loss": 0.6636,
      "step": 7626
    },
    {
      "epoch": 0.3160926685730863,
      "grad_norm": 0.4045552611351013,
      "learning_rate": 3.419743876662937e-06,
      "loss": 0.6958,
      "step": 7627
    },
    {
      "epoch": 0.31613411247876,
      "grad_norm": 0.41316014528274536,
      "learning_rate": 3.4195366571345685e-06,
      "loss": 0.6833,
      "step": 7628
    },
    {
      "epoch": 0.3161755563844337,
      "grad_norm": 0.4057060480117798,
      "learning_rate": 3.4193294376062003e-06,
      "loss": 0.7013,
      "step": 7629
    },
    {
      "epoch": 0.31621700029010735,
      "grad_norm": 0.41709405183792114,
      "learning_rate": 3.419122218077832e-06,
      "loss": 0.7163,
      "step": 7630
    },
    {
      "epoch": 0.31625844419578103,
      "grad_norm": 0.4237930178642273,
      "learning_rate": 3.4189149985494635e-06,
      "loss": 0.6481,
      "step": 7631
    },
    {
      "epoch": 0.3162998881014547,
      "grad_norm": 0.4220014214515686,
      "learning_rate": 3.4187077790210953e-06,
      "loss": 0.7122,
      "step": 7632
    },
    {
      "epoch": 0.31634133200712833,
      "grad_norm": 0.41987738013267517,
      "learning_rate": 3.4185005594927267e-06,
      "loss": 0.7341,
      "step": 7633
    },
    {
      "epoch": 0.316382775912802,
      "grad_norm": 0.4061524271965027,
      "learning_rate": 3.4182933399643585e-06,
      "loss": 0.7395,
      "step": 7634
    },
    {
      "epoch": 0.3164242198184757,
      "grad_norm": 0.42098337411880493,
      "learning_rate": 3.41808612043599e-06,
      "loss": 0.7239,
      "step": 7635
    },
    {
      "epoch": 0.31646566372414936,
      "grad_norm": 0.4204918146133423,
      "learning_rate": 3.4178789009076217e-06,
      "loss": 0.6733,
      "step": 7636
    },
    {
      "epoch": 0.31650710762982304,
      "grad_norm": 0.4350239038467407,
      "learning_rate": 3.417671681379253e-06,
      "loss": 0.7383,
      "step": 7637
    },
    {
      "epoch": 0.3165485515354967,
      "grad_norm": 0.4121612012386322,
      "learning_rate": 3.4174644618508853e-06,
      "loss": 0.72,
      "step": 7638
    },
    {
      "epoch": 0.3165899954411704,
      "grad_norm": 0.41749104857444763,
      "learning_rate": 3.417257242322517e-06,
      "loss": 0.7292,
      "step": 7639
    },
    {
      "epoch": 0.31663143934684407,
      "grad_norm": 0.3663554787635803,
      "learning_rate": 3.4170500227941485e-06,
      "loss": 0.676,
      "step": 7640
    },
    {
      "epoch": 0.3166728832525177,
      "grad_norm": 0.4047391712665558,
      "learning_rate": 3.4168428032657803e-06,
      "loss": 0.6873,
      "step": 7641
    },
    {
      "epoch": 0.31671432715819137,
      "grad_norm": 0.44748350977897644,
      "learning_rate": 3.4166355837374117e-06,
      "loss": 0.7056,
      "step": 7642
    },
    {
      "epoch": 0.31675577106386504,
      "grad_norm": 0.45415085554122925,
      "learning_rate": 3.4164283642090435e-06,
      "loss": 0.6814,
      "step": 7643
    },
    {
      "epoch": 0.3167972149695387,
      "grad_norm": 0.46014922857284546,
      "learning_rate": 3.416221144680675e-06,
      "loss": 0.7327,
      "step": 7644
    },
    {
      "epoch": 0.3168386588752124,
      "grad_norm": 0.44210028648376465,
      "learning_rate": 3.4160139251523067e-06,
      "loss": 0.7363,
      "step": 7645
    },
    {
      "epoch": 0.3168801027808861,
      "grad_norm": 0.39721691608428955,
      "learning_rate": 3.415806705623938e-06,
      "loss": 0.6648,
      "step": 7646
    },
    {
      "epoch": 0.31692154668655975,
      "grad_norm": 0.4421832263469696,
      "learning_rate": 3.41559948609557e-06,
      "loss": 0.7488,
      "step": 7647
    },
    {
      "epoch": 0.31696299059223343,
      "grad_norm": 0.41342490911483765,
      "learning_rate": 3.4153922665672017e-06,
      "loss": 0.7532,
      "step": 7648
    },
    {
      "epoch": 0.3170044344979071,
      "grad_norm": 0.44288313388824463,
      "learning_rate": 3.415185047038833e-06,
      "loss": 0.7102,
      "step": 7649
    },
    {
      "epoch": 0.3170458784035807,
      "grad_norm": 0.375647634267807,
      "learning_rate": 3.414977827510465e-06,
      "loss": 0.678,
      "step": 7650
    },
    {
      "epoch": 0.3170873223092544,
      "grad_norm": 0.3705850839614868,
      "learning_rate": 3.4147706079820963e-06,
      "loss": 0.6506,
      "step": 7651
    },
    {
      "epoch": 0.3171287662149281,
      "grad_norm": 0.4182378351688385,
      "learning_rate": 3.414563388453728e-06,
      "loss": 0.7155,
      "step": 7652
    },
    {
      "epoch": 0.31717021012060176,
      "grad_norm": 0.37628793716430664,
      "learning_rate": 3.4143561689253595e-06,
      "loss": 0.6556,
      "step": 7653
    },
    {
      "epoch": 0.31721165402627544,
      "grad_norm": 0.3786916732788086,
      "learning_rate": 3.4141489493969913e-06,
      "loss": 0.6709,
      "step": 7654
    },
    {
      "epoch": 0.3172530979319491,
      "grad_norm": 0.39087650179862976,
      "learning_rate": 3.4139417298686227e-06,
      "loss": 0.708,
      "step": 7655
    },
    {
      "epoch": 0.3172945418376228,
      "grad_norm": 0.3908731937408447,
      "learning_rate": 3.413734510340255e-06,
      "loss": 0.7054,
      "step": 7656
    },
    {
      "epoch": 0.31733598574329647,
      "grad_norm": 0.4352070689201355,
      "learning_rate": 3.4135272908118867e-06,
      "loss": 0.7251,
      "step": 7657
    },
    {
      "epoch": 0.31737742964897014,
      "grad_norm": 0.3858291804790497,
      "learning_rate": 3.413320071283518e-06,
      "loss": 0.6615,
      "step": 7658
    },
    {
      "epoch": 0.31741887355464377,
      "grad_norm": 0.49656426906585693,
      "learning_rate": 3.41311285175515e-06,
      "loss": 0.688,
      "step": 7659
    },
    {
      "epoch": 0.31746031746031744,
      "grad_norm": 0.3792707622051239,
      "learning_rate": 3.4129056322267813e-06,
      "loss": 0.6516,
      "step": 7660
    },
    {
      "epoch": 0.3175017613659911,
      "grad_norm": 0.41389650106430054,
      "learning_rate": 3.412698412698413e-06,
      "loss": 0.7004,
      "step": 7661
    },
    {
      "epoch": 0.3175432052716648,
      "grad_norm": 0.46229544281959534,
      "learning_rate": 3.4124911931700445e-06,
      "loss": 0.7537,
      "step": 7662
    },
    {
      "epoch": 0.3175846491773385,
      "grad_norm": 0.39482879638671875,
      "learning_rate": 3.4122839736416763e-06,
      "loss": 0.6746,
      "step": 7663
    },
    {
      "epoch": 0.31762609308301215,
      "grad_norm": 0.41032156348228455,
      "learning_rate": 3.412076754113308e-06,
      "loss": 0.6869,
      "step": 7664
    },
    {
      "epoch": 0.31766753698868583,
      "grad_norm": 0.4453418254852295,
      "learning_rate": 3.4118695345849395e-06,
      "loss": 0.7162,
      "step": 7665
    },
    {
      "epoch": 0.3177089808943595,
      "grad_norm": 0.42089638113975525,
      "learning_rate": 3.4116623150565713e-06,
      "loss": 0.7058,
      "step": 7666
    },
    {
      "epoch": 0.3177504248000332,
      "grad_norm": 0.43652886152267456,
      "learning_rate": 3.4114550955282027e-06,
      "loss": 0.7461,
      "step": 7667
    },
    {
      "epoch": 0.3177918687057068,
      "grad_norm": 0.45390647649765015,
      "learning_rate": 3.4112478759998345e-06,
      "loss": 0.7411,
      "step": 7668
    },
    {
      "epoch": 0.3178333126113805,
      "grad_norm": 0.3878803551197052,
      "learning_rate": 3.411040656471466e-06,
      "loss": 0.6814,
      "step": 7669
    },
    {
      "epoch": 0.31787475651705416,
      "grad_norm": 0.4003022015094757,
      "learning_rate": 3.4108334369430977e-06,
      "loss": 0.6626,
      "step": 7670
    },
    {
      "epoch": 0.31791620042272783,
      "grad_norm": 0.3987198770046234,
      "learning_rate": 3.410626217414729e-06,
      "loss": 0.6571,
      "step": 7671
    },
    {
      "epoch": 0.3179576443284015,
      "grad_norm": 0.40823686122894287,
      "learning_rate": 3.4104189978863613e-06,
      "loss": 0.6941,
      "step": 7672
    },
    {
      "epoch": 0.3179990882340752,
      "grad_norm": 0.40657660365104675,
      "learning_rate": 3.410211778357993e-06,
      "loss": 0.741,
      "step": 7673
    },
    {
      "epoch": 0.31804053213974887,
      "grad_norm": 0.44792336225509644,
      "learning_rate": 3.4100045588296245e-06,
      "loss": 0.738,
      "step": 7674
    },
    {
      "epoch": 0.31808197604542254,
      "grad_norm": 0.4176732301712036,
      "learning_rate": 3.4097973393012563e-06,
      "loss": 0.7351,
      "step": 7675
    },
    {
      "epoch": 0.31812341995109616,
      "grad_norm": 0.41925331950187683,
      "learning_rate": 3.4095901197728877e-06,
      "loss": 0.7174,
      "step": 7676
    },
    {
      "epoch": 0.31816486385676984,
      "grad_norm": 0.4190998673439026,
      "learning_rate": 3.4093829002445195e-06,
      "loss": 0.7007,
      "step": 7677
    },
    {
      "epoch": 0.3182063077624435,
      "grad_norm": 0.40392783284187317,
      "learning_rate": 3.409175680716151e-06,
      "loss": 0.6675,
      "step": 7678
    },
    {
      "epoch": 0.3182477516681172,
      "grad_norm": 0.42512425780296326,
      "learning_rate": 3.4089684611877827e-06,
      "loss": 0.7351,
      "step": 7679
    },
    {
      "epoch": 0.3182891955737909,
      "grad_norm": 0.4120903015136719,
      "learning_rate": 3.408761241659414e-06,
      "loss": 0.6718,
      "step": 7680
    },
    {
      "epoch": 0.31833063947946455,
      "grad_norm": 0.432298481464386,
      "learning_rate": 3.408554022131046e-06,
      "loss": 0.7559,
      "step": 7681
    },
    {
      "epoch": 0.3183720833851382,
      "grad_norm": 0.43162813782691956,
      "learning_rate": 3.4083468026026777e-06,
      "loss": 0.73,
      "step": 7682
    },
    {
      "epoch": 0.3184135272908119,
      "grad_norm": 0.44006428122520447,
      "learning_rate": 3.408139583074309e-06,
      "loss": 0.7114,
      "step": 7683
    },
    {
      "epoch": 0.3184549711964856,
      "grad_norm": 0.45050138235092163,
      "learning_rate": 3.407932363545941e-06,
      "loss": 0.71,
      "step": 7684
    },
    {
      "epoch": 0.3184964151021592,
      "grad_norm": 0.4031083583831787,
      "learning_rate": 3.4077251440175723e-06,
      "loss": 0.6733,
      "step": 7685
    },
    {
      "epoch": 0.3185378590078329,
      "grad_norm": 0.44399768114089966,
      "learning_rate": 3.407517924489204e-06,
      "loss": 0.7546,
      "step": 7686
    },
    {
      "epoch": 0.31857930291350656,
      "grad_norm": 0.3915066719055176,
      "learning_rate": 3.4073107049608355e-06,
      "loss": 0.6843,
      "step": 7687
    },
    {
      "epoch": 0.31862074681918023,
      "grad_norm": 0.4070793092250824,
      "learning_rate": 3.4071034854324677e-06,
      "loss": 0.7227,
      "step": 7688
    },
    {
      "epoch": 0.3186621907248539,
      "grad_norm": 0.42345187067985535,
      "learning_rate": 3.4068962659040987e-06,
      "loss": 0.7073,
      "step": 7689
    },
    {
      "epoch": 0.3187036346305276,
      "grad_norm": 0.3930114209651947,
      "learning_rate": 3.406689046375731e-06,
      "loss": 0.7393,
      "step": 7690
    },
    {
      "epoch": 0.31874507853620127,
      "grad_norm": 0.40518513321876526,
      "learning_rate": 3.4064818268473627e-06,
      "loss": 0.7024,
      "step": 7691
    },
    {
      "epoch": 0.31878652244187494,
      "grad_norm": 0.39365819096565247,
      "learning_rate": 3.406274607318994e-06,
      "loss": 0.6772,
      "step": 7692
    },
    {
      "epoch": 0.3188279663475486,
      "grad_norm": 0.4130285680294037,
      "learning_rate": 3.406067387790626e-06,
      "loss": 0.6395,
      "step": 7693
    },
    {
      "epoch": 0.31886941025322224,
      "grad_norm": 0.4543284475803375,
      "learning_rate": 3.4058601682622573e-06,
      "loss": 0.7036,
      "step": 7694
    },
    {
      "epoch": 0.3189108541588959,
      "grad_norm": 0.43021634221076965,
      "learning_rate": 3.405652948733889e-06,
      "loss": 0.6968,
      "step": 7695
    },
    {
      "epoch": 0.3189522980645696,
      "grad_norm": 0.4006253182888031,
      "learning_rate": 3.4054457292055205e-06,
      "loss": 0.7056,
      "step": 7696
    },
    {
      "epoch": 0.31899374197024327,
      "grad_norm": 0.40772032737731934,
      "learning_rate": 3.4052385096771523e-06,
      "loss": 0.6943,
      "step": 7697
    },
    {
      "epoch": 0.31903518587591695,
      "grad_norm": 0.43194499611854553,
      "learning_rate": 3.4050312901487837e-06,
      "loss": 0.698,
      "step": 7698
    },
    {
      "epoch": 0.3190766297815906,
      "grad_norm": 0.43614739179611206,
      "learning_rate": 3.4048240706204155e-06,
      "loss": 0.6924,
      "step": 7699
    },
    {
      "epoch": 0.3191180736872643,
      "grad_norm": 0.3805427849292755,
      "learning_rate": 3.4046168510920473e-06,
      "loss": 0.6582,
      "step": 7700
    },
    {
      "epoch": 0.319159517592938,
      "grad_norm": 0.42810508608818054,
      "learning_rate": 3.4044096315636787e-06,
      "loss": 0.7832,
      "step": 7701
    },
    {
      "epoch": 0.3192009614986116,
      "grad_norm": 0.4575166702270508,
      "learning_rate": 3.4042024120353105e-06,
      "loss": 0.7075,
      "step": 7702
    },
    {
      "epoch": 0.3192424054042853,
      "grad_norm": 0.4254007041454315,
      "learning_rate": 3.403995192506942e-06,
      "loss": 0.6859,
      "step": 7703
    },
    {
      "epoch": 0.31928384930995896,
      "grad_norm": 0.41642487049102783,
      "learning_rate": 3.4037879729785737e-06,
      "loss": 0.677,
      "step": 7704
    },
    {
      "epoch": 0.31932529321563263,
      "grad_norm": 0.3975803554058075,
      "learning_rate": 3.403580753450205e-06,
      "loss": 0.6422,
      "step": 7705
    },
    {
      "epoch": 0.3193667371213063,
      "grad_norm": 0.381899356842041,
      "learning_rate": 3.4033735339218373e-06,
      "loss": 0.6702,
      "step": 7706
    },
    {
      "epoch": 0.31940818102698,
      "grad_norm": 0.4365960657596588,
      "learning_rate": 3.4031663143934683e-06,
      "loss": 0.7109,
      "step": 7707
    },
    {
      "epoch": 0.31944962493265366,
      "grad_norm": 0.42308947443962097,
      "learning_rate": 3.4029590948651005e-06,
      "loss": 0.7271,
      "step": 7708
    },
    {
      "epoch": 0.31949106883832734,
      "grad_norm": 0.4323892295360565,
      "learning_rate": 3.4027518753367323e-06,
      "loss": 0.688,
      "step": 7709
    },
    {
      "epoch": 0.319532512744001,
      "grad_norm": 0.41515398025512695,
      "learning_rate": 3.4025446558083637e-06,
      "loss": 0.7493,
      "step": 7710
    },
    {
      "epoch": 0.31957395664967464,
      "grad_norm": 0.40579625964164734,
      "learning_rate": 3.4023374362799955e-06,
      "loss": 0.6913,
      "step": 7711
    },
    {
      "epoch": 0.3196154005553483,
      "grad_norm": 0.3911464214324951,
      "learning_rate": 3.402130216751627e-06,
      "loss": 0.6416,
      "step": 7712
    },
    {
      "epoch": 0.319656844461022,
      "grad_norm": 0.491960346698761,
      "learning_rate": 3.4019229972232587e-06,
      "loss": 0.7424,
      "step": 7713
    },
    {
      "epoch": 0.31969828836669567,
      "grad_norm": 0.4209902584552765,
      "learning_rate": 3.40171577769489e-06,
      "loss": 0.6841,
      "step": 7714
    },
    {
      "epoch": 0.31973973227236935,
      "grad_norm": 0.4620499610900879,
      "learning_rate": 3.401508558166522e-06,
      "loss": 0.7905,
      "step": 7715
    },
    {
      "epoch": 0.319781176178043,
      "grad_norm": 0.4331238269805908,
      "learning_rate": 3.4013013386381533e-06,
      "loss": 0.7689,
      "step": 7716
    },
    {
      "epoch": 0.3198226200837167,
      "grad_norm": 0.4361741244792938,
      "learning_rate": 3.401094119109785e-06,
      "loss": 0.7283,
      "step": 7717
    },
    {
      "epoch": 0.3198640639893904,
      "grad_norm": 0.3759622871875763,
      "learning_rate": 3.400886899581417e-06,
      "loss": 0.7046,
      "step": 7718
    },
    {
      "epoch": 0.31990550789506406,
      "grad_norm": 0.4229751527309418,
      "learning_rate": 3.4006796800530483e-06,
      "loss": 0.7019,
      "step": 7719
    },
    {
      "epoch": 0.3199469518007377,
      "grad_norm": 0.40966150164604187,
      "learning_rate": 3.40047246052468e-06,
      "loss": 0.7168,
      "step": 7720
    },
    {
      "epoch": 0.31998839570641135,
      "grad_norm": 0.39007389545440674,
      "learning_rate": 3.4002652409963115e-06,
      "loss": 0.7107,
      "step": 7721
    },
    {
      "epoch": 0.32002983961208503,
      "grad_norm": 0.3803401291370392,
      "learning_rate": 3.4000580214679437e-06,
      "loss": 0.6495,
      "step": 7722
    },
    {
      "epoch": 0.3200712835177587,
      "grad_norm": 0.4482669234275818,
      "learning_rate": 3.3998508019395747e-06,
      "loss": 0.7312,
      "step": 7723
    },
    {
      "epoch": 0.3201127274234324,
      "grad_norm": 0.4213234782218933,
      "learning_rate": 3.399643582411207e-06,
      "loss": 0.6666,
      "step": 7724
    },
    {
      "epoch": 0.32015417132910606,
      "grad_norm": 0.4391288161277771,
      "learning_rate": 3.3994363628828387e-06,
      "loss": 0.7227,
      "step": 7725
    },
    {
      "epoch": 0.32019561523477974,
      "grad_norm": 0.37178272008895874,
      "learning_rate": 3.39922914335447e-06,
      "loss": 0.6552,
      "step": 7726
    },
    {
      "epoch": 0.3202370591404534,
      "grad_norm": 0.4243561327457428,
      "learning_rate": 3.399021923826102e-06,
      "loss": 0.7279,
      "step": 7727
    },
    {
      "epoch": 0.3202785030461271,
      "grad_norm": 0.42968595027923584,
      "learning_rate": 3.3988147042977333e-06,
      "loss": 0.7424,
      "step": 7728
    },
    {
      "epoch": 0.3203199469518007,
      "grad_norm": 0.3924400210380554,
      "learning_rate": 3.398607484769365e-06,
      "loss": 0.6787,
      "step": 7729
    },
    {
      "epoch": 0.3203613908574744,
      "grad_norm": 0.39011120796203613,
      "learning_rate": 3.3984002652409965e-06,
      "loss": 0.7041,
      "step": 7730
    },
    {
      "epoch": 0.32040283476314807,
      "grad_norm": 0.4148184061050415,
      "learning_rate": 3.3981930457126283e-06,
      "loss": 0.7153,
      "step": 7731
    },
    {
      "epoch": 0.32044427866882175,
      "grad_norm": 0.40022215247154236,
      "learning_rate": 3.3979858261842597e-06,
      "loss": 0.629,
      "step": 7732
    },
    {
      "epoch": 0.3204857225744954,
      "grad_norm": 0.4232144355773926,
      "learning_rate": 3.3977786066558915e-06,
      "loss": 0.7671,
      "step": 7733
    },
    {
      "epoch": 0.3205271664801691,
      "grad_norm": 0.38317179679870605,
      "learning_rate": 3.3975713871275233e-06,
      "loss": 0.7117,
      "step": 7734
    },
    {
      "epoch": 0.3205686103858428,
      "grad_norm": 0.4519965350627899,
      "learning_rate": 3.3973641675991547e-06,
      "loss": 0.7112,
      "step": 7735
    },
    {
      "epoch": 0.32061005429151646,
      "grad_norm": 0.45592793822288513,
      "learning_rate": 3.3971569480707865e-06,
      "loss": 0.7454,
      "step": 7736
    },
    {
      "epoch": 0.3206514981971901,
      "grad_norm": 0.4002833068370819,
      "learning_rate": 3.396949728542418e-06,
      "loss": 0.6779,
      "step": 7737
    },
    {
      "epoch": 0.32069294210286375,
      "grad_norm": 0.445517897605896,
      "learning_rate": 3.3967425090140497e-06,
      "loss": 0.7224,
      "step": 7738
    },
    {
      "epoch": 0.32073438600853743,
      "grad_norm": 0.37436121702194214,
      "learning_rate": 3.396535289485681e-06,
      "loss": 0.6455,
      "step": 7739
    },
    {
      "epoch": 0.3207758299142111,
      "grad_norm": 0.4186210632324219,
      "learning_rate": 3.3963280699573133e-06,
      "loss": 0.7043,
      "step": 7740
    },
    {
      "epoch": 0.3208172738198848,
      "grad_norm": 0.42991721630096436,
      "learning_rate": 3.3961208504289443e-06,
      "loss": 0.7201,
      "step": 7741
    },
    {
      "epoch": 0.32085871772555846,
      "grad_norm": 0.4248584806919098,
      "learning_rate": 3.3959136309005765e-06,
      "loss": 0.7576,
      "step": 7742
    },
    {
      "epoch": 0.32090016163123214,
      "grad_norm": 0.3895716667175293,
      "learning_rate": 3.3957064113722083e-06,
      "loss": 0.6924,
      "step": 7743
    },
    {
      "epoch": 0.3209416055369058,
      "grad_norm": 0.405115008354187,
      "learning_rate": 3.3954991918438397e-06,
      "loss": 0.6803,
      "step": 7744
    },
    {
      "epoch": 0.3209830494425795,
      "grad_norm": 0.41199082136154175,
      "learning_rate": 3.3952919723154715e-06,
      "loss": 0.6843,
      "step": 7745
    },
    {
      "epoch": 0.3210244933482531,
      "grad_norm": 0.4533744752407074,
      "learning_rate": 3.395084752787103e-06,
      "loss": 0.7114,
      "step": 7746
    },
    {
      "epoch": 0.3210659372539268,
      "grad_norm": 0.49249210953712463,
      "learning_rate": 3.3948775332587347e-06,
      "loss": 0.7623,
      "step": 7747
    },
    {
      "epoch": 0.32110738115960047,
      "grad_norm": 0.4302029311656952,
      "learning_rate": 3.394670313730366e-06,
      "loss": 0.738,
      "step": 7748
    },
    {
      "epoch": 0.32114882506527415,
      "grad_norm": 0.4031790494918823,
      "learning_rate": 3.394463094201998e-06,
      "loss": 0.6936,
      "step": 7749
    },
    {
      "epoch": 0.3211902689709478,
      "grad_norm": 0.42486757040023804,
      "learning_rate": 3.3942558746736293e-06,
      "loss": 0.6865,
      "step": 7750
    },
    {
      "epoch": 0.3212317128766215,
      "grad_norm": 0.42226642370224,
      "learning_rate": 3.394048655145261e-06,
      "loss": 0.7073,
      "step": 7751
    },
    {
      "epoch": 0.3212731567822952,
      "grad_norm": 0.4190801978111267,
      "learning_rate": 3.393841435616893e-06,
      "loss": 0.7295,
      "step": 7752
    },
    {
      "epoch": 0.32131460068796885,
      "grad_norm": 0.3991812765598297,
      "learning_rate": 3.3936342160885243e-06,
      "loss": 0.6738,
      "step": 7753
    },
    {
      "epoch": 0.32135604459364253,
      "grad_norm": 0.40885910391807556,
      "learning_rate": 3.393426996560156e-06,
      "loss": 0.6794,
      "step": 7754
    },
    {
      "epoch": 0.32139748849931615,
      "grad_norm": 0.4149489402770996,
      "learning_rate": 3.3932197770317875e-06,
      "loss": 0.7,
      "step": 7755
    },
    {
      "epoch": 0.32143893240498983,
      "grad_norm": 0.39827853441238403,
      "learning_rate": 3.3930125575034197e-06,
      "loss": 0.6685,
      "step": 7756
    },
    {
      "epoch": 0.3214803763106635,
      "grad_norm": 0.3996274173259735,
      "learning_rate": 3.3928053379750507e-06,
      "loss": 0.6794,
      "step": 7757
    },
    {
      "epoch": 0.3215218202163372,
      "grad_norm": 0.4261578321456909,
      "learning_rate": 3.392598118446683e-06,
      "loss": 0.7463,
      "step": 7758
    },
    {
      "epoch": 0.32156326412201086,
      "grad_norm": 0.4278627634048462,
      "learning_rate": 3.392390898918314e-06,
      "loss": 0.7573,
      "step": 7759
    },
    {
      "epoch": 0.32160470802768454,
      "grad_norm": 0.5598564743995667,
      "learning_rate": 3.392183679389946e-06,
      "loss": 0.7283,
      "step": 7760
    },
    {
      "epoch": 0.3216461519333582,
      "grad_norm": 0.4112405776977539,
      "learning_rate": 3.391976459861578e-06,
      "loss": 0.6989,
      "step": 7761
    },
    {
      "epoch": 0.3216875958390319,
      "grad_norm": 0.43529078364372253,
      "learning_rate": 3.3917692403332093e-06,
      "loss": 0.6851,
      "step": 7762
    },
    {
      "epoch": 0.3217290397447055,
      "grad_norm": 0.4011897146701813,
      "learning_rate": 3.391562020804841e-06,
      "loss": 0.7075,
      "step": 7763
    },
    {
      "epoch": 0.3217704836503792,
      "grad_norm": 0.40968990325927734,
      "learning_rate": 3.3913548012764725e-06,
      "loss": 0.7292,
      "step": 7764
    },
    {
      "epoch": 0.32181192755605287,
      "grad_norm": 0.4049062430858612,
      "learning_rate": 3.3911475817481043e-06,
      "loss": 0.7002,
      "step": 7765
    },
    {
      "epoch": 0.32185337146172655,
      "grad_norm": 0.42602822184562683,
      "learning_rate": 3.3909403622197357e-06,
      "loss": 0.7131,
      "step": 7766
    },
    {
      "epoch": 0.3218948153674002,
      "grad_norm": 0.4166877269744873,
      "learning_rate": 3.3907331426913675e-06,
      "loss": 0.7592,
      "step": 7767
    },
    {
      "epoch": 0.3219362592730739,
      "grad_norm": 0.46128278970718384,
      "learning_rate": 3.390525923162999e-06,
      "loss": 0.7183,
      "step": 7768
    },
    {
      "epoch": 0.3219777031787476,
      "grad_norm": 0.4284079670906067,
      "learning_rate": 3.3903187036346307e-06,
      "loss": 0.6708,
      "step": 7769
    },
    {
      "epoch": 0.32201914708442125,
      "grad_norm": 0.3837057948112488,
      "learning_rate": 3.3901114841062625e-06,
      "loss": 0.6599,
      "step": 7770
    },
    {
      "epoch": 0.32206059099009493,
      "grad_norm": 0.43382352590560913,
      "learning_rate": 3.389904264577894e-06,
      "loss": 0.7249,
      "step": 7771
    },
    {
      "epoch": 0.32210203489576855,
      "grad_norm": 0.42883485555648804,
      "learning_rate": 3.3896970450495257e-06,
      "loss": 0.7001,
      "step": 7772
    },
    {
      "epoch": 0.32214347880144223,
      "grad_norm": 0.4049244821071625,
      "learning_rate": 3.389489825521157e-06,
      "loss": 0.7267,
      "step": 7773
    },
    {
      "epoch": 0.3221849227071159,
      "grad_norm": 0.4542079269886017,
      "learning_rate": 3.3892826059927893e-06,
      "loss": 0.7319,
      "step": 7774
    },
    {
      "epoch": 0.3222263666127896,
      "grad_norm": 0.41243627667427063,
      "learning_rate": 3.3890753864644203e-06,
      "loss": 0.7002,
      "step": 7775
    },
    {
      "epoch": 0.32226781051846326,
      "grad_norm": 0.4199390709400177,
      "learning_rate": 3.3888681669360525e-06,
      "loss": 0.7174,
      "step": 7776
    },
    {
      "epoch": 0.32230925442413694,
      "grad_norm": 0.4294116497039795,
      "learning_rate": 3.3886609474076843e-06,
      "loss": 0.707,
      "step": 7777
    },
    {
      "epoch": 0.3223506983298106,
      "grad_norm": 0.43246713280677795,
      "learning_rate": 3.3884537278793157e-06,
      "loss": 0.686,
      "step": 7778
    },
    {
      "epoch": 0.3223921422354843,
      "grad_norm": 0.40518781542778015,
      "learning_rate": 3.3882465083509475e-06,
      "loss": 0.6604,
      "step": 7779
    },
    {
      "epoch": 0.32243358614115797,
      "grad_norm": 0.4680253863334656,
      "learning_rate": 3.388039288822579e-06,
      "loss": 0.7688,
      "step": 7780
    },
    {
      "epoch": 0.3224750300468316,
      "grad_norm": 0.39582008123397827,
      "learning_rate": 3.3878320692942107e-06,
      "loss": 0.6768,
      "step": 7781
    },
    {
      "epoch": 0.32251647395250527,
      "grad_norm": 0.39064061641693115,
      "learning_rate": 3.387624849765842e-06,
      "loss": 0.6823,
      "step": 7782
    },
    {
      "epoch": 0.32255791785817894,
      "grad_norm": 0.4117547571659088,
      "learning_rate": 3.387417630237474e-06,
      "loss": 0.7107,
      "step": 7783
    },
    {
      "epoch": 0.3225993617638526,
      "grad_norm": 0.4325833022594452,
      "learning_rate": 3.3872104107091053e-06,
      "loss": 0.691,
      "step": 7784
    },
    {
      "epoch": 0.3226408056695263,
      "grad_norm": 0.43816766142845154,
      "learning_rate": 3.387003191180737e-06,
      "loss": 0.6528,
      "step": 7785
    },
    {
      "epoch": 0.3226822495752,
      "grad_norm": 0.41078343987464905,
      "learning_rate": 3.386795971652369e-06,
      "loss": 0.6763,
      "step": 7786
    },
    {
      "epoch": 0.32272369348087365,
      "grad_norm": 0.46850088238716125,
      "learning_rate": 3.3865887521240003e-06,
      "loss": 0.7546,
      "step": 7787
    },
    {
      "epoch": 0.32276513738654733,
      "grad_norm": 0.3931032419204712,
      "learning_rate": 3.386381532595632e-06,
      "loss": 0.6455,
      "step": 7788
    },
    {
      "epoch": 0.322806581292221,
      "grad_norm": 0.41496604681015015,
      "learning_rate": 3.3861743130672635e-06,
      "loss": 0.7097,
      "step": 7789
    },
    {
      "epoch": 0.32284802519789463,
      "grad_norm": 0.4326360821723938,
      "learning_rate": 3.3859670935388957e-06,
      "loss": 0.714,
      "step": 7790
    },
    {
      "epoch": 0.3228894691035683,
      "grad_norm": 0.4061557352542877,
      "learning_rate": 3.3857598740105267e-06,
      "loss": 0.6421,
      "step": 7791
    },
    {
      "epoch": 0.322930913009242,
      "grad_norm": 0.4397035241127014,
      "learning_rate": 3.385552654482159e-06,
      "loss": 0.783,
      "step": 7792
    },
    {
      "epoch": 0.32297235691491566,
      "grad_norm": 0.43344151973724365,
      "learning_rate": 3.38534543495379e-06,
      "loss": 0.6632,
      "step": 7793
    },
    {
      "epoch": 0.32301380082058934,
      "grad_norm": 0.39462539553642273,
      "learning_rate": 3.385138215425422e-06,
      "loss": 0.7057,
      "step": 7794
    },
    {
      "epoch": 0.323055244726263,
      "grad_norm": 0.4175146520137787,
      "learning_rate": 3.384930995897054e-06,
      "loss": 0.684,
      "step": 7795
    },
    {
      "epoch": 0.3230966886319367,
      "grad_norm": 0.4242911636829376,
      "learning_rate": 3.3847237763686853e-06,
      "loss": 0.6768,
      "step": 7796
    },
    {
      "epoch": 0.32313813253761037,
      "grad_norm": 0.3953508138656616,
      "learning_rate": 3.384516556840317e-06,
      "loss": 0.6902,
      "step": 7797
    },
    {
      "epoch": 0.323179576443284,
      "grad_norm": 0.42934197187423706,
      "learning_rate": 3.3843093373119485e-06,
      "loss": 0.6953,
      "step": 7798
    },
    {
      "epoch": 0.32322102034895767,
      "grad_norm": 0.44775304198265076,
      "learning_rate": 3.3841021177835803e-06,
      "loss": 0.7391,
      "step": 7799
    },
    {
      "epoch": 0.32326246425463134,
      "grad_norm": 0.4067079424858093,
      "learning_rate": 3.3838948982552117e-06,
      "loss": 0.7035,
      "step": 7800
    },
    {
      "epoch": 0.323303908160305,
      "grad_norm": 0.43671005964279175,
      "learning_rate": 3.3836876787268435e-06,
      "loss": 0.7458,
      "step": 7801
    },
    {
      "epoch": 0.3233453520659787,
      "grad_norm": 0.42380115389823914,
      "learning_rate": 3.383480459198475e-06,
      "loss": 0.7039,
      "step": 7802
    },
    {
      "epoch": 0.3233867959716524,
      "grad_norm": 0.4287997782230377,
      "learning_rate": 3.3832732396701067e-06,
      "loss": 0.7029,
      "step": 7803
    },
    {
      "epoch": 0.32342823987732605,
      "grad_norm": 0.47196006774902344,
      "learning_rate": 3.3830660201417385e-06,
      "loss": 0.7653,
      "step": 7804
    },
    {
      "epoch": 0.32346968378299973,
      "grad_norm": 0.4171411991119385,
      "learning_rate": 3.38285880061337e-06,
      "loss": 0.6947,
      "step": 7805
    },
    {
      "epoch": 0.3235111276886734,
      "grad_norm": 0.4181687533855438,
      "learning_rate": 3.3826515810850017e-06,
      "loss": 0.709,
      "step": 7806
    },
    {
      "epoch": 0.323552571594347,
      "grad_norm": 0.4172302484512329,
      "learning_rate": 3.382444361556633e-06,
      "loss": 0.7303,
      "step": 7807
    },
    {
      "epoch": 0.3235940155000207,
      "grad_norm": 0.39349624514579773,
      "learning_rate": 3.3822371420282653e-06,
      "loss": 0.6726,
      "step": 7808
    },
    {
      "epoch": 0.3236354594056944,
      "grad_norm": 0.4072658121585846,
      "learning_rate": 3.3820299224998963e-06,
      "loss": 0.7061,
      "step": 7809
    },
    {
      "epoch": 0.32367690331136806,
      "grad_norm": 0.41630786657333374,
      "learning_rate": 3.3818227029715285e-06,
      "loss": 0.7384,
      "step": 7810
    },
    {
      "epoch": 0.32371834721704174,
      "grad_norm": 0.42587098479270935,
      "learning_rate": 3.38161548344316e-06,
      "loss": 0.7151,
      "step": 7811
    },
    {
      "epoch": 0.3237597911227154,
      "grad_norm": 0.3995664715766907,
      "learning_rate": 3.3814082639147917e-06,
      "loss": 0.6643,
      "step": 7812
    },
    {
      "epoch": 0.3238012350283891,
      "grad_norm": 0.43177494406700134,
      "learning_rate": 3.3812010443864235e-06,
      "loss": 0.7104,
      "step": 7813
    },
    {
      "epoch": 0.32384267893406277,
      "grad_norm": 0.4404847025871277,
      "learning_rate": 3.380993824858055e-06,
      "loss": 0.7566,
      "step": 7814
    },
    {
      "epoch": 0.32388412283973644,
      "grad_norm": 0.4310963451862335,
      "learning_rate": 3.3807866053296867e-06,
      "loss": 0.6676,
      "step": 7815
    },
    {
      "epoch": 0.32392556674541007,
      "grad_norm": 0.4372182786464691,
      "learning_rate": 3.380579385801318e-06,
      "loss": 0.7456,
      "step": 7816
    },
    {
      "epoch": 0.32396701065108374,
      "grad_norm": 0.4283396005630493,
      "learning_rate": 3.38037216627295e-06,
      "loss": 0.7419,
      "step": 7817
    },
    {
      "epoch": 0.3240084545567574,
      "grad_norm": 0.40074852108955383,
      "learning_rate": 3.3801649467445813e-06,
      "loss": 0.7122,
      "step": 7818
    },
    {
      "epoch": 0.3240498984624311,
      "grad_norm": 0.4354152977466583,
      "learning_rate": 3.379957727216213e-06,
      "loss": 0.7269,
      "step": 7819
    },
    {
      "epoch": 0.3240913423681048,
      "grad_norm": 0.4093610644340515,
      "learning_rate": 3.3797505076878445e-06,
      "loss": 0.6879,
      "step": 7820
    },
    {
      "epoch": 0.32413278627377845,
      "grad_norm": 0.4472990930080414,
      "learning_rate": 3.3795432881594763e-06,
      "loss": 0.6843,
      "step": 7821
    },
    {
      "epoch": 0.3241742301794521,
      "grad_norm": 0.4268062114715576,
      "learning_rate": 3.379336068631108e-06,
      "loss": 0.6139,
      "step": 7822
    },
    {
      "epoch": 0.3242156740851258,
      "grad_norm": 0.4400743544101715,
      "learning_rate": 3.3791288491027395e-06,
      "loss": 0.7249,
      "step": 7823
    },
    {
      "epoch": 0.3242571179907994,
      "grad_norm": 0.42358335852622986,
      "learning_rate": 3.3789216295743717e-06,
      "loss": 0.6815,
      "step": 7824
    },
    {
      "epoch": 0.3242985618964731,
      "grad_norm": 0.4414313733577728,
      "learning_rate": 3.3787144100460027e-06,
      "loss": 0.7075,
      "step": 7825
    },
    {
      "epoch": 0.3243400058021468,
      "grad_norm": 0.4234028160572052,
      "learning_rate": 3.378507190517635e-06,
      "loss": 0.7173,
      "step": 7826
    },
    {
      "epoch": 0.32438144970782046,
      "grad_norm": 0.4101279377937317,
      "learning_rate": 3.3782999709892663e-06,
      "loss": 0.7098,
      "step": 7827
    },
    {
      "epoch": 0.32442289361349413,
      "grad_norm": 0.423184335231781,
      "learning_rate": 3.378092751460898e-06,
      "loss": 0.6576,
      "step": 7828
    },
    {
      "epoch": 0.3244643375191678,
      "grad_norm": 0.41824302077293396,
      "learning_rate": 3.3778855319325295e-06,
      "loss": 0.6261,
      "step": 7829
    },
    {
      "epoch": 0.3245057814248415,
      "grad_norm": 0.428160697221756,
      "learning_rate": 3.3776783124041613e-06,
      "loss": 0.7188,
      "step": 7830
    },
    {
      "epoch": 0.32454722533051517,
      "grad_norm": 0.40940743684768677,
      "learning_rate": 3.377471092875793e-06,
      "loss": 0.7662,
      "step": 7831
    },
    {
      "epoch": 0.32458866923618884,
      "grad_norm": 0.47021254897117615,
      "learning_rate": 3.3772638733474245e-06,
      "loss": 0.7878,
      "step": 7832
    },
    {
      "epoch": 0.32463011314186246,
      "grad_norm": 0.42245328426361084,
      "learning_rate": 3.3770566538190563e-06,
      "loss": 0.7271,
      "step": 7833
    },
    {
      "epoch": 0.32467155704753614,
      "grad_norm": 0.42557018995285034,
      "learning_rate": 3.3768494342906877e-06,
      "loss": 0.7012,
      "step": 7834
    },
    {
      "epoch": 0.3247130009532098,
      "grad_norm": 0.43708133697509766,
      "learning_rate": 3.3766422147623195e-06,
      "loss": 0.6951,
      "step": 7835
    },
    {
      "epoch": 0.3247544448588835,
      "grad_norm": 0.46211957931518555,
      "learning_rate": 3.376434995233951e-06,
      "loss": 0.7648,
      "step": 7836
    },
    {
      "epoch": 0.3247958887645572,
      "grad_norm": 0.3861435055732727,
      "learning_rate": 3.3762277757055827e-06,
      "loss": 0.6456,
      "step": 7837
    },
    {
      "epoch": 0.32483733267023085,
      "grad_norm": 0.38586193323135376,
      "learning_rate": 3.3760205561772145e-06,
      "loss": 0.6901,
      "step": 7838
    },
    {
      "epoch": 0.3248787765759045,
      "grad_norm": 0.4112374484539032,
      "learning_rate": 3.375813336648846e-06,
      "loss": 0.6674,
      "step": 7839
    },
    {
      "epoch": 0.3249202204815782,
      "grad_norm": 0.3697628974914551,
      "learning_rate": 3.3756061171204777e-06,
      "loss": 0.6833,
      "step": 7840
    },
    {
      "epoch": 0.3249616643872519,
      "grad_norm": 0.45484375953674316,
      "learning_rate": 3.375398897592109e-06,
      "loss": 0.7439,
      "step": 7841
    },
    {
      "epoch": 0.3250031082929255,
      "grad_norm": 0.4441000521183014,
      "learning_rate": 3.3751916780637413e-06,
      "loss": 0.7068,
      "step": 7842
    },
    {
      "epoch": 0.3250445521985992,
      "grad_norm": 0.4480298161506653,
      "learning_rate": 3.3749844585353723e-06,
      "loss": 0.7402,
      "step": 7843
    },
    {
      "epoch": 0.32508599610427286,
      "grad_norm": 0.3955453634262085,
      "learning_rate": 3.3747772390070045e-06,
      "loss": 0.658,
      "step": 7844
    },
    {
      "epoch": 0.32512744000994653,
      "grad_norm": 0.41058430075645447,
      "learning_rate": 3.374570019478636e-06,
      "loss": 0.6842,
      "step": 7845
    },
    {
      "epoch": 0.3251688839156202,
      "grad_norm": 0.41125938296318054,
      "learning_rate": 3.3743627999502677e-06,
      "loss": 0.7217,
      "step": 7846
    },
    {
      "epoch": 0.3252103278212939,
      "grad_norm": 0.4203391373157501,
      "learning_rate": 3.3741555804218995e-06,
      "loss": 0.6921,
      "step": 7847
    },
    {
      "epoch": 0.32525177172696756,
      "grad_norm": 0.43751221895217896,
      "learning_rate": 3.373948360893531e-06,
      "loss": 0.7444,
      "step": 7848
    },
    {
      "epoch": 0.32529321563264124,
      "grad_norm": 0.434929758310318,
      "learning_rate": 3.3737411413651627e-06,
      "loss": 0.7935,
      "step": 7849
    },
    {
      "epoch": 0.32533465953831486,
      "grad_norm": 0.44635188579559326,
      "learning_rate": 3.373533921836794e-06,
      "loss": 0.731,
      "step": 7850
    },
    {
      "epoch": 0.32537610344398854,
      "grad_norm": 0.45439258217811584,
      "learning_rate": 3.373326702308426e-06,
      "loss": 0.7327,
      "step": 7851
    },
    {
      "epoch": 0.3254175473496622,
      "grad_norm": 0.4235478937625885,
      "learning_rate": 3.3731194827800573e-06,
      "loss": 0.6847,
      "step": 7852
    },
    {
      "epoch": 0.3254589912553359,
      "grad_norm": 0.41780802607536316,
      "learning_rate": 3.372912263251689e-06,
      "loss": 0.6902,
      "step": 7853
    },
    {
      "epoch": 0.32550043516100957,
      "grad_norm": 0.39976009726524353,
      "learning_rate": 3.3727050437233205e-06,
      "loss": 0.6765,
      "step": 7854
    },
    {
      "epoch": 0.32554187906668325,
      "grad_norm": 0.40875163674354553,
      "learning_rate": 3.3724978241949523e-06,
      "loss": 0.7148,
      "step": 7855
    },
    {
      "epoch": 0.3255833229723569,
      "grad_norm": 0.40105557441711426,
      "learning_rate": 3.372290604666584e-06,
      "loss": 0.6748,
      "step": 7856
    },
    {
      "epoch": 0.3256247668780306,
      "grad_norm": 0.39805611968040466,
      "learning_rate": 3.3720833851382155e-06,
      "loss": 0.7112,
      "step": 7857
    },
    {
      "epoch": 0.3256662107837043,
      "grad_norm": 0.4086209833621979,
      "learning_rate": 3.3718761656098477e-06,
      "loss": 0.744,
      "step": 7858
    },
    {
      "epoch": 0.3257076546893779,
      "grad_norm": 0.4500320255756378,
      "learning_rate": 3.3716689460814787e-06,
      "loss": 0.7458,
      "step": 7859
    },
    {
      "epoch": 0.3257490985950516,
      "grad_norm": 0.42177948355674744,
      "learning_rate": 3.371461726553111e-06,
      "loss": 0.7153,
      "step": 7860
    },
    {
      "epoch": 0.32579054250072526,
      "grad_norm": 0.4524421691894531,
      "learning_rate": 3.3712545070247423e-06,
      "loss": 0.6619,
      "step": 7861
    },
    {
      "epoch": 0.32583198640639893,
      "grad_norm": 0.4058699905872345,
      "learning_rate": 3.371047287496374e-06,
      "loss": 0.6265,
      "step": 7862
    },
    {
      "epoch": 0.3258734303120726,
      "grad_norm": 0.457205206155777,
      "learning_rate": 3.3708400679680055e-06,
      "loss": 0.7405,
      "step": 7863
    },
    {
      "epoch": 0.3259148742177463,
      "grad_norm": 0.4275912642478943,
      "learning_rate": 3.3706328484396373e-06,
      "loss": 0.74,
      "step": 7864
    },
    {
      "epoch": 0.32595631812341996,
      "grad_norm": 0.41272756457328796,
      "learning_rate": 3.370425628911269e-06,
      "loss": 0.645,
      "step": 7865
    },
    {
      "epoch": 0.32599776202909364,
      "grad_norm": 0.3926021158695221,
      "learning_rate": 3.3702184093829005e-06,
      "loss": 0.6965,
      "step": 7866
    },
    {
      "epoch": 0.3260392059347673,
      "grad_norm": 0.39076176285743713,
      "learning_rate": 3.3700111898545323e-06,
      "loss": 0.6956,
      "step": 7867
    },
    {
      "epoch": 0.32608064984044094,
      "grad_norm": 0.42536666989326477,
      "learning_rate": 3.3698039703261637e-06,
      "loss": 0.6896,
      "step": 7868
    },
    {
      "epoch": 0.3261220937461146,
      "grad_norm": 0.3805718421936035,
      "learning_rate": 3.3695967507977955e-06,
      "loss": 0.6987,
      "step": 7869
    },
    {
      "epoch": 0.3261635376517883,
      "grad_norm": 0.3970760703086853,
      "learning_rate": 3.369389531269427e-06,
      "loss": 0.7068,
      "step": 7870
    },
    {
      "epoch": 0.32620498155746197,
      "grad_norm": 0.44920942187309265,
      "learning_rate": 3.3691823117410587e-06,
      "loss": 0.7263,
      "step": 7871
    },
    {
      "epoch": 0.32624642546313565,
      "grad_norm": 0.42040398716926575,
      "learning_rate": 3.36897509221269e-06,
      "loss": 0.7151,
      "step": 7872
    },
    {
      "epoch": 0.3262878693688093,
      "grad_norm": 0.4255923926830292,
      "learning_rate": 3.368767872684322e-06,
      "loss": 0.7292,
      "step": 7873
    },
    {
      "epoch": 0.326329313274483,
      "grad_norm": 0.41248470544815063,
      "learning_rate": 3.368560653155954e-06,
      "loss": 0.7375,
      "step": 7874
    },
    {
      "epoch": 0.3263707571801567,
      "grad_norm": 0.42222097516059875,
      "learning_rate": 3.368353433627585e-06,
      "loss": 0.7206,
      "step": 7875
    },
    {
      "epoch": 0.32641220108583036,
      "grad_norm": 0.389122873544693,
      "learning_rate": 3.3681462140992173e-06,
      "loss": 0.6531,
      "step": 7876
    },
    {
      "epoch": 0.326453644991504,
      "grad_norm": 0.44389376044273376,
      "learning_rate": 3.3679389945708483e-06,
      "loss": 0.7,
      "step": 7877
    },
    {
      "epoch": 0.32649508889717765,
      "grad_norm": 0.39713814854621887,
      "learning_rate": 3.3677317750424805e-06,
      "loss": 0.719,
      "step": 7878
    },
    {
      "epoch": 0.32653653280285133,
      "grad_norm": 0.4517744779586792,
      "learning_rate": 3.367524555514112e-06,
      "loss": 0.7649,
      "step": 7879
    },
    {
      "epoch": 0.326577976708525,
      "grad_norm": 0.4171895980834961,
      "learning_rate": 3.3673173359857437e-06,
      "loss": 0.696,
      "step": 7880
    },
    {
      "epoch": 0.3266194206141987,
      "grad_norm": 0.46063557267189026,
      "learning_rate": 3.367110116457375e-06,
      "loss": 0.7676,
      "step": 7881
    },
    {
      "epoch": 0.32666086451987236,
      "grad_norm": 0.4462823271751404,
      "learning_rate": 3.366902896929007e-06,
      "loss": 0.7163,
      "step": 7882
    },
    {
      "epoch": 0.32670230842554604,
      "grad_norm": 0.4188307225704193,
      "learning_rate": 3.3666956774006387e-06,
      "loss": 0.696,
      "step": 7883
    },
    {
      "epoch": 0.3267437523312197,
      "grad_norm": 0.41473913192749023,
      "learning_rate": 3.36648845787227e-06,
      "loss": 0.6676,
      "step": 7884
    },
    {
      "epoch": 0.32678519623689334,
      "grad_norm": 0.42960110306739807,
      "learning_rate": 3.366281238343902e-06,
      "loss": 0.687,
      "step": 7885
    },
    {
      "epoch": 0.326826640142567,
      "grad_norm": 0.40732842683792114,
      "learning_rate": 3.3660740188155333e-06,
      "loss": 0.6691,
      "step": 7886
    },
    {
      "epoch": 0.3268680840482407,
      "grad_norm": 0.43849077820777893,
      "learning_rate": 3.365866799287165e-06,
      "loss": 0.7089,
      "step": 7887
    },
    {
      "epoch": 0.32690952795391437,
      "grad_norm": 0.4181516170501709,
      "learning_rate": 3.3656595797587965e-06,
      "loss": 0.6753,
      "step": 7888
    },
    {
      "epoch": 0.32695097185958805,
      "grad_norm": 0.4406169652938843,
      "learning_rate": 3.3654523602304283e-06,
      "loss": 0.7026,
      "step": 7889
    },
    {
      "epoch": 0.3269924157652617,
      "grad_norm": 0.4158298671245575,
      "learning_rate": 3.3652451407020597e-06,
      "loss": 0.6913,
      "step": 7890
    },
    {
      "epoch": 0.3270338596709354,
      "grad_norm": 0.44252318143844604,
      "learning_rate": 3.3650379211736915e-06,
      "loss": 0.7432,
      "step": 7891
    },
    {
      "epoch": 0.3270753035766091,
      "grad_norm": 0.42553484439849854,
      "learning_rate": 3.3648307016453237e-06,
      "loss": 0.7247,
      "step": 7892
    },
    {
      "epoch": 0.32711674748228275,
      "grad_norm": 0.42196255922317505,
      "learning_rate": 3.3646234821169547e-06,
      "loss": 0.7103,
      "step": 7893
    },
    {
      "epoch": 0.3271581913879564,
      "grad_norm": 0.42846500873565674,
      "learning_rate": 3.364416262588587e-06,
      "loss": 0.7192,
      "step": 7894
    },
    {
      "epoch": 0.32719963529363005,
      "grad_norm": 0.395893394947052,
      "learning_rate": 3.3642090430602183e-06,
      "loss": 0.6829,
      "step": 7895
    },
    {
      "epoch": 0.32724107919930373,
      "grad_norm": 0.41526922583580017,
      "learning_rate": 3.36400182353185e-06,
      "loss": 0.7207,
      "step": 7896
    },
    {
      "epoch": 0.3272825231049774,
      "grad_norm": 0.3700304627418518,
      "learning_rate": 3.3637946040034815e-06,
      "loss": 0.6886,
      "step": 7897
    },
    {
      "epoch": 0.3273239670106511,
      "grad_norm": 0.41648176312446594,
      "learning_rate": 3.3635873844751133e-06,
      "loss": 0.7234,
      "step": 7898
    },
    {
      "epoch": 0.32736541091632476,
      "grad_norm": 0.43109601736068726,
      "learning_rate": 3.363380164946745e-06,
      "loss": 0.7903,
      "step": 7899
    },
    {
      "epoch": 0.32740685482199844,
      "grad_norm": 0.4177792966365814,
      "learning_rate": 3.3631729454183765e-06,
      "loss": 0.7197,
      "step": 7900
    },
    {
      "epoch": 0.3274482987276721,
      "grad_norm": 0.4625702500343323,
      "learning_rate": 3.3629657258900083e-06,
      "loss": 0.7349,
      "step": 7901
    },
    {
      "epoch": 0.3274897426333458,
      "grad_norm": 0.39170798659324646,
      "learning_rate": 3.3627585063616397e-06,
      "loss": 0.707,
      "step": 7902
    },
    {
      "epoch": 0.3275311865390194,
      "grad_norm": 0.4053094983100891,
      "learning_rate": 3.3625512868332715e-06,
      "loss": 0.7285,
      "step": 7903
    },
    {
      "epoch": 0.3275726304446931,
      "grad_norm": 0.4121449291706085,
      "learning_rate": 3.362344067304903e-06,
      "loss": 0.7043,
      "step": 7904
    },
    {
      "epoch": 0.32761407435036677,
      "grad_norm": 0.43291008472442627,
      "learning_rate": 3.3621368477765347e-06,
      "loss": 0.6731,
      "step": 7905
    },
    {
      "epoch": 0.32765551825604045,
      "grad_norm": 0.41992101073265076,
      "learning_rate": 3.361929628248166e-06,
      "loss": 0.7046,
      "step": 7906
    },
    {
      "epoch": 0.3276969621617141,
      "grad_norm": 0.4211881756782532,
      "learning_rate": 3.361722408719798e-06,
      "loss": 0.7202,
      "step": 7907
    },
    {
      "epoch": 0.3277384060673878,
      "grad_norm": 0.4671046733856201,
      "learning_rate": 3.36151518919143e-06,
      "loss": 0.7828,
      "step": 7908
    },
    {
      "epoch": 0.3277798499730615,
      "grad_norm": 0.4107086658477783,
      "learning_rate": 3.361307969663061e-06,
      "loss": 0.6829,
      "step": 7909
    },
    {
      "epoch": 0.32782129387873515,
      "grad_norm": 0.39492759108543396,
      "learning_rate": 3.3611007501346933e-06,
      "loss": 0.676,
      "step": 7910
    },
    {
      "epoch": 0.3278627377844088,
      "grad_norm": 0.41530975699424744,
      "learning_rate": 3.3608935306063243e-06,
      "loss": 0.6831,
      "step": 7911
    },
    {
      "epoch": 0.32790418169008245,
      "grad_norm": 0.39978304505348206,
      "learning_rate": 3.3606863110779565e-06,
      "loss": 0.718,
      "step": 7912
    },
    {
      "epoch": 0.32794562559575613,
      "grad_norm": 0.42126598954200745,
      "learning_rate": 3.360479091549588e-06,
      "loss": 0.7046,
      "step": 7913
    },
    {
      "epoch": 0.3279870695014298,
      "grad_norm": 0.4389324188232422,
      "learning_rate": 3.3602718720212197e-06,
      "loss": 0.7102,
      "step": 7914
    },
    {
      "epoch": 0.3280285134071035,
      "grad_norm": 0.4299924671649933,
      "learning_rate": 3.360064652492851e-06,
      "loss": 0.6799,
      "step": 7915
    },
    {
      "epoch": 0.32806995731277716,
      "grad_norm": 0.4222584366798401,
      "learning_rate": 3.359857432964483e-06,
      "loss": 0.6741,
      "step": 7916
    },
    {
      "epoch": 0.32811140121845084,
      "grad_norm": 0.44245246052742004,
      "learning_rate": 3.3596502134361147e-06,
      "loss": 0.7117,
      "step": 7917
    },
    {
      "epoch": 0.3281528451241245,
      "grad_norm": 0.39610254764556885,
      "learning_rate": 3.359442993907746e-06,
      "loss": 0.7102,
      "step": 7918
    },
    {
      "epoch": 0.3281942890297982,
      "grad_norm": 0.406148225069046,
      "learning_rate": 3.359235774379378e-06,
      "loss": 0.7131,
      "step": 7919
    },
    {
      "epoch": 0.3282357329354718,
      "grad_norm": 0.40056148171424866,
      "learning_rate": 3.3590285548510093e-06,
      "loss": 0.7061,
      "step": 7920
    },
    {
      "epoch": 0.3282771768411455,
      "grad_norm": 0.41547709703445435,
      "learning_rate": 3.358821335322641e-06,
      "loss": 0.7144,
      "step": 7921
    },
    {
      "epoch": 0.32831862074681917,
      "grad_norm": 0.4306262731552124,
      "learning_rate": 3.3586141157942725e-06,
      "loss": 0.741,
      "step": 7922
    },
    {
      "epoch": 0.32836006465249284,
      "grad_norm": 0.40384969115257263,
      "learning_rate": 3.3584068962659043e-06,
      "loss": 0.6812,
      "step": 7923
    },
    {
      "epoch": 0.3284015085581665,
      "grad_norm": 0.38736841082572937,
      "learning_rate": 3.3581996767375357e-06,
      "loss": 0.707,
      "step": 7924
    },
    {
      "epoch": 0.3284429524638402,
      "grad_norm": 0.3965844511985779,
      "learning_rate": 3.3579924572091675e-06,
      "loss": 0.6904,
      "step": 7925
    },
    {
      "epoch": 0.3284843963695139,
      "grad_norm": 0.4387584924697876,
      "learning_rate": 3.3577852376807997e-06,
      "loss": 0.6973,
      "step": 7926
    },
    {
      "epoch": 0.32852584027518755,
      "grad_norm": 0.40261057019233704,
      "learning_rate": 3.3575780181524307e-06,
      "loss": 0.6661,
      "step": 7927
    },
    {
      "epoch": 0.32856728418086123,
      "grad_norm": 0.6934207081794739,
      "learning_rate": 3.357370798624063e-06,
      "loss": 0.6597,
      "step": 7928
    },
    {
      "epoch": 0.32860872808653485,
      "grad_norm": 0.4315175414085388,
      "learning_rate": 3.3571635790956943e-06,
      "loss": 0.6582,
      "step": 7929
    },
    {
      "epoch": 0.32865017199220853,
      "grad_norm": 0.46121904253959656,
      "learning_rate": 3.356956359567326e-06,
      "loss": 0.7393,
      "step": 7930
    },
    {
      "epoch": 0.3286916158978822,
      "grad_norm": 0.4215610921382904,
      "learning_rate": 3.3567491400389575e-06,
      "loss": 0.6438,
      "step": 7931
    },
    {
      "epoch": 0.3287330598035559,
      "grad_norm": 0.4495464861392975,
      "learning_rate": 3.3565419205105893e-06,
      "loss": 0.7747,
      "step": 7932
    },
    {
      "epoch": 0.32877450370922956,
      "grad_norm": 0.40555819869041443,
      "learning_rate": 3.3563347009822207e-06,
      "loss": 0.6471,
      "step": 7933
    },
    {
      "epoch": 0.32881594761490324,
      "grad_norm": 0.42446568608283997,
      "learning_rate": 3.3561274814538525e-06,
      "loss": 0.7407,
      "step": 7934
    },
    {
      "epoch": 0.3288573915205769,
      "grad_norm": 0.4170495867729187,
      "learning_rate": 3.3559202619254843e-06,
      "loss": 0.6912,
      "step": 7935
    },
    {
      "epoch": 0.3288988354262506,
      "grad_norm": 0.4335777461528778,
      "learning_rate": 3.3557130423971157e-06,
      "loss": 0.7092,
      "step": 7936
    },
    {
      "epoch": 0.32894027933192427,
      "grad_norm": 0.4119087755680084,
      "learning_rate": 3.3555058228687475e-06,
      "loss": 0.6326,
      "step": 7937
    },
    {
      "epoch": 0.3289817232375979,
      "grad_norm": 0.41475436091423035,
      "learning_rate": 3.355298603340379e-06,
      "loss": 0.6975,
      "step": 7938
    },
    {
      "epoch": 0.32902316714327157,
      "grad_norm": 0.41199180483818054,
      "learning_rate": 3.3550913838120107e-06,
      "loss": 0.7263,
      "step": 7939
    },
    {
      "epoch": 0.32906461104894524,
      "grad_norm": 0.40641072392463684,
      "learning_rate": 3.354884164283642e-06,
      "loss": 0.662,
      "step": 7940
    },
    {
      "epoch": 0.3291060549546189,
      "grad_norm": 0.430259644985199,
      "learning_rate": 3.354676944755274e-06,
      "loss": 0.7074,
      "step": 7941
    },
    {
      "epoch": 0.3291474988602926,
      "grad_norm": 0.38428762555122375,
      "learning_rate": 3.3544697252269053e-06,
      "loss": 0.693,
      "step": 7942
    },
    {
      "epoch": 0.3291889427659663,
      "grad_norm": 0.389007031917572,
      "learning_rate": 3.354262505698537e-06,
      "loss": 0.7434,
      "step": 7943
    },
    {
      "epoch": 0.32923038667163995,
      "grad_norm": 0.4037756323814392,
      "learning_rate": 3.3540552861701693e-06,
      "loss": 0.6674,
      "step": 7944
    },
    {
      "epoch": 0.32927183057731363,
      "grad_norm": 0.3862777650356293,
      "learning_rate": 3.3538480666418003e-06,
      "loss": 0.6833,
      "step": 7945
    },
    {
      "epoch": 0.32931327448298725,
      "grad_norm": 0.3925323188304901,
      "learning_rate": 3.3536408471134325e-06,
      "loss": 0.7014,
      "step": 7946
    },
    {
      "epoch": 0.3293547183886609,
      "grad_norm": 0.40981143712997437,
      "learning_rate": 3.353433627585064e-06,
      "loss": 0.668,
      "step": 7947
    },
    {
      "epoch": 0.3293961622943346,
      "grad_norm": 0.41306132078170776,
      "learning_rate": 3.3532264080566957e-06,
      "loss": 0.676,
      "step": 7948
    },
    {
      "epoch": 0.3294376062000083,
      "grad_norm": 0.40375250577926636,
      "learning_rate": 3.353019188528327e-06,
      "loss": 0.6605,
      "step": 7949
    },
    {
      "epoch": 0.32947905010568196,
      "grad_norm": 0.4091317355632782,
      "learning_rate": 3.352811968999959e-06,
      "loss": 0.6715,
      "step": 7950
    },
    {
      "epoch": 0.32952049401135564,
      "grad_norm": 0.4747426211833954,
      "learning_rate": 3.3526047494715903e-06,
      "loss": 0.6719,
      "step": 7951
    },
    {
      "epoch": 0.3295619379170293,
      "grad_norm": 0.47717249393463135,
      "learning_rate": 3.352397529943222e-06,
      "loss": 0.7605,
      "step": 7952
    },
    {
      "epoch": 0.329603381822703,
      "grad_norm": 0.4491930902004242,
      "learning_rate": 3.352190310414854e-06,
      "loss": 0.7102,
      "step": 7953
    },
    {
      "epoch": 0.32964482572837667,
      "grad_norm": 0.40875571966171265,
      "learning_rate": 3.3519830908864853e-06,
      "loss": 0.7361,
      "step": 7954
    },
    {
      "epoch": 0.3296862696340503,
      "grad_norm": 0.4491513669490814,
      "learning_rate": 3.351775871358117e-06,
      "loss": 0.7107,
      "step": 7955
    },
    {
      "epoch": 0.32972771353972397,
      "grad_norm": 0.3997478485107422,
      "learning_rate": 3.3515686518297485e-06,
      "loss": 0.6857,
      "step": 7956
    },
    {
      "epoch": 0.32976915744539764,
      "grad_norm": 0.42646029591560364,
      "learning_rate": 3.3513614323013803e-06,
      "loss": 0.6965,
      "step": 7957
    },
    {
      "epoch": 0.3298106013510713,
      "grad_norm": 0.4220108091831207,
      "learning_rate": 3.3511542127730117e-06,
      "loss": 0.7163,
      "step": 7958
    },
    {
      "epoch": 0.329852045256745,
      "grad_norm": 0.4226192533969879,
      "learning_rate": 3.3509469932446435e-06,
      "loss": 0.7607,
      "step": 7959
    },
    {
      "epoch": 0.3298934891624187,
      "grad_norm": 0.39844655990600586,
      "learning_rate": 3.3507397737162758e-06,
      "loss": 0.6787,
      "step": 7960
    },
    {
      "epoch": 0.32993493306809235,
      "grad_norm": 0.4171030521392822,
      "learning_rate": 3.3505325541879067e-06,
      "loss": 0.6893,
      "step": 7961
    },
    {
      "epoch": 0.32997637697376603,
      "grad_norm": 0.4193304181098938,
      "learning_rate": 3.350325334659539e-06,
      "loss": 0.7031,
      "step": 7962
    },
    {
      "epoch": 0.3300178208794397,
      "grad_norm": 0.39458978176116943,
      "learning_rate": 3.3501181151311703e-06,
      "loss": 0.696,
      "step": 7963
    },
    {
      "epoch": 0.3300592647851133,
      "grad_norm": 0.39896777272224426,
      "learning_rate": 3.349910895602802e-06,
      "loss": 0.7295,
      "step": 7964
    },
    {
      "epoch": 0.330100708690787,
      "grad_norm": 0.4222836196422577,
      "learning_rate": 3.3497036760744335e-06,
      "loss": 0.726,
      "step": 7965
    },
    {
      "epoch": 0.3301421525964607,
      "grad_norm": 0.4341786503791809,
      "learning_rate": 3.3494964565460653e-06,
      "loss": 0.7219,
      "step": 7966
    },
    {
      "epoch": 0.33018359650213436,
      "grad_norm": 0.37430259585380554,
      "learning_rate": 3.3492892370176967e-06,
      "loss": 0.6487,
      "step": 7967
    },
    {
      "epoch": 0.33022504040780803,
      "grad_norm": 0.41345641016960144,
      "learning_rate": 3.3490820174893285e-06,
      "loss": 0.6907,
      "step": 7968
    },
    {
      "epoch": 0.3302664843134817,
      "grad_norm": 0.3859317898750305,
      "learning_rate": 3.3488747979609603e-06,
      "loss": 0.6832,
      "step": 7969
    },
    {
      "epoch": 0.3303079282191554,
      "grad_norm": 0.42162033915519714,
      "learning_rate": 3.3486675784325917e-06,
      "loss": 0.6761,
      "step": 7970
    },
    {
      "epoch": 0.33034937212482907,
      "grad_norm": 0.40062597393989563,
      "learning_rate": 3.3484603589042235e-06,
      "loss": 0.6755,
      "step": 7971
    },
    {
      "epoch": 0.3303908160305027,
      "grad_norm": 0.39600634574890137,
      "learning_rate": 3.348253139375855e-06,
      "loss": 0.6921,
      "step": 7972
    },
    {
      "epoch": 0.33043225993617636,
      "grad_norm": 0.44521716237068176,
      "learning_rate": 3.3480459198474867e-06,
      "loss": 0.7075,
      "step": 7973
    },
    {
      "epoch": 0.33047370384185004,
      "grad_norm": 0.4115220904350281,
      "learning_rate": 3.347838700319118e-06,
      "loss": 0.7109,
      "step": 7974
    },
    {
      "epoch": 0.3305151477475237,
      "grad_norm": 0.45592111349105835,
      "learning_rate": 3.34763148079075e-06,
      "loss": 0.696,
      "step": 7975
    },
    {
      "epoch": 0.3305565916531974,
      "grad_norm": 0.45308563113212585,
      "learning_rate": 3.3474242612623813e-06,
      "loss": 0.7649,
      "step": 7976
    },
    {
      "epoch": 0.3305980355588711,
      "grad_norm": 0.3784540295600891,
      "learning_rate": 3.347217041734013e-06,
      "loss": 0.6897,
      "step": 7977
    },
    {
      "epoch": 0.33063947946454475,
      "grad_norm": 0.4064774513244629,
      "learning_rate": 3.3470098222056454e-06,
      "loss": 0.6992,
      "step": 7978
    },
    {
      "epoch": 0.3306809233702184,
      "grad_norm": 0.39900341629981995,
      "learning_rate": 3.3468026026772763e-06,
      "loss": 0.6704,
      "step": 7979
    },
    {
      "epoch": 0.3307223672758921,
      "grad_norm": 0.41627398133277893,
      "learning_rate": 3.3465953831489085e-06,
      "loss": 0.6719,
      "step": 7980
    },
    {
      "epoch": 0.3307638111815657,
      "grad_norm": 0.45161765813827515,
      "learning_rate": 3.34638816362054e-06,
      "loss": 0.6938,
      "step": 7981
    },
    {
      "epoch": 0.3308052550872394,
      "grad_norm": 0.49716946482658386,
      "learning_rate": 3.3461809440921717e-06,
      "loss": 0.7283,
      "step": 7982
    },
    {
      "epoch": 0.3308466989929131,
      "grad_norm": 0.3851609528064728,
      "learning_rate": 3.345973724563803e-06,
      "loss": 0.7109,
      "step": 7983
    },
    {
      "epoch": 0.33088814289858676,
      "grad_norm": 0.39520880579948425,
      "learning_rate": 3.345766505035435e-06,
      "loss": 0.7156,
      "step": 7984
    },
    {
      "epoch": 0.33092958680426043,
      "grad_norm": 0.449020117521286,
      "learning_rate": 3.3455592855070663e-06,
      "loss": 0.7263,
      "step": 7985
    },
    {
      "epoch": 0.3309710307099341,
      "grad_norm": 0.43262919783592224,
      "learning_rate": 3.345352065978698e-06,
      "loss": 0.6943,
      "step": 7986
    },
    {
      "epoch": 0.3310124746156078,
      "grad_norm": 0.4304153025150299,
      "learning_rate": 3.34514484645033e-06,
      "loss": 0.7317,
      "step": 7987
    },
    {
      "epoch": 0.33105391852128147,
      "grad_norm": 0.38980814814567566,
      "learning_rate": 3.3449376269219613e-06,
      "loss": 0.6842,
      "step": 7988
    },
    {
      "epoch": 0.33109536242695514,
      "grad_norm": 0.44170382618904114,
      "learning_rate": 3.344730407393593e-06,
      "loss": 0.7313,
      "step": 7989
    },
    {
      "epoch": 0.33113680633262876,
      "grad_norm": 0.40107449889183044,
      "learning_rate": 3.3445231878652245e-06,
      "loss": 0.6997,
      "step": 7990
    },
    {
      "epoch": 0.33117825023830244,
      "grad_norm": 0.38986918330192566,
      "learning_rate": 3.3443159683368563e-06,
      "loss": 0.7222,
      "step": 7991
    },
    {
      "epoch": 0.3312196941439761,
      "grad_norm": 0.40724360942840576,
      "learning_rate": 3.3441087488084877e-06,
      "loss": 0.7181,
      "step": 7992
    },
    {
      "epoch": 0.3312611380496498,
      "grad_norm": 0.4041632115840912,
      "learning_rate": 3.3439015292801195e-06,
      "loss": 0.7472,
      "step": 7993
    },
    {
      "epoch": 0.33130258195532347,
      "grad_norm": 0.4228819012641907,
      "learning_rate": 3.343694309751751e-06,
      "loss": 0.7063,
      "step": 7994
    },
    {
      "epoch": 0.33134402586099715,
      "grad_norm": 0.4062122106552124,
      "learning_rate": 3.3434870902233827e-06,
      "loss": 0.7089,
      "step": 7995
    },
    {
      "epoch": 0.3313854697666708,
      "grad_norm": 0.4325317442417145,
      "learning_rate": 3.343279870695015e-06,
      "loss": 0.7194,
      "step": 7996
    },
    {
      "epoch": 0.3314269136723445,
      "grad_norm": 0.6145809292793274,
      "learning_rate": 3.3430726511666463e-06,
      "loss": 0.661,
      "step": 7997
    },
    {
      "epoch": 0.3314683575780182,
      "grad_norm": 0.43180668354034424,
      "learning_rate": 3.342865431638278e-06,
      "loss": 0.749,
      "step": 7998
    },
    {
      "epoch": 0.3315098014836918,
      "grad_norm": 0.43506962060928345,
      "learning_rate": 3.3426582121099095e-06,
      "loss": 0.7119,
      "step": 7999
    },
    {
      "epoch": 0.3315512453893655,
      "grad_norm": 0.43424952030181885,
      "learning_rate": 3.3424509925815413e-06,
      "loss": 0.6917,
      "step": 8000
    },
    {
      "epoch": 0.33159268929503916,
      "grad_norm": 0.41163647174835205,
      "learning_rate": 3.3422437730531727e-06,
      "loss": 0.7488,
      "step": 8001
    },
    {
      "epoch": 0.33163413320071283,
      "grad_norm": 0.3894725441932678,
      "learning_rate": 3.3420365535248045e-06,
      "loss": 0.7036,
      "step": 8002
    },
    {
      "epoch": 0.3316755771063865,
      "grad_norm": 0.402765691280365,
      "learning_rate": 3.341829333996436e-06,
      "loss": 0.7283,
      "step": 8003
    },
    {
      "epoch": 0.3317170210120602,
      "grad_norm": 0.46130695939064026,
      "learning_rate": 3.3416221144680677e-06,
      "loss": 0.7168,
      "step": 8004
    },
    {
      "epoch": 0.33175846491773386,
      "grad_norm": 0.39759549498558044,
      "learning_rate": 3.3414148949396995e-06,
      "loss": 0.636,
      "step": 8005
    },
    {
      "epoch": 0.33179990882340754,
      "grad_norm": 0.40847623348236084,
      "learning_rate": 3.341207675411331e-06,
      "loss": 0.7113,
      "step": 8006
    },
    {
      "epoch": 0.33184135272908116,
      "grad_norm": 0.42142438888549805,
      "learning_rate": 3.3410004558829627e-06,
      "loss": 0.7239,
      "step": 8007
    },
    {
      "epoch": 0.33188279663475484,
      "grad_norm": 0.41945144534111023,
      "learning_rate": 3.340793236354594e-06,
      "loss": 0.7556,
      "step": 8008
    },
    {
      "epoch": 0.3319242405404285,
      "grad_norm": 0.41355016827583313,
      "learning_rate": 3.340586016826226e-06,
      "loss": 0.652,
      "step": 8009
    },
    {
      "epoch": 0.3319656844461022,
      "grad_norm": 0.43096327781677246,
      "learning_rate": 3.3403787972978573e-06,
      "loss": 0.7314,
      "step": 8010
    },
    {
      "epoch": 0.33200712835177587,
      "grad_norm": 0.4124286472797394,
      "learning_rate": 3.340171577769489e-06,
      "loss": 0.7734,
      "step": 8011
    },
    {
      "epoch": 0.33204857225744955,
      "grad_norm": 0.4384773373603821,
      "learning_rate": 3.3399643582411205e-06,
      "loss": 0.71,
      "step": 8012
    },
    {
      "epoch": 0.3320900161631232,
      "grad_norm": 0.38740506768226624,
      "learning_rate": 3.3397571387127527e-06,
      "loss": 0.6921,
      "step": 8013
    },
    {
      "epoch": 0.3321314600687969,
      "grad_norm": 0.40994343161582947,
      "learning_rate": 3.3395499191843846e-06,
      "loss": 0.6926,
      "step": 8014
    },
    {
      "epoch": 0.3321729039744706,
      "grad_norm": 0.4233691394329071,
      "learning_rate": 3.339342699656016e-06,
      "loss": 0.6865,
      "step": 8015
    },
    {
      "epoch": 0.3322143478801442,
      "grad_norm": 0.4100450873374939,
      "learning_rate": 3.3391354801276477e-06,
      "loss": 0.6619,
      "step": 8016
    },
    {
      "epoch": 0.3322557917858179,
      "grad_norm": 0.43379610776901245,
      "learning_rate": 3.338928260599279e-06,
      "loss": 0.7024,
      "step": 8017
    },
    {
      "epoch": 0.33229723569149155,
      "grad_norm": 0.45424965023994446,
      "learning_rate": 3.338721041070911e-06,
      "loss": 0.7092,
      "step": 8018
    },
    {
      "epoch": 0.33233867959716523,
      "grad_norm": 0.3991762101650238,
      "learning_rate": 3.3385138215425423e-06,
      "loss": 0.6206,
      "step": 8019
    },
    {
      "epoch": 0.3323801235028389,
      "grad_norm": 0.4172744154930115,
      "learning_rate": 3.338306602014174e-06,
      "loss": 0.7134,
      "step": 8020
    },
    {
      "epoch": 0.3324215674085126,
      "grad_norm": 0.40873926877975464,
      "learning_rate": 3.338099382485806e-06,
      "loss": 0.7091,
      "step": 8021
    },
    {
      "epoch": 0.33246301131418626,
      "grad_norm": 0.40145280957221985,
      "learning_rate": 3.3378921629574373e-06,
      "loss": 0.7051,
      "step": 8022
    },
    {
      "epoch": 0.33250445521985994,
      "grad_norm": 0.40661489963531494,
      "learning_rate": 3.337684943429069e-06,
      "loss": 0.6621,
      "step": 8023
    },
    {
      "epoch": 0.3325458991255336,
      "grad_norm": 0.4530370235443115,
      "learning_rate": 3.3374777239007005e-06,
      "loss": 0.6912,
      "step": 8024
    },
    {
      "epoch": 0.33258734303120724,
      "grad_norm": 0.47130855917930603,
      "learning_rate": 3.3372705043723323e-06,
      "loss": 0.7631,
      "step": 8025
    },
    {
      "epoch": 0.3326287869368809,
      "grad_norm": 0.4066319167613983,
      "learning_rate": 3.3370632848439637e-06,
      "loss": 0.6731,
      "step": 8026
    },
    {
      "epoch": 0.3326702308425546,
      "grad_norm": 0.4065719246864319,
      "learning_rate": 3.3368560653155955e-06,
      "loss": 0.6581,
      "step": 8027
    },
    {
      "epoch": 0.33271167474822827,
      "grad_norm": 0.41268390417099,
      "learning_rate": 3.336648845787227e-06,
      "loss": 0.7188,
      "step": 8028
    },
    {
      "epoch": 0.33275311865390195,
      "grad_norm": 0.39808282256126404,
      "learning_rate": 3.3364416262588587e-06,
      "loss": 0.7048,
      "step": 8029
    },
    {
      "epoch": 0.3327945625595756,
      "grad_norm": 0.38352862000465393,
      "learning_rate": 3.336234406730491e-06,
      "loss": 0.6505,
      "step": 8030
    },
    {
      "epoch": 0.3328360064652493,
      "grad_norm": 0.42006441950798035,
      "learning_rate": 3.3360271872021223e-06,
      "loss": 0.6383,
      "step": 8031
    },
    {
      "epoch": 0.332877450370923,
      "grad_norm": 0.4129132926464081,
      "learning_rate": 3.335819967673754e-06,
      "loss": 0.6748,
      "step": 8032
    },
    {
      "epoch": 0.3329188942765966,
      "grad_norm": 0.4353955388069153,
      "learning_rate": 3.3356127481453855e-06,
      "loss": 0.6965,
      "step": 8033
    },
    {
      "epoch": 0.3329603381822703,
      "grad_norm": 0.6213743686676025,
      "learning_rate": 3.3354055286170173e-06,
      "loss": 0.7064,
      "step": 8034
    },
    {
      "epoch": 0.33300178208794395,
      "grad_norm": 0.5031179785728455,
      "learning_rate": 3.3351983090886487e-06,
      "loss": 0.7524,
      "step": 8035
    },
    {
      "epoch": 0.33304322599361763,
      "grad_norm": 0.410678505897522,
      "learning_rate": 3.3349910895602805e-06,
      "loss": 0.6936,
      "step": 8036
    },
    {
      "epoch": 0.3330846698992913,
      "grad_norm": 0.4394047260284424,
      "learning_rate": 3.334783870031912e-06,
      "loss": 0.7131,
      "step": 8037
    },
    {
      "epoch": 0.333126113804965,
      "grad_norm": 0.40699484944343567,
      "learning_rate": 3.3345766505035437e-06,
      "loss": 0.7354,
      "step": 8038
    },
    {
      "epoch": 0.33316755771063866,
      "grad_norm": 0.40327244997024536,
      "learning_rate": 3.3343694309751755e-06,
      "loss": 0.6936,
      "step": 8039
    },
    {
      "epoch": 0.33320900161631234,
      "grad_norm": 0.4139373302459717,
      "learning_rate": 3.334162211446807e-06,
      "loss": 0.7258,
      "step": 8040
    },
    {
      "epoch": 0.333250445521986,
      "grad_norm": 0.4306830167770386,
      "learning_rate": 3.3339549919184387e-06,
      "loss": 0.7273,
      "step": 8041
    },
    {
      "epoch": 0.33329188942765964,
      "grad_norm": 0.42024728655815125,
      "learning_rate": 3.33374777239007e-06,
      "loss": 0.7168,
      "step": 8042
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.39892956614494324,
      "learning_rate": 3.333540552861702e-06,
      "loss": 0.7073,
      "step": 8043
    },
    {
      "checkpoint_runtime": 208.4844
    },
    {
      "epoch": 0.333374777239007,
      "grad_norm": 0.3866526782512665,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.6536,
      "step": 8044
    },
    {
      "epoch": 0.33341622114468067,
      "grad_norm": 0.4210661053657532,
      "learning_rate": 3.333126113804965e-06,
      "loss": 0.6608,
      "step": 8045
    },
    {
      "epoch": 0.33345766505035435,
      "grad_norm": 0.437215119600296,
      "learning_rate": 3.3329188942765965e-06,
      "loss": 0.6973,
      "step": 8046
    },
    {
      "epoch": 0.333499108956028,
      "grad_norm": 0.3843875229358673,
      "learning_rate": 3.3327116747482288e-06,
      "loss": 0.6906,
      "step": 8047
    },
    {
      "epoch": 0.3335405528617017,
      "grad_norm": 0.40978240966796875,
      "learning_rate": 3.3325044552198606e-06,
      "loss": 0.7297,
      "step": 8048
    },
    {
      "epoch": 0.3335819967673754,
      "grad_norm": 0.41272062063217163,
      "learning_rate": 3.332297235691492e-06,
      "loss": 0.6727,
      "step": 8049
    },
    {
      "epoch": 0.33362344067304905,
      "grad_norm": 0.42441073060035706,
      "learning_rate": 3.3320900161631238e-06,
      "loss": 0.7423,
      "step": 8050
    },
    {
      "epoch": 0.3336648845787227,
      "grad_norm": 0.4119737148284912,
      "learning_rate": 3.331882796634755e-06,
      "loss": 0.7327,
      "step": 8051
    },
    {
      "epoch": 0.33370632848439635,
      "grad_norm": 0.41272541880607605,
      "learning_rate": 3.331675577106387e-06,
      "loss": 0.6826,
      "step": 8052
    },
    {
      "epoch": 0.33374777239007003,
      "grad_norm": 0.38396868109703064,
      "learning_rate": 3.3314683575780183e-06,
      "loss": 0.6704,
      "step": 8053
    },
    {
      "epoch": 0.3337892162957437,
      "grad_norm": 0.4227725863456726,
      "learning_rate": 3.33126113804965e-06,
      "loss": 0.7148,
      "step": 8054
    },
    {
      "epoch": 0.3338306602014174,
      "grad_norm": 0.4149216413497925,
      "learning_rate": 3.3310539185212815e-06,
      "loss": 0.7008,
      "step": 8055
    },
    {
      "epoch": 0.33387210410709106,
      "grad_norm": 0.45245110988616943,
      "learning_rate": 3.3308466989929133e-06,
      "loss": 0.7629,
      "step": 8056
    },
    {
      "epoch": 0.33391354801276474,
      "grad_norm": 0.4650709331035614,
      "learning_rate": 3.330639479464545e-06,
      "loss": 0.7327,
      "step": 8057
    },
    {
      "epoch": 0.3339549919184384,
      "grad_norm": 0.4578043520450592,
      "learning_rate": 3.3304322599361765e-06,
      "loss": 0.7603,
      "step": 8058
    },
    {
      "epoch": 0.33399643582411204,
      "grad_norm": 0.42209184169769287,
      "learning_rate": 3.3302250404078083e-06,
      "loss": 0.7404,
      "step": 8059
    },
    {
      "epoch": 0.3340378797297857,
      "grad_norm": 0.40769118070602417,
      "learning_rate": 3.3300178208794397e-06,
      "loss": 0.6794,
      "step": 8060
    },
    {
      "epoch": 0.3340793236354594,
      "grad_norm": 0.4127490222454071,
      "learning_rate": 3.3298106013510715e-06,
      "loss": 0.665,
      "step": 8061
    },
    {
      "epoch": 0.33412076754113307,
      "grad_norm": 0.4418104290962219,
      "learning_rate": 3.329603381822703e-06,
      "loss": 0.7222,
      "step": 8062
    },
    {
      "epoch": 0.33416221144680675,
      "grad_norm": 0.38394927978515625,
      "learning_rate": 3.3293961622943347e-06,
      "loss": 0.6786,
      "step": 8063
    },
    {
      "epoch": 0.3342036553524804,
      "grad_norm": 0.47822070121765137,
      "learning_rate": 3.329188942765966e-06,
      "loss": 0.6959,
      "step": 8064
    },
    {
      "epoch": 0.3342450992581541,
      "grad_norm": 0.4010542035102844,
      "learning_rate": 3.3289817232375984e-06,
      "loss": 0.7151,
      "step": 8065
    },
    {
      "epoch": 0.3342865431638278,
      "grad_norm": 0.43703991174697876,
      "learning_rate": 3.32877450370923e-06,
      "loss": 0.7244,
      "step": 8066
    },
    {
      "epoch": 0.33432798706950145,
      "grad_norm": 0.406598299741745,
      "learning_rate": 3.3285672841808615e-06,
      "loss": 0.6538,
      "step": 8067
    },
    {
      "epoch": 0.3343694309751751,
      "grad_norm": 0.4346665143966675,
      "learning_rate": 3.3283600646524934e-06,
      "loss": 0.7294,
      "step": 8068
    },
    {
      "epoch": 0.33441087488084875,
      "grad_norm": 0.3880501091480255,
      "learning_rate": 3.3281528451241247e-06,
      "loss": 0.651,
      "step": 8069
    },
    {
      "epoch": 0.33445231878652243,
      "grad_norm": 0.43470486998558044,
      "learning_rate": 3.3279456255957566e-06,
      "loss": 0.7113,
      "step": 8070
    },
    {
      "epoch": 0.3344937626921961,
      "grad_norm": 0.4513741731643677,
      "learning_rate": 3.327738406067388e-06,
      "loss": 0.766,
      "step": 8071
    },
    {
      "epoch": 0.3345352065978698,
      "grad_norm": 0.3939751982688904,
      "learning_rate": 3.3275311865390197e-06,
      "loss": 0.699,
      "step": 8072
    },
    {
      "epoch": 0.33457665050354346,
      "grad_norm": 0.46824631094932556,
      "learning_rate": 3.327323967010651e-06,
      "loss": 0.7915,
      "step": 8073
    },
    {
      "epoch": 0.33461809440921714,
      "grad_norm": 0.4007798433303833,
      "learning_rate": 3.327116747482283e-06,
      "loss": 0.7129,
      "step": 8074
    },
    {
      "epoch": 0.3346595383148908,
      "grad_norm": 0.39239993691444397,
      "learning_rate": 3.3269095279539147e-06,
      "loss": 0.6271,
      "step": 8075
    },
    {
      "epoch": 0.3347009822205645,
      "grad_norm": 0.4036101698875427,
      "learning_rate": 3.326702308425546e-06,
      "loss": 0.6714,
      "step": 8076
    },
    {
      "epoch": 0.3347424261262381,
      "grad_norm": 0.42456308007240295,
      "learning_rate": 3.326495088897178e-06,
      "loss": 0.6897,
      "step": 8077
    },
    {
      "epoch": 0.3347838700319118,
      "grad_norm": 0.38069862127304077,
      "learning_rate": 3.3262878693688093e-06,
      "loss": 0.7139,
      "step": 8078
    },
    {
      "epoch": 0.33482531393758547,
      "grad_norm": 0.43674829602241516,
      "learning_rate": 3.326080649840441e-06,
      "loss": 0.6842,
      "step": 8079
    },
    {
      "epoch": 0.33486675784325914,
      "grad_norm": 0.3938743472099304,
      "learning_rate": 3.3258734303120725e-06,
      "loss": 0.6982,
      "step": 8080
    },
    {
      "epoch": 0.3349082017489328,
      "grad_norm": 0.4254458546638489,
      "learning_rate": 3.3256662107837048e-06,
      "loss": 0.7183,
      "step": 8081
    },
    {
      "epoch": 0.3349496456546065,
      "grad_norm": 0.43785691261291504,
      "learning_rate": 3.3254589912553366e-06,
      "loss": 0.7653,
      "step": 8082
    },
    {
      "epoch": 0.3349910895602802,
      "grad_norm": 0.3793331980705261,
      "learning_rate": 3.325251771726968e-06,
      "loss": 0.702,
      "step": 8083
    },
    {
      "epoch": 0.33503253346595385,
      "grad_norm": 0.463458389043808,
      "learning_rate": 3.3250445521985998e-06,
      "loss": 0.6414,
      "step": 8084
    },
    {
      "epoch": 0.33507397737162753,
      "grad_norm": 0.4247121810913086,
      "learning_rate": 3.324837332670231e-06,
      "loss": 0.6907,
      "step": 8085
    },
    {
      "epoch": 0.33511542127730115,
      "grad_norm": 0.44032782316207886,
      "learning_rate": 3.324630113141863e-06,
      "loss": 0.731,
      "step": 8086
    },
    {
      "epoch": 0.33515686518297483,
      "grad_norm": 0.4050964117050171,
      "learning_rate": 3.3244228936134943e-06,
      "loss": 0.6572,
      "step": 8087
    },
    {
      "epoch": 0.3351983090886485,
      "grad_norm": 0.3922918140888214,
      "learning_rate": 3.324215674085126e-06,
      "loss": 0.6924,
      "step": 8088
    },
    {
      "epoch": 0.3352397529943222,
      "grad_norm": 0.4149254560470581,
      "learning_rate": 3.3240084545567575e-06,
      "loss": 0.7607,
      "step": 8089
    },
    {
      "epoch": 0.33528119689999586,
      "grad_norm": 0.5018079876899719,
      "learning_rate": 3.3238012350283893e-06,
      "loss": 0.7526,
      "step": 8090
    },
    {
      "epoch": 0.33532264080566954,
      "grad_norm": 0.37619051337242126,
      "learning_rate": 3.323594015500021e-06,
      "loss": 0.6567,
      "step": 8091
    },
    {
      "epoch": 0.3353640847113432,
      "grad_norm": 0.41074663400650024,
      "learning_rate": 3.3233867959716525e-06,
      "loss": 0.7114,
      "step": 8092
    },
    {
      "epoch": 0.3354055286170169,
      "grad_norm": 0.3720768690109253,
      "learning_rate": 3.3231795764432843e-06,
      "loss": 0.6676,
      "step": 8093
    },
    {
      "epoch": 0.3354469725226905,
      "grad_norm": 0.47178930044174194,
      "learning_rate": 3.3229723569149157e-06,
      "loss": 0.7031,
      "step": 8094
    },
    {
      "epoch": 0.3354884164283642,
      "grad_norm": 0.4005848169326782,
      "learning_rate": 3.3227651373865475e-06,
      "loss": 0.7202,
      "step": 8095
    },
    {
      "epoch": 0.33552986033403787,
      "grad_norm": 0.46273112297058105,
      "learning_rate": 3.322557917858179e-06,
      "loss": 0.688,
      "step": 8096
    },
    {
      "epoch": 0.33557130423971154,
      "grad_norm": 0.41446107625961304,
      "learning_rate": 3.3223506983298107e-06,
      "loss": 0.7024,
      "step": 8097
    },
    {
      "epoch": 0.3356127481453852,
      "grad_norm": 0.40680378675460815,
      "learning_rate": 3.322143478801442e-06,
      "loss": 0.6638,
      "step": 8098
    },
    {
      "epoch": 0.3356541920510589,
      "grad_norm": 0.43839231133461,
      "learning_rate": 3.3219362592730744e-06,
      "loss": 0.7456,
      "step": 8099
    },
    {
      "epoch": 0.3356956359567326,
      "grad_norm": 0.4191763997077942,
      "learning_rate": 3.321729039744706e-06,
      "loss": 0.6825,
      "step": 8100
    },
    {
      "epoch": 0.33573707986240625,
      "grad_norm": 0.4088877737522125,
      "learning_rate": 3.3215218202163376e-06,
      "loss": 0.6689,
      "step": 8101
    },
    {
      "epoch": 0.33577852376807993,
      "grad_norm": 0.3875562250614166,
      "learning_rate": 3.3213146006879694e-06,
      "loss": 0.6389,
      "step": 8102
    },
    {
      "epoch": 0.33581996767375355,
      "grad_norm": 0.4289272427558899,
      "learning_rate": 3.3211073811596007e-06,
      "loss": 0.7537,
      "step": 8103
    },
    {
      "epoch": 0.3358614115794272,
      "grad_norm": 0.4227634370326996,
      "learning_rate": 3.3209001616312326e-06,
      "loss": 0.6819,
      "step": 8104
    },
    {
      "epoch": 0.3359028554851009,
      "grad_norm": 0.39751020073890686,
      "learning_rate": 3.320692942102864e-06,
      "loss": 0.709,
      "step": 8105
    },
    {
      "epoch": 0.3359442993907746,
      "grad_norm": 0.4498789608478546,
      "learning_rate": 3.3204857225744958e-06,
      "loss": 0.7274,
      "step": 8106
    },
    {
      "epoch": 0.33598574329644826,
      "grad_norm": 0.4376446306705475,
      "learning_rate": 3.320278503046127e-06,
      "loss": 0.6887,
      "step": 8107
    },
    {
      "epoch": 0.33602718720212194,
      "grad_norm": 0.4088791012763977,
      "learning_rate": 3.320071283517759e-06,
      "loss": 0.6881,
      "step": 8108
    },
    {
      "epoch": 0.3360686311077956,
      "grad_norm": 0.38699325919151306,
      "learning_rate": 3.3198640639893908e-06,
      "loss": 0.6748,
      "step": 8109
    },
    {
      "epoch": 0.3361100750134693,
      "grad_norm": 0.4062426686286926,
      "learning_rate": 3.319656844461022e-06,
      "loss": 0.6898,
      "step": 8110
    },
    {
      "epoch": 0.33615151891914297,
      "grad_norm": 0.39614295959472656,
      "learning_rate": 3.319449624932654e-06,
      "loss": 0.6908,
      "step": 8111
    },
    {
      "epoch": 0.3361929628248166,
      "grad_norm": 0.43543753027915955,
      "learning_rate": 3.3192424054042853e-06,
      "loss": 0.676,
      "step": 8112
    },
    {
      "epoch": 0.33623440673049027,
      "grad_norm": 0.42491409182548523,
      "learning_rate": 3.319035185875917e-06,
      "loss": 0.7034,
      "step": 8113
    },
    {
      "epoch": 0.33627585063616394,
      "grad_norm": 0.41535019874572754,
      "learning_rate": 3.3188279663475485e-06,
      "loss": 0.682,
      "step": 8114
    },
    {
      "epoch": 0.3363172945418376,
      "grad_norm": 0.43163424730300903,
      "learning_rate": 3.3186207468191808e-06,
      "loss": 0.7275,
      "step": 8115
    },
    {
      "epoch": 0.3363587384475113,
      "grad_norm": 0.4573400318622589,
      "learning_rate": 3.3184135272908117e-06,
      "loss": 0.7288,
      "step": 8116
    },
    {
      "epoch": 0.336400182353185,
      "grad_norm": 0.42940300703048706,
      "learning_rate": 3.318206307762444e-06,
      "loss": 0.7278,
      "step": 8117
    },
    {
      "epoch": 0.33644162625885865,
      "grad_norm": 0.4591856598854065,
      "learning_rate": 3.3179990882340758e-06,
      "loss": 0.7571,
      "step": 8118
    },
    {
      "epoch": 0.3364830701645323,
      "grad_norm": 0.41584473848342896,
      "learning_rate": 3.317791868705707e-06,
      "loss": 0.6931,
      "step": 8119
    },
    {
      "epoch": 0.33652451407020595,
      "grad_norm": 0.41130807995796204,
      "learning_rate": 3.317584649177339e-06,
      "loss": 0.6749,
      "step": 8120
    },
    {
      "epoch": 0.3365659579758796,
      "grad_norm": 0.3965443968772888,
      "learning_rate": 3.3173774296489703e-06,
      "loss": 0.7008,
      "step": 8121
    },
    {
      "epoch": 0.3366074018815533,
      "grad_norm": 0.3974188268184662,
      "learning_rate": 3.317170210120602e-06,
      "loss": 0.7258,
      "step": 8122
    },
    {
      "epoch": 0.336648845787227,
      "grad_norm": 0.39916959404945374,
      "learning_rate": 3.3169629905922335e-06,
      "loss": 0.6582,
      "step": 8123
    },
    {
      "epoch": 0.33669028969290066,
      "grad_norm": 0.3768806457519531,
      "learning_rate": 3.3167557710638654e-06,
      "loss": 0.6825,
      "step": 8124
    },
    {
      "epoch": 0.33673173359857433,
      "grad_norm": 0.3995022177696228,
      "learning_rate": 3.3165485515354967e-06,
      "loss": 0.6902,
      "step": 8125
    },
    {
      "epoch": 0.336773177504248,
      "grad_norm": 0.3937276005744934,
      "learning_rate": 3.3163413320071285e-06,
      "loss": 0.6937,
      "step": 8126
    },
    {
      "epoch": 0.3368146214099217,
      "grad_norm": 0.4300495684146881,
      "learning_rate": 3.3161341124787604e-06,
      "loss": 0.6649,
      "step": 8127
    },
    {
      "epoch": 0.33685606531559537,
      "grad_norm": 0.3997376263141632,
      "learning_rate": 3.3159268929503917e-06,
      "loss": 0.7019,
      "step": 8128
    },
    {
      "epoch": 0.336897509221269,
      "grad_norm": 0.41976574063301086,
      "learning_rate": 3.3157196734220236e-06,
      "loss": 0.6833,
      "step": 8129
    },
    {
      "epoch": 0.33693895312694266,
      "grad_norm": 0.43948033452033997,
      "learning_rate": 3.315512453893655e-06,
      "loss": 0.708,
      "step": 8130
    },
    {
      "epoch": 0.33698039703261634,
      "grad_norm": 0.41532713174819946,
      "learning_rate": 3.3153052343652867e-06,
      "loss": 0.7451,
      "step": 8131
    },
    {
      "epoch": 0.33702184093829,
      "grad_norm": 0.39756307005882263,
      "learning_rate": 3.315098014836918e-06,
      "loss": 0.6951,
      "step": 8132
    },
    {
      "epoch": 0.3370632848439637,
      "grad_norm": 0.3887263238430023,
      "learning_rate": 3.3148907953085504e-06,
      "loss": 0.6749,
      "step": 8133
    },
    {
      "epoch": 0.3371047287496374,
      "grad_norm": 0.472686231136322,
      "learning_rate": 3.314683575780182e-06,
      "loss": 0.7683,
      "step": 8134
    },
    {
      "epoch": 0.33714617265531105,
      "grad_norm": 0.3917160928249359,
      "learning_rate": 3.3144763562518136e-06,
      "loss": 0.6693,
      "step": 8135
    },
    {
      "epoch": 0.3371876165609847,
      "grad_norm": 0.41691985726356506,
      "learning_rate": 3.3142691367234454e-06,
      "loss": 0.6761,
      "step": 8136
    },
    {
      "epoch": 0.3372290604666584,
      "grad_norm": 0.4337606132030487,
      "learning_rate": 3.3140619171950768e-06,
      "loss": 0.6863,
      "step": 8137
    },
    {
      "epoch": 0.337270504372332,
      "grad_norm": 0.46069976687431335,
      "learning_rate": 3.3138546976667086e-06,
      "loss": 0.6646,
      "step": 8138
    },
    {
      "epoch": 0.3373119482780057,
      "grad_norm": 0.43618836998939514,
      "learning_rate": 3.31364747813834e-06,
      "loss": 0.7468,
      "step": 8139
    },
    {
      "epoch": 0.3373533921836794,
      "grad_norm": 0.4228168725967407,
      "learning_rate": 3.3134402586099718e-06,
      "loss": 0.7051,
      "step": 8140
    },
    {
      "epoch": 0.33739483608935306,
      "grad_norm": 0.3810685873031616,
      "learning_rate": 3.313233039081603e-06,
      "loss": 0.6849,
      "step": 8141
    },
    {
      "epoch": 0.33743627999502673,
      "grad_norm": 0.4250097870826721,
      "learning_rate": 3.313025819553235e-06,
      "loss": 0.6946,
      "step": 8142
    },
    {
      "epoch": 0.3374777239007004,
      "grad_norm": 0.4589986205101013,
      "learning_rate": 3.3128186000248668e-06,
      "loss": 0.7705,
      "step": 8143
    },
    {
      "epoch": 0.3375191678063741,
      "grad_norm": 0.4005773961544037,
      "learning_rate": 3.312611380496498e-06,
      "loss": 0.7175,
      "step": 8144
    },
    {
      "epoch": 0.33756061171204776,
      "grad_norm": 0.3853548765182495,
      "learning_rate": 3.31240416096813e-06,
      "loss": 0.7271,
      "step": 8145
    },
    {
      "epoch": 0.33760205561772144,
      "grad_norm": 0.45562827587127686,
      "learning_rate": 3.3121969414397613e-06,
      "loss": 0.7161,
      "step": 8146
    },
    {
      "epoch": 0.33764349952339506,
      "grad_norm": 0.42512333393096924,
      "learning_rate": 3.311989721911393e-06,
      "loss": 0.7346,
      "step": 8147
    },
    {
      "epoch": 0.33768494342906874,
      "grad_norm": 0.4211041033267975,
      "learning_rate": 3.3117825023830245e-06,
      "loss": 0.694,
      "step": 8148
    },
    {
      "epoch": 0.3377263873347424,
      "grad_norm": 0.4500575661659241,
      "learning_rate": 3.3115752828546568e-06,
      "loss": 0.731,
      "step": 8149
    },
    {
      "epoch": 0.3377678312404161,
      "grad_norm": 0.39083340764045715,
      "learning_rate": 3.3113680633262877e-06,
      "loss": 0.7092,
      "step": 8150
    },
    {
      "epoch": 0.33780927514608977,
      "grad_norm": 0.39939233660697937,
      "learning_rate": 3.31116084379792e-06,
      "loss": 0.7086,
      "step": 8151
    },
    {
      "epoch": 0.33785071905176345,
      "grad_norm": 0.4183955490589142,
      "learning_rate": 3.3109536242695518e-06,
      "loss": 0.6832,
      "step": 8152
    },
    {
      "epoch": 0.3378921629574371,
      "grad_norm": 0.3965153694152832,
      "learning_rate": 3.310746404741183e-06,
      "loss": 0.6741,
      "step": 8153
    },
    {
      "epoch": 0.3379336068631108,
      "grad_norm": 0.41451096534729004,
      "learning_rate": 3.310539185212815e-06,
      "loss": 0.7263,
      "step": 8154
    },
    {
      "epoch": 0.3379750507687844,
      "grad_norm": 0.3950093686580658,
      "learning_rate": 3.3103319656844464e-06,
      "loss": 0.6492,
      "step": 8155
    },
    {
      "epoch": 0.3380164946744581,
      "grad_norm": 0.38639557361602783,
      "learning_rate": 3.310124746156078e-06,
      "loss": 0.6812,
      "step": 8156
    },
    {
      "epoch": 0.3380579385801318,
      "grad_norm": 0.4128146469593048,
      "learning_rate": 3.3099175266277095e-06,
      "loss": 0.6797,
      "step": 8157
    },
    {
      "epoch": 0.33809938248580546,
      "grad_norm": 0.4176913797855377,
      "learning_rate": 3.3097103070993414e-06,
      "loss": 0.7368,
      "step": 8158
    },
    {
      "epoch": 0.33814082639147913,
      "grad_norm": 0.3906860947608948,
      "learning_rate": 3.3095030875709727e-06,
      "loss": 0.6829,
      "step": 8159
    },
    {
      "epoch": 0.3381822702971528,
      "grad_norm": 0.44545209407806396,
      "learning_rate": 3.3092958680426046e-06,
      "loss": 0.7727,
      "step": 8160
    },
    {
      "epoch": 0.3382237142028265,
      "grad_norm": 0.40641355514526367,
      "learning_rate": 3.3090886485142364e-06,
      "loss": 0.687,
      "step": 8161
    },
    {
      "epoch": 0.33826515810850016,
      "grad_norm": 0.44399794936180115,
      "learning_rate": 3.3088814289858677e-06,
      "loss": 0.7136,
      "step": 8162
    },
    {
      "epoch": 0.33830660201417384,
      "grad_norm": 0.4144401550292969,
      "learning_rate": 3.3086742094574996e-06,
      "loss": 0.6897,
      "step": 8163
    },
    {
      "epoch": 0.33834804591984746,
      "grad_norm": 0.425601989030838,
      "learning_rate": 3.308466989929131e-06,
      "loss": 0.6718,
      "step": 8164
    },
    {
      "epoch": 0.33838948982552114,
      "grad_norm": 0.4186052978038788,
      "learning_rate": 3.3082597704007628e-06,
      "loss": 0.7003,
      "step": 8165
    },
    {
      "epoch": 0.3384309337311948,
      "grad_norm": 0.4085485637187958,
      "learning_rate": 3.308052550872394e-06,
      "loss": 0.6952,
      "step": 8166
    },
    {
      "epoch": 0.3384723776368685,
      "grad_norm": 0.4444997012615204,
      "learning_rate": 3.3078453313440264e-06,
      "loss": 0.7903,
      "step": 8167
    },
    {
      "epoch": 0.33851382154254217,
      "grad_norm": 0.3986816704273224,
      "learning_rate": 3.3076381118156573e-06,
      "loss": 0.7078,
      "step": 8168
    },
    {
      "epoch": 0.33855526544821585,
      "grad_norm": 0.4308570325374603,
      "learning_rate": 3.3074308922872896e-06,
      "loss": 0.6919,
      "step": 8169
    },
    {
      "epoch": 0.3385967093538895,
      "grad_norm": 0.4258674383163452,
      "learning_rate": 3.3072236727589214e-06,
      "loss": 0.6978,
      "step": 8170
    },
    {
      "epoch": 0.3386381532595632,
      "grad_norm": 0.40323829650878906,
      "learning_rate": 3.3070164532305528e-06,
      "loss": 0.6868,
      "step": 8171
    },
    {
      "epoch": 0.3386795971652369,
      "grad_norm": 0.42620015144348145,
      "learning_rate": 3.3068092337021846e-06,
      "loss": 0.6639,
      "step": 8172
    },
    {
      "epoch": 0.3387210410709105,
      "grad_norm": 0.42764487862586975,
      "learning_rate": 3.306602014173816e-06,
      "loss": 0.6992,
      "step": 8173
    },
    {
      "epoch": 0.3387624849765842,
      "grad_norm": 0.421510249376297,
      "learning_rate": 3.3063947946454478e-06,
      "loss": 0.7146,
      "step": 8174
    },
    {
      "epoch": 0.33880392888225785,
      "grad_norm": 0.3898261785507202,
      "learning_rate": 3.306187575117079e-06,
      "loss": 0.6804,
      "step": 8175
    },
    {
      "epoch": 0.33884537278793153,
      "grad_norm": 0.39174729585647583,
      "learning_rate": 3.305980355588711e-06,
      "loss": 0.7266,
      "step": 8176
    },
    {
      "epoch": 0.3388868166936052,
      "grad_norm": 0.42321160435676575,
      "learning_rate": 3.3057731360603423e-06,
      "loss": 0.7103,
      "step": 8177
    },
    {
      "epoch": 0.3389282605992789,
      "grad_norm": 0.42934203147888184,
      "learning_rate": 3.305565916531974e-06,
      "loss": 0.6912,
      "step": 8178
    },
    {
      "epoch": 0.33896970450495256,
      "grad_norm": 0.39762037992477417,
      "learning_rate": 3.305358697003606e-06,
      "loss": 0.7605,
      "step": 8179
    },
    {
      "epoch": 0.33901114841062624,
      "grad_norm": 0.4020342230796814,
      "learning_rate": 3.3051514774752373e-06,
      "loss": 0.7334,
      "step": 8180
    },
    {
      "epoch": 0.33905259231629986,
      "grad_norm": 0.4472717344760895,
      "learning_rate": 3.304944257946869e-06,
      "loss": 0.6804,
      "step": 8181
    },
    {
      "epoch": 0.33909403622197354,
      "grad_norm": 0.44548308849334717,
      "learning_rate": 3.3047370384185005e-06,
      "loss": 0.6987,
      "step": 8182
    },
    {
      "epoch": 0.3391354801276472,
      "grad_norm": 0.4155973792076111,
      "learning_rate": 3.3045298188901328e-06,
      "loss": 0.6606,
      "step": 8183
    },
    {
      "epoch": 0.3391769240333209,
      "grad_norm": 0.3973948657512665,
      "learning_rate": 3.3043225993617637e-06,
      "loss": 0.6282,
      "step": 8184
    },
    {
      "epoch": 0.33921836793899457,
      "grad_norm": 0.4547775089740753,
      "learning_rate": 3.304115379833396e-06,
      "loss": 0.7273,
      "step": 8185
    },
    {
      "epoch": 0.33925981184466825,
      "grad_norm": 0.38149547576904297,
      "learning_rate": 3.3039081603050274e-06,
      "loss": 0.7378,
      "step": 8186
    },
    {
      "epoch": 0.3393012557503419,
      "grad_norm": 0.4121350347995758,
      "learning_rate": 3.303700940776659e-06,
      "loss": 0.6863,
      "step": 8187
    },
    {
      "epoch": 0.3393426996560156,
      "grad_norm": 0.39098861813545227,
      "learning_rate": 3.303493721248291e-06,
      "loss": 0.7151,
      "step": 8188
    },
    {
      "epoch": 0.3393841435616893,
      "grad_norm": 0.4331980049610138,
      "learning_rate": 3.3032865017199224e-06,
      "loss": 0.7578,
      "step": 8189
    },
    {
      "epoch": 0.3394255874673629,
      "grad_norm": 0.41539841890335083,
      "learning_rate": 3.303079282191554e-06,
      "loss": 0.7334,
      "step": 8190
    },
    {
      "epoch": 0.3394670313730366,
      "grad_norm": 0.4201783835887909,
      "learning_rate": 3.3028720626631856e-06,
      "loss": 0.6716,
      "step": 8191
    },
    {
      "epoch": 0.33950847527871025,
      "grad_norm": 0.43180710077285767,
      "learning_rate": 3.3026648431348174e-06,
      "loss": 0.7063,
      "step": 8192
    },
    {
      "epoch": 0.33954991918438393,
      "grad_norm": 0.43275007605552673,
      "learning_rate": 3.3024576236064487e-06,
      "loss": 0.7788,
      "step": 8193
    },
    {
      "epoch": 0.3395913630900576,
      "grad_norm": 0.4269428551197052,
      "learning_rate": 3.3022504040780806e-06,
      "loss": 0.6964,
      "step": 8194
    },
    {
      "epoch": 0.3396328069957313,
      "grad_norm": 0.4455830752849579,
      "learning_rate": 3.3020431845497124e-06,
      "loss": 0.7417,
      "step": 8195
    },
    {
      "epoch": 0.33967425090140496,
      "grad_norm": 0.41200563311576843,
      "learning_rate": 3.3018359650213438e-06,
      "loss": 0.6835,
      "step": 8196
    },
    {
      "epoch": 0.33971569480707864,
      "grad_norm": 0.4117705523967743,
      "learning_rate": 3.3016287454929756e-06,
      "loss": 0.688,
      "step": 8197
    },
    {
      "epoch": 0.3397571387127523,
      "grad_norm": 0.4359980821609497,
      "learning_rate": 3.301421525964607e-06,
      "loss": 0.7114,
      "step": 8198
    },
    {
      "epoch": 0.33979858261842594,
      "grad_norm": 0.4097535014152527,
      "learning_rate": 3.3012143064362388e-06,
      "loss": 0.7246,
      "step": 8199
    },
    {
      "epoch": 0.3398400265240996,
      "grad_norm": 0.3824756443500519,
      "learning_rate": 3.30100708690787e-06,
      "loss": 0.6653,
      "step": 8200
    },
    {
      "epoch": 0.3398814704297733,
      "grad_norm": 0.40654703974723816,
      "learning_rate": 3.3007998673795024e-06,
      "loss": 0.6749,
      "step": 8201
    },
    {
      "epoch": 0.33992291433544697,
      "grad_norm": 0.380788654088974,
      "learning_rate": 3.3005926478511333e-06,
      "loss": 0.717,
      "step": 8202
    },
    {
      "epoch": 0.33996435824112065,
      "grad_norm": 0.4020877182483673,
      "learning_rate": 3.3003854283227656e-06,
      "loss": 0.7227,
      "step": 8203
    },
    {
      "epoch": 0.3400058021467943,
      "grad_norm": 0.43380260467529297,
      "learning_rate": 3.3001782087943974e-06,
      "loss": 0.7212,
      "step": 8204
    },
    {
      "epoch": 0.340047246052468,
      "grad_norm": 0.40868085622787476,
      "learning_rate": 3.2999709892660288e-06,
      "loss": 0.688,
      "step": 8205
    },
    {
      "epoch": 0.3400886899581417,
      "grad_norm": 0.422498881816864,
      "learning_rate": 3.2997637697376606e-06,
      "loss": 0.7029,
      "step": 8206
    },
    {
      "epoch": 0.3401301338638153,
      "grad_norm": 0.4094950258731842,
      "learning_rate": 3.299556550209292e-06,
      "loss": 0.6577,
      "step": 8207
    },
    {
      "epoch": 0.340171577769489,
      "grad_norm": 0.42016324400901794,
      "learning_rate": 3.2993493306809238e-06,
      "loss": 0.6655,
      "step": 8208
    },
    {
      "epoch": 0.34021302167516265,
      "grad_norm": 0.45435062050819397,
      "learning_rate": 3.299142111152555e-06,
      "loss": 0.6927,
      "step": 8209
    },
    {
      "epoch": 0.34025446558083633,
      "grad_norm": 0.40360575914382935,
      "learning_rate": 3.298934891624187e-06,
      "loss": 0.7007,
      "step": 8210
    },
    {
      "epoch": 0.34029590948651,
      "grad_norm": 0.37984806299209595,
      "learning_rate": 3.2987276720958184e-06,
      "loss": 0.7148,
      "step": 8211
    },
    {
      "epoch": 0.3403373533921837,
      "grad_norm": 0.4794977009296417,
      "learning_rate": 3.29852045256745e-06,
      "loss": 0.731,
      "step": 8212
    },
    {
      "epoch": 0.34037879729785736,
      "grad_norm": 0.4282992482185364,
      "learning_rate": 3.298313233039082e-06,
      "loss": 0.6874,
      "step": 8213
    },
    {
      "epoch": 0.34042024120353104,
      "grad_norm": 0.4196062982082367,
      "learning_rate": 3.2981060135107134e-06,
      "loss": 0.7432,
      "step": 8214
    },
    {
      "epoch": 0.3404616851092047,
      "grad_norm": 0.3992667496204376,
      "learning_rate": 3.297898793982345e-06,
      "loss": 0.6899,
      "step": 8215
    },
    {
      "epoch": 0.34050312901487834,
      "grad_norm": 0.40477341413497925,
      "learning_rate": 3.2976915744539765e-06,
      "loss": 0.6877,
      "step": 8216
    },
    {
      "epoch": 0.340544572920552,
      "grad_norm": 0.41400623321533203,
      "learning_rate": 3.2974843549256088e-06,
      "loss": 0.6278,
      "step": 8217
    },
    {
      "epoch": 0.3405860168262257,
      "grad_norm": 0.4152536988258362,
      "learning_rate": 3.2972771353972397e-06,
      "loss": 0.7126,
      "step": 8218
    },
    {
      "epoch": 0.34062746073189937,
      "grad_norm": 0.40164583921432495,
      "learning_rate": 3.297069915868872e-06,
      "loss": 0.6846,
      "step": 8219
    },
    {
      "epoch": 0.34066890463757304,
      "grad_norm": 0.39710527658462524,
      "learning_rate": 3.2968626963405034e-06,
      "loss": 0.6595,
      "step": 8220
    },
    {
      "epoch": 0.3407103485432467,
      "grad_norm": 0.43746107816696167,
      "learning_rate": 3.296655476812135e-06,
      "loss": 0.7063,
      "step": 8221
    },
    {
      "epoch": 0.3407517924489204,
      "grad_norm": 0.4980984330177307,
      "learning_rate": 3.296448257283767e-06,
      "loss": 0.7029,
      "step": 8222
    },
    {
      "epoch": 0.3407932363545941,
      "grad_norm": 0.4424000084400177,
      "learning_rate": 3.2962410377553984e-06,
      "loss": 0.7306,
      "step": 8223
    },
    {
      "epoch": 0.34083468026026775,
      "grad_norm": 0.4248712658882141,
      "learning_rate": 3.29603381822703e-06,
      "loss": 0.7373,
      "step": 8224
    },
    {
      "epoch": 0.3408761241659414,
      "grad_norm": 0.42607027292251587,
      "learning_rate": 3.2958265986986616e-06,
      "loss": 0.7131,
      "step": 8225
    },
    {
      "epoch": 0.34091756807161505,
      "grad_norm": 0.4254146218299866,
      "learning_rate": 3.2956193791702934e-06,
      "loss": 0.6804,
      "step": 8226
    },
    {
      "epoch": 0.34095901197728873,
      "grad_norm": 0.4113422930240631,
      "learning_rate": 3.2954121596419248e-06,
      "loss": 0.6729,
      "step": 8227
    },
    {
      "epoch": 0.3410004558829624,
      "grad_norm": 0.39332330226898193,
      "learning_rate": 3.2952049401135566e-06,
      "loss": 0.6873,
      "step": 8228
    },
    {
      "epoch": 0.3410418997886361,
      "grad_norm": 0.39834925532341003,
      "learning_rate": 3.294997720585188e-06,
      "loss": 0.7175,
      "step": 8229
    },
    {
      "epoch": 0.34108334369430976,
      "grad_norm": 0.40276190638542175,
      "learning_rate": 3.2947905010568198e-06,
      "loss": 0.7305,
      "step": 8230
    },
    {
      "epoch": 0.34112478759998344,
      "grad_norm": 0.4111160635948181,
      "learning_rate": 3.2945832815284516e-06,
      "loss": 0.7031,
      "step": 8231
    },
    {
      "epoch": 0.3411662315056571,
      "grad_norm": 0.4187011420726776,
      "learning_rate": 3.294376062000083e-06,
      "loss": 0.7031,
      "step": 8232
    },
    {
      "epoch": 0.3412076754113308,
      "grad_norm": 0.4183242619037628,
      "learning_rate": 3.294168842471715e-06,
      "loss": 0.7898,
      "step": 8233
    },
    {
      "epoch": 0.3412491193170044,
      "grad_norm": 0.41397011280059814,
      "learning_rate": 3.293961622943346e-06,
      "loss": 0.7209,
      "step": 8234
    },
    {
      "epoch": 0.3412905632226781,
      "grad_norm": 0.39874985814094543,
      "learning_rate": 3.2937544034149784e-06,
      "loss": 0.6709,
      "step": 8235
    },
    {
      "epoch": 0.34133200712835177,
      "grad_norm": 0.4511902630329132,
      "learning_rate": 3.2935471838866093e-06,
      "loss": 0.7522,
      "step": 8236
    },
    {
      "epoch": 0.34137345103402544,
      "grad_norm": 0.4246155321598053,
      "learning_rate": 3.2933399643582416e-06,
      "loss": 0.7366,
      "step": 8237
    },
    {
      "epoch": 0.3414148949396991,
      "grad_norm": 0.40551620721817017,
      "learning_rate": 3.293132744829873e-06,
      "loss": 0.62,
      "step": 8238
    },
    {
      "epoch": 0.3414563388453728,
      "grad_norm": 0.397664338350296,
      "learning_rate": 3.2929255253015048e-06,
      "loss": 0.6785,
      "step": 8239
    },
    {
      "epoch": 0.3414977827510465,
      "grad_norm": 0.42914676666259766,
      "learning_rate": 3.2927183057731366e-06,
      "loss": 0.6448,
      "step": 8240
    },
    {
      "epoch": 0.34153922665672015,
      "grad_norm": 0.41293951869010925,
      "learning_rate": 3.292511086244768e-06,
      "loss": 0.7228,
      "step": 8241
    },
    {
      "epoch": 0.3415806705623938,
      "grad_norm": 0.393697589635849,
      "learning_rate": 3.2923038667163998e-06,
      "loss": 0.6726,
      "step": 8242
    },
    {
      "epoch": 0.34162211446806745,
      "grad_norm": 0.4054800868034363,
      "learning_rate": 3.292096647188031e-06,
      "loss": 0.6938,
      "step": 8243
    },
    {
      "epoch": 0.3416635583737411,
      "grad_norm": 0.4179021716117859,
      "learning_rate": 3.291889427659663e-06,
      "loss": 0.707,
      "step": 8244
    },
    {
      "epoch": 0.3417050022794148,
      "grad_norm": 0.39516451954841614,
      "learning_rate": 3.2916822081312944e-06,
      "loss": 0.7258,
      "step": 8245
    },
    {
      "epoch": 0.3417464461850885,
      "grad_norm": 0.4341980814933777,
      "learning_rate": 3.291474988602926e-06,
      "loss": 0.7153,
      "step": 8246
    },
    {
      "epoch": 0.34178789009076216,
      "grad_norm": 0.42921265959739685,
      "learning_rate": 3.2912677690745576e-06,
      "loss": 0.7168,
      "step": 8247
    },
    {
      "epoch": 0.34182933399643584,
      "grad_norm": 0.4249199628829956,
      "learning_rate": 3.2910605495461894e-06,
      "loss": 0.6873,
      "step": 8248
    },
    {
      "epoch": 0.3418707779021095,
      "grad_norm": 0.407155841588974,
      "learning_rate": 3.290853330017821e-06,
      "loss": 0.7266,
      "step": 8249
    },
    {
      "epoch": 0.3419122218077832,
      "grad_norm": 0.4311218559741974,
      "learning_rate": 3.2906461104894526e-06,
      "loss": 0.7786,
      "step": 8250
    },
    {
      "epoch": 0.3419536657134568,
      "grad_norm": 0.4178151488304138,
      "learning_rate": 3.2904388909610848e-06,
      "loss": 0.7017,
      "step": 8251
    },
    {
      "epoch": 0.3419951096191305,
      "grad_norm": 0.4007262587547302,
      "learning_rate": 3.2902316714327157e-06,
      "loss": 0.6318,
      "step": 8252
    },
    {
      "epoch": 0.34203655352480417,
      "grad_norm": 0.43141937255859375,
      "learning_rate": 3.290024451904348e-06,
      "loss": 0.709,
      "step": 8253
    },
    {
      "epoch": 0.34207799743047784,
      "grad_norm": 0.3759288787841797,
      "learning_rate": 3.2898172323759794e-06,
      "loss": 0.6646,
      "step": 8254
    },
    {
      "epoch": 0.3421194413361515,
      "grad_norm": 0.4037187993526459,
      "learning_rate": 3.289610012847611e-06,
      "loss": 0.7047,
      "step": 8255
    },
    {
      "epoch": 0.3421608852418252,
      "grad_norm": 0.4008956849575043,
      "learning_rate": 3.289402793319243e-06,
      "loss": 0.6821,
      "step": 8256
    },
    {
      "epoch": 0.3422023291474989,
      "grad_norm": 0.4044203758239746,
      "learning_rate": 3.2891955737908744e-06,
      "loss": 0.699,
      "step": 8257
    },
    {
      "epoch": 0.34224377305317255,
      "grad_norm": 0.45174455642700195,
      "learning_rate": 3.288988354262506e-06,
      "loss": 0.8027,
      "step": 8258
    },
    {
      "epoch": 0.34228521695884623,
      "grad_norm": 0.41861510276794434,
      "learning_rate": 3.2887811347341376e-06,
      "loss": 0.7006,
      "step": 8259
    },
    {
      "epoch": 0.34232666086451985,
      "grad_norm": 0.4350740313529968,
      "learning_rate": 3.2885739152057694e-06,
      "loss": 0.7085,
      "step": 8260
    },
    {
      "epoch": 0.3423681047701935,
      "grad_norm": 0.47519373893737793,
      "learning_rate": 3.2883666956774008e-06,
      "loss": 0.7324,
      "step": 8261
    },
    {
      "epoch": 0.3424095486758672,
      "grad_norm": 0.38128402829170227,
      "learning_rate": 3.2881594761490326e-06,
      "loss": 0.6436,
      "step": 8262
    },
    {
      "epoch": 0.3424509925815409,
      "grad_norm": 0.42504337430000305,
      "learning_rate": 3.287952256620664e-06,
      "loss": 0.6771,
      "step": 8263
    },
    {
      "epoch": 0.34249243648721456,
      "grad_norm": 0.4140465557575226,
      "learning_rate": 3.2877450370922958e-06,
      "loss": 0.6879,
      "step": 8264
    },
    {
      "epoch": 0.34253388039288823,
      "grad_norm": 0.39626458287239075,
      "learning_rate": 3.2875378175639276e-06,
      "loss": 0.6776,
      "step": 8265
    },
    {
      "epoch": 0.3425753242985619,
      "grad_norm": 0.4329952895641327,
      "learning_rate": 3.287330598035559e-06,
      "loss": 0.6667,
      "step": 8266
    },
    {
      "epoch": 0.3426167682042356,
      "grad_norm": 0.3829418122768402,
      "learning_rate": 3.287123378507191e-06,
      "loss": 0.6383,
      "step": 8267
    },
    {
      "epoch": 0.3426582121099092,
      "grad_norm": 0.3914972245693207,
      "learning_rate": 3.286916158978822e-06,
      "loss": 0.6609,
      "step": 8268
    },
    {
      "epoch": 0.3426996560155829,
      "grad_norm": 0.4153934419155121,
      "learning_rate": 3.2867089394504544e-06,
      "loss": 0.6263,
      "step": 8269
    },
    {
      "epoch": 0.34274109992125656,
      "grad_norm": 0.4246017634868622,
      "learning_rate": 3.2865017199220854e-06,
      "loss": 0.6985,
      "step": 8270
    },
    {
      "epoch": 0.34278254382693024,
      "grad_norm": 0.453042209148407,
      "learning_rate": 3.2862945003937176e-06,
      "loss": 0.7169,
      "step": 8271
    },
    {
      "epoch": 0.3428239877326039,
      "grad_norm": 0.4430462121963501,
      "learning_rate": 3.286087280865349e-06,
      "loss": 0.7234,
      "step": 8272
    },
    {
      "epoch": 0.3428654316382776,
      "grad_norm": 0.3992575705051422,
      "learning_rate": 3.2858800613369808e-06,
      "loss": 0.6934,
      "step": 8273
    },
    {
      "epoch": 0.3429068755439513,
      "grad_norm": 0.4461498558521271,
      "learning_rate": 3.2856728418086126e-06,
      "loss": 0.6786,
      "step": 8274
    },
    {
      "epoch": 0.34294831944962495,
      "grad_norm": 0.4139283001422882,
      "learning_rate": 3.285465622280244e-06,
      "loss": 0.696,
      "step": 8275
    },
    {
      "epoch": 0.3429897633552986,
      "grad_norm": 0.41413089632987976,
      "learning_rate": 3.2852584027518758e-06,
      "loss": 0.7145,
      "step": 8276
    },
    {
      "epoch": 0.34303120726097225,
      "grad_norm": 0.4272351562976837,
      "learning_rate": 3.285051183223507e-06,
      "loss": 0.6809,
      "step": 8277
    },
    {
      "epoch": 0.3430726511666459,
      "grad_norm": 0.3883785903453827,
      "learning_rate": 3.284843963695139e-06,
      "loss": 0.6477,
      "step": 8278
    },
    {
      "epoch": 0.3431140950723196,
      "grad_norm": 0.4192541241645813,
      "learning_rate": 3.2846367441667704e-06,
      "loss": 0.7327,
      "step": 8279
    },
    {
      "epoch": 0.3431555389779933,
      "grad_norm": 0.3897496461868286,
      "learning_rate": 3.284429524638402e-06,
      "loss": 0.7375,
      "step": 8280
    },
    {
      "epoch": 0.34319698288366696,
      "grad_norm": 0.42063048481941223,
      "learning_rate": 3.2842223051100336e-06,
      "loss": 0.6785,
      "step": 8281
    },
    {
      "epoch": 0.34323842678934063,
      "grad_norm": 0.4189550280570984,
      "learning_rate": 3.2840150855816654e-06,
      "loss": 0.7202,
      "step": 8282
    },
    {
      "epoch": 0.3432798706950143,
      "grad_norm": 0.4480151832103729,
      "learning_rate": 3.283807866053297e-06,
      "loss": 0.7236,
      "step": 8283
    },
    {
      "epoch": 0.343321314600688,
      "grad_norm": 0.4498073160648346,
      "learning_rate": 3.2836006465249286e-06,
      "loss": 0.7314,
      "step": 8284
    },
    {
      "epoch": 0.34336275850636166,
      "grad_norm": 0.4089435935020447,
      "learning_rate": 3.283393426996561e-06,
      "loss": 0.7251,
      "step": 8285
    },
    {
      "epoch": 0.3434042024120353,
      "grad_norm": 0.37917083501815796,
      "learning_rate": 3.2831862074681918e-06,
      "loss": 0.7032,
      "step": 8286
    },
    {
      "epoch": 0.34344564631770896,
      "grad_norm": 0.43270036578178406,
      "learning_rate": 3.282978987939824e-06,
      "loss": 0.7083,
      "step": 8287
    },
    {
      "epoch": 0.34348709022338264,
      "grad_norm": 0.43415552377700806,
      "learning_rate": 3.2827717684114554e-06,
      "loss": 0.6958,
      "step": 8288
    },
    {
      "epoch": 0.3435285341290563,
      "grad_norm": 0.4249191880226135,
      "learning_rate": 3.282564548883087e-06,
      "loss": 0.6906,
      "step": 8289
    },
    {
      "epoch": 0.34356997803473,
      "grad_norm": 0.44363370537757874,
      "learning_rate": 3.2823573293547186e-06,
      "loss": 0.7551,
      "step": 8290
    },
    {
      "epoch": 0.34361142194040367,
      "grad_norm": 0.4261104166507721,
      "learning_rate": 3.2821501098263504e-06,
      "loss": 0.7084,
      "step": 8291
    },
    {
      "epoch": 0.34365286584607735,
      "grad_norm": 0.3948059678077698,
      "learning_rate": 3.281942890297982e-06,
      "loss": 0.6882,
      "step": 8292
    },
    {
      "epoch": 0.343694309751751,
      "grad_norm": 0.4130396246910095,
      "learning_rate": 3.2817356707696136e-06,
      "loss": 0.7051,
      "step": 8293
    },
    {
      "epoch": 0.3437357536574247,
      "grad_norm": 0.39997172355651855,
      "learning_rate": 3.2815284512412454e-06,
      "loss": 0.6543,
      "step": 8294
    },
    {
      "epoch": 0.3437771975630983,
      "grad_norm": 0.420317679643631,
      "learning_rate": 3.2813212317128768e-06,
      "loss": 0.6746,
      "step": 8295
    },
    {
      "epoch": 0.343818641468772,
      "grad_norm": 0.3757094442844391,
      "learning_rate": 3.2811140121845086e-06,
      "loss": 0.6511,
      "step": 8296
    },
    {
      "epoch": 0.3438600853744457,
      "grad_norm": 0.4113697409629822,
      "learning_rate": 3.28090679265614e-06,
      "loss": 0.6956,
      "step": 8297
    },
    {
      "epoch": 0.34390152928011936,
      "grad_norm": 0.4157564043998718,
      "learning_rate": 3.2806995731277718e-06,
      "loss": 0.752,
      "step": 8298
    },
    {
      "epoch": 0.34394297318579303,
      "grad_norm": 0.4178808331489563,
      "learning_rate": 3.280492353599403e-06,
      "loss": 0.728,
      "step": 8299
    },
    {
      "epoch": 0.3439844170914667,
      "grad_norm": 0.46545225381851196,
      "learning_rate": 3.280285134071035e-06,
      "loss": 0.741,
      "step": 8300
    },
    {
      "epoch": 0.3440258609971404,
      "grad_norm": 0.42063915729522705,
      "learning_rate": 3.280077914542667e-06,
      "loss": 0.6644,
      "step": 8301
    },
    {
      "epoch": 0.34406730490281406,
      "grad_norm": 0.41480180621147156,
      "learning_rate": 3.279870695014298e-06,
      "loss": 0.717,
      "step": 8302
    },
    {
      "epoch": 0.3441087488084877,
      "grad_norm": 0.42868340015411377,
      "learning_rate": 3.2796634754859304e-06,
      "loss": 0.6947,
      "step": 8303
    },
    {
      "epoch": 0.34415019271416136,
      "grad_norm": 0.4138825237751007,
      "learning_rate": 3.2794562559575614e-06,
      "loss": 0.7395,
      "step": 8304
    },
    {
      "epoch": 0.34419163661983504,
      "grad_norm": 0.431780070066452,
      "learning_rate": 3.2792490364291936e-06,
      "loss": 0.6931,
      "step": 8305
    },
    {
      "epoch": 0.3442330805255087,
      "grad_norm": 0.4359157085418701,
      "learning_rate": 3.279041816900825e-06,
      "loss": 0.6797,
      "step": 8306
    },
    {
      "epoch": 0.3442745244311824,
      "grad_norm": 0.4280223548412323,
      "learning_rate": 3.2788345973724568e-06,
      "loss": 0.6849,
      "step": 8307
    },
    {
      "epoch": 0.34431596833685607,
      "grad_norm": 0.4720728397369385,
      "learning_rate": 3.278627377844088e-06,
      "loss": 0.7041,
      "step": 8308
    },
    {
      "epoch": 0.34435741224252975,
      "grad_norm": 0.405681312084198,
      "learning_rate": 3.27842015831572e-06,
      "loss": 0.6765,
      "step": 8309
    },
    {
      "epoch": 0.3443988561482034,
      "grad_norm": 0.40840592980384827,
      "learning_rate": 3.2782129387873518e-06,
      "loss": 0.6729,
      "step": 8310
    },
    {
      "epoch": 0.3444403000538771,
      "grad_norm": 0.42388543486595154,
      "learning_rate": 3.278005719258983e-06,
      "loss": 0.6259,
      "step": 8311
    },
    {
      "epoch": 0.3444817439595507,
      "grad_norm": 0.39172250032424927,
      "learning_rate": 3.277798499730615e-06,
      "loss": 0.6824,
      "step": 8312
    },
    {
      "epoch": 0.3445231878652244,
      "grad_norm": 0.4206143617630005,
      "learning_rate": 3.2775912802022464e-06,
      "loss": 0.6844,
      "step": 8313
    },
    {
      "epoch": 0.3445646317708981,
      "grad_norm": 0.44540008902549744,
      "learning_rate": 3.277384060673878e-06,
      "loss": 0.6616,
      "step": 8314
    },
    {
      "epoch": 0.34460607567657175,
      "grad_norm": 0.3946239948272705,
      "learning_rate": 3.2771768411455096e-06,
      "loss": 0.6799,
      "step": 8315
    },
    {
      "epoch": 0.34464751958224543,
      "grad_norm": 0.390486478805542,
      "learning_rate": 3.2769696216171414e-06,
      "loss": 0.7371,
      "step": 8316
    },
    {
      "epoch": 0.3446889634879191,
      "grad_norm": 0.41588276624679565,
      "learning_rate": 3.276762402088773e-06,
      "loss": 0.707,
      "step": 8317
    },
    {
      "epoch": 0.3447304073935928,
      "grad_norm": 0.3903566002845764,
      "learning_rate": 3.2765551825604046e-06,
      "loss": 0.6761,
      "step": 8318
    },
    {
      "epoch": 0.34477185129926646,
      "grad_norm": 0.44057101011276245,
      "learning_rate": 3.276347963032037e-06,
      "loss": 0.7183,
      "step": 8319
    },
    {
      "epoch": 0.34481329520494014,
      "grad_norm": 0.4107791781425476,
      "learning_rate": 3.2761407435036678e-06,
      "loss": 0.6859,
      "step": 8320
    },
    {
      "epoch": 0.34485473911061376,
      "grad_norm": 0.41709867119789124,
      "learning_rate": 3.2759335239753e-06,
      "loss": 0.7407,
      "step": 8321
    },
    {
      "epoch": 0.34489618301628744,
      "grad_norm": 0.4087441861629486,
      "learning_rate": 3.2757263044469314e-06,
      "loss": 0.6833,
      "step": 8322
    },
    {
      "epoch": 0.3449376269219611,
      "grad_norm": 0.40378430485725403,
      "learning_rate": 3.275519084918563e-06,
      "loss": 0.7603,
      "step": 8323
    },
    {
      "epoch": 0.3449790708276348,
      "grad_norm": 0.4217969477176666,
      "learning_rate": 3.2753118653901946e-06,
      "loss": 0.7312,
      "step": 8324
    },
    {
      "epoch": 0.34502051473330847,
      "grad_norm": 0.42874574661254883,
      "learning_rate": 3.2751046458618264e-06,
      "loss": 0.7067,
      "step": 8325
    },
    {
      "epoch": 0.34506195863898215,
      "grad_norm": 0.44342219829559326,
      "learning_rate": 3.274897426333458e-06,
      "loss": 0.752,
      "step": 8326
    },
    {
      "epoch": 0.3451034025446558,
      "grad_norm": 0.3970843553543091,
      "learning_rate": 3.2746902068050896e-06,
      "loss": 0.6833,
      "step": 8327
    },
    {
      "epoch": 0.3451448464503295,
      "grad_norm": 0.42490053176879883,
      "learning_rate": 3.2744829872767214e-06,
      "loss": 0.7639,
      "step": 8328
    },
    {
      "epoch": 0.3451862903560031,
      "grad_norm": 0.42897534370422363,
      "learning_rate": 3.2742757677483528e-06,
      "loss": 0.6715,
      "step": 8329
    },
    {
      "epoch": 0.3452277342616768,
      "grad_norm": 0.357544869184494,
      "learning_rate": 3.2740685482199846e-06,
      "loss": 0.6862,
      "step": 8330
    },
    {
      "epoch": 0.3452691781673505,
      "grad_norm": 0.4055792987346649,
      "learning_rate": 3.273861328691616e-06,
      "loss": 0.696,
      "step": 8331
    },
    {
      "epoch": 0.34531062207302415,
      "grad_norm": 0.421024352312088,
      "learning_rate": 3.2736541091632478e-06,
      "loss": 0.708,
      "step": 8332
    },
    {
      "epoch": 0.34535206597869783,
      "grad_norm": 0.40851712226867676,
      "learning_rate": 3.273446889634879e-06,
      "loss": 0.699,
      "step": 8333
    },
    {
      "epoch": 0.3453935098843715,
      "grad_norm": 0.4174291789531708,
      "learning_rate": 3.273239670106511e-06,
      "loss": 0.6897,
      "step": 8334
    },
    {
      "epoch": 0.3454349537900452,
      "grad_norm": 0.3878898322582245,
      "learning_rate": 3.273032450578143e-06,
      "loss": 0.7041,
      "step": 8335
    },
    {
      "epoch": 0.34547639769571886,
      "grad_norm": 0.43906232714653015,
      "learning_rate": 3.272825231049774e-06,
      "loss": 0.6943,
      "step": 8336
    },
    {
      "epoch": 0.34551784160139254,
      "grad_norm": 0.3986448049545288,
      "learning_rate": 3.2726180115214064e-06,
      "loss": 0.6846,
      "step": 8337
    },
    {
      "epoch": 0.34555928550706616,
      "grad_norm": 0.41709455847740173,
      "learning_rate": 3.2724107919930374e-06,
      "loss": 0.6897,
      "step": 8338
    },
    {
      "epoch": 0.34560072941273984,
      "grad_norm": 0.4260007441043854,
      "learning_rate": 3.2722035724646696e-06,
      "loss": 0.7209,
      "step": 8339
    },
    {
      "epoch": 0.3456421733184135,
      "grad_norm": 0.41488856077194214,
      "learning_rate": 3.271996352936301e-06,
      "loss": 0.675,
      "step": 8340
    },
    {
      "epoch": 0.3456836172240872,
      "grad_norm": 0.43570229411125183,
      "learning_rate": 3.271789133407933e-06,
      "loss": 0.7344,
      "step": 8341
    },
    {
      "epoch": 0.34572506112976087,
      "grad_norm": 0.420254647731781,
      "learning_rate": 3.271581913879564e-06,
      "loss": 0.6814,
      "step": 8342
    },
    {
      "epoch": 0.34576650503543455,
      "grad_norm": 0.40473467111587524,
      "learning_rate": 3.271374694351196e-06,
      "loss": 0.6948,
      "step": 8343
    },
    {
      "epoch": 0.3458079489411082,
      "grad_norm": 0.3946726620197296,
      "learning_rate": 3.271167474822828e-06,
      "loss": 0.6949,
      "step": 8344
    },
    {
      "epoch": 0.3458493928467819,
      "grad_norm": 0.44714558124542236,
      "learning_rate": 3.270960255294459e-06,
      "loss": 0.7278,
      "step": 8345
    },
    {
      "epoch": 0.3458908367524556,
      "grad_norm": 0.41146141290664673,
      "learning_rate": 3.270753035766091e-06,
      "loss": 0.7119,
      "step": 8346
    },
    {
      "epoch": 0.3459322806581292,
      "grad_norm": 0.4135577082633972,
      "learning_rate": 3.2705458162377224e-06,
      "loss": 0.7122,
      "step": 8347
    },
    {
      "epoch": 0.3459737245638029,
      "grad_norm": 0.4159775376319885,
      "learning_rate": 3.270338596709354e-06,
      "loss": 0.719,
      "step": 8348
    },
    {
      "epoch": 0.34601516846947655,
      "grad_norm": 0.43465274572372437,
      "learning_rate": 3.2701313771809856e-06,
      "loss": 0.7042,
      "step": 8349
    },
    {
      "epoch": 0.34605661237515023,
      "grad_norm": 0.4419252872467041,
      "learning_rate": 3.2699241576526174e-06,
      "loss": 0.7295,
      "step": 8350
    },
    {
      "epoch": 0.3460980562808239,
      "grad_norm": 0.40720805525779724,
      "learning_rate": 3.2697169381242488e-06,
      "loss": 0.7052,
      "step": 8351
    },
    {
      "epoch": 0.3461395001864976,
      "grad_norm": 0.4039510488510132,
      "learning_rate": 3.2695097185958806e-06,
      "loss": 0.7197,
      "step": 8352
    },
    {
      "epoch": 0.34618094409217126,
      "grad_norm": 0.3667064607143402,
      "learning_rate": 3.269302499067513e-06,
      "loss": 0.6331,
      "step": 8353
    },
    {
      "epoch": 0.34622238799784494,
      "grad_norm": 0.4098520576953888,
      "learning_rate": 3.2690952795391438e-06,
      "loss": 0.7036,
      "step": 8354
    },
    {
      "epoch": 0.3462638319035186,
      "grad_norm": 0.39566749334335327,
      "learning_rate": 3.268888060010776e-06,
      "loss": 0.7028,
      "step": 8355
    },
    {
      "epoch": 0.34630527580919224,
      "grad_norm": 0.4372621178627014,
      "learning_rate": 3.2686808404824074e-06,
      "loss": 0.7126,
      "step": 8356
    },
    {
      "epoch": 0.3463467197148659,
      "grad_norm": 0.41523173451423645,
      "learning_rate": 3.268473620954039e-06,
      "loss": 0.6899,
      "step": 8357
    },
    {
      "epoch": 0.3463881636205396,
      "grad_norm": 0.3875006139278412,
      "learning_rate": 3.2682664014256706e-06,
      "loss": 0.6699,
      "step": 8358
    },
    {
      "epoch": 0.34642960752621327,
      "grad_norm": 0.4124395251274109,
      "learning_rate": 3.2680591818973024e-06,
      "loss": 0.6887,
      "step": 8359
    },
    {
      "epoch": 0.34647105143188694,
      "grad_norm": 0.422855019569397,
      "learning_rate": 3.2678519623689338e-06,
      "loss": 0.7189,
      "step": 8360
    },
    {
      "epoch": 0.3465124953375606,
      "grad_norm": 0.4258534014225006,
      "learning_rate": 3.2676447428405656e-06,
      "loss": 0.7263,
      "step": 8361
    },
    {
      "epoch": 0.3465539392432343,
      "grad_norm": 0.40111246705055237,
      "learning_rate": 3.2674375233121974e-06,
      "loss": 0.7114,
      "step": 8362
    },
    {
      "epoch": 0.346595383148908,
      "grad_norm": 0.38914307951927185,
      "learning_rate": 3.2672303037838288e-06,
      "loss": 0.6735,
      "step": 8363
    },
    {
      "epoch": 0.3466368270545816,
      "grad_norm": 0.40925362706184387,
      "learning_rate": 3.2670230842554606e-06,
      "loss": 0.6815,
      "step": 8364
    },
    {
      "epoch": 0.3466782709602553,
      "grad_norm": 0.4222832918167114,
      "learning_rate": 3.266815864727092e-06,
      "loss": 0.6906,
      "step": 8365
    },
    {
      "epoch": 0.34671971486592895,
      "grad_norm": 0.43050020933151245,
      "learning_rate": 3.2666086451987238e-06,
      "loss": 0.6782,
      "step": 8366
    },
    {
      "epoch": 0.34676115877160263,
      "grad_norm": 0.43644094467163086,
      "learning_rate": 3.266401425670355e-06,
      "loss": 0.7517,
      "step": 8367
    },
    {
      "epoch": 0.3468026026772763,
      "grad_norm": 0.38347160816192627,
      "learning_rate": 3.266194206141987e-06,
      "loss": 0.6982,
      "step": 8368
    },
    {
      "epoch": 0.34684404658295,
      "grad_norm": 0.4532240331172943,
      "learning_rate": 3.2659869866136184e-06,
      "loss": 0.7197,
      "step": 8369
    },
    {
      "epoch": 0.34688549048862366,
      "grad_norm": 0.41144469380378723,
      "learning_rate": 3.26577976708525e-06,
      "loss": 0.6802,
      "step": 8370
    },
    {
      "epoch": 0.34692693439429734,
      "grad_norm": 0.4099407494068146,
      "learning_rate": 3.2655725475568824e-06,
      "loss": 0.7197,
      "step": 8371
    },
    {
      "epoch": 0.346968378299971,
      "grad_norm": 0.4614093005657196,
      "learning_rate": 3.265365328028514e-06,
      "loss": 0.7495,
      "step": 8372
    },
    {
      "epoch": 0.34700982220564464,
      "grad_norm": 0.41772952675819397,
      "learning_rate": 3.2651581085001456e-06,
      "loss": 0.7375,
      "step": 8373
    },
    {
      "epoch": 0.3470512661113183,
      "grad_norm": 0.401076078414917,
      "learning_rate": 3.264950888971777e-06,
      "loss": 0.614,
      "step": 8374
    },
    {
      "epoch": 0.347092710016992,
      "grad_norm": 0.41139674186706543,
      "learning_rate": 3.264743669443409e-06,
      "loss": 0.7246,
      "step": 8375
    },
    {
      "epoch": 0.34713415392266567,
      "grad_norm": 0.40465858578681946,
      "learning_rate": 3.26453644991504e-06,
      "loss": 0.696,
      "step": 8376
    },
    {
      "epoch": 0.34717559782833934,
      "grad_norm": 0.41757339239120483,
      "learning_rate": 3.264329230386672e-06,
      "loss": 0.7158,
      "step": 8377
    },
    {
      "epoch": 0.347217041734013,
      "grad_norm": 0.4280861020088196,
      "learning_rate": 3.264122010858304e-06,
      "loss": 0.708,
      "step": 8378
    },
    {
      "epoch": 0.3472584856396867,
      "grad_norm": 0.42357343435287476,
      "learning_rate": 3.263914791329935e-06,
      "loss": 0.7068,
      "step": 8379
    },
    {
      "epoch": 0.3472999295453604,
      "grad_norm": 0.3917806148529053,
      "learning_rate": 3.263707571801567e-06,
      "loss": 0.6782,
      "step": 8380
    },
    {
      "epoch": 0.34734137345103405,
      "grad_norm": 0.4386560320854187,
      "learning_rate": 3.2635003522731984e-06,
      "loss": 0.7183,
      "step": 8381
    },
    {
      "epoch": 0.3473828173567077,
      "grad_norm": 0.4434758126735687,
      "learning_rate": 3.26329313274483e-06,
      "loss": 0.7498,
      "step": 8382
    },
    {
      "epoch": 0.34742426126238135,
      "grad_norm": 0.4160711169242859,
      "learning_rate": 3.2630859132164616e-06,
      "loss": 0.6963,
      "step": 8383
    },
    {
      "epoch": 0.34746570516805503,
      "grad_norm": 0.4238876402378082,
      "learning_rate": 3.2628786936880934e-06,
      "loss": 0.7671,
      "step": 8384
    },
    {
      "epoch": 0.3475071490737287,
      "grad_norm": 0.4074348509311676,
      "learning_rate": 3.2626714741597248e-06,
      "loss": 0.74,
      "step": 8385
    },
    {
      "epoch": 0.3475485929794024,
      "grad_norm": 0.4182678163051605,
      "learning_rate": 3.2624642546313566e-06,
      "loss": 0.7551,
      "step": 8386
    },
    {
      "epoch": 0.34759003688507606,
      "grad_norm": 0.40148499608039856,
      "learning_rate": 3.262257035102989e-06,
      "loss": 0.7255,
      "step": 8387
    },
    {
      "epoch": 0.34763148079074974,
      "grad_norm": 0.39570698142051697,
      "learning_rate": 3.2620498155746198e-06,
      "loss": 0.6349,
      "step": 8388
    },
    {
      "epoch": 0.3476729246964234,
      "grad_norm": 0.44137296080589294,
      "learning_rate": 3.261842596046252e-06,
      "loss": 0.66,
      "step": 8389
    },
    {
      "epoch": 0.34771436860209703,
      "grad_norm": 0.38619688153266907,
      "learning_rate": 3.2616353765178834e-06,
      "loss": 0.6779,
      "step": 8390
    },
    {
      "epoch": 0.3477558125077707,
      "grad_norm": 0.3878011405467987,
      "learning_rate": 3.261428156989515e-06,
      "loss": 0.6719,
      "step": 8391
    },
    {
      "epoch": 0.3477972564134444,
      "grad_norm": 0.4060506522655487,
      "learning_rate": 3.2612209374611466e-06,
      "loss": 0.6801,
      "step": 8392
    },
    {
      "epoch": 0.34783870031911807,
      "grad_norm": 0.47176098823547363,
      "learning_rate": 3.2610137179327784e-06,
      "loss": 0.7344,
      "step": 8393
    },
    {
      "epoch": 0.34788014422479174,
      "grad_norm": 0.4307069778442383,
      "learning_rate": 3.2608064984044098e-06,
      "loss": 0.6584,
      "step": 8394
    },
    {
      "epoch": 0.3479215881304654,
      "grad_norm": 0.39372867345809937,
      "learning_rate": 3.2605992788760416e-06,
      "loss": 0.7014,
      "step": 8395
    },
    {
      "epoch": 0.3479630320361391,
      "grad_norm": 0.42050257325172424,
      "learning_rate": 3.2603920593476734e-06,
      "loss": 0.666,
      "step": 8396
    },
    {
      "epoch": 0.3480044759418128,
      "grad_norm": 0.3794766068458557,
      "learning_rate": 3.2601848398193048e-06,
      "loss": 0.6598,
      "step": 8397
    },
    {
      "epoch": 0.34804591984748645,
      "grad_norm": 0.39259037375450134,
      "learning_rate": 3.2599776202909366e-06,
      "loss": 0.6631,
      "step": 8398
    },
    {
      "epoch": 0.3480873637531601,
      "grad_norm": 0.3963391184806824,
      "learning_rate": 3.259770400762568e-06,
      "loss": 0.6823,
      "step": 8399
    },
    {
      "epoch": 0.34812880765883375,
      "grad_norm": 0.4069724678993225,
      "learning_rate": 3.2595631812342e-06,
      "loss": 0.7026,
      "step": 8400
    },
    {
      "epoch": 0.3481702515645074,
      "grad_norm": 0.41394898295402527,
      "learning_rate": 3.259355961705831e-06,
      "loss": 0.7729,
      "step": 8401
    },
    {
      "epoch": 0.3482116954701811,
      "grad_norm": 0.4288797080516815,
      "learning_rate": 3.259148742177463e-06,
      "loss": 0.6653,
      "step": 8402
    },
    {
      "epoch": 0.3482531393758548,
      "grad_norm": 0.401050865650177,
      "learning_rate": 3.2589415226490944e-06,
      "loss": 0.6594,
      "step": 8403
    },
    {
      "epoch": 0.34829458328152846,
      "grad_norm": 0.40262600779533386,
      "learning_rate": 3.258734303120726e-06,
      "loss": 0.6937,
      "step": 8404
    },
    {
      "epoch": 0.34833602718720214,
      "grad_norm": 0.3903612792491913,
      "learning_rate": 3.2585270835923584e-06,
      "loss": 0.6422,
      "step": 8405
    },
    {
      "epoch": 0.3483774710928758,
      "grad_norm": 0.44177308678627014,
      "learning_rate": 3.25831986406399e-06,
      "loss": 0.71,
      "step": 8406
    },
    {
      "epoch": 0.3484189149985495,
      "grad_norm": 0.4801948368549347,
      "learning_rate": 3.2581126445356216e-06,
      "loss": 0.6855,
      "step": 8407
    },
    {
      "epoch": 0.3484603589042231,
      "grad_norm": 0.4117943048477173,
      "learning_rate": 3.257905425007253e-06,
      "loss": 0.6697,
      "step": 8408
    },
    {
      "epoch": 0.3485018028098968,
      "grad_norm": 0.40243151783943176,
      "learning_rate": 3.257698205478885e-06,
      "loss": 0.7283,
      "step": 8409
    },
    {
      "epoch": 0.34854324671557046,
      "grad_norm": 0.4275542199611664,
      "learning_rate": 3.257490985950516e-06,
      "loss": 0.7394,
      "step": 8410
    },
    {
      "epoch": 0.34858469062124414,
      "grad_norm": 0.4015789330005646,
      "learning_rate": 3.257283766422148e-06,
      "loss": 0.6936,
      "step": 8411
    },
    {
      "epoch": 0.3486261345269178,
      "grad_norm": 0.4187641143798828,
      "learning_rate": 3.2570765468937794e-06,
      "loss": 0.6703,
      "step": 8412
    },
    {
      "epoch": 0.3486675784325915,
      "grad_norm": 0.43096086382865906,
      "learning_rate": 3.256869327365411e-06,
      "loss": 0.7397,
      "step": 8413
    },
    {
      "epoch": 0.3487090223382652,
      "grad_norm": 0.39312028884887695,
      "learning_rate": 3.256662107837043e-06,
      "loss": 0.7458,
      "step": 8414
    },
    {
      "epoch": 0.34875046624393885,
      "grad_norm": 0.44121694564819336,
      "learning_rate": 3.2564548883086744e-06,
      "loss": 0.6812,
      "step": 8415
    },
    {
      "epoch": 0.34879191014961247,
      "grad_norm": 0.42621079087257385,
      "learning_rate": 3.256247668780306e-06,
      "loss": 0.6949,
      "step": 8416
    },
    {
      "epoch": 0.34883335405528615,
      "grad_norm": 0.4094623029232025,
      "learning_rate": 3.2560404492519376e-06,
      "loss": 0.7236,
      "step": 8417
    },
    {
      "epoch": 0.3488747979609598,
      "grad_norm": 0.42307811975479126,
      "learning_rate": 3.2558332297235694e-06,
      "loss": 0.7197,
      "step": 8418
    },
    {
      "epoch": 0.3489162418666335,
      "grad_norm": 0.4493318796157837,
      "learning_rate": 3.2556260101952008e-06,
      "loss": 0.7439,
      "step": 8419
    },
    {
      "epoch": 0.3489576857723072,
      "grad_norm": 0.4237443804740906,
      "learning_rate": 3.2554187906668326e-06,
      "loss": 0.7102,
      "step": 8420
    },
    {
      "epoch": 0.34899912967798086,
      "grad_norm": 0.43112456798553467,
      "learning_rate": 3.255211571138464e-06,
      "loss": 0.7421,
      "step": 8421
    },
    {
      "epoch": 0.34904057358365453,
      "grad_norm": 0.42541006207466125,
      "learning_rate": 3.2550043516100958e-06,
      "loss": 0.7278,
      "step": 8422
    },
    {
      "epoch": 0.3490820174893282,
      "grad_norm": 0.3866768777370453,
      "learning_rate": 3.254797132081728e-06,
      "loss": 0.6809,
      "step": 8423
    },
    {
      "epoch": 0.3491234613950019,
      "grad_norm": 0.4028589725494385,
      "learning_rate": 3.2545899125533594e-06,
      "loss": 0.6862,
      "step": 8424
    },
    {
      "epoch": 0.3491649053006755,
      "grad_norm": 0.4218546748161316,
      "learning_rate": 3.254382693024991e-06,
      "loss": 0.6948,
      "step": 8425
    },
    {
      "epoch": 0.3492063492063492,
      "grad_norm": 0.43646878004074097,
      "learning_rate": 3.2541754734966226e-06,
      "loss": 0.7109,
      "step": 8426
    },
    {
      "epoch": 0.34924779311202286,
      "grad_norm": 0.39836540818214417,
      "learning_rate": 3.2539682539682544e-06,
      "loss": 0.7147,
      "step": 8427
    },
    {
      "epoch": 0.34928923701769654,
      "grad_norm": 0.40244823694229126,
      "learning_rate": 3.2537610344398858e-06,
      "loss": 0.6931,
      "step": 8428
    },
    {
      "epoch": 0.3493306809233702,
      "grad_norm": 0.39114394783973694,
      "learning_rate": 3.2535538149115176e-06,
      "loss": 0.7086,
      "step": 8429
    },
    {
      "epoch": 0.3493721248290439,
      "grad_norm": 0.4192892014980316,
      "learning_rate": 3.253346595383149e-06,
      "loss": 0.7026,
      "step": 8430
    },
    {
      "epoch": 0.3494135687347176,
      "grad_norm": 0.40353110432624817,
      "learning_rate": 3.253139375854781e-06,
      "loss": 0.7236,
      "step": 8431
    },
    {
      "epoch": 0.34945501264039125,
      "grad_norm": 0.42545658349990845,
      "learning_rate": 3.2529321563264126e-06,
      "loss": 0.7097,
      "step": 8432
    },
    {
      "epoch": 0.3494964565460649,
      "grad_norm": 0.4859357178211212,
      "learning_rate": 3.252724936798044e-06,
      "loss": 0.7271,
      "step": 8433
    },
    {
      "epoch": 0.34953790045173855,
      "grad_norm": 0.398715078830719,
      "learning_rate": 3.252517717269676e-06,
      "loss": 0.6731,
      "step": 8434
    },
    {
      "epoch": 0.3495793443574122,
      "grad_norm": 0.40515193343162537,
      "learning_rate": 3.252310497741307e-06,
      "loss": 0.6179,
      "step": 8435
    },
    {
      "epoch": 0.3496207882630859,
      "grad_norm": 0.39331522583961487,
      "learning_rate": 3.252103278212939e-06,
      "loss": 0.6698,
      "step": 8436
    },
    {
      "epoch": 0.3496622321687596,
      "grad_norm": 0.4430435299873352,
      "learning_rate": 3.2518960586845704e-06,
      "loss": 0.7278,
      "step": 8437
    },
    {
      "epoch": 0.34970367607443326,
      "grad_norm": 0.4872400462627411,
      "learning_rate": 3.251688839156202e-06,
      "loss": 0.7754,
      "step": 8438
    },
    {
      "epoch": 0.34974511998010693,
      "grad_norm": 0.40427038073539734,
      "learning_rate": 3.2514816196278344e-06,
      "loss": 0.7258,
      "step": 8439
    },
    {
      "epoch": 0.3497865638857806,
      "grad_norm": 0.403890997171402,
      "learning_rate": 3.251274400099466e-06,
      "loss": 0.6477,
      "step": 8440
    },
    {
      "epoch": 0.3498280077914543,
      "grad_norm": 0.3872579038143158,
      "learning_rate": 3.2510671805710976e-06,
      "loss": 0.6703,
      "step": 8441
    },
    {
      "epoch": 0.34986945169712796,
      "grad_norm": 0.37973129749298096,
      "learning_rate": 3.250859961042729e-06,
      "loss": 0.6924,
      "step": 8442
    },
    {
      "epoch": 0.3499108956028016,
      "grad_norm": 0.42740708589553833,
      "learning_rate": 3.250652741514361e-06,
      "loss": 0.7212,
      "step": 8443
    },
    {
      "epoch": 0.34995233950847526,
      "grad_norm": 0.4297677278518677,
      "learning_rate": 3.250445521985992e-06,
      "loss": 0.72,
      "step": 8444
    },
    {
      "epoch": 0.34999378341414894,
      "grad_norm": 0.4436231553554535,
      "learning_rate": 3.250238302457624e-06,
      "loss": 0.7188,
      "step": 8445
    },
    {
      "epoch": 0.3500352273198226,
      "grad_norm": 0.4219163656234741,
      "learning_rate": 3.2500310829292554e-06,
      "loss": 0.7209,
      "step": 8446
    },
    {
      "epoch": 0.3500766712254963,
      "grad_norm": 0.4114866256713867,
      "learning_rate": 3.249823863400887e-06,
      "loss": 0.666,
      "step": 8447
    },
    {
      "epoch": 0.35011811513116997,
      "grad_norm": 0.4007662534713745,
      "learning_rate": 3.249616643872519e-06,
      "loss": 0.7061,
      "step": 8448
    },
    {
      "epoch": 0.35015955903684365,
      "grad_norm": 0.41682374477386475,
      "learning_rate": 3.2494094243441504e-06,
      "loss": 0.6858,
      "step": 8449
    },
    {
      "epoch": 0.3502010029425173,
      "grad_norm": 0.4100598394870758,
      "learning_rate": 3.249202204815782e-06,
      "loss": 0.718,
      "step": 8450
    },
    {
      "epoch": 0.35024244684819095,
      "grad_norm": 0.38997456431388855,
      "learning_rate": 3.2489949852874136e-06,
      "loss": 0.6973,
      "step": 8451
    },
    {
      "epoch": 0.3502838907538646,
      "grad_norm": 0.38735389709472656,
      "learning_rate": 3.2487877657590454e-06,
      "loss": 0.6819,
      "step": 8452
    },
    {
      "epoch": 0.3503253346595383,
      "grad_norm": 0.42380228638648987,
      "learning_rate": 3.2485805462306768e-06,
      "loss": 0.7572,
      "step": 8453
    },
    {
      "epoch": 0.350366778565212,
      "grad_norm": 0.40912526845932007,
      "learning_rate": 3.2483733267023086e-06,
      "loss": 0.6731,
      "step": 8454
    },
    {
      "epoch": 0.35040822247088566,
      "grad_norm": 0.39777782559394836,
      "learning_rate": 3.24816610717394e-06,
      "loss": 0.6802,
      "step": 8455
    },
    {
      "epoch": 0.35044966637655933,
      "grad_norm": 0.40776360034942627,
      "learning_rate": 3.2479588876455718e-06,
      "loss": 0.6731,
      "step": 8456
    },
    {
      "epoch": 0.350491110282233,
      "grad_norm": 0.43636733293533325,
      "learning_rate": 3.247751668117204e-06,
      "loss": 0.6462,
      "step": 8457
    },
    {
      "epoch": 0.3505325541879067,
      "grad_norm": 0.4054887890815735,
      "learning_rate": 3.2475444485888354e-06,
      "loss": 0.7383,
      "step": 8458
    },
    {
      "epoch": 0.35057399809358036,
      "grad_norm": 0.41566285490989685,
      "learning_rate": 3.2473372290604672e-06,
      "loss": 0.7163,
      "step": 8459
    },
    {
      "epoch": 0.350615441999254,
      "grad_norm": 0.4094327390193939,
      "learning_rate": 3.2471300095320986e-06,
      "loss": 0.7261,
      "step": 8460
    },
    {
      "epoch": 0.35065688590492766,
      "grad_norm": 0.40990570187568665,
      "learning_rate": 3.2469227900037304e-06,
      "loss": 0.657,
      "step": 8461
    },
    {
      "epoch": 0.35069832981060134,
      "grad_norm": 0.42081236839294434,
      "learning_rate": 3.246715570475362e-06,
      "loss": 0.7111,
      "step": 8462
    },
    {
      "epoch": 0.350739773716275,
      "grad_norm": 0.425610214471817,
      "learning_rate": 3.2465083509469936e-06,
      "loss": 0.6892,
      "step": 8463
    },
    {
      "epoch": 0.3507812176219487,
      "grad_norm": 0.4749346673488617,
      "learning_rate": 3.246301131418625e-06,
      "loss": 0.6796,
      "step": 8464
    },
    {
      "epoch": 0.35082266152762237,
      "grad_norm": 0.3954388201236725,
      "learning_rate": 3.246093911890257e-06,
      "loss": 0.6506,
      "step": 8465
    },
    {
      "epoch": 0.35086410543329605,
      "grad_norm": 0.4021048843860626,
      "learning_rate": 3.2458866923618886e-06,
      "loss": 0.71,
      "step": 8466
    },
    {
      "epoch": 0.3509055493389697,
      "grad_norm": 0.41527876257896423,
      "learning_rate": 3.24567947283352e-06,
      "loss": 0.7135,
      "step": 8467
    },
    {
      "epoch": 0.3509469932446434,
      "grad_norm": 0.4124913811683655,
      "learning_rate": 3.245472253305152e-06,
      "loss": 0.7153,
      "step": 8468
    },
    {
      "epoch": 0.350988437150317,
      "grad_norm": 0.40956151485443115,
      "learning_rate": 3.245265033776783e-06,
      "loss": 0.708,
      "step": 8469
    },
    {
      "epoch": 0.3510298810559907,
      "grad_norm": 0.39759552478790283,
      "learning_rate": 3.245057814248415e-06,
      "loss": 0.6559,
      "step": 8470
    },
    {
      "epoch": 0.3510713249616644,
      "grad_norm": 0.40804463624954224,
      "learning_rate": 3.2448505947200464e-06,
      "loss": 0.7258,
      "step": 8471
    },
    {
      "epoch": 0.35111276886733805,
      "grad_norm": 0.3893631100654602,
      "learning_rate": 3.244643375191678e-06,
      "loss": 0.7445,
      "step": 8472
    },
    {
      "epoch": 0.35115421277301173,
      "grad_norm": 0.4003143012523651,
      "learning_rate": 3.2444361556633096e-06,
      "loss": 0.6803,
      "step": 8473
    },
    {
      "epoch": 0.3511956566786854,
      "grad_norm": 0.373654842376709,
      "learning_rate": 3.244228936134942e-06,
      "loss": 0.6555,
      "step": 8474
    },
    {
      "epoch": 0.3512371005843591,
      "grad_norm": 0.4565506875514984,
      "learning_rate": 3.2440217166065736e-06,
      "loss": 0.6962,
      "step": 8475
    },
    {
      "epoch": 0.35127854449003276,
      "grad_norm": 0.41194650530815125,
      "learning_rate": 3.243814497078205e-06,
      "loss": 0.7195,
      "step": 8476
    },
    {
      "epoch": 0.3513199883957064,
      "grad_norm": 0.40757957100868225,
      "learning_rate": 3.243607277549837e-06,
      "loss": 0.717,
      "step": 8477
    },
    {
      "epoch": 0.35136143230138006,
      "grad_norm": 0.4242447018623352,
      "learning_rate": 3.243400058021468e-06,
      "loss": 0.7292,
      "step": 8478
    },
    {
      "epoch": 0.35140287620705374,
      "grad_norm": 0.3864867687225342,
      "learning_rate": 3.2431928384931e-06,
      "loss": 0.7109,
      "step": 8479
    },
    {
      "epoch": 0.3514443201127274,
      "grad_norm": 0.44120100140571594,
      "learning_rate": 3.2429856189647314e-06,
      "loss": 0.6929,
      "step": 8480
    },
    {
      "epoch": 0.3514857640184011,
      "grad_norm": 0.43264901638031006,
      "learning_rate": 3.242778399436363e-06,
      "loss": 0.744,
      "step": 8481
    },
    {
      "epoch": 0.35152720792407477,
      "grad_norm": 0.4202939569950104,
      "learning_rate": 3.2425711799079946e-06,
      "loss": 0.6807,
      "step": 8482
    },
    {
      "epoch": 0.35156865182974845,
      "grad_norm": 0.3774070143699646,
      "learning_rate": 3.2423639603796264e-06,
      "loss": 0.6539,
      "step": 8483
    },
    {
      "epoch": 0.3516100957354221,
      "grad_norm": 0.37859976291656494,
      "learning_rate": 3.242156740851258e-06,
      "loss": 0.6604,
      "step": 8484
    },
    {
      "epoch": 0.3516515396410958,
      "grad_norm": 0.40818437933921814,
      "learning_rate": 3.2419495213228896e-06,
      "loss": 0.7036,
      "step": 8485
    },
    {
      "epoch": 0.3516929835467694,
      "grad_norm": 0.4266854524612427,
      "learning_rate": 3.2417423017945214e-06,
      "loss": 0.6805,
      "step": 8486
    },
    {
      "epoch": 0.3517344274524431,
      "grad_norm": 0.41126519441604614,
      "learning_rate": 3.241535082266153e-06,
      "loss": 0.6772,
      "step": 8487
    },
    {
      "epoch": 0.3517758713581168,
      "grad_norm": 0.4264627695083618,
      "learning_rate": 3.2413278627377846e-06,
      "loss": 0.7246,
      "step": 8488
    },
    {
      "epoch": 0.35181731526379045,
      "grad_norm": 0.43510106205940247,
      "learning_rate": 3.241120643209416e-06,
      "loss": 0.6871,
      "step": 8489
    },
    {
      "epoch": 0.35185875916946413,
      "grad_norm": 0.4071009159088135,
      "learning_rate": 3.240913423681048e-06,
      "loss": 0.7305,
      "step": 8490
    },
    {
      "epoch": 0.3519002030751378,
      "grad_norm": 0.4376145899295807,
      "learning_rate": 3.24070620415268e-06,
      "loss": 0.7239,
      "step": 8491
    },
    {
      "epoch": 0.3519416469808115,
      "grad_norm": 0.39137589931488037,
      "learning_rate": 3.2404989846243114e-06,
      "loss": 0.6686,
      "step": 8492
    },
    {
      "epoch": 0.35198309088648516,
      "grad_norm": 0.4495477080345154,
      "learning_rate": 3.2402917650959432e-06,
      "loss": 0.738,
      "step": 8493
    },
    {
      "epoch": 0.35202453479215884,
      "grad_norm": 0.4391402304172516,
      "learning_rate": 3.2400845455675746e-06,
      "loss": 0.7415,
      "step": 8494
    },
    {
      "epoch": 0.35206597869783246,
      "grad_norm": 0.43204179406166077,
      "learning_rate": 3.2398773260392064e-06,
      "loss": 0.7378,
      "step": 8495
    },
    {
      "epoch": 0.35210742260350614,
      "grad_norm": 0.4314158260822296,
      "learning_rate": 3.239670106510838e-06,
      "loss": 0.7002,
      "step": 8496
    },
    {
      "epoch": 0.3521488665091798,
      "grad_norm": 0.40520891547203064,
      "learning_rate": 3.2394628869824696e-06,
      "loss": 0.7117,
      "step": 8497
    },
    {
      "epoch": 0.3521903104148535,
      "grad_norm": 0.41262534260749817,
      "learning_rate": 3.239255667454101e-06,
      "loss": 0.723,
      "step": 8498
    },
    {
      "epoch": 0.35223175432052717,
      "grad_norm": 0.3972526490688324,
      "learning_rate": 3.239048447925733e-06,
      "loss": 0.7563,
      "step": 8499
    },
    {
      "epoch": 0.35227319822620085,
      "grad_norm": 0.41168370842933655,
      "learning_rate": 3.2388412283973646e-06,
      "loss": 0.7366,
      "step": 8500
    },
    {
      "epoch": 0.3523146421318745,
      "grad_norm": 0.3810071349143982,
      "learning_rate": 3.238634008868996e-06,
      "loss": 0.6591,
      "step": 8501
    },
    {
      "epoch": 0.3523560860375482,
      "grad_norm": 0.38791337609291077,
      "learning_rate": 3.238426789340628e-06,
      "loss": 0.6993,
      "step": 8502
    },
    {
      "epoch": 0.3523975299432219,
      "grad_norm": 0.42793065309524536,
      "learning_rate": 3.238219569812259e-06,
      "loss": 0.6914,
      "step": 8503
    },
    {
      "epoch": 0.3524389738488955,
      "grad_norm": 0.4107297360897064,
      "learning_rate": 3.238012350283891e-06,
      "loss": 0.7444,
      "step": 8504
    },
    {
      "epoch": 0.3524804177545692,
      "grad_norm": 0.3796500265598297,
      "learning_rate": 3.2378051307555224e-06,
      "loss": 0.6554,
      "step": 8505
    },
    {
      "epoch": 0.35252186166024285,
      "grad_norm": 0.4484458863735199,
      "learning_rate": 3.237597911227154e-06,
      "loss": 0.7281,
      "step": 8506
    },
    {
      "epoch": 0.35256330556591653,
      "grad_norm": 0.42883116006851196,
      "learning_rate": 3.2373906916987856e-06,
      "loss": 0.7,
      "step": 8507
    },
    {
      "epoch": 0.3526047494715902,
      "grad_norm": 0.3968334197998047,
      "learning_rate": 3.237183472170418e-06,
      "loss": 0.7134,
      "step": 8508
    },
    {
      "epoch": 0.3526461933772639,
      "grad_norm": 0.4121285378932953,
      "learning_rate": 3.2369762526420496e-06,
      "loss": 0.6968,
      "step": 8509
    },
    {
      "epoch": 0.35268763728293756,
      "grad_norm": 0.39721494913101196,
      "learning_rate": 3.236769033113681e-06,
      "loss": 0.6797,
      "step": 8510
    },
    {
      "epoch": 0.35272908118861124,
      "grad_norm": 0.4493896961212158,
      "learning_rate": 3.236561813585313e-06,
      "loss": 0.6765,
      "step": 8511
    },
    {
      "epoch": 0.35277052509428486,
      "grad_norm": 0.4610176682472229,
      "learning_rate": 3.236354594056944e-06,
      "loss": 0.6783,
      "step": 8512
    },
    {
      "epoch": 0.35281196899995854,
      "grad_norm": 0.41171789169311523,
      "learning_rate": 3.236147374528576e-06,
      "loss": 0.7322,
      "step": 8513
    },
    {
      "epoch": 0.3528534129056322,
      "grad_norm": 0.4150131940841675,
      "learning_rate": 3.2359401550002074e-06,
      "loss": 0.6993,
      "step": 8514
    },
    {
      "epoch": 0.3528948568113059,
      "grad_norm": 0.41514599323272705,
      "learning_rate": 3.235732935471839e-06,
      "loss": 0.7449,
      "step": 8515
    },
    {
      "epoch": 0.35293630071697957,
      "grad_norm": 0.5728110074996948,
      "learning_rate": 3.2355257159434706e-06,
      "loss": 0.7488,
      "step": 8516
    },
    {
      "epoch": 0.35297774462265324,
      "grad_norm": 0.3947862982749939,
      "learning_rate": 3.2353184964151024e-06,
      "loss": 0.678,
      "step": 8517
    },
    {
      "epoch": 0.3530191885283269,
      "grad_norm": 0.4125240743160248,
      "learning_rate": 3.2351112768867342e-06,
      "loss": 0.6694,
      "step": 8518
    },
    {
      "epoch": 0.3530606324340006,
      "grad_norm": 0.437519371509552,
      "learning_rate": 3.2349040573583656e-06,
      "loss": 0.6956,
      "step": 8519
    },
    {
      "epoch": 0.3531020763396743,
      "grad_norm": 0.42778247594833374,
      "learning_rate": 3.2346968378299974e-06,
      "loss": 0.7063,
      "step": 8520
    },
    {
      "epoch": 0.3531435202453479,
      "grad_norm": 0.40842971205711365,
      "learning_rate": 3.234489618301629e-06,
      "loss": 0.7034,
      "step": 8521
    },
    {
      "epoch": 0.3531849641510216,
      "grad_norm": 0.4150451421737671,
      "learning_rate": 3.2342823987732606e-06,
      "loss": 0.6895,
      "step": 8522
    },
    {
      "epoch": 0.35322640805669525,
      "grad_norm": 0.41860222816467285,
      "learning_rate": 3.234075179244892e-06,
      "loss": 0.7441,
      "step": 8523
    },
    {
      "epoch": 0.35326785196236893,
      "grad_norm": 0.4254886209964752,
      "learning_rate": 3.233867959716524e-06,
      "loss": 0.689,
      "step": 8524
    },
    {
      "epoch": 0.3533092958680426,
      "grad_norm": 0.3855765163898468,
      "learning_rate": 3.233660740188155e-06,
      "loss": 0.6509,
      "step": 8525
    },
    {
      "epoch": 0.3533507397737163,
      "grad_norm": 0.4651561677455902,
      "learning_rate": 3.2334535206597874e-06,
      "loss": 0.7357,
      "step": 8526
    },
    {
      "epoch": 0.35339218367938996,
      "grad_norm": 0.39483535289764404,
      "learning_rate": 3.2332463011314192e-06,
      "loss": 0.6572,
      "step": 8527
    },
    {
      "epoch": 0.35343362758506364,
      "grad_norm": 0.4444069564342499,
      "learning_rate": 3.2330390816030506e-06,
      "loss": 0.7341,
      "step": 8528
    },
    {
      "epoch": 0.3534750714907373,
      "grad_norm": 0.3630324602127075,
      "learning_rate": 3.2328318620746824e-06,
      "loss": 0.6625,
      "step": 8529
    },
    {
      "epoch": 0.35351651539641094,
      "grad_norm": 0.38817235827445984,
      "learning_rate": 3.232624642546314e-06,
      "loss": 0.6934,
      "step": 8530
    },
    {
      "epoch": 0.3535579593020846,
      "grad_norm": 0.4378013610839844,
      "learning_rate": 3.2324174230179456e-06,
      "loss": 0.7349,
      "step": 8531
    },
    {
      "epoch": 0.3535994032077583,
      "grad_norm": 0.43015584349632263,
      "learning_rate": 3.232210203489577e-06,
      "loss": 0.7133,
      "step": 8532
    },
    {
      "epoch": 0.35364084711343197,
      "grad_norm": 0.40005186200141907,
      "learning_rate": 3.232002983961209e-06,
      "loss": 0.6703,
      "step": 8533
    },
    {
      "epoch": 0.35368229101910564,
      "grad_norm": 0.4355786442756653,
      "learning_rate": 3.23179576443284e-06,
      "loss": 0.6956,
      "step": 8534
    },
    {
      "epoch": 0.3537237349247793,
      "grad_norm": 0.43644124269485474,
      "learning_rate": 3.231588544904472e-06,
      "loss": 0.7417,
      "step": 8535
    },
    {
      "epoch": 0.353765178830453,
      "grad_norm": 0.39708253741264343,
      "learning_rate": 3.231381325376104e-06,
      "loss": 0.7185,
      "step": 8536
    },
    {
      "epoch": 0.3538066227361267,
      "grad_norm": 0.4361204504966736,
      "learning_rate": 3.231174105847735e-06,
      "loss": 0.6833,
      "step": 8537
    },
    {
      "epoch": 0.3538480666418003,
      "grad_norm": 0.41886967420578003,
      "learning_rate": 3.230966886319367e-06,
      "loss": 0.662,
      "step": 8538
    },
    {
      "epoch": 0.353889510547474,
      "grad_norm": 0.4427701532840729,
      "learning_rate": 3.2307596667909984e-06,
      "loss": 0.7168,
      "step": 8539
    },
    {
      "epoch": 0.35393095445314765,
      "grad_norm": 0.39517053961753845,
      "learning_rate": 3.23055244726263e-06,
      "loss": 0.7024,
      "step": 8540
    },
    {
      "epoch": 0.3539723983588213,
      "grad_norm": 0.45205774903297424,
      "learning_rate": 3.2303452277342616e-06,
      "loss": 0.7,
      "step": 8541
    },
    {
      "epoch": 0.354013842264495,
      "grad_norm": 0.4129032492637634,
      "learning_rate": 3.230138008205894e-06,
      "loss": 0.6774,
      "step": 8542
    },
    {
      "epoch": 0.3540552861701687,
      "grad_norm": 0.4169149100780487,
      "learning_rate": 3.2299307886775248e-06,
      "loss": 0.7366,
      "step": 8543
    },
    {
      "epoch": 0.35409673007584236,
      "grad_norm": 0.4418199360370636,
      "learning_rate": 3.229723569149157e-06,
      "loss": 0.7131,
      "step": 8544
    },
    {
      "epoch": 0.35413817398151604,
      "grad_norm": 0.40951967239379883,
      "learning_rate": 3.229516349620789e-06,
      "loss": 0.7048,
      "step": 8545
    },
    {
      "epoch": 0.3541796178871897,
      "grad_norm": 0.4091774821281433,
      "learning_rate": 3.2293091300924202e-06,
      "loss": 0.6365,
      "step": 8546
    },
    {
      "epoch": 0.35422106179286333,
      "grad_norm": 0.4138348400592804,
      "learning_rate": 3.229101910564052e-06,
      "loss": 0.7162,
      "step": 8547
    },
    {
      "epoch": 0.354262505698537,
      "grad_norm": 0.47510600090026855,
      "learning_rate": 3.2288946910356834e-06,
      "loss": 0.6571,
      "step": 8548
    },
    {
      "epoch": 0.3543039496042107,
      "grad_norm": 0.42789509892463684,
      "learning_rate": 3.2286874715073152e-06,
      "loss": 0.7278,
      "step": 8549
    },
    {
      "epoch": 0.35434539350988437,
      "grad_norm": 0.4559313952922821,
      "learning_rate": 3.2284802519789466e-06,
      "loss": 0.7485,
      "step": 8550
    },
    {
      "epoch": 0.35438683741555804,
      "grad_norm": 0.38186684250831604,
      "learning_rate": 3.2282730324505784e-06,
      "loss": 0.661,
      "step": 8551
    },
    {
      "epoch": 0.3544282813212317,
      "grad_norm": 0.41352972388267517,
      "learning_rate": 3.2280658129222102e-06,
      "loss": 0.7067,
      "step": 8552
    },
    {
      "epoch": 0.3544697252269054,
      "grad_norm": 0.40542006492614746,
      "learning_rate": 3.2278585933938416e-06,
      "loss": 0.6514,
      "step": 8553
    },
    {
      "epoch": 0.3545111691325791,
      "grad_norm": 0.43535369634628296,
      "learning_rate": 3.2276513738654734e-06,
      "loss": 0.7267,
      "step": 8554
    },
    {
      "epoch": 0.35455261303825275,
      "grad_norm": 0.4038558006286621,
      "learning_rate": 3.227444154337105e-06,
      "loss": 0.6693,
      "step": 8555
    },
    {
      "epoch": 0.3545940569439264,
      "grad_norm": 0.4128554165363312,
      "learning_rate": 3.2272369348087366e-06,
      "loss": 0.6991,
      "step": 8556
    },
    {
      "epoch": 0.35463550084960005,
      "grad_norm": 0.41730573773384094,
      "learning_rate": 3.227029715280368e-06,
      "loss": 0.7166,
      "step": 8557
    },
    {
      "epoch": 0.3546769447552737,
      "grad_norm": 0.3999996781349182,
      "learning_rate": 3.2268224957520002e-06,
      "loss": 0.6047,
      "step": 8558
    },
    {
      "epoch": 0.3547183886609474,
      "grad_norm": 0.4316307604312897,
      "learning_rate": 3.226615276223631e-06,
      "loss": 0.7075,
      "step": 8559
    },
    {
      "epoch": 0.3547598325666211,
      "grad_norm": 0.4348883330821991,
      "learning_rate": 3.2264080566952634e-06,
      "loss": 0.6794,
      "step": 8560
    },
    {
      "epoch": 0.35480127647229476,
      "grad_norm": 0.3705008029937744,
      "learning_rate": 3.2262008371668952e-06,
      "loss": 0.6492,
      "step": 8561
    },
    {
      "epoch": 0.35484272037796843,
      "grad_norm": 0.4168749451637268,
      "learning_rate": 3.2259936176385266e-06,
      "loss": 0.7024,
      "step": 8562
    },
    {
      "epoch": 0.3548841642836421,
      "grad_norm": 0.4245136082172394,
      "learning_rate": 3.2257863981101584e-06,
      "loss": 0.7092,
      "step": 8563
    },
    {
      "epoch": 0.3549256081893158,
      "grad_norm": 0.4210855960845947,
      "learning_rate": 3.22557917858179e-06,
      "loss": 0.6637,
      "step": 8564
    },
    {
      "epoch": 0.3549670520949894,
      "grad_norm": 0.4014700949192047,
      "learning_rate": 3.2253719590534216e-06,
      "loss": 0.7012,
      "step": 8565
    },
    {
      "epoch": 0.3550084960006631,
      "grad_norm": 0.4253336191177368,
      "learning_rate": 3.225164739525053e-06,
      "loss": 0.7317,
      "step": 8566
    },
    {
      "epoch": 0.35504993990633676,
      "grad_norm": 0.4420667290687561,
      "learning_rate": 3.224957519996685e-06,
      "loss": 0.7163,
      "step": 8567
    },
    {
      "epoch": 0.35509138381201044,
      "grad_norm": 0.503119170665741,
      "learning_rate": 3.224750300468316e-06,
      "loss": 0.7354,
      "step": 8568
    },
    {
      "epoch": 0.3551328277176841,
      "grad_norm": 0.4166845977306366,
      "learning_rate": 3.224543080939948e-06,
      "loss": 0.7067,
      "step": 8569
    },
    {
      "epoch": 0.3551742716233578,
      "grad_norm": 0.4471542239189148,
      "learning_rate": 3.22433586141158e-06,
      "loss": 0.7554,
      "step": 8570
    },
    {
      "epoch": 0.3552157155290315,
      "grad_norm": 0.40410467982292175,
      "learning_rate": 3.224128641883211e-06,
      "loss": 0.697,
      "step": 8571
    },
    {
      "epoch": 0.35525715943470515,
      "grad_norm": 0.3820880055427551,
      "learning_rate": 3.223921422354843e-06,
      "loss": 0.7217,
      "step": 8572
    },
    {
      "epoch": 0.35529860334037877,
      "grad_norm": 0.40566423535346985,
      "learning_rate": 3.2237142028264744e-06,
      "loss": 0.7223,
      "step": 8573
    },
    {
      "epoch": 0.35534004724605245,
      "grad_norm": 0.4551563858985901,
      "learning_rate": 3.223506983298106e-06,
      "loss": 0.6537,
      "step": 8574
    },
    {
      "epoch": 0.3553814911517261,
      "grad_norm": 0.380825400352478,
      "learning_rate": 3.2232997637697376e-06,
      "loss": 0.6946,
      "step": 8575
    },
    {
      "epoch": 0.3554229350573998,
      "grad_norm": 0.43815889954566956,
      "learning_rate": 3.22309254424137e-06,
      "loss": 0.7397,
      "step": 8576
    },
    {
      "epoch": 0.3554643789630735,
      "grad_norm": 0.37048688530921936,
      "learning_rate": 3.222885324713001e-06,
      "loss": 0.665,
      "step": 8577
    },
    {
      "epoch": 0.35550582286874716,
      "grad_norm": 0.4148445725440979,
      "learning_rate": 3.222678105184633e-06,
      "loss": 0.7197,
      "step": 8578
    },
    {
      "epoch": 0.35554726677442083,
      "grad_norm": 0.3876015841960907,
      "learning_rate": 3.222470885656265e-06,
      "loss": 0.663,
      "step": 8579
    },
    {
      "epoch": 0.3555887106800945,
      "grad_norm": 0.4169434905052185,
      "learning_rate": 3.2222636661278962e-06,
      "loss": 0.6793,
      "step": 8580
    },
    {
      "epoch": 0.3556301545857682,
      "grad_norm": 0.4030241072177887,
      "learning_rate": 3.222056446599528e-06,
      "loss": 0.687,
      "step": 8581
    },
    {
      "epoch": 0.3556715984914418,
      "grad_norm": 0.4049224555492401,
      "learning_rate": 3.2218492270711594e-06,
      "loss": 0.7068,
      "step": 8582
    },
    {
      "epoch": 0.3557130423971155,
      "grad_norm": 0.40973329544067383,
      "learning_rate": 3.2216420075427912e-06,
      "loss": 0.6455,
      "step": 8583
    },
    {
      "epoch": 0.35575448630278916,
      "grad_norm": 0.4468746483325958,
      "learning_rate": 3.2214347880144226e-06,
      "loss": 0.6919,
      "step": 8584
    },
    {
      "epoch": 0.35579593020846284,
      "grad_norm": 0.4454256594181061,
      "learning_rate": 3.2212275684860544e-06,
      "loss": 0.7124,
      "step": 8585
    },
    {
      "epoch": 0.3558373741141365,
      "grad_norm": 0.42078959941864014,
      "learning_rate": 3.221020348957686e-06,
      "loss": 0.6946,
      "step": 8586
    },
    {
      "epoch": 0.3558788180198102,
      "grad_norm": 0.41023966670036316,
      "learning_rate": 3.2208131294293176e-06,
      "loss": 0.6919,
      "step": 8587
    },
    {
      "epoch": 0.35592026192548387,
      "grad_norm": 0.41594240069389343,
      "learning_rate": 3.2206059099009494e-06,
      "loss": 0.6804,
      "step": 8588
    },
    {
      "epoch": 0.35596170583115755,
      "grad_norm": 0.45461198687553406,
      "learning_rate": 3.220398690372581e-06,
      "loss": 0.718,
      "step": 8589
    },
    {
      "epoch": 0.3560031497368312,
      "grad_norm": 0.4051576256752014,
      "learning_rate": 3.2201914708442126e-06,
      "loss": 0.6826,
      "step": 8590
    },
    {
      "epoch": 0.35604459364250485,
      "grad_norm": 0.42591792345046997,
      "learning_rate": 3.219984251315844e-06,
      "loss": 0.7386,
      "step": 8591
    },
    {
      "epoch": 0.3560860375481785,
      "grad_norm": 0.4453498423099518,
      "learning_rate": 3.2197770317874762e-06,
      "loss": 0.6726,
      "step": 8592
    },
    {
      "epoch": 0.3561274814538522,
      "grad_norm": 0.42274242639541626,
      "learning_rate": 3.219569812259107e-06,
      "loss": 0.7446,
      "step": 8593
    },
    {
      "epoch": 0.3561689253595259,
      "grad_norm": 0.4164092242717743,
      "learning_rate": 3.2193625927307394e-06,
      "loss": 0.748,
      "step": 8594
    },
    {
      "epoch": 0.35621036926519956,
      "grad_norm": 0.4614359736442566,
      "learning_rate": 3.2191553732023704e-06,
      "loss": 0.6766,
      "step": 8595
    },
    {
      "epoch": 0.35625181317087323,
      "grad_norm": 0.386618435382843,
      "learning_rate": 3.2189481536740026e-06,
      "loss": 0.701,
      "step": 8596
    },
    {
      "epoch": 0.3562932570765469,
      "grad_norm": 0.43639281392097473,
      "learning_rate": 3.2187409341456344e-06,
      "loss": 0.719,
      "step": 8597
    },
    {
      "epoch": 0.3563347009822206,
      "grad_norm": 0.37950024008750916,
      "learning_rate": 3.218533714617266e-06,
      "loss": 0.6694,
      "step": 8598
    },
    {
      "epoch": 0.3563761448878942,
      "grad_norm": 0.412626713514328,
      "learning_rate": 3.2183264950888976e-06,
      "loss": 0.7018,
      "step": 8599
    },
    {
      "epoch": 0.3564175887935679,
      "grad_norm": 0.4036887288093567,
      "learning_rate": 3.218119275560529e-06,
      "loss": 0.6736,
      "step": 8600
    },
    {
      "epoch": 0.35645903269924156,
      "grad_norm": 0.46518674492836,
      "learning_rate": 3.217912056032161e-06,
      "loss": 0.7415,
      "step": 8601
    },
    {
      "epoch": 0.35650047660491524,
      "grad_norm": 0.4361576437950134,
      "learning_rate": 3.217704836503792e-06,
      "loss": 0.748,
      "step": 8602
    },
    {
      "epoch": 0.3565419205105889,
      "grad_norm": 0.3992605209350586,
      "learning_rate": 3.217497616975424e-06,
      "loss": 0.7231,
      "step": 8603
    },
    {
      "epoch": 0.3565833644162626,
      "grad_norm": 0.40660595893859863,
      "learning_rate": 3.2172903974470554e-06,
      "loss": 0.7028,
      "step": 8604
    },
    {
      "epoch": 0.35662480832193627,
      "grad_norm": 0.41146934032440186,
      "learning_rate": 3.2170831779186872e-06,
      "loss": 0.6912,
      "step": 8605
    },
    {
      "epoch": 0.35666625222760995,
      "grad_norm": 0.40343713760375977,
      "learning_rate": 3.216875958390319e-06,
      "loss": 0.7017,
      "step": 8606
    },
    {
      "epoch": 0.3567076961332836,
      "grad_norm": 0.399984747171402,
      "learning_rate": 3.2166687388619504e-06,
      "loss": 0.6969,
      "step": 8607
    },
    {
      "epoch": 0.35674914003895725,
      "grad_norm": 0.3853943347930908,
      "learning_rate": 3.2164615193335822e-06,
      "loss": 0.6526,
      "step": 8608
    },
    {
      "epoch": 0.3567905839446309,
      "grad_norm": 0.4272754490375519,
      "learning_rate": 3.2162542998052136e-06,
      "loss": 0.7002,
      "step": 8609
    },
    {
      "epoch": 0.3568320278503046,
      "grad_norm": 0.4493042826652527,
      "learning_rate": 3.216047080276846e-06,
      "loss": 0.7954,
      "step": 8610
    },
    {
      "epoch": 0.3568734717559783,
      "grad_norm": 0.43379130959510803,
      "learning_rate": 3.215839860748477e-06,
      "loss": 0.7742,
      "step": 8611
    },
    {
      "epoch": 0.35691491566165195,
      "grad_norm": 0.40505853295326233,
      "learning_rate": 3.215632641220109e-06,
      "loss": 0.6909,
      "step": 8612
    },
    {
      "epoch": 0.35695635956732563,
      "grad_norm": 0.3881370425224304,
      "learning_rate": 3.215425421691741e-06,
      "loss": 0.6785,
      "step": 8613
    },
    {
      "epoch": 0.3569978034729993,
      "grad_norm": 0.38113144040107727,
      "learning_rate": 3.2152182021633722e-06,
      "loss": 0.6445,
      "step": 8614
    },
    {
      "epoch": 0.357039247378673,
      "grad_norm": 0.4128870964050293,
      "learning_rate": 3.215010982635004e-06,
      "loss": 0.7576,
      "step": 8615
    },
    {
      "epoch": 0.35708069128434666,
      "grad_norm": 0.4382694363594055,
      "learning_rate": 3.2148037631066354e-06,
      "loss": 0.7031,
      "step": 8616
    },
    {
      "epoch": 0.3571221351900203,
      "grad_norm": 0.4418732821941376,
      "learning_rate": 3.2145965435782672e-06,
      "loss": 0.7463,
      "step": 8617
    },
    {
      "epoch": 0.35716357909569396,
      "grad_norm": 0.4117302894592285,
      "learning_rate": 3.2143893240498986e-06,
      "loss": 0.7251,
      "step": 8618
    },
    {
      "epoch": 0.35720502300136764,
      "grad_norm": 0.4213348627090454,
      "learning_rate": 3.2141821045215304e-06,
      "loss": 0.7144,
      "step": 8619
    },
    {
      "epoch": 0.3572464669070413,
      "grad_norm": 0.43918853998184204,
      "learning_rate": 3.213974884993162e-06,
      "loss": 0.7439,
      "step": 8620
    },
    {
      "epoch": 0.357287910812715,
      "grad_norm": 0.43314146995544434,
      "learning_rate": 3.2137676654647936e-06,
      "loss": 0.691,
      "step": 8621
    },
    {
      "epoch": 0.35732935471838867,
      "grad_norm": 0.3981378674507141,
      "learning_rate": 3.2135604459364254e-06,
      "loss": 0.7205,
      "step": 8622
    },
    {
      "epoch": 0.35737079862406235,
      "grad_norm": 0.40885764360427856,
      "learning_rate": 3.213353226408057e-06,
      "loss": 0.6951,
      "step": 8623
    },
    {
      "epoch": 0.357412242529736,
      "grad_norm": 0.4133684039115906,
      "learning_rate": 3.2131460068796886e-06,
      "loss": 0.6925,
      "step": 8624
    },
    {
      "epoch": 0.35745368643540965,
      "grad_norm": 0.45643529295921326,
      "learning_rate": 3.21293878735132e-06,
      "loss": 0.7441,
      "step": 8625
    },
    {
      "epoch": 0.3574951303410833,
      "grad_norm": 0.42970994114875793,
      "learning_rate": 3.2127315678229522e-06,
      "loss": 0.6926,
      "step": 8626
    },
    {
      "epoch": 0.357536574246757,
      "grad_norm": 0.3958750069141388,
      "learning_rate": 3.212524348294583e-06,
      "loss": 0.6519,
      "step": 8627
    },
    {
      "epoch": 0.3575780181524307,
      "grad_norm": 0.4271100163459778,
      "learning_rate": 3.2123171287662154e-06,
      "loss": 0.7026,
      "step": 8628
    },
    {
      "epoch": 0.35761946205810435,
      "grad_norm": 0.4285823106765747,
      "learning_rate": 3.2121099092378464e-06,
      "loss": 0.6956,
      "step": 8629
    },
    {
      "epoch": 0.35766090596377803,
      "grad_norm": 0.41392239928245544,
      "learning_rate": 3.2119026897094786e-06,
      "loss": 0.7029,
      "step": 8630
    },
    {
      "epoch": 0.3577023498694517,
      "grad_norm": 0.4328571856021881,
      "learning_rate": 3.2116954701811104e-06,
      "loss": 0.705,
      "step": 8631
    },
    {
      "epoch": 0.3577437937751254,
      "grad_norm": 0.40659859776496887,
      "learning_rate": 3.211488250652742e-06,
      "loss": 0.6848,
      "step": 8632
    },
    {
      "epoch": 0.35778523768079906,
      "grad_norm": 0.4014585614204407,
      "learning_rate": 3.2112810311243736e-06,
      "loss": 0.6593,
      "step": 8633
    },
    {
      "epoch": 0.3578266815864727,
      "grad_norm": 0.374635249376297,
      "learning_rate": 3.211073811596005e-06,
      "loss": 0.6284,
      "step": 8634
    },
    {
      "epoch": 0.35786812549214636,
      "grad_norm": 0.4336332082748413,
      "learning_rate": 3.210866592067637e-06,
      "loss": 0.7231,
      "step": 8635
    },
    {
      "epoch": 0.35790956939782004,
      "grad_norm": 0.39702433347702026,
      "learning_rate": 3.2106593725392682e-06,
      "loss": 0.6583,
      "step": 8636
    },
    {
      "epoch": 0.3579510133034937,
      "grad_norm": 0.4217739999294281,
      "learning_rate": 3.2104521530109e-06,
      "loss": 0.7622,
      "step": 8637
    },
    {
      "epoch": 0.3579924572091674,
      "grad_norm": 0.411190390586853,
      "learning_rate": 3.2102449334825314e-06,
      "loss": 0.718,
      "step": 8638
    },
    {
      "epoch": 0.35803390111484107,
      "grad_norm": 0.40480706095695496,
      "learning_rate": 3.2100377139541632e-06,
      "loss": 0.7252,
      "step": 8639
    },
    {
      "epoch": 0.35807534502051475,
      "grad_norm": 0.40707555413246155,
      "learning_rate": 3.209830494425795e-06,
      "loss": 0.678,
      "step": 8640
    },
    {
      "epoch": 0.3581167889261884,
      "grad_norm": 0.4020068049430847,
      "learning_rate": 3.2096232748974264e-06,
      "loss": 0.7224,
      "step": 8641
    },
    {
      "epoch": 0.3581582328318621,
      "grad_norm": 0.41944706439971924,
      "learning_rate": 3.2094160553690582e-06,
      "loss": 0.6989,
      "step": 8642
    },
    {
      "epoch": 0.3581996767375357,
      "grad_norm": 0.41154971718788147,
      "learning_rate": 3.2092088358406896e-06,
      "loss": 0.7004,
      "step": 8643
    },
    {
      "epoch": 0.3582411206432094,
      "grad_norm": 0.4606510102748871,
      "learning_rate": 3.209001616312322e-06,
      "loss": 0.7495,
      "step": 8644
    },
    {
      "epoch": 0.3582825645488831,
      "grad_norm": 0.44434985518455505,
      "learning_rate": 3.208794396783953e-06,
      "loss": 0.668,
      "step": 8645
    },
    {
      "epoch": 0.35832400845455675,
      "grad_norm": 0.4028402864933014,
      "learning_rate": 3.208587177255585e-06,
      "loss": 0.7634,
      "step": 8646
    },
    {
      "epoch": 0.35836545236023043,
      "grad_norm": 0.4163174629211426,
      "learning_rate": 3.2083799577272164e-06,
      "loss": 0.7288,
      "step": 8647
    },
    {
      "epoch": 0.3584068962659041,
      "grad_norm": 0.39894750714302063,
      "learning_rate": 3.2081727381988482e-06,
      "loss": 0.7269,
      "step": 8648
    },
    {
      "epoch": 0.3584483401715778,
      "grad_norm": 0.4147360622882843,
      "learning_rate": 3.20796551867048e-06,
      "loss": 0.708,
      "step": 8649
    },
    {
      "epoch": 0.35848978407725146,
      "grad_norm": 0.45082521438598633,
      "learning_rate": 3.2077582991421114e-06,
      "loss": 0.7551,
      "step": 8650
    },
    {
      "epoch": 0.35853122798292514,
      "grad_norm": 0.4069189727306366,
      "learning_rate": 3.2075510796137432e-06,
      "loss": 0.6669,
      "step": 8651
    },
    {
      "epoch": 0.35857267188859876,
      "grad_norm": 0.43804097175598145,
      "learning_rate": 3.2073438600853746e-06,
      "loss": 0.7057,
      "step": 8652
    },
    {
      "epoch": 0.35861411579427244,
      "grad_norm": 0.41555774211883545,
      "learning_rate": 3.2071366405570064e-06,
      "loss": 0.709,
      "step": 8653
    },
    {
      "epoch": 0.3586555596999461,
      "grad_norm": 0.4156549870967865,
      "learning_rate": 3.206929421028638e-06,
      "loss": 0.6732,
      "step": 8654
    },
    {
      "epoch": 0.3586970036056198,
      "grad_norm": 0.38702520728111267,
      "learning_rate": 3.2067222015002696e-06,
      "loss": 0.6552,
      "step": 8655
    },
    {
      "epoch": 0.35873844751129347,
      "grad_norm": 0.418766587972641,
      "learning_rate": 3.206514981971901e-06,
      "loss": 0.7734,
      "step": 8656
    },
    {
      "epoch": 0.35877989141696714,
      "grad_norm": 0.43494096398353577,
      "learning_rate": 3.206307762443533e-06,
      "loss": 0.7148,
      "step": 8657
    },
    {
      "epoch": 0.3588213353226408,
      "grad_norm": 0.3983537256717682,
      "learning_rate": 3.2061005429151646e-06,
      "loss": 0.6417,
      "step": 8658
    },
    {
      "epoch": 0.3588627792283145,
      "grad_norm": 0.4036007821559906,
      "learning_rate": 3.205893323386796e-06,
      "loss": 0.6461,
      "step": 8659
    },
    {
      "epoch": 0.3589042231339881,
      "grad_norm": 0.3809744417667389,
      "learning_rate": 3.2056861038584282e-06,
      "loss": 0.6978,
      "step": 8660
    },
    {
      "epoch": 0.3589456670396618,
      "grad_norm": 0.4268876016139984,
      "learning_rate": 3.205478884330059e-06,
      "loss": 0.7219,
      "step": 8661
    },
    {
      "epoch": 0.3589871109453355,
      "grad_norm": 0.4100329577922821,
      "learning_rate": 3.2052716648016914e-06,
      "loss": 0.7072,
      "step": 8662
    },
    {
      "epoch": 0.35902855485100915,
      "grad_norm": 0.42140743136405945,
      "learning_rate": 3.2050644452733224e-06,
      "loss": 0.7166,
      "step": 8663
    },
    {
      "epoch": 0.35906999875668283,
      "grad_norm": 0.4297003746032715,
      "learning_rate": 3.2048572257449546e-06,
      "loss": 0.7456,
      "step": 8664
    },
    {
      "epoch": 0.3591114426623565,
      "grad_norm": 0.39545580744743347,
      "learning_rate": 3.204650006216586e-06,
      "loss": 0.6914,
      "step": 8665
    },
    {
      "epoch": 0.3591528865680302,
      "grad_norm": 0.4164530336856842,
      "learning_rate": 3.204442786688218e-06,
      "loss": 0.6655,
      "step": 8666
    },
    {
      "epoch": 0.35919433047370386,
      "grad_norm": 0.40562406182289124,
      "learning_rate": 3.2042355671598496e-06,
      "loss": 0.687,
      "step": 8667
    },
    {
      "epoch": 0.35923577437937754,
      "grad_norm": 0.43582430481910706,
      "learning_rate": 3.204028347631481e-06,
      "loss": 0.7136,
      "step": 8668
    },
    {
      "epoch": 0.35927721828505116,
      "grad_norm": 0.4351360499858856,
      "learning_rate": 3.203821128103113e-06,
      "loss": 0.7544,
      "step": 8669
    },
    {
      "epoch": 0.35931866219072484,
      "grad_norm": 0.3984261453151703,
      "learning_rate": 3.2036139085747442e-06,
      "loss": 0.6794,
      "step": 8670
    },
    {
      "epoch": 0.3593601060963985,
      "grad_norm": 0.4076157510280609,
      "learning_rate": 3.203406689046376e-06,
      "loss": 0.718,
      "step": 8671
    },
    {
      "epoch": 0.3594015500020722,
      "grad_norm": 0.40390744805336,
      "learning_rate": 3.2031994695180074e-06,
      "loss": 0.6804,
      "step": 8672
    },
    {
      "epoch": 0.35944299390774587,
      "grad_norm": 0.39340588450431824,
      "learning_rate": 3.2029922499896392e-06,
      "loss": 0.696,
      "step": 8673
    },
    {
      "epoch": 0.35948443781341954,
      "grad_norm": 0.43475353717803955,
      "learning_rate": 3.202785030461271e-06,
      "loss": 0.6855,
      "step": 8674
    },
    {
      "epoch": 0.3595258817190932,
      "grad_norm": 0.48821792006492615,
      "learning_rate": 3.2025778109329024e-06,
      "loss": 0.7532,
      "step": 8675
    },
    {
      "epoch": 0.3595673256247669,
      "grad_norm": 0.4022676944732666,
      "learning_rate": 3.2023705914045342e-06,
      "loss": 0.6978,
      "step": 8676
    },
    {
      "epoch": 0.3596087695304406,
      "grad_norm": 0.37009915709495544,
      "learning_rate": 3.2021633718761656e-06,
      "loss": 0.7012,
      "step": 8677
    },
    {
      "epoch": 0.3596502134361142,
      "grad_norm": 0.4057188034057617,
      "learning_rate": 3.201956152347798e-06,
      "loss": 0.7129,
      "step": 8678
    },
    {
      "epoch": 0.3596916573417879,
      "grad_norm": 0.4646468758583069,
      "learning_rate": 3.201748932819429e-06,
      "loss": 0.7102,
      "step": 8679
    },
    {
      "epoch": 0.35973310124746155,
      "grad_norm": 0.4313954710960388,
      "learning_rate": 3.201541713291061e-06,
      "loss": 0.733,
      "step": 8680
    },
    {
      "epoch": 0.35977454515313523,
      "grad_norm": 0.4056831896305084,
      "learning_rate": 3.2013344937626924e-06,
      "loss": 0.6432,
      "step": 8681
    },
    {
      "epoch": 0.3598159890588089,
      "grad_norm": 0.3931620121002197,
      "learning_rate": 3.2011272742343242e-06,
      "loss": 0.6626,
      "step": 8682
    },
    {
      "epoch": 0.3598574329644826,
      "grad_norm": 0.41706568002700806,
      "learning_rate": 3.200920054705956e-06,
      "loss": 0.7039,
      "step": 8683
    },
    {
      "epoch": 0.35989887687015626,
      "grad_norm": 0.42725566029548645,
      "learning_rate": 3.2007128351775874e-06,
      "loss": 0.6671,
      "step": 8684
    },
    {
      "epoch": 0.35994032077582994,
      "grad_norm": 0.41366472840309143,
      "learning_rate": 3.2005056156492192e-06,
      "loss": 0.7019,
      "step": 8685
    },
    {
      "epoch": 0.35998176468150356,
      "grad_norm": 0.4153638780117035,
      "learning_rate": 3.2002983961208506e-06,
      "loss": 0.7222,
      "step": 8686
    },
    {
      "epoch": 0.36002320858717723,
      "grad_norm": 0.41646304726600647,
      "learning_rate": 3.2000911765924824e-06,
      "loss": 0.7517,
      "step": 8687
    },
    {
      "epoch": 0.3600646524928509,
      "grad_norm": 0.3975159227848053,
      "learning_rate": 3.199883957064114e-06,
      "loss": 0.7152,
      "step": 8688
    },
    {
      "epoch": 0.3601060963985246,
      "grad_norm": 0.41991302371025085,
      "learning_rate": 3.1996767375357456e-06,
      "loss": 0.7251,
      "step": 8689
    },
    {
      "epoch": 0.36014754030419827,
      "grad_norm": 0.44278642535209656,
      "learning_rate": 3.199469518007377e-06,
      "loss": 0.6681,
      "step": 8690
    },
    {
      "epoch": 0.36018898420987194,
      "grad_norm": 0.4114539325237274,
      "learning_rate": 3.199262298479009e-06,
      "loss": 0.6626,
      "step": 8691
    },
    {
      "epoch": 0.3602304281155456,
      "grad_norm": 0.41678017377853394,
      "learning_rate": 3.1990550789506406e-06,
      "loss": 0.6844,
      "step": 8692
    },
    {
      "epoch": 0.3602718720212193,
      "grad_norm": 0.42319753766059875,
      "learning_rate": 3.198847859422272e-06,
      "loss": 0.7671,
      "step": 8693
    },
    {
      "epoch": 0.360313315926893,
      "grad_norm": 0.4015193283557892,
      "learning_rate": 3.1986406398939043e-06,
      "loss": 0.7136,
      "step": 8694
    },
    {
      "epoch": 0.3603547598325666,
      "grad_norm": 0.39908382296562195,
      "learning_rate": 3.1984334203655352e-06,
      "loss": 0.7009,
      "step": 8695
    },
    {
      "epoch": 0.3603962037382403,
      "grad_norm": 0.3875944912433624,
      "learning_rate": 3.1982262008371674e-06,
      "loss": 0.6591,
      "step": 8696
    },
    {
      "epoch": 0.36043764764391395,
      "grad_norm": 0.4470199644565582,
      "learning_rate": 3.1980189813087984e-06,
      "loss": 0.74,
      "step": 8697
    },
    {
      "epoch": 0.3604790915495876,
      "grad_norm": 0.40394559502601624,
      "learning_rate": 3.1978117617804306e-06,
      "loss": 0.6561,
      "step": 8698
    },
    {
      "epoch": 0.3605205354552613,
      "grad_norm": 0.40429189801216125,
      "learning_rate": 3.197604542252062e-06,
      "loss": 0.7131,
      "step": 8699
    },
    {
      "epoch": 0.360561979360935,
      "grad_norm": 0.39882224798202515,
      "learning_rate": 3.197397322723694e-06,
      "loss": 0.6637,
      "step": 8700
    },
    {
      "epoch": 0.36060342326660866,
      "grad_norm": 0.38423794507980347,
      "learning_rate": 3.1971901031953256e-06,
      "loss": 0.6968,
      "step": 8701
    },
    {
      "epoch": 0.36064486717228234,
      "grad_norm": 0.39836516976356506,
      "learning_rate": 3.196982883666957e-06,
      "loss": 0.7012,
      "step": 8702
    },
    {
      "epoch": 0.360686311077956,
      "grad_norm": 0.4412648677825928,
      "learning_rate": 3.196775664138589e-06,
      "loss": 0.7405,
      "step": 8703
    },
    {
      "epoch": 0.36072775498362963,
      "grad_norm": 0.4252113699913025,
      "learning_rate": 3.1965684446102202e-06,
      "loss": 0.7109,
      "step": 8704
    },
    {
      "epoch": 0.3607691988893033,
      "grad_norm": 0.459979772567749,
      "learning_rate": 3.196361225081852e-06,
      "loss": 0.699,
      "step": 8705
    },
    {
      "epoch": 0.360810642794977,
      "grad_norm": 0.39731279015541077,
      "learning_rate": 3.1961540055534834e-06,
      "loss": 0.7068,
      "step": 8706
    },
    {
      "epoch": 0.36085208670065066,
      "grad_norm": 0.41199877858161926,
      "learning_rate": 3.1959467860251152e-06,
      "loss": 0.6902,
      "step": 8707
    },
    {
      "epoch": 0.36089353060632434,
      "grad_norm": 0.3849584460258484,
      "learning_rate": 3.1957395664967466e-06,
      "loss": 0.65,
      "step": 8708
    },
    {
      "epoch": 0.360934974511998,
      "grad_norm": 0.3979337513446808,
      "learning_rate": 3.1955323469683784e-06,
      "loss": 0.6581,
      "step": 8709
    },
    {
      "epoch": 0.3609764184176717,
      "grad_norm": 0.4135061800479889,
      "learning_rate": 3.1953251274400102e-06,
      "loss": 0.6871,
      "step": 8710
    },
    {
      "epoch": 0.3610178623233454,
      "grad_norm": 0.3670703172683716,
      "learning_rate": 3.1951179079116416e-06,
      "loss": 0.6415,
      "step": 8711
    },
    {
      "epoch": 0.36105930622901905,
      "grad_norm": 0.4223484694957733,
      "learning_rate": 3.194910688383274e-06,
      "loss": 0.6906,
      "step": 8712
    },
    {
      "epoch": 0.36110075013469267,
      "grad_norm": 0.4482652246952057,
      "learning_rate": 3.194703468854905e-06,
      "loss": 0.6763,
      "step": 8713
    },
    {
      "epoch": 0.36114219404036635,
      "grad_norm": 0.4217328727245331,
      "learning_rate": 3.194496249326537e-06,
      "loss": 0.7723,
      "step": 8714
    },
    {
      "epoch": 0.36118363794604,
      "grad_norm": 0.3905523717403412,
      "learning_rate": 3.1942890297981684e-06,
      "loss": 0.6614,
      "step": 8715
    },
    {
      "epoch": 0.3612250818517137,
      "grad_norm": 0.3970358967781067,
      "learning_rate": 3.1940818102698002e-06,
      "loss": 0.7126,
      "step": 8716
    },
    {
      "epoch": 0.3612665257573874,
      "grad_norm": 0.3950396776199341,
      "learning_rate": 3.1938745907414316e-06,
      "loss": 0.6947,
      "step": 8717
    },
    {
      "epoch": 0.36130796966306106,
      "grad_norm": 0.4351550340652466,
      "learning_rate": 3.1936673712130634e-06,
      "loss": 0.6982,
      "step": 8718
    },
    {
      "epoch": 0.36134941356873473,
      "grad_norm": 0.4230193793773651,
      "learning_rate": 3.1934601516846952e-06,
      "loss": 0.6946,
      "step": 8719
    },
    {
      "epoch": 0.3613908574744084,
      "grad_norm": 0.4132295846939087,
      "learning_rate": 3.1932529321563266e-06,
      "loss": 0.6943,
      "step": 8720
    },
    {
      "epoch": 0.36143230138008203,
      "grad_norm": 0.3997012674808502,
      "learning_rate": 3.1930457126279584e-06,
      "loss": 0.6906,
      "step": 8721
    },
    {
      "epoch": 0.3614737452857557,
      "grad_norm": 0.4293724596500397,
      "learning_rate": 3.19283849309959e-06,
      "loss": 0.6619,
      "step": 8722
    },
    {
      "epoch": 0.3615151891914294,
      "grad_norm": 0.41585633158683777,
      "learning_rate": 3.1926312735712216e-06,
      "loss": 0.6763,
      "step": 8723
    },
    {
      "epoch": 0.36155663309710306,
      "grad_norm": 0.3797011375427246,
      "learning_rate": 3.192424054042853e-06,
      "loss": 0.6694,
      "step": 8724
    },
    {
      "epoch": 0.36159807700277674,
      "grad_norm": 0.4137294888496399,
      "learning_rate": 3.192216834514485e-06,
      "loss": 0.6443,
      "step": 8725
    },
    {
      "epoch": 0.3616395209084504,
      "grad_norm": 0.38571393489837646,
      "learning_rate": 3.1920096149861162e-06,
      "loss": 0.6699,
      "step": 8726
    },
    {
      "epoch": 0.3616809648141241,
      "grad_norm": 0.46915021538734436,
      "learning_rate": 3.191802395457748e-06,
      "loss": 0.7356,
      "step": 8727
    },
    {
      "epoch": 0.36172240871979777,
      "grad_norm": 0.42679134011268616,
      "learning_rate": 3.1915951759293803e-06,
      "loss": 0.6858,
      "step": 8728
    },
    {
      "epoch": 0.36176385262547145,
      "grad_norm": 0.44958364963531494,
      "learning_rate": 3.1913879564010112e-06,
      "loss": 0.6925,
      "step": 8729
    },
    {
      "epoch": 0.36180529653114507,
      "grad_norm": 0.47745299339294434,
      "learning_rate": 3.1911807368726435e-06,
      "loss": 0.76,
      "step": 8730
    },
    {
      "epoch": 0.36184674043681875,
      "grad_norm": 0.4011189341545105,
      "learning_rate": 3.190973517344275e-06,
      "loss": 0.6897,
      "step": 8731
    },
    {
      "epoch": 0.3618881843424924,
      "grad_norm": 0.44328612089157104,
      "learning_rate": 3.1907662978159066e-06,
      "loss": 0.7223,
      "step": 8732
    },
    {
      "epoch": 0.3619296282481661,
      "grad_norm": 0.4299856126308441,
      "learning_rate": 3.190559078287538e-06,
      "loss": 0.679,
      "step": 8733
    },
    {
      "epoch": 0.3619710721538398,
      "grad_norm": 0.4424896538257599,
      "learning_rate": 3.19035185875917e-06,
      "loss": 0.714,
      "step": 8734
    },
    {
      "epoch": 0.36201251605951346,
      "grad_norm": 0.4129107892513275,
      "learning_rate": 3.1901446392308017e-06,
      "loss": 0.7917,
      "step": 8735
    },
    {
      "epoch": 0.36205395996518713,
      "grad_norm": 0.3794506788253784,
      "learning_rate": 3.189937419702433e-06,
      "loss": 0.684,
      "step": 8736
    },
    {
      "epoch": 0.3620954038708608,
      "grad_norm": 0.38367196917533875,
      "learning_rate": 3.189730200174065e-06,
      "loss": 0.6549,
      "step": 8737
    },
    {
      "epoch": 0.3621368477765345,
      "grad_norm": 0.38777387142181396,
      "learning_rate": 3.1895229806456962e-06,
      "loss": 0.6519,
      "step": 8738
    },
    {
      "epoch": 0.3621782916822081,
      "grad_norm": 0.39142030477523804,
      "learning_rate": 3.189315761117328e-06,
      "loss": 0.7188,
      "step": 8739
    },
    {
      "epoch": 0.3622197355878818,
      "grad_norm": 0.38989362120628357,
      "learning_rate": 3.1891085415889594e-06,
      "loss": 0.6436,
      "step": 8740
    },
    {
      "epoch": 0.36226117949355546,
      "grad_norm": 0.3616463243961334,
      "learning_rate": 3.1889013220605912e-06,
      "loss": 0.6604,
      "step": 8741
    },
    {
      "epoch": 0.36230262339922914,
      "grad_norm": 0.3997287452220917,
      "learning_rate": 3.1886941025322226e-06,
      "loss": 0.6497,
      "step": 8742
    },
    {
      "epoch": 0.3623440673049028,
      "grad_norm": 0.43689417839050293,
      "learning_rate": 3.1884868830038544e-06,
      "loss": 0.7478,
      "step": 8743
    },
    {
      "epoch": 0.3623855112105765,
      "grad_norm": 0.3861250579357147,
      "learning_rate": 3.1882796634754867e-06,
      "loss": 0.728,
      "step": 8744
    },
    {
      "epoch": 0.36242695511625017,
      "grad_norm": 0.4277266263961792,
      "learning_rate": 3.1880724439471176e-06,
      "loss": 0.7107,
      "step": 8745
    },
    {
      "epoch": 0.36246839902192385,
      "grad_norm": 0.41886186599731445,
      "learning_rate": 3.18786522441875e-06,
      "loss": 0.6541,
      "step": 8746
    },
    {
      "epoch": 0.36250984292759747,
      "grad_norm": 0.40659788250923157,
      "learning_rate": 3.187658004890381e-06,
      "loss": 0.6239,
      "step": 8747
    },
    {
      "epoch": 0.36255128683327115,
      "grad_norm": 0.4191431701183319,
      "learning_rate": 3.187450785362013e-06,
      "loss": 0.73,
      "step": 8748
    },
    {
      "epoch": 0.3625927307389448,
      "grad_norm": 0.38583990931510925,
      "learning_rate": 3.1872435658336444e-06,
      "loss": 0.6482,
      "step": 8749
    },
    {
      "epoch": 0.3626341746446185,
      "grad_norm": 0.4314139187335968,
      "learning_rate": 3.1870363463052762e-06,
      "loss": 0.7511,
      "step": 8750
    },
    {
      "epoch": 0.3626756185502922,
      "grad_norm": 0.46310287714004517,
      "learning_rate": 3.1868291267769076e-06,
      "loss": 0.7014,
      "step": 8751
    },
    {
      "epoch": 0.36271706245596586,
      "grad_norm": 0.39220088720321655,
      "learning_rate": 3.1866219072485394e-06,
      "loss": 0.701,
      "step": 8752
    },
    {
      "epoch": 0.36275850636163953,
      "grad_norm": 0.3864094018936157,
      "learning_rate": 3.1864146877201713e-06,
      "loss": 0.7006,
      "step": 8753
    },
    {
      "epoch": 0.3627999502673132,
      "grad_norm": 0.4193985164165497,
      "learning_rate": 3.1862074681918026e-06,
      "loss": 0.6805,
      "step": 8754
    },
    {
      "epoch": 0.3628413941729869,
      "grad_norm": 0.400595098733902,
      "learning_rate": 3.1860002486634344e-06,
      "loss": 0.7012,
      "step": 8755
    },
    {
      "epoch": 0.3628828380786605,
      "grad_norm": 0.3874930143356323,
      "learning_rate": 3.185793029135066e-06,
      "loss": 0.687,
      "step": 8756
    },
    {
      "epoch": 0.3629242819843342,
      "grad_norm": 0.4066002368927002,
      "learning_rate": 3.1855858096066976e-06,
      "loss": 0.6914,
      "step": 8757
    },
    {
      "epoch": 0.36296572589000786,
      "grad_norm": 0.43422219157218933,
      "learning_rate": 3.185378590078329e-06,
      "loss": 0.7583,
      "step": 8758
    },
    {
      "epoch": 0.36300716979568154,
      "grad_norm": 0.408501535654068,
      "learning_rate": 3.185171370549961e-06,
      "loss": 0.7239,
      "step": 8759
    },
    {
      "epoch": 0.3630486137013552,
      "grad_norm": 0.39842933416366577,
      "learning_rate": 3.1849641510215922e-06,
      "loss": 0.7192,
      "step": 8760
    },
    {
      "epoch": 0.3630900576070289,
      "grad_norm": 0.38772615790367126,
      "learning_rate": 3.184756931493224e-06,
      "loss": 0.6636,
      "step": 8761
    },
    {
      "epoch": 0.36313150151270257,
      "grad_norm": 0.4501410722732544,
      "learning_rate": 3.1845497119648563e-06,
      "loss": 0.7351,
      "step": 8762
    },
    {
      "epoch": 0.36317294541837625,
      "grad_norm": 0.41095712780952454,
      "learning_rate": 3.1843424924364872e-06,
      "loss": 0.6346,
      "step": 8763
    },
    {
      "epoch": 0.3632143893240499,
      "grad_norm": 0.43201744556427,
      "learning_rate": 3.1841352729081195e-06,
      "loss": 0.7205,
      "step": 8764
    },
    {
      "epoch": 0.36325583322972355,
      "grad_norm": 0.4419722259044647,
      "learning_rate": 3.183928053379751e-06,
      "loss": 0.7068,
      "step": 8765
    },
    {
      "epoch": 0.3632972771353972,
      "grad_norm": 0.3976384699344635,
      "learning_rate": 3.1837208338513827e-06,
      "loss": 0.6809,
      "step": 8766
    },
    {
      "epoch": 0.3633387210410709,
      "grad_norm": 0.4190293848514557,
      "learning_rate": 3.183513614323014e-06,
      "loss": 0.6963,
      "step": 8767
    },
    {
      "epoch": 0.3633801649467446,
      "grad_norm": 0.4072466194629669,
      "learning_rate": 3.183306394794646e-06,
      "loss": 0.6639,
      "step": 8768
    },
    {
      "epoch": 0.36342160885241825,
      "grad_norm": 0.44544410705566406,
      "learning_rate": 3.1830991752662772e-06,
      "loss": 0.733,
      "step": 8769
    },
    {
      "epoch": 0.36346305275809193,
      "grad_norm": 0.4002475142478943,
      "learning_rate": 3.182891955737909e-06,
      "loss": 0.708,
      "step": 8770
    },
    {
      "epoch": 0.3635044966637656,
      "grad_norm": 0.4335148334503174,
      "learning_rate": 3.182684736209541e-06,
      "loss": 0.6892,
      "step": 8771
    },
    {
      "epoch": 0.3635459405694393,
      "grad_norm": 0.4222176671028137,
      "learning_rate": 3.1824775166811722e-06,
      "loss": 0.7067,
      "step": 8772
    },
    {
      "epoch": 0.3635873844751129,
      "grad_norm": 0.3817721903324127,
      "learning_rate": 3.182270297152804e-06,
      "loss": 0.6665,
      "step": 8773
    },
    {
      "epoch": 0.3636288283807866,
      "grad_norm": 0.3962269723415375,
      "learning_rate": 3.1820630776244354e-06,
      "loss": 0.6609,
      "step": 8774
    },
    {
      "epoch": 0.36367027228646026,
      "grad_norm": 0.43797460198402405,
      "learning_rate": 3.1818558580960672e-06,
      "loss": 0.6808,
      "step": 8775
    },
    {
      "epoch": 0.36371171619213394,
      "grad_norm": 0.4102444052696228,
      "learning_rate": 3.1816486385676986e-06,
      "loss": 0.6425,
      "step": 8776
    },
    {
      "epoch": 0.3637531600978076,
      "grad_norm": 0.42108339071273804,
      "learning_rate": 3.1814414190393304e-06,
      "loss": 0.7212,
      "step": 8777
    },
    {
      "epoch": 0.3637946040034813,
      "grad_norm": 0.4083385467529297,
      "learning_rate": 3.181234199510962e-06,
      "loss": 0.6426,
      "step": 8778
    },
    {
      "epoch": 0.36383604790915497,
      "grad_norm": 0.4140072166919708,
      "learning_rate": 3.1810269799825936e-06,
      "loss": 0.7356,
      "step": 8779
    },
    {
      "epoch": 0.36387749181482865,
      "grad_norm": 0.43164199590682983,
      "learning_rate": 3.180819760454226e-06,
      "loss": 0.6614,
      "step": 8780
    },
    {
      "epoch": 0.3639189357205023,
      "grad_norm": 0.4276571273803711,
      "learning_rate": 3.180612540925857e-06,
      "loss": 0.686,
      "step": 8781
    },
    {
      "epoch": 0.36396037962617594,
      "grad_norm": 0.4281634986400604,
      "learning_rate": 3.180405321397489e-06,
      "loss": 0.7313,
      "step": 8782
    },
    {
      "epoch": 0.3640018235318496,
      "grad_norm": 0.4151059687137604,
      "learning_rate": 3.1801981018691204e-06,
      "loss": 0.7006,
      "step": 8783
    },
    {
      "epoch": 0.3640432674375233,
      "grad_norm": 0.42673927545547485,
      "learning_rate": 3.1799908823407523e-06,
      "loss": 0.7039,
      "step": 8784
    },
    {
      "epoch": 0.364084711343197,
      "grad_norm": 0.41177552938461304,
      "learning_rate": 3.1797836628123836e-06,
      "loss": 0.6797,
      "step": 8785
    },
    {
      "epoch": 0.36412615524887065,
      "grad_norm": 0.38462021946907043,
      "learning_rate": 3.1795764432840154e-06,
      "loss": 0.6368,
      "step": 8786
    },
    {
      "epoch": 0.36416759915454433,
      "grad_norm": 0.39418941736221313,
      "learning_rate": 3.179369223755647e-06,
      "loss": 0.6567,
      "step": 8787
    },
    {
      "epoch": 0.364209043060218,
      "grad_norm": 0.46565166115760803,
      "learning_rate": 3.1791620042272786e-06,
      "loss": 0.7227,
      "step": 8788
    },
    {
      "epoch": 0.3642504869658917,
      "grad_norm": 0.3941397964954376,
      "learning_rate": 3.1789547846989105e-06,
      "loss": 0.6953,
      "step": 8789
    },
    {
      "epoch": 0.36429193087156536,
      "grad_norm": 0.39258432388305664,
      "learning_rate": 3.178747565170542e-06,
      "loss": 0.677,
      "step": 8790
    },
    {
      "epoch": 0.364333374777239,
      "grad_norm": 0.39421311020851135,
      "learning_rate": 3.1785403456421736e-06,
      "loss": 0.7083,
      "step": 8791
    },
    {
      "epoch": 0.36437481868291266,
      "grad_norm": 0.4141696095466614,
      "learning_rate": 3.178333126113805e-06,
      "loss": 0.7153,
      "step": 8792
    },
    {
      "epoch": 0.36441626258858634,
      "grad_norm": 0.42644450068473816,
      "learning_rate": 3.178125906585437e-06,
      "loss": 0.6473,
      "step": 8793
    },
    {
      "epoch": 0.36445770649426,
      "grad_norm": 0.4411250352859497,
      "learning_rate": 3.1779186870570682e-06,
      "loss": 0.6887,
      "step": 8794
    },
    {
      "epoch": 0.3644991503999337,
      "grad_norm": 0.3941774070262909,
      "learning_rate": 3.1777114675287e-06,
      "loss": 0.6901,
      "step": 8795
    },
    {
      "epoch": 0.36454059430560737,
      "grad_norm": 0.39575690031051636,
      "learning_rate": 3.1775042480003323e-06,
      "loss": 0.667,
      "step": 8796
    },
    {
      "epoch": 0.36458203821128105,
      "grad_norm": 0.4060383141040802,
      "learning_rate": 3.1772970284719632e-06,
      "loss": 0.7465,
      "step": 8797
    },
    {
      "epoch": 0.3646234821169547,
      "grad_norm": 0.4219093918800354,
      "learning_rate": 3.1770898089435955e-06,
      "loss": 0.7051,
      "step": 8798
    },
    {
      "epoch": 0.3646649260226284,
      "grad_norm": 0.3760469853878021,
      "learning_rate": 3.176882589415227e-06,
      "loss": 0.7008,
      "step": 8799
    },
    {
      "epoch": 0.364706369928302,
      "grad_norm": 0.40530356764793396,
      "learning_rate": 3.1766753698868587e-06,
      "loss": 0.6921,
      "step": 8800
    },
    {
      "epoch": 0.3647478138339757,
      "grad_norm": 0.42373690009117126,
      "learning_rate": 3.17646815035849e-06,
      "loss": 0.7507,
      "step": 8801
    },
    {
      "epoch": 0.3647892577396494,
      "grad_norm": 0.3819704055786133,
      "learning_rate": 3.176260930830122e-06,
      "loss": 0.6948,
      "step": 8802
    },
    {
      "epoch": 0.36483070164532305,
      "grad_norm": 0.43700453639030457,
      "learning_rate": 3.1760537113017532e-06,
      "loss": 0.6687,
      "step": 8803
    },
    {
      "epoch": 0.36487214555099673,
      "grad_norm": 0.4160009026527405,
      "learning_rate": 3.175846491773385e-06,
      "loss": 0.6682,
      "step": 8804
    },
    {
      "epoch": 0.3649135894566704,
      "grad_norm": 0.42203670740127563,
      "learning_rate": 3.175639272245017e-06,
      "loss": 0.7375,
      "step": 8805
    },
    {
      "epoch": 0.3649550333623441,
      "grad_norm": 0.39568477869033813,
      "learning_rate": 3.1754320527166482e-06,
      "loss": 0.6873,
      "step": 8806
    },
    {
      "epoch": 0.36499647726801776,
      "grad_norm": 0.4489230513572693,
      "learning_rate": 3.17522483318828e-06,
      "loss": 0.7803,
      "step": 8807
    },
    {
      "epoch": 0.3650379211736914,
      "grad_norm": 0.41700926423072815,
      "learning_rate": 3.1750176136599114e-06,
      "loss": 0.7126,
      "step": 8808
    },
    {
      "epoch": 0.36507936507936506,
      "grad_norm": 0.41486063599586487,
      "learning_rate": 3.1748103941315432e-06,
      "loss": 0.6799,
      "step": 8809
    },
    {
      "epoch": 0.36512080898503874,
      "grad_norm": 0.4383470118045807,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 0.7456,
      "step": 8810
    },
    {
      "epoch": 0.3651622528907124,
      "grad_norm": 0.43047183752059937,
      "learning_rate": 3.1743959550748064e-06,
      "loss": 0.689,
      "step": 8811
    },
    {
      "epoch": 0.3652036967963861,
      "grad_norm": 0.4064778983592987,
      "learning_rate": 3.174188735546438e-06,
      "loss": 0.7117,
      "step": 8812
    },
    {
      "epoch": 0.36524514070205977,
      "grad_norm": 0.4036770462989807,
      "learning_rate": 3.1739815160180696e-06,
      "loss": 0.6799,
      "step": 8813
    },
    {
      "epoch": 0.36528658460773344,
      "grad_norm": 0.4272207021713257,
      "learning_rate": 3.173774296489702e-06,
      "loss": 0.743,
      "step": 8814
    },
    {
      "epoch": 0.3653280285134071,
      "grad_norm": 0.4056776165962219,
      "learning_rate": 3.173567076961333e-06,
      "loss": 0.7356,
      "step": 8815
    },
    {
      "epoch": 0.3653694724190808,
      "grad_norm": 0.3883136212825775,
      "learning_rate": 3.173359857432965e-06,
      "loss": 0.684,
      "step": 8816
    },
    {
      "epoch": 0.3654109163247544,
      "grad_norm": 0.38638612627983093,
      "learning_rate": 3.1731526379045965e-06,
      "loss": 0.6984,
      "step": 8817
    },
    {
      "epoch": 0.3654523602304281,
      "grad_norm": 0.4543929100036621,
      "learning_rate": 3.1729454183762283e-06,
      "loss": 0.8047,
      "step": 8818
    },
    {
      "epoch": 0.3654938041361018,
      "grad_norm": 0.41637372970581055,
      "learning_rate": 3.1727381988478596e-06,
      "loss": 0.658,
      "step": 8819
    },
    {
      "epoch": 0.36553524804177545,
      "grad_norm": 0.4713640511035919,
      "learning_rate": 3.1725309793194915e-06,
      "loss": 0.754,
      "step": 8820
    },
    {
      "epoch": 0.36557669194744913,
      "grad_norm": 0.42243054509162903,
      "learning_rate": 3.172323759791123e-06,
      "loss": 0.7238,
      "step": 8821
    },
    {
      "epoch": 0.3656181358531228,
      "grad_norm": 0.42175623774528503,
      "learning_rate": 3.1721165402627547e-06,
      "loss": 0.6578,
      "step": 8822
    },
    {
      "epoch": 0.3656595797587965,
      "grad_norm": 0.44663870334625244,
      "learning_rate": 3.1719093207343865e-06,
      "loss": 0.7307,
      "step": 8823
    },
    {
      "epoch": 0.36570102366447016,
      "grad_norm": 0.40727609395980835,
      "learning_rate": 3.171702101206018e-06,
      "loss": 0.6915,
      "step": 8824
    },
    {
      "epoch": 0.36574246757014384,
      "grad_norm": 0.4353761076927185,
      "learning_rate": 3.1714948816776497e-06,
      "loss": 0.7288,
      "step": 8825
    },
    {
      "epoch": 0.36578391147581746,
      "grad_norm": 0.42904141545295715,
      "learning_rate": 3.171287662149281e-06,
      "loss": 0.6238,
      "step": 8826
    },
    {
      "epoch": 0.36582535538149114,
      "grad_norm": 0.4006114602088928,
      "learning_rate": 3.171080442620913e-06,
      "loss": 0.722,
      "step": 8827
    },
    {
      "epoch": 0.3658667992871648,
      "grad_norm": 0.39533430337905884,
      "learning_rate": 3.1708732230925442e-06,
      "loss": 0.7078,
      "step": 8828
    },
    {
      "epoch": 0.3659082431928385,
      "grad_norm": 0.3962975740432739,
      "learning_rate": 3.170666003564176e-06,
      "loss": 0.6937,
      "step": 8829
    },
    {
      "epoch": 0.36594968709851217,
      "grad_norm": 0.4026281237602234,
      "learning_rate": 3.1704587840358074e-06,
      "loss": 0.6426,
      "step": 8830
    },
    {
      "epoch": 0.36599113100418584,
      "grad_norm": 0.39985892176628113,
      "learning_rate": 3.1702515645074392e-06,
      "loss": 0.6353,
      "step": 8831
    },
    {
      "epoch": 0.3660325749098595,
      "grad_norm": 0.40309709310531616,
      "learning_rate": 3.1700443449790715e-06,
      "loss": 0.6909,
      "step": 8832
    },
    {
      "epoch": 0.3660740188155332,
      "grad_norm": 0.38339272141456604,
      "learning_rate": 3.169837125450703e-06,
      "loss": 0.7322,
      "step": 8833
    },
    {
      "epoch": 0.3661154627212068,
      "grad_norm": 0.44847530126571655,
      "learning_rate": 3.1696299059223347e-06,
      "loss": 0.6941,
      "step": 8834
    },
    {
      "epoch": 0.3661569066268805,
      "grad_norm": 0.4052031636238098,
      "learning_rate": 3.169422686393966e-06,
      "loss": 0.6294,
      "step": 8835
    },
    {
      "epoch": 0.3661983505325542,
      "grad_norm": 0.40281644463539124,
      "learning_rate": 3.169215466865598e-06,
      "loss": 0.7249,
      "step": 8836
    },
    {
      "epoch": 0.36623979443822785,
      "grad_norm": 0.39112725853919983,
      "learning_rate": 3.1690082473372292e-06,
      "loss": 0.7258,
      "step": 8837
    },
    {
      "epoch": 0.3662812383439015,
      "grad_norm": 0.4365958869457245,
      "learning_rate": 3.168801027808861e-06,
      "loss": 0.7205,
      "step": 8838
    },
    {
      "epoch": 0.3663226822495752,
      "grad_norm": 0.40034088492393494,
      "learning_rate": 3.1685938082804924e-06,
      "loss": 0.7083,
      "step": 8839
    },
    {
      "epoch": 0.3663641261552489,
      "grad_norm": 0.440376877784729,
      "learning_rate": 3.1683865887521243e-06,
      "loss": 0.6729,
      "step": 8840
    },
    {
      "epoch": 0.36640557006092256,
      "grad_norm": 0.4406425952911377,
      "learning_rate": 3.168179369223756e-06,
      "loss": 0.6744,
      "step": 8841
    },
    {
      "epoch": 0.36644701396659624,
      "grad_norm": 0.3999883234500885,
      "learning_rate": 3.1679721496953874e-06,
      "loss": 0.6677,
      "step": 8842
    },
    {
      "epoch": 0.36648845787226986,
      "grad_norm": 0.43860435485839844,
      "learning_rate": 3.1677649301670193e-06,
      "loss": 0.7212,
      "step": 8843
    },
    {
      "epoch": 0.36652990177794353,
      "grad_norm": 0.43907806277275085,
      "learning_rate": 3.1675577106386506e-06,
      "loss": 0.6941,
      "step": 8844
    },
    {
      "epoch": 0.3665713456836172,
      "grad_norm": 0.4345446825027466,
      "learning_rate": 3.1673504911102824e-06,
      "loss": 0.7002,
      "step": 8845
    },
    {
      "epoch": 0.3666127895892909,
      "grad_norm": 0.4227834641933441,
      "learning_rate": 3.167143271581914e-06,
      "loss": 0.7002,
      "step": 8846
    },
    {
      "epoch": 0.36665423349496457,
      "grad_norm": 0.3927697539329529,
      "learning_rate": 3.1669360520535456e-06,
      "loss": 0.668,
      "step": 8847
    },
    {
      "epoch": 0.36669567740063824,
      "grad_norm": 0.4078225791454315,
      "learning_rate": 3.166728832525177e-06,
      "loss": 0.7678,
      "step": 8848
    },
    {
      "epoch": 0.3667371213063119,
      "grad_norm": 0.4158647060394287,
      "learning_rate": 3.166521612996809e-06,
      "loss": 0.741,
      "step": 8849
    },
    {
      "epoch": 0.3667785652119856,
      "grad_norm": 0.37287968397140503,
      "learning_rate": 3.166314393468441e-06,
      "loss": 0.6517,
      "step": 8850
    },
    {
      "epoch": 0.3668200091176593,
      "grad_norm": 0.39171770215034485,
      "learning_rate": 3.1661071739400725e-06,
      "loss": 0.6335,
      "step": 8851
    },
    {
      "epoch": 0.3668614530233329,
      "grad_norm": 0.45891904830932617,
      "learning_rate": 3.1658999544117043e-06,
      "loss": 0.7556,
      "step": 8852
    },
    {
      "epoch": 0.3669028969290066,
      "grad_norm": 0.42991599440574646,
      "learning_rate": 3.1656927348833357e-06,
      "loss": 0.6675,
      "step": 8853
    },
    {
      "epoch": 0.36694434083468025,
      "grad_norm": 0.41664254665374756,
      "learning_rate": 3.1654855153549675e-06,
      "loss": 0.7004,
      "step": 8854
    },
    {
      "epoch": 0.3669857847403539,
      "grad_norm": 0.43574804067611694,
      "learning_rate": 3.165278295826599e-06,
      "loss": 0.7046,
      "step": 8855
    },
    {
      "epoch": 0.3670272286460276,
      "grad_norm": 0.41894927620887756,
      "learning_rate": 3.1650710762982307e-06,
      "loss": 0.7124,
      "step": 8856
    },
    {
      "epoch": 0.3670686725517013,
      "grad_norm": 0.3938758075237274,
      "learning_rate": 3.1648638567698625e-06,
      "loss": 0.688,
      "step": 8857
    },
    {
      "epoch": 0.36711011645737496,
      "grad_norm": 0.4141032099723816,
      "learning_rate": 3.164656637241494e-06,
      "loss": 0.7043,
      "step": 8858
    },
    {
      "epoch": 0.36715156036304863,
      "grad_norm": 0.4135150611400604,
      "learning_rate": 3.1644494177131257e-06,
      "loss": 0.7212,
      "step": 8859
    },
    {
      "epoch": 0.3671930042687223,
      "grad_norm": 0.4092787504196167,
      "learning_rate": 3.164242198184757e-06,
      "loss": 0.6729,
      "step": 8860
    },
    {
      "epoch": 0.36723444817439593,
      "grad_norm": 0.4133279323577881,
      "learning_rate": 3.164034978656389e-06,
      "loss": 0.7372,
      "step": 8861
    },
    {
      "epoch": 0.3672758920800696,
      "grad_norm": 0.38868510723114014,
      "learning_rate": 3.1638277591280202e-06,
      "loss": 0.6643,
      "step": 8862
    },
    {
      "epoch": 0.3673173359857433,
      "grad_norm": 0.424174040555954,
      "learning_rate": 3.163620539599652e-06,
      "loss": 0.7004,
      "step": 8863
    },
    {
      "epoch": 0.36735877989141696,
      "grad_norm": 0.42038533091545105,
      "learning_rate": 3.1634133200712834e-06,
      "loss": 0.7422,
      "step": 8864
    },
    {
      "epoch": 0.36740022379709064,
      "grad_norm": 0.3929160535335541,
      "learning_rate": 3.1632061005429152e-06,
      "loss": 0.6885,
      "step": 8865
    },
    {
      "epoch": 0.3674416677027643,
      "grad_norm": 0.41162413358688354,
      "learning_rate": 3.1629988810145475e-06,
      "loss": 0.7068,
      "step": 8866
    },
    {
      "epoch": 0.367483111608438,
      "grad_norm": 0.3875158429145813,
      "learning_rate": 3.162791661486179e-06,
      "loss": 0.7146,
      "step": 8867
    },
    {
      "epoch": 0.3675245555141117,
      "grad_norm": 0.4303911030292511,
      "learning_rate": 3.1625844419578107e-06,
      "loss": 0.6968,
      "step": 8868
    },
    {
      "epoch": 0.3675659994197853,
      "grad_norm": 0.37923362851142883,
      "learning_rate": 3.162377222429442e-06,
      "loss": 0.6641,
      "step": 8869
    },
    {
      "epoch": 0.36760744332545897,
      "grad_norm": 0.3993144631385803,
      "learning_rate": 3.162170002901074e-06,
      "loss": 0.731,
      "step": 8870
    },
    {
      "epoch": 0.36764888723113265,
      "grad_norm": 0.40559154748916626,
      "learning_rate": 3.1619627833727053e-06,
      "loss": 0.7188,
      "step": 8871
    },
    {
      "epoch": 0.3676903311368063,
      "grad_norm": 0.40088963508605957,
      "learning_rate": 3.161755563844337e-06,
      "loss": 0.6777,
      "step": 8872
    },
    {
      "epoch": 0.36773177504248,
      "grad_norm": 0.4128590226173401,
      "learning_rate": 3.1615483443159684e-06,
      "loss": 0.6943,
      "step": 8873
    },
    {
      "epoch": 0.3677732189481537,
      "grad_norm": 0.4353982210159302,
      "learning_rate": 3.1613411247876003e-06,
      "loss": 0.7417,
      "step": 8874
    },
    {
      "epoch": 0.36781466285382736,
      "grad_norm": 0.4161624610424042,
      "learning_rate": 3.161133905259232e-06,
      "loss": 0.6355,
      "step": 8875
    },
    {
      "epoch": 0.36785610675950103,
      "grad_norm": 0.41683724522590637,
      "learning_rate": 3.1609266857308635e-06,
      "loss": 0.7651,
      "step": 8876
    },
    {
      "epoch": 0.3678975506651747,
      "grad_norm": 0.44149574637413025,
      "learning_rate": 3.1607194662024953e-06,
      "loss": 0.7206,
      "step": 8877
    },
    {
      "epoch": 0.36793899457084833,
      "grad_norm": 0.36669236421585083,
      "learning_rate": 3.1605122466741266e-06,
      "loss": 0.6589,
      "step": 8878
    },
    {
      "epoch": 0.367980438476522,
      "grad_norm": 0.42263463139533997,
      "learning_rate": 3.1603050271457585e-06,
      "loss": 0.697,
      "step": 8879
    },
    {
      "epoch": 0.3680218823821957,
      "grad_norm": 0.39665165543556213,
      "learning_rate": 3.16009780761739e-06,
      "loss": 0.6768,
      "step": 8880
    },
    {
      "epoch": 0.36806332628786936,
      "grad_norm": 0.4220811426639557,
      "learning_rate": 3.1598905880890217e-06,
      "loss": 0.7172,
      "step": 8881
    },
    {
      "epoch": 0.36810477019354304,
      "grad_norm": 0.4566970765590668,
      "learning_rate": 3.159683368560653e-06,
      "loss": 0.7004,
      "step": 8882
    },
    {
      "epoch": 0.3681462140992167,
      "grad_norm": 0.42729124426841736,
      "learning_rate": 3.159476149032285e-06,
      "loss": 0.702,
      "step": 8883
    },
    {
      "epoch": 0.3681876580048904,
      "grad_norm": 0.39075806736946106,
      "learning_rate": 3.159268929503917e-06,
      "loss": 0.7207,
      "step": 8884
    },
    {
      "epoch": 0.36822910191056407,
      "grad_norm": 0.40214312076568604,
      "learning_rate": 3.1590617099755485e-06,
      "loss": 0.7302,
      "step": 8885
    },
    {
      "epoch": 0.36827054581623775,
      "grad_norm": 0.3944583833217621,
      "learning_rate": 3.1588544904471803e-06,
      "loss": 0.7052,
      "step": 8886
    },
    {
      "epoch": 0.36831198972191137,
      "grad_norm": 0.4106513559818268,
      "learning_rate": 3.1586472709188117e-06,
      "loss": 0.7422,
      "step": 8887
    },
    {
      "epoch": 0.36835343362758505,
      "grad_norm": 0.41970962285995483,
      "learning_rate": 3.1584400513904435e-06,
      "loss": 0.7058,
      "step": 8888
    },
    {
      "epoch": 0.3683948775332587,
      "grad_norm": 0.3840161859989166,
      "learning_rate": 3.158232831862075e-06,
      "loss": 0.6943,
      "step": 8889
    },
    {
      "epoch": 0.3684363214389324,
      "grad_norm": 0.4291200339794159,
      "learning_rate": 3.1580256123337067e-06,
      "loss": 0.7437,
      "step": 8890
    },
    {
      "epoch": 0.3684777653446061,
      "grad_norm": 0.42865267395973206,
      "learning_rate": 3.157818392805338e-06,
      "loss": 0.7101,
      "step": 8891
    },
    {
      "epoch": 0.36851920925027976,
      "grad_norm": 0.4136693775653839,
      "learning_rate": 3.15761117327697e-06,
      "loss": 0.7096,
      "step": 8892
    },
    {
      "epoch": 0.36856065315595343,
      "grad_norm": 0.4085708558559418,
      "learning_rate": 3.1574039537486017e-06,
      "loss": 0.6813,
      "step": 8893
    },
    {
      "epoch": 0.3686020970616271,
      "grad_norm": 0.3818548619747162,
      "learning_rate": 3.157196734220233e-06,
      "loss": 0.6488,
      "step": 8894
    },
    {
      "epoch": 0.36864354096730073,
      "grad_norm": 0.415693998336792,
      "learning_rate": 3.156989514691865e-06,
      "loss": 0.6547,
      "step": 8895
    },
    {
      "epoch": 0.3686849848729744,
      "grad_norm": 0.38231855630874634,
      "learning_rate": 3.1567822951634962e-06,
      "loss": 0.6954,
      "step": 8896
    },
    {
      "epoch": 0.3687264287786481,
      "grad_norm": 0.4177792966365814,
      "learning_rate": 3.156575075635128e-06,
      "loss": 0.7069,
      "step": 8897
    },
    {
      "epoch": 0.36876787268432176,
      "grad_norm": 0.41496288776397705,
      "learning_rate": 3.1563678561067594e-06,
      "loss": 0.6851,
      "step": 8898
    },
    {
      "epoch": 0.36880931658999544,
      "grad_norm": 0.38058000802993774,
      "learning_rate": 3.1561606365783913e-06,
      "loss": 0.6969,
      "step": 8899
    },
    {
      "epoch": 0.3688507604956691,
      "grad_norm": 0.4469859004020691,
      "learning_rate": 3.1559534170500226e-06,
      "loss": 0.7708,
      "step": 8900
    },
    {
      "epoch": 0.3688922044013428,
      "grad_norm": 0.4495726525783539,
      "learning_rate": 3.155746197521655e-06,
      "loss": 0.6992,
      "step": 8901
    },
    {
      "epoch": 0.36893364830701647,
      "grad_norm": 0.44434696435928345,
      "learning_rate": 3.1555389779932867e-06,
      "loss": 0.6707,
      "step": 8902
    },
    {
      "epoch": 0.36897509221269015,
      "grad_norm": 0.3963526487350464,
      "learning_rate": 3.155331758464918e-06,
      "loss": 0.6733,
      "step": 8903
    },
    {
      "epoch": 0.36901653611836377,
      "grad_norm": 0.5381492376327515,
      "learning_rate": 3.15512453893655e-06,
      "loss": 0.7307,
      "step": 8904
    },
    {
      "epoch": 0.36905798002403745,
      "grad_norm": 0.4107595682144165,
      "learning_rate": 3.1549173194081813e-06,
      "loss": 0.6766,
      "step": 8905
    },
    {
      "epoch": 0.3690994239297111,
      "grad_norm": 0.3853423595428467,
      "learning_rate": 3.154710099879813e-06,
      "loss": 0.71,
      "step": 8906
    },
    {
      "epoch": 0.3691408678353848,
      "grad_norm": 0.44488653540611267,
      "learning_rate": 3.1545028803514445e-06,
      "loss": 0.7334,
      "step": 8907
    },
    {
      "epoch": 0.3691823117410585,
      "grad_norm": 0.42180588841438293,
      "learning_rate": 3.1542956608230763e-06,
      "loss": 0.646,
      "step": 8908
    },
    {
      "epoch": 0.36922375564673215,
      "grad_norm": 0.4292296767234802,
      "learning_rate": 3.154088441294708e-06,
      "loss": 0.6484,
      "step": 8909
    },
    {
      "epoch": 0.36926519955240583,
      "grad_norm": 0.43941107392311096,
      "learning_rate": 3.1538812217663395e-06,
      "loss": 0.6658,
      "step": 8910
    },
    {
      "epoch": 0.3693066434580795,
      "grad_norm": 0.4024118483066559,
      "learning_rate": 3.1536740022379713e-06,
      "loss": 0.7249,
      "step": 8911
    },
    {
      "epoch": 0.3693480873637532,
      "grad_norm": 0.39081859588623047,
      "learning_rate": 3.1534667827096027e-06,
      "loss": 0.6953,
      "step": 8912
    },
    {
      "epoch": 0.3693895312694268,
      "grad_norm": 0.4163510203361511,
      "learning_rate": 3.1532595631812345e-06,
      "loss": 0.7219,
      "step": 8913
    },
    {
      "epoch": 0.3694309751751005,
      "grad_norm": 0.38302725553512573,
      "learning_rate": 3.153052343652866e-06,
      "loss": 0.687,
      "step": 8914
    },
    {
      "epoch": 0.36947241908077416,
      "grad_norm": 0.422585129737854,
      "learning_rate": 3.1528451241244977e-06,
      "loss": 0.6946,
      "step": 8915
    },
    {
      "epoch": 0.36951386298644784,
      "grad_norm": 0.4248790442943573,
      "learning_rate": 3.152637904596129e-06,
      "loss": 0.7201,
      "step": 8916
    },
    {
      "epoch": 0.3695553068921215,
      "grad_norm": 0.3838922381401062,
      "learning_rate": 3.1524306850677613e-06,
      "loss": 0.6603,
      "step": 8917
    },
    {
      "epoch": 0.3695967507977952,
      "grad_norm": 0.4402380585670471,
      "learning_rate": 3.152223465539393e-06,
      "loss": 0.7576,
      "step": 8918
    },
    {
      "epoch": 0.36963819470346887,
      "grad_norm": 0.4050540626049042,
      "learning_rate": 3.1520162460110245e-06,
      "loss": 0.6968,
      "step": 8919
    },
    {
      "epoch": 0.36967963860914255,
      "grad_norm": 0.4492555856704712,
      "learning_rate": 3.1518090264826563e-06,
      "loss": 0.6975,
      "step": 8920
    },
    {
      "epoch": 0.3697210825148162,
      "grad_norm": 0.40193501114845276,
      "learning_rate": 3.1516018069542877e-06,
      "loss": 0.7595,
      "step": 8921
    },
    {
      "epoch": 0.36976252642048985,
      "grad_norm": 0.43714073300361633,
      "learning_rate": 3.1513945874259195e-06,
      "loss": 0.7666,
      "step": 8922
    },
    {
      "epoch": 0.3698039703261635,
      "grad_norm": 0.43069228529930115,
      "learning_rate": 3.151187367897551e-06,
      "loss": 0.6665,
      "step": 8923
    },
    {
      "epoch": 0.3698454142318372,
      "grad_norm": 0.44574522972106934,
      "learning_rate": 3.1509801483691827e-06,
      "loss": 0.7229,
      "step": 8924
    },
    {
      "epoch": 0.3698868581375109,
      "grad_norm": 0.4311257302761078,
      "learning_rate": 3.150772928840814e-06,
      "loss": 0.6721,
      "step": 8925
    },
    {
      "epoch": 0.36992830204318455,
      "grad_norm": 0.41908079385757446,
      "learning_rate": 3.150565709312446e-06,
      "loss": 0.697,
      "step": 8926
    },
    {
      "epoch": 0.36996974594885823,
      "grad_norm": 0.4305960536003113,
      "learning_rate": 3.1503584897840777e-06,
      "loss": 0.754,
      "step": 8927
    },
    {
      "epoch": 0.3700111898545319,
      "grad_norm": 0.40490633249282837,
      "learning_rate": 3.150151270255709e-06,
      "loss": 0.6755,
      "step": 8928
    },
    {
      "epoch": 0.3700526337602056,
      "grad_norm": 0.3791857361793518,
      "learning_rate": 3.149944050727341e-06,
      "loss": 0.6689,
      "step": 8929
    },
    {
      "epoch": 0.3700940776658792,
      "grad_norm": 0.4229313135147095,
      "learning_rate": 3.1497368311989723e-06,
      "loss": 0.712,
      "step": 8930
    },
    {
      "epoch": 0.3701355215715529,
      "grad_norm": 0.43068942427635193,
      "learning_rate": 3.149529611670604e-06,
      "loss": 0.7083,
      "step": 8931
    },
    {
      "epoch": 0.37017696547722656,
      "grad_norm": 0.41415274143218994,
      "learning_rate": 3.1493223921422354e-06,
      "loss": 0.6925,
      "step": 8932
    },
    {
      "epoch": 0.37021840938290024,
      "grad_norm": 0.45212414860725403,
      "learning_rate": 3.1491151726138673e-06,
      "loss": 0.6646,
      "step": 8933
    },
    {
      "epoch": 0.3702598532885739,
      "grad_norm": 0.4104657769203186,
      "learning_rate": 3.1489079530854986e-06,
      "loss": 0.6316,
      "step": 8934
    },
    {
      "epoch": 0.3703012971942476,
      "grad_norm": 0.413814902305603,
      "learning_rate": 3.148700733557131e-06,
      "loss": 0.7446,
      "step": 8935
    },
    {
      "epoch": 0.37034274109992127,
      "grad_norm": 0.3918263614177704,
      "learning_rate": 3.1484935140287627e-06,
      "loss": 0.6893,
      "step": 8936
    },
    {
      "epoch": 0.37038418500559495,
      "grad_norm": 0.40154901146888733,
      "learning_rate": 3.148286294500394e-06,
      "loss": 0.6958,
      "step": 8937
    },
    {
      "epoch": 0.3704256289112686,
      "grad_norm": 0.44967883825302124,
      "learning_rate": 3.148079074972026e-06,
      "loss": 0.7539,
      "step": 8938
    },
    {
      "epoch": 0.37046707281694224,
      "grad_norm": 0.37649351358413696,
      "learning_rate": 3.1478718554436573e-06,
      "loss": 0.6957,
      "step": 8939
    },
    {
      "epoch": 0.3705085167226159,
      "grad_norm": 0.41725289821624756,
      "learning_rate": 3.147664635915289e-06,
      "loss": 0.689,
      "step": 8940
    },
    {
      "epoch": 0.3705499606282896,
      "grad_norm": 0.4126502275466919,
      "learning_rate": 3.1474574163869205e-06,
      "loss": 0.719,
      "step": 8941
    },
    {
      "epoch": 0.3705914045339633,
      "grad_norm": 0.4377695918083191,
      "learning_rate": 3.1472501968585523e-06,
      "loss": 0.739,
      "step": 8942
    },
    {
      "epoch": 0.37063284843963695,
      "grad_norm": 0.38311824202537537,
      "learning_rate": 3.1470429773301837e-06,
      "loss": 0.6895,
      "step": 8943
    },
    {
      "epoch": 0.37067429234531063,
      "grad_norm": 0.40987834334373474,
      "learning_rate": 3.1468357578018155e-06,
      "loss": 0.7278,
      "step": 8944
    },
    {
      "epoch": 0.3707157362509843,
      "grad_norm": 0.39136096835136414,
      "learning_rate": 3.1466285382734473e-06,
      "loss": 0.729,
      "step": 8945
    },
    {
      "epoch": 0.370757180156658,
      "grad_norm": 0.42354851961135864,
      "learning_rate": 3.1464213187450787e-06,
      "loss": 0.6946,
      "step": 8946
    },
    {
      "epoch": 0.37079862406233166,
      "grad_norm": 0.43437498807907104,
      "learning_rate": 3.1462140992167105e-06,
      "loss": 0.6929,
      "step": 8947
    },
    {
      "epoch": 0.3708400679680053,
      "grad_norm": 0.39896121621131897,
      "learning_rate": 3.146006879688342e-06,
      "loss": 0.7188,
      "step": 8948
    },
    {
      "epoch": 0.37088151187367896,
      "grad_norm": 0.3920890688896179,
      "learning_rate": 3.1457996601599737e-06,
      "loss": 0.6741,
      "step": 8949
    },
    {
      "epoch": 0.37092295577935264,
      "grad_norm": 0.4017693102359772,
      "learning_rate": 3.145592440631605e-06,
      "loss": 0.7036,
      "step": 8950
    },
    {
      "epoch": 0.3709643996850263,
      "grad_norm": 0.4487254321575165,
      "learning_rate": 3.1453852211032373e-06,
      "loss": 0.6853,
      "step": 8951
    },
    {
      "epoch": 0.3710058435907,
      "grad_norm": 0.41148504614830017,
      "learning_rate": 3.1451780015748682e-06,
      "loss": 0.6652,
      "step": 8952
    },
    {
      "epoch": 0.37104728749637367,
      "grad_norm": 0.40088164806365967,
      "learning_rate": 3.1449707820465005e-06,
      "loss": 0.7283,
      "step": 8953
    },
    {
      "epoch": 0.37108873140204734,
      "grad_norm": 0.39436858892440796,
      "learning_rate": 3.1447635625181323e-06,
      "loss": 0.6843,
      "step": 8954
    },
    {
      "epoch": 0.371130175307721,
      "grad_norm": 0.4027118384838104,
      "learning_rate": 3.1445563429897637e-06,
      "loss": 0.7324,
      "step": 8955
    },
    {
      "epoch": 0.37117161921339464,
      "grad_norm": 0.42992323637008667,
      "learning_rate": 3.1443491234613955e-06,
      "loss": 0.7239,
      "step": 8956
    },
    {
      "epoch": 0.3712130631190683,
      "grad_norm": 0.45619744062423706,
      "learning_rate": 3.144141903933027e-06,
      "loss": 0.7329,
      "step": 8957
    },
    {
      "epoch": 0.371254507024742,
      "grad_norm": 0.41691407561302185,
      "learning_rate": 3.1439346844046587e-06,
      "loss": 0.6731,
      "step": 8958
    },
    {
      "epoch": 0.3712959509304157,
      "grad_norm": 0.41546350717544556,
      "learning_rate": 3.14372746487629e-06,
      "loss": 0.6968,
      "step": 8959
    },
    {
      "epoch": 0.37133739483608935,
      "grad_norm": 0.44289126992225647,
      "learning_rate": 3.143520245347922e-06,
      "loss": 0.7498,
      "step": 8960
    },
    {
      "epoch": 0.37137883874176303,
      "grad_norm": 0.40682074427604675,
      "learning_rate": 3.1433130258195533e-06,
      "loss": 0.7139,
      "step": 8961
    },
    {
      "epoch": 0.3714202826474367,
      "grad_norm": 0.3976593613624573,
      "learning_rate": 3.143105806291185e-06,
      "loss": 0.678,
      "step": 8962
    },
    {
      "epoch": 0.3714617265531104,
      "grad_norm": 0.44312337040901184,
      "learning_rate": 3.142898586762817e-06,
      "loss": 0.6653,
      "step": 8963
    },
    {
      "epoch": 0.37150317045878406,
      "grad_norm": 0.44188857078552246,
      "learning_rate": 3.1426913672344483e-06,
      "loss": 0.7041,
      "step": 8964
    },
    {
      "epoch": 0.3715446143644577,
      "grad_norm": 0.4210462272167206,
      "learning_rate": 3.14248414770608e-06,
      "loss": 0.6686,
      "step": 8965
    },
    {
      "epoch": 0.37158605827013136,
      "grad_norm": 0.43581393361091614,
      "learning_rate": 3.1422769281777115e-06,
      "loss": 0.6913,
      "step": 8966
    },
    {
      "epoch": 0.37162750217580504,
      "grad_norm": 0.4365854263305664,
      "learning_rate": 3.1420697086493433e-06,
      "loss": 0.7068,
      "step": 8967
    },
    {
      "epoch": 0.3716689460814787,
      "grad_norm": 0.40731990337371826,
      "learning_rate": 3.1418624891209746e-06,
      "loss": 0.6987,
      "step": 8968
    },
    {
      "epoch": 0.3717103899871524,
      "grad_norm": 0.437568724155426,
      "learning_rate": 3.141655269592607e-06,
      "loss": 0.7211,
      "step": 8969
    },
    {
      "epoch": 0.37175183389282607,
      "grad_norm": 0.4120711088180542,
      "learning_rate": 3.1414480500642387e-06,
      "loss": 0.7236,
      "step": 8970
    },
    {
      "epoch": 0.37179327779849974,
      "grad_norm": 0.42031511664390564,
      "learning_rate": 3.14124083053587e-06,
      "loss": 0.7183,
      "step": 8971
    },
    {
      "epoch": 0.3718347217041734,
      "grad_norm": 0.4255237877368927,
      "learning_rate": 3.141033611007502e-06,
      "loss": 0.744,
      "step": 8972
    },
    {
      "epoch": 0.3718761656098471,
      "grad_norm": 0.429678738117218,
      "learning_rate": 3.1408263914791333e-06,
      "loss": 0.7146,
      "step": 8973
    },
    {
      "epoch": 0.3719176095155207,
      "grad_norm": 0.4080464839935303,
      "learning_rate": 3.140619171950765e-06,
      "loss": 0.6656,
      "step": 8974
    },
    {
      "epoch": 0.3719590534211944,
      "grad_norm": 0.40447452664375305,
      "learning_rate": 3.1404119524223965e-06,
      "loss": 0.6836,
      "step": 8975
    },
    {
      "epoch": 0.3720004973268681,
      "grad_norm": 0.4139817953109741,
      "learning_rate": 3.1402047328940283e-06,
      "loss": 0.6909,
      "step": 8976
    },
    {
      "epoch": 0.37204194123254175,
      "grad_norm": 0.4272252321243286,
      "learning_rate": 3.1399975133656597e-06,
      "loss": 0.7529,
      "step": 8977
    },
    {
      "epoch": 0.37208338513821543,
      "grad_norm": 0.40480709075927734,
      "learning_rate": 3.1397902938372915e-06,
      "loss": 0.6465,
      "step": 8978
    },
    {
      "epoch": 0.3721248290438891,
      "grad_norm": 0.4349636435508728,
      "learning_rate": 3.1395830743089233e-06,
      "loss": 0.7134,
      "step": 8979
    },
    {
      "epoch": 0.3721662729495628,
      "grad_norm": 0.3812727630138397,
      "learning_rate": 3.1393758547805547e-06,
      "loss": 0.6343,
      "step": 8980
    },
    {
      "epoch": 0.37220771685523646,
      "grad_norm": 0.4279041886329651,
      "learning_rate": 3.1391686352521865e-06,
      "loss": 0.7083,
      "step": 8981
    },
    {
      "epoch": 0.3722491607609101,
      "grad_norm": 0.43525397777557373,
      "learning_rate": 3.138961415723818e-06,
      "loss": 0.7056,
      "step": 8982
    },
    {
      "epoch": 0.37229060466658376,
      "grad_norm": 0.41355806589126587,
      "learning_rate": 3.1387541961954497e-06,
      "loss": 0.7041,
      "step": 8983
    },
    {
      "epoch": 0.37233204857225743,
      "grad_norm": 0.4455762505531311,
      "learning_rate": 3.138546976667081e-06,
      "loss": 0.7184,
      "step": 8984
    },
    {
      "epoch": 0.3723734924779311,
      "grad_norm": 0.40540534257888794,
      "learning_rate": 3.1383397571387133e-06,
      "loss": 0.6278,
      "step": 8985
    },
    {
      "epoch": 0.3724149363836048,
      "grad_norm": 0.4474935233592987,
      "learning_rate": 3.1381325376103442e-06,
      "loss": 0.7366,
      "step": 8986
    },
    {
      "epoch": 0.37245638028927847,
      "grad_norm": 0.50785893201828,
      "learning_rate": 3.1379253180819765e-06,
      "loss": 0.8015,
      "step": 8987
    },
    {
      "epoch": 0.37249782419495214,
      "grad_norm": 0.44993510842323303,
      "learning_rate": 3.1377180985536083e-06,
      "loss": 0.7634,
      "step": 8988
    },
    {
      "epoch": 0.3725392681006258,
      "grad_norm": 0.44682446122169495,
      "learning_rate": 3.1375108790252397e-06,
      "loss": 0.7126,
      "step": 8989
    },
    {
      "epoch": 0.3725807120062995,
      "grad_norm": 0.4277103841304779,
      "learning_rate": 3.1373036594968715e-06,
      "loss": 0.7437,
      "step": 8990
    },
    {
      "epoch": 0.3726221559119731,
      "grad_norm": 0.387192964553833,
      "learning_rate": 3.137096439968503e-06,
      "loss": 0.681,
      "step": 8991
    },
    {
      "epoch": 0.3726635998176468,
      "grad_norm": 0.4618108868598938,
      "learning_rate": 3.1368892204401347e-06,
      "loss": 0.7266,
      "step": 8992
    },
    {
      "epoch": 0.3727050437233205,
      "grad_norm": 0.3832768499851227,
      "learning_rate": 3.136682000911766e-06,
      "loss": 0.731,
      "step": 8993
    },
    {
      "epoch": 0.37274648762899415,
      "grad_norm": 0.4146955907344818,
      "learning_rate": 3.136474781383398e-06,
      "loss": 0.7415,
      "step": 8994
    },
    {
      "epoch": 0.3727879315346678,
      "grad_norm": 0.4216214716434479,
      "learning_rate": 3.1362675618550293e-06,
      "loss": 0.7126,
      "step": 8995
    },
    {
      "epoch": 0.3728293754403415,
      "grad_norm": 0.4020933508872986,
      "learning_rate": 3.136060342326661e-06,
      "loss": 0.6865,
      "step": 8996
    },
    {
      "epoch": 0.3728708193460152,
      "grad_norm": 0.44254156947135925,
      "learning_rate": 3.135853122798293e-06,
      "loss": 0.7759,
      "step": 8997
    },
    {
      "epoch": 0.37291226325168886,
      "grad_norm": 0.3965075612068176,
      "learning_rate": 3.1356459032699243e-06,
      "loss": 0.7112,
      "step": 8998
    },
    {
      "epoch": 0.37295370715736254,
      "grad_norm": 0.4085024893283844,
      "learning_rate": 3.135438683741556e-06,
      "loss": 0.7026,
      "step": 8999
    },
    {
      "epoch": 0.37299515106303616,
      "grad_norm": 0.3847168982028961,
      "learning_rate": 3.1352314642131875e-06,
      "loss": 0.6571,
      "step": 9000
    },
    {
      "epoch": 0.37303659496870983,
      "grad_norm": 0.4005201458930969,
      "learning_rate": 3.1350242446848193e-06,
      "loss": 0.6799,
      "step": 9001
    },
    {
      "epoch": 0.3730780388743835,
      "grad_norm": 0.38419410586357117,
      "learning_rate": 3.1348170251564507e-06,
      "loss": 0.6427,
      "step": 9002
    },
    {
      "epoch": 0.3731194827800572,
      "grad_norm": 0.4034765362739563,
      "learning_rate": 3.134609805628083e-06,
      "loss": 0.7407,
      "step": 9003
    },
    {
      "epoch": 0.37316092668573086,
      "grad_norm": 0.42471420764923096,
      "learning_rate": 3.134402586099714e-06,
      "loss": 0.6909,
      "step": 9004
    },
    {
      "epoch": 0.37320237059140454,
      "grad_norm": 0.3986184000968933,
      "learning_rate": 3.134195366571346e-06,
      "loss": 0.6698,
      "step": 9005
    },
    {
      "epoch": 0.3732438144970782,
      "grad_norm": 0.3673023581504822,
      "learning_rate": 3.133988147042978e-06,
      "loss": 0.6846,
      "step": 9006
    },
    {
      "epoch": 0.3732852584027519,
      "grad_norm": 0.4386202096939087,
      "learning_rate": 3.1337809275146093e-06,
      "loss": 0.7034,
      "step": 9007
    },
    {
      "epoch": 0.3733267023084256,
      "grad_norm": 0.42089977860450745,
      "learning_rate": 3.133573707986241e-06,
      "loss": 0.7126,
      "step": 9008
    },
    {
      "epoch": 0.3733681462140992,
      "grad_norm": 0.42322877049446106,
      "learning_rate": 3.1333664884578725e-06,
      "loss": 0.7217,
      "step": 9009
    },
    {
      "epoch": 0.37340959011977287,
      "grad_norm": 0.4157218933105469,
      "learning_rate": 3.1331592689295043e-06,
      "loss": 0.6995,
      "step": 9010
    },
    {
      "epoch": 0.37345103402544655,
      "grad_norm": 0.4435606598854065,
      "learning_rate": 3.1329520494011357e-06,
      "loss": 0.7151,
      "step": 9011
    },
    {
      "epoch": 0.3734924779311202,
      "grad_norm": 0.42205163836479187,
      "learning_rate": 3.1327448298727675e-06,
      "loss": 0.7203,
      "step": 9012
    },
    {
      "epoch": 0.3735339218367939,
      "grad_norm": 0.4289886951446533,
      "learning_rate": 3.132537610344399e-06,
      "loss": 0.7288,
      "step": 9013
    },
    {
      "epoch": 0.3735753657424676,
      "grad_norm": 0.44541287422180176,
      "learning_rate": 3.1323303908160307e-06,
      "loss": 0.6755,
      "step": 9014
    },
    {
      "epoch": 0.37361680964814126,
      "grad_norm": 0.42502719163894653,
      "learning_rate": 3.1321231712876625e-06,
      "loss": 0.6941,
      "step": 9015
    },
    {
      "epoch": 0.37365825355381493,
      "grad_norm": 0.418663889169693,
      "learning_rate": 3.131915951759294e-06,
      "loss": 0.7354,
      "step": 9016
    },
    {
      "epoch": 0.37369969745948856,
      "grad_norm": 0.37722355127334595,
      "learning_rate": 3.1317087322309257e-06,
      "loss": 0.7615,
      "step": 9017
    },
    {
      "epoch": 0.37374114136516223,
      "grad_norm": 0.4065714180469513,
      "learning_rate": 3.131501512702557e-06,
      "loss": 0.6987,
      "step": 9018
    },
    {
      "epoch": 0.3737825852708359,
      "grad_norm": 0.37011241912841797,
      "learning_rate": 3.1312942931741893e-06,
      "loss": 0.6885,
      "step": 9019
    },
    {
      "epoch": 0.3738240291765096,
      "grad_norm": 0.39469170570373535,
      "learning_rate": 3.1310870736458203e-06,
      "loss": 0.6821,
      "step": 9020
    },
    {
      "epoch": 0.37386547308218326,
      "grad_norm": 0.4192628264427185,
      "learning_rate": 3.1308798541174525e-06,
      "loss": 0.6526,
      "step": 9021
    },
    {
      "epoch": 0.37390691698785694,
      "grad_norm": 0.44984716176986694,
      "learning_rate": 3.1306726345890835e-06,
      "loss": 0.7432,
      "step": 9022
    },
    {
      "epoch": 0.3739483608935306,
      "grad_norm": 0.4296228289604187,
      "learning_rate": 3.1304654150607157e-06,
      "loss": 0.6796,
      "step": 9023
    },
    {
      "epoch": 0.3739898047992043,
      "grad_norm": 0.40176934003829956,
      "learning_rate": 3.1302581955323475e-06,
      "loss": 0.6724,
      "step": 9024
    },
    {
      "epoch": 0.37403124870487797,
      "grad_norm": 0.580371618270874,
      "learning_rate": 3.130050976003979e-06,
      "loss": 0.7333,
      "step": 9025
    },
    {
      "epoch": 0.3740726926105516,
      "grad_norm": 0.40399980545043945,
      "learning_rate": 3.1298437564756107e-06,
      "loss": 0.6982,
      "step": 9026
    },
    {
      "epoch": 0.37411413651622527,
      "grad_norm": 0.47145065665245056,
      "learning_rate": 3.129636536947242e-06,
      "loss": 0.8025,
      "step": 9027
    },
    {
      "epoch": 0.37415558042189895,
      "grad_norm": 0.4262302815914154,
      "learning_rate": 3.129429317418874e-06,
      "loss": 0.7062,
      "step": 9028
    },
    {
      "epoch": 0.3741970243275726,
      "grad_norm": 0.40217873454093933,
      "learning_rate": 3.1292220978905053e-06,
      "loss": 0.7089,
      "step": 9029
    },
    {
      "epoch": 0.3742384682332463,
      "grad_norm": 0.41054466366767883,
      "learning_rate": 3.129014878362137e-06,
      "loss": 0.6703,
      "step": 9030
    },
    {
      "epoch": 0.37427991213892,
      "grad_norm": 0.42367956042289734,
      "learning_rate": 3.128807658833769e-06,
      "loss": 0.771,
      "step": 9031
    },
    {
      "epoch": 0.37432135604459366,
      "grad_norm": 0.41689419746398926,
      "learning_rate": 3.1286004393054003e-06,
      "loss": 0.6696,
      "step": 9032
    },
    {
      "epoch": 0.37436279995026733,
      "grad_norm": 0.4569893777370453,
      "learning_rate": 3.128393219777032e-06,
      "loss": 0.7959,
      "step": 9033
    },
    {
      "epoch": 0.374404243855941,
      "grad_norm": 0.4198036789894104,
      "learning_rate": 3.1281860002486635e-06,
      "loss": 0.6592,
      "step": 9034
    },
    {
      "epoch": 0.37444568776161463,
      "grad_norm": 0.4172281324863434,
      "learning_rate": 3.1279787807202953e-06,
      "loss": 0.6841,
      "step": 9035
    },
    {
      "epoch": 0.3744871316672883,
      "grad_norm": 0.41084471344947815,
      "learning_rate": 3.1277715611919267e-06,
      "loss": 0.7244,
      "step": 9036
    },
    {
      "epoch": 0.374528575572962,
      "grad_norm": 0.41479793190956116,
      "learning_rate": 3.127564341663559e-06,
      "loss": 0.6646,
      "step": 9037
    },
    {
      "epoch": 0.37457001947863566,
      "grad_norm": 0.36531442403793335,
      "learning_rate": 3.12735712213519e-06,
      "loss": 0.7004,
      "step": 9038
    },
    {
      "epoch": 0.37461146338430934,
      "grad_norm": 0.39806339144706726,
      "learning_rate": 3.127149902606822e-06,
      "loss": 0.6904,
      "step": 9039
    },
    {
      "epoch": 0.374652907289983,
      "grad_norm": 0.3861761689186096,
      "learning_rate": 3.126942683078454e-06,
      "loss": 0.6917,
      "step": 9040
    },
    {
      "epoch": 0.3746943511956567,
      "grad_norm": 0.4125358462333679,
      "learning_rate": 3.1267354635500853e-06,
      "loss": 0.7041,
      "step": 9041
    },
    {
      "epoch": 0.37473579510133037,
      "grad_norm": 0.3887813687324524,
      "learning_rate": 3.126528244021717e-06,
      "loss": 0.6926,
      "step": 9042
    },
    {
      "epoch": 0.374777239007004,
      "grad_norm": 0.41943198442459106,
      "learning_rate": 3.1263210244933485e-06,
      "loss": 0.6904,
      "step": 9043
    },
    {
      "epoch": 0.37481868291267767,
      "grad_norm": 0.40313640236854553,
      "learning_rate": 3.1261138049649803e-06,
      "loss": 0.6785,
      "step": 9044
    },
    {
      "epoch": 0.37486012681835135,
      "grad_norm": 0.42146387696266174,
      "learning_rate": 3.1259065854366117e-06,
      "loss": 0.6936,
      "step": 9045
    },
    {
      "epoch": 0.374901570724025,
      "grad_norm": 0.39755356311798096,
      "learning_rate": 3.1256993659082435e-06,
      "loss": 0.731,
      "step": 9046
    },
    {
      "epoch": 0.3749430146296987,
      "grad_norm": 0.3908047378063202,
      "learning_rate": 3.125492146379875e-06,
      "loss": 0.684,
      "step": 9047
    },
    {
      "epoch": 0.3749844585353724,
      "grad_norm": 0.4660780429840088,
      "learning_rate": 3.1252849268515067e-06,
      "loss": 0.7664,
      "step": 9048
    },
    {
      "epoch": 0.37502590244104606,
      "grad_norm": 0.3982694745063782,
      "learning_rate": 3.1250777073231385e-06,
      "loss": 0.6882,
      "step": 9049
    },
    {
      "epoch": 0.37506734634671973,
      "grad_norm": 0.3845243752002716,
      "learning_rate": 3.12487048779477e-06,
      "loss": 0.6697,
      "step": 9050
    },
    {
      "epoch": 0.3751087902523934,
      "grad_norm": 0.40967074036598206,
      "learning_rate": 3.1246632682664017e-06,
      "loss": 0.7224,
      "step": 9051
    },
    {
      "epoch": 0.37515023415806703,
      "grad_norm": 0.45407307147979736,
      "learning_rate": 3.124456048738033e-06,
      "loss": 0.6774,
      "step": 9052
    },
    {
      "epoch": 0.3751916780637407,
      "grad_norm": 0.36823195219039917,
      "learning_rate": 3.1242488292096653e-06,
      "loss": 0.6522,
      "step": 9053
    },
    {
      "epoch": 0.3752331219694144,
      "grad_norm": 0.3910224735736847,
      "learning_rate": 3.1240416096812963e-06,
      "loss": 0.7097,
      "step": 9054
    },
    {
      "epoch": 0.37527456587508806,
      "grad_norm": 0.4064963459968567,
      "learning_rate": 3.1238343901529285e-06,
      "loss": 0.6512,
      "step": 9055
    },
    {
      "epoch": 0.37531600978076174,
      "grad_norm": 0.3889906704425812,
      "learning_rate": 3.12362717062456e-06,
      "loss": 0.6978,
      "step": 9056
    },
    {
      "epoch": 0.3753574536864354,
      "grad_norm": 0.3777136206626892,
      "learning_rate": 3.1234199510961917e-06,
      "loss": 0.6639,
      "step": 9057
    },
    {
      "epoch": 0.3753988975921091,
      "grad_norm": 0.39697420597076416,
      "learning_rate": 3.1232127315678235e-06,
      "loss": 0.666,
      "step": 9058
    },
    {
      "epoch": 0.37544034149778277,
      "grad_norm": 0.41608989238739014,
      "learning_rate": 3.123005512039455e-06,
      "loss": 0.762,
      "step": 9059
    },
    {
      "epoch": 0.37548178540345645,
      "grad_norm": 0.4434530436992645,
      "learning_rate": 3.1227982925110867e-06,
      "loss": 0.7117,
      "step": 9060
    },
    {
      "epoch": 0.37552322930913007,
      "grad_norm": 0.4080751836299896,
      "learning_rate": 3.122591072982718e-06,
      "loss": 0.7278,
      "step": 9061
    },
    {
      "epoch": 0.37556467321480375,
      "grad_norm": 0.4407501816749573,
      "learning_rate": 3.12238385345435e-06,
      "loss": 0.7041,
      "step": 9062
    },
    {
      "epoch": 0.3756061171204774,
      "grad_norm": 0.4107012450695038,
      "learning_rate": 3.1221766339259813e-06,
      "loss": 0.7275,
      "step": 9063
    },
    {
      "epoch": 0.3756475610261511,
      "grad_norm": 0.402629554271698,
      "learning_rate": 3.121969414397613e-06,
      "loss": 0.7366,
      "step": 9064
    },
    {
      "epoch": 0.3756890049318248,
      "grad_norm": 0.4513614773750305,
      "learning_rate": 3.1217621948692445e-06,
      "loss": 0.7582,
      "step": 9065
    },
    {
      "epoch": 0.37573044883749845,
      "grad_norm": 0.4121415615081787,
      "learning_rate": 3.1215549753408763e-06,
      "loss": 0.7017,
      "step": 9066
    },
    {
      "epoch": 0.37577189274317213,
      "grad_norm": 0.4087435305118561,
      "learning_rate": 3.121347755812508e-06,
      "loss": 0.6853,
      "step": 9067
    },
    {
      "epoch": 0.3758133366488458,
      "grad_norm": 0.42026081681251526,
      "learning_rate": 3.1211405362841395e-06,
      "loss": 0.6765,
      "step": 9068
    },
    {
      "epoch": 0.3758547805545195,
      "grad_norm": 0.4215841293334961,
      "learning_rate": 3.1209333167557713e-06,
      "loss": 0.7017,
      "step": 9069
    },
    {
      "epoch": 0.3758962244601931,
      "grad_norm": 0.42803895473480225,
      "learning_rate": 3.1207260972274027e-06,
      "loss": 0.7014,
      "step": 9070
    },
    {
      "epoch": 0.3759376683658668,
      "grad_norm": 0.3890891671180725,
      "learning_rate": 3.120518877699035e-06,
      "loss": 0.6689,
      "step": 9071
    },
    {
      "epoch": 0.37597911227154046,
      "grad_norm": 0.3961104154586792,
      "learning_rate": 3.120311658170666e-06,
      "loss": 0.6892,
      "step": 9072
    },
    {
      "epoch": 0.37602055617721414,
      "grad_norm": 0.41163522005081177,
      "learning_rate": 3.120104438642298e-06,
      "loss": 0.7423,
      "step": 9073
    },
    {
      "epoch": 0.3760620000828878,
      "grad_norm": 0.3853738605976105,
      "learning_rate": 3.1198972191139295e-06,
      "loss": 0.6439,
      "step": 9074
    },
    {
      "epoch": 0.3761034439885615,
      "grad_norm": 0.43779686093330383,
      "learning_rate": 3.1196899995855613e-06,
      "loss": 0.7368,
      "step": 9075
    },
    {
      "epoch": 0.37614488789423517,
      "grad_norm": 0.38855960965156555,
      "learning_rate": 3.119482780057193e-06,
      "loss": 0.6627,
      "step": 9076
    },
    {
      "epoch": 0.37618633179990885,
      "grad_norm": 0.4142540395259857,
      "learning_rate": 3.1192755605288245e-06,
      "loss": 0.6985,
      "step": 9077
    },
    {
      "epoch": 0.37622777570558247,
      "grad_norm": 0.4327195882797241,
      "learning_rate": 3.1190683410004563e-06,
      "loss": 0.6885,
      "step": 9078
    },
    {
      "epoch": 0.37626921961125614,
      "grad_norm": 0.4253315329551697,
      "learning_rate": 3.1188611214720877e-06,
      "loss": 0.6682,
      "step": 9079
    },
    {
      "epoch": 0.3763106635169298,
      "grad_norm": 0.42176294326782227,
      "learning_rate": 3.1186539019437195e-06,
      "loss": 0.7368,
      "step": 9080
    },
    {
      "epoch": 0.3763521074226035,
      "grad_norm": 0.43868452310562134,
      "learning_rate": 3.118446682415351e-06,
      "loss": 0.6815,
      "step": 9081
    },
    {
      "epoch": 0.3763935513282772,
      "grad_norm": 0.42157405614852905,
      "learning_rate": 3.1182394628869827e-06,
      "loss": 0.7676,
      "step": 9082
    },
    {
      "epoch": 0.37643499523395085,
      "grad_norm": 0.4277568757534027,
      "learning_rate": 3.118032243358614e-06,
      "loss": 0.6919,
      "step": 9083
    },
    {
      "epoch": 0.37647643913962453,
      "grad_norm": 0.42543360590934753,
      "learning_rate": 3.117825023830246e-06,
      "loss": 0.6975,
      "step": 9084
    },
    {
      "epoch": 0.3765178830452982,
      "grad_norm": 0.38533830642700195,
      "learning_rate": 3.1176178043018777e-06,
      "loss": 0.6488,
      "step": 9085
    },
    {
      "epoch": 0.3765593269509719,
      "grad_norm": 0.4000158905982971,
      "learning_rate": 3.117410584773509e-06,
      "loss": 0.6829,
      "step": 9086
    },
    {
      "epoch": 0.3766007708566455,
      "grad_norm": 0.4085211753845215,
      "learning_rate": 3.1172033652451413e-06,
      "loss": 0.6871,
      "step": 9087
    },
    {
      "epoch": 0.3766422147623192,
      "grad_norm": 0.41819900274276733,
      "learning_rate": 3.1169961457167723e-06,
      "loss": 0.679,
      "step": 9088
    },
    {
      "epoch": 0.37668365866799286,
      "grad_norm": 0.37898722290992737,
      "learning_rate": 3.1167889261884045e-06,
      "loss": 0.6273,
      "step": 9089
    },
    {
      "epoch": 0.37672510257366654,
      "grad_norm": 0.3954674303531647,
      "learning_rate": 3.116581706660036e-06,
      "loss": 0.6678,
      "step": 9090
    },
    {
      "epoch": 0.3767665464793402,
      "grad_norm": 0.4099103510379791,
      "learning_rate": 3.1163744871316677e-06,
      "loss": 0.674,
      "step": 9091
    },
    {
      "epoch": 0.3768079903850139,
      "grad_norm": 0.4313313961029053,
      "learning_rate": 3.1161672676032995e-06,
      "loss": 0.693,
      "step": 9092
    },
    {
      "epoch": 0.37684943429068757,
      "grad_norm": 0.4157784879207611,
      "learning_rate": 3.115960048074931e-06,
      "loss": 0.7014,
      "step": 9093
    },
    {
      "epoch": 0.37689087819636125,
      "grad_norm": 0.38549521565437317,
      "learning_rate": 3.1157528285465627e-06,
      "loss": 0.7111,
      "step": 9094
    },
    {
      "epoch": 0.3769323221020349,
      "grad_norm": 0.4178113341331482,
      "learning_rate": 3.115545609018194e-06,
      "loss": 0.7021,
      "step": 9095
    },
    {
      "epoch": 0.37697376600770854,
      "grad_norm": 0.4093519151210785,
      "learning_rate": 3.115338389489826e-06,
      "loss": 0.6753,
      "step": 9096
    },
    {
      "epoch": 0.3770152099133822,
      "grad_norm": 0.46477174758911133,
      "learning_rate": 3.1151311699614573e-06,
      "loss": 0.7068,
      "step": 9097
    },
    {
      "epoch": 0.3770566538190559,
      "grad_norm": 0.431284636259079,
      "learning_rate": 3.114923950433089e-06,
      "loss": 0.7107,
      "step": 9098
    },
    {
      "epoch": 0.3770980977247296,
      "grad_norm": 0.4034000635147095,
      "learning_rate": 3.1147167309047205e-06,
      "loss": 0.6477,
      "step": 9099
    },
    {
      "epoch": 0.37713954163040325,
      "grad_norm": 0.40878334641456604,
      "learning_rate": 3.1145095113763523e-06,
      "loss": 0.7117,
      "step": 9100
    },
    {
      "epoch": 0.37718098553607693,
      "grad_norm": 0.41085487604141235,
      "learning_rate": 3.114302291847984e-06,
      "loss": 0.7378,
      "step": 9101
    },
    {
      "epoch": 0.3772224294417506,
      "grad_norm": 0.40366291999816895,
      "learning_rate": 3.1140950723196155e-06,
      "loss": 0.7061,
      "step": 9102
    },
    {
      "epoch": 0.3772638733474243,
      "grad_norm": 0.39071395993232727,
      "learning_rate": 3.1138878527912477e-06,
      "loss": 0.6768,
      "step": 9103
    },
    {
      "epoch": 0.3773053172530979,
      "grad_norm": 0.4179964065551758,
      "learning_rate": 3.1136806332628787e-06,
      "loss": 0.7324,
      "step": 9104
    },
    {
      "epoch": 0.3773467611587716,
      "grad_norm": 0.42246708273887634,
      "learning_rate": 3.113473413734511e-06,
      "loss": 0.7065,
      "step": 9105
    },
    {
      "epoch": 0.37738820506444526,
      "grad_norm": 0.4216391444206238,
      "learning_rate": 3.113266194206142e-06,
      "loss": 0.74,
      "step": 9106
    },
    {
      "epoch": 0.37742964897011894,
      "grad_norm": 0.40119192004203796,
      "learning_rate": 3.113058974677774e-06,
      "loss": 0.6599,
      "step": 9107
    },
    {
      "epoch": 0.3774710928757926,
      "grad_norm": 0.39321181178092957,
      "learning_rate": 3.1128517551494055e-06,
      "loss": 0.703,
      "step": 9108
    },
    {
      "epoch": 0.3775125367814663,
      "grad_norm": 0.39705103635787964,
      "learning_rate": 3.1126445356210373e-06,
      "loss": 0.71,
      "step": 9109
    },
    {
      "epoch": 0.37755398068713997,
      "grad_norm": 0.3890209197998047,
      "learning_rate": 3.112437316092669e-06,
      "loss": 0.7162,
      "step": 9110
    },
    {
      "epoch": 0.37759542459281364,
      "grad_norm": 0.410936564207077,
      "learning_rate": 3.1122300965643005e-06,
      "loss": 0.6965,
      "step": 9111
    },
    {
      "epoch": 0.3776368684984873,
      "grad_norm": 0.4227541387081146,
      "learning_rate": 3.1120228770359323e-06,
      "loss": 0.7002,
      "step": 9112
    },
    {
      "epoch": 0.37767831240416094,
      "grad_norm": 0.41310614347457886,
      "learning_rate": 3.1118156575075637e-06,
      "loss": 0.7417,
      "step": 9113
    },
    {
      "epoch": 0.3777197563098346,
      "grad_norm": 0.39821621775627136,
      "learning_rate": 3.1116084379791955e-06,
      "loss": 0.6428,
      "step": 9114
    },
    {
      "epoch": 0.3777612002155083,
      "grad_norm": 0.3745613992214203,
      "learning_rate": 3.111401218450827e-06,
      "loss": 0.6638,
      "step": 9115
    },
    {
      "epoch": 0.377802644121182,
      "grad_norm": 0.3949117362499237,
      "learning_rate": 3.1111939989224587e-06,
      "loss": 0.671,
      "step": 9116
    },
    {
      "epoch": 0.37784408802685565,
      "grad_norm": 0.4343055486679077,
      "learning_rate": 3.11098677939409e-06,
      "loss": 0.6991,
      "step": 9117
    },
    {
      "epoch": 0.37788553193252933,
      "grad_norm": 0.37441444396972656,
      "learning_rate": 3.110779559865722e-06,
      "loss": 0.6653,
      "step": 9118
    },
    {
      "epoch": 0.377926975838203,
      "grad_norm": 0.41116878390312195,
      "learning_rate": 3.1105723403373537e-06,
      "loss": 0.6334,
      "step": 9119
    },
    {
      "epoch": 0.3779684197438767,
      "grad_norm": 0.4206935167312622,
      "learning_rate": 3.110365120808985e-06,
      "loss": 0.7507,
      "step": 9120
    },
    {
      "epoch": 0.37800986364955036,
      "grad_norm": 0.5626989006996155,
      "learning_rate": 3.1101579012806173e-06,
      "loss": 0.7302,
      "step": 9121
    },
    {
      "epoch": 0.378051307555224,
      "grad_norm": 0.44472408294677734,
      "learning_rate": 3.1099506817522483e-06,
      "loss": 0.7343,
      "step": 9122
    },
    {
      "epoch": 0.37809275146089766,
      "grad_norm": 0.43195515871047974,
      "learning_rate": 3.1097434622238805e-06,
      "loss": 0.7478,
      "step": 9123
    },
    {
      "epoch": 0.37813419536657134,
      "grad_norm": 0.3992812931537628,
      "learning_rate": 3.109536242695512e-06,
      "loss": 0.699,
      "step": 9124
    },
    {
      "epoch": 0.378175639272245,
      "grad_norm": 0.3875252604484558,
      "learning_rate": 3.1093290231671437e-06,
      "loss": 0.6951,
      "step": 9125
    },
    {
      "epoch": 0.3782170831779187,
      "grad_norm": 0.43861815333366394,
      "learning_rate": 3.109121803638775e-06,
      "loss": 0.6458,
      "step": 9126
    },
    {
      "epoch": 0.37825852708359237,
      "grad_norm": 0.38638928532600403,
      "learning_rate": 3.108914584110407e-06,
      "loss": 0.6531,
      "step": 9127
    },
    {
      "epoch": 0.37829997098926604,
      "grad_norm": 0.42017653584480286,
      "learning_rate": 3.1087073645820387e-06,
      "loss": 0.7073,
      "step": 9128
    },
    {
      "epoch": 0.3783414148949397,
      "grad_norm": 0.40330740809440613,
      "learning_rate": 3.10850014505367e-06,
      "loss": 0.7257,
      "step": 9129
    },
    {
      "epoch": 0.3783828588006134,
      "grad_norm": 0.442602276802063,
      "learning_rate": 3.108292925525302e-06,
      "loss": 0.7546,
      "step": 9130
    },
    {
      "epoch": 0.378424302706287,
      "grad_norm": 0.43920981884002686,
      "learning_rate": 3.1080857059969333e-06,
      "loss": 0.6884,
      "step": 9131
    },
    {
      "epoch": 0.3784657466119607,
      "grad_norm": 0.4300270676612854,
      "learning_rate": 3.107878486468565e-06,
      "loss": 0.678,
      "step": 9132
    },
    {
      "epoch": 0.3785071905176344,
      "grad_norm": 0.40002313256263733,
      "learning_rate": 3.1076712669401965e-06,
      "loss": 0.6848,
      "step": 9133
    },
    {
      "epoch": 0.37854863442330805,
      "grad_norm": 0.4457443654537201,
      "learning_rate": 3.1074640474118283e-06,
      "loss": 0.7006,
      "step": 9134
    },
    {
      "epoch": 0.3785900783289817,
      "grad_norm": 0.4265090525150299,
      "learning_rate": 3.1072568278834597e-06,
      "loss": 0.7227,
      "step": 9135
    },
    {
      "epoch": 0.3786315222346554,
      "grad_norm": 0.39798253774642944,
      "learning_rate": 3.1070496083550915e-06,
      "loss": 0.6573,
      "step": 9136
    },
    {
      "epoch": 0.3786729661403291,
      "grad_norm": 0.42372873425483704,
      "learning_rate": 3.1068423888267237e-06,
      "loss": 0.7654,
      "step": 9137
    },
    {
      "epoch": 0.37871441004600276,
      "grad_norm": 0.42440658807754517,
      "learning_rate": 3.1066351692983547e-06,
      "loss": 0.7156,
      "step": 9138
    },
    {
      "epoch": 0.3787558539516764,
      "grad_norm": 0.4401095509529114,
      "learning_rate": 3.106427949769987e-06,
      "loss": 0.6403,
      "step": 9139
    },
    {
      "epoch": 0.37879729785735006,
      "grad_norm": 0.4060695469379425,
      "learning_rate": 3.106220730241618e-06,
      "loss": 0.6844,
      "step": 9140
    },
    {
      "epoch": 0.37883874176302373,
      "grad_norm": 0.4003825783729553,
      "learning_rate": 3.10601351071325e-06,
      "loss": 0.6892,
      "step": 9141
    },
    {
      "epoch": 0.3788801856686974,
      "grad_norm": 0.403488427400589,
      "learning_rate": 3.1058062911848815e-06,
      "loss": 0.6914,
      "step": 9142
    },
    {
      "epoch": 0.3789216295743711,
      "grad_norm": 0.40524524450302124,
      "learning_rate": 3.1055990716565133e-06,
      "loss": 0.6655,
      "step": 9143
    },
    {
      "epoch": 0.37896307348004477,
      "grad_norm": 0.42342856526374817,
      "learning_rate": 3.1053918521281447e-06,
      "loss": 0.738,
      "step": 9144
    },
    {
      "epoch": 0.37900451738571844,
      "grad_norm": 0.4484630823135376,
      "learning_rate": 3.1051846325997765e-06,
      "loss": 0.7518,
      "step": 9145
    },
    {
      "epoch": 0.3790459612913921,
      "grad_norm": 0.4479706287384033,
      "learning_rate": 3.1049774130714083e-06,
      "loss": 0.7754,
      "step": 9146
    },
    {
      "epoch": 0.3790874051970658,
      "grad_norm": 0.3792915344238281,
      "learning_rate": 3.1047701935430397e-06,
      "loss": 0.6783,
      "step": 9147
    },
    {
      "epoch": 0.3791288491027394,
      "grad_norm": 0.4135064482688904,
      "learning_rate": 3.1045629740146715e-06,
      "loss": 0.6904,
      "step": 9148
    },
    {
      "epoch": 0.3791702930084131,
      "grad_norm": 0.4051499366760254,
      "learning_rate": 3.104355754486303e-06,
      "loss": 0.7429,
      "step": 9149
    },
    {
      "epoch": 0.37921173691408677,
      "grad_norm": 0.4056798815727234,
      "learning_rate": 3.1041485349579347e-06,
      "loss": 0.6238,
      "step": 9150
    },
    {
      "epoch": 0.37925318081976045,
      "grad_norm": 0.40963971614837646,
      "learning_rate": 3.103941315429566e-06,
      "loss": 0.6687,
      "step": 9151
    },
    {
      "epoch": 0.3792946247254341,
      "grad_norm": 0.4232853651046753,
      "learning_rate": 3.103734095901198e-06,
      "loss": 0.7424,
      "step": 9152
    },
    {
      "epoch": 0.3793360686311078,
      "grad_norm": 0.44976383447647095,
      "learning_rate": 3.1035268763728297e-06,
      "loss": 0.8,
      "step": 9153
    },
    {
      "epoch": 0.3793775125367815,
      "grad_norm": 0.3789380192756653,
      "learning_rate": 3.103319656844461e-06,
      "loss": 0.688,
      "step": 9154
    },
    {
      "epoch": 0.37941895644245516,
      "grad_norm": 0.4339161813259125,
      "learning_rate": 3.1031124373160933e-06,
      "loss": 0.7151,
      "step": 9155
    },
    {
      "epoch": 0.37946040034812883,
      "grad_norm": 0.3916315734386444,
      "learning_rate": 3.1029052177877243e-06,
      "loss": 0.6511,
      "step": 9156
    },
    {
      "epoch": 0.37950184425380246,
      "grad_norm": 0.42720597982406616,
      "learning_rate": 3.1026979982593565e-06,
      "loss": 0.6603,
      "step": 9157
    },
    {
      "epoch": 0.37954328815947613,
      "grad_norm": 0.4029950797557831,
      "learning_rate": 3.102490778730988e-06,
      "loss": 0.666,
      "step": 9158
    },
    {
      "epoch": 0.3795847320651498,
      "grad_norm": 0.41230687499046326,
      "learning_rate": 3.1022835592026197e-06,
      "loss": 0.6763,
      "step": 9159
    },
    {
      "epoch": 0.3796261759708235,
      "grad_norm": 0.4137158989906311,
      "learning_rate": 3.102076339674251e-06,
      "loss": 0.6852,
      "step": 9160
    },
    {
      "epoch": 0.37966761987649716,
      "grad_norm": 0.3936541676521301,
      "learning_rate": 3.101869120145883e-06,
      "loss": 0.684,
      "step": 9161
    },
    {
      "epoch": 0.37970906378217084,
      "grad_norm": 0.4540264904499054,
      "learning_rate": 3.1016619006175147e-06,
      "loss": 0.6782,
      "step": 9162
    },
    {
      "epoch": 0.3797505076878445,
      "grad_norm": 0.4015125036239624,
      "learning_rate": 3.101454681089146e-06,
      "loss": 0.7256,
      "step": 9163
    },
    {
      "epoch": 0.3797919515935182,
      "grad_norm": 0.43519648909568787,
      "learning_rate": 3.101247461560778e-06,
      "loss": 0.7952,
      "step": 9164
    },
    {
      "epoch": 0.3798333954991918,
      "grad_norm": 0.43334752321243286,
      "learning_rate": 3.1010402420324093e-06,
      "loss": 0.7122,
      "step": 9165
    },
    {
      "epoch": 0.3798748394048655,
      "grad_norm": 0.38858523964881897,
      "learning_rate": 3.100833022504041e-06,
      "loss": 0.6384,
      "step": 9166
    },
    {
      "epoch": 0.37991628331053917,
      "grad_norm": 0.4041828215122223,
      "learning_rate": 3.1006258029756725e-06,
      "loss": 0.6899,
      "step": 9167
    },
    {
      "epoch": 0.37995772721621285,
      "grad_norm": 0.4047773778438568,
      "learning_rate": 3.1004185834473043e-06,
      "loss": 0.73,
      "step": 9168
    },
    {
      "epoch": 0.3799991711218865,
      "grad_norm": 0.41954758763313293,
      "learning_rate": 3.1002113639189357e-06,
      "loss": 0.7112,
      "step": 9169
    },
    {
      "epoch": 0.3800406150275602,
      "grad_norm": 0.40262526273727417,
      "learning_rate": 3.1000041443905675e-06,
      "loss": 0.7061,
      "step": 9170
    },
    {
      "epoch": 0.3800820589332339,
      "grad_norm": 0.4172843098640442,
      "learning_rate": 3.0997969248621997e-06,
      "loss": 0.6473,
      "step": 9171
    },
    {
      "epoch": 0.38012350283890756,
      "grad_norm": 0.39049574732780457,
      "learning_rate": 3.0995897053338307e-06,
      "loss": 0.6387,
      "step": 9172
    },
    {
      "epoch": 0.38016494674458123,
      "grad_norm": 0.4358871877193451,
      "learning_rate": 3.099382485805463e-06,
      "loss": 0.6462,
      "step": 9173
    },
    {
      "epoch": 0.38020639065025486,
      "grad_norm": 0.41856205463409424,
      "learning_rate": 3.099175266277094e-06,
      "loss": 0.7256,
      "step": 9174
    },
    {
      "epoch": 0.38024783455592853,
      "grad_norm": 0.4508495330810547,
      "learning_rate": 3.098968046748726e-06,
      "loss": 0.7565,
      "step": 9175
    },
    {
      "epoch": 0.3802892784616022,
      "grad_norm": 0.4072088599205017,
      "learning_rate": 3.0987608272203575e-06,
      "loss": 0.6831,
      "step": 9176
    },
    {
      "epoch": 0.3803307223672759,
      "grad_norm": 0.437509149312973,
      "learning_rate": 3.0985536076919893e-06,
      "loss": 0.7272,
      "step": 9177
    },
    {
      "epoch": 0.38037216627294956,
      "grad_norm": 0.3919816017150879,
      "learning_rate": 3.0983463881636207e-06,
      "loss": 0.658,
      "step": 9178
    },
    {
      "epoch": 0.38041361017862324,
      "grad_norm": 0.41366225481033325,
      "learning_rate": 3.0981391686352525e-06,
      "loss": 0.6792,
      "step": 9179
    },
    {
      "epoch": 0.3804550540842969,
      "grad_norm": 0.3940708637237549,
      "learning_rate": 3.0979319491068843e-06,
      "loss": 0.6787,
      "step": 9180
    },
    {
      "epoch": 0.3804964979899706,
      "grad_norm": 0.37057119607925415,
      "learning_rate": 3.0977247295785157e-06,
      "loss": 0.6355,
      "step": 9181
    },
    {
      "epoch": 0.38053794189564427,
      "grad_norm": 0.4065597653388977,
      "learning_rate": 3.0975175100501475e-06,
      "loss": 0.6897,
      "step": 9182
    },
    {
      "epoch": 0.3805793858013179,
      "grad_norm": 0.41408172249794006,
      "learning_rate": 3.097310290521779e-06,
      "loss": 0.7371,
      "step": 9183
    },
    {
      "epoch": 0.38062082970699157,
      "grad_norm": 0.4121535122394562,
      "learning_rate": 3.0971030709934107e-06,
      "loss": 0.6934,
      "step": 9184
    },
    {
      "epoch": 0.38066227361266525,
      "grad_norm": 0.39426106214523315,
      "learning_rate": 3.096895851465042e-06,
      "loss": 0.6777,
      "step": 9185
    },
    {
      "epoch": 0.3807037175183389,
      "grad_norm": 0.398344486951828,
      "learning_rate": 3.096688631936674e-06,
      "loss": 0.7153,
      "step": 9186
    },
    {
      "epoch": 0.3807451614240126,
      "grad_norm": 0.4373199939727783,
      "learning_rate": 3.0964814124083053e-06,
      "loss": 0.7561,
      "step": 9187
    },
    {
      "epoch": 0.3807866053296863,
      "grad_norm": 0.38807064294815063,
      "learning_rate": 3.096274192879937e-06,
      "loss": 0.7073,
      "step": 9188
    },
    {
      "epoch": 0.38082804923535996,
      "grad_norm": 0.41266536712646484,
      "learning_rate": 3.0960669733515693e-06,
      "loss": 0.6698,
      "step": 9189
    },
    {
      "epoch": 0.38086949314103363,
      "grad_norm": 0.451168417930603,
      "learning_rate": 3.0958597538232003e-06,
      "loss": 0.7844,
      "step": 9190
    },
    {
      "epoch": 0.38091093704670725,
      "grad_norm": 0.41523683071136475,
      "learning_rate": 3.0956525342948325e-06,
      "loss": 0.6826,
      "step": 9191
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.4772147536277771,
      "learning_rate": 3.095445314766464e-06,
      "loss": 0.723,
      "step": 9192
    },
    {
      "epoch": 0.3809938248580546,
      "grad_norm": 0.3959874212741852,
      "learning_rate": 3.0952380952380957e-06,
      "loss": 0.6663,
      "step": 9193
    },
    {
      "epoch": 0.3810352687637283,
      "grad_norm": 0.4092823266983032,
      "learning_rate": 3.095030875709727e-06,
      "loss": 0.674,
      "step": 9194
    },
    {
      "epoch": 0.38107671266940196,
      "grad_norm": 0.38881716132164,
      "learning_rate": 3.094823656181359e-06,
      "loss": 0.6752,
      "step": 9195
    },
    {
      "epoch": 0.38111815657507564,
      "grad_norm": 0.41186651587486267,
      "learning_rate": 3.0946164366529903e-06,
      "loss": 0.6913,
      "step": 9196
    },
    {
      "epoch": 0.3811596004807493,
      "grad_norm": 0.38933441042900085,
      "learning_rate": 3.094409217124622e-06,
      "loss": 0.6606,
      "step": 9197
    },
    {
      "epoch": 0.381201044386423,
      "grad_norm": 0.46014562249183655,
      "learning_rate": 3.094201997596254e-06,
      "loss": 0.7891,
      "step": 9198
    },
    {
      "epoch": 0.38124248829209667,
      "grad_norm": 0.43342795968055725,
      "learning_rate": 3.0939947780678853e-06,
      "loss": 0.7053,
      "step": 9199
    },
    {
      "epoch": 0.3812839321977703,
      "grad_norm": 0.3876701593399048,
      "learning_rate": 3.093787558539517e-06,
      "loss": 0.717,
      "step": 9200
    },
    {
      "epoch": 0.38132537610344397,
      "grad_norm": 0.42016223073005676,
      "learning_rate": 3.0935803390111485e-06,
      "loss": 0.6826,
      "step": 9201
    },
    {
      "epoch": 0.38136682000911765,
      "grad_norm": 0.3845817446708679,
      "learning_rate": 3.0933731194827803e-06,
      "loss": 0.6818,
      "step": 9202
    },
    {
      "epoch": 0.3814082639147913,
      "grad_norm": 0.42730796337127686,
      "learning_rate": 3.0931658999544117e-06,
      "loss": 0.6796,
      "step": 9203
    },
    {
      "epoch": 0.381449707820465,
      "grad_norm": 0.39858508110046387,
      "learning_rate": 3.0929586804260435e-06,
      "loss": 0.6812,
      "step": 9204
    },
    {
      "epoch": 0.3814911517261387,
      "grad_norm": 0.41656821966171265,
      "learning_rate": 3.092751460897675e-06,
      "loss": 0.6733,
      "step": 9205
    },
    {
      "epoch": 0.38153259563181235,
      "grad_norm": 0.4087032377719879,
      "learning_rate": 3.0925442413693067e-06,
      "loss": 0.6782,
      "step": 9206
    },
    {
      "epoch": 0.38157403953748603,
      "grad_norm": 0.4199017286300659,
      "learning_rate": 3.092337021840939e-06,
      "loss": 0.6802,
      "step": 9207
    },
    {
      "epoch": 0.3816154834431597,
      "grad_norm": 0.4345358908176422,
      "learning_rate": 3.09212980231257e-06,
      "loss": 0.7092,
      "step": 9208
    },
    {
      "epoch": 0.38165692734883333,
      "grad_norm": 0.4565702974796295,
      "learning_rate": 3.091922582784202e-06,
      "loss": 0.6724,
      "step": 9209
    },
    {
      "epoch": 0.381698371254507,
      "grad_norm": 0.40829774737358093,
      "learning_rate": 3.0917153632558335e-06,
      "loss": 0.7113,
      "step": 9210
    },
    {
      "epoch": 0.3817398151601807,
      "grad_norm": 0.3997948467731476,
      "learning_rate": 3.0915081437274653e-06,
      "loss": 0.6882,
      "step": 9211
    },
    {
      "epoch": 0.38178125906585436,
      "grad_norm": 0.39756909012794495,
      "learning_rate": 3.0913009241990967e-06,
      "loss": 0.6965,
      "step": 9212
    },
    {
      "epoch": 0.38182270297152804,
      "grad_norm": 0.42741507291793823,
      "learning_rate": 3.0910937046707285e-06,
      "loss": 0.7021,
      "step": 9213
    },
    {
      "epoch": 0.3818641468772017,
      "grad_norm": 0.41837838292121887,
      "learning_rate": 3.0908864851423603e-06,
      "loss": 0.6982,
      "step": 9214
    },
    {
      "epoch": 0.3819055907828754,
      "grad_norm": 0.3680271506309509,
      "learning_rate": 3.0906792656139917e-06,
      "loss": 0.6803,
      "step": 9215
    },
    {
      "epoch": 0.38194703468854907,
      "grad_norm": 0.47932693362236023,
      "learning_rate": 3.0904720460856235e-06,
      "loss": 0.7446,
      "step": 9216
    },
    {
      "epoch": 0.38198847859422275,
      "grad_norm": 0.5347210764884949,
      "learning_rate": 3.090264826557255e-06,
      "loss": 0.7231,
      "step": 9217
    },
    {
      "epoch": 0.38202992249989637,
      "grad_norm": 0.39970850944519043,
      "learning_rate": 3.0900576070288867e-06,
      "loss": 0.7344,
      "step": 9218
    },
    {
      "epoch": 0.38207136640557005,
      "grad_norm": 0.4219069182872772,
      "learning_rate": 3.089850387500518e-06,
      "loss": 0.6705,
      "step": 9219
    },
    {
      "epoch": 0.3821128103112437,
      "grad_norm": 0.40626099705696106,
      "learning_rate": 3.08964316797215e-06,
      "loss": 0.7156,
      "step": 9220
    },
    {
      "epoch": 0.3821542542169174,
      "grad_norm": 0.37770986557006836,
      "learning_rate": 3.0894359484437813e-06,
      "loss": 0.6879,
      "step": 9221
    },
    {
      "epoch": 0.3821956981225911,
      "grad_norm": 0.36803457140922546,
      "learning_rate": 3.089228728915413e-06,
      "loss": 0.6553,
      "step": 9222
    },
    {
      "epoch": 0.38223714202826475,
      "grad_norm": 0.423929363489151,
      "learning_rate": 3.0890215093870453e-06,
      "loss": 0.6608,
      "step": 9223
    },
    {
      "epoch": 0.38227858593393843,
      "grad_norm": 0.3884040415287018,
      "learning_rate": 3.0888142898586763e-06,
      "loss": 0.7126,
      "step": 9224
    },
    {
      "epoch": 0.3823200298396121,
      "grad_norm": 0.4234277307987213,
      "learning_rate": 3.0886070703303085e-06,
      "loss": 0.6619,
      "step": 9225
    },
    {
      "epoch": 0.38236147374528573,
      "grad_norm": 0.4217928946018219,
      "learning_rate": 3.08839985080194e-06,
      "loss": 0.7173,
      "step": 9226
    },
    {
      "epoch": 0.3824029176509594,
      "grad_norm": 0.40673696994781494,
      "learning_rate": 3.0881926312735717e-06,
      "loss": 0.6736,
      "step": 9227
    },
    {
      "epoch": 0.3824443615566331,
      "grad_norm": 0.41197019815444946,
      "learning_rate": 3.087985411745203e-06,
      "loss": 0.6749,
      "step": 9228
    },
    {
      "epoch": 0.38248580546230676,
      "grad_norm": 0.4270392954349518,
      "learning_rate": 3.087778192216835e-06,
      "loss": 0.7128,
      "step": 9229
    },
    {
      "epoch": 0.38252724936798044,
      "grad_norm": 0.409603476524353,
      "learning_rate": 3.0875709726884663e-06,
      "loss": 0.693,
      "step": 9230
    },
    {
      "epoch": 0.3825686932736541,
      "grad_norm": 0.44367167353630066,
      "learning_rate": 3.087363753160098e-06,
      "loss": 0.7283,
      "step": 9231
    },
    {
      "epoch": 0.3826101371793278,
      "grad_norm": 0.39005792140960693,
      "learning_rate": 3.08715653363173e-06,
      "loss": 0.6313,
      "step": 9232
    },
    {
      "epoch": 0.38265158108500147,
      "grad_norm": 0.43114331364631653,
      "learning_rate": 3.0869493141033613e-06,
      "loss": 0.6875,
      "step": 9233
    },
    {
      "epoch": 0.38269302499067515,
      "grad_norm": 0.42688754200935364,
      "learning_rate": 3.086742094574993e-06,
      "loss": 0.7119,
      "step": 9234
    },
    {
      "epoch": 0.38273446889634877,
      "grad_norm": 0.4607378840446472,
      "learning_rate": 3.0865348750466245e-06,
      "loss": 0.7593,
      "step": 9235
    },
    {
      "epoch": 0.38277591280202244,
      "grad_norm": 0.4205532670021057,
      "learning_rate": 3.0863276555182563e-06,
      "loss": 0.7437,
      "step": 9236
    },
    {
      "epoch": 0.3828173567076961,
      "grad_norm": 0.3958719074726105,
      "learning_rate": 3.0861204359898877e-06,
      "loss": 0.7041,
      "step": 9237
    },
    {
      "epoch": 0.3828588006133698,
      "grad_norm": 0.41612759232521057,
      "learning_rate": 3.0859132164615195e-06,
      "loss": 0.6743,
      "step": 9238
    },
    {
      "epoch": 0.3829002445190435,
      "grad_norm": 0.4589601755142212,
      "learning_rate": 3.085705996933151e-06,
      "loss": 0.7297,
      "step": 9239
    },
    {
      "epoch": 0.38294168842471715,
      "grad_norm": 0.45214006304740906,
      "learning_rate": 3.0854987774047827e-06,
      "loss": 0.7485,
      "step": 9240
    },
    {
      "epoch": 0.38298313233039083,
      "grad_norm": 0.40532243251800537,
      "learning_rate": 3.085291557876415e-06,
      "loss": 0.6909,
      "step": 9241
    },
    {
      "epoch": 0.3830245762360645,
      "grad_norm": 0.3932921588420868,
      "learning_rate": 3.085084338348046e-06,
      "loss": 0.687,
      "step": 9242
    },
    {
      "epoch": 0.3830660201417382,
      "grad_norm": 0.40643736720085144,
      "learning_rate": 3.084877118819678e-06,
      "loss": 0.6965,
      "step": 9243
    },
    {
      "epoch": 0.3831074640474118,
      "grad_norm": 0.449027419090271,
      "learning_rate": 3.0846698992913095e-06,
      "loss": 0.6666,
      "step": 9244
    },
    {
      "epoch": 0.3831489079530855,
      "grad_norm": 0.3998550474643707,
      "learning_rate": 3.0844626797629413e-06,
      "loss": 0.701,
      "step": 9245
    },
    {
      "epoch": 0.38319035185875916,
      "grad_norm": 0.41012856364250183,
      "learning_rate": 3.0842554602345727e-06,
      "loss": 0.6414,
      "step": 9246
    },
    {
      "epoch": 0.38323179576443284,
      "grad_norm": 0.41098618507385254,
      "learning_rate": 3.0840482407062045e-06,
      "loss": 0.6624,
      "step": 9247
    },
    {
      "epoch": 0.3832732396701065,
      "grad_norm": 0.46740084886550903,
      "learning_rate": 3.083841021177836e-06,
      "loss": 0.7201,
      "step": 9248
    },
    {
      "epoch": 0.3833146835757802,
      "grad_norm": 0.4086371064186096,
      "learning_rate": 3.0836338016494677e-06,
      "loss": 0.666,
      "step": 9249
    },
    {
      "epoch": 0.38335612748145387,
      "grad_norm": 0.3859868049621582,
      "learning_rate": 3.0834265821210995e-06,
      "loss": 0.6307,
      "step": 9250
    },
    {
      "epoch": 0.38339757138712754,
      "grad_norm": 0.4064895808696747,
      "learning_rate": 3.083219362592731e-06,
      "loss": 0.7031,
      "step": 9251
    },
    {
      "epoch": 0.38343901529280117,
      "grad_norm": 0.39568623900413513,
      "learning_rate": 3.0830121430643627e-06,
      "loss": 0.6993,
      "step": 9252
    },
    {
      "epoch": 0.38348045919847484,
      "grad_norm": 0.3965052664279938,
      "learning_rate": 3.082804923535994e-06,
      "loss": 0.6677,
      "step": 9253
    },
    {
      "epoch": 0.3835219031041485,
      "grad_norm": 0.3991289734840393,
      "learning_rate": 3.082597704007626e-06,
      "loss": 0.6921,
      "step": 9254
    },
    {
      "epoch": 0.3835633470098222,
      "grad_norm": 0.4161168038845062,
      "learning_rate": 3.0823904844792573e-06,
      "loss": 0.6987,
      "step": 9255
    },
    {
      "epoch": 0.3836047909154959,
      "grad_norm": 0.3837801218032837,
      "learning_rate": 3.082183264950889e-06,
      "loss": 0.7041,
      "step": 9256
    },
    {
      "epoch": 0.38364623482116955,
      "grad_norm": 0.3896898627281189,
      "learning_rate": 3.0819760454225205e-06,
      "loss": 0.7351,
      "step": 9257
    },
    {
      "epoch": 0.38368767872684323,
      "grad_norm": 0.3969760239124298,
      "learning_rate": 3.0817688258941523e-06,
      "loss": 0.6835,
      "step": 9258
    },
    {
      "epoch": 0.3837291226325169,
      "grad_norm": 0.4045853614807129,
      "learning_rate": 3.0815616063657845e-06,
      "loss": 0.666,
      "step": 9259
    },
    {
      "epoch": 0.3837705665381906,
      "grad_norm": 0.4116005599498749,
      "learning_rate": 3.081354386837416e-06,
      "loss": 0.6964,
      "step": 9260
    },
    {
      "epoch": 0.3838120104438642,
      "grad_norm": 0.3917616009712219,
      "learning_rate": 3.0811471673090477e-06,
      "loss": 0.6885,
      "step": 9261
    },
    {
      "epoch": 0.3838534543495379,
      "grad_norm": 0.41883930563926697,
      "learning_rate": 3.080939947780679e-06,
      "loss": 0.6703,
      "step": 9262
    },
    {
      "epoch": 0.38389489825521156,
      "grad_norm": 0.3914545178413391,
      "learning_rate": 3.080732728252311e-06,
      "loss": 0.7054,
      "step": 9263
    },
    {
      "epoch": 0.38393634216088524,
      "grad_norm": 0.4153364896774292,
      "learning_rate": 3.0805255087239423e-06,
      "loss": 0.7294,
      "step": 9264
    },
    {
      "epoch": 0.3839777860665589,
      "grad_norm": 0.4503365159034729,
      "learning_rate": 3.080318289195574e-06,
      "loss": 0.7317,
      "step": 9265
    },
    {
      "epoch": 0.3840192299722326,
      "grad_norm": 0.38273900747299194,
      "learning_rate": 3.080111069667206e-06,
      "loss": 0.7111,
      "step": 9266
    },
    {
      "epoch": 0.38406067387790627,
      "grad_norm": 0.39001408219337463,
      "learning_rate": 3.0799038501388373e-06,
      "loss": 0.7439,
      "step": 9267
    },
    {
      "epoch": 0.38410211778357994,
      "grad_norm": 0.4026496112346649,
      "learning_rate": 3.079696630610469e-06,
      "loss": 0.708,
      "step": 9268
    },
    {
      "epoch": 0.3841435616892536,
      "grad_norm": 0.42652666568756104,
      "learning_rate": 3.0794894110821005e-06,
      "loss": 0.7251,
      "step": 9269
    },
    {
      "epoch": 0.38418500559492724,
      "grad_norm": 0.40743565559387207,
      "learning_rate": 3.0792821915537323e-06,
      "loss": 0.7258,
      "step": 9270
    },
    {
      "epoch": 0.3842264495006009,
      "grad_norm": 0.42058518528938293,
      "learning_rate": 3.0790749720253637e-06,
      "loss": 0.6884,
      "step": 9271
    },
    {
      "epoch": 0.3842678934062746,
      "grad_norm": 0.4118545949459076,
      "learning_rate": 3.0788677524969955e-06,
      "loss": 0.6855,
      "step": 9272
    },
    {
      "epoch": 0.3843093373119483,
      "grad_norm": 0.4530578553676605,
      "learning_rate": 3.078660532968627e-06,
      "loss": 0.6875,
      "step": 9273
    },
    {
      "epoch": 0.38435078121762195,
      "grad_norm": 0.42843106389045715,
      "learning_rate": 3.0784533134402587e-06,
      "loss": 0.7013,
      "step": 9274
    },
    {
      "epoch": 0.3843922251232956,
      "grad_norm": 0.45375871658325195,
      "learning_rate": 3.078246093911891e-06,
      "loss": 0.7146,
      "step": 9275
    },
    {
      "epoch": 0.3844336690289693,
      "grad_norm": 0.39169248938560486,
      "learning_rate": 3.0780388743835223e-06,
      "loss": 0.674,
      "step": 9276
    },
    {
      "epoch": 0.384475112934643,
      "grad_norm": 0.47202402353286743,
      "learning_rate": 3.077831654855154e-06,
      "loss": 0.729,
      "step": 9277
    },
    {
      "epoch": 0.38451655684031666,
      "grad_norm": 0.47285759449005127,
      "learning_rate": 3.0776244353267855e-06,
      "loss": 0.7554,
      "step": 9278
    },
    {
      "epoch": 0.3845580007459903,
      "grad_norm": 0.39169469475746155,
      "learning_rate": 3.0774172157984173e-06,
      "loss": 0.6794,
      "step": 9279
    },
    {
      "epoch": 0.38459944465166396,
      "grad_norm": 0.41335371136665344,
      "learning_rate": 3.0772099962700487e-06,
      "loss": 0.7026,
      "step": 9280
    },
    {
      "epoch": 0.38464088855733763,
      "grad_norm": 0.5122267603874207,
      "learning_rate": 3.0770027767416805e-06,
      "loss": 0.7996,
      "step": 9281
    },
    {
      "epoch": 0.3846823324630113,
      "grad_norm": 0.42029279470443726,
      "learning_rate": 3.076795557213312e-06,
      "loss": 0.7041,
      "step": 9282
    },
    {
      "epoch": 0.384723776368685,
      "grad_norm": 0.38627052307128906,
      "learning_rate": 3.0765883376849437e-06,
      "loss": 0.718,
      "step": 9283
    },
    {
      "epoch": 0.38476522027435867,
      "grad_norm": 0.4266931712627411,
      "learning_rate": 3.0763811181565755e-06,
      "loss": 0.6802,
      "step": 9284
    },
    {
      "epoch": 0.38480666418003234,
      "grad_norm": 0.4115358293056488,
      "learning_rate": 3.076173898628207e-06,
      "loss": 0.6675,
      "step": 9285
    },
    {
      "epoch": 0.384848108085706,
      "grad_norm": 0.42079469561576843,
      "learning_rate": 3.0759666790998387e-06,
      "loss": 0.6609,
      "step": 9286
    },
    {
      "epoch": 0.38488955199137964,
      "grad_norm": 0.4105003774166107,
      "learning_rate": 3.07575945957147e-06,
      "loss": 0.6641,
      "step": 9287
    },
    {
      "epoch": 0.3849309958970533,
      "grad_norm": 0.44400209188461304,
      "learning_rate": 3.075552240043102e-06,
      "loss": 0.6768,
      "step": 9288
    },
    {
      "epoch": 0.384972439802727,
      "grad_norm": 0.40054166316986084,
      "learning_rate": 3.0753450205147333e-06,
      "loss": 0.6992,
      "step": 9289
    },
    {
      "epoch": 0.3850138837084007,
      "grad_norm": 0.3749813139438629,
      "learning_rate": 3.075137800986365e-06,
      "loss": 0.686,
      "step": 9290
    },
    {
      "epoch": 0.38505532761407435,
      "grad_norm": 0.4300565719604492,
      "learning_rate": 3.0749305814579965e-06,
      "loss": 0.7028,
      "step": 9291
    },
    {
      "epoch": 0.385096771519748,
      "grad_norm": 0.4431581497192383,
      "learning_rate": 3.0747233619296283e-06,
      "loss": 0.7161,
      "step": 9292
    },
    {
      "epoch": 0.3851382154254217,
      "grad_norm": 0.47178760170936584,
      "learning_rate": 3.0745161424012605e-06,
      "loss": 0.6978,
      "step": 9293
    },
    {
      "epoch": 0.3851796593310954,
      "grad_norm": 0.39275801181793213,
      "learning_rate": 3.074308922872892e-06,
      "loss": 0.7205,
      "step": 9294
    },
    {
      "epoch": 0.38522110323676906,
      "grad_norm": 0.41197440028190613,
      "learning_rate": 3.0741017033445237e-06,
      "loss": 0.6871,
      "step": 9295
    },
    {
      "epoch": 0.3852625471424427,
      "grad_norm": 0.3886524438858032,
      "learning_rate": 3.073894483816155e-06,
      "loss": 0.7612,
      "step": 9296
    },
    {
      "epoch": 0.38530399104811636,
      "grad_norm": 0.44530779123306274,
      "learning_rate": 3.073687264287787e-06,
      "loss": 0.7716,
      "step": 9297
    },
    {
      "epoch": 0.38534543495379003,
      "grad_norm": 0.3905171751976013,
      "learning_rate": 3.0734800447594183e-06,
      "loss": 0.6514,
      "step": 9298
    },
    {
      "epoch": 0.3853868788594637,
      "grad_norm": 0.4377938508987427,
      "learning_rate": 3.07327282523105e-06,
      "loss": 0.7163,
      "step": 9299
    },
    {
      "epoch": 0.3854283227651374,
      "grad_norm": 0.4164804220199585,
      "learning_rate": 3.0730656057026815e-06,
      "loss": 0.6943,
      "step": 9300
    },
    {
      "epoch": 0.38546976667081106,
      "grad_norm": 0.3963632583618164,
      "learning_rate": 3.0728583861743133e-06,
      "loss": 0.7355,
      "step": 9301
    },
    {
      "epoch": 0.38551121057648474,
      "grad_norm": 0.42046797275543213,
      "learning_rate": 3.072651166645945e-06,
      "loss": 0.6958,
      "step": 9302
    },
    {
      "epoch": 0.3855526544821584,
      "grad_norm": 0.4171767830848694,
      "learning_rate": 3.0724439471175765e-06,
      "loss": 0.6887,
      "step": 9303
    },
    {
      "epoch": 0.3855940983878321,
      "grad_norm": 0.39495649933815,
      "learning_rate": 3.0722367275892083e-06,
      "loss": 0.7014,
      "step": 9304
    },
    {
      "epoch": 0.3856355422935057,
      "grad_norm": 0.41342851519584656,
      "learning_rate": 3.0720295080608397e-06,
      "loss": 0.688,
      "step": 9305
    },
    {
      "epoch": 0.3856769861991794,
      "grad_norm": 0.3821893036365509,
      "learning_rate": 3.0718222885324715e-06,
      "loss": 0.681,
      "step": 9306
    },
    {
      "epoch": 0.38571843010485307,
      "grad_norm": 0.4187532663345337,
      "learning_rate": 3.071615069004103e-06,
      "loss": 0.6915,
      "step": 9307
    },
    {
      "epoch": 0.38575987401052675,
      "grad_norm": 0.40716350078582764,
      "learning_rate": 3.0714078494757347e-06,
      "loss": 0.6821,
      "step": 9308
    },
    {
      "epoch": 0.3858013179162004,
      "grad_norm": 0.3995026648044586,
      "learning_rate": 3.071200629947366e-06,
      "loss": 0.6769,
      "step": 9309
    },
    {
      "epoch": 0.3858427618218741,
      "grad_norm": 0.39887571334838867,
      "learning_rate": 3.0709934104189983e-06,
      "loss": 0.6995,
      "step": 9310
    },
    {
      "epoch": 0.3858842057275478,
      "grad_norm": 0.4234156608581543,
      "learning_rate": 3.07078619089063e-06,
      "loss": 0.7073,
      "step": 9311
    },
    {
      "epoch": 0.38592564963322146,
      "grad_norm": 0.42823249101638794,
      "learning_rate": 3.0705789713622615e-06,
      "loss": 0.7744,
      "step": 9312
    },
    {
      "epoch": 0.3859670935388951,
      "grad_norm": 0.4211607575416565,
      "learning_rate": 3.0703717518338933e-06,
      "loss": 0.7649,
      "step": 9313
    },
    {
      "epoch": 0.38600853744456876,
      "grad_norm": 0.4031028747558594,
      "learning_rate": 3.0701645323055247e-06,
      "loss": 0.6672,
      "step": 9314
    },
    {
      "epoch": 0.38604998135024243,
      "grad_norm": 0.43502265214920044,
      "learning_rate": 3.0699573127771565e-06,
      "loss": 0.7006,
      "step": 9315
    },
    {
      "epoch": 0.3860914252559161,
      "grad_norm": 0.38088634610176086,
      "learning_rate": 3.069750093248788e-06,
      "loss": 0.7067,
      "step": 9316
    },
    {
      "epoch": 0.3861328691615898,
      "grad_norm": 0.3934900760650635,
      "learning_rate": 3.0695428737204197e-06,
      "loss": 0.7119,
      "step": 9317
    },
    {
      "epoch": 0.38617431306726346,
      "grad_norm": 0.42896568775177,
      "learning_rate": 3.069335654192051e-06,
      "loss": 0.7524,
      "step": 9318
    },
    {
      "epoch": 0.38621575697293714,
      "grad_norm": 0.4162254333496094,
      "learning_rate": 3.069128434663683e-06,
      "loss": 0.7139,
      "step": 9319
    },
    {
      "epoch": 0.3862572008786108,
      "grad_norm": 0.41852423548698425,
      "learning_rate": 3.0689212151353147e-06,
      "loss": 0.7051,
      "step": 9320
    },
    {
      "epoch": 0.3862986447842845,
      "grad_norm": 0.4078312814235687,
      "learning_rate": 3.068713995606946e-06,
      "loss": 0.757,
      "step": 9321
    },
    {
      "epoch": 0.3863400886899581,
      "grad_norm": 0.4128841161727905,
      "learning_rate": 3.068506776078578e-06,
      "loss": 0.7336,
      "step": 9322
    },
    {
      "epoch": 0.3863815325956318,
      "grad_norm": 0.4214540719985962,
      "learning_rate": 3.0682995565502093e-06,
      "loss": 0.697,
      "step": 9323
    },
    {
      "epoch": 0.38642297650130547,
      "grad_norm": 0.5146307349205017,
      "learning_rate": 3.068092337021841e-06,
      "loss": 0.6804,
      "step": 9324
    },
    {
      "epoch": 0.38646442040697915,
      "grad_norm": 0.40413782000541687,
      "learning_rate": 3.0678851174934725e-06,
      "loss": 0.686,
      "step": 9325
    },
    {
      "epoch": 0.3865058643126528,
      "grad_norm": 0.4414723217487335,
      "learning_rate": 3.0676778979651043e-06,
      "loss": 0.7346,
      "step": 9326
    },
    {
      "epoch": 0.3865473082183265,
      "grad_norm": 0.41432738304138184,
      "learning_rate": 3.0674706784367365e-06,
      "loss": 0.7241,
      "step": 9327
    },
    {
      "epoch": 0.3865887521240002,
      "grad_norm": 0.42227959632873535,
      "learning_rate": 3.067263458908368e-06,
      "loss": 0.7246,
      "step": 9328
    },
    {
      "epoch": 0.38663019602967386,
      "grad_norm": 0.4370814859867096,
      "learning_rate": 3.0670562393799997e-06,
      "loss": 0.7522,
      "step": 9329
    },
    {
      "epoch": 0.38667163993534753,
      "grad_norm": 0.4009360074996948,
      "learning_rate": 3.066849019851631e-06,
      "loss": 0.672,
      "step": 9330
    },
    {
      "epoch": 0.38671308384102115,
      "grad_norm": 0.3789265751838684,
      "learning_rate": 3.066641800323263e-06,
      "loss": 0.6594,
      "step": 9331
    },
    {
      "epoch": 0.38675452774669483,
      "grad_norm": 0.43974459171295166,
      "learning_rate": 3.0664345807948943e-06,
      "loss": 0.7202,
      "step": 9332
    },
    {
      "epoch": 0.3867959716523685,
      "grad_norm": 0.41657185554504395,
      "learning_rate": 3.066227361266526e-06,
      "loss": 0.7183,
      "step": 9333
    },
    {
      "epoch": 0.3868374155580422,
      "grad_norm": 0.4277246296405792,
      "learning_rate": 3.0660201417381575e-06,
      "loss": 0.6992,
      "step": 9334
    },
    {
      "epoch": 0.38687885946371586,
      "grad_norm": 0.39929816126823425,
      "learning_rate": 3.0658129222097893e-06,
      "loss": 0.6129,
      "step": 9335
    },
    {
      "epoch": 0.38692030336938954,
      "grad_norm": 0.3911859691143036,
      "learning_rate": 3.065605702681421e-06,
      "loss": 0.6646,
      "step": 9336
    },
    {
      "epoch": 0.3869617472750632,
      "grad_norm": 0.42812100052833557,
      "learning_rate": 3.0653984831530525e-06,
      "loss": 0.7207,
      "step": 9337
    },
    {
      "epoch": 0.3870031911807369,
      "grad_norm": 0.4009913504123688,
      "learning_rate": 3.0651912636246843e-06,
      "loss": 0.6772,
      "step": 9338
    },
    {
      "epoch": 0.3870446350864105,
      "grad_norm": 0.4087466895580292,
      "learning_rate": 3.0649840440963157e-06,
      "loss": 0.6508,
      "step": 9339
    },
    {
      "epoch": 0.3870860789920842,
      "grad_norm": 0.40619394183158875,
      "learning_rate": 3.0647768245679475e-06,
      "loss": 0.6945,
      "step": 9340
    },
    {
      "epoch": 0.38712752289775787,
      "grad_norm": 0.43778157234191895,
      "learning_rate": 3.064569605039579e-06,
      "loss": 0.651,
      "step": 9341
    },
    {
      "epoch": 0.38716896680343155,
      "grad_norm": 0.45212897658348083,
      "learning_rate": 3.0643623855112107e-06,
      "loss": 0.7206,
      "step": 9342
    },
    {
      "epoch": 0.3872104107091052,
      "grad_norm": 0.3748480975627899,
      "learning_rate": 3.064155165982842e-06,
      "loss": 0.6938,
      "step": 9343
    },
    {
      "epoch": 0.3872518546147789,
      "grad_norm": 0.4283382296562195,
      "learning_rate": 3.0639479464544743e-06,
      "loss": 0.6681,
      "step": 9344
    },
    {
      "epoch": 0.3872932985204526,
      "grad_norm": 0.431734561920166,
      "learning_rate": 3.063740726926106e-06,
      "loss": 0.6798,
      "step": 9345
    },
    {
      "epoch": 0.38733474242612626,
      "grad_norm": 0.3874494731426239,
      "learning_rate": 3.0635335073977375e-06,
      "loss": 0.6772,
      "step": 9346
    },
    {
      "epoch": 0.38737618633179993,
      "grad_norm": 0.4058365523815155,
      "learning_rate": 3.0633262878693693e-06,
      "loss": 0.6943,
      "step": 9347
    },
    {
      "epoch": 0.38741763023747355,
      "grad_norm": 0.43409669399261475,
      "learning_rate": 3.0631190683410007e-06,
      "loss": 0.6814,
      "step": 9348
    },
    {
      "epoch": 0.38745907414314723,
      "grad_norm": 0.44051212072372437,
      "learning_rate": 3.0629118488126325e-06,
      "loss": 0.7214,
      "step": 9349
    },
    {
      "epoch": 0.3875005180488209,
      "grad_norm": 0.4288006126880646,
      "learning_rate": 3.062704629284264e-06,
      "loss": 0.6978,
      "step": 9350
    },
    {
      "epoch": 0.3875419619544946,
      "grad_norm": 0.4143591523170471,
      "learning_rate": 3.0624974097558957e-06,
      "loss": 0.6953,
      "step": 9351
    },
    {
      "epoch": 0.38758340586016826,
      "grad_norm": 0.3979547619819641,
      "learning_rate": 3.062290190227527e-06,
      "loss": 0.6562,
      "step": 9352
    },
    {
      "epoch": 0.38762484976584194,
      "grad_norm": 0.40608513355255127,
      "learning_rate": 3.062082970699159e-06,
      "loss": 0.6945,
      "step": 9353
    },
    {
      "epoch": 0.3876662936715156,
      "grad_norm": 0.4167138338088989,
      "learning_rate": 3.0618757511707907e-06,
      "loss": 0.65,
      "step": 9354
    },
    {
      "epoch": 0.3877077375771893,
      "grad_norm": 0.4133530855178833,
      "learning_rate": 3.061668531642422e-06,
      "loss": 0.6798,
      "step": 9355
    },
    {
      "epoch": 0.38774918148286297,
      "grad_norm": 0.43597152829170227,
      "learning_rate": 3.061461312114054e-06,
      "loss": 0.6968,
      "step": 9356
    },
    {
      "epoch": 0.3877906253885366,
      "grad_norm": 0.4552972912788391,
      "learning_rate": 3.0612540925856853e-06,
      "loss": 0.7023,
      "step": 9357
    },
    {
      "epoch": 0.38783206929421027,
      "grad_norm": 0.45744043588638306,
      "learning_rate": 3.061046873057317e-06,
      "loss": 0.7249,
      "step": 9358
    },
    {
      "epoch": 0.38787351319988395,
      "grad_norm": 0.38768458366394043,
      "learning_rate": 3.0608396535289485e-06,
      "loss": 0.6654,
      "step": 9359
    },
    {
      "epoch": 0.3879149571055576,
      "grad_norm": 0.44565173983573914,
      "learning_rate": 3.0606324340005803e-06,
      "loss": 0.7227,
      "step": 9360
    },
    {
      "epoch": 0.3879564010112313,
      "grad_norm": 0.4418325424194336,
      "learning_rate": 3.0604252144722117e-06,
      "loss": 0.6963,
      "step": 9361
    },
    {
      "epoch": 0.387997844916905,
      "grad_norm": 0.4202401340007782,
      "learning_rate": 3.060217994943844e-06,
      "loss": 0.7322,
      "step": 9362
    },
    {
      "epoch": 0.38803928882257865,
      "grad_norm": 0.4099680185317993,
      "learning_rate": 3.0600107754154757e-06,
      "loss": 0.7207,
      "step": 9363
    },
    {
      "epoch": 0.38808073272825233,
      "grad_norm": 0.42437639832496643,
      "learning_rate": 3.059803555887107e-06,
      "loss": 0.7019,
      "step": 9364
    },
    {
      "epoch": 0.388122176633926,
      "grad_norm": 0.4297334551811218,
      "learning_rate": 3.059596336358739e-06,
      "loss": 0.6953,
      "step": 9365
    },
    {
      "epoch": 0.38816362053959963,
      "grad_norm": 0.3909590542316437,
      "learning_rate": 3.0593891168303703e-06,
      "loss": 0.6721,
      "step": 9366
    },
    {
      "epoch": 0.3882050644452733,
      "grad_norm": 0.3923332989215851,
      "learning_rate": 3.059181897302002e-06,
      "loss": 0.6709,
      "step": 9367
    },
    {
      "epoch": 0.388246508350947,
      "grad_norm": 0.4225355386734009,
      "learning_rate": 3.0589746777736335e-06,
      "loss": 0.6786,
      "step": 9368
    },
    {
      "epoch": 0.38828795225662066,
      "grad_norm": 0.40075549483299255,
      "learning_rate": 3.0587674582452653e-06,
      "loss": 0.7092,
      "step": 9369
    },
    {
      "epoch": 0.38832939616229434,
      "grad_norm": 0.395334392786026,
      "learning_rate": 3.0585602387168967e-06,
      "loss": 0.7083,
      "step": 9370
    },
    {
      "epoch": 0.388370840067968,
      "grad_norm": 0.4742256999015808,
      "learning_rate": 3.0583530191885285e-06,
      "loss": 0.7449,
      "step": 9371
    },
    {
      "epoch": 0.3884122839736417,
      "grad_norm": 0.440755695104599,
      "learning_rate": 3.0581457996601603e-06,
      "loss": 0.7703,
      "step": 9372
    },
    {
      "epoch": 0.38845372787931537,
      "grad_norm": 0.39307084679603577,
      "learning_rate": 3.0579385801317917e-06,
      "loss": 0.7002,
      "step": 9373
    },
    {
      "epoch": 0.388495171784989,
      "grad_norm": 0.3954385817050934,
      "learning_rate": 3.0577313606034235e-06,
      "loss": 0.7085,
      "step": 9374
    },
    {
      "epoch": 0.38853661569066267,
      "grad_norm": 0.3915391266345978,
      "learning_rate": 3.057524141075055e-06,
      "loss": 0.6803,
      "step": 9375
    },
    {
      "epoch": 0.38857805959633634,
      "grad_norm": 0.42253270745277405,
      "learning_rate": 3.0573169215466867e-06,
      "loss": 0.7141,
      "step": 9376
    },
    {
      "epoch": 0.38861950350201,
      "grad_norm": 0.43275704979896545,
      "learning_rate": 3.057109702018318e-06,
      "loss": 0.7012,
      "step": 9377
    },
    {
      "epoch": 0.3886609474076837,
      "grad_norm": 0.48151740431785583,
      "learning_rate": 3.0569024824899503e-06,
      "loss": 0.7449,
      "step": 9378
    },
    {
      "epoch": 0.3887023913133574,
      "grad_norm": 0.42160600423812866,
      "learning_rate": 3.0566952629615813e-06,
      "loss": 0.6963,
      "step": 9379
    },
    {
      "epoch": 0.38874383521903105,
      "grad_norm": 0.4075644314289093,
      "learning_rate": 3.0564880434332135e-06,
      "loss": 0.6858,
      "step": 9380
    },
    {
      "epoch": 0.38878527912470473,
      "grad_norm": 0.43163609504699707,
      "learning_rate": 3.0562808239048453e-06,
      "loss": 0.6967,
      "step": 9381
    },
    {
      "epoch": 0.3888267230303784,
      "grad_norm": 0.4000757336616516,
      "learning_rate": 3.0560736043764767e-06,
      "loss": 0.6614,
      "step": 9382
    },
    {
      "epoch": 0.38886816693605203,
      "grad_norm": 0.41001972556114197,
      "learning_rate": 3.0558663848481085e-06,
      "loss": 0.675,
      "step": 9383
    },
    {
      "epoch": 0.3889096108417257,
      "grad_norm": 0.4375309646129608,
      "learning_rate": 3.05565916531974e-06,
      "loss": 0.718,
      "step": 9384
    },
    {
      "epoch": 0.3889510547473994,
      "grad_norm": 0.47871777415275574,
      "learning_rate": 3.0554519457913717e-06,
      "loss": 0.7468,
      "step": 9385
    },
    {
      "epoch": 0.38899249865307306,
      "grad_norm": 0.4396761953830719,
      "learning_rate": 3.055244726263003e-06,
      "loss": 0.6819,
      "step": 9386
    },
    {
      "epoch": 0.38903394255874674,
      "grad_norm": 0.43953385949134827,
      "learning_rate": 3.055037506734635e-06,
      "loss": 0.6902,
      "step": 9387
    },
    {
      "epoch": 0.3890753864644204,
      "grad_norm": 0.42892351746559143,
      "learning_rate": 3.0548302872062667e-06,
      "loss": 0.7288,
      "step": 9388
    },
    {
      "epoch": 0.3891168303700941,
      "grad_norm": 0.4042990803718567,
      "learning_rate": 3.054623067677898e-06,
      "loss": 0.6667,
      "step": 9389
    },
    {
      "epoch": 0.38915827427576777,
      "grad_norm": 0.39697572588920593,
      "learning_rate": 3.05441584814953e-06,
      "loss": 0.6851,
      "step": 9390
    },
    {
      "epoch": 0.38919971818144145,
      "grad_norm": 0.3929368853569031,
      "learning_rate": 3.0542086286211613e-06,
      "loss": 0.6702,
      "step": 9391
    },
    {
      "epoch": 0.38924116208711507,
      "grad_norm": 0.4014310836791992,
      "learning_rate": 3.054001409092793e-06,
      "loss": 0.6858,
      "step": 9392
    },
    {
      "epoch": 0.38928260599278874,
      "grad_norm": 0.408161997795105,
      "learning_rate": 3.0537941895644245e-06,
      "loss": 0.6667,
      "step": 9393
    },
    {
      "epoch": 0.3893240498984624,
      "grad_norm": 0.40276798605918884,
      "learning_rate": 3.0535869700360563e-06,
      "loss": 0.6556,
      "step": 9394
    },
    {
      "epoch": 0.3893654938041361,
      "grad_norm": 0.43274080753326416,
      "learning_rate": 3.0533797505076877e-06,
      "loss": 0.7205,
      "step": 9395
    },
    {
      "epoch": 0.3894069377098098,
      "grad_norm": 0.4182130992412567,
      "learning_rate": 3.05317253097932e-06,
      "loss": 0.7208,
      "step": 9396
    },
    {
      "epoch": 0.38944838161548345,
      "grad_norm": 0.4255117177963257,
      "learning_rate": 3.0529653114509517e-06,
      "loss": 0.7148,
      "step": 9397
    },
    {
      "epoch": 0.38948982552115713,
      "grad_norm": 0.40732160210609436,
      "learning_rate": 3.052758091922583e-06,
      "loss": 0.6854,
      "step": 9398
    },
    {
      "epoch": 0.3895312694268308,
      "grad_norm": 0.4692497253417969,
      "learning_rate": 3.052550872394215e-06,
      "loss": 0.7415,
      "step": 9399
    },
    {
      "epoch": 0.38957271333250443,
      "grad_norm": 0.4151582717895508,
      "learning_rate": 3.0523436528658463e-06,
      "loss": 0.722,
      "step": 9400
    },
    {
      "epoch": 0.3896141572381781,
      "grad_norm": 0.41852709650993347,
      "learning_rate": 3.052136433337478e-06,
      "loss": 0.7573,
      "step": 9401
    },
    {
      "epoch": 0.3896556011438518,
      "grad_norm": 0.42075392603874207,
      "learning_rate": 3.0519292138091095e-06,
      "loss": 0.7275,
      "step": 9402
    },
    {
      "epoch": 0.38969704504952546,
      "grad_norm": 0.4144735336303711,
      "learning_rate": 3.0517219942807413e-06,
      "loss": 0.7144,
      "step": 9403
    },
    {
      "epoch": 0.38973848895519914,
      "grad_norm": 0.3844081461429596,
      "learning_rate": 3.0515147747523727e-06,
      "loss": 0.6539,
      "step": 9404
    },
    {
      "epoch": 0.3897799328608728,
      "grad_norm": 0.4397536516189575,
      "learning_rate": 3.0513075552240045e-06,
      "loss": 0.6885,
      "step": 9405
    },
    {
      "epoch": 0.3898213767665465,
      "grad_norm": 0.3842150568962097,
      "learning_rate": 3.0511003356956363e-06,
      "loss": 0.6689,
      "step": 9406
    },
    {
      "epoch": 0.38986282067222017,
      "grad_norm": 0.4209131896495819,
      "learning_rate": 3.0508931161672677e-06,
      "loss": 0.7139,
      "step": 9407
    },
    {
      "epoch": 0.38990426457789384,
      "grad_norm": 0.3816525638103485,
      "learning_rate": 3.0506858966388995e-06,
      "loss": 0.7064,
      "step": 9408
    },
    {
      "epoch": 0.38994570848356747,
      "grad_norm": 0.43802395462989807,
      "learning_rate": 3.050478677110531e-06,
      "loss": 0.6863,
      "step": 9409
    },
    {
      "epoch": 0.38998715238924114,
      "grad_norm": 0.40442538261413574,
      "learning_rate": 3.0502714575821627e-06,
      "loss": 0.6554,
      "step": 9410
    },
    {
      "epoch": 0.3900285962949148,
      "grad_norm": 0.41163915395736694,
      "learning_rate": 3.050064238053794e-06,
      "loss": 0.6738,
      "step": 9411
    },
    {
      "epoch": 0.3900700402005885,
      "grad_norm": 0.42640355229377747,
      "learning_rate": 3.0498570185254263e-06,
      "loss": 0.7239,
      "step": 9412
    },
    {
      "epoch": 0.3901114841062622,
      "grad_norm": 0.41415268182754517,
      "learning_rate": 3.0496497989970573e-06,
      "loss": 0.704,
      "step": 9413
    },
    {
      "epoch": 0.39015292801193585,
      "grad_norm": 0.4178018271923065,
      "learning_rate": 3.0494425794686895e-06,
      "loss": 0.6835,
      "step": 9414
    },
    {
      "epoch": 0.39019437191760953,
      "grad_norm": 0.42922040820121765,
      "learning_rate": 3.0492353599403214e-06,
      "loss": 0.7175,
      "step": 9415
    },
    {
      "epoch": 0.3902358158232832,
      "grad_norm": 0.40202462673187256,
      "learning_rate": 3.0490281404119527e-06,
      "loss": 0.7068,
      "step": 9416
    },
    {
      "epoch": 0.3902772597289569,
      "grad_norm": 0.4150748550891876,
      "learning_rate": 3.0488209208835845e-06,
      "loss": 0.7247,
      "step": 9417
    },
    {
      "epoch": 0.3903187036346305,
      "grad_norm": 0.3927296996116638,
      "learning_rate": 3.048613701355216e-06,
      "loss": 0.6863,
      "step": 9418
    },
    {
      "epoch": 0.3903601475403042,
      "grad_norm": 0.42835643887519836,
      "learning_rate": 3.0484064818268477e-06,
      "loss": 0.6763,
      "step": 9419
    },
    {
      "epoch": 0.39040159144597786,
      "grad_norm": 0.39233919978141785,
      "learning_rate": 3.048199262298479e-06,
      "loss": 0.6855,
      "step": 9420
    },
    {
      "epoch": 0.39044303535165154,
      "grad_norm": 0.4676865041255951,
      "learning_rate": 3.047992042770111e-06,
      "loss": 0.698,
      "step": 9421
    },
    {
      "epoch": 0.3904844792573252,
      "grad_norm": 0.43749746680259705,
      "learning_rate": 3.0477848232417423e-06,
      "loss": 0.7002,
      "step": 9422
    },
    {
      "epoch": 0.3905259231629989,
      "grad_norm": 0.4187760353088379,
      "learning_rate": 3.047577603713374e-06,
      "loss": 0.7021,
      "step": 9423
    },
    {
      "epoch": 0.39056736706867257,
      "grad_norm": 0.40054646134376526,
      "learning_rate": 3.047370384185006e-06,
      "loss": 0.7048,
      "step": 9424
    },
    {
      "epoch": 0.39060881097434624,
      "grad_norm": 0.3844476044178009,
      "learning_rate": 3.0471631646566373e-06,
      "loss": 0.6562,
      "step": 9425
    },
    {
      "epoch": 0.3906502548800199,
      "grad_norm": 0.4560074806213379,
      "learning_rate": 3.046955945128269e-06,
      "loss": 0.7402,
      "step": 9426
    },
    {
      "epoch": 0.39069169878569354,
      "grad_norm": 0.422397255897522,
      "learning_rate": 3.0467487255999005e-06,
      "loss": 0.697,
      "step": 9427
    },
    {
      "epoch": 0.3907331426913672,
      "grad_norm": 0.401538610458374,
      "learning_rate": 3.0465415060715323e-06,
      "loss": 0.7356,
      "step": 9428
    },
    {
      "epoch": 0.3907745865970409,
      "grad_norm": 0.38236844539642334,
      "learning_rate": 3.0463342865431637e-06,
      "loss": 0.6874,
      "step": 9429
    },
    {
      "epoch": 0.3908160305027146,
      "grad_norm": 0.40216246247291565,
      "learning_rate": 3.046127067014796e-06,
      "loss": 0.6558,
      "step": 9430
    },
    {
      "epoch": 0.39085747440838825,
      "grad_norm": 0.40618130564689636,
      "learning_rate": 3.045919847486427e-06,
      "loss": 0.6813,
      "step": 9431
    },
    {
      "epoch": 0.3908989183140619,
      "grad_norm": 0.3617019057273865,
      "learning_rate": 3.045712627958059e-06,
      "loss": 0.6788,
      "step": 9432
    },
    {
      "epoch": 0.3909403622197356,
      "grad_norm": 0.40182480216026306,
      "learning_rate": 3.045505408429691e-06,
      "loss": 0.6317,
      "step": 9433
    },
    {
      "epoch": 0.3909818061254093,
      "grad_norm": 0.40691280364990234,
      "learning_rate": 3.0452981889013223e-06,
      "loss": 0.7026,
      "step": 9434
    },
    {
      "epoch": 0.3910232500310829,
      "grad_norm": 0.3901875913143158,
      "learning_rate": 3.045090969372954e-06,
      "loss": 0.6703,
      "step": 9435
    },
    {
      "epoch": 0.3910646939367566,
      "grad_norm": 0.37969711422920227,
      "learning_rate": 3.0448837498445855e-06,
      "loss": 0.6105,
      "step": 9436
    },
    {
      "epoch": 0.39110613784243026,
      "grad_norm": 0.4128420054912567,
      "learning_rate": 3.0446765303162173e-06,
      "loss": 0.7029,
      "step": 9437
    },
    {
      "epoch": 0.39114758174810393,
      "grad_norm": 0.40657541155815125,
      "learning_rate": 3.0444693107878487e-06,
      "loss": 0.6719,
      "step": 9438
    },
    {
      "epoch": 0.3911890256537776,
      "grad_norm": 0.3873860538005829,
      "learning_rate": 3.0442620912594805e-06,
      "loss": 0.6594,
      "step": 9439
    },
    {
      "epoch": 0.3912304695594513,
      "grad_norm": 0.43893229961395264,
      "learning_rate": 3.044054871731112e-06,
      "loss": 0.6809,
      "step": 9440
    },
    {
      "epoch": 0.39127191346512497,
      "grad_norm": 0.43747207522392273,
      "learning_rate": 3.0438476522027437e-06,
      "loss": 0.6719,
      "step": 9441
    },
    {
      "epoch": 0.39131335737079864,
      "grad_norm": 0.4196889102458954,
      "learning_rate": 3.0436404326743755e-06,
      "loss": 0.7444,
      "step": 9442
    },
    {
      "epoch": 0.3913548012764723,
      "grad_norm": 0.375454306602478,
      "learning_rate": 3.043433213146007e-06,
      "loss": 0.6641,
      "step": 9443
    },
    {
      "epoch": 0.39139624518214594,
      "grad_norm": 0.4401792883872986,
      "learning_rate": 3.0432259936176387e-06,
      "loss": 0.7008,
      "step": 9444
    },
    {
      "epoch": 0.3914376890878196,
      "grad_norm": 0.3918755054473877,
      "learning_rate": 3.04301877408927e-06,
      "loss": 0.6759,
      "step": 9445
    },
    {
      "epoch": 0.3914791329934933,
      "grad_norm": 0.4113456606864929,
      "learning_rate": 3.0428115545609024e-06,
      "loss": 0.7659,
      "step": 9446
    },
    {
      "epoch": 0.39152057689916697,
      "grad_norm": 0.4200038015842438,
      "learning_rate": 3.0426043350325333e-06,
      "loss": 0.7122,
      "step": 9447
    },
    {
      "epoch": 0.39156202080484065,
      "grad_norm": 0.42115336656570435,
      "learning_rate": 3.0423971155041655e-06,
      "loss": 0.6528,
      "step": 9448
    },
    {
      "epoch": 0.3916034647105143,
      "grad_norm": 0.3892390727996826,
      "learning_rate": 3.0421898959757974e-06,
      "loss": 0.6572,
      "step": 9449
    },
    {
      "epoch": 0.391644908616188,
      "grad_norm": 0.4148411154747009,
      "learning_rate": 3.0419826764474287e-06,
      "loss": 0.686,
      "step": 9450
    },
    {
      "epoch": 0.3916863525218617,
      "grad_norm": 0.3985777199268341,
      "learning_rate": 3.0417754569190606e-06,
      "loss": 0.6809,
      "step": 9451
    },
    {
      "epoch": 0.39172779642753536,
      "grad_norm": 0.43128225207328796,
      "learning_rate": 3.041568237390692e-06,
      "loss": 0.6885,
      "step": 9452
    },
    {
      "epoch": 0.391769240333209,
      "grad_norm": 0.40669140219688416,
      "learning_rate": 3.0413610178623237e-06,
      "loss": 0.7219,
      "step": 9453
    },
    {
      "epoch": 0.39181068423888266,
      "grad_norm": 0.3818778395652771,
      "learning_rate": 3.041153798333955e-06,
      "loss": 0.7009,
      "step": 9454
    },
    {
      "epoch": 0.39185212814455633,
      "grad_norm": 0.4299337863922119,
      "learning_rate": 3.040946578805587e-06,
      "loss": 0.7102,
      "step": 9455
    },
    {
      "epoch": 0.39189357205023,
      "grad_norm": 0.4290864169597626,
      "learning_rate": 3.0407393592772183e-06,
      "loss": 0.6941,
      "step": 9456
    },
    {
      "epoch": 0.3919350159559037,
      "grad_norm": 0.40424028038978577,
      "learning_rate": 3.04053213974885e-06,
      "loss": 0.6664,
      "step": 9457
    },
    {
      "epoch": 0.39197645986157736,
      "grad_norm": 0.4697413146495819,
      "learning_rate": 3.040324920220482e-06,
      "loss": 0.7214,
      "step": 9458
    },
    {
      "epoch": 0.39201790376725104,
      "grad_norm": 0.370175838470459,
      "learning_rate": 3.0401177006921133e-06,
      "loss": 0.7195,
      "step": 9459
    },
    {
      "epoch": 0.3920593476729247,
      "grad_norm": 0.44440561532974243,
      "learning_rate": 3.039910481163745e-06,
      "loss": 0.7089,
      "step": 9460
    },
    {
      "epoch": 0.39210079157859834,
      "grad_norm": 0.40980175137519836,
      "learning_rate": 3.0397032616353765e-06,
      "loss": 0.6456,
      "step": 9461
    },
    {
      "epoch": 0.392142235484272,
      "grad_norm": 0.3858970105648041,
      "learning_rate": 3.0394960421070088e-06,
      "loss": 0.6976,
      "step": 9462
    },
    {
      "epoch": 0.3921836793899457,
      "grad_norm": 0.4140380024909973,
      "learning_rate": 3.0392888225786397e-06,
      "loss": 0.7024,
      "step": 9463
    },
    {
      "epoch": 0.39222512329561937,
      "grad_norm": 0.43206754326820374,
      "learning_rate": 3.039081603050272e-06,
      "loss": 0.6665,
      "step": 9464
    },
    {
      "epoch": 0.39226656720129305,
      "grad_norm": 0.4300233721733093,
      "learning_rate": 3.038874383521903e-06,
      "loss": 0.6993,
      "step": 9465
    },
    {
      "epoch": 0.3923080111069667,
      "grad_norm": 0.38279590010643005,
      "learning_rate": 3.038667163993535e-06,
      "loss": 0.7249,
      "step": 9466
    },
    {
      "epoch": 0.3923494550126404,
      "grad_norm": 0.39058512449264526,
      "learning_rate": 3.038459944465167e-06,
      "loss": 0.6982,
      "step": 9467
    },
    {
      "epoch": 0.3923908989183141,
      "grad_norm": 0.42312824726104736,
      "learning_rate": 3.0382527249367983e-06,
      "loss": 0.7031,
      "step": 9468
    },
    {
      "epoch": 0.39243234282398776,
      "grad_norm": 0.3944706320762634,
      "learning_rate": 3.03804550540843e-06,
      "loss": 0.7324,
      "step": 9469
    },
    {
      "epoch": 0.3924737867296614,
      "grad_norm": 0.43116700649261475,
      "learning_rate": 3.0378382858800615e-06,
      "loss": 0.7305,
      "step": 9470
    },
    {
      "epoch": 0.39251523063533506,
      "grad_norm": 0.4367968440055847,
      "learning_rate": 3.0376310663516933e-06,
      "loss": 0.7554,
      "step": 9471
    },
    {
      "epoch": 0.39255667454100873,
      "grad_norm": 0.4336731433868408,
      "learning_rate": 3.0374238468233247e-06,
      "loss": 0.7664,
      "step": 9472
    },
    {
      "epoch": 0.3925981184466824,
      "grad_norm": 0.4240306615829468,
      "learning_rate": 3.0372166272949565e-06,
      "loss": 0.6698,
      "step": 9473
    },
    {
      "epoch": 0.3926395623523561,
      "grad_norm": 0.41871407628059387,
      "learning_rate": 3.037009407766588e-06,
      "loss": 0.6815,
      "step": 9474
    },
    {
      "epoch": 0.39268100625802976,
      "grad_norm": 0.405497282743454,
      "learning_rate": 3.0368021882382197e-06,
      "loss": 0.6941,
      "step": 9475
    },
    {
      "epoch": 0.39272245016370344,
      "grad_norm": 0.4034014344215393,
      "learning_rate": 3.0365949687098515e-06,
      "loss": 0.6526,
      "step": 9476
    },
    {
      "epoch": 0.3927638940693771,
      "grad_norm": 0.3778819143772125,
      "learning_rate": 3.036387749181483e-06,
      "loss": 0.6829,
      "step": 9477
    },
    {
      "epoch": 0.3928053379750508,
      "grad_norm": 0.39154472947120667,
      "learning_rate": 3.0361805296531147e-06,
      "loss": 0.6809,
      "step": 9478
    },
    {
      "epoch": 0.3928467818807244,
      "grad_norm": 0.458553671836853,
      "learning_rate": 3.035973310124746e-06,
      "loss": 0.7729,
      "step": 9479
    },
    {
      "epoch": 0.3928882257863981,
      "grad_norm": 0.40662333369255066,
      "learning_rate": 3.0357660905963784e-06,
      "loss": 0.7083,
      "step": 9480
    },
    {
      "epoch": 0.39292966969207177,
      "grad_norm": 0.3842395842075348,
      "learning_rate": 3.0355588710680093e-06,
      "loss": 0.6417,
      "step": 9481
    },
    {
      "epoch": 0.39297111359774545,
      "grad_norm": 0.40851306915283203,
      "learning_rate": 3.0353516515396416e-06,
      "loss": 0.6973,
      "step": 9482
    },
    {
      "epoch": 0.3930125575034191,
      "grad_norm": 0.42207080125808716,
      "learning_rate": 3.035144432011273e-06,
      "loss": 0.7305,
      "step": 9483
    },
    {
      "epoch": 0.3930540014090928,
      "grad_norm": 0.3744085431098938,
      "learning_rate": 3.0349372124829047e-06,
      "loss": 0.6938,
      "step": 9484
    },
    {
      "epoch": 0.3930954453147665,
      "grad_norm": 0.3872545659542084,
      "learning_rate": 3.0347299929545366e-06,
      "loss": 0.6309,
      "step": 9485
    },
    {
      "epoch": 0.39313688922044016,
      "grad_norm": 0.4180368185043335,
      "learning_rate": 3.034522773426168e-06,
      "loss": 0.6697,
      "step": 9486
    },
    {
      "epoch": 0.39317833312611383,
      "grad_norm": 0.4022526741027832,
      "learning_rate": 3.0343155538977998e-06,
      "loss": 0.6858,
      "step": 9487
    },
    {
      "epoch": 0.39321977703178745,
      "grad_norm": 0.41308555006980896,
      "learning_rate": 3.034108334369431e-06,
      "loss": 0.679,
      "step": 9488
    },
    {
      "epoch": 0.39326122093746113,
      "grad_norm": 0.39015960693359375,
      "learning_rate": 3.033901114841063e-06,
      "loss": 0.687,
      "step": 9489
    },
    {
      "epoch": 0.3933026648431348,
      "grad_norm": 0.4057956337928772,
      "learning_rate": 3.0336938953126943e-06,
      "loss": 0.7466,
      "step": 9490
    },
    {
      "epoch": 0.3933441087488085,
      "grad_norm": 0.41703540086746216,
      "learning_rate": 3.033486675784326e-06,
      "loss": 0.6656,
      "step": 9491
    },
    {
      "epoch": 0.39338555265448216,
      "grad_norm": 0.4397125244140625,
      "learning_rate": 3.0332794562559575e-06,
      "loss": 0.7239,
      "step": 9492
    },
    {
      "epoch": 0.39342699656015584,
      "grad_norm": 0.40248948335647583,
      "learning_rate": 3.0330722367275893e-06,
      "loss": 0.6904,
      "step": 9493
    },
    {
      "epoch": 0.3934684404658295,
      "grad_norm": 0.4140368103981018,
      "learning_rate": 3.032865017199221e-06,
      "loss": 0.7349,
      "step": 9494
    },
    {
      "epoch": 0.3935098843715032,
      "grad_norm": 0.4492703676223755,
      "learning_rate": 3.0326577976708525e-06,
      "loss": 0.7371,
      "step": 9495
    },
    {
      "epoch": 0.3935513282771768,
      "grad_norm": 0.3933362066745758,
      "learning_rate": 3.0324505781424848e-06,
      "loss": 0.6658,
      "step": 9496
    },
    {
      "epoch": 0.3935927721828505,
      "grad_norm": 0.4190264642238617,
      "learning_rate": 3.0322433586141157e-06,
      "loss": 0.6958,
      "step": 9497
    },
    {
      "epoch": 0.39363421608852417,
      "grad_norm": 0.4406547248363495,
      "learning_rate": 3.032036139085748e-06,
      "loss": 0.7687,
      "step": 9498
    },
    {
      "epoch": 0.39367565999419785,
      "grad_norm": 0.4731440246105194,
      "learning_rate": 3.031828919557379e-06,
      "loss": 0.7628,
      "step": 9499
    },
    {
      "epoch": 0.3937171038998715,
      "grad_norm": 0.4237787127494812,
      "learning_rate": 3.031621700029011e-06,
      "loss": 0.6875,
      "step": 9500
    },
    {
      "epoch": 0.3937585478055452,
      "grad_norm": 0.4177541732788086,
      "learning_rate": 3.0314144805006425e-06,
      "loss": 0.6465,
      "step": 9501
    },
    {
      "epoch": 0.3937999917112189,
      "grad_norm": 0.4523256719112396,
      "learning_rate": 3.0312072609722743e-06,
      "loss": 0.7993,
      "step": 9502
    },
    {
      "epoch": 0.39384143561689255,
      "grad_norm": 0.38543811440467834,
      "learning_rate": 3.031000041443906e-06,
      "loss": 0.6826,
      "step": 9503
    },
    {
      "epoch": 0.39388287952256623,
      "grad_norm": 0.41366326808929443,
      "learning_rate": 3.0307928219155375e-06,
      "loss": 0.6575,
      "step": 9504
    },
    {
      "epoch": 0.39392432342823985,
      "grad_norm": 0.4158187806606293,
      "learning_rate": 3.0305856023871694e-06,
      "loss": 0.7203,
      "step": 9505
    },
    {
      "epoch": 0.39396576733391353,
      "grad_norm": 0.4178304076194763,
      "learning_rate": 3.0303783828588007e-06,
      "loss": 0.7109,
      "step": 9506
    },
    {
      "epoch": 0.3940072112395872,
      "grad_norm": 0.4592568874359131,
      "learning_rate": 3.0301711633304325e-06,
      "loss": 0.7341,
      "step": 9507
    },
    {
      "epoch": 0.3940486551452609,
      "grad_norm": 0.4132538437843323,
      "learning_rate": 3.029963943802064e-06,
      "loss": 0.6838,
      "step": 9508
    },
    {
      "epoch": 0.39409009905093456,
      "grad_norm": 0.4246460497379303,
      "learning_rate": 3.0297567242736957e-06,
      "loss": 0.7031,
      "step": 9509
    },
    {
      "epoch": 0.39413154295660824,
      "grad_norm": 0.40939202904701233,
      "learning_rate": 3.0295495047453276e-06,
      "loss": 0.6609,
      "step": 9510
    },
    {
      "epoch": 0.3941729868622819,
      "grad_norm": 0.40757328271865845,
      "learning_rate": 3.029342285216959e-06,
      "loss": 0.7036,
      "step": 9511
    },
    {
      "epoch": 0.3942144307679556,
      "grad_norm": 0.4184506833553314,
      "learning_rate": 3.0291350656885907e-06,
      "loss": 0.6727,
      "step": 9512
    },
    {
      "epoch": 0.39425587467362927,
      "grad_norm": 0.47774237394332886,
      "learning_rate": 3.028927846160222e-06,
      "loss": 0.7834,
      "step": 9513
    },
    {
      "epoch": 0.3942973185793029,
      "grad_norm": 0.45306167006492615,
      "learning_rate": 3.0287206266318544e-06,
      "loss": 0.745,
      "step": 9514
    },
    {
      "epoch": 0.39433876248497657,
      "grad_norm": 0.4820023477077484,
      "learning_rate": 3.0285134071034853e-06,
      "loss": 0.787,
      "step": 9515
    },
    {
      "epoch": 0.39438020639065025,
      "grad_norm": 0.4054049551486969,
      "learning_rate": 3.0283061875751176e-06,
      "loss": 0.6511,
      "step": 9516
    },
    {
      "epoch": 0.3944216502963239,
      "grad_norm": 0.39879316091537476,
      "learning_rate": 3.028098968046749e-06,
      "loss": 0.6914,
      "step": 9517
    },
    {
      "epoch": 0.3944630942019976,
      "grad_norm": 0.40437743067741394,
      "learning_rate": 3.0278917485183808e-06,
      "loss": 0.6873,
      "step": 9518
    },
    {
      "epoch": 0.3945045381076713,
      "grad_norm": 0.41723304986953735,
      "learning_rate": 3.0276845289900126e-06,
      "loss": 0.7087,
      "step": 9519
    },
    {
      "epoch": 0.39454598201334495,
      "grad_norm": 0.4016387164592743,
      "learning_rate": 3.027477309461644e-06,
      "loss": 0.6846,
      "step": 9520
    },
    {
      "epoch": 0.39458742591901863,
      "grad_norm": 0.38109537959098816,
      "learning_rate": 3.0272700899332758e-06,
      "loss": 0.7073,
      "step": 9521
    },
    {
      "epoch": 0.39462886982469225,
      "grad_norm": 0.431786447763443,
      "learning_rate": 3.027062870404907e-06,
      "loss": 0.7078,
      "step": 9522
    },
    {
      "epoch": 0.39467031373036593,
      "grad_norm": 0.4154272675514221,
      "learning_rate": 3.026855650876539e-06,
      "loss": 0.6707,
      "step": 9523
    },
    {
      "epoch": 0.3947117576360396,
      "grad_norm": 0.45969444513320923,
      "learning_rate": 3.0266484313481703e-06,
      "loss": 0.7339,
      "step": 9524
    },
    {
      "epoch": 0.3947532015417133,
      "grad_norm": 0.43442776799201965,
      "learning_rate": 3.026441211819802e-06,
      "loss": 0.668,
      "step": 9525
    },
    {
      "epoch": 0.39479464544738696,
      "grad_norm": 0.4378511607646942,
      "learning_rate": 3.0262339922914335e-06,
      "loss": 0.7317,
      "step": 9526
    },
    {
      "epoch": 0.39483608935306064,
      "grad_norm": 0.41116365790367126,
      "learning_rate": 3.0260267727630653e-06,
      "loss": 0.7185,
      "step": 9527
    },
    {
      "epoch": 0.3948775332587343,
      "grad_norm": 0.42327770590782166,
      "learning_rate": 3.025819553234697e-06,
      "loss": 0.708,
      "step": 9528
    },
    {
      "epoch": 0.394918977164408,
      "grad_norm": 0.4089825451374054,
      "learning_rate": 3.0256123337063285e-06,
      "loss": 0.7256,
      "step": 9529
    },
    {
      "epoch": 0.39496042107008167,
      "grad_norm": 0.38536033034324646,
      "learning_rate": 3.0254051141779608e-06,
      "loss": 0.656,
      "step": 9530
    },
    {
      "epoch": 0.3950018649757553,
      "grad_norm": 0.4033794701099396,
      "learning_rate": 3.0251978946495917e-06,
      "loss": 0.6902,
      "step": 9531
    },
    {
      "epoch": 0.39504330888142897,
      "grad_norm": 0.40971848368644714,
      "learning_rate": 3.024990675121224e-06,
      "loss": 0.6523,
      "step": 9532
    },
    {
      "epoch": 0.39508475278710264,
      "grad_norm": 0.437128484249115,
      "learning_rate": 3.024783455592855e-06,
      "loss": 0.7283,
      "step": 9533
    },
    {
      "epoch": 0.3951261966927763,
      "grad_norm": 0.4298624098300934,
      "learning_rate": 3.024576236064487e-06,
      "loss": 0.7429,
      "step": 9534
    },
    {
      "epoch": 0.39516764059845,
      "grad_norm": 0.4121999740600586,
      "learning_rate": 3.0243690165361185e-06,
      "loss": 0.6997,
      "step": 9535
    },
    {
      "epoch": 0.3952090845041237,
      "grad_norm": 0.39793020486831665,
      "learning_rate": 3.0241617970077504e-06,
      "loss": 0.6801,
      "step": 9536
    },
    {
      "epoch": 0.39525052840979735,
      "grad_norm": 0.4143294095993042,
      "learning_rate": 3.023954577479382e-06,
      "loss": 0.7744,
      "step": 9537
    },
    {
      "epoch": 0.39529197231547103,
      "grad_norm": 0.41114315390586853,
      "learning_rate": 3.0237473579510135e-06,
      "loss": 0.7107,
      "step": 9538
    },
    {
      "epoch": 0.3953334162211447,
      "grad_norm": 0.4264541566371918,
      "learning_rate": 3.0235401384226454e-06,
      "loss": 0.708,
      "step": 9539
    },
    {
      "epoch": 0.39537486012681833,
      "grad_norm": 0.4418397545814514,
      "learning_rate": 3.0233329188942767e-06,
      "loss": 0.702,
      "step": 9540
    },
    {
      "epoch": 0.395416304032492,
      "grad_norm": 0.42302101850509644,
      "learning_rate": 3.0231256993659086e-06,
      "loss": 0.6978,
      "step": 9541
    },
    {
      "epoch": 0.3954577479381657,
      "grad_norm": 0.437695175409317,
      "learning_rate": 3.02291847983754e-06,
      "loss": 0.7102,
      "step": 9542
    },
    {
      "epoch": 0.39549919184383936,
      "grad_norm": 0.4156973361968994,
      "learning_rate": 3.0227112603091717e-06,
      "loss": 0.6909,
      "step": 9543
    },
    {
      "epoch": 0.39554063574951304,
      "grad_norm": 0.4250630736351013,
      "learning_rate": 3.022504040780803e-06,
      "loss": 0.6746,
      "step": 9544
    },
    {
      "epoch": 0.3955820796551867,
      "grad_norm": 0.43384137749671936,
      "learning_rate": 3.022296821252435e-06,
      "loss": 0.7444,
      "step": 9545
    },
    {
      "epoch": 0.3956235235608604,
      "grad_norm": 0.38809671998023987,
      "learning_rate": 3.0220896017240668e-06,
      "loss": 0.6821,
      "step": 9546
    },
    {
      "epoch": 0.39566496746653407,
      "grad_norm": 0.42618754506111145,
      "learning_rate": 3.021882382195698e-06,
      "loss": 0.6899,
      "step": 9547
    },
    {
      "epoch": 0.3957064113722077,
      "grad_norm": 0.41692298650741577,
      "learning_rate": 3.0216751626673304e-06,
      "loss": 0.6802,
      "step": 9548
    },
    {
      "epoch": 0.39574785527788137,
      "grad_norm": 0.377252995967865,
      "learning_rate": 3.0214679431389613e-06,
      "loss": 0.6914,
      "step": 9549
    },
    {
      "epoch": 0.39578929918355504,
      "grad_norm": 0.426179975271225,
      "learning_rate": 3.0212607236105936e-06,
      "loss": 0.698,
      "step": 9550
    },
    {
      "epoch": 0.3958307430892287,
      "grad_norm": 0.40155890583992004,
      "learning_rate": 3.021053504082225e-06,
      "loss": 0.6512,
      "step": 9551
    },
    {
      "epoch": 0.3958721869949024,
      "grad_norm": 0.4274677336215973,
      "learning_rate": 3.0208462845538568e-06,
      "loss": 0.6467,
      "step": 9552
    },
    {
      "epoch": 0.3959136309005761,
      "grad_norm": 0.4084275960922241,
      "learning_rate": 3.020639065025488e-06,
      "loss": 0.7053,
      "step": 9553
    },
    {
      "epoch": 0.39595507480624975,
      "grad_norm": 0.41739505529403687,
      "learning_rate": 3.02043184549712e-06,
      "loss": 0.6515,
      "step": 9554
    },
    {
      "epoch": 0.39599651871192343,
      "grad_norm": 0.40556082129478455,
      "learning_rate": 3.0202246259687518e-06,
      "loss": 0.7195,
      "step": 9555
    },
    {
      "epoch": 0.3960379626175971,
      "grad_norm": 0.39795422554016113,
      "learning_rate": 3.020017406440383e-06,
      "loss": 0.6766,
      "step": 9556
    },
    {
      "epoch": 0.3960794065232707,
      "grad_norm": 0.4124256372451782,
      "learning_rate": 3.019810186912015e-06,
      "loss": 0.7236,
      "step": 9557
    },
    {
      "epoch": 0.3961208504289444,
      "grad_norm": 0.3810274600982666,
      "learning_rate": 3.0196029673836463e-06,
      "loss": 0.6608,
      "step": 9558
    },
    {
      "epoch": 0.3961622943346181,
      "grad_norm": 0.425054669380188,
      "learning_rate": 3.019395747855278e-06,
      "loss": 0.7024,
      "step": 9559
    },
    {
      "epoch": 0.39620373824029176,
      "grad_norm": 0.37433454394340515,
      "learning_rate": 3.0191885283269095e-06,
      "loss": 0.6387,
      "step": 9560
    },
    {
      "epoch": 0.39624518214596544,
      "grad_norm": 0.404678076505661,
      "learning_rate": 3.0189813087985413e-06,
      "loss": 0.6754,
      "step": 9561
    },
    {
      "epoch": 0.3962866260516391,
      "grad_norm": 0.4291474223136902,
      "learning_rate": 3.0187740892701727e-06,
      "loss": 0.7246,
      "step": 9562
    },
    {
      "epoch": 0.3963280699573128,
      "grad_norm": 0.4419684112071991,
      "learning_rate": 3.0185668697418045e-06,
      "loss": 0.728,
      "step": 9563
    },
    {
      "epoch": 0.39636951386298647,
      "grad_norm": 0.40887850522994995,
      "learning_rate": 3.0183596502134368e-06,
      "loss": 0.6334,
      "step": 9564
    },
    {
      "epoch": 0.39641095776866014,
      "grad_norm": 0.38945356011390686,
      "learning_rate": 3.0181524306850677e-06,
      "loss": 0.697,
      "step": 9565
    },
    {
      "epoch": 0.39645240167433377,
      "grad_norm": 0.40308645367622375,
      "learning_rate": 3.0179452111567e-06,
      "loss": 0.7693,
      "step": 9566
    },
    {
      "epoch": 0.39649384558000744,
      "grad_norm": 0.4002532362937927,
      "learning_rate": 3.017737991628331e-06,
      "loss": 0.6704,
      "step": 9567
    },
    {
      "epoch": 0.3965352894856811,
      "grad_norm": 0.4021584987640381,
      "learning_rate": 3.017530772099963e-06,
      "loss": 0.6865,
      "step": 9568
    },
    {
      "epoch": 0.3965767333913548,
      "grad_norm": 0.47005054354667664,
      "learning_rate": 3.0173235525715946e-06,
      "loss": 0.6838,
      "step": 9569
    },
    {
      "epoch": 0.3966181772970285,
      "grad_norm": 0.4345030188560486,
      "learning_rate": 3.0171163330432264e-06,
      "loss": 0.7421,
      "step": 9570
    },
    {
      "epoch": 0.39665962120270215,
      "grad_norm": 0.4128609001636505,
      "learning_rate": 3.016909113514858e-06,
      "loss": 0.6935,
      "step": 9571
    },
    {
      "epoch": 0.3967010651083758,
      "grad_norm": 0.4201023578643799,
      "learning_rate": 3.0167018939864896e-06,
      "loss": 0.7466,
      "step": 9572
    },
    {
      "epoch": 0.3967425090140495,
      "grad_norm": 0.39556774497032166,
      "learning_rate": 3.0164946744581214e-06,
      "loss": 0.6703,
      "step": 9573
    },
    {
      "epoch": 0.3967839529197232,
      "grad_norm": 0.42128506302833557,
      "learning_rate": 3.0162874549297528e-06,
      "loss": 0.6877,
      "step": 9574
    },
    {
      "epoch": 0.3968253968253968,
      "grad_norm": 0.43170836567878723,
      "learning_rate": 3.0160802354013846e-06,
      "loss": 0.7119,
      "step": 9575
    },
    {
      "epoch": 0.3968668407310705,
      "grad_norm": 0.4017358422279358,
      "learning_rate": 3.015873015873016e-06,
      "loss": 0.6826,
      "step": 9576
    },
    {
      "epoch": 0.39690828463674416,
      "grad_norm": 0.5477772951126099,
      "learning_rate": 3.0156657963446478e-06,
      "loss": 0.6798,
      "step": 9577
    },
    {
      "epoch": 0.39694972854241783,
      "grad_norm": 0.4157739281654358,
      "learning_rate": 3.015458576816279e-06,
      "loss": 0.7335,
      "step": 9578
    },
    {
      "epoch": 0.3969911724480915,
      "grad_norm": 0.4020553529262543,
      "learning_rate": 3.015251357287911e-06,
      "loss": 0.6973,
      "step": 9579
    },
    {
      "epoch": 0.3970326163537652,
      "grad_norm": 0.42026329040527344,
      "learning_rate": 3.0150441377595428e-06,
      "loss": 0.6835,
      "step": 9580
    },
    {
      "epoch": 0.39707406025943887,
      "grad_norm": 0.38915637135505676,
      "learning_rate": 3.014836918231174e-06,
      "loss": 0.734,
      "step": 9581
    },
    {
      "epoch": 0.39711550416511254,
      "grad_norm": 0.39482253789901733,
      "learning_rate": 3.0146296987028064e-06,
      "loss": 0.7061,
      "step": 9582
    },
    {
      "epoch": 0.39715694807078616,
      "grad_norm": 0.4192039668560028,
      "learning_rate": 3.0144224791744373e-06,
      "loss": 0.6772,
      "step": 9583
    },
    {
      "epoch": 0.39719839197645984,
      "grad_norm": 0.3737242519855499,
      "learning_rate": 3.0142152596460696e-06,
      "loss": 0.6333,
      "step": 9584
    },
    {
      "epoch": 0.3972398358821335,
      "grad_norm": 0.4051286280155182,
      "learning_rate": 3.014008040117701e-06,
      "loss": 0.7433,
      "step": 9585
    },
    {
      "epoch": 0.3972812797878072,
      "grad_norm": 0.428166002035141,
      "learning_rate": 3.0138008205893328e-06,
      "loss": 0.7012,
      "step": 9586
    },
    {
      "epoch": 0.3973227236934809,
      "grad_norm": 0.41494107246398926,
      "learning_rate": 3.013593601060964e-06,
      "loss": 0.7216,
      "step": 9587
    },
    {
      "epoch": 0.39736416759915455,
      "grad_norm": 0.3880235552787781,
      "learning_rate": 3.013386381532596e-06,
      "loss": 0.7021,
      "step": 9588
    },
    {
      "epoch": 0.3974056115048282,
      "grad_norm": 0.41936543583869934,
      "learning_rate": 3.0131791620042278e-06,
      "loss": 0.6404,
      "step": 9589
    },
    {
      "epoch": 0.3974470554105019,
      "grad_norm": 0.4009304642677307,
      "learning_rate": 3.012971942475859e-06,
      "loss": 0.7158,
      "step": 9590
    },
    {
      "epoch": 0.3974884993161756,
      "grad_norm": 0.44974878430366516,
      "learning_rate": 3.012764722947491e-06,
      "loss": 0.7216,
      "step": 9591
    },
    {
      "epoch": 0.3975299432218492,
      "grad_norm": 0.37596216797828674,
      "learning_rate": 3.0125575034191224e-06,
      "loss": 0.7107,
      "step": 9592
    },
    {
      "epoch": 0.3975713871275229,
      "grad_norm": 0.38925701379776,
      "learning_rate": 3.012350283890754e-06,
      "loss": 0.6355,
      "step": 9593
    },
    {
      "epoch": 0.39761283103319656,
      "grad_norm": 0.4065350890159607,
      "learning_rate": 3.0121430643623855e-06,
      "loss": 0.6947,
      "step": 9594
    },
    {
      "epoch": 0.39765427493887023,
      "grad_norm": 0.3931446373462677,
      "learning_rate": 3.0119358448340174e-06,
      "loss": 0.7021,
      "step": 9595
    },
    {
      "epoch": 0.3976957188445439,
      "grad_norm": 0.4282339811325073,
      "learning_rate": 3.0117286253056487e-06,
      "loss": 0.6649,
      "step": 9596
    },
    {
      "epoch": 0.3977371627502176,
      "grad_norm": 0.41863736510276794,
      "learning_rate": 3.0115214057772805e-06,
      "loss": 0.729,
      "step": 9597
    },
    {
      "epoch": 0.39777860665589126,
      "grad_norm": 0.4263836443424225,
      "learning_rate": 3.0113141862489128e-06,
      "loss": 0.7188,
      "step": 9598
    },
    {
      "epoch": 0.39782005056156494,
      "grad_norm": 0.44368159770965576,
      "learning_rate": 3.0111069667205437e-06,
      "loss": 0.7651,
      "step": 9599
    },
    {
      "epoch": 0.3978614944672386,
      "grad_norm": 0.3832560181617737,
      "learning_rate": 3.010899747192176e-06,
      "loss": 0.6891,
      "step": 9600
    },
    {
      "epoch": 0.39790293837291224,
      "grad_norm": 0.4258100986480713,
      "learning_rate": 3.0106925276638074e-06,
      "loss": 0.6936,
      "step": 9601
    },
    {
      "epoch": 0.3979443822785859,
      "grad_norm": 0.39879173040390015,
      "learning_rate": 3.010485308135439e-06,
      "loss": 0.6781,
      "step": 9602
    },
    {
      "epoch": 0.3979858261842596,
      "grad_norm": 0.3927847445011139,
      "learning_rate": 3.0102780886070706e-06,
      "loss": 0.688,
      "step": 9603
    },
    {
      "epoch": 0.39802727008993327,
      "grad_norm": 0.44034093618392944,
      "learning_rate": 3.0100708690787024e-06,
      "loss": 0.7589,
      "step": 9604
    },
    {
      "epoch": 0.39806871399560695,
      "grad_norm": 0.423814594745636,
      "learning_rate": 3.0098636495503338e-06,
      "loss": 0.7043,
      "step": 9605
    },
    {
      "epoch": 0.3981101579012806,
      "grad_norm": 0.4956913888454437,
      "learning_rate": 3.0096564300219656e-06,
      "loss": 0.7086,
      "step": 9606
    },
    {
      "epoch": 0.3981516018069543,
      "grad_norm": 0.39377278089523315,
      "learning_rate": 3.0094492104935974e-06,
      "loss": 0.6738,
      "step": 9607
    },
    {
      "epoch": 0.398193045712628,
      "grad_norm": 0.44385913014411926,
      "learning_rate": 3.0092419909652288e-06,
      "loss": 0.6451,
      "step": 9608
    },
    {
      "epoch": 0.3982344896183016,
      "grad_norm": 0.4152202010154724,
      "learning_rate": 3.0090347714368606e-06,
      "loss": 0.7893,
      "step": 9609
    },
    {
      "epoch": 0.3982759335239753,
      "grad_norm": 0.4080865681171417,
      "learning_rate": 3.008827551908492e-06,
      "loss": 0.6689,
      "step": 9610
    },
    {
      "epoch": 0.39831737742964896,
      "grad_norm": 0.41871216893196106,
      "learning_rate": 3.0086203323801238e-06,
      "loss": 0.6625,
      "step": 9611
    },
    {
      "epoch": 0.39835882133532263,
      "grad_norm": 0.43666625022888184,
      "learning_rate": 3.008413112851755e-06,
      "loss": 0.7366,
      "step": 9612
    },
    {
      "epoch": 0.3984002652409963,
      "grad_norm": 0.4114949405193329,
      "learning_rate": 3.008205893323387e-06,
      "loss": 0.6797,
      "step": 9613
    },
    {
      "epoch": 0.39844170914667,
      "grad_norm": 0.41424646973609924,
      "learning_rate": 3.0079986737950183e-06,
      "loss": 0.7253,
      "step": 9614
    },
    {
      "epoch": 0.39848315305234366,
      "grad_norm": 0.40461185574531555,
      "learning_rate": 3.00779145426665e-06,
      "loss": 0.6953,
      "step": 9615
    },
    {
      "epoch": 0.39852459695801734,
      "grad_norm": 0.4125845432281494,
      "learning_rate": 3.0075842347382824e-06,
      "loss": 0.6825,
      "step": 9616
    },
    {
      "epoch": 0.398566040863691,
      "grad_norm": 0.4023011326789856,
      "learning_rate": 3.0073770152099133e-06,
      "loss": 0.7158,
      "step": 9617
    },
    {
      "epoch": 0.39860748476936464,
      "grad_norm": 0.3855823278427124,
      "learning_rate": 3.0071697956815456e-06,
      "loss": 0.6633,
      "step": 9618
    },
    {
      "epoch": 0.3986489286750383,
      "grad_norm": 0.397495836019516,
      "learning_rate": 3.006962576153177e-06,
      "loss": 0.645,
      "step": 9619
    },
    {
      "epoch": 0.398690372580712,
      "grad_norm": 0.42676928639411926,
      "learning_rate": 3.0067553566248088e-06,
      "loss": 0.6462,
      "step": 9620
    },
    {
      "epoch": 0.39873181648638567,
      "grad_norm": 0.4323368966579437,
      "learning_rate": 3.00654813709644e-06,
      "loss": 0.748,
      "step": 9621
    },
    {
      "epoch": 0.39877326039205935,
      "grad_norm": 0.40364721417427063,
      "learning_rate": 3.006340917568072e-06,
      "loss": 0.7197,
      "step": 9622
    },
    {
      "epoch": 0.398814704297733,
      "grad_norm": 0.40984681248664856,
      "learning_rate": 3.0061336980397038e-06,
      "loss": 0.7123,
      "step": 9623
    },
    {
      "epoch": 0.3988561482034067,
      "grad_norm": 0.42342615127563477,
      "learning_rate": 3.005926478511335e-06,
      "loss": 0.6814,
      "step": 9624
    },
    {
      "epoch": 0.3988975921090804,
      "grad_norm": 0.4154376685619354,
      "learning_rate": 3.005719258982967e-06,
      "loss": 0.682,
      "step": 9625
    },
    {
      "epoch": 0.39893903601475406,
      "grad_norm": 0.40213608741760254,
      "learning_rate": 3.0055120394545984e-06,
      "loss": 0.7097,
      "step": 9626
    },
    {
      "epoch": 0.3989804799204277,
      "grad_norm": 0.43730056285858154,
      "learning_rate": 3.00530481992623e-06,
      "loss": 0.7336,
      "step": 9627
    },
    {
      "epoch": 0.39902192382610135,
      "grad_norm": 0.3922622799873352,
      "learning_rate": 3.0050976003978616e-06,
      "loss": 0.6943,
      "step": 9628
    },
    {
      "epoch": 0.39906336773177503,
      "grad_norm": 0.4234660267829895,
      "learning_rate": 3.0048903808694934e-06,
      "loss": 0.6409,
      "step": 9629
    },
    {
      "epoch": 0.3991048116374487,
      "grad_norm": 0.4117945432662964,
      "learning_rate": 3.0046831613411247e-06,
      "loss": 0.6968,
      "step": 9630
    },
    {
      "epoch": 0.3991462555431224,
      "grad_norm": 0.4291854798793793,
      "learning_rate": 3.0044759418127566e-06,
      "loss": 0.7083,
      "step": 9631
    },
    {
      "epoch": 0.39918769944879606,
      "grad_norm": 0.4228399097919464,
      "learning_rate": 3.0042687222843888e-06,
      "loss": 0.6851,
      "step": 9632
    },
    {
      "epoch": 0.39922914335446974,
      "grad_norm": 0.43061158061027527,
      "learning_rate": 3.0040615027560198e-06,
      "loss": 0.7168,
      "step": 9633
    },
    {
      "epoch": 0.3992705872601434,
      "grad_norm": 0.44774770736694336,
      "learning_rate": 3.003854283227652e-06,
      "loss": 0.7216,
      "step": 9634
    },
    {
      "epoch": 0.3993120311658171,
      "grad_norm": 0.4778788387775421,
      "learning_rate": 3.0036470636992834e-06,
      "loss": 0.7458,
      "step": 9635
    },
    {
      "epoch": 0.3993534750714907,
      "grad_norm": 0.405484676361084,
      "learning_rate": 3.003439844170915e-06,
      "loss": 0.6963,
      "step": 9636
    },
    {
      "epoch": 0.3993949189771644,
      "grad_norm": 0.4083288609981537,
      "learning_rate": 3.0032326246425466e-06,
      "loss": 0.667,
      "step": 9637
    },
    {
      "epoch": 0.39943636288283807,
      "grad_norm": 0.4395633339881897,
      "learning_rate": 3.0030254051141784e-06,
      "loss": 0.731,
      "step": 9638
    },
    {
      "epoch": 0.39947780678851175,
      "grad_norm": 0.46842095255851746,
      "learning_rate": 3.0028181855858098e-06,
      "loss": 0.6823,
      "step": 9639
    },
    {
      "epoch": 0.3995192506941854,
      "grad_norm": 0.4304969012737274,
      "learning_rate": 3.0026109660574416e-06,
      "loss": 0.6257,
      "step": 9640
    },
    {
      "epoch": 0.3995606945998591,
      "grad_norm": 0.42316973209381104,
      "learning_rate": 3.0024037465290734e-06,
      "loss": 0.698,
      "step": 9641
    },
    {
      "epoch": 0.3996021385055328,
      "grad_norm": 0.42654502391815186,
      "learning_rate": 3.0021965270007048e-06,
      "loss": 0.7192,
      "step": 9642
    },
    {
      "epoch": 0.39964358241120645,
      "grad_norm": 0.45171642303466797,
      "learning_rate": 3.0019893074723366e-06,
      "loss": 0.7605,
      "step": 9643
    },
    {
      "epoch": 0.3996850263168801,
      "grad_norm": 0.43299993872642517,
      "learning_rate": 3.001782087943968e-06,
      "loss": 0.7301,
      "step": 9644
    },
    {
      "epoch": 0.39972647022255375,
      "grad_norm": 0.41101953387260437,
      "learning_rate": 3.0015748684155998e-06,
      "loss": 0.6953,
      "step": 9645
    },
    {
      "epoch": 0.39976791412822743,
      "grad_norm": 0.4370548129081726,
      "learning_rate": 3.001367648887231e-06,
      "loss": 0.7325,
      "step": 9646
    },
    {
      "epoch": 0.3998093580339011,
      "grad_norm": 0.4835684299468994,
      "learning_rate": 3.001160429358863e-06,
      "loss": 0.7172,
      "step": 9647
    },
    {
      "epoch": 0.3998508019395748,
      "grad_norm": 0.4168919622898102,
      "learning_rate": 3.0009532098304943e-06,
      "loss": 0.6801,
      "step": 9648
    },
    {
      "epoch": 0.39989224584524846,
      "grad_norm": 0.42068737745285034,
      "learning_rate": 3.000745990302126e-06,
      "loss": 0.7227,
      "step": 9649
    },
    {
      "epoch": 0.39993368975092214,
      "grad_norm": 0.426327109336853,
      "learning_rate": 3.0005387707737584e-06,
      "loss": 0.703,
      "step": 9650
    },
    {
      "epoch": 0.3999751336565958,
      "grad_norm": 0.4372767210006714,
      "learning_rate": 3.0003315512453894e-06,
      "loss": 0.7252,
      "step": 9651
    },
    {
      "epoch": 0.4000165775622695,
      "grad_norm": 0.39417994022369385,
      "learning_rate": 3.0001243317170216e-06,
      "loss": 0.7095,
      "step": 9652
    },
    {
      "epoch": 0.4000580214679431,
      "grad_norm": 0.45287519693374634,
      "learning_rate": 2.999917112188653e-06,
      "loss": 0.7617,
      "step": 9653
    },
    {
      "epoch": 0.4000994653736168,
      "grad_norm": 0.39047256112098694,
      "learning_rate": 2.9997098926602848e-06,
      "loss": 0.6921,
      "step": 9654
    },
    {
      "epoch": 0.40014090927929047,
      "grad_norm": 0.43514031171798706,
      "learning_rate": 2.999502673131916e-06,
      "loss": 0.7384,
      "step": 9655
    },
    {
      "epoch": 0.40018235318496415,
      "grad_norm": 0.43791842460632324,
      "learning_rate": 2.999295453603548e-06,
      "loss": 0.6835,
      "step": 9656
    },
    {
      "epoch": 0.4002237970906378,
      "grad_norm": 0.365914523601532,
      "learning_rate": 2.9990882340751794e-06,
      "loss": 0.6929,
      "step": 9657
    },
    {
      "epoch": 0.4002652409963115,
      "grad_norm": 0.385794073343277,
      "learning_rate": 2.998881014546811e-06,
      "loss": 0.6487,
      "step": 9658
    },
    {
      "epoch": 0.4003066849019852,
      "grad_norm": 0.4497039020061493,
      "learning_rate": 2.998673795018443e-06,
      "loss": 0.6887,
      "step": 9659
    },
    {
      "epoch": 0.40034812880765885,
      "grad_norm": 0.3803664743900299,
      "learning_rate": 2.9984665754900744e-06,
      "loss": 0.667,
      "step": 9660
    },
    {
      "epoch": 0.40038957271333253,
      "grad_norm": 0.40676215291023254,
      "learning_rate": 2.998259355961706e-06,
      "loss": 0.7122,
      "step": 9661
    },
    {
      "epoch": 0.40043101661900615,
      "grad_norm": 0.5099861025810242,
      "learning_rate": 2.9980521364333376e-06,
      "loss": 0.7947,
      "step": 9662
    },
    {
      "epoch": 0.40047246052467983,
      "grad_norm": 0.4264516830444336,
      "learning_rate": 2.9978449169049694e-06,
      "loss": 0.6931,
      "step": 9663
    },
    {
      "epoch": 0.4005139044303535,
      "grad_norm": 0.40377357602119446,
      "learning_rate": 2.9976376973766008e-06,
      "loss": 0.7192,
      "step": 9664
    },
    {
      "epoch": 0.4005553483360272,
      "grad_norm": 0.40250930190086365,
      "learning_rate": 2.9974304778482326e-06,
      "loss": 0.6815,
      "step": 9665
    },
    {
      "epoch": 0.40059679224170086,
      "grad_norm": 0.4150809347629547,
      "learning_rate": 2.997223258319864e-06,
      "loss": 0.6731,
      "step": 9666
    },
    {
      "epoch": 0.40063823614737454,
      "grad_norm": 0.3994804322719574,
      "learning_rate": 2.9970160387914958e-06,
      "loss": 0.6965,
      "step": 9667
    },
    {
      "epoch": 0.4006796800530482,
      "grad_norm": 0.4262216091156006,
      "learning_rate": 2.996808819263128e-06,
      "loss": 0.7173,
      "step": 9668
    },
    {
      "epoch": 0.4007211239587219,
      "grad_norm": 0.4071671962738037,
      "learning_rate": 2.9966015997347594e-06,
      "loss": 0.6593,
      "step": 9669
    },
    {
      "epoch": 0.4007625678643955,
      "grad_norm": 0.4282101094722748,
      "learning_rate": 2.996394380206391e-06,
      "loss": 0.6907,
      "step": 9670
    },
    {
      "epoch": 0.4008040117700692,
      "grad_norm": 0.3996165096759796,
      "learning_rate": 2.9961871606780226e-06,
      "loss": 0.7192,
      "step": 9671
    },
    {
      "epoch": 0.40084545567574287,
      "grad_norm": 0.4529103934764862,
      "learning_rate": 2.9959799411496544e-06,
      "loss": 0.7654,
      "step": 9672
    },
    {
      "epoch": 0.40088689958141654,
      "grad_norm": 0.3931501805782318,
      "learning_rate": 2.9957727216212858e-06,
      "loss": 0.6749,
      "step": 9673
    },
    {
      "epoch": 0.4009283434870902,
      "grad_norm": 0.42044511437416077,
      "learning_rate": 2.9955655020929176e-06,
      "loss": 0.7065,
      "step": 9674
    },
    {
      "epoch": 0.4009697873927639,
      "grad_norm": 0.4241681694984436,
      "learning_rate": 2.995358282564549e-06,
      "loss": 0.735,
      "step": 9675
    },
    {
      "epoch": 0.4010112312984376,
      "grad_norm": 0.37090548872947693,
      "learning_rate": 2.9951510630361808e-06,
      "loss": 0.6632,
      "step": 9676
    },
    {
      "epoch": 0.40105267520411125,
      "grad_norm": 0.41652169823646545,
      "learning_rate": 2.9949438435078126e-06,
      "loss": 0.7278,
      "step": 9677
    },
    {
      "epoch": 0.40109411910978493,
      "grad_norm": 0.3859569728374481,
      "learning_rate": 2.994736623979444e-06,
      "loss": 0.663,
      "step": 9678
    },
    {
      "epoch": 0.40113556301545855,
      "grad_norm": 0.40043866634368896,
      "learning_rate": 2.9945294044510758e-06,
      "loss": 0.668,
      "step": 9679
    },
    {
      "epoch": 0.40117700692113223,
      "grad_norm": 0.4198809266090393,
      "learning_rate": 2.994322184922707e-06,
      "loss": 0.6982,
      "step": 9680
    },
    {
      "epoch": 0.4012184508268059,
      "grad_norm": 0.3963892161846161,
      "learning_rate": 2.994114965394339e-06,
      "loss": 0.7035,
      "step": 9681
    },
    {
      "epoch": 0.4012598947324796,
      "grad_norm": 0.40620893239974976,
      "learning_rate": 2.9939077458659704e-06,
      "loss": 0.7153,
      "step": 9682
    },
    {
      "epoch": 0.40130133863815326,
      "grad_norm": 0.409356951713562,
      "learning_rate": 2.993700526337602e-06,
      "loss": 0.691,
      "step": 9683
    },
    {
      "epoch": 0.40134278254382694,
      "grad_norm": 0.40759673714637756,
      "learning_rate": 2.9934933068092344e-06,
      "loss": 0.6593,
      "step": 9684
    },
    {
      "epoch": 0.4013842264495006,
      "grad_norm": 0.4362272620201111,
      "learning_rate": 2.9932860872808654e-06,
      "loss": 0.6971,
      "step": 9685
    },
    {
      "epoch": 0.4014256703551743,
      "grad_norm": 0.4030139744281769,
      "learning_rate": 2.9930788677524976e-06,
      "loss": 0.6505,
      "step": 9686
    },
    {
      "epoch": 0.40146711426084797,
      "grad_norm": 0.44515135884284973,
      "learning_rate": 2.992871648224129e-06,
      "loss": 0.7523,
      "step": 9687
    },
    {
      "epoch": 0.4015085581665216,
      "grad_norm": 0.38346153497695923,
      "learning_rate": 2.9926644286957608e-06,
      "loss": 0.6327,
      "step": 9688
    },
    {
      "epoch": 0.40155000207219527,
      "grad_norm": 0.41850608587265015,
      "learning_rate": 2.992457209167392e-06,
      "loss": 0.6801,
      "step": 9689
    },
    {
      "epoch": 0.40159144597786894,
      "grad_norm": 0.3929440975189209,
      "learning_rate": 2.992249989639024e-06,
      "loss": 0.7134,
      "step": 9690
    },
    {
      "epoch": 0.4016328898835426,
      "grad_norm": 0.4518125653266907,
      "learning_rate": 2.9920427701106554e-06,
      "loss": 0.7428,
      "step": 9691
    },
    {
      "epoch": 0.4016743337892163,
      "grad_norm": 0.4130164384841919,
      "learning_rate": 2.991835550582287e-06,
      "loss": 0.6401,
      "step": 9692
    },
    {
      "epoch": 0.40171577769489,
      "grad_norm": 0.43333011865615845,
      "learning_rate": 2.991628331053919e-06,
      "loss": 0.7095,
      "step": 9693
    },
    {
      "epoch": 0.40175722160056365,
      "grad_norm": 0.4257073402404785,
      "learning_rate": 2.9914211115255504e-06,
