{
  "best_metric": 0.8249050225524125,
  "best_model_checkpoint": "/data/hungnm/unisentiment/roberta-large-sentiment/checkpoint-12296",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 15370,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016265452179570592,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.2297,
      "step": 5
    },
    {
      "epoch": 0.0032530904359141183,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.2375,
      "step": 10
    },
    {
      "epoch": 0.004879635653871178,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.2213,
      "step": 15
    },
    {
      "epoch": 0.006506180871828237,
      "grad_norm": 3.86765718460083,
      "learning_rate": 9.74025974025974e-07,
      "loss": 2.2313,
      "step": 20
    },
    {
      "epoch": 0.008132726089785295,
      "grad_norm": 1.735704779624939,
      "learning_rate": 2.5974025974025976e-06,
      "loss": 2.2234,
      "step": 25
    },
    {
      "epoch": 0.009759271307742356,
      "grad_norm": 1.5706150531768799,
      "learning_rate": 4.220779220779221e-06,
      "loss": 2.1965,
      "step": 30
    },
    {
      "epoch": 0.011385816525699415,
      "grad_norm": 2.025402069091797,
      "learning_rate": 5.844155844155844e-06,
      "loss": 2.1842,
      "step": 35
    },
    {
      "epoch": 0.013012361743656473,
      "grad_norm": 2.8359689712524414,
      "learning_rate": 7.467532467532468e-06,
      "loss": 2.1539,
      "step": 40
    },
    {
      "epoch": 0.014638906961613532,
      "grad_norm": 3.062162399291992,
      "learning_rate": 9.090909090909091e-06,
      "loss": 2.1127,
      "step": 45
    },
    {
      "epoch": 0.01626545217957059,
      "grad_norm": 4.7635908126831055,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 2.0376,
      "step": 50
    },
    {
      "epoch": 0.017891997397527653,
      "grad_norm": 5.502624988555908,
      "learning_rate": 1.2337662337662339e-05,
      "loss": 1.853,
      "step": 55
    },
    {
      "epoch": 0.01951854261548471,
      "grad_norm": 8.860891342163086,
      "learning_rate": 1.396103896103896e-05,
      "loss": 1.6277,
      "step": 60
    },
    {
      "epoch": 0.02114508783344177,
      "grad_norm": 13.536232948303223,
      "learning_rate": 1.5584415584415583e-05,
      "loss": 1.4906,
      "step": 65
    },
    {
      "epoch": 0.02277163305139883,
      "grad_norm": 4.987203598022461,
      "learning_rate": 1.7207792207792208e-05,
      "loss": 1.4451,
      "step": 70
    },
    {
      "epoch": 0.024398178269355888,
      "grad_norm": 6.489861488342285,
      "learning_rate": 1.8831168831168833e-05,
      "loss": 1.3596,
      "step": 75
    },
    {
      "epoch": 0.026024723487312947,
      "grad_norm": 6.009546756744385,
      "learning_rate": 2.0454545454545457e-05,
      "loss": 1.3021,
      "step": 80
    },
    {
      "epoch": 0.027651268705270005,
      "grad_norm": 5.856492042541504,
      "learning_rate": 2.207792207792208e-05,
      "loss": 1.2303,
      "step": 85
    },
    {
      "epoch": 0.029277813923227064,
      "grad_norm": 8.079117774963379,
      "learning_rate": 2.3701298701298703e-05,
      "loss": 1.2756,
      "step": 90
    },
    {
      "epoch": 0.030904359141184126,
      "grad_norm": 9.356736183166504,
      "learning_rate": 2.5324675324675325e-05,
      "loss": 1.2231,
      "step": 95
    },
    {
      "epoch": 0.03253090435914118,
      "grad_norm": 7.978050708770752,
      "learning_rate": 2.694805194805195e-05,
      "loss": 1.2171,
      "step": 100
    },
    {
      "epoch": 0.034157449577098244,
      "grad_norm": 9.788860321044922,
      "learning_rate": 2.857142857142857e-05,
      "loss": 1.3057,
      "step": 105
    },
    {
      "epoch": 0.035783994795055306,
      "grad_norm": 6.810769081115723,
      "learning_rate": 3.01948051948052e-05,
      "loss": 1.2105,
      "step": 110
    },
    {
      "epoch": 0.03741054001301236,
      "grad_norm": 7.437067985534668,
      "learning_rate": 3.181818181818182e-05,
      "loss": 1.2065,
      "step": 115
    },
    {
      "epoch": 0.03903708523096942,
      "grad_norm": 8.67028522491455,
      "learning_rate": 3.344155844155844e-05,
      "loss": 1.1346,
      "step": 120
    },
    {
      "epoch": 0.04066363044892648,
      "grad_norm": 9.523406982421875,
      "learning_rate": 3.506493506493507e-05,
      "loss": 1.1662,
      "step": 125
    },
    {
      "epoch": 0.04229017566688354,
      "grad_norm": 4.4032816886901855,
      "learning_rate": 3.668831168831169e-05,
      "loss": 1.1708,
      "step": 130
    },
    {
      "epoch": 0.043916720884840596,
      "grad_norm": 5.519209861755371,
      "learning_rate": 3.831168831168831e-05,
      "loss": 1.1403,
      "step": 135
    },
    {
      "epoch": 0.04554326610279766,
      "grad_norm": 5.634235382080078,
      "learning_rate": 3.993506493506494e-05,
      "loss": 1.1536,
      "step": 140
    },
    {
      "epoch": 0.04716981132075472,
      "grad_norm": 6.300098419189453,
      "learning_rate": 4.155844155844156e-05,
      "loss": 1.1277,
      "step": 145
    },
    {
      "epoch": 0.048796356538711776,
      "grad_norm": 4.605157375335693,
      "learning_rate": 4.318181818181819e-05,
      "loss": 1.1163,
      "step": 150
    },
    {
      "epoch": 0.05042290175666884,
      "grad_norm": 5.781686305999756,
      "learning_rate": 4.4805194805194805e-05,
      "loss": 1.1107,
      "step": 155
    },
    {
      "epoch": 0.05204944697462589,
      "grad_norm": 4.554281234741211,
      "learning_rate": 4.642857142857143e-05,
      "loss": 1.1387,
      "step": 160
    },
    {
      "epoch": 0.053675992192582955,
      "grad_norm": 10.010477066040039,
      "learning_rate": 4.8051948051948054e-05,
      "loss": 1.1531,
      "step": 165
    },
    {
      "epoch": 0.05530253741054001,
      "grad_norm": 6.158892631530762,
      "learning_rate": 4.967532467532468e-05,
      "loss": 1.1396,
      "step": 170
    },
    {
      "epoch": 0.05692908262849707,
      "grad_norm": 6.746368408203125,
      "learning_rate": 4.999999520430831e-05,
      "loss": 1.1243,
      "step": 175
    },
    {
      "epoch": 0.05855562784645413,
      "grad_norm": 7.575245380401611,
      "learning_rate": 4.999997389012675e-05,
      "loss": 1.1174,
      "step": 180
    },
    {
      "epoch": 0.06018217306441119,
      "grad_norm": 9.75899887084961,
      "learning_rate": 4.99999232689698e-05,
      "loss": 1.123,
      "step": 185
    },
    {
      "epoch": 0.06180871828236825,
      "grad_norm": 6.071502685546875,
      "learning_rate": 4.9999846005164544e-05,
      "loss": 1.0794,
      "step": 190
    },
    {
      "epoch": 0.06343526350032531,
      "grad_norm": 4.638003826141357,
      "learning_rate": 4.999974209879331e-05,
      "loss": 1.1059,
      "step": 195
    },
    {
      "epoch": 0.06506180871828236,
      "grad_norm": 4.056344985961914,
      "learning_rate": 4.999961154996685e-05,
      "loss": 1.0373,
      "step": 200
    },
    {
      "epoch": 0.06668835393623943,
      "grad_norm": 10.81039810180664,
      "learning_rate": 4.999945435882428e-05,
      "loss": 1.0483,
      "step": 205
    },
    {
      "epoch": 0.06831489915419649,
      "grad_norm": 9.315641403198242,
      "learning_rate": 4.999927052553313e-05,
      "loss": 1.0723,
      "step": 210
    },
    {
      "epoch": 0.06994144437215355,
      "grad_norm": 8.322559356689453,
      "learning_rate": 4.9999060050289286e-05,
      "loss": 1.1134,
      "step": 215
    },
    {
      "epoch": 0.07156798959011061,
      "grad_norm": 5.152672290802002,
      "learning_rate": 4.999882293331708e-05,
      "loss": 1.0625,
      "step": 220
    },
    {
      "epoch": 0.07319453480806766,
      "grad_norm": 3.750262498855591,
      "learning_rate": 4.999855917486921e-05,
      "loss": 1.025,
      "step": 225
    },
    {
      "epoch": 0.07482108002602472,
      "grad_norm": 4.812891483306885,
      "learning_rate": 4.999826877522675e-05,
      "loss": 1.0774,
      "step": 230
    },
    {
      "epoch": 0.07644762524398178,
      "grad_norm": 5.638612747192383,
      "learning_rate": 4.999795173469919e-05,
      "loss": 1.026,
      "step": 235
    },
    {
      "epoch": 0.07807417046193885,
      "grad_norm": 4.525393486022949,
      "learning_rate": 4.99976080536244e-05,
      "loss": 1.0333,
      "step": 240
    },
    {
      "epoch": 0.0797007156798959,
      "grad_norm": 14.256878852844238,
      "learning_rate": 4.9997237732368645e-05,
      "loss": 1.0749,
      "step": 245
    },
    {
      "epoch": 0.08132726089785296,
      "grad_norm": 7.804468154907227,
      "learning_rate": 4.9996840771326584e-05,
      "loss": 1.0461,
      "step": 250
    },
    {
      "epoch": 0.08295380611581002,
      "grad_norm": 3.510770559310913,
      "learning_rate": 4.999641717092126e-05,
      "loss": 1.0215,
      "step": 255
    },
    {
      "epoch": 0.08458035133376708,
      "grad_norm": 6.548313140869141,
      "learning_rate": 4.99959669316041e-05,
      "loss": 0.9953,
      "step": 260
    },
    {
      "epoch": 0.08620689655172414,
      "grad_norm": 3.891099691390991,
      "learning_rate": 4.999549005385494e-05,
      "loss": 1.0075,
      "step": 265
    },
    {
      "epoch": 0.08783344176968119,
      "grad_norm": 3.341900587081909,
      "learning_rate": 4.999498653818199e-05,
      "loss": 1.0451,
      "step": 270
    },
    {
      "epoch": 0.08945998698763825,
      "grad_norm": 3.6907620429992676,
      "learning_rate": 4.999445638512185e-05,
      "loss": 1.0313,
      "step": 275
    },
    {
      "epoch": 0.09108653220559532,
      "grad_norm": 4.23262357711792,
      "learning_rate": 4.99938995952395e-05,
      "loss": 1.0493,
      "step": 280
    },
    {
      "epoch": 0.09271307742355238,
      "grad_norm": 3.793381929397583,
      "learning_rate": 4.9993316169128334e-05,
      "loss": 1.0103,
      "step": 285
    },
    {
      "epoch": 0.09433962264150944,
      "grad_norm": 5.088762283325195,
      "learning_rate": 4.99927061074101e-05,
      "loss": 1.0206,
      "step": 290
    },
    {
      "epoch": 0.09596616785946649,
      "grad_norm": 4.209216594696045,
      "learning_rate": 4.999206941073496e-05,
      "loss": 1.0018,
      "step": 295
    },
    {
      "epoch": 0.09759271307742355,
      "grad_norm": 3.5634896755218506,
      "learning_rate": 4.9991406079781424e-05,
      "loss": 0.9936,
      "step": 300
    },
    {
      "epoch": 0.09921925829538061,
      "grad_norm": 6.785153388977051,
      "learning_rate": 4.999071611525643e-05,
      "loss": 1.0299,
      "step": 305
    },
    {
      "epoch": 0.10084580351333768,
      "grad_norm": 4.595156192779541,
      "learning_rate": 4.998999951789528e-05,
      "loss": 1.0238,
      "step": 310
    },
    {
      "epoch": 0.10247234873129472,
      "grad_norm": 4.024753570556641,
      "learning_rate": 4.998925628846164e-05,
      "loss": 1.0202,
      "step": 315
    },
    {
      "epoch": 0.10409889394925179,
      "grad_norm": 4.274604320526123,
      "learning_rate": 4.9988486427747606e-05,
      "loss": 1.0169,
      "step": 320
    },
    {
      "epoch": 0.10572543916720885,
      "grad_norm": 5.742299556732178,
      "learning_rate": 4.99876899365736e-05,
      "loss": 0.9837,
      "step": 325
    },
    {
      "epoch": 0.10735198438516591,
      "grad_norm": 4.468045711517334,
      "learning_rate": 4.998686681578846e-05,
      "loss": 1.0671,
      "step": 330
    },
    {
      "epoch": 0.10897852960312297,
      "grad_norm": 3.6818110942840576,
      "learning_rate": 4.998601706626938e-05,
      "loss": 1.0505,
      "step": 335
    },
    {
      "epoch": 0.11060507482108002,
      "grad_norm": 2.8949925899505615,
      "learning_rate": 4.9985140688921975e-05,
      "loss": 1.02,
      "step": 340
    },
    {
      "epoch": 0.11223162003903708,
      "grad_norm": 3.744718551635742,
      "learning_rate": 4.9984237684680194e-05,
      "loss": 1.0275,
      "step": 345
    },
    {
      "epoch": 0.11385816525699415,
      "grad_norm": 6.084460735321045,
      "learning_rate": 4.998330805450636e-05,
      "loss": 0.9692,
      "step": 350
    },
    {
      "epoch": 0.11548471047495121,
      "grad_norm": 3.124788522720337,
      "learning_rate": 4.998235179939122e-05,
      "loss": 0.9594,
      "step": 355
    },
    {
      "epoch": 0.11711125569290826,
      "grad_norm": 3.5672078132629395,
      "learning_rate": 4.998136892035382e-05,
      "loss": 0.9882,
      "step": 360
    },
    {
      "epoch": 0.11873780091086532,
      "grad_norm": 3.500018835067749,
      "learning_rate": 4.998035941844167e-05,
      "loss": 0.9527,
      "step": 365
    },
    {
      "epoch": 0.12036434612882238,
      "grad_norm": 3.729696035385132,
      "learning_rate": 4.997932329473058e-05,
      "loss": 1.0004,
      "step": 370
    },
    {
      "epoch": 0.12199089134677944,
      "grad_norm": 4.352784156799316,
      "learning_rate": 4.997826055032476e-05,
      "loss": 0.9475,
      "step": 375
    },
    {
      "epoch": 0.1236174365647365,
      "grad_norm": 5.5154523849487305,
      "learning_rate": 4.99771711863568e-05,
      "loss": 0.9892,
      "step": 380
    },
    {
      "epoch": 0.12524398178269355,
      "grad_norm": 5.252265930175781,
      "learning_rate": 4.997605520398762e-05,
      "loss": 0.9563,
      "step": 385
    },
    {
      "epoch": 0.12687052700065063,
      "grad_norm": 3.895646810531616,
      "learning_rate": 4.9974912604406554e-05,
      "loss": 1.0064,
      "step": 390
    },
    {
      "epoch": 0.12849707221860768,
      "grad_norm": 43.6920166015625,
      "learning_rate": 4.997374338883127e-05,
      "loss": 0.9715,
      "step": 395
    },
    {
      "epoch": 0.13012361743656473,
      "grad_norm": 3.9318182468414307,
      "learning_rate": 4.9972547558507815e-05,
      "loss": 0.9883,