      "loss": 0.6899,
      "step": 9694
    },
    {
      "epoch": 0.40179866550623733,
      "grad_norm": 0.4309791326522827,
      "learning_rate": 2.991213891997182e-06,
      "loss": 0.644,
      "step": 9695
    },
    {
      "epoch": 0.401840109411911,
      "grad_norm": 0.4238462746143341,
      "learning_rate": 2.9910066724688136e-06,
      "loss": 0.7166,
      "step": 9696
    },
    {
      "epoch": 0.4018815533175846,
      "grad_norm": 0.37391698360443115,
      "learning_rate": 2.9907994529404454e-06,
      "loss": 0.7034,
      "step": 9697
    },
    {
      "epoch": 0.4019229972232583,
      "grad_norm": 0.39448216557502747,
      "learning_rate": 2.9905922334120768e-06,
      "loss": 0.6835,
      "step": 9698
    },
    {
      "epoch": 0.401964441128932,
      "grad_norm": 0.4246854782104492,
      "learning_rate": 2.9903850138837086e-06,
      "loss": 0.6943,
      "step": 9699
    },
    {
      "epoch": 0.40200588503460566,
      "grad_norm": 0.47914642095565796,
      "learning_rate": 2.99017779435534e-06,
      "loss": 0.7328,
      "step": 9700
    },
    {
      "epoch": 0.40204732894027934,
      "grad_norm": 0.4108937382698059,
      "learning_rate": 2.9899705748269718e-06,
      "loss": 0.7351,
      "step": 9701
    },
    {
      "epoch": 0.402088772845953,
      "grad_norm": 0.47547468543052673,
      "learning_rate": 2.989763355298604e-06,
      "loss": 0.7439,
      "step": 9702
    },
    {
      "epoch": 0.4021302167516267,
      "grad_norm": 0.3836901783943176,
      "learning_rate": 2.9895561357702354e-06,
      "loss": 0.698,
      "step": 9703
    },
    {
      "epoch": 0.40217166065730037,
      "grad_norm": 0.4137960374355316,
      "learning_rate": 2.989348916241867e-06,
      "loss": 0.6604,
      "step": 9704
    },
    {
      "epoch": 0.402213104562974,
      "grad_norm": 0.4403729736804962,
      "learning_rate": 2.9891416967134986e-06,
      "loss": 0.7084,
      "step": 9705
    },
    {
      "epoch": 0.40225454846864767,
      "grad_norm": 0.38235464692115784,
      "learning_rate": 2.9889344771851304e-06,
      "loss": 0.6726,
      "step": 9706
    },
    {
      "epoch": 0.40229599237432134,
      "grad_norm": 0.4330444037914276,
      "learning_rate": 2.9887272576567618e-06,
      "loss": 0.689,
      "step": 9707
    },
    {
      "epoch": 0.402337436279995,
      "grad_norm": 0.41801586747169495,
      "learning_rate": 2.9885200381283936e-06,
      "loss": 0.6974,
      "step": 9708
    },
    {
      "epoch": 0.4023788801856687,
      "grad_norm": 0.411640465259552,
      "learning_rate": 2.988312818600025e-06,
      "loss": 0.6982,
      "step": 9709
    },
    {
      "epoch": 0.4024203240913424,
      "grad_norm": 0.45801541209220886,
      "learning_rate": 2.9881055990716568e-06,
      "loss": 0.688,
      "step": 9710
    },
    {
      "epoch": 0.40246176799701605,
      "grad_norm": 0.4810921847820282,
      "learning_rate": 2.9878983795432886e-06,
      "loss": 0.7605,
      "step": 9711
    },
    {
      "epoch": 0.40250321190268973,
      "grad_norm": 0.4124468266963959,
      "learning_rate": 2.98769116001492e-06,
      "loss": 0.6692,
      "step": 9712
    },
    {
      "epoch": 0.4025446558083634,
      "grad_norm": 0.407736599445343,
      "learning_rate": 2.9874839404865518e-06,
      "loss": 0.7416,
      "step": 9713
    },
    {
      "epoch": 0.402586099714037,
      "grad_norm": 0.43104997277259827,
      "learning_rate": 2.987276720958183e-06,
      "loss": 0.6719,
      "step": 9714
    },
    {
      "epoch": 0.4026275436197107,
      "grad_norm": 0.3871555030345917,
      "learning_rate": 2.987069501429815e-06,
      "loss": 0.6791,
      "step": 9715
    },
    {
      "epoch": 0.4026689875253844,
      "grad_norm": 0.43013012409210205,
      "learning_rate": 2.9868622819014464e-06,
      "loss": 0.7112,
      "step": 9716
    },
    {
      "epoch": 0.40271043143105806,
      "grad_norm": 0.390768826007843,
      "learning_rate": 2.986655062373078e-06,
      "loss": 0.7297,
      "step": 9717
    },
    {
      "epoch": 0.40275187533673174,
      "grad_norm": 0.41533708572387695,
      "learning_rate": 2.9864478428447096e-06,
      "loss": 0.6846,
      "step": 9718
    },
    {
      "epoch": 0.4027933192424054,
      "grad_norm": 0.4183333218097687,
      "learning_rate": 2.9862406233163414e-06,
      "loss": 0.66,
      "step": 9719
    },
    {
      "epoch": 0.4028347631480791,
      "grad_norm": 0.42538776993751526,
      "learning_rate": 2.9860334037879736e-06,
      "loss": 0.7168,
      "step": 9720
    },
    {
      "epoch": 0.40287620705375277,
      "grad_norm": 0.3932821750640869,
      "learning_rate": 2.985826184259605e-06,
      "loss": 0.7111,
      "step": 9721
    },
    {
      "epoch": 0.40291765095942644,
      "grad_norm": 0.40207764506340027,
      "learning_rate": 2.985618964731237e-06,
      "loss": 0.725,
      "step": 9722
    },
    {
      "epoch": 0.40295909486510006,
      "grad_norm": 0.3952674865722656,
      "learning_rate": 2.985411745202868e-06,
      "loss": 0.637,
      "step": 9723
    },
    {
      "epoch": 0.40300053877077374,
      "grad_norm": 0.3992679715156555,
      "learning_rate": 2.9852045256745e-06,
      "loss": 0.6704,
      "step": 9724
    },
    {
      "epoch": 0.4030419826764474,
      "grad_norm": 0.4539913535118103,
      "learning_rate": 2.9849973061461314e-06,
      "loss": 0.7115,
      "step": 9725
    },
    {
      "epoch": 0.4030834265821211,
      "grad_norm": 0.4348430037498474,
      "learning_rate": 2.984790086617763e-06,
      "loss": 0.7009,
      "step": 9726
    },
    {
      "epoch": 0.4031248704877948,
      "grad_norm": 0.4055812358856201,
      "learning_rate": 2.9845828670893946e-06,
      "loss": 0.6998,
      "step": 9727
    },
    {
      "epoch": 0.40316631439346845,
      "grad_norm": 0.4385034441947937,
      "learning_rate": 2.9843756475610264e-06,
      "loss": 0.7321,
      "step": 9728
    },
    {
      "epoch": 0.4032077582991421,
      "grad_norm": 0.44498491287231445,
      "learning_rate": 2.984168428032658e-06,
      "loss": 0.7411,
      "step": 9729
    },
    {
      "epoch": 0.4032492022048158,
      "grad_norm": 0.39831238985061646,
      "learning_rate": 2.9839612085042896e-06,
      "loss": 0.681,
      "step": 9730
    },
    {
      "epoch": 0.4032906461104894,
      "grad_norm": 0.4489978551864624,
      "learning_rate": 2.9837539889759214e-06,
      "loss": 0.7593,
      "step": 9731
    },
    {
      "epoch": 0.4033320900161631,
      "grad_norm": 0.43996408581733704,
      "learning_rate": 2.9835467694475528e-06,
      "loss": 0.7134,
      "step": 9732
    },
    {
      "epoch": 0.4033735339218368,
      "grad_norm": 0.4582103192806244,
      "learning_rate": 2.9833395499191846e-06,
      "loss": 0.7085,
      "step": 9733
    },
    {
      "epoch": 0.40341497782751046,
      "grad_norm": 0.4034340977668762,
      "learning_rate": 2.983132330390816e-06,
      "loss": 0.7085,
      "step": 9734
    },
    {
      "epoch": 0.40345642173318413,
      "grad_norm": 0.390758216381073,
      "learning_rate": 2.9829251108624478e-06,
      "loss": 0.6543,
      "step": 9735
    },
    {
      "epoch": 0.4034978656388578,
      "grad_norm": 0.4019414782524109,
      "learning_rate": 2.982717891334079e-06,
      "loss": 0.6099,
      "step": 9736
    },
    {
      "epoch": 0.4035393095445315,
      "grad_norm": 0.3674839735031128,
      "learning_rate": 2.9825106718057114e-06,
      "loss": 0.6615,
      "step": 9737
    },
    {
      "epoch": 0.40358075345020517,
      "grad_norm": 0.39651063084602356,
      "learning_rate": 2.982303452277343e-06,
      "loss": 0.7063,
      "step": 9738
    },
    {
      "epoch": 0.40362219735587884,
      "grad_norm": 0.45791327953338623,
      "learning_rate": 2.9820962327489746e-06,
      "loss": 0.7656,
      "step": 9739
    },
    {
      "epoch": 0.40366364126155246,
      "grad_norm": 0.4377143979072571,
      "learning_rate": 2.9818890132206064e-06,
      "loss": 0.6709,
      "step": 9740
    },
    {
      "epoch": 0.40370508516722614,
      "grad_norm": 0.4366690516471863,
      "learning_rate": 2.9816817936922378e-06,
      "loss": 0.7251,
      "step": 9741
    },
    {
      "epoch": 0.4037465290728998,
      "grad_norm": 0.45985424518585205,
      "learning_rate": 2.9814745741638696e-06,
      "loss": 0.7292,
      "step": 9742
    },
    {
      "epoch": 0.4037879729785735,
      "grad_norm": 0.41890010237693787,
      "learning_rate": 2.981267354635501e-06,
      "loss": 0.7183,
      "step": 9743
    },
    {
      "epoch": 0.40382941688424717,
      "grad_norm": 0.4206309914588928,
      "learning_rate": 2.9810601351071328e-06,
      "loss": 0.717,
      "step": 9744
    },
    {
      "epoch": 0.40387086078992085,
      "grad_norm": 0.40354251861572266,
      "learning_rate": 2.9808529155787646e-06,
      "loss": 0.6577,
      "step": 9745
    },
    {
      "epoch": 0.4039123046955945,
      "grad_norm": 0.41606658697128296,
      "learning_rate": 2.980645696050396e-06,
      "loss": 0.6763,
      "step": 9746
    },
    {
      "epoch": 0.4039537486012682,
      "grad_norm": 0.40557873249053955,
      "learning_rate": 2.9804384765220278e-06,
      "loss": 0.6495,
      "step": 9747
    },
    {
      "epoch": 0.4039951925069419,
      "grad_norm": 0.4139299988746643,
      "learning_rate": 2.980231256993659e-06,
      "loss": 0.6874,
      "step": 9748
    },
    {
      "epoch": 0.4040366364126155,
      "grad_norm": 0.4618210792541504,
      "learning_rate": 2.980024037465291e-06,
      "loss": 0.728,
      "step": 9749
    },
    {
      "epoch": 0.4040780803182892,
      "grad_norm": 0.4006025493144989,
      "learning_rate": 2.9798168179369224e-06,
      "loss": 0.6624,
      "step": 9750
    },
    {
      "epoch": 0.40411952422396286,
      "grad_norm": 0.41140055656433105,
      "learning_rate": 2.979609598408554e-06,
      "loss": 0.6853,
      "step": 9751
    },
    {
      "epoch": 0.40416096812963653,
      "grad_norm": 0.3993538022041321,
      "learning_rate": 2.9794023788801856e-06,
      "loss": 0.7245,
      "step": 9752
    },
    {
      "epoch": 0.4042024120353102,
      "grad_norm": 0.4154256284236908,
      "learning_rate": 2.9791951593518174e-06,
      "loss": 0.7375,
      "step": 9753
    },
    {
      "epoch": 0.4042438559409839,
      "grad_norm": 0.38683274388313293,
      "learning_rate": 2.9789879398234496e-06,
      "loss": 0.6671,
      "step": 9754
    },
    {
      "epoch": 0.40428529984665756,
      "grad_norm": 0.398532509803772,
      "learning_rate": 2.978780720295081e-06,
      "loss": 0.7321,
      "step": 9755
    },
    {
      "epoch": 0.40432674375233124,
      "grad_norm": 0.38536715507507324,
      "learning_rate": 2.978573500766713e-06,
      "loss": 0.6814,
      "step": 9756
    },
    {
      "epoch": 0.40436818765800486,
      "grad_norm": 0.41558557748794556,
      "learning_rate": 2.978366281238344e-06,
      "loss": 0.6821,
      "step": 9757
    },
    {
      "epoch": 0.40440963156367854,
      "grad_norm": 0.39867663383483887,
      "learning_rate": 2.978159061709976e-06,
      "loss": 0.7092,
      "step": 9758
    },
    {
      "epoch": 0.4044510754693522,
      "grad_norm": 0.41024357080459595,
      "learning_rate": 2.9779518421816074e-06,
      "loss": 0.686,
      "step": 9759
    },
    {
      "epoch": 0.4044925193750259,
      "grad_norm": 0.3874228894710541,
      "learning_rate": 2.977744622653239e-06,
      "loss": 0.6919,
      "step": 9760
    },
    {
      "epoch": 0.40453396328069957,
      "grad_norm": 0.3987756371498108,
      "learning_rate": 2.9775374031248706e-06,
      "loss": 0.647,
      "step": 9761
    },
    {
      "epoch": 0.40457540718637325,
      "grad_norm": 0.384509414434433,
      "learning_rate": 2.9773301835965024e-06,
      "loss": 0.6512,
      "step": 9762
    },
    {
      "epoch": 0.4046168510920469,
      "grad_norm": 0.4130660891532898,
      "learning_rate": 2.977122964068134e-06,
      "loss": 0.6956,
      "step": 9763
    },
    {
      "epoch": 0.4046582949977206,
      "grad_norm": 0.41469961404800415,
      "learning_rate": 2.9769157445397656e-06,
      "loss": 0.6439,
      "step": 9764
    },
    {
      "epoch": 0.4046997389033943,
      "grad_norm": 0.43319326639175415,
      "learning_rate": 2.9767085250113974e-06,
      "loss": 0.7203,
      "step": 9765
    },
    {
      "epoch": 0.4047411828090679,
      "grad_norm": 0.42562583088874817,
      "learning_rate": 2.9765013054830288e-06,
      "loss": 0.7291,
      "step": 9766
    },
    {
      "epoch": 0.4047826267147416,
      "grad_norm": 0.3917831778526306,
      "learning_rate": 2.9762940859546606e-06,
      "loss": 0.6954,
      "step": 9767
    },
    {
      "epoch": 0.40482407062041526,
      "grad_norm": 0.4059503376483917,
      "learning_rate": 2.976086866426292e-06,
      "loss": 0.6973,
      "step": 9768
    },
    {
      "epoch": 0.40486551452608893,
      "grad_norm": 0.415391206741333,
      "learning_rate": 2.9758796468979238e-06,
      "loss": 0.7131,
      "step": 9769
    },
    {
      "epoch": 0.4049069584317626,
      "grad_norm": 0.4100469648838043,
      "learning_rate": 2.975672427369555e-06,
      "loss": 0.6978,
      "step": 9770
    },
    {
      "epoch": 0.4049484023374363,
      "grad_norm": 0.4291653335094452,
      "learning_rate": 2.9754652078411874e-06,
      "loss": 0.689,
      "step": 9771
    },
    {
      "epoch": 0.40498984624310996,
      "grad_norm": 0.4807572066783905,
      "learning_rate": 2.975257988312819e-06,
      "loss": 0.6653,
      "step": 9772
    },
    {
      "epoch": 0.40503129014878364,
      "grad_norm": 0.3876575529575348,
      "learning_rate": 2.9750507687844506e-06,
      "loss": 0.6962,
      "step": 9773
    },
    {
      "epoch": 0.4050727340544573,
      "grad_norm": 0.42250585556030273,
      "learning_rate": 2.9748435492560824e-06,
      "loss": 0.6829,
      "step": 9774
    },
    {
      "epoch": 0.40511417796013094,
      "grad_norm": 0.3840040862560272,
      "learning_rate": 2.9746363297277138e-06,
      "loss": 0.6383,
      "step": 9775
    },
    {
      "epoch": 0.4051556218658046,
      "grad_norm": 0.4058098793029785,
      "learning_rate": 2.9744291101993456e-06,
      "loss": 0.667,
      "step": 9776
    },
    {
      "epoch": 0.4051970657714783,
      "grad_norm": 0.4196680188179016,
      "learning_rate": 2.974221890670977e-06,
      "loss": 0.6843,
      "step": 9777
    },
    {
      "epoch": 0.40523850967715197,
      "grad_norm": 0.41006314754486084,
      "learning_rate": 2.9740146711426088e-06,
      "loss": 0.6853,
      "step": 9778
    },
    {
      "epoch": 0.40527995358282565,
      "grad_norm": 0.42810681462287903,
      "learning_rate": 2.97380745161424e-06,
      "loss": 0.7095,
      "step": 9779
    },
    {
      "epoch": 0.4053213974884993,
      "grad_norm": 0.3791421055793762,
      "learning_rate": 2.973600232085872e-06,
      "loss": 0.6659,
      "step": 9780
    },
    {
      "epoch": 0.405362841394173,
      "grad_norm": 0.4133743941783905,
      "learning_rate": 2.973393012557504e-06,
      "loss": 0.688,
      "step": 9781
    },
    {
      "epoch": 0.4054042852998467,
      "grad_norm": 0.39762014150619507,
      "learning_rate": 2.973185793029135e-06,
      "loss": 0.6975,
      "step": 9782
    },
    {
      "epoch": 0.40544572920552036,
      "grad_norm": 0.432325154542923,
      "learning_rate": 2.972978573500767e-06,
      "loss": 0.6964,
      "step": 9783
    },
    {
      "epoch": 0.405487173111194,
      "grad_norm": 0.3798847794532776,
      "learning_rate": 2.9727713539723984e-06,
      "loss": 0.6218,
      "step": 9784
    },
    {
      "epoch": 0.40552861701686765,
      "grad_norm": 0.40270090103149414,
      "learning_rate": 2.97256413444403e-06,
      "loss": 0.6697,
      "step": 9785
    },
    {
      "epoch": 0.40557006092254133,
      "grad_norm": 0.39199671149253845,
      "learning_rate": 2.9723569149156616e-06,
      "loss": 0.7109,
      "step": 9786
    },
    {
      "epoch": 0.405611504828215,
      "grad_norm": 0.3734724819660187,
      "learning_rate": 2.972149695387294e-06,
      "loss": 0.6576,
      "step": 9787
    },
    {
      "epoch": 0.4056529487338887,
      "grad_norm": 0.422929584980011,
      "learning_rate": 2.9719424758589248e-06,
      "loss": 0.7195,
      "step": 9788
    },
    {
      "epoch": 0.40569439263956236,
      "grad_norm": 0.3927742838859558,
      "learning_rate": 2.971735256330557e-06,
      "loss": 0.6973,
      "step": 9789
    },
    {
      "epoch": 0.40573583654523604,
      "grad_norm": 0.41834065318107605,
      "learning_rate": 2.971528036802189e-06,
      "loss": 0.7046,
      "step": 9790
    },
    {
      "epoch": 0.4057772804509097,
      "grad_norm": 0.40593087673187256,
      "learning_rate": 2.97132081727382e-06,
      "loss": 0.7168,
      "step": 9791
    },
    {
      "epoch": 0.40581872435658334,
      "grad_norm": 0.4382356107234955,
      "learning_rate": 2.971113597745452e-06,
      "loss": 0.7053,
      "step": 9792
    },
    {
      "epoch": 0.405860168262257,
      "grad_norm": 0.43486127257347107,
      "learning_rate": 2.9709063782170834e-06,
      "loss": 0.6655,
      "step": 9793
    },
    {
      "epoch": 0.4059016121679307,
      "grad_norm": 0.4384447932243347,
      "learning_rate": 2.970699158688715e-06,
      "loss": 0.6996,
      "step": 9794
    },
    {
      "epoch": 0.40594305607360437,
      "grad_norm": 0.37911680340766907,
      "learning_rate": 2.9704919391603466e-06,
      "loss": 0.6279,
      "step": 9795
    },
    {
      "epoch": 0.40598449997927805,
      "grad_norm": 0.4113192558288574,
      "learning_rate": 2.9702847196319784e-06,
      "loss": 0.7108,
      "step": 9796
    },
    {
      "epoch": 0.4060259438849517,
      "grad_norm": 0.3932170271873474,
      "learning_rate": 2.9700775001036098e-06,
      "loss": 0.6826,
      "step": 9797
    },
    {
      "epoch": 0.4060673877906254,
      "grad_norm": 0.4573267996311188,
      "learning_rate": 2.9698702805752416e-06,
      "loss": 0.7081,
      "step": 9798
    },
    {
      "epoch": 0.4061088316962991,
      "grad_norm": 0.38210076093673706,
      "learning_rate": 2.9696630610468734e-06,
      "loss": 0.6199,
      "step": 9799
    },
    {
      "epoch": 0.40615027560197275,
      "grad_norm": 0.43597087264060974,
      "learning_rate": 2.9694558415185048e-06,
      "loss": 0.7279,
      "step": 9800
    },
    {
      "epoch": 0.4061917195076464,
      "grad_norm": 0.42689868807792664,
      "learning_rate": 2.9692486219901366e-06,
      "loss": 0.7275,
      "step": 9801
    },
    {
      "epoch": 0.40623316341332005,
      "grad_norm": 0.4760863482952118,
      "learning_rate": 2.969041402461768e-06,
      "loss": 0.6757,
      "step": 9802
    },
    {
      "epoch": 0.40627460731899373,
      "grad_norm": 0.3733714818954468,
      "learning_rate": 2.9688341829333998e-06,
      "loss": 0.6475,
      "step": 9803
    },
    {
      "epoch": 0.4063160512246674,
      "grad_norm": 0.39665886759757996,
      "learning_rate": 2.968626963405031e-06,
      "loss": 0.6772,
      "step": 9804
    },
    {
      "epoch": 0.4063574951303411,
      "grad_norm": 0.44022735953330994,
      "learning_rate": 2.9684197438766634e-06,
      "loss": 0.6558,
      "step": 9805
    },
    {
      "epoch": 0.40639893903601476,
      "grad_norm": 0.41237711906433105,
      "learning_rate": 2.968212524348295e-06,
      "loss": 0.6951,
      "step": 9806
    },
    {
      "epoch": 0.40644038294168844,
      "grad_norm": 0.3845568299293518,
      "learning_rate": 2.9680053048199266e-06,
      "loss": 0.644,
      "step": 9807
    },
    {
      "epoch": 0.4064818268473621,
      "grad_norm": 0.3948551118373871,
      "learning_rate": 2.9677980852915584e-06,
      "loss": 0.6838,
      "step": 9808
    },
    {
      "epoch": 0.4065232707530358,
      "grad_norm": 0.4554884433746338,
      "learning_rate": 2.96759086576319e-06,
      "loss": 0.7437,
      "step": 9809
    },
    {
      "epoch": 0.4065647146587094,
      "grad_norm": 0.4134681522846222,
      "learning_rate": 2.9673836462348216e-06,
      "loss": 0.6843,
      "step": 9810
    },
    {
      "epoch": 0.4066061585643831,
      "grad_norm": 0.37341564893722534,
      "learning_rate": 2.967176426706453e-06,
      "loss": 0.6799,
      "step": 9811
    },
    {
      "epoch": 0.40664760247005677,
      "grad_norm": 0.40386343002319336,
      "learning_rate": 2.966969207178085e-06,
      "loss": 0.6948,
      "step": 9812
    },
    {
      "epoch": 0.40668904637573045,
      "grad_norm": 0.3902512192726135,
      "learning_rate": 2.966761987649716e-06,
      "loss": 0.6678,
      "step": 9813
    },
    {
      "epoch": 0.4067304902814041,
      "grad_norm": 0.4293665885925293,
      "learning_rate": 2.966554768121348e-06,
      "loss": 0.7361,
      "step": 9814
    },
    {
      "epoch": 0.4067719341870778,
      "grad_norm": 0.44609761238098145,
      "learning_rate": 2.96634754859298e-06,
      "loss": 0.7173,
      "step": 9815
    },
    {
      "epoch": 0.4068133780927515,
      "grad_norm": 0.43193814158439636,
      "learning_rate": 2.966140329064611e-06,
      "loss": 0.7697,
      "step": 9816
    },
    {
      "epoch": 0.40685482199842515,
      "grad_norm": 0.40452417731285095,
      "learning_rate": 2.965933109536243e-06,
      "loss": 0.6707,
      "step": 9817
    },
    {
      "epoch": 0.4068962659040988,
      "grad_norm": 0.41765305399894714,
      "learning_rate": 2.9657258900078744e-06,
      "loss": 0.7388,
      "step": 9818
    },
    {
      "epoch": 0.40693770980977245,
      "grad_norm": 0.4309346675872803,
      "learning_rate": 2.965518670479506e-06,
      "loss": 0.6675,
      "step": 9819
    },
    {
      "epoch": 0.40697915371544613,
      "grad_norm": 0.3872416019439697,
      "learning_rate": 2.9653114509511376e-06,
      "loss": 0.7013,
      "step": 9820
    },
    {
      "epoch": 0.4070205976211198,
      "grad_norm": 0.3716718256473541,
      "learning_rate": 2.96510423142277e-06,
      "loss": 0.6863,
      "step": 9821
    },
    {
      "epoch": 0.4070620415267935,
      "grad_norm": 0.41562730073928833,
      "learning_rate": 2.9648970118944008e-06,
      "loss": 0.7166,
      "step": 9822
    },
    {
      "epoch": 0.40710348543246716,
      "grad_norm": 0.4011440575122833,
      "learning_rate": 2.964689792366033e-06,
      "loss": 0.7153,
      "step": 9823
    },
    {
      "epoch": 0.40714492933814084,
      "grad_norm": 0.4375945031642914,
      "learning_rate": 2.964482572837665e-06,
      "loss": 0.7494,
      "step": 9824
    },
    {
      "epoch": 0.4071863732438145,
      "grad_norm": 0.4255179166793823,
      "learning_rate": 2.964275353309296e-06,
      "loss": 0.708,
      "step": 9825
    },
    {
      "epoch": 0.4072278171494882,
      "grad_norm": 0.40039414167404175,
      "learning_rate": 2.964068133780928e-06,
      "loss": 0.6779,
      "step": 9826
    },
    {
      "epoch": 0.4072692610551618,
      "grad_norm": 0.4201604723930359,
      "learning_rate": 2.9638609142525594e-06,
      "loss": 0.6969,
      "step": 9827
    },
    {
      "epoch": 0.4073107049608355,
      "grad_norm": 0.4251379072666168,
      "learning_rate": 2.963653694724191e-06,
      "loss": 0.7157,
      "step": 9828
    },
    {
      "epoch": 0.40735214886650917,
      "grad_norm": 0.4018380641937256,
      "learning_rate": 2.9634464751958226e-06,
      "loss": 0.6987,
      "step": 9829
    },
    {
      "epoch": 0.40739359277218284,
      "grad_norm": 0.41972389817237854,
      "learning_rate": 2.9632392556674544e-06,
      "loss": 0.7166,
      "step": 9830
    },
    {
      "epoch": 0.4074350366778565,
      "grad_norm": 0.46676549315452576,
      "learning_rate": 2.9630320361390858e-06,
      "loss": 0.6982,
      "step": 9831
    },
    {
      "epoch": 0.4074764805835302,
      "grad_norm": 0.4330548942089081,
      "learning_rate": 2.9628248166107176e-06,
      "loss": 0.7063,
      "step": 9832
    },
    {
      "epoch": 0.4075179244892039,
      "grad_norm": 0.38913047313690186,
      "learning_rate": 2.9626175970823494e-06,
      "loss": 0.6644,
      "step": 9833
    },
    {
      "epoch": 0.40755936839487755,
      "grad_norm": 0.4217050075531006,
      "learning_rate": 2.9624103775539808e-06,
      "loss": 0.6458,
      "step": 9834
    },
    {
      "epoch": 0.40760081230055123,
      "grad_norm": 0.421800434589386,
      "learning_rate": 2.9622031580256126e-06,
      "loss": 0.6853,
      "step": 9835
    },
    {
      "epoch": 0.40764225620622485,
      "grad_norm": 0.38288357853889465,
      "learning_rate": 2.961995938497244e-06,
      "loss": 0.6719,
      "step": 9836
    },
    {
      "epoch": 0.40768370011189853,
      "grad_norm": 0.4018184244632721,
      "learning_rate": 2.9617887189688758e-06,
      "loss": 0.7068,
      "step": 9837
    },
    {
      "epoch": 0.4077251440175722,
      "grad_norm": 0.4016752243041992,
      "learning_rate": 2.961581499440507e-06,
      "loss": 0.7034,
      "step": 9838
    },
    {
      "epoch": 0.4077665879232459,
      "grad_norm": 0.4374559819698334,
      "learning_rate": 2.9613742799121394e-06,
      "loss": 0.7039,
      "step": 9839
    },
    {
      "epoch": 0.40780803182891956,
      "grad_norm": 0.3954083025455475,
      "learning_rate": 2.9611670603837704e-06,
      "loss": 0.6494,
      "step": 9840
    },
    {
      "epoch": 0.40784947573459324,
      "grad_norm": 0.3769851326942444,
      "learning_rate": 2.9609598408554026e-06,
      "loss": 0.6768,
      "step": 9841
    },
    {
      "epoch": 0.4078909196402669,
      "grad_norm": 0.4377748370170593,
      "learning_rate": 2.9607526213270344e-06,
      "loss": 0.6736,
      "step": 9842
    },
    {
      "epoch": 0.4079323635459406,
      "grad_norm": 0.42467209696769714,
      "learning_rate": 2.960545401798666e-06,
      "loss": 0.6893,
      "step": 9843
    },
    {
      "epoch": 0.40797380745161427,
      "grad_norm": 0.4180726408958435,
      "learning_rate": 2.9603381822702976e-06,
      "loss": 0.6813,
      "step": 9844
    },
    {
      "epoch": 0.4080152513572879,
      "grad_norm": 0.3715086281299591,
      "learning_rate": 2.960130962741929e-06,
      "loss": 0.6371,
      "step": 9845
    },
    {
      "epoch": 0.40805669526296157,
      "grad_norm": 0.465617835521698,
      "learning_rate": 2.959923743213561e-06,
      "loss": 0.7207,
      "step": 9846
    },
    {
      "epoch": 0.40809813916863524,
      "grad_norm": 0.3955141007900238,
      "learning_rate": 2.959716523685192e-06,
      "loss": 0.6697,
      "step": 9847
    },
    {
      "epoch": 0.4081395830743089,
      "grad_norm": 0.40128272771835327,
      "learning_rate": 2.959509304156824e-06,
      "loss": 0.7043,
      "step": 9848
    },
    {
      "epoch": 0.4081810269799826,
      "grad_norm": 0.40640193223953247,
      "learning_rate": 2.9593020846284554e-06,
      "loss": 0.6284,
      "step": 9849
    },
    {
      "epoch": 0.4082224708856563,
      "grad_norm": 0.4107244312763214,
      "learning_rate": 2.959094865100087e-06,
      "loss": 0.6943,
      "step": 9850
    },
    {
      "epoch": 0.40826391479132995,
      "grad_norm": 0.4154118299484253,
      "learning_rate": 2.958887645571719e-06,
      "loss": 0.661,
      "step": 9851
    },
    {
      "epoch": 0.40830535869700363,
      "grad_norm": 0.39897677302360535,
      "learning_rate": 2.9586804260433504e-06,
      "loss": 0.6726,
      "step": 9852
    },
    {
      "epoch": 0.40834680260267725,
      "grad_norm": 0.40769824385643005,
      "learning_rate": 2.958473206514982e-06,
      "loss": 0.6705,
      "step": 9853
    },
    {
      "epoch": 0.4083882465083509,
      "grad_norm": 0.4450128972530365,
      "learning_rate": 2.9582659869866136e-06,
      "loss": 0.7395,
      "step": 9854
    },
    {
      "epoch": 0.4084296904140246,
      "grad_norm": 0.41259756684303284,
      "learning_rate": 2.958058767458246e-06,
      "loss": 0.7351,
      "step": 9855
    },
    {
      "epoch": 0.4084711343196983,
      "grad_norm": 0.4322452247142792,
      "learning_rate": 2.9578515479298768e-06,
      "loss": 0.6863,
      "step": 9856
    },
    {
      "epoch": 0.40851257822537196,
      "grad_norm": 0.4103268086910248,
      "learning_rate": 2.957644328401509e-06,
      "loss": 0.6294,
      "step": 9857
    },
    {
      "epoch": 0.40855402213104564,
      "grad_norm": 0.4311973452568054,
      "learning_rate": 2.95743710887314e-06,
      "loss": 0.6741,
      "step": 9858
    },
    {
      "epoch": 0.4085954660367193,
      "grad_norm": 0.3895007073879242,
      "learning_rate": 2.957229889344772e-06,
      "loss": 0.708,
      "step": 9859
    },
    {
      "epoch": 0.408636909942393,
      "grad_norm": 0.41853252053260803,
      "learning_rate": 2.957022669816404e-06,
      "loss": 0.7349,
      "step": 9860
    },
    {
      "epoch": 0.40867835384806667,
      "grad_norm": 0.41222086548805237,
      "learning_rate": 2.9568154502880354e-06,
      "loss": 0.6895,
      "step": 9861
    },
    {
      "epoch": 0.4087197977537403,
      "grad_norm": 0.4088389277458191,
      "learning_rate": 2.956608230759667e-06,
      "loss": 0.6392,
      "step": 9862
    },
    {
      "epoch": 0.40876124165941397,
      "grad_norm": 0.44018176198005676,
      "learning_rate": 2.9564010112312986e-06,
      "loss": 0.7249,
      "step": 9863
    },
    {
      "epoch": 0.40880268556508764,
      "grad_norm": 0.46858349442481995,
      "learning_rate": 2.9561937917029304e-06,
      "loss": 0.7843,
      "step": 9864
    },
    {
      "epoch": 0.4088441294707613,
      "grad_norm": 0.4034697115421295,
      "learning_rate": 2.9559865721745618e-06,
      "loss": 0.6829,
      "step": 9865
    },
    {
      "epoch": 0.408885573376435,
      "grad_norm": 0.3867449462413788,
      "learning_rate": 2.9557793526461936e-06,
      "loss": 0.6399,
      "step": 9866
    },
    {
      "epoch": 0.4089270172821087,
      "grad_norm": 0.4132276177406311,
      "learning_rate": 2.9555721331178254e-06,
      "loss": 0.6216,
      "step": 9867
    },
    {
      "epoch": 0.40896846118778235,
      "grad_norm": 0.43972450494766235,
      "learning_rate": 2.955364913589457e-06,
      "loss": 0.6926,
      "step": 9868
    },
    {
      "epoch": 0.409009905093456,
      "grad_norm": 0.3830566108226776,
      "learning_rate": 2.9551576940610886e-06,
      "loss": 0.6716,
      "step": 9869
    },
    {
      "epoch": 0.4090513489991297,
      "grad_norm": 0.4093276560306549,
      "learning_rate": 2.95495047453272e-06,
      "loss": 0.6647,
      "step": 9870
    },
    {
      "epoch": 0.4090927929048033,
      "grad_norm": 0.40445640683174133,
      "learning_rate": 2.954743255004352e-06,
      "loss": 0.6628,
      "step": 9871
    },
    {
      "epoch": 0.409134236810477,
      "grad_norm": 0.4183807969093323,
      "learning_rate": 2.954536035475983e-06,
      "loss": 0.6802,
      "step": 9872
    },
    {
      "epoch": 0.4091756807161507,
      "grad_norm": 0.40704983472824097,
      "learning_rate": 2.9543288159476154e-06,
      "loss": 0.7356,
      "step": 9873
    },
    {
      "epoch": 0.40921712462182436,
      "grad_norm": 0.3970617949962616,
      "learning_rate": 2.9541215964192464e-06,
      "loss": 0.7427,
      "step": 9874
    },
    {
      "epoch": 0.40925856852749803,
      "grad_norm": 0.42581823468208313,
      "learning_rate": 2.9539143768908786e-06,
      "loss": 0.7668,
      "step": 9875
    },
    {
      "epoch": 0.4093000124331717,
      "grad_norm": 0.3967044949531555,
      "learning_rate": 2.9537071573625104e-06,
      "loss": 0.678,
      "step": 9876
    },
    {
      "epoch": 0.4093414563388454,
      "grad_norm": 0.3935525417327881,
      "learning_rate": 2.953499937834142e-06,
      "loss": 0.679,
      "step": 9877
    },
    {
      "epoch": 0.40938290024451907,
      "grad_norm": 0.383810818195343,
      "learning_rate": 2.9532927183057736e-06,
      "loss": 0.6588,
      "step": 9878
    },
    {
      "epoch": 0.4094243441501927,
      "grad_norm": 0.40502607822418213,
      "learning_rate": 2.953085498777405e-06,
      "loss": 0.7495,
      "step": 9879
    },
    {
      "epoch": 0.40946578805586636,
      "grad_norm": 0.43088823556900024,
      "learning_rate": 2.952878279249037e-06,
      "loss": 0.6658,
      "step": 9880
    },
    {
      "epoch": 0.40950723196154004,
      "grad_norm": 0.4080725610256195,
      "learning_rate": 2.952671059720668e-06,
      "loss": 0.7075,
      "step": 9881
    },
    {
      "epoch": 0.4095486758672137,
      "grad_norm": 0.36060792207717896,
      "learning_rate": 2.9524638401923e-06,
      "loss": 0.6119,
      "step": 9882
    },
    {
      "epoch": 0.4095901197728874,
      "grad_norm": 0.40297508239746094,
      "learning_rate": 2.9522566206639314e-06,
      "loss": 0.6689,
      "step": 9883
    },
    {
      "epoch": 0.4096315636785611,
      "grad_norm": 0.4059356153011322,
      "learning_rate": 2.952049401135563e-06,
      "loss": 0.6775,
      "step": 9884
    },
    {
      "epoch": 0.40967300758423475,
      "grad_norm": 0.43801403045654297,
      "learning_rate": 2.951842181607195e-06,
      "loss": 0.6689,
      "step": 9885
    },
    {
      "epoch": 0.4097144514899084,
      "grad_norm": 0.40025022625923157,
      "learning_rate": 2.9516349620788264e-06,
      "loss": 0.6647,
      "step": 9886
    },
    {
      "epoch": 0.4097558953955821,
      "grad_norm": 0.40739428997039795,
      "learning_rate": 2.951427742550458e-06,
      "loss": 0.7041,
      "step": 9887
    },
    {
      "epoch": 0.4097973393012557,
      "grad_norm": 0.4218859076499939,
      "learning_rate": 2.9512205230220896e-06,
      "loss": 0.7006,
      "step": 9888
    },
    {
      "epoch": 0.4098387832069294,
      "grad_norm": 0.4065995514392853,
      "learning_rate": 2.951013303493722e-06,
      "loss": 0.6674,
      "step": 9889
    },
    {
      "epoch": 0.4098802271126031,
      "grad_norm": 0.4310034215450287,
      "learning_rate": 2.9508060839653528e-06,
      "loss": 0.6708,
      "step": 9890
    },
    {
      "epoch": 0.40992167101827676,
      "grad_norm": 0.4128533601760864,
      "learning_rate": 2.950598864436985e-06,
      "loss": 0.6901,
      "step": 9891
    },
    {
      "epoch": 0.40996311492395043,
      "grad_norm": 0.4101572036743164,
      "learning_rate": 2.950391644908616e-06,
      "loss": 0.6492,
      "step": 9892
    },
    {
      "epoch": 0.4100045588296241,
      "grad_norm": 0.3957703709602356,
      "learning_rate": 2.950184425380248e-06,
      "loss": 0.6899,
      "step": 9893
    },
    {
      "epoch": 0.4100460027352978,
      "grad_norm": 0.3723556697368622,
      "learning_rate": 2.94997720585188e-06,
      "loss": 0.6715,
      "step": 9894
    },
    {
      "epoch": 0.41008744664097146,
      "grad_norm": 0.3845115303993225,
      "learning_rate": 2.9497699863235114e-06,
      "loss": 0.6562,
      "step": 9895
    },
    {
      "epoch": 0.41012889054664514,
      "grad_norm": 0.4314626455307007,
      "learning_rate": 2.949562766795143e-06,
      "loss": 0.7449,
      "step": 9896
    },
    {
      "epoch": 0.41017033445231876,
      "grad_norm": 0.43843644857406616,
      "learning_rate": 2.9493555472667746e-06,
      "loss": 0.6987,
      "step": 9897
    },
    {
      "epoch": 0.41021177835799244,
      "grad_norm": 0.4094424247741699,
      "learning_rate": 2.9491483277384064e-06,
      "loss": 0.6987,
      "step": 9898
    },
    {
      "epoch": 0.4102532222636661,
      "grad_norm": 0.4344218373298645,
      "learning_rate": 2.948941108210038e-06,
      "loss": 0.6379,
      "step": 9899
    },
    {
      "epoch": 0.4102946661693398,
      "grad_norm": 0.41544845700263977,
      "learning_rate": 2.9487338886816696e-06,
      "loss": 0.6914,
      "step": 9900
    },
    {
      "epoch": 0.41033611007501347,
      "grad_norm": 0.3984636068344116,
      "learning_rate": 2.948526669153301e-06,
      "loss": 0.6799,
      "step": 9901
    },
    {
      "epoch": 0.41037755398068715,
      "grad_norm": 0.38979098200798035,
      "learning_rate": 2.948319449624933e-06,
      "loss": 0.7173,
      "step": 9902
    },
    {
      "epoch": 0.4104189978863608,
      "grad_norm": 0.41933101415634155,
      "learning_rate": 2.9481122300965646e-06,
      "loss": 0.6862,
      "step": 9903
    },
    {
      "epoch": 0.4104604417920345,
      "grad_norm": 0.41371187567710876,
      "learning_rate": 2.947905010568196e-06,
      "loss": 0.7002,
      "step": 9904
    },
    {
      "epoch": 0.4105018856977081,
      "grad_norm": 0.43397021293640137,
      "learning_rate": 2.947697791039828e-06,
      "loss": 0.637,
      "step": 9905
    },
    {
      "epoch": 0.4105433296033818,
      "grad_norm": 0.39864104986190796,
      "learning_rate": 2.947490571511459e-06,
      "loss": 0.6938,
      "step": 9906
    },
    {
      "epoch": 0.4105847735090555,
      "grad_norm": 0.40557411313056946,
      "learning_rate": 2.9472833519830914e-06,
      "loss": 0.6316,
      "step": 9907
    },
    {
      "epoch": 0.41062621741472916,
      "grad_norm": 0.4153580069541931,
      "learning_rate": 2.9470761324547224e-06,
      "loss": 0.7166,
      "step": 9908
    },
    {
      "epoch": 0.41066766132040283,
      "grad_norm": 0.40662312507629395,
      "learning_rate": 2.9468689129263546e-06,
      "loss": 0.6587,
      "step": 9909
    },
    {
      "epoch": 0.4107091052260765,
      "grad_norm": 0.38802796602249146,
      "learning_rate": 2.946661693397986e-06,
      "loss": 0.6758,
      "step": 9910
    },
    {
      "epoch": 0.4107505491317502,
      "grad_norm": 0.4412783980369568,
      "learning_rate": 2.946454473869618e-06,
      "loss": 0.7275,
      "step": 9911
    },
    {
      "epoch": 0.41079199303742386,
      "grad_norm": 0.371125191450119,
      "learning_rate": 2.9462472543412496e-06,
      "loss": 0.6433,
      "step": 9912
    },
    {
      "epoch": 0.41083343694309754,
      "grad_norm": 0.3777885437011719,
      "learning_rate": 2.946040034812881e-06,
      "loss": 0.645,
      "step": 9913
    },
    {
      "epoch": 0.41087488084877116,
      "grad_norm": 0.393840491771698,
      "learning_rate": 2.945832815284513e-06,
      "loss": 0.6832,
      "step": 9914
    },
    {
      "epoch": 0.41091632475444484,
      "grad_norm": 0.39863622188568115,
      "learning_rate": 2.945625595756144e-06,
      "loss": 0.667,
      "step": 9915
    },
    {
      "epoch": 0.4109577686601185,
      "grad_norm": 0.39298927783966064,
      "learning_rate": 2.945418376227776e-06,
      "loss": 0.6853,
      "step": 9916
    },
    {
      "epoch": 0.4109992125657922,
      "grad_norm": 0.4325486421585083,
      "learning_rate": 2.9452111566994074e-06,
      "loss": 0.6512,
      "step": 9917
    },
    {
      "epoch": 0.41104065647146587,
      "grad_norm": 0.4007813632488251,
      "learning_rate": 2.945003937171039e-06,
      "loss": 0.6919,
      "step": 9918
    },
    {
      "epoch": 0.41108210037713955,
      "grad_norm": 0.43355464935302734,
      "learning_rate": 2.9447967176426706e-06,
      "loss": 0.7268,
      "step": 9919
    },
    {
      "epoch": 0.4111235442828132,
      "grad_norm": 0.3767062723636627,
      "learning_rate": 2.9445894981143024e-06,
      "loss": 0.6606,
      "step": 9920
    },
    {
      "epoch": 0.4111649881884869,
      "grad_norm": 0.4500974714756012,
      "learning_rate": 2.944382278585934e-06,
      "loss": 0.6826,
      "step": 9921
    },
    {
      "epoch": 0.4112064320941606,
      "grad_norm": 0.4071853458881378,
      "learning_rate": 2.9441750590575656e-06,
      "loss": 0.6824,
      "step": 9922
    },
    {
      "epoch": 0.4112478759998342,
      "grad_norm": 0.4739421010017395,
      "learning_rate": 2.943967839529198e-06,
      "loss": 0.7356,
      "step": 9923
    },
    {
      "epoch": 0.4112893199055079,
      "grad_norm": 0.4158670902252197,
      "learning_rate": 2.9437606200008288e-06,
      "loss": 0.6599,
      "step": 9924
    },
    {
      "epoch": 0.41133076381118155,
      "grad_norm": 0.40996891260147095,
      "learning_rate": 2.943553400472461e-06,
      "loss": 0.7335,
      "step": 9925
    },
    {
      "epoch": 0.41137220771685523,
      "grad_norm": 0.4265899956226349,
      "learning_rate": 2.943346180944092e-06,
      "loss": 0.751,
      "step": 9926
    },
    {
      "epoch": 0.4114136516225289,
      "grad_norm": 0.3901168704032898,
      "learning_rate": 2.9431389614157242e-06,
      "loss": 0.6665,
      "step": 9927
    },
    {
      "epoch": 0.4114550955282026,
      "grad_norm": 0.4476945400238037,
      "learning_rate": 2.942931741887356e-06,
      "loss": 0.7175,
      "step": 9928
    },
    {
      "epoch": 0.41149653943387626,
      "grad_norm": 0.4273155927658081,
      "learning_rate": 2.9427245223589874e-06,
      "loss": 0.7363,
      "step": 9929
    },
    {
      "epoch": 0.41153798333954994,
      "grad_norm": 0.4015730619430542,
      "learning_rate": 2.9425173028306192e-06,
      "loss": 0.6642,
      "step": 9930
    },
    {
      "epoch": 0.4115794272452236,
      "grad_norm": 0.4184058606624603,
      "learning_rate": 2.9423100833022506e-06,
      "loss": 0.751,
      "step": 9931
    },
    {
      "epoch": 0.41162087115089724,
      "grad_norm": 0.4010770320892334,
      "learning_rate": 2.9421028637738824e-06,
      "loss": 0.7415,
      "step": 9932
    },
    {
      "epoch": 0.4116623150565709,
      "grad_norm": 0.4162043333053589,
      "learning_rate": 2.941895644245514e-06,
      "loss": 0.6912,
      "step": 9933
    },
    {
      "epoch": 0.4117037589622446,
      "grad_norm": 0.4069697856903076,
      "learning_rate": 2.9416884247171456e-06,
      "loss": 0.7273,
      "step": 9934
    },
    {
      "epoch": 0.41174520286791827,
      "grad_norm": 0.4305668771266937,
      "learning_rate": 2.941481205188777e-06,
      "loss": 0.6912,
      "step": 9935
    },
    {
      "epoch": 0.41178664677359195,
      "grad_norm": 0.4122347831726074,
      "learning_rate": 2.941273985660409e-06,
      "loss": 0.6635,
      "step": 9936
    },
    {
      "epoch": 0.4118280906792656,
      "grad_norm": 0.41147279739379883,
      "learning_rate": 2.9410667661320406e-06,
      "loss": 0.7229,
      "step": 9937
    },
    {
      "epoch": 0.4118695345849393,
      "grad_norm": 0.4109627902507782,
      "learning_rate": 2.940859546603672e-06,
      "loss": 0.7708,
      "step": 9938
    },
    {
      "epoch": 0.411910978490613,
      "grad_norm": 0.42420098185539246,
      "learning_rate": 2.940652327075304e-06,
      "loss": 0.7279,
      "step": 9939
    },
    {
      "epoch": 0.4119524223962866,
      "grad_norm": 0.4074649214744568,
      "learning_rate": 2.940445107546935e-06,
      "loss": 0.7277,
      "step": 9940
    },
    {
      "epoch": 0.4119938663019603,
      "grad_norm": 0.4052943289279938,
      "learning_rate": 2.9402378880185674e-06,
      "loss": 0.6741,
      "step": 9941
    },
    {
      "epoch": 0.41203531020763395,
      "grad_norm": 0.4130780100822449,
      "learning_rate": 2.9400306684901984e-06,
      "loss": 0.6853,
      "step": 9942
    },
    {
      "epoch": 0.41207675411330763,
      "grad_norm": 0.40979427099227905,
      "learning_rate": 2.9398234489618306e-06,
      "loss": 0.6733,
      "step": 9943
    },
    {
      "epoch": 0.4121181980189813,
      "grad_norm": 0.41698014736175537,
      "learning_rate": 2.939616229433462e-06,
      "loss": 0.698,
      "step": 9944
    },
    {
      "epoch": 0.412159641924655,
      "grad_norm": 0.3715636134147644,
      "learning_rate": 2.939409009905094e-06,
      "loss": 0.6418,
      "step": 9945
    },
    {
      "epoch": 0.41220108583032866,
      "grad_norm": 0.4079609811306,
      "learning_rate": 2.9392017903767256e-06,
      "loss": 0.6791,
      "step": 9946
    },
    {
      "epoch": 0.41224252973600234,
      "grad_norm": 0.40526172518730164,
      "learning_rate": 2.938994570848357e-06,
      "loss": 0.6763,
      "step": 9947
    },
    {
      "epoch": 0.412283973641676,
      "grad_norm": 0.4079827070236206,
      "learning_rate": 2.938787351319989e-06,
      "loss": 0.7605,
      "step": 9948
    },
    {
      "epoch": 0.41232541754734964,
      "grad_norm": 0.43996623158454895,
      "learning_rate": 2.93858013179162e-06,
      "loss": 0.744,
      "step": 9949
    },
    {
      "epoch": 0.4123668614530233,
      "grad_norm": 0.4004502594470978,
      "learning_rate": 2.938372912263252e-06,
      "loss": 0.728,
      "step": 9950
    },
    {
      "epoch": 0.412408305358697,
      "grad_norm": 0.4523896276950836,
      "learning_rate": 2.9381656927348834e-06,
      "loss": 0.7148,
      "step": 9951
    },
    {
      "epoch": 0.41244974926437067,
      "grad_norm": 0.44025474786758423,
      "learning_rate": 2.937958473206515e-06,
      "loss": 0.7358,
      "step": 9952
    },
    {
      "epoch": 0.41249119317004435,
      "grad_norm": 0.39966803789138794,
      "learning_rate": 2.9377512536781466e-06,
      "loss": 0.6677,
      "step": 9953
    },
    {
      "epoch": 0.412532637075718,
      "grad_norm": 0.39091306924819946,
      "learning_rate": 2.9375440341497784e-06,
      "loss": 0.6733,
      "step": 9954
    },
    {
      "epoch": 0.4125740809813917,
      "grad_norm": 0.41826990246772766,
      "learning_rate": 2.93733681462141e-06,
      "loss": 0.6892,
      "step": 9955
    },
    {
      "epoch": 0.4126155248870654,
      "grad_norm": 0.4154479205608368,
      "learning_rate": 2.9371295950930416e-06,
      "loss": 0.7212,
      "step": 9956
    },
    {
      "epoch": 0.41265696879273905,
      "grad_norm": 0.46661099791526794,
      "learning_rate": 2.936922375564674e-06,
      "loss": 0.793,
      "step": 9957
    },
    {
      "epoch": 0.4126984126984127,
      "grad_norm": 0.42165571451187134,
      "learning_rate": 2.936715156036305e-06,
      "loss": 0.703,
      "step": 9958
    },
    {
      "epoch": 0.41273985660408635,
      "grad_norm": 0.3981415331363678,
      "learning_rate": 2.936507936507937e-06,
      "loss": 0.6792,
      "step": 9959
    },
    {
      "epoch": 0.41278130050976003,
      "grad_norm": 0.43455350399017334,
      "learning_rate": 2.9363007169795684e-06,
      "loss": 0.733,
      "step": 9960
    },
    {
      "epoch": 0.4128227444154337,
      "grad_norm": 0.40184104442596436,
      "learning_rate": 2.9360934974512002e-06,
      "loss": 0.7087,
      "step": 9961
    },
    {
      "epoch": 0.4128641883211074,
      "grad_norm": 0.39984843134880066,
      "learning_rate": 2.9358862779228316e-06,
      "loss": 0.7166,
      "step": 9962
    },
    {
      "epoch": 0.41290563222678106,
      "grad_norm": 0.41432371735572815,
      "learning_rate": 2.9356790583944634e-06,
      "loss": 0.6831,
      "step": 9963
    },
    {
      "epoch": 0.41294707613245474,
      "grad_norm": 0.43040016293525696,
      "learning_rate": 2.9354718388660952e-06,
      "loss": 0.6819,
      "step": 9964
    },
    {
      "epoch": 0.4129885200381284,
      "grad_norm": 0.4298746585845947,
      "learning_rate": 2.9352646193377266e-06,
      "loss": 0.7622,
      "step": 9965
    },
    {
      "epoch": 0.41302996394380204,
      "grad_norm": 0.3972429633140564,
      "learning_rate": 2.9350573998093584e-06,
      "loss": 0.6382,
      "step": 9966
    },
    {
      "epoch": 0.4130714078494757,
      "grad_norm": 0.4548075497150421,
      "learning_rate": 2.93485018028099e-06,
      "loss": 0.6855,
      "step": 9967
    },
    {
      "epoch": 0.4131128517551494,
      "grad_norm": 0.3880721926689148,
      "learning_rate": 2.9346429607526216e-06,
      "loss": 0.6516,
      "step": 9968
    },
    {
      "epoch": 0.41315429566082307,
      "grad_norm": 0.41648736596107483,
      "learning_rate": 2.934435741224253e-06,
      "loss": 0.7219,
      "step": 9969
    },
    {
      "epoch": 0.41319573956649674,
      "grad_norm": 0.4189252257347107,
      "learning_rate": 2.934228521695885e-06,
      "loss": 0.6552,
      "step": 9970
    },
    {
      "epoch": 0.4132371834721704,
      "grad_norm": 0.3922632932662964,
      "learning_rate": 2.934021302167516e-06,
      "loss": 0.6396,
      "step": 9971
    },
    {
      "epoch": 0.4132786273778441,
      "grad_norm": 0.4077017903327942,
      "learning_rate": 2.933814082639148e-06,
      "loss": 0.6536,
      "step": 9972
    },
    {
      "epoch": 0.4133200712835178,
      "grad_norm": 0.413200318813324,
      "learning_rate": 2.9336068631107802e-06,
      "loss": 0.7556,
      "step": 9973
    },
    {
      "epoch": 0.41336151518919145,
      "grad_norm": 0.4114530682563782,
      "learning_rate": 2.933399643582411e-06,
      "loss": 0.7104,
      "step": 9974
    },
    {
      "epoch": 0.4134029590948651,
      "grad_norm": 0.4124280512332916,
      "learning_rate": 2.9331924240540434e-06,
      "loss": 0.6719,
      "step": 9975
    },
    {
      "epoch": 0.41344440300053875,
      "grad_norm": 0.3713845908641815,
      "learning_rate": 2.9329852045256744e-06,
      "loss": 0.7056,
      "step": 9976
    },
    {
      "epoch": 0.41348584690621243,
      "grad_norm": 0.39752280712127686,
      "learning_rate": 2.9327779849973066e-06,
      "loss": 0.7122,
      "step": 9977
    },
    {
      "epoch": 0.4135272908118861,
      "grad_norm": 0.39454489946365356,
      "learning_rate": 2.932570765468938e-06,
      "loss": 0.7393,
      "step": 9978
    },
    {
      "epoch": 0.4135687347175598,
      "grad_norm": 0.4232577383518219,
      "learning_rate": 2.93236354594057e-06,
      "loss": 0.6969,
      "step": 9979
    },
    {
      "epoch": 0.41361017862323346,
      "grad_norm": 0.41344892978668213,
      "learning_rate": 2.932156326412201e-06,
      "loss": 0.6617,
      "step": 9980
    },
    {
      "epoch": 0.41365162252890714,
      "grad_norm": 0.3872958719730377,
      "learning_rate": 2.931949106883833e-06,
      "loss": 0.6481,
      "step": 9981
    },
    {
      "epoch": 0.4136930664345808,
      "grad_norm": 0.44460973143577576,
      "learning_rate": 2.931741887355465e-06,
      "loss": 0.7537,
      "step": 9982
    },
    {
      "epoch": 0.4137345103402545,
      "grad_norm": 0.42787396907806396,
      "learning_rate": 2.931534667827096e-06,
      "loss": 0.6687,
      "step": 9983
    },
    {
      "epoch": 0.4137759542459281,
      "grad_norm": 0.42375168204307556,
      "learning_rate": 2.931327448298728e-06,
      "loss": 0.679,
      "step": 9984
    },
    {
      "epoch": 0.4138173981516018,
      "grad_norm": 0.4151184558868408,
      "learning_rate": 2.9311202287703594e-06,
      "loss": 0.6755,
      "step": 9985
    },
    {
      "epoch": 0.41385884205727547,
      "grad_norm": 0.44001683592796326,
      "learning_rate": 2.9309130092419912e-06,
      "loss": 0.6985,
      "step": 9986
    },
    {
      "epoch": 0.41390028596294914,
      "grad_norm": 0.42662665247917175,
      "learning_rate": 2.9307057897136226e-06,
      "loss": 0.6543,
      "step": 9987
    },
    {
      "epoch": 0.4139417298686228,
      "grad_norm": 0.42190679907798767,
      "learning_rate": 2.9304985701852544e-06,
      "loss": 0.7054,
      "step": 9988
    },
    {
      "epoch": 0.4139831737742965,
      "grad_norm": 0.3966737985610962,
      "learning_rate": 2.9302913506568862e-06,
      "loss": 0.7322,
      "step": 9989
    },
    {
      "epoch": 0.4140246176799702,
      "grad_norm": 0.4179009199142456,
      "learning_rate": 2.9300841311285176e-06,
      "loss": 0.6818,
      "step": 9990
    },
    {
      "epoch": 0.41406606158564385,
      "grad_norm": 0.43420839309692383,
      "learning_rate": 2.92987691160015e-06,
      "loss": 0.6742,
      "step": 9991
    },
    {
      "epoch": 0.41410750549131753,
      "grad_norm": 0.4090658724308014,
      "learning_rate": 2.929669692071781e-06,
      "loss": 0.7119,
      "step": 9992
    },
    {
      "epoch": 0.41414894939699115,
      "grad_norm": 0.4331103563308716,
      "learning_rate": 2.929462472543413e-06,
      "loss": 0.7126,
      "step": 9993
    },
    {
      "epoch": 0.4141903933026648,
      "grad_norm": 0.3838220536708832,
      "learning_rate": 2.9292552530150444e-06,
      "loss": 0.6747,
      "step": 9994
    },
    {
      "epoch": 0.4142318372083385,
      "grad_norm": 0.39047226309776306,
      "learning_rate": 2.9290480334866762e-06,
      "loss": 0.6993,
      "step": 9995
    },
    {
      "epoch": 0.4142732811140122,
      "grad_norm": 0.42338618636131287,
      "learning_rate": 2.9288408139583076e-06,
      "loss": 0.6611,
      "step": 9996
    },
    {
      "epoch": 0.41431472501968586,
      "grad_norm": 0.4173765778541565,
      "learning_rate": 2.9286335944299394e-06,
      "loss": 0.7207,
      "step": 9997
    },
    {
      "epoch": 0.41435616892535954,
      "grad_norm": 0.4694630801677704,
      "learning_rate": 2.9284263749015712e-06,
      "loss": 0.7632,
      "step": 9998
    },
    {
      "epoch": 0.4143976128310332,
      "grad_norm": 0.3794528841972351,
      "learning_rate": 2.9282191553732026e-06,
      "loss": 0.6506,
      "step": 9999
    },
    {
      "epoch": 0.4144390567367069,
      "grad_norm": 0.4125378727912903,
      "learning_rate": 2.9280119358448344e-06,
      "loss": 0.7036,
      "step": 10000
    },
    {
      "epoch": 0.4144805006423805,
      "grad_norm": 0.479382187128067,
      "learning_rate": 2.927804716316466e-06,
      "loss": 0.7251,
      "step": 10001
    },
    {
      "epoch": 0.4145219445480542,
      "grad_norm": 0.43535634875297546,
      "learning_rate": 2.9275974967880976e-06,
      "loss": 0.7097,
      "step": 10002
    },
    {
      "epoch": 0.41456338845372787,
      "grad_norm": 0.41060003638267517,
      "learning_rate": 2.927390277259729e-06,
      "loss": 0.6906,
      "step": 10003
    },
    {
      "epoch": 0.41460483235940154,
      "grad_norm": 0.4216634929180145,
      "learning_rate": 2.927183057731361e-06,
      "loss": 0.6709,
      "step": 10004
    },
    {
      "epoch": 0.4146462762650752,
      "grad_norm": 0.44468292593955994,
      "learning_rate": 2.926975838202992e-06,
      "loss": 0.6991,
      "step": 10005
    },
    {
      "epoch": 0.4146877201707489,
      "grad_norm": 0.4194738268852234,
      "learning_rate": 2.926768618674624e-06,
      "loss": 0.7322,
      "step": 10006
    },
    {
      "epoch": 0.4147291640764226,
      "grad_norm": 0.4366951882839203,
      "learning_rate": 2.9265613991462562e-06,
      "loss": 0.7334,
      "step": 10007
    },
    {
      "epoch": 0.41477060798209625,
      "grad_norm": 0.42715564370155334,
      "learning_rate": 2.926354179617887e-06,
      "loss": 0.6815,
      "step": 10008
    },
    {
      "epoch": 0.41481205188776993,
      "grad_norm": 0.3871564269065857,
      "learning_rate": 2.9261469600895194e-06,
      "loss": 0.6904,
      "step": 10009
    },
    {
      "epoch": 0.41485349579344355,
      "grad_norm": 0.4171812832355499,
      "learning_rate": 2.9259397405611504e-06,
      "loss": 0.7393,
      "step": 10010
    },
    {
      "epoch": 0.4148949396991172,
      "grad_norm": 0.4639853537082672,
      "learning_rate": 2.9257325210327826e-06,
      "loss": 0.752,
      "step": 10011
    },
    {
      "epoch": 0.4149363836047909,
      "grad_norm": 0.438038170337677,
      "learning_rate": 2.925525301504414e-06,
      "loss": 0.7021,
      "step": 10012
    },
    {
      "epoch": 0.4149778275104646,
      "grad_norm": 0.3876878321170807,
      "learning_rate": 2.925318081976046e-06,
      "loss": 0.673,
      "step": 10013
    },
    {
      "epoch": 0.41501927141613826,
      "grad_norm": 0.39537274837493896,
      "learning_rate": 2.925110862447677e-06,
      "loss": 0.6532,
      "step": 10014
    },
    {
      "epoch": 0.41506071532181193,
      "grad_norm": 0.3916131556034088,
      "learning_rate": 2.924903642919309e-06,
      "loss": 0.6873,
      "step": 10015
    },
    {
      "epoch": 0.4151021592274856,
      "grad_norm": 0.5355849862098694,
      "learning_rate": 2.924696423390941e-06,
      "loss": 0.679,
      "step": 10016
    },
    {
      "epoch": 0.4151436031331593,
      "grad_norm": 0.44493067264556885,
      "learning_rate": 2.9244892038625722e-06,
      "loss": 0.7637,
      "step": 10017
    },
    {
      "epoch": 0.41518504703883297,
      "grad_norm": 0.43440911173820496,
      "learning_rate": 2.924281984334204e-06,
      "loss": 0.7151,
      "step": 10018
    },
    {
      "epoch": 0.4152264909445066,
      "grad_norm": 0.39024344086647034,
      "learning_rate": 2.9240747648058354e-06,
      "loss": 0.6465,
      "step": 10019
    },
    {
      "epoch": 0.41526793485018026,
      "grad_norm": 0.37725114822387695,
      "learning_rate": 2.9238675452774672e-06,
      "loss": 0.7029,
      "step": 10020
    },
    {
      "epoch": 0.41530937875585394,
      "grad_norm": 0.3950027525424957,
      "learning_rate": 2.9236603257490986e-06,
      "loss": 0.7231,
      "step": 10021
    },
    {
      "epoch": 0.4153508226615276,
      "grad_norm": 0.41263076663017273,
      "learning_rate": 2.9234531062207304e-06,
      "loss": 0.7034,
      "step": 10022
    },
    {
      "epoch": 0.4153922665672013,
      "grad_norm": 0.4066719114780426,
      "learning_rate": 2.923245886692362e-06,
      "loss": 0.6309,
      "step": 10023
    },
    {
      "epoch": 0.415433710472875,
      "grad_norm": 0.4050705134868622,
      "learning_rate": 2.9230386671639936e-06,
      "loss": 0.6836,
      "step": 10024
    },
    {
      "epoch": 0.41547515437854865,
      "grad_norm": 0.38551566004753113,
      "learning_rate": 2.922831447635626e-06,
      "loss": 0.6692,
      "step": 10025
    },
    {
      "epoch": 0.4155165982842223,
      "grad_norm": 0.39364781975746155,
      "learning_rate": 2.922624228107257e-06,
      "loss": 0.6399,
      "step": 10026
    },
    {
      "epoch": 0.41555804218989595,
      "grad_norm": 0.39220213890075684,
      "learning_rate": 2.922417008578889e-06,
      "loss": 0.7007,
      "step": 10027
    },
    {
      "epoch": 0.4155994860955696,
      "grad_norm": 0.4132976233959198,
      "learning_rate": 2.9222097890505204e-06,
      "loss": 0.7498,
      "step": 10028
    },
    {
      "epoch": 0.4156409300012433,
      "grad_norm": 0.42848655581474304,
      "learning_rate": 2.9220025695221522e-06,
      "loss": 0.6899,
      "step": 10029
    },
    {
      "epoch": 0.415682373906917,
      "grad_norm": 0.4121173620223999,
      "learning_rate": 2.9217953499937836e-06,
      "loss": 0.6914,
      "step": 10030
    },
    {
      "epoch": 0.41572381781259066,
      "grad_norm": 0.38896945118904114,
      "learning_rate": 2.9215881304654154e-06,
      "loss": 0.7085,
      "step": 10031
    },
    {
      "epoch": 0.41576526171826433,
      "grad_norm": 0.4360326826572418,
      "learning_rate": 2.921380910937047e-06,
      "loss": 0.719,
      "step": 10032
    },
    {
      "epoch": 0.415806705623938,
      "grad_norm": 0.43854182958602905,
      "learning_rate": 2.9211736914086786e-06,
      "loss": 0.7231,
      "step": 10033
    },
    {
      "epoch": 0.4158481495296117,
      "grad_norm": 0.42000943422317505,
      "learning_rate": 2.9209664718803104e-06,
      "loss": 0.6848,
      "step": 10034
    },
    {
      "epoch": 0.41588959343528537,
      "grad_norm": 0.44673657417297363,
      "learning_rate": 2.920759252351942e-06,
      "loss": 0.7191,
      "step": 10035
    },
    {
      "epoch": 0.415931037340959,
      "grad_norm": 0.4248892664909363,
      "learning_rate": 2.9205520328235736e-06,
      "loss": 0.7278,
      "step": 10036
    },
    {
      "epoch": 0.41597248124663266,
      "grad_norm": 0.4134279191493988,
      "learning_rate": 2.920344813295205e-06,
      "loss": 0.6899,
      "step": 10037
    },
    {
      "epoch": 0.41601392515230634,
      "grad_norm": 0.4307844340801239,
      "learning_rate": 2.920137593766837e-06,
      "loss": 0.7169,
      "step": 10038
    },
    {
      "epoch": 0.41605536905798,
      "grad_norm": 0.4218250811100006,
      "learning_rate": 2.919930374238468e-06,
      "loss": 0.7063,
      "step": 10039
    },
    {
      "epoch": 0.4160968129636537,
      "grad_norm": 0.4126237630844116,
      "learning_rate": 2.9197231547101e-06,
      "loss": 0.6775,
      "step": 10040
    },
    {
      "epoch": 0.41613825686932737,
      "grad_norm": 0.4284915626049042,
      "learning_rate": 2.9195159351817322e-06,
      "loss": 0.6892,
      "step": 10041
    },
    {
      "epoch": 0.41617970077500105,
      "grad_norm": 0.4493291676044464,
      "learning_rate": 2.919308715653363e-06,
      "loss": 0.707,
      "step": 10042
    },
    {
      "epoch": 0.4162211446806747,
      "grad_norm": 0.38464421033859253,
      "learning_rate": 2.9191014961249954e-06,
      "loss": 0.645,
      "step": 10043
    },
    {
      "epoch": 0.4162625885863484,
      "grad_norm": 0.40138405561447144,
      "learning_rate": 2.9188942765966264e-06,
      "loss": 0.689,
      "step": 10044
    },
    {
      "epoch": 0.416304032492022,
      "grad_norm": 0.41934534907341003,
      "learning_rate": 2.9186870570682586e-06,
      "loss": 0.6815,
      "step": 10045
    },
    {
      "epoch": 0.4163454763976957,
      "grad_norm": 0.40376022458076477,
      "learning_rate": 2.91847983753989e-06,
      "loss": 0.7148,
      "step": 10046
    },
    {
      "epoch": 0.4163869203033694,
      "grad_norm": 0.4188496768474579,
      "learning_rate": 2.918272618011522e-06,
      "loss": 0.6438,
      "step": 10047
    },
    {
      "epoch": 0.41642836420904306,
      "grad_norm": 0.4080359637737274,
      "learning_rate": 2.9180653984831532e-06,
      "loss": 0.6993,
      "step": 10048
    },
    {
      "epoch": 0.41646980811471673,
      "grad_norm": 0.38120919466018677,
      "learning_rate": 2.917858178954785e-06,
      "loss": 0.6711,
      "step": 10049
    },
    {
      "epoch": 0.4165112520203904,
      "grad_norm": 0.41138529777526855,
      "learning_rate": 2.917650959426417e-06,
      "loss": 0.7163,
      "step": 10050
    },
    {
      "epoch": 0.4165526959260641,
      "grad_norm": 0.41981860995292664,
      "learning_rate": 2.9174437398980482e-06,
      "loss": 0.6647,
      "step": 10051
    },
    {
      "epoch": 0.41659413983173776,
      "grad_norm": 0.4471815228462219,
      "learning_rate": 2.91723652036968e-06,
      "loss": 0.6936,
      "step": 10052
    },
    {
      "epoch": 0.41663558373741144,
      "grad_norm": 0.3866643011569977,
      "learning_rate": 2.9170293008413114e-06,
      "loss": 0.6129,
      "step": 10053
    },
    {
      "epoch": 0.41667702764308506,
      "grad_norm": 0.3875521123409271,
      "learning_rate": 2.9168220813129432e-06,
      "loss": 0.6122,
      "step": 10054
    },
    {
      "epoch": 0.41671847154875874,
      "grad_norm": 0.41306057572364807,
      "learning_rate": 2.9166148617845746e-06,
      "loss": 0.6545,
      "step": 10055
    },
    {
      "epoch": 0.4167599154544324,
      "grad_norm": 0.41437965631484985,
      "learning_rate": 2.9164076422562064e-06,
      "loss": 0.6672,
      "step": 10056
    },
    {
      "epoch": 0.4168013593601061,
      "grad_norm": 0.395694762468338,
      "learning_rate": 2.916200422727838e-06,
      "loss": 0.6559,
      "step": 10057
    },
    {
      "epoch": 0.41684280326577977,
      "grad_norm": 0.4070577621459961,
      "learning_rate": 2.9159932031994696e-06,
      "loss": 0.6918,
      "step": 10058
    },
    {
      "epoch": 0.41688424717145345,
      "grad_norm": 0.43152493238449097,
      "learning_rate": 2.915785983671102e-06,
      "loss": 0.6803,
      "step": 10059
    },
    {
      "epoch": 0.4169256910771271,
      "grad_norm": 0.4536620080471039,
      "learning_rate": 2.915578764142733e-06,
      "loss": 0.7019,
      "step": 10060
    },
    {
      "epoch": 0.4169671349828008,
      "grad_norm": 0.4446908235549927,
      "learning_rate": 2.915371544614365e-06,
      "loss": 0.6952,
      "step": 10061
    },
    {
      "epoch": 0.4170085788884744,
      "grad_norm": 0.41655266284942627,
      "learning_rate": 2.9151643250859964e-06,
      "loss": 0.6783,
      "step": 10062
    },
    {
      "epoch": 0.4170500227941481,
      "grad_norm": 0.4246901869773865,
      "learning_rate": 2.9149571055576282e-06,
      "loss": 0.7532,
      "step": 10063
    },
    {
      "epoch": 0.4170914666998218,
      "grad_norm": 0.4338078796863556,
      "learning_rate": 2.9147498860292596e-06,
      "loss": 0.6992,
      "step": 10064
    },
    {
      "epoch": 0.41713291060549545,
      "grad_norm": 0.3923001289367676,
      "learning_rate": 2.9145426665008914e-06,
      "loss": 0.7367,
      "step": 10065
    },
    {
      "epoch": 0.41717435451116913,
      "grad_norm": 0.5137569308280945,
      "learning_rate": 2.914335446972523e-06,
      "loss": 0.7358,
      "step": 10066
    },
    {
      "epoch": 0.4172157984168428,
      "grad_norm": 0.3998924493789673,
      "learning_rate": 2.9141282274441546e-06,
      "loss": 0.6965,
      "step": 10067
    },
    {
      "epoch": 0.4172572423225165,
      "grad_norm": 0.40006181597709656,
      "learning_rate": 2.9139210079157864e-06,
      "loss": 0.7104,
      "step": 10068
    },
    {
      "epoch": 0.41729868622819016,
      "grad_norm": 0.4219914972782135,
      "learning_rate": 2.913713788387418e-06,
      "loss": 0.7056,
      "step": 10069
    },
    {
      "epoch": 0.41734013013386384,
      "grad_norm": 0.3953186571598053,
      "learning_rate": 2.9135065688590496e-06,
      "loss": 0.7297,
      "step": 10070
    },
    {
      "epoch": 0.41738157403953746,
      "grad_norm": 0.39155176281929016,
      "learning_rate": 2.913299349330681e-06,
      "loss": 0.6799,
      "step": 10071
    },
    {
      "epoch": 0.41742301794521114,
      "grad_norm": 0.3963681161403656,
      "learning_rate": 2.913092129802313e-06,
      "loss": 0.6285,
      "step": 10072
    },
    {
      "epoch": 0.4174644618508848,
      "grad_norm": 0.4096900522708893,
      "learning_rate": 2.912884910273944e-06,
      "loss": 0.6799,
      "step": 10073
    },
    {
      "epoch": 0.4175059057565585,
      "grad_norm": 0.40416139364242554,
      "learning_rate": 2.912677690745576e-06,
      "loss": 0.696,
      "step": 10074
    },
    {
      "epoch": 0.41754734966223217,
      "grad_norm": 0.444085955619812,
      "learning_rate": 2.9124704712172074e-06,
      "loss": 0.7034,
      "step": 10075
    },
    {
      "epoch": 0.41758879356790585,
      "grad_norm": 0.4189859628677368,
      "learning_rate": 2.9122632516888392e-06,
      "loss": 0.6899,
      "step": 10076
    },
    {
      "epoch": 0.4176302374735795,
      "grad_norm": 0.40397050976753235,
      "learning_rate": 2.9120560321604714e-06,
      "loss": 0.7283,
      "step": 10077
    },
    {
      "epoch": 0.4176716813792532,
      "grad_norm": 0.4646121859550476,
      "learning_rate": 2.9118488126321024e-06,
      "loss": 0.7408,
      "step": 10078
    },
    {
      "epoch": 0.4177131252849269,
      "grad_norm": 0.3946421146392822,
      "learning_rate": 2.9116415931037346e-06,
      "loss": 0.717,
      "step": 10079
    },
    {
      "epoch": 0.4177545691906005,
      "grad_norm": 0.42847490310668945,
      "learning_rate": 2.911434373575366e-06,
      "loss": 0.668,
      "step": 10080
    },
    {
      "epoch": 0.4177960130962742,
      "grad_norm": 0.4313769042491913,
      "learning_rate": 2.911227154046998e-06,
      "loss": 0.7117,
      "step": 10081
    },
    {
      "epoch": 0.41783745700194785,
      "grad_norm": 0.4347526431083679,
      "learning_rate": 2.9110199345186292e-06,
      "loss": 0.7015,
      "step": 10082
    },
    {
      "epoch": 0.41787890090762153,
      "grad_norm": 0.43561851978302,
      "learning_rate": 2.910812714990261e-06,
      "loss": 0.6584,
      "step": 10083
    },
    {
      "epoch": 0.4179203448132952,
      "grad_norm": 0.40543240308761597,
      "learning_rate": 2.9106054954618924e-06,
      "loss": 0.6255,
      "step": 10084
    },
    {
      "epoch": 0.4179617887189689,
      "grad_norm": 0.4019544720649719,
      "learning_rate": 2.9103982759335242e-06,
      "loss": 0.6289,
      "step": 10085
    },
    {
      "epoch": 0.41800323262464256,
      "grad_norm": 0.4045496881008148,
      "learning_rate": 2.910191056405156e-06,
      "loss": 0.6978,
      "step": 10086
    },
    {
      "epoch": 0.41804467653031624,
      "grad_norm": 0.4120318293571472,
      "learning_rate": 2.9099838368767874e-06,
      "loss": 0.6687,
      "step": 10087
    },
    {
      "epoch": 0.41808612043598986,
      "grad_norm": 0.4046080410480499,
      "learning_rate": 2.9097766173484192e-06,
      "loss": 0.6638,
      "step": 10088
    },
    {
      "epoch": 0.41812756434166354,
      "grad_norm": 0.4070160388946533,
      "learning_rate": 2.9095693978200506e-06,
      "loss": 0.6783,
      "step": 10089
    },
    {
      "epoch": 0.4181690082473372,
      "grad_norm": 0.4111829698085785,
      "learning_rate": 2.9093621782916824e-06,
      "loss": 0.7019,
      "step": 10090
    },
    {
      "epoch": 0.4182104521530109,
      "grad_norm": 0.4016530215740204,
      "learning_rate": 2.909154958763314e-06,
      "loss": 0.6498,
      "step": 10091
    },
    {
      "epoch": 0.41825189605868457,
      "grad_norm": 0.3862515091896057,
      "learning_rate": 2.9089477392349456e-06,
      "loss": 0.6609,
      "step": 10092
    },
    {
      "epoch": 0.41829333996435825,
      "grad_norm": 0.4537737965583801,
      "learning_rate": 2.908740519706577e-06,
      "loss": 0.7473,
      "step": 10093
    },
    {
      "epoch": 0.4183347838700319,
      "grad_norm": 0.4077194631099701,
      "learning_rate": 2.908533300178209e-06,
      "loss": 0.7039,
      "step": 10094
    },
    {
      "epoch": 0.4183762277757056,
      "grad_norm": 0.42547231912612915,
      "learning_rate": 2.908326080649841e-06,
      "loss": 0.7688,
      "step": 10095
    },
    {
      "epoch": 0.4184176716813793,
      "grad_norm": 0.38500475883483887,
      "learning_rate": 2.9081188611214724e-06,
      "loss": 0.64,
      "step": 10096
    },
    {
      "epoch": 0.4184591155870529,
      "grad_norm": 0.4095021188259125,
      "learning_rate": 2.9079116415931042e-06,
      "loss": 0.6846,
      "step": 10097
    },
    {
      "epoch": 0.4185005594927266,
      "grad_norm": 0.4133777618408203,
      "learning_rate": 2.9077044220647356e-06,
      "loss": 0.6697,
      "step": 10098
    },
    {
      "epoch": 0.41854200339840025,
      "grad_norm": 0.4071449935436249,
      "learning_rate": 2.9074972025363674e-06,
      "loss": 0.6575,
      "step": 10099
    },
    {
      "epoch": 0.41858344730407393,
      "grad_norm": 0.40374892950057983,
      "learning_rate": 2.907289983007999e-06,
      "loss": 0.678,
      "step": 10100
    },
    {
      "epoch": 0.4186248912097476,
      "grad_norm": 0.4224300980567932,
      "learning_rate": 2.9070827634796306e-06,
      "loss": 0.7646,
      "step": 10101
    },
    {
      "epoch": 0.4186663351154213,
      "grad_norm": 0.424853652715683,
      "learning_rate": 2.9068755439512624e-06,
      "loss": 0.724,
      "step": 10102
    },
    {
      "epoch": 0.41870777902109496,
      "grad_norm": 0.3898879587650299,
      "learning_rate": 2.906668324422894e-06,
      "loss": 0.687,
      "step": 10103
    },
    {
      "epoch": 0.41874922292676864,
      "grad_norm": 0.42142367362976074,
      "learning_rate": 2.9064611048945256e-06,
      "loss": 0.7504,
      "step": 10104
    },
    {
      "epoch": 0.4187906668324423,
      "grad_norm": 0.39617568254470825,
      "learning_rate": 2.906253885366157e-06,
      "loss": 0.6882,
      "step": 10105
    },
    {
      "epoch": 0.41883211073811594,
      "grad_norm": 0.4162764549255371,
      "learning_rate": 2.906046665837789e-06,
      "loss": 0.6709,
      "step": 10106
    },
    {
      "epoch": 0.4188735546437896,
      "grad_norm": 0.45892590284347534,
      "learning_rate": 2.9058394463094202e-06,
      "loss": 0.6831,
      "step": 10107
    },
    {
      "epoch": 0.4189149985494633,
      "grad_norm": 0.4267515242099762,
      "learning_rate": 2.905632226781052e-06,
      "loss": 0.6858,
      "step": 10108
    },
    {
      "epoch": 0.41895644245513697,
      "grad_norm": 0.4441525340080261,
      "learning_rate": 2.9054250072526834e-06,
      "loss": 0.7308,
      "step": 10109
    },
    {
      "epoch": 0.41899788636081065,
      "grad_norm": 0.4077686071395874,
      "learning_rate": 2.9052177877243152e-06,
      "loss": 0.7095,
      "step": 10110
    },
    {
      "epoch": 0.4190393302664843,
      "grad_norm": 0.4197358191013336,
      "learning_rate": 2.9050105681959475e-06,
      "loss": 0.6647,
      "step": 10111
    },
    {
      "epoch": 0.419080774172158,
      "grad_norm": 0.4275217652320862,
      "learning_rate": 2.9048033486675784e-06,
      "loss": 0.72,
      "step": 10112
    },
    {
      "epoch": 0.4191222180778317,
      "grad_norm": 0.4126758873462677,
      "learning_rate": 2.9045961291392106e-06,
      "loss": 0.7151,
      "step": 10113
    },
    {
      "epoch": 0.4191636619835053,
      "grad_norm": 0.4319170117378235,
      "learning_rate": 2.904388909610842e-06,
      "loss": 0.7312,
      "step": 10114
    },
    {
      "epoch": 0.419205105889179,
      "grad_norm": 0.39468684792518616,
      "learning_rate": 2.904181690082474e-06,
      "loss": 0.6582,
      "step": 10115
    },
    {
      "epoch": 0.41924654979485265,
      "grad_norm": 0.4175708591938019,
      "learning_rate": 2.9039744705541052e-06,
      "loss": 0.6917,
      "step": 10116
    },
    {
      "epoch": 0.41928799370052633,
      "grad_norm": 0.4118190407752991,
      "learning_rate": 2.903767251025737e-06,
      "loss": 0.6755,
      "step": 10117
    },
    {
      "epoch": 0.4193294376062,
      "grad_norm": 0.4232187569141388,
      "learning_rate": 2.9035600314973684e-06,
      "loss": 0.6902,
      "step": 10118
    },
    {
      "epoch": 0.4193708815118737,
      "grad_norm": 0.41559267044067383,
      "learning_rate": 2.9033528119690002e-06,
      "loss": 0.7065,
      "step": 10119
    },
    {
      "epoch": 0.41941232541754736,
      "grad_norm": 0.40015387535095215,
      "learning_rate": 2.903145592440632e-06,
      "loss": 0.7622,
      "step": 10120
    },
    {
      "epoch": 0.41945376932322104,
      "grad_norm": 0.43168577551841736,
      "learning_rate": 2.9029383729122634e-06,
      "loss": 0.7197,
      "step": 10121
    },
    {
      "epoch": 0.4194952132288947,
      "grad_norm": 0.4424954950809479,
      "learning_rate": 2.9027311533838952e-06,
      "loss": 0.7102,
      "step": 10122
    },
    {
      "epoch": 0.41953665713456834,
      "grad_norm": 0.39829349517822266,
      "learning_rate": 2.9025239338555266e-06,
      "loss": 0.6477,
      "step": 10123
    },
    {
      "epoch": 0.419578101040242,
      "grad_norm": 0.4464055895805359,
      "learning_rate": 2.9023167143271584e-06,
      "loss": 0.7225,
      "step": 10124
    },
    {
      "epoch": 0.4196195449459157,
      "grad_norm": 0.44539156556129456,
      "learning_rate": 2.90210949479879e-06,
      "loss": 0.767,
      "step": 10125
    },
    {
      "epoch": 0.41966098885158937,
      "grad_norm": 0.41021695733070374,
      "learning_rate": 2.9019022752704216e-06,
      "loss": 0.6956,
      "step": 10126
    },
    {
      "epoch": 0.41970243275726304,
      "grad_norm": 0.42815688252449036,
      "learning_rate": 2.901695055742053e-06,
      "loss": 0.6829,
      "step": 10127
    },
    {
      "epoch": 0.4197438766629367,
      "grad_norm": 0.4615345895290375,
      "learning_rate": 2.901487836213685e-06,
      "loss": 0.7056,
      "step": 10128
    },
    {
      "epoch": 0.4197853205686104,
      "grad_norm": 0.4075317084789276,
      "learning_rate": 2.901280616685317e-06,
      "loss": 0.6912,
      "step": 10129
    },
    {
      "epoch": 0.4198267644742841,
      "grad_norm": 0.4276561141014099,
      "learning_rate": 2.9010733971569484e-06,
      "loss": 0.674,
      "step": 10130
    },
    {
      "epoch": 0.41986820837995775,
      "grad_norm": 0.41672348976135254,
      "learning_rate": 2.9008661776285802e-06,
      "loss": 0.6958,
      "step": 10131
    },
    {
      "epoch": 0.4199096522856314,
      "grad_norm": 0.3936053216457367,
      "learning_rate": 2.9006589581002116e-06,
      "loss": 0.688,
      "step": 10132
    },
    {
      "epoch": 0.41995109619130505,
      "grad_norm": 0.4158685505390167,
      "learning_rate": 2.9004517385718434e-06,
      "loss": 0.6726,
      "step": 10133
    },
    {
      "epoch": 0.41999254009697873,
      "grad_norm": 0.41040393710136414,
      "learning_rate": 2.900244519043475e-06,
      "loss": 0.6804,
      "step": 10134
    },
    {
      "epoch": 0.4200339840026524,
      "grad_norm": 0.4078502058982849,
      "learning_rate": 2.9000372995151066e-06,
      "loss": 0.7456,
      "step": 10135
    },
    {
      "epoch": 0.4200754279083261,
      "grad_norm": 0.3787294030189514,
      "learning_rate": 2.899830079986738e-06,
      "loss": 0.6664,
      "step": 10136
    },
    {
      "epoch": 0.42011687181399976,
      "grad_norm": 0.4465276300907135,
      "learning_rate": 2.89962286045837e-06,
      "loss": 0.7246,
      "step": 10137
    },
    {
      "epoch": 0.42015831571967344,
      "grad_norm": 0.41634276509284973,
      "learning_rate": 2.8994156409300016e-06,
      "loss": 0.691,
      "step": 10138
    },
    {
      "epoch": 0.4201997596253471,
      "grad_norm": 0.41570356488227844,
      "learning_rate": 2.899208421401633e-06,
      "loss": 0.6851,
      "step": 10139
    },
    {
      "epoch": 0.4202412035310208,
      "grad_norm": 0.41125667095184326,
      "learning_rate": 2.899001201873265e-06,
      "loss": 0.6753,
      "step": 10140
    },
    {
      "epoch": 0.4202826474366944,
      "grad_norm": 0.43010514974594116,
      "learning_rate": 2.8987939823448962e-06,
      "loss": 0.7109,
      "step": 10141
    },
    {
      "epoch": 0.4203240913423681,
      "grad_norm": 0.4447622299194336,
      "learning_rate": 2.898586762816528e-06,
      "loss": 0.6852,
      "step": 10142
    },
    {
      "epoch": 0.42036553524804177,
      "grad_norm": 0.37587252259254456,
      "learning_rate": 2.8983795432881594e-06,
      "loss": 0.6855,
      "step": 10143
    },
    {
      "epoch": 0.42040697915371544,
      "grad_norm": 0.47261160612106323,
      "learning_rate": 2.8981723237597912e-06,
      "loss": 0.7773,
      "step": 10144
    },
    {
      "epoch": 0.4204484230593891,
      "grad_norm": 0.39476269483566284,
      "learning_rate": 2.8979651042314226e-06,
      "loss": 0.6777,
      "step": 10145
    },
    {
      "epoch": 0.4204898669650628,
      "grad_norm": 0.42408743500709534,
      "learning_rate": 2.897757884703055e-06,
      "loss": 0.6736,
      "step": 10146
    },
    {
      "epoch": 0.4205313108707365,
      "grad_norm": 0.41875147819519043,
      "learning_rate": 2.8975506651746867e-06,
      "loss": 0.7395,
      "step": 10147
    },
    {
      "epoch": 0.42057275477641015,
      "grad_norm": 0.3818807303905487,
      "learning_rate": 2.897343445646318e-06,
      "loss": 0.6786,
      "step": 10148
    },
    {
      "epoch": 0.4206141986820838,
      "grad_norm": 0.39929676055908203,
      "learning_rate": 2.89713622611795e-06,
      "loss": 0.7102,
      "step": 10149
    },
    {
      "epoch": 0.42065564258775745,
      "grad_norm": 0.39484116435050964,
      "learning_rate": 2.8969290065895812e-06,
      "loss": 0.6819,
      "step": 10150
    },
    {
      "epoch": 0.4206970864934311,
      "grad_norm": 0.44920578598976135,
      "learning_rate": 2.896721787061213e-06,
      "loss": 0.7023,
      "step": 10151
    },
    {
      "epoch": 0.4207385303991048,
      "grad_norm": 0.48473748564720154,
      "learning_rate": 2.8965145675328444e-06,
      "loss": 0.8225,
      "step": 10152
    },
    {
      "epoch": 0.4207799743047785,
      "grad_norm": 0.3916252553462982,
      "learning_rate": 2.8963073480044762e-06,
      "loss": 0.6824,
      "step": 10153
    },
    {
      "epoch": 0.42082141821045216,
      "grad_norm": 0.4219130873680115,
      "learning_rate": 2.8961001284761076e-06,
      "loss": 0.7069,
      "step": 10154
    },
    {
      "epoch": 0.42086286211612584,
      "grad_norm": 0.37896451354026794,
      "learning_rate": 2.8958929089477394e-06,
      "loss": 0.6726,
      "step": 10155
    },
    {
      "epoch": 0.4209043060217995,
      "grad_norm": 0.432675302028656,
      "learning_rate": 2.8956856894193712e-06,
      "loss": 0.7346,
      "step": 10156
    },
    {
      "epoch": 0.4209457499274732,
      "grad_norm": 0.42689788341522217,
      "learning_rate": 2.8954784698910026e-06,
      "loss": 0.6852,
      "step": 10157
    },
    {
      "epoch": 0.4209871938331468,
      "grad_norm": 0.40822523832321167,
      "learning_rate": 2.8952712503626344e-06,
      "loss": 0.7198,
      "step": 10158
    },
    {
      "epoch": 0.4210286377388205,
      "grad_norm": 0.4555254578590393,
      "learning_rate": 2.895064030834266e-06,
      "loss": 0.6924,
      "step": 10159
    },
    {
      "epoch": 0.42107008164449417,
      "grad_norm": 0.43470826745033264,
      "learning_rate": 2.8948568113058976e-06,
      "loss": 0.7238,
      "step": 10160
    },
    {
      "epoch": 0.42111152555016784,
      "grad_norm": 0.37628692388534546,
      "learning_rate": 2.894649591777529e-06,
      "loss": 0.6229,
      "step": 10161
    },
    {
      "epoch": 0.4211529694558415,
      "grad_norm": 0.4038245677947998,
      "learning_rate": 2.894442372249161e-06,
      "loss": 0.7654,
      "step": 10162
    },
    {
      "epoch": 0.4211944133615152,
      "grad_norm": 0.4021161198616028,
      "learning_rate": 2.894235152720793e-06,
      "loss": 0.6816,
      "step": 10163
    },
    {
      "epoch": 0.4212358572671889,
      "grad_norm": 0.47103646397590637,
      "learning_rate": 2.8940279331924244e-06,
      "loss": 0.6826,
      "step": 10164
    },
    {
      "epoch": 0.42127730117286255,
      "grad_norm": 0.43137219548225403,
      "learning_rate": 2.8938207136640563e-06,
      "loss": 0.6509,
      "step": 10165
    },
    {
      "epoch": 0.4213187450785362,
      "grad_norm": 0.4132159352302551,
      "learning_rate": 2.8936134941356876e-06,
      "loss": 0.6533,
      "step": 10166
    },
    {
      "epoch": 0.42136018898420985,
      "grad_norm": 0.4073375463485718,
      "learning_rate": 2.8934062746073194e-06,
      "loss": 0.6733,
      "step": 10167
    },
    {
      "epoch": 0.4214016328898835,
      "grad_norm": 0.39615103602409363,
      "learning_rate": 2.893199055078951e-06,
      "loss": 0.6301,
      "step": 10168
    },
    {
      "epoch": 0.4214430767955572,
      "grad_norm": 0.4119548499584198,
      "learning_rate": 2.8929918355505826e-06,
      "loss": 0.7258,
      "step": 10169
    },
    {
      "epoch": 0.4214845207012309,
      "grad_norm": 0.3968375325202942,
      "learning_rate": 2.892784616022214e-06,
      "loss": 0.6851,
      "step": 10170
    },
    {
      "epoch": 0.42152596460690456,
      "grad_norm": 0.42875370383262634,
      "learning_rate": 2.892577396493846e-06,
      "loss": 0.7073,
      "step": 10171
    },
    {
      "epoch": 0.42156740851257823,
      "grad_norm": 0.40341463685035706,
      "learning_rate": 2.8923701769654776e-06,
      "loss": 0.7263,
      "step": 10172
    },
    {
      "epoch": 0.4216088524182519,
      "grad_norm": 0.40994176268577576,
      "learning_rate": 2.892162957437109e-06,
      "loss": 0.6838,
      "step": 10173
    },
    {
      "epoch": 0.4216502963239256,
      "grad_norm": 0.3969689905643463,
      "learning_rate": 2.891955737908741e-06,
      "loss": 0.6956,
      "step": 10174
    },
    {
      "epoch": 0.4216917402295992,
      "grad_norm": 0.4012397229671478,
      "learning_rate": 2.8917485183803722e-06,
      "loss": 0.7178,
      "step": 10175
    },
    {
      "epoch": 0.4217331841352729,
      "grad_norm": 0.41515862941741943,
      "learning_rate": 2.891541298852004e-06,
      "loss": 0.6968,
      "step": 10176
    },
    {
      "epoch": 0.42177462804094656,
      "grad_norm": 0.4045315384864807,
      "learning_rate": 2.8913340793236354e-06,
      "loss": 0.637,
      "step": 10177
    },
    {
      "epoch": 0.42181607194662024,
      "grad_norm": 0.4065553843975067,
      "learning_rate": 2.8911268597952672e-06,
      "loss": 0.731,
      "step": 10178
    },
    {
      "epoch": 0.4218575158522939,
      "grad_norm": 0.38114413619041443,
      "learning_rate": 2.8909196402668986e-06,
      "loss": 0.6787,
      "step": 10179
    },
    {
      "epoch": 0.4218989597579676,
      "grad_norm": 0.4144408106803894,
      "learning_rate": 2.890712420738531e-06,
      "loss": 0.7239,
      "step": 10180
    },
    {
      "epoch": 0.4219404036636413,
      "grad_norm": 0.3942946791648865,
      "learning_rate": 2.8905052012101627e-06,
      "loss": 0.6406,
      "step": 10181
    },
    {
      "epoch": 0.42198184756931495,
      "grad_norm": 0.41020676493644714,
      "learning_rate": 2.890297981681794e-06,
      "loss": 0.6953,
      "step": 10182
    },
    {
      "epoch": 0.4220232914749886,
      "grad_norm": 0.4050442576408386,
      "learning_rate": 2.890090762153426e-06,
      "loss": 0.6769,
      "step": 10183
    },
    {
      "epoch": 0.42206473538066225,
      "grad_norm": 0.42163223028182983,
      "learning_rate": 2.8898835426250572e-06,
      "loss": 0.6957,
      "step": 10184
    },
    {
      "epoch": 0.4221061792863359,
      "grad_norm": 0.39362090826034546,
      "learning_rate": 2.889676323096689e-06,
      "loss": 0.6694,
      "step": 10185
    },
    {
      "epoch": 0.4221476231920096,
      "grad_norm": 0.38748443126678467,
      "learning_rate": 2.8894691035683204e-06,
      "loss": 0.6913,
      "step": 10186
    },
    {
      "epoch": 0.4221890670976833,
      "grad_norm": 0.46111661195755005,
      "learning_rate": 2.8892618840399522e-06,
      "loss": 0.7261,
      "step": 10187
    },
    {
      "epoch": 0.42223051100335696,
      "grad_norm": 0.4202723205089569,
      "learning_rate": 2.8890546645115836e-06,
      "loss": 0.7014,
      "step": 10188
    },
    {
      "epoch": 0.42227195490903063,
      "grad_norm": 0.40815672278404236,
      "learning_rate": 2.8888474449832154e-06,
      "loss": 0.7235,
      "step": 10189
    },
    {
      "epoch": 0.4223133988147043,
      "grad_norm": 0.44557347893714905,
      "learning_rate": 2.8886402254548472e-06,
      "loss": 0.7305,
      "step": 10190
    },
    {
      "epoch": 0.422354842720378,
      "grad_norm": 0.41423219442367554,
      "learning_rate": 2.8884330059264786e-06,
      "loss": 0.6337,
      "step": 10191
    },
    {
      "epoch": 0.42239628662605166,
      "grad_norm": 0.399832546710968,
      "learning_rate": 2.8882257863981104e-06,
      "loss": 0.6868,
      "step": 10192
    },
    {
      "epoch": 0.4224377305317253,
      "grad_norm": 0.3639524579048157,
      "learning_rate": 2.888018566869742e-06,
      "loss": 0.6476,
      "step": 10193
    },
    {
      "epoch": 0.42247917443739896,
      "grad_norm": 0.40864095091819763,
      "learning_rate": 2.8878113473413736e-06,
      "loss": 0.7261,
      "step": 10194
    },
    {
      "epoch": 0.42252061834307264,
      "grad_norm": 0.4145399034023285,
      "learning_rate": 2.887604127813005e-06,
      "loss": 0.6655,
      "step": 10195
    },
    {
      "epoch": 0.4225620622487463,
      "grad_norm": 0.4135926365852356,
      "learning_rate": 2.887396908284637e-06,
      "loss": 0.6855,
      "step": 10196
    },
    {
      "epoch": 0.42260350615442,
      "grad_norm": 0.42498481273651123,
      "learning_rate": 2.8871896887562682e-06,
      "loss": 0.7341,
      "step": 10197
    },
    {
      "epoch": 0.42264495006009367,
      "grad_norm": 0.4186903238296509,
      "learning_rate": 2.8869824692279005e-06,
      "loss": 0.7028,
      "step": 10198
    },
    {
      "epoch": 0.42268639396576735,
      "grad_norm": 0.4143699109554291,
      "learning_rate": 2.8867752496995323e-06,
      "loss": 0.78,
      "step": 10199
    },
    {
      "epoch": 0.422727837871441,
      "grad_norm": 0.41902509331703186,
      "learning_rate": 2.8865680301711636e-06,
      "loss": 0.6488,
      "step": 10200
    },
    {
      "epoch": 0.4227692817771147,
      "grad_norm": 0.396794855594635,
      "learning_rate": 2.8863608106427955e-06,
      "loss": 0.6522,
      "step": 10201
    },
    {
      "epoch": 0.4228107256827883,
      "grad_norm": 0.41808241605758667,
      "learning_rate": 2.886153591114427e-06,
      "loss": 0.6593,
      "step": 10202
    },
    {
      "epoch": 0.422852169588462,
      "grad_norm": 0.3930683434009552,
      "learning_rate": 2.8859463715860587e-06,
      "loss": 0.7261,
      "step": 10203
    },
    {
      "epoch": 0.4228936134941357,
      "grad_norm": 0.41105759143829346,
      "learning_rate": 2.88573915205769e-06,
      "loss": 0.6797,
      "step": 10204
    },
    {
      "epoch": 0.42293505739980936,
      "grad_norm": 0.3948060870170593,
      "learning_rate": 2.885531932529322e-06,
      "loss": 0.6885,
      "step": 10205
    },
    {
      "epoch": 0.42297650130548303,
      "grad_norm": 0.4425502419471741,
      "learning_rate": 2.8853247130009532e-06,
      "loss": 0.6951,
      "step": 10206
    },
    {
      "epoch": 0.4230179452111567,
      "grad_norm": 0.441812127828598,
      "learning_rate": 2.885117493472585e-06,
      "loss": 0.6698,
      "step": 10207
    },
    {
      "epoch": 0.4230593891168304,
      "grad_norm": 0.41703009605407715,
      "learning_rate": 2.884910273944217e-06,
      "loss": 0.6709,
      "step": 10208
    },
    {
      "epoch": 0.42310083302250406,
      "grad_norm": 0.37838214635849,
      "learning_rate": 2.8847030544158482e-06,
      "loss": 0.6591,
      "step": 10209
    },
    {
      "epoch": 0.4231422769281777,
      "grad_norm": 0.4456915259361267,
      "learning_rate": 2.88449583488748e-06,
      "loss": 0.7206,
      "step": 10210
    },
    {
      "epoch": 0.42318372083385136,
      "grad_norm": 0.39847251772880554,
      "learning_rate": 2.8842886153591114e-06,
      "loss": 0.7074,
      "step": 10211
    },
    {
      "epoch": 0.42322516473952504,
      "grad_norm": 0.3975691497325897,
      "learning_rate": 2.8840813958307432e-06,
      "loss": 0.7183,
      "step": 10212
    },
    {
      "epoch": 0.4232666086451987,
      "grad_norm": 0.4412553906440735,
      "learning_rate": 2.8838741763023746e-06,
      "loss": 0.6831,
      "step": 10213
    },
    {
      "epoch": 0.4233080525508724,
      "grad_norm": 0.3839682638645172,
      "learning_rate": 2.883666956774007e-06,
      "loss": 0.6415,
      "step": 10214
    },
    {
      "epoch": 0.42334949645654607,
      "grad_norm": 0.4261326193809509,
      "learning_rate": 2.883459737245638e-06,
      "loss": 0.6592,
      "step": 10215
    },
    {
      "epoch": 0.42339094036221975,
      "grad_norm": 0.4193367063999176,
      "learning_rate": 2.88325251771727e-06,
      "loss": 0.7007,
      "step": 10216
    },
    {
      "epoch": 0.4234323842678934,
      "grad_norm": 0.4272805154323578,
      "learning_rate": 2.883045298188902e-06,
      "loss": 0.72,
      "step": 10217
    },
    {
      "epoch": 0.4234738281735671,
      "grad_norm": 0.4257102608680725,
      "learning_rate": 2.8828380786605332e-06,
      "loss": 0.7686,
      "step": 10218
    },
    {
      "epoch": 0.4235152720792407,
      "grad_norm": 0.40702876448631287,
      "learning_rate": 2.882630859132165e-06,
      "loss": 0.7122,
      "step": 10219
    },
    {
      "epoch": 0.4235567159849144,
      "grad_norm": 0.4129182994365692,
      "learning_rate": 2.8824236396037964e-06,
      "loss": 0.6683,
      "step": 10220
    },
    {
      "epoch": 0.4235981598905881,
      "grad_norm": 0.4500209391117096,
      "learning_rate": 2.8822164200754283e-06,
      "loss": 0.7084,
      "step": 10221
    },
    {
      "epoch": 0.42363960379626175,
      "grad_norm": 0.41604897379875183,
      "learning_rate": 2.8820092005470596e-06,
      "loss": 0.6832,
      "step": 10222
    },
    {
      "epoch": 0.42368104770193543,
      "grad_norm": 0.39230993390083313,
      "learning_rate": 2.8818019810186914e-06,
      "loss": 0.7218,
      "step": 10223
    },
    {
      "epoch": 0.4237224916076091,
      "grad_norm": 0.3749372959136963,
      "learning_rate": 2.8815947614903233e-06,
      "loss": 0.6801,
      "step": 10224
    },
    {
      "epoch": 0.4237639355132828,
      "grad_norm": 0.4048909842967987,
      "learning_rate": 2.8813875419619546e-06,
      "loss": 0.6174,
      "step": 10225
    },
    {
      "epoch": 0.42380537941895646,
      "grad_norm": 0.4248400330543518,
      "learning_rate": 2.8811803224335865e-06,
      "loss": 0.7142,
      "step": 10226
    },
    {
      "epoch": 0.42384682332463014,
      "grad_norm": 0.39567089080810547,
      "learning_rate": 2.880973102905218e-06,
      "loss": 0.7222,
      "step": 10227
    },
    {
      "epoch": 0.42388826723030376,
      "grad_norm": 0.41997814178466797,
      "learning_rate": 2.8807658833768496e-06,
      "loss": 0.7324,
      "step": 10228
    },
    {
      "epoch": 0.42392971113597744,
      "grad_norm": 0.3850817382335663,
      "learning_rate": 2.880558663848481e-06,
      "loss": 0.6646,
      "step": 10229
    },
    {
      "epoch": 0.4239711550416511,
      "grad_norm": 0.3971218466758728,
      "learning_rate": 2.880351444320113e-06,
      "loss": 0.6935,
      "step": 10230
    },
    {
      "epoch": 0.4240125989473248,
      "grad_norm": 0.39058709144592285,
      "learning_rate": 2.8801442247917442e-06,
      "loss": 0.6748,
      "step": 10231
    },
    {
      "epoch": 0.42405404285299847,
      "grad_norm": 0.4227049648761749,
      "learning_rate": 2.8799370052633765e-06,
      "loss": 0.6399,
      "step": 10232
    },
    {
      "epoch": 0.42409548675867215,
      "grad_norm": 0.42481720447540283,
      "learning_rate": 2.8797297857350083e-06,
      "loss": 0.6937,
      "step": 10233
    },
    {
      "epoch": 0.4241369306643458,
      "grad_norm": 0.39221954345703125,
      "learning_rate": 2.8795225662066397e-06,
      "loss": 0.6625,
      "step": 10234
    },
    {
      "epoch": 0.4241783745700195,
      "grad_norm": 0.4086552560329437,
      "learning_rate": 2.8793153466782715e-06,
      "loss": 0.7483,
      "step": 10235
    },
    {
      "epoch": 0.4242198184756931,
      "grad_norm": 0.41614705324172974,
      "learning_rate": 2.879108127149903e-06,
      "loss": 0.6652,
      "step": 10236
    },
    {
      "epoch": 0.4242612623813668,
      "grad_norm": 0.4312654733657837,
      "learning_rate": 2.8789009076215347e-06,
      "loss": 0.6996,
      "step": 10237
    },
    {
      "epoch": 0.4243027062870405,
      "grad_norm": 0.3950953781604767,
      "learning_rate": 2.878693688093166e-06,
      "loss": 0.6656,
      "step": 10238
    },
    {
      "epoch": 0.42434415019271415,
      "grad_norm": 0.3886812925338745,
      "learning_rate": 2.878486468564798e-06,
      "loss": 0.6171,
      "step": 10239
    },
    {
      "epoch": 0.42438559409838783,
      "grad_norm": 0.42413538694381714,
      "learning_rate": 2.8782792490364292e-06,
      "loss": 0.6663,
      "step": 10240
    },
    {
      "epoch": 0.4244270380040615,
      "grad_norm": 0.3936503827571869,
      "learning_rate": 2.878072029508061e-06,
      "loss": 0.7002,
      "step": 10241
    },
    {
      "epoch": 0.4244684819097352,
      "grad_norm": 0.384168803691864,
      "learning_rate": 2.877864809979693e-06,
      "loss": 0.6533,
      "step": 10242
    },
    {
      "epoch": 0.42450992581540886,
      "grad_norm": 0.4433716833591461,
      "learning_rate": 2.8776575904513242e-06,
      "loss": 0.7679,
      "step": 10243
    },
    {
      "epoch": 0.42455136972108254,
      "grad_norm": 0.42682769894599915,
      "learning_rate": 2.877450370922956e-06,
      "loss": 0.6594,
      "step": 10244
    },
    {
      "epoch": 0.42459281362675616,
      "grad_norm": 0.3907032012939453,
      "learning_rate": 2.8772431513945874e-06,
      "loss": 0.6868,
      "step": 10245
    },
    {
      "epoch": 0.42463425753242984,
      "grad_norm": 0.39520198106765747,
      "learning_rate": 2.8770359318662192e-06,
      "loss": 0.6619,
      "step": 10246
    },
    {
      "epoch": 0.4246757014381035,
      "grad_norm": 0.3837250769138336,
      "learning_rate": 2.8768287123378506e-06,
      "loss": 0.6553,
      "step": 10247
    },
    {
      "epoch": 0.4247171453437772,
      "grad_norm": 0.42277634143829346,
      "learning_rate": 2.876621492809483e-06,
      "loss": 0.6958,
      "step": 10248
    },
    {
      "epoch": 0.42475858924945087,
      "grad_norm": 0.38644087314605713,
      "learning_rate": 2.876414273281114e-06,
      "loss": 0.6849,
      "step": 10249
    },
    {
      "epoch": 0.42480003315512455,
      "grad_norm": 0.3962564468383789,
      "learning_rate": 2.876207053752746e-06,
      "loss": 0.7198,
      "step": 10250
    },
    {
      "epoch": 0.4248414770607982,
      "grad_norm": 0.4509621560573578,
      "learning_rate": 2.875999834224378e-06,
      "loss": 0.7019,
      "step": 10251
    },
    {
      "epoch": 0.4248829209664719,
      "grad_norm": 0.4079335331916809,
      "learning_rate": 2.8757926146960093e-06,
      "loss": 0.6799,
      "step": 10252
    },
    {
      "epoch": 0.4249243648721456,
      "grad_norm": 0.41734832525253296,
      "learning_rate": 2.875585395167641e-06,
      "loss": 0.6185,
      "step": 10253
    },
    {
      "epoch": 0.4249658087778192,
      "grad_norm": 0.4063252806663513,
      "learning_rate": 2.8753781756392724e-06,
      "loss": 0.6694,
      "step": 10254
    },
    {
      "epoch": 0.4250072526834929,
      "grad_norm": 0.40629255771636963,
      "learning_rate": 2.8751709561109043e-06,
      "loss": 0.6873,
      "step": 10255
    },
    {
      "epoch": 0.42504869658916655,
      "grad_norm": 0.42590293288230896,
      "learning_rate": 2.8749637365825356e-06,
      "loss": 0.7593,
      "step": 10256
    },
    {
      "epoch": 0.42509014049484023,
      "grad_norm": 0.42158129811286926,
      "learning_rate": 2.8747565170541675e-06,
      "loss": 0.702,
      "step": 10257
    },
    {
      "epoch": 0.4251315844005139,
      "grad_norm": 0.3932560980319977,
      "learning_rate": 2.874549297525799e-06,
      "loss": 0.6526,
      "step": 10258
    },
    {
      "epoch": 0.4251730283061876,
      "grad_norm": 0.40466374158859253,
      "learning_rate": 2.8743420779974306e-06,
      "loss": 0.692,
      "step": 10259
    },
    {
      "epoch": 0.42521447221186126,
      "grad_norm": 0.4045518934726715,
      "learning_rate": 2.8741348584690625e-06,
      "loss": 0.6906,
      "step": 10260
    },
    {
      "epoch": 0.42525591611753494,
      "grad_norm": 0.40885889530181885,
      "learning_rate": 2.873927638940694e-06,
      "loss": 0.7017,
      "step": 10261
    },
    {
      "epoch": 0.4252973600232086,
      "grad_norm": 0.43215063214302063,
      "learning_rate": 2.8737204194123257e-06,
      "loss": 0.6454,
      "step": 10262
    },
    {
      "epoch": 0.42533880392888224,
      "grad_norm": 0.46865829825401306,
      "learning_rate": 2.873513199883957e-06,
      "loss": 0.743,
      "step": 10263
    },
    {
      "epoch": 0.4253802478345559,
      "grad_norm": 0.3963261544704437,
      "learning_rate": 2.873305980355589e-06,
      "loss": 0.6936,
      "step": 10264
    },
    {
      "epoch": 0.4254216917402296,
      "grad_norm": 0.39299359917640686,
      "learning_rate": 2.8730987608272202e-06,
      "loss": 0.6573,
      "step": 10265
    },
    {
      "epoch": 0.42546313564590327,
      "grad_norm": 0.40829357504844666,
      "learning_rate": 2.8728915412988525e-06,
      "loss": 0.6823,
      "step": 10266
    },
    {
      "epoch": 0.42550457955157694,
      "grad_norm": 0.4267268776893616,
      "learning_rate": 2.8726843217704834e-06,
      "loss": 0.7211,
      "step": 10267
    },
    {
      "epoch": 0.4255460234572506,
      "grad_norm": 0.43717238306999207,
      "learning_rate": 2.8724771022421157e-06,
      "loss": 0.7153,
      "step": 10268
    },
    {
      "epoch": 0.4255874673629243,
      "grad_norm": 0.38898205757141113,
      "learning_rate": 2.8722698827137475e-06,
      "loss": 0.6183,
      "step": 10269
    },
    {
      "epoch": 0.425628911268598,
      "grad_norm": 0.40572547912597656,
      "learning_rate": 2.872062663185379e-06,
      "loss": 0.7458,
      "step": 10270
    },
    {
      "epoch": 0.4256703551742716,
      "grad_norm": 0.4216849207878113,
      "learning_rate": 2.8718554436570107e-06,
      "loss": 0.7795,
      "step": 10271
    },
    {
      "epoch": 0.4257117990799453,
      "grad_norm": 0.407711386680603,
      "learning_rate": 2.871648224128642e-06,
      "loss": 0.7449,
      "step": 10272
    },
    {
      "epoch": 0.42575324298561895,
      "grad_norm": 0.37762150168418884,
      "learning_rate": 2.871441004600274e-06,
      "loss": 0.6973,
      "step": 10273
    },
    {
      "epoch": 0.42579468689129263,
      "grad_norm": 0.3813624083995819,
      "learning_rate": 2.8712337850719052e-06,
      "loss": 0.6504,
      "step": 10274
    },
    {
      "epoch": 0.4258361307969663,
      "grad_norm": 0.4144291877746582,
      "learning_rate": 2.871026565543537e-06,
      "loss": 0.675,
      "step": 10275
    },
    {
      "epoch": 0.42587757470264,
      "grad_norm": 0.4193122088909149,
      "learning_rate": 2.8708193460151684e-06,
      "loss": 0.6975,
      "step": 10276
    },
    {
      "epoch": 0.42591901860831366,
      "grad_norm": 0.3827277719974518,
      "learning_rate": 2.8706121264868002e-06,
      "loss": 0.6722,
      "step": 10277
    },
    {
      "epoch": 0.42596046251398734,
      "grad_norm": 0.4192068576812744,
      "learning_rate": 2.870404906958432e-06,
      "loss": 0.7405,
      "step": 10278
    },
    {
      "epoch": 0.426001906419661,
      "grad_norm": 0.36999237537384033,
      "learning_rate": 2.8701976874300634e-06,
      "loss": 0.6798,
      "step": 10279
    },
    {
      "epoch": 0.42604335032533464,
      "grad_norm": 0.4163936376571655,
      "learning_rate": 2.8699904679016953e-06,
      "loss": 0.7681,
      "step": 10280
    },
    {
      "epoch": 0.4260847942310083,
      "grad_norm": 0.39415282011032104,
      "learning_rate": 2.8697832483733266e-06,
      "loss": 0.7061,
      "step": 10281
    },
    {
      "epoch": 0.426126238136682,
      "grad_norm": 0.4454629719257355,
      "learning_rate": 2.869576028844959e-06,
      "loss": 0.6639,
      "step": 10282
    },
    {
      "epoch": 0.42616768204235567,
      "grad_norm": 0.40792116522789,
      "learning_rate": 2.86936880931659e-06,
      "loss": 0.6849,
      "step": 10283
    },
    {
      "epoch": 0.42620912594802934,
      "grad_norm": 0.4383392333984375,
      "learning_rate": 2.869161589788222e-06,
      "loss": 0.7009,
      "step": 10284
    },
    {
      "epoch": 0.426250569853703,
      "grad_norm": 0.3914540708065033,
      "learning_rate": 2.868954370259854e-06,
      "loss": 0.653,
      "step": 10285
    },
    {
      "epoch": 0.4262920137593767,
      "grad_norm": 1.3565726280212402,
      "learning_rate": 2.8687471507314853e-06,
      "loss": 0.6313,
      "step": 10286
    },
    {
      "epoch": 0.4263334576650504,
      "grad_norm": 0.39975056052207947,
      "learning_rate": 2.868539931203117e-06,
      "loss": 0.7212,
      "step": 10287
    },
    {
      "epoch": 0.42637490157072405,
      "grad_norm": 0.3931676745414734,
      "learning_rate": 2.8683327116747485e-06,
      "loss": 0.7012,
      "step": 10288
    },
    {
      "epoch": 0.4264163454763977,
      "grad_norm": 0.4522618055343628,
      "learning_rate": 2.8681254921463803e-06,
      "loss": 0.7581,
      "step": 10289
    },
    {
      "epoch": 0.42645778938207135,
      "grad_norm": 0.4114205539226532,
      "learning_rate": 2.8679182726180116e-06,
      "loss": 0.6775,
      "step": 10290
    },
    {
      "epoch": 0.426499233287745,
      "grad_norm": 0.4485328793525696,
      "learning_rate": 2.8677110530896435e-06,
      "loss": 0.7458,
      "step": 10291
    },
    {
      "epoch": 0.4265406771934187,
      "grad_norm": 0.45354917645454407,
      "learning_rate": 2.867503833561275e-06,
      "loss": 0.6826,
      "step": 10292
    },
    {
      "epoch": 0.4265821210990924,
      "grad_norm": 0.3831586241722107,
      "learning_rate": 2.8672966140329067e-06,
      "loss": 0.684,
      "step": 10293
    },
    {
      "epoch": 0.42662356500476606,
      "grad_norm": 0.4138661026954651,
      "learning_rate": 2.8670893945045385e-06,
      "loss": 0.7246,
      "step": 10294
    },
    {
      "epoch": 0.42666500891043974,
      "grad_norm": 0.4374277591705322,
      "learning_rate": 2.86688217497617e-06,
      "loss": 0.6868,
      "step": 10295
    },
    {
      "epoch": 0.4267064528161134,
      "grad_norm": 0.4301886558532715,
      "learning_rate": 2.8666749554478017e-06,
      "loss": 0.7124,
      "step": 10296
    },
    {
      "epoch": 0.42674789672178703,
      "grad_norm": 0.41223520040512085,
      "learning_rate": 2.866467735919433e-06,
      "loss": 0.6858,
      "step": 10297
    },
    {
      "epoch": 0.4267893406274607,
      "grad_norm": 0.39161986112594604,
      "learning_rate": 2.866260516391065e-06,
      "loss": 0.6907,
      "step": 10298
    },
    {
      "epoch": 0.4268307845331344,
      "grad_norm": 0.39950069785118103,
      "learning_rate": 2.8660532968626962e-06,
      "loss": 0.6638,
      "step": 10299
    },
    {
      "epoch": 0.42687222843880807,
      "grad_norm": 0.42116594314575195,
      "learning_rate": 2.8658460773343285e-06,
      "loss": 0.7405,
      "step": 10300
    },
    {
      "epoch": 0.42691367234448174,
      "grad_norm": 0.39591485261917114,
      "learning_rate": 2.8656388578059594e-06,
      "loss": 0.719,
      "step": 10301
    },
    {
      "epoch": 0.4269551162501554,
      "grad_norm": 0.42279744148254395,
      "learning_rate": 2.8654316382775917e-06,
      "loss": 0.7117,
      "step": 10302
    },
    {
      "epoch": 0.4269965601558291,
      "grad_norm": 0.4172136187553406,
      "learning_rate": 2.8652244187492235e-06,
      "loss": 0.7041,
      "step": 10303
    },
    {
      "epoch": 0.4270380040615028,
      "grad_norm": 0.44062983989715576,
      "learning_rate": 2.865017199220855e-06,
      "loss": 0.7278,
      "step": 10304
    },
    {
      "epoch": 0.42707944796717645,
      "grad_norm": 0.4178955852985382,
      "learning_rate": 2.8648099796924867e-06,
      "loss": 0.7178,
      "step": 10305
    },
    {
      "epoch": 0.4271208918728501,
      "grad_norm": 0.39628198742866516,
      "learning_rate": 2.864602760164118e-06,
      "loss": 0.71,
      "step": 10306
    },
    {
      "epoch": 0.42716233577852375,
      "grad_norm": 0.358176052570343,
      "learning_rate": 2.86439554063575e-06,
      "loss": 0.6376,
      "step": 10307
    },
    {
      "epoch": 0.4272037796841974,
      "grad_norm": 0.40806353092193604,
      "learning_rate": 2.8641883211073812e-06,
      "loss": 0.6686,
      "step": 10308
    },
    {
      "epoch": 0.4272452235898711,
      "grad_norm": 0.4033411145210266,
      "learning_rate": 2.863981101579013e-06,
      "loss": 0.7014,
      "step": 10309
    },
    {
      "epoch": 0.4272866674955448,
      "grad_norm": 0.4035930633544922,
      "learning_rate": 2.8637738820506444e-06,
      "loss": 0.6735,
      "step": 10310
    },
    {
      "epoch": 0.42732811140121846,
      "grad_norm": 0.38307318091392517,
      "learning_rate": 2.8635666625222763e-06,
      "loss": 0.6473,
      "step": 10311
    },
    {
      "epoch": 0.42736955530689213,
      "grad_norm": 0.4092990756034851,
      "learning_rate": 2.863359442993908e-06,
      "loss": 0.6892,
      "step": 10312
    },
    {
      "epoch": 0.4274109992125658,
      "grad_norm": 0.41131725907325745,
      "learning_rate": 2.8631522234655394e-06,
      "loss": 0.6963,
      "step": 10313
    },
    {
      "epoch": 0.4274524431182395,
      "grad_norm": 0.4274854362010956,
      "learning_rate": 2.8629450039371713e-06,
      "loss": 0.7202,
      "step": 10314
    },
    {
      "epoch": 0.4274938870239131,
      "grad_norm": 0.4018362760543823,
      "learning_rate": 2.8627377844088026e-06,
      "loss": 0.692,
      "step": 10315
    },
    {
      "epoch": 0.4275353309295868,
      "grad_norm": 0.38827648758888245,
      "learning_rate": 2.862530564880435e-06,
      "loss": 0.6609,
      "step": 10316
    },
    {
      "epoch": 0.42757677483526046,
      "grad_norm": 0.41456443071365356,
      "learning_rate": 2.862323345352066e-06,
      "loss": 0.7686,
      "step": 10317
    },
    {
      "epoch": 0.42761821874093414,
      "grad_norm": 0.45351704955101013,
      "learning_rate": 2.862116125823698e-06,
      "loss": 0.7185,
      "step": 10318
    },
    {
      "epoch": 0.4276596626466078,
      "grad_norm": 0.4366154670715332,
      "learning_rate": 2.8619089062953295e-06,
      "loss": 0.7505,
      "step": 10319
    },
    {
      "epoch": 0.4277011065522815,
      "grad_norm": 0.42559099197387695,
      "learning_rate": 2.8617016867669613e-06,
      "loss": 0.7468,
      "step": 10320
    },
    {
      "epoch": 0.4277425504579552,
      "grad_norm": 0.44532063603401184,
      "learning_rate": 2.861494467238593e-06,
      "loss": 0.7395,
      "step": 10321
    },
    {
      "epoch": 0.42778399436362885,
      "grad_norm": 0.3997159004211426,
      "learning_rate": 2.8612872477102245e-06,
      "loss": 0.6895,
      "step": 10322
    },
    {
      "epoch": 0.42782543826930247,
      "grad_norm": 0.39214181900024414,
      "learning_rate": 2.8610800281818563e-06,
      "loss": 0.6453,
      "step": 10323
    },
    {
      "epoch": 0.42786688217497615,
      "grad_norm": 0.4151853024959564,
      "learning_rate": 2.8608728086534877e-06,
      "loss": 0.7383,
      "step": 10324
    },
    {
      "epoch": 0.4279083260806498,
      "grad_norm": 0.4224843382835388,
      "learning_rate": 2.8606655891251195e-06,
      "loss": 0.744,
      "step": 10325
    },
    {
      "epoch": 0.4279497699863235,
      "grad_norm": 0.3751247227191925,
      "learning_rate": 2.860458369596751e-06,
      "loss": 0.6599,
      "step": 10326
    },
    {
      "epoch": 0.4279912138919972,
      "grad_norm": 0.4261116087436676,
      "learning_rate": 2.8602511500683827e-06,
      "loss": 0.7114,
      "step": 10327
    },
    {
      "epoch": 0.42803265779767086,
      "grad_norm": 0.36136800050735474,
      "learning_rate": 2.860043930540014e-06,
      "loss": 0.6694,
      "step": 10328
    },
    {
      "epoch": 0.42807410170334453,
      "grad_norm": 0.40281742811203003,
      "learning_rate": 2.859836711011646e-06,
      "loss": 0.7125,
      "step": 10329
    },
    {
      "epoch": 0.4281155456090182,
      "grad_norm": 0.4250551462173462,
      "learning_rate": 2.8596294914832777e-06,
      "loss": 0.7083,
      "step": 10330
    },
    {
      "epoch": 0.4281569895146919,
      "grad_norm": 0.4115763008594513,
      "learning_rate": 2.859422271954909e-06,
      "loss": 0.6924,
      "step": 10331
    },
    {
      "epoch": 0.4281984334203655,
      "grad_norm": 0.43523451685905457,
      "learning_rate": 2.8592150524265413e-06,
      "loss": 0.7241,
      "step": 10332
    },
    {
      "epoch": 0.4282398773260392,
      "grad_norm": 0.3943975865840912,
      "learning_rate": 2.8590078328981722e-06,
      "loss": 0.6353,
      "step": 10333
    },
    {
      "epoch": 0.42828132123171286,
      "grad_norm": 0.43322503566741943,
      "learning_rate": 2.8588006133698045e-06,
      "loss": 0.6724,
      "step": 10334
    },
    {
      "epoch": 0.42832276513738654,
      "grad_norm": 0.5009081363677979,
      "learning_rate": 2.8585933938414354e-06,
      "loss": 0.7222,
      "step": 10335
    },
    {
      "epoch": 0.4283642090430602,
      "grad_norm": 0.43241265416145325,
      "learning_rate": 2.8583861743130677e-06,
      "loss": 0.6643,
      "step": 10336
    },
    {
      "epoch": 0.4284056529487339,
      "grad_norm": 0.41388100385665894,
      "learning_rate": 2.858178954784699e-06,
      "loss": 0.7316,
      "step": 10337
    },
    {
      "epoch": 0.42844709685440757,
      "grad_norm": 0.403164267539978,
      "learning_rate": 2.857971735256331e-06,
      "loss": 0.6941,
      "step": 10338
    },
    {
      "epoch": 0.42848854076008125,
      "grad_norm": 0.4207041561603546,
      "learning_rate": 2.8577645157279627e-06,
      "loss": 0.6661,
      "step": 10339
    },
    {
      "epoch": 0.4285299846657549,
      "grad_norm": 0.41981980204582214,
      "learning_rate": 2.857557296199594e-06,
      "loss": 0.6494,
      "step": 10340
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.4165239632129669,
      "learning_rate": 2.857350076671226e-06,
      "loss": 0.6901,
      "step": 10341
    },
    {
      "epoch": 0.4286128724771022,
      "grad_norm": 0.4078351557254791,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.688,
      "step": 10342
    },
    {
      "epoch": 0.4286543163827759,
      "grad_norm": 0.4032996594905853,
      "learning_rate": 2.856935637614489e-06,
      "loss": 0.7085,
      "step": 10343
    },
    {
      "epoch": 0.4286957602884496,
      "grad_norm": 0.40226784348487854,
      "learning_rate": 2.8567284180861205e-06,
      "loss": 0.694,
      "step": 10344
    },
    {
      "epoch": 0.42873720419412326,
      "grad_norm": 0.4037446081638336,
      "learning_rate": 2.8565211985577523e-06,
      "loss": 0.6671,
      "step": 10345
    },
    {
      "epoch": 0.42877864809979693,
      "grad_norm": 0.42036697268486023,
      "learning_rate": 2.856313979029384e-06,
      "loss": 0.6985,
      "step": 10346
    },
    {
      "epoch": 0.4288200920054706,
      "grad_norm": 0.4235241711139679,
      "learning_rate": 2.8561067595010155e-06,
      "loss": 0.7065,
      "step": 10347
    },
    {
      "epoch": 0.4288615359111443,
      "grad_norm": 0.3868952691555023,
      "learning_rate": 2.8558995399726473e-06,
      "loss": 0.6929,
      "step": 10348
    },
    {
      "epoch": 0.42890297981681796,
      "grad_norm": 0.4019016921520233,
      "learning_rate": 2.8556923204442786e-06,
      "loss": 0.6971,
      "step": 10349
    },
    {
      "epoch": 0.4289444237224916,
      "grad_norm": 0.41189122200012207,
      "learning_rate": 2.855485100915911e-06,
      "loss": 0.7515,
      "step": 10350
    },
    {
      "epoch": 0.42898586762816526,
      "grad_norm": 0.43131476640701294,
      "learning_rate": 2.855277881387542e-06,
      "loss": 0.6801,
      "step": 10351
    },
    {
      "epoch": 0.42902731153383894,
      "grad_norm": 0.3869524300098419,
      "learning_rate": 2.855070661859174e-06,
      "loss": 0.7052,
      "step": 10352
    },
    {
      "epoch": 0.4290687554395126,
      "grad_norm": 0.4453834593296051,
      "learning_rate": 2.8548634423308055e-06,
      "loss": 0.6528,
      "step": 10353
    },
    {
      "epoch": 0.4291101993451863,
      "grad_norm": 0.41838037967681885,
      "learning_rate": 2.8546562228024373e-06,
      "loss": 0.6708,
      "step": 10354
    },
    {
      "epoch": 0.42915164325085997,
      "grad_norm": 0.409089595079422,
      "learning_rate": 2.854449003274069e-06,
      "loss": 0.7175,
      "step": 10355
    },
    {
      "epoch": 0.42919308715653365,
      "grad_norm": 0.44469740986824036,
      "learning_rate": 2.8542417837457005e-06,
      "loss": 0.7241,
      "step": 10356
    },
    {
      "epoch": 0.4292345310622073,
      "grad_norm": 0.3896887004375458,
      "learning_rate": 2.8540345642173323e-06,
      "loss": 0.6871,
      "step": 10357
    },
    {
      "epoch": 0.42927597496788095,
      "grad_norm": 0.37660789489746094,
      "learning_rate": 2.8538273446889637e-06,
      "loss": 0.6422,
      "step": 10358
    },
    {
      "epoch": 0.4293174188735546,
      "grad_norm": 0.3995661735534668,
      "learning_rate": 2.8536201251605955e-06,
      "loss": 0.6704,
      "step": 10359
    },
    {
      "epoch": 0.4293588627792283,
      "grad_norm": 0.45961079001426697,
      "learning_rate": 2.853412905632227e-06,
      "loss": 0.782,
      "step": 10360
    },
    {
      "epoch": 0.429400306684902,
      "grad_norm": 0.4425617754459381,
      "learning_rate": 2.8532056861038587e-06,
      "loss": 0.7329,
      "step": 10361
    },
    {
      "epoch": 0.42944175059057565,
      "grad_norm": 0.388287752866745,
      "learning_rate": 2.85299846657549e-06,
      "loss": 0.6654,
      "step": 10362
    },
    {
      "epoch": 0.42948319449624933,
      "grad_norm": 0.4254777133464813,
      "learning_rate": 2.852791247047122e-06,
      "loss": 0.6649,
      "step": 10363
    },
    {
      "epoch": 0.429524638401923,
      "grad_norm": 0.43070361018180847,
      "learning_rate": 2.8525840275187537e-06,
      "loss": 0.6343,
      "step": 10364
    },
    {
      "epoch": 0.4295660823075967,
      "grad_norm": 0.3828522264957428,
      "learning_rate": 2.852376807990385e-06,
      "loss": 0.6719,
      "step": 10365
    },
    {
      "epoch": 0.42960752621327036,
      "grad_norm": 0.4857997000217438,
      "learning_rate": 2.8521695884620173e-06,
      "loss": 0.781,
      "step": 10366
    },
    {
      "epoch": 0.429648970118944,
      "grad_norm": 0.4035874605178833,
      "learning_rate": 2.8519623689336482e-06,
      "loss": 0.6904,
      "step": 10367
    },
    {
      "epoch": 0.42969041402461766,
      "grad_norm": 0.41508448123931885,
      "learning_rate": 2.8517551494052805e-06,
      "loss": 0.6844,
      "step": 10368
    },
    {
      "epoch": 0.42973185793029134,
      "grad_norm": 0.3784126043319702,
      "learning_rate": 2.8515479298769114e-06,
      "loss": 0.6697,
      "step": 10369
    },
    {
      "epoch": 0.429773301835965,
      "grad_norm": 0.4123397469520569,
      "learning_rate": 2.8513407103485437e-06,
      "loss": 0.6909,
      "step": 10370
    },
    {
      "epoch": 0.4298147457416387,
      "grad_norm": 0.4014924168586731,
      "learning_rate": 2.851133490820175e-06,
      "loss": 0.6503,
      "step": 10371
    },
    {
      "epoch": 0.42985618964731237,
      "grad_norm": 0.3760715425014496,
      "learning_rate": 2.850926271291807e-06,
      "loss": 0.6814,
      "step": 10372
    },
    {
      "epoch": 0.42989763355298605,
      "grad_norm": 0.39710837602615356,
      "learning_rate": 2.8507190517634387e-06,
      "loss": 0.7124,
      "step": 10373
    },
    {
      "epoch": 0.4299390774586597,
      "grad_norm": 0.38073188066482544,
      "learning_rate": 2.85051183223507e-06,
      "loss": 0.7207,
      "step": 10374
    },
    {
      "epoch": 0.4299805213643334,
      "grad_norm": 0.39431479573249817,
      "learning_rate": 2.850304612706702e-06,
      "loss": 0.718,
      "step": 10375
    },
    {
      "epoch": 0.430021965270007,
      "grad_norm": 0.3984322249889374,
      "learning_rate": 2.8500973931783333e-06,
      "loss": 0.6854,
      "step": 10376
    },
    {
      "epoch": 0.4300634091756807,
      "grad_norm": 0.4100247919559479,
      "learning_rate": 2.849890173649965e-06,
      "loss": 0.7086,
      "step": 10377
    },
    {
      "epoch": 0.4301048530813544,
      "grad_norm": 0.40601712465286255,
      "learning_rate": 2.8496829541215965e-06,
      "loss": 0.6569,
      "step": 10378
    },
    {
      "epoch": 0.43014629698702805,
      "grad_norm": 0.39086848497390747,
      "learning_rate": 2.8494757345932283e-06,
      "loss": 0.7195,
      "step": 10379
    },
    {
      "epoch": 0.43018774089270173,
      "grad_norm": 0.41443049907684326,
      "learning_rate": 2.8492685150648597e-06,
      "loss": 0.6763,
      "step": 10380
    },
    {
      "epoch": 0.4302291847983754,
      "grad_norm": 0.41948074102401733,
      "learning_rate": 2.8490612955364915e-06,
      "loss": 0.7395,
      "step": 10381
    },
    {
      "epoch": 0.4302706287040491,
      "grad_norm": 0.4203872084617615,
      "learning_rate": 2.8488540760081233e-06,
      "loss": 0.6766,
      "step": 10382
    },
    {
      "epoch": 0.43031207260972276,
      "grad_norm": 0.41850173473358154,
      "learning_rate": 2.8486468564797547e-06,
      "loss": 0.6774,
      "step": 10383
    },
    {
      "epoch": 0.4303535165153964,
      "grad_norm": 0.4498981535434723,
      "learning_rate": 2.848439636951387e-06,
      "loss": 0.7539,
      "step": 10384
    },
    {
      "epoch": 0.43039496042107006,
      "grad_norm": 0.4484434127807617,
      "learning_rate": 2.848232417423018e-06,
      "loss": 0.7385,
      "step": 10385
    },
    {
      "epoch": 0.43043640432674374,
      "grad_norm": 0.41074103116989136,
      "learning_rate": 2.84802519789465e-06,
      "loss": 0.6959,
      "step": 10386
    },
    {
      "epoch": 0.4304778482324174,
      "grad_norm": 0.4210585355758667,
      "learning_rate": 2.8478179783662815e-06,
      "loss": 0.7473,
      "step": 10387
    },
    {
      "epoch": 0.4305192921380911,
      "grad_norm": 0.4294824004173279,
      "learning_rate": 2.8476107588379133e-06,
      "loss": 0.729,
      "step": 10388
    },
    {
      "epoch": 0.43056073604376477,
      "grad_norm": 0.401900976896286,
      "learning_rate": 2.8474035393095447e-06,
      "loss": 0.7144,
      "step": 10389
    },
    {
      "epoch": 0.43060217994943845,
      "grad_norm": 0.41022762656211853,
      "learning_rate": 2.8471963197811765e-06,
      "loss": 0.6653,
      "step": 10390
    },
    {
      "epoch": 0.4306436238551121,
      "grad_norm": 0.45915380120277405,
      "learning_rate": 2.8469891002528083e-06,
      "loss": 0.7119,
      "step": 10391
    },
    {
      "epoch": 0.4306850677607858,
      "grad_norm": 0.4486527144908905,
      "learning_rate": 2.8467818807244397e-06,
      "loss": 0.7689,
      "step": 10392
    },
    {
      "epoch": 0.4307265116664594,
      "grad_norm": 0.4198085367679596,
      "learning_rate": 2.8465746611960715e-06,
      "loss": 0.6575,
      "step": 10393
    },
    {
      "epoch": 0.4307679555721331,
      "grad_norm": 0.4376690685749054,
      "learning_rate": 2.846367441667703e-06,
      "loss": 0.7307,
      "step": 10394
    },
    {
      "epoch": 0.4308093994778068,
      "grad_norm": 0.3829568028450012,
      "learning_rate": 2.8461602221393347e-06,
      "loss": 0.6804,
      "step": 10395
    },
    {
      "epoch": 0.43085084338348045,
      "grad_norm": 0.445376455783844,
      "learning_rate": 2.845953002610966e-06,
      "loss": 0.7283,
      "step": 10396
    },
    {
      "epoch": 0.43089228728915413,
      "grad_norm": 0.42636096477508545,
      "learning_rate": 2.845745783082598e-06,
      "loss": 0.7225,
      "step": 10397
    },
    {
      "epoch": 0.4309337311948278,
      "grad_norm": 0.3926697373390198,
      "learning_rate": 2.8455385635542297e-06,
      "loss": 0.6819,
      "step": 10398
    },
    {
      "epoch": 0.4309751751005015,
      "grad_norm": 0.383539080619812,
      "learning_rate": 2.845331344025861e-06,
      "loss": 0.6531,
      "step": 10399
    },
    {
      "epoch": 0.43101661900617516,
      "grad_norm": 0.3896278440952301,
      "learning_rate": 2.8451241244974933e-06,
      "loss": 0.686,
      "step": 10400
    },
    {
      "epoch": 0.43105806291184884,
      "grad_norm": 0.45750442147254944,
      "learning_rate": 2.8449169049691243e-06,
      "loss": 0.7634,
      "step": 10401
    },
    {
      "epoch": 0.43109950681752246,
      "grad_norm": 0.5514282584190369,
      "learning_rate": 2.8447096854407565e-06,
      "loss": 0.6888,
      "step": 10402
    },
    {
      "epoch": 0.43114095072319614,
      "grad_norm": 0.4909672141075134,
      "learning_rate": 2.8445024659123875e-06,
      "loss": 0.7512,
      "step": 10403
    },
    {
      "epoch": 0.4311823946288698,
      "grad_norm": 0.40350115299224854,
      "learning_rate": 2.8442952463840197e-06,
      "loss": 0.6833,
      "step": 10404
    },
    {
      "epoch": 0.4312238385345435,
      "grad_norm": 0.4067443311214447,
      "learning_rate": 2.844088026855651e-06,
      "loss": 0.6758,
      "step": 10405
    },
    {
      "epoch": 0.43126528244021717,
      "grad_norm": 0.4207201600074768,
      "learning_rate": 2.843880807327283e-06,
      "loss": 0.6987,
      "step": 10406
    },
    {
      "epoch": 0.43130672634589085,
      "grad_norm": 0.5194283723831177,
      "learning_rate": 2.8436735877989147e-06,
      "loss": 0.7173,
      "step": 10407
    },
    {
      "epoch": 0.4313481702515645,
      "grad_norm": 0.41775834560394287,
      "learning_rate": 2.843466368270546e-06,
      "loss": 0.6826,
      "step": 10408
    },
    {
      "epoch": 0.4313896141572382,
      "grad_norm": 0.4213453531265259,
      "learning_rate": 2.843259148742178e-06,
      "loss": 0.7417,
      "step": 10409
    },
    {
      "epoch": 0.4314310580629119,
      "grad_norm": 0.4384973645210266,
      "learning_rate": 2.8430519292138093e-06,
      "loss": 0.7024,
      "step": 10410
    },
    {
      "epoch": 0.4314725019685855,
      "grad_norm": 0.4062466323375702,
      "learning_rate": 2.842844709685441e-06,
      "loss": 0.7256,
      "step": 10411
    },
    {
      "epoch": 0.4315139458742592,
      "grad_norm": 0.45456886291503906,
      "learning_rate": 2.8426374901570725e-06,
      "loss": 0.7163,
      "step": 10412
    },
    {
      "epoch": 0.43155538977993285,
      "grad_norm": 0.3806185722351074,
      "learning_rate": 2.8424302706287043e-06,
      "loss": 0.6923,
      "step": 10413
    },
    {
      "epoch": 0.43159683368560653,
      "grad_norm": 0.3872608244419098,
      "learning_rate": 2.8422230511003357e-06,
      "loss": 0.6941,
      "step": 10414
    },
    {
      "epoch": 0.4316382775912802,
      "grad_norm": 0.3723561465740204,
      "learning_rate": 2.8420158315719675e-06,
      "loss": 0.6619,
      "step": 10415
    },
    {
      "epoch": 0.4316797214969539,
      "grad_norm": 0.41451072692871094,
      "learning_rate": 2.8418086120435993e-06,
      "loss": 0.6829,
      "step": 10416
    },
    {
      "epoch": 0.43172116540262756,
      "grad_norm": 0.3857555091381073,
      "learning_rate": 2.8416013925152307e-06,
      "loss": 0.6677,
      "step": 10417
    },
    {
      "epoch": 0.43176260930830124,
      "grad_norm": 0.42607617378234863,
      "learning_rate": 2.841394172986863e-06,
      "loss": 0.6768,
      "step": 10418
    },
    {
      "epoch": 0.43180405321397486,
      "grad_norm": 0.4787174463272095,
      "learning_rate": 2.841186953458494e-06,
      "loss": 0.7783,
      "step": 10419
    },
    {
      "epoch": 0.43184549711964854,
      "grad_norm": 0.453541100025177,
      "learning_rate": 2.840979733930126e-06,
      "loss": 0.7217,
      "step": 10420
    },
    {
      "epoch": 0.4318869410253222,
      "grad_norm": 0.41865450143814087,
      "learning_rate": 2.8407725144017575e-06,
      "loss": 0.7148,
      "step": 10421
    },
    {
      "epoch": 0.4319283849309959,
      "grad_norm": 0.4176312983036041,
      "learning_rate": 2.8405652948733893e-06,
      "loss": 0.6924,
      "step": 10422
    },
    {
      "epoch": 0.43196982883666957,
      "grad_norm": 0.40825653076171875,
      "learning_rate": 2.8403580753450207e-06,
      "loss": 0.6279,
      "step": 10423
    },
    {
      "epoch": 0.43201127274234324,
      "grad_norm": 0.4302830696105957,
      "learning_rate": 2.8401508558166525e-06,
      "loss": 0.6958,
      "step": 10424
    },
    {
      "epoch": 0.4320527166480169,
      "grad_norm": 0.42328062653541565,
      "learning_rate": 2.8399436362882843e-06,
      "loss": 0.7087,
      "step": 10425
    },
    {
      "epoch": 0.4320941605536906,
      "grad_norm": 0.4232436716556549,
      "learning_rate": 2.8397364167599157e-06,
      "loss": 0.7058,
      "step": 10426
    },
    {
      "epoch": 0.4321356044593643,
      "grad_norm": 0.4146150052547455,
      "learning_rate": 2.8395291972315475e-06,
      "loss": 0.7168,
      "step": 10427
    },
    {
      "epoch": 0.4321770483650379,
      "grad_norm": 0.39475908875465393,
      "learning_rate": 2.839321977703179e-06,
      "loss": 0.7042,
      "step": 10428
    },
    {
      "epoch": 0.4322184922707116,
      "grad_norm": 0.4133051931858063,
      "learning_rate": 2.8391147581748107e-06,
      "loss": 0.7473,
      "step": 10429
    },
    {
      "epoch": 0.43225993617638525,
      "grad_norm": 0.4265364408493042,
      "learning_rate": 2.838907538646442e-06,
      "loss": 0.7161,
      "step": 10430
    },
    {
      "epoch": 0.43230138008205893,
      "grad_norm": 0.47007066011428833,
      "learning_rate": 2.838700319118074e-06,
      "loss": 0.6924,
      "step": 10431
    },
    {
      "epoch": 0.4323428239877326,
      "grad_norm": 0.38882359862327576,
      "learning_rate": 2.8384930995897053e-06,
      "loss": 0.6256,
      "step": 10432
    },
    {
      "epoch": 0.4323842678934063,
      "grad_norm": 0.4489215910434723,
      "learning_rate": 2.838285880061337e-06,
      "loss": 0.7015,
      "step": 10433
    },
    {
      "epoch": 0.43242571179907996,
      "grad_norm": 0.42065873742103577,
      "learning_rate": 2.8380786605329693e-06,
      "loss": 0.74,
      "step": 10434
    },
    {
      "epoch": 0.43246715570475364,
      "grad_norm": 0.42804017663002014,
      "learning_rate": 2.8378714410046003e-06,
      "loss": 0.6963,
      "step": 10435
    },
    {
      "epoch": 0.4325085996104273,
      "grad_norm": 0.45132169127464294,
      "learning_rate": 2.8376642214762325e-06,
      "loss": 0.6914,
      "step": 10436
    },
    {
      "epoch": 0.43255004351610093,
      "grad_norm": 0.38495537638664246,
      "learning_rate": 2.8374570019478635e-06,
      "loss": 0.7031,
      "step": 10437
    },
    {
      "epoch": 0.4325914874217746,
      "grad_norm": 0.4353320002555847,
      "learning_rate": 2.8372497824194957e-06,
      "loss": 0.6973,
      "step": 10438
    },
    {
      "epoch": 0.4326329313274483,
      "grad_norm": 0.3904997408390045,
      "learning_rate": 2.837042562891127e-06,
      "loss": 0.6505,
      "step": 10439
    },
    {
      "epoch": 0.43267437523312197,
      "grad_norm": 0.41821181774139404,
      "learning_rate": 2.836835343362759e-06,
      "loss": 0.6642,
      "step": 10440
    },
    {
      "epoch": 0.43271581913879564,
      "grad_norm": 0.42200586199760437,
      "learning_rate": 2.8366281238343903e-06,
      "loss": 0.6707,
      "step": 10441
    },
    {
      "epoch": 0.4327572630444693,
      "grad_norm": 0.45376840233802795,
      "learning_rate": 2.836420904306022e-06,
      "loss": 0.7063,
      "step": 10442
    },
    {
      "epoch": 0.432798706950143,
      "grad_norm": 0.4244289696216583,
      "learning_rate": 2.836213684777654e-06,
      "loss": 0.6737,
      "step": 10443
    },
    {
      "epoch": 0.4328401508558167,
      "grad_norm": 0.4336882531642914,
      "learning_rate": 2.8360064652492853e-06,
      "loss": 0.7075,
      "step": 10444
    },
    {
      "epoch": 0.4328815947614903,
      "grad_norm": 0.4346788823604584,
      "learning_rate": 2.835799245720917e-06,
      "loss": 0.6704,
      "step": 10445
    },
    {
      "epoch": 0.432923038667164,
      "grad_norm": 0.4044596552848816,
      "learning_rate": 2.8355920261925485e-06,
      "loss": 0.6975,
      "step": 10446
    },
    {
      "epoch": 0.43296448257283765,
      "grad_norm": 0.39014607667922974,
      "learning_rate": 2.8353848066641803e-06,
      "loss": 0.6404,
      "step": 10447
    },
    {
      "epoch": 0.4330059264785113,
      "grad_norm": 0.37580567598342896,
      "learning_rate": 2.8351775871358117e-06,
      "loss": 0.6658,
      "step": 10448
    },
    {
      "epoch": 0.433047370384185,
      "grad_norm": 0.42023444175720215,
      "learning_rate": 2.8349703676074435e-06,
      "loss": 0.7043,
      "step": 10449
    },
    {
      "epoch": 0.4330888142898587,
      "grad_norm": 0.41422587633132935,
      "learning_rate": 2.834763148079075e-06,
      "loss": 0.6907,
      "step": 10450
    },
    {
      "epoch": 0.43313025819553236,
      "grad_norm": 0.39521753787994385,
      "learning_rate": 2.8345559285507067e-06,
      "loss": 0.6328,
      "step": 10451
    },
    {
      "epoch": 0.43317170210120604,
      "grad_norm": 0.3909316062927246,
      "learning_rate": 2.834348709022339e-06,
      "loss": 0.6357,
      "step": 10452
    },
    {
      "epoch": 0.4332131460068797,
      "grad_norm": 0.4118797481060028,
      "learning_rate": 2.83414148949397e-06,
      "loss": 0.6495,
      "step": 10453
    },
    {
      "epoch": 0.43325458991255333,
      "grad_norm": 0.44610312581062317,
      "learning_rate": 2.833934269965602e-06,
      "loss": 0.7161,
      "step": 10454
    },
    {
      "epoch": 0.433296033818227,
      "grad_norm": 0.40244585275650024,
      "learning_rate": 2.8337270504372335e-06,
      "loss": 0.6704,
      "step": 10455
    },
    {
      "epoch": 0.4333374777239007,
      "grad_norm": 0.4081888496875763,
      "learning_rate": 2.8335198309088653e-06,
      "loss": 0.6914,
      "step": 10456
    },
    {
      "epoch": 0.43337892162957437,
      "grad_norm": 0.42663708329200745,
      "learning_rate": 2.8333126113804967e-06,
      "loss": 0.6818,
      "step": 10457
    },
    {
      "epoch": 0.43342036553524804,
      "grad_norm": 0.3722307085990906,
      "learning_rate": 2.8331053918521285e-06,
      "loss": 0.6589,
      "step": 10458
    },
    {
      "epoch": 0.4334618094409217,
      "grad_norm": 0.4462888240814209,
      "learning_rate": 2.8328981723237603e-06,
      "loss": 0.6964,
      "step": 10459
    },
    {
      "epoch": 0.4335032533465954,
      "grad_norm": 0.4314737319946289,
      "learning_rate": 2.8326909527953917e-06,
      "loss": 0.6882,
      "step": 10460
    },
    {
      "epoch": 0.4335446972522691,
      "grad_norm": 0.39039912819862366,
      "learning_rate": 2.8324837332670235e-06,
      "loss": 0.6716,
      "step": 10461
    },
    {
      "epoch": 0.43358614115794275,
      "grad_norm": 0.42251133918762207,
      "learning_rate": 2.832276513738655e-06,
      "loss": 0.7087,
      "step": 10462
    },
    {
      "epoch": 0.43362758506361637,
      "grad_norm": 0.36282774806022644,
      "learning_rate": 2.8320692942102867e-06,
      "loss": 0.6202,
      "step": 10463
    },
    {
      "epoch": 0.43366902896929005,
      "grad_norm": 0.42430540919303894,
      "learning_rate": 2.831862074681918e-06,
      "loss": 0.7068,
      "step": 10464
    },
    {
      "epoch": 0.4337104728749637,
      "grad_norm": 0.43808457255363464,
      "learning_rate": 2.83165485515355e-06,
      "loss": 0.687,
      "step": 10465
    },
    {
      "epoch": 0.4337519167806374,
      "grad_norm": 0.4150921106338501,
      "learning_rate": 2.8314476356251813e-06,
      "loss": 0.7036,
      "step": 10466
    },
    {
      "epoch": 0.4337933606863111,
      "grad_norm": 0.39958906173706055,
      "learning_rate": 2.831240416096813e-06,
      "loss": 0.6436,
      "step": 10467
    },
    {
      "epoch": 0.43383480459198476,
      "grad_norm": 0.37656062841415405,
      "learning_rate": 2.8310331965684453e-06,
      "loss": 0.6648,
      "step": 10468
    },
    {
      "epoch": 0.43387624849765843,
      "grad_norm": 0.399624228477478,
      "learning_rate": 2.8308259770400763e-06,
      "loss": 0.7317,
      "step": 10469
    },
    {
      "epoch": 0.4339176924033321,
      "grad_norm": 0.4033322036266327,
      "learning_rate": 2.8306187575117085e-06,
      "loss": 0.6682,
      "step": 10470
    },
    {
      "epoch": 0.43395913630900573,
      "grad_norm": 0.41484978795051575,
      "learning_rate": 2.8304115379833395e-06,
      "loss": 0.6672,
      "step": 10471
    },
    {
      "epoch": 0.4340005802146794,
      "grad_norm": 0.38175979256629944,
      "learning_rate": 2.8302043184549717e-06,
      "loss": 0.6749,
      "step": 10472
    },
    {
      "epoch": 0.4340420241203531,
      "grad_norm": 0.3942864239215851,
      "learning_rate": 2.829997098926603e-06,
      "loss": 0.6844,
      "step": 10473
    },
    {
      "epoch": 0.43408346802602676,
      "grad_norm": 0.38024571537971497,
      "learning_rate": 2.829789879398235e-06,
      "loss": 0.656,
      "step": 10474
    },
    {
      "epoch": 0.43412491193170044,
      "grad_norm": 0.3977097272872925,
      "learning_rate": 2.8295826598698663e-06,
      "loss": 0.6895,
      "step": 10475
    },
    {
      "epoch": 0.4341663558373741,
      "grad_norm": 0.39658039808273315,
      "learning_rate": 2.829375440341498e-06,
      "loss": 0.6581,
      "step": 10476
    },
    {
      "epoch": 0.4342077997430478,
      "grad_norm": 0.4145636260509491,
      "learning_rate": 2.82916822081313e-06,
      "loss": 0.6465,
      "step": 10477
    },
    {
      "epoch": 0.4342492436487215,
      "grad_norm": 0.4419283866882324,
      "learning_rate": 2.8289610012847613e-06,
      "loss": 0.7417,
      "step": 10478
    },
    {
      "epoch": 0.43429068755439515,
      "grad_norm": 0.3772049844264984,
      "learning_rate": 2.828753781756393e-06,
      "loss": 0.6864,
      "step": 10479
    },
    {
      "epoch": 0.43433213146006877,
      "grad_norm": 0.3961716294288635,
      "learning_rate": 2.8285465622280245e-06,
      "loss": 0.6528,
      "step": 10480
    },
    {
      "epoch": 0.43437357536574245,
      "grad_norm": 0.43045318126678467,
      "learning_rate": 2.8283393426996563e-06,
      "loss": 0.6707,
      "step": 10481
    },
    {
      "epoch": 0.4344150192714161,
      "grad_norm": 0.42050227522850037,
      "learning_rate": 2.8281321231712877e-06,
      "loss": 0.7131,
      "step": 10482
    },
    {
      "epoch": 0.4344564631770898,
      "grad_norm": 0.39020708203315735,
      "learning_rate": 2.8279249036429195e-06,
      "loss": 0.6646,
      "step": 10483
    },
    {
      "epoch": 0.4344979070827635,
      "grad_norm": 0.432284414768219,
      "learning_rate": 2.827717684114551e-06,
      "loss": 0.6824,
      "step": 10484
    },
    {
      "epoch": 0.43453935098843716,
      "grad_norm": 0.396045058965683,
      "learning_rate": 2.8275104645861827e-06,
      "loss": 0.6782,
      "step": 10485
    },
    {
      "epoch": 0.43458079489411083,
      "grad_norm": 0.41037964820861816,
      "learning_rate": 2.827303245057815e-06,
      "loss": 0.7098,
      "step": 10486
    },
    {
      "epoch": 0.4346222387997845,
      "grad_norm": 0.38621383905410767,
      "learning_rate": 2.827096025529446e-06,
      "loss": 0.7354,
      "step": 10487
    },
    {
      "epoch": 0.4346636827054582,
      "grad_norm": 0.39662718772888184,
      "learning_rate": 2.826888806001078e-06,
      "loss": 0.6588,
      "step": 10488
    },
    {
      "epoch": 0.4347051266111318,
      "grad_norm": 0.41946738958358765,
      "learning_rate": 2.8266815864727095e-06,
      "loss": 0.6823,
      "step": 10489
    },
    {
      "epoch": 0.4347465705168055,
      "grad_norm": 0.433270663022995,
      "learning_rate": 2.8264743669443413e-06,
      "loss": 0.7177,
      "step": 10490
    },
    {
      "epoch": 0.43478801442247916,
      "grad_norm": 0.4116487503051758,
      "learning_rate": 2.8262671474159727e-06,
      "loss": 0.6906,
      "step": 10491
    },
    {
      "epoch": 0.43482945832815284,
      "grad_norm": 0.3993769586086273,
      "learning_rate": 2.8260599278876045e-06,
      "loss": 0.6345,
      "step": 10492
    },
    {
      "epoch": 0.4348709022338265,
      "grad_norm": 0.424850195646286,
      "learning_rate": 2.825852708359236e-06,
      "loss": 0.7167,
      "step": 10493
    },
    {
      "epoch": 0.4349123461395002,
      "grad_norm": 0.402822345495224,
      "learning_rate": 2.8256454888308677e-06,
      "loss": 0.7317,
      "step": 10494
    },
    {
      "epoch": 0.43495379004517387,
      "grad_norm": 0.4139024019241333,
      "learning_rate": 2.8254382693024995e-06,
      "loss": 0.6396,
      "step": 10495
    },
    {
      "epoch": 0.43499523395084755,
      "grad_norm": 0.43480122089385986,
      "learning_rate": 2.825231049774131e-06,
      "loss": 0.6902,
      "step": 10496
    },
    {
      "epoch": 0.4350366778565212,
      "grad_norm": 0.40440455079078674,
      "learning_rate": 2.8250238302457627e-06,
      "loss": 0.6963,
      "step": 10497
    },
    {
      "epoch": 0.43507812176219485,
      "grad_norm": 0.4423624873161316,
      "learning_rate": 2.824816610717394e-06,
      "loss": 0.6774,
      "step": 10498
    },
    {
      "epoch": 0.4351195656678685,
      "grad_norm": 0.40952593088150024,
      "learning_rate": 2.824609391189026e-06,
      "loss": 0.7397,
      "step": 10499
    },
    {
      "epoch": 0.4351610095735422,
      "grad_norm": 0.40536072850227356,
      "learning_rate": 2.8244021716606573e-06,
      "loss": 0.718,
      "step": 10500
    },
    {
      "epoch": 0.4352024534792159,
      "grad_norm": 0.4244198203086853,
      "learning_rate": 2.824194952132289e-06,
      "loss": 0.7441,
      "step": 10501
    },
    {
      "epoch": 0.43524389738488956,
      "grad_norm": 0.3927190601825714,
      "learning_rate": 2.8239877326039205e-06,
      "loss": 0.6692,
      "step": 10502
    },
    {
      "epoch": 0.43528534129056323,
      "grad_norm": 0.42186030745506287,
      "learning_rate": 2.8237805130755523e-06,
      "loss": 0.7111,
      "step": 10503
    },
    {
      "epoch": 0.4353267851962369,
      "grad_norm": 0.42222970724105835,
      "learning_rate": 2.8235732935471845e-06,
      "loss": 0.7449,
      "step": 10504
    },
    {
      "epoch": 0.4353682291019106,
      "grad_norm": 0.37797150015830994,
      "learning_rate": 2.823366074018816e-06,
      "loss": 0.6899,
      "step": 10505
    },
    {
      "epoch": 0.4354096730075842,
      "grad_norm": 0.42747077345848083,
      "learning_rate": 2.8231588544904477e-06,
      "loss": 0.663,
      "step": 10506
    },
    {
      "epoch": 0.4354511169132579,
      "grad_norm": 0.41385143995285034,
      "learning_rate": 2.822951634962079e-06,
      "loss": 0.7111,
      "step": 10507
    },
    {
      "epoch": 0.43549256081893156,
      "grad_norm": 0.4049971103668213,
      "learning_rate": 2.822744415433711e-06,
      "loss": 0.6587,
      "step": 10508
    },
    {
      "epoch": 0.43553400472460524,
      "grad_norm": 0.4067429304122925,
      "learning_rate": 2.8225371959053423e-06,
      "loss": 0.7156,
      "step": 10509
    },
    {
      "epoch": 0.4355754486302789,
      "grad_norm": 0.44558238983154297,
      "learning_rate": 2.822329976376974e-06,
      "loss": 0.6871,
      "step": 10510
    },
    {
      "epoch": 0.4356168925359526,
      "grad_norm": 0.45163431763648987,
      "learning_rate": 2.8221227568486055e-06,
      "loss": 0.7032,
      "step": 10511
    },
    {
      "epoch": 0.43565833644162627,
      "grad_norm": 0.39426249265670776,
      "learning_rate": 2.8219155373202373e-06,
      "loss": 0.7511,
      "step": 10512
    },
    {
      "epoch": 0.43569978034729995,
      "grad_norm": 0.39733532071113586,
      "learning_rate": 2.821708317791869e-06,
      "loss": 0.6276,
      "step": 10513
    },
    {
      "epoch": 0.4357412242529736,
      "grad_norm": 0.4135626256465912,
      "learning_rate": 2.8215010982635005e-06,
      "loss": 0.6664,
      "step": 10514
    },
    {
      "epoch": 0.43578266815864725,
      "grad_norm": 0.3550359606742859,
      "learning_rate": 2.8212938787351323e-06,
      "loss": 0.6376,
      "step": 10515
    },
    {
      "epoch": 0.4358241120643209,
      "grad_norm": 0.401189923286438,
      "learning_rate": 2.8210866592067637e-06,
      "loss": 0.6859,
      "step": 10516
    },
    {
      "epoch": 0.4358655559699946,
      "grad_norm": 0.41182178258895874,
      "learning_rate": 2.8208794396783955e-06,
      "loss": 0.6853,
      "step": 10517
    },
    {
      "epoch": 0.4359069998756683,
      "grad_norm": 0.42777812480926514,
      "learning_rate": 2.820672220150027e-06,
      "loss": 0.7268,
      "step": 10518
    },
    {
      "epoch": 0.43594844378134195,
      "grad_norm": 0.3913375735282898,
      "learning_rate": 2.8204650006216587e-06,
      "loss": 0.6448,
      "step": 10519
    },
    {
      "epoch": 0.43598988768701563,
      "grad_norm": 0.44535398483276367,
      "learning_rate": 2.820257781093291e-06,
      "loss": 0.6997,
      "step": 10520
    },
    {
      "epoch": 0.4360313315926893,
      "grad_norm": 0.4279990792274475,
      "learning_rate": 2.820050561564922e-06,
      "loss": 0.6793,
      "step": 10521
    },
    {
      "epoch": 0.436072775498363,
      "grad_norm": 0.3860864043235779,
      "learning_rate": 2.819843342036554e-06,
      "loss": 0.6599,
      "step": 10522
    },
    {
      "epoch": 0.43611421940403666,
      "grad_norm": 0.41974228620529175,
      "learning_rate": 2.8196361225081855e-06,
      "loss": 0.7134,
      "step": 10523
    },
    {
      "epoch": 0.4361556633097103,
      "grad_norm": 0.3703858554363251,
      "learning_rate": 2.8194289029798173e-06,
      "loss": 0.6343,
      "step": 10524
    },
    {
      "epoch": 0.43619710721538396,
      "grad_norm": 0.3997543752193451,
      "learning_rate": 2.8192216834514487e-06,
      "loss": 0.678,
      "step": 10525
    },
    {
      "epoch": 0.43623855112105764,
      "grad_norm": 0.43000444769859314,
      "learning_rate": 2.8190144639230805e-06,
      "loss": 0.6931,
      "step": 10526
    },
    {
      "epoch": 0.4362799950267313,
      "grad_norm": 0.4271995723247528,
      "learning_rate": 2.818807244394712e-06,
      "loss": 0.7231,
      "step": 10527
    },
    {
      "epoch": 0.436321438932405,
      "grad_norm": 0.41930752992630005,
      "learning_rate": 2.8186000248663437e-06,
      "loss": 0.699,
      "step": 10528
    },
    {
      "epoch": 0.43636288283807867,
      "grad_norm": 0.4193907380104065,
      "learning_rate": 2.8183928053379755e-06,
      "loss": 0.7102,
      "step": 10529
    },
    {
      "epoch": 0.43640432674375235,
      "grad_norm": 0.3911597728729248,
      "learning_rate": 2.818185585809607e-06,
      "loss": 0.6589,
      "step": 10530
    },
    {
      "epoch": 0.436445770649426,
      "grad_norm": 0.40989765524864197,
      "learning_rate": 2.8179783662812387e-06,
      "loss": 0.719,
      "step": 10531
    },
    {
      "epoch": 0.43648721455509965,
      "grad_norm": 0.4120161235332489,
      "learning_rate": 2.81777114675287e-06,
      "loss": 0.7134,
      "step": 10532
    },
    {
      "epoch": 0.4365286584607733,
      "grad_norm": 0.45054763555526733,
      "learning_rate": 2.817563927224502e-06,
      "loss": 0.7251,
      "step": 10533
    },
    {
      "epoch": 0.436570102366447,
      "grad_norm": 0.40333986282348633,
      "learning_rate": 2.8173567076961333e-06,
      "loss": 0.6725,
      "step": 10534
    },
    {
      "epoch": 0.4366115462721207,
      "grad_norm": 0.46302980184555054,
      "learning_rate": 2.817149488167765e-06,
      "loss": 0.7195,
      "step": 10535
    },
    {
      "epoch": 0.43665299017779435,
      "grad_norm": 0.4543590247631073,
      "learning_rate": 2.8169422686393965e-06,
      "loss": 0.6653,
      "step": 10536
    },
    {
      "epoch": 0.43669443408346803,
      "grad_norm": 0.4041726291179657,
      "learning_rate": 2.8167350491110283e-06,
      "loss": 0.7134,
      "step": 10537
    },
    {
      "epoch": 0.4367358779891417,
      "grad_norm": 0.4104674160480499,
      "learning_rate": 2.8165278295826605e-06,
      "loss": 0.6461,
      "step": 10538
    },
    {
      "epoch": 0.4367773218948154,
      "grad_norm": 0.4113777279853821,
      "learning_rate": 2.816320610054292e-06,
      "loss": 0.6572,
      "step": 10539
    },
    {
      "epoch": 0.43681876580048906,
      "grad_norm": 0.41088980436325073,
      "learning_rate": 2.8161133905259237e-06,
      "loss": 0.6776,
      "step": 10540
    },
    {
      "epoch": 0.4368602097061627,
      "grad_norm": 0.4095016419887543,
      "learning_rate": 2.815906170997555e-06,
      "loss": 0.693,
      "step": 10541
    },
    {
      "epoch": 0.43690165361183636,
      "grad_norm": 0.4238571524620056,
      "learning_rate": 2.815698951469187e-06,
      "loss": 0.6946,
      "step": 10542
    },
    {
      "epoch": 0.43694309751751004,
      "grad_norm": 0.4106113910675049,
      "learning_rate": 2.8154917319408183e-06,
      "loss": 0.7159,
      "step": 10543
    },
    {
      "epoch": 0.4369845414231837,
      "grad_norm": 0.3952290117740631,
      "learning_rate": 2.81528451241245e-06,
      "loss": 0.696,
      "step": 10544
    },
    {
      "epoch": 0.4370259853288574,
      "grad_norm": 0.4206458330154419,
      "learning_rate": 2.8150772928840815e-06,
      "loss": 0.6693,
      "step": 10545
    },
    {
      "epoch": 0.43706742923453107,
      "grad_norm": 0.41807064414024353,
      "learning_rate": 2.8148700733557133e-06,
      "loss": 0.6824,
      "step": 10546
    },
    {
      "epoch": 0.43710887314020475,
      "grad_norm": 0.3907195031642914,
      "learning_rate": 2.814662853827345e-06,
      "loss": 0.6686,
      "step": 10547
    },
    {
      "epoch": 0.4371503170458784,
      "grad_norm": 0.39504575729370117,
      "learning_rate": 2.8144556342989765e-06,
      "loss": 0.6621,
      "step": 10548
    },
    {
      "epoch": 0.4371917609515521,
      "grad_norm": 0.375581294298172,
      "learning_rate": 2.8142484147706083e-06,
      "loss": 0.6389,
      "step": 10549
    },
    {
      "epoch": 0.4372332048572257,
      "grad_norm": 0.4173888564109802,
      "learning_rate": 2.8140411952422397e-06,
      "loss": 0.6973,
      "step": 10550
    },
    {
      "epoch": 0.4372746487628994,
      "grad_norm": 0.4885769486427307,
      "learning_rate": 2.8138339757138715e-06,
      "loss": 0.7322,
      "step": 10551
    },
    {
      "epoch": 0.4373160926685731,
      "grad_norm": 0.4213376045227051,
      "learning_rate": 2.813626756185503e-06,
      "loss": 0.7437,
      "step": 10552
    },
    {
      "epoch": 0.43735753657424675,
      "grad_norm": 0.40378329157829285,
      "learning_rate": 2.8134195366571347e-06,
      "loss": 0.6732,
      "step": 10553
    },
    {
      "epoch": 0.43739898047992043,
      "grad_norm": 0.40505045652389526,
      "learning_rate": 2.813212317128766e-06,
      "loss": 0.7404,
      "step": 10554
    },
    {
      "epoch": 0.4374404243855941,
      "grad_norm": 0.4088124930858612,
      "learning_rate": 2.813005097600398e-06,
      "loss": 0.7034,
      "step": 10555
    },
    {
      "epoch": 0.4374818682912678,
      "grad_norm": 0.38331568241119385,
      "learning_rate": 2.81279787807203e-06,
      "loss": 0.6522,
      "step": 10556
    },
    {
      "epoch": 0.43752331219694146,
      "grad_norm": 0.45342105627059937,
      "learning_rate": 2.8125906585436615e-06,
      "loss": 0.7258,
      "step": 10557
    },
    {
      "epoch": 0.43756475610261514,
      "grad_norm": 0.43386366963386536,
      "learning_rate": 2.8123834390152933e-06,
      "loss": 0.738,
      "step": 10558
    },
    {
      "epoch": 0.43760620000828876,
      "grad_norm": 0.403163343667984,
      "learning_rate": 2.8121762194869247e-06,
      "loss": 0.6968,
      "step": 10559
    },
    {
      "epoch": 0.43764764391396244,
      "grad_norm": 0.4145965874195099,
      "learning_rate": 2.8119689999585565e-06,
      "loss": 0.6768,
      "step": 10560
    },
    {
      "epoch": 0.4376890878196361,
      "grad_norm": 0.415755033493042,
      "learning_rate": 2.811761780430188e-06,
      "loss": 0.6965,
      "step": 10561
    },
    {
      "epoch": 0.4377305317253098,
      "grad_norm": 0.4121474325656891,
      "learning_rate": 2.8115545609018197e-06,
      "loss": 0.6301,
      "step": 10562
    },
    {
      "epoch": 0.43777197563098347,
      "grad_norm": 0.4487161338329315,
      "learning_rate": 2.811347341373451e-06,
      "loss": 0.6838,
      "step": 10563
    },
    {
      "epoch": 0.43781341953665714,
      "grad_norm": 0.4149439334869385,
      "learning_rate": 2.811140121845083e-06,
      "loss": 0.6677,
      "step": 10564
    },
    {
      "epoch": 0.4378548634423308,
      "grad_norm": 0.47265148162841797,
      "learning_rate": 2.8109329023167147e-06,
      "loss": 0.7224,
      "step": 10565
    },
    {
      "epoch": 0.4378963073480045,
      "grad_norm": 0.3953212797641754,
      "learning_rate": 2.810725682788346e-06,
      "loss": 0.696,
      "step": 10566
    },
    {
      "epoch": 0.4379377512536781,
      "grad_norm": 0.4550792872905731,
      "learning_rate": 2.810518463259978e-06,
      "loss": 0.6948,
      "step": 10567
    },
    {
      "epoch": 0.4379791951593518,
      "grad_norm": 0.4062862992286682,
      "learning_rate": 2.8103112437316093e-06,
      "loss": 0.6743,
      "step": 10568
    },
    {
      "epoch": 0.4380206390650255,
      "grad_norm": 0.42095720767974854,
      "learning_rate": 2.810104024203241e-06,
      "loss": 0.6003,
      "step": 10569
    },
    {
      "epoch": 0.43806208297069915,
      "grad_norm": 0.40781882405281067,
      "learning_rate": 2.8098968046748725e-06,
      "loss": 0.7029,
      "step": 10570
    },
    {
      "epoch": 0.43810352687637283,
      "grad_norm": 0.4319226145744324,
      "learning_rate": 2.8096895851465043e-06,
      "loss": 0.6348,
      "step": 10571
    },
    {
      "epoch": 0.4381449707820465,
      "grad_norm": 0.4180569052696228,
      "learning_rate": 2.8094823656181357e-06,
      "loss": 0.7301,
      "step": 10572
    },
    {
      "epoch": 0.4381864146877202,
      "grad_norm": 0.4475403130054474,
      "learning_rate": 2.809275146089768e-06,
      "loss": 0.7253,
      "step": 10573
    },
    {
      "epoch": 0.43822785859339386,
      "grad_norm": 0.44447505474090576,
      "learning_rate": 2.8090679265613997e-06,
      "loss": 0.7538,
      "step": 10574
    },
    {
      "epoch": 0.43826930249906754,
      "grad_norm": 0.3841472268104553,
      "learning_rate": 2.808860707033031e-06,
      "loss": 0.6349,
      "step": 10575
    },
    {
      "epoch": 0.43831074640474116,
      "grad_norm": 0.41565394401550293,
      "learning_rate": 2.808653487504663e-06,
      "loss": 0.6914,
      "step": 10576
    },
    {
      "epoch": 0.43835219031041484,
      "grad_norm": 0.37992823123931885,
      "learning_rate": 2.8084462679762943e-06,
      "loss": 0.6667,
      "step": 10577
    },
    {
      "epoch": 0.4383936342160885,
      "grad_norm": 0.4416390657424927,
      "learning_rate": 2.808239048447926e-06,
      "loss": 0.7446,
      "step": 10578
    },
    {
      "epoch": 0.4384350781217622,
      "grad_norm": 0.41148489713668823,
      "learning_rate": 2.8080318289195575e-06,
      "loss": 0.7119,
      "step": 10579
    },
    {
      "epoch": 0.43847652202743587,
      "grad_norm": 0.4190036356449127,
      "learning_rate": 2.8078246093911893e-06,
      "loss": 0.6683,
      "step": 10580
    },
    {
      "epoch": 0.43851796593310954,
      "grad_norm": 0.4412931203842163,
      "learning_rate": 2.807617389862821e-06,
      "loss": 0.6141,
      "step": 10581
    },
    {
      "epoch": 0.4385594098387832,
      "grad_norm": 0.41133633255958557,
      "learning_rate": 2.8074101703344525e-06,
      "loss": 0.7148,
      "step": 10582
    },
    {
      "epoch": 0.4386008537444569,
      "grad_norm": 0.4217776358127594,
      "learning_rate": 2.8072029508060843e-06,
      "loss": 0.7068,
      "step": 10583
    },
    {
      "epoch": 0.4386422976501306,
      "grad_norm": 0.4044494926929474,
      "learning_rate": 2.8069957312777157e-06,
      "loss": 0.6781,
      "step": 10584
    },
    {
      "epoch": 0.4386837415558042,
      "grad_norm": 0.4196532070636749,
      "learning_rate": 2.8067885117493475e-06,
      "loss": 0.6812,
      "step": 10585
    },
    {
      "epoch": 0.4387251854614779,
      "grad_norm": 0.40873807668685913,
      "learning_rate": 2.806581292220979e-06,
      "loss": 0.6827,
      "step": 10586
    },
    {
      "epoch": 0.43876662936715155,
      "grad_norm": 0.39048802852630615,
      "learning_rate": 2.8063740726926107e-06,
      "loss": 0.6289,
      "step": 10587
    },
    {
      "epoch": 0.4388080732728252,
      "grad_norm": 0.39452847838401794,
      "learning_rate": 2.806166853164242e-06,
      "loss": 0.7085,
      "step": 10588
    },
    {
      "epoch": 0.4388495171784989,
      "grad_norm": 0.4516887664794922,
      "learning_rate": 2.805959633635874e-06,
      "loss": 0.7197,
      "step": 10589
    },
    {
      "epoch": 0.4388909610841726,
      "grad_norm": 0.41256222128868103,
      "learning_rate": 2.805752414107506e-06,
      "loss": 0.7128,
      "step": 10590
    },
    {
      "epoch": 0.43893240498984626,
      "grad_norm": 0.4292495548725128,
      "learning_rate": 2.8055451945791375e-06,
      "loss": 0.7139,
      "step": 10591
    },
    {
      "epoch": 0.43897384889551994,
      "grad_norm": 0.3860321342945099,
      "learning_rate": 2.8053379750507693e-06,
      "loss": 0.7107,
      "step": 10592
    },
    {
      "epoch": 0.43901529280119356,
      "grad_norm": 0.3742777407169342,
      "learning_rate": 2.8051307555224007e-06,
      "loss": 0.6594,
      "step": 10593
    },
    {
      "epoch": 0.43905673670686723,
      "grad_norm": 0.425051212310791,
      "learning_rate": 2.8049235359940325e-06,
      "loss": 0.6935,
      "step": 10594
    },
    {
      "epoch": 0.4390981806125409,
      "grad_norm": 0.4327065944671631,
      "learning_rate": 2.804716316465664e-06,
      "loss": 0.6772,
      "step": 10595
    },
    {
      "epoch": 0.4391396245182146,
      "grad_norm": 0.3994385600090027,
      "learning_rate": 2.8045090969372957e-06,
      "loss": 0.7175,
      "step": 10596
    },
    {
      "epoch": 0.43918106842388827,
      "grad_norm": 0.44643959403038025,
      "learning_rate": 2.804301877408927e-06,
      "loss": 0.7716,
      "step": 10597
    },
    {
      "epoch": 0.43922251232956194,
      "grad_norm": 0.40110304951667786,
      "learning_rate": 2.804094657880559e-06,
      "loss": 0.6873,
      "step": 10598
    },
    {
      "epoch": 0.4392639562352356,
      "grad_norm": 0.45577630400657654,
      "learning_rate": 2.8038874383521907e-06,
      "loss": 0.7759,
      "step": 10599
    },
    {
      "epoch": 0.4393054001409093,
      "grad_norm": 0.4214892089366913,
      "learning_rate": 2.803680218823822e-06,
      "loss": 0.6566,
      "step": 10600
    },
    {
      "epoch": 0.439346844046583,
      "grad_norm": 0.4173821806907654,
      "learning_rate": 2.803472999295454e-06,
      "loss": 0.6888,
      "step": 10601
    },
    {
      "epoch": 0.4393882879522566,
      "grad_norm": 0.40902063250541687,
      "learning_rate": 2.8032657797670853e-06,
      "loss": 0.7188,
      "step": 10602
    },
    {
      "epoch": 0.4394297318579303,
      "grad_norm": 0.4221217930316925,
      "learning_rate": 2.803058560238717e-06,
      "loss": 0.7158,
      "step": 10603
    },
    {
      "epoch": 0.43947117576360395,
      "grad_norm": 0.4052228331565857,
      "learning_rate": 2.8028513407103485e-06,
      "loss": 0.7188,
      "step": 10604
    },
    {
      "epoch": 0.4395126196692776,
      "grad_norm": 0.3986262381076813,
      "learning_rate": 2.8026441211819803e-06,
      "loss": 0.72,
      "step": 10605
    },
    {
      "epoch": 0.4395540635749513,
      "grad_norm": 0.42035117745399475,
      "learning_rate": 2.8024369016536117e-06,
      "loss": 0.6899,
      "step": 10606
    },
    {
      "epoch": 0.439595507480625,
      "grad_norm": 0.4032190442085266,
      "learning_rate": 2.802229682125244e-06,
      "loss": 0.7178,
      "step": 10607
    },
    {
      "epoch": 0.43963695138629866,
      "grad_norm": 0.4030471444129944,
      "learning_rate": 2.8020224625968757e-06,
      "loss": 0.6398,
      "step": 10608
    },
    {
      "epoch": 0.43967839529197233,
      "grad_norm": 0.44073107838630676,
      "learning_rate": 2.801815243068507e-06,
      "loss": 0.7056,
      "step": 10609
    },
    {
      "epoch": 0.439719839197646,
      "grad_norm": 0.3942532539367676,
      "learning_rate": 2.801608023540139e-06,
      "loss": 0.6741,
      "step": 10610
    },
    {
      "epoch": 0.43976128310331963,
      "grad_norm": 0.4088027775287628,
      "learning_rate": 2.8014008040117703e-06,
      "loss": 0.688,
      "step": 10611
    },
    {
      "epoch": 0.4398027270089933,
      "grad_norm": 0.4064330458641052,
      "learning_rate": 2.801193584483402e-06,
      "loss": 0.6643,
      "step": 10612
    },
    {
      "epoch": 0.439844170914667,
      "grad_norm": 0.47182697057724,
      "learning_rate": 2.8009863649550335e-06,
      "loss": 0.6802,
      "step": 10613
    },
    {
      "epoch": 0.43988561482034066,
      "grad_norm": 0.4067172408103943,
      "learning_rate": 2.8007791454266653e-06,
      "loss": 0.6436,
      "step": 10614
    },
    {
      "epoch": 0.43992705872601434,
      "grad_norm": 0.39407485723495483,
      "learning_rate": 2.8005719258982967e-06,
      "loss": 0.6866,
      "step": 10615
    },
    {
      "epoch": 0.439968502631688,
      "grad_norm": 0.4145001173019409,
      "learning_rate": 2.8003647063699285e-06,
      "loss": 0.6649,
      "step": 10616
    },
    {
      "epoch": 0.4400099465373617,
      "grad_norm": 0.40470680594444275,
      "learning_rate": 2.8001574868415603e-06,
      "loss": 0.6641,
      "step": 10617
    },
    {
      "epoch": 0.4400513904430354,
      "grad_norm": 0.4191378057003021,
      "learning_rate": 2.7999502673131917e-06,
      "loss": 0.6985,
      "step": 10618
    },
    {
      "epoch": 0.44009283434870905,
      "grad_norm": 0.42289814352989197,
      "learning_rate": 2.7997430477848235e-06,
      "loss": 0.7471,
      "step": 10619
    },
    {
      "epoch": 0.44013427825438267,
      "grad_norm": 0.398616224527359,
      "learning_rate": 2.799535828256455e-06,
      "loss": 0.693,
      "step": 10620
    },
    {
      "epoch": 0.44017572216005635,
      "grad_norm": 0.3996465802192688,
      "learning_rate": 2.7993286087280867e-06,
      "loss": 0.6963,
      "step": 10621
    },
    {
      "epoch": 0.44021716606573,
      "grad_norm": 0.39495009183883667,
      "learning_rate": 2.799121389199718e-06,
      "loss": 0.6992,
      "step": 10622
    },
    {
      "epoch": 0.4402586099714037,
      "grad_norm": 0.43602702021598816,
      "learning_rate": 2.79891416967135e-06,
      "loss": 0.6582,
      "step": 10623
    },
    {
      "epoch": 0.4403000538770774,
      "grad_norm": 0.4120887815952301,
      "learning_rate": 2.7987069501429813e-06,
      "loss": 0.7324,
      "step": 10624
    },
    {
      "epoch": 0.44034149778275106,
      "grad_norm": 0.4042099118232727,
      "learning_rate": 2.7984997306146135e-06,
      "loss": 0.6731,
      "step": 10625
    },
    {
      "epoch": 0.44038294168842473,
      "grad_norm": 0.40953895449638367,
      "learning_rate": 2.7982925110862453e-06,
      "loss": 0.7229,
      "step": 10626
    },
    {
      "epoch": 0.4404243855940984,
      "grad_norm": 0.4370701313018799,
      "learning_rate": 2.7980852915578767e-06,
      "loss": 0.7239,
      "step": 10627
    },
    {
      "epoch": 0.44046582949977203,
      "grad_norm": 0.3957687020301819,
      "learning_rate": 2.7978780720295085e-06,
      "loss": 0.6882,
      "step": 10628
    },
    {
      "epoch": 0.4405072734054457,
      "grad_norm": 0.43923306465148926,
      "learning_rate": 2.79767085250114e-06,
      "loss": 0.7203,
      "step": 10629
    },
    {
      "epoch": 0.4405487173111194,
      "grad_norm": 0.4286334216594696,
      "learning_rate": 2.7974636329727717e-06,
      "loss": 0.6553,
      "step": 10630
    },
    {
      "epoch": 0.44059016121679306,
      "grad_norm": 0.4045356512069702,
      "learning_rate": 2.797256413444403e-06,
      "loss": 0.6677,
      "step": 10631
    },
    {
      "epoch": 0.44063160512246674,
      "grad_norm": 0.425481915473938,
      "learning_rate": 2.797049193916035e-06,
      "loss": 0.6757,
      "step": 10632
    },
    {
      "epoch": 0.4406730490281404,
      "grad_norm": 0.4431073069572449,
      "learning_rate": 2.7968419743876663e-06,
      "loss": 0.676,
      "step": 10633
    },
    {
      "epoch": 0.4407144929338141,
      "grad_norm": 0.39554330706596375,
      "learning_rate": 2.796634754859298e-06,
      "loss": 0.7019,
      "step": 10634
    },
    {
      "epoch": 0.44075593683948777,
      "grad_norm": 0.4320926070213318,
      "learning_rate": 2.79642753533093e-06,
      "loss": 0.7622,
      "step": 10635
    },
    {
      "epoch": 0.44079738074516145,
      "grad_norm": 0.407745897769928,
      "learning_rate": 2.7962203158025613e-06,
      "loss": 0.7046,
      "step": 10636
    },
    {
      "epoch": 0.44083882465083507,
      "grad_norm": 0.43103858828544617,
      "learning_rate": 2.796013096274193e-06,
      "loss": 0.7179,
      "step": 10637
    },
    {
      "epoch": 0.44088026855650875,
      "grad_norm": 0.4133353531360626,
      "learning_rate": 2.7958058767458245e-06,
      "loss": 0.7064,
      "step": 10638
    },
    {
      "epoch": 0.4409217124621824,
      "grad_norm": 0.40666496753692627,
      "learning_rate": 2.7955986572174563e-06,
      "loss": 0.6741,
      "step": 10639
    },
    {
      "epoch": 0.4409631563678561,
      "grad_norm": 0.41530531644821167,
      "learning_rate": 2.7953914376890877e-06,
      "loss": 0.7004,
      "step": 10640
    },
    {
      "epoch": 0.4410046002735298,
      "grad_norm": 0.4393569231033325,
      "learning_rate": 2.79518421816072e-06,
      "loss": 0.7197,
      "step": 10641
    },
    {
      "epoch": 0.44104604417920346,
      "grad_norm": 0.4229731261730194,
      "learning_rate": 2.7949769986323517e-06,
      "loss": 0.7461,
      "step": 10642
    },
    {
      "epoch": 0.44108748808487713,
      "grad_norm": 0.41027265787124634,
      "learning_rate": 2.794769779103983e-06,
      "loss": 0.7251,
      "step": 10643
    },
    {
      "epoch": 0.4411289319905508,
      "grad_norm": 0.5027011036872864,
      "learning_rate": 2.794562559575615e-06,
      "loss": 0.6592,
      "step": 10644
    },
    {
      "epoch": 0.4411703758962245,
      "grad_norm": 0.39715152978897095,
      "learning_rate": 2.7943553400472463e-06,
      "loss": 0.6826,
      "step": 10645
    },
    {
      "epoch": 0.4412118198018981,
      "grad_norm": 0.42920026183128357,
      "learning_rate": 2.794148120518878e-06,
      "loss": 0.6713,
      "step": 10646
    },
    {
      "epoch": 0.4412532637075718,
      "grad_norm": 0.38504379987716675,
      "learning_rate": 2.7939409009905095e-06,
      "loss": 0.6515,
      "step": 10647
    },
    {
      "epoch": 0.44129470761324546,
      "grad_norm": 0.4209655821323395,
      "learning_rate": 2.7937336814621413e-06,
      "loss": 0.6528,
      "step": 10648
    },
    {
      "epoch": 0.44133615151891914,
      "grad_norm": 0.41192448139190674,
      "learning_rate": 2.7935264619337727e-06,
      "loss": 0.699,
      "step": 10649
    },
    {
      "epoch": 0.4413775954245928,
      "grad_norm": 0.46252885460853577,
      "learning_rate": 2.7933192424054045e-06,
      "loss": 0.7214,
      "step": 10650
    },
    {
      "epoch": 0.4414190393302665,
      "grad_norm": 0.4012692868709564,
      "learning_rate": 2.7931120228770363e-06,
      "loss": 0.7068,
      "step": 10651
    },
    {
      "epoch": 0.44146048323594017,
      "grad_norm": 0.4071120023727417,
      "learning_rate": 2.7929048033486677e-06,
      "loss": 0.6722,
      "step": 10652
    },
    {
      "epoch": 0.44150192714161385,
      "grad_norm": 0.3719985783100128,
      "learning_rate": 2.7926975838202995e-06,
      "loss": 0.6228,
      "step": 10653
    },
    {
      "epoch": 0.44154337104728747,
      "grad_norm": 0.40987682342529297,
      "learning_rate": 2.792490364291931e-06,
      "loss": 0.6926,
      "step": 10654
    },
    {
      "epoch": 0.44158481495296115,
      "grad_norm": 0.3787139058113098,
      "learning_rate": 2.7922831447635627e-06,
      "loss": 0.6658,
      "step": 10655
    },
    {
      "epoch": 0.4416262588586348,
      "grad_norm": 0.4065896272659302,
      "learning_rate": 2.792075925235194e-06,
      "loss": 0.7168,
      "step": 10656
    },
    {
      "epoch": 0.4416677027643085,
      "grad_norm": 0.4119814336299896,
      "learning_rate": 2.791868705706826e-06,
      "loss": 0.6492,
      "step": 10657
    },
    {
      "epoch": 0.4417091466699822,
      "grad_norm": 0.44896841049194336,
      "learning_rate": 2.7916614861784573e-06,
      "loss": 0.752,
      "step": 10658
    },
    {
      "epoch": 0.44175059057565585,
      "grad_norm": 0.439333438873291,
      "learning_rate": 2.7914542666500895e-06,
      "loss": 0.6864,
      "step": 10659
    },
    {
      "epoch": 0.44179203448132953,
      "grad_norm": 0.4057009518146515,
      "learning_rate": 2.7912470471217213e-06,
      "loss": 0.7283,
      "step": 10660
    },
    {
      "epoch": 0.4418334783870032,
      "grad_norm": 0.4217979609966278,
      "learning_rate": 2.7910398275933527e-06,
      "loss": 0.7349,
      "step": 10661
    },
    {
      "epoch": 0.4418749222926769,
      "grad_norm": 0.40723666548728943,
      "learning_rate": 2.7908326080649845e-06,
      "loss": 0.7119,
      "step": 10662
    },
    {
      "epoch": 0.4419163661983505,
      "grad_norm": 0.41493383049964905,
      "learning_rate": 2.790625388536616e-06,
      "loss": 0.7471,
      "step": 10663
    },
    {
      "epoch": 0.4419578101040242,
      "grad_norm": 0.4638195037841797,
      "learning_rate": 2.7904181690082477e-06,
      "loss": 0.756,
      "step": 10664
    },
    {
      "epoch": 0.44199925400969786,
      "grad_norm": 0.39149704575538635,
      "learning_rate": 2.790210949479879e-06,
      "loss": 0.6951,
      "step": 10665
    },
    {
      "epoch": 0.44204069791537154,
      "grad_norm": 0.41711366176605225,
      "learning_rate": 2.790003729951511e-06,
      "loss": 0.6803,
      "step": 10666
    },
    {
      "epoch": 0.4420821418210452,
      "grad_norm": 0.4307255744934082,
      "learning_rate": 2.7897965104231423e-06,
      "loss": 0.7043,
      "step": 10667
    },
    {
      "epoch": 0.4421235857267189,
      "grad_norm": 0.4422287046909332,
      "learning_rate": 2.789589290894774e-06,
      "loss": 0.7371,
      "step": 10668
    },
    {
      "epoch": 0.44216502963239257,
      "grad_norm": 0.4224344491958618,
      "learning_rate": 2.789382071366406e-06,
      "loss": 0.7251,
      "step": 10669
    },
    {
      "epoch": 0.44220647353806625,
      "grad_norm": 0.3960646390914917,
      "learning_rate": 2.7891748518380373e-06,
      "loss": 0.7131,
      "step": 10670
    },
    {
      "epoch": 0.4422479174437399,
      "grad_norm": 0.4138723313808441,
      "learning_rate": 2.788967632309669e-06,
      "loss": 0.7058,
      "step": 10671
    },
    {
      "epoch": 0.44228936134941355,
      "grad_norm": 0.43399345874786377,
      "learning_rate": 2.7887604127813005e-06,
      "loss": 0.7166,
      "step": 10672
    },
    {
      "epoch": 0.4423308052550872,
      "grad_norm": 0.3994515836238861,
      "learning_rate": 2.7885531932529323e-06,
      "loss": 0.6901,
      "step": 10673
    },
    {
      "epoch": 0.4423722491607609,
      "grad_norm": 0.41371437907218933,
      "learning_rate": 2.7883459737245637e-06,
      "loss": 0.6648,
      "step": 10674
    },
    {
      "epoch": 0.4424136930664346,
      "grad_norm": 0.4053540527820587,
      "learning_rate": 2.788138754196196e-06,
      "loss": 0.6399,
      "step": 10675
    },
    {
      "epoch": 0.44245513697210825,
      "grad_norm": 0.44357797503471375,
      "learning_rate": 2.787931534667827e-06,
      "loss": 0.749,
      "step": 10676
    },
    {
      "epoch": 0.44249658087778193,
      "grad_norm": 0.37790152430534363,
      "learning_rate": 2.787724315139459e-06,
      "loss": 0.6646,
      "step": 10677
    },
    {
      "epoch": 0.4425380247834556,
      "grad_norm": 0.43542614579200745,
      "learning_rate": 2.787517095611091e-06,
      "loss": 0.7091,
      "step": 10678
    },
    {
      "epoch": 0.4425794686891293,
      "grad_norm": 0.3899983763694763,
      "learning_rate": 2.7873098760827223e-06,
      "loss": 0.6743,
      "step": 10679
    },
    {
      "epoch": 0.4426209125948029,
      "grad_norm": 0.41934314370155334,
      "learning_rate": 2.787102656554354e-06,
      "loss": 0.679,
      "step": 10680
    },
    {
      "epoch": 0.4426623565004766,
      "grad_norm": 0.4647693634033203,
      "learning_rate": 2.7868954370259855e-06,
      "loss": 0.7148,
      "step": 10681
    },
    {
      "epoch": 0.44270380040615026,
      "grad_norm": 0.4176410436630249,
      "learning_rate": 2.7866882174976173e-06,
      "loss": 0.6902,
      "step": 10682
    },
    {
      "epoch": 0.44274524431182394,
      "grad_norm": 0.4041560888290405,
      "learning_rate": 2.7864809979692487e-06,
      "loss": 0.6658,
      "step": 10683
    },
    {
      "epoch": 0.4427866882174976,
      "grad_norm": 0.40555325150489807,
      "learning_rate": 2.7862737784408805e-06,
      "loss": 0.6956,
      "step": 10684
    },
    {
      "epoch": 0.4428281321231713,
      "grad_norm": 0.3938453197479248,
      "learning_rate": 2.786066558912512e-06,
      "loss": 0.6721,
      "step": 10685
    },
    {
      "epoch": 0.44286957602884497,
      "grad_norm": 0.4473835229873657,
      "learning_rate": 2.7858593393841437e-06,
      "loss": 0.7321,
      "step": 10686
    },
    {
      "epoch": 0.44291101993451865,
      "grad_norm": 0.3867129385471344,
      "learning_rate": 2.7856521198557755e-06,
      "loss": 0.6763,
      "step": 10687
    },
    {
      "epoch": 0.4429524638401923,
      "grad_norm": 0.3999990224838257,
      "learning_rate": 2.785444900327407e-06,
      "loss": 0.7057,
      "step": 10688
    },
    {
      "epoch": 0.44299390774586594,
      "grad_norm": 0.3968227803707123,
      "learning_rate": 2.7852376807990387e-06,
      "loss": 0.6956,
      "step": 10689
    },
    {
      "epoch": 0.4430353516515396,
      "grad_norm": 0.45817697048187256,
      "learning_rate": 2.78503046127067e-06,
      "loss": 0.6852,
      "step": 10690
    },
    {
      "epoch": 0.4430767955572133,
      "grad_norm": 0.4271210730075836,
      "learning_rate": 2.7848232417423023e-06,
      "loss": 0.6251,
      "step": 10691
    },
    {
      "epoch": 0.443118239462887,
      "grad_norm": 0.39607587456703186,
      "learning_rate": 2.7846160222139333e-06,
      "loss": 0.7278,
      "step": 10692
    },
    {
      "epoch": 0.44315968336856065,
      "grad_norm": 0.37491244077682495,
      "learning_rate": 2.7844088026855655e-06,
      "loss": 0.6195,
      "step": 10693
    },
    {
      "epoch": 0.44320112727423433,
      "grad_norm": 0.391167014837265,
      "learning_rate": 2.7842015831571965e-06,
      "loss": 0.6362,
      "step": 10694
    },
    {
      "epoch": 0.443242571179908,
      "grad_norm": 0.39960363507270813,
      "learning_rate": 2.7839943636288287e-06,
      "loss": 0.6765,
      "step": 10695
    },
    {
      "epoch": 0.4432840150855817,
      "grad_norm": 0.42391476035118103,
      "learning_rate": 2.7837871441004605e-06,
      "loss": 0.6929,
      "step": 10696
    },
    {
      "epoch": 0.44332545899125536,
      "grad_norm": 0.3722314238548279,
      "learning_rate": 2.783579924572092e-06,
      "loss": 0.625,
      "step": 10697
    },
    {
      "epoch": 0.443366902896929,
      "grad_norm": 0.38690996170043945,
      "learning_rate": 2.7833727050437237e-06,
      "loss": 0.6162,
      "step": 10698
    },
    {
      "epoch": 0.44340834680260266,
      "grad_norm": 0.4186747670173645,
      "learning_rate": 2.783165485515355e-06,
      "loss": 0.7336,
      "step": 10699
    },
    {
      "epoch": 0.44344979070827634,
      "grad_norm": 0.3984531760215759,
      "learning_rate": 2.782958265986987e-06,
      "loss": 0.7257,
      "step": 10700
    },
    {
      "epoch": 0.44349123461395,
      "grad_norm": 0.38653764128685,
      "learning_rate": 2.7827510464586183e-06,
      "loss": 0.7065,
      "step": 10701
    },
    {
      "epoch": 0.4435326785196237,
      "grad_norm": 0.4031617343425751,
      "learning_rate": 2.78254382693025e-06,
      "loss": 0.6616,
      "step": 10702
    },
    {
      "epoch": 0.44357412242529737,
      "grad_norm": 0.3965320885181427,
      "learning_rate": 2.782336607401882e-06,
      "loss": 0.6956,
      "step": 10703
    },
    {
      "epoch": 0.44361556633097105,
      "grad_norm": 0.41684433817863464,
      "learning_rate": 2.7821293878735133e-06,
      "loss": 0.6912,
      "step": 10704
    },
    {
      "epoch": 0.4436570102366447,
      "grad_norm": 0.41227489709854126,
      "learning_rate": 2.781922168345145e-06,
      "loss": 0.6964,
      "step": 10705
    },
    {
      "epoch": 0.4436984541423184,
      "grad_norm": 0.4462659955024719,
      "learning_rate": 2.7817149488167765e-06,
      "loss": 0.6986,
      "step": 10706
    },
    {
      "epoch": 0.443739898047992,
      "grad_norm": 0.40475383400917053,
      "learning_rate": 2.7815077292884083e-06,
      "loss": 0.6833,
      "step": 10707
    },
    {
      "epoch": 0.4437813419536657,
      "grad_norm": 0.5172575116157532,
      "learning_rate": 2.7813005097600397e-06,
      "loss": 0.7681,
      "step": 10708
    },
    {
      "epoch": 0.4438227858593394,
      "grad_norm": 0.4176976680755615,
      "learning_rate": 2.781093290231672e-06,
      "loss": 0.74,
      "step": 10709
    },
    {
      "epoch": 0.44386422976501305,
      "grad_norm": 0.3735900819301605,
      "learning_rate": 2.780886070703303e-06,
      "loss": 0.616,
      "step": 10710
    },
    {
      "epoch": 0.44390567367068673,
      "grad_norm": 0.41185203194618225,
      "learning_rate": 2.780678851174935e-06,
      "loss": 0.7158,
      "step": 10711
    },
    {
      "epoch": 0.4439471175763604,
      "grad_norm": 0.40176498889923096,
      "learning_rate": 2.780471631646567e-06,
      "loss": 0.7191,
      "step": 10712
    },
    {
      "epoch": 0.4439885614820341,
      "grad_norm": 0.38810068368911743,
      "learning_rate": 2.7802644121181983e-06,
      "loss": 0.7056,
      "step": 10713
    },
    {
      "epoch": 0.44403000538770776,
      "grad_norm": 0.41772010922431946,
      "learning_rate": 2.78005719258983e-06,
      "loss": 0.7268,
      "step": 10714
    },
    {
      "epoch": 0.4440714492933814,
      "grad_norm": 0.397176593542099,
      "learning_rate": 2.7798499730614615e-06,
      "loss": 0.6462,
      "step": 10715
    },
    {
      "epoch": 0.44411289319905506,
      "grad_norm": 0.41932404041290283,
      "learning_rate": 2.7796427535330933e-06,
      "loss": 0.6812,
      "step": 10716
    },
    {
      "epoch": 0.44415433710472874,
      "grad_norm": 0.4088246822357178,
      "learning_rate": 2.7794355340047247e-06,
      "loss": 0.7056,
      "step": 10717
    },
    {
      "epoch": 0.4441957810104024,
      "grad_norm": 0.42907828092575073,
      "learning_rate": 2.7792283144763565e-06,
      "loss": 0.7156,
      "step": 10718
    },
    {
      "epoch": 0.4442372249160761,
      "grad_norm": 0.42180898785591125,
      "learning_rate": 2.779021094947988e-06,
      "loss": 0.7346,
      "step": 10719
    },
    {
      "epoch": 0.44427866882174977,
      "grad_norm": 0.4063490927219391,
      "learning_rate": 2.7788138754196197e-06,
      "loss": 0.7537,
      "step": 10720
    },
    {
      "epoch": 0.44432011272742344,
      "grad_norm": 0.4329044222831726,
      "learning_rate": 2.7786066558912515e-06,
      "loss": 0.7051,
      "step": 10721
    },
    {
      "epoch": 0.4443615566330971,
      "grad_norm": 0.3947679102420807,
      "learning_rate": 2.778399436362883e-06,
      "loss": 0.7032,
      "step": 10722
    },
    {
      "epoch": 0.4444030005387708,
      "grad_norm": 0.43656572699546814,
      "learning_rate": 2.7781922168345147e-06,
      "loss": 0.6718,
      "step": 10723
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.45558103919029236,
      "learning_rate": 2.777984997306146e-06,
      "loss": 0.6835,
      "step": 10724
    },
    {
      "epoch": 0.4444858883501181,
      "grad_norm": 0.40221020579338074,
      "learning_rate": 2.7777777777777783e-06,
      "loss": 0.7114,
      "step": 10725
    },
    {
      "epoch": 0.4445273322557918,
      "grad_norm": 0.4129464030265808,
      "learning_rate": 2.7775705582494093e-06,
      "loss": 0.6801,
      "step": 10726
    },
    {
      "epoch": 0.44456877616146545,
      "grad_norm": 0.39348557591438293,
      "learning_rate": 2.7773633387210415e-06,
      "loss": 0.6881,
      "step": 10727
    },
    {
      "epoch": 0.44461022006713913,
      "grad_norm": 0.3803649842739105,
      "learning_rate": 2.7771561191926725e-06,
      "loss": 0.6702,
      "step": 10728
    },
    {
      "epoch": 0.4446516639728128,
      "grad_norm": 0.43088364601135254,
      "learning_rate": 2.7769488996643047e-06,
      "loss": 0.7524,
      "step": 10729
    },
    {
      "epoch": 0.4446931078784865,
      "grad_norm": 0.39187121391296387,
      "learning_rate": 2.7767416801359365e-06,
      "loss": 0.6829,
      "step": 10730
    },
    {
      "epoch": 0.44473455178416016,
      "grad_norm": 0.4070812463760376,
      "learning_rate": 2.776534460607568e-06,
      "loss": 0.7019,
      "step": 10731
    },
    {
      "epoch": 0.44477599568983384,
      "grad_norm": 0.3851306438446045,
      "learning_rate": 2.7763272410791997e-06,
      "loss": 0.6519,
      "step": 10732
    },
    {
      "epoch": 0.44481743959550746,
      "grad_norm": 0.4127126932144165,
      "learning_rate": 2.776120021550831e-06,
      "loss": 0.7158,
      "step": 10733
    },
    {
      "epoch": 0.44485888350118113,
      "grad_norm": 0.4145079255104065,
      "learning_rate": 2.775912802022463e-06,
      "loss": 0.6888,
      "step": 10734
    },
    {
      "epoch": 0.4449003274068548,
      "grad_norm": 0.383944034576416,
      "learning_rate": 2.7757055824940943e-06,
      "loss": 0.6554,
      "step": 10735
    },
    {
      "epoch": 0.4449417713125285,
      "grad_norm": 0.42479604482650757,
      "learning_rate": 2.775498362965726e-06,
      "loss": 0.7365,
      "step": 10736
    },
    {
      "epoch": 0.44498321521820217,
      "grad_norm": 0.4285183548927307,
      "learning_rate": 2.7752911434373575e-06,
      "loss": 0.7349,
      "step": 10737
    },
    {
      "epoch": 0.44502465912387584,
      "grad_norm": 0.4095458686351776,
      "learning_rate": 2.7750839239089893e-06,
      "loss": 0.7029,
      "step": 10738
    },
    {
      "epoch": 0.4450661030295495,
      "grad_norm": 0.38396984338760376,
      "learning_rate": 2.774876704380621e-06,
      "loss": 0.6465,
      "step": 10739
    },
    {
      "epoch": 0.4451075469352232,
      "grad_norm": 0.42741823196411133,
      "learning_rate": 2.7746694848522525e-06,
      "loss": 0.7263,
      "step": 10740
    },
    {
      "epoch": 0.4451489908408968,
      "grad_norm": 0.4390709102153778,
      "learning_rate": 2.7744622653238843e-06,
      "loss": 0.7129,
      "step": 10741
    },
    {
      "epoch": 0.4451904347465705,
      "grad_norm": 0.43229514360427856,
      "learning_rate": 2.7742550457955157e-06,
      "loss": 0.6899,
      "step": 10742
    },
    {
      "epoch": 0.4452318786522442,
      "grad_norm": 0.38738396763801575,
      "learning_rate": 2.774047826267148e-06,
      "loss": 0.6582,
      "step": 10743
    },
    {
      "epoch": 0.44527332255791785,
      "grad_norm": 0.35839617252349854,
      "learning_rate": 2.773840606738779e-06,
      "loss": 0.6165,
      "step": 10744
    },
    {
      "epoch": 0.4453147664635915,
      "grad_norm": 0.4070381820201874,
      "learning_rate": 2.773633387210411e-06,
      "loss": 0.6869,
      "step": 10745
    },
    {
      "epoch": 0.4453562103692652,
      "grad_norm": 0.4540160298347473,
      "learning_rate": 2.7734261676820425e-06,
      "loss": 0.6921,
      "step": 10746
    },
    {
      "epoch": 0.4453976542749389,
      "grad_norm": 0.4162324368953705,
      "learning_rate": 2.7732189481536743e-06,
      "loss": 0.7141,
      "step": 10747
    },
    {
      "epoch": 0.44543909818061256,
      "grad_norm": 0.3760441541671753,
      "learning_rate": 2.773011728625306e-06,
      "loss": 0.6422,
      "step": 10748
    },
    {
      "epoch": 0.44548054208628624,
      "grad_norm": 0.4429740309715271,
      "learning_rate": 2.7728045090969375e-06,
      "loss": 0.6882,
      "step": 10749
    },
    {
      "epoch": 0.44552198599195986,
      "grad_norm": 0.4257534146308899,
      "learning_rate": 2.7725972895685693e-06,
      "loss": 0.6846,
      "step": 10750
    },
    {
      "epoch": 0.44556342989763353,
      "grad_norm": 0.44154950976371765,
      "learning_rate": 2.7723900700402007e-06,
      "loss": 0.719,
      "step": 10751
    },
    {
      "epoch": 0.4456048738033072,
      "grad_norm": 0.4083128571510315,
      "learning_rate": 2.7721828505118325e-06,
      "loss": 0.6667,
      "step": 10752
    },
    {
      "epoch": 0.4456463177089809,
      "grad_norm": 0.4325467050075531,
      "learning_rate": 2.771975630983464e-06,
      "loss": 0.7561,
      "step": 10753
    },
    {
      "epoch": 0.44568776161465457,
      "grad_norm": 0.4141205847263336,
      "learning_rate": 2.7717684114550957e-06,
      "loss": 0.675,
      "step": 10754
    },
    {
      "epoch": 0.44572920552032824,
      "grad_norm": 0.515047013759613,
      "learning_rate": 2.7715611919267275e-06,
      "loss": 0.7239,
      "step": 10755
    },
    {
      "epoch": 0.4457706494260019,
      "grad_norm": 0.39870646595954895,
      "learning_rate": 2.771353972398359e-06,
      "loss": 0.675,
      "step": 10756
    },
    {
      "epoch": 0.4458120933316756,
      "grad_norm": 0.44386667013168335,
      "learning_rate": 2.7711467528699907e-06,
      "loss": 0.6868,
      "step": 10757
    },
    {
      "epoch": 0.4458535372373493,
      "grad_norm": 0.3994561731815338,
      "learning_rate": 2.770939533341622e-06,
      "loss": 0.6216,
      "step": 10758
    },
    {
      "epoch": 0.4458949811430229,
      "grad_norm": 0.3937024772167206,
      "learning_rate": 2.7707323138132543e-06,
      "loss": 0.637,
      "step": 10759
    },
    {
      "epoch": 0.44593642504869657,
      "grad_norm": 0.4203580915927887,
      "learning_rate": 2.7705250942848853e-06,
      "loss": 0.6421,
      "step": 10760
    },
    {
      "epoch": 0.44597786895437025,
      "grad_norm": 0.39245256781578064,
      "learning_rate": 2.7703178747565175e-06,
      "loss": 0.6584,
      "step": 10761
    },
    {
      "epoch": 0.4460193128600439,
      "grad_norm": 0.41881120204925537,
      "learning_rate": 2.7701106552281485e-06,
      "loss": 0.6958,
      "step": 10762
    },
    {
      "epoch": 0.4460607567657176,
      "grad_norm": 0.40717312693595886,
      "learning_rate": 2.7699034356997807e-06,
      "loss": 0.6978,
      "step": 10763
    },
    {
      "epoch": 0.4461022006713913,
      "grad_norm": 0.4443020522594452,
      "learning_rate": 2.7696962161714125e-06,
      "loss": 0.7031,
      "step": 10764
    },
    {
      "epoch": 0.44614364457706496,
      "grad_norm": 0.43122488260269165,
      "learning_rate": 2.769488996643044e-06,
      "loss": 0.7351,
      "step": 10765
    },
    {
      "epoch": 0.44618508848273863,
      "grad_norm": 0.5521938800811768,
      "learning_rate": 2.7692817771146757e-06,
      "loss": 0.7028,
      "step": 10766
    },
    {
      "epoch": 0.4462265323884123,
      "grad_norm": 0.4305494427680969,
      "learning_rate": 2.769074557586307e-06,
      "loss": 0.6821,
      "step": 10767
    },
    {
      "epoch": 0.44626797629408593,
      "grad_norm": 0.42370104789733887,
      "learning_rate": 2.768867338057939e-06,
      "loss": 0.6643,
      "step": 10768
    },
    {
      "epoch": 0.4463094201997596,
      "grad_norm": 0.4145101308822632,
      "learning_rate": 2.7686601185295703e-06,
      "loss": 0.6875,
      "step": 10769
    },
    {
      "epoch": 0.4463508641054333,
      "grad_norm": 0.4131934642791748,
      "learning_rate": 2.768452899001202e-06,
      "loss": 0.7046,
      "step": 10770
    },
    {
      "epoch": 0.44639230801110696,
      "grad_norm": 0.4026491045951843,
      "learning_rate": 2.7682456794728335e-06,
      "loss": 0.6897,
      "step": 10771
    },
    {
      "epoch": 0.44643375191678064,
      "grad_norm": 0.49821916222572327,
      "learning_rate": 2.7680384599444653e-06,
      "loss": 0.7076,
      "step": 10772
    },
    {
      "epoch": 0.4464751958224543,
      "grad_norm": 0.41298797726631165,
      "learning_rate": 2.767831240416097e-06,
      "loss": 0.6926,
      "step": 10773
    },
    {
      "epoch": 0.446516639728128,
      "grad_norm": 0.4178711175918579,
      "learning_rate": 2.7676240208877285e-06,
      "loss": 0.6548,
      "step": 10774
    },
    {
      "epoch": 0.4465580836338017,
      "grad_norm": 0.4024145305156708,
      "learning_rate": 2.7674168013593603e-06,
      "loss": 0.7007,
      "step": 10775
    },
    {
      "epoch": 0.4465995275394753,
      "grad_norm": 0.45866072177886963,
      "learning_rate": 2.7672095818309917e-06,
      "loss": 0.7324,
      "step": 10776
    },
    {
      "epoch": 0.44664097144514897,
      "grad_norm": 0.41210514307022095,
      "learning_rate": 2.767002362302624e-06,
      "loss": 0.6987,
      "step": 10777
    },
    {
      "epoch": 0.44668241535082265,
      "grad_norm": 0.4201638698577881,
      "learning_rate": 2.766795142774255e-06,
      "loss": 0.6897,
      "step": 10778
    },
    {
      "epoch": 0.4467238592564963,
      "grad_norm": 0.4367576241493225,
      "learning_rate": 2.766587923245887e-06,
      "loss": 0.6899,
      "step": 10779
    },
    {
      "epoch": 0.44676530316217,
      "grad_norm": 0.3872509300708771,
      "learning_rate": 2.7663807037175185e-06,
      "loss": 0.6868,
      "step": 10780
    },
    {
      "epoch": 0.4468067470678437,
      "grad_norm": 0.41382497549057007,
      "learning_rate": 2.7661734841891503e-06,
      "loss": 0.7222,
      "step": 10781
    },
    {
      "epoch": 0.44684819097351736,
      "grad_norm": 0.4649643301963806,
      "learning_rate": 2.765966264660782e-06,
      "loss": 0.7117,
      "step": 10782
    },
    {
      "epoch": 0.44688963487919103,
      "grad_norm": 0.4160113036632538,
      "learning_rate": 2.7657590451324135e-06,
      "loss": 0.7095,
      "step": 10783
    },
    {
      "epoch": 0.4469310787848647,
      "grad_norm": 0.4087352454662323,
      "learning_rate": 2.7655518256040453e-06,
      "loss": 0.689,
      "step": 10784
    },
    {
      "epoch": 0.44697252269053833,
      "grad_norm": 0.41340169310569763,
      "learning_rate": 2.7653446060756767e-06,
      "loss": 0.6531,
      "step": 10785
    },
    {
      "epoch": 0.447013966596212,
      "grad_norm": 0.412334144115448,
      "learning_rate": 2.7651373865473085e-06,
      "loss": 0.6602,
      "step": 10786
    },
    {
      "epoch": 0.4470554105018857,
      "grad_norm": 0.4670717418193817,
      "learning_rate": 2.76493016701894e-06,
      "loss": 0.7279,
      "step": 10787
    },
    {
      "epoch": 0.44709685440755936,
      "grad_norm": 0.36499109864234924,
      "learning_rate": 2.7647229474905717e-06,
      "loss": 0.6853,
      "step": 10788
    },
    {
      "epoch": 0.44713829831323304,
      "grad_norm": 0.3761703670024872,
      "learning_rate": 2.764515727962203e-06,
      "loss": 0.6628,
      "step": 10789
    },
    {
      "epoch": 0.4471797422189067,
      "grad_norm": 0.42338645458221436,
      "learning_rate": 2.764308508433835e-06,
      "loss": 0.7263,
      "step": 10790
    },
    {
      "epoch": 0.4472211861245804,
      "grad_norm": 0.4451175630092621,
      "learning_rate": 2.7641012889054667e-06,
      "loss": 0.6879,
      "step": 10791
    },
    {
      "epoch": 0.44726263003025407,
      "grad_norm": 0.4271060526371002,
      "learning_rate": 2.763894069377098e-06,
      "loss": 0.7205,
      "step": 10792
    },
    {
      "epoch": 0.44730407393592775,
      "grad_norm": 0.42221876978874207,
      "learning_rate": 2.7636868498487303e-06,
      "loss": 0.6808,
      "step": 10793
    },
    {
      "epoch": 0.44734551784160137,
      "grad_norm": 0.3987949788570404,
      "learning_rate": 2.7634796303203613e-06,
      "loss": 0.6838,
      "step": 10794
    },
    {
      "epoch": 0.44738696174727505,
      "grad_norm": 0.4065544605255127,
      "learning_rate": 2.7632724107919935e-06,
      "loss": 0.6777,
      "step": 10795
    },
    {
      "epoch": 0.4474284056529487,
      "grad_norm": 0.41755664348602295,
      "learning_rate": 2.7630651912636245e-06,
      "loss": 0.6404,
      "step": 10796
    },
    {
      "epoch": 0.4474698495586224,
      "grad_norm": 0.40265128016471863,
      "learning_rate": 2.7628579717352567e-06,
      "loss": 0.6895,
      "step": 10797
    },
    {
      "epoch": 0.4475112934642961,
      "grad_norm": 0.43838223814964294,
      "learning_rate": 2.762650752206888e-06,
      "loss": 0.6682,
      "step": 10798
    },
    {
      "epoch": 0.44755273736996976,
      "grad_norm": 0.44894281029701233,
      "learning_rate": 2.76244353267852e-06,
      "loss": 0.7244,
      "step": 10799
    },
    {
      "epoch": 0.44759418127564343,
      "grad_norm": 0.5027337670326233,
      "learning_rate": 2.7622363131501517e-06,
      "loss": 0.7837,
      "step": 10800
    },
    {
      "epoch": 0.4476356251813171,
      "grad_norm": 0.4085533320903778,
      "learning_rate": 2.762029093621783e-06,
      "loss": 0.6663,
      "step": 10801
    },
    {
      "epoch": 0.44767706908699073,
      "grad_norm": 0.40160036087036133,
      "learning_rate": 2.761821874093415e-06,
      "loss": 0.6848,
      "step": 10802
    },
    {
      "epoch": 0.4477185129926644,
      "grad_norm": 0.43014684319496155,
      "learning_rate": 2.7616146545650463e-06,
      "loss": 0.7092,
      "step": 10803
    },
    {
      "epoch": 0.4477599568983381,
      "grad_norm": 0.38239920139312744,
      "learning_rate": 2.761407435036678e-06,
      "loss": 0.677,
      "step": 10804
    },
    {
      "epoch": 0.44780140080401176,
      "grad_norm": 0.4521341919898987,
      "learning_rate": 2.7612002155083095e-06,
      "loss": 0.7076,
      "step": 10805
    },
    {
      "epoch": 0.44784284470968544,
      "grad_norm": 0.42911645770072937,
      "learning_rate": 2.7609929959799413e-06,
      "loss": 0.6722,
      "step": 10806
    },
    {
      "epoch": 0.4478842886153591,
      "grad_norm": 0.3787784278392792,
      "learning_rate": 2.7607857764515727e-06,
      "loss": 0.661,
      "step": 10807
    },
    {
      "epoch": 0.4479257325210328,
      "grad_norm": 0.42653194069862366,
      "learning_rate": 2.7605785569232045e-06,
      "loss": 0.6699,
      "step": 10808
    },
    {
      "epoch": 0.44796717642670647,
      "grad_norm": 0.42119038105010986,
      "learning_rate": 2.7603713373948363e-06,
      "loss": 0.7219,
      "step": 10809
    },
    {
      "epoch": 0.44800862033238015,
      "grad_norm": 0.39014604687690735,
      "learning_rate": 2.7601641178664677e-06,
      "loss": 0.6979,
      "step": 10810
    },
    {
      "epoch": 0.44805006423805377,
      "grad_norm": 0.40384072065353394,
      "learning_rate": 2.7599568983381e-06,
      "loss": 0.677,
      "step": 10811
    },
    {
      "epoch": 0.44809150814372745,
      "grad_norm": 0.40486863255500793,
      "learning_rate": 2.759749678809731e-06,
      "loss": 0.6704,
      "step": 10812
    },
    {
      "epoch": 0.4481329520494011,
      "grad_norm": 0.4096280038356781,
      "learning_rate": 2.759542459281363e-06,
      "loss": 0.7083,
      "step": 10813
    },
    {
      "epoch": 0.4481743959550748,
      "grad_norm": 0.37814274430274963,
      "learning_rate": 2.7593352397529945e-06,
      "loss": 0.6326,
      "step": 10814
    },
    {
      "epoch": 0.4482158398607485,
      "grad_norm": 0.40980005264282227,
      "learning_rate": 2.7591280202246263e-06,
      "loss": 0.7097,
      "step": 10815
    },
    {
      "epoch": 0.44825728376642215,
      "grad_norm": 0.3945491909980774,
      "learning_rate": 2.758920800696258e-06,
      "loss": 0.6631,
      "step": 10816
    },
    {
      "epoch": 0.44829872767209583,
      "grad_norm": 0.439034640789032,
      "learning_rate": 2.7587135811678895e-06,
      "loss": 0.743,
      "step": 10817
    },
    {
      "epoch": 0.4483401715777695,
      "grad_norm": 0.4239235520362854,
      "learning_rate": 2.7585063616395213e-06,
      "loss": 0.6991,
      "step": 10818
    },
    {
      "epoch": 0.4483816154834432,
      "grad_norm": 0.41628003120422363,
      "learning_rate": 2.7582991421111527e-06,
      "loss": 0.7466,
      "step": 10819
    },
    {
      "epoch": 0.4484230593891168,
      "grad_norm": 0.40465855598449707,
      "learning_rate": 2.7580919225827845e-06,
      "loss": 0.6418,
      "step": 10820
    },
    {
      "epoch": 0.4484645032947905,
      "grad_norm": 0.398535817861557,
      "learning_rate": 2.757884703054416e-06,
      "loss": 0.6512,
      "step": 10821
    },
    {
      "epoch": 0.44850594720046416,
      "grad_norm": 0.4544258415699005,
      "learning_rate": 2.7576774835260477e-06,
      "loss": 0.6915,
      "step": 10822
    },
    {
      "epoch": 0.44854739110613784,
      "grad_norm": 0.42401474714279175,
      "learning_rate": 2.757470263997679e-06,
      "loss": 0.6809,
      "step": 10823
    },
    {
      "epoch": 0.4485888350118115,
      "grad_norm": 0.4807281792163849,
      "learning_rate": 2.757263044469311e-06,
      "loss": 0.7727,
      "step": 10824
    },
    {
      "epoch": 0.4486302789174852,
      "grad_norm": 0.4088771343231201,
      "learning_rate": 2.7570558249409427e-06,
      "loss": 0.7041,
      "step": 10825
    },
    {
      "epoch": 0.44867172282315887,
      "grad_norm": 0.4210512340068817,
      "learning_rate": 2.756848605412574e-06,
      "loss": 0.6868,
      "step": 10826
    },
    {
      "epoch": 0.44871316672883255,
      "grad_norm": 0.4039902091026306,
      "learning_rate": 2.7566413858842064e-06,
      "loss": 0.7212,
      "step": 10827
    },
    {
      "epoch": 0.4487546106345062,
      "grad_norm": 0.391928493976593,
      "learning_rate": 2.7564341663558373e-06,
      "loss": 0.7405,
      "step": 10828
    },
    {
      "epoch": 0.44879605454017985,
      "grad_norm": 0.4118398129940033,
      "learning_rate": 2.7562269468274695e-06,
      "loss": 0.7057,
      "step": 10829
    },
    {
      "epoch": 0.4488374984458535,
      "grad_norm": 0.39546898007392883,
      "learning_rate": 2.756019727299101e-06,
      "loss": 0.6532,
      "step": 10830
    },
    {
      "epoch": 0.4488789423515272,
      "grad_norm": 0.37725499272346497,
      "learning_rate": 2.7558125077707327e-06,
      "loss": 0.7217,
      "step": 10831
    },
    {
      "epoch": 0.4489203862572009,
      "grad_norm": 0.3909328579902649,
      "learning_rate": 2.755605288242364e-06,
      "loss": 0.686,
      "step": 10832
    },
    {
      "epoch": 0.44896183016287455,
      "grad_norm": 0.4160630404949188,
      "learning_rate": 2.755398068713996e-06,
      "loss": 0.6803,
      "step": 10833
    },
    {
      "epoch": 0.44900327406854823,
      "grad_norm": 0.4155634045600891,
      "learning_rate": 2.7551908491856277e-06,
      "loss": 0.7227,
      "step": 10834
    },
    {
      "epoch": 0.4490447179742219,
      "grad_norm": 0.4120108485221863,
      "learning_rate": 2.754983629657259e-06,
      "loss": 0.667,
      "step": 10835
    },
    {
      "epoch": 0.4490861618798956,
      "grad_norm": 0.40564781427383423,
      "learning_rate": 2.754776410128891e-06,
      "loss": 0.7156,
      "step": 10836
    },
    {
      "epoch": 0.4491276057855692,
      "grad_norm": 0.42689386010169983,
      "learning_rate": 2.7545691906005223e-06,
      "loss": 0.7219,
      "step": 10837
    },
    {
      "epoch": 0.4491690496912429,
      "grad_norm": 0.4072379469871521,
      "learning_rate": 2.754361971072154e-06,
      "loss": 0.679,
      "step": 10838
    },
    {
      "epoch": 0.44921049359691656,
      "grad_norm": 0.3911210298538208,
      "learning_rate": 2.7541547515437855e-06,
      "loss": 0.6743,
      "step": 10839
    },
    {
      "epoch": 0.44925193750259024,
      "grad_norm": 0.4427466094493866,
      "learning_rate": 2.7539475320154173e-06,
      "loss": 0.7186,
      "step": 10840
    },
    {
      "epoch": 0.4492933814082639,
      "grad_norm": 0.4513285160064697,
      "learning_rate": 2.7537403124870487e-06,
      "loss": 0.7183,
      "step": 10841
    },
    {
      "epoch": 0.4493348253139376,
      "grad_norm": 0.4227806031703949,
      "learning_rate": 2.7535330929586805e-06,
      "loss": 0.6394,
      "step": 10842
    },
    {
      "epoch": 0.44937626921961127,
      "grad_norm": 0.4032141864299774,
      "learning_rate": 2.7533258734303123e-06,
      "loss": 0.6542,
      "step": 10843
    },
    {
      "epoch": 0.44941771312528495,
      "grad_norm": 0.40027546882629395,
      "learning_rate": 2.7531186539019437e-06,
      "loss": 0.7113,
      "step": 10844
    },
    {
      "epoch": 0.4494591570309586,
      "grad_norm": 0.39920344948768616,
      "learning_rate": 2.752911434373576e-06,
      "loss": 0.6345,
      "step": 10845
    },
    {
      "epoch": 0.44950060093663224,
      "grad_norm": 0.38025155663490295,
      "learning_rate": 2.752704214845207e-06,
      "loss": 0.7003,
      "step": 10846
    },
    {
      "epoch": 0.4495420448423059,
      "grad_norm": 0.4052783250808716,
      "learning_rate": 2.752496995316839e-06,
      "loss": 0.6857,
      "step": 10847
    },
    {
      "epoch": 0.4495834887479796,
      "grad_norm": 0.4205634891986847,
      "learning_rate": 2.7522897757884705e-06,
      "loss": 0.7039,
      "step": 10848
    },
    {
      "epoch": 0.4496249326536533,
      "grad_norm": 0.42052412033081055,
      "learning_rate": 2.7520825562601023e-06,
      "loss": 0.7014,
      "step": 10849
    },
    {
      "epoch": 0.44966637655932695,
      "grad_norm": 0.39881160855293274,
      "learning_rate": 2.7518753367317337e-06,
      "loss": 0.6975,
      "step": 10850
    },
    {
      "epoch": 0.44970782046500063,
      "grad_norm": 0.4214608073234558,
      "learning_rate": 2.7516681172033655e-06,
      "loss": 0.6985,
      "step": 10851
    },
    {
      "epoch": 0.4497492643706743,
      "grad_norm": 0.4114598333835602,
      "learning_rate": 2.7514608976749973e-06,
      "loss": 0.645,
      "step": 10852
    },
    {
      "epoch": 0.449790708276348,
      "grad_norm": 0.4306933283805847,
      "learning_rate": 2.7512536781466287e-06,
      "loss": 0.6613,
      "step": 10853
    },
    {
      "epoch": 0.44983215218202166,
      "grad_norm": 0.3998236358165741,
      "learning_rate": 2.7510464586182605e-06,
      "loss": 0.6853,
      "step": 10854
    },
    {
      "epoch": 0.4498735960876953,
      "grad_norm": 0.42441365122795105,
      "learning_rate": 2.750839239089892e-06,
      "loss": 0.6489,
      "step": 10855
    },
    {
      "epoch": 0.44991503999336896,
      "grad_norm": 0.42009133100509644,
      "learning_rate": 2.7506320195615237e-06,
      "loss": 0.7478,
      "step": 10856
    },
    {
      "epoch": 0.44995648389904264,
      "grad_norm": 0.4120667278766632,
      "learning_rate": 2.750424800033155e-06,
      "loss": 0.6775,
      "step": 10857
    },
    {
      "epoch": 0.4499979278047163,
      "grad_norm": 0.37560510635375977,
      "learning_rate": 2.750217580504787e-06,
      "loss": 0.6279,
      "step": 10858
    },
    {
      "epoch": 0.45003937171039,
      "grad_norm": 0.4543132185935974,
      "learning_rate": 2.7500103609764183e-06,
      "loss": 0.7448,
      "step": 10859
    },
    {
      "epoch": 0.45008081561606367,
      "grad_norm": 0.40168115496635437,
      "learning_rate": 2.74980314144805e-06,
      "loss": 0.6597,
      "step": 10860
    },
    {
      "epoch": 0.45012225952173734,
      "grad_norm": 0.3960631191730499,
      "learning_rate": 2.7495959219196824e-06,
      "loss": 0.667,
      "step": 10861
    },
    {
      "epoch": 0.450163703427411,
      "grad_norm": 0.38308054208755493,
      "learning_rate": 2.7493887023913133e-06,
      "loss": 0.6339,
      "step": 10862
    },
    {
      "epoch": 0.45020514733308464,
      "grad_norm": 0.44173336029052734,
      "learning_rate": 2.7491814828629456e-06,
      "loss": 0.6772,
      "step": 10863
    },
    {
      "epoch": 0.4502465912387583,
      "grad_norm": 0.42553815245628357,
      "learning_rate": 2.748974263334577e-06,
      "loss": 0.7214,
      "step": 10864
    },
    {
      "epoch": 0.450288035144432,
      "grad_norm": 0.4157269299030304,
      "learning_rate": 2.7487670438062087e-06,
      "loss": 0.7295,
      "step": 10865
    },
    {
      "epoch": 0.4503294790501057,
      "grad_norm": 0.40996667742729187,
      "learning_rate": 2.74855982427784e-06,
      "loss": 0.7051,
      "step": 10866
    },
    {
      "epoch": 0.45037092295577935,
      "grad_norm": 0.5329059958457947,
      "learning_rate": 2.748352604749472e-06,
      "loss": 0.7686,
      "step": 10867
    },
    {
      "epoch": 0.45041236686145303,
      "grad_norm": 0.39402318000793457,
      "learning_rate": 2.7481453852211033e-06,
      "loss": 0.7112,
      "step": 10868
    },
    {
      "epoch": 0.4504538107671267,
      "grad_norm": 0.393494188785553,
      "learning_rate": 2.747938165692735e-06,
      "loss": 0.6924,
      "step": 10869
    },
    {
      "epoch": 0.4504952546728004,
      "grad_norm": 0.39389529824256897,
      "learning_rate": 2.747730946164367e-06,
      "loss": 0.6735,
      "step": 10870
    },
    {
      "epoch": 0.45053669857847406,
      "grad_norm": 0.4094820022583008,
      "learning_rate": 2.7475237266359983e-06,
      "loss": 0.672,
      "step": 10871
    },
    {
      "epoch": 0.4505781424841477,
      "grad_norm": 0.4184476435184479,
      "learning_rate": 2.74731650710763e-06,
      "loss": 0.7101,
      "step": 10872
    },
    {
      "epoch": 0.45061958638982136,
      "grad_norm": 0.36338573694229126,
      "learning_rate": 2.7471092875792615e-06,
      "loss": 0.6755,
      "step": 10873
    },
    {
      "epoch": 0.45066103029549504,
      "grad_norm": 0.4082651436328888,
      "learning_rate": 2.7469020680508933e-06,
      "loss": 0.686,
      "step": 10874
    },
    {
      "epoch": 0.4507024742011687,
      "grad_norm": 0.4405486583709717,
      "learning_rate": 2.7466948485225247e-06,
      "loss": 0.6973,
      "step": 10875
    },
    {
      "epoch": 0.4507439181068424,
      "grad_norm": 0.4147479236125946,
      "learning_rate": 2.7464876289941565e-06,
      "loss": 0.7164,
      "step": 10876
    },
    {
      "epoch": 0.45078536201251607,
      "grad_norm": 0.4813583791255951,
      "learning_rate": 2.7462804094657888e-06,
      "loss": 0.7621,
      "step": 10877
    },
    {
      "epoch": 0.45082680591818974,
      "grad_norm": 0.3963519036769867,
      "learning_rate": 2.7460731899374197e-06,
      "loss": 0.6882,
      "step": 10878
    },
    {
      "epoch": 0.4508682498238634,
      "grad_norm": 0.41709020733833313,
      "learning_rate": 2.745865970409052e-06,
      "loss": 0.7108,
      "step": 10879
    },
    {
      "epoch": 0.4509096937295371,
      "grad_norm": 0.41503480076789856,
      "learning_rate": 2.745658750880683e-06,
      "loss": 0.7058,
      "step": 10880
    },
    {
      "epoch": 0.4509511376352107,
      "grad_norm": 0.4125930964946747,
      "learning_rate": 2.745451531352315e-06,
      "loss": 0.7446,
      "step": 10881
    },
    {
      "epoch": 0.4509925815408844,
      "grad_norm": 0.4199207127094269,
      "learning_rate": 2.7452443118239465e-06,
      "loss": 0.6688,
      "step": 10882
    },
    {
      "epoch": 0.4510340254465581,
      "grad_norm": 0.39728984236717224,
      "learning_rate": 2.7450370922955783e-06,
      "loss": 0.6505,
      "step": 10883
    },
    {
      "epoch": 0.45107546935223175,
      "grad_norm": 0.3735872209072113,
      "learning_rate": 2.7448298727672097e-06,
      "loss": 0.6976,
      "step": 10884
    },
    {
      "epoch": 0.4511169132579054,
      "grad_norm": 0.41844668984413147,
      "learning_rate": 2.7446226532388415e-06,
      "loss": 0.7563,
      "step": 10885
    },
    {
      "epoch": 0.4511583571635791,
      "grad_norm": 0.4364495277404785,
      "learning_rate": 2.7444154337104734e-06,
      "loss": 0.7158,
      "step": 10886
    },
    {
      "epoch": 0.4511998010692528,
      "grad_norm": 0.40021952986717224,
      "learning_rate": 2.7442082141821047e-06,
      "loss": 0.6798,
      "step": 10887
    },
    {
      "epoch": 0.45124124497492646,
      "grad_norm": 0.4262683689594269,
      "learning_rate": 2.7440009946537365e-06,
      "loss": 0.7659,
      "step": 10888
    },
    {
      "epoch": 0.4512826888806001,
      "grad_norm": 0.38293272256851196,
      "learning_rate": 2.743793775125368e-06,
      "loss": 0.6045,
      "step": 10889
    },
    {
      "epoch": 0.45132413278627376,
      "grad_norm": 0.39113572239875793,
      "learning_rate": 2.7435865555969997e-06,
      "loss": 0.6401,
      "step": 10890
    },
    {
      "epoch": 0.45136557669194743,
      "grad_norm": 0.41732561588287354,
      "learning_rate": 2.743379336068631e-06,
      "loss": 0.6855,
      "step": 10891
    },
    {
      "epoch": 0.4514070205976211,
      "grad_norm": 0.4109300374984741,
      "learning_rate": 2.743172116540263e-06,
      "loss": 0.7167,
      "step": 10892
    },
    {
      "epoch": 0.4514484645032948,
      "grad_norm": 0.47449973225593567,
      "learning_rate": 2.7429648970118943e-06,
      "loss": 0.7256,
      "step": 10893
    },
    {
      "epoch": 0.45148990840896847,
      "grad_norm": 0.41985392570495605,
      "learning_rate": 2.742757677483526e-06,
      "loss": 0.7074,
      "step": 10894
    },
    {
      "epoch": 0.45153135231464214,
      "grad_norm": 0.40595000982284546,
      "learning_rate": 2.7425504579551584e-06,
      "loss": 0.7311,
      "step": 10895
    },
    {
      "epoch": 0.4515727962203158,
      "grad_norm": 0.40797045826911926,
      "learning_rate": 2.7423432384267893e-06,
      "loss": 0.7209,
      "step": 10896
    },
    {
      "epoch": 0.4516142401259895,
      "grad_norm": 0.40827038884162903,
      "learning_rate": 2.7421360188984216e-06,
      "loss": 0.7429,
      "step": 10897
    },
    {
      "epoch": 0.4516556840316631,
      "grad_norm": 0.42651981115341187,
      "learning_rate": 2.741928799370053e-06,
      "loss": 0.6561,
      "step": 10898
    },
    {
      "epoch": 0.4516971279373368,
      "grad_norm": 0.410020112991333,
      "learning_rate": 2.7417215798416848e-06,
      "loss": 0.6678,
      "step": 10899
    },
    {
      "epoch": 0.4517385718430105,
      "grad_norm": 0.4196649491786957,
      "learning_rate": 2.741514360313316e-06,
      "loss": 0.7085,
      "step": 10900
    },
    {
      "epoch": 0.45178001574868415,
      "grad_norm": 0.38706499338150024,
      "learning_rate": 2.741307140784948e-06,
      "loss": 0.646,
      "step": 10901
    },
    {
      "epoch": 0.4518214596543578,
      "grad_norm": 0.44052794575691223,
      "learning_rate": 2.7410999212565793e-06,
      "loss": 0.7144,
      "step": 10902
    },
    {
      "epoch": 0.4518629035600315,
      "grad_norm": 0.37969788908958435,
      "learning_rate": 2.740892701728211e-06,
      "loss": 0.6819,
      "step": 10903
    },
    {
      "epoch": 0.4519043474657052,
      "grad_norm": 0.47429054975509644,
      "learning_rate": 2.740685482199843e-06,
      "loss": 0.7134,
      "step": 10904
    },
    {
      "epoch": 0.45194579137137886,
      "grad_norm": 0.3993639051914215,
      "learning_rate": 2.7404782626714743e-06,
      "loss": 0.6755,
      "step": 10905
    },
    {
      "epoch": 0.45198723527705253,
      "grad_norm": 0.41313016414642334,
      "learning_rate": 2.740271043143106e-06,
      "loss": 0.6721,
      "step": 10906
    },
    {
      "epoch": 0.45202867918272616,
      "grad_norm": 0.4232945740222931,
      "learning_rate": 2.7400638236147375e-06,
      "loss": 0.7156,
      "step": 10907
    },
    {
      "epoch": 0.45207012308839983,
      "grad_norm": 0.38364794850349426,
      "learning_rate": 2.7398566040863693e-06,
      "loss": 0.6497,
      "step": 10908
    },
    {
      "epoch": 0.4521115669940735,
      "grad_norm": 0.38954055309295654,
      "learning_rate": 2.7396493845580007e-06,
      "loss": 0.6604,
      "step": 10909
    },
    {
      "epoch": 0.4521530108997472,
      "grad_norm": 0.43266937136650085,
      "learning_rate": 2.7394421650296325e-06,
      "loss": 0.7166,
      "step": 10910
    },
    {
      "epoch": 0.45219445480542086,
      "grad_norm": 0.39658209681510925,
      "learning_rate": 2.739234945501264e-06,
      "loss": 0.665,
      "step": 10911
    },
    {
      "epoch": 0.45223589871109454,
      "grad_norm": 0.40220561623573303,
      "learning_rate": 2.7390277259728957e-06,
      "loss": 0.6971,
      "step": 10912
    },
    {
      "epoch": 0.4522773426167682,
      "grad_norm": 0.4212718605995178,
      "learning_rate": 2.738820506444528e-06,
      "loss": 0.7075,
      "step": 10913
    },
    {
      "epoch": 0.4523187865224419,
      "grad_norm": 0.42087072134017944,
      "learning_rate": 2.738613286916159e-06,
      "loss": 0.6399,
      "step": 10914
    },
    {
      "epoch": 0.4523602304281156,
      "grad_norm": 0.40563976764678955,
      "learning_rate": 2.738406067387791e-06,
      "loss": 0.6985,
      "step": 10915
    },
    {
      "epoch": 0.4524016743337892,
      "grad_norm": 0.411584734916687,
      "learning_rate": 2.7381988478594225e-06,
      "loss": 0.7087,
      "step": 10916
    },
    {
      "epoch": 0.45244311823946287,
      "grad_norm": 0.44316041469573975,
      "learning_rate": 2.7379916283310544e-06,
      "loss": 0.7076,
      "step": 10917
    },
    {
      "epoch": 0.45248456214513655,
      "grad_norm": 0.41748175024986267,
      "learning_rate": 2.7377844088026857e-06,
      "loss": 0.6927,
      "step": 10918
    },
    {
      "epoch": 0.4525260060508102,
      "grad_norm": 0.4268921911716461,
      "learning_rate": 2.7375771892743175e-06,
      "loss": 0.7339,
      "step": 10919
    },
    {
      "epoch": 0.4525674499564839,
      "grad_norm": 0.3742809295654297,
      "learning_rate": 2.737369969745949e-06,
      "loss": 0.6539,
      "step": 10920
    },
    {
      "epoch": 0.4526088938621576,
      "grad_norm": 0.4144238829612732,
      "learning_rate": 2.7371627502175807e-06,
      "loss": 0.7051,
      "step": 10921
    },
    {
      "epoch": 0.45265033776783126,
      "grad_norm": 0.4325546324253082,
      "learning_rate": 2.7369555306892126e-06,
      "loss": 0.7336,
      "step": 10922
    },
    {
      "epoch": 0.45269178167350493,
      "grad_norm": 0.4144049882888794,
      "learning_rate": 2.736748311160844e-06,
      "loss": 0.6678,
      "step": 10923
    },
    {
      "epoch": 0.45273322557917856,
      "grad_norm": 0.4398965537548065,
      "learning_rate": 2.7365410916324757e-06,
      "loss": 0.718,
      "step": 10924
    },
    {
      "epoch": 0.45277466948485223,
      "grad_norm": 0.381059467792511,
      "learning_rate": 2.736333872104107e-06,
      "loss": 0.6669,
      "step": 10925
    },
    {
      "epoch": 0.4528161133905259,
      "grad_norm": 0.4741170108318329,
      "learning_rate": 2.736126652575739e-06,
      "loss": 0.7087,
      "step": 10926
    },
    {
      "epoch": 0.4528575572961996,
      "grad_norm": 0.418230801820755,
      "learning_rate": 2.7359194330473703e-06,
      "loss": 0.6981,
      "step": 10927
    },
    {
      "epoch": 0.45289900120187326,
      "grad_norm": 0.4087010622024536,
      "learning_rate": 2.735712213519002e-06,
      "loss": 0.739,
      "step": 10928
    },
    {
      "epoch": 0.45294044510754694,
      "grad_norm": 0.4169796407222748,
      "learning_rate": 2.7355049939906335e-06,
      "loss": 0.6387,
      "step": 10929
    },
    {
      "epoch": 0.4529818890132206,
      "grad_norm": 0.4080181419849396,
      "learning_rate": 2.7352977744622653e-06,
      "loss": 0.6541,
      "step": 10930
    },
    {
      "epoch": 0.4530233329188943,
      "grad_norm": 0.39472997188568115,
      "learning_rate": 2.7350905549338976e-06,
      "loss": 0.658,
      "step": 10931
    },
    {
      "epoch": 0.45306477682456797,
      "grad_norm": 0.4234772026538849,
      "learning_rate": 2.734883335405529e-06,
      "loss": 0.7158,
      "step": 10932
    },
    {
      "epoch": 0.4531062207302416,
      "grad_norm": 0.47047293186187744,
      "learning_rate": 2.7346761158771608e-06,
      "loss": 0.7998,
      "step": 10933
    },
    {
      "epoch": 0.45314766463591527,
      "grad_norm": 0.39611148834228516,
      "learning_rate": 2.734468896348792e-06,
      "loss": 0.6993,
      "step": 10934
    },
    {
      "epoch": 0.45318910854158895,
      "grad_norm": 0.42588523030281067,
      "learning_rate": 2.734261676820424e-06,
      "loss": 0.6927,
      "step": 10935
    },
    {
      "epoch": 0.4532305524472626,
      "grad_norm": 0.4029209613800049,
      "learning_rate": 2.7340544572920553e-06,
      "loss": 0.6797,
      "step": 10936
    },
    {
      "epoch": 0.4532719963529363,
      "grad_norm": 0.39266279339790344,
      "learning_rate": 2.733847237763687e-06,
      "loss": 0.696,
      "step": 10937
    },
    {
      "epoch": 0.45331344025861,
      "grad_norm": 0.4313693642616272,
      "learning_rate": 2.733640018235319e-06,
      "loss": 0.7412,
      "step": 10938
    },
    {
      "epoch": 0.45335488416428366,
      "grad_norm": 0.42305421829223633,
      "learning_rate": 2.7334327987069503e-06,
      "loss": 0.6987,
      "step": 10939
    },
    {
      "epoch": 0.45339632806995733,
      "grad_norm": 0.39201682806015015,
      "learning_rate": 2.733225579178582e-06,
      "loss": 0.6311,
      "step": 10940
    },
    {
      "epoch": 0.453437771975631,
      "grad_norm": 0.3959994912147522,
      "learning_rate": 2.7330183596502135e-06,
      "loss": 0.7034,
      "step": 10941
    },
    {
      "epoch": 0.45347921588130463,
      "grad_norm": 0.4310702383518219,
      "learning_rate": 2.7328111401218453e-06,
      "loss": 0.6826,
      "step": 10942
    },
    {
      "epoch": 0.4535206597869783,
      "grad_norm": 0.39889174699783325,
      "learning_rate": 2.7326039205934767e-06,
      "loss": 0.6494,
      "step": 10943
    },
    {
      "epoch": 0.453562103692652,
      "grad_norm": 0.44837307929992676,
      "learning_rate": 2.7323967010651085e-06,
      "loss": 0.7244,
      "step": 10944
    },
    {
      "epoch": 0.45360354759832566,
      "grad_norm": 0.41153982281684875,
      "learning_rate": 2.73218948153674e-06,
      "loss": 0.6564,
      "step": 10945
    },
    {
      "epoch": 0.45364499150399934,
      "grad_norm": 0.4167439043521881,
      "learning_rate": 2.7319822620083717e-06,
      "loss": 0.6675,
      "step": 10946
    },
    {
      "epoch": 0.453686435409673,
      "grad_norm": 0.4112125337123871,
      "learning_rate": 2.731775042480004e-06,
      "loss": 0.7319,
      "step": 10947
    },
    {
      "epoch": 0.4537278793153467,
      "grad_norm": 0.3924325704574585,
      "learning_rate": 2.731567822951635e-06,
      "loss": 0.7112,
      "step": 10948
    },
    {
      "epoch": 0.45376932322102037,
      "grad_norm": 0.41091832518577576,
      "learning_rate": 2.731360603423267e-06,
      "loss": 0.6785,
      "step": 10949
    },
    {
      "epoch": 0.453810767126694,
      "grad_norm": 0.4320807456970215,
      "learning_rate": 2.7311533838948986e-06,
      "loss": 0.6936,
      "step": 10950
    },
    {
      "epoch": 0.45385221103236767,
      "grad_norm": 0.3961375653743744,
      "learning_rate": 2.7309461643665304e-06,
      "loss": 0.7095,
      "step": 10951
    },
    {
      "epoch": 0.45389365493804135,
      "grad_norm": 0.41803792119026184,
      "learning_rate": 2.7307389448381617e-06,
      "loss": 0.6689,
      "step": 10952
    },
    {
      "epoch": 0.453935098843715,
      "grad_norm": 0.41552016139030457,
      "learning_rate": 2.7305317253097936e-06,
      "loss": 0.7207,
      "step": 10953
    },
    {
      "epoch": 0.4539765427493887,
      "grad_norm": 0.3932361602783203,
      "learning_rate": 2.730324505781425e-06,
      "loss": 0.6637,
      "step": 10954
    },
    {
      "epoch": 0.4540179866550624,
      "grad_norm": 0.40272071957588196,
      "learning_rate": 2.7301172862530568e-06,
      "loss": 0.6996,
      "step": 10955
    },
    {
      "epoch": 0.45405943056073605,
      "grad_norm": 0.47219178080558777,
      "learning_rate": 2.7299100667246886e-06,
      "loss": 0.7329,
      "step": 10956
    },
    {
      "epoch": 0.45410087446640973,
      "grad_norm": 0.4067838490009308,
      "learning_rate": 2.72970284719632e-06,
      "loss": 0.7036,
      "step": 10957
    },
    {
      "epoch": 0.4541423183720834,
      "grad_norm": 0.417812317609787,
      "learning_rate": 2.7294956276679518e-06,
      "loss": 0.6721,
      "step": 10958
    },
    {
      "epoch": 0.45418376227775703,
      "grad_norm": 0.4504159986972809,
      "learning_rate": 2.729288408139583e-06,
      "loss": 0.6643,
      "step": 10959
    },
    {
      "epoch": 0.4542252061834307,
      "grad_norm": 0.4014267325401306,
      "learning_rate": 2.729081188611215e-06,
      "loss": 0.7046,
      "step": 10960
    },
    {
      "epoch": 0.4542666500891044,
      "grad_norm": 0.42609161138534546,
      "learning_rate": 2.7288739690828463e-06,
      "loss": 0.6552,
      "step": 10961
    },
    {
      "epoch": 0.45430809399477806,
      "grad_norm": 0.45836910605430603,
      "learning_rate": 2.728666749554478e-06,
      "loss": 0.7549,
      "step": 10962
    },
    {
      "epoch": 0.45434953790045174,
      "grad_norm": 0.4530719816684723,
      "learning_rate": 2.7284595300261095e-06,
      "loss": 0.7288,
      "step": 10963
    },
    {
      "epoch": 0.4543909818061254,
      "grad_norm": 0.3804531693458557,
      "learning_rate": 2.7282523104977413e-06,
      "loss": 0.6489,
      "step": 10964
    },
    {
      "epoch": 0.4544324257117991,
      "grad_norm": 0.39621302485466003,
      "learning_rate": 2.7280450909693736e-06,
      "loss": 0.6587,
      "step": 10965
    },
    {
      "epoch": 0.45447386961747277,
      "grad_norm": 0.40300020575523376,
      "learning_rate": 2.727837871441005e-06,
      "loss": 0.6772,
      "step": 10966
    },
    {
      "epoch": 0.45451531352314645,
      "grad_norm": 0.3925478458404541,
      "learning_rate": 2.7276306519126368e-06,
      "loss": 0.7024,
      "step": 10967
    },
    {
      "epoch": 0.45455675742882007,
      "grad_norm": 0.4346516728401184,
      "learning_rate": 2.727423432384268e-06,
      "loss": 0.7327,
      "step": 10968
    },
    {
      "epoch": 0.45459820133449375,
      "grad_norm": 0.4222603738307953,
      "learning_rate": 2.7272162128559e-06,
      "loss": 0.7429,
      "step": 10969
    },
    {
      "epoch": 0.4546396452401674,
      "grad_norm": 0.422123521566391,
      "learning_rate": 2.7270089933275313e-06,
      "loss": 0.7382,
      "step": 10970
    },
    {
      "epoch": 0.4546810891458411,
      "grad_norm": 0.4419358968734741,
      "learning_rate": 2.726801773799163e-06,
      "loss": 0.6896,
      "step": 10971
    },
    {
      "epoch": 0.4547225330515148,
      "grad_norm": 0.42431479692459106,
      "learning_rate": 2.7265945542707945e-06,
      "loss": 0.739,
      "step": 10972
    },
    {
      "epoch": 0.45476397695718845,
      "grad_norm": 0.415476530790329,
      "learning_rate": 2.7263873347424264e-06,
      "loss": 0.7039,
      "step": 10973
    },
    {
      "epoch": 0.45480542086286213,
      "grad_norm": 0.413877934217453,
      "learning_rate": 2.726180115214058e-06,
      "loss": 0.7408,
      "step": 10974
    },
    {
      "epoch": 0.4548468647685358,
      "grad_norm": 0.42428210377693176,
      "learning_rate": 2.7259728956856895e-06,
      "loss": 0.6986,
      "step": 10975
    },
    {
      "epoch": 0.4548883086742095,
      "grad_norm": 0.41334787011146545,
      "learning_rate": 2.7257656761573214e-06,
      "loss": 0.6752,
      "step": 10976
    },
    {
      "epoch": 0.4549297525798831,
      "grad_norm": 0.43593302369117737,
      "learning_rate": 2.7255584566289527e-06,
      "loss": 0.7058,
      "step": 10977
    },
    {
      "epoch": 0.4549711964855568,
      "grad_norm": 0.4018903374671936,
      "learning_rate": 2.7253512371005845e-06,
      "loss": 0.7281,
      "step": 10978
    },
    {
      "epoch": 0.45501264039123046,
      "grad_norm": 0.4166567325592041,
      "learning_rate": 2.725144017572216e-06,
      "loss": 0.6733,
      "step": 10979
    },
    {
      "epoch": 0.45505408429690414,
      "grad_norm": 0.4349125027656555,
      "learning_rate": 2.7249367980438477e-06,
      "loss": 0.7083,
      "step": 10980
    },
    {
      "epoch": 0.4550955282025778,
      "grad_norm": 0.41710150241851807,
      "learning_rate": 2.724729578515479e-06,
      "loss": 0.6893,
      "step": 10981
    },
    {
      "epoch": 0.4551369721082515,
      "grad_norm": 0.38541314005851746,
      "learning_rate": 2.724522358987111e-06,
      "loss": 0.6462,
      "step": 10982
    },
    {
      "epoch": 0.45517841601392517,
      "grad_norm": 0.42610839009284973,
      "learning_rate": 2.724315139458743e-06,
      "loss": 0.7142,
      "step": 10983
    },
    {
      "epoch": 0.45521985991959885,
      "grad_norm": 0.3879481852054596,
      "learning_rate": 2.7241079199303746e-06,
      "loss": 0.663,
      "step": 10984
    },
    {
      "epoch": 0.45526130382527247,
      "grad_norm": 0.416305810213089,
      "learning_rate": 2.7239007004020064e-06,
      "loss": 0.6918,
      "step": 10985
    },
    {
      "epoch": 0.45530274773094614,
      "grad_norm": 0.40580788254737854,
      "learning_rate": 2.7236934808736378e-06,
      "loss": 0.6692,
      "step": 10986
    },
    {
      "epoch": 0.4553441916366198,
      "grad_norm": 0.40694156289100647,
      "learning_rate": 2.7234862613452696e-06,
      "loss": 0.6472,
      "step": 10987
    },
    {
      "epoch": 0.4553856355422935,
      "grad_norm": 0.4609130620956421,
      "learning_rate": 2.723279041816901e-06,
      "loss": 0.6803,
      "step": 10988
    },
    {
      "epoch": 0.4554270794479672,
      "grad_norm": 0.3797556161880493,
      "learning_rate": 2.7230718222885328e-06,
      "loss": 0.6997,
      "step": 10989
    },
    {
      "epoch": 0.45546852335364085,
      "grad_norm": 0.3887242376804352,
      "learning_rate": 2.722864602760164e-06,
      "loss": 0.7407,
      "step": 10990
    },
    {
      "epoch": 0.45550996725931453,
      "grad_norm": 0.4107593894004822,
      "learning_rate": 2.722657383231796e-06,
      "loss": 0.6715,
      "step": 10991
    },
    {
      "epoch": 0.4555514111649882,
      "grad_norm": 0.42508140206336975,
      "learning_rate": 2.7224501637034278e-06,
      "loss": 0.7725,
      "step": 10992
    },
    {
      "epoch": 0.4555928550706619,
      "grad_norm": 0.4022448658943176,
      "learning_rate": 2.722242944175059e-06,
      "loss": 0.6921,
      "step": 10993
    },
    {
      "epoch": 0.4556342989763355,
      "grad_norm": 0.41692981123924255,
      "learning_rate": 2.722035724646691e-06,
      "loss": 0.7209,
      "step": 10994
    },
    {
      "epoch": 0.4556757428820092,
      "grad_norm": 0.4232257008552551,
      "learning_rate": 2.7218285051183223e-06,
      "loss": 0.6902,
      "step": 10995
    },
    {
      "epoch": 0.45571718678768286,
      "grad_norm": 0.3977949023246765,
      "learning_rate": 2.721621285589954e-06,
      "loss": 0.6636,
      "step": 10996
    },
    {
      "epoch": 0.45575863069335654,
      "grad_norm": 0.40580636262893677,
      "learning_rate": 2.7214140660615855e-06,
      "loss": 0.6636,
      "step": 10997
    },
    {
      "epoch": 0.4558000745990302,
      "grad_norm": 0.4000755846500397,
      "learning_rate": 2.7212068465332173e-06,
      "loss": 0.6726,
      "step": 10998
    },
    {
      "epoch": 0.4558415185047039,
      "grad_norm": 0.39793479442596436,
      "learning_rate": 2.7209996270048496e-06,
      "loss": 0.6692,
      "step": 10999
    },
    {
      "epoch": 0.45588296241037757,
      "grad_norm": 0.4175099730491638,
      "learning_rate": 2.720792407476481e-06,
      "loss": 0.7214,
      "step": 11000
    },
    {
      "epoch": 0.45592440631605125,
      "grad_norm": 0.3890663683414459,
      "learning_rate": 2.7205851879481128e-06,
      "loss": 0.6875,
      "step": 11001
    },
    {
      "epoch": 0.4559658502217249,
      "grad_norm": 0.420482873916626,
      "learning_rate": 2.720377968419744e-06,
      "loss": 0.7307,
      "step": 11002
    },
    {
      "epoch": 0.45600729412739854,
      "grad_norm": 0.4144669771194458,
      "learning_rate": 2.720170748891376e-06,
      "loss": 0.7368,
      "step": 11003
    },
    {
      "epoch": 0.4560487380330722,
      "grad_norm": 0.4443295896053314,
      "learning_rate": 2.7199635293630074e-06,
      "loss": 0.7135,
      "step": 11004
    },
    {
      "epoch": 0.4560901819387459,
      "grad_norm": 0.40316635370254517,
      "learning_rate": 2.719756309834639e-06,
      "loss": 0.6985,
      "step": 11005
    },
    {
      "epoch": 0.4561316258444196,
      "grad_norm": 0.39365121722221375,
      "learning_rate": 2.7195490903062705e-06,
      "loss": 0.6649,
      "step": 11006
    },
    {
      "epoch": 0.45617306975009325,
      "grad_norm": 0.43081071972846985,
      "learning_rate": 2.7193418707779024e-06,
      "loss": 0.7306,
      "step": 11007
    },
    {
      "epoch": 0.45621451365576693,
      "grad_norm": 0.4056808054447174,
      "learning_rate": 2.719134651249534e-06,
      "loss": 0.6746,
      "step": 11008
    },
    {
      "epoch": 0.4562559575614406,
      "grad_norm": 0.39404046535491943,
      "learning_rate": 2.7189274317211656e-06,
      "loss": 0.6693,
      "step": 11009
    },
    {
      "epoch": 0.4562974014671143,
      "grad_norm": 0.3803918659687042,
      "learning_rate": 2.7187202121927974e-06,
      "loss": 0.687,
      "step": 11010
    },
    {
      "epoch": 0.4563388453727879,
      "grad_norm": 0.39889997243881226,
      "learning_rate": 2.7185129926644287e-06,
      "loss": 0.6818,
      "step": 11011
    },
    {
      "epoch": 0.4563802892784616,
      "grad_norm": 0.3972780108451843,
      "learning_rate": 2.7183057731360606e-06,
      "loss": 0.6542,
      "step": 11012
    },
    {
      "epoch": 0.45642173318413526,
      "grad_norm": 0.4153781533241272,
      "learning_rate": 2.718098553607692e-06,
      "loss": 0.7039,
      "step": 11013
    },
    {
      "epoch": 0.45646317708980894,
      "grad_norm": 0.4415527284145355,
      "learning_rate": 2.7178913340793238e-06,
      "loss": 0.6836,
      "step": 11014
    },
    {
      "epoch": 0.4565046209954826,
      "grad_norm": 0.41895467042922974,
      "learning_rate": 2.717684114550955e-06,
      "loss": 0.7496,
      "step": 11015
    },
    {
      "epoch": 0.4565460649011563,
      "grad_norm": 0.37936803698539734,
      "learning_rate": 2.7174768950225874e-06,
      "loss": 0.6295,
      "step": 11016
    },
    {
      "epoch": 0.45658750880682997,
      "grad_norm": 0.4163251519203186,
      "learning_rate": 2.717269675494219e-06,
      "loss": 0.6478,
      "step": 11017
    },
    {
      "epoch": 0.45662895271250364,
      "grad_norm": 0.3695342242717743,
      "learning_rate": 2.7170624559658506e-06,
      "loss": 0.6758,
      "step": 11018
    },
    {
      "epoch": 0.4566703966181773,
      "grad_norm": 0.4327620565891266,
      "learning_rate": 2.7168552364374824e-06,
      "loss": 0.7075,
      "step": 11019
    },
    {
      "epoch": 0.45671184052385094,
      "grad_norm": 0.4081602990627289,
      "learning_rate": 2.7166480169091138e-06,
      "loss": 0.6647,
      "step": 11020
    },
    {
      "epoch": 0.4567532844295246,
      "grad_norm": 0.39616838097572327,
      "learning_rate": 2.7164407973807456e-06,
      "loss": 0.6938,
      "step": 11021
    },
    {
      "epoch": 0.4567947283351983,
      "grad_norm": 0.4005618691444397,
      "learning_rate": 2.716233577852377e-06,
      "loss": 0.6682,
      "step": 11022
    },
    {
      "epoch": 0.456836172240872,
      "grad_norm": 0.3990097939968109,
      "learning_rate": 2.7160263583240088e-06,
      "loss": 0.708,
      "step": 11023
    },
    {
      "epoch": 0.45687761614654565,
      "grad_norm": 0.4063732326030731,
      "learning_rate": 2.71581913879564e-06,
      "loss": 0.6995,
      "step": 11024
    },
    {
      "epoch": 0.45691906005221933,
      "grad_norm": 0.4292859137058258,
      "learning_rate": 2.715611919267272e-06,
      "loss": 0.7429,
      "step": 11025
    },
    {
      "epoch": 0.456960503957893,
      "grad_norm": 0.4322301745414734,
      "learning_rate": 2.7154046997389038e-06,
      "loss": 0.6583,
      "step": 11026
    },
    {
      "epoch": 0.4570019478635667,
      "grad_norm": 0.39700940251350403,
      "learning_rate": 2.715197480210535e-06,
      "loss": 0.6831,
      "step": 11027
    },
    {
      "epoch": 0.45704339176924036,
      "grad_norm": 0.41327154636383057,
      "learning_rate": 2.714990260682167e-06,
      "loss": 0.7007,
      "step": 11028
    },
    {
      "epoch": 0.457084835674914,
      "grad_norm": 0.3812848627567291,
      "learning_rate": 2.7147830411537983e-06,
      "loss": 0.6865,
      "step": 11029
    },
    {
      "epoch": 0.45712627958058766,
      "grad_norm": 0.4133034646511078,
      "learning_rate": 2.71457582162543e-06,
      "loss": 0.6852,
      "step": 11030
    },
    {
      "epoch": 0.45716772348626133,
      "grad_norm": 0.42187151312828064,
      "learning_rate": 2.7143686020970615e-06,
      "loss": 0.698,
      "step": 11031
    },
    {
      "epoch": 0.457209167391935,
      "grad_norm": 0.3703000247478485,
      "learning_rate": 2.7141613825686934e-06,
      "loss": 0.6816,
      "step": 11032
    },
    {
      "epoch": 0.4572506112976087,
      "grad_norm": 0.45361965894699097,
      "learning_rate": 2.7139541630403247e-06,
      "loss": 0.707,
      "step": 11033
    },
    {
      "epoch": 0.45729205520328237,
      "grad_norm": 0.3807827830314636,
      "learning_rate": 2.713746943511957e-06,
      "loss": 0.7258,
      "step": 11034
    },
    {
      "epoch": 0.45733349910895604,
      "grad_norm": 0.3980445861816406,
      "learning_rate": 2.7135397239835888e-06,
      "loss": 0.6884,
      "step": 11035
    },
    {
      "epoch": 0.4573749430146297,
      "grad_norm": 0.37374451756477356,
      "learning_rate": 2.71333250445522e-06,
      "loss": 0.6396,
      "step": 11036
    },
    {
      "epoch": 0.45741638692030334,
      "grad_norm": 0.3747153878211975,
      "learning_rate": 2.713125284926852e-06,
      "loss": 0.6721,
      "step": 11037
    },
    {
      "epoch": 0.457457830825977,
      "grad_norm": 0.43894830346107483,
      "learning_rate": 2.7129180653984834e-06,
      "loss": 0.7479,
      "step": 11038
    },
    {
      "epoch": 0.4574992747316507,
      "grad_norm": 0.39191851019859314,
      "learning_rate": 2.712710845870115e-06,
      "loss": 0.6581,
      "step": 11039
    },
    {
      "epoch": 0.4575407186373244,
      "grad_norm": 0.4144415259361267,
      "learning_rate": 2.7125036263417466e-06,
      "loss": 0.728,
      "step": 11040
    },
    {
      "epoch": 0.45758216254299805,
      "grad_norm": 0.4504624307155609,
      "learning_rate": 2.7122964068133784e-06,
      "loss": 0.6775,
      "step": 11041
    },
    {
      "epoch": 0.4576236064486717,
      "grad_norm": 0.4267991781234741,
      "learning_rate": 2.7120891872850097e-06,
      "loss": 0.72,
      "step": 11042
    },
    {
      "epoch": 0.4576650503543454,
      "grad_norm": 0.43962401151657104,
      "learning_rate": 2.7118819677566416e-06,
      "loss": 0.7076,
      "step": 11043
    },
    {
      "epoch": 0.4577064942600191,
      "grad_norm": 0.39161714911460876,
      "learning_rate": 2.7116747482282734e-06,
      "loss": 0.6533,
      "step": 11044
    },
    {
      "epoch": 0.45774793816569276,
      "grad_norm": 0.38124167919158936,
      "learning_rate": 2.7114675286999048e-06,
      "loss": 0.6962,
      "step": 11045
    },
    {
      "epoch": 0.4577893820713664,
      "grad_norm": 0.4161529242992401,
      "learning_rate": 2.7112603091715366e-06,
      "loss": 0.7544,
      "step": 11046
    },
    {
      "epoch": 0.45783082597704006,
      "grad_norm": 0.4410530924797058,
      "learning_rate": 2.711053089643168e-06,
      "loss": 0.6686,
      "step": 11047
    },
    {
      "epoch": 0.45787226988271373,
      "grad_norm": 0.5071407556533813,
      "learning_rate": 2.7108458701147998e-06,
      "loss": 0.6958,
      "step": 11048
    },
    {
      "epoch": 0.4579137137883874,
      "grad_norm": 0.4109571874141693,
      "learning_rate": 2.710638650586431e-06,
      "loss": 0.6957,
      "step": 11049
    },
    {
      "epoch": 0.4579551576940611,
      "grad_norm": 0.47870010137557983,
      "learning_rate": 2.7104314310580634e-06,
      "loss": 0.7141,
      "step": 11050
    },
    {
      "epoch": 0.45799660159973477,
      "grad_norm": 0.37814196944236755,
      "learning_rate": 2.7102242115296943e-06,
      "loss": 0.6667,
      "step": 11051
    },
    {
      "epoch": 0.45803804550540844,
      "grad_norm": 0.4037674367427826,
      "learning_rate": 2.7100169920013266e-06,
      "loss": 0.7023,
      "step": 11052
    },
    {
      "epoch": 0.4580794894110821,
      "grad_norm": 0.40082207322120667,
      "learning_rate": 2.7098097724729584e-06,
      "loss": 0.7109,
      "step": 11053
    },
    {
      "epoch": 0.4581209333167558,
      "grad_norm": 0.4080982804298401,
      "learning_rate": 2.7096025529445898e-06,
      "loss": 0.6539,
      "step": 11054
    },
    {
      "epoch": 0.4581623772224294,
      "grad_norm": 0.4657817482948303,
      "learning_rate": 2.7093953334162216e-06,
      "loss": 0.694,
      "step": 11055
    },
    {
      "epoch": 0.4582038211281031,
      "grad_norm": 0.4042403995990753,
      "learning_rate": 2.709188113887853e-06,
      "loss": 0.6467,
      "step": 11056
    },
    {
      "epoch": 0.45824526503377677,
      "grad_norm": 0.39588502049446106,
      "learning_rate": 2.7089808943594848e-06,
      "loss": 0.6543,
      "step": 11057
    },
    {
      "epoch": 0.45828670893945045,
      "grad_norm": 0.4253116548061371,
      "learning_rate": 2.708773674831116e-06,
      "loss": 0.7163,
      "step": 11058
    },
    {
      "epoch": 0.4583281528451241,
      "grad_norm": 0.3875236511230469,
      "learning_rate": 2.708566455302748e-06,
      "loss": 0.6321,
      "step": 11059
    },
    {
      "epoch": 0.4583695967507978,
      "grad_norm": 0.41057148575782776,
      "learning_rate": 2.7083592357743798e-06,
      "loss": 0.6836,
      "step": 11060
    },
    {
      "epoch": 0.4584110406564715,
      "grad_norm": 0.4294869601726532,
      "learning_rate": 2.708152016246011e-06,
      "loss": 0.6649,
      "step": 11061
    },
    {
      "epoch": 0.45845248456214516,
      "grad_norm": 0.43284764885902405,
      "learning_rate": 2.707944796717643e-06,
      "loss": 0.6907,
      "step": 11062
    },
    {
      "epoch": 0.45849392846781883,
      "grad_norm": 0.40644901990890503,
      "learning_rate": 2.7077375771892744e-06,
      "loss": 0.6876,
      "step": 11063
    },
    {
      "epoch": 0.45853537237349246,
      "grad_norm": 0.4286835193634033,
      "learning_rate": 2.707530357660906e-06,
      "loss": 0.7295,
      "step": 11064
    },
    {
      "epoch": 0.45857681627916613,
      "grad_norm": 0.38506999611854553,
      "learning_rate": 2.7073231381325375e-06,
      "loss": 0.6964,
      "step": 11065
    },
    {
      "epoch": 0.4586182601848398,
      "grad_norm": 0.40911439061164856,
      "learning_rate": 2.7071159186041694e-06,
      "loss": 0.6885,
      "step": 11066
    },
    {
      "epoch": 0.4586597040905135,
      "grad_norm": 0.40578508377075195,
      "learning_rate": 2.7069086990758007e-06,
      "loss": 0.6792,
      "step": 11067
    },
    {
      "epoch": 0.45870114799618716,
      "grad_norm": 0.415713906288147,
      "learning_rate": 2.706701479547433e-06,
      "loss": 0.7029,
      "step": 11068
    },
    {
      "epoch": 0.45874259190186084,
      "grad_norm": 0.40949469804763794,
      "learning_rate": 2.7064942600190648e-06,
      "loss": 0.751,
      "step": 11069
    },
    {
      "epoch": 0.4587840358075345,
      "grad_norm": 0.4660669267177582,
      "learning_rate": 2.706287040490696e-06,
      "loss": 0.6812,
      "step": 11070
    },
    {
      "epoch": 0.4588254797132082,
      "grad_norm": 0.42611274123191833,
      "learning_rate": 2.706079820962328e-06,
      "loss": 0.6489,
      "step": 11071
    },
    {
      "epoch": 0.4588669236188818,
      "grad_norm": 0.4142078757286072,
      "learning_rate": 2.7058726014339594e-06,
      "loss": 0.6943,
      "step": 11072
    },
    {
      "epoch": 0.4589083675245555,
      "grad_norm": 0.4200544059276581,
      "learning_rate": 2.705665381905591e-06,
      "loss": 0.6875,
      "step": 11073
    },
    {
      "epoch": 0.45894981143022917,
      "grad_norm": 0.4499918520450592,
      "learning_rate": 2.7054581623772226e-06,
      "loss": 0.7078,
      "step": 11074
    },
    {
      "epoch": 0.45899125533590285,
      "grad_norm": 0.40911298990249634,
      "learning_rate": 2.7052509428488544e-06,
      "loss": 0.658,
      "step": 11075
    },
    {
      "epoch": 0.4590326992415765,
      "grad_norm": 0.4008288085460663,
      "learning_rate": 2.7050437233204858e-06,
      "loss": 0.6582,
      "step": 11076
    },
    {
      "epoch": 0.4590741431472502,
      "grad_norm": 0.4159731864929199,
      "learning_rate": 2.7048365037921176e-06,
      "loss": 0.6538,
      "step": 11077
    },
    {
      "epoch": 0.4591155870529239,
      "grad_norm": 0.39277616143226624,
      "learning_rate": 2.7046292842637494e-06,
      "loss": 0.6495,
      "step": 11078
    },
    {
      "epoch": 0.45915703095859756,
      "grad_norm": 0.4392695426940918,
      "learning_rate": 2.7044220647353808e-06,
      "loss": 0.696,
      "step": 11079
    },
    {
      "epoch": 0.45919847486427123,
      "grad_norm": 0.4126371741294861,
      "learning_rate": 2.7042148452070126e-06,
      "loss": 0.6919,
      "step": 11080
    },
    {
      "epoch": 0.45923991876994485,
      "grad_norm": 0.3887927830219269,
      "learning_rate": 2.704007625678644e-06,
      "loss": 0.6835,
      "step": 11081
    },
    {
      "epoch": 0.45928136267561853,
      "grad_norm": 0.3839223086833954,
      "learning_rate": 2.7038004061502758e-06,
      "loss": 0.693,
      "step": 11082
    },
    {
      "epoch": 0.4593228065812922,
      "grad_norm": 0.38559722900390625,
      "learning_rate": 2.703593186621907e-06,
      "loss": 0.702,
      "step": 11083
    },
    {
      "epoch": 0.4593642504869659,
      "grad_norm": 0.4121198058128357,
      "learning_rate": 2.7033859670935394e-06,
      "loss": 0.7231,
      "step": 11084
    },
    {
      "epoch": 0.45940569439263956,
      "grad_norm": 0.39820775389671326,
      "learning_rate": 2.7031787475651703e-06,
      "loss": 0.7014,
      "step": 11085
    },
    {
      "epoch": 0.45944713829831324,
      "grad_norm": 0.38914281129837036,
      "learning_rate": 2.7029715280368026e-06,
      "loss": 0.6814,
      "step": 11086
    },
    {
      "epoch": 0.4594885822039869,
      "grad_norm": 0.3773519694805145,
      "learning_rate": 2.7027643085084344e-06,
      "loss": 0.6855,
      "step": 11087
    },
    {
      "epoch": 0.4595300261096606,
      "grad_norm": 0.43106913566589355,
      "learning_rate": 2.7025570889800658e-06,
      "loss": 0.6604,
      "step": 11088
    },
    {
      "epoch": 0.45957147001533427,
      "grad_norm": 0.4008869528770447,
      "learning_rate": 2.7023498694516976e-06,
      "loss": 0.6794,
      "step": 11089
    },
    {
      "epoch": 0.4596129139210079,
      "grad_norm": 0.41924574971199036,
      "learning_rate": 2.702142649923329e-06,
      "loss": 0.6448,
      "step": 11090
    },
    {
      "epoch": 0.45965435782668157,
      "grad_norm": 0.40154513716697693,
      "learning_rate": 2.7019354303949608e-06,
      "loss": 0.729,
      "step": 11091
    },
    {
      "epoch": 0.45969580173235525,
      "grad_norm": 0.40835919976234436,
      "learning_rate": 2.701728210866592e-06,
      "loss": 0.6349,
      "step": 11092
    },
    {
      "epoch": 0.4597372456380289,
      "grad_norm": 0.41681328415870667,
      "learning_rate": 2.701520991338224e-06,
      "loss": 0.7322,
      "step": 11093
    },
    {
      "epoch": 0.4597786895437026,
      "grad_norm": 0.4502260088920593,
      "learning_rate": 2.7013137718098554e-06,
      "loss": 0.6733,
      "step": 11094
    },
    {
      "epoch": 0.4598201334493763,
      "grad_norm": 0.37679728865623474,
      "learning_rate": 2.701106552281487e-06,
      "loss": 0.647,
      "step": 11095
    },
    {
      "epoch": 0.45986157735504996,
      "grad_norm": 0.40633487701416016,
      "learning_rate": 2.700899332753119e-06,
      "loss": 0.6885,
      "step": 11096
    },
    {
      "epoch": 0.45990302126072363,
      "grad_norm": 0.4069505035877228,
      "learning_rate": 2.7006921132247504e-06,
      "loss": 0.6829,
      "step": 11097
    },
    {
      "epoch": 0.45994446516639725,
      "grad_norm": 0.44154998660087585,
      "learning_rate": 2.700484893696382e-06,
      "loss": 0.6951,
      "step": 11098
    },
    {
      "epoch": 0.45998590907207093,
      "grad_norm": 0.41747498512268066,
      "learning_rate": 2.7002776741680136e-06,
      "loss": 0.688,
      "step": 11099
    },
    {
      "epoch": 0.4600273529777446,
      "grad_norm": 0.43520429730415344,
      "learning_rate": 2.7000704546396454e-06,
      "loss": 0.7227,
      "step": 11100
    },
    {
      "epoch": 0.4600687968834183,
      "grad_norm": 0.4050680696964264,
      "learning_rate": 2.6998632351112767e-06,
      "loss": 0.689,
      "step": 11101
    },
    {
      "epoch": 0.46011024078909196,
      "grad_norm": 0.3939696252346039,
      "learning_rate": 2.699656015582909e-06,
      "loss": 0.6672,
      "step": 11102
    },
    {
      "epoch": 0.46015168469476564,
      "grad_norm": 0.4252401888370514,
      "learning_rate": 2.69944879605454e-06,
      "loss": 0.7075,
      "step": 11103
    },
    {
      "epoch": 0.4601931286004393,
      "grad_norm": 0.4103040099143982,
      "learning_rate": 2.699241576526172e-06,
      "loss": 0.6995,
      "step": 11104
    },
    {
      "epoch": 0.460234572506113,
      "grad_norm": 0.4442475140094757,
      "learning_rate": 2.699034356997804e-06,
      "loss": 0.7183,
      "step": 11105
    },
    {
      "epoch": 0.46027601641178667,
      "grad_norm": 0.44601768255233765,
      "learning_rate": 2.6988271374694354e-06,
      "loss": 0.7185,
      "step": 11106
    },
    {
      "epoch": 0.4603174603174603,
      "grad_norm": 0.3692470192909241,
      "learning_rate": 2.698619917941067e-06,
      "loss": 0.684,
      "step": 11107
    },
    {
      "epoch": 0.46035890422313397,
      "grad_norm": 0.429762601852417,
      "learning_rate": 2.6984126984126986e-06,
      "loss": 0.7566,
      "step": 11108
    },
    {
      "epoch": 0.46040034812880765,
      "grad_norm": 0.3877778947353363,
      "learning_rate": 2.6982054788843304e-06,
      "loss": 0.6655,
      "step": 11109
    },
    {
      "epoch": 0.4604417920344813,
      "grad_norm": 0.40138059854507446,
      "learning_rate": 2.6979982593559618e-06,
      "loss": 0.7009,
      "step": 11110
    },
    {
      "epoch": 0.460483235940155,
      "grad_norm": 0.41679006814956665,
      "learning_rate": 2.6977910398275936e-06,
      "loss": 0.6556,
      "step": 11111
    },
    {
      "epoch": 0.4605246798458287,
      "grad_norm": 0.44560715556144714,
      "learning_rate": 2.697583820299225e-06,
      "loss": 0.7043,
      "step": 11112
    },
    {
      "epoch": 0.46056612375150235,
      "grad_norm": 0.40604913234710693,
      "learning_rate": 2.6973766007708568e-06,
      "loss": 0.6547,
      "step": 11113
    },
    {
      "epoch": 0.46060756765717603,
      "grad_norm": 0.4061640202999115,
      "learning_rate": 2.6971693812424886e-06,
      "loss": 0.6774,
      "step": 11114
    },
    {
      "epoch": 0.4606490115628497,
      "grad_norm": 0.41733548045158386,
      "learning_rate": 2.69696216171412e-06,
      "loss": 0.6938,
      "step": 11115
    },
    {
      "epoch": 0.46069045546852333,
      "grad_norm": 0.3916509747505188,
      "learning_rate": 2.6967549421857518e-06,
      "loss": 0.7092,
      "step": 11116
    },
    {
      "epoch": 0.460731899374197,
      "grad_norm": 0.4092314839363098,
      "learning_rate": 2.696547722657383e-06,
      "loss": 0.6329,
      "step": 11117
    },
    {
      "epoch": 0.4607733432798707,
      "grad_norm": 0.44937360286712646,
      "learning_rate": 2.6963405031290154e-06,
      "loss": 0.6863,
      "step": 11118
    },
    {
      "epoch": 0.46081478718554436,
      "grad_norm": 0.40027719736099243,
      "learning_rate": 2.6961332836006463e-06,
      "loss": 0.6897,
      "step": 11119
    },
    {
      "epoch": 0.46085623109121804,
      "grad_norm": 0.39452603459358215,
      "learning_rate": 2.6959260640722786e-06,
      "loss": 0.6991,
      "step": 11120
    },
    {
      "epoch": 0.4608976749968917,
      "grad_norm": 0.38913100957870483,
      "learning_rate": 2.6957188445439104e-06,
      "loss": 0.6405,
      "step": 11121
    },
    {
      "epoch": 0.4609391189025654,
      "grad_norm": 0.4196983277797699,
      "learning_rate": 2.6955116250155418e-06,
      "loss": 0.7178,
      "step": 11122
    },
    {
      "epoch": 0.46098056280823907,
      "grad_norm": 0.40368756651878357,
      "learning_rate": 2.6953044054871736e-06,
      "loss": 0.7119,
      "step": 11123
    },
    {
      "epoch": 0.46102200671391275,
      "grad_norm": 0.4047142565250397,
      "learning_rate": 2.695097185958805e-06,
      "loss": 0.6758,
      "step": 11124
    },
    {
      "epoch": 0.46106345061958637,
      "grad_norm": 0.39317867159843445,
      "learning_rate": 2.6948899664304368e-06,
      "loss": 0.703,
      "step": 11125
    },
    {
      "epoch": 0.46110489452526005,
      "grad_norm": 0.42687445878982544,
      "learning_rate": 2.694682746902068e-06,
      "loss": 0.7212,
      "step": 11126
    },
    {
      "epoch": 0.4611463384309337,
      "grad_norm": 0.3832915723323822,
      "learning_rate": 2.6944755273737e-06,
      "loss": 0.6981,
      "step": 11127
    },
    {
      "epoch": 0.4611877823366074,
      "grad_norm": 0.38915538787841797,
      "learning_rate": 2.6942683078453314e-06,
      "loss": 0.6904,
      "step": 11128
    },
    {
      "epoch": 0.4612292262422811,
      "grad_norm": 0.41386979818344116,
      "learning_rate": 2.694061088316963e-06,
      "loss": 0.7229,
      "step": 11129
    },
    {
      "epoch": 0.46127067014795475,
      "grad_norm": 0.4211133122444153,
      "learning_rate": 2.693853868788595e-06,
      "loss": 0.7134,
      "step": 11130
    },
    {
      "epoch": 0.46131211405362843,
      "grad_norm": 0.41509777307510376,
      "learning_rate": 2.6936466492602264e-06,
      "loss": 0.7051,
      "step": 11131
    },
    {
      "epoch": 0.4613535579593021,
      "grad_norm": 0.4328169524669647,
      "learning_rate": 2.693439429731858e-06,
      "loss": 0.647,
      "step": 11132
    },
    {
      "epoch": 0.46139500186497573,
      "grad_norm": 0.41960057616233826,
      "learning_rate": 2.6932322102034896e-06,
      "loss": 0.708,
      "step": 11133
    },
    {
      "epoch": 0.4614364457706494,
      "grad_norm": 0.41000857949256897,
      "learning_rate": 2.6930249906751214e-06,
      "loss": 0.7445,
      "step": 11134
    },
    {
      "epoch": 0.4614778896763231,
      "grad_norm": 0.4023102819919586,
      "learning_rate": 2.6928177711467528e-06,
      "loss": 0.7253,
      "step": 11135
    },
    {
      "epoch": 0.46151933358199676,
      "grad_norm": 0.3988569974899292,
      "learning_rate": 2.692610551618385e-06,
      "loss": 0.6494,
      "step": 11136
    },
    {
      "epoch": 0.46156077748767044,
      "grad_norm": 0.4133259952068329,
      "learning_rate": 2.692403332090016e-06,
      "loss": 0.6929,
      "step": 11137
    },
    {
      "epoch": 0.4616022213933441,
      "grad_norm": 0.4169234037399292,
      "learning_rate": 2.692196112561648e-06,
      "loss": 0.6973,
      "step": 11138
    },
    {
      "epoch": 0.4616436652990178,
      "grad_norm": 0.40685009956359863,
      "learning_rate": 2.69198889303328e-06,
      "loss": 0.6678,
      "step": 11139
    },
    {
      "epoch": 0.46168510920469147,
      "grad_norm": 0.4177339971065521,
      "learning_rate": 2.6917816735049114e-06,
      "loss": 0.6475,
      "step": 11140
    },
    {
      "epoch": 0.46172655311036515,
      "grad_norm": 0.4211559295654297,
      "learning_rate": 2.691574453976543e-06,
      "loss": 0.7119,
      "step": 11141
    },
    {
      "epoch": 0.46176799701603877,
      "grad_norm": 0.3892132341861725,
      "learning_rate": 2.6913672344481746e-06,
      "loss": 0.6901,
      "step": 11142
    },
    {
      "epoch": 0.46180944092171244,
      "grad_norm": 0.39078250527381897,
      "learning_rate": 2.6911600149198064e-06,
      "loss": 0.6045,
      "step": 11143
    },
    {
      "epoch": 0.4618508848273861,
      "grad_norm": 0.42629191279411316,
      "learning_rate": 2.6909527953914378e-06,
      "loss": 0.7203,
      "step": 11144
    },
    {
      "epoch": 0.4618923287330598,
      "grad_norm": 0.3952932357788086,
      "learning_rate": 2.6907455758630696e-06,
      "loss": 0.6184,
      "step": 11145
    },
    {
      "epoch": 0.4619337726387335,
      "grad_norm": 0.4084804654121399,
      "learning_rate": 2.690538356334701e-06,
      "loss": 0.6791,
      "step": 11146
    },
    {
      "epoch": 0.46197521654440715,
      "grad_norm": 0.40639325976371765,
      "learning_rate": 2.6903311368063328e-06,
      "loss": 0.709,
      "step": 11147
    },
    {
      "epoch": 0.46201666045008083,
      "grad_norm": 0.3865593671798706,
      "learning_rate": 2.6901239172779646e-06,
      "loss": 0.7188,
      "step": 11148
    },
    {
      "epoch": 0.4620581043557545,
      "grad_norm": 0.40918588638305664,
      "learning_rate": 2.689916697749596e-06,
      "loss": 0.6631,
      "step": 11149
    },
    {
      "epoch": 0.4620995482614282,
      "grad_norm": 0.4436027705669403,
      "learning_rate": 2.6897094782212278e-06,
      "loss": 0.739,
      "step": 11150
    },
    {
      "epoch": 0.4621409921671018,
      "grad_norm": 0.4142022430896759,
      "learning_rate": 2.689502258692859e-06,
      "loss": 0.7046,
      "step": 11151
    },
    {
      "epoch": 0.4621824360727755,
      "grad_norm": 0.4158035218715668,
      "learning_rate": 2.6892950391644914e-06,
      "loss": 0.6892,
      "step": 11152
    },
    {
      "epoch": 0.46222387997844916,
      "grad_norm": 0.4455680847167969,
      "learning_rate": 2.6890878196361224e-06,
      "loss": 0.7122,
      "step": 11153
    },
    {
      "epoch": 0.46226532388412284,
      "grad_norm": 0.40724867582321167,
      "learning_rate": 2.6888806001077546e-06,
      "loss": 0.7146,
      "step": 11154
    },
    {
      "epoch": 0.4623067677897965,
      "grad_norm": 0.4016520380973816,
      "learning_rate": 2.6886733805793856e-06,
      "loss": 0.686,
      "step": 11155
    },
    {
      "epoch": 0.4623482116954702,
      "grad_norm": 0.41035157442092896,
      "learning_rate": 2.6884661610510178e-06,
      "loss": 0.6741,
      "step": 11156
    },
    {
      "epoch": 0.46238965560114387,
      "grad_norm": 0.403272420167923,
      "learning_rate": 2.6882589415226496e-06,
      "loss": 0.6548,
      "step": 11157
    },
    {
      "epoch": 0.46243109950681754,
      "grad_norm": 0.42731815576553345,
      "learning_rate": 2.688051721994281e-06,
      "loss": 0.7454,
      "step": 11158
    },
    {
      "epoch": 0.46247254341249117,
      "grad_norm": 0.4065122902393341,
      "learning_rate": 2.6878445024659128e-06,
      "loss": 0.677,
      "step": 11159
    },
    {
      "epoch": 0.46251398731816484,
      "grad_norm": 0.38682854175567627,
      "learning_rate": 2.687637282937544e-06,
      "loss": 0.626,
      "step": 11160
    },
    {
      "epoch": 0.4625554312238385,
      "grad_norm": 0.43228790163993835,
      "learning_rate": 2.687430063409176e-06,
      "loss": 0.6841,
      "step": 11161
    },
    {
      "epoch": 0.4625968751295122,
      "grad_norm": 0.3832246661186218,
      "learning_rate": 2.6872228438808074e-06,
      "loss": 0.649,
      "step": 11162
    },
    {
      "epoch": 0.4626383190351859,
      "grad_norm": 0.44277992844581604,
      "learning_rate": 2.687015624352439e-06,
      "loss": 0.6976,
      "step": 11163
    },
    {
      "epoch": 0.46267976294085955,
      "grad_norm": 0.41466712951660156,
      "learning_rate": 2.6868084048240706e-06,
      "loss": 0.6913,
      "step": 11164
    },
    {
      "epoch": 0.46272120684653323,
      "grad_norm": 0.37682676315307617,
      "learning_rate": 2.6866011852957024e-06,
      "loss": 0.6935,
      "step": 11165
    },
    {
      "epoch": 0.4627626507522069,
      "grad_norm": 0.38312166929244995,
      "learning_rate": 2.686393965767334e-06,
      "loss": 0.686,
      "step": 11166
    },
    {
      "epoch": 0.4628040946578806,
      "grad_norm": 0.40527021884918213,
      "learning_rate": 2.6861867462389656e-06,
      "loss": 0.6643,
      "step": 11167
    },
    {
      "epoch": 0.4628455385635542,
      "grad_norm": 0.3953838646411896,
      "learning_rate": 2.6859795267105974e-06,
      "loss": 0.62,
      "step": 11168
    },
    {
      "epoch": 0.4628869824692279,
      "grad_norm": 0.4136052429676056,
      "learning_rate": 2.6857723071822288e-06,
      "loss": 0.6934,
      "step": 11169
    },
    {
      "epoch": 0.46292842637490156,
      "grad_norm": 0.4169231653213501,
      "learning_rate": 2.685565087653861e-06,
      "loss": 0.7235,
      "step": 11170
    },
    {
      "epoch": 0.46296987028057524,
      "grad_norm": 0.4529958665370941,
      "learning_rate": 2.685357868125492e-06,
      "loss": 0.7178,
      "step": 11171
    },
    {
      "epoch": 0.4630113141862489,
      "grad_norm": 0.42302411794662476,
      "learning_rate": 2.685150648597124e-06,
      "loss": 0.6804,
      "step": 11172
    },
    {
      "epoch": 0.4630527580919226,
      "grad_norm": 0.4239014983177185,
      "learning_rate": 2.684943429068756e-06,
      "loss": 0.7234,
      "step": 11173
    },
    {
      "epoch": 0.46309420199759627,
      "grad_norm": 0.4115825593471527,
      "learning_rate": 2.6847362095403874e-06,
      "loss": 0.667,
      "step": 11174
    },
    {
      "epoch": 0.46313564590326994,
      "grad_norm": 0.41107621788978577,
      "learning_rate": 2.684528990012019e-06,
      "loss": 0.7029,
      "step": 11175
    },
    {
      "epoch": 0.4631770898089436,
      "grad_norm": 0.37145453691482544,
      "learning_rate": 2.6843217704836506e-06,
      "loss": 0.7002,
      "step": 11176
    },
    {
      "epoch": 0.46321853371461724,
      "grad_norm": 0.5006809234619141,
      "learning_rate": 2.6841145509552824e-06,
      "loss": 0.7411,
      "step": 11177
    },
    {
      "epoch": 0.4632599776202909,
      "grad_norm": 0.4266074597835541,
      "learning_rate": 2.6839073314269138e-06,
      "loss": 0.6973,
      "step": 11178
    },
    {
      "epoch": 0.4633014215259646,
      "grad_norm": 0.3802022635936737,
      "learning_rate": 2.6837001118985456e-06,
      "loss": 0.6582,
      "step": 11179
    },
    {
      "epoch": 0.4633428654316383,
      "grad_norm": 0.38080036640167236,
      "learning_rate": 2.683492892370177e-06,
      "loss": 0.6433,
      "step": 11180
    },
    {
      "epoch": 0.46338430933731195,
      "grad_norm": 0.4272611439228058,
      "learning_rate": 2.6832856728418088e-06,
      "loss": 0.7402,
      "step": 11181
    },
    {
      "epoch": 0.4634257532429856,
      "grad_norm": 0.4095204770565033,
      "learning_rate": 2.6830784533134406e-06,
      "loss": 0.6753,
      "step": 11182
    },
    {
      "epoch": 0.4634671971486593,
      "grad_norm": 0.39807790517807007,
      "learning_rate": 2.682871233785072e-06,
      "loss": 0.6157,
      "step": 11183
    },
    {
      "epoch": 0.463508641054333,
      "grad_norm": 0.3881725072860718,
      "learning_rate": 2.6826640142567038e-06,
      "loss": 0.668,
      "step": 11184
    },
    {
      "epoch": 0.46355008496000666,
      "grad_norm": 0.4209165871143341,
      "learning_rate": 2.682456794728335e-06,
      "loss": 0.6758,
      "step": 11185
    },
    {
      "epoch": 0.4635915288656803,
      "grad_norm": 0.4349238872528076,
      "learning_rate": 2.6822495751999674e-06,
      "loss": 0.6682,
      "step": 11186
    },
    {
      "epoch": 0.46363297277135396,
      "grad_norm": 0.3940873146057129,
      "learning_rate": 2.6820423556715984e-06,
      "loss": 0.6687,
      "step": 11187
    },
    {
      "epoch": 0.46367441667702763,
      "grad_norm": 0.4084358215332031,
      "learning_rate": 2.6818351361432306e-06,
      "loss": 0.6936,
      "step": 11188
    },
    {
      "epoch": 0.4637158605827013,
      "grad_norm": 0.39920103549957275,
      "learning_rate": 2.681627916614862e-06,
      "loss": 0.6771,
      "step": 11189
    },
    {
      "epoch": 0.463757304488375,
      "grad_norm": 0.39023837447166443,
      "learning_rate": 2.681420697086494e-06,
      "loss": 0.7234,
      "step": 11190
    },
    {
      "epoch": 0.46379874839404867,
      "grad_norm": 0.4536566734313965,
      "learning_rate": 2.6812134775581256e-06,
      "loss": 0.7559,
      "step": 11191
    },
    {
      "epoch": 0.46384019229972234,
      "grad_norm": 0.40461117029190063,
      "learning_rate": 2.681006258029757e-06,
      "loss": 0.7167,
      "step": 11192
    },
    {
      "epoch": 0.463881636205396,
      "grad_norm": 0.41064390540122986,
      "learning_rate": 2.680799038501389e-06,
      "loss": 0.7327,
      "step": 11193
    },
    {
      "epoch": 0.46392308011106964,
      "grad_norm": 0.4117027521133423,
      "learning_rate": 2.68059181897302e-06,
      "loss": 0.676,
      "step": 11194
    },
    {
      "epoch": 0.4639645240167433,
      "grad_norm": 0.42216360569000244,
      "learning_rate": 2.680384599444652e-06,
      "loss": 0.6895,
      "step": 11195
    },
    {
      "epoch": 0.464005967922417,
      "grad_norm": 0.41259244084358215,
      "learning_rate": 2.6801773799162834e-06,
      "loss": 0.7015,
      "step": 11196
    },
    {
      "epoch": 0.4640474118280907,
      "grad_norm": 0.4024483859539032,
      "learning_rate": 2.679970160387915e-06,
      "loss": 0.6859,
      "step": 11197
    },
    {
      "epoch": 0.46408885573376435,
      "grad_norm": 0.4277755916118622,
      "learning_rate": 2.6797629408595466e-06,
      "loss": 0.6758,
      "step": 11198
    },
    {
      "epoch": 0.464130299639438,
      "grad_norm": 0.45284128189086914,
      "learning_rate": 2.6795557213311784e-06,
      "loss": 0.7598,
      "step": 11199
    },
    {
      "epoch": 0.4641717435451117,
      "grad_norm": 0.419006884098053,
      "learning_rate": 2.67934850180281e-06,
      "loss": 0.6592,
      "step": 11200
    },
    {
      "epoch": 0.4642131874507854,
      "grad_norm": 0.4490922689437866,
      "learning_rate": 2.6791412822744416e-06,
      "loss": 0.76,
      "step": 11201
    },
    {
      "epoch": 0.46425463135645906,
      "grad_norm": 0.4028441309928894,
      "learning_rate": 2.678934062746074e-06,
      "loss": 0.7141,
      "step": 11202
    },
    {
      "epoch": 0.4642960752621327,
      "grad_norm": 0.4196886420249939,
      "learning_rate": 2.6787268432177048e-06,
      "loss": 0.7167,
      "step": 11203
    },
    {
      "epoch": 0.46433751916780636,
      "grad_norm": 0.3859315514564514,
      "learning_rate": 2.678519623689337e-06,
      "loss": 0.6827,
      "step": 11204
    },
    {
      "epoch": 0.46437896307348003,
      "grad_norm": 0.40718165040016174,
      "learning_rate": 2.678312404160968e-06,
      "loss": 0.6705,
      "step": 11205
    },
    {
      "epoch": 0.4644204069791537,
      "grad_norm": 0.4229024350643158,
      "learning_rate": 2.6781051846326e-06,
      "loss": 0.6963,
      "step": 11206
    },
    {
      "epoch": 0.4644618508848274,
      "grad_norm": 0.4159987270832062,
      "learning_rate": 2.6778979651042316e-06,
      "loss": 0.6324,
      "step": 11207
    },
    {
      "epoch": 0.46450329479050106,
      "grad_norm": 0.3814437985420227,
      "learning_rate": 2.6776907455758634e-06,
      "loss": 0.629,
      "step": 11208
    },
    {
      "epoch": 0.46454473869617474,
      "grad_norm": 0.3941105008125305,
      "learning_rate": 2.677483526047495e-06,
      "loss": 0.7175,
      "step": 11209
    },
    {
      "epoch": 0.4645861826018484,
      "grad_norm": 0.4210870563983917,
      "learning_rate": 2.6772763065191266e-06,
      "loss": 0.6755,
      "step": 11210
    },
    {
      "epoch": 0.4646276265075221,
      "grad_norm": 0.40596073865890503,
      "learning_rate": 2.6770690869907584e-06,
      "loss": 0.7295,
      "step": 11211
    },
    {
      "epoch": 0.4646690704131957,
      "grad_norm": 0.42053118348121643,
      "learning_rate": 2.6768618674623898e-06,
      "loss": 0.7454,
      "step": 11212
    },
    {
      "epoch": 0.4647105143188694,
      "grad_norm": 0.4199621379375458,
      "learning_rate": 2.6766546479340216e-06,
      "loss": 0.6776,
      "step": 11213
    },
    {
      "epoch": 0.46475195822454307,
      "grad_norm": 0.4141105115413666,
      "learning_rate": 2.676447428405653e-06,
      "loss": 0.6987,
      "step": 11214
    },
    {
      "epoch": 0.46479340213021675,
      "grad_norm": 0.37793076038360596,
      "learning_rate": 2.6762402088772848e-06,
      "loss": 0.6753,
      "step": 11215
    },
    {
      "epoch": 0.4648348460358904,
      "grad_norm": 0.448324590921402,
      "learning_rate": 2.676032989348916e-06,
      "loss": 0.748,
      "step": 11216
    },
    {
      "epoch": 0.4648762899415641,
      "grad_norm": 0.40591832995414734,
      "learning_rate": 2.675825769820548e-06,
      "loss": 0.6754,
      "step": 11217
    },
    {
      "epoch": 0.4649177338472378,
      "grad_norm": 0.3897480368614197,
      "learning_rate": 2.6756185502921798e-06,
      "loss": 0.629,
      "step": 11218
    },
    {
      "epoch": 0.46495917775291146,
      "grad_norm": 0.42676663398742676,
      "learning_rate": 2.675411330763811e-06,
      "loss": 0.6923,
      "step": 11219
    },
    {
      "epoch": 0.4650006216585851,
      "grad_norm": 0.4161882698535919,
      "learning_rate": 2.6752041112354434e-06,
      "loss": 0.6992,
      "step": 11220
    },
    {
      "epoch": 0.46504206556425876,
      "grad_norm": 0.4115017354488373,
      "learning_rate": 2.6749968917070744e-06,
      "loss": 0.7063,
      "step": 11221
    },
    {
      "epoch": 0.46508350946993243,
      "grad_norm": 0.4191448390483856,
      "learning_rate": 2.6747896721787066e-06,
      "loss": 0.6975,
      "step": 11222
    },
    {
      "epoch": 0.4651249533756061,
      "grad_norm": 0.46519437432289124,
      "learning_rate": 2.674582452650338e-06,
      "loss": 0.6843,
      "step": 11223
    },
    {
      "epoch": 0.4651663972812798,
      "grad_norm": 0.4161173403263092,
      "learning_rate": 2.67437523312197e-06,
      "loss": 0.675,
      "step": 11224
    },
    {
      "epoch": 0.46520784118695346,
      "grad_norm": 0.3683862090110779,
      "learning_rate": 2.674168013593601e-06,
      "loss": 0.6545,
      "step": 11225
    },
    {
      "epoch": 0.46524928509262714,
      "grad_norm": 0.41165271401405334,
      "learning_rate": 2.673960794065233e-06,
      "loss": 0.6991,
      "step": 11226
    },
    {
      "epoch": 0.4652907289983008,
      "grad_norm": 0.42277175188064575,
      "learning_rate": 2.673753574536865e-06,
      "loss": 0.6743,
      "step": 11227
    },
    {
      "epoch": 0.4653321729039745,
      "grad_norm": 0.43579211831092834,
      "learning_rate": 2.673546355008496e-06,
      "loss": 0.7025,
      "step": 11228
    },
    {
      "epoch": 0.4653736168096481,
      "grad_norm": 0.3758833110332489,
      "learning_rate": 2.673339135480128e-06,
      "loss": 0.7156,
      "step": 11229
    },
    {
      "epoch": 0.4654150607153218,
      "grad_norm": 0.4107937216758728,
      "learning_rate": 2.6731319159517594e-06,
      "loss": 0.6841,
      "step": 11230
    },
    {
      "epoch": 0.46545650462099547,
      "grad_norm": 0.4381679594516754,
      "learning_rate": 2.672924696423391e-06,
      "loss": 0.7385,
      "step": 11231
    },
    {
      "epoch": 0.46549794852666915,
      "grad_norm": 0.4296608567237854,
      "learning_rate": 2.6727174768950226e-06,
      "loss": 0.7539,
      "step": 11232
    },
    {
      "epoch": 0.4655393924323428,
      "grad_norm": 0.4551069736480713,
      "learning_rate": 2.6725102573666544e-06,
      "loss": 0.7246,
      "step": 11233
    },
    {
      "epoch": 0.4655808363380165,
      "grad_norm": 0.3940431475639343,
      "learning_rate": 2.672303037838286e-06,
      "loss": 0.6309,
      "step": 11234
    },
    {
      "epoch": 0.4656222802436902,
      "grad_norm": 0.4092904031276703,
      "learning_rate": 2.6720958183099176e-06,
      "loss": 0.6848,
      "step": 11235
    },
    {
      "epoch": 0.46566372414936386,
      "grad_norm": 0.40257173776626587,
      "learning_rate": 2.67188859878155e-06,
      "loss": 0.7053,
      "step": 11236
    },
    {
      "epoch": 0.46570516805503753,
      "grad_norm": 0.4007249176502228,
      "learning_rate": 2.6716813792531808e-06,
      "loss": 0.752,
      "step": 11237
    },
    {
      "epoch": 0.46574661196071115,
      "grad_norm": 0.39026397466659546,
      "learning_rate": 2.671474159724813e-06,
      "loss": 0.6464,
      "step": 11238
    },
    {
      "epoch": 0.46578805586638483,
      "grad_norm": 0.40933090448379517,
      "learning_rate": 2.671266940196444e-06,
      "loss": 0.6733,
      "step": 11239
    },
    {
      "epoch": 0.4658294997720585,
      "grad_norm": 0.4237338602542877,
      "learning_rate": 2.671059720668076e-06,
      "loss": 0.6774,
      "step": 11240
    },
    {
      "epoch": 0.4658709436777322,
      "grad_norm": 0.40366876125335693,
      "learning_rate": 2.6708525011397076e-06,
      "loss": 0.6589,
      "step": 11241
    },
    {
      "epoch": 0.46591238758340586,
      "grad_norm": 0.40464428067207336,
      "learning_rate": 2.6706452816113394e-06,
      "loss": 0.6843,
      "step": 11242
    },
    {
      "epoch": 0.46595383148907954,
      "grad_norm": 0.43191391229629517,
      "learning_rate": 2.670438062082971e-06,
      "loss": 0.6853,
      "step": 11243
    },
    {
      "epoch": 0.4659952753947532,
      "grad_norm": 0.42226117849349976,
      "learning_rate": 2.6702308425546026e-06,
      "loss": 0.6442,
      "step": 11244
    },
    {
      "epoch": 0.4660367193004269,
      "grad_norm": 0.3899506628513336,
      "learning_rate": 2.6700236230262344e-06,
      "loss": 0.703,
      "step": 11245
    },
    {
      "epoch": 0.4660781632061005,
      "grad_norm": 0.4078708291053772,
      "learning_rate": 2.6698164034978658e-06,
      "loss": 0.7422,
      "step": 11246
    },
    {
      "epoch": 0.4661196071117742,
      "grad_norm": 0.4090125262737274,
      "learning_rate": 2.6696091839694976e-06,
      "loss": 0.683,
      "step": 11247
    },
    {
      "epoch": 0.46616105101744787,
      "grad_norm": 0.40038812160491943,
      "learning_rate": 2.669401964441129e-06,
      "loss": 0.6703,
      "step": 11248
    },
    {
      "epoch": 0.46620249492312155,
      "grad_norm": 0.39683952927589417,
      "learning_rate": 2.669194744912761e-06,
      "loss": 0.7209,
      "step": 11249
    },
    {
      "epoch": 0.4662439388287952,
      "grad_norm": 0.3966313898563385,
      "learning_rate": 2.668987525384392e-06,
      "loss": 0.7024,
      "step": 11250
    },
    {
      "epoch": 0.4662853827344689,
      "grad_norm": 0.43369200825691223,
      "learning_rate": 2.668780305856024e-06,
      "loss": 0.7231,
      "step": 11251
    },
    {
      "epoch": 0.4663268266401426,
      "grad_norm": 0.3945905566215515,
      "learning_rate": 2.668573086327656e-06,
      "loss": 0.7018,
      "step": 11252
    },
    {
      "epoch": 0.46636827054581625,
      "grad_norm": 0.42186418175697327,
      "learning_rate": 2.668365866799287e-06,
      "loss": 0.7322,
      "step": 11253
    },
    {
      "epoch": 0.46640971445148993,
      "grad_norm": 0.37021923065185547,
      "learning_rate": 2.6681586472709194e-06,
      "loss": 0.6963,
      "step": 11254
    },
    {
      "epoch": 0.46645115835716355,
      "grad_norm": 0.44502708315849304,
      "learning_rate": 2.6679514277425504e-06,
      "loss": 0.6873,
      "step": 11255
    },
    {
      "epoch": 0.46649260226283723,
      "grad_norm": 0.4111633598804474,
      "learning_rate": 2.6677442082141826e-06,
      "loss": 0.7158,
      "step": 11256
    },
    {
      "epoch": 0.4665340461685109,
      "grad_norm": 0.3981391191482544,
      "learning_rate": 2.667536988685814e-06,
      "loss": 0.6333,
      "step": 11257
    },
    {
      "epoch": 0.4665754900741846,
      "grad_norm": 0.4147587716579437,
      "learning_rate": 2.667329769157446e-06,
      "loss": 0.7153,
      "step": 11258
    },
    {
      "epoch": 0.46661693397985826,
      "grad_norm": 0.36941131949424744,
      "learning_rate": 2.667122549629077e-06,
      "loss": 0.6287,
      "step": 11259
    },
    {
      "epoch": 0.46665837788553194,
      "grad_norm": 0.4135850965976715,
      "learning_rate": 2.666915330100709e-06,
      "loss": 0.7053,
      "step": 11260
    },
    {
      "epoch": 0.4666998217912056,
      "grad_norm": 0.4090293347835541,
      "learning_rate": 2.666708110572341e-06,
      "loss": 0.694,
      "step": 11261
    },
    {
      "epoch": 0.4667412656968793,
      "grad_norm": 0.39830052852630615,
      "learning_rate": 2.666500891043972e-06,
      "loss": 0.7053,
      "step": 11262
    },
    {
      "epoch": 0.46678270960255297,
      "grad_norm": 0.40726739168167114,
      "learning_rate": 2.666293671515604e-06,
      "loss": 0.6465,
      "step": 11263
    },
    {
      "epoch": 0.4668241535082266,
      "grad_norm": 0.39864087104797363,
      "learning_rate": 2.6660864519872354e-06,
      "loss": 0.6907,
      "step": 11264
    },
    {
      "epoch": 0.46686559741390027,
      "grad_norm": 0.4562920331954956,
      "learning_rate": 2.665879232458867e-06,
      "loss": 0.749,
      "step": 11265
    },
    {
      "epoch": 0.46690704131957395,
      "grad_norm": 0.39818036556243896,
      "learning_rate": 2.6656720129304986e-06,
      "loss": 0.6475,
      "step": 11266
    },
    {
      "epoch": 0.4669484852252476,
      "grad_norm": 0.4208519458770752,
      "learning_rate": 2.6654647934021304e-06,
      "loss": 0.6595,
      "step": 11267
    },
    {
      "epoch": 0.4669899291309213,
      "grad_norm": 0.41602328419685364,
      "learning_rate": 2.6652575738737618e-06,
      "loss": 0.6725,
      "step": 11268
    },
    {
      "epoch": 0.467031373036595,
      "grad_norm": 0.4702516794204712,
      "learning_rate": 2.6650503543453936e-06,
      "loss": 0.7239,
      "step": 11269
    },
    {
      "epoch": 0.46707281694226865,
      "grad_norm": 0.41449037194252014,
      "learning_rate": 2.664843134817026e-06,
      "loss": 0.719,
      "step": 11270
    },
    {
      "epoch": 0.46711426084794233,
      "grad_norm": 0.4287029504776001,
      "learning_rate": 2.6646359152886568e-06,
      "loss": 0.6885,
      "step": 11271
    },
    {
      "epoch": 0.467155704753616,
      "grad_norm": 0.41523489356040955,
      "learning_rate": 2.664428695760289e-06,
      "loss": 0.7067,
      "step": 11272
    },
    {
      "epoch": 0.46719714865928963,
      "grad_norm": 0.44421476125717163,
      "learning_rate": 2.66422147623192e-06,
      "loss": 0.6849,
      "step": 11273
    },
    {
      "epoch": 0.4672385925649633,
      "grad_norm": 0.41227471828460693,
      "learning_rate": 2.664014256703552e-06,
      "loss": 0.6965,
      "step": 11274
    },
    {
      "epoch": 0.467280036470637,
      "grad_norm": 0.43132659792900085,
      "learning_rate": 2.6638070371751836e-06,
      "loss": 0.7288,
      "step": 11275
    },
    {
      "epoch": 0.46732148037631066,
      "grad_norm": 0.4015710651874542,
      "learning_rate": 2.6635998176468154e-06,
      "loss": 0.6609,
      "step": 11276
    },
    {
      "epoch": 0.46736292428198434,
      "grad_norm": 0.41580522060394287,
      "learning_rate": 2.6633925981184468e-06,
      "loss": 0.682,
      "step": 11277
    },
    {
      "epoch": 0.467404368187658,
      "grad_norm": 0.4613843262195587,
      "learning_rate": 2.6631853785900786e-06,
      "loss": 0.6816,
      "step": 11278
    },
    {
      "epoch": 0.4674458120933317,
      "grad_norm": 0.41593068838119507,
      "learning_rate": 2.6629781590617104e-06,
      "loss": 0.6643,
      "step": 11279
    },
    {
      "epoch": 0.46748725599900537,
      "grad_norm": 0.41972148418426514,
      "learning_rate": 2.662770939533342e-06,
      "loss": 0.6765,
      "step": 11280
    },
    {
      "epoch": 0.467528699904679,
      "grad_norm": 0.4007648527622223,
      "learning_rate": 2.6625637200049736e-06,
      "loss": 0.6956,
      "step": 11281
    },
    {
      "epoch": 0.46757014381035267,
      "grad_norm": 0.3916735053062439,
      "learning_rate": 2.662356500476605e-06,
      "loss": 0.6693,
      "step": 11282
    },
    {
      "epoch": 0.46761158771602634,
      "grad_norm": 0.42664656043052673,
      "learning_rate": 2.662149280948237e-06,
      "loss": 0.6869,
      "step": 11283
    },
    {
      "epoch": 0.4676530316217,
      "grad_norm": 0.4421672523021698,
      "learning_rate": 2.661942061419868e-06,
      "loss": 0.755,
      "step": 11284
    },
    {
      "epoch": 0.4676944755273737,
      "grad_norm": 0.3841400742530823,
      "learning_rate": 2.6617348418915e-06,
      "loss": 0.686,
      "step": 11285
    },
    {
      "epoch": 0.4677359194330474,
      "grad_norm": 0.39451977610588074,
      "learning_rate": 2.6615276223631314e-06,
      "loss": 0.6675,
      "step": 11286
    },
    {
      "epoch": 0.46777736333872105,
      "grad_norm": 0.4100334048271179,
      "learning_rate": 2.661320402834763e-06,
      "loss": 0.6713,
      "step": 11287
    },
    {
      "epoch": 0.46781880724439473,
      "grad_norm": 0.3835384249687195,
      "learning_rate": 2.6611131833063954e-06,
      "loss": 0.655,
      "step": 11288
    },
    {
      "epoch": 0.4678602511500684,
      "grad_norm": 0.38414764404296875,
      "learning_rate": 2.6609059637780264e-06,
      "loss": 0.6655,
      "step": 11289
    },
    {
      "epoch": 0.46790169505574203,
      "grad_norm": 0.43537354469299316,
      "learning_rate": 2.6606987442496586e-06,
      "loss": 0.686,
      "step": 11290
    },
    {
      "epoch": 0.4679431389614157,
      "grad_norm": 0.41374656558036804,
      "learning_rate": 2.66049152472129e-06,
      "loss": 0.6711,
      "step": 11291
    },
    {
      "epoch": 0.4679845828670894,
      "grad_norm": 0.43200749158859253,
      "learning_rate": 2.660284305192922e-06,
      "loss": 0.6501,
      "step": 11292
    },
    {
      "epoch": 0.46802602677276306,
      "grad_norm": 0.39816364645957947,
      "learning_rate": 2.660077085664553e-06,
      "loss": 0.6589,
      "step": 11293
    },
    {
      "epoch": 0.46806747067843674,
      "grad_norm": 0.42823320627212524,
      "learning_rate": 2.659869866136185e-06,
      "loss": 0.7546,
      "step": 11294
    },
    {
      "epoch": 0.4681089145841104,
      "grad_norm": 0.40132007002830505,
      "learning_rate": 2.659662646607817e-06,
      "loss": 0.6503,
      "step": 11295
    },
    {
      "epoch": 0.4681503584897841,
      "grad_norm": 0.4054487943649292,
      "learning_rate": 2.659455427079448e-06,
      "loss": 0.6741,
      "step": 11296
    },
    {
      "epoch": 0.46819180239545777,
      "grad_norm": 0.404396116733551,
      "learning_rate": 2.65924820755108e-06,
      "loss": 0.6852,
      "step": 11297
    },
    {
      "epoch": 0.46823324630113144,
      "grad_norm": 0.40296295285224915,
      "learning_rate": 2.6590409880227114e-06,
      "loss": 0.6823,
      "step": 11298
    },
    {
      "epoch": 0.46827469020680507,
      "grad_norm": 0.3938201069831848,
      "learning_rate": 2.658833768494343e-06,
      "loss": 0.6423,
      "step": 11299
    },
    {
      "epoch": 0.46831613411247874,
      "grad_norm": 0.3931317925453186,
      "learning_rate": 2.6586265489659746e-06,
      "loss": 0.6826,
      "step": 11300
    },
    {
      "epoch": 0.4683575780181524,
      "grad_norm": 0.4503451883792877,
      "learning_rate": 2.6584193294376064e-06,
      "loss": 0.7451,
      "step": 11301
    },
    {
      "epoch": 0.4683990219238261,
      "grad_norm": 0.4446885585784912,
      "learning_rate": 2.6582121099092378e-06,
      "loss": 0.6798,
      "step": 11302
    },
    {
      "epoch": 0.4684404658294998,
      "grad_norm": 0.4059770703315735,
      "learning_rate": 2.6580048903808696e-06,
      "loss": 0.6915,
      "step": 11303
    },
    {
      "epoch": 0.46848190973517345,
      "grad_norm": 0.4197401702404022,
      "learning_rate": 2.657797670852502e-06,
      "loss": 0.6545,
      "step": 11304
    },
    {
      "epoch": 0.46852335364084713,
      "grad_norm": 0.4146048426628113,
      "learning_rate": 2.6575904513241328e-06,
      "loss": 0.7275,
      "step": 11305
    },
    {
      "epoch": 0.4685647975465208,
      "grad_norm": 0.40017473697662354,
      "learning_rate": 2.657383231795765e-06,
      "loss": 0.6351,
      "step": 11306
    },
    {
      "epoch": 0.4686062414521944,
      "grad_norm": 0.4120080769062042,
      "learning_rate": 2.657176012267396e-06,
      "loss": 0.6461,
      "step": 11307
    },
    {
      "epoch": 0.4686476853578681,
      "grad_norm": 0.4195287525653839,
      "learning_rate": 2.6569687927390282e-06,
      "loss": 0.6262,
      "step": 11308
    },
    {
      "epoch": 0.4686891292635418,
      "grad_norm": 0.3913799226284027,
      "learning_rate": 2.6567615732106596e-06,
      "loss": 0.689,
      "step": 11309
    },
    {
      "epoch": 0.46873057316921546,
      "grad_norm": 0.4036743640899658,
      "learning_rate": 2.6565543536822914e-06,
      "loss": 0.6924,
      "step": 11310
    },
    {
      "epoch": 0.46877201707488914,
      "grad_norm": 0.4109649956226349,
      "learning_rate": 2.656347134153923e-06,
      "loss": 0.6567,
      "step": 11311
    },
    {
      "epoch": 0.4688134609805628,
      "grad_norm": 0.42733505368232727,
      "learning_rate": 2.6561399146255546e-06,
      "loss": 0.6675,
      "step": 11312
    },
    {
      "epoch": 0.4688549048862365,
      "grad_norm": 0.41861337423324585,
      "learning_rate": 2.6559326950971864e-06,
      "loss": 0.722,
      "step": 11313
    },
    {
      "epoch": 0.46889634879191017,
      "grad_norm": 0.3882550895214081,
      "learning_rate": 2.655725475568818e-06,
      "loss": 0.644,
      "step": 11314
    },
    {
      "epoch": 0.46893779269758384,
      "grad_norm": 0.39804166555404663,
      "learning_rate": 2.6555182560404496e-06,
      "loss": 0.6884,
      "step": 11315
    },
    {
      "epoch": 0.46897923660325747,
      "grad_norm": 0.40709370374679565,
      "learning_rate": 2.655311036512081e-06,
      "loss": 0.6963,
      "step": 11316
    },
    {
      "epoch": 0.46902068050893114,
      "grad_norm": 0.4176960587501526,
      "learning_rate": 2.655103816983713e-06,
      "loss": 0.6904,
      "step": 11317
    },
    {
      "epoch": 0.4690621244146048,
      "grad_norm": 0.4087452292442322,
      "learning_rate": 2.654896597455344e-06,
      "loss": 0.7083,
      "step": 11318
    },
    {
      "epoch": 0.4691035683202785,
      "grad_norm": 0.4135391116142273,
      "learning_rate": 2.654689377926976e-06,
      "loss": 0.6893,
      "step": 11319
    },
    {
      "epoch": 0.4691450122259522,
      "grad_norm": 0.4051947295665741,
      "learning_rate": 2.6544821583986074e-06,
      "loss": 0.7078,
      "step": 11320
    },
    {
      "epoch": 0.46918645613162585,
      "grad_norm": 0.458431601524353,
      "learning_rate": 2.654274938870239e-06,
      "loss": 0.7285,
      "step": 11321
    },
    {
      "epoch": 0.46922790003729953,
      "grad_norm": 0.3892737030982971,
      "learning_rate": 2.6540677193418714e-06,
      "loss": 0.6636,
      "step": 11322
    },
    {
      "epoch": 0.4692693439429732,
      "grad_norm": 0.42468512058258057,
      "learning_rate": 2.6538604998135024e-06,
      "loss": 0.6952,
      "step": 11323
    },
    {
      "epoch": 0.4693107878486469,
      "grad_norm": 0.3670247793197632,
      "learning_rate": 2.6536532802851346e-06,
      "loss": 0.6285,
      "step": 11324
    },
    {
      "epoch": 0.4693522317543205,
      "grad_norm": 0.428126722574234,
      "learning_rate": 2.653446060756766e-06,
      "loss": 0.7373,
      "step": 11325
    },
    {
      "epoch": 0.4693936756599942,
      "grad_norm": 0.4004904329776764,
      "learning_rate": 2.653238841228398e-06,
      "loss": 0.6351,
      "step": 11326
    },
    {
      "epoch": 0.46943511956566786,
      "grad_norm": 0.4415145814418793,
      "learning_rate": 2.653031621700029e-06,
      "loss": 0.7209,
      "step": 11327
    },
    {
      "epoch": 0.46947656347134153,
      "grad_norm": 0.4037712812423706,
      "learning_rate": 2.652824402171661e-06,
      "loss": 0.7222,
      "step": 11328
    },
    {
      "epoch": 0.4695180073770152,
      "grad_norm": 0.40898892283439636,
      "learning_rate": 2.6526171826432924e-06,
      "loss": 0.6545,
      "step": 11329
    },
    {
      "epoch": 0.4695594512826889,
      "grad_norm": 0.4330713748931885,
      "learning_rate": 2.652409963114924e-06,
      "loss": 0.6597,
      "step": 11330
    },
    {
      "epoch": 0.46960089518836257,
      "grad_norm": 0.3888359069824219,
      "learning_rate": 2.652202743586556e-06,
      "loss": 0.6787,
      "step": 11331
    },
    {
      "epoch": 0.46964233909403624,
      "grad_norm": 0.43450841307640076,
      "learning_rate": 2.6519955240581874e-06,
      "loss": 0.683,
      "step": 11332
    },
    {
      "epoch": 0.4696837829997099,
      "grad_norm": 0.3883165717124939,
      "learning_rate": 2.651788304529819e-06,
      "loss": 0.6405,
      "step": 11333
    },
    {
      "epoch": 0.46972522690538354,
      "grad_norm": 0.4168032705783844,
      "learning_rate": 2.6515810850014506e-06,
      "loss": 0.6758,
      "step": 11334
    },
    {
      "epoch": 0.4697666708110572,
      "grad_norm": 0.3981277644634247,
      "learning_rate": 2.6513738654730824e-06,
      "loss": 0.5984,
      "step": 11335
    },
    {
      "epoch": 0.4698081147167309,
      "grad_norm": 0.463856041431427,
      "learning_rate": 2.6511666459447138e-06,
      "loss": 0.6638,
      "step": 11336
    },
    {
      "epoch": 0.4698495586224046,
      "grad_norm": 0.403384804725647,
      "learning_rate": 2.6509594264163456e-06,
      "loss": 0.6726,
      "step": 11337
    },
    {
      "epoch": 0.46989100252807825,
      "grad_norm": 0.4381128251552582,
      "learning_rate": 2.650752206887977e-06,
      "loss": 0.6826,
      "step": 11338
    },
    {
      "epoch": 0.4699324464337519,
      "grad_norm": 0.40064650774002075,
      "learning_rate": 2.650544987359609e-06,
      "loss": 0.6924,
      "step": 11339
    },
    {
      "epoch": 0.4699738903394256,
      "grad_norm": 0.4129202663898468,
      "learning_rate": 2.650337767831241e-06,
      "loss": 0.6779,
      "step": 11340
    },
    {
      "epoch": 0.4700153342450993,
      "grad_norm": 0.4021745026111603,
      "learning_rate": 2.650130548302872e-06,
      "loss": 0.6616,
      "step": 11341
    },
    {
      "epoch": 0.4700567781507729,
      "grad_norm": 0.423465758562088,
      "learning_rate": 2.6499233287745042e-06,
      "loss": 0.6913,
      "step": 11342
    },
    {
      "epoch": 0.4700982220564466,
      "grad_norm": 0.38998717069625854,
      "learning_rate": 2.6497161092461356e-06,
      "loss": 0.675,
      "step": 11343
    },
    {
      "epoch": 0.47013966596212026,
      "grad_norm": 0.3829076290130615,
      "learning_rate": 2.6495088897177674e-06,
      "loss": 0.7096,
      "step": 11344
    },
    {
      "epoch": 0.47018110986779393,
      "grad_norm": 0.39907750487327576,
      "learning_rate": 2.649301670189399e-06,
      "loss": 0.704,
      "step": 11345
    },
    {
      "epoch": 0.4702225537734676,
      "grad_norm": 0.4129486680030823,
      "learning_rate": 2.6490944506610306e-06,
      "loss": 0.7075,
      "step": 11346
    },
    {
      "epoch": 0.4702639976791413,
      "grad_norm": 0.40914076566696167,
      "learning_rate": 2.648887231132662e-06,
      "loss": 0.6627,
      "step": 11347
    },
    {
      "epoch": 0.47030544158481496,
      "grad_norm": 0.4120344817638397,
      "learning_rate": 2.648680011604294e-06,
      "loss": 0.7181,
      "step": 11348
    },
    {
      "epoch": 0.47034688549048864,
      "grad_norm": 0.4412282109260559,
      "learning_rate": 2.6484727920759256e-06,
      "loss": 0.7185,
      "step": 11349
    },
    {
      "epoch": 0.4703883293961623,
      "grad_norm": 0.4474334418773651,
      "learning_rate": 2.648265572547557e-06,
      "loss": 0.6941,
      "step": 11350
    },
    {
      "epoch": 0.47042977330183594,
      "grad_norm": 0.45297300815582275,
      "learning_rate": 2.648058353019189e-06,
      "loss": 0.7136,
      "step": 11351
    },
    {
      "epoch": 0.4704712172075096,
      "grad_norm": 0.38765719532966614,
      "learning_rate": 2.64785113349082e-06,
      "loss": 0.6868,
      "step": 11352
    },
    {
      "epoch": 0.4705126611131833,
      "grad_norm": 0.45698076486587524,
      "learning_rate": 2.647643913962452e-06,
      "loss": 0.6785,
      "step": 11353
    },
    {
      "epoch": 0.47055410501885697,
      "grad_norm": 0.4141662120819092,
      "learning_rate": 2.6474366944340834e-06,
      "loss": 0.652,
      "step": 11354
    },
    {
      "epoch": 0.47059554892453065,
      "grad_norm": 0.4493732154369354,
      "learning_rate": 2.647229474905715e-06,
      "loss": 0.6783,
      "step": 11355
    },
    {
      "epoch": 0.4706369928302043,
      "grad_norm": 0.43075716495513916,
      "learning_rate": 2.6470222553773474e-06,
      "loss": 0.6956,
      "step": 11356
    },
    {
      "epoch": 0.470678436735878,
      "grad_norm": 0.39770838618278503,
      "learning_rate": 2.6468150358489784e-06,
      "loss": 0.688,
      "step": 11357
    },
    {
      "epoch": 0.4707198806415517,
      "grad_norm": 0.4198416471481323,
      "learning_rate": 2.6466078163206106e-06,
      "loss": 0.7207,
      "step": 11358
    },
    {
      "epoch": 0.47076132454722536,
      "grad_norm": 0.4094206988811493,
      "learning_rate": 2.646400596792242e-06,
      "loss": 0.7051,
      "step": 11359
    },
    {
      "epoch": 0.470802768452899,
      "grad_norm": 0.4325757622718811,
      "learning_rate": 2.646193377263874e-06,
      "loss": 0.7098,
      "step": 11360
    },
    {
      "epoch": 0.47084421235857266,
      "grad_norm": 0.4404744505882263,
      "learning_rate": 2.645986157735505e-06,
      "loss": 0.6726,
      "step": 11361
    },
    {
      "epoch": 0.47088565626424633,
      "grad_norm": 0.44359090924263,
      "learning_rate": 2.645778938207137e-06,
      "loss": 0.7461,
      "step": 11362
    },
    {
      "epoch": 0.47092710016992,
      "grad_norm": 0.39852702617645264,
      "learning_rate": 2.6455717186787684e-06,
      "loss": 0.6752,
      "step": 11363
    },
    {
      "epoch": 0.4709685440755937,
      "grad_norm": 0.44236046075820923,
      "learning_rate": 2.6453644991504e-06,
      "loss": 0.6946,
      "step": 11364
    },
    {
      "epoch": 0.47100998798126736,
      "grad_norm": 0.43115130066871643,
      "learning_rate": 2.645157279622032e-06,
      "loss": 0.6604,
      "step": 11365
    },
    {
      "epoch": 0.47105143188694104,
      "grad_norm": 0.41119763255119324,
      "learning_rate": 2.6449500600936634e-06,
      "loss": 0.7151,
      "step": 11366
    },
    {
      "epoch": 0.4710928757926147,
      "grad_norm": 0.39191895723342896,
      "learning_rate": 2.6447428405652952e-06,
      "loss": 0.6604,
      "step": 11367
    },
    {
      "epoch": 0.47113431969828834,
      "grad_norm": 0.4050554633140564,
      "learning_rate": 2.6445356210369266e-06,
      "loss": 0.7427,
      "step": 11368
    },
    {
      "epoch": 0.471175763603962,
      "grad_norm": 0.3746495544910431,
      "learning_rate": 2.6443284015085584e-06,
      "loss": 0.6302,
      "step": 11369
    },
    {
      "epoch": 0.4712172075096357,
      "grad_norm": 0.4161635935306549,
      "learning_rate": 2.64412118198019e-06,
      "loss": 0.7184,
      "step": 11370
    },
    {
      "epoch": 0.47125865141530937,
      "grad_norm": 0.38502535223960876,
      "learning_rate": 2.6439139624518216e-06,
      "loss": 0.6564,
      "step": 11371
    },
    {
      "epoch": 0.47130009532098305,
      "grad_norm": 0.39421147108078003,
      "learning_rate": 2.643706742923453e-06,
      "loss": 0.6769,
      "step": 11372
    },
    {
      "epoch": 0.4713415392266567,
      "grad_norm": 0.42189276218414307,
      "learning_rate": 2.643499523395085e-06,
      "loss": 0.7478,
      "step": 11373
    },
    {
      "epoch": 0.4713829831323304,
      "grad_norm": 0.40506711602211,
      "learning_rate": 2.643292303866717e-06,
      "loss": 0.7043,
      "step": 11374
    },
    {
      "epoch": 0.4714244270380041,
      "grad_norm": 0.45878490805625916,
      "learning_rate": 2.6430850843383484e-06,
      "loss": 0.7019,
      "step": 11375
    },
    {
      "epoch": 0.47146587094367776,
      "grad_norm": 0.3727363049983978,
      "learning_rate": 2.6428778648099802e-06,
      "loss": 0.6613,
      "step": 11376
    },
    {
      "epoch": 0.4715073148493514,
      "grad_norm": 0.41939523816108704,
      "learning_rate": 2.6426706452816116e-06,
      "loss": 0.6809,
      "step": 11377
    },
    {
      "epoch": 0.47154875875502505,
      "grad_norm": 0.42505860328674316,
      "learning_rate": 2.6424634257532434e-06,
      "loss": 0.6594,
      "step": 11378
    },
    {
      "epoch": 0.47159020266069873,
      "grad_norm": 0.4066566228866577,
      "learning_rate": 2.642256206224875e-06,
      "loss": 0.6833,
      "step": 11379
    },
    {
      "epoch": 0.4716316465663724,
      "grad_norm": 0.4299336075782776,
      "learning_rate": 2.6420489866965066e-06,
      "loss": 0.7344,
      "step": 11380
    },
    {
      "epoch": 0.4716730904720461,
      "grad_norm": 0.39767220616340637,
      "learning_rate": 2.641841767168138e-06,
      "loss": 0.6564,
      "step": 11381
    },
    {
      "epoch": 0.47171453437771976,
      "grad_norm": 0.41492271423339844,
      "learning_rate": 2.64163454763977e-06,
      "loss": 0.6855,
      "step": 11382
    },
    {
      "epoch": 0.47175597828339344,
      "grad_norm": 0.4342914819717407,
      "learning_rate": 2.6414273281114016e-06,
      "loss": 0.6885,
      "step": 11383
    },
    {
      "epoch": 0.4717974221890671,
      "grad_norm": 0.41960659623146057,
      "learning_rate": 2.641220108583033e-06,
      "loss": 0.6648,
      "step": 11384
    },
    {
      "epoch": 0.4718388660947408,
      "grad_norm": 0.40715715289115906,
      "learning_rate": 2.641012889054665e-06,
      "loss": 0.6924,
      "step": 11385
    },
    {
      "epoch": 0.4718803100004144,
      "grad_norm": 0.4295501708984375,
      "learning_rate": 2.640805669526296e-06,
      "loss": 0.705,
      "step": 11386
    },
    {
      "epoch": 0.4719217539060881,
      "grad_norm": 0.4021407663822174,
      "learning_rate": 2.640598449997928e-06,
      "loss": 0.6931,
      "step": 11387
    },
    {
      "epoch": 0.47196319781176177,
      "grad_norm": 0.4279367923736572,
      "learning_rate": 2.6403912304695594e-06,
      "loss": 0.7128,
      "step": 11388
    },
    {
      "epoch": 0.47200464171743545,
      "grad_norm": 0.4057573080062866,
      "learning_rate": 2.640184010941191e-06,
      "loss": 0.6875,
      "step": 11389
    },
    {
      "epoch": 0.4720460856231091,
      "grad_norm": 0.3977506458759308,
      "learning_rate": 2.6399767914128226e-06,
      "loss": 0.7109,
      "step": 11390
    },
    {
      "epoch": 0.4720875295287828,
      "grad_norm": 0.4762437641620636,
      "learning_rate": 2.6397695718844544e-06,
      "loss": 0.7216,
      "step": 11391
    },
    {
      "epoch": 0.4721289734344565,
      "grad_norm": 0.3883860409259796,
      "learning_rate": 2.6395623523560866e-06,
      "loss": 0.6344,
      "step": 11392
    },
    {
      "epoch": 0.47217041734013016,
      "grad_norm": 0.3890821933746338,
      "learning_rate": 2.639355132827718e-06,
      "loss": 0.7249,
      "step": 11393
    },
    {
      "epoch": 0.47221186124580383,
      "grad_norm": 0.4166756570339203,
      "learning_rate": 2.63914791329935e-06,
      "loss": 0.7306,
      "step": 11394
    },
    {
      "epoch": 0.47225330515147745,
      "grad_norm": 0.40263593196868896,
      "learning_rate": 2.638940693770981e-06,
      "loss": 0.7439,
      "step": 11395
    },
    {
      "epoch": 0.47229474905715113,
      "grad_norm": 0.4387344717979431,
      "learning_rate": 2.638733474242613e-06,
      "loss": 0.679,
      "step": 11396
    },
    {
      "epoch": 0.4723361929628248,
      "grad_norm": 0.43287423253059387,
      "learning_rate": 2.6385262547142444e-06,
      "loss": 0.6836,
      "step": 11397
    },
    {
      "epoch": 0.4723776368684985,
      "grad_norm": 0.4239111542701721,
      "learning_rate": 2.6383190351858762e-06,
      "loss": 0.6729,
      "step": 11398
    },
    {
      "epoch": 0.47241908077417216,
      "grad_norm": 0.43029096722602844,
      "learning_rate": 2.6381118156575076e-06,
      "loss": 0.6935,
      "step": 11399
    },
    {
      "epoch": 0.47246052467984584,
      "grad_norm": 0.4543099105358124,
      "learning_rate": 2.6379045961291394e-06,
      "loss": 0.6943,
      "step": 11400
    },
    {
      "epoch": 0.4725019685855195,
      "grad_norm": 0.38043293356895447,
      "learning_rate": 2.6376973766007712e-06,
      "loss": 0.6678,
      "step": 11401
    },
    {
      "epoch": 0.4725434124911932,
      "grad_norm": 0.43285271525382996,
      "learning_rate": 2.6374901570724026e-06,
      "loss": 0.7224,
      "step": 11402
    },
    {
      "epoch": 0.4725848563968668,
      "grad_norm": 0.3954186737537384,
      "learning_rate": 2.6372829375440344e-06,
      "loss": 0.6588,
      "step": 11403
    },
    {
      "epoch": 0.4726263003025405,
      "grad_norm": 0.42266973853111267,
      "learning_rate": 2.637075718015666e-06,
      "loss": 0.7273,
      "step": 11404
    },
    {
      "epoch": 0.47266774420821417,
      "grad_norm": 0.41446173191070557,
      "learning_rate": 2.6368684984872976e-06,
      "loss": 0.7228,
      "step": 11405
    },
    {
      "epoch": 0.47270918811388785,
      "grad_norm": 0.39462971687316895,
      "learning_rate": 2.636661278958929e-06,
      "loss": 0.7029,
      "step": 11406
    },
    {
      "epoch": 0.4727506320195615,
      "grad_norm": 0.5781558752059937,
      "learning_rate": 2.636454059430561e-06,
      "loss": 0.761,
      "step": 11407
    },
    {
      "epoch": 0.4727920759252352,
      "grad_norm": 0.4346928894519806,
      "learning_rate": 2.636246839902192e-06,
      "loss": 0.6896,
      "step": 11408
    },
    {
      "epoch": 0.4728335198309089,
      "grad_norm": 0.3628773093223572,
      "learning_rate": 2.6360396203738244e-06,
      "loss": 0.6316,
      "step": 11409
    },
    {
      "epoch": 0.47287496373658255,
      "grad_norm": 0.42471399903297424,
      "learning_rate": 2.6358324008454562e-06,
      "loss": 0.6713,
      "step": 11410
    },
    {
      "epoch": 0.47291640764225623,
      "grad_norm": 0.4325403869152069,
      "learning_rate": 2.6356251813170876e-06,
      "loss": 0.697,
      "step": 11411
    },
    {
      "epoch": 0.47295785154792985,
      "grad_norm": 0.4441758394241333,
      "learning_rate": 2.6354179617887194e-06,
      "loss": 0.7239,
      "step": 11412
    },
    {
      "epoch": 0.47299929545360353,
      "grad_norm": 0.41379514336586,
      "learning_rate": 2.635210742260351e-06,
      "loss": 0.7007,
      "step": 11413
    },
    {
      "epoch": 0.4730407393592772,
      "grad_norm": 0.41520535945892334,
      "learning_rate": 2.6350035227319826e-06,
      "loss": 0.6503,
      "step": 11414
    },
    {
      "epoch": 0.4730821832649509,
      "grad_norm": 0.3927014470100403,
      "learning_rate": 2.634796303203614e-06,
      "loss": 0.6407,
      "step": 11415
    },
    {
      "epoch": 0.47312362717062456,
      "grad_norm": 0.4386177062988281,
      "learning_rate": 2.634589083675246e-06,
      "loss": 0.7126,
      "step": 11416
    },
    {
      "epoch": 0.47316507107629824,
      "grad_norm": 0.45869630575180054,
      "learning_rate": 2.6343818641468776e-06,
      "loss": 0.7122,
      "step": 11417
    },
    {
      "epoch": 0.4732065149819719,
      "grad_norm": 0.44832679629325867,
      "learning_rate": 2.634174644618509e-06,
      "loss": 0.7068,
      "step": 11418
    },
    {
      "epoch": 0.4732479588876456,
      "grad_norm": 0.42184901237487793,
      "learning_rate": 2.633967425090141e-06,
      "loss": 0.71,
      "step": 11419
    },
    {
      "epoch": 0.47328940279331927,
      "grad_norm": 0.3978688418865204,
      "learning_rate": 2.633760205561772e-06,
      "loss": 0.7041,
      "step": 11420
    },
    {
      "epoch": 0.4733308466989929,
      "grad_norm": 0.4360009729862213,
      "learning_rate": 2.633552986033404e-06,
      "loss": 0.6682,
      "step": 11421
    },
    {
      "epoch": 0.47337229060466657,
      "grad_norm": 0.3902774155139923,
      "learning_rate": 2.6333457665050354e-06,
      "loss": 0.7019,
      "step": 11422
    },
    {
      "epoch": 0.47341373451034024,
      "grad_norm": 0.44248345494270325,
      "learning_rate": 2.633138546976667e-06,
      "loss": 0.7135,
      "step": 11423
    },
    {
      "epoch": 0.4734551784160139,
      "grad_norm": 0.4178068935871124,
      "learning_rate": 2.6329313274482986e-06,
      "loss": 0.6536,
      "step": 11424
    },
    {
      "epoch": 0.4734966223216876,
      "grad_norm": 0.43097278475761414,
      "learning_rate": 2.6327241079199304e-06,
      "loss": 0.6677,
      "step": 11425
    },
    {
      "epoch": 0.4735380662273613,
      "grad_norm": 0.3958195447921753,
      "learning_rate": 2.6325168883915626e-06,
      "loss": 0.6882,
      "step": 11426
    },
    {
      "epoch": 0.47357951013303495,
      "grad_norm": 0.40992850065231323,
      "learning_rate": 2.632309668863194e-06,
      "loss": 0.6515,
      "step": 11427
    },
    {
      "epoch": 0.47362095403870863,
      "grad_norm": 0.44090375304222107,
      "learning_rate": 2.632102449334826e-06,
      "loss": 0.7048,
      "step": 11428
    },
    {
      "epoch": 0.47366239794438225,
      "grad_norm": 0.43152981996536255,
      "learning_rate": 2.6318952298064572e-06,
      "loss": 0.7144,
      "step": 11429
    },
    {
      "epoch": 0.47370384185005593,
      "grad_norm": 0.40817901492118835,
      "learning_rate": 2.631688010278089e-06,
      "loss": 0.6611,
      "step": 11430
    },
    {
      "epoch": 0.4737452857557296,
      "grad_norm": 0.4613114595413208,
      "learning_rate": 2.6314807907497204e-06,
      "loss": 0.7307,
      "step": 11431
    },
    {
      "epoch": 0.4737867296614033,
      "grad_norm": 0.41390907764434814,
      "learning_rate": 2.6312735712213522e-06,
      "loss": 0.703,
      "step": 11432
    },
    {
      "epoch": 0.47382817356707696,
      "grad_norm": 0.4283941090106964,
      "learning_rate": 2.6310663516929836e-06,
      "loss": 0.698,
      "step": 11433
    },
    {
      "epoch": 0.47386961747275064,
      "grad_norm": 0.4182806611061096,
      "learning_rate": 2.6308591321646154e-06,
      "loss": 0.6571,
      "step": 11434
    },
    {
      "epoch": 0.4739110613784243,
      "grad_norm": 0.43244579434394836,
      "learning_rate": 2.6306519126362472e-06,
      "loss": 0.7693,
      "step": 11435
    },
    {
      "epoch": 0.473952505284098,
      "grad_norm": 0.4175407886505127,
      "learning_rate": 2.6304446931078786e-06,
      "loss": 0.723,
      "step": 11436
    },
    {
      "epoch": 0.47399394918977167,
      "grad_norm": 0.36724182963371277,
      "learning_rate": 2.6302374735795104e-06,
      "loss": 0.667,
      "step": 11437
    },
    {
      "epoch": 0.4740353930954453,
      "grad_norm": 0.4450782239437103,
      "learning_rate": 2.630030254051142e-06,
      "loss": 0.6901,
      "step": 11438
    },
    {
      "epoch": 0.47407683700111897,
      "grad_norm": 0.42237043380737305,
      "learning_rate": 2.6298230345227736e-06,
      "loss": 0.7263,
      "step": 11439
    },
    {
      "epoch": 0.47411828090679264,
      "grad_norm": 0.42181065678596497,
      "learning_rate": 2.629615814994405e-06,
      "loss": 0.6958,
      "step": 11440
    },
    {
      "epoch": 0.4741597248124663,
      "grad_norm": 0.3753145933151245,
      "learning_rate": 2.629408595466037e-06,
      "loss": 0.6858,
      "step": 11441
    },
    {
      "epoch": 0.47420116871814,
      "grad_norm": 0.4241846203804016,
      "learning_rate": 2.629201375937668e-06,
      "loss": 0.719,
      "step": 11442
    },
    {
      "epoch": 0.4742426126238137,
      "grad_norm": 0.3956572711467743,
      "learning_rate": 2.6289941564093004e-06,
      "loss": 0.689,
      "step": 11443
    },
    {
      "epoch": 0.47428405652948735,
      "grad_norm": 0.43075209856033325,
      "learning_rate": 2.6287869368809322e-06,
      "loss": 0.6636,
      "step": 11444
    },
    {
      "epoch": 0.47432550043516103,
      "grad_norm": 0.42895960807800293,
      "learning_rate": 2.6285797173525636e-06,
      "loss": 0.7026,
      "step": 11445
    },
    {
      "epoch": 0.4743669443408347,
      "grad_norm": 0.4139692187309265,
      "learning_rate": 2.6283724978241954e-06,
      "loss": 0.6531,
      "step": 11446
    },
    {
      "epoch": 0.47440838824650833,
      "grad_norm": 0.40550312399864197,
      "learning_rate": 2.628165278295827e-06,
      "loss": 0.7053,
      "step": 11447
    },
    {
      "epoch": 0.474449832152182,
      "grad_norm": 0.4048577845096588,
      "learning_rate": 2.6279580587674586e-06,
      "loss": 0.7164,
      "step": 11448
    },
    {
      "epoch": 0.4744912760578557,
      "grad_norm": 0.40656283497810364,
      "learning_rate": 2.62775083923909e-06,
      "loss": 0.7125,
      "step": 11449
    },
    {
      "epoch": 0.47453271996352936,
      "grad_norm": 0.4224267899990082,
      "learning_rate": 2.627543619710722e-06,
      "loss": 0.6992,
      "step": 11450
    },
    {
      "epoch": 0.47457416386920304,
      "grad_norm": 0.4325571060180664,
      "learning_rate": 2.627336400182353e-06,
      "loss": 0.6758,
      "step": 11451
    },
    {
      "epoch": 0.4746156077748767,
      "grad_norm": 0.45373156666755676,
      "learning_rate": 2.627129180653985e-06,
      "loss": 0.7144,
      "step": 11452
    },
    {
      "epoch": 0.4746570516805504,
      "grad_norm": 0.4058187007904053,
      "learning_rate": 2.626921961125617e-06,
      "loss": 0.6714,
      "step": 11453
    },
    {
      "epoch": 0.47469849558622407,
      "grad_norm": 0.4216562807559967,
      "learning_rate": 2.626714741597248e-06,
      "loss": 0.6478,
      "step": 11454
    },
    {
      "epoch": 0.4747399394918977,
      "grad_norm": 0.3842514455318451,
      "learning_rate": 2.62650752206888e-06,
      "loss": 0.6758,
      "step": 11455
    },
    {
      "epoch": 0.47478138339757137,
      "grad_norm": 0.41620180010795593,
      "learning_rate": 2.6263003025405114e-06,
      "loss": 0.7096,
      "step": 11456
    },
    {
      "epoch": 0.47482282730324504,
      "grad_norm": 0.40963149070739746,
      "learning_rate": 2.6260930830121432e-06,
      "loss": 0.6937,
      "step": 11457
    },
    {
      "epoch": 0.4748642712089187,
      "grad_norm": 0.4420025944709778,
      "learning_rate": 2.6258858634837746e-06,
      "loss": 0.699,
      "step": 11458
    },
    {
      "epoch": 0.4749057151145924,
      "grad_norm": 0.42657753825187683,
      "learning_rate": 2.6256786439554064e-06,
      "loss": 0.7046,
      "step": 11459
    },
    {
      "epoch": 0.4749471590202661,
      "grad_norm": 0.4237903952598572,
      "learning_rate": 2.625471424427038e-06,
      "loss": 0.6451,
      "step": 11460
    },
    {
      "epoch": 0.47498860292593975,
      "grad_norm": 0.41626524925231934,
      "learning_rate": 2.62526420489867e-06,
      "loss": 0.6755,
      "step": 11461
    },
    {
      "epoch": 0.47503004683161343,
      "grad_norm": 0.4227791130542755,
      "learning_rate": 2.625056985370302e-06,
      "loss": 0.6653,
      "step": 11462
    },
    {
      "epoch": 0.4750714907372871,
      "grad_norm": 0.4050874710083008,
      "learning_rate": 2.6248497658419332e-06,
      "loss": 0.6952,
      "step": 11463
    },
    {
      "epoch": 0.4751129346429607,
      "grad_norm": 0.3859892189502716,
      "learning_rate": 2.624642546313565e-06,
      "loss": 0.7024,
      "step": 11464
    },
    {
      "epoch": 0.4751543785486344,
      "grad_norm": 0.4008544385433197,
      "learning_rate": 2.6244353267851964e-06,
      "loss": 0.7363,
      "step": 11465
    },
    {
      "epoch": 0.4751958224543081,
      "grad_norm": 0.4063122272491455,
      "learning_rate": 2.6242281072568282e-06,
      "loss": 0.6624,
      "step": 11466
    },
    {
      "epoch": 0.47523726635998176,
      "grad_norm": 0.39847856760025024,
      "learning_rate": 2.6240208877284596e-06,
      "loss": 0.6963,
      "step": 11467
    },
    {
      "epoch": 0.47527871026565544,
      "grad_norm": 0.41683730483055115,
      "learning_rate": 2.6238136682000914e-06,
      "loss": 0.6912,
      "step": 11468
    },
    {
      "epoch": 0.4753201541713291,
      "grad_norm": 0.41590383648872375,
      "learning_rate": 2.623606448671723e-06,
      "loss": 0.7046,
      "step": 11469
    },
    {
      "epoch": 0.4753615980770028,
      "grad_norm": 0.40955421328544617,
      "learning_rate": 2.6233992291433546e-06,
      "loss": 0.7017,
      "step": 11470
    },
    {
      "epoch": 0.47540304198267647,
      "grad_norm": 0.4093576669692993,
      "learning_rate": 2.6231920096149864e-06,
      "loss": 0.6967,
      "step": 11471
    },
    {
      "epoch": 0.47544448588835014,
      "grad_norm": 0.4248949885368347,
      "learning_rate": 2.622984790086618e-06,
      "loss": 0.6453,
      "step": 11472
    },
    {
      "epoch": 0.47548592979402377,
      "grad_norm": 0.39768165349960327,
      "learning_rate": 2.6227775705582496e-06,
      "loss": 0.703,
      "step": 11473
    },
    {
      "epoch": 0.47552737369969744,
      "grad_norm": 0.4450720548629761,
      "learning_rate": 2.622570351029881e-06,
      "loss": 0.7585,
      "step": 11474
    },
    {
      "epoch": 0.4755688176053711,
      "grad_norm": 0.3806042969226837,
      "learning_rate": 2.622363131501513e-06,
      "loss": 0.6703,
      "step": 11475
    },
    {
      "epoch": 0.4756102615110448,
      "grad_norm": 0.39418166875839233,
      "learning_rate": 2.622155911973144e-06,
      "loss": 0.687,
      "step": 11476
    },
    {
      "epoch": 0.4756517054167185,
      "grad_norm": 0.39723825454711914,
      "learning_rate": 2.6219486924447764e-06,
      "loss": 0.6663,
      "step": 11477
    },
    {
      "epoch": 0.47569314932239215,
      "grad_norm": 0.4337148368358612,
      "learning_rate": 2.6217414729164082e-06,
      "loss": 0.7588,
      "step": 11478
    },
    {
      "epoch": 0.4757345932280658,
      "grad_norm": 0.4224752187728882,
      "learning_rate": 2.6215342533880396e-06,
      "loss": 0.6997,
      "step": 11479
    },
    {
      "epoch": 0.4757760371337395,
      "grad_norm": 0.4324043095111847,
      "learning_rate": 2.6213270338596714e-06,
      "loss": 0.7441,
      "step": 11480
    },
    {
      "epoch": 0.4758174810394132,
      "grad_norm": 0.4136542081832886,
      "learning_rate": 2.621119814331303e-06,
      "loss": 0.6804,
      "step": 11481
    },
    {
      "epoch": 0.4758589249450868,
      "grad_norm": 0.3994368612766266,
      "learning_rate": 2.6209125948029346e-06,
      "loss": 0.6791,
      "step": 11482
    },
    {
      "epoch": 0.4759003688507605,
      "grad_norm": 0.384559690952301,
      "learning_rate": 2.620705375274566e-06,
      "loss": 0.6587,
      "step": 11483
    },
    {
      "epoch": 0.47594181275643416,
      "grad_norm": 0.42351388931274414,
      "learning_rate": 2.620498155746198e-06,
      "loss": 0.7024,
      "step": 11484
    },
    {
      "epoch": 0.47598325666210783,
      "grad_norm": 0.468805730342865,
      "learning_rate": 2.6202909362178292e-06,
      "loss": 0.7351,
      "step": 11485
    },
    {
      "epoch": 0.4760247005677815,
      "grad_norm": 0.41827571392059326,
      "learning_rate": 2.620083716689461e-06,
      "loss": 0.658,
      "step": 11486
    },
    {
      "epoch": 0.4760661444734552,
      "grad_norm": 0.41353392601013184,
      "learning_rate": 2.619876497161093e-06,
      "loss": 0.6924,
      "step": 11487
    },
    {
      "epoch": 0.47610758837912887,
      "grad_norm": 0.4172934889793396,
      "learning_rate": 2.6196692776327242e-06,
      "loss": 0.6514,
      "step": 11488
    },
    {
      "epoch": 0.47614903228480254,
      "grad_norm": 0.40974968671798706,
      "learning_rate": 2.619462058104356e-06,
      "loss": 0.6882,
      "step": 11489
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 0.4307625889778137,
      "learning_rate": 2.6192548385759874e-06,
      "loss": 0.6771,
      "step": 11490
    },
    {
      "epoch": 0.47623192009614984,
      "grad_norm": 0.4092712700366974,
      "learning_rate": 2.6190476190476192e-06,
      "loss": 0.7092,
      "step": 11491
    },
    {
      "epoch": 0.4762733640018235,
      "grad_norm": 0.41111546754837036,
      "learning_rate": 2.6188403995192506e-06,
      "loss": 0.7383,
      "step": 11492
    },
    {
      "epoch": 0.4763148079074972,
      "grad_norm": 0.4058970510959625,
      "learning_rate": 2.6186331799908824e-06,
      "loss": 0.6741,
      "step": 11493
    },
    {
      "epoch": 0.4763562518131709,
      "grad_norm": 0.394513875246048,
      "learning_rate": 2.618425960462514e-06,
      "loss": 0.7212,
      "step": 11494
    },
    {
      "epoch": 0.47639769571884455,
      "grad_norm": 0.3905352056026459,
      "learning_rate": 2.618218740934146e-06,
      "loss": 0.677,
      "step": 11495
    },
    {
      "epoch": 0.4764391396245182,
      "grad_norm": 0.4193161427974701,
      "learning_rate": 2.618011521405778e-06,
      "loss": 0.6306,
      "step": 11496
    },
    {
      "epoch": 0.4764805835301919,
      "grad_norm": 0.39033299684524536,
      "learning_rate": 2.6178043018774092e-06,
      "loss": 0.6677,
      "step": 11497
    },
    {
      "epoch": 0.4765220274358656,
      "grad_norm": 0.3951054811477661,
      "learning_rate": 2.617597082349041e-06,
      "loss": 0.7228,
      "step": 11498
    },
    {
      "epoch": 0.4765634713415392,
      "grad_norm": 0.4232650101184845,
      "learning_rate": 2.6173898628206724e-06,
      "loss": 0.6946,
      "step": 11499
    },
    {
      "epoch": 0.4766049152472129,
      "grad_norm": 0.44773146510124207,
      "learning_rate": 2.6171826432923042e-06,
      "loss": 0.702,
      "step": 11500
    },
    {
      "epoch": 0.47664635915288656,
      "grad_norm": 0.42612895369529724,
      "learning_rate": 2.6169754237639356e-06,
      "loss": 0.6279,
      "step": 11501
    },
    {
      "epoch": 0.47668780305856023,
      "grad_norm": 0.42460209131240845,
      "learning_rate": 2.6167682042355674e-06,
      "loss": 0.7095,
      "step": 11502
    },
    {
      "epoch": 0.4767292469642339,
      "grad_norm": 0.4598274528980255,
      "learning_rate": 2.616560984707199e-06,
      "loss": 0.7108,
      "step": 11503
    },
    {
      "epoch": 0.4767706908699076,
      "grad_norm": 0.422524631023407,
      "learning_rate": 2.6163537651788306e-06,
      "loss": 0.7527,
      "step": 11504
    },
    {
      "epoch": 0.47681213477558126,
      "grad_norm": 0.388109028339386,
      "learning_rate": 2.6161465456504624e-06,
      "loss": 0.6042,
      "step": 11505
    },
    {
      "epoch": 0.47685357868125494,
      "grad_norm": 0.4475538432598114,
      "learning_rate": 2.615939326122094e-06,
      "loss": 0.6968,
      "step": 11506
    },
    {
      "epoch": 0.4768950225869286,
      "grad_norm": 0.3924456834793091,
      "learning_rate": 2.6157321065937256e-06,
      "loss": 0.7114,
      "step": 11507
    },
    {
      "epoch": 0.47693646649260224,
      "grad_norm": 0.42523661255836487,
      "learning_rate": 2.615524887065357e-06,
      "loss": 0.7,
      "step": 11508
    },
    {
      "epoch": 0.4769779103982759,
      "grad_norm": 0.4303314983844757,
      "learning_rate": 2.615317667536989e-06,
      "loss": 0.7148,
      "step": 11509
    },
    {
      "epoch": 0.4770193543039496,
      "grad_norm": 0.3966044783592224,
      "learning_rate": 2.61511044800862e-06,
      "loss": 0.6282,
      "step": 11510
    },
    {
      "epoch": 0.47706079820962327,
      "grad_norm": 0.39889660477638245,
      "learning_rate": 2.6149032284802524e-06,
      "loss": 0.6777,
      "step": 11511
    },
    {
      "epoch": 0.47710224211529695,
      "grad_norm": 0.44521674513816833,
      "learning_rate": 2.6146960089518834e-06,
      "loss": 0.7261,
      "step": 11512
    },
    {
      "epoch": 0.4771436860209706,
      "grad_norm": 0.41493889689445496,
      "learning_rate": 2.6144887894235156e-06,
      "loss": 0.6982,
      "step": 11513
    },
    {
      "epoch": 0.4771851299266443,
      "grad_norm": 0.42919066548347473,
      "learning_rate": 2.6142815698951474e-06,
      "loss": 0.6696,
      "step": 11514
    },
    {
      "epoch": 0.477226573832318,
      "grad_norm": 0.4288454055786133,
      "learning_rate": 2.614074350366779e-06,
      "loss": 0.678,
      "step": 11515
    },
    {
      "epoch": 0.4772680177379916,
      "grad_norm": 0.47752150893211365,
      "learning_rate": 2.6138671308384106e-06,
      "loss": 0.7003,
      "step": 11516
    },
    {
      "epoch": 0.4773094616436653,
      "grad_norm": 0.46381378173828125,
      "learning_rate": 2.613659911310042e-06,
      "loss": 0.7012,
      "step": 11517
    },
    {
      "epoch": 0.47735090554933896,
      "grad_norm": 0.38335275650024414,
      "learning_rate": 2.613452691781674e-06,
      "loss": 0.6526,
      "step": 11518
    },
    {
      "epoch": 0.47739234945501263,
      "grad_norm": 0.40139323472976685,
      "learning_rate": 2.6132454722533052e-06,
      "loss": 0.7058,
      "step": 11519
    },
    {
      "epoch": 0.4774337933606863,
      "grad_norm": 0.3763595521450043,
      "learning_rate": 2.613038252724937e-06,
      "loss": 0.6389,
      "step": 11520
    },
    {
      "epoch": 0.47747523726636,
      "grad_norm": 0.424081951379776,
      "learning_rate": 2.6128310331965684e-06,
      "loss": 0.752,
      "step": 11521
    },
    {
      "epoch": 0.47751668117203366,
      "grad_norm": 0.39727458357810974,
      "learning_rate": 2.6126238136682002e-06,
      "loss": 0.7297,
      "step": 11522
    },
    {
      "epoch": 0.47755812507770734,
      "grad_norm": 0.42372527718544006,
      "learning_rate": 2.612416594139832e-06,
      "loss": 0.6857,
      "step": 11523
    },
    {
      "epoch": 0.477599568983381,
      "grad_norm": 0.4009189307689667,
      "learning_rate": 2.6122093746114634e-06,
      "loss": 0.6145,
      "step": 11524
    },
    {
      "epoch": 0.47764101288905464,
      "grad_norm": 0.40530744194984436,
      "learning_rate": 2.6120021550830952e-06,
      "loss": 0.7256,
      "step": 11525
    },
    {
      "epoch": 0.4776824567947283,
      "grad_norm": 0.40740352869033813,
      "learning_rate": 2.6117949355547266e-06,
      "loss": 0.6549,
      "step": 11526
    },
    {
      "epoch": 0.477723900700402,
      "grad_norm": 0.3994278907775879,
      "learning_rate": 2.6115877160263584e-06,
      "loss": 0.6532,
      "step": 11527
    },
    {
      "epoch": 0.47776534460607567,
      "grad_norm": 0.4334213137626648,
      "learning_rate": 2.61138049649799e-06,
      "loss": 0.7283,
      "step": 11528
    },
    {
      "epoch": 0.47780678851174935,
      "grad_norm": 0.4287005066871643,
      "learning_rate": 2.611173276969622e-06,
      "loss": 0.6858,
      "step": 11529
    },
    {
      "epoch": 0.477848232417423,
      "grad_norm": 0.43473875522613525,
      "learning_rate": 2.610966057441254e-06,
      "loss": 0.698,
      "step": 11530
    },
    {
      "epoch": 0.4778896763230967,
      "grad_norm": 0.41809961199760437,
      "learning_rate": 2.6107588379128852e-06,
      "loss": 0.7327,
      "step": 11531
    },
    {
      "epoch": 0.4779311202287704,
      "grad_norm": 0.4139997363090515,
      "learning_rate": 2.610551618384517e-06,
      "loss": 0.7046,
      "step": 11532
    },
    {
      "epoch": 0.47797256413444406,
      "grad_norm": 0.40067538619041443,
      "learning_rate": 2.6103443988561484e-06,
      "loss": 0.6567,
      "step": 11533
    },
    {
      "epoch": 0.4780140080401177,
      "grad_norm": 0.4183250665664673,
      "learning_rate": 2.6101371793277802e-06,
      "loss": 0.6875,
      "step": 11534
    },
    {
      "epoch": 0.47805545194579135,
      "grad_norm": 0.4032343626022339,
      "learning_rate": 2.6099299597994116e-06,
      "loss": 0.6714,
      "step": 11535
    },
    {
      "epoch": 0.47809689585146503,
      "grad_norm": 0.3915208876132965,
      "learning_rate": 2.6097227402710434e-06,
      "loss": 0.7112,
      "step": 11536
    },
    {
      "epoch": 0.4781383397571387,
      "grad_norm": 0.40662887692451477,
      "learning_rate": 2.609515520742675e-06,
      "loss": 0.7385,
      "step": 11537
    },
    {
      "epoch": 0.4781797836628124,
      "grad_norm": 0.41005921363830566,
      "learning_rate": 2.6093083012143066e-06,
      "loss": 0.6675,
      "step": 11538
    },
    {
      "epoch": 0.47822122756848606,
      "grad_norm": 0.4286688268184662,
      "learning_rate": 2.6091010816859384e-06,
      "loss": 0.7927,
      "step": 11539
    },
    {
      "epoch": 0.47826267147415974,
      "grad_norm": 0.3635653257369995,
      "learning_rate": 2.60889386215757e-06,
      "loss": 0.6177,
      "step": 11540
    },
    {
      "epoch": 0.4783041153798334,
      "grad_norm": 0.41436442732810974,
      "learning_rate": 2.6086866426292016e-06,
      "loss": 0.6978,
      "step": 11541
    },
    {
      "epoch": 0.4783455592855071,
      "grad_norm": 0.40024450421333313,
      "learning_rate": 2.608479423100833e-06,
      "loss": 0.6677,
      "step": 11542
    },
    {
      "epoch": 0.4783870031911807,
      "grad_norm": 0.3978344798088074,
      "learning_rate": 2.608272203572465e-06,
      "loss": 0.7074,
      "step": 11543
    },
    {
      "epoch": 0.4784284470968544,
      "grad_norm": 0.3991069197654724,
      "learning_rate": 2.6080649840440962e-06,
      "loss": 0.7031,
      "step": 11544
    },
    {
      "epoch": 0.47846989100252807,
      "grad_norm": 0.4107877016067505,
      "learning_rate": 2.6078577645157284e-06,
      "loss": 0.6694,
      "step": 11545
    },
    {
      "epoch": 0.47851133490820175,
      "grad_norm": 0.4001396596431732,
      "learning_rate": 2.6076505449873594e-06,
      "loss": 0.6569,
      "step": 11546
    },
    {
      "epoch": 0.4785527788138754,
      "grad_norm": 0.429826557636261,
      "learning_rate": 2.6074433254589916e-06,
      "loss": 0.6931,
      "step": 11547
    },
    {
      "epoch": 0.4785942227195491,
      "grad_norm": 0.4147481322288513,
      "learning_rate": 2.6072361059306235e-06,
      "loss": 0.7114,
      "step": 11548
    },
    {
      "epoch": 0.4786356666252228,
      "grad_norm": 0.4371446371078491,
      "learning_rate": 2.607028886402255e-06,
      "loss": 0.6787,
      "step": 11549
    },
    {
      "epoch": 0.47867711053089645,
      "grad_norm": 0.45762690901756287,
      "learning_rate": 2.6068216668738866e-06,
      "loss": 0.7488,
      "step": 11550
    },
    {
      "epoch": 0.4787185544365701,
      "grad_norm": 0.4165171980857849,
      "learning_rate": 2.606614447345518e-06,
      "loss": 0.7351,
      "step": 11551
    },
    {
      "epoch": 0.47875999834224375,
      "grad_norm": 0.4067695140838623,
      "learning_rate": 2.60640722781715e-06,
      "loss": 0.6504,
      "step": 11552
    },
    {
      "epoch": 0.47880144224791743,
      "grad_norm": 0.4237878620624542,
      "learning_rate": 2.6062000082887812e-06,
      "loss": 0.6578,
      "step": 11553
    },
    {
      "epoch": 0.4788428861535911,
      "grad_norm": 0.3869917690753937,
      "learning_rate": 2.605992788760413e-06,
      "loss": 0.7045,
      "step": 11554
    },
    {
      "epoch": 0.4788843300592648,
      "grad_norm": 0.4245646893978119,
      "learning_rate": 2.6057855692320444e-06,
      "loss": 0.7065,
      "step": 11555
    },
    {
      "epoch": 0.47892577396493846,
      "grad_norm": 0.4330706298351288,
      "learning_rate": 2.6055783497036762e-06,
      "loss": 0.6816,
      "step": 11556
    },
    {
      "epoch": 0.47896721787061214,
      "grad_norm": 0.39767351746559143,
      "learning_rate": 2.605371130175308e-06,
      "loss": 0.7034,
      "step": 11557
    },
    {
      "epoch": 0.4790086617762858,
      "grad_norm": 0.40080204606056213,
      "learning_rate": 2.6051639106469394e-06,
      "loss": 0.6189,
      "step": 11558
    },
    {
      "epoch": 0.4790501056819595,
      "grad_norm": 0.4052805006504059,
      "learning_rate": 2.6049566911185712e-06,
      "loss": 0.6616,
      "step": 11559
    },
    {
      "epoch": 0.4790915495876331,
      "grad_norm": 0.38503092527389526,
      "learning_rate": 2.6047494715902026e-06,
      "loss": 0.6852,
      "step": 11560
    },
    {
      "epoch": 0.4791329934933068,
      "grad_norm": 0.4146961271762848,
      "learning_rate": 2.604542252061835e-06,
      "loss": 0.6802,
      "step": 11561
    },
    {
      "epoch": 0.47917443739898047,
      "grad_norm": 0.38257667422294617,
      "learning_rate": 2.604335032533466e-06,
      "loss": 0.655,
      "step": 11562
    },
    {
      "epoch": 0.47921588130465415,
      "grad_norm": 0.39498648047447205,
      "learning_rate": 2.604127813005098e-06,
      "loss": 0.6709,
      "step": 11563
    },
    {
      "epoch": 0.4792573252103278,
      "grad_norm": 0.4321293532848358,
      "learning_rate": 2.603920593476729e-06,
      "loss": 0.7101,
      "step": 11564
    },
    {
      "epoch": 0.4792987691160015,
      "grad_norm": 0.44235503673553467,
      "learning_rate": 2.6037133739483612e-06,
      "loss": 0.7268,
      "step": 11565
    },
    {
      "epoch": 0.4793402130216752,
      "grad_norm": 0.4245663583278656,
      "learning_rate": 2.603506154419993e-06,
      "loss": 0.7089,
      "step": 11566
    },
    {
      "epoch": 0.47938165692734885,
      "grad_norm": 0.43888047337532043,
      "learning_rate": 2.6032989348916244e-06,
      "loss": 0.7197,
      "step": 11567
    },
    {
      "epoch": 0.47942310083302253,
      "grad_norm": 0.4368878901004791,
      "learning_rate": 2.6030917153632562e-06,
      "loss": 0.7214,
      "step": 11568
    },
    {
      "epoch": 0.47946454473869615,
      "grad_norm": 0.3895343244075775,
      "learning_rate": 2.6028844958348876e-06,
      "loss": 0.668,
      "step": 11569
    },
    {
      "epoch": 0.47950598864436983,
      "grad_norm": 0.41012686491012573,
      "learning_rate": 2.6026772763065194e-06,
      "loss": 0.6682,
      "step": 11570
    },
    {
      "epoch": 0.4795474325500435,
      "grad_norm": 0.41477519273757935,
      "learning_rate": 2.602470056778151e-06,
      "loss": 0.7085,
      "step": 11571
    },
    {
      "epoch": 0.4795888764557172,
      "grad_norm": 0.4126252233982086,
      "learning_rate": 2.6022628372497826e-06,
      "loss": 0.6868,
      "step": 11572
    },
    {
      "epoch": 0.47963032036139086,
      "grad_norm": 0.409091979265213,
      "learning_rate": 2.602055617721414e-06,
      "loss": 0.6798,
      "step": 11573
    },
    {
      "epoch": 0.47967176426706454,
      "grad_norm": 0.4318733215332031,
      "learning_rate": 2.601848398193046e-06,
      "loss": 0.6891,
      "step": 11574
    },
    {
      "epoch": 0.4797132081727382,
      "grad_norm": 0.4004359245300293,
      "learning_rate": 2.6016411786646776e-06,
      "loss": 0.6709,
      "step": 11575
    },
    {
      "epoch": 0.4797546520784119,
      "grad_norm": 0.3953864276409149,
      "learning_rate": 2.601433959136309e-06,
      "loss": 0.707,
      "step": 11576
    },
    {
      "epoch": 0.4797960959840855,
      "grad_norm": 0.39123499393463135,
      "learning_rate": 2.601226739607941e-06,
      "loss": 0.644,
      "step": 11577
    },
    {
      "epoch": 0.4798375398897592,
      "grad_norm": 0.4328339993953705,
      "learning_rate": 2.6010195200795722e-06,
      "loss": 0.692,
      "step": 11578
    },
    {
      "epoch": 0.47987898379543287,
      "grad_norm": 0.39897745847702026,
      "learning_rate": 2.6008123005512045e-06,
      "loss": 0.6733,
      "step": 11579
    },
    {
      "epoch": 0.47992042770110654,
      "grad_norm": 0.39828118681907654,
      "learning_rate": 2.6006050810228354e-06,
      "loss": 0.6562,
      "step": 11580
    },
    {
      "epoch": 0.4799618716067802,
      "grad_norm": 0.39140596985816956,
      "learning_rate": 2.6003978614944676e-06,
      "loss": 0.6201,
      "step": 11581
    },
    {
      "epoch": 0.4800033155124539,
      "grad_norm": 0.38618969917297363,
      "learning_rate": 2.600190641966099e-06,
      "loss": 0.6478,
      "step": 11582
    },
    {
      "epoch": 0.4800447594181276,
      "grad_norm": 0.4119326174259186,
      "learning_rate": 2.599983422437731e-06,
      "loss": 0.7139,
      "step": 11583
    },
    {
      "epoch": 0.48008620332380125,
      "grad_norm": 0.3926195502281189,
      "learning_rate": 2.5997762029093627e-06,
      "loss": 0.696,
      "step": 11584
    },
    {
      "epoch": 0.48012764722947493,
      "grad_norm": 0.4104214608669281,
      "learning_rate": 2.599568983380994e-06,
      "loss": 0.6909,
      "step": 11585
    },
    {
      "epoch": 0.48016909113514855,
      "grad_norm": 0.4070115089416504,
      "learning_rate": 2.599361763852626e-06,
      "loss": 0.7871,
      "step": 11586
    },
    {
      "epoch": 0.48021053504082223,
      "grad_norm": 0.4437679946422577,
      "learning_rate": 2.5991545443242572e-06,
      "loss": 0.6465,
      "step": 11587
    },
    {
      "epoch": 0.4802519789464959,
      "grad_norm": 0.4363630712032318,
      "learning_rate": 2.598947324795889e-06,
      "loss": 0.6528,
      "step": 11588
    },
    {
      "epoch": 0.4802934228521696,
      "grad_norm": 0.4051705598831177,
      "learning_rate": 2.5987401052675204e-06,
      "loss": 0.6622,
      "step": 11589
    },
    {
      "epoch": 0.48033486675784326,
      "grad_norm": 0.41794949769973755,
      "learning_rate": 2.5985328857391522e-06,
      "loss": 0.658,
      "step": 11590
    },
    {
      "epoch": 0.48037631066351694,
      "grad_norm": 0.3889048099517822,
      "learning_rate": 2.598325666210784e-06,
      "loss": 0.6843,
      "step": 11591
    },
    {
      "epoch": 0.4804177545691906,
      "grad_norm": 0.4050370752811432,
      "learning_rate": 2.5981184466824154e-06,
      "loss": 0.6624,
      "step": 11592
    },
    {
      "epoch": 0.4804591984748643,
      "grad_norm": 0.3932451903820038,
      "learning_rate": 2.5979112271540472e-06,
      "loss": 0.686,
      "step": 11593
    },
    {
      "epoch": 0.48050064238053797,
      "grad_norm": 0.3967250883579254,
      "learning_rate": 2.5977040076256786e-06,
      "loss": 0.6841,
      "step": 11594
    },
    {
      "epoch": 0.4805420862862116,
      "grad_norm": 0.4180542230606079,
      "learning_rate": 2.597496788097311e-06,
      "loss": 0.6305,
      "step": 11595
    },
    {
      "epoch": 0.48058353019188527,
      "grad_norm": 0.3803776502609253,
      "learning_rate": 2.597289568568942e-06,
      "loss": 0.692,
      "step": 11596
    },
    {
      "epoch": 0.48062497409755894,
      "grad_norm": 0.3814626634120941,
      "learning_rate": 2.597082349040574e-06,
      "loss": 0.7144,
      "step": 11597
    },
    {
      "epoch": 0.4806664180032326,
      "grad_norm": 0.39796778559684753,
      "learning_rate": 2.596875129512205e-06,
      "loss": 0.6339,
      "step": 11598
    },
    {
      "epoch": 0.4807078619089063,
      "grad_norm": 0.4293161630630493,
      "learning_rate": 2.5966679099838372e-06,
      "loss": 0.7506,
      "step": 11599
    },
    {
      "epoch": 0.48074930581458,
      "grad_norm": 0.3981829583644867,
      "learning_rate": 2.596460690455469e-06,
      "loss": 0.6451,
      "step": 11600
    },
    {
      "epoch": 0.48079074972025365,
      "grad_norm": 0.4225258529186249,
      "learning_rate": 2.5962534709271004e-06,
      "loss": 0.7122,
      "step": 11601
    },
    {
      "epoch": 0.48083219362592733,
      "grad_norm": 0.40936288237571716,
      "learning_rate": 2.5960462513987323e-06,
      "loss": 0.678,
      "step": 11602
    },
    {
      "epoch": 0.48087363753160095,
      "grad_norm": 0.38916948437690735,
      "learning_rate": 2.5958390318703636e-06,
      "loss": 0.626,
      "step": 11603
    },
    {
      "epoch": 0.4809150814372746,
      "grad_norm": 0.42521753907203674,
      "learning_rate": 2.5956318123419954e-06,
      "loss": 0.7061,
      "step": 11604
    },
    {
      "epoch": 0.4809565253429483,
      "grad_norm": 0.4177034795284271,
      "learning_rate": 2.595424592813627e-06,
      "loss": 0.7356,
      "step": 11605
    },
    {
      "epoch": 0.480997969248622,
      "grad_norm": 0.45508503913879395,
      "learning_rate": 2.5952173732852586e-06,
      "loss": 0.7056,
      "step": 11606
    },
    {
      "epoch": 0.48103941315429566,
      "grad_norm": 0.41210442781448364,
      "learning_rate": 2.59501015375689e-06,
      "loss": 0.6846,
      "step": 11607
    },
    {
      "epoch": 0.48108085705996934,
      "grad_norm": 0.47408780455589294,
      "learning_rate": 2.594802934228522e-06,
      "loss": 0.6836,
      "step": 11608
    },
    {
      "epoch": 0.481122300965643,
      "grad_norm": 0.386417955160141,
      "learning_rate": 2.5945957147001536e-06,
      "loss": 0.6782,
      "step": 11609
    },
    {
      "epoch": 0.4811637448713167,
      "grad_norm": 0.41070273518562317,
      "learning_rate": 2.594388495171785e-06,
      "loss": 0.7341,
      "step": 11610
    },
    {
      "epoch": 0.48120518877699037,
      "grad_norm": 0.47239434719085693,
      "learning_rate": 2.594181275643417e-06,
      "loss": 0.6901,
      "step": 11611
    },
    {
      "epoch": 0.481246632682664,
      "grad_norm": 0.39068931341171265,
      "learning_rate": 2.5939740561150482e-06,
      "loss": 0.6508,
      "step": 11612
    },
    {
      "epoch": 0.48128807658833767,
      "grad_norm": 0.4528723955154419,
      "learning_rate": 2.5937668365866805e-06,
      "loss": 0.7166,
      "step": 11613
    },
    {
      "epoch": 0.48132952049401134,
      "grad_norm": 0.3947731554508209,
      "learning_rate": 2.5935596170583114e-06,
      "loss": 0.6752,
      "step": 11614
    },
    {
      "epoch": 0.481370964399685,
      "grad_norm": 0.3788282871246338,
      "learning_rate": 2.5933523975299437e-06,
      "loss": 0.6826,
      "step": 11615
    },
    {
      "epoch": 0.4814124083053587,
      "grad_norm": 0.42978227138519287,
      "learning_rate": 2.593145178001575e-06,
      "loss": 0.7317,
      "step": 11616
    },
    {
      "epoch": 0.4814538522110324,
      "grad_norm": 0.40099453926086426,
      "learning_rate": 2.592937958473207e-06,
      "loss": 0.6721,
      "step": 11617
    },
    {
      "epoch": 0.48149529611670605,
      "grad_norm": 0.42128872871398926,
      "learning_rate": 2.5927307389448387e-06,
      "loss": 0.6758,
      "step": 11618
    },
    {
      "epoch": 0.48153674002237973,
      "grad_norm": 0.46361130475997925,
      "learning_rate": 2.59252351941647e-06,
      "loss": 0.7297,
      "step": 11619
    },
    {
      "epoch": 0.4815781839280534,
      "grad_norm": 0.40562406182289124,
      "learning_rate": 2.592316299888102e-06,
      "loss": 0.6813,
      "step": 11620
    },
    {
      "epoch": 0.481619627833727,
      "grad_norm": 0.4406891167163849,
      "learning_rate": 2.5921090803597332e-06,
      "loss": 0.7249,
      "step": 11621
    },
    {
      "epoch": 0.4816610717394007,
      "grad_norm": 0.45224037766456604,
      "learning_rate": 2.591901860831365e-06,
      "loss": 0.6941,
      "step": 11622
    },
    {
      "epoch": 0.4817025156450744,
      "grad_norm": 0.4184279441833496,
      "learning_rate": 2.5916946413029964e-06,
      "loss": 0.7012,
      "step": 11623
    },
    {
      "epoch": 0.48174395955074806,
      "grad_norm": 0.4488790035247803,
      "learning_rate": 2.5914874217746282e-06,
      "loss": 0.7092,
      "step": 11624
    },
    {
      "epoch": 0.48178540345642173,
      "grad_norm": 0.42170652747154236,
      "learning_rate": 2.5912802022462596e-06,
      "loss": 0.6256,
      "step": 11625
    },
    {
      "epoch": 0.4818268473620954,
      "grad_norm": 0.39791765809059143,
      "learning_rate": 2.5910729827178914e-06,
      "loss": 0.6582,
      "step": 11626
    },
    {
      "epoch": 0.4818682912677691,
      "grad_norm": 0.4144909679889679,
      "learning_rate": 2.5908657631895232e-06,
      "loss": 0.7446,
      "step": 11627
    },
    {
      "epoch": 0.48190973517344277,
      "grad_norm": 0.4263695180416107,
      "learning_rate": 2.5906585436611546e-06,
      "loss": 0.6865,
      "step": 11628
    },
    {
      "epoch": 0.48195117907911644,
      "grad_norm": 0.4539168179035187,
      "learning_rate": 2.590451324132787e-06,
      "loss": 0.726,
      "step": 11629
    },
    {
      "epoch": 0.48199262298479006,
      "grad_norm": 0.40417903661727905,
      "learning_rate": 2.590244104604418e-06,
      "loss": 0.6663,
      "step": 11630
    },
    {
      "epoch": 0.48203406689046374,
      "grad_norm": 0.4610176682472229,
      "learning_rate": 2.59003688507605e-06,
      "loss": 0.7583,
      "step": 11631
    },
    {
      "epoch": 0.4820755107961374,
      "grad_norm": 0.40591952204704285,
      "learning_rate": 2.589829665547681e-06,
      "loss": 0.6887,
      "step": 11632
    },
    {
      "epoch": 0.4821169547018111,
      "grad_norm": 0.39120566844940186,
      "learning_rate": 2.5896224460193133e-06,
      "loss": 0.6831,
      "step": 11633
    },
    {
      "epoch": 0.4821583986074848,
      "grad_norm": 0.5742559432983398,
      "learning_rate": 2.5894152264909446e-06,
      "loss": 0.6721,
      "step": 11634
    },
    {
      "epoch": 0.48219984251315845,
      "grad_norm": 0.42650851607322693,
      "learning_rate": 2.5892080069625764e-06,
      "loss": 0.7072,
      "step": 11635
    },
    {
      "epoch": 0.4822412864188321,
      "grad_norm": 0.41156336665153503,
      "learning_rate": 2.5890007874342083e-06,
      "loss": 0.7036,
      "step": 11636
    },
    {
      "epoch": 0.4822827303245058,
      "grad_norm": 0.4662439525127411,
      "learning_rate": 2.5887935679058396e-06,
      "loss": 0.7468,
      "step": 11637
    },
    {
      "epoch": 0.4823241742301794,
      "grad_norm": 0.4466845393180847,
      "learning_rate": 2.5885863483774715e-06,
      "loss": 0.6948,
      "step": 11638
    },
    {
      "epoch": 0.4823656181358531,
      "grad_norm": 0.4165245592594147,
      "learning_rate": 2.588379128849103e-06,
      "loss": 0.6847,
      "step": 11639
    },
    {
      "epoch": 0.4824070620415268,
      "grad_norm": 0.46678534150123596,
      "learning_rate": 2.5881719093207346e-06,
      "loss": 0.6913,
      "step": 11640
    },
    {
      "epoch": 0.48244850594720046,
      "grad_norm": 0.39842164516448975,
      "learning_rate": 2.587964689792366e-06,
      "loss": 0.7168,
      "step": 11641
    },
    {
      "epoch": 0.48248994985287413,
      "grad_norm": 0.403902143239975,
      "learning_rate": 2.587757470263998e-06,
      "loss": 0.6333,
      "step": 11642
    },
    {
      "epoch": 0.4825313937585478,
      "grad_norm": 0.39395859837532043,
      "learning_rate": 2.5875502507356292e-06,
      "loss": 0.6449,
      "step": 11643
    },
    {
      "epoch": 0.4825728376642215,
      "grad_norm": 0.3754208981990814,
      "learning_rate": 2.587343031207261e-06,
      "loss": 0.6331,
      "step": 11644
    },
    {
      "epoch": 0.48261428156989516,
      "grad_norm": 0.41033515334129333,
      "learning_rate": 2.587135811678893e-06,
      "loss": 0.672,
      "step": 11645
    },
    {
      "epoch": 0.48265572547556884,
      "grad_norm": 0.4282844662666321,
      "learning_rate": 2.5869285921505242e-06,
      "loss": 0.6808,
      "step": 11646
    },
    {
      "epoch": 0.48269716938124246,
      "grad_norm": 0.4066835343837738,
      "learning_rate": 2.5867213726221565e-06,
      "loss": 0.6833,
      "step": 11647
    },
    {
      "epoch": 0.48273861328691614,
      "grad_norm": 0.4274231195449829,
      "learning_rate": 2.5865141530937874e-06,
      "loss": 0.7034,
      "step": 11648
    },
    {
      "epoch": 0.4827800571925898,
      "grad_norm": 0.4119492471218109,
      "learning_rate": 2.5863069335654197e-06,
      "loss": 0.6686,
      "step": 11649
    },
    {
      "epoch": 0.4828215010982635,
      "grad_norm": 0.4257236123085022,
      "learning_rate": 2.586099714037051e-06,
      "loss": 0.6958,
      "step": 11650
    },
    {
      "epoch": 0.48286294500393717,
      "grad_norm": 0.42284804582595825,
      "learning_rate": 2.585892494508683e-06,
      "loss": 0.6877,
      "step": 11651
    },
    {
      "epoch": 0.48290438890961085,
      "grad_norm": 0.4185241162776947,
      "learning_rate": 2.5856852749803147e-06,
      "loss": 0.6724,
      "step": 11652
    },
    {
      "epoch": 0.4829458328152845,
      "grad_norm": 0.41104406118392944,
      "learning_rate": 2.585478055451946e-06,
      "loss": 0.6943,
      "step": 11653
    },
    {
      "epoch": 0.4829872767209582,
      "grad_norm": 0.4149235188961029,
      "learning_rate": 2.585270835923578e-06,
      "loss": 0.73,
      "step": 11654
    },
    {
      "epoch": 0.4830287206266319,
      "grad_norm": 0.39431366324424744,
      "learning_rate": 2.5850636163952092e-06,
      "loss": 0.6572,
      "step": 11655
    },
    {
      "epoch": 0.4830701645323055,
      "grad_norm": 0.4001201391220093,
      "learning_rate": 2.584856396866841e-06,
      "loss": 0.6852,
      "step": 11656
    },
    {
      "epoch": 0.4831116084379792,
      "grad_norm": 0.3522079586982727,
      "learning_rate": 2.5846491773384724e-06,
      "loss": 0.6086,
      "step": 11657
    },
    {
      "epoch": 0.48315305234365286,
      "grad_norm": 0.4063052833080292,
      "learning_rate": 2.5844419578101042e-06,
      "loss": 0.667,
      "step": 11658
    },
    {
      "epoch": 0.48319449624932653,
      "grad_norm": 0.3912616968154907,
      "learning_rate": 2.5842347382817356e-06,
      "loss": 0.6987,
      "step": 11659
    },
    {
      "epoch": 0.4832359401550002,
      "grad_norm": 0.3868068754673004,
      "learning_rate": 2.5840275187533674e-06,
      "loss": 0.6947,
      "step": 11660
    },
    {
      "epoch": 0.4832773840606739,
      "grad_norm": 0.4248347878456116,
      "learning_rate": 2.5838202992249993e-06,
      "loss": 0.6887,
      "step": 11661
    },
    {
      "epoch": 0.48331882796634756,
      "grad_norm": 0.4194730520248413,
      "learning_rate": 2.5836130796966306e-06,
      "loss": 0.6689,
      "step": 11662
    },
    {
      "epoch": 0.48336027187202124,
      "grad_norm": 0.4166310727596283,
      "learning_rate": 2.583405860168263e-06,
      "loss": 0.6963,
      "step": 11663
    },
    {
      "epoch": 0.48340171577769486,
      "grad_norm": 0.39779525995254517,
      "learning_rate": 2.583198640639894e-06,
      "loss": 0.6797,
      "step": 11664
    },
    {
      "epoch": 0.48344315968336854,
      "grad_norm": 0.39168864488601685,
      "learning_rate": 2.582991421111526e-06,
      "loss": 0.6604,
      "step": 11665
    },
    {
      "epoch": 0.4834846035890422,
      "grad_norm": 0.4259786009788513,
      "learning_rate": 2.582784201583157e-06,
      "loss": 0.6765,
      "step": 11666
    },
    {
      "epoch": 0.4835260474947159,
      "grad_norm": 0.4194049537181854,
      "learning_rate": 2.5825769820547893e-06,
      "loss": 0.7499,
      "step": 11667
    },
    {
      "epoch": 0.48356749140038957,
      "grad_norm": 0.38810208439826965,
      "learning_rate": 2.5823697625264206e-06,
      "loss": 0.6552,
      "step": 11668
    },
    {
      "epoch": 0.48360893530606325,
      "grad_norm": 0.42150184512138367,
      "learning_rate": 2.5821625429980525e-06,
      "loss": 0.7168,
      "step": 11669
    },
    {
      "epoch": 0.4836503792117369,
      "grad_norm": 0.4383477568626404,
      "learning_rate": 2.5819553234696843e-06,
      "loss": 0.7258,
      "step": 11670
    },
    {
      "epoch": 0.4836918231174106,
      "grad_norm": 0.39602476358413696,
      "learning_rate": 2.5817481039413156e-06,
      "loss": 0.7439,
      "step": 11671
    },
    {
      "epoch": 0.4837332670230843,
      "grad_norm": 0.4621061384677887,
      "learning_rate": 2.5815408844129475e-06,
      "loss": 0.6511,
      "step": 11672
    },
    {
      "epoch": 0.4837747109287579,
      "grad_norm": 0.41699016094207764,
      "learning_rate": 2.581333664884579e-06,
      "loss": 0.6528,
      "step": 11673
    },
    {
      "epoch": 0.4838161548344316,
      "grad_norm": 0.3604426383972168,
      "learning_rate": 2.5811264453562107e-06,
      "loss": 0.6536,
      "step": 11674
    },
    {
      "epoch": 0.48385759874010525,
      "grad_norm": 0.384452760219574,
      "learning_rate": 2.580919225827842e-06,
      "loss": 0.666,
      "step": 11675
    },
    {
      "epoch": 0.48389904264577893,
      "grad_norm": 0.4200873076915741,
      "learning_rate": 2.580712006299474e-06,
      "loss": 0.6874,
      "step": 11676
    },
    {
      "epoch": 0.4839404865514526,
      "grad_norm": 0.39917704463005066,
      "learning_rate": 2.5805047867711052e-06,
      "loss": 0.6606,
      "step": 11677
    },
    {
      "epoch": 0.4839819304571263,
      "grad_norm": 0.4574318528175354,
      "learning_rate": 2.580297567242737e-06,
      "loss": 0.6968,
      "step": 11678
    },
    {
      "epoch": 0.48402337436279996,
      "grad_norm": 0.3878830373287201,
      "learning_rate": 2.580090347714369e-06,
      "loss": 0.6544,
      "step": 11679
    },
    {
      "epoch": 0.48406481826847364,
      "grad_norm": 0.40189918875694275,
      "learning_rate": 2.5798831281860002e-06,
      "loss": 0.6499,
      "step": 11680
    },
    {
      "epoch": 0.4841062621741473,
      "grad_norm": 0.4005547761917114,
      "learning_rate": 2.5796759086576325e-06,
      "loss": 0.6677,
      "step": 11681
    },
    {
      "epoch": 0.48414770607982094,
      "grad_norm": 0.42312827706336975,
      "learning_rate": 2.5794686891292634e-06,
      "loss": 0.6282,
      "step": 11682
    },
    {
      "epoch": 0.4841891499854946,
      "grad_norm": 0.415216326713562,
      "learning_rate": 2.5792614696008957e-06,
      "loss": 0.6173,
      "step": 11683
    },
    {
      "epoch": 0.4842305938911683,
      "grad_norm": 0.42040297389030457,
      "learning_rate": 2.579054250072527e-06,
      "loss": 0.6399,
      "step": 11684
    },
    {
      "epoch": 0.48427203779684197,
      "grad_norm": 0.44499635696411133,
      "learning_rate": 2.578847030544159e-06,
      "loss": 0.7555,
      "step": 11685
    },
    {
      "epoch": 0.48431348170251565,
      "grad_norm": 0.4037047326564789,
      "learning_rate": 2.5786398110157902e-06,
      "loss": 0.667,
      "step": 11686
    },
    {
      "epoch": 0.4843549256081893,
      "grad_norm": 0.39928266406059265,
      "learning_rate": 2.578432591487422e-06,
      "loss": 0.6932,
      "step": 11687
    },
    {
      "epoch": 0.484396369513863,
      "grad_norm": 0.41333672404289246,
      "learning_rate": 2.578225371959054e-06,
      "loss": 0.6865,
      "step": 11688
    },
    {
      "epoch": 0.4844378134195367,
      "grad_norm": 0.4288559854030609,
      "learning_rate": 2.5780181524306853e-06,
      "loss": 0.6841,
      "step": 11689
    },
    {
      "epoch": 0.48447925732521036,
      "grad_norm": 0.3841112554073334,
      "learning_rate": 2.577810932902317e-06,
      "loss": 0.6763,
      "step": 11690
    },
    {
      "epoch": 0.484520701230884,
      "grad_norm": 0.41102245450019836,
      "learning_rate": 2.5776037133739484e-06,
      "loss": 0.6953,
      "step": 11691
    },
    {
      "epoch": 0.48456214513655765,
      "grad_norm": 0.44632160663604736,
      "learning_rate": 2.5773964938455803e-06,
      "loss": 0.6627,
      "step": 11692
    },
    {
      "epoch": 0.48460358904223133,
      "grad_norm": 0.4115983843803406,
      "learning_rate": 2.5771892743172116e-06,
      "loss": 0.6542,
      "step": 11693
    },
    {
      "epoch": 0.484645032947905,
      "grad_norm": 0.40611201524734497,
      "learning_rate": 2.5769820547888434e-06,
      "loss": 0.7737,
      "step": 11694
    },
    {
      "epoch": 0.4846864768535787,
      "grad_norm": 0.4429548680782318,
      "learning_rate": 2.576774835260475e-06,
      "loss": 0.7345,
      "step": 11695
    },
    {
      "epoch": 0.48472792075925236,
      "grad_norm": 0.39729732275009155,
      "learning_rate": 2.5765676157321066e-06,
      "loss": 0.6865,
      "step": 11696
    },
    {
      "epoch": 0.48476936466492604,
      "grad_norm": 0.42085039615631104,
      "learning_rate": 2.576360396203739e-06,
      "loss": 0.676,
      "step": 11697
    },
    {
      "epoch": 0.4848108085705997,
      "grad_norm": 0.40861254930496216,
      "learning_rate": 2.57615317667537e-06,
      "loss": 0.6149,
      "step": 11698
    },
    {
      "epoch": 0.48485225247627334,
      "grad_norm": 0.41195034980773926,
      "learning_rate": 2.575945957147002e-06,
      "loss": 0.647,
      "step": 11699
    },
    {
      "epoch": 0.484893696381947,
      "grad_norm": 0.46981099247932434,
      "learning_rate": 2.575738737618633e-06,
      "loss": 0.7092,
      "step": 11700
    },
    {
      "epoch": 0.4849351402876207,
      "grad_norm": 0.40118664503097534,
      "learning_rate": 2.5755315180902653e-06,
      "loss": 0.6641,
      "step": 11701
    },
    {
      "epoch": 0.48497658419329437,
      "grad_norm": 0.3987587094306946,
      "learning_rate": 2.5753242985618967e-06,
      "loss": 0.6421,
      "step": 11702
    },
    {
      "epoch": 0.48501802809896805,
      "grad_norm": 0.42760929465293884,
      "learning_rate": 2.5751170790335285e-06,
      "loss": 0.6533,
      "step": 11703
    },
    {
      "epoch": 0.4850594720046417,
      "grad_norm": 0.43486782908439636,
      "learning_rate": 2.57490985950516e-06,
      "loss": 0.6799,
      "step": 11704
    },
    {
      "epoch": 0.4851009159103154,
      "grad_norm": 0.4044019281864166,
      "learning_rate": 2.5747026399767917e-06,
      "loss": 0.6807,
      "step": 11705
    },
    {
      "epoch": 0.4851423598159891,
      "grad_norm": 0.4227464199066162,
      "learning_rate": 2.5744954204484235e-06,
      "loss": 0.6863,
      "step": 11706
    },
    {
      "epoch": 0.48518380372166275,
      "grad_norm": 0.4053824841976166,
      "learning_rate": 2.574288200920055e-06,
      "loss": 0.6708,
      "step": 11707
    },
    {
      "epoch": 0.4852252476273364,
      "grad_norm": 0.4071059823036194,
      "learning_rate": 2.5740809813916867e-06,
      "loss": 0.6897,
      "step": 11708
    },
    {
      "epoch": 0.48526669153301005,
      "grad_norm": 0.419656366109848,
      "learning_rate": 2.573873761863318e-06,
      "loss": 0.6445,
      "step": 11709
    },
    {
      "epoch": 0.48530813543868373,
      "grad_norm": 0.4550735056400299,
      "learning_rate": 2.57366654233495e-06,
      "loss": 0.7273,
      "step": 11710
    },
    {
      "epoch": 0.4853495793443574,
      "grad_norm": 0.3978002369403839,
      "learning_rate": 2.5734593228065812e-06,
      "loss": 0.7452,
      "step": 11711
    },
    {
      "epoch": 0.4853910232500311,
      "grad_norm": 0.43315574526786804,
      "learning_rate": 2.573252103278213e-06,
      "loss": 0.7375,
      "step": 11712
    },
    {
      "epoch": 0.48543246715570476,
      "grad_norm": 0.42050817608833313,
      "learning_rate": 2.573044883749845e-06,
      "loss": 0.6714,
      "step": 11713
    },
    {
      "epoch": 0.48547391106137844,
      "grad_norm": 0.38777080178260803,
      "learning_rate": 2.5728376642214762e-06,
      "loss": 0.6661,
      "step": 11714
    },
    {
      "epoch": 0.4855153549670521,
      "grad_norm": 0.42340484261512756,
      "learning_rate": 2.5726304446931085e-06,
      "loss": 0.7441,
      "step": 11715
    },
    {
      "epoch": 0.4855567988727258,
      "grad_norm": 0.40363290905952454,
      "learning_rate": 2.5724232251647394e-06,
      "loss": 0.6958,
      "step": 11716
    },
    {
      "epoch": 0.4855982427783994,
      "grad_norm": 0.396987646818161,
      "learning_rate": 2.5722160056363717e-06,
      "loss": 0.7152,
      "step": 11717
    },
    {
      "epoch": 0.4856396866840731,
      "grad_norm": 0.40842631459236145,
      "learning_rate": 2.572008786108003e-06,
      "loss": 0.6514,
      "step": 11718
    },
    {
      "epoch": 0.48568113058974677,
      "grad_norm": 0.4362861216068268,
      "learning_rate": 2.571801566579635e-06,
      "loss": 0.7524,
      "step": 11719
    },
    {
      "epoch": 0.48572257449542044,
      "grad_norm": 0.4214134216308594,
      "learning_rate": 2.5715943470512663e-06,
      "loss": 0.735,
      "step": 11720
    },
    {
      "epoch": 0.4857640184010941,
      "grad_norm": 0.3909485340118408,
      "learning_rate": 2.571387127522898e-06,
      "loss": 0.6918,
      "step": 11721
    },
    {
      "epoch": 0.4858054623067678,
      "grad_norm": 0.4207461178302765,
      "learning_rate": 2.57117990799453e-06,
      "loss": 0.7517,
      "step": 11722
    },
    {
      "epoch": 0.4858469062124415,
      "grad_norm": 0.46485665440559387,
      "learning_rate": 2.5709726884661613e-06,
      "loss": 0.7922,
      "step": 11723
    },
    {
      "epoch": 0.48588835011811515,
      "grad_norm": 0.423723965883255,
      "learning_rate": 2.570765468937793e-06,
      "loss": 0.637,
      "step": 11724
    },
    {
      "epoch": 0.4859297940237888,
      "grad_norm": 0.4169079065322876,
      "learning_rate": 2.5705582494094245e-06,
      "loss": 0.653,
      "step": 11725
    },
    {
      "epoch": 0.48597123792946245,
      "grad_norm": 0.42474526166915894,
      "learning_rate": 2.5703510298810563e-06,
      "loss": 0.657,
      "step": 11726
    },
    {
      "epoch": 0.48601268183513613,
      "grad_norm": 0.42716607451438904,
      "learning_rate": 2.5701438103526876e-06,
      "loss": 0.6808,
      "step": 11727
    },
    {
      "epoch": 0.4860541257408098,
      "grad_norm": 0.382269948720932,
      "learning_rate": 2.5699365908243195e-06,
      "loss": 0.6683,
      "step": 11728
    },
    {
      "epoch": 0.4860955696464835,
      "grad_norm": 0.4053051769733429,
      "learning_rate": 2.569729371295951e-06,
      "loss": 0.6904,
      "step": 11729
    },
    {
      "epoch": 0.48613701355215716,
      "grad_norm": 0.42488184571266174,
      "learning_rate": 2.5695221517675826e-06,
      "loss": 0.6719,
      "step": 11730
    },
    {
      "epoch": 0.48617845745783084,
      "grad_norm": 0.45039093494415283,
      "learning_rate": 2.569314932239215e-06,
      "loss": 0.7241,
      "step": 11731
    },
    {
      "epoch": 0.4862199013635045,
      "grad_norm": 0.44674497842788696,
      "learning_rate": 2.569107712710846e-06,
      "loss": 0.7108,
      "step": 11732
    },
    {
      "epoch": 0.4862613452691782,
      "grad_norm": 0.4420705735683441,
      "learning_rate": 2.568900493182478e-06,
      "loss": 0.7159,
      "step": 11733
    },
    {
      "epoch": 0.4863027891748518,
      "grad_norm": 0.4540080726146698,
      "learning_rate": 2.5686932736541095e-06,
      "loss": 0.7336,
      "step": 11734
    },
    {
      "epoch": 0.4863442330805255,
      "grad_norm": 0.4138307273387909,
      "learning_rate": 2.5684860541257413e-06,
      "loss": 0.6202,
      "step": 11735
    },
    {
      "epoch": 0.48638567698619917,
      "grad_norm": 0.4144209027290344,
      "learning_rate": 2.5682788345973727e-06,
      "loss": 0.6497,
      "step": 11736
    },
    {
      "epoch": 0.48642712089187284,
      "grad_norm": 0.40233033895492554,
      "learning_rate": 2.5680716150690045e-06,
      "loss": 0.6519,
      "step": 11737
    },
    {
      "epoch": 0.4864685647975465,
      "grad_norm": 0.38257160782814026,
      "learning_rate": 2.567864395540636e-06,
      "loss": 0.6746,
      "step": 11738
    },
    {
      "epoch": 0.4865100087032202,
      "grad_norm": 0.434166818857193,
      "learning_rate": 2.5676571760122677e-06,
      "loss": 0.7242,
      "step": 11739
    },
    {
      "epoch": 0.4865514526088939,
      "grad_norm": 0.4222361743450165,
      "learning_rate": 2.5674499564838995e-06,
      "loss": 0.7065,
      "step": 11740
    },
    {
      "epoch": 0.48659289651456755,
      "grad_norm": 0.46821728348731995,
      "learning_rate": 2.567242736955531e-06,
      "loss": 0.7258,
      "step": 11741
    },
    {
      "epoch": 0.48663434042024123,
      "grad_norm": 0.39915820956230164,
      "learning_rate": 2.5670355174271627e-06,
      "loss": 0.7117,
      "step": 11742
    },
    {
      "epoch": 0.48667578432591485,
      "grad_norm": 0.4079072177410126,
      "learning_rate": 2.566828297898794e-06,
      "loss": 0.6936,
      "step": 11743
    },
    {
      "epoch": 0.48671722823158853,
      "grad_norm": 0.42718204855918884,
      "learning_rate": 2.566621078370426e-06,
      "loss": 0.6787,
      "step": 11744
    },
    {
      "epoch": 0.4867586721372622,
      "grad_norm": 0.46704432368278503,
      "learning_rate": 2.5664138588420572e-06,
      "loss": 0.6738,
      "step": 11745
    },
    {
      "epoch": 0.4868001160429359,
      "grad_norm": 0.4072502851486206,
      "learning_rate": 2.566206639313689e-06,
      "loss": 0.6561,
      "step": 11746
    },
    {
      "epoch": 0.48684155994860956,
      "grad_norm": 0.5002769827842712,
      "learning_rate": 2.5659994197853204e-06,
      "loss": 0.7214,
      "step": 11747
    },
    {
      "epoch": 0.48688300385428324,
      "grad_norm": 0.4059838652610779,
      "learning_rate": 2.5657922002569523e-06,
      "loss": 0.6935,
      "step": 11748
    },
    {
      "epoch": 0.4869244477599569,
      "grad_norm": 0.4504377543926239,
      "learning_rate": 2.5655849807285845e-06,
      "loss": 0.7427,
      "step": 11749
    },
    {
      "epoch": 0.4869658916656306,
      "grad_norm": 0.4694659113883972,
      "learning_rate": 2.5653777612002154e-06,
      "loss": 0.7417,
      "step": 11750
    },
    {
      "epoch": 0.48700733557130427,
      "grad_norm": 0.41655439138412476,
      "learning_rate": 2.5651705416718477e-06,
      "loss": 0.72,
      "step": 11751
    },
    {
      "epoch": 0.4870487794769779,
      "grad_norm": 0.429232656955719,
      "learning_rate": 2.564963322143479e-06,
      "loss": 0.6376,
      "step": 11752
    },
    {
      "epoch": 0.48709022338265157,
      "grad_norm": 0.36466559767723083,