      "step": 400
    },
    {
      "epoch": 0.1317501626545218,
      "grad_norm": 9.373114585876465,
      "learning_rate": 4.99713251147106e-05,
      "loss": 0.9842,
      "step": 405
    },
    {
      "epoch": 0.13337670787247885,
      "grad_norm": 4.192168712615967,
      "learning_rate": 4.997007605874239e-05,
      "loss": 0.986,
      "step": 410
    },
    {
      "epoch": 0.13500325309043593,
      "grad_norm": 7.416944980621338,
      "learning_rate": 4.996880039193431e-05,
      "loss": 0.965,
      "step": 415
    },
    {
      "epoch": 0.13662979830839297,
      "grad_norm": 3.9326727390289307,
      "learning_rate": 4.996749811564586e-05,
      "loss": 0.9931,
      "step": 420
    },
    {
      "epoch": 0.13825634352635002,
      "grad_norm": 5.82494592666626,
      "learning_rate": 4.996616923126488e-05,
      "loss": 0.9522,
      "step": 425
    },
    {
      "epoch": 0.1398828887443071,
      "grad_norm": 2.137333869934082,
      "learning_rate": 4.996481374020759e-05,
      "loss": 0.9943,
      "step": 430
    },
    {
      "epoch": 0.14150943396226415,
      "grad_norm": 3.4468958377838135,
      "learning_rate": 4.996343164391853e-05,
      "loss": 0.9767,
      "step": 435
    },
    {
      "epoch": 0.14313597918022122,
      "grad_norm": 19.23995590209961,
      "learning_rate": 4.9962022943870626e-05,
      "loss": 0.9562,
      "step": 440
    },
    {
      "epoch": 0.14476252439817827,
      "grad_norm": 3.0280990600585938,
      "learning_rate": 4.9960587641565125e-05,
      "loss": 0.981,
      "step": 445
    },
    {
      "epoch": 0.14638906961613532,
      "grad_norm": 2.9482223987579346,
      "learning_rate": 4.995912573853166e-05,
      "loss": 0.9449,
      "step": 450
    },
    {
      "epoch": 0.1480156148340924,
      "grad_norm": 5.7567243576049805,
      "learning_rate": 4.9957637236328195e-05,
      "loss": 0.9919,
      "step": 455
    },
    {
      "epoch": 0.14964216005204944,
      "grad_norm": 3.0479254722595215,
      "learning_rate": 4.995612213654103e-05,
      "loss": 0.9837,
      "step": 460
    },
    {
      "epoch": 0.1512687052700065,
      "grad_norm": 3.3615362644195557,
      "learning_rate": 4.995458044078482e-05,
      "loss": 0.9584,
      "step": 465
    },
    {
      "epoch": 0.15289525048796357,
      "grad_norm": 22.171850204467773,
      "learning_rate": 4.995301215070257e-05,
      "loss": 0.9984,
      "step": 470
    },
    {
      "epoch": 0.15452179570592062,
      "grad_norm": 4.525479316711426,
      "learning_rate": 4.9951417267965626e-05,
      "loss": 1.0301,
      "step": 475
    },
    {
      "epoch": 0.1561483409238777,
      "grad_norm": 5.372246265411377,
      "learning_rate": 4.9949795794273664e-05,
      "loss": 0.9911,
      "step": 480
    },
    {
      "epoch": 0.15777488614183474,
      "grad_norm": 2.3989129066467285,
      "learning_rate": 4.99481477313547e-05,
      "loss": 0.96,
      "step": 485
    },
    {
      "epoch": 0.1594014313597918,
      "grad_norm": 3.0757484436035156,
      "learning_rate": 4.994647308096509e-05,
      "loss": 0.9738,
      "step": 490
    },
    {
      "epoch": 0.16102797657774887,
      "grad_norm": 3.538886308670044,
      "learning_rate": 4.9944771844889524e-05,
      "loss": 0.9492,
      "step": 495
    },
    {
      "epoch": 0.16265452179570591,
      "grad_norm": 3.5516517162323,
      "learning_rate": 4.994304402494104e-05,
      "loss": 0.9463,
      "step": 500
    },
    {
      "epoch": 0.164281067013663,
      "grad_norm": 4.078283786773682,
      "learning_rate": 4.994128962296097e-05,
      "loss": 0.9378,
      "step": 505
    },
    {
      "epoch": 0.16590761223162004,
      "grad_norm": 2.795860767364502,
      "learning_rate": 4.993950864081901e-05,
      "loss": 0.9979,
      "step": 510
    },
    {
      "epoch": 0.1675341574495771,
      "grad_norm": 2.6040124893188477,
      "learning_rate": 4.9937701080413165e-05,
      "loss": 0.97,
      "step": 515
    },
    {
      "epoch": 0.16916070266753416,
      "grad_norm": 3.229245185852051,
      "learning_rate": 4.993586694366977e-05,
      "loss": 0.9673,
      "step": 520
    },
    {
      "epoch": 0.1707872478854912,
      "grad_norm": 4.013242244720459,
      "learning_rate": 4.993400623254347e-05,
      "loss": 0.9578,
      "step": 525
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 2.899711847305298,
      "learning_rate": 4.993249853141837e-05,
      "loss": 0.959,
      "step": 530
    },
    {
      "epoch": 0.17404033832140534,
      "grad_norm": 3.2729837894439697,
      "learning_rate": 4.9930589991419e-05,
      "loss": 0.9437,
      "step": 535
    },
    {
      "epoch": 0.17566688353936238,
      "grad_norm": 3.1390457153320312,
      "learning_rate": 4.992865488266043e-05,
      "loss": 0.9274,
      "step": 540
    },
    {
      "epoch": 0.17729342875731946,
      "grad_norm": 2.9977355003356934,
      "learning_rate": 4.9926693207204925e-05,
      "loss": 0.9704,
      "step": 545
    },
    {
      "epoch": 0.1789199739752765,
      "grad_norm": 3.9077603816986084,
      "learning_rate": 4.9924704967143064e-05,
      "loss": 0.9489,
      "step": 550
    },
    {
      "epoch": 0.18054651919323358,
      "grad_norm": 3.813920736312866,
      "learning_rate": 4.992269016459373e-05,
      "loss": 0.9854,
      "step": 555
    },
    {
      "epoch": 0.18217306441119063,
      "grad_norm": 2.9923019409179688,
      "learning_rate": 4.9920648801704103e-05,
      "loss": 0.9514,
      "step": 560
    },
    {
      "epoch": 0.18379960962914768,
      "grad_norm": 3.6704702377319336,
      "learning_rate": 4.991858088064971e-05,
      "loss": 0.9483,
      "step": 565
    },
    {
      "epoch": 0.18542615484710476,
      "grad_norm": 4.187371730804443,
      "learning_rate": 4.991648640363434e-05,
      "loss": 0.9206,
      "step": 570
    },
    {
      "epoch": 0.1870527000650618,
      "grad_norm": 3.446115732192993,
      "learning_rate": 4.991436537289009e-05,
      "loss": 0.9495,
      "step": 575
    },
    {
      "epoch": 0.18867924528301888,
      "grad_norm": 3.385150671005249,
      "learning_rate": 4.9912217790677365e-05,
      "loss": 0.9166,
      "step": 580
    },
    {
      "epoch": 0.19030579050097593,
      "grad_norm": 3.2357215881347656,
      "learning_rate": 4.991004365928487e-05,
      "loss": 0.9566,
      "step": 585
    },
    {
      "epoch": 0.19193233571893298,
      "grad_norm": 5.720197677612305,
      "learning_rate": 4.990784298102959e-05,
      "loss": 0.9416,
      "step": 590
    },
    {
      "epoch": 0.19355888093689005,
      "grad_norm": 3.472399950027466,
      "learning_rate": 4.99056157582568e-05,
      "loss": 0.9407,
      "step": 595
    },
    {
      "epoch": 0.1951854261548471,
      "grad_norm": 3.562246561050415,
      "learning_rate": 4.9903361993340095e-05,
      "loss": 0.9659,
      "step": 600
    },
    {
      "epoch": 0.19681197137280415,
      "grad_norm": 2.6203455924987793,
      "learning_rate": 4.9901081688681314e-05,
      "loss": 0.9354,
      "step": 605
    },
    {
      "epoch": 0.19843851659076123,
      "grad_norm": 2.9699368476867676,
      "learning_rate": 4.989877484671061e-05,
      "loss": 0.9478,
      "step": 610
    },
    {
      "epoch": 0.20006506180871828,
      "grad_norm": 2.1798095703125,
      "learning_rate": 4.989644146988639e-05,
      "loss": 0.9689,
      "step": 615
    },
    {
      "epoch": 0.20169160702667535,
      "grad_norm": 2.909578561782837,
      "learning_rate": 4.989408156069537e-05,
      "loss": 0.9187,
      "step": 620
    },
    {
      "epoch": 0.2033181522446324,
      "grad_norm": 2.7465949058532715,
      "learning_rate": 4.989169512165253e-05,
      "loss": 0.9443,
      "step": 625
    },
    {
      "epoch": 0.20494469746258945,
      "grad_norm": 4.120491981506348,
      "learning_rate": 4.988928215530111e-05,
      "loss": 0.9183,
      "step": 630
    },
    {
      "epoch": 0.20657124268054652,
      "grad_norm": 5.750351428985596,
      "learning_rate": 4.988684266421263e-05,
      "loss": 0.9721,
      "step": 635
    },
    {
      "epoch": 0.20819778789850357,
      "grad_norm": 2.510326862335205,
      "learning_rate": 4.9884376650986874e-05,
      "loss": 0.9456,
      "step": 640
    },
    {
      "epoch": 0.20982433311646065,
      "grad_norm": 3.090721845626831,
      "learning_rate": 4.988188411825191e-05,
      "loss": 0.9333,
      "step": 645
    },
    {
      "epoch": 0.2114508783344177,
      "grad_norm": 2.2550766468048096,
      "learning_rate": 4.987936506866405e-05,
      "loss": 0.9336,
      "step": 650
    },
    {
      "epoch": 0.21307742355237475,
      "grad_norm": 3.066791534423828,
      "learning_rate": 4.987681950490786e-05,
      "loss": 0.9349,
      "step": 655
    },
    {
      "epoch": 0.21470396877033182,
      "grad_norm": 2.512131929397583,
      "learning_rate": 4.987424742969616e-05,
      "loss": 0.9353,
      "step": 660
    },
    {
      "epoch": 0.21633051398828887,
      "grad_norm": 3.988386869430542,
      "learning_rate": 4.987164884577007e-05,
      "loss": 0.9514,
      "step": 665
    },
    {
      "epoch": 0.21795705920624595,
      "grad_norm": 3.2770578861236572,
      "learning_rate": 4.986902375589889e-05,
      "loss": 0.9342,
      "step": 670
    },
    {
      "epoch": 0.219583604424203,
      "grad_norm": 3.2448511123657227,
      "learning_rate": 4.986637216288021e-05,
      "loss": 0.9482,
      "step": 675
    },
    {
      "epoch": 0.22121014964216004,
      "grad_norm": 2.492069721221924,
      "learning_rate": 4.986369406953988e-05,
      "loss": 0.9299,
      "step": 680
    },
    {
      "epoch": 0.22283669486011712,
      "grad_norm": 4.076034069061279,
      "learning_rate": 4.986098947873195e-05,
      "loss": 0.9075,
      "step": 685
    },
    {
      "epoch": 0.22446324007807417,
      "grad_norm": 2.4998836517333984,
      "learning_rate": 4.985825839333871e-05,
      "loss": 0.9472,
      "step": 690
    },
    {
      "epoch": 0.22608978529603124,
      "grad_norm": 7.26285457611084,
      "learning_rate": 4.985550081627074e-05,
      "loss": 0.9604,
      "step": 695
    },
    {
      "epoch": 0.2277163305139883,
      "grad_norm": 3.53657865524292,
      "learning_rate": 4.985271675046679e-05,
      "loss": 0.9694,
      "step": 700
    },
    {
      "epoch": 0.22934287573194534,
      "grad_norm": 2.3773930072784424,
      "learning_rate": 4.984990619889387e-05,
      "loss": 0.906,
      "step": 705
    },
    {
      "epoch": 0.23096942094990242,
      "grad_norm": 3.7686069011688232,
      "learning_rate": 4.984706916454721e-05,
      "loss": 0.9665,
      "step": 710
    },
    {
      "epoch": 0.23259596616785946,
      "grad_norm": 2.416747570037842,
      "learning_rate": 4.984420565045027e-05,
      "loss": 0.947,
      "step": 715
    },
    {
      "epoch": 0.2342225113858165,
      "grad_norm": 3.5830488204956055,
      "learning_rate": 4.984131565965472e-05,
      "loss": 0.9246,
      "step": 720
    },
    {
      "epoch": 0.2358490566037736,
      "grad_norm": 3.2117176055908203,
      "learning_rate": 4.983839919524045e-05,
      "loss": 0.942,
      "step": 725
    },
    {
      "epoch": 0.23747560182173064,
      "grad_norm": 3.1787447929382324,
      "learning_rate": 4.983545626031555e-05,
      "loss": 0.9533,
      "step": 730
    },
    {
      "epoch": 0.2391021470396877,
      "grad_norm": 2.5831379890441895,
      "learning_rate": 4.983248685801636e-05,
      "loss": 0.9443,
      "step": 735
    },
    {
      "epoch": 0.24072869225764476,
      "grad_norm": 7.841883659362793,
      "learning_rate": 4.982949099150738e-05,
      "loss": 0.9042,
      "step": 740
    },
    {
      "epoch": 0.2423552374756018,
      "grad_norm": 2.721116781234741,
      "learning_rate": 4.982646866398133e-05,
      "loss": 0.9232,
      "step": 745
    },
    {
      "epoch": 0.24398178269355889,
      "grad_norm": 2.3474032878875732,
      "learning_rate": 4.982341987865914e-05,
      "loss": 0.9731,
      "step": 750
    },
    {
      "epoch": 0.24560832791151593,
      "grad_norm": 2.8456294536590576,
      "learning_rate": 4.9820344638789926e-05,
      "loss": 0.9234,
      "step": 755
    },
    {
      "epoch": 0.247234873129473,
      "grad_norm": 3.1040987968444824,
      "learning_rate": 4.981724294765101e-05,
      "loss": 0.9508,
      "step": 760
    },
    {
      "epoch": 0.24886141834743006,
      "grad_norm": 4.044954776763916,
      "learning_rate": 4.9814114808547876e-05,
      "loss": 0.9593,
      "step": 765
    },
    {
      "epoch": 0.2504879635653871,
      "grad_norm": 3.3007290363311768,
      "learning_rate": 4.981096022481422e-05,
      "loss": 0.9174,
      "step": 770
    },
    {
      "epoch": 0.2521145087833442,
      "grad_norm": 3.6367475986480713,
      "learning_rate": 4.980777919981191e-05,
      "loss": 0.9472,
      "step": 775
    },
    {
      "epoch": 0.25374105400130126,
      "grad_norm": 18.750322341918945,
      "learning_rate": 4.980457173693099e-05,
      "loss": 0.9101,
      "step": 780
    },
    {
      "epoch": 0.2553675992192583,
      "grad_norm": 9.816338539123535,
      "learning_rate": 4.980133783958969e-05,
      "loss": 0.9439,
      "step": 785
    },
    {
      "epoch": 0.25699414443721535,
      "grad_norm": 2.4749770164489746,
      "learning_rate": 4.9798077511234396e-05,
      "loss": 0.9198,
      "step": 790
    },
    {
      "epoch": 0.25862068965517243,
      "grad_norm": 2.2600533962249756,
      "learning_rate": 4.979479075533967e-05,
      "loss": 0.9095,
      "step": 795
    },
    {
      "epoch": 0.26024723487312945,
      "grad_norm": 3.450012445449829,
      "learning_rate": 4.9791477575408254e-05,
      "loss": 0.9263,
      "step": 800
    },
    {
      "epoch": 0.2618737800910865,