      "learning_rate": 2.564756102615111e-06,
      "loss": 0.636,
      "step": 11753
    },
    {
      "epoch": 0.48713166728832524,
      "grad_norm": 0.37916186451911926,
      "learning_rate": 2.5645488830867423e-06,
      "loss": 0.6637,
      "step": 11754
    },
    {
      "epoch": 0.4871731111939989,
      "grad_norm": 0.4096997082233429,
      "learning_rate": 2.564341663558374e-06,
      "loss": 0.7573,
      "step": 11755
    },
    {
      "epoch": 0.4872145550996726,
      "grad_norm": 0.4438663125038147,
      "learning_rate": 2.5641344440300055e-06,
      "loss": 0.7476,
      "step": 11756
    },
    {
      "epoch": 0.4872559990053463,
      "grad_norm": 0.39736124873161316,
      "learning_rate": 2.5639272245016373e-06,
      "loss": 0.6628,
      "step": 11757
    },
    {
      "epoch": 0.48729744291101995,
      "grad_norm": 0.3926561176776886,
      "learning_rate": 2.563720004973269e-06,
      "loss": 0.6973,
      "step": 11758
    },
    {
      "epoch": 0.48733888681669363,
      "grad_norm": 0.40299558639526367,
      "learning_rate": 2.5635127854449005e-06,
      "loss": 0.6882,
      "step": 11759
    },
    {
      "epoch": 0.48738033072236725,
      "grad_norm": 0.41009944677352905,
      "learning_rate": 2.5633055659165323e-06,
      "loss": 0.7012,
      "step": 11760
    },
    {
      "epoch": 0.4874217746280409,
      "grad_norm": 0.4384874403476715,
      "learning_rate": 2.5630983463881637e-06,
      "loss": 0.6853,
      "step": 11761
    },
    {
      "epoch": 0.4874632185337146,
      "grad_norm": 0.43121787905693054,
      "learning_rate": 2.5628911268597955e-06,
      "loss": 0.7137,
      "step": 11762
    },
    {
      "epoch": 0.4875046624393883,
      "grad_norm": 0.4062303304672241,
      "learning_rate": 2.562683907331427e-06,
      "loss": 0.7051,
      "step": 11763
    },
    {
      "epoch": 0.48754610634506196,
      "grad_norm": 0.4356639087200165,
      "learning_rate": 2.5624766878030587e-06,
      "loss": 0.6785,
      "step": 11764
    },
    {
      "epoch": 0.48758755025073564,
      "grad_norm": 0.39018702507019043,
      "learning_rate": 2.56226946827469e-06,
      "loss": 0.7072,
      "step": 11765
    },
    {
      "epoch": 0.4876289941564093,
      "grad_norm": 0.38774725794792175,
      "learning_rate": 2.562062248746322e-06,
      "loss": 0.6543,
      "step": 11766
    },
    {
      "epoch": 0.487670438062083,
      "grad_norm": 0.41465339064598083,
      "learning_rate": 2.561855029217954e-06,
      "loss": 0.6765,
      "step": 11767
    },
    {
      "epoch": 0.48771188196775667,
      "grad_norm": 0.41915664076805115,
      "learning_rate": 2.5616478096895855e-06,
      "loss": 0.731,
      "step": 11768
    },
    {
      "epoch": 0.4877533258734303,
      "grad_norm": 0.41065385937690735,
      "learning_rate": 2.5614405901612173e-06,
      "loss": 0.6531,
      "step": 11769
    },
    {
      "epoch": 0.48779476977910396,
      "grad_norm": 0.4317989945411682,
      "learning_rate": 2.5612333706328487e-06,
      "loss": 0.6862,
      "step": 11770
    },
    {
      "epoch": 0.48783621368477764,
      "grad_norm": 0.43733957409858704,
      "learning_rate": 2.5610261511044805e-06,
      "loss": 0.7183,
      "step": 11771
    },
    {
      "epoch": 0.4878776575904513,
      "grad_norm": 0.42367860674858093,
      "learning_rate": 2.560818931576112e-06,
      "loss": 0.688,
      "step": 11772
    },
    {
      "epoch": 0.487919101496125,
      "grad_norm": 0.4038856625556946,
      "learning_rate": 2.5606117120477437e-06,
      "loss": 0.6648,
      "step": 11773
    },
    {
      "epoch": 0.4879605454017987,
      "grad_norm": 0.4097851812839508,
      "learning_rate": 2.5604044925193755e-06,
      "loss": 0.7021,
      "step": 11774
    },
    {
      "epoch": 0.48800198930747235,
      "grad_norm": 0.39602458477020264,
      "learning_rate": 2.560197272991007e-06,
      "loss": 0.6697,
      "step": 11775
    },
    {
      "epoch": 0.488043433213146,
      "grad_norm": 0.43369197845458984,
      "learning_rate": 2.5599900534626387e-06,
      "loss": 0.662,
      "step": 11776
    },
    {
      "epoch": 0.4880848771188197,
      "grad_norm": 0.42835086584091187,
      "learning_rate": 2.55978283393427e-06,
      "loss": 0.6663,
      "step": 11777
    },
    {
      "epoch": 0.4881263210244933,
      "grad_norm": 0.3784225881099701,
      "learning_rate": 2.559575614405902e-06,
      "loss": 0.6608,
      "step": 11778
    },
    {
      "epoch": 0.488167764930167,
      "grad_norm": 0.49114200472831726,
      "learning_rate": 2.5593683948775333e-06,
      "loss": 0.7217,
      "step": 11779
    },
    {
      "epoch": 0.4882092088358407,
      "grad_norm": 0.423460990190506,
      "learning_rate": 2.559161175349165e-06,
      "loss": 0.675,
      "step": 11780
    },
    {
      "epoch": 0.48825065274151436,
      "grad_norm": 0.4231064021587372,
      "learning_rate": 2.5589539558207964e-06,
      "loss": 0.6616,
      "step": 11781
    },
    {
      "epoch": 0.48829209664718803,
      "grad_norm": 0.3901020884513855,
      "learning_rate": 2.5587467362924283e-06,
      "loss": 0.6527,
      "step": 11782
    },
    {
      "epoch": 0.4883335405528617,
      "grad_norm": 0.42021113634109497,
      "learning_rate": 2.5585395167640605e-06,
      "loss": 0.7344,
      "step": 11783
    },
    {
      "epoch": 0.4883749844585354,
      "grad_norm": 0.4223724603652954,
      "learning_rate": 2.5583322972356915e-06,
      "loss": 0.6892,
      "step": 11784
    },
    {
      "epoch": 0.48841642836420907,
      "grad_norm": 0.41570302844047546,
      "learning_rate": 2.5581250777073237e-06,
      "loss": 0.6814,
      "step": 11785
    },
    {
      "epoch": 0.4884578722698827,
      "grad_norm": 0.3912476599216461,
      "learning_rate": 2.557917858178955e-06,
      "loss": 0.7214,
      "step": 11786
    },
    {
      "epoch": 0.48849931617555636,
      "grad_norm": 0.41298478841781616,
      "learning_rate": 2.557710638650587e-06,
      "loss": 0.645,
      "step": 11787
    },
    {
      "epoch": 0.48854076008123004,
      "grad_norm": 0.4091927111148834,
      "learning_rate": 2.5575034191222183e-06,
      "loss": 0.7034,
      "step": 11788
    },
    {
      "epoch": 0.4885822039869037,
      "grad_norm": 0.3971502184867859,
      "learning_rate": 2.55729619959385e-06,
      "loss": 0.6464,
      "step": 11789
    },
    {
      "epoch": 0.4886236478925774,
      "grad_norm": 0.4347737729549408,
      "learning_rate": 2.5570889800654815e-06,
      "loss": 0.7012,
      "step": 11790
    },
    {
      "epoch": 0.48866509179825107,
      "grad_norm": 0.4490567743778229,
      "learning_rate": 2.5568817605371133e-06,
      "loss": 0.6892,
      "step": 11791
    },
    {
      "epoch": 0.48870653570392475,
      "grad_norm": 0.4031572639942169,
      "learning_rate": 2.556674541008745e-06,
      "loss": 0.7236,
      "step": 11792
    },
    {
      "epoch": 0.4887479796095984,
      "grad_norm": 0.36614537239074707,
      "learning_rate": 2.5564673214803765e-06,
      "loss": 0.6156,
      "step": 11793
    },
    {
      "epoch": 0.4887894235152721,
      "grad_norm": 0.42122191190719604,
      "learning_rate": 2.5562601019520083e-06,
      "loss": 0.7244,
      "step": 11794
    },
    {
      "epoch": 0.4888308674209457,
      "grad_norm": 0.40989986062049866,
      "learning_rate": 2.5560528824236397e-06,
      "loss": 0.6771,
      "step": 11795
    },
    {
      "epoch": 0.4888723113266194,
      "grad_norm": 0.3913651406764984,
      "learning_rate": 2.5558456628952715e-06,
      "loss": 0.6169,
      "step": 11796
    },
    {
      "epoch": 0.4889137552322931,
      "grad_norm": 0.38704216480255127,
      "learning_rate": 2.555638443366903e-06,
      "loss": 0.7042,
      "step": 11797
    },
    {
      "epoch": 0.48895519913796676,
      "grad_norm": 0.410615473985672,
      "learning_rate": 2.5554312238385347e-06,
      "loss": 0.6433,
      "step": 11798
    },
    {
      "epoch": 0.48899664304364043,
      "grad_norm": 0.45726287364959717,
      "learning_rate": 2.555224004310166e-06,
      "loss": 0.6908,
      "step": 11799
    },
    {
      "epoch": 0.4890380869493141,
      "grad_norm": 0.429104208946228,
      "learning_rate": 2.555016784781798e-06,
      "loss": 0.6578,
      "step": 11800
    },
    {
      "epoch": 0.4890795308549878,
      "grad_norm": 0.4122019410133362,
      "learning_rate": 2.55480956525343e-06,
      "loss": 0.6213,
      "step": 11801
    },
    {
      "epoch": 0.48912097476066146,
      "grad_norm": 0.4092191755771637,
      "learning_rate": 2.5546023457250615e-06,
      "loss": 0.6329,
      "step": 11802
    },
    {
      "epoch": 0.48916241866633514,
      "grad_norm": 0.41506391763687134,
      "learning_rate": 2.5543951261966933e-06,
      "loss": 0.6989,
      "step": 11803
    },
    {
      "epoch": 0.48920386257200876,
      "grad_norm": 0.4621204435825348,
      "learning_rate": 2.5541879066683247e-06,
      "loss": 0.7395,
      "step": 11804
    },
    {
      "epoch": 0.48924530647768244,
      "grad_norm": 0.3812274932861328,
      "learning_rate": 2.5539806871399565e-06,
      "loss": 0.6593,
      "step": 11805
    },
    {
      "epoch": 0.4892867503833561,
      "grad_norm": 0.432704895734787,
      "learning_rate": 2.553773467611588e-06,
      "loss": 0.7329,
      "step": 11806
    },
    {
      "epoch": 0.4893281942890298,
      "grad_norm": 0.39898422360420227,
      "learning_rate": 2.5535662480832197e-06,
      "loss": 0.6775,
      "step": 11807
    },
    {
      "epoch": 0.48936963819470347,
      "grad_norm": 0.38203686475753784,
      "learning_rate": 2.553359028554851e-06,
      "loss": 0.7493,
      "step": 11808
    },
    {
      "epoch": 0.48941108210037715,
      "grad_norm": 0.44070830941200256,
      "learning_rate": 2.553151809026483e-06,
      "loss": 0.6707,
      "step": 11809
    },
    {
      "epoch": 0.4894525260060508,
      "grad_norm": 0.4193633794784546,
      "learning_rate": 2.5529445894981147e-06,
      "loss": 0.6736,
      "step": 11810
    },
    {
      "epoch": 0.4894939699117245,
      "grad_norm": 0.47975507378578186,
      "learning_rate": 2.552737369969746e-06,
      "loss": 0.7681,
      "step": 11811
    },
    {
      "epoch": 0.4895354138173981,
      "grad_norm": 0.41110509634017944,
      "learning_rate": 2.552530150441378e-06,
      "loss": 0.6852,
      "step": 11812
    },
    {
      "epoch": 0.4895768577230718,
      "grad_norm": 0.4707271456718445,
      "learning_rate": 2.5523229309130093e-06,
      "loss": 0.7383,
      "step": 11813
    },
    {
      "epoch": 0.4896183016287455,
      "grad_norm": 0.4056098163127899,
      "learning_rate": 2.552115711384641e-06,
      "loss": 0.6951,
      "step": 11814
    },
    {
      "epoch": 0.48965974553441916,
      "grad_norm": 0.4092637896537781,
      "learning_rate": 2.5519084918562725e-06,
      "loss": 0.6829,
      "step": 11815
    },
    {
      "epoch": 0.48970118944009283,
      "grad_norm": 0.4322759807109833,
      "learning_rate": 2.5517012723279043e-06,
      "loss": 0.6638,
      "step": 11816
    },
    {
      "epoch": 0.4897426333457665,
      "grad_norm": 0.42654603719711304,
      "learning_rate": 2.5514940527995356e-06,
      "loss": 0.6951,
      "step": 11817
    },
    {
      "epoch": 0.4897840772514402,
      "grad_norm": 0.4356617331504822,
      "learning_rate": 2.5512868332711675e-06,
      "loss": 0.7192,
      "step": 11818
    },
    {
      "epoch": 0.48982552115711386,
      "grad_norm": 0.418536901473999,
      "learning_rate": 2.5510796137427997e-06,
      "loss": 0.6965,
      "step": 11819
    },
    {
      "epoch": 0.48986696506278754,
      "grad_norm": 0.4005710482597351,
      "learning_rate": 2.550872394214431e-06,
      "loss": 0.6799,
      "step": 11820
    },
    {
      "epoch": 0.48990840896846116,
      "grad_norm": 0.3959464430809021,
      "learning_rate": 2.550665174686063e-06,
      "loss": 0.6576,
      "step": 11821
    },
    {
      "epoch": 0.48994985287413484,
      "grad_norm": 0.4176786541938782,
      "learning_rate": 2.5504579551576943e-06,
      "loss": 0.7063,
      "step": 11822
    },
    {
      "epoch": 0.4899912967798085,
      "grad_norm": 0.3999761641025543,
      "learning_rate": 2.550250735629326e-06,
      "loss": 0.6488,
      "step": 11823
    },
    {
      "epoch": 0.4900327406854822,
      "grad_norm": 0.4032444655895233,
      "learning_rate": 2.5500435161009575e-06,
      "loss": 0.6943,
      "step": 11824
    },
    {
      "epoch": 0.49007418459115587,
      "grad_norm": 0.42853134870529175,
      "learning_rate": 2.5498362965725893e-06,
      "loss": 0.6951,
      "step": 11825
    },
    {
      "epoch": 0.49011562849682955,
      "grad_norm": 0.36227530241012573,
      "learning_rate": 2.5496290770442207e-06,
      "loss": 0.6724,
      "step": 11826
    },
    {
      "epoch": 0.4901570724025032,
      "grad_norm": 0.4039766788482666,
      "learning_rate": 2.5494218575158525e-06,
      "loss": 0.7124,
      "step": 11827
    },
    {
      "epoch": 0.4901985163081769,
      "grad_norm": 0.39769476652145386,
      "learning_rate": 2.5492146379874843e-06,
      "loss": 0.6241,
      "step": 11828
    },
    {
      "epoch": 0.4902399602138506,
      "grad_norm": 0.3912244439125061,
      "learning_rate": 2.5490074184591157e-06,
      "loss": 0.6667,
      "step": 11829
    },
    {
      "epoch": 0.4902814041195242,
      "grad_norm": 0.41921713948249817,
      "learning_rate": 2.5488001989307475e-06,
      "loss": 0.7174,
      "step": 11830
    },
    {
      "epoch": 0.4903228480251979,
      "grad_norm": 0.40436655282974243,
      "learning_rate": 2.548592979402379e-06,
      "loss": 0.6875,
      "step": 11831
    },
    {
      "epoch": 0.49036429193087155,
      "grad_norm": 0.4021992087364197,
      "learning_rate": 2.5483857598740107e-06,
      "loss": 0.7246,
      "step": 11832
    },
    {
      "epoch": 0.49040573583654523,
      "grad_norm": 0.3978753685951233,
      "learning_rate": 2.548178540345642e-06,
      "loss": 0.6903,
      "step": 11833
    },
    {
      "epoch": 0.4904471797422189,
      "grad_norm": 0.3932604491710663,
      "learning_rate": 2.547971320817274e-06,
      "loss": 0.6526,
      "step": 11834
    },
    {
      "epoch": 0.4904886236478926,
      "grad_norm": 0.3858346939086914,
      "learning_rate": 2.547764101288906e-06,
      "loss": 0.6802,
      "step": 11835
    },
    {
      "epoch": 0.49053006755356626,
      "grad_norm": 0.4108521640300751,
      "learning_rate": 2.5475568817605375e-06,
      "loss": 0.7217,
      "step": 11836
    },
    {
      "epoch": 0.49057151145923994,
      "grad_norm": 0.4297684133052826,
      "learning_rate": 2.5473496622321693e-06,
      "loss": 0.739,
      "step": 11837
    },
    {
      "epoch": 0.4906129553649136,
      "grad_norm": 0.37183231115341187,
      "learning_rate": 2.5471424427038007e-06,
      "loss": 0.6621,
      "step": 11838
    },
    {
      "epoch": 0.49065439927058724,
      "grad_norm": 0.41974401473999023,
      "learning_rate": 2.5469352231754325e-06,
      "loss": 0.6919,
      "step": 11839
    },
    {
      "epoch": 0.4906958431762609,
      "grad_norm": 0.39542335271835327,
      "learning_rate": 2.546728003647064e-06,
      "loss": 0.7153,
      "step": 11840
    },
    {
      "epoch": 0.4907372870819346,
      "grad_norm": 0.43556898832321167,
      "learning_rate": 2.5465207841186957e-06,
      "loss": 0.6771,
      "step": 11841
    },
    {
      "epoch": 0.49077873098760827,
      "grad_norm": 0.4343653619289398,
      "learning_rate": 2.546313564590327e-06,
      "loss": 0.6765,
      "step": 11842
    },
    {
      "epoch": 0.49082017489328195,
      "grad_norm": 0.4049076437950134,
      "learning_rate": 2.546106345061959e-06,
      "loss": 0.666,
      "step": 11843
    },
    {
      "epoch": 0.4908616187989556,
      "grad_norm": 0.40400511026382446,
      "learning_rate": 2.5458991255335907e-06,
      "loss": 0.6682,
      "step": 11844
    },
    {
      "epoch": 0.4909030627046293,
      "grad_norm": 0.4484359920024872,
      "learning_rate": 2.545691906005222e-06,
      "loss": 0.6814,
      "step": 11845
    },
    {
      "epoch": 0.490944506610303,
      "grad_norm": 0.4321199655532837,
      "learning_rate": 2.545484686476854e-06,
      "loss": 0.7141,
      "step": 11846
    },
    {
      "epoch": 0.4909859505159766,
      "grad_norm": 0.3911685049533844,
      "learning_rate": 2.5452774669484853e-06,
      "loss": 0.6875,
      "step": 11847
    },
    {
      "epoch": 0.4910273944216503,
      "grad_norm": 0.41462430357933044,
      "learning_rate": 2.545070247420117e-06,
      "loss": 0.7178,
      "step": 11848
    },
    {
      "epoch": 0.49106883832732395,
      "grad_norm": 0.3773510158061981,
      "learning_rate": 2.5448630278917485e-06,
      "loss": 0.6526,
      "step": 11849
    },
    {
      "epoch": 0.49111028223299763,
      "grad_norm": 0.3944261074066162,
      "learning_rate": 2.5446558083633803e-06,
      "loss": 0.7092,
      "step": 11850
    },
    {
      "epoch": 0.4911517261386713,
      "grad_norm": 0.42144206166267395,
      "learning_rate": 2.5444485888350117e-06,
      "loss": 0.707,
      "step": 11851
    },
    {
      "epoch": 0.491193170044345,
      "grad_norm": 0.42240363359451294,
      "learning_rate": 2.5442413693066435e-06,
      "loss": 0.6796,
      "step": 11852
    },
    {
      "epoch": 0.49123461395001866,
      "grad_norm": 0.4180384874343872,
      "learning_rate": 2.5440341497782757e-06,
      "loss": 0.696,
      "step": 11853
    },
    {
      "epoch": 0.49127605785569234,
      "grad_norm": 0.4444087743759155,
      "learning_rate": 2.543826930249907e-06,
      "loss": 0.7529,
      "step": 11854
    },
    {
      "epoch": 0.491317501761366,
      "grad_norm": 0.4109300971031189,
      "learning_rate": 2.543619710721539e-06,
      "loss": 0.6666,
      "step": 11855
    },
    {
      "epoch": 0.49135894566703964,
      "grad_norm": 0.42331811785697937,
      "learning_rate": 2.5434124911931703e-06,
      "loss": 0.6768,
      "step": 11856
    },
    {
      "epoch": 0.4914003895727133,
      "grad_norm": 0.4209948182106018,
      "learning_rate": 2.543205271664802e-06,
      "loss": 0.6848,
      "step": 11857
    },
    {
      "epoch": 0.491441833478387,
      "grad_norm": 0.4185831844806671,
      "learning_rate": 2.5429980521364335e-06,
      "loss": 0.6792,
      "step": 11858
    },
    {
      "epoch": 0.49148327738406067,
      "grad_norm": 0.44093117117881775,
      "learning_rate": 2.5427908326080653e-06,
      "loss": 0.7213,
      "step": 11859
    },
    {
      "epoch": 0.49152472128973435,
      "grad_norm": 0.4204893112182617,
      "learning_rate": 2.5425836130796967e-06,
      "loss": 0.6877,
      "step": 11860
    },
    {
      "epoch": 0.491566165195408,
      "grad_norm": 0.5034067630767822,
      "learning_rate": 2.5423763935513285e-06,
      "loss": 0.7206,
      "step": 11861
    },
    {
      "epoch": 0.4916076091010817,
      "grad_norm": 0.39760157465934753,
      "learning_rate": 2.5421691740229603e-06,
      "loss": 0.6691,
      "step": 11862
    },
    {
      "epoch": 0.4916490530067554,
      "grad_norm": 0.3997891843318939,
      "learning_rate": 2.5419619544945917e-06,
      "loss": 0.6218,
      "step": 11863
    },
    {
      "epoch": 0.49169049691242905,
      "grad_norm": 0.4092789888381958,
      "learning_rate": 2.5417547349662235e-06,
      "loss": 0.6833,
      "step": 11864
    },
    {
      "epoch": 0.4917319408181027,
      "grad_norm": 0.39059022068977356,
      "learning_rate": 2.541547515437855e-06,
      "loss": 0.6664,
      "step": 11865
    },
    {
      "epoch": 0.49177338472377635,
      "grad_norm": 0.3996644914150238,
      "learning_rate": 2.5413402959094867e-06,
      "loss": 0.6194,
      "step": 11866
    },
    {
      "epoch": 0.49181482862945003,
      "grad_norm": 0.39793941378593445,
      "learning_rate": 2.541133076381118e-06,
      "loss": 0.686,
      "step": 11867
    },
    {
      "epoch": 0.4918562725351237,
      "grad_norm": 0.4015215039253235,
      "learning_rate": 2.54092585685275e-06,
      "loss": 0.7085,
      "step": 11868
    },
    {
      "epoch": 0.4918977164407974,
      "grad_norm": 0.3885466754436493,
      "learning_rate": 2.5407186373243813e-06,
      "loss": 0.6401,
      "step": 11869
    },
    {
      "epoch": 0.49193916034647106,
      "grad_norm": 0.43044522404670715,
      "learning_rate": 2.5405114177960135e-06,
      "loss": 0.7222,
      "step": 11870
    },
    {
      "epoch": 0.49198060425214474,
      "grad_norm": 0.4216560125350952,
      "learning_rate": 2.5403041982676453e-06,
      "loss": 0.7023,
      "step": 11871
    },
    {
      "epoch": 0.4920220481578184,
      "grad_norm": 0.4033070504665375,
      "learning_rate": 2.5400969787392767e-06,
      "loss": 0.6626,
      "step": 11872
    },
    {
      "epoch": 0.49206349206349204,
      "grad_norm": 0.4591406285762787,
      "learning_rate": 2.5398897592109085e-06,
      "loss": 0.7756,
      "step": 11873
    },
    {
      "epoch": 0.4921049359691657,
      "grad_norm": 0.46172305941581726,
      "learning_rate": 2.53968253968254e-06,
      "loss": 0.7234,
      "step": 11874
    },
    {
      "epoch": 0.4921463798748394,
      "grad_norm": 0.3872503936290741,
      "learning_rate": 2.5394753201541717e-06,
      "loss": 0.6917,
      "step": 11875
    },
    {
      "epoch": 0.49218782378051307,
      "grad_norm": 0.4404793977737427,
      "learning_rate": 2.539268100625803e-06,
      "loss": 0.7593,
      "step": 11876
    },
    {
      "epoch": 0.49222926768618674,
      "grad_norm": 0.4223960340023041,
      "learning_rate": 2.539060881097435e-06,
      "loss": 0.6919,
      "step": 11877
    },
    {
      "epoch": 0.4922707115918604,
      "grad_norm": 0.3922606110572815,
      "learning_rate": 2.5388536615690663e-06,
      "loss": 0.6484,
      "step": 11878
    },
    {
      "epoch": 0.4923121554975341,
      "grad_norm": 0.4334293305873871,
      "learning_rate": 2.538646442040698e-06,
      "loss": 0.7112,
      "step": 11879
    },
    {
      "epoch": 0.4923535994032078,
      "grad_norm": 0.41647428274154663,
      "learning_rate": 2.53843922251233e-06,
      "loss": 0.6829,
      "step": 11880
    },
    {
      "epoch": 0.49239504330888145,
      "grad_norm": 0.4099753201007843,
      "learning_rate": 2.5382320029839613e-06,
      "loss": 0.6545,
      "step": 11881
    },
    {
      "epoch": 0.4924364872145551,
      "grad_norm": 0.3943389654159546,
      "learning_rate": 2.538024783455593e-06,
      "loss": 0.6642,
      "step": 11882
    },
    {
      "epoch": 0.49247793112022875,
      "grad_norm": 0.4057413935661316,
      "learning_rate": 2.5378175639272245e-06,
      "loss": 0.6758,
      "step": 11883
    },
    {
      "epoch": 0.49251937502590243,
      "grad_norm": 0.3960435688495636,
      "learning_rate": 2.5376103443988563e-06,
      "loss": 0.655,
      "step": 11884
    },
    {
      "epoch": 0.4925608189315761,
      "grad_norm": 0.4294629693031311,
      "learning_rate": 2.5374031248704877e-06,
      "loss": 0.6821,
      "step": 11885
    },
    {
      "epoch": 0.4926022628372498,
      "grad_norm": 0.4214281737804413,
      "learning_rate": 2.5371959053421195e-06,
      "loss": 0.6667,
      "step": 11886
    },
    {
      "epoch": 0.49264370674292346,
      "grad_norm": 0.42540642619132996,
      "learning_rate": 2.5369886858137517e-06,
      "loss": 0.6602,
      "step": 11887
    },
    {
      "epoch": 0.49268515064859714,
      "grad_norm": 0.4040733575820923,
      "learning_rate": 2.536781466285383e-06,
      "loss": 0.7051,
      "step": 11888
    },
    {
      "epoch": 0.4927265945542708,
      "grad_norm": 0.40961313247680664,
      "learning_rate": 2.536574246757015e-06,
      "loss": 0.6539,
      "step": 11889
    },
    {
      "epoch": 0.4927680384599445,
      "grad_norm": 0.45013532042503357,
      "learning_rate": 2.5363670272286463e-06,
      "loss": 0.7177,
      "step": 11890
    },
    {
      "epoch": 0.4928094823656181,
      "grad_norm": 0.40048396587371826,
      "learning_rate": 2.536159807700278e-06,
      "loss": 0.6259,
      "step": 11891
    },
    {
      "epoch": 0.4928509262712918,
      "grad_norm": 0.4677923321723938,
      "learning_rate": 2.5359525881719095e-06,
      "loss": 0.7701,
      "step": 11892
    },
    {
      "epoch": 0.49289237017696547,
      "grad_norm": 0.375237375497818,
      "learning_rate": 2.5357453686435413e-06,
      "loss": 0.6663,
      "step": 11893
    },
    {
      "epoch": 0.49293381408263914,
      "grad_norm": 0.4697551131248474,
      "learning_rate": 2.5355381491151727e-06,
      "loss": 0.7939,
      "step": 11894
    },
    {
      "epoch": 0.4929752579883128,
      "grad_norm": 0.4318310022354126,
      "learning_rate": 2.5353309295868045e-06,
      "loss": 0.6587,
      "step": 11895
    },
    {
      "epoch": 0.4930167018939865,
      "grad_norm": 0.3958035111427307,
      "learning_rate": 2.5351237100584363e-06,
      "loss": 0.6575,
      "step": 11896
    },
    {
      "epoch": 0.4930581457996602,
      "grad_norm": 0.40231743454933167,
      "learning_rate": 2.5349164905300677e-06,
      "loss": 0.6603,
      "step": 11897
    },
    {
      "epoch": 0.49309958970533385,
      "grad_norm": 0.3843839168548584,
      "learning_rate": 2.5347092710016995e-06,
      "loss": 0.6792,
      "step": 11898
    },
    {
      "epoch": 0.49314103361100753,
      "grad_norm": 0.4204179644584656,
      "learning_rate": 2.534502051473331e-06,
      "loss": 0.682,
      "step": 11899
    },
    {
      "epoch": 0.49318247751668115,
      "grad_norm": 0.3666963279247284,
      "learning_rate": 2.5342948319449627e-06,
      "loss": 0.6653,
      "step": 11900
    },
    {
      "epoch": 0.4932239214223548,
      "grad_norm": 0.3982766270637512,
      "learning_rate": 2.534087612416594e-06,
      "loss": 0.6748,
      "step": 11901
    },
    {
      "epoch": 0.4932653653280285,
      "grad_norm": 0.40631696581840515,
      "learning_rate": 2.533880392888226e-06,
      "loss": 0.7026,
      "step": 11902
    },
    {
      "epoch": 0.4933068092337022,
      "grad_norm": 0.405801922082901,
      "learning_rate": 2.5336731733598573e-06,
      "loss": 0.7266,
      "step": 11903
    },
    {
      "epoch": 0.49334825313937586,
      "grad_norm": 0.3891429305076599,
      "learning_rate": 2.5334659538314895e-06,
      "loss": 0.6575,
      "step": 11904
    },
    {
      "epoch": 0.49338969704504954,
      "grad_norm": 0.4176947772502899,
      "learning_rate": 2.5332587343031213e-06,
      "loss": 0.6965,
      "step": 11905
    },
    {
      "epoch": 0.4934311409507232,
      "grad_norm": 0.3754618763923645,
      "learning_rate": 2.5330515147747527e-06,
      "loss": 0.6799,
      "step": 11906
    },
    {
      "epoch": 0.4934725848563969,
      "grad_norm": 0.41250160336494446,
      "learning_rate": 2.5328442952463845e-06,
      "loss": 0.6953,
      "step": 11907
    },
    {
      "epoch": 0.4935140287620705,
      "grad_norm": 0.45370081067085266,
      "learning_rate": 2.532637075718016e-06,
      "loss": 0.749,
      "step": 11908
    },
    {
      "epoch": 0.4935554726677442,
      "grad_norm": 0.41599783301353455,
      "learning_rate": 2.5324298561896477e-06,
      "loss": 0.7106,
      "step": 11909
    },
    {
      "epoch": 0.49359691657341787,
      "grad_norm": 0.3919639587402344,
      "learning_rate": 2.532222636661279e-06,
      "loss": 0.647,
      "step": 11910
    },
    {
      "epoch": 0.49363836047909154,
      "grad_norm": 0.4156576991081238,
      "learning_rate": 2.532015417132911e-06,
      "loss": 0.6996,
      "step": 11911
    },
    {
      "epoch": 0.4936798043847652,
      "grad_norm": 0.40890729427337646,
      "learning_rate": 2.5318081976045423e-06,
      "loss": 0.6455,
      "step": 11912
    },
    {
      "epoch": 0.4937212482904389,
      "grad_norm": 0.4306601583957672,
      "learning_rate": 2.531600978076174e-06,
      "loss": 0.6378,
      "step": 11913
    },
    {
      "epoch": 0.4937626921961126,
      "grad_norm": 0.402927964925766,
      "learning_rate": 2.531393758547806e-06,
      "loss": 0.6974,
      "step": 11914
    },
    {
      "epoch": 0.49380413610178625,
      "grad_norm": 0.435237318277359,
      "learning_rate": 2.5311865390194373e-06,
      "loss": 0.7483,
      "step": 11915
    },
    {
      "epoch": 0.49384558000745993,
      "grad_norm": 0.4264313578605652,
      "learning_rate": 2.530979319491069e-06,
      "loss": 0.6821,
      "step": 11916
    },
    {
      "epoch": 0.49388702391313355,
      "grad_norm": 0.42379096150398254,
      "learning_rate": 2.5307720999627005e-06,
      "loss": 0.7137,
      "step": 11917
    },
    {
      "epoch": 0.4939284678188072,
      "grad_norm": 0.4034231901168823,
      "learning_rate": 2.5305648804343323e-06,
      "loss": 0.6807,
      "step": 11918
    },
    {
      "epoch": 0.4939699117244809,
      "grad_norm": 0.43287500739097595,
      "learning_rate": 2.5303576609059637e-06,
      "loss": 0.7908,
      "step": 11919
    },
    {
      "epoch": 0.4940113556301546,
      "grad_norm": 0.4099683463573456,
      "learning_rate": 2.530150441377596e-06,
      "loss": 0.728,
      "step": 11920
    },
    {
      "epoch": 0.49405279953582826,
      "grad_norm": 0.3925293982028961,
      "learning_rate": 2.529943221849227e-06,
      "loss": 0.6821,
      "step": 11921
    },
    {
      "epoch": 0.49409424344150193,
      "grad_norm": 0.4048459231853485,
      "learning_rate": 2.529736002320859e-06,
      "loss": 0.6989,
      "step": 11922
    },
    {
      "epoch": 0.4941356873471756,
      "grad_norm": 0.4153275787830353,
      "learning_rate": 2.529528782792491e-06,
      "loss": 0.6876,
      "step": 11923
    },
    {
      "epoch": 0.4941771312528493,
      "grad_norm": 0.4109082818031311,
      "learning_rate": 2.5293215632641223e-06,
      "loss": 0.7103,
      "step": 11924
    },
    {
      "epoch": 0.49421857515852297,
      "grad_norm": 0.3953899145126343,
      "learning_rate": 2.529114343735754e-06,
      "loss": 0.637,
      "step": 11925
    },
    {
      "epoch": 0.4942600190641966,
      "grad_norm": 0.3811253607273102,
      "learning_rate": 2.5289071242073855e-06,
      "loss": 0.6733,
      "step": 11926
    },
    {
      "epoch": 0.49430146296987026,
      "grad_norm": 0.38809067010879517,
      "learning_rate": 2.5286999046790173e-06,
      "loss": 0.7112,
      "step": 11927
    },
    {
      "epoch": 0.49434290687554394,
      "grad_norm": 0.4251621961593628,
      "learning_rate": 2.5284926851506487e-06,
      "loss": 0.7329,
      "step": 11928
    },
    {
      "epoch": 0.4943843507812176,
      "grad_norm": 0.3860369622707367,
      "learning_rate": 2.5282854656222805e-06,
      "loss": 0.6737,
      "step": 11929
    },
    {
      "epoch": 0.4944257946868913,
      "grad_norm": 0.3976783752441406,
      "learning_rate": 2.528078246093912e-06,
      "loss": 0.6541,
      "step": 11930
    },
    {
      "epoch": 0.494467238592565,
      "grad_norm": 0.38773834705352783,
      "learning_rate": 2.5278710265655437e-06,
      "loss": 0.6941,
      "step": 11931
    },
    {
      "epoch": 0.49450868249823865,
      "grad_norm": 0.4206503927707672,
      "learning_rate": 2.5276638070371755e-06,
      "loss": 0.7126,
      "step": 11932
    },
    {
      "epoch": 0.4945501264039123,
      "grad_norm": 0.41528499126434326,
      "learning_rate": 2.527456587508807e-06,
      "loss": 0.624,
      "step": 11933
    },
    {
      "epoch": 0.49459157030958595,
      "grad_norm": 0.3745574653148651,
      "learning_rate": 2.5272493679804387e-06,
      "loss": 0.6881,
      "step": 11934
    },
    {
      "epoch": 0.4946330142152596,
      "grad_norm": 0.4035796523094177,
      "learning_rate": 2.52704214845207e-06,
      "loss": 0.6477,
      "step": 11935
    },
    {
      "epoch": 0.4946744581209333,
      "grad_norm": 0.41015711426734924,
      "learning_rate": 2.526834928923702e-06,
      "loss": 0.7039,
      "step": 11936
    },
    {
      "epoch": 0.494715902026607,
      "grad_norm": 0.40707623958587646,
      "learning_rate": 2.5266277093953333e-06,
      "loss": 0.684,
      "step": 11937
    },
    {
      "epoch": 0.49475734593228066,
      "grad_norm": 0.4093051254749298,
      "learning_rate": 2.5264204898669655e-06,
      "loss": 0.7123,
      "step": 11938
    },
    {
      "epoch": 0.49479878983795433,
      "grad_norm": 0.4050680100917816,
      "learning_rate": 2.5262132703385965e-06,
      "loss": 0.684,
      "step": 11939
    },
    {
      "epoch": 0.494840233743628,
      "grad_norm": 0.4116615653038025,
      "learning_rate": 2.5260060508102287e-06,
      "loss": 0.6968,
      "step": 11940
    },
    {
      "epoch": 0.4948816776493017,
      "grad_norm": 0.42350536584854126,
      "learning_rate": 2.5257988312818605e-06,
      "loss": 0.7393,
      "step": 11941
    },
    {
      "epoch": 0.49492312155497536,
      "grad_norm": 0.4255143702030182,
      "learning_rate": 2.525591611753492e-06,
      "loss": 0.6296,
      "step": 11942
    },
    {
      "epoch": 0.494964565460649,
      "grad_norm": 0.40906867384910583,
      "learning_rate": 2.5253843922251237e-06,
      "loss": 0.6611,
      "step": 11943
    },
    {
      "epoch": 0.49500600936632266,
      "grad_norm": 0.4224621653556824,
      "learning_rate": 2.525177172696755e-06,
      "loss": 0.6786,
      "step": 11944
    },
    {
      "epoch": 0.49504745327199634,
      "grad_norm": 0.4067190885543823,
      "learning_rate": 2.524969953168387e-06,
      "loss": 0.6687,
      "step": 11945
    },
    {
      "epoch": 0.49508889717767,
      "grad_norm": 0.4177899956703186,
      "learning_rate": 2.5247627336400183e-06,
      "loss": 0.7036,
      "step": 11946
    },
    {
      "epoch": 0.4951303410833437,
      "grad_norm": 0.422874391078949,
      "learning_rate": 2.52455551411165e-06,
      "loss": 0.7231,
      "step": 11947
    },
    {
      "epoch": 0.49517178498901737,
      "grad_norm": 0.4232557415962219,
      "learning_rate": 2.524348294583282e-06,
      "loss": 0.6777,
      "step": 11948
    },
    {
      "epoch": 0.49521322889469105,
      "grad_norm": 0.41736260056495667,
      "learning_rate": 2.5241410750549133e-06,
      "loss": 0.7006,
      "step": 11949
    },
    {
      "epoch": 0.4952546728003647,
      "grad_norm": 0.4092543423175812,
      "learning_rate": 2.523933855526545e-06,
      "loss": 0.6897,
      "step": 11950
    },
    {
      "epoch": 0.4952961167060384,
      "grad_norm": 0.3978733718395233,
      "learning_rate": 2.5237266359981765e-06,
      "loss": 0.6851,
      "step": 11951
    },
    {
      "epoch": 0.495337560611712,
      "grad_norm": 0.4310804307460785,
      "learning_rate": 2.5235194164698083e-06,
      "loss": 0.6772,
      "step": 11952
    },
    {
      "epoch": 0.4953790045173857,
      "grad_norm": 0.38709092140197754,
      "learning_rate": 2.5233121969414397e-06,
      "loss": 0.6436,
      "step": 11953
    },
    {
      "epoch": 0.4954204484230594,
      "grad_norm": 0.38359251618385315,
      "learning_rate": 2.523104977413072e-06,
      "loss": 0.6632,
      "step": 11954
    },
    {
      "epoch": 0.49546189232873306,
      "grad_norm": 0.396247923374176,
      "learning_rate": 2.522897757884703e-06,
      "loss": 0.6625,
      "step": 11955
    },
    {
      "epoch": 0.49550333623440673,
      "grad_norm": 0.39747077226638794,
      "learning_rate": 2.522690538356335e-06,
      "loss": 0.6692,
      "step": 11956
    },
    {
      "epoch": 0.4955447801400804,
      "grad_norm": 0.3933205306529999,
      "learning_rate": 2.522483318827967e-06,
      "loss": 0.6923,
      "step": 11957
    },
    {
      "epoch": 0.4955862240457541,
      "grad_norm": 0.4358789324760437,
      "learning_rate": 2.5222760992995983e-06,
      "loss": 0.6799,
      "step": 11958
    },
    {
      "epoch": 0.49562766795142776,
      "grad_norm": 0.42939484119415283,
      "learning_rate": 2.52206887977123e-06,
      "loss": 0.6967,
      "step": 11959
    },
    {
      "epoch": 0.49566911185710144,
      "grad_norm": 0.43787631392478943,
      "learning_rate": 2.5218616602428615e-06,
      "loss": 0.6904,
      "step": 11960
    },
    {
      "epoch": 0.49571055576277506,
      "grad_norm": 0.3841419816017151,
      "learning_rate": 2.5216544407144933e-06,
      "loss": 0.6865,
      "step": 11961
    },
    {
      "epoch": 0.49575199966844874,
      "grad_norm": 0.4146205484867096,
      "learning_rate": 2.5214472211861247e-06,
      "loss": 0.7205,
      "step": 11962
    },
    {
      "epoch": 0.4957934435741224,
      "grad_norm": 0.40564408898353577,
      "learning_rate": 2.5212400016577565e-06,
      "loss": 0.6401,
      "step": 11963
    },
    {
      "epoch": 0.4958348874797961,
      "grad_norm": 0.4023638665676117,
      "learning_rate": 2.521032782129388e-06,
      "loss": 0.6882,
      "step": 11964
    },
    {
      "epoch": 0.49587633138546977,
      "grad_norm": 0.45414650440216064,
      "learning_rate": 2.5208255626010197e-06,
      "loss": 0.6908,
      "step": 11965
    },
    {
      "epoch": 0.49591777529114345,
      "grad_norm": 0.3852802515029907,
      "learning_rate": 2.5206183430726515e-06,
      "loss": 0.6484,
      "step": 11966
    },
    {
      "epoch": 0.4959592191968171,
      "grad_norm": 0.4603707194328308,
      "learning_rate": 2.520411123544283e-06,
      "loss": 0.7444,
      "step": 11967
    },
    {
      "epoch": 0.4960006631024908,
      "grad_norm": 0.4256884753704071,
      "learning_rate": 2.5202039040159147e-06,
      "loss": 0.7069,
      "step": 11968
    },
    {
      "epoch": 0.4960421070081644,
      "grad_norm": 0.42455315589904785,
      "learning_rate": 2.519996684487546e-06,
      "loss": 0.7349,
      "step": 11969
    },
    {
      "epoch": 0.4960835509138381,
      "grad_norm": 0.3844889998435974,
      "learning_rate": 2.519789464959178e-06,
      "loss": 0.7068,
      "step": 11970
    },
    {
      "epoch": 0.4961249948195118,
      "grad_norm": 0.3990664780139923,
      "learning_rate": 2.5195822454308093e-06,
      "loss": 0.6486,
      "step": 11971
    },
    {
      "epoch": 0.49616643872518545,
      "grad_norm": 0.41621580719947815,
      "learning_rate": 2.5193750259024415e-06,
      "loss": 0.671,
      "step": 11972
    },
    {
      "epoch": 0.49620788263085913,
      "grad_norm": 0.42466944456100464,
      "learning_rate": 2.5191678063740725e-06,
      "loss": 0.7122,
      "step": 11973
    },
    {
      "epoch": 0.4962493265365328,
      "grad_norm": 0.4108840227127075,
      "learning_rate": 2.5189605868457047e-06,
      "loss": 0.6799,
      "step": 11974
    },
    {
      "epoch": 0.4962907704422065,
      "grad_norm": 0.38877516984939575,
      "learning_rate": 2.5187533673173365e-06,
      "loss": 0.7058,
      "step": 11975
    },
    {
      "epoch": 0.49633221434788016,
      "grad_norm": 0.5147877931594849,
      "learning_rate": 2.518546147788968e-06,
      "loss": 0.6914,
      "step": 11976
    },
    {
      "epoch": 0.49637365825355384,
      "grad_norm": 0.4093729257583618,
      "learning_rate": 2.5183389282605997e-06,
      "loss": 0.7366,
      "step": 11977
    },
    {
      "epoch": 0.49641510215922746,
      "grad_norm": 0.41924306750297546,
      "learning_rate": 2.518131708732231e-06,
      "loss": 0.699,
      "step": 11978
    },
    {
      "epoch": 0.49645654606490114,
      "grad_norm": 0.42035865783691406,
      "learning_rate": 2.517924489203863e-06,
      "loss": 0.6772,
      "step": 11979
    },
    {
      "epoch": 0.4964979899705748,
      "grad_norm": 0.4675120711326599,
      "learning_rate": 2.5177172696754943e-06,
      "loss": 0.6583,
      "step": 11980
    },
    {
      "epoch": 0.4965394338762485,
      "grad_norm": 0.3918024003505707,
      "learning_rate": 2.517510050147126e-06,
      "loss": 0.6754,
      "step": 11981
    },
    {
      "epoch": 0.49658087778192217,
      "grad_norm": 0.41508054733276367,
      "learning_rate": 2.5173028306187575e-06,
      "loss": 0.6735,
      "step": 11982
    },
    {
      "epoch": 0.49662232168759585,
      "grad_norm": 0.35878026485443115,
      "learning_rate": 2.5170956110903893e-06,
      "loss": 0.6425,
      "step": 11983
    },
    {
      "epoch": 0.4966637655932695,
      "grad_norm": 0.3775622248649597,
      "learning_rate": 2.516888391562021e-06,
      "loss": 0.6436,
      "step": 11984
    },
    {
      "epoch": 0.4967052094989432,
      "grad_norm": 0.46326592564582825,
      "learning_rate": 2.5166811720336525e-06,
      "loss": 0.7006,
      "step": 11985
    },
    {
      "epoch": 0.4967466534046169,
      "grad_norm": 0.42945584654808044,
      "learning_rate": 2.5164739525052843e-06,
      "loss": 0.6573,
      "step": 11986
    },
    {
      "epoch": 0.4967880973102905,
      "grad_norm": 0.3993362784385681,
      "learning_rate": 2.5162667329769157e-06,
      "loss": 0.7063,
      "step": 11987
    },
    {
      "epoch": 0.4968295412159642,
      "grad_norm": 0.45028671622276306,
      "learning_rate": 2.516059513448548e-06,
      "loss": 0.7297,
      "step": 11988
    },
    {
      "epoch": 0.49687098512163785,
      "grad_norm": 0.3696768283843994,
      "learning_rate": 2.515852293920179e-06,
      "loss": 0.6998,
      "step": 11989
    },
    {
      "epoch": 0.49691242902731153,
      "grad_norm": 0.4076455533504486,
      "learning_rate": 2.515645074391811e-06,
      "loss": 0.6786,
      "step": 11990
    },
    {
      "epoch": 0.4969538729329852,
      "grad_norm": 0.43897855281829834,
      "learning_rate": 2.515437854863442e-06,
      "loss": 0.7318,
      "step": 11991
    },
    {
      "epoch": 0.4969953168386589,
      "grad_norm": 0.4200321435928345,
      "learning_rate": 2.5152306353350743e-06,
      "loss": 0.7026,
      "step": 11992
    },
    {
      "epoch": 0.49703676074433256,
      "grad_norm": 0.3863934874534607,
      "learning_rate": 2.515023415806706e-06,
      "loss": 0.6931,
      "step": 11993
    },
    {
      "epoch": 0.49707820465000624,
      "grad_norm": 0.4182608425617218,
      "learning_rate": 2.5148161962783375e-06,
      "loss": 0.7134,
      "step": 11994
    },
    {
      "epoch": 0.49711964855567986,
      "grad_norm": 0.39725589752197266,
      "learning_rate": 2.5146089767499693e-06,
      "loss": 0.7151,
      "step": 11995
    },
    {
      "epoch": 0.49716109246135354,
      "grad_norm": 0.40578627586364746,
      "learning_rate": 2.5144017572216007e-06,
      "loss": 0.679,
      "step": 11996
    },
    {
      "epoch": 0.4972025363670272,
      "grad_norm": 0.3958173990249634,
      "learning_rate": 2.5141945376932325e-06,
      "loss": 0.6537,
      "step": 11997
    },
    {
      "epoch": 0.4972439802727009,
      "grad_norm": 0.3821972608566284,
      "learning_rate": 2.513987318164864e-06,
      "loss": 0.6395,
      "step": 11998
    },
    {
      "epoch": 0.49728542417837457,
      "grad_norm": 0.42043787240982056,
      "learning_rate": 2.5137800986364957e-06,
      "loss": 0.7158,
      "step": 11999
    },
    {
      "epoch": 0.49732686808404825,
      "grad_norm": 0.396319717168808,
      "learning_rate": 2.513572879108127e-06,
      "loss": 0.7014,
      "step": 12000
    },
    {
      "epoch": 0.4973683119897219,
      "grad_norm": 0.40823930501937866,
      "learning_rate": 2.513365659579759e-06,
      "loss": 0.6682,
      "step": 12001
    },
    {
      "epoch": 0.4974097558953956,
      "grad_norm": 0.40652742981910706,
      "learning_rate": 2.5131584400513907e-06,
      "loss": 0.6765,
      "step": 12002
    },
    {
      "epoch": 0.4974511998010693,
      "grad_norm": 0.43065810203552246,
      "learning_rate": 2.512951220523022e-06,
      "loss": 0.7185,
      "step": 12003
    },
    {
      "epoch": 0.4974926437067429,
      "grad_norm": 0.3975633978843689,
      "learning_rate": 2.512744000994654e-06,
      "loss": 0.6215,
      "step": 12004
    },
    {
      "epoch": 0.4975340876124166,
      "grad_norm": 0.4000629186630249,
      "learning_rate": 2.5125367814662853e-06,
      "loss": 0.6833,
      "step": 12005
    },
    {
      "epoch": 0.49757553151809025,
      "grad_norm": 0.42067578434944153,
      "learning_rate": 2.5123295619379175e-06,
      "loss": 0.6821,
      "step": 12006
    },
    {
      "epoch": 0.49761697542376393,
      "grad_norm": 0.411510705947876,
      "learning_rate": 2.5121223424095485e-06,
      "loss": 0.7305,
      "step": 12007
    },
    {
      "epoch": 0.4976584193294376,
      "grad_norm": 0.44052401185035706,
      "learning_rate": 2.5119151228811807e-06,
      "loss": 0.709,
      "step": 12008
    },
    {
      "epoch": 0.4976998632351113,
      "grad_norm": 0.41767334938049316,
      "learning_rate": 2.5117079033528125e-06,
      "loss": 0.6865,
      "step": 12009
    },
    {
      "epoch": 0.49774130714078496,
      "grad_norm": 0.44363266229629517,
      "learning_rate": 2.511500683824444e-06,
      "loss": 0.7024,
      "step": 12010
    },
    {
      "epoch": 0.49778275104645864,
      "grad_norm": 0.432852178812027,
      "learning_rate": 2.5112934642960757e-06,
      "loss": 0.7156,
      "step": 12011
    },
    {
      "epoch": 0.4978241949521323,
      "grad_norm": 0.4528099298477173,
      "learning_rate": 2.511086244767707e-06,
      "loss": 0.6547,
      "step": 12012
    },
    {
      "epoch": 0.49786563885780594,
      "grad_norm": 0.42078375816345215,
      "learning_rate": 2.510879025239339e-06,
      "loss": 0.668,
      "step": 12013
    },
    {
      "epoch": 0.4979070827634796,
      "grad_norm": 0.4412442445755005,
      "learning_rate": 2.5106718057109703e-06,
      "loss": 0.7344,
      "step": 12014
    },
    {
      "epoch": 0.4979485266691533,
      "grad_norm": 0.39522692561149597,
      "learning_rate": 2.510464586182602e-06,
      "loss": 0.6765,
      "step": 12015
    },
    {
      "epoch": 0.49798997057482697,
      "grad_norm": 0.44233158230781555,
      "learning_rate": 2.5102573666542335e-06,
      "loss": 0.6833,
      "step": 12016
    },
    {
      "epoch": 0.49803141448050064,
      "grad_norm": 0.39103421568870544,
      "learning_rate": 2.5100501471258653e-06,
      "loss": 0.7344,
      "step": 12017
    },
    {
      "epoch": 0.4980728583861743,
      "grad_norm": 0.40546971559524536,
      "learning_rate": 2.509842927597497e-06,
      "loss": 0.6697,
      "step": 12018
    },
    {
      "epoch": 0.498114302291848,
      "grad_norm": 0.4304859936237335,
      "learning_rate": 2.5096357080691285e-06,
      "loss": 0.71,
      "step": 12019
    },
    {
      "epoch": 0.4981557461975217,
      "grad_norm": 0.412144273519516,
      "learning_rate": 2.5094284885407603e-06,
      "loss": 0.6788,
      "step": 12020
    },
    {
      "epoch": 0.4981971901031953,
      "grad_norm": 0.4435880482196808,
      "learning_rate": 2.5092212690123917e-06,
      "loss": 0.7385,
      "step": 12021
    },
    {
      "epoch": 0.498238634008869,
      "grad_norm": 0.3919011354446411,
      "learning_rate": 2.509014049484024e-06,
      "loss": 0.6604,
      "step": 12022
    },
    {
      "epoch": 0.49828007791454265,
      "grad_norm": 0.41757670044898987,
      "learning_rate": 2.508806829955655e-06,
      "loss": 0.6935,
      "step": 12023
    },
    {
      "epoch": 0.49832152182021633,
      "grad_norm": 0.39835837483406067,
      "learning_rate": 2.508599610427287e-06,
      "loss": 0.7051,
      "step": 12024
    },
    {
      "epoch": 0.49836296572589,
      "grad_norm": 0.4181268513202667,
      "learning_rate": 2.508392390898918e-06,
      "loss": 0.6749,
      "step": 12025
    },
    {
      "epoch": 0.4984044096315637,
      "grad_norm": 0.434345543384552,
      "learning_rate": 2.5081851713705503e-06,
      "loss": 0.7385,
      "step": 12026
    },
    {
      "epoch": 0.49844585353723736,
      "grad_norm": 0.410093754529953,
      "learning_rate": 2.507977951842182e-06,
      "loss": 0.6626,
      "step": 12027
    },
    {
      "epoch": 0.49848729744291104,
      "grad_norm": 0.4303359091281891,
      "learning_rate": 2.5077707323138135e-06,
      "loss": 0.702,
      "step": 12028
    },
    {
      "epoch": 0.4985287413485847,
      "grad_norm": 0.4118058681488037,
      "learning_rate": 2.5075635127854453e-06,
      "loss": 0.6572,
      "step": 12029
    },
    {
      "epoch": 0.49857018525425834,
      "grad_norm": 0.4294072985649109,
      "learning_rate": 2.5073562932570767e-06,
      "loss": 0.7041,
      "step": 12030
    },
    {
      "epoch": 0.498611629159932,
      "grad_norm": 0.443994402885437,
      "learning_rate": 2.5071490737287085e-06,
      "loss": 0.7014,
      "step": 12031
    },
    {
      "epoch": 0.4986530730656057,
      "grad_norm": 0.40063682198524475,
      "learning_rate": 2.50694185420034e-06,
      "loss": 0.6846,
      "step": 12032
    },
    {
      "epoch": 0.49869451697127937,
      "grad_norm": 0.41289472579956055,
      "learning_rate": 2.5067346346719717e-06,
      "loss": 0.6338,
      "step": 12033
    },
    {
      "epoch": 0.49873596087695304,
      "grad_norm": 0.4603697955608368,
      "learning_rate": 2.506527415143603e-06,
      "loss": 0.7336,
      "step": 12034
    },
    {
      "epoch": 0.4987774047826267,
      "grad_norm": 0.4044231176376343,
      "learning_rate": 2.506320195615235e-06,
      "loss": 0.6689,
      "step": 12035
    },
    {
      "epoch": 0.4988188486883004,
      "grad_norm": 0.4238329231739044,
      "learning_rate": 2.5061129760868667e-06,
      "loss": 0.6941,
      "step": 12036
    },
    {
      "epoch": 0.4988602925939741,
      "grad_norm": 0.4304075241088867,
      "learning_rate": 2.505905756558498e-06,
      "loss": 0.7002,
      "step": 12037
    },
    {
      "epoch": 0.49890173649964775,
      "grad_norm": 0.4487854838371277,
      "learning_rate": 2.50569853703013e-06,
      "loss": 0.7612,
      "step": 12038
    },
    {
      "epoch": 0.4989431804053214,
      "grad_norm": 0.40556544065475464,
      "learning_rate": 2.5054913175017613e-06,
      "loss": 0.6733,
      "step": 12039
    },
    {
      "epoch": 0.49898462431099505,
      "grad_norm": 0.4324802756309509,
      "learning_rate": 2.5052840979733935e-06,
      "loss": 0.7173,
      "step": 12040
    },
    {
      "epoch": 0.49902606821666873,
      "grad_norm": 0.4120703935623169,
      "learning_rate": 2.5050768784450245e-06,
      "loss": 0.639,
      "step": 12041
    },
    {
      "epoch": 0.4990675121223424,
      "grad_norm": 0.39611753821372986,
      "learning_rate": 2.5048696589166567e-06,
      "loss": 0.677,
      "step": 12042
    },
    {
      "epoch": 0.4991089560280161,
      "grad_norm": 0.39439111948013306,
      "learning_rate": 2.504662439388288e-06,
      "loss": 0.6763,
      "step": 12043
    },
    {
      "epoch": 0.49915039993368976,
      "grad_norm": 0.41333523392677307,
      "learning_rate": 2.50445521985992e-06,
      "loss": 0.6274,
      "step": 12044
    },
    {
      "epoch": 0.49919184383936344,
      "grad_norm": 0.41353872418403625,
      "learning_rate": 2.5042480003315517e-06,
      "loss": 0.729,
      "step": 12045
    },
    {
      "epoch": 0.4992332877450371,
      "grad_norm": 0.4190951883792877,
      "learning_rate": 2.504040780803183e-06,
      "loss": 0.6963,
      "step": 12046
    },
    {
      "epoch": 0.4992747316507108,
      "grad_norm": 0.4346992075443268,
      "learning_rate": 2.503833561274815e-06,
      "loss": 0.7247,
      "step": 12047
    },
    {
      "epoch": 0.4993161755563844,
      "grad_norm": 0.40948501229286194,
      "learning_rate": 2.5036263417464463e-06,
      "loss": 0.6721,
      "step": 12048
    },
    {
      "epoch": 0.4993576194620581,
      "grad_norm": 0.41514986753463745,
      "learning_rate": 2.503419122218078e-06,
      "loss": 0.6984,
      "step": 12049
    },
    {
      "epoch": 0.49939906336773177,
      "grad_norm": 0.39792993664741516,
      "learning_rate": 2.5032119026897095e-06,
      "loss": 0.6943,
      "step": 12050
    },
    {
      "epoch": 0.49944050727340544,
      "grad_norm": 0.43684446811676025,
      "learning_rate": 2.5030046831613413e-06,
      "loss": 0.6915,
      "step": 12051
    },
    {
      "epoch": 0.4994819511790791,
      "grad_norm": 0.40352702140808105,
      "learning_rate": 2.5027974636329727e-06,
      "loss": 0.6451,
      "step": 12052
    },
    {
      "epoch": 0.4995233950847528,
      "grad_norm": 0.4345749318599701,
      "learning_rate": 2.5025902441046045e-06,
      "loss": 0.6938,
      "step": 12053
    },
    {
      "epoch": 0.4995648389904265,
      "grad_norm": 0.41432181000709534,
      "learning_rate": 2.5023830245762363e-06,
      "loss": 0.6434,
      "step": 12054
    },
    {
      "epoch": 0.49960628289610015,
      "grad_norm": 0.3918708264827728,
      "learning_rate": 2.5021758050478677e-06,
      "loss": 0.6842,
      "step": 12055
    },
    {
      "epoch": 0.4996477268017738,
      "grad_norm": 0.3903842270374298,
      "learning_rate": 2.5019685855195e-06,
      "loss": 0.6638,
      "step": 12056
    },
    {
      "epoch": 0.49968917070744745,
      "grad_norm": 0.4764392077922821,
      "learning_rate": 2.501761365991131e-06,
      "loss": 0.6969,
      "step": 12057
    },
    {
      "epoch": 0.4997306146131211,
      "grad_norm": 0.417617529630661,
      "learning_rate": 2.501554146462763e-06,
      "loss": 0.6781,
      "step": 12058
    },
    {
      "epoch": 0.4997720585187948,
      "grad_norm": 0.40553587675094604,
      "learning_rate": 2.5013469269343945e-06,
      "loss": 0.6853,
      "step": 12059
    },
    {
      "epoch": 0.4998135024244685,
      "grad_norm": 0.42576566338539124,
      "learning_rate": 2.5011397074060263e-06,
      "loss": 0.689,
      "step": 12060
    },
    {
      "epoch": 0.49985494633014216,
      "grad_norm": 0.39877843856811523,
      "learning_rate": 2.5009324878776577e-06,
      "loss": 0.7032,
      "step": 12061
    },
    {
      "epoch": 0.49989639023581584,
      "grad_norm": 0.4165988862514496,
      "learning_rate": 2.5007252683492895e-06,
      "loss": 0.7024,
      "step": 12062
    },
    {
      "epoch": 0.4999378341414895,
      "grad_norm": 0.3979125916957855,
      "learning_rate": 2.5005180488209213e-06,
      "loss": 0.6545,
      "step": 12063
    },
    {
      "epoch": 0.4999792780471632,
      "grad_norm": 0.39304810762405396,
      "learning_rate": 2.5003108292925527e-06,
      "loss": 0.6475,
      "step": 12064
    },
    {
      "epoch": 0.5000207219528369,
      "grad_norm": 0.3913843035697937,
      "learning_rate": 2.5001036097641845e-06,
      "loss": 0.6804,
      "step": 12065
    },
    {
      "epoch": 0.5000621658585105,
      "grad_norm": 0.392557829618454,
      "learning_rate": 2.4998963902358163e-06,
      "loss": 0.6113,
      "step": 12066
    },
    {
      "epoch": 0.5001036097641842,
      "grad_norm": 0.4007066488265991,
      "learning_rate": 2.4996891707074477e-06,
      "loss": 0.7057,
      "step": 12067
    },
    {
      "epoch": 0.5001450536698578,
      "grad_norm": 0.3967163562774658,
      "learning_rate": 2.4994819511790795e-06,
      "loss": 0.689,
      "step": 12068
    },
    {
      "epoch": 0.5001864975755315,
      "grad_norm": 0.4334842264652252,
      "learning_rate": 2.499274731650711e-06,
      "loss": 0.6893,
      "step": 12069
    },
    {
      "epoch": 0.5002279414812052,
      "grad_norm": 0.44219881296157837,
      "learning_rate": 2.4990675121223427e-06,
      "loss": 0.6875,
      "step": 12070
    },
    {
      "epoch": 0.5002693853868788,
      "grad_norm": 0.39369404315948486,
      "learning_rate": 2.498860292593974e-06,
      "loss": 0.6688,
      "step": 12071
    },
    {
      "epoch": 0.5003108292925526,
      "grad_norm": 0.40580645203590393,
      "learning_rate": 2.498653073065606e-06,
      "loss": 0.6802,
      "step": 12072
    },
    {
      "epoch": 0.5003522731982262,
      "grad_norm": 0.39442992210388184,
      "learning_rate": 2.4984458535372373e-06,
      "loss": 0.652,
      "step": 12073
    },
    {
      "epoch": 0.5003937171038999,
      "grad_norm": 0.38728371262550354,
      "learning_rate": 2.498238634008869e-06,
      "loss": 0.6328,
      "step": 12074
    },
    {
      "epoch": 0.5004351610095735,
      "grad_norm": 0.4222020208835602,
      "learning_rate": 2.498031414480501e-06,
      "loss": 0.7429,
      "step": 12075
    },
    {
      "epoch": 0.5004766049152473,
      "grad_norm": 0.3780023753643036,
      "learning_rate": 2.4978241949521327e-06,
      "loss": 0.6898,
      "step": 12076
    },
    {
      "epoch": 0.5005180488209209,
      "grad_norm": 0.38240793347358704,
      "learning_rate": 2.497616975423764e-06,
      "loss": 0.6938,
      "step": 12077
    },
    {
      "epoch": 0.5005594927265945,
      "grad_norm": 0.48400115966796875,
      "learning_rate": 2.497409755895396e-06,
      "loss": 0.7175,
      "step": 12078
    },
    {
      "epoch": 0.5006009366322682,
      "grad_norm": 0.422498881816864,
      "learning_rate": 2.4972025363670273e-06,
      "loss": 0.7158,
      "step": 12079
    },
    {
      "epoch": 0.5006423805379419,
      "grad_norm": 0.40404435992240906,
      "learning_rate": 2.496995316838659e-06,
      "loss": 0.6372,
      "step": 12080
    },
    {
      "epoch": 0.5006838244436156,
      "grad_norm": 0.4048217535018921,
      "learning_rate": 2.4967880973102905e-06,
      "loss": 0.6633,
      "step": 12081
    },
    {
      "epoch": 0.5007252683492892,
      "grad_norm": 0.465869665145874,
      "learning_rate": 2.4965808777819223e-06,
      "loss": 0.7322,
      "step": 12082
    },
    {
      "epoch": 0.5007667122549629,
      "grad_norm": 0.3907471001148224,
      "learning_rate": 2.496373658253554e-06,
      "loss": 0.6674,
      "step": 12083
    },
    {
      "epoch": 0.5008081561606366,
      "grad_norm": 0.4338460862636566,
      "learning_rate": 2.496166438725186e-06,
      "loss": 0.7615,
      "step": 12084
    },
    {
      "epoch": 0.5008496000663103,
      "grad_norm": 0.41220352053642273,
      "learning_rate": 2.4959592191968173e-06,
      "loss": 0.625,
      "step": 12085
    },
    {
      "epoch": 0.5008910439719839,
      "grad_norm": 0.3711015284061432,
      "learning_rate": 2.495751999668449e-06,
      "loss": 0.632,
      "step": 12086
    },
    {
      "epoch": 0.5009324878776575,
      "grad_norm": 0.3978126049041748,
      "learning_rate": 2.4955447801400805e-06,
      "loss": 0.6635,
      "step": 12087
    },
    {
      "epoch": 0.5009739317833313,
      "grad_norm": 0.3690616488456726,
      "learning_rate": 2.4953375606117123e-06,
      "loss": 0.6504,
      "step": 12088
    },
    {
      "epoch": 0.5010153756890049,
      "grad_norm": 0.43138614296913147,
      "learning_rate": 2.4951303410833437e-06,
      "loss": 0.6978,
      "step": 12089
    },
    {
      "epoch": 0.5010568195946786,
      "grad_norm": 0.38003459572792053,
      "learning_rate": 2.4949231215549755e-06,
      "loss": 0.7004,
      "step": 12090
    },
    {
      "epoch": 0.5010982635003522,
      "grad_norm": 0.41197532415390015,
      "learning_rate": 2.494715902026607e-06,
      "loss": 0.6915,
      "step": 12091
    },
    {
      "epoch": 0.501139707406026,
      "grad_norm": 0.399027019739151,
      "learning_rate": 2.494508682498239e-06,
      "loss": 0.7029,
      "step": 12092
    },
    {
      "epoch": 0.5011811513116996,
      "grad_norm": 0.3783413767814636,
      "learning_rate": 2.4943014629698705e-06,
      "loss": 0.6754,
      "step": 12093
    },
    {
      "epoch": 0.5012225952173733,
      "grad_norm": 0.3827938437461853,
      "learning_rate": 2.4940942434415023e-06,
      "loss": 0.6885,
      "step": 12094
    },
    {
      "epoch": 0.501264039123047,
      "grad_norm": 0.42406222224235535,
      "learning_rate": 2.4938870239131337e-06,
      "loss": 0.7034,
      "step": 12095
    },
    {
      "epoch": 0.5013054830287206,
      "grad_norm": 0.44488710165023804,
      "learning_rate": 2.4936798043847655e-06,
      "loss": 0.6843,
      "step": 12096
    },
    {
      "epoch": 0.5013469269343943,
      "grad_norm": 0.4248327612876892,
      "learning_rate": 2.493472584856397e-06,
      "loss": 0.7112,
      "step": 12097
    },
    {
      "epoch": 0.5013883708400679,
      "grad_norm": 0.41129153966903687,
      "learning_rate": 2.4932653653280287e-06,
      "loss": 0.6947,
      "step": 12098
    },
    {
      "epoch": 0.5014298147457417,
      "grad_norm": 0.3921922445297241,
      "learning_rate": 2.49305814579966e-06,
      "loss": 0.6155,
      "step": 12099
    },
    {
      "epoch": 0.5014712586514153,
      "grad_norm": 0.42110082507133484,
      "learning_rate": 2.492850926271292e-06,
      "loss": 0.6636,
      "step": 12100
    },
    {
      "epoch": 0.501512702557089,
      "grad_norm": 0.3675326406955719,
      "learning_rate": 2.4926437067429237e-06,
      "loss": 0.6815,
      "step": 12101
    },
    {
      "epoch": 0.5015541464627626,
      "grad_norm": 0.41612574458122253,
      "learning_rate": 2.4924364872145555e-06,
      "loss": 0.6869,
      "step": 12102
    },
    {
      "epoch": 0.5015955903684364,
      "grad_norm": 0.3758845329284668,
      "learning_rate": 2.492229267686187e-06,
      "loss": 0.6715,
      "step": 12103
    },
    {
      "epoch": 0.50163703427411,
      "grad_norm": 0.4302389323711395,
      "learning_rate": 2.4920220481578187e-06,
      "loss": 0.6724,
      "step": 12104
    },
    {
      "epoch": 0.5016784781797836,
      "grad_norm": 0.40757232904434204,
      "learning_rate": 2.49181482862945e-06,
      "loss": 0.6792,
      "step": 12105
    },
    {
      "epoch": 0.5017199220854573,
      "grad_norm": 0.4076420068740845,
      "learning_rate": 2.491607609101082e-06,
      "loss": 0.7021,
      "step": 12106
    },
    {
      "epoch": 0.501761365991131,
      "grad_norm": 0.5484263896942139,
      "learning_rate": 2.4914003895727133e-06,
      "loss": 0.7571,
      "step": 12107
    },
    {
      "epoch": 0.5018028098968047,
      "grad_norm": 0.40297257900238037,
      "learning_rate": 2.491193170044345e-06,
      "loss": 0.71,
      "step": 12108
    },
    {
      "epoch": 0.5018442538024783,
      "grad_norm": 0.4634743332862854,
      "learning_rate": 2.4909859505159765e-06,
      "loss": 0.8298,
      "step": 12109
    },
    {
      "epoch": 0.5018856977081521,
      "grad_norm": 0.42236819863319397,
      "learning_rate": 2.4907787309876087e-06,
      "loss": 0.7231,
      "step": 12110
    },
    {
      "epoch": 0.5019271416138257,
      "grad_norm": 0.4364050030708313,
      "learning_rate": 2.49057151145924e-06,
      "loss": 0.6849,
      "step": 12111
    },
    {
      "epoch": 0.5019685855194994,
      "grad_norm": 0.4014566242694855,
      "learning_rate": 2.490364291930872e-06,
      "loss": 0.6598,
      "step": 12112
    },
    {
      "epoch": 0.502010029425173,
      "grad_norm": 0.4068412780761719,
      "learning_rate": 2.4901570724025033e-06,
      "loss": 0.6827,
      "step": 12113
    },
    {
      "epoch": 0.5020514733308467,
      "grad_norm": 0.42305973172187805,
      "learning_rate": 2.489949852874135e-06,
      "loss": 0.6848,
      "step": 12114
    },
    {
      "epoch": 0.5020929172365204,
      "grad_norm": 0.4540117383003235,
      "learning_rate": 2.4897426333457665e-06,
      "loss": 0.6947,
      "step": 12115
    },
    {
      "epoch": 0.502134361142194,
      "grad_norm": 0.43764641880989075,
      "learning_rate": 2.4895354138173983e-06,
      "loss": 0.6855,
      "step": 12116
    },
    {
      "epoch": 0.5021758050478677,
      "grad_norm": 0.3906167149543762,
      "learning_rate": 2.4893281942890297e-06,
      "loss": 0.7195,
      "step": 12117
    },
    {
      "epoch": 0.5022172489535414,
      "grad_norm": 0.39837321639060974,
      "learning_rate": 2.489120974760662e-06,
      "loss": 0.6879,
      "step": 12118
    },
    {
      "epoch": 0.5022586928592151,
      "grad_norm": 0.3770964741706848,
      "learning_rate": 2.4889137552322933e-06,
      "loss": 0.6602,
      "step": 12119
    },
    {
      "epoch": 0.5023001367648887,
      "grad_norm": 0.41646072268486023,
      "learning_rate": 2.488706535703925e-06,
      "loss": 0.6665,
      "step": 12120
    },
    {
      "epoch": 0.5023415806705624,
      "grad_norm": 0.42203015089035034,
      "learning_rate": 2.4884993161755565e-06,
      "loss": 0.668,
      "step": 12121
    },
    {
      "epoch": 0.5023830245762361,
      "grad_norm": 0.396767795085907,
      "learning_rate": 2.4882920966471883e-06,
      "loss": 0.6805,
      "step": 12122
    },
    {
      "epoch": 0.5024244684819097,
      "grad_norm": 0.4128914475440979,
      "learning_rate": 2.4880848771188197e-06,
      "loss": 0.6931,
      "step": 12123
    },
    {
      "epoch": 0.5024659123875834,
      "grad_norm": 0.4730176031589508,
      "learning_rate": 2.4878776575904515e-06,
      "loss": 0.7441,
      "step": 12124
    },
    {
      "epoch": 0.502507356293257,
      "grad_norm": 0.4274924099445343,
      "learning_rate": 2.487670438062083e-06,
      "loss": 0.7002,
      "step": 12125
    },
    {
      "epoch": 0.5025488001989308,
      "grad_norm": 0.42004191875457764,
      "learning_rate": 2.4874632185337147e-06,
      "loss": 0.7004,
      "step": 12126
    },
    {
      "epoch": 0.5025902441046044,
      "grad_norm": 0.4135735332965851,
      "learning_rate": 2.4872559990053465e-06,
      "loss": 0.6881,
      "step": 12127
    },
    {
      "epoch": 0.5026316880102781,
      "grad_norm": 0.3836991786956787,
      "learning_rate": 2.4870487794769783e-06,
      "loss": 0.6882,
      "step": 12128
    },
    {
      "epoch": 0.5026731319159518,
      "grad_norm": 0.3888266682624817,
      "learning_rate": 2.4868415599486097e-06,
      "loss": 0.7007,
      "step": 12129
    },
    {
      "epoch": 0.5027145758216254,
      "grad_norm": 0.4015958309173584,
      "learning_rate": 2.4866343404202415e-06,
      "loss": 0.6921,
      "step": 12130
    },
    {
      "epoch": 0.5027560197272991,
      "grad_norm": 0.4097941219806671,
      "learning_rate": 2.486427120891873e-06,
      "loss": 0.696,
      "step": 12131
    },
    {
      "epoch": 0.5027974636329727,
      "grad_norm": 0.40690216422080994,
      "learning_rate": 2.4862199013635047e-06,
      "loss": 0.71,
      "step": 12132
    },
    {
      "epoch": 0.5028389075386465,
      "grad_norm": 0.41904616355895996,
      "learning_rate": 2.486012681835136e-06,
      "loss": 0.6675,
      "step": 12133
    },
    {
      "epoch": 0.5028803514443201,
      "grad_norm": 0.39571309089660645,
      "learning_rate": 2.485805462306768e-06,
      "loss": 0.6617,
      "step": 12134
    },
    {
      "epoch": 0.5029217953499938,
      "grad_norm": 0.4086185097694397,
      "learning_rate": 2.4855982427783993e-06,
      "loss": 0.6926,
      "step": 12135
    },
    {
      "epoch": 0.5029632392556674,
      "grad_norm": 0.42128175497055054,
      "learning_rate": 2.4853910232500315e-06,
      "loss": 0.6322,
      "step": 12136
    },
    {
      "epoch": 0.5030046831613412,
      "grad_norm": 0.40756407380104065,
      "learning_rate": 2.485183803721663e-06,
      "loss": 0.666,
      "step": 12137
    },
    {
      "epoch": 0.5030461270670148,
      "grad_norm": 0.42321527004241943,
      "learning_rate": 2.4849765841932947e-06,
      "loss": 0.6799,
      "step": 12138
    },
    {
      "epoch": 0.5030875709726884,
      "grad_norm": 0.4023328125476837,
      "learning_rate": 2.484769364664926e-06,
      "loss": 0.6775,
      "step": 12139
    },
    {
      "epoch": 0.5031290148783621,
      "grad_norm": 0.3966747522354126,
      "learning_rate": 2.484562145136558e-06,
      "loss": 0.7302,
      "step": 12140
    },
    {
      "epoch": 0.5031704587840358,
      "grad_norm": 0.40463635325431824,
      "learning_rate": 2.4843549256081893e-06,
      "loss": 0.7241,
      "step": 12141
    },
    {
      "epoch": 0.5032119026897095,
      "grad_norm": 0.4534701704978943,
      "learning_rate": 2.484147706079821e-06,
      "loss": 0.7119,
      "step": 12142
    },
    {
      "epoch": 0.5032533465953831,
      "grad_norm": 0.41324228048324585,
      "learning_rate": 2.4839404865514525e-06,
      "loss": 0.7239,
      "step": 12143
    },
    {
      "epoch": 0.5032947905010569,
      "grad_norm": 0.40443310141563416,
      "learning_rate": 2.4837332670230847e-06,
      "loss": 0.7285,
      "step": 12144
    },
    {
      "epoch": 0.5033362344067305,
      "grad_norm": 0.44884440302848816,
      "learning_rate": 2.483526047494716e-06,
      "loss": 0.7341,
      "step": 12145
    },
    {
      "epoch": 0.5033776783124042,
      "grad_norm": 0.4210149049758911,
      "learning_rate": 2.483318827966348e-06,
      "loss": 0.7056,
      "step": 12146
    },
    {
      "epoch": 0.5034191222180778,
      "grad_norm": 0.40544062852859497,
      "learning_rate": 2.4831116084379793e-06,
      "loss": 0.652,
      "step": 12147
    },
    {
      "epoch": 0.5034605661237515,
      "grad_norm": 0.405346155166626,
      "learning_rate": 2.482904388909611e-06,
      "loss": 0.6924,
      "step": 12148
    },
    {
      "epoch": 0.5035020100294252,
      "grad_norm": 0.38160941004753113,
      "learning_rate": 2.4826971693812425e-06,
      "loss": 0.7017,
      "step": 12149
    },
    {
      "epoch": 0.5035434539350988,
      "grad_norm": 0.4462243616580963,
      "learning_rate": 2.4824899498528743e-06,
      "loss": 0.6898,
      "step": 12150
    },
    {
      "epoch": 0.5035848978407725,
      "grad_norm": 0.3990682363510132,
      "learning_rate": 2.4822827303245057e-06,
      "loss": 0.7202,
      "step": 12151
    },
    {
      "epoch": 0.5036263417464462,
      "grad_norm": 0.3961663842201233,
      "learning_rate": 2.4820755107961375e-06,
      "loss": 0.6577,
      "step": 12152
    },
    {
      "epoch": 0.5036677856521199,
      "grad_norm": 0.43611231446266174,
      "learning_rate": 2.4818682912677693e-06,
      "loss": 0.6826,
      "step": 12153
    },
    {
      "epoch": 0.5037092295577935,
      "grad_norm": 0.402664452791214,
      "learning_rate": 2.481661071739401e-06,
      "loss": 0.6847,
      "step": 12154
    },
    {
      "epoch": 0.5037506734634672,
      "grad_norm": 0.41429275274276733,
      "learning_rate": 2.4814538522110325e-06,
      "loss": 0.7056,
      "step": 12155
    },
    {
      "epoch": 0.5037921173691409,
      "grad_norm": 0.44511136412620544,
      "learning_rate": 2.4812466326826643e-06,
      "loss": 0.6981,
      "step": 12156
    },
    {
      "epoch": 0.5038335612748145,
      "grad_norm": 0.4469084143638611,
      "learning_rate": 2.4810394131542957e-06,
      "loss": 0.7123,
      "step": 12157
    },
    {
      "epoch": 0.5038750051804882,
      "grad_norm": 0.43722930550575256,
      "learning_rate": 2.4808321936259275e-06,
      "loss": 0.6898,
      "step": 12158
    },
    {
      "epoch": 0.5039164490861618,
      "grad_norm": 0.41285890340805054,
      "learning_rate": 2.480624974097559e-06,
      "loss": 0.6572,
      "step": 12159
    },
    {
      "epoch": 0.5039578929918356,
      "grad_norm": 0.40923431515693665,
      "learning_rate": 2.4804177545691907e-06,
      "loss": 0.7046,
      "step": 12160
    },
    {
      "epoch": 0.5039993368975092,
      "grad_norm": 0.6008912324905396,
      "learning_rate": 2.4802105350408225e-06,
      "loss": 0.7225,
      "step": 12161
    },
    {
      "epoch": 0.5040407808031829,
      "grad_norm": 0.4196130335330963,
      "learning_rate": 2.4800033155124543e-06,
      "loss": 0.7048,
      "step": 12162
    },
    {
      "epoch": 0.5040822247088566,
      "grad_norm": 0.4156739115715027,
      "learning_rate": 2.4797960959840857e-06,
      "loss": 0.7347,
      "step": 12163
    },
    {
      "epoch": 0.5041236686145303,
      "grad_norm": 0.42003384232521057,
      "learning_rate": 2.4795888764557175e-06,
      "loss": 0.6791,
      "step": 12164
    },
    {
      "epoch": 0.5041651125202039,
      "grad_norm": 0.420210599899292,
      "learning_rate": 2.479381656927349e-06,
      "loss": 0.6566,
      "step": 12165
    },
    {
      "epoch": 0.5042065564258775,
      "grad_norm": 0.4262768626213074,
      "learning_rate": 2.4791744373989807e-06,
      "loss": 0.7219,
      "step": 12166
    },
    {
      "epoch": 0.5042480003315513,
      "grad_norm": 0.43248069286346436,
      "learning_rate": 2.478967217870612e-06,
      "loss": 0.7002,
      "step": 12167
    },
    {
      "epoch": 0.5042894442372249,
      "grad_norm": 0.4152391254901886,
      "learning_rate": 2.478759998342244e-06,
      "loss": 0.7025,
      "step": 12168
    },
    {
      "epoch": 0.5043308881428986,
      "grad_norm": 0.42801567912101746,
      "learning_rate": 2.4785527788138757e-06,
      "loss": 0.6771,
      "step": 12169
    },
    {
      "epoch": 0.5043723320485722,
      "grad_norm": 0.39395156502723694,
      "learning_rate": 2.4783455592855075e-06,
      "loss": 0.688,
      "step": 12170
    },
    {
      "epoch": 0.504413775954246,
      "grad_norm": 0.489948034286499,
      "learning_rate": 2.478138339757139e-06,
      "loss": 0.7026,
      "step": 12171
    },
    {
      "epoch": 0.5044552198599196,
      "grad_norm": 0.4414099156856537,
      "learning_rate": 2.4779311202287707e-06,
      "loss": 0.7036,
      "step": 12172
    },
    {
      "epoch": 0.5044966637655933,
      "grad_norm": 0.38853156566619873,
      "learning_rate": 2.477723900700402e-06,
      "loss": 0.6208,
      "step": 12173
    },
    {
      "epoch": 0.5045381076712669,
      "grad_norm": 0.3727603256702423,
      "learning_rate": 2.477516681172034e-06,
      "loss": 0.6548,
      "step": 12174
    },
    {
      "epoch": 0.5045795515769406,
      "grad_norm": 0.4071371853351593,
      "learning_rate": 2.4773094616436653e-06,
      "loss": 0.6714,
      "step": 12175
    },
    {
      "epoch": 0.5046209954826143,
      "grad_norm": 0.40217652916908264,
      "learning_rate": 2.477102242115297e-06,
      "loss": 0.6838,
      "step": 12176
    },
    {
      "epoch": 0.5046624393882879,
      "grad_norm": 0.41850778460502625,
      "learning_rate": 2.4768950225869285e-06,
      "loss": 0.7446,
      "step": 12177
    },
    {
      "epoch": 0.5047038832939617,
      "grad_norm": 0.48072972893714905,
      "learning_rate": 2.4766878030585603e-06,
      "loss": 0.7186,
      "step": 12178
    },
    {
      "epoch": 0.5047453271996353,
      "grad_norm": 0.4234667122364044,
      "learning_rate": 2.476480583530192e-06,
      "loss": 0.7129,
      "step": 12179
    },
    {
      "epoch": 0.504786771105309,
      "grad_norm": 0.4175702631473541,
      "learning_rate": 2.476273364001824e-06,
      "loss": 0.7106,
      "step": 12180
    },
    {
      "epoch": 0.5048282150109826,
      "grad_norm": 0.4012846052646637,
      "learning_rate": 2.4760661444734553e-06,
      "loss": 0.696,
      "step": 12181
    },
    {
      "epoch": 0.5048696589166564,
      "grad_norm": 0.46382564306259155,
      "learning_rate": 2.475858924945087e-06,
      "loss": 0.6608,
      "step": 12182
    },
    {
      "epoch": 0.50491110282233,
      "grad_norm": 0.4569469392299652,
      "learning_rate": 2.4756517054167185e-06,
      "loss": 0.7263,
      "step": 12183
    },
    {
      "epoch": 0.5049525467280036,
      "grad_norm": 0.4094659686088562,
      "learning_rate": 2.4754444858883503e-06,
      "loss": 0.6797,
      "step": 12184
    },
    {
      "epoch": 0.5049939906336773,
      "grad_norm": 0.41655343770980835,
      "learning_rate": 2.4752372663599817e-06,
      "loss": 0.6924,
      "step": 12185
    },
    {
      "epoch": 0.505035434539351,
      "grad_norm": 0.4058151841163635,
      "learning_rate": 2.4750300468316135e-06,
      "loss": 0.7302,
      "step": 12186
    },
    {
      "epoch": 0.5050768784450247,
      "grad_norm": 0.40714722871780396,
      "learning_rate": 2.4748228273032453e-06,
      "loss": 0.73,
      "step": 12187
    },
    {
      "epoch": 0.5051183223506983,
      "grad_norm": 0.3988172709941864,
      "learning_rate": 2.474615607774877e-06,
      "loss": 0.6741,
      "step": 12188
    },
    {
      "epoch": 0.505159766256372,
      "grad_norm": 0.4347361624240875,
      "learning_rate": 2.4744083882465085e-06,
      "loss": 0.6609,
      "step": 12189
    },
    {
      "epoch": 0.5052012101620457,
      "grad_norm": 0.4218306243419647,
      "learning_rate": 2.4742011687181403e-06,
      "loss": 0.686,
      "step": 12190
    },
    {
      "epoch": 0.5052426540677193,
      "grad_norm": 0.41292136907577515,
      "learning_rate": 2.4739939491897717e-06,
      "loss": 0.6637,
      "step": 12191
    },
    {
      "epoch": 0.505284097973393,
      "grad_norm": 0.4476361572742462,
      "learning_rate": 2.4737867296614035e-06,
      "loss": 0.66,
      "step": 12192
    },
    {
      "epoch": 0.5053255418790666,
      "grad_norm": 0.4280683696269989,
      "learning_rate": 2.473579510133035e-06,
      "loss": 0.6466,
      "step": 12193
    },
    {
      "epoch": 0.5053669857847404,
      "grad_norm": 0.389683336019516,
      "learning_rate": 2.4733722906046667e-06,
      "loss": 0.6467,
      "step": 12194
    },
    {
      "epoch": 0.505408429690414,
      "grad_norm": 0.4336313307285309,
      "learning_rate": 2.4731650710762985e-06,
      "loss": 0.7227,
      "step": 12195
    },
    {
      "epoch": 0.5054498735960877,
      "grad_norm": 0.41722410917282104,
      "learning_rate": 2.47295785154793e-06,
      "loss": 0.6914,
      "step": 12196
    },
    {
      "epoch": 0.5054913175017614,
      "grad_norm": 0.41265353560447693,
      "learning_rate": 2.4727506320195617e-06,
      "loss": 0.7083,
      "step": 12197
    },
    {
      "epoch": 0.5055327614074351,
      "grad_norm": 0.42749983072280884,
      "learning_rate": 2.4725434124911935e-06,
      "loss": 0.6467,
      "step": 12198
    },
    {
      "epoch": 0.5055742053131087,
      "grad_norm": 0.4008595645427704,
      "learning_rate": 2.472336192962825e-06,
      "loss": 0.6747,
      "step": 12199
    },
    {
      "epoch": 0.5056156492187823,
      "grad_norm": 0.40904802083969116,
      "learning_rate": 2.4721289734344567e-06,
      "loss": 0.6927,
      "step": 12200
    },
    {
      "epoch": 0.5056570931244561,
      "grad_norm": 0.43915098905563354,
      "learning_rate": 2.471921753906088e-06,
      "loss": 0.7239,
      "step": 12201
    },
    {
      "epoch": 0.5056985370301297,
      "grad_norm": 0.4118737280368805,
      "learning_rate": 2.47171453437772e-06,
      "loss": 0.6975,
      "step": 12202
    },
    {
      "epoch": 0.5057399809358034,
      "grad_norm": 0.40852171182632446,
      "learning_rate": 2.4715073148493517e-06,
      "loss": 0.6674,
      "step": 12203
    },
    {
      "epoch": 0.505781424841477,
      "grad_norm": 0.42807379364967346,
      "learning_rate": 2.471300095320983e-06,
      "loss": 0.7115,
      "step": 12204
    },
    {
      "epoch": 0.5058228687471508,
      "grad_norm": 0.4124264717102051,
      "learning_rate": 2.471092875792615e-06,
      "loss": 0.6548,
      "step": 12205
    },
    {
      "epoch": 0.5058643126528244,
      "grad_norm": 0.44087740778923035,
      "learning_rate": 2.4708856562642467e-06,
      "loss": 0.6951,
      "step": 12206
    },
    {
      "epoch": 0.5059057565584981,
      "grad_norm": 0.46886372566223145,
      "learning_rate": 2.470678436735878e-06,
      "loss": 0.7434,
      "step": 12207
    },
    {
      "epoch": 0.5059472004641717,
      "grad_norm": 0.44389280676841736,
      "learning_rate": 2.47047121720751e-06,
      "loss": 0.6847,
      "step": 12208
    },
    {
      "epoch": 0.5059886443698454,
      "grad_norm": 0.46807995438575745,
      "learning_rate": 2.4702639976791413e-06,
      "loss": 0.7605,
      "step": 12209
    },
    {
      "epoch": 0.5060300882755191,
      "grad_norm": 0.4185270369052887,
      "learning_rate": 2.470056778150773e-06,
      "loss": 0.6715,
      "step": 12210
    },
    {
      "epoch": 0.5060715321811927,
      "grad_norm": 0.4210631251335144,
      "learning_rate": 2.4698495586224045e-06,
      "loss": 0.6771,
      "step": 12211
    },
    {
      "epoch": 0.5061129760868665,
      "grad_norm": 0.3961467444896698,
      "learning_rate": 2.4696423390940363e-06,
      "loss": 0.7131,
      "step": 12212
    },
    {
      "epoch": 0.5061544199925401,
      "grad_norm": 0.4135449528694153,
      "learning_rate": 2.469435119565668e-06,
      "loss": 0.7065,
      "step": 12213
    },
    {
      "epoch": 0.5061958638982138,
      "grad_norm": 0.43862423300743103,
      "learning_rate": 2.4692279000373e-06,
      "loss": 0.705,
      "step": 12214
    },
    {
      "epoch": 0.5062373078038874,
      "grad_norm": 0.39867350459098816,
      "learning_rate": 2.4690206805089313e-06,
      "loss": 0.6504,
      "step": 12215
    },
    {
      "epoch": 0.5062787517095612,
      "grad_norm": 0.41621169447898865,
      "learning_rate": 2.468813460980563e-06,
      "loss": 0.7286,
      "step": 12216
    },
    {
      "epoch": 0.5063201956152348,
      "grad_norm": 0.4095323085784912,
      "learning_rate": 2.4686062414521945e-06,
      "loss": 0.6757,
      "step": 12217
    },
    {
      "epoch": 0.5063616395209084,
      "grad_norm": 0.42133840918540955,
      "learning_rate": 2.4683990219238263e-06,
      "loss": 0.6973,
      "step": 12218
    },
    {
      "epoch": 0.5064030834265821,
      "grad_norm": 0.404493123292923,
      "learning_rate": 2.4681918023954577e-06,
      "loss": 0.656,
      "step": 12219
    },
    {
      "epoch": 0.5064445273322558,
      "grad_norm": 0.4059443473815918,
      "learning_rate": 2.4679845828670895e-06,
      "loss": 0.6831,
      "step": 12220
    },
    {
      "epoch": 0.5064859712379295,
      "grad_norm": 0.4434875249862671,
      "learning_rate": 2.4677773633387213e-06,
      "loss": 0.7377,
      "step": 12221
    },
    {
      "epoch": 0.5065274151436031,
      "grad_norm": 0.3694967031478882,
      "learning_rate": 2.4675701438103527e-06,
      "loss": 0.675,
      "step": 12222
    },
    {
      "epoch": 0.5065688590492768,
      "grad_norm": 0.4084167182445526,
      "learning_rate": 2.4673629242819845e-06,
      "loss": 0.7253,
      "step": 12223
    },
    {
      "epoch": 0.5066103029549505,
      "grad_norm": 0.4680980145931244,
      "learning_rate": 2.4671557047536163e-06,
      "loss": 0.674,
      "step": 12224
    },
    {
      "epoch": 0.5066517468606242,
      "grad_norm": 0.4182599186897278,
      "learning_rate": 2.4669484852252477e-06,
      "loss": 0.6877,
      "step": 12225
    },
    {
      "epoch": 0.5066931907662978,
      "grad_norm": 0.4306640923023224,
      "learning_rate": 2.4667412656968795e-06,
      "loss": 0.6904,
      "step": 12226
    },
    {
      "epoch": 0.5067346346719714,
      "grad_norm": 0.3828088045120239,
      "learning_rate": 2.466534046168511e-06,
      "loss": 0.7129,
      "step": 12227
    },
    {
      "epoch": 0.5067760785776452,
      "grad_norm": 0.4040340781211853,
      "learning_rate": 2.4663268266401427e-06,
      "loss": 0.6958,
      "step": 12228
    },
    {
      "epoch": 0.5068175224833188,
      "grad_norm": 0.4159044623374939,
      "learning_rate": 2.4661196071117745e-06,
      "loss": 0.6895,
      "step": 12229
    },
    {
      "epoch": 0.5068589663889925,
      "grad_norm": 0.42758458852767944,
      "learning_rate": 2.465912387583406e-06,
      "loss": 0.6572,
      "step": 12230
    },
    {
      "epoch": 0.5069004102946661,
      "grad_norm": 0.3837548792362213,
      "learning_rate": 2.4657051680550377e-06,
      "loss": 0.6752,
      "step": 12231
    },
    {
      "epoch": 0.5069418542003399,
      "grad_norm": 0.3906620144844055,
      "learning_rate": 2.4654979485266695e-06,
      "loss": 0.7191,
      "step": 12232
    },
    {
      "epoch": 0.5069832981060135,
      "grad_norm": 0.38963863253593445,
      "learning_rate": 2.465290728998301e-06,
      "loss": 0.657,
      "step": 12233
    },
    {
      "epoch": 0.5070247420116872,
      "grad_norm": 0.39764219522476196,
      "learning_rate": 2.4650835094699327e-06,
      "loss": 0.7117,
      "step": 12234
    },
    {
      "epoch": 0.5070661859173609,
      "grad_norm": 0.3864278495311737,
      "learning_rate": 2.464876289941564e-06,
      "loss": 0.6969,
      "step": 12235
    },
    {
      "epoch": 0.5071076298230345,
      "grad_norm": 0.4151107668876648,
      "learning_rate": 2.464669070413196e-06,
      "loss": 0.7247,
      "step": 12236
    },
    {
      "epoch": 0.5071490737287082,
      "grad_norm": 0.3893643915653229,
      "learning_rate": 2.4644618508848277e-06,
      "loss": 0.6558,
      "step": 12237
    },
    {
      "epoch": 0.5071905176343818,
      "grad_norm": 0.4144800305366516,
      "learning_rate": 2.464254631356459e-06,
      "loss": 0.6611,
      "step": 12238
    },
    {
      "epoch": 0.5072319615400556,
      "grad_norm": 0.4151061177253723,
      "learning_rate": 2.464047411828091e-06,
      "loss": 0.688,
      "step": 12239
    },
    {
      "epoch": 0.5072734054457292,
      "grad_norm": 0.41983118653297424,
      "learning_rate": 2.4638401922997227e-06,
      "loss": 0.6702,
      "step": 12240
    },
    {
      "epoch": 0.5073148493514029,
      "grad_norm": 0.4059462249279022,
      "learning_rate": 2.463632972771354e-06,
      "loss": 0.6826,
      "step": 12241
    },
    {
      "epoch": 0.5073562932570765,
      "grad_norm": 0.43690910935401917,
      "learning_rate": 2.463425753242986e-06,
      "loss": 0.7006,
      "step": 12242
    },
    {
      "epoch": 0.5073977371627503,
      "grad_norm": 0.40925687551498413,
      "learning_rate": 2.4632185337146173e-06,
      "loss": 0.7004,
      "step": 12243
    },
    {
      "epoch": 0.5074391810684239,
      "grad_norm": 0.40462225675582886,
      "learning_rate": 2.463011314186249e-06,
      "loss": 0.676,
      "step": 12244
    },
    {
      "epoch": 0.5074806249740975,
      "grad_norm": 0.438272625207901,
      "learning_rate": 2.462804094657881e-06,
      "loss": 0.7189,
      "step": 12245
    },
    {
      "epoch": 0.5075220688797712,
      "grad_norm": 0.40205591917037964,
      "learning_rate": 2.4625968751295123e-06,
      "loss": 0.7114,
      "step": 12246
    },
    {
      "epoch": 0.5075635127854449,
      "grad_norm": 0.42673230171203613,
      "learning_rate": 2.462389655601144e-06,
      "loss": 0.6738,
      "step": 12247
    },
    {
      "epoch": 0.5076049566911186,
      "grad_norm": 0.4513818621635437,
      "learning_rate": 2.4621824360727755e-06,
      "loss": 0.67,
      "step": 12248
    },
    {
      "epoch": 0.5076464005967922,
      "grad_norm": 0.40852463245391846,
      "learning_rate": 2.4619752165444073e-06,
      "loss": 0.6833,
      "step": 12249
    },
    {
      "epoch": 0.507687844502466,
      "grad_norm": 0.41575610637664795,
      "learning_rate": 2.461767997016039e-06,
      "loss": 0.6873,
      "step": 12250
    },
    {
      "epoch": 0.5077292884081396,
      "grad_norm": 0.43960052728652954,
      "learning_rate": 2.4615607774876705e-06,
      "loss": 0.7,
      "step": 12251
    },
    {
      "epoch": 0.5077707323138132,
      "grad_norm": 0.4234411418437958,
      "learning_rate": 2.4613535579593023e-06,
      "loss": 0.6927,
      "step": 12252
    },
    {
      "epoch": 0.5078121762194869,
      "grad_norm": 0.40765973925590515,
      "learning_rate": 2.4611463384309337e-06,
      "loss": 0.6688,
      "step": 12253
    },
    {
      "epoch": 0.5078536201251606,
      "grad_norm": 0.4090679883956909,
      "learning_rate": 2.4609391189025655e-06,
      "loss": 0.7419,
      "step": 12254
    },
    {
      "epoch": 0.5078950640308343,
      "grad_norm": 0.4289027452468872,
      "learning_rate": 2.4607318993741973e-06,
      "loss": 0.7227,
      "step": 12255
    },
    {
      "epoch": 0.5079365079365079,
      "grad_norm": 0.422525018453598,
      "learning_rate": 2.4605246798458287e-06,
      "loss": 0.6652,
      "step": 12256
    },
    {
      "epoch": 0.5079779518421816,
      "grad_norm": 0.4026212990283966,
      "learning_rate": 2.4603174603174605e-06,
      "loss": 0.7076,
      "step": 12257
    },
    {
      "epoch": 0.5080193957478553,
      "grad_norm": 0.43463334441185,
      "learning_rate": 2.4601102407890923e-06,
      "loss": 0.7288,
      "step": 12258
    },
    {
      "epoch": 0.508060839653529,
      "grad_norm": 0.4179787039756775,
      "learning_rate": 2.4599030212607237e-06,
      "loss": 0.6624,
      "step": 12259
    },
    {
      "epoch": 0.5081022835592026,
      "grad_norm": 0.413021445274353,
      "learning_rate": 2.4596958017323555e-06,
      "loss": 0.6636,
      "step": 12260
    },
    {
      "epoch": 0.5081437274648762,
      "grad_norm": 0.3945936858654022,
      "learning_rate": 2.459488582203987e-06,
      "loss": 0.6554,
      "step": 12261
    },
    {
      "epoch": 0.50818517137055,
      "grad_norm": 0.4498036205768585,
      "learning_rate": 2.4592813626756187e-06,
      "loss": 0.7036,
      "step": 12262
    },
    {
      "epoch": 0.5082266152762236,
      "grad_norm": 0.4390327036380768,
      "learning_rate": 2.4590741431472505e-06,
      "loss": 0.7441,
      "step": 12263
    },
    {
      "epoch": 0.5082680591818973,
      "grad_norm": 0.4191127121448517,
      "learning_rate": 2.458866923618882e-06,
      "loss": 0.7249,
      "step": 12264
    },
    {
      "epoch": 0.508309503087571,
      "grad_norm": 0.3868004381656647,
      "learning_rate": 2.4586597040905137e-06,
      "loss": 0.6829,
      "step": 12265
    },
    {
      "epoch": 0.5083509469932447,
      "grad_norm": 0.4275377690792084,
      "learning_rate": 2.4584524845621455e-06,
      "loss": 0.6907,
      "step": 12266
    },
    {
      "epoch": 0.5083923908989183,
      "grad_norm": 0.4002785384654999,
      "learning_rate": 2.458245265033777e-06,
      "loss": 0.6843,
      "step": 12267
    },
    {
      "epoch": 0.508433834804592,
      "grad_norm": 0.3869546353816986,
      "learning_rate": 2.4580380455054087e-06,
      "loss": 0.6877,
      "step": 12268
    },
    {
      "epoch": 0.5084752787102657,
      "grad_norm": 0.39479702711105347,
      "learning_rate": 2.45783082597704e-06,
      "loss": 0.7141,
      "step": 12269
    },
    {
      "epoch": 0.5085167226159393,
      "grad_norm": 0.4116244912147522,
      "learning_rate": 2.457623606448672e-06,
      "loss": 0.7008,
      "step": 12270
    },
    {
      "epoch": 0.508558166521613,
      "grad_norm": 0.37338438630104065,
      "learning_rate": 2.4574163869203037e-06,
      "loss": 0.6489,
      "step": 12271
    },
    {
      "epoch": 0.5085996104272866,
      "grad_norm": 0.4580710828304291,
      "learning_rate": 2.457209167391935e-06,
      "loss": 0.6877,
      "step": 12272
    },
    {
      "epoch": 0.5086410543329604,
      "grad_norm": 0.41053277254104614,
      "learning_rate": 2.457001947863567e-06,
      "loss": 0.674,
      "step": 12273
    },
    {
      "epoch": 0.508682498238634,
      "grad_norm": 0.4166455566883087,
      "learning_rate": 2.4567947283351983e-06,
      "loss": 0.6978,
      "step": 12274
    },
    {
      "epoch": 0.5087239421443077,
      "grad_norm": 0.3813936710357666,
      "learning_rate": 2.45658750880683e-06,
      "loss": 0.6067,
      "step": 12275
    },
    {
      "epoch": 0.5087653860499813,
      "grad_norm": 0.4155506193637848,
      "learning_rate": 2.456380289278462e-06,
      "loss": 0.6857,
      "step": 12276
    },
    {
      "epoch": 0.5088068299556551,
      "grad_norm": 0.39879536628723145,
      "learning_rate": 2.4561730697500933e-06,
      "loss": 0.6553,
      "step": 12277
    },
    {
      "epoch": 0.5088482738613287,
      "grad_norm": 0.3778477907180786,
      "learning_rate": 2.455965850221725e-06,
      "loss": 0.6458,
      "step": 12278
    },
    {
      "epoch": 0.5088897177670023,
      "grad_norm": 0.4185250401496887,
      "learning_rate": 2.455758630693357e-06,
      "loss": 0.7308,
      "step": 12279
    },
    {
      "epoch": 0.508931161672676,
      "grad_norm": 0.43229997158050537,
      "learning_rate": 2.4555514111649883e-06,
      "loss": 0.6755,
      "step": 12280
    },
    {
      "epoch": 0.5089726055783497,
      "grad_norm": 0.4224403202533722,
      "learning_rate": 2.45534419163662e-06,
      "loss": 0.7854,
      "step": 12281
    },
    {
      "epoch": 0.5090140494840234,
      "grad_norm": 0.3853486180305481,
      "learning_rate": 2.4551369721082515e-06,
      "loss": 0.6765,
      "step": 12282
    },
    {
      "epoch": 0.509055493389697,
      "grad_norm": 0.38584086298942566,
      "learning_rate": 2.4549297525798833e-06,
      "loss": 0.6238,
      "step": 12283
    },
    {
      "epoch": 0.5090969372953708,
      "grad_norm": 0.3991621732711792,
      "learning_rate": 2.454722533051515e-06,
      "loss": 0.6505,
      "step": 12284
    },
    {
      "epoch": 0.5091383812010444,
      "grad_norm": 0.41635560989379883,
      "learning_rate": 2.4545153135231465e-06,
      "loss": 0.6821,
      "step": 12285
    },
    {
      "epoch": 0.5091798251067181,
      "grad_norm": 0.41656365990638733,
      "learning_rate": 2.4543080939947783e-06,
      "loss": 0.6396,
      "step": 12286
    },
    {
      "epoch": 0.5092212690123917,
      "grad_norm": 0.408433735370636,
      "learning_rate": 2.4541008744664097e-06,
      "loss": 0.6729,
      "step": 12287
    },
    {
      "epoch": 0.5092627129180654,
      "grad_norm": 0.45683589577674866,
      "learning_rate": 2.4538936549380415e-06,
      "loss": 0.7201,
      "step": 12288
    },
    {
      "epoch": 0.5093041568237391,
      "grad_norm": 0.39597707986831665,
      "learning_rate": 2.4536864354096733e-06,
      "loss": 0.6816,
      "step": 12289
    },
    {
      "epoch": 0.5093456007294127,
      "grad_norm": 0.4139714241027832,
      "learning_rate": 2.4534792158813047e-06,
      "loss": 0.7043,
      "step": 12290
    },
    {
      "epoch": 0.5093870446350864,
      "grad_norm": 0.4132087528705597,
      "learning_rate": 2.4532719963529365e-06,
      "loss": 0.696,
      "step": 12291
    },
    {
      "epoch": 0.5094284885407601,
      "grad_norm": 0.40727344155311584,
      "learning_rate": 2.4530647768245683e-06,
      "loss": 0.6954,
      "step": 12292
    },
    {
      "epoch": 0.5094699324464338,
      "grad_norm": 0.36722633242607117,
      "learning_rate": 2.4528575572961997e-06,
      "loss": 0.6692,
      "step": 12293
    },
    {
      "epoch": 0.5095113763521074,
      "grad_norm": 0.37520572543144226,
      "learning_rate": 2.4526503377678315e-06,
      "loss": 0.6541,
      "step": 12294
    },
    {
      "epoch": 0.5095528202577811,
      "grad_norm": 0.4189264476299286,
      "learning_rate": 2.452443118239463e-06,
      "loss": 0.6835,
      "step": 12295
    },
    {
      "epoch": 0.5095942641634548,
      "grad_norm": 0.40546074509620667,
      "learning_rate": 2.4522358987110947e-06,
      "loss": 0.6658,
      "step": 12296
    },
    {
      "epoch": 0.5096357080691284,
      "grad_norm": 0.40954652428627014,
      "learning_rate": 2.4520286791827265e-06,
      "loss": 0.6721,
      "step": 12297
    },
    {
      "epoch": 0.5096771519748021,
      "grad_norm": 0.41790667176246643,
      "learning_rate": 2.451821459654358e-06,
      "loss": 0.6489,
      "step": 12298
    },
    {
      "epoch": 0.5097185958804757,
      "grad_norm": 0.4155614674091339,
      "learning_rate": 2.4516142401259897e-06,
      "loss": 0.6882,
      "step": 12299
    },
    {
      "epoch": 0.5097600397861495,
      "grad_norm": 0.41406548023223877,
      "learning_rate": 2.451407020597621e-06,
      "loss": 0.6646,
      "step": 12300
    },
    {
      "epoch": 0.5098014836918231,
      "grad_norm": 0.399674654006958,
      "learning_rate": 2.451199801069253e-06,
      "loss": 0.6666,
      "step": 12301
    },
    {
      "epoch": 0.5098429275974968,
      "grad_norm": 0.42055946588516235,
      "learning_rate": 2.4509925815408847e-06,
      "loss": 0.6614,
      "step": 12302
    },
    {
      "epoch": 0.5098843715031705,
      "grad_norm": 0.40756893157958984,
      "learning_rate": 2.450785362012516e-06,
      "loss": 0.6582,
      "step": 12303
    },
    {
      "epoch": 0.5099258154088441,
      "grad_norm": 0.42666637897491455,
      "learning_rate": 2.450578142484148e-06,
      "loss": 0.6499,
      "step": 12304
    },
    {
      "epoch": 0.5099672593145178,
      "grad_norm": 0.4141885042190552,
      "learning_rate": 2.4503709229557797e-06,
      "loss": 0.6683,
      "step": 12305
    },
    {
      "epoch": 0.5100087032201914,
      "grad_norm": 0.4307863414287567,
      "learning_rate": 2.450163703427411e-06,
      "loss": 0.7729,
      "step": 12306
    },
    {
      "epoch": 0.5100501471258652,
      "grad_norm": 0.3775271773338318,
      "learning_rate": 2.449956483899043e-06,
      "loss": 0.6486,
      "step": 12307
    },
    {
      "epoch": 0.5100915910315388,
      "grad_norm": 0.3676570653915405,
      "learning_rate": 2.4497492643706743e-06,
      "loss": 0.6211,
      "step": 12308
    },
    {
      "epoch": 0.5101330349372125,
      "grad_norm": 0.4327489137649536,
      "learning_rate": 2.449542044842306e-06,
      "loss": 0.7007,
      "step": 12309
    },
    {
      "epoch": 0.5101744788428861,
      "grad_norm": 0.4338235557079315,
      "learning_rate": 2.449334825313938e-06,
      "loss": 0.7275,
      "step": 12310
    },
    {
      "epoch": 0.5102159227485599,
      "grad_norm": 0.4016983211040497,
      "learning_rate": 2.4491276057855693e-06,
      "loss": 0.7014,
      "step": 12311
    },
    {
      "epoch": 0.5102573666542335,
      "grad_norm": 0.3973780870437622,
      "learning_rate": 2.448920386257201e-06,
      "loss": 0.687,
      "step": 12312
    },
    {
      "epoch": 0.5102988105599071,
      "grad_norm": 0.39833512902259827,
      "learning_rate": 2.448713166728833e-06,
      "loss": 0.7417,
      "step": 12313
    },
    {
      "epoch": 0.5103402544655808,
      "grad_norm": 0.4238496720790863,
      "learning_rate": 2.4485059472004643e-06,
      "loss": 0.6938,
      "step": 12314
    },
    {
      "epoch": 0.5103816983712545,
      "grad_norm": 0.38524329662323,
      "learning_rate": 2.448298727672096e-06,
      "loss": 0.6777,
      "step": 12315
    },
    {
      "epoch": 0.5104231422769282,
      "grad_norm": 0.45752766728401184,
      "learning_rate": 2.4480915081437275e-06,
      "loss": 0.717,
      "step": 12316
    },
    {
      "epoch": 0.5104645861826018,
      "grad_norm": 0.4319753348827362,
      "learning_rate": 2.4478842886153593e-06,
      "loss": 0.6914,
      "step": 12317
    },
    {
      "epoch": 0.5105060300882756,
      "grad_norm": 0.38236337900161743,
      "learning_rate": 2.4476770690869907e-06,
      "loss": 0.6315,
      "step": 12318
    },
    {
      "epoch": 0.5105474739939492,
      "grad_norm": 0.39100182056427,
      "learning_rate": 2.4474698495586225e-06,
      "loss": 0.6584,
      "step": 12319
    },
    {
      "epoch": 0.5105889178996229,
      "grad_norm": 0.3938053846359253,
      "learning_rate": 2.4472626300302543e-06,
      "loss": 0.7285,
      "step": 12320
    },
    {
      "epoch": 0.5106303618052965,
      "grad_norm": 0.40324538946151733,
      "learning_rate": 2.4470554105018857e-06,
      "loss": 0.6021,
      "step": 12321
    },
    {
      "epoch": 0.5106718057109702,
      "grad_norm": 0.3964911699295044,
      "learning_rate": 2.4468481909735175e-06,
      "loss": 0.6666,
      "step": 12322
    },
    {
      "epoch": 0.5107132496166439,
      "grad_norm": 0.4506520926952362,
      "learning_rate": 2.4466409714451493e-06,
      "loss": 0.7021,
      "step": 12323
    },
    {
      "epoch": 0.5107546935223175,
      "grad_norm": 0.4161345660686493,
      "learning_rate": 2.4464337519167807e-06,
      "loss": 0.6439,
      "step": 12324
    },
    {
      "epoch": 0.5107961374279912,
      "grad_norm": 0.4516972601413727,
      "learning_rate": 2.4462265323884125e-06,
      "loss": 0.7361,
      "step": 12325
    },
    {
      "epoch": 0.5108375813336649,
      "grad_norm": 0.39624130725860596,
      "learning_rate": 2.446019312860044e-06,
      "loss": 0.6708,
      "step": 12326
    },
    {
      "epoch": 0.5108790252393386,
      "grad_norm": 0.3733086585998535,
      "learning_rate": 2.4458120933316757e-06,
      "loss": 0.6357,
      "step": 12327
    },
    {
      "epoch": 0.5109204691450122,
      "grad_norm": 0.4160917401313782,
      "learning_rate": 2.4456048738033075e-06,
      "loss": 0.6958,
      "step": 12328
    },
    {
      "epoch": 0.510961913050686,
      "grad_norm": 0.4151105284690857,
      "learning_rate": 2.445397654274939e-06,
      "loss": 0.6924,
      "step": 12329
    },
    {
      "epoch": 0.5110033569563596,
      "grad_norm": 0.4358883798122406,
      "learning_rate": 2.4451904347465707e-06,
      "loss": 0.6896,
      "step": 12330
    },
    {
      "epoch": 0.5110448008620332,
      "grad_norm": 0.4502808749675751,
      "learning_rate": 2.4449832152182026e-06,
      "loss": 0.7,
      "step": 12331
    },
    {
      "epoch": 0.5110862447677069,
      "grad_norm": 0.4560929536819458,
      "learning_rate": 2.444775995689834e-06,
      "loss": 0.6884,
      "step": 12332
    },
    {
      "epoch": 0.5111276886733805,
      "grad_norm": 0.42196333408355713,
      "learning_rate": 2.4445687761614657e-06,
      "loss": 0.6858,
      "step": 12333
    },
    {
      "epoch": 0.5111691325790543,
      "grad_norm": 0.4276515543460846,
      "learning_rate": 2.444361556633097e-06,
      "loss": 0.6895,
      "step": 12334
    },
    {
      "epoch": 0.5112105764847279,
      "grad_norm": 0.40929850935935974,
      "learning_rate": 2.444154337104729e-06,
      "loss": 0.698,
      "step": 12335
    },
    {
      "epoch": 0.5112520203904016,
      "grad_norm": 0.41483771800994873,
      "learning_rate": 2.4439471175763608e-06,
      "loss": 0.7358,
      "step": 12336
    },
    {
      "epoch": 0.5112934642960753,
      "grad_norm": 0.42353618144989014,
      "learning_rate": 2.443739898047992e-06,
      "loss": 0.6816,
      "step": 12337
    },
    {
      "epoch": 0.511334908201749,
      "grad_norm": 0.42023879289627075,
      "learning_rate": 2.443532678519624e-06,
      "loss": 0.7073,
      "step": 12338
    },
    {
      "epoch": 0.5113763521074226,
      "grad_norm": 0.4273877441883087,
      "learning_rate": 2.4433254589912558e-06,
      "loss": 0.6796,
      "step": 12339
    },
    {
      "epoch": 0.5114177960130962,
      "grad_norm": 0.4168732464313507,
      "learning_rate": 2.443118239462887e-06,
      "loss": 0.6931,
      "step": 12340
    },
    {
      "epoch": 0.51145923991877,
      "grad_norm": 0.3826819062232971,
      "learning_rate": 2.442911019934519e-06,
      "loss": 0.689,
      "step": 12341
    },
    {
      "epoch": 0.5115006838244436,
      "grad_norm": 0.3854614198207855,
      "learning_rate": 2.4427038004061503e-06,
      "loss": 0.6512,
      "step": 12342
    },
    {
      "epoch": 0.5115421277301173,
      "grad_norm": 0.4086402356624603,
      "learning_rate": 2.442496580877782e-06,
      "loss": 0.6786,
      "step": 12343
    },
    {
      "epoch": 0.5115835716357909,
      "grad_norm": 0.39128780364990234,
      "learning_rate": 2.4422893613494135e-06,
      "loss": 0.6592,
      "step": 12344
    },
    {
      "epoch": 0.5116250155414647,
      "grad_norm": 0.4435243308544159,
      "learning_rate": 2.4420821418210453e-06,
      "loss": 0.7549,
      "step": 12345
    },
    {
      "epoch": 0.5116664594471383,
      "grad_norm": 0.43374961614608765,
      "learning_rate": 2.441874922292677e-06,
      "loss": 0.7158,
      "step": 12346
    },
    {
      "epoch": 0.511707903352812,
      "grad_norm": 0.4074364900588989,
      "learning_rate": 2.441667702764309e-06,
      "loss": 0.7075,
      "step": 12347
    },
    {
      "epoch": 0.5117493472584856,
      "grad_norm": 0.4237951934337616,
      "learning_rate": 2.4414604832359403e-06,
      "loss": 0.7222,
      "step": 12348
    },
    {
      "epoch": 0.5117907911641593,
      "grad_norm": 0.427463173866272,
      "learning_rate": 2.441253263707572e-06,
      "loss": 0.7053,
      "step": 12349
    },
    {
      "epoch": 0.511832235069833,
      "grad_norm": 0.41484227776527405,
      "learning_rate": 2.4410460441792035e-06,
      "loss": 0.6968,
      "step": 12350
    },
    {
      "epoch": 0.5118736789755066,
      "grad_norm": 0.4204866290092468,
      "learning_rate": 2.4408388246508353e-06,
      "loss": 0.7314,
      "step": 12351
    },
    {
      "epoch": 0.5119151228811804,
      "grad_norm": 0.38757947087287903,
      "learning_rate": 2.4406316051224667e-06,
      "loss": 0.7034,
      "step": 12352
    },
    {
      "epoch": 0.511956566786854,
      "grad_norm": 0.4288354218006134,
      "learning_rate": 2.4404243855940985e-06,
      "loss": 0.7188,
      "step": 12353
    },
    {
      "epoch": 0.5119980106925277,
      "grad_norm": 0.3931960463523865,
      "learning_rate": 2.4402171660657304e-06,
      "loss": 0.6987,
      "step": 12354
    },
    {
      "epoch": 0.5120394545982013,
      "grad_norm": 0.40989845991134644,
      "learning_rate": 2.440009946537362e-06,
      "loss": 0.6633,
      "step": 12355
    },
    {
      "epoch": 0.5120808985038751,
      "grad_norm": 0.38906097412109375,
      "learning_rate": 2.4398027270089935e-06,
      "loss": 0.6985,
      "step": 12356
    },
    {
      "epoch": 0.5121223424095487,
      "grad_norm": 0.42476212978363037,
      "learning_rate": 2.4395955074806254e-06,
      "loss": 0.6906,
      "step": 12357
    },
    {
      "epoch": 0.5121637863152223,
      "grad_norm": 0.41792193055152893,
      "learning_rate": 2.4393882879522567e-06,
      "loss": 0.752,
      "step": 12358
    },
    {
      "epoch": 0.512205230220896,
      "grad_norm": 0.4364200532436371,
      "learning_rate": 2.4391810684238886e-06,
      "loss": 0.7156,
      "step": 12359
    },
    {
      "epoch": 0.5122466741265697,
      "grad_norm": 0.38577136397361755,
      "learning_rate": 2.43897384889552e-06,
      "loss": 0.6508,
      "step": 12360
    },
    {
      "epoch": 0.5122881180322434,
      "grad_norm": 0.4027463495731354,
      "learning_rate": 2.4387666293671517e-06,
      "loss": 0.7175,
      "step": 12361
    },
    {
      "epoch": 0.512329561937917,
      "grad_norm": 0.39810699224472046,
      "learning_rate": 2.4385594098387836e-06,
      "loss": 0.6482,
      "step": 12362
    },
    {
      "epoch": 0.5123710058435907,
      "grad_norm": 0.43842020630836487,
      "learning_rate": 2.438352190310415e-06,
      "loss": 0.6777,
      "step": 12363
    },
    {
      "epoch": 0.5124124497492644,
      "grad_norm": 0.4122227728366852,
      "learning_rate": 2.4381449707820467e-06,
      "loss": 0.6602,
      "step": 12364
    },
    {
      "epoch": 0.512453893654938,
      "grad_norm": 0.387963205575943,
      "learning_rate": 2.4379377512536786e-06,
      "loss": 0.6875,
      "step": 12365
    },
    {
      "epoch": 0.5124953375606117,
      "grad_norm": 0.43270790576934814,
      "learning_rate": 2.43773053172531e-06,
      "loss": 0.6964,
      "step": 12366
    },
    {
      "epoch": 0.5125367814662853,
      "grad_norm": 0.40784206986427307,
      "learning_rate": 2.4375233121969418e-06,
      "loss": 0.6649,
      "step": 12367
    },
    {
      "epoch": 0.5125782253719591,
      "grad_norm": 0.4311182200908661,
      "learning_rate": 2.437316092668573e-06,
      "loss": 0.7292,
      "step": 12368
    },
    {
      "epoch": 0.5126196692776327,
      "grad_norm": 0.4617283344268799,
      "learning_rate": 2.437108873140205e-06,
      "loss": 0.8054,
      "step": 12369
    },
    {
      "epoch": 0.5126611131833064,
      "grad_norm": 0.4167940020561218,
      "learning_rate": 2.4369016536118363e-06,
      "loss": 0.6571,
      "step": 12370
    },
    {
      "epoch": 0.51270255708898,
      "grad_norm": 0.460943341255188,
      "learning_rate": 2.436694434083468e-06,
      "loss": 0.6902,
      "step": 12371
    },
    {
      "epoch": 0.5127440009946538,
      "grad_norm": 0.4228794574737549,
      "learning_rate": 2.4364872145551e-06,
      "loss": 0.7069,
      "step": 12372
    },
    {
      "epoch": 0.5127854449003274,
      "grad_norm": 0.39585429430007935,
      "learning_rate": 2.4362799950267318e-06,
      "loss": 0.6624,
      "step": 12373
    },
    {
      "epoch": 0.512826888806001,
      "grad_norm": 0.3891880214214325,
      "learning_rate": 2.436072775498363e-06,
      "loss": 0.6737,
      "step": 12374
    },
    {
      "epoch": 0.5128683327116748,
      "grad_norm": 0.41229698061943054,
      "learning_rate": 2.435865555969995e-06,
      "loss": 0.6978,
      "step": 12375
    },
    {
      "epoch": 0.5129097766173484,
      "grad_norm": 0.5377939939498901,
      "learning_rate": 2.4356583364416263e-06,
      "loss": 0.7338,
      "step": 12376
    },
    {
      "epoch": 0.5129512205230221,
      "grad_norm": 0.3782595694065094,
      "learning_rate": 2.435451116913258e-06,
      "loss": 0.6515,
      "step": 12377
    },
    {
      "epoch": 0.5129926644286957,
      "grad_norm": 0.42360132932662964,
      "learning_rate": 2.4352438973848895e-06,
      "loss": 0.7092,
      "step": 12378
    },
    {
      "epoch": 0.5130341083343695,
      "grad_norm": 0.44392064213752747,
      "learning_rate": 2.4350366778565213e-06,
      "loss": 0.7302,
      "step": 12379
    },
    {
      "epoch": 0.5130755522400431,
      "grad_norm": 0.4285444915294647,
      "learning_rate": 2.434829458328153e-06,
      "loss": 0.6786,
      "step": 12380
    },
    {
      "epoch": 0.5131169961457168,
      "grad_norm": 0.42923787236213684,
      "learning_rate": 2.434622238799785e-06,
      "loss": 0.683,
      "step": 12381
    },
    {
      "epoch": 0.5131584400513904,
      "grad_norm": 0.3952816426753998,
      "learning_rate": 2.4344150192714163e-06,
      "loss": 0.7085,
      "step": 12382
    },
    {
      "epoch": 0.5131998839570641,
      "grad_norm": 0.3907700181007385,
      "learning_rate": 2.434207799743048e-06,
      "loss": 0.677,
      "step": 12383
    },
    {
      "epoch": 0.5132413278627378,
      "grad_norm": 0.4267285466194153,
      "learning_rate": 2.4340005802146795e-06,
      "loss": 0.7092,
      "step": 12384
    },
    {
      "epoch": 0.5132827717684114,
      "grad_norm": 0.37499916553497314,
      "learning_rate": 2.4337933606863114e-06,
      "loss": 0.6089,
      "step": 12385
    },
    {
      "epoch": 0.5133242156740851,
      "grad_norm": 0.4341261386871338,
      "learning_rate": 2.4335861411579427e-06,
      "loss": 0.6753,
      "step": 12386
    },
    {
      "epoch": 0.5133656595797588,
      "grad_norm": 0.38155174255371094,
      "learning_rate": 2.4333789216295745e-06,
      "loss": 0.6492,
      "step": 12387
    },
    {
      "epoch": 0.5134071034854325,
      "grad_norm": 0.41241884231567383,
      "learning_rate": 2.4331717021012064e-06,
      "loss": 0.6774,
      "step": 12388
    },
    {
      "epoch": 0.5134485473911061,
      "grad_norm": 0.4269993305206299,
      "learning_rate": 2.432964482572838e-06,
      "loss": 0.6416,
      "step": 12389
    },
    {
      "epoch": 0.5134899912967799,
      "grad_norm": 0.4279613196849823,
      "learning_rate": 2.4327572630444696e-06,
      "loss": 0.6787,
      "step": 12390
    },
    {
      "epoch": 0.5135314352024535,
      "grad_norm": 0.42685380578041077,
      "learning_rate": 2.4325500435161014e-06,
      "loss": 0.6768,
      "step": 12391
    },
    {
      "epoch": 0.5135728791081271,
      "grad_norm": 0.3986414670944214,
      "learning_rate": 2.4323428239877327e-06,
      "loss": 0.7012,
      "step": 12392
    },
    {
      "epoch": 0.5136143230138008,
      "grad_norm": 0.4117792546749115,
      "learning_rate": 2.4321356044593646e-06,
      "loss": 0.7258,
      "step": 12393
    },
    {
      "epoch": 0.5136557669194745,
      "grad_norm": 0.42331960797309875,
      "learning_rate": 2.431928384930996e-06,
      "loss": 0.7141,
      "step": 12394
    },
    {
      "epoch": 0.5136972108251482,
      "grad_norm": 0.40305784344673157,
      "learning_rate": 2.4317211654026278e-06,
      "loss": 0.6221,
      "step": 12395
    },
    {
      "epoch": 0.5137386547308218,
      "grad_norm": 0.40559908747673035,
      "learning_rate": 2.431513945874259e-06,
      "loss": 0.6831,
      "step": 12396
    },
    {
      "epoch": 0.5137800986364955,
      "grad_norm": 0.423515260219574,
      "learning_rate": 2.431306726345891e-06,
      "loss": 0.7366,
      "step": 12397
    },
    {
      "epoch": 0.5138215425421692,
      "grad_norm": 0.43370333313941956,
      "learning_rate": 2.4310995068175228e-06,
      "loss": 0.7384,
      "step": 12398
    },
    {
      "epoch": 0.5138629864478429,
      "grad_norm": 0.3805336058139801,
      "learning_rate": 2.4308922872891546e-06,
      "loss": 0.7029,
      "step": 12399
    },
    {
      "epoch": 0.5139044303535165,
      "grad_norm": 0.4202497899532318,
      "learning_rate": 2.430685067760786e-06,
      "loss": 0.6996,
      "step": 12400
    },
    {
      "epoch": 0.5139458742591901,
      "grad_norm": 0.37977880239486694,
      "learning_rate": 2.4304778482324178e-06,
      "loss": 0.6873,
      "step": 12401
    },
    {
      "epoch": 0.5139873181648639,
      "grad_norm": 0.3970392644405365,
      "learning_rate": 2.430270628704049e-06,
      "loss": 0.6785,
      "step": 12402
    },
    {
      "epoch": 0.5140287620705375,
      "grad_norm": 0.3944105803966522,
      "learning_rate": 2.430063409175681e-06,
      "loss": 0.6763,
      "step": 12403
    },
    {
      "epoch": 0.5140702059762112,
      "grad_norm": 0.40997523069381714,
      "learning_rate": 2.4298561896473123e-06,
      "loss": 0.6132,
      "step": 12404
    },
    {
      "epoch": 0.5141116498818848,
      "grad_norm": 0.3899269998073578,
      "learning_rate": 2.429648970118944e-06,
      "loss": 0.7026,
      "step": 12405
    },
    {
      "epoch": 0.5141530937875586,
      "grad_norm": 0.3951246440410614,
      "learning_rate": 2.429441750590576e-06,
      "loss": 0.6248,
      "step": 12406
    },
    {
      "epoch": 0.5141945376932322,
      "grad_norm": 0.41713371872901917,
      "learning_rate": 2.4292345310622078e-06,
      "loss": 0.6768,
      "step": 12407
    },
    {
      "epoch": 0.5142359815989059,
      "grad_norm": 0.38698330521583557,
      "learning_rate": 2.429027311533839e-06,
      "loss": 0.6526,
      "step": 12408
    },
    {
      "epoch": 0.5142774255045796,
      "grad_norm": 0.4173656702041626,
      "learning_rate": 2.428820092005471e-06,
      "loss": 0.6713,
      "step": 12409
    },
    {
      "epoch": 0.5143188694102532,
      "grad_norm": 0.40929093956947327,
      "learning_rate": 2.4286128724771023e-06,
      "loss": 0.7102,
      "step": 12410
    },
    {
      "epoch": 0.5143603133159269,
      "grad_norm": 0.4007910192012787,
      "learning_rate": 2.428405652948734e-06,
      "loss": 0.6577,
      "step": 12411
    },
    {
      "epoch": 0.5144017572216005,
      "grad_norm": 0.4513412117958069,
      "learning_rate": 2.4281984334203655e-06,
      "loss": 0.74,
      "step": 12412
    },
    {
      "epoch": 0.5144432011272743,
      "grad_norm": 0.431240051984787,
      "learning_rate": 2.4279912138919974e-06,
      "loss": 0.6517,
      "step": 12413
    },
    {
      "epoch": 0.5144846450329479,
      "grad_norm": 0.390553742647171,
      "learning_rate": 2.427783994363629e-06,
      "loss": 0.7197,
      "step": 12414
    },
    {
      "epoch": 0.5145260889386216,
      "grad_norm": 0.44447460770606995,
      "learning_rate": 2.427576774835261e-06,
      "loss": 0.7241,
      "step": 12415
    },
    {
      "epoch": 0.5145675328442952,
      "grad_norm": 0.41056710481643677,
      "learning_rate": 2.4273695553068924e-06,
      "loss": 0.6636,
      "step": 12416
    },
    {
      "epoch": 0.514608976749969,
      "grad_norm": 0.3975561261177063,
      "learning_rate": 2.427162335778524e-06,
      "loss": 0.6892,
      "step": 12417
    },
    {
      "epoch": 0.5146504206556426,
      "grad_norm": 0.4347476065158844,
      "learning_rate": 2.4269551162501556e-06,
      "loss": 0.7458,
      "step": 12418
    },
    {
      "epoch": 0.5146918645613162,
      "grad_norm": 0.39927423000335693,
      "learning_rate": 2.4267478967217874e-06,
      "loss": 0.663,
      "step": 12419
    },
    {
      "epoch": 0.51473330846699,
      "grad_norm": 0.4162602424621582,
      "learning_rate": 2.4265406771934187e-06,
      "loss": 0.6921,
      "step": 12420
    },
    {
      "epoch": 0.5147747523726636,
      "grad_norm": 0.4358236491680145,
      "learning_rate": 2.4263334576650506e-06,
      "loss": 0.7126,
      "step": 12421
    },
    {
      "epoch": 0.5148161962783373,
      "grad_norm": 0.45137056708335876,
      "learning_rate": 2.426126238136682e-06,
      "loss": 0.679,
      "step": 12422
    },
    {
      "epoch": 0.5148576401840109,
      "grad_norm": 0.43010976910591125,
      "learning_rate": 2.425919018608314e-06,
      "loss": 0.7275,
      "step": 12423
    },
    {
      "epoch": 0.5148990840896847,
      "grad_norm": 0.4156637191772461,
      "learning_rate": 2.4257117990799456e-06,
      "loss": 0.6356,
      "step": 12424
    },
    {
      "epoch": 0.5149405279953583,
      "grad_norm": 0.3929874897003174,
      "learning_rate": 2.4255045795515774e-06,
      "loss": 0.6714,
      "step": 12425
    },
    {
      "epoch": 0.5149819719010319,
      "grad_norm": 0.4154079854488373,
      "learning_rate": 2.4252973600232088e-06,
      "loss": 0.713,
      "step": 12426
    },
    {
      "epoch": 0.5150234158067056,
      "grad_norm": 0.41203850507736206,
      "learning_rate": 2.4250901404948406e-06,
      "loss": 0.7222,
      "step": 12427
    },
    {
      "epoch": 0.5150648597123793,
      "grad_norm": 0.42554187774658203,
      "learning_rate": 2.424882920966472e-06,
      "loss": 0.6814,
      "step": 12428
    },
    {
      "epoch": 0.515106303618053,
      "grad_norm": 0.4256628155708313,
      "learning_rate": 2.4246757014381038e-06,
      "loss": 0.729,
      "step": 12429
    },
    {
      "epoch": 0.5151477475237266,
      "grad_norm": 0.3902890980243683,
      "learning_rate": 2.424468481909735e-06,
      "loss": 0.6405,
      "step": 12430
    },
    {
      "epoch": 0.5151891914294003,
      "grad_norm": 0.40703219175338745,
      "learning_rate": 2.424261262381367e-06,
      "loss": 0.7228,
      "step": 12431
    },
    {
      "epoch": 0.515230635335074,
      "grad_norm": 0.3920605778694153,
      "learning_rate": 2.4240540428529988e-06,
      "loss": 0.666,
      "step": 12432
    },
    {
      "epoch": 0.5152720792407477,
      "grad_norm": 0.3911963403224945,
      "learning_rate": 2.4238468233246306e-06,
      "loss": 0.6814,
      "step": 12433
    },
    {
      "epoch": 0.5153135231464213,
      "grad_norm": 0.4035681188106537,
      "learning_rate": 2.423639603796262e-06,
      "loss": 0.6431,
      "step": 12434
    },
    {
      "epoch": 0.5153549670520949,
      "grad_norm": 0.41798722743988037,
      "learning_rate": 2.4234323842678938e-06,
      "loss": 0.6931,
      "step": 12435
    },
    {
      "epoch": 0.5153964109577687,
      "grad_norm": 0.41263723373413086,
      "learning_rate": 2.423225164739525e-06,
      "loss": 0.632,
      "step": 12436
    },
    {
      "epoch": 0.5154378548634423,
      "grad_norm": 0.4507071077823639,
      "learning_rate": 2.423017945211157e-06,
      "loss": 0.6801,
      "step": 12437
    },
    {
      "epoch": 0.515479298769116,
      "grad_norm": 0.41934531927108765,
      "learning_rate": 2.4228107256827883e-06,
      "loss": 0.7023,
      "step": 12438
    },
    {
      "epoch": 0.5155207426747896,
      "grad_norm": 0.40594613552093506,
      "learning_rate": 2.42260350615442e-06,
      "loss": 0.7009,
      "step": 12439
    },
    {
      "epoch": 0.5155621865804634,
      "grad_norm": 0.4075371026992798,
      "learning_rate": 2.422396286626052e-06,
      "loss": 0.6614,
      "step": 12440
    },
    {
      "epoch": 0.515603630486137,
      "grad_norm": 0.40703001618385315,
      "learning_rate": 2.4221890670976838e-06,
      "loss": 0.6458,
      "step": 12441
    },
    {
      "epoch": 0.5156450743918107,
      "grad_norm": 0.41371577978134155,
      "learning_rate": 2.421981847569315e-06,
      "loss": 0.6851,
      "step": 12442
    },
    {
      "epoch": 0.5156865182974844,
      "grad_norm": 0.43757346272468567,
      "learning_rate": 2.421774628040947e-06,
      "loss": 0.6637,
      "step": 12443
    },
    {
      "epoch": 0.515727962203158,
      "grad_norm": 0.4561006724834442,
      "learning_rate": 2.4215674085125784e-06,
      "loss": 0.7083,
      "step": 12444
    },
    {
      "epoch": 0.5157694061088317,
      "grad_norm": 0.3714505434036255,
      "learning_rate": 2.42136018898421e-06,
      "loss": 0.6426,
      "step": 12445
    },
    {
      "epoch": 0.5158108500145053,
      "grad_norm": 0.4047311246395111,
      "learning_rate": 2.4211529694558415e-06,
      "loss": 0.6888,
      "step": 12446
    },
    {
      "epoch": 0.5158522939201791,
      "grad_norm": 0.4280291795730591,
      "learning_rate": 2.4209457499274734e-06,
      "loss": 0.7139,
      "step": 12447
    },
    {
      "epoch": 0.5158937378258527,
      "grad_norm": 0.4457102119922638,
      "learning_rate": 2.4207385303991047e-06,
      "loss": 0.6888,
      "step": 12448
    },
    {
      "epoch": 0.5159351817315264,
      "grad_norm": 0.4048601984977722,
      "learning_rate": 2.420531310870737e-06,
      "loss": 0.672,
      "step": 12449
    },
    {
      "epoch": 0.5159766256372,
      "grad_norm": 0.438904345035553,
      "learning_rate": 2.4203240913423684e-06,
      "loss": 0.6816,
      "step": 12450
    },
    {
      "epoch": 0.5160180695428738,
      "grad_norm": 0.427717387676239,
      "learning_rate": 2.420116871814e-06,
      "loss": 0.6766,
      "step": 12451
    },
    {
      "epoch": 0.5160595134485474,
      "grad_norm": 0.4113977551460266,
      "learning_rate": 2.4199096522856316e-06,
      "loss": 0.6586,
      "step": 12452
    },
    {
      "epoch": 0.516100957354221,
      "grad_norm": 0.4249646067619324,
      "learning_rate": 2.4197024327572634e-06,
      "loss": 0.656,
      "step": 12453
    },
    {
      "epoch": 0.5161424012598947,
      "grad_norm": 0.37517818808555603,
      "learning_rate": 2.4194952132288948e-06,
      "loss": 0.6938,
      "step": 12454
    },
    {
      "epoch": 0.5161838451655684,
      "grad_norm": 0.42449942231178284,
      "learning_rate": 2.4192879937005266e-06,
      "loss": 0.6747,
      "step": 12455
    },
    {
      "epoch": 0.5162252890712421,
      "grad_norm": 0.39077767729759216,
      "learning_rate": 2.419080774172158e-06,
      "loss": 0.6672,
      "step": 12456
    },
    {
      "epoch": 0.5162667329769157,
      "grad_norm": 0.3730625808238983,
      "learning_rate": 2.4188735546437898e-06,
      "loss": 0.657,
      "step": 12457
    },
    {
      "epoch": 0.5163081768825895,
      "grad_norm": 0.39072945713996887,
      "learning_rate": 2.4186663351154216e-06,
      "loss": 0.6265,
      "step": 12458
    },
    {
      "epoch": 0.5163496207882631,
      "grad_norm": 0.4303812086582184,
      "learning_rate": 2.4184591155870534e-06,
      "loss": 0.708,
      "step": 12459
    },
    {
      "epoch": 0.5163910646939368,
      "grad_norm": 0.42370739579200745,
      "learning_rate": 2.4182518960586848e-06,
      "loss": 0.712,
      "step": 12460
    },
    {
      "epoch": 0.5164325085996104,
      "grad_norm": 0.38857710361480713,
      "learning_rate": 2.4180446765303166e-06,
      "loss": 0.7075,
      "step": 12461
    },
    {
      "epoch": 0.516473952505284,
      "grad_norm": 0.4039013981819153,
      "learning_rate": 2.417837457001948e-06,
      "loss": 0.7056,
      "step": 12462
    },
    {
      "epoch": 0.5165153964109578,
      "grad_norm": 0.43030688166618347,
      "learning_rate": 2.4176302374735798e-06,
      "loss": 0.699,
      "step": 12463
    },
    {
      "epoch": 0.5165568403166314,
      "grad_norm": 0.41584914922714233,
      "learning_rate": 2.417423017945211e-06,
      "loss": 0.6921,
      "step": 12464
    },
    {
      "epoch": 0.5165982842223051,
      "grad_norm": 0.4191140830516815,
      "learning_rate": 2.417215798416843e-06,
      "loss": 0.6541,
      "step": 12465
    },
    {
      "epoch": 0.5166397281279788,
      "grad_norm": 0.4151654541492462,
      "learning_rate": 2.4170085788884743e-06,
      "loss": 0.6835,
      "step": 12466
    },
    {
      "epoch": 0.5166811720336525,
      "grad_norm": 0.4528132975101471,
      "learning_rate": 2.4168013593601066e-06,
      "loss": 0.7583,
      "step": 12467
    },
    {
      "epoch": 0.5167226159393261,
      "grad_norm": 0.39025992155075073,
      "learning_rate": 2.416594139831738e-06,
      "loss": 0.6472,
      "step": 12468
    },
    {
      "epoch": 0.5167640598449998,
      "grad_norm": 0.4446750283241272,
      "learning_rate": 2.4163869203033698e-06,
      "loss": 0.6902,
      "step": 12469
    },
    {
      "epoch": 0.5168055037506735,
      "grad_norm": 0.4111630618572235,
      "learning_rate": 2.416179700775001e-06,
      "loss": 0.6537,
      "step": 12470
    },
    {
      "epoch": 0.5168469476563471,
      "grad_norm": 0.4135884642601013,
      "learning_rate": 2.415972481246633e-06,
      "loss": 0.6786,
      "step": 12471
    },
    {
      "epoch": 0.5168883915620208,
      "grad_norm": 0.4348341226577759,
      "learning_rate": 2.4157652617182644e-06,
      "loss": 0.6729,
      "step": 12472
    },
    {
      "epoch": 0.5169298354676944,
      "grad_norm": 0.40917840600013733,
      "learning_rate": 2.415558042189896e-06,
      "loss": 0.61,
      "step": 12473
    },
    {
      "epoch": 0.5169712793733682,
      "grad_norm": 0.4171946346759796,
      "learning_rate": 2.4153508226615275e-06,
      "loss": 0.6383,
      "step": 12474
    },
    {
      "epoch": 0.5170127232790418,
      "grad_norm": 0.41365399956703186,
      "learning_rate": 2.4151436031331598e-06,
      "loss": 0.6841,
      "step": 12475
    },
    {
      "epoch": 0.5170541671847155,
      "grad_norm": 0.42516109347343445,
      "learning_rate": 2.414936383604791e-06,
      "loss": 0.7195,
      "step": 12476
    },
    {
      "epoch": 0.5170956110903892,
      "grad_norm": 0.43032506108283997,
      "learning_rate": 2.414729164076423e-06,
      "loss": 0.6857,
      "step": 12477
    },
    {
      "epoch": 0.5171370549960629,
      "grad_norm": 0.45707836747169495,
      "learning_rate": 2.4145219445480544e-06,
      "loss": 0.7461,
      "step": 12478
    },
    {
      "epoch": 0.5171784989017365,
      "grad_norm": 0.41527754068374634,
      "learning_rate": 2.414314725019686e-06,
      "loss": 0.6799,
      "step": 12479
    },
    {
      "epoch": 0.5172199428074101,
      "grad_norm": 0.38724255561828613,
      "learning_rate": 2.4141075054913176e-06,
      "loss": 0.6934,
      "step": 12480
    },
    {
      "epoch": 0.5172613867130839,
      "grad_norm": 0.39848360419273376,
      "learning_rate": 2.4139002859629494e-06,
      "loss": 0.7003,
      "step": 12481
    },
    {
      "epoch": 0.5173028306187575,
      "grad_norm": 0.40624484419822693,
      "learning_rate": 2.4136930664345807e-06,
      "loss": 0.6903,
      "step": 12482
    },
    {
      "epoch": 0.5173442745244312,
      "grad_norm": 0.3953806757926941,
      "learning_rate": 2.4134858469062126e-06,
      "loss": 0.718,
      "step": 12483
    },
    {
      "epoch": 0.5173857184301048,
      "grad_norm": 0.4093821942806244,
      "learning_rate": 2.4132786273778444e-06,
      "loss": 0.6654,
      "step": 12484
    },
    {
      "epoch": 0.5174271623357786,
      "grad_norm": 0.44131216406822205,
      "learning_rate": 2.413071407849476e-06,
      "loss": 0.7002,
      "step": 12485
    },
    {
      "epoch": 0.5174686062414522,
      "grad_norm": 0.40850284695625305,
      "learning_rate": 2.4128641883211076e-06,
      "loss": 0.639,
      "step": 12486
    },
    {
      "epoch": 0.5175100501471258,
      "grad_norm": 0.43210911750793457,
      "learning_rate": 2.4126569687927394e-06,
      "loss": 0.6929,
      "step": 12487
    },
    {
      "epoch": 0.5175514940527995,
      "grad_norm": 0.4152526557445526,
      "learning_rate": 2.4124497492643708e-06,
      "loss": 0.6838,
      "step": 12488
    },
    {
      "epoch": 0.5175929379584732,
      "grad_norm": 0.41131117939949036,
      "learning_rate": 2.4122425297360026e-06,
      "loss": 0.7144,
      "step": 12489
    },
    {
      "epoch": 0.5176343818641469,
      "grad_norm": 0.4109557271003723,
      "learning_rate": 2.412035310207634e-06,
      "loss": 0.687,
      "step": 12490
    },
    {
      "epoch": 0.5176758257698205,
      "grad_norm": 0.4735833406448364,
      "learning_rate": 2.4118280906792658e-06,
      "loss": 0.6871,
      "step": 12491
    },
    {
      "epoch": 0.5177172696754943,
      "grad_norm": 0.4004615247249603,
      "learning_rate": 2.411620871150897e-06,
      "loss": 0.7136,
      "step": 12492
    },
    {
      "epoch": 0.5177587135811679,
      "grad_norm": 0.42039746046066284,
      "learning_rate": 2.4114136516225294e-06,
      "loss": 0.6754,
      "step": 12493
    },
    {
      "epoch": 0.5178001574868416,
      "grad_norm": 0.4239159822463989,
      "learning_rate": 2.4112064320941608e-06,
      "loss": 0.6909,
      "step": 12494
    },
    {
      "epoch": 0.5178416013925152,
      "grad_norm": 0.40937042236328125,
      "learning_rate": 2.4109992125657926e-06,
      "loss": 0.696,
      "step": 12495
    },
    {
      "epoch": 0.5178830452981888,
      "grad_norm": 0.43401363492012024,
      "learning_rate": 2.410791993037424e-06,
      "loss": 0.6262,
      "step": 12496
    },
    {
      "epoch": 0.5179244892038626,
      "grad_norm": 0.3982897698879242,
      "learning_rate": 2.4105847735090558e-06,
      "loss": 0.6704,
      "step": 12497
    },
    {
      "epoch": 0.5179659331095362,
      "grad_norm": 0.4086677134037018,
      "learning_rate": 2.410377553980687e-06,
      "loss": 0.6906,
      "step": 12498
    },
    {
      "epoch": 0.5180073770152099,
      "grad_norm": 0.42691051959991455,
      "learning_rate": 2.410170334452319e-06,
      "loss": 0.7454,
      "step": 12499
    },
    {
      "epoch": 0.5180488209208836,
      "grad_norm": 0.40323272347450256,
      "learning_rate": 2.4099631149239504e-06,
      "loss": 0.6511,
      "step": 12500
    },
    {
      "epoch": 0.5180902648265573,
      "grad_norm": 0.36057355999946594,
      "learning_rate": 2.4097558953955826e-06,
      "loss": 0.6216,
      "step": 12501
    },
    {
      "epoch": 0.5181317087322309,
      "grad_norm": 0.41945764422416687,
      "learning_rate": 2.409548675867214e-06,
      "loss": 0.6553,
      "step": 12502
    },
    {
      "epoch": 0.5181731526379046,
      "grad_norm": 0.41368958353996277,
      "learning_rate": 2.4093414563388458e-06,
      "loss": 0.6809,
      "step": 12503
    },
    {
      "epoch": 0.5182145965435783,
      "grad_norm": 0.4240632653236389,
      "learning_rate": 2.409134236810477e-06,
      "loss": 0.6995,
      "step": 12504
    },
    {
      "epoch": 0.5182560404492519,
      "grad_norm": 0.412204772233963,
      "learning_rate": 2.408927017282109e-06,
      "loss": 0.6866,
      "step": 12505
    },
    {
      "epoch": 0.5182974843549256,
      "grad_norm": 0.40530160069465637,
      "learning_rate": 2.4087197977537404e-06,
      "loss": 0.6797,
      "step": 12506
    },
    {
      "epoch": 0.5183389282605992,
      "grad_norm": 0.40192726254463196,
      "learning_rate": 2.408512578225372e-06,
      "loss": 0.655,
      "step": 12507
    },
    {
      "epoch": 0.518380372166273,
      "grad_norm": 0.408189058303833,
      "learning_rate": 2.4083053586970036e-06,
      "loss": 0.7059,
      "step": 12508
    },
    {
      "epoch": 0.5184218160719466,
      "grad_norm": 0.4096115827560425,
      "learning_rate": 2.4080981391686354e-06,
      "loss": 0.6708,
      "step": 12509
    },
    {
      "epoch": 0.5184632599776203,
      "grad_norm": 0.421370267868042,
      "learning_rate": 2.407890919640267e-06,
      "loss": 0.6831,
      "step": 12510
    },
    {
      "epoch": 0.518504703883294,
      "grad_norm": 0.3952125310897827,
      "learning_rate": 2.407683700111899e-06,
      "loss": 0.6335,
      "step": 12511
    },
    {
      "epoch": 0.5185461477889677,
      "grad_norm": 0.4815824329853058,
      "learning_rate": 2.4074764805835304e-06,
      "loss": 0.689,
      "step": 12512
    },
    {
      "epoch": 0.5185875916946413,
      "grad_norm": 0.43765345215797424,
      "learning_rate": 2.407269261055162e-06,
      "loss": 0.6829,
      "step": 12513
    },
    {
      "epoch": 0.5186290356003149,
      "grad_norm": 0.40729421377182007,
      "learning_rate": 2.4070620415267936e-06,
      "loss": 0.6852,
      "step": 12514
    },
    {
      "epoch": 0.5186704795059887,
      "grad_norm": 0.39820703864097595,
      "learning_rate": 2.4068548219984254e-06,
      "loss": 0.688,
      "step": 12515
    },
    {
      "epoch": 0.5187119234116623,
      "grad_norm": 0.41707485914230347,
      "learning_rate": 2.4066476024700568e-06,
      "loss": 0.6785,
      "step": 12516
    },
    {
      "epoch": 0.518753367317336,
      "grad_norm": 0.4507840871810913,
      "learning_rate": 2.4064403829416886e-06,
      "loss": 0.7305,
      "step": 12517
    },
    {
      "epoch": 0.5187948112230096,
      "grad_norm": 0.4300510585308075,
      "learning_rate": 2.40623316341332e-06,
      "loss": 0.7275,
      "step": 12518
    },
    {
      "epoch": 0.5188362551286834,
      "grad_norm": 0.43752312660217285,
      "learning_rate": 2.406025943884952e-06,
      "loss": 0.6829,
      "step": 12519
    },
    {
      "epoch": 0.518877699034357,
      "grad_norm": 0.436367928981781,
      "learning_rate": 2.4058187243565836e-06,
      "loss": 0.671,
      "step": 12520
    },
    {
      "epoch": 0.5189191429400307,
      "grad_norm": 0.39381086826324463,
      "learning_rate": 2.4056115048282154e-06,
      "loss": 0.6594,
      "step": 12521
    },
    {
      "epoch": 0.5189605868457043,
      "grad_norm": 0.3898296356201172,
      "learning_rate": 2.4054042852998468e-06,
      "loss": 0.6641,
      "step": 12522
    },
    {
      "epoch": 0.519002030751378,
      "grad_norm": 0.3970431685447693,
      "learning_rate": 2.4051970657714786e-06,
      "loss": 0.6682,
      "step": 12523
    },
    {
      "epoch": 0.5190434746570517,
      "grad_norm": 0.3919110596179962,
      "learning_rate": 2.40498984624311e-06,
      "loss": 0.6379,
      "step": 12524
    },
    {
      "epoch": 0.5190849185627253,
      "grad_norm": 0.36473989486694336,
      "learning_rate": 2.4047826267147418e-06,
      "loss": 0.6638,
      "step": 12525
    },
    {
      "epoch": 0.519126362468399,
      "grad_norm": 0.4072173237800598,
      "learning_rate": 2.404575407186373e-06,
      "loss": 0.7156,
      "step": 12526
    },
    {
      "epoch": 0.5191678063740727,
      "grad_norm": 0.39302361011505127,
      "learning_rate": 2.404368187658005e-06,
      "loss": 0.6898,
      "step": 12527
    },
    {
      "epoch": 0.5192092502797464,
      "grad_norm": 0.3886263370513916,
      "learning_rate": 2.4041609681296368e-06,
      "loss": 0.6519,
      "step": 12528
    },
    {
      "epoch": 0.51925069418542,
      "grad_norm": 0.432587206363678,
      "learning_rate": 2.4039537486012686e-06,
      "loss": 0.7781,
      "step": 12529
    },
    {
      "epoch": 0.5192921380910938,
      "grad_norm": 0.4729192852973938,
      "learning_rate": 2.4037465290729e-06,
      "loss": 0.6837,
      "step": 12530
    },
    {
      "epoch": 0.5193335819967674,
      "grad_norm": 0.4396415948867798,
      "learning_rate": 2.4035393095445318e-06,
      "loss": 0.7422,
      "step": 12531
    },
    {
      "epoch": 0.519375025902441,
      "grad_norm": 0.45288777351379395,
      "learning_rate": 2.403332090016163e-06,
      "loss": 0.6451,
      "step": 12532
    },
    {
      "epoch": 0.5194164698081147,
      "grad_norm": 0.43435826897621155,
      "learning_rate": 2.403124870487795e-06,
      "loss": 0.6577,
      "step": 12533
    },
    {
      "epoch": 0.5194579137137884,
      "grad_norm": 0.40854042768478394,
      "learning_rate": 2.4029176509594264e-06,
      "loss": 0.6938,
      "step": 12534
    },
    {
      "epoch": 0.5194993576194621,
      "grad_norm": 0.37687379121780396,
      "learning_rate": 2.402710431431058e-06,
      "loss": 0.6818,
      "step": 12535
    },
    {
      "epoch": 0.5195408015251357,
      "grad_norm": 0.4056571424007416,
      "learning_rate": 2.40250321190269e-06,
      "loss": 0.637,
      "step": 12536
    },
    {
      "epoch": 0.5195822454308094,
      "grad_norm": 0.43971315026283264,
      "learning_rate": 2.4022959923743218e-06,
      "loss": 0.6849,
      "step": 12537
    },
    {
      "epoch": 0.5196236893364831,
      "grad_norm": 0.4217950105667114,
      "learning_rate": 2.402088772845953e-06,
      "loss": 0.6694,
      "step": 12538
    },
    {
      "epoch": 0.5196651332421568,
      "grad_norm": 0.4526236951351166,
      "learning_rate": 2.401881553317585e-06,
      "loss": 0.709,
      "step": 12539
    },
    {
      "epoch": 0.5197065771478304,
      "grad_norm": 0.3883284628391266,
      "learning_rate": 2.4016743337892164e-06,
      "loss": 0.6967,
      "step": 12540
    },
    {
      "epoch": 0.519748021053504,
      "grad_norm": 0.38360580801963806,
      "learning_rate": 2.401467114260848e-06,
      "loss": 0.5983,
      "step": 12541
    },
    {
      "epoch": 0.5197894649591778,
      "grad_norm": 0.4257265329360962,
      "learning_rate": 2.4012598947324796e-06,
      "loss": 0.6892,
      "step": 12542
    },
    {
      "epoch": 0.5198309088648514,
      "grad_norm": 0.43864619731903076,
      "learning_rate": 2.4010526752041114e-06,
      "loss": 0.7433,
      "step": 12543
    },
    {
      "epoch": 0.5198723527705251,
      "grad_norm": 0.3953695595264435,
      "learning_rate": 2.4008454556757428e-06,
      "loss": 0.718,
      "step": 12544
    },
    {
      "epoch": 0.5199137966761987,
      "grad_norm": 0.4083937704563141,
      "learning_rate": 2.400638236147375e-06,
      "loss": 0.6782,
      "step": 12545
    },
    {
      "epoch": 0.5199552405818725,
      "grad_norm": 0.383492648601532,
      "learning_rate": 2.4004310166190064e-06,
      "loss": 0.6672,
      "step": 12546
    },
    {
      "epoch": 0.5199966844875461,
      "grad_norm": 0.4017408788204193,
      "learning_rate": 2.400223797090638e-06,
      "loss": 0.6809,
      "step": 12547
    },
    {
      "epoch": 0.5200381283932197,
      "grad_norm": 0.4100874662399292,
      "learning_rate": 2.4000165775622696e-06,
      "loss": 0.6572,
      "step": 12548
    },
    {
      "epoch": 0.5200795722988935,
      "grad_norm": 0.3927365839481354,
      "learning_rate": 2.3998093580339014e-06,
      "loss": 0.6967,
      "step": 12549
    },
    {
      "epoch": 0.5201210162045671,
      "grad_norm": 0.38179871439933777,
      "learning_rate": 2.3996021385055328e-06,
      "loss": 0.7109,
      "step": 12550
    },
    {
      "epoch": 0.5201624601102408,
      "grad_norm": 0.4520033895969391,
      "learning_rate": 2.3993949189771646e-06,
      "loss": 0.6787,
      "step": 12551
    },
    {
      "epoch": 0.5202039040159144,
      "grad_norm": 0.3848505914211273,
      "learning_rate": 2.399187699448796e-06,
      "loss": 0.6555,
      "step": 12552
    },
    {
      "epoch": 0.5202453479215882,
      "grad_norm": 0.43456435203552246,
      "learning_rate": 2.3989804799204278e-06,
      "loss": 0.7583,
      "step": 12553
    },
    {
      "epoch": 0.5202867918272618,
      "grad_norm": 0.37724238634109497,
      "learning_rate": 2.3987732603920596e-06,
      "loss": 0.6694,
      "step": 12554
    },
    {
      "epoch": 0.5203282357329355,
      "grad_norm": 0.4380587935447693,
      "learning_rate": 2.3985660408636914e-06,
      "loss": 0.6631,
      "step": 12555
    },
    {
      "epoch": 0.5203696796386091,
      "grad_norm": 0.38457006216049194,
      "learning_rate": 2.3983588213353228e-06,
      "loss": 0.6522,
      "step": 12556
    },
    {
      "epoch": 0.5204111235442828,
      "grad_norm": 0.4093158543109894,
      "learning_rate": 2.3981516018069546e-06,
      "loss": 0.6672,
      "step": 12557
    },
    {
      "epoch": 0.5204525674499565,
      "grad_norm": 0.40698012709617615,
      "learning_rate": 2.397944382278586e-06,
      "loss": 0.7083,
      "step": 12558
    },
    {
      "epoch": 0.5204940113556301,
      "grad_norm": 0.38175714015960693,
      "learning_rate": 2.3977371627502178e-06,
      "loss": 0.6394,
      "step": 12559
    },
    {
      "epoch": 0.5205354552613038,
      "grad_norm": 0.3884532153606415,
      "learning_rate": 2.397529943221849e-06,
      "loss": 0.6873,
      "step": 12560
    },
    {
      "epoch": 0.5205768991669775,
      "grad_norm": 0.4071963131427765,
      "learning_rate": 2.397322723693481e-06,
      "loss": 0.7072,
      "step": 12561
    },
    {
      "epoch": 0.5206183430726512,
      "grad_norm": 0.41138336062431335,
      "learning_rate": 2.3971155041651128e-06,
      "loss": 0.7156,
      "step": 12562
    },
    {
      "epoch": 0.5206597869783248,
      "grad_norm": 0.43066808581352234,
      "learning_rate": 2.3969082846367446e-06,
      "loss": 0.6946,
      "step": 12563
    },
    {
      "epoch": 0.5207012308839986,
      "grad_norm": 0.42132607102394104,
      "learning_rate": 2.396701065108376e-06,
      "loss": 0.657,
      "step": 12564
    },
    {
      "epoch": 0.5207426747896722,
      "grad_norm": 0.391268253326416,
      "learning_rate": 2.3964938455800078e-06,
      "loss": 0.6702,
      "step": 12565
    },
    {
      "epoch": 0.5207841186953458,
      "grad_norm": 0.43152138590812683,
      "learning_rate": 2.396286626051639e-06,
      "loss": 0.6707,
      "step": 12566
    },
    {
      "epoch": 0.5208255626010195,
      "grad_norm": 0.4148610830307007,
      "learning_rate": 2.396079406523271e-06,
      "loss": 0.6992,
      "step": 12567
    },
    {
      "epoch": 0.5208670065066932,
      "grad_norm": 0.41824865341186523,
      "learning_rate": 2.3958721869949024e-06,
      "loss": 0.6852,
      "step": 12568
    },
    {
      "epoch": 0.5209084504123669,
      "grad_norm": 0.43457284569740295,
      "learning_rate": 2.395664967466534e-06,
      "loss": 0.7435,
      "step": 12569
    },
    {
      "epoch": 0.5209498943180405,
      "grad_norm": 0.4613175094127655,
      "learning_rate": 2.3954577479381656e-06,
      "loss": 0.7152,
      "step": 12570
    },
    {
      "epoch": 0.5209913382237142,
      "grad_norm": 0.4210778474807739,
      "learning_rate": 2.395250528409798e-06,
      "loss": 0.7275,
      "step": 12571
    },
    {
      "epoch": 0.5210327821293879,
      "grad_norm": 0.42776134610176086,
      "learning_rate": 2.395043308881429e-06,
      "loss": 0.6748,
      "step": 12572
    },
    {
      "epoch": 0.5210742260350616,
      "grad_norm": 0.3835519850254059,
      "learning_rate": 2.394836089353061e-06,
      "loss": 0.6061,
      "step": 12573
    },
    {
      "epoch": 0.5211156699407352,
      "grad_norm": 0.4283444583415985,
      "learning_rate": 2.3946288698246924e-06,
      "loss": 0.6697,
      "step": 12574
    },
    {
      "epoch": 0.5211571138464088,
      "grad_norm": 0.3993901312351227,
      "learning_rate": 2.394421650296324e-06,
      "loss": 0.6556,
      "step": 12575
    },
    {
      "epoch": 0.5211985577520826,
      "grad_norm": 0.41993531584739685,
      "learning_rate": 2.3942144307679556e-06,
      "loss": 0.6781,
      "step": 12576
    },
    {
      "epoch": 0.5212400016577562,
      "grad_norm": 0.4229331314563751,
      "learning_rate": 2.3940072112395874e-06,
      "loss": 0.608,
      "step": 12577
    },
    {
      "epoch": 0.5212814455634299,
      "grad_norm": 0.3984411954879761,
      "learning_rate": 2.3937999917112188e-06,
      "loss": 0.6804,
      "step": 12578
    },
    {
      "epoch": 0.5213228894691035,
      "grad_norm": 0.40265849232673645,
      "learning_rate": 2.3935927721828506e-06,
      "loss": 0.7041,
      "step": 12579
    },
    {
      "epoch": 0.5213643333747773,
      "grad_norm": 0.3807474970817566,
      "learning_rate": 2.3933855526544824e-06,
      "loss": 0.681,
      "step": 12580
    },
    {
      "epoch": 0.5214057772804509,
      "grad_norm": 0.44563746452331543,
      "learning_rate": 2.393178333126114e-06,
      "loss": 0.7412,
      "step": 12581
    },
    {
      "epoch": 0.5214472211861246,
      "grad_norm": 0.43464934825897217,
      "learning_rate": 2.3929711135977456e-06,
      "loss": 0.6731,
      "step": 12582
    },
    {
      "epoch": 0.5214886650917983,
      "grad_norm": 0.46325355768203735,
      "learning_rate": 2.3927638940693774e-06,
      "loss": 0.7142,
      "step": 12583
    },
    {
      "epoch": 0.5215301089974719,
      "grad_norm": 0.4135199189186096,
      "learning_rate": 2.3925566745410088e-06,
      "loss": 0.6268,
      "step": 12584
    },
    {
      "epoch": 0.5215715529031456,
      "grad_norm": 0.43282660841941833,
      "learning_rate": 2.3923494550126406e-06,
      "loss": 0.6584,
      "step": 12585
    },
    {
      "epoch": 0.5216129968088192,
      "grad_norm": 0.4272555112838745,
      "learning_rate": 2.392142235484272e-06,
      "loss": 0.6584,
      "step": 12586
    },
    {
      "epoch": 0.521654440714493,
      "grad_norm": 0.4186498522758484,
      "learning_rate": 2.3919350159559038e-06,
      "loss": 0.708,
      "step": 12587
    },
    {
      "epoch": 0.5216958846201666,
      "grad_norm": 0.38211435079574585,
      "learning_rate": 2.3917277964275356e-06,
      "loss": 0.6737,
      "step": 12588
    },
    {
      "epoch": 0.5217373285258403,
      "grad_norm": 0.3978312611579895,
      "learning_rate": 2.3915205768991674e-06,
      "loss": 0.6486,
      "step": 12589
    },
    {
      "epoch": 0.5217787724315139,
      "grad_norm": 0.4126531779766083,
      "learning_rate": 2.3913133573707988e-06,
      "loss": 0.6926,
      "step": 12590
    },
    {
      "epoch": 0.5218202163371877,
      "grad_norm": 0.40570902824401855,
      "learning_rate": 2.3911061378424306e-06,
      "loss": 0.6479,
      "step": 12591
    },
    {
      "epoch": 0.5218616602428613,
      "grad_norm": 0.41340118646621704,
      "learning_rate": 2.390898918314062e-06,
      "loss": 0.6962,
      "step": 12592
    },
    {
      "epoch": 0.5219031041485349,
      "grad_norm": 0.4114457666873932,
      "learning_rate": 2.3906916987856938e-06,
      "loss": 0.6985,
      "step": 12593
    },
    {
      "epoch": 0.5219445480542086,
      "grad_norm": 0.40483883023262024,
      "learning_rate": 2.390484479257325e-06,
      "loss": 0.6935,
      "step": 12594
    },
    {
      "epoch": 0.5219859919598823,
      "grad_norm": 0.4975307583808899,
      "learning_rate": 2.390277259728957e-06,
      "loss": 0.7463,
      "step": 12595
    },
    {
      "epoch": 0.522027435865556,
      "grad_norm": 0.40892666578292847,
      "learning_rate": 2.3900700402005888e-06,
      "loss": 0.6826,
      "step": 12596
    },
    {
      "epoch": 0.5220688797712296,
      "grad_norm": 0.40424248576164246,
      "learning_rate": 2.3898628206722206e-06,
      "loss": 0.6951,
      "step": 12597
    },
    {
      "epoch": 0.5221103236769034,
      "grad_norm": 0.42291751503944397,
      "learning_rate": 2.389655601143852e-06,
      "loss": 0.6948,
      "step": 12598
    },
    {
      "epoch": 0.522151767582577,
      "grad_norm": 0.4212879240512848,
      "learning_rate": 2.3894483816154838e-06,
      "loss": 0.7129,
      "step": 12599
    },
    {
      "epoch": 0.5221932114882507,
      "grad_norm": 0.4208809435367584,
      "learning_rate": 2.389241162087115e-06,
      "loss": 0.6791,
      "step": 12600
    },
    {
      "epoch": 0.5222346553939243,
      "grad_norm": 0.4288797080516815,
      "learning_rate": 2.389033942558747e-06,
      "loss": 0.6738,
      "step": 12601
    },
    {
      "epoch": 0.522276099299598,
      "grad_norm": 0.3962031900882721,
      "learning_rate": 2.3888267230303784e-06,
      "loss": 0.6794,
      "step": 12602
    },
    {
      "epoch": 0.5223175432052717,
      "grad_norm": 0.44279807806015015,
      "learning_rate": 2.38861950350201e-06,
      "loss": 0.7302,
      "step": 12603
    },
    {
      "epoch": 0.5223589871109453,
      "grad_norm": 0.41808339953422546,
      "learning_rate": 2.388412283973642e-06,
      "loss": 0.6888,
      "step": 12604
    },
    {
      "epoch": 0.522400431016619,
      "grad_norm": 0.4430859386920929,
      "learning_rate": 2.3882050644452734e-06,
      "loss": 0.6876,
      "step": 12605
    },
    {
      "epoch": 0.5224418749222927,
      "grad_norm": 0.39555037021636963,
      "learning_rate": 2.387997844916905e-06,
      "loss": 0.6825,
      "step": 12606
    },
    {
      "epoch": 0.5224833188279664,
      "grad_norm": 0.4349897801876068,
      "learning_rate": 2.387790625388537e-06,
      "loss": 0.6875,
      "step": 12607
    },
    {
      "epoch": 0.52252476273364,
      "grad_norm": 0.4097515046596527,
      "learning_rate": 2.3875834058601684e-06,
      "loss": 0.6334,
      "step": 12608
    },
    {
      "epoch": 0.5225662066393136,
      "grad_norm": 0.3824368715286255,
      "learning_rate": 2.3873761863318e-06,
      "loss": 0.6366,
      "step": 12609
    },
    {
      "epoch": 0.5226076505449874,
      "grad_norm": 0.4280630052089691,
      "learning_rate": 2.3871689668034316e-06,
      "loss": 0.6962,
      "step": 12610
    },
    {
      "epoch": 0.522649094450661,
      "grad_norm": 0.4225119650363922,
      "learning_rate": 2.3869617472750634e-06,
      "loss": 0.6407,
      "step": 12611
    },
    {
      "epoch": 0.5226905383563347,
      "grad_norm": 0.44540655612945557,
      "learning_rate": 2.3867545277466948e-06,
      "loss": 0.7183,
      "step": 12612
    },
    {
      "epoch": 0.5227319822620083,
      "grad_norm": 0.373508483171463,
      "learning_rate": 2.3865473082183266e-06,
      "loss": 0.7096,
      "step": 12613
    },
    {
      "epoch": 0.5227734261676821,
      "grad_norm": 0.37904226779937744,
      "learning_rate": 2.3863400886899584e-06,
      "loss": 0.6919,
      "step": 12614
    },
    {
      "epoch": 0.5228148700733557,
      "grad_norm": 0.4333645701408386,
      "learning_rate": 2.38613286916159e-06,
      "loss": 0.74,
      "step": 12615
    },
    {
      "epoch": 0.5228563139790294,
      "grad_norm": 0.425849974155426,
      "learning_rate": 2.3859256496332216e-06,
      "loss": 0.6704,
      "step": 12616
    },
    {
      "epoch": 0.522897757884703,
      "grad_norm": 0.42296767234802246,
      "learning_rate": 2.3857184301048534e-06,
      "loss": 0.7142,
      "step": 12617
    },
    {
      "epoch": 0.5229392017903767,
      "grad_norm": 0.4082709550857544,
      "learning_rate": 2.3855112105764848e-06,
      "loss": 0.6183,
      "step": 12618
    },
    {
      "epoch": 0.5229806456960504,
      "grad_norm": 0.3941473960876465,
      "learning_rate": 2.3853039910481166e-06,
      "loss": 0.6584,
      "step": 12619
    },
    {
      "epoch": 0.523022089601724,
      "grad_norm": 0.4375854730606079,
      "learning_rate": 2.385096771519748e-06,
      "loss": 0.7272,
      "step": 12620
    },
    {
      "epoch": 0.5230635335073978,
      "grad_norm": 0.40812429785728455,
      "learning_rate": 2.3848895519913798e-06,
      "loss": 0.688,
      "step": 12621
    },
    {
      "epoch": 0.5231049774130714,
      "grad_norm": 0.4617955982685089,
      "learning_rate": 2.3846823324630116e-06,
      "loss": 0.7085,
      "step": 12622
    },
    {
      "epoch": 0.5231464213187451,
      "grad_norm": 0.45311516523361206,
      "learning_rate": 2.3844751129346434e-06,
      "loss": 0.7429,
      "step": 12623
    },
    {
      "epoch": 0.5231878652244187,
      "grad_norm": 0.4297454357147217,
      "learning_rate": 2.3842678934062748e-06,
      "loss": 0.6945,
      "step": 12624
    },
    {
      "epoch": 0.5232293091300925,
      "grad_norm": 0.41116654872894287,
      "learning_rate": 2.3840606738779066e-06,
      "loss": 0.6337,
      "step": 12625
    },
    {
      "epoch": 0.5232707530357661,
      "grad_norm": 0.38406509160995483,
      "learning_rate": 2.383853454349538e-06,
      "loss": 0.6455,
      "step": 12626
    },
    {
      "epoch": 0.5233121969414397,
      "grad_norm": 0.3692176342010498,
      "learning_rate": 2.3836462348211698e-06,
      "loss": 0.6707,
      "step": 12627
    },
    {
      "epoch": 0.5233536408471134,
      "grad_norm": 0.44967779517173767,
      "learning_rate": 2.383439015292801e-06,
      "loss": 0.7078,
      "step": 12628
    },
    {
      "epoch": 0.5233950847527871,
      "grad_norm": 0.4245588481426239,
      "learning_rate": 2.383231795764433e-06,
      "loss": 0.6593,
      "step": 12629
    },
    {
      "epoch": 0.5234365286584608,
      "grad_norm": 0.41723015904426575,
      "learning_rate": 2.383024576236065e-06,
      "loss": 0.6547,
      "step": 12630
    },
    {
      "epoch": 0.5234779725641344,
      "grad_norm": 0.42626410722732544,
      "learning_rate": 2.382817356707696e-06,
      "loss": 0.661,
      "step": 12631
    },
    {
      "epoch": 0.5235194164698082,
      "grad_norm": 0.3979733884334564,
      "learning_rate": 2.382610137179328e-06,
      "loss": 0.6725,
      "step": 12632
    },
    {
      "epoch": 0.5235608603754818,
      "grad_norm": 0.37187525629997253,
      "learning_rate": 2.38240291765096e-06,
      "loss": 0.6332,
      "step": 12633
    },
    {
      "epoch": 0.5236023042811555,
      "grad_norm": 0.4311836063861847,
      "learning_rate": 2.382195698122591e-06,
      "loss": 0.6809,
      "step": 12634
    },
    {
      "epoch": 0.5236437481868291,
      "grad_norm": 0.4104052782058716,
      "learning_rate": 2.381988478594223e-06,
      "loss": 0.6768,
      "step": 12635
    },
    {
      "epoch": 0.5236851920925027,
      "grad_norm": 0.41102686524391174,
      "learning_rate": 2.3817812590658544e-06,
      "loss": 0.646,
      "step": 12636
    },
    {
      "epoch": 0.5237266359981765,
      "grad_norm": 0.44484105706214905,
      "learning_rate": 2.381574039537486e-06,
      "loss": 0.6914,
      "step": 12637
    },
    {
      "epoch": 0.5237680799038501,
      "grad_norm": 0.4190810024738312,
      "learning_rate": 2.381366820009118e-06,
      "loss": 0.7161,
      "step": 12638
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 0.43017104268074036,
      "learning_rate": 2.3811596004807494e-06,
      "loss": 0.7102,
      "step": 12639
    },
    {
      "epoch": 0.5238509677151975,
      "grad_norm": 0.4285981059074402,
      "learning_rate": 2.380952380952381e-06,
      "loss": 0.7449,
      "step": 12640
    },
    {
      "epoch": 0.5238924116208712,
      "grad_norm": 0.412391722202301,
      "learning_rate": 2.380745161424013e-06,
      "loss": 0.7083,
      "step": 12641
    },
    {
      "epoch": 0.5239338555265448,
      "grad_norm": 0.45212551951408386,
      "learning_rate": 2.3805379418956444e-06,
      "loss": 0.7373,
      "step": 12642
    },
    {
      "epoch": 0.5239752994322185,
      "grad_norm": 0.4177693724632263,
      "learning_rate": 2.380330722367276e-06,
      "loss": 0.7188,
      "step": 12643
    },
    {
      "epoch": 0.5240167433378922,
      "grad_norm": 0.39702433347702026,
      "learning_rate": 2.3801235028389076e-06,
      "loss": 0.6466,
      "step": 12644
    },
    {
      "epoch": 0.5240581872435658,
      "grad_norm": 0.40061143040657043,
      "learning_rate": 2.3799162833105394e-06,
      "loss": 0.6869,
      "step": 12645
    },
    {
      "epoch": 0.5240996311492395,
      "grad_norm": 0.4022364914417267,
      "learning_rate": 2.3797090637821708e-06,
      "loss": 0.6891,
      "step": 12646
    },
    {
      "epoch": 0.5241410750549131,
      "grad_norm": 0.42555657029151917,
      "learning_rate": 2.3795018442538026e-06,
      "loss": 0.6802,
      "step": 12647
    },
    {
      "epoch": 0.5241825189605869,
      "grad_norm": 0.4297129213809967,
      "learning_rate": 2.3792946247254344e-06,
      "loss": 0.7517,
      "step": 12648
    },
    {
      "epoch": 0.5242239628662605,
      "grad_norm": 0.3912673592567444,
      "learning_rate": 2.379087405197066e-06,
      "loss": 0.7084,
      "step": 12649
    },
    {
      "epoch": 0.5242654067719342,
      "grad_norm": 0.3703274428844452,
      "learning_rate": 2.3788801856686976e-06,
      "loss": 0.6738,
      "step": 12650
    },
    {
      "epoch": 0.5243068506776078,
      "grad_norm": 0.43826237320899963,
      "learning_rate": 2.3786729661403294e-06,
      "loss": 0.6863,
      "step": 12651
    },
    {
      "epoch": 0.5243482945832816,
      "grad_norm": 0.42081063985824585,
      "learning_rate": 2.3784657466119608e-06,
      "loss": 0.6898,
      "step": 12652
    },
    {
      "epoch": 0.5243897384889552,
      "grad_norm": 0.3877864480018616,
      "learning_rate": 2.3782585270835926e-06,
      "loss": 0.6293,
      "step": 12653
    },
    {
      "epoch": 0.5244311823946288,
      "grad_norm": 0.3883492648601532,
      "learning_rate": 2.378051307555224e-06,
      "loss": 0.624,
      "step": 12654
    },
    {
      "epoch": 0.5244726263003026,
      "grad_norm": 0.45694699883461,
      "learning_rate": 2.3778440880268558e-06,
      "loss": 0.6716,
      "step": 12655
    },
    {
      "epoch": 0.5245140702059762,
      "grad_norm": 0.4060195982456207,
      "learning_rate": 2.3776368684984876e-06,
      "loss": 0.7324,
      "step": 12656
    },
    {
      "epoch": 0.5245555141116499,
      "grad_norm": 0.4088493883609772,
      "learning_rate": 2.377429648970119e-06,
      "loss": 0.656,
      "step": 12657
    },
    {
      "epoch": 0.5245969580173235,
      "grad_norm": 0.40790894627571106,
      "learning_rate": 2.3772224294417508e-06,
      "loss": 0.7104,
      "step": 12658
    },
    {
      "epoch": 0.5246384019229973,
      "grad_norm": 0.40414130687713623,
      "learning_rate": 2.3770152099133826e-06,
      "loss": 0.686,
      "step": 12659
    },
    {
      "epoch": 0.5246798458286709,
      "grad_norm": 0.37157467007637024,
      "learning_rate": 2.376807990385014e-06,
      "loss": 0.6274,
      "step": 12660
    },
    {
      "epoch": 0.5247212897343445,
      "grad_norm": 0.40897637605667114,
      "learning_rate": 2.376600770856646e-06,
      "loss": 0.6926,
      "step": 12661
    },
    {
      "epoch": 0.5247627336400182,
      "grad_norm": 0.43411290645599365,
      "learning_rate": 2.376393551328277e-06,
      "loss": 0.7273,
      "step": 12662
    },
    {
      "epoch": 0.5248041775456919,
      "grad_norm": 0.43205928802490234,
      "learning_rate": 2.376186331799909e-06,
      "loss": 0.7333,
      "step": 12663
    },
    {
      "epoch": 0.5248456214513656,
      "grad_norm": 0.4218291938304901,
      "learning_rate": 2.375979112271541e-06,
      "loss": 0.6642,
      "step": 12664
    },
    {
      "epoch": 0.5248870653570392,
      "grad_norm": 0.4380809962749481,
      "learning_rate": 2.375771892743172e-06,
      "loss": 0.6683,
      "step": 12665
    },
    {
      "epoch": 0.524928509262713,
      "grad_norm": 0.37009352445602417,
      "learning_rate": 2.375564673214804e-06,
      "loss": 0.6428,
      "step": 12666
    },
    {
      "epoch": 0.5249699531683866,
      "grad_norm": 0.4073854386806488,
      "learning_rate": 2.375357453686436e-06,
      "loss": 0.7047,
      "step": 12667
    },
    {
      "epoch": 0.5250113970740603,
      "grad_norm": 0.3952029049396515,
      "learning_rate": 2.375150234158067e-06,
      "loss": 0.6676,
      "step": 12668
    },
    {
      "epoch": 0.5250528409797339,
      "grad_norm": 0.43570664525032043,
      "learning_rate": 2.374943014629699e-06,
      "loss": 0.7493,
      "step": 12669
    },
    {
      "epoch": 0.5250942848854075,
      "grad_norm": 0.3945210874080658,
      "learning_rate": 2.3747357951013304e-06,
      "loss": 0.6476,
      "step": 12670
    },
    {
      "epoch": 0.5251357287910813,
      "grad_norm": 0.40418878197669983,
      "learning_rate": 2.374528575572962e-06,
      "loss": 0.6658,
      "step": 12671
    },
    {
      "epoch": 0.5251771726967549,
      "grad_norm": 0.4028280973434448,
      "learning_rate": 2.374321356044594e-06,
      "loss": 0.7146,
      "step": 12672
    },
    {
      "epoch": 0.5252186166024286,
      "grad_norm": 0.41503721475601196,
      "learning_rate": 2.3741141365162254e-06,
      "loss": 0.6957,
      "step": 12673
    },
    {
      "epoch": 0.5252600605081023,
      "grad_norm": 0.4172116219997406,
      "learning_rate": 2.373906916987857e-06,
      "loss": 0.6301,
      "step": 12674
    },
    {
      "epoch": 0.525301504413776,
      "grad_norm": 0.4189743995666504,
      "learning_rate": 2.3736996974594886e-06,
      "loss": 0.6731,
      "step": 12675
    },
    {
      "epoch": 0.5253429483194496,
      "grad_norm": 0.4130241870880127,
      "learning_rate": 2.3734924779311204e-06,
      "loss": 0.6619,
      "step": 12676
    },
    {
      "epoch": 0.5253843922251233,
      "grad_norm": 0.4559548795223236,
      "learning_rate": 2.373285258402752e-06,
      "loss": 0.6565,
      "step": 12677
    },
    {
      "epoch": 0.525425836130797,
      "grad_norm": 0.4367161989212036,
      "learning_rate": 2.3730780388743836e-06,
      "loss": 0.6676,
      "step": 12678
    },
    {
      "epoch": 0.5254672800364706,
      "grad_norm": 0.4063808023929596,
      "learning_rate": 2.3728708193460154e-06,
      "loss": 0.689,
      "step": 12679
    },
    {
      "epoch": 0.5255087239421443,
      "grad_norm": 0.409132719039917,
      "learning_rate": 2.3726635998176468e-06,
      "loss": 0.6951,
      "step": 12680
    },
    {
      "epoch": 0.5255501678478179,
      "grad_norm": 0.3958336412906647,
      "learning_rate": 2.3724563802892786e-06,
      "loss": 0.668,
      "step": 12681
    },
    {
      "epoch": 0.5255916117534917,
      "grad_norm": 0.39782172441482544,
      "learning_rate": 2.3722491607609104e-06,
      "loss": 0.7225,
      "step": 12682
    },
    {
      "epoch": 0.5256330556591653,
      "grad_norm": 0.3923264443874359,
      "learning_rate": 2.3720419412325418e-06,
      "loss": 0.6702,
      "step": 12683
    },
    {
      "epoch": 0.525674499564839,
      "grad_norm": 0.42885860800743103,
      "learning_rate": 2.3718347217041736e-06,
      "loss": 0.7006,
      "step": 12684
    },
    {
      "epoch": 0.5257159434705126,
      "grad_norm": 0.4408898949623108,
      "learning_rate": 2.3716275021758054e-06,
      "loss": 0.7454,
      "step": 12685
    },
    {
      "epoch": 0.5257573873761864,
      "grad_norm": 0.4405261278152466,
      "learning_rate": 2.3714202826474368e-06,
      "loss": 0.6964,
      "step": 12686
    },
    {
      "epoch": 0.52579883128186,
      "grad_norm": 0.3965229094028473,
      "learning_rate": 2.3712130631190686e-06,
      "loss": 0.674,
      "step": 12687
    },
    {
      "epoch": 0.5258402751875336,
      "grad_norm": 0.41549956798553467,
      "learning_rate": 2.3710058435907e-06,
      "loss": 0.6814,
      "step": 12688
    },
    {
      "epoch": 0.5258817190932074,
      "grad_norm": 0.40055760741233826,
      "learning_rate": 2.370798624062332e-06,
      "loss": 0.697,
      "step": 12689
    },
    {
      "epoch": 0.525923162998881,
      "grad_norm": 0.40844807028770447,
      "learning_rate": 2.3705914045339636e-06,
      "loss": 0.7136,
      "step": 12690
    },
    {
      "epoch": 0.5259646069045547,
      "grad_norm": 0.4323974549770355,
      "learning_rate": 2.370384185005595e-06,
      "loss": 0.6965,
      "step": 12691
    },
    {
      "epoch": 0.5260060508102283,
      "grad_norm": 0.44082149863243103,
      "learning_rate": 2.370176965477227e-06,
      "loss": 0.6786,
      "step": 12692
    },
    {
      "epoch": 0.5260474947159021,
      "grad_norm": 0.4313060939311981,
      "learning_rate": 2.3699697459488586e-06,
      "loss": 0.6973,
      "step": 12693
    },
    {
      "epoch": 0.5260889386215757,
      "grad_norm": 0.3962545692920685,
      "learning_rate": 2.36976252642049e-06,
      "loss": 0.6483,
      "step": 12694
    },
    {
      "epoch": 0.5261303825272494,
      "grad_norm": 0.42420676350593567,
      "learning_rate": 2.369555306892122e-06,
      "loss": 0.6704,
      "step": 12695
    },
    {
      "epoch": 0.526171826432923,
      "grad_norm": 0.40030965209007263,
      "learning_rate": 2.369348087363753e-06,
      "loss": 0.7094,
      "step": 12696
    },
    {
      "epoch": 0.5262132703385967,
      "grad_norm": 0.4156250059604645,
      "learning_rate": 2.369140867835385e-06,
      "loss": 0.6364,
      "step": 12697
    },
    {
      "epoch": 0.5262547142442704,
      "grad_norm": 0.4453713893890381,
      "learning_rate": 2.368933648307017e-06,
      "loss": 0.6234,
      "step": 12698
    },
    {
      "epoch": 0.526296158149944,
      "grad_norm": 0.47580769658088684,
      "learning_rate": 2.368726428778648e-06,
      "loss": 0.7421,
      "step": 12699
    },
    {
      "epoch": 0.5263376020556177,
      "grad_norm": 0.39875853061676025,
      "learning_rate": 2.36851920925028e-06,
      "loss": 0.675,
      "step": 12700
    },
    {
      "epoch": 0.5263790459612914,
      "grad_norm": 0.3920235335826874,
      "learning_rate": 2.3683119897219114e-06,
      "loss": 0.6654,
      "step": 12701
    },
    {
      "epoch": 0.5264204898669651,
      "grad_norm": 0.397657573223114,
      "learning_rate": 2.368104770193543e-06,
      "loss": 0.6575,
      "step": 12702
    },
    {
      "epoch": 0.5264619337726387,
      "grad_norm": 0.3951939642429352,
      "learning_rate": 2.367897550665175e-06,
      "loss": 0.6737,
      "step": 12703
    },
    {
      "epoch": 0.5265033776783125,
      "grad_norm": 0.4291022717952728,
      "learning_rate": 2.3676903311368064e-06,
      "loss": 0.6667,
      "step": 12704
    },
    {
      "epoch": 0.5265448215839861,
      "grad_norm": 0.43796488642692566,
      "learning_rate": 2.367483111608438e-06,
      "loss": 0.6949,
      "step": 12705
    },
    {
      "epoch": 0.5265862654896597,
      "grad_norm": 0.42261624336242676,
      "learning_rate": 2.36727589208007e-06,
      "loss": 0.7107,
      "step": 12706
    },
    {
      "epoch": 0.5266277093953334,
      "grad_norm": 0.41933420300483704,
      "learning_rate": 2.3670686725517014e-06,
      "loss": 0.7512,
      "step": 12707
    },
    {
      "epoch": 0.526669153301007,
      "grad_norm": 0.4342271089553833,
      "learning_rate": 2.366861453023333e-06,
      "loss": 0.7194,
      "step": 12708
    },
    {
      "epoch": 0.5267105972066808,
      "grad_norm": 0.42758506536483765,
      "learning_rate": 2.3666542334949646e-06,
      "loss": 0.7561,
      "step": 12709
    },
    {
      "epoch": 0.5267520411123544,
      "grad_norm": 0.4329321086406708,
      "learning_rate": 2.3664470139665964e-06,
      "loss": 0.7002,
      "step": 12710
    },
    {
      "epoch": 0.5267934850180281,
      "grad_norm": 0.39651286602020264,
      "learning_rate": 2.366239794438228e-06,
      "loss": 0.6471,
      "step": 12711
    },
    {
      "epoch": 0.5268349289237018,
      "grad_norm": 0.44623979926109314,
      "learning_rate": 2.3660325749098596e-06,
      "loss": 0.7144,
      "step": 12712
    },
    {
      "epoch": 0.5268763728293755,
      "grad_norm": 0.41242149472236633,
      "learning_rate": 2.3658253553814914e-06,
      "loss": 0.656,
      "step": 12713
    },
    {
      "epoch": 0.5269178167350491,
      "grad_norm": 0.4180530607700348,
      "learning_rate": 2.365618135853123e-06,
      "loss": 0.7018,
      "step": 12714
    },
    {
      "epoch": 0.5269592606407227,
      "grad_norm": 0.4093191921710968,
      "learning_rate": 2.3654109163247546e-06,
      "loss": 0.7043,
      "step": 12715
    },
    {
      "epoch": 0.5270007045463965,
      "grad_norm": 0.4319854974746704,
      "learning_rate": 2.3652036967963864e-06,
      "loss": 0.6598,
      "step": 12716
    },
    {
      "epoch": 0.5270421484520701,
      "grad_norm": 0.3933837115764618,
      "learning_rate": 2.3649964772680178e-06,
      "loss": 0.6396,
      "step": 12717
    },
    {
      "epoch": 0.5270835923577438,
      "grad_norm": 0.40750330686569214,
      "learning_rate": 2.3647892577396496e-06,
      "loss": 0.6809,
      "step": 12718
    },
    {
      "epoch": 0.5271250362634174,
      "grad_norm": 0.40311792492866516,
      "learning_rate": 2.3645820382112814e-06,
      "loss": 0.6787,
      "step": 12719
    },
    {
      "epoch": 0.5271664801690912,
      "grad_norm": 0.4228418469429016,
      "learning_rate": 2.364374818682913e-06,
      "loss": 0.6565,
      "step": 12720
    },
    {
      "epoch": 0.5272079240747648,
      "grad_norm": 0.4425879120826721,
      "learning_rate": 2.3641675991545446e-06,
      "loss": 0.6514,
      "step": 12721
    },
    {
      "epoch": 0.5272493679804384,
      "grad_norm": 0.41079095005989075,
      "learning_rate": 2.363960379626176e-06,
      "loss": 0.6942,
      "step": 12722
    },
    {
      "epoch": 0.5272908118861122,
      "grad_norm": 0.41754844784736633,
      "learning_rate": 2.363753160097808e-06,
      "loss": 0.6658,
      "step": 12723
    },
    {
      "epoch": 0.5273322557917858,
      "grad_norm": 0.41991499066352844,
      "learning_rate": 2.3635459405694396e-06,
      "loss": 0.7134,
      "step": 12724
    },
    {
      "epoch": 0.5273736996974595,
      "grad_norm": 0.4074188470840454,
      "learning_rate": 2.363338721041071e-06,
      "loss": 0.6262,
      "step": 12725
    },
    {
      "epoch": 0.5274151436031331,
      "grad_norm": 0.41648292541503906,
      "learning_rate": 2.363131501512703e-06,
      "loss": 0.6594,
      "step": 12726
    },
    {
      "epoch": 0.5274565875088069,
      "grad_norm": 0.4286017417907715,
      "learning_rate": 2.362924281984334e-06,
      "loss": 0.665,
      "step": 12727
    },
    {
      "epoch": 0.5274980314144805,
      "grad_norm": 0.40575817227363586,
      "learning_rate": 2.362717062455966e-06,
      "loss": 0.6743,
      "step": 12728
    },
    {
      "epoch": 0.5275394753201542,
      "grad_norm": 0.3965182602405548,
      "learning_rate": 2.362509842927598e-06,
      "loss": 0.6604,
      "step": 12729
    },
    {
      "epoch": 0.5275809192258278,
      "grad_norm": 0.419206440448761,
      "learning_rate": 2.362302623399229e-06,
      "loss": 0.6644,
      "step": 12730
    },
    {
      "epoch": 0.5276223631315015,
      "grad_norm": 0.4203120768070221,
      "learning_rate": 2.362095403870861e-06,
      "loss": 0.7168,
      "step": 12731
    },
    {
      "epoch": 0.5276638070371752,
      "grad_norm": 0.3980722725391388,
      "learning_rate": 2.361888184342493e-06,
      "loss": 0.634,
      "step": 12732
    },
    {
      "epoch": 0.5277052509428488,
      "grad_norm": 0.4063666760921478,
      "learning_rate": 2.361680964814124e-06,
      "loss": 0.693,
      "step": 12733
    },
    {
      "epoch": 0.5277466948485225,
      "grad_norm": 0.43318161368370056,
      "learning_rate": 2.361473745285756e-06,
      "loss": 0.7529,
      "step": 12734
    },
    {
      "epoch": 0.5277881387541962,
      "grad_norm": 0.44466325640678406,
      "learning_rate": 2.3612665257573874e-06,
      "loss": 0.7136,
      "step": 12735
    },
    {
      "epoch": 0.5278295826598699,
      "grad_norm": 0.3778674900531769,
      "learning_rate": 2.361059306229019e-06,
      "loss": 0.6633,
      "step": 12736
    },
    {
      "epoch": 0.5278710265655435,
      "grad_norm": 0.419139564037323,
      "learning_rate": 2.360852086700651e-06,
      "loss": 0.6665,
      "step": 12737
    },
    {
      "epoch": 0.5279124704712173,
      "grad_norm": 0.4465898871421814,
      "learning_rate": 2.3606448671722824e-06,
      "loss": 0.7339,
      "step": 12738
    },
    {
      "epoch": 0.5279539143768909,
      "grad_norm": 0.41023901104927063,
      "learning_rate": 2.360437647643914e-06,
      "loss": 0.667,
      "step": 12739
    },
    {
      "epoch": 0.5279953582825645,
      "grad_norm": 0.4159390330314636,
      "learning_rate": 2.360230428115546e-06,
      "loss": 0.6816,
      "step": 12740
    },
    {
      "epoch": 0.5280368021882382,
      "grad_norm": 0.41323333978652954,
      "learning_rate": 2.3600232085871774e-06,
      "loss": 0.7079,
      "step": 12741
    },
    {
      "epoch": 0.5280782460939119,
      "grad_norm": 0.4255991280078888,
      "learning_rate": 2.359815989058809e-06,
      "loss": 0.7102,
      "step": 12742
    },
    {
      "epoch": 0.5281196899995856,
      "grad_norm": 0.4253883957862854,
      "learning_rate": 2.3596087695304406e-06,
      "loss": 0.6848,
      "step": 12743
    },
    {
      "epoch": 0.5281611339052592,
      "grad_norm": 0.4293748438358307,
      "learning_rate": 2.3594015500020724e-06,
      "loss": 0.7014,
      "step": 12744
    },
    {
      "epoch": 0.5282025778109329,
      "grad_norm": 0.43246424198150635,
      "learning_rate": 2.359194330473704e-06,
      "loss": 0.6976,
      "step": 12745
    },
    {
      "epoch": 0.5282440217166066,
      "grad_norm": 0.41858580708503723,
      "learning_rate": 2.3589871109453356e-06,
      "loss": 0.6807,
      "step": 12746
    },
    {
      "epoch": 0.5282854656222803,
      "grad_norm": 0.406931072473526,
      "learning_rate": 2.3587798914169674e-06,
      "loss": 0.6676,
      "step": 12747
    },
    {
      "epoch": 0.5283269095279539,
      "grad_norm": 0.414328396320343,
      "learning_rate": 2.3585726718885992e-06,
      "loss": 0.6589,
      "step": 12748
    },
    {
      "epoch": 0.5283683534336275,
      "grad_norm": 0.4091304838657379,
      "learning_rate": 2.3583654523602306e-06,
      "loss": 0.6273,
      "step": 12749
    },
    {
      "epoch": 0.5284097973393013,
      "grad_norm": 0.44330504536628723,
      "learning_rate": 2.3581582328318624e-06,
      "loss": 0.6709,
      "step": 12750
    },
    {
      "epoch": 0.5284512412449749,
      "grad_norm": 0.3971948027610779,
      "learning_rate": 2.357951013303494e-06,
      "loss": 0.6653,
      "step": 12751
    },
    {
      "epoch": 0.5284926851506486,
      "grad_norm": 0.39212262630462646,
      "learning_rate": 2.3577437937751256e-06,
      "loss": 0.6772,
      "step": 12752
    },
    {
      "epoch": 0.5285341290563222,
      "grad_norm": 0.4163492023944855,
      "learning_rate": 2.357536574246757e-06,
      "loss": 0.6704,
      "step": 12753
    },
    {
      "epoch": 0.528575572961996,
      "grad_norm": 0.4218030869960785,
      "learning_rate": 2.357329354718389e-06,
      "loss": 0.7046,
      "step": 12754
    },
    {
      "epoch": 0.5286170168676696,
      "grad_norm": 0.38087084889411926,
      "learning_rate": 2.3571221351900206e-06,
      "loss": 0.6598,
      "step": 12755
    },
    {
      "epoch": 0.5286584607733433,
      "grad_norm": 0.4195363223552704,
      "learning_rate": 2.356914915661652e-06,
      "loss": 0.6904,
      "step": 12756
    },
    {
      "epoch": 0.528699904679017,
      "grad_norm": 0.4067707657814026,
      "learning_rate": 2.356707696133284e-06,
      "loss": 0.6478,
      "step": 12757
    },
    {
      "epoch": 0.5287413485846906,
      "grad_norm": 0.4406799077987671,
      "learning_rate": 2.3565004766049156e-06,
      "loss": 0.6699,
      "step": 12758
    },
    {
      "epoch": 0.5287827924903643,
      "grad_norm": 0.38076919317245483,
      "learning_rate": 2.356293257076547e-06,
      "loss": 0.681,
      "step": 12759
    },
    {
      "epoch": 0.5288242363960379,
      "grad_norm": 0.4239473342895508,
      "learning_rate": 2.356086037548179e-06,
      "loss": 0.7322,
      "step": 12760
    },
    {
      "epoch": 0.5288656803017117,
      "grad_norm": 0.40087074041366577,
      "learning_rate": 2.35587881801981e-06,
      "loss": 0.6646,
      "step": 12761
    },
    {
      "epoch": 0.5289071242073853,
      "grad_norm": 0.4074096381664276,
      "learning_rate": 2.355671598491442e-06,
      "loss": 0.6765,
      "step": 12762
    },
    {
      "epoch": 0.528948568113059,
      "grad_norm": 0.4054841697216034,
      "learning_rate": 2.355464378963074e-06,
      "loss": 0.7202,
      "step": 12763
    },
    {
      "epoch": 0.5289900120187326,
      "grad_norm": 0.4648331105709076,
      "learning_rate": 2.355257159434705e-06,
      "loss": 0.7073,
      "step": 12764
    },
    {
      "epoch": 0.5290314559244064,
      "grad_norm": 0.39438101649284363,
      "learning_rate": 2.355049939906337e-06,
      "loss": 0.7,
      "step": 12765
    },
    {
      "epoch": 0.52907289983008,
      "grad_norm": 0.3795521855354309,
      "learning_rate": 2.354842720377969e-06,
      "loss": 0.7054,
      "step": 12766
    },
    {
      "epoch": 0.5291143437357536,
      "grad_norm": 0.39583802223205566,
      "learning_rate": 2.3546355008496e-06,
      "loss": 0.6954,
      "step": 12767
    },
    {
      "epoch": 0.5291557876414273,
      "grad_norm": 0.43929022550582886,
      "learning_rate": 2.354428281321232e-06,
      "loss": 0.7345,
      "step": 12768
    },
    {
      "epoch": 0.529197231547101,
      "grad_norm": 0.41069528460502625,
      "learning_rate": 2.3542210617928634e-06,
      "loss": 0.6525,
      "step": 12769
    },
    {
      "epoch": 0.5292386754527747,
      "grad_norm": 0.4104939103126526,
      "learning_rate": 2.354013842264495e-06,
      "loss": 0.6709,
      "step": 12770
    },
    {
      "epoch": 0.5292801193584483,
      "grad_norm": 0.407794713973999,
      "learning_rate": 2.353806622736127e-06,
      "loss": 0.6614,
      "step": 12771
    },
    {
      "epoch": 0.529321563264122,
      "grad_norm": 0.4212910532951355,
      "learning_rate": 2.3535994032077584e-06,
      "loss": 0.6936,
      "step": 12772
    },
    {
      "epoch": 0.5293630071697957,
      "grad_norm": 0.39720138907432556,
      "learning_rate": 2.35339218367939e-06,
      "loss": 0.7473,
      "step": 12773
    },
    {
      "epoch": 0.5294044510754694,
      "grad_norm": 0.39428165555000305,
      "learning_rate": 2.353184964151022e-06,
      "loss": 0.6509,
      "step": 12774
    },
    {
      "epoch": 0.529445894981143,
      "grad_norm": 0.39720964431762695,
      "learning_rate": 2.3529777446226534e-06,
      "loss": 0.744,
      "step": 12775
    },
    {
      "epoch": 0.5294873388868166,
      "grad_norm": 0.42615777254104614,
      "learning_rate": 2.3527705250942852e-06,
      "loss": 0.7457,
      "step": 12776
    },
    {
      "epoch": 0.5295287827924904,
      "grad_norm": 0.4502609074115753,
      "learning_rate": 2.3525633055659166e-06,
      "loss": 0.6932,
      "step": 12777
    },
    {
      "epoch": 0.529570226698164,
      "grad_norm": 0.4430423974990845,
      "learning_rate": 2.3523560860375484e-06,
      "loss": 0.6794,
      "step": 12778
    },
    {
      "epoch": 0.5296116706038377,
      "grad_norm": 0.382313996553421,
      "learning_rate": 2.35214886650918e-06,
      "loss": 0.7288,
      "step": 12779
    },
    {
      "epoch": 0.5296531145095114,
      "grad_norm": 0.43854355812072754,
      "learning_rate": 2.3519416469808116e-06,
      "loss": 0.6622,
      "step": 12780
    },
    {
      "epoch": 0.5296945584151851,
      "grad_norm": 0.4560701847076416,
      "learning_rate": 2.3517344274524434e-06,
      "loss": 0.7444,
      "step": 12781
    },
    {
      "epoch": 0.5297360023208587,
      "grad_norm": 0.4166525602340698,
      "learning_rate": 2.3515272079240752e-06,
      "loss": 0.7034,
      "step": 12782
    },
    {
      "epoch": 0.5297774462265323,
      "grad_norm": 0.3945890963077545,
      "learning_rate": 2.3513199883957066e-06,
      "loss": 0.6558,
      "step": 12783
    },
    {
      "epoch": 0.5298188901322061,
      "grad_norm": 0.4028562307357788,
      "learning_rate": 2.3511127688673384e-06,
      "loss": 0.7168,
      "step": 12784
    },
    {
      "epoch": 0.5298603340378797,
      "grad_norm": 0.4088309109210968,
      "learning_rate": 2.35090554933897e-06,
      "loss": 0.6787,
      "step": 12785
    },
    {
      "epoch": 0.5299017779435534,
      "grad_norm": 0.394471675157547,
      "learning_rate": 2.3506983298106016e-06,
      "loss": 0.64,
      "step": 12786
    },
    {
      "epoch": 0.529943221849227,
      "grad_norm": 0.38554680347442627,
      "learning_rate": 2.350491110282233e-06,
      "loss": 0.6705,
      "step": 12787
    },
    {
      "epoch": 0.5299846657549008,
      "grad_norm": 0.3892744183540344,
      "learning_rate": 2.350283890753865e-06,
      "loss": 0.6633,
      "step": 12788
    },
    {
      "epoch": 0.5300261096605744,
      "grad_norm": 0.3607287108898163,
      "learning_rate": 2.3500766712254966e-06,
      "loss": 0.6156,
      "step": 12789
    },
    {
      "epoch": 0.5300675535662481,
      "grad_norm": 0.4060533940792084,
      "learning_rate": 2.3498694516971284e-06,
      "loss": 0.6641,
      "step": 12790
    },
    {
      "epoch": 0.5301089974719218,
      "grad_norm": 0.4194815456867218,
      "learning_rate": 2.34966223216876e-06,
      "loss": 0.7363,
      "step": 12791
    },
    {
      "epoch": 0.5301504413775954,
      "grad_norm": 0.40061917901039124,
      "learning_rate": 2.3494550126403916e-06,
      "loss": 0.7128,
      "step": 12792
    },
    {
      "epoch": 0.5301918852832691,
      "grad_norm": 0.441508948802948,
      "learning_rate": 2.349247793112023e-06,
      "loss": 0.6781,
      "step": 12793
    },
    {
      "epoch": 0.5302333291889427,
      "grad_norm": 0.39830511808395386,
      "learning_rate": 2.349040573583655e-06,
      "loss": 0.6753,
      "step": 12794
    },
    {
      "epoch": 0.5302747730946165,
      "grad_norm": 0.4329967498779297,
      "learning_rate": 2.348833354055286e-06,
      "loss": 0.7028,
      "step": 12795
    },
    {
      "epoch": 0.5303162170002901,
      "grad_norm": 0.40524542331695557,
      "learning_rate": 2.348626134526918e-06,
      "loss": 0.6747,
      "step": 12796
    },
    {
      "epoch": 0.5303576609059638,
      "grad_norm": 0.3959845304489136,
      "learning_rate": 2.34841891499855e-06,
      "loss": 0.7095,
      "step": 12797
    },
    {
      "epoch": 0.5303991048116374,
      "grad_norm": 0.420844703912735,
      "learning_rate": 2.348211695470181e-06,
      "loss": 0.6611,
      "step": 12798
    },
    {
      "epoch": 0.5304405487173112,
      "grad_norm": 0.4011492431163788,
      "learning_rate": 2.348004475941813e-06,
      "loss": 0.7131,
      "step": 12799
    },
    {
      "epoch": 0.5304819926229848,
      "grad_norm": 0.3901159465312958,
      "learning_rate": 2.347797256413445e-06,
      "loss": 0.684,
      "step": 12800
    },
    {
      "epoch": 0.5305234365286584,
      "grad_norm": 0.409708708524704,
      "learning_rate": 2.347590036885076e-06,
      "loss": 0.6779,
      "step": 12801
    },
    {
      "epoch": 0.5305648804343321,
      "grad_norm": 0.4444390535354614,
      "learning_rate": 2.347382817356708e-06,
      "loss": 0.7104,
      "step": 12802
    },
    {
      "epoch": 0.5306063243400058,
      "grad_norm": 0.40092501044273376,
      "learning_rate": 2.3471755978283394e-06,
      "loss": 0.6616,
      "step": 12803
    },
    {
      "epoch": 0.5306477682456795,
      "grad_norm": 0.4154167175292969,
      "learning_rate": 2.346968378299971e-06,
      "loss": 0.6578,
      "step": 12804
    },
    {
      "epoch": 0.5306892121513531,
      "grad_norm": 0.4358821213245392,
      "learning_rate": 2.3467611587716026e-06,
      "loss": 0.7517,
      "step": 12805
    },
    {
      "epoch": 0.5307306560570269,
      "grad_norm": 0.4447653591632843,
      "learning_rate": 2.3465539392432344e-06,
      "loss": 0.7336,
      "step": 12806
    },
    {
      "epoch": 0.5307720999627005,
      "grad_norm": 0.3950003385543823,
      "learning_rate": 2.3463467197148662e-06,
      "loss": 0.7269,
      "step": 12807
    },
    {
      "epoch": 0.5308135438683742,
      "grad_norm": 0.37115126848220825,
      "learning_rate": 2.346139500186498e-06,
      "loss": 0.6614,
      "step": 12808
    },
    {
      "epoch": 0.5308549877740478,
      "grad_norm": 0.4073227643966675,
      "learning_rate": 2.3459322806581294e-06,
      "loss": 0.7185,
      "step": 12809
    },
    {
      "epoch": 0.5308964316797214,
      "grad_norm": 0.3942599594593048,
      "learning_rate": 2.3457250611297612e-06,
      "loss": 0.6995,
      "step": 12810
    },
    {
      "epoch": 0.5309378755853952,
      "grad_norm": 0.43371403217315674,
      "learning_rate": 2.3455178416013926e-06,
      "loss": 0.7314,
      "step": 12811
    },
    {
      "epoch": 0.5309793194910688,
      "grad_norm": 0.40421175956726074,
      "learning_rate": 2.3453106220730244e-06,
      "loss": 0.6592,
      "step": 12812
    },
    {
      "epoch": 0.5310207633967425,
      "grad_norm": 0.3891434371471405,
      "learning_rate": 2.345103402544656e-06,
      "loss": 0.6693,
      "step": 12813
    },
    {
      "epoch": 0.5310622073024162,
      "grad_norm": 0.450286865234375,
      "learning_rate": 2.3448961830162876e-06,
      "loss": 0.662,
      "step": 12814
    },
    {
      "epoch": 0.5311036512080899,
      "grad_norm": 0.3891145884990692,
      "learning_rate": 2.3446889634879194e-06,
      "loss": 0.6337,
      "step": 12815
    },
    {
      "epoch": 0.5311450951137635,
      "grad_norm": 0.41346439719200134,
      "learning_rate": 2.3444817439595512e-06,
      "loss": 0.6487,
      "step": 12816
    },
    {
      "epoch": 0.5311865390194372,
      "grad_norm": 0.4364015758037567,
      "learning_rate": 2.3442745244311826e-06,
      "loss": 0.6799,
      "step": 12817
    },
    {
      "epoch": 0.5312279829251109,
      "grad_norm": 0.4026208221912384,
      "learning_rate": 2.3440673049028144e-06,
      "loss": 0.6807,
      "step": 12818
    },
    {
      "epoch": 0.5312694268307845,
      "grad_norm": 0.410280704498291,
      "learning_rate": 2.343860085374446e-06,
      "loss": 0.6699,
      "step": 12819
    },
    {
      "epoch": 0.5313108707364582,
      "grad_norm": 0.399504691362381,
      "learning_rate": 2.3436528658460776e-06,
      "loss": 0.6875,
      "step": 12820
    },
    {
      "epoch": 0.5313523146421318,
      "grad_norm": 0.44626593589782715,
      "learning_rate": 2.343445646317709e-06,
      "loss": 0.6777,
      "step": 12821
    },
    {
      "epoch": 0.5313937585478056,
      "grad_norm": 0.39921003580093384,
      "learning_rate": 2.343238426789341e-06,
      "loss": 0.6747,
      "step": 12822
    },
    {
      "epoch": 0.5314352024534792,
      "grad_norm": 0.4259141981601715,
      "learning_rate": 2.343031207260972e-06,
      "loss": 0.6655,
      "step": 12823
    },
    {
      "epoch": 0.5314766463591529,
      "grad_norm": 0.38230735063552856,
      "learning_rate": 2.3428239877326044e-06,
      "loss": 0.6328,
      "step": 12824
    },
    {
      "epoch": 0.5315180902648265,
      "grad_norm": 0.4410659670829773,
      "learning_rate": 2.342616768204236e-06,
      "loss": 0.6464,
      "step": 12825
    },
    {
      "epoch": 0.5315595341705003,
      "grad_norm": 0.4279730021953583,
      "learning_rate": 2.3424095486758676e-06,
      "loss": 0.635,
      "step": 12826
    },
    {
      "epoch": 0.5316009780761739,
      "grad_norm": 0.40151217579841614,
      "learning_rate": 2.342202329147499e-06,
      "loss": 0.7274,
      "step": 12827
    },
    {
      "epoch": 0.5316424219818475,
      "grad_norm": 0.44970500469207764,
      "learning_rate": 2.341995109619131e-06,
      "loss": 0.6853,
      "step": 12828
    },
    {
      "epoch": 0.5316838658875213,
      "grad_norm": 0.37925848364830017,
      "learning_rate": 2.341787890090762e-06,
      "loss": 0.7417,
      "step": 12829
    },
    {
      "epoch": 0.5317253097931949,
      "grad_norm": 0.4270873963832855,
      "learning_rate": 2.341580670562394e-06,
      "loss": 0.6987,
      "step": 12830
    },
    {
      "epoch": 0.5317667536988686,
      "grad_norm": 0.4409238398075104,
      "learning_rate": 2.3413734510340254e-06,
      "loss": 0.6285,
      "step": 12831
    },
    {
      "epoch": 0.5318081976045422,
      "grad_norm": 0.41487544775009155,
      "learning_rate": 2.341166231505657e-06,
      "loss": 0.7117,
      "step": 12832
    },
    {
      "epoch": 0.531849641510216,
      "grad_norm": 0.40720516443252563,
      "learning_rate": 2.340959011977289e-06,
      "loss": 0.6917,
      "step": 12833
    },
    {
      "epoch": 0.5318910854158896,
      "grad_norm": 0.44723257422447205,
      "learning_rate": 2.340751792448921e-06,
      "loss": 0.6538,
      "step": 12834
    },
    {
      "epoch": 0.5319325293215633,
      "grad_norm": 0.4195338189601898,
      "learning_rate": 2.3405445729205522e-06,
      "loss": 0.6726,
      "step": 12835
    },
    {
      "epoch": 0.5319739732272369,
      "grad_norm": 0.4372144043445587,
      "learning_rate": 2.340337353392184e-06,
      "loss": 0.6698,
      "step": 12836
    },
    {
      "epoch": 0.5320154171329106,
      "grad_norm": 0.380083292722702,
      "learning_rate": 2.3401301338638154e-06,
      "loss": 0.6514,
      "step": 12837
    },
    {
      "epoch": 0.5320568610385843,
      "grad_norm": 0.4389864206314087,
      "learning_rate": 2.3399229143354472e-06,
      "loss": 0.6982,
      "step": 12838
    },
    {
      "epoch": 0.5320983049442579,
      "grad_norm": 0.4643430709838867,
      "learning_rate": 2.3397156948070786e-06,
      "loss": 0.6929,
      "step": 12839
    },
    {
      "epoch": 0.5321397488499316,
      "grad_norm": 0.4307169020175934,
      "learning_rate": 2.3395084752787104e-06,
      "loss": 0.6705,
      "step": 12840
    },
    {
      "epoch": 0.5321811927556053,
      "grad_norm": 0.3644630014896393,
      "learning_rate": 2.3393012557503422e-06,
      "loss": 0.6692,
      "step": 12841
    },
    {
      "epoch": 0.532222636661279,
      "grad_norm": 0.39188143610954285,
      "learning_rate": 2.339094036221974e-06,
      "loss": 0.6528,
      "step": 12842
    },
    {
      "epoch": 0.5322640805669526,
      "grad_norm": 0.4374130666255951,
      "learning_rate": 2.3388868166936054e-06,
      "loss": 0.7328,
      "step": 12843
    },
    {
      "epoch": 0.5323055244726262,
      "grad_norm": 0.4222624599933624,
      "learning_rate": 2.3386795971652372e-06,
      "loss": 0.6958,
      "step": 12844
    },
    {
      "epoch": 0.5323469683783,
      "grad_norm": 0.40722373127937317,
      "learning_rate": 2.3384723776368686e-06,
      "loss": 0.7096,
      "step": 12845
    },
    {
      "epoch": 0.5323884122839736,
      "grad_norm": 0.43191152811050415,
      "learning_rate": 2.3382651581085004e-06,
      "loss": 0.7136,
      "step": 12846
    },
    {
      "epoch": 0.5324298561896473,
      "grad_norm": 0.41238442063331604,
      "learning_rate": 2.338057938580132e-06,
      "loss": 0.6904,
      "step": 12847
    },
    {
      "epoch": 0.532471300095321,
      "grad_norm": 0.39594894647598267,
      "learning_rate": 2.3378507190517636e-06,
      "loss": 0.6998,
      "step": 12848
    },
    {
      "epoch": 0.5325127440009947,
      "grad_norm": 0.3915773630142212,
      "learning_rate": 2.337643499523395e-06,
      "loss": 0.6641,
      "step": 12849
    },
    {
      "epoch": 0.5325541879066683,
      "grad_norm": 0.4006320536136627,
      "learning_rate": 2.3374362799950272e-06,
      "loss": 0.6722,
      "step": 12850
    },
    {
      "epoch": 0.532595631812342,
      "grad_norm": 0.4360206723213196,
      "learning_rate": 2.3372290604666586e-06,
      "loss": 0.699,
      "step": 12851
    },
    {
      "epoch": 0.5326370757180157,
      "grad_norm": 0.4406997263431549,
      "learning_rate": 2.3370218409382904e-06,
      "loss": 0.683,
      "step": 12852
    },
    {
      "epoch": 0.5326785196236893,
      "grad_norm": 0.41712766885757446,
      "learning_rate": 2.336814621409922e-06,
      "loss": 0.6932,
      "step": 12853
    },
    {
      "epoch": 0.532719963529363,
      "grad_norm": 0.4048572778701782,
      "learning_rate": 2.3366074018815536e-06,
      "loss": 0.7043,
      "step": 12854
    },
    {
      "epoch": 0.5327614074350366,
      "grad_norm": 0.41389012336730957,
      "learning_rate": 2.336400182353185e-06,
      "loss": 0.6814,
      "step": 12855
    },
    {
      "epoch": 0.5328028513407104,
      "grad_norm": 0.4192982017993927,
      "learning_rate": 2.336192962824817e-06,
      "loss": 0.6752,
      "step": 12856
    },
    {
      "epoch": 0.532844295246384,
      "grad_norm": 0.42561307549476624,
      "learning_rate": 2.335985743296448e-06,
      "loss": 0.7051,
      "step": 12857
    },
    {
      "epoch": 0.5328857391520577,
      "grad_norm": 0.3976946473121643,
      "learning_rate": 2.3357785237680804e-06,
      "loss": 0.6765,
      "step": 12858
    },
    {
      "epoch": 0.5329271830577313,
      "grad_norm": 0.4183340072631836,
      "learning_rate": 2.335571304239712e-06,
      "loss": 0.6635,
      "step": 12859
    },
    {
      "epoch": 0.5329686269634051,
      "grad_norm": 0.4339468479156494,
      "learning_rate": 2.3353640847113436e-06,
      "loss": 0.6851,
      "step": 12860
    },
    {
      "epoch": 0.5330100708690787,
      "grad_norm": 0.4078211784362793,
      "learning_rate": 2.335156865182975e-06,
      "loss": 0.6407,
      "step": 12861
    },
    {
      "epoch": 0.5330515147747523,
      "grad_norm": 0.4416934549808502,
      "learning_rate": 2.334949645654607e-06,
      "loss": 0.7,
      "step": 12862
    },
    {
      "epoch": 0.533092958680426,
      "grad_norm": 0.43863627314567566,
      "learning_rate": 2.334742426126238e-06,
      "loss": 0.6904,
      "step": 12863
    },
    {
      "epoch": 0.5331344025860997,
      "grad_norm": 0.3993176817893982,
      "learning_rate": 2.33453520659787e-06,
      "loss": 0.6235,
      "step": 12864
    },
    {
      "epoch": 0.5331758464917734,
      "grad_norm": 0.3826536536216736,
      "learning_rate": 2.3343279870695014e-06,
      "loss": 0.6876,
      "step": 12865
    },
    {
      "epoch": 0.533217290397447,
      "grad_norm": 0.3884011507034302,
      "learning_rate": 2.3341207675411332e-06,
      "loss": 0.652,
      "step": 12866
    },
    {
      "epoch": 0.5332587343031208,
      "grad_norm": 0.4264260530471802,
      "learning_rate": 2.333913548012765e-06,
      "loss": 0.7207,
      "step": 12867
    },
    {
      "epoch": 0.5333001782087944,
      "grad_norm": 0.4118197560310364,
      "learning_rate": 2.333706328484397e-06,
      "loss": 0.6823,
      "step": 12868
    },
    {
      "epoch": 0.5333416221144681,
      "grad_norm": 0.3943627178668976,
      "learning_rate": 2.3334991089560282e-06,
      "loss": 0.7007,
      "step": 12869
    },
    {
      "epoch": 0.5333830660201417,
      "grad_norm": 0.42532142996788025,
      "learning_rate": 2.33329188942766e-06,
      "loss": 0.7336,
      "step": 12870
    },
    {
      "epoch": 0.5334245099258154,
      "grad_norm": 0.4259392023086548,
      "learning_rate": 2.3330846698992914e-06,
      "loss": 0.679,
      "step": 12871
    },
    {
      "epoch": 0.5334659538314891,
      "grad_norm": 0.43316420912742615,
      "learning_rate": 2.3328774503709232e-06,
      "loss": 0.6692,
      "step": 12872
    },
    {
      "epoch": 0.5335073977371627,
      "grad_norm": 0.39270442724227905,
      "learning_rate": 2.3326702308425546e-06,
      "loss": 0.6836,
      "step": 12873
    },
    {
      "epoch": 0.5335488416428364,
      "grad_norm": 0.39156314730644226,
      "learning_rate": 2.3324630113141864e-06,
      "loss": 0.7212,
      "step": 12874
    },
    {
      "epoch": 0.5335902855485101,
      "grad_norm": 0.4253687858581543,
      "learning_rate": 2.332255791785818e-06,
      "loss": 0.6375,
      "step": 12875
    },
    {
      "epoch": 0.5336317294541838,
      "grad_norm": 0.38181617856025696,
      "learning_rate": 2.33204857225745e-06,
      "loss": 0.6726,
      "step": 12876
    },
    {
      "epoch": 0.5336731733598574,
      "grad_norm": 0.4157556891441345,
      "learning_rate": 2.3318413527290814e-06,
      "loss": 0.7169,
      "step": 12877
    },
    {
      "epoch": 0.5337146172655312,
      "grad_norm": 0.41273123025894165,
      "learning_rate": 2.3316341332007132e-06,
      "loss": 0.6705,
      "step": 12878
    },
    {
      "epoch": 0.5337560611712048,
      "grad_norm": 0.42082834243774414,
      "learning_rate": 2.3314269136723446e-06,
      "loss": 0.6857,
      "step": 12879
    },
    {
      "epoch": 0.5337975050768784,
      "grad_norm": 0.41242021322250366,
      "learning_rate": 2.3312196941439764e-06,
      "loss": 0.6882,
      "step": 12880
    },
    {
      "epoch": 0.5338389489825521,
      "grad_norm": 0.3927420973777771,
      "learning_rate": 2.331012474615608e-06,
      "loss": 0.6721,
      "step": 12881
    },
    {
      "epoch": 0.5338803928882258,
      "grad_norm": 0.4605309069156647,
      "learning_rate": 2.3308052550872396e-06,
      "loss": 0.7207,
      "step": 12882
    },
    {
      "epoch": 0.5339218367938995,
      "grad_norm": 0.45265206694602966,
      "learning_rate": 2.330598035558871e-06,
      "loss": 0.6671,
      "step": 12883
    },
    {
      "epoch": 0.5339632806995731,
      "grad_norm": 0.42870867252349854,
      "learning_rate": 2.330390816030503e-06,
      "loss": 0.7317,
      "step": 12884
    },
    {
      "epoch": 0.5340047246052468,
      "grad_norm": 0.410745769739151,
      "learning_rate": 2.3301835965021346e-06,
      "loss": 0.7073,
      "step": 12885
    },
    {
      "epoch": 0.5340461685109205,
      "grad_norm": 0.3888387382030487,
      "learning_rate": 2.3299763769737664e-06,
      "loss": 0.6732,
      "step": 12886
    },
    {
      "epoch": 0.5340876124165942,
      "grad_norm": 0.390068382024765,
      "learning_rate": 2.329769157445398e-06,
      "loss": 0.6716,
      "step": 12887
    },
    {
      "epoch": 0.5341290563222678,
      "grad_norm": 0.39509692788124084,
      "learning_rate": 2.3295619379170296e-06,
      "loss": 0.7114,
      "step": 12888
    },
    {
      "epoch": 0.5341705002279414,
      "grad_norm": 0.4518096148967743,
      "learning_rate": 2.329354718388661e-06,
      "loss": 0.7043,
      "step": 12889
    },
    {
      "epoch": 0.5342119441336152,
      "grad_norm": 0.4201490879058838,
      "learning_rate": 2.329147498860293e-06,
      "loss": 0.6642,
      "step": 12890
    },
    {
      "epoch": 0.5342533880392888,
      "grad_norm": 0.39589551091194153,
      "learning_rate": 2.328940279331924e-06,
      "loss": 0.6792,
      "step": 12891
    },
    {
      "epoch": 0.5342948319449625,
      "grad_norm": 0.38695114850997925,
      "learning_rate": 2.328733059803556e-06,
      "loss": 0.7246,
      "step": 12892
    },
    {
      "epoch": 0.5343362758506361,
      "grad_norm": 0.4321115016937256,
      "learning_rate": 2.328525840275188e-06,
      "loss": 0.6835,
      "step": 12893
    },
    {
      "epoch": 0.5343777197563099,
      "grad_norm": 0.4100632071495056,
      "learning_rate": 2.3283186207468196e-06,
      "loss": 0.6437,
      "step": 12894
    },
    {
      "epoch": 0.5344191636619835,
      "grad_norm": 0.4007258117198944,
      "learning_rate": 2.328111401218451e-06,
      "loss": 0.6823,
      "step": 12895
    },
    {
      "epoch": 0.5344606075676572,
      "grad_norm": 0.4309125542640686,
      "learning_rate": 2.327904181690083e-06,
      "loss": 0.6941,
      "step": 12896
    },
    {
      "epoch": 0.5345020514733309,
      "grad_norm": 0.3988111913204193,
      "learning_rate": 2.3276969621617142e-06,
      "loss": 0.6477,
      "step": 12897
    },
    {
      "epoch": 0.5345434953790045,
      "grad_norm": 0.3994053602218628,
      "learning_rate": 2.327489742633346e-06,
      "loss": 0.6633,
      "step": 12898
    },
    {
      "epoch": 0.5345849392846782,
      "grad_norm": 0.4092133045196533,
      "learning_rate": 2.3272825231049774e-06,
      "loss": 0.6558,
      "step": 12899
    },
    {
      "epoch": 0.5346263831903518,
      "grad_norm": 0.3984118103981018,
      "learning_rate": 2.3270753035766092e-06,
      "loss": 0.6448,
      "step": 12900
    },
    {
      "epoch": 0.5346678270960256,
      "grad_norm": 0.4333925247192383,
      "learning_rate": 2.3268680840482406e-06,
      "loss": 0.6746,
      "step": 12901
    },
    {
      "epoch": 0.5347092710016992,
      "grad_norm": 0.39333856105804443,
      "learning_rate": 2.326660864519873e-06,
      "loss": 0.6797,
      "step": 12902
    },
    {
      "epoch": 0.5347507149073729,
      "grad_norm": 0.42644229531288147,
      "learning_rate": 2.3264536449915042e-06,
      "loss": 0.7175,
      "step": 12903
    },
    {
      "epoch": 0.5347921588130465,
      "grad_norm": 0.45764684677124023,
      "learning_rate": 2.326246425463136e-06,
      "loss": 0.6582,
      "step": 12904
    },
    {
      "epoch": 0.5348336027187202,
      "grad_norm": 0.3673115074634552,
      "learning_rate": 2.3260392059347674e-06,
      "loss": 0.6013,
      "step": 12905
    },
    {
      "epoch": 0.5348750466243939,
      "grad_norm": 0.4003862738609314,
      "learning_rate": 2.3258319864063992e-06,
      "loss": 0.6985,
      "step": 12906
    },
    {
      "epoch": 0.5349164905300675,
      "grad_norm": 0.3869233727455139,
      "learning_rate": 2.3256247668780306e-06,
      "loss": 0.6875,
      "step": 12907
    },
    {
      "epoch": 0.5349579344357412,
      "grad_norm": 0.4463527798652649,
      "learning_rate": 2.3254175473496624e-06,
      "loss": 0.7288,
      "step": 12908
    },
    {
      "epoch": 0.5349993783414149,
      "grad_norm": 0.40733829140663147,
      "learning_rate": 2.325210327821294e-06,
      "loss": 0.6958,
      "step": 12909
    },
    {
      "epoch": 0.5350408222470886,
      "grad_norm": 0.4178887903690338,
      "learning_rate": 2.3250031082929256e-06,
      "loss": 0.7068,
      "step": 12910
    },
    {
      "epoch": 0.5350822661527622,
      "grad_norm": 0.43788233399391174,
      "learning_rate": 2.3247958887645574e-06,
      "loss": 0.7014,
      "step": 12911
    },
    {
      "epoch": 0.535123710058436,
      "grad_norm": 0.3919357359409332,
      "learning_rate": 2.3245886692361892e-06,
      "loss": 0.6715,
      "step": 12912
    },
    {
      "epoch": 0.5351651539641096,
      "grad_norm": 0.40420424938201904,
      "learning_rate": 2.3243814497078206e-06,
      "loss": 0.7185,
      "step": 12913
    },
    {
      "epoch": 0.5352065978697832,
      "grad_norm": 0.41175714135169983,
      "learning_rate": 2.3241742301794524e-06,
      "loss": 0.6589,
      "step": 12914
    },
    {
      "epoch": 0.5352480417754569,
      "grad_norm": 0.4010791778564453,
      "learning_rate": 2.323967010651084e-06,
      "loss": 0.6362,
      "step": 12915
    },
    {
      "epoch": 0.5352894856811306,
      "grad_norm": 0.3885844945907593,
      "learning_rate": 2.3237597911227156e-06,
      "loss": 0.688,
      "step": 12916
    },
    {
      "epoch": 0.5353309295868043,
      "grad_norm": 0.44161322712898254,
      "learning_rate": 2.323552571594347e-06,
      "loss": 0.7446,
      "step": 12917
    },
    {
      "epoch": 0.5353723734924779,
      "grad_norm": 0.4468418061733246,
      "learning_rate": 2.323345352065979e-06,
      "loss": 0.6985,
      "step": 12918
    },
    {
      "epoch": 0.5354138173981516,
      "grad_norm": 0.4260421097278595,
      "learning_rate": 2.3231381325376106e-06,
      "loss": 0.656,
      "step": 12919
    },
    {
      "epoch": 0.5354552613038253,
      "grad_norm": 0.36235854029655457,
      "learning_rate": 2.3229309130092424e-06,
      "loss": 0.6028,
      "step": 12920
    },
    {
      "epoch": 0.535496705209499,
      "grad_norm": 0.4143977165222168,
      "learning_rate": 2.322723693480874e-06,
      "loss": 0.7085,
      "step": 12921
    },
    {
      "epoch": 0.5355381491151726,
      "grad_norm": 0.40639543533325195,
      "learning_rate": 2.3225164739525056e-06,
      "loss": 0.688,
      "step": 12922
    },
    {
      "epoch": 0.5355795930208462,
      "grad_norm": 0.40311139822006226,
      "learning_rate": 2.322309254424137e-06,
      "loss": 0.6824,
      "step": 12923
    },
    {
      "epoch": 0.53562103692652,
      "grad_norm": 0.43853384256362915,
      "learning_rate": 2.322102034895769e-06,
      "loss": 0.7562,
      "step": 12924
    },
    {
      "epoch": 0.5356624808321936,
      "grad_norm": 0.4531860053539276,
      "learning_rate": 2.3218948153674002e-06,
      "loss": 0.7606,
      "step": 12925
    },
    {
      "epoch": 0.5357039247378673,
      "grad_norm": 0.40941768884658813,
      "learning_rate": 2.321687595839032e-06,
      "loss": 0.7014,
      "step": 12926
    },
    {
      "epoch": 0.5357453686435409,
      "grad_norm": 0.40554338693618774,
      "learning_rate": 2.3214803763106634e-06,
      "loss": 0.7173,
      "step": 12927
    },
    {
      "epoch": 0.5357868125492147,
      "grad_norm": 0.40341389179229736,
      "learning_rate": 2.3212731567822956e-06,
      "loss": 0.6444,
      "step": 12928
    },
    {
      "epoch": 0.5358282564548883,
      "grad_norm": 0.44897788763046265,
      "learning_rate": 2.321065937253927e-06,
      "loss": 0.6959,
      "step": 12929
    },
    {
      "epoch": 0.535869700360562,
      "grad_norm": 0.420905739068985,
      "learning_rate": 2.320858717725559e-06,
      "loss": 0.6906,
      "step": 12930
    },
    {
      "epoch": 0.5359111442662357,
      "grad_norm": 0.4090760350227356,
      "learning_rate": 2.3206514981971902e-06,
      "loss": 0.6523,
      "step": 12931
    },
    {
      "epoch": 0.5359525881719093,
      "grad_norm": 0.40057849884033203,
      "learning_rate": 2.320444278668822e-06,
      "loss": 0.6506,
      "step": 12932
    },
    {
      "epoch": 0.535994032077583,
      "grad_norm": 0.4215586185455322,
      "learning_rate": 2.3202370591404534e-06,
      "loss": 0.6974,
      "step": 12933
    },
    {
      "epoch": 0.5360354759832566,
      "grad_norm": 0.400571346282959,
      "learning_rate": 2.3200298396120852e-06,
      "loss": 0.6593,
      "step": 12934
    },
    {
      "epoch": 0.5360769198889304,
      "grad_norm": 0.36831507086753845,
      "learning_rate": 2.3198226200837166e-06,
      "loss": 0.5818,
      "step": 12935
    },
    {
      "epoch": 0.536118363794604,
      "grad_norm": 0.4073820114135742,
      "learning_rate": 2.3196154005553484e-06,
      "loss": 0.6714,
      "step": 12936
    },
    {
      "epoch": 0.5361598077002777,
      "grad_norm": 0.3813076615333557,
      "learning_rate": 2.3194081810269802e-06,
      "loss": 0.645,
      "step": 12937
    },
    {
      "epoch": 0.5362012516059513,
      "grad_norm": 0.39923980832099915,
      "learning_rate": 2.319200961498612e-06,
      "loss": 0.646,
      "step": 12938
    },
    {
      "epoch": 0.5362426955116251,
      "grad_norm": 0.42261260747909546,
      "learning_rate": 2.3189937419702434e-06,
      "loss": 0.6794,
      "step": 12939
    },
    {
      "epoch": 0.5362841394172987,
      "grad_norm": 0.4347153604030609,
      "learning_rate": 2.3187865224418752e-06,
      "loss": 0.7045,
      "step": 12940
    },
    {
      "epoch": 0.5363255833229723,
      "grad_norm": 0.43233221769332886,
      "learning_rate": 2.3185793029135066e-06,
      "loss": 0.6848,
      "step": 12941
    },
    {
      "epoch": 0.536367027228646,
      "grad_norm": 0.4340842664241791,
      "learning_rate": 2.3183720833851384e-06,
      "loss": 0.7057,
      "step": 12942
    },
    {
      "epoch": 0.5364084711343197,
      "grad_norm": 0.3948730528354645,
      "learning_rate": 2.31816486385677e-06,
      "loss": 0.5989,
      "step": 12943
    },
    {
      "epoch": 0.5364499150399934,
      "grad_norm": 0.42701467871665955,
      "learning_rate": 2.3179576443284016e-06,
      "loss": 0.7151,
      "step": 12944
    },
    {
      "epoch": 0.536491358945667,
      "grad_norm": 0.42283880710601807,
      "learning_rate": 2.3177504248000334e-06,
      "loss": 0.6971,
      "step": 12945
    },
    {
      "epoch": 0.5365328028513408,
      "grad_norm": 0.44455471634864807,
      "learning_rate": 2.3175432052716652e-06,
      "loss": 0.7498,
      "step": 12946
    },
    {
      "epoch": 0.5365742467570144,
      "grad_norm": 0.42121943831443787,
      "learning_rate": 2.3173359857432966e-06,
      "loss": 0.6836,
      "step": 12947
    },
    {
      "epoch": 0.5366156906626881,
      "grad_norm": 0.4126746952533722,
      "learning_rate": 2.3171287662149284e-06,
      "loss": 0.6343,
      "step": 12948
    },
    {
      "epoch": 0.5366571345683617,
      "grad_norm": 0.43476393818855286,
      "learning_rate": 2.31692154668656e-06,
      "loss": 0.6781,
      "step": 12949
    },
    {
      "epoch": 0.5366985784740353,
      "grad_norm": 0.3706677556037903,
      "learning_rate": 2.3167143271581916e-06,
      "loss": 0.6541,
      "step": 12950
    },
    {
      "epoch": 0.5367400223797091,
      "grad_norm": 0.43021148443222046,
      "learning_rate": 2.316507107629823e-06,
      "loss": 0.7,
      "step": 12951
    },
    {
      "epoch": 0.5367814662853827,
      "grad_norm": 0.4301229417324066,
      "learning_rate": 2.316299888101455e-06,
      "loss": 0.7422,
      "step": 12952
    },
    {
      "epoch": 0.5368229101910564,
      "grad_norm": 0.3899882137775421,
      "learning_rate": 2.3160926685730862e-06,
      "loss": 0.636,
      "step": 12953
    },
    {
      "epoch": 0.5368643540967301,
      "grad_norm": 0.3972240388393402,
      "learning_rate": 2.3158854490447184e-06,
      "loss": 0.6757,
      "step": 12954
    },
    {
      "epoch": 0.5369057980024038,
      "grad_norm": 0.39857980608940125,
      "learning_rate": 2.31567822951635e-06,
      "loss": 0.6572,
      "step": 12955
    },
    {
      "epoch": 0.5369472419080774,
      "grad_norm": 0.42433953285217285,
      "learning_rate": 2.3154710099879816e-06,
      "loss": 0.6927,
      "step": 12956
    },
    {
      "epoch": 0.5369886858137511,
      "grad_norm": 0.4580519497394562,
      "learning_rate": 2.315263790459613e-06,
      "loss": 0.7014,
      "step": 12957
    },
    {
      "epoch": 0.5370301297194248,
      "grad_norm": 0.38678693771362305,
      "learning_rate": 2.315056570931245e-06,
      "loss": 0.6218,
      "step": 12958
    },
    {
      "epoch": 0.5370715736250984,
      "grad_norm": 0.45386382937431335,
      "learning_rate": 2.3148493514028762e-06,
      "loss": 0.7103,
      "step": 12959
    },
    {
      "epoch": 0.5371130175307721,
      "grad_norm": 0.3997993767261505,
      "learning_rate": 2.314642131874508e-06,
      "loss": 0.6548,
      "step": 12960
    },
    {
      "epoch": 0.5371544614364457,
      "grad_norm": 0.40142855048179626,
      "learning_rate": 2.3144349123461394e-06,
      "loss": 0.6785,
      "step": 12961
    },
    {
      "epoch": 0.5371959053421195,
      "grad_norm": 0.4283410310745239,
      "learning_rate": 2.3142276928177712e-06,
      "loss": 0.6876,
      "step": 12962
    },
    {
      "epoch": 0.5372373492477931,
      "grad_norm": 0.3978479504585266,
      "learning_rate": 2.314020473289403e-06,
      "loss": 0.6908,
      "step": 12963
    },
    {
      "epoch": 0.5372787931534668,
      "grad_norm": 0.40223902463912964,
      "learning_rate": 2.313813253761035e-06,
      "loss": 0.6826,
      "step": 12964
    },
    {
      "epoch": 0.5373202370591404,
      "grad_norm": 0.41629084944725037,
      "learning_rate": 2.3136060342326662e-06,
      "loss": 0.7134,
      "step": 12965
    },
    {
      "epoch": 0.5373616809648141,
      "grad_norm": 0.4367891252040863,
      "learning_rate": 2.313398814704298e-06,
      "loss": 0.7087,
      "step": 12966
    },
    {
      "epoch": 0.5374031248704878,
      "grad_norm": 0.39007118344306946,
      "learning_rate": 2.3131915951759294e-06,
      "loss": 0.6567,
      "step": 12967
    },
    {
      "epoch": 0.5374445687761614,
      "grad_norm": 0.3793109357357025,
      "learning_rate": 2.3129843756475612e-06,
      "loss": 0.6497,
      "step": 12968
    },
    {
      "epoch": 0.5374860126818352,
      "grad_norm": 0.39812222123146057,
      "learning_rate": 2.3127771561191926e-06,
      "loss": 0.6345,
      "step": 12969
    },
    {
      "epoch": 0.5375274565875088,
      "grad_norm": 0.4116498827934265,
      "learning_rate": 2.3125699365908244e-06,
      "loss": 0.67,
      "step": 12970
    },
    {
      "epoch": 0.5375689004931825,
      "grad_norm": 0.45759615302085876,
      "learning_rate": 2.312362717062456e-06,
      "loss": 0.7313,
      "step": 12971
    },
    {
      "epoch": 0.5376103443988561,
      "grad_norm": 0.42657917737960815,
      "learning_rate": 2.312155497534088e-06,
      "loss": 0.666,
      "step": 12972
    },
    {
      "epoch": 0.5376517883045299,
      "grad_norm": 0.3863115906715393,
      "learning_rate": 2.3119482780057194e-06,
      "loss": 0.6904,
      "step": 12973
    },
    {
      "epoch": 0.5376932322102035,
      "grad_norm": 0.41953253746032715,
      "learning_rate": 2.3117410584773512e-06,
      "loss": 0.7051,
      "step": 12974
    },
    {
      "epoch": 0.5377346761158771,
      "grad_norm": 0.41634392738342285,
      "learning_rate": 2.3115338389489826e-06,
      "loss": 0.6764,
      "step": 12975
    },
    {
      "epoch": 0.5377761200215508,
      "grad_norm": 0.4194655120372772,
      "learning_rate": 2.3113266194206144e-06,
      "loss": 0.6388,
      "step": 12976
    },
    {
      "epoch": 0.5378175639272245,
      "grad_norm": 0.44342562556266785,
      "learning_rate": 2.311119399892246e-06,
      "loss": 0.6992,
      "step": 12977
    },
    {
      "epoch": 0.5378590078328982,
      "grad_norm": 0.3915787637233734,
      "learning_rate": 2.3109121803638776e-06,
      "loss": 0.66,
      "step": 12978
    },
    {
      "epoch": 0.5379004517385718,
      "grad_norm": 0.4389720857143402,
      "learning_rate": 2.310704960835509e-06,
      "loss": 0.7151,
      "step": 12979
    },
    {
      "epoch": 0.5379418956442455,
      "grad_norm": 0.4177413582801819,
      "learning_rate": 2.3104977413071412e-06,
      "loss": 0.71,
      "step": 12980
    },
    {
      "epoch": 0.5379833395499192,
      "grad_norm": 0.3874242603778839,
      "learning_rate": 2.3102905217787726e-06,
      "loss": 0.6283,
      "step": 12981
    },
    {
      "epoch": 0.5380247834555929,
      "grad_norm": 0.42663660645484924,
      "learning_rate": 2.3100833022504044e-06,
      "loss": 0.7231,
      "step": 12982
    },
    {
      "epoch": 0.5380662273612665,
      "grad_norm": 0.40798890590667725,
      "learning_rate": 2.309876082722036e-06,
      "loss": 0.6487,
      "step": 12983
    },
    {
      "epoch": 0.5381076712669401,
      "grad_norm": 0.4156686067581177,
      "learning_rate": 2.3096688631936676e-06,
      "loss": 0.6783,
      "step": 12984
    },
    {
      "epoch": 0.5381491151726139,
      "grad_norm": 0.42959898710250854,
      "learning_rate": 2.309461643665299e-06,
      "loss": 0.7039,
      "step": 12985
    },
    {
      "epoch": 0.5381905590782875,
      "grad_norm": 0.3906037211418152,
      "learning_rate": 2.309254424136931e-06,
      "loss": 0.636,
      "step": 12986
    },
    {
      "epoch": 0.5382320029839612,
      "grad_norm": 0.4188729226589203,
      "learning_rate": 2.3090472046085622e-06,
      "loss": 0.6715,
      "step": 12987
    },
    {
      "epoch": 0.5382734468896349,
      "grad_norm": 0.4583487808704376,
      "learning_rate": 2.308839985080194e-06,
      "loss": 0.7466,
      "step": 12988
    },
    {
      "epoch": 0.5383148907953086,
      "grad_norm": 0.3973032236099243,
      "learning_rate": 2.308632765551826e-06,
      "loss": 0.6965,
      "step": 12989
    },
    {
      "epoch": 0.5383563347009822,
      "grad_norm": 0.3715957999229431,
      "learning_rate": 2.3084255460234576e-06,
      "loss": 0.6655,
      "step": 12990
    },
    {
      "epoch": 0.5383977786066559,
      "grad_norm": 0.4440552294254303,
      "learning_rate": 2.308218326495089e-06,
      "loss": 0.6726,
      "step": 12991
    },
    {
      "epoch": 0.5384392225123296,
      "grad_norm": 0.3799477517604828,
      "learning_rate": 2.308011106966721e-06,
      "loss": 0.6692,
      "step": 12992
    },
    {
      "epoch": 0.5384806664180032,
      "grad_norm": 0.4284555912017822,
      "learning_rate": 2.3078038874383522e-06,
      "loss": 0.7047,
      "step": 12993
    },
    {
      "epoch": 0.5385221103236769,
      "grad_norm": 0.41752544045448303,
      "learning_rate": 2.307596667909984e-06,
      "loss": 0.6764,
      "step": 12994
    },
    {
      "epoch": 0.5385635542293505,
      "grad_norm": 0.4005182981491089,
      "learning_rate": 2.3073894483816154e-06,
      "loss": 0.6997,
      "step": 12995
    },
    {
      "epoch": 0.5386049981350243,
      "grad_norm": 0.42117759585380554,
      "learning_rate": 2.3071822288532472e-06,
      "loss": 0.6832,
      "step": 12996
    },
    {
      "epoch": 0.5386464420406979,
      "grad_norm": 0.4225638210773468,
      "learning_rate": 2.306975009324879e-06,
      "loss": 0.6997,
      "step": 12997
    },
    {
      "epoch": 0.5386878859463716,
      "grad_norm": 0.4211042821407318,
      "learning_rate": 2.306767789796511e-06,
      "loss": 0.7388,
      "step": 12998
    },
    {
      "epoch": 0.5387293298520452,
      "grad_norm": 0.43044334650039673,
      "learning_rate": 2.3065605702681422e-06,
      "loss": 0.6791,
      "step": 12999
    },
    {
      "epoch": 0.538770773757719,
      "grad_norm": 0.3610936105251312,
      "learning_rate": 2.306353350739774e-06,
      "loss": 0.6117,
      "step": 13000
    },
    {
      "epoch": 0.5388122176633926,
      "grad_norm": 0.41754060983657837,
      "learning_rate": 2.3061461312114054e-06,
      "loss": 0.6658,
      "step": 13001
    },
    {
      "epoch": 0.5388536615690662,
      "grad_norm": 0.4171181917190552,
      "learning_rate": 2.3059389116830372e-06,
      "loss": 0.6783,
      "step": 13002
    },
    {
      "epoch": 0.53889510547474,
      "grad_norm": 0.43441829085350037,
      "learning_rate": 2.3057316921546686e-06,
      "loss": 0.6483,
      "step": 13003
    },
    {
      "epoch": 0.5389365493804136,
      "grad_norm": 0.4103885591030121,
      "learning_rate": 2.3055244726263004e-06,
      "loss": 0.7065,
      "step": 13004
    },
    {
      "epoch": 0.5389779932860873,
      "grad_norm": 0.4269576370716095,
      "learning_rate": 2.305317253097932e-06,
      "loss": 0.6709,
      "step": 13005
    },
    {
      "epoch": 0.5390194371917609,
      "grad_norm": 0.4174657166004181,
      "learning_rate": 2.305110033569564e-06,
      "loss": 0.6273,
      "step": 13006
    },
    {
      "epoch": 0.5390608810974347,
      "grad_norm": 0.43940219283103943,
      "learning_rate": 2.3049028140411954e-06,
      "loss": 0.717,
      "step": 13007
    },
    {
      "epoch": 0.5391023250031083,
      "grad_norm": 0.4273514747619629,
      "learning_rate": 2.3046955945128272e-06,
      "loss": 0.7415,
      "step": 13008
    },
    {
      "epoch": 0.539143768908782,
      "grad_norm": 0.42901578545570374,
      "learning_rate": 2.3044883749844586e-06,
      "loss": 0.6608,
      "step": 13009
    },
    {
      "epoch": 0.5391852128144556,
      "grad_norm": 0.4408095180988312,
      "learning_rate": 2.3042811554560904e-06,
      "loss": 0.6877,
      "step": 13010
    },
    {
      "epoch": 0.5392266567201293,
      "grad_norm": 0.4260903596878052,
      "learning_rate": 2.304073935927722e-06,
      "loss": 0.6853,
      "step": 13011
    },
    {
      "epoch": 0.539268100625803,
      "grad_norm": 0.4214371144771576,
      "learning_rate": 2.3038667163993536e-06,
      "loss": 0.7219,
      "step": 13012
    },
    {
      "epoch": 0.5393095445314766,
      "grad_norm": 0.3758764863014221,
      "learning_rate": 2.303659496870985e-06,
      "loss": 0.6512,
      "step": 13013
    },
    {
      "epoch": 0.5393509884371503,
      "grad_norm": 0.4344061315059662,
      "learning_rate": 2.303452277342617e-06,
      "loss": 0.7063,
      "step": 13014
    },
    {
      "epoch": 0.539392432342824,
      "grad_norm": 0.4331554174423218,
      "learning_rate": 2.3032450578142486e-06,
      "loss": 0.7026,
      "step": 13015
    },
    {
      "epoch": 0.5394338762484977,
      "grad_norm": 0.4216785430908203,
      "learning_rate": 2.3030378382858804e-06,
      "loss": 0.6227,
      "step": 13016
    },
    {
      "epoch": 0.5394753201541713,
      "grad_norm": 0.4197157919406891,
      "learning_rate": 2.302830618757512e-06,
      "loss": 0.666,
      "step": 13017
    },
    {
      "epoch": 0.5395167640598449,
      "grad_norm": 0.3976689279079437,
      "learning_rate": 2.3026233992291436e-06,
      "loss": 0.6589,
      "step": 13018
    },
    {
      "epoch": 0.5395582079655187,
      "grad_norm": 0.39127814769744873,
      "learning_rate": 2.302416179700775e-06,
      "loss": 0.7114,
      "step": 13019
    },
    {
      "epoch": 0.5395996518711923,
      "grad_norm": 0.4111287593841553,
      "learning_rate": 2.302208960172407e-06,
      "loss": 0.6707,
      "step": 13020
    },
    {
      "epoch": 0.539641095776866,
      "grad_norm": 0.42271688580513,
      "learning_rate": 2.3020017406440382e-06,
      "loss": 0.7271,
      "step": 13021
    },
    {
      "epoch": 0.5396825396825397,
      "grad_norm": 0.41561102867126465,
      "learning_rate": 2.30179452111567e-06,
      "loss": 0.6709,
      "step": 13022
    },
    {
      "epoch": 0.5397239835882134,
      "grad_norm": 0.39351606369018555,
      "learning_rate": 2.301587301587302e-06,
      "loss": 0.6759,
      "step": 13023
    },
    {
      "epoch": 0.539765427493887,
      "grad_norm": 0.3768216371536255,
      "learning_rate": 2.3013800820589337e-06,
      "loss": 0.6565,
      "step": 13024
    },
    {
      "epoch": 0.5398068713995607,
      "grad_norm": 0.4016602337360382,
      "learning_rate": 2.301172862530565e-06,
      "loss": 0.6904,
      "step": 13025
    },
    {
      "epoch": 0.5398483153052344,
      "grad_norm": 0.395116925239563,
      "learning_rate": 2.300965643002197e-06,
      "loss": 0.6925,
      "step": 13026
    },
    {
      "epoch": 0.539889759210908,
      "grad_norm": 0.3856351375579834,
      "learning_rate": 2.3007584234738282e-06,
      "loss": 0.644,
      "step": 13027
    },
    {
      "epoch": 0.5399312031165817,
      "grad_norm": 0.4040999412536621,
      "learning_rate": 2.30055120394546e-06,
      "loss": 0.7013,
      "step": 13028
    },
    {
      "epoch": 0.5399726470222553,
      "grad_norm": 0.3965044319629669,
      "learning_rate": 2.3003439844170914e-06,
      "loss": 0.6924,
      "step": 13029
    },
    {
      "epoch": 0.5400140909279291,
      "grad_norm": 0.40864816308021545,
      "learning_rate": 2.3001367648887232e-06,
      "loss": 0.7048,
      "step": 13030
    },
    {
      "epoch": 0.5400555348336027,
      "grad_norm": 0.4118366241455078,
      "learning_rate": 2.299929545360355e-06,
      "loss": 0.6825,
      "step": 13031
    },
    {
      "epoch": 0.5400969787392764,
      "grad_norm": 0.42553189396858215,
      "learning_rate": 2.2997223258319864e-06,
      "loss": 0.7012,
      "step": 13032
    },
    {
      "epoch": 0.54013842264495,
      "grad_norm": 0.45001623034477234,
      "learning_rate": 2.2995151063036182e-06,
      "loss": 0.6578,
      "step": 13033
    },
    {
      "epoch": 0.5401798665506238,
      "grad_norm": 0.43337246775627136,
      "learning_rate": 2.29930788677525e-06,
      "loss": 0.6746,
      "step": 13034
    },
    {
      "epoch": 0.5402213104562974,
      "grad_norm": 0.4214021563529968,
      "learning_rate": 2.2991006672468814e-06,
      "loss": 0.7346,
      "step": 13035
    },
    {
      "epoch": 0.540262754361971,
      "grad_norm": 0.456148236989975,
      "learning_rate": 2.2988934477185132e-06,
      "loss": 0.7015,
      "step": 13036
    },
    {
      "epoch": 0.5403041982676448,
      "grad_norm": 0.47132354974746704,
      "learning_rate": 2.2986862281901446e-06,
      "loss": 0.7229,
      "step": 13037
    },
    {
      "epoch": 0.5403456421733184,
      "grad_norm": 0.4156130850315094,
      "learning_rate": 2.2984790086617764e-06,
      "loss": 0.6733,
      "step": 13038
    },
    {
      "epoch": 0.5403870860789921,
      "grad_norm": 0.4088706076145172,
      "learning_rate": 2.2982717891334082e-06,
      "loss": 0.6875,
      "step": 13039
    },
    {
      "epoch": 0.5404285299846657,
      "grad_norm": 0.3998040556907654,
      "learning_rate": 2.2980645696050396e-06,
      "loss": 0.7097,
      "step": 13040
    },
    {
      "epoch": 0.5404699738903395,
      "grad_norm": 0.39529040455818176,
      "learning_rate": 2.2978573500766714e-06,
      "loss": 0.6826,
      "step": 13041
    },
    {
      "epoch": 0.5405114177960131,
      "grad_norm": 0.37041836977005005,
      "learning_rate": 2.2976501305483033e-06,
      "loss": 0.6418,
      "step": 13042
    },
    {
      "epoch": 0.5405528617016868,
      "grad_norm": 0.4389377236366272,
      "learning_rate": 2.2974429110199346e-06,
      "loss": 0.7157,
      "step": 13043
    },
    {
      "epoch": 0.5405943056073604,
      "grad_norm": 0.427168071269989,
      "learning_rate": 2.2972356914915664e-06,
      "loss": 0.7134,
      "step": 13044
    },
    {
      "epoch": 0.5406357495130341,
      "grad_norm": 0.3949218988418579,
      "learning_rate": 2.297028471963198e-06,
      "loss": 0.6924,
      "step": 13045
    },
    {
      "epoch": 0.5406771934187078,
      "grad_norm": 0.3986227810382843,
      "learning_rate": 2.2968212524348296e-06,
      "loss": 0.6403,
      "step": 13046
    },
    {
      "epoch": 0.5407186373243814,
      "grad_norm": 0.3988339304924011,
      "learning_rate": 2.296614032906461e-06,
      "loss": 0.6826,
      "step": 13047
    },
    {
      "epoch": 0.5407600812300551,
      "grad_norm": 0.43006977438926697,
      "learning_rate": 2.296406813378093e-06,
      "loss": 0.7433,
      "step": 13048
    },
    {
      "epoch": 0.5408015251357288,
      "grad_norm": 0.47492456436157227,
      "learning_rate": 2.2961995938497246e-06,
      "loss": 0.7815,
      "step": 13049
    },
    {
      "epoch": 0.5408429690414025,
      "grad_norm": 0.4300835132598877,
      "learning_rate": 2.2959923743213565e-06,
      "loss": 0.6898,
      "step": 13050
    },
    {
      "epoch": 0.5408844129470761,
      "grad_norm": 0.3882294297218323,
      "learning_rate": 2.295785154792988e-06,
      "loss": 0.6832,
      "step": 13051
    },
    {
      "epoch": 0.5409258568527499,
      "grad_norm": 0.43443813920021057,
      "learning_rate": 2.2955779352646197e-06,
      "loss": 0.6836,
      "step": 13052
    },
    {
      "epoch": 0.5409673007584235,
      "grad_norm": 0.4253323972225189,
      "learning_rate": 2.295370715736251e-06,
      "loss": 0.7162,
      "step": 13053
    },
    {
      "epoch": 0.5410087446640971,
      "grad_norm": 0.41642314195632935,
      "learning_rate": 2.295163496207883e-06,
      "loss": 0.7063,
      "step": 13054
    },
    {
      "epoch": 0.5410501885697708,
      "grad_norm": 0.40942347049713135,
      "learning_rate": 2.2949562766795142e-06,
      "loss": 0.7388,
      "step": 13055
    },
    {
      "epoch": 0.5410916324754445,
      "grad_norm": 0.4091856777667999,
      "learning_rate": 2.294749057151146e-06,
      "loss": 0.6875,
      "step": 13056
    },
    {
      "epoch": 0.5411330763811182,
      "grad_norm": 0.43276774883270264,
      "learning_rate": 2.294541837622778e-06,
      "loss": 0.7205,
      "step": 13057
    },
    {
      "epoch": 0.5411745202867918,
      "grad_norm": 0.4142039716243744,
      "learning_rate": 2.2943346180944092e-06,
      "loss": 0.6831,
      "step": 13058
    },
    {
      "epoch": 0.5412159641924655,
      "grad_norm": 0.41540977358818054,
      "learning_rate": 2.294127398566041e-06,
      "loss": 0.6842,
      "step": 13059
    },
    {
      "epoch": 0.5412574080981392,
      "grad_norm": 0.412065327167511,
      "learning_rate": 2.293920179037673e-06,
      "loss": 0.6858,
      "step": 13060
    },
    {
      "epoch": 0.5412988520038129,
      "grad_norm": 0.39477625489234924,
      "learning_rate": 2.2937129595093042e-06,
      "loss": 0.6562,
      "step": 13061
    },
    {
      "epoch": 0.5413402959094865,
      "grad_norm": 0.39126113057136536,
      "learning_rate": 2.293505739980936e-06,
      "loss": 0.6746,
      "step": 13062
    },
    {
      "epoch": 0.5413817398151601,
      "grad_norm": 0.43605178594589233,
      "learning_rate": 2.2932985204525674e-06,
      "loss": 0.6919,
      "step": 13063
    },
    {
      "epoch": 0.5414231837208339,
      "grad_norm": 0.4126887023448944,
      "learning_rate": 2.2930913009241992e-06,
      "loss": 0.6813,
      "step": 13064
    },
    {
      "epoch": 0.5414646276265075,
      "grad_norm": 0.4009850323200226,
      "learning_rate": 2.292884081395831e-06,
      "loss": 0.6904,
      "step": 13065
    },
    {
      "epoch": 0.5415060715321812,
      "grad_norm": 0.43473395705223083,
      "learning_rate": 2.2926768618674624e-06,
      "loss": 0.6543,
      "step": 13066
    },
    {
      "epoch": 0.5415475154378548,
      "grad_norm": 0.4203380346298218,
      "learning_rate": 2.2924696423390942e-06,
      "loss": 0.701,
      "step": 13067
    },
    {
      "epoch": 0.5415889593435286,
      "grad_norm": 0.3945593237876892,
      "learning_rate": 2.292262422810726e-06,
      "loss": 0.6633,
      "step": 13068
    },
    {
      "epoch": 0.5416304032492022,
      "grad_norm": 0.4337891936302185,
      "learning_rate": 2.2920552032823574e-06,
      "loss": 0.6993,
      "step": 13069
    },
    {
      "epoch": 0.5416718471548759,
      "grad_norm": 0.3984043300151825,
      "learning_rate": 2.2918479837539893e-06,
      "loss": 0.6451,
      "step": 13070
    },
    {
      "epoch": 0.5417132910605496,
      "grad_norm": 0.39748844504356384,
      "learning_rate": 2.2916407642256206e-06,
      "loss": 0.6968,
      "step": 13071
    },
    {
      "epoch": 0.5417547349662232,
      "grad_norm": 0.40924501419067383,
      "learning_rate": 2.2914335446972524e-06,
      "loss": 0.7,
      "step": 13072
    },
    {
      "epoch": 0.5417961788718969,
      "grad_norm": 0.4150150716304779,
      "learning_rate": 2.2912263251688843e-06,
      "loss": 0.6799,
      "step": 13073
    },
    {
      "epoch": 0.5418376227775705,
      "grad_norm": 0.40606558322906494,
      "learning_rate": 2.2910191056405156e-06,
      "loss": 0.6582,
      "step": 13074
    },
    {
      "epoch": 0.5418790666832443,
      "grad_norm": 0.39317601919174194,
      "learning_rate": 2.2908118861121474e-06,
      "loss": 0.6757,
      "step": 13075
    },
    {
      "epoch": 0.5419205105889179,
      "grad_norm": 0.3806534707546234,
      "learning_rate": 2.2906046665837793e-06,
      "loss": 0.6742,
      "step": 13076
    },
    {
      "epoch": 0.5419619544945916,
      "grad_norm": 0.44157591462135315,
      "learning_rate": 2.2903974470554106e-06,
      "loss": 0.6985,
      "step": 13077
    },
    {
      "epoch": 0.5420033984002652,
      "grad_norm": 0.46515941619873047,
      "learning_rate": 2.2901902275270425e-06,
      "loss": 0.7184,
      "step": 13078
    },
    {
      "epoch": 0.5420448423059389,
      "grad_norm": 0.45570215582847595,
      "learning_rate": 2.289983007998674e-06,
      "loss": 0.6932,
      "step": 13079
    },
    {
      "epoch": 0.5420862862116126,
      "grad_norm": 0.4363085925579071,
      "learning_rate": 2.2897757884703056e-06,
      "loss": 0.6932,
      "step": 13080
    },
    {
      "epoch": 0.5421277301172862,
      "grad_norm": 0.38323503732681274,
      "learning_rate": 2.289568568941937e-06,
      "loss": 0.6365,
      "step": 13081
    },
    {
      "epoch": 0.5421691740229599,
      "grad_norm": 0.4592910706996918,
      "learning_rate": 2.289361349413569e-06,
      "loss": 0.7134,
      "step": 13082
    },
    {
      "epoch": 0.5422106179286336,
      "grad_norm": 0.4227254092693329,
      "learning_rate": 2.2891541298852007e-06,
      "loss": 0.6667,
      "step": 13083
    },
    {
      "epoch": 0.5422520618343073,
      "grad_norm": 0.4155708849430084,
      "learning_rate": 2.288946910356832e-06,
      "loss": 0.6907,
      "step": 13084
    },
    {
      "epoch": 0.5422935057399809,
      "grad_norm": 0.42054152488708496,
      "learning_rate": 2.288739690828464e-06,
      "loss": 0.6886,
      "step": 13085
    },
    {
      "epoch": 0.5423349496456547,
      "grad_norm": 0.41738632321357727,
      "learning_rate": 2.2885324713000957e-06,
      "loss": 0.682,
      "step": 13086
    },
    {
      "epoch": 0.5423763935513283,
      "grad_norm": 0.4195530414581299,
      "learning_rate": 2.288325251771727e-06,
      "loss": 0.7346,
      "step": 13087
    },
    {
      "epoch": 0.5424178374570019,
      "grad_norm": 0.38131043314933777,
      "learning_rate": 2.288118032243359e-06,
      "loss": 0.6802,
      "step": 13088
    },
    {
      "epoch": 0.5424592813626756,
      "grad_norm": 0.3956884741783142,
      "learning_rate": 2.2879108127149902e-06,
      "loss": 0.6836,
      "step": 13089
    },
    {
      "epoch": 0.5425007252683492,
      "grad_norm": 0.4074039161205292,
      "learning_rate": 2.287703593186622e-06,
      "loss": 0.7024,
      "step": 13090
    },
    {
      "epoch": 0.542542169174023,
      "grad_norm": 0.41772302985191345,
      "learning_rate": 2.287496373658254e-06,
      "loss": 0.6877,
      "step": 13091
    },
    {
      "epoch": 0.5425836130796966,
      "grad_norm": 0.4184700846672058,
      "learning_rate": 2.2872891541298852e-06,
      "loss": 0.6973,
      "step": 13092
    },
    {
      "epoch": 0.5426250569853703,
      "grad_norm": 0.3956819772720337,
      "learning_rate": 2.287081934601517e-06,
      "loss": 0.7009,
      "step": 13093
    },
    {
      "epoch": 0.542666500891044,
      "grad_norm": 0.4205375909805298,
      "learning_rate": 2.286874715073149e-06,
      "loss": 0.6975,
      "step": 13094
    },
    {
      "epoch": 0.5427079447967177,
      "grad_norm": 0.42318397760391235,
      "learning_rate": 2.2866674955447802e-06,
      "loss": 0.6733,
      "step": 13095
    },
    {
      "epoch": 0.5427493887023913,
      "grad_norm": 0.38692793250083923,
      "learning_rate": 2.286460276016412e-06,
      "loss": 0.6378,
      "step": 13096
    },
    {
      "epoch": 0.5427908326080649,
      "grad_norm": 0.4002832770347595,
      "learning_rate": 2.2862530564880434e-06,
      "loss": 0.6361,
      "step": 13097
    },
    {
      "epoch": 0.5428322765137387,
      "grad_norm": 0.40159061551094055,
      "learning_rate": 2.2860458369596752e-06,
      "loss": 0.7712,
      "step": 13098
    },
    {
      "epoch": 0.5428737204194123,
      "grad_norm": 0.40783387422561646,
      "learning_rate": 2.285838617431307e-06,
      "loss": 0.6782,
      "step": 13099
    },
    {
      "epoch": 0.542915164325086,
      "grad_norm": 0.38862863183021545,
      "learning_rate": 2.2856313979029384e-06,
      "loss": 0.6938,
      "step": 13100
    },
    {
      "epoch": 0.5429566082307596,
      "grad_norm": 0.42051997780799866,
      "learning_rate": 2.2854241783745703e-06,
      "loss": 0.6851,
      "step": 13101
    },
    {
      "epoch": 0.5429980521364334,
      "grad_norm": 0.3822377920150757,
      "learning_rate": 2.285216958846202e-06,
      "loss": 0.6453,
      "step": 13102
    },
    {
      "epoch": 0.543039496042107,
      "grad_norm": 0.42325931787490845,
      "learning_rate": 2.2850097393178334e-06,
      "loss": 0.6246,
      "step": 13103
    },
    {
      "epoch": 0.5430809399477807,
      "grad_norm": 0.4015417695045471,
      "learning_rate": 2.2848025197894653e-06,
      "loss": 0.6528,
      "step": 13104
    },
    {
      "epoch": 0.5431223838534543,
      "grad_norm": 0.37978196144104004,
      "learning_rate": 2.2845953002610966e-06,
      "loss": 0.6548,
      "step": 13105
    },
    {
      "epoch": 0.543163827759128,
      "grad_norm": 0.4474363923072815,
      "learning_rate": 2.2843880807327285e-06,
      "loss": 0.7147,
      "step": 13106
    },
    {
      "epoch": 0.5432052716648017,
      "grad_norm": 0.38809752464294434,
      "learning_rate": 2.2841808612043603e-06,
      "loss": 0.6755,
      "step": 13107
    },
    {
      "epoch": 0.5432467155704753,
      "grad_norm": 0.40150949358940125,
      "learning_rate": 2.2839736416759916e-06,
      "loss": 0.6897,
      "step": 13108
    },
    {
      "epoch": 0.5432881594761491,
      "grad_norm": 0.3872639834880829,
      "learning_rate": 2.2837664221476235e-06,
      "loss": 0.7091,
      "step": 13109
    },
    {
      "epoch": 0.5433296033818227,
      "grad_norm": 0.4326481819152832,
      "learning_rate": 2.283559202619255e-06,
      "loss": 0.7122,
      "step": 13110
    },
    {
      "epoch": 0.5433710472874964,
      "grad_norm": 0.40404340624809265,
      "learning_rate": 2.2833519830908867e-06,
      "loss": 0.6909,
      "step": 13111
    },
    {
      "epoch": 0.54341249119317,
      "grad_norm": 0.4216577112674713,
      "learning_rate": 2.2831447635625185e-06,
      "loss": 0.689,
      "step": 13112
    },
    {
      "epoch": 0.5434539350988438,
      "grad_norm": 0.39716944098472595,
      "learning_rate": 2.28293754403415e-06,
      "loss": 0.67,
      "step": 13113
    },
    {
      "epoch": 0.5434953790045174,
      "grad_norm": 0.4619634449481964,
      "learning_rate": 2.2827303245057817e-06,
      "loss": 0.6847,
      "step": 13114
    },
    {
      "epoch": 0.543536822910191,
      "grad_norm": 0.40128597617149353,
      "learning_rate": 2.282523104977413e-06,
      "loss": 0.6565,
      "step": 13115
    },
    {
      "epoch": 0.5435782668158647,
      "grad_norm": 0.4360540807247162,
      "learning_rate": 2.282315885449045e-06,
      "loss": 0.6587,
      "step": 13116
    },
    {
      "epoch": 0.5436197107215384,
      "grad_norm": 0.40819254517555237,
      "learning_rate": 2.2821086659206767e-06,
      "loss": 0.7219,
      "step": 13117
    },
    {
      "epoch": 0.5436611546272121,
      "grad_norm": 0.4125180244445801,
      "learning_rate": 2.281901446392308e-06,
      "loss": 0.6675,
      "step": 13118
    },
    {
      "epoch": 0.5437025985328857,
      "grad_norm": 0.430863618850708,
      "learning_rate": 2.28169422686394e-06,
      "loss": 0.6963,
      "step": 13119
    },
    {
      "epoch": 0.5437440424385594,
      "grad_norm": 0.35979706048965454,
      "learning_rate": 2.2814870073355717e-06,
      "loss": 0.6492,
      "step": 13120
    },
    {
      "epoch": 0.5437854863442331,
      "grad_norm": 0.43866249918937683,
      "learning_rate": 2.281279787807203e-06,
      "loss": 0.7395,
      "step": 13121
    },
    {
      "epoch": 0.5438269302499068,
      "grad_norm": 0.4313240647315979,
      "learning_rate": 2.281072568278835e-06,
      "loss": 0.6444,
      "step": 13122
    },
    {
      "epoch": 0.5438683741555804,
      "grad_norm": 0.4026164412498474,
      "learning_rate": 2.2808653487504662e-06,
      "loss": 0.6144,
      "step": 13123
    },
    {
      "epoch": 0.543909818061254,
      "grad_norm": 0.3814598321914673,
      "learning_rate": 2.280658129222098e-06,
      "loss": 0.6433,
      "step": 13124
    },
    {
      "epoch": 0.5439512619669278,
      "grad_norm": 0.4032510221004486,
      "learning_rate": 2.28045090969373e-06,
      "loss": 0.6511,
      "step": 13125
    },
    {
      "epoch": 0.5439927058726014,
      "grad_norm": 0.3973713517189026,
      "learning_rate": 2.2802436901653612e-06,
      "loss": 0.677,
      "step": 13126
    },
    {
      "epoch": 0.5440341497782751,
      "grad_norm": 0.39668890833854675,
      "learning_rate": 2.280036470636993e-06,
      "loss": 0.6393,
      "step": 13127
    },
    {
      "epoch": 0.5440755936839488,
      "grad_norm": 0.42134392261505127,
      "learning_rate": 2.279829251108625e-06,
      "loss": 0.718,
      "step": 13128
    },
    {
      "epoch": 0.5441170375896225,
      "grad_norm": 0.38493219017982483,
      "learning_rate": 2.2796220315802563e-06,
      "loss": 0.6833,
      "step": 13129
    },
    {
      "epoch": 0.5441584814952961,
      "grad_norm": 0.40522825717926025,
      "learning_rate": 2.279414812051888e-06,
      "loss": 0.7354,
      "step": 13130
    },
    {
      "epoch": 0.5441999254009698,
      "grad_norm": 0.43816831707954407,
      "learning_rate": 2.2792075925235194e-06,
      "loss": 0.7561,
      "step": 13131
    },
    {
      "epoch": 0.5442413693066435,
      "grad_norm": 0.4227769076824188,
      "learning_rate": 2.2790003729951513e-06,
      "loss": 0.6897,
      "step": 13132
    },
    {
      "epoch": 0.5442828132123171,
      "grad_norm": 0.4518188238143921,
      "learning_rate": 2.278793153466783e-06,
      "loss": 0.7461,
      "step": 13133
    },
    {
      "epoch": 0.5443242571179908,
      "grad_norm": 0.5409524440765381,
      "learning_rate": 2.2785859339384144e-06,
      "loss": 0.6655,
      "step": 13134
    },
    {
      "epoch": 0.5443657010236644,
      "grad_norm": 0.4639953374862671,
      "learning_rate": 2.2783787144100463e-06,
      "loss": 0.7366,
      "step": 13135
    },
    {
      "epoch": 0.5444071449293382,
      "grad_norm": 0.41584667563438416,
      "learning_rate": 2.2781714948816776e-06,
      "loss": 0.6802,
      "step": 13136
    },
    {
      "epoch": 0.5444485888350118,
      "grad_norm": 0.4627249538898468,
      "learning_rate": 2.2779642753533095e-06,
      "loss": 0.7185,
      "step": 13137
    },
    {
      "epoch": 0.5444900327406855,
      "grad_norm": 0.40552303194999695,
      "learning_rate": 2.2777570558249413e-06,
      "loss": 0.6664,
      "step": 13138
    },
    {
      "epoch": 0.5445314766463591,
      "grad_norm": 0.4061635136604309,
      "learning_rate": 2.2775498362965726e-06,
      "loss": 0.6912,
      "step": 13139
    },
    {
      "epoch": 0.5445729205520328,
      "grad_norm": 0.4271814525127411,
      "learning_rate": 2.2773426167682045e-06,
      "loss": 0.7148,
      "step": 13140
    },
    {
      "epoch": 0.5446143644577065,
      "grad_norm": 0.44705790281295776,
      "learning_rate": 2.2771353972398363e-06,
      "loss": 0.707,
      "step": 13141
    },
    {
      "epoch": 0.5446558083633801,
      "grad_norm": 0.39973461627960205,
      "learning_rate": 2.2769281777114677e-06,
      "loss": 0.7122,
      "step": 13142
    },
    {
      "epoch": 0.5446972522690539,
      "grad_norm": 0.4686422049999237,
      "learning_rate": 2.2767209581830995e-06,
      "loss": 0.7623,
      "step": 13143
    },
    {
      "epoch": 0.5447386961747275,
      "grad_norm": 0.41152653098106384,
      "learning_rate": 2.276513738654731e-06,
      "loss": 0.6787,
      "step": 13144
    },
    {
      "epoch": 0.5447801400804012,
      "grad_norm": 0.4192121922969818,
      "learning_rate": 2.2763065191263627e-06,
      "loss": 0.6924,
      "step": 13145
    },
    {
      "epoch": 0.5448215839860748,
      "grad_norm": 0.4294746518135071,
      "learning_rate": 2.2760992995979945e-06,
      "loss": 0.7168,
      "step": 13146
    },
    {
      "epoch": 0.5448630278917486,
      "grad_norm": 0.37345921993255615,
      "learning_rate": 2.275892080069626e-06,
      "loss": 0.6898,
      "step": 13147
    },
    {
      "epoch": 0.5449044717974222,
      "grad_norm": 0.42853042483329773,
      "learning_rate": 2.2756848605412577e-06,
      "loss": 0.6921,
      "step": 13148
    },
    {
      "epoch": 0.5449459157030958,
      "grad_norm": 0.43928271532058716,
      "learning_rate": 2.2754776410128895e-06,
      "loss": 0.657,
      "step": 13149
    },
    {
      "epoch": 0.5449873596087695,
      "grad_norm": 0.3801010549068451,
      "learning_rate": 2.275270421484521e-06,
      "loss": 0.6858,
      "step": 13150
    },
    {
      "epoch": 0.5450288035144432,
      "grad_norm": 0.41162487864494324,
      "learning_rate": 2.2750632019561527e-06,
      "loss": 0.6892,
      "step": 13151
    },
    {
      "epoch": 0.5450702474201169,
      "grad_norm": 0.4198351502418518,
      "learning_rate": 2.274855982427784e-06,
      "loss": 0.6488,
      "step": 13152
    },
    {
      "epoch": 0.5451116913257905,
      "grad_norm": 0.446866512298584,
      "learning_rate": 2.274648762899416e-06,
      "loss": 0.6951,
      "step": 13153
    },
    {
      "epoch": 0.5451531352314642,
      "grad_norm": 0.4092423915863037,
      "learning_rate": 2.2744415433710477e-06,
      "loss": 0.6873,
      "step": 13154
    },
    {
      "epoch": 0.5451945791371379,
      "grad_norm": 0.4121929705142975,
      "learning_rate": 2.274234323842679e-06,
      "loss": 0.6512,
      "step": 13155
    },
    {
      "epoch": 0.5452360230428116,
      "grad_norm": 0.41316282749176025,
      "learning_rate": 2.274027104314311e-06,
      "loss": 0.6753,
      "step": 13156
    },
    {
      "epoch": 0.5452774669484852,
      "grad_norm": 0.4032987654209137,
      "learning_rate": 2.2738198847859422e-06,
      "loss": 0.7125,
      "step": 13157
    },
    {
      "epoch": 0.5453189108541588,
      "grad_norm": 0.391738623380661,
      "learning_rate": 2.273612665257574e-06,
      "loss": 0.6499,
      "step": 13158
    },
    {
      "epoch": 0.5453603547598326,
      "grad_norm": 0.41248148679733276,
      "learning_rate": 2.273405445729206e-06,
      "loss": 0.6793,
      "step": 13159
    },
    {
      "epoch": 0.5454017986655062,
      "grad_norm": 0.45113813877105713,
      "learning_rate": 2.2731982262008373e-06,
      "loss": 0.7449,
      "step": 13160
    },
    {
      "epoch": 0.5454432425711799,
      "grad_norm": 0.4089980125427246,
      "learning_rate": 2.272991006672469e-06,
      "loss": 0.6821,
      "step": 13161
    },
    {
      "epoch": 0.5454846864768536,
      "grad_norm": 0.39982855319976807,
      "learning_rate": 2.2727837871441004e-06,
      "loss": 0.6772,
      "step": 13162
    },
    {
      "epoch": 0.5455261303825273,
      "grad_norm": 0.4435079097747803,
      "learning_rate": 2.2725765676157323e-06,
      "loss": 0.6919,
      "step": 13163
    },
    {
      "epoch": 0.5455675742882009,
      "grad_norm": 0.4188969135284424,
      "learning_rate": 2.272369348087364e-06,
      "loss": 0.6659,
      "step": 13164
    },
    {
      "epoch": 0.5456090181938746,
      "grad_norm": 0.38747015595436096,
      "learning_rate": 2.2721621285589955e-06,
      "loss": 0.6436,
      "step": 13165
    },
    {
      "epoch": 0.5456504620995483,
      "grad_norm": 0.40399423241615295,
      "learning_rate": 2.2719549090306273e-06,
      "loss": 0.6725,
      "step": 13166
    },
    {
      "epoch": 0.5456919060052219,
      "grad_norm": 0.4200986325740814,
      "learning_rate": 2.271747689502259e-06,
      "loss": 0.7147,
      "step": 13167
    },
    {
      "epoch": 0.5457333499108956,
      "grad_norm": 0.4272199869155884,
      "learning_rate": 2.2715404699738905e-06,
      "loss": 0.7373,
      "step": 13168
    },
    {
      "epoch": 0.5457747938165692,
      "grad_norm": 0.5077555179595947,
      "learning_rate": 2.2713332504455223e-06,
      "loss": 0.738,
      "step": 13169
    },
    {
      "epoch": 0.545816237722243,
      "grad_norm": 0.3806942403316498,
      "learning_rate": 2.2711260309171537e-06,
      "loss": 0.6208,
      "step": 13170
    },
    {
      "epoch": 0.5458576816279166,
      "grad_norm": 0.3827495574951172,
      "learning_rate": 2.2709188113887855e-06,
      "loss": 0.6858,
      "step": 13171
    },
    {
      "epoch": 0.5458991255335903,
      "grad_norm": 0.3979736566543579,
      "learning_rate": 2.2707115918604173e-06,
      "loss": 0.6787,
      "step": 13172
    },
    {
      "epoch": 0.545940569439264,
      "grad_norm": 0.4549827575683594,
      "learning_rate": 2.2705043723320487e-06,
      "loss": 0.7212,
      "step": 13173
    },
    {
      "epoch": 0.5459820133449377,
      "grad_norm": 0.4281477928161621,
      "learning_rate": 2.2702971528036805e-06,
      "loss": 0.7002,
      "step": 13174
    },
    {
      "epoch": 0.5460234572506113,
      "grad_norm": 0.4416317343711853,
      "learning_rate": 2.2700899332753123e-06,
      "loss": 0.6508,
      "step": 13175
    },
    {
      "epoch": 0.5460649011562849,
      "grad_norm": 0.4436587989330292,
      "learning_rate": 2.2698827137469437e-06,
      "loss": 0.6719,
      "step": 13176
    },
    {
      "epoch": 0.5461063450619587,
      "grad_norm": 0.4420471489429474,
      "learning_rate": 2.2696754942185755e-06,
      "loss": 0.6978,
      "step": 13177
    },
    {
      "epoch": 0.5461477889676323,
      "grad_norm": 0.40397176146507263,
      "learning_rate": 2.269468274690207e-06,
      "loss": 0.6769,
      "step": 13178
    },
    {
      "epoch": 0.546189232873306,
      "grad_norm": 0.4033898115158081,
      "learning_rate": 2.2692610551618387e-06,
      "loss": 0.6987,
      "step": 13179
    },
    {
      "epoch": 0.5462306767789796,
      "grad_norm": 0.4174647927284241,
      "learning_rate": 2.26905383563347e-06,
      "loss": 0.6802,
      "step": 13180
    },
    {
      "epoch": 0.5462721206846534,
      "grad_norm": 0.45528751611709595,
      "learning_rate": 2.268846616105102e-06,
      "loss": 0.6853,
      "step": 13181
    },
    {
      "epoch": 0.546313564590327,
      "grad_norm": 0.4134081304073334,
      "learning_rate": 2.2686393965767337e-06,
      "loss": 0.7244,
      "step": 13182
    },
    {
      "epoch": 0.5463550084960007,
      "grad_norm": 0.377729207277298,
      "learning_rate": 2.2684321770483655e-06,
      "loss": 0.6685,
      "step": 13183
    },
    {
      "epoch": 0.5463964524016743,
      "grad_norm": 0.3915073871612549,
      "learning_rate": 2.268224957519997e-06,
      "loss": 0.6958,
      "step": 13184
    },
    {
      "epoch": 0.546437896307348,
      "grad_norm": 0.38253268599510193,
      "learning_rate": 2.2680177379916287e-06,
      "loss": 0.6853,
      "step": 13185
    },
    {
      "epoch": 0.5464793402130217,
      "grad_norm": 0.41328439116477966,
      "learning_rate": 2.26781051846326e-06,
      "loss": 0.6653,
      "step": 13186
    },
    {
      "epoch": 0.5465207841186953,
      "grad_norm": 0.4042830765247345,
      "learning_rate": 2.267603298934892e-06,
      "loss": 0.6847,
      "step": 13187
    },
    {
      "epoch": 0.546562228024369,
      "grad_norm": 0.4443589746952057,
      "learning_rate": 2.2673960794065233e-06,
      "loss": 0.7046,
      "step": 13188
    },
    {
      "epoch": 0.5466036719300427,
      "grad_norm": 0.41989824175834656,
      "learning_rate": 2.267188859878155e-06,
      "loss": 0.6438,
      "step": 13189
    },
    {
      "epoch": 0.5466451158357164,
      "grad_norm": 0.4113078713417053,
      "learning_rate": 2.266981640349787e-06,
      "loss": 0.6528,
      "step": 13190
    },
    {
      "epoch": 0.54668655974139,
      "grad_norm": 0.42277461290359497,
      "learning_rate": 2.2667744208214183e-06,
      "loss": 0.6592,
      "step": 13191
    },
    {
      "epoch": 0.5467280036470638,
      "grad_norm": 0.4090423583984375,
      "learning_rate": 2.26656720129305e-06,
      "loss": 0.6849,
      "step": 13192
    },
    {
      "epoch": 0.5467694475527374,
      "grad_norm": 0.43814489245414734,
      "learning_rate": 2.266359981764682e-06,
      "loss": 0.662,
      "step": 13193
    },
    {
      "epoch": 0.546810891458411,
      "grad_norm": 0.4450767934322357,
      "learning_rate": 2.2661527622363133e-06,
      "loss": 0.7466,
      "step": 13194
    },
    {
      "epoch": 0.5468523353640847,
      "grad_norm": 0.4039347767829895,
      "learning_rate": 2.265945542707945e-06,
      "loss": 0.6963,
      "step": 13195
    },
    {
      "epoch": 0.5468937792697584,
      "grad_norm": 0.39635682106018066,
      "learning_rate": 2.2657383231795765e-06,
      "loss": 0.7201,
      "step": 13196
    },
    {
      "epoch": 0.5469352231754321,
      "grad_norm": 0.41493159532546997,
      "learning_rate": 2.2655311036512083e-06,
      "loss": 0.6906,
      "step": 13197
    },
    {
      "epoch": 0.5469766670811057,
      "grad_norm": 0.43690812587738037,
      "learning_rate": 2.26532388412284e-06,
      "loss": 0.7153,
      "step": 13198
    },
    {
      "epoch": 0.5470181109867794,
      "grad_norm": 0.44056639075279236,
      "learning_rate": 2.2651166645944715e-06,
      "loss": 0.6562,
      "step": 13199
    },
    {
      "epoch": 0.5470595548924531,
      "grad_norm": 0.41806933283805847,
      "learning_rate": 2.2649094450661033e-06,
      "loss": 0.6848,
      "step": 13200
    },
    {
      "epoch": 0.5471009987981267,
      "grad_norm": 0.3945106565952301,
      "learning_rate": 2.264702225537735e-06,
      "loss": 0.6643,
      "step": 13201
    },
    {
      "epoch": 0.5471424427038004,
      "grad_norm": 0.4359179735183716,
      "learning_rate": 2.2644950060093665e-06,
      "loss": 0.7153,
      "step": 13202
    },
    {
      "epoch": 0.547183886609474,
      "grad_norm": 0.43853121995925903,
      "learning_rate": 2.2642877864809983e-06,
      "loss": 0.6973,
      "step": 13203
    },
    {
      "epoch": 0.5472253305151478,
      "grad_norm": 0.44614842534065247,
      "learning_rate": 2.2640805669526297e-06,
      "loss": 0.6396,
      "step": 13204
    },
    {
      "epoch": 0.5472667744208214,
      "grad_norm": 0.44142046570777893,
      "learning_rate": 2.2638733474242615e-06,
      "loss": 0.6945,
      "step": 13205
    },
    {
      "epoch": 0.5473082183264951,
      "grad_norm": 0.8966847658157349,
      "learning_rate": 2.263666127895893e-06,
      "loss": 0.6576,
      "step": 13206
    },
    {
      "epoch": 0.5473496622321687,
      "grad_norm": 0.39809247851371765,
      "learning_rate": 2.2634589083675247e-06,
      "loss": 0.7126,
      "step": 13207
    },
    {
      "epoch": 0.5473911061378425,
      "grad_norm": 0.431170791387558,
      "learning_rate": 2.2632516888391565e-06,
      "loss": 0.6633,
      "step": 13208
    },
    {
      "epoch": 0.5474325500435161,
      "grad_norm": 0.41954490542411804,
      "learning_rate": 2.2630444693107883e-06,
      "loss": 0.6902,
      "step": 13209
    },
    {
      "epoch": 0.5474739939491897,
      "grad_norm": 0.452528178691864,
      "learning_rate": 2.2628372497824197e-06,
      "loss": 0.6694,
      "step": 13210
    },
    {
      "epoch": 0.5475154378548635,
      "grad_norm": 0.4167577028274536,
      "learning_rate": 2.2626300302540515e-06,
      "loss": 0.627,
      "step": 13211
    },
    {
      "epoch": 0.5475568817605371,
      "grad_norm": 0.41400668025016785,
      "learning_rate": 2.262422810725683e-06,
      "loss": 0.6821,
      "step": 13212
    },
    {
      "epoch": 0.5475983256662108,
      "grad_norm": 0.3875214755535126,
      "learning_rate": 2.2622155911973147e-06,
      "loss": 0.6517,
      "step": 13213
    },
    {
      "epoch": 0.5476397695718844,
      "grad_norm": 0.38791581988334656,
      "learning_rate": 2.262008371668946e-06,
      "loss": 0.6417,
      "step": 13214
    },
    {
      "epoch": 0.5476812134775582,
      "grad_norm": 0.39880916476249695,
      "learning_rate": 2.261801152140578e-06,
      "loss": 0.6558,
      "step": 13215
    },
    {
      "epoch": 0.5477226573832318,
      "grad_norm": 0.3855287432670593,
      "learning_rate": 2.2615939326122097e-06,
      "loss": 0.7129,
      "step": 13216
    },
    {
      "epoch": 0.5477641012889055,
      "grad_norm": 0.39640384912490845,
      "learning_rate": 2.2613867130838415e-06,
      "loss": 0.6631,
      "step": 13217
    },
    {
      "epoch": 0.5478055451945791,
      "grad_norm": 0.4335428476333618,
      "learning_rate": 2.261179493555473e-06,
      "loss": 0.6565,
      "step": 13218
    },
    {
      "epoch": 0.5478469891002528,
      "grad_norm": 0.4185199737548828,
      "learning_rate": 2.2609722740271047e-06,
      "loss": 0.6544,
      "step": 13219
    },
    {
      "epoch": 0.5478884330059265,
      "grad_norm": 0.44557300209999084,
      "learning_rate": 2.260765054498736e-06,
      "loss": 0.6923,
      "step": 13220
    },
    {
      "epoch": 0.5479298769116001,
      "grad_norm": 0.4069538116455078,
      "learning_rate": 2.260557834970368e-06,
      "loss": 0.6913,
      "step": 13221
    },
    {
      "epoch": 0.5479713208172738,
      "grad_norm": 0.39331233501434326,
      "learning_rate": 2.2603506154419993e-06,
      "loss": 0.6794,
      "step": 13222
    },
    {
      "epoch": 0.5480127647229475,
      "grad_norm": 0.3724033236503601,
      "learning_rate": 2.260143395913631e-06,
      "loss": 0.6251,
      "step": 13223
    },
    {
      "epoch": 0.5480542086286212,
      "grad_norm": 0.4439789354801178,
      "learning_rate": 2.259936176385263e-06,
      "loss": 0.7126,
      "step": 13224
    },
    {
      "epoch": 0.5480956525342948,
      "grad_norm": 0.443485289812088,
      "learning_rate": 2.2597289568568947e-06,
      "loss": 0.6924,
      "step": 13225
    },
    {
      "epoch": 0.5481370964399686,
      "grad_norm": 0.4384426474571228,
      "learning_rate": 2.259521737328526e-06,
      "loss": 0.7039,
      "step": 13226
    },
    {
      "epoch": 0.5481785403456422,
      "grad_norm": 0.44989538192749023,
      "learning_rate": 2.259314517800158e-06,
      "loss": 0.7468,
      "step": 13227
    },
    {
      "epoch": 0.5482199842513158,
      "grad_norm": 0.4054316282272339,
      "learning_rate": 2.2591072982717893e-06,
      "loss": 0.6938,
      "step": 13228
    },
    {
      "epoch": 0.5482614281569895,
      "grad_norm": 0.42390158772468567,
      "learning_rate": 2.258900078743421e-06,
      "loss": 0.7091,
      "step": 13229
    },
    {
      "epoch": 0.5483028720626631,
      "grad_norm": 0.4133685827255249,
      "learning_rate": 2.2586928592150525e-06,
      "loss": 0.7507,
      "step": 13230
    },
    {
      "epoch": 0.5483443159683369,
      "grad_norm": 0.41284993290901184,
      "learning_rate": 2.2584856396866843e-06,
      "loss": 0.7224,
      "step": 13231
    },
    {
      "epoch": 0.5483857598740105,
      "grad_norm": 0.4839017689228058,
      "learning_rate": 2.2582784201583157e-06,
      "loss": 0.6425,
      "step": 13232
    },
    {
      "epoch": 0.5484272037796842,
      "grad_norm": 0.42022016644477844,
      "learning_rate": 2.2580712006299475e-06,
      "loss": 0.6659,
      "step": 13233
    },
    {
      "epoch": 0.5484686476853579,
      "grad_norm": 0.39305222034454346,
      "learning_rate": 2.2578639811015793e-06,
      "loss": 0.6298,
      "step": 13234
    },
    {
      "epoch": 0.5485100915910316,
      "grad_norm": 0.4334964454174042,
      "learning_rate": 2.257656761573211e-06,
      "loss": 0.6958,
      "step": 13235
    },
    {
      "epoch": 0.5485515354967052,
      "grad_norm": 0.4476449489593506,
      "learning_rate": 2.2574495420448425e-06,
      "loss": 0.7383,
      "step": 13236
    },
    {
      "epoch": 0.5485929794023788,
      "grad_norm": 0.4088231921195984,
      "learning_rate": 2.2572423225164743e-06,
      "loss": 0.6968,
      "step": 13237
    },
    {
      "epoch": 0.5486344233080526,
      "grad_norm": 0.40412062406539917,
      "learning_rate": 2.2570351029881057e-06,
      "loss": 0.6982,
      "step": 13238
    },
    {
      "epoch": 0.5486758672137262,
      "grad_norm": 0.4167626202106476,
      "learning_rate": 2.2568278834597375e-06,
      "loss": 0.673,
      "step": 13239
    },
    {
      "epoch": 0.5487173111193999,
      "grad_norm": 0.4079391062259674,
      "learning_rate": 2.256620663931369e-06,
      "loss": 0.6562,
      "step": 13240
    },
    {
      "epoch": 0.5487587550250735,
      "grad_norm": 0.4207518994808197,
      "learning_rate": 2.2564134444030007e-06,
      "loss": 0.6934,
      "step": 13241
    },
    {
      "epoch": 0.5488001989307473,
      "grad_norm": 0.4155116081237793,
      "learning_rate": 2.2562062248746325e-06,
      "loss": 0.7029,
      "step": 13242
    },
    {
      "epoch": 0.5488416428364209,
      "grad_norm": 0.40218666195869446,
      "learning_rate": 2.2559990053462643e-06,
      "loss": 0.6495,
      "step": 13243
    },
    {
      "epoch": 0.5488830867420946,
      "grad_norm": 0.3839605450630188,
      "learning_rate": 2.2557917858178957e-06,
      "loss": 0.6687,
      "step": 13244
    },
    {
      "epoch": 0.5489245306477682,
      "grad_norm": 0.4050431549549103,
      "learning_rate": 2.2555845662895275e-06,
      "loss": 0.696,
      "step": 13245
    },
    {
      "epoch": 0.5489659745534419,
      "grad_norm": 0.41162705421447754,
      "learning_rate": 2.255377346761159e-06,
      "loss": 0.6327,
      "step": 13246
    },
    {
      "epoch": 0.5490074184591156,
      "grad_norm": 0.4483102560043335,
      "learning_rate": 2.2551701272327907e-06,
      "loss": 0.6671,
      "step": 13247
    },
    {
      "epoch": 0.5490488623647892,
      "grad_norm": 0.3960977792739868,
      "learning_rate": 2.254962907704422e-06,
      "loss": 0.6934,
      "step": 13248
    },
    {
      "epoch": 0.549090306270463,
      "grad_norm": 0.4219824969768524,
      "learning_rate": 2.254755688176054e-06,
      "loss": 0.6908,
      "step": 13249
    },
    {
      "epoch": 0.5491317501761366,
      "grad_norm": 0.3973199129104614,
      "learning_rate": 2.2545484686476857e-06,
      "loss": 0.645,
      "step": 13250
    },
    {
      "epoch": 0.5491731940818103,
      "grad_norm": 0.4612855017185211,
      "learning_rate": 2.2543412491193175e-06,
      "loss": 0.6764,
      "step": 13251
    },
    {
      "epoch": 0.5492146379874839,
      "grad_norm": 0.384584903717041,
      "learning_rate": 2.254134029590949e-06,
      "loss": 0.6985,
      "step": 13252
    },
    {
      "epoch": 0.5492560818931577,
      "grad_norm": 0.45565611124038696,
      "learning_rate": 2.2539268100625807e-06,
      "loss": 0.7134,
      "step": 13253
    },
    {
      "epoch": 0.5492975257988313,
      "grad_norm": 0.4270487427711487,
      "learning_rate": 2.253719590534212e-06,
      "loss": 0.7317,
      "step": 13254
    },
    {
      "epoch": 0.5493389697045049,
      "grad_norm": 0.4032677412033081,
      "learning_rate": 2.253512371005844e-06,
      "loss": 0.6855,
      "step": 13255
    },
    {
      "epoch": 0.5493804136101786,
      "grad_norm": 0.38839104771614075,
      "learning_rate": 2.2533051514774753e-06,
      "loss": 0.6567,
      "step": 13256
    },
    {
      "epoch": 0.5494218575158523,
      "grad_norm": 0.4111059606075287,
      "learning_rate": 2.253097931949107e-06,
      "loss": 0.6907,
      "step": 13257
    },
    {
      "epoch": 0.549463301421526,
      "grad_norm": 0.4253048598766327,
      "learning_rate": 2.2528907124207385e-06,
      "loss": 0.7207,
      "step": 13258
    },
    {
      "epoch": 0.5495047453271996,
      "grad_norm": 0.4157562553882599,
      "learning_rate": 2.2526834928923707e-06,
      "loss": 0.6483,
      "step": 13259
    },
    {
      "epoch": 0.5495461892328733,
      "grad_norm": 0.42656418681144714,
      "learning_rate": 2.252476273364002e-06,
      "loss": 0.6869,
      "step": 13260
    },
    {
      "epoch": 0.549587633138547,
      "grad_norm": 0.36805260181427,
      "learning_rate": 2.252269053835634e-06,
      "loss": 0.6427,
      "step": 13261
    },
    {
      "epoch": 0.5496290770442206,
      "grad_norm": 0.41254109144210815,
      "learning_rate": 2.2520618343072653e-06,
      "loss": 0.6276,
      "step": 13262
    },
    {
      "epoch": 0.5496705209498943,
      "grad_norm": 0.3962934911251068,
      "learning_rate": 2.251854614778897e-06,
      "loss": 0.67,
      "step": 13263
    },
    {
      "epoch": 0.549711964855568,
      "grad_norm": 0.39917755126953125,
      "learning_rate": 2.2516473952505285e-06,
      "loss": 0.6926,
      "step": 13264
    },
    {
      "epoch": 0.5497534087612417,
      "grad_norm": 0.43411973118782043,
      "learning_rate": 2.2514401757221603e-06,
      "loss": 0.6863,
      "step": 13265
    },
    {
      "epoch": 0.5497948526669153,
      "grad_norm": 0.4580281674861908,
      "learning_rate": 2.2512329561937917e-06,
      "loss": 0.7317,
      "step": 13266
    },
    {
      "epoch": 0.549836296572589,
      "grad_norm": 0.40566927194595337,
      "learning_rate": 2.2510257366654235e-06,
      "loss": 0.6594,
      "step": 13267
    },
    {
      "epoch": 0.5498777404782627,
      "grad_norm": 0.4104520380496979,
      "learning_rate": 2.2508185171370553e-06,
      "loss": 0.7013,
      "step": 13268
    },
    {
      "epoch": 0.5499191843839364,
      "grad_norm": 0.4234115481376648,
      "learning_rate": 2.250611297608687e-06,
      "loss": 0.7188,
      "step": 13269
    },
    {
      "epoch": 0.54996062828961,
      "grad_norm": 0.40120241045951843,
      "learning_rate": 2.2504040780803185e-06,
      "loss": 0.7178,
      "step": 13270
    },
    {
      "epoch": 0.5500020721952836,
      "grad_norm": 0.4030784070491791,
      "learning_rate": 2.2501968585519503e-06,
      "loss": 0.6787,
      "step": 13271
    },
    {
      "epoch": 0.5500435161009574,
      "grad_norm": 0.40126433968544006,
      "learning_rate": 2.2499896390235817e-06,
      "loss": 0.6936,
      "step": 13272
    },
    {
      "epoch": 0.550084960006631,
      "grad_norm": 0.3900946080684662,
      "learning_rate": 2.2497824194952135e-06,
      "loss": 0.6633,
      "step": 13273
    },
    {
      "epoch": 0.5501264039123047,
      "grad_norm": 0.39549195766448975,
      "learning_rate": 2.249575199966845e-06,
      "loss": 0.6417,
      "step": 13274
    },
    {
      "epoch": 0.5501678478179783,
      "grad_norm": 0.3908202350139618,
      "learning_rate": 2.2493679804384767e-06,
      "loss": 0.6134,
      "step": 13275
    },
    {
      "epoch": 0.5502092917236521,
      "grad_norm": 0.39231616258621216,
      "learning_rate": 2.2491607609101085e-06,
      "loss": 0.6418,
      "step": 13276
    },
    {
      "epoch": 0.5502507356293257,
      "grad_norm": 0.39186495542526245,
      "learning_rate": 2.2489535413817403e-06,
      "loss": 0.6826,
      "step": 13277
    },
    {
      "epoch": 0.5502921795349994,
      "grad_norm": 0.4371165335178375,
      "learning_rate": 2.2487463218533717e-06,
      "loss": 0.71,
      "step": 13278
    },
    {
      "epoch": 0.550333623440673,
      "grad_norm": 0.42693352699279785,
      "learning_rate": 2.2485391023250035e-06,
      "loss": 0.6573,
      "step": 13279
    },
    {
      "epoch": 0.5503750673463467,
      "grad_norm": 0.44119614362716675,
      "learning_rate": 2.248331882796635e-06,
      "loss": 0.7068,
      "step": 13280
    },
    {
      "epoch": 0.5504165112520204,
      "grad_norm": 0.39260008931159973,
      "learning_rate": 2.2481246632682667e-06,
      "loss": 0.7062,
      "step": 13281
    },
    {
      "epoch": 0.550457955157694,
      "grad_norm": 0.41641637682914734,
      "learning_rate": 2.247917443739898e-06,
      "loss": 0.6481,
      "step": 13282
    },
    {
      "epoch": 0.5504993990633678,
      "grad_norm": 0.49975037574768066,
      "learning_rate": 2.24771022421153e-06,
      "loss": 0.6426,
      "step": 13283
    },
    {
      "epoch": 0.5505408429690414,
      "grad_norm": 0.3906380534172058,
      "learning_rate": 2.2475030046831613e-06,
      "loss": 0.7134,
      "step": 13284
    },
    {
      "epoch": 0.5505822868747151,
      "grad_norm": 0.4052414894104004,
      "learning_rate": 2.2472957851547935e-06,
      "loss": 0.6818,
      "step": 13285
    },
    {
      "epoch": 0.5506237307803887,
      "grad_norm": 0.4188137650489807,
      "learning_rate": 2.247088565626425e-06,
      "loss": 0.7053,
      "step": 13286
    },
    {
      "epoch": 0.5506651746860625,
      "grad_norm": 0.4122617840766907,
      "learning_rate": 2.2468813460980567e-06,
      "loss": 0.6829,
      "step": 13287
    },
    {
      "epoch": 0.5507066185917361,
      "grad_norm": 0.4464053213596344,
      "learning_rate": 2.246674126569688e-06,
      "loss": 0.7141,
      "step": 13288
    },
    {
      "epoch": 0.5507480624974097,
      "grad_norm": 0.428396999835968,
      "learning_rate": 2.24646690704132e-06,
      "loss": 0.7168,
      "step": 13289
    },
    {
      "epoch": 0.5507895064030834,
      "grad_norm": 0.40052929520606995,
      "learning_rate": 2.2462596875129513e-06,
      "loss": 0.6953,
      "step": 13290
    },
    {
      "epoch": 0.5508309503087571,
      "grad_norm": 0.4280436635017395,
      "learning_rate": 2.246052467984583e-06,
      "loss": 0.6721,
      "step": 13291
    },
    {
      "epoch": 0.5508723942144308,
      "grad_norm": 0.41729235649108887,
      "learning_rate": 2.2458452484562145e-06,
      "loss": 0.7041,
      "step": 13292
    },
    {
      "epoch": 0.5509138381201044,
      "grad_norm": 0.42635929584503174,
      "learning_rate": 2.2456380289278463e-06,
      "loss": 0.7134,
      "step": 13293
    },
    {
      "epoch": 0.5509552820257781,
      "grad_norm": 0.4178532361984253,
      "learning_rate": 2.245430809399478e-06,
      "loss": 0.7241,
      "step": 13294
    },
    {
      "epoch": 0.5509967259314518,
      "grad_norm": 0.39530080556869507,
      "learning_rate": 2.24522358987111e-06,
      "loss": 0.688,
      "step": 13295
    },
    {
      "epoch": 0.5510381698371255,
      "grad_norm": 0.4052293598651886,
      "learning_rate": 2.2450163703427413e-06,
      "loss": 0.6963,
      "step": 13296
    },
    {
      "epoch": 0.5510796137427991,
      "grad_norm": 0.4075140357017517,
      "learning_rate": 2.244809150814373e-06,
      "loss": 0.677,
      "step": 13297
    },
    {
      "epoch": 0.5511210576484727,
      "grad_norm": 0.40253809094429016,
      "learning_rate": 2.2446019312860045e-06,
      "loss": 0.6547,
      "step": 13298
    },
    {
      "epoch": 0.5511625015541465,
      "grad_norm": 0.4387081265449524,
      "learning_rate": 2.2443947117576363e-06,
      "loss": 0.6918,
      "step": 13299
    },
    {
      "epoch": 0.5512039454598201,
      "grad_norm": 0.40122395753860474,
      "learning_rate": 2.2441874922292677e-06,
      "loss": 0.6665,
      "step": 13300
    },
    {
      "epoch": 0.5512453893654938,
      "grad_norm": 0.40657299757003784,
      "learning_rate": 2.2439802727008995e-06,
      "loss": 0.708,
      "step": 13301
    },
    {
      "epoch": 0.5512868332711675,
      "grad_norm": 0.38221287727355957,
      "learning_rate": 2.2437730531725313e-06,
      "loss": 0.6476,
      "step": 13302
    },
    {
      "epoch": 0.5513282771768412,
      "grad_norm": 0.39684078097343445,
      "learning_rate": 2.243565833644163e-06,
      "loss": 0.7073,
      "step": 13303
    },
    {
      "epoch": 0.5513697210825148,
      "grad_norm": 0.42343392968177795,
      "learning_rate": 2.2433586141157945e-06,
      "loss": 0.7365,
      "step": 13304
    },
    {
      "epoch": 0.5514111649881885,
      "grad_norm": 0.4186185300350189,
      "learning_rate": 2.2431513945874263e-06,
      "loss": 0.6666,
      "step": 13305
    },
    {
      "epoch": 0.5514526088938622,
      "grad_norm": 0.4141731858253479,
      "learning_rate": 2.2429441750590577e-06,
      "loss": 0.666,
      "step": 13306
    },
    {
      "epoch": 0.5514940527995358,
      "grad_norm": 0.4046260416507721,
      "learning_rate": 2.2427369555306895e-06,
      "loss": 0.6678,
      "step": 13307
    },
    {
      "epoch": 0.5515354967052095,
      "grad_norm": 0.40718501806259155,
      "learning_rate": 2.242529736002321e-06,
      "loss": 0.6548,
      "step": 13308
    },
    {
      "epoch": 0.5515769406108831,
      "grad_norm": 0.4141186475753784,
      "learning_rate": 2.2423225164739527e-06,
      "loss": 0.6973,
      "step": 13309
    },
    {
      "epoch": 0.5516183845165569,
      "grad_norm": 0.3941904604434967,
      "learning_rate": 2.242115296945584e-06,
      "loss": 0.6531,
      "step": 13310
    },
    {
      "epoch": 0.5516598284222305,
      "grad_norm": 0.37318119406700134,
      "learning_rate": 2.2419080774172163e-06,
      "loss": 0.6586,
      "step": 13311
    },
    {
      "epoch": 0.5517012723279042,
      "grad_norm": 0.40590736269950867,
      "learning_rate": 2.2417008578888477e-06,
      "loss": 0.6931,
      "step": 13312
    },
    {
      "epoch": 0.5517427162335778,
      "grad_norm": 0.43241578340530396,
      "learning_rate": 2.2414936383604795e-06,
      "loss": 0.738,
      "step": 13313
    },
    {
      "epoch": 0.5517841601392516,
      "grad_norm": 0.430996298789978,
      "learning_rate": 2.241286418832111e-06,
      "loss": 0.6921,
      "step": 13314
    },
    {
      "epoch": 0.5518256040449252,
      "grad_norm": 0.4031265079975128,
      "learning_rate": 2.2410791993037427e-06,
      "loss": 0.693,
      "step": 13315
    },
    {
      "epoch": 0.5518670479505988,
      "grad_norm": 0.43074992299079895,
      "learning_rate": 2.240871979775374e-06,
      "loss": 0.6172,
      "step": 13316
    },
    {
      "epoch": 0.5519084918562726,
      "grad_norm": 0.4218733608722687,
      "learning_rate": 2.240664760247006e-06,
      "loss": 0.7051,
      "step": 13317
    },
    {
      "epoch": 0.5519499357619462,
      "grad_norm": 0.38145074248313904,
      "learning_rate": 2.2404575407186373e-06,
      "loss": 0.6907,
      "step": 13318
    },
    {
      "epoch": 0.5519913796676199,
      "grad_norm": 0.40226516127586365,
      "learning_rate": 2.240250321190269e-06,
      "loss": 0.7002,
      "step": 13319
    },
    {
      "epoch": 0.5520328235732935,
      "grad_norm": 0.4473637640476227,
      "learning_rate": 2.240043101661901e-06,
      "loss": 0.7427,
      "step": 13320
    },
    {
      "epoch": 0.5520742674789673,
      "grad_norm": 0.3917519152164459,
      "learning_rate": 2.2398358821335327e-06,
      "loss": 0.6997,
      "step": 13321
    },
    {
      "epoch": 0.5521157113846409,
      "grad_norm": 0.3805406391620636,
      "learning_rate": 2.239628662605164e-06,
      "loss": 0.6067,
      "step": 13322
    },
    {
      "epoch": 0.5521571552903145,
      "grad_norm": 0.38390281796455383,
      "learning_rate": 2.239421443076796e-06,
      "loss": 0.6407,
      "step": 13323
    },
    {
      "epoch": 0.5521985991959882,
      "grad_norm": 0.395438551902771,
      "learning_rate": 2.2392142235484273e-06,
      "loss": 0.6602,
      "step": 13324
    },
    {
      "epoch": 0.5522400431016619,
      "grad_norm": 0.39681094884872437,
      "learning_rate": 2.239007004020059e-06,
      "loss": 0.6621,
      "step": 13325
    },
    {
      "epoch": 0.5522814870073356,
      "grad_norm": 0.4080069065093994,
      "learning_rate": 2.2387997844916905e-06,
      "loss": 0.6978,
      "step": 13326
    },
    {
      "epoch": 0.5523229309130092,
      "grad_norm": 0.41217124462127686,
      "learning_rate": 2.2385925649633223e-06,
      "loss": 0.7383,
      "step": 13327
    },
    {
      "epoch": 0.552364374818683,
      "grad_norm": 0.40536928176879883,
      "learning_rate": 2.2383853454349537e-06,
      "loss": 0.6564,
      "step": 13328
    },
    {
      "epoch": 0.5524058187243566,
      "grad_norm": 0.40895333886146545,
      "learning_rate": 2.238178125906586e-06,
      "loss": 0.7205,
      "step": 13329
    },
    {
      "epoch": 0.5524472626300303,
      "grad_norm": 0.39871978759765625,
      "learning_rate": 2.2379709063782173e-06,
      "loss": 0.6755,
      "step": 13330
    },
    {
      "epoch": 0.5524887065357039,
      "grad_norm": 0.41123121976852417,
      "learning_rate": 2.237763686849849e-06,
      "loss": 0.7473,
      "step": 13331
    },
    {
      "epoch": 0.5525301504413775,
      "grad_norm": 0.40804433822631836,
      "learning_rate": 2.2375564673214805e-06,
      "loss": 0.6782,
      "step": 13332
    },
    {
      "epoch": 0.5525715943470513,
      "grad_norm": 0.4030790627002716,
      "learning_rate": 2.2373492477931123e-06,
      "loss": 0.6848,
      "step": 13333
    },
    {
      "epoch": 0.5526130382527249,
      "grad_norm": 0.3813791275024414,
      "learning_rate": 2.2371420282647437e-06,
      "loss": 0.6738,
      "step": 13334
    },
    {
      "epoch": 0.5526544821583986,
      "grad_norm": 0.43365219235420227,
      "learning_rate": 2.2369348087363755e-06,
      "loss": 0.6936,
      "step": 13335
    },
    {
      "epoch": 0.5526959260640723,
      "grad_norm": 0.43022313714027405,
      "learning_rate": 2.236727589208007e-06,
      "loss": 0.7285,
      "step": 13336
    },
    {
      "epoch": 0.552737369969746,
      "grad_norm": 0.4141666293144226,
      "learning_rate": 2.236520369679639e-06,
      "loss": 0.6554,
      "step": 13337
    },
    {
      "epoch": 0.5527788138754196,
      "grad_norm": 0.39094865322113037,
      "learning_rate": 2.2363131501512705e-06,
      "loss": 0.6609,
      "step": 13338
    },
    {
      "epoch": 0.5528202577810933,
      "grad_norm": 0.40162554383277893,
      "learning_rate": 2.2361059306229023e-06,
      "loss": 0.613,
      "step": 13339
    },
    {
      "epoch": 0.552861701686767,
      "grad_norm": 0.4159516990184784,
      "learning_rate": 2.2358987110945337e-06,
      "loss": 0.6792,
      "step": 13340
    },
    {
      "epoch": 0.5529031455924406,
      "grad_norm": 0.4038596749305725,
      "learning_rate": 2.2356914915661655e-06,
      "loss": 0.6833,
      "step": 13341
    },
    {
      "epoch": 0.5529445894981143,
      "grad_norm": 0.3969716429710388,
      "learning_rate": 2.235484272037797e-06,
      "loss": 0.6929,
      "step": 13342
    },
    {
      "epoch": 0.5529860334037879,
      "grad_norm": 0.4287538528442383,
      "learning_rate": 2.2352770525094287e-06,
      "loss": 0.7012,
      "step": 13343
    },
    {
      "epoch": 0.5530274773094617,
      "grad_norm": 0.4244667887687683,
      "learning_rate": 2.23506983298106e-06,
      "loss": 0.672,
      "step": 13344
    },
    {
      "epoch": 0.5530689212151353,
      "grad_norm": 0.41264766454696655,
      "learning_rate": 2.234862613452692e-06,
      "loss": 0.6907,
      "step": 13345
    },
    {
      "epoch": 0.553110365120809,
      "grad_norm": 0.375772088766098,
      "learning_rate": 2.2346553939243237e-06,
      "loss": 0.6744,
      "step": 13346
    },
    {
      "epoch": 0.5531518090264826,
      "grad_norm": 0.4508207440376282,
      "learning_rate": 2.2344481743959555e-06,
      "loss": 0.7206,
      "step": 13347
    },
    {
      "epoch": 0.5531932529321564,
      "grad_norm": 0.4071665108203888,
      "learning_rate": 2.234240954867587e-06,
      "loss": 0.6775,
      "step": 13348
    },
    {
      "epoch": 0.55323469683783,
      "grad_norm": 0.4171450734138489,
      "learning_rate": 2.2340337353392187e-06,
      "loss": 0.6787,
      "step": 13349
    },
    {
      "epoch": 0.5532761407435036,
      "grad_norm": 0.4202789068222046,
      "learning_rate": 2.23382651581085e-06,
      "loss": 0.6863,
      "step": 13350
    },
    {
      "epoch": 0.5533175846491774,
      "grad_norm": 0.43829914927482605,
      "learning_rate": 2.233619296282482e-06,
      "loss": 0.6877,
      "step": 13351
    },
    {
      "epoch": 0.553359028554851,
      "grad_norm": 0.40467071533203125,
      "learning_rate": 2.2334120767541133e-06,
      "loss": 0.6951,
      "step": 13352
    },
    {
      "epoch": 0.5534004724605247,
      "grad_norm": 0.3977621793746948,
      "learning_rate": 2.233204857225745e-06,
      "loss": 0.676,
      "step": 13353
    },
    {
      "epoch": 0.5534419163661983,
      "grad_norm": 0.38484591245651245,
      "learning_rate": 2.2329976376973765e-06,
      "loss": 0.6423,
      "step": 13354
    },
    {
      "epoch": 0.5534833602718721,
      "grad_norm": 0.44045478105545044,
      "learning_rate": 2.2327904181690087e-06,
      "loss": 0.7419,
      "step": 13355
    },
    {
      "epoch": 0.5535248041775457,
      "grad_norm": 0.38302960991859436,
      "learning_rate": 2.23258319864064e-06,
      "loss": 0.6158,
      "step": 13356
    },
    {
      "epoch": 0.5535662480832194,
      "grad_norm": 0.4010436236858368,
      "learning_rate": 2.232375979112272e-06,
      "loss": 0.6357,
      "step": 13357
    },
    {
      "epoch": 0.553607691988893,
      "grad_norm": 0.42946529388427734,
      "learning_rate": 2.2321687595839033e-06,
      "loss": 0.6854,
      "step": 13358
    },
    {
      "epoch": 0.5536491358945667,
      "grad_norm": 0.4361298084259033,
      "learning_rate": 2.231961540055535e-06,
      "loss": 0.6263,
      "step": 13359
    },
    {
      "epoch": 0.5536905798002404,
      "grad_norm": 0.43426117300987244,
      "learning_rate": 2.2317543205271665e-06,
      "loss": 0.6722,
      "step": 13360
    },
    {
      "epoch": 0.553732023705914,
      "grad_norm": 0.43082141876220703,
      "learning_rate": 2.2315471009987983e-06,
      "loss": 0.7673,
      "step": 13361
    },
    {
      "epoch": 0.5537734676115877,
      "grad_norm": 0.4068543314933777,
      "learning_rate": 2.2313398814704297e-06,
      "loss": 0.6659,
      "step": 13362
    },
    {
      "epoch": 0.5538149115172614,
      "grad_norm": 0.40030407905578613,
      "learning_rate": 2.231132661942062e-06,
      "loss": 0.6687,
      "step": 13363
    },
    {
      "epoch": 0.5538563554229351,
      "grad_norm": 0.40095648169517517,
      "learning_rate": 2.2309254424136933e-06,
      "loss": 0.6782,
      "step": 13364
    },
    {
      "epoch": 0.5538977993286087,
      "grad_norm": 0.49406012892723083,
      "learning_rate": 2.230718222885325e-06,
      "loss": 0.7488,
      "step": 13365
    },
    {
      "epoch": 0.5539392432342825,
      "grad_norm": 0.3838699460029602,
      "learning_rate": 2.2305110033569565e-06,
      "loss": 0.6638,
      "step": 13366
    },
    {
      "epoch": 0.5539806871399561,
      "grad_norm": 0.4086756110191345,
      "learning_rate": 2.2303037838285883e-06,
      "loss": 0.6797,
      "step": 13367
    },
    {
      "epoch": 0.5540221310456297,
      "grad_norm": 0.39085254073143005,
      "learning_rate": 2.2300965643002197e-06,
      "loss": 0.6222,
      "step": 13368
    },
    {
      "epoch": 0.5540635749513034,
      "grad_norm": 0.41899600625038147,
      "learning_rate": 2.2298893447718515e-06,
      "loss": 0.6958,
      "step": 13369
    },
    {
      "epoch": 0.554105018856977,
      "grad_norm": 0.4285520315170288,
      "learning_rate": 2.229682125243483e-06,
      "loss": 0.6976,
      "step": 13370
    },
    {
      "epoch": 0.5541464627626508,
      "grad_norm": 0.39842110872268677,
      "learning_rate": 2.2294749057151147e-06,
      "loss": 0.6702,
      "step": 13371
    },
    {
      "epoch": 0.5541879066683244,
      "grad_norm": 0.38456428050994873,
      "learning_rate": 2.2292676861867465e-06,
      "loss": 0.6259,
      "step": 13372
    },
    {
      "epoch": 0.5542293505739981,
      "grad_norm": 0.3744107484817505,
      "learning_rate": 2.2290604666583783e-06,
      "loss": 0.6565,
      "step": 13373
    },
    {
      "epoch": 0.5542707944796718,
      "grad_norm": 0.4025529623031616,
      "learning_rate": 2.2288532471300097e-06,
      "loss": 0.7324,
      "step": 13374
    },
    {
      "epoch": 0.5543122383853455,
      "grad_norm": 0.4216817021369934,
      "learning_rate": 2.2286460276016415e-06,
      "loss": 0.7095,
      "step": 13375
    },
    {
      "epoch": 0.5543536822910191,
      "grad_norm": 0.3994043171405792,
      "learning_rate": 2.228438808073273e-06,
      "loss": 0.6736,
      "step": 13376
    },
    {
      "epoch": 0.5543951261966927,
      "grad_norm": 0.38003724813461304,
      "learning_rate": 2.2282315885449047e-06,
      "loss": 0.7078,
      "step": 13377
    },
    {
      "epoch": 0.5544365701023665,
      "grad_norm": 0.4104180634021759,
      "learning_rate": 2.228024369016536e-06,
      "loss": 0.6694,
      "step": 13378
    },
    {
      "epoch": 0.5544780140080401,
      "grad_norm": 0.43352511525154114,
      "learning_rate": 2.227817149488168e-06,
      "loss": 0.7068,
      "step": 13379
    },
    {
      "epoch": 0.5545194579137138,
      "grad_norm": 0.414272278547287,
      "learning_rate": 2.2276099299597993e-06,
      "loss": 0.6548,
      "step": 13380
    },
    {
      "epoch": 0.5545609018193874,
      "grad_norm": 0.40755319595336914,
      "learning_rate": 2.2274027104314315e-06,
      "loss": 0.7058,
      "step": 13381
    },
    {
      "epoch": 0.5546023457250612,
      "grad_norm": 0.4305060803890228,
      "learning_rate": 2.227195490903063e-06,
      "loss": 0.7056,
      "step": 13382
    },
    {
      "epoch": 0.5546437896307348,
      "grad_norm": 0.3956022262573242,
      "learning_rate": 2.2269882713746947e-06,
      "loss": 0.6588,
      "step": 13383
    },
    {
      "epoch": 0.5546852335364084,
      "grad_norm": 0.4256497025489807,
      "learning_rate": 2.226781051846326e-06,
      "loss": 0.649,
      "step": 13384
    },
    {
      "epoch": 0.5547266774420822,
      "grad_norm": 0.4398941695690155,
      "learning_rate": 2.226573832317958e-06,
      "loss": 0.719,
      "step": 13385
    },
    {
      "epoch": 0.5547681213477558,
      "grad_norm": 0.39232495427131653,
      "learning_rate": 2.2263666127895893e-06,
      "loss": 0.6964,
      "step": 13386
    },
    {
      "epoch": 0.5548095652534295,
      "grad_norm": 0.4180092215538025,
      "learning_rate": 2.226159393261221e-06,
      "loss": 0.6829,
      "step": 13387
    },
    {
      "epoch": 0.5548510091591031,
      "grad_norm": 0.4017854332923889,
      "learning_rate": 2.2259521737328525e-06,
      "loss": 0.6421,
      "step": 13388
    },
    {
      "epoch": 0.5548924530647769,
      "grad_norm": 0.3935394585132599,
      "learning_rate": 2.2257449542044843e-06,
      "loss": 0.6541,
      "step": 13389
    },
    {
      "epoch": 0.5549338969704505,
      "grad_norm": 0.4190226197242737,
      "learning_rate": 2.225537734676116e-06,
      "loss": 0.7004,
      "step": 13390
    },
    {
      "epoch": 0.5549753408761242,
      "grad_norm": 0.4204018712043762,
      "learning_rate": 2.225330515147748e-06,
      "loss": 0.6387,
      "step": 13391
    },
    {
      "epoch": 0.5550167847817978,
      "grad_norm": 0.42090368270874023,
      "learning_rate": 2.2251232956193793e-06,
      "loss": 0.6611,
      "step": 13392
    },
    {
      "epoch": 0.5550582286874715,
      "grad_norm": 0.4075533449649811,
      "learning_rate": 2.224916076091011e-06,
      "loss": 0.6379,
      "step": 13393
    },
    {
      "epoch": 0.5550996725931452,
      "grad_norm": 0.431539386510849,
      "learning_rate": 2.2247088565626425e-06,
      "loss": 0.6959,
      "step": 13394
    },
    {
      "epoch": 0.5551411164988188,
      "grad_norm": 0.4098464250564575,
      "learning_rate": 2.2245016370342743e-06,
      "loss": 0.6372,
      "step": 13395
    },
    {
      "epoch": 0.5551825604044925,
      "grad_norm": 0.4535987377166748,
      "learning_rate": 2.2242944175059057e-06,
      "loss": 0.7329,
      "step": 13396
    },
    {
      "epoch": 0.5552240043101662,
      "grad_norm": 0.4095035791397095,
      "learning_rate": 2.2240871979775375e-06,
      "loss": 0.6621,
      "step": 13397
    },
    {
      "epoch": 0.5552654482158399,
      "grad_norm": 0.38087335228919983,
      "learning_rate": 2.2238799784491693e-06,
      "loss": 0.6787,
      "step": 13398
    },
    {
      "epoch": 0.5553068921215135,
      "grad_norm": 0.42278924584388733,
      "learning_rate": 2.223672758920801e-06,
      "loss": 0.7124,
      "step": 13399
    },
    {
      "epoch": 0.5553483360271873,
      "grad_norm": 0.425914466381073,
      "learning_rate": 2.2234655393924325e-06,
      "loss": 0.7185,
      "step": 13400
    },
    {
      "epoch": 0.5553897799328609,
      "grad_norm": 0.4198751151561737,
      "learning_rate": 2.2232583198640643e-06,
      "loss": 0.696,
      "step": 13401
    },
    {
      "epoch": 0.5554312238385345,
      "grad_norm": 0.4298326075077057,
      "learning_rate": 2.2230511003356957e-06,
      "loss": 0.6584,
      "step": 13402
    },
    {
      "epoch": 0.5554726677442082,
      "grad_norm": 0.4123218357563019,
      "learning_rate": 2.2228438808073275e-06,
      "loss": 0.7009,
      "step": 13403
    },
    {
      "epoch": 0.5555141116498818,
      "grad_norm": 0.4475235939025879,
      "learning_rate": 2.222636661278959e-06,
      "loss": 0.6807,
      "step": 13404
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.44484037160873413,
      "learning_rate": 2.2224294417505907e-06,
      "loss": 0.6478,
      "step": 13405
    },
    {
      "epoch": 0.5555969994612292,
      "grad_norm": 0.45554664731025696,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.7017,
      "step": 13406
    },
    {
      "epoch": 0.5556384433669029,
      "grad_norm": 0.4124221205711365,
      "learning_rate": 2.2220150026938543e-06,
      "loss": 0.6881,
      "step": 13407
    },
    {
      "epoch": 0.5556798872725766,
      "grad_norm": 0.4102628529071808,
      "learning_rate": 2.2218077831654857e-06,
      "loss": 0.7053,
      "step": 13408
    },
    {
      "epoch": 0.5557213311782503,
      "grad_norm": 0.4004441499710083,
      "learning_rate": 2.2216005636371175e-06,
      "loss": 0.6232,
      "step": 13409
    },
    {
      "epoch": 0.5557627750839239,
      "grad_norm": 0.3827805519104004,
      "learning_rate": 2.221393344108749e-06,
      "loss": 0.6565,
      "step": 13410
    },
    {
      "epoch": 0.5558042189895975,
      "grad_norm": 0.4330747127532959,
      "learning_rate": 2.2211861245803807e-06,
      "loss": 0.6655,
      "step": 13411
    },
    {
      "epoch": 0.5558456628952713,
      "grad_norm": 0.42457517981529236,
      "learning_rate": 2.220978905052012e-06,
      "loss": 0.7222,
      "step": 13412
    },
    {
      "epoch": 0.5558871068009449,
      "grad_norm": 0.42491310834884644,
      "learning_rate": 2.220771685523644e-06,
      "loss": 0.6335,
      "step": 13413
    },
    {
      "epoch": 0.5559285507066186,
      "grad_norm": 0.3926023244857788,
      "learning_rate": 2.2205644659952753e-06,
      "loss": 0.6913,
      "step": 13414
    },
    {
      "epoch": 0.5559699946122922,
      "grad_norm": 0.39806604385375977,
      "learning_rate": 2.220357246466907e-06,
      "loss": 0.6642,
      "step": 13415
    },
    {
      "epoch": 0.556011438517966,
      "grad_norm": 0.3912261426448822,
      "learning_rate": 2.220150026938539e-06,
      "loss": 0.694,
      "step": 13416
    },
    {
      "epoch": 0.5560528824236396,
      "grad_norm": 0.4065779447555542,
      "learning_rate": 2.2199428074101707e-06,
      "loss": 0.6438,
      "step": 13417
    },
    {
      "epoch": 0.5560943263293133,
      "grad_norm": 0.41474640369415283,
      "learning_rate": 2.219735587881802e-06,
      "loss": 0.6702,
      "step": 13418
    },
    {
      "epoch": 0.556135770234987,
      "grad_norm": 0.40799984335899353,
      "learning_rate": 2.219528368353434e-06,
      "loss": 0.6831,
      "step": 13419
    },
    {
      "epoch": 0.5561772141406606,
      "grad_norm": 0.386981725692749,
      "learning_rate": 2.2193211488250653e-06,
      "loss": 0.6487,
      "step": 13420
    },
    {
      "epoch": 0.5562186580463343,
      "grad_norm": 0.4154641330242157,
      "learning_rate": 2.219113929296697e-06,
      "loss": 0.647,
      "step": 13421
    },
    {
      "epoch": 0.5562601019520079,
      "grad_norm": 0.4300985038280487,
      "learning_rate": 2.2189067097683285e-06,
      "loss": 0.6721,
      "step": 13422
    },
    {
      "epoch": 0.5563015458576817,
      "grad_norm": 0.4128133952617645,
      "learning_rate": 2.2186994902399603e-06,
      "loss": 0.6978,
      "step": 13423
    },
    {
      "epoch": 0.5563429897633553,
      "grad_norm": 0.3875913619995117,
      "learning_rate": 2.218492270711592e-06,
      "loss": 0.6897,
      "step": 13424
    },
    {
      "epoch": 0.556384433669029,
      "grad_norm": 0.3978937566280365,
      "learning_rate": 2.218285051183224e-06,
      "loss": 0.7024,
      "step": 13425
    },
    {
      "epoch": 0.5564258775747026,
      "grad_norm": 0.4331439435482025,
      "learning_rate": 2.2180778316548553e-06,
      "loss": 0.7061,
      "step": 13426
    },
    {
      "epoch": 0.5564673214803764,
      "grad_norm": 0.4399000108242035,
      "learning_rate": 2.217870612126487e-06,
      "loss": 0.6729,
      "step": 13427
    },
    {
      "epoch": 0.55650876538605,
      "grad_norm": 0.40281373262405396,
      "learning_rate": 2.2176633925981185e-06,
      "loss": 0.6663,
      "step": 13428
    },
    {
      "epoch": 0.5565502092917236,
      "grad_norm": 0.3918147385120392,
      "learning_rate": 2.2174561730697503e-06,
      "loss": 0.6843,
      "step": 13429
    },
    {
      "epoch": 0.5565916531973973,
      "grad_norm": 0.43212953209877014,
      "learning_rate": 2.2172489535413817e-06,
      "loss": 0.6829,
      "step": 13430
    },
    {
      "epoch": 0.556633097103071,
      "grad_norm": 0.3771352767944336,
      "learning_rate": 2.2170417340130135e-06,
      "loss": 0.6494,
      "step": 13431
    },
    {
      "epoch": 0.5566745410087447,
      "grad_norm": 0.3982904255390167,
      "learning_rate": 2.2168345144846453e-06,
      "loss": 0.7012,
      "step": 13432
    },
    {
      "epoch": 0.5567159849144183,
      "grad_norm": 0.3984405994415283,
      "learning_rate": 2.216627294956277e-06,
      "loss": 0.6512,
      "step": 13433
    },
    {
      "epoch": 0.556757428820092,
      "grad_norm": 0.4121124744415283,
      "learning_rate": 2.2164200754279085e-06,
      "loss": 0.6899,
      "step": 13434
    },
    {
      "epoch": 0.5567988727257657,
      "grad_norm": 0.4117922782897949,
      "learning_rate": 2.2162128558995403e-06,
      "loss": 0.6898,
      "step": 13435
    },
    {
      "epoch": 0.5568403166314393,
      "grad_norm": 0.4423362612724304,
      "learning_rate": 2.2160056363711717e-06,
      "loss": 0.6992,
      "step": 13436
    },
    {
      "epoch": 0.556881760537113,
      "grad_norm": 0.4063716232776642,
      "learning_rate": 2.2157984168428035e-06,
      "loss": 0.6932,
      "step": 13437
    },
    {
      "epoch": 0.5569232044427866,
      "grad_norm": 0.40465348958969116,
      "learning_rate": 2.215591197314435e-06,
      "loss": 0.6439,
      "step": 13438
    },
    {
      "epoch": 0.5569646483484604,
      "grad_norm": 0.4145529270172119,
      "learning_rate": 2.2153839777860667e-06,
      "loss": 0.6752,
      "step": 13439
    },
    {
      "epoch": 0.557006092254134,
      "grad_norm": 0.49542340636253357,
      "learning_rate": 2.215176758257698e-06,
      "loss": 0.7131,
      "step": 13440
    },
    {
      "epoch": 0.5570475361598077,
      "grad_norm": 0.43253093957901,
      "learning_rate": 2.21496953872933e-06,
      "loss": 0.7047,
      "step": 13441
    },
    {
      "epoch": 0.5570889800654814,
      "grad_norm": 0.4001549184322357,
      "learning_rate": 2.2147623192009617e-06,
      "loss": 0.6738,
      "step": 13442
    },
    {
      "epoch": 0.5571304239711551,
      "grad_norm": 0.3856462836265564,
      "learning_rate": 2.2145550996725935e-06,
      "loss": 0.6427,
      "step": 13443
    },
    {
      "epoch": 0.5571718678768287,
      "grad_norm": 0.36404281854629517,
      "learning_rate": 2.214347880144225e-06,
      "loss": 0.6241,
      "step": 13444
    },
    {
      "epoch": 0.5572133117825023,
      "grad_norm": 0.39453092217445374,
      "learning_rate": 2.2141406606158567e-06,
      "loss": 0.7024,
      "step": 13445
    },
    {
      "epoch": 0.5572547556881761,
      "grad_norm": 0.41613492369651794,
      "learning_rate": 2.213933441087488e-06,
      "loss": 0.6711,
      "step": 13446
    },
    {
      "epoch": 0.5572961995938497,
      "grad_norm": 0.4494464695453644,
      "learning_rate": 2.21372622155912e-06,
      "loss": 0.7129,
      "step": 13447
    },
    {
      "epoch": 0.5573376434995234,
      "grad_norm": 0.4255913496017456,
      "learning_rate": 2.2135190020307513e-06,
      "loss": 0.6926,
      "step": 13448
    },
    {
      "epoch": 0.557379087405197,
      "grad_norm": 0.407270222902298,
      "learning_rate": 2.213311782502383e-06,
      "loss": 0.6613,
      "step": 13449
    },
    {
      "epoch": 0.5574205313108708,
      "grad_norm": 0.40119943022727966,
      "learning_rate": 2.213104562974015e-06,
      "loss": 0.678,
      "step": 13450
    },
    {
      "epoch": 0.5574619752165444,
      "grad_norm": 0.37802988290786743,
      "learning_rate": 2.2128973434456467e-06,
      "loss": 0.6285,
      "step": 13451
    },
    {
      "epoch": 0.5575034191222181,
      "grad_norm": 0.40623581409454346,
      "learning_rate": 2.212690123917278e-06,
      "loss": 0.6956,
      "step": 13452
    },
    {
      "epoch": 0.5575448630278917,
      "grad_norm": 0.44938820600509644,
      "learning_rate": 2.21248290438891e-06,
      "loss": 0.7327,
      "step": 13453
    },
    {
      "epoch": 0.5575863069335654,
      "grad_norm": 0.3923630118370056,
      "learning_rate": 2.2122756848605413e-06,
      "loss": 0.6724,
      "step": 13454
    },
    {
      "epoch": 0.5576277508392391,
      "grad_norm": 0.46525418758392334,
      "learning_rate": 2.212068465332173e-06,
      "loss": 0.7512,
      "step": 13455
    },
    {
      "epoch": 0.5576691947449127,
      "grad_norm": 0.3949797451496124,
      "learning_rate": 2.2118612458038045e-06,
      "loss": 0.6484,
      "step": 13456
    },
    {
      "epoch": 0.5577106386505865,
      "grad_norm": 0.4021584689617157,
      "learning_rate": 2.2116540262754363e-06,
      "loss": 0.7385,
      "step": 13457
    },
    {
      "epoch": 0.5577520825562601,
      "grad_norm": 0.4018990695476532,
      "learning_rate": 2.211446806747068e-06,
      "loss": 0.6512,
      "step": 13458
    },
    {
      "epoch": 0.5577935264619338,
      "grad_norm": 0.4004579484462738,
      "learning_rate": 2.2112395872187e-06,
      "loss": 0.7473,
      "step": 13459
    },
    {
      "epoch": 0.5578349703676074,
      "grad_norm": 0.40786340832710266,
      "learning_rate": 2.2110323676903313e-06,
      "loss": 0.7065,
      "step": 13460
    },
    {
      "epoch": 0.5578764142732812,
      "grad_norm": 0.4096389412879944,
      "learning_rate": 2.210825148161963e-06,
      "loss": 0.6504,
      "step": 13461
    },
    {
      "epoch": 0.5579178581789548,
      "grad_norm": 0.40300965309143066,
      "learning_rate": 2.2106179286335945e-06,
      "loss": 0.7253,
      "step": 13462
    },
    {
      "epoch": 0.5579593020846284,
      "grad_norm": 0.4301587641239166,
      "learning_rate": 2.2104107091052263e-06,
      "loss": 0.7092,
      "step": 13463
    },
    {
      "epoch": 0.5580007459903021,
      "grad_norm": 0.4605104625225067,
      "learning_rate": 2.2102034895768577e-06,
      "loss": 0.7021,
      "step": 13464
    },
    {
      "epoch": 0.5580421898959758,
      "grad_norm": 0.3844835162162781,
      "learning_rate": 2.2099962700484895e-06,
      "loss": 0.6652,
      "step": 13465
    },
    {
      "epoch": 0.5580836338016495,
      "grad_norm": 0.39506399631500244,
      "learning_rate": 2.2097890505201213e-06,
      "loss": 0.7017,
      "step": 13466
    },
    {
      "epoch": 0.5581250777073231,
      "grad_norm": 0.431597501039505,
      "learning_rate": 2.2095818309917527e-06,
      "loss": 0.6343,
      "step": 13467
    },
    {
      "epoch": 0.5581665216129968,
      "grad_norm": 0.4331507980823517,
      "learning_rate": 2.2093746114633845e-06,
      "loss": 0.7197,
      "step": 13468
    },
    {
      "epoch": 0.5582079655186705,
      "grad_norm": 0.41448405385017395,
      "learning_rate": 2.2091673919350163e-06,
      "loss": 0.6917,
      "step": 13469
    },
    {
      "epoch": 0.5582494094243442,
      "grad_norm": 0.42382121086120605,
      "learning_rate": 2.2089601724066477e-06,
      "loss": 0.6398,
      "step": 13470
    },
    {
      "epoch": 0.5582908533300178,
      "grad_norm": 0.390537828207016,
      "learning_rate": 2.2087529528782795e-06,
      "loss": 0.6421,
      "step": 13471
    },
    {
      "epoch": 0.5583322972356914,
      "grad_norm": 0.41863149404525757,
      "learning_rate": 2.208545733349911e-06,
      "loss": 0.6694,
      "step": 13472
    },
    {
      "epoch": 0.5583737411413652,
      "grad_norm": 0.44366341829299927,
      "learning_rate": 2.2083385138215427e-06,
      "loss": 0.6776,
      "step": 13473
    },
    {
      "epoch": 0.5584151850470388,
      "grad_norm": 0.41243746876716614,
      "learning_rate": 2.2081312942931745e-06,
      "loss": 0.7064,
      "step": 13474
    },
    {
      "epoch": 0.5584566289527125,
      "grad_norm": 0.3989080488681793,
      "learning_rate": 2.207924074764806e-06,
      "loss": 0.6655,
      "step": 13475
    },
    {
      "epoch": 0.5584980728583862,
      "grad_norm": 0.43861666321754456,
      "learning_rate": 2.2077168552364377e-06,
      "loss": 0.6903,
      "step": 13476
    },
    {
      "epoch": 0.5585395167640599,
      "grad_norm": 0.39581409096717834,
      "learning_rate": 2.2075096357080695e-06,
      "loss": 0.671,
      "step": 13477
    },
    {
      "epoch": 0.5585809606697335,
      "grad_norm": 0.40590617060661316,
      "learning_rate": 2.207302416179701e-06,
      "loss": 0.6526,
      "step": 13478
    },
    {
      "epoch": 0.5586224045754072,
      "grad_norm": 0.41925570368766785,
      "learning_rate": 2.2070951966513327e-06,
      "loss": 0.6869,
      "step": 13479
    },
    {
      "epoch": 0.5586638484810809,
      "grad_norm": 0.40661245584487915,
      "learning_rate": 2.206887977122964e-06,
      "loss": 0.6802,
      "step": 13480
    },
    {
      "epoch": 0.5587052923867545,
      "grad_norm": 0.40267014503479004,
      "learning_rate": 2.206680757594596e-06,
      "loss": 0.6696,
      "step": 13481
    },
    {
      "epoch": 0.5587467362924282,
      "grad_norm": 0.4519302248954773,
      "learning_rate": 2.2064735380662273e-06,
      "loss": 0.717,
      "step": 13482
    },
    {
      "epoch": 0.5587881801981018,
      "grad_norm": 0.3907933235168457,
      "learning_rate": 2.206266318537859e-06,
      "loss": 0.6992,
      "step": 13483
    },
    {
      "epoch": 0.5588296241037756,
      "grad_norm": 0.4434550106525421,
      "learning_rate": 2.206059099009491e-06,
      "loss": 0.6979,
      "step": 13484
    },
    {
      "epoch": 0.5588710680094492,
      "grad_norm": 0.4492528438568115,
      "learning_rate": 2.2058518794811227e-06,
      "loss": 0.7117,
      "step": 13485
    },
    {
      "epoch": 0.5589125119151229,
      "grad_norm": 0.44632333517074585,
      "learning_rate": 2.205644659952754e-06,
      "loss": 0.7374,
      "step": 13486
    },
    {
      "epoch": 0.5589539558207965,
      "grad_norm": 0.39630189538002014,
      "learning_rate": 2.205437440424386e-06,
      "loss": 0.6914,
      "step": 13487
    },
    {
      "epoch": 0.5589953997264703,
      "grad_norm": 0.426482230424881,
      "learning_rate": 2.2052302208960173e-06,
      "loss": 0.6332,
      "step": 13488
    },
    {
      "epoch": 0.5590368436321439,
      "grad_norm": 0.4317478835582733,
      "learning_rate": 2.205023001367649e-06,
      "loss": 0.7627,
      "step": 13489
    },
    {
      "epoch": 0.5590782875378175,
      "grad_norm": 0.4025958180427551,
      "learning_rate": 2.2048157818392805e-06,
      "loss": 0.6543,
      "step": 13490
    },
    {
      "epoch": 0.5591197314434913,
      "grad_norm": 0.4113334119319916,
      "learning_rate": 2.2046085623109123e-06,
      "loss": 0.6853,
      "step": 13491
    },
    {
      "epoch": 0.5591611753491649,
      "grad_norm": 0.3804090917110443,
      "learning_rate": 2.204401342782544e-06,
      "loss": 0.6903,
      "step": 13492
    },
    {
      "epoch": 0.5592026192548386,
      "grad_norm": 0.42987972497940063,
      "learning_rate": 2.2041941232541755e-06,
      "loss": 0.7095,
      "step": 13493
    },
    {
      "epoch": 0.5592440631605122,
      "grad_norm": 0.39299386739730835,
      "learning_rate": 2.2039869037258073e-06,
      "loss": 0.6493,
      "step": 13494
    },
    {
      "epoch": 0.559285507066186,
      "grad_norm": 0.39975452423095703,
      "learning_rate": 2.203779684197439e-06,
      "loss": 0.6755,
      "step": 13495
    },
    {
      "epoch": 0.5593269509718596,
      "grad_norm": 0.4169398844242096,
      "learning_rate": 2.2035724646690705e-06,
      "loss": 0.6431,
      "step": 13496
    },
    {
      "epoch": 0.5593683948775332,
      "grad_norm": 0.3925769329071045,
      "learning_rate": 2.2033652451407023e-06,
      "loss": 0.6604,
      "step": 13497
    },
    {
      "epoch": 0.5594098387832069,
      "grad_norm": 0.3608100712299347,
      "learning_rate": 2.2031580256123337e-06,
      "loss": 0.6648,
      "step": 13498
    },
    {
      "epoch": 0.5594512826888806,
      "grad_norm": 0.415886789560318,
      "learning_rate": 2.2029508060839655e-06,
      "loss": 0.6871,
      "step": 13499
    },
    {
      "epoch": 0.5594927265945543,
      "grad_norm": 0.4178953170776367,
      "learning_rate": 2.2027435865555973e-06,
      "loss": 0.7006,
      "step": 13500
    },
    {
      "epoch": 0.5595341705002279,
      "grad_norm": 0.39562898874282837,
      "learning_rate": 2.2025363670272287e-06,
      "loss": 0.6331,
      "step": 13501
    },
    {
      "epoch": 0.5595756144059016,
      "grad_norm": 0.4548182785511017,
      "learning_rate": 2.2023291474988605e-06,
      "loss": 0.7021,
      "step": 13502
    },
    {
      "epoch": 0.5596170583115753,
      "grad_norm": 0.42953452467918396,
      "learning_rate": 2.2021219279704923e-06,
      "loss": 0.739,
      "step": 13503
    },
    {
      "epoch": 0.559658502217249,
      "grad_norm": 0.3848244547843933,
      "learning_rate": 2.2019147084421237e-06,
      "loss": 0.6604,
      "step": 13504
    },
    {
      "epoch": 0.5596999461229226,
      "grad_norm": 0.45081838965415955,
      "learning_rate": 2.2017074889137555e-06,
      "loss": 0.6697,
      "step": 13505
    },
    {
      "epoch": 0.5597413900285962,
      "grad_norm": 0.38033729791641235,
      "learning_rate": 2.201500269385387e-06,
      "loss": 0.6462,
      "step": 13506
    },
    {
      "epoch": 0.55978283393427,
      "grad_norm": 0.42892688512802124,
      "learning_rate": 2.2012930498570187e-06,
      "loss": 0.6722,
      "step": 13507
    },
    {
      "epoch": 0.5598242778399436,
      "grad_norm": 0.3980858623981476,
      "learning_rate": 2.2010858303286505e-06,
      "loss": 0.7052,
      "step": 13508
    },
    {
      "epoch": 0.5598657217456173,
      "grad_norm": 0.42280325293540955,
      "learning_rate": 2.200878610800282e-06,
      "loss": 0.7065,
      "step": 13509
    },
    {
      "epoch": 0.559907165651291,
      "grad_norm": 0.38851016759872437,
      "learning_rate": 2.2006713912719137e-06,
      "loss": 0.7131,
      "step": 13510
    },
    {
      "epoch": 0.5599486095569647,
      "grad_norm": 0.43656688928604126,
      "learning_rate": 2.2004641717435455e-06,
      "loss": 0.6797,
      "step": 13511
    },
    {
      "epoch": 0.5599900534626383,
      "grad_norm": 0.41167059540748596,
      "learning_rate": 2.200256952215177e-06,
      "loss": 0.6598,
      "step": 13512
    },
    {
      "epoch": 0.560031497368312,
      "grad_norm": 0.48567354679107666,
      "learning_rate": 2.2000497326868087e-06,
      "loss": 0.6899,
      "step": 13513
    },
    {
      "epoch": 0.5600729412739857,
      "grad_norm": 0.4317682087421417,
      "learning_rate": 2.19984251315844e-06,
      "loss": 0.6748,
      "step": 13514
    },
    {
      "epoch": 0.5601143851796593,
      "grad_norm": 0.47222238779067993,
      "learning_rate": 2.199635293630072e-06,
      "loss": 0.6973,
      "step": 13515
    },
    {
      "epoch": 0.560155829085333,
      "grad_norm": 0.4293147921562195,
      "learning_rate": 2.1994280741017033e-06,
      "loss": 0.6655,
      "step": 13516
    },
    {
      "epoch": 0.5601972729910066,
      "grad_norm": 0.404299259185791,
      "learning_rate": 2.199220854573335e-06,
      "loss": 0.6638,
      "step": 13517
    },
    {
      "epoch": 0.5602387168966804,
      "grad_norm": 0.399644672870636,
      "learning_rate": 2.199013635044967e-06,
      "loss": 0.6677,
      "step": 13518
    },
    {
      "epoch": 0.560280160802354,
      "grad_norm": 0.4088263213634491,
      "learning_rate": 2.1988064155165983e-06,
      "loss": 0.7126,
      "step": 13519
    },
    {
      "epoch": 0.5603216047080277,
      "grad_norm": 0.38518163561820984,
      "learning_rate": 2.19859919598823e-06,
      "loss": 0.6458,
      "step": 13520
    },
    {
      "epoch": 0.5603630486137013,
      "grad_norm": 0.4290962219238281,
      "learning_rate": 2.198391976459862e-06,
      "loss": 0.6698,
      "step": 13521
    },
    {
      "epoch": 0.5604044925193751,
      "grad_norm": 0.44641464948654175,
      "learning_rate": 2.1981847569314933e-06,
      "loss": 0.71,
      "step": 13522
    },
    {
      "epoch": 0.5604459364250487,
      "grad_norm": 0.40030407905578613,
      "learning_rate": 2.197977537403125e-06,
      "loss": 0.6549,
      "step": 13523
    },
    {
      "epoch": 0.5604873803307223,
      "grad_norm": 0.43866628408432007,
      "learning_rate": 2.1977703178747565e-06,
      "loss": 0.7046,
      "step": 13524
    },
    {
      "epoch": 0.560528824236396,
      "grad_norm": 0.3886450231075287,
      "learning_rate": 2.1975630983463883e-06,
      "loss": 0.6985,
      "step": 13525
    },
    {
      "epoch": 0.5605702681420697,
      "grad_norm": 0.3982860743999481,
      "learning_rate": 2.19735587881802e-06,
      "loss": 0.6184,
      "step": 13526
    },
    {
      "epoch": 0.5606117120477434,
      "grad_norm": 0.40546050667762756,
      "learning_rate": 2.1971486592896515e-06,
      "loss": 0.6575,
      "step": 13527
    },
    {
      "epoch": 0.560653155953417,
      "grad_norm": 0.43203985691070557,
      "learning_rate": 2.1969414397612833e-06,
      "loss": 0.6145,
      "step": 13528
    },
    {
      "epoch": 0.5606945998590908,
      "grad_norm": 0.45873114466667175,
      "learning_rate": 2.196734220232915e-06,
      "loss": 0.7437,
      "step": 13529
    },
    {
      "epoch": 0.5607360437647644,
      "grad_norm": 0.3934037387371063,
      "learning_rate": 2.1965270007045465e-06,
      "loss": 0.6167,
      "step": 13530
    },
    {
      "epoch": 0.5607774876704381,
      "grad_norm": 0.42633056640625,
      "learning_rate": 2.1963197811761783e-06,
      "loss": 0.7183,
      "step": 13531
    },
    {
      "epoch": 0.5608189315761117,
      "grad_norm": 0.37371826171875,
      "learning_rate": 2.1961125616478097e-06,
      "loss": 0.6919,
      "step": 13532
    },
    {
      "epoch": 0.5608603754817854,
      "grad_norm": 0.40359827876091003,
      "learning_rate": 2.1959053421194415e-06,
      "loss": 0.712,
      "step": 13533
    },
    {
      "epoch": 0.5609018193874591,
      "grad_norm": 0.3969515264034271,
      "learning_rate": 2.1956981225910733e-06,
      "loss": 0.6078,
      "step": 13534
    },
    {
      "epoch": 0.5609432632931327,
      "grad_norm": 0.39733344316482544,
      "learning_rate": 2.1954909030627047e-06,
      "loss": 0.6575,
      "step": 13535
    },
    {
      "epoch": 0.5609847071988064,
      "grad_norm": 0.43039190769195557,
      "learning_rate": 2.1952836835343365e-06,
      "loss": 0.6627,
      "step": 13536
    },
    {
      "epoch": 0.5610261511044801,
      "grad_norm": 0.42951494455337524,
      "learning_rate": 2.195076464005968e-06,
      "loss": 0.7037,
      "step": 13537
    },
    {
      "epoch": 0.5610675950101538,
      "grad_norm": 0.40532562136650085,
      "learning_rate": 2.1948692444775997e-06,
      "loss": 0.7158,
      "step": 13538
    },
    {
      "epoch": 0.5611090389158274,
      "grad_norm": 0.4063150882720947,
      "learning_rate": 2.1946620249492315e-06,
      "loss": 0.6858,
      "step": 13539
    },
    {
      "epoch": 0.5611504828215012,
      "grad_norm": 0.38967394828796387,
      "learning_rate": 2.194454805420863e-06,
      "loss": 0.6356,
      "step": 13540
    },
    {
      "epoch": 0.5611919267271748,
      "grad_norm": 0.3996219336986542,
      "learning_rate": 2.1942475858924947e-06,
      "loss": 0.728,
      "step": 13541
    },
    {
      "epoch": 0.5612333706328484,
      "grad_norm": 0.37087124586105347,
      "learning_rate": 2.1940403663641265e-06,
      "loss": 0.7058,
      "step": 13542
    },
    {
      "epoch": 0.5612748145385221,
      "grad_norm": 0.41605815291404724,
      "learning_rate": 2.193833146835758e-06,
      "loss": 0.6902,
      "step": 13543
    },
    {
      "epoch": 0.5613162584441957,
      "grad_norm": 0.41409724950790405,
      "learning_rate": 2.1936259273073897e-06,
      "loss": 0.6687,
      "step": 13544
    },
    {
      "epoch": 0.5613577023498695,
      "grad_norm": 0.39960646629333496,
      "learning_rate": 2.193418707779021e-06,
      "loss": 0.6833,
      "step": 13545
    },
    {
      "epoch": 0.5613991462555431,
      "grad_norm": 0.4606805443763733,
      "learning_rate": 2.193211488250653e-06,
      "loss": 0.7052,
      "step": 13546
    },
    {
      "epoch": 0.5614405901612168,
      "grad_norm": 0.4342947006225586,
      "learning_rate": 2.1930042687222847e-06,
      "loss": 0.6829,
      "step": 13547
    },
    {
      "epoch": 0.5614820340668905,
      "grad_norm": 0.39312323927879333,
      "learning_rate": 2.192797049193916e-06,
      "loss": 0.6843,
      "step": 13548
    },
    {
      "epoch": 0.5615234779725642,
      "grad_norm": 0.4197835326194763,
      "learning_rate": 2.192589829665548e-06,
      "loss": 0.7122,
      "step": 13549
    },
    {
      "epoch": 0.5615649218782378,
      "grad_norm": 0.4010487496852875,
      "learning_rate": 2.1923826101371793e-06,
      "loss": 0.6394,
      "step": 13550
    },
    {
      "epoch": 0.5616063657839114,
      "grad_norm": 0.38023194670677185,
      "learning_rate": 2.192175390608811e-06,
      "loss": 0.6494,
      "step": 13551
    },
    {
      "epoch": 0.5616478096895852,
      "grad_norm": 0.3523508906364441,
      "learning_rate": 2.191968171080443e-06,
      "loss": 0.6522,
      "step": 13552
    },
    {
      "epoch": 0.5616892535952588,
      "grad_norm": 0.40138936042785645,
      "learning_rate": 2.1917609515520743e-06,
      "loss": 0.6376,
      "step": 13553
    },
    {
      "epoch": 0.5617306975009325,
      "grad_norm": 0.41854575276374817,
      "learning_rate": 2.191553732023706e-06,
      "loss": 0.6841,
      "step": 13554
    },
    {
      "epoch": 0.5617721414066061,
      "grad_norm": 0.3792282044887543,
      "learning_rate": 2.191346512495338e-06,
      "loss": 0.686,
      "step": 13555
    },
    {
      "epoch": 0.5618135853122799,
      "grad_norm": 0.4282967448234558,
      "learning_rate": 2.1911392929669693e-06,
      "loss": 0.7009,
      "step": 13556
    },
    {
      "epoch": 0.5618550292179535,
      "grad_norm": 0.41302335262298584,
      "learning_rate": 2.190932073438601e-06,
      "loss": 0.691,
      "step": 13557
    },
    {
      "epoch": 0.5618964731236271,
      "grad_norm": 0.39084434509277344,
      "learning_rate": 2.1907248539102325e-06,
      "loss": 0.6556,
      "step": 13558
    },
    {
      "epoch": 0.5619379170293008,
      "grad_norm": 0.4302116632461548,
      "learning_rate": 2.1905176343818643e-06,
      "loss": 0.7017,
      "step": 13559
    },
    {
      "epoch": 0.5619793609349745,
      "grad_norm": 0.4295697510242462,
      "learning_rate": 2.190310414853496e-06,
      "loss": 0.6951,
      "step": 13560
    },
    {
      "epoch": 0.5620208048406482,
      "grad_norm": 0.4126332402229309,
      "learning_rate": 2.1901031953251275e-06,
      "loss": 0.6764,
      "step": 13561
    },
    {
      "epoch": 0.5620622487463218,
      "grad_norm": 0.38885870575904846,
      "learning_rate": 2.1898959757967593e-06,
      "loss": 0.7075,
      "step": 13562
    },
    {
      "epoch": 0.5621036926519956,
      "grad_norm": 0.46495676040649414,
      "learning_rate": 2.1896887562683907e-06,
      "loss": 0.7583,
      "step": 13563
    },
    {
      "epoch": 0.5621451365576692,
      "grad_norm": 0.4231015741825104,
      "learning_rate": 2.1894815367400225e-06,
      "loss": 0.6581,
      "step": 13564
    },
    {
      "epoch": 0.5621865804633429,
      "grad_norm": 0.39117616415023804,
      "learning_rate": 2.1892743172116543e-06,
      "loss": 0.6722,
      "step": 13565
    },
    {
      "epoch": 0.5622280243690165,
      "grad_norm": 0.42484748363494873,
      "learning_rate": 2.1890670976832857e-06,
      "loss": 0.6843,
      "step": 13566
    },
    {
      "epoch": 0.5622694682746902,
      "grad_norm": 0.4442344009876251,
      "learning_rate": 2.1888598781549175e-06,
      "loss": 0.6982,
      "step": 13567
    },
    {
      "epoch": 0.5623109121803639,
      "grad_norm": 0.4921358525753021,
      "learning_rate": 2.1886526586265493e-06,
      "loss": 0.7661,
      "step": 13568
    },
    {
      "epoch": 0.5623523560860375,
      "grad_norm": 0.4076802134513855,
      "learning_rate": 2.1884454390981807e-06,
      "loss": 0.6641,
      "step": 13569
    },
    {
      "epoch": 0.5623937999917112,
      "grad_norm": 0.38109737634658813,
      "learning_rate": 2.1882382195698125e-06,
      "loss": 0.6285,
      "step": 13570
    },
    {
      "epoch": 0.5624352438973849,
      "grad_norm": 0.4086001217365265,
      "learning_rate": 2.188031000041444e-06,
      "loss": 0.6786,
      "step": 13571
    },
    {
      "epoch": 0.5624766878030586,
      "grad_norm": 0.39237698912620544,
      "learning_rate": 2.1878237805130757e-06,
      "loss": 0.6426,
      "step": 13572
    },
    {
      "epoch": 0.5625181317087322,
      "grad_norm": 0.43000200390815735,
      "learning_rate": 2.1876165609847075e-06,
      "loss": 0.7117,
      "step": 13573
    },
    {
      "epoch": 0.562559575614406,
      "grad_norm": 0.3930050730705261,
      "learning_rate": 2.187409341456339e-06,
      "loss": 0.7026,
      "step": 13574
    },
    {
      "epoch": 0.5626010195200796,
      "grad_norm": 0.42346909642219543,
      "learning_rate": 2.1872021219279707e-06,
      "loss": 0.7075,
      "step": 13575
    },
    {
      "epoch": 0.5626424634257532,
      "grad_norm": 0.3909718990325928,
      "learning_rate": 2.1869949023996025e-06,
      "loss": 0.6968,
      "step": 13576
    },
    {
      "epoch": 0.5626839073314269,
      "grad_norm": 0.43160635232925415,
      "learning_rate": 2.186787682871234e-06,
      "loss": 0.6934,
      "step": 13577
    },
    {
      "epoch": 0.5627253512371005,
      "grad_norm": 0.4162079989910126,
      "learning_rate": 2.1865804633428657e-06,
      "loss": 0.7036,
      "step": 13578
    },
    {
      "epoch": 0.5627667951427743,
      "grad_norm": 0.4165841042995453,
      "learning_rate": 2.186373243814497e-06,
      "loss": 0.6854,
      "step": 13579
    },
    {
      "epoch": 0.5628082390484479,
      "grad_norm": 0.40297171473503113,
      "learning_rate": 2.186166024286129e-06,
      "loss": 0.6818,
      "step": 13580
    },
    {
      "epoch": 0.5628496829541216,
      "grad_norm": 0.44012030959129333,
      "learning_rate": 2.1859588047577607e-06,
      "loss": 0.696,
      "step": 13581
    },
    {
      "epoch": 0.5628911268597953,
      "grad_norm": 0.40899658203125,
      "learning_rate": 2.185751585229392e-06,
      "loss": 0.6912,
      "step": 13582
    },
    {
      "epoch": 0.562932570765469,
      "grad_norm": 0.438486784696579,
      "learning_rate": 2.185544365701024e-06,
      "loss": 0.6721,
      "step": 13583
    },
    {
      "epoch": 0.5629740146711426,
      "grad_norm": 0.4379560351371765,
      "learning_rate": 2.1853371461726557e-06,
      "loss": 0.6831,
      "step": 13584
    },
    {
      "epoch": 0.5630154585768162,
      "grad_norm": 0.43165647983551025,
      "learning_rate": 2.185129926644287e-06,
      "loss": 0.7225,
      "step": 13585
    },
    {
      "epoch": 0.56305690248249,
      "grad_norm": 0.43417444825172424,
      "learning_rate": 2.184922707115919e-06,
      "loss": 0.7329,
      "step": 13586
    },
    {
      "epoch": 0.5630983463881636,
      "grad_norm": 0.41414275765419006,
      "learning_rate": 2.1847154875875503e-06,
      "loss": 0.6658,
      "step": 13587
    },
    {
      "epoch": 0.5631397902938373,
      "grad_norm": 0.36921992897987366,
      "learning_rate": 2.184508268059182e-06,
      "loss": 0.6631,
      "step": 13588
    },
    {
      "epoch": 0.5631812341995109,
      "grad_norm": 0.4376254379749298,
      "learning_rate": 2.1843010485308135e-06,
      "loss": 0.7074,
      "step": 13589
    },
    {
      "epoch": 0.5632226781051847,
      "grad_norm": 0.3795234262943268,
      "learning_rate": 2.1840938290024453e-06,
      "loss": 0.6577,
      "step": 13590
    },
    {
      "epoch": 0.5632641220108583,
      "grad_norm": 0.41531500220298767,
      "learning_rate": 2.183886609474077e-06,
      "loss": 0.7317,
      "step": 13591
    },
    {
      "epoch": 0.563305565916532,
      "grad_norm": 0.4200824499130249,
      "learning_rate": 2.1836793899457085e-06,
      "loss": 0.6783,
      "step": 13592
    },
    {
      "epoch": 0.5633470098222056,
      "grad_norm": 0.4127451777458191,
      "learning_rate": 2.1834721704173403e-06,
      "loss": 0.6967,
      "step": 13593
    },
    {
      "epoch": 0.5633884537278793,
      "grad_norm": 0.432097464799881,
      "learning_rate": 2.183264950888972e-06,
      "loss": 0.697,
      "step": 13594
    },
    {
      "epoch": 0.563429897633553,
      "grad_norm": 0.419119268655777,
      "learning_rate": 2.1830577313606035e-06,
      "loss": 0.6628,
      "step": 13595
    },
    {
      "epoch": 0.5634713415392266,
      "grad_norm": 0.46469399333000183,
      "learning_rate": 2.1828505118322353e-06,
      "loss": 0.7124,
      "step": 13596
    },
    {
      "epoch": 0.5635127854449004,
      "grad_norm": 0.42279893159866333,
      "learning_rate": 2.1826432923038667e-06,
      "loss": 0.7249,
      "step": 13597
    },
    {
      "epoch": 0.563554229350574,
      "grad_norm": 0.42545968294143677,
      "learning_rate": 2.1824360727754985e-06,
      "loss": 0.739,
      "step": 13598
    },
    {
      "epoch": 0.5635956732562477,
      "grad_norm": 0.417981892824173,
      "learning_rate": 2.1822288532471303e-06,
      "loss": 0.7002,
      "step": 13599
    },
    {
      "epoch": 0.5636371171619213,
      "grad_norm": 0.41238218545913696,
      "learning_rate": 2.1820216337187617e-06,
      "loss": 0.636,
      "step": 13600
    },
    {
      "epoch": 0.5636785610675951,
      "grad_norm": 0.41787609457969666,
      "learning_rate": 2.1818144141903935e-06,
      "loss": 0.6713,
      "step": 13601
    },
    {
      "epoch": 0.5637200049732687,
      "grad_norm": 0.4048405885696411,
      "learning_rate": 2.1816071946620253e-06,
      "loss": 0.6621,
      "step": 13602
    },
    {
      "epoch": 0.5637614488789423,
      "grad_norm": 0.3973867893218994,
      "learning_rate": 2.1813999751336567e-06,
      "loss": 0.6768,
      "step": 13603
    },
    {
      "epoch": 0.563802892784616,
      "grad_norm": 0.4192880094051361,
      "learning_rate": 2.1811927556052885e-06,
      "loss": 0.6827,
      "step": 13604
    },
    {
      "epoch": 0.5638443366902897,
      "grad_norm": 0.40680718421936035,
      "learning_rate": 2.18098553607692e-06,
      "loss": 0.6987,
      "step": 13605
    },
    {
      "epoch": 0.5638857805959634,
      "grad_norm": 0.3910078704357147,
      "learning_rate": 2.1807783165485517e-06,
      "loss": 0.6764,
      "step": 13606
    },
    {
      "epoch": 0.563927224501637,
      "grad_norm": 0.39003387093544006,
      "learning_rate": 2.1805710970201835e-06,
      "loss": 0.6926,
      "step": 13607
    },
    {
      "epoch": 0.5639686684073107,
      "grad_norm": 0.409464955329895,
      "learning_rate": 2.180363877491815e-06,
      "loss": 0.7141,
      "step": 13608
    },
    {
      "epoch": 0.5640101123129844,
      "grad_norm": 0.3894730806350708,
      "learning_rate": 2.1801566579634467e-06,
      "loss": 0.6536,
      "step": 13609
    },
    {
      "epoch": 0.5640515562186581,
      "grad_norm": 0.4156385362148285,
      "learning_rate": 2.1799494384350785e-06,
      "loss": 0.7189,
      "step": 13610
    },
    {
      "epoch": 0.5640930001243317,
      "grad_norm": 0.3780062198638916,
      "learning_rate": 2.17974221890671e-06,
      "loss": 0.6509,
      "step": 13611
    },
    {
      "epoch": 0.5641344440300053,
      "grad_norm": 0.3846317231655121,
      "learning_rate": 2.1795349993783417e-06,
      "loss": 0.6589,
      "step": 13612
    },
    {
      "epoch": 0.5641758879356791,
      "grad_norm": 0.45072177052497864,
      "learning_rate": 2.179327779849973e-06,
      "loss": 0.6807,
      "step": 13613
    },
    {
      "epoch": 0.5642173318413527,
      "grad_norm": 0.42618077993392944,
      "learning_rate": 2.179120560321605e-06,
      "loss": 0.7559,
      "step": 13614
    },
    {
      "epoch": 0.5642587757470264,
      "grad_norm": 0.42198190093040466,
      "learning_rate": 2.1789133407932363e-06,
      "loss": 0.7437,
      "step": 13615
    },
    {
      "epoch": 0.5643002196527,
      "grad_norm": 0.42946889996528625,
      "learning_rate": 2.178706121264868e-06,
      "loss": 0.6769,
      "step": 13616
    },
    {
      "epoch": 0.5643416635583738,
      "grad_norm": 0.41309380531311035,
      "learning_rate": 2.1784989017365e-06,
      "loss": 0.6541,
      "step": 13617
    },
    {
      "epoch": 0.5643831074640474,
      "grad_norm": 0.41217321157455444,
      "learning_rate": 2.1782916822081317e-06,
      "loss": 0.7053,
      "step": 13618
    },
    {
      "epoch": 0.564424551369721,
      "grad_norm": 0.4197849929332733,
      "learning_rate": 2.178084462679763e-06,
      "loss": 0.6619,
      "step": 13619
    },
    {
      "epoch": 0.5644659952753948,
      "grad_norm": 0.3985639214515686,
      "learning_rate": 2.177877243151395e-06,
      "loss": 0.6697,
      "step": 13620
    },
    {
      "epoch": 0.5645074391810684,
      "grad_norm": 0.5045851469039917,
      "learning_rate": 2.1776700236230263e-06,
      "loss": 0.7589,
      "step": 13621
    },
    {
      "epoch": 0.5645488830867421,
      "grad_norm": 0.3928382992744446,
      "learning_rate": 2.177462804094658e-06,
      "loss": 0.6602,
      "step": 13622
    },
    {
      "epoch": 0.5645903269924157,
      "grad_norm": 0.40125736594200134,
      "learning_rate": 2.1772555845662895e-06,
      "loss": 0.6973,
      "step": 13623
    },
    {
      "epoch": 0.5646317708980895,
      "grad_norm": 0.38337793946266174,
      "learning_rate": 2.1770483650379213e-06,
      "loss": 0.6484,
      "step": 13624
    },
    {
      "epoch": 0.5646732148037631,
      "grad_norm": 0.41787585616111755,
      "learning_rate": 2.176841145509553e-06,
      "loss": 0.6863,
      "step": 13625
    },
    {
      "epoch": 0.5647146587094368,
      "grad_norm": 0.4291447699069977,
      "learning_rate": 2.1766339259811845e-06,
      "loss": 0.653,
      "step": 13626
    },
    {
      "epoch": 0.5647561026151104,
      "grad_norm": 0.4179460108280182,
      "learning_rate": 2.1764267064528163e-06,
      "loss": 0.7097,
      "step": 13627
    },
    {
      "epoch": 0.5647975465207841,
      "grad_norm": 0.4421672821044922,
      "learning_rate": 2.176219486924448e-06,
      "loss": 0.6846,
      "step": 13628
    },
    {
      "epoch": 0.5648389904264578,
      "grad_norm": 0.43097344040870667,
      "learning_rate": 2.1760122673960795e-06,
      "loss": 0.6749,
      "step": 13629
    },
    {
      "epoch": 0.5648804343321314,
      "grad_norm": 0.3987502157688141,
      "learning_rate": 2.1758050478677113e-06,
      "loss": 0.6821,
      "step": 13630
    },
    {
      "epoch": 0.5649218782378052,
      "grad_norm": 0.41023147106170654,
      "learning_rate": 2.1755978283393427e-06,
      "loss": 0.6837,
      "step": 13631
    },
    {
      "epoch": 0.5649633221434788,
      "grad_norm": 0.4206581115722656,
      "learning_rate": 2.1753906088109745e-06,
      "loss": 0.72,
      "step": 13632
    },
    {
      "epoch": 0.5650047660491525,
      "grad_norm": 0.38721954822540283,
      "learning_rate": 2.1751833892826063e-06,
      "loss": 0.6873,
      "step": 13633
    },
    {
      "epoch": 0.5650462099548261,
      "grad_norm": 0.4084838926792145,
      "learning_rate": 2.1749761697542377e-06,
      "loss": 0.7052,
      "step": 13634
    },
    {
      "epoch": 0.5650876538604999,
      "grad_norm": 0.4202449321746826,
      "learning_rate": 2.1747689502258695e-06,
      "loss": 0.6937,
      "step": 13635
    },
    {
      "epoch": 0.5651290977661735,
      "grad_norm": 0.42408743500709534,
      "learning_rate": 2.1745617306975013e-06,
      "loss": 0.7532,
      "step": 13636
    },
    {
      "epoch": 0.5651705416718471,
      "grad_norm": 0.43342190980911255,
      "learning_rate": 2.1743545111691327e-06,
      "loss": 0.6487,
      "step": 13637
    },
    {
      "epoch": 0.5652119855775208,
      "grad_norm": 0.40214189887046814,
      "learning_rate": 2.1741472916407645e-06,
      "loss": 0.6683,
      "step": 13638
    },
    {
      "epoch": 0.5652534294831945,
      "grad_norm": 0.43389853835105896,
      "learning_rate": 2.173940072112396e-06,
      "loss": 0.6746,
      "step": 13639
    },
    {
      "epoch": 0.5652948733888682,
      "grad_norm": 0.40097397565841675,
      "learning_rate": 2.1737328525840277e-06,
      "loss": 0.6161,
      "step": 13640
    },
    {
      "epoch": 0.5653363172945418,
      "grad_norm": 0.42159661650657654,
      "learning_rate": 2.173525633055659e-06,
      "loss": 0.6465,
      "step": 13641
    },
    {
      "epoch": 0.5653777612002155,
      "grad_norm": 0.3988941013813019,
      "learning_rate": 2.173318413527291e-06,
      "loss": 0.6567,
      "step": 13642
    },
    {
      "epoch": 0.5654192051058892,
      "grad_norm": 0.4075339138507843,
      "learning_rate": 2.1731111939989227e-06,
      "loss": 0.6696,
      "step": 13643
    },
    {
      "epoch": 0.5654606490115629,
      "grad_norm": 0.4084129333496094,
      "learning_rate": 2.1729039744705545e-06,
      "loss": 0.6882,
      "step": 13644
    },
    {
      "epoch": 0.5655020929172365,
      "grad_norm": 0.4367200434207916,
      "learning_rate": 2.172696754942186e-06,
      "loss": 0.7271,
      "step": 13645
    },
    {
      "epoch": 0.5655435368229101,
      "grad_norm": 0.4443969428539276,
      "learning_rate": 2.1724895354138177e-06,
      "loss": 0.7247,
      "step": 13646
    },
    {
      "epoch": 0.5655849807285839,
      "grad_norm": 0.399844765663147,
      "learning_rate": 2.172282315885449e-06,
      "loss": 0.6094,
      "step": 13647
    },
    {
      "epoch": 0.5656264246342575,
      "grad_norm": 0.43936195969581604,
      "learning_rate": 2.172075096357081e-06,
      "loss": 0.6501,
      "step": 13648
    },
    {
      "epoch": 0.5656678685399312,
      "grad_norm": 0.44922569394111633,
      "learning_rate": 2.1718678768287123e-06,
      "loss": 0.77,
      "step": 13649
    },
    {
      "epoch": 0.5657093124456049,
      "grad_norm": 0.403138130903244,
      "learning_rate": 2.171660657300344e-06,
      "loss": 0.7223,
      "step": 13650
    },
    {
      "epoch": 0.5657507563512786,
      "grad_norm": 0.3982759416103363,
      "learning_rate": 2.171453437771976e-06,
      "loss": 0.6179,
      "step": 13651
    },
    {
      "epoch": 0.5657922002569522,
      "grad_norm": 0.47598057985305786,
      "learning_rate": 2.1712462182436077e-06,
      "loss": 0.7263,
      "step": 13652
    },
    {
      "epoch": 0.5658336441626259,
      "grad_norm": 0.40538671612739563,
      "learning_rate": 2.171038998715239e-06,
      "loss": 0.6686,
      "step": 13653
    },
    {
      "epoch": 0.5658750880682996,
      "grad_norm": 0.3769107162952423,
      "learning_rate": 2.170831779186871e-06,
      "loss": 0.666,
      "step": 13654
    },
    {
      "epoch": 0.5659165319739732,
      "grad_norm": 0.4285000264644623,
      "learning_rate": 2.1706245596585023e-06,
      "loss": 0.7588,
      "step": 13655
    },
    {
      "epoch": 0.5659579758796469,
      "grad_norm": 0.42035356163978577,
      "learning_rate": 2.170417340130134e-06,
      "loss": 0.7063,
      "step": 13656
    },
    {
      "epoch": 0.5659994197853205,
      "grad_norm": 0.409315288066864,
      "learning_rate": 2.1702101206017655e-06,
      "loss": 0.6619,
      "step": 13657
    },
    {
      "epoch": 0.5660408636909943,
      "grad_norm": 0.41570279002189636,
      "learning_rate": 2.1700029010733973e-06,
      "loss": 0.6914,
      "step": 13658
    },
    {
      "epoch": 0.5660823075966679,
      "grad_norm": 0.4277610182762146,
      "learning_rate": 2.1697956815450287e-06,
      "loss": 0.6931,
      "step": 13659
    },
    {
      "epoch": 0.5661237515023416,
      "grad_norm": 0.45485353469848633,
      "learning_rate": 2.169588462016661e-06,
      "loss": 0.7485,
      "step": 13660
    },
    {
      "epoch": 0.5661651954080152,
      "grad_norm": 0.41995564103126526,
      "learning_rate": 2.1693812424882923e-06,
      "loss": 0.6598,
      "step": 13661
    },
    {
      "epoch": 0.566206639313689,
      "grad_norm": 0.3717178404331207,
      "learning_rate": 2.169174022959924e-06,
      "loss": 0.6459,
      "step": 13662
    },
    {
      "epoch": 0.5662480832193626,
      "grad_norm": 0.45705389976501465,
      "learning_rate": 2.1689668034315555e-06,
      "loss": 0.7139,
      "step": 13663
    },
    {
      "epoch": 0.5662895271250362,
      "grad_norm": 0.39327868819236755,
      "learning_rate": 2.1687595839031873e-06,
      "loss": 0.6796,
      "step": 13664
    },
    {
      "epoch": 0.56633097103071,
      "grad_norm": 0.4053371250629425,
      "learning_rate": 2.1685523643748187e-06,
      "loss": 0.7068,
      "step": 13665
    },
    {
      "epoch": 0.5663724149363836,
      "grad_norm": 0.39311107993125916,
      "learning_rate": 2.1683451448464505e-06,
      "loss": 0.666,
      "step": 13666
    },
    {
      "epoch": 0.5664138588420573,
      "grad_norm": 0.4119820296764374,
      "learning_rate": 2.168137925318082e-06,
      "loss": 0.6135,
      "step": 13667
    },
    {
      "epoch": 0.5664553027477309,
      "grad_norm": 0.42161619663238525,
      "learning_rate": 2.1679307057897137e-06,
      "loss": 0.7018,
      "step": 13668
    },
    {
      "epoch": 0.5664967466534047,
      "grad_norm": 0.40414151549339294,
      "learning_rate": 2.1677234862613455e-06,
      "loss": 0.705,
      "step": 13669
    },
    {
      "epoch": 0.5665381905590783,
      "grad_norm": 0.41660094261169434,
      "learning_rate": 2.1675162667329773e-06,
      "loss": 0.698,
      "step": 13670
    },
    {
      "epoch": 0.566579634464752,
      "grad_norm": 0.4019080102443695,
      "learning_rate": 2.1673090472046087e-06,
      "loss": 0.6747,
      "step": 13671
    },
    {
      "epoch": 0.5666210783704256,
      "grad_norm": 0.42354291677474976,
      "learning_rate": 2.1671018276762405e-06,
      "loss": 0.6637,
      "step": 13672
    },
    {
      "epoch": 0.5666625222760993,
      "grad_norm": 0.39122623205184937,
      "learning_rate": 2.166894608147872e-06,
      "loss": 0.6497,
      "step": 13673
    },
    {
      "epoch": 0.566703966181773,
      "grad_norm": 0.3951494097709656,
      "learning_rate": 2.1666873886195037e-06,
      "loss": 0.6309,
      "step": 13674
    },
    {
      "epoch": 0.5667454100874466,
      "grad_norm": 0.41508620977401733,
      "learning_rate": 2.166480169091135e-06,
      "loss": 0.6573,
      "step": 13675
    },
    {
      "epoch": 0.5667868539931203,
      "grad_norm": 0.42675304412841797,
      "learning_rate": 2.166272949562767e-06,
      "loss": 0.7111,
      "step": 13676
    },
    {
      "epoch": 0.566828297898794,
      "grad_norm": 0.3958274722099304,
      "learning_rate": 2.1660657300343987e-06,
      "loss": 0.6514,
      "step": 13677
    },
    {
      "epoch": 0.5668697418044677,
      "grad_norm": 0.4230049252510071,
      "learning_rate": 2.1658585105060305e-06,
      "loss": 0.671,
      "step": 13678
    },
    {
      "epoch": 0.5669111857101413,
      "grad_norm": 0.3982155919075012,
      "learning_rate": 2.165651290977662e-06,
      "loss": 0.7126,
      "step": 13679
    },
    {
      "epoch": 0.5669526296158149,
      "grad_norm": 0.39908599853515625,
      "learning_rate": 2.1654440714492937e-06,
      "loss": 0.6863,
      "step": 13680
    },
    {
      "epoch": 0.5669940735214887,
      "grad_norm": 0.38187214732170105,
      "learning_rate": 2.165236851920925e-06,
      "loss": 0.6233,
      "step": 13681
    },
    {
      "epoch": 0.5670355174271623,
      "grad_norm": 0.4256102442741394,
      "learning_rate": 2.165029632392557e-06,
      "loss": 0.6489,
      "step": 13682
    },
    {
      "epoch": 0.567076961332836,
      "grad_norm": 0.40588271617889404,
      "learning_rate": 2.1648224128641883e-06,
      "loss": 0.6528,
      "step": 13683
    },
    {
      "epoch": 0.5671184052385096,
      "grad_norm": 0.459468275308609,
      "learning_rate": 2.16461519333582e-06,
      "loss": 0.7408,
      "step": 13684
    },
    {
      "epoch": 0.5671598491441834,
      "grad_norm": 0.40608805418014526,
      "learning_rate": 2.1644079738074515e-06,
      "loss": 0.7405,
      "step": 13685
    },
    {
      "epoch": 0.567201293049857,
      "grad_norm": 0.4137836694717407,
      "learning_rate": 2.1642007542790837e-06,
      "loss": 0.6968,
      "step": 13686
    },
    {
      "epoch": 0.5672427369555307,
      "grad_norm": 0.40074464678764343,
      "learning_rate": 2.163993534750715e-06,
      "loss": 0.6981,
      "step": 13687
    },
    {
      "epoch": 0.5672841808612044,
      "grad_norm": 0.4191500246524811,
      "learning_rate": 2.163786315222347e-06,
      "loss": 0.7041,
      "step": 13688
    },
    {
      "epoch": 0.567325624766878,
      "grad_norm": 0.39591097831726074,
      "learning_rate": 2.1635790956939783e-06,
      "loss": 0.6797,
      "step": 13689
    },
    {
      "epoch": 0.5673670686725517,
      "grad_norm": 0.42156359553337097,
      "learning_rate": 2.16337187616561e-06,
      "loss": 0.7107,
      "step": 13690
    },
    {
      "epoch": 0.5674085125782253,
      "grad_norm": 0.4297883212566376,
      "learning_rate": 2.1631646566372415e-06,
      "loss": 0.7062,
      "step": 13691
    },
    {
      "epoch": 0.5674499564838991,
      "grad_norm": 0.39979955554008484,
      "learning_rate": 2.1629574371088733e-06,
      "loss": 0.6594,
      "step": 13692
    },
    {
      "epoch": 0.5674914003895727,
      "grad_norm": 0.3831852376461029,
      "learning_rate": 2.1627502175805047e-06,
      "loss": 0.6766,
      "step": 13693
    },
    {
      "epoch": 0.5675328442952464,
      "grad_norm": 0.4017987549304962,
      "learning_rate": 2.162542998052137e-06,
      "loss": 0.6586,
      "step": 13694
    },
    {
      "epoch": 0.56757428820092,
      "grad_norm": 0.39799532294273376,
      "learning_rate": 2.1623357785237683e-06,
      "loss": 0.6877,
      "step": 13695
    },
    {
      "epoch": 0.5676157321065938,
      "grad_norm": 0.4323866665363312,
      "learning_rate": 2.1621285589954e-06,
      "loss": 0.6799,
      "step": 13696
    },
    {
      "epoch": 0.5676571760122674,
      "grad_norm": 0.45041754841804504,
      "learning_rate": 2.1619213394670315e-06,
      "loss": 0.7537,
      "step": 13697
    },
    {
      "epoch": 0.567698619917941,
      "grad_norm": 0.4311353862285614,
      "learning_rate": 2.1617141199386633e-06,
      "loss": 0.6831,
      "step": 13698
    },
    {
      "epoch": 0.5677400638236147,
      "grad_norm": 0.391366571187973,
      "learning_rate": 2.1615069004102947e-06,
      "loss": 0.6667,
      "step": 13699
    },
    {
      "epoch": 0.5677815077292884,
      "grad_norm": 0.41093045473098755,
      "learning_rate": 2.1612996808819265e-06,
      "loss": 0.6666,
      "step": 13700
    },
    {
      "epoch": 0.5678229516349621,
      "grad_norm": 0.40989986062049866,
      "learning_rate": 2.161092461353558e-06,
      "loss": 0.673,
      "step": 13701
    },
    {
      "epoch": 0.5678643955406357,
      "grad_norm": 0.41437962651252747,
      "learning_rate": 2.1608852418251897e-06,
      "loss": 0.6793,
      "step": 13702
    },
    {
      "epoch": 0.5679058394463095,
      "grad_norm": 0.426816463470459,
      "learning_rate": 2.1606780222968215e-06,
      "loss": 0.7134,
      "step": 13703
    },
    {
      "epoch": 0.5679472833519831,
      "grad_norm": 0.408489853143692,
      "learning_rate": 2.1604708027684534e-06,
      "loss": 0.7332,
      "step": 13704
    },
    {
      "epoch": 0.5679887272576568,
      "grad_norm": 0.39941567182540894,
      "learning_rate": 2.1602635832400847e-06,
      "loss": 0.6719,
      "step": 13705
    },
    {
      "epoch": 0.5680301711633304,
      "grad_norm": 0.4368651509284973,
      "learning_rate": 2.1600563637117165e-06,
      "loss": 0.6829,
      "step": 13706
    },
    {
      "epoch": 0.568071615069004,
      "grad_norm": 0.4045748710632324,
      "learning_rate": 2.159849144183348e-06,
      "loss": 0.7046,
      "step": 13707
    },
    {
      "epoch": 0.5681130589746778,
      "grad_norm": 0.39960777759552,
      "learning_rate": 2.1596419246549797e-06,
      "loss": 0.739,
      "step": 13708
    },
    {
      "epoch": 0.5681545028803514,
      "grad_norm": 0.4345571994781494,
      "learning_rate": 2.159434705126611e-06,
      "loss": 0.7012,
      "step": 13709
    },
    {
      "epoch": 0.5681959467860251,
      "grad_norm": 0.41364234685897827,
      "learning_rate": 2.159227485598243e-06,
      "loss": 0.6837,
      "step": 13710
    },
    {
      "epoch": 0.5682373906916988,
      "grad_norm": 0.4114598035812378,
      "learning_rate": 2.1590202660698743e-06,
      "loss": 0.7018,
      "step": 13711
    },
    {
      "epoch": 0.5682788345973725,
      "grad_norm": 0.40810200572013855,
      "learning_rate": 2.1588130465415066e-06,
      "loss": 0.6921,
      "step": 13712
    },
    {
      "epoch": 0.5683202785030461,
      "grad_norm": 0.41379567980766296,
      "learning_rate": 2.158605827013138e-06,
      "loss": 0.6821,
      "step": 13713
    },
    {
      "epoch": 0.5683617224087198,
      "grad_norm": 0.42374250292778015,
      "learning_rate": 2.1583986074847697e-06,
      "loss": 0.6943,
      "step": 13714
    },
    {
      "epoch": 0.5684031663143935,
      "grad_norm": 0.44670814275741577,
      "learning_rate": 2.158191387956401e-06,
      "loss": 0.6809,
      "step": 13715
    },
    {
      "epoch": 0.5684446102200671,
      "grad_norm": 0.40701714158058167,
      "learning_rate": 2.157984168428033e-06,
      "loss": 0.6671,
      "step": 13716
    },
    {
      "epoch": 0.5684860541257408,
      "grad_norm": 0.3853479325771332,
      "learning_rate": 2.1577769488996643e-06,
      "loss": 0.7104,
      "step": 13717
    },
    {
      "epoch": 0.5685274980314144,
      "grad_norm": 0.46662768721580505,
      "learning_rate": 2.157569729371296e-06,
      "loss": 0.6747,
      "step": 13718
    },
    {
      "epoch": 0.5685689419370882,
      "grad_norm": 0.39216598868370056,
      "learning_rate": 2.1573625098429275e-06,
      "loss": 0.6648,
      "step": 13719
    },
    {
      "epoch": 0.5686103858427618,
      "grad_norm": 0.4249568283557892,
      "learning_rate": 2.1571552903145598e-06,
      "loss": 0.7056,
      "step": 13720
    },
    {
      "epoch": 0.5686518297484355,
      "grad_norm": 0.4172024428844452,
      "learning_rate": 2.156948070786191e-06,
      "loss": 0.6733,
      "step": 13721
    },
    {
      "epoch": 0.5686932736541092,
      "grad_norm": 0.4098511338233948,
      "learning_rate": 2.156740851257823e-06,
      "loss": 0.6484,
      "step": 13722
    },
    {
      "epoch": 0.5687347175597829,
      "grad_norm": 0.3991549015045166,
      "learning_rate": 2.1565336317294543e-06,
      "loss": 0.6499,
      "step": 13723
    },
    {
      "epoch": 0.5687761614654565,
      "grad_norm": 0.4105137586593628,
      "learning_rate": 2.156326412201086e-06,
      "loss": 0.6633,
      "step": 13724
    },
    {
      "epoch": 0.5688176053711301,
      "grad_norm": 0.45932409167289734,
      "learning_rate": 2.1561191926727175e-06,
      "loss": 0.701,
      "step": 13725
    },
    {
      "epoch": 0.5688590492768039,
      "grad_norm": 0.41389814019203186,
      "learning_rate": 2.1559119731443493e-06,
      "loss": 0.655,
      "step": 13726
    },
    {
      "epoch": 0.5689004931824775,
      "grad_norm": 0.40950700640678406,
      "learning_rate": 2.1557047536159807e-06,
      "loss": 0.681,
      "step": 13727
    },
    {
      "epoch": 0.5689419370881512,
      "grad_norm": 0.4194731116294861,
      "learning_rate": 2.1554975340876125e-06,
      "loss": 0.7085,
      "step": 13728
    },
    {
      "epoch": 0.5689833809938248,
      "grad_norm": 0.4191286861896515,
      "learning_rate": 2.1552903145592443e-06,
      "loss": 0.6691,
      "step": 13729
    },
    {
      "epoch": 0.5690248248994986,
      "grad_norm": 0.4002556800842285,
      "learning_rate": 2.155083095030876e-06,
      "loss": 0.7159,
      "step": 13730
    },
    {
      "epoch": 0.5690662688051722,
      "grad_norm": 0.4023125469684601,
      "learning_rate": 2.1548758755025075e-06,
      "loss": 0.6554,
      "step": 13731
    },
    {
      "epoch": 0.5691077127108459,
      "grad_norm": 0.445223867893219,
      "learning_rate": 2.1546686559741393e-06,
      "loss": 0.6445,
      "step": 13732
    },
    {
      "epoch": 0.5691491566165195,
      "grad_norm": 0.4298916459083557,
      "learning_rate": 2.1544614364457707e-06,
      "loss": 0.7014,
      "step": 13733
    },
    {
      "epoch": 0.5691906005221932,
      "grad_norm": 0.42163944244384766,
      "learning_rate": 2.1542542169174025e-06,
      "loss": 0.6787,
      "step": 13734
    },
    {
      "epoch": 0.5692320444278669,
      "grad_norm": 0.41559505462646484,
      "learning_rate": 2.154046997389034e-06,
      "loss": 0.6796,
      "step": 13735
    },
    {
      "epoch": 0.5692734883335405,
      "grad_norm": 0.39828386902809143,
      "learning_rate": 2.1538397778606657e-06,
      "loss": 0.728,
      "step": 13736
    },
    {
      "epoch": 0.5693149322392143,
      "grad_norm": 0.3913366496562958,
      "learning_rate": 2.153632558332297e-06,
      "loss": 0.6708,
      "step": 13737
    },
    {
      "epoch": 0.5693563761448879,
      "grad_norm": 0.4238438904285431,
      "learning_rate": 2.1534253388039294e-06,
      "loss": 0.7235,
      "step": 13738
    },
    {
      "epoch": 0.5693978200505616,
      "grad_norm": 0.42004093527793884,
      "learning_rate": 2.1532181192755607e-06,
      "loss": 0.6587,
      "step": 13739
    },
    {
      "epoch": 0.5694392639562352,
      "grad_norm": 0.41418296098709106,
      "learning_rate": 2.1530108997471926e-06,
      "loss": 0.6498,
      "step": 13740
    },
    {
      "epoch": 0.5694807078619089,
      "grad_norm": 0.3728832006454468,
      "learning_rate": 2.152803680218824e-06,
      "loss": 0.5989,
      "step": 13741
    },
    {
      "epoch": 0.5695221517675826,
      "grad_norm": 0.37687769532203674,
      "learning_rate": 2.1525964606904557e-06,
      "loss": 0.6122,
      "step": 13742
    },
    {
      "epoch": 0.5695635956732562,
      "grad_norm": 0.4326122999191284,
      "learning_rate": 2.152389241162087e-06,
      "loss": 0.6514,
      "step": 13743
    },
    {
      "epoch": 0.5696050395789299,
      "grad_norm": 0.41060319542884827,
      "learning_rate": 2.152182021633719e-06,
      "loss": 0.6667,
      "step": 13744
    },
    {
      "epoch": 0.5696464834846036,
      "grad_norm": 0.38674196600914,
      "learning_rate": 2.1519748021053503e-06,
      "loss": 0.6929,
      "step": 13745
    },
    {
      "epoch": 0.5696879273902773,
      "grad_norm": 0.4441470801830292,
      "learning_rate": 2.151767582576982e-06,
      "loss": 0.6776,
      "step": 13746
    },
    {
      "epoch": 0.5697293712959509,
      "grad_norm": 0.38636910915374756,
      "learning_rate": 2.151560363048614e-06,
      "loss": 0.6777,
      "step": 13747
    },
    {
      "epoch": 0.5697708152016246,
      "grad_norm": 0.44262778759002686,
      "learning_rate": 2.1513531435202458e-06,
      "loss": 0.672,
      "step": 13748
    },
    {
      "epoch": 0.5698122591072983,
      "grad_norm": 0.4004232883453369,
      "learning_rate": 2.151145923991877e-06,
      "loss": 0.6592,
      "step": 13749
    },
    {
      "epoch": 0.5698537030129719,
      "grad_norm": 0.3744255304336548,
      "learning_rate": 2.150938704463509e-06,
      "loss": 0.6207,
      "step": 13750
    },
    {
      "epoch": 0.5698951469186456,
      "grad_norm": 0.3982994258403778,
      "learning_rate": 2.1507314849351403e-06,
      "loss": 0.6565,
      "step": 13751
    },
    {
      "epoch": 0.5699365908243192,
      "grad_norm": 0.3812541961669922,
      "learning_rate": 2.150524265406772e-06,
      "loss": 0.657,
      "step": 13752
    },
    {
      "epoch": 0.569978034729993,
      "grad_norm": 0.44724100828170776,
      "learning_rate": 2.1503170458784035e-06,
      "loss": 0.7054,
      "step": 13753
    },
    {
      "epoch": 0.5700194786356666,
      "grad_norm": 0.3762550354003906,
      "learning_rate": 2.1501098263500353e-06,
      "loss": 0.668,
      "step": 13754
    },
    {
      "epoch": 0.5700609225413403,
      "grad_norm": 0.3970949947834015,
      "learning_rate": 2.149902606821667e-06,
      "loss": 0.6481,
      "step": 13755
    },
    {
      "epoch": 0.570102366447014,
      "grad_norm": 0.47364190220832825,
      "learning_rate": 2.149695387293299e-06,
      "loss": 0.6887,
      "step": 13756
    },
    {
      "epoch": 0.5701438103526877,
      "grad_norm": 0.41700872778892517,
      "learning_rate": 2.1494881677649303e-06,
      "loss": 0.6436,
      "step": 13757
    },
    {
      "epoch": 0.5701852542583613,
      "grad_norm": 0.43467146158218384,
      "learning_rate": 2.149280948236562e-06,
      "loss": 0.678,
      "step": 13758
    },
    {
      "epoch": 0.5702266981640349,
      "grad_norm": 0.38637763261795044,
      "learning_rate": 2.1490737287081935e-06,
      "loss": 0.6899,
      "step": 13759
    },
    {
      "epoch": 0.5702681420697087,
      "grad_norm": 0.43201443552970886,
      "learning_rate": 2.1488665091798253e-06,
      "loss": 0.6729,
      "step": 13760
    },
    {
      "epoch": 0.5703095859753823,
      "grad_norm": 0.4037129282951355,
      "learning_rate": 2.1486592896514567e-06,
      "loss": 0.6642,
      "step": 13761
    },
    {
      "epoch": 0.570351029881056,
      "grad_norm": 0.41157811880111694,
      "learning_rate": 2.1484520701230885e-06,
      "loss": 0.6854,
      "step": 13762
    },
    {
      "epoch": 0.5703924737867296,
      "grad_norm": 0.44422221183776855,
      "learning_rate": 2.14824485059472e-06,
      "loss": 0.7046,
      "step": 13763
    },
    {
      "epoch": 0.5704339176924034,
      "grad_norm": 0.4180927574634552,
      "learning_rate": 2.148037631066352e-06,
      "loss": 0.6779,
      "step": 13764
    },
    {
      "epoch": 0.570475361598077,
      "grad_norm": 0.40046486258506775,
      "learning_rate": 2.1478304115379835e-06,
      "loss": 0.6335,
      "step": 13765
    },
    {
      "epoch": 0.5705168055037507,
      "grad_norm": 0.41191163659095764,
      "learning_rate": 2.1476231920096154e-06,
      "loss": 0.6509,
      "step": 13766
    },
    {
      "epoch": 0.5705582494094243,
      "grad_norm": 0.379047155380249,
      "learning_rate": 2.1474159724812467e-06,
      "loss": 0.7106,
      "step": 13767
    },
    {
      "epoch": 0.570599693315098,
      "grad_norm": 0.39221251010894775,
      "learning_rate": 2.1472087529528785e-06,
      "loss": 0.6707,
      "step": 13768
    },
    {
      "epoch": 0.5706411372207717,
      "grad_norm": 0.4121342599391937,
      "learning_rate": 2.14700153342451e-06,
      "loss": 0.6826,
      "step": 13769
    },
    {
      "epoch": 0.5706825811264453,
      "grad_norm": 0.41001054644584656,
      "learning_rate": 2.1467943138961417e-06,
      "loss": 0.6635,
      "step": 13770
    },
    {
      "epoch": 0.570724025032119,
      "grad_norm": 0.40033721923828125,
      "learning_rate": 2.146587094367773e-06,
      "loss": 0.6334,
      "step": 13771
    },
    {
      "epoch": 0.5707654689377927,
      "grad_norm": 0.4297398030757904,
      "learning_rate": 2.146379874839405e-06,
      "loss": 0.7141,
      "step": 13772
    },
    {
      "epoch": 0.5708069128434664,
      "grad_norm": 0.4179666042327881,
      "learning_rate": 2.1461726553110367e-06,
      "loss": 0.6857,
      "step": 13773
    },
    {
      "epoch": 0.57084835674914,
      "grad_norm": 0.41863933205604553,
      "learning_rate": 2.1459654357826686e-06,
      "loss": 0.6777,
      "step": 13774
    },
    {
      "epoch": 0.5708898006548138,
      "grad_norm": 0.39949366450309753,
      "learning_rate": 2.1457582162543e-06,
      "loss": 0.702,
      "step": 13775
    },
    {
      "epoch": 0.5709312445604874,
      "grad_norm": 0.4511047601699829,
      "learning_rate": 2.1455509967259318e-06,
      "loss": 0.7673,
      "step": 13776
    },
    {
      "epoch": 0.570972688466161,
      "grad_norm": 0.4202826917171478,
      "learning_rate": 2.145343777197563e-06,
      "loss": 0.7074,
      "step": 13777
    },
    {
      "epoch": 0.5710141323718347,
      "grad_norm": 0.38963907957077026,
      "learning_rate": 2.145136557669195e-06,
      "loss": 0.7131,
      "step": 13778
    },
    {
      "epoch": 0.5710555762775084,
      "grad_norm": 0.400794118642807,
      "learning_rate": 2.1449293381408263e-06,
      "loss": 0.6281,
      "step": 13779
    },
    {
      "epoch": 0.5710970201831821,
      "grad_norm": 0.41980409622192383,
      "learning_rate": 2.144722118612458e-06,
      "loss": 0.6696,
      "step": 13780
    },
    {
      "epoch": 0.5711384640888557,
      "grad_norm": 0.4336474537849426,
      "learning_rate": 2.14451489908409e-06,
      "loss": 0.6592,
      "step": 13781
    },
    {
      "epoch": 0.5711799079945294,
      "grad_norm": 0.41137534379959106,
      "learning_rate": 2.1443076795557218e-06,
      "loss": 0.7068,
      "step": 13782
    },
    {
      "epoch": 0.5712213519002031,
      "grad_norm": 0.4011869430541992,
      "learning_rate": 2.144100460027353e-06,
      "loss": 0.6423,
      "step": 13783
    },
    {
      "epoch": 0.5712627958058768,
      "grad_norm": 0.40804818272590637,
      "learning_rate": 2.143893240498985e-06,
      "loss": 0.6698,
      "step": 13784
    },
    {
      "epoch": 0.5713042397115504,
      "grad_norm": 0.38665348291397095,
      "learning_rate": 2.1436860209706163e-06,
      "loss": 0.7109,
      "step": 13785
    },
    {
      "epoch": 0.571345683617224,
      "grad_norm": 0.4442654252052307,
      "learning_rate": 2.143478801442248e-06,
      "loss": 0.6573,
      "step": 13786
    },
    {
      "epoch": 0.5713871275228978,
      "grad_norm": 0.42189088463783264,
      "learning_rate": 2.1432715819138795e-06,
      "loss": 0.6826,
      "step": 13787
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.40870416164398193,
      "learning_rate": 2.1430643623855113e-06,
      "loss": 0.6851,
      "step": 13788
    },
    {
      "epoch": 0.5714700153342451,
      "grad_norm": 0.4245125353336334,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 0.6802,
      "step": 13789
    },
    {
      "epoch": 0.5715114592399188,
      "grad_norm": 0.39853158593177795,
      "learning_rate": 2.142649923328775e-06,
      "loss": 0.6565,
      "step": 13790
    },
    {
      "epoch": 0.5715529031455925,
      "grad_norm": 0.41548410058021545,
      "learning_rate": 2.1424427038004063e-06,
      "loss": 0.6812,
      "step": 13791
    },
    {
      "epoch": 0.5715943470512661,
      "grad_norm": 0.45132410526275635,
      "learning_rate": 2.142235484272038e-06,
      "loss": 0.7573,
      "step": 13792
    },
    {
      "epoch": 0.5716357909569397,
      "grad_norm": 0.42637890577316284,
      "learning_rate": 2.1420282647436695e-06,
      "loss": 0.687,
      "step": 13793
    },
    {
      "epoch": 0.5716772348626135,
      "grad_norm": 0.41940370202064514,
      "learning_rate": 2.1418210452153014e-06,
      "loss": 0.6538,
      "step": 13794
    },
    {
      "epoch": 0.5717186787682871,
      "grad_norm": 0.4022725224494934,
      "learning_rate": 2.1416138256869327e-06,
      "loss": 0.6652,
      "step": 13795
    },
    {
      "epoch": 0.5717601226739608,
      "grad_norm": 0.4475555121898651,
      "learning_rate": 2.1414066061585645e-06,
      "loss": 0.7107,
      "step": 13796
    },
    {
      "epoch": 0.5718015665796344,
      "grad_norm": 0.4290507137775421,
      "learning_rate": 2.141199386630196e-06,
      "loss": 0.7019,
      "step": 13797
    },
    {
      "epoch": 0.5718430104853082,
      "grad_norm": 0.4369785189628601,
      "learning_rate": 2.1409921671018277e-06,
      "loss": 0.6628,
      "step": 13798
    },
    {
      "epoch": 0.5718844543909818,
      "grad_norm": 0.4405632019042969,
      "learning_rate": 2.1407849475734596e-06,
      "loss": 0.6472,
      "step": 13799
    },
    {
      "epoch": 0.5719258982966555,
      "grad_norm": 0.4329068958759308,
      "learning_rate": 2.1405777280450914e-06,
      "loss": 0.693,
      "step": 13800
    },
    {
      "epoch": 0.5719673422023291,
      "grad_norm": 0.4378667175769806,
      "learning_rate": 2.1403705085167227e-06,
      "loss": 0.7378,
      "step": 13801
    },
    {
      "epoch": 0.5720087861080028,
      "grad_norm": 0.4579119086265564,
      "learning_rate": 2.1401632889883546e-06,
      "loss": 0.699,
      "step": 13802
    },
    {
      "epoch": 0.5720502300136765,
      "grad_norm": 0.42940330505371094,
      "learning_rate": 2.139956069459986e-06,
      "loss": 0.6899,
      "step": 13803
    },
    {
      "epoch": 0.5720916739193501,
      "grad_norm": 0.4073435068130493,
      "learning_rate": 2.1397488499316177e-06,
      "loss": 0.6431,
      "step": 13804
    },
    {
      "epoch": 0.5721331178250239,
      "grad_norm": 0.40492361783981323,
      "learning_rate": 2.139541630403249e-06,
      "loss": 0.6919,
      "step": 13805
    },
    {
      "epoch": 0.5721745617306975,
      "grad_norm": 0.39874666929244995,
      "learning_rate": 2.139334410874881e-06,
      "loss": 0.6517,
      "step": 13806
    },
    {
      "epoch": 0.5722160056363712,
      "grad_norm": 0.4402739107608795,
      "learning_rate": 2.1391271913465123e-06,
      "loss": 0.707,
      "step": 13807
    },
    {
      "epoch": 0.5722574495420448,
      "grad_norm": 0.4273405075073242,
      "learning_rate": 2.1389199718181446e-06,
      "loss": 0.7058,
      "step": 13808
    },
    {
      "epoch": 0.5722988934477186,
      "grad_norm": 0.4061214327812195,
      "learning_rate": 2.138712752289776e-06,
      "loss": 0.6553,
      "step": 13809
    },
    {
      "epoch": 0.5723403373533922,
      "grad_norm": 0.3824925422668457,
      "learning_rate": 2.1385055327614078e-06,
      "loss": 0.5905,
      "step": 13810
    },
    {
      "epoch": 0.5723817812590658,
      "grad_norm": 0.4068290889263153,
      "learning_rate": 2.138298313233039e-06,
      "loss": 0.6754,
      "step": 13811
    },
    {
      "epoch": 0.5724232251647395,
      "grad_norm": 0.40650874376296997,
      "learning_rate": 2.138091093704671e-06,
      "loss": 0.649,
      "step": 13812
    },
    {
      "epoch": 0.5724646690704132,
      "grad_norm": 0.40890997648239136,
      "learning_rate": 2.1378838741763023e-06,
      "loss": 0.6941,
      "step": 13813
    },
    {
      "epoch": 0.5725061129760869,
      "grad_norm": 0.44699788093566895,
      "learning_rate": 2.137676654647934e-06,
      "loss": 0.7012,
      "step": 13814
    },
    {
      "epoch": 0.5725475568817605,