      "grad_norm": 2.614621877670288,
      "learning_rate": 4.9788137974971006e-05,
      "loss": 0.9321,
      "step": 805
    },
    {
      "epoch": 0.2635003253090436,
      "grad_norm": 2.4043772220611572,
      "learning_rate": 4.9784771957586995e-05,
      "loss": 0.9241,
      "step": 810
    },
    {
      "epoch": 0.2651268705270006,
      "grad_norm": 3.927534818649292,
      "learning_rate": 4.97813795268434e-05,
      "loss": 0.9428,
      "step": 815
    },
    {
      "epoch": 0.2667534157449577,
      "grad_norm": 3.2626078128814697,
      "learning_rate": 4.977796068635558e-05,
      "loss": 0.936,
      "step": 820
    },
    {
      "epoch": 0.2683799609629148,
      "grad_norm": 2.7497739791870117,
      "learning_rate": 4.977451543976701e-05,
      "loss": 0.9176,
      "step": 825
    },
    {
      "epoch": 0.27000650618087185,
      "grad_norm": 2.8794078826904297,
      "learning_rate": 4.9771043790749335e-05,
      "loss": 0.9346,
      "step": 830
    },
    {
      "epoch": 0.2716330513988289,
      "grad_norm": 3.0631301403045654,
      "learning_rate": 4.976754574300231e-05,
      "loss": 0.9391,
      "step": 835
    },
    {
      "epoch": 0.27325959661678595,
      "grad_norm": 3.0575807094573975,
      "learning_rate": 4.9764021300253844e-05,
      "loss": 0.9176,
      "step": 840
    },
    {
      "epoch": 0.274886141834743,
      "grad_norm": 2.0632474422454834,
      "learning_rate": 4.976047046625997e-05,
      "loss": 0.9371,
      "step": 845
    },
    {
      "epoch": 0.27651268705270005,
      "grad_norm": 3.8703184127807617,
      "learning_rate": 4.975689324480484e-05,
      "loss": 0.9675,
      "step": 850
    },
    {
      "epoch": 0.2781392322706571,
      "grad_norm": 2.6094343662261963,
      "learning_rate": 4.975328963970073e-05,
      "loss": 0.929,
      "step": 855
    },
    {
      "epoch": 0.2797657774886142,
      "grad_norm": 3.5149149894714355,
      "learning_rate": 4.9749659654788036e-05,
      "loss": 0.9091,
      "step": 860
    },
    {
      "epoch": 0.2813923227065712,
      "grad_norm": 2.295640468597412,
      "learning_rate": 4.9746003293935275e-05,
      "loss": 0.9224,
      "step": 865
    },
    {
      "epoch": 0.2830188679245283,
      "grad_norm": 57.37953186035156,
      "learning_rate": 4.974232056103906e-05,
      "loss": 0.9183,
      "step": 870
    },
    {
      "epoch": 0.28464541314248537,
      "grad_norm": 3.7314531803131104,
      "learning_rate": 4.973861146002411e-05,
      "loss": 0.8797,
      "step": 875
    },
    {
      "epoch": 0.28627195836044245,
      "grad_norm": 12.3214693069458,
      "learning_rate": 4.973487599484324e-05,
      "loss": 0.9734,
      "step": 880
    },
    {
      "epoch": 0.28789850357839947,
      "grad_norm": 3.5420007705688477,
      "learning_rate": 4.973111416947739e-05,
      "loss": 0.9167,
      "step": 885
    },
    {
      "epoch": 0.28952504879635654,
      "grad_norm": 3.150508403778076,
      "learning_rate": 4.972732598793556e-05,
      "loss": 0.9057,
      "step": 890
    },
    {
      "epoch": 0.2911515940143136,
      "grad_norm": 4.188449382781982,
      "learning_rate": 4.972351145425485e-05,
      "loss": 0.9125,
      "step": 895
    },
    {
      "epoch": 0.29277813923227064,
      "grad_norm": 4.404074192047119,
      "learning_rate": 4.9719670572500444e-05,
      "loss": 0.9528,
      "step": 900
    },
    {
      "epoch": 0.2944046844502277,
      "grad_norm": 2.478856325149536,
      "learning_rate": 4.97158033467656e-05,
      "loss": 0.9075,
      "step": 905
    },
    {
      "epoch": 0.2960312296681848,
      "grad_norm": 4.190792560577393,
      "learning_rate": 4.9711909781171676e-05,
      "loss": 0.9282,
      "step": 910
    },
    {
      "epoch": 0.2976577748861418,
      "grad_norm": 3.2189950942993164,
      "learning_rate": 4.970798987986805e-05,
      "loss": 0.9494,
      "step": 915
    },
    {
      "epoch": 0.2992843201040989,
      "grad_norm": 2.2690813541412354,
      "learning_rate": 4.970404364703222e-05,
      "loss": 0.9373,
      "step": 920
    },
    {
      "epoch": 0.30091086532205596,
      "grad_norm": 3.4929933547973633,
      "learning_rate": 4.970007108686972e-05,
      "loss": 0.9275,
      "step": 925
    },
    {
      "epoch": 0.302537410540013,
      "grad_norm": 2.6433072090148926,
      "learning_rate": 4.9696072203614134e-05,
      "loss": 0.9124,
      "step": 930
    },
    {
      "epoch": 0.30416395575797006,
      "grad_norm": 2.269691228866577,
      "learning_rate": 4.969204700152712e-05,
      "loss": 0.9081,
      "step": 935
    },
    {
      "epoch": 0.30579050097592714,
      "grad_norm": 2.6683006286621094,
      "learning_rate": 4.9687995484898365e-05,
      "loss": 0.9004,
      "step": 940
    },
    {
      "epoch": 0.3074170461938842,
      "grad_norm": 6.1091837882995605,
      "learning_rate": 4.9683917658045606e-05,
      "loss": 0.9467,
      "step": 945
    },
    {
      "epoch": 0.30904359141184123,
      "grad_norm": 2.08435320854187,
      "learning_rate": 4.9679813525314635e-05,
      "loss": 0.8773,
      "step": 950
    },
    {
      "epoch": 0.3106701366297983,
      "grad_norm": 2.508965492248535,
      "learning_rate": 4.967568309107925e-05,
      "loss": 0.9117,
      "step": 955
    },
    {
      "epoch": 0.3122966818477554,
      "grad_norm": 2.267766237258911,
      "learning_rate": 4.967152635974129e-05,
      "loss": 0.8895,
      "step": 960
    },
    {
      "epoch": 0.3139232270657124,
      "grad_norm": 2.4633257389068604,
      "learning_rate": 4.966734333573063e-05,
      "loss": 0.8829,
      "step": 965
    },
    {
      "epoch": 0.3155497722836695,
      "grad_norm": 2.442094326019287,
      "learning_rate": 4.966313402350516e-05,
      "loss": 0.8979,
      "step": 970
    },
    {
      "epoch": 0.31717631750162656,
      "grad_norm": 3.596240282058716,
      "learning_rate": 4.965889842755077e-05,
      "loss": 0.9094,
      "step": 975
    },
    {
      "epoch": 0.3188028627195836,
      "grad_norm": 2.0434067249298096,
      "learning_rate": 4.965463655238139e-05,
      "loss": 0.9131,
      "step": 980
    },
    {
      "epoch": 0.32042940793754066,
      "grad_norm": 3.0639724731445312,
      "learning_rate": 4.965034840253893e-05,
      "loss": 0.9041,
      "step": 985
    },
    {
      "epoch": 0.32205595315549773,
      "grad_norm": 3.173212766647339,
      "learning_rate": 4.964603398259331e-05,
      "loss": 0.8974,
      "step": 990
    },
    {
      "epoch": 0.3236824983734548,
      "grad_norm": 2.5176949501037598,
      "learning_rate": 4.9641693297142455e-05,
      "loss": 0.9196,
      "step": 995
    },
    {
      "epoch": 0.32530904359141183,
      "grad_norm": 2.001845359802246,
      "learning_rate": 4.9637326350812266e-05,
      "loss": 0.8881,
      "step": 1000
    },
    {
      "epoch": 0.3269355888093689,
      "grad_norm": 3.1144118309020996,
      "learning_rate": 4.963293314825663e-05,
      "loss": 0.8751,
      "step": 1005
    },
    {
      "epoch": 0.328562134027326,
      "grad_norm": 2.621720552444458,
      "learning_rate": 4.962851369415744e-05,
      "loss": 0.9223,
      "step": 1010
    },
    {
      "epoch": 0.330188679245283,
      "grad_norm": 2.399784564971924,
      "learning_rate": 4.962406799322454e-05,
      "loss": 0.9049,
      "step": 1015
    },
    {
      "epoch": 0.3318152244632401,
      "grad_norm": 2.089038610458374,
      "learning_rate": 4.961959605019576e-05,
      "loss": 0.9285,
      "step": 1020
    },
    {
      "epoch": 0.33344176968119715,
      "grad_norm": 2.1323084831237793,
      "learning_rate": 4.961509786983689e-05,
      "loss": 0.9157,
      "step": 1025
    },
    {
      "epoch": 0.3350683148991542,
      "grad_norm": 2.186891794204712,
      "learning_rate": 4.961057345694167e-05,
      "loss": 0.9167,
      "step": 1030
    },
    {
      "epoch": 0.33669486011711125,
      "grad_norm": 2.111762762069702,
      "learning_rate": 4.9606022816331824e-05,
      "loss": 0.907,
      "step": 1035
    },
    {
      "epoch": 0.3383214053350683,
      "grad_norm": 2.2242021560668945,
      "learning_rate": 4.960144595285701e-05,
      "loss": 0.9126,
      "step": 1040
    },
    {
      "epoch": 0.33994795055302535,
      "grad_norm": 2.575448751449585,
      "learning_rate": 4.959684287139482e-05,
      "loss": 0.9559,
      "step": 1045
    },
    {
      "epoch": 0.3415744957709824,
      "grad_norm": 2.4608261585235596,
      "learning_rate": 4.959221357685081e-05,
      "loss": 0.9058,
      "step": 1050
    },
    {
      "epoch": 0.3432010409889395,
      "grad_norm": 2.738116502761841,
      "learning_rate": 4.9587558074158464e-05,
      "loss": 0.9519,
      "step": 1055
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 2.4760334491729736,
      "learning_rate": 4.958287636827919e-05,
      "loss": 0.9204,
      "step": 1060
    },
    {
      "epoch": 0.3464541314248536,
      "grad_norm": 2.2455074787139893,
      "learning_rate": 4.9578168464202324e-05,
      "loss": 0.8996,
      "step": 1065
    },
    {
      "epoch": 0.34808067664281067,
      "grad_norm": 2.1432058811187744,
      "learning_rate": 4.9573434366945124e-05,
      "loss": 0.9308,
      "step": 1070
    },
    {
      "epoch": 0.34970722186076775,
      "grad_norm": 1.8311374187469482,
      "learning_rate": 4.956867408155277e-05,
      "loss": 0.9238,
      "step": 1075
    },
    {
      "epoch": 0.35133376707872477,
      "grad_norm": 2.4586679935455322,
      "learning_rate": 4.956388761309832e-05,
      "loss": 0.8943,
      "step": 1080
    },
    {
      "epoch": 0.35296031229668184,
      "grad_norm": 2.1968910694122314,
      "learning_rate": 4.955907496668279e-05,
      "loss": 0.9492,
      "step": 1085
    },
    {
      "epoch": 0.3545868575146389,
      "grad_norm": 4.167201519012451,
      "learning_rate": 4.955423614743503e-05,
      "loss": 0.9119,
      "step": 1090
    },
    {
      "epoch": 0.35621340273259594,
      "grad_norm": 3.1522021293640137,
      "learning_rate": 4.954937116051183e-05,
      "loss": 0.9077,
      "step": 1095
    },
    {
      "epoch": 0.357839947950553,
      "grad_norm": 2.009462356567383,
      "learning_rate": 4.954448001109785e-05,
      "loss": 0.8913,
      "step": 1100
    },
    {
      "epoch": 0.3594664931685101,
      "grad_norm": 2.839548110961914,
      "learning_rate": 4.953956270440563e-05,
      "loss": 0.9048,
      "step": 1105
    },
    {
      "epoch": 0.36109303838646717,
      "grad_norm": 2.8811404705047607,
      "learning_rate": 4.953461924567559e-05,
      "loss": 0.9697,
      "step": 1110
    },
    {
      "epoch": 0.3627195836044242,
      "grad_norm": 2.85508394241333,
      "learning_rate": 4.952964964017602e-05,
      "loss": 0.9328,
      "step": 1115
    },
    {
      "epoch": 0.36434612882238127,
      "grad_norm": 2.6920166015625,
      "learning_rate": 4.952465389320307e-05,
      "loss": 0.9193,
      "step": 1120
    },
    {
      "epoch": 0.36597267404033834,
      "grad_norm": 2.714299201965332,
      "learning_rate": 4.951963201008076e-05,
      "loss": 0.896,
      "step": 1125
    },
    {
      "epoch": 0.36759921925829536,
      "grad_norm": 2.70967960357666,
      "learning_rate": 4.951458399616096e-05,
      "loss": 0.8776,
      "step": 1130
    },
    {
      "epoch": 0.36922576447625244,
      "grad_norm": 2.452511787414551,
      "learning_rate": 4.9509509856823376e-05,
      "loss": 0.9047,
      "step": 1135
    },
    {
      "epoch": 0.3708523096942095,
      "grad_norm": 2.4412052631378174,
      "learning_rate": 4.9504409597475565e-05,
      "loss": 0.9192,
      "step": 1140
    },
    {
      "epoch": 0.37247885491216653,
      "grad_norm": 2.8084144592285156,
      "learning_rate": 4.949928322355293e-05,
      "loss": 0.8547,
      "step": 1145
    },
    {
      "epoch": 0.3741054001301236,
      "grad_norm": 3.60684871673584,
      "learning_rate": 4.949413074051868e-05,
      "loss": 0.8987,
      "step": 1150
    },
    {
      "epoch": 0.3757319453480807,
      "grad_norm": 14.40071964263916,
      "learning_rate": 4.948895215386388e-05,
      "loss": 0.8831,
      "step": 1155
    },
    {
      "epoch": 0.37735849056603776,
      "grad_norm": 2.2433459758758545,
      "learning_rate": 4.948374746910739e-05,
      "loss": 0.9195,
      "step": 1160
    },
    {
      "epoch": 0.3789850357839948,
      "grad_norm": 2.2205615043640137,
      "learning_rate": 4.947851669179589e-05,
      "loss": 0.9168,
      "step": 1165
    },
    {
      "epoch": 0.38061158100195186,
      "grad_norm": 2.3830015659332275,
      "learning_rate": 4.9473259827503865e-05,
      "loss": 0.9033,
      "step": 1170
    },
    {
      "epoch": 0.38223812621990894,
      "grad_norm": 2.8027889728546143,
      "learning_rate": 4.946797688183361e-05,
      "loss": 0.8746,
      "step": 1175
    },
    {
      "epoch": 0.38386467143786596,
      "grad_norm": 4.316250324249268,
      "learning_rate": 4.94626678604152e-05,
      "loss": 0.888,
      "step": 1180
    },
    {
      "epoch": 0.38549121665582303,
      "grad_norm": 2.363643169403076,
      "learning_rate": 4.945733276890652e-05,
      "loss": 0.8916,
      "step": 1185
    },
    {
      "epoch": 0.3871177618737801,
      "grad_norm": 2.065160036087036,
      "learning_rate": 4.945197161299321e-05,
      "loss": 0.8829,
      "step": 1190
    },
    {
      "epoch": 0.38874430709173713,
      "grad_norm": 2.287912130355835,
      "learning_rate": 4.944658439838872e-05,
      "loss": 0.9027,
      "step": 1195
    },
    {
      "epoch": 0.3903708523096942,
      "grad_norm": 2.525421142578125,
      "learning_rate": 4.9441171130834245e-05,
      "loss": 0.8787,
      "step": 1200
    },
    {
      "epoch": 0.3919973975276513,
      "grad_norm": 3.762483835220337,