      "grad_norm": 0.4492000937461853,
      "learning_rate": 2.1374694351195655e-06,
      "loss": 0.7224,
      "step": 13815
    },
    {
      "epoch": 0.5725890007874342,
      "grad_norm": 0.4396720230579376,
      "learning_rate": 2.1372622155911978e-06,
      "loss": 0.6687,
      "step": 13816
    },
    {
      "epoch": 0.5726304446931079,
      "grad_norm": 0.42881765961647034,
      "learning_rate": 2.137054996062829e-06,
      "loss": 0.6665,
      "step": 13817
    },
    {
      "epoch": 0.5726718885987816,
      "grad_norm": 0.47013214230537415,
      "learning_rate": 2.136847776534461e-06,
      "loss": 0.7179,
      "step": 13818
    },
    {
      "epoch": 0.5727133325044552,
      "grad_norm": 0.424792617559433,
      "learning_rate": 2.1366405570060923e-06,
      "loss": 0.7047,
      "step": 13819
    },
    {
      "epoch": 0.5727547764101288,
      "grad_norm": 0.40681713819503784,
      "learning_rate": 2.136433337477724e-06,
      "loss": 0.6531,
      "step": 13820
    },
    {
      "epoch": 0.5727962203158026,
      "grad_norm": 0.4198211431503296,
      "learning_rate": 2.1362261179493555e-06,
      "loss": 0.7065,
      "step": 13821
    },
    {
      "epoch": 0.5728376642214762,
      "grad_norm": 0.42295414209365845,
      "learning_rate": 2.1360188984209874e-06,
      "loss": 0.6975,
      "step": 13822
    },
    {
      "epoch": 0.5728791081271499,
      "grad_norm": 0.4182445704936981,
      "learning_rate": 2.1358116788926187e-06,
      "loss": 0.7019,
      "step": 13823
    },
    {
      "epoch": 0.5729205520328235,
      "grad_norm": 0.43964332342147827,
      "learning_rate": 2.1356044593642505e-06,
      "loss": 0.7148,
      "step": 13824
    },
    {
      "epoch": 0.5729619959384973,
      "grad_norm": 0.46299123764038086,
      "learning_rate": 2.1353972398358824e-06,
      "loss": 0.72,
      "step": 13825
    },
    {
      "epoch": 0.5730034398441709,
      "grad_norm": 0.44356223940849304,
      "learning_rate": 2.135190020307514e-06,
      "loss": 0.6831,
      "step": 13826
    },
    {
      "epoch": 0.5730448837498446,
      "grad_norm": 0.3931635320186615,
      "learning_rate": 2.1349828007791455e-06,
      "loss": 0.6936,
      "step": 13827
    },
    {
      "epoch": 0.5730863276555183,
      "grad_norm": 0.4080120325088501,
      "learning_rate": 2.1347755812507774e-06,
      "loss": 0.6963,
      "step": 13828
    },
    {
      "epoch": 0.5731277715611919,
      "grad_norm": 0.411540687084198,
      "learning_rate": 2.1345683617224087e-06,
      "loss": 0.6394,
      "step": 13829
    },
    {
      "epoch": 0.5731692154668656,
      "grad_norm": 0.40596064925193787,
      "learning_rate": 2.1343611421940406e-06,
      "loss": 0.6508,
      "step": 13830
    },
    {
      "epoch": 0.5732106593725392,
      "grad_norm": 0.3929804265499115,
      "learning_rate": 2.134153922665672e-06,
      "loss": 0.6572,
      "step": 13831
    },
    {
      "epoch": 0.573252103278213,
      "grad_norm": 0.4054909944534302,
      "learning_rate": 2.1339467031373037e-06,
      "loss": 0.667,
      "step": 13832
    },
    {
      "epoch": 0.5732935471838866,
      "grad_norm": 0.4164449870586395,
      "learning_rate": 2.1337394836089356e-06,
      "loss": 0.6875,
      "step": 13833
    },
    {
      "epoch": 0.5733349910895603,
      "grad_norm": 0.4241515100002289,
      "learning_rate": 2.1335322640805674e-06,
      "loss": 0.6956,
      "step": 13834
    },
    {
      "epoch": 0.5733764349952339,
      "grad_norm": 0.39518874883651733,
      "learning_rate": 2.1333250445521988e-06,
      "loss": 0.6748,
      "step": 13835
    },
    {
      "epoch": 0.5734178789009077,
      "grad_norm": 0.438277006149292,
      "learning_rate": 2.1331178250238306e-06,
      "loss": 0.7003,
      "step": 13836
    },
    {
      "epoch": 0.5734593228065813,
      "grad_norm": 0.41692548990249634,
      "learning_rate": 2.132910605495462e-06,
      "loss": 0.7333,
      "step": 13837
    },
    {
      "epoch": 0.5735007667122549,
      "grad_norm": 0.392770379781723,
      "learning_rate": 2.1327033859670938e-06,
      "loss": 0.6005,
      "step": 13838
    },
    {
      "epoch": 0.5735422106179286,
      "grad_norm": 0.4963643550872803,
      "learning_rate": 2.132496166438725e-06,
      "loss": 0.7476,
      "step": 13839
    },
    {
      "epoch": 0.5735836545236023,
      "grad_norm": 0.46543484926223755,
      "learning_rate": 2.132288946910357e-06,
      "loss": 0.6995,
      "step": 13840
    },
    {
      "epoch": 0.573625098429276,
      "grad_norm": 0.42931169271469116,
      "learning_rate": 2.1320817273819883e-06,
      "loss": 0.7004,
      "step": 13841
    },
    {
      "epoch": 0.5736665423349496,
      "grad_norm": 0.442949503660202,
      "learning_rate": 2.1318745078536206e-06,
      "loss": 0.7385,
      "step": 13842
    },
    {
      "epoch": 0.5737079862406234,
      "grad_norm": 0.4072267711162567,
      "learning_rate": 2.131667288325252e-06,
      "loss": 0.6479,
      "step": 13843
    },
    {
      "epoch": 0.573749430146297,
      "grad_norm": 0.4108417332172394,
      "learning_rate": 2.1314600687968838e-06,
      "loss": 0.7329,
      "step": 13844
    },
    {
      "epoch": 0.5737908740519707,
      "grad_norm": 0.4182269275188446,
      "learning_rate": 2.131252849268515e-06,
      "loss": 0.6821,
      "step": 13845
    },
    {
      "epoch": 0.5738323179576443,
      "grad_norm": 0.4258277714252472,
      "learning_rate": 2.131045629740147e-06,
      "loss": 0.6788,
      "step": 13846
    },
    {
      "epoch": 0.573873761863318,
      "grad_norm": 0.4154553711414337,
      "learning_rate": 2.1308384102117783e-06,
      "loss": 0.6749,
      "step": 13847
    },
    {
      "epoch": 0.5739152057689917,
      "grad_norm": 0.42243313789367676,
      "learning_rate": 2.13063119068341e-06,
      "loss": 0.6995,
      "step": 13848
    },
    {
      "epoch": 0.5739566496746653,
      "grad_norm": 0.430134117603302,
      "learning_rate": 2.1304239711550415e-06,
      "loss": 0.6849,
      "step": 13849
    },
    {
      "epoch": 0.573998093580339,
      "grad_norm": 0.40155595541000366,
      "learning_rate": 2.1302167516266733e-06,
      "loss": 0.6785,
      "step": 13850
    },
    {
      "epoch": 0.5740395374860127,
      "grad_norm": 0.44085079431533813,
      "learning_rate": 2.130009532098305e-06,
      "loss": 0.6726,
      "step": 13851
    },
    {
      "epoch": 0.5740809813916864,
      "grad_norm": 0.41348183155059814,
      "learning_rate": 2.129802312569937e-06,
      "loss": 0.7339,
      "step": 13852
    },
    {
      "epoch": 0.57412242529736,
      "grad_norm": 0.4230993092060089,
      "learning_rate": 2.1295950930415684e-06,
      "loss": 0.7393,
      "step": 13853
    },
    {
      "epoch": 0.5741638692030336,
      "grad_norm": 0.39222705364227295,
      "learning_rate": 2.1293878735132e-06,
      "loss": 0.6124,
      "step": 13854
    },
    {
      "epoch": 0.5742053131087074,
      "grad_norm": 0.4003104865550995,
      "learning_rate": 2.1291806539848315e-06,
      "loss": 0.662,
      "step": 13855
    },
    {
      "epoch": 0.574246757014381,
      "grad_norm": 0.43984320759773254,
      "learning_rate": 2.1289734344564634e-06,
      "loss": 0.7003,
      "step": 13856
    },
    {
      "epoch": 0.5742882009200547,
      "grad_norm": 0.41393032670021057,
      "learning_rate": 2.1287662149280947e-06,
      "loss": 0.656,
      "step": 13857
    },
    {
      "epoch": 0.5743296448257283,
      "grad_norm": 0.40789929032325745,
      "learning_rate": 2.1285589953997266e-06,
      "loss": 0.649,
      "step": 13858
    },
    {
      "epoch": 0.5743710887314021,
      "grad_norm": 0.4211776554584503,
      "learning_rate": 2.1283517758713584e-06,
      "loss": 0.6843,
      "step": 13859
    },
    {
      "epoch": 0.5744125326370757,
      "grad_norm": 0.4022696018218994,
      "learning_rate": 2.12814455634299e-06,
      "loss": 0.6799,
      "step": 13860
    },
    {
      "epoch": 0.5744539765427494,
      "grad_norm": 0.4194873869419098,
      "learning_rate": 2.1279373368146216e-06,
      "loss": 0.7008,
      "step": 13861
    },
    {
      "epoch": 0.574495420448423,
      "grad_norm": 0.45870786905288696,
      "learning_rate": 2.1277301172862534e-06,
      "loss": 0.6429,
      "step": 13862
    },
    {
      "epoch": 0.5745368643540967,
      "grad_norm": 0.4796798825263977,
      "learning_rate": 2.1275228977578848e-06,
      "loss": 0.6804,
      "step": 13863
    },
    {
      "epoch": 0.5745783082597704,
      "grad_norm": 0.42008277773857117,
      "learning_rate": 2.1273156782295166e-06,
      "loss": 0.6816,
      "step": 13864
    },
    {
      "epoch": 0.574619752165444,
      "grad_norm": 0.5288835167884827,
      "learning_rate": 2.127108458701148e-06,
      "loss": 0.7085,
      "step": 13865
    },
    {
      "epoch": 0.5746611960711178,
      "grad_norm": 0.37532493472099304,
      "learning_rate": 2.1269012391727798e-06,
      "loss": 0.655,
      "step": 13866
    },
    {
      "epoch": 0.5747026399767914,
      "grad_norm": 0.42376279830932617,
      "learning_rate": 2.1266940196444116e-06,
      "loss": 0.7367,
      "step": 13867
    },
    {
      "epoch": 0.5747440838824651,
      "grad_norm": 0.41587236523628235,
      "learning_rate": 2.1264868001160434e-06,
      "loss": 0.6924,
      "step": 13868
    },
    {
      "epoch": 0.5747855277881387,
      "grad_norm": 0.4085468649864197,
      "learning_rate": 2.1262795805876748e-06,
      "loss": 0.674,
      "step": 13869
    },
    {
      "epoch": 0.5748269716938125,
      "grad_norm": 0.42116639018058777,
      "learning_rate": 2.1260723610593066e-06,
      "loss": 0.6665,
      "step": 13870
    },
    {
      "epoch": 0.5748684155994861,
      "grad_norm": 0.40271204710006714,
      "learning_rate": 2.125865141530938e-06,
      "loss": 0.6528,
      "step": 13871
    },
    {
      "epoch": 0.5749098595051597,
      "grad_norm": 0.3918752670288086,
      "learning_rate": 2.1256579220025698e-06,
      "loss": 0.6481,
      "step": 13872
    },
    {
      "epoch": 0.5749513034108334,
      "grad_norm": 0.42959052324295044,
      "learning_rate": 2.125450702474201e-06,
      "loss": 0.6517,
      "step": 13873
    },
    {
      "epoch": 0.5749927473165071,
      "grad_norm": 0.429431289434433,
      "learning_rate": 2.125243482945833e-06,
      "loss": 0.6992,
      "step": 13874
    },
    {
      "epoch": 0.5750341912221808,
      "grad_norm": 0.4196905195713043,
      "learning_rate": 2.1250362634174643e-06,
      "loss": 0.689,
      "step": 13875
    },
    {
      "epoch": 0.5750756351278544,
      "grad_norm": 0.3998565077781677,
      "learning_rate": 2.124829043889096e-06,
      "loss": 0.6993,
      "step": 13876
    },
    {
      "epoch": 0.5751170790335282,
      "grad_norm": 0.39429956674575806,
      "learning_rate": 2.124621824360728e-06,
      "loss": 0.6664,
      "step": 13877
    },
    {
      "epoch": 0.5751585229392018,
      "grad_norm": 0.40669935941696167,
      "learning_rate": 2.1244146048323598e-06,
      "loss": 0.6694,
      "step": 13878
    },
    {
      "epoch": 0.5751999668448755,
      "grad_norm": 0.41118210554122925,
      "learning_rate": 2.124207385303991e-06,
      "loss": 0.6812,
      "step": 13879
    },
    {
      "epoch": 0.5752414107505491,
      "grad_norm": 0.38151469826698303,
      "learning_rate": 2.124000165775623e-06,
      "loss": 0.6501,
      "step": 13880
    },
    {
      "epoch": 0.5752828546562228,
      "grad_norm": 0.42089179158210754,
      "learning_rate": 2.1237929462472544e-06,
      "loss": 0.6371,
      "step": 13881
    },
    {
      "epoch": 0.5753242985618965,
      "grad_norm": 0.37730225920677185,
      "learning_rate": 2.123585726718886e-06,
      "loss": 0.6096,
      "step": 13882
    },
    {
      "epoch": 0.5753657424675701,
      "grad_norm": 0.39145150780677795,
      "learning_rate": 2.1233785071905175e-06,
      "loss": 0.6379,
      "step": 13883
    },
    {
      "epoch": 0.5754071863732438,
      "grad_norm": 0.38824498653411865,
      "learning_rate": 2.1231712876621494e-06,
      "loss": 0.6388,
      "step": 13884
    },
    {
      "epoch": 0.5754486302789175,
      "grad_norm": 0.4227767884731293,
      "learning_rate": 2.122964068133781e-06,
      "loss": 0.7136,
      "step": 13885
    },
    {
      "epoch": 0.5754900741845912,
      "grad_norm": 0.42485684156417847,
      "learning_rate": 2.122756848605413e-06,
      "loss": 0.7224,
      "step": 13886
    },
    {
      "epoch": 0.5755315180902648,
      "grad_norm": 0.423724502325058,
      "learning_rate": 2.1225496290770444e-06,
      "loss": 0.6672,
      "step": 13887
    },
    {
      "epoch": 0.5755729619959385,
      "grad_norm": 0.41348734498023987,
      "learning_rate": 2.122342409548676e-06,
      "loss": 0.66,
      "step": 13888
    },
    {
      "epoch": 0.5756144059016122,
      "grad_norm": 0.4288453161716461,
      "learning_rate": 2.1221351900203076e-06,
      "loss": 0.6956,
      "step": 13889
    },
    {
      "epoch": 0.5756558498072858,
      "grad_norm": 0.4332231283187866,
      "learning_rate": 2.1219279704919394e-06,
      "loss": 0.6665,
      "step": 13890
    },
    {
      "epoch": 0.5756972937129595,
      "grad_norm": 0.424470454454422,
      "learning_rate": 2.1217207509635707e-06,
      "loss": 0.7085,
      "step": 13891
    },
    {
      "epoch": 0.5757387376186331,
      "grad_norm": 0.39516445994377136,
      "learning_rate": 2.1215135314352026e-06,
      "loss": 0.6473,
      "step": 13892
    },
    {
      "epoch": 0.5757801815243069,
      "grad_norm": 0.3949078917503357,
      "learning_rate": 2.1213063119068344e-06,
      "loss": 0.7131,
      "step": 13893
    },
    {
      "epoch": 0.5758216254299805,
      "grad_norm": 0.40617242455482483,
      "learning_rate": 2.1210990923784658e-06,
      "loss": 0.6826,
      "step": 13894
    },
    {
      "epoch": 0.5758630693356542,
      "grad_norm": 0.4483875632286072,
      "learning_rate": 2.1208918728500976e-06,
      "loss": 0.698,
      "step": 13895
    },
    {
      "epoch": 0.5759045132413279,
      "grad_norm": 0.40919235348701477,
      "learning_rate": 2.1206846533217294e-06,
      "loss": 0.6431,
      "step": 13896
    },
    {
      "epoch": 0.5759459571470016,
      "grad_norm": 0.40146422386169434,
      "learning_rate": 2.1204774337933608e-06,
      "loss": 0.7385,
      "step": 13897
    },
    {
      "epoch": 0.5759874010526752,
      "grad_norm": 0.39921867847442627,
      "learning_rate": 2.1202702142649926e-06,
      "loss": 0.6965,
      "step": 13898
    },
    {
      "epoch": 0.5760288449583488,
      "grad_norm": 0.36559563875198364,
      "learning_rate": 2.120062994736624e-06,
      "loss": 0.6464,
      "step": 13899
    },
    {
      "epoch": 0.5760702888640226,
      "grad_norm": 0.3731408417224884,
      "learning_rate": 2.1198557752082558e-06,
      "loss": 0.6831,
      "step": 13900
    },
    {
      "epoch": 0.5761117327696962,
      "grad_norm": 0.4240875542163849,
      "learning_rate": 2.1196485556798876e-06,
      "loss": 0.7101,
      "step": 13901
    },
    {
      "epoch": 0.5761531766753699,
      "grad_norm": 0.4716005325317383,
      "learning_rate": 2.119441336151519e-06,
      "loss": 0.7327,
      "step": 13902
    },
    {
      "epoch": 0.5761946205810435,
      "grad_norm": 0.442240834236145,
      "learning_rate": 2.1192341166231508e-06,
      "loss": 0.7054,
      "step": 13903
    },
    {
      "epoch": 0.5762360644867173,
      "grad_norm": 0.3637550473213196,
      "learning_rate": 2.1190268970947826e-06,
      "loss": 0.6183,
      "step": 13904
    },
    {
      "epoch": 0.5762775083923909,
      "grad_norm": 0.3845212459564209,
      "learning_rate": 2.118819677566414e-06,
      "loss": 0.6786,
      "step": 13905
    },
    {
      "epoch": 0.5763189522980646,
      "grad_norm": 0.3772569000720978,
      "learning_rate": 2.1186124580380458e-06,
      "loss": 0.6543,
      "step": 13906
    },
    {
      "epoch": 0.5763603962037382,
      "grad_norm": 0.411650151014328,
      "learning_rate": 2.118405238509677e-06,
      "loss": 0.7021,
      "step": 13907
    },
    {
      "epoch": 0.5764018401094119,
      "grad_norm": 0.40500912070274353,
      "learning_rate": 2.118198018981309e-06,
      "loss": 0.6726,
      "step": 13908
    },
    {
      "epoch": 0.5764432840150856,
      "grad_norm": 0.377911776304245,
      "learning_rate": 2.1179907994529403e-06,
      "loss": 0.6035,
      "step": 13909
    },
    {
      "epoch": 0.5764847279207592,
      "grad_norm": 0.38176068663597107,
      "learning_rate": 2.117783579924572e-06,
      "loss": 0.7012,
      "step": 13910
    },
    {
      "epoch": 0.576526171826433,
      "grad_norm": 0.47060608863830566,
      "learning_rate": 2.117576360396204e-06,
      "loss": 0.7239,
      "step": 13911
    },
    {
      "epoch": 0.5765676157321066,
      "grad_norm": 0.3953971862792969,
      "learning_rate": 2.1173691408678358e-06,
      "loss": 0.6682,
      "step": 13912
    },
    {
      "epoch": 0.5766090596377803,
      "grad_norm": 0.40225908160209656,
      "learning_rate": 2.117161921339467e-06,
      "loss": 0.6663,
      "step": 13913
    },
    {
      "epoch": 0.5766505035434539,
      "grad_norm": 0.46893617510795593,
      "learning_rate": 2.116954701811099e-06,
      "loss": 0.7341,
      "step": 13914
    },
    {
      "epoch": 0.5766919474491276,
      "grad_norm": 0.4166678190231323,
      "learning_rate": 2.1167474822827304e-06,
      "loss": 0.7495,
      "step": 13915
    },
    {
      "epoch": 0.5767333913548013,
      "grad_norm": 0.4028873145580292,
      "learning_rate": 2.116540262754362e-06,
      "loss": 0.6929,
      "step": 13916
    },
    {
      "epoch": 0.5767748352604749,
      "grad_norm": 0.396801620721817,
      "learning_rate": 2.1163330432259936e-06,
      "loss": 0.7092,
      "step": 13917
    },
    {
      "epoch": 0.5768162791661486,
      "grad_norm": 0.4214009940624237,
      "learning_rate": 2.1161258236976254e-06,
      "loss": 0.6917,
      "step": 13918
    },
    {
      "epoch": 0.5768577230718223,
      "grad_norm": 0.38364648818969727,
      "learning_rate": 2.115918604169257e-06,
      "loss": 0.6406,
      "step": 13919
    },
    {
      "epoch": 0.576899166977496,
      "grad_norm": 0.37897318601608276,
      "learning_rate": 2.1157113846408886e-06,
      "loss": 0.6926,
      "step": 13920
    },
    {
      "epoch": 0.5769406108831696,
      "grad_norm": 0.39252403378486633,
      "learning_rate": 2.1155041651125204e-06,
      "loss": 0.6772,
      "step": 13921
    },
    {
      "epoch": 0.5769820547888433,
      "grad_norm": 0.39853599667549133,
      "learning_rate": 2.115296945584152e-06,
      "loss": 0.6996,
      "step": 13922
    },
    {
      "epoch": 0.577023498694517,
      "grad_norm": 0.41421592235565186,
      "learning_rate": 2.1150897260557836e-06,
      "loss": 0.6614,
      "step": 13923
    },
    {
      "epoch": 0.5770649426001906,
      "grad_norm": 0.414110004901886,
      "learning_rate": 2.1148825065274154e-06,
      "loss": 0.7024,
      "step": 13924
    },
    {
      "epoch": 0.5771063865058643,
      "grad_norm": 0.38315826654434204,
      "learning_rate": 2.1146752869990468e-06,
      "loss": 0.6564,
      "step": 13925
    },
    {
      "epoch": 0.5771478304115379,
      "grad_norm": 0.42277082800865173,
      "learning_rate": 2.1144680674706786e-06,
      "loss": 0.6936,
      "step": 13926
    },
    {
      "epoch": 0.5771892743172117,
      "grad_norm": 0.42121320962905884,
      "learning_rate": 2.1142608479423104e-06,
      "loss": 0.6827,
      "step": 13927
    },
    {
      "epoch": 0.5772307182228853,
      "grad_norm": 0.39676469564437866,
      "learning_rate": 2.1140536284139418e-06,
      "loss": 0.6777,
      "step": 13928
    },
    {
      "epoch": 0.577272162128559,
      "grad_norm": 0.4426287114620209,
      "learning_rate": 2.1138464088855736e-06,
      "loss": 0.6866,
      "step": 13929
    },
    {
      "epoch": 0.5773136060342327,
      "grad_norm": 0.445340096950531,
      "learning_rate": 2.1136391893572054e-06,
      "loss": 0.7344,
      "step": 13930
    },
    {
      "epoch": 0.5773550499399064,
      "grad_norm": 0.4261067509651184,
      "learning_rate": 2.1134319698288368e-06,
      "loss": 0.6808,
      "step": 13931
    },
    {
      "epoch": 0.57739649384558,
      "grad_norm": 0.4126306176185608,
      "learning_rate": 2.1132247503004686e-06,
      "loss": 0.6567,
      "step": 13932
    },
    {
      "epoch": 0.5774379377512536,
      "grad_norm": 0.40023842453956604,
      "learning_rate": 2.1130175307721e-06,
      "loss": 0.6672,
      "step": 13933
    },
    {
      "epoch": 0.5774793816569274,
      "grad_norm": 0.42071107029914856,
      "learning_rate": 2.1128103112437318e-06,
      "loss": 0.6779,
      "step": 13934
    },
    {
      "epoch": 0.577520825562601,
      "grad_norm": 0.4039819538593292,
      "learning_rate": 2.1126030917153636e-06,
      "loss": 0.6672,
      "step": 13935
    },
    {
      "epoch": 0.5775622694682747,
      "grad_norm": 0.37775951623916626,
      "learning_rate": 2.112395872186995e-06,
      "loss": 0.6362,
      "step": 13936
    },
    {
      "epoch": 0.5776037133739483,
      "grad_norm": 0.392974317073822,
      "learning_rate": 2.1121886526586268e-06,
      "loss": 0.7028,
      "step": 13937
    },
    {
      "epoch": 0.5776451572796221,
      "grad_norm": 0.40517041087150574,
      "learning_rate": 2.1119814331302586e-06,
      "loss": 0.6847,
      "step": 13938
    },
    {
      "epoch": 0.5776866011852957,
      "grad_norm": 0.42223483324050903,
      "learning_rate": 2.11177421360189e-06,
      "loss": 0.6447,
      "step": 13939
    },
    {
      "epoch": 0.5777280450909694,
      "grad_norm": 0.41729047894477844,
      "learning_rate": 2.1115669940735218e-06,
      "loss": 0.6838,
      "step": 13940
    },
    {
      "epoch": 0.577769488996643,
      "grad_norm": 0.37146613001823425,
      "learning_rate": 2.111359774545153e-06,
      "loss": 0.6379,
      "step": 13941
    },
    {
      "epoch": 0.5778109329023167,
      "grad_norm": 0.4196239113807678,
      "learning_rate": 2.111152555016785e-06,
      "loss": 0.7222,
      "step": 13942
    },
    {
      "epoch": 0.5778523768079904,
      "grad_norm": 0.4621763527393341,
      "learning_rate": 2.1109453354884168e-06,
      "loss": 0.7058,
      "step": 13943
    },
    {
      "epoch": 0.577893820713664,
      "grad_norm": 0.4035604000091553,
      "learning_rate": 2.110738115960048e-06,
      "loss": 0.6885,
      "step": 13944
    },
    {
      "epoch": 0.5779352646193378,
      "grad_norm": 0.4347308576107025,
      "learning_rate": 2.11053089643168e-06,
      "loss": 0.6709,
      "step": 13945
    },
    {
      "epoch": 0.5779767085250114,
      "grad_norm": 0.40753793716430664,
      "learning_rate": 2.1103236769033114e-06,
      "loss": 0.6503,
      "step": 13946
    },
    {
      "epoch": 0.5780181524306851,
      "grad_norm": 0.4089842140674591,
      "learning_rate": 2.110116457374943e-06,
      "loss": 0.6884,
      "step": 13947
    },
    {
      "epoch": 0.5780595963363587,
      "grad_norm": 0.3859905004501343,
      "learning_rate": 2.109909237846575e-06,
      "loss": 0.7175,
      "step": 13948
    },
    {
      "epoch": 0.5781010402420325,
      "grad_norm": 0.39515092968940735,
      "learning_rate": 2.1097020183182064e-06,
      "loss": 0.673,
      "step": 13949
    },
    {
      "epoch": 0.5781424841477061,
      "grad_norm": 0.3759388029575348,
      "learning_rate": 2.109494798789838e-06,
      "loss": 0.6719,
      "step": 13950
    },
    {
      "epoch": 0.5781839280533797,
      "grad_norm": 0.39402976632118225,
      "learning_rate": 2.1092875792614696e-06,
      "loss": 0.7001,
      "step": 13951
    },
    {
      "epoch": 0.5782253719590534,
      "grad_norm": 0.44954344630241394,
      "learning_rate": 2.1090803597331014e-06,
      "loss": 0.7028,
      "step": 13952
    },
    {
      "epoch": 0.5782668158647271,
      "grad_norm": 0.41271165013313293,
      "learning_rate": 2.108873140204733e-06,
      "loss": 0.7,
      "step": 13953
    },
    {
      "epoch": 0.5783082597704008,
      "grad_norm": 0.4078815281391144,
      "learning_rate": 2.1086659206763646e-06,
      "loss": 0.6899,
      "step": 13954
    },
    {
      "epoch": 0.5783497036760744,
      "grad_norm": 0.41494059562683105,
      "learning_rate": 2.1084587011479964e-06,
      "loss": 0.6602,
      "step": 13955
    },
    {
      "epoch": 0.5783911475817481,
      "grad_norm": 0.41329336166381836,
      "learning_rate": 2.108251481619628e-06,
      "loss": 0.7163,
      "step": 13956
    },
    {
      "epoch": 0.5784325914874218,
      "grad_norm": 0.4477458894252777,
      "learning_rate": 2.1080442620912596e-06,
      "loss": 0.658,
      "step": 13957
    },
    {
      "epoch": 0.5784740353930955,
      "grad_norm": 0.41085171699523926,
      "learning_rate": 2.1078370425628914e-06,
      "loss": 0.6288,
      "step": 13958
    },
    {
      "epoch": 0.5785154792987691,
      "grad_norm": 0.4114709794521332,
      "learning_rate": 2.1076298230345228e-06,
      "loss": 0.6348,
      "step": 13959
    },
    {
      "epoch": 0.5785569232044427,
      "grad_norm": 0.41593390703201294,
      "learning_rate": 2.1074226035061546e-06,
      "loss": 0.7344,
      "step": 13960
    },
    {
      "epoch": 0.5785983671101165,
      "grad_norm": 0.4081556797027588,
      "learning_rate": 2.1072153839777864e-06,
      "loss": 0.6809,
      "step": 13961
    },
    {
      "epoch": 0.5786398110157901,
      "grad_norm": 0.4164953827857971,
      "learning_rate": 2.1070081644494178e-06,
      "loss": 0.7029,
      "step": 13962
    },
    {
      "epoch": 0.5786812549214638,
      "grad_norm": 0.39169254899024963,
      "learning_rate": 2.1068009449210496e-06,
      "loss": 0.6732,
      "step": 13963
    },
    {
      "epoch": 0.5787226988271374,
      "grad_norm": 0.43679386377334595,
      "learning_rate": 2.1065937253926814e-06,
      "loss": 0.658,
      "step": 13964
    },
    {
      "epoch": 0.5787641427328112,
      "grad_norm": 0.3995024561882019,
      "learning_rate": 2.1063865058643128e-06,
      "loss": 0.6558,
      "step": 13965
    },
    {
      "epoch": 0.5788055866384848,
      "grad_norm": 0.39450815320014954,
      "learning_rate": 2.1061792863359446e-06,
      "loss": 0.6512,
      "step": 13966
    },
    {
      "epoch": 0.5788470305441585,
      "grad_norm": 0.39770016074180603,
      "learning_rate": 2.105972066807576e-06,
      "loss": 0.644,
      "step": 13967
    },
    {
      "epoch": 0.5788884744498322,
      "grad_norm": 0.4254592955112457,
      "learning_rate": 2.1057648472792078e-06,
      "loss": 0.661,
      "step": 13968
    },
    {
      "epoch": 0.5789299183555058,
      "grad_norm": 0.4329374432563782,
      "learning_rate": 2.1055576277508396e-06,
      "loss": 0.7463,
      "step": 13969
    },
    {
      "epoch": 0.5789713622611795,
      "grad_norm": 0.411597341299057,
      "learning_rate": 2.105350408222471e-06,
      "loss": 0.6606,
      "step": 13970
    },
    {
      "epoch": 0.5790128061668531,
      "grad_norm": 0.4047931432723999,
      "learning_rate": 2.1051431886941028e-06,
      "loss": 0.7437,
      "step": 13971
    },
    {
      "epoch": 0.5790542500725269,
      "grad_norm": 0.43294963240623474,
      "learning_rate": 2.104935969165734e-06,
      "loss": 0.7292,
      "step": 13972
    },
    {
      "epoch": 0.5790956939782005,
      "grad_norm": 0.37362104654312134,
      "learning_rate": 2.104728749637366e-06,
      "loss": 0.636,
      "step": 13973
    },
    {
      "epoch": 0.5791371378838742,
      "grad_norm": 0.4424699544906616,
      "learning_rate": 2.1045215301089978e-06,
      "loss": 0.6877,
      "step": 13974
    },
    {
      "epoch": 0.5791785817895478,
      "grad_norm": 0.42981305718421936,
      "learning_rate": 2.104314310580629e-06,
      "loss": 0.6904,
      "step": 13975
    },
    {
      "epoch": 0.5792200256952215,
      "grad_norm": 0.44261765480041504,
      "learning_rate": 2.104107091052261e-06,
      "loss": 0.661,
      "step": 13976
    },
    {
      "epoch": 0.5792614696008952,
      "grad_norm": 0.3819520175457001,
      "learning_rate": 2.1038998715238928e-06,
      "loss": 0.6655,
      "step": 13977
    },
    {
      "epoch": 0.5793029135065688,
      "grad_norm": 0.41112220287323,
      "learning_rate": 2.103692651995524e-06,
      "loss": 0.6726,
      "step": 13978
    },
    {
      "epoch": 0.5793443574122425,
      "grad_norm": 0.4099596440792084,
      "learning_rate": 2.103485432467156e-06,
      "loss": 0.6781,
      "step": 13979
    },
    {
      "epoch": 0.5793858013179162,
      "grad_norm": 0.42493852972984314,
      "learning_rate": 2.1032782129387874e-06,
      "loss": 0.7344,
      "step": 13980
    },
    {
      "epoch": 0.5794272452235899,
      "grad_norm": 0.414516419172287,
      "learning_rate": 2.103070993410419e-06,
      "loss": 0.7002,
      "step": 13981
    },
    {
      "epoch": 0.5794686891292635,
      "grad_norm": 0.4188615381717682,
      "learning_rate": 2.102863773882051e-06,
      "loss": 0.7136,
      "step": 13982
    },
    {
      "epoch": 0.5795101330349373,
      "grad_norm": 0.4281579256057739,
      "learning_rate": 2.1026565543536824e-06,
      "loss": 0.639,
      "step": 13983
    },
    {
      "epoch": 0.5795515769406109,
      "grad_norm": 0.39307263493537903,
      "learning_rate": 2.102449334825314e-06,
      "loss": 0.6086,
      "step": 13984
    },
    {
      "epoch": 0.5795930208462845,
      "grad_norm": 0.4173050820827484,
      "learning_rate": 2.1022421152969456e-06,
      "loss": 0.707,
      "step": 13985
    },
    {
      "epoch": 0.5796344647519582,
      "grad_norm": 0.41683340072631836,
      "learning_rate": 2.1020348957685774e-06,
      "loss": 0.6954,
      "step": 13986
    },
    {
      "epoch": 0.5796759086576319,
      "grad_norm": 0.41456449031829834,
      "learning_rate": 2.101827676240209e-06,
      "loss": 0.6599,
      "step": 13987
    },
    {
      "epoch": 0.5797173525633056,
      "grad_norm": 0.43093299865722656,
      "learning_rate": 2.1016204567118406e-06,
      "loss": 0.6927,
      "step": 13988
    },
    {
      "epoch": 0.5797587964689792,
      "grad_norm": 0.3923068940639496,
      "learning_rate": 2.1014132371834724e-06,
      "loss": 0.6962,
      "step": 13989
    },
    {
      "epoch": 0.5798002403746529,
      "grad_norm": 0.4108723998069763,
      "learning_rate": 2.101206017655104e-06,
      "loss": 0.6964,
      "step": 13990
    },
    {
      "epoch": 0.5798416842803266,
      "grad_norm": 0.3935836851596832,
      "learning_rate": 2.1009987981267356e-06,
      "loss": 0.7115,
      "step": 13991
    },
    {
      "epoch": 0.5798831281860003,
      "grad_norm": 0.46721214056015015,
      "learning_rate": 2.1007915785983674e-06,
      "loss": 0.7024,
      "step": 13992
    },
    {
      "epoch": 0.5799245720916739,
      "grad_norm": 0.4238993525505066,
      "learning_rate": 2.1005843590699988e-06,
      "loss": 0.6588,
      "step": 13993
    },
    {
      "epoch": 0.5799660159973475,
      "grad_norm": 0.4162410795688629,
      "learning_rate": 2.1003771395416306e-06,
      "loss": 0.6853,
      "step": 13994
    },
    {
      "epoch": 0.5800074599030213,
      "grad_norm": 0.4014846086502075,
      "learning_rate": 2.1001699200132624e-06,
      "loss": 0.7139,
      "step": 13995
    },
    {
      "epoch": 0.5800489038086949,
      "grad_norm": 0.3958783745765686,
      "learning_rate": 2.0999627004848938e-06,
      "loss": 0.6814,
      "step": 13996
    },
    {
      "epoch": 0.5800903477143686,
      "grad_norm": 0.42397916316986084,
      "learning_rate": 2.0997554809565256e-06,
      "loss": 0.6693,
      "step": 13997
    },
    {
      "epoch": 0.5801317916200422,
      "grad_norm": 0.4234289228916168,
      "learning_rate": 2.099548261428157e-06,
      "loss": 0.6841,
      "step": 13998
    },
    {
      "epoch": 0.580173235525716,
      "grad_norm": 0.4102562665939331,
      "learning_rate": 2.0993410418997888e-06,
      "loss": 0.661,
      "step": 13999
    },
    {
      "epoch": 0.5802146794313896,
      "grad_norm": 0.4328835904598236,
      "learning_rate": 2.0991338223714206e-06,
      "loss": 0.6965,
      "step": 14000
    },
    {
      "epoch": 0.5802561233370633,
      "grad_norm": 0.40245530009269714,
      "learning_rate": 2.098926602843052e-06,
      "loss": 0.668,
      "step": 14001
    },
    {
      "epoch": 0.580297567242737,
      "grad_norm": 0.4202079474925995,
      "learning_rate": 2.0987193833146838e-06,
      "loss": 0.6792,
      "step": 14002
    },
    {
      "epoch": 0.5803390111484106,
      "grad_norm": 0.4056127071380615,
      "learning_rate": 2.0985121637863156e-06,
      "loss": 0.6166,
      "step": 14003
    },
    {
      "epoch": 0.5803804550540843,
      "grad_norm": 0.3893952965736389,
      "learning_rate": 2.098304944257947e-06,
      "loss": 0.7041,
      "step": 14004
    },
    {
      "epoch": 0.5804218989597579,
      "grad_norm": 0.4257561266422272,
      "learning_rate": 2.0980977247295788e-06,
      "loss": 0.7095,
      "step": 14005
    },
    {
      "epoch": 0.5804633428654317,
      "grad_norm": 0.40364140272140503,
      "learning_rate": 2.09789050520121e-06,
      "loss": 0.6917,
      "step": 14006
    },
    {
      "epoch": 0.5805047867711053,
      "grad_norm": 0.42216676473617554,
      "learning_rate": 2.097683285672842e-06,
      "loss": 0.6649,
      "step": 14007
    },
    {
      "epoch": 0.580546230676779,
      "grad_norm": 0.42590832710266113,
      "learning_rate": 2.0974760661444738e-06,
      "loss": 0.6851,
      "step": 14008
    },
    {
      "epoch": 0.5805876745824526,
      "grad_norm": 0.43560436367988586,
      "learning_rate": 2.097268846616105e-06,
      "loss": 0.6949,
      "step": 14009
    },
    {
      "epoch": 0.5806291184881264,
      "grad_norm": 0.46034926176071167,
      "learning_rate": 2.097061627087737e-06,
      "loss": 0.7258,
      "step": 14010
    },
    {
      "epoch": 0.5806705623938,
      "grad_norm": 0.39238816499710083,
      "learning_rate": 2.096854407559369e-06,
      "loss": 0.6702,
      "step": 14011
    },
    {
      "epoch": 0.5807120062994736,
      "grad_norm": 0.458793967962265,
      "learning_rate": 2.096647188031e-06,
      "loss": 0.6923,
      "step": 14012
    },
    {
      "epoch": 0.5807534502051473,
      "grad_norm": 0.44276076555252075,
      "learning_rate": 2.096439968502632e-06,
      "loss": 0.7288,
      "step": 14013
    },
    {
      "epoch": 0.580794894110821,
      "grad_norm": 0.43984249234199524,
      "learning_rate": 2.0962327489742634e-06,
      "loss": 0.6885,
      "step": 14014
    },
    {
      "epoch": 0.5808363380164947,
      "grad_norm": 0.4300515353679657,
      "learning_rate": 2.096025529445895e-06,
      "loss": 0.6884,
      "step": 14015
    },
    {
      "epoch": 0.5808777819221683,
      "grad_norm": 0.38118037581443787,
      "learning_rate": 2.0958183099175266e-06,
      "loss": 0.6621,
      "step": 14016
    },
    {
      "epoch": 0.5809192258278421,
      "grad_norm": 0.4052985906600952,
      "learning_rate": 2.0956110903891584e-06,
      "loss": 0.6935,
      "step": 14017
    },
    {
      "epoch": 0.5809606697335157,
      "grad_norm": 0.4254544973373413,
      "learning_rate": 2.09540387086079e-06,
      "loss": 0.7144,
      "step": 14018
    },
    {
      "epoch": 0.5810021136391894,
      "grad_norm": 0.43935781717300415,
      "learning_rate": 2.095196651332422e-06,
      "loss": 0.646,
      "step": 14019
    },
    {
      "epoch": 0.581043557544863,
      "grad_norm": 0.4152074456214905,
      "learning_rate": 2.0949894318040534e-06,
      "loss": 0.6792,
      "step": 14020
    },
    {
      "epoch": 0.5810850014505367,
      "grad_norm": 0.4134312570095062,
      "learning_rate": 2.094782212275685e-06,
      "loss": 0.7014,
      "step": 14021
    },
    {
      "epoch": 0.5811264453562104,
      "grad_norm": 0.41419047117233276,
      "learning_rate": 2.0945749927473166e-06,
      "loss": 0.6827,
      "step": 14022
    },
    {
      "epoch": 0.581167889261884,
      "grad_norm": 0.4122260808944702,
      "learning_rate": 2.0943677732189484e-06,
      "loss": 0.6721,
      "step": 14023
    },
    {
      "epoch": 0.5812093331675577,
      "grad_norm": 0.43012627959251404,
      "learning_rate": 2.0941605536905798e-06,
      "loss": 0.7273,
      "step": 14024
    },
    {
      "epoch": 0.5812507770732314,
      "grad_norm": 0.4375002086162567,
      "learning_rate": 2.0939533341622116e-06,
      "loss": 0.7246,
      "step": 14025
    },
    {
      "epoch": 0.5812922209789051,
      "grad_norm": 0.426711767911911,
      "learning_rate": 2.0937461146338434e-06,
      "loss": 0.6973,
      "step": 14026
    },
    {
      "epoch": 0.5813336648845787,
      "grad_norm": 0.4303346872329712,
      "learning_rate": 2.0935388951054748e-06,
      "loss": 0.6963,
      "step": 14027
    },
    {
      "epoch": 0.5813751087902524,
      "grad_norm": 0.4281289279460907,
      "learning_rate": 2.0933316755771066e-06,
      "loss": 0.709,
      "step": 14028
    },
    {
      "epoch": 0.5814165526959261,
      "grad_norm": 0.40809744596481323,
      "learning_rate": 2.0931244560487384e-06,
      "loss": 0.6946,
      "step": 14029
    },
    {
      "epoch": 0.5814579966015997,
      "grad_norm": 0.41791069507598877,
      "learning_rate": 2.0929172365203698e-06,
      "loss": 0.6626,
      "step": 14030
    },
    {
      "epoch": 0.5814994405072734,
      "grad_norm": 0.3951282799243927,
      "learning_rate": 2.0927100169920016e-06,
      "loss": 0.636,
      "step": 14031
    },
    {
      "epoch": 0.581540884412947,
      "grad_norm": 0.4155261814594269,
      "learning_rate": 2.092502797463633e-06,
      "loss": 0.6512,
      "step": 14032
    },
    {
      "epoch": 0.5815823283186208,
      "grad_norm": 0.41479218006134033,
      "learning_rate": 2.0922955779352648e-06,
      "loss": 0.6729,
      "step": 14033
    },
    {
      "epoch": 0.5816237722242944,
      "grad_norm": 0.44920557737350464,
      "learning_rate": 2.0920883584068966e-06,
      "loss": 0.647,
      "step": 14034
    },
    {
      "epoch": 0.5816652161299681,
      "grad_norm": 0.40076035261154175,
      "learning_rate": 2.091881138878528e-06,
      "loss": 0.7493,
      "step": 14035
    },
    {
      "epoch": 0.5817066600356418,
      "grad_norm": 0.3887951076030731,
      "learning_rate": 2.0916739193501598e-06,
      "loss": 0.6597,
      "step": 14036
    },
    {
      "epoch": 0.5817481039413154,
      "grad_norm": 0.377981573343277,
      "learning_rate": 2.0914666998217916e-06,
      "loss": 0.6577,
      "step": 14037
    },
    {
      "epoch": 0.5817895478469891,
      "grad_norm": 0.3933688700199127,
      "learning_rate": 2.091259480293423e-06,
      "loss": 0.6609,
      "step": 14038
    },
    {
      "epoch": 0.5818309917526627,
      "grad_norm": 0.4089418947696686,
      "learning_rate": 2.091052260765055e-06,
      "loss": 0.696,
      "step": 14039
    },
    {
      "epoch": 0.5818724356583365,
      "grad_norm": 0.44607532024383545,
      "learning_rate": 2.090845041236686e-06,
      "loss": 0.7231,
      "step": 14040
    },
    {
      "epoch": 0.5819138795640101,
      "grad_norm": 0.40787380933761597,
      "learning_rate": 2.090637821708318e-06,
      "loss": 0.7097,
      "step": 14041
    },
    {
      "epoch": 0.5819553234696838,
      "grad_norm": 0.3894723653793335,
      "learning_rate": 2.0904306021799494e-06,
      "loss": 0.7095,
      "step": 14042
    },
    {
      "epoch": 0.5819967673753574,
      "grad_norm": 0.412477046251297,
      "learning_rate": 2.090223382651581e-06,
      "loss": 0.6914,
      "step": 14043
    },
    {
      "epoch": 0.5820382112810312,
      "grad_norm": 0.4269634187221527,
      "learning_rate": 2.090016163123213e-06,
      "loss": 0.7019,
      "step": 14044
    },
    {
      "epoch": 0.5820796551867048,
      "grad_norm": 0.39182087779045105,
      "learning_rate": 2.089808943594845e-06,
      "loss": 0.6982,
      "step": 14045
    },
    {
      "epoch": 0.5821210990923784,
      "grad_norm": 0.421166330575943,
      "learning_rate": 2.089601724066476e-06,
      "loss": 0.6863,
      "step": 14046
    },
    {
      "epoch": 0.5821625429980521,
      "grad_norm": 0.3929734528064728,
      "learning_rate": 2.089394504538108e-06,
      "loss": 0.675,
      "step": 14047
    },
    {
      "epoch": 0.5822039869037258,
      "grad_norm": 0.38887181878089905,
      "learning_rate": 2.0891872850097394e-06,
      "loss": 0.6443,
      "step": 14048
    },
    {
      "epoch": 0.5822454308093995,
      "grad_norm": 0.36460423469543457,
      "learning_rate": 2.088980065481371e-06,
      "loss": 0.6357,
      "step": 14049
    },
    {
      "epoch": 0.5822868747150731,
      "grad_norm": 0.40347644686698914,
      "learning_rate": 2.0887728459530026e-06,
      "loss": 0.6467,
      "step": 14050
    },
    {
      "epoch": 0.5823283186207469,
      "grad_norm": 0.4432804584503174,
      "learning_rate": 2.0885656264246344e-06,
      "loss": 0.6707,
      "step": 14051
    },
    {
      "epoch": 0.5823697625264205,
      "grad_norm": 0.45212048292160034,
      "learning_rate": 2.088358406896266e-06,
      "loss": 0.6661,
      "step": 14052
    },
    {
      "epoch": 0.5824112064320942,
      "grad_norm": 0.4463343024253845,
      "learning_rate": 2.088151187367898e-06,
      "loss": 0.6898,
      "step": 14053
    },
    {
      "epoch": 0.5824526503377678,
      "grad_norm": 0.391155868768692,
      "learning_rate": 2.0879439678395294e-06,
      "loss": 0.6729,
      "step": 14054
    },
    {
      "epoch": 0.5824940942434415,
      "grad_norm": 0.4303790032863617,
      "learning_rate": 2.087736748311161e-06,
      "loss": 0.709,
      "step": 14055
    },
    {
      "epoch": 0.5825355381491152,
      "grad_norm": 0.41908955574035645,
      "learning_rate": 2.0875295287827926e-06,
      "loss": 0.7419,
      "step": 14056
    },
    {
      "epoch": 0.5825769820547888,
      "grad_norm": 0.3932686150074005,
      "learning_rate": 2.0873223092544244e-06,
      "loss": 0.6476,
      "step": 14057
    },
    {
      "epoch": 0.5826184259604625,
      "grad_norm": 0.4087761640548706,
      "learning_rate": 2.0871150897260558e-06,
      "loss": 0.6523,
      "step": 14058
    },
    {
      "epoch": 0.5826598698661362,
      "grad_norm": 0.392632395029068,
      "learning_rate": 2.0869078701976876e-06,
      "loss": 0.6705,
      "step": 14059
    },
    {
      "epoch": 0.5827013137718099,
      "grad_norm": 0.43129754066467285,
      "learning_rate": 2.0867006506693194e-06,
      "loss": 0.7214,
      "step": 14060
    },
    {
      "epoch": 0.5827427576774835,
      "grad_norm": 0.4126093089580536,
      "learning_rate": 2.0864934311409508e-06,
      "loss": 0.6641,
      "step": 14061
    },
    {
      "epoch": 0.5827842015831572,
      "grad_norm": 0.42566001415252686,
      "learning_rate": 2.0862862116125826e-06,
      "loss": 0.6702,
      "step": 14062
    },
    {
      "epoch": 0.5828256454888309,
      "grad_norm": 0.4286690950393677,
      "learning_rate": 2.0860789920842144e-06,
      "loss": 0.6797,
      "step": 14063
    },
    {
      "epoch": 0.5828670893945045,
      "grad_norm": 0.39859017729759216,
      "learning_rate": 2.0858717725558458e-06,
      "loss": 0.5922,
      "step": 14064
    },
    {
      "epoch": 0.5829085333001782,
      "grad_norm": 0.4190301299095154,
      "learning_rate": 2.0856645530274776e-06,
      "loss": 0.665,
      "step": 14065
    },
    {
      "epoch": 0.5829499772058518,
      "grad_norm": 0.39661139249801636,
      "learning_rate": 2.085457333499109e-06,
      "loss": 0.6641,
      "step": 14066
    },
    {
      "epoch": 0.5829914211115256,
      "grad_norm": 0.42437002062797546,
      "learning_rate": 2.0852501139707408e-06,
      "loss": 0.6414,
      "step": 14067
    },
    {
      "epoch": 0.5830328650171992,
      "grad_norm": 0.4082057476043701,
      "learning_rate": 2.085042894442372e-06,
      "loss": 0.7206,
      "step": 14068
    },
    {
      "epoch": 0.5830743089228729,
      "grad_norm": 0.39763063192367554,
      "learning_rate": 2.084835674914004e-06,
      "loss": 0.6653,
      "step": 14069
    },
    {
      "epoch": 0.5831157528285466,
      "grad_norm": 0.4525987505912781,
      "learning_rate": 2.084628455385636e-06,
      "loss": 0.7134,
      "step": 14070
    },
    {
      "epoch": 0.5831571967342203,
      "grad_norm": 0.4134485423564911,
      "learning_rate": 2.0844212358572676e-06,
      "loss": 0.6985,
      "step": 14071
    },
    {
      "epoch": 0.5831986406398939,
      "grad_norm": 0.4152374565601349,
      "learning_rate": 2.084214016328899e-06,
      "loss": 0.7048,
      "step": 14072
    },
    {
      "epoch": 0.5832400845455675,
      "grad_norm": 0.4049505293369293,
      "learning_rate": 2.084006796800531e-06,
      "loss": 0.7151,
      "step": 14073
    },
    {
      "epoch": 0.5832815284512413,
      "grad_norm": 0.43901824951171875,
      "learning_rate": 2.083799577272162e-06,
      "loss": 0.6777,
      "step": 14074
    },
    {
      "epoch": 0.5833229723569149,
      "grad_norm": 0.4292544424533844,
      "learning_rate": 2.083592357743794e-06,
      "loss": 0.7249,
      "step": 14075
    },
    {
      "epoch": 0.5833644162625886,
      "grad_norm": 0.4330682158470154,
      "learning_rate": 2.0833851382154254e-06,
      "loss": 0.696,
      "step": 14076
    },
    {
      "epoch": 0.5834058601682622,
      "grad_norm": 0.4338514506816864,
      "learning_rate": 2.083177918687057e-06,
      "loss": 0.6658,
      "step": 14077
    },
    {
      "epoch": 0.583447304073936,
      "grad_norm": 0.4213811755180359,
      "learning_rate": 2.082970699158689e-06,
      "loss": 0.6731,
      "step": 14078
    },
    {
      "epoch": 0.5834887479796096,
      "grad_norm": 0.41014501452445984,
      "learning_rate": 2.082763479630321e-06,
      "loss": 0.703,
      "step": 14079
    },
    {
      "epoch": 0.5835301918852833,
      "grad_norm": 0.41408291459083557,
      "learning_rate": 2.082556260101952e-06,
      "loss": 0.7018,
      "step": 14080
    },
    {
      "epoch": 0.5835716357909569,
      "grad_norm": 0.40027403831481934,
      "learning_rate": 2.082349040573584e-06,
      "loss": 0.7004,
      "step": 14081
    },
    {
      "epoch": 0.5836130796966306,
      "grad_norm": 0.4236155152320862,
      "learning_rate": 2.0821418210452154e-06,
      "loss": 0.662,
      "step": 14082
    },
    {
      "epoch": 0.5836545236023043,
      "grad_norm": 0.46726009249687195,
      "learning_rate": 2.081934601516847e-06,
      "loss": 0.7764,
      "step": 14083
    },
    {
      "epoch": 0.5836959675079779,
      "grad_norm": 0.3876621425151825,
      "learning_rate": 2.0817273819884786e-06,
      "loss": 0.6401,
      "step": 14084
    },
    {
      "epoch": 0.5837374114136517,
      "grad_norm": 0.4253304898738861,
      "learning_rate": 2.0815201624601104e-06,
      "loss": 0.6727,
      "step": 14085
    },
    {
      "epoch": 0.5837788553193253,
      "grad_norm": 0.40640705823898315,
      "learning_rate": 2.081312942931742e-06,
      "loss": 0.688,
      "step": 14086
    },
    {
      "epoch": 0.583820299224999,
      "grad_norm": 0.37806129455566406,
      "learning_rate": 2.081105723403374e-06,
      "loss": 0.7046,
      "step": 14087
    },
    {
      "epoch": 0.5838617431306726,
      "grad_norm": 0.4474796950817108,
      "learning_rate": 2.0808985038750054e-06,
      "loss": 0.7021,
      "step": 14088
    },
    {
      "epoch": 0.5839031870363464,
      "grad_norm": 0.42185741662979126,
      "learning_rate": 2.080691284346637e-06,
      "loss": 0.7468,
      "step": 14089
    },
    {
      "epoch": 0.58394463094202,
      "grad_norm": 0.4272300899028778,
      "learning_rate": 2.0804840648182686e-06,
      "loss": 0.707,
      "step": 14090
    },
    {
      "epoch": 0.5839860748476936,
      "grad_norm": 0.4150744378566742,
      "learning_rate": 2.0802768452899004e-06,
      "loss": 0.6898,
      "step": 14091
    },
    {
      "epoch": 0.5840275187533673,
      "grad_norm": 0.42361462116241455,
      "learning_rate": 2.0800696257615318e-06,
      "loss": 0.6718,
      "step": 14092
    },
    {
      "epoch": 0.584068962659041,
      "grad_norm": 0.4008960723876953,
      "learning_rate": 2.0798624062331636e-06,
      "loss": 0.7017,
      "step": 14093
    },
    {
      "epoch": 0.5841104065647147,
      "grad_norm": 0.4131130576133728,
      "learning_rate": 2.079655186704795e-06,
      "loss": 0.6704,
      "step": 14094
    },
    {
      "epoch": 0.5841518504703883,
      "grad_norm": 0.4030848741531372,
      "learning_rate": 2.0794479671764268e-06,
      "loss": 0.6887,
      "step": 14095
    },
    {
      "epoch": 0.584193294376062,
      "grad_norm": 0.44151321053504944,
      "learning_rate": 2.0792407476480586e-06,
      "loss": 0.7126,
      "step": 14096
    },
    {
      "epoch": 0.5842347382817357,
      "grad_norm": 0.4266393482685089,
      "learning_rate": 2.0790335281196904e-06,
      "loss": 0.7125,
      "step": 14097
    },
    {
      "epoch": 0.5842761821874093,
      "grad_norm": 0.39625313878059387,
      "learning_rate": 2.078826308591322e-06,
      "loss": 0.6713,
      "step": 14098
    },
    {
      "epoch": 0.584317626093083,
      "grad_norm": 0.4193464517593384,
      "learning_rate": 2.0786190890629536e-06,
      "loss": 0.6829,
      "step": 14099
    },
    {
      "epoch": 0.5843590699987566,
      "grad_norm": 0.38401955366134644,
      "learning_rate": 2.078411869534585e-06,
      "loss": 0.6537,
      "step": 14100
    },
    {
      "epoch": 0.5844005139044304,
      "grad_norm": 0.41671890020370483,
      "learning_rate": 2.078204650006217e-06,
      "loss": 0.7051,
      "step": 14101
    },
    {
      "epoch": 0.584441957810104,
      "grad_norm": 0.4694873094558716,
      "learning_rate": 2.077997430477848e-06,
      "loss": 0.7837,
      "step": 14102
    },
    {
      "epoch": 0.5844834017157777,
      "grad_norm": 0.39645621180534363,
      "learning_rate": 2.07779021094948e-06,
      "loss": 0.6787,
      "step": 14103
    },
    {
      "epoch": 0.5845248456214513,
      "grad_norm": 0.43636074662208557,
      "learning_rate": 2.077582991421112e-06,
      "loss": 0.6642,
      "step": 14104
    },
    {
      "epoch": 0.5845662895271251,
      "grad_norm": 0.4149452745914459,
      "learning_rate": 2.0773757718927436e-06,
      "loss": 0.7064,
      "step": 14105
    },
    {
      "epoch": 0.5846077334327987,
      "grad_norm": 0.3590863347053528,
      "learning_rate": 2.077168552364375e-06,
      "loss": 0.6582,
      "step": 14106
    },
    {
      "epoch": 0.5846491773384723,
      "grad_norm": 0.3875824213027954,
      "learning_rate": 2.076961332836007e-06,
      "loss": 0.6342,
      "step": 14107
    },
    {
      "epoch": 0.5846906212441461,
      "grad_norm": 0.3891555964946747,
      "learning_rate": 2.076754113307638e-06,
      "loss": 0.6814,
      "step": 14108
    },
    {
      "epoch": 0.5847320651498197,
      "grad_norm": 0.4402729868888855,
      "learning_rate": 2.07654689377927e-06,
      "loss": 0.73,
      "step": 14109
    },
    {
      "epoch": 0.5847735090554934,
      "grad_norm": 0.4177851378917694,
      "learning_rate": 2.0763396742509014e-06,
      "loss": 0.6543,
      "step": 14110
    },
    {
      "epoch": 0.584814952961167,
      "grad_norm": 0.4017607569694519,
      "learning_rate": 2.076132454722533e-06,
      "loss": 0.6793,
      "step": 14111
    },
    {
      "epoch": 0.5848563968668408,
      "grad_norm": 0.39510414004325867,
      "learning_rate": 2.075925235194165e-06,
      "loss": 0.6372,
      "step": 14112
    },
    {
      "epoch": 0.5848978407725144,
      "grad_norm": 0.39579302072525024,
      "learning_rate": 2.075718015665797e-06,
      "loss": 0.6782,
      "step": 14113
    },
    {
      "epoch": 0.5849392846781881,
      "grad_norm": 0.43034523725509644,
      "learning_rate": 2.075510796137428e-06,
      "loss": 0.6582,
      "step": 14114
    },
    {
      "epoch": 0.5849807285838617,
      "grad_norm": 0.39743393659591675,
      "learning_rate": 2.07530357660906e-06,
      "loss": 0.6073,
      "step": 14115
    },
    {
      "epoch": 0.5850221724895354,
      "grad_norm": 0.45189177989959717,
      "learning_rate": 2.0750963570806914e-06,
      "loss": 0.7058,
      "step": 14116
    },
    {
      "epoch": 0.5850636163952091,
      "grad_norm": 0.4064171612262726,
      "learning_rate": 2.074889137552323e-06,
      "loss": 0.6562,
      "step": 14117
    },
    {
      "epoch": 0.5851050603008827,
      "grad_norm": 0.37119773030281067,
      "learning_rate": 2.0746819180239546e-06,
      "loss": 0.6084,
      "step": 14118
    },
    {
      "epoch": 0.5851465042065565,
      "grad_norm": 0.40333500504493713,
      "learning_rate": 2.0744746984955864e-06,
      "loss": 0.6833,
      "step": 14119
    },
    {
      "epoch": 0.5851879481122301,
      "grad_norm": 0.4300459623336792,
      "learning_rate": 2.0742674789672178e-06,
      "loss": 0.6809,
      "step": 14120
    },
    {
      "epoch": 0.5852293920179038,
      "grad_norm": 0.3845449388027191,
      "learning_rate": 2.07406025943885e-06,
      "loss": 0.6061,
      "step": 14121
    },
    {
      "epoch": 0.5852708359235774,
      "grad_norm": 0.37670525908470154,
      "learning_rate": 2.0738530399104814e-06,
      "loss": 0.6179,
      "step": 14122
    },
    {
      "epoch": 0.5853122798292512,
      "grad_norm": 0.3845716118812561,
      "learning_rate": 2.073645820382113e-06,
      "loss": 0.6604,
      "step": 14123
    },
    {
      "epoch": 0.5853537237349248,
      "grad_norm": 0.4287910759449005,
      "learning_rate": 2.0734386008537446e-06,
      "loss": 0.6799,
      "step": 14124
    },
    {
      "epoch": 0.5853951676405984,
      "grad_norm": 0.41032832860946655,
      "learning_rate": 2.0732313813253764e-06,
      "loss": 0.6327,
      "step": 14125
    },
    {
      "epoch": 0.5854366115462721,
      "grad_norm": 0.4158497750759125,
      "learning_rate": 2.0730241617970078e-06,
      "loss": 0.6594,
      "step": 14126
    },
    {
      "epoch": 0.5854780554519458,
      "grad_norm": 0.4010414481163025,
      "learning_rate": 2.0728169422686396e-06,
      "loss": 0.6772,
      "step": 14127
    },
    {
      "epoch": 0.5855194993576195,
      "grad_norm": 0.4181547462940216,
      "learning_rate": 2.072609722740271e-06,
      "loss": 0.6838,
      "step": 14128
    },
    {
      "epoch": 0.5855609432632931,
      "grad_norm": 0.4231301248073578,
      "learning_rate": 2.072402503211903e-06,
      "loss": 0.6812,
      "step": 14129
    },
    {
      "epoch": 0.5856023871689668,
      "grad_norm": 0.39424529671669006,
      "learning_rate": 2.0721952836835346e-06,
      "loss": 0.6697,
      "step": 14130
    },
    {
      "epoch": 0.5856438310746405,
      "grad_norm": 0.3838825821876526,
      "learning_rate": 2.0719880641551664e-06,
      "loss": 0.6622,
      "step": 14131
    },
    {
      "epoch": 0.5856852749803142,
      "grad_norm": 0.42421436309814453,
      "learning_rate": 2.071780844626798e-06,
      "loss": 0.6764,
      "step": 14132
    },
    {
      "epoch": 0.5857267188859878,
      "grad_norm": 0.4287389814853668,
      "learning_rate": 2.0715736250984296e-06,
      "loss": 0.7095,
      "step": 14133
    },
    {
      "epoch": 0.5857681627916614,
      "grad_norm": 0.4140169620513916,
      "learning_rate": 2.071366405570061e-06,
      "loss": 0.6687,
      "step": 14134
    },
    {
      "epoch": 0.5858096066973352,
      "grad_norm": 0.4443388283252716,
      "learning_rate": 2.071159186041693e-06,
      "loss": 0.6719,
      "step": 14135
    },
    {
      "epoch": 0.5858510506030088,
      "grad_norm": 0.4401494562625885,
      "learning_rate": 2.070951966513324e-06,
      "loss": 0.7019,
      "step": 14136
    },
    {
      "epoch": 0.5858924945086825,
      "grad_norm": 0.4076954424381256,
      "learning_rate": 2.070744746984956e-06,
      "loss": 0.7429,
      "step": 14137
    },
    {
      "epoch": 0.5859339384143561,
      "grad_norm": 0.37651047110557556,
      "learning_rate": 2.070537527456588e-06,
      "loss": 0.6368,
      "step": 14138
    },
    {
      "epoch": 0.5859753823200299,
      "grad_norm": 0.39002805948257446,
      "learning_rate": 2.0703303079282196e-06,
      "loss": 0.6721,
      "step": 14139
    },
    {
      "epoch": 0.5860168262257035,
      "grad_norm": 0.4409191608428955,
      "learning_rate": 2.070123088399851e-06,
      "loss": 0.6902,
      "step": 14140
    },
    {
      "epoch": 0.5860582701313772,
      "grad_norm": 0.4271860718727112,
      "learning_rate": 2.069915868871483e-06,
      "loss": 0.6837,
      "step": 14141
    },
    {
      "epoch": 0.5860997140370509,
      "grad_norm": 0.42550286650657654,
      "learning_rate": 2.069708649343114e-06,
      "loss": 0.6964,
      "step": 14142
    },
    {
      "epoch": 0.5861411579427245,
      "grad_norm": 0.4464608430862427,
      "learning_rate": 2.069501429814746e-06,
      "loss": 0.6812,
      "step": 14143
    },
    {
      "epoch": 0.5861826018483982,
      "grad_norm": 0.3930616080760956,
      "learning_rate": 2.0692942102863774e-06,
      "loss": 0.6404,
      "step": 14144
    },
    {
      "epoch": 0.5862240457540718,
      "grad_norm": 0.39899328351020813,
      "learning_rate": 2.069086990758009e-06,
      "loss": 0.7302,
      "step": 14145
    },
    {
      "epoch": 0.5862654896597456,
      "grad_norm": 0.43091467022895813,
      "learning_rate": 2.0688797712296406e-06,
      "loss": 0.6946,
      "step": 14146
    },
    {
      "epoch": 0.5863069335654192,
      "grad_norm": 0.4289466142654419,
      "learning_rate": 2.068672551701273e-06,
      "loss": 0.6608,
      "step": 14147
    },
    {
      "epoch": 0.5863483774710929,
      "grad_norm": 0.4070889949798584,
      "learning_rate": 2.068465332172904e-06,
      "loss": 0.6678,
      "step": 14148
    },
    {
      "epoch": 0.5863898213767665,
      "grad_norm": 0.3978257477283478,
      "learning_rate": 2.068258112644536e-06,
      "loss": 0.6858,
      "step": 14149
    },
    {
      "epoch": 0.5864312652824402,
      "grad_norm": 0.388560950756073,
      "learning_rate": 2.0680508931161674e-06,
      "loss": 0.7251,
      "step": 14150
    },
    {
      "epoch": 0.5864727091881139,
      "grad_norm": 0.4146293103694916,
      "learning_rate": 2.067843673587799e-06,
      "loss": 0.7009,
      "step": 14151
    },
    {
      "epoch": 0.5865141530937875,
      "grad_norm": 0.41662558913230896,
      "learning_rate": 2.0676364540594306e-06,
      "loss": 0.6726,
      "step": 14152
    },
    {
      "epoch": 0.5865555969994612,
      "grad_norm": 0.423968106508255,
      "learning_rate": 2.0674292345310624e-06,
      "loss": 0.6919,
      "step": 14153
    },
    {
      "epoch": 0.5865970409051349,
      "grad_norm": 0.46350449323654175,
      "learning_rate": 2.0672220150026938e-06,
      "loss": 0.6415,
      "step": 14154
    },
    {
      "epoch": 0.5866384848108086,
      "grad_norm": 0.41528844833374023,
      "learning_rate": 2.0670147954743256e-06,
      "loss": 0.6743,
      "step": 14155
    },
    {
      "epoch": 0.5866799287164822,
      "grad_norm": 0.4210805296897888,
      "learning_rate": 2.0668075759459574e-06,
      "loss": 0.6669,
      "step": 14156
    },
    {
      "epoch": 0.586721372622156,
      "grad_norm": 0.44741693139076233,
      "learning_rate": 2.0666003564175892e-06,
      "loss": 0.6649,
      "step": 14157
    },
    {
      "epoch": 0.5867628165278296,
      "grad_norm": 0.3907623887062073,
      "learning_rate": 2.0663931368892206e-06,
      "loss": 0.6217,
      "step": 14158
    },
    {
      "epoch": 0.5868042604335032,
      "grad_norm": 0.4113163948059082,
      "learning_rate": 2.0661859173608524e-06,
      "loss": 0.6947,
      "step": 14159
    },
    {
      "epoch": 0.5868457043391769,
      "grad_norm": 0.3909599184989929,
      "learning_rate": 2.065978697832484e-06,
      "loss": 0.6433,
      "step": 14160
    },
    {
      "epoch": 0.5868871482448506,
      "grad_norm": 0.4082099199295044,
      "learning_rate": 2.0657714783041156e-06,
      "loss": 0.6646,
      "step": 14161
    },
    {
      "epoch": 0.5869285921505243,
      "grad_norm": 0.4048808813095093,
      "learning_rate": 2.065564258775747e-06,
      "loss": 0.6425,
      "step": 14162
    },
    {
      "epoch": 0.5869700360561979,
      "grad_norm": 0.4205514192581177,
      "learning_rate": 2.065357039247379e-06,
      "loss": 0.6774,
      "step": 14163
    },
    {
      "epoch": 0.5870114799618716,
      "grad_norm": 0.42061248421669006,
      "learning_rate": 2.06514981971901e-06,
      "loss": 0.7256,
      "step": 14164
    },
    {
      "epoch": 0.5870529238675453,
      "grad_norm": 0.4397643506526947,
      "learning_rate": 2.0649426001906424e-06,
      "loss": 0.7102,
      "step": 14165
    },
    {
      "epoch": 0.587094367773219,
      "grad_norm": 0.41509339213371277,
      "learning_rate": 2.064735380662274e-06,
      "loss": 0.7024,
      "step": 14166
    },
    {
      "epoch": 0.5871358116788926,
      "grad_norm": 0.41387051343917847,
      "learning_rate": 2.0645281611339056e-06,
      "loss": 0.668,
      "step": 14167
    },
    {
      "epoch": 0.5871772555845662,
      "grad_norm": 0.40030667185783386,
      "learning_rate": 2.064320941605537e-06,
      "loss": 0.6914,
      "step": 14168
    },
    {
      "epoch": 0.58721869949024,
      "grad_norm": 0.4140687882900238,
      "learning_rate": 2.064113722077169e-06,
      "loss": 0.6741,
      "step": 14169
    },
    {
      "epoch": 0.5872601433959136,
      "grad_norm": 0.40622541308403015,
      "learning_rate": 2.0639065025488e-06,
      "loss": 0.6731,
      "step": 14170
    },
    {
      "epoch": 0.5873015873015873,
      "grad_norm": 0.41607508063316345,
      "learning_rate": 2.063699283020432e-06,
      "loss": 0.6686,
      "step": 14171
    },
    {
      "epoch": 0.587343031207261,
      "grad_norm": 0.38662946224212646,
      "learning_rate": 2.0634920634920634e-06,
      "loss": 0.668,
      "step": 14172
    },
    {
      "epoch": 0.5873844751129347,
      "grad_norm": 0.44469600915908813,
      "learning_rate": 2.0632848439636956e-06,
      "loss": 0.7322,
      "step": 14173
    },
    {
      "epoch": 0.5874259190186083,
      "grad_norm": 0.40012022852897644,
      "learning_rate": 2.063077624435327e-06,
      "loss": 0.6555,
      "step": 14174
    },
    {
      "epoch": 0.587467362924282,
      "grad_norm": 0.4459618926048279,
      "learning_rate": 2.062870404906959e-06,
      "loss": 0.7169,
      "step": 14175
    },
    {
      "epoch": 0.5875088068299557,
      "grad_norm": 0.42265668511390686,
      "learning_rate": 2.06266318537859e-06,
      "loss": 0.6715,
      "step": 14176
    },
    {
      "epoch": 0.5875502507356293,
      "grad_norm": 0.40955111384391785,
      "learning_rate": 2.062455965850222e-06,
      "loss": 0.6659,
      "step": 14177
    },
    {
      "epoch": 0.587591694641303,
      "grad_norm": 0.39265507459640503,
      "learning_rate": 2.0622487463218534e-06,
      "loss": 0.6783,
      "step": 14178
    },
    {
      "epoch": 0.5876331385469766,
      "grad_norm": 0.4169926643371582,
      "learning_rate": 2.062041526793485e-06,
      "loss": 0.6649,
      "step": 14179
    },
    {
      "epoch": 0.5876745824526504,
      "grad_norm": 0.3616180419921875,
      "learning_rate": 2.0618343072651166e-06,
      "loss": 0.6393,
      "step": 14180
    },
    {
      "epoch": 0.587716026358324,
      "grad_norm": 0.3996827304363251,
      "learning_rate": 2.0616270877367484e-06,
      "loss": 0.6549,
      "step": 14181
    },
    {
      "epoch": 0.5877574702639977,
      "grad_norm": 0.4011535048484802,
      "learning_rate": 2.06141986820838e-06,
      "loss": 0.7051,
      "step": 14182
    },
    {
      "epoch": 0.5877989141696713,
      "grad_norm": 0.44790586829185486,
      "learning_rate": 2.061212648680012e-06,
      "loss": 0.7067,
      "step": 14183
    },
    {
      "epoch": 0.5878403580753451,
      "grad_norm": 0.4208473861217499,
      "learning_rate": 2.0610054291516434e-06,
      "loss": 0.6904,
      "step": 14184
    },
    {
      "epoch": 0.5878818019810187,
      "grad_norm": 0.43749651312828064,
      "learning_rate": 2.060798209623275e-06,
      "loss": 0.7058,
      "step": 14185
    },
    {
      "epoch": 0.5879232458866923,
      "grad_norm": 0.48676642775535583,
      "learning_rate": 2.0605909900949066e-06,
      "loss": 0.7285,
      "step": 14186
    },
    {
      "epoch": 0.587964689792366,
      "grad_norm": 0.4017510414123535,
      "learning_rate": 2.0603837705665384e-06,
      "loss": 0.6812,
      "step": 14187
    },
    {
      "epoch": 0.5880061336980397,
      "grad_norm": 0.42195677757263184,
      "learning_rate": 2.06017655103817e-06,
      "loss": 0.6521,
      "step": 14188
    },
    {
      "epoch": 0.5880475776037134,
      "grad_norm": 0.416572242975235,
      "learning_rate": 2.0599693315098016e-06,
      "loss": 0.6793,
      "step": 14189
    },
    {
      "epoch": 0.588089021509387,
      "grad_norm": 0.3998049199581146,
      "learning_rate": 2.059762111981433e-06,
      "loss": 0.717,
      "step": 14190
    },
    {
      "epoch": 0.5881304654150608,
      "grad_norm": 0.4125073254108429,
      "learning_rate": 2.0595548924530652e-06,
      "loss": 0.6288,
      "step": 14191
    },
    {
      "epoch": 0.5881719093207344,
      "grad_norm": 0.39791011810302734,
      "learning_rate": 2.0593476729246966e-06,
      "loss": 0.7048,
      "step": 14192
    },
    {
      "epoch": 0.5882133532264081,
      "grad_norm": 0.43792593479156494,
      "learning_rate": 2.0591404533963284e-06,
      "loss": 0.6848,
      "step": 14193
    },
    {
      "epoch": 0.5882547971320817,
      "grad_norm": 0.4008736312389374,
      "learning_rate": 2.05893323386796e-06,
      "loss": 0.6805,
      "step": 14194
    },
    {
      "epoch": 0.5882962410377554,
      "grad_norm": 0.43492022156715393,
      "learning_rate": 2.0587260143395916e-06,
      "loss": 0.7092,
      "step": 14195
    },
    {
      "epoch": 0.5883376849434291,
      "grad_norm": 0.41053488850593567,
      "learning_rate": 2.058518794811223e-06,
      "loss": 0.6882,
      "step": 14196
    },
    {
      "epoch": 0.5883791288491027,
      "grad_norm": 0.4040428400039673,
      "learning_rate": 2.058311575282855e-06,
      "loss": 0.7021,
      "step": 14197
    },
    {
      "epoch": 0.5884205727547764,
      "grad_norm": 0.3852170705795288,
      "learning_rate": 2.058104355754486e-06,
      "loss": 0.6099,
      "step": 14198
    },
    {
      "epoch": 0.5884620166604501,
      "grad_norm": 0.42070117592811584,
      "learning_rate": 2.0578971362261184e-06,
      "loss": 0.6958,
      "step": 14199
    },
    {
      "epoch": 0.5885034605661238,
      "grad_norm": 0.38195353746414185,
      "learning_rate": 2.05768991669775e-06,
      "loss": 0.6316,
      "step": 14200
    },
    {
      "epoch": 0.5885449044717974,
      "grad_norm": 0.4075537919998169,
      "learning_rate": 2.0574826971693816e-06,
      "loss": 0.688,
      "step": 14201
    },
    {
      "epoch": 0.5885863483774711,
      "grad_norm": 0.418608158826828,
      "learning_rate": 2.057275477641013e-06,
      "loss": 0.7302,
      "step": 14202
    },
    {
      "epoch": 0.5886277922831448,
      "grad_norm": 0.42793625593185425,
      "learning_rate": 2.057068258112645e-06,
      "loss": 0.7595,
      "step": 14203
    },
    {
      "epoch": 0.5886692361888184,
      "grad_norm": 0.5078359842300415,
      "learning_rate": 2.056861038584276e-06,
      "loss": 0.6714,
      "step": 14204
    },
    {
      "epoch": 0.5887106800944921,
      "grad_norm": 0.4163070619106293,
      "learning_rate": 2.056653819055908e-06,
      "loss": 0.6855,
      "step": 14205
    },
    {
      "epoch": 0.5887521240001657,
      "grad_norm": 0.4410068988800049,
      "learning_rate": 2.0564465995275394e-06,
      "loss": 0.6711,
      "step": 14206
    },
    {
      "epoch": 0.5887935679058395,
      "grad_norm": 0.4828309714794159,
      "learning_rate": 2.056239379999171e-06,
      "loss": 0.6816,
      "step": 14207
    },
    {
      "epoch": 0.5888350118115131,
      "grad_norm": 0.4332621991634369,
      "learning_rate": 2.056032160470803e-06,
      "loss": 0.739,
      "step": 14208
    },
    {
      "epoch": 0.5888764557171868,
      "grad_norm": 0.3900606036186218,
      "learning_rate": 2.055824940942435e-06,
      "loss": 0.6449,
      "step": 14209
    },
    {
      "epoch": 0.5889178996228605,
      "grad_norm": 0.39068275690078735,
      "learning_rate": 2.055617721414066e-06,
      "loss": 0.6165,
      "step": 14210
    },
    {
      "epoch": 0.5889593435285341,
      "grad_norm": 0.39331457018852234,
      "learning_rate": 2.055410501885698e-06,
      "loss": 0.7122,
      "step": 14211
    },
    {
      "epoch": 0.5890007874342078,
      "grad_norm": 0.43049222230911255,
      "learning_rate": 2.0552032823573294e-06,
      "loss": 0.6737,
      "step": 14212
    },
    {
      "epoch": 0.5890422313398814,
      "grad_norm": 0.4973275661468506,
      "learning_rate": 2.054996062828961e-06,
      "loss": 0.6823,
      "step": 14213
    },
    {
      "epoch": 0.5890836752455552,
      "grad_norm": 0.4248104393482208,
      "learning_rate": 2.0547888433005926e-06,
      "loss": 0.6973,
      "step": 14214
    },
    {
      "epoch": 0.5891251191512288,
      "grad_norm": 0.4152061343193054,
      "learning_rate": 2.0545816237722244e-06,
      "loss": 0.6758,
      "step": 14215
    },
    {
      "epoch": 0.5891665630569025,
      "grad_norm": 0.42476075887680054,
      "learning_rate": 2.054374404243856e-06,
      "loss": 0.7295,
      "step": 14216
    },
    {
      "epoch": 0.5892080069625761,
      "grad_norm": 0.43457522988319397,
      "learning_rate": 2.054167184715488e-06,
      "loss": 0.7045,
      "step": 14217
    },
    {
      "epoch": 0.5892494508682499,
      "grad_norm": 0.43032944202423096,
      "learning_rate": 2.0539599651871194e-06,
      "loss": 0.6405,
      "step": 14218
    },
    {
      "epoch": 0.5892908947739235,
      "grad_norm": 0.4209657907485962,
      "learning_rate": 2.0537527456587512e-06,
      "loss": 0.6852,
      "step": 14219
    },
    {
      "epoch": 0.5893323386795971,
      "grad_norm": 0.4044792950153351,
      "learning_rate": 2.0535455261303826e-06,
      "loss": 0.6681,
      "step": 14220
    },
    {
      "epoch": 0.5893737825852708,
      "grad_norm": 0.4333532452583313,
      "learning_rate": 2.0533383066020144e-06,
      "loss": 0.7072,
      "step": 14221
    },
    {
      "epoch": 0.5894152264909445,
      "grad_norm": 0.3966088593006134,
      "learning_rate": 2.053131087073646e-06,
      "loss": 0.6648,
      "step": 14222
    },
    {
      "epoch": 0.5894566703966182,
      "grad_norm": 0.4488344192504883,
      "learning_rate": 2.0529238675452776e-06,
      "loss": 0.7272,
      "step": 14223
    },
    {
      "epoch": 0.5894981143022918,
      "grad_norm": 0.40928444266319275,
      "learning_rate": 2.052716648016909e-06,
      "loss": 0.6787,
      "step": 14224
    },
    {
      "epoch": 0.5895395582079656,
      "grad_norm": 0.39797958731651306,
      "learning_rate": 2.052509428488541e-06,
      "loss": 0.6606,
      "step": 14225
    },
    {
      "epoch": 0.5895810021136392,
      "grad_norm": 0.4518476724624634,
      "learning_rate": 2.0523022089601726e-06,
      "loss": 0.7538,
      "step": 14226
    },
    {
      "epoch": 0.5896224460193129,
      "grad_norm": 0.3876931071281433,
      "learning_rate": 2.0520949894318044e-06,
      "loss": 0.632,
      "step": 14227
    },
    {
      "epoch": 0.5896638899249865,
      "grad_norm": 0.40297946333885193,
      "learning_rate": 2.051887769903436e-06,
      "loss": 0.6755,
      "step": 14228
    },
    {
      "epoch": 0.5897053338306601,
      "grad_norm": 0.38023218512535095,
      "learning_rate": 2.0516805503750676e-06,
      "loss": 0.6863,
      "step": 14229
    },
    {
      "epoch": 0.5897467777363339,
      "grad_norm": 0.3974701762199402,
      "learning_rate": 2.051473330846699e-06,
      "loss": 0.665,
      "step": 14230
    },
    {
      "epoch": 0.5897882216420075,
      "grad_norm": 0.3757016062736511,
      "learning_rate": 2.051266111318331e-06,
      "loss": 0.6733,
      "step": 14231
    },
    {
      "epoch": 0.5898296655476812,
      "grad_norm": 0.42480897903442383,
      "learning_rate": 2.051058891789962e-06,
      "loss": 0.6477,
      "step": 14232
    },
    {
      "epoch": 0.5898711094533549,
      "grad_norm": 0.3969673216342926,
      "learning_rate": 2.050851672261594e-06,
      "loss": 0.6265,
      "step": 14233
    },
    {
      "epoch": 0.5899125533590286,
      "grad_norm": 0.4334273934364319,
      "learning_rate": 2.050644452733226e-06,
      "loss": 0.6522,
      "step": 14234
    },
    {
      "epoch": 0.5899539972647022,
      "grad_norm": 0.38328373432159424,
      "learning_rate": 2.0504372332048576e-06,
      "loss": 0.6353,
      "step": 14235
    },
    {
      "epoch": 0.589995441170376,
      "grad_norm": 0.4175913631916046,
      "learning_rate": 2.050230013676489e-06,
      "loss": 0.6569,
      "step": 14236
    },
    {
      "epoch": 0.5900368850760496,
      "grad_norm": 0.4121232032775879,
      "learning_rate": 2.050022794148121e-06,
      "loss": 0.6967,
      "step": 14237
    },
    {
      "epoch": 0.5900783289817232,
      "grad_norm": 0.4246392548084259,
      "learning_rate": 2.049815574619752e-06,
      "loss": 0.6494,
      "step": 14238
    },
    {
      "epoch": 0.5901197728873969,
      "grad_norm": 0.4085242748260498,
      "learning_rate": 2.049608355091384e-06,
      "loss": 0.6484,
      "step": 14239
    },
    {
      "epoch": 0.5901612167930705,
      "grad_norm": 0.3751453757286072,
      "learning_rate": 2.0494011355630154e-06,
      "loss": 0.6704,
      "step": 14240
    },
    {
      "epoch": 0.5902026606987443,
      "grad_norm": 0.4172389507293701,
      "learning_rate": 2.049193916034647e-06,
      "loss": 0.6914,
      "step": 14241
    },
    {
      "epoch": 0.5902441046044179,
      "grad_norm": 0.38432854413986206,
      "learning_rate": 2.0489866965062786e-06,
      "loss": 0.6768,
      "step": 14242
    },
    {
      "epoch": 0.5902855485100916,
      "grad_norm": 0.4108058214187622,
      "learning_rate": 2.048779476977911e-06,
      "loss": 0.6688,
      "step": 14243
    },
    {
      "epoch": 0.5903269924157653,
      "grad_norm": 0.3812786340713501,
      "learning_rate": 2.048572257449542e-06,
      "loss": 0.6836,
      "step": 14244
    },
    {
      "epoch": 0.590368436321439,
      "grad_norm": 0.40980976819992065,
      "learning_rate": 2.048365037921174e-06,
      "loss": 0.6814,
      "step": 14245
    },
    {
      "epoch": 0.5904098802271126,
      "grad_norm": 0.42524954676628113,
      "learning_rate": 2.0481578183928054e-06,
      "loss": 0.6675,
      "step": 14246
    },
    {
      "epoch": 0.5904513241327862,
      "grad_norm": 0.38327327370643616,
      "learning_rate": 2.0479505988644372e-06,
      "loss": 0.6255,
      "step": 14247
    },
    {
      "epoch": 0.59049276803846,
      "grad_norm": 0.4526336193084717,
      "learning_rate": 2.0477433793360686e-06,
      "loss": 0.699,
      "step": 14248
    },
    {
      "epoch": 0.5905342119441336,
      "grad_norm": 0.39851605892181396,
      "learning_rate": 2.0475361598077004e-06,
      "loss": 0.6843,
      "step": 14249
    },
    {
      "epoch": 0.5905756558498073,
      "grad_norm": 0.42890194058418274,
      "learning_rate": 2.047328940279332e-06,
      "loss": 0.6862,
      "step": 14250
    },
    {
      "epoch": 0.5906170997554809,
      "grad_norm": 0.4282785952091217,
      "learning_rate": 2.0471217207509636e-06,
      "loss": 0.7461,
      "step": 14251
    },
    {
      "epoch": 0.5906585436611547,
      "grad_norm": 0.38665252923965454,
      "learning_rate": 2.0469145012225954e-06,
      "loss": 0.641,
      "step": 14252
    },
    {
      "epoch": 0.5906999875668283,
      "grad_norm": 0.43085816502571106,
      "learning_rate": 2.0467072816942272e-06,
      "loss": 0.6613,
      "step": 14253
    },
    {
      "epoch": 0.590741431472502,
      "grad_norm": 0.4050680994987488,
      "learning_rate": 2.0465000621658586e-06,
      "loss": 0.6207,
      "step": 14254
    },
    {
      "epoch": 0.5907828753781756,
      "grad_norm": 0.4099087119102478,
      "learning_rate": 2.0462928426374904e-06,
      "loss": 0.6528,
      "step": 14255
    },
    {
      "epoch": 0.5908243192838493,
      "grad_norm": 0.4211459457874298,
      "learning_rate": 2.046085623109122e-06,
      "loss": 0.6531,
      "step": 14256
    },
    {
      "epoch": 0.590865763189523,
      "grad_norm": 0.4365153908729553,
      "learning_rate": 2.0458784035807536e-06,
      "loss": 0.6667,
      "step": 14257
    },
    {
      "epoch": 0.5909072070951966,
      "grad_norm": 0.42215296626091003,
      "learning_rate": 2.045671184052385e-06,
      "loss": 0.6401,
      "step": 14258
    },
    {
      "epoch": 0.5909486510008704,
      "grad_norm": 0.4347688555717468,
      "learning_rate": 2.045463964524017e-06,
      "loss": 0.7175,
      "step": 14259
    },
    {
      "epoch": 0.590990094906544,
      "grad_norm": 0.39373746514320374,
      "learning_rate": 2.0452567449956486e-06,
      "loss": 0.6174,
      "step": 14260
    },
    {
      "epoch": 0.5910315388122177,
      "grad_norm": 0.4205906093120575,
      "learning_rate": 2.0450495254672804e-06,
      "loss": 0.6654,
      "step": 14261
    },
    {
      "epoch": 0.5910729827178913,
      "grad_norm": 0.44716185331344604,
      "learning_rate": 2.044842305938912e-06,
      "loss": 0.7771,
      "step": 14262
    },
    {
      "epoch": 0.5911144266235651,
      "grad_norm": 0.3970456123352051,
      "learning_rate": 2.0446350864105436e-06,
      "loss": 0.5822,
      "step": 14263
    },
    {
      "epoch": 0.5911558705292387,
      "grad_norm": 0.4157412052154541,
      "learning_rate": 2.044427866882175e-06,
      "loss": 0.7002,
      "step": 14264
    },
    {
      "epoch": 0.5911973144349123,
      "grad_norm": 0.42289999127388,
      "learning_rate": 2.044220647353807e-06,
      "loss": 0.6569,
      "step": 14265
    },
    {
      "epoch": 0.591238758340586,
      "grad_norm": 0.3884695768356323,
      "learning_rate": 2.044013427825438e-06,
      "loss": 0.6506,
      "step": 14266
    },
    {
      "epoch": 0.5912802022462597,
      "grad_norm": 0.4208281636238098,
      "learning_rate": 2.04380620829707e-06,
      "loss": 0.6958,
      "step": 14267
    },
    {
      "epoch": 0.5913216461519334,
      "grad_norm": 0.41270431876182556,
      "learning_rate": 2.043598988768702e-06,
      "loss": 0.6796,
      "step": 14268
    },
    {
      "epoch": 0.591363090057607,
      "grad_norm": 0.37933364510536194,
      "learning_rate": 2.0433917692403336e-06,
      "loss": 0.6858,
      "step": 14269
    },
    {
      "epoch": 0.5914045339632807,
      "grad_norm": 0.4286758601665497,
      "learning_rate": 2.043184549711965e-06,
      "loss": 0.6788,
      "step": 14270
    },
    {
      "epoch": 0.5914459778689544,
      "grad_norm": 0.39950719475746155,
      "learning_rate": 2.042977330183597e-06,
      "loss": 0.6906,
      "step": 14271
    },
    {
      "epoch": 0.591487421774628,
      "grad_norm": 0.41800302267074585,
      "learning_rate": 2.042770110655228e-06,
      "loss": 0.6768,
      "step": 14272
    },
    {
      "epoch": 0.5915288656803017,
      "grad_norm": 0.4239901006221771,
      "learning_rate": 2.04256289112686e-06,
      "loss": 0.6614,
      "step": 14273
    },
    {
      "epoch": 0.5915703095859753,
      "grad_norm": 0.40485909581184387,
      "learning_rate": 2.0423556715984914e-06,
      "loss": 0.6812,
      "step": 14274
    },
    {
      "epoch": 0.5916117534916491,
      "grad_norm": 0.379594624042511,
      "learning_rate": 2.0421484520701232e-06,
      "loss": 0.6519,
      "step": 14275
    },
    {
      "epoch": 0.5916531973973227,
      "grad_norm": 0.42538881301879883,
      "learning_rate": 2.0419412325417546e-06,
      "loss": 0.6735,
      "step": 14276
    },
    {
      "epoch": 0.5916946413029964,
      "grad_norm": 0.37060701847076416,
      "learning_rate": 2.0417340130133864e-06,
      "loss": 0.5886,
      "step": 14277
    },
    {
      "epoch": 0.59173608520867,
      "grad_norm": 0.427942156791687,
      "learning_rate": 2.0415267934850182e-06,
      "loss": 0.6492,
      "step": 14278
    },
    {
      "epoch": 0.5917775291143438,
      "grad_norm": 0.42558640241622925,
      "learning_rate": 2.04131957395665e-06,
      "loss": 0.6973,
      "step": 14279
    },
    {
      "epoch": 0.5918189730200174,
      "grad_norm": 0.42972731590270996,
      "learning_rate": 2.0411123544282814e-06,
      "loss": 0.6775,
      "step": 14280
    },
    {
      "epoch": 0.591860416925691,
      "grad_norm": 0.40738192200660706,
      "learning_rate": 2.0409051348999132e-06,
      "loss": 0.7073,
      "step": 14281
    },
    {
      "epoch": 0.5919018608313648,
      "grad_norm": 0.405779629945755,
      "learning_rate": 2.0406979153715446e-06,
      "loss": 0.6504,
      "step": 14282
    },
    {
      "epoch": 0.5919433047370384,
      "grad_norm": 0.4060389995574951,
      "learning_rate": 2.0404906958431764e-06,
      "loss": 0.6484,
      "step": 14283
    },
    {
      "epoch": 0.5919847486427121,
      "grad_norm": 0.4020824134349823,
      "learning_rate": 2.040283476314808e-06,
      "loss": 0.6729,
      "step": 14284
    },
    {
      "epoch": 0.5920261925483857,
      "grad_norm": 0.4039686620235443,
      "learning_rate": 2.0400762567864396e-06,
      "loss": 0.7025,
      "step": 14285
    },
    {
      "epoch": 0.5920676364540595,
      "grad_norm": 0.3836396634578705,
      "learning_rate": 2.0398690372580714e-06,
      "loss": 0.6731,
      "step": 14286
    },
    {
      "epoch": 0.5921090803597331,
      "grad_norm": 0.3903673589229584,
      "learning_rate": 2.0396618177297032e-06,
      "loss": 0.6946,
      "step": 14287
    },
    {
      "epoch": 0.5921505242654068,
      "grad_norm": 0.4101863503456116,
      "learning_rate": 2.0394545982013346e-06,
      "loss": 0.7,
      "step": 14288
    },
    {
      "epoch": 0.5921919681710804,
      "grad_norm": 0.4006079435348511,
      "learning_rate": 2.0392473786729664e-06,
      "loss": 0.6334,
      "step": 14289
    },
    {
      "epoch": 0.5922334120767541,
      "grad_norm": 0.43651461601257324,
      "learning_rate": 2.039040159144598e-06,
      "loss": 0.7045,
      "step": 14290
    },
    {
      "epoch": 0.5922748559824278,
      "grad_norm": 0.39669862389564514,
      "learning_rate": 2.0388329396162296e-06,
      "loss": 0.7063,
      "step": 14291
    },
    {
      "epoch": 0.5923162998881014,
      "grad_norm": 0.3962988555431366,
      "learning_rate": 2.038625720087861e-06,
      "loss": 0.699,
      "step": 14292
    },
    {
      "epoch": 0.5923577437937751,
      "grad_norm": 0.4120485186576843,
      "learning_rate": 2.038418500559493e-06,
      "loss": 0.6752,
      "step": 14293
    },
    {
      "epoch": 0.5923991876994488,
      "grad_norm": 0.39418941736221313,
      "learning_rate": 2.0382112810311246e-06,
      "loss": 0.6908,
      "step": 14294
    },
    {
      "epoch": 0.5924406316051225,
      "grad_norm": 0.4442867934703827,
      "learning_rate": 2.0380040615027564e-06,
      "loss": 0.7605,
      "step": 14295
    },
    {
      "epoch": 0.5924820755107961,
      "grad_norm": 0.3840016722679138,
      "learning_rate": 2.037796841974388e-06,
      "loss": 0.6521,
      "step": 14296
    },
    {
      "epoch": 0.5925235194164699,
      "grad_norm": 0.3866652846336365,
      "learning_rate": 2.0375896224460196e-06,
      "loss": 0.6837,
      "step": 14297
    },
    {
      "epoch": 0.5925649633221435,
      "grad_norm": 0.4449007213115692,
      "learning_rate": 2.037382402917651e-06,
      "loss": 0.7148,
      "step": 14298
    },
    {
      "epoch": 0.5926064072278171,
      "grad_norm": 0.40215498208999634,
      "learning_rate": 2.037175183389283e-06,
      "loss": 0.682,
      "step": 14299
    },
    {
      "epoch": 0.5926478511334908,
      "grad_norm": 0.40515628457069397,
      "learning_rate": 2.036967963860914e-06,
      "loss": 0.6914,
      "step": 14300
    },
    {
      "epoch": 0.5926892950391645,
      "grad_norm": 0.403740257024765,
      "learning_rate": 2.036760744332546e-06,
      "loss": 0.6713,
      "step": 14301
    },
    {
      "epoch": 0.5927307389448382,
      "grad_norm": 0.4088153541088104,
      "learning_rate": 2.036553524804178e-06,
      "loss": 0.6692,
      "step": 14302
    },
    {
      "epoch": 0.5927721828505118,
      "grad_norm": 0.4251941442489624,
      "learning_rate": 2.036346305275809e-06,
      "loss": 0.6888,
      "step": 14303
    },
    {
      "epoch": 0.5928136267561855,
      "grad_norm": 0.4285546839237213,
      "learning_rate": 2.036139085747441e-06,
      "loss": 0.7031,
      "step": 14304
    },
    {
      "epoch": 0.5928550706618592,
      "grad_norm": 0.37559422850608826,
      "learning_rate": 2.035931866219073e-06,
      "loss": 0.6919,
      "step": 14305
    },
    {
      "epoch": 0.5928965145675329,
      "grad_norm": 0.37371811270713806,
      "learning_rate": 2.0357246466907042e-06,
      "loss": 0.6575,
      "step": 14306
    },
    {
      "epoch": 0.5929379584732065,
      "grad_norm": 0.40498480200767517,
      "learning_rate": 2.035517427162336e-06,
      "loss": 0.6324,
      "step": 14307
    },
    {
      "epoch": 0.5929794023788801,
      "grad_norm": 0.435937762260437,
      "learning_rate": 2.0353102076339674e-06,
      "loss": 0.6826,
      "step": 14308
    },
    {
      "epoch": 0.5930208462845539,
      "grad_norm": 0.3992414176464081,
      "learning_rate": 2.0351029881055992e-06,
      "loss": 0.6014,
      "step": 14309
    },
    {
      "epoch": 0.5930622901902275,
      "grad_norm": 0.4087291657924652,
      "learning_rate": 2.0348957685772306e-06,
      "loss": 0.6705,
      "step": 14310
    },
    {
      "epoch": 0.5931037340959012,
      "grad_norm": 0.41874590516090393,
      "learning_rate": 2.0346885490488624e-06,
      "loss": 0.7043,
      "step": 14311
    },
    {
      "epoch": 0.5931451780015748,
      "grad_norm": 0.40523335337638855,
      "learning_rate": 2.0344813295204942e-06,
      "loss": 0.6589,
      "step": 14312
    },
    {
      "epoch": 0.5931866219072486,
      "grad_norm": 0.4835190176963806,
      "learning_rate": 2.034274109992126e-06,
      "loss": 0.7689,
      "step": 14313
    },
    {
      "epoch": 0.5932280658129222,
      "grad_norm": 0.41533568501472473,
      "learning_rate": 2.0340668904637574e-06,
      "loss": 0.7424,
      "step": 14314
    },
    {
      "epoch": 0.5932695097185959,
      "grad_norm": 0.46231433749198914,
      "learning_rate": 2.0338596709353892e-06,
      "loss": 0.7261,
      "step": 14315
    },
    {
      "epoch": 0.5933109536242696,
      "grad_norm": 0.4329625964164734,
      "learning_rate": 2.0336524514070206e-06,
      "loss": 0.7162,
      "step": 14316
    },
    {
      "epoch": 0.5933523975299432,
      "grad_norm": 0.4204372465610504,
      "learning_rate": 2.0334452318786524e-06,
      "loss": 0.7168,
      "step": 14317
    },
    {
      "epoch": 0.5933938414356169,
      "grad_norm": 0.4126313328742981,
      "learning_rate": 2.033238012350284e-06,
      "loss": 0.6572,
      "step": 14318
    },
    {
      "epoch": 0.5934352853412905,
      "grad_norm": 0.3935967981815338,
      "learning_rate": 2.0330307928219156e-06,
      "loss": 0.6528,
      "step": 14319
    },
    {
      "epoch": 0.5934767292469643,
      "grad_norm": 0.42331936955451965,
      "learning_rate": 2.0328235732935474e-06,
      "loss": 0.6707,
      "step": 14320
    },
    {
      "epoch": 0.5935181731526379,
      "grad_norm": 0.46018701791763306,
      "learning_rate": 2.0326163537651792e-06,
      "loss": 0.6904,
      "step": 14321
    },
    {
      "epoch": 0.5935596170583116,
      "grad_norm": 0.40509653091430664,
      "learning_rate": 2.0324091342368106e-06,
      "loss": 0.6567,
      "step": 14322
    },
    {
      "epoch": 0.5936010609639852,
      "grad_norm": 0.39532938599586487,
      "learning_rate": 2.0322019147084424e-06,
      "loss": 0.6555,
      "step": 14323
    },
    {
      "epoch": 0.593642504869659,
      "grad_norm": 0.38313978910446167,
      "learning_rate": 2.031994695180074e-06,
      "loss": 0.7109,
      "step": 14324
    },
    {
      "epoch": 0.5936839487753326,
      "grad_norm": 0.44797050952911377,
      "learning_rate": 2.0317874756517056e-06,
      "loss": 0.674,
      "step": 14325
    },
    {
      "epoch": 0.5937253926810062,
      "grad_norm": 0.41121363639831543,
      "learning_rate": 2.031580256123337e-06,
      "loss": 0.7063,
      "step": 14326
    },
    {
      "epoch": 0.59376683658668,
      "grad_norm": 0.40165460109710693,
      "learning_rate": 2.031373036594969e-06,
      "loss": 0.6633,
      "step": 14327
    },
    {
      "epoch": 0.5938082804923536,
      "grad_norm": 0.3795650601387024,
      "learning_rate": 2.0311658170666006e-06,
      "loss": 0.6321,
      "step": 14328
    },
    {
      "epoch": 0.5938497243980273,
      "grad_norm": 0.417758047580719,
      "learning_rate": 2.030958597538232e-06,
      "loss": 0.6343,
      "step": 14329
    },
    {
      "epoch": 0.5938911683037009,
      "grad_norm": 0.4302422106266022,
      "learning_rate": 2.030751378009864e-06,
      "loss": 0.6797,
      "step": 14330
    },
    {
      "epoch": 0.5939326122093747,
      "grad_norm": 0.4040430784225464,
      "learning_rate": 2.0305441584814956e-06,
      "loss": 0.6449,
      "step": 14331
    },
    {
      "epoch": 0.5939740561150483,
      "grad_norm": 0.43822744488716125,
      "learning_rate": 2.030336938953127e-06,
      "loss": 0.6913,
      "step": 14332
    },
    {
      "epoch": 0.5940155000207219,
      "grad_norm": 0.39323458075523376,
      "learning_rate": 2.030129719424759e-06,
      "loss": 0.6752,
      "step": 14333
    },
    {
      "epoch": 0.5940569439263956,
      "grad_norm": 0.4040904641151428,
      "learning_rate": 2.0299224998963902e-06,
      "loss": 0.6654,
      "step": 14334
    },
    {
      "epoch": 0.5940983878320693,
      "grad_norm": 0.4468677341938019,
      "learning_rate": 2.029715280368022e-06,
      "loss": 0.6488,
      "step": 14335
    },
    {
      "epoch": 0.594139831737743,
      "grad_norm": 0.3944384455680847,
      "learning_rate": 2.029508060839654e-06,
      "loss": 0.6292,
      "step": 14336
    },
    {
      "epoch": 0.5941812756434166,
      "grad_norm": 0.39719298481941223,
      "learning_rate": 2.0293008413112852e-06,
      "loss": 0.6509,
      "step": 14337
    },
    {
      "epoch": 0.5942227195490903,
      "grad_norm": 0.42398104071617126,
      "learning_rate": 2.029093621782917e-06,
      "loss": 0.653,
      "step": 14338
    },
    {
      "epoch": 0.594264163454764,
      "grad_norm": 0.43199828267097473,
      "learning_rate": 2.028886402254549e-06,
      "loss": 0.6903,
      "step": 14339
    },
    {
      "epoch": 0.5943056073604377,
      "grad_norm": 0.41490867733955383,
      "learning_rate": 2.0286791827261802e-06,
      "loss": 0.7046,
      "step": 14340
    },
    {
      "epoch": 0.5943470512661113,
      "grad_norm": 0.3960203528404236,
      "learning_rate": 2.028471963197812e-06,
      "loss": 0.6527,
      "step": 14341
    },
    {
      "epoch": 0.5943884951717849,
      "grad_norm": 0.39756324887275696,
      "learning_rate": 2.0282647436694434e-06,
      "loss": 0.6592,
      "step": 14342
    },
    {
      "epoch": 0.5944299390774587,
      "grad_norm": 0.4263734221458435,
      "learning_rate": 2.0280575241410752e-06,
      "loss": 0.6831,
      "step": 14343
    },
    {
      "epoch": 0.5944713829831323,
      "grad_norm": 0.43112796545028687,
      "learning_rate": 2.0278503046127066e-06,
      "loss": 0.6477,
      "step": 14344
    },
    {
      "epoch": 0.594512826888806,
      "grad_norm": 0.4024803042411804,
      "learning_rate": 2.0276430850843384e-06,
      "loss": 0.6594,
      "step": 14345
    },
    {
      "epoch": 0.5945542707944796,
      "grad_norm": 0.43984800577163696,
      "learning_rate": 2.0274358655559702e-06,
      "loss": 0.7222,
      "step": 14346
    },
    {
      "epoch": 0.5945957147001534,
      "grad_norm": 0.4023248255252838,
      "learning_rate": 2.027228646027602e-06,
      "loss": 0.6915,
      "step": 14347
    },
    {
      "epoch": 0.594637158605827,
      "grad_norm": 0.38585853576660156,
      "learning_rate": 2.0270214264992334e-06,
      "loss": 0.6384,
      "step": 14348
    },
    {
      "epoch": 0.5946786025115007,
      "grad_norm": 0.40259602665901184,
      "learning_rate": 2.0268142069708652e-06,
      "loss": 0.7063,
      "step": 14349
    },
    {
      "epoch": 0.5947200464171744,
      "grad_norm": 0.3822415769100189,
      "learning_rate": 2.0266069874424966e-06,
      "loss": 0.6473,
      "step": 14350
    },
    {
      "epoch": 0.594761490322848,
      "grad_norm": 0.38178327679634094,
      "learning_rate": 2.0263997679141284e-06,
      "loss": 0.6323,
      "step": 14351
    },
    {
      "epoch": 0.5948029342285217,
      "grad_norm": 0.4423728287220001,
      "learning_rate": 2.02619254838576e-06,
      "loss": 0.733,
      "step": 14352
    },
    {
      "epoch": 0.5948443781341953,
      "grad_norm": 0.4118400812149048,
      "learning_rate": 2.0259853288573916e-06,
      "loss": 0.6709,
      "step": 14353
    },
    {
      "epoch": 0.5948858220398691,
      "grad_norm": 0.3957553803920746,
      "learning_rate": 2.0257781093290234e-06,
      "loss": 0.657,
      "step": 14354
    },
    {
      "epoch": 0.5949272659455427,
      "grad_norm": 0.4174366891384125,
      "learning_rate": 2.025570889800655e-06,
      "loss": 0.6953,
      "step": 14355
    },
    {
      "epoch": 0.5949687098512164,
      "grad_norm": 0.4107711613178253,
      "learning_rate": 2.0253636702722866e-06,
      "loss": 0.6799,
      "step": 14356
    },
    {
      "epoch": 0.59501015375689,
      "grad_norm": 0.4277287721633911,
      "learning_rate": 2.0251564507439184e-06,
      "loss": 0.6824,
      "step": 14357
    },
    {
      "epoch": 0.5950515976625638,
      "grad_norm": 0.38754233717918396,
      "learning_rate": 2.02494923121555e-06,
      "loss": 0.6927,
      "step": 14358
    },
    {
      "epoch": 0.5950930415682374,
      "grad_norm": 0.402620404958725,
      "learning_rate": 2.0247420116871816e-06,
      "loss": 0.6436,
      "step": 14359
    },
    {
      "epoch": 0.595134485473911,
      "grad_norm": 0.41613373160362244,
      "learning_rate": 2.024534792158813e-06,
      "loss": 0.6388,
      "step": 14360
    },
    {
      "epoch": 0.5951759293795847,
      "grad_norm": 0.40306299924850464,
      "learning_rate": 2.024327572630445e-06,
      "loss": 0.6335,
      "step": 14361
    },
    {
      "epoch": 0.5952173732852584,
      "grad_norm": 0.4143815040588379,
      "learning_rate": 2.0241203531020766e-06,
      "loss": 0.6877,
      "step": 14362
    },
    {
      "epoch": 0.5952588171909321,
      "grad_norm": 0.4262343645095825,
      "learning_rate": 2.023913133573708e-06,
      "loss": 0.729,
      "step": 14363
    },
    {
      "epoch": 0.5953002610966057,
      "grad_norm": 0.4170742332935333,
      "learning_rate": 2.02370591404534e-06,
      "loss": 0.6917,
      "step": 14364
    },
    {
      "epoch": 0.5953417050022795,
      "grad_norm": 0.3892902433872223,
      "learning_rate": 2.0234986945169716e-06,
      "loss": 0.6572,
      "step": 14365
    },
    {
      "epoch": 0.5953831489079531,
      "grad_norm": 0.410569965839386,
      "learning_rate": 2.023291474988603e-06,
      "loss": 0.6975,
      "step": 14366
    },
    {
      "epoch": 0.5954245928136268,
      "grad_norm": 0.4237006902694702,
      "learning_rate": 2.023084255460235e-06,
      "loss": 0.6448,
      "step": 14367
    },
    {
      "epoch": 0.5954660367193004,
      "grad_norm": 0.4280565083026886,
      "learning_rate": 2.0228770359318662e-06,
      "loss": 0.6605,
      "step": 14368
    },
    {
      "epoch": 0.595507480624974,
      "grad_norm": 0.40148481726646423,
      "learning_rate": 2.022669816403498e-06,
      "loss": 0.6316,
      "step": 14369
    },
    {
      "epoch": 0.5955489245306478,
      "grad_norm": 0.4305514097213745,
      "learning_rate": 2.02246259687513e-06,
      "loss": 0.6743,
      "step": 14370
    },
    {
      "epoch": 0.5955903684363214,
      "grad_norm": 0.3879026770591736,
      "learning_rate": 2.0222553773467612e-06,
      "loss": 0.6876,
      "step": 14371
    },
    {
      "epoch": 0.5956318123419951,
      "grad_norm": 0.4012109637260437,
      "learning_rate": 2.022048157818393e-06,
      "loss": 0.6307,
      "step": 14372
    },
    {
      "epoch": 0.5956732562476688,
      "grad_norm": 0.4105861485004425,
      "learning_rate": 2.0218409382900244e-06,
      "loss": 0.6792,
      "step": 14373
    },
    {
      "epoch": 0.5957147001533425,
      "grad_norm": 0.43092086911201477,
      "learning_rate": 2.0216337187616562e-06,
      "loss": 0.7158,
      "step": 14374
    },
    {
      "epoch": 0.5957561440590161,
      "grad_norm": 0.42445626854896545,
      "learning_rate": 2.021426499233288e-06,
      "loss": 0.6763,
      "step": 14375
    },
    {
      "epoch": 0.5957975879646898,
      "grad_norm": 0.3859517276287079,
      "learning_rate": 2.0212192797049194e-06,
      "loss": 0.6846,
      "step": 14376
    },
    {
      "epoch": 0.5958390318703635,
      "grad_norm": 0.4226225018501282,
      "learning_rate": 2.0210120601765512e-06,
      "loss": 0.6646,
      "step": 14377
    },
    {
      "epoch": 0.5958804757760371,
      "grad_norm": 0.4132973551750183,
      "learning_rate": 2.020804840648183e-06,
      "loss": 0.7074,
      "step": 14378
    },
    {
      "epoch": 0.5959219196817108,
      "grad_norm": 0.3989001512527466,
      "learning_rate": 2.0205976211198144e-06,
      "loss": 0.6801,
      "step": 14379
    },
    {
      "epoch": 0.5959633635873844,
      "grad_norm": 0.38745129108428955,
      "learning_rate": 2.0203904015914462e-06,
      "loss": 0.6508,
      "step": 14380
    },
    {
      "epoch": 0.5960048074930582,
      "grad_norm": 0.4824681282043457,
      "learning_rate": 2.0201831820630776e-06,
      "loss": 0.7175,
      "step": 14381
    },
    {
      "epoch": 0.5960462513987318,
      "grad_norm": 0.37309813499450684,
      "learning_rate": 2.0199759625347094e-06,
      "loss": 0.6705,
      "step": 14382
    },
    {
      "epoch": 0.5960876953044055,
      "grad_norm": 0.4019153118133545,
      "learning_rate": 2.0197687430063412e-06,
      "loss": 0.72,
      "step": 14383
    },
    {
      "epoch": 0.5961291392100792,
      "grad_norm": 0.39924439787864685,
      "learning_rate": 2.0195615234779726e-06,
      "loss": 0.665,
      "step": 14384
    },
    {
      "epoch": 0.5961705831157529,
      "grad_norm": 0.41613510251045227,
      "learning_rate": 2.0193543039496044e-06,
      "loss": 0.675,
      "step": 14385
    },
    {
      "epoch": 0.5962120270214265,
      "grad_norm": 0.43853211402893066,
      "learning_rate": 2.019147084421236e-06,
      "loss": 0.7218,
      "step": 14386
    },
    {
      "epoch": 0.5962534709271001,
      "grad_norm": 0.42514365911483765,
      "learning_rate": 2.0189398648928676e-06,
      "loss": 0.6516,
      "step": 14387
    },
    {
      "epoch": 0.5962949148327739,
      "grad_norm": 0.3866899907588959,
      "learning_rate": 2.0187326453644994e-06,
      "loss": 0.6561,
      "step": 14388
    },
    {
      "epoch": 0.5963363587384475,
      "grad_norm": 0.4057970345020294,
      "learning_rate": 2.018525425836131e-06,
      "loss": 0.7063,
      "step": 14389
    },
    {
      "epoch": 0.5963778026441212,
      "grad_norm": 0.40204766392707825,
      "learning_rate": 2.0183182063077626e-06,
      "loss": 0.6464,
      "step": 14390
    },
    {
      "epoch": 0.5964192465497948,
      "grad_norm": 0.45571938157081604,
      "learning_rate": 2.0181109867793944e-06,
      "loss": 0.7185,
      "step": 14391
    },
    {
      "epoch": 0.5964606904554686,
      "grad_norm": 0.41326677799224854,
      "learning_rate": 2.017903767251026e-06,
      "loss": 0.7009,
      "step": 14392
    },
    {
      "epoch": 0.5965021343611422,
      "grad_norm": 0.4094535708427429,
      "learning_rate": 2.0176965477226576e-06,
      "loss": 0.7144,
      "step": 14393
    },
    {
      "epoch": 0.5965435782668158,
      "grad_norm": 0.4272117614746094,
      "learning_rate": 2.017489328194289e-06,
      "loss": 0.6992,
      "step": 14394
    },
    {
      "epoch": 0.5965850221724895,
      "grad_norm": 0.4230491518974304,
      "learning_rate": 2.017282108665921e-06,
      "loss": 0.6652,
      "step": 14395
    },
    {
      "epoch": 0.5966264660781632,
      "grad_norm": 0.43777748942375183,
      "learning_rate": 2.0170748891375526e-06,
      "loss": 0.7042,
      "step": 14396
    },
    {
      "epoch": 0.5966679099838369,
      "grad_norm": 0.40034741163253784,
      "learning_rate": 2.016867669609184e-06,
      "loss": 0.6714,
      "step": 14397
    },
    {
      "epoch": 0.5967093538895105,
      "grad_norm": 0.4250170886516571,
      "learning_rate": 2.016660450080816e-06,
      "loss": 0.6992,
      "step": 14398
    },
    {
      "epoch": 0.5967507977951843,
      "grad_norm": 0.41979166865348816,
      "learning_rate": 2.0164532305524472e-06,
      "loss": 0.6688,
      "step": 14399
    },
    {
      "epoch": 0.5967922417008579,
      "grad_norm": 0.42394569516181946,
      "learning_rate": 2.016246011024079e-06,
      "loss": 0.7651,
      "step": 14400
    },
    {
      "epoch": 0.5968336856065316,
      "grad_norm": 0.4051402807235718,
      "learning_rate": 2.016038791495711e-06,
      "loss": 0.6473,
      "step": 14401
    },
    {
      "epoch": 0.5968751295122052,
      "grad_norm": 0.40897953510284424,
      "learning_rate": 2.0158315719673422e-06,
      "loss": 0.7021,
      "step": 14402
    },
    {
      "epoch": 0.5969165734178788,
      "grad_norm": 0.447746604681015,
      "learning_rate": 2.015624352438974e-06,
      "loss": 0.7339,
      "step": 14403
    },
    {
      "epoch": 0.5969580173235526,
      "grad_norm": 0.41407710313796997,
      "learning_rate": 2.015417132910606e-06,
      "loss": 0.6638,
      "step": 14404
    },
    {
      "epoch": 0.5969994612292262,
      "grad_norm": 0.3679262697696686,
      "learning_rate": 2.0152099133822372e-06,
      "loss": 0.6106,
      "step": 14405
    },
    {
      "epoch": 0.5970409051348999,
      "grad_norm": 0.3960670530796051,
      "learning_rate": 2.015002693853869e-06,
      "loss": 0.6482,
      "step": 14406
    },
    {
      "epoch": 0.5970823490405736,
      "grad_norm": 0.40117040276527405,
      "learning_rate": 2.0147954743255004e-06,
      "loss": 0.6354,
      "step": 14407
    },
    {
      "epoch": 0.5971237929462473,
      "grad_norm": 0.39229118824005127,
      "learning_rate": 2.0145882547971322e-06,
      "loss": 0.6498,
      "step": 14408
    },
    {
      "epoch": 0.5971652368519209,
      "grad_norm": 0.37588411569595337,
      "learning_rate": 2.014381035268764e-06,
      "loss": 0.6459,
      "step": 14409
    },
    {
      "epoch": 0.5972066807575946,
      "grad_norm": 0.40859487652778625,
      "learning_rate": 2.0141738157403954e-06,
      "loss": 0.6667,
      "step": 14410
    },
    {
      "epoch": 0.5972481246632683,
      "grad_norm": 0.3877080976963043,
      "learning_rate": 2.0139665962120272e-06,
      "loss": 0.7,
      "step": 14411
    },
    {
      "epoch": 0.5972895685689419,
      "grad_norm": 0.4293051064014435,
      "learning_rate": 2.013759376683659e-06,
      "loss": 0.6704,
      "step": 14412
    },
    {
      "epoch": 0.5973310124746156,
      "grad_norm": 0.40750938653945923,
      "learning_rate": 2.0135521571552904e-06,
      "loss": 0.7,
      "step": 14413
    },
    {
      "epoch": 0.5973724563802892,
      "grad_norm": 0.38827887177467346,
      "learning_rate": 2.0133449376269222e-06,
      "loss": 0.6345,
      "step": 14414
    },
    {
      "epoch": 0.597413900285963,
      "grad_norm": 0.40926864743232727,
      "learning_rate": 2.0131377180985536e-06,
      "loss": 0.7163,
      "step": 14415
    },
    {
      "epoch": 0.5974553441916366,
      "grad_norm": 0.3905308246612549,
      "learning_rate": 2.0129304985701854e-06,
      "loss": 0.6689,
      "step": 14416
    },
    {
      "epoch": 0.5974967880973103,
      "grad_norm": 0.3963407874107361,
      "learning_rate": 2.0127232790418172e-06,
      "loss": 0.6273,
      "step": 14417
    },
    {
      "epoch": 0.597538232002984,
      "grad_norm": 0.40965408086776733,
      "learning_rate": 2.0125160595134486e-06,
      "loss": 0.6429,
      "step": 14418
    },
    {
      "epoch": 0.5975796759086577,
      "grad_norm": 0.7684953808784485,
      "learning_rate": 2.0123088399850804e-06,
      "loss": 0.6788,
      "step": 14419
    },
    {
      "epoch": 0.5976211198143313,
      "grad_norm": 0.43025216460227966,
      "learning_rate": 2.012101620456712e-06,
      "loss": 0.6809,
      "step": 14420
    },
    {
      "epoch": 0.5976625637200049,
      "grad_norm": 0.4062935709953308,
      "learning_rate": 2.0118944009283436e-06,
      "loss": 0.6544,
      "step": 14421
    },
    {
      "epoch": 0.5977040076256787,
      "grad_norm": 0.4146495461463928,
      "learning_rate": 2.0116871813999754e-06,
      "loss": 0.6593,
      "step": 14422
    },
    {
      "epoch": 0.5977454515313523,
      "grad_norm": 0.4446333646774292,
      "learning_rate": 2.011479961871607e-06,
      "loss": 0.7305,
      "step": 14423
    },
    {
      "epoch": 0.597786895437026,
      "grad_norm": 0.36641445755958557,
      "learning_rate": 2.0112727423432386e-06,
      "loss": 0.6379,
      "step": 14424
    },
    {
      "epoch": 0.5978283393426996,
      "grad_norm": 0.41936540603637695,
      "learning_rate": 2.01106552281487e-06,
      "loss": 0.7214,
      "step": 14425
    },
    {
      "epoch": 0.5978697832483734,
      "grad_norm": 0.3986991047859192,
      "learning_rate": 2.010858303286502e-06,
      "loss": 0.6935,
      "step": 14426
    },
    {
      "epoch": 0.597911227154047,
      "grad_norm": 0.43886488676071167,
      "learning_rate": 2.0106510837581336e-06,
      "loss": 0.6721,
      "step": 14427
    },
    {
      "epoch": 0.5979526710597207,
      "grad_norm": 0.38111263513565063,
      "learning_rate": 2.010443864229765e-06,
      "loss": 0.6093,
      "step": 14428
    },
    {
      "epoch": 0.5979941149653943,
      "grad_norm": 0.39000430703163147,
      "learning_rate": 2.010236644701397e-06,
      "loss": 0.6649,
      "step": 14429
    },
    {
      "epoch": 0.598035558871068,
      "grad_norm": 0.44159260392189026,
      "learning_rate": 2.0100294251730286e-06,
      "loss": 0.6953,
      "step": 14430
    },
    {
      "epoch": 0.5980770027767417,
      "grad_norm": 0.40329355001449585,
      "learning_rate": 2.00982220564466e-06,
      "loss": 0.6765,
      "step": 14431
    },
    {
      "epoch": 0.5981184466824153,
      "grad_norm": 0.41361838579177856,
      "learning_rate": 2.009614986116292e-06,
      "loss": 0.7236,
      "step": 14432
    },
    {
      "epoch": 0.598159890588089,
      "grad_norm": 0.42035967111587524,
      "learning_rate": 2.0094077665879232e-06,
      "loss": 0.6798,
      "step": 14433
    },
    {
      "epoch": 0.5982013344937627,
      "grad_norm": 0.3935181200504303,
      "learning_rate": 2.009200547059555e-06,
      "loss": 0.6393,
      "step": 14434
    },
    {
      "epoch": 0.5982427783994364,
      "grad_norm": 0.4304755628108978,
      "learning_rate": 2.008993327531187e-06,
      "loss": 0.686,
      "step": 14435
    },
    {
      "epoch": 0.59828422230511,
      "grad_norm": 0.43664875626564026,
      "learning_rate": 2.0087861080028182e-06,
      "loss": 0.6815,
      "step": 14436
    },
    {
      "epoch": 0.5983256662107838,
      "grad_norm": 0.42373546957969666,
      "learning_rate": 2.00857888847445e-06,
      "loss": 0.6608,
      "step": 14437
    },
    {
      "epoch": 0.5983671101164574,
      "grad_norm": 0.4077330529689789,
      "learning_rate": 2.008371668946082e-06,
      "loss": 0.6611,
      "step": 14438
    },
    {
      "epoch": 0.598408554022131,
      "grad_norm": 0.456106573343277,
      "learning_rate": 2.0081644494177132e-06,
      "loss": 0.7551,
      "step": 14439
    },
    {
      "epoch": 0.5984499979278047,
      "grad_norm": 0.3921237885951996,
      "learning_rate": 2.007957229889345e-06,
      "loss": 0.668,
      "step": 14440
    },
    {
      "epoch": 0.5984914418334784,
      "grad_norm": 0.44003742933273315,
      "learning_rate": 2.0077500103609764e-06,
      "loss": 0.7216,
      "step": 14441
    },
    {
      "epoch": 0.5985328857391521,
      "grad_norm": 0.39004865288734436,
      "learning_rate": 2.0075427908326082e-06,
      "loss": 0.6405,
      "step": 14442
    },
    {
      "epoch": 0.5985743296448257,
      "grad_norm": 0.4339534342288971,
      "learning_rate": 2.00733557130424e-06,
      "loss": 0.6891,
      "step": 14443
    },
    {
      "epoch": 0.5986157735504994,
      "grad_norm": 0.4058506488800049,
      "learning_rate": 2.0071283517758714e-06,
      "loss": 0.6937,
      "step": 14444
    },
    {
      "epoch": 0.5986572174561731,
      "grad_norm": 0.41155415773391724,
      "learning_rate": 2.0069211322475032e-06,
      "loss": 0.6907,
      "step": 14445
    },
    {
      "epoch": 0.5986986613618468,
      "grad_norm": 0.3971969485282898,
      "learning_rate": 2.006713912719135e-06,
      "loss": 0.6659,
      "step": 14446
    },
    {
      "epoch": 0.5987401052675204,
      "grad_norm": 0.4143161177635193,
      "learning_rate": 2.0065066931907664e-06,
      "loss": 0.6398,
      "step": 14447
    },
    {
      "epoch": 0.598781549173194,
      "grad_norm": 0.4407001733779907,
      "learning_rate": 2.0062994736623982e-06,
      "loss": 0.6849,
      "step": 14448
    },
    {
      "epoch": 0.5988229930788678,
      "grad_norm": 0.3922001123428345,
      "learning_rate": 2.0060922541340296e-06,
      "loss": 0.698,
      "step": 14449
    },
    {
      "epoch": 0.5988644369845414,
      "grad_norm": 0.4262036979198456,
      "learning_rate": 2.0058850346056614e-06,
      "loss": 0.6666,
      "step": 14450
    },
    {
      "epoch": 0.5989058808902151,
      "grad_norm": 0.4321430027484894,
      "learning_rate": 2.005677815077293e-06,
      "loss": 0.697,
      "step": 14451
    },
    {
      "epoch": 0.5989473247958887,
      "grad_norm": 0.4366486072540283,
      "learning_rate": 2.0054705955489246e-06,
      "loss": 0.682,
      "step": 14452
    },
    {
      "epoch": 0.5989887687015625,
      "grad_norm": 0.44760021567344666,
      "learning_rate": 2.0052633760205564e-06,
      "loss": 0.713,
      "step": 14453
    },
    {
      "epoch": 0.5990302126072361,
      "grad_norm": 0.4016120433807373,
      "learning_rate": 2.0050561564921883e-06,
      "loss": 0.6594,
      "step": 14454
    },
    {
      "epoch": 0.5990716565129097,
      "grad_norm": 0.4284817576408386,
      "learning_rate": 2.0048489369638196e-06,
      "loss": 0.6284,
      "step": 14455
    },
    {
      "epoch": 0.5991131004185835,
      "grad_norm": 0.3893854022026062,
      "learning_rate": 2.0046417174354514e-06,
      "loss": 0.667,
      "step": 14456
    },
    {
      "epoch": 0.5991545443242571,
      "grad_norm": 0.5569701790809631,
      "learning_rate": 2.004434497907083e-06,
      "loss": 0.6703,
      "step": 14457
    },
    {
      "epoch": 0.5991959882299308,
      "grad_norm": 0.41954365372657776,
      "learning_rate": 2.0042272783787146e-06,
      "loss": 0.6752,
      "step": 14458
    },
    {
      "epoch": 0.5992374321356044,
      "grad_norm": 0.37531939148902893,
      "learning_rate": 2.004020058850346e-06,
      "loss": 0.6366,
      "step": 14459
    },
    {
      "epoch": 0.5992788760412782,
      "grad_norm": 0.3718879222869873,
      "learning_rate": 2.003812839321978e-06,
      "loss": 0.6395,
      "step": 14460
    },
    {
      "epoch": 0.5993203199469518,
      "grad_norm": 0.46860793232917786,
      "learning_rate": 2.0036056197936096e-06,
      "loss": 0.6649,
      "step": 14461
    },
    {
      "epoch": 0.5993617638526255,
      "grad_norm": 0.4035191535949707,
      "learning_rate": 2.003398400265241e-06,
      "loss": 0.6808,
      "step": 14462
    },
    {
      "epoch": 0.5994032077582991,
      "grad_norm": 0.3986499607563019,
      "learning_rate": 2.003191180736873e-06,
      "loss": 0.6819,
      "step": 14463
    },
    {
      "epoch": 0.5994446516639728,
      "grad_norm": 0.41130176186561584,
      "learning_rate": 2.0029839612085047e-06,
      "loss": 0.6858,
      "step": 14464
    },
    {
      "epoch": 0.5994860955696465,
      "grad_norm": 0.3908282220363617,
      "learning_rate": 2.002776741680136e-06,
      "loss": 0.7253,
      "step": 14465
    },
    {
      "epoch": 0.5995275394753201,
      "grad_norm": 0.3756026029586792,
      "learning_rate": 2.002569522151768e-06,
      "loss": 0.6636,
      "step": 14466
    },
    {
      "epoch": 0.5995689833809938,
      "grad_norm": 0.418349951505661,
      "learning_rate": 2.0023623026233992e-06,
      "loss": 0.7065,
      "step": 14467
    },
    {
      "epoch": 0.5996104272866675,
      "grad_norm": 0.40787529945373535,
      "learning_rate": 2.002155083095031e-06,
      "loss": 0.6614,
      "step": 14468
    },
    {
      "epoch": 0.5996518711923412,
      "grad_norm": 0.43600380420684814,
      "learning_rate": 2.001947863566663e-06,
      "loss": 0.7219,
      "step": 14469
    },
    {
      "epoch": 0.5996933150980148,
      "grad_norm": 0.436775803565979,
      "learning_rate": 2.0017406440382942e-06,
      "loss": 0.6736,
      "step": 14470
    },
    {
      "epoch": 0.5997347590036886,
      "grad_norm": 0.3963083326816559,
      "learning_rate": 2.001533424509926e-06,
      "loss": 0.6956,
      "step": 14471
    },
    {
      "epoch": 0.5997762029093622,
      "grad_norm": 0.4196145236492157,
      "learning_rate": 2.001326204981558e-06,
      "loss": 0.6656,
      "step": 14472
    },
    {
      "epoch": 0.5998176468150358,
      "grad_norm": 0.3632254898548126,
      "learning_rate": 2.0011189854531892e-06,
      "loss": 0.616,
      "step": 14473
    },
    {
      "epoch": 0.5998590907207095,
      "grad_norm": 0.4482094347476959,
      "learning_rate": 2.000911765924821e-06,
      "loss": 0.6941,
      "step": 14474
    },
    {
      "epoch": 0.5999005346263832,
      "grad_norm": 0.406034916639328,
      "learning_rate": 2.0007045463964524e-06,
      "loss": 0.6583,
      "step": 14475
    },
    {
      "epoch": 0.5999419785320569,
      "grad_norm": 0.39717498421669006,
      "learning_rate": 2.0004973268680842e-06,
      "loss": 0.6941,
      "step": 14476
    },
    {
      "epoch": 0.5999834224377305,
      "grad_norm": 0.41896647214889526,
      "learning_rate": 2.0002901073397156e-06,
      "loss": 0.6892,
      "step": 14477
    },
    {
      "epoch": 0.6000248663434042,
      "grad_norm": 0.4048294126987457,
      "learning_rate": 2.0000828878113474e-06,
      "loss": 0.7074,
      "step": 14478
    },
    {
      "epoch": 0.6000663102490779,
      "grad_norm": 0.4020443260669708,
      "learning_rate": 1.9998756682829792e-06,
      "loss": 0.6508,
      "step": 14479
    },
    {
      "epoch": 0.6001077541547516,
      "grad_norm": 0.40686270594596863,
      "learning_rate": 1.999668448754611e-06,
      "loss": 0.7131,
      "step": 14480
    },
    {
      "epoch": 0.6001491980604252,
      "grad_norm": 0.3927430510520935,
      "learning_rate": 1.9994612292262424e-06,
      "loss": 0.6857,
      "step": 14481
    },
    {
      "epoch": 0.6001906419660988,
      "grad_norm": 0.440328449010849,
      "learning_rate": 1.9992540096978743e-06,
      "loss": 0.7278,
      "step": 14482
    },
    {
      "epoch": 0.6002320858717726,
      "grad_norm": 0.40897300839424133,
      "learning_rate": 1.9990467901695056e-06,
      "loss": 0.642,
      "step": 14483
    },
    {
      "epoch": 0.6002735297774462,
      "grad_norm": 0.43869850039482117,
      "learning_rate": 1.9988395706411374e-06,
      "loss": 0.6395,
      "step": 14484
    },
    {
      "epoch": 0.6003149736831199,
      "grad_norm": 0.407537579536438,
      "learning_rate": 1.998632351112769e-06,
      "loss": 0.6978,
      "step": 14485
    },
    {
      "epoch": 0.6003564175887935,
      "grad_norm": 0.40908050537109375,
      "learning_rate": 1.9984251315844006e-06,
      "loss": 0.6882,
      "step": 14486
    },
    {
      "epoch": 0.6003978614944673,
      "grad_norm": 0.42139047384262085,
      "learning_rate": 1.9982179120560325e-06,
      "loss": 0.6516,
      "step": 14487
    },
    {
      "epoch": 0.6004393054001409,
      "grad_norm": 0.42272502183914185,
      "learning_rate": 1.9980106925276643e-06,
      "loss": 0.7177,
      "step": 14488
    },
    {
      "epoch": 0.6004807493058146,
      "grad_norm": 0.40550610423088074,
      "learning_rate": 1.9978034729992956e-06,
      "loss": 0.7009,
      "step": 14489
    },
    {
      "epoch": 0.6005221932114883,
      "grad_norm": 0.4227939546108246,
      "learning_rate": 1.9975962534709275e-06,
      "loss": 0.7003,
      "step": 14490
    },
    {
      "epoch": 0.6005636371171619,
      "grad_norm": 0.41016584634780884,
      "learning_rate": 1.997389033942559e-06,
      "loss": 0.6743,
      "step": 14491
    },
    {
      "epoch": 0.6006050810228356,
      "grad_norm": 0.4081968069076538,
      "learning_rate": 1.9971818144141907e-06,
      "loss": 0.6488,
      "step": 14492
    },
    {
      "epoch": 0.6006465249285092,
      "grad_norm": 0.39581453800201416,
      "learning_rate": 1.996974594885822e-06,
      "loss": 0.6826,
      "step": 14493
    },
    {
      "epoch": 0.600687968834183,
      "grad_norm": 0.4807567894458771,
      "learning_rate": 1.996767375357454e-06,
      "loss": 0.7126,
      "step": 14494
    },
    {
      "epoch": 0.6007294127398566,
      "grad_norm": 0.3886212408542633,
      "learning_rate": 1.9965601558290857e-06,
      "loss": 0.6493,
      "step": 14495
    },
    {
      "epoch": 0.6007708566455303,
      "grad_norm": 0.4078727960586548,
      "learning_rate": 1.996352936300717e-06,
      "loss": 0.6946,
      "step": 14496
    },
    {
      "epoch": 0.6008123005512039,
      "grad_norm": 0.4337066411972046,
      "learning_rate": 1.996145716772349e-06,
      "loss": 0.7166,
      "step": 14497
    },
    {
      "epoch": 0.6008537444568777,
      "grad_norm": 0.39529719948768616,
      "learning_rate": 1.9959384972439807e-06,
      "loss": 0.6611,
      "step": 14498
    },
    {
      "epoch": 0.6008951883625513,
      "grad_norm": 0.39812585711479187,
      "learning_rate": 1.995731277715612e-06,
      "loss": 0.6567,
      "step": 14499
    },
    {
      "epoch": 0.6009366322682249,
      "grad_norm": 0.37952789664268494,
      "learning_rate": 1.995524058187244e-06,
      "loss": 0.6606,
      "step": 14500
    },
    {
      "epoch": 0.6009780761738986,
      "grad_norm": 0.4111202657222748,
      "learning_rate": 1.9953168386588752e-06,
      "loss": 0.6787,
      "step": 14501
    },
    {
      "epoch": 0.6010195200795723,
      "grad_norm": 0.4247913658618927,
      "learning_rate": 1.995109619130507e-06,
      "loss": 0.6807,
      "step": 14502
    },
    {
      "epoch": 0.601060963985246,
      "grad_norm": 0.3853343427181244,
      "learning_rate": 1.9949023996021384e-06,
      "loss": 0.6595,
      "step": 14503
    },
    {
      "epoch": 0.6011024078909196,
      "grad_norm": 0.4247717261314392,
      "learning_rate": 1.9946951800737702e-06,
      "loss": 0.7144,
      "step": 14504
    },
    {
      "epoch": 0.6011438517965934,
      "grad_norm": 0.4129229784011841,
      "learning_rate": 1.994487960545402e-06,
      "loss": 0.6875,
      "step": 14505
    },
    {
      "epoch": 0.601185295702267,
      "grad_norm": 0.4251071512699127,
      "learning_rate": 1.994280741017034e-06,
      "loss": 0.6934,
      "step": 14506
    },
    {
      "epoch": 0.6012267396079407,
      "grad_norm": 0.41461607813835144,
      "learning_rate": 1.9940735214886652e-06,
      "loss": 0.7327,
      "step": 14507
    },
    {
      "epoch": 0.6012681835136143,
      "grad_norm": 0.40719062089920044,
      "learning_rate": 1.993866301960297e-06,
      "loss": 0.6981,
      "step": 14508
    },
    {
      "epoch": 0.601309627419288,
      "grad_norm": 0.4209258556365967,
      "learning_rate": 1.9936590824319284e-06,
      "loss": 0.6876,
      "step": 14509
    },
    {
      "epoch": 0.6013510713249617,
      "grad_norm": 0.43294888734817505,
      "learning_rate": 1.9934518629035603e-06,
      "loss": 0.6602,
      "step": 14510
    },
    {
      "epoch": 0.6013925152306353,
      "grad_norm": 0.38607361912727356,
      "learning_rate": 1.9932446433751916e-06,
      "loss": 0.6853,
      "step": 14511
    },
    {
      "epoch": 0.601433959136309,
      "grad_norm": 0.41113215684890747,
      "learning_rate": 1.9930374238468234e-06,
      "loss": 0.6787,
      "step": 14512
    },
    {
      "epoch": 0.6014754030419827,
      "grad_norm": 0.39486944675445557,
      "learning_rate": 1.9928302043184553e-06,
      "loss": 0.6849,
      "step": 14513
    },
    {
      "epoch": 0.6015168469476564,
      "grad_norm": 0.5530759692192078,
      "learning_rate": 1.992622984790087e-06,
      "loss": 0.6517,
      "step": 14514
    },
    {
      "epoch": 0.60155829085333,
      "grad_norm": 0.3923383355140686,
      "learning_rate": 1.9924157652617185e-06,
      "loss": 0.7029,
      "step": 14515
    },
    {
      "epoch": 0.6015997347590036,
      "grad_norm": 0.4193214774131775,
      "learning_rate": 1.9922085457333503e-06,
      "loss": 0.6738,
      "step": 14516
    },
    {
      "epoch": 0.6016411786646774,
      "grad_norm": 0.41727590560913086,
      "learning_rate": 1.9920013262049816e-06,
      "loss": 0.6937,
      "step": 14517
    },
    {
      "epoch": 0.601682622570351,
      "grad_norm": 0.3999008536338806,
      "learning_rate": 1.9917941066766135e-06,
      "loss": 0.6741,
      "step": 14518
    },
    {
      "epoch": 0.6017240664760247,
      "grad_norm": 0.42372938990592957,
      "learning_rate": 1.991586887148245e-06,
      "loss": 0.6792,
      "step": 14519
    },
    {
      "epoch": 0.6017655103816983,
      "grad_norm": 0.4153328835964203,
      "learning_rate": 1.9913796676198766e-06,
      "loss": 0.6887,
      "step": 14520
    },
    {
      "epoch": 0.6018069542873721,
      "grad_norm": 0.43652811646461487,
      "learning_rate": 1.991172448091508e-06,
      "loss": 0.6991,
      "step": 14521
    },
    {
      "epoch": 0.6018483981930457,
      "grad_norm": 0.4087596535682678,
      "learning_rate": 1.9909652285631403e-06,
      "loss": 0.6622,
      "step": 14522
    },
    {
      "epoch": 0.6018898420987194,
      "grad_norm": 0.46677032113075256,
      "learning_rate": 1.9907580090347717e-06,
      "loss": 0.6914,
      "step": 14523
    },
    {
      "epoch": 0.601931286004393,
      "grad_norm": 0.40632516145706177,
      "learning_rate": 1.9905507895064035e-06,
      "loss": 0.652,
      "step": 14524
    },
    {
      "epoch": 0.6019727299100667,
      "grad_norm": 0.4029214382171631,
      "learning_rate": 1.990343569978035e-06,
      "loss": 0.7244,
      "step": 14525
    },
    {
      "epoch": 0.6020141738157404,
      "grad_norm": 0.4357822835445404,
      "learning_rate": 1.9901363504496667e-06,
      "loss": 0.666,
      "step": 14526
    },
    {
      "epoch": 0.602055617721414,
      "grad_norm": 0.40490397810935974,
      "learning_rate": 1.989929130921298e-06,
      "loss": 0.6598,
      "step": 14527
    },
    {
      "epoch": 0.6020970616270878,
      "grad_norm": 0.3937639892101288,
      "learning_rate": 1.98972191139293e-06,
      "loss": 0.6931,
      "step": 14528
    },
    {
      "epoch": 0.6021385055327614,
      "grad_norm": 0.4619409143924713,
      "learning_rate": 1.9895146918645612e-06,
      "loss": 0.7493,
      "step": 14529
    },
    {
      "epoch": 0.6021799494384351,
      "grad_norm": 0.42518556118011475,
      "learning_rate": 1.989307472336193e-06,
      "loss": 0.7072,
      "step": 14530
    },
    {
      "epoch": 0.6022213933441087,
      "grad_norm": 0.4330361783504486,
      "learning_rate": 1.989100252807825e-06,
      "loss": 0.6902,
      "step": 14531
    },
    {
      "epoch": 0.6022628372497825,
      "grad_norm": 0.4245862662792206,
      "learning_rate": 1.9888930332794567e-06,
      "loss": 0.7644,
      "step": 14532
    },
    {
      "epoch": 0.6023042811554561,
      "grad_norm": 0.4168750047683716,
      "learning_rate": 1.988685813751088e-06,
      "loss": 0.6797,
      "step": 14533
    },
    {
      "epoch": 0.6023457250611297,
      "grad_norm": 0.3907606303691864,
      "learning_rate": 1.98847859422272e-06,
      "loss": 0.6884,
      "step": 14534
    },
    {
      "epoch": 0.6023871689668034,
      "grad_norm": 0.4206346273422241,
      "learning_rate": 1.9882713746943512e-06,
      "loss": 0.7175,
      "step": 14535
    },
    {
      "epoch": 0.6024286128724771,
      "grad_norm": 0.4371950626373291,
      "learning_rate": 1.988064155165983e-06,
      "loss": 0.6782,
      "step": 14536
    },
    {
      "epoch": 0.6024700567781508,
      "grad_norm": 0.41218826174736023,
      "learning_rate": 1.9878569356376144e-06,
      "loss": 0.6677,
      "step": 14537
    },
    {
      "epoch": 0.6025115006838244,
      "grad_norm": 0.423749178647995,
      "learning_rate": 1.9876497161092462e-06,
      "loss": 0.6658,
      "step": 14538
    },
    {
      "epoch": 0.6025529445894982,
      "grad_norm": 0.4343227446079254,
      "learning_rate": 1.987442496580878e-06,
      "loss": 0.698,
      "step": 14539
    },
    {
      "epoch": 0.6025943884951718,
      "grad_norm": 0.39605939388275146,
      "learning_rate": 1.98723527705251e-06,
      "loss": 0.738,
      "step": 14540
    },
    {
      "epoch": 0.6026358324008455,
      "grad_norm": 0.3900437355041504,
      "learning_rate": 1.9870280575241413e-06,
      "loss": 0.6831,
      "step": 14541
    },
    {
      "epoch": 0.6026772763065191,
      "grad_norm": 0.40614962577819824,
      "learning_rate": 1.986820837995773e-06,
      "loss": 0.698,
      "step": 14542
    },
    {
      "epoch": 0.6027187202121927,
      "grad_norm": 0.40726831555366516,
      "learning_rate": 1.9866136184674044e-06,
      "loss": 0.7039,
      "step": 14543
    },
    {
      "epoch": 0.6027601641178665,
      "grad_norm": 0.4281916618347168,
      "learning_rate": 1.9864063989390363e-06,
      "loss": 0.663,
      "step": 14544
    },
    {
      "epoch": 0.6028016080235401,
      "grad_norm": 0.4359636604785919,
      "learning_rate": 1.9861991794106676e-06,
      "loss": 0.7024,
      "step": 14545
    },
    {
      "epoch": 0.6028430519292138,
      "grad_norm": 0.43118205666542053,
      "learning_rate": 1.9859919598822995e-06,
      "loss": 0.7151,
      "step": 14546
    },
    {
      "epoch": 0.6028844958348875,
      "grad_norm": 0.4598493278026581,
      "learning_rate": 1.985784740353931e-06,
      "loss": 0.7561,
      "step": 14547
    },
    {
      "epoch": 0.6029259397405612,
      "grad_norm": 0.43564271926879883,
      "learning_rate": 1.985577520825563e-06,
      "loss": 0.6676,
      "step": 14548
    },
    {
      "epoch": 0.6029673836462348,
      "grad_norm": 0.3962418735027313,
      "learning_rate": 1.9853703012971945e-06,
      "loss": 0.6897,
      "step": 14549
    },
    {
      "epoch": 0.6030088275519085,
      "grad_norm": 0.43822038173675537,
      "learning_rate": 1.9851630817688263e-06,
      "loss": 0.6912,
      "step": 14550
    },
    {
      "epoch": 0.6030502714575822,
      "grad_norm": 0.4078572392463684,
      "learning_rate": 1.9849558622404577e-06,
      "loss": 0.6803,
      "step": 14551
    },
    {
      "epoch": 0.6030917153632558,
      "grad_norm": 0.3871171176433563,
      "learning_rate": 1.9847486427120895e-06,
      "loss": 0.6785,
      "step": 14552
    },
    {
      "epoch": 0.6031331592689295,
      "grad_norm": 0.4164421260356903,
      "learning_rate": 1.984541423183721e-06,
      "loss": 0.642,
      "step": 14553
    },
    {
      "epoch": 0.6031746031746031,
      "grad_norm": 0.4098902642726898,
      "learning_rate": 1.9843342036553527e-06,
      "loss": 0.632,
      "step": 14554
    },
    {
      "epoch": 0.6032160470802769,
      "grad_norm": 0.4154486656188965,
      "learning_rate": 1.984126984126984e-06,
      "loss": 0.6725,
      "step": 14555
    },
    {
      "epoch": 0.6032574909859505,
      "grad_norm": 0.47231099009513855,
      "learning_rate": 1.9839197645986163e-06,
      "loss": 0.6755,
      "step": 14556
    },
    {
      "epoch": 0.6032989348916242,
      "grad_norm": 0.5256350636482239,
      "learning_rate": 1.9837125450702477e-06,
      "loss": 0.6489,
      "step": 14557
    },
    {
      "epoch": 0.6033403787972978,
      "grad_norm": 0.3788302540779114,
      "learning_rate": 1.9835053255418795e-06,
      "loss": 0.6382,
      "step": 14558
    },
    {
      "epoch": 0.6033818227029716,
      "grad_norm": 0.399811327457428,
      "learning_rate": 1.983298106013511e-06,
      "loss": 0.7139,
      "step": 14559
    },
    {
      "epoch": 0.6034232666086452,
      "grad_norm": 0.3467707633972168,
      "learning_rate": 1.9830908864851427e-06,
      "loss": 0.5829,
      "step": 14560
    },
    {
      "epoch": 0.6034647105143188,
      "grad_norm": 0.4133845269680023,
      "learning_rate": 1.982883666956774e-06,
      "loss": 0.6897,
      "step": 14561
    },
    {
      "epoch": 0.6035061544199926,
      "grad_norm": 0.423615962266922,
      "learning_rate": 1.982676447428406e-06,
      "loss": 0.6881,
      "step": 14562
    },
    {
      "epoch": 0.6035475983256662,
      "grad_norm": 0.3892373740673065,
      "learning_rate": 1.9824692279000372e-06,
      "loss": 0.6481,
      "step": 14563
    },
    {
      "epoch": 0.6035890422313399,
      "grad_norm": 0.4169505536556244,
      "learning_rate": 1.982262008371669e-06,
      "loss": 0.667,
      "step": 14564
    },
    {
      "epoch": 0.6036304861370135,
      "grad_norm": 0.3884958028793335,
      "learning_rate": 1.982054788843301e-06,
      "loss": 0.6415,
      "step": 14565
    },
    {
      "epoch": 0.6036719300426873,
      "grad_norm": 0.4303205609321594,
      "learning_rate": 1.9818475693149327e-06,
      "loss": 0.7024,
      "step": 14566
    },
    {
      "epoch": 0.6037133739483609,
      "grad_norm": 0.4007004499435425,
      "learning_rate": 1.981640349786564e-06,
      "loss": 0.7046,
      "step": 14567
    },
    {
      "epoch": 0.6037548178540345,
      "grad_norm": 0.4031788110733032,
      "learning_rate": 1.981433130258196e-06,
      "loss": 0.7051,
      "step": 14568
    },
    {
      "epoch": 0.6037962617597082,
      "grad_norm": 0.40699395537376404,
      "learning_rate": 1.9812259107298273e-06,
      "loss": 0.6539,
      "step": 14569
    },
    {
      "epoch": 0.6038377056653819,
      "grad_norm": 0.4458667039871216,
      "learning_rate": 1.981018691201459e-06,
      "loss": 0.7366,
      "step": 14570
    },
    {
      "epoch": 0.6038791495710556,
      "grad_norm": 0.44263502955436707,
      "learning_rate": 1.9808114716730904e-06,
      "loss": 0.7013,
      "step": 14571
    },
    {
      "epoch": 0.6039205934767292,
      "grad_norm": 0.4185800552368164,
      "learning_rate": 1.9806042521447223e-06,
      "loss": 0.6775,
      "step": 14572
    },
    {
      "epoch": 0.603962037382403,
      "grad_norm": 0.4165576994419098,
      "learning_rate": 1.9803970326163536e-06,
      "loss": 0.6711,
      "step": 14573
    },
    {
      "epoch": 0.6040034812880766,
      "grad_norm": 0.4288649559020996,
      "learning_rate": 1.980189813087986e-06,
      "loss": 0.7314,
      "step": 14574
    },
    {
      "epoch": 0.6040449251937503,
      "grad_norm": 0.4109044373035431,
      "learning_rate": 1.9799825935596173e-06,
      "loss": 0.7043,
      "step": 14575
    },
    {
      "epoch": 0.6040863690994239,
      "grad_norm": 0.43378946185112,
      "learning_rate": 1.979775374031249e-06,
      "loss": 0.6718,
      "step": 14576
    },
    {
      "epoch": 0.6041278130050975,
      "grad_norm": 0.40689921379089355,
      "learning_rate": 1.9795681545028805e-06,
      "loss": 0.682,
      "step": 14577
    },
    {
      "epoch": 0.6041692569107713,
      "grad_norm": 0.4111175537109375,
      "learning_rate": 1.9793609349745123e-06,
      "loss": 0.7153,
      "step": 14578
    },
    {
      "epoch": 0.6042107008164449,
      "grad_norm": 0.42727532982826233,
      "learning_rate": 1.9791537154461436e-06,
      "loss": 0.6909,
      "step": 14579
    },
    {
      "epoch": 0.6042521447221186,
      "grad_norm": 0.3969823718070984,
      "learning_rate": 1.9789464959177755e-06,
      "loss": 0.6895,
      "step": 14580
    },
    {
      "epoch": 0.6042935886277923,
      "grad_norm": 0.3935965895652771,
      "learning_rate": 1.978739276389407e-06,
      "loss": 0.6522,
      "step": 14581
    },
    {
      "epoch": 0.604335032533466,
      "grad_norm": 0.43941396474838257,
      "learning_rate": 1.9785320568610387e-06,
      "loss": 0.6952,
      "step": 14582
    },
    {
      "epoch": 0.6043764764391396,
      "grad_norm": 0.41477498412132263,
      "learning_rate": 1.9783248373326705e-06,
      "loss": 0.6844,
      "step": 14583
    },
    {
      "epoch": 0.6044179203448133,
      "grad_norm": 0.39874598383903503,
      "learning_rate": 1.9781176178043023e-06,
      "loss": 0.6285,
      "step": 14584
    },
    {
      "epoch": 0.604459364250487,
      "grad_norm": 0.39784592390060425,
      "learning_rate": 1.9779103982759337e-06,
      "loss": 0.6653,
      "step": 14585
    },
    {
      "epoch": 0.6045008081561606,
      "grad_norm": 0.4260646104812622,
      "learning_rate": 1.9777031787475655e-06,
      "loss": 0.6904,
      "step": 14586
    },
    {
      "epoch": 0.6045422520618343,
      "grad_norm": 0.430759996175766,
      "learning_rate": 1.977495959219197e-06,
      "loss": 0.741,
      "step": 14587
    },
    {
      "epoch": 0.6045836959675079,
      "grad_norm": 0.42136672139167786,
      "learning_rate": 1.9772887396908287e-06,
      "loss": 0.6552,
      "step": 14588
    },
    {
      "epoch": 0.6046251398731817,
      "grad_norm": 0.40124770998954773,
      "learning_rate": 1.97708152016246e-06,
      "loss": 0.6461,
      "step": 14589
    },
    {
      "epoch": 0.6046665837788553,
      "grad_norm": 0.40887436270713806,
      "learning_rate": 1.976874300634092e-06,
      "loss": 0.6331,
      "step": 14590
    },
    {
      "epoch": 0.604708027684529,
      "grad_norm": 0.43172261118888855,
      "learning_rate": 1.9766670811057237e-06,
      "loss": 0.6543,
      "step": 14591
    },
    {
      "epoch": 0.6047494715902026,
      "grad_norm": 0.43241265416145325,
      "learning_rate": 1.9764598615773555e-06,
      "loss": 0.7156,
      "step": 14592
    },
    {
      "epoch": 0.6047909154958764,
      "grad_norm": 0.42058658599853516,
      "learning_rate": 1.976252642048987e-06,
      "loss": 0.7065,
      "step": 14593
    },
    {
      "epoch": 0.60483235940155,
      "grad_norm": 0.40181276202201843,
      "learning_rate": 1.9760454225206187e-06,
      "loss": 0.6285,
      "step": 14594
    },
    {
      "epoch": 0.6048738033072236,
      "grad_norm": 0.40688827633857727,
      "learning_rate": 1.97583820299225e-06,
      "loss": 0.709,
      "step": 14595
    },
    {
      "epoch": 0.6049152472128974,
      "grad_norm": 0.3901410698890686,
      "learning_rate": 1.975630983463882e-06,
      "loss": 0.6766,
      "step": 14596
    },
    {
      "epoch": 0.604956691118571,
      "grad_norm": 0.4416082799434662,
      "learning_rate": 1.9754237639355132e-06,
      "loss": 0.7454,
      "step": 14597
    },
    {
      "epoch": 0.6049981350242447,
      "grad_norm": 0.40549105405807495,
      "learning_rate": 1.975216544407145e-06,
      "loss": 0.7012,
      "step": 14598
    },
    {
      "epoch": 0.6050395789299183,
      "grad_norm": 0.4303135573863983,
      "learning_rate": 1.9750093248787764e-06,
      "loss": 0.689,
      "step": 14599
    },
    {
      "epoch": 0.6050810228355921,
      "grad_norm": 0.45073187351226807,
      "learning_rate": 1.9748021053504087e-06,
      "loss": 0.73,
      "step": 14600
    },
    {
      "epoch": 0.6051224667412657,
      "grad_norm": 0.3766886591911316,
      "learning_rate": 1.97459488582204e-06,
      "loss": 0.6681,
      "step": 14601
    },
    {
      "epoch": 0.6051639106469394,
      "grad_norm": 0.43008551001548767,
      "learning_rate": 1.974387666293672e-06,
      "loss": 0.6895,
      "step": 14602
    },
    {
      "epoch": 0.605205354552613,
      "grad_norm": 0.4343551695346832,
      "learning_rate": 1.9741804467653033e-06,
      "loss": 0.7064,
      "step": 14603
    },
    {
      "epoch": 0.6052467984582867,
      "grad_norm": 0.4243672788143158,
      "learning_rate": 1.973973227236935e-06,
      "loss": 0.6992,
      "step": 14604
    },
    {
      "epoch": 0.6052882423639604,
      "grad_norm": 0.4331203103065491,
      "learning_rate": 1.9737660077085665e-06,
      "loss": 0.6613,
      "step": 14605
    },
    {
      "epoch": 0.605329686269634,
      "grad_norm": 0.42671963572502136,
      "learning_rate": 1.9735587881801983e-06,
      "loss": 0.6786,
      "step": 14606
    },
    {
      "epoch": 0.6053711301753077,
      "grad_norm": 0.4072897136211395,
      "learning_rate": 1.9733515686518296e-06,
      "loss": 0.6954,
      "step": 14607
    },
    {
      "epoch": 0.6054125740809814,
      "grad_norm": 0.44126808643341064,
      "learning_rate": 1.9731443491234615e-06,
      "loss": 0.6709,
      "step": 14608
    },
    {
      "epoch": 0.6054540179866551,
      "grad_norm": 0.4323432743549347,
      "learning_rate": 1.9729371295950933e-06,
      "loss": 0.6956,
      "step": 14609
    },
    {
      "epoch": 0.6054954618923287,
      "grad_norm": 0.3924527168273926,
      "learning_rate": 1.972729910066725e-06,
      "loss": 0.6937,
      "step": 14610
    },
    {
      "epoch": 0.6055369057980025,
      "grad_norm": 0.4285413920879364,
      "learning_rate": 1.9725226905383565e-06,
      "loss": 0.7222,
      "step": 14611
    },
    {
      "epoch": 0.6055783497036761,
      "grad_norm": 0.43098345398902893,
      "learning_rate": 1.9723154710099883e-06,
      "loss": 0.6912,
      "step": 14612
    },
    {
      "epoch": 0.6056197936093497,
      "grad_norm": 0.40945395827293396,
      "learning_rate": 1.9721082514816197e-06,
      "loss": 0.6238,
      "step": 14613
    },
    {
      "epoch": 0.6056612375150234,
      "grad_norm": 0.4322804808616638,
      "learning_rate": 1.9719010319532515e-06,
      "loss": 0.6721,
      "step": 14614
    },
    {
      "epoch": 0.605702681420697,
      "grad_norm": 0.3902067542076111,
      "learning_rate": 1.971693812424883e-06,
      "loss": 0.6871,
      "step": 14615
    },
    {
      "epoch": 0.6057441253263708,
      "grad_norm": 0.4011251628398895,
      "learning_rate": 1.9714865928965147e-06,
      "loss": 0.6912,
      "step": 14616
    },
    {
      "epoch": 0.6057855692320444,
      "grad_norm": 0.3997582495212555,
      "learning_rate": 1.9712793733681465e-06,
      "loss": 0.6847,
      "step": 14617
    },
    {
      "epoch": 0.6058270131377181,
      "grad_norm": 0.42463749647140503,
      "learning_rate": 1.9710721538397783e-06,
      "loss": 0.7384,
      "step": 14618
    },
    {
      "epoch": 0.6058684570433918,
      "grad_norm": 0.38292741775512695,
      "learning_rate": 1.9708649343114097e-06,
      "loss": 0.6478,
      "step": 14619
    },
    {
      "epoch": 0.6059099009490655,
      "grad_norm": 0.4069931209087372,
      "learning_rate": 1.9706577147830415e-06,
      "loss": 0.7119,
      "step": 14620
    },
    {
      "epoch": 0.6059513448547391,
      "grad_norm": 0.41083765029907227,
      "learning_rate": 1.970450495254673e-06,
      "loss": 0.641,
      "step": 14621
    },
    {
      "epoch": 0.6059927887604127,
      "grad_norm": 0.4617087244987488,
      "learning_rate": 1.9702432757263047e-06,
      "loss": 0.715,
      "step": 14622
    },
    {
      "epoch": 0.6060342326660865,
      "grad_norm": 0.44889751076698303,
      "learning_rate": 1.970036056197936e-06,
      "loss": 0.6981,
      "step": 14623
    },
    {
      "epoch": 0.6060756765717601,
      "grad_norm": 0.40447843074798584,
      "learning_rate": 1.969828836669568e-06,
      "loss": 0.7297,
      "step": 14624
    },
    {
      "epoch": 0.6061171204774338,
      "grad_norm": 0.4242262840270996,
      "learning_rate": 1.9696216171411992e-06,
      "loss": 0.7227,
      "step": 14625
    },
    {
      "epoch": 0.6061585643831074,
      "grad_norm": 0.39803069829940796,
      "learning_rate": 1.9694143976128315e-06,
      "loss": 0.6542,
      "step": 14626
    },
    {
      "epoch": 0.6062000082887812,
      "grad_norm": 0.3962688148021698,
      "learning_rate": 1.969207178084463e-06,
      "loss": 0.6766,
      "step": 14627
    },
    {
      "epoch": 0.6062414521944548,
      "grad_norm": 0.42824968695640564,
      "learning_rate": 1.9689999585560947e-06,
      "loss": 0.6677,
      "step": 14628
    },
    {
      "epoch": 0.6062828961001284,
      "grad_norm": 0.37805047631263733,
      "learning_rate": 1.968792739027726e-06,
      "loss": 0.6382,
      "step": 14629
    },
    {
      "epoch": 0.6063243400058022,
      "grad_norm": 0.39055702090263367,
      "learning_rate": 1.968585519499358e-06,
      "loss": 0.66,
      "step": 14630
    },
    {
      "epoch": 0.6063657839114758,
      "grad_norm": 0.43569448590278625,
      "learning_rate": 1.9683782999709893e-06,
      "loss": 0.7026,
      "step": 14631
    },
    {
      "epoch": 0.6064072278171495,
      "grad_norm": 0.3735261857509613,
      "learning_rate": 1.968171080442621e-06,
      "loss": 0.6677,
      "step": 14632
    },
    {
      "epoch": 0.6064486717228231,
      "grad_norm": 0.3640260696411133,
      "learning_rate": 1.9679638609142525e-06,
      "loss": 0.6362,
      "step": 14633
    },
    {
      "epoch": 0.6064901156284969,
      "grad_norm": 0.4008565843105316,
      "learning_rate": 1.9677566413858843e-06,
      "loss": 0.7051,
      "step": 14634
    },
    {
      "epoch": 0.6065315595341705,
      "grad_norm": 0.36494311690330505,
      "learning_rate": 1.967549421857516e-06,
      "loss": 0.6465,
      "step": 14635
    },
    {
      "epoch": 0.6065730034398442,
      "grad_norm": 0.4230460822582245,
      "learning_rate": 1.967342202329148e-06,
      "loss": 0.6949,
      "step": 14636
    },
    {
      "epoch": 0.6066144473455178,
      "grad_norm": 0.40854892134666443,
      "learning_rate": 1.9671349828007793e-06,
      "loss": 0.6903,
      "step": 14637
    },
    {
      "epoch": 0.6066558912511915,
      "grad_norm": 0.4415653944015503,
      "learning_rate": 1.966927763272411e-06,
      "loss": 0.656,
      "step": 14638
    },
    {
      "epoch": 0.6066973351568652,
      "grad_norm": 0.3898528814315796,
      "learning_rate": 1.9667205437440425e-06,
      "loss": 0.6653,
      "step": 14639
    },
    {
      "epoch": 0.6067387790625388,
      "grad_norm": 0.43967920541763306,
      "learning_rate": 1.9665133242156743e-06,
      "loss": 0.7502,
      "step": 14640
    },
    {
      "epoch": 0.6067802229682125,
      "grad_norm": 0.38948407769203186,
      "learning_rate": 1.9663061046873057e-06,
      "loss": 0.5983,
      "step": 14641
    },
    {
      "epoch": 0.6068216668738862,
      "grad_norm": 0.40707844495773315,
      "learning_rate": 1.9660988851589375e-06,
      "loss": 0.6031,
      "step": 14642
    },
    {
      "epoch": 0.6068631107795599,
      "grad_norm": 0.41765865683555603,
      "learning_rate": 1.9658916656305693e-06,
      "loss": 0.6714,
      "step": 14643
    },
    {
      "epoch": 0.6069045546852335,
      "grad_norm": 0.37631139159202576,
      "learning_rate": 1.965684446102201e-06,
      "loss": 0.6663,
      "step": 14644
    },
    {
      "epoch": 0.6069459985909073,
      "grad_norm": 0.3965761661529541,
      "learning_rate": 1.9654772265738325e-06,
      "loss": 0.7217,
      "step": 14645
    },
    {
      "epoch": 0.6069874424965809,
      "grad_norm": 0.37779703736305237,
      "learning_rate": 1.9652700070454643e-06,
      "loss": 0.6621,
      "step": 14646
    },
    {
      "epoch": 0.6070288864022545,
      "grad_norm": 0.4053974747657776,
      "learning_rate": 1.9650627875170957e-06,
      "loss": 0.6436,
      "step": 14647
    },
    {
      "epoch": 0.6070703303079282,
      "grad_norm": 0.39378875494003296,
      "learning_rate": 1.9648555679887275e-06,
      "loss": 0.6672,
      "step": 14648
    },
    {
      "epoch": 0.6071117742136019,
      "grad_norm": 0.4237856864929199,
      "learning_rate": 1.964648348460359e-06,
      "loss": 0.656,
      "step": 14649
    },
    {
      "epoch": 0.6071532181192756,
      "grad_norm": 0.39871126413345337,
      "learning_rate": 1.9644411289319907e-06,
      "loss": 0.6418,
      "step": 14650
    },
    {
      "epoch": 0.6071946620249492,
      "grad_norm": 0.4464757740497589,
      "learning_rate": 1.964233909403622e-06,
      "loss": 0.73,
      "step": 14651
    },
    {
      "epoch": 0.6072361059306229,
      "grad_norm": 0.4395763576030731,
      "learning_rate": 1.9640266898752543e-06,
      "loss": 0.704,
      "step": 14652
    },
    {
      "epoch": 0.6072775498362966,
      "grad_norm": 0.4002503752708435,
      "learning_rate": 1.9638194703468857e-06,
      "loss": 0.7141,
      "step": 14653
    },
    {
      "epoch": 0.6073189937419703,
      "grad_norm": 0.43121153116226196,
      "learning_rate": 1.9636122508185175e-06,
      "loss": 0.6525,
      "step": 14654
    },
    {
      "epoch": 0.6073604376476439,
      "grad_norm": 0.3972659111022949,
      "learning_rate": 1.963405031290149e-06,
      "loss": 0.639,
      "step": 14655
    },
    {
      "epoch": 0.6074018815533175,
      "grad_norm": 0.4078911244869232,
      "learning_rate": 1.9631978117617807e-06,
      "loss": 0.6934,
      "step": 14656
    },
    {
      "epoch": 0.6074433254589913,
      "grad_norm": 0.4000316262245178,
      "learning_rate": 1.962990592233412e-06,
      "loss": 0.7064,
      "step": 14657
    },
    {
      "epoch": 0.6074847693646649,
      "grad_norm": 0.3996182382106781,
      "learning_rate": 1.962783372705044e-06,
      "loss": 0.6946,
      "step": 14658
    },
    {
      "epoch": 0.6075262132703386,
      "grad_norm": 0.37282899022102356,
      "learning_rate": 1.9625761531766753e-06,
      "loss": 0.6648,
      "step": 14659
    },
    {
      "epoch": 0.6075676571760122,
      "grad_norm": 0.4299367070198059,
      "learning_rate": 1.962368933648307e-06,
      "loss": 0.6787,
      "step": 14660
    },
    {
      "epoch": 0.607609101081686,
      "grad_norm": 0.3983380198478699,
      "learning_rate": 1.962161714119939e-06,
      "loss": 0.6683,
      "step": 14661
    },
    {
      "epoch": 0.6076505449873596,
      "grad_norm": 0.419456422328949,
      "learning_rate": 1.9619544945915707e-06,
      "loss": 0.6985,
      "step": 14662
    },
    {
      "epoch": 0.6076919888930333,
      "grad_norm": 0.4241117537021637,
      "learning_rate": 1.961747275063202e-06,
      "loss": 0.739,
      "step": 14663
    },
    {
      "epoch": 0.607733432798707,
      "grad_norm": 0.3844147324562073,
      "learning_rate": 1.961540055534834e-06,
      "loss": 0.6271,
      "step": 14664
    },
    {
      "epoch": 0.6077748767043806,
      "grad_norm": 0.4263402819633484,
      "learning_rate": 1.9613328360064653e-06,
      "loss": 0.6388,
      "step": 14665
    },
    {
      "epoch": 0.6078163206100543,
      "grad_norm": 0.4393508732318878,
      "learning_rate": 1.961125616478097e-06,
      "loss": 0.6653,
      "step": 14666
    },
    {
      "epoch": 0.6078577645157279,
      "grad_norm": 0.42251884937286377,
      "learning_rate": 1.9609183969497285e-06,
      "loss": 0.6652,
      "step": 14667
    },
    {
      "epoch": 0.6078992084214017,
      "grad_norm": 0.4040488004684448,
      "learning_rate": 1.9607111774213603e-06,
      "loss": 0.6877,
      "step": 14668
    },
    {
      "epoch": 0.6079406523270753,
      "grad_norm": 0.41173335909843445,
      "learning_rate": 1.9605039578929917e-06,
      "loss": 0.6636,
      "step": 14669
    },
    {
      "epoch": 0.607982096232749,
      "grad_norm": 0.4789021611213684,
      "learning_rate": 1.960296738364624e-06,
      "loss": 0.7444,
      "step": 14670
    },
    {
      "epoch": 0.6080235401384226,
      "grad_norm": 0.40368106961250305,
      "learning_rate": 1.9600895188362553e-06,
      "loss": 0.6854,
      "step": 14671
    },
    {
      "epoch": 0.6080649840440964,
      "grad_norm": 0.40846213698387146,
      "learning_rate": 1.959882299307887e-06,
      "loss": 0.6909,
      "step": 14672
    },
    {
      "epoch": 0.60810642794977,
      "grad_norm": 0.36752641201019287,
      "learning_rate": 1.9596750797795185e-06,
      "loss": 0.6768,
      "step": 14673
    },
    {
      "epoch": 0.6081478718554436,
      "grad_norm": 0.41921013593673706,
      "learning_rate": 1.9594678602511503e-06,
      "loss": 0.6769,
      "step": 14674
    },
    {
      "epoch": 0.6081893157611173,
      "grad_norm": 0.3896467983722687,
      "learning_rate": 1.9592606407227817e-06,
      "loss": 0.6337,
      "step": 14675
    },
    {
      "epoch": 0.608230759666791,
      "grad_norm": 0.4102485775947571,
      "learning_rate": 1.9590534211944135e-06,
      "loss": 0.6882,
      "step": 14676
    },
    {
      "epoch": 0.6082722035724647,
      "grad_norm": 0.45134979486465454,
      "learning_rate": 1.958846201666045e-06,
      "loss": 0.6953,
      "step": 14677
    },
    {
      "epoch": 0.6083136474781383,
      "grad_norm": 0.4052216410636902,
      "learning_rate": 1.958638982137677e-06,
      "loss": 0.6211,
      "step": 14678
    },
    {
      "epoch": 0.608355091383812,
      "grad_norm": 0.4310140311717987,
      "learning_rate": 1.9584317626093085e-06,
      "loss": 0.7188,
      "step": 14679
    },
    {
      "epoch": 0.6083965352894857,
      "grad_norm": 0.4156208634376526,
      "learning_rate": 1.9582245430809403e-06,
      "loss": 0.679,
      "step": 14680
    },
    {
      "epoch": 0.6084379791951594,
      "grad_norm": 0.41505077481269836,
      "learning_rate": 1.9580173235525717e-06,
      "loss": 0.7222,
      "step": 14681
    },
    {
      "epoch": 0.608479423100833,
      "grad_norm": 0.42604413628578186,
      "learning_rate": 1.9578101040242035e-06,
      "loss": 0.6813,
      "step": 14682
    },
    {
      "epoch": 0.6085208670065066,
      "grad_norm": 0.43553459644317627,
      "learning_rate": 1.957602884495835e-06,
      "loss": 0.7175,
      "step": 14683
    },
    {
      "epoch": 0.6085623109121804,
      "grad_norm": 0.3961941599845886,
      "learning_rate": 1.9573956649674667e-06,
      "loss": 0.6522,
      "step": 14684
    },
    {
      "epoch": 0.608603754817854,
      "grad_norm": 0.4804244339466095,
      "learning_rate": 1.957188445439098e-06,
      "loss": 0.689,
      "step": 14685
    },
    {
      "epoch": 0.6086451987235277,
      "grad_norm": 0.3801030218601227,
      "learning_rate": 1.95698122591073e-06,
      "loss": 0.6589,
      "step": 14686
    },
    {
      "epoch": 0.6086866426292014,
      "grad_norm": 0.40263184905052185,
      "learning_rate": 1.9567740063823617e-06,
      "loss": 0.6545,
      "step": 14687
    },
    {
      "epoch": 0.6087280865348751,
      "grad_norm": 0.40567082166671753,
      "learning_rate": 1.9565667868539935e-06,
      "loss": 0.6658,
      "step": 14688
    },
    {
      "epoch": 0.6087695304405487,
      "grad_norm": 0.41482266783714294,
      "learning_rate": 1.956359567325625e-06,
      "loss": 0.6892,
      "step": 14689
    },
    {
      "epoch": 0.6088109743462223,
      "grad_norm": 0.42148539423942566,
      "learning_rate": 1.9561523477972567e-06,
      "loss": 0.6512,
      "step": 14690
    },
    {
      "epoch": 0.6088524182518961,
      "grad_norm": 0.3934229612350464,
      "learning_rate": 1.955945128268888e-06,
      "loss": 0.6576,
      "step": 14691
    },
    {
      "epoch": 0.6088938621575697,
      "grad_norm": 0.4452611207962036,
      "learning_rate": 1.95573790874052e-06,
      "loss": 0.6483,
      "step": 14692
    },
    {
      "epoch": 0.6089353060632434,
      "grad_norm": 0.39639490842819214,
      "learning_rate": 1.9555306892121513e-06,
      "loss": 0.6788,
      "step": 14693
    },
    {
      "epoch": 0.608976749968917,
      "grad_norm": 0.4151207506656647,
      "learning_rate": 1.955323469683783e-06,
      "loss": 0.6626,
      "step": 14694
    },
    {
      "epoch": 0.6090181938745908,
      "grad_norm": 0.42069268226623535,
      "learning_rate": 1.955116250155415e-06,
      "loss": 0.6582,
      "step": 14695
    },
    {
      "epoch": 0.6090596377802644,
      "grad_norm": 0.4175778329372406,
      "learning_rate": 1.9549090306270467e-06,
      "loss": 0.708,
      "step": 14696
    },
    {
      "epoch": 0.6091010816859381,
      "grad_norm": 0.42149442434310913,
      "learning_rate": 1.954701811098678e-06,
      "loss": 0.6527,
      "step": 14697
    },
    {
      "epoch": 0.6091425255916117,
      "grad_norm": 0.46400123834609985,
      "learning_rate": 1.95449459157031e-06,
      "loss": 0.7235,
      "step": 14698
    },
    {
      "epoch": 0.6091839694972854,
      "grad_norm": 0.4213758707046509,
      "learning_rate": 1.9542873720419413e-06,
      "loss": 0.7017,
      "step": 14699
    },
    {
      "epoch": 0.6092254134029591,
      "grad_norm": 0.40203556418418884,
      "learning_rate": 1.954080152513573e-06,
      "loss": 0.6819,
      "step": 14700
    },
    {
      "epoch": 0.6092668573086327,
      "grad_norm": 0.42237186431884766,
      "learning_rate": 1.9538729329852045e-06,
      "loss": 0.7161,
      "step": 14701
    },
    {
      "epoch": 0.6093083012143065,
      "grad_norm": 0.39565232396125793,
      "learning_rate": 1.9536657134568363e-06,
      "loss": 0.639,
      "step": 14702
    },
    {
      "epoch": 0.6093497451199801,
      "grad_norm": 0.41250309348106384,
      "learning_rate": 1.953458493928468e-06,
      "loss": 0.728,
      "step": 14703
    },
    {
      "epoch": 0.6093911890256538,
      "grad_norm": 0.3725629448890686,
      "learning_rate": 1.9532512744001e-06,
      "loss": 0.6351,
      "step": 14704
    },
    {
      "epoch": 0.6094326329313274,
      "grad_norm": 0.42929139733314514,
      "learning_rate": 1.9530440548717313e-06,
      "loss": 0.7141,
      "step": 14705
    },
    {
      "epoch": 0.6094740768370012,
      "grad_norm": 0.4003793001174927,
      "learning_rate": 1.952836835343363e-06,
      "loss": 0.6859,
      "step": 14706
    },
    {
      "epoch": 0.6095155207426748,
      "grad_norm": 0.43394553661346436,
      "learning_rate": 1.9526296158149945e-06,
      "loss": 0.6421,
      "step": 14707
    },
    {
      "epoch": 0.6095569646483484,
      "grad_norm": 0.4174637496471405,
      "learning_rate": 1.9524223962866263e-06,
      "loss": 0.6929,
      "step": 14708
    },
    {
      "epoch": 0.6095984085540221,
      "grad_norm": 0.46011701226234436,
      "learning_rate": 1.9522151767582577e-06,
      "loss": 0.7133,
      "step": 14709
    },
    {
      "epoch": 0.6096398524596958,
      "grad_norm": 0.4015706479549408,
      "learning_rate": 1.9520079572298895e-06,
      "loss": 0.6746,
      "step": 14710
    },
    {
      "epoch": 0.6096812963653695,
      "grad_norm": 0.4048427939414978,
      "learning_rate": 1.951800737701521e-06,
      "loss": 0.6288,
      "step": 14711
    },
    {
      "epoch": 0.6097227402710431,
      "grad_norm": 0.42263224720954895,
      "learning_rate": 1.9515935181731527e-06,
      "loss": 0.6617,
      "step": 14712
    },
    {
      "epoch": 0.6097641841767169,
      "grad_norm": 0.4037068486213684,
      "learning_rate": 1.9513862986447845e-06,
      "loss": 0.7188,
      "step": 14713
    },
    {
      "epoch": 0.6098056280823905,
      "grad_norm": 0.40357834100723267,
      "learning_rate": 1.9511790791164163e-06,
      "loss": 0.6393,
      "step": 14714
    },
    {
      "epoch": 0.6098470719880642,
      "grad_norm": 0.3954143822193146,
      "learning_rate": 1.9509718595880477e-06,
      "loss": 0.7173,
      "step": 14715
    },
    {
      "epoch": 0.6098885158937378,
      "grad_norm": 0.39964383840560913,
      "learning_rate": 1.9507646400596795e-06,
      "loss": 0.6263,
      "step": 14716
    },
    {
      "epoch": 0.6099299597994114,
      "grad_norm": 0.4229471981525421,
      "learning_rate": 1.950557420531311e-06,
      "loss": 0.6464,
      "step": 14717
    },
    {
      "epoch": 0.6099714037050852,
      "grad_norm": 0.3831164836883545,
      "learning_rate": 1.9503502010029427e-06,
      "loss": 0.6255,
      "step": 14718
    },
    {
      "epoch": 0.6100128476107588,
      "grad_norm": 0.39974895119667053,
      "learning_rate": 1.950142981474574e-06,
      "loss": 0.6611,
      "step": 14719
    },
    {
      "epoch": 0.6100542915164325,
      "grad_norm": 0.41141578555107117,
      "learning_rate": 1.949935761946206e-06,
      "loss": 0.6562,
      "step": 14720
    },
    {
      "epoch": 0.6100957354221062,
      "grad_norm": 0.4257267713546753,
      "learning_rate": 1.9497285424178377e-06,
      "loss": 0.7202,
      "step": 14721
    },
    {
      "epoch": 0.6101371793277799,
      "grad_norm": 0.4029546082019806,
      "learning_rate": 1.9495213228894695e-06,
      "loss": 0.6764,
      "step": 14722
    },
    {
      "epoch": 0.6101786232334535,
      "grad_norm": 0.4060714542865753,
      "learning_rate": 1.949314103361101e-06,
      "loss": 0.7137,
      "step": 14723
    },
    {
      "epoch": 0.6102200671391272,
      "grad_norm": 0.4145147204399109,
      "learning_rate": 1.9491068838327327e-06,
      "loss": 0.6779,
      "step": 14724
    },
    {
      "epoch": 0.6102615110448009,
      "grad_norm": 0.4077827036380768,
      "learning_rate": 1.948899664304364e-06,
      "loss": 0.6392,
      "step": 14725
    },
    {
      "epoch": 0.6103029549504745,
      "grad_norm": 0.4191453754901886,
      "learning_rate": 1.948692444775996e-06,
      "loss": 0.7249,
      "step": 14726
    },
    {
      "epoch": 0.6103443988561482,
      "grad_norm": 0.42236271500587463,
      "learning_rate": 1.9484852252476273e-06,
      "loss": 0.6591,
      "step": 14727
    },
    {
      "epoch": 0.6103858427618218,
      "grad_norm": 0.39032596349716187,
      "learning_rate": 1.948278005719259e-06,
      "loss": 0.6349,
      "step": 14728
    },
    {
      "epoch": 0.6104272866674956,
      "grad_norm": 0.4207191467285156,
      "learning_rate": 1.948070786190891e-06,
      "loss": 0.6525,
      "step": 14729
    },
    {
      "epoch": 0.6104687305731692,
      "grad_norm": 0.41551247239112854,
      "learning_rate": 1.9478635666625223e-06,
      "loss": 0.6556,
      "step": 14730
    },
    {
      "epoch": 0.6105101744788429,
      "grad_norm": 0.3975074887275696,
      "learning_rate": 1.947656347134154e-06,
      "loss": 0.6377,
      "step": 14731
    },
    {
      "epoch": 0.6105516183845165,
      "grad_norm": 0.40849095582962036,
      "learning_rate": 1.947449127605786e-06,
      "loss": 0.7202,
      "step": 14732
    },
    {
      "epoch": 0.6105930622901903,
      "grad_norm": 0.42856234312057495,
      "learning_rate": 1.9472419080774173e-06,
      "loss": 0.731,
      "step": 14733
    },
    {
      "epoch": 0.6106345061958639,
      "grad_norm": 0.3826061487197876,
      "learning_rate": 1.947034688549049e-06,
      "loss": 0.6271,
      "step": 14734
    },
    {
      "epoch": 0.6106759501015375,
      "grad_norm": 0.4518601596355438,
      "learning_rate": 1.9468274690206805e-06,
      "loss": 0.6938,
      "step": 14735
    },
    {
      "epoch": 0.6107173940072113,
      "grad_norm": 0.43080198764801025,
      "learning_rate": 1.9466202494923123e-06,
      "loss": 0.6666,
      "step": 14736
    },
    {
      "epoch": 0.6107588379128849,
      "grad_norm": 0.41246476769447327,
      "learning_rate": 1.946413029963944e-06,
      "loss": 0.6385,
      "step": 14737
    },
    {
      "epoch": 0.6108002818185586,
      "grad_norm": 0.4633418619632721,
      "learning_rate": 1.9462058104355755e-06,
      "loss": 0.6709,
      "step": 14738
    },
    {
      "epoch": 0.6108417257242322,
      "grad_norm": 0.381747305393219,
      "learning_rate": 1.9459985909072073e-06,
      "loss": 0.6484,
      "step": 14739
    },
    {
      "epoch": 0.610883169629906,
      "grad_norm": 0.40145793557167053,
      "learning_rate": 1.945791371378839e-06,
      "loss": 0.6951,
      "step": 14740
    },
    {
      "epoch": 0.6109246135355796,
      "grad_norm": 0.3760605454444885,
      "learning_rate": 1.9455841518504705e-06,
      "loss": 0.6711,
      "step": 14741
    },
    {
      "epoch": 0.6109660574412533,
      "grad_norm": 0.4248223304748535,
      "learning_rate": 1.9453769323221023e-06,
      "loss": 0.6812,
      "step": 14742
    },
    {
      "epoch": 0.6110075013469269,
      "grad_norm": 0.41525113582611084,
      "learning_rate": 1.9451697127937337e-06,
      "loss": 0.6963,
      "step": 14743
    },
    {
      "epoch": 0.6110489452526006,
      "grad_norm": 0.4122079014778137,
      "learning_rate": 1.9449624932653655e-06,
      "loss": 0.6904,
      "step": 14744
    },
    {
      "epoch": 0.6110903891582743,
      "grad_norm": 0.4312734603881836,
      "learning_rate": 1.944755273736997e-06,
      "loss": 0.6884,
      "step": 14745
    },
    {
      "epoch": 0.6111318330639479,
      "grad_norm": 0.4094502925872803,
      "learning_rate": 1.9445480542086287e-06,
      "loss": 0.6588,
      "step": 14746
    },
    {
      "epoch": 0.6111732769696216,
      "grad_norm": 0.37969970703125,
      "learning_rate": 1.9443408346802605e-06,
      "loss": 0.6785,
      "step": 14747
    },
    {
      "epoch": 0.6112147208752953,
      "grad_norm": 0.43941259384155273,
      "learning_rate": 1.9441336151518923e-06,
      "loss": 0.7341,
      "step": 14748
    },
    {
      "epoch": 0.611256164780969,
      "grad_norm": 0.42629697918891907,
      "learning_rate": 1.9439263956235237e-06,
      "loss": 0.6454,
      "step": 14749
    },
    {
      "epoch": 0.6112976086866426,
      "grad_norm": 0.39709287881851196,
      "learning_rate": 1.9437191760951555e-06,
      "loss": 0.6837,
      "step": 14750
    },
    {
      "epoch": 0.6113390525923162,
      "grad_norm": 0.43552830815315247,
      "learning_rate": 1.943511956566787e-06,
      "loss": 0.7227,
      "step": 14751
    },
    {
      "epoch": 0.61138049649799,
      "grad_norm": 0.4556885063648224,
      "learning_rate": 1.9433047370384187e-06,
      "loss": 0.6604,
      "step": 14752
    },
    {
      "epoch": 0.6114219404036636,
      "grad_norm": 0.4166381359100342,
      "learning_rate": 1.94309751751005e-06,
      "loss": 0.6516,
      "step": 14753
    },
    {
      "epoch": 0.6114633843093373,
      "grad_norm": 0.412934809923172,
      "learning_rate": 1.942890297981682e-06,
      "loss": 0.7112,
      "step": 14754
    },
    {
      "epoch": 0.611504828215011,
      "grad_norm": 0.42269274592399597,
      "learning_rate": 1.9426830784533137e-06,
      "loss": 0.7017,
      "step": 14755
    },
    {
      "epoch": 0.6115462721206847,
      "grad_norm": 0.40668049454689026,
      "learning_rate": 1.942475858924945e-06,
      "loss": 0.7124,
      "step": 14756
    },
    {
      "epoch": 0.6115877160263583,
      "grad_norm": 0.46549752354621887,
      "learning_rate": 1.942268639396577e-06,
      "loss": 0.6997,
      "step": 14757
    },
    {
      "epoch": 0.611629159932032,
      "grad_norm": 0.43218499422073364,
      "learning_rate": 1.9420614198682087e-06,
      "loss": 0.6184,
      "step": 14758
    },
    {
      "epoch": 0.6116706038377057,
      "grad_norm": 0.47965919971466064,
      "learning_rate": 1.94185420033984e-06,
      "loss": 0.7339,
      "step": 14759
    },
    {
      "epoch": 0.6117120477433793,
      "grad_norm": 0.4407515227794647,
      "learning_rate": 1.941646980811472e-06,
      "loss": 0.731,
      "step": 14760
    },
    {
      "epoch": 0.611753491649053,
      "grad_norm": 0.42993903160095215,
      "learning_rate": 1.9414397612831033e-06,
      "loss": 0.683,
      "step": 14761
    },
    {
      "epoch": 0.6117949355547266,
      "grad_norm": 0.3894854187965393,
      "learning_rate": 1.941232541754735e-06,
      "loss": 0.6549,
      "step": 14762
    },
    {
      "epoch": 0.6118363794604004,
      "grad_norm": 0.4283939003944397,
      "learning_rate": 1.941025322226367e-06,
      "loss": 0.6792,
      "step": 14763
    },
    {
      "epoch": 0.611877823366074,
      "grad_norm": 0.39911699295043945,
      "learning_rate": 1.9408181026979983e-06,
      "loss": 0.6931,
      "step": 14764
    },
    {
      "epoch": 0.6119192672717477,
      "grad_norm": 0.40781930088996887,
      "learning_rate": 1.94061088316963e-06,
      "loss": 0.7069,
      "step": 14765
    },
    {
      "epoch": 0.6119607111774213,
      "grad_norm": 0.4218117296695709,
      "learning_rate": 1.940403663641262e-06,
      "loss": 0.694,
      "step": 14766
    },
    {
      "epoch": 0.6120021550830951,
      "grad_norm": 0.44536203145980835,
      "learning_rate": 1.9401964441128933e-06,
      "loss": 0.6798,
      "step": 14767
    },
    {
      "epoch": 0.6120435989887687,
      "grad_norm": 0.4527055323123932,
      "learning_rate": 1.939989224584525e-06,
      "loss": 0.7269,
      "step": 14768
    },
    {
      "epoch": 0.6120850428944423,
      "grad_norm": 0.4155116677284241,
      "learning_rate": 1.9397820050561565e-06,
      "loss": 0.6353,
      "step": 14769
    },
    {
      "epoch": 0.612126486800116,
      "grad_norm": 0.3994438052177429,
      "learning_rate": 1.9395747855277883e-06,
      "loss": 0.675,
      "step": 14770
    },
    {
      "epoch": 0.6121679307057897,
      "grad_norm": 0.440207302570343,
      "learning_rate": 1.93936756599942e-06,
      "loss": 0.6967,
      "step": 14771
    },
    {
      "epoch": 0.6122093746114634,
      "grad_norm": 0.42069166898727417,
      "learning_rate": 1.9391603464710515e-06,
      "loss": 0.7585,
      "step": 14772
    },
    {
      "epoch": 0.612250818517137,
      "grad_norm": 0.41963914036750793,
      "learning_rate": 1.9389531269426833e-06,
      "loss": 0.6858,
      "step": 14773
    },
    {
      "epoch": 0.6122922624228108,
      "grad_norm": 0.43636664748191833,
      "learning_rate": 1.938745907414315e-06,
      "loss": 0.6995,
      "step": 14774
    },
    {
      "epoch": 0.6123337063284844,
      "grad_norm": 0.4307825565338135,
      "learning_rate": 1.9385386878859465e-06,
      "loss": 0.6653,
      "step": 14775
    },
    {
      "epoch": 0.6123751502341581,
      "grad_norm": 0.40891873836517334,
      "learning_rate": 1.9383314683575783e-06,
      "loss": 0.702,
      "step": 14776
    },
    {
      "epoch": 0.6124165941398317,
      "grad_norm": 0.3961033225059509,
      "learning_rate": 1.9381242488292097e-06,
      "loss": 0.6343,
      "step": 14777
    },
    {
      "epoch": 0.6124580380455054,
      "grad_norm": 0.41844549775123596,
      "learning_rate": 1.9379170293008415e-06,
      "loss": 0.6797,
      "step": 14778
    },
    {
      "epoch": 0.6124994819511791,
      "grad_norm": 0.39802807569503784,
      "learning_rate": 1.937709809772473e-06,
      "loss": 0.6855,
      "step": 14779
    },
    {
      "epoch": 0.6125409258568527,
      "grad_norm": 0.41358545422554016,
      "learning_rate": 1.9375025902441047e-06,
      "loss": 0.6729,
      "step": 14780
    },
    {
      "epoch": 0.6125823697625264,
      "grad_norm": 0.4196234941482544,
      "learning_rate": 1.9372953707157365e-06,
      "loss": 0.645,
      "step": 14781
    },
    {
      "epoch": 0.6126238136682001,
      "grad_norm": 0.4037451446056366,
      "learning_rate": 1.937088151187368e-06,
      "loss": 0.7357,
      "step": 14782
    },
    {
      "epoch": 0.6126652575738738,
      "grad_norm": 0.3709436357021332,
      "learning_rate": 1.9368809316589997e-06,
      "loss": 0.6184,
      "step": 14783
    },
    {
      "epoch": 0.6127067014795474,
      "grad_norm": 0.43360579013824463,
      "learning_rate": 1.9366737121306315e-06,
      "loss": 0.7009,
      "step": 14784
    },
    {
      "epoch": 0.6127481453852212,
      "grad_norm": 0.40229877829551697,
      "learning_rate": 1.936466492602263e-06,
      "loss": 0.621,
      "step": 14785
    },
    {
      "epoch": 0.6127895892908948,
      "grad_norm": 0.4053835868835449,
      "learning_rate": 1.9362592730738947e-06,
      "loss": 0.6846,
      "step": 14786
    },
    {
      "epoch": 0.6128310331965684,
      "grad_norm": 0.41047248244285583,
      "learning_rate": 1.936052053545526e-06,
      "loss": 0.6677,
      "step": 14787
    },
    {
      "epoch": 0.6128724771022421,
      "grad_norm": 0.42960086464881897,
      "learning_rate": 1.935844834017158e-06,
      "loss": 0.7406,
      "step": 14788
    },
    {
      "epoch": 0.6129139210079158,
      "grad_norm": 0.44884783029556274,
      "learning_rate": 1.9356376144887897e-06,
      "loss": 0.6868,
      "step": 14789
    },
    {
      "epoch": 0.6129553649135895,
      "grad_norm": 0.4040799140930176,
      "learning_rate": 1.935430394960421e-06,
      "loss": 0.6816,
      "step": 14790
    },
    {
      "epoch": 0.6129968088192631,
      "grad_norm": 0.4729327857494354,
      "learning_rate": 1.935223175432053e-06,
      "loss": 0.771,
      "step": 14791
    },
    {
      "epoch": 0.6130382527249368,
      "grad_norm": 0.3861408829689026,
      "learning_rate": 1.9350159559036847e-06,
      "loss": 0.6472,
      "step": 14792
    },
    {
      "epoch": 0.6130796966306105,
      "grad_norm": 0.41926270723342896,
      "learning_rate": 1.934808736375316e-06,
      "loss": 0.7162,
      "step": 14793
    },
    {
      "epoch": 0.6131211405362842,
      "grad_norm": 0.4058879315853119,
      "learning_rate": 1.934601516846948e-06,
      "loss": 0.6503,
      "step": 14794
    },
    {
      "epoch": 0.6131625844419578,
      "grad_norm": 0.4022805988788605,
      "learning_rate": 1.9343942973185793e-06,
      "loss": 0.6672,
      "step": 14795
    },
    {
      "epoch": 0.6132040283476314,
      "grad_norm": 0.3895222544670105,
      "learning_rate": 1.934187077790211e-06,
      "loss": 0.6167,
      "step": 14796
    },
    {
      "epoch": 0.6132454722533052,
      "grad_norm": 0.4324630796909332,
      "learning_rate": 1.933979858261843e-06,
      "loss": 0.7085,
      "step": 14797
    },
    {
      "epoch": 0.6132869161589788,
      "grad_norm": 0.38302984833717346,
      "learning_rate": 1.9337726387334743e-06,
      "loss": 0.6218,
      "step": 14798
    },
    {
      "epoch": 0.6133283600646525,
      "grad_norm": 0.38687869906425476,
      "learning_rate": 1.933565419205106e-06,
      "loss": 0.6348,
      "step": 14799
    },
    {
      "epoch": 0.6133698039703261,
      "grad_norm": 0.43772369623184204,
      "learning_rate": 1.933358199676738e-06,
      "loss": 0.7412,
      "step": 14800
    },
    {
      "epoch": 0.6134112478759999,
      "grad_norm": 0.4139874279499054,
      "learning_rate": 1.9331509801483693e-06,
      "loss": 0.6993,
      "step": 14801
    },
    {
      "epoch": 0.6134526917816735,
      "grad_norm": 0.40119612216949463,
      "learning_rate": 1.932943760620001e-06,
      "loss": 0.679,
      "step": 14802
    },
    {
      "epoch": 0.6134941356873472,
      "grad_norm": 0.411485493183136,
      "learning_rate": 1.9327365410916325e-06,
      "loss": 0.6689,
      "step": 14803
    },
    {
      "epoch": 0.6135355795930209,
      "grad_norm": 0.35470786690711975,
      "learning_rate": 1.9325293215632643e-06,
      "loss": 0.678,
      "step": 14804
    },
    {
      "epoch": 0.6135770234986945,
      "grad_norm": 0.4190214276313782,
      "learning_rate": 1.932322102034896e-06,
      "loss": 0.6953,
      "step": 14805
    },
    {
      "epoch": 0.6136184674043682,
      "grad_norm": 0.415465384721756,
      "learning_rate": 1.9321148825065275e-06,
      "loss": 0.6909,
      "step": 14806
    },
    {
      "epoch": 0.6136599113100418,
      "grad_norm": 0.394703209400177,
      "learning_rate": 1.9319076629781593e-06,
      "loss": 0.7031,
      "step": 14807
    },
    {
      "epoch": 0.6137013552157156,
      "grad_norm": 0.41245317459106445,
      "learning_rate": 1.9317004434497907e-06,
      "loss": 0.6533,
      "step": 14808
    },
    {
      "epoch": 0.6137427991213892,
      "grad_norm": 0.4211778938770294,
      "learning_rate": 1.9314932239214225e-06,
      "loss": 0.6455,
      "step": 14809
    },
    {
      "epoch": 0.6137842430270629,
      "grad_norm": 0.3993346393108368,
      "learning_rate": 1.9312860043930543e-06,
      "loss": 0.6135,
      "step": 14810
    },
    {
      "epoch": 0.6138256869327365,
      "grad_norm": 0.4276604652404785,
      "learning_rate": 1.9310787848646857e-06,
      "loss": 0.7346,
      "step": 14811
    },
    {
      "epoch": 0.6138671308384102,
      "grad_norm": 0.4268699288368225,
      "learning_rate": 1.9308715653363175e-06,
      "loss": 0.6561,
      "step": 14812
    },
    {
      "epoch": 0.6139085747440839,
      "grad_norm": 0.4180394113063812,
      "learning_rate": 1.9306643458079493e-06,
      "loss": 0.6792,
      "step": 14813
    },
    {
      "epoch": 0.6139500186497575,
      "grad_norm": 0.412332147359848,
      "learning_rate": 1.9304571262795807e-06,
      "loss": 0.6461,
      "step": 14814
    },
    {
      "epoch": 0.6139914625554312,
      "grad_norm": 0.39873701333999634,
      "learning_rate": 1.9302499067512125e-06,
      "loss": 0.6033,
      "step": 14815
    },
    {
      "epoch": 0.6140329064611049,
      "grad_norm": 0.4248320460319519,
      "learning_rate": 1.930042687222844e-06,
      "loss": 0.6696,
      "step": 14816
    },
    {
      "epoch": 0.6140743503667786,
      "grad_norm": 0.42183688282966614,
      "learning_rate": 1.9298354676944757e-06,
      "loss": 0.6688,
      "step": 14817
    },
    {
      "epoch": 0.6141157942724522,
      "grad_norm": 0.4743417501449585,
      "learning_rate": 1.9296282481661075e-06,
      "loss": 0.76,
      "step": 14818
    },
    {
      "epoch": 0.614157238178126,
      "grad_norm": 0.4166763722896576,
      "learning_rate": 1.929421028637739e-06,
      "loss": 0.6864,
      "step": 14819
    },
    {
      "epoch": 0.6141986820837996,
      "grad_norm": 0.44243475794792175,
      "learning_rate": 1.9292138091093707e-06,
      "loss": 0.699,
      "step": 14820
    },
    {
      "epoch": 0.6142401259894732,
      "grad_norm": 0.42346176505088806,
      "learning_rate": 1.929006589581002e-06,
      "loss": 0.6879,
      "step": 14821
    },
    {
      "epoch": 0.6142815698951469,
      "grad_norm": 0.39672473073005676,
      "learning_rate": 1.928799370052634e-06,
      "loss": 0.6372,
      "step": 14822
    },
    {
      "epoch": 0.6143230138008205,
      "grad_norm": 0.40520238876342773,
      "learning_rate": 1.9285921505242657e-06,
      "loss": 0.6537,
      "step": 14823
    },
    {
      "epoch": 0.6143644577064943,
      "grad_norm": 0.41232243180274963,
      "learning_rate": 1.928384930995897e-06,
      "loss": 0.674,
      "step": 14824
    },
    {
      "epoch": 0.6144059016121679,
      "grad_norm": 0.4242294728755951,
      "learning_rate": 1.928177711467529e-06,
      "loss": 0.6809,
      "step": 14825
    },
    {
      "epoch": 0.6144473455178416,
      "grad_norm": 0.42215973138809204,
      "learning_rate": 1.9279704919391607e-06,
      "loss": 0.6526,
      "step": 14826
    },
    {
      "epoch": 0.6144887894235153,
      "grad_norm": 0.39736199378967285,
      "learning_rate": 1.927763272410792e-06,
      "loss": 0.7258,
      "step": 14827
    },
    {
      "epoch": 0.614530233329189,
      "grad_norm": 0.4220333397388458,
      "learning_rate": 1.927556052882424e-06,
      "loss": 0.6699,
      "step": 14828
    },
    {
      "epoch": 0.6145716772348626,
      "grad_norm": 0.37379756569862366,
      "learning_rate": 1.9273488333540553e-06,
      "loss": 0.6736,
      "step": 14829
    },
    {
      "epoch": 0.6146131211405362,
      "grad_norm": 0.4303602874279022,
      "learning_rate": 1.927141613825687e-06,
      "loss": 0.6771,
      "step": 14830
    },
    {
      "epoch": 0.61465456504621,
      "grad_norm": 0.38644057512283325,
      "learning_rate": 1.926934394297319e-06,
      "loss": 0.6578,
      "step": 14831
    },
    {
      "epoch": 0.6146960089518836,
      "grad_norm": 0.4041668474674225,
      "learning_rate": 1.9267271747689503e-06,
      "loss": 0.7058,
      "step": 14832
    },
    {
      "epoch": 0.6147374528575573,
      "grad_norm": 0.4768451750278473,
      "learning_rate": 1.926519955240582e-06,
      "loss": 0.7351,
      "step": 14833
    },
    {
      "epoch": 0.6147788967632309,
      "grad_norm": 0.45298323035240173,
      "learning_rate": 1.9263127357122135e-06,
      "loss": 0.6958,
      "step": 14834
    },
    {
      "epoch": 0.6148203406689047,
      "grad_norm": 0.4267730414867401,
      "learning_rate": 1.9261055161838453e-06,
      "loss": 0.6726,
      "step": 14835
    },
    {
      "epoch": 0.6148617845745783,
      "grad_norm": 0.3876250982284546,
      "learning_rate": 1.925898296655477e-06,
      "loss": 0.6481,
      "step": 14836
    },
    {
      "epoch": 0.614903228480252,
      "grad_norm": 0.44948717951774597,
      "learning_rate": 1.9256910771271085e-06,
      "loss": 0.6877,
      "step": 14837
    },
    {
      "epoch": 0.6149446723859257,
      "grad_norm": 0.444139301776886,
      "learning_rate": 1.9254838575987403e-06,
      "loss": 0.7229,
      "step": 14838
    },
    {
      "epoch": 0.6149861162915993,
      "grad_norm": 0.38683393597602844,
      "learning_rate": 1.925276638070372e-06,
      "loss": 0.6497,
      "step": 14839
    },
    {
      "epoch": 0.615027560197273,
      "grad_norm": 0.44527480006217957,
      "learning_rate": 1.9250694185420035e-06,
      "loss": 0.6726,
      "step": 14840
    },
    {
      "epoch": 0.6150690041029466,
      "grad_norm": 0.45373091101646423,
      "learning_rate": 1.9248621990136353e-06,
      "loss": 0.6779,
      "step": 14841
    },
    {
      "epoch": 0.6151104480086204,
      "grad_norm": 0.38458240032196045,
      "learning_rate": 1.9246549794852667e-06,
      "loss": 0.6538,
      "step": 14842
    },
    {
      "epoch": 0.615151891914294,
      "grad_norm": 0.4040895104408264,
      "learning_rate": 1.9244477599568985e-06,
      "loss": 0.6624,
      "step": 14843
    },
    {
      "epoch": 0.6151933358199677,
      "grad_norm": 0.43295571208000183,
      "learning_rate": 1.9242405404285303e-06,
      "loss": 0.6943,
      "step": 14844
    },
    {
      "epoch": 0.6152347797256413,
      "grad_norm": 0.4030100405216217,
      "learning_rate": 1.9240333209001617e-06,
      "loss": 0.6733,
      "step": 14845
    },
    {
      "epoch": 0.6152762236313151,
      "grad_norm": 0.4356665313243866,
      "learning_rate": 1.9238261013717935e-06,
      "loss": 0.6639,
      "step": 14846
    },
    {
      "epoch": 0.6153176675369887,
      "grad_norm": 0.4142886996269226,
      "learning_rate": 1.9236188818434253e-06,
      "loss": 0.6759,
      "step": 14847
    },
    {
      "epoch": 0.6153591114426623,
      "grad_norm": 0.4127577245235443,
      "learning_rate": 1.9234116623150567e-06,
      "loss": 0.6764,
      "step": 14848
    },
    {
      "epoch": 0.615400555348336,
      "grad_norm": 0.42041096091270447,
      "learning_rate": 1.9232044427866885e-06,
      "loss": 0.6532,
      "step": 14849
    },
    {
      "epoch": 0.6154419992540097,
      "grad_norm": 0.4055597484111786,
      "learning_rate": 1.92299722325832e-06,
      "loss": 0.7266,
      "step": 14850
    },
    {
      "epoch": 0.6154834431596834,
      "grad_norm": 0.44921913743019104,
      "learning_rate": 1.9227900037299517e-06,
      "loss": 0.6921,
      "step": 14851
    },
    {
      "epoch": 0.615524887065357,
      "grad_norm": 0.41817083954811096,
      "learning_rate": 1.9225827842015835e-06,
      "loss": 0.7317,
      "step": 14852
    },
    {
      "epoch": 0.6155663309710308,
      "grad_norm": 0.40512707829475403,
      "learning_rate": 1.922375564673215e-06,
      "loss": 0.6655,
      "step": 14853
    },
    {
      "epoch": 0.6156077748767044,
      "grad_norm": 0.41038820147514343,
      "learning_rate": 1.9221683451448467e-06,
      "loss": 0.6577,
      "step": 14854
    },
    {
      "epoch": 0.6156492187823781,
      "grad_norm": 0.4313015341758728,
      "learning_rate": 1.921961125616478e-06,
      "loss": 0.7192,
      "step": 14855
    },
    {
      "epoch": 0.6156906626880517,
      "grad_norm": 0.4224589169025421,
      "learning_rate": 1.92175390608811e-06,
      "loss": 0.6978,
      "step": 14856
    },
    {
      "epoch": 0.6157321065937253,
      "grad_norm": 0.4516347348690033,
      "learning_rate": 1.9215466865597417e-06,
      "loss": 0.7161,
      "step": 14857
    },
    {
      "epoch": 0.6157735504993991,
      "grad_norm": 0.3848966658115387,
      "learning_rate": 1.921339467031373e-06,
      "loss": 0.7047,
      "step": 14858
    },
    {
      "epoch": 0.6158149944050727,
      "grad_norm": 0.40078938007354736,
      "learning_rate": 1.921132247503005e-06,
      "loss": 0.6639,
      "step": 14859
    },
    {
      "epoch": 0.6158564383107464,
      "grad_norm": 0.4002012312412262,
      "learning_rate": 1.9209250279746363e-06,
      "loss": 0.6707,
      "step": 14860
    },
    {
      "epoch": 0.6158978822164201,
      "grad_norm": 0.41139712929725647,
      "learning_rate": 1.920717808446268e-06,
      "loss": 0.6652,
      "step": 14861
    },
    {
      "epoch": 0.6159393261220938,
      "grad_norm": 0.4415040910243988,
      "learning_rate": 1.9205105889179e-06,
      "loss": 0.6973,
      "step": 14862
    },
    {
      "epoch": 0.6159807700277674,
      "grad_norm": 0.39361584186553955,
      "learning_rate": 1.9203033693895313e-06,
      "loss": 0.6699,
      "step": 14863
    },
    {
      "epoch": 0.6160222139334411,
      "grad_norm": 0.40452995896339417,
      "learning_rate": 1.920096149861163e-06,
      "loss": 0.6863,
      "step": 14864
    },
    {
      "epoch": 0.6160636578391148,
      "grad_norm": 0.4272562861442566,
      "learning_rate": 1.919888930332795e-06,
      "loss": 0.688,
      "step": 14865
    },
    {
      "epoch": 0.6161051017447884,
      "grad_norm": 0.3984299600124359,
      "learning_rate": 1.9196817108044263e-06,
      "loss": 0.6877,
      "step": 14866
    },
    {
      "epoch": 0.6161465456504621,
      "grad_norm": 0.4425066113471985,
      "learning_rate": 1.919474491276058e-06,
      "loss": 0.7156,
      "step": 14867
    },
    {
      "epoch": 0.6161879895561357,
      "grad_norm": 0.4061056673526764,
      "learning_rate": 1.9192672717476895e-06,
      "loss": 0.6573,
      "step": 14868
    },
    {
      "epoch": 0.6162294334618095,
      "grad_norm": 0.4346085786819458,
      "learning_rate": 1.9190600522193213e-06,
      "loss": 0.6713,
      "step": 14869
    },
    {
      "epoch": 0.6162708773674831,
      "grad_norm": 0.40673112869262695,
      "learning_rate": 1.918852832690953e-06,
      "loss": 0.6685,
      "step": 14870
    },
    {
      "epoch": 0.6163123212731568,
      "grad_norm": 0.4579155743122101,
      "learning_rate": 1.9186456131625845e-06,
      "loss": 0.6896,
      "step": 14871
    },
    {
      "epoch": 0.6163537651788304,
      "grad_norm": 0.40163636207580566,
      "learning_rate": 1.9184383936342163e-06,
      "loss": 0.6322,
      "step": 14872
    },
    {
      "epoch": 0.6163952090845041,
      "grad_norm": 0.41845765709877014,
      "learning_rate": 1.918231174105848e-06,
      "loss": 0.6711,
      "step": 14873
    },
    {
      "epoch": 0.6164366529901778,
      "grad_norm": 0.42747995257377625,
      "learning_rate": 1.9180239545774795e-06,
      "loss": 0.696,
      "step": 14874
    },
    {
      "epoch": 0.6164780968958514,
      "grad_norm": 0.42356839776039124,
      "learning_rate": 1.9178167350491113e-06,
      "loss": 0.7166,
      "step": 14875
    },
    {
      "epoch": 0.6165195408015252,
      "grad_norm": 0.3664929270744324,
      "learning_rate": 1.9176095155207427e-06,
      "loss": 0.6681,
      "step": 14876
    },
    {
      "epoch": 0.6165609847071988,
      "grad_norm": 0.41942086815834045,
      "learning_rate": 1.9174022959923745e-06,
      "loss": 0.7156,
      "step": 14877
    },
    {
      "epoch": 0.6166024286128725,
      "grad_norm": 0.42870232462882996,
      "learning_rate": 1.917195076464006e-06,
      "loss": 0.7419,
      "step": 14878
    },
    {
      "epoch": 0.6166438725185461,
      "grad_norm": 0.3751935660839081,
      "learning_rate": 1.9169878569356377e-06,
      "loss": 0.646,
      "step": 14879
    },
    {
      "epoch": 0.6166853164242199,
      "grad_norm": 0.4315398037433624,
      "learning_rate": 1.9167806374072695e-06,
      "loss": 0.6926,
      "step": 14880
    },
    {
      "epoch": 0.6167267603298935,
      "grad_norm": 0.4243287444114685,
      "learning_rate": 1.9165734178789013e-06,
      "loss": 0.7197,
      "step": 14881
    },
    {
      "epoch": 0.6167682042355671,
      "grad_norm": 0.45076388120651245,
      "learning_rate": 1.9163661983505327e-06,
      "loss": 0.6943,
      "step": 14882
    },
    {
      "epoch": 0.6168096481412408,
      "grad_norm": 0.40727728605270386,
      "learning_rate": 1.9161589788221645e-06,
      "loss": 0.6779,
      "step": 14883
    },
    {
      "epoch": 0.6168510920469145,
      "grad_norm": 0.39620494842529297,
      "learning_rate": 1.915951759293796e-06,
      "loss": 0.6665,
      "step": 14884
    },
    {
      "epoch": 0.6168925359525882,
      "grad_norm": 0.5452051758766174,
      "learning_rate": 1.9157445397654277e-06,
      "loss": 0.7035,
      "step": 14885
    },
    {
      "epoch": 0.6169339798582618,
      "grad_norm": 0.38104408979415894,
      "learning_rate": 1.915537320237059e-06,
      "loss": 0.6692,
      "step": 14886
    },
    {
      "epoch": 0.6169754237639355,
      "grad_norm": 0.4062226116657257,
      "learning_rate": 1.915330100708691e-06,
      "loss": 0.6772,
      "step": 14887
    },
    {
      "epoch": 0.6170168676696092,
      "grad_norm": 0.41058653593063354,
      "learning_rate": 1.9151228811803227e-06,
      "loss": 0.7271,
      "step": 14888
    },
    {
      "epoch": 0.6170583115752829,
      "grad_norm": 0.41350728273391724,
      "learning_rate": 1.9149156616519545e-06,
      "loss": 0.6797,
      "step": 14889
    },
    {
      "epoch": 0.6170997554809565,
      "grad_norm": 0.4004344940185547,
      "learning_rate": 1.914708442123586e-06,
      "loss": 0.6262,
      "step": 14890
    },
    {
      "epoch": 0.6171411993866301,
      "grad_norm": 0.38142356276512146,
      "learning_rate": 1.9145012225952177e-06,
      "loss": 0.608,
      "step": 14891
    },
    {
      "epoch": 0.6171826432923039,
      "grad_norm": 0.421258807182312,
      "learning_rate": 1.914294003066849e-06,
      "loss": 0.6786,
      "step": 14892
    },
    {
      "epoch": 0.6172240871979775,
      "grad_norm": 0.40107259154319763,
      "learning_rate": 1.914086783538481e-06,
      "loss": 0.6498,
      "step": 14893
    },
    {
      "epoch": 0.6172655311036512,
      "grad_norm": 0.41604548692703247,
      "learning_rate": 1.9138795640101123e-06,
      "loss": 0.662,
      "step": 14894
    },
    {
      "epoch": 0.6173069750093249,
      "grad_norm": 0.3963643014431,
      "learning_rate": 1.913672344481744e-06,
      "loss": 0.6425,
      "step": 14895
    },
    {
      "epoch": 0.6173484189149986,
      "grad_norm": 0.38042542338371277,
      "learning_rate": 1.913465124953376e-06,
      "loss": 0.6752,
      "step": 14896
    },
    {
      "epoch": 0.6173898628206722,
      "grad_norm": 0.404562383890152,
      "learning_rate": 1.9132579054250073e-06,
      "loss": 0.6432,
      "step": 14897
    },
    {
      "epoch": 0.6174313067263459,
      "grad_norm": 0.4414012134075165,
      "learning_rate": 1.913050685896639e-06,
      "loss": 0.7288,
      "step": 14898
    },
    {
      "epoch": 0.6174727506320196,
      "grad_norm": 0.38333478569984436,
      "learning_rate": 1.912843466368271e-06,
      "loss": 0.645,
      "step": 14899
    },
    {
      "epoch": 0.6175141945376932,
      "grad_norm": 0.40459883213043213,
      "learning_rate": 1.9126362468399023e-06,
      "loss": 0.6426,
      "step": 14900
    },
    {
      "epoch": 0.6175556384433669,
      "grad_norm": 0.37731632590293884,
      "learning_rate": 1.912429027311534e-06,
      "loss": 0.6219,
      "step": 14901
    },
    {
      "epoch": 0.6175970823490405,
      "grad_norm": 0.4283466041088104,
      "learning_rate": 1.9122218077831655e-06,
      "loss": 0.7366,
      "step": 14902
    },
    {
      "epoch": 0.6176385262547143,
      "grad_norm": 0.422166645526886,
      "learning_rate": 1.9120145882547973e-06,
      "loss": 0.7004,
      "step": 14903
    },
    {
      "epoch": 0.6176799701603879,
      "grad_norm": 0.4033878445625305,
      "learning_rate": 1.9118073687264287e-06,
      "loss": 0.6685,
      "step": 14904
    },
    {
      "epoch": 0.6177214140660616,
      "grad_norm": 0.41786786913871765,
      "learning_rate": 1.9116001491980605e-06,
      "loss": 0.6709,
      "step": 14905
    },
    {
      "epoch": 0.6177628579717352,
      "grad_norm": 0.41466015577316284,
      "learning_rate": 1.9113929296696923e-06,
      "loss": 0.6608,
      "step": 14906
    },
    {
      "epoch": 0.617804301877409,
      "grad_norm": 0.4175218939781189,
      "learning_rate": 1.911185710141324e-06,
      "loss": 0.678,
      "step": 14907
    },
    {
      "epoch": 0.6178457457830826,
      "grad_norm": 0.3884677588939667,
      "learning_rate": 1.9109784906129555e-06,
      "loss": 0.687,
      "step": 14908
    },
    {
      "epoch": 0.6178871896887562,
      "grad_norm": 0.40910816192626953,
      "learning_rate": 1.9107712710845873e-06,
      "loss": 0.6479,
      "step": 14909
    },
    {
      "epoch": 0.61792863359443,
      "grad_norm": 0.4066471457481384,
      "learning_rate": 1.9105640515562187e-06,
      "loss": 0.6575,
      "step": 14910
    },
    {
      "epoch": 0.6179700775001036,
      "grad_norm": 0.44696930050849915,
      "learning_rate": 1.9103568320278505e-06,
      "loss": 0.6727,
      "step": 14911
    },
    {
      "epoch": 0.6180115214057773,
      "grad_norm": 0.42891427874565125,
      "learning_rate": 1.910149612499482e-06,
      "loss": 0.6761,
      "step": 14912
    },
    {
      "epoch": 0.6180529653114509,
      "grad_norm": 0.40125611424446106,
      "learning_rate": 1.9099423929711137e-06,
      "loss": 0.6354,
      "step": 14913
    },
    {
      "epoch": 0.6180944092171247,
      "grad_norm": 0.4250132739543915,
      "learning_rate": 1.9097351734427455e-06,
      "loss": 0.6689,
      "step": 14914
    },
    {
      "epoch": 0.6181358531227983,
      "grad_norm": 0.4029315114021301,
      "learning_rate": 1.9095279539143773e-06,
      "loss": 0.6781,
      "step": 14915
    },
    {
      "epoch": 0.618177297028472,
      "grad_norm": 0.415741890668869,
      "learning_rate": 1.9093207343860087e-06,
      "loss": 0.6262,
      "step": 14916
    },
    {
      "epoch": 0.6182187409341456,
      "grad_norm": 0.47684839367866516,
      "learning_rate": 1.9091135148576405e-06,
      "loss": 0.7717,
      "step": 14917
    },
    {
      "epoch": 0.6182601848398193,
      "grad_norm": 0.4134511351585388,
      "learning_rate": 1.908906295329272e-06,
      "loss": 0.7207,
      "step": 14918
    },
    {
      "epoch": 0.618301628745493,
      "grad_norm": 0.42611750960350037,
      "learning_rate": 1.9086990758009037e-06,
      "loss": 0.7107,
      "step": 14919
    },
    {
      "epoch": 0.6183430726511666,
      "grad_norm": 0.41064801812171936,
      "learning_rate": 1.908491856272535e-06,
      "loss": 0.7402,
      "step": 14920
    },
    {
      "epoch": 0.6183845165568403,
      "grad_norm": 0.4163835942745209,
      "learning_rate": 1.908284636744167e-06,
      "loss": 0.6388,
      "step": 14921
    },
    {
      "epoch": 0.618425960462514,
      "grad_norm": 0.43049153685569763,
      "learning_rate": 1.9080774172157987e-06,
      "loss": 0.7017,
      "step": 14922
    },
    {
      "epoch": 0.6184674043681877,
      "grad_norm": 0.42513760924339294,
      "learning_rate": 1.9078701976874305e-06,
      "loss": 0.6736,
      "step": 14923
    },
    {
      "epoch": 0.6185088482738613,
      "grad_norm": 0.4017379879951477,
      "learning_rate": 1.907662978159062e-06,
      "loss": 0.634,
      "step": 14924
    },
    {
      "epoch": 0.6185502921795349,
      "grad_norm": 0.44294729828834534,
      "learning_rate": 1.9074557586306937e-06,
      "loss": 0.675,
      "step": 14925
    },
    {
      "epoch": 0.6185917360852087,
      "grad_norm": 0.42675837874412537,
      "learning_rate": 1.907248539102325e-06,
      "loss": 0.6782,
      "step": 14926
    },
    {
      "epoch": 0.6186331799908823,
      "grad_norm": 0.4348944127559662,
      "learning_rate": 1.9070413195739567e-06,
      "loss": 0.6625,
      "step": 14927
    },
    {
      "epoch": 0.618674623896556,
      "grad_norm": 0.403538316488266,
      "learning_rate": 1.9068341000455883e-06,
      "loss": 0.6495,
      "step": 14928
    },
    {
      "epoch": 0.6187160678022297,
      "grad_norm": 0.3947971761226654,
      "learning_rate": 1.9066268805172201e-06,
      "loss": 0.6926,
      "step": 14929
    },
    {
      "epoch": 0.6187575117079034,
      "grad_norm": 0.38076117634773254,
      "learning_rate": 1.9064196609888517e-06,
      "loss": 0.6578,
      "step": 14930
    },
    {
      "epoch": 0.618798955613577,
      "grad_norm": 0.40068864822387695,
      "learning_rate": 1.9062124414604835e-06,
      "loss": 0.6851,
      "step": 14931
    },
    {
      "epoch": 0.6188403995192507,
      "grad_norm": 0.40060731768608093,
      "learning_rate": 1.9060052219321151e-06,
      "loss": 0.6152,
      "step": 14932
    },
    {
      "epoch": 0.6188818434249244,
      "grad_norm": 0.4181751310825348,
      "learning_rate": 1.9057980024037467e-06,
      "loss": 0.6448,
      "step": 14933
    },
    {
      "epoch": 0.618923287330598,
      "grad_norm": 0.49117594957351685,
      "learning_rate": 1.9055907828753783e-06,
      "loss": 0.6498,
      "step": 14934
    },
    {
      "epoch": 0.6189647312362717,
      "grad_norm": 0.422296941280365,
      "learning_rate": 1.90538356334701e-06,
      "loss": 0.7025,
      "step": 14935
    },
    {
      "epoch": 0.6190061751419453,
      "grad_norm": 0.4141998887062073,
      "learning_rate": 1.9051763438186415e-06,
      "loss": 0.6737,
      "step": 14936
    },
    {
      "epoch": 0.6190476190476191,
      "grad_norm": 0.4630289375782013,
      "learning_rate": 1.904969124290273e-06,
      "loss": 0.749,
      "step": 14937
    },
    {
      "epoch": 0.6190890629532927,
      "grad_norm": 0.3958395719528198,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.6467,
      "step": 14938
    },
    {
      "epoch": 0.6191305068589664,
      "grad_norm": 0.4212743639945984,
      "learning_rate": 1.9045546852335365e-06,
      "loss": 0.729,
      "step": 14939
    },
    {
      "epoch": 0.61917195076464,
      "grad_norm": 0.3834397494792938,
      "learning_rate": 1.9043474657051683e-06,
      "loss": 0.6473,
      "step": 14940
    },
    {
      "epoch": 0.6192133946703138,
      "grad_norm": 0.4089038372039795,
      "learning_rate": 1.9041402461768e-06,
      "loss": 0.6636,
      "step": 14941
    },
    {
      "epoch": 0.6192548385759874,
      "grad_norm": 0.4103119373321533,
      "learning_rate": 1.9039330266484315e-06,
      "loss": 0.6396,
      "step": 14942
    },
    {
      "epoch": 0.619296282481661,
      "grad_norm": 0.3904162049293518,
      "learning_rate": 1.9037258071200631e-06,
      "loss": 0.6376,
      "step": 14943
    },
    {
      "epoch": 0.6193377263873348,
      "grad_norm": 0.46241432428359985,
      "learning_rate": 1.9035185875916947e-06,
      "loss": 0.6804,
      "step": 14944
    },
    {
      "epoch": 0.6193791702930084,
      "grad_norm": 0.4125785827636719,
      "learning_rate": 1.9033113680633263e-06,
      "loss": 0.7063,
      "step": 14945
    },
    {
      "epoch": 0.6194206141986821,
      "grad_norm": 0.42722296714782715,
      "learning_rate": 1.9031041485349581e-06,
      "loss": 0.7131,
      "step": 14946
    },
    {
      "epoch": 0.6194620581043557,
      "grad_norm": 0.39716655015945435,
      "learning_rate": 1.9028969290065897e-06,
      "loss": 0.6526,
      "step": 14947
    },
    {
      "epoch": 0.6195035020100295,
      "grad_norm": 0.44048047065734863,
      "learning_rate": 1.9026897094782215e-06,
      "loss": 0.7021,
      "step": 14948
    },
    {
      "epoch": 0.6195449459157031,
      "grad_norm": 0.43738025426864624,
      "learning_rate": 1.9024824899498531e-06,
      "loss": 0.7559,
      "step": 14949
    },
    {
      "epoch": 0.6195863898213768,
      "grad_norm": 0.4228183329105377,
      "learning_rate": 1.9022752704214847e-06,
      "loss": 0.6721,
      "step": 14950
    },
    {
      "epoch": 0.6196278337270504,
      "grad_norm": 0.3939340114593506,
      "learning_rate": 1.9020680508931163e-06,
      "loss": 0.6486,
      "step": 14951
    },
    {
      "epoch": 0.6196692776327241,
      "grad_norm": 0.4195728600025177,
      "learning_rate": 1.901860831364748e-06,
      "loss": 0.6777,
      "step": 14952
    },
    {
      "epoch": 0.6197107215383978,
      "grad_norm": 0.4379664361476898,
      "learning_rate": 1.9016536118363795e-06,
      "loss": 0.6599,
      "step": 14953
    },
    {
      "epoch": 0.6197521654440714,
      "grad_norm": 0.44955509901046753,
      "learning_rate": 1.901446392308011e-06,
      "loss": 0.6971,
      "step": 14954
    },
    {
      "epoch": 0.6197936093497451,
      "grad_norm": 0.3924221992492676,
      "learning_rate": 1.901239172779643e-06,
      "loss": 0.6616,
      "step": 14955
    },
    {
      "epoch": 0.6198350532554188,
      "grad_norm": 0.8724302649497986,
      "learning_rate": 1.9010319532512745e-06,
      "loss": 0.6545,
      "step": 14956
    },
    {
      "epoch": 0.6198764971610925,
      "grad_norm": 0.3908036947250366,
      "learning_rate": 1.9008247337229063e-06,
      "loss": 0.6577,
      "step": 14957
    },
    {
      "epoch": 0.6199179410667661,
      "grad_norm": 0.4001157581806183,
      "learning_rate": 1.900617514194538e-06,
      "loss": 0.6919,
      "step": 14958
    },
    {
      "epoch": 0.6199593849724399,
      "grad_norm": 0.44106099009513855,
      "learning_rate": 1.9004102946661695e-06,
      "loss": 0.6902,
      "step": 14959
    },
    {
      "epoch": 0.6200008288781135,
      "grad_norm": 0.42071396112442017,
      "learning_rate": 1.9002030751378011e-06,
      "loss": 0.6913,
      "step": 14960
    },
    {
      "epoch": 0.6200422727837871,
      "grad_norm": 0.42299771308898926,
      "learning_rate": 1.8999958556094327e-06,
      "loss": 0.6606,
      "step": 14961
    },
    {
      "epoch": 0.6200837166894608,
      "grad_norm": 0.3929431736469269,
      "learning_rate": 1.8997886360810643e-06,
      "loss": 0.6619,
      "step": 14962
    },
    {
      "epoch": 0.6201251605951345,
      "grad_norm": 0.3986915647983551,
      "learning_rate": 1.8995814165526961e-06,
      "loss": 0.6357,
      "step": 14963
    },
    {
      "epoch": 0.6201666045008082,
      "grad_norm": 0.4467551112174988,
      "learning_rate": 1.8993741970243277e-06,
      "loss": 0.7078,
      "step": 14964
    },
    {
      "epoch": 0.6202080484064818,
      "grad_norm": 0.3988366425037384,
      "learning_rate": 1.8991669774959593e-06,
      "loss": 0.6376,
      "step": 14965
    },
    {
      "epoch": 0.6202494923121555,
      "grad_norm": 0.3880581557750702,
      "learning_rate": 1.8989597579675911e-06,
      "loss": 0.6836,
      "step": 14966
    },
    {
      "epoch": 0.6202909362178292,
      "grad_norm": 0.3960559368133545,
      "learning_rate": 1.8987525384392227e-06,
      "loss": 0.6541,
      "step": 14967
    },
    {
      "epoch": 0.6203323801235029,
      "grad_norm": 0.4429948627948761,
      "learning_rate": 1.8985453189108543e-06,
      "loss": 0.6792,
      "step": 14968
    },
    {
      "epoch": 0.6203738240291765,
      "grad_norm": 0.3930130898952484,
      "learning_rate": 1.898338099382486e-06,
      "loss": 0.6628,
      "step": 14969
    },
    {
      "epoch": 0.6204152679348501,
      "grad_norm": 0.4203483462333679,
      "learning_rate": 1.8981308798541175e-06,
      "loss": 0.7119,
      "step": 14970
    },
    {
      "epoch": 0.6204567118405239,
      "grad_norm": 0.38294193148612976,
      "learning_rate": 1.897923660325749e-06,
      "loss": 0.6827,
      "step": 14971
    },
    {
      "epoch": 0.6204981557461975,
      "grad_norm": 0.4226776659488678,
      "learning_rate": 1.897716440797381e-06,
      "loss": 0.6859,
      "step": 14972
    },
    {
      "epoch": 0.6205395996518712,
      "grad_norm": 0.37658610939979553,
      "learning_rate": 1.8975092212690125e-06,
      "loss": 0.6184,
      "step": 14973
    },
    {
      "epoch": 0.6205810435575448,
      "grad_norm": 0.3817873001098633,
      "learning_rate": 1.8973020017406443e-06,
      "loss": 0.6797,
      "step": 14974
    },
    {
      "epoch": 0.6206224874632186,
      "grad_norm": 0.3894873857498169,
      "learning_rate": 1.897094782212276e-06,
      "loss": 0.653,
      "step": 14975
    },
    {
      "epoch": 0.6206639313688922,
      "grad_norm": 0.43013253808021545,
      "learning_rate": 1.8968875626839075e-06,
      "loss": 0.7168,
      "step": 14976
    },
    {
      "epoch": 0.6207053752745659,
      "grad_norm": 0.41388076543807983,
      "learning_rate": 1.8966803431555391e-06,
      "loss": 0.6772,
      "step": 14977
    },
    {
      "epoch": 0.6207468191802396,
      "grad_norm": 0.39460816979408264,
      "learning_rate": 1.8964731236271707e-06,
      "loss": 0.6904,
      "step": 14978
    },
    {
      "epoch": 0.6207882630859132,
      "grad_norm": 0.4248746335506439,
      "learning_rate": 1.8962659040988023e-06,
      "loss": 0.6145,
      "step": 14979
    },
    {
      "epoch": 0.6208297069915869,
      "grad_norm": 0.4171145558357239,
      "learning_rate": 1.8960586845704341e-06,
      "loss": 0.6592,
      "step": 14980
    },
    {
      "epoch": 0.6208711508972605,
      "grad_norm": 0.39695125818252563,
      "learning_rate": 1.8958514650420657e-06,
      "loss": 0.6548,
      "step": 14981
    },
    {
      "epoch": 0.6209125948029343,
      "grad_norm": 0.44437888264656067,
      "learning_rate": 1.8956442455136973e-06,
      "loss": 0.7383,
      "step": 14982
    },
    {
      "epoch": 0.6209540387086079,
      "grad_norm": 0.44190120697021484,
      "learning_rate": 1.8954370259853291e-06,
      "loss": 0.6953,
      "step": 14983
    },
    {
      "epoch": 0.6209954826142816,
      "grad_norm": 0.4153902530670166,
      "learning_rate": 1.8952298064569607e-06,
      "loss": 0.6688,
      "step": 14984
    },
    {
      "epoch": 0.6210369265199552,
      "grad_norm": 0.433437317609787,
      "learning_rate": 1.8950225869285923e-06,
      "loss": 0.6873,
      "step": 14985
    },
    {
      "epoch": 0.6210783704256289,
      "grad_norm": 0.4019441306591034,
      "learning_rate": 1.894815367400224e-06,
      "loss": 0.6418,
      "step": 14986
    },
    {
      "epoch": 0.6211198143313026,
      "grad_norm": 0.4344424903392792,
      "learning_rate": 1.8946081478718555e-06,
      "loss": 0.667,
      "step": 14987
    },
    {
      "epoch": 0.6211612582369762,
      "grad_norm": 0.3985518217086792,
      "learning_rate": 1.8944009283434871e-06,
      "loss": 0.7131,
      "step": 14988
    },
    {
      "epoch": 0.6212027021426499,
      "grad_norm": 0.41022828221321106,
      "learning_rate": 1.894193708815119e-06,
      "loss": 0.6017,
      "step": 14989
    },
    {
      "epoch": 0.6212441460483236,
      "grad_norm": 0.45377737283706665,
      "learning_rate": 1.8939864892867505e-06,
      "loss": 0.7004,
      "step": 14990
    },
    {
      "epoch": 0.6212855899539973,
      "grad_norm": 0.4057566523551941,
      "learning_rate": 1.8937792697583821e-06,
      "loss": 0.6775,
      "step": 14991
    },
    {
      "epoch": 0.6213270338596709,
      "grad_norm": 0.4433775842189789,
      "learning_rate": 1.893572050230014e-06,
      "loss": 0.7225,
      "step": 14992
    },
    {
      "epoch": 0.6213684777653447,
      "grad_norm": 0.39640748500823975,
      "learning_rate": 1.8933648307016455e-06,
      "loss": 0.7086,
      "step": 14993
    },
    {
      "epoch": 0.6214099216710183,
      "grad_norm": 0.41678476333618164,
      "learning_rate": 1.8931576111732771e-06,
      "loss": 0.6775,
      "step": 14994
    },
    {
      "epoch": 0.6214513655766919,
      "grad_norm": 0.39584657549858093,
      "learning_rate": 1.8929503916449087e-06,
      "loss": 0.6914,
      "step": 14995
    },
    {
      "epoch": 0.6214928094823656,
      "grad_norm": 0.43262630701065063,
      "learning_rate": 1.8927431721165403e-06,
      "loss": 0.6804,
      "step": 14996
    },
    {
      "epoch": 0.6215342533880392,
      "grad_norm": 0.4005282521247864,
      "learning_rate": 1.8925359525881721e-06,
      "loss": 0.6304,
      "step": 14997
    },
    {
      "epoch": 0.621575697293713,
      "grad_norm": 0.41645312309265137,
      "learning_rate": 1.8923287330598037e-06,
      "loss": 0.7037,
      "step": 14998
    },
    {
      "epoch": 0.6216171411993866,
      "grad_norm": 0.38411542773246765,
      "learning_rate": 1.8921215135314353e-06,
      "loss": 0.6193,
      "step": 14999
    },
    {
      "epoch": 0.6216585851050603,
      "grad_norm": 0.41326069831848145,
      "learning_rate": 1.8919142940030671e-06,
      "loss": 0.6997,
      "step": 15000
    },
    {
      "epoch": 0.621700029010734,
      "grad_norm": 0.41125306487083435,
      "learning_rate": 1.8917070744746987e-06,
      "loss": 0.6681,
      "step": 15001
    },
    {
      "epoch": 0.6217414729164077,
      "grad_norm": 0.40434789657592773,
      "learning_rate": 1.8914998549463303e-06,
      "loss": 0.6708,
      "step": 15002
    },
    {
      "epoch": 0.6217829168220813,
      "grad_norm": 0.42135167121887207,
      "learning_rate": 1.891292635417962e-06,
      "loss": 0.7191,
      "step": 15003
    },
    {
      "epoch": 0.6218243607277549,
      "grad_norm": 0.4209967851638794,
      "learning_rate": 1.8910854158895935e-06,
      "loss": 0.7266,
      "step": 15004
    },
    {
      "epoch": 0.6218658046334287,
      "grad_norm": 0.40049102902412415,
      "learning_rate": 1.8908781963612251e-06,
      "loss": 0.7117,
      "step": 15005
    },
    {
      "epoch": 0.6219072485391023,
      "grad_norm": 0.4341377317905426,
      "learning_rate": 1.890670976832857e-06,
      "loss": 0.6531,
      "step": 15006
    },
    {
      "epoch": 0.621948692444776,
      "grad_norm": 0.41807329654693604,
      "learning_rate": 1.8904637573044885e-06,
      "loss": 0.6945,
      "step": 15007
    },
    {
      "epoch": 0.6219901363504496,
      "grad_norm": 0.4158197343349457,
      "learning_rate": 1.8902565377761201e-06,
      "loss": 0.6727,
      "step": 15008
    },
    {
      "epoch": 0.6220315802561234,
      "grad_norm": 0.3971371054649353,
      "learning_rate": 1.890049318247752e-06,
      "loss": 0.6394,
      "step": 15009
    },
    {
      "epoch": 0.622073024161797,
      "grad_norm": 0.4433320462703705,
      "learning_rate": 1.8898420987193835e-06,
      "loss": 0.6926,
      "step": 15010
    },
    {
      "epoch": 0.6221144680674707,
      "grad_norm": 0.40059325098991394,
      "learning_rate": 1.8896348791910151e-06,
      "loss": 0.6646,
      "step": 15011
    },
    {
      "epoch": 0.6221559119731443,
      "grad_norm": 0.42483288049697876,
      "learning_rate": 1.8894276596626467e-06,
      "loss": 0.6694,
      "step": 15012
    },
    {
      "epoch": 0.622197355878818,
      "grad_norm": 0.48396921157836914,
      "learning_rate": 1.8892204401342783e-06,
      "loss": 0.6907,
      "step": 15013
    },
    {
      "epoch": 0.6222387997844917,
      "grad_norm": 0.3992592394351959,
      "learning_rate": 1.8890132206059101e-06,
      "loss": 0.6373,
      "step": 15014
    },
    {
      "epoch": 0.6222802436901653,
      "grad_norm": 0.3913261592388153,
      "learning_rate": 1.8888060010775417e-06,
      "loss": 0.6519,
      "step": 15015
    },
    {
      "epoch": 0.6223216875958391,
      "grad_norm": 0.4558086693286896,
      "learning_rate": 1.8885987815491733e-06,
      "loss": 0.6808,
      "step": 15016
    },
    {
      "epoch": 0.6223631315015127,
      "grad_norm": 0.39488786458969116,
      "learning_rate": 1.888391562020805e-06,
      "loss": 0.6951,
      "step": 15017
    },
    {
      "epoch": 0.6224045754071864,
      "grad_norm": 0.41554129123687744,
      "learning_rate": 1.8881843424924367e-06,
      "loss": 0.7256,
      "step": 15018
    },
    {
      "epoch": 0.62244601931286,
      "grad_norm": 0.41894760727882385,
      "learning_rate": 1.8879771229640683e-06,
      "loss": 0.7087,
      "step": 15019
    },
    {
      "epoch": 0.6224874632185338,
      "grad_norm": 0.4097542464733124,
      "learning_rate": 1.8877699034357e-06,
      "loss": 0.6152,
      "step": 15020
    },
    {
      "epoch": 0.6225289071242074,
      "grad_norm": 0.43736496567726135,
      "learning_rate": 1.8875626839073315e-06,
      "loss": 0.7388,
      "step": 15021
    },
    {
      "epoch": 0.622570351029881,
      "grad_norm": 0.44094017148017883,
      "learning_rate": 1.8873554643789633e-06,
      "loss": 0.6211,
      "step": 15022
    },
    {
      "epoch": 0.6226117949355547,
      "grad_norm": 0.41009101271629333,
      "learning_rate": 1.887148244850595e-06,
      "loss": 0.6707,
      "step": 15023
    },
    {
      "epoch": 0.6226532388412284,
      "grad_norm": 0.41505664587020874,
      "learning_rate": 1.8869410253222265e-06,
      "loss": 0.6647,
      "step": 15024
    },
    {
      "epoch": 0.6226946827469021,
      "grad_norm": 0.42578113079071045,
      "learning_rate": 1.8867338057938581e-06,
      "loss": 0.6624,
      "step": 15025
    },
    {
      "epoch": 0.6227361266525757,
      "grad_norm": 0.4175681173801422,
      "learning_rate": 1.8865265862654897e-06,
      "loss": 0.6897,
      "step": 15026
    },
    {
      "epoch": 0.6227775705582494,
      "grad_norm": 0.402174174785614,
      "learning_rate": 1.8863193667371215e-06,
      "loss": 0.6423,
      "step": 15027
    },
    {
      "epoch": 0.6228190144639231,
      "grad_norm": 0.40228110551834106,
      "learning_rate": 1.8861121472087531e-06,
      "loss": 0.6448,
      "step": 15028
    },
    {
      "epoch": 0.6228604583695968,
      "grad_norm": 0.3848807215690613,
      "learning_rate": 1.8859049276803847e-06,
      "loss": 0.6399,
      "step": 15029
    },
    {
      "epoch": 0.6229019022752704,
      "grad_norm": 0.4526704251766205,
      "learning_rate": 1.8856977081520163e-06,
      "loss": 0.7163,
      "step": 15030
    },
    {
      "epoch": 0.622943346180944,
      "grad_norm": 0.39039579033851624,
      "learning_rate": 1.8854904886236481e-06,
      "loss": 0.6676,
      "step": 15031
    },
    {
      "epoch": 0.6229847900866178,
      "grad_norm": 0.41271987557411194,
      "learning_rate": 1.8852832690952797e-06,
      "loss": 0.6526,
      "step": 15032
    },
    {
      "epoch": 0.6230262339922914,
      "grad_norm": 0.447249174118042,
      "learning_rate": 1.8850760495669113e-06,
      "loss": 0.6683,
      "step": 15033
    },
    {
      "epoch": 0.6230676778979651,
      "grad_norm": 0.4017137885093689,
      "learning_rate": 1.884868830038543e-06,
      "loss": 0.6472,
      "step": 15034
    },
    {
      "epoch": 0.6231091218036388,
      "grad_norm": 0.3883378803730011,
      "learning_rate": 1.8846616105101747e-06,
      "loss": 0.6613,
      "step": 15035
    },
    {
      "epoch": 0.6231505657093125,
      "grad_norm": 0.4125553369522095,
      "learning_rate": 1.8844543909818063e-06,
      "loss": 0.7063,
      "step": 15036
    },
    {
      "epoch": 0.6231920096149861,
      "grad_norm": 0.39273759722709656,
      "learning_rate": 1.884247171453438e-06,
      "loss": 0.6382,
      "step": 15037
    },
    {
      "epoch": 0.6232334535206598,
      "grad_norm": 0.4009930491447449,
      "learning_rate": 1.8840399519250695e-06,
      "loss": 0.7129,
      "step": 15038
    },
    {
      "epoch": 0.6232748974263335,
      "grad_norm": 0.4156873822212219,
      "learning_rate": 1.8838327323967013e-06,
      "loss": 0.6938,
      "step": 15039
    },
    {
      "epoch": 0.6233163413320071,
      "grad_norm": 0.38735291361808777,
      "learning_rate": 1.883625512868333e-06,
      "loss": 0.6812,
      "step": 15040
    },
    {
      "epoch": 0.6233577852376808,
      "grad_norm": 0.398480087518692,
      "learning_rate": 1.8834182933399645e-06,
      "loss": 0.6405,
      "step": 15041
    },
    {
      "epoch": 0.6233992291433544,
      "grad_norm": 0.4149356186389923,
      "learning_rate": 1.8832110738115961e-06,
      "loss": 0.671,
      "step": 15042
    },
    {
      "epoch": 0.6234406730490282,
      "grad_norm": 0.44876888394355774,
      "learning_rate": 1.8830038542832277e-06,
      "loss": 0.7008,
      "step": 15043
    },
    {
      "epoch": 0.6234821169547018,
      "grad_norm": 0.3913521468639374,
      "learning_rate": 1.8827966347548595e-06,
      "loss": 0.6301,
      "step": 15044
    },
    {
      "epoch": 0.6235235608603755,
      "grad_norm": 0.4093199670314789,
      "learning_rate": 1.8825894152264911e-06,
      "loss": 0.6268,
      "step": 15045
    },
    {
      "epoch": 0.6235650047660491,
      "grad_norm": 0.42487987875938416,
      "learning_rate": 1.8823821956981227e-06,
      "loss": 0.671,
      "step": 15046
    },
    {
      "epoch": 0.6236064486717228,
      "grad_norm": 0.4048291742801666,
      "learning_rate": 1.8821749761697543e-06,
      "loss": 0.6765,
      "step": 15047
    },
    {
      "epoch": 0.6236478925773965,
      "grad_norm": 0.41267016530036926,
      "learning_rate": 1.8819677566413861e-06,
      "loss": 0.6121,
      "step": 15048
    },
    {
      "epoch": 0.6236893364830701,
      "grad_norm": 0.4013306796550751,
      "learning_rate": 1.8817605371130177e-06,
      "loss": 0.6578,
      "step": 15049
    },
    {
      "epoch": 0.6237307803887439,
      "grad_norm": 0.4232058525085449,
      "learning_rate": 1.8815533175846493e-06,
      "loss": 0.649,
      "step": 15050
    },
    {
      "epoch": 0.6237722242944175,
      "grad_norm": 0.4371868371963501,
      "learning_rate": 1.881346098056281e-06,
      "loss": 0.6807,
      "step": 15051
    },
    {
      "epoch": 0.6238136682000912,
      "grad_norm": 0.4070182740688324,
      "learning_rate": 1.8811388785279125e-06,
      "loss": 0.71,
      "step": 15052
    },
    {
      "epoch": 0.6238551121057648,
      "grad_norm": 0.4002971351146698,
      "learning_rate": 1.8809316589995443e-06,
      "loss": 0.7266,
      "step": 15053
    },
    {
      "epoch": 0.6238965560114386,
      "grad_norm": 0.44876009225845337,
      "learning_rate": 1.880724439471176e-06,
      "loss": 0.6953,
      "step": 15054
    },
    {
      "epoch": 0.6239379999171122,
      "grad_norm": 0.4260658919811249,
      "learning_rate": 1.8805172199428075e-06,
      "loss": 0.729,
      "step": 15055
    },
    {
      "epoch": 0.6239794438227858,
      "grad_norm": 0.4081728756427765,
      "learning_rate": 1.8803100004144393e-06,
      "loss": 0.658,
      "step": 15056
    },
    {
      "epoch": 0.6240208877284595,
      "grad_norm": 0.44687339663505554,
      "learning_rate": 1.880102780886071e-06,
      "loss": 0.6918,
      "step": 15057
    },
    {
      "epoch": 0.6240623316341332,
      "grad_norm": 0.45699000358581543,
      "learning_rate": 1.8798955613577025e-06,
      "loss": 0.7054,
      "step": 15058
    },
    {
      "epoch": 0.6241037755398069,
      "grad_norm": 0.39785438776016235,
      "learning_rate": 1.8796883418293341e-06,
      "loss": 0.6437,
      "step": 15059
    },
    {
      "epoch": 0.6241452194454805,
      "grad_norm": 0.41500505805015564,
      "learning_rate": 1.8794811223009657e-06,
      "loss": 0.7405,
      "step": 15060
    },
    {
      "epoch": 0.6241866633511542,
      "grad_norm": 0.3978816270828247,
      "learning_rate": 1.8792739027725975e-06,
      "loss": 0.7246,
      "step": 15061
    },
    {
      "epoch": 0.6242281072568279,
      "grad_norm": 0.41105613112449646,
      "learning_rate": 1.8790666832442291e-06,
      "loss": 0.6635,
      "step": 15062
    },
    {
      "epoch": 0.6242695511625016,
      "grad_norm": 0.3859884440898895,
      "learning_rate": 1.8788594637158607e-06,
      "loss": 0.7081,
      "step": 15063
    },
    {
      "epoch": 0.6243109950681752,
      "grad_norm": 0.4172648787498474,
      "learning_rate": 1.8786522441874923e-06,
      "loss": 0.7197,
      "step": 15064
    },
    {
      "epoch": 0.6243524389738488,
      "grad_norm": 0.452157199382782,
      "learning_rate": 1.8784450246591241e-06,
      "loss": 0.6871,
      "step": 15065
    },
    {
      "epoch": 0.6243938828795226,
      "grad_norm": 0.4066287875175476,
      "learning_rate": 1.8782378051307557e-06,
      "loss": 0.7166,
      "step": 15066
    },
    {
      "epoch": 0.6244353267851962,
      "grad_norm": 0.3994321823120117,
      "learning_rate": 1.8780305856023873e-06,
      "loss": 0.6371,
      "step": 15067
    },
    {
      "epoch": 0.6244767706908699,
      "grad_norm": 0.39441168308258057,
      "learning_rate": 1.877823366074019e-06,
      "loss": 0.6793,
      "step": 15068
    },
    {
      "epoch": 0.6245182145965436,
      "grad_norm": 0.3962359130382538,
      "learning_rate": 1.8776161465456505e-06,
      "loss": 0.6476,
      "step": 15069
    },
    {
      "epoch": 0.6245596585022173,
      "grad_norm": 0.4303707778453827,
      "learning_rate": 1.8774089270172823e-06,
      "loss": 0.6835,
      "step": 15070
    },
    {
      "epoch": 0.6246011024078909,
      "grad_norm": 0.40143388509750366,
      "learning_rate": 1.877201707488914e-06,
      "loss": 0.6473,
      "step": 15071
    },
    {
      "epoch": 0.6246425463135646,
      "grad_norm": 0.36913296580314636,
      "learning_rate": 1.8769944879605455e-06,
      "loss": 0.6344,
      "step": 15072
    },
    {
      "epoch": 0.6246839902192383,
      "grad_norm": 0.43045252561569214,
      "learning_rate": 1.8767872684321773e-06,
      "loss": 0.6953,
      "step": 15073
    },
    {
      "epoch": 0.6247254341249119,
      "grad_norm": 0.4206538796424866,
      "learning_rate": 1.876580048903809e-06,
      "loss": 0.6646,
      "step": 15074
    },
    {
      "epoch": 0.6247668780305856,
      "grad_norm": 0.43381690979003906,
      "learning_rate": 1.8763728293754405e-06,
      "loss": 0.731,
      "step": 15075
    },
    {
      "epoch": 0.6248083219362592,
      "grad_norm": 0.4109366536140442,
      "learning_rate": 1.8761656098470721e-06,
      "loss": 0.6802,
      "step": 15076
    },
    {
      "epoch": 0.624849765841933,
      "grad_norm": 0.39540112018585205,
      "learning_rate": 1.8759583903187037e-06,
      "loss": 0.6395,
      "step": 15077
    },
    {
      "epoch": 0.6248912097476066,
      "grad_norm": 0.4225304126739502,
      "learning_rate": 1.8757511707903353e-06,
      "loss": 0.6775,
      "step": 15078
    },
    {
      "epoch": 0.6249326536532803,
      "grad_norm": 0.4204528331756592,
      "learning_rate": 1.8755439512619671e-06,
      "loss": 0.7078,
      "step": 15079
    },
    {
      "epoch": 0.624974097558954,
      "grad_norm": 0.3918272852897644,
      "learning_rate": 1.8753367317335987e-06,
      "loss": 0.6548,
      "step": 15080
    },
    {
      "epoch": 0.6250155414646277,
      "grad_norm": 0.40808728337287903,
      "learning_rate": 1.8751295122052303e-06,
      "loss": 0.6567,
      "step": 15081
    },
    {
      "epoch": 0.6250569853703013,
      "grad_norm": 0.40610620379447937,
      "learning_rate": 1.8749222926768621e-06,
      "loss": 0.6636,
      "step": 15082
    },
    {
      "epoch": 0.6250984292759749,
      "grad_norm": 0.4115828275680542,
      "learning_rate": 1.8747150731484937e-06,
      "loss": 0.6501,
      "step": 15083
    },
    {
      "epoch": 0.6251398731816487,
      "grad_norm": 0.3989485800266266,
      "learning_rate": 1.8745078536201253e-06,
      "loss": 0.6448,
      "step": 15084
    },
    {
      "epoch": 0.6251813170873223,
      "grad_norm": 0.4250113070011139,
      "learning_rate": 1.874300634091757e-06,
      "loss": 0.7057,
      "step": 15085
    },
    {
      "epoch": 0.625222760992996,
      "grad_norm": 0.4170987606048584,
      "learning_rate": 1.8740934145633885e-06,
      "loss": 0.636,
      "step": 15086
    },
    {
      "epoch": 0.6252642048986696,
      "grad_norm": 0.4141305387020111,
      "learning_rate": 1.8738861950350201e-06,
      "loss": 0.639,
      "step": 15087
    },
    {
      "epoch": 0.6253056488043434,
      "grad_norm": 0.4290854334831238,
      "learning_rate": 1.873678975506652e-06,
      "loss": 0.6741,
      "step": 15088
    },
    {
      "epoch": 0.625347092710017,
      "grad_norm": 0.42094510793685913,
      "learning_rate": 1.8734717559782835e-06,
      "loss": 0.6975,
      "step": 15089
    },
    {
      "epoch": 0.6253885366156907,
      "grad_norm": 0.39019322395324707,
      "learning_rate": 1.8732645364499153e-06,
      "loss": 0.635,
      "step": 15090
    },
    {
      "epoch": 0.6254299805213643,
      "grad_norm": 0.386406809091568,
      "learning_rate": 1.873057316921547e-06,
      "loss": 0.7102,
      "step": 15091
    },
    {
      "epoch": 0.625471424427038,
      "grad_norm": 0.4018535315990448,
      "learning_rate": 1.8728500973931785e-06,
      "loss": 0.6487,
      "step": 15092
    },
    {
      "epoch": 0.6255128683327117,
      "grad_norm": 0.3832484185695648,
      "learning_rate": 1.8726428778648101e-06,
      "loss": 0.6288,
      "step": 15093
    },
    {
      "epoch": 0.6255543122383853,
      "grad_norm": 0.4227793216705322,
      "learning_rate": 1.8724356583364417e-06,
      "loss": 0.6934,
      "step": 15094
    },
    {
      "epoch": 0.625595756144059,
      "grad_norm": 0.417961448431015,
      "learning_rate": 1.8722284388080733e-06,
      "loss": 0.6837,
      "step": 15095
    },
    {
      "epoch": 0.6256372000497327,
      "grad_norm": 0.3924984931945801,
      "learning_rate": 1.8720212192797051e-06,
      "loss": 0.687,
      "step": 15096
    },
    {
      "epoch": 0.6256786439554064,
      "grad_norm": 0.40902218222618103,
      "learning_rate": 1.8718139997513367e-06,
      "loss": 0.6909,
      "step": 15097
    },
    {
      "epoch": 0.62572008786108,
      "grad_norm": 0.4203513562679291,
      "learning_rate": 1.8716067802229683e-06,
      "loss": 0.6975,
      "step": 15098
    },
    {
      "epoch": 0.6257615317667538,
      "grad_norm": 0.4369301199913025,
      "learning_rate": 1.8713995606946001e-06,
      "loss": 0.6747,
      "step": 15099
    },
    {
      "epoch": 0.6258029756724274,
      "grad_norm": 0.44389697909355164,
      "learning_rate": 1.8711923411662317e-06,
      "loss": 0.7258,
      "step": 15100
    },
    {
      "epoch": 0.625844419578101,
      "grad_norm": 0.3983655869960785,
      "learning_rate": 1.8709851216378633e-06,
      "loss": 0.7212,
      "step": 15101
    },
    {
      "epoch": 0.6258858634837747,
      "grad_norm": 0.4208645820617676,
      "learning_rate": 1.870777902109495e-06,
      "loss": 0.678,
      "step": 15102
    },
    {
      "epoch": 0.6259273073894484,
      "grad_norm": 0.40043142437934875,
      "learning_rate": 1.8705706825811265e-06,
      "loss": 0.6244,
      "step": 15103
    },
    {
      "epoch": 0.6259687512951221,
      "grad_norm": 0.3910313844680786,
      "learning_rate": 1.8703634630527581e-06,
      "loss": 0.7004,
      "step": 15104
    },
    {
      "epoch": 0.6260101952007957,
      "grad_norm": 0.3992200493812561,
      "learning_rate": 1.87015624352439e-06,
      "loss": 0.6951,
      "step": 15105
    },
    {
      "epoch": 0.6260516391064694,
      "grad_norm": 0.40548646450042725,
      "learning_rate": 1.8699490239960215e-06,
      "loss": 0.6779,
      "step": 15106
    },
    {
      "epoch": 0.6260930830121431,
      "grad_norm": 0.4485556185245514,
      "learning_rate": 1.8697418044676533e-06,
      "loss": 0.6847,
      "step": 15107
    },
    {
      "epoch": 0.6261345269178167,
      "grad_norm": 0.4766916334629059,
      "learning_rate": 1.869534584939285e-06,
      "loss": 0.7166,
      "step": 15108
    },
    {
      "epoch": 0.6261759708234904,
      "grad_norm": 0.3784847855567932,
      "learning_rate": 1.8693273654109165e-06,
      "loss": 0.6455,
      "step": 15109
    },
    {
      "epoch": 0.626217414729164,
      "grad_norm": 0.40612301230430603,
      "learning_rate": 1.8691201458825481e-06,
      "loss": 0.6523,
      "step": 15110
    },
    {
      "epoch": 0.6262588586348378,
      "grad_norm": 0.3945150077342987,
      "learning_rate": 1.8689129263541797e-06,
      "loss": 0.6362,
      "step": 15111
    },
    {
      "epoch": 0.6263003025405114,
      "grad_norm": 0.40678516030311584,
      "learning_rate": 1.8687057068258113e-06,
      "loss": 0.6758,
      "step": 15112
    },
    {
      "epoch": 0.6263417464461851,
      "grad_norm": 0.40930014848709106,
      "learning_rate": 1.868498487297443e-06,
      "loss": 0.6381,
      "step": 15113
    },
    {
      "epoch": 0.6263831903518587,
      "grad_norm": 0.41044357419013977,
      "learning_rate": 1.8682912677690747e-06,
      "loss": 0.7061,
      "step": 15114
    },
    {
      "epoch": 0.6264246342575325,
      "grad_norm": 0.4284885823726654,
      "learning_rate": 1.8680840482407065e-06,
      "loss": 0.7136,
      "step": 15115
    },
    {
      "epoch": 0.6264660781632061,
      "grad_norm": 0.42167797684669495,
      "learning_rate": 1.8678768287123381e-06,
      "loss": 0.7148,
      "step": 15116
    },
    {
      "epoch": 0.6265075220688797,
      "grad_norm": 0.4144901931285858,
      "learning_rate": 1.8676696091839697e-06,
      "loss": 0.6299,
      "step": 15117
    },
    {
      "epoch": 0.6265489659745535,
      "grad_norm": 0.45195692777633667,
      "learning_rate": 1.8674623896556013e-06,
      "loss": 0.7139,
      "step": 15118
    },
    {
      "epoch": 0.6265904098802271,
      "grad_norm": 0.4080878496170044,
      "learning_rate": 1.867255170127233e-06,
      "loss": 0.7197,
      "step": 15119
    },
    {
      "epoch": 0.6266318537859008,
      "grad_norm": 0.4332001209259033,
      "learning_rate": 1.8670479505988645e-06,
      "loss": 0.6882,
      "step": 15120
    },
    {
      "epoch": 0.6266732976915744,
      "grad_norm": 0.4235941767692566,
      "learning_rate": 1.8668407310704961e-06,
      "loss": 0.6592,
      "step": 15121
    },
    {
      "epoch": 0.6267147415972482,
      "grad_norm": 0.43270188570022583,
      "learning_rate": 1.866633511542128e-06,
      "loss": 0.6587,
      "step": 15122
    },
    {
      "epoch": 0.6267561855029218,
      "grad_norm": 0.43156272172927856,
      "learning_rate": 1.8664262920137595e-06,
      "loss": 0.697,
      "step": 15123
    },
    {
      "epoch": 0.6267976294085955,
      "grad_norm": 0.43355482816696167,
      "learning_rate": 1.8662190724853913e-06,
      "loss": 0.6638,
      "step": 15124
    },
    {
      "epoch": 0.6268390733142691,
      "grad_norm": 0.42373040318489075,
      "learning_rate": 1.866011852957023e-06,
      "loss": 0.6896,
      "step": 15125
    },
    {
      "epoch": 0.6268805172199428,
      "grad_norm": 0.40422338247299194,
      "learning_rate": 1.8658046334286545e-06,
      "loss": 0.6807,
      "step": 15126
    },
    {
      "epoch": 0.6269219611256165,
      "grad_norm": 0.4284586012363434,
      "learning_rate": 1.8655974139002861e-06,
      "loss": 0.7472,
      "step": 15127
    },
    {
      "epoch": 0.6269634050312901,
      "grad_norm": 0.38522276282310486,
      "learning_rate": 1.8653901943719177e-06,
      "loss": 0.6427,
      "step": 15128
    },
    {
      "epoch": 0.6270048489369638,
      "grad_norm": 0.4155053496360779,
      "learning_rate": 1.8651829748435493e-06,
      "loss": 0.7136,
      "step": 15129
    },
    {
      "epoch": 0.6270462928426375,
      "grad_norm": 0.41273677349090576,
      "learning_rate": 1.864975755315181e-06,
      "loss": 0.6053,
      "step": 15130
    },
    {
      "epoch": 0.6270877367483112,
      "grad_norm": 0.3879847228527069,
      "learning_rate": 1.8647685357868127e-06,
      "loss": 0.6958,
      "step": 15131
    },
    {
      "epoch": 0.6271291806539848,
      "grad_norm": 0.417024165391922,
      "learning_rate": 1.8645613162584445e-06,
      "loss": 0.6804,
      "step": 15132
    },
    {
      "epoch": 0.6271706245596586,
      "grad_norm": 0.41300007700920105,
      "learning_rate": 1.8643540967300761e-06,
      "loss": 0.6676,
      "step": 15133
    },
    {
      "epoch": 0.6272120684653322,
      "grad_norm": 0.39110273122787476,
      "learning_rate": 1.8641468772017077e-06,
      "loss": 0.676,
      "step": 15134
    },
    {
      "epoch": 0.6272535123710058,
      "grad_norm": 0.43686729669570923,
      "learning_rate": 1.8639396576733393e-06,
      "loss": 0.66,
      "step": 15135
    },
    {
      "epoch": 0.6272949562766795,
      "grad_norm": 0.4129782021045685,
      "learning_rate": 1.863732438144971e-06,
      "loss": 0.6675,
      "step": 15136
    },
    {
      "epoch": 0.6273364001823531,
      "grad_norm": 0.39705002307891846,
      "learning_rate": 1.8635252186166025e-06,
      "loss": 0.655,
      "step": 15137
    },
    {
      "epoch": 0.6273778440880269,
      "grad_norm": 0.3971523642539978,
      "learning_rate": 1.8633179990882341e-06,
      "loss": 0.6273,
      "step": 15138
    },
    {
      "epoch": 0.6274192879937005,
      "grad_norm": 0.39130452275276184,
      "learning_rate": 1.8631107795598657e-06,
      "loss": 0.6935,
      "step": 15139
    },
    {
      "epoch": 0.6274607318993742,
      "grad_norm": 0.3793759047985077,
      "learning_rate": 1.8629035600314975e-06,
      "loss": 0.6208,
      "step": 15140
    },
    {
      "epoch": 0.6275021758050479,
      "grad_norm": 0.4395447075366974,
      "learning_rate": 1.8626963405031293e-06,
      "loss": 0.6508,
      "step": 15141
    },
    {
      "epoch": 0.6275436197107216,
      "grad_norm": 0.37019193172454834,
      "learning_rate": 1.862489120974761e-06,
      "loss": 0.64,
      "step": 15142
    },
    {
      "epoch": 0.6275850636163952,
      "grad_norm": 0.41842857003211975,
      "learning_rate": 1.8622819014463925e-06,
      "loss": 0.6714,
      "step": 15143
    },
    {
      "epoch": 0.6276265075220688,
      "grad_norm": 0.4311401844024658,
      "learning_rate": 1.8620746819180241e-06,
      "loss": 0.7139,
      "step": 15144
    },
    {
      "epoch": 0.6276679514277426,
      "grad_norm": 0.45155060291290283,
      "learning_rate": 1.8618674623896557e-06,
      "loss": 0.6622,
      "step": 15145
    },
    {
      "epoch": 0.6277093953334162,
      "grad_norm": 0.41654863953590393,
      "learning_rate": 1.8616602428612873e-06,
      "loss": 0.6814,
      "step": 15146
    },
    {
      "epoch": 0.6277508392390899,
      "grad_norm": 0.3885108530521393,
      "learning_rate": 1.861453023332919e-06,
      "loss": 0.6542,
      "step": 15147
    },
    {
      "epoch": 0.6277922831447635,
      "grad_norm": 0.4091849625110626,
      "learning_rate": 1.8612458038045505e-06,
      "loss": 0.7192,
      "step": 15148
    },
    {
      "epoch": 0.6278337270504373,
      "grad_norm": 0.4416104555130005,
      "learning_rate": 1.8610385842761825e-06,
      "loss": 0.6902,
      "step": 15149
    },
    {
      "epoch": 0.6278751709561109,
      "grad_norm": 0.4733984172344208,
      "learning_rate": 1.8608313647478141e-06,
      "loss": 0.7188,
      "step": 15150
    },
    {
      "epoch": 0.6279166148617846,
      "grad_norm": 0.4048673212528229,
      "learning_rate": 1.8606241452194457e-06,
      "loss": 0.7031,
      "step": 15151
    },
    {
      "epoch": 0.6279580587674582,
      "grad_norm": 0.4496244490146637,
      "learning_rate": 1.8604169256910773e-06,
      "loss": 0.6943,
      "step": 15152
    },
    {
      "epoch": 0.6279995026731319,
      "grad_norm": 0.4038473665714264,
      "learning_rate": 1.860209706162709e-06,
      "loss": 0.7095,
      "step": 15153
    },
    {
      "epoch": 0.6280409465788056,
      "grad_norm": 0.396587997674942,
      "learning_rate": 1.8600024866343405e-06,
      "loss": 0.6833,
      "step": 15154
    },
    {
      "epoch": 0.6280823904844792,
      "grad_norm": 0.42150723934173584,
      "learning_rate": 1.8597952671059721e-06,
      "loss": 0.6842,
      "step": 15155
    },
    {
      "epoch": 0.628123834390153,
      "grad_norm": 0.44472354650497437,
      "learning_rate": 1.8595880475776037e-06,
      "loss": 0.6594,
      "step": 15156
    },
    {
      "epoch": 0.6281652782958266,
      "grad_norm": 0.4122158885002136,
      "learning_rate": 1.8593808280492355e-06,
      "loss": 0.6926,
      "step": 15157
    },
    {
      "epoch": 0.6282067222015003,
      "grad_norm": 0.42942675948143005,
      "learning_rate": 1.8591736085208673e-06,
      "loss": 0.6375,
      "step": 15158
    },
    {
      "epoch": 0.6282481661071739,
      "grad_norm": 0.3868406414985657,
      "learning_rate": 1.858966388992499e-06,
      "loss": 0.7087,
      "step": 15159
    },
    {
      "epoch": 0.6282896100128477,
      "grad_norm": 0.4299485385417938,
      "learning_rate": 1.8587591694641305e-06,
      "loss": 0.6677,
      "step": 15160
    },
    {
      "epoch": 0.6283310539185213,
      "grad_norm": 0.37471336126327515,
      "learning_rate": 1.8585519499357621e-06,
      "loss": 0.6267,
      "step": 15161
    },
    {
      "epoch": 0.6283724978241949,
      "grad_norm": 0.40147238969802856,
      "learning_rate": 1.8583447304073937e-06,
      "loss": 0.6475,
      "step": 15162
    },
    {
      "epoch": 0.6284139417298686,
      "grad_norm": 0.4224829375743866,
      "learning_rate": 1.8581375108790253e-06,
      "loss": 0.7129,
      "step": 15163
    },
    {
      "epoch": 0.6284553856355423,
      "grad_norm": 0.45749062299728394,
      "learning_rate": 1.857930291350657e-06,
      "loss": 0.7151,
      "step": 15164
    },
    {
      "epoch": 0.628496829541216,
      "grad_norm": 0.425036758184433,
      "learning_rate": 1.8577230718222885e-06,
      "loss": 0.6497,
      "step": 15165
    },
    {
      "epoch": 0.6285382734468896,
      "grad_norm": 0.4190179109573364,
      "learning_rate": 1.8575158522939206e-06,
      "loss": 0.6812,
      "step": 15166
    },
    {
      "epoch": 0.6285797173525633,
      "grad_norm": 0.3665629029273987,
      "learning_rate": 1.8573086327655521e-06,
      "loss": 0.6416,
      "step": 15167
    },
    {
      "epoch": 0.628621161258237,
      "grad_norm": 0.40463849902153015,
      "learning_rate": 1.8571014132371837e-06,
      "loss": 0.626,
      "step": 15168
    },
    {
      "epoch": 0.6286626051639106,
      "grad_norm": 0.395557165145874,
      "learning_rate": 1.8568941937088153e-06,
      "loss": 0.7063,
      "step": 15169
    },
    {
      "epoch": 0.6287040490695843,
      "grad_norm": 0.42815476655960083,
      "learning_rate": 1.856686974180447e-06,
      "loss": 0.6788,
      "step": 15170
    },
    {
      "epoch": 0.628745492975258,
      "grad_norm": 0.4760686755180359,
      "learning_rate": 1.8564797546520785e-06,
      "loss": 0.7314,
      "step": 15171
    },
    {
      "epoch": 0.6287869368809317,
      "grad_norm": 0.3949381411075592,
      "learning_rate": 1.8562725351237101e-06,
      "loss": 0.6464,
      "step": 15172
    },
    {
      "epoch": 0.6288283807866053,
      "grad_norm": 0.43162328004837036,
      "learning_rate": 1.8560653155953417e-06,
      "loss": 0.6986,
      "step": 15173
    },
    {
      "epoch": 0.628869824692279,
      "grad_norm": 0.42263323068618774,
      "learning_rate": 1.8558580960669733e-06,
      "loss": 0.7025,
      "step": 15174
    },
    {
      "epoch": 0.6289112685979527,
      "grad_norm": 0.40511828660964966,
      "learning_rate": 1.8556508765386054e-06,
      "loss": 0.6407,
      "step": 15175
    },
    {
      "epoch": 0.6289527125036264,
      "grad_norm": 0.3812329173088074,
      "learning_rate": 1.855443657010237e-06,
      "loss": 0.6875,
      "step": 15176
    },
    {
      "epoch": 0.6289941564093,
      "grad_norm": 0.3953416347503662,
      "learning_rate": 1.8552364374818685e-06,
      "loss": 0.7083,
      "step": 15177
    },
    {
      "epoch": 0.6290356003149736,
      "grad_norm": 0.398917555809021,
      "learning_rate": 1.8550292179535001e-06,
      "loss": 0.7025,
      "step": 15178
    },
    {
      "epoch": 0.6290770442206474,
      "grad_norm": 0.44467058777809143,
      "learning_rate": 1.8548219984251317e-06,
      "loss": 0.6532,
      "step": 15179
    },
    {
      "epoch": 0.629118488126321,
      "grad_norm": 0.4180728793144226,
      "learning_rate": 1.8546147788967633e-06,
      "loss": 0.6653,
      "step": 15180
    },
    {
      "epoch": 0.6291599320319947,
      "grad_norm": 0.4180881381034851,
      "learning_rate": 1.854407559368395e-06,
      "loss": 0.6609,
      "step": 15181
    },
    {
      "epoch": 0.6292013759376683,
      "grad_norm": 0.39680027961730957,
      "learning_rate": 1.8542003398400265e-06,
      "loss": 0.6814,
      "step": 15182
    },
    {
      "epoch": 0.6292428198433421,
      "grad_norm": 0.3958665430545807,
      "learning_rate": 1.8539931203116586e-06,
      "loss": 0.6393,
      "step": 15183
    },
    {
      "epoch": 0.6292842637490157,
      "grad_norm": 0.42513954639434814,
      "learning_rate": 1.8537859007832902e-06,
      "loss": 0.6758,
      "step": 15184
    },
    {
      "epoch": 0.6293257076546894,
      "grad_norm": 0.4442698061466217,
      "learning_rate": 1.8535786812549218e-06,
      "loss": 0.676,
      "step": 15185
    },
    {
      "epoch": 0.629367151560363,
      "grad_norm": 0.4365846514701843,
      "learning_rate": 1.8533714617265533e-06,
      "loss": 0.7424,
      "step": 15186
    },
    {
      "epoch": 0.6294085954660367,
      "grad_norm": 0.4201393723487854,
      "learning_rate": 1.853164242198185e-06,
      "loss": 0.6702,
      "step": 15187
    },
    {
      "epoch": 0.6294500393717104,
      "grad_norm": 0.4058719575405121,
      "learning_rate": 1.8529570226698165e-06,
      "loss": 0.6915,
      "step": 15188
    },
    {
      "epoch": 0.629491483277384,
      "grad_norm": 0.4285821318626404,
      "learning_rate": 1.8527498031414481e-06,
      "loss": 0.6866,
      "step": 15189
    },
    {
      "epoch": 0.6295329271830578,
      "grad_norm": 0.41526636481285095,
      "learning_rate": 1.8525425836130797e-06,
      "loss": 0.6919,
      "step": 15190
    },
    {
      "epoch": 0.6295743710887314,
      "grad_norm": 0.4664246737957001,
      "learning_rate": 1.8523353640847113e-06,
      "loss": 0.7064,
      "step": 15191
    },
    {
      "epoch": 0.6296158149944051,
      "grad_norm": 0.43230971693992615,
      "learning_rate": 1.8521281445563434e-06,
      "loss": 0.6716,
      "step": 15192
    },
    {
      "epoch": 0.6296572589000787,
      "grad_norm": 0.4061552584171295,
      "learning_rate": 1.851920925027975e-06,
      "loss": 0.6531,
      "step": 15193
    },
    {
      "epoch": 0.6296987028057525,
      "grad_norm": 0.3810943365097046,
      "learning_rate": 1.8517137054996066e-06,
      "loss": 0.6641,
      "step": 15194
    },
    {
      "epoch": 0.6297401467114261,
      "grad_norm": 0.48444515466690063,
      "learning_rate": 1.8515064859712381e-06,
      "loss": 0.6594,
      "step": 15195
    },
    {
      "epoch": 0.6297815906170997,
      "grad_norm": 0.41792500019073486,
      "learning_rate": 1.8512992664428697e-06,
      "loss": 0.6818,
      "step": 15196
    },
    {
      "epoch": 0.6298230345227734,
      "grad_norm": 0.421718955039978,
      "learning_rate": 1.8510920469145013e-06,
      "loss": 0.6433,
      "step": 15197
    },
    {
      "epoch": 0.6298644784284471,
      "grad_norm": 0.4035363793373108,
      "learning_rate": 1.850884827386133e-06,
      "loss": 0.7075,
      "step": 15198
    },
    {
      "epoch": 0.6299059223341208,
      "grad_norm": 0.40614035725593567,
      "learning_rate": 1.8506776078577645e-06,
      "loss": 0.6941,
      "step": 15199
    },
    {
      "epoch": 0.6299473662397944,
      "grad_norm": 0.4248254597187042,
      "learning_rate": 1.8504703883293961e-06,
      "loss": 0.62,
      "step": 15200
    },
    {
      "epoch": 0.6299888101454681,
      "grad_norm": 0.4242332875728607,
      "learning_rate": 1.8502631688010282e-06,
      "loss": 0.7292,
      "step": 15201
    },
    {
      "epoch": 0.6300302540511418,
      "grad_norm": 0.4281512200832367,
      "learning_rate": 1.8500559492726598e-06,
      "loss": 0.6765,
      "step": 15202
    },
    {
      "epoch": 0.6300716979568155,
      "grad_norm": 0.407836377620697,
      "learning_rate": 1.8498487297442914e-06,
      "loss": 0.6685,
      "step": 15203
    },
    {
      "epoch": 0.6301131418624891,
      "grad_norm": 0.43116500973701477,
      "learning_rate": 1.849641510215923e-06,
      "loss": 0.7086,
      "step": 15204
    },
    {
      "epoch": 0.6301545857681627,
      "grad_norm": 0.4226404130458832,
      "learning_rate": 1.8494342906875545e-06,
      "loss": 0.7434,
      "step": 15205
    },
    {
      "epoch": 0.6301960296738365,
      "grad_norm": 0.4231027364730835,
      "learning_rate": 1.8492270711591861e-06,
      "loss": 0.7186,
      "step": 15206
    },
    {
      "epoch": 0.6302374735795101,
      "grad_norm": 0.42640358209609985,
      "learning_rate": 1.8490198516308177e-06,
      "loss": 0.6899,
      "step": 15207
    },
    {
      "epoch": 0.6302789174851838,
      "grad_norm": 0.44813916087150574,
      "learning_rate": 1.8488126321024493e-06,
      "loss": 0.6973,
      "step": 15208
    },
    {
      "epoch": 0.6303203613908575,
      "grad_norm": 0.38663163781166077,
      "learning_rate": 1.8486054125740814e-06,
      "loss": 0.7173,
      "step": 15209
    },
    {
      "epoch": 0.6303618052965312,
      "grad_norm": 0.41024643182754517,
      "learning_rate": 1.848398193045713e-06,
      "loss": 0.65,
      "step": 15210
    },
    {
      "epoch": 0.6304032492022048,
      "grad_norm": 0.41117218136787415,
      "learning_rate": 1.8481909735173446e-06,
      "loss": 0.6982,
      "step": 15211
    },
    {
      "epoch": 0.6304446931078785,
      "grad_norm": 0.41058972477912903,
      "learning_rate": 1.8479837539889762e-06,
      "loss": 0.6865,
      "step": 15212
    },
    {
      "epoch": 0.6304861370135522,
      "grad_norm": 0.4117130637168884,
      "learning_rate": 1.8477765344606077e-06,
      "loss": 0.6892,
      "step": 15213
    },
    {
      "epoch": 0.6305275809192258,
      "grad_norm": 0.41765522956848145,
      "learning_rate": 1.8475693149322393e-06,
      "loss": 0.6382,
      "step": 15214
    },
    {
      "epoch": 0.6305690248248995,
      "grad_norm": 0.40810832381248474,
      "learning_rate": 1.847362095403871e-06,
      "loss": 0.6602,
      "step": 15215
    },
    {
      "epoch": 0.6306104687305731,
      "grad_norm": 0.40406930446624756,
      "learning_rate": 1.8471548758755025e-06,
      "loss": 0.6541,
      "step": 15216
    },
    {
      "epoch": 0.6306519126362469,
      "grad_norm": 0.39908838272094727,
      "learning_rate": 1.8469476563471341e-06,
      "loss": 0.6044,
      "step": 15217
    },
    {
      "epoch": 0.6306933565419205,
      "grad_norm": 0.4255857765674591,
      "learning_rate": 1.8467404368187662e-06,
      "loss": 0.7188,
      "step": 15218
    },
    {
      "epoch": 0.6307348004475942,
      "grad_norm": 0.43373048305511475,
      "learning_rate": 1.8465332172903978e-06,
      "loss": 0.6893,
      "step": 15219
    },
    {
      "epoch": 0.6307762443532678,
      "grad_norm": 0.42076992988586426,
      "learning_rate": 1.8463259977620294e-06,
      "loss": 0.7458,
      "step": 15220
    },
    {
      "epoch": 0.6308176882589416,
      "grad_norm": 0.37135007977485657,
      "learning_rate": 1.846118778233661e-06,
      "loss": 0.6047,
      "step": 15221
    },
    {
      "epoch": 0.6308591321646152,
      "grad_norm": 0.40021198987960815,
      "learning_rate": 1.8459115587052925e-06,
      "loss": 0.637,
      "step": 15222
    },
    {
      "epoch": 0.6309005760702888,
      "grad_norm": 0.41561463475227356,
      "learning_rate": 1.8457043391769241e-06,
      "loss": 0.6511,
      "step": 15223
    },
    {
      "epoch": 0.6309420199759626,
      "grad_norm": 0.41267129778862,
      "learning_rate": 1.8454971196485557e-06,
      "loss": 0.7021,
      "step": 15224
    },
    {
      "epoch": 0.6309834638816362,
      "grad_norm": 0.40497538447380066,
      "learning_rate": 1.8452899001201873e-06,
      "loss": 0.6499,
      "step": 15225
    },
    {
      "epoch": 0.6310249077873099,
      "grad_norm": 0.40299052000045776,
      "learning_rate": 1.845082680591819e-06,
      "loss": 0.6724,
      "step": 15226
    },
    {
      "epoch": 0.6310663516929835,
      "grad_norm": 0.42921754717826843,
      "learning_rate": 1.844875461063451e-06,
      "loss": 0.6539,
      "step": 15227
    },
    {
      "epoch": 0.6311077955986573,
      "grad_norm": 0.41832417249679565,
      "learning_rate": 1.8446682415350826e-06,
      "loss": 0.6912,
      "step": 15228
    },
    {
      "epoch": 0.6311492395043309,
      "grad_norm": 0.4399922788143158,
      "learning_rate": 1.8444610220067142e-06,
      "loss": 0.697,
      "step": 15229
    },
    {
      "epoch": 0.6311906834100045,
      "grad_norm": 0.4036673903465271,
      "learning_rate": 1.8442538024783458e-06,
      "loss": 0.7183,
      "step": 15230
    },
    {
      "epoch": 0.6312321273156782,
      "grad_norm": 0.38633206486701965,
      "learning_rate": 1.8440465829499773e-06,
      "loss": 0.7314,
      "step": 15231
    },
    {
      "epoch": 0.6312735712213519,
      "grad_norm": 0.4601365923881531,
      "learning_rate": 1.843839363421609e-06,
      "loss": 0.7257,
      "step": 15232
    },
    {
      "epoch": 0.6313150151270256,
      "grad_norm": 0.42420706152915955,
      "learning_rate": 1.8436321438932405e-06,
      "loss": 0.6193,
      "step": 15233
    },
    {
      "epoch": 0.6313564590326992,
      "grad_norm": 0.4008655846118927,
      "learning_rate": 1.8434249243648721e-06,
      "loss": 0.6926,
      "step": 15234
    },
    {
      "epoch": 0.631397902938373,
      "grad_norm": 0.4010036587715149,
      "learning_rate": 1.8432177048365037e-06,
      "loss": 0.6395,
      "step": 15235
    },
    {
      "epoch": 0.6314393468440466,
      "grad_norm": 0.44225746393203735,
      "learning_rate": 1.8430104853081358e-06,
      "loss": 0.7395,
      "step": 15236
    },
    {
      "epoch": 0.6314807907497203,
      "grad_norm": 0.39972952008247375,
      "learning_rate": 1.8428032657797674e-06,
      "loss": 0.6681,
      "step": 15237
    },
    {
      "epoch": 0.6315222346553939,
      "grad_norm": 0.4018649160861969,
      "learning_rate": 1.842596046251399e-06,
      "loss": 0.6624,
      "step": 15238
    },
    {
      "epoch": 0.6315636785610675,
      "grad_norm": 0.41557028889656067,
      "learning_rate": 1.8423888267230306e-06,
      "loss": 0.6838,
      "step": 15239
    },
    {
      "epoch": 0.6316051224667413,
      "grad_norm": 0.443386435508728,
      "learning_rate": 1.8421816071946621e-06,
      "loss": 0.7234,
      "step": 15240
    },
    {
      "epoch": 0.6316465663724149,
      "grad_norm": 0.3995409309864044,
      "learning_rate": 1.8419743876662937e-06,
      "loss": 0.6422,
      "step": 15241
    },
    {
      "epoch": 0.6316880102780886,
      "grad_norm": 0.41352614760398865,
      "learning_rate": 1.8417671681379253e-06,
      "loss": 0.6329,
      "step": 15242
    },
    {
      "epoch": 0.6317294541837623,
      "grad_norm": 0.4218639135360718,
      "learning_rate": 1.841559948609557e-06,
      "loss": 0.7217,
      "step": 15243
    },
    {
      "epoch": 0.631770898089436,
      "grad_norm": 0.4205247461795807,
      "learning_rate": 1.841352729081189e-06,
      "loss": 0.7363,
      "step": 15244
    },
    {
      "epoch": 0.6318123419951096,
      "grad_norm": 0.427081435918808,
      "learning_rate": 1.8411455095528206e-06,
      "loss": 0.6747,
      "step": 15245
    },
    {
      "epoch": 0.6318537859007833,
      "grad_norm": 0.40872371196746826,
      "learning_rate": 1.8409382900244522e-06,
      "loss": 0.6393,
      "step": 15246
    },
    {
      "epoch": 0.631895229806457,
      "grad_norm": 0.4331202507019043,
      "learning_rate": 1.8407310704960838e-06,
      "loss": 0.6864,
      "step": 15247
    },
    {
      "epoch": 0.6319366737121306,
      "grad_norm": 0.4182370603084564,
      "learning_rate": 1.8405238509677154e-06,
      "loss": 0.7489,
      "step": 15248
    },
    {
      "epoch": 0.6319781176178043,
      "grad_norm": 0.3782501220703125,
      "learning_rate": 1.840316631439347e-06,
      "loss": 0.662,
      "step": 15249
    },
    {
      "epoch": 0.6320195615234779,
      "grad_norm": 0.4139811098575592,
      "learning_rate": 1.8401094119109785e-06,
      "loss": 0.6851,
      "step": 15250
    },
    {
      "epoch": 0.6320610054291517,
      "grad_norm": 0.3864934742450714,
      "learning_rate": 1.8399021923826101e-06,
      "loss": 0.6639,
      "step": 15251
    },
    {
      "epoch": 0.6321024493348253,
      "grad_norm": 0.4106742739677429,
      "learning_rate": 1.8396949728542417e-06,
      "loss": 0.667,
      "step": 15252
    },
    {
      "epoch": 0.632143893240499,
      "grad_norm": 0.3970029652118683,
      "learning_rate": 1.8394877533258738e-06,
      "loss": 0.6279,
      "step": 15253
    },
    {
      "epoch": 0.6321853371461726,
      "grad_norm": 0.40461331605911255,
      "learning_rate": 1.8392805337975054e-06,
      "loss": 0.6738,
      "step": 15254
    },
    {
      "epoch": 0.6322267810518464,
      "grad_norm": 0.42556217312812805,
      "learning_rate": 1.839073314269137e-06,
      "loss": 0.71,
      "step": 15255
    },
    {
      "epoch": 0.63226822495752,
      "grad_norm": 0.4029914140701294,
      "learning_rate": 1.8388660947407686e-06,
      "loss": 0.6836,
      "step": 15256
    },
    {
      "epoch": 0.6323096688631936,
      "grad_norm": 0.42458978295326233,
      "learning_rate": 1.8386588752124002e-06,
      "loss": 0.7009,
      "step": 15257
    },
    {
      "epoch": 0.6323511127688674,
      "grad_norm": 0.4208468794822693,
      "learning_rate": 1.8384516556840317e-06,
      "loss": 0.6475,
      "step": 15258
    },
    {
      "epoch": 0.632392556674541,
      "grad_norm": 0.448294997215271,
      "learning_rate": 1.8382444361556633e-06,
      "loss": 0.6774,
      "step": 15259
    },
    {
      "epoch": 0.6324340005802147,
      "grad_norm": 0.43901365995407104,
      "learning_rate": 1.838037216627295e-06,
      "loss": 0.6459,
      "step": 15260
    },
    {
      "epoch": 0.6324754444858883,
      "grad_norm": 0.4139244854450226,
      "learning_rate": 1.8378299970989265e-06,
      "loss": 0.6479,
      "step": 15261
    },
    {
      "epoch": 0.6325168883915621,
      "grad_norm": 0.4156998097896576,
      "learning_rate": 1.8376227775705586e-06,
      "loss": 0.6459,
      "step": 15262
    },
    {
      "epoch": 0.6325583322972357,
      "grad_norm": 0.4145723879337311,
      "learning_rate": 1.8374155580421902e-06,
      "loss": 0.6858,
      "step": 15263
    },
    {
      "epoch": 0.6325997762029094,
      "grad_norm": 0.39287474751472473,
      "learning_rate": 1.8372083385138218e-06,
      "loss": 0.6941,
      "step": 15264
    },
    {
      "epoch": 0.632641220108583,
      "grad_norm": 0.4358798861503601,
      "learning_rate": 1.8370011189854534e-06,
      "loss": 0.676,
      "step": 15265
    },
    {
      "epoch": 0.6326826640142567,
      "grad_norm": 0.47609537839889526,
      "learning_rate": 1.836793899457085e-06,
      "loss": 0.6711,
      "step": 15266
    },
    {
      "epoch": 0.6327241079199304,
      "grad_norm": 0.41265836358070374,
      "learning_rate": 1.8365866799287165e-06,
      "loss": 0.6995,
      "step": 15267
    },
    {
      "epoch": 0.632765551825604,
      "grad_norm": 0.42632484436035156,
      "learning_rate": 1.8363794604003481e-06,
      "loss": 0.7131,
      "step": 15268
    },
    {
      "epoch": 0.6328069957312777,
      "grad_norm": 0.4007568359375,
      "learning_rate": 1.8361722408719797e-06,
      "loss": 0.6609,
      "step": 15269
    },
    {
      "epoch": 0.6328484396369514,
      "grad_norm": 0.42806461453437805,
      "learning_rate": 1.8359650213436118e-06,
      "loss": 0.6315,
      "step": 15270
    },
    {
      "epoch": 0.6328898835426251,
      "grad_norm": 0.4497212767601013,
      "learning_rate": 1.8357578018152434e-06,
      "loss": 0.7407,
      "step": 15271
    },
    {
      "epoch": 0.6329313274482987,
      "grad_norm": 0.435983270406723,
      "learning_rate": 1.835550582286875e-06,
      "loss": 0.6835,
      "step": 15272
    },
    {
      "epoch": 0.6329727713539725,
      "grad_norm": 0.39427804946899414,
      "learning_rate": 1.8353433627585066e-06,
      "loss": 0.7013,
      "step": 15273
    },
    {
      "epoch": 0.6330142152596461,
      "grad_norm": 0.41245415806770325,
      "learning_rate": 1.8351361432301382e-06,
      "loss": 0.6345,
      "step": 15274
    },
    {
      "epoch": 0.6330556591653197,
      "grad_norm": 0.3692905008792877,
      "learning_rate": 1.8349289237017698e-06,
      "loss": 0.6273,
      "step": 15275
    },
    {
      "epoch": 0.6330971030709934,
      "grad_norm": 0.40609487891197205,
      "learning_rate": 1.8347217041734014e-06,
      "loss": 0.6486,
      "step": 15276
    },
    {
      "epoch": 0.633138546976667,
      "grad_norm": 0.4210486114025116,
      "learning_rate": 1.834514484645033e-06,
      "loss": 0.6552,
      "step": 15277
    },
    {
      "epoch": 0.6331799908823408,
      "grad_norm": 0.4253741204738617,
      "learning_rate": 1.8343072651166645e-06,
      "loss": 0.6888,
      "step": 15278
    },
    {
      "epoch": 0.6332214347880144,
      "grad_norm": 0.4268137514591217,
      "learning_rate": 1.8341000455882966e-06,
      "loss": 0.6389,
      "step": 15279
    },
    {
      "epoch": 0.6332628786936881,
      "grad_norm": 0.40833765268325806,
      "learning_rate": 1.8338928260599282e-06,
      "loss": 0.714,
      "step": 15280
    },
    {
      "epoch": 0.6333043225993618,
      "grad_norm": 0.3846840262413025,
      "learning_rate": 1.8336856065315598e-06,
      "loss": 0.6646,
      "step": 15281
    },
    {
      "epoch": 0.6333457665050354,
      "grad_norm": 0.44544732570648193,
      "learning_rate": 1.8334783870031914e-06,
      "loss": 0.7074,
      "step": 15282
    },
    {
      "epoch": 0.6333872104107091,
      "grad_norm": 0.45123475790023804,
      "learning_rate": 1.833271167474823e-06,
      "loss": 0.7048,
      "step": 15283
    },
    {
      "epoch": 0.6334286543163827,
      "grad_norm": 0.4004644453525543,
      "learning_rate": 1.8330639479464546e-06,
      "loss": 0.7075,
      "step": 15284
    },
    {
      "epoch": 0.6334700982220565,
      "grad_norm": 0.3853166401386261,
      "learning_rate": 1.8328567284180862e-06,
      "loss": 0.6354,
      "step": 15285
    },
    {
      "epoch": 0.6335115421277301,
      "grad_norm": 0.41783902049064636,
      "learning_rate": 1.8326495088897177e-06,
      "loss": 0.6665,
      "step": 15286
    },
    {
      "epoch": 0.6335529860334038,
      "grad_norm": 0.4098933935165405,
      "learning_rate": 1.8324422893613493e-06,
      "loss": 0.7092,
      "step": 15287
    },
    {
      "epoch": 0.6335944299390774,
      "grad_norm": 0.4107879102230072,
      "learning_rate": 1.8322350698329814e-06,
      "loss": 0.6567,
      "step": 15288
    },
    {
      "epoch": 0.6336358738447512,
      "grad_norm": 0.4225985109806061,
      "learning_rate": 1.832027850304613e-06,
      "loss": 0.6829,
      "step": 15289
    },
    {
      "epoch": 0.6336773177504248,
      "grad_norm": 0.40843454003334045,
      "learning_rate": 1.8318206307762446e-06,
      "loss": 0.6473,
      "step": 15290
    },
    {
      "epoch": 0.6337187616560984,
      "grad_norm": 0.37326687574386597,
      "learning_rate": 1.8316134112478762e-06,
      "loss": 0.6145,
      "step": 15291
    },
    {
      "epoch": 0.6337602055617721,
      "grad_norm": 0.4108583629131317,
      "learning_rate": 1.8314061917195078e-06,
      "loss": 0.6711,
      "step": 15292
    },
    {
      "epoch": 0.6338016494674458,
      "grad_norm": 0.39427095651626587,
      "learning_rate": 1.8311989721911394e-06,
      "loss": 0.6506,
      "step": 15293
    },
    {
      "epoch": 0.6338430933731195,
      "grad_norm": 0.4485965371131897,
      "learning_rate": 1.830991752662771e-06,
      "loss": 0.6656,
      "step": 15294
    },
    {
      "epoch": 0.6338845372787931,
      "grad_norm": 0.4352828562259674,
      "learning_rate": 1.8307845331344025e-06,
      "loss": 0.6816,
      "step": 15295
    },
    {
      "epoch": 0.6339259811844669,
      "grad_norm": 0.4076905846595764,
      "learning_rate": 1.8305773136060341e-06,
      "loss": 0.6594,
      "step": 15296
    },
    {
      "epoch": 0.6339674250901405,
      "grad_norm": 0.3939104974269867,
      "learning_rate": 1.8303700940776662e-06,
      "loss": 0.6646,
      "step": 15297
    },
    {
      "epoch": 0.6340088689958142,
      "grad_norm": 0.4331432580947876,
      "learning_rate": 1.8301628745492978e-06,
      "loss": 0.6991,
      "step": 15298
    },
    {
      "epoch": 0.6340503129014878,
      "grad_norm": 0.423004150390625,
      "learning_rate": 1.8299556550209294e-06,
      "loss": 0.6346,
      "step": 15299
    },
    {
      "epoch": 0.6340917568071615,
      "grad_norm": 0.48215097188949585,
      "learning_rate": 1.829748435492561e-06,
      "loss": 0.6926,
      "step": 15300
    },
    {
      "epoch": 0.6341332007128352,
      "grad_norm": 0.4022946357727051,
      "learning_rate": 1.8295412159641926e-06,
      "loss": 0.6652,
      "step": 15301
    },
    {
      "epoch": 0.6341746446185088,
      "grad_norm": 0.4577620029449463,
      "learning_rate": 1.8293339964358242e-06,
      "loss": 0.6606,
      "step": 15302
    },
    {
      "epoch": 0.6342160885241825,
      "grad_norm": 0.4438524842262268,
      "learning_rate": 1.8291267769074558e-06,
      "loss": 0.7141,
      "step": 15303
    },
    {
      "epoch": 0.6342575324298562,
      "grad_norm": 0.41563931107521057,
      "learning_rate": 1.8289195573790873e-06,
      "loss": 0.7114,
      "step": 15304
    },
    {
      "epoch": 0.6342989763355299,
      "grad_norm": 0.4010871350765228,
      "learning_rate": 1.8287123378507194e-06,
      "loss": 0.6757,
      "step": 15305
    },
    {
      "epoch": 0.6343404202412035,
      "grad_norm": 0.4367063045501709,
      "learning_rate": 1.828505118322351e-06,
      "loss": 0.6665,
      "step": 15306
    },
    {
      "epoch": 0.6343818641468773,
      "grad_norm": 0.40178465843200684,
      "learning_rate": 1.8282978987939826e-06,
      "loss": 0.6816,
      "step": 15307
    },
    {
      "epoch": 0.6344233080525509,
      "grad_norm": 0.4690416753292084,
      "learning_rate": 1.8280906792656142e-06,
      "loss": 0.7463,
      "step": 15308
    },
    {
      "epoch": 0.6344647519582245,
      "grad_norm": 0.40547385811805725,
      "learning_rate": 1.8278834597372458e-06,
      "loss": 0.7329,
      "step": 15309
    },
    {
      "epoch": 0.6345061958638982,
      "grad_norm": 0.4524579644203186,
      "learning_rate": 1.8276762402088774e-06,
      "loss": 0.7212,
      "step": 15310
    },
    {
      "epoch": 0.6345476397695718,
      "grad_norm": 0.40013644099235535,
      "learning_rate": 1.827469020680509e-06,
      "loss": 0.6991,
      "step": 15311
    },
    {
      "epoch": 0.6345890836752456,
      "grad_norm": 0.4508562386035919,
      "learning_rate": 1.8272618011521406e-06,
      "loss": 0.7095,
      "step": 15312
    },
    {
      "epoch": 0.6346305275809192,
      "grad_norm": 0.39066797494888306,
      "learning_rate": 1.8270545816237721e-06,
      "loss": 0.6245,
      "step": 15313
    },
    {
      "epoch": 0.6346719714865929,
      "grad_norm": 0.409310519695282,
      "learning_rate": 1.8268473620954042e-06,
      "loss": 0.6384,
      "step": 15314
    },
    {
      "epoch": 0.6347134153922666,
      "grad_norm": 0.38556766510009766,
      "learning_rate": 1.8266401425670358e-06,
      "loss": 0.6814,
      "step": 15315
    },
    {
      "epoch": 0.6347548592979403,
      "grad_norm": 0.3967755436897278,
      "learning_rate": 1.8264329230386674e-06,
      "loss": 0.6226,
      "step": 15316
    },
    {
      "epoch": 0.6347963032036139,
      "grad_norm": 0.42205455899238586,
      "learning_rate": 1.826225703510299e-06,
      "loss": 0.6769,
      "step": 15317
    },
    {
      "epoch": 0.6348377471092875,
      "grad_norm": 0.43861979246139526,
      "learning_rate": 1.8260184839819306e-06,
      "loss": 0.6919,
      "step": 15318
    },
    {
      "epoch": 0.6348791910149613,
      "grad_norm": 0.4533439874649048,
      "learning_rate": 1.8258112644535622e-06,
      "loss": 0.756,
      "step": 15319
    },
    {
      "epoch": 0.6349206349206349,
      "grad_norm": 0.46745434403419495,
      "learning_rate": 1.8256040449251938e-06,
      "loss": 0.793,
      "step": 15320
    },
    {
      "epoch": 0.6349620788263086,
      "grad_norm": 0.39336535334587097,
      "learning_rate": 1.8253968253968254e-06,
      "loss": 0.7023,
      "step": 15321
    },
    {
      "epoch": 0.6350035227319822,
      "grad_norm": 0.39952343702316284,
      "learning_rate": 1.8251896058684572e-06,
      "loss": 0.6448,
      "step": 15322
    },
    {
      "epoch": 0.635044966637656,
      "grad_norm": 0.43429267406463623,
      "learning_rate": 1.824982386340089e-06,
      "loss": 0.7242,
      "step": 15323
    },
    {
      "epoch": 0.6350864105433296,
      "grad_norm": 0.4304617643356323,
      "learning_rate": 1.8247751668117206e-06,
      "loss": 0.7037,
      "step": 15324
    },
    {
      "epoch": 0.6351278544490033,
      "grad_norm": 0.39618027210235596,
      "learning_rate": 1.8245679472833522e-06,
      "loss": 0.7019,
      "step": 15325
    },
    {
      "epoch": 0.635169298354677,
      "grad_norm": 0.4075004756450653,
      "learning_rate": 1.8243607277549838e-06,
      "loss": 0.678,
      "step": 15326
    },
    {
      "epoch": 0.6352107422603506,
      "grad_norm": 0.3968529999256134,
      "learning_rate": 1.8241535082266154e-06,
      "loss": 0.6562,
      "step": 15327
    },
    {
      "epoch": 0.6352521861660243,
      "grad_norm": 0.39338448643684387,
      "learning_rate": 1.823946288698247e-06,
      "loss": 0.6641,
      "step": 15328
    },
    {
      "epoch": 0.6352936300716979,
      "grad_norm": 0.4805199205875397,
      "learning_rate": 1.8237390691698786e-06,
      "loss": 0.7144,
      "step": 15329
    },
    {
      "epoch": 0.6353350739773717,
      "grad_norm": 0.40466493368148804,
      "learning_rate": 1.8235318496415102e-06,
      "loss": 0.7489,
      "step": 15330
    },
    {
      "epoch": 0.6353765178830453,
      "grad_norm": 0.4038875102996826,
      "learning_rate": 1.8233246301131422e-06,
      "loss": 0.6357,
      "step": 15331
    },
    {
      "epoch": 0.635417961788719,
      "grad_norm": 0.3885483741760254,
      "learning_rate": 1.8231174105847738e-06,
      "loss": 0.6814,
      "step": 15332
    },
    {
      "epoch": 0.6354594056943926,
      "grad_norm": 0.389848917722702,
      "learning_rate": 1.8229101910564054e-06,
      "loss": 0.6665,
      "step": 15333
    },
    {
      "epoch": 0.6355008496000664,
      "grad_norm": 0.41061657667160034,
      "learning_rate": 1.822702971528037e-06,
      "loss": 0.7104,
      "step": 15334
    },
    {
      "epoch": 0.63554229350574,
      "grad_norm": 0.40850281715393066,
      "learning_rate": 1.8224957519996686e-06,
      "loss": 0.7009,
      "step": 15335
    },
    {
      "epoch": 0.6355837374114136,
      "grad_norm": 0.46247005462646484,
      "learning_rate": 1.8222885324713002e-06,
      "loss": 0.6838,
      "step": 15336
    },
    {
      "epoch": 0.6356251813170873,
      "grad_norm": 0.4104243814945221,
      "learning_rate": 1.8220813129429318e-06,
      "loss": 0.6658,
      "step": 15337
    },
    {
      "epoch": 0.635666625222761,
      "grad_norm": 0.44252797961235046,
      "learning_rate": 1.8218740934145634e-06,
      "loss": 0.7246,
      "step": 15338
    },
    {
      "epoch": 0.6357080691284347,
      "grad_norm": 0.3962167501449585,
      "learning_rate": 1.8216668738861952e-06,
      "loss": 0.7439,
      "step": 15339
    },
    {
      "epoch": 0.6357495130341083,
      "grad_norm": 0.41960608959198,
      "learning_rate": 1.821459654357827e-06,
      "loss": 0.7339,
      "step": 15340
    },
    {
      "epoch": 0.635790956939782,
      "grad_norm": 0.39101892709732056,
      "learning_rate": 1.8212524348294586e-06,
      "loss": 0.6436,
      "step": 15341
    },
    {
      "epoch": 0.6358324008454557,
      "grad_norm": 0.43188178539276123,
      "learning_rate": 1.8210452153010902e-06,
      "loss": 0.6812,
      "step": 15342
    },
    {
      "epoch": 0.6358738447511293,
      "grad_norm": 0.3930559754371643,
      "learning_rate": 1.8208379957727218e-06,
      "loss": 0.6914,
      "step": 15343
    },
    {
      "epoch": 0.635915288656803,
      "grad_norm": 0.3822891116142273,
      "learning_rate": 1.8206307762443534e-06,
      "loss": 0.67,
      "step": 15344
    },
    {
      "epoch": 0.6359567325624766,
      "grad_norm": 0.43734651803970337,
      "learning_rate": 1.820423556715985e-06,
      "loss": 0.6444,
      "step": 15345
    },
    {
      "epoch": 0.6359981764681504,
      "grad_norm": 0.40735191106796265,
      "learning_rate": 1.8202163371876166e-06,
      "loss": 0.6306,
      "step": 15346
    },
    {
      "epoch": 0.636039620373824,
      "grad_norm": 0.4107307493686676,
      "learning_rate": 1.8200091176592482e-06,
      "loss": 0.7025,
      "step": 15347
    },
    {
      "epoch": 0.6360810642794977,
      "grad_norm": 0.41425636410713196,
      "learning_rate": 1.81980189813088e-06,
      "loss": 0.6489,
      "step": 15348
    },
    {
      "epoch": 0.6361225081851714,
      "grad_norm": 0.4198528528213501,
      "learning_rate": 1.8195946786025118e-06,
      "loss": 0.6743,
      "step": 15349
    },
    {
      "epoch": 0.6361639520908451,
      "grad_norm": 0.39457428455352783,
      "learning_rate": 1.8193874590741434e-06,
      "loss": 0.6542,
      "step": 15350
    },
    {
      "epoch": 0.6362053959965187,
      "grad_norm": 0.41537606716156006,
      "learning_rate": 1.819180239545775e-06,
      "loss": 0.6692,
      "step": 15351
    },
    {
      "epoch": 0.6362468399021923,
      "grad_norm": 0.4152434468269348,
      "learning_rate": 1.8189730200174066e-06,
      "loss": 0.6741,
      "step": 15352
    },
    {
      "epoch": 0.6362882838078661,
      "grad_norm": 0.4022214114665985,
      "learning_rate": 1.8187658004890382e-06,
      "loss": 0.6478,
      "step": 15353
    },
    {
      "epoch": 0.6363297277135397,
      "grad_norm": 0.4001022279262543,
      "learning_rate": 1.8185585809606698e-06,
      "loss": 0.6461,
      "step": 15354
    },
    {
      "epoch": 0.6363711716192134,
      "grad_norm": 0.4324760138988495,
      "learning_rate": 1.8183513614323014e-06,
      "loss": 0.7422,
      "step": 15355
    },
    {
      "epoch": 0.636412615524887,
      "grad_norm": 0.4440074861049652,
      "learning_rate": 1.8181441419039332e-06,
      "loss": 0.7327,
      "step": 15356
    },
    {
      "epoch": 0.6364540594305608,
      "grad_norm": 0.3978829085826874,
      "learning_rate": 1.8179369223755648e-06,
      "loss": 0.6724,
      "step": 15357
    },
    {
      "epoch": 0.6364955033362344,
      "grad_norm": 0.41842445731163025,
      "learning_rate": 1.8177297028471966e-06,
      "loss": 0.6865,
      "step": 15358
    },
    {
      "epoch": 0.6365369472419081,
      "grad_norm": 0.4144447445869446,
      "learning_rate": 1.8175224833188282e-06,
      "loss": 0.7278,
      "step": 15359
    },
    {
      "epoch": 0.6365783911475817,
      "grad_norm": 0.47591978311538696,
      "learning_rate": 1.8173152637904598e-06,
      "loss": 0.6877,
      "step": 15360
    },
    {
      "epoch": 0.6366198350532554,
      "grad_norm": 0.41524603962898254,
      "learning_rate": 1.8171080442620914e-06,
      "loss": 0.712,
      "step": 15361
    },
    {
      "epoch": 0.6366612789589291,
      "grad_norm": 0.4098089635372162,
      "learning_rate": 1.816900824733723e-06,
      "loss": 0.7083,
      "step": 15362
    },
    {
      "epoch": 0.6367027228646027,
      "grad_norm": 0.43261098861694336,
      "learning_rate": 1.8166936052053546e-06,
      "loss": 0.6821,
      "step": 15363
    },
    {
      "epoch": 0.6367441667702765,
      "grad_norm": 0.407537579536438,
      "learning_rate": 1.8164863856769864e-06,
      "loss": 0.6654,
      "step": 15364
    },
    {
      "epoch": 0.6367856106759501,
      "grad_norm": 0.42200544476509094,
      "learning_rate": 1.816279166148618e-06,
      "loss": 0.661,
      "step": 15365
    },
    {
      "epoch": 0.6368270545816238,
      "grad_norm": 0.42285263538360596,
      "learning_rate": 1.8160719466202498e-06,
      "loss": 0.698,
      "step": 15366
    },
    {
      "epoch": 0.6368684984872974,
      "grad_norm": 0.42784354090690613,
      "learning_rate": 1.8158647270918814e-06,
      "loss": 0.642,
      "step": 15367
    },
    {
      "epoch": 0.6369099423929712,
      "grad_norm": 0.3939831852912903,
      "learning_rate": 1.815657507563513e-06,
      "loss": 0.6266,
      "step": 15368
    },
    {
      "epoch": 0.6369513862986448,
      "grad_norm": 0.43385496735572815,
      "learning_rate": 1.8154502880351446e-06,
      "loss": 0.725,
      "step": 15369
    },
    {
      "epoch": 0.6369928302043184,
      "grad_norm": 0.4425962269306183,
      "learning_rate": 1.8152430685067762e-06,
      "loss": 0.7201,
      "step": 15370
    },
    {
      "epoch": 0.6370342741099921,
      "grad_norm": 0.43750327825546265,
      "learning_rate": 1.8150358489784078e-06,
      "loss": 0.6599,
      "step": 15371
    },
    {
      "epoch": 0.6370757180156658,
      "grad_norm": 0.4344983696937561,
      "learning_rate": 1.8148286294500394e-06,
      "loss": 0.6779,
      "step": 15372
    },
    {
      "epoch": 0.6371171619213395,
      "grad_norm": 0.38473379611968994,
      "learning_rate": 1.8146214099216712e-06,
      "loss": 0.6439,
      "step": 15373
    },
    {
      "epoch": 0.6371586058270131,
      "grad_norm": 0.39684540033340454,
      "learning_rate": 1.8144141903933028e-06,
      "loss": 0.6893,
      "step": 15374
    },
    {
      "epoch": 0.6372000497326868,
      "grad_norm": 0.4194018542766571,
      "learning_rate": 1.8142069708649346e-06,
      "loss": 0.6987,
      "step": 15375
    },
    {
      "epoch": 0.6372414936383605,
      "grad_norm": 0.4072990119457245,
      "learning_rate": 1.8139997513365662e-06,
      "loss": 0.6483,
      "step": 15376
    },
    {
      "epoch": 0.6372829375440342,
      "grad_norm": 0.41205328702926636,
      "learning_rate": 1.8137925318081978e-06,
      "loss": 0.6992,
      "step": 15377
    },
    {
      "epoch": 0.6373243814497078,
      "grad_norm": 0.5244513154029846,
      "learning_rate": 1.8135853122798294e-06,
      "loss": 0.7211,
      "step": 15378
    },
    {
      "epoch": 0.6373658253553814,
      "grad_norm": 0.38881808519363403,
      "learning_rate": 1.813378092751461e-06,
      "loss": 0.6494,
      "step": 15379
    },
    {
      "epoch": 0.6374072692610552,
      "grad_norm": 0.42637771368026733,
      "learning_rate": 1.8131708732230926e-06,
      "loss": 0.6914,
      "step": 15380
    },
    {
      "epoch": 0.6374487131667288,
      "grad_norm": 0.4013627767562866,
      "learning_rate": 1.8129636536947244e-06,
      "loss": 0.7068,
      "step": 15381
    },
    {
      "epoch": 0.6374901570724025,
      "grad_norm": 0.41826945543289185,
      "learning_rate": 1.812756434166356e-06,
      "loss": 0.6996,
      "step": 15382
    },
    {
      "epoch": 0.6375316009780762,
      "grad_norm": 0.40788719058036804,
      "learning_rate": 1.8125492146379876e-06,
      "loss": 0.6304,
      "step": 15383
    },
    {
      "epoch": 0.6375730448837499,
      "grad_norm": 0.3821995258331299,
      "learning_rate": 1.8123419951096194e-06,
      "loss": 0.6422,
      "step": 15384
    },
    {
      "epoch": 0.6376144887894235,
      "grad_norm": 0.40059471130371094,
      "learning_rate": 1.812134775581251e-06,
      "loss": 0.6824,
      "step": 15385
    },
    {
      "epoch": 0.6376559326950972,
      "grad_norm": 0.37301573157310486,
      "learning_rate": 1.8119275560528826e-06,
      "loss": 0.6437,
      "step": 15386
    },
    {
      "epoch": 0.6376973766007709,
      "grad_norm": 0.4050006866455078,
      "learning_rate": 1.8117203365245142e-06,
      "loss": 0.7366,
      "step": 15387
    },
    {
      "epoch": 0.6377388205064445,
      "grad_norm": 0.40898439288139343,
      "learning_rate": 1.8115131169961458e-06,
      "loss": 0.691,
      "step": 15388
    },
    {
      "epoch": 0.6377802644121182,
      "grad_norm": 0.4083981513977051,
      "learning_rate": 1.8113058974677774e-06,
      "loss": 0.6412,
      "step": 15389
    },
    {
      "epoch": 0.6378217083177918,
      "grad_norm": 0.3893134891986847,
      "learning_rate": 1.8110986779394092e-06,
      "loss": 0.609,
      "step": 15390
    },
    {
      "epoch": 0.6378631522234656,
      "grad_norm": 0.5936991572380066,
      "learning_rate": 1.8108914584110408e-06,
      "loss": 0.7109,
      "step": 15391
    },
    {
      "epoch": 0.6379045961291392,
      "grad_norm": 0.42896243929862976,
      "learning_rate": 1.8106842388826726e-06,
      "loss": 0.71,
      "step": 15392
    },
    {
      "epoch": 0.6379460400348129,
      "grad_norm": 0.407651424407959,
      "learning_rate": 1.8104770193543042e-06,
      "loss": 0.6548,
      "step": 15393
    },
    {
      "epoch": 0.6379874839404865,
      "grad_norm": 0.4001621603965759,
      "learning_rate": 1.8102697998259358e-06,
      "loss": 0.6809,
      "step": 15394
    },
    {
      "epoch": 0.6380289278461603,
      "grad_norm": 0.4125313460826874,
      "learning_rate": 1.8100625802975674e-06,
      "loss": 0.6721,
      "step": 15395
    },
    {
      "epoch": 0.6380703717518339,
      "grad_norm": 0.3769037425518036,
      "learning_rate": 1.809855360769199e-06,
      "loss": 0.6353,
      "step": 15396
    },
    {
      "epoch": 0.6381118156575075,
      "grad_norm": 0.43375149369239807,
      "learning_rate": 1.8096481412408306e-06,
      "loss": 0.7324,
      "step": 15397
    },
    {
      "epoch": 0.6381532595631813,
      "grad_norm": 0.4265619218349457,
      "learning_rate": 1.8094409217124624e-06,
      "loss": 0.7422,
      "step": 15398
    },
    {
      "epoch": 0.6381947034688549,
      "grad_norm": 0.3876926898956299,
      "learning_rate": 1.809233702184094e-06,
      "loss": 0.6541,
      "step": 15399
    },
    {
      "epoch": 0.6382361473745286,
      "grad_norm": 0.43742167949676514,
      "learning_rate": 1.8090264826557256e-06,
      "loss": 0.7141,
      "step": 15400
    },
    {
      "epoch": 0.6382775912802022,
      "grad_norm": 0.46639204025268555,
      "learning_rate": 1.8088192631273574e-06,
      "loss": 0.6757,
      "step": 15401
    },
    {
      "epoch": 0.638319035185876,
      "grad_norm": 0.42468637228012085,
      "learning_rate": 1.808612043598989e-06,
      "loss": 0.641,
      "step": 15402
    },
    {
      "epoch": 0.6383604790915496,
      "grad_norm": 0.3939315974712372,
      "learning_rate": 1.8084048240706206e-06,
      "loss": 0.6604,
      "step": 15403
    },
    {
      "epoch": 0.6384019229972232,
      "grad_norm": 0.40822020173072815,
      "learning_rate": 1.8081976045422522e-06,
      "loss": 0.6709,
      "step": 15404
    },
    {
      "epoch": 0.6384433669028969,
      "grad_norm": 0.3999190032482147,
      "learning_rate": 1.8079903850138838e-06,
      "loss": 0.6987,
      "step": 15405
    },
    {
      "epoch": 0.6384848108085706,
      "grad_norm": 0.38102737069129944,
      "learning_rate": 1.8077831654855154e-06,
      "loss": 0.6504,
      "step": 15406
    },
    {
      "epoch": 0.6385262547142443,
      "grad_norm": 0.3887810707092285,
      "learning_rate": 1.8075759459571472e-06,
      "loss": 0.6445,
      "step": 15407
    },
    {
      "epoch": 0.6385676986199179,
      "grad_norm": 0.42529791593551636,
      "learning_rate": 1.8073687264287788e-06,
      "loss": 0.6335,
      "step": 15408
    },
    {
      "epoch": 0.6386091425255916,
      "grad_norm": 0.4241181015968323,
      "learning_rate": 1.8071615069004104e-06,
      "loss": 0.6785,
      "step": 15409
    },
    {
      "epoch": 0.6386505864312653,
      "grad_norm": 0.4130811095237732,
      "learning_rate": 1.8069542873720422e-06,
      "loss": 0.6819,
      "step": 15410
    },
    {
      "epoch": 0.638692030336939,
      "grad_norm": 0.4061991274356842,
      "learning_rate": 1.8067470678436738e-06,
      "loss": 0.6498,
      "step": 15411
    },
    {
      "epoch": 0.6387334742426126,
      "grad_norm": 0.42730075120925903,
      "learning_rate": 1.8065398483153054e-06,
      "loss": 0.6409,
      "step": 15412
    },
    {
      "epoch": 0.6387749181482862,
      "grad_norm": 0.403927743434906,
      "learning_rate": 1.806332628786937e-06,
      "loss": 0.65,
      "step": 15413
    },
    {
      "epoch": 0.63881636205396,
      "grad_norm": 0.40547341108322144,
      "learning_rate": 1.8061254092585686e-06,
      "loss": 0.6718,
      "step": 15414
    },
    {
      "epoch": 0.6388578059596336,
      "grad_norm": 0.4219434857368469,
      "learning_rate": 1.8059181897302004e-06,
      "loss": 0.6656,
      "step": 15415
    },
    {
      "epoch": 0.6388992498653073,
      "grad_norm": 0.4398444890975952,
      "learning_rate": 1.805710970201832e-06,
      "loss": 0.6421,
      "step": 15416
    },
    {
      "epoch": 0.638940693770981,
      "grad_norm": 0.38409423828125,
      "learning_rate": 1.8055037506734636e-06,
      "loss": 0.6638,
      "step": 15417
    },
    {
      "epoch": 0.6389821376766547,
      "grad_norm": 0.3883430063724518,
      "learning_rate": 1.8052965311450954e-06,
      "loss": 0.6196,
      "step": 15418
    },
    {
      "epoch": 0.6390235815823283,
      "grad_norm": 0.3925822675228119,
      "learning_rate": 1.805089311616727e-06,
      "loss": 0.6156,
      "step": 15419
    },
    {
      "epoch": 0.639065025488002,
      "grad_norm": 0.4166000783443451,
      "learning_rate": 1.8048820920883586e-06,
      "loss": 0.6487,
      "step": 15420
    },
    {
      "epoch": 0.6391064693936757,
      "grad_norm": 0.4378117024898529,
      "learning_rate": 1.8046748725599902e-06,
      "loss": 0.696,
      "step": 15421
    },
    {
      "epoch": 0.6391479132993493,
      "grad_norm": 0.43523991107940674,
      "learning_rate": 1.8044676530316218e-06,
      "loss": 0.6813,
      "step": 15422
    },
    {
      "epoch": 0.639189357205023,
      "grad_norm": 0.4110928773880005,
      "learning_rate": 1.8042604335032534e-06,
      "loss": 0.7312,
      "step": 15423
    },
    {
      "epoch": 0.6392308011106966,
      "grad_norm": 0.4211960732936859,
      "learning_rate": 1.8040532139748852e-06,
      "loss": 0.6719,
      "step": 15424
    },
    {
      "epoch": 0.6392722450163704,
      "grad_norm": 0.42485764622688293,
      "learning_rate": 1.8038459944465168e-06,
      "loss": 0.7656,
      "step": 15425
    },
    {
      "epoch": 0.639313688922044,
      "grad_norm": 0.4234153628349304,
      "learning_rate": 1.8036387749181484e-06,
      "loss": 0.6285,
      "step": 15426
    },
    {
      "epoch": 0.6393551328277177,
      "grad_norm": 0.40785476565361023,
      "learning_rate": 1.8034315553897802e-06,
      "loss": 0.6797,
      "step": 15427
    },
    {
      "epoch": 0.6393965767333913,
      "grad_norm": 0.42951107025146484,
      "learning_rate": 1.8032243358614118e-06,
      "loss": 0.7104,
      "step": 15428
    },
    {
      "epoch": 0.6394380206390651,
      "grad_norm": 0.4540995955467224,
      "learning_rate": 1.8030171163330434e-06,
      "loss": 0.6962,
      "step": 15429
    },
    {
      "epoch": 0.6394794645447387,
      "grad_norm": 0.4313414990901947,
      "learning_rate": 1.802809896804675e-06,
      "loss": 0.6952,
      "step": 15430
    },
    {
      "epoch": 0.6395209084504123,
      "grad_norm": 0.4329187572002411,
      "learning_rate": 1.8026026772763066e-06,
      "loss": 0.6963,
      "step": 15431
    },
    {
      "epoch": 0.639562352356086,
      "grad_norm": 0.4532982409000397,
      "learning_rate": 1.8023954577479384e-06,
      "loss": 0.652,
      "step": 15432
    },
    {
      "epoch": 0.6396037962617597,
      "grad_norm": 0.4197540581226349,
      "learning_rate": 1.80218823821957e-06,
      "loss": 0.7086,
      "step": 15433
    },
    {
      "epoch": 0.6396452401674334,
      "grad_norm": 0.4483310282230377,
      "learning_rate": 1.8019810186912016e-06,
      "loss": 0.7339,
      "step": 15434
    },
    {
      "epoch": 0.639686684073107,
      "grad_norm": 0.45394623279571533,
      "learning_rate": 1.8017737991628332e-06,
      "loss": 0.6919,
      "step": 15435
    },
    {
      "epoch": 0.6397281279787808,
      "grad_norm": 0.5343771576881409,
      "learning_rate": 1.801566579634465e-06,
      "loss": 0.6495,
      "step": 15436
    },
    {
      "epoch": 0.6397695718844544,
      "grad_norm": 0.3990635275840759,
      "learning_rate": 1.8013593601060966e-06,
      "loss": 0.6919,
      "step": 15437
    },
    {
      "epoch": 0.6398110157901281,
      "grad_norm": 0.43411561846733093,
      "learning_rate": 1.8011521405777282e-06,
      "loss": 0.6812,
      "step": 15438
    },
    {
      "epoch": 0.6398524596958017,
      "grad_norm": 0.3796842694282532,
      "learning_rate": 1.8009449210493598e-06,
      "loss": 0.6494,
      "step": 15439
    },
    {
      "epoch": 0.6398939036014754,
      "grad_norm": 0.42167574167251587,
      "learning_rate": 1.8007377015209914e-06,
      "loss": 0.7529,
      "step": 15440
    },
    {
      "epoch": 0.6399353475071491,
      "grad_norm": 0.43358561396598816,
      "learning_rate": 1.8005304819926232e-06,
      "loss": 0.7549,
      "step": 15441
    },
    {
      "epoch": 0.6399767914128227,
      "grad_norm": 0.38262295722961426,
      "learning_rate": 1.8003232624642548e-06,
      "loss": 0.6375,
      "step": 15442
    },
    {
      "epoch": 0.6400182353184964,
      "grad_norm": 0.4293767511844635,
      "learning_rate": 1.8001160429358864e-06,
      "loss": 0.6791,
      "step": 15443
    },
    {
      "epoch": 0.6400596792241701,
      "grad_norm": 0.3715209364891052,
      "learning_rate": 1.799908823407518e-06,
      "loss": 0.6565,
      "step": 15444
    },
    {
      "epoch": 0.6401011231298438,
      "grad_norm": 0.41066911816596985,
      "learning_rate": 1.7997016038791498e-06,
      "loss": 0.7035,
      "step": 15445
    },
    {
      "epoch": 0.6401425670355174,
      "grad_norm": 0.43244168162345886,
      "learning_rate": 1.7994943843507814e-06,
      "loss": 0.6906,
      "step": 15446
    },
    {
      "epoch": 0.6401840109411912,
      "grad_norm": 0.42778491973876953,
      "learning_rate": 1.799287164822413e-06,
      "loss": 0.6898,
      "step": 15447
    },
    {
      "epoch": 0.6402254548468648,
      "grad_norm": 0.3971628248691559,
      "learning_rate": 1.7990799452940446e-06,
      "loss": 0.6458,
      "step": 15448
    },
    {
      "epoch": 0.6402668987525384,
      "grad_norm": 0.4122277796268463,
      "learning_rate": 1.7988727257656764e-06,
      "loss": 0.6842,
      "step": 15449
    },
    {
      "epoch": 0.6403083426582121,
      "grad_norm": 0.4397014379501343,
      "learning_rate": 1.798665506237308e-06,
      "loss": 0.7244,
      "step": 15450
    },
    {
      "epoch": 0.6403497865638857,
      "grad_norm": 0.3966304659843445,
      "learning_rate": 1.7984582867089396e-06,
      "loss": 0.655,
      "step": 15451
    },
    {
      "epoch": 0.6403912304695595,
      "grad_norm": 0.4156041145324707,
      "learning_rate": 1.7982510671805712e-06,
      "loss": 0.6803,
      "step": 15452
    },
    {
      "epoch": 0.6404326743752331,
      "grad_norm": 0.40549612045288086,
      "learning_rate": 1.798043847652203e-06,
      "loss": 0.6841,
      "step": 15453
    },
    {
      "epoch": 0.6404741182809068,
      "grad_norm": 0.38895317912101746,
      "learning_rate": 1.7978366281238346e-06,
      "loss": 0.6455,
      "step": 15454
    },
    {
      "epoch": 0.6405155621865805,
      "grad_norm": 0.47054755687713623,
      "learning_rate": 1.7976294085954662e-06,
      "loss": 0.7312,
      "step": 15455
    },
    {
      "epoch": 0.6405570060922542,
      "grad_norm": 0.4385436773300171,
      "learning_rate": 1.7974221890670978e-06,
      "loss": 0.7212,
      "step": 15456
    },
    {
      "epoch": 0.6405984499979278,
      "grad_norm": 0.4112132489681244,
      "learning_rate": 1.7972149695387296e-06,
      "loss": 0.6431,
      "step": 15457
    },
    {
      "epoch": 0.6406398939036014,
      "grad_norm": 0.4475666582584381,
      "learning_rate": 1.7970077500103612e-06,
      "loss": 0.6406,
      "step": 15458
    },
    {
      "epoch": 0.6406813378092752,
      "grad_norm": 0.3766789734363556,
      "learning_rate": 1.7968005304819928e-06,
      "loss": 0.6399,
      "step": 15459
    },
    {
      "epoch": 0.6407227817149488,
      "grad_norm": 0.396084189414978,
      "learning_rate": 1.7965933109536244e-06,
      "loss": 0.6602,
      "step": 15460
    },
    {
      "epoch": 0.6407642256206225,
      "grad_norm": 0.40319427847862244,
      "learning_rate": 1.796386091425256e-06,
      "loss": 0.6887,
      "step": 15461
    },
    {
      "epoch": 0.6408056695262961,
      "grad_norm": 0.4469258189201355,
      "learning_rate": 1.7961788718968878e-06,
      "loss": 0.6804,
      "step": 15462
    },
    {
      "epoch": 0.6408471134319699,
      "grad_norm": 0.41355183720588684,
      "learning_rate": 1.7959716523685194e-06,
      "loss": 0.636,
      "step": 15463
    },
    {
      "epoch": 0.6408885573376435,
      "grad_norm": 0.38520270586013794,
      "learning_rate": 1.795764432840151e-06,
      "loss": 0.6719,
      "step": 15464
    },
    {
      "epoch": 0.6409300012433171,
      "grad_norm": 0.42115798592567444,
      "learning_rate": 1.7955572133117826e-06,
      "loss": 0.7009,
      "step": 15465
    },
    {
      "epoch": 0.6409714451489908,
      "grad_norm": 0.39818131923675537,
      "learning_rate": 1.7953499937834144e-06,
      "loss": 0.6853,
      "step": 15466
    },
    {
      "epoch": 0.6410128890546645,
      "grad_norm": 0.41175755858421326,
      "learning_rate": 1.795142774255046e-06,
      "loss": 0.7642,
      "step": 15467
    },
    {
      "epoch": 0.6410543329603382,
      "grad_norm": 0.41653695702552795,
      "learning_rate": 1.7949355547266776e-06,
      "loss": 0.7021,
      "step": 15468
    },
    {
      "epoch": 0.6410957768660118,
      "grad_norm": 0.41070571541786194,
      "learning_rate": 1.7947283351983092e-06,
      "loss": 0.6785,
      "step": 15469
    },
    {
      "epoch": 0.6411372207716856,
      "grad_norm": 0.41294312477111816,
      "learning_rate": 1.7945211156699408e-06,
      "loss": 0.6848,
      "step": 15470
    },
    {
      "epoch": 0.6411786646773592,
      "grad_norm": 0.4428417384624481,
      "learning_rate": 1.7943138961415726e-06,
      "loss": 0.6898,
      "step": 15471
    },
    {
      "epoch": 0.6412201085830329,
      "grad_norm": 0.3724454939365387,
      "learning_rate": 1.7941066766132042e-06,
      "loss": 0.6388,
      "step": 15472
    },
    {
      "epoch": 0.6412615524887065,
      "grad_norm": 0.40590357780456543,
      "learning_rate": 1.7938994570848358e-06,
      "loss": 0.687,
      "step": 15473
    },
    {
      "epoch": 0.6413029963943802,
      "grad_norm": 0.4373098909854889,
      "learning_rate": 1.7936922375564676e-06,
      "loss": 0.6765,
      "step": 15474
    },
    {
      "epoch": 0.6413444403000539,
      "grad_norm": 0.41632959246635437,
      "learning_rate": 1.7934850180280992e-06,
      "loss": 0.6895,
      "step": 15475
    },
    {
      "epoch": 0.6413858842057275,
      "grad_norm": 0.4230007827281952,
      "learning_rate": 1.7932777984997308e-06,
      "loss": 0.6812,
      "step": 15476
    },
    {
      "epoch": 0.6414273281114012,
      "grad_norm": 0.4226545989513397,
      "learning_rate": 1.7930705789713624e-06,
      "loss": 0.6305,
      "step": 15477
    },
    {
      "epoch": 0.6414687720170749,
      "grad_norm": 0.4460834264755249,
      "learning_rate": 1.792863359442994e-06,
      "loss": 0.7042,
      "step": 15478
    },
    {
      "epoch": 0.6415102159227486,
      "grad_norm": 0.43822285532951355,
      "learning_rate": 1.7926561399146258e-06,
      "loss": 0.7664,
      "step": 15479
    },
    {
      "epoch": 0.6415516598284222,
      "grad_norm": 0.402290016412735,
      "learning_rate": 1.7924489203862574e-06,
      "loss": 0.6897,
      "step": 15480
    },
    {
      "epoch": 0.641593103734096,
      "grad_norm": 0.4117305874824524,
      "learning_rate": 1.792241700857889e-06,
      "loss": 0.6429,
      "step": 15481
    },
    {
      "epoch": 0.6416345476397696,
      "grad_norm": 0.39917096495628357,
      "learning_rate": 1.7920344813295206e-06,
      "loss": 0.6766,
      "step": 15482
    },
    {
      "epoch": 0.6416759915454432,
      "grad_norm": 0.429167777299881,
      "learning_rate": 1.7918272618011524e-06,
      "loss": 0.6743,
      "step": 15483
    },
    {
      "epoch": 0.6417174354511169,
      "grad_norm": 0.3926008343696594,
      "learning_rate": 1.791620042272784e-06,
      "loss": 0.6526,
      "step": 15484
    },
    {
      "epoch": 0.6417588793567905,
      "grad_norm": 0.45266926288604736,
      "learning_rate": 1.7914128227444156e-06,
      "loss": 0.7312,
      "step": 15485
    },
    {
      "epoch": 0.6418003232624643,
      "grad_norm": 0.4257446825504303,
      "learning_rate": 1.7912056032160472e-06,
      "loss": 0.7351,
      "step": 15486
    },
    {
      "epoch": 0.6418417671681379,
      "grad_norm": 0.39170026779174805,
      "learning_rate": 1.7909983836876788e-06,
      "loss": 0.6102,
      "step": 15487
    },
    {
      "epoch": 0.6418832110738116,
      "grad_norm": 0.4572887420654297,
      "learning_rate": 1.7907911641593106e-06,
      "loss": 0.6777,
      "step": 15488
    },
    {
      "epoch": 0.6419246549794853,
      "grad_norm": 0.40008044242858887,
      "learning_rate": 1.7905839446309422e-06,
      "loss": 0.6733,
      "step": 15489
    },
    {
      "epoch": 0.641966098885159,
      "grad_norm": 0.40949276089668274,
      "learning_rate": 1.7903767251025738e-06,
      "loss": 0.675,
      "step": 15490
    },
    {
      "epoch": 0.6420075427908326,
      "grad_norm": 0.3931456208229065,
      "learning_rate": 1.7901695055742056e-06,
      "loss": 0.6401,
      "step": 15491
    },
    {
      "epoch": 0.6420489866965062,
      "grad_norm": 0.42496979236602783,
      "learning_rate": 1.7899622860458372e-06,
      "loss": 0.6729,
      "step": 15492
    },
    {
      "epoch": 0.64209043060218,
      "grad_norm": 0.4187045097351074,
      "learning_rate": 1.7897550665174688e-06,
      "loss": 0.6855,
      "step": 15493
    },
    {
      "epoch": 0.6421318745078536,
      "grad_norm": 0.3931962549686432,
      "learning_rate": 1.7895478469891004e-06,
      "loss": 0.7234,
      "step": 15494
    },
    {
      "epoch": 0.6421733184135273,
      "grad_norm": 0.49752750992774963,
      "learning_rate": 1.789340627460732e-06,
      "loss": 0.7021,
      "step": 15495
    },
    {
      "epoch": 0.6422147623192009,
      "grad_norm": 0.39239534735679626,
      "learning_rate": 1.7891334079323636e-06,
      "loss": 0.6768,
      "step": 15496
    },
    {
      "epoch": 0.6422562062248747,
      "grad_norm": 0.43353843688964844,
      "learning_rate": 1.7889261884039954e-06,
      "loss": 0.6987,
      "step": 15497
    },
    {
      "epoch": 0.6422976501305483,
      "grad_norm": 0.39212489128112793,
      "learning_rate": 1.788718968875627e-06,
      "loss": 0.6252,
      "step": 15498
    },
    {
      "epoch": 0.642339094036222,
      "grad_norm": 0.4245157837867737,
      "learning_rate": 1.7885117493472586e-06,
      "loss": 0.7163,
      "step": 15499
    },
    {
      "epoch": 0.6423805379418956,
      "grad_norm": 0.38970744609832764,
      "learning_rate": 1.7883045298188904e-06,
      "loss": 0.6636,
      "step": 15500
    },
    {
      "epoch": 0.6424219818475693,
      "grad_norm": 0.4229969084262848,
      "learning_rate": 1.788097310290522e-06,
      "loss": 0.7322,
      "step": 15501
    },
    {
      "epoch": 0.642463425753243,
      "grad_norm": 0.39066338539123535,
      "learning_rate": 1.7878900907621536e-06,
      "loss": 0.6418,
      "step": 15502
    },
    {
      "epoch": 0.6425048696589166,
      "grad_norm": 0.44052889943122864,
      "learning_rate": 1.7876828712337852e-06,
      "loss": 0.679,
      "step": 15503
    },
    {
      "epoch": 0.6425463135645904,
      "grad_norm": 0.39470165967941284,
      "learning_rate": 1.7874756517054168e-06,
      "loss": 0.6566,
      "step": 15504
    },
    {
      "epoch": 0.642587757470264,
      "grad_norm": 0.4400657117366791,
      "learning_rate": 1.7872684321770484e-06,
      "loss": 0.6807,
      "step": 15505
    },
    {
      "epoch": 0.6426292013759377,
      "grad_norm": 0.4967343211174011,
      "learning_rate": 1.7870612126486802e-06,
      "loss": 0.6709,
      "step": 15506
    },
    {
      "epoch": 0.6426706452816113,
      "grad_norm": 0.40817248821258545,
      "learning_rate": 1.7868539931203118e-06,
      "loss": 0.7083,
      "step": 15507
    },
    {
      "epoch": 0.6427120891872851,
      "grad_norm": 0.43578851222991943,
      "learning_rate": 1.7866467735919436e-06,
      "loss": 0.6965,
      "step": 15508
    },
    {
      "epoch": 0.6427535330929587,
      "grad_norm": 0.40248626470565796,
      "learning_rate": 1.7864395540635752e-06,
      "loss": 0.6531,
      "step": 15509
    },
    {
      "epoch": 0.6427949769986323,
      "grad_norm": 0.4236171245574951,
      "learning_rate": 1.7862323345352068e-06,
      "loss": 0.6663,
      "step": 15510
    },
    {
      "epoch": 0.642836420904306,
      "grad_norm": 0.42424914240837097,
      "learning_rate": 1.7860251150068384e-06,
      "loss": 0.6782,
      "step": 15511
    },
    {
      "epoch": 0.6428778648099797,
      "grad_norm": 0.4155040383338928,
      "learning_rate": 1.78581789547847e-06,
      "loss": 0.6709,
      "step": 15512
    },
    {
      "epoch": 0.6429193087156534,
      "grad_norm": 0.4125572144985199,
      "learning_rate": 1.7856106759501016e-06,
      "loss": 0.6631,
      "step": 15513
    },
    {
      "epoch": 0.642960752621327,
      "grad_norm": 0.40927591919898987,
      "learning_rate": 1.7854034564217334e-06,
      "loss": 0.6448,
      "step": 15514
    },
    {
      "epoch": 0.6430021965270007,
      "grad_norm": 0.39371320605278015,
      "learning_rate": 1.785196236893365e-06,
      "loss": 0.6768,
      "step": 15515
    },
    {
      "epoch": 0.6430436404326744,
      "grad_norm": 0.4258314073085785,
      "learning_rate": 1.7849890173649966e-06,
      "loss": 0.6561,
      "step": 15516
    },
    {
      "epoch": 0.6430850843383481,
      "grad_norm": 0.408025324344635,
      "learning_rate": 1.7847817978366284e-06,
      "loss": 0.6665,
      "step": 15517
    },
    {
      "epoch": 0.6431265282440217,
      "grad_norm": 0.4315021336078644,
      "learning_rate": 1.78457457830826e-06,
      "loss": 0.7023,
      "step": 15518
    },
    {
      "epoch": 0.6431679721496953,
      "grad_norm": 0.4089753329753876,
      "learning_rate": 1.7843673587798916e-06,
      "loss": 0.6844,
      "step": 15519
    },
    {
      "epoch": 0.6432094160553691,
      "grad_norm": 0.41673681139945984,
      "learning_rate": 1.7841601392515232e-06,
      "loss": 0.7124,
      "step": 15520
    },
    {
      "epoch": 0.6432508599610427,
      "grad_norm": 0.4039592444896698,
      "learning_rate": 1.7839529197231548e-06,
      "loss": 0.6387,
      "step": 15521
    },
    {
      "epoch": 0.6432923038667164,
      "grad_norm": 0.4000694453716278,
      "learning_rate": 1.7837457001947864e-06,
      "loss": 0.6621,
      "step": 15522
    },
    {
      "epoch": 0.64333374777239,
      "grad_norm": 0.44936466217041016,
      "learning_rate": 1.7835384806664182e-06,
      "loss": 0.7476,
      "step": 15523
    },
    {
      "epoch": 0.6433751916780638,
      "grad_norm": 0.41147127747535706,
      "learning_rate": 1.7833312611380498e-06,
      "loss": 0.6982,
      "step": 15524
    },
    {
      "epoch": 0.6434166355837374,
      "grad_norm": 0.41451770067214966,
      "learning_rate": 1.7831240416096816e-06,
      "loss": 0.6868,
      "step": 15525
    },
    {
      "epoch": 0.643458079489411,
      "grad_norm": 0.4154144823551178,
      "learning_rate": 1.7829168220813132e-06,
      "loss": 0.6482,
      "step": 15526
    },
    {
      "epoch": 0.6434995233950848,
      "grad_norm": 0.42121171951293945,
      "learning_rate": 1.7827096025529448e-06,
      "loss": 0.6877,
      "step": 15527
    },
    {
      "epoch": 0.6435409673007584,
      "grad_norm": 0.4155755639076233,
      "learning_rate": 1.7825023830245764e-06,
      "loss": 0.6541,
      "step": 15528
    },
    {
      "epoch": 0.6435824112064321,
      "grad_norm": 0.3846248388290405,
      "learning_rate": 1.782295163496208e-06,
      "loss": 0.6345,
      "step": 15529
    },
    {
      "epoch": 0.6436238551121057,
      "grad_norm": 0.4115012288093567,
      "learning_rate": 1.7820879439678396e-06,
      "loss": 0.7029,
      "step": 15530
    },
    {
      "epoch": 0.6436652990177795,
      "grad_norm": 0.40665191411972046,
      "learning_rate": 1.7818807244394712e-06,
      "loss": 0.6768,
      "step": 15531
    },
    {
      "epoch": 0.6437067429234531,
      "grad_norm": 0.3928937315940857,
      "learning_rate": 1.781673504911103e-06,
      "loss": 0.6427,
      "step": 15532
    },
    {
      "epoch": 0.6437481868291268,
      "grad_norm": 0.4138561487197876,
      "learning_rate": 1.7814662853827346e-06,
      "loss": 0.6768,
      "step": 15533
    },
    {
      "epoch": 0.6437896307348004,
      "grad_norm": 0.43224290013313293,
      "learning_rate": 1.7812590658543664e-06,
      "loss": 0.6418,
      "step": 15534
    },
    {
      "epoch": 0.6438310746404741,
      "grad_norm": 0.3962171673774719,
      "learning_rate": 1.781051846325998e-06,
      "loss": 0.6545,
      "step": 15535
    },
    {
      "epoch": 0.6438725185461478,
      "grad_norm": 0.41967320442199707,
      "learning_rate": 1.7808446267976296e-06,
      "loss": 0.6729,
      "step": 15536
    },
    {
      "epoch": 0.6439139624518214,
      "grad_norm": 0.39832425117492676,
      "learning_rate": 1.7806374072692612e-06,
      "loss": 0.6377,
      "step": 15537
    },
    {
      "epoch": 0.6439554063574952,
      "grad_norm": 0.37731462717056274,
      "learning_rate": 1.7804301877408928e-06,
      "loss": 0.6256,
      "step": 15538
    },
    {
      "epoch": 0.6439968502631688,
      "grad_norm": 0.4135761559009552,
      "learning_rate": 1.7802229682125244e-06,
      "loss": 0.6572,
      "step": 15539
    },
    {
      "epoch": 0.6440382941688425,
      "grad_norm": 0.3860820531845093,
      "learning_rate": 1.7800157486841562e-06,
      "loss": 0.6365,
      "step": 15540
    },
    {
      "epoch": 0.6440797380745161,
      "grad_norm": 0.44271138310432434,
      "learning_rate": 1.7798085291557878e-06,
      "loss": 0.7278,
      "step": 15541
    },
    {
      "epoch": 0.6441211819801899,
      "grad_norm": 0.4696919322013855,
      "learning_rate": 1.7796013096274196e-06,
      "loss": 0.6932,
      "step": 15542
    },
    {
      "epoch": 0.6441626258858635,
      "grad_norm": 0.384356826543808,
      "learning_rate": 1.7793940900990512e-06,
      "loss": 0.6958,
      "step": 15543
    },
    {
      "epoch": 0.6442040697915371,
      "grad_norm": 0.3896111845970154,
      "learning_rate": 1.7791868705706828e-06,
      "loss": 0.6805,
      "step": 15544
    },
    {
      "epoch": 0.6442455136972108,
      "grad_norm": 0.4447305500507355,
      "learning_rate": 1.7789796510423144e-06,
      "loss": 0.7117,
      "step": 15545
    },
    {
      "epoch": 0.6442869576028845,
      "grad_norm": 0.4154475927352905,
      "learning_rate": 1.778772431513946e-06,
      "loss": 0.6862,
      "step": 15546
    },
    {
      "epoch": 0.6443284015085582,
      "grad_norm": 0.4060879349708557,
      "learning_rate": 1.7785652119855776e-06,
      "loss": 0.679,
      "step": 15547
    },
    {
      "epoch": 0.6443698454142318,
      "grad_norm": 0.45053189992904663,
      "learning_rate": 1.7783579924572092e-06,
      "loss": 0.7358,
      "step": 15548
    },
    {
      "epoch": 0.6444112893199055,
      "grad_norm": 0.4175051748752594,
      "learning_rate": 1.778150772928841e-06,
      "loss": 0.6824,
      "step": 15549
    },
    {
      "epoch": 0.6444527332255792,
      "grad_norm": 0.3870816230773926,
      "learning_rate": 1.7779435534004726e-06,
      "loss": 0.6592,
      "step": 15550
    },
    {
      "epoch": 0.6444941771312529,
      "grad_norm": 0.41827765107154846,
      "learning_rate": 1.7777363338721044e-06,
      "loss": 0.6664,
      "step": 15551
    },
    {
      "epoch": 0.6445356210369265,
      "grad_norm": 0.39840561151504517,
      "learning_rate": 1.777529114343736e-06,
      "loss": 0.6875,
      "step": 15552
    },
    {
      "epoch": 0.6445770649426001,
      "grad_norm": 0.4228484332561493,
      "learning_rate": 1.7773218948153676e-06,
      "loss": 0.705,
      "step": 15553
    },
    {
      "epoch": 0.6446185088482739,
      "grad_norm": 0.38958340883255005,
      "learning_rate": 1.7771146752869992e-06,
      "loss": 0.5992,
      "step": 15554
    },
    {
      "epoch": 0.6446599527539475,
      "grad_norm": 0.45174264907836914,
      "learning_rate": 1.7769074557586308e-06,
      "loss": 0.7788,
      "step": 15555
    },
    {
      "epoch": 0.6447013966596212,
      "grad_norm": 0.40604180097579956,
      "learning_rate": 1.7767002362302624e-06,
      "loss": 0.6975,
      "step": 15556
    },
    {
      "epoch": 0.6447428405652949,
      "grad_norm": 0.44124436378479004,
      "learning_rate": 1.776493016701894e-06,
      "loss": 0.7383,
      "step": 15557
    },
    {
      "epoch": 0.6447842844709686,
      "grad_norm": 0.40224844217300415,
      "learning_rate": 1.7762857971735258e-06,
      "loss": 0.6538,
      "step": 15558
    },
    {
      "epoch": 0.6448257283766422,
      "grad_norm": 0.4187207818031311,
      "learning_rate": 1.7760785776451576e-06,
      "loss": 0.701,
      "step": 15559
    },
    {
      "epoch": 0.6448671722823159,
      "grad_norm": 0.4450726807117462,
      "learning_rate": 1.7758713581167892e-06,
      "loss": 0.7512,
      "step": 15560
    },
    {
      "epoch": 0.6449086161879896,
      "grad_norm": 0.43539997935295105,
      "learning_rate": 1.7756641385884208e-06,
      "loss": 0.7194,
      "step": 15561
    },
    {
      "epoch": 0.6449500600936632,
      "grad_norm": 0.41573211550712585,
      "learning_rate": 1.7754569190600524e-06,
      "loss": 0.6765,
      "step": 15562
    },
    {
      "epoch": 0.6449915039993369,
      "grad_norm": 0.38446757197380066,
      "learning_rate": 1.775249699531684e-06,
      "loss": 0.6702,
      "step": 15563
    },
    {
      "epoch": 0.6450329479050105,
      "grad_norm": 0.423725962638855,
      "learning_rate": 1.7750424800033156e-06,
      "loss": 0.6636,
      "step": 15564
    },
    {
      "epoch": 0.6450743918106843,
      "grad_norm": 0.40703558921813965,
      "learning_rate": 1.7748352604749472e-06,
      "loss": 0.7012,
      "step": 15565
    },
    {
      "epoch": 0.6451158357163579,
      "grad_norm": 0.40120285749435425,
      "learning_rate": 1.774628040946579e-06,
      "loss": 0.6537,
      "step": 15566
    },
    {
      "epoch": 0.6451572796220316,
      "grad_norm": 0.3993944525718689,
      "learning_rate": 1.7744208214182108e-06,
      "loss": 0.6748,
      "step": 15567
    },
    {
      "epoch": 0.6451987235277052,
      "grad_norm": 0.4148927628993988,
      "learning_rate": 1.7742136018898424e-06,
      "loss": 0.6919,
      "step": 15568
    },
    {
      "epoch": 0.645240167433379,
      "grad_norm": 0.42864009737968445,
      "learning_rate": 1.774006382361474e-06,
      "loss": 0.6876,
      "step": 15569
    },
    {
      "epoch": 0.6452816113390526,
      "grad_norm": 0.4156241714954376,
      "learning_rate": 1.7737991628331056e-06,
      "loss": 0.6442,
      "step": 15570
    },
    {
      "epoch": 0.6453230552447262,
      "grad_norm": 0.4176127314567566,
      "learning_rate": 1.7735919433047372e-06,
      "loss": 0.705,
      "step": 15571
    },
    {
      "epoch": 0.6453644991504,
      "grad_norm": 0.4039718210697174,
      "learning_rate": 1.7733847237763688e-06,
      "loss": 0.6605,
      "step": 15572
    },
    {
      "epoch": 0.6454059430560736,
      "grad_norm": 0.40566107630729675,
      "learning_rate": 1.7731775042480004e-06,
      "loss": 0.6827,
      "step": 15573
    },
    {
      "epoch": 0.6454473869617473,
      "grad_norm": 0.39986708760261536,
      "learning_rate": 1.772970284719632e-06,
      "loss": 0.6602,
      "step": 15574
    },
    {
      "epoch": 0.6454888308674209,
      "grad_norm": 0.4003998637199402,
      "learning_rate": 1.7727630651912638e-06,
      "loss": 0.6602,
      "step": 15575
    },
    {
      "epoch": 0.6455302747730947,
      "grad_norm": 0.4174047112464905,
      "learning_rate": 1.7725558456628956e-06,
      "loss": 0.6719,
      "step": 15576
    },
    {
      "epoch": 0.6455717186787683,
      "grad_norm": 0.4078565239906311,
      "learning_rate": 1.7723486261345272e-06,
      "loss": 0.6418,
      "step": 15577
    },
    {
      "epoch": 0.645613162584442,
      "grad_norm": 0.4062003195285797,
      "learning_rate": 1.7721414066061588e-06,
      "loss": 0.7168,
      "step": 15578
    },
    {
      "epoch": 0.6456546064901156,
      "grad_norm": 0.4140835106372833,
      "learning_rate": 1.7719341870777904e-06,
      "loss": 0.6842,
      "step": 15579
    },
    {
      "epoch": 0.6456960503957893,
      "grad_norm": 0.4167834520339966,
      "learning_rate": 1.771726967549422e-06,
      "loss": 0.646,
      "step": 15580
    },
    {
      "epoch": 0.645737494301463,
      "grad_norm": 0.4254632294178009,
      "learning_rate": 1.7715197480210536e-06,
      "loss": 0.6473,
      "step": 15581
    },
    {
      "epoch": 0.6457789382071366,
      "grad_norm": 0.4167327284812927,
      "learning_rate": 1.7713125284926852e-06,
      "loss": 0.6802,
      "step": 15582
    },
    {
      "epoch": 0.6458203821128103,
      "grad_norm": 0.4255084991455078,
      "learning_rate": 1.7711053089643168e-06,
      "loss": 0.7144,
      "step": 15583
    },
    {
      "epoch": 0.645861826018484,
      "grad_norm": 0.3984299600124359,
      "learning_rate": 1.7708980894359488e-06,
      "loss": 0.6471,
      "step": 15584
    },
    {
      "epoch": 0.6459032699241577,
      "grad_norm": 0.40233105421066284,
      "learning_rate": 1.7706908699075804e-06,
      "loss": 0.6642,
      "step": 15585
    },
    {
      "epoch": 0.6459447138298313,
      "grad_norm": 0.42302951216697693,
      "learning_rate": 1.770483650379212e-06,
      "loss": 0.7412,
      "step": 15586
    },
    {
      "epoch": 0.6459861577355049,
      "grad_norm": 0.3434576094150543,
      "learning_rate": 1.7702764308508436e-06,
      "loss": 0.609,
      "step": 15587
    },
    {
      "epoch": 0.6460276016411787,
      "grad_norm": 0.4061698317527771,
      "learning_rate": 1.7700692113224752e-06,
      "loss": 0.6542,
      "step": 15588
    },
    {
      "epoch": 0.6460690455468523,
      "grad_norm": 0.4212305247783661,
      "learning_rate": 1.7698619917941068e-06,
      "loss": 0.639,
      "step": 15589
    },
    {
      "epoch": 0.646110489452526,
      "grad_norm": 0.42416831851005554,
      "learning_rate": 1.7696547722657384e-06,
      "loss": 0.6503,
      "step": 15590
    },
    {
      "epoch": 0.6461519333581996,
      "grad_norm": 0.41662371158599854,
      "learning_rate": 1.76944755273737e-06,
      "loss": 0.6392,
      "step": 15591
    },
    {
      "epoch": 0.6461933772638734,
      "grad_norm": 0.395205557346344,
      "learning_rate": 1.7692403332090016e-06,
      "loss": 0.601,
      "step": 15592
    },
    {
      "epoch": 0.646234821169547,
      "grad_norm": 0.433176726102829,
      "learning_rate": 1.7690331136806336e-06,
      "loss": 0.7008,
      "step": 15593
    },
    {
      "epoch": 0.6462762650752207,
      "grad_norm": 0.4153425693511963,
      "learning_rate": 1.7688258941522652e-06,
      "loss": 0.707,
      "step": 15594
    },
    {
      "epoch": 0.6463177089808944,
      "grad_norm": 0.42204827070236206,
      "learning_rate": 1.7686186746238968e-06,
      "loss": 0.7263,
      "step": 15595
    },
    {
      "epoch": 0.646359152886568,
      "grad_norm": 0.39451178908348083,
      "learning_rate": 1.7684114550955284e-06,
      "loss": 0.6919,
      "step": 15596
    },
    {
      "epoch": 0.6464005967922417,
      "grad_norm": 0.41760891675949097,
      "learning_rate": 1.76820423556716e-06,
      "loss": 0.6826,
      "step": 15597
    },
    {
      "epoch": 0.6464420406979153,
      "grad_norm": 0.4631892740726471,
      "learning_rate": 1.7679970160387916e-06,
      "loss": 0.6758,
      "step": 15598
    },
    {
      "epoch": 0.6464834846035891,
      "grad_norm": 0.44400617480278015,
      "learning_rate": 1.7677897965104232e-06,
      "loss": 0.7285,
      "step": 15599
    },
    {
      "epoch": 0.6465249285092627,
      "grad_norm": 0.4115111827850342,
      "learning_rate": 1.7675825769820548e-06,
      "loss": 0.6974,
      "step": 15600
    },
    {
      "epoch": 0.6465663724149364,
      "grad_norm": 0.4206889867782593,
      "learning_rate": 1.7673753574536868e-06,
      "loss": 0.7229,
      "step": 15601
    },
    {
      "epoch": 0.64660781632061,
      "grad_norm": 0.4195168912410736,
      "learning_rate": 1.7671681379253184e-06,
      "loss": 0.6982,
      "step": 15602
    },
    {
      "epoch": 0.6466492602262838,
      "grad_norm": 0.4469691216945648,
      "learning_rate": 1.76696091839695e-06,
      "loss": 0.6855,
      "step": 15603
    },
    {
      "epoch": 0.6466907041319574,
      "grad_norm": 0.42844244837760925,
      "learning_rate": 1.7667536988685816e-06,
      "loss": 0.6981,
      "step": 15604
    },
    {
      "epoch": 0.646732148037631,
      "grad_norm": 0.40317049622535706,
      "learning_rate": 1.7665464793402132e-06,
      "loss": 0.6484,
      "step": 15605
    },
    {
      "epoch": 0.6467735919433047,
      "grad_norm": 0.3886716067790985,
      "learning_rate": 1.7663392598118448e-06,
      "loss": 0.6538,
      "step": 15606
    },
    {
      "epoch": 0.6468150358489784,
      "grad_norm": 0.41700631380081177,
      "learning_rate": 1.7661320402834764e-06,
      "loss": 0.7236,
      "step": 15607
    },
    {
      "epoch": 0.6468564797546521,
      "grad_norm": 0.4060257077217102,
      "learning_rate": 1.765924820755108e-06,
      "loss": 0.7065,
      "step": 15608
    },
    {
      "epoch": 0.6468979236603257,
      "grad_norm": 0.45118248462677,
      "learning_rate": 1.7657176012267396e-06,
      "loss": 0.6687,
      "step": 15609
    },
    {
      "epoch": 0.6469393675659995,
      "grad_norm": 0.39940473437309265,
      "learning_rate": 1.7655103816983716e-06,
      "loss": 0.6024,
      "step": 15610
    },
    {
      "epoch": 0.6469808114716731,
      "grad_norm": 0.4820694327354431,
      "learning_rate": 1.7653031621700032e-06,
      "loss": 0.7668,
      "step": 15611
    },
    {
      "epoch": 0.6470222553773468,
      "grad_norm": 0.41874057054519653,
      "learning_rate": 1.7650959426416348e-06,
      "loss": 0.7288,
      "step": 15612
    },
    {
      "epoch": 0.6470636992830204,
      "grad_norm": 0.42237457633018494,
      "learning_rate": 1.7648887231132664e-06,
      "loss": 0.6687,
      "step": 15613
    },
    {
      "epoch": 0.647105143188694,
      "grad_norm": 0.38758614659309387,
      "learning_rate": 1.764681503584898e-06,
      "loss": 0.6298,
      "step": 15614
    },
    {
      "epoch": 0.6471465870943678,
      "grad_norm": 0.4139144718647003,
      "learning_rate": 1.7644742840565296e-06,
      "loss": 0.675,
      "step": 15615
    },
    {
      "epoch": 0.6471880310000414,
      "grad_norm": 0.44925493001937866,
      "learning_rate": 1.7642670645281612e-06,
      "loss": 0.7003,
      "step": 15616
    },
    {
      "epoch": 0.6472294749057151,
      "grad_norm": 0.4387260377407074,
      "learning_rate": 1.7640598449997928e-06,
      "loss": 0.679,
      "step": 15617
    },
    {
      "epoch": 0.6472709188113888,
      "grad_norm": 0.418400377035141,
      "learning_rate": 1.7638526254714244e-06,
      "loss": 0.6614,
      "step": 15618
    },
    {
      "epoch": 0.6473123627170625,
      "grad_norm": 0.411334753036499,
      "learning_rate": 1.7636454059430564e-06,
      "loss": 0.6787,
      "step": 15619
    },
    {
      "epoch": 0.6473538066227361,
      "grad_norm": 0.4391150176525116,
      "learning_rate": 1.763438186414688e-06,
      "loss": 0.7002,
      "step": 15620
    },
    {
      "epoch": 0.6473952505284098,
      "grad_norm": 0.42732009291648865,
      "learning_rate": 1.7632309668863196e-06,
      "loss": 0.6635,
      "step": 15621
    },
    {
      "epoch": 0.6474366944340835,
      "grad_norm": 0.4103679060935974,
      "learning_rate": 1.7630237473579512e-06,
      "loss": 0.668,
      "step": 15622
    },
    {
      "epoch": 0.6474781383397571,
      "grad_norm": 0.4332858622074127,
      "learning_rate": 1.7628165278295828e-06,
      "loss": 0.7334,
      "step": 15623
    },
    {
      "epoch": 0.6475195822454308,
      "grad_norm": 0.3918190896511078,
      "learning_rate": 1.7626093083012144e-06,
      "loss": 0.7061,
      "step": 15624
    },
    {
      "epoch": 0.6475610261511044,
      "grad_norm": 0.4244137108325958,
      "learning_rate": 1.762402088772846e-06,
      "loss": 0.6875,
      "step": 15625
    },
    {
      "epoch": 0.6476024700567782,
      "grad_norm": 0.4220355153083801,
      "learning_rate": 1.7621948692444776e-06,
      "loss": 0.7102,
      "step": 15626
    },
    {
      "epoch": 0.6476439139624518,
      "grad_norm": 0.4369968771934509,
      "learning_rate": 1.7619876497161096e-06,
      "loss": 0.6819,
      "step": 15627
    },
    {
      "epoch": 0.6476853578681255,
      "grad_norm": 0.4172671139240265,
      "learning_rate": 1.7617804301877412e-06,
      "loss": 0.7235,
      "step": 15628
    },
    {
      "epoch": 0.6477268017737992,
      "grad_norm": 0.42183375358581543,
      "learning_rate": 1.7615732106593728e-06,
      "loss": 0.6409,
      "step": 15629
    },
    {
      "epoch": 0.6477682456794729,
      "grad_norm": 0.4293377697467804,
      "learning_rate": 1.7613659911310044e-06,
      "loss": 0.6443,
      "step": 15630
    },
    {
      "epoch": 0.6478096895851465,
      "grad_norm": 0.4015118479728699,
      "learning_rate": 1.761158771602636e-06,
      "loss": 0.6604,
      "step": 15631
    },
    {
      "epoch": 0.6478511334908201,
      "grad_norm": 0.40238243341445923,
      "learning_rate": 1.7609515520742676e-06,
      "loss": 0.64,
      "step": 15632
    },
    {
      "epoch": 0.6478925773964939,
      "grad_norm": 0.4070974290370941,
      "learning_rate": 1.7607443325458992e-06,
      "loss": 0.6384,
      "step": 15633
    },
    {
      "epoch": 0.6479340213021675,
      "grad_norm": 0.4365856945514679,
      "learning_rate": 1.7605371130175308e-06,
      "loss": 0.7476,
      "step": 15634
    },
    {
      "epoch": 0.6479754652078412,
      "grad_norm": 0.44453659653663635,
      "learning_rate": 1.7603298934891624e-06,
      "loss": 0.6794,
      "step": 15635
    },
    {
      "epoch": 0.6480169091135148,
      "grad_norm": 0.4177300035953522,
      "learning_rate": 1.7601226739607944e-06,
      "loss": 0.7341,
      "step": 15636
    },
    {
      "epoch": 0.6480583530191886,
      "grad_norm": 0.4283050000667572,
      "learning_rate": 1.759915454432426e-06,
      "loss": 0.7031,
      "step": 15637
    },
    {
      "epoch": 0.6480997969248622,
      "grad_norm": 0.4999232590198517,
      "learning_rate": 1.7597082349040576e-06,
      "loss": 0.6847,
      "step": 15638
    },
    {
      "epoch": 0.6481412408305359,
      "grad_norm": 0.4562135636806488,
      "learning_rate": 1.7595010153756892e-06,
      "loss": 0.7196,
      "step": 15639
    },
    {
      "epoch": 0.6481826847362095,
      "grad_norm": 0.4090512692928314,
      "learning_rate": 1.7592937958473208e-06,
      "loss": 0.6272,
      "step": 15640
    },
    {
      "epoch": 0.6482241286418832,
      "grad_norm": 0.405238538980484,
      "learning_rate": 1.7590865763189524e-06,
      "loss": 0.7229,
      "step": 15641
    },
    {
      "epoch": 0.6482655725475569,
      "grad_norm": 0.3943670988082886,
      "learning_rate": 1.758879356790584e-06,
      "loss": 0.6609,
      "step": 15642
    },
    {
      "epoch": 0.6483070164532305,
      "grad_norm": 0.4016822278499603,
      "learning_rate": 1.7586721372622156e-06,
      "loss": 0.6753,
      "step": 15643
    },
    {
      "epoch": 0.6483484603589043,
      "grad_norm": 0.39707261323928833,
      "learning_rate": 1.7584649177338472e-06,
      "loss": 0.7124,
      "step": 15644
    },
    {
      "epoch": 0.6483899042645779,
      "grad_norm": 0.4021337926387787,
      "learning_rate": 1.7582576982054792e-06,
      "loss": 0.6509,
      "step": 15645
    },
    {
      "epoch": 0.6484313481702516,
      "grad_norm": 0.4227968156337738,
      "learning_rate": 1.7580504786771108e-06,
      "loss": 0.6411,
      "step": 15646
    },
    {
      "epoch": 0.6484727920759252,
      "grad_norm": 0.42698341608047485,
      "learning_rate": 1.7578432591487424e-06,
      "loss": 0.6746,
      "step": 15647
    },
    {
      "epoch": 0.6485142359815989,
      "grad_norm": 0.42486900091171265,
      "learning_rate": 1.757636039620374e-06,
      "loss": 0.6552,
      "step": 15648
    },
    {
      "epoch": 0.6485556798872726,
      "grad_norm": 0.4552864730358124,
      "learning_rate": 1.7574288200920056e-06,
      "loss": 0.691,
      "step": 15649
    },
    {
      "epoch": 0.6485971237929462,
      "grad_norm": 0.4610636532306671,
      "learning_rate": 1.7572216005636372e-06,
      "loss": 0.731,
      "step": 15650
    },
    {
      "epoch": 0.6486385676986199,
      "grad_norm": 0.4052690863609314,
      "learning_rate": 1.7570143810352688e-06,
      "loss": 0.6772,
      "step": 15651
    },
    {
      "epoch": 0.6486800116042936,
      "grad_norm": 0.4564581513404846,
      "learning_rate": 1.7568071615069004e-06,
      "loss": 0.6815,
      "step": 15652
    },
    {
      "epoch": 0.6487214555099673,
      "grad_norm": 0.41304343938827515,
      "learning_rate": 1.756599941978532e-06,
      "loss": 0.7079,
      "step": 15653
    },
    {
      "epoch": 0.6487628994156409,
      "grad_norm": 0.42861250042915344,
      "learning_rate": 1.756392722450164e-06,
      "loss": 0.6873,
      "step": 15654
    },
    {
      "epoch": 0.6488043433213146,
      "grad_norm": 0.41023650765419006,
      "learning_rate": 1.7561855029217956e-06,
      "loss": 0.6597,
      "step": 15655
    },
    {
      "epoch": 0.6488457872269883,
      "grad_norm": 0.38327041268348694,
      "learning_rate": 1.7559782833934272e-06,
      "loss": 0.6863,
      "step": 15656
    },
    {
      "epoch": 0.6488872311326619,
      "grad_norm": 0.3939375579357147,
      "learning_rate": 1.7557710638650588e-06,
      "loss": 0.6462,
      "step": 15657
    },
    {
      "epoch": 0.6489286750383356,
      "grad_norm": 0.41441771388053894,
      "learning_rate": 1.7555638443366904e-06,
      "loss": 0.67,
      "step": 15658
    },
    {
      "epoch": 0.6489701189440092,
      "grad_norm": 0.5000901818275452,
      "learning_rate": 1.755356624808322e-06,
      "loss": 0.7087,
      "step": 15659
    },
    {
      "epoch": 0.649011562849683,
      "grad_norm": 0.4338549077510834,
      "learning_rate": 1.7551494052799536e-06,
      "loss": 0.665,
      "step": 15660
    },
    {
      "epoch": 0.6490530067553566,
      "grad_norm": 0.4083094000816345,
      "learning_rate": 1.7549421857515852e-06,
      "loss": 0.7249,
      "step": 15661
    },
    {
      "epoch": 0.6490944506610303,
      "grad_norm": 0.442543625831604,
      "learning_rate": 1.7547349662232172e-06,
      "loss": 0.6471,
      "step": 15662
    },
    {
      "epoch": 0.649135894566704,
      "grad_norm": 0.39648306369781494,
      "learning_rate": 1.7545277466948488e-06,
      "loss": 0.7051,
      "step": 15663
    },
    {
      "epoch": 0.6491773384723777,
      "grad_norm": 0.41213417053222656,
      "learning_rate": 1.7543205271664804e-06,
      "loss": 0.6473,
      "step": 15664
    },
    {
      "epoch": 0.6492187823780513,
      "grad_norm": 0.3966955840587616,
      "learning_rate": 1.754113307638112e-06,
      "loss": 0.666,
      "step": 15665
    },
    {
      "epoch": 0.6492602262837249,
      "grad_norm": 0.41007545590400696,
      "learning_rate": 1.7539060881097436e-06,
      "loss": 0.6792,
      "step": 15666
    },
    {
      "epoch": 0.6493016701893987,
      "grad_norm": 0.39393675327301025,
      "learning_rate": 1.7536988685813752e-06,
      "loss": 0.6428,
      "step": 15667
    },
    {
      "epoch": 0.6493431140950723,
      "grad_norm": 0.3899173140525818,
      "learning_rate": 1.7534916490530068e-06,
      "loss": 0.6326,
      "step": 15668
    },
    {
      "epoch": 0.649384558000746,
      "grad_norm": 0.4753111004829407,
      "learning_rate": 1.7532844295246384e-06,
      "loss": 0.7156,
      "step": 15669
    },
    {
      "epoch": 0.6494260019064196,
      "grad_norm": 0.4707523286342621,
      "learning_rate": 1.75307720999627e-06,
      "loss": 0.7354,
      "step": 15670
    },
    {
      "epoch": 0.6494674458120934,
      "grad_norm": 0.38695940375328064,
      "learning_rate": 1.752869990467902e-06,
      "loss": 0.7065,
      "step": 15671
    },
    {
      "epoch": 0.649508889717767,
      "grad_norm": 0.42865437269210815,
      "learning_rate": 1.7526627709395336e-06,
      "loss": 0.6553,
      "step": 15672
    },
    {
      "epoch": 0.6495503336234407,
      "grad_norm": 0.41782253980636597,
      "learning_rate": 1.7524555514111652e-06,
      "loss": 0.6182,
      "step": 15673
    },
    {
      "epoch": 0.6495917775291143,
      "grad_norm": 0.40565982460975647,
      "learning_rate": 1.7522483318827968e-06,
      "loss": 0.6554,
      "step": 15674
    },
    {
      "epoch": 0.649633221434788,
      "grad_norm": 0.4111839234828949,
      "learning_rate": 1.7520411123544284e-06,
      "loss": 0.6344,
      "step": 15675
    },
    {
      "epoch": 0.6496746653404617,
      "grad_norm": 0.4392800033092499,
      "learning_rate": 1.75183389282606e-06,
      "loss": 0.729,
      "step": 15676
    },
    {
      "epoch": 0.6497161092461353,
      "grad_norm": 0.3998664319515228,
      "learning_rate": 1.7516266732976916e-06,
      "loss": 0.7295,
      "step": 15677
    },
    {
      "epoch": 0.649757553151809,
      "grad_norm": 0.40909072756767273,
      "learning_rate": 1.7514194537693232e-06,
      "loss": 0.6852,
      "step": 15678
    },
    {
      "epoch": 0.6497989970574827,
      "grad_norm": 0.38243603706359863,
      "learning_rate": 1.7512122342409548e-06,
      "loss": 0.6797,
      "step": 15679
    },
    {
      "epoch": 0.6498404409631564,
      "grad_norm": 0.5085466504096985,
      "learning_rate": 1.7510050147125868e-06,
      "loss": 0.6902,
      "step": 15680
    },
    {
      "epoch": 0.64988188486883,
      "grad_norm": 0.45818427205085754,
      "learning_rate": 1.7507977951842184e-06,
      "loss": 0.6636,
      "step": 15681
    },
    {
      "epoch": 0.6499233287745038,
      "grad_norm": 0.4107896685600281,
      "learning_rate": 1.75059057565585e-06,
      "loss": 0.6848,
      "step": 15682
    },
    {
      "epoch": 0.6499647726801774,
      "grad_norm": 0.42201653122901917,
      "learning_rate": 1.7503833561274816e-06,
      "loss": 0.7029,
      "step": 15683
    },
    {
      "epoch": 0.650006216585851,
      "grad_norm": 0.38527920842170715,
      "learning_rate": 1.7501761365991132e-06,
      "loss": 0.6305,
      "step": 15684
    },
    {
      "epoch": 0.6500476604915247,
      "grad_norm": 0.39661917090415955,
      "learning_rate": 1.7499689170707448e-06,
      "loss": 0.6362,
      "step": 15685
    },
    {
      "epoch": 0.6500891043971984,
      "grad_norm": 0.42612549662590027,
      "learning_rate": 1.7497616975423764e-06,
      "loss": 0.6174,
      "step": 15686
    },
    {
      "epoch": 0.6501305483028721,
      "grad_norm": 0.4097004234790802,
      "learning_rate": 1.749554478014008e-06,
      "loss": 0.6559,
      "step": 15687
    },
    {
      "epoch": 0.6501719922085457,
      "grad_norm": 0.44937238097190857,
      "learning_rate": 1.74934725848564e-06,
      "loss": 0.7092,
      "step": 15688
    },
    {
      "epoch": 0.6502134361142194,
      "grad_norm": 0.42180415987968445,
      "learning_rate": 1.7491400389572716e-06,
      "loss": 0.6865,
      "step": 15689
    },
    {
      "epoch": 0.6502548800198931,
      "grad_norm": 0.4132656157016754,
      "learning_rate": 1.7489328194289032e-06,
      "loss": 0.6841,
      "step": 15690
    },
    {
      "epoch": 0.6502963239255668,
      "grad_norm": 0.3920235335826874,
      "learning_rate": 1.7487255999005348e-06,
      "loss": 0.6725,
      "step": 15691
    },
    {
      "epoch": 0.6503377678312404,
      "grad_norm": 0.40079450607299805,
      "learning_rate": 1.7485183803721664e-06,
      "loss": 0.6591,
      "step": 15692
    },
    {
      "epoch": 0.650379211736914,
      "grad_norm": 0.40417376160621643,
      "learning_rate": 1.748311160843798e-06,
      "loss": 0.7151,
      "step": 15693
    },
    {
      "epoch": 0.6504206556425878,
      "grad_norm": 0.3965234160423279,
      "learning_rate": 1.7481039413154296e-06,
      "loss": 0.6196,
      "step": 15694
    },
    {
      "epoch": 0.6504620995482614,
      "grad_norm": 0.4023759067058563,
      "learning_rate": 1.7478967217870612e-06,
      "loss": 0.678,
      "step": 15695
    },
    {
      "epoch": 0.6505035434539351,
      "grad_norm": 0.4178605079650879,
      "learning_rate": 1.7476895022586928e-06,
      "loss": 0.6666,
      "step": 15696
    },
    {
      "epoch": 0.6505449873596088,
      "grad_norm": 0.47339579463005066,
      "learning_rate": 1.7474822827303248e-06,
      "loss": 0.701,
      "step": 15697
    },
    {
      "epoch": 0.6505864312652825,
      "grad_norm": 0.4555041790008545,
      "learning_rate": 1.7472750632019564e-06,
      "loss": 0.6528,
      "step": 15698
    },
    {
      "epoch": 0.6506278751709561,
      "grad_norm": 0.4053938388824463,
      "learning_rate": 1.747067843673588e-06,
      "loss": 0.6787,
      "step": 15699
    },
    {
      "epoch": 0.6506693190766297,
      "grad_norm": 0.43695059418678284,
      "learning_rate": 1.7468606241452196e-06,
      "loss": 0.6567,
      "step": 15700
    },
    {
      "epoch": 0.6507107629823035,
      "grad_norm": 0.39927127957344055,
      "learning_rate": 1.7466534046168512e-06,
      "loss": 0.6936,
      "step": 15701
    },
    {
      "epoch": 0.6507522068879771,
      "grad_norm": 0.40479668974876404,
      "learning_rate": 1.7464461850884828e-06,
      "loss": 0.703,
      "step": 15702
    },
    {
      "epoch": 0.6507936507936508,
      "grad_norm": 0.42943885922431946,
      "learning_rate": 1.7462389655601144e-06,
      "loss": 0.6963,
      "step": 15703
    },
    {
      "epoch": 0.6508350946993244,
      "grad_norm": 0.3916890621185303,
      "learning_rate": 1.746031746031746e-06,
      "loss": 0.6968,
      "step": 15704
    },
    {
      "epoch": 0.6508765386049982,
      "grad_norm": 0.4107820391654968,
      "learning_rate": 1.7458245265033776e-06,
      "loss": 0.6675,
      "step": 15705
    },
    {
      "epoch": 0.6509179825106718,
      "grad_norm": 0.3747807443141937,
      "learning_rate": 1.7456173069750096e-06,
      "loss": 0.6281,
      "step": 15706
    },
    {
      "epoch": 0.6509594264163455,
      "grad_norm": 0.40859073400497437,
      "learning_rate": 1.7454100874466412e-06,
      "loss": 0.6841,
      "step": 15707
    },
    {
      "epoch": 0.6510008703220191,
      "grad_norm": 0.39724966883659363,
      "learning_rate": 1.7452028679182728e-06,
      "loss": 0.6636,
      "step": 15708
    },
    {
      "epoch": 0.6510423142276928,
      "grad_norm": 0.4036860764026642,
      "learning_rate": 1.7449956483899044e-06,
      "loss": 0.6777,
      "step": 15709
    },
    {
      "epoch": 0.6510837581333665,
      "grad_norm": 0.4448586106300354,
      "learning_rate": 1.744788428861536e-06,
      "loss": 0.7024,
      "step": 15710
    },
    {
      "epoch": 0.6511252020390401,
      "grad_norm": 0.41343891620635986,
      "learning_rate": 1.7445812093331676e-06,
      "loss": 0.6592,
      "step": 15711
    },
    {
      "epoch": 0.6511666459447139,
      "grad_norm": 0.387207955121994,
      "learning_rate": 1.7443739898047992e-06,
      "loss": 0.6584,
      "step": 15712
    },
    {
      "epoch": 0.6512080898503875,
      "grad_norm": 0.4037681221961975,
      "learning_rate": 1.7441667702764308e-06,
      "loss": 0.6748,
      "step": 15713
    },
    {
      "epoch": 0.6512495337560612,
      "grad_norm": 0.5111191272735596,
      "learning_rate": 1.7439595507480624e-06,
      "loss": 0.6509,
      "step": 15714
    },
    {
      "epoch": 0.6512909776617348,
      "grad_norm": 0.4426088035106659,
      "learning_rate": 1.7437523312196944e-06,
      "loss": 0.6965,
      "step": 15715
    },
    {
      "epoch": 0.6513324215674086,
      "grad_norm": 0.4261202812194824,
      "learning_rate": 1.743545111691326e-06,
      "loss": 0.7212,
      "step": 15716
    },
    {
      "epoch": 0.6513738654730822,
      "grad_norm": 0.43412360548973083,
      "learning_rate": 1.7433378921629576e-06,
      "loss": 0.6752,
      "step": 15717
    },
    {
      "epoch": 0.6514153093787558,
      "grad_norm": 0.39340782165527344,
      "learning_rate": 1.7431306726345892e-06,
      "loss": 0.6741,
      "step": 15718
    },
    {
      "epoch": 0.6514567532844295,
      "grad_norm": 0.40273159742355347,
      "learning_rate": 1.7429234531062208e-06,
      "loss": 0.7104,
      "step": 15719
    },
    {
      "epoch": 0.6514981971901032,
      "grad_norm": 0.43781721591949463,
      "learning_rate": 1.7427162335778524e-06,
      "loss": 0.72,
      "step": 15720
    },
    {
      "epoch": 0.6515396410957769,
      "grad_norm": 0.4172431528568268,
      "learning_rate": 1.742509014049484e-06,
      "loss": 0.7109,
      "step": 15721
    },
    {
      "epoch": 0.6515810850014505,
      "grad_norm": 0.39169105887413025,
      "learning_rate": 1.7423017945211156e-06,
      "loss": 0.6504,
      "step": 15722
    },
    {
      "epoch": 0.6516225289071242,
      "grad_norm": 0.44551441073417664,
      "learning_rate": 1.7420945749927476e-06,
      "loss": 0.7129,
      "step": 15723
    },
    {
      "epoch": 0.6516639728127979,
      "grad_norm": 0.4175686538219452,
      "learning_rate": 1.7418873554643792e-06,
      "loss": 0.6648,
      "step": 15724
    },
    {
      "epoch": 0.6517054167184716,
      "grad_norm": 0.423713356256485,
      "learning_rate": 1.7416801359360108e-06,
      "loss": 0.6682,
      "step": 15725
    },
    {
      "epoch": 0.6517468606241452,
      "grad_norm": 0.39901238679885864,
      "learning_rate": 1.7414729164076424e-06,
      "loss": 0.7092,
      "step": 15726
    },
    {
      "epoch": 0.6517883045298188,
      "grad_norm": 0.43976593017578125,
      "learning_rate": 1.741265696879274e-06,
      "loss": 0.7295,
      "step": 15727
    },
    {
      "epoch": 0.6518297484354926,
      "grad_norm": 0.4272323548793793,
      "learning_rate": 1.7410584773509056e-06,
      "loss": 0.6616,
      "step": 15728
    },
    {
      "epoch": 0.6518711923411662,
      "grad_norm": 0.42126983404159546,
      "learning_rate": 1.7408512578225372e-06,
      "loss": 0.6682,
      "step": 15729
    },
    {
      "epoch": 0.6519126362468399,
      "grad_norm": 0.413254052400589,
      "learning_rate": 1.7406440382941688e-06,
      "loss": 0.6818,
      "step": 15730
    },
    {
      "epoch": 0.6519540801525135,
      "grad_norm": 0.40079787373542786,
      "learning_rate": 1.7404368187658004e-06,
      "loss": 0.668,
      "step": 15731
    },
    {
      "epoch": 0.6519955240581873,
      "grad_norm": 0.44727012515068054,
      "learning_rate": 1.7402295992374324e-06,
      "loss": 0.6926,
      "step": 15732
    },
    {
      "epoch": 0.6520369679638609,
      "grad_norm": 0.43767988681793213,
      "learning_rate": 1.740022379709064e-06,
      "loss": 0.6671,
      "step": 15733
    },
    {
      "epoch": 0.6520784118695346,
      "grad_norm": 0.42886170744895935,
      "learning_rate": 1.7398151601806956e-06,
      "loss": 0.7085,
      "step": 15734
    },
    {
      "epoch": 0.6521198557752083,
      "grad_norm": 0.4242027997970581,
      "learning_rate": 1.7396079406523272e-06,
      "loss": 0.6851,
      "step": 15735
    },
    {
      "epoch": 0.6521612996808819,
      "grad_norm": 0.4369575083255768,
      "learning_rate": 1.7394007211239588e-06,
      "loss": 0.687,
      "step": 15736
    },
    {
      "epoch": 0.6522027435865556,
      "grad_norm": 0.44731587171554565,
      "learning_rate": 1.7391935015955904e-06,
      "loss": 0.7048,
      "step": 15737
    },
    {
      "epoch": 0.6522441874922292,
      "grad_norm": 0.4220747947692871,
      "learning_rate": 1.738986282067222e-06,
      "loss": 0.6747,
      "step": 15738
    },
    {
      "epoch": 0.652285631397903,
      "grad_norm": 0.45188236236572266,
      "learning_rate": 1.7387790625388536e-06,
      "loss": 0.7251,
      "step": 15739
    },
    {
      "epoch": 0.6523270753035766,
      "grad_norm": 0.4787285029888153,
      "learning_rate": 1.7385718430104854e-06,
      "loss": 0.708,
      "step": 15740
    },
    {
      "epoch": 0.6523685192092503,
      "grad_norm": 0.40460318326950073,
      "learning_rate": 1.7383646234821172e-06,
      "loss": 0.6807,
      "step": 15741
    },
    {
      "epoch": 0.6524099631149239,
      "grad_norm": 0.42794525623321533,
      "learning_rate": 1.7381574039537488e-06,
      "loss": 0.6592,
      "step": 15742
    },
    {
      "epoch": 0.6524514070205977,
      "grad_norm": 0.4368503987789154,
      "learning_rate": 1.7379501844253804e-06,
      "loss": 0.7113,
      "step": 15743
    },
    {
      "epoch": 0.6524928509262713,
      "grad_norm": 0.39578866958618164,
      "learning_rate": 1.737742964897012e-06,
      "loss": 0.6418,
      "step": 15744
    },
    {
      "epoch": 0.6525342948319449,
      "grad_norm": 0.394930899143219,
      "learning_rate": 1.7375357453686436e-06,
      "loss": 0.679,
      "step": 15745
    },
    {
      "epoch": 0.6525757387376186,
      "grad_norm": 0.39836016297340393,
      "learning_rate": 1.7373285258402752e-06,
      "loss": 0.6707,
      "step": 15746
    },
    {
      "epoch": 0.6526171826432923,
      "grad_norm": 0.4076630771160126,
      "learning_rate": 1.7371213063119068e-06,
      "loss": 0.6771,
      "step": 15747
    },
    {
      "epoch": 0.652658626548966,
      "grad_norm": 0.4014841914176941,
      "learning_rate": 1.7369140867835384e-06,
      "loss": 0.6897,
      "step": 15748
    },
    {
      "epoch": 0.6527000704546396,
      "grad_norm": 0.4401458501815796,
      "learning_rate": 1.7367068672551704e-06,
      "loss": 0.6854,
      "step": 15749
    },
    {
      "epoch": 0.6527415143603134,
      "grad_norm": 0.41009724140167236,
      "learning_rate": 1.736499647726802e-06,
      "loss": 0.6951,
      "step": 15750
    },
    {
      "epoch": 0.652782958265987,
      "grad_norm": 0.42843538522720337,
      "learning_rate": 1.7362924281984336e-06,
      "loss": 0.7234,
      "step": 15751
    },
    {
      "epoch": 0.6528244021716607,
      "grad_norm": 0.41356161236763,
      "learning_rate": 1.7360852086700652e-06,
      "loss": 0.6855,
      "step": 15752
    },
    {
      "epoch": 0.6528658460773343,
      "grad_norm": 0.395937979221344,
      "learning_rate": 1.7358779891416968e-06,
      "loss": 0.6322,
      "step": 15753
    },
    {
      "epoch": 0.652907289983008,
      "grad_norm": 0.4456738829612732,
      "learning_rate": 1.7356707696133284e-06,
      "loss": 0.667,
      "step": 15754
    },
    {
      "epoch": 0.6529487338886817,
      "grad_norm": 0.4107266068458557,
      "learning_rate": 1.73546355008496e-06,
      "loss": 0.6565,
      "step": 15755
    },
    {
      "epoch": 0.6529901777943553,
      "grad_norm": 0.3851933479309082,
      "learning_rate": 1.7352563305565916e-06,
      "loss": 0.6565,
      "step": 15756
    },
    {
      "epoch": 0.653031621700029,
      "grad_norm": 0.426571249961853,
      "learning_rate": 1.7350491110282234e-06,
      "loss": 0.7231,
      "step": 15757
    },
    {
      "epoch": 0.6530730656057027,
      "grad_norm": 0.4126175045967102,
      "learning_rate": 1.7348418914998552e-06,
      "loss": 0.6329,
      "step": 15758
    },
    {
      "epoch": 0.6531145095113764,
      "grad_norm": 0.4429250955581665,
      "learning_rate": 1.7346346719714868e-06,
      "loss": 0.6556,
      "step": 15759
    },
    {
      "epoch": 0.65315595341705,
      "grad_norm": 0.4172758460044861,
      "learning_rate": 1.7344274524431184e-06,
      "loss": 0.6943,
      "step": 15760
    },
    {
      "epoch": 0.6531973973227236,
      "grad_norm": 0.414285272359848,
      "learning_rate": 1.73422023291475e-06,
      "loss": 0.6609,
      "step": 15761
    },
    {
      "epoch": 0.6532388412283974,
      "grad_norm": 0.39814522862434387,
      "learning_rate": 1.7340130133863816e-06,
      "loss": 0.6471,
      "step": 15762
    },
    {
      "epoch": 0.653280285134071,
      "grad_norm": 0.4019055664539337,
      "learning_rate": 1.7338057938580132e-06,
      "loss": 0.697,
      "step": 15763
    },
    {
      "epoch": 0.6533217290397447,
      "grad_norm": 0.4140705168247223,
      "learning_rate": 1.7335985743296448e-06,
      "loss": 0.6805,
      "step": 15764
    },
    {
      "epoch": 0.6533631729454183,
      "grad_norm": 0.3930293321609497,
      "learning_rate": 1.7333913548012764e-06,
      "loss": 0.6785,
      "step": 15765
    },
    {
      "epoch": 0.6534046168510921,
      "grad_norm": 0.41667789220809937,
      "learning_rate": 1.7331841352729082e-06,
      "loss": 0.7036,
      "step": 15766
    },
    {
      "epoch": 0.6534460607567657,
      "grad_norm": 0.404541939496994,
      "learning_rate": 1.73297691574454e-06,
      "loss": 0.6879,
      "step": 15767
    },
    {
      "epoch": 0.6534875046624394,
      "grad_norm": 0.38597381114959717,
      "learning_rate": 1.7327696962161716e-06,
      "loss": 0.6548,
      "step": 15768
    },
    {
      "epoch": 0.653528948568113,
      "grad_norm": 0.5062927007675171,
      "learning_rate": 1.7325624766878032e-06,
      "loss": 0.6995,
      "step": 15769
    },
    {
      "epoch": 0.6535703924737867,
      "grad_norm": 0.44508200883865356,
      "learning_rate": 1.7323552571594348e-06,
      "loss": 0.6768,
      "step": 15770
    },
    {
      "epoch": 0.6536118363794604,
      "grad_norm": 0.3669632077217102,
      "learning_rate": 1.7321480376310664e-06,
      "loss": 0.6306,
      "step": 15771
    },
    {
      "epoch": 0.653653280285134,
      "grad_norm": 0.3990861177444458,
      "learning_rate": 1.731940818102698e-06,
      "loss": 0.626,
      "step": 15772
    },
    {
      "epoch": 0.6536947241908078,
      "grad_norm": 0.41137459874153137,
      "learning_rate": 1.7317335985743296e-06,
      "loss": 0.7163,
      "step": 15773
    },
    {
      "epoch": 0.6537361680964814,
      "grad_norm": 0.409091979265213,
      "learning_rate": 1.7315263790459614e-06,
      "loss": 0.6924,
      "step": 15774
    },
    {
      "epoch": 0.6537776120021551,
      "grad_norm": 0.4474322199821472,
      "learning_rate": 1.7313191595175932e-06,
      "loss": 0.6869,
      "step": 15775
    },
    {
      "epoch": 0.6538190559078287,
      "grad_norm": 0.4054987132549286,
      "learning_rate": 1.7311119399892248e-06,
      "loss": 0.6887,
      "step": 15776
    },
    {
      "epoch": 0.6538604998135025,
      "grad_norm": 0.4308711290359497,
      "learning_rate": 1.7309047204608564e-06,
      "loss": 0.6984,
      "step": 15777
    },
    {
      "epoch": 0.6539019437191761,
      "grad_norm": 0.3708193004131317,
      "learning_rate": 1.730697500932488e-06,
      "loss": 0.6377,
      "step": 15778
    },
    {
      "epoch": 0.6539433876248497,
      "grad_norm": 0.39931997656822205,
      "learning_rate": 1.7304902814041196e-06,
      "loss": 0.6846,
      "step": 15779
    },
    {
      "epoch": 0.6539848315305234,
      "grad_norm": 0.4054282307624817,
      "learning_rate": 1.7302830618757512e-06,
      "loss": 0.6185,
      "step": 15780
    },
    {
      "epoch": 0.6540262754361971,
      "grad_norm": 0.3906192481517792,
      "learning_rate": 1.7300758423473828e-06,
      "loss": 0.5784,
      "step": 15781
    },
    {
      "epoch": 0.6540677193418708,
      "grad_norm": 0.40949708223342896,
      "learning_rate": 1.7298686228190144e-06,
      "loss": 0.645,
      "step": 15782
    },
    {
      "epoch": 0.6541091632475444,
      "grad_norm": 0.4361708164215088,
      "learning_rate": 1.7296614032906462e-06,
      "loss": 0.7278,
      "step": 15783
    },
    {
      "epoch": 0.6541506071532182,
      "grad_norm": 0.4104963541030884,
      "learning_rate": 1.729454183762278e-06,
      "loss": 0.7242,
      "step": 15784
    },
    {
      "epoch": 0.6541920510588918,
      "grad_norm": 0.39452454447746277,
      "learning_rate": 1.7292469642339096e-06,
      "loss": 0.6606,
      "step": 15785
    },
    {
      "epoch": 0.6542334949645655,
      "grad_norm": 0.3919631242752075,
      "learning_rate": 1.7290397447055412e-06,
      "loss": 0.6147,
      "step": 15786
    },
    {
      "epoch": 0.6542749388702391,
      "grad_norm": 0.5125225782394409,
      "learning_rate": 1.7288325251771728e-06,
      "loss": 0.7137,
      "step": 15787
    },
    {
      "epoch": 0.6543163827759128,
      "grad_norm": 0.4215662181377411,
      "learning_rate": 1.7286253056488044e-06,
      "loss": 0.6575,
      "step": 15788
    },
    {
      "epoch": 0.6543578266815865,
      "grad_norm": 0.36606648564338684,
      "learning_rate": 1.728418086120436e-06,
      "loss": 0.635,
      "step": 15789
    },
    {
      "epoch": 0.6543992705872601,
      "grad_norm": 0.3957790732383728,
      "learning_rate": 1.7282108665920676e-06,
      "loss": 0.6935,
      "step": 15790
    },
    {
      "epoch": 0.6544407144929338,
      "grad_norm": 0.4405267834663391,
      "learning_rate": 1.7280036470636994e-06,
      "loss": 0.6677,
      "step": 15791
    },
    {
      "epoch": 0.6544821583986075,
      "grad_norm": 0.41332128643989563,
      "learning_rate": 1.727796427535331e-06,
      "loss": 0.6958,
      "step": 15792
    },
    {
      "epoch": 0.6545236023042812,
      "grad_norm": 0.4082602560520172,
      "learning_rate": 1.7275892080069628e-06,
      "loss": 0.5955,
      "step": 15793
    },
    {
      "epoch": 0.6545650462099548,
      "grad_norm": 0.4165429472923279,
      "learning_rate": 1.7273819884785944e-06,
      "loss": 0.6996,
      "step": 15794
    },
    {
      "epoch": 0.6546064901156285,
      "grad_norm": 0.40681174397468567,
      "learning_rate": 1.727174768950226e-06,
      "loss": 0.6165,
      "step": 15795
    },
    {
      "epoch": 0.6546479340213022,
      "grad_norm": 0.40062281489372253,
      "learning_rate": 1.7269675494218576e-06,
      "loss": 0.6785,
      "step": 15796
    },
    {
      "epoch": 0.6546893779269758,
      "grad_norm": 0.4208965301513672,
      "learning_rate": 1.7267603298934892e-06,
      "loss": 0.7029,
      "step": 15797
    },
    {
      "epoch": 0.6547308218326495,
      "grad_norm": 0.42980527877807617,
      "learning_rate": 1.7265531103651208e-06,
      "loss": 0.6875,
      "step": 15798
    },
    {
      "epoch": 0.6547722657383231,
      "grad_norm": 0.40014299750328064,
      "learning_rate": 1.7263458908367524e-06,
      "loss": 0.6665,
      "step": 15799
    },
    {
      "epoch": 0.6548137096439969,
      "grad_norm": 0.40552690625190735,
      "learning_rate": 1.7261386713083842e-06,
      "loss": 0.6929,
      "step": 15800
    },
    {
      "epoch": 0.6548551535496705,
      "grad_norm": 0.4124605059623718,
      "learning_rate": 1.7259314517800158e-06,
      "loss": 0.6472,
      "step": 15801
    },
    {
      "epoch": 0.6548965974553442,
      "grad_norm": 0.40545380115509033,
      "learning_rate": 1.7257242322516476e-06,
      "loss": 0.6793,
      "step": 15802
    },
    {
      "epoch": 0.6549380413610179,
      "grad_norm": 0.40259498357772827,
      "learning_rate": 1.7255170127232792e-06,
      "loss": 0.6975,
      "step": 15803
    },
    {
      "epoch": 0.6549794852666916,
      "grad_norm": 0.40491700172424316,
      "learning_rate": 1.7253097931949108e-06,
      "loss": 0.7051,
      "step": 15804
    },
    {
      "epoch": 0.6550209291723652,
      "grad_norm": 0.40791767835617065,
      "learning_rate": 1.7251025736665424e-06,
      "loss": 0.7026,
      "step": 15805
    },
    {
      "epoch": 0.6550623730780388,
      "grad_norm": 0.44549524784088135,
      "learning_rate": 1.724895354138174e-06,
      "loss": 0.71,
      "step": 15806
    },
    {
      "epoch": 0.6551038169837126,
      "grad_norm": 0.4575514793395996,
      "learning_rate": 1.7246881346098056e-06,
      "loss": 0.709,
      "step": 15807
    },
    {
      "epoch": 0.6551452608893862,
      "grad_norm": 0.4460285007953644,
      "learning_rate": 1.7244809150814374e-06,
      "loss": 0.6729,
      "step": 15808
    },
    {
      "epoch": 0.6551867047950599,
      "grad_norm": 0.43750715255737305,
      "learning_rate": 1.724273695553069e-06,
      "loss": 0.7068,
      "step": 15809
    },
    {
      "epoch": 0.6552281487007335,
      "grad_norm": 0.4101375341415405,
      "learning_rate": 1.7240664760247008e-06,
      "loss": 0.6461,
      "step": 15810
    },
    {
      "epoch": 0.6552695926064073,
      "grad_norm": 0.4052460193634033,
      "learning_rate": 1.7238592564963324e-06,
      "loss": 0.7023,
      "step": 15811
    },
    {
      "epoch": 0.6553110365120809,
      "grad_norm": 0.408568799495697,
      "learning_rate": 1.723652036967964e-06,
      "loss": 0.6372,
      "step": 15812
    },
    {
      "epoch": 0.6553524804177546,
      "grad_norm": 0.3925889730453491,
      "learning_rate": 1.7234448174395956e-06,
      "loss": 0.6581,
      "step": 15813
    },
    {
      "epoch": 0.6553939243234282,
      "grad_norm": 0.4190484583377838,
      "learning_rate": 1.7232375979112272e-06,
      "loss": 0.6832,
      "step": 15814
    },
    {
      "epoch": 0.6554353682291019,
      "grad_norm": 0.4159119725227356,
      "learning_rate": 1.7230303783828588e-06,
      "loss": 0.6613,
      "step": 15815
    },
    {
      "epoch": 0.6554768121347756,
      "grad_norm": 0.40684086084365845,
      "learning_rate": 1.7228231588544906e-06,
      "loss": 0.6302,
      "step": 15816
    },
    {
      "epoch": 0.6555182560404492,
      "grad_norm": 0.4096267819404602,
      "learning_rate": 1.7226159393261222e-06,
      "loss": 0.717,
      "step": 15817
    },
    {
      "epoch": 0.655559699946123,
      "grad_norm": 0.43406757712364197,
      "learning_rate": 1.7224087197977538e-06,
      "loss": 0.7068,
      "step": 15818
    },
    {
      "epoch": 0.6556011438517966,
      "grad_norm": 0.385793536901474,
      "learning_rate": 1.7222015002693856e-06,
      "loss": 0.7109,
      "step": 15819
    },
    {
      "epoch": 0.6556425877574703,
      "grad_norm": 0.424255907535553,
      "learning_rate": 1.7219942807410172e-06,
      "loss": 0.7064,
      "step": 15820
    },
    {
      "epoch": 0.6556840316631439,
      "grad_norm": 0.4171079397201538,
      "learning_rate": 1.7217870612126488e-06,
      "loss": 0.7078,
      "step": 15821
    },
    {
      "epoch": 0.6557254755688176,
      "grad_norm": 0.40043315291404724,
      "learning_rate": 1.7215798416842804e-06,
      "loss": 0.6997,
      "step": 15822
    },
    {
      "epoch": 0.6557669194744913,
      "grad_norm": 0.4386458992958069,
      "learning_rate": 1.721372622155912e-06,
      "loss": 0.7202,
      "step": 15823
    },
    {
      "epoch": 0.6558083633801649,
      "grad_norm": 0.39670661091804504,
      "learning_rate": 1.7211654026275436e-06,
      "loss": 0.6692,
      "step": 15824
    },
    {
      "epoch": 0.6558498072858386,
      "grad_norm": 0.42229771614074707,
      "learning_rate": 1.7209581830991754e-06,
      "loss": 0.6913,
      "step": 15825
    },
    {
      "epoch": 0.6558912511915123,
      "grad_norm": 0.3951636850833893,
      "learning_rate": 1.720750963570807e-06,
      "loss": 0.6675,
      "step": 15826
    },
    {
      "epoch": 0.655932695097186,
      "grad_norm": 0.4452580511569977,
      "learning_rate": 1.7205437440424386e-06,
      "loss": 0.707,
      "step": 15827
    },
    {
      "epoch": 0.6559741390028596,
      "grad_norm": 0.4179422855377197,
      "learning_rate": 1.7203365245140704e-06,
      "loss": 0.6278,
      "step": 15828
    },
    {
      "epoch": 0.6560155829085333,
      "grad_norm": 0.4243967533111572,
      "learning_rate": 1.720129304985702e-06,
      "loss": 0.7197,
      "step": 15829
    },
    {
      "epoch": 0.656057026814207,
      "grad_norm": 0.4189925193786621,
      "learning_rate": 1.7199220854573336e-06,
      "loss": 0.6897,
      "step": 15830
    },
    {
      "epoch": 0.6560984707198806,
      "grad_norm": 0.4216445982456207,
      "learning_rate": 1.7197148659289652e-06,
      "loss": 0.75,
      "step": 15831
    },
    {
      "epoch": 0.6561399146255543,
      "grad_norm": 0.41182741522789,
      "learning_rate": 1.7195076464005968e-06,
      "loss": 0.6885,
      "step": 15832
    },
    {
      "epoch": 0.6561813585312279,
      "grad_norm": 0.4273031949996948,
      "learning_rate": 1.7193004268722286e-06,
      "loss": 0.7373,
      "step": 15833
    },
    {
      "epoch": 0.6562228024369017,
      "grad_norm": 0.4141566753387451,
      "learning_rate": 1.7190932073438602e-06,
      "loss": 0.7141,
      "step": 15834
    },
    {
      "epoch": 0.6562642463425753,
      "grad_norm": 0.3880942165851593,
      "learning_rate": 1.7188859878154918e-06,
      "loss": 0.6788,
      "step": 15835
    },
    {
      "epoch": 0.656305690248249,
      "grad_norm": 0.4042184054851532,
      "learning_rate": 1.7186787682871236e-06,
      "loss": 0.6592,
      "step": 15836
    },
    {
      "epoch": 0.6563471341539227,
      "grad_norm": 0.46124470233917236,
      "learning_rate": 1.7184715487587552e-06,
      "loss": 0.7415,
      "step": 15837
    },
    {
      "epoch": 0.6563885780595964,
      "grad_norm": 0.40873071551322937,
      "learning_rate": 1.7182643292303868e-06,
      "loss": 0.6603,
      "step": 15838
    },
    {
      "epoch": 0.65643002196527,
      "grad_norm": 0.420430988073349,
      "learning_rate": 1.7180571097020184e-06,
      "loss": 0.6494,
      "step": 15839
    },
    {
      "epoch": 0.6564714658709436,
      "grad_norm": 0.4463101923465729,
      "learning_rate": 1.71784989017365e-06,
      "loss": 0.6919,
      "step": 15840
    },
    {
      "epoch": 0.6565129097766174,
      "grad_norm": 0.3935699760913849,
      "learning_rate": 1.7176426706452816e-06,
      "loss": 0.5975,
      "step": 15841
    },
    {
      "epoch": 0.656554353682291,
      "grad_norm": 0.4163416028022766,
      "learning_rate": 1.7174354511169134e-06,
      "loss": 0.6686,
      "step": 15842
    },
    {
      "epoch": 0.6565957975879647,
      "grad_norm": 0.4248247742652893,
      "learning_rate": 1.717228231588545e-06,
      "loss": 0.6904,
      "step": 15843
    },
    {
      "epoch": 0.6566372414936383,
      "grad_norm": 0.3977970778942108,
      "learning_rate": 1.7170210120601766e-06,
      "loss": 0.6377,
      "step": 15844
    },
    {
      "epoch": 0.6566786853993121,
      "grad_norm": 0.3912927210330963,
      "learning_rate": 1.7168137925318084e-06,
      "loss": 0.6549,
      "step": 15845
    },
    {
      "epoch": 0.6567201293049857,
      "grad_norm": 0.4295571446418762,
      "learning_rate": 1.71660657300344e-06,
      "loss": 0.6997,
      "step": 15846
    },
    {
      "epoch": 0.6567615732106594,
      "grad_norm": 0.415941447019577,
      "learning_rate": 1.7163993534750716e-06,
      "loss": 0.6807,
      "step": 15847
    },
    {
      "epoch": 0.656803017116333,
      "grad_norm": 0.5575341582298279,
      "learning_rate": 1.7161921339467032e-06,
      "loss": 0.6826,
      "step": 15848
    },
    {
      "epoch": 0.6568444610220067,
      "grad_norm": 0.49588531255722046,
      "learning_rate": 1.7159849144183348e-06,
      "loss": 0.6343,
      "step": 15849
    },
    {
      "epoch": 0.6568859049276804,
      "grad_norm": 0.4204784631729126,
      "learning_rate": 1.7157776948899666e-06,
      "loss": 0.7034,
      "step": 15850
    },
    {
      "epoch": 0.656927348833354,
      "grad_norm": 0.38081130385398865,
      "learning_rate": 1.7155704753615982e-06,
      "loss": 0.7002,
      "step": 15851
    },
    {
      "epoch": 0.6569687927390278,
      "grad_norm": 0.40828636288642883,
      "learning_rate": 1.7153632558332298e-06,
      "loss": 0.6985,
      "step": 15852
    },
    {
      "epoch": 0.6570102366447014,
      "grad_norm": 0.37971195578575134,
      "learning_rate": 1.7151560363048614e-06,
      "loss": 0.6223,
      "step": 15853
    },
    {
      "epoch": 0.6570516805503751,
      "grad_norm": 0.43649357557296753,
      "learning_rate": 1.7149488167764932e-06,
      "loss": 0.697,
      "step": 15854
    },
    {
      "epoch": 0.6570931244560487,
      "grad_norm": 0.5322659015655518,
      "learning_rate": 1.7147415972481248e-06,
      "loss": 0.6671,
      "step": 15855
    },
    {
      "epoch": 0.6571345683617225,
      "grad_norm": 0.3835306167602539,
      "learning_rate": 1.7145343777197564e-06,
      "loss": 0.6781,
      "step": 15856
    },
    {
      "epoch": 0.6571760122673961,
      "grad_norm": 0.42388784885406494,
      "learning_rate": 1.714327158191388e-06,
      "loss": 0.6567,
      "step": 15857
    },
    {
      "epoch": 0.6572174561730697,
      "grad_norm": 0.4194665551185608,
      "learning_rate": 1.7141199386630196e-06,
      "loss": 0.6995,
      "step": 15858
    },
    {
      "epoch": 0.6572589000787434,
      "grad_norm": 0.39259710907936096,
      "learning_rate": 1.7139127191346514e-06,
      "loss": 0.6517,
      "step": 15859
    },
    {
      "epoch": 0.6573003439844171,
      "grad_norm": 0.3966829180717468,
      "learning_rate": 1.713705499606283e-06,
      "loss": 0.6904,
      "step": 15860
    },
    {
      "epoch": 0.6573417878900908,
      "grad_norm": 0.40633827447891235,
      "learning_rate": 1.7134982800779146e-06,
      "loss": 0.6731,
      "step": 15861
    },
    {
      "epoch": 0.6573832317957644,
      "grad_norm": 0.4272167980670929,
      "learning_rate": 1.7132910605495462e-06,
      "loss": 0.6674,
      "step": 15862
    },
    {
      "epoch": 0.6574246757014381,
      "grad_norm": 0.4242865741252899,
      "learning_rate": 1.713083841021178e-06,
      "loss": 0.6796,
      "step": 15863
    },
    {
      "epoch": 0.6574661196071118,
      "grad_norm": 0.3944559395313263,
      "learning_rate": 1.7128766214928096e-06,
      "loss": 0.6628,
      "step": 15864
    },
    {
      "epoch": 0.6575075635127855,
      "grad_norm": 0.39071494340896606,
      "learning_rate": 1.7126694019644412e-06,
      "loss": 0.6533,
      "step": 15865
    },
    {
      "epoch": 0.6575490074184591,
      "grad_norm": 0.40022996068000793,
      "learning_rate": 1.7124621824360728e-06,
      "loss": 0.6685,
      "step": 15866
    },
    {
      "epoch": 0.6575904513241327,
      "grad_norm": 0.38033562898635864,
      "learning_rate": 1.7122549629077046e-06,
      "loss": 0.6238,
      "step": 15867
    },
    {
      "epoch": 0.6576318952298065,
      "grad_norm": 0.4154983162879944,
      "learning_rate": 1.7120477433793362e-06,
      "loss": 0.6594,
      "step": 15868
    },
    {
      "epoch": 0.6576733391354801,
      "grad_norm": 0.3927316963672638,
      "learning_rate": 1.7118405238509678e-06,
      "loss": 0.6697,
      "step": 15869
    },
    {
      "epoch": 0.6577147830411538,
      "grad_norm": 0.40432554483413696,
      "learning_rate": 1.7116333043225994e-06,
      "loss": 0.7039,
      "step": 15870
    },
    {
      "epoch": 0.6577562269468274,
      "grad_norm": 0.38262805342674255,
      "learning_rate": 1.7114260847942312e-06,
      "loss": 0.6771,
      "step": 15871
    },
    {
      "epoch": 0.6577976708525012,
      "grad_norm": 0.43259990215301514,
      "learning_rate": 1.7112188652658628e-06,
      "loss": 0.6448,
      "step": 15872
    },
    {
      "epoch": 0.6578391147581748,
      "grad_norm": 0.4208540618419647,
      "learning_rate": 1.7110116457374944e-06,
      "loss": 0.6934,
      "step": 15873
    },
    {
      "epoch": 0.6578805586638485,
      "grad_norm": 0.4065565764904022,
      "learning_rate": 1.710804426209126e-06,
      "loss": 0.6309,
      "step": 15874
    },
    {
      "epoch": 0.6579220025695222,
      "grad_norm": 0.4515267610549927,
      "learning_rate": 1.7105972066807576e-06,
      "loss": 0.6863,
      "step": 15875
    },
    {
      "epoch": 0.6579634464751958,