      "learning_rate": 4.943573181609876e-05,
      "loss": 0.9301,
      "step": 1205
    },
    {
      "epoch": 0.3936239427456083,
      "grad_norm": 2.5889155864715576,
      "learning_rate": 4.943026645997898e-05,
      "loss": 0.915,
      "step": 1210
    },
    {
      "epoch": 0.3952504879635654,
      "grad_norm": 2.531707525253296,
      "learning_rate": 4.9424775068299404e-05,
      "loss": 0.9419,
      "step": 1215
    },
    {
      "epoch": 0.39687703318152245,
      "grad_norm": 5.207334041595459,
      "learning_rate": 4.941925764691224e-05,
      "loss": 0.8952,
      "step": 1220
    },
    {
      "epoch": 0.39850357839947953,
      "grad_norm": 2.425893545150757,
      "learning_rate": 4.941371420169746e-05,
      "loss": 0.9191,
      "step": 1225
    },
    {
      "epoch": 0.40013012361743655,
      "grad_norm": 2.961516857147217,
      "learning_rate": 4.940814473856278e-05,
      "loss": 0.9196,
      "step": 1230
    },
    {
      "epoch": 0.4017566688353936,
      "grad_norm": 1.8956583738327026,
      "learning_rate": 4.940254926344361e-05,
      "loss": 0.8787,
      "step": 1235
    },
    {
      "epoch": 0.4033832140533507,
      "grad_norm": 2.169212579727173,
      "learning_rate": 4.9396927782303105e-05,
      "loss": 0.9207,
      "step": 1240
    },
    {
      "epoch": 0.4050097592713077,
      "grad_norm": 2.522149085998535,
      "learning_rate": 4.939128030113213e-05,
      "loss": 0.8792,
      "step": 1245
    },
    {
      "epoch": 0.4066363044892648,
      "grad_norm": 2.043013095855713,
      "learning_rate": 4.938560682594925e-05,
      "loss": 0.9033,
      "step": 1250
    },
    {
      "epoch": 0.4082628497072219,
      "grad_norm": 2.008324384689331,
      "learning_rate": 4.9379907362800756e-05,
      "loss": 0.8957,
      "step": 1255
    },
    {
      "epoch": 0.4098893949251789,
      "grad_norm": 1.997517704963684,
      "learning_rate": 4.937418191776061e-05,
      "loss": 0.8688,
      "step": 1260
    },
    {
      "epoch": 0.41151594014313597,
      "grad_norm": 3.0802457332611084,
      "learning_rate": 4.936843049693046e-05,
      "loss": 0.8748,
      "step": 1265
    },
    {
      "epoch": 0.41314248536109305,
      "grad_norm": 9.878500938415527,
      "learning_rate": 4.936265310643967e-05,
      "loss": 0.914,
      "step": 1270
    },
    {
      "epoch": 0.4147690305790501,
      "grad_norm": 2.237401008605957,
      "learning_rate": 4.935684975244525e-05,
      "loss": 0.9122,
      "step": 1275
    },
    {
      "epoch": 0.41639557579700714,
      "grad_norm": 2.210939407348633,
      "learning_rate": 4.9351020441131876e-05,
      "loss": 0.8811,
      "step": 1280
    },
    {
      "epoch": 0.4180221210149642,
      "grad_norm": 2.0587236881256104,
      "learning_rate": 4.9345165178711904e-05,
      "loss": 0.8936,
      "step": 1285
    },
    {
      "epoch": 0.4196486662329213,
      "grad_norm": 1.8404570817947388,
      "learning_rate": 4.933928397142536e-05,
      "loss": 0.9308,
      "step": 1290
    },
    {
      "epoch": 0.4212752114508783,
      "grad_norm": 1.9954590797424316,
      "learning_rate": 4.9333376825539864e-05,
      "loss": 0.8949,
      "step": 1295
    },
    {
      "epoch": 0.4229017566688354,
      "grad_norm": 2.556403636932373,
      "learning_rate": 4.932744374735075e-05,
      "loss": 0.8938,
      "step": 1300
    },
    {
      "epoch": 0.42452830188679247,
      "grad_norm": 1.964281439781189,
      "learning_rate": 4.932148474318094e-05,
      "loss": 0.8739,
      "step": 1305
    },
    {
      "epoch": 0.4261548471047495,
      "grad_norm": 1.882667899131775,
      "learning_rate": 4.9315499819381004e-05,
      "loss": 0.8931,
      "step": 1310
    },
    {
      "epoch": 0.42778139232270657,
      "grad_norm": 3.4836301803588867,
      "learning_rate": 4.930948898232912e-05,
      "loss": 0.9002,
      "step": 1315
    },
    {
      "epoch": 0.42940793754066364,
      "grad_norm": 2.2451624870300293,
      "learning_rate": 4.93034522384311e-05,
      "loss": 0.9097,
      "step": 1320
    },
    {
      "epoch": 0.43103448275862066,
      "grad_norm": 2.171105146408081,
      "learning_rate": 4.929738959412037e-05,
      "loss": 0.9215,
      "step": 1325
    },
    {
      "epoch": 0.43266102797657774,
      "grad_norm": 2.0800368785858154,
      "learning_rate": 4.9291301055857916e-05,
      "loss": 0.901,
      "step": 1330
    },
    {
      "epoch": 0.4342875731945348,
      "grad_norm": 1.9696561098098755,
      "learning_rate": 4.9285186630132376e-05,
      "loss": 0.8899,
      "step": 1335
    },
    {
      "epoch": 0.4359141184124919,
      "grad_norm": 1.8826249837875366,
      "learning_rate": 4.9279046323459934e-05,
      "loss": 0.877,
      "step": 1340
    },
    {
      "epoch": 0.4375406636304489,
      "grad_norm": 4.076726913452148,
      "learning_rate": 4.927288014238438e-05,
      "loss": 0.8946,
      "step": 1345
    },
    {
      "epoch": 0.439167208848406,
      "grad_norm": 2.221543550491333,
      "learning_rate": 4.9266688093477066e-05,
      "loss": 0.8925,
      "step": 1350
    },
    {
      "epoch": 0.44079375406636306,
      "grad_norm": 2.182318925857544,
      "learning_rate": 4.926047018333691e-05,
      "loss": 0.9199,
      "step": 1355
    },
    {
      "epoch": 0.4424202992843201,
      "grad_norm": 3.5430712699890137,
      "learning_rate": 4.925422641859041e-05,
      "loss": 0.9061,
      "step": 1360
    },
    {
      "epoch": 0.44404684450227716,
      "grad_norm": 3.27327036857605,
      "learning_rate": 4.92479568058916e-05,
      "loss": 0.9312,
      "step": 1365
    },
    {
      "epoch": 0.44567338972023424,
      "grad_norm": 2.458996534347534,
      "learning_rate": 4.924166135192206e-05,
      "loss": 0.9184,
      "step": 1370
    },
    {
      "epoch": 0.44729993493819126,
      "grad_norm": 2.3121581077575684,
      "learning_rate": 4.923534006339091e-05,
      "loss": 0.8966,
      "step": 1375
    },
    {
      "epoch": 0.44892648015614833,
      "grad_norm": 2.582960367202759,
      "learning_rate": 4.92289929470348e-05,
      "loss": 0.8672,
      "step": 1380
    },
    {
      "epoch": 0.4505530253741054,
      "grad_norm": 4.599107265472412,
      "learning_rate": 4.922262000961793e-05,
      "loss": 0.8858,
      "step": 1385
    },
    {
      "epoch": 0.4521795705920625,
      "grad_norm": 2.6778221130371094,
      "learning_rate": 4.9216221257931984e-05,
      "loss": 0.8866,
      "step": 1390
    },
    {
      "epoch": 0.4538061158100195,
      "grad_norm": 2.472867965698242,
      "learning_rate": 4.920979669879617e-05,
      "loss": 0.9032,
      "step": 1395
    },
    {
      "epoch": 0.4554326610279766,
      "grad_norm": 2.3189878463745117,
      "learning_rate": 4.920334633905721e-05,
      "loss": 0.8592,
      "step": 1400
    },
    {
      "epoch": 0.45705920624593366,
      "grad_norm": 2.355407953262329,