      "grad_norm": 0.39399778842926025,
      "learning_rate": 1.7103899871523894e-06,
      "loss": 0.6019,
      "step": 15876
    },
    {
      "epoch": 0.6580048903808695,
      "grad_norm": 0.38645172119140625,
      "learning_rate": 1.710182767624021e-06,
      "loss": 0.6571,
      "step": 15877
    },
    {
      "epoch": 0.6580463342865431,
      "grad_norm": 0.40489277243614197,
      "learning_rate": 1.7099755480956526e-06,
      "loss": 0.6517,
      "step": 15878
    },
    {
      "epoch": 0.6580877781922169,
      "grad_norm": 0.40738117694854736,
      "learning_rate": 1.7097683285672842e-06,
      "loss": 0.7021,
      "step": 15879
    },
    {
      "epoch": 0.6581292220978905,
      "grad_norm": 0.40787044167518616,
      "learning_rate": 1.709561109038916e-06,
      "loss": 0.6906,
      "step": 15880
    },
    {
      "epoch": 0.6581706660035642,
      "grad_norm": 0.38955286145210266,
      "learning_rate": 1.7093538895105476e-06,
      "loss": 0.7096,
      "step": 15881
    },
    {
      "epoch": 0.6582121099092378,
      "grad_norm": 0.4115619957447052,
      "learning_rate": 1.7091466699821792e-06,
      "loss": 0.6593,
      "step": 15882
    },
    {
      "epoch": 0.6582535538149115,
      "grad_norm": 0.39343875646591187,
      "learning_rate": 1.7089394504538108e-06,
      "loss": 0.6759,
      "step": 15883
    },
    {
      "epoch": 0.6582949977205852,
      "grad_norm": 0.40498554706573486,
      "learning_rate": 1.7087322309254426e-06,
      "loss": 0.668,
      "step": 15884
    },
    {
      "epoch": 0.6583364416262588,
      "grad_norm": 0.3864211142063141,
      "learning_rate": 1.7085250113970742e-06,
      "loss": 0.6594,
      "step": 15885
    },
    {
      "epoch": 0.6583778855319325,
      "grad_norm": 0.4145214259624481,
      "learning_rate": 1.7083177918687058e-06,
      "loss": 0.7396,
      "step": 15886
    },
    {
      "epoch": 0.6584193294376062,
      "grad_norm": 0.3966812789440155,
      "learning_rate": 1.7081105723403374e-06,
      "loss": 0.645,
      "step": 15887
    },
    {
      "epoch": 0.6584607733432799,
      "grad_norm": 0.45376506447792053,
      "learning_rate": 1.707903352811969e-06,
      "loss": 0.7477,
      "step": 15888
    },
    {
      "epoch": 0.6585022172489535,
      "grad_norm": 0.42083802819252014,
      "learning_rate": 1.7076961332836008e-06,
      "loss": 0.6702,
      "step": 15889
    },
    {
      "epoch": 0.6585436611546273,
      "grad_norm": 0.3996073305606842,
      "learning_rate": 1.7074889137552324e-06,
      "loss": 0.6117,
      "step": 15890
    },
    {
      "epoch": 0.6585851050603009,
      "grad_norm": 0.4056505560874939,
      "learning_rate": 1.707281694226864e-06,
      "loss": 0.6304,
      "step": 15891
    },
    {
      "epoch": 0.6586265489659745,
      "grad_norm": 0.4683209955692291,
      "learning_rate": 1.7070744746984956e-06,
      "loss": 0.6975,
      "step": 15892
    },
    {
      "epoch": 0.6586679928716482,
      "grad_norm": 0.3804751932621002,
      "learning_rate": 1.7068672551701274e-06,
      "loss": 0.6525,
      "step": 15893
    },
    {
      "epoch": 0.6587094367773219,
      "grad_norm": 0.3976973295211792,
      "learning_rate": 1.706660035641759e-06,
      "loss": 0.6733,
      "step": 15894
    },
    {
      "epoch": 0.6587508806829956,
      "grad_norm": 0.4610307216644287,
      "learning_rate": 1.7064528161133906e-06,
      "loss": 0.7383,
      "step": 15895
    },
    {
      "epoch": 0.6587923245886692,
      "grad_norm": 0.3912046551704407,
      "learning_rate": 1.7062455965850222e-06,
      "loss": 0.6665,
      "step": 15896
    },
    {
      "epoch": 0.6588337684943429,
      "grad_norm": 0.3910112977027893,
      "learning_rate": 1.706038377056654e-06,
      "loss": 0.6439,
      "step": 15897
    },
    {
      "epoch": 0.6588752124000166,
      "grad_norm": 0.43328994512557983,
      "learning_rate": 1.7058311575282856e-06,
      "loss": 0.6835,
      "step": 15898
    },
    {
      "epoch": 0.6589166563056903,
      "grad_norm": 0.43128570914268494,
      "learning_rate": 1.7056239379999172e-06,
      "loss": 0.6743,
      "step": 15899
    },
    {
      "epoch": 0.6589581002113639,
      "grad_norm": 0.4320971369743347,
      "learning_rate": 1.7054167184715488e-06,
      "loss": 0.6869,
      "step": 15900
    },
    {
      "epoch": 0.6589995441170375,
      "grad_norm": 0.41274017095565796,
      "learning_rate": 1.7052094989431806e-06,
      "loss": 0.66,
      "step": 15901
    },
    {
      "epoch": 0.6590409880227113,
      "grad_norm": 0.44481131434440613,
      "learning_rate": 1.7050022794148122e-06,
      "loss": 0.6871,
      "step": 15902
    },
    {
      "epoch": 0.6590824319283849,
      "grad_norm": 0.3837624192237854,
      "learning_rate": 1.7047950598864438e-06,
      "loss": 0.6777,
      "step": 15903
    },
    {
      "epoch": 0.6591238758340586,
      "grad_norm": 0.4376017153263092,
      "learning_rate": 1.7045878403580754e-06,
      "loss": 0.7148,
      "step": 15904
    },
    {
      "epoch": 0.6591653197397322,
      "grad_norm": 0.4535057246685028,
      "learning_rate": 1.704380620829707e-06,
      "loss": 0.7454,
      "step": 15905
    },
    {
      "epoch": 0.659206763645406,
      "grad_norm": 0.4428234398365021,
      "learning_rate": 1.7041734013013388e-06,
      "loss": 0.6389,
      "step": 15906
    },
    {
      "epoch": 0.6592482075510796,
      "grad_norm": 0.43806129693984985,
      "learning_rate": 1.7039661817729704e-06,
      "loss": 0.7351,
      "step": 15907
    },
    {
      "epoch": 0.6592896514567533,
      "grad_norm": 0.43979784846305847,
      "learning_rate": 1.703758962244602e-06,
      "loss": 0.6935,
      "step": 15908
    },
    {
      "epoch": 0.659331095362427,
      "grad_norm": 0.416564404964447,
      "learning_rate": 1.7035517427162339e-06,
      "loss": 0.6406,
      "step": 15909
    },
    {
      "epoch": 0.6593725392681006,
      "grad_norm": 0.3931574523448944,
      "learning_rate": 1.7033445231878654e-06,
      "loss": 0.647,
      "step": 15910
    },
    {
      "epoch": 0.6594139831737743,
      "grad_norm": 0.38810956478118896,
      "learning_rate": 1.703137303659497e-06,
      "loss": 0.6935,
      "step": 15911
    },
    {
      "epoch": 0.6594554270794479,
      "grad_norm": 0.38679274916648865,
      "learning_rate": 1.7029300841311286e-06,
      "loss": 0.6147,
      "step": 15912
    },
    {
      "epoch": 0.6594968709851217,
      "grad_norm": 0.41585877537727356,
      "learning_rate": 1.7027228646027602e-06,
      "loss": 0.7,
      "step": 15913
    },
    {
      "epoch": 0.6595383148907953,
      "grad_norm": 0.40813443064689636,
      "learning_rate": 1.7025156450743918e-06,
      "loss": 0.6643,
      "step": 15914
    },
    {
      "epoch": 0.659579758796469,
      "grad_norm": 0.44207730889320374,
      "learning_rate": 1.7023084255460236e-06,
      "loss": 0.6857,
      "step": 15915
    },
    {
      "epoch": 0.6596212027021426,
      "grad_norm": 0.4151112139225006,
      "learning_rate": 1.7021012060176552e-06,
      "loss": 0.6924,
      "step": 15916
    },
    {
      "epoch": 0.6596626466078164,
      "grad_norm": 0.4181790351867676,
      "learning_rate": 1.7018939864892868e-06,
      "loss": 0.6512,
      "step": 15917
    },
    {
      "epoch": 0.65970409051349,
      "grad_norm": 0.3908064663410187,
      "learning_rate": 1.7016867669609187e-06,
      "loss": 0.6754,
      "step": 15918
    },
    {
      "epoch": 0.6597455344191636,
      "grad_norm": 0.4371959865093231,
      "learning_rate": 1.7014795474325502e-06,
      "loss": 0.7026,
      "step": 15919
    },
    {
      "epoch": 0.6597869783248373,
      "grad_norm": 0.4230654239654541,
      "learning_rate": 1.7012723279041818e-06,
      "loss": 0.6748,
      "step": 15920
    },
    {
      "epoch": 0.659828422230511,
      "grad_norm": 0.3909659683704376,
      "learning_rate": 1.7010651083758134e-06,
      "loss": 0.629,
      "step": 15921
    },
    {
      "epoch": 0.6598698661361847,
      "grad_norm": 0.39861851930618286,
      "learning_rate": 1.700857888847445e-06,
      "loss": 0.6663,
      "step": 15922
    },
    {
      "epoch": 0.6599113100418583,
      "grad_norm": 0.4235279858112335,
      "learning_rate": 1.7006506693190766e-06,
      "loss": 0.6687,
      "step": 15923
    },
    {
      "epoch": 0.6599527539475321,
      "grad_norm": 0.43056735396385193,
      "learning_rate": 1.7004434497907084e-06,
      "loss": 0.6729,
      "step": 15924
    },
    {
      "epoch": 0.6599941978532057,
      "grad_norm": 0.4405744969844818,
      "learning_rate": 1.70023623026234e-06,
      "loss": 0.6794,
      "step": 15925
    },
    {
      "epoch": 0.6600356417588794,
      "grad_norm": 0.41619405150413513,
      "learning_rate": 1.7000290107339719e-06,
      "loss": 0.6654,
      "step": 15926
    },
    {
      "epoch": 0.660077085664553,
      "grad_norm": 0.38647568225860596,
      "learning_rate": 1.6998217912056035e-06,
      "loss": 0.6709,
      "step": 15927
    },
    {
      "epoch": 0.6601185295702267,
      "grad_norm": 0.4220942556858063,
      "learning_rate": 1.699614571677235e-06,
      "loss": 0.6788,
      "step": 15928
    },
    {
      "epoch": 0.6601599734759004,
      "grad_norm": 0.44393622875213623,
      "learning_rate": 1.6994073521488666e-06,
      "loss": 0.7236,
      "step": 15929
    },
    {
      "epoch": 0.660201417381574,
      "grad_norm": 0.3824441432952881,
      "learning_rate": 1.6992001326204982e-06,
      "loss": 0.6794,
      "step": 15930
    },
    {
      "epoch": 0.6602428612872477,
      "grad_norm": 0.40702781081199646,
      "learning_rate": 1.6989929130921298e-06,
      "loss": 0.6897,
      "step": 15931
    },
    {
      "epoch": 0.6602843051929214,
      "grad_norm": 0.40720781683921814,
      "learning_rate": 1.6987856935637617e-06,
      "loss": 0.6589,
      "step": 15932
    },
    {
      "epoch": 0.6603257490985951,
      "grad_norm": 0.4167475998401642,
      "learning_rate": 1.6985784740353932e-06,
      "loss": 0.665,
      "step": 15933
    },
    {
      "epoch": 0.6603671930042687,
      "grad_norm": 0.39371156692504883,
      "learning_rate": 1.6983712545070248e-06,
      "loss": 0.6323,
      "step": 15934
    },
    {
      "epoch": 0.6604086369099424,
      "grad_norm": 0.3834646940231323,
      "learning_rate": 1.6981640349786567e-06,
      "loss": 0.6857,
      "step": 15935
    },
    {
      "epoch": 0.6604500808156161,
      "grad_norm": 0.3804472088813782,
      "learning_rate": 1.6979568154502883e-06,
      "loss": 0.62,
      "step": 15936
    },
    {
      "epoch": 0.6604915247212897,
      "grad_norm": 0.4005916714668274,
      "learning_rate": 1.6977495959219199e-06,
      "loss": 0.6201,
      "step": 15937
    },
    {
      "epoch": 0.6605329686269634,
      "grad_norm": 0.4113697409629822,
      "learning_rate": 1.6975423763935514e-06,
      "loss": 0.6401,
      "step": 15938
    },
    {
      "epoch": 0.660574412532637,
      "grad_norm": 0.39684680104255676,
      "learning_rate": 1.697335156865183e-06,
      "loss": 0.6804,
      "step": 15939
    },
    {
      "epoch": 0.6606158564383108,
      "grad_norm": 0.42072048783302307,
      "learning_rate": 1.6971279373368146e-06,
      "loss": 0.6742,
      "step": 15940
    },
    {
      "epoch": 0.6606573003439844,
      "grad_norm": 0.4549179673194885,
      "learning_rate": 1.6969207178084465e-06,
      "loss": 0.6667,
      "step": 15941
    },
    {
      "epoch": 0.6606987442496581,
      "grad_norm": 0.4198736846446991,
      "learning_rate": 1.696713498280078e-06,
      "loss": 0.6803,
      "step": 15942
    },
    {
      "epoch": 0.6607401881553318,
      "grad_norm": 0.4072663486003876,
      "learning_rate": 1.6965062787517099e-06,
      "loss": 0.6143,
      "step": 15943
    },
    {
      "epoch": 0.6607816320610054,
      "grad_norm": 0.42493024468421936,
      "learning_rate": 1.6962990592233415e-06,
      "loss": 0.6804,
      "step": 15944
    },
    {
      "epoch": 0.6608230759666791,
      "grad_norm": 0.46426454186439514,
      "learning_rate": 1.696091839694973e-06,
      "loss": 0.6821,
      "step": 15945
    },
    {
      "epoch": 0.6608645198723527,
      "grad_norm": 0.40335360169410706,
      "learning_rate": 1.6958846201666047e-06,
      "loss": 0.699,
      "step": 15946
    },
    {
      "epoch": 0.6609059637780265,
      "grad_norm": 0.39876291155815125,
      "learning_rate": 1.6956774006382362e-06,
      "loss": 0.667,
      "step": 15947
    },
    {
      "epoch": 0.6609474076837001,
      "grad_norm": 0.40360960364341736,
      "learning_rate": 1.6954701811098678e-06,
      "loss": 0.6815,
      "step": 15948
    },
    {
      "epoch": 0.6609888515893738,
      "grad_norm": 0.4324796497821808,
      "learning_rate": 1.6952629615814994e-06,
      "loss": 0.6776,
      "step": 15949
    },
    {
      "epoch": 0.6610302954950474,
      "grad_norm": 0.3844256103038788,
      "learning_rate": 1.6950557420531313e-06,
      "loss": 0.6344,
      "step": 15950
    },
    {
      "epoch": 0.6610717394007212,
      "grad_norm": 0.3923284709453583,
      "learning_rate": 1.6948485225247628e-06,
      "loss": 0.6514,
      "step": 15951
    },
    {
      "epoch": 0.6611131833063948,
      "grad_norm": 0.4134000837802887,
      "learning_rate": 1.6946413029963947e-06,
      "loss": 0.6976,
      "step": 15952
    },
    {
      "epoch": 0.6611546272120684,
      "grad_norm": 0.391956627368927,
      "learning_rate": 1.6944340834680263e-06,
      "loss": 0.7101,
      "step": 15953
    },
    {
      "epoch": 0.6611960711177421,
      "grad_norm": 0.43002066016197205,
      "learning_rate": 1.6942268639396579e-06,
      "loss": 0.6836,
      "step": 15954
    },
    {
      "epoch": 0.6612375150234158,
      "grad_norm": 0.3834863007068634,
      "learning_rate": 1.6940196444112895e-06,
      "loss": 0.663,
      "step": 15955
    },
    {
      "epoch": 0.6612789589290895,
      "grad_norm": 0.42112401127815247,
      "learning_rate": 1.693812424882921e-06,
      "loss": 0.6636,
      "step": 15956
    },
    {
      "epoch": 0.6613204028347631,
      "grad_norm": 0.42690062522888184,
      "learning_rate": 1.6936052053545526e-06,
      "loss": 0.6875,
      "step": 15957
    },
    {
      "epoch": 0.6613618467404369,
      "grad_norm": 0.4197480082511902,
      "learning_rate": 1.6933979858261845e-06,
      "loss": 0.6628,
      "step": 15958
    },
    {
      "epoch": 0.6614032906461105,
      "grad_norm": 0.40382689237594604,
      "learning_rate": 1.693190766297816e-06,
      "loss": 0.6641,
      "step": 15959
    },
    {
      "epoch": 0.6614447345517842,
      "grad_norm": 0.408812016248703,
      "learning_rate": 1.6929835467694479e-06,
      "loss": 0.6428,
      "step": 15960
    },
    {
      "epoch": 0.6614861784574578,
      "grad_norm": 0.42023780941963196,
      "learning_rate": 1.6927763272410795e-06,
      "loss": 0.6415,
      "step": 15961
    },
    {
      "epoch": 0.6615276223631315,
      "grad_norm": 0.42537063360214233,
      "learning_rate": 1.692569107712711e-06,
      "loss": 0.6483,
      "step": 15962
    },
    {
      "epoch": 0.6615690662688052,
      "grad_norm": 0.37976691126823425,
      "learning_rate": 1.6923618881843427e-06,
      "loss": 0.6464,
      "step": 15963
    },
    {
      "epoch": 0.6616105101744788,
      "grad_norm": 0.3804509937763214,
      "learning_rate": 1.6921546686559743e-06,
      "loss": 0.6699,
      "step": 15964
    },
    {
      "epoch": 0.6616519540801525,
      "grad_norm": 0.4056214690208435,
      "learning_rate": 1.6919474491276058e-06,
      "loss": 0.6644,
      "step": 15965
    },
    {
      "epoch": 0.6616933979858262,
      "grad_norm": 0.44680482149124146,
      "learning_rate": 1.6917402295992374e-06,
      "loss": 0.684,
      "step": 15966
    },
    {
      "epoch": 0.6617348418914999,
      "grad_norm": 0.44334855675697327,
      "learning_rate": 1.6915330100708693e-06,
      "loss": 0.7544,
      "step": 15967
    },
    {
      "epoch": 0.6617762857971735,
      "grad_norm": 0.4248839318752289,
      "learning_rate": 1.6913257905425009e-06,
      "loss": 0.7054,
      "step": 15968
    },
    {
      "epoch": 0.6618177297028472,
      "grad_norm": 0.40264639258384705,
      "learning_rate": 1.6911185710141327e-06,
      "loss": 0.6372,
      "step": 15969
    },
    {
      "epoch": 0.6618591736085209,
      "grad_norm": 0.44528117775917053,
      "learning_rate": 1.6909113514857643e-06,
      "loss": 0.6794,
      "step": 15970
    },
    {
      "epoch": 0.6619006175141945,
      "grad_norm": 0.4198397099971771,
      "learning_rate": 1.6907041319573959e-06,
      "loss": 0.6733,
      "step": 15971
    },
    {
      "epoch": 0.6619420614198682,
      "grad_norm": 0.4407023787498474,
      "learning_rate": 1.6904969124290275e-06,
      "loss": 0.7764,
      "step": 15972
    },
    {
      "epoch": 0.6619835053255418,
      "grad_norm": 0.4075586199760437,
      "learning_rate": 1.690289692900659e-06,
      "loss": 0.6359,
      "step": 15973
    },
    {
      "epoch": 0.6620249492312156,
      "grad_norm": 0.427508682012558,
      "learning_rate": 1.6900824733722906e-06,
      "loss": 0.7397,
      "step": 15974
    },
    {
      "epoch": 0.6620663931368892,
      "grad_norm": 0.38765883445739746,
      "learning_rate": 1.6898752538439222e-06,
      "loss": 0.6344,
      "step": 15975
    },
    {
      "epoch": 0.6621078370425629,
      "grad_norm": 0.38913047313690186,
      "learning_rate": 1.689668034315554e-06,
      "loss": 0.6936,
      "step": 15976
    },
    {
      "epoch": 0.6621492809482366,
      "grad_norm": 0.3855162560939789,
      "learning_rate": 1.6894608147871859e-06,
      "loss": 0.7083,
      "step": 15977
    },
    {
      "epoch": 0.6621907248539103,
      "grad_norm": 0.42753905057907104,
      "learning_rate": 1.6892535952588175e-06,
      "loss": 0.7003,
      "step": 15978
    },
    {
      "epoch": 0.6622321687595839,
      "grad_norm": 0.35751551389694214,
      "learning_rate": 1.689046375730449e-06,
      "loss": 0.652,
      "step": 15979
    },
    {
      "epoch": 0.6622736126652575,
      "grad_norm": 0.4125056266784668,
      "learning_rate": 1.6888391562020807e-06,
      "loss": 0.679,
      "step": 15980
    },
    {
      "epoch": 0.6623150565709313,
      "grad_norm": 0.4588261544704437,
      "learning_rate": 1.6886319366737123e-06,
      "loss": 0.6451,
      "step": 15981
    },
    {
      "epoch": 0.6623565004766049,
      "grad_norm": 0.4142706096172333,
      "learning_rate": 1.6884247171453439e-06,
      "loss": 0.6566,
      "step": 15982
    },
    {
      "epoch": 0.6623979443822786,
      "grad_norm": 0.4337237477302551,
      "learning_rate": 1.6882174976169754e-06,
      "loss": 0.6987,
      "step": 15983
    },
    {
      "epoch": 0.6624393882879522,
      "grad_norm": 0.37395796179771423,
      "learning_rate": 1.6880102780886073e-06,
      "loss": 0.6373,
      "step": 15984
    },
    {
      "epoch": 0.662480832193626,
      "grad_norm": 0.4059516489505768,
      "learning_rate": 1.6878030585602389e-06,
      "loss": 0.6509,
      "step": 15985
    },
    {
      "epoch": 0.6625222760992996,
      "grad_norm": 0.4157456159591675,
      "learning_rate": 1.6875958390318707e-06,
      "loss": 0.6637,
      "step": 15986
    },
    {
      "epoch": 0.6625637200049733,
      "grad_norm": 0.43141451478004456,
      "learning_rate": 1.6873886195035023e-06,
      "loss": 0.678,
      "step": 15987
    },
    {
      "epoch": 0.6626051639106469,
      "grad_norm": 0.46040433645248413,
      "learning_rate": 1.6871813999751339e-06,
      "loss": 0.7114,
      "step": 15988
    },
    {
      "epoch": 0.6626466078163206,
      "grad_norm": 0.3955116271972656,
      "learning_rate": 1.6869741804467655e-06,
      "loss": 0.6833,
      "step": 15989
    },
    {
      "epoch": 0.6626880517219943,
      "grad_norm": 0.4257911443710327,
      "learning_rate": 1.686766960918397e-06,
      "loss": 0.6278,
      "step": 15990
    },
    {
      "epoch": 0.6627294956276679,
      "grad_norm": 0.4683906137943268,
      "learning_rate": 1.6865597413900287e-06,
      "loss": 0.6965,
      "step": 15991
    },
    {
      "epoch": 0.6627709395333417,
      "grad_norm": 0.3924520015716553,
      "learning_rate": 1.6863525218616602e-06,
      "loss": 0.6294,
      "step": 15992
    },
    {
      "epoch": 0.6628123834390153,
      "grad_norm": 0.4638044238090515,
      "learning_rate": 1.686145302333292e-06,
      "loss": 0.6873,
      "step": 15993
    },
    {
      "epoch": 0.662853827344689,
      "grad_norm": 0.4742959439754486,
      "learning_rate": 1.6859380828049239e-06,
      "loss": 0.7397,
      "step": 15994
    },
    {
      "epoch": 0.6628952712503626,
      "grad_norm": 0.3841947019100189,
      "learning_rate": 1.6857308632765555e-06,
      "loss": 0.6431,
      "step": 15995
    },
    {
      "epoch": 0.6629367151560364,
      "grad_norm": 0.4122883081436157,
      "learning_rate": 1.685523643748187e-06,
      "loss": 0.6649,
      "step": 15996
    },
    {
      "epoch": 0.66297815906171,
      "grad_norm": 0.41568756103515625,
      "learning_rate": 1.6853164242198187e-06,
      "loss": 0.6763,
      "step": 15997
    },
    {
      "epoch": 0.6630196029673836,
      "grad_norm": 0.3982507884502411,
      "learning_rate": 1.6851092046914503e-06,
      "loss": 0.6459,
      "step": 15998
    },
    {
      "epoch": 0.6630610468730573,
      "grad_norm": 0.4323634207248688,
      "learning_rate": 1.6849019851630819e-06,
      "loss": 0.7063,
      "step": 15999
    },
    {
      "epoch": 0.663102490778731,
      "grad_norm": 0.4259624183177948,
      "learning_rate": 1.6846947656347135e-06,
      "loss": 0.687,
      "step": 16000
    },
    {
      "epoch": 0.6631439346844047,
      "grad_norm": 0.4316518306732178,
      "learning_rate": 1.684487546106345e-06,
      "loss": 0.6763,
      "step": 16001
    },
    {
      "epoch": 0.6631853785900783,
      "grad_norm": 0.403502494096756,
      "learning_rate": 1.684280326577977e-06,
      "loss": 0.6892,
      "step": 16002
    },
    {
      "epoch": 0.663226822495752,
      "grad_norm": 0.42343929409980774,
      "learning_rate": 1.6840731070496087e-06,
      "loss": 0.656,
      "step": 16003
    },
    {
      "epoch": 0.6632682664014257,
      "grad_norm": 0.42163679003715515,
      "learning_rate": 1.6838658875212403e-06,
      "loss": 0.731,
      "step": 16004
    },
    {
      "epoch": 0.6633097103070993,
      "grad_norm": 0.381375253200531,
      "learning_rate": 1.6836586679928719e-06,
      "loss": 0.6681,
      "step": 16005
    },
    {
      "epoch": 0.663351154212773,
      "grad_norm": 0.4233826696872711,
      "learning_rate": 1.6834514484645035e-06,
      "loss": 0.6768,
      "step": 16006
    },
    {
      "epoch": 0.6633925981184466,
      "grad_norm": 0.42052093148231506,
      "learning_rate": 1.683244228936135e-06,
      "loss": 0.7349,
      "step": 16007
    },
    {
      "epoch": 0.6634340420241204,
      "grad_norm": 0.43697795271873474,
      "learning_rate": 1.6830370094077667e-06,
      "loss": 0.7327,
      "step": 16008
    },
    {
      "epoch": 0.663475485929794,
      "grad_norm": 0.43963536620140076,
      "learning_rate": 1.6828297898793983e-06,
      "loss": 0.6729,
      "step": 16009
    },
    {
      "epoch": 0.6635169298354677,
      "grad_norm": 0.41651421785354614,
      "learning_rate": 1.6826225703510298e-06,
      "loss": 0.6672,
      "step": 16010
    },
    {
      "epoch": 0.6635583737411413,
      "grad_norm": 0.4033149480819702,
      "learning_rate": 1.6824153508226619e-06,
      "loss": 0.6641,
      "step": 16011
    },
    {
      "epoch": 0.6635998176468151,
      "grad_norm": 0.40479135513305664,
      "learning_rate": 1.6822081312942935e-06,
      "loss": 0.6951,
      "step": 16012
    },
    {
      "epoch": 0.6636412615524887,
      "grad_norm": 0.395394504070282,
      "learning_rate": 1.682000911765925e-06,
      "loss": 0.6714,
      "step": 16013
    },
    {
      "epoch": 0.6636827054581623,
      "grad_norm": 0.41213956475257874,
      "learning_rate": 1.6817936922375567e-06,
      "loss": 0.6383,
      "step": 16014
    },
    {
      "epoch": 0.6637241493638361,
      "grad_norm": 0.42876607179641724,
      "learning_rate": 1.6815864727091883e-06,
      "loss": 0.6636,
      "step": 16015
    },
    {
      "epoch": 0.6637655932695097,
      "grad_norm": 0.40472525358200073,
      "learning_rate": 1.6813792531808199e-06,
      "loss": 0.6509,
      "step": 16016
    },
    {
      "epoch": 0.6638070371751834,
      "grad_norm": 0.4018842875957489,
      "learning_rate": 1.6811720336524515e-06,
      "loss": 0.656,
      "step": 16017
    },
    {
      "epoch": 0.663848481080857,
      "grad_norm": 0.45379096269607544,
      "learning_rate": 1.680964814124083e-06,
      "loss": 0.6395,
      "step": 16018
    },
    {
      "epoch": 0.6638899249865308,
      "grad_norm": 0.4383485019207001,
      "learning_rate": 1.680757594595715e-06,
      "loss": 0.7437,
      "step": 16019
    },
    {
      "epoch": 0.6639313688922044,
      "grad_norm": 0.43282946944236755,
      "learning_rate": 1.6805503750673467e-06,
      "loss": 0.6921,
      "step": 16020
    },
    {
      "epoch": 0.6639728127978781,
      "grad_norm": 0.4040639102458954,
      "learning_rate": 1.6803431555389783e-06,
      "loss": 0.6621,
      "step": 16021
    },
    {
      "epoch": 0.6640142567035517,
      "grad_norm": 0.411870539188385,
      "learning_rate": 1.6801359360106099e-06,
      "loss": 0.6697,
      "step": 16022
    },
    {
      "epoch": 0.6640557006092254,
      "grad_norm": 0.42130282521247864,
      "learning_rate": 1.6799287164822415e-06,
      "loss": 0.6752,
      "step": 16023
    },
    {
      "epoch": 0.6640971445148991,
      "grad_norm": 0.3882516622543335,
      "learning_rate": 1.679721496953873e-06,
      "loss": 0.6571,
      "step": 16024
    },
    {
      "epoch": 0.6641385884205727,
      "grad_norm": 0.40047913789749146,
      "learning_rate": 1.6795142774255047e-06,
      "loss": 0.6327,
      "step": 16025
    },
    {
      "epoch": 0.6641800323262465,
      "grad_norm": 0.40959978103637695,
      "learning_rate": 1.6793070578971363e-06,
      "loss": 0.6689,
      "step": 16026
    },
    {
      "epoch": 0.6642214762319201,
      "grad_norm": 0.4312475025653839,
      "learning_rate": 1.6790998383687679e-06,
      "loss": 0.6646,
      "step": 16027
    },
    {
      "epoch": 0.6642629201375938,
      "grad_norm": 0.38784950971603394,
      "learning_rate": 1.6788926188403999e-06,
      "loss": 0.6848,
      "step": 16028
    },
    {
      "epoch": 0.6643043640432674,
      "grad_norm": 0.3920278251171112,
      "learning_rate": 1.6786853993120315e-06,
      "loss": 0.6897,
      "step": 16029
    },
    {
      "epoch": 0.6643458079489412,
      "grad_norm": 0.4259209930896759,
      "learning_rate": 1.678478179783663e-06,
      "loss": 0.7073,
      "step": 16030
    },
    {
      "epoch": 0.6643872518546148,
      "grad_norm": 0.42099717259407043,
      "learning_rate": 1.6782709602552947e-06,
      "loss": 0.7007,
      "step": 16031
    },
    {
      "epoch": 0.6644286957602884,
      "grad_norm": 0.4049660265445709,
      "learning_rate": 1.6780637407269263e-06,
      "loss": 0.7053,
      "step": 16032
    },
    {
      "epoch": 0.6644701396659621,
      "grad_norm": 0.3923211991786957,
      "learning_rate": 1.6778565211985579e-06,
      "loss": 0.6736,
      "step": 16033
    },
    {
      "epoch": 0.6645115835716358,
      "grad_norm": 0.4169822931289673,
      "learning_rate": 1.6776493016701895e-06,
      "loss": 0.6494,
      "step": 16034
    },
    {
      "epoch": 0.6645530274773095,
      "grad_norm": 0.42023107409477234,
      "learning_rate": 1.677442082141821e-06,
      "loss": 0.7021,
      "step": 16035
    },
    {
      "epoch": 0.6645944713829831,
      "grad_norm": 0.4183354377746582,
      "learning_rate": 1.6772348626134527e-06,
      "loss": 0.6786,
      "step": 16036
    },
    {
      "epoch": 0.6646359152886568,
      "grad_norm": 0.3896489441394806,
      "learning_rate": 1.6770276430850847e-06,
      "loss": 0.6848,
      "step": 16037
    },
    {
      "epoch": 0.6646773591943305,
      "grad_norm": 0.4182365834712982,
      "learning_rate": 1.6768204235567163e-06,
      "loss": 0.665,
      "step": 16038
    },
    {
      "epoch": 0.6647188031000042,
      "grad_norm": 0.4500863254070282,
      "learning_rate": 1.6766132040283479e-06,
      "loss": 0.7019,
      "step": 16039
    },
    {
      "epoch": 0.6647602470056778,
      "grad_norm": 0.40082263946533203,
      "learning_rate": 1.6764059844999795e-06,
      "loss": 0.7444,
      "step": 16040
    },
    {
      "epoch": 0.6648016909113514,
      "grad_norm": 0.4252960681915283,
      "learning_rate": 1.676198764971611e-06,
      "loss": 0.6992,
      "step": 16041
    },
    {
      "epoch": 0.6648431348170252,
      "grad_norm": 0.45064249634742737,
      "learning_rate": 1.6759915454432427e-06,
      "loss": 0.6973,
      "step": 16042
    },
    {
      "epoch": 0.6648845787226988,
      "grad_norm": 0.3930206000804901,
      "learning_rate": 1.6757843259148743e-06,
      "loss": 0.679,
      "step": 16043
    },
    {
      "epoch": 0.6649260226283725,
      "grad_norm": 0.3674457371234894,
      "learning_rate": 1.6755771063865059e-06,
      "loss": 0.6426,
      "step": 16044
    },
    {
      "epoch": 0.6649674665340461,
      "grad_norm": 0.4044419527053833,
      "learning_rate": 1.6753698868581379e-06,
      "loss": 0.6428,
      "step": 16045
    },
    {
      "epoch": 0.6650089104397199,
      "grad_norm": 0.43544474244117737,
      "learning_rate": 1.6751626673297695e-06,
      "loss": 0.7529,
      "step": 16046
    },
    {
      "epoch": 0.6650503543453935,
      "grad_norm": 0.40776705741882324,
      "learning_rate": 1.674955447801401e-06,
      "loss": 0.7068,
      "step": 16047
    },
    {
      "epoch": 0.6650917982510672,
      "grad_norm": 0.42460039258003235,
      "learning_rate": 1.6747482282730327e-06,
      "loss": 0.7185,
      "step": 16048
    },
    {
      "epoch": 0.6651332421567409,
      "grad_norm": 0.43609604239463806,
      "learning_rate": 1.6745410087446643e-06,
      "loss": 0.7278,
      "step": 16049
    },
    {
      "epoch": 0.6651746860624145,
      "grad_norm": 0.41036221385002136,
      "learning_rate": 1.6743337892162959e-06,
      "loss": 0.6145,
      "step": 16050
    },
    {
      "epoch": 0.6652161299680882,
      "grad_norm": 0.4103671610355377,
      "learning_rate": 1.6741265696879275e-06,
      "loss": 0.6915,
      "step": 16051
    },
    {
      "epoch": 0.6652575738737618,
      "grad_norm": 0.45543572306632996,
      "learning_rate": 1.673919350159559e-06,
      "loss": 0.7053,
      "step": 16052
    },
    {
      "epoch": 0.6652990177794356,
      "grad_norm": 0.41510987281799316,
      "learning_rate": 1.6737121306311907e-06,
      "loss": 0.6608,
      "step": 16053
    },
    {
      "epoch": 0.6653404616851092,
      "grad_norm": 0.4028263986110687,
      "learning_rate": 1.6735049111028227e-06,
      "loss": 0.6244,
      "step": 16054
    },
    {
      "epoch": 0.6653819055907829,
      "grad_norm": 0.43574294447898865,
      "learning_rate": 1.6732976915744543e-06,
      "loss": 0.7166,
      "step": 16055
    },
    {
      "epoch": 0.6654233494964565,
      "grad_norm": 0.40508994460105896,
      "learning_rate": 1.6730904720460859e-06,
      "loss": 0.6774,
      "step": 16056
    },
    {
      "epoch": 0.6654647934021302,
      "grad_norm": 0.4149644374847412,
      "learning_rate": 1.6728832525177175e-06,
      "loss": 0.6564,
      "step": 16057
    },
    {
      "epoch": 0.6655062373078039,
      "grad_norm": 0.4047861099243164,
      "learning_rate": 1.672676032989349e-06,
      "loss": 0.6609,
      "step": 16058
    },
    {
      "epoch": 0.6655476812134775,
      "grad_norm": 0.39124172925949097,
      "learning_rate": 1.6724688134609807e-06,
      "loss": 0.6243,
      "step": 16059
    },
    {
      "epoch": 0.6655891251191512,
      "grad_norm": 0.4387703537940979,
      "learning_rate": 1.6722615939326123e-06,
      "loss": 0.729,
      "step": 16060
    },
    {
      "epoch": 0.6656305690248249,
      "grad_norm": 0.39764660596847534,
      "learning_rate": 1.6720543744042439e-06,
      "loss": 0.6512,
      "step": 16061
    },
    {
      "epoch": 0.6656720129304986,
      "grad_norm": 0.4250905215740204,
      "learning_rate": 1.6718471548758755e-06,
      "loss": 0.7319,
      "step": 16062
    },
    {
      "epoch": 0.6657134568361722,
      "grad_norm": 0.42997339367866516,
      "learning_rate": 1.6716399353475075e-06,
      "loss": 0.6418,
      "step": 16063
    },
    {
      "epoch": 0.665754900741846,
      "grad_norm": 0.38097551465034485,
      "learning_rate": 1.671432715819139e-06,
      "loss": 0.6078,
      "step": 16064
    },
    {
      "epoch": 0.6657963446475196,
      "grad_norm": 0.42145785689353943,
      "learning_rate": 1.6712254962907707e-06,
      "loss": 0.7346,
      "step": 16065
    },
    {
      "epoch": 0.6658377885531932,
      "grad_norm": 0.39531630277633667,
      "learning_rate": 1.6710182767624023e-06,
      "loss": 0.6514,
      "step": 16066
    },
    {
      "epoch": 0.6658792324588669,
      "grad_norm": 0.4041268229484558,
      "learning_rate": 1.6708110572340339e-06,
      "loss": 0.6733,
      "step": 16067
    },
    {
      "epoch": 0.6659206763645406,
      "grad_norm": 0.4416980445384979,
      "learning_rate": 1.6706038377056655e-06,
      "loss": 0.6494,
      "step": 16068
    },
    {
      "epoch": 0.6659621202702143,
      "grad_norm": 0.4193469285964966,
      "learning_rate": 1.670396618177297e-06,
      "loss": 0.6912,
      "step": 16069
    },
    {
      "epoch": 0.6660035641758879,
      "grad_norm": 0.4195697605609894,
      "learning_rate": 1.6701893986489287e-06,
      "loss": 0.7168,
      "step": 16070
    },
    {
      "epoch": 0.6660450080815616,
      "grad_norm": 0.41283613443374634,
      "learning_rate": 1.6699821791205603e-06,
      "loss": 0.7352,
      "step": 16071
    },
    {
      "epoch": 0.6660864519872353,
      "grad_norm": 0.4048596918582916,
      "learning_rate": 1.6697749595921923e-06,
      "loss": 0.6902,
      "step": 16072
    },
    {
      "epoch": 0.666127895892909,
      "grad_norm": 0.39398664236068726,
      "learning_rate": 1.6695677400638239e-06,
      "loss": 0.7106,
      "step": 16073
    },
    {
      "epoch": 0.6661693397985826,
      "grad_norm": 0.3825371265411377,
      "learning_rate": 1.6693605205354555e-06,
      "loss": 0.6458,
      "step": 16074
    },
    {
      "epoch": 0.6662107837042562,
      "grad_norm": 0.45259997248649597,
      "learning_rate": 1.669153301007087e-06,
      "loss": 0.7255,
      "step": 16075
    },
    {
      "epoch": 0.66625222760993,
      "grad_norm": 0.41928672790527344,
      "learning_rate": 1.6689460814787187e-06,
      "loss": 0.6669,
      "step": 16076
    },
    {
      "epoch": 0.6662936715156036,
      "grad_norm": 0.408802330493927,
      "learning_rate": 1.6687388619503503e-06,
      "loss": 0.6746,
      "step": 16077
    },
    {
      "epoch": 0.6663351154212773,
      "grad_norm": 0.4325581192970276,
      "learning_rate": 1.6685316424219819e-06,
      "loss": 0.724,
      "step": 16078
    },
    {
      "epoch": 0.666376559326951,
      "grad_norm": 0.3957660496234894,
      "learning_rate": 1.6683244228936135e-06,
      "loss": 0.7019,
      "step": 16079
    },
    {
      "epoch": 0.6664180032326247,
      "grad_norm": 0.4133421778678894,
      "learning_rate": 1.6681172033652455e-06,
      "loss": 0.6909,
      "step": 16080
    },
    {
      "epoch": 0.6664594471382983,
      "grad_norm": 0.4407144784927368,
      "learning_rate": 1.667909983836877e-06,
      "loss": 0.7502,
      "step": 16081
    },
    {
      "epoch": 0.666500891043972,
      "grad_norm": 0.4287929832935333,
      "learning_rate": 1.6677027643085087e-06,
      "loss": 0.752,
      "step": 16082
    },
    {
      "epoch": 0.6665423349496457,
      "grad_norm": 0.3846110999584198,
      "learning_rate": 1.6674955447801403e-06,
      "loss": 0.6394,
      "step": 16083
    },
    {
      "epoch": 0.6665837788553193,
      "grad_norm": 0.4137878715991974,
      "learning_rate": 1.6672883252517719e-06,
      "loss": 0.6752,
      "step": 16084
    },
    {
      "epoch": 0.666625222760993,
      "grad_norm": 0.4137948751449585,
      "learning_rate": 1.6670811057234035e-06,
      "loss": 0.6696,
      "step": 16085
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.41198962926864624,
      "learning_rate": 1.666873886195035e-06,
      "loss": 0.72,
      "step": 16086
    },
    {
      "checkpoint_runtime": 206.5332
    },
    {
      "epoch": 0.6667081105723404,
      "grad_norm": 0.3782370984554291,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.6387,
      "step": 16087
    },
    {
      "epoch": 0.666749554478014,
      "grad_norm": 0.42298710346221924,
      "learning_rate": 1.6664594471382983e-06,
      "loss": 0.6467,
      "step": 16088
    },
    {
      "epoch": 0.6667909983836877,
      "grad_norm": 0.41521352529525757,
      "learning_rate": 1.6662522276099303e-06,
      "loss": 0.6682,
      "step": 16089
    },
    {
      "epoch": 0.6668324422893613,
      "grad_norm": 0.4355306029319763,
      "learning_rate": 1.6660450080815619e-06,
      "loss": 0.7256,
      "step": 16090
    },
    {
      "epoch": 0.6668738861950351,
      "grad_norm": 0.45535486936569214,
      "learning_rate": 1.6658377885531935e-06,
      "loss": 0.6934,
      "step": 16091
    },
    {
      "epoch": 0.6669153301007087,
      "grad_norm": 0.3881808817386627,
      "learning_rate": 1.665630569024825e-06,
      "loss": 0.6741,
      "step": 16092
    },
    {
      "epoch": 0.6669567740063823,
      "grad_norm": 0.41817671060562134,
      "learning_rate": 1.6654233494964567e-06,
      "loss": 0.7003,
      "step": 16093
    },
    {
      "epoch": 0.666998217912056,
      "grad_norm": 0.4169234335422516,
      "learning_rate": 1.6652161299680883e-06,
      "loss": 0.6702,
      "step": 16094
    },
    {
      "epoch": 0.6670396618177297,
      "grad_norm": 0.3980676531791687,
      "learning_rate": 1.6650089104397199e-06,
      "loss": 0.662,
      "step": 16095
    },
    {
      "epoch": 0.6670811057234034,
      "grad_norm": 0.45285820960998535,
      "learning_rate": 1.6648016909113515e-06,
      "loss": 0.6627,
      "step": 16096
    },
    {
      "epoch": 0.667122549629077,
      "grad_norm": 0.4042116701602936,
      "learning_rate": 1.664594471382983e-06,
      "loss": 0.705,
      "step": 16097
    },
    {
      "epoch": 0.6671639935347508,
      "grad_norm": 0.4135550856590271,
      "learning_rate": 1.664387251854615e-06,
      "loss": 0.6527,
      "step": 16098
    },
    {
      "epoch": 0.6672054374404244,
      "grad_norm": 0.3868297338485718,
      "learning_rate": 1.6641800323262467e-06,
      "loss": 0.6555,
      "step": 16099
    },
    {
      "epoch": 0.6672468813460981,
      "grad_norm": 0.3735204041004181,
      "learning_rate": 1.6639728127978783e-06,
      "loss": 0.6294,
      "step": 16100
    },
    {
      "epoch": 0.6672883252517717,
      "grad_norm": 0.42280200123786926,
      "learning_rate": 1.6637655932695099e-06,
      "loss": 0.6832,
      "step": 16101
    },
    {
      "epoch": 0.6673297691574454,
      "grad_norm": 0.46520352363586426,
      "learning_rate": 1.6635583737411415e-06,
      "loss": 0.72,
      "step": 16102
    },
    {
      "epoch": 0.6673712130631191,
      "grad_norm": 0.4183734357357025,
      "learning_rate": 1.663351154212773e-06,
      "loss": 0.6389,
      "step": 16103
    },
    {
      "epoch": 0.6674126569687927,
      "grad_norm": 0.42776432633399963,
      "learning_rate": 1.6631439346844047e-06,
      "loss": 0.6499,
      "step": 16104
    },
    {
      "epoch": 0.6674541008744664,
      "grad_norm": 0.3861633837223053,
      "learning_rate": 1.6629367151560363e-06,
      "loss": 0.6812,
      "step": 16105
    },
    {
      "epoch": 0.6674955447801401,
      "grad_norm": 0.40811535716056824,
      "learning_rate": 1.6627294956276683e-06,
      "loss": 0.6699,
      "step": 16106
    },
    {
      "epoch": 0.6675369886858138,
      "grad_norm": 0.42259132862091064,
      "learning_rate": 1.6625222760992999e-06,
      "loss": 0.6531,
      "step": 16107
    },
    {
      "epoch": 0.6675784325914874,
      "grad_norm": 0.4174489974975586,
      "learning_rate": 1.6623150565709315e-06,
      "loss": 0.6829,
      "step": 16108
    },
    {
      "epoch": 0.6676198764971611,
      "grad_norm": 0.4301969110965729,
      "learning_rate": 1.662107837042563e-06,
      "loss": 0.6689,
      "step": 16109
    },
    {
      "epoch": 0.6676613204028348,
      "grad_norm": 0.4433270990848541,
      "learning_rate": 1.6619006175141947e-06,
      "loss": 0.7388,
      "step": 16110
    },
    {
      "epoch": 0.6677027643085084,
      "grad_norm": 0.4464190900325775,
      "learning_rate": 1.6616933979858263e-06,
      "loss": 0.6688,
      "step": 16111
    },
    {
      "epoch": 0.6677442082141821,
      "grad_norm": 0.44650769233703613,
      "learning_rate": 1.6614861784574579e-06,
      "loss": 0.7051,
      "step": 16112
    },
    {
      "epoch": 0.6677856521198557,
      "grad_norm": 0.41957297921180725,
      "learning_rate": 1.6612789589290895e-06,
      "loss": 0.6997,
      "step": 16113
    },
    {
      "epoch": 0.6678270960255295,
      "grad_norm": 0.43037837743759155,
      "learning_rate": 1.661071739400721e-06,
      "loss": 0.6857,
      "step": 16114
    },
    {
      "epoch": 0.6678685399312031,
      "grad_norm": 0.435438334941864,
      "learning_rate": 1.660864519872353e-06,
      "loss": 0.7158,
      "step": 16115
    },
    {
      "epoch": 0.6679099838368768,
      "grad_norm": 0.4308931529521942,
      "learning_rate": 1.6606573003439847e-06,
      "loss": 0.6692,
      "step": 16116
    },
    {
      "epoch": 0.6679514277425505,
      "grad_norm": 0.42331191897392273,
      "learning_rate": 1.6604500808156163e-06,
      "loss": 0.6602,
      "step": 16117
    },
    {
      "epoch": 0.6679928716482241,
      "grad_norm": 0.4257618188858032,
      "learning_rate": 1.6602428612872479e-06,
      "loss": 0.7251,
      "step": 16118
    },
    {
      "epoch": 0.6680343155538978,
      "grad_norm": 0.41592153906822205,
      "learning_rate": 1.6600356417588795e-06,
      "loss": 0.6929,
      "step": 16119
    },
    {
      "epoch": 0.6680757594595714,
      "grad_norm": 0.45938268303871155,
      "learning_rate": 1.659828422230511e-06,
      "loss": 0.7739,
      "step": 16120
    },
    {
      "epoch": 0.6681172033652452,
      "grad_norm": 0.44558459520339966,
      "learning_rate": 1.6596212027021427e-06,
      "loss": 0.6475,
      "step": 16121
    },
    {
      "epoch": 0.6681586472709188,
      "grad_norm": 0.39787960052490234,
      "learning_rate": 1.6594139831737743e-06,
      "loss": 0.6838,
      "step": 16122
    },
    {
      "epoch": 0.6682000911765925,
      "grad_norm": 0.4259253740310669,
      "learning_rate": 1.6592067636454059e-06,
      "loss": 0.7268,
      "step": 16123
    },
    {
      "epoch": 0.6682415350822661,
      "grad_norm": 0.40844789147377014,
      "learning_rate": 1.6589995441170379e-06,
      "loss": 0.6992,
      "step": 16124
    },
    {
      "epoch": 0.6682829789879399,
      "grad_norm": 0.4130089581012726,
      "learning_rate": 1.6587923245886695e-06,
      "loss": 0.6613,
      "step": 16125
    },
    {
      "epoch": 0.6683244228936135,
      "grad_norm": 0.40782657265663147,
      "learning_rate": 1.658585105060301e-06,
      "loss": 0.7076,
      "step": 16126
    },
    {
      "epoch": 0.6683658667992871,
      "grad_norm": 0.3920920491218567,
      "learning_rate": 1.6583778855319327e-06,
      "loss": 0.6813,
      "step": 16127
    },
    {
      "epoch": 0.6684073107049608,
      "grad_norm": 0.40434354543685913,
      "learning_rate": 1.6581706660035643e-06,
      "loss": 0.6547,
      "step": 16128
    },
    {
      "epoch": 0.6684487546106345,
      "grad_norm": 0.38019701838493347,
      "learning_rate": 1.6579634464751959e-06,
      "loss": 0.587,
      "step": 16129
    },
    {
      "epoch": 0.6684901985163082,
      "grad_norm": 0.4412388801574707,
      "learning_rate": 1.6577562269468275e-06,
      "loss": 0.6936,
      "step": 16130
    },
    {
      "epoch": 0.6685316424219818,
      "grad_norm": 0.3654560148715973,
      "learning_rate": 1.657549007418459e-06,
      "loss": 0.6685,
      "step": 16131
    },
    {
      "epoch": 0.6685730863276556,
      "grad_norm": 0.4146740734577179,
      "learning_rate": 1.657341787890091e-06,
      "loss": 0.6553,
      "step": 16132
    },
    {
      "epoch": 0.6686145302333292,
      "grad_norm": 0.4086534082889557,
      "learning_rate": 1.6571345683617227e-06,
      "loss": 0.6763,
      "step": 16133
    },
    {
      "epoch": 0.6686559741390029,
      "grad_norm": 0.43128788471221924,
      "learning_rate": 1.6569273488333543e-06,
      "loss": 0.6589,
      "step": 16134
    },
    {
      "epoch": 0.6686974180446765,
      "grad_norm": 0.4087723195552826,
      "learning_rate": 1.6567201293049859e-06,
      "loss": 0.6735,
      "step": 16135
    },
    {
      "epoch": 0.6687388619503501,
      "grad_norm": 0.43764543533325195,
      "learning_rate": 1.6565129097766175e-06,
      "loss": 0.7008,
      "step": 16136
    },
    {
      "epoch": 0.6687803058560239,
      "grad_norm": 0.4277664124965668,
      "learning_rate": 1.656305690248249e-06,
      "loss": 0.6943,
      "step": 16137
    },
    {
      "epoch": 0.6688217497616975,
      "grad_norm": 0.38755127787590027,
      "learning_rate": 1.6560984707198807e-06,
      "loss": 0.6885,
      "step": 16138
    },
    {
      "epoch": 0.6688631936673712,
      "grad_norm": 0.4323432445526123,
      "learning_rate": 1.6558912511915123e-06,
      "loss": 0.708,
      "step": 16139
    },
    {
      "epoch": 0.6689046375730449,
      "grad_norm": 0.390971302986145,
      "learning_rate": 1.6556840316631439e-06,
      "loss": 0.6669,
      "step": 16140
    },
    {
      "epoch": 0.6689460814787186,
      "grad_norm": 0.39258337020874023,
      "learning_rate": 1.6554768121347759e-06,
      "loss": 0.6962,
      "step": 16141
    },
    {
      "epoch": 0.6689875253843922,
      "grad_norm": 0.39720359444618225,
      "learning_rate": 1.6552695926064075e-06,
      "loss": 0.677,
      "step": 16142
    },
    {
      "epoch": 0.669028969290066,
      "grad_norm": 0.42048683762550354,
      "learning_rate": 1.655062373078039e-06,
      "loss": 0.6591,
      "step": 16143
    },
    {
      "epoch": 0.6690704131957396,
      "grad_norm": 0.3739628493785858,
      "learning_rate": 1.6548551535496707e-06,
      "loss": 0.6371,
      "step": 16144
    },
    {
      "epoch": 0.6691118571014132,
      "grad_norm": 0.4006481468677521,
      "learning_rate": 1.6546479340213023e-06,
      "loss": 0.6851,
      "step": 16145
    },
    {
      "epoch": 0.6691533010070869,
      "grad_norm": 0.3969648778438568,
      "learning_rate": 1.6544407144929339e-06,
      "loss": 0.624,
      "step": 16146
    },
    {
      "epoch": 0.6691947449127605,
      "grad_norm": 0.399095743894577,
      "learning_rate": 1.6542334949645655e-06,
      "loss": 0.6572,
      "step": 16147
    },
    {
      "epoch": 0.6692361888184343,
      "grad_norm": 0.4084377884864807,
      "learning_rate": 1.654026275436197e-06,
      "loss": 0.6627,
      "step": 16148
    },
    {
      "epoch": 0.6692776327241079,
      "grad_norm": 0.408769428730011,
      "learning_rate": 1.6538190559078287e-06,
      "loss": 0.6638,
      "step": 16149
    },
    {
      "epoch": 0.6693190766297816,
      "grad_norm": 0.3968919515609741,
      "learning_rate": 1.6536118363794607e-06,
      "loss": 0.6172,
      "step": 16150
    },
    {
      "epoch": 0.6693605205354553,
      "grad_norm": 0.40839144587516785,
      "learning_rate": 1.6534046168510923e-06,
      "loss": 0.6583,
      "step": 16151
    },
    {
      "epoch": 0.669401964441129,
      "grad_norm": 0.42067572474479675,
      "learning_rate": 1.6531973973227239e-06,
      "loss": 0.6727,
      "step": 16152
    },
    {
      "epoch": 0.6694434083468026,
      "grad_norm": 0.5413890480995178,
      "learning_rate": 1.6529901777943555e-06,
      "loss": 0.7273,
      "step": 16153
    },
    {
      "epoch": 0.6694848522524762,
      "grad_norm": 0.40024593472480774,
      "learning_rate": 1.652782958265987e-06,
      "loss": 0.661,
      "step": 16154
    },
    {
      "epoch": 0.66952629615815,
      "grad_norm": 0.4183686673641205,
      "learning_rate": 1.6525757387376187e-06,
      "loss": 0.6366,
      "step": 16155
    },
    {
      "epoch": 0.6695677400638236,
      "grad_norm": 0.37812820076942444,
      "learning_rate": 1.6523685192092503e-06,
      "loss": 0.6239,
      "step": 16156
    },
    {
      "epoch": 0.6696091839694973,
      "grad_norm": 0.4074612557888031,
      "learning_rate": 1.6521612996808819e-06,
      "loss": 0.6366,
      "step": 16157
    },
    {
      "epoch": 0.6696506278751709,
      "grad_norm": 0.4239770770072937,
      "learning_rate": 1.6519540801525137e-06,
      "loss": 0.6835,
      "step": 16158
    },
    {
      "epoch": 0.6696920717808447,
      "grad_norm": 0.4043435752391815,
      "learning_rate": 1.6517468606241455e-06,
      "loss": 0.6848,
      "step": 16159
    },
    {
      "epoch": 0.6697335156865183,
      "grad_norm": 0.433847039937973,
      "learning_rate": 1.651539641095777e-06,
      "loss": 0.626,
      "step": 16160
    },
    {
      "epoch": 0.669774959592192,
      "grad_norm": 0.4534812569618225,
      "learning_rate": 1.6513324215674087e-06,
      "loss": 0.7311,
      "step": 16161
    },
    {
      "epoch": 0.6698164034978656,
      "grad_norm": 0.4000806510448456,
      "learning_rate": 1.6511252020390403e-06,
      "loss": 0.7021,
      "step": 16162
    },
    {
      "epoch": 0.6698578474035393,
      "grad_norm": 0.4079470634460449,
      "learning_rate": 1.6509179825106719e-06,
      "loss": 0.6663,
      "step": 16163
    },
    {
      "epoch": 0.669899291309213,
      "grad_norm": 0.44350677728652954,
      "learning_rate": 1.6507107629823035e-06,
      "loss": 0.6385,
      "step": 16164
    },
    {
      "epoch": 0.6699407352148866,
      "grad_norm": 0.3970186710357666,
      "learning_rate": 1.650503543453935e-06,
      "loss": 0.6418,
      "step": 16165
    },
    {
      "epoch": 0.6699821791205604,
      "grad_norm": 0.44802772998809814,
      "learning_rate": 1.6502963239255667e-06,
      "loss": 0.7186,
      "step": 16166
    },
    {
      "epoch": 0.670023623026234,
      "grad_norm": 0.37707698345184326,
      "learning_rate": 1.6500891043971987e-06,
      "loss": 0.6653,
      "step": 16167
    },
    {
      "epoch": 0.6700650669319077,
      "grad_norm": 0.4406111240386963,
      "learning_rate": 1.6498818848688303e-06,
      "loss": 0.7322,
      "step": 16168
    },
    {
      "epoch": 0.6701065108375813,
      "grad_norm": 0.3813234865665436,
      "learning_rate": 1.6496746653404619e-06,
      "loss": 0.7019,
      "step": 16169
    },
    {
      "epoch": 0.6701479547432551,
      "grad_norm": 0.4211023151874542,
      "learning_rate": 1.6494674458120935e-06,
      "loss": 0.6459,
      "step": 16170
    },
    {
      "epoch": 0.6701893986489287,
      "grad_norm": 0.43281981348991394,
      "learning_rate": 1.649260226283725e-06,
      "loss": 0.6621,
      "step": 16171
    },
    {
      "epoch": 0.6702308425546023,
      "grad_norm": 0.41704243421554565,
      "learning_rate": 1.6490530067553567e-06,
      "loss": 0.7269,
      "step": 16172
    },
    {
      "epoch": 0.670272286460276,
      "grad_norm": 0.40744224190711975,
      "learning_rate": 1.6488457872269883e-06,
      "loss": 0.6689,
      "step": 16173
    },
    {
      "epoch": 0.6703137303659497,
      "grad_norm": 0.4072561264038086,
      "learning_rate": 1.6486385676986199e-06,
      "loss": 0.6731,
      "step": 16174
    },
    {
      "epoch": 0.6703551742716234,
      "grad_norm": 0.3993113338947296,
      "learning_rate": 1.6484313481702517e-06,
      "loss": 0.6678,
      "step": 16175
    },
    {
      "epoch": 0.670396618177297,
      "grad_norm": 0.40756484866142273,
      "learning_rate": 1.6482241286418835e-06,
      "loss": 0.7014,
      "step": 16176
    },
    {
      "epoch": 0.6704380620829707,
      "grad_norm": 0.3986349403858185,
      "learning_rate": 1.648016909113515e-06,
      "loss": 0.6455,
      "step": 16177
    },
    {
      "epoch": 0.6704795059886444,
      "grad_norm": 0.4195970594882965,
      "learning_rate": 1.6478096895851467e-06,
      "loss": 0.7222,
      "step": 16178
    },
    {
      "epoch": 0.670520949894318,
      "grad_norm": 0.4104878008365631,
      "learning_rate": 1.6476024700567783e-06,
      "loss": 0.6975,
      "step": 16179
    },
    {
      "epoch": 0.6705623937999917,
      "grad_norm": 0.4368089735507965,
      "learning_rate": 1.6473952505284099e-06,
      "loss": 0.6934,
      "step": 16180
    },
    {
      "epoch": 0.6706038377056653,
      "grad_norm": 0.36301952600479126,
      "learning_rate": 1.6471880310000415e-06,
      "loss": 0.6632,
      "step": 16181
    },
    {
      "epoch": 0.6706452816113391,
      "grad_norm": 0.4034351110458374,
      "learning_rate": 1.646980811471673e-06,
      "loss": 0.5869,
      "step": 16182
    },
    {
      "epoch": 0.6706867255170127,
      "grad_norm": 0.4272559583187103,
      "learning_rate": 1.6467735919433047e-06,
      "loss": 0.7006,
      "step": 16183
    },
    {
      "epoch": 0.6707281694226864,
      "grad_norm": 0.38148510456085205,
      "learning_rate": 1.6465663724149365e-06,
      "loss": 0.6191,
      "step": 16184
    },
    {
      "epoch": 0.67076961332836,
      "grad_norm": 0.43508726358413696,
      "learning_rate": 1.6463591528865683e-06,
      "loss": 0.6609,
      "step": 16185
    },
    {
      "epoch": 0.6708110572340338,
      "grad_norm": 0.4198850691318512,
      "learning_rate": 1.6461519333581999e-06,
      "loss": 0.6925,
      "step": 16186
    },
    {
      "epoch": 0.6708525011397074,
      "grad_norm": 0.38852623105049133,
      "learning_rate": 1.6459447138298315e-06,
      "loss": 0.6342,
      "step": 16187
    },
    {
      "epoch": 0.670893945045381,
      "grad_norm": 0.4669853746891022,
      "learning_rate": 1.645737494301463e-06,
      "loss": 0.7495,
      "step": 16188
    },
    {
      "epoch": 0.6709353889510548,
      "grad_norm": 0.3950718343257904,
      "learning_rate": 1.6455302747730947e-06,
      "loss": 0.6306,
      "step": 16189
    },
    {
      "epoch": 0.6709768328567284,
      "grad_norm": 0.4189806282520294,
      "learning_rate": 1.6453230552447263e-06,
      "loss": 0.6941,
      "step": 16190
    },
    {
      "epoch": 0.6710182767624021,
      "grad_norm": 0.4150543510913849,
      "learning_rate": 1.6451158357163579e-06,
      "loss": 0.6221,
      "step": 16191
    },
    {
      "epoch": 0.6710597206680757,
      "grad_norm": 0.40631791949272156,
      "learning_rate": 1.6449086161879897e-06,
      "loss": 0.6909,
      "step": 16192
    },
    {
      "epoch": 0.6711011645737495,
      "grad_norm": 0.4215424656867981,
      "learning_rate": 1.6447013966596215e-06,
      "loss": 0.7144,
      "step": 16193
    },
    {
      "epoch": 0.6711426084794231,
      "grad_norm": 0.39717501401901245,
      "learning_rate": 1.644494177131253e-06,
      "loss": 0.6669,
      "step": 16194
    },
    {
      "epoch": 0.6711840523850968,
      "grad_norm": 0.3882301449775696,
      "learning_rate": 1.6442869576028847e-06,
      "loss": 0.6549,
      "step": 16195
    },
    {
      "epoch": 0.6712254962907704,
      "grad_norm": 0.40179410576820374,
      "learning_rate": 1.6440797380745163e-06,
      "loss": 0.6177,
      "step": 16196
    },
    {
      "epoch": 0.6712669401964441,
      "grad_norm": 0.4116314947605133,
      "learning_rate": 1.6438725185461479e-06,
      "loss": 0.6688,
      "step": 16197
    },
    {
      "epoch": 0.6713083841021178,
      "grad_norm": 0.40976348519325256,
      "learning_rate": 1.6436652990177795e-06,
      "loss": 0.6497,
      "step": 16198
    },
    {
      "epoch": 0.6713498280077914,
      "grad_norm": 0.4308205246925354,
      "learning_rate": 1.643458079489411e-06,
      "loss": 0.6731,
      "step": 16199
    },
    {
      "epoch": 0.6713912719134651,
      "grad_norm": 0.42541491985321045,
      "learning_rate": 1.6432508599610427e-06,
      "loss": 0.688,
      "step": 16200
    },
    {
      "epoch": 0.6714327158191388,
      "grad_norm": 0.405973881483078,
      "learning_rate": 1.6430436404326745e-06,
      "loss": 0.6946,
      "step": 16201
    },
    {
      "epoch": 0.6714741597248125,
      "grad_norm": 0.41808444261550903,
      "learning_rate": 1.6428364209043063e-06,
      "loss": 0.6643,
      "step": 16202
    },
    {
      "epoch": 0.6715156036304861,
      "grad_norm": 0.42289915680885315,
      "learning_rate": 1.6426292013759379e-06,
      "loss": 0.6948,
      "step": 16203
    },
    {
      "epoch": 0.6715570475361599,
      "grad_norm": 0.38629665970802307,
      "learning_rate": 1.6424219818475695e-06,
      "loss": 0.6179,
      "step": 16204
    },
    {
      "epoch": 0.6715984914418335,
      "grad_norm": 0.4205452799797058,
      "learning_rate": 1.642214762319201e-06,
      "loss": 0.6248,
      "step": 16205
    },
    {
      "epoch": 0.6716399353475071,
      "grad_norm": 0.3938872218132019,
      "learning_rate": 1.6420075427908327e-06,
      "loss": 0.6367,
      "step": 16206
    },
    {
      "epoch": 0.6716813792531808,
      "grad_norm": 0.43419158458709717,
      "learning_rate": 1.6418003232624643e-06,
      "loss": 0.6558,
      "step": 16207
    },
    {
      "epoch": 0.6717228231588545,
      "grad_norm": 0.4543806314468384,
      "learning_rate": 1.6415931037340959e-06,
      "loss": 0.7085,
      "step": 16208
    },
    {
      "epoch": 0.6717642670645282,
      "grad_norm": 0.4216350317001343,
      "learning_rate": 1.6413858842057277e-06,
      "loss": 0.6613,
      "step": 16209
    },
    {
      "epoch": 0.6718057109702018,
      "grad_norm": 0.44536542892456055,
      "learning_rate": 1.6411786646773593e-06,
      "loss": 0.6895,
      "step": 16210
    },
    {
      "epoch": 0.6718471548758755,
      "grad_norm": 0.4421224296092987,
      "learning_rate": 1.640971445148991e-06,
      "loss": 0.678,
      "step": 16211
    },
    {
      "epoch": 0.6718885987815492,
      "grad_norm": 0.41273733973503113,
      "learning_rate": 1.6407642256206227e-06,
      "loss": 0.6544,
      "step": 16212
    },
    {
      "epoch": 0.6719300426872229,
      "grad_norm": 0.40923386812210083,
      "learning_rate": 1.6405570060922543e-06,
      "loss": 0.6619,
      "step": 16213
    },
    {
      "epoch": 0.6719714865928965,
      "grad_norm": 0.4133892059326172,
      "learning_rate": 1.6403497865638859e-06,
      "loss": 0.6826,
      "step": 16214
    },
    {
      "epoch": 0.6720129304985701,
      "grad_norm": 0.3920084536075592,
      "learning_rate": 1.6401425670355175e-06,
      "loss": 0.6458,
      "step": 16215
    },
    {
      "epoch": 0.6720543744042439,
      "grad_norm": 0.40323612093925476,
      "learning_rate": 1.639935347507149e-06,
      "loss": 0.635,
      "step": 16216
    },
    {
      "epoch": 0.6720958183099175,
      "grad_norm": 0.4171929657459259,
      "learning_rate": 1.6397281279787807e-06,
      "loss": 0.6707,
      "step": 16217
    },
    {
      "epoch": 0.6721372622155912,
      "grad_norm": 0.44549378752708435,
      "learning_rate": 1.6395209084504125e-06,
      "loss": 0.6748,
      "step": 16218
    },
    {
      "epoch": 0.6721787061212648,
      "grad_norm": 0.4428058862686157,
      "learning_rate": 1.639313688922044e-06,
      "loss": 0.683,
      "step": 16219
    },
    {
      "epoch": 0.6722201500269386,
      "grad_norm": 0.4401097297668457,
      "learning_rate": 1.6391064693936759e-06,
      "loss": 0.7292,
      "step": 16220
    },
    {
      "epoch": 0.6722615939326122,
      "grad_norm": 0.4201399087905884,
      "learning_rate": 1.6388992498653075e-06,
      "loss": 0.7075,
      "step": 16221
    },
    {
      "epoch": 0.6723030378382859,
      "grad_norm": 0.3983350396156311,
      "learning_rate": 1.638692030336939e-06,
      "loss": 0.6359,
      "step": 16222
    },
    {
      "epoch": 0.6723444817439596,
      "grad_norm": 0.4110371470451355,
      "learning_rate": 1.6384848108085707e-06,
      "loss": 0.6111,
      "step": 16223
    },
    {
      "epoch": 0.6723859256496332,
      "grad_norm": 0.4090495705604553,
      "learning_rate": 1.6382775912802023e-06,
      "loss": 0.6917,
      "step": 16224
    },
    {
      "epoch": 0.6724273695553069,
      "grad_norm": 0.4375980794429779,
      "learning_rate": 1.6380703717518339e-06,
      "loss": 0.656,
      "step": 16225
    },
    {
      "epoch": 0.6724688134609805,
      "grad_norm": 0.3786022663116455,
      "learning_rate": 1.6378631522234657e-06,
      "loss": 0.5862,
      "step": 16226
    },
    {
      "epoch": 0.6725102573666543,
      "grad_norm": 0.40136271715164185,
      "learning_rate": 1.6376559326950973e-06,
      "loss": 0.6282,
      "step": 16227
    },
    {
      "epoch": 0.6725517012723279,
      "grad_norm": 0.3963063061237335,
      "learning_rate": 1.637448713166729e-06,
      "loss": 0.6832,
      "step": 16228
    },
    {
      "epoch": 0.6725931451780016,
      "grad_norm": 0.44079920649528503,
      "learning_rate": 1.6372414936383607e-06,
      "loss": 0.6821,
      "step": 16229
    },
    {
      "epoch": 0.6726345890836752,
      "grad_norm": 0.38565924763679504,
      "learning_rate": 1.6370342741099923e-06,
      "loss": 0.6611,
      "step": 16230
    },
    {
      "epoch": 0.672676032989349,
      "grad_norm": 0.43100571632385254,
      "learning_rate": 1.6368270545816239e-06,
      "loss": 0.6582,
      "step": 16231
    },
    {
      "epoch": 0.6727174768950226,
      "grad_norm": 0.4247682988643646,
      "learning_rate": 1.6366198350532555e-06,
      "loss": 0.6597,
      "step": 16232
    },
    {
      "epoch": 0.6727589208006962,
      "grad_norm": 0.42204317450523376,
      "learning_rate": 1.636412615524887e-06,
      "loss": 0.6775,
      "step": 16233
    },
    {
      "epoch": 0.67280036470637,
      "grad_norm": 0.45549553632736206,
      "learning_rate": 1.6362053959965187e-06,
      "loss": 0.707,
      "step": 16234
    },
    {
      "epoch": 0.6728418086120436,
      "grad_norm": 0.41855502128601074,
      "learning_rate": 1.6359981764681505e-06,
      "loss": 0.6909,
      "step": 16235
    },
    {
      "epoch": 0.6728832525177173,
      "grad_norm": 0.4504358768463135,
      "learning_rate": 1.635790956939782e-06,
      "loss": 0.6772,
      "step": 16236
    },
    {
      "epoch": 0.6729246964233909,
      "grad_norm": 0.4598044157028198,
      "learning_rate": 1.635583737411414e-06,
      "loss": 0.6921,
      "step": 16237
    },
    {
      "epoch": 0.6729661403290647,
      "grad_norm": 0.4012659788131714,
      "learning_rate": 1.6353765178830455e-06,
      "loss": 0.6499,
      "step": 16238
    },
    {
      "epoch": 0.6730075842347383,
      "grad_norm": 0.42355167865753174,
      "learning_rate": 1.635169298354677e-06,
      "loss": 0.6709,
      "step": 16239
    },
    {
      "epoch": 0.6730490281404119,
      "grad_norm": 0.4301091730594635,
      "learning_rate": 1.6349620788263087e-06,
      "loss": 0.6774,
      "step": 16240
    },
    {
      "epoch": 0.6730904720460856,
      "grad_norm": 0.4485968351364136,
      "learning_rate": 1.6347548592979403e-06,
      "loss": 0.7341,
      "step": 16241
    },
    {
      "epoch": 0.6731319159517593,
      "grad_norm": 0.4498499631881714,
      "learning_rate": 1.6345476397695719e-06,
      "loss": 0.7295,
      "step": 16242
    },
    {
      "epoch": 0.673173359857433,
      "grad_norm": 0.4515078067779541,
      "learning_rate": 1.6343404202412037e-06,
      "loss": 0.6929,
      "step": 16243
    },
    {
      "epoch": 0.6732148037631066,
      "grad_norm": 0.40700146555900574,
      "learning_rate": 1.6341332007128353e-06,
      "loss": 0.6604,
      "step": 16244
    },
    {
      "epoch": 0.6732562476687803,
      "grad_norm": 0.44087740778923035,
      "learning_rate": 1.6339259811844669e-06,
      "loss": 0.6985,
      "step": 16245
    },
    {
      "epoch": 0.673297691574454,
      "grad_norm": 0.42361631989479065,
      "learning_rate": 1.6337187616560987e-06,
      "loss": 0.6396,
      "step": 16246
    },
    {
      "epoch": 0.6733391354801277,
      "grad_norm": 0.4367403984069824,
      "learning_rate": 1.6335115421277303e-06,
      "loss": 0.6477,
      "step": 16247
    },
    {
      "epoch": 0.6733805793858013,
      "grad_norm": 0.38674861192703247,
      "learning_rate": 1.6333043225993619e-06,
      "loss": 0.6917,
      "step": 16248
    },
    {
      "epoch": 0.6734220232914749,
      "grad_norm": 0.3819606304168701,
      "learning_rate": 1.6330971030709935e-06,
      "loss": 0.6569,
      "step": 16249
    },
    {
      "epoch": 0.6734634671971487,
      "grad_norm": 0.413939505815506,
      "learning_rate": 1.632889883542625e-06,
      "loss": 0.666,
      "step": 16250
    },
    {
      "epoch": 0.6735049111028223,
      "grad_norm": 0.4057898223400116,
      "learning_rate": 1.632682664014257e-06,
      "loss": 0.6681,
      "step": 16251
    },
    {
      "epoch": 0.673546355008496,
      "grad_norm": 0.4033510982990265,
      "learning_rate": 1.6324754444858885e-06,
      "loss": 0.6508,
      "step": 16252
    },
    {
      "epoch": 0.6735877989141696,
      "grad_norm": 0.3945682942867279,
      "learning_rate": 1.63226822495752e-06,
      "loss": 0.6672,
      "step": 16253
    },
    {
      "epoch": 0.6736292428198434,
      "grad_norm": 0.44846194982528687,
      "learning_rate": 1.632061005429152e-06,
      "loss": 0.7268,
      "step": 16254
    },
    {
      "epoch": 0.673670686725517,
      "grad_norm": 0.41785702109336853,
      "learning_rate": 1.6318537859007835e-06,
      "loss": 0.6791,
      "step": 16255
    },
    {
      "epoch": 0.6737121306311907,
      "grad_norm": 0.4068128764629364,
      "learning_rate": 1.631646566372415e-06,
      "loss": 0.7002,
      "step": 16256
    },
    {
      "epoch": 0.6737535745368644,
      "grad_norm": 0.39389556646347046,
      "learning_rate": 1.6314393468440467e-06,
      "loss": 0.6519,
      "step": 16257
    },
    {
      "epoch": 0.673795018442538,
      "grad_norm": 0.3985666334629059,
      "learning_rate": 1.6312321273156783e-06,
      "loss": 0.6454,
      "step": 16258
    },
    {
      "epoch": 0.6738364623482117,
      "grad_norm": 0.4097382724285126,
      "learning_rate": 1.6310249077873099e-06,
      "loss": 0.6548,
      "step": 16259
    },
    {
      "epoch": 0.6738779062538853,
      "grad_norm": 0.3841261565685272,
      "learning_rate": 1.6308176882589417e-06,
      "loss": 0.6792,
      "step": 16260
    },
    {
      "epoch": 0.6739193501595591,
      "grad_norm": 0.4110111892223358,
      "learning_rate": 1.6306104687305733e-06,
      "loss": 0.6952,
      "step": 16261
    },
    {
      "epoch": 0.6739607940652327,
      "grad_norm": 0.3988364338874817,
      "learning_rate": 1.6304032492022049e-06,
      "loss": 0.6238,
      "step": 16262
    },
    {
      "epoch": 0.6740022379709064,
      "grad_norm": 0.4527319669723511,
      "learning_rate": 1.6301960296738367e-06,
      "loss": 0.7438,
      "step": 16263
    },
    {
      "epoch": 0.67404368187658,
      "grad_norm": 0.44866862893104553,
      "learning_rate": 1.6299888101454683e-06,
      "loss": 0.6898,
      "step": 16264
    },
    {
      "epoch": 0.6740851257822538,
      "grad_norm": 0.4020460844039917,
      "learning_rate": 1.6297815906171e-06,
      "loss": 0.6614,
      "step": 16265
    },
    {
      "epoch": 0.6741265696879274,
      "grad_norm": 0.40324339270591736,
      "learning_rate": 1.6295743710887315e-06,
      "loss": 0.6399,
      "step": 16266
    },
    {
      "epoch": 0.674168013593601,
      "grad_norm": 0.3819420039653778,
      "learning_rate": 1.629367151560363e-06,
      "loss": 0.6808,
      "step": 16267
    },
    {
      "epoch": 0.6742094574992747,
      "grad_norm": 0.4149378836154938,
      "learning_rate": 1.629159932031995e-06,
      "loss": 0.7164,
      "step": 16268
    },
    {
      "epoch": 0.6742509014049484,
      "grad_norm": 0.40158915519714355,
      "learning_rate": 1.6289527125036265e-06,
      "loss": 0.7136,
      "step": 16269
    },
    {
      "epoch": 0.6742923453106221,
      "grad_norm": 0.3849627673625946,
      "learning_rate": 1.628745492975258e-06,
      "loss": 0.6185,
      "step": 16270
    },
    {
      "epoch": 0.6743337892162957,
      "grad_norm": 0.45467665791511536,
      "learning_rate": 1.6285382734468897e-06,
      "loss": 0.7211,
      "step": 16271
    },
    {
      "epoch": 0.6743752331219695,
      "grad_norm": 0.4521908760070801,
      "learning_rate": 1.6283310539185215e-06,
      "loss": 0.7085,
      "step": 16272
    },
    {
      "epoch": 0.6744166770276431,
      "grad_norm": 0.4169001281261444,
      "learning_rate": 1.628123834390153e-06,
      "loss": 0.6376,
      "step": 16273
    },
    {
      "epoch": 0.6744581209333168,
      "grad_norm": 0.44982749223709106,
      "learning_rate": 1.6279166148617847e-06,
      "loss": 0.7058,
      "step": 16274
    },
    {
      "epoch": 0.6744995648389904,
      "grad_norm": 0.4299849569797516,
      "learning_rate": 1.6277093953334163e-06,
      "loss": 0.7378,
      "step": 16275
    },
    {
      "epoch": 0.674541008744664,
      "grad_norm": 0.418659508228302,
      "learning_rate": 1.6275021758050479e-06,
      "loss": 0.7139,
      "step": 16276
    },
    {
      "epoch": 0.6745824526503378,
      "grad_norm": 0.4244844615459442,
      "learning_rate": 1.6272949562766797e-06,
      "loss": 0.7012,
      "step": 16277
    },
    {
      "epoch": 0.6746238965560114,
      "grad_norm": 0.4675801694393158,
      "learning_rate": 1.6270877367483113e-06,
      "loss": 0.7019,
      "step": 16278
    },
    {
      "epoch": 0.6746653404616851,
      "grad_norm": 0.39383813738822937,
      "learning_rate": 1.6268805172199429e-06,
      "loss": 0.5985,
      "step": 16279
    },
    {
      "epoch": 0.6747067843673588,
      "grad_norm": 0.42497989535331726,
      "learning_rate": 1.6266732976915745e-06,
      "loss": 0.7156,
      "step": 16280
    },
    {
      "epoch": 0.6747482282730325,
      "grad_norm": 0.4233720600605011,
      "learning_rate": 1.6264660781632063e-06,
      "loss": 0.6801,
      "step": 16281
    },
    {
      "epoch": 0.6747896721787061,
      "grad_norm": 0.4023624658584595,
      "learning_rate": 1.626258858634838e-06,
      "loss": 0.637,
      "step": 16282
    },
    {
      "epoch": 0.6748311160843798,
      "grad_norm": 0.3951246738433838,
      "learning_rate": 1.6260516391064695e-06,
      "loss": 0.6704,
      "step": 16283
    },
    {
      "epoch": 0.6748725599900535,
      "grad_norm": 0.3980317711830139,
      "learning_rate": 1.625844419578101e-06,
      "loss": 0.6638,
      "step": 16284
    },
    {
      "epoch": 0.6749140038957271,
      "grad_norm": 0.43908804655075073,
      "learning_rate": 1.625637200049733e-06,
      "loss": 0.6793,
      "step": 16285
    },
    {
      "epoch": 0.6749554478014008,
      "grad_norm": 0.4163043797016144,
      "learning_rate": 1.6254299805213645e-06,
      "loss": 0.7032,
      "step": 16286
    },
    {
      "epoch": 0.6749968917070744,
      "grad_norm": 0.4263400733470917,
      "learning_rate": 1.625222760992996e-06,
      "loss": 0.6843,
      "step": 16287
    },
    {
      "epoch": 0.6750383356127482,
      "grad_norm": 0.37725648283958435,
      "learning_rate": 1.6250155414646277e-06,
      "loss": 0.6519,
      "step": 16288
    },
    {
      "epoch": 0.6750797795184218,
      "grad_norm": 0.39993736147880554,
      "learning_rate": 1.6248083219362595e-06,
      "loss": 0.686,
      "step": 16289
    },
    {
      "epoch": 0.6751212234240955,
      "grad_norm": 0.39305904507637024,
      "learning_rate": 1.624601102407891e-06,
      "loss": 0.6774,
      "step": 16290
    },
    {
      "epoch": 0.6751626673297692,
      "grad_norm": 0.4104795753955841,
      "learning_rate": 1.6243938828795227e-06,
      "loss": 0.7332,
      "step": 16291
    },
    {
      "epoch": 0.6752041112354429,
      "grad_norm": 0.4128848612308502,
      "learning_rate": 1.6241866633511543e-06,
      "loss": 0.625,
      "step": 16292
    },
    {
      "epoch": 0.6752455551411165,
      "grad_norm": 0.4621288478374481,
      "learning_rate": 1.6239794438227859e-06,
      "loss": 0.6459,
      "step": 16293
    },
    {
      "epoch": 0.6752869990467901,
      "grad_norm": 0.38027289509773254,
      "learning_rate": 1.6237722242944177e-06,
      "loss": 0.6458,
      "step": 16294
    },
    {
      "epoch": 0.6753284429524639,
      "grad_norm": 0.4220151901245117,
      "learning_rate": 1.6235650047660493e-06,
      "loss": 0.6895,
      "step": 16295
    },
    {
      "epoch": 0.6753698868581375,
      "grad_norm": 0.417141318321228,
      "learning_rate": 1.623357785237681e-06,
      "loss": 0.7129,
      "step": 16296
    },
    {
      "epoch": 0.6754113307638112,
      "grad_norm": 0.4627132713794708,
      "learning_rate": 1.6231505657093125e-06,
      "loss": 0.7146,
      "step": 16297
    },
    {
      "epoch": 0.6754527746694848,
      "grad_norm": 0.45222043991088867,
      "learning_rate": 1.6229433461809443e-06,
      "loss": 0.7344,
      "step": 16298
    },
    {
      "epoch": 0.6754942185751586,
      "grad_norm": 0.4135313332080841,
      "learning_rate": 1.622736126652576e-06,
      "loss": 0.6919,
      "step": 16299
    },
    {
      "epoch": 0.6755356624808322,
      "grad_norm": 0.40488457679748535,
      "learning_rate": 1.6225289071242075e-06,
      "loss": 0.6931,
      "step": 16300
    },
    {
      "epoch": 0.6755771063865058,
      "grad_norm": 0.4216037690639496,
      "learning_rate": 1.622321687595839e-06,
      "loss": 0.7334,
      "step": 16301
    },
    {
      "epoch": 0.6756185502921795,
      "grad_norm": 0.41567814350128174,
      "learning_rate": 1.622114468067471e-06,
      "loss": 0.691,
      "step": 16302
    },
    {
      "epoch": 0.6756599941978532,
      "grad_norm": 0.41751399636268616,
      "learning_rate": 1.6219072485391025e-06,
      "loss": 0.6677,
      "step": 16303
    },
    {
      "epoch": 0.6757014381035269,
      "grad_norm": 0.44869473576545715,
      "learning_rate": 1.621700029010734e-06,
      "loss": 0.6925,
      "step": 16304
    },
    {
      "epoch": 0.6757428820092005,
      "grad_norm": 0.4266965687274933,
      "learning_rate": 1.6214928094823657e-06,
      "loss": 0.7043,
      "step": 16305
    },
    {
      "epoch": 0.6757843259148743,
      "grad_norm": 0.40986594557762146,
      "learning_rate": 1.6212855899539973e-06,
      "loss": 0.6956,
      "step": 16306
    },
    {
      "epoch": 0.6758257698205479,
      "grad_norm": 0.47800666093826294,
      "learning_rate": 1.621078370425629e-06,
      "loss": 0.6982,
      "step": 16307
    },
    {
      "epoch": 0.6758672137262216,
      "grad_norm": 0.4086482524871826,
      "learning_rate": 1.6208711508972607e-06,
      "loss": 0.7305,
      "step": 16308
    },
    {
      "epoch": 0.6759086576318952,
      "grad_norm": 0.44619372487068176,
      "learning_rate": 1.6206639313688923e-06,
      "loss": 0.6656,
      "step": 16309
    },
    {
      "epoch": 0.6759501015375688,
      "grad_norm": 0.4284505248069763,
      "learning_rate": 1.620456711840524e-06,
      "loss": 0.6781,
      "step": 16310
    },
    {
      "epoch": 0.6759915454432426,
      "grad_norm": 0.4361781179904938,
      "learning_rate": 1.6202494923121557e-06,
      "loss": 0.6724,
      "step": 16311
    },
    {
      "epoch": 0.6760329893489162,
      "grad_norm": 0.44303861260414124,
      "learning_rate": 1.6200422727837873e-06,
      "loss": 0.6614,
      "step": 16312
    },
    {
      "epoch": 0.6760744332545899,
      "grad_norm": 0.4088652431964874,
      "learning_rate": 1.619835053255419e-06,
      "loss": 0.6334,
      "step": 16313
    },
    {
      "epoch": 0.6761158771602636,
      "grad_norm": 0.4180946946144104,
      "learning_rate": 1.6196278337270505e-06,
      "loss": 0.6797,
      "step": 16314
    },
    {
      "epoch": 0.6761573210659373,
      "grad_norm": 0.3916303217411041,
      "learning_rate": 1.6194206141986823e-06,
      "loss": 0.6414,
      "step": 16315
    },
    {
      "epoch": 0.6761987649716109,
      "grad_norm": 0.4205370545387268,
      "learning_rate": 1.619213394670314e-06,
      "loss": 0.6371,
      "step": 16316
    },
    {
      "epoch": 0.6762402088772846,
      "grad_norm": 0.4024277627468109,
      "learning_rate": 1.6190061751419455e-06,
      "loss": 0.6995,
      "step": 16317
    },
    {
      "epoch": 0.6762816527829583,
      "grad_norm": 0.4250434339046478,
      "learning_rate": 1.618798955613577e-06,
      "loss": 0.6921,
      "step": 16318
    },
    {
      "epoch": 0.6763230966886319,
      "grad_norm": 0.41153907775878906,
      "learning_rate": 1.618591736085209e-06,
      "loss": 0.6835,
      "step": 16319
    },
    {
      "epoch": 0.6763645405943056,
      "grad_norm": 0.4041583240032196,
      "learning_rate": 1.6183845165568405e-06,
      "loss": 0.6863,
      "step": 16320
    },
    {
      "epoch": 0.6764059844999792,
      "grad_norm": 0.4305073618888855,
      "learning_rate": 1.618177297028472e-06,
      "loss": 0.6351,
      "step": 16321
    },
    {
      "epoch": 0.676447428405653,
      "grad_norm": 0.3942486047744751,
      "learning_rate": 1.6179700775001037e-06,
      "loss": 0.6493,
      "step": 16322
    },
    {
      "epoch": 0.6764888723113266,
      "grad_norm": 0.4173102080821991,
      "learning_rate": 1.6177628579717353e-06,
      "loss": 0.7092,
      "step": 16323
    },
    {
      "epoch": 0.6765303162170003,
      "grad_norm": 0.4446958303451538,
      "learning_rate": 1.6175556384433671e-06,
      "loss": 0.6725,
      "step": 16324
    },
    {
      "epoch": 0.676571760122674,
      "grad_norm": 0.4260506331920624,
      "learning_rate": 1.6173484189149987e-06,
      "loss": 0.6759,
      "step": 16325
    },
    {
      "epoch": 0.6766132040283477,
      "grad_norm": 0.42306116223335266,
      "learning_rate": 1.6171411993866303e-06,
      "loss": 0.6675,
      "step": 16326
    },
    {
      "epoch": 0.6766546479340213,
      "grad_norm": 0.42986342310905457,
      "learning_rate": 1.616933979858262e-06,
      "loss": 0.6997,
      "step": 16327
    },
    {
      "epoch": 0.6766960918396949,
      "grad_norm": 0.4706627428531647,
      "learning_rate": 1.6167267603298937e-06,
      "loss": 0.7153,
      "step": 16328
    },
    {
      "epoch": 0.6767375357453687,
      "grad_norm": 0.4231176972389221,
      "learning_rate": 1.6165195408015253e-06,
      "loss": 0.7085,
      "step": 16329
    },
    {
      "epoch": 0.6767789796510423,
      "grad_norm": 0.45094263553619385,
      "learning_rate": 1.616312321273157e-06,
      "loss": 0.7087,
      "step": 16330
    },
    {
      "epoch": 0.676820423556716,
      "grad_norm": 0.40936052799224854,
      "learning_rate": 1.6161051017447885e-06,
      "loss": 0.6331,
      "step": 16331
    },
    {
      "epoch": 0.6768618674623896,
      "grad_norm": 0.3899785280227661,
      "learning_rate": 1.61589788221642e-06,
      "loss": 0.6772,
      "step": 16332
    },
    {
      "epoch": 0.6769033113680634,
      "grad_norm": 0.4100188612937927,
      "learning_rate": 1.615690662688052e-06,
      "loss": 0.6163,
      "step": 16333
    },
    {
      "epoch": 0.676944755273737,
      "grad_norm": 0.3940403461456299,
      "learning_rate": 1.6154834431596835e-06,
      "loss": 0.6554,
      "step": 16334
    },
    {
      "epoch": 0.6769861991794107,
      "grad_norm": 0.38139110803604126,
      "learning_rate": 1.615276223631315e-06,
      "loss": 0.642,
      "step": 16335
    },
    {
      "epoch": 0.6770276430850843,
      "grad_norm": 0.4233373701572418,
      "learning_rate": 1.615069004102947e-06,
      "loss": 0.7378,
      "step": 16336
    },
    {
      "epoch": 0.677069086990758,
      "grad_norm": 0.40375572443008423,
      "learning_rate": 1.6148617845745785e-06,
      "loss": 0.7084,
      "step": 16337
    },
    {
      "epoch": 0.6771105308964317,
      "grad_norm": 0.42493799328804016,
      "learning_rate": 1.6146545650462101e-06,
      "loss": 0.6952,
      "step": 16338
    },
    {
      "epoch": 0.6771519748021053,
      "grad_norm": 0.42446815967559814,
      "learning_rate": 1.6144473455178417e-06,
      "loss": 0.6812,
      "step": 16339
    },
    {
      "epoch": 0.677193418707779,
      "grad_norm": 0.43217477202415466,
      "learning_rate": 1.6142401259894733e-06,
      "loss": 0.7056,
      "step": 16340
    },
    {
      "epoch": 0.6772348626134527,
      "grad_norm": 0.39489418268203735,
      "learning_rate": 1.6140329064611051e-06,
      "loss": 0.6627,
      "step": 16341
    },
    {
      "epoch": 0.6772763065191264,
      "grad_norm": 0.45600074529647827,
      "learning_rate": 1.6138256869327367e-06,
      "loss": 0.7402,
      "step": 16342
    },
    {
      "epoch": 0.6773177504248,
      "grad_norm": 0.4149990677833557,
      "learning_rate": 1.6136184674043683e-06,
      "loss": 0.6376,
      "step": 16343
    },
    {
      "epoch": 0.6773591943304738,
      "grad_norm": 0.43980997800827026,
      "learning_rate": 1.6134112478760001e-06,
      "loss": 0.6875,
      "step": 16344
    },
    {
      "epoch": 0.6774006382361474,
      "grad_norm": 0.42125609517097473,
      "learning_rate": 1.6132040283476317e-06,
      "loss": 0.6913,
      "step": 16345
    },
    {
      "epoch": 0.677442082141821,
      "grad_norm": 0.44827336072921753,
      "learning_rate": 1.6129968088192633e-06,
      "loss": 0.7043,
      "step": 16346
    },
    {
      "epoch": 0.6774835260474947,
      "grad_norm": 0.41174080967903137,
      "learning_rate": 1.612789589290895e-06,
      "loss": 0.6453,
      "step": 16347
    },
    {
      "epoch": 0.6775249699531684,
      "grad_norm": 0.4401751160621643,
      "learning_rate": 1.6125823697625265e-06,
      "loss": 0.7715,
      "step": 16348
    },
    {
      "epoch": 0.6775664138588421,
      "grad_norm": 0.39284059405326843,
      "learning_rate": 1.612375150234158e-06,
      "loss": 0.668,
      "step": 16349
    },
    {
      "epoch": 0.6776078577645157,
      "grad_norm": 0.392184853553772,
      "learning_rate": 1.61216793070579e-06,
      "loss": 0.6143,
      "step": 16350
    },
    {
      "epoch": 0.6776493016701894,
      "grad_norm": 0.4336661398410797,
      "learning_rate": 1.6119607111774215e-06,
      "loss": 0.7201,
      "step": 16351
    },
    {
      "epoch": 0.6776907455758631,
      "grad_norm": 0.41449442505836487,
      "learning_rate": 1.611753491649053e-06,
      "loss": 0.6814,
      "step": 16352
    },
    {
      "epoch": 0.6777321894815368,
      "grad_norm": 0.405932754278183,
      "learning_rate": 1.611546272120685e-06,
      "loss": 0.6533,
      "step": 16353
    },
    {
      "epoch": 0.6777736333872104,
      "grad_norm": 0.43539363145828247,
      "learning_rate": 1.6113390525923165e-06,
      "loss": 0.6742,
      "step": 16354
    },
    {
      "epoch": 0.677815077292884,
      "grad_norm": 0.4115121364593506,
      "learning_rate": 1.6111318330639481e-06,
      "loss": 0.6416,
      "step": 16355
    },
    {
      "epoch": 0.6778565211985578,
      "grad_norm": 0.4653797149658203,
      "learning_rate": 1.6109246135355797e-06,
      "loss": 0.6671,
      "step": 16356
    },
    {
      "epoch": 0.6778979651042314,
      "grad_norm": 0.3799360692501068,
      "learning_rate": 1.6107173940072113e-06,
      "loss": 0.6876,
      "step": 16357
    },
    {
      "epoch": 0.6779394090099051,
      "grad_norm": 0.4085388481616974,
      "learning_rate": 1.610510174478843e-06,
      "loss": 0.6417,
      "step": 16358
    },
    {
      "epoch": 0.6779808529155787,
      "grad_norm": 0.4052393138408661,
      "learning_rate": 1.6103029549504747e-06,
      "loss": 0.6354,
      "step": 16359
    },
    {
      "epoch": 0.6780222968212525,
      "grad_norm": 0.3867664933204651,
      "learning_rate": 1.6100957354221063e-06,
      "loss": 0.6368,
      "step": 16360
    },
    {
      "epoch": 0.6780637407269261,
      "grad_norm": 0.419985830783844,
      "learning_rate": 1.6098885158937381e-06,
      "loss": 0.6982,
      "step": 16361
    },
    {
      "epoch": 0.6781051846325997,
      "grad_norm": 0.4375365674495697,
      "learning_rate": 1.6096812963653697e-06,
      "loss": 0.6902,
      "step": 16362
    },
    {
      "epoch": 0.6781466285382735,
      "grad_norm": 0.42397716641426086,
      "learning_rate": 1.6094740768370013e-06,
      "loss": 0.663,
      "step": 16363
    },
    {
      "epoch": 0.6781880724439471,
      "grad_norm": 0.4689071476459503,
      "learning_rate": 1.609266857308633e-06,
      "loss": 0.728,
      "step": 16364
    },
    {
      "epoch": 0.6782295163496208,
      "grad_norm": 0.4131699502468109,
      "learning_rate": 1.6090596377802645e-06,
      "loss": 0.6672,
      "step": 16365
    },
    {
      "epoch": 0.6782709602552944,
      "grad_norm": 0.4080883860588074,
      "learning_rate": 1.608852418251896e-06,
      "loss": 0.6611,
      "step": 16366
    },
    {
      "epoch": 0.6783124041609682,
      "grad_norm": 0.42712077498435974,
      "learning_rate": 1.6086451987235277e-06,
      "loss": 0.7373,
      "step": 16367
    },
    {
      "epoch": 0.6783538480666418,
      "grad_norm": 0.43096664547920227,
      "learning_rate": 1.6084379791951595e-06,
      "loss": 0.7115,
      "step": 16368
    },
    {
      "epoch": 0.6783952919723155,
      "grad_norm": 0.4470101594924927,
      "learning_rate": 1.6082307596667911e-06,
      "loss": 0.6733,
      "step": 16369
    },
    {
      "epoch": 0.6784367358779891,
      "grad_norm": 0.40488073229789734,
      "learning_rate": 1.608023540138423e-06,
      "loss": 0.6436,
      "step": 16370
    },
    {
      "epoch": 0.6784781797836628,
      "grad_norm": 0.37891021370887756,
      "learning_rate": 1.6078163206100545e-06,
      "loss": 0.6249,
      "step": 16371
    },
    {
      "epoch": 0.6785196236893365,
      "grad_norm": 0.43865838646888733,
      "learning_rate": 1.6076091010816861e-06,
      "loss": 0.7073,
      "step": 16372
    },
    {
      "epoch": 0.6785610675950101,
      "grad_norm": 0.43869486451148987,
      "learning_rate": 1.6074018815533177e-06,
      "loss": 0.6827,
      "step": 16373
    },
    {
      "epoch": 0.6786025115006838,
      "grad_norm": 0.3984203338623047,
      "learning_rate": 1.6071946620249493e-06,
      "loss": 0.658,
      "step": 16374
    },
    {
      "epoch": 0.6786439554063575,
      "grad_norm": 0.45380374789237976,
      "learning_rate": 1.606987442496581e-06,
      "loss": 0.7141,
      "step": 16375
    },
    {
      "epoch": 0.6786853993120312,
      "grad_norm": 0.4323147237300873,
      "learning_rate": 1.6067802229682127e-06,
      "loss": 0.6031,
      "step": 16376
    },
    {
      "epoch": 0.6787268432177048,
      "grad_norm": 0.41525501012802124,
      "learning_rate": 1.6065730034398443e-06,
      "loss": 0.7329,
      "step": 16377
    },
    {
      "epoch": 0.6787682871233786,
      "grad_norm": 0.40813004970550537,
      "learning_rate": 1.6063657839114761e-06,
      "loss": 0.6813,
      "step": 16378
    },
    {
      "epoch": 0.6788097310290522,
      "grad_norm": 0.4367498457431793,
      "learning_rate": 1.6061585643831077e-06,
      "loss": 0.7474,
      "step": 16379
    },
    {
      "epoch": 0.6788511749347258,
      "grad_norm": 0.4303681254386902,
      "learning_rate": 1.6059513448547393e-06,
      "loss": 0.6897,
      "step": 16380
    },
    {
      "epoch": 0.6788926188403995,
      "grad_norm": 0.39185333251953125,
      "learning_rate": 1.605744125326371e-06,
      "loss": 0.6603,
      "step": 16381
    },
    {
      "epoch": 0.6789340627460732,
      "grad_norm": 0.4145040214061737,
      "learning_rate": 1.6055369057980025e-06,
      "loss": 0.6353,
      "step": 16382
    },
    {
      "epoch": 0.6789755066517469,
      "grad_norm": 0.4737611711025238,
      "learning_rate": 1.6053296862696341e-06,
      "loss": 0.6804,
      "step": 16383
    },
    {
      "epoch": 0.6790169505574205,
      "grad_norm": 0.39113378524780273,
      "learning_rate": 1.6051224667412657e-06,
      "loss": 0.7124,
      "step": 16384
    },
    {
      "epoch": 0.6790583944630942,
      "grad_norm": 0.4277935326099396,
      "learning_rate": 1.6049152472128975e-06,
      "loss": 0.6947,
      "step": 16385
    },
    {
      "epoch": 0.6790998383687679,
      "grad_norm": 0.39183101058006287,
      "learning_rate": 1.6047080276845291e-06,
      "loss": 0.7061,
      "step": 16386
    },
    {
      "epoch": 0.6791412822744416,
      "grad_norm": 0.45061150193214417,
      "learning_rate": 1.604500808156161e-06,
      "loss": 0.6682,
      "step": 16387
    },
    {
      "epoch": 0.6791827261801152,
      "grad_norm": 0.4232819080352783,
      "learning_rate": 1.6042935886277925e-06,
      "loss": 0.6459,
      "step": 16388
    },
    {
      "epoch": 0.6792241700857888,
      "grad_norm": 0.42031776905059814,
      "learning_rate": 1.6040863690994241e-06,
      "loss": 0.6907,
      "step": 16389
    },
    {
      "epoch": 0.6792656139914626,
      "grad_norm": 0.4279818832874298,
      "learning_rate": 1.6038791495710557e-06,
      "loss": 0.6531,
      "step": 16390
    },
    {
      "epoch": 0.6793070578971362,
      "grad_norm": 0.3922806680202484,
      "learning_rate": 1.6036719300426873e-06,
      "loss": 0.6721,
      "step": 16391
    },
    {
      "epoch": 0.6793485018028099,
      "grad_norm": 0.44862061738967896,
      "learning_rate": 1.603464710514319e-06,
      "loss": 0.6506,
      "step": 16392
    },
    {
      "epoch": 0.6793899457084835,
      "grad_norm": 0.4305098354816437,
      "learning_rate": 1.6032574909859505e-06,
      "loss": 0.6833,
      "step": 16393
    },
    {
      "epoch": 0.6794313896141573,
      "grad_norm": 0.40128201246261597,
      "learning_rate": 1.6030502714575823e-06,
      "loss": 0.6747,
      "step": 16394
    },
    {
      "epoch": 0.6794728335198309,
      "grad_norm": 0.45804712176322937,
      "learning_rate": 1.6028430519292141e-06,
      "loss": 0.7432,
      "step": 16395
    },
    {
      "epoch": 0.6795142774255046,
      "grad_norm": 0.43453139066696167,
      "learning_rate": 1.6026358324008457e-06,
      "loss": 0.6606,
      "step": 16396
    },
    {
      "epoch": 0.6795557213311783,
      "grad_norm": 0.4176333546638489,
      "learning_rate": 1.6024286128724773e-06,
      "loss": 0.6892,
      "step": 16397
    },
    {
      "epoch": 0.6795971652368519,
      "grad_norm": 0.38999608159065247,
      "learning_rate": 1.602221393344109e-06,
      "loss": 0.6334,
      "step": 16398
    },
    {
      "epoch": 0.6796386091425256,
      "grad_norm": 0.3954404294490814,
      "learning_rate": 1.6020141738157405e-06,
      "loss": 0.6724,
      "step": 16399
    },
    {
      "epoch": 0.6796800530481992,
      "grad_norm": 0.39918017387390137,
      "learning_rate": 1.6018069542873721e-06,
      "loss": 0.6584,
      "step": 16400
    },
    {
      "epoch": 0.679721496953873,
      "grad_norm": 0.3800272047519684,
      "learning_rate": 1.6015997347590037e-06,
      "loss": 0.5919,
      "step": 16401
    },
    {
      "epoch": 0.6797629408595466,
      "grad_norm": 0.4102345407009125,
      "learning_rate": 1.6013925152306355e-06,
      "loss": 0.6677,
      "step": 16402
    },
    {
      "epoch": 0.6798043847652203,
      "grad_norm": 0.470389723777771,
      "learning_rate": 1.6011852957022671e-06,
      "loss": 0.6833,
      "step": 16403
    },
    {
      "epoch": 0.6798458286708939,
      "grad_norm": 0.4658568799495697,
      "learning_rate": 1.600978076173899e-06,
      "loss": 0.6716,
      "step": 16404
    },
    {
      "epoch": 0.6798872725765677,
      "grad_norm": 0.4668389558792114,
      "learning_rate": 1.6007708566455305e-06,
      "loss": 0.6666,
      "step": 16405
    },
    {
      "epoch": 0.6799287164822413,
      "grad_norm": 0.41219428181648254,
      "learning_rate": 1.6005636371171621e-06,
      "loss": 0.6957,
      "step": 16406
    },
    {
      "epoch": 0.6799701603879149,
      "grad_norm": 0.41921865940093994,
      "learning_rate": 1.6003564175887937e-06,
      "loss": 0.6343,
      "step": 16407
    },
    {
      "epoch": 0.6800116042935886,
      "grad_norm": 0.45205870270729065,
      "learning_rate": 1.6001491980604253e-06,
      "loss": 0.7302,
      "step": 16408
    },
    {
      "epoch": 0.6800530481992623,
      "grad_norm": 0.40544790029525757,
      "learning_rate": 1.599941978532057e-06,
      "loss": 0.6827,
      "step": 16409
    },
    {
      "epoch": 0.680094492104936,
      "grad_norm": 0.4359225332736969,
      "learning_rate": 1.5997347590036885e-06,
      "loss": 0.6851,
      "step": 16410
    },
    {
      "epoch": 0.6801359360106096,
      "grad_norm": 0.3854822814464569,
      "learning_rate": 1.5995275394753203e-06,
      "loss": 0.6687,
      "step": 16411
    },
    {
      "epoch": 0.6801773799162834,
      "grad_norm": 0.4356563091278076,
      "learning_rate": 1.5993203199469521e-06,
      "loss": 0.6915,
      "step": 16412
    },
    {
      "epoch": 0.680218823821957,
      "grad_norm": 0.4016231596469879,
      "learning_rate": 1.5991131004185837e-06,
      "loss": 0.6888,
      "step": 16413
    },
    {
      "epoch": 0.6802602677276306,
      "grad_norm": 0.45952433347702026,
      "learning_rate": 1.5989058808902153e-06,
      "loss": 0.6317,
      "step": 16414
    },
    {
      "epoch": 0.6803017116333043,
      "grad_norm": 0.40454673767089844,
      "learning_rate": 1.598698661361847e-06,
      "loss": 0.6283,
      "step": 16415
    },
    {
      "epoch": 0.680343155538978,
      "grad_norm": 0.4314827620983124,
      "learning_rate": 1.5984914418334785e-06,
      "loss": 0.668,
      "step": 16416
    },
    {
      "epoch": 0.6803845994446517,
      "grad_norm": 0.38920295238494873,
      "learning_rate": 1.5982842223051101e-06,
      "loss": 0.6718,
      "step": 16417
    },
    {
      "epoch": 0.6804260433503253,
      "grad_norm": 0.3890742361545563,
      "learning_rate": 1.5980770027767417e-06,
      "loss": 0.6465,
      "step": 16418
    },
    {
      "epoch": 0.680467487255999,
      "grad_norm": 0.40719839930534363,
      "learning_rate": 1.5978697832483733e-06,
      "loss": 0.718,
      "step": 16419
    },
    {
      "epoch": 0.6805089311616727,
      "grad_norm": 0.4480153024196625,
      "learning_rate": 1.5976625637200051e-06,
      "loss": 0.6699,
      "step": 16420
    },
    {
      "epoch": 0.6805503750673464,
      "grad_norm": 0.4204968214035034,
      "learning_rate": 1.597455344191637e-06,
      "loss": 0.6669,
      "step": 16421
    },
    {
      "epoch": 0.68059181897302,
      "grad_norm": 0.39290159940719604,
      "learning_rate": 1.5972481246632685e-06,
      "loss": 0.6714,
      "step": 16422
    },
    {
      "epoch": 0.6806332628786936,
      "grad_norm": 0.4116114377975464,
      "learning_rate": 1.5970409051349001e-06,
      "loss": 0.6796,
      "step": 16423
    },
    {
      "epoch": 0.6806747067843674,
      "grad_norm": 0.4009172320365906,
      "learning_rate": 1.5968336856065317e-06,
      "loss": 0.7012,
      "step": 16424
    },
    {
      "epoch": 0.680716150690041,
      "grad_norm": 0.423123836517334,
      "learning_rate": 1.5966264660781633e-06,
      "loss": 0.6826,
      "step": 16425
    },
    {
      "epoch": 0.6807575945957147,
      "grad_norm": 0.387929767370224,
      "learning_rate": 1.596419246549795e-06,
      "loss": 0.6619,
      "step": 16426
    },
    {
      "epoch": 0.6807990385013883,
      "grad_norm": 0.4412389099597931,
      "learning_rate": 1.5962120270214265e-06,
      "loss": 0.6962,
      "step": 16427
    },
    {
      "epoch": 0.6808404824070621,
      "grad_norm": 0.440782368183136,
      "learning_rate": 1.5960048074930581e-06,
      "loss": 0.7002,
      "step": 16428
    },
    {
      "epoch": 0.6808819263127357,
      "grad_norm": 0.41289159655570984,
      "learning_rate": 1.5957975879646901e-06,
      "loss": 0.7097,
      "step": 16429
    },
    {
      "epoch": 0.6809233702184094,
      "grad_norm": 0.38834673166275024,
      "learning_rate": 1.5955903684363217e-06,
      "loss": 0.6785,
      "step": 16430
    },
    {
      "epoch": 0.680964814124083,
      "grad_norm": 0.42564669251441956,
      "learning_rate": 1.5953831489079533e-06,
      "loss": 0.6555,
      "step": 16431
    },
    {
      "epoch": 0.6810062580297567,
      "grad_norm": 0.4332869350910187,
      "learning_rate": 1.595175929379585e-06,
      "loss": 0.657,
      "step": 16432
    },
    {
      "epoch": 0.6810477019354304,
      "grad_norm": 0.4403079152107239,
      "learning_rate": 1.5949687098512165e-06,
      "loss": 0.6875,
      "step": 16433
    },
    {
      "epoch": 0.681089145841104,
      "grad_norm": 0.4539627730846405,
      "learning_rate": 1.5947614903228481e-06,
      "loss": 0.759,
      "step": 16434
    },
    {
      "epoch": 0.6811305897467778,
      "grad_norm": 0.41863948106765747,
      "learning_rate": 1.5945542707944797e-06,
      "loss": 0.6464,
      "step": 16435
    },
    {
      "epoch": 0.6811720336524514,
      "grad_norm": 0.43567246198654175,
      "learning_rate": 1.5943470512661113e-06,
      "loss": 0.7043,
      "step": 16436
    },
    {
      "epoch": 0.6812134775581251,
      "grad_norm": 0.4339817762374878,
      "learning_rate": 1.5941398317377433e-06,
      "loss": 0.6624,
      "step": 16437
    },
    {
      "epoch": 0.6812549214637987,
      "grad_norm": 0.41677218675613403,
      "learning_rate": 1.593932612209375e-06,
      "loss": 0.6487,
      "step": 16438
    },
    {
      "epoch": 0.6812963653694725,
      "grad_norm": 0.4437923729419708,
      "learning_rate": 1.5937253926810065e-06,
      "loss": 0.6965,
      "step": 16439
    },
    {
      "epoch": 0.6813378092751461,
      "grad_norm": 0.3804916739463806,
      "learning_rate": 1.5935181731526381e-06,
      "loss": 0.6079,
      "step": 16440
    },
    {
      "epoch": 0.6813792531808197,
      "grad_norm": 0.3881668448448181,
      "learning_rate": 1.5933109536242697e-06,
      "loss": 0.637,
      "step": 16441
    },
    {
      "epoch": 0.6814206970864934,
      "grad_norm": 0.4043217599391937,
      "learning_rate": 1.5931037340959013e-06,
      "loss": 0.6866,
      "step": 16442
    },
    {
      "epoch": 0.6814621409921671,
      "grad_norm": 0.3965768814086914,
      "learning_rate": 1.592896514567533e-06,
      "loss": 0.6495,
      "step": 16443
    },
    {
      "epoch": 0.6815035848978408,
      "grad_norm": 0.3855952024459839,
      "learning_rate": 1.5926892950391645e-06,
      "loss": 0.6659,
      "step": 16444
    },
    {
      "epoch": 0.6815450288035144,
      "grad_norm": 0.445781946182251,
      "learning_rate": 1.5924820755107961e-06,
      "loss": 0.6599,
      "step": 16445
    },
    {
      "epoch": 0.6815864727091882,
      "grad_norm": 0.454216331243515,
      "learning_rate": 1.5922748559824281e-06,
      "loss": 0.7072,
      "step": 16446
    },
    {
      "epoch": 0.6816279166148618,
      "grad_norm": 0.4069503843784332,
      "learning_rate": 1.5920676364540597e-06,
      "loss": 0.6682,
      "step": 16447
    },
    {
      "epoch": 0.6816693605205355,
      "grad_norm": 0.4295148253440857,
      "learning_rate": 1.5918604169256913e-06,
      "loss": 0.6738,
      "step": 16448
    },
    {
      "epoch": 0.6817108044262091,
      "grad_norm": 0.39556998014450073,
      "learning_rate": 1.591653197397323e-06,
      "loss": 0.6241,
      "step": 16449
    },
    {
      "epoch": 0.6817522483318827,
      "grad_norm": 0.4372722804546356,
      "learning_rate": 1.5914459778689545e-06,
      "loss": 0.6685,
      "step": 16450
    },
    {
      "epoch": 0.6817936922375565,
      "grad_norm": 0.40175312757492065,
      "learning_rate": 1.5912387583405861e-06,
      "loss": 0.6339,
      "step": 16451
    },
    {
      "epoch": 0.6818351361432301,
      "grad_norm": 0.4111756682395935,
      "learning_rate": 1.5910315388122177e-06,
      "loss": 0.6394,
      "step": 16452
    },
    {
      "epoch": 0.6818765800489038,
      "grad_norm": 0.3894212245941162,
      "learning_rate": 1.5908243192838493e-06,
      "loss": 0.6433,
      "step": 16453
    },
    {
      "epoch": 0.6819180239545775,
      "grad_norm": 0.39975839853286743,
      "learning_rate": 1.590617099755481e-06,
      "loss": 0.6362,
      "step": 16454
    },
    {
      "epoch": 0.6819594678602512,
      "grad_norm": 0.4323562979698181,
      "learning_rate": 1.590409880227113e-06,
      "loss": 0.7339,
      "step": 16455
    },
    {
      "epoch": 0.6820009117659248,
      "grad_norm": 0.3968159556388855,
      "learning_rate": 1.5902026606987445e-06,
      "loss": 0.6748,
      "step": 16456
    },
    {
      "epoch": 0.6820423556715985,
      "grad_norm": 0.4343950152397156,
      "learning_rate": 1.5899954411703761e-06,
      "loss": 0.6993,
      "step": 16457
    },
    {
      "epoch": 0.6820837995772722,
      "grad_norm": 0.3977610468864441,
      "learning_rate": 1.5897882216420077e-06,
      "loss": 0.6163,
      "step": 16458
    },
    {
      "epoch": 0.6821252434829458,
      "grad_norm": 0.44916510581970215,
      "learning_rate": 1.5895810021136393e-06,
      "loss": 0.6542,
      "step": 16459
    },
    {
      "epoch": 0.6821666873886195,
      "grad_norm": 0.43771079182624817,
      "learning_rate": 1.589373782585271e-06,
      "loss": 0.704,
      "step": 16460
    },
    {
      "epoch": 0.6822081312942931,
      "grad_norm": 0.3977387547492981,
      "learning_rate": 1.5891665630569025e-06,
      "loss": 0.6414,
      "step": 16461
    },
    {
      "epoch": 0.6822495751999669,
      "grad_norm": 0.42292988300323486,
      "learning_rate": 1.5889593435285341e-06,
      "loss": 0.6833,
      "step": 16462
    },
    {
      "epoch": 0.6822910191056405,
      "grad_norm": 0.3783145546913147,
      "learning_rate": 1.5887521240001661e-06,
      "loss": 0.6665,
      "step": 16463
    },
    {
      "epoch": 0.6823324630113142,
      "grad_norm": 0.42651253938674927,
      "learning_rate": 1.5885449044717977e-06,
      "loss": 0.6747,
      "step": 16464
    },
    {
      "epoch": 0.6823739069169878,
      "grad_norm": 0.4274154305458069,
      "learning_rate": 1.5883376849434293e-06,
      "loss": 0.6541,
      "step": 16465
    },
    {
      "epoch": 0.6824153508226616,
      "grad_norm": 0.4113521873950958,
      "learning_rate": 1.588130465415061e-06,
      "loss": 0.6912,
      "step": 16466
    },
    {
      "epoch": 0.6824567947283352,
      "grad_norm": 0.3884619474411011,
      "learning_rate": 1.5879232458866925e-06,
      "loss": 0.7103,
      "step": 16467
    },
    {
      "epoch": 0.6824982386340088,
      "grad_norm": 0.40528884530067444,
      "learning_rate": 1.5877160263583241e-06,
      "loss": 0.67,
      "step": 16468
    },
    {
      "epoch": 0.6825396825396826,
      "grad_norm": 0.43685516715049744,
      "learning_rate": 1.5875088068299557e-06,
      "loss": 0.6743,
      "step": 16469
    },
    {
      "epoch": 0.6825811264453562,
      "grad_norm": 0.43480271100997925,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 0.651,
      "step": 16470
    },
    {
      "epoch": 0.6826225703510299,
      "grad_norm": 0.44695931673049927,
      "learning_rate": 1.587094367773219e-06,
      "loss": 0.6974,
      "step": 16471
    },
    {
      "epoch": 0.6826640142567035,
      "grad_norm": 0.38569626212120056,
      "learning_rate": 1.586887148244851e-06,
      "loss": 0.6494,
      "step": 16472
    },
    {
      "epoch": 0.6827054581623773,
      "grad_norm": 0.4279003441333771,
      "learning_rate": 1.5866799287164825e-06,
      "loss": 0.6913,
      "step": 16473
    },
    {
      "epoch": 0.6827469020680509,
      "grad_norm": 0.42841625213623047,
      "learning_rate": 1.5864727091881141e-06,
      "loss": 0.6838,
      "step": 16474
    },
    {
      "epoch": 0.6827883459737245,
      "grad_norm": 0.39657533168792725,
      "learning_rate": 1.5862654896597457e-06,
      "loss": 0.6587,
      "step": 16475
    },
    {
      "epoch": 0.6828297898793982,
      "grad_norm": 0.46557366847991943,
      "learning_rate": 1.5860582701313773e-06,
      "loss": 0.688,
      "step": 16476
    },
    {
      "epoch": 0.6828712337850719,
      "grad_norm": 0.4088340699672699,
      "learning_rate": 1.585851050603009e-06,
      "loss": 0.6521,
      "step": 16477
    },
    {
      "epoch": 0.6829126776907456,
      "grad_norm": 0.39520496129989624,
      "learning_rate": 1.5856438310746405e-06,
      "loss": 0.65,
      "step": 16478
    },
    {
      "epoch": 0.6829541215964192,
      "grad_norm": 0.43614083528518677,
      "learning_rate": 1.5854366115462721e-06,
      "loss": 0.7324,
      "step": 16479
    },
    {
      "epoch": 0.682995565502093,
      "grad_norm": 0.4328020215034485,
      "learning_rate": 1.5852293920179037e-06,
      "loss": 0.7106,
      "step": 16480
    },
    {
      "epoch": 0.6830370094077666,
      "grad_norm": 0.4099116325378418,
      "learning_rate": 1.5850221724895357e-06,
      "loss": 0.6519,
      "step": 16481
    },
    {
      "epoch": 0.6830784533134403,
      "grad_norm": 0.43008145689964294,
      "learning_rate": 1.5848149529611673e-06,
      "loss": 0.6464,
      "step": 16482
    },
    {
      "epoch": 0.6831198972191139,
      "grad_norm": 0.42204973101615906,
      "learning_rate": 1.584607733432799e-06,
      "loss": 0.6729,
      "step": 16483
    },
    {
      "epoch": 0.6831613411247875,
      "grad_norm": 0.40035882592201233,
      "learning_rate": 1.5844005139044305e-06,
      "loss": 0.6619,
      "step": 16484
    },
    {
      "epoch": 0.6832027850304613,
      "grad_norm": 0.4388119876384735,
      "learning_rate": 1.5841932943760621e-06,
      "loss": 0.6975,
      "step": 16485
    },
    {
      "epoch": 0.6832442289361349,
      "grad_norm": 0.46647024154663086,
      "learning_rate": 1.5839860748476937e-06,
      "loss": 0.7454,
      "step": 16486
    },
    {
      "epoch": 0.6832856728418086,
      "grad_norm": 0.37079963088035583,
      "learning_rate": 1.5837788553193253e-06,
      "loss": 0.6313,
      "step": 16487
    },
    {
      "epoch": 0.6833271167474823,
      "grad_norm": 0.45674821734428406,
      "learning_rate": 1.583571635790957e-06,
      "loss": 0.6571,
      "step": 16488
    },
    {
      "epoch": 0.683368560653156,
      "grad_norm": 0.4435444474220276,
      "learning_rate": 1.5833644162625885e-06,
      "loss": 0.6857,
      "step": 16489
    },
    {
      "epoch": 0.6834100045588296,
      "grad_norm": 0.4087064862251282,
      "learning_rate": 1.5831571967342205e-06,
      "loss": 0.6564,
      "step": 16490
    },
    {
      "epoch": 0.6834514484645033,
      "grad_norm": 0.42712050676345825,
      "learning_rate": 1.5829499772058521e-06,
      "loss": 0.647,
      "step": 16491
    },
    {
      "epoch": 0.683492892370177,
      "grad_norm": 0.3875502645969391,
      "learning_rate": 1.5827427576774837e-06,
      "loss": 0.7181,
      "step": 16492
    },
    {
      "epoch": 0.6835343362758506,
      "grad_norm": 0.43457236886024475,
      "learning_rate": 1.5825355381491153e-06,
      "loss": 0.6504,
      "step": 16493
    },
    {
      "epoch": 0.6835757801815243,
      "grad_norm": 0.4282146394252777,
      "learning_rate": 1.582328318620747e-06,
      "loss": 0.6628,
      "step": 16494
    },
    {
      "epoch": 0.6836172240871979,
      "grad_norm": 0.42648109793663025,
      "learning_rate": 1.5821210990923785e-06,
      "loss": 0.6797,
      "step": 16495
    },
    {
      "epoch": 0.6836586679928717,
      "grad_norm": 0.432401567697525,
      "learning_rate": 1.5819138795640101e-06,
      "loss": 0.6802,
      "step": 16496
    },
    {
      "epoch": 0.6837001118985453,
      "grad_norm": 0.41156721115112305,
      "learning_rate": 1.5817066600356417e-06,
      "loss": 0.7217,
      "step": 16497
    },
    {
      "epoch": 0.683741555804219,
      "grad_norm": 0.4197638928890228,
      "learning_rate": 1.5814994405072737e-06,
      "loss": 0.6382,
      "step": 16498
    },
    {
      "epoch": 0.6837829997098926,
      "grad_norm": 0.4296789765357971,
      "learning_rate": 1.5812922209789053e-06,
      "loss": 0.6733,
      "step": 16499
    },
    {
      "epoch": 0.6838244436155664,
      "grad_norm": 0.39848610758781433,
      "learning_rate": 1.581085001450537e-06,
      "loss": 0.6176,
      "step": 16500
    },
    {
      "epoch": 0.68386588752124,
      "grad_norm": 0.3919709026813507,
      "learning_rate": 1.5808777819221685e-06,
      "loss": 0.6495,
      "step": 16501
    },
    {
      "epoch": 0.6839073314269136,
      "grad_norm": 0.4141601324081421,
      "learning_rate": 1.5806705623938001e-06,
      "loss": 0.7084,
      "step": 16502
    },
    {
      "epoch": 0.6839487753325874,
      "grad_norm": 0.40558919310569763,
      "learning_rate": 1.5804633428654317e-06,
      "loss": 0.6359,
      "step": 16503
    },
    {
      "epoch": 0.683990219238261,
      "grad_norm": 0.39816582202911377,
      "learning_rate": 1.5802561233370633e-06,
      "loss": 0.6287,
      "step": 16504
    },
    {
      "epoch": 0.6840316631439347,
      "grad_norm": 0.41478803753852844,
      "learning_rate": 1.580048903808695e-06,
      "loss": 0.6727,
      "step": 16505
    },
    {
      "epoch": 0.6840731070496083,
      "grad_norm": 0.3917585611343384,
      "learning_rate": 1.5798416842803265e-06,
      "loss": 0.64,
      "step": 16506
    },
    {
      "epoch": 0.6841145509552821,
      "grad_norm": 0.38947901129722595,
      "learning_rate": 1.5796344647519585e-06,
      "loss": 0.6396,
      "step": 16507
    },
    {
      "epoch": 0.6841559948609557,
      "grad_norm": 0.39654383063316345,
      "learning_rate": 1.5794272452235901e-06,
      "loss": 0.7181,
      "step": 16508
    },
    {
      "epoch": 0.6841974387666294,
      "grad_norm": 0.39379623532295227,
      "learning_rate": 1.5792200256952217e-06,
      "loss": 0.6619,
      "step": 16509
    },
    {
      "epoch": 0.684238882672303,
      "grad_norm": 0.42536720633506775,
      "learning_rate": 1.5790128061668533e-06,
      "loss": 0.7007,
      "step": 16510
    },
    {
      "epoch": 0.6842803265779767,
      "grad_norm": 0.41144341230392456,
      "learning_rate": 1.578805586638485e-06,
      "loss": 0.6689,
      "step": 16511
    },
    {
      "epoch": 0.6843217704836504,
      "grad_norm": 0.4173933267593384,
      "learning_rate": 1.5785983671101165e-06,
      "loss": 0.7174,
      "step": 16512
    },
    {
      "epoch": 0.684363214389324,
      "grad_norm": 0.3956053853034973,
      "learning_rate": 1.5783911475817481e-06,
      "loss": 0.6897,
      "step": 16513
    },
    {
      "epoch": 0.6844046582949977,
      "grad_norm": 0.4237261414527893,
      "learning_rate": 1.5781839280533797e-06,
      "loss": 0.686,
      "step": 16514
    },
    {
      "epoch": 0.6844461022006714,
      "grad_norm": 0.37724512815475464,
      "learning_rate": 1.5779767085250113e-06,
      "loss": 0.6365,
      "step": 16515
    },
    {
      "epoch": 0.6844875461063451,
      "grad_norm": 0.41833508014678955,
      "learning_rate": 1.5777694889966433e-06,
      "loss": 0.6486,
      "step": 16516
    },
    {
      "epoch": 0.6845289900120187,
      "grad_norm": 0.3877980709075928,
      "learning_rate": 1.577562269468275e-06,
      "loss": 0.6691,
      "step": 16517
    },
    {
      "epoch": 0.6845704339176925,
      "grad_norm": 0.39981138706207275,
      "learning_rate": 1.5773550499399065e-06,
      "loss": 0.6868,
      "step": 16518
    },
    {
      "epoch": 0.6846118778233661,
      "grad_norm": 0.405407577753067,
      "learning_rate": 1.5771478304115381e-06,
      "loss": 0.6838,
      "step": 16519
    },
    {
      "epoch": 0.6846533217290397,
      "grad_norm": 0.3803228735923767,
      "learning_rate": 1.5769406108831697e-06,
      "loss": 0.6328,
      "step": 16520
    },
    {
      "epoch": 0.6846947656347134,
      "grad_norm": 0.39798641204833984,
      "learning_rate": 1.5767333913548013e-06,
      "loss": 0.7107,
      "step": 16521
    },
    {
      "epoch": 0.684736209540387,
      "grad_norm": 0.390122652053833,
      "learning_rate": 1.576526171826433e-06,
      "loss": 0.629,
      "step": 16522
    },
    {
      "epoch": 0.6847776534460608,
      "grad_norm": 0.3843939006328583,
      "learning_rate": 1.5763189522980645e-06,
      "loss": 0.6329,
      "step": 16523
    },
    {
      "epoch": 0.6848190973517344,
      "grad_norm": 0.42882293462753296,
      "learning_rate": 1.5761117327696965e-06,
      "loss": 0.6968,
      "step": 16524
    },
    {
      "epoch": 0.6848605412574081,
      "grad_norm": 0.38911494612693787,
      "learning_rate": 1.5759045132413281e-06,
      "loss": 0.6613,
      "step": 16525
    },
    {
      "epoch": 0.6849019851630818,
      "grad_norm": 0.3814469277858734,
      "learning_rate": 1.5756972937129597e-06,
      "loss": 0.6724,
      "step": 16526
    },
    {
      "epoch": 0.6849434290687555,
      "grad_norm": 0.46253183484077454,
      "learning_rate": 1.5754900741845913e-06,
      "loss": 0.6598,
      "step": 16527
    },
    {
      "epoch": 0.6849848729744291,
      "grad_norm": 0.4089282155036926,
      "learning_rate": 1.575282854656223e-06,
      "loss": 0.6528,
      "step": 16528
    },
    {
      "epoch": 0.6850263168801027,
      "grad_norm": 0.43547695875167847,
      "learning_rate": 1.5750756351278545e-06,
      "loss": 0.6802,
      "step": 16529
    },
    {
      "epoch": 0.6850677607857765,
      "grad_norm": 0.46075403690338135,
      "learning_rate": 1.5748684155994861e-06,
      "loss": 0.644,
      "step": 16530
    },
    {
      "epoch": 0.6851092046914501,
      "grad_norm": 0.3681713938713074,
      "learning_rate": 1.5746611960711177e-06,
      "loss": 0.6615,
      "step": 16531
    },
    {
      "epoch": 0.6851506485971238,
      "grad_norm": 0.41443437337875366,
      "learning_rate": 1.5744539765427493e-06,
      "loss": 0.6586,
      "step": 16532
    },
    {
      "epoch": 0.6851920925027974,
      "grad_norm": 0.4202670156955719,
      "learning_rate": 1.5742467570143813e-06,
      "loss": 0.6835,
      "step": 16533
    },
    {
      "epoch": 0.6852335364084712,
      "grad_norm": 0.41073766350746155,
      "learning_rate": 1.574039537486013e-06,
      "loss": 0.6445,
      "step": 16534
    },
    {
      "epoch": 0.6852749803141448,
      "grad_norm": 0.4130845069885254,
      "learning_rate": 1.5738323179576445e-06,
      "loss": 0.6738,
      "step": 16535
    },
    {
      "epoch": 0.6853164242198184,
      "grad_norm": 0.4134502708911896,
      "learning_rate": 1.5736250984292761e-06,
      "loss": 0.6697,
      "step": 16536
    },
    {
      "epoch": 0.6853578681254922,
      "grad_norm": 0.3807763457298279,
      "learning_rate": 1.5734178789009077e-06,
      "loss": 0.6543,
      "step": 16537
    },
    {
      "epoch": 0.6853993120311658,
      "grad_norm": 0.4259345829486847,
      "learning_rate": 1.5732106593725393e-06,
      "loss": 0.6823,
      "step": 16538
    },
    {
      "epoch": 0.6854407559368395,
      "grad_norm": 0.4215068221092224,
      "learning_rate": 1.573003439844171e-06,
      "loss": 0.6819,
      "step": 16539
    },
    {
      "epoch": 0.6854821998425131,
      "grad_norm": 0.45479243993759155,
      "learning_rate": 1.5727962203158025e-06,
      "loss": 0.6824,
      "step": 16540
    },
    {
      "epoch": 0.6855236437481869,
      "grad_norm": 0.42038097977638245,
      "learning_rate": 1.5725890007874341e-06,
      "loss": 0.7203,
      "step": 16541
    },
    {
      "epoch": 0.6855650876538605,
      "grad_norm": 0.4235568642616272,
      "learning_rate": 1.5723817812590661e-06,
      "loss": 0.701,
      "step": 16542
    },
    {
      "epoch": 0.6856065315595342,
      "grad_norm": 0.4088001251220703,
      "learning_rate": 1.5721745617306977e-06,
      "loss": 0.7339,
      "step": 16543
    },
    {
      "epoch": 0.6856479754652078,
      "grad_norm": 0.37654632329940796,
      "learning_rate": 1.5719673422023293e-06,
      "loss": 0.6416,
      "step": 16544
    },
    {
      "epoch": 0.6856894193708815,
      "grad_norm": 0.3778918981552124,
      "learning_rate": 1.571760122673961e-06,
      "loss": 0.6454,
      "step": 16545
    },
    {
      "epoch": 0.6857308632765552,
      "grad_norm": 0.4051094055175781,
      "learning_rate": 1.5715529031455925e-06,
      "loss": 0.6429,
      "step": 16546
    },
    {
      "epoch": 0.6857723071822288,
      "grad_norm": 0.41533297300338745,
      "learning_rate": 1.5713456836172241e-06,
      "loss": 0.6442,
      "step": 16547
    },
    {
      "epoch": 0.6858137510879025,
      "grad_norm": 0.41721484065055847,
      "learning_rate": 1.5711384640888557e-06,
      "loss": 0.6323,
      "step": 16548
    },
    {
      "epoch": 0.6858551949935762,
      "grad_norm": 0.38716697692871094,
      "learning_rate": 1.5709312445604873e-06,
      "loss": 0.6571,
      "step": 16549
    },
    {
      "epoch": 0.6858966388992499,
      "grad_norm": 0.43903231620788574,
      "learning_rate": 1.5707240250321193e-06,
      "loss": 0.7168,
      "step": 16550
    },
    {
      "epoch": 0.6859380828049235,
      "grad_norm": 0.39671075344085693,
      "learning_rate": 1.570516805503751e-06,
      "loss": 0.644,
      "step": 16551
    },
    {
      "epoch": 0.6859795267105973,
      "grad_norm": 0.4761524796485901,
      "learning_rate": 1.5703095859753825e-06,
      "loss": 0.7354,
      "step": 16552
    },
    {
      "epoch": 0.6860209706162709,
      "grad_norm": 0.4064832031726837,
      "learning_rate": 1.5701023664470141e-06,
      "loss": 0.6594,
      "step": 16553
    },
    {
      "epoch": 0.6860624145219445,
      "grad_norm": 0.4024033844470978,
      "learning_rate": 1.5698951469186457e-06,
      "loss": 0.7124,
      "step": 16554
    },
    {
      "epoch": 0.6861038584276182,
      "grad_norm": 0.40506869554519653,
      "learning_rate": 1.5696879273902773e-06,
      "loss": 0.6566,
      "step": 16555
    },
    {
      "epoch": 0.6861453023332919,
      "grad_norm": 0.4111260771751404,
      "learning_rate": 1.569480707861909e-06,
      "loss": 0.6458,
      "step": 16556
    },
    {
      "epoch": 0.6861867462389656,
      "grad_norm": 0.42981472611427307,
      "learning_rate": 1.5692734883335405e-06,
      "loss": 0.6594,
      "step": 16557
    },
    {
      "epoch": 0.6862281901446392,
      "grad_norm": 0.3832322657108307,
      "learning_rate": 1.5690662688051721e-06,
      "loss": 0.6486,
      "step": 16558
    },
    {
      "epoch": 0.6862696340503129,
      "grad_norm": 0.4087298512458801,
      "learning_rate": 1.5688590492768041e-06,
      "loss": 0.665,
      "step": 16559
    },
    {
      "epoch": 0.6863110779559866,
      "grad_norm": 0.39591184258461,
      "learning_rate": 1.5686518297484357e-06,
      "loss": 0.7036,
      "step": 16560
    },
    {
      "epoch": 0.6863525218616603,
      "grad_norm": 0.4204745590686798,
      "learning_rate": 1.5684446102200673e-06,
      "loss": 0.6853,
      "step": 16561
    },
    {
      "epoch": 0.6863939657673339,
      "grad_norm": 0.3913266360759735,
      "learning_rate": 1.568237390691699e-06,
      "loss": 0.6178,
      "step": 16562
    },
    {
      "epoch": 0.6864354096730075,
      "grad_norm": 0.4209268391132355,
      "learning_rate": 1.5680301711633305e-06,
      "loss": 0.6661,
      "step": 16563
    },
    {
      "epoch": 0.6864768535786813,
      "grad_norm": 0.44486355781555176,
      "learning_rate": 1.5678229516349621e-06,
      "loss": 0.6576,
      "step": 16564
    },
    {
      "epoch": 0.6865182974843549,
      "grad_norm": 0.4286119043827057,
      "learning_rate": 1.5676157321065937e-06,
      "loss": 0.7102,
      "step": 16565
    },
    {
      "epoch": 0.6865597413900286,
      "grad_norm": 0.4500056207180023,
      "learning_rate": 1.5674085125782253e-06,
      "loss": 0.6621,
      "step": 16566
    },
    {
      "epoch": 0.6866011852957022,
      "grad_norm": 0.39222705364227295,
      "learning_rate": 1.567201293049857e-06,
      "loss": 0.689,
      "step": 16567
    },
    {
      "epoch": 0.686642629201376,
      "grad_norm": 0.4452320635318756,
      "learning_rate": 1.566994073521489e-06,
      "loss": 0.7114,
      "step": 16568
    },
    {
      "epoch": 0.6866840731070496,
      "grad_norm": 0.3967004716396332,
      "learning_rate": 1.5667868539931205e-06,
      "loss": 0.6643,
      "step": 16569
    },
    {
      "epoch": 0.6867255170127233,
      "grad_norm": 0.4173629879951477,
      "learning_rate": 1.5665796344647521e-06,
      "loss": 0.6283,
      "step": 16570
    },
    {
      "epoch": 0.686766960918397,
      "grad_norm": 0.4006367623806,
      "learning_rate": 1.5663724149363837e-06,
      "loss": 0.6176,
      "step": 16571
    },
    {
      "epoch": 0.6868084048240706,
      "grad_norm": 0.37428367137908936,
      "learning_rate": 1.5661651954080153e-06,
      "loss": 0.6014,
      "step": 16572
    },
    {
      "epoch": 0.6868498487297443,
      "grad_norm": 0.4641467034816742,
      "learning_rate": 1.565957975879647e-06,
      "loss": 0.7308,
      "step": 16573
    },
    {
      "epoch": 0.6868912926354179,
      "grad_norm": 0.4359675943851471,
      "learning_rate": 1.5657507563512785e-06,
      "loss": 0.6833,
      "step": 16574
    },
    {
      "epoch": 0.6869327365410917,
      "grad_norm": 0.4111281633377075,
      "learning_rate": 1.5655435368229101e-06,
      "loss": 0.6729,
      "step": 16575
    },
    {
      "epoch": 0.6869741804467653,
      "grad_norm": 0.398801326751709,
      "learning_rate": 1.5653363172945417e-06,
      "loss": 0.6554,
      "step": 16576
    },
    {
      "epoch": 0.687015624352439,
      "grad_norm": 0.4150601327419281,
      "learning_rate": 1.5651290977661737e-06,
      "loss": 0.6782,
      "step": 16577
    },
    {
      "epoch": 0.6870570682581126,
      "grad_norm": 0.40952086448669434,
      "learning_rate": 1.5649218782378053e-06,
      "loss": 0.7075,
      "step": 16578
    },
    {
      "epoch": 0.6870985121637864,
      "grad_norm": 0.41787293553352356,
      "learning_rate": 1.564714658709437e-06,
      "loss": 0.6836,
      "step": 16579
    },
    {
      "epoch": 0.68713995606946,
      "grad_norm": 0.4947410821914673,
      "learning_rate": 1.5645074391810685e-06,
      "loss": 0.7079,
      "step": 16580
    },
    {
      "epoch": 0.6871813999751336,
      "grad_norm": 0.47498029470443726,
      "learning_rate": 1.5643002196527001e-06,
      "loss": 0.7661,
      "step": 16581
    },
    {
      "epoch": 0.6872228438808073,
      "grad_norm": 0.4191833436489105,
      "learning_rate": 1.5640930001243317e-06,
      "loss": 0.6766,
      "step": 16582
    },
    {
      "epoch": 0.687264287786481,
      "grad_norm": 0.38336294889450073,
      "learning_rate": 1.5638857805959633e-06,
      "loss": 0.6667,
      "step": 16583
    },
    {
      "epoch": 0.6873057316921547,
      "grad_norm": 0.41743946075439453,
      "learning_rate": 1.563678561067595e-06,
      "loss": 0.6149,
      "step": 16584
    },
    {
      "epoch": 0.6873471755978283,
      "grad_norm": 0.4376504123210907,
      "learning_rate": 1.563471341539227e-06,
      "loss": 0.6603,
      "step": 16585
    },
    {
      "epoch": 0.687388619503502,
      "grad_norm": 0.4002896249294281,
      "learning_rate": 1.5632641220108585e-06,
      "loss": 0.698,
      "step": 16586
    },
    {
      "epoch": 0.6874300634091757,
      "grad_norm": 0.4106179177761078,
      "learning_rate": 1.5630569024824901e-06,
      "loss": 0.6782,
      "step": 16587
    },
    {
      "epoch": 0.6874715073148494,
      "grad_norm": 0.39783352613449097,
      "learning_rate": 1.5628496829541217e-06,
      "loss": 0.6683,
      "step": 16588
    },
    {
      "epoch": 0.687512951220523,
      "grad_norm": 0.40603554248809814,
      "learning_rate": 1.5626424634257533e-06,
      "loss": 0.6742,
      "step": 16589
    },
    {
      "epoch": 0.6875543951261966,
      "grad_norm": 0.39988526701927185,
      "learning_rate": 1.562435243897385e-06,
      "loss": 0.678,
      "step": 16590
    },
    {
      "epoch": 0.6875958390318704,
      "grad_norm": 0.4415309727191925,
      "learning_rate": 1.5622280243690165e-06,
      "loss": 0.7517,
      "step": 16591
    },
    {
      "epoch": 0.687637282937544,
      "grad_norm": 0.4360368251800537,
      "learning_rate": 1.5620208048406481e-06,
      "loss": 0.6565,
      "step": 16592
    },
    {
      "epoch": 0.6876787268432177,
      "grad_norm": 0.4377506673336029,
      "learning_rate": 1.56181358531228e-06,
      "loss": 0.7119,
      "step": 16593
    },
    {
      "epoch": 0.6877201707488914,
      "grad_norm": 0.4718312621116638,
      "learning_rate": 1.5616063657839117e-06,
      "loss": 0.7297,
      "step": 16594
    },
    {
      "epoch": 0.6877616146545651,
      "grad_norm": 0.41062983870506287,
      "learning_rate": 1.5613991462555433e-06,
      "loss": 0.6667,
      "step": 16595
    },
    {
      "epoch": 0.6878030585602387,
      "grad_norm": 0.3887721300125122,
      "learning_rate": 1.561191926727175e-06,
      "loss": 0.7039,
      "step": 16596
    },
    {
      "epoch": 0.6878445024659123,
      "grad_norm": 0.37392768263816833,
      "learning_rate": 1.5609847071988065e-06,
      "loss": 0.6484,
      "step": 16597
    },
    {
      "epoch": 0.6878859463715861,
      "grad_norm": 0.4421646296977997,
      "learning_rate": 1.5607774876704381e-06,
      "loss": 0.6299,
      "step": 16598
    },
    {
      "epoch": 0.6879273902772597,
      "grad_norm": 0.43053150177001953,
      "learning_rate": 1.5605702681420697e-06,
      "loss": 0.6981,
      "step": 16599
    },
    {
      "epoch": 0.6879688341829334,
      "grad_norm": 0.4001030921936035,
      "learning_rate": 1.5603630486137013e-06,
      "loss": 0.6431,
      "step": 16600
    },
    {
      "epoch": 0.688010278088607,
      "grad_norm": 0.4896659553050995,
      "learning_rate": 1.560155829085333e-06,
      "loss": 0.6539,
      "step": 16601
    },
    {
      "epoch": 0.6880517219942808,
      "grad_norm": 0.46579355001449585,
      "learning_rate": 1.5599486095569647e-06,
      "loss": 0.7017,
      "step": 16602
    },
    {
      "epoch": 0.6880931658999544,
      "grad_norm": 0.5946474075317383,
      "learning_rate": 1.5597413900285965e-06,
      "loss": 0.6731,
      "step": 16603
    },
    {
      "epoch": 0.6881346098056281,
      "grad_norm": 0.41856658458709717,
      "learning_rate": 1.5595341705002281e-06,
      "loss": 0.6332,
      "step": 16604
    },
    {
      "epoch": 0.6881760537113017,
      "grad_norm": 0.4070376455783844,
      "learning_rate": 1.5593269509718597e-06,
      "loss": 0.7124,
      "step": 16605
    },
    {
      "epoch": 0.6882174976169754,
      "grad_norm": 0.4237927794456482,
      "learning_rate": 1.5591197314434913e-06,
      "loss": 0.6603,
      "step": 16606
    },
    {
      "epoch": 0.6882589415226491,
      "grad_norm": 0.4384535551071167,
      "learning_rate": 1.558912511915123e-06,
      "loss": 0.6399,
      "step": 16607
    },
    {
      "epoch": 0.6883003854283227,
      "grad_norm": 0.41120848059654236,
      "learning_rate": 1.5587052923867545e-06,
      "loss": 0.7314,
      "step": 16608
    },
    {
      "epoch": 0.6883418293339965,
      "grad_norm": 0.47084200382232666,
      "learning_rate": 1.5584980728583861e-06,
      "loss": 0.7021,
      "step": 16609
    },
    {
      "epoch": 0.6883832732396701,
      "grad_norm": 0.42838045954704285,
      "learning_rate": 1.558290853330018e-06,
      "loss": 0.6853,
      "step": 16610
    },
    {
      "epoch": 0.6884247171453438,
      "grad_norm": 0.42581799626350403,
      "learning_rate": 1.5580836338016498e-06,
      "loss": 0.6422,
      "step": 16611
    },
    {
      "epoch": 0.6884661610510174,
      "grad_norm": 0.4468276798725128,
      "learning_rate": 1.5578764142732813e-06,
      "loss": 0.7737,
      "step": 16612
    },
    {
      "epoch": 0.6885076049566912,
      "grad_norm": 0.42627325654029846,
      "learning_rate": 1.557669194744913e-06,
      "loss": 0.6744,
      "step": 16613
    },
    {
      "epoch": 0.6885490488623648,
      "grad_norm": 0.3897863030433655,
      "learning_rate": 1.5574619752165445e-06,
      "loss": 0.6317,
      "step": 16614
    },
    {
      "epoch": 0.6885904927680384,
      "grad_norm": 0.38529959321022034,
      "learning_rate": 1.5572547556881761e-06,
      "loss": 0.6831,
      "step": 16615
    },
    {
      "epoch": 0.6886319366737121,
      "grad_norm": 0.41123276948928833,
      "learning_rate": 1.5570475361598077e-06,
      "loss": 0.696,
      "step": 16616
    },
    {
      "epoch": 0.6886733805793858,
      "grad_norm": 0.44052428007125854,
      "learning_rate": 1.5568403166314393e-06,
      "loss": 0.7007,
      "step": 16617
    },
    {
      "epoch": 0.6887148244850595,
      "grad_norm": 0.433239221572876,
      "learning_rate": 1.556633097103071e-06,
      "loss": 0.6866,
      "step": 16618
    },
    {
      "epoch": 0.6887562683907331,
      "grad_norm": 0.3594861626625061,
      "learning_rate": 1.5564258775747027e-06,
      "loss": 0.6045,
      "step": 16619
    },
    {
      "epoch": 0.6887977122964068,
      "grad_norm": 0.40516790747642517,
      "learning_rate": 1.5562186580463346e-06,
      "loss": 0.6643,
      "step": 16620
    },
    {
      "epoch": 0.6888391562020805,
      "grad_norm": 0.40015071630477905,
      "learning_rate": 1.5560114385179661e-06,
      "loss": 0.6758,
      "step": 16621
    },
    {
      "epoch": 0.6888806001077542,
      "grad_norm": 0.40229904651641846,
      "learning_rate": 1.5558042189895977e-06,
      "loss": 0.6394,
      "step": 16622
    },
    {
      "epoch": 0.6889220440134278,
      "grad_norm": 0.4143989682197571,
      "learning_rate": 1.5555969994612293e-06,
      "loss": 0.6694,
      "step": 16623
    },
    {
      "epoch": 0.6889634879191014,
      "grad_norm": 0.45119255781173706,
      "learning_rate": 1.555389779932861e-06,
      "loss": 0.6772,
      "step": 16624
    },
    {
      "epoch": 0.6890049318247752,
      "grad_norm": 0.3928352892398834,
      "learning_rate": 1.5551825604044925e-06,
      "loss": 0.6389,
      "step": 16625
    },
    {
      "epoch": 0.6890463757304488,
      "grad_norm": 0.4000597894191742,
      "learning_rate": 1.5549753408761241e-06,
      "loss": 0.7,
      "step": 16626
    },
    {
      "epoch": 0.6890878196361225,
      "grad_norm": 0.3982619643211365,
      "learning_rate": 1.554768121347756e-06,
      "loss": 0.639,
      "step": 16627
    },
    {
      "epoch": 0.6891292635417962,
      "grad_norm": 0.44259390234947205,
      "learning_rate": 1.5545609018193875e-06,
      "loss": 0.686,
      "step": 16628
    },
    {
      "epoch": 0.6891707074474699,
      "grad_norm": 0.4073920249938965,
      "learning_rate": 1.5543536822910194e-06,
      "loss": 0.7192,
      "step": 16629
    },
    {
      "epoch": 0.6892121513531435,
      "grad_norm": 0.4161735475063324,
      "learning_rate": 1.554146462762651e-06,
      "loss": 0.6526,
      "step": 16630
    },
    {
      "epoch": 0.6892535952588172,
      "grad_norm": 0.40179237723350525,
      "learning_rate": 1.5539392432342825e-06,
      "loss": 0.6736,
      "step": 16631
    },
    {
      "epoch": 0.6892950391644909,
      "grad_norm": 0.3989693224430084,
      "learning_rate": 1.5537320237059141e-06,
      "loss": 0.649,
      "step": 16632
    },
    {
      "epoch": 0.6893364830701645,
      "grad_norm": 0.41954126954078674,
      "learning_rate": 1.5535248041775457e-06,
      "loss": 0.7083,
      "step": 16633
    },
    {
      "epoch": 0.6893779269758382,
      "grad_norm": 0.39503049850463867,
      "learning_rate": 1.5533175846491773e-06,
      "loss": 0.6772,
      "step": 16634
    },
    {
      "epoch": 0.6894193708815118,
      "grad_norm": 0.39476102590560913,
      "learning_rate": 1.553110365120809e-06,
      "loss": 0.6627,
      "step": 16635
    },
    {
      "epoch": 0.6894608147871856,
      "grad_norm": 0.43434375524520874,
      "learning_rate": 1.5529031455924407e-06,
      "loss": 0.675,
      "step": 16636
    },
    {
      "epoch": 0.6895022586928592,
      "grad_norm": 0.41737160086631775,
      "learning_rate": 1.5526959260640723e-06,
      "loss": 0.6492,
      "step": 16637
    },
    {
      "epoch": 0.6895437025985329,
      "grad_norm": 0.3905733525753021,
      "learning_rate": 1.5524887065357042e-06,
      "loss": 0.6401,
      "step": 16638
    },
    {
      "epoch": 0.6895851465042065,
      "grad_norm": 0.4114612638950348,
      "learning_rate": 1.5522814870073358e-06,
      "loss": 0.6372,
      "step": 16639
    },
    {
      "epoch": 0.6896265904098803,
      "grad_norm": 0.4302862882614136,
      "learning_rate": 1.5520742674789673e-06,
      "loss": 0.6765,
      "step": 16640
    },
    {
      "epoch": 0.6896680343155539,
      "grad_norm": 0.4326808452606201,
      "learning_rate": 1.551867047950599e-06,
      "loss": 0.6886,
      "step": 16641
    },
    {
      "epoch": 0.6897094782212275,
      "grad_norm": 0.4120123088359833,
      "learning_rate": 1.5516598284222305e-06,
      "loss": 0.6305,
      "step": 16642
    },
    {
      "epoch": 0.6897509221269013,
      "grad_norm": 0.3964245319366455,
      "learning_rate": 1.5514526088938621e-06,
      "loss": 0.6846,
      "step": 16643
    },
    {
      "epoch": 0.6897923660325749,
      "grad_norm": 0.42278286814689636,
      "learning_rate": 1.551245389365494e-06,
      "loss": 0.7108,
      "step": 16644
    },
    {
      "epoch": 0.6898338099382486,
      "grad_norm": 0.4320739805698395,
      "learning_rate": 1.5510381698371255e-06,
      "loss": 0.656,
      "step": 16645
    },
    {
      "epoch": 0.6898752538439222,
      "grad_norm": 0.42302489280700684,
      "learning_rate": 1.5508309503087574e-06,
      "loss": 0.6881,
      "step": 16646
    },
    {
      "epoch": 0.689916697749596,
      "grad_norm": 0.42559999227523804,
      "learning_rate": 1.550623730780389e-06,
      "loss": 0.6848,
      "step": 16647
    },
    {
      "epoch": 0.6899581416552696,
      "grad_norm": 0.42445310950279236,
      "learning_rate": 1.5504165112520206e-06,
      "loss": 0.665,
      "step": 16648
    },
    {
      "epoch": 0.6899995855609433,
      "grad_norm": 0.4359683692455292,
      "learning_rate": 1.5502092917236521e-06,
      "loss": 0.6562,
      "step": 16649
    },
    {
      "epoch": 0.6900410294666169,
      "grad_norm": 0.4109833836555481,
      "learning_rate": 1.5500020721952837e-06,
      "loss": 0.6897,
      "step": 16650
    },
    {
      "epoch": 0.6900824733722906,
      "grad_norm": 0.4030595123767853,
      "learning_rate": 1.5497948526669153e-06,
      "loss": 0.6656,
      "step": 16651
    },
    {
      "epoch": 0.6901239172779643,
      "grad_norm": 0.4421241581439972,
      "learning_rate": 1.549587633138547e-06,
      "loss": 0.653,
      "step": 16652
    },
    {
      "epoch": 0.6901653611836379,
      "grad_norm": 0.4107217788696289,
      "learning_rate": 1.5493804136101787e-06,
      "loss": 0.6704,
      "step": 16653
    },
    {
      "epoch": 0.6902068050893116,
      "grad_norm": 0.4132261276245117,
      "learning_rate": 1.5491731940818103e-06,
      "loss": 0.6769,
      "step": 16654
    },
    {
      "epoch": 0.6902482489949853,
      "grad_norm": 0.40674081444740295,
      "learning_rate": 1.5489659745534422e-06,
      "loss": 0.6672,
      "step": 16655
    },
    {
      "epoch": 0.690289692900659,
      "grad_norm": 0.41795578598976135,
      "learning_rate": 1.5487587550250738e-06,
      "loss": 0.7114,
      "step": 16656
    },
    {
      "epoch": 0.6903311368063326,
      "grad_norm": 0.4428230822086334,
      "learning_rate": 1.5485515354967054e-06,
      "loss": 0.7246,
      "step": 16657
    },
    {
      "epoch": 0.6903725807120062,
      "grad_norm": 0.3923572301864624,
      "learning_rate": 1.548344315968337e-06,
      "loss": 0.6683,
      "step": 16658
    },
    {
      "epoch": 0.69041402461768,
      "grad_norm": 0.43894433975219727,
      "learning_rate": 1.5481370964399685e-06,
      "loss": 0.6902,
      "step": 16659
    },
    {
      "epoch": 0.6904554685233536,
      "grad_norm": 0.4109112024307251,
      "learning_rate": 1.5479298769116001e-06,
      "loss": 0.6685,
      "step": 16660
    },
    {
      "epoch": 0.6904969124290273,
      "grad_norm": 0.41837331652641296,
      "learning_rate": 1.547722657383232e-06,
      "loss": 0.7336,
      "step": 16661
    },
    {
      "epoch": 0.690538356334701,
      "grad_norm": 0.40523672103881836,
      "learning_rate": 1.5475154378548635e-06,
      "loss": 0.684,
      "step": 16662
    },
    {
      "epoch": 0.6905798002403747,
      "grad_norm": 0.4326830506324768,
      "learning_rate": 1.5473082183264951e-06,
      "loss": 0.7327,
      "step": 16663
    },
    {
      "epoch": 0.6906212441460483,
      "grad_norm": 0.4027855396270752,
      "learning_rate": 1.547100998798127e-06,
      "loss": 0.6208,
      "step": 16664
    },
    {
      "epoch": 0.690662688051722,
      "grad_norm": 0.3977644443511963,
      "learning_rate": 1.5468937792697586e-06,
      "loss": 0.6669,
      "step": 16665
    },
    {
      "epoch": 0.6907041319573957,
      "grad_norm": 0.45860642194747925,
      "learning_rate": 1.5466865597413902e-06,
      "loss": 0.7107,
      "step": 16666
    },
    {
      "epoch": 0.6907455758630693,
      "grad_norm": 0.39222654700279236,
      "learning_rate": 1.5464793402130217e-06,
      "loss": 0.6348,
      "step": 16667
    },
    {
      "epoch": 0.690787019768743,
      "grad_norm": 0.45074620842933655,
      "learning_rate": 1.5462721206846533e-06,
      "loss": 0.7476,
      "step": 16668
    },
    {
      "epoch": 0.6908284636744166,
      "grad_norm": 0.4063180685043335,
      "learning_rate": 1.546064901156285e-06,
      "loss": 0.6587,
      "step": 16669
    },
    {
      "epoch": 0.6908699075800904,
      "grad_norm": 0.40964141488075256,
      "learning_rate": 1.5458576816279168e-06,
      "loss": 0.6631,
      "step": 16670
    },
    {
      "epoch": 0.690911351485764,
      "grad_norm": 0.42978546023368835,
      "learning_rate": 1.5456504620995483e-06,
      "loss": 0.6721,
      "step": 16671
    },
    {
      "epoch": 0.6909527953914377,
      "grad_norm": 0.4148223400115967,
      "learning_rate": 1.5454432425711802e-06,
      "loss": 0.7114,
      "step": 16672
    },
    {
      "epoch": 0.6909942392971113,
      "grad_norm": 0.40420371294021606,
      "learning_rate": 1.5452360230428118e-06,
      "loss": 0.6989,
      "step": 16673
    },
    {
      "epoch": 0.6910356832027851,
      "grad_norm": 0.39026448130607605,
      "learning_rate": 1.5450288035144434e-06,
      "loss": 0.6561,
      "step": 16674
    },
    {
      "epoch": 0.6910771271084587,
      "grad_norm": 0.37360256910324097,
      "learning_rate": 1.544821583986075e-06,
      "loss": 0.6057,
      "step": 16675
    },
    {
      "epoch": 0.6911185710141323,
      "grad_norm": 0.3920608460903168,
      "learning_rate": 1.5446143644577065e-06,
      "loss": 0.678,
      "step": 16676
    },
    {
      "epoch": 0.691160014919806,
      "grad_norm": 0.4223755896091461,
      "learning_rate": 1.5444071449293381e-06,
      "loss": 0.7416,
      "step": 16677
    },
    {
      "epoch": 0.6912014588254797,
      "grad_norm": 0.4065549075603485,
      "learning_rate": 1.54419992540097e-06,
      "loss": 0.7261,
      "step": 16678
    },
    {
      "epoch": 0.6912429027311534,
      "grad_norm": 0.39833980798721313,
      "learning_rate": 1.5439927058726016e-06,
      "loss": 0.6622,
      "step": 16679
    },
    {
      "epoch": 0.691284346636827,
      "grad_norm": 0.4140157997608185,
      "learning_rate": 1.5437854863442331e-06,
      "loss": 0.649,
      "step": 16680
    },
    {
      "epoch": 0.6913257905425008,
      "grad_norm": 0.42184978723526,
      "learning_rate": 1.543578266815865e-06,
      "loss": 0.688,
      "step": 16681
    },
    {
      "epoch": 0.6913672344481744,
      "grad_norm": 0.39209532737731934,
      "learning_rate": 1.5433710472874966e-06,
      "loss": 0.6313,
      "step": 16682
    },
    {
      "epoch": 0.6914086783538481,
      "grad_norm": 0.4041922688484192,
      "learning_rate": 1.5431638277591282e-06,
      "loss": 0.6193,
      "step": 16683
    },
    {
      "epoch": 0.6914501222595217,
      "grad_norm": 0.40010058879852295,
      "learning_rate": 1.5429566082307598e-06,
      "loss": 0.6494,
      "step": 16684
    },
    {
      "epoch": 0.6914915661651954,
      "grad_norm": 0.42123520374298096,
      "learning_rate": 1.5427493887023913e-06,
      "loss": 0.662,
      "step": 16685
    },
    {
      "epoch": 0.6915330100708691,
      "grad_norm": 0.4396835267543793,
      "learning_rate": 1.542542169174023e-06,
      "loss": 0.6631,
      "step": 16686
    },
    {
      "epoch": 0.6915744539765427,
      "grad_norm": 0.4054194688796997,
      "learning_rate": 1.5423349496456548e-06,
      "loss": 0.6172,
      "step": 16687
    },
    {
      "epoch": 0.6916158978822164,
      "grad_norm": 0.4175710380077362,
      "learning_rate": 1.5421277301172864e-06,
      "loss": 0.6493,
      "step": 16688
    },
    {
      "epoch": 0.6916573417878901,
      "grad_norm": 0.43800342082977295,
      "learning_rate": 1.541920510588918e-06,
      "loss": 0.7166,
      "step": 16689
    },
    {
      "epoch": 0.6916987856935638,
      "grad_norm": 0.47093433141708374,
      "learning_rate": 1.5417132910605498e-06,
      "loss": 0.667,
      "step": 16690
    },
    {
      "epoch": 0.6917402295992374,
      "grad_norm": 0.43820780515670776,
      "learning_rate": 1.5415060715321814e-06,
      "loss": 0.7324,
      "step": 16691
    },
    {
      "epoch": 0.6917816735049112,
      "grad_norm": 0.4080968499183655,
      "learning_rate": 1.541298852003813e-06,
      "loss": 0.6655,
      "step": 16692
    },
    {
      "epoch": 0.6918231174105848,
      "grad_norm": 0.40456321835517883,
      "learning_rate": 1.5410916324754446e-06,
      "loss": 0.6959,
      "step": 16693
    },
    {
      "epoch": 0.6918645613162584,
      "grad_norm": 0.410861074924469,
      "learning_rate": 1.5408844129470761e-06,
      "loss": 0.6814,
      "step": 16694
    },
    {
      "epoch": 0.6919060052219321,
      "grad_norm": 0.44444844126701355,
      "learning_rate": 1.540677193418708e-06,
      "loss": 0.6663,
      "step": 16695
    },
    {
      "epoch": 0.6919474491276058,
      "grad_norm": 0.40606433153152466,
      "learning_rate": 1.5404699738903396e-06,
      "loss": 0.6707,
      "step": 16696
    },
    {
      "epoch": 0.6919888930332795,
      "grad_norm": 0.4321768283843994,
      "learning_rate": 1.5402627543619712e-06,
      "loss": 0.6799,
      "step": 16697
    },
    {
      "epoch": 0.6920303369389531,
      "grad_norm": 0.44445449113845825,
      "learning_rate": 1.540055534833603e-06,
      "loss": 0.6613,
      "step": 16698
    },
    {
      "epoch": 0.6920717808446268,
      "grad_norm": 0.4180211126804352,
      "learning_rate": 1.5398483153052346e-06,
      "loss": 0.642,
      "step": 16699
    },
    {
      "epoch": 0.6921132247503005,
      "grad_norm": 0.4246850907802582,
      "learning_rate": 1.5396410957768662e-06,
      "loss": 0.7034,
      "step": 16700
    },
    {
      "epoch": 0.6921546686559742,
      "grad_norm": 0.42065727710723877,
      "learning_rate": 1.5394338762484978e-06,
      "loss": 0.682,
      "step": 16701
    },
    {
      "epoch": 0.6921961125616478,
      "grad_norm": 0.4103626012802124,
      "learning_rate": 1.5392266567201294e-06,
      "loss": 0.6591,
      "step": 16702
    },
    {
      "epoch": 0.6922375564673214,
      "grad_norm": 0.4429527521133423,
      "learning_rate": 1.5390194371917612e-06,
      "loss": 0.6428,
      "step": 16703
    },
    {
      "epoch": 0.6922790003729952,
      "grad_norm": 0.4589152932167053,
      "learning_rate": 1.5388122176633928e-06,
      "loss": 0.6707,
      "step": 16704
    },
    {
      "epoch": 0.6923204442786688,
      "grad_norm": 0.4238857328891754,
      "learning_rate": 1.5386049981350244e-06,
      "loss": 0.699,
      "step": 16705
    },
    {
      "epoch": 0.6923618881843425,
      "grad_norm": 0.41641494631767273,
      "learning_rate": 1.538397778606656e-06,
      "loss": 0.6094,
      "step": 16706
    },
    {
      "epoch": 0.6924033320900161,
      "grad_norm": 0.42477741837501526,
      "learning_rate": 1.5381905590782878e-06,
      "loss": 0.6698,
      "step": 16707
    },
    {
      "epoch": 0.6924447759956899,
      "grad_norm": 0.40682101249694824,
      "learning_rate": 1.5379833395499194e-06,
      "loss": 0.6687,
      "step": 16708
    },
    {
      "epoch": 0.6924862199013635,
      "grad_norm": 0.43637245893478394,
      "learning_rate": 1.537776120021551e-06,
      "loss": 0.6957,
      "step": 16709
    },
    {
      "epoch": 0.6925276638070372,
      "grad_norm": 0.413145512342453,
      "learning_rate": 1.5375689004931826e-06,
      "loss": 0.7073,
      "step": 16710
    },
    {
      "epoch": 0.6925691077127109,
      "grad_norm": 0.40275755524635315,
      "learning_rate": 1.5373616809648142e-06,
      "loss": 0.6587,
      "step": 16711
    },
    {
      "epoch": 0.6926105516183845,
      "grad_norm": 0.4630493223667145,
      "learning_rate": 1.537154461436446e-06,
      "loss": 0.6951,
      "step": 16712
    },
    {
      "epoch": 0.6926519955240582,
      "grad_norm": 0.3768662214279175,
      "learning_rate": 1.5369472419080776e-06,
      "loss": 0.708,
      "step": 16713
    },
    {
      "epoch": 0.6926934394297318,
      "grad_norm": 0.4116923213005066,
      "learning_rate": 1.5367400223797092e-06,
      "loss": 0.6056,
      "step": 16714
    },
    {
      "epoch": 0.6927348833354056,
      "grad_norm": 0.42894136905670166,
      "learning_rate": 1.5365328028513408e-06,
      "loss": 0.6782,
      "step": 16715
    },
    {
      "epoch": 0.6927763272410792,
      "grad_norm": 0.4153902530670166,
      "learning_rate": 1.5363255833229726e-06,
      "loss": 0.6897,
      "step": 16716
    },
    {
      "epoch": 0.6928177711467529,
      "grad_norm": 0.43107399344444275,
      "learning_rate": 1.5361183637946042e-06,
      "loss": 0.7162,
      "step": 16717
    },
    {
      "epoch": 0.6928592150524265,
      "grad_norm": 0.39449232816696167,
      "learning_rate": 1.5359111442662358e-06,
      "loss": 0.6477,
      "step": 16718
    },
    {
      "epoch": 0.6929006589581002,
      "grad_norm": 0.45432373881340027,
      "learning_rate": 1.5357039247378674e-06,
      "loss": 0.6836,
      "step": 16719
    },
    {
      "epoch": 0.6929421028637739,
      "grad_norm": 0.42863452434539795,
      "learning_rate": 1.5354967052094992e-06,
      "loss": 0.7517,
      "step": 16720
    },
    {
      "epoch": 0.6929835467694475,
      "grad_norm": 0.4364698827266693,
      "learning_rate": 1.5352894856811308e-06,
      "loss": 0.6488,
      "step": 16721
    },
    {
      "epoch": 0.6930249906751212,
      "grad_norm": 0.37991440296173096,
      "learning_rate": 1.5350822661527624e-06,
      "loss": 0.6276,
      "step": 16722
    },
    {
      "epoch": 0.6930664345807949,
      "grad_norm": 0.4021202623844147,
      "learning_rate": 1.534875046624394e-06,
      "loss": 0.7272,
      "step": 16723
    },
    {
      "epoch": 0.6931078784864686,
      "grad_norm": 0.4254559874534607,
      "learning_rate": 1.5346678270960256e-06,
      "loss": 0.7212,
      "step": 16724
    },
    {
      "epoch": 0.6931493223921422,
      "grad_norm": 0.41879189014434814,
      "learning_rate": 1.5344606075676574e-06,
      "loss": 0.6898,
      "step": 16725
    },
    {
      "epoch": 0.693190766297816,
      "grad_norm": 0.4046343266963959,
      "learning_rate": 1.534253388039289e-06,
      "loss": 0.6903,
      "step": 16726
    },
    {
      "epoch": 0.6932322102034896,
      "grad_norm": 0.4003014862537384,
      "learning_rate": 1.5340461685109206e-06,
      "loss": 0.6875,
      "step": 16727
    },
    {
      "epoch": 0.6932736541091632,
      "grad_norm": 0.40442928671836853,
      "learning_rate": 1.5338389489825522e-06,
      "loss": 0.6763,
      "step": 16728
    },
    {
      "epoch": 0.6933150980148369,
      "grad_norm": 0.4136636555194855,
      "learning_rate": 1.533631729454184e-06,
      "loss": 0.6602,
      "step": 16729
    },
    {
      "epoch": 0.6933565419205105,
      "grad_norm": 0.39174598455429077,
      "learning_rate": 1.5334245099258156e-06,
      "loss": 0.6753,
      "step": 16730
    },
    {
      "epoch": 0.6933979858261843,
      "grad_norm": 0.4188128411769867,
      "learning_rate": 1.5332172903974472e-06,
      "loss": 0.6621,
      "step": 16731
    },
    {
      "epoch": 0.6934394297318579,
      "grad_norm": 0.42641279101371765,
      "learning_rate": 1.5330100708690788e-06,
      "loss": 0.6665,
      "step": 16732
    },
    {
      "epoch": 0.6934808736375316,
      "grad_norm": 0.4473981261253357,
      "learning_rate": 1.5328028513407106e-06,
      "loss": 0.7188,
      "step": 16733
    },
    {
      "epoch": 0.6935223175432053,
      "grad_norm": 0.45557376742362976,
      "learning_rate": 1.5325956318123422e-06,
      "loss": 0.7129,
      "step": 16734
    },
    {
      "epoch": 0.693563761448879,
      "grad_norm": 0.4230588674545288,
      "learning_rate": 1.5323884122839738e-06,
      "loss": 0.6581,
      "step": 16735
    },
    {
      "epoch": 0.6936052053545526,
      "grad_norm": 0.38401761651039124,
      "learning_rate": 1.5321811927556054e-06,
      "loss": 0.6869,
      "step": 16736
    },
    {
      "epoch": 0.6936466492602262,
      "grad_norm": 0.4082692563533783,
      "learning_rate": 1.5319739732272372e-06,
      "loss": 0.6768,
      "step": 16737
    },
    {
      "epoch": 0.6936880931659,
      "grad_norm": 0.39676159620285034,
      "learning_rate": 1.5317667536988688e-06,
      "loss": 0.677,
      "step": 16738
    },
    {
      "epoch": 0.6937295370715736,
      "grad_norm": 0.4020856022834778,
      "learning_rate": 1.5315595341705004e-06,
      "loss": 0.7216,
      "step": 16739
    },
    {
      "epoch": 0.6937709809772473,
      "grad_norm": 0.4189518094062805,
      "learning_rate": 1.531352314642132e-06,
      "loss": 0.6577,
      "step": 16740
    },
    {
      "epoch": 0.6938124248829209,
      "grad_norm": 0.39654698967933655,
      "learning_rate": 1.5311450951137636e-06,
      "loss": 0.6797,
      "step": 16741
    },
    {
      "epoch": 0.6938538687885947,
      "grad_norm": 0.4032168984413147,
      "learning_rate": 1.5309378755853954e-06,
      "loss": 0.6045,
      "step": 16742
    },
    {
      "epoch": 0.6938953126942683,
      "grad_norm": 0.389242947101593,
      "learning_rate": 1.530730656057027e-06,
      "loss": 0.6562,
      "step": 16743
    },
    {
      "epoch": 0.693936756599942,
      "grad_norm": 0.4361919164657593,
      "learning_rate": 1.5305234365286586e-06,
      "loss": 0.6545,
      "step": 16744
    },
    {
      "epoch": 0.6939782005056156,
      "grad_norm": 0.3848486840724945,
      "learning_rate": 1.5303162170002902e-06,
      "loss": 0.6483,
      "step": 16745
    },
    {
      "epoch": 0.6940196444112893,
      "grad_norm": 0.39408209919929504,
      "learning_rate": 1.530108997471922e-06,
      "loss": 0.6943,
      "step": 16746
    },
    {
      "epoch": 0.694061088316963,
      "grad_norm": 0.418505996465683,
      "learning_rate": 1.5299017779435536e-06,
      "loss": 0.6986,
      "step": 16747
    },
    {
      "epoch": 0.6941025322226366,
      "grad_norm": 0.40993648767471313,
      "learning_rate": 1.5296945584151852e-06,
      "loss": 0.6577,
      "step": 16748
    },
    {
      "epoch": 0.6941439761283104,
      "grad_norm": 0.4200143814086914,
      "learning_rate": 1.5294873388868168e-06,
      "loss": 0.7361,
      "step": 16749
    },
    {
      "epoch": 0.694185420033984,
      "grad_norm": 0.41309866309165955,
      "learning_rate": 1.5292801193584484e-06,
      "loss": 0.7214,
      "step": 16750
    },
    {
      "epoch": 0.6942268639396577,
      "grad_norm": 0.41637322306632996,
      "learning_rate": 1.5290728998300802e-06,
      "loss": 0.6411,
      "step": 16751
    },
    {
      "epoch": 0.6942683078453313,
      "grad_norm": 0.4240413308143616,
      "learning_rate": 1.5288656803017118e-06,
      "loss": 0.6815,
      "step": 16752
    },
    {
      "epoch": 0.6943097517510051,
      "grad_norm": 0.41010725498199463,
      "learning_rate": 1.5286584607733434e-06,
      "loss": 0.6743,
      "step": 16753
    },
    {
      "epoch": 0.6943511956566787,
      "grad_norm": 0.4233991205692291,
      "learning_rate": 1.5284512412449752e-06,
      "loss": 0.7251,
      "step": 16754
    },
    {
      "epoch": 0.6943926395623523,
      "grad_norm": 0.4166107773780823,
      "learning_rate": 1.5282440217166068e-06,
      "loss": 0.6938,
      "step": 16755
    },
    {
      "epoch": 0.694434083468026,
      "grad_norm": 0.4158385097980499,
      "learning_rate": 1.5280368021882384e-06,
      "loss": 0.6731,
      "step": 16756
    },
    {
      "epoch": 0.6944755273736997,
      "grad_norm": 0.4258190393447876,
      "learning_rate": 1.52782958265987e-06,
      "loss": 0.6572,
      "step": 16757
    },
    {
      "epoch": 0.6945169712793734,
      "grad_norm": 0.4180450737476349,
      "learning_rate": 1.5276223631315016e-06,
      "loss": 0.6733,
      "step": 16758
    },
    {
      "epoch": 0.694558415185047,
      "grad_norm": 0.41732221841812134,
      "learning_rate": 1.5274151436031334e-06,
      "loss": 0.6731,
      "step": 16759
    },
    {
      "epoch": 0.6945998590907208,
      "grad_norm": 0.41948410868644714,
      "learning_rate": 1.527207924074765e-06,
      "loss": 0.6478,
      "step": 16760
    },
    {
      "epoch": 0.6946413029963944,
      "grad_norm": 0.39481037855148315,
      "learning_rate": 1.5270007045463966e-06,
      "loss": 0.7046,
      "step": 16761
    },
    {
      "epoch": 0.6946827469020681,
      "grad_norm": 0.41728588938713074,
      "learning_rate": 1.5267934850180282e-06,
      "loss": 0.6885,
      "step": 16762
    },
    {
      "epoch": 0.6947241908077417,
      "grad_norm": 0.41394034028053284,
      "learning_rate": 1.52658626548966e-06,
      "loss": 0.6515,
      "step": 16763
    },
    {
      "epoch": 0.6947656347134153,
      "grad_norm": 0.4248751401901245,
      "learning_rate": 1.5263790459612916e-06,
      "loss": 0.709,
      "step": 16764
    },
    {
      "epoch": 0.6948070786190891,
      "grad_norm": 0.3850202262401581,
      "learning_rate": 1.5261718264329232e-06,
      "loss": 0.6534,
      "step": 16765
    },
    {
      "epoch": 0.6948485225247627,
      "grad_norm": 0.44443872570991516,
      "learning_rate": 1.5259646069045548e-06,
      "loss": 0.7002,
      "step": 16766
    },
    {
      "epoch": 0.6948899664304364,
      "grad_norm": 0.4599267840385437,
      "learning_rate": 1.5257573873761864e-06,
      "loss": 0.6987,
      "step": 16767
    },
    {
      "epoch": 0.6949314103361101,
      "grad_norm": 0.4172816276550293,
      "learning_rate": 1.5255501678478182e-06,
      "loss": 0.6709,
      "step": 16768
    },
    {
      "epoch": 0.6949728542417838,
      "grad_norm": 0.4471736252307892,
      "learning_rate": 1.5253429483194498e-06,
      "loss": 0.696,
      "step": 16769
    },
    {
      "epoch": 0.6950142981474574,
      "grad_norm": 0.4285956919193268,
      "learning_rate": 1.5251357287910814e-06,
      "loss": 0.6322,
      "step": 16770
    },
    {
      "epoch": 0.6950557420531311,
      "grad_norm": 0.39196693897247314,
      "learning_rate": 1.5249285092627132e-06,
      "loss": 0.6858,
      "step": 16771
    },
    {
      "epoch": 0.6950971859588048,
      "grad_norm": 0.39466962218284607,
      "learning_rate": 1.5247212897343448e-06,
      "loss": 0.6902,
      "step": 16772
    },
    {
      "epoch": 0.6951386298644784,
      "grad_norm": 0.43944862484931946,
      "learning_rate": 1.5245140702059764e-06,
      "loss": 0.6694,
      "step": 16773
    },
    {
      "epoch": 0.6951800737701521,
      "grad_norm": 0.4425968825817108,
      "learning_rate": 1.524306850677608e-06,
      "loss": 0.6812,
      "step": 16774
    },
    {
      "epoch": 0.6952215176758257,
      "grad_norm": 0.4463377296924591,
      "learning_rate": 1.5240996311492396e-06,
      "loss": 0.6783,
      "step": 16775
    },
    {
      "epoch": 0.6952629615814995,
      "grad_norm": 0.37946656346321106,
      "learning_rate": 1.5238924116208712e-06,
      "loss": 0.6787,
      "step": 16776
    },
    {
      "epoch": 0.6953044054871731,
      "grad_norm": 0.4161308705806732,
      "learning_rate": 1.523685192092503e-06,
      "loss": 0.6813,
      "step": 16777
    },
    {
      "epoch": 0.6953458493928468,
      "grad_norm": 0.4392637610435486,
      "learning_rate": 1.5234779725641346e-06,
      "loss": 0.6874,
      "step": 16778
    },
    {
      "epoch": 0.6953872932985204,
      "grad_norm": 0.4265557825565338,
      "learning_rate": 1.5232707530357662e-06,
      "loss": 0.6833,
      "step": 16779
    },
    {
      "epoch": 0.6954287372041941,
      "grad_norm": 0.4445050060749054,
      "learning_rate": 1.523063533507398e-06,
      "loss": 0.6892,
      "step": 16780
    },
    {
      "epoch": 0.6954701811098678,
      "grad_norm": 0.42713114619255066,
      "learning_rate": 1.5228563139790296e-06,
      "loss": 0.6965,
      "step": 16781
    },
    {
      "epoch": 0.6955116250155414,
      "grad_norm": 0.3750733733177185,
      "learning_rate": 1.5226490944506612e-06,
      "loss": 0.6562,
      "step": 16782
    },
    {
      "epoch": 0.6955530689212152,
      "grad_norm": 0.42506036162376404,
      "learning_rate": 1.5224418749222928e-06,
      "loss": 0.6718,
      "step": 16783
    },
    {
      "epoch": 0.6955945128268888,
      "grad_norm": 0.42093586921691895,
      "learning_rate": 1.5222346553939244e-06,
      "loss": 0.7211,
      "step": 16784
    },
    {
      "epoch": 0.6956359567325625,
      "grad_norm": 0.3898966312408447,
      "learning_rate": 1.522027435865556e-06,
      "loss": 0.6814,
      "step": 16785
    },
    {
      "epoch": 0.6956774006382361,
      "grad_norm": 0.4490966796875,
      "learning_rate": 1.5218202163371878e-06,
      "loss": 0.6709,
      "step": 16786
    },
    {
      "epoch": 0.6957188445439099,
      "grad_norm": 0.4603123664855957,
      "learning_rate": 1.5216129968088194e-06,
      "loss": 0.7029,
      "step": 16787
    },
    {
      "epoch": 0.6957602884495835,
      "grad_norm": 0.4312245547771454,
      "learning_rate": 1.5214057772804512e-06,
      "loss": 0.7288,
      "step": 16788
    },
    {
      "epoch": 0.6958017323552571,
      "grad_norm": 0.4495743215084076,
      "learning_rate": 1.5211985577520828e-06,
      "loss": 0.7168,
      "step": 16789
    },
    {
      "epoch": 0.6958431762609308,
      "grad_norm": 0.42483657598495483,
      "learning_rate": 1.5209913382237144e-06,
      "loss": 0.6938,
      "step": 16790
    },
    {
      "epoch": 0.6958846201666045,
      "grad_norm": 0.3786216080188751,
      "learning_rate": 1.520784118695346e-06,
      "loss": 0.6505,
      "step": 16791
    },
    {
      "epoch": 0.6959260640722782,
      "grad_norm": 0.4108757972717285,
      "learning_rate": 1.5205768991669776e-06,
      "loss": 0.6202,
      "step": 16792
    },
    {
      "epoch": 0.6959675079779518,
      "grad_norm": 0.42579811811447144,
      "learning_rate": 1.5203696796386092e-06,
      "loss": 0.6873,
      "step": 16793
    },
    {
      "epoch": 0.6960089518836255,
      "grad_norm": 0.4422059655189514,
      "learning_rate": 1.520162460110241e-06,
      "loss": 0.6714,
      "step": 16794
    },
    {
      "epoch": 0.6960503957892992,
      "grad_norm": 0.3904128074645996,
      "learning_rate": 1.5199552405818726e-06,
      "loss": 0.657,
      "step": 16795
    },
    {
      "epoch": 0.6960918396949729,
      "grad_norm": 0.42671409249305725,
      "learning_rate": 1.5197480210535044e-06,
      "loss": 0.7234,
      "step": 16796
    },
    {
      "epoch": 0.6961332836006465,
      "grad_norm": 0.450268417596817,
      "learning_rate": 1.519540801525136e-06,
      "loss": 0.7026,
      "step": 16797
    },
    {
      "epoch": 0.6961747275063201,
      "grad_norm": 0.43762993812561035,
      "learning_rate": 1.5193335819967676e-06,
      "loss": 0.6362,
      "step": 16798
    },
    {
      "epoch": 0.6962161714119939,
      "grad_norm": 0.40904274582862854,
      "learning_rate": 1.5191263624683992e-06,
      "loss": 0.661,
      "step": 16799
    },
    {
      "epoch": 0.6962576153176675,
      "grad_norm": 0.4229663610458374,
      "learning_rate": 1.5189191429400308e-06,
      "loss": 0.7126,
      "step": 16800
    },
    {
      "epoch": 0.6962990592233412,
      "grad_norm": 0.4693845808506012,
      "learning_rate": 1.5187119234116624e-06,
      "loss": 0.7423,
      "step": 16801
    },
    {
      "epoch": 0.6963405031290149,
      "grad_norm": 0.38161811232566833,
      "learning_rate": 1.518504703883294e-06,
      "loss": 0.6063,
      "step": 16802
    },
    {
      "epoch": 0.6963819470346886,
      "grad_norm": 0.4070481061935425,
      "learning_rate": 1.5182974843549258e-06,
      "loss": 0.6562,
      "step": 16803
    },
    {
      "epoch": 0.6964233909403622,
      "grad_norm": 0.42256149649620056,
      "learning_rate": 1.5180902648265574e-06,
      "loss": 0.6943,
      "step": 16804
    },
    {
      "epoch": 0.6964648348460359,
      "grad_norm": 0.3968719244003296,
      "learning_rate": 1.5178830452981892e-06,
      "loss": 0.6415,
      "step": 16805
    },
    {
      "epoch": 0.6965062787517096,
      "grad_norm": 0.4523000717163086,
      "learning_rate": 1.5176758257698208e-06,
      "loss": 0.7175,
      "step": 16806
    },
    {
      "epoch": 0.6965477226573832,
      "grad_norm": 0.40705400705337524,
      "learning_rate": 1.5174686062414524e-06,
      "loss": 0.6466,
      "step": 16807
    },
    {
      "epoch": 0.6965891665630569,
      "grad_norm": 0.39460861682891846,
      "learning_rate": 1.517261386713084e-06,
      "loss": 0.6168,
      "step": 16808
    },
    {
      "epoch": 0.6966306104687305,
      "grad_norm": 0.3971591293811798,
      "learning_rate": 1.5170541671847156e-06,
      "loss": 0.6415,
      "step": 16809
    },
    {
      "epoch": 0.6966720543744043,
      "grad_norm": 0.42209601402282715,
      "learning_rate": 1.5168469476563472e-06,
      "loss": 0.6923,
      "step": 16810
    },
    {
      "epoch": 0.6967134982800779,
      "grad_norm": 0.4125151038169861,
      "learning_rate": 1.5166397281279788e-06,
      "loss": 0.6766,
      "step": 16811
    },
    {
      "epoch": 0.6967549421857516,
      "grad_norm": 0.39692404866218567,
      "learning_rate": 1.5164325085996106e-06,
      "loss": 0.6671,
      "step": 16812
    },
    {
      "epoch": 0.6967963860914252,
      "grad_norm": 0.4617835581302643,
      "learning_rate": 1.5162252890712424e-06,
      "loss": 0.6876,
      "step": 16813
    },
    {
      "epoch": 0.696837829997099,
      "grad_norm": 0.4173271656036377,
      "learning_rate": 1.516018069542874e-06,
      "loss": 0.6738,
      "step": 16814
    },
    {
      "epoch": 0.6968792739027726,
      "grad_norm": 0.4262498915195465,
      "learning_rate": 1.5158108500145056e-06,
      "loss": 0.7068,
      "step": 16815
    },
    {
      "epoch": 0.6969207178084462,
      "grad_norm": 0.48951685428619385,
      "learning_rate": 1.5156036304861372e-06,
      "loss": 0.7024,
      "step": 16816
    },
    {
      "epoch": 0.69696216171412,
      "grad_norm": 0.4242796301841736,
      "learning_rate": 1.5153964109577688e-06,
      "loss": 0.6973,
      "step": 16817
    },
    {
      "epoch": 0.6970036056197936,
      "grad_norm": 0.3965175449848175,
      "learning_rate": 1.5151891914294004e-06,
      "loss": 0.6591,
      "step": 16818
    },
    {
      "epoch": 0.6970450495254673,
      "grad_norm": 0.42197319865226746,
      "learning_rate": 1.514981971901032e-06,
      "loss": 0.6859,
      "step": 16819
    },
    {
      "epoch": 0.6970864934311409,
      "grad_norm": 0.3900800347328186,
      "learning_rate": 1.5147747523726638e-06,
      "loss": 0.6915,
      "step": 16820
    },
    {
      "epoch": 0.6971279373368147,
      "grad_norm": 0.40391185879707336,
      "learning_rate": 1.5145675328442954e-06,
      "loss": 0.6527,
      "step": 16821
    },
    {
      "epoch": 0.6971693812424883,
      "grad_norm": 0.39946305751800537,
      "learning_rate": 1.5143603133159272e-06,
      "loss": 0.6371,
      "step": 16822
    },
    {
      "epoch": 0.697210825148162,
      "grad_norm": 0.4062614142894745,
      "learning_rate": 1.5141530937875588e-06,
      "loss": 0.6676,
      "step": 16823
    },
    {
      "epoch": 0.6972522690538356,
      "grad_norm": 0.42598265409469604,
      "learning_rate": 1.5139458742591904e-06,
      "loss": 0.708,
      "step": 16824
    },
    {
      "epoch": 0.6972937129595093,
      "grad_norm": 0.415072500705719,
      "learning_rate": 1.513738654730822e-06,
      "loss": 0.6641,
      "step": 16825
    },
    {
      "epoch": 0.697335156865183,
      "grad_norm": 0.44375693798065186,
      "learning_rate": 1.5135314352024536e-06,
      "loss": 0.6763,
      "step": 16826
    },
    {
      "epoch": 0.6973766007708566,
      "grad_norm": 0.4355472922325134,
      "learning_rate": 1.5133242156740852e-06,
      "loss": 0.7122,
      "step": 16827
    },
    {
      "epoch": 0.6974180446765303,
      "grad_norm": 0.42468228936195374,
      "learning_rate": 1.5131169961457168e-06,
      "loss": 0.6796,
      "step": 16828
    },
    {
      "epoch": 0.697459488582204,
      "grad_norm": 0.41289567947387695,
      "learning_rate": 1.5129097766173486e-06,
      "loss": 0.6519,
      "step": 16829
    },
    {
      "epoch": 0.6975009324878777,
      "grad_norm": 0.38233935832977295,
      "learning_rate": 1.5127025570889804e-06,
      "loss": 0.6549,
      "step": 16830
    },
    {
      "epoch": 0.6975423763935513,
      "grad_norm": 0.41228243708610535,
      "learning_rate": 1.512495337560612e-06,
      "loss": 0.6597,
      "step": 16831
    },
    {
      "epoch": 0.6975838202992249,
      "grad_norm": 0.42211341857910156,
      "learning_rate": 1.5122881180322436e-06,
      "loss": 0.7007,
      "step": 16832
    },
    {
      "epoch": 0.6976252642048987,
      "grad_norm": 0.41498398780822754,
      "learning_rate": 1.5120808985038752e-06,
      "loss": 0.6923,
      "step": 16833
    },
    {
      "epoch": 0.6976667081105723,
      "grad_norm": 0.4597736895084381,
      "learning_rate": 1.5118736789755068e-06,
      "loss": 0.7036,
      "step": 16834
    },
    {
      "epoch": 0.697708152016246,
      "grad_norm": 0.3906548023223877,
      "learning_rate": 1.5116664594471384e-06,
      "loss": 0.6792,
      "step": 16835
    },
    {
      "epoch": 0.6977495959219197,
      "grad_norm": 0.4018573462963104,
      "learning_rate": 1.51145923991877e-06,
      "loss": 0.6367,
      "step": 16836
    },
    {
      "epoch": 0.6977910398275934,
      "grad_norm": 0.41260090470314026,
      "learning_rate": 1.5112520203904016e-06,
      "loss": 0.7161,
      "step": 16837
    },
    {
      "epoch": 0.697832483733267,
      "grad_norm": 0.40674540400505066,
      "learning_rate": 1.5110448008620334e-06,
      "loss": 0.7073,
      "step": 16838
    },
    {
      "epoch": 0.6978739276389407,
      "grad_norm": 0.41956791281700134,
      "learning_rate": 1.5108375813336652e-06,
      "loss": 0.6954,
      "step": 16839
    },
    {
      "epoch": 0.6979153715446144,
      "grad_norm": 0.40299227833747864,
      "learning_rate": 1.5106303618052968e-06,
      "loss": 0.6693,
      "step": 16840
    },
    {
      "epoch": 0.697956815450288,
      "grad_norm": 0.46463340520858765,
      "learning_rate": 1.5104231422769284e-06,
      "loss": 0.77,
      "step": 16841
    },
    {
      "epoch": 0.6979982593559617,
      "grad_norm": 0.3964928686618805,
      "learning_rate": 1.51021592274856e-06,
      "loss": 0.6885,
      "step": 16842
    },
    {
      "epoch": 0.6980397032616353,
      "grad_norm": 0.41980281472206116,
      "learning_rate": 1.5100087032201916e-06,
      "loss": 0.6509,
      "step": 16843
    },
    {
      "epoch": 0.6980811471673091,
      "grad_norm": 0.4317905902862549,
      "learning_rate": 1.5098014836918232e-06,
      "loss": 0.6395,
      "step": 16844
    },
    {
      "epoch": 0.6981225910729827,
      "grad_norm": 0.40261349081993103,
      "learning_rate": 1.5095942641634548e-06,
      "loss": 0.6709,
      "step": 16845
    },
    {
      "epoch": 0.6981640349786564,
      "grad_norm": 0.43510785698890686,
      "learning_rate": 1.5093870446350864e-06,
      "loss": 0.6847,
      "step": 16846
    },
    {
      "epoch": 0.69820547888433,
      "grad_norm": 0.42978084087371826,
      "learning_rate": 1.5091798251067184e-06,
      "loss": 0.6941,
      "step": 16847
    },
    {
      "epoch": 0.6982469227900038,
      "grad_norm": 0.3732667565345764,
      "learning_rate": 1.50897260557835e-06,
      "loss": 0.635,
      "step": 16848
    },
    {
      "epoch": 0.6982883666956774,
      "grad_norm": 0.4033155143260956,
      "learning_rate": 1.5087653860499816e-06,
      "loss": 0.6013,
      "step": 16849
    },
    {
      "epoch": 0.698329810601351,
      "grad_norm": 0.3825395107269287,
      "learning_rate": 1.5085581665216132e-06,
      "loss": 0.652,
      "step": 16850
    },
    {
      "epoch": 0.6983712545070248,
      "grad_norm": 0.38410305976867676,
      "learning_rate": 1.5083509469932448e-06,
      "loss": 0.6948,
      "step": 16851
    },
    {
      "epoch": 0.6984126984126984,
      "grad_norm": 0.38327160477638245,
      "learning_rate": 1.5081437274648764e-06,
      "loss": 0.6569,
      "step": 16852
    },
    {
      "epoch": 0.6984541423183721,
      "grad_norm": 0.4063495099544525,
      "learning_rate": 1.507936507936508e-06,
      "loss": 0.645,
      "step": 16853
    },
    {
      "epoch": 0.6984955862240457,
      "grad_norm": 0.3928200602531433,
      "learning_rate": 1.5077292884081396e-06,
      "loss": 0.6973,
      "step": 16854
    },
    {
      "epoch": 0.6985370301297195,
      "grad_norm": 0.41994142532348633,
      "learning_rate": 1.5075220688797714e-06,
      "loss": 0.6743,
      "step": 16855
    },
    {
      "epoch": 0.6985784740353931,
      "grad_norm": 0.38647305965423584,
      "learning_rate": 1.5073148493514032e-06,
      "loss": 0.6689,
      "step": 16856
    },
    {
      "epoch": 0.6986199179410668,
      "grad_norm": 0.43377214670181274,
      "learning_rate": 1.5071076298230348e-06,
      "loss": 0.7073,
      "step": 16857
    },
    {
      "epoch": 0.6986613618467404,
      "grad_norm": 0.432190865278244,
      "learning_rate": 1.5069004102946664e-06,
      "loss": 0.6925,
      "step": 16858
    },
    {
      "epoch": 0.6987028057524141,
      "grad_norm": 0.4318296015262604,
      "learning_rate": 1.506693190766298e-06,
      "loss": 0.6904,
      "step": 16859
    },
    {
      "epoch": 0.6987442496580878,
      "grad_norm": 0.4050523340702057,
      "learning_rate": 1.5064859712379296e-06,
      "loss": 0.6589,
      "step": 16860
    },
    {
      "epoch": 0.6987856935637614,
      "grad_norm": 0.40000247955322266,
      "learning_rate": 1.5062787517095612e-06,
      "loss": 0.6682,
      "step": 16861
    },
    {
      "epoch": 0.6988271374694351,
      "grad_norm": 0.406726598739624,
      "learning_rate": 1.5060715321811928e-06,
      "loss": 0.6846,
      "step": 16862
    },
    {
      "epoch": 0.6988685813751088,
      "grad_norm": 0.4119182825088501,
      "learning_rate": 1.5058643126528244e-06,
      "loss": 0.658,
      "step": 16863
    },
    {
      "epoch": 0.6989100252807825,
      "grad_norm": 0.4022960364818573,
      "learning_rate": 1.5056570931244564e-06,
      "loss": 0.6583,
      "step": 16864
    },
    {
      "epoch": 0.6989514691864561,
      "grad_norm": 0.3845854699611664,
      "learning_rate": 1.505449873596088e-06,
      "loss": 0.6912,
      "step": 16865
    },
    {
      "epoch": 0.6989929130921299,
      "grad_norm": 0.3953777849674225,
      "learning_rate": 1.5052426540677196e-06,
      "loss": 0.6174,
      "step": 16866
    },
    {
      "epoch": 0.6990343569978035,
      "grad_norm": 0.4566531181335449,
      "learning_rate": 1.5050354345393512e-06,
      "loss": 0.6931,
      "step": 16867
    },
    {
      "epoch": 0.6990758009034771,
      "grad_norm": 0.4375271201133728,
      "learning_rate": 1.5048282150109828e-06,
      "loss": 0.6443,
      "step": 16868
    },
    {
      "epoch": 0.6991172448091508,
      "grad_norm": 0.42125600576400757,
      "learning_rate": 1.5046209954826144e-06,
      "loss": 0.6792,
      "step": 16869
    },
    {
      "epoch": 0.6991586887148245,
      "grad_norm": 0.40954530239105225,
      "learning_rate": 1.504413775954246e-06,
      "loss": 0.6616,
      "step": 16870
    },
    {
      "epoch": 0.6992001326204982,
      "grad_norm": 0.4000544250011444,
      "learning_rate": 1.5042065564258776e-06,
      "loss": 0.6407,
      "step": 16871
    },
    {
      "epoch": 0.6992415765261718,
      "grad_norm": 0.4300186038017273,
      "learning_rate": 1.5039993368975092e-06,
      "loss": 0.7112,
      "step": 16872
    },
    {
      "epoch": 0.6992830204318455,
      "grad_norm": 0.44238343834877014,
      "learning_rate": 1.5037921173691412e-06,
      "loss": 0.6779,
      "step": 16873
    },
    {
      "epoch": 0.6993244643375192,
      "grad_norm": 0.406878799200058,
      "learning_rate": 1.5035848978407728e-06,
      "loss": 0.6429,
      "step": 16874
    },
    {
      "epoch": 0.6993659082431929,
      "grad_norm": 0.4408002197742462,
      "learning_rate": 1.5033776783124044e-06,
      "loss": 0.684,
      "step": 16875
    },
    {
      "epoch": 0.6994073521488665,
      "grad_norm": 0.414987176656723,
      "learning_rate": 1.503170458784036e-06,
      "loss": 0.6971,
      "step": 16876
    },
    {
      "epoch": 0.6994487960545401,
      "grad_norm": 0.5833993554115295,
      "learning_rate": 1.5029632392556676e-06,
      "loss": 0.6301,
      "step": 16877
    },
    {
      "epoch": 0.6994902399602139,
      "grad_norm": 0.4130208492279053,
      "learning_rate": 1.5027560197272992e-06,
      "loss": 0.6592,
      "step": 16878
    },
    {
      "epoch": 0.6995316838658875,
      "grad_norm": 0.4715719223022461,
      "learning_rate": 1.5025488001989308e-06,
      "loss": 0.6772,
      "step": 16879
    },
    {
      "epoch": 0.6995731277715612,
      "grad_norm": 0.39166057109832764,
      "learning_rate": 1.5023415806705624e-06,
      "loss": 0.668,
      "step": 16880
    },
    {
      "epoch": 0.6996145716772348,
      "grad_norm": 0.41646552085876465,
      "learning_rate": 1.5021343611421944e-06,
      "loss": 0.687,
      "step": 16881
    },
    {
      "epoch": 0.6996560155829086,
      "grad_norm": 0.43060335516929626,
      "learning_rate": 1.501927141613826e-06,
      "loss": 0.6909,
      "step": 16882
    },
    {
      "epoch": 0.6996974594885822,
      "grad_norm": 0.4006277322769165,
      "learning_rate": 1.5017199220854576e-06,
      "loss": 0.6444,
      "step": 16883
    },
    {
      "epoch": 0.6997389033942559,
      "grad_norm": 0.40410366654396057,
      "learning_rate": 1.5015127025570892e-06,
      "loss": 0.6575,
      "step": 16884
    },
    {
      "epoch": 0.6997803472999296,
      "grad_norm": 0.4085201919078827,
      "learning_rate": 1.5013054830287208e-06,
      "loss": 0.7256,
      "step": 16885
    },
    {
      "epoch": 0.6998217912056032,
      "grad_norm": 0.3965468108654022,
      "learning_rate": 1.5010982635003524e-06,
      "loss": 0.6622,
      "step": 16886
    },
    {
      "epoch": 0.6998632351112769,
      "grad_norm": 0.3922404646873474,
      "learning_rate": 1.500891043971984e-06,
      "loss": 0.6522,
      "step": 16887
    },
    {
      "epoch": 0.6999046790169505,
      "grad_norm": 0.4113626778125763,
      "learning_rate": 1.5006838244436156e-06,
      "loss": 0.6141,
      "step": 16888
    },
    {
      "epoch": 0.6999461229226243,
      "grad_norm": 0.3975231945514679,
      "learning_rate": 1.5004766049152472e-06,
      "loss": 0.6646,
      "step": 16889
    },
    {
      "epoch": 0.6999875668282979,
      "grad_norm": 0.40247538685798645,
      "learning_rate": 1.5002693853868792e-06,
      "loss": 0.6548,
      "step": 16890
    },
    {
      "epoch": 0.7000290107339716,
      "grad_norm": 0.3989601135253906,
      "learning_rate": 1.5000621658585108e-06,
      "loss": 0.6954,
      "step": 16891
    },
    {
      "epoch": 0.7000704546396452,
      "grad_norm": 0.42328518629074097,
      "learning_rate": 1.4998549463301424e-06,
      "loss": 0.6854,
      "step": 16892
    },
    {
      "epoch": 0.7001118985453189,
      "grad_norm": 0.40688613057136536,
      "learning_rate": 1.499647726801774e-06,
      "loss": 0.6627,
      "step": 16893
    },
    {
      "epoch": 0.7001533424509926,
      "grad_norm": 0.4467812180519104,
      "learning_rate": 1.4994405072734056e-06,
      "loss": 0.6686,
      "step": 16894
    },
    {
      "epoch": 0.7001947863566662,
      "grad_norm": 0.3951134979724884,
      "learning_rate": 1.4992332877450372e-06,
      "loss": 0.6543,
      "step": 16895
    },
    {
      "epoch": 0.7002362302623399,
      "grad_norm": 0.3915548622608185,
      "learning_rate": 1.4990260682166688e-06,
      "loss": 0.6696,
      "step": 16896
    },
    {
      "epoch": 0.7002776741680136,
      "grad_norm": 0.4269884526729584,
      "learning_rate": 1.4988188486883004e-06,
      "loss": 0.6206,
      "step": 16897
    },
    {
      "epoch": 0.7003191180736873,
      "grad_norm": 0.42337566614151,
      "learning_rate": 1.498611629159932e-06,
      "loss": 0.6331,
      "step": 16898
    },
    {
      "epoch": 0.7003605619793609,
      "grad_norm": 0.4235631227493286,
      "learning_rate": 1.498404409631564e-06,
      "loss": 0.6876,
      "step": 16899
    },
    {
      "epoch": 0.7004020058850347,
      "grad_norm": 0.41312679648399353,
      "learning_rate": 1.4981971901031956e-06,
      "loss": 0.657,
      "step": 16900
    },
    {
      "epoch": 0.7004434497907083,
      "grad_norm": 0.46847036480903625,
      "learning_rate": 1.4979899705748272e-06,
      "loss": 0.6699,
      "step": 16901
    },
    {
      "epoch": 0.7004848936963819,
      "grad_norm": 0.42012014985084534,
      "learning_rate": 1.4977827510464588e-06,
      "loss": 0.6663,
      "step": 16902
    },
    {
      "epoch": 0.7005263376020556,
      "grad_norm": 0.5709120631217957,
      "learning_rate": 1.4975755315180904e-06,
      "loss": 0.7129,
      "step": 16903
    },
    {
      "epoch": 0.7005677815077292,
      "grad_norm": 0.4306354224681854,
      "learning_rate": 1.497368311989722e-06,
      "loss": 0.6906,
      "step": 16904
    },
    {
      "epoch": 0.700609225413403,
      "grad_norm": 0.4431023597717285,
      "learning_rate": 1.4971610924613536e-06,
      "loss": 0.6997,
      "step": 16905
    },
    {
      "epoch": 0.7006506693190766,
      "grad_norm": 0.43004193902015686,
      "learning_rate": 1.4969538729329852e-06,
      "loss": 0.6741,
      "step": 16906
    },
    {
      "epoch": 0.7006921132247503,
      "grad_norm": 0.40990740060806274,
      "learning_rate": 1.4967466534046172e-06,
      "loss": 0.6495,
      "step": 16907
    },
    {
      "epoch": 0.700733557130424,
      "grad_norm": 0.3952885568141937,
      "learning_rate": 1.4965394338762488e-06,
      "loss": 0.6335,
      "step": 16908
    },
    {
      "epoch": 0.7007750010360977,
      "grad_norm": 0.4075387120246887,
      "learning_rate": 1.4963322143478804e-06,
      "loss": 0.6484,
      "step": 16909
    },
    {
      "epoch": 0.7008164449417713,
      "grad_norm": 0.40629440546035767,
      "learning_rate": 1.496124994819512e-06,
      "loss": 0.6713,
      "step": 16910
    },
    {
      "epoch": 0.7008578888474449,
      "grad_norm": 0.41931506991386414,
      "learning_rate": 1.4959177752911436e-06,
      "loss": 0.649,
      "step": 16911
    },
    {
      "epoch": 0.7008993327531187,
      "grad_norm": 0.41404739022254944,
      "learning_rate": 1.4957105557627752e-06,
      "loss": 0.6538,
      "step": 16912
    },
    {
      "epoch": 0.7009407766587923,
      "grad_norm": 0.4480193853378296,
      "learning_rate": 1.4955033362344068e-06,
      "loss": 0.7324,
      "step": 16913
    },
    {
      "epoch": 0.700982220564466,
      "grad_norm": 0.41701433062553406,
      "learning_rate": 1.4952961167060384e-06,
      "loss": 0.7278,
      "step": 16914
    },
    {
      "epoch": 0.7010236644701396,
      "grad_norm": 0.3988775312900543,
      "learning_rate": 1.49508889717767e-06,
      "loss": 0.6853,
      "step": 16915
    },
    {
      "epoch": 0.7010651083758134,
      "grad_norm": 0.4116692543029785,
      "learning_rate": 1.494881677649302e-06,
      "loss": 0.6633,
      "step": 16916
    },
    {
      "epoch": 0.701106552281487,
      "grad_norm": 0.43091800808906555,
      "learning_rate": 1.4946744581209336e-06,
      "loss": 0.6558,
      "step": 16917
    },
    {
      "epoch": 0.7011479961871607,
      "grad_norm": 0.4412301480770111,
      "learning_rate": 1.4944672385925652e-06,
      "loss": 0.6689,
      "step": 16918
    },
    {
      "epoch": 0.7011894400928343,
      "grad_norm": 0.43008169531822205,
      "learning_rate": 1.4942600190641968e-06,
      "loss": 0.7034,
      "step": 16919
    },
    {
      "epoch": 0.701230883998508,
      "grad_norm": 0.43792837858200073,
      "learning_rate": 1.4940527995358284e-06,
      "loss": 0.7284,
      "step": 16920
    },
    {
      "epoch": 0.7012723279041817,
      "grad_norm": 0.3991003930568695,
      "learning_rate": 1.49384558000746e-06,
      "loss": 0.6267,
      "step": 16921
    },
    {
      "epoch": 0.7013137718098553,
      "grad_norm": 0.40029212832450867,
      "learning_rate": 1.4936383604790916e-06,
      "loss": 0.6514,
      "step": 16922
    },
    {
      "epoch": 0.7013552157155291,
      "grad_norm": 0.4225305914878845,
      "learning_rate": 1.4934311409507232e-06,
      "loss": 0.6763,
      "step": 16923
    },
    {
      "epoch": 0.7013966596212027,
      "grad_norm": 0.4079318940639496,
      "learning_rate": 1.4932239214223548e-06,
      "loss": 0.6503,
      "step": 16924
    },
    {
      "epoch": 0.7014381035268764,
      "grad_norm": 0.46222391724586487,
      "learning_rate": 1.4930167018939868e-06,
      "loss": 0.6248,
      "step": 16925
    },
    {
      "epoch": 0.70147954743255,
      "grad_norm": 0.41309854388237,
      "learning_rate": 1.4928094823656184e-06,
      "loss": 0.72,
      "step": 16926
    },
    {
      "epoch": 0.7015209913382238,
      "grad_norm": 0.4318609833717346,
      "learning_rate": 1.49260226283725e-06,
      "loss": 0.6915,
      "step": 16927
    },
    {
      "epoch": 0.7015624352438974,
      "grad_norm": 0.3991871178150177,
      "learning_rate": 1.4923950433088816e-06,
      "loss": 0.6627,
      "step": 16928
    },
    {
      "epoch": 0.701603879149571,
      "grad_norm": 0.4203322231769562,
      "learning_rate": 1.4921878237805132e-06,
      "loss": 0.7135,
      "step": 16929
    },
    {
      "epoch": 0.7016453230552447,
      "grad_norm": 0.44132688641548157,
      "learning_rate": 1.4919806042521448e-06,
      "loss": 0.6299,
      "step": 16930
    },
    {
      "epoch": 0.7016867669609184,
      "grad_norm": 0.43255603313446045,
      "learning_rate": 1.4917733847237764e-06,
      "loss": 0.6721,
      "step": 16931
    },
    {
      "epoch": 0.7017282108665921,
      "grad_norm": 0.4208125174045563,
      "learning_rate": 1.491566165195408e-06,
      "loss": 0.6793,
      "step": 16932
    },
    {
      "epoch": 0.7017696547722657,
      "grad_norm": 0.4287651479244232,
      "learning_rate": 1.4913589456670396e-06,
      "loss": 0.6636,
      "step": 16933
    },
    {
      "epoch": 0.7018110986779394,
      "grad_norm": 0.4042983055114746,
      "learning_rate": 1.4911517261386716e-06,
      "loss": 0.6798,
      "step": 16934
    },
    {
      "epoch": 0.7018525425836131,
      "grad_norm": 0.44416797161102295,
      "learning_rate": 1.4909445066103032e-06,
      "loss": 0.7109,
      "step": 16935
    },
    {
      "epoch": 0.7018939864892868,
      "grad_norm": 0.4055388569831848,
      "learning_rate": 1.4907372870819348e-06,
      "loss": 0.7058,
      "step": 16936
    },
    {
      "epoch": 0.7019354303949604,
      "grad_norm": 0.48726895451545715,
      "learning_rate": 1.4905300675535664e-06,
      "loss": 0.7354,
      "step": 16937
    },
    {
      "epoch": 0.701976874300634,
      "grad_norm": 0.40984928607940674,
      "learning_rate": 1.490322848025198e-06,
      "loss": 0.665,
      "step": 16938
    },
    {
      "epoch": 0.7020183182063078,
      "grad_norm": 0.4156568646430969,
      "learning_rate": 1.4901156284968296e-06,
      "loss": 0.6515,
      "step": 16939
    },
    {
      "epoch": 0.7020597621119814,
      "grad_norm": 0.4230828881263733,
      "learning_rate": 1.4899084089684612e-06,
      "loss": 0.6665,
      "step": 16940
    },
    {
      "epoch": 0.7021012060176551,
      "grad_norm": 0.41777509450912476,
      "learning_rate": 1.4897011894400928e-06,
      "loss": 0.6549,
      "step": 16941
    },
    {
      "epoch": 0.7021426499233288,
      "grad_norm": 0.41468846797943115,
      "learning_rate": 1.4894939699117248e-06,
      "loss": 0.653,
      "step": 16942
    },
    {
      "epoch": 0.7021840938290025,
      "grad_norm": 0.41629305481910706,
      "learning_rate": 1.4892867503833564e-06,
      "loss": 0.6626,
      "step": 16943
    },
    {
      "epoch": 0.7022255377346761,
      "grad_norm": 0.39117059111595154,
      "learning_rate": 1.489079530854988e-06,
      "loss": 0.6298,
      "step": 16944
    },
    {
      "epoch": 0.7022669816403498,
      "grad_norm": 0.41990095376968384,
      "learning_rate": 1.4888723113266196e-06,
      "loss": 0.6995,
      "step": 16945
    },
    {
      "epoch": 0.7023084255460235,
      "grad_norm": 0.43583956360816956,
      "learning_rate": 1.4886650917982512e-06,
      "loss": 0.688,
      "step": 16946
    },
    {
      "epoch": 0.7023498694516971,
      "grad_norm": 0.4226759970188141,
      "learning_rate": 1.4884578722698828e-06,
      "loss": 0.6426,
      "step": 16947
    },
    {
      "epoch": 0.7023913133573708,
      "grad_norm": 0.38450413942337036,
      "learning_rate": 1.4882506527415144e-06,
      "loss": 0.6559,
      "step": 16948
    },
    {
      "epoch": 0.7024327572630444,
      "grad_norm": 0.42348501086235046,
      "learning_rate": 1.488043433213146e-06,
      "loss": 0.6617,
      "step": 16949
    },
    {
      "epoch": 0.7024742011687182,
      "grad_norm": 0.4035886526107788,
      "learning_rate": 1.4878362136847776e-06,
      "loss": 0.6581,
      "step": 16950
    },
    {
      "epoch": 0.7025156450743918,
      "grad_norm": 0.44935572147369385,
      "learning_rate": 1.4876289941564096e-06,
      "loss": 0.6742,
      "step": 16951
    },
    {
      "epoch": 0.7025570889800655,
      "grad_norm": 0.38853031396865845,
      "learning_rate": 1.4874217746280412e-06,
      "loss": 0.6335,
      "step": 16952
    },
    {
      "epoch": 0.7025985328857391,
      "grad_norm": 0.41631633043289185,
      "learning_rate": 1.4872145550996728e-06,
      "loss": 0.6658,
      "step": 16953
    },
    {
      "epoch": 0.7026399767914128,
      "grad_norm": 0.4389462172985077,
      "learning_rate": 1.4870073355713044e-06,
      "loss": 0.7032,
      "step": 16954
    },
    {
      "epoch": 0.7026814206970865,
      "grad_norm": 0.42487213015556335,
      "learning_rate": 1.486800116042936e-06,
      "loss": 0.6472,
      "step": 16955
    },
    {
      "epoch": 0.7027228646027601,
      "grad_norm": 0.41566380858421326,
      "learning_rate": 1.4865928965145676e-06,
      "loss": 0.6411,
      "step": 16956
    },
    {
      "epoch": 0.7027643085084339,
      "grad_norm": 0.3981218934059143,
      "learning_rate": 1.4863856769861992e-06,
      "loss": 0.6149,
      "step": 16957
    },
    {
      "epoch": 0.7028057524141075,
      "grad_norm": 0.44129303097724915,
      "learning_rate": 1.4861784574578308e-06,
      "loss": 0.6501,
      "step": 16958
    },
    {
      "epoch": 0.7028471963197812,
      "grad_norm": 0.4113207459449768,
      "learning_rate": 1.4859712379294624e-06,
      "loss": 0.6938,
      "step": 16959
    },
    {
      "epoch": 0.7028886402254548,
      "grad_norm": 0.41196146607398987,
      "learning_rate": 1.4857640184010944e-06,
      "loss": 0.6528,
      "step": 16960
    },
    {
      "epoch": 0.7029300841311286,
      "grad_norm": 0.4329908490180969,
      "learning_rate": 1.485556798872726e-06,
      "loss": 0.6578,
      "step": 16961
    },
    {
      "epoch": 0.7029715280368022,
      "grad_norm": 0.45901238918304443,
      "learning_rate": 1.4853495793443576e-06,
      "loss": 0.6781,
      "step": 16962
    },
    {
      "epoch": 0.7030129719424758,
      "grad_norm": 0.41950342059135437,
      "learning_rate": 1.4851423598159892e-06,
      "loss": 0.6868,
      "step": 16963
    },
    {
      "epoch": 0.7030544158481495,
      "grad_norm": 0.39756476879119873,
      "learning_rate": 1.4849351402876208e-06,
      "loss": 0.6584,
      "step": 16964
    },
    {
      "epoch": 0.7030958597538232,
      "grad_norm": 0.3989443778991699,
      "learning_rate": 1.4847279207592524e-06,
      "loss": 0.6432,
      "step": 16965
    },
    {
      "epoch": 0.7031373036594969,
      "grad_norm": 0.41724640130996704,
      "learning_rate": 1.484520701230884e-06,
      "loss": 0.6348,
      "step": 16966
    },
    {
      "epoch": 0.7031787475651705,
      "grad_norm": 0.39071863889694214,
      "learning_rate": 1.4843134817025156e-06,
      "loss": 0.6661,
      "step": 16967
    },
    {
      "epoch": 0.7032201914708442,
      "grad_norm": 0.393300324678421,
      "learning_rate": 1.4841062621741476e-06,
      "loss": 0.6433,
      "step": 16968
    },
    {
      "epoch": 0.7032616353765179,
      "grad_norm": 0.3562166690826416,
      "learning_rate": 1.4838990426457792e-06,
      "loss": 0.611,
      "step": 16969
    },
    {
      "epoch": 0.7033030792821916,
      "grad_norm": 0.3994918167591095,
      "learning_rate": 1.4836918231174108e-06,
      "loss": 0.6146,
      "step": 16970
    },
    {
      "epoch": 0.7033445231878652,
      "grad_norm": 0.4227902889251709,
      "learning_rate": 1.4834846035890424e-06,
      "loss": 0.6611,
      "step": 16971
    },
    {
      "epoch": 0.7033859670935388,
      "grad_norm": 0.38050928711891174,
      "learning_rate": 1.483277384060674e-06,
      "loss": 0.6443,
      "step": 16972
    },
    {
      "epoch": 0.7034274109992126,
      "grad_norm": 0.410273939371109,
      "learning_rate": 1.4830701645323056e-06,
      "loss": 0.7322,
      "step": 16973
    },
    {
      "epoch": 0.7034688549048862,
      "grad_norm": 0.45212045311927795,
      "learning_rate": 1.4828629450039372e-06,
      "loss": 0.6539,
      "step": 16974
    },
    {
      "epoch": 0.7035102988105599,
      "grad_norm": 0.4202477037906647,
      "learning_rate": 1.4826557254755688e-06,
      "loss": 0.6753,
      "step": 16975
    },
    {
      "epoch": 0.7035517427162336,
      "grad_norm": 0.4380984306335449,
      "learning_rate": 1.4824485059472004e-06,
      "loss": 0.71,
      "step": 16976
    },
    {
      "epoch": 0.7035931866219073,
      "grad_norm": 0.42050158977508545,
      "learning_rate": 1.4822412864188324e-06,
      "loss": 0.6537,
      "step": 16977
    },
    {
      "epoch": 0.7036346305275809,
      "grad_norm": 0.388355553150177,
      "learning_rate": 1.482034066890464e-06,
      "loss": 0.6428,
      "step": 16978
    },
    {
      "epoch": 0.7036760744332546,
      "grad_norm": 0.40893515944480896,
      "learning_rate": 1.4818268473620956e-06,
      "loss": 0.6838,
      "step": 16979
    },
    {
      "epoch": 0.7037175183389283,
      "grad_norm": 0.3791602849960327,
      "learning_rate": 1.4816196278337272e-06,
      "loss": 0.6667,
      "step": 16980
    },
    {
      "epoch": 0.7037589622446019,
      "grad_norm": 0.3972429633140564,
      "learning_rate": 1.4814124083053588e-06,
      "loss": 0.6608,
      "step": 16981
    },
    {
      "epoch": 0.7038004061502756,
      "grad_norm": 0.39118531346321106,
      "learning_rate": 1.4812051887769904e-06,
      "loss": 0.6744,
      "step": 16982
    },
    {
      "epoch": 0.7038418500559492,
      "grad_norm": 0.3992242217063904,
      "learning_rate": 1.480997969248622e-06,
      "loss": 0.6438,
      "step": 16983
    },
    {
      "epoch": 0.703883293961623,
      "grad_norm": 0.3970489203929901,
      "learning_rate": 1.4807907497202536e-06,
      "loss": 0.6727,
      "step": 16984
    },
    {
      "epoch": 0.7039247378672966,
      "grad_norm": 0.4176301658153534,
      "learning_rate": 1.4805835301918852e-06,
      "loss": 0.6736,
      "step": 16985
    },
    {
      "epoch": 0.7039661817729703,
      "grad_norm": 0.42490169405937195,
      "learning_rate": 1.4803763106635172e-06,
      "loss": 0.6642,
      "step": 16986
    },
    {
      "epoch": 0.704007625678644,
      "grad_norm": 0.4063190519809723,
      "learning_rate": 1.4801690911351488e-06,
      "loss": 0.7094,
      "step": 16987
    },
    {
      "epoch": 0.7040490695843177,
      "grad_norm": 0.42902156710624695,
      "learning_rate": 1.4799618716067804e-06,
      "loss": 0.6313,
      "step": 16988
    },
    {
      "epoch": 0.7040905134899913,
      "grad_norm": 0.4297036826610565,
      "learning_rate": 1.479754652078412e-06,
      "loss": 0.6306,
      "step": 16989
    },
    {
      "epoch": 0.7041319573956649,
      "grad_norm": 0.44268572330474854,
      "learning_rate": 1.4795474325500436e-06,
      "loss": 0.7174,
      "step": 16990
    },
    {
      "epoch": 0.7041734013013387,
      "grad_norm": 0.3757169544696808,
      "learning_rate": 1.4793402130216752e-06,
      "loss": 0.6384,
      "step": 16991
    },
    {
      "epoch": 0.7042148452070123,
      "grad_norm": 0.3899397552013397,
      "learning_rate": 1.4791329934933068e-06,
      "loss": 0.5986,
      "step": 16992
    },
    {
      "epoch": 0.704256289112686,
      "grad_norm": 0.46488726139068604,
      "learning_rate": 1.4789257739649384e-06,
      "loss": 0.6893,
      "step": 16993
    },
    {
      "epoch": 0.7042977330183596,
      "grad_norm": 0.4335681200027466,
      "learning_rate": 1.47871855443657e-06,
      "loss": 0.6301,
      "step": 16994
    },
    {
      "epoch": 0.7043391769240334,
      "grad_norm": 0.4316413104534149,
      "learning_rate": 1.478511334908202e-06,
      "loss": 0.6376,
      "step": 16995
    },
    {
      "epoch": 0.704380620829707,
      "grad_norm": 0.39548900723457336,
      "learning_rate": 1.4783041153798336e-06,
      "loss": 0.6609,
      "step": 16996
    },
    {
      "epoch": 0.7044220647353807,
      "grad_norm": 0.41640374064445496,
      "learning_rate": 1.4780968958514652e-06,
      "loss": 0.7007,
      "step": 16997
    },
    {
      "epoch": 0.7044635086410543,
      "grad_norm": 0.40604373812675476,
      "learning_rate": 1.4778896763230968e-06,
      "loss": 0.6991,
      "step": 16998
    },
    {
      "epoch": 0.704504952546728,
      "grad_norm": 0.4296743869781494,
      "learning_rate": 1.4776824567947284e-06,
      "loss": 0.6903,
      "step": 16999
    },
    {
      "epoch": 0.7045463964524017,
      "grad_norm": 0.45505931973457336,
      "learning_rate": 1.47747523726636e-06,
      "loss": 0.7207,
      "step": 17000
    },
    {
      "epoch": 0.7045878403580753,
      "grad_norm": 0.4159430265426636,
      "learning_rate": 1.4772680177379916e-06,
      "loss": 0.6559,
      "step": 17001
    },
    {
      "epoch": 0.704629284263749,
      "grad_norm": 0.41849228739738464,
      "learning_rate": 1.4770607982096232e-06,
      "loss": 0.6816,
      "step": 17002
    },
    {
      "epoch": 0.7046707281694227,
      "grad_norm": 0.3905252516269684,
      "learning_rate": 1.4768535786812552e-06,
      "loss": 0.6168,
      "step": 17003
    },
    {
      "epoch": 0.7047121720750964,
      "grad_norm": 0.4428134858608246,
      "learning_rate": 1.4766463591528868e-06,
      "loss": 0.689,
      "step": 17004
    },
    {
      "epoch": 0.70475361598077,
      "grad_norm": 0.4219529330730438,
      "learning_rate": 1.4764391396245184e-06,
      "loss": 0.6653,
      "step": 17005
    },
    {
      "epoch": 0.7047950598864438,
      "grad_norm": 0.41161489486694336,
      "learning_rate": 1.47623192009615e-06,
      "loss": 0.7155,
      "step": 17006
    },
    {
      "epoch": 0.7048365037921174,
      "grad_norm": 0.42839911580085754,
      "learning_rate": 1.4760247005677816e-06,
      "loss": 0.6329,
      "step": 17007
    },
    {
      "epoch": 0.704877947697791,
      "grad_norm": 0.4130896031856537,
      "learning_rate": 1.4758174810394132e-06,
      "loss": 0.7053,
      "step": 17008
    },
    {
      "epoch": 0.7049193916034647,
      "grad_norm": 0.4535704553127289,
      "learning_rate": 1.4756102615110448e-06,
      "loss": 0.7151,
      "step": 17009
    },
    {
      "epoch": 0.7049608355091384,
      "grad_norm": 0.41720089316368103,
      "learning_rate": 1.4754030419826764e-06,
      "loss": 0.6949,
      "step": 17010
    },
    {
      "epoch": 0.7050022794148121,
      "grad_norm": 0.4244033396244049,
      "learning_rate": 1.475195822454308e-06,
      "loss": 0.6821,
      "step": 17011
    },
    {
      "epoch": 0.7050437233204857,
      "grad_norm": 0.42856988310813904,
      "learning_rate": 1.47498860292594e-06,
      "loss": 0.6565,
      "step": 17012
    },
    {
      "epoch": 0.7050851672261594,
      "grad_norm": 0.4642801582813263,
      "learning_rate": 1.4747813833975716e-06,
      "loss": 0.7529,
      "step": 17013
    },
    {
      "epoch": 0.7051266111318331,
      "grad_norm": 0.4119976758956909,
      "learning_rate": 1.4745741638692032e-06,
      "loss": 0.6482,
      "step": 17014
    },
    {
      "epoch": 0.7051680550375067,
      "grad_norm": 0.38056135177612305,
      "learning_rate": 1.4743669443408348e-06,
      "loss": 0.6421,
      "step": 17015
    },
    {
      "epoch": 0.7052094989431804,
      "grad_norm": 0.40900319814682007,
      "learning_rate": 1.4741597248124664e-06,
      "loss": 0.6687,
      "step": 17016
    },
    {
      "epoch": 0.705250942848854,
      "grad_norm": 0.4239286780357361,
      "learning_rate": 1.473952505284098e-06,
      "loss": 0.6488,
      "step": 17017
    },
    {
      "epoch": 0.7052923867545278,
      "grad_norm": 0.4327636957168579,
      "learning_rate": 1.4737452857557296e-06,
      "loss": 0.6494,
      "step": 17018
    },
    {
      "epoch": 0.7053338306602014,
      "grad_norm": 0.5016306638717651,
      "learning_rate": 1.4735380662273612e-06,
      "loss": 0.6963,
      "step": 17019
    },
    {
      "epoch": 0.7053752745658751,
      "grad_norm": 0.43588921427726746,
      "learning_rate": 1.473330846698993e-06,
      "loss": 0.6738,
      "step": 17020
    },
    {
      "epoch": 0.7054167184715487,
      "grad_norm": 0.4024278223514557,
      "learning_rate": 1.4731236271706248e-06,
      "loss": 0.6494,
      "step": 17021
    },
    {
      "epoch": 0.7054581623772225,
      "grad_norm": 0.3988310694694519,
      "learning_rate": 1.4729164076422564e-06,
      "loss": 0.6169,
      "step": 17022
    },
    {
      "epoch": 0.7054996062828961,
      "grad_norm": 0.4274252951145172,
      "learning_rate": 1.472709188113888e-06,
      "loss": 0.6699,
      "step": 17023
    },
    {
      "epoch": 0.7055410501885697,
      "grad_norm": 0.43692880868911743,
      "learning_rate": 1.4725019685855196e-06,
      "loss": 0.6687,
      "step": 17024
    },
    {
      "epoch": 0.7055824940942435,
      "grad_norm": 0.4472159445285797,
      "learning_rate": 1.4722947490571512e-06,
      "loss": 0.7314,
      "step": 17025
    },
    {
      "epoch": 0.7056239379999171,
      "grad_norm": 0.47058236598968506,
      "learning_rate": 1.4720875295287828e-06,
      "loss": 0.6857,
      "step": 17026
    },
    {
      "epoch": 0.7056653819055908,
      "grad_norm": 0.4456748962402344,
      "learning_rate": 1.4718803100004144e-06,
      "loss": 0.6776,
      "step": 17027
    },
    {
      "epoch": 0.7057068258112644,
      "grad_norm": 0.396790087223053,
      "learning_rate": 1.471673090472046e-06,
      "loss": 0.6501,
      "step": 17028
    },
    {
      "epoch": 0.7057482697169382,
      "grad_norm": 0.4563027620315552,
      "learning_rate": 1.471465870943678e-06,
      "loss": 0.6908,
      "step": 17029
    },
    {
      "epoch": 0.7057897136226118,
      "grad_norm": 0.4618462324142456,
      "learning_rate": 1.4712586514153096e-06,
      "loss": 0.7585,
      "step": 17030
    },
    {
      "epoch": 0.7058311575282855,
      "grad_norm": 0.4485417306423187,
      "learning_rate": 1.4710514318869412e-06,
      "loss": 0.6924,
      "step": 17031
    },
    {
      "epoch": 0.7058726014339591,
      "grad_norm": 0.4422416090965271,
      "learning_rate": 1.4708442123585728e-06,
      "loss": 0.7087,
      "step": 17032
    },
    {
      "epoch": 0.7059140453396328,
      "grad_norm": 0.4310205578804016,
      "learning_rate": 1.4706369928302044e-06,
      "loss": 0.6956,
      "step": 17033
    },
    {
      "epoch": 0.7059554892453065,
      "grad_norm": 0.3919341564178467,
      "learning_rate": 1.470429773301836e-06,
      "loss": 0.6569,
      "step": 17034
    },
    {
      "epoch": 0.7059969331509801,
      "grad_norm": 0.4226980209350586,
      "learning_rate": 1.4702225537734676e-06,
      "loss": 0.6877,
      "step": 17035
    },
    {
      "epoch": 0.7060383770566538,
      "grad_norm": 0.40782156586647034,
      "learning_rate": 1.4700153342450992e-06,
      "loss": 0.6448,
      "step": 17036
    },
    {
      "epoch": 0.7060798209623275,
      "grad_norm": 0.40488287806510925,
      "learning_rate": 1.469808114716731e-06,
      "loss": 0.6787,
      "step": 17037
    },
    {
      "epoch": 0.7061212648680012,
      "grad_norm": 0.3945687711238861,
      "learning_rate": 1.4696008951883628e-06,
      "loss": 0.6329,
      "step": 17038
    },
    {
      "epoch": 0.7061627087736748,
      "grad_norm": 0.3957037627696991,
      "learning_rate": 1.4693936756599944e-06,
      "loss": 0.693,
      "step": 17039
    },
    {
      "epoch": 0.7062041526793486,
      "grad_norm": 0.37580937147140503,
      "learning_rate": 1.469186456131626e-06,
      "loss": 0.6639,
      "step": 17040
    },
    {
      "epoch": 0.7062455965850222,
      "grad_norm": 0.43958523869514465,
      "learning_rate": 1.4689792366032576e-06,
      "loss": 0.689,
      "step": 17041
    },
    {
      "epoch": 0.7062870404906958,
      "grad_norm": 0.4280121624469757,
      "learning_rate": 1.4687720170748892e-06,
      "loss": 0.6523,
      "step": 17042
    },
    {
      "epoch": 0.7063284843963695,
      "grad_norm": 0.39378538727760315,
      "learning_rate": 1.4685647975465208e-06,
      "loss": 0.6317,
      "step": 17043
    },
    {
      "epoch": 0.7063699283020431,
      "grad_norm": 0.40225014090538025,
      "learning_rate": 1.4683575780181524e-06,
      "loss": 0.6304,
      "step": 17044
    },
    {
      "epoch": 0.7064113722077169,
      "grad_norm": 0.47212979197502136,
      "learning_rate": 1.4681503584897842e-06,
      "loss": 0.7336,
      "step": 17045
    },
    {
      "epoch": 0.7064528161133905,
      "grad_norm": 0.39396536350250244,
      "learning_rate": 1.4679431389614158e-06,
      "loss": 0.6509,
      "step": 17046
    },
    {
      "epoch": 0.7064942600190642,
      "grad_norm": 0.39323511719703674,
      "learning_rate": 1.4677359194330476e-06,
      "loss": 0.6667,
      "step": 17047
    },
    {
      "epoch": 0.7065357039247379,
      "grad_norm": 0.460580438375473,
      "learning_rate": 1.4675286999046792e-06,
      "loss": 0.715,
      "step": 17048
    },
    {
      "epoch": 0.7065771478304116,
      "grad_norm": 0.4254401624202728,
      "learning_rate": 1.4673214803763108e-06,
      "loss": 0.7581,
      "step": 17049
    },
    {
      "epoch": 0.7066185917360852,
      "grad_norm": 0.3968885540962219,
      "learning_rate": 1.4671142608479424e-06,
      "loss": 0.6185,
      "step": 17050
    },
    {
      "epoch": 0.7066600356417588,
      "grad_norm": 0.3994959890842438,
      "learning_rate": 1.466907041319574e-06,
      "loss": 0.6378,
      "step": 17051
    },
    {
      "epoch": 0.7067014795474326,
      "grad_norm": 0.4218335747718811,
      "learning_rate": 1.4666998217912056e-06,
      "loss": 0.7278,
      "step": 17052
    },
    {
      "epoch": 0.7067429234531062,
      "grad_norm": 0.4707273244857788,
      "learning_rate": 1.4664926022628372e-06,
      "loss": 0.6509,
      "step": 17053
    },
    {
      "epoch": 0.7067843673587799,
      "grad_norm": 0.3982565999031067,
      "learning_rate": 1.466285382734469e-06,
      "loss": 0.6462,
      "step": 17054
    },
    {
      "epoch": 0.7068258112644535,
      "grad_norm": 0.3947368562221527,
      "learning_rate": 1.4660781632061006e-06,
      "loss": 0.7189,
      "step": 17055
    },
    {
      "epoch": 0.7068672551701273,
      "grad_norm": 0.4142375886440277,
      "learning_rate": 1.4658709436777324e-06,
      "loss": 0.6594,
      "step": 17056
    },
    {
      "epoch": 0.7069086990758009,
      "grad_norm": 0.40377792716026306,
      "learning_rate": 1.465663724149364e-06,
      "loss": 0.6804,
      "step": 17057
    },
    {
      "epoch": 0.7069501429814746,
      "grad_norm": 0.40233200788497925,
      "learning_rate": 1.4654565046209956e-06,
      "loss": 0.6265,
      "step": 17058
    },
    {
      "epoch": 0.7069915868871482,
      "grad_norm": 0.3936782479286194,
      "learning_rate": 1.4652492850926272e-06,
      "loss": 0.6267,
      "step": 17059
    },
    {
      "epoch": 0.7070330307928219,
      "grad_norm": 0.4427107572555542,
      "learning_rate": 1.4650420655642588e-06,
      "loss": 0.698,
      "step": 17060
    },
    {
      "epoch": 0.7070744746984956,
      "grad_norm": 0.4007430672645569,
      "learning_rate": 1.4648348460358904e-06,
      "loss": 0.6765,
      "step": 17061
    },
    {
      "epoch": 0.7071159186041692,
      "grad_norm": 0.39708080887794495,
      "learning_rate": 1.4646276265075222e-06,
      "loss": 0.6517,
      "step": 17062
    },
    {
      "epoch": 0.707157362509843,
      "grad_norm": 0.4321609437465668,
      "learning_rate": 1.4644204069791538e-06,
      "loss": 0.7058,
      "step": 17063
    },
    {
      "epoch": 0.7071988064155166,
      "grad_norm": 0.4112613797187805,
      "learning_rate": 1.4642131874507856e-06,
      "loss": 0.6758,
      "step": 17064
    },
    {
      "epoch": 0.7072402503211903,
      "grad_norm": 0.4398932158946991,
      "learning_rate": 1.4640059679224172e-06,
      "loss": 0.687,
      "step": 17065
    },
    {
      "epoch": 0.7072816942268639,
      "grad_norm": 0.4064832031726837,
      "learning_rate": 1.4637987483940488e-06,
      "loss": 0.6769,
      "step": 17066
    },
    {
      "epoch": 0.7073231381325377,
      "grad_norm": 0.4149221181869507,
      "learning_rate": 1.4635915288656804e-06,
      "loss": 0.613,
      "step": 17067
    },
    {
      "epoch": 0.7073645820382113,
      "grad_norm": 0.4063790440559387,
      "learning_rate": 1.463384309337312e-06,
      "loss": 0.6802,
      "step": 17068
    },
    {
      "epoch": 0.7074060259438849,
      "grad_norm": 0.42275065183639526,
      "learning_rate": 1.4631770898089436e-06,
      "loss": 0.6938,
      "step": 17069
    },
    {
      "epoch": 0.7074474698495586,
      "grad_norm": 0.4306286573410034,
      "learning_rate": 1.4629698702805752e-06,
      "loss": 0.6743,
      "step": 17070
    },
    {
      "epoch": 0.7074889137552323,
      "grad_norm": 0.41605764627456665,
      "learning_rate": 1.462762650752207e-06,
      "loss": 0.6355,
      "step": 17071
    },
    {
      "epoch": 0.707530357660906,
      "grad_norm": 0.4109753668308258,
      "learning_rate": 1.4625554312238386e-06,
      "loss": 0.6904,
      "step": 17072
    },
    {
      "epoch": 0.7075718015665796,
      "grad_norm": 0.3909955322742462,
      "learning_rate": 1.4623482116954704e-06,
      "loss": 0.6299,
      "step": 17073
    },
    {
      "epoch": 0.7076132454722533,
      "grad_norm": 0.3955157399177551,
      "learning_rate": 1.462140992167102e-06,
      "loss": 0.6339,
      "step": 17074
    },
    {
      "epoch": 0.707654689377927,
      "grad_norm": 0.4407317340373993,
      "learning_rate": 1.4619337726387336e-06,
      "loss": 0.6602,
      "step": 17075
    },
    {
      "epoch": 0.7076961332836006,
      "grad_norm": 0.44630587100982666,
      "learning_rate": 1.4617265531103652e-06,
      "loss": 0.6783,
      "step": 17076
    },
    {
      "epoch": 0.7077375771892743,
      "grad_norm": 0.43595075607299805,
      "learning_rate": 1.4615193335819968e-06,
      "loss": 0.7275,
      "step": 17077
    },
    {
      "epoch": 0.707779021094948,
      "grad_norm": 0.3888038098812103,
      "learning_rate": 1.4613121140536284e-06,
      "loss": 0.6553,
      "step": 17078
    },
    {
      "epoch": 0.7078204650006217,
      "grad_norm": 0.4243892431259155,
      "learning_rate": 1.4611048945252602e-06,
      "loss": 0.6301,
      "step": 17079
    },
    {
      "epoch": 0.7078619089062953,
      "grad_norm": 0.44253790378570557,
      "learning_rate": 1.4608976749968918e-06,
      "loss": 0.6824,
      "step": 17080
    },
    {
      "epoch": 0.707903352811969,
      "grad_norm": 0.37912118434906006,
      "learning_rate": 1.4606904554685234e-06,
      "loss": 0.6451,
      "step": 17081
    },
    {
      "epoch": 0.7079447967176427,
      "grad_norm": 0.4390753507614136,
      "learning_rate": 1.4604832359401552e-06,
      "loss": 0.6846,
      "step": 17082
    },
    {
      "epoch": 0.7079862406233164,
      "grad_norm": 0.3990453779697418,
      "learning_rate": 1.4602760164117868e-06,
      "loss": 0.6683,
      "step": 17083
    },
    {
      "epoch": 0.70802768452899,
      "grad_norm": 0.3845057785511017,
      "learning_rate": 1.4600687968834184e-06,
      "loss": 0.6385,
      "step": 17084
    },
    {
      "epoch": 0.7080691284346636,
      "grad_norm": 0.41920962929725647,
      "learning_rate": 1.45986157735505e-06,
      "loss": 0.729,
      "step": 17085
    },
    {
      "epoch": 0.7081105723403374,
      "grad_norm": 0.3904169201850891,
      "learning_rate": 1.4596543578266816e-06,
      "loss": 0.658,
      "step": 17086
    },
    {
      "epoch": 0.708152016246011,
      "grad_norm": 0.38504117727279663,
      "learning_rate": 1.4594471382983132e-06,
      "loss": 0.6554,
      "step": 17087
    },
    {
      "epoch": 0.7081934601516847,
      "grad_norm": 0.3803388774394989,
      "learning_rate": 1.459239918769945e-06,
      "loss": 0.6277,
      "step": 17088
    },
    {
      "epoch": 0.7082349040573583,
      "grad_norm": 0.47713714838027954,
      "learning_rate": 1.4590326992415766e-06,
      "loss": 0.7305,
      "step": 17089
    },
    {
      "epoch": 0.7082763479630321,
      "grad_norm": 0.4066689908504486,
      "learning_rate": 1.4588254797132084e-06,
      "loss": 0.6588,
      "step": 17090
    },
    {
      "epoch": 0.7083177918687057,
      "grad_norm": 0.4041980803012848,
      "learning_rate": 1.45861826018484e-06,
      "loss": 0.6775,
      "step": 17091
    },
    {
      "epoch": 0.7083592357743794,
      "grad_norm": 0.4055139720439911,
      "learning_rate": 1.4584110406564716e-06,
      "loss": 0.6799,
      "step": 17092
    },
    {
      "epoch": 0.708400679680053,
      "grad_norm": 0.3667786121368408,
      "learning_rate": 1.4582038211281032e-06,
      "loss": 0.6416,
      "step": 17093
    },
    {
      "epoch": 0.7084421235857267,
      "grad_norm": 0.41678696870803833,
      "learning_rate": 1.4579966015997348e-06,
      "loss": 0.6559,
      "step": 17094
    },
    {
      "epoch": 0.7084835674914004,
      "grad_norm": 0.3906075656414032,
      "learning_rate": 1.4577893820713664e-06,
      "loss": 0.6453,
      "step": 17095
    },
    {
      "epoch": 0.708525011397074,
      "grad_norm": 0.41323980689048767,
      "learning_rate": 1.4575821625429982e-06,
      "loss": 0.6798,
      "step": 17096
    },
    {
      "epoch": 0.7085664553027478,
      "grad_norm": 0.3746788203716278,
      "learning_rate": 1.4573749430146298e-06,
      "loss": 0.6724,
      "step": 17097
    },
    {
      "epoch": 0.7086078992084214,
      "grad_norm": 0.45917192101478577,
      "learning_rate": 1.4571677234862614e-06,
      "loss": 0.6748,
      "step": 17098
    },
    {
      "epoch": 0.7086493431140951,
      "grad_norm": 0.39618241786956787,
      "learning_rate": 1.4569605039578932e-06,
      "loss": 0.6692,
      "step": 17099
    },
    {
      "epoch": 0.7086907870197687,
      "grad_norm": 0.4390541613101959,
      "learning_rate": 1.4567532844295248e-06,
      "loss": 0.641,
      "step": 17100
    },
    {
      "epoch": 0.7087322309254425,
      "grad_norm": 0.43311506509780884,
      "learning_rate": 1.4565460649011564e-06,
      "loss": 0.6797,
      "step": 17101
    },
    {
      "epoch": 0.7087736748311161,
      "grad_norm": 0.38658228516578674,
      "learning_rate": 1.456338845372788e-06,
      "loss": 0.6289,
      "step": 17102
    },
    {
      "epoch": 0.7088151187367897,
      "grad_norm": 0.43291813135147095,
      "learning_rate": 1.4561316258444196e-06,
      "loss": 0.6902,
      "step": 17103
    },
    {
      "epoch": 0.7088565626424634,
      "grad_norm": 0.41580280661582947,
      "learning_rate": 1.4559244063160512e-06,
      "loss": 0.6732,
      "step": 17104
    },
    {
      "epoch": 0.7088980065481371,
      "grad_norm": 0.4244527220726013,
      "learning_rate": 1.455717186787683e-06,
      "loss": 0.6561,
      "step": 17105
    },
    {
      "epoch": 0.7089394504538108,
      "grad_norm": 0.41677457094192505,
      "learning_rate": 1.4555099672593146e-06,
      "loss": 0.6775,
      "step": 17106
    },
    {
      "epoch": 0.7089808943594844,
      "grad_norm": 0.43259790539741516,
      "learning_rate": 1.4553027477309462e-06,
      "loss": 0.6995,
      "step": 17107
    },
    {
      "epoch": 0.7090223382651581,
      "grad_norm": 0.38651514053344727,
      "learning_rate": 1.455095528202578e-06,
      "loss": 0.6445,
      "step": 17108
    },
    {
      "epoch": 0.7090637821708318,
      "grad_norm": 0.429196834564209,
      "learning_rate": 1.4548883086742096e-06,
      "loss": 0.7271,
      "step": 17109
    },
    {
      "epoch": 0.7091052260765055,
      "grad_norm": 0.4176829159259796,
      "learning_rate": 1.4546810891458412e-06,
      "loss": 0.6772,
      "step": 17110
    },
    {
      "epoch": 0.7091466699821791,
      "grad_norm": 0.460344523191452,
      "learning_rate": 1.4544738696174728e-06,
      "loss": 0.7085,
      "step": 17111
    },
    {
      "epoch": 0.7091881138878527,
      "grad_norm": 0.3839628994464874,
      "learning_rate": 1.4542666500891044e-06,
      "loss": 0.7034,
      "step": 17112
    },
    {
      "epoch": 0.7092295577935265,
      "grad_norm": 0.4124516248703003,
      "learning_rate": 1.4540594305607362e-06,
      "loss": 0.6892,
      "step": 17113
    },
    {
      "epoch": 0.7092710016992001,
      "grad_norm": 0.4253595471382141,
      "learning_rate": 1.4538522110323678e-06,
      "loss": 0.6819,
      "step": 17114
    },
    {
      "epoch": 0.7093124456048738,
      "grad_norm": 0.4169895648956299,
      "learning_rate": 1.4536449915039994e-06,
      "loss": 0.7084,
      "step": 17115
    },
    {
      "epoch": 0.7093538895105475,
      "grad_norm": 0.4271676242351532,
      "learning_rate": 1.4534377719756312e-06,
      "loss": 0.7556,
      "step": 17116
    },
    {
      "epoch": 0.7093953334162212,
      "grad_norm": 0.4364808201789856,
      "learning_rate": 1.4532305524472628e-06,
      "loss": 0.6449,
      "step": 17117
    },
    {
      "epoch": 0.7094367773218948,
      "grad_norm": 0.44272512197494507,
      "learning_rate": 1.4530233329188944e-06,
      "loss": 0.6505,
      "step": 17118
    },
    {
      "epoch": 0.7094782212275685,
      "grad_norm": 0.42349377274513245,
      "learning_rate": 1.452816113390526e-06,
      "loss": 0.6873,
      "step": 17119
    },
    {
      "epoch": 0.7095196651332422,
      "grad_norm": 0.41405540704727173,
      "learning_rate": 1.4526088938621576e-06,
      "loss": 0.6627,
      "step": 17120
    },
    {
      "epoch": 0.7095611090389158,
      "grad_norm": 0.39842119812965393,
      "learning_rate": 1.4524016743337892e-06,
      "loss": 0.6957,
      "step": 17121
    },
    {
      "epoch": 0.7096025529445895,
      "grad_norm": 0.4041319191455841,
      "learning_rate": 1.452194454805421e-06,
      "loss": 0.6924,
      "step": 17122
    },
    {
      "epoch": 0.7096439968502631,
      "grad_norm": 0.44509357213974,
      "learning_rate": 1.4519872352770526e-06,
      "loss": 0.6395,
      "step": 17123
    },
    {
      "epoch": 0.7096854407559369,
      "grad_norm": 0.47146058082580566,
      "learning_rate": 1.4517800157486842e-06,
      "loss": 0.7155,
      "step": 17124
    },
    {
      "epoch": 0.7097268846616105,
      "grad_norm": 0.40336716175079346,
      "learning_rate": 1.451572796220316e-06,
      "loss": 0.679,
      "step": 17125
    },
    {
      "epoch": 0.7097683285672842,
      "grad_norm": 0.3946537971496582,
      "learning_rate": 1.4513655766919476e-06,
      "loss": 0.637,
      "step": 17126
    },
    {
      "epoch": 0.7098097724729578,
      "grad_norm": 0.41641882061958313,
      "learning_rate": 1.4511583571635792e-06,
      "loss": 0.7004,
      "step": 17127
    },
    {
      "epoch": 0.7098512163786316,
      "grad_norm": 0.38467055559158325,
      "learning_rate": 1.4509511376352108e-06,
      "loss": 0.7043,
      "step": 17128
    },
    {
      "epoch": 0.7098926602843052,
      "grad_norm": 0.4395214021205902,
      "learning_rate": 1.4507439181068424e-06,
      "loss": 0.6794,
      "step": 17129
    },
    {
      "epoch": 0.7099341041899788,
      "grad_norm": 0.4153783917427063,
      "learning_rate": 1.4505366985784742e-06,
      "loss": 0.6769,
      "step": 17130
    },
    {
      "epoch": 0.7099755480956526,
      "grad_norm": 0.44801390171051025,
      "learning_rate": 1.4503294790501058e-06,
      "loss": 0.6903,
      "step": 17131
    },
    {
      "epoch": 0.7100169920013262,
      "grad_norm": 0.44084593653678894,
      "learning_rate": 1.4501222595217374e-06,
      "loss": 0.6797,
      "step": 17132
    },
    {
      "epoch": 0.7100584359069999,
      "grad_norm": 0.43808794021606445,
      "learning_rate": 1.449915039993369e-06,
      "loss": 0.7211,
      "step": 17133
    },
    {
      "epoch": 0.7100998798126735,
      "grad_norm": 0.4207003712654114,
      "learning_rate": 1.4497078204650008e-06,
      "loss": 0.6899,
      "step": 17134
    },
    {
      "epoch": 0.7101413237183473,
      "grad_norm": 0.38609179854393005,
      "learning_rate": 1.4495006009366324e-06,
      "loss": 0.6251,
      "step": 17135
    },
    {
      "epoch": 0.7101827676240209,
      "grad_norm": 0.4192427694797516,
      "learning_rate": 1.449293381408264e-06,
      "loss": 0.6793,
      "step": 17136
    },
    {
      "epoch": 0.7102242115296945,
      "grad_norm": 0.42130979895591736,
      "learning_rate": 1.4490861618798956e-06,
      "loss": 0.6743,
      "step": 17137
    },
    {
      "epoch": 0.7102656554353682,
      "grad_norm": 0.4154570996761322,
      "learning_rate": 1.4488789423515274e-06,
      "loss": 0.6456,
      "step": 17138
    },
    {
      "epoch": 0.7103070993410419,
      "grad_norm": 0.40507709980010986,
      "learning_rate": 1.448671722823159e-06,
      "loss": 0.7285,
      "step": 17139
    },
    {
      "epoch": 0.7103485432467156,
      "grad_norm": 0.42757031321525574,
      "learning_rate": 1.4484645032947906e-06,
      "loss": 0.6388,
      "step": 17140
    },
    {
      "epoch": 0.7103899871523892,
      "grad_norm": 0.3910718560218811,
      "learning_rate": 1.4482572837664222e-06,
      "loss": 0.661,
      "step": 17141
    },
    {
      "epoch": 0.710431431058063,
      "grad_norm": 0.4253712594509125,
      "learning_rate": 1.4480500642380538e-06,
      "loss": 0.6499,
      "step": 17142
    },
    {
      "epoch": 0.7104728749637366,
      "grad_norm": 0.3651489019393921,
      "learning_rate": 1.4478428447096856e-06,
      "loss": 0.6265,
      "step": 17143
    },
    {
      "epoch": 0.7105143188694103,
      "grad_norm": 0.407944917678833,
      "learning_rate": 1.4476356251813172e-06,
      "loss": 0.7019,
      "step": 17144
    },
    {
      "epoch": 0.7105557627750839,
      "grad_norm": 0.4259357154369354,
      "learning_rate": 1.4474284056529488e-06,
      "loss": 0.7537,
      "step": 17145
    },
    {
      "epoch": 0.7105972066807575,
      "grad_norm": 0.40328076481819153,
      "learning_rate": 1.4472211861245804e-06,
      "loss": 0.6704,
      "step": 17146
    },
    {
      "epoch": 0.7106386505864313,
      "grad_norm": 0.43832820653915405,
      "learning_rate": 1.4470139665962122e-06,
      "loss": 0.6389,
      "step": 17147
    },
    {
      "epoch": 0.7106800944921049,
      "grad_norm": 0.41793403029441833,
      "learning_rate": 1.4468067470678438e-06,
      "loss": 0.7264,
      "step": 17148
    },
    {
      "epoch": 0.7107215383977786,
      "grad_norm": 0.42761752009391785,
      "learning_rate": 1.4465995275394754e-06,
      "loss": 0.7018,
      "step": 17149
    },
    {
      "epoch": 0.7107629823034523,
      "grad_norm": 0.41330400109291077,
      "learning_rate": 1.446392308011107e-06,
      "loss": 0.6656,
      "step": 17150
    },
    {
      "epoch": 0.710804426209126,
      "grad_norm": 0.4136297106742859,
      "learning_rate": 1.4461850884827388e-06,
      "loss": 0.6294,
      "step": 17151
    },
    {
      "epoch": 0.7108458701147996,
      "grad_norm": 0.4349151849746704,
      "learning_rate": 1.4459778689543704e-06,
      "loss": 0.6934,
      "step": 17152
    },
    {
      "epoch": 0.7108873140204733,
      "grad_norm": 0.4104061722755432,
      "learning_rate": 1.445770649426002e-06,
      "loss": 0.6287,
      "step": 17153
    },
    {
      "epoch": 0.710928757926147,
      "grad_norm": 0.4348065257072449,
      "learning_rate": 1.4455634298976336e-06,
      "loss": 0.7236,
      "step": 17154
    },
    {
      "epoch": 0.7109702018318206,
      "grad_norm": 0.41015976667404175,
      "learning_rate": 1.4453562103692654e-06,
      "loss": 0.6581,
      "step": 17155
    },
    {
      "epoch": 0.7110116457374943,
      "grad_norm": 0.3860315978527069,
      "learning_rate": 1.445148990840897e-06,
      "loss": 0.6606,
      "step": 17156
    },
    {
      "epoch": 0.7110530896431679,
      "grad_norm": 0.42880141735076904,
      "learning_rate": 1.4449417713125286e-06,
      "loss": 0.6698,
      "step": 17157
    },
    {
      "epoch": 0.7110945335488417,
      "grad_norm": 0.39286312460899353,
      "learning_rate": 1.4447345517841602e-06,
      "loss": 0.6528,
      "step": 17158
    },
    {
      "epoch": 0.7111359774545153,
      "grad_norm": 0.3640916347503662,
      "learning_rate": 1.4445273322557918e-06,
      "loss": 0.5814,
      "step": 17159
    },
    {
      "epoch": 0.711177421360189,
      "grad_norm": 0.4011084735393524,
      "learning_rate": 1.4443201127274236e-06,
      "loss": 0.6968,
      "step": 17160
    },
    {
      "epoch": 0.7112188652658626,
      "grad_norm": 0.38492611050605774,
      "learning_rate": 1.4441128931990552e-06,
      "loss": 0.6752,
      "step": 17161
    },
    {
      "epoch": 0.7112603091715364,
      "grad_norm": 0.423849493265152,
      "learning_rate": 1.4439056736706868e-06,
      "loss": 0.7087,
      "step": 17162
    },
    {
      "epoch": 0.71130175307721,
      "grad_norm": 0.39665883779525757,
      "learning_rate": 1.4436984541423184e-06,
      "loss": 0.6516,
      "step": 17163
    },
    {
      "epoch": 0.7113431969828836,
      "grad_norm": 0.40594109892845154,
      "learning_rate": 1.4434912346139502e-06,
      "loss": 0.6578,
      "step": 17164
    },
    {
      "epoch": 0.7113846408885574,
      "grad_norm": 0.3938899636268616,
      "learning_rate": 1.4432840150855818e-06,
      "loss": 0.6234,
      "step": 17165
    },
    {
      "epoch": 0.711426084794231,
      "grad_norm": 0.4268961548805237,
      "learning_rate": 1.4430767955572134e-06,
      "loss": 0.6715,
      "step": 17166
    },
    {
      "epoch": 0.7114675286999047,
      "grad_norm": 0.3874188959598541,
      "learning_rate": 1.442869576028845e-06,
      "loss": 0.6031,
      "step": 17167
    },
    {
      "epoch": 0.7115089726055783,
      "grad_norm": 0.4064997434616089,
      "learning_rate": 1.4426623565004766e-06,
      "loss": 0.6934,
      "step": 17168
    },
    {
      "epoch": 0.7115504165112521,
      "grad_norm": 0.47106385231018066,
      "learning_rate": 1.4424551369721084e-06,
      "loss": 0.6675,
      "step": 17169
    },
    {
      "epoch": 0.7115918604169257,
      "grad_norm": 0.40281856060028076,
      "learning_rate": 1.44224791744374e-06,
      "loss": 0.6554,
      "step": 17170
    },
    {
      "epoch": 0.7116333043225994,
      "grad_norm": 0.48936718702316284,
      "learning_rate": 1.4420406979153716e-06,
      "loss": 0.7,
      "step": 17171
    },
    {
      "epoch": 0.711674748228273,
      "grad_norm": 0.41593289375305176,
      "learning_rate": 1.4418334783870034e-06,
      "loss": 0.6951,
      "step": 17172
    },
    {
      "epoch": 0.7117161921339467,
      "grad_norm": 0.39680179953575134,
      "learning_rate": 1.441626258858635e-06,
      "loss": 0.6985,
      "step": 17173
    },
    {
      "epoch": 0.7117576360396204,
      "grad_norm": 0.4302290976047516,
      "learning_rate": 1.4414190393302666e-06,
      "loss": 0.696,
      "step": 17174
    },
    {
      "epoch": 0.711799079945294,
      "grad_norm": 0.45055392384529114,
      "learning_rate": 1.4412118198018982e-06,
      "loss": 0.7241,
      "step": 17175
    },
    {
      "epoch": 0.7118405238509677,
      "grad_norm": 0.41072559356689453,
      "learning_rate": 1.4410046002735298e-06,
      "loss": 0.6782,
      "step": 17176
    },
    {
      "epoch": 0.7118819677566414,
      "grad_norm": 0.41165706515312195,
      "learning_rate": 1.4407973807451616e-06,
      "loss": 0.6366,
      "step": 17177
    },
    {
      "epoch": 0.7119234116623151,
      "grad_norm": 0.4233558773994446,
      "learning_rate": 1.4405901612167932e-06,
      "loss": 0.7349,
      "step": 17178
    },
    {
      "epoch": 0.7119648555679887,
      "grad_norm": 0.4239099621772766,
      "learning_rate": 1.4403829416884248e-06,
      "loss": 0.6941,
      "step": 17179
    },
    {
      "epoch": 0.7120062994736625,
      "grad_norm": 0.4017597436904907,
      "learning_rate": 1.4401757221600564e-06,
      "loss": 0.6864,
      "step": 17180
    },
    {
      "epoch": 0.7120477433793361,
      "grad_norm": 0.41544127464294434,
      "learning_rate": 1.4399685026316882e-06,
      "loss": 0.6912,
      "step": 17181
    },
    {
      "epoch": 0.7120891872850097,
      "grad_norm": 0.41935113072395325,
      "learning_rate": 1.4397612831033198e-06,
      "loss": 0.7124,
      "step": 17182
    },
    {
      "epoch": 0.7121306311906834,
      "grad_norm": 0.4116427004337311,
      "learning_rate": 1.4395540635749514e-06,
      "loss": 0.6257,
      "step": 17183
    },
    {
      "epoch": 0.712172075096357,
      "grad_norm": 0.3983500003814697,
      "learning_rate": 1.439346844046583e-06,
      "loss": 0.6316,
      "step": 17184
    },
    {
      "epoch": 0.7122135190020308,
      "grad_norm": 0.4283016622066498,
      "learning_rate": 1.4391396245182146e-06,
      "loss": 0.6697,
      "step": 17185
    },
    {
      "epoch": 0.7122549629077044,
      "grad_norm": 0.4285280108451843,
      "learning_rate": 1.4389324049898464e-06,
      "loss": 0.6304,
      "step": 17186
    },
    {
      "epoch": 0.7122964068133781,
      "grad_norm": 0.3857199549674988,
      "learning_rate": 1.438725185461478e-06,
      "loss": 0.6162,
      "step": 17187
    },
    {
      "epoch": 0.7123378507190518,
      "grad_norm": 0.42128753662109375,
      "learning_rate": 1.4385179659331096e-06,
      "loss": 0.7351,
      "step": 17188
    },
    {
      "epoch": 0.7123792946247254,
      "grad_norm": 0.4033653140068054,
      "learning_rate": 1.4383107464047414e-06,
      "loss": 0.6455,
      "step": 17189
    },
    {
      "epoch": 0.7124207385303991,
      "grad_norm": 0.39610305428504944,
      "learning_rate": 1.438103526876373e-06,
      "loss": 0.6907,
      "step": 17190
    },
    {
      "epoch": 0.7124621824360727,
      "grad_norm": 0.3711470365524292,
      "learning_rate": 1.4378963073480046e-06,
      "loss": 0.6562,
      "step": 17191
    },
    {
      "epoch": 0.7125036263417465,
      "grad_norm": 0.4262102544307709,
      "learning_rate": 1.4376890878196362e-06,
      "loss": 0.6953,
      "step": 17192
    },
    {
      "epoch": 0.7125450702474201,
      "grad_norm": 0.4469776153564453,
      "learning_rate": 1.4374818682912678e-06,
      "loss": 0.6724,
      "step": 17193
    },
    {
      "epoch": 0.7125865141530938,
      "grad_norm": 0.3950137495994568,
      "learning_rate": 1.4372746487628994e-06,
      "loss": 0.6243,
      "step": 17194
    },
    {
      "epoch": 0.7126279580587674,
      "grad_norm": 0.3959225118160248,
      "learning_rate": 1.4370674292345312e-06,
      "loss": 0.6711,
      "step": 17195
    },
    {
      "epoch": 0.7126694019644412,
      "grad_norm": 0.42926108837127686,
      "learning_rate": 1.4368602097061628e-06,
      "loss": 0.7019,
      "step": 17196
    },
    {
      "epoch": 0.7127108458701148,
      "grad_norm": 0.4032299816608429,
      "learning_rate": 1.4366529901777944e-06,
      "loss": 0.658,
      "step": 17197
    },
    {
      "epoch": 0.7127522897757884,
      "grad_norm": 0.403743177652359,
      "learning_rate": 1.4364457706494262e-06,
      "loss": 0.6777,
      "step": 17198
    },
    {
      "epoch": 0.7127937336814621,
      "grad_norm": 0.39242538809776306,
      "learning_rate": 1.4362385511210578e-06,
      "loss": 0.6731,
      "step": 17199
    },
    {
      "epoch": 0.7128351775871358,
      "grad_norm": 0.4026254713535309,
      "learning_rate": 1.4360313315926894e-06,
      "loss": 0.6554,
      "step": 17200
    },
    {
      "epoch": 0.7128766214928095,
      "grad_norm": 0.39283907413482666,
      "learning_rate": 1.435824112064321e-06,
      "loss": 0.6555,
      "step": 17201
    },
    {
      "epoch": 0.7129180653984831,
      "grad_norm": 0.4268520176410675,
      "learning_rate": 1.4356168925359526e-06,
      "loss": 0.673,
      "step": 17202
    },
    {
      "epoch": 0.7129595093041569,
      "grad_norm": 0.3949146568775177,
      "learning_rate": 1.4354096730075842e-06,
      "loss": 0.6725,
      "step": 17203
    },
    {
      "epoch": 0.7130009532098305,
      "grad_norm": 0.4011481702327728,
      "learning_rate": 1.435202453479216e-06,
      "loss": 0.6685,
      "step": 17204
    },
    {
      "epoch": 0.7130423971155042,
      "grad_norm": 0.4004656672477722,
      "learning_rate": 1.4349952339508476e-06,
      "loss": 0.658,
      "step": 17205
    },
    {
      "epoch": 0.7130838410211778,
      "grad_norm": 0.38832539319992065,
      "learning_rate": 1.4347880144224794e-06,
      "loss": 0.6816,
      "step": 17206
    },
    {
      "epoch": 0.7131252849268515,
      "grad_norm": 0.4332389235496521,
      "learning_rate": 1.434580794894111e-06,
      "loss": 0.6782,
      "step": 17207
    },
    {
      "epoch": 0.7131667288325252,
      "grad_norm": 0.4023863673210144,
      "learning_rate": 1.4343735753657426e-06,
      "loss": 0.6721,
      "step": 17208
    },
    {
      "epoch": 0.7132081727381988,
      "grad_norm": 0.40926694869995117,
      "learning_rate": 1.4341663558373742e-06,
      "loss": 0.6533,
      "step": 17209
    },
    {
      "epoch": 0.7132496166438725,
      "grad_norm": 0.45090505480766296,
      "learning_rate": 1.4339591363090058e-06,
      "loss": 0.6849,
      "step": 17210
    },
    {
      "epoch": 0.7132910605495462,
      "grad_norm": 0.41792985796928406,
      "learning_rate": 1.4337519167806374e-06,
      "loss": 0.5996,
      "step": 17211
    },
    {
      "epoch": 0.7133325044552199,
      "grad_norm": 0.4013620615005493,
      "learning_rate": 1.4335446972522692e-06,
      "loss": 0.6077,
      "step": 17212
    },
    {
      "epoch": 0.7133739483608935,
      "grad_norm": 0.4184472858905792,
      "learning_rate": 1.4333374777239008e-06,
      "loss": 0.7068,
      "step": 17213
    },
    {
      "epoch": 0.7134153922665672,
      "grad_norm": 0.3798205256462097,
      "learning_rate": 1.4331302581955324e-06,
      "loss": 0.6216,
      "step": 17214
    },
    {
      "epoch": 0.7134568361722409,
      "grad_norm": 0.41796883940696716,
      "learning_rate": 1.4329230386671642e-06,
      "loss": 0.7,
      "step": 17215
    },
    {
      "epoch": 0.7134982800779145,
      "grad_norm": 0.4258931875228882,
      "learning_rate": 1.4327158191387958e-06,
      "loss": 0.6682,
      "step": 17216
    },
    {
      "epoch": 0.7135397239835882,
      "grad_norm": 0.42291250824928284,
      "learning_rate": 1.4325085996104274e-06,
      "loss": 0.6877,
      "step": 17217
    },
    {
      "epoch": 0.7135811678892618,
      "grad_norm": 0.4162435233592987,
      "learning_rate": 1.432301380082059e-06,
      "loss": 0.7123,
      "step": 17218
    },
    {
      "epoch": 0.7136226117949356,
      "grad_norm": 0.4066869616508484,
      "learning_rate": 1.4320941605536906e-06,
      "loss": 0.6738,
      "step": 17219
    },
    {
      "epoch": 0.7136640557006092,
      "grad_norm": 0.39517447352409363,
      "learning_rate": 1.4318869410253222e-06,
      "loss": 0.6409,
      "step": 17220
    },
    {
      "epoch": 0.7137054996062829,
      "grad_norm": 0.42086559534072876,
      "learning_rate": 1.431679721496954e-06,
      "loss": 0.6699,
      "step": 17221
    },
    {
      "epoch": 0.7137469435119566,
      "grad_norm": 0.3757900297641754,
      "learning_rate": 1.4314725019685856e-06,
      "loss": 0.6602,
      "step": 17222
    },
    {
      "epoch": 0.7137883874176303,
      "grad_norm": 0.3955202102661133,
      "learning_rate": 1.4312652824402174e-06,
      "loss": 0.6301,
      "step": 17223
    },
    {
      "epoch": 0.7138298313233039,
      "grad_norm": 0.4265369772911072,
      "learning_rate": 1.431058062911849e-06,
      "loss": 0.6851,
      "step": 17224
    },
    {
      "epoch": 0.7138712752289775,
      "grad_norm": 0.4501838684082031,
      "learning_rate": 1.4308508433834806e-06,
      "loss": 0.6528,
      "step": 17225
    },
    {
      "epoch": 0.7139127191346513,
      "grad_norm": 0.38698020577430725,
      "learning_rate": 1.4306436238551122e-06,
      "loss": 0.6587,
      "step": 17226
    },
    {
      "epoch": 0.7139541630403249,
      "grad_norm": 0.4086521863937378,
      "learning_rate": 1.4304364043267438e-06,
      "loss": 0.6498,
      "step": 17227
    },
    {
      "epoch": 0.7139956069459986,
      "grad_norm": 0.4133686423301697,
      "learning_rate": 1.4302291847983754e-06,
      "loss": 0.6514,
      "step": 17228
    },
    {
      "epoch": 0.7140370508516722,
      "grad_norm": 0.4076843857765198,
      "learning_rate": 1.430021965270007e-06,
      "loss": 0.6722,
      "step": 17229
    },
    {
      "epoch": 0.714078494757346,
      "grad_norm": 0.38485851883888245,
      "learning_rate": 1.4298147457416388e-06,
      "loss": 0.6227,
      "step": 17230
    },
    {
      "epoch": 0.7141199386630196,
      "grad_norm": 0.388555645942688,
      "learning_rate": 1.4296075262132706e-06,
      "loss": 0.7004,
      "step": 17231
    },
    {
      "epoch": 0.7141613825686933,
      "grad_norm": 0.41111770272254944,
      "learning_rate": 1.4294003066849022e-06,
      "loss": 0.6534,
      "step": 17232
    },
    {
      "epoch": 0.714202826474367,
      "grad_norm": 0.41902655363082886,
      "learning_rate": 1.4291930871565338e-06,
      "loss": 0.6831,
      "step": 17233
    },
    {
      "epoch": 0.7142442703800406,
      "grad_norm": 0.4162984788417816,
      "learning_rate": 1.4289858676281654e-06,
      "loss": 0.6687,
      "step": 17234
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.4067620635032654,
      "learning_rate": 1.428778648099797e-06,
      "loss": 0.6279,
      "step": 17235
    },
    {
      "epoch": 0.7143271581913879,
      "grad_norm": 0.38521045446395874,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.6353,
      "step": 17236
    },
    {
      "epoch": 0.7143686020970617,
      "grad_norm": 0.45835113525390625,
      "learning_rate": 1.4283642090430602e-06,
      "loss": 0.7493,
      "step": 17237
    },
    {
      "epoch": 0.7144100460027353,
      "grad_norm": 0.3873526155948639,
      "learning_rate": 1.428156989514692e-06,
      "loss": 0.6814,
      "step": 17238
    },
    {
      "epoch": 0.714451489908409,
      "grad_norm": 0.3855592608451843,
      "learning_rate": 1.4279497699863236e-06,
      "loss": 0.6904,
      "step": 17239
    },
    {
      "epoch": 0.7144929338140826,
      "grad_norm": 0.4106155037879944,
      "learning_rate": 1.4277425504579554e-06,
      "loss": 0.6538,
      "step": 17240
    },
    {
      "epoch": 0.7145343777197564,
      "grad_norm": 0.4222412705421448,
      "learning_rate": 1.427535330929587e-06,
      "loss": 0.7169,
      "step": 17241
    },
    {
      "epoch": 0.71457582162543,
      "grad_norm": 0.4217895567417145,
      "learning_rate": 1.4273281114012186e-06,
      "loss": 0.7034,
      "step": 17242
    },
    {
      "epoch": 0.7146172655311036,
      "grad_norm": 0.39750197529792786,
      "learning_rate": 1.4271208918728502e-06,
      "loss": 0.6324,
      "step": 17243
    },
    {
      "epoch": 0.7146587094367773,
      "grad_norm": 0.4085679054260254,
      "learning_rate": 1.4269136723444818e-06,
      "loss": 0.6622,
      "step": 17244
    },
    {
      "epoch": 0.714700153342451,
      "grad_norm": 0.45610931515693665,
      "learning_rate": 1.4267064528161134e-06,
      "loss": 0.679,
      "step": 17245
    },
    {
      "epoch": 0.7147415972481247,
      "grad_norm": 0.4353027641773224,
      "learning_rate": 1.426499233287745e-06,
      "loss": 0.6956,
      "step": 17246
    },
    {
      "epoch": 0.7147830411537983,
      "grad_norm": 0.4007241427898407,
      "learning_rate": 1.4262920137593768e-06,
      "loss": 0.6664,
      "step": 17247
    },
    {
      "epoch": 0.714824485059472,
      "grad_norm": 0.432422012090683,
      "learning_rate": 1.4260847942310086e-06,
      "loss": 0.6848,
      "step": 17248
    },
    {
      "epoch": 0.7148659289651457,
      "grad_norm": 0.4303676187992096,
      "learning_rate": 1.4258775747026402e-06,
      "loss": 0.7233,
      "step": 17249
    },
    {
      "epoch": 0.7149073728708193,
      "grad_norm": 0.38789495825767517,
      "learning_rate": 1.4256703551742718e-06,
      "loss": 0.658,
      "step": 17250
    },
    {
      "epoch": 0.714948816776493,
      "grad_norm": 0.4143153131008148,
      "learning_rate": 1.4254631356459034e-06,
      "loss": 0.6671,
      "step": 17251
    },
    {
      "epoch": 0.7149902606821666,
      "grad_norm": 0.4264090359210968,
      "learning_rate": 1.425255916117535e-06,
      "loss": 0.6056,
      "step": 17252
    },
    {
      "epoch": 0.7150317045878404,
      "grad_norm": 0.40375521779060364,
      "learning_rate": 1.4250486965891666e-06,
      "loss": 0.6565,
      "step": 17253
    },
    {
      "epoch": 0.715073148493514,
      "grad_norm": 0.4146019518375397,
      "learning_rate": 1.4248414770607982e-06,
      "loss": 0.6548,
      "step": 17254
    },
    {
      "epoch": 0.7151145923991877,
      "grad_norm": 0.3905279338359833,
      "learning_rate": 1.4246342575324298e-06,
      "loss": 0.613,
      "step": 17255
    },
    {
      "epoch": 0.7151560363048614,
      "grad_norm": 0.39962294697761536,
      "learning_rate": 1.4244270380040616e-06,
      "loss": 0.6721,
      "step": 17256
    },
    {
      "epoch": 0.7151974802105351,
      "grad_norm": 0.37589672207832336,
      "learning_rate": 1.4242198184756934e-06,
      "loss": 0.6001,
      "step": 17257
    },
    {
      "epoch": 0.7152389241162087,
      "grad_norm": 0.390352338552475,
      "learning_rate": 1.424012598947325e-06,
      "loss": 0.6802,
      "step": 17258
    },
    {
      "epoch": 0.7152803680218823,
      "grad_norm": 0.4050734043121338,
      "learning_rate": 1.4238053794189566e-06,
      "loss": 0.678,
      "step": 17259
    },
    {
      "epoch": 0.7153218119275561,
      "grad_norm": 0.41170430183410645,
      "learning_rate": 1.4235981598905882e-06,
      "loss": 0.6812,
      "step": 17260
    },
    {
      "epoch": 0.7153632558332297,
      "grad_norm": 0.4183063209056854,
      "learning_rate": 1.4233909403622198e-06,
      "loss": 0.7076,
      "step": 17261
    },
    {
      "epoch": 0.7154046997389034,
      "grad_norm": 0.4043883681297302,
      "learning_rate": 1.4231837208338514e-06,
      "loss": 0.6229,
      "step": 17262
    },
    {
      "epoch": 0.715446143644577,
      "grad_norm": 0.41671022772789,
      "learning_rate": 1.422976501305483e-06,
      "loss": 0.679,
      "step": 17263
    },
    {
      "epoch": 0.7154875875502508,
      "grad_norm": 0.457964152097702,
      "learning_rate": 1.4227692817771148e-06,
      "loss": 0.6597,
      "step": 17264
    },
    {
      "epoch": 0.7155290314559244,
      "grad_norm": 0.3986662030220032,
      "learning_rate": 1.4225620622487466e-06,
      "loss": 0.6687,
      "step": 17265
    },
    {
      "epoch": 0.7155704753615981,
      "grad_norm": 0.4151467978954315,
      "learning_rate": 1.4223548427203782e-06,
      "loss": 0.6589,
      "step": 17266
    },
    {
      "epoch": 0.7156119192672717,
      "grad_norm": 0.4237269163131714,
      "learning_rate": 1.4221476231920098e-06,
      "loss": 0.7004,
      "step": 17267
    },
    {
      "epoch": 0.7156533631729454,
      "grad_norm": 0.38333553075790405,
      "learning_rate": 1.4219404036636414e-06,
      "loss": 0.6654,
      "step": 17268
    },
    {
      "epoch": 0.7156948070786191,
      "grad_norm": 0.41093388199806213,
      "learning_rate": 1.421733184135273e-06,
      "loss": 0.7029,
      "step": 17269
    },
    {
      "epoch": 0.7157362509842927,
      "grad_norm": 0.40810373425483704,
      "learning_rate": 1.4215259646069046e-06,
      "loss": 0.6208,
      "step": 17270
    },
    {
      "epoch": 0.7157776948899665,
      "grad_norm": 0.42177239060401917,
      "learning_rate": 1.4213187450785362e-06,
      "loss": 0.6782,
      "step": 17271
    },
    {
      "epoch": 0.7158191387956401,
      "grad_norm": 0.4325515627861023,
      "learning_rate": 1.4211115255501678e-06,
      "loss": 0.7255,
      "step": 17272
    },
    {
      "epoch": 0.7158605827013138,
      "grad_norm": 0.40350043773651123,
      "learning_rate": 1.4209043060217996e-06,
      "loss": 0.6281,
      "step": 17273
    },
    {
      "epoch": 0.7159020266069874,
      "grad_norm": 0.4112646281719208,
      "learning_rate": 1.4206970864934314e-06,
      "loss": 0.6873,
      "step": 17274
    },
    {
      "epoch": 0.7159434705126612,
      "grad_norm": 0.4365899860858917,
      "learning_rate": 1.420489866965063e-06,
      "loss": 0.7063,
      "step": 17275
    },
    {
      "epoch": 0.7159849144183348,
      "grad_norm": 0.4231681227684021,
      "learning_rate": 1.4202826474366946e-06,
      "loss": 0.739,
      "step": 17276
    },
    {
      "epoch": 0.7160263583240084,
      "grad_norm": 0.41030699014663696,
      "learning_rate": 1.4200754279083262e-06,
      "loss": 0.6469,
      "step": 17277
    },
    {
      "epoch": 0.7160678022296821,
      "grad_norm": 0.40123194456100464,
      "learning_rate": 1.4198682083799578e-06,
      "loss": 0.6342,
      "step": 17278
    },
    {
      "epoch": 0.7161092461353558,
      "grad_norm": 0.42994168400764465,
      "learning_rate": 1.4196609888515894e-06,
      "loss": 0.6526,
      "step": 17279
    },
    {
      "epoch": 0.7161506900410295,
      "grad_norm": 0.39958655834198,
      "learning_rate": 1.419453769323221e-06,
      "loss": 0.6462,
      "step": 17280
    },
    {
      "epoch": 0.7161921339467031,
      "grad_norm": 0.41059520840644836,
      "learning_rate": 1.4192465497948526e-06,
      "loss": 0.6014,
      "step": 17281
    },
    {
      "epoch": 0.7162335778523768,
      "grad_norm": 0.38499975204467773,
      "learning_rate": 1.4190393302664846e-06,
      "loss": 0.644,
      "step": 17282
    },
    {
      "epoch": 0.7162750217580505,
      "grad_norm": 0.3916405737400055,
      "learning_rate": 1.4188321107381162e-06,
      "loss": 0.5753,
      "step": 17283
    },
    {
      "epoch": 0.7163164656637242,
      "grad_norm": 0.3972073197364807,
      "learning_rate": 1.4186248912097478e-06,
      "loss": 0.6235,
      "step": 17284
    },
    {
      "epoch": 0.7163579095693978,
      "grad_norm": 0.430368036031723,
      "learning_rate": 1.4184176716813794e-06,
      "loss": 0.7181,
      "step": 17285
    },
    {
      "epoch": 0.7163993534750714,
      "grad_norm": 0.40028277039527893,
      "learning_rate": 1.418210452153011e-06,
      "loss": 0.6892,
      "step": 17286
    },
    {
      "epoch": 0.7164407973807452,
      "grad_norm": 0.4451952874660492,
      "learning_rate": 1.4180032326246426e-06,
      "loss": 0.6951,
      "step": 17287
    },
    {
      "epoch": 0.7164822412864188,
      "grad_norm": 0.38671648502349854,
      "learning_rate": 1.4177960130962742e-06,
      "loss": 0.718,
      "step": 17288
    },
    {
      "epoch": 0.7165236851920925,
      "grad_norm": 0.4029669165611267,
      "learning_rate": 1.4175887935679058e-06,
      "loss": 0.6615,
      "step": 17289
    },
    {
      "epoch": 0.7165651290977662,
      "grad_norm": 0.42925021052360535,
      "learning_rate": 1.4173815740395374e-06,
      "loss": 0.668,
      "step": 17290
    },
    {
      "epoch": 0.7166065730034399,
      "grad_norm": 0.403926819562912,
      "learning_rate": 1.4171743545111695e-06,
      "loss": 0.6506,
      "step": 17291
    },
    {
      "epoch": 0.7166480169091135,
      "grad_norm": 0.3993407189846039,
      "learning_rate": 1.416967134982801e-06,
      "loss": 0.6747,
      "step": 17292
    },
    {
      "epoch": 0.7166894608147872,
      "grad_norm": 0.4468505084514618,
      "learning_rate": 1.4167599154544326e-06,
      "loss": 0.6531,
      "step": 17293
    },
    {
      "epoch": 0.7167309047204609,
      "grad_norm": 0.4041764736175537,
      "learning_rate": 1.4165526959260642e-06,
      "loss": 0.6957,
      "step": 17294
    },
    {
      "epoch": 0.7167723486261345,
      "grad_norm": 0.4140792191028595,
      "learning_rate": 1.4163454763976958e-06,
      "loss": 0.7102,
      "step": 17295
    },
    {
      "epoch": 0.7168137925318082,
      "grad_norm": 0.4193328022956848,
      "learning_rate": 1.4161382568693274e-06,
      "loss": 0.7029,
      "step": 17296
    },
    {
      "epoch": 0.7168552364374818,
      "grad_norm": 0.41624847054481506,
      "learning_rate": 1.415931037340959e-06,
      "loss": 0.7363,
      "step": 17297
    },
    {
      "epoch": 0.7168966803431556,
      "grad_norm": 0.41988906264305115,
      "learning_rate": 1.4157238178125906e-06,
      "loss": 0.7302,
      "step": 17298
    },
    {
      "epoch": 0.7169381242488292,
      "grad_norm": 0.4258204996585846,
      "learning_rate": 1.4155165982842227e-06,
      "loss": 0.675,
      "step": 17299
    },
    {
      "epoch": 0.7169795681545029,
      "grad_norm": 0.4020434617996216,
      "learning_rate": 1.4153093787558543e-06,
      "loss": 0.6556,
      "step": 17300
    },
    {
      "epoch": 0.7170210120601765,
      "grad_norm": 0.39730560779571533,
      "learning_rate": 1.4151021592274858e-06,
      "loss": 0.6381,
      "step": 17301
    },
    {
      "epoch": 0.7170624559658503,
      "grad_norm": 0.44688811898231506,
      "learning_rate": 1.4148949396991174e-06,
      "loss": 0.7285,
      "step": 17302
    },
    {
      "epoch": 0.7171038998715239,
      "grad_norm": 0.40782302618026733,
      "learning_rate": 1.414687720170749e-06,
      "loss": 0.7,
      "step": 17303
    },
    {
      "epoch": 0.7171453437771975,
      "grad_norm": 0.4457809329032898,
      "learning_rate": 1.4144805006423806e-06,
      "loss": 0.7,
      "step": 17304
    },
    {
      "epoch": 0.7171867876828713,
      "grad_norm": 0.42020443081855774,
      "learning_rate": 1.4142732811140122e-06,
      "loss": 0.7075,
      "step": 17305
    },
    {
      "epoch": 0.7172282315885449,
      "grad_norm": 0.41487154364585876,
      "learning_rate": 1.4140660615856438e-06,
      "loss": 0.6418,
      "step": 17306
    },
    {
      "epoch": 0.7172696754942186,
      "grad_norm": 0.4093276560306549,
      "learning_rate": 1.4138588420572754e-06,
      "loss": 0.665,
      "step": 17307
    },
    {
      "epoch": 0.7173111193998922,
      "grad_norm": 0.4286203384399414,
      "learning_rate": 1.4136516225289075e-06,
      "loss": 0.6326,
      "step": 17308
    },
    {
      "epoch": 0.717352563305566,
      "grad_norm": 0.39587169885635376,
      "learning_rate": 1.413444403000539e-06,
      "loss": 0.7081,
      "step": 17309
    },
    {
      "epoch": 0.7173940072112396,
      "grad_norm": 0.4468456506729126,
      "learning_rate": 1.4132371834721706e-06,
      "loss": 0.6846,
      "step": 17310
    },
    {
      "epoch": 0.7174354511169132,
      "grad_norm": 0.3953080177307129,
      "learning_rate": 1.4130299639438022e-06,
      "loss": 0.676,
      "step": 17311
    },
    {
      "epoch": 0.7174768950225869,
      "grad_norm": 0.3981952369213104,
      "learning_rate": 1.4128227444154338e-06,
      "loss": 0.6567,
      "step": 17312
    },
    {
      "epoch": 0.7175183389282606,
      "grad_norm": 0.3728790879249573,
      "learning_rate": 1.4126155248870654e-06,
      "loss": 0.6091,
      "step": 17313
    },
    {
      "epoch": 0.7175597828339343,
      "grad_norm": 0.38949382305145264,
      "learning_rate": 1.412408305358697e-06,
      "loss": 0.6062,
      "step": 17314
    },
    {
      "epoch": 0.7176012267396079,
      "grad_norm": 0.38943153619766235,
      "learning_rate": 1.4122010858303286e-06,
      "loss": 0.6392,
      "step": 17315
    },
    {
      "epoch": 0.7176426706452816,
      "grad_norm": 0.4260689914226532,
      "learning_rate": 1.4119938663019602e-06,
      "loss": 0.7114,
      "step": 17316
    },
    {
      "epoch": 0.7176841145509553,
      "grad_norm": 0.38348516821861267,
      "learning_rate": 1.4117866467735923e-06,
      "loss": 0.6091,
      "step": 17317
    },
    {
      "epoch": 0.717725558456629,
      "grad_norm": 0.41743743419647217,
      "learning_rate": 1.4115794272452239e-06,
      "loss": 0.6361,
      "step": 17318
    },
    {
      "epoch": 0.7177670023623026,
      "grad_norm": 0.3997074067592621,
      "learning_rate": 1.4113722077168554e-06,
      "loss": 0.653,
      "step": 17319
    },
    {
      "epoch": 0.7178084462679762,
      "grad_norm": 0.40662381052970886,
      "learning_rate": 1.411164988188487e-06,
      "loss": 0.6714,
      "step": 17320
    },
    {
      "epoch": 0.71784989017365,
      "grad_norm": 0.4267542362213135,
      "learning_rate": 1.4109577686601186e-06,
      "loss": 0.671,
      "step": 17321
    },
    {
      "epoch": 0.7178913340793236,
      "grad_norm": 0.4198610782623291,
      "learning_rate": 1.4107505491317502e-06,
      "loss": 0.6573,
      "step": 17322
    },
    {
      "epoch": 0.7179327779849973,
      "grad_norm": 0.40717971324920654,
      "learning_rate": 1.4105433296033818e-06,
      "loss": 0.6753,
      "step": 17323
    },
    {
      "epoch": 0.717974221890671,
      "grad_norm": 0.4537484645843506,
      "learning_rate": 1.4103361100750134e-06,
      "loss": 0.6842,
      "step": 17324
    },
    {
      "epoch": 0.7180156657963447,
      "grad_norm": 0.4328794479370117,
      "learning_rate": 1.4101288905466455e-06,
      "loss": 0.6232,
      "step": 17325
    },
    {
      "epoch": 0.7180571097020183,
      "grad_norm": 0.43901193141937256,
      "learning_rate": 1.409921671018277e-06,
      "loss": 0.7168,
      "step": 17326
    },
    {
      "epoch": 0.718098553607692,
      "grad_norm": 0.4031955897808075,
      "learning_rate": 1.4097144514899087e-06,
      "loss": 0.7026,
      "step": 17327
    },
    {
      "epoch": 0.7181399975133657,
      "grad_norm": 0.43417415022850037,
      "learning_rate": 1.4095072319615402e-06,
      "loss": 0.6554,
      "step": 17328
    },
    {
      "epoch": 0.7181814414190393,
      "grad_norm": 0.4246547222137451,
      "learning_rate": 1.4093000124331718e-06,
      "loss": 0.6907,
      "step": 17329
    },
    {
      "epoch": 0.718222885324713,
      "grad_norm": 0.4580361843109131,
      "learning_rate": 1.4090927929048034e-06,
      "loss": 0.6958,
      "step": 17330
    },
    {
      "epoch": 0.7182643292303866,
      "grad_norm": 0.44306379556655884,
      "learning_rate": 1.408885573376435e-06,
      "loss": 0.7146,
      "step": 17331
    },
    {
      "epoch": 0.7183057731360604,
      "grad_norm": 0.42164909839630127,
      "learning_rate": 1.4086783538480666e-06,
      "loss": 0.7065,
      "step": 17332
    },
    {
      "epoch": 0.718347217041734,
      "grad_norm": 0.4326670467853546,
      "learning_rate": 1.4084711343196982e-06,
      "loss": 0.6595,
      "step": 17333
    },
    {
      "epoch": 0.7183886609474077,
      "grad_norm": 0.4053807556629181,
      "learning_rate": 1.4082639147913303e-06,
      "loss": 0.6731,
      "step": 17334
    },
    {
      "epoch": 0.7184301048530813,
      "grad_norm": 0.4254748225212097,
      "learning_rate": 1.4080566952629619e-06,
      "loss": 0.6477,
      "step": 17335
    },
    {
      "epoch": 0.7184715487587551,
      "grad_norm": 0.39833804965019226,
      "learning_rate": 1.4078494757345935e-06,
      "loss": 0.6852,
      "step": 17336
    },
    {
      "epoch": 0.7185129926644287,
      "grad_norm": 0.41857248544692993,
      "learning_rate": 1.407642256206225e-06,
      "loss": 0.6484,
      "step": 17337
    },
    {
      "epoch": 0.7185544365701023,
      "grad_norm": 0.40050336718559265,
      "learning_rate": 1.4074350366778566e-06,
      "loss": 0.6471,
      "step": 17338
    },
    {
      "epoch": 0.718595880475776,
      "grad_norm": 0.39058953523635864,
      "learning_rate": 1.4072278171494882e-06,
      "loss": 0.6655,
      "step": 17339
    },
    {
      "epoch": 0.7186373243814497,
      "grad_norm": 0.39312630891799927,
      "learning_rate": 1.4070205976211198e-06,
      "loss": 0.6481,
      "step": 17340
    },
    {
      "epoch": 0.7186787682871234,
      "grad_norm": 0.3944080173969269,
      "learning_rate": 1.4068133780927514e-06,
      "loss": 0.6212,
      "step": 17341
    },
    {
      "epoch": 0.718720212192797,
      "grad_norm": 0.3970586359500885,
      "learning_rate": 1.406606158564383e-06,
      "loss": 0.6305,
      "step": 17342
    },
    {
      "epoch": 0.7187616560984708,
      "grad_norm": 0.43630170822143555,
      "learning_rate": 1.406398939036015e-06,
      "loss": 0.707,
      "step": 17343
    },
    {
      "epoch": 0.7188031000041444,
      "grad_norm": 0.40251633524894714,
      "learning_rate": 1.4061917195076467e-06,
      "loss": 0.646,
      "step": 17344
    },
    {
      "epoch": 0.7188445439098181,
      "grad_norm": 0.42355361580848694,
      "learning_rate": 1.4059844999792783e-06,
      "loss": 0.707,
      "step": 17345
    },
    {
      "epoch": 0.7188859878154917,
      "grad_norm": 0.3875163495540619,
      "learning_rate": 1.4057772804509098e-06,
      "loss": 0.6731,
      "step": 17346
    },
    {
      "epoch": 0.7189274317211654,
      "grad_norm": 0.42768093943595886,
      "learning_rate": 1.4055700609225414e-06,
      "loss": 0.6456,
      "step": 17347
    },
    {
      "epoch": 0.7189688756268391,
      "grad_norm": 0.42245447635650635,
      "learning_rate": 1.405362841394173e-06,
      "loss": 0.6521,
      "step": 17348
    },
    {
      "epoch": 0.7190103195325127,
      "grad_norm": 0.4353671967983246,
      "learning_rate": 1.4051556218658046e-06,
      "loss": 0.7292,
      "step": 17349
    },
    {
      "epoch": 0.7190517634381864,
      "grad_norm": 0.45536428689956665,
      "learning_rate": 1.4049484023374362e-06,
      "loss": 0.6943,
      "step": 17350
    },
    {
      "epoch": 0.7190932073438601,
      "grad_norm": 0.4157581627368927,
      "learning_rate": 1.4047411828090678e-06,
      "loss": 0.6274,
      "step": 17351
    },
    {
      "epoch": 0.7191346512495338,
      "grad_norm": 0.42575499415397644,
      "learning_rate": 1.4045339632806999e-06,
      "loss": 0.7222,
      "step": 17352
    },
    {
      "epoch": 0.7191760951552074,
      "grad_norm": 0.44246596097946167,
      "learning_rate": 1.4043267437523315e-06,
      "loss": 0.6748,
      "step": 17353
    },
    {
      "epoch": 0.7192175390608812,
      "grad_norm": 0.40144142508506775,
      "learning_rate": 1.404119524223963e-06,
      "loss": 0.6716,
      "step": 17354
    },
    {
      "epoch": 0.7192589829665548,
      "grad_norm": 0.4067979156970978,
      "learning_rate": 1.4039123046955946e-06,
      "loss": 0.6147,
      "step": 17355
    },
    {
      "epoch": 0.7193004268722284,
      "grad_norm": 0.408854603767395,
      "learning_rate": 1.4037050851672262e-06,
      "loss": 0.6663,
      "step": 17356
    },
    {
      "epoch": 0.7193418707779021,
      "grad_norm": 0.42338109016418457,
      "learning_rate": 1.4034978656388578e-06,
      "loss": 0.6508,
      "step": 17357
    },
    {
      "epoch": 0.7193833146835757,
      "grad_norm": 0.4001927673816681,
      "learning_rate": 1.4032906461104894e-06,
      "loss": 0.6415,
      "step": 17358
    },
    {
      "epoch": 0.7194247585892495,
      "grad_norm": 0.41887569427490234,
      "learning_rate": 1.403083426582121e-06,
      "loss": 0.6768,
      "step": 17359
    },
    {
      "epoch": 0.7194662024949231,
      "grad_norm": 0.39018580317497253,
      "learning_rate": 1.402876207053753e-06,
      "loss": 0.6296,
      "step": 17360
    },
    {
      "epoch": 0.7195076464005968,
      "grad_norm": 0.44853100180625916,
      "learning_rate": 1.4026689875253847e-06,
      "loss": 0.7053,
      "step": 17361
    },
    {
      "epoch": 0.7195490903062705,
      "grad_norm": 0.39521580934524536,
      "learning_rate": 1.4024617679970163e-06,
      "loss": 0.665,
      "step": 17362
    },
    {
      "epoch": 0.7195905342119442,
      "grad_norm": 0.389913946390152,
      "learning_rate": 1.4022545484686479e-06,
      "loss": 0.6477,
      "step": 17363
    },
    {
      "epoch": 0.7196319781176178,
      "grad_norm": 0.3807331621646881,
      "learning_rate": 1.4020473289402794e-06,
      "loss": 0.6251,
      "step": 17364
    },
    {
      "epoch": 0.7196734220232914,
      "grad_norm": 0.4107075035572052,
      "learning_rate": 1.401840109411911e-06,
      "loss": 0.6401,
      "step": 17365
    },
    {
      "epoch": 0.7197148659289652,
      "grad_norm": 0.4643884301185608,
      "learning_rate": 1.4016328898835426e-06,
      "loss": 0.7307,
      "step": 17366
    },
    {
      "epoch": 0.7197563098346388,
      "grad_norm": 0.4370321035385132,
      "learning_rate": 1.4014256703551742e-06,
      "loss": 0.6801,
      "step": 17367
    },
    {
      "epoch": 0.7197977537403125,
      "grad_norm": 0.43240052461624146,
      "learning_rate": 1.4012184508268058e-06,
      "loss": 0.6348,
      "step": 17368
    },
    {
      "epoch": 0.7198391976459861,
      "grad_norm": 0.41039302945137024,
      "learning_rate": 1.4010112312984379e-06,
      "loss": 0.6488,
      "step": 17369
    },
    {
      "epoch": 0.7198806415516599,
      "grad_norm": 0.42606598138809204,
      "learning_rate": 1.4008040117700695e-06,
      "loss": 0.6907,
      "step": 17370
    },
    {
      "epoch": 0.7199220854573335,
      "grad_norm": 0.3793336749076843,
      "learning_rate": 1.400596792241701e-06,
      "loss": 0.6514,
      "step": 17371
    },
    {
      "epoch": 0.7199635293630071,
      "grad_norm": 0.4357008635997772,
      "learning_rate": 1.4003895727133327e-06,
      "loss": 0.6796,
      "step": 17372
    },
    {
      "epoch": 0.7200049732686808,
      "grad_norm": 0.38215315341949463,
      "learning_rate": 1.4001823531849642e-06,
      "loss": 0.6132,
      "step": 17373
    },
    {
      "epoch": 0.7200464171743545,
      "grad_norm": 0.4166969954967499,
      "learning_rate": 1.3999751336565958e-06,
      "loss": 0.679,
      "step": 17374
    },
    {
      "epoch": 0.7200878610800282,
      "grad_norm": 0.4241994321346283,
      "learning_rate": 1.3997679141282274e-06,
      "loss": 0.6874,
      "step": 17375
    },
    {
      "epoch": 0.7201293049857018,
      "grad_norm": 0.42964982986450195,
      "learning_rate": 1.399560694599859e-06,
      "loss": 0.6931,
      "step": 17376
    },
    {
      "epoch": 0.7201707488913756,
      "grad_norm": 0.4299412965774536,
      "learning_rate": 1.3993534750714906e-06,
      "loss": 0.6727,
      "step": 17377
    },
    {
      "epoch": 0.7202121927970492,
      "grad_norm": 0.39656901359558105,
      "learning_rate": 1.3991462555431227e-06,
      "loss": 0.6135,
      "step": 17378
    },
    {
      "epoch": 0.7202536367027229,
      "grad_norm": 0.4515514671802521,
      "learning_rate": 1.3989390360147543e-06,
      "loss": 0.6731,
      "step": 17379
    },
    {
      "epoch": 0.7202950806083965,
      "grad_norm": 0.4034215211868286,
      "learning_rate": 1.3987318164863859e-06,
      "loss": 0.6526,
      "step": 17380
    },
    {
      "epoch": 0.7203365245140702,
      "grad_norm": 0.4588306248188019,
      "learning_rate": 1.3985245969580175e-06,
      "loss": 0.7166,
      "step": 17381
    },
    {
      "epoch": 0.7203779684197439,
      "grad_norm": 0.4249785244464874,
      "learning_rate": 1.398317377429649e-06,
      "loss": 0.6841,
      "step": 17382
    },
    {
      "epoch": 0.7204194123254175,
      "grad_norm": 0.40178245306015015,
      "learning_rate": 1.3981101579012806e-06,
      "loss": 0.6841,
      "step": 17383
    },
    {
      "epoch": 0.7204608562310912,
      "grad_norm": 0.3789336681365967,
      "learning_rate": 1.3979029383729122e-06,
      "loss": 0.6211,
      "step": 17384
    },
    {
      "epoch": 0.7205023001367649,
      "grad_norm": 0.4140903651714325,
      "learning_rate": 1.3976957188445438e-06,
      "loss": 0.6425,
      "step": 17385
    },
    {
      "epoch": 0.7205437440424386,
      "grad_norm": 0.404009073972702,
      "learning_rate": 1.3974884993161759e-06,
      "loss": 0.6903,
      "step": 17386
    },
    {
      "epoch": 0.7205851879481122,
      "grad_norm": 0.45678526163101196,
      "learning_rate": 1.3972812797878075e-06,
      "loss": 0.6659,
      "step": 17387
    },
    {
      "epoch": 0.720626631853786,
      "grad_norm": 0.4432964324951172,
      "learning_rate": 1.397074060259439e-06,
      "loss": 0.7249,
      "step": 17388
    },
    {
      "epoch": 0.7206680757594596,
      "grad_norm": 0.41829541325569153,
      "learning_rate": 1.3968668407310707e-06,
      "loss": 0.6658,
      "step": 17389
    },
    {
      "epoch": 0.7207095196651332,
      "grad_norm": 0.42822718620300293,
      "learning_rate": 1.3966596212027023e-06,
      "loss": 0.6453,
      "step": 17390
    },
    {
      "epoch": 0.7207509635708069,
      "grad_norm": 0.3907715976238251,
      "learning_rate": 1.3964524016743338e-06,
      "loss": 0.6774,
      "step": 17391
    },
    {
      "epoch": 0.7207924074764805,
      "grad_norm": 0.4679035544395447,
      "learning_rate": 1.3962451821459654e-06,
      "loss": 0.7021,
      "step": 17392
    },
    {
      "epoch": 0.7208338513821543,
      "grad_norm": 0.45747002959251404,
      "learning_rate": 1.396037962617597e-06,
      "loss": 0.7384,
      "step": 17393
    },
    {
      "epoch": 0.7208752952878279,
      "grad_norm": 0.4033137857913971,
      "learning_rate": 1.3958307430892286e-06,
      "loss": 0.6794,
      "step": 17394
    },
    {
      "epoch": 0.7209167391935016,
      "grad_norm": 0.45067644119262695,
      "learning_rate": 1.3956235235608607e-06,
      "loss": 0.7114,
      "step": 17395
    },
    {
      "epoch": 0.7209581830991753,
      "grad_norm": 0.3820972740650177,
      "learning_rate": 1.3954163040324923e-06,
      "loss": 0.6348,
      "step": 17396
    },
    {
      "epoch": 0.720999627004849,
      "grad_norm": 0.42485058307647705,
      "learning_rate": 1.3952090845041239e-06,
      "loss": 0.6766,
      "step": 17397
    },
    {
      "epoch": 0.7210410709105226,
      "grad_norm": 0.4396957457065582,
      "learning_rate": 1.3950018649757555e-06,
      "loss": 0.6793,
      "step": 17398
    },
    {
      "epoch": 0.7210825148161962,
      "grad_norm": 0.3938605487346649,
      "learning_rate": 1.394794645447387e-06,
      "loss": 0.6862,
      "step": 17399
    },
    {
      "epoch": 0.72112395872187,
      "grad_norm": 0.4191858172416687,
      "learning_rate": 1.3945874259190187e-06,
      "loss": 0.7136,
      "step": 17400
    },
    {
      "epoch": 0.7211654026275436,
      "grad_norm": 0.4656555652618408,
      "learning_rate": 1.3943802063906502e-06,
      "loss": 0.7263,
      "step": 17401
    },
    {
      "epoch": 0.7212068465332173,
      "grad_norm": 0.4497932195663452,
      "learning_rate": 1.3941729868622818e-06,
      "loss": 0.6488,
      "step": 17402
    },
    {
      "epoch": 0.7212482904388909,
      "grad_norm": 0.45916157960891724,
      "learning_rate": 1.3939657673339134e-06,
      "loss": 0.6682,
      "step": 17403
    },
    {
      "epoch": 0.7212897343445647,
      "grad_norm": 0.4112190902233124,
      "learning_rate": 1.3937585478055455e-06,
      "loss": 0.6881,
      "step": 17404
    },
    {
      "epoch": 0.7213311782502383,
      "grad_norm": 0.38071346282958984,
      "learning_rate": 1.393551328277177e-06,
      "loss": 0.6703,
      "step": 17405
    },
    {
      "epoch": 0.721372622155912,
      "grad_norm": 0.4386751055717468,
      "learning_rate": 1.3933441087488087e-06,
      "loss": 0.6342,
      "step": 17406
    },
    {
      "epoch": 0.7214140660615856,
      "grad_norm": 0.4358382225036621,
      "learning_rate": 1.3931368892204403e-06,
      "loss": 0.7031,
      "step": 17407
    },
    {
      "epoch": 0.7214555099672593,
      "grad_norm": 0.4239400029182434,
      "learning_rate": 1.3929296696920719e-06,
      "loss": 0.6095,
      "step": 17408
    },
    {
      "epoch": 0.721496953872933,
      "grad_norm": 0.3753144443035126,
      "learning_rate": 1.3927224501637035e-06,
      "loss": 0.6804,
      "step": 17409
    },
    {
      "epoch": 0.7215383977786066,
      "grad_norm": 0.38834258913993835,
      "learning_rate": 1.392515230635335e-06,
      "loss": 0.6718,
      "step": 17410
    },
    {
      "epoch": 0.7215798416842804,
      "grad_norm": 0.40292462706565857,
      "learning_rate": 1.3923080111069666e-06,
      "loss": 0.6414,
      "step": 17411
    },
    {
      "epoch": 0.721621285589954,
      "grad_norm": 0.43153274059295654,
      "learning_rate": 1.3921007915785982e-06,
      "loss": 0.7106,
      "step": 17412
    },
    {
      "epoch": 0.7216627294956277,
      "grad_norm": 0.39198118448257446,
      "learning_rate": 1.3918935720502303e-06,
      "loss": 0.6172,
      "step": 17413
    },
    {
      "epoch": 0.7217041734013013,
      "grad_norm": 0.4350663721561432,
      "learning_rate": 1.3916863525218619e-06,
      "loss": 0.6273,
      "step": 17414
    },
    {
      "epoch": 0.7217456173069751,
      "grad_norm": 0.39641499519348145,
      "learning_rate": 1.3914791329934935e-06,
      "loss": 0.6427,
      "step": 17415
    },
    {
      "epoch": 0.7217870612126487,
      "grad_norm": 0.43966054916381836,
      "learning_rate": 1.391271913465125e-06,
      "loss": 0.7163,
      "step": 17416
    },
    {
      "epoch": 0.7218285051183223,
      "grad_norm": 0.40090662240982056,
      "learning_rate": 1.3910646939367567e-06,
      "loss": 0.6404,
      "step": 17417
    },
    {
      "epoch": 0.721869949023996,
      "grad_norm": 0.42804622650146484,
      "learning_rate": 1.3908574744083883e-06,
      "loss": 0.6621,
      "step": 17418
    },
    {
      "epoch": 0.7219113929296697,
      "grad_norm": 0.39885470271110535,
      "learning_rate": 1.3906502548800198e-06,
      "loss": 0.656,
      "step": 17419
    },
    {
      "epoch": 0.7219528368353434,
      "grad_norm": 0.4252878427505493,
      "learning_rate": 1.3904430353516514e-06,
      "loss": 0.7488,
      "step": 17420
    },
    {
      "epoch": 0.721994280741017,
      "grad_norm": 0.424329936504364,
      "learning_rate": 1.3902358158232835e-06,
      "loss": 0.6688,
      "step": 17421
    },
    {
      "epoch": 0.7220357246466907,
      "grad_norm": 0.4304868280887604,
      "learning_rate": 1.390028596294915e-06,
      "loss": 0.6442,
      "step": 17422
    },
    {
      "epoch": 0.7220771685523644,
      "grad_norm": 0.46468403935432434,
      "learning_rate": 1.3898213767665467e-06,
      "loss": 0.6812,
      "step": 17423
    },
    {
      "epoch": 0.7221186124580381,
      "grad_norm": 0.4022739827632904,
      "learning_rate": 1.3896141572381783e-06,
      "loss": 0.6503,
      "step": 17424
    },
    {
      "epoch": 0.7221600563637117,
      "grad_norm": 0.4108484983444214,
      "learning_rate": 1.3894069377098099e-06,
      "loss": 0.678,
      "step": 17425
    },
    {
      "epoch": 0.7222015002693853,
      "grad_norm": 0.40491557121276855,
      "learning_rate": 1.3891997181814415e-06,
      "loss": 0.6445,
      "step": 17426
    },
    {
      "epoch": 0.7222429441750591,
      "grad_norm": 0.4266613721847534,
      "learning_rate": 1.388992498653073e-06,
      "loss": 0.6813,
      "step": 17427
    },
    {
      "epoch": 0.7222843880807327,
      "grad_norm": 0.4279089570045471,
      "learning_rate": 1.3887852791247046e-06,
      "loss": 0.7163,
      "step": 17428
    },
    {
      "epoch": 0.7223258319864064,
      "grad_norm": 0.3997454345226288,
      "learning_rate": 1.3885780595963362e-06,
      "loss": 0.6721,
      "step": 17429
    },
    {
      "epoch": 0.72236727589208,
      "grad_norm": 0.4164501130580902,
      "learning_rate": 1.3883708400679683e-06,
      "loss": 0.6871,
      "step": 17430
    },
    {
      "epoch": 0.7224087197977538,
      "grad_norm": 0.39458733797073364,
      "learning_rate": 1.3881636205395999e-06,
      "loss": 0.6342,
      "step": 17431
    },
    {
      "epoch": 0.7224501637034274,
      "grad_norm": 0.42602452635765076,
      "learning_rate": 1.3879564010112315e-06,
      "loss": 0.6781,
      "step": 17432
    },
    {
      "epoch": 0.722491607609101,
      "grad_norm": 0.40976107120513916,
      "learning_rate": 1.387749181482863e-06,
      "loss": 0.6068,
      "step": 17433
    },
    {
      "epoch": 0.7225330515147748,
      "grad_norm": 0.4227009415626526,
      "learning_rate": 1.3875419619544947e-06,
      "loss": 0.6719,
      "step": 17434
    },
    {
      "epoch": 0.7225744954204484,
      "grad_norm": 0.4550377428531647,
      "learning_rate": 1.3873347424261263e-06,
      "loss": 0.7133,
      "step": 17435
    },
    {
      "epoch": 0.7226159393261221,
      "grad_norm": 0.4322700500488281,
      "learning_rate": 1.3871275228977579e-06,
      "loss": 0.6755,
      "step": 17436
    },
    {
      "epoch": 0.7226573832317957,
      "grad_norm": 0.4307835102081299,
      "learning_rate": 1.3869203033693894e-06,
      "loss": 0.6885,
      "step": 17437
    },
    {
      "epoch": 0.7226988271374695,
      "grad_norm": 0.4248655140399933,
      "learning_rate": 1.3867130838410213e-06,
      "loss": 0.6736,
      "step": 17438
    },
    {
      "epoch": 0.7227402710431431,
      "grad_norm": 0.4186408519744873,
      "learning_rate": 1.386505864312653e-06,
      "loss": 0.6873,
      "step": 17439
    },
    {
      "epoch": 0.7227817149488168,
      "grad_norm": 0.40119022130966187,
      "learning_rate": 1.3862986447842847e-06,
      "loss": 0.6685,
      "step": 17440
    },
    {
      "epoch": 0.7228231588544904,
      "grad_norm": 0.4152214229106903,
      "learning_rate": 1.3860914252559163e-06,
      "loss": 0.6788,
      "step": 17441
    },
    {
      "epoch": 0.7228646027601641,
      "grad_norm": 0.3952624201774597,
      "learning_rate": 1.3858842057275479e-06,
      "loss": 0.652,
      "step": 17442
    },
    {
      "epoch": 0.7229060466658378,
      "grad_norm": 0.3806418776512146,
      "learning_rate": 1.3856769861991795e-06,
      "loss": 0.6597,
      "step": 17443
    },
    {
      "epoch": 0.7229474905715114,
      "grad_norm": 0.3794182538986206,
      "learning_rate": 1.385469766670811e-06,
      "loss": 0.637,
      "step": 17444
    },
    {
      "epoch": 0.7229889344771852,
      "grad_norm": 0.42944058775901794,
      "learning_rate": 1.3852625471424427e-06,
      "loss": 0.6852,
      "step": 17445
    },
    {
      "epoch": 0.7230303783828588,
      "grad_norm": 0.43929627537727356,
      "learning_rate": 1.3850553276140742e-06,
      "loss": 0.6161,
      "step": 17446
    },
    {
      "epoch": 0.7230718222885325,
      "grad_norm": 0.4343337416648865,
      "learning_rate": 1.3848481080857063e-06,
      "loss": 0.6794,
      "step": 17447
    },
    {
      "epoch": 0.7231132661942061,
      "grad_norm": 0.4136296212673187,
      "learning_rate": 1.3846408885573379e-06,
      "loss": 0.6921,
      "step": 17448
    },
    {
      "epoch": 0.7231547100998799,
      "grad_norm": 0.4051271080970764,
      "learning_rate": 1.3844336690289695e-06,
      "loss": 0.6455,
      "step": 17449
    },
    {
      "epoch": 0.7231961540055535,
      "grad_norm": 0.46531012654304504,
      "learning_rate": 1.384226449500601e-06,
      "loss": 0.6897,
      "step": 17450
    },
    {
      "epoch": 0.7232375979112271,
      "grad_norm": 0.3971078097820282,
      "learning_rate": 1.3840192299722327e-06,
      "loss": 0.6283,
      "step": 17451
    },
    {
      "epoch": 0.7232790418169008,
      "grad_norm": 0.4227045476436615,
      "learning_rate": 1.3838120104438643e-06,
      "loss": 0.668,
      "step": 17452
    },
    {
      "epoch": 0.7233204857225745,
      "grad_norm": 0.38663744926452637,
      "learning_rate": 1.3836047909154959e-06,
      "loss": 0.6387,
      "step": 17453
    },
    {
      "epoch": 0.7233619296282482,
      "grad_norm": 0.39325106143951416,
      "learning_rate": 1.3833975713871275e-06,
      "loss": 0.6163,
      "step": 17454
    },
    {
      "epoch": 0.7234033735339218,
      "grad_norm": 0.4134993255138397,
      "learning_rate": 1.3831903518587593e-06,
      "loss": 0.6733,
      "step": 17455
    },
    {
      "epoch": 0.7234448174395955,
      "grad_norm": 0.41026341915130615,
      "learning_rate": 1.382983132330391e-06,
      "loss": 0.6637,
      "step": 17456
    },
    {
      "epoch": 0.7234862613452692,
      "grad_norm": 0.4146248698234558,
      "learning_rate": 1.3827759128020227e-06,
      "loss": 0.6873,
      "step": 17457
    },
    {
      "epoch": 0.7235277052509429,
      "grad_norm": 0.4182088375091553,
      "learning_rate": 1.3825686932736543e-06,
      "loss": 0.6658,
      "step": 17458
    },
    {
      "epoch": 0.7235691491566165,
      "grad_norm": 0.37575212121009827,
      "learning_rate": 1.3823614737452859e-06,
      "loss": 0.6208,
      "step": 17459
    },
    {
      "epoch": 0.7236105930622901,
      "grad_norm": 0.4248543083667755,
      "learning_rate": 1.3821542542169175e-06,
      "loss": 0.6736,
      "step": 17460
    },
    {
      "epoch": 0.7236520369679639,
      "grad_norm": 0.3931140899658203,
      "learning_rate": 1.381947034688549e-06,
      "loss": 0.6249,
      "step": 17461
    },
    {
      "epoch": 0.7236934808736375,
      "grad_norm": 0.47457680106163025,
      "learning_rate": 1.3817398151601807e-06,
      "loss": 0.7053,
      "step": 17462
    },
    {
      "epoch": 0.7237349247793112,
      "grad_norm": 0.3681237995624542,
      "learning_rate": 1.3815325956318123e-06,
      "loss": 0.6289,
      "step": 17463
    },
    {
      "epoch": 0.7237763686849848,
      "grad_norm": 0.39734208583831787,
      "learning_rate": 1.381325376103444e-06,
      "loss": 0.6608,
      "step": 17464
    },
    {
      "epoch": 0.7238178125906586,
      "grad_norm": 0.4295814633369446,
      "learning_rate": 1.3811181565750759e-06,
      "loss": 0.7007,
      "step": 17465
    },
    {
      "epoch": 0.7238592564963322,
      "grad_norm": 0.4449225068092346,
      "learning_rate": 1.3809109370467075e-06,
      "loss": 0.705,
      "step": 17466
    },
    {
      "epoch": 0.7239007004020059,
      "grad_norm": 0.3936707377433777,
      "learning_rate": 1.380703717518339e-06,
      "loss": 0.6432,
      "step": 17467
    },
    {
      "epoch": 0.7239421443076796,
      "grad_norm": 0.4330900013446808,
      "learning_rate": 1.3804964979899707e-06,
      "loss": 0.6963,
      "step": 17468
    },
    {
      "epoch": 0.7239835882133532,
      "grad_norm": 0.40855249762535095,
      "learning_rate": 1.3802892784616023e-06,
      "loss": 0.6946,
      "step": 17469
    },
    {
      "epoch": 0.7240250321190269,
      "grad_norm": 0.40544137358665466,
      "learning_rate": 1.3800820589332339e-06,
      "loss": 0.6625,
      "step": 17470
    },
    {
      "epoch": 0.7240664760247005,
      "grad_norm": 0.4054180681705475,
      "learning_rate": 1.3798748394048655e-06,
      "loss": 0.6678,
      "step": 17471
    },
    {
      "epoch": 0.7241079199303743,
      "grad_norm": 0.4636663794517517,
      "learning_rate": 1.3796676198764973e-06,
      "loss": 0.7051,
      "step": 17472
    },
    {
      "epoch": 0.7241493638360479,
      "grad_norm": 0.39145541191101074,
      "learning_rate": 1.379460400348129e-06,
      "loss": 0.7006,
      "step": 17473
    },
    {
      "epoch": 0.7241908077417216,
      "grad_norm": 0.3848302364349365,
      "learning_rate": 1.3792531808197607e-06,
      "loss": 0.6835,
      "step": 17474
    },
    {
      "epoch": 0.7242322516473952,
      "grad_norm": 0.40552088618278503,
      "learning_rate": 1.3790459612913923e-06,
      "loss": 0.6522,
      "step": 17475
    },
    {
      "epoch": 0.724273695553069,
      "grad_norm": 0.3503236472606659,
      "learning_rate": 1.3788387417630239e-06,
      "loss": 0.5884,
      "step": 17476
    },
    {
      "epoch": 0.7243151394587426,
      "grad_norm": 0.4048171043395996,
      "learning_rate": 1.3786315222346555e-06,
      "loss": 0.6774,
      "step": 17477
    },
    {
      "epoch": 0.7243565833644162,
      "grad_norm": 0.42481541633605957,
      "learning_rate": 1.378424302706287e-06,
      "loss": 0.6759,
      "step": 17478
    },
    {
      "epoch": 0.72439802727009,
      "grad_norm": 0.39376363158226013,
      "learning_rate": 1.3782170831779187e-06,
      "loss": 0.6152,
      "step": 17479
    },
    {
      "epoch": 0.7244394711757636,
      "grad_norm": 0.393154114484787,
      "learning_rate": 1.3780098636495505e-06,
      "loss": 0.6886,
      "step": 17480
    },
    {
      "epoch": 0.7244809150814373,
      "grad_norm": 0.42169472575187683,
      "learning_rate": 1.377802644121182e-06,
      "loss": 0.6632,
      "step": 17481
    },
    {
      "epoch": 0.7245223589871109,
      "grad_norm": 0.44769179821014404,
      "learning_rate": 1.3775954245928139e-06,
      "loss": 0.6904,
      "step": 17482
    },
    {
      "epoch": 0.7245638028927847,
      "grad_norm": 0.41905510425567627,
      "learning_rate": 1.3773882050644455e-06,
      "loss": 0.6106,
      "step": 17483
    },
    {
      "epoch": 0.7246052467984583,
      "grad_norm": 0.4448399245738983,
      "learning_rate": 1.377180985536077e-06,
      "loss": 0.701,
      "step": 17484
    },
    {
      "epoch": 0.724646690704132,
      "grad_norm": 0.4232357144355774,
      "learning_rate": 1.3769737660077087e-06,
      "loss": 0.6691,
      "step": 17485
    },
    {
      "epoch": 0.7246881346098056,
      "grad_norm": 0.43226203322410583,
      "learning_rate": 1.3767665464793403e-06,
      "loss": 0.6943,
      "step": 17486
    },
    {
      "epoch": 0.7247295785154793,
      "grad_norm": 0.4052192270755768,
      "learning_rate": 1.3765593269509719e-06,
      "loss": 0.6826,
      "step": 17487
    },
    {
      "epoch": 0.724771022421153,
      "grad_norm": 0.3941108286380768,
      "learning_rate": 1.3763521074226035e-06,
      "loss": 0.5892,
      "step": 17488
    },
    {
      "epoch": 0.7248124663268266,
      "grad_norm": 0.40341296792030334,
      "learning_rate": 1.3761448878942353e-06,
      "loss": 0.6149,
      "step": 17489
    },
    {
      "epoch": 0.7248539102325003,
      "grad_norm": 0.41525524854660034,
      "learning_rate": 1.3759376683658669e-06,
      "loss": 0.6555,
      "step": 17490
    },
    {
      "epoch": 0.724895354138174,
      "grad_norm": 0.45307183265686035,
      "learning_rate": 1.3757304488374987e-06,
      "loss": 0.7219,
      "step": 17491
    },
    {
      "epoch": 0.7249367980438477,
      "grad_norm": 0.41568824648857117,
      "learning_rate": 1.3755232293091303e-06,
      "loss": 0.6758,
      "step": 17492
    },
    {
      "epoch": 0.7249782419495213,
      "grad_norm": 0.38459259271621704,
      "learning_rate": 1.3753160097807619e-06,
      "loss": 0.6287,
      "step": 17493
    },
    {
      "epoch": 0.7250196858551949,
      "grad_norm": 0.39925843477249146,
      "learning_rate": 1.3751087902523935e-06,
      "loss": 0.6636,
      "step": 17494
    },
    {
      "epoch": 0.7250611297608687,
      "grad_norm": 0.37855085730552673,
      "learning_rate": 1.374901570724025e-06,
      "loss": 0.6729,
      "step": 17495
    },
    {
      "epoch": 0.7251025736665423,
      "grad_norm": 0.4009250998497009,
      "learning_rate": 1.3746943511956567e-06,
      "loss": 0.6204,
      "step": 17496
    },
    {
      "epoch": 0.725144017572216,
      "grad_norm": 0.4213401675224304,
      "learning_rate": 1.3744871316672885e-06,
      "loss": 0.6658,
      "step": 17497
    },
    {
      "epoch": 0.7251854614778896,
      "grad_norm": 0.4160116910934448,
      "learning_rate": 1.37427991213892e-06,
      "loss": 0.6577,
      "step": 17498
    },
    {
      "epoch": 0.7252269053835634,
      "grad_norm": 0.40867945551872253,
      "learning_rate": 1.3740726926105517e-06,
      "loss": 0.6694,
      "step": 17499
    },
    {
      "epoch": 0.725268349289237,
      "grad_norm": 0.41269201040267944,
      "learning_rate": 1.3738654730821835e-06,
      "loss": 0.7019,
      "step": 17500
    },
    {
      "epoch": 0.7253097931949107,
      "grad_norm": 0.43230336904525757,
      "learning_rate": 1.373658253553815e-06,
      "loss": 0.6516,
      "step": 17501
    },
    {
      "epoch": 0.7253512371005844,
      "grad_norm": 0.4137008488178253,
      "learning_rate": 1.3734510340254467e-06,
      "loss": 0.6697,
      "step": 17502
    },
    {
      "epoch": 0.725392681006258,
      "grad_norm": 0.4572915732860565,
      "learning_rate": 1.3732438144970783e-06,
      "loss": 0.7029,
      "step": 17503
    },
    {
      "epoch": 0.7254341249119317,
      "grad_norm": 0.43488359451293945,
      "learning_rate": 1.3730365949687099e-06,
      "loss": 0.6803,
      "step": 17504
    },
    {
      "epoch": 0.7254755688176053,
      "grad_norm": 0.39924558997154236,
      "learning_rate": 1.3728293754403415e-06,
      "loss": 0.6279,
      "step": 17505
    },
    {
      "epoch": 0.7255170127232791,
      "grad_norm": 0.4372648000717163,
      "learning_rate": 1.3726221559119733e-06,
      "loss": 0.6963,
      "step": 17506
    },
    {
      "epoch": 0.7255584566289527,
      "grad_norm": 0.4213685393333435,
      "learning_rate": 1.3724149363836049e-06,
      "loss": 0.7173,
      "step": 17507
    },
    {
      "epoch": 0.7255999005346264,
      "grad_norm": 0.405083566904068,
      "learning_rate": 1.3722077168552367e-06,
      "loss": 0.6528,
      "step": 17508
    },
    {
      "epoch": 0.7256413444403,
      "grad_norm": 0.4007328450679779,
      "learning_rate": 1.3720004973268683e-06,
      "loss": 0.661,
      "step": 17509
    },
    {
      "epoch": 0.7256827883459738,
      "grad_norm": 0.44553348422050476,
      "learning_rate": 1.3717932777984999e-06,
      "loss": 0.7239,
      "step": 17510
    },
    {
      "epoch": 0.7257242322516474,
      "grad_norm": 0.4048614203929901,
      "learning_rate": 1.3715860582701315e-06,
      "loss": 0.6403,
      "step": 17511
    },
    {
      "epoch": 0.725765676157321,
      "grad_norm": 0.4508402347564697,
      "learning_rate": 1.371378838741763e-06,
      "loss": 0.7141,
      "step": 17512
    },
    {
      "epoch": 0.7258071200629947,
      "grad_norm": 0.41269198060035706,
      "learning_rate": 1.3711716192133947e-06,
      "loss": 0.6573,
      "step": 17513
    },
    {
      "epoch": 0.7258485639686684,
      "grad_norm": 0.442448228597641,
      "learning_rate": 1.3709643996850265e-06,
      "loss": 0.6686,
      "step": 17514
    },
    {
      "epoch": 0.7258900078743421,
      "grad_norm": 0.3905901610851288,
      "learning_rate": 1.370757180156658e-06,
      "loss": 0.7144,
      "step": 17515
    },
    {
      "epoch": 0.7259314517800157,
      "grad_norm": 0.42178258299827576,
      "learning_rate": 1.3705499606282897e-06,
      "loss": 0.624,
      "step": 17516
    },
    {
      "epoch": 0.7259728956856895,
      "grad_norm": 0.42789697647094727,
      "learning_rate": 1.3703427410999215e-06,
      "loss": 0.717,
      "step": 17517
    },
    {
      "epoch": 0.7260143395913631,
      "grad_norm": 0.40790408849716187,
      "learning_rate": 1.370135521571553e-06,
      "loss": 0.6204,
      "step": 17518
    },
    {
      "epoch": 0.7260557834970368,
      "grad_norm": 0.4066314995288849,
      "learning_rate": 1.3699283020431847e-06,
      "loss": 0.6973,
      "step": 17519
    },
    {
      "epoch": 0.7260972274027104,
      "grad_norm": 0.4601806700229645,
      "learning_rate": 1.3697210825148163e-06,
      "loss": 0.6786,
      "step": 17520
    },
    {
      "epoch": 0.726138671308384,
      "grad_norm": 0.41776859760284424,
      "learning_rate": 1.3695138629864479e-06,
      "loss": 0.6711,
      "step": 17521
    },
    {
      "epoch": 0.7261801152140578,
      "grad_norm": 0.3985465168952942,
      "learning_rate": 1.3693066434580795e-06,
      "loss": 0.6703,
      "step": 17522
    },
    {
      "epoch": 0.7262215591197314,
      "grad_norm": 0.43471282720565796,
      "learning_rate": 1.3690994239297113e-06,
      "loss": 0.6799,
      "step": 17523
    },
    {
      "epoch": 0.7262630030254051,
      "grad_norm": 0.4013284146785736,
      "learning_rate": 1.3688922044013429e-06,
      "loss": 0.6896,
      "step": 17524
    },
    {
      "epoch": 0.7263044469310788,
      "grad_norm": 0.4212099313735962,
      "learning_rate": 1.3686849848729745e-06,
      "loss": 0.6815,
      "step": 17525
    },
    {
      "epoch": 0.7263458908367525,
      "grad_norm": 0.4064970314502716,
      "learning_rate": 1.3684777653446063e-06,
      "loss": 0.6553,
      "step": 17526
    },
    {
      "epoch": 0.7263873347424261,
      "grad_norm": 0.42406266927719116,
      "learning_rate": 1.3682705458162379e-06,
      "loss": 0.72,
      "step": 17527
    },
    {
      "epoch": 0.7264287786480998,
      "grad_norm": 0.4300520718097687,
      "learning_rate": 1.3680633262878695e-06,
      "loss": 0.6919,
      "step": 17528
    },
    {
      "epoch": 0.7264702225537735,
      "grad_norm": 0.4189138114452362,
      "learning_rate": 1.367856106759501e-06,
      "loss": 0.6626,
      "step": 17529
    },
    {
      "epoch": 0.7265116664594471,
      "grad_norm": 0.40742337703704834,
      "learning_rate": 1.3676488872311327e-06,
      "loss": 0.6467,
      "step": 17530
    },
    {
      "epoch": 0.7265531103651208,
      "grad_norm": 0.39453136920928955,
      "learning_rate": 1.3674416677027645e-06,
      "loss": 0.6093,
      "step": 17531
    },
    {
      "epoch": 0.7265945542707944,
      "grad_norm": 0.3954002559185028,
      "learning_rate": 1.367234448174396e-06,
      "loss": 0.6655,
      "step": 17532
    },
    {
      "epoch": 0.7266359981764682,
      "grad_norm": 0.3810215890407562,
      "learning_rate": 1.3670272286460277e-06,
      "loss": 0.6763,
      "step": 17533
    },
    {
      "epoch": 0.7266774420821418,
      "grad_norm": 0.41593435406684875,
      "learning_rate": 1.3668200091176595e-06,
      "loss": 0.646,
      "step": 17534
    },
    {
      "epoch": 0.7267188859878155,
      "grad_norm": 0.4103779196739197,
      "learning_rate": 1.366612789589291e-06,
      "loss": 0.7041,
      "step": 17535
    },
    {
      "epoch": 0.7267603298934892,
      "grad_norm": 0.4070605933666229,
      "learning_rate": 1.3664055700609227e-06,
      "loss": 0.6302,
      "step": 17536
    },
    {
      "epoch": 0.7268017737991629,
      "grad_norm": 0.39627110958099365,
      "learning_rate": 1.3661983505325543e-06,
      "loss": 0.6998,
      "step": 17537
    },
    {
      "epoch": 0.7268432177048365,
      "grad_norm": 0.3947973847389221,
      "learning_rate": 1.3659911310041859e-06,
      "loss": 0.6761,
      "step": 17538
    },
    {
      "epoch": 0.7268846616105101,
      "grad_norm": 0.41158583760261536,
      "learning_rate": 1.3657839114758175e-06,
      "loss": 0.6313,
      "step": 17539
    },
    {
      "epoch": 0.7269261055161839,
      "grad_norm": 0.4412710964679718,
      "learning_rate": 1.3655766919474493e-06,
      "loss": 0.6859,
      "step": 17540
    },
    {
      "epoch": 0.7269675494218575,
      "grad_norm": 0.3694186210632324,
      "learning_rate": 1.3653694724190809e-06,
      "loss": 0.6188,
      "step": 17541
    },
    {
      "epoch": 0.7270089933275312,
      "grad_norm": 0.4904603362083435,
      "learning_rate": 1.3651622528907125e-06,
      "loss": 0.6853,
      "step": 17542
    },
    {
      "epoch": 0.7270504372332048,
      "grad_norm": 0.42320895195007324,
      "learning_rate": 1.3649550333623443e-06,
      "loss": 0.7369,
      "step": 17543
    },
    {
      "epoch": 0.7270918811388786,
      "grad_norm": 0.4059513807296753,
      "learning_rate": 1.3647478138339759e-06,
      "loss": 0.6735,
      "step": 17544
    },
    {
      "epoch": 0.7271333250445522,
      "grad_norm": 0.3872096836566925,
      "learning_rate": 1.3645405943056075e-06,
      "loss": 0.6746,
      "step": 17545
    },
    {
      "epoch": 0.7271747689502258,
      "grad_norm": 0.39571961760520935,
      "learning_rate": 1.364333374777239e-06,
      "loss": 0.7006,
      "step": 17546
    },
    {
      "epoch": 0.7272162128558995,
      "grad_norm": 0.44586309790611267,
      "learning_rate": 1.3641261552488707e-06,
      "loss": 0.7292,
      "step": 17547
    },
    {
      "epoch": 0.7272576567615732,
      "grad_norm": 0.43289631605148315,
      "learning_rate": 1.3639189357205025e-06,
      "loss": 0.6934,
      "step": 17548
    },
    {
      "epoch": 0.7272991006672469,
      "grad_norm": 0.4167085587978363,
      "learning_rate": 1.363711716192134e-06,
      "loss": 0.6963,
      "step": 17549
    },
    {
      "epoch": 0.7273405445729205,
      "grad_norm": 0.4044798016548157,
      "learning_rate": 1.3635044966637657e-06,
      "loss": 0.6321,
      "step": 17550
    },
    {
      "epoch": 0.7273819884785943,
      "grad_norm": 0.4146758019924164,
      "learning_rate": 1.3632972771353973e-06,
      "loss": 0.6246,
      "step": 17551
    },
    {
      "epoch": 0.7274234323842679,
      "grad_norm": 0.397849977016449,
      "learning_rate": 1.363090057607029e-06,
      "loss": 0.6382,
      "step": 17552
    },
    {
      "epoch": 0.7274648762899416,
      "grad_norm": 0.40415680408477783,
      "learning_rate": 1.3628828380786607e-06,
      "loss": 0.6494,
      "step": 17553
    },
    {
      "epoch": 0.7275063201956152,
      "grad_norm": 0.42717739939689636,
      "learning_rate": 1.3626756185502923e-06,
      "loss": 0.6932,
      "step": 17554
    },
    {
      "epoch": 0.7275477641012889,
      "grad_norm": 0.41101178526878357,
      "learning_rate": 1.3624683990219239e-06,
      "loss": 0.7151,
      "step": 17555
    },
    {
      "epoch": 0.7275892080069626,
      "grad_norm": 0.40928471088409424,
      "learning_rate": 1.3622611794935555e-06,
      "loss": 0.6447,
      "step": 17556
    },
    {
      "epoch": 0.7276306519126362,
      "grad_norm": 0.4402622878551483,
      "learning_rate": 1.3620539599651873e-06,
      "loss": 0.6796,
      "step": 17557
    },
    {
      "epoch": 0.7276720958183099,
      "grad_norm": 0.4760821461677551,
      "learning_rate": 1.3618467404368189e-06,
      "loss": 0.7002,
      "step": 17558
    },
    {
      "epoch": 0.7277135397239836,
      "grad_norm": 0.38029366731643677,
      "learning_rate": 1.3616395209084505e-06,
      "loss": 0.6292,
      "step": 17559
    },
    {
      "epoch": 0.7277549836296573,
      "grad_norm": 0.4174540340900421,
      "learning_rate": 1.361432301380082e-06,
      "loss": 0.658,
      "step": 17560
    },
    {
      "epoch": 0.7277964275353309,
      "grad_norm": 0.4315658509731293,
      "learning_rate": 1.3612250818517139e-06,
      "loss": 0.6744,
      "step": 17561
    },
    {
      "epoch": 0.7278378714410046,
      "grad_norm": 0.3992787003517151,
      "learning_rate": 1.3610178623233455e-06,
      "loss": 0.6451,
      "step": 17562
    },
    {
      "epoch": 0.7278793153466783,
      "grad_norm": 0.42945557832717896,
      "learning_rate": 1.360810642794977e-06,
      "loss": 0.7207,
      "step": 17563
    },
    {
      "epoch": 0.7279207592523519,
      "grad_norm": 0.4137974977493286,
      "learning_rate": 1.3606034232666087e-06,
      "loss": 0.656,
      "step": 17564
    },
    {
      "epoch": 0.7279622031580256,
      "grad_norm": 0.4560147821903229,
      "learning_rate": 1.3603962037382405e-06,
      "loss": 0.6602,
      "step": 17565
    },
    {
      "epoch": 0.7280036470636992,
      "grad_norm": 0.4137943387031555,
      "learning_rate": 1.360188984209872e-06,
      "loss": 0.7002,
      "step": 17566
    },
    {
      "epoch": 0.728045090969373,
      "grad_norm": 0.4310949742794037,
      "learning_rate": 1.3599817646815037e-06,
      "loss": 0.6711,
      "step": 17567
    },
    {
      "epoch": 0.7280865348750466,
      "grad_norm": 0.41603103280067444,
      "learning_rate": 1.3597745451531353e-06,
      "loss": 0.657,
      "step": 17568
    },
    {
      "epoch": 0.7281279787807203,
      "grad_norm": 0.4125109612941742,
      "learning_rate": 1.359567325624767e-06,
      "loss": 0.6594,
      "step": 17569
    },
    {
      "epoch": 0.728169422686394,
      "grad_norm": 0.4280267059803009,
      "learning_rate": 1.3593601060963987e-06,
      "loss": 0.6871,
      "step": 17570
    },
    {
      "epoch": 0.7282108665920677,
      "grad_norm": 0.38965243101119995,
      "learning_rate": 1.3591528865680303e-06,
      "loss": 0.6276,
      "step": 17571
    },
    {
      "epoch": 0.7282523104977413,
      "grad_norm": 0.4093649983406067,
      "learning_rate": 1.3589456670396619e-06,
      "loss": 0.689,
      "step": 17572
    },
    {
      "epoch": 0.7282937544034149,
      "grad_norm": 0.4561555087566376,
      "learning_rate": 1.3587384475112937e-06,
      "loss": 0.6891,
      "step": 17573
    },
    {
      "epoch": 0.7283351983090887,
      "grad_norm": 0.42749494314193726,
      "learning_rate": 1.3585312279829253e-06,
      "loss": 0.6737,
      "step": 17574
    },
    {
      "epoch": 0.7283766422147623,
      "grad_norm": 0.4202386140823364,
      "learning_rate": 1.3583240084545569e-06,
      "loss": 0.6113,
      "step": 17575
    },
    {
      "epoch": 0.728418086120436,
      "grad_norm": 0.3888832628726959,
      "learning_rate": 1.3581167889261885e-06,
      "loss": 0.7107,
      "step": 17576
    },
    {
      "epoch": 0.7284595300261096,
      "grad_norm": 0.4538652002811432,
      "learning_rate": 1.35790956939782e-06,
      "loss": 0.6643,
      "step": 17577
    },
    {
      "epoch": 0.7285009739317834,
      "grad_norm": 0.43097323179244995,
      "learning_rate": 1.3577023498694519e-06,
      "loss": 0.6478,
      "step": 17578
    },
    {
      "epoch": 0.728542417837457,
      "grad_norm": 0.43960240483283997,
      "learning_rate": 1.3574951303410835e-06,
      "loss": 0.6627,
      "step": 17579
    },
    {
      "epoch": 0.7285838617431307,
      "grad_norm": 0.4214801788330078,
      "learning_rate": 1.357287910812715e-06,
      "loss": 0.6461,
      "step": 17580
    },
    {
      "epoch": 0.7286253056488043,
      "grad_norm": 0.44164422154426575,
      "learning_rate": 1.3570806912843467e-06,
      "loss": 0.702,
      "step": 17581
    },
    {
      "epoch": 0.728666749554478,
      "grad_norm": 0.41146859526634216,
      "learning_rate": 1.3568734717559785e-06,
      "loss": 0.618,
      "step": 17582
    },
    {
      "epoch": 0.7287081934601517,
      "grad_norm": 0.42332640290260315,
      "learning_rate": 1.35666625222761e-06,
      "loss": 0.686,
      "step": 17583
    },
    {
      "epoch": 0.7287496373658253,
      "grad_norm": 0.4047934114933014,
      "learning_rate": 1.3564590326992417e-06,
      "loss": 0.6536,
      "step": 17584
    },
    {
      "epoch": 0.728791081271499,
      "grad_norm": 0.39197513461112976,
      "learning_rate": 1.3562518131708733e-06,
      "loss": 0.6379,
      "step": 17585
    },
    {
      "epoch": 0.7288325251771727,
      "grad_norm": 0.4133414626121521,
      "learning_rate": 1.3560445936425049e-06,
      "loss": 0.6965,
      "step": 17586
    },
    {
      "epoch": 0.7288739690828464,
      "grad_norm": 0.4677366614341736,
      "learning_rate": 1.3558373741141367e-06,
      "loss": 0.6368,
      "step": 17587
    },
    {
      "epoch": 0.72891541298852,
      "grad_norm": 0.383446604013443,
      "learning_rate": 1.3556301545857683e-06,
      "loss": 0.6034,
      "step": 17588
    },
    {
      "epoch": 0.7289568568941938,
      "grad_norm": 0.46991679072380066,
      "learning_rate": 1.3554229350573999e-06,
      "loss": 0.719,
      "step": 17589
    },
    {
      "epoch": 0.7289983007998674,
      "grad_norm": 0.4198354482650757,
      "learning_rate": 1.3552157155290317e-06,
      "loss": 0.6687,
      "step": 17590
    },
    {
      "epoch": 0.729039744705541,
      "grad_norm": 0.40832632780075073,
      "learning_rate": 1.3550084960006633e-06,
      "loss": 0.6702,
      "step": 17591
    },
    {
      "epoch": 0.7290811886112147,
      "grad_norm": 0.3995884656906128,
      "learning_rate": 1.3548012764722949e-06,
      "loss": 0.6315,
      "step": 17592
    },
    {
      "epoch": 0.7291226325168884,
      "grad_norm": 0.37485557794570923,
      "learning_rate": 1.3545940569439265e-06,
      "loss": 0.636,
      "step": 17593
    },
    {
      "epoch": 0.7291640764225621,
      "grad_norm": 0.39036667346954346,
      "learning_rate": 1.354386837415558e-06,
      "loss": 0.6782,
      "step": 17594
    },
    {
      "epoch": 0.7292055203282357,
      "grad_norm": 0.4236733317375183,
      "learning_rate": 1.3541796178871899e-06,
      "loss": 0.6836,
      "step": 17595
    },
    {
      "epoch": 0.7292469642339094,
      "grad_norm": 0.41850417852401733,
      "learning_rate": 1.3539723983588215e-06,
      "loss": 0.6467,
      "step": 17596
    },
    {
      "epoch": 0.7292884081395831,
      "grad_norm": 0.4037933051586151,
      "learning_rate": 1.353765178830453e-06,
      "loss": 0.6322,
      "step": 17597
    },
    {
      "epoch": 0.7293298520452568,
      "grad_norm": 0.4398779273033142,
      "learning_rate": 1.3535579593020847e-06,
      "loss": 0.6364,
      "step": 17598
    },
    {
      "epoch": 0.7293712959509304,
      "grad_norm": 0.3938024938106537,
      "learning_rate": 1.3533507397737165e-06,
      "loss": 0.6942,
      "step": 17599
    },
    {
      "epoch": 0.729412739856604,
      "grad_norm": 0.39998334646224976,
      "learning_rate": 1.353143520245348e-06,
      "loss": 0.6399,
      "step": 17600
    },
    {
      "epoch": 0.7294541837622778,
      "grad_norm": 0.42932459712028503,
      "learning_rate": 1.3529363007169797e-06,
      "loss": 0.6669,
      "step": 17601
    },
    {
      "epoch": 0.7294956276679514,
      "grad_norm": 0.39297598600387573,
      "learning_rate": 1.3527290811886113e-06,
      "loss": 0.6851,
      "step": 17602
    },
    {
      "epoch": 0.7295370715736251,
      "grad_norm": 0.42736852169036865,
      "learning_rate": 1.3525218616602429e-06,
      "loss": 0.7251,
      "step": 17603
    },
    {
      "epoch": 0.7295785154792988,
      "grad_norm": 0.4003267288208008,
      "learning_rate": 1.3523146421318747e-06,
      "loss": 0.6398,
      "step": 17604
    },
    {
      "epoch": 0.7296199593849725,
      "grad_norm": 0.38511863350868225,
      "learning_rate": 1.3521074226035063e-06,
      "loss": 0.6622,
      "step": 17605
    },
    {
      "epoch": 0.7296614032906461,
      "grad_norm": 0.41178011894226074,
      "learning_rate": 1.3519002030751379e-06,
      "loss": 0.6613,
      "step": 17606
    },
    {
      "epoch": 0.7297028471963197,
      "grad_norm": 0.4080874025821686,
      "learning_rate": 1.3516929835467697e-06,
      "loss": 0.6743,
      "step": 17607
    },
    {
      "epoch": 0.7297442911019935,
      "grad_norm": 0.4005504548549652,
      "learning_rate": 1.3514857640184013e-06,
      "loss": 0.6632,
      "step": 17608
    },
    {
      "epoch": 0.7297857350076671,
      "grad_norm": 0.39349642395973206,
      "learning_rate": 1.3512785444900329e-06,
      "loss": 0.6863,
      "step": 17609
    },
    {
      "epoch": 0.7298271789133408,
      "grad_norm": 0.3907824456691742,
      "learning_rate": 1.3510713249616645e-06,
      "loss": 0.6852,
      "step": 17610
    },
    {
      "epoch": 0.7298686228190144,
      "grad_norm": 0.3895411491394043,
      "learning_rate": 1.350864105433296e-06,
      "loss": 0.6487,
      "step": 17611
    },
    {
      "epoch": 0.7299100667246882,
      "grad_norm": 0.41018497943878174,
      "learning_rate": 1.3506568859049277e-06,
      "loss": 0.6935,
      "step": 17612
    },
    {
      "epoch": 0.7299515106303618,
      "grad_norm": 0.3805798888206482,
      "learning_rate": 1.3504496663765595e-06,
      "loss": 0.6595,
      "step": 17613
    },
    {
      "epoch": 0.7299929545360355,
      "grad_norm": 0.44125795364379883,
      "learning_rate": 1.350242446848191e-06,
      "loss": 0.7202,
      "step": 17614
    },
    {
      "epoch": 0.7300343984417091,
      "grad_norm": 0.4274123013019562,
      "learning_rate": 1.3500352273198227e-06,
      "loss": 0.6914,
      "step": 17615
    },
    {
      "epoch": 0.7300758423473828,
      "grad_norm": 0.4148405194282532,
      "learning_rate": 1.3498280077914545e-06,
      "loss": 0.6628,
      "step": 17616
    },
    {
      "epoch": 0.7301172862530565,
      "grad_norm": 0.43427303433418274,
      "learning_rate": 1.349620788263086e-06,
      "loss": 0.7036,
      "step": 17617
    },
    {
      "epoch": 0.7301587301587301,
      "grad_norm": 0.42457371950149536,
      "learning_rate": 1.3494135687347177e-06,
      "loss": 0.6797,
      "step": 17618
    },
    {
      "epoch": 0.7302001740644039,
      "grad_norm": 0.41605550050735474,
      "learning_rate": 1.3492063492063493e-06,
      "loss": 0.6931,
      "step": 17619
    },
    {
      "epoch": 0.7302416179700775,
      "grad_norm": 0.46486198902130127,
      "learning_rate": 1.3489991296779809e-06,
      "loss": 0.73,
      "step": 17620
    },
    {
      "epoch": 0.7302830618757512,
      "grad_norm": 0.4173734784126282,
      "learning_rate": 1.3487919101496125e-06,
      "loss": 0.6841,
      "step": 17621
    },
    {
      "epoch": 0.7303245057814248,
      "grad_norm": 0.45792585611343384,
      "learning_rate": 1.3485846906212443e-06,
      "loss": 0.7151,
      "step": 17622
    },
    {
      "epoch": 0.7303659496870986,
      "grad_norm": 0.4369369447231293,
      "learning_rate": 1.3483774710928759e-06,
      "loss": 0.6807,
      "step": 17623
    },
    {
      "epoch": 0.7304073935927722,
      "grad_norm": 0.4144294261932373,
      "learning_rate": 1.3481702515645077e-06,
      "loss": 0.6727,
      "step": 17624
    },
    {
      "epoch": 0.7304488374984458,
      "grad_norm": 0.39227136969566345,
      "learning_rate": 1.3479630320361393e-06,
      "loss": 0.6464,
      "step": 17625
    },
    {
      "epoch": 0.7304902814041195,
      "grad_norm": 0.4045243561267853,
      "learning_rate": 1.3477558125077709e-06,
      "loss": 0.661,
      "step": 17626
    },
    {
      "epoch": 0.7305317253097932,
      "grad_norm": 0.46463215351104736,
      "learning_rate": 1.3475485929794025e-06,
      "loss": 0.668,
      "step": 17627
    },
    {
      "epoch": 0.7305731692154669,
      "grad_norm": 0.3973020613193512,
      "learning_rate": 1.347341373451034e-06,
      "loss": 0.6691,
      "step": 17628
    },
    {
      "epoch": 0.7306146131211405,
      "grad_norm": 0.4016871154308319,
      "learning_rate": 1.3471341539226657e-06,
      "loss": 0.6472,
      "step": 17629
    },
    {
      "epoch": 0.7306560570268142,
      "grad_norm": 0.42157474160194397,
      "learning_rate": 1.3469269343942975e-06,
      "loss": 0.6981,
      "step": 17630
    },
    {
      "epoch": 0.7306975009324879,
      "grad_norm": 0.4324299395084381,
      "learning_rate": 1.346719714865929e-06,
      "loss": 0.6753,
      "step": 17631
    },
    {
      "epoch": 0.7307389448381616,
      "grad_norm": 0.40075892210006714,
      "learning_rate": 1.3465124953375607e-06,
      "loss": 0.7078,
      "step": 17632
    },
    {
      "epoch": 0.7307803887438352,
      "grad_norm": 0.4062173068523407,
      "learning_rate": 1.3463052758091925e-06,
      "loss": 0.6398,
      "step": 17633
    },
    {
      "epoch": 0.7308218326495088,
      "grad_norm": 0.40266022086143494,
      "learning_rate": 1.346098056280824e-06,
      "loss": 0.6924,
      "step": 17634
    },
    {
      "epoch": 0.7308632765551826,
      "grad_norm": 0.41210249066352844,
      "learning_rate": 1.3458908367524557e-06,
      "loss": 0.6958,
      "step": 17635
    },
    {
      "epoch": 0.7309047204608562,
      "grad_norm": 0.48352545499801636,
      "learning_rate": 1.3456836172240873e-06,
      "loss": 0.7185,
      "step": 17636
    },
    {
      "epoch": 0.7309461643665299,
      "grad_norm": 0.3903990089893341,
      "learning_rate": 1.3454763976957189e-06,
      "loss": 0.689,
      "step": 17637
    },
    {
      "epoch": 0.7309876082722035,
      "grad_norm": 0.4011651575565338,
      "learning_rate": 1.3452691781673505e-06,
      "loss": 0.666,
      "step": 17638
    },
    {
      "epoch": 0.7310290521778773,
      "grad_norm": 0.43048587441444397,
      "learning_rate": 1.3450619586389823e-06,
      "loss": 0.6989,
      "step": 17639
    },
    {
      "epoch": 0.7310704960835509,
      "grad_norm": 0.41844263672828674,
      "learning_rate": 1.3448547391106139e-06,
      "loss": 0.7045,
      "step": 17640
    },
    {
      "epoch": 0.7311119399892246,
      "grad_norm": 0.41407373547554016,
      "learning_rate": 1.3446475195822457e-06,
      "loss": 0.6934,
      "step": 17641
    },
    {
      "epoch": 0.7311533838948983,
      "grad_norm": 0.4160930812358856,
      "learning_rate": 1.3444403000538773e-06,
      "loss": 0.7498,
      "step": 17642
    },
    {
      "epoch": 0.7311948278005719,
      "grad_norm": 0.42814838886260986,
      "learning_rate": 1.3442330805255089e-06,
      "loss": 0.6853,
      "step": 17643
    },
    {
      "epoch": 0.7312362717062456,
      "grad_norm": 0.4032931625843048,
      "learning_rate": 1.3440258609971405e-06,
      "loss": 0.6472,
      "step": 17644
    },
    {
      "epoch": 0.7312777156119192,
      "grad_norm": 0.41552022099494934,
      "learning_rate": 1.343818641468772e-06,
      "loss": 0.606,
      "step": 17645
    },
    {
      "epoch": 0.731319159517593,
      "grad_norm": 0.4641438126564026,
      "learning_rate": 1.3436114219404037e-06,
      "loss": 0.6931,
      "step": 17646
    },
    {
      "epoch": 0.7313606034232666,
      "grad_norm": 0.40952804684638977,
      "learning_rate": 1.3434042024120353e-06,
      "loss": 0.6746,
      "step": 17647
    },
    {
      "epoch": 0.7314020473289403,
      "grad_norm": 0.38006022572517395,
      "learning_rate": 1.343196982883667e-06,
      "loss": 0.6178,
      "step": 17648
    },
    {
      "epoch": 0.7314434912346139,
      "grad_norm": 0.3994552195072174,
      "learning_rate": 1.3429897633552987e-06,
      "loss": 0.6553,
      "step": 17649
    },
    {
      "epoch": 0.7314849351402877,
      "grad_norm": 0.41635221242904663,
      "learning_rate": 1.3427825438269305e-06,
      "loss": 0.6617,
      "step": 17650
    },
    {
      "epoch": 0.7315263790459613,
      "grad_norm": 0.3968122899532318,
      "learning_rate": 1.342575324298562e-06,
      "loss": 0.6456,
      "step": 17651
    },
    {
      "epoch": 0.7315678229516349,
      "grad_norm": 0.4343999922275543,
      "learning_rate": 1.3423681047701937e-06,
      "loss": 0.709,
      "step": 17652
    },
    {
      "epoch": 0.7316092668573086,
      "grad_norm": 0.4228668808937073,
      "learning_rate": 1.3421608852418253e-06,
      "loss": 0.6821,
      "step": 17653
    },
    {
      "epoch": 0.7316507107629823,
      "grad_norm": 0.36743924021720886,
      "learning_rate": 1.3419536657134569e-06,
      "loss": 0.592,
      "step": 17654
    },
    {
      "epoch": 0.731692154668656,
      "grad_norm": 0.4461619555950165,
      "learning_rate": 1.3417464461850885e-06,
      "loss": 0.7029,
      "step": 17655
    },
    {
      "epoch": 0.7317335985743296,
      "grad_norm": 0.40960729122161865,
      "learning_rate": 1.3415392266567203e-06,
      "loss": 0.6041,
      "step": 17656
    },
    {
      "epoch": 0.7317750424800034,
      "grad_norm": 0.4142327904701233,
      "learning_rate": 1.3413320071283519e-06,
      "loss": 0.7129,
      "step": 17657
    },
    {
      "epoch": 0.731816486385677,
      "grad_norm": 0.43525567650794983,
      "learning_rate": 1.3411247875999837e-06,
      "loss": 0.6698,
      "step": 17658
    },
    {
      "epoch": 0.7318579302913507,
      "grad_norm": 0.4089859426021576,
      "learning_rate": 1.3409175680716153e-06,
      "loss": 0.6976,
      "step": 17659
    },
    {
      "epoch": 0.7318993741970243,
      "grad_norm": 0.39958620071411133,
      "learning_rate": 1.340710348543247e-06,
      "loss": 0.6334,
      "step": 17660
    },
    {
      "epoch": 0.731940818102698,
      "grad_norm": 0.43112292885780334,
      "learning_rate": 1.3405031290148785e-06,
      "loss": 0.7085,
      "step": 17661
    },
    {
      "epoch": 0.7319822620083717,
      "grad_norm": 0.4369018077850342,
      "learning_rate": 1.34029590948651e-06,
      "loss": 0.7159,
      "step": 17662
    },
    {
      "epoch": 0.7320237059140453,
      "grad_norm": 0.40243029594421387,
      "learning_rate": 1.3400886899581417e-06,
      "loss": 0.6298,
      "step": 17663
    },
    {
      "epoch": 0.732065149819719,
      "grad_norm": 0.3876744210720062,
      "learning_rate": 1.3398814704297733e-06,
      "loss": 0.6379,
      "step": 17664
    },
    {
      "epoch": 0.7321065937253927,
      "grad_norm": 0.4240640103816986,
      "learning_rate": 1.339674250901405e-06,
      "loss": 0.6869,
      "step": 17665
    },
    {
      "epoch": 0.7321480376310664,
      "grad_norm": 0.3975006639957428,
      "learning_rate": 1.339467031373037e-06,
      "loss": 0.6504,
      "step": 17666
    },
    {
      "epoch": 0.73218948153674,
      "grad_norm": 0.41036680340766907,
      "learning_rate": 1.3392598118446685e-06,
      "loss": 0.6185,
      "step": 17667
    },
    {
      "epoch": 0.7322309254424136,
      "grad_norm": 0.4121469557285309,
      "learning_rate": 1.3390525923163e-06,
      "loss": 0.6896,
      "step": 17668
    },
    {
      "epoch": 0.7322723693480874,
      "grad_norm": 0.4292762875556946,
      "learning_rate": 1.3388453727879317e-06,
      "loss": 0.66,
      "step": 17669
    },
    {
      "epoch": 0.732313813253761,
      "grad_norm": 0.4432283937931061,
      "learning_rate": 1.3386381532595633e-06,
      "loss": 0.7134,
      "step": 17670
    },
    {
      "epoch": 0.7323552571594347,
      "grad_norm": 0.4057601988315582,
      "learning_rate": 1.3384309337311949e-06,
      "loss": 0.635,
      "step": 17671
    },
    {
      "epoch": 0.7323967010651083,
      "grad_norm": 0.41985541582107544,
      "learning_rate": 1.3382237142028265e-06,
      "loss": 0.6897,
      "step": 17672
    },
    {
      "epoch": 0.7324381449707821,
      "grad_norm": 0.41044801473617554,
      "learning_rate": 1.338016494674458e-06,
      "loss": 0.666,
      "step": 17673
    },
    {
      "epoch": 0.7324795888764557,
      "grad_norm": 0.4280877113342285,
      "learning_rate": 1.3378092751460899e-06,
      "loss": 0.6909,
      "step": 17674
    },
    {
      "epoch": 0.7325210327821294,
      "grad_norm": 0.4212762117385864,
      "learning_rate": 1.3376020556177217e-06,
      "loss": 0.6967,
      "step": 17675
    },
    {
      "epoch": 0.732562476687803,
      "grad_norm": 0.4144017696380615,
      "learning_rate": 1.3373948360893533e-06,
      "loss": 0.6558,
      "step": 17676
    },
    {
      "epoch": 0.7326039205934767,
      "grad_norm": 0.4300427734851837,
      "learning_rate": 1.337187616560985e-06,
      "loss": 0.7228,
      "step": 17677
    },
    {
      "epoch": 0.7326453644991504,
      "grad_norm": 0.4473978281021118,
      "learning_rate": 1.3369803970326165e-06,
      "loss": 0.6409,
      "step": 17678
    },
    {
      "epoch": 0.732686808404824,
      "grad_norm": 0.41310301423072815,
      "learning_rate": 1.336773177504248e-06,
      "loss": 0.6411,
      "step": 17679
    },
    {
      "epoch": 0.7327282523104978,
      "grad_norm": 0.42003491520881653,
      "learning_rate": 1.3365659579758797e-06,
      "loss": 0.7157,
      "step": 17680
    },
    {
      "epoch": 0.7327696962161714,
      "grad_norm": 0.42352771759033203,
      "learning_rate": 1.3363587384475113e-06,
      "loss": 0.6516,
      "step": 17681
    },
    {
      "epoch": 0.7328111401218451,
      "grad_norm": 0.41207677125930786,
      "learning_rate": 1.336151518919143e-06,
      "loss": 0.634,
      "step": 17682
    },
    {
      "epoch": 0.7328525840275187,
      "grad_norm": 0.44245922565460205,
      "learning_rate": 1.335944299390775e-06,
      "loss": 0.749,
      "step": 17683
    },
    {
      "epoch": 0.7328940279331925,
      "grad_norm": 0.40678828954696655,
      "learning_rate": 1.3357370798624065e-06,
      "loss": 0.689,
      "step": 17684
    },
    {
      "epoch": 0.7329354718388661,
      "grad_norm": 0.40013769268989563,
      "learning_rate": 1.335529860334038e-06,
      "loss": 0.6642,
      "step": 17685
    },
    {
      "epoch": 0.7329769157445397,
      "grad_norm": 0.42995795607566833,
      "learning_rate": 1.3353226408056697e-06,
      "loss": 0.665,
      "step": 17686
    },
    {
      "epoch": 0.7330183596502134,
      "grad_norm": 0.45039281249046326,
      "learning_rate": 1.3351154212773013e-06,
      "loss": 0.6732,
      "step": 17687
    },
    {
      "epoch": 0.7330598035558871,
      "grad_norm": 0.43420520424842834,
      "learning_rate": 1.3349082017489329e-06,
      "loss": 0.6678,
      "step": 17688
    },
    {
      "epoch": 0.7331012474615608,
      "grad_norm": 0.4256800711154938,
      "learning_rate": 1.3347009822205645e-06,
      "loss": 0.6322,
      "step": 17689
    },
    {
      "epoch": 0.7331426913672344,
      "grad_norm": 0.39028656482696533,
      "learning_rate": 1.334493762692196e-06,
      "loss": 0.6454,
      "step": 17690
    },
    {
      "epoch": 0.7331841352729082,
      "grad_norm": 0.4413905739784241,
      "learning_rate": 1.334286543163828e-06,
      "loss": 0.6537,
      "step": 17691
    },
    {
      "epoch": 0.7332255791785818,
      "grad_norm": 0.4470168650150299,
      "learning_rate": 1.3340793236354597e-06,
      "loss": 0.689,
      "step": 17692
    },
    {
      "epoch": 0.7332670230842555,
      "grad_norm": 0.43094855546951294,
      "learning_rate": 1.3338721041070913e-06,
      "loss": 0.6716,
      "step": 17693
    },
    {
      "epoch": 0.7333084669899291,
      "grad_norm": 0.4629989564418793,
      "learning_rate": 1.333664884578723e-06,
      "loss": 0.7798,
      "step": 17694
    },
    {
      "epoch": 0.7333499108956028,
      "grad_norm": 0.43301618099212646,
      "learning_rate": 1.3334576650503545e-06,
      "loss": 0.6951,
      "step": 17695
    },
    {
      "epoch": 0.7333913548012765,
      "grad_norm": 0.4382196068763733,
      "learning_rate": 1.333250445521986e-06,
      "loss": 0.6774,
      "step": 17696
    },
    {
      "epoch": 0.7334327987069501,
      "grad_norm": 0.4087275266647339,
      "learning_rate": 1.3330432259936177e-06,
      "loss": 0.6686,
      "step": 17697
    },
    {
      "epoch": 0.7334742426126238,
      "grad_norm": 0.3955472707748413,
      "learning_rate": 1.3328360064652493e-06,
      "loss": 0.5831,
      "step": 17698
    },
    {
      "epoch": 0.7335156865182975,
      "grad_norm": 0.37742936611175537,
      "learning_rate": 1.3326287869368809e-06,
      "loss": 0.6017,
      "step": 17699
    },
    {
      "epoch": 0.7335571304239712,
      "grad_norm": 0.42822912335395813,
      "learning_rate": 1.332421567408513e-06,
      "loss": 0.6592,
      "step": 17700
    },
    {
      "epoch": 0.7335985743296448,
      "grad_norm": 0.421403169631958,
      "learning_rate": 1.3322143478801445e-06,
      "loss": 0.6464,
      "step": 17701
    },
    {
      "epoch": 0.7336400182353185,
      "grad_norm": 0.42078259587287903,
      "learning_rate": 1.332007128351776e-06,
      "loss": 0.6348,
      "step": 17702
    },
    {
      "epoch": 0.7336814621409922,
      "grad_norm": 0.4165763258934021,
      "learning_rate": 1.3317999088234077e-06,
      "loss": 0.6729,
      "step": 17703
    },
    {
      "epoch": 0.7337229060466658,
      "grad_norm": 0.4066922664642334,
      "learning_rate": 1.3315926892950393e-06,
      "loss": 0.6567,
      "step": 17704
    },
    {
      "epoch": 0.7337643499523395,
      "grad_norm": 0.42212092876434326,
      "learning_rate": 1.331385469766671e-06,
      "loss": 0.6749,
      "step": 17705
    },
    {
      "epoch": 0.7338057938580131,
      "grad_norm": 0.4051072597503662,
      "learning_rate": 1.3311782502383025e-06,
      "loss": 0.6683,
      "step": 17706
    },
    {
      "epoch": 0.7338472377636869,
      "grad_norm": 0.4200262725353241,
      "learning_rate": 1.330971030709934e-06,
      "loss": 0.7274,
      "step": 17707
    },
    {
      "epoch": 0.7338886816693605,
      "grad_norm": 0.45273247361183167,
      "learning_rate": 1.3307638111815657e-06,
      "loss": 0.6904,
      "step": 17708
    },
    {
      "epoch": 0.7339301255750342,
      "grad_norm": 0.4229804575443268,
      "learning_rate": 1.3305565916531977e-06,
      "loss": 0.6708,
      "step": 17709
    },
    {
      "epoch": 0.7339715694807079,
      "grad_norm": 0.4317142963409424,
      "learning_rate": 1.3303493721248293e-06,
      "loss": 0.7004,
      "step": 17710
    },
    {
      "epoch": 0.7340130133863816,
      "grad_norm": 0.43891704082489014,
      "learning_rate": 1.330142152596461e-06,
      "loss": 0.6663,
      "step": 17711
    },
    {
      "epoch": 0.7340544572920552,
      "grad_norm": 0.4107625484466553,
      "learning_rate": 1.3299349330680925e-06,
      "loss": 0.693,
      "step": 17712
    },
    {
      "epoch": 0.7340959011977288,
      "grad_norm": 0.4235994517803192,
      "learning_rate": 1.329727713539724e-06,
      "loss": 0.6418,
      "step": 17713
    },
    {
      "epoch": 0.7341373451034026,
      "grad_norm": 0.418617844581604,
      "learning_rate": 1.3295204940113557e-06,
      "loss": 0.6763,
      "step": 17714
    },
    {
      "epoch": 0.7341787890090762,
      "grad_norm": 0.41659700870513916,
      "learning_rate": 1.3293132744829873e-06,
      "loss": 0.6938,
      "step": 17715
    },
    {
      "epoch": 0.7342202329147499,
      "grad_norm": 0.42446398735046387,
      "learning_rate": 1.3291060549546189e-06,
      "loss": 0.6687,
      "step": 17716
    },
    {
      "epoch": 0.7342616768204235,
      "grad_norm": 0.3918251395225525,
      "learning_rate": 1.328898835426251e-06,
      "loss": 0.7107,
      "step": 17717
    },
    {
      "epoch": 0.7343031207260973,
      "grad_norm": 0.40055739879608154,
      "learning_rate": 1.3286916158978825e-06,
      "loss": 0.6331,
      "step": 17718
    },
    {
      "epoch": 0.7343445646317709,
      "grad_norm": 0.39058953523635864,
      "learning_rate": 1.3284843963695141e-06,
      "loss": 0.6235,
      "step": 17719
    },
    {
      "epoch": 0.7343860085374446,
      "grad_norm": 0.425293892621994,
      "learning_rate": 1.3282771768411457e-06,
      "loss": 0.6321,
      "step": 17720
    },
    {
      "epoch": 0.7344274524431182,
      "grad_norm": 0.4224386513233185,
      "learning_rate": 1.3280699573127773e-06,
      "loss": 0.6627,
      "step": 17721
    },
    {
      "epoch": 0.7344688963487919,
      "grad_norm": 0.4451560378074646,
      "learning_rate": 1.327862737784409e-06,
      "loss": 0.6941,
      "step": 17722
    },
    {
      "epoch": 0.7345103402544656,
      "grad_norm": 0.41918396949768066,
      "learning_rate": 1.3276555182560405e-06,
      "loss": 0.6611,
      "step": 17723
    },
    {
      "epoch": 0.7345517841601392,
      "grad_norm": 0.4141343832015991,
      "learning_rate": 1.327448298727672e-06,
      "loss": 0.6503,
      "step": 17724
    },
    {
      "epoch": 0.734593228065813,
      "grad_norm": 0.4357835054397583,
      "learning_rate": 1.3272410791993037e-06,
      "loss": 0.7053,
      "step": 17725
    },
    {
      "epoch": 0.7346346719714866,
      "grad_norm": 0.40157321095466614,
      "learning_rate": 1.3270338596709357e-06,
      "loss": 0.616,
      "step": 17726
    },
    {
      "epoch": 0.7346761158771603,
      "grad_norm": 0.4196839928627014,
      "learning_rate": 1.3268266401425673e-06,
      "loss": 0.6702,
      "step": 17727
    },
    {
      "epoch": 0.7347175597828339,
      "grad_norm": 0.4422299861907959,
      "learning_rate": 1.326619420614199e-06,
      "loss": 0.709,
      "step": 17728
    },
    {
      "epoch": 0.7347590036885076,
      "grad_norm": 0.376321405172348,
      "learning_rate": 1.3264122010858305e-06,
      "loss": 0.6165,
      "step": 17729
    },
    {
      "epoch": 0.7348004475941813,
      "grad_norm": 0.39355507493019104,
      "learning_rate": 1.326204981557462e-06,
      "loss": 0.6024,
      "step": 17730
    },
    {
      "epoch": 0.7348418914998549,
      "grad_norm": 0.43227970600128174,
      "learning_rate": 1.3259977620290937e-06,
      "loss": 0.6986,
      "step": 17731
    },
    {
      "epoch": 0.7348833354055286,
      "grad_norm": 0.424333781003952,
      "learning_rate": 1.3257905425007253e-06,
      "loss": 0.7397,
      "step": 17732
    },
    {
      "epoch": 0.7349247793112023,
      "grad_norm": 0.46666836738586426,
      "learning_rate": 1.3255833229723569e-06,
      "loss": 0.6829,
      "step": 17733
    },
    {
      "epoch": 0.734966223216876,
      "grad_norm": 0.39106830954551697,
      "learning_rate": 1.3253761034439885e-06,
      "loss": 0.6382,
      "step": 17734
    },
    {
      "epoch": 0.7350076671225496,
      "grad_norm": 0.42597174644470215,
      "learning_rate": 1.3251688839156205e-06,
      "loss": 0.6682,
      "step": 17735
    },
    {
      "epoch": 0.7350491110282233,
      "grad_norm": 0.3749087154865265,
      "learning_rate": 1.3249616643872521e-06,
      "loss": 0.6415,
      "step": 17736
    },
    {
      "epoch": 0.735090554933897,
      "grad_norm": 0.5266129970550537,
      "learning_rate": 1.3247544448588837e-06,
      "loss": 0.7036,
      "step": 17737
    },
    {
      "epoch": 0.7351319988395706,
      "grad_norm": 0.43422597646713257,
      "learning_rate": 1.3245472253305153e-06,
      "loss": 0.6479,
      "step": 17738
    },
    {
      "epoch": 0.7351734427452443,
      "grad_norm": 0.41420039534568787,
      "learning_rate": 1.324340005802147e-06,
      "loss": 0.6653,
      "step": 17739
    },
    {
      "epoch": 0.7352148866509179,
      "grad_norm": 0.3859080970287323,
      "learning_rate": 1.3241327862737785e-06,
      "loss": 0.6262,
      "step": 17740
    },
    {
      "epoch": 0.7352563305565917,
      "grad_norm": 0.43890058994293213,
      "learning_rate": 1.32392556674541e-06,
      "loss": 0.6637,
      "step": 17741
    },
    {
      "epoch": 0.7352977744622653,
      "grad_norm": 0.43299928307533264,
      "learning_rate": 1.3237183472170417e-06,
      "loss": 0.6539,
      "step": 17742
    },
    {
      "epoch": 0.735339218367939,
      "grad_norm": 0.394838809967041,
      "learning_rate": 1.3235111276886737e-06,
      "loss": 0.6952,
      "step": 17743
    },
    {
      "epoch": 0.7353806622736127,
      "grad_norm": 0.40062767267227173,
      "learning_rate": 1.3233039081603053e-06,
      "loss": 0.663,
      "step": 17744
    },
    {
      "epoch": 0.7354221061792864,
      "grad_norm": 0.4012412130832672,
      "learning_rate": 1.323096688631937e-06,
      "loss": 0.6458,
      "step": 17745
    },
    {
      "epoch": 0.73546355008496,
      "grad_norm": 0.40164998173713684,
      "learning_rate": 1.3228894691035685e-06,
      "loss": 0.661,
      "step": 17746
    },
    {
      "epoch": 0.7355049939906336,
      "grad_norm": 0.40151891112327576,
      "learning_rate": 1.3226822495752e-06,
      "loss": 0.6573,
      "step": 17747
    },
    {
      "epoch": 0.7355464378963074,
      "grad_norm": 0.40262463688850403,
      "learning_rate": 1.3224750300468317e-06,
      "loss": 0.6587,
      "step": 17748
    },
    {
      "epoch": 0.735587881801981,
      "grad_norm": 0.44928982853889465,
      "learning_rate": 1.3222678105184633e-06,
      "loss": 0.6361,
      "step": 17749
    },
    {
      "epoch": 0.7356293257076547,
      "grad_norm": 0.4095138907432556,
      "learning_rate": 1.322060590990095e-06,
      "loss": 0.6519,
      "step": 17750
    },
    {
      "epoch": 0.7356707696133283,
      "grad_norm": 0.42883414030075073,
      "learning_rate": 1.3218533714617265e-06,
      "loss": 0.6818,
      "step": 17751
    },
    {
      "epoch": 0.7357122135190021,
      "grad_norm": 0.3955925703048706,
      "learning_rate": 1.3216461519333585e-06,
      "loss": 0.6384,
      "step": 17752
    },
    {
      "epoch": 0.7357536574246757,
      "grad_norm": 0.41250914335250854,
      "learning_rate": 1.3214389324049901e-06,
      "loss": 0.6998,
      "step": 17753
    },
    {
      "epoch": 0.7357951013303494,
      "grad_norm": 0.4283217191696167,
      "learning_rate": 1.3212317128766217e-06,
      "loss": 0.7072,
      "step": 17754
    },
    {
      "epoch": 0.735836545236023,
      "grad_norm": 0.4297679364681244,
      "learning_rate": 1.3210244933482533e-06,
      "loss": 0.6514,
      "step": 17755
    },
    {
      "epoch": 0.7358779891416967,
      "grad_norm": 0.4424307644367218,
      "learning_rate": 1.320817273819885e-06,
      "loss": 0.6594,
      "step": 17756
    },
    {
      "epoch": 0.7359194330473704,
      "grad_norm": 0.43793049454689026,
      "learning_rate": 1.3206100542915165e-06,
      "loss": 0.6619,
      "step": 17757
    },
    {
      "epoch": 0.735960876953044,
      "grad_norm": 0.40330901741981506,
      "learning_rate": 1.320402834763148e-06,
      "loss": 0.6523,
      "step": 17758
    },
    {
      "epoch": 0.7360023208587178,
      "grad_norm": 0.4621729552745819,
      "learning_rate": 1.3201956152347797e-06,
      "loss": 0.7006,
      "step": 17759
    },
    {
      "epoch": 0.7360437647643914,
      "grad_norm": 0.4140281081199646,
      "learning_rate": 1.3199883957064113e-06,
      "loss": 0.6938,
      "step": 17760
    },
    {
      "epoch": 0.7360852086700651,
      "grad_norm": 0.3956485390663147,
      "learning_rate": 1.3197811761780433e-06,
      "loss": 0.6387,
      "step": 17761
    },
    {
      "epoch": 0.7361266525757387,
      "grad_norm": 0.44890764355659485,
      "learning_rate": 1.319573956649675e-06,
      "loss": 0.7249,
      "step": 17762
    },
    {
      "epoch": 0.7361680964814125,
      "grad_norm": 0.4246966242790222,
      "learning_rate": 1.3193667371213065e-06,
      "loss": 0.6782,
      "step": 17763
    },
    {
      "epoch": 0.7362095403870861,
      "grad_norm": 0.4175665080547333,
      "learning_rate": 1.3191595175929381e-06,
      "loss": 0.6437,
      "step": 17764
    },
    {
      "epoch": 0.7362509842927597,
      "grad_norm": 0.4314151406288147,
      "learning_rate": 1.3189522980645697e-06,
      "loss": 0.7461,
      "step": 17765
    },
    {
      "epoch": 0.7362924281984334,
      "grad_norm": 0.47177544236183167,
      "learning_rate": 1.3187450785362013e-06,
      "loss": 0.6191,
      "step": 17766
    },
    {
      "epoch": 0.7363338721041071,
      "grad_norm": 0.3895148038864136,
      "learning_rate": 1.318537859007833e-06,
      "loss": 0.6273,
      "step": 17767
    },
    {
      "epoch": 0.7363753160097808,
      "grad_norm": 0.38011813163757324,
      "learning_rate": 1.3183306394794645e-06,
      "loss": 0.6434,
      "step": 17768
    },
    {
      "epoch": 0.7364167599154544,
      "grad_norm": 0.44096285104751587,
      "learning_rate": 1.318123419951096e-06,
      "loss": 0.7034,
      "step": 17769
    },
    {
      "epoch": 0.7364582038211281,
      "grad_norm": 0.40571075677871704,
      "learning_rate": 1.3179162004227281e-06,
      "loss": 0.6322,
      "step": 17770
    },
    {
      "epoch": 0.7364996477268018,
      "grad_norm": 0.43218541145324707,
      "learning_rate": 1.3177089808943597e-06,
      "loss": 0.7397,
      "step": 17771
    },
    {
      "epoch": 0.7365410916324755,
      "grad_norm": 0.4716487526893616,
      "learning_rate": 1.3175017613659913e-06,
      "loss": 0.6415,
      "step": 17772
    },
    {
      "epoch": 0.7365825355381491,
      "grad_norm": 0.3943225145339966,
      "learning_rate": 1.317294541837623e-06,
      "loss": 0.6624,
      "step": 17773
    },
    {
      "epoch": 0.7366239794438227,
      "grad_norm": 0.3842735290527344,
      "learning_rate": 1.3170873223092545e-06,
      "loss": 0.6616,
      "step": 17774
    },
    {
      "epoch": 0.7366654233494965,
      "grad_norm": 0.4227282404899597,
      "learning_rate": 1.316880102780886e-06,
      "loss": 0.6594,
      "step": 17775
    },
    {
      "epoch": 0.7367068672551701,
      "grad_norm": 0.45664024353027344,
      "learning_rate": 1.3166728832525177e-06,
      "loss": 0.6428,
      "step": 17776
    },
    {
      "epoch": 0.7367483111608438,
      "grad_norm": 0.3969261050224304,
      "learning_rate": 1.3164656637241493e-06,
      "loss": 0.665,
      "step": 17777
    },
    {
      "epoch": 0.7367897550665174,
      "grad_norm": 0.4578079879283905,
      "learning_rate": 1.3162584441957813e-06,
      "loss": 0.6499,
      "step": 17778
    },
    {
      "epoch": 0.7368311989721912,
      "grad_norm": 0.38991278409957886,
      "learning_rate": 1.316051224667413e-06,
      "loss": 0.6611,
      "step": 17779
    },
    {
      "epoch": 0.7368726428778648,
      "grad_norm": 0.41884148120880127,
      "learning_rate": 1.3158440051390445e-06,
      "loss": 0.6992,
      "step": 17780
    },
    {
      "epoch": 0.7369140867835385,
      "grad_norm": 0.3985823690891266,
      "learning_rate": 1.3156367856106761e-06,
      "loss": 0.641,
      "step": 17781
    },
    {
      "epoch": 0.7369555306892122,
      "grad_norm": 0.47274893522262573,
      "learning_rate": 1.3154295660823077e-06,
      "loss": 0.7241,
      "step": 17782
    },
    {
      "epoch": 0.7369969745948858,
      "grad_norm": 0.4704985022544861,
      "learning_rate": 1.3152223465539393e-06,
      "loss": 0.6985,
      "step": 17783
    },
    {
      "epoch": 0.7370384185005595,
      "grad_norm": 0.43171727657318115,
      "learning_rate": 1.315015127025571e-06,
      "loss": 0.6492,
      "step": 17784
    },
    {
      "epoch": 0.7370798624062331,
      "grad_norm": 0.4260539710521698,
      "learning_rate": 1.3148079074972025e-06,
      "loss": 0.6875,
      "step": 17785
    },
    {
      "epoch": 0.7371213063119069,
      "grad_norm": 0.4236333668231964,
      "learning_rate": 1.314600687968834e-06,
      "loss": 0.6624,
      "step": 17786
    },
    {
      "epoch": 0.7371627502175805,
      "grad_norm": 0.4468648135662079,
      "learning_rate": 1.3143934684404661e-06,
      "loss": 0.6665,
      "step": 17787
    },
    {
      "epoch": 0.7372041941232542,
      "grad_norm": 0.3918130397796631,
      "learning_rate": 1.3141862489120977e-06,
      "loss": 0.6626,
      "step": 17788
    },
    {
      "epoch": 0.7372456380289278,
      "grad_norm": 0.4088038504123688,
      "learning_rate": 1.3139790293837293e-06,
      "loss": 0.6394,
      "step": 17789
    },
    {
      "epoch": 0.7372870819346015,
      "grad_norm": 0.3960227370262146,
      "learning_rate": 1.313771809855361e-06,
      "loss": 0.6317,
      "step": 17790
    },
    {
      "epoch": 0.7373285258402752,
      "grad_norm": 0.4270232319831848,
      "learning_rate": 1.3135645903269925e-06,
      "loss": 0.6884,
      "step": 17791
    },
    {
      "epoch": 0.7373699697459488,
      "grad_norm": 0.41873347759246826,
      "learning_rate": 1.313357370798624e-06,
      "loss": 0.6772,
      "step": 17792
    },
    {
      "epoch": 0.7374114136516225,
      "grad_norm": 0.4039892256259918,
      "learning_rate": 1.3131501512702557e-06,
      "loss": 0.6594,
      "step": 17793
    },
    {
      "epoch": 0.7374528575572962,
      "grad_norm": 0.41404640674591064,
      "learning_rate": 1.3129429317418873e-06,
      "loss": 0.6627,
      "step": 17794
    },
    {
      "epoch": 0.7374943014629699,
      "grad_norm": 0.4048686921596527,
      "learning_rate": 1.312735712213519e-06,
      "loss": 0.6337,
      "step": 17795
    },
    {
      "epoch": 0.7375357453686435,
      "grad_norm": 0.43747371435165405,
      "learning_rate": 1.312528492685151e-06,
      "loss": 0.6777,
      "step": 17796
    },
    {
      "epoch": 0.7375771892743173,
      "grad_norm": 0.37258535623550415,
      "learning_rate": 1.3123212731567825e-06,
      "loss": 0.6621,
      "step": 17797
    },
    {
      "epoch": 0.7376186331799909,
      "grad_norm": 0.42875465750694275,
      "learning_rate": 1.3121140536284141e-06,
      "loss": 0.6936,
      "step": 17798
    },
    {
      "epoch": 0.7376600770856645,
      "grad_norm": 0.45931416749954224,
      "learning_rate": 1.3119068341000457e-06,
      "loss": 0.7268,
      "step": 17799
    },
    {
      "epoch": 0.7377015209913382,
      "grad_norm": 0.43424221873283386,
      "learning_rate": 1.3116996145716773e-06,
      "loss": 0.6775,
      "step": 17800
    },
    {
      "epoch": 0.7377429648970119,
      "grad_norm": 0.430513471364975,
      "learning_rate": 1.311492395043309e-06,
      "loss": 0.66,
      "step": 17801
    },
    {
      "epoch": 0.7377844088026856,
      "grad_norm": 0.3857762813568115,
      "learning_rate": 1.3112851755149405e-06,
      "loss": 0.6638,
      "step": 17802
    },
    {
      "epoch": 0.7378258527083592,
      "grad_norm": 0.42836034297943115,
      "learning_rate": 1.311077955986572e-06,
      "loss": 0.65,
      "step": 17803
    },
    {
      "epoch": 0.7378672966140329,
      "grad_norm": 0.403128057718277,
      "learning_rate": 1.3108707364582041e-06,
      "loss": 0.7083,
      "step": 17804
    },
    {
      "epoch": 0.7379087405197066,
      "grad_norm": 0.41026806831359863,
      "learning_rate": 1.3106635169298357e-06,
      "loss": 0.6323,
      "step": 17805
    },
    {
      "epoch": 0.7379501844253803,
      "grad_norm": 0.43012359738349915,
      "learning_rate": 1.3104562974014673e-06,
      "loss": 0.6992,
      "step": 17806
    },
    {
      "epoch": 0.7379916283310539,
      "grad_norm": 0.3895985782146454,
      "learning_rate": 1.310249077873099e-06,
      "loss": 0.6317,
      "step": 17807
    },
    {
      "epoch": 0.7380330722367275,
      "grad_norm": 0.42365992069244385,
      "learning_rate": 1.3100418583447305e-06,
      "loss": 0.6577,
      "step": 17808
    },
    {
      "epoch": 0.7380745161424013,
      "grad_norm": 0.4378974735736847,
      "learning_rate": 1.3098346388163621e-06,
      "loss": 0.6804,
      "step": 17809
    },
    {
      "epoch": 0.7381159600480749,
      "grad_norm": 0.4388437569141388,
      "learning_rate": 1.3096274192879937e-06,
      "loss": 0.7006,
      "step": 17810
    },
    {
      "epoch": 0.7381574039537486,
      "grad_norm": 0.4295605421066284,
      "learning_rate": 1.3094201997596253e-06,
      "loss": 0.7024,
      "step": 17811
    },
    {
      "epoch": 0.7381988478594222,
      "grad_norm": 0.44909030199050903,
      "learning_rate": 1.309212980231257e-06,
      "loss": 0.6926,
      "step": 17812
    },
    {
      "epoch": 0.738240291765096,
      "grad_norm": 0.4091121554374695,
      "learning_rate": 1.309005760702889e-06,
      "loss": 0.6702,
      "step": 17813
    },
    {
      "epoch": 0.7382817356707696,
      "grad_norm": 0.4011324346065521,
      "learning_rate": 1.3087985411745205e-06,
      "loss": 0.6848,
      "step": 17814
    },
    {
      "epoch": 0.7383231795764433,
      "grad_norm": 0.4642154276371002,
      "learning_rate": 1.3085913216461521e-06,
      "loss": 0.656,
      "step": 17815
    },
    {
      "epoch": 0.738364623482117,
      "grad_norm": 0.4398166537284851,
      "learning_rate": 1.3083841021177837e-06,
      "loss": 0.6774,
      "step": 17816
    },
    {
      "epoch": 0.7384060673877906,
      "grad_norm": 0.4138544499874115,
      "learning_rate": 1.3081768825894153e-06,
      "loss": 0.6895,
      "step": 17817
    },
    {
      "epoch": 0.7384475112934643,
      "grad_norm": 0.4339129626750946,
      "learning_rate": 1.307969663061047e-06,
      "loss": 0.688,
      "step": 17818
    },
    {
      "epoch": 0.7384889551991379,
      "grad_norm": 0.39784836769104004,
      "learning_rate": 1.3077624435326785e-06,
      "loss": 0.6719,
      "step": 17819
    },
    {
      "epoch": 0.7385303991048117,
      "grad_norm": 0.41233399510383606,
      "learning_rate": 1.30755522400431e-06,
      "loss": 0.6079,
      "step": 17820
    },
    {
      "epoch": 0.7385718430104853,
      "grad_norm": 0.45124754309654236,
      "learning_rate": 1.3073480044759417e-06,
      "loss": 0.7363,
      "step": 17821
    },
    {
      "epoch": 0.738613286916159,
      "grad_norm": 0.39800024032592773,
      "learning_rate": 1.3071407849475737e-06,
      "loss": 0.6488,
      "step": 17822
    },
    {
      "epoch": 0.7386547308218326,
      "grad_norm": 0.41092902421951294,
      "learning_rate": 1.3069335654192053e-06,
      "loss": 0.6541,
      "step": 17823
    },
    {
      "epoch": 0.7386961747275064,
      "grad_norm": 0.39009666442871094,
      "learning_rate": 1.306726345890837e-06,
      "loss": 0.6707,
      "step": 17824
    },
    {
      "epoch": 0.73873761863318,
      "grad_norm": 0.3923057019710541,
      "learning_rate": 1.3065191263624685e-06,
      "loss": 0.6541,
      "step": 17825
    },
    {
      "epoch": 0.7387790625388536,
      "grad_norm": 0.3749821186065674,
      "learning_rate": 1.3063119068341001e-06,
      "loss": 0.6619,
      "step": 17826
    },
    {
      "epoch": 0.7388205064445273,
      "grad_norm": 0.4053066074848175,
      "learning_rate": 1.3061046873057317e-06,
      "loss": 0.6917,
      "step": 17827
    },
    {
      "epoch": 0.738861950350201,
      "grad_norm": 0.4341716766357422,
      "learning_rate": 1.3058974677773633e-06,
      "loss": 0.6584,
      "step": 17828
    },
    {
      "epoch": 0.7389033942558747,
      "grad_norm": 0.36815115809440613,
      "learning_rate": 1.305690248248995e-06,
      "loss": 0.6223,
      "step": 17829
    },
    {
      "epoch": 0.7389448381615483,
      "grad_norm": 0.3982944190502167,
      "learning_rate": 1.305483028720627e-06,
      "loss": 0.6804,
      "step": 17830
    },
    {
      "epoch": 0.7389862820672221,
      "grad_norm": 0.4061671197414398,
      "learning_rate": 1.3052758091922585e-06,
      "loss": 0.6686,
      "step": 17831
    },
    {
      "epoch": 0.7390277259728957,
      "grad_norm": 0.4052245020866394,
      "learning_rate": 1.3050685896638901e-06,
      "loss": 0.6683,
      "step": 17832
    },
    {
      "epoch": 0.7390691698785694,
      "grad_norm": 0.39896178245544434,
      "learning_rate": 1.3048613701355217e-06,
      "loss": 0.6899,
      "step": 17833
    },
    {
      "epoch": 0.739110613784243,
      "grad_norm": 0.4067554771900177,
      "learning_rate": 1.3046541506071533e-06,
      "loss": 0.6266,
      "step": 17834
    },
    {
      "epoch": 0.7391520576899167,
      "grad_norm": 0.44535553455352783,
      "learning_rate": 1.304446931078785e-06,
      "loss": 0.7183,
      "step": 17835
    },
    {
      "epoch": 0.7391935015955904,
      "grad_norm": 0.39614152908325195,
      "learning_rate": 1.3042397115504165e-06,
      "loss": 0.6327,
      "step": 17836
    },
    {
      "epoch": 0.739234945501264,
      "grad_norm": 0.4104803800582886,
      "learning_rate": 1.3040324920220481e-06,
      "loss": 0.6754,
      "step": 17837
    },
    {
      "epoch": 0.7392763894069377,
      "grad_norm": 0.44602707028388977,
      "learning_rate": 1.3038252724936797e-06,
      "loss": 0.6852,
      "step": 17838
    },
    {
      "epoch": 0.7393178333126114,
      "grad_norm": 0.4120709300041199,
      "learning_rate": 1.3036180529653117e-06,
      "loss": 0.6488,
      "step": 17839
    },
    {
      "epoch": 0.7393592772182851,
      "grad_norm": 0.40208005905151367,
      "learning_rate": 1.3034108334369433e-06,
      "loss": 0.6582,
      "step": 17840
    },
    {
      "epoch": 0.7394007211239587,
      "grad_norm": 0.3729678690433502,
      "learning_rate": 1.303203613908575e-06,
      "loss": 0.6875,
      "step": 17841
    },
    {
      "epoch": 0.7394421650296324,
      "grad_norm": 0.3816264867782593,
      "learning_rate": 1.3029963943802065e-06,
      "loss": 0.667,
      "step": 17842
    },
    {
      "epoch": 0.7394836089353061,
      "grad_norm": 0.4855034649372101,
      "learning_rate": 1.3027891748518381e-06,
      "loss": 0.7576,
      "step": 17843
    },
    {
      "epoch": 0.7395250528409797,
      "grad_norm": 0.4666666090488434,
      "learning_rate": 1.3025819553234697e-06,
      "loss": 0.7017,
      "step": 17844
    },
    {
      "epoch": 0.7395664967466534,
      "grad_norm": 0.4072892367839813,
      "learning_rate": 1.3023747357951013e-06,
      "loss": 0.6514,
      "step": 17845
    },
    {
      "epoch": 0.739607940652327,
      "grad_norm": 0.464616060256958,
      "learning_rate": 1.302167516266733e-06,
      "loss": 0.6953,
      "step": 17846
    },
    {
      "epoch": 0.7396493845580008,
      "grad_norm": 0.4091990888118744,
      "learning_rate": 1.3019602967383645e-06,
      "loss": 0.6558,
      "step": 17847
    },
    {
      "epoch": 0.7396908284636744,
      "grad_norm": 0.4112918972969055,
      "learning_rate": 1.3017530772099965e-06,
      "loss": 0.6617,
      "step": 17848
    },
    {
      "epoch": 0.7397322723693481,
      "grad_norm": 0.39194589853286743,
      "learning_rate": 1.3015458576816281e-06,
      "loss": 0.6643,
      "step": 17849
    },
    {
      "epoch": 0.7397737162750218,
      "grad_norm": 0.438322514295578,
      "learning_rate": 1.3013386381532597e-06,
      "loss": 0.6544,
      "step": 17850
    },
    {
      "epoch": 0.7398151601806954,
      "grad_norm": 0.4379824697971344,
      "learning_rate": 1.3011314186248913e-06,
      "loss": 0.6677,
      "step": 17851
    },
    {
      "epoch": 0.7398566040863691,
      "grad_norm": 0.4417119324207306,
      "learning_rate": 1.300924199096523e-06,
      "loss": 0.7056,
      "step": 17852
    },
    {
      "epoch": 0.7398980479920427,
      "grad_norm": 0.4286735951900482,
      "learning_rate": 1.3007169795681545e-06,
      "loss": 0.7324,
      "step": 17853
    },
    {
      "epoch": 0.7399394918977165,
      "grad_norm": 0.4006495773792267,
      "learning_rate": 1.3005097600397861e-06,
      "loss": 0.6689,
      "step": 17854
    },
    {
      "epoch": 0.7399809358033901,
      "grad_norm": 0.5095283389091492,
      "learning_rate": 1.3003025405114177e-06,
      "loss": 0.671,
      "step": 17855
    },
    {
      "epoch": 0.7400223797090638,
      "grad_norm": 0.421403706073761,
      "learning_rate": 1.3000953209830495e-06,
      "loss": 0.6909,
      "step": 17856
    },
    {
      "epoch": 0.7400638236147374,
      "grad_norm": 0.4157628118991852,
      "learning_rate": 1.2998881014546813e-06,
      "loss": 0.729,
      "step": 17857
    },
    {
      "epoch": 0.7401052675204112,
      "grad_norm": 0.3860557973384857,
      "learning_rate": 1.299680881926313e-06,
      "loss": 0.6342,
      "step": 17858
    },
    {
      "epoch": 0.7401467114260848,
      "grad_norm": 0.4295955002307892,
      "learning_rate": 1.2994736623979445e-06,
      "loss": 0.6633,
      "step": 17859
    },
    {
      "epoch": 0.7401881553317584,
      "grad_norm": 0.44841268658638,
      "learning_rate": 1.2992664428695761e-06,
      "loss": 0.6821,
      "step": 17860
    },
    {
      "epoch": 0.7402295992374321,
      "grad_norm": 0.4424310028553009,
      "learning_rate": 1.2990592233412077e-06,
      "loss": 0.7317,
      "step": 17861
    },
    {
      "epoch": 0.7402710431431058,
      "grad_norm": 0.42950788140296936,
      "learning_rate": 1.2988520038128393e-06,
      "loss": 0.6353,
      "step": 17862
    },
    {
      "epoch": 0.7403124870487795,
      "grad_norm": 0.459308385848999,
      "learning_rate": 1.298644784284471e-06,
      "loss": 0.7374,
      "step": 17863
    },
    {
      "epoch": 0.7403539309544531,
      "grad_norm": 0.41612425446510315,
      "learning_rate": 1.2984375647561025e-06,
      "loss": 0.6201,
      "step": 17864
    },
    {
      "epoch": 0.7403953748601269,
      "grad_norm": 0.3812253773212433,
      "learning_rate": 1.2982303452277345e-06,
      "loss": 0.6882,
      "step": 17865
    },
    {
      "epoch": 0.7404368187658005,
      "grad_norm": 0.4150495231151581,
      "learning_rate": 1.2980231256993661e-06,
      "loss": 0.6058,
      "step": 17866
    },
    {
      "epoch": 0.7404782626714742,
      "grad_norm": 0.43041861057281494,
      "learning_rate": 1.2978159061709977e-06,
      "loss": 0.6814,
      "step": 17867
    },
    {
      "epoch": 0.7405197065771478,
      "grad_norm": 0.4236215054988861,
      "learning_rate": 1.2976086866426293e-06,
      "loss": 0.6477,
      "step": 17868
    },
    {
      "epoch": 0.7405611504828215,
      "grad_norm": 0.40639829635620117,
      "learning_rate": 1.297401467114261e-06,
      "loss": 0.6544,
      "step": 17869
    },
    {
      "epoch": 0.7406025943884952,
      "grad_norm": 0.4288565218448639,
      "learning_rate": 1.2971942475858925e-06,
      "loss": 0.6696,
      "step": 17870
    },
    {
      "epoch": 0.7406440382941688,
      "grad_norm": 0.4309433102607727,
      "learning_rate": 1.2969870280575241e-06,
      "loss": 0.6738,
      "step": 17871
    },
    {
      "epoch": 0.7406854821998425,
      "grad_norm": 0.4066268801689148,
      "learning_rate": 1.2967798085291557e-06,
      "loss": 0.6578,
      "step": 17872
    },
    {
      "epoch": 0.7407269261055162,
      "grad_norm": 0.400766521692276,
      "learning_rate": 1.2965725890007875e-06,
      "loss": 0.6633,
      "step": 17873
    },
    {
      "epoch": 0.7407683700111899,
      "grad_norm": 0.4134844243526459,
      "learning_rate": 1.2963653694724193e-06,
      "loss": 0.6411,
      "step": 17874
    },
    {
      "epoch": 0.7408098139168635,
      "grad_norm": 0.4513557255268097,
      "learning_rate": 1.296158149944051e-06,
      "loss": 0.6978,
      "step": 17875
    },
    {
      "epoch": 0.7408512578225372,
      "grad_norm": 0.4455469250679016,
      "learning_rate": 1.2959509304156825e-06,
      "loss": 0.6899,
      "step": 17876
    },
    {
      "epoch": 0.7408927017282109,
      "grad_norm": 0.401604026556015,
      "learning_rate": 1.2957437108873141e-06,
      "loss": 0.6682,
      "step": 17877
    },
    {
      "epoch": 0.7409341456338845,
      "grad_norm": 0.42244118452072144,
      "learning_rate": 1.2955364913589457e-06,
      "loss": 0.6348,
      "step": 17878
    },
    {
      "epoch": 0.7409755895395582,
      "grad_norm": 0.3995290696620941,
      "learning_rate": 1.2953292718305773e-06,
      "loss": 0.63,
      "step": 17879
    },
    {
      "epoch": 0.7410170334452318,
      "grad_norm": 0.4102252125740051,
      "learning_rate": 1.295122052302209e-06,
      "loss": 0.6632,
      "step": 17880
    },
    {
      "epoch": 0.7410584773509056,
      "grad_norm": 0.36665990948677063,
      "learning_rate": 1.2949148327738405e-06,
      "loss": 0.6511,
      "step": 17881
    },
    {
      "epoch": 0.7410999212565792,
      "grad_norm": 0.3950020968914032,
      "learning_rate": 1.2947076132454723e-06,
      "loss": 0.629,
      "step": 17882
    },
    {
      "epoch": 0.7411413651622529,
      "grad_norm": 0.40033072233200073,
      "learning_rate": 1.2945003937171041e-06,
      "loss": 0.6572,
      "step": 17883
    },
    {
      "epoch": 0.7411828090679266,
      "grad_norm": 0.40632957220077515,
      "learning_rate": 1.2942931741887357e-06,
      "loss": 0.6794,
      "step": 17884
    },
    {
      "epoch": 0.7412242529736003,
      "grad_norm": 0.45877355337142944,
      "learning_rate": 1.2940859546603673e-06,
      "loss": 0.7332,
      "step": 17885
    },
    {
      "epoch": 0.7412656968792739,
      "grad_norm": 0.4210985004901886,
      "learning_rate": 1.293878735131999e-06,
      "loss": 0.6479,
      "step": 17886
    },
    {
      "epoch": 0.7413071407849475,
      "grad_norm": 0.40694260597229004,
      "learning_rate": 1.2936715156036305e-06,
      "loss": 0.6807,
      "step": 17887
    },
    {
      "epoch": 0.7413485846906213,
      "grad_norm": 0.4106813967227936,
      "learning_rate": 1.2934642960752621e-06,
      "loss": 0.7045,
      "step": 17888
    },
    {
      "epoch": 0.7413900285962949,
      "grad_norm": 0.4084658920764923,
      "learning_rate": 1.2932570765468937e-06,
      "loss": 0.7092,
      "step": 17889
    },
    {
      "epoch": 0.7414314725019686,
      "grad_norm": 0.39607176184654236,
      "learning_rate": 1.2930498570185255e-06,
      "loss": 0.6851,
      "step": 17890
    },
    {
      "epoch": 0.7414729164076422,
      "grad_norm": 0.38920292258262634,
      "learning_rate": 1.2928426374901573e-06,
      "loss": 0.6582,
      "step": 17891
    },
    {
      "epoch": 0.741514360313316,
      "grad_norm": 0.38739362359046936,
      "learning_rate": 1.292635417961789e-06,
      "loss": 0.6216,
      "step": 17892
    },
    {
      "epoch": 0.7415558042189896,
      "grad_norm": 0.4156036078929901,
      "learning_rate": 1.2924281984334205e-06,
      "loss": 0.6426,
      "step": 17893
    },
    {
      "epoch": 0.7415972481246633,
      "grad_norm": 0.39145782589912415,
      "learning_rate": 1.2922209789050521e-06,
      "loss": 0.6056,
      "step": 17894
    },
    {
      "epoch": 0.7416386920303369,
      "grad_norm": 0.4178645610809326,
      "learning_rate": 1.2920137593766837e-06,
      "loss": 0.72,
      "step": 17895
    },
    {
      "epoch": 0.7416801359360106,
      "grad_norm": 0.5193049907684326,
      "learning_rate": 1.2918065398483153e-06,
      "loss": 0.6686,
      "step": 17896
    },
    {
      "epoch": 0.7417215798416843,
      "grad_norm": 0.4042234420776367,
      "learning_rate": 1.291599320319947e-06,
      "loss": 0.6798,
      "step": 17897
    },
    {
      "epoch": 0.7417630237473579,
      "grad_norm": 0.41255268454551697,
      "learning_rate": 1.2913921007915785e-06,
      "loss": 0.6931,
      "step": 17898
    },
    {
      "epoch": 0.7418044676530317,
      "grad_norm": 0.3810236155986786,
      "learning_rate": 1.2911848812632103e-06,
      "loss": 0.6436,
      "step": 17899
    },
    {
      "epoch": 0.7418459115587053,
      "grad_norm": 0.394985169172287,
      "learning_rate": 1.2909776617348421e-06,
      "loss": 0.6833,
      "step": 17900
    },
    {
      "epoch": 0.741887355464379,
      "grad_norm": 0.4578222334384918,
      "learning_rate": 1.2907704422064737e-06,
      "loss": 0.6598,
      "step": 17901
    },
    {
      "epoch": 0.7419287993700526,
      "grad_norm": 0.4051802456378937,
      "learning_rate": 1.2905632226781053e-06,
      "loss": 0.7041,
      "step": 17902
    },
    {
      "epoch": 0.7419702432757264,
      "grad_norm": 0.44669273495674133,
      "learning_rate": 1.290356003149737e-06,
      "loss": 0.6682,
      "step": 17903
    },
    {
      "epoch": 0.7420116871814,
      "grad_norm": 0.40941309928894043,
      "learning_rate": 1.2901487836213685e-06,
      "loss": 0.6798,
      "step": 17904
    },
    {
      "epoch": 0.7420531310870736,
      "grad_norm": 0.4025413393974304,
      "learning_rate": 1.2899415640930001e-06,
      "loss": 0.6277,
      "step": 17905
    },
    {
      "epoch": 0.7420945749927473,
      "grad_norm": 0.4081211984157562,
      "learning_rate": 1.2897343445646317e-06,
      "loss": 0.6469,
      "step": 17906
    },
    {
      "epoch": 0.742136018898421,
      "grad_norm": 0.44304975867271423,
      "learning_rate": 1.2895271250362635e-06,
      "loss": 0.6666,
      "step": 17907
    },
    {
      "epoch": 0.7421774628040947,
      "grad_norm": 0.4066527187824249,
      "learning_rate": 1.2893199055078951e-06,
      "loss": 0.6685,
      "step": 17908
    },
    {
      "epoch": 0.7422189067097683,
      "grad_norm": 0.42509549856185913,
      "learning_rate": 1.289112685979527e-06,
      "loss": 0.7051,
      "step": 17909
    },
    {
      "epoch": 0.742260350615442,
      "grad_norm": 0.406710147857666,
      "learning_rate": 1.2889054664511585e-06,
      "loss": 0.637,
      "step": 17910
    },
    {
      "epoch": 0.7423017945211157,
      "grad_norm": 0.399396687746048,
      "learning_rate": 1.2886982469227901e-06,
      "loss": 0.6361,
      "step": 17911
    },
    {
      "epoch": 0.7423432384267893,
      "grad_norm": 0.4481818377971649,
      "learning_rate": 1.2884910273944217e-06,
      "loss": 0.6251,
      "step": 17912
    },
    {
      "epoch": 0.742384682332463,
      "grad_norm": 0.40548276901245117,
      "learning_rate": 1.2882838078660533e-06,
      "loss": 0.637,
      "step": 17913
    },
    {
      "epoch": 0.7424261262381366,
      "grad_norm": 0.4383731186389923,
      "learning_rate": 1.288076588337685e-06,
      "loss": 0.6917,
      "step": 17914
    },
    {
      "epoch": 0.7424675701438104,
      "grad_norm": 0.4054107367992401,
      "learning_rate": 1.2878693688093165e-06,
      "loss": 0.6633,
      "step": 17915
    },
    {
      "epoch": 0.742509014049484,
      "grad_norm": 0.4100343883037567,
      "learning_rate": 1.2876621492809483e-06,
      "loss": 0.687,
      "step": 17916
    },
    {
      "epoch": 0.7425504579551577,
      "grad_norm": 0.45086705684661865,
      "learning_rate": 1.28745492975258e-06,
      "loss": 0.7053,
      "step": 17917
    },
    {
      "epoch": 0.7425919018608313,
      "grad_norm": 0.37883898615837097,
      "learning_rate": 1.2872477102242117e-06,
      "loss": 0.6555,
      "step": 17918
    },
    {
      "epoch": 0.7426333457665051,
      "grad_norm": 0.39629262685775757,
      "learning_rate": 1.2870404906958433e-06,
      "loss": 0.6755,
      "step": 17919
    },
    {
      "epoch": 0.7426747896721787,
      "grad_norm": 0.41545966267585754,
      "learning_rate": 1.286833271167475e-06,
      "loss": 0.6841,
      "step": 17920
    },
    {
      "epoch": 0.7427162335778523,
      "grad_norm": 0.40257102251052856,
      "learning_rate": 1.2866260516391065e-06,
      "loss": 0.6635,
      "step": 17921
    },
    {
      "epoch": 0.7427576774835261,
      "grad_norm": 0.44758176803588867,
      "learning_rate": 1.2864188321107381e-06,
      "loss": 0.6904,
      "step": 17922
    },
    {
      "epoch": 0.7427991213891997,
      "grad_norm": 0.444850891828537,
      "learning_rate": 1.2862116125823697e-06,
      "loss": 0.6779,
      "step": 17923
    },
    {
      "epoch": 0.7428405652948734,
      "grad_norm": 0.440321147441864,
      "learning_rate": 1.2860043930540015e-06,
      "loss": 0.683,
      "step": 17924
    },
    {
      "epoch": 0.742882009200547,
      "grad_norm": 0.38065704703330994,
      "learning_rate": 1.2857971735256331e-06,
      "loss": 0.6257,
      "step": 17925
    },
    {
      "epoch": 0.7429234531062208,
      "grad_norm": 0.48940539360046387,
      "learning_rate": 1.285589953997265e-06,
      "loss": 0.7411,
      "step": 17926
    },
    {
      "epoch": 0.7429648970118944,
      "grad_norm": 0.40706580877304077,
      "learning_rate": 1.2853827344688965e-06,
      "loss": 0.698,
      "step": 17927
    },
    {
      "epoch": 0.7430063409175681,
      "grad_norm": 0.45417356491088867,
      "learning_rate": 1.2851755149405281e-06,
      "loss": 0.6814,
      "step": 17928
    },
    {
      "epoch": 0.7430477848232417,
      "grad_norm": 0.41347941756248474,
      "learning_rate": 1.2849682954121597e-06,
      "loss": 0.7095,
      "step": 17929
    },
    {
      "epoch": 0.7430892287289154,
      "grad_norm": 0.41316214203834534,
      "learning_rate": 1.2847610758837913e-06,
      "loss": 0.644,
      "step": 17930
    },
    {
      "epoch": 0.7431306726345891,
      "grad_norm": 0.4080352187156677,
      "learning_rate": 1.284553856355423e-06,
      "loss": 0.7072,
      "step": 17931
    },
    {
      "epoch": 0.7431721165402627,
      "grad_norm": 0.4291748106479645,
      "learning_rate": 1.2843466368270547e-06,
      "loss": 0.6503,
      "step": 17932
    },
    {
      "epoch": 0.7432135604459364,
      "grad_norm": 0.40580302476882935,
      "learning_rate": 1.2841394172986863e-06,
      "loss": 0.6953,
      "step": 17933
    },
    {
      "epoch": 0.7432550043516101,
      "grad_norm": 0.441336452960968,
      "learning_rate": 1.283932197770318e-06,
      "loss": 0.7158,
      "step": 17934
    },
    {
      "epoch": 0.7432964482572838,
      "grad_norm": 0.42175135016441345,
      "learning_rate": 1.2837249782419497e-06,
      "loss": 0.6641,
      "step": 17935
    },
    {
      "epoch": 0.7433378921629574,
      "grad_norm": 0.4386596381664276,
      "learning_rate": 1.2835177587135813e-06,
      "loss": 0.7073,
      "step": 17936
    },
    {
      "epoch": 0.7433793360686312,
      "grad_norm": 0.3928120732307434,
      "learning_rate": 1.283310539185213e-06,
      "loss": 0.6169,
      "step": 17937
    },
    {
      "epoch": 0.7434207799743048,
      "grad_norm": 0.38779523968696594,
      "learning_rate": 1.2831033196568445e-06,
      "loss": 0.6965,
      "step": 17938
    },
    {
      "epoch": 0.7434622238799784,
      "grad_norm": 0.4124859869480133,
      "learning_rate": 1.2828961001284761e-06,
      "loss": 0.691,
      "step": 17939
    },
    {
      "epoch": 0.7435036677856521,
      "grad_norm": 0.4071972370147705,
      "learning_rate": 1.2826888806001077e-06,
      "loss": 0.6802,
      "step": 17940
    },
    {
      "epoch": 0.7435451116913258,
      "grad_norm": 0.40728041529655457,
      "learning_rate": 1.2824816610717395e-06,
      "loss": 0.6719,
      "step": 17941
    },
    {
      "epoch": 0.7435865555969995,
      "grad_norm": 0.4035493731498718,
      "learning_rate": 1.2822744415433711e-06,
      "loss": 0.6348,
      "step": 17942
    },
    {
      "epoch": 0.7436279995026731,
      "grad_norm": 0.48423340916633606,
      "learning_rate": 1.2820672220150027e-06,
      "loss": 0.6938,
      "step": 17943
    },
    {
      "epoch": 0.7436694434083468,
      "grad_norm": 0.42305925488471985,
      "learning_rate": 1.2818600024866345e-06,
      "loss": 0.6431,
      "step": 17944
    },
    {
      "epoch": 0.7437108873140205,
      "grad_norm": 0.4159092605113983,
      "learning_rate": 1.2816527829582661e-06,
      "loss": 0.6229,
      "step": 17945
    },
    {
      "epoch": 0.7437523312196942,
      "grad_norm": 0.445028692483902,
      "learning_rate": 1.2814455634298977e-06,
      "loss": 0.6569,
      "step": 17946
    },
    {
      "epoch": 0.7437937751253678,
      "grad_norm": 0.4062625467777252,
      "learning_rate": 1.2812383439015293e-06,
      "loss": 0.6719,
      "step": 17947
    },
    {
      "epoch": 0.7438352190310414,
      "grad_norm": 0.4103319048881531,
      "learning_rate": 1.281031124373161e-06,
      "loss": 0.6799,
      "step": 17948
    },
    {
      "epoch": 0.7438766629367152,
      "grad_norm": 0.4249497950077057,
      "learning_rate": 1.2808239048447927e-06,
      "loss": 0.7114,
      "step": 17949
    },
    {
      "epoch": 0.7439181068423888,
      "grad_norm": 0.44548988342285156,
      "learning_rate": 1.2806166853164243e-06,
      "loss": 0.6892,
      "step": 17950
    },
    {
      "epoch": 0.7439595507480625,
      "grad_norm": 0.37914037704467773,
      "learning_rate": 1.280409465788056e-06,
      "loss": 0.6487,
      "step": 17951
    },
    {
      "epoch": 0.7440009946537361,
      "grad_norm": 0.47239696979522705,
      "learning_rate": 1.2802022462596877e-06,
      "loss": 0.7035,
      "step": 17952
    },
    {
      "epoch": 0.7440424385594099,
      "grad_norm": 0.4493926763534546,
      "learning_rate": 1.2799950267313193e-06,
      "loss": 0.7102,
      "step": 17953
    },
    {
      "epoch": 0.7440838824650835,
      "grad_norm": 0.4170963168144226,
      "learning_rate": 1.279787807202951e-06,
      "loss": 0.6731,
      "step": 17954
    },
    {
      "epoch": 0.7441253263707572,
      "grad_norm": 0.3886983096599579,
      "learning_rate": 1.2795805876745825e-06,
      "loss": 0.6208,
      "step": 17955
    },
    {
      "epoch": 0.7441667702764309,
      "grad_norm": 0.4078763723373413,
      "learning_rate": 1.2793733681462141e-06,
      "loss": 0.6466,
      "step": 17956
    },
    {
      "epoch": 0.7442082141821045,
      "grad_norm": 0.40421056747436523,
      "learning_rate": 1.2791661486178457e-06,
      "loss": 0.6508,
      "step": 17957
    },
    {
      "epoch": 0.7442496580877782,
      "grad_norm": 0.4042975604534149,
      "learning_rate": 1.2789589290894775e-06,
      "loss": 0.6443,
      "step": 17958
    },
    {
      "epoch": 0.7442911019934518,
      "grad_norm": 0.4701588749885559,
      "learning_rate": 1.2787517095611091e-06,
      "loss": 0.6626,
      "step": 17959
    },
    {
      "epoch": 0.7443325458991256,
      "grad_norm": 0.43352964520454407,
      "learning_rate": 1.2785444900327407e-06,
      "loss": 0.7665,
      "step": 17960
    },
    {
      "epoch": 0.7443739898047992,
      "grad_norm": 0.3970031440258026,
      "learning_rate": 1.2783372705043725e-06,
      "loss": 0.6658,
      "step": 17961
    },
    {
      "epoch": 0.7444154337104729,
      "grad_norm": 0.4194445312023163,
      "learning_rate": 1.2781300509760041e-06,
      "loss": 0.6653,
      "step": 17962
    },
    {
      "epoch": 0.7444568776161465,
      "grad_norm": 0.40324634313583374,
      "learning_rate": 1.2779228314476357e-06,
      "loss": 0.6177,
      "step": 17963
    },
    {
      "epoch": 0.7444983215218202,
      "grad_norm": 0.38328710198402405,
      "learning_rate": 1.2777156119192673e-06,
      "loss": 0.6924,
      "step": 17964
    },
    {
      "epoch": 0.7445397654274939,
      "grad_norm": 0.43711209297180176,
      "learning_rate": 1.277508392390899e-06,
      "loss": 0.6967,
      "step": 17965
    },
    {
      "epoch": 0.7445812093331675,
      "grad_norm": 0.45097804069519043,
      "learning_rate": 1.2773011728625307e-06,
      "loss": 0.7024,
      "step": 17966
    },
    {
      "epoch": 0.7446226532388412,
      "grad_norm": 0.40400218963623047,
      "learning_rate": 1.2770939533341623e-06,
      "loss": 0.6613,
      "step": 17967
    },
    {
      "epoch": 0.7446640971445149,
      "grad_norm": 0.43957170844078064,
      "learning_rate": 1.276886733805794e-06,
      "loss": 0.6565,
      "step": 17968
    },
    {
      "epoch": 0.7447055410501886,
      "grad_norm": 0.3820628821849823,
      "learning_rate": 1.2766795142774255e-06,
      "loss": 0.6412,
      "step": 17969
    },
    {
      "epoch": 0.7447469849558622,
      "grad_norm": 0.39659222960472107,
      "learning_rate": 1.2764722947490573e-06,
      "loss": 0.6184,
      "step": 17970
    },
    {
      "epoch": 0.744788428861536,
      "grad_norm": 0.4141824543476105,
      "learning_rate": 1.276265075220689e-06,
      "loss": 0.6173,
      "step": 17971
    },
    {
      "epoch": 0.7448298727672096,
      "grad_norm": 0.4708532691001892,
      "learning_rate": 1.2760578556923205e-06,
      "loss": 0.6914,
      "step": 17972
    },
    {
      "epoch": 0.7448713166728832,
      "grad_norm": 0.40130093693733215,
      "learning_rate": 1.2758506361639521e-06,
      "loss": 0.6447,
      "step": 17973
    },
    {
      "epoch": 0.7449127605785569,
      "grad_norm": 0.431858092546463,
      "learning_rate": 1.2756434166355837e-06,
      "loss": 0.6677,
      "step": 17974
    },
    {
      "epoch": 0.7449542044842306,
      "grad_norm": 0.3850264549255371,
      "learning_rate": 1.2754361971072155e-06,
      "loss": 0.6567,
      "step": 17975
    },
    {
      "epoch": 0.7449956483899043,
      "grad_norm": 0.40212851762771606,
      "learning_rate": 1.2752289775788471e-06,
      "loss": 0.6533,
      "step": 17976
    },
    {
      "epoch": 0.7450370922955779,
      "grad_norm": 0.40164652466773987,
      "learning_rate": 1.2750217580504787e-06,
      "loss": 0.6368,
      "step": 17977
    },
    {
      "epoch": 0.7450785362012516,
      "grad_norm": 0.42696163058280945,
      "learning_rate": 1.2748145385221103e-06,
      "loss": 0.6661,
      "step": 17978
    },
    {
      "epoch": 0.7451199801069253,
      "grad_norm": 0.4164913594722748,
      "learning_rate": 1.2746073189937421e-06,
      "loss": 0.6803,
      "step": 17979
    },
    {
      "epoch": 0.745161424012599,
      "grad_norm": 0.39903637766838074,
      "learning_rate": 1.2744000994653737e-06,
      "loss": 0.6675,
      "step": 17980
    },
    {
      "epoch": 0.7452028679182726,
      "grad_norm": 0.4060525596141815,
      "learning_rate": 1.2741928799370053e-06,
      "loss": 0.6534,
      "step": 17981
    },
    {
      "epoch": 0.7452443118239462,
      "grad_norm": 0.46491873264312744,
      "learning_rate": 1.273985660408637e-06,
      "loss": 0.7365,
      "step": 17982
    },
    {
      "epoch": 0.74528575572962,
      "grad_norm": 0.3856048583984375,
      "learning_rate": 1.2737784408802687e-06,
      "loss": 0.656,
      "step": 17983
    },
    {
      "epoch": 0.7453271996352936,
      "grad_norm": 0.3953703045845032,
      "learning_rate": 1.2735712213519003e-06,
      "loss": 0.6663,
      "step": 17984
    },
    {
      "epoch": 0.7453686435409673,
      "grad_norm": 0.39112481474876404,
      "learning_rate": 1.273364001823532e-06,
      "loss": 0.687,
      "step": 17985
    },
    {
      "epoch": 0.745410087446641,
      "grad_norm": 0.44401276111602783,
      "learning_rate": 1.2731567822951635e-06,
      "loss": 0.6697,
      "step": 17986
    },
    {
      "epoch": 0.7454515313523147,
      "grad_norm": 0.41151028871536255,
      "learning_rate": 1.2729495627667953e-06,
      "loss": 0.6749,
      "step": 17987
    },
    {
      "epoch": 0.7454929752579883,
      "grad_norm": 0.4557395279407501,
      "learning_rate": 1.272742343238427e-06,
      "loss": 0.7485,
      "step": 17988
    },
    {
      "epoch": 0.745534419163662,
      "grad_norm": 0.3865765631198883,
      "learning_rate": 1.2725351237100585e-06,
      "loss": 0.5995,
      "step": 17989
    },
    {
      "epoch": 0.7455758630693357,
      "grad_norm": 0.4425818920135498,
      "learning_rate": 1.2723279041816901e-06,
      "loss": 0.6631,
      "step": 17990
    },
    {
      "epoch": 0.7456173069750093,
      "grad_norm": 0.3990689814090729,
      "learning_rate": 1.2721206846533217e-06,
      "loss": 0.741,
      "step": 17991
    },
    {
      "epoch": 0.745658750880683,
      "grad_norm": 0.41478872299194336,
      "learning_rate": 1.2719134651249535e-06,
      "loss": 0.6328,
      "step": 17992
    },
    {
      "epoch": 0.7457001947863566,
      "grad_norm": 0.4028747081756592,
      "learning_rate": 1.2717062455965851e-06,
      "loss": 0.6755,
      "step": 17993
    },
    {
      "epoch": 0.7457416386920304,
      "grad_norm": 0.3823970556259155,
      "learning_rate": 1.2714990260682167e-06,
      "loss": 0.671,
      "step": 17994
    },
    {
      "epoch": 0.745783082597704,
      "grad_norm": 0.39892029762268066,
      "learning_rate": 1.2712918065398483e-06,
      "loss": 0.6694,
      "step": 17995
    },
    {
      "epoch": 0.7458245265033777,
      "grad_norm": 0.424058198928833,
      "learning_rate": 1.2710845870114801e-06,
      "loss": 0.6779,
      "step": 17996
    },
    {
      "epoch": 0.7458659704090513,
      "grad_norm": 0.41004472970962524,
      "learning_rate": 1.2708773674831117e-06,
      "loss": 0.6887,
      "step": 17997
    },
    {
      "epoch": 0.7459074143147251,
      "grad_norm": 0.39678674936294556,
      "learning_rate": 1.2706701479547433e-06,
      "loss": 0.5887,
      "step": 17998
    },
    {
      "epoch": 0.7459488582203987,
      "grad_norm": 0.3990894556045532,
      "learning_rate": 1.270462928426375e-06,
      "loss": 0.6483,
      "step": 17999
    },
    {
      "epoch": 0.7459903021260723,
      "grad_norm": 0.4123969078063965,
      "learning_rate": 1.2702557088980067e-06,
      "loss": 0.7126,
      "step": 18000
    },
    {
      "epoch": 0.746031746031746,
      "grad_norm": 0.43104955554008484,
      "learning_rate": 1.2700484893696383e-06,
      "loss": 0.7192,
      "step": 18001
    },
    {
      "epoch": 0.7460731899374197,
      "grad_norm": 0.3994036614894867,
      "learning_rate": 1.26984126984127e-06,
      "loss": 0.693,
      "step": 18002
    },
    {
      "epoch": 0.7461146338430934,
      "grad_norm": 0.4198873043060303,
      "learning_rate": 1.2696340503129015e-06,
      "loss": 0.6831,
      "step": 18003
    },
    {
      "epoch": 0.746156077748767,
      "grad_norm": 0.44471997022628784,
      "learning_rate": 1.2694268307845331e-06,
      "loss": 0.6573,
      "step": 18004
    },
    {
      "epoch": 0.7461975216544408,
      "grad_norm": 0.4223025143146515,
      "learning_rate": 1.269219611256165e-06,
      "loss": 0.686,
      "step": 18005
    },
    {
      "epoch": 0.7462389655601144,
      "grad_norm": 0.40300053358078003,
      "learning_rate": 1.2690123917277965e-06,
      "loss": 0.696,
      "step": 18006
    },
    {
      "epoch": 0.7462804094657881,
      "grad_norm": 0.39616361260414124,
      "learning_rate": 1.2688051721994281e-06,
      "loss": 0.6848,
      "step": 18007
    },
    {
      "epoch": 0.7463218533714617,
      "grad_norm": 0.40903550386428833,
      "learning_rate": 1.2685979526710597e-06,
      "loss": 0.6792,
      "step": 18008
    },
    {
      "epoch": 0.7463632972771354,
      "grad_norm": 0.44018179178237915,
      "learning_rate": 1.2683907331426915e-06,
      "loss": 0.66,
      "step": 18009
    },
    {
      "epoch": 0.7464047411828091,
      "grad_norm": 0.43438011407852173,
      "learning_rate": 1.2681835136143231e-06,
      "loss": 0.7068,
      "step": 18010
    },
    {
      "epoch": 0.7464461850884827,
      "grad_norm": 0.42794495820999146,
      "learning_rate": 1.2679762940859547e-06,
      "loss": 0.6783,
      "step": 18011
    },
    {
      "epoch": 0.7464876289941564,
      "grad_norm": 0.4277538061141968,
      "learning_rate": 1.2677690745575863e-06,
      "loss": 0.6783,
      "step": 18012
    },
    {
      "epoch": 0.7465290728998301,
      "grad_norm": 0.41058090329170227,
      "learning_rate": 1.2675618550292181e-06,
      "loss": 0.6754,
      "step": 18013
    },
    {
      "epoch": 0.7465705168055038,
      "grad_norm": 0.449241042137146,
      "learning_rate": 1.2673546355008497e-06,
      "loss": 0.7073,
      "step": 18014
    },
    {
      "epoch": 0.7466119607111774,
      "grad_norm": 0.4004034698009491,
      "learning_rate": 1.2671474159724813e-06,
      "loss": 0.6575,
      "step": 18015
    },
    {
      "epoch": 0.7466534046168511,
      "grad_norm": 0.38303661346435547,
      "learning_rate": 1.266940196444113e-06,
      "loss": 0.6738,
      "step": 18016
    },
    {
      "epoch": 0.7466948485225248,
      "grad_norm": 0.3967627286911011,
      "learning_rate": 1.2667329769157447e-06,
      "loss": 0.6816,
      "step": 18017
    },
    {
      "epoch": 0.7467362924281984,
      "grad_norm": 0.39240753650665283,
      "learning_rate": 1.2665257573873763e-06,
      "loss": 0.6499,
      "step": 18018
    },
    {
      "epoch": 0.7467777363338721,
      "grad_norm": 0.37365520000457764,
      "learning_rate": 1.266318537859008e-06,
      "loss": 0.6182,
      "step": 18019
    },
    {
      "epoch": 0.7468191802395457,
      "grad_norm": 0.39595508575439453,
      "learning_rate": 1.2661113183306395e-06,
      "loss": 0.6383,
      "step": 18020
    },
    {
      "epoch": 0.7468606241452195,
      "grad_norm": 0.4256483018398285,
      "learning_rate": 1.2659040988022711e-06,
      "loss": 0.6814,
      "step": 18021
    },
    {
      "epoch": 0.7469020680508931,
      "grad_norm": 0.4003860056400299,
      "learning_rate": 1.265696879273903e-06,
      "loss": 0.6572,
      "step": 18022
    },
    {
      "epoch": 0.7469435119565668,
      "grad_norm": 0.44488412141799927,
      "learning_rate": 1.2654896597455345e-06,
      "loss": 0.7125,
      "step": 18023
    },
    {
      "epoch": 0.7469849558622405,
      "grad_norm": 0.41488760709762573,
      "learning_rate": 1.2652824402171661e-06,
      "loss": 0.6648,
      "step": 18024
    },
    {
      "epoch": 0.7470263997679141,
      "grad_norm": 0.41693955659866333,
      "learning_rate": 1.265075220688798e-06,
      "loss": 0.7166,
      "step": 18025
    },
    {
      "epoch": 0.7470678436735878,
      "grad_norm": 0.47288286685943604,
      "learning_rate": 1.2648680011604295e-06,
      "loss": 0.689,
      "step": 18026
    },
    {
      "epoch": 0.7471092875792614,
      "grad_norm": 0.4486538767814636,
      "learning_rate": 1.2646607816320611e-06,
      "loss": 0.6997,
      "step": 18027
    },
    {
      "epoch": 0.7471507314849352,
      "grad_norm": 0.38557592034339905,
      "learning_rate": 1.2644535621036927e-06,
      "loss": 0.6643,
      "step": 18028
    },
    {
      "epoch": 0.7471921753906088,
      "grad_norm": 0.42697638273239136,
      "learning_rate": 1.2642463425753243e-06,
      "loss": 0.6384,
      "step": 18029
    },
    {
      "epoch": 0.7472336192962825,
      "grad_norm": 0.4298316538333893,
      "learning_rate": 1.264039123046956e-06,
      "loss": 0.7166,
      "step": 18030
    },
    {
      "epoch": 0.7472750632019561,
      "grad_norm": 0.4366237223148346,
      "learning_rate": 1.2638319035185877e-06,
      "loss": 0.774,
      "step": 18031
    },
    {
      "epoch": 0.7473165071076299,
      "grad_norm": 0.3822805881500244,
      "learning_rate": 1.2636246839902193e-06,
      "loss": 0.6539,
      "step": 18032
    },
    {
      "epoch": 0.7473579510133035,
      "grad_norm": 0.4416252374649048,
      "learning_rate": 1.263417464461851e-06,
      "loss": 0.6661,
      "step": 18033
    },
    {
      "epoch": 0.7473993949189771,
      "grad_norm": 0.41685259342193604,
      "learning_rate": 1.2632102449334827e-06,
      "loss": 0.6807,
      "step": 18034
    },
    {
      "epoch": 0.7474408388246508,
      "grad_norm": 0.4027719497680664,
      "learning_rate": 1.2630030254051143e-06,
      "loss": 0.6689,
      "step": 18035
    },
    {
      "epoch": 0.7474822827303245,
      "grad_norm": 0.3914279639720917,
      "learning_rate": 1.262795805876746e-06,
      "loss": 0.6635,
      "step": 18036
    },
    {
      "epoch": 0.7475237266359982,
      "grad_norm": 0.4317137598991394,
      "learning_rate": 1.2625885863483775e-06,
      "loss": 0.6521,
      "step": 18037
    },
    {
      "epoch": 0.7475651705416718,
      "grad_norm": 0.4281472861766815,
      "learning_rate": 1.2623813668200091e-06,
      "loss": 0.6542,
      "step": 18038
    },
    {
      "epoch": 0.7476066144473456,
      "grad_norm": 0.4050513207912445,
      "learning_rate": 1.262174147291641e-06,
      "loss": 0.6332,
      "step": 18039
    },
    {
      "epoch": 0.7476480583530192,
      "grad_norm": 0.42278122901916504,
      "learning_rate": 1.2619669277632725e-06,
      "loss": 0.6649,
      "step": 18040
    },
    {
      "epoch": 0.7476895022586929,
      "grad_norm": 0.4415608048439026,
      "learning_rate": 1.2617597082349041e-06,
      "loss": 0.6833,
      "step": 18041
    },
    {
      "epoch": 0.7477309461643665,
      "grad_norm": 0.4337461292743683,
      "learning_rate": 1.261552488706536e-06,
      "loss": 0.6838,
      "step": 18042
    },
    {
      "epoch": 0.7477723900700401,
      "grad_norm": 0.41062915325164795,
      "learning_rate": 1.2613452691781675e-06,
      "loss": 0.6548,
      "step": 18043
    },
    {
      "epoch": 0.7478138339757139,
      "grad_norm": 0.4038822054862976,
      "learning_rate": 1.2611380496497991e-06,
      "loss": 0.6841,
      "step": 18044
    },
    {
      "epoch": 0.7478552778813875,
      "grad_norm": 0.40963801741600037,
      "learning_rate": 1.2609308301214307e-06,
      "loss": 0.6713,
      "step": 18045
    },
    {
      "epoch": 0.7478967217870612,
      "grad_norm": 0.4199795126914978,
      "learning_rate": 1.2607236105930623e-06,
      "loss": 0.7594,
      "step": 18046
    },
    {
      "epoch": 0.7479381656927349,
      "grad_norm": 0.399190217256546,
      "learning_rate": 1.260516391064694e-06,
      "loss": 0.6516,
      "step": 18047
    },
    {
      "epoch": 0.7479796095984086,
      "grad_norm": 0.4055478870868683,
      "learning_rate": 1.2603091715363257e-06,
      "loss": 0.6759,
      "step": 18048
    },
    {
      "epoch": 0.7480210535040822,
      "grad_norm": 0.41443192958831787,
      "learning_rate": 1.2601019520079573e-06,
      "loss": 0.6418,
      "step": 18049
    },
    {
      "epoch": 0.7480624974097559,
      "grad_norm": 0.38799428939819336,
      "learning_rate": 1.259894732479589e-06,
      "loss": 0.6208,
      "step": 18050
    },
    {
      "epoch": 0.7481039413154296,
      "grad_norm": 0.4371360242366791,
      "learning_rate": 1.2596875129512208e-06,
      "loss": 0.6873,
      "step": 18051
    },
    {
      "epoch": 0.7481453852211032,
      "grad_norm": 0.3942939341068268,
      "learning_rate": 1.2594802934228524e-06,
      "loss": 0.6792,
      "step": 18052
    },
    {
      "epoch": 0.7481868291267769,
      "grad_norm": 0.408089280128479,
      "learning_rate": 1.259273073894484e-06,
      "loss": 0.7136,
      "step": 18053
    },
    {
      "epoch": 0.7482282730324505,
      "grad_norm": 0.393614798784256,
      "learning_rate": 1.2590658543661155e-06,
      "loss": 0.6497,
      "step": 18054
    },
    {
      "epoch": 0.7482697169381243,
      "grad_norm": 0.3956488072872162,
      "learning_rate": 1.2588586348377471e-06,
      "loss": 0.6613,
      "step": 18055
    },
    {
      "epoch": 0.7483111608437979,
      "grad_norm": 0.3749770522117615,
      "learning_rate": 1.2586514153093787e-06,
      "loss": 0.6278,
      "step": 18056
    },
    {
      "epoch": 0.7483526047494716,
      "grad_norm": 0.3833193778991699,
      "learning_rate": 1.2584441957810105e-06,
      "loss": 0.593,
      "step": 18057
    },
    {
      "epoch": 0.7483940486551452,
      "grad_norm": 0.42074131965637207,
      "learning_rate": 1.2582369762526421e-06,
      "loss": 0.6802,
      "step": 18058
    },
    {
      "epoch": 0.748435492560819,
      "grad_norm": 0.42019370198249817,
      "learning_rate": 1.258029756724274e-06,
      "loss": 0.6914,
      "step": 18059
    },
    {
      "epoch": 0.7484769364664926,
      "grad_norm": 0.4084164798259735,
      "learning_rate": 1.2578225371959056e-06,
      "loss": 0.6626,
      "step": 18060
    },
    {
      "epoch": 0.7485183803721662,
      "grad_norm": 0.43233099579811096,
      "learning_rate": 1.2576153176675372e-06,
      "loss": 0.6737,
      "step": 18061
    },
    {
      "epoch": 0.74855982427784,
      "grad_norm": 0.4396106004714966,
      "learning_rate": 1.2574080981391687e-06,
      "loss": 0.6855,
      "step": 18062
    },
    {
      "epoch": 0.7486012681835136,
      "grad_norm": 0.4270753562450409,
      "learning_rate": 1.2572008786108003e-06,
      "loss": 0.6877,
      "step": 18063
    },
    {
      "epoch": 0.7486427120891873,
      "grad_norm": 0.3926583528518677,
      "learning_rate": 1.256993659082432e-06,
      "loss": 0.6663,
      "step": 18064
    },
    {
      "epoch": 0.7486841559948609,
      "grad_norm": 0.40158137679100037,
      "learning_rate": 1.2567864395540635e-06,
      "loss": 0.6919,
      "step": 18065
    },
    {
      "epoch": 0.7487255999005347,
      "grad_norm": 0.39559832215309143,
      "learning_rate": 1.2565792200256953e-06,
      "loss": 0.7156,
      "step": 18066
    },
    {
      "epoch": 0.7487670438062083,
      "grad_norm": 0.3891668915748596,
      "learning_rate": 1.256372000497327e-06,
      "loss": 0.5944,
      "step": 18067
    },
    {
      "epoch": 0.748808487711882,
      "grad_norm": 0.39854809641838074,
      "learning_rate": 1.2561647809689588e-06,
      "loss": 0.667,
      "step": 18068
    },
    {
      "epoch": 0.7488499316175556,
      "grad_norm": 0.48829442262649536,
      "learning_rate": 1.2559575614405904e-06,
      "loss": 0.7683,
      "step": 18069
    },
    {
      "epoch": 0.7488913755232293,
      "grad_norm": 0.41078442335128784,
      "learning_rate": 1.255750341912222e-06,
      "loss": 0.6406,
      "step": 18070
    },
    {
      "epoch": 0.748932819428903,
      "grad_norm": 0.40055447816848755,
      "learning_rate": 1.2555431223838535e-06,
      "loss": 0.6487,
      "step": 18071
    },
    {
      "epoch": 0.7489742633345766,
      "grad_norm": 0.424227237701416,
      "learning_rate": 1.2553359028554851e-06,
      "loss": 0.6575,
      "step": 18072
    },
    {
      "epoch": 0.7490157072402504,
      "grad_norm": 0.4163024425506592,
      "learning_rate": 1.2551286833271167e-06,
      "loss": 0.6853,
      "step": 18073
    },
    {
      "epoch": 0.749057151145924,
      "grad_norm": 0.4982098937034607,
      "learning_rate": 1.2549214637987486e-06,
      "loss": 0.6943,
      "step": 18074
    },
    {
      "epoch": 0.7490985950515977,
      "grad_norm": 0.42212527990341187,
      "learning_rate": 1.2547142442703801e-06,
      "loss": 0.6694,
      "step": 18075
    },
    {
      "epoch": 0.7491400389572713,
      "grad_norm": 0.41553330421447754,
      "learning_rate": 1.254507024742012e-06,
      "loss": 0.611,
      "step": 18076
    },
    {
      "epoch": 0.7491814828629451,
      "grad_norm": 0.3973764181137085,
      "learning_rate": 1.2542998052136436e-06,
      "loss": 0.6399,
      "step": 18077
    },
    {
      "epoch": 0.7492229267686187,
      "grad_norm": 0.49291864037513733,
      "learning_rate": 1.2540925856852752e-06,
      "loss": 0.7334,
      "step": 18078
    },
    {
      "epoch": 0.7492643706742923,
      "grad_norm": 0.42273131012916565,
      "learning_rate": 1.2538853661569068e-06,
      "loss": 0.6779,
      "step": 18079
    },
    {
      "epoch": 0.749305814579966,
      "grad_norm": 0.41080018877983093,
      "learning_rate": 1.2536781466285383e-06,
      "loss": 0.6726,
      "step": 18080
    },
    {
      "epoch": 0.7493472584856397,
      "grad_norm": 0.41369596123695374,
      "learning_rate": 1.25347092710017e-06,
      "loss": 0.6833,
      "step": 18081
    },
    {
      "epoch": 0.7493887023913134,
      "grad_norm": 0.41463503241539,
      "learning_rate": 1.2532637075718015e-06,
      "loss": 0.6587,
      "step": 18082
    },
    {
      "epoch": 0.749430146296987,
      "grad_norm": 0.47034165263175964,
      "learning_rate": 1.2530564880434334e-06,
      "loss": 0.6359,
      "step": 18083
    },
    {
      "epoch": 0.7494715902026607,
      "grad_norm": 0.4203554093837738,
      "learning_rate": 1.252849268515065e-06,
      "loss": 0.6602,
      "step": 18084
    },
    {
      "epoch": 0.7495130341083344,
      "grad_norm": 0.40883710980415344,
      "learning_rate": 1.2526420489866968e-06,
      "loss": 0.6598,
      "step": 18085
    },
    {
      "epoch": 0.749554478014008,
      "grad_norm": 0.43225979804992676,
      "learning_rate": 1.2524348294583284e-06,
      "loss": 0.6248,
      "step": 18086
    },
    {
      "epoch": 0.7495959219196817,
      "grad_norm": 0.4079810380935669,
      "learning_rate": 1.25222760992996e-06,
      "loss": 0.6628,
      "step": 18087
    },
    {
      "epoch": 0.7496373658253553,
      "grad_norm": 0.42913246154785156,
      "learning_rate": 1.2520203904015916e-06,
      "loss": 0.7161,
      "step": 18088
    },
    {
      "epoch": 0.7496788097310291,
      "grad_norm": 0.41871094703674316,
      "learning_rate": 1.2518131708732231e-06,
      "loss": 0.6808,
      "step": 18089
    },
    {
      "epoch": 0.7497202536367027,
      "grad_norm": 0.43216896057128906,
      "learning_rate": 1.2516059513448547e-06,
      "loss": 0.6924,
      "step": 18090
    },
    {
      "epoch": 0.7497616975423764,
      "grad_norm": 0.4284150004386902,
      "learning_rate": 1.2513987318164863e-06,
      "loss": 0.6167,
      "step": 18091
    },
    {
      "epoch": 0.74980314144805,
      "grad_norm": 0.4093926250934601,
      "learning_rate": 1.2511915122881182e-06,
      "loss": 0.698,
      "step": 18092
    },
    {
      "epoch": 0.7498445853537238,
      "grad_norm": 0.4449462592601776,
      "learning_rate": 1.25098429275975e-06,
      "loss": 0.6599,
      "step": 18093
    },
    {
      "epoch": 0.7498860292593974,
      "grad_norm": 0.39144787192344666,
      "learning_rate": 1.2507770732313816e-06,
      "loss": 0.6229,
      "step": 18094
    },
    {
      "epoch": 0.749927473165071,
      "grad_norm": 0.42938515543937683,
      "learning_rate": 1.2505698537030132e-06,
      "loss": 0.7241,
      "step": 18095
    },
    {
      "epoch": 0.7499689170707448,
      "grad_norm": 0.40155020356178284,
      "learning_rate": 1.2503626341746448e-06,
      "loss": 0.6609,
      "step": 18096
    },
    {
      "epoch": 0.7500103609764184,
      "grad_norm": 0.4145388603210449,
      "learning_rate": 1.2501554146462764e-06,
      "loss": 0.7012,
      "step": 18097
    },
    {
      "epoch": 0.7500518048820921,
      "grad_norm": 0.4284125864505768,
      "learning_rate": 1.2499481951179082e-06,
      "loss": 0.6949,
      "step": 18098
    },
    {
      "epoch": 0.7500932487877657,
      "grad_norm": 0.4008921682834625,
      "learning_rate": 1.2497409755895398e-06,
      "loss": 0.6569,
      "step": 18099
    },
    {
      "epoch": 0.7501346926934395,
      "grad_norm": 0.40656721591949463,
      "learning_rate": 1.2495337560611714e-06,
      "loss": 0.6879,
      "step": 18100
    },
    {
      "epoch": 0.7501761365991131,
      "grad_norm": 0.3756991922855377,
      "learning_rate": 1.249326536532803e-06,
      "loss": 0.6637,
      "step": 18101
    },
    {
      "epoch": 0.7502175805047868,
      "grad_norm": 0.4266617000102997,
      "learning_rate": 1.2491193170044346e-06,
      "loss": 0.707,
      "step": 18102
    },
    {
      "epoch": 0.7502590244104604,
      "grad_norm": 0.3844257593154907,
      "learning_rate": 1.2489120974760664e-06,
      "loss": 0.6318,
      "step": 18103
    },
    {
      "epoch": 0.7503004683161341,
      "grad_norm": 0.44730669260025024,
      "learning_rate": 1.248704877947698e-06,
      "loss": 0.6829,
      "step": 18104
    },
    {
      "epoch": 0.7503419122218078,
      "grad_norm": 0.40021583437919617,
      "learning_rate": 1.2484976584193296e-06,
      "loss": 0.699,
      "step": 18105
    },
    {
      "epoch": 0.7503833561274814,
      "grad_norm": 0.37868693470954895,
      "learning_rate": 1.2482904388909612e-06,
      "loss": 0.6517,
      "step": 18106
    },
    {
      "epoch": 0.7504248000331551,
      "grad_norm": 0.3893684446811676,
      "learning_rate": 1.248083219362593e-06,
      "loss": 0.6608,
      "step": 18107
    },
    {
      "epoch": 0.7504662439388288,
      "grad_norm": 0.41274234652519226,
      "learning_rate": 1.2478759998342246e-06,
      "loss": 0.6819,
      "step": 18108
    },
    {
      "epoch": 0.7505076878445025,
      "grad_norm": 0.4035905599594116,
      "learning_rate": 1.2476687803058562e-06,
      "loss": 0.6495,
      "step": 18109
    },
    {
      "epoch": 0.7505491317501761,
      "grad_norm": 0.415272980928421,
      "learning_rate": 1.2474615607774878e-06,
      "loss": 0.6495,
      "step": 18110
    },
    {
      "epoch": 0.7505905756558499,
      "grad_norm": 0.43477997183799744,
      "learning_rate": 1.2472543412491196e-06,
      "loss": 0.6788,
      "step": 18111
    },
    {
      "epoch": 0.7506320195615235,
      "grad_norm": 0.43373411893844604,
      "learning_rate": 1.2470471217207512e-06,
      "loss": 0.6726,
      "step": 18112
    },
    {
      "epoch": 0.7506734634671971,
      "grad_norm": 0.39661386609077454,
      "learning_rate": 1.2468399021923828e-06,
      "loss": 0.7021,
      "step": 18113
    },
    {
      "epoch": 0.7507149073728708,
      "grad_norm": 0.4066760241985321,
      "learning_rate": 1.2466326826640144e-06,
      "loss": 0.6407,
      "step": 18114
    },
    {
      "epoch": 0.7507563512785445,
      "grad_norm": 0.4205028712749481,
      "learning_rate": 1.246425463135646e-06,
      "loss": 0.6793,
      "step": 18115
    },
    {
      "epoch": 0.7507977951842182,
      "grad_norm": 0.4091859757900238,
      "learning_rate": 1.2462182436072778e-06,
      "loss": 0.6455,
      "step": 18116
    },
    {
      "epoch": 0.7508392390898918,
      "grad_norm": 0.44470667839050293,
      "learning_rate": 1.2460110240789094e-06,
      "loss": 0.6698,
      "step": 18117
    },
    {
      "epoch": 0.7508806829955655,
      "grad_norm": 0.40063872933387756,
      "learning_rate": 1.245803804550541e-06,
      "loss": 0.6523,
      "step": 18118
    },
    {
      "epoch": 0.7509221269012392,
      "grad_norm": 0.40479418635368347,
      "learning_rate": 1.2455965850221726e-06,
      "loss": 0.6649,
      "step": 18119
    },
    {
      "epoch": 0.7509635708069129,
      "grad_norm": 0.4068812429904938,
      "learning_rate": 1.2453893654938044e-06,
      "loss": 0.696,
      "step": 18120
    },
    {
      "epoch": 0.7510050147125865,
      "grad_norm": 0.4208543300628662,
      "learning_rate": 1.245182145965436e-06,
      "loss": 0.6858,
      "step": 18121
    },
    {
      "epoch": 0.7510464586182601,
      "grad_norm": 0.4267224669456482,
      "learning_rate": 1.2449749264370676e-06,
      "loss": 0.6775,
      "step": 18122
    },
    {
      "epoch": 0.7510879025239339,
      "grad_norm": 0.4049726128578186,
      "learning_rate": 1.2447677069086992e-06,
      "loss": 0.7092,
      "step": 18123
    },
    {
      "epoch": 0.7511293464296075,
      "grad_norm": 0.43642547726631165,
      "learning_rate": 1.244560487380331e-06,
      "loss": 0.6289,
      "step": 18124
    },
    {
      "epoch": 0.7511707903352812,
      "grad_norm": 0.4001486003398895,
      "learning_rate": 1.2443532678519626e-06,
      "loss": 0.6582,
      "step": 18125
    },
    {
      "epoch": 0.7512122342409548,
      "grad_norm": 0.4406491816043854,
      "learning_rate": 1.2441460483235942e-06,
      "loss": 0.7123,
      "step": 18126
    },
    {
      "epoch": 0.7512536781466286,
      "grad_norm": 0.38485848903656006,
      "learning_rate": 1.2439388287952258e-06,
      "loss": 0.6326,
      "step": 18127
    },
    {
      "epoch": 0.7512951220523022,
      "grad_norm": 0.4575856626033783,
      "learning_rate": 1.2437316092668574e-06,
      "loss": 0.6323,
      "step": 18128
    },
    {
      "epoch": 0.7513365659579759,
      "grad_norm": 0.36781466007232666,
      "learning_rate": 1.2435243897384892e-06,
      "loss": 0.6643,
      "step": 18129
    },
    {
      "epoch": 0.7513780098636496,
      "grad_norm": 0.41730374097824097,
      "learning_rate": 1.2433171702101208e-06,
      "loss": 0.6599,
      "step": 18130
    },
    {
      "epoch": 0.7514194537693232,
      "grad_norm": 0.4034154415130615,
      "learning_rate": 1.2431099506817524e-06,
      "loss": 0.6648,
      "step": 18131
    },
    {
      "epoch": 0.7514608976749969,
      "grad_norm": 0.3895029127597809,
      "learning_rate": 1.242902731153384e-06,
      "loss": 0.5907,
      "step": 18132
    },
    {
      "epoch": 0.7515023415806705,
      "grad_norm": 0.38178855180740356,
      "learning_rate": 1.2426955116250158e-06,
      "loss": 0.6859,
      "step": 18133
    },
    {
      "epoch": 0.7515437854863443,
      "grad_norm": 0.4604616165161133,
      "learning_rate": 1.2424882920966474e-06,
      "loss": 0.7395,
      "step": 18134
    },
    {
      "epoch": 0.7515852293920179,
      "grad_norm": 0.41205278038978577,
      "learning_rate": 1.242281072568279e-06,
      "loss": 0.6758,
      "step": 18135
    },
    {
      "epoch": 0.7516266732976916,
      "grad_norm": 0.3968134820461273,
      "learning_rate": 1.2420738530399106e-06,
      "loss": 0.6614,
      "step": 18136
    },
    {
      "epoch": 0.7516681172033652,
      "grad_norm": 0.4180302321910858,
      "learning_rate": 1.2418666335115424e-06,
      "loss": 0.6379,
      "step": 18137
    },
    {
      "epoch": 0.751709561109039,
      "grad_norm": 0.4360416531562805,
      "learning_rate": 1.241659413983174e-06,
      "loss": 0.7087,
      "step": 18138
    },
    {
      "epoch": 0.7517510050147126,
      "grad_norm": 0.4283049404621124,
      "learning_rate": 1.2414521944548056e-06,
      "loss": 0.6364,
      "step": 18139
    },
    {
      "epoch": 0.7517924489203862,
      "grad_norm": 0.43910738825798035,
      "learning_rate": 1.2412449749264372e-06,
      "loss": 0.7266,
      "step": 18140
    },
    {
      "epoch": 0.75183389282606,
      "grad_norm": 0.42952442169189453,
      "learning_rate": 1.2410377553980688e-06,
      "loss": 0.686,
      "step": 18141
    },
    {
      "epoch": 0.7518753367317336,
      "grad_norm": 0.39954304695129395,
      "learning_rate": 1.2408305358697006e-06,
      "loss": 0.6956,
      "step": 18142
    },
    {
      "epoch": 0.7519167806374073,
      "grad_norm": 0.43185874819755554,
      "learning_rate": 1.2406233163413322e-06,
      "loss": 0.637,
      "step": 18143
    },
    {
      "epoch": 0.7519582245430809,
      "grad_norm": 0.420297235250473,
      "learning_rate": 1.2404160968129638e-06,
      "loss": 0.644,
      "step": 18144
    },
    {
      "epoch": 0.7519996684487547,
      "grad_norm": 0.4051879644393921,
      "learning_rate": 1.2402088772845954e-06,
      "loss": 0.6731,
      "step": 18145
    },
    {
      "epoch": 0.7520411123544283,
      "grad_norm": 0.4346986711025238,
      "learning_rate": 1.2400016577562272e-06,
      "loss": 0.7161,
      "step": 18146
    },
    {
      "epoch": 0.7520825562601019,
      "grad_norm": 0.39401689171791077,
      "learning_rate": 1.2397944382278588e-06,
      "loss": 0.6234,
      "step": 18147
    },
    {
      "epoch": 0.7521240001657756,
      "grad_norm": 0.4115256667137146,
      "learning_rate": 1.2395872186994904e-06,
      "loss": 0.6378,
      "step": 18148
    },
    {
      "epoch": 0.7521654440714493,
      "grad_norm": 0.41027265787124634,
      "learning_rate": 1.239379999171122e-06,
      "loss": 0.6699,
      "step": 18149
    },
    {
      "epoch": 0.752206887977123,
      "grad_norm": 0.42018914222717285,
      "learning_rate": 1.2391727796427538e-06,
      "loss": 0.6675,
      "step": 18150
    },
    {
      "epoch": 0.7522483318827966,
      "grad_norm": 0.4364220201969147,
      "learning_rate": 1.2389655601143854e-06,
      "loss": 0.6323,
      "step": 18151
    },
    {
      "epoch": 0.7522897757884703,
      "grad_norm": 0.3931707441806793,
      "learning_rate": 1.238758340586017e-06,
      "loss": 0.6407,
      "step": 18152
    },
    {
      "epoch": 0.752331219694144,
      "grad_norm": 0.4244634211063385,
      "learning_rate": 1.2385511210576486e-06,
      "loss": 0.6783,
      "step": 18153
    },
    {
      "epoch": 0.7523726635998177,
      "grad_norm": 0.4203733503818512,
      "learning_rate": 1.2383439015292802e-06,
      "loss": 0.6515,
      "step": 18154
    },
    {
      "epoch": 0.7524141075054913,
      "grad_norm": 0.4208974242210388,
      "learning_rate": 1.238136682000912e-06,
      "loss": 0.6531,
      "step": 18155
    },
    {
      "epoch": 0.7524555514111649,
      "grad_norm": 0.4043782353401184,
      "learning_rate": 1.2379294624725436e-06,
      "loss": 0.6619,
      "step": 18156
    },
    {
      "epoch": 0.7524969953168387,
      "grad_norm": 0.42378875613212585,
      "learning_rate": 1.2377222429441752e-06,
      "loss": 0.6334,
      "step": 18157
    },
    {
      "epoch": 0.7525384392225123,
      "grad_norm": 0.3850429654121399,
      "learning_rate": 1.2375150234158068e-06,
      "loss": 0.6434,
      "step": 18158
    },
    {
      "epoch": 0.752579883128186,
      "grad_norm": 0.43882259726524353,
      "learning_rate": 1.2373078038874386e-06,
      "loss": 0.678,
      "step": 18159
    },
    {
      "epoch": 0.7526213270338596,
      "grad_norm": 0.42201563715934753,
      "learning_rate": 1.2371005843590702e-06,
      "loss": 0.6846,
      "step": 18160
    },
    {
      "epoch": 0.7526627709395334,
      "grad_norm": 0.4154178500175476,
      "learning_rate": 1.2368933648307018e-06,
      "loss": 0.7286,
      "step": 18161
    },
    {
      "epoch": 0.752704214845207,
      "grad_norm": 0.3873840868473053,
      "learning_rate": 1.2366861453023334e-06,
      "loss": 0.6908,
      "step": 18162
    },
    {
      "epoch": 0.7527456587508807,
      "grad_norm": 0.4223342835903168,
      "learning_rate": 1.236478925773965e-06,
      "loss": 0.6902,
      "step": 18163
    },
    {
      "epoch": 0.7527871026565544,
      "grad_norm": 0.4386080205440521,
      "learning_rate": 1.2362717062455968e-06,
      "loss": 0.6694,
      "step": 18164
    },
    {
      "epoch": 0.752828546562228,
      "grad_norm": 0.4245038330554962,
      "learning_rate": 1.2360644867172284e-06,
      "loss": 0.6783,
      "step": 18165
    },
    {
      "epoch": 0.7528699904679017,
      "grad_norm": 0.4469040632247925,
      "learning_rate": 1.23585726718886e-06,
      "loss": 0.6506,
      "step": 18166
    },
    {
      "epoch": 0.7529114343735753,
      "grad_norm": 0.4272489547729492,
      "learning_rate": 1.2356500476604916e-06,
      "loss": 0.6946,
      "step": 18167
    },
    {
      "epoch": 0.7529528782792491,
      "grad_norm": 0.43511152267456055,
      "learning_rate": 1.2354428281321234e-06,
      "loss": 0.6968,
      "step": 18168
    },
    {
      "epoch": 0.7529943221849227,
      "grad_norm": 0.4406329393386841,
      "learning_rate": 1.235235608603755e-06,
      "loss": 0.7256,
      "step": 18169
    },
    {
      "epoch": 0.7530357660905964,
      "grad_norm": 0.3928154706954956,
      "learning_rate": 1.2350283890753866e-06,
      "loss": 0.6577,
      "step": 18170
    },
    {
      "epoch": 0.75307720999627,
      "grad_norm": 0.4237310588359833,
      "learning_rate": 1.2348211695470182e-06,
      "loss": 0.6931,
      "step": 18171
    },
    {
      "epoch": 0.7531186539019438,
      "grad_norm": 0.5531282424926758,
      "learning_rate": 1.23461395001865e-06,
      "loss": 0.7029,
      "step": 18172
    },
    {
      "epoch": 0.7531600978076174,
      "grad_norm": 0.4064716398715973,
      "learning_rate": 1.2344067304902816e-06,
      "loss": 0.6449,
      "step": 18173
    },
    {
      "epoch": 0.753201541713291,
      "grad_norm": 0.3712570369243622,
      "learning_rate": 1.2341995109619132e-06,
      "loss": 0.6521,
      "step": 18174
    },
    {
      "epoch": 0.7532429856189647,
      "grad_norm": 0.3940562903881073,
      "learning_rate": 1.2339922914335448e-06,
      "loss": 0.7268,
      "step": 18175
    },
    {
      "epoch": 0.7532844295246384,
      "grad_norm": 0.39350855350494385,
      "learning_rate": 1.2337850719051764e-06,
      "loss": 0.6851,
      "step": 18176
    },
    {
      "epoch": 0.7533258734303121,
      "grad_norm": 0.4416342079639435,
      "learning_rate": 1.2335778523768082e-06,
      "loss": 0.6606,
      "step": 18177
    },
    {
      "epoch": 0.7533673173359857,
      "grad_norm": 0.40940192341804504,
      "learning_rate": 1.2333706328484398e-06,
      "loss": 0.6306,
      "step": 18178
    },
    {
      "epoch": 0.7534087612416595,
      "grad_norm": 0.4198229908943176,
      "learning_rate": 1.2331634133200714e-06,
      "loss": 0.642,
      "step": 18179
    },
    {
      "epoch": 0.7534502051473331,
      "grad_norm": 0.4120827913284302,
      "learning_rate": 1.232956193791703e-06,
      "loss": 0.7017,
      "step": 18180
    },
    {
      "epoch": 0.7534916490530068,
      "grad_norm": 0.38771459460258484,
      "learning_rate": 1.2327489742633348e-06,
      "loss": 0.5988,
      "step": 18181
    },
    {
      "epoch": 0.7535330929586804,
      "grad_norm": 0.44402164220809937,
      "learning_rate": 1.2325417547349664e-06,
      "loss": 0.696,
      "step": 18182
    },
    {
      "epoch": 0.753574536864354,
      "grad_norm": 0.44455617666244507,
      "learning_rate": 1.232334535206598e-06,
      "loss": 0.6869,
      "step": 18183
    },
    {
      "epoch": 0.7536159807700278,
      "grad_norm": 0.39785236120224,
      "learning_rate": 1.2321273156782296e-06,
      "loss": 0.6753,
      "step": 18184
    },
    {
      "epoch": 0.7536574246757014,
      "grad_norm": 0.38366401195526123,
      "learning_rate": 1.2319200961498614e-06,
      "loss": 0.604,
      "step": 18185
    },
    {
      "epoch": 0.7536988685813751,
      "grad_norm": 0.4519854187965393,
      "learning_rate": 1.231712876621493e-06,
      "loss": 0.7439,
      "step": 18186
    },
    {
      "epoch": 0.7537403124870488,
      "grad_norm": 0.4359987676143646,
      "learning_rate": 1.2315056570931246e-06,
      "loss": 0.6548,
      "step": 18187
    },
    {
      "epoch": 0.7537817563927225,
      "grad_norm": 0.42344972491264343,
      "learning_rate": 1.2312984375647562e-06,
      "loss": 0.6278,
      "step": 18188
    },
    {
      "epoch": 0.7538232002983961,
      "grad_norm": 0.4253320097923279,
      "learning_rate": 1.2310912180363878e-06,
      "loss": 0.6747,
      "step": 18189
    },
    {
      "epoch": 0.7538646442040698,
      "grad_norm": 0.42661672830581665,
      "learning_rate": 1.2308839985080196e-06,
      "loss": 0.6863,
      "step": 18190
    },
    {
      "epoch": 0.7539060881097435,
      "grad_norm": 0.41609615087509155,
      "learning_rate": 1.2306767789796512e-06,
      "loss": 0.6807,
      "step": 18191
    },
    {
      "epoch": 0.7539475320154171,
      "grad_norm": 0.4082472622394562,
      "learning_rate": 1.2304695594512828e-06,
      "loss": 0.6409,
      "step": 18192
    },
    {
      "epoch": 0.7539889759210908,
      "grad_norm": 0.40783607959747314,
      "learning_rate": 1.2302623399229144e-06,
      "loss": 0.7031,
      "step": 18193
    },
    {
      "epoch": 0.7540304198267644,
      "grad_norm": 0.414442241191864,
      "learning_rate": 1.2300551203945462e-06,
      "loss": 0.6461,
      "step": 18194
    },
    {
      "epoch": 0.7540718637324382,
      "grad_norm": 0.42578059434890747,
      "learning_rate": 1.2298479008661778e-06,
      "loss": 0.7225,
      "step": 18195
    },
    {
      "epoch": 0.7541133076381118,
      "grad_norm": 0.42140722274780273,
      "learning_rate": 1.2296406813378094e-06,
      "loss": 0.668,
      "step": 18196
    },
    {
      "epoch": 0.7541547515437855,
      "grad_norm": 0.44172757863998413,
      "learning_rate": 1.229433461809441e-06,
      "loss": 0.7159,
      "step": 18197
    },
    {
      "epoch": 0.7541961954494592,
      "grad_norm": 0.4471127688884735,
      "learning_rate": 1.2292262422810728e-06,
      "loss": 0.6715,
      "step": 18198
    },
    {
      "epoch": 0.7542376393551329,
      "grad_norm": 0.48301398754119873,
      "learning_rate": 1.2290190227527044e-06,
      "loss": 0.7354,
      "step": 18199
    },
    {
      "epoch": 0.7542790832608065,
      "grad_norm": 0.38618719577789307,
      "learning_rate": 1.228811803224336e-06,
      "loss": 0.6012,
      "step": 18200
    },
    {
      "epoch": 0.7543205271664801,
      "grad_norm": 0.4431675374507904,
      "learning_rate": 1.2286045836959676e-06,
      "loss": 0.7437,
      "step": 18201
    },
    {
      "epoch": 0.7543619710721539,
      "grad_norm": 0.42908886075019836,
      "learning_rate": 1.2283973641675992e-06,
      "loss": 0.6815,
      "step": 18202
    },
    {
      "epoch": 0.7544034149778275,
      "grad_norm": 0.4171474874019623,
      "learning_rate": 1.228190144639231e-06,
      "loss": 0.6636,
      "step": 18203
    },
    {
      "epoch": 0.7544448588835012,
      "grad_norm": 0.43771809339523315,
      "learning_rate": 1.2279829251108626e-06,
      "loss": 0.6777,
      "step": 18204
    },
    {
      "epoch": 0.7544863027891748,
      "grad_norm": 0.3687293231487274,
      "learning_rate": 1.2277757055824942e-06,
      "loss": 0.6604,
      "step": 18205
    },
    {
      "epoch": 0.7545277466948486,
      "grad_norm": 0.40466344356536865,
      "learning_rate": 1.2275684860541258e-06,
      "loss": 0.6802,
      "step": 18206
    },
    {
      "epoch": 0.7545691906005222,
      "grad_norm": 0.4073166251182556,
      "learning_rate": 1.2273612665257576e-06,
      "loss": 0.6816,
      "step": 18207
    },
    {
      "epoch": 0.7546106345061958,
      "grad_norm": 0.4285399615764618,
      "learning_rate": 1.2271540469973892e-06,
      "loss": 0.7079,
      "step": 18208
    },
    {
      "epoch": 0.7546520784118695,
      "grad_norm": 0.4258655309677124,
      "learning_rate": 1.2269468274690208e-06,
      "loss": 0.6793,
      "step": 18209
    },
    {
      "epoch": 0.7546935223175432,
      "grad_norm": 0.4392007291316986,
      "learning_rate": 1.2267396079406524e-06,
      "loss": 0.6782,
      "step": 18210
    },
    {
      "epoch": 0.7547349662232169,
      "grad_norm": 0.4320068657398224,
      "learning_rate": 1.2265323884122842e-06,
      "loss": 0.6587,
      "step": 18211
    },
    {
      "epoch": 0.7547764101288905,
      "grad_norm": 0.4275197982788086,
      "learning_rate": 1.2263251688839158e-06,
      "loss": 0.6915,
      "step": 18212
    },
    {
      "epoch": 0.7548178540345643,
      "grad_norm": 0.4353669285774231,
      "learning_rate": 1.2261179493555474e-06,
      "loss": 0.6586,
      "step": 18213
    },
    {
      "epoch": 0.7548592979402379,
      "grad_norm": 0.39792516827583313,
      "learning_rate": 1.225910729827179e-06,
      "loss": 0.6119,
      "step": 18214
    },
    {
      "epoch": 0.7549007418459116,
      "grad_norm": 0.46582213044166565,
      "learning_rate": 1.2257035102988106e-06,
      "loss": 0.6675,
      "step": 18215
    },
    {
      "epoch": 0.7549421857515852,
      "grad_norm": 0.37943220138549805,
      "learning_rate": 1.2254962907704424e-06,
      "loss": 0.6373,
      "step": 18216
    },
    {
      "epoch": 0.7549836296572588,
      "grad_norm": 0.43685299158096313,
      "learning_rate": 1.225289071242074e-06,
      "loss": 0.6526,
      "step": 18217
    },
    {
      "epoch": 0.7550250735629326,
      "grad_norm": 0.40441015362739563,
      "learning_rate": 1.2250818517137056e-06,
      "loss": 0.626,
      "step": 18218
    },
    {
      "epoch": 0.7550665174686062,
      "grad_norm": 0.42112988233566284,
      "learning_rate": 1.2248746321853372e-06,
      "loss": 0.6433,
      "step": 18219
    },
    {
      "epoch": 0.7551079613742799,
      "grad_norm": 0.4789241552352905,
      "learning_rate": 1.224667412656969e-06,
      "loss": 0.7092,
      "step": 18220
    },
    {
      "epoch": 0.7551494052799536,
      "grad_norm": 0.4105861186981201,
      "learning_rate": 1.2244601931286006e-06,
      "loss": 0.6697,
      "step": 18221
    },
    {
      "epoch": 0.7551908491856273,
      "grad_norm": 0.3997950553894043,
      "learning_rate": 1.2242529736002322e-06,
      "loss": 0.6599,
      "step": 18222
    },
    {
      "epoch": 0.7552322930913009,
      "grad_norm": 0.4342069625854492,
      "learning_rate": 1.2240457540718638e-06,
      "loss": 0.7213,
      "step": 18223
    },
    {
      "epoch": 0.7552737369969746,
      "grad_norm": 0.4278947412967682,
      "learning_rate": 1.2238385345434954e-06,
      "loss": 0.6559,
      "step": 18224
    },
    {
      "epoch": 0.7553151809026483,
      "grad_norm": 0.4252760410308838,
      "learning_rate": 1.2236313150151272e-06,
      "loss": 0.6602,
      "step": 18225
    },
    {
      "epoch": 0.7553566248083219,
      "grad_norm": 0.3982548117637634,
      "learning_rate": 1.2234240954867588e-06,
      "loss": 0.6593,
      "step": 18226
    },
    {
      "epoch": 0.7553980687139956,
      "grad_norm": 0.4037231206893921,
      "learning_rate": 1.2232168759583904e-06,
      "loss": 0.6345,
      "step": 18227
    },
    {
      "epoch": 0.7554395126196692,
      "grad_norm": 0.4367417097091675,
      "learning_rate": 1.223009656430022e-06,
      "loss": 0.6672,
      "step": 18228
    },
    {
      "epoch": 0.755480956525343,
      "grad_norm": 0.40735286474227905,
      "learning_rate": 1.2228024369016538e-06,
      "loss": 0.6902,
      "step": 18229
    },
    {
      "epoch": 0.7555224004310166,
      "grad_norm": 0.397272527217865,
      "learning_rate": 1.2225952173732854e-06,
      "loss": 0.6533,
      "step": 18230
    },
    {
      "epoch": 0.7555638443366903,
      "grad_norm": 0.4441736042499542,
      "learning_rate": 1.222387997844917e-06,
      "loss": 0.6996,
      "step": 18231
    },
    {
      "epoch": 0.755605288242364,
      "grad_norm": 0.43984255194664,
      "learning_rate": 1.2221807783165486e-06,
      "loss": 0.6748,
      "step": 18232
    },
    {
      "epoch": 0.7556467321480377,
      "grad_norm": 0.4073004424571991,
      "learning_rate": 1.2219735587881804e-06,
      "loss": 0.6951,
      "step": 18233
    },
    {
      "epoch": 0.7556881760537113,
      "grad_norm": 0.4917154908180237,
      "learning_rate": 1.221766339259812e-06,
      "loss": 0.6804,
      "step": 18234
    },
    {
      "epoch": 0.7557296199593849,
      "grad_norm": 0.3972444534301758,
      "learning_rate": 1.2215591197314436e-06,
      "loss": 0.6576,
      "step": 18235
    },
    {
      "epoch": 0.7557710638650587,
      "grad_norm": 0.422237753868103,
      "learning_rate": 1.2213519002030752e-06,
      "loss": 0.6764,
      "step": 18236
    },
    {
      "epoch": 0.7558125077707323,
      "grad_norm": 0.37819766998291016,
      "learning_rate": 1.2211446806747068e-06,
      "loss": 0.6469,
      "step": 18237
    },
    {
      "epoch": 0.755853951676406,
      "grad_norm": 0.42419740557670593,
      "learning_rate": 1.2209374611463386e-06,
      "loss": 0.7075,
      "step": 18238
    },
    {
      "epoch": 0.7558953955820796,
      "grad_norm": 0.39829257130622864,
      "learning_rate": 1.2207302416179702e-06,
      "loss": 0.6418,
      "step": 18239
    },
    {
      "epoch": 0.7559368394877534,
      "grad_norm": 0.4384540319442749,
      "learning_rate": 1.2205230220896018e-06,
      "loss": 0.6525,
      "step": 18240
    },
    {
      "epoch": 0.755978283393427,
      "grad_norm": 0.3931398093700409,
      "learning_rate": 1.2203158025612334e-06,
      "loss": 0.6881,
      "step": 18241
    },
    {
      "epoch": 0.7560197272991007,
      "grad_norm": 0.4000490605831146,
      "learning_rate": 1.2201085830328652e-06,
      "loss": 0.6613,
      "step": 18242
    },
    {
      "epoch": 0.7560611712047743,
      "grad_norm": 0.4376785159111023,
      "learning_rate": 1.2199013635044968e-06,
      "loss": 0.6873,
      "step": 18243
    },
    {
      "epoch": 0.756102615110448,
      "grad_norm": 0.41143882274627686,
      "learning_rate": 1.2196941439761284e-06,
      "loss": 0.6169,
      "step": 18244
    },
    {
      "epoch": 0.7561440590161217,
      "grad_norm": 0.4073342978954315,
      "learning_rate": 1.21948692444776e-06,
      "loss": 0.663,
      "step": 18245
    },
    {
      "epoch": 0.7561855029217953,
      "grad_norm": 0.4308210015296936,
      "learning_rate": 1.2192797049193918e-06,
      "loss": 0.6985,
      "step": 18246
    },
    {
      "epoch": 0.756226946827469,
      "grad_norm": 0.3765134811401367,
      "learning_rate": 1.2190724853910234e-06,
      "loss": 0.6425,
      "step": 18247
    },
    {
      "epoch": 0.7562683907331427,
      "grad_norm": 0.4012081027030945,
      "learning_rate": 1.218865265862655e-06,
      "loss": 0.6855,
      "step": 18248
    },
    {
      "epoch": 0.7563098346388164,
      "grad_norm": 0.3964138329029083,
      "learning_rate": 1.2186580463342866e-06,
      "loss": 0.6771,
      "step": 18249
    },
    {
      "epoch": 0.75635127854449,
      "grad_norm": 0.41609707474708557,
      "learning_rate": 1.2184508268059182e-06,
      "loss": 0.6288,
      "step": 18250
    },
    {
      "epoch": 0.7563927224501638,
      "grad_norm": 0.40993911027908325,
      "learning_rate": 1.21824360727755e-06,
      "loss": 0.6721,
      "step": 18251
    },
    {
      "epoch": 0.7564341663558374,
      "grad_norm": 0.40015077590942383,
      "learning_rate": 1.2180363877491816e-06,
      "loss": 0.6396,
      "step": 18252
    },
    {
      "epoch": 0.756475610261511,
      "grad_norm": 0.4193427562713623,
      "learning_rate": 1.2178291682208132e-06,
      "loss": 0.6763,
      "step": 18253
    },
    {
      "epoch": 0.7565170541671847,
      "grad_norm": 0.3957098126411438,
      "learning_rate": 1.2176219486924448e-06,
      "loss": 0.6892,
      "step": 18254
    },
    {
      "epoch": 0.7565584980728584,
      "grad_norm": 0.42201468348503113,
      "learning_rate": 1.2174147291640766e-06,
      "loss": 0.6495,
      "step": 18255
    },
    {
      "epoch": 0.7565999419785321,
      "grad_norm": 0.4222988784313202,
      "learning_rate": 1.2172075096357082e-06,
      "loss": 0.6743,
      "step": 18256
    },
    {
      "epoch": 0.7566413858842057,
      "grad_norm": 0.40637075901031494,
      "learning_rate": 1.2170002901073398e-06,
      "loss": 0.6611,
      "step": 18257
    },
    {
      "epoch": 0.7566828297898794,
      "grad_norm": 0.42748862504959106,
      "learning_rate": 1.2167930705789714e-06,
      "loss": 0.674,
      "step": 18258
    },
    {
      "epoch": 0.7567242736955531,
      "grad_norm": 0.42896154522895813,
      "learning_rate": 1.2165858510506032e-06,
      "loss": 0.6821,
      "step": 18259
    },
    {
      "epoch": 0.7567657176012268,
      "grad_norm": 0.40537121891975403,
      "learning_rate": 1.2163786315222348e-06,
      "loss": 0.6561,
      "step": 18260
    },
    {
      "epoch": 0.7568071615069004,
      "grad_norm": 0.46956291794776917,
      "learning_rate": 1.2161714119938664e-06,
      "loss": 0.741,
      "step": 18261
    },
    {
      "epoch": 0.756848605412574,
      "grad_norm": 0.43676742911338806,
      "learning_rate": 1.215964192465498e-06,
      "loss": 0.6871,
      "step": 18262
    },
    {
      "epoch": 0.7568900493182478,
      "grad_norm": 0.47088801860809326,
      "learning_rate": 1.2157569729371296e-06,
      "loss": 0.7427,
      "step": 18263
    },
    {
      "epoch": 0.7569314932239214,
      "grad_norm": 0.40483951568603516,
      "learning_rate": 1.2155497534087614e-06,
      "loss": 0.7,
      "step": 18264
    },
    {
      "epoch": 0.7569729371295951,
      "grad_norm": 0.4142512083053589,
      "learning_rate": 1.215342533880393e-06,
      "loss": 0.679,
      "step": 18265
    },
    {
      "epoch": 0.7570143810352687,
      "grad_norm": 0.4070943593978882,
      "learning_rate": 1.2151353143520246e-06,
      "loss": 0.6826,
      "step": 18266
    },
    {
      "epoch": 0.7570558249409425,
      "grad_norm": 0.37281253933906555,
      "learning_rate": 1.2149280948236562e-06,
      "loss": 0.6375,
      "step": 18267
    },
    {
      "epoch": 0.7570972688466161,
      "grad_norm": 0.4313116669654846,
      "learning_rate": 1.214720875295288e-06,
      "loss": 0.6458,
      "step": 18268
    },
    {
      "epoch": 0.7571387127522897,
      "grad_norm": 0.4331933259963989,
      "learning_rate": 1.2145136557669196e-06,
      "loss": 0.6594,
      "step": 18269
    },
    {
      "epoch": 0.7571801566579635,
      "grad_norm": 0.4115048050880432,
      "learning_rate": 1.2143064362385512e-06,
      "loss": 0.677,
      "step": 18270
    },
    {
      "epoch": 0.7572216005636371,
      "grad_norm": 0.44158029556274414,
      "learning_rate": 1.2140992167101828e-06,
      "loss": 0.6884,
      "step": 18271
    },
    {
      "epoch": 0.7572630444693108,
      "grad_norm": 0.4029923677444458,
      "learning_rate": 1.2138919971818146e-06,
      "loss": 0.6719,
      "step": 18272
    },
    {
      "epoch": 0.7573044883749844,
      "grad_norm": 0.42345482110977173,
      "learning_rate": 1.2136847776534462e-06,
      "loss": 0.6584,
      "step": 18273
    },
    {
      "epoch": 0.7573459322806582,
      "grad_norm": 0.4061579406261444,
      "learning_rate": 1.2134775581250778e-06,
      "loss": 0.6648,
      "step": 18274
    },
    {
      "epoch": 0.7573873761863318,
      "grad_norm": 0.4210527241230011,
      "learning_rate": 1.2132703385967094e-06,
      "loss": 0.6815,
      "step": 18275
    },
    {
      "epoch": 0.7574288200920055,
      "grad_norm": 0.37955883145332336,
      "learning_rate": 1.213063119068341e-06,
      "loss": 0.6475,
      "step": 18276
    },
    {
      "epoch": 0.7574702639976791,
      "grad_norm": 0.40023282170295715,
      "learning_rate": 1.2128558995399728e-06,
      "loss": 0.6849,
      "step": 18277
    },
    {
      "epoch": 0.7575117079033528,
      "grad_norm": 0.40901604294776917,
      "learning_rate": 1.2126486800116044e-06,
      "loss": 0.687,
      "step": 18278
    },
    {
      "epoch": 0.7575531518090265,
      "grad_norm": 0.4351852834224701,
      "learning_rate": 1.212441460483236e-06,
      "loss": 0.6682,
      "step": 18279
    },
    {
      "epoch": 0.7575945957147001,
      "grad_norm": 0.40971407294273376,
      "learning_rate": 1.2122342409548676e-06,
      "loss": 0.6212,
      "step": 18280
    },
    {
      "epoch": 0.7576360396203738,
      "grad_norm": 0.42162537574768066,
      "learning_rate": 1.2120270214264994e-06,
      "loss": 0.691,
      "step": 18281
    },
    {
      "epoch": 0.7576774835260475,
      "grad_norm": 0.40328946709632874,
      "learning_rate": 1.211819801898131e-06,
      "loss": 0.6921,
      "step": 18282
    },
    {
      "epoch": 0.7577189274317212,
      "grad_norm": 0.4126608073711395,
      "learning_rate": 1.2116125823697626e-06,
      "loss": 0.6707,
      "step": 18283
    },
    {
      "epoch": 0.7577603713373948,
      "grad_norm": 0.4197458028793335,
      "learning_rate": 1.2114053628413942e-06,
      "loss": 0.6921,
      "step": 18284
    },
    {
      "epoch": 0.7578018152430686,
      "grad_norm": 0.40786877274513245,
      "learning_rate": 1.211198143313026e-06,
      "loss": 0.6714,
      "step": 18285
    },
    {
      "epoch": 0.7578432591487422,
      "grad_norm": 0.4516717493534088,
      "learning_rate": 1.2109909237846576e-06,
      "loss": 0.6602,
      "step": 18286
    },
    {
      "epoch": 0.7578847030544158,
      "grad_norm": 0.44697824120521545,
      "learning_rate": 1.2107837042562892e-06,
      "loss": 0.6938,
      "step": 18287
    },
    {
      "epoch": 0.7579261469600895,
      "grad_norm": 0.397945374250412,
      "learning_rate": 1.2105764847279208e-06,
      "loss": 0.6517,
      "step": 18288
    },
    {
      "epoch": 0.7579675908657632,
      "grad_norm": 0.4477408826351166,
      "learning_rate": 1.2103692651995524e-06,
      "loss": 0.7263,
      "step": 18289
    },
    {
      "epoch": 0.7580090347714369,
      "grad_norm": 0.40338435769081116,
      "learning_rate": 1.2101620456711842e-06,
      "loss": 0.6824,
      "step": 18290
    },
    {
      "epoch": 0.7580504786771105,
      "grad_norm": 0.46308666467666626,
      "learning_rate": 1.2099548261428158e-06,
      "loss": 0.6993,
      "step": 18291
    },
    {
      "epoch": 0.7580919225827842,
      "grad_norm": 0.4168059527873993,
      "learning_rate": 1.2097476066144474e-06,
      "loss": 0.634,
      "step": 18292
    },
    {
      "epoch": 0.7581333664884579,
      "grad_norm": 0.40961647033691406,
      "learning_rate": 1.209540387086079e-06,
      "loss": 0.655,
      "step": 18293
    },
    {
      "epoch": 0.7581748103941316,
      "grad_norm": 0.4179156720638275,
      "learning_rate": 1.2093331675577108e-06,
      "loss": 0.6958,
      "step": 18294
    },
    {
      "epoch": 0.7582162542998052,
      "grad_norm": 0.43189772963523865,
      "learning_rate": 1.2091259480293424e-06,
      "loss": 0.7048,
      "step": 18295
    },
    {
      "epoch": 0.7582576982054788,
      "grad_norm": 0.3966561555862427,
      "learning_rate": 1.208918728500974e-06,
      "loss": 0.6561,
      "step": 18296
    },
    {
      "epoch": 0.7582991421111526,
      "grad_norm": 0.4419609606266022,
      "learning_rate": 1.2087115089726056e-06,
      "loss": 0.6429,
      "step": 18297
    },
    {
      "epoch": 0.7583405860168262,
      "grad_norm": 0.3821149170398712,
      "learning_rate": 1.2085042894442372e-06,
      "loss": 0.6206,
      "step": 18298
    },
    {
      "epoch": 0.7583820299224999,
      "grad_norm": 0.3998735547065735,
      "learning_rate": 1.208297069915869e-06,
      "loss": 0.691,
      "step": 18299
    },
    {
      "epoch": 0.7584234738281735,
      "grad_norm": 0.40938466787338257,
      "learning_rate": 1.2080898503875006e-06,
      "loss": 0.6595,
      "step": 18300
    },
    {
      "epoch": 0.7584649177338473,
      "grad_norm": 0.3774470090866089,
      "learning_rate": 1.2078826308591322e-06,
      "loss": 0.6379,
      "step": 18301
    },
    {
      "epoch": 0.7585063616395209,
      "grad_norm": 0.39445528388023376,
      "learning_rate": 1.2076754113307638e-06,
      "loss": 0.6323,
      "step": 18302
    },
    {
      "epoch": 0.7585478055451946,
      "grad_norm": 0.3761560916900635,
      "learning_rate": 1.2074681918023956e-06,
      "loss": 0.5693,
      "step": 18303
    },
    {
      "epoch": 0.7585892494508683,
      "grad_norm": 0.4001929461956024,
      "learning_rate": 1.2072609722740272e-06,
      "loss": 0.6836,
      "step": 18304
    },
    {
      "epoch": 0.7586306933565419,
      "grad_norm": 0.42420458793640137,
      "learning_rate": 1.2070537527456588e-06,
      "loss": 0.6941,
      "step": 18305
    },
    {
      "epoch": 0.7586721372622156,
      "grad_norm": 0.44530537724494934,
      "learning_rate": 1.2068465332172904e-06,
      "loss": 0.6677,
      "step": 18306
    },
    {
      "epoch": 0.7587135811678892,
      "grad_norm": 0.408319354057312,
      "learning_rate": 1.2066393136889222e-06,
      "loss": 0.678,
      "step": 18307
    },
    {
      "epoch": 0.758755025073563,
      "grad_norm": 0.421370267868042,
      "learning_rate": 1.2064320941605538e-06,
      "loss": 0.7195,
      "step": 18308
    },
    {
      "epoch": 0.7587964689792366,
      "grad_norm": 0.37915274500846863,
      "learning_rate": 1.2062248746321854e-06,
      "loss": 0.6678,
      "step": 18309
    },
    {
      "epoch": 0.7588379128849103,
      "grad_norm": 0.46476539969444275,
      "learning_rate": 1.206017655103817e-06,
      "loss": 0.7148,
      "step": 18310
    },
    {
      "epoch": 0.7588793567905839,
      "grad_norm": 0.4160342216491699,
      "learning_rate": 1.2058104355754486e-06,
      "loss": 0.6646,
      "step": 18311
    },
    {
      "epoch": 0.7589208006962577,
      "grad_norm": 0.4177989661693573,
      "learning_rate": 1.2056032160470804e-06,
      "loss": 0.6849,
      "step": 18312
    },
    {
      "epoch": 0.7589622446019313,
      "grad_norm": 0.3905975818634033,
      "learning_rate": 1.205395996518712e-06,
      "loss": 0.6335,
      "step": 18313
    },
    {
      "epoch": 0.7590036885076049,
      "grad_norm": 0.40524470806121826,
      "learning_rate": 1.2051887769903436e-06,
      "loss": 0.6509,
      "step": 18314
    },
    {
      "epoch": 0.7590451324132786,
      "grad_norm": 0.401454359292984,
      "learning_rate": 1.2049815574619752e-06,
      "loss": 0.649,
      "step": 18315
    },
    {
      "epoch": 0.7590865763189523,
      "grad_norm": 0.42186182737350464,
      "learning_rate": 1.204774337933607e-06,
      "loss": 0.6375,
      "step": 18316
    },
    {
      "epoch": 0.759128020224626,
      "grad_norm": 0.41850799322128296,
      "learning_rate": 1.2045671184052386e-06,
      "loss": 0.7015,
      "step": 18317
    },
    {
      "epoch": 0.7591694641302996,
      "grad_norm": 0.45139169692993164,
      "learning_rate": 1.2043598988768702e-06,
      "loss": 0.6967,
      "step": 18318
    },
    {
      "epoch": 0.7592109080359734,
      "grad_norm": 0.4083717465400696,
      "learning_rate": 1.2041526793485018e-06,
      "loss": 0.6482,
      "step": 18319
    },
    {
      "epoch": 0.759252351941647,
      "grad_norm": 0.3886021077632904,
      "learning_rate": 1.2039454598201336e-06,
      "loss": 0.666,
      "step": 18320
    },
    {
      "epoch": 0.7592937958473206,
      "grad_norm": 0.4052678048610687,
      "learning_rate": 1.2037382402917652e-06,
      "loss": 0.6685,
      "step": 18321
    },
    {
      "epoch": 0.7593352397529943,
      "grad_norm": 0.41507044434547424,
      "learning_rate": 1.2035310207633968e-06,
      "loss": 0.6888,
      "step": 18322
    },
    {
      "epoch": 0.759376683658668,
      "grad_norm": 0.40326935052871704,
      "learning_rate": 1.2033238012350284e-06,
      "loss": 0.6558,
      "step": 18323
    },
    {
      "epoch": 0.7594181275643417,
      "grad_norm": 0.4302123188972473,
      "learning_rate": 1.20311658170666e-06,
      "loss": 0.7517,
      "step": 18324
    },
    {
      "epoch": 0.7594595714700153,
      "grad_norm": 0.4274311065673828,
      "learning_rate": 1.2029093621782918e-06,
      "loss": 0.6716,
      "step": 18325
    },
    {
      "epoch": 0.759501015375689,
      "grad_norm": 0.4138660728931427,
      "learning_rate": 1.2027021426499234e-06,
      "loss": 0.6858,
      "step": 18326
    },
    {
      "epoch": 0.7595424592813627,
      "grad_norm": 0.4584431052207947,
      "learning_rate": 1.202494923121555e-06,
      "loss": 0.6763,
      "step": 18327
    },
    {
      "epoch": 0.7595839031870364,
      "grad_norm": 0.4378434121608734,
      "learning_rate": 1.2022877035931866e-06,
      "loss": 0.7101,
      "step": 18328
    },
    {
      "epoch": 0.75962534709271,
      "grad_norm": 0.43136072158813477,
      "learning_rate": 1.2020804840648184e-06,
      "loss": 0.6812,
      "step": 18329
    },
    {
      "epoch": 0.7596667909983836,
      "grad_norm": 0.4279765486717224,
      "learning_rate": 1.20187326453645e-06,
      "loss": 0.6841,
      "step": 18330
    },
    {
      "epoch": 0.7597082349040574,
      "grad_norm": 0.41779157519340515,
      "learning_rate": 1.2016660450080816e-06,
      "loss": 0.7,
      "step": 18331
    },
    {
      "epoch": 0.759749678809731,
      "grad_norm": 0.40034160017967224,
      "learning_rate": 1.2014588254797132e-06,
      "loss": 0.6382,
      "step": 18332
    },
    {
      "epoch": 0.7597911227154047,
      "grad_norm": 0.4380475878715515,
      "learning_rate": 1.201251605951345e-06,
      "loss": 0.6812,
      "step": 18333
    },
    {
      "epoch": 0.7598325666210783,
      "grad_norm": 0.47501856088638306,
      "learning_rate": 1.2010443864229766e-06,
      "loss": 0.6825,
      "step": 18334
    },
    {
      "epoch": 0.7598740105267521,
      "grad_norm": 0.4194132685661316,
      "learning_rate": 1.2008371668946082e-06,
      "loss": 0.6776,
      "step": 18335
    },
    {
      "epoch": 0.7599154544324257,
      "grad_norm": 0.4407571852207184,
      "learning_rate": 1.2006299473662398e-06,
      "loss": 0.668,
      "step": 18336
    },
    {
      "epoch": 0.7599568983380994,
      "grad_norm": 0.42548617720603943,
      "learning_rate": 1.2004227278378714e-06,
      "loss": 0.6802,
      "step": 18337
    },
    {
      "epoch": 0.759998342243773,
      "grad_norm": 0.4193524122238159,
      "learning_rate": 1.2002155083095032e-06,
      "loss": 0.6853,
      "step": 18338
    },
    {
      "epoch": 0.7600397861494467,
      "grad_norm": 0.4022304117679596,
      "learning_rate": 1.2000082887811348e-06,
      "loss": 0.6377,
      "step": 18339
    },
    {
      "epoch": 0.7600812300551204,
      "grad_norm": 0.4295608401298523,
      "learning_rate": 1.1998010692527664e-06,
      "loss": 0.7039,
      "step": 18340
    },
    {
      "epoch": 0.760122673960794,
      "grad_norm": 0.39926278591156006,
      "learning_rate": 1.199593849724398e-06,
      "loss": 0.6918,
      "step": 18341
    },
    {
      "epoch": 0.7601641178664678,
      "grad_norm": 0.38446712493896484,
      "learning_rate": 1.1993866301960298e-06,
      "loss": 0.6587,
      "step": 18342
    },
    {
      "epoch": 0.7602055617721414,
      "grad_norm": 0.41846922039985657,
      "learning_rate": 1.1991794106676614e-06,
      "loss": 0.6566,
      "step": 18343
    },
    {
      "epoch": 0.7602470056778151,
      "grad_norm": 0.4314047694206238,
      "learning_rate": 1.198972191139293e-06,
      "loss": 0.73,
      "step": 18344
    },
    {
      "epoch": 0.7602884495834887,
      "grad_norm": 0.4201318919658661,
      "learning_rate": 1.1987649716109246e-06,
      "loss": 0.6624,
      "step": 18345
    },
    {
      "epoch": 0.7603298934891625,
      "grad_norm": 0.45999568700790405,
      "learning_rate": 1.1985577520825564e-06,
      "loss": 0.6759,
      "step": 18346
    },
    {
      "epoch": 0.7603713373948361,
      "grad_norm": 0.4237552881240845,
      "learning_rate": 1.198350532554188e-06,
      "loss": 0.6609,
      "step": 18347
    },
    {
      "epoch": 0.7604127813005097,
      "grad_norm": 0.40629950165748596,
      "learning_rate": 1.1981433130258196e-06,
      "loss": 0.7065,
      "step": 18348
    },
    {
      "epoch": 0.7604542252061834,
      "grad_norm": 0.43076226115226746,
      "learning_rate": 1.1979360934974512e-06,
      "loss": 0.6921,
      "step": 18349
    },
    {
      "epoch": 0.7604956691118571,
      "grad_norm": 0.46152153611183167,
      "learning_rate": 1.1977288739690828e-06,
      "loss": 0.7126,
      "step": 18350
    },
    {
      "epoch": 0.7605371130175308,
      "grad_norm": 0.40662476420402527,
      "learning_rate": 1.1975216544407146e-06,
      "loss": 0.6907,
      "step": 18351
    },
    {
      "epoch": 0.7605785569232044,
      "grad_norm": 0.39291083812713623,
      "learning_rate": 1.1973144349123462e-06,
      "loss": 0.6345,
      "step": 18352
    },
    {
      "epoch": 0.7606200008288782,
      "grad_norm": 0.41597530245780945,
      "learning_rate": 1.1971072153839778e-06,
      "loss": 0.6819,
      "step": 18353
    },
    {
      "epoch": 0.7606614447345518,
      "grad_norm": 0.3715381622314453,
      "learning_rate": 1.1968999958556094e-06,
      "loss": 0.6166,
      "step": 18354
    },
    {
      "epoch": 0.7607028886402255,
      "grad_norm": 0.3759414851665497,
      "learning_rate": 1.1966927763272412e-06,
      "loss": 0.6288,
      "step": 18355
    },
    {
      "epoch": 0.7607443325458991,
      "grad_norm": 0.4244822859764099,
      "learning_rate": 1.1964855567988728e-06,
      "loss": 0.6931,
      "step": 18356
    },
    {
      "epoch": 0.7607857764515727,
      "grad_norm": 0.41699737310409546,
      "learning_rate": 1.1962783372705044e-06,
      "loss": 0.7198,
      "step": 18357
    },
    {
      "epoch": 0.7608272203572465,
      "grad_norm": 0.38307371735572815,
      "learning_rate": 1.196071117742136e-06,
      "loss": 0.6262,
      "step": 18358
    },
    {
      "epoch": 0.7608686642629201,
      "grad_norm": 0.4011707007884979,
      "learning_rate": 1.1958638982137678e-06,
      "loss": 0.658,
      "step": 18359
    },
    {
      "epoch": 0.7609101081685938,
      "grad_norm": 0.43034785985946655,
      "learning_rate": 1.1956566786853994e-06,
      "loss": 0.7031,
      "step": 18360
    },
    {
      "epoch": 0.7609515520742675,
      "grad_norm": 0.43973734974861145,
      "learning_rate": 1.195449459157031e-06,
      "loss": 0.6937,
      "step": 18361
    },
    {
      "epoch": 0.7609929959799412,
      "grad_norm": 0.38501331210136414,
      "learning_rate": 1.1952422396286626e-06,
      "loss": 0.6895,
      "step": 18362
    },
    {
      "epoch": 0.7610344398856148,
      "grad_norm": 0.4142957925796509,
      "learning_rate": 1.1950350201002944e-06,
      "loss": 0.6885,
      "step": 18363
    },
    {
      "epoch": 0.7610758837912885,
      "grad_norm": 0.4294329285621643,
      "learning_rate": 1.194827800571926e-06,
      "loss": 0.7302,
      "step": 18364
    },
    {
      "epoch": 0.7611173276969622,
      "grad_norm": 0.4126407504081726,
      "learning_rate": 1.1946205810435576e-06,
      "loss": 0.6763,
      "step": 18365
    },
    {
      "epoch": 0.7611587716026358,
      "grad_norm": 0.40298980474472046,
      "learning_rate": 1.1944133615151892e-06,
      "loss": 0.676,
      "step": 18366
    },
    {
      "epoch": 0.7612002155083095,
      "grad_norm": 0.40391093492507935,
      "learning_rate": 1.194206141986821e-06,
      "loss": 0.6821,
      "step": 18367
    },
    {
      "epoch": 0.7612416594139831,
      "grad_norm": 0.43327081203460693,
      "learning_rate": 1.1939989224584526e-06,
      "loss": 0.6842,
      "step": 18368
    },
    {
      "epoch": 0.7612831033196569,
      "grad_norm": 0.4301246702671051,
      "learning_rate": 1.1937917029300842e-06,
      "loss": 0.6619,
      "step": 18369
    },
    {
      "epoch": 0.7613245472253305,
      "grad_norm": 0.4142049252986908,
      "learning_rate": 1.1935844834017158e-06,
      "loss": 0.6592,
      "step": 18370
    },
    {
      "epoch": 0.7613659911310042,
      "grad_norm": 0.4257356822490692,
      "learning_rate": 1.1933772638733474e-06,
      "loss": 0.6987,
      "step": 18371
    },
    {
      "epoch": 0.7614074350366778,
      "grad_norm": 0.4078195095062256,
      "learning_rate": 1.1931700443449792e-06,
      "loss": 0.7117,
      "step": 18372
    },
    {
      "epoch": 0.7614488789423516,
      "grad_norm": 0.4216165542602539,
      "learning_rate": 1.1929628248166108e-06,
      "loss": 0.7107,
      "step": 18373
    },
    {
      "epoch": 0.7614903228480252,
      "grad_norm": 0.4060041606426239,
      "learning_rate": 1.1927556052882424e-06,
      "loss": 0.6427,
      "step": 18374
    },
    {
      "epoch": 0.7615317667536988,
      "grad_norm": 0.44099846482276917,
      "learning_rate": 1.192548385759874e-06,
      "loss": 0.7004,
      "step": 18375
    },
    {
      "epoch": 0.7615732106593726,
      "grad_norm": 0.4212326407432556,
      "learning_rate": 1.1923411662315058e-06,
      "loss": 0.7178,
      "step": 18376
    },
    {
      "epoch": 0.7616146545650462,
      "grad_norm": 0.4202856123447418,
      "learning_rate": 1.1921339467031374e-06,
      "loss": 0.6892,
      "step": 18377
    },
    {
      "epoch": 0.7616560984707199,
      "grad_norm": 0.43056002259254456,
      "learning_rate": 1.191926727174769e-06,
      "loss": 0.7247,
      "step": 18378
    },
    {
      "epoch": 0.7616975423763935,
      "grad_norm": 0.4548676311969757,
      "learning_rate": 1.1917195076464006e-06,
      "loss": 0.6635,
      "step": 18379
    },
    {
      "epoch": 0.7617389862820673,
      "grad_norm": 0.4552077054977417,
      "learning_rate": 1.1915122881180324e-06,
      "loss": 0.7075,
      "step": 18380
    },
    {
      "epoch": 0.7617804301877409,
      "grad_norm": 0.40756136178970337,
      "learning_rate": 1.191305068589664e-06,
      "loss": 0.6361,
      "step": 18381
    },
    {
      "epoch": 0.7618218740934145,
      "grad_norm": 0.4407983422279358,
      "learning_rate": 1.1910978490612956e-06,
      "loss": 0.6906,
      "step": 18382
    },
    {
      "epoch": 0.7618633179990882,
      "grad_norm": 0.40262487530708313,
      "learning_rate": 1.1908906295329272e-06,
      "loss": 0.6904,
      "step": 18383
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 0.38154393434524536,
      "learning_rate": 1.190683410004559e-06,
      "loss": 0.6598,
      "step": 18384
    },
    {
      "epoch": 0.7619462058104356,
      "grad_norm": 0.3998153805732727,
      "learning_rate": 1.1904761904761906e-06,
      "loss": 0.6675,
      "step": 18385
    },
    {
      "epoch": 0.7619876497161092,
      "grad_norm": 0.4057264029979706,
      "learning_rate": 1.1902689709478222e-06,
      "loss": 0.6885,
      "step": 18386
    },
    {
      "epoch": 0.762029093621783,
      "grad_norm": 0.40950122475624084,
      "learning_rate": 1.1900617514194538e-06,
      "loss": 0.6589,
      "step": 18387
    },
    {
      "epoch": 0.7620705375274566,
      "grad_norm": 0.43461716175079346,
      "learning_rate": 1.1898545318910854e-06,
      "loss": 0.6902,
      "step": 18388
    },
    {
      "epoch": 0.7621119814331303,
      "grad_norm": 0.4395633935928345,
      "learning_rate": 1.1896473123627172e-06,
      "loss": 0.6974,
      "step": 18389
    },
    {
      "epoch": 0.7621534253388039,
      "grad_norm": 0.3979886770248413,
      "learning_rate": 1.1894400928343488e-06,
      "loss": 0.6149,
      "step": 18390
    },
    {
      "epoch": 0.7621948692444775,
      "grad_norm": 0.48363152146339417,
      "learning_rate": 1.1892328733059804e-06,
      "loss": 0.7729,
      "step": 18391
    },
    {
      "epoch": 0.7622363131501513,
      "grad_norm": 0.43511006236076355,
      "learning_rate": 1.189025653777612e-06,
      "loss": 0.6926,
      "step": 18392
    },
    {
      "epoch": 0.7622777570558249,
      "grad_norm": 0.41377049684524536,
      "learning_rate": 1.1888184342492438e-06,
      "loss": 0.6338,
      "step": 18393
    },
    {
      "epoch": 0.7623192009614986,
      "grad_norm": 0.43079859018325806,
      "learning_rate": 1.1886112147208754e-06,
      "loss": 0.7416,
      "step": 18394
    },
    {
      "epoch": 0.7623606448671723,
      "grad_norm": 0.4158470034599304,
      "learning_rate": 1.188403995192507e-06,
      "loss": 0.684,
      "step": 18395
    },
    {
      "epoch": 0.762402088772846,
      "grad_norm": 0.3965252637863159,
      "learning_rate": 1.1881967756641386e-06,
      "loss": 0.6682,
      "step": 18396
    },
    {
      "epoch": 0.7624435326785196,
      "grad_norm": 0.4280250668525696,
      "learning_rate": 1.1879895561357704e-06,
      "loss": 0.7129,
      "step": 18397
    },
    {
      "epoch": 0.7624849765841933,
      "grad_norm": 0.3984662592411041,
      "learning_rate": 1.187782336607402e-06,
      "loss": 0.7018,
      "step": 18398
    },
    {
      "epoch": 0.762526420489867,
      "grad_norm": 0.40761762857437134,
      "learning_rate": 1.1875751170790336e-06,
      "loss": 0.6538,
      "step": 18399
    },
    {
      "epoch": 0.7625678643955406,
      "grad_norm": 0.4361613988876343,
      "learning_rate": 1.1873678975506652e-06,
      "loss": 0.6909,
      "step": 18400
    },
    {
      "epoch": 0.7626093083012143,
      "grad_norm": 0.3918582797050476,
      "learning_rate": 1.187160678022297e-06,
      "loss": 0.6599,
      "step": 18401
    },
    {
      "epoch": 0.7626507522068879,
      "grad_norm": 0.39397314190864563,
      "learning_rate": 1.1869534584939286e-06,
      "loss": 0.6316,
      "step": 18402
    },
    {
      "epoch": 0.7626921961125617,
      "grad_norm": 0.4001719653606415,
      "learning_rate": 1.1867462389655602e-06,
      "loss": 0.6482,
      "step": 18403
    },
    {
      "epoch": 0.7627336400182353,
      "grad_norm": 0.39547988772392273,
      "learning_rate": 1.1865390194371918e-06,
      "loss": 0.615,
      "step": 18404
    },
    {
      "epoch": 0.762775083923909,
      "grad_norm": 0.3844884932041168,
      "learning_rate": 1.1863317999088234e-06,
      "loss": 0.6655,
      "step": 18405
    },
    {
      "epoch": 0.7628165278295826,
      "grad_norm": 0.41524896025657654,
      "learning_rate": 1.1861245803804552e-06,
      "loss": 0.6772,
      "step": 18406
    },
    {
      "epoch": 0.7628579717352564,
      "grad_norm": 0.42783123254776,
      "learning_rate": 1.1859173608520868e-06,
      "loss": 0.6722,
      "step": 18407
    },
    {
      "epoch": 0.76289941564093,
      "grad_norm": 0.43278905749320984,
      "learning_rate": 1.1857101413237184e-06,
      "loss": 0.6589,
      "step": 18408
    },
    {
      "epoch": 0.7629408595466036,
      "grad_norm": 0.4240676760673523,
      "learning_rate": 1.18550292179535e-06,
      "loss": 0.6602,
      "step": 18409
    },
    {
      "epoch": 0.7629823034522774,
      "grad_norm": 0.41042715311050415,
      "learning_rate": 1.1852957022669818e-06,
      "loss": 0.6708,
      "step": 18410
    },
    {
      "epoch": 0.763023747357951,
      "grad_norm": 0.39893192052841187,
      "learning_rate": 1.1850884827386134e-06,
      "loss": 0.6569,
      "step": 18411
    },
    {
      "epoch": 0.7630651912636247,
      "grad_norm": 0.41617295145988464,
      "learning_rate": 1.184881263210245e-06,
      "loss": 0.6455,
      "step": 18412
    },
    {
      "epoch": 0.7631066351692983,
      "grad_norm": 0.42428210377693176,
      "learning_rate": 1.1846740436818766e-06,
      "loss": 0.698,
      "step": 18413
    },
    {
      "epoch": 0.7631480790749721,
      "grad_norm": 0.44222119450569153,
      "learning_rate": 1.1844668241535084e-06,
      "loss": 0.6927,
      "step": 18414
    },
    {
      "epoch": 0.7631895229806457,
      "grad_norm": 0.43572643399238586,
      "learning_rate": 1.18425960462514e-06,
      "loss": 0.7134,
      "step": 18415
    },
    {
      "epoch": 0.7632309668863194,
      "grad_norm": 0.42910078167915344,
      "learning_rate": 1.1840523850967716e-06,
      "loss": 0.7007,
      "step": 18416
    },
    {
      "epoch": 0.763272410791993,
      "grad_norm": 0.41192805767059326,
      "learning_rate": 1.1838451655684032e-06,
      "loss": 0.6672,
      "step": 18417
    },
    {
      "epoch": 0.7633138546976667,
      "grad_norm": 0.4022907316684723,
      "learning_rate": 1.183637946040035e-06,
      "loss": 0.6373,
      "step": 18418
    },
    {
      "epoch": 0.7633552986033404,
      "grad_norm": 0.4238658547401428,
      "learning_rate": 1.1834307265116666e-06,
      "loss": 0.6873,
      "step": 18419
    },
    {
      "epoch": 0.763396742509014,
      "grad_norm": 0.40032732486724854,
      "learning_rate": 1.1832235069832982e-06,
      "loss": 0.6631,
      "step": 18420
    },
    {
      "epoch": 0.7634381864146877,
      "grad_norm": 0.4157845377922058,
      "learning_rate": 1.1830162874549298e-06,
      "loss": 0.6398,
      "step": 18421
    },
    {
      "epoch": 0.7634796303203614,
      "grad_norm": 0.4107213616371155,
      "learning_rate": 1.1828090679265616e-06,
      "loss": 0.7031,
      "step": 18422
    },
    {
      "epoch": 0.7635210742260351,
      "grad_norm": 0.444105327129364,
      "learning_rate": 1.1826018483981932e-06,
      "loss": 0.718,
      "step": 18423
    },
    {
      "epoch": 0.7635625181317087,
      "grad_norm": 0.4231577515602112,
      "learning_rate": 1.1823946288698248e-06,
      "loss": 0.6942,
      "step": 18424
    },
    {
      "epoch": 0.7636039620373825,
      "grad_norm": 0.4133392572402954,
      "learning_rate": 1.1821874093414564e-06,
      "loss": 0.6575,
      "step": 18425
    },
    {
      "epoch": 0.7636454059430561,
      "grad_norm": 0.4107622802257538,
      "learning_rate": 1.181980189813088e-06,
      "loss": 0.6644,
      "step": 18426
    },
    {
      "epoch": 0.7636868498487297,
      "grad_norm": 0.43001824617385864,
      "learning_rate": 1.1817729702847198e-06,
      "loss": 0.6426,
      "step": 18427
    },
    {
      "epoch": 0.7637282937544034,
      "grad_norm": 0.4136884808540344,
      "learning_rate": 1.1815657507563514e-06,
      "loss": 0.7131,
      "step": 18428
    },
    {
      "epoch": 0.763769737660077,
      "grad_norm": 0.3851294219493866,
      "learning_rate": 1.181358531227983e-06,
      "loss": 0.6063,
      "step": 18429
    },
    {
      "epoch": 0.7638111815657508,
      "grad_norm": 0.4164036214351654,
      "learning_rate": 1.1811513116996146e-06,
      "loss": 0.7271,
      "step": 18430
    },
    {
      "epoch": 0.7638526254714244,
      "grad_norm": 0.4233842194080353,
      "learning_rate": 1.1809440921712464e-06,
      "loss": 0.6768,
      "step": 18431
    },
    {
      "epoch": 0.7638940693770981,
      "grad_norm": 0.4609467089176178,
      "learning_rate": 1.180736872642878e-06,
      "loss": 0.6528,
      "step": 18432
    },
    {
      "epoch": 0.7639355132827718,
      "grad_norm": 0.42858242988586426,
      "learning_rate": 1.1805296531145096e-06,
      "loss": 0.6564,
      "step": 18433
    },
    {
      "epoch": 0.7639769571884455,
      "grad_norm": 0.4150758385658264,
      "learning_rate": 1.1803224335861412e-06,
      "loss": 0.6674,
      "step": 18434
    },
    {
      "epoch": 0.7640184010941191,
      "grad_norm": 0.4111219346523285,
      "learning_rate": 1.180115214057773e-06,
      "loss": 0.6182,
      "step": 18435
    },
    {
      "epoch": 0.7640598449997927,
      "grad_norm": 0.413320928812027,
      "learning_rate": 1.1799079945294046e-06,
      "loss": 0.6726,
      "step": 18436
    },
    {
      "epoch": 0.7641012889054665,
      "grad_norm": 0.4055173397064209,
      "learning_rate": 1.1797007750010362e-06,
      "loss": 0.6526,
      "step": 18437
    },
    {
      "epoch": 0.7641427328111401,
      "grad_norm": 0.4086606800556183,
      "learning_rate": 1.1794935554726678e-06,
      "loss": 0.6664,
      "step": 18438
    },
    {
      "epoch": 0.7641841767168138,
      "grad_norm": 0.43589872121810913,
      "learning_rate": 1.1792863359442996e-06,
      "loss": 0.7075,
      "step": 18439
    },
    {
      "epoch": 0.7642256206224874,
      "grad_norm": 0.40485116839408875,
      "learning_rate": 1.1790791164159312e-06,
      "loss": 0.6895,
      "step": 18440
    },
    {
      "epoch": 0.7642670645281612,
      "grad_norm": 0.42566031217575073,
      "learning_rate": 1.1788718968875628e-06,
      "loss": 0.7073,
      "step": 18441
    },
    {
      "epoch": 0.7643085084338348,
      "grad_norm": 0.42485877871513367,
      "learning_rate": 1.1786646773591944e-06,
      "loss": 0.6808,
      "step": 18442
    },
    {
      "epoch": 0.7643499523395084,
      "grad_norm": 0.3711492419242859,
      "learning_rate": 1.178457457830826e-06,
      "loss": 0.5986,
      "step": 18443
    },
    {
      "epoch": 0.7643913962451822,
      "grad_norm": 0.43041500449180603,
      "learning_rate": 1.1782502383024578e-06,
      "loss": 0.6887,
      "step": 18444
    },
    {
      "epoch": 0.7644328401508558,
      "grad_norm": 0.4731006324291229,
      "learning_rate": 1.1780430187740894e-06,
      "loss": 0.665,
      "step": 18445
    },
    {
      "epoch": 0.7644742840565295,
      "grad_norm": 0.42910102009773254,
      "learning_rate": 1.177835799245721e-06,
      "loss": 0.6924,
      "step": 18446
    },
    {
      "epoch": 0.7645157279622031,
      "grad_norm": 0.455193430185318,
      "learning_rate": 1.1776285797173526e-06,
      "loss": 0.6975,
      "step": 18447
    },
    {
      "epoch": 0.7645571718678769,
      "grad_norm": 0.42143282294273376,
      "learning_rate": 1.1774213601889844e-06,
      "loss": 0.6909,
      "step": 18448
    },
    {
      "epoch": 0.7645986157735505,
      "grad_norm": 0.38527652621269226,
      "learning_rate": 1.177214140660616e-06,
      "loss": 0.6627,
      "step": 18449
    },
    {
      "epoch": 0.7646400596792242,
      "grad_norm": 0.3927207291126251,
      "learning_rate": 1.1770069211322476e-06,
      "loss": 0.6536,
      "step": 18450
    },
    {
      "epoch": 0.7646815035848978,
      "grad_norm": 0.4658842980861664,
      "learning_rate": 1.1767997016038792e-06,
      "loss": 0.6792,
      "step": 18451
    },
    {
      "epoch": 0.7647229474905715,
      "grad_norm": 0.41170987486839294,
      "learning_rate": 1.176592482075511e-06,
      "loss": 0.6454,
      "step": 18452
    },
    {
      "epoch": 0.7647643913962452,
      "grad_norm": 0.44729188084602356,
      "learning_rate": 1.1763852625471426e-06,
      "loss": 0.6562,
      "step": 18453
    },
    {
      "epoch": 0.7648058353019188,
      "grad_norm": 0.4143582880496979,
      "learning_rate": 1.1761780430187742e-06,
      "loss": 0.6487,
      "step": 18454
    },
    {
      "epoch": 0.7648472792075925,
      "grad_norm": 0.41327160596847534,
      "learning_rate": 1.1759708234904058e-06,
      "loss": 0.6753,
      "step": 18455
    },
    {
      "epoch": 0.7648887231132662,
      "grad_norm": 0.4269656836986542,
      "learning_rate": 1.1757636039620376e-06,
      "loss": 0.7073,
      "step": 18456
    },
    {
      "epoch": 0.7649301670189399,
      "grad_norm": 0.38702014088630676,
      "learning_rate": 1.1755563844336692e-06,
      "loss": 0.6414,
      "step": 18457
    },
    {
      "epoch": 0.7649716109246135,
      "grad_norm": 0.3879673182964325,
      "learning_rate": 1.1753491649053008e-06,
      "loss": 0.644,
      "step": 18458
    },
    {
      "epoch": 0.7650130548302873,
      "grad_norm": 0.39750558137893677,
      "learning_rate": 1.1751419453769324e-06,
      "loss": 0.6602,
      "step": 18459
    },
    {
      "epoch": 0.7650544987359609,
      "grad_norm": 0.395530104637146,
      "learning_rate": 1.1749347258485642e-06,
      "loss": 0.6448,
      "step": 18460
    },
    {
      "epoch": 0.7650959426416345,
      "grad_norm": 0.42407405376434326,
      "learning_rate": 1.1747275063201958e-06,
      "loss": 0.6112,
      "step": 18461
    },
    {
      "epoch": 0.7651373865473082,
      "grad_norm": 0.43486905097961426,
      "learning_rate": 1.1745202867918274e-06,
      "loss": 0.6525,
      "step": 18462
    },
    {
      "epoch": 0.7651788304529819,
      "grad_norm": 0.4844317138195038,
      "learning_rate": 1.174313067263459e-06,
      "loss": 0.7026,
      "step": 18463
    },
    {
      "epoch": 0.7652202743586556,
      "grad_norm": 0.38536015152931213,
      "learning_rate": 1.1741058477350906e-06,
      "loss": 0.6443,
      "step": 18464
    },
    {
      "epoch": 0.7652617182643292,
      "grad_norm": 0.3893115520477295,
      "learning_rate": 1.1738986282067224e-06,
      "loss": 0.6158,
      "step": 18465
    },
    {
      "epoch": 0.7653031621700029,
      "grad_norm": 0.4041883945465088,
      "learning_rate": 1.173691408678354e-06,
      "loss": 0.6658,
      "step": 18466
    },
    {
      "epoch": 0.7653446060756766,
      "grad_norm": 0.4559694826602936,
      "learning_rate": 1.1734841891499856e-06,
      "loss": 0.6636,
      "step": 18467
    },
    {
      "epoch": 0.7653860499813503,
      "grad_norm": 0.37963739037513733,
      "learning_rate": 1.1732769696216172e-06,
      "loss": 0.6163,
      "step": 18468
    },
    {
      "epoch": 0.7654274938870239,
      "grad_norm": 0.3712363541126251,
      "learning_rate": 1.173069750093249e-06,
      "loss": 0.6195,
      "step": 18469
    },
    {
      "epoch": 0.7654689377926975,
      "grad_norm": 0.4640085995197296,
      "learning_rate": 1.1728625305648806e-06,
      "loss": 0.6665,
      "step": 18470
    },
    {
      "epoch": 0.7655103816983713,
      "grad_norm": 0.4123537540435791,
      "learning_rate": 1.1726553110365122e-06,
      "loss": 0.6627,
      "step": 18471
    },
    {
      "epoch": 0.7655518256040449,
      "grad_norm": 0.3673608601093292,
      "learning_rate": 1.1724480915081438e-06,
      "loss": 0.6521,
      "step": 18472
    },
    {
      "epoch": 0.7655932695097186,
      "grad_norm": 0.39640936255455017,
      "learning_rate": 1.1722408719797756e-06,
      "loss": 0.6624,
      "step": 18473
    },
    {
      "epoch": 0.7656347134153922,
      "grad_norm": 0.37369483709335327,
      "learning_rate": 1.1720336524514072e-06,
      "loss": 0.6525,
      "step": 18474
    },
    {
      "epoch": 0.765676157321066,
      "grad_norm": 0.38504764437675476,
      "learning_rate": 1.1718264329230388e-06,
      "loss": 0.6244,
      "step": 18475
    },
    {
      "epoch": 0.7657176012267396,
      "grad_norm": 0.4304869472980499,
      "learning_rate": 1.1716192133946704e-06,
      "loss": 0.6904,
      "step": 18476
    },
    {
      "epoch": 0.7657590451324133,
      "grad_norm": 0.37363943457603455,
      "learning_rate": 1.1714119938663022e-06,
      "loss": 0.6263,
      "step": 18477
    },
    {
      "epoch": 0.765800489038087,
      "grad_norm": 0.3924352526664734,
      "learning_rate": 1.1712047743379338e-06,
      "loss": 0.6182,
      "step": 18478
    },
    {
      "epoch": 0.7658419329437606,
      "grad_norm": 0.406818151473999,
      "learning_rate": 1.1709975548095654e-06,
      "loss": 0.6556,
      "step": 18479
    },
    {
      "epoch": 0.7658833768494343,
      "grad_norm": 0.3982245922088623,
      "learning_rate": 1.170790335281197e-06,
      "loss": 0.6305,
      "step": 18480
    },
    {
      "epoch": 0.7659248207551079,
      "grad_norm": 0.4387570321559906,
      "learning_rate": 1.1705831157528286e-06,
      "loss": 0.6798,
      "step": 18481
    },
    {
      "epoch": 0.7659662646607817,
      "grad_norm": 0.40468600392341614,
      "learning_rate": 1.1703758962244604e-06,
      "loss": 0.6432,
      "step": 18482
    },
    {
      "epoch": 0.7660077085664553,
      "grad_norm": 0.4335421919822693,
      "learning_rate": 1.170168676696092e-06,
      "loss": 0.6388,
      "step": 18483
    },
    {
      "epoch": 0.766049152472129,
      "grad_norm": 0.4113207459449768,
      "learning_rate": 1.1699614571677236e-06,
      "loss": 0.6445,
      "step": 18484
    },
    {
      "epoch": 0.7660905963778026,
      "grad_norm": 0.41679611802101135,
      "learning_rate": 1.1697542376393552e-06,
      "loss": 0.6608,
      "step": 18485
    },
    {
      "epoch": 0.7661320402834764,
      "grad_norm": 0.44585224986076355,
      "learning_rate": 1.169547018110987e-06,
      "loss": 0.7113,
      "step": 18486
    },
    {
      "epoch": 0.76617348418915,
      "grad_norm": 0.45966386795043945,
      "learning_rate": 1.1693397985826186e-06,
      "loss": 0.7385,
      "step": 18487
    },
    {
      "epoch": 0.7662149280948236,
      "grad_norm": 0.45024600625038147,
      "learning_rate": 1.1691325790542502e-06,
      "loss": 0.7181,
      "step": 18488
    },
    {
      "epoch": 0.7662563720004973,
      "grad_norm": 0.40605056285858154,
      "learning_rate": 1.1689253595258818e-06,
      "loss": 0.6372,
      "step": 18489
    },
    {
      "epoch": 0.766297815906171,
      "grad_norm": 0.4288059175014496,
      "learning_rate": 1.1687181399975136e-06,
      "loss": 0.694,
      "step": 18490
    },
    {
      "epoch": 0.7663392598118447,
      "grad_norm": 0.4235321581363678,
      "learning_rate": 1.1685109204691452e-06,
      "loss": 0.6887,
      "step": 18491
    },
    {
      "epoch": 0.7663807037175183,
      "grad_norm": 0.4153278172016144,
      "learning_rate": 1.1683037009407768e-06,
      "loss": 0.6536,
      "step": 18492
    },
    {
      "epoch": 0.766422147623192,
      "grad_norm": 0.3742377460002899,
      "learning_rate": 1.1680964814124084e-06,
      "loss": 0.6241,
      "step": 18493
    },
    {
      "epoch": 0.7664635915288657,
      "grad_norm": 0.4051739573478699,
      "learning_rate": 1.1678892618840402e-06,
      "loss": 0.6682,
      "step": 18494
    },
    {
      "epoch": 0.7665050354345394,
      "grad_norm": 0.41767990589141846,
      "learning_rate": 1.1676820423556718e-06,
      "loss": 0.6536,
      "step": 18495
    },
    {
      "epoch": 0.766546479340213,
      "grad_norm": 0.4442001283168793,
      "learning_rate": 1.1674748228273034e-06,
      "loss": 0.6741,
      "step": 18496
    },
    {
      "epoch": 0.7665879232458866,
      "grad_norm": 0.44730162620544434,
      "learning_rate": 1.167267603298935e-06,
      "loss": 0.7124,
      "step": 18497
    },
    {
      "epoch": 0.7666293671515604,
      "grad_norm": 0.4113895297050476,
      "learning_rate": 1.1670603837705666e-06,
      "loss": 0.6086,
      "step": 18498
    },
    {
      "epoch": 0.766670811057234,
      "grad_norm": 0.3887301981449127,
      "learning_rate": 1.1668531642421984e-06,
      "loss": 0.6648,
      "step": 18499
    },
    {
      "epoch": 0.7667122549629077,
      "grad_norm": 0.42825308442115784,
      "learning_rate": 1.16664594471383e-06,
      "loss": 0.6331,
      "step": 18500
    },
    {
      "epoch": 0.7667536988685814,
      "grad_norm": 0.4334266185760498,
      "learning_rate": 1.1664387251854616e-06,
      "loss": 0.7128,
      "step": 18501
    },
    {
      "epoch": 0.7667951427742551,
      "grad_norm": 0.41184303164482117,
      "learning_rate": 1.1662315056570932e-06,
      "loss": 0.6627,
      "step": 18502
    },
    {
      "epoch": 0.7668365866799287,
      "grad_norm": 0.42716002464294434,
      "learning_rate": 1.166024286128725e-06,
      "loss": 0.657,
      "step": 18503
    },
    {
      "epoch": 0.7668780305856023,
      "grad_norm": 0.4353923201560974,
      "learning_rate": 1.1658170666003566e-06,
      "loss": 0.6699,
      "step": 18504
    },
    {
      "epoch": 0.7669194744912761,
      "grad_norm": 0.40347740054130554,
      "learning_rate": 1.1656098470719882e-06,
      "loss": 0.6812,
      "step": 18505
    },
    {
      "epoch": 0.7669609183969497,
      "grad_norm": 0.41010141372680664,
      "learning_rate": 1.1654026275436198e-06,
      "loss": 0.6416,
      "step": 18506
    },
    {
      "epoch": 0.7670023623026234,
      "grad_norm": 0.4176444709300995,
      "learning_rate": 1.1651954080152514e-06,
      "loss": 0.6664,
      "step": 18507
    },
    {
      "epoch": 0.767043806208297,
      "grad_norm": 0.3686901032924652,
      "learning_rate": 1.1649881884868832e-06,
      "loss": 0.6389,
      "step": 18508
    },
    {
      "epoch": 0.7670852501139708,
      "grad_norm": 0.43918144702911377,
      "learning_rate": 1.1647809689585148e-06,
      "loss": 0.6884,
      "step": 18509
    },
    {
      "epoch": 0.7671266940196444,
      "grad_norm": 0.4338878393173218,
      "learning_rate": 1.1645737494301464e-06,
      "loss": 0.7166,
      "step": 18510
    },
    {
      "epoch": 0.7671681379253181,
      "grad_norm": 0.4349721372127533,
      "learning_rate": 1.164366529901778e-06,
      "loss": 0.6688,
      "step": 18511
    },
    {
      "epoch": 0.7672095818309917,
      "grad_norm": 0.40066230297088623,
      "learning_rate": 1.1641593103734098e-06,
      "loss": 0.6277,
      "step": 18512
    },
    {
      "epoch": 0.7672510257366654,
      "grad_norm": 0.39259424805641174,
      "learning_rate": 1.1639520908450414e-06,
      "loss": 0.6562,
      "step": 18513
    },
    {
      "epoch": 0.7672924696423391,
      "grad_norm": 0.4056980013847351,
      "learning_rate": 1.163744871316673e-06,
      "loss": 0.6245,
      "step": 18514
    },
    {
      "epoch": 0.7673339135480127,
      "grad_norm": 0.41037675738334656,
      "learning_rate": 1.1635376517883046e-06,
      "loss": 0.6943,
      "step": 18515
    },
    {
      "epoch": 0.7673753574536865,
      "grad_norm": 0.40309709310531616,
      "learning_rate": 1.1633304322599364e-06,
      "loss": 0.6935,
      "step": 18516
    },
    {
      "epoch": 0.7674168013593601,
      "grad_norm": 0.3900870680809021,
      "learning_rate": 1.163123212731568e-06,
      "loss": 0.655,
      "step": 18517
    },
    {
      "epoch": 0.7674582452650338,
      "grad_norm": 0.443907231092453,
      "learning_rate": 1.1629159932031996e-06,
      "loss": 0.7085,
      "step": 18518
    },
    {
      "epoch": 0.7674996891707074,
      "grad_norm": 0.3788963854312897,
      "learning_rate": 1.1627087736748312e-06,
      "loss": 0.6425,
      "step": 18519
    },
    {
      "epoch": 0.7675411330763812,
      "grad_norm": 0.3821406960487366,
      "learning_rate": 1.1625015541464628e-06,
      "loss": 0.6785,
      "step": 18520
    },
    {
      "epoch": 0.7675825769820548,
      "grad_norm": 0.4338715672492981,
      "learning_rate": 1.1622943346180946e-06,
      "loss": 0.6886,
      "step": 18521
    },
    {
      "epoch": 0.7676240208877284,
      "grad_norm": 0.40271928906440735,
      "learning_rate": 1.1620871150897262e-06,
      "loss": 0.6396,
      "step": 18522
    },
    {
      "epoch": 0.7676654647934021,
      "grad_norm": 0.4060998558998108,
      "learning_rate": 1.1618798955613578e-06,
      "loss": 0.6859,
      "step": 18523
    },
    {
      "epoch": 0.7677069086990758,
      "grad_norm": 0.42858368158340454,
      "learning_rate": 1.1616726760329894e-06,
      "loss": 0.6621,
      "step": 18524
    },
    {
      "epoch": 0.7677483526047495,
      "grad_norm": 0.367145299911499,
      "learning_rate": 1.1614654565046212e-06,
      "loss": 0.6489,
      "step": 18525
    },
    {
      "epoch": 0.7677897965104231,
      "grad_norm": 0.4178408980369568,
      "learning_rate": 1.1612582369762528e-06,
      "loss": 0.7119,
      "step": 18526
    },
    {
      "epoch": 0.7678312404160968,
      "grad_norm": 0.4580093324184418,
      "learning_rate": 1.1610510174478844e-06,
      "loss": 0.7937,
      "step": 18527
    },
    {
      "epoch": 0.7678726843217705,
      "grad_norm": 0.4395301043987274,
      "learning_rate": 1.160843797919516e-06,
      "loss": 0.7051,
      "step": 18528
    },
    {
      "epoch": 0.7679141282274442,
      "grad_norm": 0.39206668734550476,
      "learning_rate": 1.1606365783911478e-06,
      "loss": 0.6702,
      "step": 18529
    },
    {
      "epoch": 0.7679555721331178,
      "grad_norm": 0.409267783164978,
      "learning_rate": 1.1604293588627794e-06,
      "loss": 0.6729,
      "step": 18530
    },
    {
      "epoch": 0.7679970160387914,
      "grad_norm": 0.43972328305244446,
      "learning_rate": 1.160222139334411e-06,
      "loss": 0.6597,
      "step": 18531
    },
    {
      "epoch": 0.7680384599444652,
      "grad_norm": 0.40109190344810486,
      "learning_rate": 1.1600149198060426e-06,
      "loss": 0.6602,
      "step": 18532
    },
    {
      "epoch": 0.7680799038501388,
      "grad_norm": 0.39063355326652527,
      "learning_rate": 1.1598077002776742e-06,
      "loss": 0.6273,
      "step": 18533
    },
    {
      "epoch": 0.7681213477558125,
      "grad_norm": 0.44438230991363525,
      "learning_rate": 1.159600480749306e-06,
      "loss": 0.6354,
      "step": 18534
    },
    {
      "epoch": 0.7681627916614862,
      "grad_norm": 0.4230665862560272,
      "learning_rate": 1.1593932612209376e-06,
      "loss": 0.637,
      "step": 18535
    },
    {
      "epoch": 0.7682042355671599,
      "grad_norm": 0.43769678473472595,
      "learning_rate": 1.1591860416925692e-06,
      "loss": 0.7104,
      "step": 18536
    },
    {
      "epoch": 0.7682456794728335,
      "grad_norm": 0.41632193326950073,
      "learning_rate": 1.1589788221642008e-06,
      "loss": 0.7039,
      "step": 18537
    },
    {
      "epoch": 0.7682871233785072,
      "grad_norm": 0.3924875259399414,
      "learning_rate": 1.1587716026358326e-06,
      "loss": 0.6177,
      "step": 18538
    },
    {
      "epoch": 0.7683285672841809,
      "grad_norm": 0.4123399555683136,
      "learning_rate": 1.1585643831074642e-06,
      "loss": 0.6517,
      "step": 18539
    },
    {
      "epoch": 0.7683700111898545,
      "grad_norm": 0.3976581394672394,
      "learning_rate": 1.1583571635790958e-06,
      "loss": 0.6951,
      "step": 18540
    },
    {
      "epoch": 0.7684114550955282,
      "grad_norm": 0.42206886410713196,
      "learning_rate": 1.1581499440507274e-06,
      "loss": 0.6785,
      "step": 18541
    },
    {
      "epoch": 0.7684528990012018,
      "grad_norm": 0.4087122976779938,
      "learning_rate": 1.1579427245223592e-06,
      "loss": 0.6641,
      "step": 18542
    },
    {
      "epoch": 0.7684943429068756,
      "grad_norm": 0.39679965376853943,
      "learning_rate": 1.1577355049939908e-06,
      "loss": 0.6572,
      "step": 18543
    },
    {
      "epoch": 0.7685357868125492,
      "grad_norm": 0.42872047424316406,
      "learning_rate": 1.1575282854656224e-06,
      "loss": 0.6372,
      "step": 18544
    },
    {
      "epoch": 0.7685772307182229,
      "grad_norm": 0.44230228662490845,
      "learning_rate": 1.157321065937254e-06,
      "loss": 0.6653,
      "step": 18545
    },
    {
      "epoch": 0.7686186746238965,
      "grad_norm": 0.4191790521144867,
      "learning_rate": 1.1571138464088856e-06,
      "loss": 0.6431,
      "step": 18546
    },
    {
      "epoch": 0.7686601185295703,
      "grad_norm": 0.4560861885547638,
      "learning_rate": 1.1569066268805174e-06,
      "loss": 0.6555,
      "step": 18547
    },
    {
      "epoch": 0.7687015624352439,
      "grad_norm": 0.390288770198822,
      "learning_rate": 1.156699407352149e-06,
      "loss": 0.6685,
      "step": 18548
    },
    {
      "epoch": 0.7687430063409175,
      "grad_norm": 0.41414180397987366,
      "learning_rate": 1.1564921878237806e-06,
      "loss": 0.7012,
      "step": 18549
    },
    {
      "epoch": 0.7687844502465913,
      "grad_norm": 0.4463585913181305,
      "learning_rate": 1.1562849682954122e-06,
      "loss": 0.6968,
      "step": 18550
    },
    {
      "epoch": 0.7688258941522649,
      "grad_norm": 0.45496705174446106,
      "learning_rate": 1.156077748767044e-06,
      "loss": 0.6447,
      "step": 18551
    },
    {
      "epoch": 0.7688673380579386,
      "grad_norm": 0.405990332365036,
      "learning_rate": 1.1558705292386756e-06,
      "loss": 0.6873,
      "step": 18552
    },
    {
      "epoch": 0.7689087819636122,
      "grad_norm": 0.4221707880496979,
      "learning_rate": 1.1556633097103072e-06,
      "loss": 0.6873,
      "step": 18553
    },
    {
      "epoch": 0.768950225869286,
      "grad_norm": 0.40096721053123474,
      "learning_rate": 1.1554560901819388e-06,
      "loss": 0.6316,
      "step": 18554
    },
    {
      "epoch": 0.7689916697749596,
      "grad_norm": 0.41256675124168396,
      "learning_rate": 1.1552488706535706e-06,
      "loss": 0.678,
      "step": 18555
    },
    {
      "epoch": 0.7690331136806333,
      "grad_norm": 0.4267052710056305,
      "learning_rate": 1.1550416511252022e-06,
      "loss": 0.6801,
      "step": 18556
    },
    {
      "epoch": 0.7690745575863069,
      "grad_norm": 0.43435922265052795,
      "learning_rate": 1.1548344315968338e-06,
      "loss": 0.7118,
      "step": 18557
    },
    {
      "epoch": 0.7691160014919806,
      "grad_norm": 0.45345303416252136,
      "learning_rate": 1.1546272120684654e-06,
      "loss": 0.6426,
      "step": 18558
    },
    {
      "epoch": 0.7691574453976543,
      "grad_norm": 0.3971325159072876,
      "learning_rate": 1.154419992540097e-06,
      "loss": 0.6333,
      "step": 18559
    },
    {
      "epoch": 0.7691988893033279,
      "grad_norm": 0.4306395351886749,
      "learning_rate": 1.1542127730117288e-06,
      "loss": 0.6558,
      "step": 18560
    },
    {
      "epoch": 0.7692403332090016,
      "grad_norm": 0.41357091069221497,
      "learning_rate": 1.1540055534833604e-06,
      "loss": 0.6914,
      "step": 18561
    },
    {
      "epoch": 0.7692817771146753,
      "grad_norm": 0.41950181126594543,
      "learning_rate": 1.153798333954992e-06,
      "loss": 0.6665,
      "step": 18562
    },
    {
      "epoch": 0.769323221020349,
      "grad_norm": 0.43289950489997864,
      "learning_rate": 1.1535911144266236e-06,
      "loss": 0.6517,
      "step": 18563
    },
    {
      "epoch": 0.7693646649260226,
      "grad_norm": 0.4054003059864044,
      "learning_rate": 1.1533838948982554e-06,
      "loss": 0.6503,
      "step": 18564
    },
    {
      "epoch": 0.7694061088316962,
      "grad_norm": 0.3784058094024658,
      "learning_rate": 1.153176675369887e-06,
      "loss": 0.641,
      "step": 18565
    },
    {
      "epoch": 0.76944755273737,
      "grad_norm": 0.4092369079589844,
      "learning_rate": 1.1529694558415186e-06,
      "loss": 0.6354,
      "step": 18566
    },
    {
      "epoch": 0.7694889966430436,
      "grad_norm": 0.42733046412467957,
      "learning_rate": 1.1527622363131502e-06,
      "loss": 0.678,
      "step": 18567
    },
    {
      "epoch": 0.7695304405487173,
      "grad_norm": 0.4132389724254608,
      "learning_rate": 1.152555016784782e-06,
      "loss": 0.6943,
      "step": 18568
    },
    {
      "epoch": 0.769571884454391,
      "grad_norm": 0.4011296331882477,
      "learning_rate": 1.1523477972564136e-06,
      "loss": 0.5869,
      "step": 18569
    },
    {
      "epoch": 0.7696133283600647,
      "grad_norm": 0.4523649513721466,
      "learning_rate": 1.1521405777280452e-06,
      "loss": 0.7327,
      "step": 18570
    },
    {
      "epoch": 0.7696547722657383,
      "grad_norm": 0.3923201262950897,
      "learning_rate": 1.1519333581996768e-06,
      "loss": 0.6693,
      "step": 18571
    },
    {
      "epoch": 0.769696216171412,
      "grad_norm": 0.4153224527835846,
      "learning_rate": 1.1517261386713084e-06,
      "loss": 0.6848,
      "step": 18572
    },
    {
      "epoch": 0.7697376600770857,
      "grad_norm": 0.5003382563591003,
      "learning_rate": 1.1515189191429402e-06,
      "loss": 0.6887,
      "step": 18573
    },
    {
      "epoch": 0.7697791039827593,
      "grad_norm": 0.4181358218193054,
      "learning_rate": 1.1513116996145718e-06,
      "loss": 0.6614,
      "step": 18574
    },
    {
      "epoch": 0.769820547888433,
      "grad_norm": 0.4148607850074768,
      "learning_rate": 1.1511044800862034e-06,
      "loss": 0.642,
      "step": 18575
    },
    {
      "epoch": 0.7698619917941066,
      "grad_norm": 0.3932316303253174,
      "learning_rate": 1.150897260557835e-06,
      "loss": 0.5818,
      "step": 18576
    },
    {
      "epoch": 0.7699034356997804,
      "grad_norm": 0.4383046627044678,
      "learning_rate": 1.1506900410294668e-06,
      "loss": 0.7273,
      "step": 18577
    },
    {
      "epoch": 0.769944879605454,
      "grad_norm": 0.4460391402244568,
      "learning_rate": 1.1504828215010984e-06,
      "loss": 0.6467,
      "step": 18578
    },
    {
      "epoch": 0.7699863235111277,
      "grad_norm": 0.4331652522087097,
      "learning_rate": 1.15027560197273e-06,
      "loss": 0.7046,
      "step": 18579
    },
    {
      "epoch": 0.7700277674168013,
      "grad_norm": 0.4264139235019684,
      "learning_rate": 1.1500683824443616e-06,
      "loss": 0.6946,
      "step": 18580
    },
    {
      "epoch": 0.7700692113224751,
      "grad_norm": 0.41391050815582275,
      "learning_rate": 1.1498611629159932e-06,
      "loss": 0.6637,
      "step": 18581
    },
    {
      "epoch": 0.7701106552281487,
      "grad_norm": 0.45625627040863037,
      "learning_rate": 1.149653943387625e-06,
      "loss": 0.7258,
      "step": 18582
    },
    {
      "epoch": 0.7701520991338223,
      "grad_norm": 0.4479356110095978,
      "learning_rate": 1.1494467238592566e-06,
      "loss": 0.6826,
      "step": 18583
    },
    {
      "epoch": 0.770193543039496,
      "grad_norm": 0.4043210446834564,
      "learning_rate": 1.1492395043308882e-06,
      "loss": 0.6442,
      "step": 18584
    },
    {
      "epoch": 0.7702349869451697,
      "grad_norm": 0.40128234028816223,
      "learning_rate": 1.1490322848025198e-06,
      "loss": 0.6525,
      "step": 18585
    },
    {
      "epoch": 0.7702764308508434,
      "grad_norm": 0.4102870523929596,
      "learning_rate": 1.1488250652741516e-06,
      "loss": 0.6726,
      "step": 18586
    },
    {
      "epoch": 0.770317874756517,
      "grad_norm": 0.4233836233615875,
      "learning_rate": 1.1486178457457832e-06,
      "loss": 0.6914,
      "step": 18587
    },
    {
      "epoch": 0.7703593186621908,
      "grad_norm": 0.38220882415771484,
      "learning_rate": 1.1484106262174148e-06,
      "loss": 0.6382,
      "step": 18588
    },
    {
      "epoch": 0.7704007625678644,
      "grad_norm": 0.40496397018432617,
      "learning_rate": 1.1482034066890464e-06,
      "loss": 0.682,
      "step": 18589
    },
    {
      "epoch": 0.7704422064735381,
      "grad_norm": 0.4491124451160431,
      "learning_rate": 1.1479961871606782e-06,
      "loss": 0.7189,
      "step": 18590
    },
    {
      "epoch": 0.7704836503792117,
      "grad_norm": 0.40522831678390503,
      "learning_rate": 1.1477889676323098e-06,
      "loss": 0.6604,
      "step": 18591
    },
    {
      "epoch": 0.7705250942848854,
      "grad_norm": 0.433199018239975,
      "learning_rate": 1.1475817481039414e-06,
      "loss": 0.6809,
      "step": 18592
    },
    {
      "epoch": 0.7705665381905591,
      "grad_norm": 0.4345639646053314,
      "learning_rate": 1.147374528575573e-06,
      "loss": 0.6847,
      "step": 18593
    },
    {
      "epoch": 0.7706079820962327,
      "grad_norm": 0.44261783361434937,
      "learning_rate": 1.1471673090472046e-06,
      "loss": 0.7377,
      "step": 18594
    },
    {
      "epoch": 0.7706494260019064,
      "grad_norm": 0.41557615995407104,
      "learning_rate": 1.1469600895188364e-06,
      "loss": 0.6368,
      "step": 18595
    },
    {
      "epoch": 0.7706908699075801,
      "grad_norm": 0.40809065103530884,
      "learning_rate": 1.146752869990468e-06,
      "loss": 0.7234,
      "step": 18596
    },
    {
      "epoch": 0.7707323138132538,
      "grad_norm": 0.40976637601852417,
      "learning_rate": 1.1465456504620996e-06,
      "loss": 0.6381,
      "step": 18597
    },
    {
      "epoch": 0.7707737577189274,
      "grad_norm": 0.4206528663635254,
      "learning_rate": 1.1463384309337312e-06,
      "loss": 0.64,
      "step": 18598
    },
    {
      "epoch": 0.7708152016246012,
      "grad_norm": 0.4338268041610718,
      "learning_rate": 1.146131211405363e-06,
      "loss": 0.6899,
      "step": 18599
    },
    {
      "epoch": 0.7708566455302748,
      "grad_norm": 0.4137064814567566,
      "learning_rate": 1.1459239918769946e-06,
      "loss": 0.6779,
      "step": 18600
    },
    {
      "epoch": 0.7708980894359484,
      "grad_norm": 0.4082860052585602,
      "learning_rate": 1.1457167723486262e-06,
      "loss": 0.6826,
      "step": 18601
    },
    {
      "epoch": 0.7709395333416221,
      "grad_norm": 0.4037609398365021,
      "learning_rate": 1.1455095528202578e-06,
      "loss": 0.662,
      "step": 18602
    },
    {
      "epoch": 0.7709809772472958,
      "grad_norm": 0.3922029137611389,
      "learning_rate": 1.1453023332918896e-06,
      "loss": 0.6444,
      "step": 18603
    },
    {
      "epoch": 0.7710224211529695,
      "grad_norm": 0.4038182199001312,
      "learning_rate": 1.1450951137635212e-06,
      "loss": 0.677,
      "step": 18604
    },
    {
      "epoch": 0.7710638650586431,
      "grad_norm": 0.4116866886615753,
      "learning_rate": 1.1448878942351528e-06,
      "loss": 0.6696,
      "step": 18605
    },
    {
      "epoch": 0.7711053089643168,
      "grad_norm": 0.419552206993103,
      "learning_rate": 1.1446806747067844e-06,
      "loss": 0.7444,
      "step": 18606
    },
    {
      "epoch": 0.7711467528699905,
      "grad_norm": 0.40074682235717773,
      "learning_rate": 1.144473455178416e-06,
      "loss": 0.6091,
      "step": 18607
    },
    {
      "epoch": 0.7711881967756642,
      "grad_norm": 0.41989514231681824,
      "learning_rate": 1.1442662356500478e-06,
      "loss": 0.6621,
      "step": 18608
    },
    {
      "epoch": 0.7712296406813378,
      "grad_norm": 0.42825815081596375,
      "learning_rate": 1.1440590161216794e-06,
      "loss": 0.6545,
      "step": 18609
    },
    {
      "epoch": 0.7712710845870114,
      "grad_norm": 0.4134232699871063,
      "learning_rate": 1.143851796593311e-06,
      "loss": 0.6871,
      "step": 18610
    },
    {
      "epoch": 0.7713125284926852,
      "grad_norm": 0.3758434057235718,
      "learning_rate": 1.1436445770649426e-06,
      "loss": 0.6089,
      "step": 18611
    },
    {
      "epoch": 0.7713539723983588,
      "grad_norm": 0.3733452260494232,
      "learning_rate": 1.1434373575365744e-06,
      "loss": 0.6907,
      "step": 18612
    },
    {
      "epoch": 0.7713954163040325,
      "grad_norm": 0.4429336190223694,
      "learning_rate": 1.143230138008206e-06,
      "loss": 0.6959,
      "step": 18613
    },
    {
      "epoch": 0.7714368602097061,
      "grad_norm": 0.37851688265800476,
      "learning_rate": 1.1430229184798376e-06,
      "loss": 0.65,
      "step": 18614
    },
    {
      "epoch": 0.7714783041153799,
      "grad_norm": 0.4046182930469513,
      "learning_rate": 1.1428156989514692e-06,
      "loss": 0.6455,
      "step": 18615
    },
    {
      "epoch": 0.7715197480210535,
      "grad_norm": 0.44494199752807617,
      "learning_rate": 1.142608479423101e-06,
      "loss": 0.7251,
      "step": 18616
    },
    {
      "epoch": 0.7715611919267272,
      "grad_norm": 0.41044414043426514,
      "learning_rate": 1.1424012598947326e-06,
      "loss": 0.6527,
      "step": 18617
    },
    {
      "epoch": 0.7716026358324009,
      "grad_norm": 0.4215778112411499,
      "learning_rate": 1.1421940403663642e-06,
      "loss": 0.6835,
      "step": 18618
    },
    {
      "epoch": 0.7716440797380745,
      "grad_norm": 0.3994966149330139,
      "learning_rate": 1.1419868208379958e-06,
      "loss": 0.6703,
      "step": 18619
    },
    {
      "epoch": 0.7716855236437482,
      "grad_norm": 0.45467814803123474,
      "learning_rate": 1.1417796013096274e-06,
      "loss": 0.6818,
      "step": 18620
    },
    {
      "epoch": 0.7717269675494218,
      "grad_norm": 0.49117904901504517,
      "learning_rate": 1.1415723817812592e-06,
      "loss": 0.7037,
      "step": 18621
    },
    {
      "epoch": 0.7717684114550956,
      "grad_norm": 0.43307989835739136,
      "learning_rate": 1.1413651622528908e-06,
      "loss": 0.7048,
      "step": 18622
    },
    {
      "epoch": 0.7718098553607692,
      "grad_norm": 0.4177330434322357,
      "learning_rate": 1.1411579427245224e-06,
      "loss": 0.6305,
      "step": 18623
    },
    {
      "epoch": 0.7718512992664429,
      "grad_norm": 0.4392470717430115,
      "learning_rate": 1.140950723196154e-06,
      "loss": 0.6896,
      "step": 18624
    },
    {
      "epoch": 0.7718927431721165,
      "grad_norm": 0.39668378233909607,
      "learning_rate": 1.1407435036677858e-06,
      "loss": 0.66,
      "step": 18625
    },
    {
      "epoch": 0.7719341870777902,
      "grad_norm": 0.40316227078437805,
      "learning_rate": 1.1405362841394174e-06,
      "loss": 0.6266,
      "step": 18626
    },
    {
      "epoch": 0.7719756309834639,
      "grad_norm": 0.44575291872024536,
      "learning_rate": 1.140329064611049e-06,
      "loss": 0.7234,
      "step": 18627
    },
    {
      "epoch": 0.7720170748891375,
      "grad_norm": 0.44293880462646484,
      "learning_rate": 1.1401218450826806e-06,
      "loss": 0.7629,
      "step": 18628
    },
    {
      "epoch": 0.7720585187948112,
      "grad_norm": 0.4704482853412628,
      "learning_rate": 1.1399146255543124e-06,
      "loss": 0.666,
      "step": 18629
    },
    {
      "epoch": 0.7720999627004849,
      "grad_norm": 0.4036056697368622,
      "learning_rate": 1.139707406025944e-06,
      "loss": 0.6243,
      "step": 18630
    },
    {
      "epoch": 0.7721414066061586,
      "grad_norm": 0.41639819741249084,
      "learning_rate": 1.1395001864975756e-06,
      "loss": 0.6492,
      "step": 18631
    },
    {
      "epoch": 0.7721828505118322,
      "grad_norm": 0.42895159125328064,
      "learning_rate": 1.1392929669692072e-06,
      "loss": 0.6853,
      "step": 18632
    },
    {
      "epoch": 0.772224294417506,
      "grad_norm": 0.37037062644958496,
      "learning_rate": 1.1390857474408388e-06,
      "loss": 0.6193,
      "step": 18633
    },
    {
      "epoch": 0.7722657383231796,
      "grad_norm": 0.49012070894241333,
      "learning_rate": 1.1388785279124706e-06,
      "loss": 0.7498,
      "step": 18634
    },
    {
      "epoch": 0.7723071822288532,
      "grad_norm": 0.40547385811805725,
      "learning_rate": 1.1386713083841022e-06,
      "loss": 0.6821,
      "step": 18635
    },
    {
      "epoch": 0.7723486261345269,
      "grad_norm": 0.4003818929195404,
      "learning_rate": 1.1384640888557338e-06,
      "loss": 0.7014,
      "step": 18636
    },
    {
      "epoch": 0.7723900700402005,
      "grad_norm": 0.40989387035369873,
      "learning_rate": 1.1382568693273654e-06,
      "loss": 0.6407,
      "step": 18637
    },
    {
      "epoch": 0.7724315139458743,
      "grad_norm": 0.4199989140033722,
      "learning_rate": 1.1380496497989972e-06,
      "loss": 0.6506,
      "step": 18638
    },
    {
      "epoch": 0.7724729578515479,
      "grad_norm": 0.4276945888996124,
      "learning_rate": 1.1378424302706288e-06,
      "loss": 0.6554,
      "step": 18639
    },
    {
      "epoch": 0.7725144017572216,
      "grad_norm": 0.4385223984718323,
      "learning_rate": 1.1376352107422604e-06,
      "loss": 0.6053,
      "step": 18640
    },
    {
      "epoch": 0.7725558456628953,
      "grad_norm": 0.4457826018333435,
      "learning_rate": 1.137427991213892e-06,
      "loss": 0.7139,
      "step": 18641
    },
    {
      "epoch": 0.772597289568569,
      "grad_norm": 0.4264000654220581,
      "learning_rate": 1.1372207716855238e-06,
      "loss": 0.6613,
      "step": 18642
    },
    {
      "epoch": 0.7726387334742426,
      "grad_norm": 0.400253564119339,
      "learning_rate": 1.1370135521571554e-06,
      "loss": 0.6965,
      "step": 18643
    },
    {
      "epoch": 0.7726801773799162,
      "grad_norm": 0.4083994925022125,
      "learning_rate": 1.136806332628787e-06,
      "loss": 0.6445,
      "step": 18644
    },
    {
      "epoch": 0.77272162128559,
      "grad_norm": 0.4834294319152832,
      "learning_rate": 1.1365991131004186e-06,
      "loss": 0.7148,
      "step": 18645
    },
    {
      "epoch": 0.7727630651912636,
      "grad_norm": 0.41728124022483826,
      "learning_rate": 1.1363918935720502e-06,
      "loss": 0.6461,
      "step": 18646
    },
    {
      "epoch": 0.7728045090969373,
      "grad_norm": 0.3866616487503052,
      "learning_rate": 1.136184674043682e-06,
      "loss": 0.6781,
      "step": 18647
    },
    {
      "epoch": 0.7728459530026109,
      "grad_norm": 0.3816164433956146,
      "learning_rate": 1.1359774545153136e-06,
      "loss": 0.6248,
      "step": 18648
    },
    {
      "epoch": 0.7728873969082847,
      "grad_norm": 0.40736693143844604,
      "learning_rate": 1.1357702349869452e-06,
      "loss": 0.6743,
      "step": 18649
    },
    {
      "epoch": 0.7729288408139583,
      "grad_norm": 0.4172896146774292,
      "learning_rate": 1.1355630154585768e-06,
      "loss": 0.7017,
      "step": 18650
    },
    {
      "epoch": 0.772970284719632,
      "grad_norm": 0.4151565432548523,
      "learning_rate": 1.1353557959302086e-06,
      "loss": 0.6658,
      "step": 18651
    },
    {
      "epoch": 0.7730117286253056,
      "grad_norm": 0.41716209053993225,
      "learning_rate": 1.1351485764018402e-06,
      "loss": 0.6277,
      "step": 18652
    },
    {
      "epoch": 0.7730531725309793,
      "grad_norm": 0.3743089437484741,
      "learning_rate": 1.1349413568734718e-06,
      "loss": 0.6421,
      "step": 18653
    },
    {
      "epoch": 0.773094616436653,
      "grad_norm": 0.4839318096637726,
      "learning_rate": 1.1347341373451034e-06,
      "loss": 0.7744,
      "step": 18654
    },
    {
      "epoch": 0.7731360603423266,
      "grad_norm": 0.44469866156578064,
      "learning_rate": 1.134526917816735e-06,
      "loss": 0.6509,
      "step": 18655
    },
    {
      "epoch": 0.7731775042480004,
      "grad_norm": 0.4358755946159363,
      "learning_rate": 1.1343196982883668e-06,
      "loss": 0.6907,
      "step": 18656
    },
    {
      "epoch": 0.773218948153674,
      "grad_norm": 0.4016987681388855,
      "learning_rate": 1.1341124787599984e-06,
      "loss": 0.6802,
      "step": 18657
    },
    {
      "epoch": 0.7732603920593477,
      "grad_norm": 0.3807734549045563,
      "learning_rate": 1.13390525923163e-06,
      "loss": 0.6167,
      "step": 18658
    },
    {
      "epoch": 0.7733018359650213,
      "grad_norm": 0.427251398563385,
      "learning_rate": 1.1336980397032616e-06,
      "loss": 0.6669,
      "step": 18659
    },
    {
      "epoch": 0.7733432798706951,
      "grad_norm": 0.41537249088287354,
      "learning_rate": 1.1334908201748934e-06,
      "loss": 0.6511,
      "step": 18660
    },
    {
      "epoch": 0.7733847237763687,
      "grad_norm": 0.4192321002483368,
      "learning_rate": 1.133283600646525e-06,
      "loss": 0.6682,
      "step": 18661
    },
    {
      "epoch": 0.7734261676820423,
      "grad_norm": 0.40378138422966003,
      "learning_rate": 1.1330763811181566e-06,
      "loss": 0.6658,
      "step": 18662
    },
    {
      "epoch": 0.773467611587716,
      "grad_norm": 0.45574843883514404,
      "learning_rate": 1.1328691615897882e-06,
      "loss": 0.6722,
      "step": 18663
    },
    {
      "epoch": 0.7735090554933897,
      "grad_norm": 0.40728622674942017,
      "learning_rate": 1.13266194206142e-06,
      "loss": 0.6637,
      "step": 18664
    },
    {
      "epoch": 0.7735504993990634,
      "grad_norm": 0.4360423982143402,
      "learning_rate": 1.1324547225330516e-06,
      "loss": 0.672,
      "step": 18665
    },
    {
      "epoch": 0.773591943304737,
      "grad_norm": 0.39295539259910583,
      "learning_rate": 1.1322475030046832e-06,
      "loss": 0.6438,
      "step": 18666
    },
    {
      "epoch": 0.7736333872104108,
      "grad_norm": 0.43571820855140686,
      "learning_rate": 1.1320402834763148e-06,
      "loss": 0.704,
      "step": 18667
    },
    {
      "epoch": 0.7736748311160844,
      "grad_norm": 0.46852004528045654,
      "learning_rate": 1.1318330639479464e-06,
      "loss": 0.7096,
      "step": 18668
    },
    {
      "epoch": 0.7737162750217581,
      "grad_norm": 0.39040815830230713,
      "learning_rate": 1.1316258444195782e-06,
      "loss": 0.6171,
      "step": 18669
    },
    {
      "epoch": 0.7737577189274317,
      "grad_norm": 0.42720717191696167,
      "learning_rate": 1.1314186248912098e-06,
      "loss": 0.6484,
      "step": 18670
    },
    {
      "epoch": 0.7737991628331053,
      "grad_norm": 0.40399742126464844,
      "learning_rate": 1.1312114053628414e-06,
      "loss": 0.6472,
      "step": 18671
    },
    {
      "epoch": 0.7738406067387791,
      "grad_norm": 0.40572458505630493,
      "learning_rate": 1.131004185834473e-06,
      "loss": 0.6277,
      "step": 18672
    },
    {
      "epoch": 0.7738820506444527,
      "grad_norm": 0.4221031665802002,
      "learning_rate": 1.1307969663061048e-06,
      "loss": 0.7378,
      "step": 18673
    },
    {
      "epoch": 0.7739234945501264,
      "grad_norm": 0.4922735095024109,
      "learning_rate": 1.1305897467777364e-06,
      "loss": 0.6749,
      "step": 18674
    },
    {
      "epoch": 0.7739649384558001,
      "grad_norm": 0.38611453771591187,
      "learning_rate": 1.130382527249368e-06,
      "loss": 0.6338,
      "step": 18675
    },
    {
      "epoch": 0.7740063823614738,
      "grad_norm": 0.42773130536079407,
      "learning_rate": 1.1301753077209996e-06,
      "loss": 0.6699,
      "step": 18676
    },
    {
      "epoch": 0.7740478262671474,
      "grad_norm": 0.3943508267402649,
      "learning_rate": 1.1299680881926314e-06,
      "loss": 0.5994,
      "step": 18677
    },
    {
      "epoch": 0.774089270172821,
      "grad_norm": 0.4129193127155304,
      "learning_rate": 1.129760868664263e-06,
      "loss": 0.6511,
      "step": 18678
    },
    {
      "epoch": 0.7741307140784948,
      "grad_norm": 0.45398756861686707,
      "learning_rate": 1.1295536491358946e-06,
      "loss": 0.6577,
      "step": 18679
    },
    {
      "epoch": 0.7741721579841684,
      "grad_norm": 0.44203853607177734,
      "learning_rate": 1.1293464296075262e-06,
      "loss": 0.7103,
      "step": 18680
    },
    {
      "epoch": 0.7742136018898421,
      "grad_norm": 0.4224059581756592,
      "learning_rate": 1.1291392100791578e-06,
      "loss": 0.7003,
      "step": 18681
    },
    {
      "epoch": 0.7742550457955157,
      "grad_norm": 0.38337188959121704,
      "learning_rate": 1.1289319905507896e-06,
      "loss": 0.6665,
      "step": 18682
    },
    {
      "epoch": 0.7742964897011895,
      "grad_norm": 0.4135085940361023,
      "learning_rate": 1.1287247710224212e-06,
      "loss": 0.6583,
      "step": 18683
    },
    {
      "epoch": 0.7743379336068631,
      "grad_norm": 0.43615031242370605,
      "learning_rate": 1.1285175514940528e-06,
      "loss": 0.7256,
      "step": 18684
    },
    {
      "epoch": 0.7743793775125368,
      "grad_norm": 0.464132159948349,
      "learning_rate": 1.1283103319656844e-06,
      "loss": 0.6807,
      "step": 18685
    },
    {
      "epoch": 0.7744208214182104,
      "grad_norm": 0.4091275632381439,
      "learning_rate": 1.1281031124373162e-06,
      "loss": 0.6733,
      "step": 18686
    },
    {
      "epoch": 0.7744622653238841,
      "grad_norm": 0.4081200361251831,
      "learning_rate": 1.1278958929089478e-06,
      "loss": 0.6549,
      "step": 18687
    },
    {
      "epoch": 0.7745037092295578,
      "grad_norm": 0.4100535213947296,
      "learning_rate": 1.1276886733805794e-06,
      "loss": 0.6578,
      "step": 18688
    },
    {
      "epoch": 0.7745451531352314,
      "grad_norm": 0.4124448299407959,
      "learning_rate": 1.127481453852211e-06,
      "loss": 0.6598,
      "step": 18689
    },
    {
      "epoch": 0.7745865970409052,
      "grad_norm": 0.39033955335617065,
      "learning_rate": 1.1272742343238428e-06,
      "loss": 0.663,
      "step": 18690
    },
    {
      "epoch": 0.7746280409465788,
      "grad_norm": 0.3985194265842438,
      "learning_rate": 1.1270670147954744e-06,
      "loss": 0.6025,
      "step": 18691
    },
    {
      "epoch": 0.7746694848522525,
      "grad_norm": 0.44080355763435364,
      "learning_rate": 1.126859795267106e-06,
      "loss": 0.6925,
      "step": 18692
    },
    {
      "epoch": 0.7747109287579261,
      "grad_norm": 0.39983922243118286,
      "learning_rate": 1.1266525757387376e-06,
      "loss": 0.6099,
      "step": 18693
    },
    {
      "epoch": 0.7747523726635999,
      "grad_norm": 0.40150535106658936,
      "learning_rate": 1.1264453562103692e-06,
      "loss": 0.6621,
      "step": 18694
    },
    {
      "epoch": 0.7747938165692735,
      "grad_norm": 0.41600513458251953,
      "learning_rate": 1.126238136682001e-06,
      "loss": 0.6537,
      "step": 18695
    },
    {
      "epoch": 0.7748352604749471,
      "grad_norm": 0.44487765431404114,
      "learning_rate": 1.1260309171536326e-06,
      "loss": 0.6821,
      "step": 18696
    },
    {
      "epoch": 0.7748767043806208,
      "grad_norm": 0.3848533034324646,
      "learning_rate": 1.1258236976252642e-06,
      "loss": 0.6531,
      "step": 18697
    },
    {
      "epoch": 0.7749181482862945,
      "grad_norm": 0.3946925699710846,
      "learning_rate": 1.1256164780968958e-06,
      "loss": 0.6261,
      "step": 18698
    },
    {
      "epoch": 0.7749595921919682,
      "grad_norm": 0.3837633728981018,
      "learning_rate": 1.1254092585685276e-06,
      "loss": 0.6594,
      "step": 18699
    },
    {
      "epoch": 0.7750010360976418,
      "grad_norm": 0.4449877440929413,
      "learning_rate": 1.1252020390401592e-06,
      "loss": 0.6561,
      "step": 18700
    },
    {
      "epoch": 0.7750424800033155,
      "grad_norm": 0.39526981115341187,
      "learning_rate": 1.1249948195117908e-06,
      "loss": 0.6848,
      "step": 18701
    },
    {
      "epoch": 0.7750839239089892,
      "grad_norm": 0.42968806624412537,
      "learning_rate": 1.1247875999834224e-06,
      "loss": 0.6704,
      "step": 18702
    },
    {
      "epoch": 0.7751253678146629,
      "grad_norm": 0.4310893416404724,
      "learning_rate": 1.1245803804550542e-06,
      "loss": 0.6666,
      "step": 18703
    },
    {
      "epoch": 0.7751668117203365,
      "grad_norm": 0.4172687232494354,
      "learning_rate": 1.1243731609266858e-06,
      "loss": 0.6998,
      "step": 18704
    },
    {
      "epoch": 0.7752082556260101,
      "grad_norm": 0.43702420592308044,
      "learning_rate": 1.1241659413983174e-06,
      "loss": 0.6404,
      "step": 18705
    },
    {
      "epoch": 0.7752496995316839,
      "grad_norm": 0.4350883662700653,
      "learning_rate": 1.123958721869949e-06,
      "loss": 0.641,
      "step": 18706
    },
    {
      "epoch": 0.7752911434373575,
      "grad_norm": 0.4447561204433441,
      "learning_rate": 1.1237515023415806e-06,
      "loss": 0.7148,
      "step": 18707
    },
    {
      "epoch": 0.7753325873430312,
      "grad_norm": 0.46580421924591064,
      "learning_rate": 1.1235442828132124e-06,
      "loss": 0.6963,
      "step": 18708
    },
    {
      "epoch": 0.7753740312487049,
      "grad_norm": 0.42502662539482117,
      "learning_rate": 1.123337063284844e-06,
      "loss": 0.6982,
      "step": 18709
    },
    {
      "epoch": 0.7754154751543786,
      "grad_norm": 0.4118465781211853,
      "learning_rate": 1.1231298437564756e-06,
      "loss": 0.6707,
      "step": 18710
    },
    {
      "epoch": 0.7754569190600522,
      "grad_norm": 0.40466099977493286,
      "learning_rate": 1.1229226242281072e-06,
      "loss": 0.6643,
      "step": 18711
    },
    {
      "epoch": 0.7754983629657259,
      "grad_norm": 0.4172675907611847,
      "learning_rate": 1.122715404699739e-06,
      "loss": 0.698,
      "step": 18712
    },
    {
      "epoch": 0.7755398068713996,
      "grad_norm": 0.39812880754470825,
      "learning_rate": 1.1225081851713706e-06,
      "loss": 0.6582,
      "step": 18713
    },
    {
      "epoch": 0.7755812507770732,
      "grad_norm": 0.4064295291900635,
      "learning_rate": 1.1223009656430022e-06,
      "loss": 0.6532,
      "step": 18714
    },
    {
      "epoch": 0.7756226946827469,
      "grad_norm": 0.40281838178634644,
      "learning_rate": 1.1220937461146338e-06,
      "loss": 0.6597,
      "step": 18715
    },
    {
      "epoch": 0.7756641385884205,
      "grad_norm": 0.4151982367038727,
      "learning_rate": 1.1218865265862656e-06,
      "loss": 0.6963,
      "step": 18716
    },
    {
      "epoch": 0.7757055824940943,
      "grad_norm": 0.4396878778934479,
      "learning_rate": 1.1216793070578972e-06,
      "loss": 0.6725,
      "step": 18717
    },
    {
      "epoch": 0.7757470263997679,
      "grad_norm": 0.42466995120048523,
      "learning_rate": 1.1214720875295288e-06,
      "loss": 0.6895,
      "step": 18718
    },
    {
      "epoch": 0.7757884703054416,
      "grad_norm": 0.385654479265213,
      "learning_rate": 1.1212648680011604e-06,
      "loss": 0.6349,
      "step": 18719
    },
    {
      "epoch": 0.7758299142111152,
      "grad_norm": 0.3998183608055115,
      "learning_rate": 1.121057648472792e-06,
      "loss": 0.6108,
      "step": 18720
    },
    {
      "epoch": 0.775871358116789,
      "grad_norm": 0.4541303813457489,
      "learning_rate": 1.1208504289444238e-06,
      "loss": 0.7129,
      "step": 18721
    },
    {
      "epoch": 0.7759128020224626,
      "grad_norm": 0.41898781061172485,
      "learning_rate": 1.1206432094160554e-06,
      "loss": 0.6553,
      "step": 18722
    },
    {
      "epoch": 0.7759542459281362,
      "grad_norm": 0.40791329741477966,
      "learning_rate": 1.120435989887687e-06,
      "loss": 0.67,
      "step": 18723
    },
    {
      "epoch": 0.77599568983381,
      "grad_norm": 0.4193266034126282,
      "learning_rate": 1.1202287703593186e-06,
      "loss": 0.673,
      "step": 18724
    },
    {
      "epoch": 0.7760371337394836,
      "grad_norm": 0.42473992705345154,
      "learning_rate": 1.1200215508309504e-06,
      "loss": 0.719,
      "step": 18725
    },
    {
      "epoch": 0.7760785776451573,
      "grad_norm": 0.46280011534690857,
      "learning_rate": 1.119814331302582e-06,
      "loss": 0.7314,
      "step": 18726
    },
    {
      "epoch": 0.7761200215508309,
      "grad_norm": 0.41773149371147156,
      "learning_rate": 1.1196071117742136e-06,
      "loss": 0.6775,
      "step": 18727
    },
    {
      "epoch": 0.7761614654565047,
      "grad_norm": 0.41040560603141785,
      "learning_rate": 1.1193998922458452e-06,
      "loss": 0.6718,
      "step": 18728
    },
    {
      "epoch": 0.7762029093621783,
      "grad_norm": 0.4290517568588257,
      "learning_rate": 1.1191926727174768e-06,
      "loss": 0.6504,
      "step": 18729
    },
    {
      "epoch": 0.776244353267852,
      "grad_norm": 0.4477480947971344,
      "learning_rate": 1.1189854531891086e-06,
      "loss": 0.6556,
      "step": 18730
    },
    {
      "epoch": 0.7762857971735256,
      "grad_norm": 0.41496020555496216,
      "learning_rate": 1.1187782336607402e-06,
      "loss": 0.6484,
      "step": 18731
    },
    {
      "epoch": 0.7763272410791993,
      "grad_norm": 0.4248695373535156,
      "learning_rate": 1.1185710141323718e-06,
      "loss": 0.6479,
      "step": 18732
    },
    {
      "epoch": 0.776368684984873,
      "grad_norm": 0.41781800985336304,
      "learning_rate": 1.1183637946040034e-06,
      "loss": 0.6887,
      "step": 18733
    },
    {
      "epoch": 0.7764101288905466,
      "grad_norm": 0.4059661030769348,
      "learning_rate": 1.1181565750756352e-06,
      "loss": 0.6475,
      "step": 18734
    },
    {
      "epoch": 0.7764515727962203,
      "grad_norm": 0.4230389893054962,
      "learning_rate": 1.1179493555472668e-06,
      "loss": 0.6417,
      "step": 18735
    },
    {
      "epoch": 0.776493016701894,
      "grad_norm": 0.41161203384399414,
      "learning_rate": 1.1177421360188984e-06,
      "loss": 0.6526,
      "step": 18736
    },
    {
      "epoch": 0.7765344606075677,
      "grad_norm": 0.39863821864128113,
      "learning_rate": 1.11753491649053e-06,
      "loss": 0.658,
      "step": 18737
    },
    {
      "epoch": 0.7765759045132413,
      "grad_norm": 0.41206008195877075,
      "learning_rate": 1.1173276969621618e-06,
      "loss": 0.6914,
      "step": 18738
    },
    {
      "epoch": 0.7766173484189149,
      "grad_norm": 0.43053460121154785,
      "learning_rate": 1.1171204774337934e-06,
      "loss": 0.6831,
      "step": 18739
    },
    {
      "epoch": 0.7766587923245887,
      "grad_norm": 0.40393710136413574,
      "learning_rate": 1.116913257905425e-06,
      "loss": 0.6617,
      "step": 18740
    },
    {
      "epoch": 0.7767002362302623,
      "grad_norm": 0.4298970699310303,
      "learning_rate": 1.1167060383770566e-06,
      "loss": 0.6797,
      "step": 18741
    },
    {
      "epoch": 0.776741680135936,
      "grad_norm": 0.4310683608055115,
      "learning_rate": 1.1164988188486882e-06,
      "loss": 0.7332,
      "step": 18742
    },
    {
      "epoch": 0.7767831240416097,
      "grad_norm": 0.42302408814430237,
      "learning_rate": 1.11629159932032e-06,
      "loss": 0.6742,
      "step": 18743
    },
    {
      "epoch": 0.7768245679472834,
      "grad_norm": 0.42748138308525085,
      "learning_rate": 1.1160843797919516e-06,
      "loss": 0.6689,
      "step": 18744
    },
    {
      "epoch": 0.776866011852957,
      "grad_norm": 0.4327971339225769,
      "learning_rate": 1.1158771602635832e-06,
      "loss": 0.6919,
      "step": 18745
    },
    {
      "epoch": 0.7769074557586307,
      "grad_norm": 0.39091575145721436,
      "learning_rate": 1.1156699407352148e-06,
      "loss": 0.6334,
      "step": 18746
    },
    {
      "epoch": 0.7769488996643044,
      "grad_norm": 0.4027308225631714,
      "learning_rate": 1.1154627212068466e-06,
      "loss": 0.6714,
      "step": 18747
    },
    {
      "epoch": 0.776990343569978,
      "grad_norm": 0.44414129853248596,
      "learning_rate": 1.1152555016784782e-06,
      "loss": 0.6854,
      "step": 18748
    },
    {
      "epoch": 0.7770317874756517,
      "grad_norm": 0.416934996843338,
      "learning_rate": 1.1150482821501098e-06,
      "loss": 0.6931,
      "step": 18749
    },
    {
      "epoch": 0.7770732313813253,
      "grad_norm": 0.4397965371608734,
      "learning_rate": 1.1148410626217414e-06,
      "loss": 0.6407,
      "step": 18750
    },
    {
      "epoch": 0.7771146752869991,
      "grad_norm": 0.42599260807037354,
      "learning_rate": 1.1146338430933732e-06,
      "loss": 0.67,
      "step": 18751
    },
    {
      "epoch": 0.7771561191926727,
      "grad_norm": 0.3895764648914337,
      "learning_rate": 1.1144266235650048e-06,
      "loss": 0.6943,
      "step": 18752
    },
    {
      "epoch": 0.7771975630983464,
      "grad_norm": 0.38975492119789124,
      "learning_rate": 1.1142194040366364e-06,
      "loss": 0.6648,
      "step": 18753
    },
    {
      "epoch": 0.77723900700402,
      "grad_norm": 0.4198983907699585,
      "learning_rate": 1.114012184508268e-06,
      "loss": 0.719,
      "step": 18754
    },
    {
      "epoch": 0.7772804509096938,
      "grad_norm": 0.43661805987358093,
      "learning_rate": 1.1138049649798996e-06,
      "loss": 0.6411,
      "step": 18755
    },
    {
      "epoch": 0.7773218948153674,
      "grad_norm": 0.4458347260951996,
      "learning_rate": 1.1135977454515314e-06,
      "loss": 0.6844,
      "step": 18756
    },
    {
      "epoch": 0.777363338721041,
      "grad_norm": 0.40382084250450134,
      "learning_rate": 1.113390525923163e-06,
      "loss": 0.6949,
      "step": 18757
    },
    {
      "epoch": 0.7774047826267148,
      "grad_norm": 0.41136816143989563,
      "learning_rate": 1.1131833063947946e-06,
      "loss": 0.6422,
      "step": 18758
    },
    {
      "epoch": 0.7774462265323884,
      "grad_norm": 0.4279707372188568,
      "learning_rate": 1.1129760868664262e-06,
      "loss": 0.7142,
      "step": 18759
    },
    {
      "epoch": 0.7774876704380621,
      "grad_norm": 0.4160175323486328,
      "learning_rate": 1.112768867338058e-06,
      "loss": 0.6646,
      "step": 18760
    },
    {
      "epoch": 0.7775291143437357,
      "grad_norm": 0.39878320693969727,
      "learning_rate": 1.1125616478096896e-06,
      "loss": 0.626,
      "step": 18761
    },
    {
      "epoch": 0.7775705582494095,
      "grad_norm": 0.4324229061603546,
      "learning_rate": 1.1123544282813212e-06,
      "loss": 0.6831,
      "step": 18762
    },
    {
      "epoch": 0.7776120021550831,
      "grad_norm": 0.39504092931747437,
      "learning_rate": 1.1121472087529528e-06,
      "loss": 0.6339,
      "step": 18763
    },
    {
      "epoch": 0.7776534460607568,
      "grad_norm": 0.4084392189979553,
      "learning_rate": 1.1119399892245846e-06,
      "loss": 0.6624,
      "step": 18764
    },
    {
      "epoch": 0.7776948899664304,
      "grad_norm": 0.42251867055892944,
      "learning_rate": 1.1117327696962162e-06,
      "loss": 0.6842,
      "step": 18765
    },
    {
      "epoch": 0.7777363338721041,
      "grad_norm": 0.4526134133338928,
      "learning_rate": 1.1115255501678478e-06,
      "loss": 0.6309,
      "step": 18766
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 0.4456314742565155,
      "learning_rate": 1.1113183306394794e-06,
      "loss": 0.6407,
      "step": 18767
    },
    {
      "epoch": 0.7778192216834514,
      "grad_norm": 0.40142011642456055,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.6531,
      "step": 18768
    },
    {
      "epoch": 0.7778606655891251,
      "grad_norm": 0.4107346832752228,
      "learning_rate": 1.1109038915827428e-06,
      "loss": 0.6613,
      "step": 18769
    },
    {
      "epoch": 0.7779021094947988,
      "grad_norm": 0.3942998945713043,
      "learning_rate": 1.1106966720543744e-06,
      "loss": 0.6348,
      "step": 18770
    },
    {
      "epoch": 0.7779435534004725,
      "grad_norm": 0.4623507261276245,
      "learning_rate": 1.110489452526006e-06,
      "loss": 0.7013,
      "step": 18771
    },
    {
      "epoch": 0.7779849973061461,
      "grad_norm": 0.41602858901023865,
      "learning_rate": 1.1102822329976376e-06,
      "loss": 0.6753,
      "step": 18772
    },
    {
      "epoch": 0.7780264412118199,
      "grad_norm": 0.41368016600608826,
      "learning_rate": 1.1100750134692694e-06,
      "loss": 0.6238,
      "step": 18773
    },
    {
      "epoch": 0.7780678851174935,
      "grad_norm": 0.4194434881210327,
      "learning_rate": 1.109867793940901e-06,
      "loss": 0.6848,
      "step": 18774
    },
    {
      "epoch": 0.7781093290231671,
      "grad_norm": 0.47933462262153625,
      "learning_rate": 1.1096605744125326e-06,
      "loss": 0.7034,
      "step": 18775
    },
    {
      "epoch": 0.7781507729288408,
      "grad_norm": 0.4257299304008484,
      "learning_rate": 1.1094533548841642e-06,
      "loss": 0.6974,
      "step": 18776
    },
    {
      "epoch": 0.7781922168345144,
      "grad_norm": 0.4096745252609253,
      "learning_rate": 1.109246135355796e-06,
      "loss": 0.7034,
      "step": 18777
    },
    {
      "epoch": 0.7782336607401882,
      "grad_norm": 0.4046975076198578,
      "learning_rate": 1.1090389158274276e-06,
      "loss": 0.6281,
      "step": 18778
    },
    {
      "epoch": 0.7782751046458618,
      "grad_norm": 0.44408655166625977,
      "learning_rate": 1.1088316962990592e-06,
      "loss": 0.699,
      "step": 18779
    },
    {
      "epoch": 0.7783165485515355,
      "grad_norm": 0.4406507909297943,
      "learning_rate": 1.1086244767706908e-06,
      "loss": 0.6875,
      "step": 18780
    },
    {
      "epoch": 0.7783579924572092,
      "grad_norm": 0.40251192450523376,
      "learning_rate": 1.1084172572423227e-06,
      "loss": 0.6609,
      "step": 18781
    },
    {
      "epoch": 0.7783994363628829,
      "grad_norm": 0.38690513372421265,
      "learning_rate": 1.1082100377139542e-06,
      "loss": 0.6295,
      "step": 18782
    },
    {
      "epoch": 0.7784408802685565,
      "grad_norm": 0.4032716751098633,
      "learning_rate": 1.1080028181855858e-06,
      "loss": 0.6478,
      "step": 18783
    },
    {
      "epoch": 0.7784823241742301,
      "grad_norm": 0.391605406999588,
      "learning_rate": 1.1077955986572174e-06,
      "loss": 0.6782,
      "step": 18784
    },
    {
      "epoch": 0.7785237680799039,
      "grad_norm": 0.42940467596054077,
      "learning_rate": 1.107588379128849e-06,
      "loss": 0.6954,
      "step": 18785
    },
    {
      "epoch": 0.7785652119855775,
      "grad_norm": 0.47324827313423157,
      "learning_rate": 1.1073811596004808e-06,
      "loss": 0.7213,
      "step": 18786
    },
    {
      "epoch": 0.7786066558912512,
      "grad_norm": 0.4105950891971588,
      "learning_rate": 1.1071739400721124e-06,
      "loss": 0.6704,
      "step": 18787
    },
    {
      "epoch": 0.7786480997969248,
      "grad_norm": 0.43260952830314636,
      "learning_rate": 1.106966720543744e-06,
      "loss": 0.6833,
      "step": 18788
    },
    {
      "epoch": 0.7786895437025986,
      "grad_norm": 0.41201263666152954,
      "learning_rate": 1.1067595010153756e-06,
      "loss": 0.6041,
      "step": 18789
    },
    {
      "epoch": 0.7787309876082722,
      "grad_norm": 0.4050266444683075,
      "learning_rate": 1.1065522814870075e-06,
      "loss": 0.675,
      "step": 18790
    },
    {
      "epoch": 0.7787724315139459,
      "grad_norm": 0.41364237666130066,
      "learning_rate": 1.106345061958639e-06,
      "loss": 0.6405,
      "step": 18791
    },
    {
      "epoch": 0.7788138754196196,
      "grad_norm": 0.3975664973258972,
      "learning_rate": 1.1061378424302706e-06,
      "loss": 0.6594,
      "step": 18792
    },
    {
      "epoch": 0.7788553193252932,
      "grad_norm": 0.4294144809246063,
      "learning_rate": 1.1059306229019022e-06,
      "loss": 0.6395,
      "step": 18793
    },
    {
      "epoch": 0.7788967632309669,
      "grad_norm": 0.39942771196365356,
      "learning_rate": 1.105723403373534e-06,
      "loss": 0.6631,
      "step": 18794
    },
    {
      "epoch": 0.7789382071366405,
      "grad_norm": 0.4126463830471039,
      "learning_rate": 1.1055161838451656e-06,
      "loss": 0.6553,
      "step": 18795
    },
    {
      "epoch": 0.7789796510423143,
      "grad_norm": 0.4053286910057068,
      "learning_rate": 1.1053089643167972e-06,
      "loss": 0.6354,
      "step": 18796
    },
    {
      "epoch": 0.7790210949479879,
      "grad_norm": 0.3916606605052948,
      "learning_rate": 1.1051017447884288e-06,
      "loss": 0.6975,
      "step": 18797
    },
    {
      "epoch": 0.7790625388536616,
      "grad_norm": 0.42304617166519165,
      "learning_rate": 1.1048945252600607e-06,
      "loss": 0.7141,
      "step": 18798
    },
    {
      "epoch": 0.7791039827593352,
      "grad_norm": 0.39214229583740234,
      "learning_rate": 1.1046873057316923e-06,
      "loss": 0.6514,
      "step": 18799
    },
    {
      "epoch": 0.7791454266650089,
      "grad_norm": 0.4053260087966919,
      "learning_rate": 1.1044800862033238e-06,
      "loss": 0.7063,
      "step": 18800
    },
    {
      "epoch": 0.7791868705706826,
      "grad_norm": 0.38826656341552734,
      "learning_rate": 1.1042728666749554e-06,
      "loss": 0.6624,
      "step": 18801
    },
    {
      "epoch": 0.7792283144763562,
      "grad_norm": 0.4081454575061798,
      "learning_rate": 1.1040656471465873e-06,
      "loss": 0.7229,
      "step": 18802
    },
    {
      "epoch": 0.7792697583820299,
      "grad_norm": 0.4236225485801697,
      "learning_rate": 1.1038584276182189e-06,
      "loss": 0.7153,
      "step": 18803
    },
    {
      "epoch": 0.7793112022877036,
      "grad_norm": 0.3959042429924011,
      "learning_rate": 1.1036512080898504e-06,
      "loss": 0.689,
      "step": 18804
    },
    {
      "epoch": 0.7793526461933773,
      "grad_norm": 0.41856518387794495,
      "learning_rate": 1.103443988561482e-06,
      "loss": 0.6355,
      "step": 18805
    },
    {
      "epoch": 0.7793940900990509,
      "grad_norm": 0.40799281001091003,
      "learning_rate": 1.1032367690331136e-06,
      "loss": 0.6873,
      "step": 18806
    },
    {
      "epoch": 0.7794355340047247,
      "grad_norm": 0.4445628523826599,
      "learning_rate": 1.1030295495047455e-06,
      "loss": 0.663,
      "step": 18807
    },
    {
      "epoch": 0.7794769779103983,
      "grad_norm": 0.3994918167591095,
      "learning_rate": 1.102822329976377e-06,
      "loss": 0.677,
      "step": 18808
    },
    {
      "epoch": 0.7795184218160719,
      "grad_norm": 0.4263248145580292,
      "learning_rate": 1.1026151104480086e-06,
      "loss": 0.6786,
      "step": 18809
    },
    {
      "epoch": 0.7795598657217456,
      "grad_norm": 0.3844582438468933,
      "learning_rate": 1.1024078909196402e-06,
      "loss": 0.6617,
      "step": 18810
    },
    {
      "epoch": 0.7796013096274192,
      "grad_norm": 0.4311043918132782,
      "learning_rate": 1.102200671391272e-06,
      "loss": 0.6512,
      "step": 18811
    },
    {
      "epoch": 0.779642753533093,
      "grad_norm": 0.41674917936325073,
      "learning_rate": 1.1019934518629037e-06,
      "loss": 0.6864,
      "step": 18812
    },
    {
      "epoch": 0.7796841974387666,
      "grad_norm": 0.4419926702976227,
      "learning_rate": 1.1017862323345353e-06,
      "loss": 0.6744,
      "step": 18813
    },
    {
      "epoch": 0.7797256413444403,
      "grad_norm": 0.3851330578327179,
      "learning_rate": 1.1015790128061668e-06,
      "loss": 0.6531,
      "step": 18814
    },
    {
      "epoch": 0.779767085250114,
      "grad_norm": 0.3952349126338959,
      "learning_rate": 1.1013717932777987e-06,
      "loss": 0.6786,
      "step": 18815
    },
    {
      "epoch": 0.7798085291557877,
      "grad_norm": 0.44760245084762573,
      "learning_rate": 1.1011645737494303e-06,
      "loss": 0.6631,
      "step": 18816
    },
    {
      "epoch": 0.7798499730614613,
      "grad_norm": 0.42007654905319214,
      "learning_rate": 1.1009573542210619e-06,
      "loss": 0.6587,
      "step": 18817
    },
    {
      "epoch": 0.7798914169671349,
      "grad_norm": 0.39179566502571106,
      "learning_rate": 1.1007501346926934e-06,
      "loss": 0.6769,
      "step": 18818
    },
    {
      "epoch": 0.7799328608728087,
      "grad_norm": 0.39047813415527344,
      "learning_rate": 1.1005429151643253e-06,
      "loss": 0.6605,
      "step": 18819
    },
    {
      "epoch": 0.7799743047784823,
      "grad_norm": 0.40700316429138184,
      "learning_rate": 1.1003356956359569e-06,
      "loss": 0.6892,
      "step": 18820
    },
    {
      "epoch": 0.780015748684156,
      "grad_norm": 0.3896768391132355,
      "learning_rate": 1.1001284761075885e-06,
      "loss": 0.6543,
      "step": 18821
    },
    {
      "epoch": 0.7800571925898296,
      "grad_norm": 0.3938610553741455,
      "learning_rate": 1.09992125657922e-06,
      "loss": 0.6461,
      "step": 18822
    },
    {
      "epoch": 0.7800986364955034,
      "grad_norm": 0.4198577404022217,
      "learning_rate": 1.0997140370508516e-06,
      "loss": 0.6597,
      "step": 18823
    },
    {
      "epoch": 0.780140080401177,
      "grad_norm": 0.41655710339546204,
      "learning_rate": 1.0995068175224835e-06,
      "loss": 0.5928,
      "step": 18824
    },
    {
      "epoch": 0.7801815243068507,
      "grad_norm": 0.3947398364543915,
      "learning_rate": 1.099299597994115e-06,
      "loss": 0.6196,
      "step": 18825
    },
    {
      "epoch": 0.7802229682125243,
      "grad_norm": 0.40593814849853516,
      "learning_rate": 1.0990923784657467e-06,
      "loss": 0.6841,
      "step": 18826
    },
    {
      "epoch": 0.780264412118198,
      "grad_norm": 0.4076291024684906,
      "learning_rate": 1.0988851589373782e-06,
      "loss": 0.6652,
      "step": 18827
    },
    {
      "epoch": 0.7803058560238717,
      "grad_norm": 0.44146642088890076,
      "learning_rate": 1.09867793940901e-06,
      "loss": 0.6241,
      "step": 18828
    },
    {
      "epoch": 0.7803472999295453,
      "grad_norm": 0.45738741755485535,
      "learning_rate": 1.0984707198806417e-06,
      "loss": 0.6708,
      "step": 18829
    },
    {
      "epoch": 0.7803887438352191,
      "grad_norm": 0.4379986822605133,
      "learning_rate": 1.0982635003522733e-06,
      "loss": 0.6901,
      "step": 18830
    },
    {
      "epoch": 0.7804301877408927,
      "grad_norm": 0.3872493803501129,
      "learning_rate": 1.0980562808239049e-06,
      "loss": 0.6398,
      "step": 18831
    },
    {
      "epoch": 0.7804716316465664,
      "grad_norm": 0.4121597707271576,
      "learning_rate": 1.0978490612955367e-06,
      "loss": 0.6982,
      "step": 18832
    },
    {
      "epoch": 0.78051307555224,
      "grad_norm": 0.4347425401210785,
      "learning_rate": 1.0976418417671683e-06,
      "loss": 0.6331,
      "step": 18833
    },
    {
      "epoch": 0.7805545194579138,
      "grad_norm": 0.40419870615005493,
      "learning_rate": 1.0974346222387999e-06,
      "loss": 0.6638,
      "step": 18834
    },
    {
      "epoch": 0.7805959633635874,
      "grad_norm": 0.41031917929649353,
      "learning_rate": 1.0972274027104315e-06,
      "loss": 0.6846,
      "step": 18835
    },
    {
      "epoch": 0.780637407269261,
      "grad_norm": 0.411199688911438,
      "learning_rate": 1.0970201831820633e-06,
      "loss": 0.6553,
      "step": 18836
    },
    {
      "epoch": 0.7806788511749347,
      "grad_norm": 0.3981885612010956,
      "learning_rate": 1.0968129636536949e-06,
      "loss": 0.5975,
      "step": 18837
    },
    {
      "epoch": 0.7807202950806084,
      "grad_norm": 0.399404913187027,
      "learning_rate": 1.0966057441253265e-06,
      "loss": 0.6232,
      "step": 18838
    },
    {
      "epoch": 0.7807617389862821,
      "grad_norm": 0.4402633011341095,
      "learning_rate": 1.096398524596958e-06,
      "loss": 0.6982,
      "step": 18839
    },
    {
      "epoch": 0.7808031828919557,
      "grad_norm": 0.4260307848453522,
      "learning_rate": 1.0961913050685897e-06,
      "loss": 0.7595,
      "step": 18840
    },
    {
      "epoch": 0.7808446267976294,
      "grad_norm": 0.4659217596054077,
      "learning_rate": 1.0959840855402215e-06,
      "loss": 0.7234,
      "step": 18841
    },
    {
      "epoch": 0.7808860707033031,
      "grad_norm": 0.4123915433883667,
      "learning_rate": 1.095776866011853e-06,
      "loss": 0.6493,
      "step": 18842
    },
    {
      "epoch": 0.7809275146089768,
      "grad_norm": 0.4259832799434662,
      "learning_rate": 1.0955696464834847e-06,
      "loss": 0.6912,
      "step": 18843
    },
    {
      "epoch": 0.7809689585146504,
      "grad_norm": 0.4359184205532074,
      "learning_rate": 1.0953624269551163e-06,
      "loss": 0.6946,
      "step": 18844
    },
    {
      "epoch": 0.781010402420324,
      "grad_norm": 0.3814309239387512,
      "learning_rate": 1.095155207426748e-06,
      "loss": 0.6674,
      "step": 18845
    },
    {
      "epoch": 0.7810518463259978,
      "grad_norm": 0.4515441060066223,
      "learning_rate": 1.0949479878983797e-06,
      "loss": 0.6339,
      "step": 18846
    },
    {
      "epoch": 0.7810932902316714,
      "grad_norm": 0.4213552474975586,
      "learning_rate": 1.0947407683700113e-06,
      "loss": 0.7195,
      "step": 18847
    },
    {
      "epoch": 0.7811347341373451,
      "grad_norm": 0.42695024609565735,
      "learning_rate": 1.0945335488416429e-06,
      "loss": 0.6851,
      "step": 18848
    },
    {
      "epoch": 0.7811761780430188,
      "grad_norm": 0.39487844705581665,
      "learning_rate": 1.0943263293132747e-06,
      "loss": 0.6672,
      "step": 18849
    },
    {
      "epoch": 0.7812176219486925,
      "grad_norm": 0.41849663853645325,
      "learning_rate": 1.0941191097849063e-06,
      "loss": 0.6985,
      "step": 18850
    },
    {
      "epoch": 0.7812590658543661,
      "grad_norm": 0.3849886953830719,
      "learning_rate": 1.0939118902565379e-06,
      "loss": 0.6757,
      "step": 18851
    },
    {
      "epoch": 0.7813005097600398,
      "grad_norm": 0.4149416387081146,
      "learning_rate": 1.0937046707281695e-06,
      "loss": 0.709,
      "step": 18852
    },
    {
      "epoch": 0.7813419536657135,
      "grad_norm": 0.42889589071273804,
      "learning_rate": 1.0934974511998013e-06,
      "loss": 0.7078,
      "step": 18853
    },
    {
      "epoch": 0.7813833975713871,
      "grad_norm": 0.40220752358436584,
      "learning_rate": 1.0932902316714329e-06,
      "loss": 0.646,
      "step": 18854
    },
    {
      "epoch": 0.7814248414770608,
      "grad_norm": 0.4418937563896179,
      "learning_rate": 1.0930830121430645e-06,
      "loss": 0.6716,
      "step": 18855
    },
    {
      "epoch": 0.7814662853827344,
      "grad_norm": 0.3934852182865143,
      "learning_rate": 1.092875792614696e-06,
      "loss": 0.6556,
      "step": 18856
    },
    {
      "epoch": 0.7815077292884082,
      "grad_norm": 0.4429916739463806,
      "learning_rate": 1.0926685730863279e-06,
      "loss": 0.7279,
      "step": 18857
    },
    {
      "epoch": 0.7815491731940818,
      "grad_norm": 0.41080379486083984,
      "learning_rate": 1.0924613535579595e-06,
      "loss": 0.6807,
      "step": 18858
    },
    {
      "epoch": 0.7815906170997555,
      "grad_norm": 0.434968501329422,
      "learning_rate": 1.092254134029591e-06,
      "loss": 0.7236,
      "step": 18859
    },
    {
      "epoch": 0.7816320610054291,
      "grad_norm": 0.4013499319553375,
      "learning_rate": 1.0920469145012227e-06,
      "loss": 0.6716,
      "step": 18860
    },
    {
      "epoch": 0.7816735049111028,
      "grad_norm": 0.44486042857170105,
      "learning_rate": 1.0918396949728543e-06,
      "loss": 0.5999,
      "step": 18861
    },
    {
      "epoch": 0.7817149488167765,
      "grad_norm": 0.4249608814716339,
      "learning_rate": 1.091632475444486e-06,
      "loss": 0.6539,
      "step": 18862
    },
    {
      "epoch": 0.7817563927224501,
      "grad_norm": 0.3992757499217987,
      "learning_rate": 1.0914252559161177e-06,
      "loss": 0.6785,
      "step": 18863
    },
    {
      "epoch": 0.7817978366281239,
      "grad_norm": 0.4159892201423645,
      "learning_rate": 1.0912180363877493e-06,
      "loss": 0.6508,
      "step": 18864
    },
    {
      "epoch": 0.7818392805337975,
      "grad_norm": 0.38451892137527466,
      "learning_rate": 1.0910108168593809e-06,
      "loss": 0.6392,
      "step": 18865
    },
    {
      "epoch": 0.7818807244394712,
      "grad_norm": 0.46010440587997437,
      "learning_rate": 1.0908035973310127e-06,
      "loss": 0.7175,
      "step": 18866
    },
    {
      "epoch": 0.7819221683451448,
      "grad_norm": 0.40840089321136475,
      "learning_rate": 1.0905963778026443e-06,
      "loss": 0.6538,
      "step": 18867
    },
    {
      "epoch": 0.7819636122508186,
      "grad_norm": 0.4650651514530182,
      "learning_rate": 1.0903891582742759e-06,
      "loss": 0.6581,
      "step": 18868
    },
    {
      "epoch": 0.7820050561564922,
      "grad_norm": 0.4152011573314667,
      "learning_rate": 1.0901819387459075e-06,
      "loss": 0.6545,
      "step": 18869
    },
    {
      "epoch": 0.7820465000621658,
      "grad_norm": 0.45975184440612793,
      "learning_rate": 1.0899747192175393e-06,
      "loss": 0.6414,
      "step": 18870
    },
    {
      "epoch": 0.7820879439678395,
      "grad_norm": 0.39851266145706177,
      "learning_rate": 1.0897674996891709e-06,
      "loss": 0.6505,
      "step": 18871
    },
    {
      "epoch": 0.7821293878735132,
      "grad_norm": 0.38592639565467834,
      "learning_rate": 1.0895602801608025e-06,
      "loss": 0.6677,
      "step": 18872
    },
    {
      "epoch": 0.7821708317791869,
      "grad_norm": 0.42993760108947754,
      "learning_rate": 1.089353060632434e-06,
      "loss": 0.6533,
      "step": 18873
    },
    {
      "epoch": 0.7822122756848605,
      "grad_norm": 0.4233139753341675,
      "learning_rate": 1.0891458411040659e-06,
      "loss": 0.7239,
      "step": 18874
    },
    {
      "epoch": 0.7822537195905342,
      "grad_norm": 0.4279825687408447,
      "learning_rate": 1.0889386215756975e-06,
      "loss": 0.6892,
      "step": 18875
    },
    {
      "epoch": 0.7822951634962079,
      "grad_norm": 0.4484666883945465,
      "learning_rate": 1.088731402047329e-06,
      "loss": 0.7579,
      "step": 18876
    },
    {
      "epoch": 0.7823366074018816,
      "grad_norm": 0.4028109908103943,
      "learning_rate": 1.0885241825189607e-06,
      "loss": 0.6626,
      "step": 18877
    },
    {
      "epoch": 0.7823780513075552,
      "grad_norm": 0.43316707015037537,
      "learning_rate": 1.0883169629905923e-06,
      "loss": 0.6704,
      "step": 18878
    },
    {
      "epoch": 0.7824194952132288,
      "grad_norm": 0.4176410436630249,
      "learning_rate": 1.088109743462224e-06,
      "loss": 0.6703,
      "step": 18879
    },
    {
      "epoch": 0.7824609391189026,
      "grad_norm": 0.43516579270362854,
      "learning_rate": 1.0879025239338557e-06,
      "loss": 0.6943,
      "step": 18880
    },
    {
      "epoch": 0.7825023830245762,
      "grad_norm": 0.41555994749069214,
      "learning_rate": 1.0876953044054873e-06,
      "loss": 0.668,
      "step": 18881
    },
    {
      "epoch": 0.7825438269302499,
      "grad_norm": 0.40481501817703247,
      "learning_rate": 1.0874880848771189e-06,
      "loss": 0.6858,
      "step": 18882
    },
    {
      "epoch": 0.7825852708359236,
      "grad_norm": 0.42684054374694824,
      "learning_rate": 1.0872808653487507e-06,
      "loss": 0.6475,
      "step": 18883
    },
    {
      "epoch": 0.7826267147415973,
      "grad_norm": 0.4245835542678833,
      "learning_rate": 1.0870736458203823e-06,
      "loss": 0.6595,
      "step": 18884
    },
    {
      "epoch": 0.7826681586472709,
      "grad_norm": 0.38963648676872253,
      "learning_rate": 1.0868664262920139e-06,
      "loss": 0.6367,
      "step": 18885
    },
    {
      "epoch": 0.7827096025529446,
      "grad_norm": 0.4163943827152252,
      "learning_rate": 1.0866592067636455e-06,
      "loss": 0.6882,
      "step": 18886
    },
    {
      "epoch": 0.7827510464586183,
      "grad_norm": 0.4186316430568695,
      "learning_rate": 1.0864519872352773e-06,
      "loss": 0.6299,
      "step": 18887
    },
    {
      "epoch": 0.7827924903642919,
      "grad_norm": 0.4664421081542969,
      "learning_rate": 1.0862447677069089e-06,
      "loss": 0.7148,
      "step": 18888
    },
    {
      "epoch": 0.7828339342699656,
      "grad_norm": 0.4560980498790741,
      "learning_rate": 1.0860375481785405e-06,
      "loss": 0.6732,
      "step": 18889
    },
    {
      "epoch": 0.7828753781756392,
      "grad_norm": 0.41615772247314453,
      "learning_rate": 1.085830328650172e-06,
      "loss": 0.6533,
      "step": 18890
    },
    {
      "epoch": 0.782916822081313,
      "grad_norm": 0.42156779766082764,
      "learning_rate": 1.0856231091218039e-06,
      "loss": 0.6724,
      "step": 18891
    },
    {
      "epoch": 0.7829582659869866,
      "grad_norm": 0.39790910482406616,
      "learning_rate": 1.0854158895934355e-06,
      "loss": 0.6733,
      "step": 18892
    },
    {
      "epoch": 0.7829997098926603,
      "grad_norm": 0.4292767345905304,
      "learning_rate": 1.085208670065067e-06,
      "loss": 0.7246,
      "step": 18893
    },
    {
      "epoch": 0.7830411537983339,
      "grad_norm": 0.40560129284858704,
      "learning_rate": 1.0850014505366987e-06,
      "loss": 0.6714,
      "step": 18894
    },
    {
      "epoch": 0.7830825977040077,
      "grad_norm": 0.37918904423713684,
      "learning_rate": 1.0847942310083305e-06,
      "loss": 0.6311,
      "step": 18895
    },
    {
      "epoch": 0.7831240416096813,
      "grad_norm": 0.3928378224372864,
      "learning_rate": 1.084587011479962e-06,
      "loss": 0.6541,
      "step": 18896
    },
    {
      "epoch": 0.7831654855153549,
      "grad_norm": 0.4097030460834503,
      "learning_rate": 1.0843797919515937e-06,
      "loss": 0.7316,
      "step": 18897
    },
    {
      "epoch": 0.7832069294210287,
      "grad_norm": 0.4423655867576599,
      "learning_rate": 1.0841725724232253e-06,
      "loss": 0.6665,
      "step": 18898
    },
    {
      "epoch": 0.7832483733267023,
      "grad_norm": 0.4249981939792633,
      "learning_rate": 1.0839653528948569e-06,
      "loss": 0.6824,
      "step": 18899
    },
    {
      "epoch": 0.783289817232376,
      "grad_norm": 0.43631815910339355,
      "learning_rate": 1.0837581333664887e-06,
      "loss": 0.6525,
      "step": 18900
    },
    {
      "epoch": 0.7833312611380496,
      "grad_norm": 0.42427247762680054,
      "learning_rate": 1.0835509138381203e-06,
      "loss": 0.6039,
      "step": 18901
    },
    {
      "epoch": 0.7833727050437234,
      "grad_norm": 0.43400827050209045,
      "learning_rate": 1.0833436943097519e-06,
      "loss": 0.6584,
      "step": 18902
    },
    {
      "epoch": 0.783414148949397,
      "grad_norm": 0.4354948103427887,
      "learning_rate": 1.0831364747813835e-06,
      "loss": 0.6282,
      "step": 18903
    },
    {
      "epoch": 0.7834555928550707,
      "grad_norm": 0.3904609978199005,
      "learning_rate": 1.0829292552530153e-06,
      "loss": 0.6403,
      "step": 18904
    },
    {
      "epoch": 0.7834970367607443,
      "grad_norm": 0.42187222838401794,
      "learning_rate": 1.0827220357246469e-06,
      "loss": 0.6869,
      "step": 18905
    },
    {
      "epoch": 0.783538480666418,
      "grad_norm": 0.43158531188964844,
      "learning_rate": 1.0825148161962785e-06,
      "loss": 0.6466,
      "step": 18906
    },
    {
      "epoch": 0.7835799245720917,
      "grad_norm": 0.4186829924583435,
      "learning_rate": 1.08230759666791e-06,
      "loss": 0.6661,
      "step": 18907
    },
    {
      "epoch": 0.7836213684777653,
      "grad_norm": 0.4584585726261139,
      "learning_rate": 1.0821003771395419e-06,
      "loss": 0.6927,
      "step": 18908
    },
    {
      "epoch": 0.783662812383439,
      "grad_norm": 0.4755858778953552,
      "learning_rate": 1.0818931576111735e-06,
      "loss": 0.71,
      "step": 18909
    },
    {
      "epoch": 0.7837042562891127,
      "grad_norm": 0.3958633542060852,
      "learning_rate": 1.081685938082805e-06,
      "loss": 0.6328,
      "step": 18910
    },
    {
      "epoch": 0.7837457001947864,
      "grad_norm": 0.4167976677417755,
      "learning_rate": 1.0814787185544367e-06,
      "loss": 0.6659,
      "step": 18911
    },
    {
      "epoch": 0.78378714410046,
      "grad_norm": 0.3970787525177002,
      "learning_rate": 1.0812714990260685e-06,
      "loss": 0.6467,
      "step": 18912
    },
    {
      "epoch": 0.7838285880061338,
      "grad_norm": 0.41433587670326233,
      "learning_rate": 1.0810642794977e-06,
      "loss": 0.6832,
      "step": 18913
    },
    {
      "epoch": 0.7838700319118074,
      "grad_norm": 0.41205278038978577,
      "learning_rate": 1.0808570599693317e-06,
      "loss": 0.6411,
      "step": 18914
    },
    {
      "epoch": 0.783911475817481,
      "grad_norm": 0.43575945496559143,
      "learning_rate": 1.0806498404409633e-06,
      "loss": 0.6903,
      "step": 18915
    },
    {
      "epoch": 0.7839529197231547,
      "grad_norm": 0.40554696321487427,
      "learning_rate": 1.0804426209125949e-06,
      "loss": 0.6704,
      "step": 18916
    },
    {
      "epoch": 0.7839943636288284,
      "grad_norm": 0.4216777980327606,
      "learning_rate": 1.0802354013842267e-06,
      "loss": 0.6903,
      "step": 18917
    },
    {
      "epoch": 0.7840358075345021,
      "grad_norm": 0.3853289783000946,
      "learning_rate": 1.0800281818558583e-06,
      "loss": 0.6567,
      "step": 18918
    },
    {
      "epoch": 0.7840772514401757,
      "grad_norm": 0.3791319727897644,
      "learning_rate": 1.0798209623274899e-06,
      "loss": 0.6383,
      "step": 18919
    },
    {
      "epoch": 0.7841186953458494,
      "grad_norm": 0.39951223134994507,
      "learning_rate": 1.0796137427991215e-06,
      "loss": 0.6577,
      "step": 18920
    },
    {
      "epoch": 0.7841601392515231,
      "grad_norm": 0.4251583218574524,
      "learning_rate": 1.0794065232707533e-06,
      "loss": 0.6942,
      "step": 18921
    },
    {
      "epoch": 0.7842015831571967,
      "grad_norm": 0.39110326766967773,
      "learning_rate": 1.0791993037423849e-06,
      "loss": 0.6665,
      "step": 18922
    },
    {
      "epoch": 0.7842430270628704,
      "grad_norm": 0.4191063642501831,
      "learning_rate": 1.0789920842140165e-06,
      "loss": 0.6451,
      "step": 18923
    },
    {
      "epoch": 0.784284470968544,
      "grad_norm": 0.37308424711227417,
      "learning_rate": 1.078784864685648e-06,
      "loss": 0.6638,
      "step": 18924
    },
    {
      "epoch": 0.7843259148742178,
      "grad_norm": 0.4515168070793152,
      "learning_rate": 1.0785776451572799e-06,
      "loss": 0.6923,
      "step": 18925
    },
    {
      "epoch": 0.7843673587798914,
      "grad_norm": 0.4061039090156555,
      "learning_rate": 1.0783704256289115e-06,
      "loss": 0.6511,
      "step": 18926
    },
    {
      "epoch": 0.7844088026855651,
      "grad_norm": 0.4120289385318756,
      "learning_rate": 1.078163206100543e-06,
      "loss": 0.6798,
      "step": 18927
    },
    {
      "epoch": 0.7844502465912387,
      "grad_norm": 0.4173102378845215,
      "learning_rate": 1.0779559865721747e-06,
      "loss": 0.7458,
      "step": 18928
    },
    {
      "epoch": 0.7844916904969125,
      "grad_norm": 0.4414609968662262,
      "learning_rate": 1.0777487670438063e-06,
      "loss": 0.6855,
      "step": 18929
    },
    {
      "epoch": 0.7845331344025861,
      "grad_norm": 0.4301402270793915,
      "learning_rate": 1.077541547515438e-06,
      "loss": 0.6631,
      "step": 18930
    },
    {
      "epoch": 0.7845745783082597,
      "grad_norm": 0.4159611463546753,
      "learning_rate": 1.0773343279870697e-06,
      "loss": 0.6827,
      "step": 18931
    },
    {
      "epoch": 0.7846160222139335,
      "grad_norm": 0.39676064252853394,
      "learning_rate": 1.0771271084587013e-06,
      "loss": 0.647,
      "step": 18932
    },
    {
      "epoch": 0.7846574661196071,
      "grad_norm": 0.446014940738678,
      "learning_rate": 1.0769198889303329e-06,
      "loss": 0.6962,
      "step": 18933
    },
    {
      "epoch": 0.7846989100252808,
      "grad_norm": 0.44470736384391785,
      "learning_rate": 1.0767126694019647e-06,
      "loss": 0.7288,
      "step": 18934
    },
    {
      "epoch": 0.7847403539309544,
      "grad_norm": 0.46436241269111633,
      "learning_rate": 1.0765054498735963e-06,
      "loss": 0.6948,
      "step": 18935
    },
    {
      "epoch": 0.7847817978366282,
      "grad_norm": 0.4103720188140869,
      "learning_rate": 1.0762982303452279e-06,
      "loss": 0.667,
      "step": 18936
    },
    {
      "epoch": 0.7848232417423018,
      "grad_norm": 0.44861122965812683,
      "learning_rate": 1.0760910108168595e-06,
      "loss": 0.6787,
      "step": 18937
    },
    {
      "epoch": 0.7848646856479755,
      "grad_norm": 0.4206666946411133,
      "learning_rate": 1.075883791288491e-06,
      "loss": 0.7148,
      "step": 18938
    },
    {
      "epoch": 0.7849061295536491,
      "grad_norm": 0.41205519437789917,
      "learning_rate": 1.0756765717601229e-06,
      "loss": 0.7126,
      "step": 18939
    },
    {
      "epoch": 0.7849475734593228,
      "grad_norm": 0.42302823066711426,
      "learning_rate": 1.0754693522317545e-06,
      "loss": 0.7236,
      "step": 18940
    },
    {
      "epoch": 0.7849890173649965,
      "grad_norm": 0.4126865565776825,
      "learning_rate": 1.075262132703386e-06,
      "loss": 0.6769,
      "step": 18941
    },
    {
      "epoch": 0.7850304612706701,
      "grad_norm": 0.4498770833015442,
      "learning_rate": 1.0750549131750177e-06,
      "loss": 0.6456,
      "step": 18942
    },
    {
      "epoch": 0.7850719051763438,
      "grad_norm": 0.4283280074596405,
      "learning_rate": 1.0748476936466495e-06,
      "loss": 0.689,
      "step": 18943
    },
    {
      "epoch": 0.7851133490820175,
      "grad_norm": 0.4110310971736908,
      "learning_rate": 1.074640474118281e-06,
      "loss": 0.7446,
      "step": 18944
    },
    {
      "epoch": 0.7851547929876912,
      "grad_norm": 0.3797263503074646,
      "learning_rate": 1.0744332545899127e-06,
      "loss": 0.6619,
      "step": 18945
    },
    {
      "epoch": 0.7851962368933648,
      "grad_norm": 0.44340941309928894,
      "learning_rate": 1.0742260350615443e-06,
      "loss": 0.652,
      "step": 18946
    },
    {
      "epoch": 0.7852376807990386,
      "grad_norm": 0.4089241325855255,
      "learning_rate": 1.074018815533176e-06,
      "loss": 0.6622,
      "step": 18947
    },
    {
      "epoch": 0.7852791247047122,
      "grad_norm": 0.43524765968322754,
      "learning_rate": 1.0738115960048077e-06,
      "loss": 0.7211,
      "step": 18948
    },
    {
      "epoch": 0.7853205686103858,
      "grad_norm": 0.3743706941604614,
      "learning_rate": 1.0736043764764393e-06,
      "loss": 0.653,
      "step": 18949
    },
    {
      "epoch": 0.7853620125160595,
      "grad_norm": 0.41213884949684143,
      "learning_rate": 1.0733971569480709e-06,
      "loss": 0.7084,
      "step": 18950
    },
    {
      "epoch": 0.7854034564217331,
      "grad_norm": 0.42825594544410706,
      "learning_rate": 1.0731899374197025e-06,
      "loss": 0.7109,
      "step": 18951
    },
    {
      "epoch": 0.7854449003274069,
      "grad_norm": 0.4252018332481384,
      "learning_rate": 1.0729827178913343e-06,
      "loss": 0.6841,
      "step": 18952
    },
    {
      "epoch": 0.7854863442330805,
      "grad_norm": 0.4036179482936859,
      "learning_rate": 1.0727754983629659e-06,
      "loss": 0.6082,
      "step": 18953
    },
    {
      "epoch": 0.7855277881387542,
      "grad_norm": 0.42404064536094666,
      "learning_rate": 1.0725682788345975e-06,
      "loss": 0.6345,
      "step": 18954
    },
    {
      "epoch": 0.7855692320444279,
      "grad_norm": 0.4236163794994354,
      "learning_rate": 1.072361059306229e-06,
      "loss": 0.6759,
      "step": 18955
    },
    {
      "epoch": 0.7856106759501016,
      "grad_norm": 0.38917067646980286,
      "learning_rate": 1.0721538397778609e-06,
      "loss": 0.6542,
      "step": 18956
    },
    {
      "epoch": 0.7856521198557752,
      "grad_norm": 0.4344775676727295,
      "learning_rate": 1.0719466202494925e-06,
      "loss": 0.6478,
      "step": 18957
    },
    {
      "epoch": 0.7856935637614488,
      "grad_norm": 0.441770076751709,
      "learning_rate": 1.071739400721124e-06,
      "loss": 0.6466,
      "step": 18958
    },
    {
      "epoch": 0.7857350076671226,
      "grad_norm": 0.4167284369468689,
      "learning_rate": 1.0715321811927557e-06,
      "loss": 0.6929,
      "step": 18959
    },
    {
      "epoch": 0.7857764515727962,
      "grad_norm": 0.43887925148010254,
      "learning_rate": 1.0713249616643875e-06,
      "loss": 0.6753,
      "step": 18960
    },
    {
      "epoch": 0.7858178954784699,
      "grad_norm": 0.4397631883621216,
      "learning_rate": 1.071117742136019e-06,
      "loss": 0.6851,
      "step": 18961
    },
    {
      "epoch": 0.7858593393841435,
      "grad_norm": 0.3701726496219635,
      "learning_rate": 1.0709105226076507e-06,
      "loss": 0.6942,
      "step": 18962
    },
    {
      "epoch": 0.7859007832898173,
      "grad_norm": 0.47792327404022217,
      "learning_rate": 1.0707033030792823e-06,
      "loss": 0.6205,
      "step": 18963
    },
    {
      "epoch": 0.7859422271954909,
      "grad_norm": 0.45355769991874695,
      "learning_rate": 1.0704960835509139e-06,
      "loss": 0.7286,
      "step": 18964
    },
    {
      "epoch": 0.7859836711011646,
      "grad_norm": 0.4044123888015747,
      "learning_rate": 1.0702888640225457e-06,
      "loss": 0.6548,
      "step": 18965
    },
    {
      "epoch": 0.7860251150068382,
      "grad_norm": 0.4119459390640259,
      "learning_rate": 1.0700816444941773e-06,
      "loss": 0.6718,
      "step": 18966
    },
    {
      "epoch": 0.7860665589125119,
      "grad_norm": 0.4102928042411804,
      "learning_rate": 1.0698744249658089e-06,
      "loss": 0.6147,
      "step": 18967
    },
    {
      "epoch": 0.7861080028181856,
      "grad_norm": 0.38672274351119995,
      "learning_rate": 1.0696672054374405e-06,
      "loss": 0.6119,
      "step": 18968
    },
    {
      "epoch": 0.7861494467238592,
      "grad_norm": 0.3983534574508667,
      "learning_rate": 1.0694599859090723e-06,
      "loss": 0.6661,
      "step": 18969
    },
    {
      "epoch": 0.786190890629533,
      "grad_norm": 0.4630031883716583,
      "learning_rate": 1.0692527663807039e-06,
      "loss": 0.6514,
      "step": 18970
    },
    {
      "epoch": 0.7862323345352066,
      "grad_norm": 0.43525680899620056,
      "learning_rate": 1.0690455468523355e-06,
      "loss": 0.6543,
      "step": 18971
    },
    {
      "epoch": 0.7862737784408803,
      "grad_norm": 0.40419742465019226,
      "learning_rate": 1.068838327323967e-06,
      "loss": 0.6412,
      "step": 18972
    },
    {
      "epoch": 0.7863152223465539,
      "grad_norm": 0.43733981251716614,
      "learning_rate": 1.0686311077955989e-06,
      "loss": 0.7062,
      "step": 18973
    },
    {
      "epoch": 0.7863566662522277,
      "grad_norm": 0.4275733530521393,
      "learning_rate": 1.0684238882672305e-06,
      "loss": 0.6934,
      "step": 18974
    },
    {
      "epoch": 0.7863981101579013,
      "grad_norm": 0.42171716690063477,
      "learning_rate": 1.068216668738862e-06,
      "loss": 0.677,
      "step": 18975
    },
    {
      "epoch": 0.7864395540635749,
      "grad_norm": 0.4469958245754242,
      "learning_rate": 1.0680094492104937e-06,
      "loss": 0.681,
      "step": 18976
    },
    {
      "epoch": 0.7864809979692486,
      "grad_norm": 0.40395402908325195,
      "learning_rate": 1.0678022296821253e-06,
      "loss": 0.6603,
      "step": 18977
    },
    {
      "epoch": 0.7865224418749223,
      "grad_norm": 0.4354201853275299,
      "learning_rate": 1.067595010153757e-06,
      "loss": 0.7185,
      "step": 18978
    },
    {
      "epoch": 0.786563885780596,
      "grad_norm": 0.39695966243743896,
      "learning_rate": 1.0673877906253887e-06,
      "loss": 0.6674,
      "step": 18979
    },
    {
      "epoch": 0.7866053296862696,
      "grad_norm": 0.36842912435531616,
      "learning_rate": 1.0671805710970203e-06,
      "loss": 0.6396,
      "step": 18980
    },
    {
      "epoch": 0.7866467735919433,
      "grad_norm": 0.400668203830719,
      "learning_rate": 1.0669733515686519e-06,
      "loss": 0.6376,
      "step": 18981
    },
    {
      "epoch": 0.786688217497617,
      "grad_norm": 0.400261789560318,
      "learning_rate": 1.0667661320402837e-06,
      "loss": 0.6218,
      "step": 18982
    },
    {
      "epoch": 0.7867296614032906,
      "grad_norm": 0.43155601620674133,
      "learning_rate": 1.0665589125119153e-06,
      "loss": 0.7091,
      "step": 18983
    },
    {
      "epoch": 0.7867711053089643,
      "grad_norm": 0.46743783354759216,
      "learning_rate": 1.0663516929835469e-06,
      "loss": 0.6833,
      "step": 18984
    },
    {
      "epoch": 0.786812549214638,
      "grad_norm": 0.40582382678985596,
      "learning_rate": 1.0661444734551785e-06,
      "loss": 0.7268,
      "step": 18985
    },
    {
      "epoch": 0.7868539931203117,
      "grad_norm": 0.37013640999794006,
      "learning_rate": 1.0659372539268103e-06,
      "loss": 0.6492,
      "step": 18986
    },
    {
      "epoch": 0.7868954370259853,
      "grad_norm": 0.44557520747184753,
      "learning_rate": 1.0657300343984419e-06,
      "loss": 0.6757,
      "step": 18987
    },
    {
      "epoch": 0.786936880931659,
      "grad_norm": 0.38597506284713745,
      "learning_rate": 1.0655228148700735e-06,
      "loss": 0.6714,
      "step": 18988
    },
    {
      "epoch": 0.7869783248373327,
      "grad_norm": 0.41287940740585327,
      "learning_rate": 1.065315595341705e-06,
      "loss": 0.6725,
      "step": 18989
    },
    {
      "epoch": 0.7870197687430064,
      "grad_norm": 0.4253733456134796,
      "learning_rate": 1.0651083758133367e-06,
      "loss": 0.656,
      "step": 18990
    },
    {
      "epoch": 0.78706121264868,
      "grad_norm": 0.47208890318870544,
      "learning_rate": 1.0649011562849685e-06,
      "loss": 0.7074,
      "step": 18991
    },
    {
      "epoch": 0.7871026565543536,
      "grad_norm": 0.41482940316200256,
      "learning_rate": 1.0646939367566e-06,
      "loss": 0.6758,
      "step": 18992
    },
    {
      "epoch": 0.7871441004600274,
      "grad_norm": 0.453336238861084,
      "learning_rate": 1.0644867172282317e-06,
      "loss": 0.7168,
      "step": 18993
    },
    {
      "epoch": 0.787185544365701,
      "grad_norm": 0.3993060886859894,
      "learning_rate": 1.0642794976998633e-06,
      "loss": 0.6528,
      "step": 18994
    },
    {
      "epoch": 0.7872269882713747,
      "grad_norm": 0.4056169390678406,
      "learning_rate": 1.064072278171495e-06,
      "loss": 0.6921,
      "step": 18995
    },
    {
      "epoch": 0.7872684321770483,
      "grad_norm": 0.4069829285144806,
      "learning_rate": 1.0638650586431267e-06,
      "loss": 0.6958,
      "step": 18996
    },
    {
      "epoch": 0.7873098760827221,
      "grad_norm": 0.42748552560806274,
      "learning_rate": 1.0636578391147583e-06,
      "loss": 0.6855,
      "step": 18997
    },
    {
      "epoch": 0.7873513199883957,
      "grad_norm": 0.39725416898727417,
      "learning_rate": 1.0634506195863899e-06,
      "loss": 0.7271,
      "step": 18998
    },
    {
      "epoch": 0.7873927638940694,
      "grad_norm": 0.40559178590774536,
      "learning_rate": 1.0632434000580217e-06,
      "loss": 0.7312,
      "step": 18999
    },
    {
      "epoch": 0.787434207799743,
      "grad_norm": 0.38320693373680115,
      "learning_rate": 1.0630361805296533e-06,
      "loss": 0.6447,
      "step": 19000
    },
    {
      "epoch": 0.7874756517054167,
      "grad_norm": 0.4029567837715149,
      "learning_rate": 1.0628289610012849e-06,
      "loss": 0.7061,
      "step": 19001
    },
    {
      "epoch": 0.7875170956110904,
      "grad_norm": 0.4195445775985718,
      "learning_rate": 1.0626217414729165e-06,
      "loss": 0.675,
      "step": 19002
    },
    {
      "epoch": 0.787558539516764,
      "grad_norm": 0.4085678160190582,
      "learning_rate": 1.062414521944548e-06,
      "loss": 0.6804,
      "step": 19003
    },
    {
      "epoch": 0.7875999834224378,
      "grad_norm": 0.4336659908294678,
      "learning_rate": 1.0622073024161799e-06,
      "loss": 0.6992,
      "step": 19004
    },
    {
      "epoch": 0.7876414273281114,
      "grad_norm": 0.39956793189048767,
      "learning_rate": 1.0620000828878115e-06,
      "loss": 0.7,
      "step": 19005
    },
    {
      "epoch": 0.7876828712337851,
      "grad_norm": 0.4146704375743866,
      "learning_rate": 1.061792863359443e-06,
      "loss": 0.6498,
      "step": 19006
    },
    {
      "epoch": 0.7877243151394587,
      "grad_norm": 0.4378402531147003,
      "learning_rate": 1.0615856438310747e-06,
      "loss": 0.7189,
      "step": 19007
    },
    {
      "epoch": 0.7877657590451325,
      "grad_norm": 0.4073712229728699,
      "learning_rate": 1.0613784243027065e-06,
      "loss": 0.6984,
      "step": 19008
    },
    {
      "epoch": 0.7878072029508061,
      "grad_norm": 0.42775028944015503,
      "learning_rate": 1.061171204774338e-06,
      "loss": 0.6864,
      "step": 19009
    },
    {
      "epoch": 0.7878486468564797,
      "grad_norm": 0.4420085847377777,
      "learning_rate": 1.0609639852459697e-06,
      "loss": 0.6365,
      "step": 19010
    },
    {
      "epoch": 0.7878900907621534,
      "grad_norm": 0.41885870695114136,
      "learning_rate": 1.0607567657176013e-06,
      "loss": 0.7188,
      "step": 19011
    },
    {
      "epoch": 0.7879315346678271,
      "grad_norm": 0.46431753039360046,
      "learning_rate": 1.0605495461892329e-06,
      "loss": 0.7429,
      "step": 19012
    },
    {
      "epoch": 0.7879729785735008,
      "grad_norm": 0.4021746516227722,
      "learning_rate": 1.0603423266608647e-06,
      "loss": 0.6177,
      "step": 19013
    },
    {
      "epoch": 0.7880144224791744,
      "grad_norm": 0.3859046995639801,
      "learning_rate": 1.0601351071324963e-06,
      "loss": 0.6497,
      "step": 19014
    },
    {
      "epoch": 0.7880558663848481,
      "grad_norm": 0.41980859637260437,
      "learning_rate": 1.0599278876041279e-06,
      "loss": 0.6519,
      "step": 19015
    },
    {
      "epoch": 0.7880973102905218,
      "grad_norm": 0.40654247999191284,
      "learning_rate": 1.0597206680757595e-06,
      "loss": 0.6792,
      "step": 19016
    },
    {
      "epoch": 0.7881387541961955,
      "grad_norm": 0.4137214124202728,
      "learning_rate": 1.0595134485473913e-06,
      "loss": 0.6475,
      "step": 19017
    },
    {
      "epoch": 0.7881801981018691,
      "grad_norm": 0.42744868993759155,
      "learning_rate": 1.0593062290190229e-06,
      "loss": 0.6864,
      "step": 19018
    },
    {
      "epoch": 0.7882216420075427,
      "grad_norm": 0.4114372432231903,
      "learning_rate": 1.0590990094906545e-06,
      "loss": 0.6589,
      "step": 19019
    },
    {
      "epoch": 0.7882630859132165,
      "grad_norm": 0.41254138946533203,
      "learning_rate": 1.058891789962286e-06,
      "loss": 0.6875,
      "step": 19020
    },
    {
      "epoch": 0.7883045298188901,
      "grad_norm": 0.4225682020187378,
      "learning_rate": 1.0586845704339179e-06,
      "loss": 0.6896,
      "step": 19021
    },
    {
      "epoch": 0.7883459737245638,
      "grad_norm": 0.39918065071105957,
      "learning_rate": 1.0584773509055495e-06,
      "loss": 0.6899,
      "step": 19022
    },
    {
      "epoch": 0.7883874176302375,
      "grad_norm": 0.43880385160446167,
      "learning_rate": 1.058270131377181e-06,
      "loss": 0.6654,
      "step": 19023
    },
    {
      "epoch": 0.7884288615359112,
      "grad_norm": 0.42066556215286255,
      "learning_rate": 1.0580629118488127e-06,
      "loss": 0.7278,
      "step": 19024
    },
    {
      "epoch": 0.7884703054415848,
      "grad_norm": 0.39099857211112976,
      "learning_rate": 1.0578556923204443e-06,
      "loss": 0.6403,
      "step": 19025
    },
    {
      "epoch": 0.7885117493472585,
      "grad_norm": 0.4247477650642395,
      "learning_rate": 1.057648472792076e-06,
      "loss": 0.6399,
      "step": 19026
    },
    {
      "epoch": 0.7885531932529322,
      "grad_norm": 0.38323912024497986,
      "learning_rate": 1.0574412532637077e-06,
      "loss": 0.6619,
      "step": 19027
    },
    {
      "epoch": 0.7885946371586058,
      "grad_norm": 0.4923589527606964,
      "learning_rate": 1.0572340337353393e-06,
      "loss": 0.6241,
      "step": 19028
    },
    {
      "epoch": 0.7886360810642795,
      "grad_norm": 0.43588075041770935,
      "learning_rate": 1.0570268142069709e-06,
      "loss": 0.7095,
      "step": 19029
    },
    {
      "epoch": 0.7886775249699531,
      "grad_norm": 0.4013383388519287,
      "learning_rate": 1.0568195946786027e-06,
      "loss": 0.6127,
      "step": 19030
    },
    {
      "epoch": 0.7887189688756269,
      "grad_norm": 0.3968726396560669,
      "learning_rate": 1.0566123751502343e-06,
      "loss": 0.6345,
      "step": 19031
    },
    {
      "epoch": 0.7887604127813005,
      "grad_norm": 0.42208540439605713,
      "learning_rate": 1.0564051556218659e-06,
      "loss": 0.6735,
      "step": 19032
    },
    {
      "epoch": 0.7888018566869742,
      "grad_norm": 0.4201124608516693,
      "learning_rate": 1.0561979360934975e-06,
      "loss": 0.6483,
      "step": 19033
    },
    {
      "epoch": 0.7888433005926478,
      "grad_norm": 0.4163497984409332,
      "learning_rate": 1.0559907165651293e-06,
      "loss": 0.6702,
      "step": 19034
    },
    {
      "epoch": 0.7888847444983216,
      "grad_norm": 0.43796736001968384,
      "learning_rate": 1.0557834970367609e-06,
      "loss": 0.666,
      "step": 19035
    },
    {
      "epoch": 0.7889261884039952,
      "grad_norm": 0.4109412431716919,
      "learning_rate": 1.0555762775083925e-06,
      "loss": 0.6716,
      "step": 19036
    },
    {
      "epoch": 0.7889676323096688,
      "grad_norm": 0.39690208435058594,
      "learning_rate": 1.055369057980024e-06,
      "loss": 0.6512,
      "step": 19037
    },
    {
      "epoch": 0.7890090762153426,
      "grad_norm": 0.4165399968624115,
      "learning_rate": 1.0551618384516557e-06,
      "loss": 0.7332,
      "step": 19038
    },
    {
      "epoch": 0.7890505201210162,
      "grad_norm": 0.41597679257392883,
      "learning_rate": 1.0549546189232875e-06,
      "loss": 0.677,
      "step": 19039
    },
    {
      "epoch": 0.7890919640266899,
      "grad_norm": 0.406015008687973,
      "learning_rate": 1.054747399394919e-06,
      "loss": 0.6597,
      "step": 19040
    },
    {
      "epoch": 0.7891334079323635,
      "grad_norm": 0.40825799107551575,
      "learning_rate": 1.0545401798665507e-06,
      "loss": 0.611,
      "step": 19041
    },
    {
      "epoch": 0.7891748518380373,
      "grad_norm": 0.4134366810321808,
      "learning_rate": 1.0543329603381823e-06,
      "loss": 0.6941,
      "step": 19042
    },
    {
      "epoch": 0.7892162957437109,
      "grad_norm": 0.41800180077552795,
      "learning_rate": 1.054125740809814e-06,
      "loss": 0.7043,
      "step": 19043
    },
    {
      "epoch": 0.7892577396493845,
      "grad_norm": 0.38747695088386536,
      "learning_rate": 1.0539185212814457e-06,
      "loss": 0.6022,
      "step": 19044
    },
    {
      "epoch": 0.7892991835550582,
      "grad_norm": 0.42518559098243713,
      "learning_rate": 1.0537113017530773e-06,
      "loss": 0.6266,
      "step": 19045
    },
    {
      "epoch": 0.7893406274607319,
      "grad_norm": 0.4062229096889496,
      "learning_rate": 1.0535040822247089e-06,
      "loss": 0.6439,
      "step": 19046
    },
    {
      "epoch": 0.7893820713664056,
      "grad_norm": 0.43059051036834717,
      "learning_rate": 1.0532968626963407e-06,
      "loss": 0.6707,
      "step": 19047
    },
    {
      "epoch": 0.7894235152720792,
      "grad_norm": 0.4235536456108093,
      "learning_rate": 1.0530896431679723e-06,
      "loss": 0.6158,
      "step": 19048
    },
    {
      "epoch": 0.789464959177753,
      "grad_norm": 0.3986480236053467,
      "learning_rate": 1.0528824236396039e-06,
      "loss": 0.658,
      "step": 19049
    },
    {
      "epoch": 0.7895064030834266,
      "grad_norm": 0.4458043873310089,
      "learning_rate": 1.0526752041112355e-06,
      "loss": 0.675,
      "step": 19050
    },
    {
      "epoch": 0.7895478469891003,
      "grad_norm": 0.41404658555984497,
      "learning_rate": 1.052467984582867e-06,
      "loss": 0.7113,
      "step": 19051
    },
    {
      "epoch": 0.7895892908947739,
      "grad_norm": 0.4179061949253082,
      "learning_rate": 1.0522607650544989e-06,
      "loss": 0.6562,
      "step": 19052
    },
    {
      "epoch": 0.7896307348004475,
      "grad_norm": 0.4164784252643585,
      "learning_rate": 1.0520535455261305e-06,
      "loss": 0.6837,
      "step": 19053
    },
    {
      "epoch": 0.7896721787061213,
      "grad_norm": 0.40183213353157043,
      "learning_rate": 1.051846325997762e-06,
      "loss": 0.6362,
      "step": 19054
    },
    {
      "epoch": 0.7897136226117949,
      "grad_norm": 0.44516390562057495,
      "learning_rate": 1.0516391064693937e-06,
      "loss": 0.7229,
      "step": 19055
    },
    {
      "epoch": 0.7897550665174686,
      "grad_norm": 0.42284077405929565,
      "learning_rate": 1.0514318869410255e-06,
      "loss": 0.7131,
      "step": 19056
    },
    {
      "epoch": 0.7897965104231423,
      "grad_norm": 0.41007930040359497,
      "learning_rate": 1.051224667412657e-06,
      "loss": 0.635,
      "step": 19057
    },
    {
      "epoch": 0.789837954328816,
      "grad_norm": 0.45519980788230896,
      "learning_rate": 1.0510174478842887e-06,
      "loss": 0.7332,
      "step": 19058
    },
    {
      "epoch": 0.7898793982344896,
      "grad_norm": 0.4291892945766449,
      "learning_rate": 1.0508102283559203e-06,
      "loss": 0.6285,
      "step": 19059
    },
    {
      "epoch": 0.7899208421401633,
      "grad_norm": 0.3933717906475067,
      "learning_rate": 1.050603008827552e-06,
      "loss": 0.6249,
      "step": 19060
    },
    {
      "epoch": 0.789962286045837,
      "grad_norm": 0.4040065109729767,
      "learning_rate": 1.0503957892991837e-06,
      "loss": 0.6669,
      "step": 19061
    },
    {
      "epoch": 0.7900037299515106,
      "grad_norm": 0.42912599444389343,
      "learning_rate": 1.0501885697708153e-06,
      "loss": 0.6794,
      "step": 19062
    },
    {
      "epoch": 0.7900451738571843,
      "grad_norm": 0.46373358368873596,
      "learning_rate": 1.0499813502424469e-06,
      "loss": 0.7131,
      "step": 19063
    },
    {
      "epoch": 0.7900866177628579,
      "grad_norm": 0.44351908564567566,
      "learning_rate": 1.0497741307140785e-06,
      "loss": 0.6991,
      "step": 19064
    },
    {
      "epoch": 0.7901280616685317,
      "grad_norm": 0.48599907755851746,
      "learning_rate": 1.0495669111857103e-06,
      "loss": 0.7123,
      "step": 19065
    },
    {
      "epoch": 0.7901695055742053,
      "grad_norm": 0.41803210973739624,
      "learning_rate": 1.0493596916573419e-06,
      "loss": 0.6292,
      "step": 19066
    },
    {
      "epoch": 0.790210949479879,
      "grad_norm": 0.3954671621322632,
      "learning_rate": 1.0491524721289735e-06,
      "loss": 0.6187,
      "step": 19067
    },
    {
      "epoch": 0.7902523933855526,
      "grad_norm": 0.4047689139842987,
      "learning_rate": 1.048945252600605e-06,
      "loss": 0.6576,
      "step": 19068
    },
    {
      "epoch": 0.7902938372912264,
      "grad_norm": 0.42807894945144653,
      "learning_rate": 1.0487380330722369e-06,
      "loss": 0.6578,
      "step": 19069
    },
    {
      "epoch": 0.7903352811969,
      "grad_norm": 0.45076891779899597,
      "learning_rate": 1.0485308135438685e-06,
      "loss": 0.662,
      "step": 19070
    },
    {
      "epoch": 0.7903767251025736,
      "grad_norm": 0.407345712184906,
      "learning_rate": 1.0483235940155e-06,
      "loss": 0.6387,
      "step": 19071
    },
    {
      "epoch": 0.7904181690082474,
      "grad_norm": 0.38333529233932495,
      "learning_rate": 1.0481163744871317e-06,
      "loss": 0.6071,
      "step": 19072
    },
    {
      "epoch": 0.790459612913921,
      "grad_norm": 0.4433356523513794,
      "learning_rate": 1.0479091549587633e-06,
      "loss": 0.656,
      "step": 19073
    },
    {
      "epoch": 0.7905010568195947,
      "grad_norm": 0.4648955464363098,
      "learning_rate": 1.047701935430395e-06,
      "loss": 0.6985,
      "step": 19074
    },
    {
      "epoch": 0.7905425007252683,
      "grad_norm": 0.4331129193305969,
      "learning_rate": 1.0474947159020267e-06,
      "loss": 0.7375,
      "step": 19075
    },
    {
      "epoch": 0.7905839446309421,
      "grad_norm": 0.41944730281829834,
      "learning_rate": 1.0472874963736583e-06,
      "loss": 0.6921,
      "step": 19076
    },
    {
      "epoch": 0.7906253885366157,
      "grad_norm": 0.4042914807796478,
      "learning_rate": 1.0470802768452899e-06,
      "loss": 0.7076,
      "step": 19077
    },
    {
      "epoch": 0.7906668324422894,
      "grad_norm": 0.4790422320365906,
      "learning_rate": 1.0468730573169217e-06,
      "loss": 0.6522,
      "step": 19078
    },
    {
      "epoch": 0.790708276347963,
      "grad_norm": 0.400082528591156,
      "learning_rate": 1.0466658377885533e-06,
      "loss": 0.6116,
      "step": 19079
    },
    {
      "epoch": 0.7907497202536367,
      "grad_norm": 0.39321476221084595,
      "learning_rate": 1.0464586182601849e-06,
      "loss": 0.6589,
      "step": 19080
    },
    {
      "epoch": 0.7907911641593104,
      "grad_norm": 0.40992337465286255,
      "learning_rate": 1.0462513987318165e-06,
      "loss": 0.6509,
      "step": 19081
    },
    {
      "epoch": 0.790832608064984,
      "grad_norm": 0.4748751223087311,
      "learning_rate": 1.0460441792034483e-06,
      "loss": 0.7249,
      "step": 19082
    },
    {
      "epoch": 0.7908740519706577,
      "grad_norm": 0.4171788990497589,
      "learning_rate": 1.0458369596750799e-06,
      "loss": 0.6865,
      "step": 19083
    },
    {
      "epoch": 0.7909154958763314,
      "grad_norm": 0.3876623213291168,
      "learning_rate": 1.0456297401467115e-06,
      "loss": 0.6487,
      "step": 19084
    },
    {
      "epoch": 0.7909569397820051,
      "grad_norm": 0.4288181662559509,
      "learning_rate": 1.045422520618343e-06,
      "loss": 0.6594,
      "step": 19085
    },
    {
      "epoch": 0.7909983836876787,
      "grad_norm": 0.4108084440231323,
      "learning_rate": 1.0452153010899747e-06,
      "loss": 0.6753,
      "step": 19086
    },
    {
      "epoch": 0.7910398275933525,
      "grad_norm": 0.43524670600891113,
      "learning_rate": 1.0450080815616065e-06,
      "loss": 0.7097,
      "step": 19087
    },
    {
      "epoch": 0.7910812714990261,
      "grad_norm": 0.409869909286499,
      "learning_rate": 1.044800862033238e-06,
      "loss": 0.7131,
      "step": 19088
    },
    {
      "epoch": 0.7911227154046997,
      "grad_norm": 0.380825012922287,
      "learning_rate": 1.0445936425048697e-06,
      "loss": 0.6284,
      "step": 19089
    },
    {
      "epoch": 0.7911641593103734,
      "grad_norm": 0.4318774342536926,
      "learning_rate": 1.0443864229765013e-06,
      "loss": 0.704,
      "step": 19090
    },
    {
      "epoch": 0.791205603216047,
      "grad_norm": 0.43159759044647217,
      "learning_rate": 1.044179203448133e-06,
      "loss": 0.7534,
      "step": 19091
    },
    {
      "epoch": 0.7912470471217208,
      "grad_norm": 0.3986278474330902,
      "learning_rate": 1.0439719839197647e-06,
      "loss": 0.6329,
      "step": 19092
    },
    {
      "epoch": 0.7912884910273944,
      "grad_norm": 0.3839908838272095,
      "learning_rate": 1.0437647643913963e-06,
      "loss": 0.6395,
      "step": 19093
    },
    {
      "epoch": 0.7913299349330681,
      "grad_norm": 0.4027753472328186,
      "learning_rate": 1.0435575448630279e-06,
      "loss": 0.6786,
      "step": 19094
    },
    {
      "epoch": 0.7913713788387418,
      "grad_norm": 0.40776771306991577,
      "learning_rate": 1.0433503253346597e-06,
      "loss": 0.6345,
      "step": 19095
    },
    {
      "epoch": 0.7914128227444154,
      "grad_norm": 0.39865782856941223,
      "learning_rate": 1.0431431058062913e-06,
      "loss": 0.6886,
      "step": 19096
    },
    {
      "epoch": 0.7914542666500891,
      "grad_norm": 0.4060210585594177,
      "learning_rate": 1.0429358862779229e-06,
      "loss": 0.6309,
      "step": 19097
    },
    {
      "epoch": 0.7914957105557627,
      "grad_norm": 0.4109935462474823,
      "learning_rate": 1.0427286667495545e-06,
      "loss": 0.7019,
      "step": 19098
    },
    {
      "epoch": 0.7915371544614365,
      "grad_norm": 0.4249586760997772,
      "learning_rate": 1.042521447221186e-06,
      "loss": 0.7258,
      "step": 19099
    },
    {
      "epoch": 0.7915785983671101,
      "grad_norm": 0.4111277759075165,
      "learning_rate": 1.042314227692818e-06,
      "loss": 0.708,
      "step": 19100
    },
    {
      "epoch": 0.7916200422727838,
      "grad_norm": 0.4062563180923462,
      "learning_rate": 1.0421070081644495e-06,
      "loss": 0.7083,
      "step": 19101
    },
    {
      "epoch": 0.7916614861784574,
      "grad_norm": 0.4303516447544098,
      "learning_rate": 1.041899788636081e-06,
      "loss": 0.6631,
      "step": 19102
    },
    {
      "epoch": 0.7917029300841312,
      "grad_norm": 0.41718176007270813,
      "learning_rate": 1.0416925691077127e-06,
      "loss": 0.6572,
      "step": 19103
    },
    {
      "epoch": 0.7917443739898048,
      "grad_norm": 0.43008843064308167,
      "learning_rate": 1.0414853495793445e-06,
      "loss": 0.6655,
      "step": 19104
    },
    {
      "epoch": 0.7917858178954784,
      "grad_norm": 0.3673175871372223,
      "learning_rate": 1.041278130050976e-06,
      "loss": 0.679,
      "step": 19105
    },
    {
      "epoch": 0.7918272618011521,
      "grad_norm": 0.40681567788124084,
      "learning_rate": 1.0410709105226077e-06,
      "loss": 0.7119,
      "step": 19106
    },
    {
      "epoch": 0.7918687057068258,
      "grad_norm": 0.4077039062976837,
      "learning_rate": 1.0408636909942393e-06,
      "loss": 0.6633,
      "step": 19107
    },
    {
      "epoch": 0.7919101496124995,
      "grad_norm": 0.4045580327510834,
      "learning_rate": 1.040656471465871e-06,
      "loss": 0.6392,
      "step": 19108
    },
    {
      "epoch": 0.7919515935181731,
      "grad_norm": 0.41078484058380127,
      "learning_rate": 1.0404492519375027e-06,
      "loss": 0.6687,
      "step": 19109
    },
    {
      "epoch": 0.7919930374238469,
      "grad_norm": 0.3805042803287506,
      "learning_rate": 1.0402420324091343e-06,
      "loss": 0.6573,
      "step": 19110
    },
    {
      "epoch": 0.7920344813295205,
      "grad_norm": 0.38384923338890076,
      "learning_rate": 1.0400348128807659e-06,
      "loss": 0.6379,
      "step": 19111
    },
    {
      "epoch": 0.7920759252351942,
      "grad_norm": 0.3904039263725281,
      "learning_rate": 1.0398275933523975e-06,
      "loss": 0.6085,
      "step": 19112
    },
    {
      "epoch": 0.7921173691408678,
      "grad_norm": 0.4050913453102112,
      "learning_rate": 1.0396203738240293e-06,
      "loss": 0.6792,
      "step": 19113
    },
    {
      "epoch": 0.7921588130465415,
      "grad_norm": 0.40852969884872437,
      "learning_rate": 1.039413154295661e-06,
      "loss": 0.6514,
      "step": 19114
    },
    {
      "epoch": 0.7922002569522152,
      "grad_norm": 0.3969424068927765,
      "learning_rate": 1.0392059347672925e-06,
      "loss": 0.7062,
      "step": 19115
    },
    {
      "epoch": 0.7922417008578888,
      "grad_norm": 0.421100378036499,
      "learning_rate": 1.038998715238924e-06,
      "loss": 0.7146,
      "step": 19116
    },
    {
      "epoch": 0.7922831447635625,
      "grad_norm": 0.4273785650730133,
      "learning_rate": 1.038791495710556e-06,
      "loss": 0.6427,
      "step": 19117
    },
    {
      "epoch": 0.7923245886692362,
      "grad_norm": 0.4265558421611786,
      "learning_rate": 1.0385842761821875e-06,
      "loss": 0.6348,
      "step": 19118
    },
    {
      "epoch": 0.7923660325749099,
      "grad_norm": 0.4083535671234131,
      "learning_rate": 1.038377056653819e-06,
      "loss": 0.6544,
      "step": 19119
    },
    {
      "epoch": 0.7924074764805835,
      "grad_norm": 0.4063396453857422,
      "learning_rate": 1.0381698371254507e-06,
      "loss": 0.6272,
      "step": 19120
    },
    {
      "epoch": 0.7924489203862572,
      "grad_norm": 0.4282853901386261,
      "learning_rate": 1.0379626175970825e-06,
      "loss": 0.7186,
      "step": 19121
    },
    {
      "epoch": 0.7924903642919309,
      "grad_norm": 0.4389208257198334,
      "learning_rate": 1.037755398068714e-06,
      "loss": 0.7498,
      "step": 19122
    },
    {
      "epoch": 0.7925318081976045,
      "grad_norm": 0.39804983139038086,
      "learning_rate": 1.0375481785403457e-06,
      "loss": 0.687,
      "step": 19123
    },
    {
      "epoch": 0.7925732521032782,
      "grad_norm": 0.41383928060531616,
      "learning_rate": 1.0373409590119773e-06,
      "loss": 0.6573,
      "step": 19124
    },
    {
      "epoch": 0.7926146960089518,
      "grad_norm": 0.4161819815635681,
      "learning_rate": 1.0371337394836089e-06,
      "loss": 0.6837,
      "step": 19125
    },
    {
      "epoch": 0.7926561399146256,
      "grad_norm": 0.4032747149467468,
      "learning_rate": 1.0369265199552407e-06,
      "loss": 0.6675,
      "step": 19126
    },
    {
      "epoch": 0.7926975838202992,
      "grad_norm": 0.4000454843044281,
      "learning_rate": 1.0367193004268723e-06,
      "loss": 0.6792,
      "step": 19127
    },
    {
      "epoch": 0.7927390277259729,
      "grad_norm": 0.47749966382980347,
      "learning_rate": 1.0365120808985039e-06,
      "loss": 0.7383,
      "step": 19128
    },
    {
      "epoch": 0.7927804716316466,
      "grad_norm": 0.37558141350746155,
      "learning_rate": 1.0363048613701355e-06,
      "loss": 0.6378,
      "step": 19129
    },
    {
      "epoch": 0.7928219155373203,
      "grad_norm": 0.39896973967552185,
      "learning_rate": 1.0360976418417673e-06,
      "loss": 0.6128,
      "step": 19130
    },
    {
      "epoch": 0.7928633594429939,
      "grad_norm": 0.46355870366096497,
      "learning_rate": 1.035890422313399e-06,
      "loss": 0.6432,
      "step": 19131
    },
    {
      "epoch": 0.7929048033486675,
      "grad_norm": 0.3922251760959625,
      "learning_rate": 1.0356832027850305e-06,
      "loss": 0.6328,
      "step": 19132
    },
    {
      "epoch": 0.7929462472543413,
      "grad_norm": 0.4367353916168213,
      "learning_rate": 1.035475983256662e-06,
      "loss": 0.6707,
      "step": 19133
    },
    {
      "epoch": 0.7929876911600149,
      "grad_norm": 0.42394259572029114,
      "learning_rate": 1.035268763728294e-06,
      "loss": 0.6489,
      "step": 19134
    },
    {
      "epoch": 0.7930291350656886,
      "grad_norm": 0.4028220474720001,
      "learning_rate": 1.0350615441999255e-06,
      "loss": 0.6215,
      "step": 19135
    },
    {
      "epoch": 0.7930705789713622,
      "grad_norm": 0.4532501995563507,
      "learning_rate": 1.034854324671557e-06,
      "loss": 0.6732,
      "step": 19136
    },
    {
      "epoch": 0.793112022877036,
      "grad_norm": 0.40724194049835205,
      "learning_rate": 1.0346471051431887e-06,
      "loss": 0.6362,
      "step": 19137
    },
    {
      "epoch": 0.7931534667827096,
      "grad_norm": 0.43106380105018616,
      "learning_rate": 1.0344398856148203e-06,
      "loss": 0.6265,
      "step": 19138
    },
    {
      "epoch": 0.7931949106883833,
      "grad_norm": 0.4336501657962799,
      "learning_rate": 1.034232666086452e-06,
      "loss": 0.7258,
      "step": 19139
    },
    {
      "epoch": 0.793236354594057,
      "grad_norm": 0.4108738303184509,
      "learning_rate": 1.0340254465580837e-06,
      "loss": 0.6349,
      "step": 19140
    },
    {
      "epoch": 0.7932777984997306,
      "grad_norm": 0.3940304219722748,
      "learning_rate": 1.0338182270297153e-06,
      "loss": 0.6277,
      "step": 19141
    },
    {
      "epoch": 0.7933192424054043,
      "grad_norm": 0.45226559042930603,
      "learning_rate": 1.0336110075013469e-06,
      "loss": 0.6729,
      "step": 19142
    },
    {
      "epoch": 0.7933606863110779,
      "grad_norm": 0.42679503560066223,
      "learning_rate": 1.0334037879729787e-06,
      "loss": 0.6646,
      "step": 19143
    },
    {
      "epoch": 0.7934021302167517,
      "grad_norm": 0.41281846165657043,
      "learning_rate": 1.0331965684446103e-06,
      "loss": 0.6829,
      "step": 19144
    },
    {
      "epoch": 0.7934435741224253,
      "grad_norm": 0.4624081552028656,
      "learning_rate": 1.032989348916242e-06,
      "loss": 0.7023,
      "step": 19145
    },
    {
      "epoch": 0.793485018028099,
      "grad_norm": 0.42832711338996887,
      "learning_rate": 1.0327821293878735e-06,
      "loss": 0.6868,
      "step": 19146
    },
    {
      "epoch": 0.7935264619337726,
      "grad_norm": 0.3996882140636444,
      "learning_rate": 1.032574909859505e-06,
      "loss": 0.5905,
      "step": 19147
    },
    {
      "epoch": 0.7935679058394464,
      "grad_norm": 0.392010360956192,
      "learning_rate": 1.032367690331137e-06,
      "loss": 0.6793,
      "step": 19148
    },
    {
      "epoch": 0.79360934974512,
      "grad_norm": 0.4246884882450104,
      "learning_rate": 1.0321604708027685e-06,
      "loss": 0.6733,
      "step": 19149
    },
    {
      "epoch": 0.7936507936507936,
      "grad_norm": 0.4101029932498932,
      "learning_rate": 1.0319532512744e-06,
      "loss": 0.6541,
      "step": 19150
    },
    {
      "epoch": 0.7936922375564673,
      "grad_norm": 0.40647193789482117,
      "learning_rate": 1.0317460317460317e-06,
      "loss": 0.6979,
      "step": 19151
    },
    {
      "epoch": 0.793733681462141,
      "grad_norm": 0.3890632688999176,
      "learning_rate": 1.0315388122176635e-06,
      "loss": 0.6193,
      "step": 19152
    },
    {
      "epoch": 0.7937751253678147,
      "grad_norm": 0.4335087239742279,
      "learning_rate": 1.031331592689295e-06,
      "loss": 0.6156,
      "step": 19153
    },
    {
      "epoch": 0.7938165692734883,
      "grad_norm": 0.4121062755584717,
      "learning_rate": 1.0311243731609267e-06,
      "loss": 0.6553,
      "step": 19154
    },
    {
      "epoch": 0.793858013179162,
      "grad_norm": 0.4343561828136444,
      "learning_rate": 1.0309171536325583e-06,
      "loss": 0.6888,
      "step": 19155
    },
    {
      "epoch": 0.7938994570848357,
      "grad_norm": 0.394872784614563,
      "learning_rate": 1.03070993410419e-06,
      "loss": 0.6339,
      "step": 19156
    },
    {
      "epoch": 0.7939409009905093,
      "grad_norm": 0.418499618768692,
      "learning_rate": 1.0305027145758217e-06,
      "loss": 0.6666,
      "step": 19157
    },
    {
      "epoch": 0.793982344896183,
      "grad_norm": 0.39036089181900024,
      "learning_rate": 1.0302954950474533e-06,
      "loss": 0.6614,
      "step": 19158
    },
    {
      "epoch": 0.7940237888018566,
      "grad_norm": 0.4605160653591156,
      "learning_rate": 1.030088275519085e-06,
      "loss": 0.6687,
      "step": 19159
    },
    {
      "epoch": 0.7940652327075304,
      "grad_norm": 0.4489428699016571,
      "learning_rate": 1.0298810559907165e-06,
      "loss": 0.7183,
      "step": 19160
    },
    {
      "epoch": 0.794106676613204,
      "grad_norm": 0.4360344111919403,
      "learning_rate": 1.0296738364623483e-06,
      "loss": 0.6559,
      "step": 19161
    },
    {
      "epoch": 0.7941481205188777,
      "grad_norm": 0.3946727514266968,
      "learning_rate": 1.02946661693398e-06,
      "loss": 0.6885,
      "step": 19162
    },
    {
      "epoch": 0.7941895644245514,
      "grad_norm": 0.39983662962913513,
      "learning_rate": 1.0292593974056115e-06,
      "loss": 0.6736,
      "step": 19163
    },
    {
      "epoch": 0.7942310083302251,
      "grad_norm": 0.4200667142868042,
      "learning_rate": 1.029052177877243e-06,
      "loss": 0.6833,
      "step": 19164
    },
    {
      "epoch": 0.7942724522358987,
      "grad_norm": 0.4198237359523773,
      "learning_rate": 1.028844958348875e-06,
      "loss": 0.6477,
      "step": 19165
    },
    {
      "epoch": 0.7943138961415723,
      "grad_norm": 0.43376612663269043,
      "learning_rate": 1.0286377388205065e-06,
      "loss": 0.7078,
      "step": 19166
    },
    {
      "epoch": 0.7943553400472461,
      "grad_norm": 0.41372203826904297,
      "learning_rate": 1.028430519292138e-06,
      "loss": 0.6509,
      "step": 19167
    },
    {
      "epoch": 0.7943967839529197,
      "grad_norm": 0.4244420826435089,
      "learning_rate": 1.0282232997637697e-06,
      "loss": 0.6527,
      "step": 19168
    },
    {
      "epoch": 0.7944382278585934,
      "grad_norm": 0.40649351477622986,
      "learning_rate": 1.0280160802354015e-06,
      "loss": 0.7073,
      "step": 19169
    },
    {
      "epoch": 0.794479671764267,
      "grad_norm": 0.4477437436580658,
      "learning_rate": 1.027808860707033e-06,
      "loss": 0.6882,
      "step": 19170
    },
    {
      "epoch": 0.7945211156699408,
      "grad_norm": 0.39487090706825256,
      "learning_rate": 1.0276016411786647e-06,
      "loss": 0.6521,
      "step": 19171
    },
    {
      "epoch": 0.7945625595756144,
      "grad_norm": 0.4563996493816376,
      "learning_rate": 1.0273944216502963e-06,
      "loss": 0.6487,
      "step": 19172
    },
    {
      "epoch": 0.7946040034812881,
      "grad_norm": 0.42581963539123535,
      "learning_rate": 1.027187202121928e-06,
      "loss": 0.6924,
      "step": 19173
    },
    {
      "epoch": 0.7946454473869617,
      "grad_norm": 0.4447450041770935,
      "learning_rate": 1.0269799825935597e-06,
      "loss": 0.6821,
      "step": 19174
    },
    {
      "epoch": 0.7946868912926354,
      "grad_norm": 0.41711607575416565,
      "learning_rate": 1.0267727630651913e-06,
      "loss": 0.6477,
      "step": 19175
    },
    {
      "epoch": 0.7947283351983091,
      "grad_norm": 0.4189877510070801,
      "learning_rate": 1.026565543536823e-06,
      "loss": 0.7258,
      "step": 19176
    },
    {
      "epoch": 0.7947697791039827,
      "grad_norm": 0.3847160041332245,
      "learning_rate": 1.0263583240084545e-06,
      "loss": 0.6658,
      "step": 19177
    },
    {
      "epoch": 0.7948112230096565,
      "grad_norm": 0.45315617322921753,
      "learning_rate": 1.0261511044800863e-06,
      "loss": 0.7803,
      "step": 19178
    },
    {
      "epoch": 0.7948526669153301,
      "grad_norm": 0.4791221022605896,
      "learning_rate": 1.025943884951718e-06,
      "loss": 0.7,
      "step": 19179
    },
    {
      "epoch": 0.7948941108210038,
      "grad_norm": 0.45094266533851624,
      "learning_rate": 1.0257366654233495e-06,
      "loss": 0.7161,
      "step": 19180
    },
    {
      "epoch": 0.7949355547266774,
      "grad_norm": 0.41903766989707947,
      "learning_rate": 1.025529445894981e-06,
      "loss": 0.6396,
      "step": 19181
    },
    {
      "epoch": 0.7949769986323512,
      "grad_norm": 0.42651352286338806,
      "learning_rate": 1.025322226366613e-06,
      "loss": 0.6431,
      "step": 19182
    },
    {
      "epoch": 0.7950184425380248,
      "grad_norm": 0.41960132122039795,
      "learning_rate": 1.0251150068382445e-06,
      "loss": 0.6855,
      "step": 19183
    },
    {
      "epoch": 0.7950598864436984,
      "grad_norm": 0.43986159563064575,
      "learning_rate": 1.024907787309876e-06,
      "loss": 0.7014,
      "step": 19184
    },
    {
      "epoch": 0.7951013303493721,
      "grad_norm": 0.39675453305244446,
      "learning_rate": 1.0247005677815077e-06,
      "loss": 0.618,
      "step": 19185
    },
    {
      "epoch": 0.7951427742550458,
      "grad_norm": 0.4377390444278717,
      "learning_rate": 1.0244933482531393e-06,
      "loss": 0.723,
      "step": 19186
    },
    {
      "epoch": 0.7951842181607195,
      "grad_norm": 0.4442036747932434,
      "learning_rate": 1.024286128724771e-06,
      "loss": 0.6952,
      "step": 19187
    },
    {
      "epoch": 0.7952256620663931,
      "grad_norm": 0.41307970881462097,
      "learning_rate": 1.0240789091964027e-06,
      "loss": 0.6589,
      "step": 19188
    },
    {
      "epoch": 0.7952671059720668,
      "grad_norm": 0.42951148748397827,
      "learning_rate": 1.0238716896680343e-06,
      "loss": 0.6949,
      "step": 19189
    },
    {
      "epoch": 0.7953085498777405,
      "grad_norm": 0.3905729055404663,
      "learning_rate": 1.023664470139666e-06,
      "loss": 0.6451,
      "step": 19190
    },
    {
      "epoch": 0.7953499937834142,
      "grad_norm": 0.4037232995033264,
      "learning_rate": 1.0234572506112977e-06,
      "loss": 0.7296,
      "step": 19191
    },
    {
      "epoch": 0.7953914376890878,
      "grad_norm": 0.43444111943244934,
      "learning_rate": 1.0232500310829293e-06,
      "loss": 0.6639,
      "step": 19192
    },
    {
      "epoch": 0.7954328815947614,
      "grad_norm": 0.3985172212123871,
      "learning_rate": 1.023042811554561e-06,
      "loss": 0.6648,
      "step": 19193
    },
    {
      "epoch": 0.7954743255004352,
      "grad_norm": 0.46277618408203125,
      "learning_rate": 1.0228355920261925e-06,
      "loss": 0.7373,
      "step": 19194
    },
    {
      "epoch": 0.7955157694061088,
      "grad_norm": 0.3824600577354431,
      "learning_rate": 1.0226283724978243e-06,
      "loss": 0.624,
      "step": 19195
    },
    {
      "epoch": 0.7955572133117825,
      "grad_norm": 0.42047038674354553,
      "learning_rate": 1.022421152969456e-06,
      "loss": 0.6831,
      "step": 19196
    },
    {
      "epoch": 0.7955986572174562,
      "grad_norm": 0.4389949440956116,
      "learning_rate": 1.0222139334410875e-06,
      "loss": 0.6552,
      "step": 19197
    },
    {
      "epoch": 0.7956401011231299,
      "grad_norm": 0.41491270065307617,
      "learning_rate": 1.022006713912719e-06,
      "loss": 0.691,
      "step": 19198
    },
    {
      "epoch": 0.7956815450288035,
      "grad_norm": 0.42028236389160156,
      "learning_rate": 1.021799494384351e-06,
      "loss": 0.6403,
      "step": 19199
    },
    {
      "epoch": 0.7957229889344772,
      "grad_norm": 0.4287458062171936,
      "learning_rate": 1.0215922748559825e-06,
      "loss": 0.6713,
      "step": 19200
    },
    {
      "epoch": 0.7957644328401509,
      "grad_norm": 0.5021578073501587,
      "learning_rate": 1.021385055327614e-06,
      "loss": 0.7108,
      "step": 19201
    },
    {
      "epoch": 0.7958058767458245,
      "grad_norm": 0.36508241295814514,
      "learning_rate": 1.0211778357992457e-06,
      "loss": 0.6388,
      "step": 19202
    },
    {
      "epoch": 0.7958473206514982,
      "grad_norm": 0.4200110137462616,
      "learning_rate": 1.0209706162708773e-06,
      "loss": 0.7119,
      "step": 19203
    },
    {
      "epoch": 0.7958887645571718,
      "grad_norm": 0.3856765031814575,
      "learning_rate": 1.0207633967425091e-06,
      "loss": 0.6418,
      "step": 19204
    },
    {
      "epoch": 0.7959302084628456,
      "grad_norm": 0.39627358317375183,
      "learning_rate": 1.0205561772141407e-06,
      "loss": 0.7054,
      "step": 19205
    },
    {
      "epoch": 0.7959716523685192,
      "grad_norm": 0.40329477190971375,
      "learning_rate": 1.0203489576857723e-06,
      "loss": 0.6938,
      "step": 19206
    },
    {
      "epoch": 0.7960130962741929,
      "grad_norm": 0.4326789975166321,
      "learning_rate": 1.020141738157404e-06,
      "loss": 0.6495,
      "step": 19207
    },
    {
      "epoch": 0.7960545401798665,
      "grad_norm": 0.4115239977836609,
      "learning_rate": 1.0199345186290357e-06,
      "loss": 0.6654,
      "step": 19208
    },
    {
      "epoch": 0.7960959840855403,
      "grad_norm": 0.4198668599128723,
      "learning_rate": 1.0197272991006673e-06,
      "loss": 0.6797,
      "step": 19209
    },
    {
      "epoch": 0.7961374279912139,
      "grad_norm": 0.4563296437263489,
      "learning_rate": 1.019520079572299e-06,
      "loss": 0.6836,
      "step": 19210
    },
    {
      "epoch": 0.7961788718968875,
      "grad_norm": 0.3868393898010254,
      "learning_rate": 1.0193128600439305e-06,
      "loss": 0.7185,
      "step": 19211
    },
    {
      "epoch": 0.7962203158025613,
      "grad_norm": 0.3994968831539154,
      "learning_rate": 1.0191056405155623e-06,
      "loss": 0.6552,
      "step": 19212
    },
    {
      "epoch": 0.7962617597082349,
      "grad_norm": 0.41161084175109863,
      "learning_rate": 1.018898420987194e-06,
      "loss": 0.674,
      "step": 19213
    },
    {
      "epoch": 0.7963032036139086,
      "grad_norm": 0.38676247000694275,
      "learning_rate": 1.0186912014588255e-06,
      "loss": 0.6487,
      "step": 19214
    },
    {
      "epoch": 0.7963446475195822,
      "grad_norm": 0.38010066747665405,
      "learning_rate": 1.018483981930457e-06,
      "loss": 0.6394,
      "step": 19215
    },
    {
      "epoch": 0.796386091425256,
      "grad_norm": 0.4166615307331085,
      "learning_rate": 1.018276762402089e-06,
      "loss": 0.6436,
      "step": 19216
    },
    {
      "epoch": 0.7964275353309296,
      "grad_norm": 0.4359854459762573,
      "learning_rate": 1.0180695428737205e-06,
      "loss": 0.6665,
      "step": 19217
    },
    {
      "epoch": 0.7964689792366032,
      "grad_norm": 0.40236780047416687,
      "learning_rate": 1.0178623233453521e-06,
      "loss": 0.6796,
      "step": 19218
    },
    {
      "epoch": 0.7965104231422769,
      "grad_norm": 0.4294043779373169,
      "learning_rate": 1.0176551038169837e-06,
      "loss": 0.689,
      "step": 19219
    },
    {
      "epoch": 0.7965518670479506,
      "grad_norm": 0.4124505817890167,
      "learning_rate": 1.0174478842886153e-06,
      "loss": 0.7263,
      "step": 19220
    },
    {
      "epoch": 0.7965933109536243,
      "grad_norm": 0.4284639060497284,
      "learning_rate": 1.0172406647602471e-06,
      "loss": 0.6317,
      "step": 19221
    },
    {
      "epoch": 0.7966347548592979,
      "grad_norm": 0.38135308027267456,
      "learning_rate": 1.0170334452318787e-06,
      "loss": 0.5961,
      "step": 19222
    },
    {
      "epoch": 0.7966761987649716,
      "grad_norm": 0.40183383226394653,
      "learning_rate": 1.0168262257035103e-06,
      "loss": 0.6345,
      "step": 19223
    },
    {
      "epoch": 0.7967176426706453,
      "grad_norm": 0.3929379880428314,
      "learning_rate": 1.016619006175142e-06,
      "loss": 0.6358,
      "step": 19224
    },
    {
      "epoch": 0.796759086576319,
      "grad_norm": 0.4066028594970703,
      "learning_rate": 1.0164117866467737e-06,
      "loss": 0.6777,
      "step": 19225
    },
    {
      "epoch": 0.7968005304819926,
      "grad_norm": 0.4000471830368042,
      "learning_rate": 1.0162045671184053e-06,
      "loss": 0.656,
      "step": 19226
    },
    {
      "epoch": 0.7968419743876662,
      "grad_norm": 0.3860797882080078,
      "learning_rate": 1.015997347590037e-06,
      "loss": 0.6724,
      "step": 19227
    },
    {
      "epoch": 0.79688341829334,
      "grad_norm": 0.44066840410232544,
      "learning_rate": 1.0157901280616685e-06,
      "loss": 0.6477,
      "step": 19228
    },
    {
      "epoch": 0.7969248621990136,
      "grad_norm": 0.42083582282066345,
      "learning_rate": 1.0155829085333003e-06,
      "loss": 0.6974,
      "step": 19229
    },
    {
      "epoch": 0.7969663061046873,
      "grad_norm": 0.4336947798728943,
      "learning_rate": 1.015375689004932e-06,
      "loss": 0.6466,
      "step": 19230
    },
    {
      "epoch": 0.797007750010361,
      "grad_norm": 0.3749508857727051,
      "learning_rate": 1.0151684694765635e-06,
      "loss": 0.6532,
      "step": 19231
    },
    {
      "epoch": 0.7970491939160347,
      "grad_norm": 0.42087122797966003,
      "learning_rate": 1.0149612499481951e-06,
      "loss": 0.642,
      "step": 19232
    },
    {
      "epoch": 0.7970906378217083,
      "grad_norm": 0.41845813393592834,
      "learning_rate": 1.014754030419827e-06,
      "loss": 0.6338,
      "step": 19233
    },
    {
      "epoch": 0.797132081727382,
      "grad_norm": 0.43553638458251953,
      "learning_rate": 1.0145468108914585e-06,
      "loss": 0.7169,
      "step": 19234
    },
    {
      "epoch": 0.7971735256330557,
      "grad_norm": 0.5038114786148071,
      "learning_rate": 1.0143395913630901e-06,
      "loss": 0.6865,
      "step": 19235
    },
    {
      "epoch": 0.7972149695387293,
      "grad_norm": 0.40449094772338867,
      "learning_rate": 1.0141323718347217e-06,
      "loss": 0.6499,
      "step": 19236
    },
    {
      "epoch": 0.797256413444403,
      "grad_norm": 0.43818750977516174,
      "learning_rate": 1.0139251523063533e-06,
      "loss": 0.6982,
      "step": 19237
    },
    {
      "epoch": 0.7972978573500766,
      "grad_norm": 0.4346558153629303,
      "learning_rate": 1.0137179327779851e-06,
      "loss": 0.6974,
      "step": 19238
    },
    {
      "epoch": 0.7973393012557504,
      "grad_norm": 0.48006772994995117,
      "learning_rate": 1.0135107132496167e-06,
      "loss": 0.7063,
      "step": 19239
    },
    {
      "epoch": 0.797380745161424,
      "grad_norm": 0.4236416518688202,
      "learning_rate": 1.0133034937212483e-06,
      "loss": 0.6307,
      "step": 19240
    },
    {
      "epoch": 0.7974221890670977,
      "grad_norm": 0.39990922808647156,
      "learning_rate": 1.01309627419288e-06,
      "loss": 0.6396,
      "step": 19241
    },
    {
      "epoch": 0.7974636329727713,
      "grad_norm": 0.4082353711128235,
      "learning_rate": 1.0128890546645117e-06,
      "loss": 0.7078,
      "step": 19242
    },
    {
      "epoch": 0.7975050768784451,
      "grad_norm": 0.4209856688976288,
      "learning_rate": 1.0126818351361433e-06,
      "loss": 0.646,
      "step": 19243
    },
    {
      "epoch": 0.7975465207841187,
      "grad_norm": 0.4066612422466278,
      "learning_rate": 1.012474615607775e-06,
      "loss": 0.6575,
      "step": 19244
    },
    {
      "epoch": 0.7975879646897923,
      "grad_norm": 0.43028101325035095,
      "learning_rate": 1.0122673960794065e-06,
      "loss": 0.656,
      "step": 19245
    },
    {
      "epoch": 0.797629408595466,
      "grad_norm": 0.4397296905517578,
      "learning_rate": 1.0120601765510383e-06,
      "loss": 0.6755,
      "step": 19246
    },
    {
      "epoch": 0.7976708525011397,
      "grad_norm": 0.4157484173774719,
      "learning_rate": 1.01185295702267e-06,
      "loss": 0.6113,
      "step": 19247
    },
    {
      "epoch": 0.7977122964068134,
      "grad_norm": 0.4109066426753998,
      "learning_rate": 1.0116457374943015e-06,
      "loss": 0.6416,
      "step": 19248
    },
    {
      "epoch": 0.797753740312487,
      "grad_norm": 0.42129120230674744,
      "learning_rate": 1.0114385179659331e-06,
      "loss": 0.627,
      "step": 19249
    },
    {
      "epoch": 0.7977951842181608,
      "grad_norm": 0.4447652995586395,
      "learning_rate": 1.011231298437565e-06,
      "loss": 0.689,
      "step": 19250
    },
    {
      "epoch": 0.7978366281238344,
      "grad_norm": 0.4349232614040375,
      "learning_rate": 1.0110240789091965e-06,
      "loss": 0.6863,
      "step": 19251
    },
    {
      "epoch": 0.7978780720295081,
      "grad_norm": 0.42941880226135254,
      "learning_rate": 1.0108168593808281e-06,
      "loss": 0.6669,
      "step": 19252
    },
    {
      "epoch": 0.7979195159351817,
      "grad_norm": 0.3948589563369751,
      "learning_rate": 1.0106096398524597e-06,
      "loss": 0.6521,
      "step": 19253
    },
    {
      "epoch": 0.7979609598408554,
      "grad_norm": 0.42189639806747437,
      "learning_rate": 1.0104024203240915e-06,
      "loss": 0.6588,
      "step": 19254
    },
    {
      "epoch": 0.7980024037465291,
      "grad_norm": 0.4036520719528198,
      "learning_rate": 1.0101952007957231e-06,
      "loss": 0.6676,
      "step": 19255
    },
    {
      "epoch": 0.7980438476522027,
      "grad_norm": 0.40319085121154785,
      "learning_rate": 1.0099879812673547e-06,
      "loss": 0.6213,
      "step": 19256
    },
    {
      "epoch": 0.7980852915578764,
      "grad_norm": 0.4165392518043518,
      "learning_rate": 1.0097807617389863e-06,
      "loss": 0.6755,
      "step": 19257
    },
    {
      "epoch": 0.7981267354635501,
      "grad_norm": 0.40580490231513977,
      "learning_rate": 1.009573542210618e-06,
      "loss": 0.6582,
      "step": 19258
    },
    {
      "epoch": 0.7981681793692238,
      "grad_norm": 0.42870891094207764,
      "learning_rate": 1.0093663226822497e-06,
      "loss": 0.7094,
      "step": 19259
    },
    {
      "epoch": 0.7982096232748974,
      "grad_norm": 0.4271981120109558,
      "learning_rate": 1.0091591031538813e-06,
      "loss": 0.6838,
      "step": 19260
    },
    {
      "epoch": 0.7982510671805711,
      "grad_norm": 0.4052793085575104,
      "learning_rate": 1.008951883625513e-06,
      "loss": 0.6425,
      "step": 19261
    },
    {
      "epoch": 0.7982925110862448,
      "grad_norm": 0.4063340723514557,
      "learning_rate": 1.0087446640971445e-06,
      "loss": 0.6262,
      "step": 19262
    },
    {
      "epoch": 0.7983339549919184,
      "grad_norm": 0.4111064672470093,
      "learning_rate": 1.0085374445687763e-06,
      "loss": 0.7012,
      "step": 19263
    },
    {
      "epoch": 0.7983753988975921,
      "grad_norm": 0.4201134145259857,
      "learning_rate": 1.008330225040408e-06,
      "loss": 0.6301,
      "step": 19264
    },
    {
      "epoch": 0.7984168428032657,
      "grad_norm": 0.4240688979625702,
      "learning_rate": 1.0081230055120395e-06,
      "loss": 0.6682,
      "step": 19265
    },
    {
      "epoch": 0.7984582867089395,
      "grad_norm": 0.4157269597053528,
      "learning_rate": 1.0079157859836711e-06,
      "loss": 0.6768,
      "step": 19266
    },
    {
      "epoch": 0.7984997306146131,
      "grad_norm": 0.42575252056121826,
      "learning_rate": 1.007708566455303e-06,
      "loss": 0.7024,
      "step": 19267
    },
    {
      "epoch": 0.7985411745202868,
      "grad_norm": 0.4296768009662628,
      "learning_rate": 1.0075013469269345e-06,
      "loss": 0.7058,
      "step": 19268
    },
    {
      "epoch": 0.7985826184259605,
      "grad_norm": 0.3999042212963104,
      "learning_rate": 1.0072941273985661e-06,
      "loss": 0.6736,
      "step": 19269
    },
    {
      "epoch": 0.7986240623316342,
      "grad_norm": 0.465154767036438,
      "learning_rate": 1.0070869078701977e-06,
      "loss": 0.7075,
      "step": 19270
    },
    {
      "epoch": 0.7986655062373078,
      "grad_norm": 0.41514769196510315,
      "learning_rate": 1.0068796883418295e-06,
      "loss": 0.6445,
      "step": 19271
    },
    {
      "epoch": 0.7987069501429814,
      "grad_norm": 0.4592706859111786,
      "learning_rate": 1.0066724688134611e-06,
      "loss": 0.7712,
      "step": 19272
    },
    {
      "epoch": 0.7987483940486552,
      "grad_norm": 0.43338578939437866,
      "learning_rate": 1.0064652492850927e-06,
      "loss": 0.6334,
      "step": 19273
    },
    {
      "epoch": 0.7987898379543288,
      "grad_norm": 0.43260589241981506,
      "learning_rate": 1.0062580297567243e-06,
      "loss": 0.6985,
      "step": 19274
    },
    {
      "epoch": 0.7988312818600025,
      "grad_norm": 0.40319791436195374,
      "learning_rate": 1.006050810228356e-06,
      "loss": 0.6418,
      "step": 19275
    },
    {
      "epoch": 0.7988727257656761,
      "grad_norm": 0.417646586894989,
      "learning_rate": 1.0058435906999877e-06,
      "loss": 0.7026,
      "step": 19276
    },
    {
      "epoch": 0.7989141696713499,
      "grad_norm": 0.42771264910697937,
      "learning_rate": 1.0056363711716193e-06,
      "loss": 0.652,
      "step": 19277
    },
    {
      "epoch": 0.7989556135770235,
      "grad_norm": 0.4110005795955658,
      "learning_rate": 1.005429151643251e-06,
      "loss": 0.6628,
      "step": 19278
    },
    {
      "epoch": 0.7989970574826971,
      "grad_norm": 0.4150239825248718,
      "learning_rate": 1.0052219321148825e-06,
      "loss": 0.658,
      "step": 19279
    },
    {
      "epoch": 0.7990385013883708,
      "grad_norm": 0.3892100751399994,
      "learning_rate": 1.0050147125865143e-06,
      "loss": 0.6283,
      "step": 19280
    },
    {
      "epoch": 0.7990799452940445,
      "grad_norm": 0.46774381399154663,
      "learning_rate": 1.004807493058146e-06,
      "loss": 0.6985,
      "step": 19281
    },
    {
      "epoch": 0.7991213891997182,
      "grad_norm": 0.4484656751155853,
      "learning_rate": 1.0046002735297775e-06,
      "loss": 0.6995,
      "step": 19282
    },
    {
      "epoch": 0.7991628331053918,
      "grad_norm": 0.3673012852668762,
      "learning_rate": 1.0043930540014091e-06,
      "loss": 0.5863,
      "step": 19283
    },
    {
      "epoch": 0.7992042770110656,
      "grad_norm": 0.42089372873306274,
      "learning_rate": 1.004185834473041e-06,
      "loss": 0.7371,
      "step": 19284
    },
    {
      "epoch": 0.7992457209167392,
      "grad_norm": 0.5051882863044739,
      "learning_rate": 1.0039786149446725e-06,
      "loss": 0.771,
      "step": 19285
    },
    {
      "epoch": 0.7992871648224129,
      "grad_norm": 0.477998822927475,
      "learning_rate": 1.0037713954163041e-06,
      "loss": 0.705,
      "step": 19286
    },
    {
      "epoch": 0.7993286087280865,
      "grad_norm": 0.6044408679008484,
      "learning_rate": 1.0035641758879357e-06,
      "loss": 0.7295,
      "step": 19287
    },
    {
      "epoch": 0.7993700526337602,
      "grad_norm": 0.4190250635147095,
      "learning_rate": 1.0033569563595675e-06,
      "loss": 0.6799,
      "step": 19288
    },
    {
      "epoch": 0.7994114965394339,
      "grad_norm": 0.46985289454460144,
      "learning_rate": 1.0031497368311991e-06,
      "loss": 0.7078,
      "step": 19289
    },
    {
      "epoch": 0.7994529404451075,
      "grad_norm": 0.3941118121147156,
      "learning_rate": 1.0029425173028307e-06,
      "loss": 0.6658,
      "step": 19290
    },
    {
      "epoch": 0.7994943843507812,
      "grad_norm": 0.3669600188732147,
      "learning_rate": 1.0027352977744623e-06,
      "loss": 0.6312,
      "step": 19291
    },
    {
      "epoch": 0.7995358282564549,
      "grad_norm": 0.39518335461616516,
      "learning_rate": 1.0025280782460941e-06,
      "loss": 0.6454,
      "step": 19292
    },
    {
      "epoch": 0.7995772721621286,
      "grad_norm": 0.41699251532554626,
      "learning_rate": 1.0023208587177257e-06,
      "loss": 0.6641,
      "step": 19293
    },
    {
      "epoch": 0.7996187160678022,
      "grad_norm": 0.4234275817871094,
      "learning_rate": 1.0021136391893573e-06,
      "loss": 0.651,
      "step": 19294
    },
    {
      "epoch": 0.799660159973476,
      "grad_norm": 0.4194095730781555,
      "learning_rate": 1.001906419660989e-06,
      "loss": 0.6624,
      "step": 19295
    },
    {
      "epoch": 0.7997016038791496,
      "grad_norm": 0.3878420889377594,
      "learning_rate": 1.0016992001326205e-06,
      "loss": 0.6987,
      "step": 19296
    },
    {
      "epoch": 0.7997430477848232,
      "grad_norm": 0.42241519689559937,
      "learning_rate": 1.0014919806042523e-06,
      "loss": 0.6925,
      "step": 19297
    },
    {
      "epoch": 0.7997844916904969,
      "grad_norm": 0.3853236436843872,
      "learning_rate": 1.001284761075884e-06,
      "loss": 0.6838,
      "step": 19298
    },
    {
      "epoch": 0.7998259355961705,
      "grad_norm": 0.3837362229824066,
      "learning_rate": 1.0010775415475155e-06,
      "loss": 0.6135,
      "step": 19299
    },
    {
      "epoch": 0.7998673795018443,
      "grad_norm": 0.4171057343482971,
      "learning_rate": 1.0008703220191471e-06,
      "loss": 0.6453,
      "step": 19300
    },
    {
      "epoch": 0.7999088234075179,
      "grad_norm": 0.4310872554779053,
      "learning_rate": 1.000663102490779e-06,
      "loss": 0.6536,
      "step": 19301
    },
    {
      "epoch": 0.7999502673131916,
      "grad_norm": 0.41878220438957214,
      "learning_rate": 1.0004558829624105e-06,
      "loss": 0.7063,
      "step": 19302
    },
    {
      "epoch": 0.7999917112188653,
      "grad_norm": 0.42537620663642883,
      "learning_rate": 1.0002486634340421e-06,
      "loss": 0.6919,
      "step": 19303
    },
    {
      "epoch": 0.800033155124539,
      "grad_norm": 0.3814378082752228,
      "learning_rate": 1.0000414439056737e-06,
      "loss": 0.6536,
      "step": 19304
    },
    {
      "epoch": 0.8000745990302126,
      "grad_norm": 0.40808889269828796,
      "learning_rate": 9.998342243773055e-07,
      "loss": 0.6558,
      "step": 19305
    },
    {
      "epoch": 0.8001160429358862,
      "grad_norm": 0.3830588459968567,
      "learning_rate": 9.996270048489371e-07,
      "loss": 0.6656,
      "step": 19306
    },
    {
      "epoch": 0.80015748684156,
      "grad_norm": 0.4424755573272705,
      "learning_rate": 9.994197853205687e-07,
      "loss": 0.6902,
      "step": 19307
    },
    {
      "epoch": 0.8001989307472336,
      "grad_norm": 0.3978821039199829,
      "learning_rate": 9.992125657922003e-07,
      "loss": 0.6287,
      "step": 19308
    },
    {
      "epoch": 0.8002403746529073,
      "grad_norm": 0.4131268858909607,
      "learning_rate": 9.990053462638321e-07,
      "loss": 0.6476,
      "step": 19309
    },
    {
      "epoch": 0.8002818185585809,
      "grad_norm": 0.43022796511650085,
      "learning_rate": 9.987981267354637e-07,
      "loss": 0.6949,
      "step": 19310
    },
    {
      "epoch": 0.8003232624642547,
      "grad_norm": 0.38939306139945984,
      "learning_rate": 9.985909072070953e-07,
      "loss": 0.6432,
      "step": 19311
    },
    {
      "epoch": 0.8003647063699283,
      "grad_norm": 0.43666860461235046,
      "learning_rate": 9.98383687678727e-07,
      "loss": 0.6705,
      "step": 19312
    },
    {
      "epoch": 0.800406150275602,
      "grad_norm": 0.39608246088027954,
      "learning_rate": 9.981764681503585e-07,
      "loss": 0.6694,
      "step": 19313
    },
    {
      "epoch": 0.8004475941812756,
      "grad_norm": 0.40720611810684204,
      "learning_rate": 9.979692486219903e-07,
      "loss": 0.6625,
      "step": 19314
    },
    {
      "epoch": 0.8004890380869493,
      "grad_norm": 0.40317606925964355,
      "learning_rate": 9.97762029093622e-07,
      "loss": 0.6455,
      "step": 19315
    },
    {
      "epoch": 0.800530481992623,
      "grad_norm": 0.4622092545032501,
      "learning_rate": 9.975548095652535e-07,
      "loss": 0.7468,
      "step": 19316
    },
    {
      "epoch": 0.8005719258982966,
      "grad_norm": 0.4301154911518097,
      "learning_rate": 9.973475900368851e-07,
      "loss": 0.655,
      "step": 19317
    },
    {
      "epoch": 0.8006133698039704,
      "grad_norm": 0.3981773853302002,
      "learning_rate": 9.97140370508517e-07,
      "loss": 0.6176,
      "step": 19318
    },
    {
      "epoch": 0.800654813709644,
      "grad_norm": 0.4526747465133667,
      "learning_rate": 9.969331509801485e-07,
      "loss": 0.7174,
      "step": 19319
    },
    {
      "epoch": 0.8006962576153177,
      "grad_norm": 0.42840272188186646,
      "learning_rate": 9.967259314517801e-07,
      "loss": 0.679,
      "step": 19320
    },
    {
      "epoch": 0.8007377015209913,
      "grad_norm": 0.4032116234302521,
      "learning_rate": 9.965187119234117e-07,
      "loss": 0.6471,
      "step": 19321
    },
    {
      "epoch": 0.8007791454266651,
      "grad_norm": 0.436919242143631,
      "learning_rate": 9.963114923950435e-07,
      "loss": 0.6798,
      "step": 19322
    },
    {
      "epoch": 0.8008205893323387,
      "grad_norm": 0.4440711736679077,
      "learning_rate": 9.961042728666751e-07,
      "loss": 0.6509,
      "step": 19323
    },
    {
      "epoch": 0.8008620332380123,
      "grad_norm": 0.41983041167259216,
      "learning_rate": 9.958970533383067e-07,
      "loss": 0.6814,
      "step": 19324
    },
    {
      "epoch": 0.800903477143686,
      "grad_norm": 0.4386483430862427,
      "learning_rate": 9.956898338099383e-07,
      "loss": 0.6748,
      "step": 19325
    },
    {
      "epoch": 0.8009449210493597,
      "grad_norm": 0.4092095196247101,
      "learning_rate": 9.954826142815701e-07,
      "loss": 0.6581,
      "step": 19326
    },
    {
      "epoch": 0.8009863649550334,
      "grad_norm": 0.3955235779285431,
      "learning_rate": 9.952753947532017e-07,
      "loss": 0.6136,
      "step": 19327
    },
    {
      "epoch": 0.801027808860707,
      "grad_norm": 0.4116622507572174,
      "learning_rate": 9.950681752248333e-07,
      "loss": 0.7034,
      "step": 19328
    },
    {
      "epoch": 0.8010692527663807,
      "grad_norm": 0.4314057528972626,
      "learning_rate": 9.94860955696465e-07,
      "loss": 0.6418,
      "step": 19329
    },
    {
      "epoch": 0.8011106966720544,
      "grad_norm": 0.4565576910972595,
      "learning_rate": 9.946537361680965e-07,
      "loss": 0.6771,
      "step": 19330
    },
    {
      "epoch": 0.8011521405777281,
      "grad_norm": 0.3949267864227295,
      "learning_rate": 9.944465166397283e-07,
      "loss": 0.6364,
      "step": 19331
    },
    {
      "epoch": 0.8011935844834017,
      "grad_norm": 0.4094487130641937,
      "learning_rate": 9.9423929711136e-07,
      "loss": 0.6448,
      "step": 19332
    },
    {
      "epoch": 0.8012350283890753,
      "grad_norm": 0.45284906029701233,
      "learning_rate": 9.940320775829915e-07,
      "loss": 0.6914,
      "step": 19333
    },
    {
      "epoch": 0.8012764722947491,
      "grad_norm": 0.45013436675071716,
      "learning_rate": 9.938248580546231e-07,
      "loss": 0.6696,
      "step": 19334
    },
    {
      "epoch": 0.8013179162004227,
      "grad_norm": 0.47605466842651367,
      "learning_rate": 9.93617638526255e-07,
      "loss": 0.6858,
      "step": 19335
    },
    {
      "epoch": 0.8013593601060964,
      "grad_norm": 0.4308841824531555,
      "learning_rate": 9.934104189978865e-07,
      "loss": 0.6726,
      "step": 19336
    },
    {
      "epoch": 0.80140080401177,
      "grad_norm": 0.4405074417591095,
      "learning_rate": 9.932031994695181e-07,
      "loss": 0.6694,
      "step": 19337
    },
    {
      "epoch": 0.8014422479174438,
      "grad_norm": 0.4123457372188568,
      "learning_rate": 9.929959799411497e-07,
      "loss": 0.6497,
      "step": 19338
    },
    {
      "epoch": 0.8014836918231174,
      "grad_norm": 0.3954439163208008,
      "learning_rate": 9.927887604127815e-07,
      "loss": 0.6716,
      "step": 19339
    },
    {
      "epoch": 0.801525135728791,
      "grad_norm": 0.4239901900291443,
      "learning_rate": 9.925815408844131e-07,
      "loss": 0.6782,
      "step": 19340
    },
    {
      "epoch": 0.8015665796344648,
      "grad_norm": 0.39566484093666077,
      "learning_rate": 9.923743213560447e-07,
      "loss": 0.6477,
      "step": 19341
    },
    {
      "epoch": 0.8016080235401384,
      "grad_norm": 0.40866681933403015,
      "learning_rate": 9.921671018276763e-07,
      "loss": 0.632,
      "step": 19342
    },
    {
      "epoch": 0.8016494674458121,
      "grad_norm": 0.40841877460479736,
      "learning_rate": 9.919598822993081e-07,
      "loss": 0.6263,
      "step": 19343
    },
    {
      "epoch": 0.8016909113514857,
      "grad_norm": 0.4263053834438324,
      "learning_rate": 9.917526627709397e-07,
      "loss": 0.7036,
      "step": 19344
    },
    {
      "epoch": 0.8017323552571595,
      "grad_norm": 0.40292561054229736,
      "learning_rate": 9.915454432425713e-07,
      "loss": 0.6628,
      "step": 19345
    },
    {
      "epoch": 0.8017737991628331,
      "grad_norm": 0.4185163974761963,
      "learning_rate": 9.91338223714203e-07,
      "loss": 0.6853,
      "step": 19346
    },
    {
      "epoch": 0.8018152430685068,
      "grad_norm": 0.39670273661613464,
      "learning_rate": 9.911310041858345e-07,
      "loss": 0.6697,
      "step": 19347
    },
    {
      "epoch": 0.8018566869741804,
      "grad_norm": 0.3813093304634094,
      "learning_rate": 9.909237846574663e-07,
      "loss": 0.6354,
      "step": 19348
    },
    {
      "epoch": 0.8018981308798541,
      "grad_norm": 0.4442175328731537,
      "learning_rate": 9.90716565129098e-07,
      "loss": 0.6532,
      "step": 19349
    },
    {
      "epoch": 0.8019395747855278,
      "grad_norm": 0.4058321714401245,
      "learning_rate": 9.905093456007295e-07,
      "loss": 0.679,
      "step": 19350
    },
    {
      "epoch": 0.8019810186912014,
      "grad_norm": 0.4311847984790802,
      "learning_rate": 9.903021260723611e-07,
      "loss": 0.6803,
      "step": 19351
    },
    {
      "epoch": 0.8020224625968752,
      "grad_norm": 0.43009698390960693,
      "learning_rate": 9.90094906543993e-07,
      "loss": 0.6642,
      "step": 19352
    },
    {
      "epoch": 0.8020639065025488,
      "grad_norm": 0.4170438349246979,
      "learning_rate": 9.898876870156245e-07,
      "loss": 0.6372,
      "step": 19353
    },
    {
      "epoch": 0.8021053504082225,
      "grad_norm": 0.4326097369194031,
      "learning_rate": 9.896804674872561e-07,
      "loss": 0.6906,
      "step": 19354
    },
    {
      "epoch": 0.8021467943138961,
      "grad_norm": 0.41405466198921204,
      "learning_rate": 9.894732479588877e-07,
      "loss": 0.6782,
      "step": 19355
    },
    {
      "epoch": 0.8021882382195699,
      "grad_norm": 0.4072588086128235,
      "learning_rate": 9.892660284305193e-07,
      "loss": 0.6083,
      "step": 19356
    },
    {
      "epoch": 0.8022296821252435,
      "grad_norm": 0.45250436663627625,
      "learning_rate": 9.890588089021511e-07,
      "loss": 0.6698,
      "step": 19357
    },
    {
      "epoch": 0.8022711260309171,
      "grad_norm": 0.4154640734195709,
      "learning_rate": 9.888515893737827e-07,
      "loss": 0.6534,
      "step": 19358
    },
    {
      "epoch": 0.8023125699365908,
      "grad_norm": 0.4051082134246826,
      "learning_rate": 9.886443698454143e-07,
      "loss": 0.6555,
      "step": 19359
    },
    {
      "epoch": 0.8023540138422645,
      "grad_norm": 0.4125066101551056,
      "learning_rate": 9.88437150317046e-07,
      "loss": 0.6592,
      "step": 19360
    },
    {
      "epoch": 0.8023954577479382,
      "grad_norm": 0.42021554708480835,
      "learning_rate": 9.882299307886777e-07,
      "loss": 0.6765,
      "step": 19361
    },
    {
      "epoch": 0.8024369016536118,
      "grad_norm": 0.4037286937236786,
      "learning_rate": 9.880227112603093e-07,
      "loss": 0.6582,
      "step": 19362
    },
    {
      "epoch": 0.8024783455592855,
      "grad_norm": 0.435926616191864,
      "learning_rate": 9.87815491731941e-07,
      "loss": 0.6335,
      "step": 19363
    },
    {
      "epoch": 0.8025197894649592,
      "grad_norm": 0.39977800846099854,
      "learning_rate": 9.876082722035725e-07,
      "loss": 0.6516,
      "step": 19364
    },
    {
      "epoch": 0.8025612333706329,
      "grad_norm": 0.4473310708999634,
      "learning_rate": 9.874010526752043e-07,
      "loss": 0.6975,
      "step": 19365
    },
    {
      "epoch": 0.8026026772763065,
      "grad_norm": 0.38879403471946716,
      "learning_rate": 9.87193833146836e-07,
      "loss": 0.6317,
      "step": 19366
    },
    {
      "epoch": 0.8026441211819801,
      "grad_norm": 0.3910500407218933,
      "learning_rate": 9.869866136184675e-07,
      "loss": 0.6501,
      "step": 19367
    },
    {
      "epoch": 0.8026855650876539,
      "grad_norm": 0.46971505880355835,
      "learning_rate": 9.867793940900991e-07,
      "loss": 0.6902,
      "step": 19368
    },
    {
      "epoch": 0.8027270089933275,
      "grad_norm": 0.4669254422187805,
      "learning_rate": 9.865721745617307e-07,
      "loss": 0.6746,
      "step": 19369
    },
    {
      "epoch": 0.8027684528990012,
      "grad_norm": 0.41468217968940735,
      "learning_rate": 9.863649550333625e-07,
      "loss": 0.6532,
      "step": 19370
    },
    {
      "epoch": 0.8028098968046748,
      "grad_norm": 0.45452365279197693,
      "learning_rate": 9.861577355049941e-07,
      "loss": 0.7517,
      "step": 19371
    },
    {
      "epoch": 0.8028513407103486,
      "grad_norm": 0.44085124135017395,
      "learning_rate": 9.859505159766257e-07,
      "loss": 0.6741,
      "step": 19372
    },
    {
      "epoch": 0.8028927846160222,
      "grad_norm": 0.4711313247680664,
      "learning_rate": 9.857432964482573e-07,
      "loss": 0.6639,
      "step": 19373
    },
    {
      "epoch": 0.8029342285216959,
      "grad_norm": 0.4268048405647278,
      "learning_rate": 9.855360769198891e-07,
      "loss": 0.7173,
      "step": 19374
    },
    {
      "epoch": 0.8029756724273696,
      "grad_norm": 0.4161265790462494,
      "learning_rate": 9.853288573915207e-07,
      "loss": 0.6438,
      "step": 19375
    },
    {
      "epoch": 0.8030171163330432,
      "grad_norm": 0.44771715998649597,
      "learning_rate": 9.851216378631523e-07,
      "loss": 0.7151,
      "step": 19376
    },
    {
      "epoch": 0.8030585602387169,
      "grad_norm": 0.4126644730567932,
      "learning_rate": 9.84914418334784e-07,
      "loss": 0.6736,
      "step": 19377
    },
    {
      "epoch": 0.8031000041443905,
      "grad_norm": 0.44197171926498413,
      "learning_rate": 9.847071988064157e-07,
      "loss": 0.7146,
      "step": 19378
    },
    {
      "epoch": 0.8031414480500643,
      "grad_norm": 0.4149807393550873,
      "learning_rate": 9.844999792780473e-07,
      "loss": 0.6614,
      "step": 19379
    },
    {
      "epoch": 0.8031828919557379,
      "grad_norm": 0.45384860038757324,
      "learning_rate": 9.84292759749679e-07,
      "loss": 0.6643,
      "step": 19380
    },
    {
      "epoch": 0.8032243358614116,
      "grad_norm": 0.40754759311676025,
      "learning_rate": 9.840855402213105e-07,
      "loss": 0.6157,
      "step": 19381
    },
    {
      "epoch": 0.8032657797670852,
      "grad_norm": 0.413920134305954,
      "learning_rate": 9.838783206929421e-07,
      "loss": 0.6414,
      "step": 19382
    },
    {
      "epoch": 0.803307223672759,
      "grad_norm": 0.3815504014492035,
      "learning_rate": 9.83671101164574e-07,
      "loss": 0.666,
      "step": 19383
    },
    {
      "epoch": 0.8033486675784326,
      "grad_norm": 0.43437597155570984,
      "learning_rate": 9.834638816362055e-07,
      "loss": 0.7161,
      "step": 19384
    },
    {
      "epoch": 0.8033901114841062,
      "grad_norm": 0.4221706986427307,
      "learning_rate": 9.832566621078371e-07,
      "loss": 0.6672,
      "step": 19385
    },
    {
      "epoch": 0.80343155538978,
      "grad_norm": 0.4401795268058777,
      "learning_rate": 9.830494425794687e-07,
      "loss": 0.6906,
      "step": 19386
    },
    {
      "epoch": 0.8034729992954536,
      "grad_norm": 0.39666101336479187,
      "learning_rate": 9.828422230511005e-07,
      "loss": 0.6844,
      "step": 19387
    },
    {
      "epoch": 0.8035144432011273,
      "grad_norm": 0.38539233803749084,
      "learning_rate": 9.826350035227321e-07,
      "loss": 0.6394,
      "step": 19388
    },
    {
      "epoch": 0.8035558871068009,
      "grad_norm": 0.39911550283432007,
      "learning_rate": 9.824277839943637e-07,
      "loss": 0.646,
      "step": 19389
    },
    {
      "epoch": 0.8035973310124747,
      "grad_norm": 0.43002066016197205,
      "learning_rate": 9.822205644659953e-07,
      "loss": 0.6443,
      "step": 19390
    },
    {
      "epoch": 0.8036387749181483,
      "grad_norm": 0.3984774947166443,
      "learning_rate": 9.820133449376271e-07,
      "loss": 0.6462,
      "step": 19391
    },
    {
      "epoch": 0.803680218823822,
      "grad_norm": 0.4080101251602173,
      "learning_rate": 9.818061254092587e-07,
      "loss": 0.6895,
      "step": 19392
    },
    {
      "epoch": 0.8037216627294956,
      "grad_norm": 0.4028632640838623,
      "learning_rate": 9.815989058808903e-07,
      "loss": 0.6991,
      "step": 19393
    },
    {
      "epoch": 0.8037631066351693,
      "grad_norm": 0.4477446973323822,
      "learning_rate": 9.81391686352522e-07,
      "loss": 0.651,
      "step": 19394
    },
    {
      "epoch": 0.803804550540843,
      "grad_norm": 0.43847987055778503,
      "learning_rate": 9.811844668241535e-07,
      "loss": 0.6578,
      "step": 19395
    },
    {
      "epoch": 0.8038459944465166,
      "grad_norm": 0.39322584867477417,
      "learning_rate": 9.809772472957853e-07,
      "loss": 0.6725,
      "step": 19396
    },
    {
      "epoch": 0.8038874383521903,
      "grad_norm": 0.4845471680164337,
      "learning_rate": 9.80770027767417e-07,
      "loss": 0.7529,
      "step": 19397
    },
    {
      "epoch": 0.803928882257864,
      "grad_norm": 0.3780086636543274,
      "learning_rate": 9.805628082390485e-07,
      "loss": 0.6129,
      "step": 19398
    },
    {
      "epoch": 0.8039703261635377,
      "grad_norm": 0.4300594925880432,
      "learning_rate": 9.803555887106801e-07,
      "loss": 0.6809,
      "step": 19399
    },
    {
      "epoch": 0.8040117700692113,
      "grad_norm": 0.41659486293792725,
      "learning_rate": 9.80148369182312e-07,
      "loss": 0.6488,
      "step": 19400
    },
    {
      "epoch": 0.8040532139748849,
      "grad_norm": 0.4114384055137634,
      "learning_rate": 9.799411496539435e-07,
      "loss": 0.707,
      "step": 19401
    },
    {
      "epoch": 0.8040946578805587,
      "grad_norm": 0.44136449694633484,
      "learning_rate": 9.797339301255751e-07,
      "loss": 0.7041,
      "step": 19402
    },
    {
      "epoch": 0.8041361017862323,
      "grad_norm": 0.39955055713653564,
      "learning_rate": 9.795267105972067e-07,
      "loss": 0.6431,
      "step": 19403
    },
    {
      "epoch": 0.804177545691906,
      "grad_norm": 0.4023345708847046,
      "learning_rate": 9.793194910688385e-07,
      "loss": 0.6381,
      "step": 19404
    },
    {
      "epoch": 0.8042189895975796,
      "grad_norm": 0.40706032514572144,
      "learning_rate": 9.791122715404701e-07,
      "loss": 0.6621,
      "step": 19405
    },
    {
      "epoch": 0.8042604335032534,
      "grad_norm": 0.38676050305366516,
      "learning_rate": 9.789050520121017e-07,
      "loss": 0.6912,
      "step": 19406
    },
    {
      "epoch": 0.804301877408927,
      "grad_norm": 0.42181622982025146,
      "learning_rate": 9.786978324837333e-07,
      "loss": 0.6753,
      "step": 19407
    },
    {
      "epoch": 0.8043433213146007,
      "grad_norm": 0.42874929308891296,
      "learning_rate": 9.78490612955365e-07,
      "loss": 0.6904,
      "step": 19408
    },
    {
      "epoch": 0.8043847652202744,
      "grad_norm": 0.38372957706451416,
      "learning_rate": 9.782833934269967e-07,
      "loss": 0.6646,
      "step": 19409
    },
    {
      "epoch": 0.804426209125948,
      "grad_norm": 0.384045273065567,
      "learning_rate": 9.780761738986283e-07,
      "loss": 0.6046,
      "step": 19410
    },
    {
      "epoch": 0.8044676530316217,
      "grad_norm": 0.45876365900039673,
      "learning_rate": 9.7786895437026e-07,
      "loss": 0.6864,
      "step": 19411
    },
    {
      "epoch": 0.8045090969372953,
      "grad_norm": 0.39014485478401184,
      "learning_rate": 9.776617348418915e-07,
      "loss": 0.657,
      "step": 19412
    },
    {
      "epoch": 0.8045505408429691,
      "grad_norm": 0.4047342836856842,
      "learning_rate": 9.774545153135233e-07,
      "loss": 0.7008,
      "step": 19413
    },
    {
      "epoch": 0.8045919847486427,
      "grad_norm": 0.3980944752693176,
      "learning_rate": 9.77247295785155e-07,
      "loss": 0.6489,
      "step": 19414
    },
    {
      "epoch": 0.8046334286543164,
      "grad_norm": 0.5027517080307007,
      "learning_rate": 9.770400762567865e-07,
      "loss": 0.6914,
      "step": 19415
    },
    {
      "epoch": 0.80467487255999,
      "grad_norm": 0.43754228949546814,
      "learning_rate": 9.768328567284181e-07,
      "loss": 0.6543,
      "step": 19416
    },
    {
      "epoch": 0.8047163164656638,
      "grad_norm": 0.405769020318985,
      "learning_rate": 9.7662563720005e-07,
      "loss": 0.6836,
      "step": 19417
    },
    {
      "epoch": 0.8047577603713374,
      "grad_norm": 0.41399651765823364,
      "learning_rate": 9.764184176716815e-07,
      "loss": 0.6846,
      "step": 19418
    },
    {
      "epoch": 0.804799204277011,
      "grad_norm": 0.39244070649147034,
      "learning_rate": 9.762111981433131e-07,
      "loss": 0.6724,
      "step": 19419
    },
    {
      "epoch": 0.8048406481826847,
      "grad_norm": 0.4376327693462372,
      "learning_rate": 9.760039786149447e-07,
      "loss": 0.6703,
      "step": 19420
    },
    {
      "epoch": 0.8048820920883584,
      "grad_norm": 0.4239356815814972,
      "learning_rate": 9.757967590865763e-07,
      "loss": 0.6726,
      "step": 19421
    },
    {
      "epoch": 0.8049235359940321,
      "grad_norm": 0.40637609362602234,
      "learning_rate": 9.755895395582081e-07,
      "loss": 0.6954,
      "step": 19422
    },
    {
      "epoch": 0.8049649798997057,
      "grad_norm": 0.4169522821903229,
      "learning_rate": 9.753823200298397e-07,
      "loss": 0.6901,
      "step": 19423
    },
    {
      "epoch": 0.8050064238053795,
      "grad_norm": 0.4658155143260956,
      "learning_rate": 9.751751005014713e-07,
      "loss": 0.6987,
      "step": 19424
    },
    {
      "epoch": 0.8050478677110531,
      "grad_norm": 0.43301939964294434,
      "learning_rate": 9.74967880973103e-07,
      "loss": 0.6217,
      "step": 19425
    },
    {
      "epoch": 0.8050893116167268,
      "grad_norm": 0.43847575783729553,
      "learning_rate": 9.747606614447347e-07,
      "loss": 0.6384,
      "step": 19426
    },
    {
      "epoch": 0.8051307555224004,
      "grad_norm": 0.4266357719898224,
      "learning_rate": 9.745534419163663e-07,
      "loss": 0.6544,
      "step": 19427
    },
    {
      "epoch": 0.805172199428074,
      "grad_norm": 0.3924982249736786,
      "learning_rate": 9.74346222387998e-07,
      "loss": 0.686,
      "step": 19428
    },
    {
      "epoch": 0.8052136433337478,
      "grad_norm": 0.45344647765159607,
      "learning_rate": 9.741390028596295e-07,
      "loss": 0.7053,
      "step": 19429
    },
    {
      "epoch": 0.8052550872394214,
      "grad_norm": 0.39786022901535034,
      "learning_rate": 9.739317833312611e-07,
      "loss": 0.634,
      "step": 19430
    },
    {
      "epoch": 0.8052965311450951,
      "grad_norm": 0.400820791721344,
      "learning_rate": 9.73724563802893e-07,
      "loss": 0.665,
      "step": 19431
    },
    {
      "epoch": 0.8053379750507688,
      "grad_norm": 0.42308247089385986,
      "learning_rate": 9.735173442745245e-07,
      "loss": 0.6752,
      "step": 19432
    },
    {
      "epoch": 0.8053794189564425,
      "grad_norm": 0.4272763431072235,
      "learning_rate": 9.733101247461561e-07,
      "loss": 0.7058,
      "step": 19433
    },
    {
      "epoch": 0.8054208628621161,
      "grad_norm": 0.4212509095668793,
      "learning_rate": 9.731029052177877e-07,
      "loss": 0.6118,
      "step": 19434
    },
    {
      "epoch": 0.8054623067677898,
      "grad_norm": 0.3884056806564331,
      "learning_rate": 9.728956856894195e-07,
      "loss": 0.6007,
      "step": 19435
    },
    {
      "epoch": 0.8055037506734635,
      "grad_norm": 0.46689823269844055,
      "learning_rate": 9.726884661610511e-07,
      "loss": 0.6829,
      "step": 19436
    },
    {
      "epoch": 0.8055451945791371,
      "grad_norm": 0.4108678996562958,
      "learning_rate": 9.724812466326827e-07,
      "loss": 0.5872,
      "step": 19437
    },
    {
      "epoch": 0.8055866384848108,
      "grad_norm": 0.43207234144210815,
      "learning_rate": 9.722740271043143e-07,
      "loss": 0.6777,
      "step": 19438
    },
    {
      "epoch": 0.8056280823904844,
      "grad_norm": 0.4216160774230957,
      "learning_rate": 9.720668075759461e-07,
      "loss": 0.6327,
      "step": 19439
    },
    {
      "epoch": 0.8056695262961582,
      "grad_norm": 0.410572350025177,
      "learning_rate": 9.718595880475777e-07,
      "loss": 0.6543,
      "step": 19440
    },
    {
      "epoch": 0.8057109702018318,
      "grad_norm": 0.3959107995033264,
      "learning_rate": 9.716523685192093e-07,
      "loss": 0.6434,
      "step": 19441
    },
    {
      "epoch": 0.8057524141075055,
      "grad_norm": 0.4042576253414154,
      "learning_rate": 9.71445148990841e-07,
      "loss": 0.6266,
      "step": 19442
    },
    {
      "epoch": 0.8057938580131792,
      "grad_norm": 0.42086338996887207,
      "learning_rate": 9.712379294624725e-07,
      "loss": 0.6213,
      "step": 19443
    },
    {
      "epoch": 0.8058353019188529,
      "grad_norm": 0.43096107244491577,
      "learning_rate": 9.710307099341043e-07,
      "loss": 0.6206,
      "step": 19444
    },
    {
      "epoch": 0.8058767458245265,
      "grad_norm": 0.4263373017311096,
      "learning_rate": 9.70823490405736e-07,
      "loss": 0.6843,
      "step": 19445
    },
    {
      "epoch": 0.8059181897302001,
      "grad_norm": 0.3968210518360138,
      "learning_rate": 9.706162708773675e-07,
      "loss": 0.6324,
      "step": 19446
    },
    {
      "epoch": 0.8059596336358739,
      "grad_norm": 0.46057218313217163,
      "learning_rate": 9.704090513489991e-07,
      "loss": 0.7241,
      "step": 19447
    },
    {
      "epoch": 0.8060010775415475,
      "grad_norm": 0.45166391134262085,
      "learning_rate": 9.70201831820631e-07,
      "loss": 0.6716,
      "step": 19448
    },
    {
      "epoch": 0.8060425214472212,
      "grad_norm": 0.40858060121536255,
      "learning_rate": 9.699946122922625e-07,
      "loss": 0.661,
      "step": 19449
    },
    {
      "epoch": 0.8060839653528948,
      "grad_norm": 0.39239588379859924,
      "learning_rate": 9.697873927638941e-07,
      "loss": 0.5837,
      "step": 19450
    },
    {
      "epoch": 0.8061254092585686,
      "grad_norm": 0.40284907817840576,
      "learning_rate": 9.695801732355257e-07,
      "loss": 0.6699,
      "step": 19451
    },
    {
      "epoch": 0.8061668531642422,
      "grad_norm": 0.380706250667572,
      "learning_rate": 9.693729537071575e-07,
      "loss": 0.7168,
      "step": 19452
    },
    {
      "epoch": 0.8062082970699158,
      "grad_norm": 0.41635850071907043,
      "learning_rate": 9.691657341787891e-07,
      "loss": 0.6764,
      "step": 19453
    },
    {
      "epoch": 0.8062497409755895,
      "grad_norm": 0.40801382064819336,
      "learning_rate": 9.689585146504207e-07,
      "loss": 0.7042,
      "step": 19454
    },
    {
      "epoch": 0.8062911848812632,
      "grad_norm": 0.4157634973526001,
      "learning_rate": 9.687512951220523e-07,
      "loss": 0.6982,
      "step": 19455
    },
    {
      "epoch": 0.8063326287869369,
      "grad_norm": 0.3871079087257385,
      "learning_rate": 9.68544075593684e-07,
      "loss": 0.6608,
      "step": 19456
    },
    {
      "epoch": 0.8063740726926105,
      "grad_norm": 0.4332897663116455,
      "learning_rate": 9.683368560653157e-07,
      "loss": 0.7198,
      "step": 19457
    },
    {
      "epoch": 0.8064155165982843,
      "grad_norm": 0.4432371258735657,
      "learning_rate": 9.681296365369473e-07,
      "loss": 0.7292,
      "step": 19458
    },
    {
      "epoch": 0.8064569605039579,
      "grad_norm": 0.41420260071754456,
      "learning_rate": 9.67922417008579e-07,
      "loss": 0.6802,
      "step": 19459
    },
    {
      "epoch": 0.8064984044096316,
      "grad_norm": 0.3928356468677521,
      "learning_rate": 9.677151974802105e-07,
      "loss": 0.6592,
      "step": 19460
    },
    {
      "epoch": 0.8065398483153052,
      "grad_norm": 0.4358169436454773,
      "learning_rate": 9.675079779518423e-07,
      "loss": 0.676,
      "step": 19461
    },
    {
      "epoch": 0.8065812922209789,
      "grad_norm": 0.3987569510936737,
      "learning_rate": 9.67300758423474e-07,
      "loss": 0.6584,
      "step": 19462
    },
    {
      "epoch": 0.8066227361266526,
      "grad_norm": 0.4270404577255249,
      "learning_rate": 9.670935388951055e-07,
      "loss": 0.6782,
      "step": 19463
    },
    {
      "epoch": 0.8066641800323262,
      "grad_norm": 0.38269197940826416,
      "learning_rate": 9.668863193667371e-07,
      "loss": 0.6333,
      "step": 19464
    },
    {
      "epoch": 0.8067056239379999,
      "grad_norm": 0.4105305075645447,
      "learning_rate": 9.66679099838369e-07,
      "loss": 0.6708,
      "step": 19465
    },
    {
      "epoch": 0.8067470678436736,
      "grad_norm": 0.4493926763534546,
      "learning_rate": 9.664718803100005e-07,
      "loss": 0.6888,
      "step": 19466
    },
    {
      "epoch": 0.8067885117493473,
      "grad_norm": 0.44324979186058044,
      "learning_rate": 9.662646607816321e-07,
      "loss": 0.6615,
      "step": 19467
    },
    {
      "epoch": 0.8068299556550209,
      "grad_norm": 0.4452129304409027,
      "learning_rate": 9.660574412532637e-07,
      "loss": 0.7203,
      "step": 19468
    },
    {
      "epoch": 0.8068713995606946,
      "grad_norm": 0.42313480377197266,
      "learning_rate": 9.658502217248953e-07,
      "loss": 0.6858,
      "step": 19469
    },
    {
      "epoch": 0.8069128434663683,
      "grad_norm": 0.42199939489364624,
      "learning_rate": 9.656430021965271e-07,
      "loss": 0.6122,
      "step": 19470
    },
    {
      "epoch": 0.8069542873720419,
      "grad_norm": 0.40722450613975525,
      "learning_rate": 9.654357826681587e-07,
      "loss": 0.6372,
      "step": 19471
    },
    {
      "epoch": 0.8069957312777156,
      "grad_norm": 0.4141182005405426,
      "learning_rate": 9.652285631397903e-07,
      "loss": 0.6913,
      "step": 19472
    },
    {
      "epoch": 0.8070371751833892,
      "grad_norm": 0.39499810338020325,
      "learning_rate": 9.65021343611422e-07,
      "loss": 0.6057,
      "step": 19473
    },
    {
      "epoch": 0.807078619089063,
      "grad_norm": 0.3915005326271057,
      "learning_rate": 9.648141240830538e-07,
      "loss": 0.6123,
      "step": 19474
    },
    {
      "epoch": 0.8071200629947366,
      "grad_norm": 0.40600264072418213,
      "learning_rate": 9.646069045546853e-07,
      "loss": 0.6948,
      "step": 19475
    },
    {
      "epoch": 0.8071615069004103,
      "grad_norm": 0.4054807126522064,
      "learning_rate": 9.64399685026317e-07,
      "loss": 0.6506,
      "step": 19476
    },
    {
      "epoch": 0.807202950806084,
      "grad_norm": 0.3943275511264801,
      "learning_rate": 9.641924654979485e-07,
      "loss": 0.6519,
      "step": 19477
    },
    {
      "epoch": 0.8072443947117577,
      "grad_norm": 0.43208932876586914,
      "learning_rate": 9.639852459695804e-07,
      "loss": 0.6998,
      "step": 19478
    },
    {
      "epoch": 0.8072858386174313,
      "grad_norm": 0.4362471103668213,
      "learning_rate": 9.63778026441212e-07,
      "loss": 0.6509,
      "step": 19479
    },
    {
      "epoch": 0.8073272825231049,
      "grad_norm": 0.40001848340034485,
      "learning_rate": 9.635708069128435e-07,
      "loss": 0.6833,
      "step": 19480
    },
    {
      "epoch": 0.8073687264287787,
      "grad_norm": 0.42091280221939087,
      "learning_rate": 9.633635873844751e-07,
      "loss": 0.6797,
      "step": 19481
    },
    {
      "epoch": 0.8074101703344523,
      "grad_norm": 0.4162572920322418,
      "learning_rate": 9.631563678561067e-07,
      "loss": 0.6705,
      "step": 19482
    },
    {
      "epoch": 0.807451614240126,
      "grad_norm": 0.41161155700683594,
      "learning_rate": 9.629491483277386e-07,
      "loss": 0.6556,
      "step": 19483
    },
    {
      "epoch": 0.8074930581457996,
      "grad_norm": 0.42415064573287964,
      "learning_rate": 9.627419287993701e-07,
      "loss": 0.6864,
      "step": 19484
    },
    {
      "epoch": 0.8075345020514734,
      "grad_norm": 0.4010603725910187,
      "learning_rate": 9.625347092710017e-07,
      "loss": 0.6445,
      "step": 19485
    },
    {
      "epoch": 0.807575945957147,
      "grad_norm": 0.425437331199646,
      "learning_rate": 9.623274897426333e-07,
      "loss": 0.6953,
      "step": 19486
    },
    {
      "epoch": 0.8076173898628207,
      "grad_norm": 0.4074363708496094,
      "learning_rate": 9.621202702142652e-07,
      "loss": 0.6986,
      "step": 19487
    },
    {
      "epoch": 0.8076588337684943,
      "grad_norm": 0.3950572907924652,
      "learning_rate": 9.619130506858967e-07,
      "loss": 0.6141,
      "step": 19488
    },
    {
      "epoch": 0.807700277674168,
      "grad_norm": 0.42007943987846375,
      "learning_rate": 9.617058311575283e-07,
      "loss": 0.6766,
      "step": 19489
    },
    {
      "epoch": 0.8077417215798417,
      "grad_norm": 0.39537885785102844,
      "learning_rate": 9.6149861162916e-07,
      "loss": 0.6675,
      "step": 19490
    },
    {
      "epoch": 0.8077831654855153,
      "grad_norm": 0.43586283922195435,
      "learning_rate": 9.612913921007918e-07,
      "loss": 0.641,
      "step": 19491
    },
    {
      "epoch": 0.807824609391189,
      "grad_norm": 0.42393407225608826,
      "learning_rate": 9.610841725724234e-07,
      "loss": 0.6841,
      "step": 19492
    },
    {
      "epoch": 0.8078660532968627,
      "grad_norm": 0.3910629451274872,
      "learning_rate": 9.60876953044055e-07,
      "loss": 0.6633,
      "step": 19493
    },
    {
      "epoch": 0.8079074972025364,
      "grad_norm": 0.41703030467033386,
      "learning_rate": 9.606697335156865e-07,
      "loss": 0.6989,
      "step": 19494
    },
    {
      "epoch": 0.80794894110821,
      "grad_norm": 0.40369051694869995,
      "learning_rate": 9.604625139873181e-07,
      "loss": 0.6644,
      "step": 19495
    },
    {
      "epoch": 0.8079903850138838,
      "grad_norm": 0.408910870552063,
      "learning_rate": 9.6025529445895e-07,
      "loss": 0.6299,
      "step": 19496
    },
    {
      "epoch": 0.8080318289195574,
      "grad_norm": 0.42264577746391296,
      "learning_rate": 9.600480749305815e-07,
      "loss": 0.6222,
      "step": 19497
    },
    {
      "epoch": 0.808073272825231,
      "grad_norm": 0.455897718667984,
      "learning_rate": 9.598408554022131e-07,
      "loss": 0.7235,
      "step": 19498
    },
    {
      "epoch": 0.8081147167309047,
      "grad_norm": 0.40088674426078796,
      "learning_rate": 9.596336358738447e-07,
      "loss": 0.6772,
      "step": 19499
    },
    {
      "epoch": 0.8081561606365784,
      "grad_norm": 0.38747960329055786,
      "learning_rate": 9.594264163454766e-07,
      "loss": 0.6309,
      "step": 19500
    },
    {
      "epoch": 0.8081976045422521,
      "grad_norm": 0.4007011651992798,
      "learning_rate": 9.592191968171082e-07,
      "loss": 0.6134,
      "step": 19501
    },
    {
      "epoch": 0.8082390484479257,
      "grad_norm": 0.4221867322921753,
      "learning_rate": 9.590119772887397e-07,
      "loss": 0.7185,
      "step": 19502
    },
    {
      "epoch": 0.8082804923535994,
      "grad_norm": 0.3902588486671448,
      "learning_rate": 9.588047577603713e-07,
      "loss": 0.6569,
      "step": 19503
    },
    {
      "epoch": 0.8083219362592731,
      "grad_norm": 0.39849603176116943,
      "learning_rate": 9.58597538232003e-07,
      "loss": 0.6853,
      "step": 19504
    },
    {
      "epoch": 0.8083633801649468,
      "grad_norm": 0.40166008472442627,
      "learning_rate": 9.583903187036348e-07,
      "loss": 0.6406,
      "step": 19505
    },
    {
      "epoch": 0.8084048240706204,
      "grad_norm": 0.39391639828681946,
      "learning_rate": 9.581830991752663e-07,
      "loss": 0.6743,
      "step": 19506
    },
    {
      "epoch": 0.808446267976294,
      "grad_norm": 0.4134604334831238,
      "learning_rate": 9.57975879646898e-07,
      "loss": 0.6606,
      "step": 19507
    },
    {
      "epoch": 0.8084877118819678,
      "grad_norm": 0.42527467012405396,
      "learning_rate": 9.577686601185295e-07,
      "loss": 0.657,
      "step": 19508
    },
    {
      "epoch": 0.8085291557876414,
      "grad_norm": 0.4068295955657959,
      "learning_rate": 9.575614405901614e-07,
      "loss": 0.6273,
      "step": 19509
    },
    {
      "epoch": 0.8085705996933151,
      "grad_norm": 0.48065459728240967,
      "learning_rate": 9.57354221061793e-07,
      "loss": 0.7065,
      "step": 19510
    },
    {
      "epoch": 0.8086120435989888,
      "grad_norm": 0.4314187467098236,
      "learning_rate": 9.571470015334245e-07,
      "loss": 0.6752,
      "step": 19511
    },
    {
      "epoch": 0.8086534875046625,
      "grad_norm": 0.44857847690582275,
      "learning_rate": 9.569397820050561e-07,
      "loss": 0.7059,
      "step": 19512
    },
    {
      "epoch": 0.8086949314103361,
      "grad_norm": 0.4418477714061737,
      "learning_rate": 9.56732562476688e-07,
      "loss": 0.7152,
      "step": 19513
    },
    {
      "epoch": 0.8087363753160097,
      "grad_norm": 0.390018105506897,
      "learning_rate": 9.565253429483196e-07,
      "loss": 0.6215,
      "step": 19514
    },
    {
      "epoch": 0.8087778192216835,
      "grad_norm": 0.42457568645477295,
      "learning_rate": 9.563181234199512e-07,
      "loss": 0.6974,
      "step": 19515
    },
    {
      "epoch": 0.8088192631273571,
      "grad_norm": 0.40534886717796326,
      "learning_rate": 9.561109038915827e-07,
      "loss": 0.6145,
      "step": 19516
    },
    {
      "epoch": 0.8088607070330308,
      "grad_norm": 0.4359142780303955,
      "learning_rate": 9.559036843632143e-07,
      "loss": 0.6687,
      "step": 19517
    },
    {
      "epoch": 0.8089021509387044,
      "grad_norm": 0.4252561330795288,
      "learning_rate": 9.556964648348462e-07,
      "loss": 0.7236,
      "step": 19518
    },
    {
      "epoch": 0.8089435948443782,
      "grad_norm": 0.4606792628765106,
      "learning_rate": 9.554892453064778e-07,
      "loss": 0.7036,
      "step": 19519
    },
    {
      "epoch": 0.8089850387500518,
      "grad_norm": 0.403179407119751,
      "learning_rate": 9.552820257781093e-07,
      "loss": 0.6648,
      "step": 19520
    },
    {
      "epoch": 0.8090264826557255,
      "grad_norm": 0.39424610137939453,
      "learning_rate": 9.55074806249741e-07,
      "loss": 0.6053,
      "step": 19521
    },
    {
      "epoch": 0.8090679265613991,
      "grad_norm": 0.35912024974823,
      "learning_rate": 9.548675867213728e-07,
      "loss": 0.6118,
      "step": 19522
    },
    {
      "epoch": 0.8091093704670728,
      "grad_norm": 0.44607606530189514,
      "learning_rate": 9.546603671930044e-07,
      "loss": 0.6936,
      "step": 19523
    },
    {
      "epoch": 0.8091508143727465,
      "grad_norm": 0.41180214285850525,
      "learning_rate": 9.54453147664636e-07,
      "loss": 0.6327,
      "step": 19524
    },
    {
      "epoch": 0.8091922582784201,
      "grad_norm": 0.4313219487667084,
      "learning_rate": 9.542459281362675e-07,
      "loss": 0.6495,
      "step": 19525
    },
    {
      "epoch": 0.8092337021840939,
      "grad_norm": 0.43710458278656006,
      "learning_rate": 9.540387086078994e-07,
      "loss": 0.6724,
      "step": 19526
    },
    {
      "epoch": 0.8092751460897675,
      "grad_norm": 0.4097225069999695,
      "learning_rate": 9.53831489079531e-07,
      "loss": 0.6671,
      "step": 19527
    },
    {
      "epoch": 0.8093165899954412,
      "grad_norm": 0.43732449412345886,
      "learning_rate": 9.536242695511626e-07,
      "loss": 0.7074,
      "step": 19528
    },
    {
      "epoch": 0.8093580339011148,
      "grad_norm": 0.4084685444831848,
      "learning_rate": 9.534170500227941e-07,
      "loss": 0.6787,
      "step": 19529
    },
    {
      "epoch": 0.8093994778067886,
      "grad_norm": 0.44383955001831055,
      "learning_rate": 9.532098304944259e-07,
      "loss": 0.6936,
      "step": 19530
    },
    {
      "epoch": 0.8094409217124622,
      "grad_norm": 0.37607231736183167,
      "learning_rate": 9.530026109660576e-07,
      "loss": 0.5717,
      "step": 19531
    },
    {
      "epoch": 0.8094823656181358,
      "grad_norm": 0.4168631136417389,
      "learning_rate": 9.527953914376892e-07,
      "loss": 0.6262,
      "step": 19532
    },
    {
      "epoch": 0.8095238095238095,
      "grad_norm": 0.3866446912288666,
      "learning_rate": 9.525881719093208e-07,
      "loss": 0.5846,
      "step": 19533
    },
    {
      "epoch": 0.8095652534294832,
      "grad_norm": 0.44628483057022095,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.6749,
      "step": 19534
    },
    {
      "epoch": 0.8096066973351569,
      "grad_norm": 0.38900160789489746,
      "learning_rate": 9.521737328525842e-07,
      "loss": 0.632,
      "step": 19535
    },
    {
      "epoch": 0.8096481412408305,
      "grad_norm": 0.45896750688552856,
      "learning_rate": 9.519665133242158e-07,
      "loss": 0.6853,
      "step": 19536
    },
    {
      "epoch": 0.8096895851465042,
      "grad_norm": 0.44777947664260864,
      "learning_rate": 9.517592937958474e-07,
      "loss": 0.6479,
      "step": 19537
    },
    {
      "epoch": 0.8097310290521779,
      "grad_norm": 0.4207313358783722,
      "learning_rate": 9.515520742674791e-07,
      "loss": 0.6604,
      "step": 19538
    },
    {
      "epoch": 0.8097724729578516,
      "grad_norm": 0.42841997742652893,
      "learning_rate": 9.513448547391108e-07,
      "loss": 0.6726,
      "step": 19539
    },
    {
      "epoch": 0.8098139168635252,
      "grad_norm": 0.40120217204093933,
      "learning_rate": 9.511376352107424e-07,
      "loss": 0.6519,
      "step": 19540
    },
    {
      "epoch": 0.8098553607691988,
      "grad_norm": 0.4038342535495758,
      "learning_rate": 9.50930415682374e-07,
      "loss": 0.6503,
      "step": 19541
    },
    {
      "epoch": 0.8098968046748726,
      "grad_norm": 0.4423257112503052,
      "learning_rate": 9.507231961540056e-07,
      "loss": 0.72,
      "step": 19542
    },
    {
      "epoch": 0.8099382485805462,
      "grad_norm": 0.4006195068359375,
      "learning_rate": 9.505159766256373e-07,
      "loss": 0.6469,
      "step": 19543
    },
    {
      "epoch": 0.8099796924862199,
      "grad_norm": 0.4442633390426636,
      "learning_rate": 9.50308757097269e-07,
      "loss": 0.6663,
      "step": 19544
    },
    {
      "epoch": 0.8100211363918935,
      "grad_norm": 0.425112247467041,
      "learning_rate": 9.501015375689006e-07,
      "loss": 0.7031,
      "step": 19545
    },
    {
      "epoch": 0.8100625802975673,
      "grad_norm": 0.3796014189720154,
      "learning_rate": 9.498943180405322e-07,
      "loss": 0.668,
      "step": 19546
    },
    {
      "epoch": 0.8101040242032409,
      "grad_norm": 0.4757033586502075,
      "learning_rate": 9.496870985121639e-07,
      "loss": 0.748,
      "step": 19547
    },
    {
      "epoch": 0.8101454681089146,
      "grad_norm": 0.4946349263191223,
      "learning_rate": 9.494798789837956e-07,
      "loss": 0.6772,
      "step": 19548
    },
    {
      "epoch": 0.8101869120145883,
      "grad_norm": 0.4156789779663086,
      "learning_rate": 9.492726594554272e-07,
      "loss": 0.6943,
      "step": 19549
    },
    {
      "epoch": 0.8102283559202619,
      "grad_norm": 0.48004481196403503,
      "learning_rate": 9.490654399270588e-07,
      "loss": 0.73,
      "step": 19550
    },
    {
      "epoch": 0.8102697998259356,
      "grad_norm": 0.4179478883743286,
      "learning_rate": 9.488582203986905e-07,
      "loss": 0.6398,
      "step": 19551
    },
    {
      "epoch": 0.8103112437316092,
      "grad_norm": 0.4231070578098297,
      "learning_rate": 9.486510008703222e-07,
      "loss": 0.6439,
      "step": 19552
    },
    {
      "epoch": 0.810352687637283,
      "grad_norm": 0.440898060798645,
      "learning_rate": 9.484437813419538e-07,
      "loss": 0.6995,
      "step": 19553
    },
    {
      "epoch": 0.8103941315429566,
      "grad_norm": 0.39904752373695374,
      "learning_rate": 9.482365618135854e-07,
      "loss": 0.5946,
      "step": 19554
    },
    {
      "epoch": 0.8104355754486303,
      "grad_norm": 0.4150044620037079,
      "learning_rate": 9.480293422852171e-07,
      "loss": 0.6434,
      "step": 19555
    },
    {
      "epoch": 0.8104770193543039,
      "grad_norm": 0.41956767439842224,
      "learning_rate": 9.478221227568487e-07,
      "loss": 0.6498,
      "step": 19556
    },
    {
      "epoch": 0.8105184632599777,
      "grad_norm": 0.4612277150154114,
      "learning_rate": 9.476149032284804e-07,
      "loss": 0.6674,
      "step": 19557
    },
    {
      "epoch": 0.8105599071656513,
      "grad_norm": 0.4166176915168762,
      "learning_rate": 9.47407683700112e-07,
      "loss": 0.6538,
      "step": 19558
    },
    {
      "epoch": 0.8106013510713249,
      "grad_norm": 0.37055960297584534,
      "learning_rate": 9.472004641717436e-07,
      "loss": 0.5989,
      "step": 19559
    },
    {
      "epoch": 0.8106427949769986,
      "grad_norm": 0.3949413001537323,
      "learning_rate": 9.469932446433753e-07,
      "loss": 0.6338,
      "step": 19560
    },
    {
      "epoch": 0.8106842388826723,
      "grad_norm": 0.446988046169281,
      "learning_rate": 9.46786025115007e-07,
      "loss": 0.741,
      "step": 19561
    },
    {
      "epoch": 0.810725682788346,
      "grad_norm": 0.4303412437438965,
      "learning_rate": 9.465788055866386e-07,
      "loss": 0.6472,
      "step": 19562
    },
    {
      "epoch": 0.8107671266940196,
      "grad_norm": 0.41025108098983765,
      "learning_rate": 9.463715860582702e-07,
      "loss": 0.6406,
      "step": 19563
    },
    {
      "epoch": 0.8108085705996934,
      "grad_norm": 0.3790895342826843,
      "learning_rate": 9.461643665299019e-07,
      "loss": 0.6412,
      "step": 19564
    },
    {
      "epoch": 0.810850014505367,
      "grad_norm": 0.3985369801521301,
      "learning_rate": 9.459571470015336e-07,
      "loss": 0.6382,
      "step": 19565
    },
    {
      "epoch": 0.8108914584110407,
      "grad_norm": 0.44402050971984863,
      "learning_rate": 9.457499274731652e-07,
      "loss": 0.7021,
      "step": 19566
    },
    {
      "epoch": 0.8109329023167143,
      "grad_norm": 0.40355411171913147,
      "learning_rate": 9.455427079447968e-07,
      "loss": 0.6852,
      "step": 19567
    },
    {
      "epoch": 0.810974346222388,
      "grad_norm": 0.4874855577945709,
      "learning_rate": 9.453354884164285e-07,
      "loss": 0.6508,
      "step": 19568
    },
    {
      "epoch": 0.8110157901280617,
      "grad_norm": 0.41878223419189453,
      "learning_rate": 9.451282688880601e-07,
      "loss": 0.6375,
      "step": 19569
    },
    {
      "epoch": 0.8110572340337353,
      "grad_norm": 0.4299320578575134,
      "learning_rate": 9.449210493596918e-07,
      "loss": 0.679,
      "step": 19570
    },
    {
      "epoch": 0.811098677939409,
      "grad_norm": 0.3985569477081299,
      "learning_rate": 9.447138298313234e-07,
      "loss": 0.6492,
      "step": 19571
    },
    {
      "epoch": 0.8111401218450827,
      "grad_norm": 0.4231337010860443,
      "learning_rate": 9.445066103029551e-07,
      "loss": 0.6569,
      "step": 19572
    },
    {
      "epoch": 0.8111815657507564,
      "grad_norm": 0.4230033755302429,
      "learning_rate": 9.442993907745867e-07,
      "loss": 0.6664,
      "step": 19573
    },
    {
      "epoch": 0.81122300965643,
      "grad_norm": 0.40938642621040344,
      "learning_rate": 9.440921712462184e-07,
      "loss": 0.6774,
      "step": 19574
    },
    {
      "epoch": 0.8112644535621036,
      "grad_norm": 0.4231196641921997,
      "learning_rate": 9.4388495171785e-07,
      "loss": 0.6945,
      "step": 19575
    },
    {
      "epoch": 0.8113058974677774,
      "grad_norm": 0.4148619472980499,
      "learning_rate": 9.436777321894817e-07,
      "loss": 0.6716,
      "step": 19576
    },
    {
      "epoch": 0.811347341373451,
      "grad_norm": 0.43137112259864807,
      "learning_rate": 9.434705126611133e-07,
      "loss": 0.6578,
      "step": 19577
    },
    {
      "epoch": 0.8113887852791247,
      "grad_norm": 0.3822075426578522,
      "learning_rate": 9.432632931327449e-07,
      "loss": 0.6266,
      "step": 19578
    },
    {
      "epoch": 0.8114302291847983,
      "grad_norm": 0.40512531995773315,
      "learning_rate": 9.430560736043766e-07,
      "loss": 0.6707,
      "step": 19579
    },
    {
      "epoch": 0.8114716730904721,
      "grad_norm": 0.4204866290092468,
      "learning_rate": 9.428488540760082e-07,
      "loss": 0.6171,
      "step": 19580
    },
    {
      "epoch": 0.8115131169961457,
      "grad_norm": 0.4203627407550812,
      "learning_rate": 9.426416345476399e-07,
      "loss": 0.666,
      "step": 19581
    },
    {
      "epoch": 0.8115545609018194,
      "grad_norm": 0.40756741166114807,
      "learning_rate": 9.424344150192715e-07,
      "loss": 0.6772,
      "step": 19582
    },
    {
      "epoch": 0.811596004807493,
      "grad_norm": 0.3999824523925781,
      "learning_rate": 9.422271954909032e-07,
      "loss": 0.5991,
      "step": 19583
    },
    {
      "epoch": 0.8116374487131667,
      "grad_norm": 0.3941100239753723,
      "learning_rate": 9.420199759625348e-07,
      "loss": 0.6417,
      "step": 19584
    },
    {
      "epoch": 0.8116788926188404,
      "grad_norm": 0.4361357092857361,
      "learning_rate": 9.418127564341665e-07,
      "loss": 0.6719,
      "step": 19585
    },
    {
      "epoch": 0.811720336524514,
      "grad_norm": 0.4035375714302063,
      "learning_rate": 9.416055369057981e-07,
      "loss": 0.6096,
      "step": 19586
    },
    {
      "epoch": 0.8117617804301878,
      "grad_norm": 0.4277745187282562,
      "learning_rate": 9.413983173774298e-07,
      "loss": 0.7708,
      "step": 19587
    },
    {
      "epoch": 0.8118032243358614,
      "grad_norm": 0.40764865279197693,
      "learning_rate": 9.411910978490614e-07,
      "loss": 0.7007,
      "step": 19588
    },
    {
      "epoch": 0.8118446682415351,
      "grad_norm": 0.40538203716278076,
      "learning_rate": 9.409838783206931e-07,
      "loss": 0.6682,
      "step": 19589
    },
    {
      "epoch": 0.8118861121472087,
      "grad_norm": 0.4215800166130066,
      "learning_rate": 9.407766587923247e-07,
      "loss": 0.6781,
      "step": 19590
    },
    {
      "epoch": 0.8119275560528825,
      "grad_norm": 0.4068776071071625,
      "learning_rate": 9.405694392639563e-07,
      "loss": 0.7058,
      "step": 19591
    },
    {
      "epoch": 0.8119689999585561,
      "grad_norm": 0.39818522334098816,
      "learning_rate": 9.40362219735588e-07,
      "loss": 0.6674,
      "step": 19592
    },
    {
      "epoch": 0.8120104438642297,
      "grad_norm": 0.4001738429069519,
      "learning_rate": 9.401550002072197e-07,
      "loss": 0.6249,
      "step": 19593
    },
    {
      "epoch": 0.8120518877699034,
      "grad_norm": 0.4007297158241272,
      "learning_rate": 9.399477806788513e-07,
      "loss": 0.675,
      "step": 19594
    },
    {
      "epoch": 0.8120933316755771,
      "grad_norm": 0.38033580780029297,
      "learning_rate": 9.397405611504829e-07,
      "loss": 0.665,
      "step": 19595
    },
    {
      "epoch": 0.8121347755812508,
      "grad_norm": 0.4213559031486511,
      "learning_rate": 9.395333416221146e-07,
      "loss": 0.7086,
      "step": 19596
    },
    {
      "epoch": 0.8121762194869244,
      "grad_norm": 0.5391089916229248,
      "learning_rate": 9.393261220937462e-07,
      "loss": 0.7217,
      "step": 19597
    },
    {
      "epoch": 0.8122176633925982,
      "grad_norm": 0.3782764971256256,
      "learning_rate": 9.391189025653779e-07,
      "loss": 0.6614,
      "step": 19598
    },
    {
      "epoch": 0.8122591072982718,
      "grad_norm": 0.4088730216026306,
      "learning_rate": 9.389116830370095e-07,
      "loss": 0.6853,
      "step": 19599
    },
    {
      "epoch": 0.8123005512039455,
      "grad_norm": 0.42250147461891174,
      "learning_rate": 9.387044635086412e-07,
      "loss": 0.6583,
      "step": 19600
    },
    {
      "epoch": 0.8123419951096191,
      "grad_norm": 0.40646958351135254,
      "learning_rate": 9.384972439802728e-07,
      "loss": 0.6567,
      "step": 19601
    },
    {
      "epoch": 0.8123834390152928,
      "grad_norm": 0.40952181816101074,
      "learning_rate": 9.382900244519045e-07,
      "loss": 0.6521,
      "step": 19602
    },
    {
      "epoch": 0.8124248829209665,
      "grad_norm": 0.40526333451271057,
      "learning_rate": 9.380828049235361e-07,
      "loss": 0.6427,
      "step": 19603
    },
    {
      "epoch": 0.8124663268266401,
      "grad_norm": 0.41660037636756897,
      "learning_rate": 9.378755853951677e-07,
      "loss": 0.6273,
      "step": 19604
    },
    {
      "epoch": 0.8125077707323138,
      "grad_norm": 0.4488762319087982,
      "learning_rate": 9.376683658667994e-07,
      "loss": 0.6887,
      "step": 19605
    },
    {
      "epoch": 0.8125492146379875,
      "grad_norm": 0.39120998978614807,
      "learning_rate": 9.374611463384311e-07,
      "loss": 0.6421,
      "step": 19606
    },
    {
      "epoch": 0.8125906585436612,
      "grad_norm": 0.4163181781768799,
      "learning_rate": 9.372539268100627e-07,
      "loss": 0.639,
      "step": 19607
    },
    {
      "epoch": 0.8126321024493348,
      "grad_norm": 0.38665008544921875,
      "learning_rate": 9.370467072816943e-07,
      "loss": 0.6661,
      "step": 19608
    },
    {
      "epoch": 0.8126735463550085,
      "grad_norm": 0.4687749743461609,
      "learning_rate": 9.36839487753326e-07,
      "loss": 0.6779,
      "step": 19609
    },
    {
      "epoch": 0.8127149902606822,
      "grad_norm": 0.39361050724983215,
      "learning_rate": 9.366322682249577e-07,
      "loss": 0.6696,
      "step": 19610
    },
    {
      "epoch": 0.8127564341663558,
      "grad_norm": 0.4061894118785858,
      "learning_rate": 9.364250486965893e-07,
      "loss": 0.6671,
      "step": 19611
    },
    {
      "epoch": 0.8127978780720295,
      "grad_norm": 0.4217144250869751,
      "learning_rate": 9.362178291682209e-07,
      "loss": 0.7,
      "step": 19612
    },
    {
      "epoch": 0.8128393219777031,
      "grad_norm": 0.42487600445747375,
      "learning_rate": 9.360106096398526e-07,
      "loss": 0.6506,
      "step": 19613
    },
    {
      "epoch": 0.8128807658833769,
      "grad_norm": 0.38957899808883667,
      "learning_rate": 9.358033901114842e-07,
      "loss": 0.5918,
      "step": 19614
    },
    {
      "epoch": 0.8129222097890505,
      "grad_norm": 0.4197501838207245,
      "learning_rate": 9.355961705831159e-07,
      "loss": 0.6816,
      "step": 19615
    },
    {
      "epoch": 0.8129636536947242,
      "grad_norm": 0.42846062779426575,
      "learning_rate": 9.353889510547475e-07,
      "loss": 0.6448,
      "step": 19616
    },
    {
      "epoch": 0.8130050976003979,
      "grad_norm": 0.4134286642074585,
      "learning_rate": 9.351817315263791e-07,
      "loss": 0.6396,
      "step": 19617
    },
    {
      "epoch": 0.8130465415060716,
      "grad_norm": 0.43286341428756714,
      "learning_rate": 9.349745119980108e-07,
      "loss": 0.7164,
      "step": 19618
    },
    {
      "epoch": 0.8130879854117452,
      "grad_norm": 0.38709139823913574,
      "learning_rate": 9.347672924696425e-07,
      "loss": 0.6277,
      "step": 19619
    },
    {
      "epoch": 0.8131294293174188,
      "grad_norm": 0.41580572724342346,
      "learning_rate": 9.345600729412741e-07,
      "loss": 0.6628,
      "step": 19620
    },
    {
      "epoch": 0.8131708732230926,
      "grad_norm": 0.4190933406352997,
      "learning_rate": 9.343528534129057e-07,
      "loss": 0.6914,
      "step": 19621
    },
    {
      "epoch": 0.8132123171287662,
      "grad_norm": 0.416583776473999,
      "learning_rate": 9.341456338845374e-07,
      "loss": 0.6589,
      "step": 19622
    },
    {
      "epoch": 0.8132537610344399,
      "grad_norm": 0.47658082842826843,
      "learning_rate": 9.339384143561691e-07,
      "loss": 0.6763,
      "step": 19623
    },
    {
      "epoch": 0.8132952049401135,
      "grad_norm": 0.4048510491847992,
      "learning_rate": 9.337311948278007e-07,
      "loss": 0.7461,
      "step": 19624
    },
    {
      "epoch": 0.8133366488457873,
      "grad_norm": 0.40592893958091736,
      "learning_rate": 9.335239752994323e-07,
      "loss": 0.6689,
      "step": 19625
    },
    {
      "epoch": 0.8133780927514609,
      "grad_norm": 0.443429559469223,
      "learning_rate": 9.33316755771064e-07,
      "loss": 0.6906,
      "step": 19626
    },
    {
      "epoch": 0.8134195366571346,
      "grad_norm": 0.41236671805381775,
      "learning_rate": 9.331095362426957e-07,
      "loss": 0.6802,
      "step": 19627
    },
    {
      "epoch": 0.8134609805628082,
      "grad_norm": 0.4343055784702301,
      "learning_rate": 9.329023167143273e-07,
      "loss": 0.6372,
      "step": 19628
    },
    {
      "epoch": 0.8135024244684819,
      "grad_norm": 0.4300934374332428,
      "learning_rate": 9.326950971859589e-07,
      "loss": 0.7224,
      "step": 19629
    },
    {
      "epoch": 0.8135438683741556,
      "grad_norm": 0.42721492052078247,
      "learning_rate": 9.324878776575905e-07,
      "loss": 0.6458,
      "step": 19630
    },
    {
      "epoch": 0.8135853122798292,
      "grad_norm": 0.410971462726593,
      "learning_rate": 9.322806581292223e-07,
      "loss": 0.677,
      "step": 19631
    },
    {
      "epoch": 0.813626756185503,
      "grad_norm": 0.40581756830215454,
      "learning_rate": 9.320734386008539e-07,
      "loss": 0.6501,
      "step": 19632
    },
    {
      "epoch": 0.8136682000911766,
      "grad_norm": 0.41823723912239075,
      "learning_rate": 9.318662190724855e-07,
      "loss": 0.692,
      "step": 19633
    },
    {
      "epoch": 0.8137096439968503,
      "grad_norm": 0.39229342341423035,
      "learning_rate": 9.316589995441171e-07,
      "loss": 0.639,
      "step": 19634
    },
    {
      "epoch": 0.8137510879025239,
      "grad_norm": 0.4431338608264923,
      "learning_rate": 9.314517800157488e-07,
      "loss": 0.6932,
      "step": 19635
    },
    {
      "epoch": 0.8137925318081976,
      "grad_norm": 0.4214947819709778,
      "learning_rate": 9.312445604873805e-07,
      "loss": 0.6934,
      "step": 19636
    },
    {
      "epoch": 0.8138339757138713,
      "grad_norm": 0.42291656136512756,
      "learning_rate": 9.310373409590121e-07,
      "loss": 0.6956,
      "step": 19637
    },
    {
      "epoch": 0.8138754196195449,
      "grad_norm": 0.44587475061416626,
      "learning_rate": 9.308301214306437e-07,
      "loss": 0.7249,
      "step": 19638
    },
    {
      "epoch": 0.8139168635252186,
      "grad_norm": 0.40491124987602234,
      "learning_rate": 9.306229019022753e-07,
      "loss": 0.6628,
      "step": 19639
    },
    {
      "epoch": 0.8139583074308923,
      "grad_norm": 0.40952345728874207,
      "learning_rate": 9.304156823739071e-07,
      "loss": 0.6674,
      "step": 19640
    },
    {
      "epoch": 0.813999751336566,
      "grad_norm": 0.43701407313346863,
      "learning_rate": 9.302084628455387e-07,
      "loss": 0.6532,
      "step": 19641
    },
    {
      "epoch": 0.8140411952422396,
      "grad_norm": 0.3995969295501709,
      "learning_rate": 9.300012433171703e-07,
      "loss": 0.6801,
      "step": 19642
    },
    {
      "epoch": 0.8140826391479133,
      "grad_norm": 0.40720489621162415,
      "learning_rate": 9.297940237888019e-07,
      "loss": 0.6689,
      "step": 19643
    },
    {
      "epoch": 0.814124083053587,
      "grad_norm": 0.40332961082458496,
      "learning_rate": 9.295868042604337e-07,
      "loss": 0.6865,
      "step": 19644
    },
    {
      "epoch": 0.8141655269592606,
      "grad_norm": 0.3916777968406677,
      "learning_rate": 9.293795847320653e-07,
      "loss": 0.6775,
      "step": 19645
    },
    {
      "epoch": 0.8142069708649343,
      "grad_norm": 0.40694382786750793,
      "learning_rate": 9.291723652036969e-07,
      "loss": 0.6725,
      "step": 19646
    },
    {
      "epoch": 0.8142484147706079,
      "grad_norm": 0.42133644223213196,
      "learning_rate": 9.289651456753285e-07,
      "loss": 0.7185,
      "step": 19647
    },
    {
      "epoch": 0.8142898586762817,
      "grad_norm": 0.4047718346118927,
      "learning_rate": 9.287579261469603e-07,
      "loss": 0.6542,
      "step": 19648
    },
    {
      "epoch": 0.8143313025819553,
      "grad_norm": 0.4477708637714386,
      "learning_rate": 9.285507066185919e-07,
      "loss": 0.6804,
      "step": 19649
    },
    {
      "epoch": 0.814372746487629,
      "grad_norm": 0.4244292676448822,
      "learning_rate": 9.283434870902235e-07,
      "loss": 0.6427,
      "step": 19650
    },
    {
      "epoch": 0.8144141903933027,
      "grad_norm": 0.4160427749156952,
      "learning_rate": 9.281362675618551e-07,
      "loss": 0.7073,
      "step": 19651
    },
    {
      "epoch": 0.8144556342989764,
      "grad_norm": 0.4110974967479706,
      "learning_rate": 9.279290480334867e-07,
      "loss": 0.6566,
      "step": 19652
    },
    {
      "epoch": 0.81449707820465,
      "grad_norm": 0.4017619788646698,
      "learning_rate": 9.277218285051185e-07,
      "loss": 0.6842,
      "step": 19653
    },
    {
      "epoch": 0.8145385221103236,
      "grad_norm": 0.41221749782562256,
      "learning_rate": 9.275146089767501e-07,
      "loss": 0.5894,
      "step": 19654
    },
    {
      "epoch": 0.8145799660159974,
      "grad_norm": 0.4414932429790497,
      "learning_rate": 9.273073894483817e-07,
      "loss": 0.7131,
      "step": 19655
    },
    {
      "epoch": 0.814621409921671,
      "grad_norm": 0.41789546608924866,
      "learning_rate": 9.271001699200133e-07,
      "loss": 0.6713,
      "step": 19656
    },
    {
      "epoch": 0.8146628538273447,
      "grad_norm": 0.4307880103588104,
      "learning_rate": 9.268929503916451e-07,
      "loss": 0.6843,
      "step": 19657
    },
    {
      "epoch": 0.8147042977330183,
      "grad_norm": 0.42938876152038574,
      "learning_rate": 9.266857308632767e-07,
      "loss": 0.7581,
      "step": 19658
    },
    {
      "epoch": 0.8147457416386921,
      "grad_norm": 0.4760993421077728,
      "learning_rate": 9.264785113349083e-07,
      "loss": 0.7705,
      "step": 19659
    },
    {
      "epoch": 0.8147871855443657,
      "grad_norm": 0.40461939573287964,
      "learning_rate": 9.262712918065399e-07,
      "loss": 0.6443,
      "step": 19660
    },
    {
      "epoch": 0.8148286294500394,
      "grad_norm": 0.394463449716568,
      "learning_rate": 9.260640722781717e-07,
      "loss": 0.6337,
      "step": 19661
    },
    {
      "epoch": 0.814870073355713,
      "grad_norm": 0.41641008853912354,
      "learning_rate": 9.258568527498033e-07,
      "loss": 0.6821,
      "step": 19662
    },
    {
      "epoch": 0.8149115172613867,
      "grad_norm": 0.4241864085197449,
      "learning_rate": 9.256496332214349e-07,
      "loss": 0.6776,
      "step": 19663
    },
    {
      "epoch": 0.8149529611670604,
      "grad_norm": 0.404107004404068,
      "learning_rate": 9.254424136930665e-07,
      "loss": 0.6675,
      "step": 19664
    },
    {
      "epoch": 0.814994405072734,
      "grad_norm": 0.45130255818367004,
      "learning_rate": 9.252351941646981e-07,
      "loss": 0.7126,
      "step": 19665
    },
    {
      "epoch": 0.8150358489784078,
      "grad_norm": 0.43848392367362976,
      "learning_rate": 9.250279746363299e-07,
      "loss": 0.7046,
      "step": 19666
    },
    {
      "epoch": 0.8150772928840814,
      "grad_norm": 0.41398677229881287,
      "learning_rate": 9.248207551079615e-07,
      "loss": 0.6672,
      "step": 19667
    },
    {
      "epoch": 0.8151187367897551,
      "grad_norm": 0.43072229623794556,
      "learning_rate": 9.246135355795931e-07,
      "loss": 0.6445,
      "step": 19668
    },
    {
      "epoch": 0.8151601806954287,
      "grad_norm": 0.4125159978866577,
      "learning_rate": 9.244063160512247e-07,
      "loss": 0.5852,
      "step": 19669
    },
    {
      "epoch": 0.8152016246011025,
      "grad_norm": 0.42327719926834106,
      "learning_rate": 9.241990965228565e-07,
      "loss": 0.6549,
      "step": 19670
    },
    {
      "epoch": 0.8152430685067761,
      "grad_norm": 0.4482043981552124,
      "learning_rate": 9.239918769944881e-07,
      "loss": 0.6667,
      "step": 19671
    },
    {
      "epoch": 0.8152845124124497,
      "grad_norm": 0.41444191336631775,
      "learning_rate": 9.237846574661197e-07,
      "loss": 0.6251,
      "step": 19672
    },
    {
      "epoch": 0.8153259563181234,
      "grad_norm": 0.39755353331565857,
      "learning_rate": 9.235774379377513e-07,
      "loss": 0.6127,
      "step": 19673
    },
    {
      "epoch": 0.8153674002237971,
      "grad_norm": 0.43398159742355347,
      "learning_rate": 9.233702184093831e-07,
      "loss": 0.6379,
      "step": 19674
    },
    {
      "epoch": 0.8154088441294708,
      "grad_norm": 0.4340681731700897,
      "learning_rate": 9.231629988810147e-07,
      "loss": 0.6741,
      "step": 19675
    },
    {
      "epoch": 0.8154502880351444,
      "grad_norm": 0.42247021198272705,
      "learning_rate": 9.229557793526463e-07,
      "loss": 0.7024,
      "step": 19676
    },
    {
      "epoch": 0.8154917319408181,
      "grad_norm": 0.44464629888534546,
      "learning_rate": 9.227485598242779e-07,
      "loss": 0.6719,
      "step": 19677
    },
    {
      "epoch": 0.8155331758464918,
      "grad_norm": 0.408925861120224,
      "learning_rate": 9.225413402959095e-07,
      "loss": 0.6926,
      "step": 19678
    },
    {
      "epoch": 0.8155746197521655,
      "grad_norm": 0.40343138575553894,
      "learning_rate": 9.223341207675413e-07,
      "loss": 0.6951,
      "step": 19679
    },
    {
      "epoch": 0.8156160636578391,
      "grad_norm": 0.42007914185523987,
      "learning_rate": 9.221269012391729e-07,
      "loss": 0.7026,
      "step": 19680
    },
    {
      "epoch": 0.8156575075635127,
      "grad_norm": 0.39540037512779236,
      "learning_rate": 9.219196817108045e-07,
      "loss": 0.6033,
      "step": 19681
    },
    {
      "epoch": 0.8156989514691865,
      "grad_norm": 0.4160844683647156,
      "learning_rate": 9.217124621824361e-07,
      "loss": 0.6077,
      "step": 19682
    },
    {
      "epoch": 0.8157403953748601,
      "grad_norm": 0.39774826169013977,
      "learning_rate": 9.215052426540679e-07,
      "loss": 0.6387,
      "step": 19683
    },
    {
      "epoch": 0.8157818392805338,
      "grad_norm": 0.41695356369018555,
      "learning_rate": 9.212980231256995e-07,
      "loss": 0.7037,
      "step": 19684
    },
    {
      "epoch": 0.8158232831862074,
      "grad_norm": 0.4222720265388489,
      "learning_rate": 9.210908035973311e-07,
      "loss": 0.6635,
      "step": 19685
    },
    {
      "epoch": 0.8158647270918812,
      "grad_norm": 0.41020119190216064,
      "learning_rate": 9.208835840689627e-07,
      "loss": 0.6782,
      "step": 19686
    },
    {
      "epoch": 0.8159061709975548,
      "grad_norm": 0.4394177496433258,
      "learning_rate": 9.206763645405945e-07,
      "loss": 0.6615,
      "step": 19687
    },
    {
      "epoch": 0.8159476149032285,
      "grad_norm": 0.4219087064266205,
      "learning_rate": 9.204691450122261e-07,
      "loss": 0.6642,
      "step": 19688
    },
    {
      "epoch": 0.8159890588089022,
      "grad_norm": 0.3978648781776428,
      "learning_rate": 9.202619254838577e-07,
      "loss": 0.6373,
      "step": 19689
    },
    {
      "epoch": 0.8160305027145758,
      "grad_norm": 0.4171396493911743,
      "learning_rate": 9.200547059554893e-07,
      "loss": 0.6674,
      "step": 19690
    },
    {
      "epoch": 0.8160719466202495,
      "grad_norm": 0.4340713918209076,
      "learning_rate": 9.198474864271209e-07,
      "loss": 0.7319,
      "step": 19691
    },
    {
      "epoch": 0.8161133905259231,
      "grad_norm": 0.38078805804252625,
      "learning_rate": 9.196402668987527e-07,
      "loss": 0.6572,
      "step": 19692
    },
    {
      "epoch": 0.8161548344315969,
      "grad_norm": 0.3820040822029114,
      "learning_rate": 9.194330473703843e-07,
      "loss": 0.6208,
      "step": 19693
    },
    {
      "epoch": 0.8161962783372705,
      "grad_norm": 0.39858493208885193,
      "learning_rate": 9.192258278420159e-07,
      "loss": 0.6191,
      "step": 19694
    },
    {
      "epoch": 0.8162377222429442,
      "grad_norm": 0.4075811207294464,
      "learning_rate": 9.190186083136475e-07,
      "loss": 0.6375,
      "step": 19695
    },
    {
      "epoch": 0.8162791661486178,
      "grad_norm": 0.38667964935302734,
      "learning_rate": 9.188113887852793e-07,
      "loss": 0.6506,
      "step": 19696
    },
    {
      "epoch": 0.8163206100542915,
      "grad_norm": 0.4880538284778595,
      "learning_rate": 9.186041692569109e-07,
      "loss": 0.6898,
      "step": 19697
    },
    {
      "epoch": 0.8163620539599652,
      "grad_norm": 0.3847583830356598,
      "learning_rate": 9.183969497285425e-07,
      "loss": 0.6809,
      "step": 19698
    },
    {
      "epoch": 0.8164034978656388,
      "grad_norm": 0.4298837184906006,
      "learning_rate": 9.181897302001741e-07,
      "loss": 0.7065,
      "step": 19699
    },
    {
      "epoch": 0.8164449417713125,
      "grad_norm": 0.427444189786911,
      "learning_rate": 9.179825106718059e-07,
      "loss": 0.7097,
      "step": 19700
    },
    {
      "epoch": 0.8164863856769862,
      "grad_norm": 0.3810659646987915,
      "learning_rate": 9.177752911434375e-07,
      "loss": 0.6403,
      "step": 19701
    },
    {
      "epoch": 0.8165278295826599,
      "grad_norm": 0.38859644532203674,
      "learning_rate": 9.175680716150691e-07,
      "loss": 0.6388,
      "step": 19702
    },
    {
      "epoch": 0.8165692734883335,
      "grad_norm": 0.43492475152015686,
      "learning_rate": 9.173608520867007e-07,
      "loss": 0.6991,
      "step": 19703
    },
    {
      "epoch": 0.8166107173940073,
      "grad_norm": 0.3958321511745453,
      "learning_rate": 9.171536325583323e-07,
      "loss": 0.6471,
      "step": 19704
    },
    {
      "epoch": 0.8166521612996809,
      "grad_norm": 0.4117456078529358,
      "learning_rate": 9.169464130299641e-07,
      "loss": 0.6183,
      "step": 19705
    },
    {
      "epoch": 0.8166936052053545,
      "grad_norm": 0.425936758518219,
      "learning_rate": 9.167391935015957e-07,
      "loss": 0.6605,
      "step": 19706
    },
    {
      "epoch": 0.8167350491110282,
      "grad_norm": 0.4100085496902466,
      "learning_rate": 9.165319739732273e-07,
      "loss": 0.6136,
      "step": 19707
    },
    {
      "epoch": 0.8167764930167019,
      "grad_norm": 0.404861718416214,
      "learning_rate": 9.163247544448589e-07,
      "loss": 0.667,
      "step": 19708
    },
    {
      "epoch": 0.8168179369223756,
      "grad_norm": 0.41336488723754883,
      "learning_rate": 9.161175349164907e-07,
      "loss": 0.6726,
      "step": 19709
    },
    {
      "epoch": 0.8168593808280492,
      "grad_norm": 0.4177948832511902,
      "learning_rate": 9.159103153881223e-07,
      "loss": 0.6788,
      "step": 19710
    },
    {
      "epoch": 0.8169008247337229,
      "grad_norm": 0.4212389588356018,
      "learning_rate": 9.157030958597539e-07,
      "loss": 0.6902,
      "step": 19711
    },
    {
      "epoch": 0.8169422686393966,
      "grad_norm": 0.42884549498558044,
      "learning_rate": 9.154958763313855e-07,
      "loss": 0.675,
      "step": 19712
    },
    {
      "epoch": 0.8169837125450703,
      "grad_norm": 0.4176728427410126,
      "learning_rate": 9.152886568030171e-07,
      "loss": 0.6534,
      "step": 19713
    },
    {
      "epoch": 0.8170251564507439,
      "grad_norm": 0.42101776599884033,
      "learning_rate": 9.150814372746489e-07,
      "loss": 0.7009,
      "step": 19714
    },
    {
      "epoch": 0.8170666003564175,
      "grad_norm": 0.4365403652191162,
      "learning_rate": 9.148742177462805e-07,
      "loss": 0.694,
      "step": 19715
    },
    {
      "epoch": 0.8171080442620913,
      "grad_norm": 0.423768550157547,
      "learning_rate": 9.146669982179121e-07,
      "loss": 0.6965,
      "step": 19716
    },
    {
      "epoch": 0.8171494881677649,
      "grad_norm": 0.44997668266296387,
      "learning_rate": 9.144597786895437e-07,
      "loss": 0.6615,
      "step": 19717
    },
    {
      "epoch": 0.8171909320734386,
      "grad_norm": 0.4319949150085449,
      "learning_rate": 9.142525591611755e-07,
      "loss": 0.7166,
      "step": 19718
    },
    {
      "epoch": 0.8172323759791122,
      "grad_norm": 0.40280041098594666,
      "learning_rate": 9.140453396328071e-07,
      "loss": 0.6584,
      "step": 19719
    },
    {
      "epoch": 0.817273819884786,
      "grad_norm": 0.40576761960983276,
      "learning_rate": 9.138381201044387e-07,
      "loss": 0.693,
      "step": 19720
    },
    {
      "epoch": 0.8173152637904596,
      "grad_norm": 0.3668539524078369,
      "learning_rate": 9.136309005760703e-07,
      "loss": 0.6354,
      "step": 19721
    },
    {
      "epoch": 0.8173567076961333,
      "grad_norm": 0.4559946358203888,
      "learning_rate": 9.134236810477021e-07,
      "loss": 0.7173,
      "step": 19722
    },
    {
      "epoch": 0.817398151601807,
      "grad_norm": 0.44028905034065247,
      "learning_rate": 9.132164615193337e-07,
      "loss": 0.6758,
      "step": 19723
    },
    {
      "epoch": 0.8174395955074806,
      "grad_norm": 0.4746832549571991,
      "learning_rate": 9.130092419909653e-07,
      "loss": 0.718,
      "step": 19724
    },
    {
      "epoch": 0.8174810394131543,
      "grad_norm": 0.42604297399520874,
      "learning_rate": 9.128020224625969e-07,
      "loss": 0.6847,
      "step": 19725
    },
    {
      "epoch": 0.8175224833188279,
      "grad_norm": 0.4185559153556824,
      "learning_rate": 9.125948029342286e-07,
      "loss": 0.7095,
      "step": 19726
    },
    {
      "epoch": 0.8175639272245017,
      "grad_norm": 0.4090959429740906,
      "learning_rate": 9.123875834058603e-07,
      "loss": 0.6874,
      "step": 19727
    },
    {
      "epoch": 0.8176053711301753,
      "grad_norm": 0.3999204635620117,
      "learning_rate": 9.121803638774919e-07,
      "loss": 0.6525,
      "step": 19728
    },
    {
      "epoch": 0.817646815035849,
      "grad_norm": 0.4440370202064514,
      "learning_rate": 9.119731443491235e-07,
      "loss": 0.6997,
      "step": 19729
    },
    {
      "epoch": 0.8176882589415226,
      "grad_norm": 0.41823819279670715,
      "learning_rate": 9.117659248207551e-07,
      "loss": 0.6654,
      "step": 19730
    },
    {
      "epoch": 0.8177297028471964,
      "grad_norm": 0.45928406715393066,
      "learning_rate": 9.115587052923869e-07,
      "loss": 0.6968,
      "step": 19731
    },
    {
      "epoch": 0.81777114675287,
      "grad_norm": 0.41114240884780884,
      "learning_rate": 9.113514857640185e-07,
      "loss": 0.6782,
      "step": 19732
    },
    {
      "epoch": 0.8178125906585436,
      "grad_norm": 0.38109442591667175,
      "learning_rate": 9.111442662356501e-07,
      "loss": 0.6187,
      "step": 19733
    },
    {
      "epoch": 0.8178540345642173,
      "grad_norm": 0.4392819106578827,
      "learning_rate": 9.109370467072817e-07,
      "loss": 0.6345,
      "step": 19734
    },
    {
      "epoch": 0.817895478469891,
      "grad_norm": 0.40617629885673523,
      "learning_rate": 9.107298271789135e-07,
      "loss": 0.662,
      "step": 19735
    },
    {
      "epoch": 0.8179369223755647,
      "grad_norm": 0.3965899348258972,
      "learning_rate": 9.105226076505451e-07,
      "loss": 0.6698,
      "step": 19736
    },
    {
      "epoch": 0.8179783662812383,
      "grad_norm": 0.4302690029144287,
      "learning_rate": 9.103153881221767e-07,
      "loss": 0.6599,
      "step": 19737
    },
    {
      "epoch": 0.818019810186912,
      "grad_norm": 0.39805907011032104,
      "learning_rate": 9.101081685938083e-07,
      "loss": 0.6229,
      "step": 19738
    },
    {
      "epoch": 0.8180612540925857,
      "grad_norm": 0.42825451493263245,
      "learning_rate": 9.0990094906544e-07,
      "loss": 0.6709,
      "step": 19739
    },
    {
      "epoch": 0.8181026979982594,
      "grad_norm": 0.4311811327934265,
      "learning_rate": 9.096937295370717e-07,
      "loss": 0.635,
      "step": 19740
    },
    {
      "epoch": 0.818144141903933,
      "grad_norm": 0.38889995217323303,
      "learning_rate": 9.094865100087033e-07,
      "loss": 0.6764,
      "step": 19741
    },
    {
      "epoch": 0.8181855858096067,
      "grad_norm": 0.4110477864742279,
      "learning_rate": 9.092792904803349e-07,
      "loss": 0.6724,
      "step": 19742
    },
    {
      "epoch": 0.8182270297152804,
      "grad_norm": 0.40320104360580444,
      "learning_rate": 9.090720709519666e-07,
      "loss": 0.7156,
      "step": 19743
    },
    {
      "epoch": 0.818268473620954,
      "grad_norm": 0.41246506571769714,
      "learning_rate": 9.088648514235983e-07,
      "loss": 0.6665,
      "step": 19744
    },
    {
      "epoch": 0.8183099175266277,
      "grad_norm": 0.4248059391975403,
      "learning_rate": 9.086576318952299e-07,
      "loss": 0.6622,
      "step": 19745
    },
    {
      "epoch": 0.8183513614323014,
      "grad_norm": 0.3999119997024536,
      "learning_rate": 9.084504123668615e-07,
      "loss": 0.6411,
      "step": 19746
    },
    {
      "epoch": 0.8183928053379751,
      "grad_norm": 0.42367807030677795,
      "learning_rate": 9.082431928384932e-07,
      "loss": 0.7061,
      "step": 19747
    },
    {
      "epoch": 0.8184342492436487,
      "grad_norm": 0.4492710828781128,
      "learning_rate": 9.080359733101249e-07,
      "loss": 0.6753,
      "step": 19748
    },
    {
      "epoch": 0.8184756931493224,
      "grad_norm": 0.4066287577152252,
      "learning_rate": 9.078287537817565e-07,
      "loss": 0.6536,
      "step": 19749
    },
    {
      "epoch": 0.8185171370549961,
      "grad_norm": 0.4213227331638336,
      "learning_rate": 9.076215342533881e-07,
      "loss": 0.6842,
      "step": 19750
    },
    {
      "epoch": 0.8185585809606697,
      "grad_norm": 0.4452371895313263,
      "learning_rate": 9.074143147250197e-07,
      "loss": 0.6733,
      "step": 19751
    },
    {
      "epoch": 0.8186000248663434,
      "grad_norm": 0.4622397720813751,
      "learning_rate": 9.072070951966514e-07,
      "loss": 0.6978,
      "step": 19752
    },
    {
      "epoch": 0.818641468772017,
      "grad_norm": 0.3998209834098816,
      "learning_rate": 9.069998756682831e-07,
      "loss": 0.6603,
      "step": 19753
    },
    {
      "epoch": 0.8186829126776908,
      "grad_norm": 0.4282058775424957,
      "learning_rate": 9.067926561399147e-07,
      "loss": 0.6646,
      "step": 19754
    },
    {
      "epoch": 0.8187243565833644,
      "grad_norm": 0.4603007733821869,
      "learning_rate": 9.065854366115463e-07,
      "loss": 0.6516,
      "step": 19755
    },
    {
      "epoch": 0.8187658004890381,
      "grad_norm": 0.43246859312057495,
      "learning_rate": 9.06378217083178e-07,
      "loss": 0.6626,
      "step": 19756
    },
    {
      "epoch": 0.8188072443947118,
      "grad_norm": 0.43439385294914246,
      "learning_rate": 9.061709975548097e-07,
      "loss": 0.6919,
      "step": 19757
    },
    {
      "epoch": 0.8188486883003854,
      "grad_norm": 0.4170182943344116,
      "learning_rate": 9.059637780264413e-07,
      "loss": 0.6849,
      "step": 19758
    },
    {
      "epoch": 0.8188901322060591,
      "grad_norm": 0.4048578441143036,
      "learning_rate": 9.057565584980729e-07,
      "loss": 0.6179,
      "step": 19759
    },
    {
      "epoch": 0.8189315761117327,
      "grad_norm": 0.4051901400089264,
      "learning_rate": 9.055493389697046e-07,
      "loss": 0.7023,
      "step": 19760
    },
    {
      "epoch": 0.8189730200174065,
      "grad_norm": 0.39366015791893005,
      "learning_rate": 9.053421194413363e-07,
      "loss": 0.6213,
      "step": 19761
    },
    {
      "epoch": 0.8190144639230801,
      "grad_norm": 0.44939351081848145,
      "learning_rate": 9.051348999129679e-07,
      "loss": 0.6541,
      "step": 19762
    },
    {
      "epoch": 0.8190559078287538,
      "grad_norm": 0.3908756673336029,
      "learning_rate": 9.049276803845995e-07,
      "loss": 0.667,
      "step": 19763
    },
    {
      "epoch": 0.8190973517344274,
      "grad_norm": 0.43734195828437805,
      "learning_rate": 9.047204608562312e-07,
      "loss": 0.6611,
      "step": 19764
    },
    {
      "epoch": 0.8191387956401012,
      "grad_norm": 0.42125099897384644,
      "learning_rate": 9.045132413278628e-07,
      "loss": 0.658,
      "step": 19765
    },
    {
      "epoch": 0.8191802395457748,
      "grad_norm": 0.43547943234443665,
      "learning_rate": 9.043060217994945e-07,
      "loss": 0.6921,
      "step": 19766
    },
    {
      "epoch": 0.8192216834514484,
      "grad_norm": 0.39512085914611816,
      "learning_rate": 9.040988022711261e-07,
      "loss": 0.6859,
      "step": 19767
    },
    {
      "epoch": 0.8192631273571221,
      "grad_norm": 0.4024547040462494,
      "learning_rate": 9.038915827427577e-07,
      "loss": 0.6753,
      "step": 19768
    },
    {
      "epoch": 0.8193045712627958,
      "grad_norm": 0.43935248255729675,
      "learning_rate": 9.036843632143894e-07,
      "loss": 0.679,
      "step": 19769
    },
    {
      "epoch": 0.8193460151684695,
      "grad_norm": 0.4244343042373657,
      "learning_rate": 9.034771436860211e-07,
      "loss": 0.6273,
      "step": 19770
    },
    {
      "epoch": 0.8193874590741431,
      "grad_norm": 0.41510331630706787,
      "learning_rate": 9.032699241576527e-07,
      "loss": 0.6539,
      "step": 19771
    },
    {
      "epoch": 0.8194289029798169,
      "grad_norm": 0.41384971141815186,
      "learning_rate": 9.030627046292843e-07,
      "loss": 0.6639,
      "step": 19772
    },
    {
      "epoch": 0.8194703468854905,
      "grad_norm": 0.4073028564453125,
      "learning_rate": 9.02855485100916e-07,
      "loss": 0.6466,
      "step": 19773
    },
    {
      "epoch": 0.8195117907911642,
      "grad_norm": 0.4025697708129883,
      "learning_rate": 9.026482655725477e-07,
      "loss": 0.6642,
      "step": 19774
    },
    {
      "epoch": 0.8195532346968378,
      "grad_norm": 0.42282360792160034,
      "learning_rate": 9.024410460441793e-07,
      "loss": 0.6531,
      "step": 19775
    },
    {
      "epoch": 0.8195946786025115,
      "grad_norm": 0.3846050500869751,
      "learning_rate": 9.022338265158109e-07,
      "loss": 0.6298,
      "step": 19776
    },
    {
      "epoch": 0.8196361225081852,
      "grad_norm": 0.41263172030448914,
      "learning_rate": 9.020266069874426e-07,
      "loss": 0.6655,
      "step": 19777
    },
    {
      "epoch": 0.8196775664138588,
      "grad_norm": 0.3998473584651947,
      "learning_rate": 9.018193874590742e-07,
      "loss": 0.6897,
      "step": 19778
    },
    {
      "epoch": 0.8197190103195325,
      "grad_norm": 0.38915055990219116,
      "learning_rate": 9.016121679307059e-07,
      "loss": 0.6772,
      "step": 19779
    },
    {
      "epoch": 0.8197604542252062,
      "grad_norm": 0.41776904463768005,
      "learning_rate": 9.014049484023375e-07,
      "loss": 0.7057,
      "step": 19780
    },
    {
      "epoch": 0.8198018981308799,
      "grad_norm": 0.40689817070961,
      "learning_rate": 9.011977288739692e-07,
      "loss": 0.6879,
      "step": 19781
    },
    {
      "epoch": 0.8198433420365535,
      "grad_norm": 0.4390327036380768,
      "learning_rate": 9.009905093456008e-07,
      "loss": 0.6588,
      "step": 19782
    },
    {
      "epoch": 0.8198847859422272,
      "grad_norm": 0.39893630146980286,
      "learning_rate": 9.007832898172325e-07,
      "loss": 0.6384,
      "step": 19783
    },
    {
      "epoch": 0.8199262298479009,
      "grad_norm": 0.4407680928707123,
      "learning_rate": 9.005760702888641e-07,
      "loss": 0.6329,
      "step": 19784
    },
    {
      "epoch": 0.8199676737535745,
      "grad_norm": 0.4602377712726593,
      "learning_rate": 9.003688507604957e-07,
      "loss": 0.6866,
      "step": 19785
    },
    {
      "epoch": 0.8200091176592482,
      "grad_norm": 0.4353601336479187,
      "learning_rate": 9.001616312321274e-07,
      "loss": 0.7,
      "step": 19786
    },
    {
      "epoch": 0.8200505615649218,
      "grad_norm": 0.4277232885360718,
      "learning_rate": 8.99954411703759e-07,
      "loss": 0.7246,
      "step": 19787
    },
    {
      "epoch": 0.8200920054705956,
      "grad_norm": 0.3654617667198181,
      "learning_rate": 8.997471921753907e-07,
      "loss": 0.6267,
      "step": 19788
    },
    {
      "epoch": 0.8201334493762692,
      "grad_norm": 0.39362895488739014,
      "learning_rate": 8.995399726470223e-07,
      "loss": 0.6421,
      "step": 19789
    },
    {
      "epoch": 0.8201748932819429,
      "grad_norm": 0.4053902328014374,
      "learning_rate": 8.99332753118654e-07,
      "loss": 0.6904,
      "step": 19790
    },
    {
      "epoch": 0.8202163371876166,
      "grad_norm": 0.3964919149875641,
      "learning_rate": 8.991255335902856e-07,
      "loss": 0.6353,
      "step": 19791
    },
    {
      "epoch": 0.8202577810932903,
      "grad_norm": 0.452632874250412,
      "learning_rate": 8.989183140619173e-07,
      "loss": 0.7083,
      "step": 19792
    },
    {
      "epoch": 0.8202992249989639,
      "grad_norm": 0.4176661968231201,
      "learning_rate": 8.987110945335489e-07,
      "loss": 0.6471,
      "step": 19793
    },
    {
      "epoch": 0.8203406689046375,
      "grad_norm": 0.4097972810268402,
      "learning_rate": 8.985038750051806e-07,
      "loss": 0.6752,
      "step": 19794
    },
    {
      "epoch": 0.8203821128103113,
      "grad_norm": 0.38780200481414795,
      "learning_rate": 8.982966554768122e-07,
      "loss": 0.5657,
      "step": 19795
    },
    {
      "epoch": 0.8204235567159849,
      "grad_norm": 0.4135057032108307,
      "learning_rate": 8.980894359484439e-07,
      "loss": 0.6648,
      "step": 19796
    },
    {
      "epoch": 0.8204650006216586,
      "grad_norm": 0.5055879354476929,
      "learning_rate": 8.978822164200755e-07,
      "loss": 0.7061,
      "step": 19797
    },
    {
      "epoch": 0.8205064445273322,
      "grad_norm": 0.43080827593803406,
      "learning_rate": 8.976749968917072e-07,
      "loss": 0.6742,
      "step": 19798
    },
    {
      "epoch": 0.820547888433006,
      "grad_norm": 0.4254550635814667,
      "learning_rate": 8.974677773633388e-07,
      "loss": 0.6514,
      "step": 19799
    },
    {
      "epoch": 0.8205893323386796,
      "grad_norm": 0.39122870564460754,
      "learning_rate": 8.972605578349704e-07,
      "loss": 0.6213,
      "step": 19800
    },
    {
      "epoch": 0.8206307762443533,
      "grad_norm": 0.41951683163642883,
      "learning_rate": 8.970533383066021e-07,
      "loss": 0.6545,
      "step": 19801
    },
    {
      "epoch": 0.8206722201500269,
      "grad_norm": 0.42110535502433777,
      "learning_rate": 8.968461187782338e-07,
      "loss": 0.6892,
      "step": 19802
    },
    {
      "epoch": 0.8207136640557006,
      "grad_norm": 0.45328107476234436,
      "learning_rate": 8.966388992498654e-07,
      "loss": 0.6964,
      "step": 19803
    },
    {
      "epoch": 0.8207551079613743,
      "grad_norm": 0.3902305066585541,
      "learning_rate": 8.96431679721497e-07,
      "loss": 0.6481,
      "step": 19804
    },
    {
      "epoch": 0.8207965518670479,
      "grad_norm": 0.3752572536468506,
      "learning_rate": 8.962244601931287e-07,
      "loss": 0.657,
      "step": 19805
    },
    {
      "epoch": 0.8208379957727217,
      "grad_norm": 0.4239276945590973,
      "learning_rate": 8.960172406647603e-07,
      "loss": 0.7094,
      "step": 19806
    },
    {
      "epoch": 0.8208794396783953,
      "grad_norm": 0.4271276593208313,
      "learning_rate": 8.95810021136392e-07,
      "loss": 0.6378,
      "step": 19807
    },
    {
      "epoch": 0.820920883584069,
      "grad_norm": 0.3974558115005493,
      "learning_rate": 8.956028016080236e-07,
      "loss": 0.6118,
      "step": 19808
    },
    {
      "epoch": 0.8209623274897426,
      "grad_norm": 0.4485943913459778,
      "learning_rate": 8.953955820796553e-07,
      "loss": 0.6829,
      "step": 19809
    },
    {
      "epoch": 0.8210037713954162,
      "grad_norm": 0.514271080493927,
      "learning_rate": 8.951883625512869e-07,
      "loss": 0.7303,
      "step": 19810
    },
    {
      "epoch": 0.82104521530109,
      "grad_norm": 0.43207061290740967,
      "learning_rate": 8.949811430229186e-07,
      "loss": 0.6813,
      "step": 19811
    },
    {
      "epoch": 0.8210866592067636,
      "grad_norm": 0.4120883345603943,
      "learning_rate": 8.947739234945502e-07,
      "loss": 0.6613,
      "step": 19812
    },
    {
      "epoch": 0.8211281031124373,
      "grad_norm": 0.40813973546028137,
      "learning_rate": 8.945667039661818e-07,
      "loss": 0.5936,
      "step": 19813
    },
    {
      "epoch": 0.821169547018111,
      "grad_norm": 0.4333094358444214,
      "learning_rate": 8.943594844378135e-07,
      "loss": 0.6364,
      "step": 19814
    },
    {
      "epoch": 0.8212109909237847,
      "grad_norm": 0.4044613838195801,
      "learning_rate": 8.941522649094452e-07,
      "loss": 0.6963,
      "step": 19815
    },
    {
      "epoch": 0.8212524348294583,
      "grad_norm": 0.3731793463230133,
      "learning_rate": 8.939450453810768e-07,
      "loss": 0.6252,
      "step": 19816
    },
    {
      "epoch": 0.821293878735132,
      "grad_norm": 0.4420129358768463,
      "learning_rate": 8.937378258527084e-07,
      "loss": 0.718,
      "step": 19817
    },
    {
      "epoch": 0.8213353226408057,
      "grad_norm": 0.3989412784576416,
      "learning_rate": 8.935306063243401e-07,
      "loss": 0.6726,
      "step": 19818
    },
    {
      "epoch": 0.8213767665464793,
      "grad_norm": 0.413006454706192,
      "learning_rate": 8.933233867959718e-07,
      "loss": 0.6393,
      "step": 19819
    },
    {
      "epoch": 0.821418210452153,
      "grad_norm": 0.4077928960323334,
      "learning_rate": 8.931161672676034e-07,
      "loss": 0.6157,
      "step": 19820
    },
    {
      "epoch": 0.8214596543578266,
      "grad_norm": 0.42799046635627747,
      "learning_rate": 8.92908947739235e-07,
      "loss": 0.6765,
      "step": 19821
    },
    {
      "epoch": 0.8215010982635004,
      "grad_norm": 0.47029295563697815,
      "learning_rate": 8.927017282108667e-07,
      "loss": 0.7362,
      "step": 19822
    },
    {
      "epoch": 0.821542542169174,
      "grad_norm": 0.4753129780292511,
      "learning_rate": 8.924945086824983e-07,
      "loss": 0.7144,
      "step": 19823
    },
    {
      "epoch": 0.8215839860748477,
      "grad_norm": 0.4120253622531891,
      "learning_rate": 8.9228728915413e-07,
      "loss": 0.644,
      "step": 19824
    },
    {
      "epoch": 0.8216254299805213,
      "grad_norm": 0.41009610891342163,
      "learning_rate": 8.920800696257616e-07,
      "loss": 0.6392,
      "step": 19825
    },
    {
      "epoch": 0.8216668738861951,
      "grad_norm": 0.44124719500541687,
      "learning_rate": 8.918728500973932e-07,
      "loss": 0.676,
      "step": 19826
    },
    {
      "epoch": 0.8217083177918687,
      "grad_norm": 0.4321514368057251,
      "learning_rate": 8.916656305690249e-07,
      "loss": 0.6738,
      "step": 19827
    },
    {
      "epoch": 0.8217497616975423,
      "grad_norm": 0.4038153886795044,
      "learning_rate": 8.914584110406566e-07,
      "loss": 0.6721,
      "step": 19828
    },
    {
      "epoch": 0.8217912056032161,
      "grad_norm": 0.4090587794780731,
      "learning_rate": 8.912511915122882e-07,
      "loss": 0.6368,
      "step": 19829
    },
    {
      "epoch": 0.8218326495088897,
      "grad_norm": 0.4300379753112793,
      "learning_rate": 8.910439719839198e-07,
      "loss": 0.6693,
      "step": 19830
    },
    {
      "epoch": 0.8218740934145634,
      "grad_norm": 0.385168194770813,
      "learning_rate": 8.908367524555515e-07,
      "loss": 0.6685,
      "step": 19831
    },
    {
      "epoch": 0.821915537320237,
      "grad_norm": 0.4192860424518585,
      "learning_rate": 8.906295329271832e-07,
      "loss": 0.6349,
      "step": 19832
    },
    {
      "epoch": 0.8219569812259108,
      "grad_norm": 0.48932138085365295,
      "learning_rate": 8.904223133988148e-07,
      "loss": 0.7098,
      "step": 19833
    },
    {
      "epoch": 0.8219984251315844,
      "grad_norm": 0.4394828677177429,
      "learning_rate": 8.902150938704464e-07,
      "loss": 0.6969,
      "step": 19834
    },
    {
      "epoch": 0.8220398690372581,
      "grad_norm": 0.3691657781600952,
      "learning_rate": 8.900078743420781e-07,
      "loss": 0.5822,
      "step": 19835
    },
    {
      "epoch": 0.8220813129429317,
      "grad_norm": 0.40473607182502747,
      "learning_rate": 8.898006548137098e-07,
      "loss": 0.7527,
      "step": 19836
    },
    {
      "epoch": 0.8221227568486054,
      "grad_norm": 0.3698449730873108,
      "learning_rate": 8.895934352853414e-07,
      "loss": 0.6306,
      "step": 19837
    },
    {
      "epoch": 0.8221642007542791,
      "grad_norm": 0.41561150550842285,
      "learning_rate": 8.89386215756973e-07,
      "loss": 0.6312,
      "step": 19838
    },
    {
      "epoch": 0.8222056446599527,
      "grad_norm": 0.3987681269645691,
      "learning_rate": 8.891789962286046e-07,
      "loss": 0.6278,
      "step": 19839
    },
    {
      "epoch": 0.8222470885656264,
      "grad_norm": 0.41775357723236084,
      "learning_rate": 8.889717767002363e-07,
      "loss": 0.683,
      "step": 19840
    },
    {
      "epoch": 0.8222885324713001,
      "grad_norm": 0.42423057556152344,
      "learning_rate": 8.88764557171868e-07,
      "loss": 0.655,
      "step": 19841
    },
    {
      "epoch": 0.8223299763769738,
      "grad_norm": 0.428254097700119,
      "learning_rate": 8.885573376434996e-07,
      "loss": 0.6832,
      "step": 19842
    },
    {
      "epoch": 0.8223714202826474,
      "grad_norm": 0.4318723678588867,
      "learning_rate": 8.883501181151312e-07,
      "loss": 0.6799,
      "step": 19843
    },
    {
      "epoch": 0.8224128641883212,
      "grad_norm": 0.44461116194725037,
      "learning_rate": 8.881428985867629e-07,
      "loss": 0.6168,
      "step": 19844
    },
    {
      "epoch": 0.8224543080939948,
      "grad_norm": 0.42376846075057983,
      "learning_rate": 8.879356790583946e-07,
      "loss": 0.6681,
      "step": 19845
    },
    {
      "epoch": 0.8224957519996684,
      "grad_norm": 0.41388991475105286,
      "learning_rate": 8.877284595300262e-07,
      "loss": 0.6298,
      "step": 19846
    },
    {
      "epoch": 0.8225371959053421,
      "grad_norm": 0.4312250316143036,
      "learning_rate": 8.875212400016578e-07,
      "loss": 0.6539,
      "step": 19847
    },
    {
      "epoch": 0.8225786398110158,
      "grad_norm": 0.4134850800037384,
      "learning_rate": 8.873140204732895e-07,
      "loss": 0.6824,
      "step": 19848
    },
    {
      "epoch": 0.8226200837166895,
      "grad_norm": 0.43058791756629944,
      "learning_rate": 8.871068009449212e-07,
      "loss": 0.6647,
      "step": 19849
    },
    {
      "epoch": 0.8226615276223631,
      "grad_norm": 0.4457002878189087,
      "learning_rate": 8.868995814165528e-07,
      "loss": 0.6907,
      "step": 19850
    },
    {
      "epoch": 0.8227029715280368,
      "grad_norm": 0.44663041830062866,
      "learning_rate": 8.866923618881844e-07,
      "loss": 0.6697,
      "step": 19851
    },
    {
      "epoch": 0.8227444154337105,
      "grad_norm": 0.4230356216430664,
      "learning_rate": 8.86485142359816e-07,
      "loss": 0.6938,
      "step": 19852
    },
    {
      "epoch": 0.8227858593393842,
      "grad_norm": 0.3900780975818634,
      "learning_rate": 8.862779228314478e-07,
      "loss": 0.642,
      "step": 19853
    },
    {
      "epoch": 0.8228273032450578,
      "grad_norm": 0.38914573192596436,
      "learning_rate": 8.860707033030794e-07,
      "loss": 0.6621,
      "step": 19854
    },
    {
      "epoch": 0.8228687471507314,
      "grad_norm": 0.3960849940776825,
      "learning_rate": 8.85863483774711e-07,
      "loss": 0.6505,
      "step": 19855
    },
    {
      "epoch": 0.8229101910564052,
      "grad_norm": 0.4598386585712433,
      "learning_rate": 8.856562642463426e-07,
      "loss": 0.7212,
      "step": 19856
    },
    {
      "epoch": 0.8229516349620788,
      "grad_norm": 0.3934946060180664,
      "learning_rate": 8.854490447179744e-07,
      "loss": 0.6353,
      "step": 19857
    },
    {
      "epoch": 0.8229930788677525,
      "grad_norm": 0.4103161692619324,
      "learning_rate": 8.85241825189606e-07,
      "loss": 0.6143,
      "step": 19858
    },
    {
      "epoch": 0.8230345227734261,
      "grad_norm": 0.4113180637359619,
      "learning_rate": 8.850346056612376e-07,
      "loss": 0.635,
      "step": 19859
    },
    {
      "epoch": 0.8230759666790999,
      "grad_norm": 0.4891010522842407,
      "learning_rate": 8.848273861328692e-07,
      "loss": 0.6869,
      "step": 19860
    },
    {
      "epoch": 0.8231174105847735,
      "grad_norm": 0.40585508942604065,
      "learning_rate": 8.846201666045008e-07,
      "loss": 0.6776,
      "step": 19861
    },
    {
      "epoch": 0.8231588544904472,
      "grad_norm": 0.46703702211380005,
      "learning_rate": 8.844129470761326e-07,
      "loss": 0.6719,
      "step": 19862
    },
    {
      "epoch": 0.8232002983961209,
      "grad_norm": 0.4452812671661377,
      "learning_rate": 8.842057275477642e-07,
      "loss": 0.7026,
      "step": 19863
    },
    {
      "epoch": 0.8232417423017945,
      "grad_norm": 0.4191691279411316,
      "learning_rate": 8.839985080193958e-07,
      "loss": 0.6758,
      "step": 19864
    },
    {
      "epoch": 0.8232831862074682,
      "grad_norm": 0.40663719177246094,
      "learning_rate": 8.837912884910274e-07,
      "loss": 0.6904,
      "step": 19865
    },
    {
      "epoch": 0.8233246301131418,
      "grad_norm": 0.43943801522254944,
      "learning_rate": 8.835840689626592e-07,
      "loss": 0.7029,
      "step": 19866
    },
    {
      "epoch": 0.8233660740188156,
      "grad_norm": 0.40461650490760803,
      "learning_rate": 8.833768494342908e-07,
      "loss": 0.66,
      "step": 19867
    },
    {
      "epoch": 0.8234075179244892,
      "grad_norm": 0.40240442752838135,
      "learning_rate": 8.831696299059224e-07,
      "loss": 0.6465,
      "step": 19868
    },
    {
      "epoch": 0.8234489618301629,
      "grad_norm": 0.4185279309749603,
      "learning_rate": 8.82962410377554e-07,
      "loss": 0.6704,
      "step": 19869
    },
    {
      "epoch": 0.8234904057358365,
      "grad_norm": 0.4145281910896301,
      "learning_rate": 8.827551908491858e-07,
      "loss": 0.6223,
      "step": 19870
    },
    {
      "epoch": 0.8235318496415102,
      "grad_norm": 0.38802003860473633,
      "learning_rate": 8.825479713208174e-07,
      "loss": 0.6953,
      "step": 19871
    },
    {
      "epoch": 0.8235732935471839,
      "grad_norm": 0.39619219303131104,
      "learning_rate": 8.82340751792449e-07,
      "loss": 0.6381,
      "step": 19872
    },
    {
      "epoch": 0.8236147374528575,
      "grad_norm": 0.4229048490524292,
      "learning_rate": 8.821335322640806e-07,
      "loss": 0.66,
      "step": 19873
    },
    {
      "epoch": 0.8236561813585312,
      "grad_norm": 0.43544185161590576,
      "learning_rate": 8.819263127357122e-07,
      "loss": 0.6211,
      "step": 19874
    },
    {
      "epoch": 0.8236976252642049,
      "grad_norm": 0.4113274812698364,
      "learning_rate": 8.81719093207344e-07,
      "loss": 0.6707,
      "step": 19875
    },
    {
      "epoch": 0.8237390691698786,
      "grad_norm": 0.43609076738357544,
      "learning_rate": 8.815118736789756e-07,
      "loss": 0.6874,
      "step": 19876
    },
    {
      "epoch": 0.8237805130755522,
      "grad_norm": 0.42663663625717163,
      "learning_rate": 8.813046541506072e-07,
      "loss": 0.6426,
      "step": 19877
    },
    {
      "epoch": 0.823821956981226,
      "grad_norm": 0.3920465409755707,
      "learning_rate": 8.810974346222388e-07,
      "loss": 0.657,
      "step": 19878
    },
    {
      "epoch": 0.8238634008868996,
      "grad_norm": 0.4364449977874756,
      "learning_rate": 8.808902150938706e-07,
      "loss": 0.6501,
      "step": 19879
    },
    {
      "epoch": 0.8239048447925732,
      "grad_norm": 0.42272433638572693,
      "learning_rate": 8.806829955655022e-07,
      "loss": 0.7186,
      "step": 19880
    },
    {
      "epoch": 0.8239462886982469,
      "grad_norm": 0.42049479484558105,
      "learning_rate": 8.804757760371338e-07,
      "loss": 0.6589,
      "step": 19881
    },
    {
      "epoch": 0.8239877326039206,
      "grad_norm": 0.3955175578594208,
      "learning_rate": 8.802685565087654e-07,
      "loss": 0.6554,
      "step": 19882
    },
    {
      "epoch": 0.8240291765095943,
      "grad_norm": 0.3816300332546234,
      "learning_rate": 8.800613369803972e-07,
      "loss": 0.6506,
      "step": 19883
    },
    {
      "epoch": 0.8240706204152679,
      "grad_norm": 0.3778861463069916,
      "learning_rate": 8.798541174520288e-07,
      "loss": 0.5952,
      "step": 19884
    },
    {
      "epoch": 0.8241120643209416,
      "grad_norm": 0.3931732177734375,
      "learning_rate": 8.796468979236604e-07,
      "loss": 0.647,
      "step": 19885
    },
    {
      "epoch": 0.8241535082266153,
      "grad_norm": 0.4139452874660492,
      "learning_rate": 8.79439678395292e-07,
      "loss": 0.6715,
      "step": 19886
    },
    {
      "epoch": 0.824194952132289,
      "grad_norm": 0.4181923270225525,
      "learning_rate": 8.792324588669236e-07,
      "loss": 0.6465,
      "step": 19887
    },
    {
      "epoch": 0.8242363960379626,
      "grad_norm": 0.4303172826766968,
      "learning_rate": 8.790252393385554e-07,
      "loss": 0.7083,
      "step": 19888
    },
    {
      "epoch": 0.8242778399436362,
      "grad_norm": 0.40997904539108276,
      "learning_rate": 8.78818019810187e-07,
      "loss": 0.652,
      "step": 19889
    },
    {
      "epoch": 0.82431928384931,
      "grad_norm": 0.40727031230926514,
      "learning_rate": 8.786108002818186e-07,
      "loss": 0.6842,
      "step": 19890
    },
    {
      "epoch": 0.8243607277549836,
      "grad_norm": 0.39041846990585327,
      "learning_rate": 8.784035807534502e-07,
      "loss": 0.6141,
      "step": 19891
    },
    {
      "epoch": 0.8244021716606573,
      "grad_norm": 0.40774205327033997,
      "learning_rate": 8.78196361225082e-07,
      "loss": 0.6523,
      "step": 19892
    },
    {
      "epoch": 0.824443615566331,
      "grad_norm": 0.4005863666534424,
      "learning_rate": 8.779891416967136e-07,
      "loss": 0.6283,
      "step": 19893
    },
    {
      "epoch": 0.8244850594720047,
      "grad_norm": 0.5196129083633423,
      "learning_rate": 8.777819221683452e-07,
      "loss": 0.7163,
      "step": 19894
    },
    {
      "epoch": 0.8245265033776783,
      "grad_norm": 0.39395537972450256,
      "learning_rate": 8.775747026399768e-07,
      "loss": 0.6682,
      "step": 19895
    },
    {
      "epoch": 0.824567947283352,
      "grad_norm": 0.43335792422294617,
      "learning_rate": 8.773674831116086e-07,
      "loss": 0.6846,
      "step": 19896
    },
    {
      "epoch": 0.8246093911890257,
      "grad_norm": 0.4048047959804535,
      "learning_rate": 8.771602635832402e-07,
      "loss": 0.6267,
      "step": 19897
    },
    {
      "epoch": 0.8246508350946993,
      "grad_norm": 0.42945295572280884,
      "learning_rate": 8.769530440548718e-07,
      "loss": 0.6975,
      "step": 19898
    },
    {
      "epoch": 0.824692279000373,
      "grad_norm": 0.42068254947662354,
      "learning_rate": 8.767458245265034e-07,
      "loss": 0.6473,
      "step": 19899
    },
    {
      "epoch": 0.8247337229060466,
      "grad_norm": 0.40256649255752563,
      "learning_rate": 8.76538604998135e-07,
      "loss": 0.6798,
      "step": 19900
    },
    {
      "epoch": 0.8247751668117204,
      "grad_norm": 0.4270101487636566,
      "learning_rate": 8.763313854697668e-07,
      "loss": 0.6373,
      "step": 19901
    },
    {
      "epoch": 0.824816610717394,
      "grad_norm": 0.41601482033729553,
      "learning_rate": 8.761241659413984e-07,
      "loss": 0.65,
      "step": 19902
    },
    {
      "epoch": 0.8248580546230677,
      "grad_norm": 0.4278353452682495,
      "learning_rate": 8.7591694641303e-07,
      "loss": 0.6721,
      "step": 19903
    },
    {
      "epoch": 0.8248994985287413,
      "grad_norm": 0.45707967877388,
      "learning_rate": 8.757097268846616e-07,
      "loss": 0.684,
      "step": 19904
    },
    {
      "epoch": 0.8249409424344151,
      "grad_norm": 0.44562390446662903,
      "learning_rate": 8.755025073562934e-07,
      "loss": 0.6765,
      "step": 19905
    },
    {
      "epoch": 0.8249823863400887,
      "grad_norm": 0.39573732018470764,
      "learning_rate": 8.75295287827925e-07,
      "loss": 0.6519,
      "step": 19906
    },
    {
      "epoch": 0.8250238302457623,
      "grad_norm": 0.36985552310943604,
      "learning_rate": 8.750880682995566e-07,
      "loss": 0.6005,
      "step": 19907
    },
    {
      "epoch": 0.825065274151436,
      "grad_norm": 0.4180866777896881,
      "learning_rate": 8.748808487711882e-07,
      "loss": 0.6593,
      "step": 19908
    },
    {
      "epoch": 0.8251067180571097,
      "grad_norm": 0.3873366713523865,
      "learning_rate": 8.7467362924282e-07,
      "loss": 0.6841,
      "step": 19909
    },
    {
      "epoch": 0.8251481619627834,
      "grad_norm": 0.41564396023750305,
      "learning_rate": 8.744664097144516e-07,
      "loss": 0.6482,
      "step": 19910
    },
    {
      "epoch": 0.825189605868457,
      "grad_norm": 0.41485926508903503,
      "learning_rate": 8.742591901860832e-07,
      "loss": 0.6584,
      "step": 19911
    },
    {
      "epoch": 0.8252310497741308,
      "grad_norm": 0.40630820393562317,
      "learning_rate": 8.740519706577148e-07,
      "loss": 0.6506,
      "step": 19912
    },
    {
      "epoch": 0.8252724936798044,
      "grad_norm": 0.43116316199302673,
      "learning_rate": 8.738447511293464e-07,
      "loss": 0.6876,
      "step": 19913
    },
    {
      "epoch": 0.8253139375854781,
      "grad_norm": 0.4070904850959778,
      "learning_rate": 8.736375316009782e-07,
      "loss": 0.6719,
      "step": 19914
    },
    {
      "epoch": 0.8253553814911517,
      "grad_norm": 0.41669800877571106,
      "learning_rate": 8.734303120726098e-07,
      "loss": 0.6545,
      "step": 19915
    },
    {
      "epoch": 0.8253968253968254,
      "grad_norm": 0.4123237133026123,
      "learning_rate": 8.732230925442414e-07,
      "loss": 0.736,
      "step": 19916
    },
    {
      "epoch": 0.8254382693024991,
      "grad_norm": 0.4123421609401703,
      "learning_rate": 8.73015873015873e-07,
      "loss": 0.6926,
      "step": 19917
    },
    {
      "epoch": 0.8254797132081727,
      "grad_norm": 0.42681917548179626,
      "learning_rate": 8.728086534875048e-07,
      "loss": 0.6792,
      "step": 19918
    },
    {
      "epoch": 0.8255211571138464,
      "grad_norm": 0.41366487741470337,
      "learning_rate": 8.726014339591364e-07,
      "loss": 0.677,
      "step": 19919
    },
    {
      "epoch": 0.8255626010195201,
      "grad_norm": 0.40823736786842346,
      "learning_rate": 8.72394214430768e-07,
      "loss": 0.6447,
      "step": 19920
    },
    {
      "epoch": 0.8256040449251938,
      "grad_norm": 0.43181225657463074,
      "learning_rate": 8.721869949023996e-07,
      "loss": 0.6567,
      "step": 19921
    },
    {
      "epoch": 0.8256454888308674,
      "grad_norm": 0.4298512637615204,
      "learning_rate": 8.719797753740312e-07,
      "loss": 0.6351,
      "step": 19922
    },
    {
      "epoch": 0.8256869327365411,
      "grad_norm": 0.4529671370983124,
      "learning_rate": 8.71772555845663e-07,
      "loss": 0.6858,
      "step": 19923
    },
    {
      "epoch": 0.8257283766422148,
      "grad_norm": 0.42633047699928284,
      "learning_rate": 8.715653363172946e-07,
      "loss": 0.653,
      "step": 19924
    },
    {
      "epoch": 0.8257698205478884,
      "grad_norm": 0.40717771649360657,
      "learning_rate": 8.713581167889262e-07,
      "loss": 0.6741,
      "step": 19925
    },
    {
      "epoch": 0.8258112644535621,
      "grad_norm": 0.3932841122150421,
      "learning_rate": 8.711508972605578e-07,
      "loss": 0.6353,
      "step": 19926
    },
    {
      "epoch": 0.8258527083592357,
      "grad_norm": 0.39596128463745117,
      "learning_rate": 8.709436777321896e-07,
      "loss": 0.6693,
      "step": 19927
    },
    {
      "epoch": 0.8258941522649095,
      "grad_norm": 0.47697025537490845,
      "learning_rate": 8.707364582038212e-07,
      "loss": 0.7051,
      "step": 19928
    },
    {
      "epoch": 0.8259355961705831,
      "grad_norm": 0.38695260882377625,
      "learning_rate": 8.705292386754528e-07,
      "loss": 0.6315,
      "step": 19929
    },
    {
      "epoch": 0.8259770400762568,
      "grad_norm": 0.4535088539123535,
      "learning_rate": 8.703220191470844e-07,
      "loss": 0.6687,
      "step": 19930
    },
    {
      "epoch": 0.8260184839819305,
      "grad_norm": 0.4227275550365448,
      "learning_rate": 8.701147996187162e-07,
      "loss": 0.6876,
      "step": 19931
    },
    {
      "epoch": 0.8260599278876041,
      "grad_norm": 0.448953777551651,
      "learning_rate": 8.699075800903478e-07,
      "loss": 0.6597,
      "step": 19932
    },
    {
      "epoch": 0.8261013717932778,
      "grad_norm": 0.47158339619636536,
      "learning_rate": 8.697003605619794e-07,
      "loss": 0.6965,
      "step": 19933
    },
    {
      "epoch": 0.8261428156989514,
      "grad_norm": 0.4061207175254822,
      "learning_rate": 8.69493141033611e-07,
      "loss": 0.6285,
      "step": 19934
    },
    {
      "epoch": 0.8261842596046252,
      "grad_norm": 0.41902783513069153,
      "learning_rate": 8.692859215052427e-07,
      "loss": 0.6229,
      "step": 19935
    },
    {
      "epoch": 0.8262257035102988,
      "grad_norm": 0.3826889097690582,
      "learning_rate": 8.690787019768744e-07,
      "loss": 0.6694,
      "step": 19936
    },
    {
      "epoch": 0.8262671474159725,
      "grad_norm": 0.4034103453159332,
      "learning_rate": 8.68871482448506e-07,
      "loss": 0.6615,
      "step": 19937
    },
    {
      "epoch": 0.8263085913216461,
      "grad_norm": 0.4457993805408478,
      "learning_rate": 8.686642629201376e-07,
      "loss": 0.6865,
      "step": 19938
    },
    {
      "epoch": 0.8263500352273199,
      "grad_norm": 0.40942978858947754,
      "learning_rate": 8.684570433917692e-07,
      "loss": 0.6691,
      "step": 19939
    },
    {
      "epoch": 0.8263914791329935,
      "grad_norm": 0.43380528688430786,
      "learning_rate": 8.68249823863401e-07,
      "loss": 0.7126,
      "step": 19940
    },
    {
      "epoch": 0.8264329230386671,
      "grad_norm": 0.38526079058647156,
      "learning_rate": 8.680426043350326e-07,
      "loss": 0.6371,
      "step": 19941
    },
    {
      "epoch": 0.8264743669443408,
      "grad_norm": 0.40194422006607056,
      "learning_rate": 8.678353848066642e-07,
      "loss": 0.6554,
      "step": 19942
    },
    {
      "epoch": 0.8265158108500145,
      "grad_norm": 0.41399288177490234,
      "learning_rate": 8.676281652782958e-07,
      "loss": 0.7267,
      "step": 19943
    },
    {
      "epoch": 0.8265572547556882,
      "grad_norm": 0.395340234041214,
      "learning_rate": 8.674209457499276e-07,
      "loss": 0.6581,
      "step": 19944
    },
    {
      "epoch": 0.8265986986613618,
      "grad_norm": 0.3893130123615265,
      "learning_rate": 8.672137262215592e-07,
      "loss": 0.6973,
      "step": 19945
    },
    {
      "epoch": 0.8266401425670356,
      "grad_norm": 0.40726932883262634,
      "learning_rate": 8.670065066931908e-07,
      "loss": 0.6013,
      "step": 19946
    },
    {
      "epoch": 0.8266815864727092,
      "grad_norm": 0.4310166835784912,
      "learning_rate": 8.667992871648224e-07,
      "loss": 0.6962,
      "step": 19947
    },
    {
      "epoch": 0.8267230303783829,
      "grad_norm": 0.42941218614578247,
      "learning_rate": 8.665920676364541e-07,
      "loss": 0.6595,
      "step": 19948
    },
    {
      "epoch": 0.8267644742840565,
      "grad_norm": 0.3880023658275604,
      "learning_rate": 8.663848481080858e-07,
      "loss": 0.6118,
      "step": 19949
    },
    {
      "epoch": 0.8268059181897301,
      "grad_norm": 0.3884003460407257,
      "learning_rate": 8.661776285797174e-07,
      "loss": 0.6316,
      "step": 19950
    },
    {
      "epoch": 0.8268473620954039,
      "grad_norm": 0.4087783992290497,
      "learning_rate": 8.65970409051349e-07,
      "loss": 0.6067,
      "step": 19951
    },
    {
      "epoch": 0.8268888060010775,
      "grad_norm": 0.41053497791290283,
      "learning_rate": 8.657631895229807e-07,
      "loss": 0.7322,
      "step": 19952
    },
    {
      "epoch": 0.8269302499067512,
      "grad_norm": 0.4300461709499359,
      "learning_rate": 8.655559699946124e-07,
      "loss": 0.6383,
      "step": 19953
    },
    {
      "epoch": 0.8269716938124249,
      "grad_norm": 0.4125569760799408,
      "learning_rate": 8.65348750466244e-07,
      "loss": 0.6433,
      "step": 19954
    },
    {
      "epoch": 0.8270131377180986,
      "grad_norm": 0.4206247627735138,
      "learning_rate": 8.651415309378756e-07,
      "loss": 0.6488,
      "step": 19955
    },
    {
      "epoch": 0.8270545816237722,
      "grad_norm": 0.44500041007995605,
      "learning_rate": 8.649343114095072e-07,
      "loss": 0.6992,
      "step": 19956
    },
    {
      "epoch": 0.8270960255294459,
      "grad_norm": 0.4274558424949646,
      "learning_rate": 8.64727091881139e-07,
      "loss": 0.6648,
      "step": 19957
    },
    {
      "epoch": 0.8271374694351196,
      "grad_norm": 0.43079593777656555,
      "learning_rate": 8.645198723527706e-07,
      "loss": 0.71,
      "step": 19958
    },
    {
      "epoch": 0.8271789133407932,
      "grad_norm": 0.42514410614967346,
      "learning_rate": 8.643126528244022e-07,
      "loss": 0.6909,
      "step": 19959
    },
    {
      "epoch": 0.8272203572464669,
      "grad_norm": 0.3957414925098419,
      "learning_rate": 8.641054332960338e-07,
      "loss": 0.6268,
      "step": 19960
    },
    {
      "epoch": 0.8272618011521405,
      "grad_norm": 0.4250696301460266,
      "learning_rate": 8.638982137676655e-07,
      "loss": 0.6252,
      "step": 19961
    },
    {
      "epoch": 0.8273032450578143,
      "grad_norm": 0.4406365752220154,
      "learning_rate": 8.636909942392972e-07,
      "loss": 0.6565,
      "step": 19962
    },
    {
      "epoch": 0.8273446889634879,
      "grad_norm": 0.41153931617736816,
      "learning_rate": 8.634837747109288e-07,
      "loss": 0.6818,
      "step": 19963
    },
    {
      "epoch": 0.8273861328691616,
      "grad_norm": 0.4214142858982086,
      "learning_rate": 8.632765551825604e-07,
      "loss": 0.6743,
      "step": 19964
    },
    {
      "epoch": 0.8274275767748352,
      "grad_norm": 0.40621864795684814,
      "learning_rate": 8.630693356541921e-07,
      "loss": 0.6912,
      "step": 19965
    },
    {
      "epoch": 0.827469020680509,
      "grad_norm": 0.42468762397766113,
      "learning_rate": 8.628621161258238e-07,
      "loss": 0.7075,
      "step": 19966
    },
    {
      "epoch": 0.8275104645861826,
      "grad_norm": 0.39489686489105225,
      "learning_rate": 8.626548965974554e-07,
      "loss": 0.6631,
      "step": 19967
    },
    {
      "epoch": 0.8275519084918562,
      "grad_norm": 0.425238698720932,
      "learning_rate": 8.62447677069087e-07,
      "loss": 0.6444,
      "step": 19968
    },
    {
      "epoch": 0.82759335239753,
      "grad_norm": 0.4191061556339264,
      "learning_rate": 8.622404575407187e-07,
      "loss": 0.6437,
      "step": 19969
    },
    {
      "epoch": 0.8276347963032036,
      "grad_norm": 0.4298592209815979,
      "learning_rate": 8.620332380123504e-07,
      "loss": 0.6494,
      "step": 19970
    },
    {
      "epoch": 0.8276762402088773,
      "grad_norm": 0.40879446268081665,
      "learning_rate": 8.61826018483982e-07,
      "loss": 0.6229,
      "step": 19971
    },
    {
      "epoch": 0.8277176841145509,
      "grad_norm": 0.39823323488235474,
      "learning_rate": 8.616187989556136e-07,
      "loss": 0.6279,
      "step": 19972
    },
    {
      "epoch": 0.8277591280202247,
      "grad_norm": 0.3998973071575165,
      "learning_rate": 8.614115794272453e-07,
      "loss": 0.6783,
      "step": 19973
    },
    {
      "epoch": 0.8278005719258983,
      "grad_norm": 0.4314455986022949,
      "learning_rate": 8.612043598988769e-07,
      "loss": 0.6841,
      "step": 19974
    },
    {
      "epoch": 0.827842015831572,
      "grad_norm": 0.4018285572528839,
      "learning_rate": 8.609971403705086e-07,
      "loss": 0.6371,
      "step": 19975
    },
    {
      "epoch": 0.8278834597372456,
      "grad_norm": 0.44595712423324585,
      "learning_rate": 8.607899208421402e-07,
      "loss": 0.7227,
      "step": 19976
    },
    {
      "epoch": 0.8279249036429193,
      "grad_norm": 0.3940489590167999,
      "learning_rate": 8.605827013137718e-07,
      "loss": 0.655,
      "step": 19977
    },
    {
      "epoch": 0.827966347548593,
      "grad_norm": 0.46020129323005676,
      "learning_rate": 8.603754817854035e-07,
      "loss": 0.7246,
      "step": 19978
    },
    {
      "epoch": 0.8280077914542666,
      "grad_norm": 0.42290931940078735,
      "learning_rate": 8.601682622570352e-07,
      "loss": 0.6593,
      "step": 19979
    },
    {
      "epoch": 0.8280492353599403,
      "grad_norm": 0.4424780607223511,
      "learning_rate": 8.599610427286668e-07,
      "loss": 0.7007,
      "step": 19980
    },
    {
      "epoch": 0.828090679265614,
      "grad_norm": 0.4345730245113373,
      "learning_rate": 8.597538232002984e-07,
      "loss": 0.6843,
      "step": 19981
    },
    {
      "epoch": 0.8281321231712877,
      "grad_norm": 0.39915740489959717,
      "learning_rate": 8.595466036719301e-07,
      "loss": 0.6595,
      "step": 19982
    },
    {
      "epoch": 0.8281735670769613,
      "grad_norm": 0.43529045581817627,
      "learning_rate": 8.593393841435618e-07,
      "loss": 0.6647,
      "step": 19983
    },
    {
      "epoch": 0.8282150109826351,
      "grad_norm": 0.3980282247066498,
      "learning_rate": 8.591321646151934e-07,
      "loss": 0.6272,
      "step": 19984
    },
    {
      "epoch": 0.8282564548883087,
      "grad_norm": 0.425192266702652,
      "learning_rate": 8.58924945086825e-07,
      "loss": 0.6722,
      "step": 19985
    },
    {
      "epoch": 0.8282978987939823,
      "grad_norm": 0.39500176906585693,
      "learning_rate": 8.587177255584567e-07,
      "loss": 0.6597,
      "step": 19986
    },
    {
      "epoch": 0.828339342699656,
      "grad_norm": 0.4205699563026428,
      "learning_rate": 8.585105060300883e-07,
      "loss": 0.6759,
      "step": 19987
    },
    {
      "epoch": 0.8283807866053297,
      "grad_norm": 0.43331825733184814,
      "learning_rate": 8.5830328650172e-07,
      "loss": 0.6914,
      "step": 19988
    },
    {
      "epoch": 0.8284222305110034,
      "grad_norm": 0.46068274974823,
      "learning_rate": 8.580960669733516e-07,
      "loss": 0.673,
      "step": 19989
    },
    {
      "epoch": 0.828463674416677,
      "grad_norm": 0.4112700819969177,
      "learning_rate": 8.578888474449833e-07,
      "loss": 0.6748,
      "step": 19990
    },
    {
      "epoch": 0.8285051183223507,
      "grad_norm": 0.3983365297317505,
      "learning_rate": 8.576816279166149e-07,
      "loss": 0.6262,
      "step": 19991
    },
    {
      "epoch": 0.8285465622280244,
      "grad_norm": 0.4470222294330597,
      "learning_rate": 8.574744083882466e-07,
      "loss": 0.6844,
      "step": 19992
    },
    {
      "epoch": 0.828588006133698,
      "grad_norm": 0.42061251401901245,
      "learning_rate": 8.572671888598782e-07,
      "loss": 0.6584,
      "step": 19993
    },
    {
      "epoch": 0.8286294500393717,
      "grad_norm": 0.390026718378067,
      "learning_rate": 8.570599693315098e-07,
      "loss": 0.6715,
      "step": 19994
    },
    {
      "epoch": 0.8286708939450453,
      "grad_norm": 0.42781898379325867,
      "learning_rate": 8.568527498031415e-07,
      "loss": 0.692,
      "step": 19995
    },
    {
      "epoch": 0.8287123378507191,
      "grad_norm": 0.4022115170955658,
      "learning_rate": 8.566455302747731e-07,
      "loss": 0.663,
      "step": 19996
    },
    {
      "epoch": 0.8287537817563927,
      "grad_norm": 0.4376991093158722,
      "learning_rate": 8.564383107464048e-07,
      "loss": 0.6875,
      "step": 19997
    },
    {
      "epoch": 0.8287952256620664,
      "grad_norm": 0.4662567973136902,
      "learning_rate": 8.562310912180364e-07,
      "loss": 0.6503,
      "step": 19998
    },
    {
      "epoch": 0.82883666956774,
      "grad_norm": 0.3976411819458008,
      "learning_rate": 8.560238716896681e-07,
      "loss": 0.6711,
      "step": 19999
    },
    {
      "epoch": 0.8288781134734138,
      "grad_norm": 0.3967292904853821,
      "learning_rate": 8.558166521612997e-07,
      "loss": 0.6533,
      "step": 20000
    },
    {
      "epoch": 0.8289195573790874,
      "grad_norm": 0.4101928770542145,
      "learning_rate": 8.556094326329314e-07,
      "loss": 0.6526,
      "step": 20001
    },
    {
      "epoch": 0.828961001284761,
      "grad_norm": 0.41288354992866516,
      "learning_rate": 8.55402213104563e-07,
      "loss": 0.6406,
      "step": 20002
    },
    {
      "epoch": 0.8290024451904348,
      "grad_norm": 0.4233168661594391,
      "learning_rate": 8.551949935761947e-07,
      "loss": 0.7458,
      "step": 20003
    },
    {
      "epoch": 0.8290438890961084,
      "grad_norm": 0.4096802771091461,
      "learning_rate": 8.549877740478263e-07,
      "loss": 0.6472,
      "step": 20004
    },
    {
      "epoch": 0.8290853330017821,
      "grad_norm": 0.41254425048828125,
      "learning_rate": 8.54780554519458e-07,
      "loss": 0.6595,
      "step": 20005
    },
    {
      "epoch": 0.8291267769074557,
      "grad_norm": 0.4419281482696533,
      "learning_rate": 8.545733349910896e-07,
      "loss": 0.6667,
      "step": 20006
    },
    {
      "epoch": 0.8291682208131295,
      "grad_norm": 0.39667999744415283,
      "learning_rate": 8.543661154627213e-07,
      "loss": 0.641,
      "step": 20007
    },
    {
      "epoch": 0.8292096647188031,
      "grad_norm": 0.43128979206085205,
      "learning_rate": 8.541588959343529e-07,
      "loss": 0.671,
      "step": 20008
    },
    {
      "epoch": 0.8292511086244768,
      "grad_norm": 0.4288334846496582,
      "learning_rate": 8.539516764059845e-07,
      "loss": 0.7034,
      "step": 20009
    },
    {
      "epoch": 0.8292925525301504,
      "grad_norm": 0.4451741874217987,
      "learning_rate": 8.537444568776162e-07,
      "loss": 0.6748,
      "step": 20010
    },
    {
      "epoch": 0.8293339964358241,
      "grad_norm": 0.44680196046829224,
      "learning_rate": 8.535372373492478e-07,
      "loss": 0.6995,
      "step": 20011
    },
    {
      "epoch": 0.8293754403414978,
      "grad_norm": 0.41273558139801025,
      "learning_rate": 8.533300178208795e-07,
      "loss": 0.6692,
      "step": 20012
    },
    {
      "epoch": 0.8294168842471714,
      "grad_norm": 0.4450312554836273,
      "learning_rate": 8.531227982925111e-07,
      "loss": 0.7041,
      "step": 20013
    },
    {
      "epoch": 0.8294583281528451,
      "grad_norm": 0.41574862599372864,
      "learning_rate": 8.529155787641428e-07,
      "loss": 0.6361,
      "step": 20014
    },
    {
      "epoch": 0.8294997720585188,
      "grad_norm": 0.45015352964401245,
      "learning_rate": 8.527083592357744e-07,
      "loss": 0.677,
      "step": 20015
    },
    {
      "epoch": 0.8295412159641925,
      "grad_norm": 0.37761011719703674,
      "learning_rate": 8.525011397074061e-07,
      "loss": 0.666,
      "step": 20016
    },
    {
      "epoch": 0.8295826598698661,
      "grad_norm": 0.42317500710487366,
      "learning_rate": 8.522939201790377e-07,
      "loss": 0.7002,
      "step": 20017
    },
    {
      "epoch": 0.8296241037755399,
      "grad_norm": 0.40746697783470154,
      "learning_rate": 8.520867006506694e-07,
      "loss": 0.6836,
      "step": 20018
    },
    {
      "epoch": 0.8296655476812135,
      "grad_norm": 0.4092734456062317,
      "learning_rate": 8.51879481122301e-07,
      "loss": 0.6521,
      "step": 20019
    },
    {
      "epoch": 0.8297069915868871,
      "grad_norm": 0.41898098587989807,
      "learning_rate": 8.516722615939327e-07,
      "loss": 0.6643,
      "step": 20020
    },
    {
      "epoch": 0.8297484354925608,
      "grad_norm": 0.48607155680656433,
      "learning_rate": 8.514650420655643e-07,
      "loss": 0.6827,
      "step": 20021
    },
    {
      "epoch": 0.8297898793982345,
      "grad_norm": 0.3992387652397156,
      "learning_rate": 8.512578225371959e-07,
      "loss": 0.6567,
      "step": 20022
    },
    {
      "epoch": 0.8298313233039082,
      "grad_norm": 0.3944805860519409,
      "learning_rate": 8.510506030088276e-07,
      "loss": 0.6401,
      "step": 20023
    },
    {
      "epoch": 0.8298727672095818,
      "grad_norm": 0.39521360397338867,
      "learning_rate": 8.508433834804593e-07,
      "loss": 0.6459,
      "step": 20024
    },
    {
      "epoch": 0.8299142111152555,
      "grad_norm": 0.4150036871433258,
      "learning_rate": 8.506361639520909e-07,
      "loss": 0.6624,
      "step": 20025
    },
    {
      "epoch": 0.8299556550209292,
      "grad_norm": 0.4565248489379883,
      "learning_rate": 8.504289444237225e-07,
      "loss": 0.6752,
      "step": 20026
    },
    {
      "epoch": 0.8299970989266029,
      "grad_norm": 0.39699459075927734,
      "learning_rate": 8.502217248953542e-07,
      "loss": 0.7241,
      "step": 20027
    },
    {
      "epoch": 0.8300385428322765,
      "grad_norm": 0.3757595717906952,
      "learning_rate": 8.500145053669859e-07,
      "loss": 0.5988,
      "step": 20028
    },
    {
      "epoch": 0.8300799867379501,
      "grad_norm": 0.4404752552509308,
      "learning_rate": 8.498072858386175e-07,
      "loss": 0.661,
      "step": 20029
    },
    {
      "epoch": 0.8301214306436239,
      "grad_norm": 0.4492279291152954,
      "learning_rate": 8.496000663102491e-07,
      "loss": 0.6672,
      "step": 20030
    },
    {
      "epoch": 0.8301628745492975,
      "grad_norm": 0.44739478826522827,
      "learning_rate": 8.493928467818808e-07,
      "loss": 0.6711,
      "step": 20031
    },
    {
      "epoch": 0.8302043184549712,
      "grad_norm": 0.3968082070350647,
      "learning_rate": 8.491856272535124e-07,
      "loss": 0.6556,
      "step": 20032
    },
    {
      "epoch": 0.8302457623606448,
      "grad_norm": 0.41891148686408997,
      "learning_rate": 8.489784077251441e-07,
      "loss": 0.662,
      "step": 20033
    },
    {
      "epoch": 0.8302872062663186,
      "grad_norm": 0.38743114471435547,
      "learning_rate": 8.487711881967757e-07,
      "loss": 0.6343,
      "step": 20034
    },
    {
      "epoch": 0.8303286501719922,
      "grad_norm": 0.4422621428966522,
      "learning_rate": 8.485639686684073e-07,
      "loss": 0.7025,
      "step": 20035
    },
    {
      "epoch": 0.8303700940776659,
      "grad_norm": 0.45171454548835754,
      "learning_rate": 8.48356749140039e-07,
      "loss": 0.6604,
      "step": 20036
    },
    {
      "epoch": 0.8304115379833396,
      "grad_norm": 0.45018911361694336,
      "learning_rate": 8.481495296116707e-07,
      "loss": 0.7222,
      "step": 20037
    },
    {
      "epoch": 0.8304529818890132,
      "grad_norm": 0.3977010250091553,
      "learning_rate": 8.479423100833023e-07,
      "loss": 0.6531,
      "step": 20038
    },
    {
      "epoch": 0.8304944257946869,
      "grad_norm": 0.49777066707611084,
      "learning_rate": 8.477350905549339e-07,
      "loss": 0.6611,
      "step": 20039
    },
    {
      "epoch": 0.8305358697003605,
      "grad_norm": 0.4100944697856903,
      "learning_rate": 8.475278710265656e-07,
      "loss": 0.6552,
      "step": 20040
    },
    {
      "epoch": 0.8305773136060343,
      "grad_norm": 0.3915710151195526,
      "learning_rate": 8.473206514981973e-07,
      "loss": 0.6438,
      "step": 20041
    },
    {
      "epoch": 0.8306187575117079,
      "grad_norm": 0.38771945238113403,
      "learning_rate": 8.471134319698289e-07,
      "loss": 0.6447,
      "step": 20042
    },
    {
      "epoch": 0.8306602014173816,
      "grad_norm": 0.4356440603733063,
      "learning_rate": 8.469062124414605e-07,
      "loss": 0.6483,
      "step": 20043
    },
    {
      "epoch": 0.8307016453230552,
      "grad_norm": 0.4159722328186035,
      "learning_rate": 8.466989929130922e-07,
      "loss": 0.6364,
      "step": 20044
    },
    {
      "epoch": 0.830743089228729,
      "grad_norm": 0.38659438490867615,
      "learning_rate": 8.464917733847239e-07,
      "loss": 0.5859,
      "step": 20045
    },
    {
      "epoch": 0.8307845331344026,
      "grad_norm": 0.45010748505592346,
      "learning_rate": 8.462845538563555e-07,
      "loss": 0.6377,
      "step": 20046
    },
    {
      "epoch": 0.8308259770400762,
      "grad_norm": 0.41263681650161743,
      "learning_rate": 8.460773343279871e-07,
      "loss": 0.6489,
      "step": 20047
    },
    {
      "epoch": 0.83086742094575,
      "grad_norm": 0.44679760932922363,
      "learning_rate": 8.458701147996187e-07,
      "loss": 0.7117,
      "step": 20048
    },
    {
      "epoch": 0.8309088648514236,
      "grad_norm": 0.41662296652793884,
      "learning_rate": 8.456628952712504e-07,
      "loss": 0.6719,
      "step": 20049
    },
    {
      "epoch": 0.8309503087570973,
      "grad_norm": 0.4050852656364441,
      "learning_rate": 8.454556757428821e-07,
      "loss": 0.6472,
      "step": 20050
    },
    {
      "epoch": 0.8309917526627709,
      "grad_norm": 0.42484140396118164,
      "learning_rate": 8.452484562145137e-07,
      "loss": 0.7167,
      "step": 20051
    },
    {
      "epoch": 0.8310331965684447,
      "grad_norm": 0.41925716400146484,
      "learning_rate": 8.450412366861453e-07,
      "loss": 0.6787,
      "step": 20052
    },
    {
      "epoch": 0.8310746404741183,
      "grad_norm": 0.41142305731773376,
      "learning_rate": 8.44834017157777e-07,
      "loss": 0.6399,
      "step": 20053
    },
    {
      "epoch": 0.8311160843797919,
      "grad_norm": 0.42942026257514954,
      "learning_rate": 8.446267976294087e-07,
      "loss": 0.6451,
      "step": 20054
    },
    {
      "epoch": 0.8311575282854656,
      "grad_norm": 0.3716887831687927,
      "learning_rate": 8.444195781010403e-07,
      "loss": 0.6289,
      "step": 20055
    },
    {
      "epoch": 0.8311989721911393,
      "grad_norm": 0.42751744389533997,
      "learning_rate": 8.442123585726719e-07,
      "loss": 0.6757,
      "step": 20056
    },
    {
      "epoch": 0.831240416096813,
      "grad_norm": 0.3912983536720276,
      "learning_rate": 8.440051390443036e-07,
      "loss": 0.6493,
      "step": 20057
    },
    {
      "epoch": 0.8312818600024866,
      "grad_norm": 0.3984086215496063,
      "learning_rate": 8.437979195159353e-07,
      "loss": 0.6641,
      "step": 20058
    },
    {
      "epoch": 0.8313233039081603,
      "grad_norm": 0.44905218482017517,
      "learning_rate": 8.435906999875669e-07,
      "loss": 0.6716,
      "step": 20059
    },
    {
      "epoch": 0.831364747813834,
      "grad_norm": 0.4295237958431244,
      "learning_rate": 8.433834804591985e-07,
      "loss": 0.7192,
      "step": 20060
    },
    {
      "epoch": 0.8314061917195077,
      "grad_norm": 0.41920343041419983,
      "learning_rate": 8.431762609308301e-07,
      "loss": 0.6492,
      "step": 20061
    },
    {
      "epoch": 0.8314476356251813,
      "grad_norm": 0.3962598741054535,
      "learning_rate": 8.429690414024619e-07,
      "loss": 0.7192,
      "step": 20062
    },
    {
      "epoch": 0.8314890795308549,
      "grad_norm": 0.4192592203617096,
      "learning_rate": 8.427618218740935e-07,
      "loss": 0.6656,
      "step": 20063
    },
    {
      "epoch": 0.8315305234365287,
      "grad_norm": 0.4547889530658722,
      "learning_rate": 8.425546023457251e-07,
      "loss": 0.6742,
      "step": 20064
    },
    {
      "epoch": 0.8315719673422023,
      "grad_norm": 0.4208383858203888,
      "learning_rate": 8.423473828173567e-07,
      "loss": 0.6531,
      "step": 20065
    },
    {
      "epoch": 0.831613411247876,
      "grad_norm": 0.41206905245780945,
      "learning_rate": 8.421401632889885e-07,
      "loss": 0.6592,
      "step": 20066
    },
    {
      "epoch": 0.8316548551535496,
      "grad_norm": 0.45992785692214966,
      "learning_rate": 8.419329437606201e-07,
      "loss": 0.6538,
      "step": 20067
    },
    {
      "epoch": 0.8316962990592234,
      "grad_norm": 0.4289086163043976,
      "learning_rate": 8.417257242322517e-07,
      "loss": 0.6842,
      "step": 20068
    },
    {
      "epoch": 0.831737742964897,
      "grad_norm": 0.43026667833328247,
      "learning_rate": 8.415185047038833e-07,
      "loss": 0.6027,
      "step": 20069
    },
    {
      "epoch": 0.8317791868705707,
      "grad_norm": 0.44514214992523193,
      "learning_rate": 8.413112851755149e-07,
      "loss": 0.6627,
      "step": 20070
    },
    {
      "epoch": 0.8318206307762444,
      "grad_norm": 0.37040069699287415,
      "learning_rate": 8.411040656471467e-07,
      "loss": 0.6367,
      "step": 20071
    },
    {
      "epoch": 0.831862074681918,
      "grad_norm": 0.39829742908477783,
      "learning_rate": 8.408968461187783e-07,
      "loss": 0.6499,
      "step": 20072
    },
    {
      "epoch": 0.8319035185875917,
      "grad_norm": 0.430103600025177,
      "learning_rate": 8.406896265904099e-07,
      "loss": 0.686,
      "step": 20073
    },
    {
      "epoch": 0.8319449624932653,
      "grad_norm": 0.40057259798049927,
      "learning_rate": 8.404824070620415e-07,
      "loss": 0.6348,
      "step": 20074
    },
    {
      "epoch": 0.8319864063989391,
      "grad_norm": 0.4294160306453705,
      "learning_rate": 8.402751875336733e-07,
      "loss": 0.6461,
      "step": 20075
    },
    {
      "epoch": 0.8320278503046127,
      "grad_norm": 0.4689786434173584,
      "learning_rate": 8.400679680053049e-07,
      "loss": 0.7097,
      "step": 20076
    },
    {
      "epoch": 0.8320692942102864,
      "grad_norm": 0.45120856165885925,
      "learning_rate": 8.398607484769365e-07,
      "loss": 0.7344,
      "step": 20077
    },
    {
      "epoch": 0.83211073811596,
      "grad_norm": 0.3935043215751648,
      "learning_rate": 8.396535289485681e-07,
      "loss": 0.7222,
      "step": 20078
    },
    {
      "epoch": 0.8321521820216338,
      "grad_norm": 0.3996037542819977,
      "learning_rate": 8.394463094201999e-07,
      "loss": 0.6846,
      "step": 20079
    },
    {
      "epoch": 0.8321936259273074,
      "grad_norm": 0.4432406425476074,
      "learning_rate": 8.392390898918315e-07,
      "loss": 0.6857,
      "step": 20080
    },
    {
      "epoch": 0.832235069832981,
      "grad_norm": 0.40930941700935364,
      "learning_rate": 8.390318703634631e-07,
      "loss": 0.6685,
      "step": 20081
    },
    {
      "epoch": 0.8322765137386547,
      "grad_norm": 0.4179249107837677,
      "learning_rate": 8.388246508350947e-07,
      "loss": 0.6803,
      "step": 20082
    },
    {
      "epoch": 0.8323179576443284,
      "grad_norm": 0.42251846194267273,
      "learning_rate": 8.386174313067263e-07,
      "loss": 0.6898,
      "step": 20083
    },
    {
      "epoch": 0.8323594015500021,
      "grad_norm": 0.4750705659389496,
      "learning_rate": 8.384102117783581e-07,
      "loss": 0.6885,
      "step": 20084
    },
    {
      "epoch": 0.8324008454556757,
      "grad_norm": 0.4333284795284271,
      "learning_rate": 8.382029922499897e-07,
      "loss": 0.6851,
      "step": 20085
    },
    {
      "epoch": 0.8324422893613495,
      "grad_norm": 0.44305646419525146,
      "learning_rate": 8.379957727216213e-07,
      "loss": 0.6731,
      "step": 20086
    },
    {
      "epoch": 0.8324837332670231,
      "grad_norm": 0.41136986017227173,
      "learning_rate": 8.377885531932529e-07,
      "loss": 0.6903,
      "step": 20087
    },
    {
      "epoch": 0.8325251771726968,
      "grad_norm": 0.4479343891143799,
      "learning_rate": 8.375813336648847e-07,
      "loss": 0.6873,
      "step": 20088
    },
    {
      "epoch": 0.8325666210783704,
      "grad_norm": 0.44717758893966675,
      "learning_rate": 8.373741141365163e-07,
      "loss": 0.7314,
      "step": 20089
    },
    {
      "epoch": 0.832608064984044,
      "grad_norm": 0.42782285809516907,
      "learning_rate": 8.371668946081479e-07,
      "loss": 0.7343,
      "step": 20090
    },
    {
      "epoch": 0.8326495088897178,
      "grad_norm": 0.44053348898887634,
      "learning_rate": 8.369596750797795e-07,
      "loss": 0.7009,
      "step": 20091
    },
    {
      "epoch": 0.8326909527953914,
      "grad_norm": 0.45652642846107483,
      "learning_rate": 8.367524555514113e-07,
      "loss": 0.6924,
      "step": 20092
    },
    {
      "epoch": 0.8327323967010651,
      "grad_norm": 0.4076385498046875,
      "learning_rate": 8.365452360230429e-07,
      "loss": 0.6257,
      "step": 20093
    },
    {
      "epoch": 0.8327738406067388,
      "grad_norm": 0.39857017993927,
      "learning_rate": 8.363380164946745e-07,
      "loss": 0.6624,
      "step": 20094
    },
    {
      "epoch": 0.8328152845124125,
      "grad_norm": 0.4346845746040344,
      "learning_rate": 8.361307969663061e-07,
      "loss": 0.673,
      "step": 20095
    },
    {
      "epoch": 0.8328567284180861,
      "grad_norm": 0.4635133743286133,
      "learning_rate": 8.359235774379377e-07,
      "loss": 0.7104,
      "step": 20096
    },
    {
      "epoch": 0.8328981723237598,
      "grad_norm": 0.4417278468608856,
      "learning_rate": 8.357163579095695e-07,
      "loss": 0.6729,
      "step": 20097
    },
    {
      "epoch": 0.8329396162294335,
      "grad_norm": 0.4420102536678314,
      "learning_rate": 8.355091383812011e-07,
      "loss": 0.6832,
      "step": 20098
    },
    {
      "epoch": 0.8329810601351071,
      "grad_norm": 0.41820478439331055,
      "learning_rate": 8.353019188528327e-07,
      "loss": 0.6626,
      "step": 20099
    },
    {
      "epoch": 0.8330225040407808,
      "grad_norm": 0.45711323618888855,
      "learning_rate": 8.350946993244643e-07,
      "loss": 0.7045,
      "step": 20100
    },
    {
      "epoch": 0.8330639479464544,
      "grad_norm": 0.42562970519065857,
      "learning_rate": 8.348874797960961e-07,
      "loss": 0.6833,
      "step": 20101
    },
    {
      "epoch": 0.8331053918521282,
      "grad_norm": 0.43641841411590576,
      "learning_rate": 8.346802602677277e-07,
      "loss": 0.6875,
      "step": 20102
    },
    {
      "epoch": 0.8331468357578018,
      "grad_norm": 0.43513673543930054,
      "learning_rate": 8.344730407393593e-07,
      "loss": 0.6655,
      "step": 20103
    },
    {
      "epoch": 0.8331882796634755,
      "grad_norm": 0.4346787929534912,
      "learning_rate": 8.342658212109909e-07,
      "loss": 0.6555,
      "step": 20104
    },
    {
      "epoch": 0.8332297235691491,
      "grad_norm": 0.4621865153312683,
      "learning_rate": 8.340586016826227e-07,
      "loss": 0.6824,
      "step": 20105
    },
    {
      "epoch": 0.8332711674748229,
      "grad_norm": 0.42973873019218445,
      "learning_rate": 8.338513821542543e-07,
      "loss": 0.6549,
      "step": 20106
    },
    {
      "epoch": 0.8333126113804965,
      "grad_norm": 0.42873603105545044,
      "learning_rate": 8.336441626258859e-07,
      "loss": 0.6488,
      "step": 20107
    },
    {
      "epoch": 0.8333540552861701,
      "grad_norm": 0.449428915977478,
      "learning_rate": 8.334369430975175e-07,
      "loss": 0.6703,
      "step": 20108
    },
    {
      "epoch": 0.8333954991918439,
      "grad_norm": 0.4479316771030426,
      "learning_rate": 8.332297235691491e-07,
      "loss": 0.7274,
      "step": 20109
    },
    {
      "epoch": 0.8334369430975175,
      "grad_norm": 0.4759034216403961,
      "learning_rate": 8.330225040407809e-07,
      "loss": 0.7117,
      "step": 20110
    },
    {
      "epoch": 0.8334783870031912,
      "grad_norm": 0.424565851688385,
      "learning_rate": 8.328152845124125e-07,
      "loss": 0.6406,
      "step": 20111
    },
    {
      "epoch": 0.8335198309088648,
      "grad_norm": 0.4320058822631836,
      "learning_rate": 8.326080649840441e-07,
      "loss": 0.7358,
      "step": 20112
    },
    {
      "epoch": 0.8335612748145386,
      "grad_norm": 0.40586286783218384,
      "learning_rate": 8.324008454556757e-07,
      "loss": 0.62,
      "step": 20113
    },
    {
      "epoch": 0.8336027187202122,
      "grad_norm": 0.4325121343135834,
      "learning_rate": 8.321936259273075e-07,
      "loss": 0.6591,
      "step": 20114
    },
    {
      "epoch": 0.8336441626258858,
      "grad_norm": 0.4304608404636383,
      "learning_rate": 8.319864063989391e-07,
      "loss": 0.7225,
      "step": 20115
    },
    {
      "epoch": 0.8336856065315595,
      "grad_norm": 0.4105498492717743,
      "learning_rate": 8.317791868705707e-07,
      "loss": 0.6417,
      "step": 20116
    },
    {
      "epoch": 0.8337270504372332,
      "grad_norm": 0.45178529620170593,
      "learning_rate": 8.315719673422023e-07,
      "loss": 0.6707,
      "step": 20117
    },
    {
      "epoch": 0.8337684943429069,
      "grad_norm": 0.3989909887313843,
      "learning_rate": 8.313647478138341e-07,
      "loss": 0.66,
      "step": 20118
    },
    {
      "epoch": 0.8338099382485805,
      "grad_norm": 0.40666308999061584,
      "learning_rate": 8.311575282854657e-07,
      "loss": 0.6432,
      "step": 20119
    },
    {
      "epoch": 0.8338513821542543,
      "grad_norm": 0.4271012842655182,
      "learning_rate": 8.309503087570973e-07,
      "loss": 0.6152,
      "step": 20120
    },
    {
      "epoch": 0.8338928260599279,
      "grad_norm": 0.3968338668346405,
      "learning_rate": 8.307430892287289e-07,
      "loss": 0.644,
      "step": 20121
    },
    {
      "epoch": 0.8339342699656016,
      "grad_norm": 0.4402945935726166,
      "learning_rate": 8.305358697003605e-07,
      "loss": 0.7002,
      "step": 20122
    },
    {
      "epoch": 0.8339757138712752,
      "grad_norm": 0.43367862701416016,
      "learning_rate": 8.303286501719923e-07,
      "loss": 0.6829,
      "step": 20123
    },
    {
      "epoch": 0.8340171577769488,
      "grad_norm": 0.41570237278938293,
      "learning_rate": 8.301214306436239e-07,
      "loss": 0.6472,
      "step": 20124
    },
    {
      "epoch": 0.8340586016826226,
      "grad_norm": 0.4177684187889099,
      "learning_rate": 8.299142111152555e-07,
      "loss": 0.6934,
      "step": 20125
    },
    {
      "epoch": 0.8341000455882962,
      "grad_norm": 0.4139949381351471,
      "learning_rate": 8.297069915868871e-07,
      "loss": 0.6995,
      "step": 20126
    },
    {
      "epoch": 0.8341414894939699,
      "grad_norm": 0.4132566452026367,
      "learning_rate": 8.294997720585189e-07,
      "loss": 0.72,
      "step": 20127
    },
    {
      "epoch": 0.8341829333996436,
      "grad_norm": 0.42942988872528076,
      "learning_rate": 8.292925525301505e-07,
      "loss": 0.6899,
      "step": 20128
    },
    {
      "epoch": 0.8342243773053173,
      "grad_norm": 0.44176384806632996,
      "learning_rate": 8.290853330017821e-07,
      "loss": 0.6641,
      "step": 20129
    },
    {
      "epoch": 0.8342658212109909,
      "grad_norm": 0.40067216753959656,
      "learning_rate": 8.288781134734137e-07,
      "loss": 0.6798,
      "step": 20130
    },
    {
      "epoch": 0.8343072651166646,
      "grad_norm": 0.4535911977291107,
      "learning_rate": 8.286708939450455e-07,
      "loss": 0.6865,
      "step": 20131
    },
    {
      "epoch": 0.8343487090223383,
      "grad_norm": 0.40589749813079834,
      "learning_rate": 8.284636744166771e-07,
      "loss": 0.6534,
      "step": 20132
    },
    {
      "epoch": 0.8343901529280119,
      "grad_norm": 0.44679269194602966,
      "learning_rate": 8.282564548883087e-07,
      "loss": 0.6257,
      "step": 20133
    },
    {
      "epoch": 0.8344315968336856,
      "grad_norm": 0.45142728090286255,
      "learning_rate": 8.280492353599403e-07,
      "loss": 0.6528,
      "step": 20134
    },
    {
      "epoch": 0.8344730407393592,
      "grad_norm": 0.4098823368549347,
      "learning_rate": 8.278420158315719e-07,
      "loss": 0.6643,
      "step": 20135
    },
    {
      "epoch": 0.834514484645033,
      "grad_norm": 0.437261164188385,
      "learning_rate": 8.276347963032037e-07,
      "loss": 0.666,
      "step": 20136
    },
    {
      "epoch": 0.8345559285507066,
      "grad_norm": 0.41778817772865295,
      "learning_rate": 8.274275767748353e-07,
      "loss": 0.6316,
      "step": 20137
    },
    {
      "epoch": 0.8345973724563803,
      "grad_norm": 0.4401358962059021,
      "learning_rate": 8.272203572464669e-07,
      "loss": 0.7249,
      "step": 20138
    },
    {
      "epoch": 0.834638816362054,
      "grad_norm": 0.4213072657585144,
      "learning_rate": 8.270131377180985e-07,
      "loss": 0.6653,
      "step": 20139
    },
    {
      "epoch": 0.8346802602677277,
      "grad_norm": 0.387548565864563,
      "learning_rate": 8.268059181897303e-07,
      "loss": 0.6355,
      "step": 20140
    },
    {
      "epoch": 0.8347217041734013,
      "grad_norm": 0.42852258682250977,
      "learning_rate": 8.265986986613619e-07,
      "loss": 0.6786,
      "step": 20141
    },
    {
      "epoch": 0.8347631480790749,
      "grad_norm": 0.3876879811286926,
      "learning_rate": 8.263914791329935e-07,
      "loss": 0.6693,
      "step": 20142
    },
    {
      "epoch": 0.8348045919847487,
      "grad_norm": 0.42805978655815125,
      "learning_rate": 8.261842596046251e-07,
      "loss": 0.6583,
      "step": 20143
    },
    {
      "epoch": 0.8348460358904223,
      "grad_norm": 0.46535688638687134,
      "learning_rate": 8.259770400762568e-07,
      "loss": 0.6812,
      "step": 20144
    },
    {
      "epoch": 0.834887479796096,
      "grad_norm": 0.4829835891723633,
      "learning_rate": 8.257698205478885e-07,
      "loss": 0.6481,
      "step": 20145
    },
    {
      "epoch": 0.8349289237017696,
      "grad_norm": 0.4245811998844147,
      "learning_rate": 8.255626010195201e-07,
      "loss": 0.7096,
      "step": 20146
    },
    {
      "epoch": 0.8349703676074434,
      "grad_norm": 0.4103786051273346,
      "learning_rate": 8.253553814911517e-07,
      "loss": 0.634,
      "step": 20147
    },
    {
      "epoch": 0.835011811513117,
      "grad_norm": 0.40655213594436646,
      "learning_rate": 8.251481619627833e-07,
      "loss": 0.6833,
      "step": 20148
    },
    {
      "epoch": 0.8350532554187907,
      "grad_norm": 0.4138886034488678,
      "learning_rate": 8.249409424344151e-07,
      "loss": 0.6106,
      "step": 20149
    },
    {
      "epoch": 0.8350946993244643,
      "grad_norm": 0.4149421751499176,
      "learning_rate": 8.247337229060467e-07,
      "loss": 0.6593,
      "step": 20150
    },
    {
      "epoch": 0.835136143230138,
      "grad_norm": 0.3820747435092926,
      "learning_rate": 8.245265033776783e-07,
      "loss": 0.6034,
      "step": 20151
    },
    {
      "epoch": 0.8351775871358117,
      "grad_norm": 0.3988364338874817,
      "learning_rate": 8.243192838493099e-07,
      "loss": 0.6565,
      "step": 20152
    },
    {
      "epoch": 0.8352190310414853,
      "grad_norm": 0.4235917925834656,
      "learning_rate": 8.241120643209417e-07,
      "loss": 0.6433,
      "step": 20153
    },
    {
      "epoch": 0.835260474947159,
      "grad_norm": 0.4080042839050293,
      "learning_rate": 8.239048447925733e-07,
      "loss": 0.6646,
      "step": 20154
    },
    {
      "epoch": 0.8353019188528327,
      "grad_norm": 0.42446550726890564,
      "learning_rate": 8.236976252642049e-07,
      "loss": 0.6821,
      "step": 20155
    },
    {
      "epoch": 0.8353433627585064,
      "grad_norm": 0.49676021933555603,
      "learning_rate": 8.234904057358365e-07,
      "loss": 0.7761,
      "step": 20156
    },
    {
      "epoch": 0.83538480666418,
      "grad_norm": 0.4220278561115265,
      "learning_rate": 8.232831862074682e-07,
      "loss": 0.6039,
      "step": 20157
    },
    {
      "epoch": 0.8354262505698538,
      "grad_norm": 0.43039804697036743,
      "learning_rate": 8.230759666790999e-07,
      "loss": 0.6738,
      "step": 20158
    },
    {
      "epoch": 0.8354676944755274,
      "grad_norm": 0.4140724241733551,
      "learning_rate": 8.228687471507315e-07,
      "loss": 0.6416,
      "step": 20159
    },
    {
      "epoch": 0.835509138381201,
      "grad_norm": 0.4029494822025299,
      "learning_rate": 8.226615276223631e-07,
      "loss": 0.6484,
      "step": 20160
    },
    {
      "epoch": 0.8355505822868747,
      "grad_norm": 0.4469091594219208,
      "learning_rate": 8.224543080939948e-07,
      "loss": 0.7079,
      "step": 20161
    },
    {
      "epoch": 0.8355920261925484,
      "grad_norm": 0.46666160225868225,
      "learning_rate": 8.222470885656265e-07,
      "loss": 0.6572,
      "step": 20162
    },
    {
      "epoch": 0.8356334700982221,
      "grad_norm": 0.4139482378959656,
      "learning_rate": 8.220398690372581e-07,
      "loss": 0.6355,
      "step": 20163
    },
    {
      "epoch": 0.8356749140038957,
      "grad_norm": 0.40282413363456726,
      "learning_rate": 8.218326495088897e-07,
      "loss": 0.6281,
      "step": 20164
    },
    {
      "epoch": 0.8357163579095694,
      "grad_norm": 0.43178045749664307,
      "learning_rate": 8.216254299805213e-07,
      "loss": 0.6562,
      "step": 20165
    },
    {
      "epoch": 0.8357578018152431,
      "grad_norm": 0.41251131892204285,
      "learning_rate": 8.214182104521531e-07,
      "loss": 0.6219,
      "step": 20166
    },
    {
      "epoch": 0.8357992457209168,
      "grad_norm": 0.40567490458488464,
      "learning_rate": 8.212109909237847e-07,
      "loss": 0.6703,
      "step": 20167
    },
    {
      "epoch": 0.8358406896265904,
      "grad_norm": 0.4112817347049713,
      "learning_rate": 8.210037713954163e-07,
      "loss": 0.6617,
      "step": 20168
    },
    {
      "epoch": 0.835882133532264,
      "grad_norm": 0.3748422861099243,
      "learning_rate": 8.207965518670479e-07,
      "loss": 0.6201,
      "step": 20169
    },
    {
      "epoch": 0.8359235774379378,
      "grad_norm": 0.4309268295764923,
      "learning_rate": 8.205893323386796e-07,
      "loss": 0.6644,
      "step": 20170
    },
    {
      "epoch": 0.8359650213436114,
      "grad_norm": 0.4265573024749756,
      "learning_rate": 8.203821128103113e-07,
      "loss": 0.6501,
      "step": 20171
    },
    {
      "epoch": 0.8360064652492851,
      "grad_norm": 0.4016571044921875,
      "learning_rate": 8.201748932819429e-07,
      "loss": 0.6802,
      "step": 20172
    },
    {
      "epoch": 0.8360479091549587,
      "grad_norm": 0.43196430802345276,
      "learning_rate": 8.199676737535745e-07,
      "loss": 0.6606,
      "step": 20173
    },
    {
      "epoch": 0.8360893530606325,
      "grad_norm": 0.41298431158065796,
      "learning_rate": 8.197604542252062e-07,
      "loss": 0.6646,
      "step": 20174
    },
    {
      "epoch": 0.8361307969663061,
      "grad_norm": 0.4271107614040375,
      "learning_rate": 8.195532346968379e-07,
      "loss": 0.6676,
      "step": 20175
    },
    {
      "epoch": 0.8361722408719797,
      "grad_norm": 0.43281686305999756,
      "learning_rate": 8.193460151684695e-07,
      "loss": 0.6465,
      "step": 20176
    },
    {
      "epoch": 0.8362136847776535,
      "grad_norm": 0.4030953645706177,
      "learning_rate": 8.191387956401011e-07,
      "loss": 0.6869,
      "step": 20177
    },
    {
      "epoch": 0.8362551286833271,
      "grad_norm": 0.46221786737442017,
      "learning_rate": 8.189315761117328e-07,
      "loss": 0.6913,
      "step": 20178
    },
    {
      "epoch": 0.8362965725890008,
      "grad_norm": 0.45367565751075745,
      "learning_rate": 8.187243565833645e-07,
      "loss": 0.699,
      "step": 20179
    },
    {
      "epoch": 0.8363380164946744,
      "grad_norm": 0.4214944839477539,
      "learning_rate": 8.185171370549961e-07,
      "loss": 0.639,
      "step": 20180
    },
    {
      "epoch": 0.8363794604003482,
      "grad_norm": 0.4334227442741394,
      "learning_rate": 8.183099175266277e-07,
      "loss": 0.6277,
      "step": 20181
    },
    {
      "epoch": 0.8364209043060218,
      "grad_norm": 0.3942674994468689,
      "learning_rate": 8.181026979982593e-07,
      "loss": 0.6659,
      "step": 20182
    },
    {
      "epoch": 0.8364623482116955,
      "grad_norm": 0.4073808789253235,
      "learning_rate": 8.17895478469891e-07,
      "loss": 0.6912,
      "step": 20183
    },
    {
      "epoch": 0.8365037921173691,
      "grad_norm": 0.40669405460357666,
      "learning_rate": 8.176882589415227e-07,
      "loss": 0.6094,
      "step": 20184
    },
    {
      "epoch": 0.8365452360230428,
      "grad_norm": 0.44962188601493835,
      "learning_rate": 8.174810394131543e-07,
      "loss": 0.6646,
      "step": 20185
    },
    {
      "epoch": 0.8365866799287165,
      "grad_norm": 0.4204114079475403,
      "learning_rate": 8.172738198847859e-07,
      "loss": 0.6572,
      "step": 20186
    },
    {
      "epoch": 0.8366281238343901,
      "grad_norm": 0.4026535153388977,
      "learning_rate": 8.170666003564176e-07,
      "loss": 0.639,
      "step": 20187
    },
    {
      "epoch": 0.8366695677400638,
      "grad_norm": 0.4280976355075836,
      "learning_rate": 8.168593808280493e-07,
      "loss": 0.6508,
      "step": 20188
    },
    {
      "epoch": 0.8367110116457375,
      "grad_norm": 0.427510142326355,
      "learning_rate": 8.166521612996809e-07,
      "loss": 0.6372,
      "step": 20189
    },
    {
      "epoch": 0.8367524555514112,
      "grad_norm": 0.39416950941085815,
      "learning_rate": 8.164449417713125e-07,
      "loss": 0.6451,
      "step": 20190
    },
    {
      "epoch": 0.8367938994570848,
      "grad_norm": 0.38611263036727905,
      "learning_rate": 8.162377222429442e-07,
      "loss": 0.6241,
      "step": 20191
    },
    {
      "epoch": 0.8368353433627586,
      "grad_norm": 0.4174487292766571,
      "learning_rate": 8.16030502714576e-07,
      "loss": 0.6375,
      "step": 20192
    },
    {
      "epoch": 0.8368767872684322,
      "grad_norm": 0.4115034341812134,
      "learning_rate": 8.158232831862075e-07,
      "loss": 0.6537,
      "step": 20193
    },
    {
      "epoch": 0.8369182311741058,
      "grad_norm": 0.43263453245162964,
      "learning_rate": 8.156160636578391e-07,
      "loss": 0.6384,
      "step": 20194
    },
    {
      "epoch": 0.8369596750797795,
      "grad_norm": 0.4312496483325958,
      "learning_rate": 8.154088441294708e-07,
      "loss": 0.6571,
      "step": 20195
    },
    {
      "epoch": 0.8370011189854532,
      "grad_norm": 0.4276101887226105,
      "learning_rate": 8.152016246011024e-07,
      "loss": 0.6874,
      "step": 20196
    },
    {
      "epoch": 0.8370425628911269,
      "grad_norm": 0.4064944088459015,
      "learning_rate": 8.149944050727341e-07,
      "loss": 0.6412,
      "step": 20197
    },
    {
      "epoch": 0.8370840067968005,
      "grad_norm": 0.42428916692733765,
      "learning_rate": 8.147871855443657e-07,
      "loss": 0.6586,
      "step": 20198
    },
    {
      "epoch": 0.8371254507024742,
      "grad_norm": 0.41180431842803955,
      "learning_rate": 8.145799660159974e-07,
      "loss": 0.6438,
      "step": 20199
    },
    {
      "epoch": 0.8371668946081479,
      "grad_norm": 0.4133630692958832,
      "learning_rate": 8.14372746487629e-07,
      "loss": 0.6755,
      "step": 20200
    },
    {
      "epoch": 0.8372083385138216,
      "grad_norm": 0.4471549391746521,
      "learning_rate": 8.141655269592608e-07,
      "loss": 0.7253,
      "step": 20201
    },
    {
      "epoch": 0.8372497824194952,
      "grad_norm": 0.4245521128177643,
      "learning_rate": 8.139583074308923e-07,
      "loss": 0.6846,
      "step": 20202
    },
    {
      "epoch": 0.8372912263251688,
      "grad_norm": 0.4138719141483307,
      "learning_rate": 8.137510879025239e-07,
      "loss": 0.6318,
      "step": 20203
    },
    {
      "epoch": 0.8373326702308426,
      "grad_norm": 0.389619916677475,
      "learning_rate": 8.135438683741556e-07,
      "loss": 0.6403,
      "step": 20204
    },
    {
      "epoch": 0.8373741141365162,
      "grad_norm": 0.4178262948989868,
      "learning_rate": 8.133366488457872e-07,
      "loss": 0.6936,
      "step": 20205
    },
    {
      "epoch": 0.8374155580421899,
      "grad_norm": 0.4229121208190918,
      "learning_rate": 8.13129429317419e-07,
      "loss": 0.6919,
      "step": 20206
    },
    {
      "epoch": 0.8374570019478635,
      "grad_norm": 0.46110671758651733,
      "learning_rate": 8.129222097890505e-07,
      "loss": 0.6426,
      "step": 20207
    },
    {
      "epoch": 0.8374984458535373,
      "grad_norm": 0.4340898096561432,
      "learning_rate": 8.127149902606822e-07,
      "loss": 0.6816,
      "step": 20208
    },
    {
      "epoch": 0.8375398897592109,
      "grad_norm": 0.46029648184776306,
      "learning_rate": 8.125077707323138e-07,
      "loss": 0.7317,
      "step": 20209
    },
    {
      "epoch": 0.8375813336648846,
      "grad_norm": 0.4133777320384979,
      "learning_rate": 8.123005512039456e-07,
      "loss": 0.6763,
      "step": 20210
    },
    {
      "epoch": 0.8376227775705583,
      "grad_norm": 0.39068976044654846,
      "learning_rate": 8.120933316755771e-07,
      "loss": 0.6766,
      "step": 20211
    },
    {
      "epoch": 0.8376642214762319,
      "grad_norm": 0.4317997694015503,
      "learning_rate": 8.118861121472089e-07,
      "loss": 0.6823,
      "step": 20212
    },
    {
      "epoch": 0.8377056653819056,
      "grad_norm": 0.4342253506183624,
      "learning_rate": 8.116788926188404e-07,
      "loss": 0.6946,
      "step": 20213
    },
    {
      "epoch": 0.8377471092875792,
      "grad_norm": 0.45126211643218994,
      "learning_rate": 8.114716730904722e-07,
      "loss": 0.686,
      "step": 20214
    },
    {
      "epoch": 0.837788553193253,
      "grad_norm": 0.44056689739227295,
      "learning_rate": 8.112644535621037e-07,
      "loss": 0.6978,
      "step": 20215
    },
    {
      "epoch": 0.8378299970989266,
      "grad_norm": 0.38186904788017273,
      "learning_rate": 8.110572340337355e-07,
      "loss": 0.6311,
      "step": 20216
    },
    {
      "epoch": 0.8378714410046003,
      "grad_norm": 0.4326772093772888,
      "learning_rate": 8.10850014505367e-07,
      "loss": 0.703,
      "step": 20217
    },
    {
      "epoch": 0.8379128849102739,
      "grad_norm": 0.48534396290779114,
      "learning_rate": 8.106427949769986e-07,
      "loss": 0.6505,
      "step": 20218
    },
    {
      "epoch": 0.8379543288159477,
      "grad_norm": 0.4277802109718323,
      "learning_rate": 8.104355754486304e-07,
      "loss": 0.6829,
      "step": 20219
    },
    {
      "epoch": 0.8379957727216213,
      "grad_norm": 0.4214552938938141,
      "learning_rate": 8.10228355920262e-07,
      "loss": 0.6577,
      "step": 20220
    },
    {
      "epoch": 0.8380372166272949,
      "grad_norm": 0.4311408996582031,
      "learning_rate": 8.100211363918937e-07,
      "loss": 0.6984,
      "step": 20221
    },
    {
      "epoch": 0.8380786605329686,
      "grad_norm": 0.3997269570827484,
      "learning_rate": 8.098139168635252e-07,
      "loss": 0.6812,
      "step": 20222
    },
    {
      "epoch": 0.8381201044386423,
      "grad_norm": 0.40064096450805664,
      "learning_rate": 8.09606697335157e-07,
      "loss": 0.6853,
      "step": 20223
    },
    {
      "epoch": 0.838161548344316,
      "grad_norm": 0.40908172726631165,
      "learning_rate": 8.093994778067885e-07,
      "loss": 0.6423,
      "step": 20224
    },
    {
      "epoch": 0.8382029922499896,
      "grad_norm": 0.425047367811203,
      "learning_rate": 8.091922582784203e-07,
      "loss": 0.6827,
      "step": 20225
    },
    {
      "epoch": 0.8382444361556634,
      "grad_norm": 0.4360681176185608,
      "learning_rate": 8.089850387500519e-07,
      "loss": 0.6587,
      "step": 20226
    },
    {
      "epoch": 0.838285880061337,
      "grad_norm": 0.43311211466789246,
      "learning_rate": 8.087778192216836e-07,
      "loss": 0.6858,
      "step": 20227
    },
    {
      "epoch": 0.8383273239670106,
      "grad_norm": 0.415152907371521,
      "learning_rate": 8.085705996933152e-07,
      "loss": 0.6395,
      "step": 20228
    },
    {
      "epoch": 0.8383687678726843,
      "grad_norm": 0.39357760548591614,
      "learning_rate": 8.083633801649469e-07,
      "loss": 0.6826,
      "step": 20229
    },
    {
      "epoch": 0.838410211778358,
      "grad_norm": 0.4715930223464966,
      "learning_rate": 8.081561606365785e-07,
      "loss": 0.7057,
      "step": 20230
    },
    {
      "epoch": 0.8384516556840317,
      "grad_norm": 0.4347377419471741,
      "learning_rate": 8.0794894110821e-07,
      "loss": 0.6274,
      "step": 20231
    },
    {
      "epoch": 0.8384930995897053,
      "grad_norm": 0.3943631052970886,
      "learning_rate": 8.077417215798418e-07,
      "loss": 0.6631,
      "step": 20232
    },
    {
      "epoch": 0.838534543495379,
      "grad_norm": 0.474553644657135,
      "learning_rate": 8.075345020514735e-07,
      "loss": 0.6803,
      "step": 20233
    },
    {
      "epoch": 0.8385759874010527,
      "grad_norm": 0.436998188495636,
      "learning_rate": 8.073272825231051e-07,
      "loss": 0.6812,
      "step": 20234
    },
    {
      "epoch": 0.8386174313067264,
      "grad_norm": 0.4583664536476135,
      "learning_rate": 8.071200629947367e-07,
      "loss": 0.6865,
      "step": 20235
    },
    {
      "epoch": 0.8386588752124,
      "grad_norm": 0.43711233139038086,
      "learning_rate": 8.069128434663684e-07,
      "loss": 0.6471,
      "step": 20236
    },
    {
      "epoch": 0.8387003191180736,
      "grad_norm": 0.4086623191833496,
      "learning_rate": 8.067056239380001e-07,
      "loss": 0.6317,
      "step": 20237
    },
    {
      "epoch": 0.8387417630237474,
      "grad_norm": 0.43112263083457947,
      "learning_rate": 8.064984044096317e-07,
      "loss": 0.676,
      "step": 20238
    },
    {
      "epoch": 0.838783206929421,
      "grad_norm": 0.4171840250492096,
      "learning_rate": 8.062911848812633e-07,
      "loss": 0.661,
      "step": 20239
    },
    {
      "epoch": 0.8388246508350947,
      "grad_norm": 0.4216444194316864,
      "learning_rate": 8.06083965352895e-07,
      "loss": 0.6517,
      "step": 20240
    },
    {
      "epoch": 0.8388660947407683,
      "grad_norm": 0.4085068702697754,
      "learning_rate": 8.058767458245266e-07,
      "loss": 0.678,
      "step": 20241
    },
    {
      "epoch": 0.8389075386464421,
      "grad_norm": 0.4424096345901489,
      "learning_rate": 8.056695262961583e-07,
      "loss": 0.7021,
      "step": 20242
    },
    {
      "epoch": 0.8389489825521157,
      "grad_norm": 0.4305326044559479,
      "learning_rate": 8.054623067677899e-07,
      "loss": 0.6682,
      "step": 20243
    },
    {
      "epoch": 0.8389904264577894,
      "grad_norm": 0.4179757833480835,
      "learning_rate": 8.052550872394215e-07,
      "loss": 0.7206,
      "step": 20244
    },
    {
      "epoch": 0.839031870363463,
      "grad_norm": 0.4429771900177002,
      "learning_rate": 8.050478677110532e-07,
      "loss": 0.6875,
      "step": 20245
    },
    {
      "epoch": 0.8390733142691367,
      "grad_norm": 0.40724819898605347,
      "learning_rate": 8.048406481826849e-07,
      "loss": 0.6658,
      "step": 20246
    },
    {
      "epoch": 0.8391147581748104,
      "grad_norm": 0.40812182426452637,
      "learning_rate": 8.046334286543165e-07,
      "loss": 0.6858,
      "step": 20247
    },
    {
      "epoch": 0.839156202080484,
      "grad_norm": 0.4145753085613251,
      "learning_rate": 8.04426209125948e-07,
      "loss": 0.6987,
      "step": 20248
    },
    {
      "epoch": 0.8391976459861578,
      "grad_norm": 0.4020088315010071,
      "learning_rate": 8.042189895975798e-07,
      "loss": 0.6532,
      "step": 20249
    },
    {
      "epoch": 0.8392390898918314,
      "grad_norm": 0.4070190489292145,
      "learning_rate": 8.040117700692115e-07,
      "loss": 0.6599,
      "step": 20250
    },
    {
      "epoch": 0.8392805337975051,
      "grad_norm": 0.40536993741989136,
      "learning_rate": 8.038045505408431e-07,
      "loss": 0.6868,
      "step": 20251
    },
    {
      "epoch": 0.8393219777031787,
      "grad_norm": 0.45047906041145325,
      "learning_rate": 8.035973310124747e-07,
      "loss": 0.7283,
      "step": 20252
    },
    {
      "epoch": 0.8393634216088525,
      "grad_norm": 0.4204121530056,
      "learning_rate": 8.033901114841064e-07,
      "loss": 0.6484,
      "step": 20253
    },
    {
      "epoch": 0.8394048655145261,
      "grad_norm": 0.39317265152931213,
      "learning_rate": 8.031828919557381e-07,
      "loss": 0.626,
      "step": 20254
    },
    {
      "epoch": 0.8394463094201997,
      "grad_norm": 0.3943878412246704,
      "learning_rate": 8.029756724273697e-07,
      "loss": 0.5958,
      "step": 20255
    },
    {
      "epoch": 0.8394877533258734,
      "grad_norm": 0.4100732207298279,
      "learning_rate": 8.027684528990013e-07,
      "loss": 0.6348,
      "step": 20256
    },
    {
      "epoch": 0.8395291972315471,
      "grad_norm": 0.438375860452652,
      "learning_rate": 8.025612333706329e-07,
      "loss": 0.7015,
      "step": 20257
    },
    {
      "epoch": 0.8395706411372208,
      "grad_norm": 0.4063936471939087,
      "learning_rate": 8.023540138422646e-07,
      "loss": 0.645,
      "step": 20258
    },
    {
      "epoch": 0.8396120850428944,
      "grad_norm": 0.40298527479171753,
      "learning_rate": 8.021467943138963e-07,
      "loss": 0.6281,
      "step": 20259
    },
    {
      "epoch": 0.8396535289485682,
      "grad_norm": 0.4144369661808014,
      "learning_rate": 8.019395747855279e-07,
      "loss": 0.6605,
      "step": 20260
    },
    {
      "epoch": 0.8396949728542418,
      "grad_norm": 0.44305869936943054,
      "learning_rate": 8.017323552571595e-07,
      "loss": 0.6591,
      "step": 20261
    },
    {
      "epoch": 0.8397364167599155,
      "grad_norm": 0.40938323736190796,
      "learning_rate": 8.015251357287912e-07,
      "loss": 0.6613,
      "step": 20262
    },
    {
      "epoch": 0.8397778606655891,
      "grad_norm": 0.3989787697792053,
      "learning_rate": 8.013179162004229e-07,
      "loss": 0.6501,
      "step": 20263
    },
    {
      "epoch": 0.8398193045712627,
      "grad_norm": 0.3831969201564789,
      "learning_rate": 8.011106966720545e-07,
      "loss": 0.6719,
      "step": 20264
    },
    {
      "epoch": 0.8398607484769365,
      "grad_norm": 0.4422278106212616,
      "learning_rate": 8.009034771436861e-07,
      "loss": 0.6819,
      "step": 20265
    },
    {
      "epoch": 0.8399021923826101,
      "grad_norm": 0.41532623767852783,
      "learning_rate": 8.006962576153178e-07,
      "loss": 0.6451,
      "step": 20266
    },
    {
      "epoch": 0.8399436362882838,
      "grad_norm": 0.40918755531311035,
      "learning_rate": 8.004890380869495e-07,
      "loss": 0.699,
      "step": 20267
    },
    {
      "epoch": 0.8399850801939575,
      "grad_norm": 0.3815593421459198,
      "learning_rate": 8.002818185585811e-07,
      "loss": 0.7112,
      "step": 20268
    },
    {
      "epoch": 0.8400265240996312,
      "grad_norm": 0.44708389043807983,
      "learning_rate": 8.000745990302127e-07,
      "loss": 0.7285,
      "step": 20269
    },
    {
      "epoch": 0.8400679680053048,
      "grad_norm": 0.41335397958755493,
      "learning_rate": 7.998673795018443e-07,
      "loss": 0.6503,
      "step": 20270
    },
    {
      "epoch": 0.8401094119109785,
      "grad_norm": 0.3990667164325714,
      "learning_rate": 7.996601599734761e-07,
      "loss": 0.6567,
      "step": 20271
    },
    {
      "epoch": 0.8401508558166522,
      "grad_norm": 0.4109889566898346,
      "learning_rate": 7.994529404451077e-07,
      "loss": 0.6465,
      "step": 20272
    },
    {
      "epoch": 0.8401922997223258,
      "grad_norm": 0.41938143968582153,
      "learning_rate": 7.992457209167393e-07,
      "loss": 0.6226,
      "step": 20273
    },
    {
      "epoch": 0.8402337436279995,
      "grad_norm": 0.38808903098106384,
      "learning_rate": 7.990385013883709e-07,
      "loss": 0.604,
      "step": 20274
    },
    {
      "epoch": 0.8402751875336731,
      "grad_norm": 0.4040178954601288,
      "learning_rate": 7.988312818600026e-07,
      "loss": 0.6128,
      "step": 20275
    },
    {
      "epoch": 0.8403166314393469,
      "grad_norm": 0.44610920548439026,
      "learning_rate": 7.986240623316343e-07,
      "loss": 0.6892,
      "step": 20276
    },
    {
      "epoch": 0.8403580753450205,
      "grad_norm": 0.4109758138656616,
      "learning_rate": 7.984168428032659e-07,
      "loss": 0.6754,
      "step": 20277
    },
    {
      "epoch": 0.8403995192506942,
      "grad_norm": 0.4390300214290619,
      "learning_rate": 7.982096232748975e-07,
      "loss": 0.7388,
      "step": 20278
    },
    {
      "epoch": 0.8404409631563678,
      "grad_norm": 0.42315617203712463,
      "learning_rate": 7.980024037465291e-07,
      "loss": 0.6821,
      "step": 20279
    },
    {
      "epoch": 0.8404824070620416,
      "grad_norm": 0.45524272322654724,
      "learning_rate": 7.977951842181609e-07,
      "loss": 0.6749,
      "step": 20280
    },
    {
      "epoch": 0.8405238509677152,
      "grad_norm": 0.43759286403656006,
      "learning_rate": 7.975879646897925e-07,
      "loss": 0.751,
      "step": 20281
    },
    {
      "epoch": 0.8405652948733888,
      "grad_norm": 0.4256024956703186,
      "learning_rate": 7.973807451614241e-07,
      "loss": 0.6589,
      "step": 20282
    },
    {
      "epoch": 0.8406067387790626,
      "grad_norm": 0.4395647943019867,
      "learning_rate": 7.971735256330557e-07,
      "loss": 0.6945,
      "step": 20283
    },
    {
      "epoch": 0.8406481826847362,
      "grad_norm": 0.4106292128562927,
      "learning_rate": 7.969663061046875e-07,
      "loss": 0.6586,
      "step": 20284
    },
    {
      "epoch": 0.8406896265904099,
      "grad_norm": 0.3862418830394745,
      "learning_rate": 7.967590865763191e-07,
      "loss": 0.6754,
      "step": 20285
    },
    {
      "epoch": 0.8407310704960835,
      "grad_norm": 0.4498670697212219,
      "learning_rate": 7.965518670479507e-07,
      "loss": 0.6997,
      "step": 20286
    },
    {
      "epoch": 0.8407725144017573,
      "grad_norm": 0.4008539915084839,
      "learning_rate": 7.963446475195823e-07,
      "loss": 0.6748,
      "step": 20287
    },
    {
      "epoch": 0.8408139583074309,
      "grad_norm": 0.4114227294921875,
      "learning_rate": 7.961374279912141e-07,
      "loss": 0.6827,
      "step": 20288
    },
    {
      "epoch": 0.8408554022131045,
      "grad_norm": 0.4190865457057953,
      "learning_rate": 7.959302084628457e-07,
      "loss": 0.652,
      "step": 20289
    },
    {
      "epoch": 0.8408968461187782,
      "grad_norm": 0.469209760427475,
      "learning_rate": 7.957229889344773e-07,
      "loss": 0.72,
      "step": 20290
    },
    {
      "epoch": 0.8409382900244519,
      "grad_norm": 0.40231853723526,
      "learning_rate": 7.955157694061089e-07,
      "loss": 0.6794,
      "step": 20291
    },
    {
      "epoch": 0.8409797339301256,
      "grad_norm": 0.4166019856929779,
      "learning_rate": 7.953085498777405e-07,
      "loss": 0.6809,
      "step": 20292
    },
    {
      "epoch": 0.8410211778357992,
      "grad_norm": 0.3955361843109131,
      "learning_rate": 7.951013303493723e-07,
      "loss": 0.6364,
      "step": 20293
    },
    {
      "epoch": 0.841062621741473,
      "grad_norm": 0.44638872146606445,
      "learning_rate": 7.948941108210039e-07,
      "loss": 0.687,
      "step": 20294
    },
    {
      "epoch": 0.8411040656471466,
      "grad_norm": 0.4024045169353485,
      "learning_rate": 7.946868912926355e-07,
      "loss": 0.6301,
      "step": 20295
    },
    {
      "epoch": 0.8411455095528203,
      "grad_norm": 0.42070427536964417,
      "learning_rate": 7.944796717642671e-07,
      "loss": 0.6482,
      "step": 20296
    },
    {
      "epoch": 0.8411869534584939,
      "grad_norm": 0.39480817317962646,
      "learning_rate": 7.942724522358989e-07,
      "loss": 0.6609,
      "step": 20297
    },
    {
      "epoch": 0.8412283973641675,
      "grad_norm": 0.4168970286846161,
      "learning_rate": 7.940652327075305e-07,
      "loss": 0.6888,
      "step": 20298
    },
    {
      "epoch": 0.8412698412698413,
      "grad_norm": 0.43357062339782715,
      "learning_rate": 7.938580131791621e-07,
      "loss": 0.6753,
      "step": 20299
    },
    {
      "epoch": 0.8413112851755149,
      "grad_norm": 0.4303136169910431,
      "learning_rate": 7.936507936507937e-07,
      "loss": 0.678,
      "step": 20300
    },
    {
      "epoch": 0.8413527290811886,
      "grad_norm": 0.4187089204788208,
      "learning_rate": 7.934435741224255e-07,
      "loss": 0.6819,
      "step": 20301
    },
    {
      "epoch": 0.8413941729868623,
      "grad_norm": 0.46501705050468445,
      "learning_rate": 7.932363545940571e-07,
      "loss": 0.6971,
      "step": 20302
    },
    {
      "epoch": 0.841435616892536,
      "grad_norm": 0.4112091660499573,
      "learning_rate": 7.930291350656887e-07,
      "loss": 0.6692,
      "step": 20303
    },
    {
      "epoch": 0.8414770607982096,
      "grad_norm": 0.41882097721099854,
      "learning_rate": 7.928219155373203e-07,
      "loss": 0.6632,
      "step": 20304
    },
    {
      "epoch": 0.8415185047038833,
      "grad_norm": 0.40139561891555786,
      "learning_rate": 7.926146960089519e-07,
      "loss": 0.6851,
      "step": 20305
    },
    {
      "epoch": 0.841559948609557,
      "grad_norm": 0.41318365931510925,
      "learning_rate": 7.924074764805837e-07,
      "loss": 0.6484,
      "step": 20306
    },
    {
      "epoch": 0.8416013925152306,
      "grad_norm": 0.42632460594177246,
      "learning_rate": 7.922002569522153e-07,
      "loss": 0.6602,
      "step": 20307
    },
    {
      "epoch": 0.8416428364209043,
      "grad_norm": 0.44999897480010986,
      "learning_rate": 7.919930374238469e-07,
      "loss": 0.6691,
      "step": 20308
    },
    {
      "epoch": 0.8416842803265779,
      "grad_norm": 0.42027410864830017,
      "learning_rate": 7.917858178954785e-07,
      "loss": 0.6995,
      "step": 20309
    },
    {
      "epoch": 0.8417257242322517,
      "grad_norm": 0.46995192766189575,
      "learning_rate": 7.915785983671103e-07,
      "loss": 0.6521,
      "step": 20310
    },
    {
      "epoch": 0.8417671681379253,
      "grad_norm": 0.42636585235595703,
      "learning_rate": 7.913713788387419e-07,
      "loss": 0.7013,
      "step": 20311
    },
    {
      "epoch": 0.841808612043599,
      "grad_norm": 0.41778379678726196,
      "learning_rate": 7.911641593103735e-07,
      "loss": 0.7163,
      "step": 20312
    },
    {
      "epoch": 0.8418500559492726,
      "grad_norm": 0.44311270117759705,
      "learning_rate": 7.909569397820051e-07,
      "loss": 0.6772,
      "step": 20313
    },
    {
      "epoch": 0.8418914998549464,
      "grad_norm": 0.3899264931678772,
      "learning_rate": 7.907497202536369e-07,
      "loss": 0.623,
      "step": 20314
    },
    {
      "epoch": 0.84193294376062,
      "grad_norm": 0.41782352328300476,
      "learning_rate": 7.905425007252685e-07,
      "loss": 0.651,
      "step": 20315
    },
    {
      "epoch": 0.8419743876662936,
      "grad_norm": 0.37422803044319153,
      "learning_rate": 7.903352811969001e-07,
      "loss": 0.6643,
      "step": 20316
    },
    {
      "epoch": 0.8420158315719674,
      "grad_norm": 0.427614688873291,
      "learning_rate": 7.901280616685317e-07,
      "loss": 0.7292,
      "step": 20317
    },
    {
      "epoch": 0.842057275477641,
      "grad_norm": 0.4152832627296448,
      "learning_rate": 7.899208421401633e-07,
      "loss": 0.593,
      "step": 20318
    },
    {
      "epoch": 0.8420987193833147,
      "grad_norm": 0.39070409536361694,
      "learning_rate": 7.897136226117951e-07,
      "loss": 0.6697,
      "step": 20319
    },
    {
      "epoch": 0.8421401632889883,
      "grad_norm": 0.4044613838195801,
      "learning_rate": 7.895064030834267e-07,
      "loss": 0.6848,
      "step": 20320
    },
    {
      "epoch": 0.8421816071946621,
      "grad_norm": 0.4143247902393341,
      "learning_rate": 7.892991835550583e-07,
      "loss": 0.6975,
      "step": 20321
    },
    {
      "epoch": 0.8422230511003357,
      "grad_norm": 0.42672422528266907,
      "learning_rate": 7.890919640266899e-07,
      "loss": 0.6698,
      "step": 20322
    },
    {
      "epoch": 0.8422644950060094,
      "grad_norm": 0.3664911985397339,
      "learning_rate": 7.888847444983217e-07,
      "loss": 0.6427,
      "step": 20323
    },
    {
      "epoch": 0.842305938911683,
      "grad_norm": 0.4231851398944855,
      "learning_rate": 7.886775249699533e-07,
      "loss": 0.6643,
      "step": 20324
    },
    {
      "epoch": 0.8423473828173567,
      "grad_norm": 0.43241921067237854,
      "learning_rate": 7.884703054415849e-07,
      "loss": 0.6375,
      "step": 20325
    },
    {
      "epoch": 0.8423888267230304,
      "grad_norm": 0.38926368951797485,
      "learning_rate": 7.882630859132165e-07,
      "loss": 0.6272,
      "step": 20326
    },
    {
      "epoch": 0.842430270628704,
      "grad_norm": 0.38392889499664307,
      "learning_rate": 7.880558663848483e-07,
      "loss": 0.6116,
      "step": 20327
    },
    {
      "epoch": 0.8424717145343777,
      "grad_norm": 0.43238773941993713,
      "learning_rate": 7.878486468564799e-07,
      "loss": 0.6637,
      "step": 20328
    },
    {
      "epoch": 0.8425131584400514,
      "grad_norm": 0.39051949977874756,
      "learning_rate": 7.876414273281115e-07,
      "loss": 0.642,
      "step": 20329
    },
    {
      "epoch": 0.8425546023457251,
      "grad_norm": 0.4083532691001892,
      "learning_rate": 7.874342077997431e-07,
      "loss": 0.6875,
      "step": 20330
    },
    {
      "epoch": 0.8425960462513987,
      "grad_norm": 0.4511617422103882,
      "learning_rate": 7.872269882713747e-07,
      "loss": 0.6663,
      "step": 20331
    },
    {
      "epoch": 0.8426374901570725,
      "grad_norm": 0.429328590631485,
      "learning_rate": 7.870197687430065e-07,
      "loss": 0.6685,
      "step": 20332
    },
    {
      "epoch": 0.8426789340627461,
      "grad_norm": 0.4301462173461914,
      "learning_rate": 7.868125492146381e-07,
      "loss": 0.6873,
      "step": 20333
    },
    {
      "epoch": 0.8427203779684197,
      "grad_norm": 0.4104442894458771,
      "learning_rate": 7.866053296862697e-07,
      "loss": 0.6134,
      "step": 20334
    },
    {
      "epoch": 0.8427618218740934,
      "grad_norm": 0.3949783444404602,
      "learning_rate": 7.863981101579013e-07,
      "loss": 0.658,
      "step": 20335
    },
    {
      "epoch": 0.842803265779767,
      "grad_norm": 0.3894331157207489,
      "learning_rate": 7.861908906295331e-07,
      "loss": 0.6697,
      "step": 20336
    },
    {
      "epoch": 0.8428447096854408,
      "grad_norm": 0.4031327962875366,
      "learning_rate": 7.859836711011647e-07,
      "loss": 0.6454,
      "step": 20337
    },
    {
      "epoch": 0.8428861535911144,
      "grad_norm": 0.3981502950191498,
      "learning_rate": 7.857764515727963e-07,
      "loss": 0.6475,
      "step": 20338
    },
    {
      "epoch": 0.8429275974967881,
      "grad_norm": 0.4274056553840637,
      "learning_rate": 7.855692320444279e-07,
      "loss": 0.6909,
      "step": 20339
    },
    {
      "epoch": 0.8429690414024618,
      "grad_norm": 0.38113051652908325,
      "learning_rate": 7.853620125160597e-07,
      "loss": 0.6355,
      "step": 20340
    },
    {
      "epoch": 0.8430104853081355,
      "grad_norm": 0.41596877574920654,
      "learning_rate": 7.851547929876913e-07,
      "loss": 0.689,
      "step": 20341
    },
    {
      "epoch": 0.8430519292138091,
      "grad_norm": 0.451772004365921,
      "learning_rate": 7.849475734593229e-07,
      "loss": 0.6687,
      "step": 20342
    },
    {
      "epoch": 0.8430933731194827,
      "grad_norm": 0.41261178255081177,
      "learning_rate": 7.847403539309545e-07,
      "loss": 0.6761,
      "step": 20343
    },
    {
      "epoch": 0.8431348170251565,
      "grad_norm": 0.43507614731788635,
      "learning_rate": 7.845331344025861e-07,
      "loss": 0.6543,
      "step": 20344
    },
    {
      "epoch": 0.8431762609308301,
      "grad_norm": 0.36950087547302246,
      "learning_rate": 7.843259148742179e-07,
      "loss": 0.6219,
      "step": 20345
    },
    {
      "epoch": 0.8432177048365038,
      "grad_norm": 0.3994353711605072,
      "learning_rate": 7.841186953458495e-07,
      "loss": 0.6865,
      "step": 20346
    },
    {
      "epoch": 0.8432591487421774,
      "grad_norm": 0.41059935092926025,
      "learning_rate": 7.839114758174811e-07,
      "loss": 0.6384,
      "step": 20347
    },
    {
      "epoch": 0.8433005926478512,
      "grad_norm": 0.4057645797729492,
      "learning_rate": 7.837042562891127e-07,
      "loss": 0.6328,
      "step": 20348
    },
    {
      "epoch": 0.8433420365535248,
      "grad_norm": 0.434075266122818,
      "learning_rate": 7.834970367607445e-07,
      "loss": 0.6697,
      "step": 20349
    },
    {
      "epoch": 0.8433834804591984,
      "grad_norm": 0.374485582113266,
      "learning_rate": 7.832898172323761e-07,
      "loss": 0.6626,
      "step": 20350
    },
    {
      "epoch": 0.8434249243648722,
      "grad_norm": 0.45293405652046204,
      "learning_rate": 7.830825977040077e-07,
      "loss": 0.7603,
      "step": 20351
    },
    {
      "epoch": 0.8434663682705458,
      "grad_norm": 0.4895859956741333,
      "learning_rate": 7.828753781756393e-07,
      "loss": 0.7407,
      "step": 20352
    },
    {
      "epoch": 0.8435078121762195,
      "grad_norm": 0.4310372769832611,
      "learning_rate": 7.826681586472709e-07,
      "loss": 0.6704,
      "step": 20353
    },
    {
      "epoch": 0.8435492560818931,
      "grad_norm": 0.4351061284542084,
      "learning_rate": 7.824609391189027e-07,
      "loss": 0.6759,
      "step": 20354
    },
    {
      "epoch": 0.8435906999875669,
      "grad_norm": 0.42802488803863525,
      "learning_rate": 7.822537195905343e-07,
      "loss": 0.692,
      "step": 20355
    },
    {
      "epoch": 0.8436321438932405,
      "grad_norm": 0.4239374101161957,
      "learning_rate": 7.820465000621659e-07,
      "loss": 0.6714,
      "step": 20356
    },
    {
      "epoch": 0.8436735877989142,
      "grad_norm": 0.4451308846473694,
      "learning_rate": 7.818392805337975e-07,
      "loss": 0.7197,
      "step": 20357
    },
    {
      "epoch": 0.8437150317045878,
      "grad_norm": 0.427103191614151,
      "learning_rate": 7.816320610054293e-07,
      "loss": 0.6799,
      "step": 20358
    },
    {
      "epoch": 0.8437564756102615,
      "grad_norm": 0.42711034417152405,
      "learning_rate": 7.814248414770609e-07,
      "loss": 0.6897,
      "step": 20359
    },
    {
      "epoch": 0.8437979195159352,
      "grad_norm": 0.42329227924346924,
      "learning_rate": 7.812176219486925e-07,
      "loss": 0.6793,
      "step": 20360
    },
    {
      "epoch": 0.8438393634216088,
      "grad_norm": 0.3749352693557739,
      "learning_rate": 7.810104024203241e-07,
      "loss": 0.5995,
      "step": 20361
    },
    {
      "epoch": 0.8438808073272825,
      "grad_norm": 0.395255446434021,
      "learning_rate": 7.808031828919559e-07,
      "loss": 0.6316,
      "step": 20362
    },
    {
      "epoch": 0.8439222512329562,
      "grad_norm": 0.41739657521247864,
      "learning_rate": 7.805959633635875e-07,
      "loss": 0.6149,
      "step": 20363
    },
    {
      "epoch": 0.8439636951386299,
      "grad_norm": 0.416291207075119,
      "learning_rate": 7.803887438352191e-07,
      "loss": 0.698,
      "step": 20364
    },
    {
      "epoch": 0.8440051390443035,
      "grad_norm": 0.42187097668647766,
      "learning_rate": 7.801815243068507e-07,
      "loss": 0.6829,
      "step": 20365
    },
    {
      "epoch": 0.8440465829499773,
      "grad_norm": 0.4220981001853943,
      "learning_rate": 7.799743047784824e-07,
      "loss": 0.6226,
      "step": 20366
    },
    {
      "epoch": 0.8440880268556509,
      "grad_norm": 0.4218301773071289,
      "learning_rate": 7.797670852501141e-07,
      "loss": 0.6285,
      "step": 20367
    },
    {
      "epoch": 0.8441294707613245,
      "grad_norm": 0.43591782450675964,
      "learning_rate": 7.795598657217457e-07,
      "loss": 0.6646,
      "step": 20368
    },
    {
      "epoch": 0.8441709146669982,
      "grad_norm": 0.4099085330963135,
      "learning_rate": 7.793526461933773e-07,
      "loss": 0.6685,
      "step": 20369
    },
    {
      "epoch": 0.8442123585726719,
      "grad_norm": 0.43688157200813293,
      "learning_rate": 7.79145426665009e-07,
      "loss": 0.6534,
      "step": 20370
    },
    {
      "epoch": 0.8442538024783456,
      "grad_norm": 0.4327491521835327,
      "learning_rate": 7.789382071366407e-07,
      "loss": 0.6997,
      "step": 20371
    },
    {
      "epoch": 0.8442952463840192,
      "grad_norm": 0.4056679606437683,
      "learning_rate": 7.787309876082723e-07,
      "loss": 0.6401,
      "step": 20372
    },
    {
      "epoch": 0.8443366902896929,
      "grad_norm": 0.4049903452396393,
      "learning_rate": 7.785237680799039e-07,
      "loss": 0.6381,
      "step": 20373
    },
    {
      "epoch": 0.8443781341953666,
      "grad_norm": 0.4465595483779907,
      "learning_rate": 7.783165485515355e-07,
      "loss": 0.6967,
      "step": 20374
    },
    {
      "epoch": 0.8444195781010403,
      "grad_norm": 0.4046452045440674,
      "learning_rate": 7.781093290231673e-07,
      "loss": 0.7046,
      "step": 20375
    },
    {
      "epoch": 0.8444610220067139,
      "grad_norm": 0.4019264280796051,
      "learning_rate": 7.779021094947989e-07,
      "loss": 0.6641,
      "step": 20376
    },
    {
      "epoch": 0.8445024659123875,
      "grad_norm": 0.3953589200973511,
      "learning_rate": 7.776948899664305e-07,
      "loss": 0.6449,
      "step": 20377
    },
    {
      "epoch": 0.8445439098180613,
      "grad_norm": 0.3843584656715393,
      "learning_rate": 7.774876704380621e-07,
      "loss": 0.61,
      "step": 20378
    },
    {
      "epoch": 0.8445853537237349,
      "grad_norm": 0.3851132392883301,
      "learning_rate": 7.772804509096938e-07,
      "loss": 0.6539,
      "step": 20379
    },
    {
      "epoch": 0.8446267976294086,
      "grad_norm": 0.43895745277404785,
      "learning_rate": 7.770732313813255e-07,
      "loss": 0.6938,
      "step": 20380
    },
    {
      "epoch": 0.8446682415350822,
      "grad_norm": 0.42420393228530884,
      "learning_rate": 7.768660118529571e-07,
      "loss": 0.6774,
      "step": 20381
    },
    {
      "epoch": 0.844709685440756,
      "grad_norm": 0.4183800220489502,
      "learning_rate": 7.766587923245887e-07,
      "loss": 0.7079,
      "step": 20382
    },
    {
      "epoch": 0.8447511293464296,
      "grad_norm": 0.41839006543159485,
      "learning_rate": 7.764515727962204e-07,
      "loss": 0.7114,
      "step": 20383
    },
    {
      "epoch": 0.8447925732521033,
      "grad_norm": 0.4254937767982483,
      "learning_rate": 7.762443532678521e-07,
      "loss": 0.678,
      "step": 20384
    },
    {
      "epoch": 0.844834017157777,
      "grad_norm": 0.41902899742126465,
      "learning_rate": 7.760371337394837e-07,
      "loss": 0.7141,
      "step": 20385
    },
    {
      "epoch": 0.8448754610634506,
      "grad_norm": 0.4079195559024811,
      "learning_rate": 7.758299142111153e-07,
      "loss": 0.6603,
      "step": 20386
    },
    {
      "epoch": 0.8449169049691243,
      "grad_norm": 0.42082467675209045,
      "learning_rate": 7.75622694682747e-07,
      "loss": 0.6396,
      "step": 20387
    },
    {
      "epoch": 0.8449583488747979,
      "grad_norm": 0.3983731269836426,
      "learning_rate": 7.754154751543787e-07,
      "loss": 0.6548,
      "step": 20388
    },
    {
      "epoch": 0.8449997927804717,
      "grad_norm": 0.43963998556137085,
      "learning_rate": 7.752082556260103e-07,
      "loss": 0.7043,
      "step": 20389
    },
    {
      "epoch": 0.8450412366861453,
      "grad_norm": 0.40820637345314026,
      "learning_rate": 7.750010360976419e-07,
      "loss": 0.6436,
      "step": 20390
    },
    {
      "epoch": 0.845082680591819,
      "grad_norm": 0.4260731637477875,
      "learning_rate": 7.747938165692735e-07,
      "loss": 0.707,
      "step": 20391
    },
    {
      "epoch": 0.8451241244974926,
      "grad_norm": 0.41070759296417236,
      "learning_rate": 7.745865970409052e-07,
      "loss": 0.656,
      "step": 20392
    },
    {
      "epoch": 0.8451655684031664,
      "grad_norm": 0.3845791816711426,
      "learning_rate": 7.743793775125369e-07,
      "loss": 0.7061,
      "step": 20393
    },
    {
      "epoch": 0.84520701230884,
      "grad_norm": 0.37620067596435547,
      "learning_rate": 7.741721579841685e-07,
      "loss": 0.6254,
      "step": 20394
    },
    {
      "epoch": 0.8452484562145136,
      "grad_norm": 0.4091775119304657,
      "learning_rate": 7.739649384558001e-07,
      "loss": 0.6632,
      "step": 20395
    },
    {
      "epoch": 0.8452899001201873,
      "grad_norm": 0.4363259971141815,
      "learning_rate": 7.737577189274318e-07,
      "loss": 0.6471,
      "step": 20396
    },
    {
      "epoch": 0.845331344025861,
      "grad_norm": 0.4026367962360382,
      "learning_rate": 7.735504993990635e-07,
      "loss": 0.6597,
      "step": 20397
    },
    {
      "epoch": 0.8453727879315347,
      "grad_norm": 0.39174115657806396,
      "learning_rate": 7.733432798706951e-07,
      "loss": 0.6532,
      "step": 20398
    },
    {
      "epoch": 0.8454142318372083,
      "grad_norm": 0.39346399903297424,
      "learning_rate": 7.731360603423267e-07,
      "loss": 0.6852,
      "step": 20399
    },
    {
      "epoch": 0.845455675742882,
      "grad_norm": 0.41150641441345215,
      "learning_rate": 7.729288408139584e-07,
      "loss": 0.6597,
      "step": 20400
    },
    {
      "epoch": 0.8454971196485557,
      "grad_norm": 0.3971468210220337,
      "learning_rate": 7.727216212855901e-07,
      "loss": 0.6105,
      "step": 20401
    },
    {
      "epoch": 0.8455385635542294,
      "grad_norm": 0.4269232749938965,
      "learning_rate": 7.725144017572217e-07,
      "loss": 0.6918,
      "step": 20402
    },
    {
      "epoch": 0.845580007459903,
      "grad_norm": 0.44439801573753357,
      "learning_rate": 7.723071822288533e-07,
      "loss": 0.7056,
      "step": 20403
    },
    {
      "epoch": 0.8456214513655766,
      "grad_norm": 0.4329550266265869,
      "learning_rate": 7.72099962700485e-07,
      "loss": 0.6469,
      "step": 20404
    },
    {
      "epoch": 0.8456628952712504,
      "grad_norm": 0.37064599990844727,
      "learning_rate": 7.718927431721166e-07,
      "loss": 0.6388,
      "step": 20405
    },
    {
      "epoch": 0.845704339176924,
      "grad_norm": 0.5832260251045227,
      "learning_rate": 7.716855236437483e-07,
      "loss": 0.6921,
      "step": 20406
    },
    {
      "epoch": 0.8457457830825977,
      "grad_norm": 0.38717830181121826,
      "learning_rate": 7.714783041153799e-07,
      "loss": 0.6429,
      "step": 20407
    },
    {
      "epoch": 0.8457872269882714,
      "grad_norm": 0.38445159792900085,
      "learning_rate": 7.712710845870115e-07,
      "loss": 0.6371,
      "step": 20408
    },
    {
      "epoch": 0.8458286708939451,
      "grad_norm": 0.4126291871070862,
      "learning_rate": 7.710638650586432e-07,
      "loss": 0.6235,
      "step": 20409
    },
    {
      "epoch": 0.8458701147996187,
      "grad_norm": 0.4096199572086334,
      "learning_rate": 7.708566455302749e-07,
      "loss": 0.6708,
      "step": 20410
    },
    {
      "epoch": 0.8459115587052923,
      "grad_norm": 0.40687939524650574,
      "learning_rate": 7.706494260019065e-07,
      "loss": 0.687,
      "step": 20411
    },
    {
      "epoch": 0.8459530026109661,
      "grad_norm": 0.39303112030029297,
      "learning_rate": 7.704422064735381e-07,
      "loss": 0.623,
      "step": 20412
    },
    {
      "epoch": 0.8459944465166397,
      "grad_norm": 0.4197443127632141,
      "learning_rate": 7.702349869451698e-07,
      "loss": 0.6383,
      "step": 20413
    },
    {
      "epoch": 0.8460358904223134,
      "grad_norm": 0.4085114002227783,
      "learning_rate": 7.700277674168015e-07,
      "loss": 0.6508,
      "step": 20414
    },
    {
      "epoch": 0.846077334327987,
      "grad_norm": 0.4611682891845703,
      "learning_rate": 7.698205478884331e-07,
      "loss": 0.7354,
      "step": 20415
    },
    {
      "epoch": 0.8461187782336608,
      "grad_norm": 0.4064315855503082,
      "learning_rate": 7.696133283600647e-07,
      "loss": 0.6476,
      "step": 20416
    },
    {
      "epoch": 0.8461602221393344,
      "grad_norm": 0.4356051981449127,
      "learning_rate": 7.694061088316964e-07,
      "loss": 0.6798,
      "step": 20417
    },
    {
      "epoch": 0.8462016660450081,
      "grad_norm": 0.4189104437828064,
      "learning_rate": 7.69198889303328e-07,
      "loss": 0.6749,
      "step": 20418
    },
    {
      "epoch": 0.8462431099506817,
      "grad_norm": 0.5172034502029419,
      "learning_rate": 7.689916697749597e-07,
      "loss": 0.7185,
      "step": 20419
    },
    {
      "epoch": 0.8462845538563554,
      "grad_norm": 0.41513264179229736,
      "learning_rate": 7.687844502465913e-07,
      "loss": 0.6552,
      "step": 20420
    },
    {
      "epoch": 0.8463259977620291,
      "grad_norm": 0.4175315201282501,
      "learning_rate": 7.68577230718223e-07,
      "loss": 0.6852,
      "step": 20421
    },
    {
      "epoch": 0.8463674416677027,
      "grad_norm": 0.3819960355758667,
      "learning_rate": 7.683700111898546e-07,
      "loss": 0.6356,
      "step": 20422
    },
    {
      "epoch": 0.8464088855733765,
      "grad_norm": 0.4077717065811157,
      "learning_rate": 7.681627916614863e-07,
      "loss": 0.6478,
      "step": 20423
    },
    {
      "epoch": 0.8464503294790501,
      "grad_norm": 0.4417567849159241,
      "learning_rate": 7.679555721331179e-07,
      "loss": 0.6694,
      "step": 20424
    },
    {
      "epoch": 0.8464917733847238,
      "grad_norm": 0.3939686119556427,
      "learning_rate": 7.677483526047496e-07,
      "loss": 0.6677,
      "step": 20425
    },
    {
      "epoch": 0.8465332172903974,
      "grad_norm": 0.4271582365036011,
      "learning_rate": 7.675411330763812e-07,
      "loss": 0.6604,
      "step": 20426
    },
    {
      "epoch": 0.8465746611960712,
      "grad_norm": 0.4056811034679413,
      "learning_rate": 7.673339135480128e-07,
      "loss": 0.6514,
      "step": 20427
    },
    {
      "epoch": 0.8466161051017448,
      "grad_norm": 0.43984273076057434,
      "learning_rate": 7.671266940196445e-07,
      "loss": 0.6816,
      "step": 20428
    },
    {
      "epoch": 0.8466575490074184,
      "grad_norm": 0.4196353852748871,
      "learning_rate": 7.669194744912761e-07,
      "loss": 0.6813,
      "step": 20429
    },
    {
      "epoch": 0.8466989929130921,
      "grad_norm": 0.3771660625934601,
      "learning_rate": 7.667122549629078e-07,
      "loss": 0.6367,
      "step": 20430
    },
    {
      "epoch": 0.8467404368187658,
      "grad_norm": 0.4578901529312134,
      "learning_rate": 7.665050354345394e-07,
      "loss": 0.6997,
      "step": 20431
    },
    {
      "epoch": 0.8467818807244395,
      "grad_norm": 0.40670812129974365,
      "learning_rate": 7.662978159061711e-07,
      "loss": 0.6743,
      "step": 20432
    },
    {
      "epoch": 0.8468233246301131,
      "grad_norm": 0.3928362727165222,
      "learning_rate": 7.660905963778027e-07,
      "loss": 0.6326,
      "step": 20433
    },
    {
      "epoch": 0.8468647685357868,
      "grad_norm": 0.4217066168785095,
      "learning_rate": 7.658833768494344e-07,
      "loss": 0.7036,
      "step": 20434
    },
    {
      "epoch": 0.8469062124414605,
      "grad_norm": 0.39548835158348083,
      "learning_rate": 7.65676157321066e-07,
      "loss": 0.631,
      "step": 20435
    },
    {
      "epoch": 0.8469476563471342,
      "grad_norm": 0.4086071848869324,
      "learning_rate": 7.654689377926977e-07,
      "loss": 0.6477,
      "step": 20436
    },
    {
      "epoch": 0.8469891002528078,
      "grad_norm": 0.43001681566238403,
      "learning_rate": 7.652617182643293e-07,
      "loss": 0.6821,
      "step": 20437
    },
    {
      "epoch": 0.8470305441584814,
      "grad_norm": 0.43386632204055786,
      "learning_rate": 7.65054498735961e-07,
      "loss": 0.6713,
      "step": 20438
    },
    {
      "epoch": 0.8470719880641552,
      "grad_norm": 0.4168662130832672,
      "learning_rate": 7.648472792075926e-07,
      "loss": 0.7065,
      "step": 20439
    },
    {
      "epoch": 0.8471134319698288,
      "grad_norm": 0.39915046095848083,
      "learning_rate": 7.646400596792242e-07,
      "loss": 0.6168,
      "step": 20440
    },
    {
      "epoch": 0.8471548758755025,
      "grad_norm": 0.3733215928077698,
      "learning_rate": 7.644328401508559e-07,
      "loss": 0.6295,
      "step": 20441
    },
    {
      "epoch": 0.8471963197811762,
      "grad_norm": 0.4335360527038574,
      "learning_rate": 7.642256206224876e-07,
      "loss": 0.6675,
      "step": 20442
    },
    {
      "epoch": 0.8472377636868499,
      "grad_norm": 0.4043378233909607,
      "learning_rate": 7.640184010941192e-07,
      "loss": 0.6692,
      "step": 20443
    },
    {
      "epoch": 0.8472792075925235,
      "grad_norm": 0.4419574439525604,
      "learning_rate": 7.638111815657508e-07,
      "loss": 0.6327,
      "step": 20444
    },
    {
      "epoch": 0.8473206514981972,
      "grad_norm": 0.43890947103500366,
      "learning_rate": 7.636039620373825e-07,
      "loss": 0.6483,
      "step": 20445
    },
    {
      "epoch": 0.8473620954038709,
      "grad_norm": 0.4200500547885895,
      "learning_rate": 7.633967425090141e-07,
      "loss": 0.6548,
      "step": 20446
    },
    {
      "epoch": 0.8474035393095445,
      "grad_norm": 0.4187223017215729,
      "learning_rate": 7.631895229806458e-07,
      "loss": 0.7084,
      "step": 20447
    },
    {
      "epoch": 0.8474449832152182,
      "grad_norm": 0.42845991253852844,
      "learning_rate": 7.629823034522774e-07,
      "loss": 0.6686,
      "step": 20448
    },
    {
      "epoch": 0.8474864271208918,
      "grad_norm": 0.4146058261394501,
      "learning_rate": 7.627750839239091e-07,
      "loss": 0.6583,
      "step": 20449
    },
    {
      "epoch": 0.8475278710265656,
      "grad_norm": 0.4317057728767395,
      "learning_rate": 7.625678643955407e-07,
      "loss": 0.6403,
      "step": 20450
    },
    {
      "epoch": 0.8475693149322392,
      "grad_norm": 0.4178686738014221,
      "learning_rate": 7.623606448671724e-07,
      "loss": 0.6412,
      "step": 20451
    },
    {
      "epoch": 0.8476107588379129,
      "grad_norm": 0.3794616162776947,
      "learning_rate": 7.62153425338804e-07,
      "loss": 0.6102,
      "step": 20452
    },
    {
      "epoch": 0.8476522027435865,
      "grad_norm": 0.43796563148498535,
      "learning_rate": 7.619462058104356e-07,
      "loss": 0.6555,
      "step": 20453
    },
    {
      "epoch": 0.8476936466492603,
      "grad_norm": 0.4346521496772766,
      "learning_rate": 7.617389862820673e-07,
      "loss": 0.6963,
      "step": 20454
    },
    {
      "epoch": 0.8477350905549339,
      "grad_norm": 0.426464319229126,
      "learning_rate": 7.61531766753699e-07,
      "loss": 0.6826,
      "step": 20455
    },
    {
      "epoch": 0.8477765344606075,
      "grad_norm": 0.4010998606681824,
      "learning_rate": 7.613245472253306e-07,
      "loss": 0.6482,
      "step": 20456
    },
    {
      "epoch": 0.8478179783662813,
      "grad_norm": 0.41508328914642334,
      "learning_rate": 7.611173276969622e-07,
      "loss": 0.6873,
      "step": 20457
    },
    {
      "epoch": 0.8478594222719549,
      "grad_norm": 0.4321563243865967,
      "learning_rate": 7.609101081685939e-07,
      "loss": 0.6653,
      "step": 20458
    },
    {
      "epoch": 0.8479008661776286,
      "grad_norm": 0.39460548758506775,
      "learning_rate": 7.607028886402256e-07,
      "loss": 0.5974,
      "step": 20459
    },
    {
      "epoch": 0.8479423100833022,
      "grad_norm": 0.43614527583122253,
      "learning_rate": 7.604956691118572e-07,
      "loss": 0.6874,
      "step": 20460
    },
    {
      "epoch": 0.847983753988976,
      "grad_norm": 0.40487203001976013,
      "learning_rate": 7.602884495834888e-07,
      "loss": 0.6412,
      "step": 20461
    },
    {
      "epoch": 0.8480251978946496,
      "grad_norm": 0.43353933095932007,
      "learning_rate": 7.600812300551205e-07,
      "loss": 0.6316,
      "step": 20462
    },
    {
      "epoch": 0.8480666418003233,
      "grad_norm": 0.44699952006340027,
      "learning_rate": 7.598740105267522e-07,
      "loss": 0.6508,
      "step": 20463
    },
    {
      "epoch": 0.8481080857059969,
      "grad_norm": 0.4085722863674164,
      "learning_rate": 7.596667909983838e-07,
      "loss": 0.7041,
      "step": 20464
    },
    {
      "epoch": 0.8481495296116706,
      "grad_norm": 0.40354153513908386,
      "learning_rate": 7.594595714700154e-07,
      "loss": 0.6476,
      "step": 20465
    },
    {
      "epoch": 0.8481909735173443,
      "grad_norm": 0.4219331443309784,
      "learning_rate": 7.59252351941647e-07,
      "loss": 0.6973,
      "step": 20466
    },
    {
      "epoch": 0.8482324174230179,
      "grad_norm": 0.43789976835250854,
      "learning_rate": 7.590451324132787e-07,
      "loss": 0.6921,
      "step": 20467
    },
    {
      "epoch": 0.8482738613286916,
      "grad_norm": 0.40335455536842346,
      "learning_rate": 7.588379128849104e-07,
      "loss": 0.6266,
      "step": 20468
    },
    {
      "epoch": 0.8483153052343653,
      "grad_norm": 0.39297932386398315,
      "learning_rate": 7.58630693356542e-07,
      "loss": 0.6288,
      "step": 20469
    },
    {
      "epoch": 0.848356749140039,
      "grad_norm": 0.3709549307823181,
      "learning_rate": 7.584234738281736e-07,
      "loss": 0.6178,
      "step": 20470
    },
    {
      "epoch": 0.8483981930457126,
      "grad_norm": 0.3981325626373291,
      "learning_rate": 7.582162542998053e-07,
      "loss": 0.6548,
      "step": 20471
    },
    {
      "epoch": 0.8484396369513862,
      "grad_norm": 0.41424915194511414,
      "learning_rate": 7.58009034771437e-07,
      "loss": 0.6589,
      "step": 20472
    },
    {
      "epoch": 0.84848108085706,
      "grad_norm": 0.38770031929016113,
      "learning_rate": 7.578018152430686e-07,
      "loss": 0.6606,
      "step": 20473
    },
    {
      "epoch": 0.8485225247627336,
      "grad_norm": 0.46688422560691833,
      "learning_rate": 7.575945957147002e-07,
      "loss": 0.7151,
      "step": 20474
    },
    {
      "epoch": 0.8485639686684073,
      "grad_norm": 0.40260228514671326,
      "learning_rate": 7.573873761863319e-07,
      "loss": 0.6848,
      "step": 20475
    },
    {
      "epoch": 0.848605412574081,
      "grad_norm": 0.3820686638355255,
      "learning_rate": 7.571801566579636e-07,
      "loss": 0.6425,
      "step": 20476
    },
    {
      "epoch": 0.8486468564797547,
      "grad_norm": 0.4266296923160553,
      "learning_rate": 7.569729371295952e-07,
      "loss": 0.6761,
      "step": 20477
    },
    {
      "epoch": 0.8486883003854283,
      "grad_norm": 0.39699041843414307,
      "learning_rate": 7.567657176012268e-07,
      "loss": 0.6775,
      "step": 20478
    },
    {
      "epoch": 0.848729744291102,
      "grad_norm": 0.3877445161342621,
      "learning_rate": 7.565584980728584e-07,
      "loss": 0.6154,
      "step": 20479
    },
    {
      "epoch": 0.8487711881967757,
      "grad_norm": 0.42113596200942993,
      "learning_rate": 7.563512785444902e-07,
      "loss": 0.6038,
      "step": 20480
    },
    {
      "epoch": 0.8488126321024493,
      "grad_norm": 0.4289214611053467,
      "learning_rate": 7.561440590161218e-07,
      "loss": 0.6652,
      "step": 20481
    },
    {
      "epoch": 0.848854076008123,
      "grad_norm": 0.4578174352645874,
      "learning_rate": 7.559368394877534e-07,
      "loss": 0.6746,
      "step": 20482
    },
    {
      "epoch": 0.8488955199137966,
      "grad_norm": 0.4689340591430664,
      "learning_rate": 7.55729619959385e-07,
      "loss": 0.7081,
      "step": 20483
    },
    {
      "epoch": 0.8489369638194704,
      "grad_norm": 0.41705599427223206,
      "learning_rate": 7.555224004310167e-07,
      "loss": 0.6865,
      "step": 20484
    },
    {
      "epoch": 0.848978407725144,
      "grad_norm": 0.4212873876094818,
      "learning_rate": 7.553151809026484e-07,
      "loss": 0.6375,
      "step": 20485
    },
    {
      "epoch": 0.8490198516308177,
      "grad_norm": 0.44557350873947144,
      "learning_rate": 7.5510796137428e-07,
      "loss": 0.6527,
      "step": 20486
    },
    {
      "epoch": 0.8490612955364913,
      "grad_norm": 0.40809524059295654,
      "learning_rate": 7.549007418459116e-07,
      "loss": 0.6798,
      "step": 20487
    },
    {
      "epoch": 0.8491027394421651,
      "grad_norm": 0.3946968615055084,
      "learning_rate": 7.546935223175432e-07,
      "loss": 0.6733,
      "step": 20488
    },
    {
      "epoch": 0.8491441833478387,
      "grad_norm": 0.4350440502166748,
      "learning_rate": 7.54486302789175e-07,
      "loss": 0.6661,
      "step": 20489
    },
    {
      "epoch": 0.8491856272535123,
      "grad_norm": 0.4276566505432129,
      "learning_rate": 7.542790832608066e-07,
      "loss": 0.6877,
      "step": 20490
    },
    {
      "epoch": 0.849227071159186,
      "grad_norm": 0.4717964828014374,
      "learning_rate": 7.540718637324382e-07,
      "loss": 0.6794,
      "step": 20491
    },
    {
      "epoch": 0.8492685150648597,
      "grad_norm": 0.4250555634498596,
      "learning_rate": 7.538646442040698e-07,
      "loss": 0.6914,
      "step": 20492
    },
    {
      "epoch": 0.8493099589705334,
      "grad_norm": 0.415860652923584,
      "learning_rate": 7.536574246757016e-07,
      "loss": 0.7136,
      "step": 20493
    },
    {
      "epoch": 0.849351402876207,
      "grad_norm": 0.4054071009159088,
      "learning_rate": 7.534502051473332e-07,
      "loss": 0.6812,
      "step": 20494
    },
    {
      "epoch": 0.8493928467818808,
      "grad_norm": 0.3962019085884094,
      "learning_rate": 7.532429856189648e-07,
      "loss": 0.6167,
      "step": 20495
    },
    {
      "epoch": 0.8494342906875544,
      "grad_norm": 0.426248699426651,
      "learning_rate": 7.530357660905964e-07,
      "loss": 0.6975,
      "step": 20496
    },
    {
      "epoch": 0.8494757345932281,
      "grad_norm": 0.4153566062450409,
      "learning_rate": 7.528285465622282e-07,
      "loss": 0.6223,
      "step": 20497
    },
    {
      "epoch": 0.8495171784989017,
      "grad_norm": 0.45274245738983154,
      "learning_rate": 7.526213270338598e-07,
      "loss": 0.6283,
      "step": 20498
    },
    {
      "epoch": 0.8495586224045754,
      "grad_norm": 0.4107130765914917,
      "learning_rate": 7.524141075054914e-07,
      "loss": 0.666,
      "step": 20499
    },
    {
      "epoch": 0.8496000663102491,
      "grad_norm": 0.40169525146484375,
      "learning_rate": 7.52206887977123e-07,
      "loss": 0.6267,
      "step": 20500
    },
    {
      "epoch": 0.8496415102159227,
      "grad_norm": 0.41305768489837646,
      "learning_rate": 7.519996684487546e-07,
      "loss": 0.7397,
      "step": 20501
    },
    {
      "epoch": 0.8496829541215964,
      "grad_norm": 0.4273608326911926,
      "learning_rate": 7.517924489203864e-07,
      "loss": 0.7434,
      "step": 20502
    },
    {
      "epoch": 0.8497243980272701,
      "grad_norm": 0.4160315692424774,
      "learning_rate": 7.51585229392018e-07,
      "loss": 0.738,
      "step": 20503
    },
    {
      "epoch": 0.8497658419329438,
      "grad_norm": 0.4425608515739441,
      "learning_rate": 7.513780098636496e-07,
      "loss": 0.6426,
      "step": 20504
    },
    {
      "epoch": 0.8498072858386174,
      "grad_norm": 0.40396633744239807,
      "learning_rate": 7.511707903352812e-07,
      "loss": 0.6733,
      "step": 20505
    },
    {
      "epoch": 0.8498487297442912,
      "grad_norm": 0.42526212334632874,
      "learning_rate": 7.50963570806913e-07,
      "loss": 0.7002,
      "step": 20506
    },
    {
      "epoch": 0.8498901736499648,
      "grad_norm": 0.4493483006954193,
      "learning_rate": 7.507563512785446e-07,
      "loss": 0.6263,
      "step": 20507
    },
    {
      "epoch": 0.8499316175556384,
      "grad_norm": 0.4093776047229767,
      "learning_rate": 7.505491317501762e-07,
      "loss": 0.7063,
      "step": 20508
    },
    {
      "epoch": 0.8499730614613121,
      "grad_norm": 0.4123353064060211,
      "learning_rate": 7.503419122218078e-07,
      "loss": 0.6359,
      "step": 20509
    },
    {
      "epoch": 0.8500145053669858,
      "grad_norm": 0.41335126757621765,
      "learning_rate": 7.501346926934396e-07,
      "loss": 0.6841,
      "step": 20510
    },
    {
      "epoch": 0.8500559492726595,
      "grad_norm": 0.41959109902381897,
      "learning_rate": 7.499274731650712e-07,
      "loss": 0.6683,
      "step": 20511
    },
    {
      "epoch": 0.8500973931783331,
      "grad_norm": 0.40418002009391785,
      "learning_rate": 7.497202536367028e-07,
      "loss": 0.651,
      "step": 20512
    },
    {
      "epoch": 0.8501388370840068,
      "grad_norm": 0.45339900255203247,
      "learning_rate": 7.495130341083344e-07,
      "loss": 0.7083,
      "step": 20513
    },
    {
      "epoch": 0.8501802809896805,
      "grad_norm": 0.46708256006240845,
      "learning_rate": 7.49305814579966e-07,
      "loss": 0.6401,
      "step": 20514
    },
    {
      "epoch": 0.8502217248953542,
      "grad_norm": 0.4338332712650299,
      "learning_rate": 7.490985950515978e-07,
      "loss": 0.7207,
      "step": 20515
    },
    {
      "epoch": 0.8502631688010278,
      "grad_norm": 0.414694607257843,
      "learning_rate": 7.488913755232294e-07,
      "loss": 0.6696,
      "step": 20516
    },
    {
      "epoch": 0.8503046127067014,
      "grad_norm": 0.44499385356903076,
      "learning_rate": 7.48684155994861e-07,
      "loss": 0.6465,
      "step": 20517
    },
    {
      "epoch": 0.8503460566123752,
      "grad_norm": 0.3877165615558624,
      "learning_rate": 7.484769364664926e-07,
      "loss": 0.6407,
      "step": 20518
    },
    {
      "epoch": 0.8503875005180488,
      "grad_norm": 0.37963423132896423,
      "learning_rate": 7.482697169381244e-07,
      "loss": 0.594,
      "step": 20519
    },
    {
      "epoch": 0.8504289444237225,
      "grad_norm": 0.3967238664627075,
      "learning_rate": 7.48062497409756e-07,
      "loss": 0.6405,
      "step": 20520
    },
    {
      "epoch": 0.8504703883293961,
      "grad_norm": 0.4245978891849518,
      "learning_rate": 7.478552778813876e-07,
      "loss": 0.6625,
      "step": 20521
    },
    {
      "epoch": 0.8505118322350699,
      "grad_norm": 0.39641013741493225,
      "learning_rate": 7.476480583530192e-07,
      "loss": 0.6599,
      "step": 20522
    },
    {
      "epoch": 0.8505532761407435,
      "grad_norm": 0.4119395911693573,
      "learning_rate": 7.47440838824651e-07,
      "loss": 0.6602,
      "step": 20523
    },
    {
      "epoch": 0.8505947200464172,
      "grad_norm": 0.42134690284729004,
      "learning_rate": 7.472336192962826e-07,
      "loss": 0.631,
      "step": 20524
    },
    {
      "epoch": 0.8506361639520909,
      "grad_norm": 0.4474722146987915,
      "learning_rate": 7.470263997679142e-07,
      "loss": 0.6774,
      "step": 20525
    },
    {
      "epoch": 0.8506776078577645,
      "grad_norm": 0.37624427676200867,
      "learning_rate": 7.468191802395458e-07,
      "loss": 0.6644,
      "step": 20526
    },
    {
      "epoch": 0.8507190517634382,
      "grad_norm": 0.4044206440448761,
      "learning_rate": 7.466119607111774e-07,
      "loss": 0.7095,
      "step": 20527
    },
    {
      "epoch": 0.8507604956691118,
      "grad_norm": 0.40896427631378174,
      "learning_rate": 7.464047411828092e-07,
      "loss": 0.6942,
      "step": 20528
    },
    {
      "epoch": 0.8508019395747856,
      "grad_norm": 0.39833012223243713,
      "learning_rate": 7.461975216544408e-07,
      "loss": 0.6276,
      "step": 20529
    },
    {
      "epoch": 0.8508433834804592,
      "grad_norm": 0.39790770411491394,
      "learning_rate": 7.459903021260724e-07,
      "loss": 0.6538,
      "step": 20530
    },
    {
      "epoch": 0.8508848273861329,
      "grad_norm": 0.4233314096927643,
      "learning_rate": 7.45783082597704e-07,
      "loss": 0.6996,
      "step": 20531
    },
    {
      "epoch": 0.8509262712918065,
      "grad_norm": 0.3879458010196686,
      "learning_rate": 7.455758630693358e-07,
      "loss": 0.6735,
      "step": 20532
    },
    {
      "epoch": 0.8509677151974802,
      "grad_norm": 0.4220264256000519,
      "learning_rate": 7.453686435409674e-07,
      "loss": 0.6913,
      "step": 20533
    },
    {
      "epoch": 0.8510091591031539,
      "grad_norm": 0.4294629693031311,
      "learning_rate": 7.45161424012599e-07,
      "loss": 0.7162,
      "step": 20534
    },
    {
      "epoch": 0.8510506030088275,
      "grad_norm": 0.3961077034473419,
      "learning_rate": 7.449542044842306e-07,
      "loss": 0.6479,
      "step": 20535
    },
    {
      "epoch": 0.8510920469145012,
      "grad_norm": 0.42062968015670776,
      "learning_rate": 7.447469849558624e-07,
      "loss": 0.6447,
      "step": 20536
    },
    {
      "epoch": 0.8511334908201749,
      "grad_norm": 0.40641701221466064,
      "learning_rate": 7.44539765427494e-07,
      "loss": 0.6285,
      "step": 20537
    },
    {
      "epoch": 0.8511749347258486,
      "grad_norm": 0.3852499723434448,
      "learning_rate": 7.443325458991256e-07,
      "loss": 0.6493,
      "step": 20538
    },
    {
      "epoch": 0.8512163786315222,
      "grad_norm": 0.40276870131492615,
      "learning_rate": 7.441253263707572e-07,
      "loss": 0.6063,
      "step": 20539
    },
    {
      "epoch": 0.851257822537196,
      "grad_norm": 0.4118403196334839,
      "learning_rate": 7.439181068423888e-07,
      "loss": 0.5885,
      "step": 20540
    },
    {
      "epoch": 0.8512992664428696,
      "grad_norm": 0.41731491684913635,
      "learning_rate": 7.437108873140206e-07,
      "loss": 0.6685,
      "step": 20541
    },
    {
      "epoch": 0.8513407103485432,
      "grad_norm": 0.42045459151268005,
      "learning_rate": 7.435036677856522e-07,
      "loss": 0.6855,
      "step": 20542
    },
    {
      "epoch": 0.8513821542542169,
      "grad_norm": 0.4106985330581665,
      "learning_rate": 7.432964482572838e-07,
      "loss": 0.6433,
      "step": 20543
    },
    {
      "epoch": 0.8514235981598905,
      "grad_norm": 0.4170352518558502,
      "learning_rate": 7.430892287289154e-07,
      "loss": 0.666,
      "step": 20544
    },
    {
      "epoch": 0.8514650420655643,
      "grad_norm": 0.3775574266910553,
      "learning_rate": 7.428820092005472e-07,
      "loss": 0.679,
      "step": 20545
    },
    {
      "epoch": 0.8515064859712379,
      "grad_norm": 0.438890665769577,
      "learning_rate": 7.426747896721788e-07,
      "loss": 0.677,
      "step": 20546
    },
    {
      "epoch": 0.8515479298769116,
      "grad_norm": 0.44147396087646484,
      "learning_rate": 7.424675701438104e-07,
      "loss": 0.7334,
      "step": 20547
    },
    {
      "epoch": 0.8515893737825853,
      "grad_norm": 0.4002898931503296,
      "learning_rate": 7.42260350615442e-07,
      "loss": 0.6495,
      "step": 20548
    },
    {
      "epoch": 0.851630817688259,
      "grad_norm": 0.4059344232082367,
      "learning_rate": 7.420531310870738e-07,
      "loss": 0.7013,
      "step": 20549
    },
    {
      "epoch": 0.8516722615939326,
      "grad_norm": 0.456600159406662,
      "learning_rate": 7.418459115587054e-07,
      "loss": 0.6855,
      "step": 20550
    },
    {
      "epoch": 0.8517137054996062,
      "grad_norm": 0.42365702986717224,
      "learning_rate": 7.41638692030337e-07,
      "loss": 0.6334,
      "step": 20551
    },
    {
      "epoch": 0.85175514940528,
      "grad_norm": 0.396771639585495,
      "learning_rate": 7.414314725019686e-07,
      "loss": 0.6213,
      "step": 20552
    },
    {
      "epoch": 0.8517965933109536,
      "grad_norm": 0.4176287353038788,
      "learning_rate": 7.412242529736002e-07,
      "loss": 0.6459,
      "step": 20553
    },
    {
      "epoch": 0.8518380372166273,
      "grad_norm": 0.4199005663394928,
      "learning_rate": 7.41017033445232e-07,
      "loss": 0.6429,
      "step": 20554
    },
    {
      "epoch": 0.8518794811223009,
      "grad_norm": 0.4248897433280945,
      "learning_rate": 7.408098139168636e-07,
      "loss": 0.6558,
      "step": 20555
    },
    {
      "epoch": 0.8519209250279747,
      "grad_norm": 0.4089105725288391,
      "learning_rate": 7.406025943884952e-07,
      "loss": 0.697,
      "step": 20556
    },
    {
      "epoch": 0.8519623689336483,
      "grad_norm": 0.4246751368045807,
      "learning_rate": 7.403953748601268e-07,
      "loss": 0.5947,
      "step": 20557
    },
    {
      "epoch": 0.852003812839322,
      "grad_norm": 0.41488566994667053,
      "learning_rate": 7.401881553317586e-07,
      "loss": 0.7073,
      "step": 20558
    },
    {
      "epoch": 0.8520452567449956,
      "grad_norm": 0.4013586640357971,
      "learning_rate": 7.399809358033902e-07,
      "loss": 0.6832,
      "step": 20559
    },
    {
      "epoch": 0.8520867006506693,
      "grad_norm": 0.44106525182724,
      "learning_rate": 7.397737162750218e-07,
      "loss": 0.7209,
      "step": 20560
    },
    {
      "epoch": 0.852128144556343,
      "grad_norm": 0.4139322340488434,
      "learning_rate": 7.395664967466534e-07,
      "loss": 0.6907,
      "step": 20561
    },
    {
      "epoch": 0.8521695884620166,
      "grad_norm": 0.4122198820114136,
      "learning_rate": 7.39359277218285e-07,
      "loss": 0.6549,
      "step": 20562
    },
    {
      "epoch": 0.8522110323676904,
      "grad_norm": 0.43688124418258667,
      "learning_rate": 7.391520576899168e-07,
      "loss": 0.6855,
      "step": 20563
    },
    {
      "epoch": 0.852252476273364,
      "grad_norm": 0.4417873024940491,
      "learning_rate": 7.389448381615484e-07,
      "loss": 0.6907,
      "step": 20564
    },
    {
      "epoch": 0.8522939201790377,
      "grad_norm": 0.42622631788253784,
      "learning_rate": 7.3873761863318e-07,
      "loss": 0.6353,
      "step": 20565
    },
    {
      "epoch": 0.8523353640847113,
      "grad_norm": 0.3992529809474945,
      "learning_rate": 7.385303991048116e-07,
      "loss": 0.6819,
      "step": 20566
    },
    {
      "epoch": 0.8523768079903851,
      "grad_norm": 0.39934906363487244,
      "learning_rate": 7.383231795764434e-07,
      "loss": 0.6281,
      "step": 20567
    },
    {
      "epoch": 0.8524182518960587,
      "grad_norm": 0.3900816738605499,
      "learning_rate": 7.38115960048075e-07,
      "loss": 0.647,
      "step": 20568
    },
    {
      "epoch": 0.8524596958017323,
      "grad_norm": 0.4243934452533722,
      "learning_rate": 7.379087405197066e-07,
      "loss": 0.6453,
      "step": 20569
    },
    {
      "epoch": 0.852501139707406,
      "grad_norm": 0.41517138481140137,
      "learning_rate": 7.377015209913382e-07,
      "loss": 0.6384,
      "step": 20570
    },
    {
      "epoch": 0.8525425836130797,
      "grad_norm": 0.42522910237312317,
      "learning_rate": 7.3749430146297e-07,
      "loss": 0.657,
      "step": 20571
    },
    {
      "epoch": 0.8525840275187534,
      "grad_norm": 0.42327556014060974,
      "learning_rate": 7.372870819346016e-07,
      "loss": 0.718,
      "step": 20572
    },
    {
      "epoch": 0.852625471424427,
      "grad_norm": 0.4190184772014618,
      "learning_rate": 7.370798624062332e-07,
      "loss": 0.7026,
      "step": 20573
    },
    {
      "epoch": 0.8526669153301007,
      "grad_norm": 0.4596066474914551,
      "learning_rate": 7.368726428778648e-07,
      "loss": 0.729,
      "step": 20574
    },
    {
      "epoch": 0.8527083592357744,
      "grad_norm": 0.4370293617248535,
      "learning_rate": 7.366654233494965e-07,
      "loss": 0.6577,
      "step": 20575
    },
    {
      "epoch": 0.8527498031414481,
      "grad_norm": 0.47592470049858093,
      "learning_rate": 7.364582038211282e-07,
      "loss": 0.6472,
      "step": 20576
    },
    {
      "epoch": 0.8527912470471217,
      "grad_norm": 0.4237995147705078,
      "learning_rate": 7.362509842927598e-07,
      "loss": 0.6587,
      "step": 20577
    },
    {
      "epoch": 0.8528326909527953,
      "grad_norm": 0.4135417938232422,
      "learning_rate": 7.360437647643914e-07,
      "loss": 0.6089,
      "step": 20578
    },
    {
      "epoch": 0.8528741348584691,
      "grad_norm": 0.42195311188697815,
      "learning_rate": 7.35836545236023e-07,
      "loss": 0.677,
      "step": 20579
    },
    {
      "epoch": 0.8529155787641427,
      "grad_norm": 0.4275600016117096,
      "learning_rate": 7.356293257076548e-07,
      "loss": 0.6542,
      "step": 20580
    },
    {
      "epoch": 0.8529570226698164,
      "grad_norm": 0.4209749400615692,
      "learning_rate": 7.354221061792864e-07,
      "loss": 0.6752,
      "step": 20581
    },
    {
      "epoch": 0.85299846657549,
      "grad_norm": 0.4078131318092346,
      "learning_rate": 7.35214886650918e-07,
      "loss": 0.6881,
      "step": 20582
    },
    {
      "epoch": 0.8530399104811638,
      "grad_norm": 0.41180264949798584,
      "learning_rate": 7.350076671225496e-07,
      "loss": 0.688,
      "step": 20583
    },
    {
      "epoch": 0.8530813543868374,
      "grad_norm": 0.3994300365447998,
      "learning_rate": 7.348004475941814e-07,
      "loss": 0.639,
      "step": 20584
    },
    {
      "epoch": 0.853122798292511,
      "grad_norm": 0.41855666041374207,
      "learning_rate": 7.34593228065813e-07,
      "loss": 0.6479,
      "step": 20585
    },
    {
      "epoch": 0.8531642421981848,
      "grad_norm": 0.4313127398490906,
      "learning_rate": 7.343860085374446e-07,
      "loss": 0.6887,
      "step": 20586
    },
    {
      "epoch": 0.8532056861038584,
      "grad_norm": 0.40297380089759827,
      "learning_rate": 7.341787890090762e-07,
      "loss": 0.7358,
      "step": 20587
    },
    {
      "epoch": 0.8532471300095321,
      "grad_norm": 0.4103645384311676,
      "learning_rate": 7.339715694807079e-07,
      "loss": 0.6616,
      "step": 20588
    },
    {
      "epoch": 0.8532885739152057,
      "grad_norm": 0.42178410291671753,
      "learning_rate": 7.337643499523396e-07,
      "loss": 0.6399,
      "step": 20589
    },
    {
      "epoch": 0.8533300178208795,
      "grad_norm": 0.43834972381591797,
      "learning_rate": 7.335571304239712e-07,
      "loss": 0.6866,
      "step": 20590
    },
    {
      "epoch": 0.8533714617265531,
      "grad_norm": 0.40926608443260193,
      "learning_rate": 7.333499108956028e-07,
      "loss": 0.623,
      "step": 20591
    },
    {
      "epoch": 0.8534129056322268,
      "grad_norm": 0.39626017212867737,
      "learning_rate": 7.331426913672345e-07,
      "loss": 0.6733,
      "step": 20592
    },
    {
      "epoch": 0.8534543495379004,
      "grad_norm": 0.4075760245323181,
      "learning_rate": 7.329354718388662e-07,
      "loss": 0.6459,
      "step": 20593
    },
    {
      "epoch": 0.8534957934435741,
      "grad_norm": 0.4105537533760071,
      "learning_rate": 7.327282523104978e-07,
      "loss": 0.6898,
      "step": 20594
    },
    {
      "epoch": 0.8535372373492478,
      "grad_norm": 0.4362832307815552,
      "learning_rate": 7.325210327821294e-07,
      "loss": 0.6792,
      "step": 20595
    },
    {
      "epoch": 0.8535786812549214,
      "grad_norm": 0.4502408504486084,
      "learning_rate": 7.323138132537611e-07,
      "loss": 0.7025,
      "step": 20596
    },
    {
      "epoch": 0.8536201251605952,
      "grad_norm": 0.385731965303421,
      "learning_rate": 7.321065937253928e-07,
      "loss": 0.6691,
      "step": 20597
    },
    {
      "epoch": 0.8536615690662688,
      "grad_norm": 0.4145203232765198,
      "learning_rate": 7.318993741970244e-07,
      "loss": 0.6448,
      "step": 20598
    },
    {
      "epoch": 0.8537030129719425,
      "grad_norm": 0.38093337416648865,
      "learning_rate": 7.31692154668656e-07,
      "loss": 0.6366,
      "step": 20599
    },
    {
      "epoch": 0.8537444568776161,
      "grad_norm": 0.4369021952152252,
      "learning_rate": 7.314849351402876e-07,
      "loss": 0.6249,
      "step": 20600
    },
    {
      "epoch": 0.8537859007832899,
      "grad_norm": 0.4268071949481964,
      "learning_rate": 7.312777156119193e-07,
      "loss": 0.6581,
      "step": 20601
    },
    {
      "epoch": 0.8538273446889635,
      "grad_norm": 0.4084811806678772,
      "learning_rate": 7.31070496083551e-07,
      "loss": 0.6379,
      "step": 20602
    },
    {
      "epoch": 0.8538687885946371,
      "grad_norm": 0.439690500497818,
      "learning_rate": 7.308632765551826e-07,
      "loss": 0.7015,
      "step": 20603
    },
    {
      "epoch": 0.8539102325003108,
      "grad_norm": 0.44381242990493774,
      "learning_rate": 7.306560570268142e-07,
      "loss": 0.6343,
      "step": 20604
    },
    {
      "epoch": 0.8539516764059845,
      "grad_norm": 0.46103769540786743,
      "learning_rate": 7.304488374984459e-07,
      "loss": 0.6775,
      "step": 20605
    },
    {
      "epoch": 0.8539931203116582,
      "grad_norm": 0.4114983081817627,
      "learning_rate": 7.302416179700776e-07,
      "loss": 0.6346,
      "step": 20606
    },
    {
      "epoch": 0.8540345642173318,
      "grad_norm": 0.43805304169654846,
      "learning_rate": 7.300343984417092e-07,
      "loss": 0.6522,
      "step": 20607
    },
    {
      "epoch": 0.8540760081230055,
      "grad_norm": 0.4048362076282501,
      "learning_rate": 7.298271789133408e-07,
      "loss": 0.6511,
      "step": 20608
    },
    {
      "epoch": 0.8541174520286792,
      "grad_norm": 0.4062684178352356,
      "learning_rate": 7.296199593849725e-07,
      "loss": 0.6721,
      "step": 20609
    },
    {
      "epoch": 0.8541588959343529,
      "grad_norm": 0.4015860855579376,
      "learning_rate": 7.294127398566042e-07,
      "loss": 0.6643,
      "step": 20610
    },
    {
      "epoch": 0.8542003398400265,
      "grad_norm": 0.42164140939712524,
      "learning_rate": 7.292055203282358e-07,
      "loss": 0.6163,
      "step": 20611
    },
    {
      "epoch": 0.8542417837457001,
      "grad_norm": 0.40147626399993896,
      "learning_rate": 7.289983007998674e-07,
      "loss": 0.6226,
      "step": 20612
    },
    {
      "epoch": 0.8542832276513739,
      "grad_norm": 0.38960352540016174,
      "learning_rate": 7.287910812714991e-07,
      "loss": 0.6548,
      "step": 20613
    },
    {
      "epoch": 0.8543246715570475,
      "grad_norm": 0.43109095096588135,
      "learning_rate": 7.285838617431307e-07,
      "loss": 0.6254,
      "step": 20614
    },
    {
      "epoch": 0.8543661154627212,
      "grad_norm": 0.4323986768722534,
      "learning_rate": 7.283766422147624e-07,
      "loss": 0.7076,
      "step": 20615
    },
    {
      "epoch": 0.8544075593683949,
      "grad_norm": 0.37228113412857056,
      "learning_rate": 7.28169422686394e-07,
      "loss": 0.6283,
      "step": 20616
    },
    {
      "epoch": 0.8544490032740686,
      "grad_norm": 0.4250946342945099,
      "learning_rate": 7.279622031580256e-07,
      "loss": 0.6672,
      "step": 20617
    },
    {
      "epoch": 0.8544904471797422,
      "grad_norm": 0.42265087366104126,
      "learning_rate": 7.277549836296573e-07,
      "loss": 0.704,
      "step": 20618
    },
    {
      "epoch": 0.8545318910854159,
      "grad_norm": 0.3783937394618988,
      "learning_rate": 7.27547764101289e-07,
      "loss": 0.6168,
      "step": 20619
    },
    {
      "epoch": 0.8545733349910896,
      "grad_norm": 0.42652684450149536,
      "learning_rate": 7.273405445729206e-07,
      "loss": 0.6614,
      "step": 20620
    },
    {
      "epoch": 0.8546147788967632,
      "grad_norm": 0.4222947359085083,
      "learning_rate": 7.271333250445522e-07,
      "loss": 0.6978,
      "step": 20621
    },
    {
      "epoch": 0.8546562228024369,
      "grad_norm": 0.43025317788124084,
      "learning_rate": 7.269261055161839e-07,
      "loss": 0.6201,
      "step": 20622
    },
    {
      "epoch": 0.8546976667081105,
      "grad_norm": 0.4381261169910431,
      "learning_rate": 7.267188859878156e-07,
      "loss": 0.6934,
      "step": 20623
    },
    {
      "epoch": 0.8547391106137843,
      "grad_norm": 0.669762134552002,
      "learning_rate": 7.265116664594472e-07,
      "loss": 0.7092,
      "step": 20624
    },
    {
      "epoch": 0.8547805545194579,
      "grad_norm": 0.39167869091033936,
      "learning_rate": 7.263044469310788e-07,
      "loss": 0.6626,
      "step": 20625
    },
    {
      "epoch": 0.8548219984251316,
      "grad_norm": 0.378739595413208,
      "learning_rate": 7.260972274027105e-07,
      "loss": 0.6135,
      "step": 20626
    },
    {
      "epoch": 0.8548634423308052,
      "grad_norm": 0.4129163324832916,
      "learning_rate": 7.258900078743421e-07,
      "loss": 0.6935,
      "step": 20627
    },
    {
      "epoch": 0.854904886236479,
      "grad_norm": 0.4739900231361389,
      "learning_rate": 7.256827883459738e-07,
      "loss": 0.7242,
      "step": 20628
    },
    {
      "epoch": 0.8549463301421526,
      "grad_norm": 0.44078561663627625,
      "learning_rate": 7.254755688176054e-07,
      "loss": 0.7068,
      "step": 20629
    },
    {
      "epoch": 0.8549877740478262,
      "grad_norm": 0.40549206733703613,
      "learning_rate": 7.252683492892371e-07,
      "loss": 0.697,
      "step": 20630
    },
    {
      "epoch": 0.8550292179535,
      "grad_norm": 0.42405807971954346,
      "learning_rate": 7.250611297608687e-07,
      "loss": 0.6987,
      "step": 20631
    },
    {
      "epoch": 0.8550706618591736,
      "grad_norm": 0.4120432734489441,
      "learning_rate": 7.248539102325004e-07,
      "loss": 0.6793,
      "step": 20632
    },
    {
      "epoch": 0.8551121057648473,
      "grad_norm": 0.4435417056083679,
      "learning_rate": 7.24646690704132e-07,
      "loss": 0.6608,
      "step": 20633
    },
    {
      "epoch": 0.8551535496705209,
      "grad_norm": 0.4202413260936737,
      "learning_rate": 7.244394711757637e-07,
      "loss": 0.6394,
      "step": 20634
    },
    {
      "epoch": 0.8551949935761947,
      "grad_norm": 0.37648656964302063,
      "learning_rate": 7.242322516473953e-07,
      "loss": 0.6343,
      "step": 20635
    },
    {
      "epoch": 0.8552364374818683,
      "grad_norm": 0.43615004420280457,
      "learning_rate": 7.240250321190269e-07,
      "loss": 0.7203,
      "step": 20636
    },
    {
      "epoch": 0.855277881387542,
      "grad_norm": 0.4045298099517822,
      "learning_rate": 7.238178125906586e-07,
      "loss": 0.6909,
      "step": 20637
    },
    {
      "epoch": 0.8553193252932156,
      "grad_norm": 0.41749417781829834,
      "learning_rate": 7.236105930622902e-07,
      "loss": 0.6331,
      "step": 20638
    },
    {
      "epoch": 0.8553607691988893,
      "grad_norm": 0.4086287319660187,
      "learning_rate": 7.234033735339219e-07,
      "loss": 0.6853,
      "step": 20639
    },
    {
      "epoch": 0.855402213104563,
      "grad_norm": 0.39390724897384644,
      "learning_rate": 7.231961540055535e-07,
      "loss": 0.6178,
      "step": 20640
    },
    {
      "epoch": 0.8554436570102366,
      "grad_norm": 0.4174099266529083,
      "learning_rate": 7.229889344771852e-07,
      "loss": 0.6641,
      "step": 20641
    },
    {
      "epoch": 0.8554851009159103,
      "grad_norm": 0.4187566339969635,
      "learning_rate": 7.227817149488168e-07,
      "loss": 0.688,
      "step": 20642
    },
    {
      "epoch": 0.855526544821584,
      "grad_norm": 0.43028339743614197,
      "learning_rate": 7.225744954204485e-07,
      "loss": 0.7334,
      "step": 20643
    },
    {
      "epoch": 0.8555679887272577,
      "grad_norm": 0.42927947640419006,
      "learning_rate": 7.223672758920801e-07,
      "loss": 0.6547,
      "step": 20644
    },
    {
      "epoch": 0.8556094326329313,
      "grad_norm": 0.4395941197872162,
      "learning_rate": 7.221600563637118e-07,
      "loss": 0.6674,
      "step": 20645
    },
    {
      "epoch": 0.8556508765386049,
      "grad_norm": 0.4393553137779236,
      "learning_rate": 7.219528368353434e-07,
      "loss": 0.6797,
      "step": 20646
    },
    {
      "epoch": 0.8556923204442787,
      "grad_norm": 0.3965934216976166,
      "learning_rate": 7.217456173069751e-07,
      "loss": 0.6257,
      "step": 20647
    },
    {
      "epoch": 0.8557337643499523,
      "grad_norm": 0.4146082401275635,
      "learning_rate": 7.215383977786067e-07,
      "loss": 0.6667,
      "step": 20648
    },
    {
      "epoch": 0.855775208255626,
      "grad_norm": 0.44024360179901123,
      "learning_rate": 7.213311782502383e-07,
      "loss": 0.7244,
      "step": 20649
    },
    {
      "epoch": 0.8558166521612997,
      "grad_norm": 0.45884424448013306,
      "learning_rate": 7.2112395872187e-07,
      "loss": 0.6631,
      "step": 20650
    },
    {
      "epoch": 0.8558580960669734,
      "grad_norm": 0.4345189929008484,
      "learning_rate": 7.209167391935017e-07,
      "loss": 0.6469,
      "step": 20651
    },
    {
      "epoch": 0.855899539972647,
      "grad_norm": 0.3816482126712799,
      "learning_rate": 7.207095196651333e-07,
      "loss": 0.673,
      "step": 20652
    },
    {
      "epoch": 0.8559409838783207,
      "grad_norm": 0.4629276394844055,
      "learning_rate": 7.205023001367649e-07,
      "loss": 0.6785,
      "step": 20653
    },
    {
      "epoch": 0.8559824277839944,
      "grad_norm": 0.43444257974624634,
      "learning_rate": 7.202950806083966e-07,
      "loss": 0.6747,
      "step": 20654
    },
    {
      "epoch": 0.856023871689668,
      "grad_norm": 0.4620898365974426,
      "learning_rate": 7.200878610800282e-07,
      "loss": 0.6879,
      "step": 20655
    },
    {
      "epoch": 0.8560653155953417,
      "grad_norm": 0.39860424399375916,
      "learning_rate": 7.198806415516599e-07,
      "loss": 0.6031,
      "step": 20656
    },
    {
      "epoch": 0.8561067595010153,
      "grad_norm": 0.37967753410339355,
      "learning_rate": 7.196734220232915e-07,
      "loss": 0.6772,
      "step": 20657
    },
    {
      "epoch": 0.8561482034066891,
      "grad_norm": 0.44143545627593994,
      "learning_rate": 7.194662024949232e-07,
      "loss": 0.6477,
      "step": 20658
    },
    {
      "epoch": 0.8561896473123627,
      "grad_norm": 0.4243814945220947,
      "learning_rate": 7.192589829665548e-07,
      "loss": 0.6567,
      "step": 20659
    },
    {
      "epoch": 0.8562310912180364,
      "grad_norm": 0.4085336923599243,
      "learning_rate": 7.190517634381865e-07,
      "loss": 0.6567,
      "step": 20660
    },
    {
      "epoch": 0.85627253512371,
      "grad_norm": 0.40385058522224426,
      "learning_rate": 7.188445439098181e-07,
      "loss": 0.6958,
      "step": 20661
    },
    {
      "epoch": 0.8563139790293838,
      "grad_norm": 0.45761188864707947,
      "learning_rate": 7.186373243814497e-07,
      "loss": 0.6851,
      "step": 20662
    },
    {
      "epoch": 0.8563554229350574,
      "grad_norm": 0.4179224371910095,
      "learning_rate": 7.184301048530814e-07,
      "loss": 0.7062,
      "step": 20663
    },
    {
      "epoch": 0.856396866840731,
      "grad_norm": 0.3983958959579468,
      "learning_rate": 7.182228853247131e-07,
      "loss": 0.6429,
      "step": 20664
    },
    {
      "epoch": 0.8564383107464048,
      "grad_norm": 0.3968222141265869,
      "learning_rate": 7.180156657963447e-07,
      "loss": 0.6581,
      "step": 20665
    },
    {
      "epoch": 0.8564797546520784,
      "grad_norm": 0.40999695658683777,
      "learning_rate": 7.178084462679763e-07,
      "loss": 0.6382,
      "step": 20666
    },
    {
      "epoch": 0.8565211985577521,
      "grad_norm": 0.41727912425994873,
      "learning_rate": 7.17601226739608e-07,
      "loss": 0.6472,
      "step": 20667
    },
    {
      "epoch": 0.8565626424634257,
      "grad_norm": 0.3964017331600189,
      "learning_rate": 7.173940072112397e-07,
      "loss": 0.6272,
      "step": 20668
    },
    {
      "epoch": 0.8566040863690995,
      "grad_norm": 0.4236403703689575,
      "learning_rate": 7.171867876828713e-07,
      "loss": 0.6945,
      "step": 20669
    },
    {
      "epoch": 0.8566455302747731,
      "grad_norm": 0.41873735189437866,
      "learning_rate": 7.169795681545029e-07,
      "loss": 0.6702,
      "step": 20670
    },
    {
      "epoch": 0.8566869741804468,
      "grad_norm": 0.40993115305900574,
      "learning_rate": 7.167723486261346e-07,
      "loss": 0.677,
      "step": 20671
    },
    {
      "epoch": 0.8567284180861204,
      "grad_norm": 0.43196114897727966,
      "learning_rate": 7.165651290977662e-07,
      "loss": 0.6758,
      "step": 20672
    },
    {
      "epoch": 0.8567698619917941,
      "grad_norm": 0.4199584424495697,
      "learning_rate": 7.163579095693979e-07,
      "loss": 0.6819,
      "step": 20673
    },
    {
      "epoch": 0.8568113058974678,
      "grad_norm": 0.43562430143356323,
      "learning_rate": 7.161506900410295e-07,
      "loss": 0.6495,
      "step": 20674
    },
    {
      "epoch": 0.8568527498031414,
      "grad_norm": 0.46757742762565613,
      "learning_rate": 7.159434705126611e-07,
      "loss": 0.7274,
      "step": 20675
    },
    {
      "epoch": 0.8568941937088151,
      "grad_norm": 0.4230787456035614,
      "learning_rate": 7.157362509842928e-07,
      "loss": 0.646,
      "step": 20676
    },
    {
      "epoch": 0.8569356376144888,
      "grad_norm": 0.39319083094596863,
      "learning_rate": 7.155290314559245e-07,
      "loss": 0.6136,
      "step": 20677
    },
    {
      "epoch": 0.8569770815201625,
      "grad_norm": 0.39663931727409363,
      "learning_rate": 7.153218119275561e-07,
      "loss": 0.636,
      "step": 20678
    },
    {
      "epoch": 0.8570185254258361,
      "grad_norm": 0.38029035925865173,
      "learning_rate": 7.151145923991877e-07,
      "loss": 0.6768,
      "step": 20679
    },
    {
      "epoch": 0.8570599693315099,
      "grad_norm": 0.4543338418006897,
      "learning_rate": 7.149073728708194e-07,
      "loss": 0.6786,
      "step": 20680
    },
    {
      "epoch": 0.8571014132371835,
      "grad_norm": 0.4540020823478699,
      "learning_rate": 7.147001533424511e-07,
      "loss": 0.7065,
      "step": 20681
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.42281222343444824,
      "learning_rate": 7.144929338140827e-07,
      "loss": 0.7029,
      "step": 20682
    },
    {
      "epoch": 0.8571843010485308,
      "grad_norm": 0.42932766675949097,
      "learning_rate": 7.142857142857143e-07,
      "loss": 0.6892,
      "step": 20683
    },
    {
      "epoch": 0.8572257449542044,
      "grad_norm": 0.36511629819869995,
      "learning_rate": 7.14078494757346e-07,
      "loss": 0.6238,
      "step": 20684
    },
    {
      "epoch": 0.8572671888598782,
      "grad_norm": 0.47147801518440247,
      "learning_rate": 7.138712752289777e-07,
      "loss": 0.7489,
      "step": 20685
    },
    {
      "epoch": 0.8573086327655518,
      "grad_norm": 0.40835800766944885,
      "learning_rate": 7.136640557006093e-07,
      "loss": 0.6415,
      "step": 20686
    },
    {
      "epoch": 0.8573500766712255,
      "grad_norm": 0.40998151898384094,
      "learning_rate": 7.134568361722409e-07,
      "loss": 0.6782,
      "step": 20687
    },
    {
      "epoch": 0.8573915205768992,
      "grad_norm": 0.4171992838382721,
      "learning_rate": 7.132496166438725e-07,
      "loss": 0.6511,
      "step": 20688
    },
    {
      "epoch": 0.8574329644825729,
      "grad_norm": 0.435227632522583,
      "learning_rate": 7.130423971155043e-07,
      "loss": 0.6809,
      "step": 20689
    },
    {
      "epoch": 0.8574744083882465,
      "grad_norm": 0.38382989168167114,
      "learning_rate": 7.128351775871359e-07,
      "loss": 0.6067,
      "step": 20690
    },
    {
      "epoch": 0.8575158522939201,
      "grad_norm": 0.3736913800239563,
      "learning_rate": 7.126279580587675e-07,
      "loss": 0.6385,
      "step": 20691
    },
    {
      "epoch": 0.8575572961995939,
      "grad_norm": 0.3784366548061371,
      "learning_rate": 7.124207385303991e-07,
      "loss": 0.6263,
      "step": 20692
    },
    {
      "epoch": 0.8575987401052675,
      "grad_norm": 0.4555272161960602,
      "learning_rate": 7.122135190020308e-07,
      "loss": 0.653,
      "step": 20693
    },
    {
      "epoch": 0.8576401840109412,
      "grad_norm": 0.4482711851596832,
      "learning_rate": 7.120062994736625e-07,
      "loss": 0.7026,
      "step": 20694
    },
    {
      "epoch": 0.8576816279166148,
      "grad_norm": 0.40066763758659363,
      "learning_rate": 7.117990799452941e-07,
      "loss": 0.6487,
      "step": 20695
    },
    {
      "epoch": 0.8577230718222886,
      "grad_norm": 0.41767817735671997,
      "learning_rate": 7.115918604169257e-07,
      "loss": 0.6357,
      "step": 20696
    },
    {
      "epoch": 0.8577645157279622,
      "grad_norm": 0.42136138677597046,
      "learning_rate": 7.113846408885574e-07,
      "loss": 0.6824,
      "step": 20697
    },
    {
      "epoch": 0.8578059596336359,
      "grad_norm": 0.41795065999031067,
      "learning_rate": 7.111774213601891e-07,
      "loss": 0.6262,
      "step": 20698
    },
    {
      "epoch": 0.8578474035393095,
      "grad_norm": 0.42105284333229065,
      "learning_rate": 7.109702018318207e-07,
      "loss": 0.6638,
      "step": 20699
    },
    {
      "epoch": 0.8578888474449832,
      "grad_norm": 0.3958751857280731,
      "learning_rate": 7.107629823034523e-07,
      "loss": 0.6282,
      "step": 20700
    },
    {
      "epoch": 0.8579302913506569,
      "grad_norm": 0.42144879698753357,
      "learning_rate": 7.105557627750839e-07,
      "loss": 0.6101,
      "step": 20701
    },
    {
      "epoch": 0.8579717352563305,
      "grad_norm": 0.40834423899650574,
      "learning_rate": 7.103485432467157e-07,
      "loss": 0.7122,
      "step": 20702
    },
    {
      "epoch": 0.8580131791620043,
      "grad_norm": 0.4218199551105499,
      "learning_rate": 7.101413237183473e-07,
      "loss": 0.6676,
      "step": 20703
    },
    {
      "epoch": 0.8580546230676779,
      "grad_norm": 0.44073861837387085,
      "learning_rate": 7.099341041899789e-07,
      "loss": 0.6641,
      "step": 20704
    },
    {
      "epoch": 0.8580960669733516,
      "grad_norm": 0.4332108199596405,
      "learning_rate": 7.097268846616105e-07,
      "loss": 0.7095,
      "step": 20705
    },
    {
      "epoch": 0.8581375108790252,
      "grad_norm": 0.4333076477050781,
      "learning_rate": 7.095196651332423e-07,
      "loss": 0.6887,
      "step": 20706
    },
    {
      "epoch": 0.8581789547846989,
      "grad_norm": 0.4310159981250763,
      "learning_rate": 7.093124456048739e-07,
      "loss": 0.6835,
      "step": 20707
    },
    {
      "epoch": 0.8582203986903726,
      "grad_norm": 0.39767298102378845,
      "learning_rate": 7.091052260765055e-07,
      "loss": 0.6073,
      "step": 20708
    },
    {
      "epoch": 0.8582618425960462,
      "grad_norm": 0.4283129572868347,
      "learning_rate": 7.088980065481371e-07,
      "loss": 0.6294,
      "step": 20709
    },
    {
      "epoch": 0.8583032865017199,
      "grad_norm": 0.42040449380874634,
      "learning_rate": 7.086907870197687e-07,
      "loss": 0.6626,
      "step": 20710
    },
    {
      "epoch": 0.8583447304073936,
      "grad_norm": 0.443744033575058,
      "learning_rate": 7.084835674914005e-07,
      "loss": 0.7006,
      "step": 20711
    },
    {
      "epoch": 0.8583861743130673,
      "grad_norm": 0.4153057932853699,
      "learning_rate": 7.082763479630321e-07,
      "loss": 0.7144,
      "step": 20712
    },
    {
      "epoch": 0.8584276182187409,
      "grad_norm": 0.4464622139930725,
      "learning_rate": 7.080691284346637e-07,
      "loss": 0.6514,
      "step": 20713
    },
    {
      "epoch": 0.8584690621244147,
      "grad_norm": 0.4239831864833832,
      "learning_rate": 7.078619089062953e-07,
      "loss": 0.6605,
      "step": 20714
    },
    {
      "epoch": 0.8585105060300883,
      "grad_norm": 0.416917622089386,
      "learning_rate": 7.076546893779271e-07,
      "loss": 0.6705,
      "step": 20715
    },
    {
      "epoch": 0.8585519499357619,
      "grad_norm": 0.43651047348976135,
      "learning_rate": 7.074474698495587e-07,
      "loss": 0.6677,
      "step": 20716
    },
    {
      "epoch": 0.8585933938414356,
      "grad_norm": 0.41689062118530273,
      "learning_rate": 7.072402503211903e-07,
      "loss": 0.653,
      "step": 20717
    },
    {
      "epoch": 0.8586348377471092,
      "grad_norm": 0.4162941575050354,
      "learning_rate": 7.070330307928219e-07,
      "loss": 0.6329,
      "step": 20718
    },
    {
      "epoch": 0.858676281652783,
      "grad_norm": 0.4778016209602356,
      "learning_rate": 7.068258112644537e-07,
      "loss": 0.7329,
      "step": 20719
    },
    {
      "epoch": 0.8587177255584566,
      "grad_norm": 0.4138352870941162,
      "learning_rate": 7.066185917360853e-07,
      "loss": 0.6691,
      "step": 20720
    },
    {
      "epoch": 0.8587591694641303,
      "grad_norm": 0.43397819995880127,
      "learning_rate": 7.064113722077169e-07,
      "loss": 0.675,
      "step": 20721
    },
    {
      "epoch": 0.858800613369804,
      "grad_norm": 0.44667157530784607,
      "learning_rate": 7.062041526793485e-07,
      "loss": 0.6838,
      "step": 20722
    },
    {
      "epoch": 0.8588420572754777,
      "grad_norm": 0.4274214506149292,
      "learning_rate": 7.059969331509801e-07,
      "loss": 0.6847,
      "step": 20723
    },
    {
      "epoch": 0.8588835011811513,
      "grad_norm": 0.4447210729122162,
      "learning_rate": 7.057897136226119e-07,
      "loss": 0.6104,
      "step": 20724
    },
    {
      "epoch": 0.8589249450868249,
      "grad_norm": 0.440625935792923,
      "learning_rate": 7.055824940942435e-07,
      "loss": 0.7159,
      "step": 20725
    },
    {
      "epoch": 0.8589663889924987,
      "grad_norm": 0.4345286190509796,
      "learning_rate": 7.053752745658751e-07,
      "loss": 0.6782,
      "step": 20726
    },
    {
      "epoch": 0.8590078328981723,
      "grad_norm": 0.44737088680267334,
      "learning_rate": 7.051680550375067e-07,
      "loss": 0.7064,
      "step": 20727
    },
    {
      "epoch": 0.859049276803846,
      "grad_norm": 0.40545275807380676,
      "learning_rate": 7.049608355091385e-07,
      "loss": 0.6648,
      "step": 20728
    },
    {
      "epoch": 0.8590907207095196,
      "grad_norm": 0.43794143199920654,
      "learning_rate": 7.047536159807701e-07,
      "loss": 0.688,
      "step": 20729
    },
    {
      "epoch": 0.8591321646151934,
      "grad_norm": 0.4279376268386841,
      "learning_rate": 7.045463964524017e-07,
      "loss": 0.6183,
      "step": 20730
    },
    {
      "epoch": 0.859173608520867,
      "grad_norm": 0.4055424928665161,
      "learning_rate": 7.043391769240333e-07,
      "loss": 0.6619,
      "step": 20731
    },
    {
      "epoch": 0.8592150524265407,
      "grad_norm": 0.4232601523399353,
      "learning_rate": 7.041319573956651e-07,
      "loss": 0.6921,
      "step": 20732
    },
    {
      "epoch": 0.8592564963322143,
      "grad_norm": 0.3670955002307892,
      "learning_rate": 7.039247378672967e-07,
      "loss": 0.5907,
      "step": 20733
    },
    {
      "epoch": 0.859297940237888,
      "grad_norm": 0.41845718026161194,
      "learning_rate": 7.037175183389283e-07,
      "loss": 0.6978,
      "step": 20734
    },
    {
      "epoch": 0.8593393841435617,
      "grad_norm": 0.4219863712787628,
      "learning_rate": 7.035102988105599e-07,
      "loss": 0.616,
      "step": 20735
    },
    {
      "epoch": 0.8593808280492353,
      "grad_norm": 0.41098758578300476,
      "learning_rate": 7.033030792821915e-07,
      "loss": 0.6733,
      "step": 20736
    },
    {
      "epoch": 0.8594222719549091,
      "grad_norm": 0.4137190878391266,
      "learning_rate": 7.030958597538233e-07,
      "loss": 0.6747,
      "step": 20737
    },
    {
      "epoch": 0.8594637158605827,
      "grad_norm": 0.38438576459884644,
      "learning_rate": 7.028886402254549e-07,
      "loss": 0.6543,
      "step": 20738
    },
    {
      "epoch": 0.8595051597662564,
      "grad_norm": 0.3769966959953308,
      "learning_rate": 7.026814206970865e-07,
      "loss": 0.6523,
      "step": 20739
    },
    {
      "epoch": 0.85954660367193,
      "grad_norm": 0.42673033475875854,
      "learning_rate": 7.024742011687181e-07,
      "loss": 0.7256,
      "step": 20740
    },
    {
      "epoch": 0.8595880475776038,
      "grad_norm": 0.44293245673179626,
      "learning_rate": 7.022669816403499e-07,
      "loss": 0.6707,
      "step": 20741
    },
    {
      "epoch": 0.8596294914832774,
      "grad_norm": 0.4170716404914856,
      "learning_rate": 7.020597621119815e-07,
      "loss": 0.7061,
      "step": 20742
    },
    {
      "epoch": 0.859670935388951,
      "grad_norm": 0.4361152648925781,
      "learning_rate": 7.018525425836131e-07,
      "loss": 0.6967,
      "step": 20743
    },
    {
      "epoch": 0.8597123792946247,
      "grad_norm": 0.4424999952316284,
      "learning_rate": 7.016453230552447e-07,
      "loss": 0.6709,
      "step": 20744
    },
    {
      "epoch": 0.8597538232002984,
      "grad_norm": 0.41426846385002136,
      "learning_rate": 7.014381035268765e-07,
      "loss": 0.6814,
      "step": 20745
    },
    {
      "epoch": 0.8597952671059721,
      "grad_norm": 0.43202832341194153,
      "learning_rate": 7.012308839985081e-07,
      "loss": 0.6311,
      "step": 20746
    },
    {
      "epoch": 0.8598367110116457,
      "grad_norm": 0.435259610414505,
      "learning_rate": 7.010236644701397e-07,
      "loss": 0.7069,
      "step": 20747
    },
    {
      "epoch": 0.8598781549173194,
      "grad_norm": 0.4061526358127594,
      "learning_rate": 7.008164449417713e-07,
      "loss": 0.6169,
      "step": 20748
    },
    {
      "epoch": 0.8599195988229931,
      "grad_norm": 0.37928295135498047,
      "learning_rate": 7.006092254134029e-07,
      "loss": 0.6378,
      "step": 20749
    },
    {
      "epoch": 0.8599610427286668,
      "grad_norm": 0.4216063320636749,
      "learning_rate": 7.004020058850347e-07,
      "loss": 0.696,
      "step": 20750
    },
    {
      "epoch": 0.8600024866343404,
      "grad_norm": 0.4561203122138977,
      "learning_rate": 7.001947863566663e-07,
      "loss": 0.6469,
      "step": 20751
    },
    {
      "epoch": 0.860043930540014,
      "grad_norm": 0.4967528283596039,
      "learning_rate": 6.999875668282979e-07,
      "loss": 0.7694,
      "step": 20752
    },
    {
      "epoch": 0.8600853744456878,
      "grad_norm": 0.4147067964076996,
      "learning_rate": 6.997803472999295e-07,
      "loss": 0.6323,
      "step": 20753
    },
    {
      "epoch": 0.8601268183513614,
      "grad_norm": 0.4385703504085541,
      "learning_rate": 6.995731277715613e-07,
      "loss": 0.7288,
      "step": 20754
    },
    {
      "epoch": 0.8601682622570351,
      "grad_norm": 0.4238215386867523,
      "learning_rate": 6.993659082431929e-07,
      "loss": 0.645,
      "step": 20755
    },
    {
      "epoch": 0.8602097061627088,
      "grad_norm": 0.48376473784446716,
      "learning_rate": 6.991586887148245e-07,
      "loss": 0.6849,
      "step": 20756
    },
    {
      "epoch": 0.8602511500683825,
      "grad_norm": 0.42255550622940063,
      "learning_rate": 6.989514691864561e-07,
      "loss": 0.6047,
      "step": 20757
    },
    {
      "epoch": 0.8602925939740561,
      "grad_norm": 0.41737762093544006,
      "learning_rate": 6.987442496580879e-07,
      "loss": 0.6841,
      "step": 20758
    },
    {
      "epoch": 0.8603340378797298,
      "grad_norm": 0.3802061080932617,
      "learning_rate": 6.985370301297195e-07,
      "loss": 0.5896,
      "step": 20759
    },
    {
      "epoch": 0.8603754817854035,
      "grad_norm": 0.4072643518447876,
      "learning_rate": 6.983298106013511e-07,
      "loss": 0.6243,
      "step": 20760
    },
    {
      "epoch": 0.8604169256910771,
      "grad_norm": 0.4043499231338501,
      "learning_rate": 6.981225910729827e-07,
      "loss": 0.6992,
      "step": 20761
    },
    {
      "epoch": 0.8604583695967508,
      "grad_norm": 0.4156593978404999,
      "learning_rate": 6.979153715446143e-07,
      "loss": 0.6743,
      "step": 20762
    },
    {
      "epoch": 0.8604998135024244,
      "grad_norm": 0.4169057309627533,
      "learning_rate": 6.977081520162461e-07,
      "loss": 0.642,
      "step": 20763
    },
    {
      "epoch": 0.8605412574080982,
      "grad_norm": 0.461443156003952,
      "learning_rate": 6.975009324878777e-07,
      "loss": 0.647,
      "step": 20764
    },
    {
      "epoch": 0.8605827013137718,
      "grad_norm": 0.4311562180519104,
      "learning_rate": 6.972937129595093e-07,
      "loss": 0.6526,
      "step": 20765
    },
    {
      "epoch": 0.8606241452194455,
      "grad_norm": 0.3944648206233978,
      "learning_rate": 6.970864934311409e-07,
      "loss": 0.6416,
      "step": 20766
    },
    {
      "epoch": 0.8606655891251191,
      "grad_norm": 0.4345790147781372,
      "learning_rate": 6.968792739027727e-07,
      "loss": 0.6672,
      "step": 20767
    },
    {
      "epoch": 0.8607070330307928,
      "grad_norm": 0.4515756368637085,
      "learning_rate": 6.966720543744043e-07,
      "loss": 0.7393,
      "step": 20768
    },
    {
      "epoch": 0.8607484769364665,
      "grad_norm": 0.45680031180381775,
      "learning_rate": 6.964648348460359e-07,
      "loss": 0.6946,
      "step": 20769
    },
    {
      "epoch": 0.8607899208421401,
      "grad_norm": 0.42079228162765503,
      "learning_rate": 6.962576153176675e-07,
      "loss": 0.6348,
      "step": 20770
    },
    {
      "epoch": 0.8608313647478139,
      "grad_norm": 0.3976905047893524,
      "learning_rate": 6.960503957892991e-07,
      "loss": 0.6373,
      "step": 20771
    },
    {
      "epoch": 0.8608728086534875,
      "grad_norm": 0.4169802665710449,
      "learning_rate": 6.958431762609309e-07,
      "loss": 0.6427,
      "step": 20772
    },
    {
      "epoch": 0.8609142525591612,
      "grad_norm": 0.4299662709236145,
      "learning_rate": 6.956359567325625e-07,
      "loss": 0.606,
      "step": 20773
    },
    {
      "epoch": 0.8609556964648348,
      "grad_norm": 0.4135642647743225,
      "learning_rate": 6.954287372041941e-07,
      "loss": 0.7047,
      "step": 20774
    },
    {
      "epoch": 0.8609971403705086,
      "grad_norm": 0.4326569139957428,
      "learning_rate": 6.952215176758257e-07,
      "loss": 0.6757,
      "step": 20775
    },
    {
      "epoch": 0.8610385842761822,
      "grad_norm": 0.4333491623401642,
      "learning_rate": 6.950142981474575e-07,
      "loss": 0.6245,
      "step": 20776
    },
    {
      "epoch": 0.8610800281818558,
      "grad_norm": 0.4208334982395172,
      "learning_rate": 6.948070786190891e-07,
      "loss": 0.6798,
      "step": 20777
    },
    {
      "epoch": 0.8611214720875295,
      "grad_norm": 0.4477040767669678,
      "learning_rate": 6.945998590907207e-07,
      "loss": 0.6674,
      "step": 20778
    },
    {
      "epoch": 0.8611629159932032,
      "grad_norm": 0.3821337819099426,
      "learning_rate": 6.943926395623523e-07,
      "loss": 0.6428,
      "step": 20779
    },
    {
      "epoch": 0.8612043598988769,
      "grad_norm": 0.4158639907836914,
      "learning_rate": 6.941854200339841e-07,
      "loss": 0.6562,
      "step": 20780
    },
    {
      "epoch": 0.8612458038045505,
      "grad_norm": 0.39476126432418823,
      "learning_rate": 6.939782005056157e-07,
      "loss": 0.6078,
      "step": 20781
    },
    {
      "epoch": 0.8612872477102242,
      "grad_norm": 0.3789387345314026,
      "learning_rate": 6.937709809772473e-07,
      "loss": 0.6025,
      "step": 20782
    },
    {
      "epoch": 0.8613286916158979,
      "grad_norm": 0.4036412239074707,
      "learning_rate": 6.935637614488789e-07,
      "loss": 0.6697,
      "step": 20783
    },
    {
      "epoch": 0.8613701355215716,
      "grad_norm": 0.40359246730804443,
      "learning_rate": 6.933565419205106e-07,
      "loss": 0.6848,
      "step": 20784
    },
    {
      "epoch": 0.8614115794272452,
      "grad_norm": 0.40351003408432007,
      "learning_rate": 6.931493223921423e-07,
      "loss": 0.6931,
      "step": 20785
    },
    {
      "epoch": 0.8614530233329188,
      "grad_norm": 0.4472194015979767,
      "learning_rate": 6.929421028637739e-07,
      "loss": 0.7092,
      "step": 20786
    },
    {
      "epoch": 0.8614944672385926,
      "grad_norm": 0.43898746371269226,
      "learning_rate": 6.927348833354055e-07,
      "loss": 0.7178,
      "step": 20787
    },
    {
      "epoch": 0.8615359111442662,
      "grad_norm": 0.42746561765670776,
      "learning_rate": 6.925276638070371e-07,
      "loss": 0.6306,
      "step": 20788
    },
    {
      "epoch": 0.8615773550499399,
      "grad_norm": 0.41697919368743896,
      "learning_rate": 6.923204442786689e-07,
      "loss": 0.7349,
      "step": 20789
    },
    {
      "epoch": 0.8616187989556136,
      "grad_norm": 0.4274982213973999,
      "learning_rate": 6.921132247503005e-07,
      "loss": 0.6534,
      "step": 20790
    },
    {
      "epoch": 0.8616602428612873,
      "grad_norm": 0.38506534695625305,
      "learning_rate": 6.919060052219321e-07,
      "loss": 0.6274,
      "step": 20791
    },
    {
      "epoch": 0.8617016867669609,
      "grad_norm": 0.41367805004119873,
      "learning_rate": 6.916987856935637e-07,
      "loss": 0.6333,
      "step": 20792
    },
    {
      "epoch": 0.8617431306726346,
      "grad_norm": 0.4192332923412323,
      "learning_rate": 6.914915661651955e-07,
      "loss": 0.6528,
      "step": 20793
    },
    {
      "epoch": 0.8617845745783083,
      "grad_norm": 0.4149060845375061,
      "learning_rate": 6.912843466368271e-07,
      "loss": 0.6738,
      "step": 20794
    },
    {
      "epoch": 0.8618260184839819,
      "grad_norm": 0.39049437642097473,
      "learning_rate": 6.910771271084587e-07,
      "loss": 0.6201,
      "step": 20795
    },
    {
      "epoch": 0.8618674623896556,
      "grad_norm": 0.40260255336761475,
      "learning_rate": 6.908699075800903e-07,
      "loss": 0.6678,
      "step": 20796
    },
    {
      "epoch": 0.8619089062953292,
      "grad_norm": 0.4282781481742859,
      "learning_rate": 6.90662688051722e-07,
      "loss": 0.6361,
      "step": 20797
    },
    {
      "epoch": 0.861950350201003,
      "grad_norm": 0.3925972282886505,
      "learning_rate": 6.904554685233537e-07,
      "loss": 0.6587,
      "step": 20798
    },
    {
      "epoch": 0.8619917941066766,
      "grad_norm": 0.46792784333229065,
      "learning_rate": 6.902482489949853e-07,
      "loss": 0.7544,
      "step": 20799
    },
    {
      "epoch": 0.8620332380123503,
      "grad_norm": 0.3964206576347351,
      "learning_rate": 6.900410294666169e-07,
      "loss": 0.6342,
      "step": 20800
    },
    {
      "epoch": 0.8620746819180239,
      "grad_norm": 0.42005443572998047,
      "learning_rate": 6.898338099382486e-07,
      "loss": 0.7211,
      "step": 20801
    },
    {
      "epoch": 0.8621161258236977,
      "grad_norm": 0.4279838800430298,
      "learning_rate": 6.896265904098803e-07,
      "loss": 0.6705,
      "step": 20802
    },
    {
      "epoch": 0.8621575697293713,
      "grad_norm": 0.44034284353256226,
      "learning_rate": 6.894193708815119e-07,
      "loss": 0.6841,
      "step": 20803
    },
    {
      "epoch": 0.8621990136350449,
      "grad_norm": 0.4010227620601654,
      "learning_rate": 6.892121513531435e-07,
      "loss": 0.644,
      "step": 20804
    },
    {
      "epoch": 0.8622404575407187,
      "grad_norm": 0.44678133726119995,
      "learning_rate": 6.890049318247752e-07,
      "loss": 0.6697,
      "step": 20805
    },
    {
      "epoch": 0.8622819014463923,
      "grad_norm": 0.40810662508010864,
      "learning_rate": 6.887977122964069e-07,
      "loss": 0.6929,
      "step": 20806
    },
    {
      "epoch": 0.862323345352066,
      "grad_norm": 0.42817357182502747,
      "learning_rate": 6.885904927680385e-07,
      "loss": 0.7024,
      "step": 20807
    },
    {
      "epoch": 0.8623647892577396,
      "grad_norm": 0.4252474308013916,
      "learning_rate": 6.883832732396701e-07,
      "loss": 0.6604,
      "step": 20808
    },
    {
      "epoch": 0.8624062331634134,
      "grad_norm": 0.4117342531681061,
      "learning_rate": 6.881760537113017e-07,
      "loss": 0.7184,
      "step": 20809
    },
    {
      "epoch": 0.862447677069087,
      "grad_norm": 0.41226378083229065,
      "learning_rate": 6.879688341829334e-07,
      "loss": 0.645,
      "step": 20810
    },
    {
      "epoch": 0.8624891209747607,
      "grad_norm": 0.4464651346206665,
      "learning_rate": 6.877616146545651e-07,
      "loss": 0.6675,
      "step": 20811
    },
    {
      "epoch": 0.8625305648804343,
      "grad_norm": 0.38623860478401184,
      "learning_rate": 6.875543951261967e-07,
      "loss": 0.6044,
      "step": 20812
    },
    {
      "epoch": 0.862572008786108,
      "grad_norm": 0.42095017433166504,
      "learning_rate": 6.873471755978283e-07,
      "loss": 0.7186,
      "step": 20813
    },
    {
      "epoch": 0.8626134526917817,
      "grad_norm": 0.42698630690574646,
      "learning_rate": 6.8713995606946e-07,
      "loss": 0.7003,
      "step": 20814
    },
    {
      "epoch": 0.8626548965974553,
      "grad_norm": 0.40745851397514343,
      "learning_rate": 6.869327365410917e-07,
      "loss": 0.6466,
      "step": 20815
    },
    {
      "epoch": 0.862696340503129,
      "grad_norm": 0.4222664535045624,
      "learning_rate": 6.867255170127233e-07,
      "loss": 0.6536,
      "step": 20816
    },
    {
      "epoch": 0.8627377844088027,
      "grad_norm": 0.4130344092845917,
      "learning_rate": 6.865182974843549e-07,
      "loss": 0.6838,
      "step": 20817
    },
    {
      "epoch": 0.8627792283144764,
      "grad_norm": 0.42726221680641174,
      "learning_rate": 6.863110779559866e-07,
      "loss": 0.6301,
      "step": 20818
    },
    {
      "epoch": 0.86282067222015,
      "grad_norm": 0.4120759963989258,
      "learning_rate": 6.861038584276183e-07,
      "loss": 0.6868,
      "step": 20819
    },
    {
      "epoch": 0.8628621161258238,
      "grad_norm": 0.49347439408302307,
      "learning_rate": 6.858966388992499e-07,
      "loss": 0.7289,
      "step": 20820
    },
    {
      "epoch": 0.8629035600314974,
      "grad_norm": 0.4265531897544861,
      "learning_rate": 6.856894193708815e-07,
      "loss": 0.6593,
      "step": 20821
    },
    {
      "epoch": 0.862945003937171,
      "grad_norm": 0.4273729622364044,
      "learning_rate": 6.854821998425132e-07,
      "loss": 0.7057,
      "step": 20822
    },
    {
      "epoch": 0.8629864478428447,
      "grad_norm": 0.4245201349258423,
      "learning_rate": 6.852749803141448e-07,
      "loss": 0.6505,
      "step": 20823
    },
    {
      "epoch": 0.8630278917485183,
      "grad_norm": 0.4108963906764984,
      "learning_rate": 6.850677607857765e-07,
      "loss": 0.6798,
      "step": 20824
    },
    {
      "epoch": 0.8630693356541921,
      "grad_norm": 0.4078311324119568,
      "learning_rate": 6.848605412574081e-07,
      "loss": 0.6597,
      "step": 20825
    },
    {
      "epoch": 0.8631107795598657,
      "grad_norm": 0.4702009856700897,
      "learning_rate": 6.846533217290397e-07,
      "loss": 0.7085,
      "step": 20826
    },
    {
      "epoch": 0.8631522234655394,
      "grad_norm": 0.40734416246414185,
      "learning_rate": 6.844461022006714e-07,
      "loss": 0.6641,
      "step": 20827
    },
    {
      "epoch": 0.8631936673712131,
      "grad_norm": 0.4306730329990387,
      "learning_rate": 6.842388826723031e-07,
      "loss": 0.634,
      "step": 20828
    },
    {
      "epoch": 0.8632351112768867,
      "grad_norm": 0.43333733081817627,
      "learning_rate": 6.840316631439347e-07,
      "loss": 0.6533,
      "step": 20829
    },
    {
      "epoch": 0.8632765551825604,
      "grad_norm": 0.4167692959308624,
      "learning_rate": 6.838244436155663e-07,
      "loss": 0.6749,
      "step": 20830
    },
    {
      "epoch": 0.863317999088234,
      "grad_norm": 0.4385184049606323,
      "learning_rate": 6.83617224087198e-07,
      "loss": 0.7112,
      "step": 20831
    },
    {
      "epoch": 0.8633594429939078,
      "grad_norm": 0.45269450545310974,
      "learning_rate": 6.834100045588297e-07,
      "loss": 0.7019,
      "step": 20832
    },
    {
      "epoch": 0.8634008868995814,
      "grad_norm": 0.4097804129123688,
      "learning_rate": 6.832027850304613e-07,
      "loss": 0.6222,
      "step": 20833
    },
    {
      "epoch": 0.8634423308052551,
      "grad_norm": 0.4247879981994629,
      "learning_rate": 6.829955655020929e-07,
      "loss": 0.6796,
      "step": 20834
    },
    {
      "epoch": 0.8634837747109287,
      "grad_norm": 0.4304753541946411,
      "learning_rate": 6.827883459737246e-07,
      "loss": 0.6914,
      "step": 20835
    },
    {
      "epoch": 0.8635252186166025,
      "grad_norm": 0.3951328992843628,
      "learning_rate": 6.825811264453562e-07,
      "loss": 0.5956,
      "step": 20836
    },
    {
      "epoch": 0.8635666625222761,
      "grad_norm": 0.42578229308128357,
      "learning_rate": 6.823739069169879e-07,
      "loss": 0.6713,
      "step": 20837
    },
    {
      "epoch": 0.8636081064279497,
      "grad_norm": 0.4293753206729889,
      "learning_rate": 6.821666873886195e-07,
      "loss": 0.6216,
      "step": 20838
    },
    {
      "epoch": 0.8636495503336235,
      "grad_norm": 0.4271979331970215,
      "learning_rate": 6.819594678602512e-07,
      "loss": 0.7095,
      "step": 20839
    },
    {
      "epoch": 0.8636909942392971,
      "grad_norm": 0.44095200300216675,
      "learning_rate": 6.817522483318828e-07,
      "loss": 0.6804,
      "step": 20840
    },
    {
      "epoch": 0.8637324381449708,
      "grad_norm": 0.42021045088768005,
      "learning_rate": 6.815450288035145e-07,
      "loss": 0.6577,
      "step": 20841
    },
    {
      "epoch": 0.8637738820506444,
      "grad_norm": 0.4489968419075012,
      "learning_rate": 6.813378092751461e-07,
      "loss": 0.6763,
      "step": 20842
    },
    {
      "epoch": 0.8638153259563182,
      "grad_norm": 0.4154908359050751,
      "learning_rate": 6.811305897467777e-07,
      "loss": 0.6234,
      "step": 20843
    },
    {
      "epoch": 0.8638567698619918,
      "grad_norm": 0.39254358410835266,
      "learning_rate": 6.809233702184094e-07,
      "loss": 0.6383,
      "step": 20844
    },
    {
      "epoch": 0.8638982137676655,
      "grad_norm": 0.36391326785087585,
      "learning_rate": 6.80716150690041e-07,
      "loss": 0.6,
      "step": 20845
    },
    {
      "epoch": 0.8639396576733391,
      "grad_norm": 0.41372087597846985,
      "learning_rate": 6.805089311616727e-07,
      "loss": 0.6748,
      "step": 20846
    },
    {
      "epoch": 0.8639811015790128,
      "grad_norm": 0.42433053255081177,
      "learning_rate": 6.803017116333043e-07,
      "loss": 0.6451,
      "step": 20847
    },
    {
      "epoch": 0.8640225454846865,
      "grad_norm": 0.4240246117115021,
      "learning_rate": 6.80094492104936e-07,
      "loss": 0.6644,
      "step": 20848
    },
    {
      "epoch": 0.8640639893903601,
      "grad_norm": 0.4566548764705658,
      "learning_rate": 6.798872725765676e-07,
      "loss": 0.6868,
      "step": 20849
    },
    {
      "epoch": 0.8641054332960338,
      "grad_norm": 0.39275214076042175,
      "learning_rate": 6.796800530481993e-07,
      "loss": 0.652,
      "step": 20850
    },
    {
      "epoch": 0.8641468772017075,
      "grad_norm": 0.40007081627845764,
      "learning_rate": 6.794728335198309e-07,
      "loss": 0.6448,
      "step": 20851
    },
    {
      "epoch": 0.8641883211073812,
      "grad_norm": 0.42563021183013916,
      "learning_rate": 6.792656139914626e-07,
      "loss": 0.6583,
      "step": 20852
    },
    {
      "epoch": 0.8642297650130548,
      "grad_norm": 0.4261209964752197,
      "learning_rate": 6.790583944630942e-07,
      "loss": 0.655,
      "step": 20853
    },
    {
      "epoch": 0.8642712089187286,
      "grad_norm": 0.4443355202674866,
      "learning_rate": 6.788511749347259e-07,
      "loss": 0.7251,
      "step": 20854
    },
    {
      "epoch": 0.8643126528244022,
      "grad_norm": 0.4288378953933716,
      "learning_rate": 6.786439554063575e-07,
      "loss": 0.6267,
      "step": 20855
    },
    {
      "epoch": 0.8643540967300758,
      "grad_norm": 0.40441545844078064,
      "learning_rate": 6.784367358779892e-07,
      "loss": 0.657,
      "step": 20856
    },
    {
      "epoch": 0.8643955406357495,
      "grad_norm": 0.4161124527454376,
      "learning_rate": 6.782295163496208e-07,
      "loss": 0.6987,
      "step": 20857
    },
    {
      "epoch": 0.8644369845414231,
      "grad_norm": 0.42267096042633057,
      "learning_rate": 6.780222968212524e-07,
      "loss": 0.6572,
      "step": 20858
    },
    {
      "epoch": 0.8644784284470969,
      "grad_norm": 0.42510706186294556,
      "learning_rate": 6.778150772928841e-07,
      "loss": 0.6505,
      "step": 20859
    },
    {
      "epoch": 0.8645198723527705,
      "grad_norm": 0.3682912290096283,
      "learning_rate": 6.776078577645158e-07,
      "loss": 0.7031,
      "step": 20860
    },
    {
      "epoch": 0.8645613162584442,
      "grad_norm": 0.4171193838119507,
      "learning_rate": 6.774006382361474e-07,
      "loss": 0.7207,
      "step": 20861
    },
    {
      "epoch": 0.8646027601641179,
      "grad_norm": 0.4350731372833252,
      "learning_rate": 6.77193418707779e-07,
      "loss": 0.6917,
      "step": 20862
    },
    {
      "epoch": 0.8646442040697916,
      "grad_norm": 0.4271436035633087,
      "learning_rate": 6.769861991794107e-07,
      "loss": 0.6476,
      "step": 20863
    },
    {
      "epoch": 0.8646856479754652,
      "grad_norm": 0.45376431941986084,
      "learning_rate": 6.767789796510423e-07,
      "loss": 0.6948,
      "step": 20864
    },
    {
      "epoch": 0.8647270918811388,
      "grad_norm": 0.4329504370689392,
      "learning_rate": 6.76571760122674e-07,
      "loss": 0.7001,
      "step": 20865
    },
    {
      "epoch": 0.8647685357868126,
      "grad_norm": 0.400450736284256,
      "learning_rate": 6.763645405943056e-07,
      "loss": 0.6722,
      "step": 20866
    },
    {
      "epoch": 0.8648099796924862,
      "grad_norm": 0.40707460045814514,
      "learning_rate": 6.761573210659373e-07,
      "loss": 0.6364,
      "step": 20867
    },
    {
      "epoch": 0.8648514235981599,
      "grad_norm": 0.42245185375213623,
      "learning_rate": 6.759501015375689e-07,
      "loss": 0.6653,
      "step": 20868
    },
    {
      "epoch": 0.8648928675038335,
      "grad_norm": 0.4060738682746887,
      "learning_rate": 6.757428820092006e-07,
      "loss": 0.6433,
      "step": 20869
    },
    {
      "epoch": 0.8649343114095073,
      "grad_norm": 0.40571385622024536,
      "learning_rate": 6.755356624808322e-07,
      "loss": 0.6207,
      "step": 20870
    },
    {
      "epoch": 0.8649757553151809,
      "grad_norm": 0.48484402894973755,
      "learning_rate": 6.753284429524638e-07,
      "loss": 0.703,
      "step": 20871
    },
    {
      "epoch": 0.8650171992208546,
      "grad_norm": 0.430320143699646,
      "learning_rate": 6.751212234240955e-07,
      "loss": 0.7029,
      "step": 20872
    },
    {
      "epoch": 0.8650586431265282,
      "grad_norm": 0.4147166311740875,
      "learning_rate": 6.749140038957272e-07,
      "loss": 0.6689,
      "step": 20873
    },
    {
      "epoch": 0.8651000870322019,
      "grad_norm": 0.4049677550792694,
      "learning_rate": 6.747067843673588e-07,
      "loss": 0.6838,
      "step": 20874
    },
    {
      "epoch": 0.8651415309378756,
      "grad_norm": 0.4344314932823181,
      "learning_rate": 6.744995648389904e-07,
      "loss": 0.7153,
      "step": 20875
    },
    {
      "epoch": 0.8651829748435492,
      "grad_norm": 0.4016934931278229,
      "learning_rate": 6.742923453106221e-07,
      "loss": 0.6401,
      "step": 20876
    },
    {
      "epoch": 0.865224418749223,
      "grad_norm": 0.4326065182685852,
      "learning_rate": 6.740851257822538e-07,
      "loss": 0.6298,
      "step": 20877
    },
    {
      "epoch": 0.8652658626548966,
      "grad_norm": 0.40421611070632935,
      "learning_rate": 6.738779062538854e-07,
      "loss": 0.6266,
      "step": 20878
    },
    {
      "epoch": 0.8653073065605703,
      "grad_norm": 0.43931135535240173,
      "learning_rate": 6.73670686725517e-07,
      "loss": 0.6692,
      "step": 20879
    },
    {
      "epoch": 0.8653487504662439,
      "grad_norm": 0.44456881284713745,
      "learning_rate": 6.734634671971487e-07,
      "loss": 0.7,
      "step": 20880
    },
    {
      "epoch": 0.8653901943719177,
      "grad_norm": 0.4109469950199127,
      "learning_rate": 6.732562476687803e-07,
      "loss": 0.6843,
      "step": 20881
    },
    {
      "epoch": 0.8654316382775913,
      "grad_norm": 0.4496559500694275,
      "learning_rate": 6.73049028140412e-07,
      "loss": 0.6367,
      "step": 20882
    },
    {
      "epoch": 0.8654730821832649,
      "grad_norm": 0.4600655436515808,
      "learning_rate": 6.728418086120436e-07,
      "loss": 0.7131,
      "step": 20883
    },
    {
      "epoch": 0.8655145260889386,
      "grad_norm": 0.432919979095459,
      "learning_rate": 6.726345890836752e-07,
      "loss": 0.6924,
      "step": 20884
    },
    {
      "epoch": 0.8655559699946123,
      "grad_norm": 0.4070434272289276,
      "learning_rate": 6.724273695553069e-07,
      "loss": 0.62,
      "step": 20885
    },
    {
      "epoch": 0.865597413900286,
      "grad_norm": 0.4597512185573578,
      "learning_rate": 6.722201500269386e-07,
      "loss": 0.6699,
      "step": 20886
    },
    {
      "epoch": 0.8656388578059596,
      "grad_norm": 0.44829005002975464,
      "learning_rate": 6.720129304985702e-07,
      "loss": 0.6836,
      "step": 20887
    },
    {
      "epoch": 0.8656803017116333,
      "grad_norm": 0.41864582896232605,
      "learning_rate": 6.718057109702018e-07,
      "loss": 0.6965,
      "step": 20888
    },
    {
      "epoch": 0.865721745617307,
      "grad_norm": 0.4184340238571167,
      "learning_rate": 6.715984914418335e-07,
      "loss": 0.6475,
      "step": 20889
    },
    {
      "epoch": 0.8657631895229806,
      "grad_norm": 0.4089394211769104,
      "learning_rate": 6.713912719134652e-07,
      "loss": 0.7126,
      "step": 20890
    },
    {
      "epoch": 0.8658046334286543,
      "grad_norm": 0.43311527371406555,
      "learning_rate": 6.711840523850968e-07,
      "loss": 0.6779,
      "step": 20891
    },
    {
      "epoch": 0.865846077334328,
      "grad_norm": 0.4505446255207062,
      "learning_rate": 6.709768328567284e-07,
      "loss": 0.6689,
      "step": 20892
    },
    {
      "epoch": 0.8658875212400017,
      "grad_norm": 0.43238765001296997,
      "learning_rate": 6.707696133283601e-07,
      "loss": 0.7117,
      "step": 20893
    },
    {
      "epoch": 0.8659289651456753,
      "grad_norm": 0.37029340863227844,
      "learning_rate": 6.705623937999918e-07,
      "loss": 0.6482,
      "step": 20894
    },
    {
      "epoch": 0.865970409051349,
      "grad_norm": 0.45924702286720276,
      "learning_rate": 6.703551742716234e-07,
      "loss": 0.7045,
      "step": 20895
    },
    {
      "epoch": 0.8660118529570227,
      "grad_norm": 0.4316651523113251,
      "learning_rate": 6.70147954743255e-07,
      "loss": 0.6835,
      "step": 20896
    },
    {
      "epoch": 0.8660532968626964,
      "grad_norm": 0.4028855264186859,
      "learning_rate": 6.699407352148866e-07,
      "loss": 0.6827,
      "step": 20897
    },
    {
      "epoch": 0.86609474076837,
      "grad_norm": 0.41253799200057983,
      "learning_rate": 6.697335156865185e-07,
      "loss": 0.676,
      "step": 20898
    },
    {
      "epoch": 0.8661361846740436,
      "grad_norm": 0.40884727239608765,
      "learning_rate": 6.6952629615815e-07,
      "loss": 0.6208,
      "step": 20899
    },
    {
      "epoch": 0.8661776285797174,
      "grad_norm": 0.39559206366539,
      "learning_rate": 6.693190766297816e-07,
      "loss": 0.653,
      "step": 20900
    },
    {
      "epoch": 0.866219072485391,
      "grad_norm": 0.4069955348968506,
      "learning_rate": 6.691118571014132e-07,
      "loss": 0.6355,
      "step": 20901
    },
    {
      "epoch": 0.8662605163910647,
      "grad_norm": 0.41326963901519775,
      "learning_rate": 6.689046375730449e-07,
      "loss": 0.6558,
      "step": 20902
    },
    {
      "epoch": 0.8663019602967383,
      "grad_norm": 0.4666498303413391,
      "learning_rate": 6.686974180446767e-07,
      "loss": 0.7168,
      "step": 20903
    },
    {
      "epoch": 0.8663434042024121,
      "grad_norm": 0.3905469477176666,
      "learning_rate": 6.684901985163082e-07,
      "loss": 0.6877,
      "step": 20904
    },
    {
      "epoch": 0.8663848481080857,
      "grad_norm": 0.430497407913208,
      "learning_rate": 6.682829789879398e-07,
      "loss": 0.6978,
      "step": 20905
    },
    {
      "epoch": 0.8664262920137594,
      "grad_norm": 0.40694501996040344,
      "learning_rate": 6.680757594595715e-07,
      "loss": 0.649,
      "step": 20906
    },
    {
      "epoch": 0.866467735919433,
      "grad_norm": 0.4109271466732025,
      "learning_rate": 6.678685399312033e-07,
      "loss": 0.6963,
      "step": 20907
    },
    {
      "epoch": 0.8665091798251067,
      "grad_norm": 0.4173610508441925,
      "learning_rate": 6.676613204028348e-07,
      "loss": 0.6855,
      "step": 20908
    },
    {
      "epoch": 0.8665506237307804,
      "grad_norm": 0.425766259431839,
      "learning_rate": 6.674541008744664e-07,
      "loss": 0.6655,
      "step": 20909
    },
    {
      "epoch": 0.866592067636454,
      "grad_norm": 0.4009172320365906,
      "learning_rate": 6.67246881346098e-07,
      "loss": 0.6799,
      "step": 20910
    },
    {
      "epoch": 0.8666335115421278,
      "grad_norm": 0.3937987983226776,
      "learning_rate": 6.670396618177299e-07,
      "loss": 0.6449,
      "step": 20911
    },
    {
      "epoch": 0.8666749554478014,
      "grad_norm": 0.38850513100624084,
      "learning_rate": 6.668324422893615e-07,
      "loss": 0.6431,
      "step": 20912
    },
    {
      "epoch": 0.8667163993534751,
      "grad_norm": 0.4118667244911194,
      "learning_rate": 6.66625222760993e-07,
      "loss": 0.7085,
      "step": 20913
    },
    {
      "epoch": 0.8667578432591487,
      "grad_norm": 0.38834768533706665,
      "learning_rate": 6.664180032326246e-07,
      "loss": 0.6594,
      "step": 20914
    },
    {
      "epoch": 0.8667992871648225,
      "grad_norm": 0.3838385343551636,
      "learning_rate": 6.662107837042565e-07,
      "loss": 0.6642,
      "step": 20915
    },
    {
      "epoch": 0.8668407310704961,
      "grad_norm": 0.38794219493865967,
      "learning_rate": 6.66003564175888e-07,
      "loss": 0.6074,
      "step": 20916
    },
    {
      "epoch": 0.8668821749761697,
      "grad_norm": 0.3803756833076477,
      "learning_rate": 6.657963446475196e-07,
      "loss": 0.6777,
      "step": 20917
    },
    {
      "epoch": 0.8669236188818434,
      "grad_norm": 0.4118775427341461,
      "learning_rate": 6.655891251191512e-07,
      "loss": 0.6586,
      "step": 20918
    },
    {
      "epoch": 0.8669650627875171,
      "grad_norm": 0.4366982877254486,
      "learning_rate": 6.653819055907828e-07,
      "loss": 0.6821,
      "step": 20919
    },
    {
      "epoch": 0.8670065066931908,
      "grad_norm": 0.4392503798007965,
      "learning_rate": 6.651746860624147e-07,
      "loss": 0.6917,
      "step": 20920
    },
    {
      "epoch": 0.8670479505988644,
      "grad_norm": 0.4193454086780548,
      "learning_rate": 6.649674665340463e-07,
      "loss": 0.6602,
      "step": 20921
    },
    {
      "epoch": 0.8670893945045381,
      "grad_norm": 0.426750123500824,
      "learning_rate": 6.647602470056778e-07,
      "loss": 0.653,
      "step": 20922
    },
    {
      "epoch": 0.8671308384102118,
      "grad_norm": 0.4447825849056244,
      "learning_rate": 6.645530274773094e-07,
      "loss": 0.7476,
      "step": 20923
    },
    {
      "epoch": 0.8671722823158855,
      "grad_norm": 0.44760167598724365,
      "learning_rate": 6.643458079489413e-07,
      "loss": 0.6819,
      "step": 20924
    },
    {
      "epoch": 0.8672137262215591,
      "grad_norm": 0.41266539692878723,
      "learning_rate": 6.641385884205729e-07,
      "loss": 0.6975,
      "step": 20925
    },
    {
      "epoch": 0.8672551701272327,
      "grad_norm": 0.4238797724246979,
      "learning_rate": 6.639313688922044e-07,
      "loss": 0.6488,
      "step": 20926
    },
    {
      "epoch": 0.8672966140329065,
      "grad_norm": 0.4176521599292755,
      "learning_rate": 6.63724149363836e-07,
      "loss": 0.6619,
      "step": 20927
    },
    {
      "epoch": 0.8673380579385801,
      "grad_norm": 0.43469467759132385,
      "learning_rate": 6.635169298354679e-07,
      "loss": 0.6808,
      "step": 20928
    },
    {
      "epoch": 0.8673795018442538,
      "grad_norm": 0.41081467270851135,
      "learning_rate": 6.633097103070995e-07,
      "loss": 0.6744,
      "step": 20929
    },
    {
      "epoch": 0.8674209457499275,
      "grad_norm": 0.4399205446243286,
      "learning_rate": 6.63102490778731e-07,
      "loss": 0.7065,
      "step": 20930
    },
    {
      "epoch": 0.8674623896556012,
      "grad_norm": 0.4125112295150757,
      "learning_rate": 6.628952712503626e-07,
      "loss": 0.6846,
      "step": 20931
    },
    {
      "epoch": 0.8675038335612748,
      "grad_norm": 0.5592701435089111,
      "learning_rate": 6.626880517219942e-07,
      "loss": 0.7273,
      "step": 20932
    },
    {
      "epoch": 0.8675452774669485,
      "grad_norm": 0.42536336183547974,
      "learning_rate": 6.624808321936261e-07,
      "loss": 0.6655,
      "step": 20933
    },
    {
      "epoch": 0.8675867213726222,
      "grad_norm": 0.4069126546382904,
      "learning_rate": 6.622736126652577e-07,
      "loss": 0.7106,
      "step": 20934
    },
    {
      "epoch": 0.8676281652782958,
      "grad_norm": 0.4172263443470001,
      "learning_rate": 6.620663931368892e-07,
      "loss": 0.6902,
      "step": 20935
    },
    {
      "epoch": 0.8676696091839695,
      "grad_norm": 0.4257378876209259,
      "learning_rate": 6.618591736085208e-07,
      "loss": 0.7275,
      "step": 20936
    },
    {
      "epoch": 0.8677110530896431,
      "grad_norm": 0.4231618642807007,
      "learning_rate": 6.616519540801527e-07,
      "loss": 0.7102,
      "step": 20937
    },
    {
      "epoch": 0.8677524969953169,
      "grad_norm": 0.39940279722213745,
      "learning_rate": 6.614447345517843e-07,
      "loss": 0.6661,
      "step": 20938
    },
    {
      "epoch": 0.8677939409009905,
      "grad_norm": 0.3883577883243561,
      "learning_rate": 6.612375150234159e-07,
      "loss": 0.6641,
      "step": 20939
    },
    {
      "epoch": 0.8678353848066642,
      "grad_norm": 0.4278617203235626,
      "learning_rate": 6.610302954950474e-07,
      "loss": 0.6866,
      "step": 20940
    },
    {
      "epoch": 0.8678768287123378,
      "grad_norm": 0.4029589593410492,
      "learning_rate": 6.608230759666793e-07,
      "loss": 0.6941,
      "step": 20941
    },
    {
      "epoch": 0.8679182726180115,
      "grad_norm": 0.4023185968399048,
      "learning_rate": 6.606158564383109e-07,
      "loss": 0.5927,
      "step": 20942
    },
    {
      "epoch": 0.8679597165236852,
      "grad_norm": 0.3980408310890198,
      "learning_rate": 6.604086369099425e-07,
      "loss": 0.6074,
      "step": 20943
    },
    {
      "epoch": 0.8680011604293588,
      "grad_norm": 0.4162020683288574,
      "learning_rate": 6.60201417381574e-07,
      "loss": 0.6667,
      "step": 20944
    },
    {
      "epoch": 0.8680426043350326,
      "grad_norm": 0.41835200786590576,
      "learning_rate": 6.599941978532056e-07,
      "loss": 0.6866,
      "step": 20945
    },
    {
      "epoch": 0.8680840482407062,
      "grad_norm": 0.41209304332733154,
      "learning_rate": 6.597869783248375e-07,
      "loss": 0.605,
      "step": 20946
    },
    {
      "epoch": 0.8681254921463799,
      "grad_norm": 0.4571639895439148,
      "learning_rate": 6.595797587964691e-07,
      "loss": 0.6716,
      "step": 20947
    },
    {
      "epoch": 0.8681669360520535,
      "grad_norm": 0.4019971787929535,
      "learning_rate": 6.593725392681007e-07,
      "loss": 0.6605,
      "step": 20948
    },
    {
      "epoch": 0.8682083799577273,
      "grad_norm": 0.46099260449409485,
      "learning_rate": 6.591653197397322e-07,
      "loss": 0.6453,
      "step": 20949
    },
    {
      "epoch": 0.8682498238634009,
      "grad_norm": 0.41087642312049866,
      "learning_rate": 6.589581002113641e-07,
      "loss": 0.6447,
      "step": 20950
    },
    {
      "epoch": 0.8682912677690745,
      "grad_norm": 0.3926309645175934,
      "learning_rate": 6.587508806829957e-07,
      "loss": 0.6365,
      "step": 20951
    },
    {
      "epoch": 0.8683327116747482,
      "grad_norm": 0.38196465373039246,
      "learning_rate": 6.585436611546273e-07,
      "loss": 0.6526,
      "step": 20952
    },
    {
      "epoch": 0.8683741555804219,
      "grad_norm": 0.4099431037902832,
      "learning_rate": 6.583364416262589e-07,
      "loss": 0.6643,
      "step": 20953
    },
    {
      "epoch": 0.8684155994860956,
      "grad_norm": 0.42053699493408203,
      "learning_rate": 6.581292220978907e-07,
      "loss": 0.688,
      "step": 20954
    },
    {
      "epoch": 0.8684570433917692,
      "grad_norm": 0.419247031211853,
      "learning_rate": 6.579220025695223e-07,
      "loss": 0.6941,
      "step": 20955
    },
    {
      "epoch": 0.868498487297443,
      "grad_norm": 0.4384819567203522,
      "learning_rate": 6.577147830411539e-07,
      "loss": 0.6731,
      "step": 20956
    },
    {
      "epoch": 0.8685399312031166,
      "grad_norm": 0.426577091217041,
      "learning_rate": 6.575075635127855e-07,
      "loss": 0.6803,
      "step": 20957
    },
    {
      "epoch": 0.8685813751087903,
      "grad_norm": 0.40765416622161865,
      "learning_rate": 6.57300343984417e-07,
      "loss": 0.6384,
      "step": 20958
    },
    {
      "epoch": 0.8686228190144639,
      "grad_norm": 0.43082523345947266,
      "learning_rate": 6.570931244560489e-07,
      "loss": 0.6447,
      "step": 20959
    },
    {
      "epoch": 0.8686642629201375,
      "grad_norm": 0.4293767213821411,
      "learning_rate": 6.568859049276805e-07,
      "loss": 0.6843,
      "step": 20960
    },
    {
      "epoch": 0.8687057068258113,
      "grad_norm": 0.40518850088119507,
      "learning_rate": 6.56678685399312e-07,
      "loss": 0.6561,
      "step": 20961
    },
    {
      "epoch": 0.8687471507314849,
      "grad_norm": 0.43323907256126404,
      "learning_rate": 6.564714658709437e-07,
      "loss": 0.6624,
      "step": 20962
    },
    {
      "epoch": 0.8687885946371586,
      "grad_norm": 0.42290908098220825,
      "learning_rate": 6.562642463425755e-07,
      "loss": 0.6597,
      "step": 20963
    },
    {
      "epoch": 0.8688300385428323,
      "grad_norm": 0.42473524808883667,
      "learning_rate": 6.560570268142071e-07,
      "loss": 0.7131,
      "step": 20964
    },
    {
      "epoch": 0.868871482448506,
      "grad_norm": 0.4045064449310303,
      "learning_rate": 6.558498072858387e-07,
      "loss": 0.6685,
      "step": 20965
    },
    {
      "epoch": 0.8689129263541796,
      "grad_norm": 0.4135255515575409,
      "learning_rate": 6.556425877574703e-07,
      "loss": 0.6777,
      "step": 20966
    },
    {
      "epoch": 0.8689543702598533,
      "grad_norm": 0.396879643201828,
      "learning_rate": 6.554353682291021e-07,
      "loss": 0.6187,
      "step": 20967
    },
    {
      "epoch": 0.868995814165527,
      "grad_norm": 0.4325798451900482,
      "learning_rate": 6.552281487007337e-07,
      "loss": 0.6785,
      "step": 20968
    },
    {
      "epoch": 0.8690372580712006,
      "grad_norm": 0.4022548794746399,
      "learning_rate": 6.550209291723653e-07,
      "loss": 0.6643,
      "step": 20969
    },
    {
      "epoch": 0.8690787019768743,
      "grad_norm": 0.4126604497432709,
      "learning_rate": 6.548137096439969e-07,
      "loss": 0.6635,
      "step": 20970
    },
    {
      "epoch": 0.8691201458825479,
      "grad_norm": 0.42352133989334106,
      "learning_rate": 6.546064901156285e-07,
      "loss": 0.6288,
      "step": 20971
    },
    {
      "epoch": 0.8691615897882217,
      "grad_norm": 0.46328747272491455,
      "learning_rate": 6.543992705872603e-07,
      "loss": 0.7012,
      "step": 20972
    },
    {
      "epoch": 0.8692030336938953,
      "grad_norm": 0.38528740406036377,
      "learning_rate": 6.541920510588919e-07,
      "loss": 0.6105,
      "step": 20973
    },
    {
      "epoch": 0.869244477599569,
      "grad_norm": 0.4061183035373688,
      "learning_rate": 6.539848315305235e-07,
      "loss": 0.6709,
      "step": 20974
    },
    {
      "epoch": 0.8692859215052426,
      "grad_norm": 0.43848398327827454,
      "learning_rate": 6.53777612002155e-07,
      "loss": 0.7651,
      "step": 20975
    },
    {
      "epoch": 0.8693273654109164,
      "grad_norm": 0.4271588623523712,
      "learning_rate": 6.535703924737869e-07,
      "loss": 0.6932,
      "step": 20976
    },
    {
      "epoch": 0.86936880931659,
      "grad_norm": 0.4046817123889923,
      "learning_rate": 6.533631729454185e-07,
      "loss": 0.636,
      "step": 20977
    },
    {
      "epoch": 0.8694102532222636,
      "grad_norm": 0.3679453730583191,
      "learning_rate": 6.531559534170501e-07,
      "loss": 0.6003,
      "step": 20978
    },
    {
      "epoch": 0.8694516971279374,
      "grad_norm": 0.43561503291130066,
      "learning_rate": 6.529487338886817e-07,
      "loss": 0.6569,
      "step": 20979
    },
    {
      "epoch": 0.869493141033611,
      "grad_norm": 0.40223008394241333,
      "learning_rate": 6.527415143603135e-07,
      "loss": 0.6663,
      "step": 20980
    },
    {
      "epoch": 0.8695345849392847,
      "grad_norm": 0.39952352643013,
      "learning_rate": 6.525342948319451e-07,
      "loss": 0.6548,
      "step": 20981
    },
    {
      "epoch": 0.8695760288449583,
      "grad_norm": 0.4100901186466217,
      "learning_rate": 6.523270753035767e-07,
      "loss": 0.6777,
      "step": 20982
    },
    {
      "epoch": 0.8696174727506321,
      "grad_norm": 0.42813199758529663,
      "learning_rate": 6.521198557752083e-07,
      "loss": 0.7083,
      "step": 20983
    },
    {
      "epoch": 0.8696589166563057,
      "grad_norm": 0.420285701751709,
      "learning_rate": 6.519126362468399e-07,
      "loss": 0.6743,
      "step": 20984
    },
    {
      "epoch": 0.8697003605619794,
      "grad_norm": 0.5559378862380981,
      "learning_rate": 6.517054167184717e-07,
      "loss": 0.6818,
      "step": 20985
    },
    {
      "epoch": 0.869741804467653,
      "grad_norm": 0.43568581342697144,
      "learning_rate": 6.514981971901033e-07,
      "loss": 0.7173,
      "step": 20986
    },
    {
      "epoch": 0.8697832483733267,
      "grad_norm": 0.4175143837928772,
      "learning_rate": 6.512909776617349e-07,
      "loss": 0.6581,
      "step": 20987
    },
    {
      "epoch": 0.8698246922790004,
      "grad_norm": 0.4171862304210663,
      "learning_rate": 6.510837581333665e-07,
      "loss": 0.6802,
      "step": 20988
    },
    {
      "epoch": 0.869866136184674,
      "grad_norm": 0.4128933846950531,
      "learning_rate": 6.508765386049983e-07,
      "loss": 0.6733,
      "step": 20989
    },
    {
      "epoch": 0.8699075800903477,
      "grad_norm": 0.38840848207473755,
      "learning_rate": 6.506693190766299e-07,
      "loss": 0.6692,
      "step": 20990
    },
    {
      "epoch": 0.8699490239960214,
      "grad_norm": 0.4228536784648895,
      "learning_rate": 6.504620995482615e-07,
      "loss": 0.6798,
      "step": 20991
    },
    {
      "epoch": 0.8699904679016951,
      "grad_norm": 0.4262978434562683,
      "learning_rate": 6.502548800198931e-07,
      "loss": 0.7113,
      "step": 20992
    },
    {
      "epoch": 0.8700319118073687,
      "grad_norm": 0.42713019251823425,
      "learning_rate": 6.500476604915248e-07,
      "loss": 0.6509,
      "step": 20993
    },
    {
      "epoch": 0.8700733557130425,
      "grad_norm": 0.42333167791366577,
      "learning_rate": 6.498404409631565e-07,
      "loss": 0.6721,
      "step": 20994
    },
    {
      "epoch": 0.8701147996187161,
      "grad_norm": 0.45988669991493225,
      "learning_rate": 6.496332214347881e-07,
      "loss": 0.6885,
      "step": 20995
    },
    {
      "epoch": 0.8701562435243897,
      "grad_norm": 0.4318194091320038,
      "learning_rate": 6.494260019064197e-07,
      "loss": 0.7512,
      "step": 20996
    },
    {
      "epoch": 0.8701976874300634,
      "grad_norm": 0.40950191020965576,
      "learning_rate": 6.492187823780513e-07,
      "loss": 0.6694,
      "step": 20997
    },
    {
      "epoch": 0.870239131335737,
      "grad_norm": 0.39523282647132874,
      "learning_rate": 6.490115628496831e-07,
      "loss": 0.6381,
      "step": 20998
    },
    {
      "epoch": 0.8702805752414108,
      "grad_norm": 0.4065968096256256,
      "learning_rate": 6.488043433213147e-07,
      "loss": 0.6917,
      "step": 20999
    },
    {
      "epoch": 0.8703220191470844,
      "grad_norm": 0.39876100420951843,
      "learning_rate": 6.485971237929463e-07,
      "loss": 0.6447,
      "step": 21000
    },
    {
      "epoch": 0.8703634630527581,
      "grad_norm": 0.36696720123291016,
      "learning_rate": 6.483899042645779e-07,
      "loss": 0.6635,
      "step": 21001
    },
    {
      "epoch": 0.8704049069584318,
      "grad_norm": 0.405466765165329,
      "learning_rate": 6.481826847362097e-07,
      "loss": 0.6565,
      "step": 21002
    },
    {
      "epoch": 0.8704463508641054,
      "grad_norm": 0.39545005559921265,
      "learning_rate": 6.479754652078413e-07,
      "loss": 0.6193,
      "step": 21003
    },
    {
      "epoch": 0.8704877947697791,
      "grad_norm": 0.429261177778244,
      "learning_rate": 6.477682456794729e-07,
      "loss": 0.6819,
      "step": 21004
    },
    {
      "epoch": 0.8705292386754527,
      "grad_norm": 0.382962703704834,
      "learning_rate": 6.475610261511045e-07,
      "loss": 0.6793,
      "step": 21005
    },
    {
      "epoch": 0.8705706825811265,
      "grad_norm": 0.3871862292289734,
      "learning_rate": 6.473538066227362e-07,
      "loss": 0.6281,
      "step": 21006
    },
    {
      "epoch": 0.8706121264868001,
      "grad_norm": 0.4119024872779846,
      "learning_rate": 6.471465870943679e-07,
      "loss": 0.6272,
      "step": 21007
    },
    {
      "epoch": 0.8706535703924738,
      "grad_norm": 0.45024019479751587,
      "learning_rate": 6.469393675659995e-07,
      "loss": 0.6782,
      "step": 21008
    },
    {
      "epoch": 0.8706950142981474,
      "grad_norm": 0.3999943137168884,
      "learning_rate": 6.467321480376311e-07,
      "loss": 0.6731,
      "step": 21009
    },
    {
      "epoch": 0.8707364582038212,
      "grad_norm": 0.4422866404056549,
      "learning_rate": 6.465249285092628e-07,
      "loss": 0.6545,
      "step": 21010
    },
    {
      "epoch": 0.8707779021094948,
      "grad_norm": 0.4573245346546173,
      "learning_rate": 6.463177089808945e-07,
      "loss": 0.6886,
      "step": 21011
    },
    {
      "epoch": 0.8708193460151684,
      "grad_norm": 0.42297235131263733,
      "learning_rate": 6.461104894525261e-07,
      "loss": 0.6646,
      "step": 21012
    },
    {
      "epoch": 0.8708607899208421,
      "grad_norm": 0.4467531442642212,
      "learning_rate": 6.459032699241577e-07,
      "loss": 0.6794,
      "step": 21013
    },
    {
      "epoch": 0.8709022338265158,
      "grad_norm": 0.429007887840271,
      "learning_rate": 6.456960503957893e-07,
      "loss": 0.7156,
      "step": 21014
    },
    {
      "epoch": 0.8709436777321895,
      "grad_norm": 0.39854303002357483,
      "learning_rate": 6.454888308674211e-07,
      "loss": 0.7006,
      "step": 21015
    },
    {
      "epoch": 0.8709851216378631,
      "grad_norm": 0.4026521146297455,
      "learning_rate": 6.452816113390527e-07,
      "loss": 0.6826,
      "step": 21016
    },
    {
      "epoch": 0.8710265655435369,
      "grad_norm": 0.45026010274887085,
      "learning_rate": 6.450743918106843e-07,
      "loss": 0.6583,
      "step": 21017
    },
    {
      "epoch": 0.8710680094492105,
      "grad_norm": 0.41821566224098206,
      "learning_rate": 6.448671722823159e-07,
      "loss": 0.7026,
      "step": 21018
    },
    {
      "epoch": 0.8711094533548842,
      "grad_norm": 0.39174166321754456,
      "learning_rate": 6.446599527539476e-07,
      "loss": 0.673,
      "step": 21019
    },
    {
      "epoch": 0.8711508972605578,
      "grad_norm": 0.40558144450187683,
      "learning_rate": 6.444527332255793e-07,
      "loss": 0.6407,
      "step": 21020
    },
    {
      "epoch": 0.8711923411662315,
      "grad_norm": 0.42113354802131653,
      "learning_rate": 6.442455136972109e-07,
      "loss": 0.6854,
      "step": 21021
    },
    {
      "epoch": 0.8712337850719052,
      "grad_norm": 0.4299522340297699,
      "learning_rate": 6.440382941688425e-07,
      "loss": 0.6599,
      "step": 21022
    },
    {
      "epoch": 0.8712752289775788,
      "grad_norm": 0.38888901472091675,
      "learning_rate": 6.438310746404742e-07,
      "loss": 0.6404,
      "step": 21023
    },
    {
      "epoch": 0.8713166728832525,
      "grad_norm": 0.42769384384155273,
      "learning_rate": 6.436238551121059e-07,
      "loss": 0.7002,
      "step": 21024
    },
    {
      "epoch": 0.8713581167889262,
      "grad_norm": 0.3867104649543762,
      "learning_rate": 6.434166355837375e-07,
      "loss": 0.6853,
      "step": 21025
    },
    {
      "epoch": 0.8713995606945999,
      "grad_norm": 0.4051859676837921,
      "learning_rate": 6.432094160553691e-07,
      "loss": 0.6565,
      "step": 21026
    },
    {
      "epoch": 0.8714410046002735,
      "grad_norm": 0.4128601551055908,
      "learning_rate": 6.430021965270008e-07,
      "loss": 0.6271,
      "step": 21027
    },
    {
      "epoch": 0.8714824485059472,
      "grad_norm": 0.39655977487564087,
      "learning_rate": 6.427949769986325e-07,
      "loss": 0.6671,
      "step": 21028
    },
    {
      "epoch": 0.8715238924116209,
      "grad_norm": 0.43138906359672546,
      "learning_rate": 6.425877574702641e-07,
      "loss": 0.6962,
      "step": 21029
    },
    {
      "epoch": 0.8715653363172945,
      "grad_norm": 0.41463854908943176,
      "learning_rate": 6.423805379418957e-07,
      "loss": 0.6521,
      "step": 21030
    },
    {
      "epoch": 0.8716067802229682,
      "grad_norm": 0.4136026203632355,
      "learning_rate": 6.421733184135274e-07,
      "loss": 0.6509,
      "step": 21031
    },
    {
      "epoch": 0.8716482241286418,
      "grad_norm": 0.41526171565055847,
      "learning_rate": 6.41966098885159e-07,
      "loss": 0.6719,
      "step": 21032
    },
    {
      "epoch": 0.8716896680343156,
      "grad_norm": 0.44354721903800964,
      "learning_rate": 6.417588793567907e-07,
      "loss": 0.698,
      "step": 21033
    },
    {
      "epoch": 0.8717311119399892,
      "grad_norm": 0.4231584668159485,
      "learning_rate": 6.415516598284223e-07,
      "loss": 0.6531,
      "step": 21034
    },
    {
      "epoch": 0.8717725558456629,
      "grad_norm": 0.3978957533836365,
      "learning_rate": 6.413444403000539e-07,
      "loss": 0.6324,
      "step": 21035
    },
    {
      "epoch": 0.8718139997513366,
      "grad_norm": 0.39762282371520996,
      "learning_rate": 6.411372207716856e-07,
      "loss": 0.6205,
      "step": 21036
    },
    {
      "epoch": 0.8718554436570103,
      "grad_norm": 0.4275680482387543,
      "learning_rate": 6.409300012433173e-07,
      "loss": 0.6528,
      "step": 21037
    },
    {
      "epoch": 0.8718968875626839,
      "grad_norm": 0.45531949400901794,
      "learning_rate": 6.407227817149489e-07,
      "loss": 0.7322,
      "step": 21038
    },
    {
      "epoch": 0.8719383314683575,
      "grad_norm": 0.4012257754802704,
      "learning_rate": 6.405155621865805e-07,
      "loss": 0.6812,
      "step": 21039
    },
    {
      "epoch": 0.8719797753740313,
      "grad_norm": 0.40079963207244873,
      "learning_rate": 6.403083426582122e-07,
      "loss": 0.6064,
      "step": 21040
    },
    {
      "epoch": 0.8720212192797049,
      "grad_norm": 0.43314310908317566,
      "learning_rate": 6.401011231298439e-07,
      "loss": 0.6603,
      "step": 21041
    },
    {
      "epoch": 0.8720626631853786,
      "grad_norm": 0.37559404969215393,
      "learning_rate": 6.398939036014755e-07,
      "loss": 0.6086,
      "step": 21042
    },
    {
      "epoch": 0.8721041070910522,
      "grad_norm": 0.4518875777721405,
      "learning_rate": 6.396866840731071e-07,
      "loss": 0.6499,
      "step": 21043
    },
    {
      "epoch": 0.872145550996726,
      "grad_norm": 0.4204740524291992,
      "learning_rate": 6.394794645447388e-07,
      "loss": 0.655,
      "step": 21044
    },
    {
      "epoch": 0.8721869949023996,
      "grad_norm": 0.4391556978225708,
      "learning_rate": 6.392722450163704e-07,
      "loss": 0.6715,
      "step": 21045
    },
    {
      "epoch": 0.8722284388080733,
      "grad_norm": 0.3926897644996643,
      "learning_rate": 6.390650254880021e-07,
      "loss": 0.6149,
      "step": 21046
    },
    {
      "epoch": 0.872269882713747,
      "grad_norm": 0.40009206533432007,
      "learning_rate": 6.388578059596337e-07,
      "loss": 0.7205,
      "step": 21047
    },
    {
      "epoch": 0.8723113266194206,
      "grad_norm": 0.42757606506347656,
      "learning_rate": 6.386505864312654e-07,
      "loss": 0.6503,
      "step": 21048
    },
    {
      "epoch": 0.8723527705250943,
      "grad_norm": 0.4311619699001312,
      "learning_rate": 6.38443366902897e-07,
      "loss": 0.6276,
      "step": 21049
    },
    {
      "epoch": 0.8723942144307679,
      "grad_norm": 0.45492252707481384,
      "learning_rate": 6.382361473745287e-07,
      "loss": 0.6329,
      "step": 21050
    },
    {
      "epoch": 0.8724356583364417,
      "grad_norm": 0.3861817717552185,
      "learning_rate": 6.380289278461603e-07,
      "loss": 0.6641,
      "step": 21051
    },
    {
      "epoch": 0.8724771022421153,
      "grad_norm": 0.4098234474658966,
      "learning_rate": 6.378217083177919e-07,
      "loss": 0.6205,
      "step": 21052
    },
    {
      "epoch": 0.872518546147789,
      "grad_norm": 0.43558695912361145,
      "learning_rate": 6.376144887894236e-07,
      "loss": 0.6543,
      "step": 21053
    },
    {
      "epoch": 0.8725599900534626,
      "grad_norm": 0.38374799489974976,
      "learning_rate": 6.374072692610552e-07,
      "loss": 0.6018,
      "step": 21054
    },
    {
      "epoch": 0.8726014339591364,
      "grad_norm": 0.40227314829826355,
      "learning_rate": 6.372000497326869e-07,
      "loss": 0.6847,
      "step": 21055
    },
    {
      "epoch": 0.87264287786481,
      "grad_norm": 0.4054569900035858,
      "learning_rate": 6.369928302043185e-07,
      "loss": 0.6326,
      "step": 21056
    },
    {
      "epoch": 0.8726843217704836,
      "grad_norm": 0.41419917345046997,
      "learning_rate": 6.367856106759502e-07,
      "loss": 0.7007,
      "step": 21057
    },
    {
      "epoch": 0.8727257656761573,
      "grad_norm": 0.47099941968917847,
      "learning_rate": 6.365783911475818e-07,
      "loss": 0.7246,
      "step": 21058
    },
    {
      "epoch": 0.872767209581831,
      "grad_norm": 0.4324527084827423,
      "learning_rate": 6.363711716192135e-07,
      "loss": 0.6759,
      "step": 21059
    },
    {
      "epoch": 0.8728086534875047,
      "grad_norm": 0.4270777702331543,
      "learning_rate": 6.361639520908451e-07,
      "loss": 0.6792,
      "step": 21060
    },
    {
      "epoch": 0.8728500973931783,
      "grad_norm": 0.38193923234939575,
      "learning_rate": 6.359567325624768e-07,
      "loss": 0.6562,
      "step": 21061
    },
    {
      "epoch": 0.872891541298852,
      "grad_norm": 0.414305180311203,
      "learning_rate": 6.357495130341084e-07,
      "loss": 0.7297,
      "step": 21062
    },
    {
      "epoch": 0.8729329852045257,
      "grad_norm": 0.3815478980541229,
      "learning_rate": 6.355422935057401e-07,
      "loss": 0.5847,
      "step": 21063
    },
    {
      "epoch": 0.8729744291101993,
      "grad_norm": 0.4070812165737152,
      "learning_rate": 6.353350739773717e-07,
      "loss": 0.6982,
      "step": 21064
    },
    {
      "epoch": 0.873015873015873,
      "grad_norm": 0.4201981723308563,
      "learning_rate": 6.351278544490034e-07,
      "loss": 0.7017,
      "step": 21065
    },
    {
      "epoch": 0.8730573169215466,
      "grad_norm": 0.3947572112083435,
      "learning_rate": 6.34920634920635e-07,
      "loss": 0.62,
      "step": 21066
    },
    {
      "epoch": 0.8730987608272204,
      "grad_norm": 0.414168119430542,
      "learning_rate": 6.347134153922666e-07,
      "loss": 0.6602,
      "step": 21067
    },
    {
      "epoch": 0.873140204732894,
      "grad_norm": 0.43096923828125,
      "learning_rate": 6.345061958638983e-07,
      "loss": 0.649,
      "step": 21068
    },
    {
      "epoch": 0.8731816486385677,
      "grad_norm": 0.40803661942481995,
      "learning_rate": 6.342989763355299e-07,
      "loss": 0.6643,
      "step": 21069
    },
    {
      "epoch": 0.8732230925442414,
      "grad_norm": 0.40051186084747314,
      "learning_rate": 6.340917568071616e-07,
      "loss": 0.6411,
      "step": 21070
    },
    {
      "epoch": 0.8732645364499151,
      "grad_norm": 0.4332613945007324,
      "learning_rate": 6.338845372787932e-07,
      "loss": 0.6499,
      "step": 21071
    },
    {
      "epoch": 0.8733059803555887,
      "grad_norm": 0.4091212749481201,
      "learning_rate": 6.336773177504249e-07,
      "loss": 0.6309,
      "step": 21072
    },
    {
      "epoch": 0.8733474242612623,
      "grad_norm": 0.403309166431427,
      "learning_rate": 6.334700982220565e-07,
      "loss": 0.6385,
      "step": 21073
    },
    {
      "epoch": 0.8733888681669361,
      "grad_norm": 0.43522611260414124,
      "learning_rate": 6.332628786936882e-07,
      "loss": 0.7429,
      "step": 21074
    },
    {
      "epoch": 0.8734303120726097,
      "grad_norm": 0.4101661443710327,
      "learning_rate": 6.330556591653198e-07,
      "loss": 0.6569,
      "step": 21075
    },
    {
      "epoch": 0.8734717559782834,
      "grad_norm": 0.4063767194747925,
      "learning_rate": 6.328484396369515e-07,
      "loss": 0.6647,
      "step": 21076
    },
    {
      "epoch": 0.873513199883957,
      "grad_norm": 0.4405270516872406,
      "learning_rate": 6.326412201085831e-07,
      "loss": 0.7084,
      "step": 21077
    },
    {
      "epoch": 0.8735546437896308,
      "grad_norm": 0.4283601939678192,
      "learning_rate": 6.324340005802148e-07,
      "loss": 0.6158,
      "step": 21078
    },
    {
      "epoch": 0.8735960876953044,
      "grad_norm": 0.42846760153770447,
      "learning_rate": 6.322267810518464e-07,
      "loss": 0.6646,
      "step": 21079
    },
    {
      "epoch": 0.8736375316009781,
      "grad_norm": 0.40826937556266785,
      "learning_rate": 6.32019561523478e-07,
      "loss": 0.6353,
      "step": 21080
    },
    {
      "epoch": 0.8736789755066517,
      "grad_norm": 0.4547218680381775,
      "learning_rate": 6.318123419951097e-07,
      "loss": 0.7053,
      "step": 21081
    },
    {
      "epoch": 0.8737204194123254,
      "grad_norm": 0.4659343957901001,
      "learning_rate": 6.316051224667414e-07,
      "loss": 0.6785,
      "step": 21082
    },
    {
      "epoch": 0.8737618633179991,
      "grad_norm": 0.41162532567977905,
      "learning_rate": 6.31397902938373e-07,
      "loss": 0.6262,
      "step": 21083
    },
    {
      "epoch": 0.8738033072236727,
      "grad_norm": 0.43572402000427246,
      "learning_rate": 6.311906834100046e-07,
      "loss": 0.748,
      "step": 21084
    },
    {
      "epoch": 0.8738447511293465,
      "grad_norm": 0.3738400340080261,
      "learning_rate": 6.309834638816363e-07,
      "loss": 0.6163,
      "step": 21085
    },
    {
      "epoch": 0.8738861950350201,
      "grad_norm": 0.43965157866477966,
      "learning_rate": 6.30776244353268e-07,
      "loss": 0.7086,
      "step": 21086
    },
    {
      "epoch": 0.8739276389406938,
      "grad_norm": 0.42424824833869934,
      "learning_rate": 6.305690248248996e-07,
      "loss": 0.6914,
      "step": 21087
    },
    {
      "epoch": 0.8739690828463674,
      "grad_norm": 0.40504661202430725,
      "learning_rate": 6.303618052965312e-07,
      "loss": 0.6515,
      "step": 21088
    },
    {
      "epoch": 0.8740105267520412,
      "grad_norm": 0.3959163725376129,
      "learning_rate": 6.301545857681629e-07,
      "loss": 0.6543,
      "step": 21089
    },
    {
      "epoch": 0.8740519706577148,
      "grad_norm": 0.44820645451545715,
      "learning_rate": 6.299473662397945e-07,
      "loss": 0.6427,
      "step": 21090
    },
    {
      "epoch": 0.8740934145633884,
      "grad_norm": 0.4210261106491089,
      "learning_rate": 6.297401467114262e-07,
      "loss": 0.6658,
      "step": 21091
    },
    {
      "epoch": 0.8741348584690621,
      "grad_norm": 0.39452698826789856,
      "learning_rate": 6.295329271830578e-07,
      "loss": 0.6471,
      "step": 21092
    },
    {
      "epoch": 0.8741763023747358,
      "grad_norm": 0.38127636909484863,
      "learning_rate": 6.293257076546894e-07,
      "loss": 0.6726,
      "step": 21093
    },
    {
      "epoch": 0.8742177462804095,
      "grad_norm": 0.43088850378990173,
      "learning_rate": 6.291184881263211e-07,
      "loss": 0.6559,
      "step": 21094
    },
    {
      "epoch": 0.8742591901860831,
      "grad_norm": 0.43449413776397705,
      "learning_rate": 6.289112685979528e-07,
      "loss": 0.6373,
      "step": 21095
    },
    {
      "epoch": 0.8743006340917568,
      "grad_norm": 0.4328954219818115,
      "learning_rate": 6.287040490695844e-07,
      "loss": 0.6877,
      "step": 21096
    },
    {
      "epoch": 0.8743420779974305,
      "grad_norm": 0.45773303508758545,
      "learning_rate": 6.28496829541216e-07,
      "loss": 0.708,
      "step": 21097
    },
    {
      "epoch": 0.8743835219031042,
      "grad_norm": 0.3907792270183563,
      "learning_rate": 6.282896100128477e-07,
      "loss": 0.6217,
      "step": 21098
    },
    {
      "epoch": 0.8744249658087778,
      "grad_norm": 0.4148302376270294,
      "learning_rate": 6.280823904844794e-07,
      "loss": 0.631,
      "step": 21099
    },
    {
      "epoch": 0.8744664097144514,
      "grad_norm": 0.4146079123020172,
      "learning_rate": 6.27875170956111e-07,
      "loss": 0.6582,
      "step": 21100
    },
    {
      "epoch": 0.8745078536201252,
      "grad_norm": 0.4227794110774994,
      "learning_rate": 6.276679514277426e-07,
      "loss": 0.6294,
      "step": 21101
    },
    {
      "epoch": 0.8745492975257988,
      "grad_norm": 0.42427343130111694,
      "learning_rate": 6.274607318993743e-07,
      "loss": 0.6741,
      "step": 21102
    },
    {
      "epoch": 0.8745907414314725,
      "grad_norm": 0.3804783523082733,
      "learning_rate": 6.27253512371006e-07,
      "loss": 0.6018,
      "step": 21103
    },
    {
      "epoch": 0.8746321853371462,
      "grad_norm": 0.40360864996910095,
      "learning_rate": 6.270462928426376e-07,
      "loss": 0.6256,
      "step": 21104
    },
    {
      "epoch": 0.8746736292428199,
      "grad_norm": 0.42289596796035767,
      "learning_rate": 6.268390733142692e-07,
      "loss": 0.6466,
      "step": 21105
    },
    {
      "epoch": 0.8747150731484935,
      "grad_norm": 0.3998088836669922,
      "learning_rate": 6.266318537859008e-07,
      "loss": 0.66,
      "step": 21106
    },
    {
      "epoch": 0.8747565170541672,
      "grad_norm": 0.3816741704940796,
      "learning_rate": 6.264246342575325e-07,
      "loss": 0.594,
      "step": 21107
    },
    {
      "epoch": 0.8747979609598409,
      "grad_norm": 0.418963223695755,
      "learning_rate": 6.262174147291642e-07,
      "loss": 0.6201,
      "step": 21108
    },
    {
      "epoch": 0.8748394048655145,
      "grad_norm": 0.4480084180831909,
      "learning_rate": 6.260101952007958e-07,
      "loss": 0.6921,
      "step": 21109
    },
    {
      "epoch": 0.8748808487711882,
      "grad_norm": 0.38828662037849426,
      "learning_rate": 6.258029756724274e-07,
      "loss": 0.6001,
      "step": 21110
    },
    {
      "epoch": 0.8749222926768618,
      "grad_norm": 0.4004436731338501,
      "learning_rate": 6.255957561440591e-07,
      "loss": 0.6473,
      "step": 21111
    },
    {
      "epoch": 0.8749637365825356,
      "grad_norm": 0.4194926917552948,
      "learning_rate": 6.253885366156908e-07,
      "loss": 0.6829,
      "step": 21112
    },
    {
      "epoch": 0.8750051804882092,
      "grad_norm": 0.3987944424152374,
      "learning_rate": 6.251813170873224e-07,
      "loss": 0.6279,
      "step": 21113
    },
    {
      "epoch": 0.8750466243938829,
      "grad_norm": 0.3936912417411804,
      "learning_rate": 6.249740975589541e-07,
      "loss": 0.6919,
      "step": 21114
    },
    {
      "epoch": 0.8750880682995565,
      "grad_norm": 0.43640032410621643,
      "learning_rate": 6.247668780305857e-07,
      "loss": 0.6576,
      "step": 21115
    },
    {
      "epoch": 0.8751295122052303,
      "grad_norm": 0.39306315779685974,
      "learning_rate": 6.245596585022173e-07,
      "loss": 0.6439,
      "step": 21116
    },
    {
      "epoch": 0.8751709561109039,
      "grad_norm": 0.4292284846305847,
      "learning_rate": 6.24352438973849e-07,
      "loss": 0.662,
      "step": 21117
    },
    {
      "epoch": 0.8752124000165775,
      "grad_norm": 0.43263867497444153,
      "learning_rate": 6.241452194454806e-07,
      "loss": 0.679,
      "step": 21118
    },
    {
      "epoch": 0.8752538439222513,
      "grad_norm": 0.41625720262527466,
      "learning_rate": 6.239379999171123e-07,
      "loss": 0.62,
      "step": 21119
    },
    {
      "epoch": 0.8752952878279249,
      "grad_norm": 0.4633916914463043,
      "learning_rate": 6.237307803887439e-07,
      "loss": 0.7173,
      "step": 21120
    },
    {
      "epoch": 0.8753367317335986,
      "grad_norm": 0.40823599696159363,
      "learning_rate": 6.235235608603756e-07,
      "loss": 0.6501,
      "step": 21121
    },
    {
      "epoch": 0.8753781756392722,
      "grad_norm": 0.44758760929107666,
      "learning_rate": 6.233163413320072e-07,
      "loss": 0.7009,
      "step": 21122
    },
    {
      "epoch": 0.875419619544946,
      "grad_norm": 0.4339844882488251,
      "learning_rate": 6.231091218036389e-07,
      "loss": 0.6036,
      "step": 21123
    },
    {
      "epoch": 0.8754610634506196,
      "grad_norm": 0.411300927400589,
      "learning_rate": 6.229019022752705e-07,
      "loss": 0.6858,
      "step": 21124
    },
    {
      "epoch": 0.8755025073562932,
      "grad_norm": 0.4465458393096924,
      "learning_rate": 6.226946827469022e-07,
      "loss": 0.6667,
      "step": 21125
    },
    {
      "epoch": 0.8755439512619669,
      "grad_norm": 0.43640896677970886,
      "learning_rate": 6.224874632185338e-07,
      "loss": 0.6432,
      "step": 21126
    },
    {
      "epoch": 0.8755853951676406,
      "grad_norm": 0.4123358130455017,
      "learning_rate": 6.222802436901655e-07,
      "loss": 0.646,
      "step": 21127
    },
    {
      "epoch": 0.8756268390733143,
      "grad_norm": 0.4537299871444702,
      "learning_rate": 6.220730241617971e-07,
      "loss": 0.6877,
      "step": 21128
    },
    {
      "epoch": 0.8756682829789879,
      "grad_norm": 0.4167579412460327,
      "learning_rate": 6.218658046334287e-07,
      "loss": 0.6571,
      "step": 21129
    },
    {
      "epoch": 0.8757097268846616,
      "grad_norm": 0.41824257373809814,
      "learning_rate": 6.216585851050604e-07,
      "loss": 0.6554,
      "step": 21130
    },
    {
      "epoch": 0.8757511707903353,
      "grad_norm": 0.4184277057647705,
      "learning_rate": 6.21451365576692e-07,
      "loss": 0.696,
      "step": 21131
    },
    {
      "epoch": 0.875792614696009,
      "grad_norm": 0.4107036888599396,
      "learning_rate": 6.212441460483237e-07,
      "loss": 0.6631,
      "step": 21132
    },
    {
      "epoch": 0.8758340586016826,
      "grad_norm": 0.44150468707084656,
      "learning_rate": 6.210369265199553e-07,
      "loss": 0.709,
      "step": 21133
    },
    {
      "epoch": 0.8758755025073562,
      "grad_norm": 0.40125706791877747,
      "learning_rate": 6.20829706991587e-07,
      "loss": 0.6221,
      "step": 21134
    },
    {
      "epoch": 0.87591694641303,
      "grad_norm": 0.4508236050605774,
      "learning_rate": 6.206224874632186e-07,
      "loss": 0.7021,
      "step": 21135
    },
    {
      "epoch": 0.8759583903187036,
      "grad_norm": 0.4027940034866333,
      "learning_rate": 6.204152679348503e-07,
      "loss": 0.6931,
      "step": 21136
    },
    {
      "epoch": 0.8759998342243773,
      "grad_norm": 0.40112560987472534,
      "learning_rate": 6.202080484064819e-07,
      "loss": 0.6167,
      "step": 21137
    },
    {
      "epoch": 0.876041278130051,
      "grad_norm": 0.47703316807746887,
      "learning_rate": 6.200008288781136e-07,
      "loss": 0.611,
      "step": 21138
    },
    {
      "epoch": 0.8760827220357247,
      "grad_norm": 0.41878071427345276,
      "learning_rate": 6.197936093497452e-07,
      "loss": 0.6478,
      "step": 21139
    },
    {
      "epoch": 0.8761241659413983,
      "grad_norm": 0.4195132851600647,
      "learning_rate": 6.195863898213769e-07,
      "loss": 0.6746,
      "step": 21140
    },
    {
      "epoch": 0.876165609847072,
      "grad_norm": 0.41365063190460205,
      "learning_rate": 6.193791702930085e-07,
      "loss": 0.6284,
      "step": 21141
    },
    {
      "epoch": 0.8762070537527457,
      "grad_norm": 0.4118717610836029,
      "learning_rate": 6.191719507646401e-07,
      "loss": 0.6636,
      "step": 21142
    },
    {
      "epoch": 0.8762484976584193,
      "grad_norm": 0.4001978635787964,
      "learning_rate": 6.189647312362718e-07,
      "loss": 0.6678,
      "step": 21143
    },
    {
      "epoch": 0.876289941564093,
      "grad_norm": 0.40394824743270874,
      "learning_rate": 6.187575117079034e-07,
      "loss": 0.6541,
      "step": 21144
    },
    {
      "epoch": 0.8763313854697666,
      "grad_norm": 0.43684524297714233,
      "learning_rate": 6.185502921795351e-07,
      "loss": 0.6406,
      "step": 21145
    },
    {
      "epoch": 0.8763728293754404,
      "grad_norm": 0.44488877058029175,
      "learning_rate": 6.183430726511667e-07,
      "loss": 0.6777,
      "step": 21146
    },
    {
      "epoch": 0.876414273281114,
      "grad_norm": 0.41524744033813477,
      "learning_rate": 6.181358531227984e-07,
      "loss": 0.6229,
      "step": 21147
    },
    {
      "epoch": 0.8764557171867877,
      "grad_norm": 0.4127786159515381,
      "learning_rate": 6.1792863359443e-07,
      "loss": 0.6218,
      "step": 21148
    },
    {
      "epoch": 0.8764971610924613,
      "grad_norm": 0.44900187849998474,
      "learning_rate": 6.177214140660617e-07,
      "loss": 0.6418,
      "step": 21149
    },
    {
      "epoch": 0.8765386049981351,
      "grad_norm": 0.43120262026786804,
      "learning_rate": 6.175141945376933e-07,
      "loss": 0.6667,
      "step": 21150
    },
    {
      "epoch": 0.8765800489038087,
      "grad_norm": 0.42008501291275024,
      "learning_rate": 6.17306975009325e-07,
      "loss": 0.6675,
      "step": 21151
    },
    {
      "epoch": 0.8766214928094823,
      "grad_norm": 0.40160608291625977,
      "learning_rate": 6.170997554809566e-07,
      "loss": 0.6511,
      "step": 21152
    },
    {
      "epoch": 0.876662936715156,
      "grad_norm": 0.39608532190322876,
      "learning_rate": 6.168925359525882e-07,
      "loss": 0.6526,
      "step": 21153
    },
    {
      "epoch": 0.8767043806208297,
      "grad_norm": 0.424333781003952,
      "learning_rate": 6.166853164242199e-07,
      "loss": 0.6997,
      "step": 21154
    },
    {
      "epoch": 0.8767458245265034,
      "grad_norm": 0.41962510347366333,
      "learning_rate": 6.164780968958515e-07,
      "loss": 0.6531,
      "step": 21155
    },
    {
      "epoch": 0.876787268432177,
      "grad_norm": 0.4095209836959839,
      "learning_rate": 6.162708773674832e-07,
      "loss": 0.6737,
      "step": 21156
    },
    {
      "epoch": 0.8768287123378508,
      "grad_norm": 0.4183366000652313,
      "learning_rate": 6.160636578391148e-07,
      "loss": 0.6625,
      "step": 21157
    },
    {
      "epoch": 0.8768701562435244,
      "grad_norm": 0.40923160314559937,
      "learning_rate": 6.158564383107465e-07,
      "loss": 0.6329,
      "step": 21158
    },
    {
      "epoch": 0.8769116001491981,
      "grad_norm": 0.4039829671382904,
      "learning_rate": 6.156492187823781e-07,
      "loss": 0.6147,
      "step": 21159
    },
    {
      "epoch": 0.8769530440548717,
      "grad_norm": 0.4238888621330261,
      "learning_rate": 6.154419992540098e-07,
      "loss": 0.6587,
      "step": 21160
    },
    {
      "epoch": 0.8769944879605454,
      "grad_norm": 0.3870724141597748,
      "learning_rate": 6.152347797256414e-07,
      "loss": 0.6809,
      "step": 21161
    },
    {
      "epoch": 0.8770359318662191,
      "grad_norm": 0.43098098039627075,
      "learning_rate": 6.150275601972731e-07,
      "loss": 0.6646,
      "step": 21162
    },
    {
      "epoch": 0.8770773757718927,
      "grad_norm": 0.3778671622276306,
      "learning_rate": 6.148203406689047e-07,
      "loss": 0.6141,
      "step": 21163
    },
    {
      "epoch": 0.8771188196775664,
      "grad_norm": 0.41942545771598816,
      "learning_rate": 6.146131211405364e-07,
      "loss": 0.697,
      "step": 21164
    },
    {
      "epoch": 0.8771602635832401,
      "grad_norm": 0.390924334526062,
      "learning_rate": 6.14405901612168e-07,
      "loss": 0.6564,
      "step": 21165
    },
    {
      "epoch": 0.8772017074889138,
      "grad_norm": 0.4021517038345337,
      "learning_rate": 6.141986820837996e-07,
      "loss": 0.663,
      "step": 21166
    },
    {
      "epoch": 0.8772431513945874,
      "grad_norm": 0.3966197371482849,
      "learning_rate": 6.139914625554313e-07,
      "loss": 0.6219,
      "step": 21167
    },
    {
      "epoch": 0.8772845953002611,
      "grad_norm": 0.4305345118045807,
      "learning_rate": 6.137842430270629e-07,
      "loss": 0.7175,
      "step": 21168
    },
    {
      "epoch": 0.8773260392059348,
      "grad_norm": 0.44699615240097046,
      "learning_rate": 6.135770234986946e-07,
      "loss": 0.7095,
      "step": 21169
    },
    {
      "epoch": 0.8773674831116084,
      "grad_norm": 0.4194971024990082,
      "learning_rate": 6.133698039703262e-07,
      "loss": 0.6501,
      "step": 21170
    },
    {
      "epoch": 0.8774089270172821,
      "grad_norm": 0.4412505328655243,
      "learning_rate": 6.131625844419579e-07,
      "loss": 0.6674,
      "step": 21171
    },
    {
      "epoch": 0.8774503709229557,
      "grad_norm": 0.41273123025894165,
      "learning_rate": 6.129553649135895e-07,
      "loss": 0.6448,
      "step": 21172
    },
    {
      "epoch": 0.8774918148286295,
      "grad_norm": 0.4396305978298187,
      "learning_rate": 6.127481453852212e-07,
      "loss": 0.6968,
      "step": 21173
    },
    {
      "epoch": 0.8775332587343031,
      "grad_norm": 0.38865870237350464,
      "learning_rate": 6.125409258568528e-07,
      "loss": 0.6519,
      "step": 21174
    },
    {
      "epoch": 0.8775747026399768,
      "grad_norm": 0.39792460203170776,
      "learning_rate": 6.123337063284845e-07,
      "loss": 0.6466,
      "step": 21175
    },
    {
      "epoch": 0.8776161465456505,
      "grad_norm": 0.3990892767906189,
      "learning_rate": 6.121264868001161e-07,
      "loss": 0.6431,
      "step": 21176
    },
    {
      "epoch": 0.8776575904513242,
      "grad_norm": 0.39086923003196716,
      "learning_rate": 6.119192672717477e-07,
      "loss": 0.6038,
      "step": 21177
    },
    {
      "epoch": 0.8776990343569978,
      "grad_norm": 0.4340440034866333,
      "learning_rate": 6.117120477433794e-07,
      "loss": 0.6926,
      "step": 21178
    },
    {
      "epoch": 0.8777404782626714,
      "grad_norm": 0.41801774501800537,
      "learning_rate": 6.11504828215011e-07,
      "loss": 0.6538,
      "step": 21179
    },
    {
      "epoch": 0.8777819221683452,
      "grad_norm": 0.40189599990844727,
      "learning_rate": 6.112976086866427e-07,
      "loss": 0.6492,
      "step": 21180
    },
    {
      "epoch": 0.8778233660740188,
      "grad_norm": 0.4554777145385742,
      "learning_rate": 6.110903891582743e-07,
      "loss": 0.658,
      "step": 21181
    },
    {
      "epoch": 0.8778648099796925,
      "grad_norm": 0.385280579328537,
      "learning_rate": 6.10883169629906e-07,
      "loss": 0.6172,
      "step": 21182
    },
    {
      "epoch": 0.8779062538853661,
      "grad_norm": 0.3970181941986084,
      "learning_rate": 6.106759501015376e-07,
      "loss": 0.7048,
      "step": 21183
    },
    {
      "epoch": 0.8779476977910399,
      "grad_norm": 0.4412635266780853,
      "learning_rate": 6.104687305731693e-07,
      "loss": 0.6836,
      "step": 21184
    },
    {
      "epoch": 0.8779891416967135,
      "grad_norm": 0.3977997303009033,
      "learning_rate": 6.102615110448009e-07,
      "loss": 0.6342,
      "step": 21185
    },
    {
      "epoch": 0.8780305856023871,
      "grad_norm": 0.42402365803718567,
      "learning_rate": 6.100542915164326e-07,
      "loss": 0.6301,
      "step": 21186
    },
    {
      "epoch": 0.8780720295080608,
      "grad_norm": 0.4120456874370575,
      "learning_rate": 6.098470719880642e-07,
      "loss": 0.6562,
      "step": 21187
    },
    {
      "epoch": 0.8781134734137345,
      "grad_norm": 0.4150981307029724,
      "learning_rate": 6.096398524596959e-07,
      "loss": 0.6559,
      "step": 21188
    },
    {
      "epoch": 0.8781549173194082,
      "grad_norm": 0.46171069145202637,
      "learning_rate": 6.094326329313275e-07,
      "loss": 0.7285,
      "step": 21189
    },
    {
      "epoch": 0.8781963612250818,
      "grad_norm": 0.4841157793998718,
      "learning_rate": 6.092254134029591e-07,
      "loss": 0.7439,
      "step": 21190
    },
    {
      "epoch": 0.8782378051307556,
      "grad_norm": 0.3994734287261963,
      "learning_rate": 6.090181938745908e-07,
      "loss": 0.6266,
      "step": 21191
    },
    {
      "epoch": 0.8782792490364292,
      "grad_norm": 0.4370063543319702,
      "learning_rate": 6.088109743462224e-07,
      "loss": 0.7087,
      "step": 21192
    },
    {
      "epoch": 0.8783206929421029,
      "grad_norm": 0.4016975164413452,
      "learning_rate": 6.086037548178541e-07,
      "loss": 0.6544,
      "step": 21193
    },
    {
      "epoch": 0.8783621368477765,
      "grad_norm": 0.43727266788482666,
      "learning_rate": 6.083965352894857e-07,
      "loss": 0.7144,
      "step": 21194
    },
    {
      "epoch": 0.8784035807534502,
      "grad_norm": 0.4964316189289093,
      "learning_rate": 6.081893157611174e-07,
      "loss": 0.7661,
      "step": 21195
    },
    {
      "epoch": 0.8784450246591239,
      "grad_norm": 0.44300737977027893,
      "learning_rate": 6.07982096232749e-07,
      "loss": 0.6588,
      "step": 21196
    },
    {
      "epoch": 0.8784864685647975,
      "grad_norm": 0.3880154490470886,
      "learning_rate": 6.077748767043807e-07,
      "loss": 0.6753,
      "step": 21197
    },
    {
      "epoch": 0.8785279124704712,
      "grad_norm": 0.39382079243659973,
      "learning_rate": 6.075676571760123e-07,
      "loss": 0.6309,
      "step": 21198
    },
    {
      "epoch": 0.8785693563761449,
      "grad_norm": 0.4457118809223175,
      "learning_rate": 6.07360437647644e-07,
      "loss": 0.673,
      "step": 21199
    },
    {
      "epoch": 0.8786108002818186,
      "grad_norm": 0.4266894459724426,
      "learning_rate": 6.071532181192756e-07,
      "loss": 0.6327,
      "step": 21200
    },
    {
      "epoch": 0.8786522441874922,
      "grad_norm": 0.3981444835662842,
      "learning_rate": 6.069459985909073e-07,
      "loss": 0.6548,
      "step": 21201
    },
    {
      "epoch": 0.878693688093166,
      "grad_norm": 0.4332673251628876,
      "learning_rate": 6.067387790625389e-07,
      "loss": 0.6472,
      "step": 21202
    },
    {
      "epoch": 0.8787351319988396,
      "grad_norm": 0.40309593081474304,
      "learning_rate": 6.065315595341705e-07,
      "loss": 0.6715,
      "step": 21203
    },
    {
      "epoch": 0.8787765759045132,
      "grad_norm": 0.43588098883628845,
      "learning_rate": 6.063243400058022e-07,
      "loss": 0.6614,
      "step": 21204
    },
    {
      "epoch": 0.8788180198101869,
      "grad_norm": 0.3771733343601227,
      "learning_rate": 6.061171204774338e-07,
      "loss": 0.6388,
      "step": 21205
    },
    {
      "epoch": 0.8788594637158605,
      "grad_norm": 0.4000172019004822,
      "learning_rate": 6.059099009490655e-07,
      "loss": 0.6332,
      "step": 21206
    },
    {
      "epoch": 0.8789009076215343,
      "grad_norm": 0.4109748601913452,
      "learning_rate": 6.057026814206971e-07,
      "loss": 0.6387,
      "step": 21207
    },
    {
      "epoch": 0.8789423515272079,
      "grad_norm": 0.4211612641811371,
      "learning_rate": 6.054954618923288e-07,
      "loss": 0.6509,
      "step": 21208
    },
    {
      "epoch": 0.8789837954328816,
      "grad_norm": 0.41582900285720825,
      "learning_rate": 6.052882423639604e-07,
      "loss": 0.7034,
      "step": 21209
    },
    {
      "epoch": 0.8790252393385553,
      "grad_norm": 0.3982144892215729,
      "learning_rate": 6.050810228355921e-07,
      "loss": 0.5945,
      "step": 21210
    },
    {
      "epoch": 0.879066683244229,
      "grad_norm": 0.40995556116104126,
      "learning_rate": 6.048738033072237e-07,
      "loss": 0.6763,
      "step": 21211
    },
    {
      "epoch": 0.8791081271499026,
      "grad_norm": 0.5177196860313416,
      "learning_rate": 6.046665837788554e-07,
      "loss": 0.7014,
      "step": 21212
    },
    {
      "epoch": 0.8791495710555762,
      "grad_norm": 0.38519400358200073,
      "learning_rate": 6.04459364250487e-07,
      "loss": 0.6329,
      "step": 21213
    },
    {
      "epoch": 0.87919101496125,
      "grad_norm": 0.3544653058052063,
      "learning_rate": 6.042521447221186e-07,
      "loss": 0.5948,
      "step": 21214
    },
    {
      "epoch": 0.8792324588669236,
      "grad_norm": 0.41371381282806396,
      "learning_rate": 6.040449251937503e-07,
      "loss": 0.6869,
      "step": 21215
    },
    {
      "epoch": 0.8792739027725973,
      "grad_norm": 0.3963320851325989,
      "learning_rate": 6.038377056653819e-07,
      "loss": 0.6578,
      "step": 21216
    },
    {
      "epoch": 0.8793153466782709,
      "grad_norm": 0.44351184368133545,
      "learning_rate": 6.036304861370136e-07,
      "loss": 0.663,
      "step": 21217
    },
    {
      "epoch": 0.8793567905839447,
      "grad_norm": 0.42671236395835876,
      "learning_rate": 6.034232666086452e-07,
      "loss": 0.6646,
      "step": 21218
    },
    {
      "epoch": 0.8793982344896183,
      "grad_norm": 0.41453787684440613,
      "learning_rate": 6.032160470802769e-07,
      "loss": 0.6863,
      "step": 21219
    },
    {
      "epoch": 0.879439678395292,
      "grad_norm": 0.4558846652507782,
      "learning_rate": 6.030088275519085e-07,
      "loss": 0.6976,
      "step": 21220
    },
    {
      "epoch": 0.8794811223009656,
      "grad_norm": 0.3961840569972992,
      "learning_rate": 6.028016080235402e-07,
      "loss": 0.6587,
      "step": 21221
    },
    {
      "epoch": 0.8795225662066393,
      "grad_norm": 0.397318571805954,
      "learning_rate": 6.025943884951718e-07,
      "loss": 0.62,
      "step": 21222
    },
    {
      "epoch": 0.879564010112313,
      "grad_norm": 0.42686906456947327,
      "learning_rate": 6.023871689668035e-07,
      "loss": 0.6252,
      "step": 21223
    },
    {
      "epoch": 0.8796054540179866,
      "grad_norm": 0.3899343013763428,
      "learning_rate": 6.021799494384351e-07,
      "loss": 0.64,
      "step": 21224
    },
    {
      "epoch": 0.8796468979236604,
      "grad_norm": 0.4354437589645386,
      "learning_rate": 6.019727299100668e-07,
      "loss": 0.6409,
      "step": 21225
    },
    {
      "epoch": 0.879688341829334,
      "grad_norm": 0.3970591127872467,
      "learning_rate": 6.017655103816984e-07,
      "loss": 0.6385,
      "step": 21226
    },
    {
      "epoch": 0.8797297857350077,
      "grad_norm": 0.41292405128479004,
      "learning_rate": 6.0155829085333e-07,
      "loss": 0.6653,
      "step": 21227
    },
    {
      "epoch": 0.8797712296406813,
      "grad_norm": 0.42619726061820984,
      "learning_rate": 6.013510713249617e-07,
      "loss": 0.6836,
      "step": 21228
    },
    {
      "epoch": 0.8798126735463551,
      "grad_norm": 0.4213118851184845,
      "learning_rate": 6.011438517965933e-07,
      "loss": 0.6665,
      "step": 21229
    },
    {
      "epoch": 0.8798541174520287,
      "grad_norm": 0.4193699359893799,
      "learning_rate": 6.00936632268225e-07,
      "loss": 0.7043,
      "step": 21230
    },
    {
      "epoch": 0.8798955613577023,
      "grad_norm": 0.4275221526622772,
      "learning_rate": 6.007294127398566e-07,
      "loss": 0.5917,
      "step": 21231
    },
    {
      "epoch": 0.879937005263376,
      "grad_norm": 0.4113697409629822,
      "learning_rate": 6.005221932114883e-07,
      "loss": 0.6556,
      "step": 21232
    },
    {
      "epoch": 0.8799784491690497,
      "grad_norm": 0.401626318693161,
      "learning_rate": 6.003149736831199e-07,
      "loss": 0.6436,
      "step": 21233
    },
    {
      "epoch": 0.8800198930747234,
      "grad_norm": 0.42154234647750854,
      "learning_rate": 6.001077541547516e-07,
      "loss": 0.6644,
      "step": 21234
    },
    {
      "epoch": 0.880061336980397,
      "grad_norm": 0.40765756368637085,
      "learning_rate": 5.999005346263832e-07,
      "loss": 0.6636,
      "step": 21235
    },
    {
      "epoch": 0.8801027808860707,
      "grad_norm": 0.4323737323284149,
      "learning_rate": 5.996933150980149e-07,
      "loss": 0.6565,
      "step": 21236
    },
    {
      "epoch": 0.8801442247917444,
      "grad_norm": 0.42075368762016296,
      "learning_rate": 5.994860955696465e-07,
      "loss": 0.655,
      "step": 21237
    },
    {
      "epoch": 0.8801856686974181,
      "grad_norm": 0.4402667284011841,
      "learning_rate": 5.992788760412782e-07,
      "loss": 0.6195,
      "step": 21238
    },
    {
      "epoch": 0.8802271126030917,
      "grad_norm": 0.4020743668079376,
      "learning_rate": 5.990716565129098e-07,
      "loss": 0.6056,
      "step": 21239
    },
    {
      "epoch": 0.8802685565087653,
      "grad_norm": 0.4630289077758789,
      "learning_rate": 5.988644369845414e-07,
      "loss": 0.6586,
      "step": 21240
    },
    {
      "epoch": 0.8803100004144391,
      "grad_norm": 0.4193219244480133,
      "learning_rate": 5.986572174561731e-07,
      "loss": 0.6516,
      "step": 21241
    },
    {
      "epoch": 0.8803514443201127,
      "grad_norm": 0.4290623962879181,
      "learning_rate": 5.984499979278047e-07,
      "loss": 0.6736,
      "step": 21242
    },
    {
      "epoch": 0.8803928882257864,
      "grad_norm": 0.43410077691078186,
      "learning_rate": 5.982427783994364e-07,
      "loss": 0.6534,
      "step": 21243
    },
    {
      "epoch": 0.88043433213146,
      "grad_norm": 0.40461185574531555,
      "learning_rate": 5.98035558871068e-07,
      "loss": 0.6373,
      "step": 21244
    },
    {
      "epoch": 0.8804757760371338,
      "grad_norm": 0.40267953276634216,
      "learning_rate": 5.978283393426997e-07,
      "loss": 0.6079,
      "step": 21245
    },
    {
      "epoch": 0.8805172199428074,
      "grad_norm": 0.4229053556919098,
      "learning_rate": 5.976211198143313e-07,
      "loss": 0.7283,
      "step": 21246
    },
    {
      "epoch": 0.880558663848481,
      "grad_norm": 0.4041632115840912,
      "learning_rate": 5.97413900285963e-07,
      "loss": 0.6349,
      "step": 21247
    },
    {
      "epoch": 0.8806001077541548,
      "grad_norm": 0.40068912506103516,
      "learning_rate": 5.972066807575946e-07,
      "loss": 0.6384,
      "step": 21248
    },
    {
      "epoch": 0.8806415516598284,
      "grad_norm": 0.49801430106163025,
      "learning_rate": 5.969994612292263e-07,
      "loss": 0.7448,
      "step": 21249
    },
    {
      "epoch": 0.8806829955655021,
      "grad_norm": 0.39630112051963806,
      "learning_rate": 5.967922417008579e-07,
      "loss": 0.6049,
      "step": 21250
    },
    {
      "epoch": 0.8807244394711757,
      "grad_norm": 0.4059813320636749,
      "learning_rate": 5.965850221724896e-07,
      "loss": 0.6573,
      "step": 21251
    },
    {
      "epoch": 0.8807658833768495,
      "grad_norm": 0.41820117831230164,
      "learning_rate": 5.963778026441212e-07,
      "loss": 0.6666,
      "step": 21252
    },
    {
      "epoch": 0.8808073272825231,
      "grad_norm": 0.43901243805885315,
      "learning_rate": 5.961705831157529e-07,
      "loss": 0.6279,
      "step": 21253
    },
    {
      "epoch": 0.8808487711881968,
      "grad_norm": 0.4226698875427246,
      "learning_rate": 5.959633635873845e-07,
      "loss": 0.6538,
      "step": 21254
    },
    {
      "epoch": 0.8808902150938704,
      "grad_norm": 0.44338512420654297,
      "learning_rate": 5.957561440590162e-07,
      "loss": 0.6517,
      "step": 21255
    },
    {
      "epoch": 0.8809316589995441,
      "grad_norm": 0.4916036128997803,
      "learning_rate": 5.955489245306478e-07,
      "loss": 0.6963,
      "step": 21256
    },
    {
      "epoch": 0.8809731029052178,
      "grad_norm": 0.39940357208251953,
      "learning_rate": 5.953417050022795e-07,
      "loss": 0.6791,
      "step": 21257
    },
    {
      "epoch": 0.8810145468108914,
      "grad_norm": 0.43063634634017944,
      "learning_rate": 5.951344854739111e-07,
      "loss": 0.6755,
      "step": 21258
    },
    {
      "epoch": 0.8810559907165652,
      "grad_norm": 0.4553086757659912,
      "learning_rate": 5.949272659455427e-07,
      "loss": 0.6721,
      "step": 21259
    },
    {
      "epoch": 0.8810974346222388,
      "grad_norm": 0.4346065819263458,
      "learning_rate": 5.947200464171744e-07,
      "loss": 0.6426,
      "step": 21260
    },
    {
      "epoch": 0.8811388785279125,
      "grad_norm": 0.45353439450263977,
      "learning_rate": 5.94512826888806e-07,
      "loss": 0.7029,
      "step": 21261
    },
    {
      "epoch": 0.8811803224335861,
      "grad_norm": 0.4067379832267761,
      "learning_rate": 5.943056073604377e-07,
      "loss": 0.6959,
      "step": 21262
    },
    {
      "epoch": 0.8812217663392599,
      "grad_norm": 0.41099658608436584,
      "learning_rate": 5.940983878320693e-07,
      "loss": 0.6663,
      "step": 21263
    },
    {
      "epoch": 0.8812632102449335,
      "grad_norm": 0.4050459563732147,
      "learning_rate": 5.93891168303701e-07,
      "loss": 0.676,
      "step": 21264
    },
    {
      "epoch": 0.8813046541506071,
      "grad_norm": 0.43283700942993164,
      "learning_rate": 5.936839487753326e-07,
      "loss": 0.7485,
      "step": 21265
    },
    {
      "epoch": 0.8813460980562808,
      "grad_norm": 0.40667301416397095,
      "learning_rate": 5.934767292469643e-07,
      "loss": 0.641,
      "step": 21266
    },
    {
      "epoch": 0.8813875419619545,
      "grad_norm": 0.3854729235172272,
      "learning_rate": 5.932695097185959e-07,
      "loss": 0.6256,
      "step": 21267
    },
    {
      "epoch": 0.8814289858676282,
      "grad_norm": 0.42123374342918396,
      "learning_rate": 5.930622901902276e-07,
      "loss": 0.6699,
      "step": 21268
    },
    {
      "epoch": 0.8814704297733018,
      "grad_norm": 0.39144134521484375,
      "learning_rate": 5.928550706618592e-07,
      "loss": 0.6841,
      "step": 21269
    },
    {
      "epoch": 0.8815118736789755,
      "grad_norm": 0.4290797710418701,
      "learning_rate": 5.926478511334909e-07,
      "loss": 0.7054,
      "step": 21270
    },
    {
      "epoch": 0.8815533175846492,
      "grad_norm": 0.4198606312274933,
      "learning_rate": 5.924406316051225e-07,
      "loss": 0.6694,
      "step": 21271
    },
    {
      "epoch": 0.8815947614903229,
      "grad_norm": 0.4345215857028961,
      "learning_rate": 5.922334120767542e-07,
      "loss": 0.7074,
      "step": 21272
    },
    {
      "epoch": 0.8816362053959965,
      "grad_norm": 0.400138795375824,
      "learning_rate": 5.920261925483858e-07,
      "loss": 0.6615,
      "step": 21273
    },
    {
      "epoch": 0.8816776493016701,
      "grad_norm": 0.40557560324668884,
      "learning_rate": 5.918189730200175e-07,
      "loss": 0.7346,
      "step": 21274
    },
    {
      "epoch": 0.8817190932073439,
      "grad_norm": 0.45335280895233154,
      "learning_rate": 5.916117534916491e-07,
      "loss": 0.6381,
      "step": 21275
    },
    {
      "epoch": 0.8817605371130175,
      "grad_norm": 0.41442054510116577,
      "learning_rate": 5.914045339632808e-07,
      "loss": 0.7017,
      "step": 21276
    },
    {
      "epoch": 0.8818019810186912,
      "grad_norm": 0.4143434762954712,
      "learning_rate": 5.911973144349124e-07,
      "loss": 0.6693,
      "step": 21277
    },
    {
      "epoch": 0.8818434249243648,
      "grad_norm": 0.3999609053134918,
      "learning_rate": 5.90990094906544e-07,
      "loss": 0.6685,
      "step": 21278
    },
    {
      "epoch": 0.8818848688300386,
      "grad_norm": 0.4353742301464081,
      "learning_rate": 5.907828753781757e-07,
      "loss": 0.688,
      "step": 21279
    },
    {
      "epoch": 0.8819263127357122,
      "grad_norm": 0.40500181913375854,
      "learning_rate": 5.905756558498073e-07,
      "loss": 0.6571,
      "step": 21280
    },
    {
      "epoch": 0.8819677566413859,
      "grad_norm": 0.3766523599624634,
      "learning_rate": 5.90368436321439e-07,
      "loss": 0.6405,
      "step": 21281
    },
    {
      "epoch": 0.8820092005470596,
      "grad_norm": 0.40084701776504517,
      "learning_rate": 5.901612167930706e-07,
      "loss": 0.6388,
      "step": 21282
    },
    {
      "epoch": 0.8820506444527332,
      "grad_norm": 0.39553922414779663,
      "learning_rate": 5.899539972647023e-07,
      "loss": 0.671,
      "step": 21283
    },
    {
      "epoch": 0.8820920883584069,
      "grad_norm": 0.4461310803890228,
      "learning_rate": 5.897467777363339e-07,
      "loss": 0.6897,
      "step": 21284
    },
    {
      "epoch": 0.8821335322640805,
      "grad_norm": 0.4045344591140747,
      "learning_rate": 5.895395582079656e-07,
      "loss": 0.656,
      "step": 21285
    },
    {
      "epoch": 0.8821749761697543,
      "grad_norm": 0.38338449597358704,
      "learning_rate": 5.893323386795972e-07,
      "loss": 0.6578,
      "step": 21286
    },
    {
      "epoch": 0.8822164200754279,
      "grad_norm": 0.4106886386871338,
      "learning_rate": 5.891251191512289e-07,
      "loss": 0.6997,
      "step": 21287
    },
    {
      "epoch": 0.8822578639811016,
      "grad_norm": 0.6928749680519104,
      "learning_rate": 5.889178996228605e-07,
      "loss": 0.6553,
      "step": 21288
    },
    {
      "epoch": 0.8822993078867752,
      "grad_norm": 0.4545588493347168,
      "learning_rate": 5.887106800944922e-07,
      "loss": 0.6865,
      "step": 21289
    },
    {
      "epoch": 0.882340751792449,
      "grad_norm": 0.48300936818122864,
      "learning_rate": 5.885034605661238e-07,
      "loss": 0.6462,
      "step": 21290
    },
    {
      "epoch": 0.8823821956981226,
      "grad_norm": 0.40506988763809204,
      "learning_rate": 5.882962410377555e-07,
      "loss": 0.6299,
      "step": 21291
    },
    {
      "epoch": 0.8824236396037962,
      "grad_norm": 0.4171604514122009,
      "learning_rate": 5.880890215093871e-07,
      "loss": 0.7,
      "step": 21292
    },
    {
      "epoch": 0.88246508350947,
      "grad_norm": 0.40383443236351013,
      "learning_rate": 5.878818019810188e-07,
      "loss": 0.674,
      "step": 21293
    },
    {
      "epoch": 0.8825065274151436,
      "grad_norm": 0.38713788986206055,
      "learning_rate": 5.876745824526504e-07,
      "loss": 0.6255,
      "step": 21294
    },
    {
      "epoch": 0.8825479713208173,
      "grad_norm": 0.4053410291671753,
      "learning_rate": 5.874673629242821e-07,
      "loss": 0.6407,
      "step": 21295
    },
    {
      "epoch": 0.8825894152264909,
      "grad_norm": 0.39805781841278076,
      "learning_rate": 5.872601433959137e-07,
      "loss": 0.6461,
      "step": 21296
    },
    {
      "epoch": 0.8826308591321647,
      "grad_norm": 0.43926721811294556,
      "learning_rate": 5.870529238675453e-07,
      "loss": 0.6709,
      "step": 21297
    },
    {
      "epoch": 0.8826723030378383,
      "grad_norm": 0.4555812180042267,
      "learning_rate": 5.86845704339177e-07,
      "loss": 0.6035,
      "step": 21298
    },
    {
      "epoch": 0.882713746943512,
      "grad_norm": 0.4007062315940857,
      "learning_rate": 5.866384848108086e-07,
      "loss": 0.6633,
      "step": 21299
    },
    {
      "epoch": 0.8827551908491856,
      "grad_norm": 0.3850429058074951,
      "learning_rate": 5.864312652824403e-07,
      "loss": 0.6608,
      "step": 21300
    },
    {
      "epoch": 0.8827966347548593,
      "grad_norm": 0.45208337903022766,
      "learning_rate": 5.862240457540719e-07,
      "loss": 0.6851,
      "step": 21301
    },
    {
      "epoch": 0.882838078660533,
      "grad_norm": 0.4389796853065491,
      "learning_rate": 5.860168262257036e-07,
      "loss": 0.6719,
      "step": 21302
    },
    {
      "epoch": 0.8828795225662066,
      "grad_norm": 0.39704954624176025,
      "learning_rate": 5.858096066973352e-07,
      "loss": 0.6296,
      "step": 21303
    },
    {
      "epoch": 0.8829209664718803,
      "grad_norm": 0.4411863684654236,
      "learning_rate": 5.856023871689669e-07,
      "loss": 0.7302,
      "step": 21304
    },
    {
      "epoch": 0.882962410377554,
      "grad_norm": 0.3994826376438141,
      "learning_rate": 5.853951676405985e-07,
      "loss": 0.6729,
      "step": 21305
    },
    {
      "epoch": 0.8830038542832277,
      "grad_norm": 0.44235241413116455,
      "learning_rate": 5.851879481122302e-07,
      "loss": 0.6794,
      "step": 21306
    },
    {
      "epoch": 0.8830452981889013,
      "grad_norm": 0.4170055091381073,
      "learning_rate": 5.849807285838618e-07,
      "loss": 0.6632,
      "step": 21307
    },
    {
      "epoch": 0.8830867420945749,
      "grad_norm": 0.3976626992225647,
      "learning_rate": 5.847735090554935e-07,
      "loss": 0.6895,
      "step": 21308
    },
    {
      "epoch": 0.8831281860002487,
      "grad_norm": 0.42260727286338806,
      "learning_rate": 5.845662895271251e-07,
      "loss": 0.7009,
      "step": 21309
    },
    {
      "epoch": 0.8831696299059223,
      "grad_norm": 0.41299498081207275,
      "learning_rate": 5.843590699987568e-07,
      "loss": 0.6184,
      "step": 21310
    },
    {
      "epoch": 0.883211073811596,
      "grad_norm": 0.4086816608905792,
      "learning_rate": 5.841518504703884e-07,
      "loss": 0.673,
      "step": 21311
    },
    {
      "epoch": 0.8832525177172696,
      "grad_norm": 0.434195876121521,
      "learning_rate": 5.839446309420201e-07,
      "loss": 0.6509,
      "step": 21312
    },
    {
      "epoch": 0.8832939616229434,
      "grad_norm": 0.4236471652984619,
      "learning_rate": 5.837374114136517e-07,
      "loss": 0.7057,
      "step": 21313
    },
    {
      "epoch": 0.883335405528617,
      "grad_norm": 0.39274778962135315,
      "learning_rate": 5.835301918852833e-07,
      "loss": 0.6697,
      "step": 21314
    },
    {
      "epoch": 0.8833768494342907,
      "grad_norm": 0.38926005363464355,
      "learning_rate": 5.83322972356915e-07,
      "loss": 0.6917,
      "step": 21315
    },
    {
      "epoch": 0.8834182933399644,
      "grad_norm": 0.4058479964733124,
      "learning_rate": 5.831157528285466e-07,
      "loss": 0.65,
      "step": 21316
    },
    {
      "epoch": 0.883459737245638,
      "grad_norm": 0.39445558190345764,
      "learning_rate": 5.829085333001783e-07,
      "loss": 0.6544,
      "step": 21317
    },
    {
      "epoch": 0.8835011811513117,
      "grad_norm": 0.4244757294654846,
      "learning_rate": 5.827013137718099e-07,
      "loss": 0.7065,
      "step": 21318
    },
    {
      "epoch": 0.8835426250569853,
      "grad_norm": 0.43243154883384705,
      "learning_rate": 5.824940942434416e-07,
      "loss": 0.6478,
      "step": 21319
    },
    {
      "epoch": 0.8835840689626591,
      "grad_norm": 0.3923470079898834,
      "learning_rate": 5.822868747150732e-07,
      "loss": 0.6278,
      "step": 21320
    },
    {
      "epoch": 0.8836255128683327,
      "grad_norm": 0.44109416007995605,
      "learning_rate": 5.820796551867049e-07,
      "loss": 0.6639,
      "step": 21321
    },
    {
      "epoch": 0.8836669567740064,
      "grad_norm": 0.42157456278800964,
      "learning_rate": 5.818724356583365e-07,
      "loss": 0.6692,
      "step": 21322
    },
    {
      "epoch": 0.88370840067968,
      "grad_norm": 0.4430670738220215,
      "learning_rate": 5.816652161299682e-07,
      "loss": 0.655,
      "step": 21323
    },
    {
      "epoch": 0.8837498445853538,
      "grad_norm": 0.43569067120552063,
      "learning_rate": 5.814579966015998e-07,
      "loss": 0.7205,
      "step": 21324
    },
    {
      "epoch": 0.8837912884910274,
      "grad_norm": 0.42636728286743164,
      "learning_rate": 5.812507770732314e-07,
      "loss": 0.6936,
      "step": 21325
    },
    {
      "epoch": 0.883832732396701,
      "grad_norm": 0.4302840530872345,
      "learning_rate": 5.810435575448631e-07,
      "loss": 0.7065,
      "step": 21326
    },
    {
      "epoch": 0.8838741763023747,
      "grad_norm": 0.4147132635116577,
      "learning_rate": 5.808363380164947e-07,
      "loss": 0.6721,
      "step": 21327
    },
    {
      "epoch": 0.8839156202080484,
      "grad_norm": 0.38858553767204285,
      "learning_rate": 5.806291184881264e-07,
      "loss": 0.63,
      "step": 21328
    },
    {
      "epoch": 0.8839570641137221,
      "grad_norm": 0.3891463279724121,
      "learning_rate": 5.80421898959758e-07,
      "loss": 0.6357,
      "step": 21329
    },
    {
      "epoch": 0.8839985080193957,
      "grad_norm": 0.3993394672870636,
      "learning_rate": 5.802146794313897e-07,
      "loss": 0.64,
      "step": 21330
    },
    {
      "epoch": 0.8840399519250695,
      "grad_norm": 0.4401186406612396,
      "learning_rate": 5.800074599030213e-07,
      "loss": 0.6859,
      "step": 21331
    },
    {
      "epoch": 0.8840813958307431,
      "grad_norm": 0.4040793180465698,
      "learning_rate": 5.79800240374653e-07,
      "loss": 0.6753,
      "step": 21332
    },
    {
      "epoch": 0.8841228397364168,
      "grad_norm": 0.41444963216781616,
      "learning_rate": 5.795930208462846e-07,
      "loss": 0.646,
      "step": 21333
    },
    {
      "epoch": 0.8841642836420904,
      "grad_norm": 0.37345439195632935,
      "learning_rate": 5.793858013179163e-07,
      "loss": 0.6665,
      "step": 21334
    },
    {
      "epoch": 0.884205727547764,
      "grad_norm": 0.4315361976623535,
      "learning_rate": 5.791785817895479e-07,
      "loss": 0.7035,
      "step": 21335
    },
    {
      "epoch": 0.8842471714534378,
      "grad_norm": 0.3848472237586975,
      "learning_rate": 5.789713622611796e-07,
      "loss": 0.6428,
      "step": 21336
    },
    {
      "epoch": 0.8842886153591114,
      "grad_norm": 0.4175746738910675,
      "learning_rate": 5.787641427328112e-07,
      "loss": 0.6957,
      "step": 21337
    },
    {
      "epoch": 0.8843300592647851,
      "grad_norm": 0.3899771571159363,
      "learning_rate": 5.785569232044428e-07,
      "loss": 0.6012,
      "step": 21338
    },
    {
      "epoch": 0.8843715031704588,
      "grad_norm": 0.4082960784435272,
      "learning_rate": 5.783497036760745e-07,
      "loss": 0.6299,
      "step": 21339
    },
    {
      "epoch": 0.8844129470761325,
      "grad_norm": 0.40709570050239563,
      "learning_rate": 5.781424841477061e-07,
      "loss": 0.6084,
      "step": 21340
    },
    {
      "epoch": 0.8844543909818061,
      "grad_norm": 0.4515201449394226,
      "learning_rate": 5.779352646193378e-07,
      "loss": 0.6829,
      "step": 21341
    },
    {
      "epoch": 0.8844958348874798,
      "grad_norm": 0.4040144383907318,
      "learning_rate": 5.777280450909694e-07,
      "loss": 0.6584,
      "step": 21342
    },
    {
      "epoch": 0.8845372787931535,
      "grad_norm": 0.4204562306404114,
      "learning_rate": 5.775208255626011e-07,
      "loss": 0.6538,
      "step": 21343
    },
    {
      "epoch": 0.8845787226988271,
      "grad_norm": 0.41292986273765564,
      "learning_rate": 5.773136060342327e-07,
      "loss": 0.662,
      "step": 21344
    },
    {
      "epoch": 0.8846201666045008,
      "grad_norm": 0.4320838749408722,
      "learning_rate": 5.771063865058644e-07,
      "loss": 0.6692,
      "step": 21345
    },
    {
      "epoch": 0.8846616105101744,
      "grad_norm": 0.4192841053009033,
      "learning_rate": 5.76899166977496e-07,
      "loss": 0.6758,
      "step": 21346
    },
    {
      "epoch": 0.8847030544158482,
      "grad_norm": 0.40273743867874146,
      "learning_rate": 5.766919474491277e-07,
      "loss": 0.6781,
      "step": 21347
    },
    {
      "epoch": 0.8847444983215218,
      "grad_norm": 0.409249871969223,
      "learning_rate": 5.764847279207593e-07,
      "loss": 0.6335,
      "step": 21348
    },
    {
      "epoch": 0.8847859422271955,
      "grad_norm": 0.41869544982910156,
      "learning_rate": 5.76277508392391e-07,
      "loss": 0.7096,
      "step": 21349
    },
    {
      "epoch": 0.8848273861328692,
      "grad_norm": 0.41468721628189087,
      "learning_rate": 5.760702888640226e-07,
      "loss": 0.6143,
      "step": 21350
    },
    {
      "epoch": 0.8848688300385429,
      "grad_norm": 0.3962930738925934,
      "learning_rate": 5.758630693356542e-07,
      "loss": 0.6349,
      "step": 21351
    },
    {
      "epoch": 0.8849102739442165,
      "grad_norm": 0.40675088763237,
      "learning_rate": 5.756558498072859e-07,
      "loss": 0.6156,
      "step": 21352
    },
    {
      "epoch": 0.8849517178498901,
      "grad_norm": 0.4155443608760834,
      "learning_rate": 5.754486302789175e-07,
      "loss": 0.6562,
      "step": 21353
    },
    {
      "epoch": 0.8849931617555639,
      "grad_norm": 0.4477280080318451,
      "learning_rate": 5.752414107505492e-07,
      "loss": 0.6943,
      "step": 21354
    },
    {
      "epoch": 0.8850346056612375,
      "grad_norm": 0.4371253550052643,
      "learning_rate": 5.750341912221808e-07,
      "loss": 0.6763,
      "step": 21355
    },
    {
      "epoch": 0.8850760495669112,
      "grad_norm": 0.3861115574836731,
      "learning_rate": 5.748269716938125e-07,
      "loss": 0.6855,
      "step": 21356
    },
    {
      "epoch": 0.8851174934725848,
      "grad_norm": 0.41206416487693787,
      "learning_rate": 5.746197521654441e-07,
      "loss": 0.6265,
      "step": 21357
    },
    {
      "epoch": 0.8851589373782586,
      "grad_norm": 0.42784106731414795,
      "learning_rate": 5.744125326370758e-07,
      "loss": 0.6604,
      "step": 21358
    },
    {
      "epoch": 0.8852003812839322,
      "grad_norm": 0.4599573314189911,
      "learning_rate": 5.742053131087074e-07,
      "loss": 0.6892,
      "step": 21359
    },
    {
      "epoch": 0.8852418251896058,
      "grad_norm": 0.426604300737381,
      "learning_rate": 5.739980935803391e-07,
      "loss": 0.6624,
      "step": 21360
    },
    {
      "epoch": 0.8852832690952795,
      "grad_norm": 0.40245509147644043,
      "learning_rate": 5.737908740519707e-07,
      "loss": 0.6777,
      "step": 21361
    },
    {
      "epoch": 0.8853247130009532,
      "grad_norm": 0.4394230842590332,
      "learning_rate": 5.735836545236023e-07,
      "loss": 0.7209,
      "step": 21362
    },
    {
      "epoch": 0.8853661569066269,
      "grad_norm": 0.41601940989494324,
      "learning_rate": 5.73376434995234e-07,
      "loss": 0.6803,
      "step": 21363
    },
    {
      "epoch": 0.8854076008123005,
      "grad_norm": 0.4124889075756073,
      "learning_rate": 5.731692154668656e-07,
      "loss": 0.7029,
      "step": 21364
    },
    {
      "epoch": 0.8854490447179743,
      "grad_norm": 0.41583630442619324,
      "learning_rate": 5.729619959384973e-07,
      "loss": 0.6561,
      "step": 21365
    },
    {
      "epoch": 0.8854904886236479,
      "grad_norm": 0.47018930315971375,
      "learning_rate": 5.727547764101289e-07,
      "loss": 0.7701,
      "step": 21366
    },
    {
      "epoch": 0.8855319325293216,
      "grad_norm": 0.4338330328464508,
      "learning_rate": 5.725475568817606e-07,
      "loss": 0.6775,
      "step": 21367
    },
    {
      "epoch": 0.8855733764349952,
      "grad_norm": 0.43430954217910767,
      "learning_rate": 5.723403373533922e-07,
      "loss": 0.7552,
      "step": 21368
    },
    {
      "epoch": 0.8856148203406689,
      "grad_norm": 0.42488640546798706,
      "learning_rate": 5.721331178250239e-07,
      "loss": 0.6528,
      "step": 21369
    },
    {
      "epoch": 0.8856562642463426,
      "grad_norm": 0.4809923768043518,
      "learning_rate": 5.719258982966555e-07,
      "loss": 0.6588,
      "step": 21370
    },
    {
      "epoch": 0.8856977081520162,
      "grad_norm": 0.4260718822479248,
      "learning_rate": 5.717186787682872e-07,
      "loss": 0.6392,
      "step": 21371
    },
    {
      "epoch": 0.8857391520576899,
      "grad_norm": 0.42809367179870605,
      "learning_rate": 5.715114592399188e-07,
      "loss": 0.686,
      "step": 21372
    },
    {
      "epoch": 0.8857805959633636,
      "grad_norm": 0.4245171546936035,
      "learning_rate": 5.713042397115505e-07,
      "loss": 0.6946,
      "step": 21373
    },
    {
      "epoch": 0.8858220398690373,
      "grad_norm": 0.44846656918525696,
      "learning_rate": 5.710970201831821e-07,
      "loss": 0.6609,
      "step": 21374
    },
    {
      "epoch": 0.8858634837747109,
      "grad_norm": 0.44118067622184753,
      "learning_rate": 5.708898006548137e-07,
      "loss": 0.6614,
      "step": 21375
    },
    {
      "epoch": 0.8859049276803846,
      "grad_norm": 0.42059674859046936,
      "learning_rate": 5.706825811264454e-07,
      "loss": 0.6552,
      "step": 21376
    },
    {
      "epoch": 0.8859463715860583,
      "grad_norm": 0.4402751326560974,
      "learning_rate": 5.70475361598077e-07,
      "loss": 0.6614,
      "step": 21377
    },
    {
      "epoch": 0.8859878154917319,
      "grad_norm": 0.40249964594841003,
      "learning_rate": 5.702681420697087e-07,
      "loss": 0.6755,
      "step": 21378
    },
    {
      "epoch": 0.8860292593974056,
      "grad_norm": 0.4175712466239929,
      "learning_rate": 5.700609225413403e-07,
      "loss": 0.6201,
      "step": 21379
    },
    {
      "epoch": 0.8860707033030792,
      "grad_norm": 0.4190959930419922,
      "learning_rate": 5.69853703012972e-07,
      "loss": 0.6951,
      "step": 21380
    },
    {
      "epoch": 0.886112147208753,
      "grad_norm": 0.4328671395778656,
      "learning_rate": 5.696464834846036e-07,
      "loss": 0.6935,
      "step": 21381
    },
    {
      "epoch": 0.8861535911144266,
      "grad_norm": 0.4444023370742798,
      "learning_rate": 5.694392639562353e-07,
      "loss": 0.7029,
      "step": 21382
    },
    {
      "epoch": 0.8861950350201003,
      "grad_norm": 0.3819735646247864,
      "learning_rate": 5.692320444278669e-07,
      "loss": 0.6147,
      "step": 21383
    },
    {
      "epoch": 0.886236478925774,
      "grad_norm": 0.41260862350463867,
      "learning_rate": 5.690248248994986e-07,
      "loss": 0.6453,
      "step": 21384
    },
    {
      "epoch": 0.8862779228314477,
      "grad_norm": 0.4013561010360718,
      "learning_rate": 5.688176053711302e-07,
      "loss": 0.6814,
      "step": 21385
    },
    {
      "epoch": 0.8863193667371213,
      "grad_norm": 0.4620875418186188,
      "learning_rate": 5.686103858427619e-07,
      "loss": 0.7476,
      "step": 21386
    },
    {
      "epoch": 0.8863608106427949,
      "grad_norm": 0.4387966990470886,
      "learning_rate": 5.684031663143935e-07,
      "loss": 0.671,
      "step": 21387
    },
    {
      "epoch": 0.8864022545484687,
      "grad_norm": 0.40118181705474854,
      "learning_rate": 5.681959467860251e-07,
      "loss": 0.6296,
      "step": 21388
    },
    {
      "epoch": 0.8864436984541423,
      "grad_norm": 0.426576167345047,
      "learning_rate": 5.679887272576568e-07,
      "loss": 0.6582,
      "step": 21389
    },
    {
      "epoch": 0.886485142359816,
      "grad_norm": 0.3915785551071167,
      "learning_rate": 5.677815077292884e-07,
      "loss": 0.6382,
      "step": 21390
    },
    {
      "epoch": 0.8865265862654896,
      "grad_norm": 0.4120689034461975,
      "learning_rate": 5.675742882009201e-07,
      "loss": 0.6677,
      "step": 21391
    },
    {
      "epoch": 0.8865680301711634,
      "grad_norm": 0.41568928956985474,
      "learning_rate": 5.673670686725517e-07,
      "loss": 0.6726,
      "step": 21392
    },
    {
      "epoch": 0.886609474076837,
      "grad_norm": 0.4572696089744568,
      "learning_rate": 5.671598491441834e-07,
      "loss": 0.6711,
      "step": 21393
    },
    {
      "epoch": 0.8866509179825107,
      "grad_norm": 0.4576156735420227,
      "learning_rate": 5.66952629615815e-07,
      "loss": 0.6656,
      "step": 21394
    },
    {
      "epoch": 0.8866923618881843,
      "grad_norm": 0.42794856429100037,
      "learning_rate": 5.667454100874467e-07,
      "loss": 0.6865,
      "step": 21395
    },
    {
      "epoch": 0.886733805793858,
      "grad_norm": 0.42485445737838745,
      "learning_rate": 5.665381905590783e-07,
      "loss": 0.666,
      "step": 21396
    },
    {
      "epoch": 0.8867752496995317,
      "grad_norm": 0.40749481320381165,
      "learning_rate": 5.6633097103071e-07,
      "loss": 0.7169,
      "step": 21397
    },
    {
      "epoch": 0.8868166936052053,
      "grad_norm": 0.43550944328308105,
      "learning_rate": 5.661237515023416e-07,
      "loss": 0.6998,
      "step": 21398
    },
    {
      "epoch": 0.886858137510879,
      "grad_norm": 0.3828023672103882,
      "learning_rate": 5.659165319739732e-07,
      "loss": 0.647,
      "step": 21399
    },
    {
      "epoch": 0.8868995814165527,
      "grad_norm": 0.4325985908508301,
      "learning_rate": 5.657093124456049e-07,
      "loss": 0.6487,
      "step": 21400
    },
    {
      "epoch": 0.8869410253222264,
      "grad_norm": 0.46142271161079407,
      "learning_rate": 5.655020929172365e-07,
      "loss": 0.6792,
      "step": 21401
    },
    {
      "epoch": 0.8869824692279,
      "grad_norm": 0.4016042649745941,
      "learning_rate": 5.652948733888682e-07,
      "loss": 0.6659,
      "step": 21402
    },
    {
      "epoch": 0.8870239131335738,
      "grad_norm": 0.41583186388015747,
      "learning_rate": 5.650876538604998e-07,
      "loss": 0.6611,
      "step": 21403
    },
    {
      "epoch": 0.8870653570392474,
      "grad_norm": 0.4193524420261383,
      "learning_rate": 5.648804343321315e-07,
      "loss": 0.674,
      "step": 21404
    },
    {
      "epoch": 0.887106800944921,
      "grad_norm": 0.4389016926288605,
      "learning_rate": 5.646732148037631e-07,
      "loss": 0.6627,
      "step": 21405
    },
    {
      "epoch": 0.8871482448505947,
      "grad_norm": 0.41006413102149963,
      "learning_rate": 5.644659952753948e-07,
      "loss": 0.6987,
      "step": 21406
    },
    {
      "epoch": 0.8871896887562684,
      "grad_norm": 0.4075542986392975,
      "learning_rate": 5.642587757470264e-07,
      "loss": 0.6519,
      "step": 21407
    },
    {
      "epoch": 0.8872311326619421,
      "grad_norm": 0.3941805064678192,
      "learning_rate": 5.640515562186581e-07,
      "loss": 0.6753,
      "step": 21408
    },
    {
      "epoch": 0.8872725765676157,
      "grad_norm": 0.41101667284965515,
      "learning_rate": 5.638443366902897e-07,
      "loss": 0.6588,
      "step": 21409
    },
    {
      "epoch": 0.8873140204732894,
      "grad_norm": 0.4959043562412262,
      "learning_rate": 5.636371171619214e-07,
      "loss": 0.6927,
      "step": 21410
    },
    {
      "epoch": 0.8873554643789631,
      "grad_norm": 0.4106086194515228,
      "learning_rate": 5.63429897633553e-07,
      "loss": 0.6764,
      "step": 21411
    },
    {
      "epoch": 0.8873969082846368,
      "grad_norm": 0.4185197055339813,
      "learning_rate": 5.632226781051846e-07,
      "loss": 0.6982,
      "step": 21412
    },
    {
      "epoch": 0.8874383521903104,
      "grad_norm": 0.4126123785972595,
      "learning_rate": 5.630154585768163e-07,
      "loss": 0.6282,
      "step": 21413
    },
    {
      "epoch": 0.887479796095984,
      "grad_norm": 0.4600255787372589,
      "learning_rate": 5.628082390484479e-07,
      "loss": 0.6646,
      "step": 21414
    },
    {
      "epoch": 0.8875212400016578,
      "grad_norm": 0.42496591806411743,
      "learning_rate": 5.626010195200796e-07,
      "loss": 0.672,
      "step": 21415
    },
    {
      "epoch": 0.8875626839073314,
      "grad_norm": 0.42459532618522644,
      "learning_rate": 5.623937999917112e-07,
      "loss": 0.6699,
      "step": 21416
    },
    {
      "epoch": 0.8876041278130051,
      "grad_norm": 0.4087842106819153,
      "learning_rate": 5.621865804633429e-07,
      "loss": 0.6676,
      "step": 21417
    },
    {
      "epoch": 0.8876455717186787,
      "grad_norm": 0.41371971368789673,
      "learning_rate": 5.619793609349745e-07,
      "loss": 0.6721,
      "step": 21418
    },
    {
      "epoch": 0.8876870156243525,
      "grad_norm": 0.44439592957496643,
      "learning_rate": 5.617721414066062e-07,
      "loss": 0.6731,
      "step": 21419
    },
    {
      "epoch": 0.8877284595300261,
      "grad_norm": 0.4632638096809387,
      "learning_rate": 5.615649218782378e-07,
      "loss": 0.7107,
      "step": 21420
    },
    {
      "epoch": 0.8877699034356997,
      "grad_norm": 0.4751431345939636,
      "learning_rate": 5.613577023498695e-07,
      "loss": 0.7124,
      "step": 21421
    },
    {
      "epoch": 0.8878113473413735,
      "grad_norm": 0.4198697805404663,
      "learning_rate": 5.611504828215011e-07,
      "loss": 0.7062,
      "step": 21422
    },
    {
      "epoch": 0.8878527912470471,
      "grad_norm": 0.4256114363670349,
      "learning_rate": 5.609432632931328e-07,
      "loss": 0.6487,
      "step": 21423
    },
    {
      "epoch": 0.8878942351527208,
      "grad_norm": 0.44787174463272095,
      "learning_rate": 5.607360437647644e-07,
      "loss": 0.7546,
      "step": 21424
    },
    {
      "epoch": 0.8879356790583944,
      "grad_norm": 0.4393247067928314,
      "learning_rate": 5.60528824236396e-07,
      "loss": 0.6642,
      "step": 21425
    },
    {
      "epoch": 0.8879771229640682,
      "grad_norm": 0.40169140696525574,
      "learning_rate": 5.603216047080277e-07,
      "loss": 0.6677,
      "step": 21426
    },
    {
      "epoch": 0.8880185668697418,
      "grad_norm": 0.43607109785079956,
      "learning_rate": 5.601143851796593e-07,
      "loss": 0.6838,
      "step": 21427
    },
    {
      "epoch": 0.8880600107754155,
      "grad_norm": 0.40074434876441956,
      "learning_rate": 5.59907165651291e-07,
      "loss": 0.6422,
      "step": 21428
    },
    {
      "epoch": 0.8881014546810891,
      "grad_norm": 0.42218637466430664,
      "learning_rate": 5.596999461229226e-07,
      "loss": 0.6511,
      "step": 21429
    },
    {
      "epoch": 0.8881428985867628,
      "grad_norm": 0.4322926700115204,
      "learning_rate": 5.594927265945543e-07,
      "loss": 0.7112,
      "step": 21430
    },
    {
      "epoch": 0.8881843424924365,
      "grad_norm": 0.3780708611011505,
      "learning_rate": 5.592855070661859e-07,
      "loss": 0.6172,
      "step": 21431
    },
    {
      "epoch": 0.8882257863981101,
      "grad_norm": 0.4227933883666992,
      "learning_rate": 5.590782875378176e-07,
      "loss": 0.6642,
      "step": 21432
    },
    {
      "epoch": 0.8882672303037839,
      "grad_norm": 0.40627148747444153,
      "learning_rate": 5.588710680094492e-07,
      "loss": 0.67,
      "step": 21433
    },
    {
      "epoch": 0.8883086742094575,
      "grad_norm": 0.4355160593986511,
      "learning_rate": 5.586638484810809e-07,
      "loss": 0.6846,
      "step": 21434
    },
    {
      "epoch": 0.8883501181151312,
      "grad_norm": 0.4284313917160034,
      "learning_rate": 5.584566289527125e-07,
      "loss": 0.6895,
      "step": 21435
    },
    {
      "epoch": 0.8883915620208048,
      "grad_norm": 0.4165712893009186,
      "learning_rate": 5.582494094243441e-07,
      "loss": 0.6506,
      "step": 21436
    },
    {
      "epoch": 0.8884330059264786,
      "grad_norm": 0.4220913052558899,
      "learning_rate": 5.580421898959758e-07,
      "loss": 0.6646,
      "step": 21437
    },
    {
      "epoch": 0.8884744498321522,
      "grad_norm": 0.47515106201171875,
      "learning_rate": 5.578349703676074e-07,
      "loss": 0.7295,
      "step": 21438
    },
    {
      "epoch": 0.8885158937378258,
      "grad_norm": 0.42370641231536865,
      "learning_rate": 5.576277508392391e-07,
      "loss": 0.7219,
      "step": 21439
    },
    {
      "epoch": 0.8885573376434995,
      "grad_norm": 0.3939555585384369,
      "learning_rate": 5.574205313108707e-07,
      "loss": 0.6401,
      "step": 21440
    },
    {
      "epoch": 0.8885987815491732,
      "grad_norm": 0.43276670575141907,
      "learning_rate": 5.572133117825024e-07,
      "loss": 0.6936,
      "step": 21441
    },
    {
      "epoch": 0.8886402254548469,
      "grad_norm": 0.4184841215610504,
      "learning_rate": 5.57006092254134e-07,
      "loss": 0.6854,
      "step": 21442
    },
    {
      "epoch": 0.8886816693605205,
      "grad_norm": 0.4263876974582672,
      "learning_rate": 5.567988727257657e-07,
      "loss": 0.6948,
      "step": 21443
    },
    {
      "epoch": 0.8887231132661942,
      "grad_norm": 0.40944844484329224,
      "learning_rate": 5.565916531973973e-07,
      "loss": 0.6592,
      "step": 21444
    },
    {
      "epoch": 0.8887645571718679,
      "grad_norm": 0.39088836312294006,
      "learning_rate": 5.56384433669029e-07,
      "loss": 0.6974,
      "step": 21445
    },
    {
      "epoch": 0.8888060010775416,
      "grad_norm": 0.42904531955718994,
      "learning_rate": 5.561772141406606e-07,
      "loss": 0.6456,
      "step": 21446
    },
    {
      "epoch": 0.8888474449832152,
      "grad_norm": 0.4065399169921875,
      "learning_rate": 5.559699946122923e-07,
      "loss": 0.6094,
      "step": 21447
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.46251675486564636,
      "learning_rate": 5.557627750839239e-07,
      "loss": 0.6592,
      "step": 21448
    },
    {
      "epoch": 0.8889303327945626,
      "grad_norm": 0.4851667284965515,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.6625,
      "step": 21449
    },
    {
      "epoch": 0.8889717767002362,
      "grad_norm": 0.4046773314476013,
      "learning_rate": 5.553483360271872e-07,
      "loss": 0.6942,
      "step": 21450
    },
    {
      "epoch": 0.8890132206059099,
      "grad_norm": 0.38640618324279785,
      "learning_rate": 5.551411164988188e-07,
      "loss": 0.6624,
      "step": 21451
    },
    {
      "epoch": 0.8890546645115835,
      "grad_norm": 0.4353013336658478,
      "learning_rate": 5.549338969704505e-07,
      "loss": 0.6771,
      "step": 21452
    },
    {
      "epoch": 0.8890961084172573,
      "grad_norm": 0.410611629486084,
      "learning_rate": 5.547266774420821e-07,
      "loss": 0.657,
      "step": 21453
    },
    {
      "epoch": 0.8891375523229309,
      "grad_norm": 0.3647153377532959,
      "learning_rate": 5.545194579137138e-07,
      "loss": 0.5778,
      "step": 21454
    },
    {
      "epoch": 0.8891789962286046,
      "grad_norm": 0.44402986764907837,
      "learning_rate": 5.543122383853454e-07,
      "loss": 0.7234,
      "step": 21455
    },
    {
      "epoch": 0.8892204401342783,
      "grad_norm": 0.470130056142807,
      "learning_rate": 5.541050188569771e-07,
      "loss": 0.6587,
      "step": 21456
    },
    {
      "epoch": 0.8892618840399519,
      "grad_norm": 0.4010375440120697,
      "learning_rate": 5.538977993286087e-07,
      "loss": 0.6627,
      "step": 21457
    },
    {
      "epoch": 0.8893033279456256,
      "grad_norm": 0.4176090359687805,
      "learning_rate": 5.536905798002404e-07,
      "loss": 0.6447,
      "step": 21458
    },
    {
      "epoch": 0.8893447718512992,
      "grad_norm": 0.4478818476200104,
      "learning_rate": 5.53483360271872e-07,
      "loss": 0.651,
      "step": 21459
    },
    {
      "epoch": 0.889386215756973,
      "grad_norm": 0.415207177400589,
      "learning_rate": 5.532761407435037e-07,
      "loss": 0.6689,
      "step": 21460
    },
    {
      "epoch": 0.8894276596626466,
      "grad_norm": 0.48885300755500793,
      "learning_rate": 5.530689212151353e-07,
      "loss": 0.7124,
      "step": 21461
    },
    {
      "epoch": 0.8894691035683203,
      "grad_norm": 0.38643890619277954,
      "learning_rate": 5.52861701686767e-07,
      "loss": 0.6476,
      "step": 21462
    },
    {
      "epoch": 0.8895105474739939,
      "grad_norm": 0.3775377869606018,
      "learning_rate": 5.526544821583986e-07,
      "loss": 0.6235,
      "step": 21463
    },
    {
      "epoch": 0.8895519913796677,
      "grad_norm": 0.41830676794052124,
      "learning_rate": 5.524472626300303e-07,
      "loss": 0.6716,
      "step": 21464
    },
    {
      "epoch": 0.8895934352853413,
      "grad_norm": 0.413295716047287,
      "learning_rate": 5.522400431016619e-07,
      "loss": 0.7122,
      "step": 21465
    },
    {
      "epoch": 0.8896348791910149,
      "grad_norm": 0.4661704897880554,
      "learning_rate": 5.520328235732936e-07,
      "loss": 0.7109,
      "step": 21466
    },
    {
      "epoch": 0.8896763230966886,
      "grad_norm": 0.4171830117702484,
      "learning_rate": 5.518256040449252e-07,
      "loss": 0.7448,
      "step": 21467
    },
    {
      "epoch": 0.8897177670023623,
      "grad_norm": 0.4147372841835022,
      "learning_rate": 5.516183845165568e-07,
      "loss": 0.6704,
      "step": 21468
    },
    {
      "epoch": 0.889759210908036,
      "grad_norm": 0.41747069358825684,
      "learning_rate": 5.514111649881885e-07,
      "loss": 0.6707,
      "step": 21469
    },
    {
      "epoch": 0.8898006548137096,
      "grad_norm": 0.4109131395816803,
      "learning_rate": 5.512039454598201e-07,
      "loss": 0.71,
      "step": 21470
    },
    {
      "epoch": 0.8898420987193834,
      "grad_norm": 0.4317873418331146,
      "learning_rate": 5.509967259314518e-07,
      "loss": 0.6814,
      "step": 21471
    },
    {
      "epoch": 0.889883542625057,
      "grad_norm": 0.433786004781723,
      "learning_rate": 5.507895064030834e-07,
      "loss": 0.6746,
      "step": 21472
    },
    {
      "epoch": 0.8899249865307307,
      "grad_norm": 0.42062774300575256,
      "learning_rate": 5.505822868747151e-07,
      "loss": 0.6777,
      "step": 21473
    },
    {
      "epoch": 0.8899664304364043,
      "grad_norm": 0.40893206000328064,
      "learning_rate": 5.503750673463467e-07,
      "loss": 0.6804,
      "step": 21474
    },
    {
      "epoch": 0.890007874342078,
      "grad_norm": 0.412617027759552,
      "learning_rate": 5.501678478179784e-07,
      "loss": 0.646,
      "step": 21475
    },
    {
      "epoch": 0.8900493182477517,
      "grad_norm": 0.41835370659828186,
      "learning_rate": 5.4996062828961e-07,
      "loss": 0.6543,
      "step": 21476
    },
    {
      "epoch": 0.8900907621534253,
      "grad_norm": 0.40206417441368103,
      "learning_rate": 5.497534087612417e-07,
      "loss": 0.6427,
      "step": 21477
    },
    {
      "epoch": 0.890132206059099,
      "grad_norm": 0.38961368799209595,
      "learning_rate": 5.495461892328733e-07,
      "loss": 0.6504,
      "step": 21478
    },
    {
      "epoch": 0.8901736499647727,
      "grad_norm": 0.39861997961997986,
      "learning_rate": 5.49338969704505e-07,
      "loss": 0.655,
      "step": 21479
    },
    {
      "epoch": 0.8902150938704464,
      "grad_norm": 0.3833748996257782,
      "learning_rate": 5.491317501761366e-07,
      "loss": 0.6387,
      "step": 21480
    },
    {
      "epoch": 0.89025653777612,
      "grad_norm": 0.42654335498809814,
      "learning_rate": 5.489245306477683e-07,
      "loss": 0.6876,
      "step": 21481
    },
    {
      "epoch": 0.8902979816817936,
      "grad_norm": 0.4320532977581024,
      "learning_rate": 5.487173111193999e-07,
      "loss": 0.7001,
      "step": 21482
    },
    {
      "epoch": 0.8903394255874674,
      "grad_norm": 0.41375985741615295,
      "learning_rate": 5.485100915910316e-07,
      "loss": 0.6874,
      "step": 21483
    },
    {
      "epoch": 0.890380869493141,
      "grad_norm": 0.39052531123161316,
      "learning_rate": 5.483028720626632e-07,
      "loss": 0.6611,
      "step": 21484
    },
    {
      "epoch": 0.8904223133988147,
      "grad_norm": 0.4876834750175476,
      "learning_rate": 5.480956525342948e-07,
      "loss": 0.7561,
      "step": 21485
    },
    {
      "epoch": 0.8904637573044883,
      "grad_norm": 0.4162476658821106,
      "learning_rate": 5.478884330059265e-07,
      "loss": 0.6584,
      "step": 21486
    },
    {
      "epoch": 0.8905052012101621,
      "grad_norm": 0.4234643876552582,
      "learning_rate": 5.476812134775581e-07,
      "loss": 0.6205,
      "step": 21487
    },
    {
      "epoch": 0.8905466451158357,
      "grad_norm": 0.39526426792144775,
      "learning_rate": 5.474739939491898e-07,
      "loss": 0.666,
      "step": 21488
    },
    {
      "epoch": 0.8905880890215094,
      "grad_norm": 0.3973953127861023,
      "learning_rate": 5.472667744208214e-07,
      "loss": 0.6445,
      "step": 21489
    },
    {
      "epoch": 0.890629532927183,
      "grad_norm": 0.4397544264793396,
      "learning_rate": 5.470595548924531e-07,
      "loss": 0.7252,
      "step": 21490
    },
    {
      "epoch": 0.8906709768328567,
      "grad_norm": 0.4076325595378876,
      "learning_rate": 5.468523353640847e-07,
      "loss": 0.7139,
      "step": 21491
    },
    {
      "epoch": 0.8907124207385304,
      "grad_norm": 0.4012235701084137,
      "learning_rate": 5.466451158357164e-07,
      "loss": 0.6538,
      "step": 21492
    },
    {
      "epoch": 0.890753864644204,
      "grad_norm": 0.37476491928100586,
      "learning_rate": 5.46437896307348e-07,
      "loss": 0.6281,
      "step": 21493
    },
    {
      "epoch": 0.8907953085498778,
      "grad_norm": 0.4615389108657837,
      "learning_rate": 5.462306767789797e-07,
      "loss": 0.7285,
      "step": 21494
    },
    {
      "epoch": 0.8908367524555514,
      "grad_norm": 0.4027417004108429,
      "learning_rate": 5.460234572506113e-07,
      "loss": 0.661,
      "step": 21495
    },
    {
      "epoch": 0.8908781963612251,
      "grad_norm": 0.4122719168663025,
      "learning_rate": 5.45816237722243e-07,
      "loss": 0.6604,
      "step": 21496
    },
    {
      "epoch": 0.8909196402668987,
      "grad_norm": 0.41437840461730957,
      "learning_rate": 5.456090181938746e-07,
      "loss": 0.6523,
      "step": 21497
    },
    {
      "epoch": 0.8909610841725725,
      "grad_norm": 0.3941861689090729,
      "learning_rate": 5.454017986655063e-07,
      "loss": 0.65,
      "step": 21498
    },
    {
      "epoch": 0.8910025280782461,
      "grad_norm": 0.42280423641204834,
      "learning_rate": 5.451945791371379e-07,
      "loss": 0.6383,
      "step": 21499
    },
    {
      "epoch": 0.8910439719839197,
      "grad_norm": 0.5476798415184021,
      "learning_rate": 5.449873596087696e-07,
      "loss": 0.6686,
      "step": 21500
    },
    {
      "epoch": 0.8910854158895934,
      "grad_norm": 0.4216171205043793,
      "learning_rate": 5.447801400804012e-07,
      "loss": 0.676,
      "step": 21501
    },
    {
      "epoch": 0.8911268597952671,
      "grad_norm": 0.39519500732421875,
      "learning_rate": 5.445729205520329e-07,
      "loss": 0.6572,
      "step": 21502
    },
    {
      "epoch": 0.8911683037009408,
      "grad_norm": 0.424181193113327,
      "learning_rate": 5.443657010236645e-07,
      "loss": 0.7041,
      "step": 21503
    },
    {
      "epoch": 0.8912097476066144,
      "grad_norm": 0.4139166474342346,
      "learning_rate": 5.441584814952961e-07,
      "loss": 0.6399,
      "step": 21504
    },
    {
      "epoch": 0.8912511915122882,
      "grad_norm": 0.40471363067626953,
      "learning_rate": 5.439512619669278e-07,
      "loss": 0.6039,
      "step": 21505
    },
    {
      "epoch": 0.8912926354179618,
      "grad_norm": 0.3936883509159088,
      "learning_rate": 5.437440424385594e-07,
      "loss": 0.6355,
      "step": 21506
    },
    {
      "epoch": 0.8913340793236355,
      "grad_norm": 0.400671124458313,
      "learning_rate": 5.435368229101911e-07,
      "loss": 0.6658,
      "step": 21507
    },
    {
      "epoch": 0.8913755232293091,
      "grad_norm": 0.41553181409835815,
      "learning_rate": 5.433296033818227e-07,
      "loss": 0.6836,
      "step": 21508
    },
    {
      "epoch": 0.8914169671349828,
      "grad_norm": 0.3719368875026703,
      "learning_rate": 5.431223838534544e-07,
      "loss": 0.649,
      "step": 21509
    },
    {
      "epoch": 0.8914584110406565,
      "grad_norm": 0.3963344991207123,
      "learning_rate": 5.42915164325086e-07,
      "loss": 0.6299,
      "step": 21510
    },
    {
      "epoch": 0.8914998549463301,
      "grad_norm": 0.3754962682723999,
      "learning_rate": 5.427079447967177e-07,
      "loss": 0.6318,
      "step": 21511
    },
    {
      "epoch": 0.8915412988520038,
      "grad_norm": 0.4597218632698059,
      "learning_rate": 5.425007252683493e-07,
      "loss": 0.7097,
      "step": 21512
    },
    {
      "epoch": 0.8915827427576775,
      "grad_norm": 0.4156400263309479,
      "learning_rate": 5.42293505739981e-07,
      "loss": 0.6748,
      "step": 21513
    },
    {
      "epoch": 0.8916241866633512,
      "grad_norm": 0.3961472511291504,
      "learning_rate": 5.420862862116126e-07,
      "loss": 0.658,
      "step": 21514
    },
    {
      "epoch": 0.8916656305690248,
      "grad_norm": 0.44837409257888794,
      "learning_rate": 5.418790666832443e-07,
      "loss": 0.6279,
      "step": 21515
    },
    {
      "epoch": 0.8917070744746985,
      "grad_norm": 0.4305555522441864,
      "learning_rate": 5.416718471548759e-07,
      "loss": 0.6512,
      "step": 21516
    },
    {
      "epoch": 0.8917485183803722,
      "grad_norm": 0.4055415689945221,
      "learning_rate": 5.414646276265076e-07,
      "loss": 0.6332,
      "step": 21517
    },
    {
      "epoch": 0.8917899622860458,
      "grad_norm": 0.43324723839759827,
      "learning_rate": 5.412574080981392e-07,
      "loss": 0.6665,
      "step": 21518
    },
    {
      "epoch": 0.8918314061917195,
      "grad_norm": 0.43128296732902527,
      "learning_rate": 5.410501885697709e-07,
      "loss": 0.6218,
      "step": 21519
    },
    {
      "epoch": 0.8918728500973931,
      "grad_norm": 0.3806942105293274,
      "learning_rate": 5.408429690414025e-07,
      "loss": 0.6404,
      "step": 21520
    },
    {
      "epoch": 0.8919142940030669,
      "grad_norm": 0.45334523916244507,
      "learning_rate": 5.406357495130342e-07,
      "loss": 0.6849,
      "step": 21521
    },
    {
      "epoch": 0.8919557379087405,
      "grad_norm": 0.41324469447135925,
      "learning_rate": 5.404285299846658e-07,
      "loss": 0.6729,
      "step": 21522
    },
    {
      "epoch": 0.8919971818144142,
      "grad_norm": 0.4556848406791687,
      "learning_rate": 5.402213104562974e-07,
      "loss": 0.6873,
      "step": 21523
    },
    {
      "epoch": 0.8920386257200879,
      "grad_norm": 0.422860711812973,
      "learning_rate": 5.400140909279291e-07,
      "loss": 0.6829,
      "step": 21524
    },
    {
      "epoch": 0.8920800696257616,
      "grad_norm": 0.44406893849372864,
      "learning_rate": 5.398068713995607e-07,
      "loss": 0.6655,
      "step": 21525
    },
    {
      "epoch": 0.8921215135314352,
      "grad_norm": 0.43663206696510315,
      "learning_rate": 5.395996518711924e-07,
      "loss": 0.6688,
      "step": 21526
    },
    {
      "epoch": 0.8921629574371088,
      "grad_norm": 0.42670127749443054,
      "learning_rate": 5.39392432342824e-07,
      "loss": 0.6691,
      "step": 21527
    },
    {
      "epoch": 0.8922044013427826,
      "grad_norm": 0.4078049659729004,
      "learning_rate": 5.391852128144557e-07,
      "loss": 0.6648,
      "step": 21528
    },
    {
      "epoch": 0.8922458452484562,
      "grad_norm": 0.4168897271156311,
      "learning_rate": 5.389779932860873e-07,
      "loss": 0.6283,
      "step": 21529
    },
    {
      "epoch": 0.8922872891541299,
      "grad_norm": 0.4247773587703705,
      "learning_rate": 5.38770773757719e-07,
      "loss": 0.6675,
      "step": 21530
    },
    {
      "epoch": 0.8923287330598035,
      "grad_norm": 0.41870251297950745,
      "learning_rate": 5.385635542293506e-07,
      "loss": 0.6583,
      "step": 21531
    },
    {
      "epoch": 0.8923701769654773,
      "grad_norm": 0.4011913537979126,
      "learning_rate": 5.383563347009823e-07,
      "loss": 0.6646,
      "step": 21532
    },
    {
      "epoch": 0.8924116208711509,
      "grad_norm": 0.44034889340400696,
      "learning_rate": 5.381491151726139e-07,
      "loss": 0.7085,
      "step": 21533
    },
    {
      "epoch": 0.8924530647768246,
      "grad_norm": 0.41620439291000366,
      "learning_rate": 5.379418956442455e-07,
      "loss": 0.6774,
      "step": 21534
    },
    {
      "epoch": 0.8924945086824982,
      "grad_norm": 0.42178142070770264,
      "learning_rate": 5.377346761158772e-07,
      "loss": 0.6765,
      "step": 21535
    },
    {
      "epoch": 0.8925359525881719,
      "grad_norm": 0.4151946008205414,
      "learning_rate": 5.375274565875088e-07,
      "loss": 0.6395,
      "step": 21536
    },
    {
      "epoch": 0.8925773964938456,
      "grad_norm": 0.42614197731018066,
      "learning_rate": 5.373202370591405e-07,
      "loss": 0.6404,
      "step": 21537
    },
    {
      "epoch": 0.8926188403995192,
      "grad_norm": 0.4456254839897156,
      "learning_rate": 5.371130175307721e-07,
      "loss": 0.6719,
      "step": 21538
    },
    {
      "epoch": 0.892660284305193,
      "grad_norm": 0.5199732184410095,
      "learning_rate": 5.369057980024038e-07,
      "loss": 0.7312,
      "step": 21539
    },
    {
      "epoch": 0.8927017282108666,
      "grad_norm": 0.40631303191185,
      "learning_rate": 5.366985784740354e-07,
      "loss": 0.6605,
      "step": 21540
    },
    {
      "epoch": 0.8927431721165403,
      "grad_norm": 0.4384613037109375,
      "learning_rate": 5.364913589456671e-07,
      "loss": 0.686,
      "step": 21541
    },
    {
      "epoch": 0.8927846160222139,
      "grad_norm": 0.40261247754096985,
      "learning_rate": 5.362841394172987e-07,
      "loss": 0.6603,
      "step": 21542
    },
    {
      "epoch": 0.8928260599278875,
      "grad_norm": 0.4232719838619232,
      "learning_rate": 5.360769198889304e-07,
      "loss": 0.7292,
      "step": 21543
    },
    {
      "epoch": 0.8928675038335613,
      "grad_norm": 0.40407630801200867,
      "learning_rate": 5.35869700360562e-07,
      "loss": 0.6035,
      "step": 21544
    },
    {
      "epoch": 0.8929089477392349,
      "grad_norm": 0.4483436942100525,
      "learning_rate": 5.356624808321937e-07,
      "loss": 0.6797,
      "step": 21545
    },
    {
      "epoch": 0.8929503916449086,
      "grad_norm": 0.3982340395450592,
      "learning_rate": 5.354552613038253e-07,
      "loss": 0.6191,
      "step": 21546
    },
    {
      "epoch": 0.8929918355505823,
      "grad_norm": 0.39528003334999084,
      "learning_rate": 5.352480417754569e-07,
      "loss": 0.6501,
      "step": 21547
    },
    {
      "epoch": 0.893033279456256,
      "grad_norm": 0.5671359896659851,
      "learning_rate": 5.350408222470886e-07,
      "loss": 0.6998,
      "step": 21548
    },
    {
      "epoch": 0.8930747233619296,
      "grad_norm": 0.39818331599235535,
      "learning_rate": 5.348336027187202e-07,
      "loss": 0.6721,
      "step": 21549
    },
    {
      "epoch": 0.8931161672676033,
      "grad_norm": 0.41658154129981995,
      "learning_rate": 5.346263831903519e-07,
      "loss": 0.6896,
      "step": 21550
    },
    {
      "epoch": 0.893157611173277,
      "grad_norm": 0.4083831012248993,
      "learning_rate": 5.344191636619835e-07,
      "loss": 0.6437,
      "step": 21551
    },
    {
      "epoch": 0.8931990550789506,
      "grad_norm": 0.3779272735118866,
      "learning_rate": 5.342119441336152e-07,
      "loss": 0.6174,
      "step": 21552
    },
    {
      "epoch": 0.8932404989846243,
      "grad_norm": 0.40685805678367615,
      "learning_rate": 5.340047246052468e-07,
      "loss": 0.6675,
      "step": 21553
    },
    {
      "epoch": 0.8932819428902979,
      "grad_norm": 0.4362069070339203,
      "learning_rate": 5.337975050768785e-07,
      "loss": 0.6394,
      "step": 21554
    },
    {
      "epoch": 0.8933233867959717,
      "grad_norm": 0.4189010262489319,
      "learning_rate": 5.335902855485101e-07,
      "loss": 0.6486,
      "step": 21555
    },
    {
      "epoch": 0.8933648307016453,
      "grad_norm": 0.42723435163497925,
      "learning_rate": 5.333830660201418e-07,
      "loss": 0.7119,
      "step": 21556
    },
    {
      "epoch": 0.893406274607319,
      "grad_norm": 0.381091445684433,
      "learning_rate": 5.331758464917734e-07,
      "loss": 0.6226,
      "step": 21557
    },
    {
      "epoch": 0.8934477185129927,
      "grad_norm": 0.4040159285068512,
      "learning_rate": 5.329686269634051e-07,
      "loss": 0.6368,
      "step": 21558
    },
    {
      "epoch": 0.8934891624186664,
      "grad_norm": 0.41120877861976624,
      "learning_rate": 5.327614074350367e-07,
      "loss": 0.6699,
      "step": 21559
    },
    {
      "epoch": 0.89353060632434,
      "grad_norm": 0.4002573788166046,
      "learning_rate": 5.325541879066683e-07,
      "loss": 0.6218,
      "step": 21560
    },
    {
      "epoch": 0.8935720502300136,
      "grad_norm": 0.3949008285999298,
      "learning_rate": 5.323469683783e-07,
      "loss": 0.655,
      "step": 21561
    },
    {
      "epoch": 0.8936134941356874,
      "grad_norm": 0.4075847566127777,
      "learning_rate": 5.321397488499316e-07,
      "loss": 0.6555,
      "step": 21562
    },
    {
      "epoch": 0.893654938041361,
      "grad_norm": 0.42333805561065674,
      "learning_rate": 5.319325293215633e-07,
      "loss": 0.6787,
      "step": 21563
    },
    {
      "epoch": 0.8936963819470347,
      "grad_norm": 0.42599019408226013,
      "learning_rate": 5.317253097931949e-07,
      "loss": 0.6636,
      "step": 21564
    },
    {
      "epoch": 0.8937378258527083,
      "grad_norm": 0.40359124541282654,
      "learning_rate": 5.315180902648266e-07,
      "loss": 0.6493,
      "step": 21565
    },
    {
      "epoch": 0.8937792697583821,
      "grad_norm": 0.4086727201938629,
      "learning_rate": 5.313108707364582e-07,
      "loss": 0.6285,
      "step": 21566
    },
    {
      "epoch": 0.8938207136640557,
      "grad_norm": 0.4239732325077057,
      "learning_rate": 5.311036512080899e-07,
      "loss": 0.6826,
      "step": 21567
    },
    {
      "epoch": 0.8938621575697294,
      "grad_norm": 0.4134672284126282,
      "learning_rate": 5.308964316797215e-07,
      "loss": 0.709,
      "step": 21568
    },
    {
      "epoch": 0.893903601475403,
      "grad_norm": 0.4345559775829315,
      "learning_rate": 5.306892121513532e-07,
      "loss": 0.6874,
      "step": 21569
    },
    {
      "epoch": 0.8939450453810767,
      "grad_norm": 0.4147503077983856,
      "learning_rate": 5.304819926229848e-07,
      "loss": 0.6975,
      "step": 21570
    },
    {
      "epoch": 0.8939864892867504,
      "grad_norm": 0.42487600445747375,
      "learning_rate": 5.302747730946164e-07,
      "loss": 0.6799,
      "step": 21571
    },
    {
      "epoch": 0.894027933192424,
      "grad_norm": 0.3928578197956085,
      "learning_rate": 5.300675535662481e-07,
      "loss": 0.6724,
      "step": 21572
    },
    {
      "epoch": 0.8940693770980978,
      "grad_norm": 0.4153605103492737,
      "learning_rate": 5.298603340378797e-07,
      "loss": 0.6608,
      "step": 21573
    },
    {
      "epoch": 0.8941108210037714,
      "grad_norm": 0.46701285243034363,
      "learning_rate": 5.296531145095114e-07,
      "loss": 0.6914,
      "step": 21574
    },
    {
      "epoch": 0.8941522649094451,
      "grad_norm": 0.38295358419418335,
      "learning_rate": 5.29445894981143e-07,
      "loss": 0.7283,
      "step": 21575
    },
    {
      "epoch": 0.8941937088151187,
      "grad_norm": 0.4339483082294464,
      "learning_rate": 5.292386754527747e-07,
      "loss": 0.7286,
      "step": 21576
    },
    {
      "epoch": 0.8942351527207925,
      "grad_norm": 0.40063896775245667,
      "learning_rate": 5.290314559244063e-07,
      "loss": 0.635,
      "step": 21577
    },
    {
      "epoch": 0.8942765966264661,
      "grad_norm": 0.42637500166893005,
      "learning_rate": 5.28824236396038e-07,
      "loss": 0.6895,
      "step": 21578
    },
    {
      "epoch": 0.8943180405321397,
      "grad_norm": 0.3816472291946411,
      "learning_rate": 5.286170168676696e-07,
      "loss": 0.6515,
      "step": 21579
    },
    {
      "epoch": 0.8943594844378134,
      "grad_norm": 0.40653225779533386,
      "learning_rate": 5.284097973393013e-07,
      "loss": 0.6782,
      "step": 21580
    },
    {
      "epoch": 0.8944009283434871,
      "grad_norm": 0.4274885654449463,
      "learning_rate": 5.282025778109329e-07,
      "loss": 0.6753,
      "step": 21581
    },
    {
      "epoch": 0.8944423722491608,
      "grad_norm": 0.4499562978744507,
      "learning_rate": 5.279953582825646e-07,
      "loss": 0.6964,
      "step": 21582
    },
    {
      "epoch": 0.8944838161548344,
      "grad_norm": 0.40025967359542847,
      "learning_rate": 5.277881387541962e-07,
      "loss": 0.647,
      "step": 21583
    },
    {
      "epoch": 0.8945252600605081,
      "grad_norm": 0.41534289717674255,
      "learning_rate": 5.275809192258278e-07,
      "loss": 0.639,
      "step": 21584
    },
    {
      "epoch": 0.8945667039661818,
      "grad_norm": 0.40859201550483704,
      "learning_rate": 5.273736996974595e-07,
      "loss": 0.6899,
      "step": 21585
    },
    {
      "epoch": 0.8946081478718555,
      "grad_norm": 0.4099723994731903,
      "learning_rate": 5.271664801690911e-07,
      "loss": 0.6143,
      "step": 21586
    },
    {
      "epoch": 0.8946495917775291,
      "grad_norm": 0.4361809194087982,
      "learning_rate": 5.269592606407228e-07,
      "loss": 0.678,
      "step": 21587
    },
    {
      "epoch": 0.8946910356832027,
      "grad_norm": 0.4017447531223297,
      "learning_rate": 5.267520411123544e-07,
      "loss": 0.6299,
      "step": 21588
    },
    {
      "epoch": 0.8947324795888765,
      "grad_norm": 0.4370596706867218,
      "learning_rate": 5.265448215839861e-07,
      "loss": 0.6964,
      "step": 21589
    },
    {
      "epoch": 0.8947739234945501,
      "grad_norm": 0.39070379734039307,
      "learning_rate": 5.263376020556177e-07,
      "loss": 0.6411,
      "step": 21590
    },
    {
      "epoch": 0.8948153674002238,
      "grad_norm": 0.4805527925491333,
      "learning_rate": 5.261303825272494e-07,
      "loss": 0.7256,
      "step": 21591
    },
    {
      "epoch": 0.8948568113058974,
      "grad_norm": 0.436856746673584,
      "learning_rate": 5.25923162998881e-07,
      "loss": 0.7449,
      "step": 21592
    },
    {
      "epoch": 0.8948982552115712,
      "grad_norm": 0.4093371331691742,
      "learning_rate": 5.257159434705127e-07,
      "loss": 0.6842,
      "step": 21593
    },
    {
      "epoch": 0.8949396991172448,
      "grad_norm": 0.3973616063594818,
      "learning_rate": 5.255087239421443e-07,
      "loss": 0.7096,
      "step": 21594
    },
    {
      "epoch": 0.8949811430229185,
      "grad_norm": 0.4307800829410553,
      "learning_rate": 5.25301504413776e-07,
      "loss": 0.6326,
      "step": 21595
    },
    {
      "epoch": 0.8950225869285922,
      "grad_norm": 0.3821743428707123,
      "learning_rate": 5.250942848854076e-07,
      "loss": 0.6288,
      "step": 21596
    },
    {
      "epoch": 0.8950640308342658,
      "grad_norm": 0.4651239216327667,
      "learning_rate": 5.248870653570392e-07,
      "loss": 0.657,
      "step": 21597
    },
    {
      "epoch": 0.8951054747399395,
      "grad_norm": 0.43149086833000183,
      "learning_rate": 5.246798458286709e-07,
      "loss": 0.6558,
      "step": 21598
    },
    {
      "epoch": 0.8951469186456131,
      "grad_norm": 0.38808968663215637,
      "learning_rate": 5.244726263003025e-07,
      "loss": 0.6079,
      "step": 21599
    },
    {
      "epoch": 0.8951883625512869,
      "grad_norm": 0.4123295545578003,
      "learning_rate": 5.242654067719342e-07,
      "loss": 0.6024,
      "step": 21600
    },
    {
      "epoch": 0.8952298064569605,
      "grad_norm": 0.40237629413604736,
      "learning_rate": 5.240581872435658e-07,
      "loss": 0.6516,
      "step": 21601
    },
    {
      "epoch": 0.8952712503626342,
      "grad_norm": 0.44824451208114624,
      "learning_rate": 5.238509677151975e-07,
      "loss": 0.6675,
      "step": 21602
    },
    {
      "epoch": 0.8953126942683078,
      "grad_norm": 0.3879373371601105,
      "learning_rate": 5.236437481868291e-07,
      "loss": 0.6165,
      "step": 21603
    },
    {
      "epoch": 0.8953541381739815,
      "grad_norm": 0.4469955563545227,
      "learning_rate": 5.234365286584608e-07,
      "loss": 0.6924,
      "step": 21604
    },
    {
      "epoch": 0.8953955820796552,
      "grad_norm": 0.45527762174606323,
      "learning_rate": 5.232293091300924e-07,
      "loss": 0.701,
      "step": 21605
    },
    {
      "epoch": 0.8954370259853288,
      "grad_norm": 0.3921760022640228,
      "learning_rate": 5.230220896017241e-07,
      "loss": 0.6193,
      "step": 21606
    },
    {
      "epoch": 0.8954784698910025,
      "grad_norm": 0.4063795804977417,
      "learning_rate": 5.228148700733557e-07,
      "loss": 0.6562,
      "step": 21607
    },
    {
      "epoch": 0.8955199137966762,
      "grad_norm": 0.4430139362812042,
      "learning_rate": 5.226076505449873e-07,
      "loss": 0.6432,
      "step": 21608
    },
    {
      "epoch": 0.8955613577023499,
      "grad_norm": 0.41347554326057434,
      "learning_rate": 5.22400431016619e-07,
      "loss": 0.6643,
      "step": 21609
    },
    {
      "epoch": 0.8956028016080235,
      "grad_norm": 0.41163283586502075,
      "learning_rate": 5.221932114882506e-07,
      "loss": 0.6726,
      "step": 21610
    },
    {
      "epoch": 0.8956442455136973,
      "grad_norm": 0.4329853951931,
      "learning_rate": 5.219859919598823e-07,
      "loss": 0.6995,
      "step": 21611
    },
    {
      "epoch": 0.8956856894193709,
      "grad_norm": 0.40636536478996277,
      "learning_rate": 5.217787724315139e-07,
      "loss": 0.6442,
      "step": 21612
    },
    {
      "epoch": 0.8957271333250445,
      "grad_norm": 0.4101714789867401,
      "learning_rate": 5.215715529031456e-07,
      "loss": 0.6769,
      "step": 21613
    },
    {
      "epoch": 0.8957685772307182,
      "grad_norm": 0.46497735381126404,
      "learning_rate": 5.213643333747772e-07,
      "loss": 0.687,
      "step": 21614
    },
    {
      "epoch": 0.8958100211363919,
      "grad_norm": 0.41650301218032837,
      "learning_rate": 5.21157113846409e-07,
      "loss": 0.6573,
      "step": 21615
    },
    {
      "epoch": 0.8958514650420656,
      "grad_norm": 0.41922280192375183,
      "learning_rate": 5.209498943180405e-07,
      "loss": 0.6819,
      "step": 21616
    },
    {
      "epoch": 0.8958929089477392,
      "grad_norm": 0.41188955307006836,
      "learning_rate": 5.207426747896722e-07,
      "loss": 0.6069,
      "step": 21617
    },
    {
      "epoch": 0.8959343528534129,
      "grad_norm": 0.3900759518146515,
      "learning_rate": 5.205354552613038e-07,
      "loss": 0.6423,
      "step": 21618
    },
    {
      "epoch": 0.8959757967590866,
      "grad_norm": 0.41566193103790283,
      "learning_rate": 5.203282357329355e-07,
      "loss": 0.647,
      "step": 21619
    },
    {
      "epoch": 0.8960172406647603,
      "grad_norm": 0.44791263341903687,
      "learning_rate": 5.201210162045671e-07,
      "loss": 0.6772,
      "step": 21620
    },
    {
      "epoch": 0.8960586845704339,
      "grad_norm": 0.3969278931617737,
      "learning_rate": 5.199137966761987e-07,
      "loss": 0.6354,
      "step": 21621
    },
    {
      "epoch": 0.8961001284761075,
      "grad_norm": 0.41287511587142944,
      "learning_rate": 5.197065771478304e-07,
      "loss": 0.687,
      "step": 21622
    },
    {
      "epoch": 0.8961415723817813,
      "grad_norm": 0.4647110104560852,
      "learning_rate": 5.19499357619462e-07,
      "loss": 0.728,
      "step": 21623
    },
    {
      "epoch": 0.8961830162874549,
      "grad_norm": 0.4311895966529846,
      "learning_rate": 5.192921380910937e-07,
      "loss": 0.6873,
      "step": 21624
    },
    {
      "epoch": 0.8962244601931286,
      "grad_norm": 0.4035327434539795,
      "learning_rate": 5.190849185627253e-07,
      "loss": 0.6659,
      "step": 21625
    },
    {
      "epoch": 0.8962659040988022,
      "grad_norm": 0.4065888524055481,
      "learning_rate": 5.18877699034357e-07,
      "loss": 0.6176,
      "step": 21626
    },
    {
      "epoch": 0.896307348004476,
      "grad_norm": 0.4296950101852417,
      "learning_rate": 5.186704795059886e-07,
      "loss": 0.668,
      "step": 21627
    },
    {
      "epoch": 0.8963487919101496,
      "grad_norm": 0.48781752586364746,
      "learning_rate": 5.184632599776203e-07,
      "loss": 0.7815,
      "step": 21628
    },
    {
      "epoch": 0.8963902358158233,
      "grad_norm": 0.368587851524353,
      "learning_rate": 5.182560404492519e-07,
      "loss": 0.6306,
      "step": 21629
    },
    {
      "epoch": 0.896431679721497,
      "grad_norm": 0.3955225944519043,
      "learning_rate": 5.180488209208836e-07,
      "loss": 0.6531,
      "step": 21630
    },
    {
      "epoch": 0.8964731236271706,
      "grad_norm": 0.39316800236701965,
      "learning_rate": 5.178416013925152e-07,
      "loss": 0.6356,
      "step": 21631
    },
    {
      "epoch": 0.8965145675328443,
      "grad_norm": 0.4311050474643707,
      "learning_rate": 5.17634381864147e-07,
      "loss": 0.667,
      "step": 21632
    },
    {
      "epoch": 0.8965560114385179,
      "grad_norm": 0.4411413073539734,
      "learning_rate": 5.174271623357785e-07,
      "loss": 0.6326,
      "step": 21633
    },
    {
      "epoch": 0.8965974553441917,
      "grad_norm": 0.42417746782302856,
      "learning_rate": 5.172199428074101e-07,
      "loss": 0.7006,
      "step": 21634
    },
    {
      "epoch": 0.8966388992498653,
      "grad_norm": 0.3984362781047821,
      "learning_rate": 5.170127232790418e-07,
      "loss": 0.6235,
      "step": 21635
    },
    {
      "epoch": 0.896680343155539,
      "grad_norm": 0.38406893610954285,
      "learning_rate": 5.168055037506734e-07,
      "loss": 0.6444,
      "step": 21636
    },
    {
      "epoch": 0.8967217870612126,
      "grad_norm": 0.423697829246521,
      "learning_rate": 5.165982842223051e-07,
      "loss": 0.651,
      "step": 21637
    },
    {
      "epoch": 0.8967632309668864,
      "grad_norm": 0.4383939504623413,
      "learning_rate": 5.163910646939367e-07,
      "loss": 0.6458,
      "step": 21638
    },
    {
      "epoch": 0.89680467487256,
      "grad_norm": 0.41231778264045715,
      "learning_rate": 5.161838451655685e-07,
      "loss": 0.6975,
      "step": 21639
    },
    {
      "epoch": 0.8968461187782336,
      "grad_norm": 0.41430342197418213,
      "learning_rate": 5.159766256372e-07,
      "loss": 0.6658,
      "step": 21640
    },
    {
      "epoch": 0.8968875626839073,
      "grad_norm": 0.42921629548072815,
      "learning_rate": 5.157694061088318e-07,
      "loss": 0.6631,
      "step": 21641
    },
    {
      "epoch": 0.896929006589581,
      "grad_norm": 0.4215526878833771,
      "learning_rate": 5.155621865804633e-07,
      "loss": 0.6368,
      "step": 21642
    },
    {
      "epoch": 0.8969704504952547,
      "grad_norm": 0.5884553790092468,
      "learning_rate": 5.15354967052095e-07,
      "loss": 0.6708,
      "step": 21643
    },
    {
      "epoch": 0.8970118944009283,
      "grad_norm": 0.39017221331596375,
      "learning_rate": 5.151477475237266e-07,
      "loss": 0.6597,
      "step": 21644
    },
    {
      "epoch": 0.897053338306602,
      "grad_norm": 0.43384143710136414,
      "learning_rate": 5.149405279953582e-07,
      "loss": 0.6736,
      "step": 21645
    },
    {
      "epoch": 0.8970947822122757,
      "grad_norm": 0.4396486282348633,
      "learning_rate": 5.1473330846699e-07,
      "loss": 0.6499,
      "step": 21646
    },
    {
      "epoch": 0.8971362261179494,
      "grad_norm": 0.4012719988822937,
      "learning_rate": 5.145260889386215e-07,
      "loss": 0.6333,
      "step": 21647
    },
    {
      "epoch": 0.897177670023623,
      "grad_norm": 0.381801962852478,
      "learning_rate": 5.143188694102533e-07,
      "loss": 0.6539,
      "step": 21648
    },
    {
      "epoch": 0.8972191139292967,
      "grad_norm": 0.41323745250701904,
      "learning_rate": 5.141116498818848e-07,
      "loss": 0.6008,
      "step": 21649
    },
    {
      "epoch": 0.8972605578349704,
      "grad_norm": 0.4583244323730469,
      "learning_rate": 5.139044303535166e-07,
      "loss": 0.7344,
      "step": 21650
    },
    {
      "epoch": 0.897302001740644,
      "grad_norm": 0.41953304409980774,
      "learning_rate": 5.136972108251481e-07,
      "loss": 0.6482,
      "step": 21651
    },
    {
      "epoch": 0.8973434456463177,
      "grad_norm": 0.42950865626335144,
      "learning_rate": 5.134899912967799e-07,
      "loss": 0.6195,
      "step": 21652
    },
    {
      "epoch": 0.8973848895519914,
      "grad_norm": 0.4639635384082794,
      "learning_rate": 5.132827717684114e-07,
      "loss": 0.6992,
      "step": 21653
    },
    {
      "epoch": 0.8974263334576651,
      "grad_norm": 0.3812353014945984,
      "learning_rate": 5.130755522400432e-07,
      "loss": 0.6555,
      "step": 21654
    },
    {
      "epoch": 0.8974677773633387,
      "grad_norm": 0.4112999141216278,
      "learning_rate": 5.128683327116747e-07,
      "loss": 0.6699,
      "step": 21655
    },
    {
      "epoch": 0.8975092212690124,
      "grad_norm": 0.3868524432182312,
      "learning_rate": 5.126611131833065e-07,
      "loss": 0.6378,
      "step": 21656
    },
    {
      "epoch": 0.8975506651746861,
      "grad_norm": 0.43214574456214905,
      "learning_rate": 5.12453893654938e-07,
      "loss": 0.7219,
      "step": 21657
    },
    {
      "epoch": 0.8975921090803597,
      "grad_norm": 0.4264596998691559,
      "learning_rate": 5.122466741265696e-07,
      "loss": 0.6356,
      "step": 21658
    },
    {
      "epoch": 0.8976335529860334,
      "grad_norm": 0.3986511826515198,
      "learning_rate": 5.120394545982014e-07,
      "loss": 0.625,
      "step": 21659
    },
    {
      "epoch": 0.897674996891707,
      "grad_norm": 0.45593035221099854,
      "learning_rate": 5.11832235069833e-07,
      "loss": 0.6582,
      "step": 21660
    },
    {
      "epoch": 0.8977164407973808,
      "grad_norm": 0.4224576950073242,
      "learning_rate": 5.116250155414647e-07,
      "loss": 0.73,
      "step": 21661
    },
    {
      "epoch": 0.8977578847030544,
      "grad_norm": 0.4360947608947754,
      "learning_rate": 5.114177960130962e-07,
      "loss": 0.7012,
      "step": 21662
    },
    {
      "epoch": 0.8977993286087281,
      "grad_norm": 0.4460400640964508,
      "learning_rate": 5.11210576484728e-07,
      "loss": 0.6938,
      "step": 21663
    },
    {
      "epoch": 0.8978407725144018,
      "grad_norm": 0.4628468155860901,
      "learning_rate": 5.110033569563596e-07,
      "loss": 0.6288,
      "step": 21664
    },
    {
      "epoch": 0.8978822164200754,
      "grad_norm": 0.3927665650844574,
      "learning_rate": 5.107961374279913e-07,
      "loss": 0.6349,
      "step": 21665
    },
    {
      "epoch": 0.8979236603257491,
      "grad_norm": 0.40729469060897827,
      "learning_rate": 5.105889178996229e-07,
      "loss": 0.6531,
      "step": 21666
    },
    {
      "epoch": 0.8979651042314227,
      "grad_norm": 0.40527626872062683,
      "learning_rate": 5.103816983712546e-07,
      "loss": 0.6794,
      "step": 21667
    },
    {
      "epoch": 0.8980065481370965,
      "grad_norm": 0.42476168274879456,
      "learning_rate": 5.101744788428862e-07,
      "loss": 0.6943,
      "step": 21668
    },
    {
      "epoch": 0.8980479920427701,
      "grad_norm": 0.4106142222881317,
      "learning_rate": 5.099672593145179e-07,
      "loss": 0.6406,
      "step": 21669
    },
    {
      "epoch": 0.8980894359484438,
      "grad_norm": 0.4043257236480713,
      "learning_rate": 5.097600397861495e-07,
      "loss": 0.6854,
      "step": 21670
    },
    {
      "epoch": 0.8981308798541174,
      "grad_norm": 0.43675118684768677,
      "learning_rate": 5.095528202577812e-07,
      "loss": 0.644,
      "step": 21671
    },
    {
      "epoch": 0.8981723237597912,
      "grad_norm": 0.42304709553718567,
      "learning_rate": 5.093456007294128e-07,
      "loss": 0.6484,
      "step": 21672
    },
    {
      "epoch": 0.8982137676654648,
      "grad_norm": 0.38777920603752136,
      "learning_rate": 5.091383812010445e-07,
      "loss": 0.5853,
      "step": 21673
    },
    {
      "epoch": 0.8982552115711384,
      "grad_norm": 0.40734386444091797,
      "learning_rate": 5.089311616726761e-07,
      "loss": 0.667,
      "step": 21674
    },
    {
      "epoch": 0.8982966554768121,
      "grad_norm": 0.3839884102344513,
      "learning_rate": 5.087239421443077e-07,
      "loss": 0.6184,
      "step": 21675
    },
    {
      "epoch": 0.8983380993824858,
      "grad_norm": 0.4026780426502228,
      "learning_rate": 5.085167226159394e-07,
      "loss": 0.6503,
      "step": 21676
    },
    {
      "epoch": 0.8983795432881595,
      "grad_norm": 0.4248570203781128,
      "learning_rate": 5.08309503087571e-07,
      "loss": 0.6741,
      "step": 21677
    },
    {
      "epoch": 0.8984209871938331,
      "grad_norm": 0.40241941809654236,
      "learning_rate": 5.081022835592027e-07,
      "loss": 0.6475,
      "step": 21678
    },
    {
      "epoch": 0.8984624310995069,
      "grad_norm": 0.42992886900901794,
      "learning_rate": 5.078950640308343e-07,
      "loss": 0.67,
      "step": 21679
    },
    {
      "epoch": 0.8985038750051805,
      "grad_norm": 0.39304637908935547,
      "learning_rate": 5.07687844502466e-07,
      "loss": 0.6493,
      "step": 21680
    },
    {
      "epoch": 0.8985453189108542,
      "grad_norm": 0.42343226075172424,
      "learning_rate": 5.074806249740976e-07,
      "loss": 0.6458,
      "step": 21681
    },
    {
      "epoch": 0.8985867628165278,
      "grad_norm": 0.43401435017585754,
      "learning_rate": 5.072734054457293e-07,
      "loss": 0.684,
      "step": 21682
    },
    {
      "epoch": 0.8986282067222015,
      "grad_norm": 0.43685391545295715,
      "learning_rate": 5.070661859173609e-07,
      "loss": 0.6578,
      "step": 21683
    },
    {
      "epoch": 0.8986696506278752,
      "grad_norm": 0.4203037619590759,
      "learning_rate": 5.068589663889926e-07,
      "loss": 0.6521,
      "step": 21684
    },
    {
      "epoch": 0.8987110945335488,
      "grad_norm": 0.40927252173423767,
      "learning_rate": 5.066517468606242e-07,
      "loss": 0.6743,
      "step": 21685
    },
    {
      "epoch": 0.8987525384392225,
      "grad_norm": 0.4302552342414856,
      "learning_rate": 5.064445273322559e-07,
      "loss": 0.6611,
      "step": 21686
    },
    {
      "epoch": 0.8987939823448962,
      "grad_norm": 0.3948485851287842,
      "learning_rate": 5.062373078038875e-07,
      "loss": 0.6707,
      "step": 21687
    },
    {
      "epoch": 0.8988354262505699,
      "grad_norm": 0.42815446853637695,
      "learning_rate": 5.060300882755192e-07,
      "loss": 0.6884,
      "step": 21688
    },
    {
      "epoch": 0.8988768701562435,
      "grad_norm": 0.45705458521842957,
      "learning_rate": 5.058228687471508e-07,
      "loss": 0.6604,
      "step": 21689
    },
    {
      "epoch": 0.8989183140619172,
      "grad_norm": 0.38672807812690735,
      "learning_rate": 5.056156492187825e-07,
      "loss": 0.6357,
      "step": 21690
    },
    {
      "epoch": 0.8989597579675909,
      "grad_norm": 0.41418200731277466,
      "learning_rate": 5.054084296904141e-07,
      "loss": 0.729,
      "step": 21691
    },
    {
      "epoch": 0.8990012018732645,
      "grad_norm": 0.416007399559021,
      "learning_rate": 5.052012101620458e-07,
      "loss": 0.7388,
      "step": 21692
    },
    {
      "epoch": 0.8990426457789382,
      "grad_norm": 0.41595259308815,
      "learning_rate": 5.049939906336774e-07,
      "loss": 0.6658,
      "step": 21693
    },
    {
      "epoch": 0.8990840896846118,
      "grad_norm": 0.49643537402153015,
      "learning_rate": 5.04786771105309e-07,
      "loss": 0.7134,
      "step": 21694
    },
    {
      "epoch": 0.8991255335902856,
      "grad_norm": 0.4049939513206482,
      "learning_rate": 5.045795515769407e-07,
      "loss": 0.6526,
      "step": 21695
    },
    {
      "epoch": 0.8991669774959592,
      "grad_norm": 0.4014173746109009,
      "learning_rate": 5.043723320485723e-07,
      "loss": 0.6814,
      "step": 21696
    },
    {
      "epoch": 0.8992084214016329,
      "grad_norm": 0.4296542704105377,
      "learning_rate": 5.04165112520204e-07,
      "loss": 0.644,
      "step": 21697
    },
    {
      "epoch": 0.8992498653073066,
      "grad_norm": 0.4030790627002716,
      "learning_rate": 5.039578929918356e-07,
      "loss": 0.6282,
      "step": 21698
    },
    {
      "epoch": 0.8992913092129803,
      "grad_norm": 0.39832839369773865,
      "learning_rate": 5.037506734634673e-07,
      "loss": 0.6473,
      "step": 21699
    },
    {
      "epoch": 0.8993327531186539,
      "grad_norm": 0.4235720932483673,
      "learning_rate": 5.035434539350989e-07,
      "loss": 0.6575,
      "step": 21700
    },
    {
      "epoch": 0.8993741970243275,
      "grad_norm": 0.4068671762943268,
      "learning_rate": 5.033362344067306e-07,
      "loss": 0.6416,
      "step": 21701
    },
    {
      "epoch": 0.8994156409300013,
      "grad_norm": 0.39585357904434204,
      "learning_rate": 5.031290148783622e-07,
      "loss": 0.651,
      "step": 21702
    },
    {
      "epoch": 0.8994570848356749,
      "grad_norm": 0.40354934334754944,
      "learning_rate": 5.029217953499939e-07,
      "loss": 0.6775,
      "step": 21703
    },
    {
      "epoch": 0.8994985287413486,
      "grad_norm": 0.416501522064209,
      "learning_rate": 5.027145758216255e-07,
      "loss": 0.603,
      "step": 21704
    },
    {
      "epoch": 0.8995399726470222,
      "grad_norm": 0.42291387915611267,
      "learning_rate": 5.025073562932572e-07,
      "loss": 0.675,
      "step": 21705
    },
    {
      "epoch": 0.899581416552696,
      "grad_norm": 0.37700366973876953,
      "learning_rate": 5.023001367648888e-07,
      "loss": 0.6324,
      "step": 21706
    },
    {
      "epoch": 0.8996228604583696,
      "grad_norm": 0.45422127842903137,
      "learning_rate": 5.020929172365205e-07,
      "loss": 0.6661,
      "step": 21707
    },
    {
      "epoch": 0.8996643043640433,
      "grad_norm": 0.39959803223609924,
      "learning_rate": 5.018856977081521e-07,
      "loss": 0.6039,
      "step": 21708
    },
    {
      "epoch": 0.8997057482697169,
      "grad_norm": 0.4188365340232849,
      "learning_rate": 5.016784781797838e-07,
      "loss": 0.6692,
      "step": 21709
    },
    {
      "epoch": 0.8997471921753906,
      "grad_norm": 0.42786991596221924,
      "learning_rate": 5.014712586514154e-07,
      "loss": 0.6711,
      "step": 21710
    },
    {
      "epoch": 0.8997886360810643,
      "grad_norm": 0.40679505467414856,
      "learning_rate": 5.012640391230471e-07,
      "loss": 0.7031,
      "step": 21711
    },
    {
      "epoch": 0.8998300799867379,
      "grad_norm": 0.3900405466556549,
      "learning_rate": 5.010568195946787e-07,
      "loss": 0.6313,
      "step": 21712
    },
    {
      "epoch": 0.8998715238924117,
      "grad_norm": 0.3941775858402252,
      "learning_rate": 5.008496000663103e-07,
      "loss": 0.6938,
      "step": 21713
    },
    {
      "epoch": 0.8999129677980853,
      "grad_norm": 0.40731289982795715,
      "learning_rate": 5.00642380537942e-07,
      "loss": 0.6608,
      "step": 21714
    },
    {
      "epoch": 0.899954411703759,
      "grad_norm": 0.3974337875843048,
      "learning_rate": 5.004351610095736e-07,
      "loss": 0.6483,
      "step": 21715
    },
    {
      "epoch": 0.8999958556094326,
      "grad_norm": 0.42606380581855774,
      "learning_rate": 5.002279414812053e-07,
      "loss": 0.6595,
      "step": 21716
    },
    {
      "epoch": 0.9000372995151062,
      "grad_norm": 0.4259994328022003,
      "learning_rate": 5.000207219528369e-07,
      "loss": 0.6316,
      "step": 21717
    },
    {
      "epoch": 0.90007874342078,
      "grad_norm": 0.44175952672958374,
      "learning_rate": 4.998135024244686e-07,
      "loss": 0.6918,
      "step": 21718
    },
    {
      "epoch": 0.9001201873264536,
      "grad_norm": 0.3832865059375763,
      "learning_rate": 4.996062828961002e-07,
      "loss": 0.6283,
      "step": 21719
    },
    {
      "epoch": 0.9001616312321273,
      "grad_norm": 0.4191640615463257,
      "learning_rate": 4.993990633677319e-07,
      "loss": 0.7319,
      "step": 21720
    },
    {
      "epoch": 0.900203075137801,
      "grad_norm": 0.42710813879966736,
      "learning_rate": 4.991918438393635e-07,
      "loss": 0.6819,
      "step": 21721
    },
    {
      "epoch": 0.9002445190434747,
      "grad_norm": 0.4307880103588104,
      "learning_rate": 4.989846243109952e-07,
      "loss": 0.6685,
      "step": 21722
    },
    {
      "epoch": 0.9002859629491483,
      "grad_norm": 0.40458813309669495,
      "learning_rate": 4.987774047826268e-07,
      "loss": 0.6506,
      "step": 21723
    },
    {
      "epoch": 0.900327406854822,
      "grad_norm": 0.4133318066596985,
      "learning_rate": 4.985701852542585e-07,
      "loss": 0.6466,
      "step": 21724
    },
    {
      "epoch": 0.9003688507604957,
      "grad_norm": 0.4259662628173828,
      "learning_rate": 4.983629657258901e-07,
      "loss": 0.7036,
      "step": 21725
    },
    {
      "epoch": 0.9004102946661693,
      "grad_norm": 0.4297091066837311,
      "learning_rate": 4.981557461975218e-07,
      "loss": 0.6233,
      "step": 21726
    },
    {
      "epoch": 0.900451738571843,
      "grad_norm": 0.39043527841567993,
      "learning_rate": 4.979485266691534e-07,
      "loss": 0.641,
      "step": 21727
    },
    {
      "epoch": 0.9004931824775166,
      "grad_norm": 0.4170440137386322,
      "learning_rate": 4.977413071407851e-07,
      "loss": 0.6475,
      "step": 21728
    },
    {
      "epoch": 0.9005346263831904,
      "grad_norm": 0.3916419446468353,
      "learning_rate": 4.975340876124167e-07,
      "loss": 0.6041,
      "step": 21729
    },
    {
      "epoch": 0.900576070288864,
      "grad_norm": 0.43758535385131836,
      "learning_rate": 4.973268680840483e-07,
      "loss": 0.6274,
      "step": 21730
    },
    {
      "epoch": 0.9006175141945377,
      "grad_norm": 0.4131115972995758,
      "learning_rate": 4.9711964855568e-07,
      "loss": 0.699,
      "step": 21731
    },
    {
      "epoch": 0.9006589581002113,
      "grad_norm": 0.4266568422317505,
      "learning_rate": 4.969124290273116e-07,
      "loss": 0.687,
      "step": 21732
    },
    {
      "epoch": 0.9007004020058851,
      "grad_norm": 0.3950921297073364,
      "learning_rate": 4.967052094989433e-07,
      "loss": 0.6989,
      "step": 21733
    },
    {
      "epoch": 0.9007418459115587,
      "grad_norm": 0.44160592555999756,
      "learning_rate": 4.964979899705749e-07,
      "loss": 0.6877,
      "step": 21734
    },
    {
      "epoch": 0.9007832898172323,
      "grad_norm": 0.4287833273410797,
      "learning_rate": 4.962907704422066e-07,
      "loss": 0.6804,
      "step": 21735
    },
    {
      "epoch": 0.9008247337229061,
      "grad_norm": 0.3995215892791748,
      "learning_rate": 4.960835509138382e-07,
      "loss": 0.7102,
      "step": 21736
    },
    {
      "epoch": 0.9008661776285797,
      "grad_norm": 0.38380053639411926,
      "learning_rate": 4.958763313854699e-07,
      "loss": 0.6267,
      "step": 21737
    },
    {
      "epoch": 0.9009076215342534,
      "grad_norm": 0.40147364139556885,
      "learning_rate": 4.956691118571015e-07,
      "loss": 0.616,
      "step": 21738
    },
    {
      "epoch": 0.900949065439927,
      "grad_norm": 0.4564894437789917,
      "learning_rate": 4.954618923287332e-07,
      "loss": 0.6484,
      "step": 21739
    },
    {
      "epoch": 0.9009905093456008,
      "grad_norm": 0.4524424970149994,
      "learning_rate": 4.952546728003648e-07,
      "loss": 0.7368,
      "step": 21740
    },
    {
      "epoch": 0.9010319532512744,
      "grad_norm": 0.4386501908302307,
      "learning_rate": 4.950474532719965e-07,
      "loss": 0.7346,
      "step": 21741
    },
    {
      "epoch": 0.9010733971569481,
      "grad_norm": 0.3903694748878479,
      "learning_rate": 4.948402337436281e-07,
      "loss": 0.6395,
      "step": 21742
    },
    {
      "epoch": 0.9011148410626217,
      "grad_norm": 0.4003303050994873,
      "learning_rate": 4.946330142152597e-07,
      "loss": 0.6736,
      "step": 21743
    },
    {
      "epoch": 0.9011562849682954,
      "grad_norm": 0.42025068402290344,
      "learning_rate": 4.944257946868914e-07,
      "loss": 0.7606,
      "step": 21744
    },
    {
      "epoch": 0.9011977288739691,
      "grad_norm": 0.3982343375682831,
      "learning_rate": 4.94218575158523e-07,
      "loss": 0.623,
      "step": 21745
    },
    {
      "epoch": 0.9012391727796427,
      "grad_norm": 0.4200702905654907,
      "learning_rate": 4.940113556301547e-07,
      "loss": 0.6672,
      "step": 21746
    },
    {
      "epoch": 0.9012806166853164,
      "grad_norm": 0.38132521510124207,
      "learning_rate": 4.938041361017863e-07,
      "loss": 0.658,
      "step": 21747
    },
    {
      "epoch": 0.9013220605909901,
      "grad_norm": 0.4360782206058502,
      "learning_rate": 4.93596916573418e-07,
      "loss": 0.6978,
      "step": 21748
    },
    {
      "epoch": 0.9013635044966638,
      "grad_norm": 0.427347332239151,
      "learning_rate": 4.933896970450496e-07,
      "loss": 0.6709,
      "step": 21749
    },
    {
      "epoch": 0.9014049484023374,
      "grad_norm": 0.4409312605857849,
      "learning_rate": 4.931824775166813e-07,
      "loss": 0.6678,
      "step": 21750
    },
    {
      "epoch": 0.9014463923080112,
      "grad_norm": 0.4005622863769531,
      "learning_rate": 4.929752579883129e-07,
      "loss": 0.6703,
      "step": 21751
    },
    {
      "epoch": 0.9014878362136848,
      "grad_norm": 0.4347497522830963,
      "learning_rate": 4.927680384599446e-07,
      "loss": 0.697,
      "step": 21752
    },
    {
      "epoch": 0.9015292801193584,
      "grad_norm": 0.384727418422699,
      "learning_rate": 4.925608189315762e-07,
      "loss": 0.6282,
      "step": 21753
    },
    {
      "epoch": 0.9015707240250321,
      "grad_norm": 0.4598029553890228,
      "learning_rate": 4.923535994032079e-07,
      "loss": 0.6947,
      "step": 21754
    },
    {
      "epoch": 0.9016121679307058,
      "grad_norm": 0.4261048138141632,
      "learning_rate": 4.921463798748395e-07,
      "loss": 0.6271,
      "step": 21755
    },
    {
      "epoch": 0.9016536118363795,
      "grad_norm": 0.43014076352119446,
      "learning_rate": 4.919391603464711e-07,
      "loss": 0.6313,
      "step": 21756
    },
    {
      "epoch": 0.9016950557420531,
      "grad_norm": 0.42324361205101013,
      "learning_rate": 4.917319408181028e-07,
      "loss": 0.6033,
      "step": 21757
    },
    {
      "epoch": 0.9017364996477268,
      "grad_norm": 0.3989635705947876,
      "learning_rate": 4.915247212897344e-07,
      "loss": 0.6724,
      "step": 21758
    },
    {
      "epoch": 0.9017779435534005,
      "grad_norm": 0.4641168415546417,
      "learning_rate": 4.913175017613661e-07,
      "loss": 0.708,
      "step": 21759
    },
    {
      "epoch": 0.9018193874590742,
      "grad_norm": 0.47041693329811096,
      "learning_rate": 4.911102822329977e-07,
      "loss": 0.6971,
      "step": 21760
    },
    {
      "epoch": 0.9018608313647478,
      "grad_norm": 0.42518502473831177,
      "learning_rate": 4.909030627046294e-07,
      "loss": 0.6741,
      "step": 21761
    },
    {
      "epoch": 0.9019022752704214,
      "grad_norm": 0.3835727274417877,
      "learning_rate": 4.90695843176261e-07,
      "loss": 0.6494,
      "step": 21762
    },
    {
      "epoch": 0.9019437191760952,
      "grad_norm": 0.4638034701347351,
      "learning_rate": 4.904886236478927e-07,
      "loss": 0.6993,
      "step": 21763
    },
    {
      "epoch": 0.9019851630817688,
      "grad_norm": 0.4210434854030609,
      "learning_rate": 4.902814041195243e-07,
      "loss": 0.6855,
      "step": 21764
    },
    {
      "epoch": 0.9020266069874425,
      "grad_norm": 0.4069121181964874,
      "learning_rate": 4.90074184591156e-07,
      "loss": 0.6544,
      "step": 21765
    },
    {
      "epoch": 0.9020680508931161,
      "grad_norm": 0.4243365526199341,
      "learning_rate": 4.898669650627876e-07,
      "loss": 0.6738,
      "step": 21766
    },
    {
      "epoch": 0.9021094947987899,
      "grad_norm": 0.44677332043647766,
      "learning_rate": 4.896597455344193e-07,
      "loss": 0.6775,
      "step": 21767
    },
    {
      "epoch": 0.9021509387044635,
      "grad_norm": 0.3879317343235016,
      "learning_rate": 4.894525260060509e-07,
      "loss": 0.6553,
      "step": 21768
    },
    {
      "epoch": 0.9021923826101372,
      "grad_norm": 0.4478183388710022,
      "learning_rate": 4.892453064776825e-07,
      "loss": 0.679,
      "step": 21769
    },
    {
      "epoch": 0.9022338265158109,
      "grad_norm": 0.4068681001663208,
      "learning_rate": 4.890380869493142e-07,
      "loss": 0.6697,
      "step": 21770
    },
    {
      "epoch": 0.9022752704214845,
      "grad_norm": 0.4193711280822754,
      "learning_rate": 4.888308674209458e-07,
      "loss": 0.7144,
      "step": 21771
    },
    {
      "epoch": 0.9023167143271582,
      "grad_norm": 0.43674561381340027,
      "learning_rate": 4.886236478925775e-07,
      "loss": 0.6571,
      "step": 21772
    },
    {
      "epoch": 0.9023581582328318,
      "grad_norm": 0.3922615051269531,
      "learning_rate": 4.884164283642091e-07,
      "loss": 0.6565,
      "step": 21773
    },
    {
      "epoch": 0.9023996021385056,
      "grad_norm": 0.4094826281070709,
      "learning_rate": 4.882092088358408e-07,
      "loss": 0.6272,
      "step": 21774
    },
    {
      "epoch": 0.9024410460441792,
      "grad_norm": 0.4444669187068939,
      "learning_rate": 4.880019893074724e-07,
      "loss": 0.7183,
      "step": 21775
    },
    {
      "epoch": 0.9024824899498529,
      "grad_norm": 0.4068131446838379,
      "learning_rate": 4.877947697791041e-07,
      "loss": 0.661,
      "step": 21776
    },
    {
      "epoch": 0.9025239338555265,
      "grad_norm": 0.41802167892456055,
      "learning_rate": 4.875875502507357e-07,
      "loss": 0.6248,
      "step": 21777
    },
    {
      "epoch": 0.9025653777612002,
      "grad_norm": 0.4208430051803589,
      "learning_rate": 4.873803307223674e-07,
      "loss": 0.6805,
      "step": 21778
    },
    {
      "epoch": 0.9026068216668739,
      "grad_norm": 0.4063315987586975,
      "learning_rate": 4.87173111193999e-07,
      "loss": 0.5923,
      "step": 21779
    },
    {
      "epoch": 0.9026482655725475,
      "grad_norm": 0.40667447447776794,
      "learning_rate": 4.869658916656306e-07,
      "loss": 0.6437,
      "step": 21780
    },
    {
      "epoch": 0.9026897094782212,
      "grad_norm": 0.4293023347854614,
      "learning_rate": 4.867586721372623e-07,
      "loss": 0.7148,
      "step": 21781
    },
    {
      "epoch": 0.9027311533838949,
      "grad_norm": 0.41112715005874634,
      "learning_rate": 4.865514526088939e-07,
      "loss": 0.6965,
      "step": 21782
    },
    {
      "epoch": 0.9027725972895686,
      "grad_norm": 0.39250093698501587,
      "learning_rate": 4.863442330805256e-07,
      "loss": 0.6075,
      "step": 21783
    },
    {
      "epoch": 0.9028140411952422,
      "grad_norm": 0.42396581172943115,
      "learning_rate": 4.861370135521572e-07,
      "loss": 0.6631,
      "step": 21784
    },
    {
      "epoch": 0.902855485100916,
      "grad_norm": 0.3919146955013275,
      "learning_rate": 4.859297940237889e-07,
      "loss": 0.6434,
      "step": 21785
    },
    {
      "epoch": 0.9028969290065896,
      "grad_norm": 0.4356183111667633,
      "learning_rate": 4.857225744954205e-07,
      "loss": 0.6401,
      "step": 21786
    },
    {
      "epoch": 0.9029383729122632,
      "grad_norm": 0.3960706889629364,
      "learning_rate": 4.855153549670522e-07,
      "loss": 0.6464,
      "step": 21787
    },
    {
      "epoch": 0.9029798168179369,
      "grad_norm": 0.43283525109291077,
      "learning_rate": 4.853081354386838e-07,
      "loss": 0.6786,
      "step": 21788
    },
    {
      "epoch": 0.9030212607236106,
      "grad_norm": 0.4411073625087738,
      "learning_rate": 4.851009159103155e-07,
      "loss": 0.668,
      "step": 21789
    },
    {
      "epoch": 0.9030627046292843,
      "grad_norm": 0.43951287865638733,
      "learning_rate": 4.848936963819471e-07,
      "loss": 0.698,
      "step": 21790
    },
    {
      "epoch": 0.9031041485349579,
      "grad_norm": 0.391604483127594,
      "learning_rate": 4.846864768535788e-07,
      "loss": 0.6636,
      "step": 21791
    },
    {
      "epoch": 0.9031455924406316,
      "grad_norm": 0.4240918457508087,
      "learning_rate": 4.844792573252104e-07,
      "loss": 0.6539,
      "step": 21792
    },
    {
      "epoch": 0.9031870363463053,
      "grad_norm": 0.4464181959629059,
      "learning_rate": 4.84272037796842e-07,
      "loss": 0.6599,
      "step": 21793
    },
    {
      "epoch": 0.903228480251979,
      "grad_norm": 0.4299499988555908,
      "learning_rate": 4.840648182684737e-07,
      "loss": 0.7209,
      "step": 21794
    },
    {
      "epoch": 0.9032699241576526,
      "grad_norm": 0.39692774415016174,
      "learning_rate": 4.838575987401053e-07,
      "loss": 0.6685,
      "step": 21795
    },
    {
      "epoch": 0.9033113680633262,
      "grad_norm": 0.39123156666755676,
      "learning_rate": 4.83650379211737e-07,
      "loss": 0.6401,
      "step": 21796
    },
    {
      "epoch": 0.903352811969,
      "grad_norm": 0.4436837136745453,
      "learning_rate": 4.834431596833686e-07,
      "loss": 0.7235,
      "step": 21797
    },
    {
      "epoch": 0.9033942558746736,
      "grad_norm": 0.3717900514602661,
      "learning_rate": 4.832359401550003e-07,
      "loss": 0.6525,
      "step": 21798
    },
    {
      "epoch": 0.9034356997803473,
      "grad_norm": 0.41760775446891785,
      "learning_rate": 4.830287206266319e-07,
      "loss": 0.6498,
      "step": 21799
    },
    {
      "epoch": 0.903477143686021,
      "grad_norm": 0.4355441629886627,
      "learning_rate": 4.828215010982636e-07,
      "loss": 0.6812,
      "step": 21800
    },
    {
      "epoch": 0.9035185875916947,
      "grad_norm": 0.4129011034965515,
      "learning_rate": 4.826142815698952e-07,
      "loss": 0.6224,
      "step": 21801
    },
    {
      "epoch": 0.9035600314973683,
      "grad_norm": 0.44283023476600647,
      "learning_rate": 4.824070620415269e-07,
      "loss": 0.6927,
      "step": 21802
    },
    {
      "epoch": 0.903601475403042,
      "grad_norm": 0.38623684644699097,
      "learning_rate": 4.821998425131585e-07,
      "loss": 0.6517,
      "step": 21803
    },
    {
      "epoch": 0.9036429193087157,
      "grad_norm": 0.42322438955307007,
      "learning_rate": 4.819926229847902e-07,
      "loss": 0.626,
      "step": 21804
    },
    {
      "epoch": 0.9036843632143893,
      "grad_norm": 0.42495718598365784,
      "learning_rate": 4.817854034564218e-07,
      "loss": 0.6725,
      "step": 21805
    },
    {
      "epoch": 0.903725807120063,
      "grad_norm": 0.40512368083000183,
      "learning_rate": 4.815781839280534e-07,
      "loss": 0.6428,
      "step": 21806
    },
    {
      "epoch": 0.9037672510257366,
      "grad_norm": 0.4157402217388153,
      "learning_rate": 4.813709643996851e-07,
      "loss": 0.694,
      "step": 21807
    },
    {
      "epoch": 0.9038086949314104,
      "grad_norm": 0.3808349668979645,
      "learning_rate": 4.811637448713167e-07,
      "loss": 0.6125,
      "step": 21808
    },
    {
      "epoch": 0.903850138837084,
      "grad_norm": 0.4222026765346527,
      "learning_rate": 4.809565253429484e-07,
      "loss": 0.6724,
      "step": 21809
    },
    {
      "epoch": 0.9038915827427577,
      "grad_norm": 0.3946753442287445,
      "learning_rate": 4.8074930581458e-07,
      "loss": 0.6426,
      "step": 21810
    },
    {
      "epoch": 0.9039330266484313,
      "grad_norm": 0.42505571246147156,
      "learning_rate": 4.805420862862117e-07,
      "loss": 0.6462,
      "step": 21811
    },
    {
      "epoch": 0.9039744705541051,
      "grad_norm": 0.4211755394935608,
      "learning_rate": 4.803348667578433e-07,
      "loss": 0.6456,
      "step": 21812
    },
    {
      "epoch": 0.9040159144597787,
      "grad_norm": 0.42822906374931335,
      "learning_rate": 4.80127647229475e-07,
      "loss": 0.6517,
      "step": 21813
    },
    {
      "epoch": 0.9040573583654523,
      "grad_norm": 0.38781803846359253,
      "learning_rate": 4.799204277011066e-07,
      "loss": 0.5944,
      "step": 21814
    },
    {
      "epoch": 0.904098802271126,
      "grad_norm": 0.4216715395450592,
      "learning_rate": 4.797132081727383e-07,
      "loss": 0.6654,
      "step": 21815
    },
    {
      "epoch": 0.9041402461767997,
      "grad_norm": 0.41587114334106445,
      "learning_rate": 4.795059886443699e-07,
      "loss": 0.6495,
      "step": 21816
    },
    {
      "epoch": 0.9041816900824734,
      "grad_norm": 0.45223137736320496,
      "learning_rate": 4.792987691160015e-07,
      "loss": 0.6786,
      "step": 21817
    },
    {
      "epoch": 0.904223133988147,
      "grad_norm": 0.39984941482543945,
      "learning_rate": 4.790915495876332e-07,
      "loss": 0.6467,
      "step": 21818
    },
    {
      "epoch": 0.9042645778938208,
      "grad_norm": 0.5335744023323059,
      "learning_rate": 4.788843300592648e-07,
      "loss": 0.6859,
      "step": 21819
    },
    {
      "epoch": 0.9043060217994944,
      "grad_norm": 0.42930686473846436,
      "learning_rate": 4.786771105308965e-07,
      "loss": 0.6436,
      "step": 21820
    },
    {
      "epoch": 0.9043474657051681,
      "grad_norm": 0.41653481125831604,
      "learning_rate": 4.784698910025281e-07,
      "loss": 0.5924,
      "step": 21821
    },
    {
      "epoch": 0.9043889096108417,
      "grad_norm": 0.4115052819252014,
      "learning_rate": 4.782626714741598e-07,
      "loss": 0.649,
      "step": 21822
    },
    {
      "epoch": 0.9044303535165154,
      "grad_norm": 0.4311596751213074,
      "learning_rate": 4.780554519457914e-07,
      "loss": 0.6598,
      "step": 21823
    },
    {
      "epoch": 0.9044717974221891,
      "grad_norm": 0.4343392848968506,
      "learning_rate": 4.778482324174231e-07,
      "loss": 0.688,
      "step": 21824
    },
    {
      "epoch": 0.9045132413278627,
      "grad_norm": 0.383236289024353,
      "learning_rate": 4.776410128890547e-07,
      "loss": 0.6514,
      "step": 21825
    },
    {
      "epoch": 0.9045546852335364,
      "grad_norm": 0.4686683714389801,
      "learning_rate": 4.774337933606864e-07,
      "loss": 0.6909,
      "step": 21826
    },
    {
      "epoch": 0.9045961291392101,
      "grad_norm": 0.44237831234931946,
      "learning_rate": 4.77226573832318e-07,
      "loss": 0.6925,
      "step": 21827
    },
    {
      "epoch": 0.9046375730448838,
      "grad_norm": 0.4202333688735962,
      "learning_rate": 4.770193543039497e-07,
      "loss": 0.6427,
      "step": 21828
    },
    {
      "epoch": 0.9046790169505574,
      "grad_norm": 0.43009433150291443,
      "learning_rate": 4.768121347755813e-07,
      "loss": 0.6407,
      "step": 21829
    },
    {
      "epoch": 0.9047204608562311,
      "grad_norm": 0.4319974482059479,
      "learning_rate": 4.766049152472129e-07,
      "loss": 0.6824,
      "step": 21830
    },
    {
      "epoch": 0.9047619047619048,
      "grad_norm": 0.40904760360717773,
      "learning_rate": 4.763976957188446e-07,
      "loss": 0.6404,
      "step": 21831
    },
    {
      "epoch": 0.9048033486675784,
      "grad_norm": 0.4286786615848541,
      "learning_rate": 4.7619047619047623e-07,
      "loss": 0.6536,
      "step": 21832
    },
    {
      "epoch": 0.9048447925732521,
      "grad_norm": 0.415861040353775,
      "learning_rate": 4.759832566621079e-07,
      "loss": 0.6387,
      "step": 21833
    },
    {
      "epoch": 0.9048862364789257,
      "grad_norm": 0.4225499927997589,
      "learning_rate": 4.7577603713373953e-07,
      "loss": 0.6683,
      "step": 21834
    },
    {
      "epoch": 0.9049276803845995,
      "grad_norm": 0.4072605073451996,
      "learning_rate": 4.755688176053712e-07,
      "loss": 0.6633,
      "step": 21835
    },
    {
      "epoch": 0.9049691242902731,
      "grad_norm": 0.42319801449775696,
      "learning_rate": 4.753615980770028e-07,
      "loss": 0.6719,
      "step": 21836
    },
    {
      "epoch": 0.9050105681959468,
      "grad_norm": 0.4009440839290619,
      "learning_rate": 4.751543785486345e-07,
      "loss": 0.6562,
      "step": 21837
    },
    {
      "epoch": 0.9050520121016205,
      "grad_norm": 0.4773741364479065,
      "learning_rate": 4.749471590202661e-07,
      "loss": 0.6685,
      "step": 21838
    },
    {
      "epoch": 0.9050934560072941,
      "grad_norm": 0.3910205066204071,
      "learning_rate": 4.747399394918978e-07,
      "loss": 0.6338,
      "step": 21839
    },
    {
      "epoch": 0.9051348999129678,
      "grad_norm": 0.3949735462665558,
      "learning_rate": 4.745327199635294e-07,
      "loss": 0.6263,
      "step": 21840
    },
    {
      "epoch": 0.9051763438186414,
      "grad_norm": 0.4101082384586334,
      "learning_rate": 4.743255004351611e-07,
      "loss": 0.6216,
      "step": 21841
    },
    {
      "epoch": 0.9052177877243152,
      "grad_norm": 0.4428708851337433,
      "learning_rate": 4.741182809067927e-07,
      "loss": 0.6558,
      "step": 21842
    },
    {
      "epoch": 0.9052592316299888,
      "grad_norm": 0.3821789026260376,
      "learning_rate": 4.7391106137842433e-07,
      "loss": 0.5985,
      "step": 21843
    },
    {
      "epoch": 0.9053006755356625,
      "grad_norm": 0.43065911531448364,
      "learning_rate": 4.73703841850056e-07,
      "loss": 0.678,
      "step": 21844
    },
    {
      "epoch": 0.9053421194413361,
      "grad_norm": 0.4532856047153473,
      "learning_rate": 4.7349662232168763e-07,
      "loss": 0.6516,
      "step": 21845
    },
    {
      "epoch": 0.9053835633470099,
      "grad_norm": 0.519549548625946,
      "learning_rate": 4.732894027933193e-07,
      "loss": 0.6705,
      "step": 21846
    },
    {
      "epoch": 0.9054250072526835,
      "grad_norm": 0.42888814210891724,
      "learning_rate": 4.7308218326495093e-07,
      "loss": 0.6409,
      "step": 21847
    },
    {
      "epoch": 0.9054664511583571,
      "grad_norm": 0.44154980778694153,
      "learning_rate": 4.728749637365826e-07,
      "loss": 0.7161,
      "step": 21848
    },
    {
      "epoch": 0.9055078950640308,
      "grad_norm": 0.4338739812374115,
      "learning_rate": 4.7266774420821423e-07,
      "loss": 0.7103,
      "step": 21849
    },
    {
      "epoch": 0.9055493389697045,
      "grad_norm": 0.44967222213745117,
      "learning_rate": 4.724605246798459e-07,
      "loss": 0.6931,
      "step": 21850
    },
    {
      "epoch": 0.9055907828753782,
      "grad_norm": 0.40397676825523376,
      "learning_rate": 4.7225330515147753e-07,
      "loss": 0.6584,
      "step": 21851
    },
    {
      "epoch": 0.9056322267810518,
      "grad_norm": 0.41361865401268005,
      "learning_rate": 4.720460856231092e-07,
      "loss": 0.6486,
      "step": 21852
    },
    {
      "epoch": 0.9056736706867256,
      "grad_norm": 0.42927393317222595,
      "learning_rate": 4.7183886609474083e-07,
      "loss": 0.6436,
      "step": 21853
    },
    {
      "epoch": 0.9057151145923992,
      "grad_norm": 0.4177165925502777,
      "learning_rate": 4.7163164656637243e-07,
      "loss": 0.6389,
      "step": 21854
    },
    {
      "epoch": 0.9057565584980729,
      "grad_norm": 0.4381028115749359,
      "learning_rate": 4.714244270380041e-07,
      "loss": 0.6863,
      "step": 21855
    },
    {
      "epoch": 0.9057980024037465,
      "grad_norm": 0.41930004954338074,
      "learning_rate": 4.7121720750963573e-07,
      "loss": 0.6681,
      "step": 21856
    },
    {
      "epoch": 0.9058394463094201,
      "grad_norm": 0.46559232473373413,
      "learning_rate": 4.710099879812674e-07,
      "loss": 0.6669,
      "step": 21857
    },
    {
      "epoch": 0.9058808902150939,
      "grad_norm": 0.43353271484375,
      "learning_rate": 4.7080276845289903e-07,
      "loss": 0.6804,
      "step": 21858
    },
    {
      "epoch": 0.9059223341207675,
      "grad_norm": 0.4458071291446686,
      "learning_rate": 4.705955489245307e-07,
      "loss": 0.6609,
      "step": 21859
    },
    {
      "epoch": 0.9059637780264412,
      "grad_norm": 0.4055100679397583,
      "learning_rate": 4.7038832939616233e-07,
      "loss": 0.6891,
      "step": 21860
    },
    {
      "epoch": 0.9060052219321149,
      "grad_norm": 0.407653272151947,
      "learning_rate": 4.70181109867794e-07,
      "loss": 0.5922,
      "step": 21861
    },
    {
      "epoch": 0.9060466658377886,
      "grad_norm": 0.4469847083091736,
      "learning_rate": 4.6997389033942563e-07,
      "loss": 0.7156,
      "step": 21862
    },
    {
      "epoch": 0.9060881097434622,
      "grad_norm": 0.4088365435600281,
      "learning_rate": 4.697666708110573e-07,
      "loss": 0.6755,
      "step": 21863
    },
    {
      "epoch": 0.9061295536491359,
      "grad_norm": 0.40674006938934326,
      "learning_rate": 4.6955945128268893e-07,
      "loss": 0.6138,
      "step": 21864
    },
    {
      "epoch": 0.9061709975548096,
      "grad_norm": 0.4210219085216522,
      "learning_rate": 4.693522317543206e-07,
      "loss": 0.6562,
      "step": 21865
    },
    {
      "epoch": 0.9062124414604832,
      "grad_norm": 0.426429808139801,
      "learning_rate": 4.6914501222595223e-07,
      "loss": 0.636,
      "step": 21866
    },
    {
      "epoch": 0.9062538853661569,
      "grad_norm": 0.4359915554523468,
      "learning_rate": 4.6893779269758383e-07,
      "loss": 0.752,
      "step": 21867
    },
    {
      "epoch": 0.9062953292718305,
      "grad_norm": 0.4718107581138611,
      "learning_rate": 4.6873057316921553e-07,
      "loss": 0.7227,
      "step": 21868
    },
    {
      "epoch": 0.9063367731775043,
      "grad_norm": 0.4304783344268799,
      "learning_rate": 4.6852335364084713e-07,
      "loss": 0.6841,
      "step": 21869
    },
    {
      "epoch": 0.9063782170831779,
      "grad_norm": 0.4141804277896881,
      "learning_rate": 4.6831613411247883e-07,
      "loss": 0.6493,
      "step": 21870
    },
    {
      "epoch": 0.9064196609888516,
      "grad_norm": 0.45378121733665466,
      "learning_rate": 4.6810891458411043e-07,
      "loss": 0.7488,
      "step": 21871
    },
    {
      "epoch": 0.9064611048945252,
      "grad_norm": 0.40535059571266174,
      "learning_rate": 4.679016950557421e-07,
      "loss": 0.6337,
      "step": 21872
    },
    {
      "epoch": 0.906502548800199,
      "grad_norm": 0.4472917318344116,
      "learning_rate": 4.6769447552737373e-07,
      "loss": 0.6449,
      "step": 21873
    },
    {
      "epoch": 0.9065439927058726,
      "grad_norm": 0.39975082874298096,
      "learning_rate": 4.674872559990054e-07,
      "loss": 0.615,
      "step": 21874
    },
    {
      "epoch": 0.9065854366115462,
      "grad_norm": 0.38931435346603394,
      "learning_rate": 4.6728003647063703e-07,
      "loss": 0.631,
      "step": 21875
    },
    {
      "epoch": 0.90662688051722,
      "grad_norm": 0.4334350824356079,
      "learning_rate": 4.670728169422687e-07,
      "loss": 0.6769,
      "step": 21876
    },
    {
      "epoch": 0.9066683244228936,
      "grad_norm": 0.4389737546443939,
      "learning_rate": 4.6686559741390033e-07,
      "loss": 0.6421,
      "step": 21877
    },
    {
      "epoch": 0.9067097683285673,
      "grad_norm": 0.4046443700790405,
      "learning_rate": 4.66658377885532e-07,
      "loss": 0.6304,
      "step": 21878
    },
    {
      "epoch": 0.9067512122342409,
      "grad_norm": 0.40728917717933655,
      "learning_rate": 4.6645115835716363e-07,
      "loss": 0.6309,
      "step": 21879
    },
    {
      "epoch": 0.9067926561399147,
      "grad_norm": 0.41511136293411255,
      "learning_rate": 4.6624393882879523e-07,
      "loss": 0.6647,
      "step": 21880
    },
    {
      "epoch": 0.9068341000455883,
      "grad_norm": 0.5885186195373535,
      "learning_rate": 4.6603671930042694e-07,
      "loss": 0.6678,
      "step": 21881
    },
    {
      "epoch": 0.906875543951262,
      "grad_norm": 0.4216212034225464,
      "learning_rate": 4.6582949977205853e-07,
      "loss": 0.7012,
      "step": 21882
    },
    {
      "epoch": 0.9069169878569356,
      "grad_norm": 0.4141342043876648,
      "learning_rate": 4.6562228024369024e-07,
      "loss": 0.6354,
      "step": 21883
    },
    {
      "epoch": 0.9069584317626093,
      "grad_norm": 0.47348377108573914,
      "learning_rate": 4.6541506071532183e-07,
      "loss": 0.7502,
      "step": 21884
    },
    {
      "epoch": 0.906999875668283,
      "grad_norm": 0.41425594687461853,
      "learning_rate": 4.6520784118695354e-07,
      "loss": 0.6205,
      "step": 21885
    },
    {
      "epoch": 0.9070413195739566,
      "grad_norm": 0.4626140296459198,
      "learning_rate": 4.6500062165858513e-07,
      "loss": 0.6932,
      "step": 21886
    },
    {
      "epoch": 0.9070827634796303,
      "grad_norm": 0.42274755239486694,
      "learning_rate": 4.6479340213021684e-07,
      "loss": 0.6451,
      "step": 21887
    },
    {
      "epoch": 0.907124207385304,
      "grad_norm": 0.40548062324523926,
      "learning_rate": 4.6458618260184843e-07,
      "loss": 0.6821,
      "step": 21888
    },
    {
      "epoch": 0.9071656512909777,
      "grad_norm": 0.4095577299594879,
      "learning_rate": 4.6437896307348014e-07,
      "loss": 0.6631,
      "step": 21889
    },
    {
      "epoch": 0.9072070951966513,
      "grad_norm": 0.4232381582260132,
      "learning_rate": 4.6417174354511174e-07,
      "loss": 0.6317,
      "step": 21890
    },
    {
      "epoch": 0.9072485391023251,
      "grad_norm": 0.4019046127796173,
      "learning_rate": 4.6396452401674333e-07,
      "loss": 0.6663,
      "step": 21891
    },
    {
      "epoch": 0.9072899830079987,
      "grad_norm": 0.4106959402561188,
      "learning_rate": 4.6375730448837504e-07,
      "loss": 0.6738,
      "step": 21892
    },
    {
      "epoch": 0.9073314269136723,
      "grad_norm": 0.4423587918281555,
      "learning_rate": 4.6355008496000663e-07,
      "loss": 0.6569,
      "step": 21893
    },
    {
      "epoch": 0.907372870819346,
      "grad_norm": 0.4258795380592346,
      "learning_rate": 4.6334286543163834e-07,
      "loss": 0.7046,
      "step": 21894
    },
    {
      "epoch": 0.9074143147250197,
      "grad_norm": 0.4396324157714844,
      "learning_rate": 4.6313564590326993e-07,
      "loss": 0.7324,
      "step": 21895
    },
    {
      "epoch": 0.9074557586306934,
      "grad_norm": 0.4092320203781128,
      "learning_rate": 4.6292842637490164e-07,
      "loss": 0.6588,
      "step": 21896
    },
    {
      "epoch": 0.907497202536367,
      "grad_norm": 0.46263691782951355,
      "learning_rate": 4.6272120684653324e-07,
      "loss": 0.6875,
      "step": 21897
    },
    {
      "epoch": 0.9075386464420407,
      "grad_norm": 0.4211302101612091,
      "learning_rate": 4.6251398731816494e-07,
      "loss": 0.6006,
      "step": 21898
    },
    {
      "epoch": 0.9075800903477144,
      "grad_norm": 0.42266300320625305,
      "learning_rate": 4.6230676778979654e-07,
      "loss": 0.6685,
      "step": 21899
    },
    {
      "epoch": 0.907621534253388,
      "grad_norm": 0.4128749370574951,
      "learning_rate": 4.6209954826142824e-07,
      "loss": 0.6694,
      "step": 21900
    },
    {
      "epoch": 0.9076629781590617,
      "grad_norm": 0.40134397149086,
      "learning_rate": 4.6189232873305984e-07,
      "loss": 0.6202,
      "step": 21901
    },
    {
      "epoch": 0.9077044220647353,
      "grad_norm": 0.4068869650363922,
      "learning_rate": 4.6168510920469154e-07,
      "loss": 0.6194,
      "step": 21902
    },
    {
      "epoch": 0.9077458659704091,
      "grad_norm": 0.40625011920928955,
      "learning_rate": 4.6147788967632314e-07,
      "loss": 0.6892,
      "step": 21903
    },
    {
      "epoch": 0.9077873098760827,
      "grad_norm": 0.4061359465122223,
      "learning_rate": 4.6127067014795473e-07,
      "loss": 0.6736,
      "step": 21904
    },
    {
      "epoch": 0.9078287537817564,
      "grad_norm": 0.4065978229045868,
      "learning_rate": 4.6106345061958644e-07,
      "loss": 0.5699,
      "step": 21905
    },
    {
      "epoch": 0.90787019768743,
      "grad_norm": 0.45524531602859497,
      "learning_rate": 4.6085623109121804e-07,
      "loss": 0.7012,
      "step": 21906
    },
    {
      "epoch": 0.9079116415931038,
      "grad_norm": 0.39973676204681396,
      "learning_rate": 4.6064901156284974e-07,
      "loss": 0.6704,
      "step": 21907
    },
    {
      "epoch": 0.9079530854987774,
      "grad_norm": 0.44014686346054077,
      "learning_rate": 4.6044179203448134e-07,
      "loss": 0.6669,
      "step": 21908
    },
    {
      "epoch": 0.907994529404451,
      "grad_norm": 0.4096478819847107,
      "learning_rate": 4.6023457250611304e-07,
      "loss": 0.6682,
      "step": 21909
    },
    {
      "epoch": 0.9080359733101248,
      "grad_norm": 0.4659821391105652,
      "learning_rate": 4.6002735297774464e-07,
      "loss": 0.6686,
      "step": 21910
    },
    {
      "epoch": 0.9080774172157984,
      "grad_norm": 0.4361088275909424,
      "learning_rate": 4.5982013344937634e-07,
      "loss": 0.679,
      "step": 21911
    },
    {
      "epoch": 0.9081188611214721,
      "grad_norm": 0.4240829050540924,
      "learning_rate": 4.5961291392100794e-07,
      "loss": 0.6421,
      "step": 21912
    },
    {
      "epoch": 0.9081603050271457,
      "grad_norm": 0.4424685537815094,
      "learning_rate": 4.5940569439263964e-07,
      "loss": 0.7422,
      "step": 21913
    },
    {
      "epoch": 0.9082017489328195,
      "grad_norm": 0.4150722026824951,
      "learning_rate": 4.5919847486427124e-07,
      "loss": 0.6754,
      "step": 21914
    },
    {
      "epoch": 0.9082431928384931,
      "grad_norm": 0.45071372389793396,
      "learning_rate": 4.5899125533590294e-07,
      "loss": 0.6362,
      "step": 21915
    },
    {
      "epoch": 0.9082846367441668,
      "grad_norm": 0.4109066128730774,
      "learning_rate": 4.5878403580753454e-07,
      "loss": 0.665,
      "step": 21916
    },
    {
      "epoch": 0.9083260806498404,
      "grad_norm": 0.39218297600746155,
      "learning_rate": 4.5857681627916614e-07,
      "loss": 0.6315,
      "step": 21917
    },
    {
      "epoch": 0.9083675245555141,
      "grad_norm": 0.4137602150440216,
      "learning_rate": 4.5836959675079784e-07,
      "loss": 0.6589,
      "step": 21918
    },
    {
      "epoch": 0.9084089684611878,
      "grad_norm": 0.4427436292171478,
      "learning_rate": 4.5816237722242944e-07,
      "loss": 0.7095,
      "step": 21919
    },
    {
      "epoch": 0.9084504123668614,
      "grad_norm": 0.4001578390598297,
      "learning_rate": 4.5795515769406114e-07,
      "loss": 0.6466,
      "step": 21920
    },
    {
      "epoch": 0.9084918562725351,
      "grad_norm": 0.3878651261329651,
      "learning_rate": 4.5774793816569274e-07,
      "loss": 0.5944,
      "step": 21921
    },
    {
      "epoch": 0.9085333001782088,
      "grad_norm": 0.40754836797714233,
      "learning_rate": 4.5754071863732444e-07,
      "loss": 0.6643,
      "step": 21922
    },
    {
      "epoch": 0.9085747440838825,
      "grad_norm": 0.4177807867527008,
      "learning_rate": 4.5733349910895604e-07,
      "loss": 0.6506,
      "step": 21923
    },
    {
      "epoch": 0.9086161879895561,
      "grad_norm": 0.3908924460411072,
      "learning_rate": 4.5712627958058774e-07,
      "loss": 0.6453,
      "step": 21924
    },
    {
      "epoch": 0.9086576318952299,
      "grad_norm": 0.4043509364128113,
      "learning_rate": 4.5691906005221934e-07,
      "loss": 0.6527,
      "step": 21925
    },
    {
      "epoch": 0.9086990758009035,
      "grad_norm": 0.434023916721344,
      "learning_rate": 4.5671184052385104e-07,
      "loss": 0.6661,
      "step": 21926
    },
    {
      "epoch": 0.9087405197065771,
      "grad_norm": 0.4161599576473236,
      "learning_rate": 4.5650462099548264e-07,
      "loss": 0.717,
      "step": 21927
    },
    {
      "epoch": 0.9087819636122508,
      "grad_norm": 0.38105711340904236,
      "learning_rate": 4.562974014671143e-07,
      "loss": 0.5566,
      "step": 21928
    },
    {
      "epoch": 0.9088234075179245,
      "grad_norm": 0.43955835700035095,
      "learning_rate": 4.5609018193874594e-07,
      "loss": 0.647,
      "step": 21929
    },
    {
      "epoch": 0.9088648514235982,
      "grad_norm": 0.436076819896698,
      "learning_rate": 4.5588296241037754e-07,
      "loss": 0.6602,
      "step": 21930
    },
    {
      "epoch": 0.9089062953292718,
      "grad_norm": 0.4477609694004059,
      "learning_rate": 4.5567574288200924e-07,
      "loss": 0.6985,
      "step": 21931
    },
    {
      "epoch": 0.9089477392349455,
      "grad_norm": 0.39782240986824036,
      "learning_rate": 4.5546852335364084e-07,
      "loss": 0.6471,
      "step": 21932
    },
    {
      "epoch": 0.9089891831406192,
      "grad_norm": 0.3890943229198456,
      "learning_rate": 4.5526130382527254e-07,
      "loss": 0.6461,
      "step": 21933
    },
    {
      "epoch": 0.9090306270462929,
      "grad_norm": 0.3940759003162384,
      "learning_rate": 4.5505408429690414e-07,
      "loss": 0.6405,
      "step": 21934
    },
    {
      "epoch": 0.9090720709519665,
      "grad_norm": 0.4153643548488617,
      "learning_rate": 4.5484686476853584e-07,
      "loss": 0.6406,
      "step": 21935
    },
    {
      "epoch": 0.9091135148576401,
      "grad_norm": 0.39626652002334595,
      "learning_rate": 4.5463964524016744e-07,
      "loss": 0.6571,
      "step": 21936
    },
    {
      "epoch": 0.9091549587633139,
      "grad_norm": 0.4438709020614624,
      "learning_rate": 4.5443242571179914e-07,
      "loss": 0.6987,
      "step": 21937
    },
    {
      "epoch": 0.9091964026689875,
      "grad_norm": 0.4175708591938019,
      "learning_rate": 4.5422520618343074e-07,
      "loss": 0.6433,
      "step": 21938
    },
    {
      "epoch": 0.9092378465746612,
      "grad_norm": 0.385701984167099,
      "learning_rate": 4.5401798665506244e-07,
      "loss": 0.6625,
      "step": 21939
    },
    {
      "epoch": 0.9092792904803348,
      "grad_norm": 0.4024161994457245,
      "learning_rate": 4.5381076712669404e-07,
      "loss": 0.6106,
      "step": 21940
    },
    {
      "epoch": 0.9093207343860086,
      "grad_norm": 0.44621536135673523,
      "learning_rate": 4.536035475983257e-07,
      "loss": 0.7097,
      "step": 21941
    },
    {
      "epoch": 0.9093621782916822,
      "grad_norm": 0.40440747141838074,
      "learning_rate": 4.5339632806995734e-07,
      "loss": 0.6637,
      "step": 21942
    },
    {
      "epoch": 0.9094036221973559,
      "grad_norm": 0.39670151472091675,
      "learning_rate": 4.53189108541589e-07,
      "loss": 0.6516,
      "step": 21943
    },
    {
      "epoch": 0.9094450661030296,
      "grad_norm": 0.4108840525150299,
      "learning_rate": 4.5298188901322064e-07,
      "loss": 0.649,
      "step": 21944
    },
    {
      "epoch": 0.9094865100087032,
      "grad_norm": 0.4383496344089508,
      "learning_rate": 4.527746694848523e-07,
      "loss": 0.6748,
      "step": 21945
    },
    {
      "epoch": 0.9095279539143769,
      "grad_norm": 0.4075096547603607,
      "learning_rate": 4.5256744995648394e-07,
      "loss": 0.6283,
      "step": 21946
    },
    {
      "epoch": 0.9095693978200505,
      "grad_norm": 0.4298741817474365,
      "learning_rate": 4.523602304281156e-07,
      "loss": 0.6432,
      "step": 21947
    },
    {
      "epoch": 0.9096108417257243,
      "grad_norm": 0.4200645089149475,
      "learning_rate": 4.5215301089974724e-07,
      "loss": 0.6765,
      "step": 21948
    },
    {
      "epoch": 0.9096522856313979,
      "grad_norm": 0.42554399371147156,
      "learning_rate": 4.5194579137137884e-07,
      "loss": 0.6326,
      "step": 21949
    },
    {
      "epoch": 0.9096937295370716,
      "grad_norm": 0.51891028881073,
      "learning_rate": 4.5173857184301054e-07,
      "loss": 0.6926,
      "step": 21950
    },
    {
      "epoch": 0.9097351734427452,
      "grad_norm": 0.39836055040359497,
      "learning_rate": 4.5153135231464214e-07,
      "loss": 0.6775,
      "step": 21951
    },
    {
      "epoch": 0.909776617348419,
      "grad_norm": 0.44383516907691956,
      "learning_rate": 4.5132413278627385e-07,
      "loss": 0.6877,
      "step": 21952
    },
    {
      "epoch": 0.9098180612540926,
      "grad_norm": 0.38334357738494873,
      "learning_rate": 4.5111691325790544e-07,
      "loss": 0.621,
      "step": 21953
    },
    {
      "epoch": 0.9098595051597662,
      "grad_norm": 0.4602614939212799,
      "learning_rate": 4.509096937295371e-07,
      "loss": 0.6599,
      "step": 21954
    },
    {
      "epoch": 0.90990094906544,
      "grad_norm": 0.39133772253990173,
      "learning_rate": 4.5070247420116874e-07,
      "loss": 0.6481,
      "step": 21955
    },
    {
      "epoch": 0.9099423929711136,
      "grad_norm": 0.39993515610694885,
      "learning_rate": 4.504952546728004e-07,
      "loss": 0.6229,
      "step": 21956
    },
    {
      "epoch": 0.9099838368767873,
      "grad_norm": 0.4412272572517395,
      "learning_rate": 4.5028803514443204e-07,
      "loss": 0.6851,
      "step": 21957
    },
    {
      "epoch": 0.9100252807824609,
      "grad_norm": 0.4108656644821167,
      "learning_rate": 4.500808156160637e-07,
      "loss": 0.6387,
      "step": 21958
    },
    {
      "epoch": 0.9100667246881347,
      "grad_norm": 0.4274214208126068,
      "learning_rate": 4.4987359608769535e-07,
      "loss": 0.6506,
      "step": 21959
    },
    {
      "epoch": 0.9101081685938083,
      "grad_norm": 0.4025028944015503,
      "learning_rate": 4.49666376559327e-07,
      "loss": 0.6488,
      "step": 21960
    },
    {
      "epoch": 0.9101496124994819,
      "grad_norm": 0.46380704641342163,
      "learning_rate": 4.4945915703095865e-07,
      "loss": 0.7078,
      "step": 21961
    },
    {
      "epoch": 0.9101910564051556,
      "grad_norm": 0.41215527057647705,
      "learning_rate": 4.492519375025903e-07,
      "loss": 0.6287,
      "step": 21962
    },
    {
      "epoch": 0.9102325003108293,
      "grad_norm": 0.4708963632583618,
      "learning_rate": 4.4904471797422195e-07,
      "loss": 0.6975,
      "step": 21963
    },
    {
      "epoch": 0.910273944216503,
      "grad_norm": 0.4384475350379944,
      "learning_rate": 4.488374984458536e-07,
      "loss": 0.6367,
      "step": 21964
    },
    {
      "epoch": 0.9103153881221766,
      "grad_norm": 0.45328259468078613,
      "learning_rate": 4.486302789174852e-07,
      "loss": 0.6735,
      "step": 21965
    },
    {
      "epoch": 0.9103568320278503,
      "grad_norm": 0.4006426930427551,
      "learning_rate": 4.484230593891169e-07,
      "loss": 0.6594,
      "step": 21966
    },
    {
      "epoch": 0.910398275933524,
      "grad_norm": 0.45825493335723877,
      "learning_rate": 4.482158398607485e-07,
      "loss": 0.7468,
      "step": 21967
    },
    {
      "epoch": 0.9104397198391977,
      "grad_norm": 0.4354480504989624,
      "learning_rate": 4.4800862033238015e-07,
      "loss": 0.6713,
      "step": 21968
    },
    {
      "epoch": 0.9104811637448713,
      "grad_norm": 0.4258207082748413,
      "learning_rate": 4.478014008040118e-07,
      "loss": 0.6912,
      "step": 21969
    },
    {
      "epoch": 0.9105226076505449,
      "grad_norm": 0.3814166188240051,
      "learning_rate": 4.4759418127564345e-07,
      "loss": 0.6454,
      "step": 21970
    },
    {
      "epoch": 0.9105640515562187,
      "grad_norm": 0.466389536857605,
      "learning_rate": 4.473869617472751e-07,
      "loss": 0.7026,
      "step": 21971
    },
    {
      "epoch": 0.9106054954618923,
      "grad_norm": 0.42857474088668823,
      "learning_rate": 4.4717974221890675e-07,
      "loss": 0.6694,
      "step": 21972
    },
    {
      "epoch": 0.910646939367566,
      "grad_norm": 0.44757792353630066,
      "learning_rate": 4.469725226905384e-07,
      "loss": 0.7053,
      "step": 21973
    },
    {
      "epoch": 0.9106883832732396,
      "grad_norm": 0.3950522243976593,
      "learning_rate": 4.4676530316217005e-07,
      "loss": 0.6348,
      "step": 21974
    },
    {
      "epoch": 0.9107298271789134,
      "grad_norm": 0.3857004642486572,
      "learning_rate": 4.465580836338017e-07,
      "loss": 0.6711,
      "step": 21975
    },
    {
      "epoch": 0.910771271084587,
      "grad_norm": 0.40685707330703735,
      "learning_rate": 4.4635086410543335e-07,
      "loss": 0.6753,
      "step": 21976
    },
    {
      "epoch": 0.9108127149902607,
      "grad_norm": 0.42790019512176514,
      "learning_rate": 4.46143644577065e-07,
      "loss": 0.6578,
      "step": 21977
    },
    {
      "epoch": 0.9108541588959344,
      "grad_norm": 0.4190075099468231,
      "learning_rate": 4.459364250486966e-07,
      "loss": 0.7056,
      "step": 21978
    },
    {
      "epoch": 0.910895602801608,
      "grad_norm": 0.4036070704460144,
      "learning_rate": 4.457292055203283e-07,
      "loss": 0.656,
      "step": 21979
    },
    {
      "epoch": 0.9109370467072817,
      "grad_norm": 0.40884602069854736,
      "learning_rate": 4.455219859919599e-07,
      "loss": 0.6571,
      "step": 21980
    },
    {
      "epoch": 0.9109784906129553,
      "grad_norm": 0.4441774785518646,
      "learning_rate": 4.453147664635916e-07,
      "loss": 0.6443,
      "step": 21981
    },
    {
      "epoch": 0.9110199345186291,
      "grad_norm": 0.41033732891082764,
      "learning_rate": 4.451075469352232e-07,
      "loss": 0.6296,
      "step": 21982
    },
    {
      "epoch": 0.9110613784243027,
      "grad_norm": 0.4200485050678253,
      "learning_rate": 4.449003274068549e-07,
      "loss": 0.6843,
      "step": 21983
    },
    {
      "epoch": 0.9111028223299764,
      "grad_norm": 0.4122755825519562,
      "learning_rate": 4.446931078784865e-07,
      "loss": 0.6691,
      "step": 21984
    },
    {
      "epoch": 0.91114426623565,
      "grad_norm": 0.41610583662986755,
      "learning_rate": 4.4448588835011815e-07,
      "loss": 0.6558,
      "step": 21985
    },
    {
      "epoch": 0.9111857101413238,
      "grad_norm": 0.4091094136238098,
      "learning_rate": 4.442786688217498e-07,
      "loss": 0.6469,
      "step": 21986
    },
    {
      "epoch": 0.9112271540469974,
      "grad_norm": 0.4278522729873657,
      "learning_rate": 4.4407144929338145e-07,
      "loss": 0.657,
      "step": 21987
    },
    {
      "epoch": 0.911268597952671,
      "grad_norm": 0.4479804039001465,
      "learning_rate": 4.438642297650131e-07,
      "loss": 0.6539,
      "step": 21988
    },
    {
      "epoch": 0.9113100418583447,
      "grad_norm": 0.39609208703041077,
      "learning_rate": 4.4365701023664475e-07,
      "loss": 0.6855,
      "step": 21989
    },
    {
      "epoch": 0.9113514857640184,
      "grad_norm": 0.4221287667751312,
      "learning_rate": 4.434497907082764e-07,
      "loss": 0.6411,
      "step": 21990
    },
    {
      "epoch": 0.9113929296696921,
      "grad_norm": 0.41076070070266724,
      "learning_rate": 4.43242571179908e-07,
      "loss": 0.6555,
      "step": 21991
    },
    {
      "epoch": 0.9114343735753657,
      "grad_norm": 0.44083017110824585,
      "learning_rate": 4.430353516515397e-07,
      "loss": 0.6818,
      "step": 21992
    },
    {
      "epoch": 0.9114758174810395,
      "grad_norm": 0.4445989727973938,
      "learning_rate": 4.428281321231713e-07,
      "loss": 0.6917,
      "step": 21993
    },
    {
      "epoch": 0.9115172613867131,
      "grad_norm": 0.41486185789108276,
      "learning_rate": 4.42620912594803e-07,
      "loss": 0.6536,
      "step": 21994
    },
    {
      "epoch": 0.9115587052923868,
      "grad_norm": 0.43957579135894775,
      "learning_rate": 4.424136930664346e-07,
      "loss": 0.6379,
      "step": 21995
    },
    {
      "epoch": 0.9116001491980604,
      "grad_norm": 0.41291743516921997,
      "learning_rate": 4.422064735380663e-07,
      "loss": 0.6539,
      "step": 21996
    },
    {
      "epoch": 0.911641593103734,
      "grad_norm": 0.416046142578125,
      "learning_rate": 4.419992540096979e-07,
      "loss": 0.6846,
      "step": 21997
    },
    {
      "epoch": 0.9116830370094078,
      "grad_norm": 0.4313727617263794,
      "learning_rate": 4.417920344813296e-07,
      "loss": 0.6499,
      "step": 21998
    },
    {
      "epoch": 0.9117244809150814,
      "grad_norm": 0.4496467113494873,
      "learning_rate": 4.415848149529612e-07,
      "loss": 0.7135,
      "step": 21999
    },
    {
      "epoch": 0.9117659248207551,
      "grad_norm": 0.4458792805671692,
      "learning_rate": 4.413775954245929e-07,
      "loss": 0.657,
      "step": 22000
    },
    {
      "epoch": 0.9118073687264288,
      "grad_norm": 0.39992889761924744,
      "learning_rate": 4.411703758962245e-07,
      "loss": 0.658,
      "step": 22001
    },
    {
      "epoch": 0.9118488126321025,
      "grad_norm": 0.40975651144981384,
      "learning_rate": 4.409631563678561e-07,
      "loss": 0.6447,
      "step": 22002
    },
    {
      "epoch": 0.9118902565377761,
      "grad_norm": 0.5595182180404663,
      "learning_rate": 4.407559368394878e-07,
      "loss": 0.7329,
      "step": 22003
    },
    {
      "epoch": 0.9119317004434498,
      "grad_norm": 0.40182971954345703,
      "learning_rate": 4.405487173111194e-07,
      "loss": 0.6836,
      "step": 22004
    },
    {
      "epoch": 0.9119731443491235,
      "grad_norm": 0.42242270708084106,
      "learning_rate": 4.403414977827511e-07,
      "loss": 0.6414,
      "step": 22005
    },
    {
      "epoch": 0.9120145882547971,
      "grad_norm": 0.38808000087738037,
      "learning_rate": 4.401342782543827e-07,
      "loss": 0.6488,
      "step": 22006
    },
    {
      "epoch": 0.9120560321604708,
      "grad_norm": 0.3949698209762573,
      "learning_rate": 4.399270587260144e-07,
      "loss": 0.6497,
      "step": 22007
    },
    {
      "epoch": 0.9120974760661444,
      "grad_norm": 0.41160544753074646,
      "learning_rate": 4.39719839197646e-07,
      "loss": 0.6862,
      "step": 22008
    },
    {
      "epoch": 0.9121389199718182,
      "grad_norm": 0.4011923372745514,
      "learning_rate": 4.395126196692777e-07,
      "loss": 0.65,
      "step": 22009
    },
    {
      "epoch": 0.9121803638774918,
      "grad_norm": 0.4130443036556244,
      "learning_rate": 4.393054001409093e-07,
      "loss": 0.6526,
      "step": 22010
    },
    {
      "epoch": 0.9122218077831655,
      "grad_norm": 0.39396265149116516,
      "learning_rate": 4.39098180612541e-07,
      "loss": 0.6409,
      "step": 22011
    },
    {
      "epoch": 0.9122632516888391,
      "grad_norm": 0.433217853307724,
      "learning_rate": 4.388909610841726e-07,
      "loss": 0.7363,
      "step": 22012
    },
    {
      "epoch": 0.9123046955945129,
      "grad_norm": 0.4155697226524353,
      "learning_rate": 4.386837415558043e-07,
      "loss": 0.6562,
      "step": 22013
    },
    {
      "epoch": 0.9123461395001865,
      "grad_norm": 0.4628601372241974,
      "learning_rate": 4.384765220274359e-07,
      "loss": 0.6797,
      "step": 22014
    },
    {
      "epoch": 0.9123875834058601,
      "grad_norm": 0.46119269728660583,
      "learning_rate": 4.382693024990675e-07,
      "loss": 0.6533,
      "step": 22015
    },
    {
      "epoch": 0.9124290273115339,
      "grad_norm": 0.42125385999679565,
      "learning_rate": 4.380620829706992e-07,
      "loss": 0.6631,
      "step": 22016
    },
    {
      "epoch": 0.9124704712172075,
      "grad_norm": 0.4196785092353821,
      "learning_rate": 4.378548634423308e-07,
      "loss": 0.668,
      "step": 22017
    },
    {
      "epoch": 0.9125119151228812,
      "grad_norm": 0.4031142294406891,
      "learning_rate": 4.376476439139625e-07,
      "loss": 0.6443,
      "step": 22018
    },
    {
      "epoch": 0.9125533590285548,
      "grad_norm": 0.431690514087677,
      "learning_rate": 4.374404243855941e-07,
      "loss": 0.6937,
      "step": 22019
    },
    {
      "epoch": 0.9125948029342286,
      "grad_norm": 0.4232026934623718,
      "learning_rate": 4.372332048572258e-07,
      "loss": 0.609,
      "step": 22020
    },
    {
      "epoch": 0.9126362468399022,
      "grad_norm": 0.4330886900424957,
      "learning_rate": 4.370259853288574e-07,
      "loss": 0.6357,
      "step": 22021
    },
    {
      "epoch": 0.9126776907455758,
      "grad_norm": 0.38470128178596497,
      "learning_rate": 4.368187658004891e-07,
      "loss": 0.6125,
      "step": 22022
    },
    {
      "epoch": 0.9127191346512495,
      "grad_norm": 0.41551080346107483,
      "learning_rate": 4.366115462721207e-07,
      "loss": 0.6606,
      "step": 22023
    },
    {
      "epoch": 0.9127605785569232,
      "grad_norm": 0.3932059407234192,
      "learning_rate": 4.364043267437524e-07,
      "loss": 0.6472,
      "step": 22024
    },
    {
      "epoch": 0.9128020224625969,
      "grad_norm": 0.4135592579841614,
      "learning_rate": 4.36197107215384e-07,
      "loss": 0.608,
      "step": 22025
    },
    {
      "epoch": 0.9128434663682705,
      "grad_norm": 0.40718215703964233,
      "learning_rate": 4.359898876870156e-07,
      "loss": 0.6643,
      "step": 22026
    },
    {
      "epoch": 0.9128849102739443,
      "grad_norm": 0.40444228053092957,
      "learning_rate": 4.357826681586473e-07,
      "loss": 0.646,
      "step": 22027
    },
    {
      "epoch": 0.9129263541796179,
      "grad_norm": 0.4239163398742676,
      "learning_rate": 4.355754486302789e-07,
      "loss": 0.6718,
      "step": 22028
    },
    {
      "epoch": 0.9129677980852916,
      "grad_norm": 0.4534132778644562,
      "learning_rate": 4.353682291019106e-07,
      "loss": 0.7458,
      "step": 22029
    },
    {
      "epoch": 0.9130092419909652,
      "grad_norm": 0.3860693871974945,
      "learning_rate": 4.351610095735422e-07,
      "loss": 0.6191,
      "step": 22030
    },
    {
      "epoch": 0.9130506858966388,
      "grad_norm": 0.3922826051712036,
      "learning_rate": 4.349537900451739e-07,
      "loss": 0.655,
      "step": 22031
    },
    {
      "epoch": 0.9130921298023126,
      "grad_norm": 0.43598267436027527,
      "learning_rate": 4.347465705168055e-07,
      "loss": 0.6982,
      "step": 22032
    },
    {
      "epoch": 0.9131335737079862,
      "grad_norm": 0.4274459779262543,
      "learning_rate": 4.345393509884372e-07,
      "loss": 0.6753,
      "step": 22033
    },
    {
      "epoch": 0.9131750176136599,
      "grad_norm": 0.44872531294822693,
      "learning_rate": 4.343321314600688e-07,
      "loss": 0.6226,
      "step": 22034
    },
    {
      "epoch": 0.9132164615193336,
      "grad_norm": 0.4802679419517517,
      "learning_rate": 4.341249119317005e-07,
      "loss": 0.7542,
      "step": 22035
    },
    {
      "epoch": 0.9132579054250073,
      "grad_norm": 0.45738622546195984,
      "learning_rate": 4.339176924033321e-07,
      "loss": 0.7058,
      "step": 22036
    },
    {
      "epoch": 0.9132993493306809,
      "grad_norm": 0.39424073696136475,
      "learning_rate": 4.337104728749638e-07,
      "loss": 0.6538,
      "step": 22037
    },
    {
      "epoch": 0.9133407932363546,
      "grad_norm": 0.40317121148109436,
      "learning_rate": 4.335032533465954e-07,
      "loss": 0.6401,
      "step": 22038
    },
    {
      "epoch": 0.9133822371420283,
      "grad_norm": 0.40300115942955017,
      "learning_rate": 4.3329603381822706e-07,
      "loss": 0.6306,
      "step": 22039
    },
    {
      "epoch": 0.9134236810477019,
      "grad_norm": 0.41830310225486755,
      "learning_rate": 4.330888142898587e-07,
      "loss": 0.6422,
      "step": 22040
    },
    {
      "epoch": 0.9134651249533756,
      "grad_norm": 0.4213850796222687,
      "learning_rate": 4.3288159476149036e-07,
      "loss": 0.7317,
      "step": 22041
    },
    {
      "epoch": 0.9135065688590492,
      "grad_norm": 0.41475746035575867,
      "learning_rate": 4.32674375233122e-07,
      "loss": 0.6157,
      "step": 22042
    },
    {
      "epoch": 0.913548012764723,
      "grad_norm": 0.42407694458961487,
      "learning_rate": 4.324671557047536e-07,
      "loss": 0.6875,
      "step": 22043
    },
    {
      "epoch": 0.9135894566703966,
      "grad_norm": 0.4370182454586029,
      "learning_rate": 4.322599361763853e-07,
      "loss": 0.672,
      "step": 22044
    },
    {
      "epoch": 0.9136309005760703,
      "grad_norm": 0.4042162299156189,
      "learning_rate": 4.320527166480169e-07,
      "loss": 0.6309,
      "step": 22045
    },
    {
      "epoch": 0.913672344481744,
      "grad_norm": 0.4311542510986328,
      "learning_rate": 4.318454971196486e-07,
      "loss": 0.7224,
      "step": 22046
    },
    {
      "epoch": 0.9137137883874177,
      "grad_norm": 0.4370914101600647,
      "learning_rate": 4.316382775912802e-07,
      "loss": 0.689,
      "step": 22047
    },
    {
      "epoch": 0.9137552322930913,
      "grad_norm": 0.4042140245437622,
      "learning_rate": 4.314310580629119e-07,
      "loss": 0.6653,
      "step": 22048
    },
    {
      "epoch": 0.9137966761987649,
      "grad_norm": 0.41949939727783203,
      "learning_rate": 4.312238385345435e-07,
      "loss": 0.6208,
      "step": 22049
    },
    {
      "epoch": 0.9138381201044387,
      "grad_norm": 0.38674020767211914,
      "learning_rate": 4.310166190061752e-07,
      "loss": 0.6428,
      "step": 22050
    },
    {
      "epoch": 0.9138795640101123,
      "grad_norm": 0.41467535495758057,
      "learning_rate": 4.308093994778068e-07,
      "loss": 0.6255,
      "step": 22051
    },
    {
      "epoch": 0.913921007915786,
      "grad_norm": 0.42031770944595337,
      "learning_rate": 4.3060217994943846e-07,
      "loss": 0.6649,
      "step": 22052
    },
    {
      "epoch": 0.9139624518214596,
      "grad_norm": 0.4564233422279358,
      "learning_rate": 4.303949604210701e-07,
      "loss": 0.6639,
      "step": 22053
    },
    {
      "epoch": 0.9140038957271334,
      "grad_norm": 0.4058915674686432,
      "learning_rate": 4.3018774089270176e-07,
      "loss": 0.6449,
      "step": 22054
    },
    {
      "epoch": 0.914045339632807,
      "grad_norm": 0.4156801998615265,
      "learning_rate": 4.299805213643334e-07,
      "loss": 0.6193,
      "step": 22055
    },
    {
      "epoch": 0.9140867835384807,
      "grad_norm": 0.39422816038131714,
      "learning_rate": 4.2977330183596506e-07,
      "loss": 0.6525,
      "step": 22056
    },
    {
      "epoch": 0.9141282274441543,
      "grad_norm": 0.38649216294288635,
      "learning_rate": 4.295660823075967e-07,
      "loss": 0.6765,
      "step": 22057
    },
    {
      "epoch": 0.914169671349828,
      "grad_norm": 0.41150787472724915,
      "learning_rate": 4.2935886277922836e-07,
      "loss": 0.6249,
      "step": 22058
    },
    {
      "epoch": 0.9142111152555017,
      "grad_norm": 0.42460593581199646,
      "learning_rate": 4.2915164325086e-07,
      "loss": 0.6697,
      "step": 22059
    },
    {
      "epoch": 0.9142525591611753,
      "grad_norm": 0.43042606115341187,
      "learning_rate": 4.2894442372249166e-07,
      "loss": 0.6335,
      "step": 22060
    },
    {
      "epoch": 0.914294003066849,
      "grad_norm": 0.39188334345817566,
      "learning_rate": 4.287372041941233e-07,
      "loss": 0.657,
      "step": 22061
    },
    {
      "epoch": 0.9143354469725227,
      "grad_norm": 0.4087007939815521,
      "learning_rate": 4.285299846657549e-07,
      "loss": 0.668,
      "step": 22062
    },
    {
      "epoch": 0.9143768908781964,
      "grad_norm": 0.41290420293807983,
      "learning_rate": 4.2832276513738656e-07,
      "loss": 0.6279,
      "step": 22063
    },
    {
      "epoch": 0.91441833478387,
      "grad_norm": 0.45419225096702576,
      "learning_rate": 4.281155456090182e-07,
      "loss": 0.6479,
      "step": 22064
    },
    {
      "epoch": 0.9144597786895438,
      "grad_norm": 0.41928571462631226,
      "learning_rate": 4.2790832608064986e-07,
      "loss": 0.6855,
      "step": 22065
    },
    {
      "epoch": 0.9145012225952174,
      "grad_norm": 0.4196389615535736,
      "learning_rate": 4.277011065522815e-07,
      "loss": 0.6785,
      "step": 22066
    },
    {
      "epoch": 0.914542666500891,
      "grad_norm": 0.39780735969543457,
      "learning_rate": 4.2749388702391316e-07,
      "loss": 0.6121,
      "step": 22067
    },
    {
      "epoch": 0.9145841104065647,
      "grad_norm": 0.4089689552783966,
      "learning_rate": 4.272866674955448e-07,
      "loss": 0.64,
      "step": 22068
    },
    {
      "epoch": 0.9146255543122384,
      "grad_norm": 0.4306671917438507,
      "learning_rate": 4.2707944796717646e-07,
      "loss": 0.6172,
      "step": 22069
    },
    {
      "epoch": 0.9146669982179121,
      "grad_norm": 0.42286109924316406,
      "learning_rate": 4.268722284388081e-07,
      "loss": 0.605,
      "step": 22070
    },
    {
      "epoch": 0.9147084421235857,
      "grad_norm": 0.41175487637519836,
      "learning_rate": 4.2666500891043976e-07,
      "loss": 0.6726,
      "step": 22071
    },
    {
      "epoch": 0.9147498860292594,
      "grad_norm": 0.41872844099998474,
      "learning_rate": 4.264577893820714e-07,
      "loss": 0.6763,
      "step": 22072
    },
    {
      "epoch": 0.9147913299349331,
      "grad_norm": 0.4220089912414551,
      "learning_rate": 4.2625056985370306e-07,
      "loss": 0.6331,
      "step": 22073
    },
    {
      "epoch": 0.9148327738406067,
      "grad_norm": 0.4154084622859955,
      "learning_rate": 4.260433503253347e-07,
      "loss": 0.6609,
      "step": 22074
    },
    {
      "epoch": 0.9148742177462804,
      "grad_norm": 0.4035089910030365,
      "learning_rate": 4.2583613079696636e-07,
      "loss": 0.6476,
      "step": 22075
    },
    {
      "epoch": 0.914915661651954,
      "grad_norm": 0.42882323265075684,
      "learning_rate": 4.2562891126859796e-07,
      "loss": 0.6831,
      "step": 22076
    },
    {
      "epoch": 0.9149571055576278,
      "grad_norm": 0.43196332454681396,
      "learning_rate": 4.2542169174022966e-07,
      "loss": 0.6411,
      "step": 22077
    },
    {
      "epoch": 0.9149985494633014,
      "grad_norm": 0.4543108642101288,
      "learning_rate": 4.2521447221186126e-07,
      "loss": 0.6669,
      "step": 22078
    },
    {
      "epoch": 0.9150399933689751,
      "grad_norm": 0.3963073492050171,
      "learning_rate": 4.2500725268349296e-07,
      "loss": 0.6545,
      "step": 22079
    },
    {
      "epoch": 0.9150814372746487,
      "grad_norm": 0.41498783230781555,
      "learning_rate": 4.2480003315512456e-07,
      "loss": 0.679,
      "step": 22080
    },
    {
      "epoch": 0.9151228811803225,
      "grad_norm": 0.38928982615470886,
      "learning_rate": 4.245928136267562e-07,
      "loss": 0.6365,
      "step": 22081
    },
    {
      "epoch": 0.9151643250859961,
      "grad_norm": 0.4207240641117096,
      "learning_rate": 4.2438559409838786e-07,
      "loss": 0.6577,
      "step": 22082
    },
    {
      "epoch": 0.9152057689916697,
      "grad_norm": 0.4153026342391968,
      "learning_rate": 4.241783745700195e-07,
      "loss": 0.6445,
      "step": 22083
    },
    {
      "epoch": 0.9152472128973435,
      "grad_norm": 0.43801453709602356,
      "learning_rate": 4.2397115504165116e-07,
      "loss": 0.7031,
      "step": 22084
    },
    {
      "epoch": 0.9152886568030171,
      "grad_norm": 0.4115074872970581,
      "learning_rate": 4.237639355132828e-07,
      "loss": 0.6948,
      "step": 22085
    },
    {
      "epoch": 0.9153301007086908,
      "grad_norm": 0.4357074797153473,
      "learning_rate": 4.2355671598491446e-07,
      "loss": 0.6864,
      "step": 22086
    },
    {
      "epoch": 0.9153715446143644,
      "grad_norm": 0.44178858399391174,
      "learning_rate": 4.233494964565461e-07,
      "loss": 0.6357,
      "step": 22087
    },
    {
      "epoch": 0.9154129885200382,
      "grad_norm": 0.4617457091808319,
      "learning_rate": 4.2314227692817776e-07,
      "loss": 0.6183,
      "step": 22088
    },
    {
      "epoch": 0.9154544324257118,
      "grad_norm": 0.43514496088027954,
      "learning_rate": 4.2293505739980936e-07,
      "loss": 0.6959,
      "step": 22089
    },
    {
      "epoch": 0.9154958763313855,
      "grad_norm": 0.3878042995929718,
      "learning_rate": 4.2272783787144106e-07,
      "loss": 0.611,
      "step": 22090
    },
    {
      "epoch": 0.9155373202370591,
      "grad_norm": 0.4126483201980591,
      "learning_rate": 4.2252061834307266e-07,
      "loss": 0.6809,
      "step": 22091
    },
    {
      "epoch": 0.9155787641427328,
      "grad_norm": 0.4401007294654846,
      "learning_rate": 4.2231339881470437e-07,
      "loss": 0.7075,
      "step": 22092
    },
    {
      "epoch": 0.9156202080484065,
      "grad_norm": 0.4173707365989685,
      "learning_rate": 4.2210617928633596e-07,
      "loss": 0.6227,
      "step": 22093
    },
    {
      "epoch": 0.9156616519540801,
      "grad_norm": 0.4250272810459137,
      "learning_rate": 4.2189895975796767e-07,
      "loss": 0.6339,
      "step": 22094
    },
    {
      "epoch": 0.9157030958597538,
      "grad_norm": 0.38277488946914673,
      "learning_rate": 4.2169174022959926e-07,
      "loss": 0.6501,
      "step": 22095
    },
    {
      "epoch": 0.9157445397654275,
      "grad_norm": 0.45190346240997314,
      "learning_rate": 4.2148452070123097e-07,
      "loss": 0.6846,
      "step": 22096
    },
    {
      "epoch": 0.9157859836711012,
      "grad_norm": 0.40072646737098694,
      "learning_rate": 4.2127730117286256e-07,
      "loss": 0.5901,
      "step": 22097
    },
    {
      "epoch": 0.9158274275767748,
      "grad_norm": 0.40093716979026794,
      "learning_rate": 4.2107008164449427e-07,
      "loss": 0.647,
      "step": 22098
    },
    {
      "epoch": 0.9158688714824486,
      "grad_norm": 0.3733333349227905,
      "learning_rate": 4.2086286211612586e-07,
      "loss": 0.6316,
      "step": 22099
    },
    {
      "epoch": 0.9159103153881222,
      "grad_norm": 0.44776588678359985,
      "learning_rate": 4.2065564258775746e-07,
      "loss": 0.6304,
      "step": 22100
    },
    {
      "epoch": 0.9159517592937958,
      "grad_norm": 0.4183082580566406,
      "learning_rate": 4.2044842305938917e-07,
      "loss": 0.6985,
      "step": 22101
    },
    {
      "epoch": 0.9159932031994695,
      "grad_norm": 0.43177926540374756,
      "learning_rate": 4.2024120353102076e-07,
      "loss": 0.7092,
      "step": 22102
    },
    {
      "epoch": 0.9160346471051432,
      "grad_norm": 0.4306720495223999,
      "learning_rate": 4.2003398400265247e-07,
      "loss": 0.6677,
      "step": 22103
    },
    {
      "epoch": 0.9160760910108169,
      "grad_norm": 0.4406459331512451,
      "learning_rate": 4.1982676447428406e-07,
      "loss": 0.6584,
      "step": 22104
    },
    {
      "epoch": 0.9161175349164905,
      "grad_norm": 0.4169570207595825,
      "learning_rate": 4.1961954494591577e-07,
      "loss": 0.6575,
      "step": 22105
    },
    {
      "epoch": 0.9161589788221642,
      "grad_norm": 0.4203692078590393,
      "learning_rate": 4.1941232541754736e-07,
      "loss": 0.6599,
      "step": 22106
    },
    {
      "epoch": 0.9162004227278379,
      "grad_norm": 0.42578038573265076,
      "learning_rate": 4.1920510588917907e-07,
      "loss": 0.687,
      "step": 22107
    },
    {
      "epoch": 0.9162418666335116,
      "grad_norm": 0.41912221908569336,
      "learning_rate": 4.1899788636081067e-07,
      "loss": 0.6926,
      "step": 22108
    },
    {
      "epoch": 0.9162833105391852,
      "grad_norm": 0.4718819260597229,
      "learning_rate": 4.1879066683244237e-07,
      "loss": 0.71,
      "step": 22109
    },
    {
      "epoch": 0.9163247544448588,
      "grad_norm": 0.4216209352016449,
      "learning_rate": 4.1858344730407397e-07,
      "loss": 0.7125,
      "step": 22110
    },
    {
      "epoch": 0.9163661983505326,
      "grad_norm": 0.41354256868362427,
      "learning_rate": 4.1837622777570567e-07,
      "loss": 0.7076,
      "step": 22111
    },
    {
      "epoch": 0.9164076422562062,
      "grad_norm": 0.42258474230766296,
      "learning_rate": 4.1816900824733727e-07,
      "loss": 0.6454,
      "step": 22112
    },
    {
      "epoch": 0.9164490861618799,
      "grad_norm": 0.4206172525882721,
      "learning_rate": 4.1796178871896886e-07,
      "loss": 0.7007,
      "step": 22113
    },
    {
      "epoch": 0.9164905300675535,
      "grad_norm": 0.4062540531158447,
      "learning_rate": 4.1775456919060057e-07,
      "loss": 0.6541,
      "step": 22114
    },
    {
      "epoch": 0.9165319739732273,
      "grad_norm": 0.39082372188568115,
      "learning_rate": 4.1754734966223216e-07,
      "loss": 0.6409,
      "step": 22115
    },
    {
      "epoch": 0.9165734178789009,
      "grad_norm": 0.42778483033180237,
      "learning_rate": 4.1734013013386387e-07,
      "loss": 0.7219,
      "step": 22116
    },
    {
      "epoch": 0.9166148617845746,
      "grad_norm": 0.4433804154396057,
      "learning_rate": 4.1713291060549547e-07,
      "loss": 0.7152,
      "step": 22117
    },
    {
      "epoch": 0.9166563056902483,
      "grad_norm": 0.3983723521232605,
      "learning_rate": 4.1692569107712717e-07,
      "loss": 0.618,
      "step": 22118
    },
    {
      "epoch": 0.9166977495959219,
      "grad_norm": 0.42044684290885925,
      "learning_rate": 4.1671847154875877e-07,
      "loss": 0.6731,
      "step": 22119
    },
    {
      "epoch": 0.9167391935015956,
      "grad_norm": 0.3724820613861084,
      "learning_rate": 4.1651125202039047e-07,
      "loss": 0.627,
      "step": 22120
    },
    {
      "epoch": 0.9167806374072692,
      "grad_norm": 0.461250901222229,
      "learning_rate": 4.1630403249202207e-07,
      "loss": 0.6454,
      "step": 22121
    },
    {
      "epoch": 0.916822081312943,
      "grad_norm": 0.40330570936203003,
      "learning_rate": 4.1609681296365377e-07,
      "loss": 0.667,
      "step": 22122
    },
    {
      "epoch": 0.9168635252186166,
      "grad_norm": 0.40236085653305054,
      "learning_rate": 4.1588959343528537e-07,
      "loss": 0.6366,
      "step": 22123
    },
    {
      "epoch": 0.9169049691242903,
      "grad_norm": 0.47887974977493286,
      "learning_rate": 4.1568237390691707e-07,
      "loss": 0.6954,
      "step": 22124
    },
    {
      "epoch": 0.9169464130299639,
      "grad_norm": 0.4403918981552124,
      "learning_rate": 4.1547515437854867e-07,
      "loss": 0.6582,
      "step": 22125
    },
    {
      "epoch": 0.9169878569356377,
      "grad_norm": 0.40729039907455444,
      "learning_rate": 4.1526793485018027e-07,
      "loss": 0.6371,
      "step": 22126
    },
    {
      "epoch": 0.9170293008413113,
      "grad_norm": 0.4006808400154114,
      "learning_rate": 4.1506071532181197e-07,
      "loss": 0.6934,
      "step": 22127
    },
    {
      "epoch": 0.9170707447469849,
      "grad_norm": 0.4384099543094635,
      "learning_rate": 4.1485349579344357e-07,
      "loss": 0.687,
      "step": 22128
    },
    {
      "epoch": 0.9171121886526586,
      "grad_norm": 0.4215103089809418,
      "learning_rate": 4.1464627626507527e-07,
      "loss": 0.6768,
      "step": 22129
    },
    {
      "epoch": 0.9171536325583323,
      "grad_norm": 0.43386855721473694,
      "learning_rate": 4.1443905673670687e-07,
      "loss": 0.6904,
      "step": 22130
    },
    {
      "epoch": 0.917195076464006,
      "grad_norm": 0.4184907078742981,
      "learning_rate": 4.1423183720833857e-07,
      "loss": 0.6757,
      "step": 22131
    },
    {
      "epoch": 0.9172365203696796,
      "grad_norm": 0.3844473958015442,
      "learning_rate": 4.1402461767997017e-07,
      "loss": 0.6368,
      "step": 22132
    },
    {
      "epoch": 0.9172779642753534,
      "grad_norm": 0.418864905834198,
      "learning_rate": 4.1381739815160187e-07,
      "loss": 0.6106,
      "step": 22133
    },
    {
      "epoch": 0.917319408181027,
      "grad_norm": 0.4102354347705841,
      "learning_rate": 4.1361017862323347e-07,
      "loss": 0.6317,
      "step": 22134
    },
    {
      "epoch": 0.9173608520867006,
      "grad_norm": 0.41153115034103394,
      "learning_rate": 4.1340295909486517e-07,
      "loss": 0.665,
      "step": 22135
    },
    {
      "epoch": 0.9174022959923743,
      "grad_norm": 0.41083332896232605,
      "learning_rate": 4.1319573956649677e-07,
      "loss": 0.6643,
      "step": 22136
    },
    {
      "epoch": 0.917443739898048,
      "grad_norm": 0.4087878465652466,
      "learning_rate": 4.129885200381284e-07,
      "loss": 0.6724,
      "step": 22137
    },
    {
      "epoch": 0.9174851838037217,
      "grad_norm": 0.4624066650867462,
      "learning_rate": 4.1278130050976007e-07,
      "loss": 0.7079,
      "step": 22138
    },
    {
      "epoch": 0.9175266277093953,
      "grad_norm": 0.40069225430488586,
      "learning_rate": 4.1257408098139167e-07,
      "loss": 0.6512,
      "step": 22139
    },
    {
      "epoch": 0.917568071615069,
      "grad_norm": 0.4340575933456421,
      "learning_rate": 4.1236686145302337e-07,
      "loss": 0.7229,
      "step": 22140
    },
    {
      "epoch": 0.9176095155207427,
      "grad_norm": 0.43180498480796814,
      "learning_rate": 4.1215964192465497e-07,
      "loss": 0.6735,
      "step": 22141
    },
    {
      "epoch": 0.9176509594264164,
      "grad_norm": 0.45644280314445496,
      "learning_rate": 4.1195242239628667e-07,
      "loss": 0.6956,
      "step": 22142
    },
    {
      "epoch": 0.91769240333209,
      "grad_norm": 0.4217568039894104,
      "learning_rate": 4.1174520286791827e-07,
      "loss": 0.6978,
      "step": 22143
    },
    {
      "epoch": 0.9177338472377636,
      "grad_norm": 0.40324866771698,
      "learning_rate": 4.1153798333954997e-07,
      "loss": 0.662,
      "step": 22144
    },
    {
      "epoch": 0.9177752911434374,
      "grad_norm": 0.4438250958919525,
      "learning_rate": 4.1133076381118157e-07,
      "loss": 0.6971,
      "step": 22145
    },
    {
      "epoch": 0.917816735049111,
      "grad_norm": 0.4567849040031433,
      "learning_rate": 4.1112354428281327e-07,
      "loss": 0.6703,
      "step": 22146
    },
    {
      "epoch": 0.9178581789547847,
      "grad_norm": 0.4421851634979248,
      "learning_rate": 4.1091632475444487e-07,
      "loss": 0.6924,
      "step": 22147
    },
    {
      "epoch": 0.9178996228604583,
      "grad_norm": 0.37911370396614075,
      "learning_rate": 4.1070910522607657e-07,
      "loss": 0.6204,
      "step": 22148
    },
    {
      "epoch": 0.9179410667661321,
      "grad_norm": 0.3992007076740265,
      "learning_rate": 4.1050188569770817e-07,
      "loss": 0.6184,
      "step": 22149
    },
    {
      "epoch": 0.9179825106718057,
      "grad_norm": 0.370268315076828,
      "learning_rate": 4.102946661693398e-07,
      "loss": 0.618,
      "step": 22150
    },
    {
      "epoch": 0.9180239545774794,
      "grad_norm": 0.45464763045310974,
      "learning_rate": 4.1008744664097147e-07,
      "loss": 0.6882,
      "step": 22151
    },
    {
      "epoch": 0.918065398483153,
      "grad_norm": 0.3815627694129944,
      "learning_rate": 4.098802271126031e-07,
      "loss": 0.6111,
      "step": 22152
    },
    {
      "epoch": 0.9181068423888267,
      "grad_norm": 0.42523372173309326,
      "learning_rate": 4.0967300758423477e-07,
      "loss": 0.6437,
      "step": 22153
    },
    {
      "epoch": 0.9181482862945004,
      "grad_norm": 0.41412872076034546,
      "learning_rate": 4.094657880558664e-07,
      "loss": 0.6329,
      "step": 22154
    },
    {
      "epoch": 0.918189730200174,
      "grad_norm": 0.43132030963897705,
      "learning_rate": 4.0925856852749807e-07,
      "loss": 0.6401,
      "step": 22155
    },
    {
      "epoch": 0.9182311741058478,
      "grad_norm": 0.4029728174209595,
      "learning_rate": 4.0905134899912967e-07,
      "loss": 0.62,
      "step": 22156
    },
    {
      "epoch": 0.9182726180115214,
      "grad_norm": 0.5114448070526123,
      "learning_rate": 4.088441294707614e-07,
      "loss": 0.6687,
      "step": 22157
    },
    {
      "epoch": 0.9183140619171951,
      "grad_norm": 0.3804416060447693,
      "learning_rate": 4.0863690994239297e-07,
      "loss": 0.5996,
      "step": 22158
    },
    {
      "epoch": 0.9183555058228687,
      "grad_norm": 0.41832447052001953,
      "learning_rate": 4.084296904140247e-07,
      "loss": 0.6686,
      "step": 22159
    },
    {
      "epoch": 0.9183969497285425,
      "grad_norm": 0.3986947536468506,
      "learning_rate": 4.0822247088565627e-07,
      "loss": 0.6166,
      "step": 22160
    },
    {
      "epoch": 0.9184383936342161,
      "grad_norm": 0.44590428471565247,
      "learning_rate": 4.08015251357288e-07,
      "loss": 0.6731,
      "step": 22161
    },
    {
      "epoch": 0.9184798375398897,
      "grad_norm": 0.4256608188152313,
      "learning_rate": 4.0780803182891957e-07,
      "loss": 0.6646,
      "step": 22162
    },
    {
      "epoch": 0.9185212814455634,
      "grad_norm": 0.42692098021507263,
      "learning_rate": 4.076008123005512e-07,
      "loss": 0.6296,
      "step": 22163
    },
    {
      "epoch": 0.9185627253512371,
      "grad_norm": 0.4565250873565674,
      "learning_rate": 4.0739359277218287e-07,
      "loss": 0.7126,
      "step": 22164
    },
    {
      "epoch": 0.9186041692569108,
      "grad_norm": 0.39262405037879944,
      "learning_rate": 4.071863732438145e-07,
      "loss": 0.6113,
      "step": 22165
    },
    {
      "epoch": 0.9186456131625844,
      "grad_norm": 0.4332748055458069,
      "learning_rate": 4.069791537154462e-07,
      "loss": 0.6444,
      "step": 22166
    },
    {
      "epoch": 0.9186870570682582,
      "grad_norm": 0.4202212989330292,
      "learning_rate": 4.067719341870778e-07,
      "loss": 0.6565,
      "step": 22167
    },
    {
      "epoch": 0.9187285009739318,
      "grad_norm": 0.4355592131614685,
      "learning_rate": 4.065647146587095e-07,
      "loss": 0.6653,
      "step": 22168
    },
    {
      "epoch": 0.9187699448796055,
      "grad_norm": 0.4152291715145111,
      "learning_rate": 4.063574951303411e-07,
      "loss": 0.6592,
      "step": 22169
    },
    {
      "epoch": 0.9188113887852791,
      "grad_norm": 0.4618321657180786,
      "learning_rate": 4.061502756019728e-07,
      "loss": 0.7183,
      "step": 22170
    },
    {
      "epoch": 0.9188528326909527,
      "grad_norm": 0.44260236620903015,
      "learning_rate": 4.059430560736044e-07,
      "loss": 0.6848,
      "step": 22171
    },
    {
      "epoch": 0.9188942765966265,
      "grad_norm": 0.4012022912502289,
      "learning_rate": 4.057358365452361e-07,
      "loss": 0.5914,
      "step": 22172
    },
    {
      "epoch": 0.9189357205023001,
      "grad_norm": 0.4166465997695923,
      "learning_rate": 4.055286170168677e-07,
      "loss": 0.6831,
      "step": 22173
    },
    {
      "epoch": 0.9189771644079738,
      "grad_norm": 0.3947274386882782,
      "learning_rate": 4.053213974884993e-07,
      "loss": 0.6215,
      "step": 22174
    },
    {
      "epoch": 0.9190186083136475,
      "grad_norm": 0.42787227034568787,
      "learning_rate": 4.05114177960131e-07,
      "loss": 0.6644,
      "step": 22175
    },
    {
      "epoch": 0.9190600522193212,
      "grad_norm": 0.4242860674858093,
      "learning_rate": 4.049069584317626e-07,
      "loss": 0.698,
      "step": 22176
    },
    {
      "epoch": 0.9191014961249948,
      "grad_norm": 0.40455371141433716,
      "learning_rate": 4.046997389033943e-07,
      "loss": 0.5961,
      "step": 22177
    },
    {
      "epoch": 0.9191429400306685,
      "grad_norm": 0.46630680561065674,
      "learning_rate": 4.044925193750259e-07,
      "loss": 0.6616,
      "step": 22178
    },
    {
      "epoch": 0.9191843839363422,
      "grad_norm": 0.43062227964401245,
      "learning_rate": 4.042852998466576e-07,
      "loss": 0.661,
      "step": 22179
    },
    {
      "epoch": 0.9192258278420158,
      "grad_norm": 0.4337824285030365,
      "learning_rate": 4.040780803182892e-07,
      "loss": 0.67,
      "step": 22180
    },
    {
      "epoch": 0.9192672717476895,
      "grad_norm": 0.42257601022720337,
      "learning_rate": 4.038708607899209e-07,
      "loss": 0.67,
      "step": 22181
    },
    {
      "epoch": 0.9193087156533631,
      "grad_norm": 0.4307631850242615,
      "learning_rate": 4.0366364126155253e-07,
      "loss": 0.6742,
      "step": 22182
    },
    {
      "epoch": 0.9193501595590369,
      "grad_norm": 0.45422986149787903,
      "learning_rate": 4.034564217331842e-07,
      "loss": 0.6643,
      "step": 22183
    },
    {
      "epoch": 0.9193916034647105,
      "grad_norm": 0.39353176951408386,
      "learning_rate": 4.0324920220481583e-07,
      "loss": 0.6332,
      "step": 22184
    },
    {
      "epoch": 0.9194330473703842,
      "grad_norm": 0.43233948945999146,
      "learning_rate": 4.030419826764475e-07,
      "loss": 0.6454,
      "step": 22185
    },
    {
      "epoch": 0.9194744912760578,
      "grad_norm": 0.398840993642807,
      "learning_rate": 4.0283476314807913e-07,
      "loss": 0.5741,
      "step": 22186
    },
    {
      "epoch": 0.9195159351817316,
      "grad_norm": 0.3804643452167511,
      "learning_rate": 4.026275436197107e-07,
      "loss": 0.6682,
      "step": 22187
    },
    {
      "epoch": 0.9195573790874052,
      "grad_norm": 0.4021351933479309,
      "learning_rate": 4.0242032409134243e-07,
      "loss": 0.688,
      "step": 22188
    },
    {
      "epoch": 0.9195988229930788,
      "grad_norm": 0.41401660442352295,
      "learning_rate": 4.02213104562974e-07,
      "loss": 0.6674,
      "step": 22189
    },
    {
      "epoch": 0.9196402668987526,
      "grad_norm": 0.44052591919898987,
      "learning_rate": 4.0200588503460573e-07,
      "loss": 0.6726,
      "step": 22190
    },
    {
      "epoch": 0.9196817108044262,
      "grad_norm": 0.38475707173347473,
      "learning_rate": 4.0179866550623733e-07,
      "loss": 0.6694,
      "step": 22191
    },
    {
      "epoch": 0.9197231547100999,
      "grad_norm": 0.3973960280418396,
      "learning_rate": 4.0159144597786903e-07,
      "loss": 0.6128,
      "step": 22192
    },
    {
      "epoch": 0.9197645986157735,
      "grad_norm": 0.43617480993270874,
      "learning_rate": 4.0138422644950063e-07,
      "loss": 0.6523,
      "step": 22193
    },
    {
      "epoch": 0.9198060425214473,
      "grad_norm": 0.392958402633667,
      "learning_rate": 4.011770069211323e-07,
      "loss": 0.6201,
      "step": 22194
    },
    {
      "epoch": 0.9198474864271209,
      "grad_norm": 0.3944983184337616,
      "learning_rate": 4.0096978739276393e-07,
      "loss": 0.6851,
      "step": 22195
    },
    {
      "epoch": 0.9198889303327945,
      "grad_norm": 0.40340930223464966,
      "learning_rate": 4.007625678643956e-07,
      "loss": 0.6499,
      "step": 22196
    },
    {
      "epoch": 0.9199303742384682,
      "grad_norm": 0.39751264452934265,
      "learning_rate": 4.0055534833602723e-07,
      "loss": 0.6449,
      "step": 22197
    },
    {
      "epoch": 0.9199718181441419,
      "grad_norm": 0.43647652864456177,
      "learning_rate": 4.003481288076589e-07,
      "loss": 0.6189,
      "step": 22198
    },
    {
      "epoch": 0.9200132620498156,
      "grad_norm": 0.3826138377189636,
      "learning_rate": 4.0014090927929053e-07,
      "loss": 0.6426,
      "step": 22199
    },
    {
      "epoch": 0.9200547059554892,
      "grad_norm": 0.4139733910560608,
      "learning_rate": 3.9993368975092213e-07,
      "loss": 0.6978,
      "step": 22200
    },
    {
      "epoch": 0.920096149861163,
      "grad_norm": 0.4198732078075409,
      "learning_rate": 3.9972647022255383e-07,
      "loss": 0.6327,
      "step": 22201
    },
    {
      "epoch": 0.9201375937668366,
      "grad_norm": 0.3935845196247101,
      "learning_rate": 3.9951925069418543e-07,
      "loss": 0.6676,
      "step": 22202
    },
    {
      "epoch": 0.9201790376725103,
      "grad_norm": 0.4179231524467468,
      "learning_rate": 3.9931203116581713e-07,
      "loss": 0.6917,
      "step": 22203
    },
    {
      "epoch": 0.9202204815781839,
      "grad_norm": 0.4452146887779236,
      "learning_rate": 3.9910481163744873e-07,
      "loss": 0.6738,
      "step": 22204
    },
    {
      "epoch": 0.9202619254838575,
      "grad_norm": 0.43203210830688477,
      "learning_rate": 3.9889759210908043e-07,
      "loss": 0.5864,
      "step": 22205
    },
    {
      "epoch": 0.9203033693895313,
      "grad_norm": 0.37480705976486206,
      "learning_rate": 3.9869037258071203e-07,
      "loss": 0.6183,
      "step": 22206
    },
    {
      "epoch": 0.9203448132952049,
      "grad_norm": 0.4728090465068817,
      "learning_rate": 3.9848315305234373e-07,
      "loss": 0.7705,
      "step": 22207
    },
    {
      "epoch": 0.9203862572008786,
      "grad_norm": 0.41070929169654846,
      "learning_rate": 3.9827593352397533e-07,
      "loss": 0.6298,
      "step": 22208
    },
    {
      "epoch": 0.9204277011065523,
      "grad_norm": 0.4104289412498474,
      "learning_rate": 3.9806871399560703e-07,
      "loss": 0.6788,
      "step": 22209
    },
    {
      "epoch": 0.920469145012226,
      "grad_norm": 0.4289223253726959,
      "learning_rate": 3.9786149446723863e-07,
      "loss": 0.619,
      "step": 22210
    },
    {
      "epoch": 0.9205105889178996,
      "grad_norm": 0.383742094039917,
      "learning_rate": 3.9765427493887023e-07,
      "loss": 0.6434,
      "step": 22211
    },
    {
      "epoch": 0.9205520328235733,
      "grad_norm": 0.42177265882492065,
      "learning_rate": 3.9744705541050193e-07,
      "loss": 0.667,
      "step": 22212
    },
    {
      "epoch": 0.920593476729247,
      "grad_norm": 0.4245764911174774,
      "learning_rate": 3.9723983588213353e-07,
      "loss": 0.7073,
      "step": 22213
    },
    {
      "epoch": 0.9206349206349206,
      "grad_norm": 0.42813989520072937,
      "learning_rate": 3.9703261635376523e-07,
      "loss": 0.6753,
      "step": 22214
    },
    {
      "epoch": 0.9206763645405943,
      "grad_norm": 0.38868942856788635,
      "learning_rate": 3.9682539682539683e-07,
      "loss": 0.5836,
      "step": 22215
    },
    {
      "epoch": 0.9207178084462679,
      "grad_norm": 0.4881787598133087,
      "learning_rate": 3.9661817729702853e-07,
      "loss": 0.741,
      "step": 22216
    },
    {
      "epoch": 0.9207592523519417,
      "grad_norm": 0.4125223457813263,
      "learning_rate": 3.9641095776866013e-07,
      "loss": 0.6808,
      "step": 22217
    },
    {
      "epoch": 0.9208006962576153,
      "grad_norm": 0.43530043959617615,
      "learning_rate": 3.9620373824029183e-07,
      "loss": 0.6553,
      "step": 22218
    },
    {
      "epoch": 0.920842140163289,
      "grad_norm": 0.4123053550720215,
      "learning_rate": 3.9599651871192343e-07,
      "loss": 0.6707,
      "step": 22219
    },
    {
      "epoch": 0.9208835840689626,
      "grad_norm": 0.40143659710884094,
      "learning_rate": 3.9578929918355513e-07,
      "loss": 0.6642,
      "step": 22220
    },
    {
      "epoch": 0.9209250279746364,
      "grad_norm": 0.39907631278038025,
      "learning_rate": 3.9558207965518673e-07,
      "loss": 0.5948,
      "step": 22221
    },
    {
      "epoch": 0.92096647188031,
      "grad_norm": 0.4393535554409027,
      "learning_rate": 3.9537486012681843e-07,
      "loss": 0.6959,
      "step": 22222
    },
    {
      "epoch": 0.9210079157859836,
      "grad_norm": 0.4174802303314209,
      "learning_rate": 3.9516764059845003e-07,
      "loss": 0.6058,
      "step": 22223
    },
    {
      "epoch": 0.9210493596916574,
      "grad_norm": 0.41545554995536804,
      "learning_rate": 3.9496042107008163e-07,
      "loss": 0.7029,
      "step": 22224
    },
    {
      "epoch": 0.921090803597331,
      "grad_norm": 0.3938498795032501,
      "learning_rate": 3.9475320154171333e-07,
      "loss": 0.6531,
      "step": 22225
    },
    {
      "epoch": 0.9211322475030047,
      "grad_norm": 0.44151178002357483,
      "learning_rate": 3.9454598201334493e-07,
      "loss": 0.6904,
      "step": 22226
    },
    {
      "epoch": 0.9211736914086783,
      "grad_norm": 0.4208865463733673,
      "learning_rate": 3.9433876248497663e-07,
      "loss": 0.6453,
      "step": 22227
    },
    {
      "epoch": 0.9212151353143521,
      "grad_norm": 0.4188539385795593,
      "learning_rate": 3.9413154295660823e-07,
      "loss": 0.6448,
      "step": 22228
    },
    {
      "epoch": 0.9212565792200257,
      "grad_norm": 0.37795382738113403,
      "learning_rate": 3.9392432342823993e-07,
      "loss": 0.5927,
      "step": 22229
    },
    {
      "epoch": 0.9212980231256994,
      "grad_norm": 0.4400404095649719,
      "learning_rate": 3.9371710389987153e-07,
      "loss": 0.6738,
      "step": 22230
    },
    {
      "epoch": 0.921339467031373,
      "grad_norm": 0.42634958028793335,
      "learning_rate": 3.9350988437150323e-07,
      "loss": 0.6829,
      "step": 22231
    },
    {
      "epoch": 0.9213809109370467,
      "grad_norm": 0.42551514506340027,
      "learning_rate": 3.9330266484313483e-07,
      "loss": 0.6827,
      "step": 22232
    },
    {
      "epoch": 0.9214223548427204,
      "grad_norm": 0.42764368653297424,
      "learning_rate": 3.9309544531476654e-07,
      "loss": 0.6526,
      "step": 22233
    },
    {
      "epoch": 0.921463798748394,
      "grad_norm": 0.4405936598777771,
      "learning_rate": 3.9288822578639813e-07,
      "loss": 0.6869,
      "step": 22234
    },
    {
      "epoch": 0.9215052426540677,
      "grad_norm": 0.43019893765449524,
      "learning_rate": 3.9268100625802984e-07,
      "loss": 0.6882,
      "step": 22235
    },
    {
      "epoch": 0.9215466865597414,
      "grad_norm": 0.42577120661735535,
      "learning_rate": 3.9247378672966143e-07,
      "loss": 0.6783,
      "step": 22236
    },
    {
      "epoch": 0.9215881304654151,
      "grad_norm": 0.43746110796928406,
      "learning_rate": 3.9226656720129303e-07,
      "loss": 0.7174,
      "step": 22237
    },
    {
      "epoch": 0.9216295743710887,
      "grad_norm": 0.38967365026474,
      "learning_rate": 3.9205934767292473e-07,
      "loss": 0.6376,
      "step": 22238
    },
    {
      "epoch": 0.9216710182767625,
      "grad_norm": 0.4589861035346985,
      "learning_rate": 3.9185212814455633e-07,
      "loss": 0.6965,
      "step": 22239
    },
    {
      "epoch": 0.9217124621824361,
      "grad_norm": 0.4128563106060028,
      "learning_rate": 3.9164490861618804e-07,
      "loss": 0.7007,
      "step": 22240
    },
    {
      "epoch": 0.9217539060881097,
      "grad_norm": 0.43714439868927,
      "learning_rate": 3.9143768908781963e-07,
      "loss": 0.6401,
      "step": 22241
    },
    {
      "epoch": 0.9217953499937834,
      "grad_norm": 0.4154449701309204,
      "learning_rate": 3.9123046955945134e-07,
      "loss": 0.6279,
      "step": 22242
    },
    {
      "epoch": 0.921836793899457,
      "grad_norm": 0.4214541018009186,
      "learning_rate": 3.9102325003108293e-07,
      "loss": 0.6663,
      "step": 22243
    },
    {
      "epoch": 0.9218782378051308,
      "grad_norm": 0.3632116913795471,
      "learning_rate": 3.9081603050271464e-07,
      "loss": 0.6108,
      "step": 22244
    },
    {
      "epoch": 0.9219196817108044,
      "grad_norm": 0.421127051115036,
      "learning_rate": 3.9060881097434623e-07,
      "loss": 0.646,
      "step": 22245
    },
    {
      "epoch": 0.9219611256164781,
      "grad_norm": 0.4117325246334076,
      "learning_rate": 3.9040159144597794e-07,
      "loss": 0.6608,
      "step": 22246
    },
    {
      "epoch": 0.9220025695221518,
      "grad_norm": 0.37873950600624084,
      "learning_rate": 3.9019437191760953e-07,
      "loss": 0.6443,
      "step": 22247
    },
    {
      "epoch": 0.9220440134278255,
      "grad_norm": 0.3980756998062134,
      "learning_rate": 3.899871523892412e-07,
      "loss": 0.6646,
      "step": 22248
    },
    {
      "epoch": 0.9220854573334991,
      "grad_norm": 0.40306130051612854,
      "learning_rate": 3.8977993286087284e-07,
      "loss": 0.6427,
      "step": 22249
    },
    {
      "epoch": 0.9221269012391727,
      "grad_norm": 0.40291473269462585,
      "learning_rate": 3.895727133325045e-07,
      "loss": 0.6312,
      "step": 22250
    },
    {
      "epoch": 0.9221683451448465,
      "grad_norm": 0.4417590796947479,
      "learning_rate": 3.8936549380413614e-07,
      "loss": 0.6606,
      "step": 22251
    },
    {
      "epoch": 0.9222097890505201,
      "grad_norm": 0.4224981963634491,
      "learning_rate": 3.8915827427576773e-07,
      "loss": 0.6362,
      "step": 22252
    },
    {
      "epoch": 0.9222512329561938,
      "grad_norm": 0.4432431757450104,
      "learning_rate": 3.8895105474739944e-07,
      "loss": 0.6736,
      "step": 22253
    },
    {
      "epoch": 0.9222926768618674,
      "grad_norm": 0.4231889843940735,
      "learning_rate": 3.8874383521903103e-07,
      "loss": 0.6399,
      "step": 22254
    },
    {
      "epoch": 0.9223341207675412,
      "grad_norm": 0.3831135332584381,
      "learning_rate": 3.8853661569066274e-07,
      "loss": 0.646,
      "step": 22255
    },
    {
      "epoch": 0.9223755646732148,
      "grad_norm": 0.41470733284950256,
      "learning_rate": 3.8832939616229433e-07,
      "loss": 0.614,
      "step": 22256
    },
    {
      "epoch": 0.9224170085788884,
      "grad_norm": 0.4269106388092041,
      "learning_rate": 3.8812217663392604e-07,
      "loss": 0.7148,
      "step": 22257
    },
    {
      "epoch": 0.9224584524845622,
      "grad_norm": 0.43638163805007935,
      "learning_rate": 3.8791495710555764e-07,
      "loss": 0.6493,
      "step": 22258
    },
    {
      "epoch": 0.9224998963902358,
      "grad_norm": 0.45652151107788086,
      "learning_rate": 3.8770773757718934e-07,
      "loss": 0.6677,
      "step": 22259
    },
    {
      "epoch": 0.9225413402959095,
      "grad_norm": 0.434817910194397,
      "learning_rate": 3.8750051804882094e-07,
      "loss": 0.6677,
      "step": 22260
    },
    {
      "epoch": 0.9225827842015831,
      "grad_norm": 0.40999308228492737,
      "learning_rate": 3.872932985204526e-07,
      "loss": 0.7249,
      "step": 22261
    },
    {
      "epoch": 0.9226242281072569,
      "grad_norm": 0.4012381136417389,
      "learning_rate": 3.8708607899208424e-07,
      "loss": 0.6793,
      "step": 22262
    },
    {
      "epoch": 0.9226656720129305,
      "grad_norm": 0.44949448108673096,
      "learning_rate": 3.868788594637159e-07,
      "loss": 0.6729,
      "step": 22263
    },
    {
      "epoch": 0.9227071159186042,
      "grad_norm": 0.4265710413455963,
      "learning_rate": 3.8667163993534754e-07,
      "loss": 0.6295,
      "step": 22264
    },
    {
      "epoch": 0.9227485598242778,
      "grad_norm": 0.3944624066352844,
      "learning_rate": 3.864644204069792e-07,
      "loss": 0.6619,
      "step": 22265
    },
    {
      "epoch": 0.9227900037299515,
      "grad_norm": 0.4055692255496979,
      "learning_rate": 3.8625720087861084e-07,
      "loss": 0.5896,
      "step": 22266
    },
    {
      "epoch": 0.9228314476356252,
      "grad_norm": 0.4230136573314667,
      "learning_rate": 3.860499813502425e-07,
      "loss": 0.6871,
      "step": 22267
    },
    {
      "epoch": 0.9228728915412988,
      "grad_norm": 0.4314816892147064,
      "learning_rate": 3.8584276182187414e-07,
      "loss": 0.6733,
      "step": 22268
    },
    {
      "epoch": 0.9229143354469725,
      "grad_norm": 0.4521991014480591,
      "learning_rate": 3.8563554229350574e-07,
      "loss": 0.666,
      "step": 22269
    },
    {
      "epoch": 0.9229557793526462,
      "grad_norm": 0.41855400800704956,
      "learning_rate": 3.8542832276513744e-07,
      "loss": 0.7229,
      "step": 22270
    },
    {
      "epoch": 0.9229972232583199,
      "grad_norm": 0.42210933566093445,
      "learning_rate": 3.8522110323676904e-07,
      "loss": 0.71,
      "step": 22271
    },
    {
      "epoch": 0.9230386671639935,
      "grad_norm": 0.38391363620758057,
      "learning_rate": 3.8501388370840074e-07,
      "loss": 0.6051,
      "step": 22272
    },
    {
      "epoch": 0.9230801110696673,
      "grad_norm": 0.4737195372581482,
      "learning_rate": 3.8480666418003234e-07,
      "loss": 0.6715,
      "step": 22273
    },
    {
      "epoch": 0.9231215549753409,
      "grad_norm": 0.4364435374736786,
      "learning_rate": 3.84599444651664e-07,
      "loss": 0.6816,
      "step": 22274
    },
    {
      "epoch": 0.9231629988810145,
      "grad_norm": 0.44413185119628906,
      "learning_rate": 3.8439222512329564e-07,
      "loss": 0.6519,
      "step": 22275
    },
    {
      "epoch": 0.9232044427866882,
      "grad_norm": 0.4837568700313568,
      "learning_rate": 3.841850055949273e-07,
      "loss": 0.6941,
      "step": 22276
    },
    {
      "epoch": 0.9232458866923619,
      "grad_norm": 0.41548749804496765,
      "learning_rate": 3.8397778606655894e-07,
      "loss": 0.6486,
      "step": 22277
    },
    {
      "epoch": 0.9232873305980356,
      "grad_norm": 0.6428281664848328,
      "learning_rate": 3.837705665381906e-07,
      "loss": 0.6951,
      "step": 22278
    },
    {
      "epoch": 0.9233287745037092,
      "grad_norm": 0.4326765239238739,
      "learning_rate": 3.8356334700982224e-07,
      "loss": 0.6753,
      "step": 22279
    },
    {
      "epoch": 0.9233702184093829,
      "grad_norm": 0.4001523554325104,
      "learning_rate": 3.833561274814539e-07,
      "loss": 0.6816,
      "step": 22280
    },
    {
      "epoch": 0.9234116623150566,
      "grad_norm": 0.46318763494491577,
      "learning_rate": 3.8314890795308554e-07,
      "loss": 0.7075,
      "step": 22281
    },
    {
      "epoch": 0.9234531062207303,
      "grad_norm": 0.38921278715133667,
      "learning_rate": 3.829416884247172e-07,
      "loss": 0.605,
      "step": 22282
    },
    {
      "epoch": 0.9234945501264039,
      "grad_norm": 0.4211929142475128,
      "learning_rate": 3.8273446889634884e-07,
      "loss": 0.649,
      "step": 22283
    },
    {
      "epoch": 0.9235359940320775,
      "grad_norm": 0.4073669910430908,
      "learning_rate": 3.825272493679805e-07,
      "loss": 0.6866,
      "step": 22284
    },
    {
      "epoch": 0.9235774379377513,
      "grad_norm": 0.38739991188049316,
      "learning_rate": 3.823200298396121e-07,
      "loss": 0.6279,
      "step": 22285
    },
    {
      "epoch": 0.9236188818434249,
      "grad_norm": 0.39897942543029785,
      "learning_rate": 3.821128103112438e-07,
      "loss": 0.6874,
      "step": 22286
    },
    {
      "epoch": 0.9236603257490986,
      "grad_norm": 0.39754992723464966,
      "learning_rate": 3.819055907828754e-07,
      "loss": 0.6572,
      "step": 22287
    },
    {
      "epoch": 0.9237017696547722,
      "grad_norm": 0.4159892797470093,
      "learning_rate": 3.8169837125450704e-07,
      "loss": 0.672,
      "step": 22288
    },
    {
      "epoch": 0.923743213560446,
      "grad_norm": 0.4716191291809082,
      "learning_rate": 3.814911517261387e-07,
      "loss": 0.7478,
      "step": 22289
    },
    {
      "epoch": 0.9237846574661196,
      "grad_norm": 0.4174673557281494,
      "learning_rate": 3.8128393219777034e-07,
      "loss": 0.6775,
      "step": 22290
    },
    {
      "epoch": 0.9238261013717933,
      "grad_norm": 0.4108074903488159,
      "learning_rate": 3.81076712669402e-07,
      "loss": 0.6827,
      "step": 22291
    },
    {
      "epoch": 0.923867545277467,
      "grad_norm": 0.3985379636287689,
      "learning_rate": 3.8086949314103364e-07,
      "loss": 0.6844,
      "step": 22292
    },
    {
      "epoch": 0.9239089891831406,
      "grad_norm": 0.37764260172843933,
      "learning_rate": 3.806622736126653e-07,
      "loss": 0.6245,
      "step": 22293
    },
    {
      "epoch": 0.9239504330888143,
      "grad_norm": 0.39742356538772583,
      "learning_rate": 3.8045505408429694e-07,
      "loss": 0.6708,
      "step": 22294
    },
    {
      "epoch": 0.9239918769944879,
      "grad_norm": 0.3933504819869995,
      "learning_rate": 3.802478345559286e-07,
      "loss": 0.7126,
      "step": 22295
    },
    {
      "epoch": 0.9240333209001617,
      "grad_norm": 0.4181134104728699,
      "learning_rate": 3.8004061502756024e-07,
      "loss": 0.6335,
      "step": 22296
    },
    {
      "epoch": 0.9240747648058353,
      "grad_norm": 0.43315309286117554,
      "learning_rate": 3.798333954991919e-07,
      "loss": 0.7173,
      "step": 22297
    },
    {
      "epoch": 0.924116208711509,
      "grad_norm": 0.40219634771347046,
      "learning_rate": 3.796261759708235e-07,
      "loss": 0.6688,
      "step": 22298
    },
    {
      "epoch": 0.9241576526171826,
      "grad_norm": 0.41858118772506714,
      "learning_rate": 3.794189564424552e-07,
      "loss": 0.6696,
      "step": 22299
    },
    {
      "epoch": 0.9241990965228564,
      "grad_norm": 0.40735116600990295,
      "learning_rate": 3.792117369140868e-07,
      "loss": 0.6833,
      "step": 22300
    },
    {
      "epoch": 0.92424054042853,
      "grad_norm": 0.4013233482837677,
      "learning_rate": 3.790045173857185e-07,
      "loss": 0.6575,
      "step": 22301
    },
    {
      "epoch": 0.9242819843342036,
      "grad_norm": 0.43694594502449036,
      "learning_rate": 3.787972978573501e-07,
      "loss": 0.7086,
      "step": 22302
    },
    {
      "epoch": 0.9243234282398773,
      "grad_norm": 0.40606066584587097,
      "learning_rate": 3.785900783289818e-07,
      "loss": 0.6672,
      "step": 22303
    },
    {
      "epoch": 0.924364872145551,
      "grad_norm": 0.43133068084716797,
      "learning_rate": 3.783828588006134e-07,
      "loss": 0.6897,
      "step": 22304
    },
    {
      "epoch": 0.9244063160512247,
      "grad_norm": 0.42128878831863403,
      "learning_rate": 3.781756392722451e-07,
      "loss": 0.6545,
      "step": 22305
    },
    {
      "epoch": 0.9244477599568983,
      "grad_norm": 0.4442581534385681,
      "learning_rate": 3.779684197438767e-07,
      "loss": 0.6787,
      "step": 22306
    },
    {
      "epoch": 0.924489203862572,
      "grad_norm": 0.46650707721710205,
      "learning_rate": 3.7776120021550834e-07,
      "loss": 0.6731,
      "step": 22307
    },
    {
      "epoch": 0.9245306477682457,
      "grad_norm": 0.42433956265449524,
      "learning_rate": 3.7755398068714e-07,
      "loss": 0.6501,
      "step": 22308
    },
    {
      "epoch": 0.9245720916739194,
      "grad_norm": 0.44297778606414795,
      "learning_rate": 3.773467611587716e-07,
      "loss": 0.6851,
      "step": 22309
    },
    {
      "epoch": 0.924613535579593,
      "grad_norm": 0.4343014657497406,
      "learning_rate": 3.771395416304033e-07,
      "loss": 0.6331,
      "step": 22310
    },
    {
      "epoch": 0.9246549794852666,
      "grad_norm": 0.41484177112579346,
      "learning_rate": 3.769323221020349e-07,
      "loss": 0.6577,
      "step": 22311
    },
    {
      "epoch": 0.9246964233909404,
      "grad_norm": 0.4072597026824951,
      "learning_rate": 3.767251025736666e-07,
      "loss": 0.7261,
      "step": 22312
    },
    {
      "epoch": 0.924737867296614,
      "grad_norm": 0.40252262353897095,
      "learning_rate": 3.765178830452982e-07,
      "loss": 0.6587,
      "step": 22313
    },
    {
      "epoch": 0.9247793112022877,
      "grad_norm": 0.4436371624469757,
      "learning_rate": 3.763106635169299e-07,
      "loss": 0.6655,
      "step": 22314
    },
    {
      "epoch": 0.9248207551079614,
      "grad_norm": 0.38914090394973755,
      "learning_rate": 3.761034439885615e-07,
      "loss": 0.644,
      "step": 22315
    },
    {
      "epoch": 0.9248621990136351,
      "grad_norm": 0.42281731963157654,
      "learning_rate": 3.758962244601932e-07,
      "loss": 0.6711,
      "step": 22316
    },
    {
      "epoch": 0.9249036429193087,
      "grad_norm": 0.40579429268836975,
      "learning_rate": 3.756890049318248e-07,
      "loss": 0.6565,
      "step": 22317
    },
    {
      "epoch": 0.9249450868249823,
      "grad_norm": 0.4136389493942261,
      "learning_rate": 3.754817854034565e-07,
      "loss": 0.6559,
      "step": 22318
    },
    {
      "epoch": 0.9249865307306561,
      "grad_norm": 0.43421193957328796,
      "learning_rate": 3.752745658750881e-07,
      "loss": 0.6481,
      "step": 22319
    },
    {
      "epoch": 0.9250279746363297,
      "grad_norm": 0.39531201124191284,
      "learning_rate": 3.750673463467198e-07,
      "loss": 0.6343,
      "step": 22320
    },
    {
      "epoch": 0.9250694185420034,
      "grad_norm": 0.4292892813682556,
      "learning_rate": 3.748601268183514e-07,
      "loss": 0.6388,
      "step": 22321
    },
    {
      "epoch": 0.925110862447677,
      "grad_norm": 0.42729225754737854,
      "learning_rate": 3.74652907289983e-07,
      "loss": 0.6355,
      "step": 22322
    },
    {
      "epoch": 0.9251523063533508,
      "grad_norm": 0.3878740072250366,
      "learning_rate": 3.744456877616147e-07,
      "loss": 0.6288,
      "step": 22323
    },
    {
      "epoch": 0.9251937502590244,
      "grad_norm": 0.4222938120365143,
      "learning_rate": 3.742384682332463e-07,
      "loss": 0.7173,
      "step": 22324
    },
    {
      "epoch": 0.9252351941646981,
      "grad_norm": 0.45415201783180237,
      "learning_rate": 3.74031248704878e-07,
      "loss": 0.6831,
      "step": 22325
    },
    {
      "epoch": 0.9252766380703717,
      "grad_norm": 0.43415573239326477,
      "learning_rate": 3.738240291765096e-07,
      "loss": 0.6713,
      "step": 22326
    },
    {
      "epoch": 0.9253180819760454,
      "grad_norm": 0.4511180818080902,
      "learning_rate": 3.736168096481413e-07,
      "loss": 0.7261,
      "step": 22327
    },
    {
      "epoch": 0.9253595258817191,
      "grad_norm": 0.39859965443611145,
      "learning_rate": 3.734095901197729e-07,
      "loss": 0.6218,
      "step": 22328
    },
    {
      "epoch": 0.9254009697873927,
      "grad_norm": 0.415436327457428,
      "learning_rate": 3.732023705914046e-07,
      "loss": 0.6366,
      "step": 22329
    },
    {
      "epoch": 0.9254424136930665,
      "grad_norm": 0.4083605110645294,
      "learning_rate": 3.729951510630362e-07,
      "loss": 0.6324,
      "step": 22330
    },
    {
      "epoch": 0.9254838575987401,
      "grad_norm": 0.396239310503006,
      "learning_rate": 3.727879315346679e-07,
      "loss": 0.6146,
      "step": 22331
    },
    {
      "epoch": 0.9255253015044138,
      "grad_norm": 0.4331321120262146,
      "learning_rate": 3.725807120062995e-07,
      "loss": 0.6821,
      "step": 22332
    },
    {
      "epoch": 0.9255667454100874,
      "grad_norm": 0.4008884131908417,
      "learning_rate": 3.723734924779312e-07,
      "loss": 0.6613,
      "step": 22333
    },
    {
      "epoch": 0.9256081893157612,
      "grad_norm": 0.432350218296051,
      "learning_rate": 3.721662729495628e-07,
      "loss": 0.7405,
      "step": 22334
    },
    {
      "epoch": 0.9256496332214348,
      "grad_norm": 0.4215299189090729,
      "learning_rate": 3.719590534211944e-07,
      "loss": 0.6714,
      "step": 22335
    },
    {
      "epoch": 0.9256910771271084,
      "grad_norm": 0.41322043538093567,
      "learning_rate": 3.717518338928261e-07,
      "loss": 0.6182,
      "step": 22336
    },
    {
      "epoch": 0.9257325210327821,
      "grad_norm": 0.4187880754470825,
      "learning_rate": 3.715446143644577e-07,
      "loss": 0.6825,
      "step": 22337
    },
    {
      "epoch": 0.9257739649384558,
      "grad_norm": 0.39930951595306396,
      "learning_rate": 3.713373948360894e-07,
      "loss": 0.666,
      "step": 22338
    },
    {
      "epoch": 0.9258154088441295,
      "grad_norm": 0.4128972589969635,
      "learning_rate": 3.71130175307721e-07,
      "loss": 0.6642,
      "step": 22339
    },
    {
      "epoch": 0.9258568527498031,
      "grad_norm": 0.4360491633415222,
      "learning_rate": 3.709229557793527e-07,
      "loss": 0.6334,
      "step": 22340
    },
    {
      "epoch": 0.9258982966554768,
      "grad_norm": 0.4114395081996918,
      "learning_rate": 3.707157362509843e-07,
      "loss": 0.6855,
      "step": 22341
    },
    {
      "epoch": 0.9259397405611505,
      "grad_norm": 0.4181136190891266,
      "learning_rate": 3.70508516722616e-07,
      "loss": 0.6924,
      "step": 22342
    },
    {
      "epoch": 0.9259811844668242,
      "grad_norm": 0.42109012603759766,
      "learning_rate": 3.703012971942476e-07,
      "loss": 0.6553,
      "step": 22343
    },
    {
      "epoch": 0.9260226283724978,
      "grad_norm": 0.4225921630859375,
      "learning_rate": 3.700940776658793e-07,
      "loss": 0.6582,
      "step": 22344
    },
    {
      "epoch": 0.9260640722781714,
      "grad_norm": 0.4129653871059418,
      "learning_rate": 3.698868581375109e-07,
      "loss": 0.6176,
      "step": 22345
    },
    {
      "epoch": 0.9261055161838452,
      "grad_norm": 0.4387122392654419,
      "learning_rate": 3.696796386091425e-07,
      "loss": 0.7214,
      "step": 22346
    },
    {
      "epoch": 0.9261469600895188,
      "grad_norm": 0.4470447301864624,
      "learning_rate": 3.694724190807742e-07,
      "loss": 0.6636,
      "step": 22347
    },
    {
      "epoch": 0.9261884039951925,
      "grad_norm": 0.4072017967700958,
      "learning_rate": 3.692651995524058e-07,
      "loss": 0.682,
      "step": 22348
    },
    {
      "epoch": 0.9262298479008662,
      "grad_norm": 0.4034588038921356,
      "learning_rate": 3.690579800240375e-07,
      "loss": 0.6029,
      "step": 22349
    },
    {
      "epoch": 0.9262712918065399,
      "grad_norm": 0.44186776876449585,
      "learning_rate": 3.688507604956691e-07,
      "loss": 0.7224,
      "step": 22350
    },
    {
      "epoch": 0.9263127357122135,
      "grad_norm": 0.39881622791290283,
      "learning_rate": 3.686435409673008e-07,
      "loss": 0.6754,
      "step": 22351
    },
    {
      "epoch": 0.9263541796178872,
      "grad_norm": 0.43790096044540405,
      "learning_rate": 3.684363214389324e-07,
      "loss": 0.6663,
      "step": 22352
    },
    {
      "epoch": 0.9263956235235609,
      "grad_norm": 0.3885153830051422,
      "learning_rate": 3.682291019105641e-07,
      "loss": 0.6702,
      "step": 22353
    },
    {
      "epoch": 0.9264370674292345,
      "grad_norm": 0.4624767005443573,
      "learning_rate": 3.680218823821957e-07,
      "loss": 0.6471,
      "step": 22354
    },
    {
      "epoch": 0.9264785113349082,
      "grad_norm": 0.42617201805114746,
      "learning_rate": 3.678146628538274e-07,
      "loss": 0.6804,
      "step": 22355
    },
    {
      "epoch": 0.9265199552405818,
      "grad_norm": 0.41617870330810547,
      "learning_rate": 3.67607443325459e-07,
      "loss": 0.6208,
      "step": 22356
    },
    {
      "epoch": 0.9265613991462556,
      "grad_norm": 0.3823138475418091,
      "learning_rate": 3.674002237970907e-07,
      "loss": 0.6506,
      "step": 22357
    },
    {
      "epoch": 0.9266028430519292,
      "grad_norm": 0.40782982110977173,
      "learning_rate": 3.671930042687223e-07,
      "loss": 0.6117,
      "step": 22358
    },
    {
      "epoch": 0.9266442869576029,
      "grad_norm": 0.4224918484687805,
      "learning_rate": 3.6698578474035395e-07,
      "loss": 0.6823,
      "step": 22359
    },
    {
      "epoch": 0.9266857308632765,
      "grad_norm": 0.4176846146583557,
      "learning_rate": 3.667785652119856e-07,
      "loss": 0.6891,
      "step": 22360
    },
    {
      "epoch": 0.9267271747689503,
      "grad_norm": 0.4424492418766022,
      "learning_rate": 3.6657134568361725e-07,
      "loss": 0.7213,
      "step": 22361
    },
    {
      "epoch": 0.9267686186746239,
      "grad_norm": 0.40268415212631226,
      "learning_rate": 3.663641261552489e-07,
      "loss": 0.6346,
      "step": 22362
    },
    {
      "epoch": 0.9268100625802975,
      "grad_norm": 0.4163341224193573,
      "learning_rate": 3.6615690662688055e-07,
      "loss": 0.6362,
      "step": 22363
    },
    {
      "epoch": 0.9268515064859713,
      "grad_norm": 0.4012050926685333,
      "learning_rate": 3.659496870985122e-07,
      "loss": 0.6571,
      "step": 22364
    },
    {
      "epoch": 0.9268929503916449,
      "grad_norm": 0.42011502385139465,
      "learning_rate": 3.657424675701438e-07,
      "loss": 0.6448,
      "step": 22365
    },
    {
      "epoch": 0.9269343942973186,
      "grad_norm": 0.3702411353588104,
      "learning_rate": 3.655352480417755e-07,
      "loss": 0.5994,
      "step": 22366
    },
    {
      "epoch": 0.9269758382029922,
      "grad_norm": 0.4240339696407318,
      "learning_rate": 3.653280285134071e-07,
      "loss": 0.6565,
      "step": 22367
    },
    {
      "epoch": 0.927017282108666,
      "grad_norm": 0.4463580846786499,
      "learning_rate": 3.651208089850388e-07,
      "loss": 0.6693,
      "step": 22368
    },
    {
      "epoch": 0.9270587260143396,
      "grad_norm": 0.4004729092121124,
      "learning_rate": 3.649135894566704e-07,
      "loss": 0.6614,
      "step": 22369
    },
    {
      "epoch": 0.9271001699200133,
      "grad_norm": 0.39380308985710144,
      "learning_rate": 3.647063699283021e-07,
      "loss": 0.6135,
      "step": 22370
    },
    {
      "epoch": 0.9271416138256869,
      "grad_norm": 0.4014785587787628,
      "learning_rate": 3.644991503999337e-07,
      "loss": 0.6014,
      "step": 22371
    },
    {
      "epoch": 0.9271830577313606,
      "grad_norm": 0.44208085536956787,
      "learning_rate": 3.6429193087156535e-07,
      "loss": 0.6417,
      "step": 22372
    },
    {
      "epoch": 0.9272245016370343,
      "grad_norm": 0.3795013725757599,
      "learning_rate": 3.64084711343197e-07,
      "loss": 0.6302,
      "step": 22373
    },
    {
      "epoch": 0.9272659455427079,
      "grad_norm": 0.42747366428375244,
      "learning_rate": 3.6387749181482865e-07,
      "loss": 0.6625,
      "step": 22374
    },
    {
      "epoch": 0.9273073894483816,
      "grad_norm": 0.4012719690799713,
      "learning_rate": 3.636702722864603e-07,
      "loss": 0.6661,
      "step": 22375
    },
    {
      "epoch": 0.9273488333540553,
      "grad_norm": 0.41593748331069946,
      "learning_rate": 3.6346305275809195e-07,
      "loss": 0.6858,
      "step": 22376
    },
    {
      "epoch": 0.927390277259729,
      "grad_norm": 0.4109537899494171,
      "learning_rate": 3.632558332297236e-07,
      "loss": 0.6841,
      "step": 22377
    },
    {
      "epoch": 0.9274317211654026,
      "grad_norm": 0.42824918031692505,
      "learning_rate": 3.6304861370135525e-07,
      "loss": 0.7009,
      "step": 22378
    },
    {
      "epoch": 0.9274731650710762,
      "grad_norm": 0.4123459756374359,
      "learning_rate": 3.628413941729869e-07,
      "loss": 0.6736,
      "step": 22379
    },
    {
      "epoch": 0.92751460897675,
      "grad_norm": 0.45608964562416077,
      "learning_rate": 3.6263417464461855e-07,
      "loss": 0.6521,
      "step": 22380
    },
    {
      "epoch": 0.9275560528824236,
      "grad_norm": 0.5152004361152649,
      "learning_rate": 3.624269551162502e-07,
      "loss": 0.6631,
      "step": 22381
    },
    {
      "epoch": 0.9275974967880973,
      "grad_norm": 0.4209660589694977,
      "learning_rate": 3.6221973558788186e-07,
      "loss": 0.6982,
      "step": 22382
    },
    {
      "epoch": 0.927638940693771,
      "grad_norm": 0.42127805948257446,
      "learning_rate": 3.6201251605951345e-07,
      "loss": 0.6545,
      "step": 22383
    },
    {
      "epoch": 0.9276803845994447,
      "grad_norm": 0.4284980595111847,
      "learning_rate": 3.618052965311451e-07,
      "loss": 0.6825,
      "step": 22384
    },
    {
      "epoch": 0.9277218285051183,
      "grad_norm": 0.40579432249069214,
      "learning_rate": 3.6159807700277675e-07,
      "loss": 0.7092,
      "step": 22385
    },
    {
      "epoch": 0.927763272410792,
      "grad_norm": 0.40481293201446533,
      "learning_rate": 3.613908574744084e-07,
      "loss": 0.6311,
      "step": 22386
    },
    {
      "epoch": 0.9278047163164657,
      "grad_norm": 0.4197527766227722,
      "learning_rate": 3.6118363794604005e-07,
      "loss": 0.631,
      "step": 22387
    },
    {
      "epoch": 0.9278461602221393,
      "grad_norm": 0.42228105664253235,
      "learning_rate": 3.609764184176717e-07,
      "loss": 0.7075,
      "step": 22388
    },
    {
      "epoch": 0.927887604127813,
      "grad_norm": 0.38583412766456604,
      "learning_rate": 3.6076919888930336e-07,
      "loss": 0.6112,
      "step": 22389
    },
    {
      "epoch": 0.9279290480334866,
      "grad_norm": 0.42533639073371887,
      "learning_rate": 3.60561979360935e-07,
      "loss": 0.658,
      "step": 22390
    },
    {
      "epoch": 0.9279704919391604,
      "grad_norm": 0.4376464784145355,
      "learning_rate": 3.6035475983256666e-07,
      "loss": 0.6604,
      "step": 22391
    },
    {
      "epoch": 0.928011935844834,
      "grad_norm": 0.4134712517261505,
      "learning_rate": 3.601475403041983e-07,
      "loss": 0.6365,
      "step": 22392
    },
    {
      "epoch": 0.9280533797505077,
      "grad_norm": 0.4058242440223694,
      "learning_rate": 3.5994032077582996e-07,
      "loss": 0.6287,
      "step": 22393
    },
    {
      "epoch": 0.9280948236561813,
      "grad_norm": 0.40809330344200134,
      "learning_rate": 3.597331012474616e-07,
      "loss": 0.6388,
      "step": 22394
    },
    {
      "epoch": 0.9281362675618551,
      "grad_norm": 0.441999614238739,
      "learning_rate": 3.5952588171909326e-07,
      "loss": 0.7515,
      "step": 22395
    },
    {
      "epoch": 0.9281777114675287,
      "grad_norm": 0.4518662095069885,
      "learning_rate": 3.5931866219072485e-07,
      "loss": 0.6694,
      "step": 22396
    },
    {
      "epoch": 0.9282191553732023,
      "grad_norm": 0.4308764338493347,
      "learning_rate": 3.5911144266235656e-07,
      "loss": 0.6072,
      "step": 22397
    },
    {
      "epoch": 0.928260599278876,
      "grad_norm": 0.3827759623527527,
      "learning_rate": 3.5890422313398816e-07,
      "loss": 0.6428,
      "step": 22398
    },
    {
      "epoch": 0.9283020431845497,
      "grad_norm": 0.44304701685905457,
      "learning_rate": 3.5869700360561986e-07,
      "loss": 0.7048,
      "step": 22399
    },
    {
      "epoch": 0.9283434870902234,
      "grad_norm": 0.4073714017868042,
      "learning_rate": 3.5848978407725146e-07,
      "loss": 0.6301,
      "step": 22400
    },
    {
      "epoch": 0.928384930995897,
      "grad_norm": 0.4313697814941406,
      "learning_rate": 3.582825645488831e-07,
      "loss": 0.6672,
      "step": 22401
    },
    {
      "epoch": 0.9284263749015708,
      "grad_norm": 0.4722842574119568,
      "learning_rate": 3.5807534502051476e-07,
      "loss": 0.7222,
      "step": 22402
    },
    {
      "epoch": 0.9284678188072444,
      "grad_norm": 0.44513779878616333,
      "learning_rate": 3.578681254921464e-07,
      "loss": 0.6284,
      "step": 22403
    },
    {
      "epoch": 0.9285092627129181,
      "grad_norm": 0.4363020062446594,
      "learning_rate": 3.5766090596377806e-07,
      "loss": 0.6954,
      "step": 22404
    },
    {
      "epoch": 0.9285507066185917,
      "grad_norm": 0.45416370034217834,
      "learning_rate": 3.574536864354097e-07,
      "loss": 0.6967,
      "step": 22405
    },
    {
      "epoch": 0.9285921505242654,
      "grad_norm": 0.41454970836639404,
      "learning_rate": 3.5724646690704136e-07,
      "loss": 0.6663,
      "step": 22406
    },
    {
      "epoch": 0.9286335944299391,
      "grad_norm": 0.42112571001052856,
      "learning_rate": 3.57039247378673e-07,
      "loss": 0.665,
      "step": 22407
    },
    {
      "epoch": 0.9286750383356127,
      "grad_norm": 0.461509644985199,
      "learning_rate": 3.5683202785030466e-07,
      "loss": 0.662,
      "step": 22408
    },
    {
      "epoch": 0.9287164822412864,
      "grad_norm": 0.4556443393230438,
      "learning_rate": 3.5662480832193626e-07,
      "loss": 0.708,
      "step": 22409
    },
    {
      "epoch": 0.9287579261469601,
      "grad_norm": 0.4383894205093384,
      "learning_rate": 3.5641758879356796e-07,
      "loss": 0.6902,
      "step": 22410
    },
    {
      "epoch": 0.9287993700526338,
      "grad_norm": 0.43385639786720276,
      "learning_rate": 3.5621036926519956e-07,
      "loss": 0.7045,
      "step": 22411
    },
    {
      "epoch": 0.9288408139583074,
      "grad_norm": 0.4054213762283325,
      "learning_rate": 3.5600314973683126e-07,
      "loss": 0.7266,
      "step": 22412
    },
    {
      "epoch": 0.9288822578639812,
      "grad_norm": 0.4278903603553772,
      "learning_rate": 3.5579593020846286e-07,
      "loss": 0.6786,
      "step": 22413
    },
    {
      "epoch": 0.9289237017696548,
      "grad_norm": 0.4777695834636688,
      "learning_rate": 3.5558871068009456e-07,
      "loss": 0.6865,
      "step": 22414
    },
    {
      "epoch": 0.9289651456753284,
      "grad_norm": 0.4187951385974884,
      "learning_rate": 3.5538149115172616e-07,
      "loss": 0.6267,
      "step": 22415
    },
    {
      "epoch": 0.9290065895810021,
      "grad_norm": 0.45424923300743103,
      "learning_rate": 3.5517427162335786e-07,
      "loss": 0.7108,
      "step": 22416
    },
    {
      "epoch": 0.9290480334866758,
      "grad_norm": 0.4252395033836365,
      "learning_rate": 3.5496705209498946e-07,
      "loss": 0.6643,
      "step": 22417
    },
    {
      "epoch": 0.9290894773923495,
      "grad_norm": 0.41241398453712463,
      "learning_rate": 3.5475983256662116e-07,
      "loss": 0.6526,
      "step": 22418
    },
    {
      "epoch": 0.9291309212980231,
      "grad_norm": 0.38142287731170654,
      "learning_rate": 3.5455261303825276e-07,
      "loss": 0.6444,
      "step": 22419
    },
    {
      "epoch": 0.9291723652036968,
      "grad_norm": 0.43044668436050415,
      "learning_rate": 3.5434539350988436e-07,
      "loss": 0.6615,
      "step": 22420
    },
    {
      "epoch": 0.9292138091093705,
      "grad_norm": 0.39762264490127563,
      "learning_rate": 3.5413817398151606e-07,
      "loss": 0.6707,
      "step": 22421
    },
    {
      "epoch": 0.9292552530150442,
      "grad_norm": 0.4123538136482239,
      "learning_rate": 3.5393095445314766e-07,
      "loss": 0.7273,
      "step": 22422
    },
    {
      "epoch": 0.9292966969207178,
      "grad_norm": 0.4516674876213074,
      "learning_rate": 3.5372373492477936e-07,
      "loss": 0.6775,
      "step": 22423
    },
    {
      "epoch": 0.9293381408263914,
      "grad_norm": 0.3963054418563843,
      "learning_rate": 3.5351651539641096e-07,
      "loss": 0.6466,
      "step": 22424
    },
    {
      "epoch": 0.9293795847320652,
      "grad_norm": 0.38929614424705505,
      "learning_rate": 3.5330929586804266e-07,
      "loss": 0.6365,
      "step": 22425
    },
    {
      "epoch": 0.9294210286377388,
      "grad_norm": 0.41165411472320557,
      "learning_rate": 3.5310207633967426e-07,
      "loss": 0.6729,
      "step": 22426
    },
    {
      "epoch": 0.9294624725434125,
      "grad_norm": 0.42684558033943176,
      "learning_rate": 3.5289485681130596e-07,
      "loss": 0.6539,
      "step": 22427
    },
    {
      "epoch": 0.9295039164490861,
      "grad_norm": 0.4151720106601715,
      "learning_rate": 3.5268763728293756e-07,
      "loss": 0.6545,
      "step": 22428
    },
    {
      "epoch": 0.9295453603547599,
      "grad_norm": 0.4097486138343811,
      "learning_rate": 3.5248041775456926e-07,
      "loss": 0.6613,
      "step": 22429
    },
    {
      "epoch": 0.9295868042604335,
      "grad_norm": 0.4908868074417114,
      "learning_rate": 3.5227319822620086e-07,
      "loss": 0.6929,
      "step": 22430
    },
    {
      "epoch": 0.9296282481661072,
      "grad_norm": 0.3948611319065094,
      "learning_rate": 3.5206597869783256e-07,
      "loss": 0.672,
      "step": 22431
    },
    {
      "epoch": 0.9296696920717809,
      "grad_norm": 0.4200514853000641,
      "learning_rate": 3.5185875916946416e-07,
      "loss": 0.665,
      "step": 22432
    },
    {
      "epoch": 0.9297111359774545,
      "grad_norm": 0.42276307940483093,
      "learning_rate": 3.5165153964109576e-07,
      "loss": 0.7258,
      "step": 22433
    },
    {
      "epoch": 0.9297525798831282,
      "grad_norm": 0.42619457840919495,
      "learning_rate": 3.5144432011272746e-07,
      "loss": 0.7268,
      "step": 22434
    },
    {
      "epoch": 0.9297940237888018,
      "grad_norm": 0.38453376293182373,
      "learning_rate": 3.5123710058435906e-07,
      "loss": 0.6217,
      "step": 22435
    },
    {
      "epoch": 0.9298354676944756,
      "grad_norm": 0.39644718170166016,
      "learning_rate": 3.5102988105599076e-07,
      "loss": 0.6584,
      "step": 22436
    },
    {
      "epoch": 0.9298769116001492,
      "grad_norm": 0.4060280919075012,
      "learning_rate": 3.5082266152762236e-07,
      "loss": 0.6833,
      "step": 22437
    },
    {
      "epoch": 0.9299183555058229,
      "grad_norm": 0.39882612228393555,
      "learning_rate": 3.5061544199925406e-07,
      "loss": 0.6533,
      "step": 22438
    },
    {
      "epoch": 0.9299597994114965,
      "grad_norm": 0.3972475826740265,
      "learning_rate": 3.5040822247088566e-07,
      "loss": 0.6736,
      "step": 22439
    },
    {
      "epoch": 0.9300012433171702,
      "grad_norm": 0.4339313209056854,
      "learning_rate": 3.5020100294251736e-07,
      "loss": 0.6536,
      "step": 22440
    },
    {
      "epoch": 0.9300426872228439,
      "grad_norm": 0.4450444281101227,
      "learning_rate": 3.4999378341414896e-07,
      "loss": 0.6682,
      "step": 22441
    },
    {
      "epoch": 0.9300841311285175,
      "grad_norm": 0.4280235767364502,
      "learning_rate": 3.4978656388578067e-07,
      "loss": 0.6781,
      "step": 22442
    },
    {
      "epoch": 0.9301255750341912,
      "grad_norm": 0.4501829147338867,
      "learning_rate": 3.4957934435741226e-07,
      "loss": 0.6455,
      "step": 22443
    },
    {
      "epoch": 0.9301670189398649,
      "grad_norm": 0.40820568799972534,
      "learning_rate": 3.4937212482904397e-07,
      "loss": 0.6909,
      "step": 22444
    },
    {
      "epoch": 0.9302084628455386,
      "grad_norm": 0.41135138273239136,
      "learning_rate": 3.4916490530067556e-07,
      "loss": 0.6816,
      "step": 22445
    },
    {
      "epoch": 0.9302499067512122,
      "grad_norm": 0.4309143126010895,
      "learning_rate": 3.4895768577230716e-07,
      "loss": 0.5996,
      "step": 22446
    },
    {
      "epoch": 0.930291350656886,
      "grad_norm": 0.4103061556816101,
      "learning_rate": 3.4875046624393886e-07,
      "loss": 0.6311,
      "step": 22447
    },
    {
      "epoch": 0.9303327945625596,
      "grad_norm": 0.4207097589969635,
      "learning_rate": 3.4854324671557046e-07,
      "loss": 0.6221,
      "step": 22448
    },
    {
      "epoch": 0.9303742384682332,
      "grad_norm": 0.38750946521759033,
      "learning_rate": 3.4833602718720216e-07,
      "loss": 0.6354,
      "step": 22449
    },
    {
      "epoch": 0.9304156823739069,
      "grad_norm": 0.427666574716568,
      "learning_rate": 3.4812880765883376e-07,
      "loss": 0.677,
      "step": 22450
    },
    {
      "epoch": 0.9304571262795805,
      "grad_norm": 0.42890846729278564,
      "learning_rate": 3.4792158813046547e-07,
      "loss": 0.6663,
      "step": 22451
    },
    {
      "epoch": 0.9304985701852543,
      "grad_norm": 0.43159276247024536,
      "learning_rate": 3.4771436860209706e-07,
      "loss": 0.6667,
      "step": 22452
    },
    {
      "epoch": 0.9305400140909279,
      "grad_norm": 0.3827398419380188,
      "learning_rate": 3.4750714907372877e-07,
      "loss": 0.6272,
      "step": 22453
    },
    {
      "epoch": 0.9305814579966016,
      "grad_norm": 0.4345005452632904,
      "learning_rate": 3.4729992954536036e-07,
      "loss": 0.688,
      "step": 22454
    },
    {
      "epoch": 0.9306229019022753,
      "grad_norm": 0.3971203565597534,
      "learning_rate": 3.4709271001699207e-07,
      "loss": 0.6494,
      "step": 22455
    },
    {
      "epoch": 0.930664345807949,
      "grad_norm": 0.41512519121170044,
      "learning_rate": 3.4688549048862366e-07,
      "loss": 0.6655,
      "step": 22456
    },
    {
      "epoch": 0.9307057897136226,
      "grad_norm": 0.4498893916606903,
      "learning_rate": 3.466782709602553e-07,
      "loss": 0.6562,
      "step": 22457
    },
    {
      "epoch": 0.9307472336192962,
      "grad_norm": 0.42801666259765625,
      "learning_rate": 3.4647105143188696e-07,
      "loss": 0.6863,
      "step": 22458
    },
    {
      "epoch": 0.93078867752497,
      "grad_norm": 0.44655290246009827,
      "learning_rate": 3.4626383190351856e-07,
      "loss": 0.6892,
      "step": 22459
    },
    {
      "epoch": 0.9308301214306436,
      "grad_norm": 0.4149303436279297,
      "learning_rate": 3.4605661237515027e-07,
      "loss": 0.6915,
      "step": 22460
    },
    {
      "epoch": 0.9308715653363173,
      "grad_norm": 0.44622233510017395,
      "learning_rate": 3.4584939284678186e-07,
      "loss": 0.6776,
      "step": 22461
    },
    {
      "epoch": 0.9309130092419909,
      "grad_norm": 0.42673763632774353,
      "learning_rate": 3.4564217331841357e-07,
      "loss": 0.6396,
      "step": 22462
    },
    {
      "epoch": 0.9309544531476647,
      "grad_norm": 0.4450651705265045,
      "learning_rate": 3.4543495379004516e-07,
      "loss": 0.6953,
      "step": 22463
    },
    {
      "epoch": 0.9309958970533383,
      "grad_norm": 0.3957541882991791,
      "learning_rate": 3.4522773426167687e-07,
      "loss": 0.6909,
      "step": 22464
    },
    {
      "epoch": 0.931037340959012,
      "grad_norm": 0.4338967502117157,
      "learning_rate": 3.4502051473330846e-07,
      "loss": 0.6537,
      "step": 22465
    },
    {
      "epoch": 0.9310787848646856,
      "grad_norm": 0.4376203417778015,
      "learning_rate": 3.4481329520494017e-07,
      "loss": 0.6582,
      "step": 22466
    },
    {
      "epoch": 0.9311202287703593,
      "grad_norm": 0.4419507682323456,
      "learning_rate": 3.4460607567657176e-07,
      "loss": 0.7,
      "step": 22467
    },
    {
      "epoch": 0.931161672676033,
      "grad_norm": 0.39703530073165894,
      "learning_rate": 3.4439885614820347e-07,
      "loss": 0.6705,
      "step": 22468
    },
    {
      "epoch": 0.9312031165817066,
      "grad_norm": 0.4446874260902405,
      "learning_rate": 3.4419163661983507e-07,
      "loss": 0.6749,
      "step": 22469
    },
    {
      "epoch": 0.9312445604873804,
      "grad_norm": 0.4514203369617462,
      "learning_rate": 3.439844170914667e-07,
      "loss": 0.6825,
      "step": 22470
    },
    {
      "epoch": 0.931286004393054,
      "grad_norm": 0.42847374081611633,
      "learning_rate": 3.4377719756309837e-07,
      "loss": 0.6809,
      "step": 22471
    },
    {
      "epoch": 0.9313274482987277,
      "grad_norm": 0.4392583966255188,
      "learning_rate": 3.4356997803473e-07,
      "loss": 0.6964,
      "step": 22472
    },
    {
      "epoch": 0.9313688922044013,
      "grad_norm": 0.41295143961906433,
      "learning_rate": 3.4336275850636167e-07,
      "loss": 0.5714,
      "step": 22473
    },
    {
      "epoch": 0.9314103361100751,
      "grad_norm": 0.4675564765930176,
      "learning_rate": 3.431555389779933e-07,
      "loss": 0.6958,
      "step": 22474
    },
    {
      "epoch": 0.9314517800157487,
      "grad_norm": 0.39130446314811707,
      "learning_rate": 3.4294831944962497e-07,
      "loss": 0.6613,
      "step": 22475
    },
    {
      "epoch": 0.9314932239214223,
      "grad_norm": 0.44917210936546326,
      "learning_rate": 3.427410999212566e-07,
      "loss": 0.5966,
      "step": 22476
    },
    {
      "epoch": 0.931534667827096,
      "grad_norm": 0.3709532618522644,
      "learning_rate": 3.4253388039288827e-07,
      "loss": 0.6444,
      "step": 22477
    },
    {
      "epoch": 0.9315761117327697,
      "grad_norm": 0.413872092962265,
      "learning_rate": 3.4232666086451987e-07,
      "loss": 0.62,
      "step": 22478
    },
    {
      "epoch": 0.9316175556384434,
      "grad_norm": 0.4304647743701935,
      "learning_rate": 3.4211944133615157e-07,
      "loss": 0.7083,
      "step": 22479
    },
    {
      "epoch": 0.931658999544117,
      "grad_norm": 0.39772829413414,
      "learning_rate": 3.4191222180778317e-07,
      "loss": 0.6647,
      "step": 22480
    },
    {
      "epoch": 0.9317004434497907,
      "grad_norm": 0.44184941053390503,
      "learning_rate": 3.4170500227941487e-07,
      "loss": 0.6489,
      "step": 22481
    },
    {
      "epoch": 0.9317418873554644,
      "grad_norm": 0.38860422372817993,
      "learning_rate": 3.4149778275104647e-07,
      "loss": 0.6124,
      "step": 22482
    },
    {
      "epoch": 0.9317833312611381,
      "grad_norm": 0.3913264274597168,
      "learning_rate": 3.412905632226781e-07,
      "loss": 0.6427,
      "step": 22483
    },
    {
      "epoch": 0.9318247751668117,
      "grad_norm": 0.43540334701538086,
      "learning_rate": 3.4108334369430977e-07,
      "loss": 0.6736,
      "step": 22484
    },
    {
      "epoch": 0.9318662190724853,
      "grad_norm": 0.4227714240550995,
      "learning_rate": 3.408761241659414e-07,
      "loss": 0.6658,
      "step": 22485
    },
    {
      "epoch": 0.9319076629781591,
      "grad_norm": 0.42171013355255127,
      "learning_rate": 3.4066890463757307e-07,
      "loss": 0.7168,
      "step": 22486
    },
    {
      "epoch": 0.9319491068838327,
      "grad_norm": 0.45419901609420776,
      "learning_rate": 3.404616851092047e-07,
      "loss": 0.6091,
      "step": 22487
    },
    {
      "epoch": 0.9319905507895064,
      "grad_norm": 0.447968989610672,
      "learning_rate": 3.4025446558083637e-07,
      "loss": 0.6824,
      "step": 22488
    },
    {
      "epoch": 0.93203199469518,
      "grad_norm": 0.4223446846008301,
      "learning_rate": 3.40047246052468e-07,
      "loss": 0.6949,
      "step": 22489
    },
    {
      "epoch": 0.9320734386008538,
      "grad_norm": 0.3904838562011719,
      "learning_rate": 3.3984002652409967e-07,
      "loss": 0.5995,
      "step": 22490
    },
    {
      "epoch": 0.9321148825065274,
      "grad_norm": 0.4448259770870209,
      "learning_rate": 3.396328069957313e-07,
      "loss": 0.6755,
      "step": 22491
    },
    {
      "epoch": 0.932156326412201,
      "grad_norm": 0.426695317029953,
      "learning_rate": 3.3942558746736297e-07,
      "loss": 0.6599,
      "step": 22492
    },
    {
      "epoch": 0.9321977703178748,
      "grad_norm": 0.41158419847488403,
      "learning_rate": 3.392183679389946e-07,
      "loss": 0.689,
      "step": 22493
    },
    {
      "epoch": 0.9322392142235484,
      "grad_norm": 0.4201909899711609,
      "learning_rate": 3.390111484106262e-07,
      "loss": 0.656,
      "step": 22494
    },
    {
      "epoch": 0.9322806581292221,
      "grad_norm": 0.46222546696662903,
      "learning_rate": 3.388039288822579e-07,
      "loss": 0.6979,
      "step": 22495
    },
    {
      "epoch": 0.9323221020348957,
      "grad_norm": 0.4192471504211426,
      "learning_rate": 3.385967093538895e-07,
      "loss": 0.6494,
      "step": 22496
    },
    {
      "epoch": 0.9323635459405695,
      "grad_norm": 0.43279245495796204,
      "learning_rate": 3.3838948982552117e-07,
      "loss": 0.6575,
      "step": 22497
    },
    {
      "epoch": 0.9324049898462431,
      "grad_norm": 0.4200434684753418,
      "learning_rate": 3.381822702971528e-07,
      "loss": 0.7162,
      "step": 22498
    },
    {
      "epoch": 0.9324464337519168,
      "grad_norm": 0.48031294345855713,
      "learning_rate": 3.3797505076878447e-07,
      "loss": 0.6552,
      "step": 22499
    },
    {
      "epoch": 0.9324878776575904,
      "grad_norm": 0.417064905166626,
      "learning_rate": 3.377678312404161e-07,
      "loss": 0.6528,
      "step": 22500
    },
    {
      "epoch": 0.9325293215632641,
      "grad_norm": 0.40902429819107056,
      "learning_rate": 3.3756061171204777e-07,
      "loss": 0.5928,
      "step": 22501
    },
    {
      "epoch": 0.9325707654689378,
      "grad_norm": 0.42229366302490234,
      "learning_rate": 3.373533921836794e-07,
      "loss": 0.6895,
      "step": 22502
    },
    {
      "epoch": 0.9326122093746114,
      "grad_norm": 0.4400928020477295,
      "learning_rate": 3.3714617265531107e-07,
      "loss": 0.6702,
      "step": 22503
    },
    {
      "epoch": 0.9326536532802852,
      "grad_norm": 0.4230530560016632,
      "learning_rate": 3.369389531269427e-07,
      "loss": 0.7115,
      "step": 22504
    },
    {
      "epoch": 0.9326950971859588,
      "grad_norm": 0.4365421235561371,
      "learning_rate": 3.3673173359857437e-07,
      "loss": 0.672,
      "step": 22505
    },
    {
      "epoch": 0.9327365410916325,
      "grad_norm": 0.4257984757423401,
      "learning_rate": 3.36524514070206e-07,
      "loss": 0.6639,
      "step": 22506
    },
    {
      "epoch": 0.9327779849973061,
      "grad_norm": 0.41713088750839233,
      "learning_rate": 3.363172945418376e-07,
      "loss": 0.6263,
      "step": 22507
    },
    {
      "epoch": 0.9328194289029799,
      "grad_norm": 0.42098501324653625,
      "learning_rate": 3.361100750134693e-07,
      "loss": 0.6473,
      "step": 22508
    },
    {
      "epoch": 0.9328608728086535,
      "grad_norm": 0.4304363429546356,
      "learning_rate": 3.359028554851009e-07,
      "loss": 0.7419,
      "step": 22509
    },
    {
      "epoch": 0.9329023167143271,
      "grad_norm": 0.38049963116645813,
      "learning_rate": 3.356956359567326e-07,
      "loss": 0.6171,
      "step": 22510
    },
    {
      "epoch": 0.9329437606200008,
      "grad_norm": 0.4703044891357422,
      "learning_rate": 3.354884164283642e-07,
      "loss": 0.709,
      "step": 22511
    },
    {
      "epoch": 0.9329852045256745,
      "grad_norm": 0.44463521242141724,
      "learning_rate": 3.352811968999959e-07,
      "loss": 0.6324,
      "step": 22512
    },
    {
      "epoch": 0.9330266484313482,
      "grad_norm": 0.43506574630737305,
      "learning_rate": 3.350739773716275e-07,
      "loss": 0.6697,
      "step": 22513
    },
    {
      "epoch": 0.9330680923370218,
      "grad_norm": 0.40987923741340637,
      "learning_rate": 3.348667578432592e-07,
      "loss": 0.6053,
      "step": 22514
    },
    {
      "epoch": 0.9331095362426955,
      "grad_norm": 0.4395192563533783,
      "learning_rate": 3.346595383148908e-07,
      "loss": 0.6606,
      "step": 22515
    },
    {
      "epoch": 0.9331509801483692,
      "grad_norm": 0.43187767267227173,
      "learning_rate": 3.3445231878652247e-07,
      "loss": 0.7041,
      "step": 22516
    },
    {
      "epoch": 0.9331924240540429,
      "grad_norm": 0.5478802919387817,
      "learning_rate": 3.342450992581541e-07,
      "loss": 0.7148,
      "step": 22517
    },
    {
      "epoch": 0.9332338679597165,
      "grad_norm": 0.43575769662857056,
      "learning_rate": 3.340378797297858e-07,
      "loss": 0.6698,
      "step": 22518
    },
    {
      "epoch": 0.9332753118653901,
      "grad_norm": 0.3952096104621887,
      "learning_rate": 3.338306602014174e-07,
      "loss": 0.5858,
      "step": 22519
    },
    {
      "epoch": 0.9333167557710639,
      "grad_norm": 0.4112103581428528,
      "learning_rate": 3.33623440673049e-07,
      "loss": 0.625,
      "step": 22520
    },
    {
      "epoch": 0.9333581996767375,
      "grad_norm": 0.39789533615112305,
      "learning_rate": 3.334162211446807e-07,
      "loss": 0.6191,
      "step": 22521
    },
    {
      "epoch": 0.9333996435824112,
      "grad_norm": 0.3806796371936798,
      "learning_rate": 3.332090016163123e-07,
      "loss": 0.6299,
      "step": 22522
    },
    {
      "epoch": 0.9334410874880849,
      "grad_norm": 0.4431036412715912,
      "learning_rate": 3.33001782087944e-07,
      "loss": 0.6608,
      "step": 22523
    },
    {
      "epoch": 0.9334825313937586,
      "grad_norm": 0.4300888478755951,
      "learning_rate": 3.327945625595756e-07,
      "loss": 0.6429,
      "step": 22524
    },
    {
      "epoch": 0.9335239752994322,
      "grad_norm": 0.4719328284263611,
      "learning_rate": 3.3258734303120733e-07,
      "loss": 0.7184,
      "step": 22525
    },
    {
      "epoch": 0.9335654192051059,
      "grad_norm": 0.464844286441803,
      "learning_rate": 3.323801235028389e-07,
      "loss": 0.6747,
      "step": 22526
    },
    {
      "epoch": 0.9336068631107796,
      "grad_norm": 0.42974865436553955,
      "learning_rate": 3.3217290397447063e-07,
      "loss": 0.6743,
      "step": 22527
    },
    {
      "epoch": 0.9336483070164532,
      "grad_norm": 0.45739665627479553,
      "learning_rate": 3.319656844461022e-07,
      "loss": 0.6903,
      "step": 22528
    },
    {
      "epoch": 0.9336897509221269,
      "grad_norm": 0.39657679200172424,
      "learning_rate": 3.3175846491773393e-07,
      "loss": 0.6257,
      "step": 22529
    },
    {
      "epoch": 0.9337311948278005,
      "grad_norm": 0.4236172139644623,
      "learning_rate": 3.315512453893655e-07,
      "loss": 0.7118,
      "step": 22530
    },
    {
      "epoch": 0.9337726387334743,
      "grad_norm": 0.42161887884140015,
      "learning_rate": 3.313440258609971e-07,
      "loss": 0.6987,
      "step": 22531
    },
    {
      "epoch": 0.9338140826391479,
      "grad_norm": 0.40426599979400635,
      "learning_rate": 3.311368063326288e-07,
      "loss": 0.6415,
      "step": 22532
    },
    {
      "epoch": 0.9338555265448216,
      "grad_norm": 0.3915784955024719,
      "learning_rate": 3.309295868042604e-07,
      "loss": 0.6221,
      "step": 22533
    },
    {
      "epoch": 0.9338969704504952,
      "grad_norm": 0.4045061767101288,
      "learning_rate": 3.3072236727589213e-07,
      "loss": 0.6108,
      "step": 22534
    },
    {
      "epoch": 0.933938414356169,
      "grad_norm": 0.42910143733024597,
      "learning_rate": 3.305151477475237e-07,
      "loss": 0.6755,
      "step": 22535
    },
    {
      "epoch": 0.9339798582618426,
      "grad_norm": 0.41924622654914856,
      "learning_rate": 3.3030792821915543e-07,
      "loss": 0.6992,
      "step": 22536
    },
    {
      "epoch": 0.9340213021675162,
      "grad_norm": 0.4237860143184662,
      "learning_rate": 3.30100708690787e-07,
      "loss": 0.6746,
      "step": 22537
    },
    {
      "epoch": 0.93406274607319,
      "grad_norm": 0.404877245426178,
      "learning_rate": 3.2989348916241873e-07,
      "loss": 0.678,
      "step": 22538
    },
    {
      "epoch": 0.9341041899788636,
      "grad_norm": 0.413104385137558,
      "learning_rate": 3.296862696340503e-07,
      "loss": 0.7146,
      "step": 22539
    },
    {
      "epoch": 0.9341456338845373,
      "grad_norm": 0.43517807126045227,
      "learning_rate": 3.2947905010568203e-07,
      "loss": 0.6653,
      "step": 22540
    },
    {
      "epoch": 0.9341870777902109,
      "grad_norm": 0.41642457246780396,
      "learning_rate": 3.292718305773136e-07,
      "loss": 0.6696,
      "step": 22541
    },
    {
      "epoch": 0.9342285216958847,
      "grad_norm": 0.4401613473892212,
      "learning_rate": 3.2906461104894533e-07,
      "loss": 0.6564,
      "step": 22542
    },
    {
      "epoch": 0.9342699656015583,
      "grad_norm": 0.40061819553375244,
      "learning_rate": 3.2885739152057693e-07,
      "loss": 0.6029,
      "step": 22543
    },
    {
      "epoch": 0.934311409507232,
      "grad_norm": 0.44017326831817627,
      "learning_rate": 3.286501719922085e-07,
      "loss": 0.6219,
      "step": 22544
    },
    {
      "epoch": 0.9343528534129056,
      "grad_norm": 0.4017145037651062,
      "learning_rate": 3.2844295246384023e-07,
      "loss": 0.6577,
      "step": 22545
    },
    {
      "epoch": 0.9343942973185793,
      "grad_norm": 0.4330856204032898,
      "learning_rate": 3.282357329354718e-07,
      "loss": 0.6885,
      "step": 22546
    },
    {
      "epoch": 0.934435741224253,
      "grad_norm": 0.4124925136566162,
      "learning_rate": 3.2802851340710353e-07,
      "loss": 0.6338,
      "step": 22547
    },
    {
      "epoch": 0.9344771851299266,
      "grad_norm": 0.44773757457733154,
      "learning_rate": 3.278212938787351e-07,
      "loss": 0.6849,
      "step": 22548
    },
    {
      "epoch": 0.9345186290356003,
      "grad_norm": 0.3872513771057129,
      "learning_rate": 3.2761407435036683e-07,
      "loss": 0.6151,
      "step": 22549
    },
    {
      "epoch": 0.934560072941274,
      "grad_norm": 0.4315352439880371,
      "learning_rate": 3.2740685482199843e-07,
      "loss": 0.6829,
      "step": 22550
    },
    {
      "epoch": 0.9346015168469477,
      "grad_norm": 0.41650962829589844,
      "learning_rate": 3.2719963529363013e-07,
      "loss": 0.673,
      "step": 22551
    },
    {
      "epoch": 0.9346429607526213,
      "grad_norm": 0.39676204323768616,
      "learning_rate": 3.2699241576526173e-07,
      "loss": 0.6478,
      "step": 22552
    },
    {
      "epoch": 0.9346844046582949,
      "grad_norm": 0.4246250092983246,
      "learning_rate": 3.2678519623689343e-07,
      "loss": 0.6503,
      "step": 22553
    },
    {
      "epoch": 0.9347258485639687,
      "grad_norm": 0.4099022150039673,
      "learning_rate": 3.2657797670852503e-07,
      "loss": 0.667,
      "step": 22554
    },
    {
      "epoch": 0.9347672924696423,
      "grad_norm": 0.4494040608406067,
      "learning_rate": 3.2637075718015673e-07,
      "loss": 0.645,
      "step": 22555
    },
    {
      "epoch": 0.934808736375316,
      "grad_norm": 0.3926289677619934,
      "learning_rate": 3.2616353765178833e-07,
      "loss": 0.6244,
      "step": 22556
    },
    {
      "epoch": 0.9348501802809897,
      "grad_norm": 0.42267096042633057,
      "learning_rate": 3.259563181234199e-07,
      "loss": 0.624,
      "step": 22557
    },
    {
      "epoch": 0.9348916241866634,
      "grad_norm": 0.40026867389678955,
      "learning_rate": 3.2574909859505163e-07,
      "loss": 0.6619,
      "step": 22558
    },
    {
      "epoch": 0.934933068092337,
      "grad_norm": 0.4497218132019043,
      "learning_rate": 3.2554187906668323e-07,
      "loss": 0.6685,
      "step": 22559
    },
    {
      "epoch": 0.9349745119980107,
      "grad_norm": 0.40292811393737793,
      "learning_rate": 3.2533465953831493e-07,
      "loss": 0.663,
      "step": 22560
    },
    {
      "epoch": 0.9350159559036844,
      "grad_norm": 0.43290334939956665,
      "learning_rate": 3.2512744000994653e-07,
      "loss": 0.6932,
      "step": 22561
    },
    {
      "epoch": 0.935057399809358,
      "grad_norm": 0.3959939181804657,
      "learning_rate": 3.2492022048157823e-07,
      "loss": 0.6021,
      "step": 22562
    },
    {
      "epoch": 0.9350988437150317,
      "grad_norm": 0.4570123553276062,
      "learning_rate": 3.2471300095320983e-07,
      "loss": 0.6576,
      "step": 22563
    },
    {
      "epoch": 0.9351402876207053,
      "grad_norm": 0.429307222366333,
      "learning_rate": 3.2450578142484153e-07,
      "loss": 0.6775,
      "step": 22564
    },
    {
      "epoch": 0.9351817315263791,
      "grad_norm": 0.43759632110595703,
      "learning_rate": 3.2429856189647313e-07,
      "loss": 0.6755,
      "step": 22565
    },
    {
      "epoch": 0.9352231754320527,
      "grad_norm": 0.3922669589519501,
      "learning_rate": 3.2409134236810483e-07,
      "loss": 0.6909,
      "step": 22566
    },
    {
      "epoch": 0.9352646193377264,
      "grad_norm": 0.41394883394241333,
      "learning_rate": 3.2388412283973643e-07,
      "loss": 0.6711,
      "step": 22567
    },
    {
      "epoch": 0.9353060632434,
      "grad_norm": 0.43639811873435974,
      "learning_rate": 3.236769033113681e-07,
      "loss": 0.6536,
      "step": 22568
    },
    {
      "epoch": 0.9353475071490738,
      "grad_norm": 0.40334808826446533,
      "learning_rate": 3.2346968378299973e-07,
      "loss": 0.6211,
      "step": 22569
    },
    {
      "epoch": 0.9353889510547474,
      "grad_norm": 0.39216935634613037,
      "learning_rate": 3.232624642546314e-07,
      "loss": 0.6305,
      "step": 22570
    },
    {
      "epoch": 0.935430394960421,
      "grad_norm": 0.40508219599723816,
      "learning_rate": 3.2305524472626303e-07,
      "loss": 0.6725,
      "step": 22571
    },
    {
      "epoch": 0.9354718388660948,
      "grad_norm": 0.46662476658821106,
      "learning_rate": 3.2284802519789463e-07,
      "loss": 0.7003,
      "step": 22572
    },
    {
      "epoch": 0.9355132827717684,
      "grad_norm": 0.4066019356250763,
      "learning_rate": 3.2264080566952633e-07,
      "loss": 0.6772,
      "step": 22573
    },
    {
      "epoch": 0.9355547266774421,
      "grad_norm": 0.44691184163093567,
      "learning_rate": 3.2243358614115793e-07,
      "loss": 0.6202,
      "step": 22574
    },
    {
      "epoch": 0.9355961705831157,
      "grad_norm": 0.4330269396305084,
      "learning_rate": 3.2222636661278963e-07,
      "loss": 0.6873,
      "step": 22575
    },
    {
      "epoch": 0.9356376144887895,
      "grad_norm": 0.40682119131088257,
      "learning_rate": 3.2201914708442123e-07,
      "loss": 0.6425,
      "step": 22576
    },
    {
      "epoch": 0.9356790583944631,
      "grad_norm": 0.40092039108276367,
      "learning_rate": 3.2181192755605293e-07,
      "loss": 0.6183,
      "step": 22577
    },
    {
      "epoch": 0.9357205023001368,
      "grad_norm": 0.4146396219730377,
      "learning_rate": 3.2160470802768453e-07,
      "loss": 0.605,
      "step": 22578
    },
    {
      "epoch": 0.9357619462058104,
      "grad_norm": 0.38058701157569885,
      "learning_rate": 3.2139748849931623e-07,
      "loss": 0.6647,
      "step": 22579
    },
    {
      "epoch": 0.9358033901114841,
      "grad_norm": 0.42083314061164856,
      "learning_rate": 3.2119026897094783e-07,
      "loss": 0.7112,
      "step": 22580
    },
    {
      "epoch": 0.9358448340171578,
      "grad_norm": 0.3826419413089752,
      "learning_rate": 3.209830494425795e-07,
      "loss": 0.6575,
      "step": 22581
    },
    {
      "epoch": 0.9358862779228314,
      "grad_norm": 0.42726337909698486,
      "learning_rate": 3.2077582991421113e-07,
      "loss": 0.6863,
      "step": 22582
    },
    {
      "epoch": 0.9359277218285051,
      "grad_norm": 0.4279736876487732,
      "learning_rate": 3.205686103858428e-07,
      "loss": 0.6967,
      "step": 22583
    },
    {
      "epoch": 0.9359691657341788,
      "grad_norm": 0.3983338475227356,
      "learning_rate": 3.2036139085747443e-07,
      "loss": 0.6658,
      "step": 22584
    },
    {
      "epoch": 0.9360106096398525,
      "grad_norm": 0.40128687024116516,
      "learning_rate": 3.201541713291061e-07,
      "loss": 0.6497,
      "step": 22585
    },
    {
      "epoch": 0.9360520535455261,
      "grad_norm": 0.42378413677215576,
      "learning_rate": 3.1994695180073773e-07,
      "loss": 0.6321,
      "step": 22586
    },
    {
      "epoch": 0.9360934974511999,
      "grad_norm": 0.4139254689216614,
      "learning_rate": 3.197397322723694e-07,
      "loss": 0.6423,
      "step": 22587
    },
    {
      "epoch": 0.9361349413568735,
      "grad_norm": 0.4287931025028229,
      "learning_rate": 3.1953251274400103e-07,
      "loss": 0.656,
      "step": 22588
    },
    {
      "epoch": 0.9361763852625471,
      "grad_norm": 0.42412757873535156,
      "learning_rate": 3.193252932156327e-07,
      "loss": 0.7063,
      "step": 22589
    },
    {
      "epoch": 0.9362178291682208,
      "grad_norm": 0.4218384325504303,
      "learning_rate": 3.1911807368726433e-07,
      "loss": 0.6741,
      "step": 22590
    },
    {
      "epoch": 0.9362592730738944,
      "grad_norm": 0.40070295333862305,
      "learning_rate": 3.1891085415889593e-07,
      "loss": 0.6333,
      "step": 22591
    },
    {
      "epoch": 0.9363007169795682,
      "grad_norm": 0.42076897621154785,
      "learning_rate": 3.187036346305276e-07,
      "loss": 0.6736,
      "step": 22592
    },
    {
      "epoch": 0.9363421608852418,
      "grad_norm": 0.45353585481643677,
      "learning_rate": 3.1849641510215923e-07,
      "loss": 0.6747,
      "step": 22593
    },
    {
      "epoch": 0.9363836047909155,
      "grad_norm": 0.4181041121482849,
      "learning_rate": 3.182891955737909e-07,
      "loss": 0.6736,
      "step": 22594
    },
    {
      "epoch": 0.9364250486965892,
      "grad_norm": 0.43138203024864197,
      "learning_rate": 3.1808197604542253e-07,
      "loss": 0.7139,
      "step": 22595
    },
    {
      "epoch": 0.9364664926022629,
      "grad_norm": 0.45184755325317383,
      "learning_rate": 3.178747565170542e-07,
      "loss": 0.6989,
      "step": 22596
    },
    {
      "epoch": 0.9365079365079365,
      "grad_norm": 0.4023093283176422,
      "learning_rate": 3.1766753698868583e-07,
      "loss": 0.6487,
      "step": 22597
    },
    {
      "epoch": 0.9365493804136101,
      "grad_norm": 0.41347697377204895,
      "learning_rate": 3.174603174603175e-07,
      "loss": 0.6707,
      "step": 22598
    },
    {
      "epoch": 0.9365908243192839,
      "grad_norm": 0.4041733741760254,
      "learning_rate": 3.1725309793194913e-07,
      "loss": 0.6156,
      "step": 22599
    },
    {
      "epoch": 0.9366322682249575,
      "grad_norm": 0.4014715850353241,
      "learning_rate": 3.170458784035808e-07,
      "loss": 0.6373,
      "step": 22600
    },
    {
      "epoch": 0.9366737121306312,
      "grad_norm": 0.4268902540206909,
      "learning_rate": 3.1683865887521244e-07,
      "loss": 0.6792,
      "step": 22601
    },
    {
      "epoch": 0.9367151560363048,
      "grad_norm": 0.448744535446167,
      "learning_rate": 3.166314393468441e-07,
      "loss": 0.7025,
      "step": 22602
    },
    {
      "epoch": 0.9367565999419786,
      "grad_norm": 0.4212503135204315,
      "learning_rate": 3.1642421981847574e-07,
      "loss": 0.6621,
      "step": 22603
    },
    {
      "epoch": 0.9367980438476522,
      "grad_norm": 0.47785696387290955,
      "learning_rate": 3.162170002901074e-07,
      "loss": 0.7067,
      "step": 22604
    },
    {
      "epoch": 0.9368394877533259,
      "grad_norm": 0.3724975883960724,
      "learning_rate": 3.16009780761739e-07,
      "loss": 0.6205,
      "step": 22605
    },
    {
      "epoch": 0.9368809316589995,
      "grad_norm": 0.402670294046402,
      "learning_rate": 3.158025612333707e-07,
      "loss": 0.6335,
      "step": 22606
    },
    {
      "epoch": 0.9369223755646732,
      "grad_norm": 0.3873659074306488,
      "learning_rate": 3.155953417050023e-07,
      "loss": 0.6704,
      "step": 22607
    },
    {
      "epoch": 0.9369638194703469,
      "grad_norm": 0.4435826241970062,
      "learning_rate": 3.15388122176634e-07,
      "loss": 0.6842,
      "step": 22608
    },
    {
      "epoch": 0.9370052633760205,
      "grad_norm": 0.3981541097164154,
      "learning_rate": 3.151809026482656e-07,
      "loss": 0.6494,
      "step": 22609
    },
    {
      "epoch": 0.9370467072816943,
      "grad_norm": 0.4442363977432251,
      "learning_rate": 3.1497368311989724e-07,
      "loss": 0.7067,
      "step": 22610
    },
    {
      "epoch": 0.9370881511873679,
      "grad_norm": 0.4465920627117157,
      "learning_rate": 3.147664635915289e-07,
      "loss": 0.6761,
      "step": 22611
    },
    {
      "epoch": 0.9371295950930416,
      "grad_norm": 0.4386685788631439,
      "learning_rate": 3.1455924406316054e-07,
      "loss": 0.701,
      "step": 22612
    },
    {
      "epoch": 0.9371710389987152,
      "grad_norm": 0.37436750531196594,
      "learning_rate": 3.143520245347922e-07,
      "loss": 0.6237,
      "step": 22613
    },
    {
      "epoch": 0.9372124829043889,
      "grad_norm": 0.3905163109302521,
      "learning_rate": 3.1414480500642384e-07,
      "loss": 0.5999,
      "step": 22614
    },
    {
      "epoch": 0.9372539268100626,
      "grad_norm": 0.407635897397995,
      "learning_rate": 3.139375854780555e-07,
      "loss": 0.7092,
      "step": 22615
    },
    {
      "epoch": 0.9372953707157362,
      "grad_norm": 0.42757824063301086,
      "learning_rate": 3.1373036594968714e-07,
      "loss": 0.675,
      "step": 22616
    },
    {
      "epoch": 0.9373368146214099,
      "grad_norm": 0.41417235136032104,
      "learning_rate": 3.135231464213188e-07,
      "loss": 0.6869,
      "step": 22617
    },
    {
      "epoch": 0.9373782585270836,
      "grad_norm": 0.4252808392047882,
      "learning_rate": 3.133159268929504e-07,
      "loss": 0.6813,
      "step": 22618
    },
    {
      "epoch": 0.9374197024327573,
      "grad_norm": 0.4096668064594269,
      "learning_rate": 3.131087073645821e-07,
      "loss": 0.608,
      "step": 22619
    },
    {
      "epoch": 0.9374611463384309,
      "grad_norm": 0.5335261821746826,
      "learning_rate": 3.129014878362137e-07,
      "loss": 0.6775,
      "step": 22620
    },
    {
      "epoch": 0.9375025902441046,
      "grad_norm": 0.39103323221206665,
      "learning_rate": 3.126942683078454e-07,
      "loss": 0.6304,
      "step": 22621
    },
    {
      "epoch": 0.9375440341497783,
      "grad_norm": 0.444345623254776,
      "learning_rate": 3.1248704877947704e-07,
      "loss": 0.7036,
      "step": 22622
    },
    {
      "epoch": 0.9375854780554519,
      "grad_norm": 0.3969419300556183,
      "learning_rate": 3.1227982925110864e-07,
      "loss": 0.6774,
      "step": 22623
    },
    {
      "epoch": 0.9376269219611256,
      "grad_norm": 0.41894611716270447,
      "learning_rate": 3.120726097227403e-07,
      "loss": 0.6765,
      "step": 22624
    },
    {
      "epoch": 0.9376683658667992,
      "grad_norm": 0.4521391987800598,
      "learning_rate": 3.1186539019437194e-07,
      "loss": 0.687,
      "step": 22625
    },
    {
      "epoch": 0.937709809772473,
      "grad_norm": 0.4342905879020691,
      "learning_rate": 3.116581706660036e-07,
      "loss": 0.681,
      "step": 22626
    },
    {
      "epoch": 0.9377512536781466,
      "grad_norm": 0.43111875653266907,
      "learning_rate": 3.1145095113763524e-07,
      "loss": 0.684,
      "step": 22627
    },
    {
      "epoch": 0.9377926975838203,
      "grad_norm": 0.43110212683677673,
      "learning_rate": 3.112437316092669e-07,
      "loss": 0.6765,
      "step": 22628
    },
    {
      "epoch": 0.937834141489494,
      "grad_norm": 0.40399089455604553,
      "learning_rate": 3.1103651208089854e-07,
      "loss": 0.6296,
      "step": 22629
    },
    {
      "epoch": 0.9378755853951677,
      "grad_norm": 0.48418989777565,
      "learning_rate": 3.108292925525302e-07,
      "loss": 0.7214,
      "step": 22630
    },
    {
      "epoch": 0.9379170293008413,
      "grad_norm": 0.4238094091415405,
      "learning_rate": 3.1062207302416184e-07,
      "loss": 0.6979,
      "step": 22631
    },
    {
      "epoch": 0.9379584732065149,
      "grad_norm": 0.3733350932598114,
      "learning_rate": 3.104148534957935e-07,
      "loss": 0.6323,
      "step": 22632
    },
    {
      "epoch": 0.9379999171121887,
      "grad_norm": 0.41005539894104004,
      "learning_rate": 3.1020763396742514e-07,
      "loss": 0.6648,
      "step": 22633
    },
    {
      "epoch": 0.9380413610178623,
      "grad_norm": 0.41264620423316956,
      "learning_rate": 3.100004144390568e-07,
      "loss": 0.6005,
      "step": 22634
    },
    {
      "epoch": 0.938082804923536,
      "grad_norm": 0.4105056822299957,
      "learning_rate": 3.0979319491068844e-07,
      "loss": 0.6941,
      "step": 22635
    },
    {
      "epoch": 0.9381242488292096,
      "grad_norm": 0.4443499445915222,
      "learning_rate": 3.0958597538232004e-07,
      "loss": 0.6683,
      "step": 22636
    },
    {
      "epoch": 0.9381656927348834,
      "grad_norm": 0.4108963906764984,
      "learning_rate": 3.093787558539517e-07,
      "loss": 0.639,
      "step": 22637
    },
    {
      "epoch": 0.938207136640557,
      "grad_norm": 0.4111112952232361,
      "learning_rate": 3.0917153632558334e-07,
      "loss": 0.6523,
      "step": 22638
    },
    {
      "epoch": 0.9382485805462307,
      "grad_norm": 0.4012923836708069,
      "learning_rate": 3.08964316797215e-07,
      "loss": 0.6331,
      "step": 22639
    },
    {
      "epoch": 0.9382900244519043,
      "grad_norm": 0.39918848872184753,
      "learning_rate": 3.0875709726884664e-07,
      "loss": 0.6758,
      "step": 22640
    },
    {
      "epoch": 0.938331468357578,
      "grad_norm": 0.41612181067466736,
      "learning_rate": 3.085498777404783e-07,
      "loss": 0.6351,
      "step": 22641
    },
    {
      "epoch": 0.9383729122632517,
      "grad_norm": 0.41747620701789856,
      "learning_rate": 3.0834265821210994e-07,
      "loss": 0.6891,
      "step": 22642
    },
    {
      "epoch": 0.9384143561689253,
      "grad_norm": 0.4157271087169647,
      "learning_rate": 3.081354386837416e-07,
      "loss": 0.6527,
      "step": 22643
    },
    {
      "epoch": 0.9384558000745991,
      "grad_norm": 0.40750008821487427,
      "learning_rate": 3.0792821915537324e-07,
      "loss": 0.6462,
      "step": 22644
    },
    {
      "epoch": 0.9384972439802727,
      "grad_norm": 0.4657857120037079,
      "learning_rate": 3.077209996270049e-07,
      "loss": 0.7078,
      "step": 22645
    },
    {
      "epoch": 0.9385386878859464,
      "grad_norm": 0.4145195186138153,
      "learning_rate": 3.0751378009863654e-07,
      "loss": 0.6648,
      "step": 22646
    },
    {
      "epoch": 0.93858013179162,
      "grad_norm": 0.3850816488265991,
      "learning_rate": 3.073065605702682e-07,
      "loss": 0.6161,
      "step": 22647
    },
    {
      "epoch": 0.9386215756972938,
      "grad_norm": 0.4316581189632416,
      "learning_rate": 3.070993410418998e-07,
      "loss": 0.6847,
      "step": 22648
    },
    {
      "epoch": 0.9386630196029674,
      "grad_norm": 0.4093256890773773,
      "learning_rate": 3.0689212151353144e-07,
      "loss": 0.6587,
      "step": 22649
    },
    {
      "epoch": 0.938704463508641,
      "grad_norm": 0.41901537775993347,
      "learning_rate": 3.066849019851631e-07,
      "loss": 0.7061,
      "step": 22650
    },
    {
      "epoch": 0.9387459074143147,
      "grad_norm": 0.40523555874824524,
      "learning_rate": 3.0647768245679474e-07,
      "loss": 0.6927,
      "step": 22651
    },
    {
      "epoch": 0.9387873513199884,
      "grad_norm": 0.4257374107837677,
      "learning_rate": 3.062704629284264e-07,
      "loss": 0.6603,
      "step": 22652
    },
    {
      "epoch": 0.9388287952256621,
      "grad_norm": 0.40490153431892395,
      "learning_rate": 3.0606324340005804e-07,
      "loss": 0.6597,
      "step": 22653
    },
    {
      "epoch": 0.9388702391313357,
      "grad_norm": 0.4396701753139496,
      "learning_rate": 3.058560238716897e-07,
      "loss": 0.667,
      "step": 22654
    },
    {
      "epoch": 0.9389116830370094,
      "grad_norm": 0.4322687089443207,
      "learning_rate": 3.0564880434332134e-07,
      "loss": 0.7506,
      "step": 22655
    },
    {
      "epoch": 0.9389531269426831,
      "grad_norm": 0.42184755206108093,
      "learning_rate": 3.05441584814953e-07,
      "loss": 0.6685,
      "step": 22656
    },
    {
      "epoch": 0.9389945708483568,
      "grad_norm": 0.48660042881965637,
      "learning_rate": 3.0523436528658464e-07,
      "loss": 0.6975,
      "step": 22657
    },
    {
      "epoch": 0.9390360147540304,
      "grad_norm": 0.43585479259490967,
      "learning_rate": 3.050271457582163e-07,
      "loss": 0.6646,
      "step": 22658
    },
    {
      "epoch": 0.939077458659704,
      "grad_norm": 0.3921436667442322,
      "learning_rate": 3.0481992622984794e-07,
      "loss": 0.6346,
      "step": 22659
    },
    {
      "epoch": 0.9391189025653778,
      "grad_norm": 0.41703155636787415,
      "learning_rate": 3.0461270670147954e-07,
      "loss": 0.7029,
      "step": 22660
    },
    {
      "epoch": 0.9391603464710514,
      "grad_norm": 0.4036855399608612,
      "learning_rate": 3.044054871731112e-07,
      "loss": 0.698,
      "step": 22661
    },
    {
      "epoch": 0.9392017903767251,
      "grad_norm": 0.3861207067966461,
      "learning_rate": 3.0419826764474284e-07,
      "loss": 0.6543,
      "step": 22662
    },
    {
      "epoch": 0.9392432342823988,
      "grad_norm": 0.42968934774398804,
      "learning_rate": 3.039910481163745e-07,
      "loss": 0.6274,
      "step": 22663
    },
    {
      "epoch": 0.9392846781880725,
      "grad_norm": 0.45244643092155457,
      "learning_rate": 3.0378382858800614e-07,
      "loss": 0.6895,
      "step": 22664
    },
    {
      "epoch": 0.9393261220937461,
      "grad_norm": 0.4216071367263794,
      "learning_rate": 3.035766090596378e-07,
      "loss": 0.6938,
      "step": 22665
    },
    {
      "epoch": 0.9393675659994198,
      "grad_norm": 0.4145006239414215,
      "learning_rate": 3.0336938953126944e-07,
      "loss": 0.6569,
      "step": 22666
    },
    {
      "epoch": 0.9394090099050935,
      "grad_norm": 0.3545956611633301,
      "learning_rate": 3.031621700029011e-07,
      "loss": 0.6086,
      "step": 22667
    },
    {
      "epoch": 0.9394504538107671,
      "grad_norm": 0.420364648103714,
      "learning_rate": 3.0295495047453274e-07,
      "loss": 0.6499,
      "step": 22668
    },
    {
      "epoch": 0.9394918977164408,
      "grad_norm": 0.4074712097644806,
      "learning_rate": 3.027477309461644e-07,
      "loss": 0.6473,
      "step": 22669
    },
    {
      "epoch": 0.9395333416221144,
      "grad_norm": 0.452358603477478,
      "learning_rate": 3.0254051141779605e-07,
      "loss": 0.6967,
      "step": 22670
    },
    {
      "epoch": 0.9395747855277882,
      "grad_norm": 0.4092451333999634,
      "learning_rate": 3.023332918894277e-07,
      "loss": 0.6831,
      "step": 22671
    },
    {
      "epoch": 0.9396162294334618,
      "grad_norm": 0.4189174473285675,
      "learning_rate": 3.021260723610593e-07,
      "loss": 0.6687,
      "step": 22672
    },
    {
      "epoch": 0.9396576733391355,
      "grad_norm": 0.3928716778755188,
      "learning_rate": 3.0191885283269094e-07,
      "loss": 0.6487,
      "step": 22673
    },
    {
      "epoch": 0.9396991172448091,
      "grad_norm": 0.41957369446754456,
      "learning_rate": 3.017116333043226e-07,
      "loss": 0.6141,
      "step": 22674
    },
    {
      "epoch": 0.9397405611504828,
      "grad_norm": 0.44678106904029846,
      "learning_rate": 3.0150441377595424e-07,
      "loss": 0.6312,
      "step": 22675
    },
    {
      "epoch": 0.9397820050561565,
      "grad_norm": 0.4766293466091156,
      "learning_rate": 3.012971942475859e-07,
      "loss": 0.668,
      "step": 22676
    },
    {
      "epoch": 0.9398234489618301,
      "grad_norm": 0.3994397521018982,
      "learning_rate": 3.0108997471921754e-07,
      "loss": 0.6456,
      "step": 22677
    },
    {
      "epoch": 0.9398648928675039,
      "grad_norm": 0.44541677832603455,
      "learning_rate": 3.008827551908492e-07,
      "loss": 0.6414,
      "step": 22678
    },
    {
      "epoch": 0.9399063367731775,
      "grad_norm": 0.3981247544288635,
      "learning_rate": 3.0067553566248085e-07,
      "loss": 0.7126,
      "step": 22679
    },
    {
      "epoch": 0.9399477806788512,
      "grad_norm": 0.41834592819213867,
      "learning_rate": 3.004683161341125e-07,
      "loss": 0.6638,
      "step": 22680
    },
    {
      "epoch": 0.9399892245845248,
      "grad_norm": 0.5145294666290283,
      "learning_rate": 3.0026109660574415e-07,
      "loss": 0.741,
      "step": 22681
    },
    {
      "epoch": 0.9400306684901986,
      "grad_norm": 0.47122928500175476,
      "learning_rate": 3.000538770773758e-07,
      "loss": 0.6801,
      "step": 22682
    },
    {
      "epoch": 0.9400721123958722,
      "grad_norm": 0.3984459340572357,
      "learning_rate": 2.9984665754900745e-07,
      "loss": 0.621,
      "step": 22683
    },
    {
      "epoch": 0.9401135563015458,
      "grad_norm": 0.4163937568664551,
      "learning_rate": 2.996394380206391e-07,
      "loss": 0.6693,
      "step": 22684
    },
    {
      "epoch": 0.9401550002072195,
      "grad_norm": 0.4243709146976471,
      "learning_rate": 2.994322184922707e-07,
      "loss": 0.6832,
      "step": 22685
    },
    {
      "epoch": 0.9401964441128932,
      "grad_norm": 0.40118080377578735,
      "learning_rate": 2.9922499896390234e-07,
      "loss": 0.6736,
      "step": 22686
    },
    {
      "epoch": 0.9402378880185669,
      "grad_norm": 0.39724454283714294,
      "learning_rate": 2.99017779435534e-07,
      "loss": 0.605,
      "step": 22687
    },
    {
      "epoch": 0.9402793319242405,
      "grad_norm": 0.3788558840751648,
      "learning_rate": 2.9881055990716565e-07,
      "loss": 0.6029,
      "step": 22688
    },
    {
      "epoch": 0.9403207758299142,
      "grad_norm": 0.3969612717628479,
      "learning_rate": 2.986033403787973e-07,
      "loss": 0.6504,
      "step": 22689
    },
    {
      "epoch": 0.9403622197355879,
      "grad_norm": 0.4430015981197357,
      "learning_rate": 2.9839612085042895e-07,
      "loss": 0.6353,
      "step": 22690
    },
    {
      "epoch": 0.9404036636412616,
      "grad_norm": 0.433103084564209,
      "learning_rate": 2.981889013220606e-07,
      "loss": 0.661,
      "step": 22691
    },
    {
      "epoch": 0.9404451075469352,
      "grad_norm": 0.39984339475631714,
      "learning_rate": 2.9798168179369225e-07,
      "loss": 0.6357,
      "step": 22692
    },
    {
      "epoch": 0.9404865514526088,
      "grad_norm": 0.4219370484352112,
      "learning_rate": 2.977744622653239e-07,
      "loss": 0.6628,
      "step": 22693
    },
    {
      "epoch": 0.9405279953582826,
      "grad_norm": 0.4019707143306732,
      "learning_rate": 2.9756724273695555e-07,
      "loss": 0.594,
      "step": 22694
    },
    {
      "epoch": 0.9405694392639562,
      "grad_norm": 0.4206814169883728,
      "learning_rate": 2.973600232085872e-07,
      "loss": 0.6469,
      "step": 22695
    },
    {
      "epoch": 0.9406108831696299,
      "grad_norm": 0.40712857246398926,
      "learning_rate": 2.9715280368021885e-07,
      "loss": 0.6577,
      "step": 22696
    },
    {
      "epoch": 0.9406523270753036,
      "grad_norm": 0.38911762833595276,
      "learning_rate": 2.969455841518505e-07,
      "loss": 0.5895,
      "step": 22697
    },
    {
      "epoch": 0.9406937709809773,
      "grad_norm": 0.42631709575653076,
      "learning_rate": 2.9673836462348215e-07,
      "loss": 0.7179,
      "step": 22698
    },
    {
      "epoch": 0.9407352148866509,
      "grad_norm": 0.41164296865463257,
      "learning_rate": 2.965311450951138e-07,
      "loss": 0.6664,
      "step": 22699
    },
    {
      "epoch": 0.9407766587923246,
      "grad_norm": 0.40951773524284363,
      "learning_rate": 2.9632392556674545e-07,
      "loss": 0.701,
      "step": 22700
    },
    {
      "epoch": 0.9408181026979983,
      "grad_norm": 0.41902661323547363,
      "learning_rate": 2.961167060383771e-07,
      "loss": 0.5952,
      "step": 22701
    },
    {
      "epoch": 0.9408595466036719,
      "grad_norm": 0.44185319542884827,
      "learning_rate": 2.9590948651000875e-07,
      "loss": 0.6719,
      "step": 22702
    },
    {
      "epoch": 0.9409009905093456,
      "grad_norm": 0.446958988904953,
      "learning_rate": 2.957022669816404e-07,
      "loss": 0.6926,
      "step": 22703
    },
    {
      "epoch": 0.9409424344150192,
      "grad_norm": 0.4236953854560852,
      "learning_rate": 2.95495047453272e-07,
      "loss": 0.6534,
      "step": 22704
    },
    {
      "epoch": 0.940983878320693,
      "grad_norm": 0.4064887762069702,
      "learning_rate": 2.9528782792490365e-07,
      "loss": 0.6588,
      "step": 22705
    },
    {
      "epoch": 0.9410253222263666,
      "grad_norm": 0.42141008377075195,
      "learning_rate": 2.950806083965353e-07,
      "loss": 0.7013,
      "step": 22706
    },
    {
      "epoch": 0.9410667661320403,
      "grad_norm": 0.4182586967945099,
      "learning_rate": 2.9487338886816695e-07,
      "loss": 0.6973,
      "step": 22707
    },
    {
      "epoch": 0.9411082100377139,
      "grad_norm": 0.45304182171821594,
      "learning_rate": 2.946661693397986e-07,
      "loss": 0.704,
      "step": 22708
    },
    {
      "epoch": 0.9411496539433877,
      "grad_norm": 0.47902780771255493,
      "learning_rate": 2.9445894981143025e-07,
      "loss": 0.7422,
      "step": 22709
    },
    {
      "epoch": 0.9411910978490613,
      "grad_norm": 0.4208369255065918,
      "learning_rate": 2.942517302830619e-07,
      "loss": 0.6577,
      "step": 22710
    },
    {
      "epoch": 0.9412325417547349,
      "grad_norm": 0.4209844172000885,
      "learning_rate": 2.9404451075469355e-07,
      "loss": 0.7283,
      "step": 22711
    },
    {
      "epoch": 0.9412739856604087,
      "grad_norm": 0.4426361620426178,
      "learning_rate": 2.938372912263252e-07,
      "loss": 0.6523,
      "step": 22712
    },
    {
      "epoch": 0.9413154295660823,
      "grad_norm": 0.43733128905296326,
      "learning_rate": 2.9363007169795685e-07,
      "loss": 0.688,
      "step": 22713
    },
    {
      "epoch": 0.941356873471756,
      "grad_norm": 0.4314127266407013,
      "learning_rate": 2.934228521695885e-07,
      "loss": 0.6344,
      "step": 22714
    },
    {
      "epoch": 0.9413983173774296,
      "grad_norm": 0.41051992774009705,
      "learning_rate": 2.9321563264122015e-07,
      "loss": 0.6548,
      "step": 22715
    },
    {
      "epoch": 0.9414397612831034,
      "grad_norm": 0.4073718190193176,
      "learning_rate": 2.930084131128518e-07,
      "loss": 0.7301,
      "step": 22716
    },
    {
      "epoch": 0.941481205188777,
      "grad_norm": 0.40204429626464844,
      "learning_rate": 2.9280119358448345e-07,
      "loss": 0.5961,
      "step": 22717
    },
    {
      "epoch": 0.9415226490944507,
      "grad_norm": 0.4238159954547882,
      "learning_rate": 2.925939740561151e-07,
      "loss": 0.6633,
      "step": 22718
    },
    {
      "epoch": 0.9415640930001243,
      "grad_norm": 0.39215710759162903,
      "learning_rate": 2.9238675452774675e-07,
      "loss": 0.6113,
      "step": 22719
    },
    {
      "epoch": 0.941605536905798,
      "grad_norm": 0.4143853783607483,
      "learning_rate": 2.921795349993784e-07,
      "loss": 0.6362,
      "step": 22720
    },
    {
      "epoch": 0.9416469808114717,
      "grad_norm": 0.4640759527683258,
      "learning_rate": 2.9197231547101005e-07,
      "loss": 0.6946,
      "step": 22721
    },
    {
      "epoch": 0.9416884247171453,
      "grad_norm": 0.41405633091926575,
      "learning_rate": 2.9176509594264165e-07,
      "loss": 0.672,
      "step": 22722
    },
    {
      "epoch": 0.941729868622819,
      "grad_norm": 0.42910036444664,
      "learning_rate": 2.915578764142733e-07,
      "loss": 0.6741,
      "step": 22723
    },
    {
      "epoch": 0.9417713125284927,
      "grad_norm": 0.4006488025188446,
      "learning_rate": 2.9135065688590495e-07,
      "loss": 0.6862,
      "step": 22724
    },
    {
      "epoch": 0.9418127564341664,
      "grad_norm": 0.4322017431259155,
      "learning_rate": 2.911434373575366e-07,
      "loss": 0.6354,
      "step": 22725
    },
    {
      "epoch": 0.94185420033984,
      "grad_norm": 0.3877887427806854,
      "learning_rate": 2.9093621782916825e-07,
      "loss": 0.6064,
      "step": 22726
    },
    {
      "epoch": 0.9418956442455138,
      "grad_norm": 0.4459737241268158,
      "learning_rate": 2.907289983007999e-07,
      "loss": 0.6624,
      "step": 22727
    },
    {
      "epoch": 0.9419370881511874,
      "grad_norm": 0.39123016595840454,
      "learning_rate": 2.9052177877243155e-07,
      "loss": 0.6533,
      "step": 22728
    },
    {
      "epoch": 0.941978532056861,
      "grad_norm": 0.4387635290622711,
      "learning_rate": 2.903145592440632e-07,
      "loss": 0.6948,
      "step": 22729
    },
    {
      "epoch": 0.9420199759625347,
      "grad_norm": 0.48579201102256775,
      "learning_rate": 2.9010733971569485e-07,
      "loss": 0.6659,
      "step": 22730
    },
    {
      "epoch": 0.9420614198682083,
      "grad_norm": 0.40345367789268494,
      "learning_rate": 2.899001201873265e-07,
      "loss": 0.6675,
      "step": 22731
    },
    {
      "epoch": 0.9421028637738821,
      "grad_norm": 0.4185834527015686,
      "learning_rate": 2.8969290065895816e-07,
      "loss": 0.6396,
      "step": 22732
    },
    {
      "epoch": 0.9421443076795557,
      "grad_norm": 0.4086865782737732,
      "learning_rate": 2.894856811305898e-07,
      "loss": 0.6692,
      "step": 22733
    },
    {
      "epoch": 0.9421857515852294,
      "grad_norm": 0.3893241882324219,
      "learning_rate": 2.892784616022214e-07,
      "loss": 0.6245,
      "step": 22734
    },
    {
      "epoch": 0.9422271954909031,
      "grad_norm": 0.42727822065353394,
      "learning_rate": 2.8907124207385305e-07,
      "loss": 0.7101,
      "step": 22735
    },
    {
      "epoch": 0.9422686393965767,
      "grad_norm": 0.42357343435287476,
      "learning_rate": 2.888640225454847e-07,
      "loss": 0.7048,
      "step": 22736
    },
    {
      "epoch": 0.9423100833022504,
      "grad_norm": 0.41281503438949585,
      "learning_rate": 2.8865680301711635e-07,
      "loss": 0.6055,
      "step": 22737
    },
    {
      "epoch": 0.942351527207924,
      "grad_norm": 0.39140456914901733,
      "learning_rate": 2.88449583488748e-07,
      "loss": 0.626,
      "step": 22738
    },
    {
      "epoch": 0.9423929711135978,
      "grad_norm": 0.4127078950405121,
      "learning_rate": 2.8824236396037965e-07,
      "loss": 0.6504,
      "step": 22739
    },
    {
      "epoch": 0.9424344150192714,
      "grad_norm": 0.4043368399143219,
      "learning_rate": 2.880351444320113e-07,
      "loss": 0.6653,
      "step": 22740
    },
    {
      "epoch": 0.9424758589249451,
      "grad_norm": 0.4386996030807495,
      "learning_rate": 2.8782792490364296e-07,
      "loss": 0.6967,
      "step": 22741
    },
    {
      "epoch": 0.9425173028306187,
      "grad_norm": 0.3970065712928772,
      "learning_rate": 2.876207053752746e-07,
      "loss": 0.6575,
      "step": 22742
    },
    {
      "epoch": 0.9425587467362925,
      "grad_norm": 0.41672858595848083,
      "learning_rate": 2.8741348584690626e-07,
      "loss": 0.6594,
      "step": 22743
    },
    {
      "epoch": 0.9426001906419661,
      "grad_norm": 0.4536133110523224,
      "learning_rate": 2.872062663185379e-07,
      "loss": 0.693,
      "step": 22744
    },
    {
      "epoch": 0.9426416345476397,
      "grad_norm": 0.42414590716362,
      "learning_rate": 2.8699904679016956e-07,
      "loss": 0.686,
      "step": 22745
    },
    {
      "epoch": 0.9426830784533134,
      "grad_norm": 0.42621704936027527,
      "learning_rate": 2.8679182726180115e-07,
      "loss": 0.6869,
      "step": 22746
    },
    {
      "epoch": 0.9427245223589871,
      "grad_norm": 0.3844846785068512,
      "learning_rate": 2.865846077334328e-07,
      "loss": 0.6677,
      "step": 22747
    },
    {
      "epoch": 0.9427659662646608,
      "grad_norm": 0.4149191677570343,
      "learning_rate": 2.8637738820506446e-07,
      "loss": 0.6179,
      "step": 22748
    },
    {
      "epoch": 0.9428074101703344,
      "grad_norm": 0.41492363810539246,
      "learning_rate": 2.861701686766961e-07,
      "loss": 0.6765,
      "step": 22749
    },
    {
      "epoch": 0.9428488540760082,
      "grad_norm": 0.4472828507423401,
      "learning_rate": 2.8596294914832776e-07,
      "loss": 0.6107,
      "step": 22750
    },
    {
      "epoch": 0.9428902979816818,
      "grad_norm": 0.40962597727775574,
      "learning_rate": 2.857557296199594e-07,
      "loss": 0.7069,
      "step": 22751
    },
    {
      "epoch": 0.9429317418873555,
      "grad_norm": 0.43634462356567383,
      "learning_rate": 2.8554851009159106e-07,
      "loss": 0.6343,
      "step": 22752
    },
    {
      "epoch": 0.9429731857930291,
      "grad_norm": 0.4072112739086151,
      "learning_rate": 2.853412905632227e-07,
      "loss": 0.6438,
      "step": 22753
    },
    {
      "epoch": 0.9430146296987028,
      "grad_norm": 0.42363491654396057,
      "learning_rate": 2.8513407103485436e-07,
      "loss": 0.6327,
      "step": 22754
    },
    {
      "epoch": 0.9430560736043765,
      "grad_norm": 0.4420884847640991,
      "learning_rate": 2.84926851506486e-07,
      "loss": 0.6705,
      "step": 22755
    },
    {
      "epoch": 0.9430975175100501,
      "grad_norm": 0.39288148283958435,
      "learning_rate": 2.8471963197811766e-07,
      "loss": 0.6508,
      "step": 22756
    },
    {
      "epoch": 0.9431389614157238,
      "grad_norm": 0.42284977436065674,
      "learning_rate": 2.845124124497493e-07,
      "loss": 0.6479,
      "step": 22757
    },
    {
      "epoch": 0.9431804053213975,
      "grad_norm": 0.4109274446964264,
      "learning_rate": 2.8430519292138096e-07,
      "loss": 0.702,
      "step": 22758
    },
    {
      "epoch": 0.9432218492270712,
      "grad_norm": 0.3990035653114319,
      "learning_rate": 2.8409797339301256e-07,
      "loss": 0.6489,
      "step": 22759
    },
    {
      "epoch": 0.9432632931327448,
      "grad_norm": 0.41258591413497925,
      "learning_rate": 2.838907538646442e-07,
      "loss": 0.6492,
      "step": 22760
    },
    {
      "epoch": 0.9433047370384186,
      "grad_norm": 0.422921746969223,
      "learning_rate": 2.8368353433627586e-07,
      "loss": 0.6682,
      "step": 22761
    },
    {
      "epoch": 0.9433461809440922,
      "grad_norm": 0.4034198820590973,
      "learning_rate": 2.834763148079075e-07,
      "loss": 0.6427,
      "step": 22762
    },
    {
      "epoch": 0.9433876248497658,
      "grad_norm": 0.4882027506828308,
      "learning_rate": 2.8326909527953916e-07,
      "loss": 0.6904,
      "step": 22763
    },
    {
      "epoch": 0.9434290687554395,
      "grad_norm": 0.4321634769439697,
      "learning_rate": 2.830618757511708e-07,
      "loss": 0.6154,
      "step": 22764
    },
    {
      "epoch": 0.9434705126611131,
      "grad_norm": 0.41709962487220764,
      "learning_rate": 2.8285465622280246e-07,
      "loss": 0.7153,
      "step": 22765
    },
    {
      "epoch": 0.9435119565667869,
      "grad_norm": 0.407401978969574,
      "learning_rate": 2.826474366944341e-07,
      "loss": 0.6119,
      "step": 22766
    },
    {
      "epoch": 0.9435534004724605,
      "grad_norm": 0.3820617198944092,
      "learning_rate": 2.8244021716606576e-07,
      "loss": 0.6637,
      "step": 22767
    },
    {
      "epoch": 0.9435948443781342,
      "grad_norm": 0.41792768239974976,
      "learning_rate": 2.822329976376974e-07,
      "loss": 0.646,
      "step": 22768
    },
    {
      "epoch": 0.9436362882838079,
      "grad_norm": 0.44623687863349915,
      "learning_rate": 2.8202577810932906e-07,
      "loss": 0.7092,
      "step": 22769
    },
    {
      "epoch": 0.9436777321894816,
      "grad_norm": 0.4191036820411682,
      "learning_rate": 2.818185585809607e-07,
      "loss": 0.6536,
      "step": 22770
    },
    {
      "epoch": 0.9437191760951552,
      "grad_norm": 0.43237483501434326,
      "learning_rate": 2.816113390525923e-07,
      "loss": 0.6614,
      "step": 22771
    },
    {
      "epoch": 0.9437606200008288,
      "grad_norm": 0.4314989149570465,
      "learning_rate": 2.8140411952422396e-07,
      "loss": 0.6752,
      "step": 22772
    },
    {
      "epoch": 0.9438020639065026,
      "grad_norm": 0.4296998679637909,
      "learning_rate": 2.811968999958556e-07,
      "loss": 0.6774,
      "step": 22773
    },
    {
      "epoch": 0.9438435078121762,
      "grad_norm": 0.4172271192073822,
      "learning_rate": 2.8098968046748726e-07,
      "loss": 0.658,
      "step": 22774
    },
    {
      "epoch": 0.9438849517178499,
      "grad_norm": 0.4085381031036377,
      "learning_rate": 2.807824609391189e-07,
      "loss": 0.646,
      "step": 22775
    },
    {
      "epoch": 0.9439263956235235,
      "grad_norm": 0.4323885142803192,
      "learning_rate": 2.8057524141075056e-07,
      "loss": 0.7078,
      "step": 22776
    },
    {
      "epoch": 0.9439678395291973,
      "grad_norm": 0.4209352433681488,
      "learning_rate": 2.803680218823822e-07,
      "loss": 0.6238,
      "step": 22777
    },
    {
      "epoch": 0.9440092834348709,
      "grad_norm": 0.4218093752861023,
      "learning_rate": 2.8016080235401386e-07,
      "loss": 0.6708,
      "step": 22778
    },
    {
      "epoch": 0.9440507273405446,
      "grad_norm": 0.45900022983551025,
      "learning_rate": 2.799535828256455e-07,
      "loss": 0.6559,
      "step": 22779
    },
    {
      "epoch": 0.9440921712462182,
      "grad_norm": 0.44049981236457825,
      "learning_rate": 2.7974636329727716e-07,
      "loss": 0.6567,
      "step": 22780
    },
    {
      "epoch": 0.9441336151518919,
      "grad_norm": 0.4077194333076477,
      "learning_rate": 2.795391437689088e-07,
      "loss": 0.6791,
      "step": 22781
    },
    {
      "epoch": 0.9441750590575656,
      "grad_norm": 0.3887229859828949,
      "learning_rate": 2.7933192424054046e-07,
      "loss": 0.6438,
      "step": 22782
    },
    {
      "epoch": 0.9442165029632392,
      "grad_norm": 0.42631518840789795,
      "learning_rate": 2.7912470471217206e-07,
      "loss": 0.6743,
      "step": 22783
    },
    {
      "epoch": 0.944257946868913,
      "grad_norm": 0.3999282121658325,
      "learning_rate": 2.789174851838037e-07,
      "loss": 0.6665,
      "step": 22784
    },
    {
      "epoch": 0.9442993907745866,
      "grad_norm": 0.45306265354156494,
      "learning_rate": 2.7871026565543536e-07,
      "loss": 0.7222,
      "step": 22785
    },
    {
      "epoch": 0.9443408346802603,
      "grad_norm": 0.38545554876327515,
      "learning_rate": 2.78503046127067e-07,
      "loss": 0.6072,
      "step": 22786
    },
    {
      "epoch": 0.9443822785859339,
      "grad_norm": 0.44729822874069214,
      "learning_rate": 2.7829582659869866e-07,
      "loss": 0.6917,
      "step": 22787
    },
    {
      "epoch": 0.9444237224916077,
      "grad_norm": 0.4546288251876831,
      "learning_rate": 2.780886070703303e-07,
      "loss": 0.7275,
      "step": 22788
    },
    {
      "epoch": 0.9444651663972813,
      "grad_norm": 0.44708070158958435,
      "learning_rate": 2.7788138754196196e-07,
      "loss": 0.729,
      "step": 22789
    },
    {
      "epoch": 0.9445066103029549,
      "grad_norm": 0.41593965888023376,
      "learning_rate": 2.776741680135936e-07,
      "loss": 0.7064,
      "step": 22790
    },
    {
      "epoch": 0.9445480542086286,
      "grad_norm": 0.46121135354042053,
      "learning_rate": 2.7746694848522526e-07,
      "loss": 0.7017,
      "step": 22791
    },
    {
      "epoch": 0.9445894981143023,
      "grad_norm": 0.3997088670730591,
      "learning_rate": 2.772597289568569e-07,
      "loss": 0.6292,
      "step": 22792
    },
    {
      "epoch": 0.944630942019976,
      "grad_norm": 0.4015962481498718,
      "learning_rate": 2.7705250942848856e-07,
      "loss": 0.686,
      "step": 22793
    },
    {
      "epoch": 0.9446723859256496,
      "grad_norm": 0.4062272310256958,
      "learning_rate": 2.768452899001202e-07,
      "loss": 0.7415,
      "step": 22794
    },
    {
      "epoch": 0.9447138298313233,
      "grad_norm": 0.42459723353385925,
      "learning_rate": 2.7663807037175186e-07,
      "loss": 0.6528,
      "step": 22795
    },
    {
      "epoch": 0.944755273736997,
      "grad_norm": 0.41581830382347107,
      "learning_rate": 2.764308508433835e-07,
      "loss": 0.6472,
      "step": 22796
    },
    {
      "epoch": 0.9447967176426706,
      "grad_norm": 0.39395421743392944,
      "learning_rate": 2.7622363131501516e-07,
      "loss": 0.601,
      "step": 22797
    },
    {
      "epoch": 0.9448381615483443,
      "grad_norm": 0.4263230860233307,
      "learning_rate": 2.760164117866468e-07,
      "loss": 0.6388,
      "step": 22798
    },
    {
      "epoch": 0.944879605454018,
      "grad_norm": 0.4168474078178406,
      "learning_rate": 2.758091922582784e-07,
      "loss": 0.6658,
      "step": 22799
    },
    {
      "epoch": 0.9449210493596917,
      "grad_norm": 0.38852357864379883,
      "learning_rate": 2.7560197272991006e-07,
      "loss": 0.6072,
      "step": 22800
    },
    {
      "epoch": 0.9449624932653653,
      "grad_norm": 0.4184841215610504,
      "learning_rate": 2.753947532015417e-07,
      "loss": 0.6681,
      "step": 22801
    },
    {
      "epoch": 0.945003937171039,
      "grad_norm": 0.42349672317504883,
      "learning_rate": 2.7518753367317336e-07,
      "loss": 0.6328,
      "step": 22802
    },
    {
      "epoch": 0.9450453810767127,
      "grad_norm": 0.437913715839386,
      "learning_rate": 2.74980314144805e-07,
      "loss": 0.7043,
      "step": 22803
    },
    {
      "epoch": 0.9450868249823864,
      "grad_norm": 0.42048004269599915,
      "learning_rate": 2.7477309461643666e-07,
      "loss": 0.6553,
      "step": 22804
    },
    {
      "epoch": 0.94512826888806,
      "grad_norm": 0.4124428927898407,
      "learning_rate": 2.745658750880683e-07,
      "loss": 0.6924,
      "step": 22805
    },
    {
      "epoch": 0.9451697127937336,
      "grad_norm": 0.42635539174079895,
      "learning_rate": 2.7435865555969996e-07,
      "loss": 0.7002,
      "step": 22806
    },
    {
      "epoch": 0.9452111566994074,
      "grad_norm": 0.4151478111743927,
      "learning_rate": 2.741514360313316e-07,
      "loss": 0.6647,
      "step": 22807
    },
    {
      "epoch": 0.945252600605081,
      "grad_norm": 0.4213005602359772,
      "learning_rate": 2.7394421650296326e-07,
      "loss": 0.725,
      "step": 22808
    },
    {
      "epoch": 0.9452940445107547,
      "grad_norm": 0.4288274645805359,
      "learning_rate": 2.737369969745949e-07,
      "loss": 0.6467,
      "step": 22809
    },
    {
      "epoch": 0.9453354884164283,
      "grad_norm": 0.3863148093223572,
      "learning_rate": 2.7352977744622657e-07,
      "loss": 0.6394,
      "step": 22810
    },
    {
      "epoch": 0.9453769323221021,
      "grad_norm": 0.4247457683086395,
      "learning_rate": 2.733225579178582e-07,
      "loss": 0.6707,
      "step": 22811
    },
    {
      "epoch": 0.9454183762277757,
      "grad_norm": 0.3899017870426178,
      "learning_rate": 2.7311533838948987e-07,
      "loss": 0.6265,
      "step": 22812
    },
    {
      "epoch": 0.9454598201334494,
      "grad_norm": 0.4104098677635193,
      "learning_rate": 2.729081188611215e-07,
      "loss": 0.6155,
      "step": 22813
    },
    {
      "epoch": 0.945501264039123,
      "grad_norm": 0.40642639994621277,
      "learning_rate": 2.7270089933275317e-07,
      "loss": 0.6609,
      "step": 22814
    },
    {
      "epoch": 0.9455427079447967,
      "grad_norm": 0.4625271260738373,
      "learning_rate": 2.724936798043848e-07,
      "loss": 0.7,
      "step": 22815
    },
    {
      "epoch": 0.9455841518504704,
      "grad_norm": 0.40861526131629944,
      "learning_rate": 2.7228646027601647e-07,
      "loss": 0.5981,
      "step": 22816
    },
    {
      "epoch": 0.945625595756144,
      "grad_norm": 0.41150790452957153,
      "learning_rate": 2.7207924074764806e-07,
      "loss": 0.6605,
      "step": 22817
    },
    {
      "epoch": 0.9456670396618178,
      "grad_norm": 0.4292409121990204,
      "learning_rate": 2.718720212192797e-07,
      "loss": 0.6953,
      "step": 22818
    },
    {
      "epoch": 0.9457084835674914,
      "grad_norm": 0.4093433916568756,
      "learning_rate": 2.7166480169091137e-07,
      "loss": 0.6442,
      "step": 22819
    },
    {
      "epoch": 0.9457499274731651,
      "grad_norm": 0.4469381868839264,
      "learning_rate": 2.71457582162543e-07,
      "loss": 0.6758,
      "step": 22820
    },
    {
      "epoch": 0.9457913713788387,
      "grad_norm": 0.4092901945114136,
      "learning_rate": 2.7125036263417467e-07,
      "loss": 0.6187,
      "step": 22821
    },
    {
      "epoch": 0.9458328152845125,
      "grad_norm": 0.43391701579093933,
      "learning_rate": 2.710431431058063e-07,
      "loss": 0.7025,
      "step": 22822
    },
    {
      "epoch": 0.9458742591901861,
      "grad_norm": 0.4669155776500702,
      "learning_rate": 2.7083592357743797e-07,
      "loss": 0.6602,
      "step": 22823
    },
    {
      "epoch": 0.9459157030958597,
      "grad_norm": 0.39093348383903503,
      "learning_rate": 2.706287040490696e-07,
      "loss": 0.5902,
      "step": 22824
    },
    {
      "epoch": 0.9459571470015334,
      "grad_norm": 0.428759902715683,
      "learning_rate": 2.7042148452070127e-07,
      "loss": 0.7198,
      "step": 22825
    },
    {
      "epoch": 0.9459985909072071,
      "grad_norm": 0.485566645860672,
      "learning_rate": 2.702142649923329e-07,
      "loss": 0.7222,
      "step": 22826
    },
    {
      "epoch": 0.9460400348128808,
      "grad_norm": 0.39952725172042847,
      "learning_rate": 2.7000704546396457e-07,
      "loss": 0.6692,
      "step": 22827
    },
    {
      "epoch": 0.9460814787185544,
      "grad_norm": 0.5552158355712891,
      "learning_rate": 2.697998259355962e-07,
      "loss": 0.6833,
      "step": 22828
    },
    {
      "epoch": 0.9461229226242281,
      "grad_norm": 0.4321841597557068,
      "learning_rate": 2.6959260640722787e-07,
      "loss": 0.6433,
      "step": 22829
    },
    {
      "epoch": 0.9461643665299018,
      "grad_norm": 0.4281070828437805,
      "learning_rate": 2.693853868788595e-07,
      "loss": 0.6023,
      "step": 22830
    },
    {
      "epoch": 0.9462058104355755,
      "grad_norm": 0.4176303446292877,
      "learning_rate": 2.6917816735049117e-07,
      "loss": 0.5981,
      "step": 22831
    },
    {
      "epoch": 0.9462472543412491,
      "grad_norm": 0.4318159818649292,
      "learning_rate": 2.6897094782212277e-07,
      "loss": 0.6721,
      "step": 22832
    },
    {
      "epoch": 0.9462886982469227,
      "grad_norm": 0.4278686046600342,
      "learning_rate": 2.687637282937544e-07,
      "loss": 0.6924,
      "step": 22833
    },
    {
      "epoch": 0.9463301421525965,
      "grad_norm": 0.4061988592147827,
      "learning_rate": 2.6855650876538607e-07,
      "loss": 0.6475,
      "step": 22834
    },
    {
      "epoch": 0.9463715860582701,
      "grad_norm": 0.46817705035209656,
      "learning_rate": 2.683492892370177e-07,
      "loss": 0.672,
      "step": 22835
    },
    {
      "epoch": 0.9464130299639438,
      "grad_norm": 0.42470812797546387,
      "learning_rate": 2.6814206970864937e-07,
      "loss": 0.687,
      "step": 22836
    },
    {
      "epoch": 0.9464544738696175,
      "grad_norm": 0.44553080201148987,
      "learning_rate": 2.67934850180281e-07,
      "loss": 0.634,
      "step": 22837
    },
    {
      "epoch": 0.9464959177752912,
      "grad_norm": 0.43136778473854065,
      "learning_rate": 2.6772763065191267e-07,
      "loss": 0.6805,
      "step": 22838
    },
    {
      "epoch": 0.9465373616809648,
      "grad_norm": 0.4268726408481598,
      "learning_rate": 2.675204111235443e-07,
      "loss": 0.7079,
      "step": 22839
    },
    {
      "epoch": 0.9465788055866385,
      "grad_norm": 0.392160564661026,
      "learning_rate": 2.6731319159517597e-07,
      "loss": 0.5809,
      "step": 22840
    },
    {
      "epoch": 0.9466202494923122,
      "grad_norm": 0.3885416090488434,
      "learning_rate": 2.671059720668076e-07,
      "loss": 0.6221,
      "step": 22841
    },
    {
      "epoch": 0.9466616933979858,
      "grad_norm": 0.4072680175304413,
      "learning_rate": 2.6689875253843927e-07,
      "loss": 0.5942,
      "step": 22842
    },
    {
      "epoch": 0.9467031373036595,
      "grad_norm": 0.4582754969596863,
      "learning_rate": 2.666915330100709e-07,
      "loss": 0.663,
      "step": 22843
    },
    {
      "epoch": 0.9467445812093331,
      "grad_norm": 0.4269886016845703,
      "learning_rate": 2.6648431348170257e-07,
      "loss": 0.6903,
      "step": 22844
    },
    {
      "epoch": 0.9467860251150069,
      "grad_norm": 0.4284595847129822,
      "learning_rate": 2.6627709395333417e-07,
      "loss": 0.6603,
      "step": 22845
    },
    {
      "epoch": 0.9468274690206805,
      "grad_norm": 0.42095744609832764,
      "learning_rate": 2.660698744249658e-07,
      "loss": 0.6538,
      "step": 22846
    },
    {
      "epoch": 0.9468689129263542,
      "grad_norm": 0.4087023437023163,
      "learning_rate": 2.6586265489659747e-07,
      "loss": 0.6884,
      "step": 22847
    },
    {
      "epoch": 0.9469103568320278,
      "grad_norm": 0.4188758134841919,
      "learning_rate": 2.656554353682291e-07,
      "loss": 0.6233,
      "step": 22848
    },
    {
      "epoch": 0.9469518007377015,
      "grad_norm": 0.41798171401023865,
      "learning_rate": 2.6544821583986077e-07,
      "loss": 0.6653,
      "step": 22849
    },
    {
      "epoch": 0.9469932446433752,
      "grad_norm": 0.43697234988212585,
      "learning_rate": 2.652409963114924e-07,
      "loss": 0.6388,
      "step": 22850
    },
    {
      "epoch": 0.9470346885490488,
      "grad_norm": 0.47930365800857544,
      "learning_rate": 2.6503377678312407e-07,
      "loss": 0.7336,
      "step": 22851
    },
    {
      "epoch": 0.9470761324547226,
      "grad_norm": 0.40789684653282166,
      "learning_rate": 2.648265572547557e-07,
      "loss": 0.6978,
      "step": 22852
    },
    {
      "epoch": 0.9471175763603962,
      "grad_norm": 0.41941335797309875,
      "learning_rate": 2.6461933772638737e-07,
      "loss": 0.7112,
      "step": 22853
    },
    {
      "epoch": 0.9471590202660699,
      "grad_norm": 0.4139295518398285,
      "learning_rate": 2.64412118198019e-07,
      "loss": 0.624,
      "step": 22854
    },
    {
      "epoch": 0.9472004641717435,
      "grad_norm": 0.3950515687465668,
      "learning_rate": 2.6420489866965067e-07,
      "loss": 0.6213,
      "step": 22855
    },
    {
      "epoch": 0.9472419080774173,
      "grad_norm": 0.40832066535949707,
      "learning_rate": 2.639976791412823e-07,
      "loss": 0.671,
      "step": 22856
    },
    {
      "epoch": 0.9472833519830909,
      "grad_norm": 0.4133493900299072,
      "learning_rate": 2.637904596129139e-07,
      "loss": 0.6117,
      "step": 22857
    },
    {
      "epoch": 0.9473247958887645,
      "grad_norm": 0.40054136514663696,
      "learning_rate": 2.6358324008454557e-07,
      "loss": 0.6437,
      "step": 22858
    },
    {
      "epoch": 0.9473662397944382,
      "grad_norm": 0.42023786902427673,
      "learning_rate": 2.633760205561772e-07,
      "loss": 0.7009,
      "step": 22859
    },
    {
      "epoch": 0.9474076837001119,
      "grad_norm": 0.4199494421482086,
      "learning_rate": 2.6316880102780887e-07,
      "loss": 0.6765,
      "step": 22860
    },
    {
      "epoch": 0.9474491276057856,
      "grad_norm": 0.4012482166290283,
      "learning_rate": 2.629615814994405e-07,
      "loss": 0.6699,
      "step": 22861
    },
    {
      "epoch": 0.9474905715114592,
      "grad_norm": 0.3931578993797302,
      "learning_rate": 2.6275436197107217e-07,
      "loss": 0.668,
      "step": 22862
    },
    {
      "epoch": 0.947532015417133,
      "grad_norm": 0.41210150718688965,
      "learning_rate": 2.625471424427038e-07,
      "loss": 0.7166,
      "step": 22863
    },
    {
      "epoch": 0.9475734593228066,
      "grad_norm": 0.4258553981781006,
      "learning_rate": 2.6233992291433547e-07,
      "loss": 0.7047,
      "step": 22864
    },
    {
      "epoch": 0.9476149032284803,
      "grad_norm": 0.40003249049186707,
      "learning_rate": 2.621327033859671e-07,
      "loss": 0.6443,
      "step": 22865
    },
    {
      "epoch": 0.9476563471341539,
      "grad_norm": 0.4364911913871765,
      "learning_rate": 2.6192548385759877e-07,
      "loss": 0.6741,
      "step": 22866
    },
    {
      "epoch": 0.9476977910398275,
      "grad_norm": 0.44728341698646545,
      "learning_rate": 2.617182643292304e-07,
      "loss": 0.6799,
      "step": 22867
    },
    {
      "epoch": 0.9477392349455013,
      "grad_norm": 0.43094602227211,
      "learning_rate": 2.615110448008621e-07,
      "loss": 0.6697,
      "step": 22868
    },
    {
      "epoch": 0.9477806788511749,
      "grad_norm": 0.4405622184276581,
      "learning_rate": 2.6130382527249367e-07,
      "loss": 0.6689,
      "step": 22869
    },
    {
      "epoch": 0.9478221227568486,
      "grad_norm": 0.4337182343006134,
      "learning_rate": 2.610966057441253e-07,
      "loss": 0.6938,
      "step": 22870
    },
    {
      "epoch": 0.9478635666625223,
      "grad_norm": 0.4185543656349182,
      "learning_rate": 2.6088938621575697e-07,
      "loss": 0.6766,
      "step": 22871
    },
    {
      "epoch": 0.947905010568196,
      "grad_norm": 0.42378294467926025,
      "learning_rate": 2.606821666873886e-07,
      "loss": 0.6145,
      "step": 22872
    },
    {
      "epoch": 0.9479464544738696,
      "grad_norm": 0.36861488223075867,
      "learning_rate": 2.6047494715902027e-07,
      "loss": 0.6483,
      "step": 22873
    },
    {
      "epoch": 0.9479878983795433,
      "grad_norm": 0.4233914315700531,
      "learning_rate": 2.602677276306519e-07,
      "loss": 0.7084,
      "step": 22874
    },
    {
      "epoch": 0.948029342285217,
      "grad_norm": 0.40444132685661316,
      "learning_rate": 2.6006050810228357e-07,
      "loss": 0.6547,
      "step": 22875
    },
    {
      "epoch": 0.9480707861908906,
      "grad_norm": 0.4329718053340912,
      "learning_rate": 2.598532885739152e-07,
      "loss": 0.667,
      "step": 22876
    },
    {
      "epoch": 0.9481122300965643,
      "grad_norm": 0.40779465436935425,
      "learning_rate": 2.596460690455469e-07,
      "loss": 0.6305,
      "step": 22877
    },
    {
      "epoch": 0.9481536740022379,
      "grad_norm": 0.3921043276786804,
      "learning_rate": 2.594388495171785e-07,
      "loss": 0.6624,
      "step": 22878
    },
    {
      "epoch": 0.9481951179079117,
      "grad_norm": 0.4168822169303894,
      "learning_rate": 2.592316299888102e-07,
      "loss": 0.681,
      "step": 22879
    },
    {
      "epoch": 0.9482365618135853,
      "grad_norm": 0.42491114139556885,
      "learning_rate": 2.590244104604418e-07,
      "loss": 0.6523,
      "step": 22880
    },
    {
      "epoch": 0.948278005719259,
      "grad_norm": 0.4007750153541565,
      "learning_rate": 2.588171909320735e-07,
      "loss": 0.6335,
      "step": 22881
    },
    {
      "epoch": 0.9483194496249326,
      "grad_norm": 0.4305400848388672,
      "learning_rate": 2.5860997140370507e-07,
      "loss": 0.7456,
      "step": 22882
    },
    {
      "epoch": 0.9483608935306064,
      "grad_norm": 0.4000861644744873,
      "learning_rate": 2.584027518753367e-07,
      "loss": 0.6755,
      "step": 22883
    },
    {
      "epoch": 0.94840233743628,
      "grad_norm": 0.43147340416908264,
      "learning_rate": 2.5819553234696837e-07,
      "loss": 0.7332,
      "step": 22884
    },
    {
      "epoch": 0.9484437813419536,
      "grad_norm": 0.47997432947158813,
      "learning_rate": 2.579883128186e-07,
      "loss": 0.6685,
      "step": 22885
    },
    {
      "epoch": 0.9484852252476274,
      "grad_norm": 0.40064743161201477,
      "learning_rate": 2.577810932902317e-07,
      "loss": 0.6028,
      "step": 22886
    },
    {
      "epoch": 0.948526669153301,
      "grad_norm": 0.4713772237300873,
      "learning_rate": 2.575738737618633e-07,
      "loss": 0.7024,
      "step": 22887
    },
    {
      "epoch": 0.9485681130589747,
      "grad_norm": 0.3695751130580902,
      "learning_rate": 2.57366654233495e-07,
      "loss": 0.6147,
      "step": 22888
    },
    {
      "epoch": 0.9486095569646483,
      "grad_norm": 0.40204527974128723,
      "learning_rate": 2.571594347051266e-07,
      "loss": 0.6278,
      "step": 22889
    },
    {
      "epoch": 0.9486510008703221,
      "grad_norm": 0.3713224530220032,
      "learning_rate": 2.569522151767583e-07,
      "loss": 0.598,
      "step": 22890
    },
    {
      "epoch": 0.9486924447759957,
      "grad_norm": 0.44002947211265564,
      "learning_rate": 2.567449956483899e-07,
      "loss": 0.6479,
      "step": 22891
    },
    {
      "epoch": 0.9487338886816694,
      "grad_norm": 0.3893745243549347,
      "learning_rate": 2.565377761200216e-07,
      "loss": 0.6149,
      "step": 22892
    },
    {
      "epoch": 0.948775332587343,
      "grad_norm": 0.4052067697048187,
      "learning_rate": 2.5633055659165323e-07,
      "loss": 0.6476,
      "step": 22893
    },
    {
      "epoch": 0.9488167764930167,
      "grad_norm": 0.4145209491252899,
      "learning_rate": 2.561233370632848e-07,
      "loss": 0.7284,
      "step": 22894
    },
    {
      "epoch": 0.9488582203986904,
      "grad_norm": 0.3993420898914337,
      "learning_rate": 2.559161175349165e-07,
      "loss": 0.6791,
      "step": 22895
    },
    {
      "epoch": 0.948899664304364,
      "grad_norm": 0.42741960287094116,
      "learning_rate": 2.557088980065481e-07,
      "loss": 0.6432,
      "step": 22896
    },
    {
      "epoch": 0.9489411082100377,
      "grad_norm": 0.43356314301490784,
      "learning_rate": 2.555016784781798e-07,
      "loss": 0.714,
      "step": 22897
    },
    {
      "epoch": 0.9489825521157114,
      "grad_norm": 0.4041145443916321,
      "learning_rate": 2.552944589498114e-07,
      "loss": 0.6562,
      "step": 22898
    },
    {
      "epoch": 0.9490239960213851,
      "grad_norm": 0.40037861466407776,
      "learning_rate": 2.550872394214431e-07,
      "loss": 0.615,
      "step": 22899
    },
    {
      "epoch": 0.9490654399270587,
      "grad_norm": 0.4392162263393402,
      "learning_rate": 2.548800198930747e-07,
      "loss": 0.6332,
      "step": 22900
    },
    {
      "epoch": 0.9491068838327325,
      "grad_norm": 0.40276139974594116,
      "learning_rate": 2.546728003647064e-07,
      "loss": 0.6669,
      "step": 22901
    },
    {
      "epoch": 0.9491483277384061,
      "grad_norm": 0.3904343247413635,
      "learning_rate": 2.5446558083633803e-07,
      "loss": 0.6271,
      "step": 22902
    },
    {
      "epoch": 0.9491897716440797,
      "grad_norm": 0.4627732038497925,
      "learning_rate": 2.542583613079697e-07,
      "loss": 0.6963,
      "step": 22903
    },
    {
      "epoch": 0.9492312155497534,
      "grad_norm": 0.39827394485473633,
      "learning_rate": 2.5405114177960133e-07,
      "loss": 0.615,
      "step": 22904
    },
    {
      "epoch": 0.949272659455427,
      "grad_norm": 0.4118174612522125,
      "learning_rate": 2.53843922251233e-07,
      "loss": 0.6204,
      "step": 22905
    },
    {
      "epoch": 0.9493141033611008,
      "grad_norm": 0.4250461459159851,
      "learning_rate": 2.5363670272286463e-07,
      "loss": 0.6661,
      "step": 22906
    },
    {
      "epoch": 0.9493555472667744,
      "grad_norm": 0.4134157598018646,
      "learning_rate": 2.534294831944963e-07,
      "loss": 0.6985,
      "step": 22907
    },
    {
      "epoch": 0.9493969911724481,
      "grad_norm": 0.4255031645298004,
      "learning_rate": 2.5322226366612793e-07,
      "loss": 0.6219,
      "step": 22908
    },
    {
      "epoch": 0.9494384350781218,
      "grad_norm": 0.42101195454597473,
      "learning_rate": 2.530150441377596e-07,
      "loss": 0.6514,
      "step": 22909
    },
    {
      "epoch": 0.9494798789837954,
      "grad_norm": 0.43490487337112427,
      "learning_rate": 2.5280782460939123e-07,
      "loss": 0.6541,
      "step": 22910
    },
    {
      "epoch": 0.9495213228894691,
      "grad_norm": 0.4604805111885071,
      "learning_rate": 2.526006050810229e-07,
      "loss": 0.6451,
      "step": 22911
    },
    {
      "epoch": 0.9495627667951427,
      "grad_norm": 0.3938830494880676,
      "learning_rate": 2.523933855526545e-07,
      "loss": 0.6542,
      "step": 22912
    },
    {
      "epoch": 0.9496042107008165,
      "grad_norm": 0.42697417736053467,
      "learning_rate": 2.5218616602428613e-07,
      "loss": 0.6995,
      "step": 22913
    },
    {
      "epoch": 0.9496456546064901,
      "grad_norm": 0.4559043347835541,
      "learning_rate": 2.519789464959178e-07,
      "loss": 0.6899,
      "step": 22914
    },
    {
      "epoch": 0.9496870985121638,
      "grad_norm": 0.41328009963035583,
      "learning_rate": 2.5177172696754943e-07,
      "loss": 0.6429,
      "step": 22915
    },
    {
      "epoch": 0.9497285424178374,
      "grad_norm": 0.4519127905368805,
      "learning_rate": 2.515645074391811e-07,
      "loss": 0.6907,
      "step": 22916
    },
    {
      "epoch": 0.9497699863235112,
      "grad_norm": 0.4523499011993408,
      "learning_rate": 2.5135728791081273e-07,
      "loss": 0.7476,
      "step": 22917
    },
    {
      "epoch": 0.9498114302291848,
      "grad_norm": 0.4186107814311981,
      "learning_rate": 2.511500683824444e-07,
      "loss": 0.6885,
      "step": 22918
    },
    {
      "epoch": 0.9498528741348584,
      "grad_norm": 0.4260886311531067,
      "learning_rate": 2.5094284885407603e-07,
      "loss": 0.7083,
      "step": 22919
    },
    {
      "epoch": 0.9498943180405321,
      "grad_norm": 0.417978972196579,
      "learning_rate": 2.507356293257077e-07,
      "loss": 0.6451,
      "step": 22920
    },
    {
      "epoch": 0.9499357619462058,
      "grad_norm": 0.4097512662410736,
      "learning_rate": 2.5052840979733933e-07,
      "loss": 0.6539,
      "step": 22921
    },
    {
      "epoch": 0.9499772058518795,
      "grad_norm": 0.40072301030158997,
      "learning_rate": 2.50321190268971e-07,
      "loss": 0.6523,
      "step": 22922
    },
    {
      "epoch": 0.9500186497575531,
      "grad_norm": 0.4339374303817749,
      "learning_rate": 2.5011397074060263e-07,
      "loss": 0.6506,
      "step": 22923
    },
    {
      "epoch": 0.9500600936632269,
      "grad_norm": 0.4084415137767792,
      "learning_rate": 2.499067512122343e-07,
      "loss": 0.6705,
      "step": 22924
    },
    {
      "epoch": 0.9501015375689005,
      "grad_norm": 0.4454505741596222,
      "learning_rate": 2.4969953168386593e-07,
      "loss": 0.6506,
      "step": 22925
    },
    {
      "epoch": 0.9501429814745742,
      "grad_norm": 0.40532633662223816,
      "learning_rate": 2.494923121554976e-07,
      "loss": 0.6613,
      "step": 22926
    },
    {
      "epoch": 0.9501844253802478,
      "grad_norm": 0.37040647864341736,
      "learning_rate": 2.4928509262712923e-07,
      "loss": 0.6329,
      "step": 22927
    },
    {
      "epoch": 0.9502258692859215,
      "grad_norm": 0.3976791799068451,
      "learning_rate": 2.490778730987609e-07,
      "loss": 0.6458,
      "step": 22928
    },
    {
      "epoch": 0.9502673131915952,
      "grad_norm": 0.3973597288131714,
      "learning_rate": 2.4887065357039253e-07,
      "loss": 0.614,
      "step": 22929
    },
    {
      "epoch": 0.9503087570972688,
      "grad_norm": 0.41508814692497253,
      "learning_rate": 2.4866343404202413e-07,
      "loss": 0.6534,
      "step": 22930
    },
    {
      "epoch": 0.9503502010029425,
      "grad_norm": 0.4126274287700653,
      "learning_rate": 2.484562145136558e-07,
      "loss": 0.6375,
      "step": 22931
    },
    {
      "epoch": 0.9503916449086162,
      "grad_norm": 0.43353089690208435,
      "learning_rate": 2.4824899498528743e-07,
      "loss": 0.7323,
      "step": 22932
    },
    {
      "epoch": 0.9504330888142899,
      "grad_norm": 0.4166928231716156,
      "learning_rate": 2.480417754569191e-07,
      "loss": 0.6677,
      "step": 22933
    },
    {
      "epoch": 0.9504745327199635,
      "grad_norm": 0.4163173735141754,
      "learning_rate": 2.4783455592855073e-07,
      "loss": 0.6666,
      "step": 22934
    },
    {
      "epoch": 0.9505159766256372,
      "grad_norm": 0.44188305735588074,
      "learning_rate": 2.476273364001824e-07,
      "loss": 0.6826,
      "step": 22935
    },
    {
      "epoch": 0.9505574205313109,
      "grad_norm": 0.4279739260673523,
      "learning_rate": 2.4742011687181403e-07,
      "loss": 0.6399,
      "step": 22936
    },
    {
      "epoch": 0.9505988644369845,
      "grad_norm": 0.4257028102874756,
      "learning_rate": 2.472128973434457e-07,
      "loss": 0.6783,
      "step": 22937
    },
    {
      "epoch": 0.9506403083426582,
      "grad_norm": 0.4130859076976776,
      "learning_rate": 2.4700567781507733e-07,
      "loss": 0.699,
      "step": 22938
    },
    {
      "epoch": 0.9506817522483318,
      "grad_norm": 0.39376822113990784,
      "learning_rate": 2.46798458286709e-07,
      "loss": 0.6729,
      "step": 22939
    },
    {
      "epoch": 0.9507231961540056,
      "grad_norm": 0.42714056372642517,
      "learning_rate": 2.4659123875834063e-07,
      "loss": 0.642,
      "step": 22940
    },
    {
      "epoch": 0.9507646400596792,
      "grad_norm": 0.43861863017082214,
      "learning_rate": 2.463840192299723e-07,
      "loss": 0.7269,
      "step": 22941
    },
    {
      "epoch": 0.9508060839653529,
      "grad_norm": 0.42031535506248474,
      "learning_rate": 2.4617679970160394e-07,
      "loss": 0.6306,
      "step": 22942
    },
    {
      "epoch": 0.9508475278710266,
      "grad_norm": 0.3881395161151886,
      "learning_rate": 2.4596958017323553e-07,
      "loss": 0.6467,
      "step": 22943
    },
    {
      "epoch": 0.9508889717767003,
      "grad_norm": 0.46742671728134155,
      "learning_rate": 2.457623606448672e-07,
      "loss": 0.7313,
      "step": 22944
    },
    {
      "epoch": 0.9509304156823739,
      "grad_norm": 0.47064971923828125,
      "learning_rate": 2.4555514111649883e-07,
      "loss": 0.6967,
      "step": 22945
    },
    {
      "epoch": 0.9509718595880475,
      "grad_norm": 0.3907201886177063,
      "learning_rate": 2.453479215881305e-07,
      "loss": 0.6646,
      "step": 22946
    },
    {
      "epoch": 0.9510133034937213,
      "grad_norm": 0.4067206382751465,
      "learning_rate": 2.4514070205976213e-07,
      "loss": 0.675,
      "step": 22947
    },
    {
      "epoch": 0.9510547473993949,
      "grad_norm": 0.4150945246219635,
      "learning_rate": 2.449334825313938e-07,
      "loss": 0.6254,
      "step": 22948
    },
    {
      "epoch": 0.9510961913050686,
      "grad_norm": 0.4254530668258667,
      "learning_rate": 2.4472626300302543e-07,
      "loss": 0.6643,
      "step": 22949
    },
    {
      "epoch": 0.9511376352107422,
      "grad_norm": 0.44320616126060486,
      "learning_rate": 2.445190434746571e-07,
      "loss": 0.7347,
      "step": 22950
    },
    {
      "epoch": 0.951179079116416,
      "grad_norm": 0.40031999349594116,
      "learning_rate": 2.4431182394628874e-07,
      "loss": 0.6725,
      "step": 22951
    },
    {
      "epoch": 0.9512205230220896,
      "grad_norm": 0.41472044587135315,
      "learning_rate": 2.441046044179204e-07,
      "loss": 0.6758,
      "step": 22952
    },
    {
      "epoch": 0.9512619669277633,
      "grad_norm": 0.45670685172080994,
      "learning_rate": 2.4389738488955204e-07,
      "loss": 0.698,
      "step": 22953
    },
    {
      "epoch": 0.951303410833437,
      "grad_norm": 0.4573492705821991,
      "learning_rate": 2.436901653611837e-07,
      "loss": 0.6726,
      "step": 22954
    },
    {
      "epoch": 0.9513448547391106,
      "grad_norm": 0.4401511251926422,
      "learning_rate": 2.434829458328153e-07,
      "loss": 0.692,
      "step": 22955
    },
    {
      "epoch": 0.9513862986447843,
      "grad_norm": 0.4167815148830414,
      "learning_rate": 2.4327572630444693e-07,
      "loss": 0.6638,
      "step": 22956
    },
    {
      "epoch": 0.9514277425504579,
      "grad_norm": 0.44382917881011963,
      "learning_rate": 2.430685067760786e-07,
      "loss": 0.6699,
      "step": 22957
    },
    {
      "epoch": 0.9514691864561317,
      "grad_norm": 0.4075579047203064,
      "learning_rate": 2.4286128724771023e-07,
      "loss": 0.6155,
      "step": 22958
    },
    {
      "epoch": 0.9515106303618053,
      "grad_norm": 0.436689555644989,
      "learning_rate": 2.426540677193419e-07,
      "loss": 0.738,
      "step": 22959
    },
    {
      "epoch": 0.951552074267479,
      "grad_norm": 0.4287460446357727,
      "learning_rate": 2.4244684819097354e-07,
      "loss": 0.6866,
      "step": 22960
    },
    {
      "epoch": 0.9515935181731526,
      "grad_norm": 0.419484406709671,
      "learning_rate": 2.422396286626052e-07,
      "loss": 0.6858,
      "step": 22961
    },
    {
      "epoch": 0.9516349620788264,
      "grad_norm": 0.40869638323783875,
      "learning_rate": 2.4203240913423684e-07,
      "loss": 0.666,
      "step": 22962
    },
    {
      "epoch": 0.9516764059845,
      "grad_norm": 0.4054705500602722,
      "learning_rate": 2.418251896058685e-07,
      "loss": 0.6129,
      "step": 22963
    },
    {
      "epoch": 0.9517178498901736,
      "grad_norm": 0.435607373714447,
      "learning_rate": 2.4161797007750014e-07,
      "loss": 0.6481,
      "step": 22964
    },
    {
      "epoch": 0.9517592937958473,
      "grad_norm": 0.3902180790901184,
      "learning_rate": 2.414107505491318e-07,
      "loss": 0.634,
      "step": 22965
    },
    {
      "epoch": 0.951800737701521,
      "grad_norm": 0.447815865278244,
      "learning_rate": 2.4120353102076344e-07,
      "loss": 0.6857,
      "step": 22966
    },
    {
      "epoch": 0.9518421816071947,
      "grad_norm": 0.4189189076423645,
      "learning_rate": 2.409963114923951e-07,
      "loss": 0.6213,
      "step": 22967
    },
    {
      "epoch": 0.9518836255128683,
      "grad_norm": 0.4183293282985687,
      "learning_rate": 2.407890919640267e-07,
      "loss": 0.6366,
      "step": 22968
    },
    {
      "epoch": 0.951925069418542,
      "grad_norm": 0.4004623591899872,
      "learning_rate": 2.4058187243565834e-07,
      "loss": 0.6526,
      "step": 22969
    },
    {
      "epoch": 0.9519665133242157,
      "grad_norm": 0.38660067319869995,
      "learning_rate": 2.4037465290729e-07,
      "loss": 0.6337,
      "step": 22970
    },
    {
      "epoch": 0.9520079572298893,
      "grad_norm": 0.40654176473617554,
      "learning_rate": 2.4016743337892164e-07,
      "loss": 0.6285,
      "step": 22971
    },
    {
      "epoch": 0.952049401135563,
      "grad_norm": 0.42269065976142883,
      "learning_rate": 2.399602138505533e-07,
      "loss": 0.6378,
      "step": 22972
    },
    {
      "epoch": 0.9520908450412366,
      "grad_norm": 0.3705494999885559,
      "learning_rate": 2.3975299432218494e-07,
      "loss": 0.6172,
      "step": 22973
    },
    {
      "epoch": 0.9521322889469104,
      "grad_norm": 0.41681766510009766,
      "learning_rate": 2.395457747938166e-07,
      "loss": 0.6454,
      "step": 22974
    },
    {
      "epoch": 0.952173732852584,
      "grad_norm": 0.40389660000801086,
      "learning_rate": 2.3933855526544824e-07,
      "loss": 0.7,
      "step": 22975
    },
    {
      "epoch": 0.9522151767582577,
      "grad_norm": 0.4558272659778595,
      "learning_rate": 2.391313357370799e-07,
      "loss": 0.6421,
      "step": 22976
    },
    {
      "epoch": 0.9522566206639314,
      "grad_norm": 0.49605125188827515,
      "learning_rate": 2.3892411620871154e-07,
      "loss": 0.7195,
      "step": 22977
    },
    {
      "epoch": 0.9522980645696051,
      "grad_norm": 0.395453542470932,
      "learning_rate": 2.387168966803432e-07,
      "loss": 0.6638,
      "step": 22978
    },
    {
      "epoch": 0.9523395084752787,
      "grad_norm": 0.39343881607055664,
      "learning_rate": 2.3850967715197484e-07,
      "loss": 0.6213,
      "step": 22979
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 0.45467105507850647,
      "learning_rate": 2.3830245762360646e-07,
      "loss": 0.6516,
      "step": 22980
    },
    {
      "epoch": 0.9524223962866261,
      "grad_norm": 0.45640087127685547,
      "learning_rate": 2.3809523809523811e-07,
      "loss": 0.71,
      "step": 22981
    },
    {
      "epoch": 0.9524638401922997,
      "grad_norm": 0.44020524621009827,
      "learning_rate": 2.3788801856686976e-07,
      "loss": 0.7397,
      "step": 22982
    },
    {
      "epoch": 0.9525052840979734,
      "grad_norm": 0.42731040716171265,
      "learning_rate": 2.376807990385014e-07,
      "loss": 0.6517,
      "step": 22983
    },
    {
      "epoch": 0.952546728003647,
      "grad_norm": 0.4042796492576599,
      "learning_rate": 2.3747357951013304e-07,
      "loss": 0.6093,
      "step": 22984
    },
    {
      "epoch": 0.9525881719093208,
      "grad_norm": 0.40690121054649353,
      "learning_rate": 2.372663599817647e-07,
      "loss": 0.6555,
      "step": 22985
    },
    {
      "epoch": 0.9526296158149944,
      "grad_norm": 0.4129328429698944,
      "learning_rate": 2.3705914045339634e-07,
      "loss": 0.6495,
      "step": 22986
    },
    {
      "epoch": 0.9526710597206681,
      "grad_norm": 0.41668522357940674,
      "learning_rate": 2.36851920925028e-07,
      "loss": 0.6602,
      "step": 22987
    },
    {
      "epoch": 0.9527125036263417,
      "grad_norm": 0.40695318579673767,
      "learning_rate": 2.3664470139665964e-07,
      "loss": 0.6689,
      "step": 22988
    },
    {
      "epoch": 0.9527539475320154,
      "grad_norm": 0.4231056272983551,
      "learning_rate": 2.364374818682913e-07,
      "loss": 0.6636,
      "step": 22989
    },
    {
      "epoch": 0.9527953914376891,
      "grad_norm": 0.4279882311820984,
      "learning_rate": 2.3623026233992294e-07,
      "loss": 0.6952,
      "step": 22990
    },
    {
      "epoch": 0.9528368353433627,
      "grad_norm": 0.47367140650749207,
      "learning_rate": 2.360230428115546e-07,
      "loss": 0.6721,
      "step": 22991
    },
    {
      "epoch": 0.9528782792490365,
      "grad_norm": 0.39083296060562134,
      "learning_rate": 2.3581582328318621e-07,
      "loss": 0.6801,
      "step": 22992
    },
    {
      "epoch": 0.9529197231547101,
      "grad_norm": 0.4076195955276489,
      "learning_rate": 2.3560860375481786e-07,
      "loss": 0.6808,
      "step": 22993
    },
    {
      "epoch": 0.9529611670603838,
      "grad_norm": 0.3925579786300659,
      "learning_rate": 2.3540138422644952e-07,
      "loss": 0.613,
      "step": 22994
    },
    {
      "epoch": 0.9530026109660574,
      "grad_norm": 0.418381005525589,
      "learning_rate": 2.3519416469808117e-07,
      "loss": 0.6672,
      "step": 22995
    },
    {
      "epoch": 0.9530440548717312,
      "grad_norm": 0.4180576205253601,
      "learning_rate": 2.3498694516971282e-07,
      "loss": 0.65,
      "step": 22996
    },
    {
      "epoch": 0.9530854987774048,
      "grad_norm": 0.44758865237236023,
      "learning_rate": 2.3477972564134447e-07,
      "loss": 0.6959,
      "step": 22997
    },
    {
      "epoch": 0.9531269426830784,
      "grad_norm": 0.41935187578201294,
      "learning_rate": 2.3457250611297612e-07,
      "loss": 0.6936,
      "step": 22998
    },
    {
      "epoch": 0.9531683865887521,
      "grad_norm": 0.41272836923599243,
      "learning_rate": 2.3436528658460777e-07,
      "loss": 0.6594,
      "step": 22999
    },
    {
      "epoch": 0.9532098304944258,
      "grad_norm": 0.4224584698677063,
      "learning_rate": 2.3415806705623942e-07,
      "loss": 0.6449,
      "step": 23000
    },
    {
      "epoch": 0.9532512744000995,
      "grad_norm": 0.4248402416706085,
      "learning_rate": 2.3395084752787104e-07,
      "loss": 0.6957,
      "step": 23001
    },
    {
      "epoch": 0.9532927183057731,
      "grad_norm": 0.4413665235042572,
      "learning_rate": 2.337436279995027e-07,
      "loss": 0.7336,
      "step": 23002
    },
    {
      "epoch": 0.9533341622114468,
      "grad_norm": 0.4118202030658722,
      "learning_rate": 2.3353640847113434e-07,
      "loss": 0.6589,
      "step": 23003
    },
    {
      "epoch": 0.9533756061171205,
      "grad_norm": 0.4617410898208618,
      "learning_rate": 2.33329188942766e-07,
      "loss": 0.687,
      "step": 23004
    },
    {
      "epoch": 0.9534170500227942,
      "grad_norm": 0.3688286244869232,
      "learning_rate": 2.3312196941439762e-07,
      "loss": 0.6215,
      "step": 23005
    },
    {
      "epoch": 0.9534584939284678,
      "grad_norm": 0.4183925688266754,
      "learning_rate": 2.3291474988602927e-07,
      "loss": 0.6309,
      "step": 23006
    },
    {
      "epoch": 0.9534999378341414,
      "grad_norm": 0.4117042124271393,
      "learning_rate": 2.3270753035766092e-07,
      "loss": 0.6705,
      "step": 23007
    },
    {
      "epoch": 0.9535413817398152,
      "grad_norm": 0.4411187469959259,
      "learning_rate": 2.3250031082929257e-07,
      "loss": 0.6984,
      "step": 23008
    },
    {
      "epoch": 0.9535828256454888,
      "grad_norm": 0.4096396565437317,
      "learning_rate": 2.3229309130092422e-07,
      "loss": 0.6514,
      "step": 23009
    },
    {
      "epoch": 0.9536242695511625,
      "grad_norm": 0.40306371450424194,
      "learning_rate": 2.3208587177255587e-07,
      "loss": 0.625,
      "step": 23010
    },
    {
      "epoch": 0.9536657134568362,
      "grad_norm": 0.457781583070755,
      "learning_rate": 2.3187865224418752e-07,
      "loss": 0.699,
      "step": 23011
    },
    {
      "epoch": 0.9537071573625099,
      "grad_norm": 0.38474053144454956,
      "learning_rate": 2.3167143271581917e-07,
      "loss": 0.6302,
      "step": 23012
    },
    {
      "epoch": 0.9537486012681835,
      "grad_norm": 0.38026633858680725,
      "learning_rate": 2.3146421318745082e-07,
      "loss": 0.6575,
      "step": 23013
    },
    {
      "epoch": 0.9537900451738572,
      "grad_norm": 0.4118310511112213,
      "learning_rate": 2.3125699365908247e-07,
      "loss": 0.6543,
      "step": 23014
    },
    {
      "epoch": 0.9538314890795309,
      "grad_norm": 0.40507233142852783,
      "learning_rate": 2.3104977413071412e-07,
      "loss": 0.6051,
      "step": 23015
    },
    {
      "epoch": 0.9538729329852045,
      "grad_norm": 0.4533459544181824,
      "learning_rate": 2.3084255460234577e-07,
      "loss": 0.7201,
      "step": 23016
    },
    {
      "epoch": 0.9539143768908782,
      "grad_norm": 0.4498926103115082,
      "learning_rate": 2.3063533507397737e-07,
      "loss": 0.6963,
      "step": 23017
    },
    {
      "epoch": 0.9539558207965518,
      "grad_norm": 0.4431069791316986,
      "learning_rate": 2.3042811554560902e-07,
      "loss": 0.6622,
      "step": 23018
    },
    {
      "epoch": 0.9539972647022256,
      "grad_norm": 0.4086289703845978,
      "learning_rate": 2.3022089601724067e-07,
      "loss": 0.6536,
      "step": 23019
    },
    {
      "epoch": 0.9540387086078992,
      "grad_norm": 0.4287601113319397,
      "learning_rate": 2.3001367648887232e-07,
      "loss": 0.6681,
      "step": 23020
    },
    {
      "epoch": 0.9540801525135729,
      "grad_norm": 0.48039767146110535,
      "learning_rate": 2.2980645696050397e-07,
      "loss": 0.6984,
      "step": 23021
    },
    {
      "epoch": 0.9541215964192465,
      "grad_norm": 0.42725619673728943,
      "learning_rate": 2.2959923743213562e-07,
      "loss": 0.6406,
      "step": 23022
    },
    {
      "epoch": 0.9541630403249203,
      "grad_norm": 0.4328678250312805,
      "learning_rate": 2.2939201790376727e-07,
      "loss": 0.6418,
      "step": 23023
    },
    {
      "epoch": 0.9542044842305939,
      "grad_norm": 0.4361988306045532,
      "learning_rate": 2.2918479837539892e-07,
      "loss": 0.6851,
      "step": 23024
    },
    {
      "epoch": 0.9542459281362675,
      "grad_norm": 0.43480053544044495,
      "learning_rate": 2.2897757884703057e-07,
      "loss": 0.6946,
      "step": 23025
    },
    {
      "epoch": 0.9542873720419413,
      "grad_norm": 0.40493062138557434,
      "learning_rate": 2.2877035931866222e-07,
      "loss": 0.6211,
      "step": 23026
    },
    {
      "epoch": 0.9543288159476149,
      "grad_norm": 0.41657501459121704,
      "learning_rate": 2.2856313979029387e-07,
      "loss": 0.6969,
      "step": 23027
    },
    {
      "epoch": 0.9543702598532886,
      "grad_norm": 0.4044494032859802,
      "learning_rate": 2.2835592026192552e-07,
      "loss": 0.6205,
      "step": 23028
    },
    {
      "epoch": 0.9544117037589622,
      "grad_norm": 0.43438801169395447,
      "learning_rate": 2.2814870073355715e-07,
      "loss": 0.6586,
      "step": 23029
    },
    {
      "epoch": 0.954453147664636,
      "grad_norm": 0.390813946723938,
      "learning_rate": 2.2794148120518877e-07,
      "loss": 0.6815,
      "step": 23030
    },
    {
      "epoch": 0.9544945915703096,
      "grad_norm": 0.4380500316619873,
      "learning_rate": 2.2773426167682042e-07,
      "loss": 0.6812,
      "step": 23031
    },
    {
      "epoch": 0.9545360354759832,
      "grad_norm": 0.4174291789531708,
      "learning_rate": 2.2752704214845207e-07,
      "loss": 0.6625,
      "step": 23032
    },
    {
      "epoch": 0.9545774793816569,
      "grad_norm": 0.4351099729537964,
      "learning_rate": 2.2731982262008372e-07,
      "loss": 0.6438,
      "step": 23033
    },
    {
      "epoch": 0.9546189232873306,
      "grad_norm": 0.4168747365474701,
      "learning_rate": 2.2711260309171537e-07,
      "loss": 0.6958,
      "step": 23034
    },
    {
      "epoch": 0.9546603671930043,
      "grad_norm": 0.425367146730423,
      "learning_rate": 2.2690538356334702e-07,
      "loss": 0.6633,
      "step": 23035
    },
    {
      "epoch": 0.9547018110986779,
      "grad_norm": 0.4160795211791992,
      "learning_rate": 2.2669816403497867e-07,
      "loss": 0.6683,
      "step": 23036
    },
    {
      "epoch": 0.9547432550043516,
      "grad_norm": 0.4083781838417053,
      "learning_rate": 2.2649094450661032e-07,
      "loss": 0.6843,
      "step": 23037
    },
    {
      "epoch": 0.9547846989100253,
      "grad_norm": 0.427225261926651,
      "learning_rate": 2.2628372497824197e-07,
      "loss": 0.7087,
      "step": 23038
    },
    {
      "epoch": 0.954826142815699,
      "grad_norm": 0.4449876546859741,
      "learning_rate": 2.2607650544987362e-07,
      "loss": 0.6279,
      "step": 23039
    },
    {
      "epoch": 0.9548675867213726,
      "grad_norm": 0.4162338674068451,
      "learning_rate": 2.2586928592150527e-07,
      "loss": 0.7003,
      "step": 23040
    },
    {
      "epoch": 0.9549090306270462,
      "grad_norm": 0.525565505027771,
      "learning_rate": 2.2566206639313692e-07,
      "loss": 0.5938,
      "step": 23041
    },
    {
      "epoch": 0.95495047453272,
      "grad_norm": 0.39031946659088135,
      "learning_rate": 2.2545484686476855e-07,
      "loss": 0.6511,
      "step": 23042
    },
    {
      "epoch": 0.9549919184383936,
      "grad_norm": 0.419796884059906,
      "learning_rate": 2.252476273364002e-07,
      "loss": 0.6792,
      "step": 23043
    },
    {
      "epoch": 0.9550333623440673,
      "grad_norm": 0.3977256417274475,
      "learning_rate": 2.2504040780803185e-07,
      "loss": 0.6805,
      "step": 23044
    },
    {
      "epoch": 0.955074806249741,
      "grad_norm": 0.4180990159511566,
      "learning_rate": 2.248331882796635e-07,
      "loss": 0.657,
      "step": 23045
    },
    {
      "epoch": 0.9551162501554147,
      "grad_norm": 0.4230247735977173,
      "learning_rate": 2.2462596875129515e-07,
      "loss": 0.6469,
      "step": 23046
    },
    {
      "epoch": 0.9551576940610883,
      "grad_norm": 0.420587956905365,
      "learning_rate": 2.244187492229268e-07,
      "loss": 0.6693,
      "step": 23047
    },
    {
      "epoch": 0.955199137966762,
      "grad_norm": 0.42101094126701355,
      "learning_rate": 2.2421152969455845e-07,
      "loss": 0.6979,
      "step": 23048
    },
    {
      "epoch": 0.9552405818724357,
      "grad_norm": 0.40215256810188293,
      "learning_rate": 2.2400431016619007e-07,
      "loss": 0.6577,
      "step": 23049
    },
    {
      "epoch": 0.9552820257781093,
      "grad_norm": 0.43815138936042786,
      "learning_rate": 2.2379709063782172e-07,
      "loss": 0.6948,
      "step": 23050
    },
    {
      "epoch": 0.955323469683783,
      "grad_norm": 0.42006292939186096,
      "learning_rate": 2.2358987110945337e-07,
      "loss": 0.6606,
      "step": 23051
    },
    {
      "epoch": 0.9553649135894566,
      "grad_norm": 0.42261528968811035,
      "learning_rate": 2.2338265158108502e-07,
      "loss": 0.6527,
      "step": 23052
    },
    {
      "epoch": 0.9554063574951304,
      "grad_norm": 0.44373568892478943,
      "learning_rate": 2.2317543205271667e-07,
      "loss": 0.698,
      "step": 23053
    },
    {
      "epoch": 0.955447801400804,
      "grad_norm": 0.4723448157310486,
      "learning_rate": 2.229682125243483e-07,
      "loss": 0.6993,
      "step": 23054
    },
    {
      "epoch": 0.9554892453064777,
      "grad_norm": 0.4002114534378052,
      "learning_rate": 2.2276099299597995e-07,
      "loss": 0.7041,
      "step": 23055
    },
    {
      "epoch": 0.9555306892121513,
      "grad_norm": 0.40913933515548706,
      "learning_rate": 2.225537734676116e-07,
      "loss": 0.6572,
      "step": 23056
    },
    {
      "epoch": 0.9555721331178251,
      "grad_norm": 0.41568508744239807,
      "learning_rate": 2.2234655393924325e-07,
      "loss": 0.6678,
      "step": 23057
    },
    {
      "epoch": 0.9556135770234987,
      "grad_norm": 0.44930580258369446,
      "learning_rate": 2.221393344108749e-07,
      "loss": 0.7056,
      "step": 23058
    },
    {
      "epoch": 0.9556550209291723,
      "grad_norm": 0.45008140802383423,
      "learning_rate": 2.2193211488250655e-07,
      "loss": 0.6672,
      "step": 23059
    },
    {
      "epoch": 0.955696464834846,
      "grad_norm": 0.414188027381897,
      "learning_rate": 2.217248953541382e-07,
      "loss": 0.6321,
      "step": 23060
    },
    {
      "epoch": 0.9557379087405197,
      "grad_norm": 0.4118972420692444,
      "learning_rate": 2.2151767582576985e-07,
      "loss": 0.7104,
      "step": 23061
    },
    {
      "epoch": 0.9557793526461934,
      "grad_norm": 0.44999557733535767,
      "learning_rate": 2.213104562974015e-07,
      "loss": 0.7661,
      "step": 23062
    },
    {
      "epoch": 0.955820796551867,
      "grad_norm": 0.4481130838394165,
      "learning_rate": 2.2110323676903315e-07,
      "loss": 0.6376,
      "step": 23063
    },
    {
      "epoch": 0.9558622404575408,
      "grad_norm": 0.3855637013912201,
      "learning_rate": 2.208960172406648e-07,
      "loss": 0.6522,
      "step": 23064
    },
    {
      "epoch": 0.9559036843632144,
      "grad_norm": 0.3998076021671295,
      "learning_rate": 2.2068879771229645e-07,
      "loss": 0.639,
      "step": 23065
    },
    {
      "epoch": 0.9559451282688881,
      "grad_norm": 0.4299071133136749,
      "learning_rate": 2.2048157818392805e-07,
      "loss": 0.6542,
      "step": 23066
    },
    {
      "epoch": 0.9559865721745617,
      "grad_norm": 0.41824570298194885,
      "learning_rate": 2.202743586555597e-07,
      "loss": 0.6442,
      "step": 23067
    },
    {
      "epoch": 0.9560280160802354,
      "grad_norm": 0.40059101581573486,
      "learning_rate": 2.2006713912719135e-07,
      "loss": 0.6426,
      "step": 23068
    },
    {
      "epoch": 0.9560694599859091,
      "grad_norm": 0.3977491855621338,
      "learning_rate": 2.19859919598823e-07,
      "loss": 0.6306,
      "step": 23069
    },
    {
      "epoch": 0.9561109038915827,
      "grad_norm": 0.40396496653556824,
      "learning_rate": 2.1965270007045465e-07,
      "loss": 0.5979,
      "step": 23070
    },
    {
      "epoch": 0.9561523477972564,
      "grad_norm": 0.4345090985298157,
      "learning_rate": 2.194454805420863e-07,
      "loss": 0.6562,
      "step": 23071
    },
    {
      "epoch": 0.9561937917029301,
      "grad_norm": 0.3842495381832123,
      "learning_rate": 2.1923826101371795e-07,
      "loss": 0.6096,
      "step": 23072
    },
    {
      "epoch": 0.9562352356086038,
      "grad_norm": 0.47109782695770264,
      "learning_rate": 2.190310414853496e-07,
      "loss": 0.6877,
      "step": 23073
    },
    {
      "epoch": 0.9562766795142774,
      "grad_norm": 0.4509117901325226,
      "learning_rate": 2.1882382195698125e-07,
      "loss": 0.6711,
      "step": 23074
    },
    {
      "epoch": 0.9563181234199511,
      "grad_norm": 0.4120785593986511,
      "learning_rate": 2.186166024286129e-07,
      "loss": 0.6938,
      "step": 23075
    },
    {
      "epoch": 0.9563595673256248,
      "grad_norm": 0.3917151391506195,
      "learning_rate": 2.1840938290024455e-07,
      "loss": 0.6084,
      "step": 23076
    },
    {
      "epoch": 0.9564010112312984,
      "grad_norm": 0.40734294056892395,
      "learning_rate": 2.182021633718762e-07,
      "loss": 0.6909,
      "step": 23077
    },
    {
      "epoch": 0.9564424551369721,
      "grad_norm": 0.3935093879699707,
      "learning_rate": 2.179949438435078e-07,
      "loss": 0.619,
      "step": 23078
    },
    {
      "epoch": 0.9564838990426457,
      "grad_norm": 0.44054827094078064,
      "learning_rate": 2.1778772431513945e-07,
      "loss": 0.6005,
      "step": 23079
    },
    {
      "epoch": 0.9565253429483195,
      "grad_norm": 0.4292285442352295,
      "learning_rate": 2.175805047867711e-07,
      "loss": 0.6846,
      "step": 23080
    },
    {
      "epoch": 0.9565667868539931,
      "grad_norm": 0.40296241641044617,
      "learning_rate": 2.1737328525840275e-07,
      "loss": 0.662,
      "step": 23081
    },
    {
      "epoch": 0.9566082307596668,
      "grad_norm": 0.47924676537513733,
      "learning_rate": 2.171660657300344e-07,
      "loss": 0.6146,
      "step": 23082
    },
    {
      "epoch": 0.9566496746653405,
      "grad_norm": 0.3806125819683075,
      "learning_rate": 2.1695884620166605e-07,
      "loss": 0.6229,
      "step": 23083
    },
    {
      "epoch": 0.9566911185710142,
      "grad_norm": 0.3953922688961029,
      "learning_rate": 2.167516266732977e-07,
      "loss": 0.6655,
      "step": 23084
    },
    {
      "epoch": 0.9567325624766878,
      "grad_norm": 0.37801459431648254,
      "learning_rate": 2.1654440714492935e-07,
      "loss": 0.6189,
      "step": 23085
    },
    {
      "epoch": 0.9567740063823614,
      "grad_norm": 0.45688873529434204,
      "learning_rate": 2.16337187616561e-07,
      "loss": 0.6298,
      "step": 23086
    },
    {
      "epoch": 0.9568154502880352,
      "grad_norm": 0.42127177119255066,
      "learning_rate": 2.1612996808819265e-07,
      "loss": 0.6602,
      "step": 23087
    },
    {
      "epoch": 0.9568568941937088,
      "grad_norm": 0.4120872914791107,
      "learning_rate": 2.159227485598243e-07,
      "loss": 0.6592,
      "step": 23088
    },
    {
      "epoch": 0.9568983380993825,
      "grad_norm": 0.4545368552207947,
      "learning_rate": 2.1571552903145595e-07,
      "loss": 0.6045,
      "step": 23089
    },
    {
      "epoch": 0.9569397820050561,
      "grad_norm": 0.4282217025756836,
      "learning_rate": 2.155083095030876e-07,
      "loss": 0.6501,
      "step": 23090
    },
    {
      "epoch": 0.9569812259107299,
      "grad_norm": 0.45731642842292786,
      "learning_rate": 2.1530108997471923e-07,
      "loss": 0.6592,
      "step": 23091
    },
    {
      "epoch": 0.9570226698164035,
      "grad_norm": 0.38938209414482117,
      "learning_rate": 2.1509387044635088e-07,
      "loss": 0.6467,
      "step": 23092
    },
    {
      "epoch": 0.9570641137220771,
      "grad_norm": 0.39338698983192444,
      "learning_rate": 2.1488665091798253e-07,
      "loss": 0.6711,
      "step": 23093
    },
    {
      "epoch": 0.9571055576277508,
      "grad_norm": 0.40484172105789185,
      "learning_rate": 2.1467943138961418e-07,
      "loss": 0.6904,
      "step": 23094
    },
    {
      "epoch": 0.9571470015334245,
      "grad_norm": 0.39045190811157227,
      "learning_rate": 2.1447221186124583e-07,
      "loss": 0.645,
      "step": 23095
    },
    {
      "epoch": 0.9571884454390982,
      "grad_norm": 0.387788861989975,
      "learning_rate": 2.1426499233287745e-07,
      "loss": 0.6158,
      "step": 23096
    },
    {
      "epoch": 0.9572298893447718,
      "grad_norm": 0.41141822934150696,
      "learning_rate": 2.140577728045091e-07,
      "loss": 0.6646,
      "step": 23097
    },
    {
      "epoch": 0.9572713332504456,
      "grad_norm": 0.42459625005722046,
      "learning_rate": 2.1385055327614075e-07,
      "loss": 0.6808,
      "step": 23098
    },
    {
      "epoch": 0.9573127771561192,
      "grad_norm": 0.41666072607040405,
      "learning_rate": 2.136433337477724e-07,
      "loss": 0.6876,
      "step": 23099
    },
    {
      "epoch": 0.9573542210617929,
      "grad_norm": 0.38848885893821716,
      "learning_rate": 2.1343611421940406e-07,
      "loss": 0.6378,
      "step": 23100
    },
    {
      "epoch": 0.9573956649674665,
      "grad_norm": 0.4206179976463318,
      "learning_rate": 2.132288946910357e-07,
      "loss": 0.7119,
      "step": 23101
    },
    {
      "epoch": 0.9574371088731402,
      "grad_norm": 0.41347256302833557,
      "learning_rate": 2.1302167516266736e-07,
      "loss": 0.6685,
      "step": 23102
    },
    {
      "epoch": 0.9574785527788139,
      "grad_norm": 0.44028353691101074,
      "learning_rate": 2.1281445563429898e-07,
      "loss": 0.7117,
      "step": 23103
    },
    {
      "epoch": 0.9575199966844875,
      "grad_norm": 0.42882731556892395,
      "learning_rate": 2.1260723610593063e-07,
      "loss": 0.6819,
      "step": 23104
    },
    {
      "epoch": 0.9575614405901612,
      "grad_norm": 0.4082034230232239,
      "learning_rate": 2.1240001657756228e-07,
      "loss": 0.6541,
      "step": 23105
    },
    {
      "epoch": 0.9576028844958349,
      "grad_norm": 0.40479129552841187,
      "learning_rate": 2.1219279704919393e-07,
      "loss": 0.655,
      "step": 23106
    },
    {
      "epoch": 0.9576443284015086,
      "grad_norm": 0.39620548486709595,
      "learning_rate": 2.1198557752082558e-07,
      "loss": 0.6228,
      "step": 23107
    },
    {
      "epoch": 0.9576857723071822,
      "grad_norm": 0.39385801553726196,
      "learning_rate": 2.1177835799245723e-07,
      "loss": 0.6733,
      "step": 23108
    },
    {
      "epoch": 0.957727216212856,
      "grad_norm": 0.39574769139289856,
      "learning_rate": 2.1157113846408888e-07,
      "loss": 0.6376,
      "step": 23109
    },
    {
      "epoch": 0.9577686601185296,
      "grad_norm": 0.396316260099411,
      "learning_rate": 2.1136391893572053e-07,
      "loss": 0.6843,
      "step": 23110
    },
    {
      "epoch": 0.9578101040242032,
      "grad_norm": 0.3944850265979767,
      "learning_rate": 2.1115669940735218e-07,
      "loss": 0.6301,
      "step": 23111
    },
    {
      "epoch": 0.9578515479298769,
      "grad_norm": 0.42518535256385803,
      "learning_rate": 2.1094947987898383e-07,
      "loss": 0.6514,
      "step": 23112
    },
    {
      "epoch": 0.9578929918355505,
      "grad_norm": 0.4587007462978363,
      "learning_rate": 2.1074226035061548e-07,
      "loss": 0.6907,
      "step": 23113
    },
    {
      "epoch": 0.9579344357412243,
      "grad_norm": 0.4469851851463318,
      "learning_rate": 2.1053504082224713e-07,
      "loss": 0.6753,
      "step": 23114
    },
    {
      "epoch": 0.9579758796468979,
      "grad_norm": 0.42378219962120056,
      "learning_rate": 2.1032782129387873e-07,
      "loss": 0.6765,
      "step": 23115
    },
    {
      "epoch": 0.9580173235525716,
      "grad_norm": 0.3928704857826233,
      "learning_rate": 2.1012060176551038e-07,
      "loss": 0.616,
      "step": 23116
    },
    {
      "epoch": 0.9580587674582453,
      "grad_norm": 0.43857342004776,
      "learning_rate": 2.0991338223714203e-07,
      "loss": 0.6711,
      "step": 23117
    },
    {
      "epoch": 0.958100211363919,
      "grad_norm": 0.4277779757976532,
      "learning_rate": 2.0970616270877368e-07,
      "loss": 0.6951,
      "step": 23118
    },
    {
      "epoch": 0.9581416552695926,
      "grad_norm": 0.4310479164123535,
      "learning_rate": 2.0949894318040533e-07,
      "loss": 0.6996,
      "step": 23119
    },
    {
      "epoch": 0.9581830991752662,
      "grad_norm": 0.3995084762573242,
      "learning_rate": 2.0929172365203698e-07,
      "loss": 0.6211,
      "step": 23120
    },
    {
      "epoch": 0.95822454308094,
      "grad_norm": 0.40500330924987793,
      "learning_rate": 2.0908450412366863e-07,
      "loss": 0.6395,
      "step": 23121
    },
    {
      "epoch": 0.9582659869866136,
      "grad_norm": 0.3900543451309204,
      "learning_rate": 2.0887728459530028e-07,
      "loss": 0.6575,
      "step": 23122
    },
    {
      "epoch": 0.9583074308922873,
      "grad_norm": 0.4262588918209076,
      "learning_rate": 2.0867006506693193e-07,
      "loss": 0.7283,
      "step": 23123
    },
    {
      "epoch": 0.9583488747979609,
      "grad_norm": 0.42917948961257935,
      "learning_rate": 2.0846284553856358e-07,
      "loss": 0.7189,
      "step": 23124
    },
    {
      "epoch": 0.9583903187036347,
      "grad_norm": 0.3942415714263916,
      "learning_rate": 2.0825562601019523e-07,
      "loss": 0.6084,
      "step": 23125
    },
    {
      "epoch": 0.9584317626093083,
      "grad_norm": 0.4708394408226013,
      "learning_rate": 2.0804840648182689e-07,
      "loss": 0.6707,
      "step": 23126
    },
    {
      "epoch": 0.958473206514982,
      "grad_norm": 0.4115959107875824,
      "learning_rate": 2.0784118695345854e-07,
      "loss": 0.7085,
      "step": 23127
    },
    {
      "epoch": 0.9585146504206556,
      "grad_norm": 0.4072836637496948,
      "learning_rate": 2.0763396742509013e-07,
      "loss": 0.6497,
      "step": 23128
    },
    {
      "epoch": 0.9585560943263293,
      "grad_norm": 0.37720367312431335,
      "learning_rate": 2.0742674789672178e-07,
      "loss": 0.6409,
      "step": 23129
    },
    {
      "epoch": 0.958597538232003,
      "grad_norm": 0.4143693149089813,
      "learning_rate": 2.0721952836835343e-07,
      "loss": 0.7084,
      "step": 23130
    },
    {
      "epoch": 0.9586389821376766,
      "grad_norm": 0.4518071711063385,
      "learning_rate": 2.0701230883998508e-07,
      "loss": 0.6763,
      "step": 23131
    },
    {
      "epoch": 0.9586804260433504,
      "grad_norm": 0.4292539954185486,
      "learning_rate": 2.0680508931161673e-07,
      "loss": 0.6653,
      "step": 23132
    },
    {
      "epoch": 0.958721869949024,
      "grad_norm": 0.40304943919181824,
      "learning_rate": 2.0659786978324838e-07,
      "loss": 0.7002,
      "step": 23133
    },
    {
      "epoch": 0.9587633138546977,
      "grad_norm": 0.4577230215072632,
      "learning_rate": 2.0639065025488003e-07,
      "loss": 0.7302,
      "step": 23134
    },
    {
      "epoch": 0.9588047577603713,
      "grad_norm": 0.39477604627609253,
      "learning_rate": 2.0618343072651169e-07,
      "loss": 0.6445,
      "step": 23135
    },
    {
      "epoch": 0.9588462016660451,
      "grad_norm": 0.40126991271972656,
      "learning_rate": 2.0597621119814334e-07,
      "loss": 0.6195,
      "step": 23136
    },
    {
      "epoch": 0.9588876455717187,
      "grad_norm": 0.4666047692298889,
      "learning_rate": 2.0576899166977499e-07,
      "loss": 0.7156,
      "step": 23137
    },
    {
      "epoch": 0.9589290894773923,
      "grad_norm": 0.42344728112220764,
      "learning_rate": 2.0556177214140664e-07,
      "loss": 0.6887,
      "step": 23138
    },
    {
      "epoch": 0.958970533383066,
      "grad_norm": 0.3777908980846405,
      "learning_rate": 2.0535455261303829e-07,
      "loss": 0.6597,
      "step": 23139
    },
    {
      "epoch": 0.9590119772887397,
      "grad_norm": 0.4349067509174347,
      "learning_rate": 2.051473330846699e-07,
      "loss": 0.6501,
      "step": 23140
    },
    {
      "epoch": 0.9590534211944134,
      "grad_norm": 0.44103920459747314,
      "learning_rate": 2.0494011355630156e-07,
      "loss": 0.6707,
      "step": 23141
    },
    {
      "epoch": 0.959094865100087,
      "grad_norm": 0.3859243094921112,
      "learning_rate": 2.047328940279332e-07,
      "loss": 0.6287,
      "step": 23142
    },
    {
      "epoch": 0.9591363090057607,
      "grad_norm": 0.4258202314376831,
      "learning_rate": 2.0452567449956484e-07,
      "loss": 0.6042,
      "step": 23143
    },
    {
      "epoch": 0.9591777529114344,
      "grad_norm": 0.4573183059692383,
      "learning_rate": 2.0431845497119649e-07,
      "loss": 0.6853,
      "step": 23144
    },
    {
      "epoch": 0.9592191968171081,
      "grad_norm": 0.42446082830429077,
      "learning_rate": 2.0411123544282814e-07,
      "loss": 0.6934,
      "step": 23145
    },
    {
      "epoch": 0.9592606407227817,
      "grad_norm": 0.38893240690231323,
      "learning_rate": 2.0390401591445979e-07,
      "loss": 0.651,
      "step": 23146
    },
    {
      "epoch": 0.9593020846284553,
      "grad_norm": 0.41613274812698364,
      "learning_rate": 2.0369679638609144e-07,
      "loss": 0.6831,
      "step": 23147
    },
    {
      "epoch": 0.9593435285341291,
      "grad_norm": 0.40390047430992126,
      "learning_rate": 2.034895768577231e-07,
      "loss": 0.5852,
      "step": 23148
    },
    {
      "epoch": 0.9593849724398027,
      "grad_norm": 0.40002626180648804,
      "learning_rate": 2.0328235732935474e-07,
      "loss": 0.6213,
      "step": 23149
    },
    {
      "epoch": 0.9594264163454764,
      "grad_norm": 0.41857755184173584,
      "learning_rate": 2.030751378009864e-07,
      "loss": 0.6534,
      "step": 23150
    },
    {
      "epoch": 0.95946786025115,
      "grad_norm": 0.3895922601222992,
      "learning_rate": 2.0286791827261804e-07,
      "loss": 0.691,
      "step": 23151
    },
    {
      "epoch": 0.9595093041568238,
      "grad_norm": 0.4271538257598877,
      "learning_rate": 2.0266069874424966e-07,
      "loss": 0.6812,
      "step": 23152
    },
    {
      "epoch": 0.9595507480624974,
      "grad_norm": 0.4382900297641754,
      "learning_rate": 2.024534792158813e-07,
      "loss": 0.6721,
      "step": 23153
    },
    {
      "epoch": 0.959592191968171,
      "grad_norm": 0.4384820759296417,
      "learning_rate": 2.0224625968751296e-07,
      "loss": 0.6833,
      "step": 23154
    },
    {
      "epoch": 0.9596336358738448,
      "grad_norm": 0.39148852229118347,
      "learning_rate": 2.020390401591446e-07,
      "loss": 0.6587,
      "step": 23155
    },
    {
      "epoch": 0.9596750797795184,
      "grad_norm": 0.43319573998451233,
      "learning_rate": 2.0183182063077626e-07,
      "loss": 0.7266,
      "step": 23156
    },
    {
      "epoch": 0.9597165236851921,
      "grad_norm": 0.43426328897476196,
      "learning_rate": 2.0162460110240791e-07,
      "loss": 0.7261,
      "step": 23157
    },
    {
      "epoch": 0.9597579675908657,
      "grad_norm": 0.4503077268600464,
      "learning_rate": 2.0141738157403956e-07,
      "loss": 0.7109,
      "step": 23158
    },
    {
      "epoch": 0.9597994114965395,
      "grad_norm": 0.43513306975364685,
      "learning_rate": 2.0121016204567121e-07,
      "loss": 0.6322,
      "step": 23159
    },
    {
      "epoch": 0.9598408554022131,
      "grad_norm": 0.42748013138771057,
      "learning_rate": 2.0100294251730286e-07,
      "loss": 0.6545,
      "step": 23160
    },
    {
      "epoch": 0.9598822993078868,
      "grad_norm": 0.4084666073322296,
      "learning_rate": 2.0079572298893452e-07,
      "loss": 0.7018,
      "step": 23161
    },
    {
      "epoch": 0.9599237432135604,
      "grad_norm": 0.40143901109695435,
      "learning_rate": 2.0058850346056614e-07,
      "loss": 0.6278,
      "step": 23162
    },
    {
      "epoch": 0.9599651871192341,
      "grad_norm": 0.4238412082195282,
      "learning_rate": 2.003812839321978e-07,
      "loss": 0.6865,
      "step": 23163
    },
    {
      "epoch": 0.9600066310249078,
      "grad_norm": 0.42535850405693054,
      "learning_rate": 2.0017406440382944e-07,
      "loss": 0.6759,
      "step": 23164
    },
    {
      "epoch": 0.9600480749305814,
      "grad_norm": 0.4023435413837433,
      "learning_rate": 1.9996684487546106e-07,
      "loss": 0.6285,
      "step": 23165
    },
    {
      "epoch": 0.9600895188362552,
      "grad_norm": 0.4379653036594391,
      "learning_rate": 1.9975962534709271e-07,
      "loss": 0.678,
      "step": 23166
    },
    {
      "epoch": 0.9601309627419288,
      "grad_norm": 0.3986414670944214,
      "learning_rate": 1.9955240581872436e-07,
      "loss": 0.6555,
      "step": 23167
    },
    {
      "epoch": 0.9601724066476025,
      "grad_norm": 0.42397162318229675,
      "learning_rate": 1.9934518629035601e-07,
      "loss": 0.6631,
      "step": 23168
    },
    {
      "epoch": 0.9602138505532761,
      "grad_norm": 0.43795710802078247,
      "learning_rate": 1.9913796676198766e-07,
      "loss": 0.7224,
      "step": 23169
    },
    {
      "epoch": 0.9602552944589499,
      "grad_norm": 0.37666749954223633,
      "learning_rate": 1.9893074723361932e-07,
      "loss": 0.6166,
      "step": 23170
    },
    {
      "epoch": 0.9602967383646235,
      "grad_norm": 0.4144953191280365,
      "learning_rate": 1.9872352770525097e-07,
      "loss": 0.6517,
      "step": 23171
    },
    {
      "epoch": 0.9603381822702971,
      "grad_norm": 0.43220314383506775,
      "learning_rate": 1.9851630817688262e-07,
      "loss": 0.6978,
      "step": 23172
    },
    {
      "epoch": 0.9603796261759708,
      "grad_norm": 0.42193520069122314,
      "learning_rate": 1.9830908864851427e-07,
      "loss": 0.6458,
      "step": 23173
    },
    {
      "epoch": 0.9604210700816445,
      "grad_norm": 0.4525734484195709,
      "learning_rate": 1.9810186912014592e-07,
      "loss": 0.6388,
      "step": 23174
    },
    {
      "epoch": 0.9604625139873182,
      "grad_norm": 0.38898661732673645,
      "learning_rate": 1.9789464959177757e-07,
      "loss": 0.6469,
      "step": 23175
    },
    {
      "epoch": 0.9605039578929918,
      "grad_norm": 0.4502175748348236,
      "learning_rate": 1.9768743006340922e-07,
      "loss": 0.6682,
      "step": 23176
    },
    {
      "epoch": 0.9605454017986655,
      "grad_norm": 0.418739914894104,
      "learning_rate": 1.9748021053504081e-07,
      "loss": 0.7018,
      "step": 23177
    },
    {
      "epoch": 0.9605868457043392,
      "grad_norm": 0.4222378730773926,
      "learning_rate": 1.9727299100667247e-07,
      "loss": 0.6573,
      "step": 23178
    },
    {
      "epoch": 0.9606282896100129,
      "grad_norm": 0.4074171483516693,
      "learning_rate": 1.9706577147830412e-07,
      "loss": 0.6429,
      "step": 23179
    },
    {
      "epoch": 0.9606697335156865,
      "grad_norm": 0.44411569833755493,
      "learning_rate": 1.9685855194993577e-07,
      "loss": 0.649,
      "step": 23180
    },
    {
      "epoch": 0.9607111774213601,
      "grad_norm": 0.4248681664466858,
      "learning_rate": 1.9665133242156742e-07,
      "loss": 0.6376,
      "step": 23181
    },
    {
      "epoch": 0.9607526213270339,
      "grad_norm": 0.41487404704093933,
      "learning_rate": 1.9644411289319907e-07,
      "loss": 0.6588,
      "step": 23182
    },
    {
      "epoch": 0.9607940652327075,
      "grad_norm": 0.40593189001083374,
      "learning_rate": 1.9623689336483072e-07,
      "loss": 0.6688,
      "step": 23183
    },
    {
      "epoch": 0.9608355091383812,
      "grad_norm": 0.4116593301296234,
      "learning_rate": 1.9602967383646237e-07,
      "loss": 0.6313,
      "step": 23184
    },
    {
      "epoch": 0.9608769530440548,
      "grad_norm": 0.42997676134109497,
      "learning_rate": 1.9582245430809402e-07,
      "loss": 0.677,
      "step": 23185
    },
    {
      "epoch": 0.9609183969497286,
      "grad_norm": 0.4413765072822571,
      "learning_rate": 1.9561523477972567e-07,
      "loss": 0.7112,
      "step": 23186
    },
    {
      "epoch": 0.9609598408554022,
      "grad_norm": 0.41863149404525757,
      "learning_rate": 1.9540801525135732e-07,
      "loss": 0.6829,
      "step": 23187
    },
    {
      "epoch": 0.9610012847610759,
      "grad_norm": 0.42147454619407654,
      "learning_rate": 1.9520079572298897e-07,
      "loss": 0.6508,
      "step": 23188
    },
    {
      "epoch": 0.9610427286667496,
      "grad_norm": 0.38393649458885193,
      "learning_rate": 1.949935761946206e-07,
      "loss": 0.615,
      "step": 23189
    },
    {
      "epoch": 0.9610841725724232,
      "grad_norm": 0.40934789180755615,
      "learning_rate": 1.9478635666625224e-07,
      "loss": 0.671,
      "step": 23190
    },
    {
      "epoch": 0.9611256164780969,
      "grad_norm": 0.45235881209373474,
      "learning_rate": 1.9457913713788387e-07,
      "loss": 0.6771,
      "step": 23191
    },
    {
      "epoch": 0.9611670603837705,
      "grad_norm": 0.3910123109817505,
      "learning_rate": 1.9437191760951552e-07,
      "loss": 0.6278,
      "step": 23192
    },
    {
      "epoch": 0.9612085042894443,
      "grad_norm": 0.39734891057014465,
      "learning_rate": 1.9416469808114717e-07,
      "loss": 0.636,
      "step": 23193
    },
    {
      "epoch": 0.9612499481951179,
      "grad_norm": 0.4058263599872589,
      "learning_rate": 1.9395747855277882e-07,
      "loss": 0.6783,
      "step": 23194
    },
    {
      "epoch": 0.9612913921007916,
      "grad_norm": 0.412992388010025,
      "learning_rate": 1.9375025902441047e-07,
      "loss": 0.684,
      "step": 23195
    },
    {
      "epoch": 0.9613328360064652,
      "grad_norm": 0.422380268573761,
      "learning_rate": 1.9354303949604212e-07,
      "loss": 0.6432,
      "step": 23196
    },
    {
      "epoch": 0.961374279912139,
      "grad_norm": 0.4061347246170044,
      "learning_rate": 1.9333581996767377e-07,
      "loss": 0.6598,
      "step": 23197
    },
    {
      "epoch": 0.9614157238178126,
      "grad_norm": 0.430497944355011,
      "learning_rate": 1.9312860043930542e-07,
      "loss": 0.6721,
      "step": 23198
    },
    {
      "epoch": 0.9614571677234862,
      "grad_norm": 0.4027978777885437,
      "learning_rate": 1.9292138091093707e-07,
      "loss": 0.6396,
      "step": 23199
    },
    {
      "epoch": 0.96149861162916,
      "grad_norm": 0.41532155871391296,
      "learning_rate": 1.9271416138256872e-07,
      "loss": 0.6735,
      "step": 23200
    },
    {
      "epoch": 0.9615400555348336,
      "grad_norm": 0.4190988838672638,
      "learning_rate": 1.9250694185420037e-07,
      "loss": 0.6173,
      "step": 23201
    },
    {
      "epoch": 0.9615814994405073,
      "grad_norm": 0.4334971606731415,
      "learning_rate": 1.92299722325832e-07,
      "loss": 0.6802,
      "step": 23202
    },
    {
      "epoch": 0.9616229433461809,
      "grad_norm": 0.43776872754096985,
      "learning_rate": 1.9209250279746364e-07,
      "loss": 0.6747,
      "step": 23203
    },
    {
      "epoch": 0.9616643872518547,
      "grad_norm": 0.38229429721832275,
      "learning_rate": 1.918852832690953e-07,
      "loss": 0.6342,
      "step": 23204
    },
    {
      "epoch": 0.9617058311575283,
      "grad_norm": 0.378037691116333,
      "learning_rate": 1.9167806374072695e-07,
      "loss": 0.6338,
      "step": 23205
    },
    {
      "epoch": 0.9617472750632019,
      "grad_norm": 0.4342402517795563,
      "learning_rate": 1.914708442123586e-07,
      "loss": 0.6759,
      "step": 23206
    },
    {
      "epoch": 0.9617887189688756,
      "grad_norm": 0.3797514736652374,
      "learning_rate": 1.9126362468399025e-07,
      "loss": 0.6145,
      "step": 23207
    },
    {
      "epoch": 0.9618301628745493,
      "grad_norm": 0.44878053665161133,
      "learning_rate": 1.910564051556219e-07,
      "loss": 0.6819,
      "step": 23208
    },
    {
      "epoch": 0.961871606780223,
      "grad_norm": 0.395308256149292,
      "learning_rate": 1.9084918562725352e-07,
      "loss": 0.652,
      "step": 23209
    },
    {
      "epoch": 0.9619130506858966,
      "grad_norm": 0.3902066946029663,
      "learning_rate": 1.9064196609888517e-07,
      "loss": 0.6118,
      "step": 23210
    },
    {
      "epoch": 0.9619544945915703,
      "grad_norm": 0.4185203015804291,
      "learning_rate": 1.9043474657051682e-07,
      "loss": 0.6404,
      "step": 23211
    },
    {
      "epoch": 0.961995938497244,
      "grad_norm": 0.38643380999565125,
      "learning_rate": 1.9022752704214847e-07,
      "loss": 0.6516,
      "step": 23212
    },
    {
      "epoch": 0.9620373824029177,
      "grad_norm": 0.3884592056274414,
      "learning_rate": 1.9002030751378012e-07,
      "loss": 0.6237,
      "step": 23213
    },
    {
      "epoch": 0.9620788263085913,
      "grad_norm": 0.4123048782348633,
      "learning_rate": 1.8981308798541175e-07,
      "loss": 0.7129,
      "step": 23214
    },
    {
      "epoch": 0.9621202702142649,
      "grad_norm": 0.4210478365421295,
      "learning_rate": 1.896058684570434e-07,
      "loss": 0.6989,
      "step": 23215
    },
    {
      "epoch": 0.9621617141199387,
      "grad_norm": 0.45603370666503906,
      "learning_rate": 1.8939864892867505e-07,
      "loss": 0.6981,
      "step": 23216
    },
    {
      "epoch": 0.9622031580256123,
      "grad_norm": 0.41936248540878296,
      "learning_rate": 1.891914294003067e-07,
      "loss": 0.6575,
      "step": 23217
    },
    {
      "epoch": 0.962244601931286,
      "grad_norm": 0.4217749536037445,
      "learning_rate": 1.8898420987193835e-07,
      "loss": 0.6357,
      "step": 23218
    },
    {
      "epoch": 0.9622860458369596,
      "grad_norm": 0.39954033493995667,
      "learning_rate": 1.8877699034357e-07,
      "loss": 0.7048,
      "step": 23219
    },
    {
      "epoch": 0.9623274897426334,
      "grad_norm": 0.42419958114624023,
      "learning_rate": 1.8856977081520165e-07,
      "loss": 0.6323,
      "step": 23220
    },
    {
      "epoch": 0.962368933648307,
      "grad_norm": 0.4251878261566162,
      "learning_rate": 1.883625512868333e-07,
      "loss": 0.673,
      "step": 23221
    },
    {
      "epoch": 0.9624103775539807,
      "grad_norm": 0.4311068058013916,
      "learning_rate": 1.8815533175846495e-07,
      "loss": 0.6742,
      "step": 23222
    },
    {
      "epoch": 0.9624518214596544,
      "grad_norm": 0.42706409096717834,
      "learning_rate": 1.879481122300966e-07,
      "loss": 0.6907,
      "step": 23223
    },
    {
      "epoch": 0.962493265365328,
      "grad_norm": 0.430532842874527,
      "learning_rate": 1.8774089270172825e-07,
      "loss": 0.6721,
      "step": 23224
    },
    {
      "epoch": 0.9625347092710017,
      "grad_norm": 0.41283872723579407,
      "learning_rate": 1.875336731733599e-07,
      "loss": 0.6777,
      "step": 23225
    },
    {
      "epoch": 0.9625761531766753,
      "grad_norm": 0.4132024049758911,
      "learning_rate": 1.873264536449915e-07,
      "loss": 0.6156,
      "step": 23226
    },
    {
      "epoch": 0.9626175970823491,
      "grad_norm": 0.40249085426330566,
      "learning_rate": 1.8711923411662315e-07,
      "loss": 0.6277,
      "step": 23227
    },
    {
      "epoch": 0.9626590409880227,
      "grad_norm": 0.4029373228549957,
      "learning_rate": 1.869120145882548e-07,
      "loss": 0.6053,
      "step": 23228
    },
    {
      "epoch": 0.9627004848936964,
      "grad_norm": 0.38418957591056824,
      "learning_rate": 1.8670479505988645e-07,
      "loss": 0.6237,
      "step": 23229
    },
    {
      "epoch": 0.96274192879937,
      "grad_norm": 0.4161483347415924,
      "learning_rate": 1.864975755315181e-07,
      "loss": 0.6384,
      "step": 23230
    },
    {
      "epoch": 0.9627833727050438,
      "grad_norm": 0.4120124876499176,
      "learning_rate": 1.8629035600314975e-07,
      "loss": 0.6611,
      "step": 23231
    },
    {
      "epoch": 0.9628248166107174,
      "grad_norm": 0.41910380125045776,
      "learning_rate": 1.860831364747814e-07,
      "loss": 0.6692,
      "step": 23232
    },
    {
      "epoch": 0.962866260516391,
      "grad_norm": 0.40932485461235046,
      "learning_rate": 1.8587591694641305e-07,
      "loss": 0.6499,
      "step": 23233
    },
    {
      "epoch": 0.9629077044220647,
      "grad_norm": 0.4058624505996704,
      "learning_rate": 1.856686974180447e-07,
      "loss": 0.6067,
      "step": 23234
    },
    {
      "epoch": 0.9629491483277384,
      "grad_norm": 0.406186580657959,
      "learning_rate": 1.8546147788967635e-07,
      "loss": 0.7041,
      "step": 23235
    },
    {
      "epoch": 0.9629905922334121,
      "grad_norm": 0.41728949546813965,
      "learning_rate": 1.85254258361308e-07,
      "loss": 0.63,
      "step": 23236
    },
    {
      "epoch": 0.9630320361390857,
      "grad_norm": 0.41570979356765747,
      "learning_rate": 1.8504703883293965e-07,
      "loss": 0.6804,
      "step": 23237
    },
    {
      "epoch": 0.9630734800447595,
      "grad_norm": 0.4367997944355011,
      "learning_rate": 1.8483981930457125e-07,
      "loss": 0.675,
      "step": 23238
    },
    {
      "epoch": 0.9631149239504331,
      "grad_norm": 0.41493576765060425,
      "learning_rate": 1.846325997762029e-07,
      "loss": 0.6614,
      "step": 23239
    },
    {
      "epoch": 0.9631563678561068,
      "grad_norm": 0.42952725291252136,
      "learning_rate": 1.8442538024783455e-07,
      "loss": 0.6482,
      "step": 23240
    },
    {
      "epoch": 0.9631978117617804,
      "grad_norm": 0.4610096216201782,
      "learning_rate": 1.842181607194662e-07,
      "loss": 0.7068,
      "step": 23241
    },
    {
      "epoch": 0.963239255667454,
      "grad_norm": 0.4690104126930237,
      "learning_rate": 1.8401094119109785e-07,
      "loss": 0.7045,
      "step": 23242
    },
    {
      "epoch": 0.9632806995731278,
      "grad_norm": 0.43758708238601685,
      "learning_rate": 1.838037216627295e-07,
      "loss": 0.7327,
      "step": 23243
    },
    {
      "epoch": 0.9633221434788014,
      "grad_norm": 0.42984187602996826,
      "learning_rate": 1.8359650213436115e-07,
      "loss": 0.7129,
      "step": 23244
    },
    {
      "epoch": 0.9633635873844751,
      "grad_norm": 0.41961970925331116,
      "learning_rate": 1.833892826059928e-07,
      "loss": 0.6415,
      "step": 23245
    },
    {
      "epoch": 0.9634050312901488,
      "grad_norm": 0.4083646833896637,
      "learning_rate": 1.8318206307762445e-07,
      "loss": 0.6659,
      "step": 23246
    },
    {
      "epoch": 0.9634464751958225,
      "grad_norm": 0.4411815106868744,
      "learning_rate": 1.829748435492561e-07,
      "loss": 0.6304,
      "step": 23247
    },
    {
      "epoch": 0.9634879191014961,
      "grad_norm": 0.41600650548934937,
      "learning_rate": 1.8276762402088775e-07,
      "loss": 0.6564,
      "step": 23248
    },
    {
      "epoch": 0.9635293630071698,
      "grad_norm": 0.44136688113212585,
      "learning_rate": 1.825604044925194e-07,
      "loss": 0.6925,
      "step": 23249
    },
    {
      "epoch": 0.9635708069128435,
      "grad_norm": 0.43774133920669556,
      "learning_rate": 1.8235318496415105e-07,
      "loss": 0.7256,
      "step": 23250
    },
    {
      "epoch": 0.9636122508185171,
      "grad_norm": 0.41175365447998047,
      "learning_rate": 1.8214596543578268e-07,
      "loss": 0.699,
      "step": 23251
    },
    {
      "epoch": 0.9636536947241908,
      "grad_norm": 0.39821356534957886,
      "learning_rate": 1.8193874590741433e-07,
      "loss": 0.6346,
      "step": 23252
    },
    {
      "epoch": 0.9636951386298644,
      "grad_norm": 0.430891752243042,
      "learning_rate": 1.8173152637904598e-07,
      "loss": 0.6461,
      "step": 23253
    },
    {
      "epoch": 0.9637365825355382,
      "grad_norm": 0.3951548635959625,
      "learning_rate": 1.8152430685067763e-07,
      "loss": 0.6283,
      "step": 23254
    },
    {
      "epoch": 0.9637780264412118,
      "grad_norm": 0.41319188475608826,
      "learning_rate": 1.8131708732230928e-07,
      "loss": 0.6438,
      "step": 23255
    },
    {
      "epoch": 0.9638194703468855,
      "grad_norm": 0.38934096693992615,
      "learning_rate": 1.8110986779394093e-07,
      "loss": 0.6237,
      "step": 23256
    },
    {
      "epoch": 0.9638609142525592,
      "grad_norm": 0.42149388790130615,
      "learning_rate": 1.8090264826557255e-07,
      "loss": 0.6882,
      "step": 23257
    },
    {
      "epoch": 0.9639023581582329,
      "grad_norm": 0.41719764471054077,
      "learning_rate": 1.806954287372042e-07,
      "loss": 0.6344,
      "step": 23258
    },
    {
      "epoch": 0.9639438020639065,
      "grad_norm": 0.4408019483089447,
      "learning_rate": 1.8048820920883585e-07,
      "loss": 0.6823,
      "step": 23259
    },
    {
      "epoch": 0.9639852459695801,
      "grad_norm": 0.4154941439628601,
      "learning_rate": 1.802809896804675e-07,
      "loss": 0.6337,
      "step": 23260
    },
    {
      "epoch": 0.9640266898752539,
      "grad_norm": 0.41061410307884216,
      "learning_rate": 1.8007377015209915e-07,
      "loss": 0.6882,
      "step": 23261
    },
    {
      "epoch": 0.9640681337809275,
      "grad_norm": 0.4463736414909363,
      "learning_rate": 1.798665506237308e-07,
      "loss": 0.6268,
      "step": 23262
    },
    {
      "epoch": 0.9641095776866012,
      "grad_norm": 0.4064274728298187,
      "learning_rate": 1.7965933109536243e-07,
      "loss": 0.6704,
      "step": 23263
    },
    {
      "epoch": 0.9641510215922748,
      "grad_norm": 0.4128144383430481,
      "learning_rate": 1.7945211156699408e-07,
      "loss": 0.6272,
      "step": 23264
    },
    {
      "epoch": 0.9641924654979486,
      "grad_norm": 0.42582666873931885,
      "learning_rate": 1.7924489203862573e-07,
      "loss": 0.6475,
      "step": 23265
    },
    {
      "epoch": 0.9642339094036222,
      "grad_norm": 0.45964887738227844,
      "learning_rate": 1.7903767251025738e-07,
      "loss": 0.6324,
      "step": 23266
    },
    {
      "epoch": 0.9642753533092958,
      "grad_norm": 0.4536997377872467,
      "learning_rate": 1.7883045298188903e-07,
      "loss": 0.7354,
      "step": 23267
    },
    {
      "epoch": 0.9643167972149695,
      "grad_norm": 0.3932463824748993,
      "learning_rate": 1.7862323345352068e-07,
      "loss": 0.6995,
      "step": 23268
    },
    {
      "epoch": 0.9643582411206432,
      "grad_norm": 0.40317365527153015,
      "learning_rate": 1.7841601392515233e-07,
      "loss": 0.6438,
      "step": 23269
    },
    {
      "epoch": 0.9643996850263169,
      "grad_norm": 0.4453044831752777,
      "learning_rate": 1.7820879439678398e-07,
      "loss": 0.7097,
      "step": 23270
    },
    {
      "epoch": 0.9644411289319905,
      "grad_norm": 0.3931093215942383,
      "learning_rate": 1.7800157486841563e-07,
      "loss": 0.6686,
      "step": 23271
    },
    {
      "epoch": 0.9644825728376643,
      "grad_norm": 0.42329302430152893,
      "learning_rate": 1.7779435534004728e-07,
      "loss": 0.6182,
      "step": 23272
    },
    {
      "epoch": 0.9645240167433379,
      "grad_norm": 0.4212232530117035,
      "learning_rate": 1.7758713581167893e-07,
      "loss": 0.6995,
      "step": 23273
    },
    {
      "epoch": 0.9645654606490116,
      "grad_norm": 0.4006105065345764,
      "learning_rate": 1.7737991628331058e-07,
      "loss": 0.6525,
      "step": 23274
    },
    {
      "epoch": 0.9646069045546852,
      "grad_norm": 0.39013901352882385,
      "learning_rate": 1.7717269675494218e-07,
      "loss": 0.6158,
      "step": 23275
    },
    {
      "epoch": 0.9646483484603589,
      "grad_norm": 0.43686220049858093,
      "learning_rate": 1.7696547722657383e-07,
      "loss": 0.7319,
      "step": 23276
    },
    {
      "epoch": 0.9646897923660326,
      "grad_norm": 0.3823152780532837,
      "learning_rate": 1.7675825769820548e-07,
      "loss": 0.6122,
      "step": 23277
    },
    {
      "epoch": 0.9647312362717062,
      "grad_norm": 0.42826035618782043,
      "learning_rate": 1.7655103816983713e-07,
      "loss": 0.6324,
      "step": 23278
    },
    {
      "epoch": 0.9647726801773799,
      "grad_norm": 0.4264126121997833,
      "learning_rate": 1.7634381864146878e-07,
      "loss": 0.6349,
      "step": 23279
    },
    {
      "epoch": 0.9648141240830536,
      "grad_norm": 0.4255863130092621,
      "learning_rate": 1.7613659911310043e-07,
      "loss": 0.6659,
      "step": 23280
    },
    {
      "epoch": 0.9648555679887273,
      "grad_norm": 0.442272424697876,
      "learning_rate": 1.7592937958473208e-07,
      "loss": 0.6877,
      "step": 23281
    },
    {
      "epoch": 0.9648970118944009,
      "grad_norm": 0.4512612521648407,
      "learning_rate": 1.7572216005636373e-07,
      "loss": 0.6843,
      "step": 23282
    },
    {
      "epoch": 0.9649384558000746,
      "grad_norm": 0.4233115017414093,
      "learning_rate": 1.7551494052799538e-07,
      "loss": 0.679,
      "step": 23283
    },
    {
      "epoch": 0.9649798997057483,
      "grad_norm": 0.37886670231819153,
      "learning_rate": 1.7530772099962703e-07,
      "loss": 0.6379,
      "step": 23284
    },
    {
      "epoch": 0.9650213436114219,
      "grad_norm": 0.4184572994709015,
      "learning_rate": 1.7510050147125868e-07,
      "loss": 0.6436,
      "step": 23285
    },
    {
      "epoch": 0.9650627875170956,
      "grad_norm": 0.4252547025680542,
      "learning_rate": 1.7489328194289033e-07,
      "loss": 0.6758,
      "step": 23286
    },
    {
      "epoch": 0.9651042314227692,
      "grad_norm": 0.40730932354927063,
      "learning_rate": 1.7468606241452198e-07,
      "loss": 0.6527,
      "step": 23287
    },
    {
      "epoch": 0.965145675328443,
      "grad_norm": 0.4110737144947052,
      "learning_rate": 1.7447884288615358e-07,
      "loss": 0.6553,
      "step": 23288
    },
    {
      "epoch": 0.9651871192341166,
      "grad_norm": 0.43789148330688477,
      "learning_rate": 1.7427162335778523e-07,
      "loss": 0.6726,
      "step": 23289
    },
    {
      "epoch": 0.9652285631397903,
      "grad_norm": 0.43218380212783813,
      "learning_rate": 1.7406440382941688e-07,
      "loss": 0.6372,
      "step": 23290
    },
    {
      "epoch": 0.965270007045464,
      "grad_norm": 0.4686080515384674,
      "learning_rate": 1.7385718430104853e-07,
      "loss": 0.6636,
      "step": 23291
    },
    {
      "epoch": 0.9653114509511377,
      "grad_norm": 0.41321247816085815,
      "learning_rate": 1.7364996477268018e-07,
      "loss": 0.6456,
      "step": 23292
    },
    {
      "epoch": 0.9653528948568113,
      "grad_norm": 0.4297012984752655,
      "learning_rate": 1.7344274524431183e-07,
      "loss": 0.6809,
      "step": 23293
    },
    {
      "epoch": 0.9653943387624849,
      "grad_norm": 0.4195738732814789,
      "learning_rate": 1.7323552571594348e-07,
      "loss": 0.6797,
      "step": 23294
    },
    {
      "epoch": 0.9654357826681587,
      "grad_norm": 0.40984293818473816,
      "learning_rate": 1.7302830618757513e-07,
      "loss": 0.6268,
      "step": 23295
    },
    {
      "epoch": 0.9654772265738323,
      "grad_norm": 0.3901982307434082,
      "learning_rate": 1.7282108665920678e-07,
      "loss": 0.6401,
      "step": 23296
    },
    {
      "epoch": 0.965518670479506,
      "grad_norm": 0.4661453366279602,
      "learning_rate": 1.7261386713083843e-07,
      "loss": 0.7253,
      "step": 23297
    },
    {
      "epoch": 0.9655601143851796,
      "grad_norm": 0.41373497247695923,
      "learning_rate": 1.7240664760247008e-07,
      "loss": 0.6647,
      "step": 23298
    },
    {
      "epoch": 0.9656015582908534,
      "grad_norm": 0.3907544016838074,
      "learning_rate": 1.7219942807410173e-07,
      "loss": 0.6318,
      "step": 23299
    },
    {
      "epoch": 0.965643002196527,
      "grad_norm": 0.4454175531864166,
      "learning_rate": 1.7199220854573336e-07,
      "loss": 0.6675,
      "step": 23300
    },
    {
      "epoch": 0.9656844461022007,
      "grad_norm": 0.38820400834083557,
      "learning_rate": 1.71784989017365e-07,
      "loss": 0.6184,
      "step": 23301
    },
    {
      "epoch": 0.9657258900078743,
      "grad_norm": 0.39134272933006287,
      "learning_rate": 1.7157776948899666e-07,
      "loss": 0.6418,
      "step": 23302
    },
    {
      "epoch": 0.965767333913548,
      "grad_norm": 0.42655515670776367,
      "learning_rate": 1.713705499606283e-07,
      "loss": 0.6655,
      "step": 23303
    },
    {
      "epoch": 0.9658087778192217,
      "grad_norm": 0.44280579686164856,
      "learning_rate": 1.7116333043225993e-07,
      "loss": 0.6492,
      "step": 23304
    },
    {
      "epoch": 0.9658502217248953,
      "grad_norm": 0.4750725030899048,
      "learning_rate": 1.7095611090389158e-07,
      "loss": 0.6542,
      "step": 23305
    },
    {
      "epoch": 0.965891665630569,
      "grad_norm": 0.4715070426464081,
      "learning_rate": 1.7074889137552323e-07,
      "loss": 0.7612,
      "step": 23306
    },
    {
      "epoch": 0.9659331095362427,
      "grad_norm": 0.41819092631340027,
      "learning_rate": 1.7054167184715488e-07,
      "loss": 0.6112,
      "step": 23307
    },
    {
      "epoch": 0.9659745534419164,
      "grad_norm": 0.42313912510871887,
      "learning_rate": 1.7033445231878653e-07,
      "loss": 0.6313,
      "step": 23308
    },
    {
      "epoch": 0.96601599734759,
      "grad_norm": 0.43265488743782043,
      "learning_rate": 1.7012723279041818e-07,
      "loss": 0.7216,
      "step": 23309
    },
    {
      "epoch": 0.9660574412532638,
      "grad_norm": 0.43604597449302673,
      "learning_rate": 1.6992001326204984e-07,
      "loss": 0.6455,
      "step": 23310
    },
    {
      "epoch": 0.9660988851589374,
      "grad_norm": 0.421562135219574,
      "learning_rate": 1.6971279373368149e-07,
      "loss": 0.6637,
      "step": 23311
    },
    {
      "epoch": 0.966140329064611,
      "grad_norm": 0.39156481623649597,
      "learning_rate": 1.695055742053131e-07,
      "loss": 0.6224,
      "step": 23312
    },
    {
      "epoch": 0.9661817729702847,
      "grad_norm": 0.4224454164505005,
      "learning_rate": 1.6929835467694476e-07,
      "loss": 0.6807,
      "step": 23313
    },
    {
      "epoch": 0.9662232168759584,
      "grad_norm": 0.4051852822303772,
      "learning_rate": 1.690911351485764e-07,
      "loss": 0.6702,
      "step": 23314
    },
    {
      "epoch": 0.9662646607816321,
      "grad_norm": 0.3934318721294403,
      "learning_rate": 1.6888391562020806e-07,
      "loss": 0.6871,
      "step": 23315
    },
    {
      "epoch": 0.9663061046873057,
      "grad_norm": 0.4254280626773834,
      "learning_rate": 1.686766960918397e-07,
      "loss": 0.6562,
      "step": 23316
    },
    {
      "epoch": 0.9663475485929794,
      "grad_norm": 0.406938374042511,
      "learning_rate": 1.6846947656347136e-07,
      "loss": 0.71,
      "step": 23317
    },
    {
      "epoch": 0.9663889924986531,
      "grad_norm": 0.39998579025268555,
      "learning_rate": 1.68262257035103e-07,
      "loss": 0.6763,
      "step": 23318
    },
    {
      "epoch": 0.9664304364043268,
      "grad_norm": 0.407700777053833,
      "learning_rate": 1.6805503750673466e-07,
      "loss": 0.6858,
      "step": 23319
    },
    {
      "epoch": 0.9664718803100004,
      "grad_norm": 0.45109105110168457,
      "learning_rate": 1.678478179783663e-07,
      "loss": 0.6521,
      "step": 23320
    },
    {
      "epoch": 0.966513324215674,
      "grad_norm": 0.39635810256004333,
      "learning_rate": 1.6764059844999796e-07,
      "loss": 0.6857,
      "step": 23321
    },
    {
      "epoch": 0.9665547681213478,
      "grad_norm": 0.42575880885124207,
      "learning_rate": 1.674333789216296e-07,
      "loss": 0.6687,
      "step": 23322
    },
    {
      "epoch": 0.9665962120270214,
      "grad_norm": 0.39914363622665405,
      "learning_rate": 1.6722615939326124e-07,
      "loss": 0.6002,
      "step": 23323
    },
    {
      "epoch": 0.9666376559326951,
      "grad_norm": 0.4216274917125702,
      "learning_rate": 1.670189398648929e-07,
      "loss": 0.6581,
      "step": 23324
    },
    {
      "epoch": 0.9666790998383687,
      "grad_norm": 0.40114206075668335,
      "learning_rate": 1.668117203365245e-07,
      "loss": 0.6343,
      "step": 23325
    },
    {
      "epoch": 0.9667205437440425,
      "grad_norm": 0.42549943923950195,
      "learning_rate": 1.6660450080815616e-07,
      "loss": 0.6473,
      "step": 23326
    },
    {
      "epoch": 0.9667619876497161,
      "grad_norm": 0.4355872571468353,
      "learning_rate": 1.663972812797878e-07,
      "loss": 0.679,
      "step": 23327
    },
    {
      "epoch": 0.9668034315553897,
      "grad_norm": 0.4368642270565033,
      "learning_rate": 1.6619006175141946e-07,
      "loss": 0.7002,
      "step": 23328
    },
    {
      "epoch": 0.9668448754610635,
      "grad_norm": 0.39656850695610046,
      "learning_rate": 1.659828422230511e-07,
      "loss": 0.6362,
      "step": 23329
    },
    {
      "epoch": 0.9668863193667371,
      "grad_norm": 0.41979920864105225,
      "learning_rate": 1.6577562269468276e-07,
      "loss": 0.6925,
      "step": 23330
    },
    {
      "epoch": 0.9669277632724108,
      "grad_norm": 0.41821253299713135,
      "learning_rate": 1.655684031663144e-07,
      "loss": 0.689,
      "step": 23331
    },
    {
      "epoch": 0.9669692071780844,
      "grad_norm": 0.37016725540161133,
      "learning_rate": 1.6536118363794606e-07,
      "loss": 0.642,
      "step": 23332
    },
    {
      "epoch": 0.9670106510837582,
      "grad_norm": 0.38707494735717773,
      "learning_rate": 1.6515396410957771e-07,
      "loss": 0.6263,
      "step": 23333
    },
    {
      "epoch": 0.9670520949894318,
      "grad_norm": 0.4342324435710907,
      "learning_rate": 1.6494674458120936e-07,
      "loss": 0.7299,
      "step": 23334
    },
    {
      "epoch": 0.9670935388951055,
      "grad_norm": 0.39876458048820496,
      "learning_rate": 1.6473952505284101e-07,
      "loss": 0.6648,
      "step": 23335
    },
    {
      "epoch": 0.9671349828007791,
      "grad_norm": 0.43323883414268494,
      "learning_rate": 1.6453230552447266e-07,
      "loss": 0.6555,
      "step": 23336
    },
    {
      "epoch": 0.9671764267064528,
      "grad_norm": 0.4499939978122711,
      "learning_rate": 1.6432508599610426e-07,
      "loss": 0.6907,
      "step": 23337
    },
    {
      "epoch": 0.9672178706121265,
      "grad_norm": 0.4446576237678528,
      "learning_rate": 1.641178664677359e-07,
      "loss": 0.7026,
      "step": 23338
    },
    {
      "epoch": 0.9672593145178001,
      "grad_norm": 0.41455352306365967,
      "learning_rate": 1.6391064693936756e-07,
      "loss": 0.6632,
      "step": 23339
    },
    {
      "epoch": 0.9673007584234738,
      "grad_norm": 0.4042210280895233,
      "learning_rate": 1.6370342741099921e-07,
      "loss": 0.62,
      "step": 23340
    },
    {
      "epoch": 0.9673422023291475,
      "grad_norm": 0.4209439754486084,
      "learning_rate": 1.6349620788263086e-07,
      "loss": 0.6833,
      "step": 23341
    },
    {
      "epoch": 0.9673836462348212,
      "grad_norm": 0.40686357021331787,
      "learning_rate": 1.6328898835426251e-07,
      "loss": 0.6898,
      "step": 23342
    },
    {
      "epoch": 0.9674250901404948,
      "grad_norm": 0.4403996169567108,
      "learning_rate": 1.6308176882589416e-07,
      "loss": 0.6477,
      "step": 23343
    },
    {
      "epoch": 0.9674665340461686,
      "grad_norm": 0.4391769468784332,
      "learning_rate": 1.6287454929752581e-07,
      "loss": 0.6362,
      "step": 23344
    },
    {
      "epoch": 0.9675079779518422,
      "grad_norm": 0.38516345620155334,
      "learning_rate": 1.6266732976915747e-07,
      "loss": 0.6132,
      "step": 23345
    },
    {
      "epoch": 0.9675494218575158,
      "grad_norm": 0.4932960271835327,
      "learning_rate": 1.6246011024078912e-07,
      "loss": 0.7222,
      "step": 23346
    },
    {
      "epoch": 0.9675908657631895,
      "grad_norm": 0.4122534394264221,
      "learning_rate": 1.6225289071242077e-07,
      "loss": 0.6646,
      "step": 23347
    },
    {
      "epoch": 0.9676323096688632,
      "grad_norm": 0.3870895206928253,
      "learning_rate": 1.6204567118405242e-07,
      "loss": 0.6703,
      "step": 23348
    },
    {
      "epoch": 0.9676737535745369,
      "grad_norm": 0.41361358761787415,
      "learning_rate": 1.6183845165568404e-07,
      "loss": 0.6985,
      "step": 23349
    },
    {
      "epoch": 0.9677151974802105,
      "grad_norm": 0.41119641065597534,
      "learning_rate": 1.616312321273157e-07,
      "loss": 0.6746,
      "step": 23350
    },
    {
      "epoch": 0.9677566413858842,
      "grad_norm": 0.40768709778785706,
      "learning_rate": 1.6142401259894731e-07,
      "loss": 0.6663,
      "step": 23351
    },
    {
      "epoch": 0.9677980852915579,
      "grad_norm": 0.4416126012802124,
      "learning_rate": 1.6121679307057896e-07,
      "loss": 0.6838,
      "step": 23352
    },
    {
      "epoch": 0.9678395291972316,
      "grad_norm": 0.4883098006248474,
      "learning_rate": 1.6100957354221061e-07,
      "loss": 0.6515,
      "step": 23353
    },
    {
      "epoch": 0.9678809731029052,
      "grad_norm": 0.4071291983127594,
      "learning_rate": 1.6080235401384227e-07,
      "loss": 0.694,
      "step": 23354
    },
    {
      "epoch": 0.9679224170085788,
      "grad_norm": 0.4490273594856262,
      "learning_rate": 1.6059513448547392e-07,
      "loss": 0.6887,
      "step": 23355
    },
    {
      "epoch": 0.9679638609142526,
      "grad_norm": 0.46576401591300964,
      "learning_rate": 1.6038791495710557e-07,
      "loss": 0.6078,
      "step": 23356
    },
    {
      "epoch": 0.9680053048199262,
      "grad_norm": 0.3993131220340729,
      "learning_rate": 1.6018069542873722e-07,
      "loss": 0.6732,
      "step": 23357
    },
    {
      "epoch": 0.9680467487255999,
      "grad_norm": 0.40115588903427124,
      "learning_rate": 1.5997347590036887e-07,
      "loss": 0.655,
      "step": 23358
    },
    {
      "epoch": 0.9680881926312735,
      "grad_norm": 0.430477112531662,
      "learning_rate": 1.5976625637200052e-07,
      "loss": 0.6547,
      "step": 23359
    },
    {
      "epoch": 0.9681296365369473,
      "grad_norm": 0.436361700296402,
      "learning_rate": 1.5955903684363217e-07,
      "loss": 0.6436,
      "step": 23360
    },
    {
      "epoch": 0.9681710804426209,
      "grad_norm": 0.46355488896369934,
      "learning_rate": 1.593518173152638e-07,
      "loss": 0.6467,
      "step": 23361
    },
    {
      "epoch": 0.9682125243482946,
      "grad_norm": 0.39769119024276733,
      "learning_rate": 1.5914459778689544e-07,
      "loss": 0.6144,
      "step": 23362
    },
    {
      "epoch": 0.9682539682539683,
      "grad_norm": 0.4173765182495117,
      "learning_rate": 1.589373782585271e-07,
      "loss": 0.6442,
      "step": 23363
    },
    {
      "epoch": 0.9682954121596419,
      "grad_norm": 0.47269025444984436,
      "learning_rate": 1.5873015873015874e-07,
      "loss": 0.702,
      "step": 23364
    },
    {
      "epoch": 0.9683368560653156,
      "grad_norm": 0.4626195728778839,
      "learning_rate": 1.585229392017904e-07,
      "loss": 0.7061,
      "step": 23365
    },
    {
      "epoch": 0.9683782999709892,
      "grad_norm": 0.4063701927661896,
      "learning_rate": 1.5831571967342204e-07,
      "loss": 0.6123,
      "step": 23366
    },
    {
      "epoch": 0.968419743876663,
      "grad_norm": 0.41015464067459106,
      "learning_rate": 1.581085001450537e-07,
      "loss": 0.6926,
      "step": 23367
    },
    {
      "epoch": 0.9684611877823366,
      "grad_norm": 0.41619399189949036,
      "learning_rate": 1.5790128061668534e-07,
      "loss": 0.6803,
      "step": 23368
    },
    {
      "epoch": 0.9685026316880103,
      "grad_norm": 0.39935314655303955,
      "learning_rate": 1.57694061088317e-07,
      "loss": 0.689,
      "step": 23369
    },
    {
      "epoch": 0.9685440755936839,
      "grad_norm": 0.4308881163597107,
      "learning_rate": 1.5748684155994862e-07,
      "loss": 0.6656,
      "step": 23370
    },
    {
      "epoch": 0.9685855194993577,
      "grad_norm": 0.4899929463863373,
      "learning_rate": 1.5727962203158027e-07,
      "loss": 0.6705,
      "step": 23371
    },
    {
      "epoch": 0.9686269634050313,
      "grad_norm": 0.41016316413879395,
      "learning_rate": 1.5707240250321192e-07,
      "loss": 0.6702,
      "step": 23372
    },
    {
      "epoch": 0.9686684073107049,
      "grad_norm": 0.4686843156814575,
      "learning_rate": 1.5686518297484357e-07,
      "loss": 0.6711,
      "step": 23373
    },
    {
      "epoch": 0.9687098512163786,
      "grad_norm": 0.4339531362056732,
      "learning_rate": 1.566579634464752e-07,
      "loss": 0.6273,
      "step": 23374
    },
    {
      "epoch": 0.9687512951220523,
      "grad_norm": 0.3982437551021576,
      "learning_rate": 1.5645074391810684e-07,
      "loss": 0.6021,
      "step": 23375
    },
    {
      "epoch": 0.968792739027726,
      "grad_norm": 0.424327552318573,
      "learning_rate": 1.5624352438973852e-07,
      "loss": 0.676,
      "step": 23376
    },
    {
      "epoch": 0.9688341829333996,
      "grad_norm": 0.40400204062461853,
      "learning_rate": 1.5603630486137014e-07,
      "loss": 0.6531,
      "step": 23377
    },
    {
      "epoch": 0.9688756268390734,
      "grad_norm": 0.41770923137664795,
      "learning_rate": 1.558290853330018e-07,
      "loss": 0.6454,
      "step": 23378
    },
    {
      "epoch": 0.968917070744747,
      "grad_norm": 0.4862979054450989,
      "learning_rate": 1.5562186580463344e-07,
      "loss": 0.7075,
      "step": 23379
    },
    {
      "epoch": 0.9689585146504207,
      "grad_norm": 0.43865296244621277,
      "learning_rate": 1.554146462762651e-07,
      "loss": 0.6987,
      "step": 23380
    },
    {
      "epoch": 0.9689999585560943,
      "grad_norm": 0.4065636992454529,
      "learning_rate": 1.5520742674789675e-07,
      "loss": 0.671,
      "step": 23381
    },
    {
      "epoch": 0.969041402461768,
      "grad_norm": 0.42944273352622986,
      "learning_rate": 1.550002072195284e-07,
      "loss": 0.6766,
      "step": 23382
    },
    {
      "epoch": 0.9690828463674417,
      "grad_norm": 0.4478388726711273,
      "learning_rate": 1.5479298769116002e-07,
      "loss": 0.7327,
      "step": 23383
    },
    {
      "epoch": 0.9691242902731153,
      "grad_norm": 0.3859099745750427,
      "learning_rate": 1.5458576816279167e-07,
      "loss": 0.6332,
      "step": 23384
    },
    {
      "epoch": 0.969165734178789,
      "grad_norm": 0.39606973528862,
      "learning_rate": 1.5437854863442332e-07,
      "loss": 0.6318,
      "step": 23385
    },
    {
      "epoch": 0.9692071780844627,
      "grad_norm": 0.41753703355789185,
      "learning_rate": 1.5417132910605497e-07,
      "loss": 0.6798,
      "step": 23386
    },
    {
      "epoch": 0.9692486219901364,
      "grad_norm": 0.447609007358551,
      "learning_rate": 1.5396410957768662e-07,
      "loss": 0.6615,
      "step": 23387
    },
    {
      "epoch": 0.96929006589581,
      "grad_norm": 0.4227668344974518,
      "learning_rate": 1.5375689004931827e-07,
      "loss": 0.6288,
      "step": 23388
    },
    {
      "epoch": 0.9693315098014836,
      "grad_norm": 0.42959046363830566,
      "learning_rate": 1.535496705209499e-07,
      "loss": 0.6447,
      "step": 23389
    },
    {
      "epoch": 0.9693729537071574,
      "grad_norm": 0.39226147532463074,
      "learning_rate": 1.5334245099258155e-07,
      "loss": 0.6409,
      "step": 23390
    },
    {
      "epoch": 0.969414397612831,
      "grad_norm": 0.3937316834926605,
      "learning_rate": 1.531352314642132e-07,
      "loss": 0.6512,
      "step": 23391
    },
    {
      "epoch": 0.9694558415185047,
      "grad_norm": 0.4207232594490051,
      "learning_rate": 1.5292801193584485e-07,
      "loss": 0.6849,
      "step": 23392
    },
    {
      "epoch": 0.9694972854241783,
      "grad_norm": 0.45545512437820435,
      "learning_rate": 1.527207924074765e-07,
      "loss": 0.6663,
      "step": 23393
    },
    {
      "epoch": 0.9695387293298521,
      "grad_norm": 0.42424553632736206,
      "learning_rate": 1.5251357287910815e-07,
      "loss": 0.662,
      "step": 23394
    },
    {
      "epoch": 0.9695801732355257,
      "grad_norm": 0.4297972321510315,
      "learning_rate": 1.5230635335073977e-07,
      "loss": 0.6929,
      "step": 23395
    },
    {
      "epoch": 0.9696216171411994,
      "grad_norm": 0.4273180663585663,
      "learning_rate": 1.5209913382237142e-07,
      "loss": 0.6865,
      "step": 23396
    },
    {
      "epoch": 0.969663061046873,
      "grad_norm": 0.42989614605903625,
      "learning_rate": 1.5189191429400307e-07,
      "loss": 0.6635,
      "step": 23397
    },
    {
      "epoch": 0.9697045049525467,
      "grad_norm": 0.4713872969150543,
      "learning_rate": 1.5168469476563472e-07,
      "loss": 0.6807,
      "step": 23398
    },
    {
      "epoch": 0.9697459488582204,
      "grad_norm": 0.3995877802371979,
      "learning_rate": 1.5147747523726637e-07,
      "loss": 0.6591,
      "step": 23399
    },
    {
      "epoch": 0.969787392763894,
      "grad_norm": 0.4467785954475403,
      "learning_rate": 1.5127025570889802e-07,
      "loss": 0.6897,
      "step": 23400
    },
    {
      "epoch": 0.9698288366695678,
      "grad_norm": 0.4269404113292694,
      "learning_rate": 1.5106303618052965e-07,
      "loss": 0.6963,
      "step": 23401
    },
    {
      "epoch": 0.9698702805752414,
      "grad_norm": 0.40911659598350525,
      "learning_rate": 1.508558166521613e-07,
      "loss": 0.6409,
      "step": 23402
    },
    {
      "epoch": 0.9699117244809151,
      "grad_norm": 0.4664289951324463,
      "learning_rate": 1.5064859712379295e-07,
      "loss": 0.6381,
      "step": 23403
    },
    {
      "epoch": 0.9699531683865887,
      "grad_norm": 0.4427375793457031,
      "learning_rate": 1.504413775954246e-07,
      "loss": 0.6588,
      "step": 23404
    },
    {
      "epoch": 0.9699946122922625,
      "grad_norm": 0.40940624475479126,
      "learning_rate": 1.5023415806705625e-07,
      "loss": 0.6373,
      "step": 23405
    },
    {
      "epoch": 0.9700360561979361,
      "grad_norm": 0.44930630922317505,
      "learning_rate": 1.500269385386879e-07,
      "loss": 0.6866,
      "step": 23406
    },
    {
      "epoch": 0.9700775001036097,
      "grad_norm": 0.41708970069885254,
      "learning_rate": 1.4981971901031955e-07,
      "loss": 0.673,
      "step": 23407
    },
    {
      "epoch": 0.9701189440092834,
      "grad_norm": 0.40876084566116333,
      "learning_rate": 1.4961249948195117e-07,
      "loss": 0.7266,
      "step": 23408
    },
    {
      "epoch": 0.9701603879149571,
      "grad_norm": 0.4588559865951538,
      "learning_rate": 1.4940527995358282e-07,
      "loss": 0.7056,
      "step": 23409
    },
    {
      "epoch": 0.9702018318206308,
      "grad_norm": 0.4186292290687561,
      "learning_rate": 1.4919806042521447e-07,
      "loss": 0.6167,
      "step": 23410
    },
    {
      "epoch": 0.9702432757263044,
      "grad_norm": 0.4015178084373474,
      "learning_rate": 1.4899084089684612e-07,
      "loss": 0.6266,
      "step": 23411
    },
    {
      "epoch": 0.9702847196319782,
      "grad_norm": 0.43719688057899475,
      "learning_rate": 1.4878362136847777e-07,
      "loss": 0.6555,
      "step": 23412
    },
    {
      "epoch": 0.9703261635376518,
      "grad_norm": 0.423287957906723,
      "learning_rate": 1.4857640184010942e-07,
      "loss": 0.6404,
      "step": 23413
    },
    {
      "epoch": 0.9703676074433255,
      "grad_norm": 0.427774578332901,
      "learning_rate": 1.4836918231174107e-07,
      "loss": 0.6619,
      "step": 23414
    },
    {
      "epoch": 0.9704090513489991,
      "grad_norm": 0.40787872672080994,
      "learning_rate": 1.4816196278337272e-07,
      "loss": 0.6544,
      "step": 23415
    },
    {
      "epoch": 0.9704504952546728,
      "grad_norm": 0.4110839366912842,
      "learning_rate": 1.4795474325500438e-07,
      "loss": 0.6324,
      "step": 23416
    },
    {
      "epoch": 0.9704919391603465,
      "grad_norm": 0.46905484795570374,
      "learning_rate": 1.47747523726636e-07,
      "loss": 0.729,
      "step": 23417
    },
    {
      "epoch": 0.9705333830660201,
      "grad_norm": 0.41941845417022705,
      "learning_rate": 1.4754030419826765e-07,
      "loss": 0.661,
      "step": 23418
    },
    {
      "epoch": 0.9705748269716938,
      "grad_norm": 0.40946829319000244,
      "learning_rate": 1.473330846698993e-07,
      "loss": 0.644,
      "step": 23419
    },
    {
      "epoch": 0.9706162708773675,
      "grad_norm": 0.4393613636493683,
      "learning_rate": 1.4712586514153095e-07,
      "loss": 0.6509,
      "step": 23420
    },
    {
      "epoch": 0.9706577147830412,
      "grad_norm": 0.41081106662750244,
      "learning_rate": 1.469186456131626e-07,
      "loss": 0.6946,
      "step": 23421
    },
    {
      "epoch": 0.9706991586887148,
      "grad_norm": 0.3804410398006439,
      "learning_rate": 1.4671142608479425e-07,
      "loss": 0.6548,
      "step": 23422
    },
    {
      "epoch": 0.9707406025943885,
      "grad_norm": 0.4108729362487793,
      "learning_rate": 1.465042065564259e-07,
      "loss": 0.6477,
      "step": 23423
    },
    {
      "epoch": 0.9707820465000622,
      "grad_norm": 0.4490480422973633,
      "learning_rate": 1.4629698702805755e-07,
      "loss": 0.7046,
      "step": 23424
    },
    {
      "epoch": 0.9708234904057358,
      "grad_norm": 0.42290663719177246,
      "learning_rate": 1.460897674996892e-07,
      "loss": 0.6602,
      "step": 23425
    },
    {
      "epoch": 0.9708649343114095,
      "grad_norm": 0.400392085313797,
      "learning_rate": 1.4588254797132083e-07,
      "loss": 0.6521,
      "step": 23426
    },
    {
      "epoch": 0.9709063782170831,
      "grad_norm": 0.41102978587150574,
      "learning_rate": 1.4567532844295248e-07,
      "loss": 0.6268,
      "step": 23427
    },
    {
      "epoch": 0.9709478221227569,
      "grad_norm": 0.4439678192138672,
      "learning_rate": 1.4546810891458413e-07,
      "loss": 0.6189,
      "step": 23428
    },
    {
      "epoch": 0.9709892660284305,
      "grad_norm": 0.43098974227905273,
      "learning_rate": 1.4526088938621578e-07,
      "loss": 0.6694,
      "step": 23429
    },
    {
      "epoch": 0.9710307099341042,
      "grad_norm": 0.41941556334495544,
      "learning_rate": 1.4505366985784743e-07,
      "loss": 0.6556,
      "step": 23430
    },
    {
      "epoch": 0.9710721538397779,
      "grad_norm": 0.4133436679840088,
      "learning_rate": 1.4484645032947908e-07,
      "loss": 0.6183,
      "step": 23431
    },
    {
      "epoch": 0.9711135977454516,
      "grad_norm": 0.4090350866317749,
      "learning_rate": 1.446392308011107e-07,
      "loss": 0.6233,
      "step": 23432
    },
    {
      "epoch": 0.9711550416511252,
      "grad_norm": 0.39913684129714966,
      "learning_rate": 1.4443201127274235e-07,
      "loss": 0.6758,
      "step": 23433
    },
    {
      "epoch": 0.9711964855567988,
      "grad_norm": 0.3914391100406647,
      "learning_rate": 1.44224791744374e-07,
      "loss": 0.6084,
      "step": 23434
    },
    {
      "epoch": 0.9712379294624726,
      "grad_norm": 0.42958560585975647,
      "learning_rate": 1.4401757221600565e-07,
      "loss": 0.6943,
      "step": 23435
    },
    {
      "epoch": 0.9712793733681462,
      "grad_norm": 0.42146405577659607,
      "learning_rate": 1.438103526876373e-07,
      "loss": 0.6588,
      "step": 23436
    },
    {
      "epoch": 0.9713208172738199,
      "grad_norm": 0.4668481647968292,
      "learning_rate": 1.4360313315926895e-07,
      "loss": 0.7119,
      "step": 23437
    },
    {
      "epoch": 0.9713622611794935,
      "grad_norm": 0.42503249645233154,
      "learning_rate": 1.4339591363090058e-07,
      "loss": 0.6243,
      "step": 23438
    },
    {
      "epoch": 0.9714037050851673,
      "grad_norm": 0.4010186791419983,
      "learning_rate": 1.4318869410253223e-07,
      "loss": 0.6362,
      "step": 23439
    },
    {
      "epoch": 0.9714451489908409,
      "grad_norm": 0.4168928265571594,
      "learning_rate": 1.4298147457416388e-07,
      "loss": 0.6802,
      "step": 23440
    },
    {
      "epoch": 0.9714865928965146,
      "grad_norm": 0.4136193096637726,
      "learning_rate": 1.4277425504579553e-07,
      "loss": 0.6711,
      "step": 23441
    },
    {
      "epoch": 0.9715280368021882,
      "grad_norm": 0.40608352422714233,
      "learning_rate": 1.4256703551742718e-07,
      "loss": 0.649,
      "step": 23442
    },
    {
      "epoch": 0.9715694807078619,
      "grad_norm": 0.4512658715248108,
      "learning_rate": 1.4235981598905883e-07,
      "loss": 0.6934,
      "step": 23443
    },
    {
      "epoch": 0.9716109246135356,
      "grad_norm": 0.40187135338783264,
      "learning_rate": 1.4215259646069048e-07,
      "loss": 0.7305,
      "step": 23444
    },
    {
      "epoch": 0.9716523685192092,
      "grad_norm": 0.40227022767066956,
      "learning_rate": 1.419453769323221e-07,
      "loss": 0.6433,
      "step": 23445
    },
    {
      "epoch": 0.971693812424883,
      "grad_norm": 0.3988104462623596,
      "learning_rate": 1.4173815740395375e-07,
      "loss": 0.6602,
      "step": 23446
    },
    {
      "epoch": 0.9717352563305566,
      "grad_norm": 0.39173558354377747,
      "learning_rate": 1.415309378755854e-07,
      "loss": 0.6198,
      "step": 23447
    },
    {
      "epoch": 0.9717767002362303,
      "grad_norm": 0.41801536083221436,
      "learning_rate": 1.4132371834721705e-07,
      "loss": 0.6556,
      "step": 23448
    },
    {
      "epoch": 0.9718181441419039,
      "grad_norm": 0.41201284527778625,
      "learning_rate": 1.411164988188487e-07,
      "loss": 0.6741,
      "step": 23449
    },
    {
      "epoch": 0.9718595880475775,
      "grad_norm": 0.41657787561416626,
      "learning_rate": 1.4090927929048035e-07,
      "loss": 0.6713,
      "step": 23450
    },
    {
      "epoch": 0.9719010319532513,
      "grad_norm": 0.43569543957710266,
      "learning_rate": 1.4070205976211198e-07,
      "loss": 0.7495,
      "step": 23451
    },
    {
      "epoch": 0.9719424758589249,
      "grad_norm": 0.43175867199897766,
      "learning_rate": 1.4049484023374363e-07,
      "loss": 0.6631,
      "step": 23452
    },
    {
      "epoch": 0.9719839197645986,
      "grad_norm": 0.43650224804878235,
      "learning_rate": 1.4028762070537528e-07,
      "loss": 0.6761,
      "step": 23453
    },
    {
      "epoch": 0.9720253636702723,
      "grad_norm": 0.40753138065338135,
      "learning_rate": 1.4008040117700693e-07,
      "loss": 0.6531,
      "step": 23454
    },
    {
      "epoch": 0.972066807575946,
      "grad_norm": 0.46421778202056885,
      "learning_rate": 1.3987318164863858e-07,
      "loss": 0.6708,
      "step": 23455
    },
    {
      "epoch": 0.9721082514816196,
      "grad_norm": 0.4288656413555145,
      "learning_rate": 1.3966596212027023e-07,
      "loss": 0.6533,
      "step": 23456
    },
    {
      "epoch": 0.9721496953872933,
      "grad_norm": 0.4824259281158447,
      "learning_rate": 1.3945874259190185e-07,
      "loss": 0.6387,
      "step": 23457
    },
    {
      "epoch": 0.972191139292967,
      "grad_norm": 0.4637094736099243,
      "learning_rate": 1.392515230635335e-07,
      "loss": 0.7178,
      "step": 23458
    },
    {
      "epoch": 0.9722325831986406,
      "grad_norm": 0.43972375988960266,
      "learning_rate": 1.3904430353516516e-07,
      "loss": 0.6492,
      "step": 23459
    },
    {
      "epoch": 0.9722740271043143,
      "grad_norm": 0.45529282093048096,
      "learning_rate": 1.388370840067968e-07,
      "loss": 0.6975,
      "step": 23460
    },
    {
      "epoch": 0.9723154710099879,
      "grad_norm": 0.4419008791446686,
      "learning_rate": 1.3862986447842846e-07,
      "loss": 0.6423,
      "step": 23461
    },
    {
      "epoch": 0.9723569149156617,
      "grad_norm": 0.431314617395401,
      "learning_rate": 1.384226449500601e-07,
      "loss": 0.6711,
      "step": 23462
    },
    {
      "epoch": 0.9723983588213353,
      "grad_norm": 0.5047176480293274,
      "learning_rate": 1.3821542542169176e-07,
      "loss": 0.7446,
      "step": 23463
    },
    {
      "epoch": 0.972439802727009,
      "grad_norm": 0.42560112476348877,
      "learning_rate": 1.380082058933234e-07,
      "loss": 0.6559,
      "step": 23464
    },
    {
      "epoch": 0.9724812466326826,
      "grad_norm": 0.4083189070224762,
      "learning_rate": 1.3780098636495503e-07,
      "loss": 0.6489,
      "step": 23465
    },
    {
      "epoch": 0.9725226905383564,
      "grad_norm": 0.4638565182685852,
      "learning_rate": 1.3759376683658668e-07,
      "loss": 0.6484,
      "step": 23466
    },
    {
      "epoch": 0.97256413444403,
      "grad_norm": 0.394010990858078,
      "learning_rate": 1.3738654730821833e-07,
      "loss": 0.6648,
      "step": 23467
    },
    {
      "epoch": 0.9726055783497036,
      "grad_norm": 0.38691356778144836,
      "learning_rate": 1.3717932777984998e-07,
      "loss": 0.6201,
      "step": 23468
    },
    {
      "epoch": 0.9726470222553774,
      "grad_norm": 0.4198537766933441,
      "learning_rate": 1.3697210825148163e-07,
      "loss": 0.6251,
      "step": 23469
    },
    {
      "epoch": 0.972688466161051,
      "grad_norm": 0.4543820917606354,
      "learning_rate": 1.3676488872311328e-07,
      "loss": 0.6804,
      "step": 23470
    },
    {
      "epoch": 0.9727299100667247,
      "grad_norm": 0.43026572465896606,
      "learning_rate": 1.3655766919474493e-07,
      "loss": 0.7375,
      "step": 23471
    },
    {
      "epoch": 0.9727713539723983,
      "grad_norm": 0.4219870865345001,
      "learning_rate": 1.3635044966637658e-07,
      "loss": 0.6926,
      "step": 23472
    },
    {
      "epoch": 0.9728127978780721,
      "grad_norm": 0.46639543771743774,
      "learning_rate": 1.3614323013800823e-07,
      "loss": 0.6682,
      "step": 23473
    },
    {
      "epoch": 0.9728542417837457,
      "grad_norm": 0.3914889395236969,
      "learning_rate": 1.3593601060963986e-07,
      "loss": 0.6259,
      "step": 23474
    },
    {
      "epoch": 0.9728956856894194,
      "grad_norm": 0.41712695360183716,
      "learning_rate": 1.357287910812715e-07,
      "loss": 0.6748,
      "step": 23475
    },
    {
      "epoch": 0.972937129595093,
      "grad_norm": 0.4274061620235443,
      "learning_rate": 1.3552157155290316e-07,
      "loss": 0.6654,
      "step": 23476
    },
    {
      "epoch": 0.9729785735007667,
      "grad_norm": 0.4057908058166504,
      "learning_rate": 1.353143520245348e-07,
      "loss": 0.6598,
      "step": 23477
    },
    {
      "epoch": 0.9730200174064404,
      "grad_norm": 0.4555774927139282,
      "learning_rate": 1.3510713249616646e-07,
      "loss": 0.7072,
      "step": 23478
    },
    {
      "epoch": 0.973061461312114,
      "grad_norm": 0.4330986440181732,
      "learning_rate": 1.348999129677981e-07,
      "loss": 0.6842,
      "step": 23479
    },
    {
      "epoch": 0.9731029052177878,
      "grad_norm": 0.42811378836631775,
      "learning_rate": 1.3469269343942976e-07,
      "loss": 0.6814,
      "step": 23480
    },
    {
      "epoch": 0.9731443491234614,
      "grad_norm": 0.3939908444881439,
      "learning_rate": 1.3448547391106138e-07,
      "loss": 0.6355,
      "step": 23481
    },
    {
      "epoch": 0.9731857930291351,
      "grad_norm": 0.43714234232902527,
      "learning_rate": 1.3427825438269303e-07,
      "loss": 0.6729,
      "step": 23482
    },
    {
      "epoch": 0.9732272369348087,
      "grad_norm": 0.4425908327102661,
      "learning_rate": 1.3407103485432468e-07,
      "loss": 0.6401,
      "step": 23483
    },
    {
      "epoch": 0.9732686808404825,
      "grad_norm": 0.38999950885772705,
      "learning_rate": 1.3386381532595633e-07,
      "loss": 0.6245,
      "step": 23484
    },
    {
      "epoch": 0.9733101247461561,
      "grad_norm": 0.46349036693573,
      "learning_rate": 1.3365659579758798e-07,
      "loss": 0.6816,
      "step": 23485
    },
    {
      "epoch": 0.9733515686518297,
      "grad_norm": 0.41009509563446045,
      "learning_rate": 1.3344937626921964e-07,
      "loss": 0.6399,
      "step": 23486
    },
    {
      "epoch": 0.9733930125575034,
      "grad_norm": 0.3979561924934387,
      "learning_rate": 1.3324215674085129e-07,
      "loss": 0.6628,
      "step": 23487
    },
    {
      "epoch": 0.9734344564631771,
      "grad_norm": 0.3908049464225769,
      "learning_rate": 1.330349372124829e-07,
      "loss": 0.6212,
      "step": 23488
    },
    {
      "epoch": 0.9734759003688508,
      "grad_norm": 0.43977534770965576,
      "learning_rate": 1.3282771768411456e-07,
      "loss": 0.6567,
      "step": 23489
    },
    {
      "epoch": 0.9735173442745244,
      "grad_norm": 0.4015119969844818,
      "learning_rate": 1.326204981557462e-07,
      "loss": 0.6228,
      "step": 23490
    },
    {
      "epoch": 0.9735587881801981,
      "grad_norm": 0.41544198989868164,
      "learning_rate": 1.3241327862737786e-07,
      "loss": 0.6439,
      "step": 23491
    },
    {
      "epoch": 0.9736002320858718,
      "grad_norm": 0.48099830746650696,
      "learning_rate": 1.322060590990095e-07,
      "loss": 0.7253,
      "step": 23492
    },
    {
      "epoch": 0.9736416759915455,
      "grad_norm": 0.42074263095855713,
      "learning_rate": 1.3199883957064116e-07,
      "loss": 0.6509,
      "step": 23493
    },
    {
      "epoch": 0.9736831198972191,
      "grad_norm": 0.4198032021522522,
      "learning_rate": 1.3179162004227279e-07,
      "loss": 0.7036,
      "step": 23494
    },
    {
      "epoch": 0.9737245638028927,
      "grad_norm": 0.3943674862384796,
      "learning_rate": 1.3158440051390444e-07,
      "loss": 0.6572,
      "step": 23495
    },
    {
      "epoch": 0.9737660077085665,
      "grad_norm": 0.4248095154762268,
      "learning_rate": 1.3137718098553609e-07,
      "loss": 0.689,
      "step": 23496
    },
    {
      "epoch": 0.9738074516142401,
      "grad_norm": 0.4173724353313446,
      "learning_rate": 1.3116996145716774e-07,
      "loss": 0.6312,
      "step": 23497
    },
    {
      "epoch": 0.9738488955199138,
      "grad_norm": 0.4483766555786133,
      "learning_rate": 1.3096274192879939e-07,
      "loss": 0.6439,
      "step": 23498
    },
    {
      "epoch": 0.9738903394255874,
      "grad_norm": 0.4117897152900696,
      "learning_rate": 1.3075552240043104e-07,
      "loss": 0.6943,
      "step": 23499
    },
    {
      "epoch": 0.9739317833312612,
      "grad_norm": 0.4290332794189453,
      "learning_rate": 1.3054830287206266e-07,
      "loss": 0.6876,
      "step": 23500
    },
    {
      "epoch": 0.9739732272369348,
      "grad_norm": 0.3797157406806946,
      "learning_rate": 1.303410833436943e-07,
      "loss": 0.6178,
      "step": 23501
    },
    {
      "epoch": 0.9740146711426085,
      "grad_norm": 0.3939172029495239,
      "learning_rate": 1.3013386381532596e-07,
      "loss": 0.665,
      "step": 23502
    },
    {
      "epoch": 0.9740561150482822,
      "grad_norm": 0.4490468204021454,
      "learning_rate": 1.299266442869576e-07,
      "loss": 0.6869,
      "step": 23503
    },
    {
      "epoch": 0.9740975589539558,
      "grad_norm": 0.443692684173584,
      "learning_rate": 1.2971942475858926e-07,
      "loss": 0.6444,
      "step": 23504
    },
    {
      "epoch": 0.9741390028596295,
      "grad_norm": 0.47876259684562683,
      "learning_rate": 1.295122052302209e-07,
      "loss": 0.629,
      "step": 23505
    },
    {
      "epoch": 0.9741804467653031,
      "grad_norm": 0.4409225881099701,
      "learning_rate": 1.2930498570185254e-07,
      "loss": 0.693,
      "step": 23506
    },
    {
      "epoch": 0.9742218906709769,
      "grad_norm": 0.454172283411026,
      "learning_rate": 1.2909776617348419e-07,
      "loss": 0.6724,
      "step": 23507
    },
    {
      "epoch": 0.9742633345766505,
      "grad_norm": 0.410898357629776,
      "learning_rate": 1.2889054664511584e-07,
      "loss": 0.6686,
      "step": 23508
    },
    {
      "epoch": 0.9743047784823242,
      "grad_norm": 0.4191705584526062,
      "learning_rate": 1.286833271167475e-07,
      "loss": 0.6721,
      "step": 23509
    },
    {
      "epoch": 0.9743462223879978,
      "grad_norm": 0.4100308120250702,
      "learning_rate": 1.2847610758837914e-07,
      "loss": 0.6765,
      "step": 23510
    },
    {
      "epoch": 0.9743876662936715,
      "grad_norm": 0.40019726753234863,
      "learning_rate": 1.282688880600108e-07,
      "loss": 0.632,
      "step": 23511
    },
    {
      "epoch": 0.9744291101993452,
      "grad_norm": 0.4674622714519501,
      "learning_rate": 1.280616685316424e-07,
      "loss": 0.6628,
      "step": 23512
    },
    {
      "epoch": 0.9744705541050188,
      "grad_norm": 0.410868376493454,
      "learning_rate": 1.2785444900327406e-07,
      "loss": 0.6991,
      "step": 23513
    },
    {
      "epoch": 0.9745119980106925,
      "grad_norm": 0.40889397263526917,
      "learning_rate": 1.276472294749057e-07,
      "loss": 0.6384,
      "step": 23514
    },
    {
      "epoch": 0.9745534419163662,
      "grad_norm": 0.39269715547561646,
      "learning_rate": 1.2744000994653736e-07,
      "loss": 0.6394,
      "step": 23515
    },
    {
      "epoch": 0.9745948858220399,
      "grad_norm": 0.4291469156742096,
      "learning_rate": 1.2723279041816901e-07,
      "loss": 0.6755,
      "step": 23516
    },
    {
      "epoch": 0.9746363297277135,
      "grad_norm": 0.4319342374801636,
      "learning_rate": 1.2702557088980066e-07,
      "loss": 0.6724,
      "step": 23517
    },
    {
      "epoch": 0.9746777736333873,
      "grad_norm": 0.41709399223327637,
      "learning_rate": 1.2681835136143231e-07,
      "loss": 0.6733,
      "step": 23518
    },
    {
      "epoch": 0.9747192175390609,
      "grad_norm": 0.40622055530548096,
      "learning_rate": 1.2661113183306396e-07,
      "loss": 0.6843,
      "step": 23519
    },
    {
      "epoch": 0.9747606614447345,
      "grad_norm": 0.431393027305603,
      "learning_rate": 1.2640391230469561e-07,
      "loss": 0.6304,
      "step": 23520
    },
    {
      "epoch": 0.9748021053504082,
      "grad_norm": 0.41044142842292786,
      "learning_rate": 1.2619669277632724e-07,
      "loss": 0.637,
      "step": 23521
    },
    {
      "epoch": 0.9748435492560819,
      "grad_norm": 0.4152747392654419,
      "learning_rate": 1.259894732479589e-07,
      "loss": 0.6643,
      "step": 23522
    },
    {
      "epoch": 0.9748849931617556,
      "grad_norm": 0.39618974924087524,
      "learning_rate": 1.2578225371959054e-07,
      "loss": 0.6296,
      "step": 23523
    },
    {
      "epoch": 0.9749264370674292,
      "grad_norm": 0.398184597492218,
      "learning_rate": 1.255750341912222e-07,
      "loss": 0.618,
      "step": 23524
    },
    {
      "epoch": 0.9749678809731029,
      "grad_norm": 0.4166944622993469,
      "learning_rate": 1.2536781466285384e-07,
      "loss": 0.6157,
      "step": 23525
    },
    {
      "epoch": 0.9750093248787766,
      "grad_norm": 0.4399368464946747,
      "learning_rate": 1.251605951344855e-07,
      "loss": 0.6594,
      "step": 23526
    },
    {
      "epoch": 0.9750507687844503,
      "grad_norm": 0.42428773641586304,
      "learning_rate": 1.2495337560611714e-07,
      "loss": 0.6667,
      "step": 23527
    },
    {
      "epoch": 0.9750922126901239,
      "grad_norm": 0.4176606833934784,
      "learning_rate": 1.247461560777488e-07,
      "loss": 0.6824,
      "step": 23528
    },
    {
      "epoch": 0.9751336565957975,
      "grad_norm": 0.397001713514328,
      "learning_rate": 1.2453893654938044e-07,
      "loss": 0.6259,
      "step": 23529
    },
    {
      "epoch": 0.9751751005014713,
      "grad_norm": 0.47528907656669617,
      "learning_rate": 1.2433171702101207e-07,
      "loss": 0.689,
      "step": 23530
    },
    {
      "epoch": 0.9752165444071449,
      "grad_norm": 0.44263598322868347,
      "learning_rate": 1.2412449749264372e-07,
      "loss": 0.6686,
      "step": 23531
    },
    {
      "epoch": 0.9752579883128186,
      "grad_norm": 0.42034009099006653,
      "learning_rate": 1.2391727796427537e-07,
      "loss": 0.6261,
      "step": 23532
    },
    {
      "epoch": 0.9752994322184922,
      "grad_norm": 0.4552428126335144,
      "learning_rate": 1.2371005843590702e-07,
      "loss": 0.6697,
      "step": 23533
    },
    {
      "epoch": 0.975340876124166,
      "grad_norm": 0.4031771719455719,
      "learning_rate": 1.2350283890753867e-07,
      "loss": 0.6548,
      "step": 23534
    },
    {
      "epoch": 0.9753823200298396,
      "grad_norm": 0.3973510265350342,
      "learning_rate": 1.2329561937917032e-07,
      "loss": 0.6367,
      "step": 23535
    },
    {
      "epoch": 0.9754237639355133,
      "grad_norm": 0.4100991189479828,
      "learning_rate": 1.2308839985080197e-07,
      "loss": 0.6609,
      "step": 23536
    },
    {
      "epoch": 0.975465207841187,
      "grad_norm": 0.466200053691864,
      "learning_rate": 1.228811803224336e-07,
      "loss": 0.7,
      "step": 23537
    },
    {
      "epoch": 0.9755066517468606,
      "grad_norm": 0.411411315202713,
      "learning_rate": 1.2267396079406524e-07,
      "loss": 0.6377,
      "step": 23538
    },
    {
      "epoch": 0.9755480956525343,
      "grad_norm": 0.4675273001194,
      "learning_rate": 1.224667412656969e-07,
      "loss": 0.7067,
      "step": 23539
    },
    {
      "epoch": 0.9755895395582079,
      "grad_norm": 0.40684938430786133,
      "learning_rate": 1.2225952173732854e-07,
      "loss": 0.7056,
      "step": 23540
    },
    {
      "epoch": 0.9756309834638817,
      "grad_norm": 0.41750404238700867,
      "learning_rate": 1.220523022089602e-07,
      "loss": 0.6707,
      "step": 23541
    },
    {
      "epoch": 0.9756724273695553,
      "grad_norm": 0.42446863651275635,
      "learning_rate": 1.2184508268059184e-07,
      "loss": 0.6744,
      "step": 23542
    },
    {
      "epoch": 0.975713871275229,
      "grad_norm": 0.4019761383533478,
      "learning_rate": 1.2163786315222347e-07,
      "loss": 0.6672,
      "step": 23543
    },
    {
      "epoch": 0.9757553151809026,
      "grad_norm": 0.41590309143066406,
      "learning_rate": 1.2143064362385512e-07,
      "loss": 0.6265,
      "step": 23544
    },
    {
      "epoch": 0.9757967590865764,
      "grad_norm": 0.4452741742134094,
      "learning_rate": 1.2122342409548677e-07,
      "loss": 0.6348,
      "step": 23545
    },
    {
      "epoch": 0.97583820299225,
      "grad_norm": 0.42487606406211853,
      "learning_rate": 1.2101620456711842e-07,
      "loss": 0.663,
      "step": 23546
    },
    {
      "epoch": 0.9758796468979236,
      "grad_norm": 0.43896546959877014,
      "learning_rate": 1.2080898503875007e-07,
      "loss": 0.6729,
      "step": 23547
    },
    {
      "epoch": 0.9759210908035973,
      "grad_norm": 0.40278294682502747,
      "learning_rate": 1.2060176551038172e-07,
      "loss": 0.6895,
      "step": 23548
    },
    {
      "epoch": 0.975962534709271,
      "grad_norm": 0.41020047664642334,
      "learning_rate": 1.2039454598201334e-07,
      "loss": 0.6653,
      "step": 23549
    },
    {
      "epoch": 0.9760039786149447,
      "grad_norm": 0.4382156729698181,
      "learning_rate": 1.20187326453645e-07,
      "loss": 0.7051,
      "step": 23550
    },
    {
      "epoch": 0.9760454225206183,
      "grad_norm": 0.38714325428009033,
      "learning_rate": 1.1998010692527664e-07,
      "loss": 0.6046,
      "step": 23551
    },
    {
      "epoch": 0.976086866426292,
      "grad_norm": 0.4153909385204315,
      "learning_rate": 1.197728873969083e-07,
      "loss": 0.6732,
      "step": 23552
    },
    {
      "epoch": 0.9761283103319657,
      "grad_norm": 0.45773690938949585,
      "learning_rate": 1.1956566786853994e-07,
      "loss": 0.7416,
      "step": 23553
    },
    {
      "epoch": 0.9761697542376394,
      "grad_norm": 0.429177850484848,
      "learning_rate": 1.193584483401716e-07,
      "loss": 0.6677,
      "step": 23554
    },
    {
      "epoch": 0.976211198143313,
      "grad_norm": 0.45945897698402405,
      "learning_rate": 1.1915122881180323e-07,
      "loss": 0.6868,
      "step": 23555
    },
    {
      "epoch": 0.9762526420489867,
      "grad_norm": 0.41973116993904114,
      "learning_rate": 1.1894400928343488e-07,
      "loss": 0.6425,
      "step": 23556
    },
    {
      "epoch": 0.9762940859546604,
      "grad_norm": 0.40328025817871094,
      "learning_rate": 1.1873678975506652e-07,
      "loss": 0.6141,
      "step": 23557
    },
    {
      "epoch": 0.976335529860334,
      "grad_norm": 0.4115675985813141,
      "learning_rate": 1.1852957022669817e-07,
      "loss": 0.6982,
      "step": 23558
    },
    {
      "epoch": 0.9763769737660077,
      "grad_norm": 0.437727689743042,
      "learning_rate": 1.1832235069832982e-07,
      "loss": 0.6696,
      "step": 23559
    },
    {
      "epoch": 0.9764184176716814,
      "grad_norm": 0.43778878450393677,
      "learning_rate": 1.1811513116996147e-07,
      "loss": 0.6683,
      "step": 23560
    },
    {
      "epoch": 0.9764598615773551,
      "grad_norm": 0.4059181809425354,
      "learning_rate": 1.1790791164159311e-07,
      "loss": 0.6321,
      "step": 23561
    },
    {
      "epoch": 0.9765013054830287,
      "grad_norm": 0.4220151901245117,
      "learning_rate": 1.1770069211322476e-07,
      "loss": 0.6864,
      "step": 23562
    },
    {
      "epoch": 0.9765427493887024,
      "grad_norm": 0.43281692266464233,
      "learning_rate": 1.1749347258485641e-07,
      "loss": 0.6511,
      "step": 23563
    },
    {
      "epoch": 0.9765841932943761,
      "grad_norm": 0.4162597060203552,
      "learning_rate": 1.1728625305648806e-07,
      "loss": 0.6678,
      "step": 23564
    },
    {
      "epoch": 0.9766256372000497,
      "grad_norm": 0.45709019899368286,
      "learning_rate": 1.1707903352811971e-07,
      "loss": 0.7185,
      "step": 23565
    },
    {
      "epoch": 0.9766670811057234,
      "grad_norm": 0.40548428893089294,
      "learning_rate": 1.1687181399975135e-07,
      "loss": 0.6986,
      "step": 23566
    },
    {
      "epoch": 0.976708525011397,
      "grad_norm": 0.4265049993991852,
      "learning_rate": 1.16664594471383e-07,
      "loss": 0.6658,
      "step": 23567
    },
    {
      "epoch": 0.9767499689170708,
      "grad_norm": 0.3892822265625,
      "learning_rate": 1.1645737494301463e-07,
      "loss": 0.6349,
      "step": 23568
    },
    {
      "epoch": 0.9767914128227444,
      "grad_norm": 0.4082087576389313,
      "learning_rate": 1.1625015541464628e-07,
      "loss": 0.6906,
      "step": 23569
    },
    {
      "epoch": 0.9768328567284181,
      "grad_norm": 0.4802243113517761,
      "learning_rate": 1.1604293588627793e-07,
      "loss": 0.7235,
      "step": 23570
    },
    {
      "epoch": 0.9768743006340918,
      "grad_norm": 0.40916332602500916,
      "learning_rate": 1.1583571635790958e-07,
      "loss": 0.6626,
      "step": 23571
    },
    {
      "epoch": 0.9769157445397654,
      "grad_norm": 0.4045411944389343,
      "learning_rate": 1.1562849682954123e-07,
      "loss": 0.6022,
      "step": 23572
    },
    {
      "epoch": 0.9769571884454391,
      "grad_norm": 0.39164069294929504,
      "learning_rate": 1.1542127730117288e-07,
      "loss": 0.7053,
      "step": 23573
    },
    {
      "epoch": 0.9769986323511127,
      "grad_norm": 0.4402811825275421,
      "learning_rate": 1.1521405777280451e-07,
      "loss": 0.61,
      "step": 23574
    },
    {
      "epoch": 0.9770400762567865,
      "grad_norm": 0.42288073897361755,
      "learning_rate": 1.1500683824443616e-07,
      "loss": 0.6365,
      "step": 23575
    },
    {
      "epoch": 0.9770815201624601,
      "grad_norm": 0.38923558592796326,
      "learning_rate": 1.1479961871606781e-07,
      "loss": 0.6185,
      "step": 23576
    },
    {
      "epoch": 0.9771229640681338,
      "grad_norm": 0.39798402786254883,
      "learning_rate": 1.1459239918769946e-07,
      "loss": 0.6179,
      "step": 23577
    },
    {
      "epoch": 0.9771644079738074,
      "grad_norm": 0.39141547679901123,
      "learning_rate": 1.1438517965933111e-07,
      "loss": 0.6877,
      "step": 23578
    },
    {
      "epoch": 0.9772058518794812,
      "grad_norm": 0.41581299901008606,
      "learning_rate": 1.1417796013096276e-07,
      "loss": 0.6588,
      "step": 23579
    },
    {
      "epoch": 0.9772472957851548,
      "grad_norm": 0.41499218344688416,
      "learning_rate": 1.1397074060259438e-07,
      "loss": 0.6698,
      "step": 23580
    },
    {
      "epoch": 0.9772887396908284,
      "grad_norm": 0.41559165716171265,
      "learning_rate": 1.1376352107422603e-07,
      "loss": 0.6852,
      "step": 23581
    },
    {
      "epoch": 0.9773301835965021,
      "grad_norm": 0.4036775231361389,
      "learning_rate": 1.1355630154585769e-07,
      "loss": 0.6685,
      "step": 23582
    },
    {
      "epoch": 0.9773716275021758,
      "grad_norm": 0.4006056487560272,
      "learning_rate": 1.1334908201748934e-07,
      "loss": 0.688,
      "step": 23583
    },
    {
      "epoch": 0.9774130714078495,
      "grad_norm": 0.392010897397995,
      "learning_rate": 1.1314186248912099e-07,
      "loss": 0.6536,
      "step": 23584
    },
    {
      "epoch": 0.9774545153135231,
      "grad_norm": 0.41725683212280273,
      "learning_rate": 1.1293464296075264e-07,
      "loss": 0.7119,
      "step": 23585
    },
    {
      "epoch": 0.9774959592191969,
      "grad_norm": 0.42873620986938477,
      "learning_rate": 1.1272742343238427e-07,
      "loss": 0.6523,
      "step": 23586
    },
    {
      "epoch": 0.9775374031248705,
      "grad_norm": 0.45240551233291626,
      "learning_rate": 1.1252020390401592e-07,
      "loss": 0.7089,
      "step": 23587
    },
    {
      "epoch": 0.9775788470305442,
      "grad_norm": 0.3923335671424866,
      "learning_rate": 1.1231298437564757e-07,
      "loss": 0.626,
      "step": 23588
    },
    {
      "epoch": 0.9776202909362178,
      "grad_norm": 0.40390488505363464,
      "learning_rate": 1.1210576484727922e-07,
      "loss": 0.6865,
      "step": 23589
    },
    {
      "epoch": 0.9776617348418914,
      "grad_norm": 0.3931303918361664,
      "learning_rate": 1.1189854531891086e-07,
      "loss": 0.6265,
      "step": 23590
    },
    {
      "epoch": 0.9777031787475652,
      "grad_norm": 0.4041198790073395,
      "learning_rate": 1.1169132579054251e-07,
      "loss": 0.6311,
      "step": 23591
    },
    {
      "epoch": 0.9777446226532388,
      "grad_norm": 0.42545369267463684,
      "learning_rate": 1.1148410626217415e-07,
      "loss": 0.6837,
      "step": 23592
    },
    {
      "epoch": 0.9777860665589125,
      "grad_norm": 0.41677185893058777,
      "learning_rate": 1.112768867338058e-07,
      "loss": 0.6619,
      "step": 23593
    },
    {
      "epoch": 0.9778275104645862,
      "grad_norm": 0.5093568563461304,
      "learning_rate": 1.1106966720543745e-07,
      "loss": 0.6908,
      "step": 23594
    },
    {
      "epoch": 0.9778689543702599,
      "grad_norm": 0.4546039402484894,
      "learning_rate": 1.108624476770691e-07,
      "loss": 0.7075,
      "step": 23595
    },
    {
      "epoch": 0.9779103982759335,
      "grad_norm": 0.39054110646247864,
      "learning_rate": 1.1065522814870075e-07,
      "loss": 0.6074,
      "step": 23596
    },
    {
      "epoch": 0.9779518421816072,
      "grad_norm": 0.4467930793762207,
      "learning_rate": 1.104480086203324e-07,
      "loss": 0.6509,
      "step": 23597
    },
    {
      "epoch": 0.9779932860872809,
      "grad_norm": 0.39826902747154236,
      "learning_rate": 1.1024078909196402e-07,
      "loss": 0.6677,
      "step": 23598
    },
    {
      "epoch": 0.9780347299929545,
      "grad_norm": 0.42044970393180847,
      "learning_rate": 1.1003356956359567e-07,
      "loss": 0.6748,
      "step": 23599
    },
    {
      "epoch": 0.9780761738986282,
      "grad_norm": 0.40181124210357666,
      "learning_rate": 1.0982635003522733e-07,
      "loss": 0.7295,
      "step": 23600
    },
    {
      "epoch": 0.9781176178043018,
      "grad_norm": 0.4282439053058624,
      "learning_rate": 1.0961913050685898e-07,
      "loss": 0.6953,
      "step": 23601
    },
    {
      "epoch": 0.9781590617099756,
      "grad_norm": 0.3857939541339874,
      "learning_rate": 1.0941191097849063e-07,
      "loss": 0.5785,
      "step": 23602
    },
    {
      "epoch": 0.9782005056156492,
      "grad_norm": 0.4286782741546631,
      "learning_rate": 1.0920469145012228e-07,
      "loss": 0.6606,
      "step": 23603
    },
    {
      "epoch": 0.9782419495213229,
      "grad_norm": 0.43351104855537415,
      "learning_rate": 1.089974719217539e-07,
      "loss": 0.6189,
      "step": 23604
    },
    {
      "epoch": 0.9782833934269966,
      "grad_norm": 0.4434877932071686,
      "learning_rate": 1.0879025239338555e-07,
      "loss": 0.6821,
      "step": 23605
    },
    {
      "epoch": 0.9783248373326703,
      "grad_norm": 0.410850465297699,
      "learning_rate": 1.085830328650172e-07,
      "loss": 0.6222,
      "step": 23606
    },
    {
      "epoch": 0.9783662812383439,
      "grad_norm": 0.41938742995262146,
      "learning_rate": 1.0837581333664885e-07,
      "loss": 0.7091,
      "step": 23607
    },
    {
      "epoch": 0.9784077251440175,
      "grad_norm": 0.4288516640663147,
      "learning_rate": 1.081685938082805e-07,
      "loss": 0.6318,
      "step": 23608
    },
    {
      "epoch": 0.9784491690496913,
      "grad_norm": 0.41946277022361755,
      "learning_rate": 1.0796137427991215e-07,
      "loss": 0.6326,
      "step": 23609
    },
    {
      "epoch": 0.9784906129553649,
      "grad_norm": 0.5122292041778564,
      "learning_rate": 1.077541547515438e-07,
      "loss": 0.6932,
      "step": 23610
    },
    {
      "epoch": 0.9785320568610386,
      "grad_norm": 0.3869394063949585,
      "learning_rate": 1.0754693522317544e-07,
      "loss": 0.6476,
      "step": 23611
    },
    {
      "epoch": 0.9785735007667122,
      "grad_norm": 0.42109549045562744,
      "learning_rate": 1.0733971569480709e-07,
      "loss": 0.656,
      "step": 23612
    },
    {
      "epoch": 0.978614944672386,
      "grad_norm": 0.4887112081050873,
      "learning_rate": 1.0713249616643873e-07,
      "loss": 0.7107,
      "step": 23613
    },
    {
      "epoch": 0.9786563885780596,
      "grad_norm": 0.41121748089790344,
      "learning_rate": 1.0692527663807038e-07,
      "loss": 0.6124,
      "step": 23614
    },
    {
      "epoch": 0.9786978324837333,
      "grad_norm": 0.41716399788856506,
      "learning_rate": 1.0671805710970203e-07,
      "loss": 0.6892,
      "step": 23615
    },
    {
      "epoch": 0.9787392763894069,
      "grad_norm": 0.42540186643600464,
      "learning_rate": 1.0651083758133368e-07,
      "loss": 0.6877,
      "step": 23616
    },
    {
      "epoch": 0.9787807202950806,
      "grad_norm": 0.4231995642185211,
      "learning_rate": 1.0630361805296532e-07,
      "loss": 0.6741,
      "step": 23617
    },
    {
      "epoch": 0.9788221642007543,
      "grad_norm": 0.44637176394462585,
      "learning_rate": 1.0609639852459697e-07,
      "loss": 0.686,
      "step": 23618
    },
    {
      "epoch": 0.9788636081064279,
      "grad_norm": 0.41902029514312744,
      "learning_rate": 1.0588917899622862e-07,
      "loss": 0.6324,
      "step": 23619
    },
    {
      "epoch": 0.9789050520121017,
      "grad_norm": 0.4379342794418335,
      "learning_rate": 1.0568195946786027e-07,
      "loss": 0.6943,
      "step": 23620
    },
    {
      "epoch": 0.9789464959177753,
      "grad_norm": 0.4473947286605835,
      "learning_rate": 1.0547473993949192e-07,
      "loss": 0.66,
      "step": 23621
    },
    {
      "epoch": 0.978987939823449,
      "grad_norm": 0.431602418422699,
      "learning_rate": 1.0526752041112357e-07,
      "loss": 0.6853,
      "step": 23622
    },
    {
      "epoch": 0.9790293837291226,
      "grad_norm": 0.4170432984828949,
      "learning_rate": 1.0506030088275519e-07,
      "loss": 0.6508,
      "step": 23623
    },
    {
      "epoch": 0.9790708276347962,
      "grad_norm": 0.4146314561367035,
      "learning_rate": 1.0485308135438684e-07,
      "loss": 0.6431,
      "step": 23624
    },
    {
      "epoch": 0.97911227154047,
      "grad_norm": 0.40374454855918884,
      "learning_rate": 1.0464586182601849e-07,
      "loss": 0.6909,
      "step": 23625
    },
    {
      "epoch": 0.9791537154461436,
      "grad_norm": 0.386064738035202,
      "learning_rate": 1.0443864229765014e-07,
      "loss": 0.6379,
      "step": 23626
    },
    {
      "epoch": 0.9791951593518173,
      "grad_norm": 0.4148416519165039,
      "learning_rate": 1.0423142276928179e-07,
      "loss": 0.6494,
      "step": 23627
    },
    {
      "epoch": 0.979236603257491,
      "grad_norm": 0.4147054851055145,
      "learning_rate": 1.0402420324091344e-07,
      "loss": 0.6641,
      "step": 23628
    },
    {
      "epoch": 0.9792780471631647,
      "grad_norm": 0.41580909490585327,
      "learning_rate": 1.0381698371254507e-07,
      "loss": 0.6527,
      "step": 23629
    },
    {
      "epoch": 0.9793194910688383,
      "grad_norm": 0.4065316319465637,
      "learning_rate": 1.0360976418417672e-07,
      "loss": 0.6545,
      "step": 23630
    },
    {
      "epoch": 0.979360934974512,
      "grad_norm": 0.4223558306694031,
      "learning_rate": 1.0340254465580837e-07,
      "loss": 0.7079,
      "step": 23631
    },
    {
      "epoch": 0.9794023788801857,
      "grad_norm": 0.4026843309402466,
      "learning_rate": 1.0319532512744002e-07,
      "loss": 0.6715,
      "step": 23632
    },
    {
      "epoch": 0.9794438227858593,
      "grad_norm": 0.42909160256385803,
      "learning_rate": 1.0298810559907167e-07,
      "loss": 0.6948,
      "step": 23633
    },
    {
      "epoch": 0.979485266691533,
      "grad_norm": 0.42372986674308777,
      "learning_rate": 1.0278088607070332e-07,
      "loss": 0.6466,
      "step": 23634
    },
    {
      "epoch": 0.9795267105972066,
      "grad_norm": 0.4231073558330536,
      "learning_rate": 1.0257366654233496e-07,
      "loss": 0.6748,
      "step": 23635
    },
    {
      "epoch": 0.9795681545028804,
      "grad_norm": 0.42625677585601807,
      "learning_rate": 1.023664470139666e-07,
      "loss": 0.6216,
      "step": 23636
    },
    {
      "epoch": 0.979609598408554,
      "grad_norm": 0.41383159160614014,
      "learning_rate": 1.0215922748559824e-07,
      "loss": 0.6427,
      "step": 23637
    },
    {
      "epoch": 0.9796510423142277,
      "grad_norm": 0.40653562545776367,
      "learning_rate": 1.0195200795722989e-07,
      "loss": 0.6448,
      "step": 23638
    },
    {
      "epoch": 0.9796924862199013,
      "grad_norm": 0.40527260303497314,
      "learning_rate": 1.0174478842886154e-07,
      "loss": 0.6644,
      "step": 23639
    },
    {
      "epoch": 0.9797339301255751,
      "grad_norm": 0.43669235706329346,
      "learning_rate": 1.015375689004932e-07,
      "loss": 0.6689,
      "step": 23640
    },
    {
      "epoch": 0.9797753740312487,
      "grad_norm": 0.44632866978645325,
      "learning_rate": 1.0133034937212483e-07,
      "loss": 0.6846,
      "step": 23641
    },
    {
      "epoch": 0.9798168179369223,
      "grad_norm": 0.4209052622318268,
      "learning_rate": 1.0112312984375648e-07,
      "loss": 0.6644,
      "step": 23642
    },
    {
      "epoch": 0.9798582618425961,
      "grad_norm": 0.4505915641784668,
      "learning_rate": 1.0091591031538813e-07,
      "loss": 0.6555,
      "step": 23643
    },
    {
      "epoch": 0.9798997057482697,
      "grad_norm": 0.42974236607551575,
      "learning_rate": 1.0070869078701978e-07,
      "loss": 0.7119,
      "step": 23644
    },
    {
      "epoch": 0.9799411496539434,
      "grad_norm": 0.42199140787124634,
      "learning_rate": 1.0050147125865143e-07,
      "loss": 0.6677,
      "step": 23645
    },
    {
      "epoch": 0.979982593559617,
      "grad_norm": 0.39782482385635376,
      "learning_rate": 1.0029425173028307e-07,
      "loss": 0.6418,
      "step": 23646
    },
    {
      "epoch": 0.9800240374652908,
      "grad_norm": 0.4724113345146179,
      "learning_rate": 1.0008703220191472e-07,
      "loss": 0.6746,
      "step": 23647
    },
    {
      "epoch": 0.9800654813709644,
      "grad_norm": 0.4095297157764435,
      "learning_rate": 9.987981267354636e-08,
      "loss": 0.6406,
      "step": 23648
    },
    {
      "epoch": 0.9801069252766381,
      "grad_norm": 0.5389772057533264,
      "learning_rate": 9.967259314517801e-08,
      "loss": 0.6731,
      "step": 23649
    },
    {
      "epoch": 0.9801483691823117,
      "grad_norm": 0.42420363426208496,
      "learning_rate": 9.946537361680966e-08,
      "loss": 0.7046,
      "step": 23650
    },
    {
      "epoch": 0.9801898130879854,
      "grad_norm": 0.43187591433525085,
      "learning_rate": 9.925815408844131e-08,
      "loss": 0.6914,
      "step": 23651
    },
    {
      "epoch": 0.9802312569936591,
      "grad_norm": 0.4129951298236847,
      "learning_rate": 9.905093456007296e-08,
      "loss": 0.6399,
      "step": 23652
    },
    {
      "epoch": 0.9802727008993327,
      "grad_norm": 0.4367525279521942,
      "learning_rate": 9.884371503170461e-08,
      "loss": 0.6481,
      "step": 23653
    },
    {
      "epoch": 0.9803141448050064,
      "grad_norm": 0.39493945240974426,
      "learning_rate": 9.863649550333623e-08,
      "loss": 0.6892,
      "step": 23654
    },
    {
      "epoch": 0.9803555887106801,
      "grad_norm": 0.48866474628448486,
      "learning_rate": 9.842927597496788e-08,
      "loss": 0.7043,
      "step": 23655
    },
    {
      "epoch": 0.9803970326163538,
      "grad_norm": 0.4350723922252655,
      "learning_rate": 9.822205644659953e-08,
      "loss": 0.6879,
      "step": 23656
    },
    {
      "epoch": 0.9804384765220274,
      "grad_norm": 0.39586225152015686,
      "learning_rate": 9.801483691823118e-08,
      "loss": 0.6335,
      "step": 23657
    },
    {
      "epoch": 0.9804799204277012,
      "grad_norm": 0.43086689710617065,
      "learning_rate": 9.780761738986283e-08,
      "loss": 0.656,
      "step": 23658
    },
    {
      "epoch": 0.9805213643333748,
      "grad_norm": 0.4141995310783386,
      "learning_rate": 9.760039786149448e-08,
      "loss": 0.6152,
      "step": 23659
    },
    {
      "epoch": 0.9805628082390484,
      "grad_norm": 0.4413281977176666,
      "learning_rate": 9.739317833312612e-08,
      "loss": 0.6768,
      "step": 23660
    },
    {
      "epoch": 0.9806042521447221,
      "grad_norm": 0.39343810081481934,
      "learning_rate": 9.718595880475776e-08,
      "loss": 0.672,
      "step": 23661
    },
    {
      "epoch": 0.9806456960503958,
      "grad_norm": 0.3849329650402069,
      "learning_rate": 9.697873927638941e-08,
      "loss": 0.6343,
      "step": 23662
    },
    {
      "epoch": 0.9806871399560695,
      "grad_norm": 0.4134483337402344,
      "learning_rate": 9.677151974802106e-08,
      "loss": 0.6394,
      "step": 23663
    },
    {
      "epoch": 0.9807285838617431,
      "grad_norm": 0.4229297935962677,
      "learning_rate": 9.656430021965271e-08,
      "loss": 0.6985,
      "step": 23664
    },
    {
      "epoch": 0.9807700277674168,
      "grad_norm": 0.4560183584690094,
      "learning_rate": 9.635708069128436e-08,
      "loss": 0.6927,
      "step": 23665
    },
    {
      "epoch": 0.9808114716730905,
      "grad_norm": 0.40779420733451843,
      "learning_rate": 9.6149861162916e-08,
      "loss": 0.6293,
      "step": 23666
    },
    {
      "epoch": 0.9808529155787642,
      "grad_norm": 0.3912232518196106,
      "learning_rate": 9.594264163454765e-08,
      "loss": 0.6101,
      "step": 23667
    },
    {
      "epoch": 0.9808943594844378,
      "grad_norm": 0.46027958393096924,
      "learning_rate": 9.57354221061793e-08,
      "loss": 0.6963,
      "step": 23668
    },
    {
      "epoch": 0.9809358033901114,
      "grad_norm": 0.3983546197414398,
      "learning_rate": 9.552820257781095e-08,
      "loss": 0.6353,
      "step": 23669
    },
    {
      "epoch": 0.9809772472957852,
      "grad_norm": 0.44635459780693054,
      "learning_rate": 9.532098304944259e-08,
      "loss": 0.6592,
      "step": 23670
    },
    {
      "epoch": 0.9810186912014588,
      "grad_norm": 0.4508945643901825,
      "learning_rate": 9.511376352107424e-08,
      "loss": 0.6736,
      "step": 23671
    },
    {
      "epoch": 0.9810601351071325,
      "grad_norm": 0.38556066155433655,
      "learning_rate": 9.490654399270587e-08,
      "loss": 0.6107,
      "step": 23672
    },
    {
      "epoch": 0.9811015790128061,
      "grad_norm": 0.45048484206199646,
      "learning_rate": 9.469932446433752e-08,
      "loss": 0.6971,
      "step": 23673
    },
    {
      "epoch": 0.9811430229184799,
      "grad_norm": 0.44595471024513245,
      "learning_rate": 9.449210493596917e-08,
      "loss": 0.7084,
      "step": 23674
    },
    {
      "epoch": 0.9811844668241535,
      "grad_norm": 0.41767561435699463,
      "learning_rate": 9.428488540760082e-08,
      "loss": 0.6799,
      "step": 23675
    },
    {
      "epoch": 0.9812259107298272,
      "grad_norm": 0.4130467176437378,
      "learning_rate": 9.407766587923247e-08,
      "loss": 0.6689,
      "step": 23676
    },
    {
      "epoch": 0.9812673546355009,
      "grad_norm": 0.396087646484375,
      "learning_rate": 9.387044635086412e-08,
      "loss": 0.6598,
      "step": 23677
    },
    {
      "epoch": 0.9813087985411745,
      "grad_norm": 0.45541685819625854,
      "learning_rate": 9.366322682249575e-08,
      "loss": 0.6896,
      "step": 23678
    },
    {
      "epoch": 0.9813502424468482,
      "grad_norm": 0.5333490967750549,
      "learning_rate": 9.34560072941274e-08,
      "loss": 0.6093,
      "step": 23679
    },
    {
      "epoch": 0.9813916863525218,
      "grad_norm": 0.43766605854034424,
      "learning_rate": 9.324878776575905e-08,
      "loss": 0.6802,
      "step": 23680
    },
    {
      "epoch": 0.9814331302581956,
      "grad_norm": 0.45895758271217346,
      "learning_rate": 9.30415682373907e-08,
      "loss": 0.6189,
      "step": 23681
    },
    {
      "epoch": 0.9814745741638692,
      "grad_norm": 0.3990688621997833,
      "learning_rate": 9.283434870902235e-08,
      "loss": 0.6887,
      "step": 23682
    },
    {
      "epoch": 0.9815160180695429,
      "grad_norm": 0.4337676167488098,
      "learning_rate": 9.2627129180654e-08,
      "loss": 0.6653,
      "step": 23683
    },
    {
      "epoch": 0.9815574619752165,
      "grad_norm": 0.42258375883102417,
      "learning_rate": 9.241990965228562e-08,
      "loss": 0.6748,
      "step": 23684
    },
    {
      "epoch": 0.9815989058808902,
      "grad_norm": 0.3940608501434326,
      "learning_rate": 9.221269012391727e-08,
      "loss": 0.6575,
      "step": 23685
    },
    {
      "epoch": 0.9816403497865639,
      "grad_norm": 0.4222395122051239,
      "learning_rate": 9.200547059554892e-08,
      "loss": 0.6681,
      "step": 23686
    },
    {
      "epoch": 0.9816817936922375,
      "grad_norm": 0.42154744267463684,
      "learning_rate": 9.179825106718058e-08,
      "loss": 0.6351,
      "step": 23687
    },
    {
      "epoch": 0.9817232375979112,
      "grad_norm": 0.42230817675590515,
      "learning_rate": 9.159103153881223e-08,
      "loss": 0.7095,
      "step": 23688
    },
    {
      "epoch": 0.9817646815035849,
      "grad_norm": 0.4477296471595764,
      "learning_rate": 9.138381201044388e-08,
      "loss": 0.6969,
      "step": 23689
    },
    {
      "epoch": 0.9818061254092586,
      "grad_norm": 0.415199875831604,
      "learning_rate": 9.117659248207553e-08,
      "loss": 0.679,
      "step": 23690
    },
    {
      "epoch": 0.9818475693149322,
      "grad_norm": 0.39856746792793274,
      "learning_rate": 9.096937295370716e-08,
      "loss": 0.6671,
      "step": 23691
    },
    {
      "epoch": 0.981889013220606,
      "grad_norm": 0.4418445825576782,
      "learning_rate": 9.076215342533881e-08,
      "loss": 0.6207,
      "step": 23692
    },
    {
      "epoch": 0.9819304571262796,
      "grad_norm": 0.38116884231567383,
      "learning_rate": 9.055493389697046e-08,
      "loss": 0.6272,
      "step": 23693
    },
    {
      "epoch": 0.9819719010319532,
      "grad_norm": 0.37959975004196167,
      "learning_rate": 9.03477143686021e-08,
      "loss": 0.6215,
      "step": 23694
    },
    {
      "epoch": 0.9820133449376269,
      "grad_norm": 0.41990989446640015,
      "learning_rate": 9.014049484023375e-08,
      "loss": 0.717,
      "step": 23695
    },
    {
      "epoch": 0.9820547888433006,
      "grad_norm": 0.43460017442703247,
      "learning_rate": 8.99332753118654e-08,
      "loss": 0.6926,
      "step": 23696
    },
    {
      "epoch": 0.9820962327489743,
      "grad_norm": 0.46710285544395447,
      "learning_rate": 8.972605578349704e-08,
      "loss": 0.7186,
      "step": 23697
    },
    {
      "epoch": 0.9821376766546479,
      "grad_norm": 0.4233827590942383,
      "learning_rate": 8.951883625512869e-08,
      "loss": 0.5942,
      "step": 23698
    },
    {
      "epoch": 0.9821791205603216,
      "grad_norm": 0.3891182541847229,
      "learning_rate": 8.931161672676034e-08,
      "loss": 0.6486,
      "step": 23699
    },
    {
      "epoch": 0.9822205644659953,
      "grad_norm": 0.41781291365623474,
      "learning_rate": 8.910439719839199e-08,
      "loss": 0.7029,
      "step": 23700
    },
    {
      "epoch": 0.982262008371669,
      "grad_norm": 0.3945339024066925,
      "learning_rate": 8.889717767002364e-08,
      "loss": 0.6299,
      "step": 23701
    },
    {
      "epoch": 0.9823034522773426,
      "grad_norm": 0.3898824453353882,
      "learning_rate": 8.868995814165529e-08,
      "loss": 0.6118,
      "step": 23702
    },
    {
      "epoch": 0.9823448961830162,
      "grad_norm": 0.437576562166214,
      "learning_rate": 8.848273861328691e-08,
      "loss": 0.6704,
      "step": 23703
    },
    {
      "epoch": 0.98238634008869,
      "grad_norm": 0.42995303869247437,
      "learning_rate": 8.827551908491856e-08,
      "loss": 0.6552,
      "step": 23704
    },
    {
      "epoch": 0.9824277839943636,
      "grad_norm": 0.43487101793289185,
      "learning_rate": 8.806829955655022e-08,
      "loss": 0.7251,
      "step": 23705
    },
    {
      "epoch": 0.9824692279000373,
      "grad_norm": 0.42375245690345764,
      "learning_rate": 8.786108002818187e-08,
      "loss": 0.6647,
      "step": 23706
    },
    {
      "epoch": 0.982510671805711,
      "grad_norm": 0.40663692355155945,
      "learning_rate": 8.765386049981352e-08,
      "loss": 0.6224,
      "step": 23707
    },
    {
      "epoch": 0.9825521157113847,
      "grad_norm": 0.3912353515625,
      "learning_rate": 8.744664097144517e-08,
      "loss": 0.6328,
      "step": 23708
    },
    {
      "epoch": 0.9825935596170583,
      "grad_norm": 0.41251981258392334,
      "learning_rate": 8.723942144307679e-08,
      "loss": 0.6736,
      "step": 23709
    },
    {
      "epoch": 0.982635003522732,
      "grad_norm": 0.44015276432037354,
      "learning_rate": 8.703220191470844e-08,
      "loss": 0.7104,
      "step": 23710
    },
    {
      "epoch": 0.9826764474284057,
      "grad_norm": 0.3964208960533142,
      "learning_rate": 8.682498238634009e-08,
      "loss": 0.645,
      "step": 23711
    },
    {
      "epoch": 0.9827178913340793,
      "grad_norm": 0.4139605760574341,
      "learning_rate": 8.661776285797174e-08,
      "loss": 0.7083,
      "step": 23712
    },
    {
      "epoch": 0.982759335239753,
      "grad_norm": 0.40190422534942627,
      "learning_rate": 8.641054332960339e-08,
      "loss": 0.6783,
      "step": 23713
    },
    {
      "epoch": 0.9828007791454266,
      "grad_norm": 0.44008633494377136,
      "learning_rate": 8.620332380123504e-08,
      "loss": 0.6918,
      "step": 23714
    },
    {
      "epoch": 0.9828422230511004,
      "grad_norm": 0.452303946018219,
      "learning_rate": 8.599610427286668e-08,
      "loss": 0.6985,
      "step": 23715
    },
    {
      "epoch": 0.982883666956774,
      "grad_norm": 0.43062764406204224,
      "learning_rate": 8.578888474449833e-08,
      "loss": 0.6764,
      "step": 23716
    },
    {
      "epoch": 0.9829251108624477,
      "grad_norm": 0.39649298787117004,
      "learning_rate": 8.558166521612997e-08,
      "loss": 0.6416,
      "step": 23717
    },
    {
      "epoch": 0.9829665547681213,
      "grad_norm": 0.3983179032802582,
      "learning_rate": 8.537444568776162e-08,
      "loss": 0.6473,
      "step": 23718
    },
    {
      "epoch": 0.9830079986737951,
      "grad_norm": 0.45764994621276855,
      "learning_rate": 8.516722615939327e-08,
      "loss": 0.6895,
      "step": 23719
    },
    {
      "epoch": 0.9830494425794687,
      "grad_norm": 0.44658827781677246,
      "learning_rate": 8.496000663102492e-08,
      "loss": 0.6637,
      "step": 23720
    },
    {
      "epoch": 0.9830908864851423,
      "grad_norm": 0.4045303463935852,
      "learning_rate": 8.475278710265655e-08,
      "loss": 0.6318,
      "step": 23721
    },
    {
      "epoch": 0.983132330390816,
      "grad_norm": 0.436860591173172,
      "learning_rate": 8.45455675742882e-08,
      "loss": 0.676,
      "step": 23722
    },
    {
      "epoch": 0.9831737742964897,
      "grad_norm": 0.4122665524482727,
      "learning_rate": 8.433834804591986e-08,
      "loss": 0.672,
      "step": 23723
    },
    {
      "epoch": 0.9832152182021634,
      "grad_norm": 0.42395463585853577,
      "learning_rate": 8.41311285175515e-08,
      "loss": 0.7058,
      "step": 23724
    },
    {
      "epoch": 0.983256662107837,
      "grad_norm": 0.41648733615875244,
      "learning_rate": 8.392390898918316e-08,
      "loss": 0.6709,
      "step": 23725
    },
    {
      "epoch": 0.9832981060135108,
      "grad_norm": 0.4401293396949768,
      "learning_rate": 8.37166894608148e-08,
      "loss": 0.6324,
      "step": 23726
    },
    {
      "epoch": 0.9833395499191844,
      "grad_norm": 0.40695586800575256,
      "learning_rate": 8.350946993244644e-08,
      "loss": 0.6571,
      "step": 23727
    },
    {
      "epoch": 0.9833809938248581,
      "grad_norm": 0.4133570194244385,
      "learning_rate": 8.330225040407808e-08,
      "loss": 0.598,
      "step": 23728
    },
    {
      "epoch": 0.9834224377305317,
      "grad_norm": 0.4219667911529541,
      "learning_rate": 8.309503087570973e-08,
      "loss": 0.689,
      "step": 23729
    },
    {
      "epoch": 0.9834638816362054,
      "grad_norm": 0.4425598382949829,
      "learning_rate": 8.288781134734138e-08,
      "loss": 0.6871,
      "step": 23730
    },
    {
      "epoch": 0.9835053255418791,
      "grad_norm": 0.37690967321395874,
      "learning_rate": 8.268059181897303e-08,
      "loss": 0.6892,
      "step": 23731
    },
    {
      "epoch": 0.9835467694475527,
      "grad_norm": 0.43040478229522705,
      "learning_rate": 8.247337229060468e-08,
      "loss": 0.7042,
      "step": 23732
    },
    {
      "epoch": 0.9835882133532264,
      "grad_norm": 0.40907347202301025,
      "learning_rate": 8.226615276223633e-08,
      "loss": 0.6614,
      "step": 23733
    },
    {
      "epoch": 0.9836296572589001,
      "grad_norm": 0.4230988621711731,
      "learning_rate": 8.205893323386796e-08,
      "loss": 0.6285,
      "step": 23734
    },
    {
      "epoch": 0.9836711011645738,
      "grad_norm": 0.4477046728134155,
      "learning_rate": 8.185171370549961e-08,
      "loss": 0.6409,
      "step": 23735
    },
    {
      "epoch": 0.9837125450702474,
      "grad_norm": 0.46720707416534424,
      "learning_rate": 8.164449417713126e-08,
      "loss": 0.6227,
      "step": 23736
    },
    {
      "epoch": 0.9837539889759211,
      "grad_norm": 0.38927161693573,
      "learning_rate": 8.143727464876291e-08,
      "loss": 0.5819,
      "step": 23737
    },
    {
      "epoch": 0.9837954328815948,
      "grad_norm": 0.4213904142379761,
      "learning_rate": 8.123005512039456e-08,
      "loss": 0.668,
      "step": 23738
    },
    {
      "epoch": 0.9838368767872684,
      "grad_norm": 0.45237278938293457,
      "learning_rate": 8.102283559202621e-08,
      "loss": 0.7325,
      "step": 23739
    },
    {
      "epoch": 0.9838783206929421,
      "grad_norm": 0.3955310881137848,
      "learning_rate": 8.081561606365785e-08,
      "loss": 0.646,
      "step": 23740
    },
    {
      "epoch": 0.9839197645986157,
      "grad_norm": 0.4115259349346161,
      "learning_rate": 8.060839653528948e-08,
      "loss": 0.6987,
      "step": 23741
    },
    {
      "epoch": 0.9839612085042895,
      "grad_norm": 0.46165499091148376,
      "learning_rate": 8.040117700692113e-08,
      "loss": 0.6699,
      "step": 23742
    },
    {
      "epoch": 0.9840026524099631,
      "grad_norm": 0.43582576513290405,
      "learning_rate": 8.019395747855278e-08,
      "loss": 0.6699,
      "step": 23743
    },
    {
      "epoch": 0.9840440963156368,
      "grad_norm": 0.45321768522262573,
      "learning_rate": 7.998673795018443e-08,
      "loss": 0.6429,
      "step": 23744
    },
    {
      "epoch": 0.9840855402213105,
      "grad_norm": 0.4111030399799347,
      "learning_rate": 7.977951842181608e-08,
      "loss": 0.7114,
      "step": 23745
    },
    {
      "epoch": 0.9841269841269841,
      "grad_norm": 0.4149567186832428,
      "learning_rate": 7.957229889344772e-08,
      "loss": 0.6812,
      "step": 23746
    },
    {
      "epoch": 0.9841684280326578,
      "grad_norm": 0.42554032802581787,
      "learning_rate": 7.936507936507937e-08,
      "loss": 0.6606,
      "step": 23747
    },
    {
      "epoch": 0.9842098719383314,
      "grad_norm": 0.47971871495246887,
      "learning_rate": 7.915785983671102e-08,
      "loss": 0.7273,
      "step": 23748
    },
    {
      "epoch": 0.9842513158440052,
      "grad_norm": 0.46155983209609985,
      "learning_rate": 7.895064030834267e-08,
      "loss": 0.6792,
      "step": 23749
    },
    {
      "epoch": 0.9842927597496788,
      "grad_norm": 0.4305484890937805,
      "learning_rate": 7.874342077997431e-08,
      "loss": 0.6665,
      "step": 23750
    },
    {
      "epoch": 0.9843342036553525,
      "grad_norm": 0.47075098752975464,
      "learning_rate": 7.853620125160596e-08,
      "loss": 0.6619,
      "step": 23751
    },
    {
      "epoch": 0.9843756475610261,
      "grad_norm": 0.41983652114868164,
      "learning_rate": 7.83289817232376e-08,
      "loss": 0.6744,
      "step": 23752
    },
    {
      "epoch": 0.9844170914666999,
      "grad_norm": 0.41290122270584106,
      "learning_rate": 7.812176219486926e-08,
      "loss": 0.6736,
      "step": 23753
    },
    {
      "epoch": 0.9844585353723735,
      "grad_norm": 0.437527596950531,
      "learning_rate": 7.79145426665009e-08,
      "loss": 0.6599,
      "step": 23754
    },
    {
      "epoch": 0.9844999792780471,
      "grad_norm": 0.42158234119415283,
      "learning_rate": 7.770732313813255e-08,
      "loss": 0.6318,
      "step": 23755
    },
    {
      "epoch": 0.9845414231837208,
      "grad_norm": 0.43469369411468506,
      "learning_rate": 7.75001036097642e-08,
      "loss": 0.6401,
      "step": 23756
    },
    {
      "epoch": 0.9845828670893945,
      "grad_norm": 0.44089460372924805,
      "learning_rate": 7.729288408139583e-08,
      "loss": 0.7079,
      "step": 23757
    },
    {
      "epoch": 0.9846243109950682,
      "grad_norm": 0.45520657300949097,
      "learning_rate": 7.708566455302749e-08,
      "loss": 0.6923,
      "step": 23758
    },
    {
      "epoch": 0.9846657549007418,
      "grad_norm": 0.42664265632629395,
      "learning_rate": 7.687844502465914e-08,
      "loss": 0.631,
      "step": 23759
    },
    {
      "epoch": 0.9847071988064156,
      "grad_norm": 0.45176950097084045,
      "learning_rate": 7.667122549629077e-08,
      "loss": 0.7144,
      "step": 23760
    },
    {
      "epoch": 0.9847486427120892,
      "grad_norm": 0.4436737298965454,
      "learning_rate": 7.646400596792242e-08,
      "loss": 0.7117,
      "step": 23761
    },
    {
      "epoch": 0.9847900866177629,
      "grad_norm": 0.4233611524105072,
      "learning_rate": 7.625678643955407e-08,
      "loss": 0.6462,
      "step": 23762
    },
    {
      "epoch": 0.9848315305234365,
      "grad_norm": 0.40636399388313293,
      "learning_rate": 7.604956691118571e-08,
      "loss": 0.6628,
      "step": 23763
    },
    {
      "epoch": 0.9848729744291101,
      "grad_norm": 0.429129034280777,
      "learning_rate": 7.584234738281736e-08,
      "loss": 0.6664,
      "step": 23764
    },
    {
      "epoch": 0.9849144183347839,
      "grad_norm": 0.42212092876434326,
      "learning_rate": 7.563512785444901e-08,
      "loss": 0.6696,
      "step": 23765
    },
    {
      "epoch": 0.9849558622404575,
      "grad_norm": 0.44828087091445923,
      "learning_rate": 7.542790832608065e-08,
      "loss": 0.6832,
      "step": 23766
    },
    {
      "epoch": 0.9849973061461312,
      "grad_norm": 0.4451351761817932,
      "learning_rate": 7.52206887977123e-08,
      "loss": 0.6769,
      "step": 23767
    },
    {
      "epoch": 0.9850387500518049,
      "grad_norm": 0.42323923110961914,
      "learning_rate": 7.501346926934395e-08,
      "loss": 0.6631,
      "step": 23768
    },
    {
      "epoch": 0.9850801939574786,
      "grad_norm": 0.44780468940734863,
      "learning_rate": 7.480624974097559e-08,
      "loss": 0.6793,
      "step": 23769
    },
    {
      "epoch": 0.9851216378631522,
      "grad_norm": 0.4005405008792877,
      "learning_rate": 7.459903021260724e-08,
      "loss": 0.6614,
      "step": 23770
    },
    {
      "epoch": 0.9851630817688259,
      "grad_norm": 0.4254232347011566,
      "learning_rate": 7.439181068423889e-08,
      "loss": 0.6946,
      "step": 23771
    },
    {
      "epoch": 0.9852045256744996,
      "grad_norm": 0.4308984577655792,
      "learning_rate": 7.418459115587054e-08,
      "loss": 0.6669,
      "step": 23772
    },
    {
      "epoch": 0.9852459695801732,
      "grad_norm": 0.4412676990032196,
      "learning_rate": 7.397737162750219e-08,
      "loss": 0.7084,
      "step": 23773
    },
    {
      "epoch": 0.9852874134858469,
      "grad_norm": 0.4421842098236084,
      "learning_rate": 7.377015209913382e-08,
      "loss": 0.6201,
      "step": 23774
    },
    {
      "epoch": 0.9853288573915205,
      "grad_norm": 0.4159706234931946,
      "learning_rate": 7.356293257076548e-08,
      "loss": 0.6732,
      "step": 23775
    },
    {
      "epoch": 0.9853703012971943,
      "grad_norm": 0.41001251339912415,
      "learning_rate": 7.335571304239713e-08,
      "loss": 0.6621,
      "step": 23776
    },
    {
      "epoch": 0.9854117452028679,
      "grad_norm": 0.44088247418403625,
      "learning_rate": 7.314849351402878e-08,
      "loss": 0.7172,
      "step": 23777
    },
    {
      "epoch": 0.9854531891085416,
      "grad_norm": 0.4310210645198822,
      "learning_rate": 7.294127398566041e-08,
      "loss": 0.6521,
      "step": 23778
    },
    {
      "epoch": 0.9854946330142152,
      "grad_norm": 0.48532834649086,
      "learning_rate": 7.273405445729206e-08,
      "loss": 0.7316,
      "step": 23779
    },
    {
      "epoch": 0.985536076919889,
      "grad_norm": 0.43172118067741394,
      "learning_rate": 7.252683492892371e-08,
      "loss": 0.6495,
      "step": 23780
    },
    {
      "epoch": 0.9855775208255626,
      "grad_norm": 0.4010343551635742,
      "learning_rate": 7.231961540055535e-08,
      "loss": 0.6432,
      "step": 23781
    },
    {
      "epoch": 0.9856189647312362,
      "grad_norm": 0.4331321716308594,
      "learning_rate": 7.2112395872187e-08,
      "loss": 0.658,
      "step": 23782
    },
    {
      "epoch": 0.98566040863691,
      "grad_norm": 0.391863077878952,
      "learning_rate": 7.190517634381865e-08,
      "loss": 0.652,
      "step": 23783
    },
    {
      "epoch": 0.9857018525425836,
      "grad_norm": 0.46277379989624023,
      "learning_rate": 7.169795681545029e-08,
      "loss": 0.6967,
      "step": 23784
    },
    {
      "epoch": 0.9857432964482573,
      "grad_norm": 0.4070708453655243,
      "learning_rate": 7.149073728708194e-08,
      "loss": 0.6899,
      "step": 23785
    },
    {
      "epoch": 0.9857847403539309,
      "grad_norm": 0.42039719223976135,
      "learning_rate": 7.128351775871359e-08,
      "loss": 0.6412,
      "step": 23786
    },
    {
      "epoch": 0.9858261842596047,
      "grad_norm": 0.38501501083374023,
      "learning_rate": 7.107629823034524e-08,
      "loss": 0.6182,
      "step": 23787
    },
    {
      "epoch": 0.9858676281652783,
      "grad_norm": 0.44314178824424744,
      "learning_rate": 7.086907870197688e-08,
      "loss": 0.7062,
      "step": 23788
    },
    {
      "epoch": 0.985909072070952,
      "grad_norm": 0.4155070185661316,
      "learning_rate": 7.066185917360853e-08,
      "loss": 0.6428,
      "step": 23789
    },
    {
      "epoch": 0.9859505159766256,
      "grad_norm": 0.43992865085601807,
      "learning_rate": 7.045463964524018e-08,
      "loss": 0.6962,
      "step": 23790
    },
    {
      "epoch": 0.9859919598822993,
      "grad_norm": 0.4030715823173523,
      "learning_rate": 7.024742011687181e-08,
      "loss": 0.6366,
      "step": 23791
    },
    {
      "epoch": 0.986033403787973,
      "grad_norm": 0.37775757908821106,
      "learning_rate": 7.004020058850346e-08,
      "loss": 0.6199,
      "step": 23792
    },
    {
      "epoch": 0.9860748476936466,
      "grad_norm": 0.4076269268989563,
      "learning_rate": 6.983298106013512e-08,
      "loss": 0.6511,
      "step": 23793
    },
    {
      "epoch": 0.9861162915993203,
      "grad_norm": 0.39976966381073,
      "learning_rate": 6.962576153176675e-08,
      "loss": 0.6215,
      "step": 23794
    },
    {
      "epoch": 0.986157735504994,
      "grad_norm": 0.4103187322616577,
      "learning_rate": 6.94185420033984e-08,
      "loss": 0.7004,
      "step": 23795
    },
    {
      "epoch": 0.9861991794106677,
      "grad_norm": 0.44598817825317383,
      "learning_rate": 6.921132247503005e-08,
      "loss": 0.7129,
      "step": 23796
    },
    {
      "epoch": 0.9862406233163413,
      "grad_norm": 0.44460245966911316,
      "learning_rate": 6.90041029466617e-08,
      "loss": 0.6985,
      "step": 23797
    },
    {
      "epoch": 0.9862820672220151,
      "grad_norm": 0.4153294265270233,
      "learning_rate": 6.879688341829334e-08,
      "loss": 0.6163,
      "step": 23798
    },
    {
      "epoch": 0.9863235111276887,
      "grad_norm": 0.43037402629852295,
      "learning_rate": 6.858966388992499e-08,
      "loss": 0.6541,
      "step": 23799
    },
    {
      "epoch": 0.9863649550333623,
      "grad_norm": 0.4054703414440155,
      "learning_rate": 6.838244436155664e-08,
      "loss": 0.6107,
      "step": 23800
    },
    {
      "epoch": 0.986406398939036,
      "grad_norm": 0.4539373815059662,
      "learning_rate": 6.817522483318829e-08,
      "loss": 0.6816,
      "step": 23801
    },
    {
      "epoch": 0.9864478428447097,
      "grad_norm": 0.4354526102542877,
      "learning_rate": 6.796800530481993e-08,
      "loss": 0.7031,
      "step": 23802
    },
    {
      "epoch": 0.9864892867503834,
      "grad_norm": 0.41053515672683716,
      "learning_rate": 6.776078577645158e-08,
      "loss": 0.6602,
      "step": 23803
    },
    {
      "epoch": 0.986530730656057,
      "grad_norm": 0.402578204870224,
      "learning_rate": 6.755356624808323e-08,
      "loss": 0.6365,
      "step": 23804
    },
    {
      "epoch": 0.9865721745617307,
      "grad_norm": 0.40409743785858154,
      "learning_rate": 6.734634671971488e-08,
      "loss": 0.6306,
      "step": 23805
    },
    {
      "epoch": 0.9866136184674044,
      "grad_norm": 0.4055984318256378,
      "learning_rate": 6.713912719134652e-08,
      "loss": 0.6482,
      "step": 23806
    },
    {
      "epoch": 0.986655062373078,
      "grad_norm": 0.4101327359676361,
      "learning_rate": 6.693190766297817e-08,
      "loss": 0.6266,
      "step": 23807
    },
    {
      "epoch": 0.9866965062787517,
      "grad_norm": 0.4070189297199249,
      "learning_rate": 6.672468813460982e-08,
      "loss": 0.6558,
      "step": 23808
    },
    {
      "epoch": 0.9867379501844253,
      "grad_norm": 0.4296228885650635,
      "learning_rate": 6.651746860624145e-08,
      "loss": 0.688,
      "step": 23809
    },
    {
      "epoch": 0.9867793940900991,
      "grad_norm": 0.40965697169303894,
      "learning_rate": 6.63102490778731e-08,
      "loss": 0.6365,
      "step": 23810
    },
    {
      "epoch": 0.9868208379957727,
      "grad_norm": 0.390299916267395,
      "learning_rate": 6.610302954950476e-08,
      "loss": 0.6155,
      "step": 23811
    },
    {
      "epoch": 0.9868622819014464,
      "grad_norm": 0.40320396423339844,
      "learning_rate": 6.589581002113639e-08,
      "loss": 0.6552,
      "step": 23812
    },
    {
      "epoch": 0.98690372580712,
      "grad_norm": 0.38492876291275024,
      "learning_rate": 6.568859049276804e-08,
      "loss": 0.616,
      "step": 23813
    },
    {
      "epoch": 0.9869451697127938,
      "grad_norm": 0.4260987937450409,
      "learning_rate": 6.548137096439969e-08,
      "loss": 0.6617,
      "step": 23814
    },
    {
      "epoch": 0.9869866136184674,
      "grad_norm": 0.4066937267780304,
      "learning_rate": 6.527415143603133e-08,
      "loss": 0.668,
      "step": 23815
    },
    {
      "epoch": 0.987028057524141,
      "grad_norm": 0.3814796805381775,
      "learning_rate": 6.506693190766298e-08,
      "loss": 0.6317,
      "step": 23816
    },
    {
      "epoch": 0.9870695014298148,
      "grad_norm": 0.4391738772392273,
      "learning_rate": 6.485971237929463e-08,
      "loss": 0.6689,
      "step": 23817
    },
    {
      "epoch": 0.9871109453354884,
      "grad_norm": 0.42578840255737305,
      "learning_rate": 6.465249285092627e-08,
      "loss": 0.6863,
      "step": 23818
    },
    {
      "epoch": 0.9871523892411621,
      "grad_norm": 0.40593814849853516,
      "learning_rate": 6.444527332255792e-08,
      "loss": 0.6791,
      "step": 23819
    },
    {
      "epoch": 0.9871938331468357,
      "grad_norm": 0.43989697098731995,
      "learning_rate": 6.423805379418957e-08,
      "loss": 0.7062,
      "step": 23820
    },
    {
      "epoch": 0.9872352770525095,
      "grad_norm": 0.40679386258125305,
      "learning_rate": 6.40308342658212e-08,
      "loss": 0.6855,
      "step": 23821
    },
    {
      "epoch": 0.9872767209581831,
      "grad_norm": 0.4051612317562103,
      "learning_rate": 6.382361473745286e-08,
      "loss": 0.6072,
      "step": 23822
    },
    {
      "epoch": 0.9873181648638568,
      "grad_norm": 0.4297308325767517,
      "learning_rate": 6.361639520908451e-08,
      "loss": 0.6119,
      "step": 23823
    },
    {
      "epoch": 0.9873596087695304,
      "grad_norm": 0.4561173617839813,
      "learning_rate": 6.340917568071616e-08,
      "loss": 0.6948,
      "step": 23824
    },
    {
      "epoch": 0.9874010526752041,
      "grad_norm": 0.41341331601142883,
      "learning_rate": 6.320195615234781e-08,
      "loss": 0.6167,
      "step": 23825
    },
    {
      "epoch": 0.9874424965808778,
      "grad_norm": 0.43012991547584534,
      "learning_rate": 6.299473662397944e-08,
      "loss": 0.587,
      "step": 23826
    },
    {
      "epoch": 0.9874839404865514,
      "grad_norm": 0.43469130992889404,
      "learning_rate": 6.27875170956111e-08,
      "loss": 0.6733,
      "step": 23827
    },
    {
      "epoch": 0.9875253843922251,
      "grad_norm": 0.40984296798706055,
      "learning_rate": 6.258029756724275e-08,
      "loss": 0.6591,
      "step": 23828
    },
    {
      "epoch": 0.9875668282978988,
      "grad_norm": 0.4504980146884918,
      "learning_rate": 6.23730780388744e-08,
      "loss": 0.6689,
      "step": 23829
    },
    {
      "epoch": 0.9876082722035725,
      "grad_norm": 0.4669979214668274,
      "learning_rate": 6.216585851050603e-08,
      "loss": 0.6942,
      "step": 23830
    },
    {
      "epoch": 0.9876497161092461,
      "grad_norm": 0.4514341950416565,
      "learning_rate": 6.195863898213768e-08,
      "loss": 0.676,
      "step": 23831
    },
    {
      "epoch": 0.9876911600149199,
      "grad_norm": 0.4008638262748718,
      "learning_rate": 6.175141945376933e-08,
      "loss": 0.6338,
      "step": 23832
    },
    {
      "epoch": 0.9877326039205935,
      "grad_norm": 0.40999457240104675,
      "learning_rate": 6.154419992540098e-08,
      "loss": 0.6442,
      "step": 23833
    },
    {
      "epoch": 0.9877740478262671,
      "grad_norm": 0.42545345425605774,
      "learning_rate": 6.133698039703262e-08,
      "loss": 0.6472,
      "step": 23834
    },
    {
      "epoch": 0.9878154917319408,
      "grad_norm": 0.4226754605770111,
      "learning_rate": 6.112976086866427e-08,
      "loss": 0.6689,
      "step": 23835
    },
    {
      "epoch": 0.9878569356376145,
      "grad_norm": 0.4396931231021881,
      "learning_rate": 6.092254134029592e-08,
      "loss": 0.5957,
      "step": 23836
    },
    {
      "epoch": 0.9878983795432882,
      "grad_norm": 0.41814103722572327,
      "learning_rate": 6.071532181192756e-08,
      "loss": 0.6963,
      "step": 23837
    },
    {
      "epoch": 0.9879398234489618,
      "grad_norm": 0.4251059889793396,
      "learning_rate": 6.050810228355921e-08,
      "loss": 0.626,
      "step": 23838
    },
    {
      "epoch": 0.9879812673546355,
      "grad_norm": 0.46041062474250793,
      "learning_rate": 6.030088275519086e-08,
      "loss": 0.7013,
      "step": 23839
    },
    {
      "epoch": 0.9880227112603092,
      "grad_norm": 0.42614808678627014,
      "learning_rate": 6.00936632268225e-08,
      "loss": 0.6521,
      "step": 23840
    },
    {
      "epoch": 0.9880641551659829,
      "grad_norm": 0.41670337319374084,
      "learning_rate": 5.988644369845415e-08,
      "loss": 0.7168,
      "step": 23841
    },
    {
      "epoch": 0.9881055990716565,
      "grad_norm": 0.40100279450416565,
      "learning_rate": 5.96792241700858e-08,
      "loss": 0.6604,
      "step": 23842
    },
    {
      "epoch": 0.9881470429773301,
      "grad_norm": 0.46226972341537476,
      "learning_rate": 5.947200464171744e-08,
      "loss": 0.7183,
      "step": 23843
    },
    {
      "epoch": 0.9881884868830039,
      "grad_norm": 0.4439721703529358,
      "learning_rate": 5.9264785113349085e-08,
      "loss": 0.6741,
      "step": 23844
    },
    {
      "epoch": 0.9882299307886775,
      "grad_norm": 0.4113834500312805,
      "learning_rate": 5.9057565584980735e-08,
      "loss": 0.6625,
      "step": 23845
    },
    {
      "epoch": 0.9882713746943512,
      "grad_norm": 0.4051987826824188,
      "learning_rate": 5.885034605661238e-08,
      "loss": 0.629,
      "step": 23846
    },
    {
      "epoch": 0.9883128186000248,
      "grad_norm": 0.3764939308166504,
      "learning_rate": 5.864312652824403e-08,
      "loss": 0.6257,
      "step": 23847
    },
    {
      "epoch": 0.9883542625056986,
      "grad_norm": 0.4414096474647522,
      "learning_rate": 5.843590699987567e-08,
      "loss": 0.6697,
      "step": 23848
    },
    {
      "epoch": 0.9883957064113722,
      "grad_norm": 0.45066481828689575,
      "learning_rate": 5.8228687471507317e-08,
      "loss": 0.6675,
      "step": 23849
    },
    {
      "epoch": 0.9884371503170459,
      "grad_norm": 0.4306630492210388,
      "learning_rate": 5.802146794313897e-08,
      "loss": 0.6279,
      "step": 23850
    },
    {
      "epoch": 0.9884785942227196,
      "grad_norm": 0.428249716758728,
      "learning_rate": 5.781424841477062e-08,
      "loss": 0.6395,
      "step": 23851
    },
    {
      "epoch": 0.9885200381283932,
      "grad_norm": 0.39646971225738525,
      "learning_rate": 5.7607028886402254e-08,
      "loss": 0.6494,
      "step": 23852
    },
    {
      "epoch": 0.9885614820340669,
      "grad_norm": 0.3952701985836029,
      "learning_rate": 5.7399809358033905e-08,
      "loss": 0.6848,
      "step": 23853
    },
    {
      "epoch": 0.9886029259397405,
      "grad_norm": 0.4247997999191284,
      "learning_rate": 5.7192589829665555e-08,
      "loss": 0.6814,
      "step": 23854
    },
    {
      "epoch": 0.9886443698454143,
      "grad_norm": 0.4270767569541931,
      "learning_rate": 5.698537030129719e-08,
      "loss": 0.7244,
      "step": 23855
    },
    {
      "epoch": 0.9886858137510879,
      "grad_norm": 0.40708184242248535,
      "learning_rate": 5.677815077292884e-08,
      "loss": 0.6334,
      "step": 23856
    },
    {
      "epoch": 0.9887272576567616,
      "grad_norm": 0.4274381101131439,
      "learning_rate": 5.657093124456049e-08,
      "loss": 0.6943,
      "step": 23857
    },
    {
      "epoch": 0.9887687015624352,
      "grad_norm": 0.3978058099746704,
      "learning_rate": 5.6363711716192137e-08,
      "loss": 0.6088,
      "step": 23858
    },
    {
      "epoch": 0.988810145468109,
      "grad_norm": 0.40732327103614807,
      "learning_rate": 5.615649218782379e-08,
      "loss": 0.672,
      "step": 23859
    },
    {
      "epoch": 0.9888515893737826,
      "grad_norm": 0.4291216731071472,
      "learning_rate": 5.594927265945543e-08,
      "loss": 0.693,
      "step": 23860
    },
    {
      "epoch": 0.9888930332794562,
      "grad_norm": 0.39499691128730774,
      "learning_rate": 5.5742053131087074e-08,
      "loss": 0.6505,
      "step": 23861
    },
    {
      "epoch": 0.98893447718513,
      "grad_norm": 0.37826603651046753,
      "learning_rate": 5.5534833602718725e-08,
      "loss": 0.6272,
      "step": 23862
    },
    {
      "epoch": 0.9889759210908036,
      "grad_norm": 0.3967103064060211,
      "learning_rate": 5.5327614074350375e-08,
      "loss": 0.6718,
      "step": 23863
    },
    {
      "epoch": 0.9890173649964773,
      "grad_norm": 0.43328383564949036,
      "learning_rate": 5.512039454598201e-08,
      "loss": 0.673,
      "step": 23864
    },
    {
      "epoch": 0.9890588089021509,
      "grad_norm": 0.3892805278301239,
      "learning_rate": 5.491317501761366e-08,
      "loss": 0.6353,
      "step": 23865
    },
    {
      "epoch": 0.9891002528078247,
      "grad_norm": 0.4255126416683197,
      "learning_rate": 5.470595548924531e-08,
      "loss": 0.6968,
      "step": 23866
    },
    {
      "epoch": 0.9891416967134983,
      "grad_norm": 0.4232417941093445,
      "learning_rate": 5.449873596087695e-08,
      "loss": 0.6812,
      "step": 23867
    },
    {
      "epoch": 0.9891831406191719,
      "grad_norm": 0.44028931856155396,
      "learning_rate": 5.42915164325086e-08,
      "loss": 0.6375,
      "step": 23868
    },
    {
      "epoch": 0.9892245845248456,
      "grad_norm": 0.41984230279922485,
      "learning_rate": 5.408429690414025e-08,
      "loss": 0.6682,
      "step": 23869
    },
    {
      "epoch": 0.9892660284305193,
      "grad_norm": 0.4386623799800873,
      "learning_rate": 5.38770773757719e-08,
      "loss": 0.6237,
      "step": 23870
    },
    {
      "epoch": 0.989307472336193,
      "grad_norm": 0.3988359272480011,
      "learning_rate": 5.3669857847403545e-08,
      "loss": 0.5978,
      "step": 23871
    },
    {
      "epoch": 0.9893489162418666,
      "grad_norm": 0.420605331659317,
      "learning_rate": 5.346263831903519e-08,
      "loss": 0.7124,
      "step": 23872
    },
    {
      "epoch": 0.9893903601475403,
      "grad_norm": 0.44066599011421204,
      "learning_rate": 5.325541879066684e-08,
      "loss": 0.6589,
      "step": 23873
    },
    {
      "epoch": 0.989431804053214,
      "grad_norm": 0.40938159823417664,
      "learning_rate": 5.304819926229848e-08,
      "loss": 0.6804,
      "step": 23874
    },
    {
      "epoch": 0.9894732479588877,
      "grad_norm": 0.41471728682518005,
      "learning_rate": 5.284097973393013e-08,
      "loss": 0.6776,
      "step": 23875
    },
    {
      "epoch": 0.9895146918645613,
      "grad_norm": 0.3953459560871124,
      "learning_rate": 5.2633760205561783e-08,
      "loss": 0.6382,
      "step": 23876
    },
    {
      "epoch": 0.9895561357702349,
      "grad_norm": 0.42707890272140503,
      "learning_rate": 5.242654067719342e-08,
      "loss": 0.6348,
      "step": 23877
    },
    {
      "epoch": 0.9895975796759087,
      "grad_norm": 0.4542788863182068,
      "learning_rate": 5.221932114882507e-08,
      "loss": 0.6445,
      "step": 23878
    },
    {
      "epoch": 0.9896390235815823,
      "grad_norm": 0.4172305762767792,
      "learning_rate": 5.201210162045672e-08,
      "loss": 0.682,
      "step": 23879
    },
    {
      "epoch": 0.989680467487256,
      "grad_norm": 0.41861143708229065,
      "learning_rate": 5.180488209208836e-08,
      "loss": 0.677,
      "step": 23880
    },
    {
      "epoch": 0.9897219113929296,
      "grad_norm": 0.43870046734809875,
      "learning_rate": 5.159766256372001e-08,
      "loss": 0.6703,
      "step": 23881
    },
    {
      "epoch": 0.9897633552986034,
      "grad_norm": 0.4483950734138489,
      "learning_rate": 5.139044303535166e-08,
      "loss": 0.6929,
      "step": 23882
    },
    {
      "epoch": 0.989804799204277,
      "grad_norm": 0.44413384795188904,
      "learning_rate": 5.11832235069833e-08,
      "loss": 0.6893,
      "step": 23883
    },
    {
      "epoch": 0.9898462431099507,
      "grad_norm": 0.411482572555542,
      "learning_rate": 5.0976003978614947e-08,
      "loss": 0.6399,
      "step": 23884
    },
    {
      "epoch": 0.9898876870156244,
      "grad_norm": 0.42969369888305664,
      "learning_rate": 5.07687844502466e-08,
      "loss": 0.6177,
      "step": 23885
    },
    {
      "epoch": 0.989929130921298,
      "grad_norm": 0.4268505275249481,
      "learning_rate": 5.056156492187824e-08,
      "loss": 0.6855,
      "step": 23886
    },
    {
      "epoch": 0.9899705748269717,
      "grad_norm": 0.4469764232635498,
      "learning_rate": 5.035434539350989e-08,
      "loss": 0.6772,
      "step": 23887
    },
    {
      "epoch": 0.9900120187326453,
      "grad_norm": 0.40855875611305237,
      "learning_rate": 5.0147125865141535e-08,
      "loss": 0.6421,
      "step": 23888
    },
    {
      "epoch": 0.9900534626383191,
      "grad_norm": 0.49218034744262695,
      "learning_rate": 4.993990633677318e-08,
      "loss": 0.6447,
      "step": 23889
    },
    {
      "epoch": 0.9900949065439927,
      "grad_norm": 0.42165106534957886,
      "learning_rate": 4.973268680840483e-08,
      "loss": 0.6567,
      "step": 23890
    },
    {
      "epoch": 0.9901363504496664,
      "grad_norm": 0.43486782908439636,
      "learning_rate": 4.952546728003648e-08,
      "loss": 0.6763,
      "step": 23891
    },
    {
      "epoch": 0.99017779435534,
      "grad_norm": 0.4308215081691742,
      "learning_rate": 4.9318247751668116e-08,
      "loss": 0.7178,
      "step": 23892
    },
    {
      "epoch": 0.9902192382610138,
      "grad_norm": 0.4345095455646515,
      "learning_rate": 4.9111028223299767e-08,
      "loss": 0.6533,
      "step": 23893
    },
    {
      "epoch": 0.9902606821666874,
      "grad_norm": 0.4309340715408325,
      "learning_rate": 4.890380869493142e-08,
      "loss": 0.6874,
      "step": 23894
    },
    {
      "epoch": 0.990302126072361,
      "grad_norm": 0.3916526138782501,
      "learning_rate": 4.869658916656306e-08,
      "loss": 0.6406,
      "step": 23895
    },
    {
      "epoch": 0.9903435699780347,
      "grad_norm": 0.42533913254737854,
      "learning_rate": 4.8489369638194704e-08,
      "loss": 0.616,
      "step": 23896
    },
    {
      "epoch": 0.9903850138837084,
      "grad_norm": 0.45141837000846863,
      "learning_rate": 4.8282150109826355e-08,
      "loss": 0.6383,
      "step": 23897
    },
    {
      "epoch": 0.9904264577893821,
      "grad_norm": 0.4341018795967102,
      "learning_rate": 4.8074930581458e-08,
      "loss": 0.6533,
      "step": 23898
    },
    {
      "epoch": 0.9904679016950557,
      "grad_norm": 0.4130069315433502,
      "learning_rate": 4.786771105308965e-08,
      "loss": 0.6445,
      "step": 23899
    },
    {
      "epoch": 0.9905093456007295,
      "grad_norm": 0.44865158200263977,
      "learning_rate": 4.766049152472129e-08,
      "loss": 0.637,
      "step": 23900
    },
    {
      "epoch": 0.9905507895064031,
      "grad_norm": 0.43999144434928894,
      "learning_rate": 4.7453271996352936e-08,
      "loss": 0.6555,
      "step": 23901
    },
    {
      "epoch": 0.9905922334120768,
      "grad_norm": 0.38872820138931274,
      "learning_rate": 4.724605246798459e-08,
      "loss": 0.6653,
      "step": 23902
    },
    {
      "epoch": 0.9906336773177504,
      "grad_norm": 0.42373907566070557,
      "learning_rate": 4.703883293961624e-08,
      "loss": 0.6724,
      "step": 23903
    },
    {
      "epoch": 0.990675121223424,
      "grad_norm": 0.4237865209579468,
      "learning_rate": 4.6831613411247874e-08,
      "loss": 0.6896,
      "step": 23904
    },
    {
      "epoch": 0.9907165651290978,
      "grad_norm": 0.4095766246318817,
      "learning_rate": 4.6624393882879525e-08,
      "loss": 0.6765,
      "step": 23905
    },
    {
      "epoch": 0.9907580090347714,
      "grad_norm": 0.41105422377586365,
      "learning_rate": 4.6417174354511175e-08,
      "loss": 0.6429,
      "step": 23906
    },
    {
      "epoch": 0.9907994529404451,
      "grad_norm": 0.4342929720878601,
      "learning_rate": 4.620995482614281e-08,
      "loss": 0.6572,
      "step": 23907
    },
    {
      "epoch": 0.9908408968461188,
      "grad_norm": 0.4210394322872162,
      "learning_rate": 4.600273529777446e-08,
      "loss": 0.6763,
      "step": 23908
    },
    {
      "epoch": 0.9908823407517925,
      "grad_norm": 0.4643417298793793,
      "learning_rate": 4.579551576940611e-08,
      "loss": 0.6959,
      "step": 23909
    },
    {
      "epoch": 0.9909237846574661,
      "grad_norm": 0.4214555621147156,
      "learning_rate": 4.558829624103776e-08,
      "loss": 0.6649,
      "step": 23910
    },
    {
      "epoch": 0.9909652285631398,
      "grad_norm": 0.4132906198501587,
      "learning_rate": 4.538107671266941e-08,
      "loss": 0.6086,
      "step": 23911
    },
    {
      "epoch": 0.9910066724688135,
      "grad_norm": 0.41022923588752747,
      "learning_rate": 4.517385718430105e-08,
      "loss": 0.6366,
      "step": 23912
    },
    {
      "epoch": 0.9910481163744871,
      "grad_norm": 0.3995725214481354,
      "learning_rate": 4.49666376559327e-08,
      "loss": 0.7126,
      "step": 23913
    },
    {
      "epoch": 0.9910895602801608,
      "grad_norm": 0.4443933069705963,
      "learning_rate": 4.4759418127564345e-08,
      "loss": 0.6436,
      "step": 23914
    },
    {
      "epoch": 0.9911310041858344,
      "grad_norm": 0.38557112216949463,
      "learning_rate": 4.4552198599195995e-08,
      "loss": 0.6539,
      "step": 23915
    },
    {
      "epoch": 0.9911724480915082,
      "grad_norm": 0.45564794540405273,
      "learning_rate": 4.4344979070827645e-08,
      "loss": 0.6544,
      "step": 23916
    },
    {
      "epoch": 0.9912138919971818,
      "grad_norm": 0.46011221408843994,
      "learning_rate": 4.413775954245928e-08,
      "loss": 0.6785,
      "step": 23917
    },
    {
      "epoch": 0.9912553359028555,
      "grad_norm": 0.4133033752441406,
      "learning_rate": 4.393054001409093e-08,
      "loss": 0.6919,
      "step": 23918
    },
    {
      "epoch": 0.9912967798085291,
      "grad_norm": 0.4311780631542206,
      "learning_rate": 4.372332048572258e-08,
      "loss": 0.6326,
      "step": 23919
    },
    {
      "epoch": 0.9913382237142029,
      "grad_norm": 0.40672388672828674,
      "learning_rate": 4.351610095735422e-08,
      "loss": 0.6003,
      "step": 23920
    },
    {
      "epoch": 0.9913796676198765,
      "grad_norm": 0.39109909534454346,
      "learning_rate": 4.330888142898587e-08,
      "loss": 0.6613,
      "step": 23921
    },
    {
      "epoch": 0.9914211115255501,
      "grad_norm": 0.39611127972602844,
      "learning_rate": 4.310166190061752e-08,
      "loss": 0.6422,
      "step": 23922
    },
    {
      "epoch": 0.9914625554312239,
      "grad_norm": 0.3867829740047455,
      "learning_rate": 4.2894442372249165e-08,
      "loss": 0.6772,
      "step": 23923
    },
    {
      "epoch": 0.9915039993368975,
      "grad_norm": 0.4738997519016266,
      "learning_rate": 4.268722284388081e-08,
      "loss": 0.6718,
      "step": 23924
    },
    {
      "epoch": 0.9915454432425712,
      "grad_norm": 0.38147035241127014,
      "learning_rate": 4.248000331551246e-08,
      "loss": 0.6273,
      "step": 23925
    },
    {
      "epoch": 0.9915868871482448,
      "grad_norm": 0.40971487760543823,
      "learning_rate": 4.22727837871441e-08,
      "loss": 0.6577,
      "step": 23926
    },
    {
      "epoch": 0.9916283310539186,
      "grad_norm": 0.40269848704338074,
      "learning_rate": 4.206556425877575e-08,
      "loss": 0.6544,
      "step": 23927
    },
    {
      "epoch": 0.9916697749595922,
      "grad_norm": 0.3963625729084015,
      "learning_rate": 4.18583447304074e-08,
      "loss": 0.6517,
      "step": 23928
    },
    {
      "epoch": 0.9917112188652658,
      "grad_norm": 0.4332420825958252,
      "learning_rate": 4.165112520203904e-08,
      "loss": 0.6876,
      "step": 23929
    },
    {
      "epoch": 0.9917526627709395,
      "grad_norm": 0.4320157766342163,
      "learning_rate": 4.144390567367069e-08,
      "loss": 0.6746,
      "step": 23930
    },
    {
      "epoch": 0.9917941066766132,
      "grad_norm": 0.4295324981212616,
      "learning_rate": 4.123668614530234e-08,
      "loss": 0.6685,
      "step": 23931
    },
    {
      "epoch": 0.9918355505822869,
      "grad_norm": 0.4232003092765808,
      "learning_rate": 4.102946661693398e-08,
      "loss": 0.6641,
      "step": 23932
    },
    {
      "epoch": 0.9918769944879605,
      "grad_norm": 0.4093596339225769,
      "learning_rate": 4.082224708856563e-08,
      "loss": 0.6355,
      "step": 23933
    },
    {
      "epoch": 0.9919184383936342,
      "grad_norm": 0.4533938467502594,
      "learning_rate": 4.061502756019728e-08,
      "loss": 0.6539,
      "step": 23934
    },
    {
      "epoch": 0.9919598822993079,
      "grad_norm": 0.44928279519081116,
      "learning_rate": 4.040780803182892e-08,
      "loss": 0.6487,
      "step": 23935
    },
    {
      "epoch": 0.9920013262049816,
      "grad_norm": 0.4099031388759613,
      "learning_rate": 4.0200588503460566e-08,
      "loss": 0.6631,
      "step": 23936
    },
    {
      "epoch": 0.9920427701106552,
      "grad_norm": 0.43626055121421814,
      "learning_rate": 3.9993368975092217e-08,
      "loss": 0.7251,
      "step": 23937
    },
    {
      "epoch": 0.9920842140163288,
      "grad_norm": 0.4466804563999176,
      "learning_rate": 3.978614944672386e-08,
      "loss": 0.6403,
      "step": 23938
    },
    {
      "epoch": 0.9921256579220026,
      "grad_norm": 0.3976042866706848,
      "learning_rate": 3.957892991835551e-08,
      "loss": 0.6429,
      "step": 23939
    },
    {
      "epoch": 0.9921671018276762,
      "grad_norm": 0.4415148198604584,
      "learning_rate": 3.9371710389987154e-08,
      "loss": 0.6615,
      "step": 23940
    },
    {
      "epoch": 0.9922085457333499,
      "grad_norm": 0.40020158886909485,
      "learning_rate": 3.91644908616188e-08,
      "loss": 0.6219,
      "step": 23941
    },
    {
      "epoch": 0.9922499896390236,
      "grad_norm": 0.41308656334877014,
      "learning_rate": 3.895727133325045e-08,
      "loss": 0.6434,
      "step": 23942
    },
    {
      "epoch": 0.9922914335446973,
      "grad_norm": 0.39947307109832764,
      "learning_rate": 3.87500518048821e-08,
      "loss": 0.6427,
      "step": 23943
    },
    {
      "epoch": 0.9923328774503709,
      "grad_norm": 0.43069303035736084,
      "learning_rate": 3.854283227651374e-08,
      "loss": 0.6299,
      "step": 23944
    },
    {
      "epoch": 0.9923743213560446,
      "grad_norm": 0.4048948287963867,
      "learning_rate": 3.8335612748145386e-08,
      "loss": 0.6355,
      "step": 23945
    },
    {
      "epoch": 0.9924157652617183,
      "grad_norm": 0.40478643774986267,
      "learning_rate": 3.812839321977704e-08,
      "loss": 0.6652,
      "step": 23946
    },
    {
      "epoch": 0.9924572091673919,
      "grad_norm": 0.4193378686904907,
      "learning_rate": 3.792117369140868e-08,
      "loss": 0.6855,
      "step": 23947
    },
    {
      "epoch": 0.9924986530730656,
      "grad_norm": 0.41007861495018005,
      "learning_rate": 3.7713954163040324e-08,
      "loss": 0.6731,
      "step": 23948
    },
    {
      "epoch": 0.9925400969787392,
      "grad_norm": 0.4058210849761963,
      "learning_rate": 3.7506734634671975e-08,
      "loss": 0.6681,
      "step": 23949
    },
    {
      "epoch": 0.992581540884413,
      "grad_norm": 0.40121230483055115,
      "learning_rate": 3.729951510630362e-08,
      "loss": 0.6497,
      "step": 23950
    },
    {
      "epoch": 0.9926229847900866,
      "grad_norm": 0.40425416827201843,
      "learning_rate": 3.709229557793527e-08,
      "loss": 0.6248,
      "step": 23951
    },
    {
      "epoch": 0.9926644286957603,
      "grad_norm": 0.45725902915000916,
      "learning_rate": 3.688507604956691e-08,
      "loss": 0.6616,
      "step": 23952
    },
    {
      "epoch": 0.992705872601434,
      "grad_norm": 0.41074100136756897,
      "learning_rate": 3.667785652119856e-08,
      "loss": 0.6538,
      "step": 23953
    },
    {
      "epoch": 0.9927473165071077,
      "grad_norm": 0.4123689532279968,
      "learning_rate": 3.6470636992830206e-08,
      "loss": 0.6968,
      "step": 23954
    },
    {
      "epoch": 0.9927887604127813,
      "grad_norm": 0.4388999342918396,
      "learning_rate": 3.626341746446186e-08,
      "loss": 0.6357,
      "step": 23955
    },
    {
      "epoch": 0.9928302043184549,
      "grad_norm": 0.4298550486564636,
      "learning_rate": 3.60561979360935e-08,
      "loss": 0.6831,
      "step": 23956
    },
    {
      "epoch": 0.9928716482241287,
      "grad_norm": 0.4490863084793091,
      "learning_rate": 3.5848978407725144e-08,
      "loss": 0.6755,
      "step": 23957
    },
    {
      "epoch": 0.9929130921298023,
      "grad_norm": 0.4286145865917206,
      "learning_rate": 3.5641758879356795e-08,
      "loss": 0.6887,
      "step": 23958
    },
    {
      "epoch": 0.992954536035476,
      "grad_norm": 0.4001103937625885,
      "learning_rate": 3.543453935098844e-08,
      "loss": 0.6658,
      "step": 23959
    },
    {
      "epoch": 0.9929959799411496,
      "grad_norm": 0.434372216463089,
      "learning_rate": 3.522731982262009e-08,
      "loss": 0.6318,
      "step": 23960
    },
    {
      "epoch": 0.9930374238468234,
      "grad_norm": 0.43184077739715576,
      "learning_rate": 3.502010029425173e-08,
      "loss": 0.7129,
      "step": 23961
    },
    {
      "epoch": 0.993078867752497,
      "grad_norm": 0.41406378149986267,
      "learning_rate": 3.4812880765883376e-08,
      "loss": 0.6743,
      "step": 23962
    },
    {
      "epoch": 0.9931203116581707,
      "grad_norm": 0.4166565239429474,
      "learning_rate": 3.4605661237515027e-08,
      "loss": 0.7332,
      "step": 23963
    },
    {
      "epoch": 0.9931617555638443,
      "grad_norm": 0.42120128870010376,
      "learning_rate": 3.439844170914667e-08,
      "loss": 0.6836,
      "step": 23964
    },
    {
      "epoch": 0.993203199469518,
      "grad_norm": 0.37605899572372437,
      "learning_rate": 3.419122218077832e-08,
      "loss": 0.599,
      "step": 23965
    },
    {
      "epoch": 0.9932446433751917,
      "grad_norm": 0.39245766401290894,
      "learning_rate": 3.3984002652409964e-08,
      "loss": 0.6328,
      "step": 23966
    },
    {
      "epoch": 0.9932860872808653,
      "grad_norm": 0.5243255496025085,
      "learning_rate": 3.3776783124041615e-08,
      "loss": 0.6299,
      "step": 23967
    },
    {
      "epoch": 0.993327531186539,
      "grad_norm": 0.38965901732444763,
      "learning_rate": 3.356956359567326e-08,
      "loss": 0.6766,
      "step": 23968
    },
    {
      "epoch": 0.9933689750922127,
      "grad_norm": 0.3873787820339203,
      "learning_rate": 3.336234406730491e-08,
      "loss": 0.7056,
      "step": 23969
    },
    {
      "epoch": 0.9934104189978864,
      "grad_norm": 0.44185787439346313,
      "learning_rate": 3.315512453893655e-08,
      "loss": 0.6628,
      "step": 23970
    },
    {
      "epoch": 0.99345186290356,
      "grad_norm": 0.4705636501312256,
      "learning_rate": 3.2947905010568196e-08,
      "loss": 0.7012,
      "step": 23971
    },
    {
      "epoch": 0.9934933068092338,
      "grad_norm": 0.44281303882598877,
      "learning_rate": 3.2740685482199847e-08,
      "loss": 0.6128,
      "step": 23972
    },
    {
      "epoch": 0.9935347507149074,
      "grad_norm": 0.43493175506591797,
      "learning_rate": 3.253346595383149e-08,
      "loss": 0.6769,
      "step": 23973
    },
    {
      "epoch": 0.993576194620581,
      "grad_norm": 0.4483121633529663,
      "learning_rate": 3.2326246425463134e-08,
      "loss": 0.6819,
      "step": 23974
    },
    {
      "epoch": 0.9936176385262547,
      "grad_norm": 0.42820483446121216,
      "learning_rate": 3.2119026897094784e-08,
      "loss": 0.6864,
      "step": 23975
    },
    {
      "epoch": 0.9936590824319284,
      "grad_norm": 0.39589545130729675,
      "learning_rate": 3.191180736872643e-08,
      "loss": 0.6355,
      "step": 23976
    },
    {
      "epoch": 0.9937005263376021,
      "grad_norm": 0.41566401720046997,
      "learning_rate": 3.170458784035808e-08,
      "loss": 0.6631,
      "step": 23977
    },
    {
      "epoch": 0.9937419702432757,
      "grad_norm": 0.4182901084423065,
      "learning_rate": 3.149736831198972e-08,
      "loss": 0.6381,
      "step": 23978
    },
    {
      "epoch": 0.9937834141489494,
      "grad_norm": 0.4169228971004486,
      "learning_rate": 3.129014878362137e-08,
      "loss": 0.6652,
      "step": 23979
    },
    {
      "epoch": 0.9938248580546231,
      "grad_norm": 0.4141540229320526,
      "learning_rate": 3.1082929255253016e-08,
      "loss": 0.6592,
      "step": 23980
    },
    {
      "epoch": 0.9938663019602967,
      "grad_norm": 0.39370259642601013,
      "learning_rate": 3.087570972688467e-08,
      "loss": 0.6199,
      "step": 23981
    },
    {
      "epoch": 0.9939077458659704,
      "grad_norm": 0.40079283714294434,
      "learning_rate": 3.066849019851631e-08,
      "loss": 0.6102,
      "step": 23982
    },
    {
      "epoch": 0.993949189771644,
      "grad_norm": 0.4259001314640045,
      "learning_rate": 3.046127067014796e-08,
      "loss": 0.6472,
      "step": 23983
    },
    {
      "epoch": 0.9939906336773178,
      "grad_norm": 0.4223737418651581,
      "learning_rate": 3.0254051141779605e-08,
      "loss": 0.6648,
      "step": 23984
    },
    {
      "epoch": 0.9940320775829914,
      "grad_norm": 0.3827546238899231,
      "learning_rate": 3.004683161341125e-08,
      "loss": 0.6326,
      "step": 23985
    },
    {
      "epoch": 0.9940735214886651,
      "grad_norm": 0.4522757828235626,
      "learning_rate": 2.98396120850429e-08,
      "loss": 0.7242,
      "step": 23986
    },
    {
      "epoch": 0.9941149653943387,
      "grad_norm": 0.45495134592056274,
      "learning_rate": 2.9632392556674542e-08,
      "loss": 0.6852,
      "step": 23987
    },
    {
      "epoch": 0.9941564093000125,
      "grad_norm": 0.4497627317905426,
      "learning_rate": 2.942517302830619e-08,
      "loss": 0.6965,
      "step": 23988
    },
    {
      "epoch": 0.9941978532056861,
      "grad_norm": 0.4223647713661194,
      "learning_rate": 2.9217953499937836e-08,
      "loss": 0.6885,
      "step": 23989
    },
    {
      "epoch": 0.9942392971113597,
      "grad_norm": 0.42440611124038696,
      "learning_rate": 2.9010733971569483e-08,
      "loss": 0.6421,
      "step": 23990
    },
    {
      "epoch": 0.9942807410170335,
      "grad_norm": 0.4231124222278595,
      "learning_rate": 2.8803514443201127e-08,
      "loss": 0.7598,
      "step": 23991
    },
    {
      "epoch": 0.9943221849227071,
      "grad_norm": 0.41109099984169006,
      "learning_rate": 2.8596294914832778e-08,
      "loss": 0.6433,
      "step": 23992
    },
    {
      "epoch": 0.9943636288283808,
      "grad_norm": 0.414800763130188,
      "learning_rate": 2.838907538646442e-08,
      "loss": 0.6938,
      "step": 23993
    },
    {
      "epoch": 0.9944050727340544,
      "grad_norm": 0.4483809173107147,
      "learning_rate": 2.8181855858096068e-08,
      "loss": 0.6476,
      "step": 23994
    },
    {
      "epoch": 0.9944465166397282,
      "grad_norm": 0.43550732731819153,
      "learning_rate": 2.7974636329727715e-08,
      "loss": 0.6639,
      "step": 23995
    },
    {
      "epoch": 0.9944879605454018,
      "grad_norm": 0.39853668212890625,
      "learning_rate": 2.7767416801359362e-08,
      "loss": 0.6302,
      "step": 23996
    },
    {
      "epoch": 0.9945294044510755,
      "grad_norm": 0.48836749792099,
      "learning_rate": 2.7560197272991006e-08,
      "loss": 0.7449,
      "step": 23997
    },
    {
      "epoch": 0.9945708483567491,
      "grad_norm": 0.4014059603214264,
      "learning_rate": 2.7352977744622657e-08,
      "loss": 0.6176,
      "step": 23998
    },
    {
      "epoch": 0.9946122922624228,
      "grad_norm": 0.4268745183944702,
      "learning_rate": 2.71457582162543e-08,
      "loss": 0.6741,
      "step": 23999
    },
    {
      "epoch": 0.9946537361680965,
      "grad_norm": 0.3827657997608185,
      "learning_rate": 2.693853868788595e-08,
      "loss": 0.6486,
      "step": 24000
    },
    {
      "epoch": 0.9946951800737701,
      "grad_norm": 0.4078461527824402,
      "learning_rate": 2.6731319159517594e-08,
      "loss": 0.6929,
      "step": 24001
    },
    {
      "epoch": 0.9947366239794438,
      "grad_norm": 0.42268165946006775,
      "learning_rate": 2.652409963114924e-08,
      "loss": 0.6471,
      "step": 24002
    },
    {
      "epoch": 0.9947780678851175,
      "grad_norm": 0.4636204242706299,
      "learning_rate": 2.6316880102780892e-08,
      "loss": 0.5714,
      "step": 24003
    },
    {
      "epoch": 0.9948195117907912,
      "grad_norm": 0.405472069978714,
      "learning_rate": 2.6109660574412535e-08,
      "loss": 0.6396,
      "step": 24004
    },
    {
      "epoch": 0.9948609556964648,
      "grad_norm": 0.39924898743629456,
      "learning_rate": 2.590244104604418e-08,
      "loss": 0.6373,
      "step": 24005
    },
    {
      "epoch": 0.9949023996021386,
      "grad_norm": 0.4534502923488617,
      "learning_rate": 2.569522151767583e-08,
      "loss": 0.6558,
      "step": 24006
    },
    {
      "epoch": 0.9949438435078122,
      "grad_norm": 0.3909415900707245,
      "learning_rate": 2.5488001989307473e-08,
      "loss": 0.6665,
      "step": 24007
    },
    {
      "epoch": 0.9949852874134858,
      "grad_norm": 0.43003329634666443,
      "learning_rate": 2.528078246093912e-08,
      "loss": 0.6794,
      "step": 24008
    },
    {
      "epoch": 0.9950267313191595,
      "grad_norm": 0.4201449751853943,
      "learning_rate": 2.5073562932570767e-08,
      "loss": 0.6494,
      "step": 24009
    },
    {
      "epoch": 0.9950681752248332,
      "grad_norm": 0.3922904133796692,
      "learning_rate": 2.4866343404202414e-08,
      "loss": 0.6698,
      "step": 24010
    },
    {
      "epoch": 0.9951096191305069,
      "grad_norm": 0.4540359079837799,
      "learning_rate": 2.4659123875834058e-08,
      "loss": 0.6941,
      "step": 24011
    },
    {
      "epoch": 0.9951510630361805,
      "grad_norm": 0.41348156332969666,
      "learning_rate": 2.445190434746571e-08,
      "loss": 0.5879,
      "step": 24012
    },
    {
      "epoch": 0.9951925069418542,
      "grad_norm": 0.43994447588920593,
      "learning_rate": 2.4244684819097352e-08,
      "loss": 0.6561,
      "step": 24013
    },
    {
      "epoch": 0.9952339508475279,
      "grad_norm": 0.4196043610572815,
      "learning_rate": 2.4037465290729e-08,
      "loss": 0.6593,
      "step": 24014
    },
    {
      "epoch": 0.9952753947532016,
      "grad_norm": 0.40524864196777344,
      "learning_rate": 2.3830245762360646e-08,
      "loss": 0.7095,
      "step": 24015
    },
    {
      "epoch": 0.9953168386588752,
      "grad_norm": 0.40045714378356934,
      "learning_rate": 2.3623026233992293e-08,
      "loss": 0.6552,
      "step": 24016
    },
    {
      "epoch": 0.9953582825645488,
      "grad_norm": 0.4123160243034363,
      "learning_rate": 2.3415806705623937e-08,
      "loss": 0.7189,
      "step": 24017
    },
    {
      "epoch": 0.9953997264702226,
      "grad_norm": 0.43043577671051025,
      "learning_rate": 2.3208587177255587e-08,
      "loss": 0.6654,
      "step": 24018
    },
    {
      "epoch": 0.9954411703758962,
      "grad_norm": 0.41473039984703064,
      "learning_rate": 2.300136764888723e-08,
      "loss": 0.651,
      "step": 24019
    },
    {
      "epoch": 0.9954826142815699,
      "grad_norm": 0.4346679747104645,
      "learning_rate": 2.279414812051888e-08,
      "loss": 0.7111,
      "step": 24020
    },
    {
      "epoch": 0.9955240581872435,
      "grad_norm": 0.4852639436721802,
      "learning_rate": 2.2586928592150525e-08,
      "loss": 0.6508,
      "step": 24021
    },
    {
      "epoch": 0.9955655020929173,
      "grad_norm": 0.41961541771888733,
      "learning_rate": 2.2379709063782172e-08,
      "loss": 0.6517,
      "step": 24022
    },
    {
      "epoch": 0.9956069459985909,
      "grad_norm": 0.47137436270713806,
      "learning_rate": 2.2172489535413823e-08,
      "loss": 0.6381,
      "step": 24023
    },
    {
      "epoch": 0.9956483899042646,
      "grad_norm": 0.4609004855155945,
      "learning_rate": 2.1965270007045466e-08,
      "loss": 0.6813,
      "step": 24024
    },
    {
      "epoch": 0.9956898338099383,
      "grad_norm": 0.3929833173751831,
      "learning_rate": 2.175805047867711e-08,
      "loss": 0.6216,
      "step": 24025
    },
    {
      "epoch": 0.9957312777156119,
      "grad_norm": 0.43735677003860474,
      "learning_rate": 2.155083095030876e-08,
      "loss": 0.6576,
      "step": 24026
    },
    {
      "epoch": 0.9957727216212856,
      "grad_norm": 0.392007052898407,
      "learning_rate": 2.1343611421940404e-08,
      "loss": 0.6396,
      "step": 24027
    },
    {
      "epoch": 0.9958141655269592,
      "grad_norm": 0.5145682692527771,
      "learning_rate": 2.113639189357205e-08,
      "loss": 0.7427,
      "step": 24028
    },
    {
      "epoch": 0.995855609432633,
      "grad_norm": 0.39346325397491455,
      "learning_rate": 2.09291723652037e-08,
      "loss": 0.6226,
      "step": 24029
    },
    {
      "epoch": 0.9958970533383066,
      "grad_norm": 0.4105379283428192,
      "learning_rate": 2.0721952836835345e-08,
      "loss": 0.7173,
      "step": 24030
    },
    {
      "epoch": 0.9959384972439803,
      "grad_norm": 0.4506531059741974,
      "learning_rate": 2.051473330846699e-08,
      "loss": 0.6978,
      "step": 24031
    },
    {
      "epoch": 0.9959799411496539,
      "grad_norm": 0.40263912081718445,
      "learning_rate": 2.030751378009864e-08,
      "loss": 0.6287,
      "step": 24032
    },
    {
      "epoch": 0.9960213850553277,
      "grad_norm": 0.43123677372932434,
      "learning_rate": 2.0100294251730283e-08,
      "loss": 0.6704,
      "step": 24033
    },
    {
      "epoch": 0.9960628289610013,
      "grad_norm": 0.41216036677360535,
      "learning_rate": 1.989307472336193e-08,
      "loss": 0.6636,
      "step": 24034
    },
    {
      "epoch": 0.9961042728666749,
      "grad_norm": 0.4459758698940277,
      "learning_rate": 1.9685855194993577e-08,
      "loss": 0.6793,
      "step": 24035
    },
    {
      "epoch": 0.9961457167723486,
      "grad_norm": 0.3996075689792633,
      "learning_rate": 1.9478635666625224e-08,
      "loss": 0.6106,
      "step": 24036
    },
    {
      "epoch": 0.9961871606780223,
      "grad_norm": 0.41113999485969543,
      "learning_rate": 1.927141613825687e-08,
      "loss": 0.6396,
      "step": 24037
    },
    {
      "epoch": 0.996228604583696,
      "grad_norm": 0.39087584614753723,
      "learning_rate": 1.906419660988852e-08,
      "loss": 0.6481,
      "step": 24038
    },
    {
      "epoch": 0.9962700484893696,
      "grad_norm": 0.42288389801979065,
      "learning_rate": 1.8856977081520162e-08,
      "loss": 0.694,
      "step": 24039
    },
    {
      "epoch": 0.9963114923950434,
      "grad_norm": 0.4320143759250641,
      "learning_rate": 1.864975755315181e-08,
      "loss": 0.6553,
      "step": 24040
    },
    {
      "epoch": 0.996352936300717,
      "grad_norm": 0.43176206946372986,
      "learning_rate": 1.8442538024783456e-08,
      "loss": 0.6361,
      "step": 24041
    },
    {
      "epoch": 0.9963943802063906,
      "grad_norm": 0.4584255516529083,
      "learning_rate": 1.8235318496415103e-08,
      "loss": 0.6851,
      "step": 24042
    },
    {
      "epoch": 0.9964358241120643,
      "grad_norm": 0.41714248061180115,
      "learning_rate": 1.802809896804675e-08,
      "loss": 0.667,
      "step": 24043
    },
    {
      "epoch": 0.996477268017738,
      "grad_norm": 0.43292784690856934,
      "learning_rate": 1.7820879439678397e-08,
      "loss": 0.6361,
      "step": 24044
    },
    {
      "epoch": 0.9965187119234117,
      "grad_norm": 0.4160439968109131,
      "learning_rate": 1.7613659911310044e-08,
      "loss": 0.6814,
      "step": 24045
    },
    {
      "epoch": 0.9965601558290853,
      "grad_norm": 0.46269482374191284,
      "learning_rate": 1.7406440382941688e-08,
      "loss": 0.691,
      "step": 24046
    },
    {
      "epoch": 0.996601599734759,
      "grad_norm": 0.3964201509952545,
      "learning_rate": 1.7199220854573335e-08,
      "loss": 0.6453,
      "step": 24047
    },
    {
      "epoch": 0.9966430436404327,
      "grad_norm": 0.47848013043403625,
      "learning_rate": 1.6992001326204982e-08,
      "loss": 0.7058,
      "step": 24048
    },
    {
      "epoch": 0.9966844875461064,
      "grad_norm": 0.45621258020401,
      "learning_rate": 1.678478179783663e-08,
      "loss": 0.6702,
      "step": 24049
    },
    {
      "epoch": 0.99672593145178,
      "grad_norm": 0.3942944407463074,
      "learning_rate": 1.6577562269468276e-08,
      "loss": 0.6686,
      "step": 24050
    },
    {
      "epoch": 0.9967673753574536,
      "grad_norm": 0.40179166197776794,
      "learning_rate": 1.6370342741099923e-08,
      "loss": 0.6965,
      "step": 24051
    },
    {
      "epoch": 0.9968088192631274,
      "grad_norm": 0.4035951793193817,
      "learning_rate": 1.6163123212731567e-08,
      "loss": 0.6327,
      "step": 24052
    },
    {
      "epoch": 0.996850263168801,
      "grad_norm": 0.42603084444999695,
      "learning_rate": 1.5955903684363214e-08,
      "loss": 0.6746,
      "step": 24053
    },
    {
      "epoch": 0.9968917070744747,
      "grad_norm": 0.4199628233909607,
      "learning_rate": 1.574868415599486e-08,
      "loss": 0.644,
      "step": 24054
    },
    {
      "epoch": 0.9969331509801483,
      "grad_norm": 0.38900476694107056,
      "learning_rate": 1.5541464627626508e-08,
      "loss": 0.6449,
      "step": 24055
    },
    {
      "epoch": 0.9969745948858221,
      "grad_norm": 0.4637569487094879,
      "learning_rate": 1.5334245099258155e-08,
      "loss": 0.6786,
      "step": 24056
    },
    {
      "epoch": 0.9970160387914957,
      "grad_norm": 0.39429667592048645,
      "learning_rate": 1.5127025570889802e-08,
      "loss": 0.7134,
      "step": 24057
    },
    {
      "epoch": 0.9970574826971694,
      "grad_norm": 0.4038718640804291,
      "learning_rate": 1.491980604252145e-08,
      "loss": 0.6216,
      "step": 24058
    },
    {
      "epoch": 0.997098926602843,
      "grad_norm": 0.4547617733478546,
      "learning_rate": 1.4712586514153095e-08,
      "loss": 0.6256,
      "step": 24059
    },
    {
      "epoch": 0.9971403705085167,
      "grad_norm": 0.37967246770858765,
      "learning_rate": 1.4505366985784742e-08,
      "loss": 0.6302,
      "step": 24060
    },
    {
      "epoch": 0.9971818144141904,
      "grad_norm": 0.4694235920906067,
      "learning_rate": 1.4298147457416389e-08,
      "loss": 0.7017,
      "step": 24061
    },
    {
      "epoch": 0.997223258319864,
      "grad_norm": 0.4229969084262848,
      "learning_rate": 1.4090927929048034e-08,
      "loss": 0.699,
      "step": 24062
    },
    {
      "epoch": 0.9972647022255378,
      "grad_norm": 0.42660489678382874,
      "learning_rate": 1.3883708400679681e-08,
      "loss": 0.6633,
      "step": 24063
    },
    {
      "epoch": 0.9973061461312114,
      "grad_norm": 0.45063167810440063,
      "learning_rate": 1.3676488872311328e-08,
      "loss": 0.64,
      "step": 24064
    },
    {
      "epoch": 0.9973475900368851,
      "grad_norm": 0.39679086208343506,
      "learning_rate": 1.3469269343942975e-08,
      "loss": 0.644,
      "step": 24065
    },
    {
      "epoch": 0.9973890339425587,
      "grad_norm": 0.42337507009506226,
      "learning_rate": 1.326204981557462e-08,
      "loss": 0.6392,
      "step": 24066
    },
    {
      "epoch": 0.9974304778482325,
      "grad_norm": 0.41460925340652466,
      "learning_rate": 1.3054830287206268e-08,
      "loss": 0.6244,
      "step": 24067
    },
    {
      "epoch": 0.9974719217539061,
      "grad_norm": 0.4437163770198822,
      "learning_rate": 1.2847610758837915e-08,
      "loss": 0.6842,
      "step": 24068
    },
    {
      "epoch": 0.9975133656595797,
      "grad_norm": 0.4410180151462555,
      "learning_rate": 1.264039123046956e-08,
      "loss": 0.7001,
      "step": 24069
    },
    {
      "epoch": 0.9975548095652534,
      "grad_norm": 0.39628365635871887,
      "learning_rate": 1.2433171702101207e-08,
      "loss": 0.6118,
      "step": 24070
    },
    {
      "epoch": 0.9975962534709271,
      "grad_norm": 0.40094804763793945,
      "learning_rate": 1.2225952173732854e-08,
      "loss": 0.6682,
      "step": 24071
    },
    {
      "epoch": 0.9976376973766008,
      "grad_norm": 0.3953889012336731,
      "learning_rate": 1.20187326453645e-08,
      "loss": 0.623,
      "step": 24072
    },
    {
      "epoch": 0.9976791412822744,
      "grad_norm": 0.49310731887817383,
      "learning_rate": 1.1811513116996147e-08,
      "loss": 0.6781,
      "step": 24073
    },
    {
      "epoch": 0.9977205851879482,
      "grad_norm": 0.4177588224411011,
      "learning_rate": 1.1604293588627794e-08,
      "loss": 0.6173,
      "step": 24074
    },
    {
      "epoch": 0.9977620290936218,
      "grad_norm": 0.4271541237831116,
      "learning_rate": 1.139707406025944e-08,
      "loss": 0.6985,
      "step": 24075
    },
    {
      "epoch": 0.9978034729992955,
      "grad_norm": 0.4241659641265869,
      "learning_rate": 1.1189854531891086e-08,
      "loss": 0.6936,
      "step": 24076
    },
    {
      "epoch": 0.9978449169049691,
      "grad_norm": 0.41232186555862427,
      "learning_rate": 1.0982635003522733e-08,
      "loss": 0.6555,
      "step": 24077
    },
    {
      "epoch": 0.9978863608106427,
      "grad_norm": 0.40220117568969727,
      "learning_rate": 1.077541547515438e-08,
      "loss": 0.6223,
      "step": 24078
    },
    {
      "epoch": 0.9979278047163165,
      "grad_norm": 0.45000794529914856,
      "learning_rate": 1.0568195946786026e-08,
      "loss": 0.649,
      "step": 24079
    },
    {
      "epoch": 0.9979692486219901,
      "grad_norm": 0.417976975440979,
      "learning_rate": 1.0360976418417673e-08,
      "loss": 0.7122,
      "step": 24080
    },
    {
      "epoch": 0.9980106925276638,
      "grad_norm": 0.41776904463768005,
      "learning_rate": 1.015375689004932e-08,
      "loss": 0.6847,
      "step": 24081
    },
    {
      "epoch": 0.9980521364333375,
      "grad_norm": 0.4293898344039917,
      "learning_rate": 9.946537361680965e-09,
      "loss": 0.6667,
      "step": 24082
    },
    {
      "epoch": 0.9980935803390112,
      "grad_norm": 0.41448596119880676,
      "learning_rate": 9.739317833312612e-09,
      "loss": 0.6235,
      "step": 24083
    },
    {
      "epoch": 0.9981350242446848,
      "grad_norm": 0.40281033515930176,
      "learning_rate": 9.53209830494426e-09,
      "loss": 0.6519,
      "step": 24084
    },
    {
      "epoch": 0.9981764681503585,
      "grad_norm": 0.5352064371109009,
      "learning_rate": 9.324878776575905e-09,
      "loss": 0.645,
      "step": 24085
    },
    {
      "epoch": 0.9982179120560322,
      "grad_norm": 0.4493749737739563,
      "learning_rate": 9.117659248207552e-09,
      "loss": 0.6775,
      "step": 24086
    },
    {
      "epoch": 0.9982593559617058,
      "grad_norm": 0.43947115540504456,
      "learning_rate": 8.910439719839199e-09,
      "loss": 0.65,
      "step": 24087
    },
    {
      "epoch": 0.9983007998673795,
      "grad_norm": 0.4529394805431366,
      "learning_rate": 8.703220191470844e-09,
      "loss": 0.6775,
      "step": 24088
    },
    {
      "epoch": 0.9983422437730531,
      "grad_norm": 0.4309327304363251,
      "learning_rate": 8.496000663102491e-09,
      "loss": 0.6674,
      "step": 24089
    },
    {
      "epoch": 0.9983836876787269,
      "grad_norm": 0.42833074927330017,
      "learning_rate": 8.288781134734138e-09,
      "loss": 0.7041,
      "step": 24090
    },
    {
      "epoch": 0.9984251315844005,
      "grad_norm": 0.42104312777519226,
      "learning_rate": 8.081561606365784e-09,
      "loss": 0.6232,
      "step": 24091
    },
    {
      "epoch": 0.9984665754900742,
      "grad_norm": 0.47261059284210205,
      "learning_rate": 7.87434207799743e-09,
      "loss": 0.6831,
      "step": 24092
    },
    {
      "epoch": 0.9985080193957478,
      "grad_norm": 0.43331441283226013,
      "learning_rate": 7.667122549629078e-09,
      "loss": 0.6389,
      "step": 24093
    },
    {
      "epoch": 0.9985494633014216,
      "grad_norm": 0.41058290004730225,
      "learning_rate": 7.459903021260725e-09,
      "loss": 0.6838,
      "step": 24094
    },
    {
      "epoch": 0.9985909072070952,
      "grad_norm": 0.388836145401001,
      "learning_rate": 7.252683492892371e-09,
      "loss": 0.6003,
      "step": 24095
    },
    {
      "epoch": 0.9986323511127688,
      "grad_norm": 0.4189140200614929,
      "learning_rate": 7.045463964524017e-09,
      "loss": 0.6917,
      "step": 24096
    },
    {
      "epoch": 0.9986737950184426,
      "grad_norm": 0.3993547558784485,
      "learning_rate": 6.838244436155664e-09,
      "loss": 0.7163,
      "step": 24097
    },
    {
      "epoch": 0.9987152389241162,
      "grad_norm": 0.40462109446525574,
      "learning_rate": 6.63102490778731e-09,
      "loss": 0.6322,
      "step": 24098
    },
    {
      "epoch": 0.9987566828297899,
      "grad_norm": 0.41604816913604736,
      "learning_rate": 6.423805379418957e-09,
      "loss": 0.6713,
      "step": 24099
    },
    {
      "epoch": 0.9987981267354635,
      "grad_norm": 0.44099119305610657,
      "learning_rate": 6.216585851050604e-09,
      "loss": 0.6675,
      "step": 24100
    },
    {
      "epoch": 0.9988395706411373,
      "grad_norm": 0.4236350357532501,
      "learning_rate": 6.00936632268225e-09,
      "loss": 0.6332,
      "step": 24101
    },
    {
      "epoch": 0.9988810145468109,
      "grad_norm": 0.43383583426475525,
      "learning_rate": 5.802146794313897e-09,
      "loss": 0.6389,
      "step": 24102
    },
    {
      "epoch": 0.9989224584524845,
      "grad_norm": 0.3844117522239685,
      "learning_rate": 5.594927265945543e-09,
      "loss": 0.6332,
      "step": 24103
    },
    {
      "epoch": 0.9989639023581582,
      "grad_norm": 0.43639302253723145,
      "learning_rate": 5.38770773757719e-09,
      "loss": 0.6434,
      "step": 24104
    },
    {
      "epoch": 0.9990053462638319,
      "grad_norm": 0.4110928475856781,
      "learning_rate": 5.180488209208836e-09,
      "loss": 0.6632,
      "step": 24105
    },
    {
      "epoch": 0.9990467901695056,
      "grad_norm": 0.41302281618118286,
      "learning_rate": 4.9732686808404825e-09,
      "loss": 0.7277,
      "step": 24106
    },
    {
      "epoch": 0.9990882340751792,
      "grad_norm": 0.4217560291290283,
      "learning_rate": 4.76604915247213e-09,
      "loss": 0.6917,
      "step": 24107
    },
    {
      "epoch": 0.999129677980853,
      "grad_norm": 0.39875221252441406,
      "learning_rate": 4.558829624103776e-09,
      "loss": 0.6223,
      "step": 24108
    },
    {
      "epoch": 0.9991711218865266,
      "grad_norm": 0.4087948501110077,
      "learning_rate": 4.351610095735422e-09,
      "loss": 0.6492,
      "step": 24109
    },
    {
      "epoch": 0.9992125657922003,
      "grad_norm": 0.4066776633262634,
      "learning_rate": 4.144390567367069e-09,
      "loss": 0.6851,
      "step": 24110
    },
    {
      "epoch": 0.9992540096978739,
      "grad_norm": 0.4268956780433655,
      "learning_rate": 3.937171038998715e-09,
      "loss": 0.703,
      "step": 24111
    },
    {
      "epoch": 0.9992954536035475,
      "grad_norm": 0.4181530177593231,
      "learning_rate": 3.729951510630362e-09,
      "loss": 0.6836,
      "step": 24112
    },
    {
      "epoch": 0.9993368975092213,
      "grad_norm": 0.40263479948043823,
      "learning_rate": 3.5227319822620085e-09,
      "loss": 0.655,
      "step": 24113
    },
    {
      "epoch": 0.9993783414148949,
      "grad_norm": 0.4511946439743042,
      "learning_rate": 3.315512453893655e-09,
      "loss": 0.6863,
      "step": 24114
    },
    {
      "epoch": 0.9994197853205686,
      "grad_norm": 0.4255143702030182,
      "learning_rate": 3.108292925525302e-09,
      "loss": 0.7095,
      "step": 24115
    },
    {
      "epoch": 0.9994612292262423,
      "grad_norm": 0.38710400462150574,
      "learning_rate": 2.9010733971569484e-09,
      "loss": 0.6963,
      "step": 24116
    },
    {
      "epoch": 0.999502673131916,
      "grad_norm": 0.45806679129600525,
      "learning_rate": 2.693853868788595e-09,
      "loss": 0.7203,
      "step": 24117
    },
    {
      "epoch": 0.9995441170375896,
      "grad_norm": 0.4411233961582184,
      "learning_rate": 2.4866343404202413e-09,
      "loss": 0.6963,
      "step": 24118
    },
    {
      "epoch": 0.9995855609432633,
      "grad_norm": 0.47997546195983887,
      "learning_rate": 2.279414812051888e-09,
      "loss": 0.6624,
      "step": 24119
    },
    {
      "epoch": 0.999627004848937,
      "grad_norm": 0.40546709299087524,
      "learning_rate": 2.0721952836835345e-09,
      "loss": 0.6761,
      "step": 24120
    },
    {
      "epoch": 0.9996684487546106,
      "grad_norm": 0.4429965317249298,
      "learning_rate": 1.864975755315181e-09,
      "loss": 0.67,
      "step": 24121
    },
    {
      "epoch": 0.9997098926602843,
      "grad_norm": 0.4323263168334961,
      "learning_rate": 1.6577562269468276e-09,
      "loss": 0.6592,
      "step": 24122
    },
    {
      "epoch": 0.9997513365659579,
      "grad_norm": 0.42195117473602295,
      "learning_rate": 1.4505366985784742e-09,
      "loss": 0.6602,
      "step": 24123
    },
    {
      "epoch": 0.9997927804716317,
      "grad_norm": 0.4603728652000427,
      "learning_rate": 1.2433171702101206e-09,
      "loss": 0.687,
      "step": 24124
    },
    {
      "epoch": 0.9998342243773053,
      "grad_norm": 0.41849106550216675,
      "learning_rate": 1.0360976418417673e-09,
      "loss": 0.6497,
      "step": 24125
    },
    {
      "epoch": 0.999875668282979,
      "grad_norm": 0.4541945457458496,
      "learning_rate": 8.288781134734138e-10,
      "loss": 0.7271,
      "step": 24126
    },
    {
      "epoch": 0.9999171121886526,
      "grad_norm": 0.4299638271331787,
      "learning_rate": 6.216585851050603e-10,
      "loss": 0.6252,
      "step": 24127
    },
    {
      "epoch": 0.9999585560943264,
      "grad_norm": 0.4042789340019226,
      "learning_rate": 4.144390567367069e-10,
      "loss": 0.6179,
      "step": 24128
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42805585265159607,
      "learning_rate": 2.0721952836835345e-10,
      "loss": 0.5836,
      "step": 24129
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 24129,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 8043,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5173164088108148e+20,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}