      "learning_rate": 4.9196870185589304e-05,
      "loss": 0.8722,
      "step": 1405
    },
    {
      "epoch": 0.4586857514638907,
      "grad_norm": 2.983154296875,
      "learning_rate": 4.919036824529415e-05,
      "loss": 0.8677,
      "step": 1410
    },
    {
      "epoch": 0.46031229668184775,
      "grad_norm": 2.6179659366607666,
      "learning_rate": 4.918384052510092e-05,
      "loss": 0.9108,
      "step": 1415
    },
    {
      "epoch": 0.46193884189980483,
      "grad_norm": 2.4379374980926514,
      "learning_rate": 4.917728703196628e-05,
      "loss": 0.9184,
      "step": 1420
    },
    {
      "epoch": 0.46356538711776185,
      "grad_norm": 2.1980526447296143,
      "learning_rate": 4.9170707772874324e-05,
      "loss": 0.8538,
      "step": 1425
    },
    {
      "epoch": 0.4651919323357189,
      "grad_norm": 2.0506389141082764,
      "learning_rate": 4.9164102754836655e-05,
      "loss": 0.8958,
      "step": 1430
    },
    {
      "epoch": 0.466818477553676,
      "grad_norm": 2.5567352771759033,
      "learning_rate": 4.915747198489229e-05,
      "loss": 0.8834,
      "step": 1435
    },
    {
      "epoch": 0.468445022771633,
      "grad_norm": 2.502176523208618,
      "learning_rate": 4.915081547010769e-05,
      "loss": 0.8885,
      "step": 1440
    },
    {
      "epoch": 0.4700715679895901,
      "grad_norm": 2.5344648361206055,
      "learning_rate": 4.914413321757679e-05,
      "loss": 0.9421,
      "step": 1445
    },
    {
      "epoch": 0.4716981132075472,
      "grad_norm": 1.9389574527740479,
      "learning_rate": 4.913742523442091e-05,
      "loss": 0.8703,
      "step": 1450
    },
    {
      "epoch": 0.47332465842550425,
      "grad_norm": 2.2000226974487305,
      "learning_rate": 4.913069152778881e-05,
      "loss": 0.8911,
      "step": 1455
    },
    {
      "epoch": 0.4749512036434613,
      "grad_norm": 2.386732339859009,
      "learning_rate": 4.912393210485666e-05,
      "loss": 0.9334,
      "step": 1460
    },
    {
      "epoch": 0.47657774886141835,
      "grad_norm": 2.4275944232940674,
      "learning_rate": 4.911714697282806e-05,
      "loss": 0.8778,
      "step": 1465
    },
    {
      "epoch": 0.4782042940793754,
      "grad_norm": 2.639420509338379,
      "learning_rate": 4.9110336138933964e-05,
      "loss": 0.8834,
      "step": 1470
    },
    {
      "epoch": 0.47983083929733245,
      "grad_norm": 1.7982360124588013,
      "learning_rate": 4.9103499610432744e-05,
      "loss": 0.8468,
      "step": 1475
    },
    {
      "epoch": 0.4814573845152895,
      "grad_norm": 2.2488203048706055,
      "learning_rate": 4.909663739461017e-05,
      "loss": 0.8559,
      "step": 1480
    },
    {
      "epoch": 0.4830839297332466,
      "grad_norm": 2.1595704555511475,
      "learning_rate": 4.908974949877935e-05,
      "loss": 0.9202,
      "step": 1485
    },
    {
      "epoch": 0.4847104749512036,
      "grad_norm": 2.029742479324341,
      "learning_rate": 4.908283593028078e-05,
      "loss": 0.9053,
      "step": 1490
    },
    {
      "epoch": 0.4863370201691607,
      "grad_norm": 2.2849221229553223,
      "learning_rate": 4.907589669648232e-05,
      "loss": 0.8709,
      "step": 1495
    },
    {
      "epoch": 0.48796356538711777,
      "grad_norm": 2.050663709640503,
      "learning_rate": 4.9068931804779175e-05,
      "loss": 0.8895,
      "step": 1500
    },
    {
      "epoch": 0.48959011060507485,
      "grad_norm": 1.842971682548523,
      "learning_rate": 4.9061941262593886e-05,
      "loss": 0.8773,
      "step": 1505
    },
    {
      "epoch": 0.49121665582303187,
      "grad_norm": 3.2731707096099854,
      "learning_rate": 4.905492507737634e-05,
      "loss": 0.8763,
      "step": 1510
    },
    {
      "epoch": 0.49284320104098894,
      "grad_norm": 1.898478388786316,
      "learning_rate": 4.904788325660377e-05,
      "loss": 0.8569,
      "step": 1515
    },
    {
      "epoch": 0.494469746258946,
      "grad_norm": 2.5667638778686523,
      "learning_rate": 4.9040815807780676e-05,
      "loss": 0.8591,
      "step": 1520
    },
    {
      "epoch": 0.49609629147690304,
      "grad_norm": 4.3376383781433105,
      "learning_rate": 4.9033722738438924e-05,
      "loss": 0.9128,
      "step": 1525
    },
    {
      "epoch": 0.4977228366948601,
      "grad_norm": 1.9267882108688354,
      "learning_rate": 4.9026604056137664e-05,
      "loss": 0.88,
      "step": 1530
    },
    {
      "epoch": 0.4993493819128172,
      "grad_norm": 1.95156729221344,
      "learning_rate": 4.901945976846334e-05,
      "loss": 0.9075,
      "step": 1535
    },
    {
      "epoch": 0.5009759271307742,
      "grad_norm": 1.811874508857727,
      "learning_rate": 4.9013725907570775e-05,
      "loss": 0.8934,
      "step": 1540
    },
    {
      "epoch": 0.5026024723487313,
      "grad_norm": 2.121002674102783,
      "learning_rate": 4.900653554942987e-05,
      "loss": 0.8735,
      "step": 1545
    },
    {
      "epoch": 0.5042290175666884,
      "grad_norm": 2.8290669918060303,
      "learning_rate": 4.8999319607303076e-05,
      "loss": 0.904,
      "step": 1550
    },
    {
      "epoch": 0.5058555627846454,
      "grad_norm": 1.8206206560134888,
      "learning_rate": 4.899207808888051e-05,
      "loss": 0.8922,
      "step": 1555
    },
    {
      "epoch": 0.5074821080026025,
      "grad_norm": 2.283167839050293,
      "learning_rate": 4.898481100187953e-05,
      "loss": 0.8866,
      "step": 1560
    },
    {
      "epoch": 0.5091086532205595,
      "grad_norm": 3.57720685005188,
      "learning_rate": 4.897751835404471e-05,
      "loss": 0.8682,
      "step": 1565
    },
    {
      "epoch": 0.5107351984385166,
      "grad_norm": 2.4594457149505615,
      "learning_rate": 4.897020015314791e-05,
      "loss": 0.8686,
      "step": 1570
    },
    {
      "epoch": 0.5123617436564737,
      "grad_norm": 2.2869081497192383,
      "learning_rate": 4.89628564069882e-05,
      "loss": 0.8681,
      "step": 1575
    },
    {
      "epoch": 0.5139882888744307,
      "grad_norm": 2.9496114253997803,
      "learning_rate": 4.8955487123391884e-05,
      "loss": 0.9034,
      "step": 1580
    },
    {
      "epoch": 0.5156148340923877,
      "grad_norm": 2.24989652633667,
      "learning_rate": 4.8948092310212466e-05,
      "loss": 0.8814,
      "step": 1585
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 2.880655527114868,
      "learning_rate": 4.8940671975330674e-05,
      "loss": 0.8777,
      "step": 1590
    },
    {
      "epoch": 0.5188679245283019,
      "grad_norm": 1.8517874479293823,
      "learning_rate": 4.893322612665442e-05,
      "loss": 0.8696,
      "step": 1595
    },
    {
      "epoch": 0.5204944697462589,
      "grad_norm": 1.9082527160644531,
      "learning_rate": 4.8925754772118825e-05,
      "loss": 0.8732,
      "step": 1600
    },
    {
      "epoch": 0.522121014964216,
      "grad_norm": 3.6718180179595947,
      "learning_rate": 4.891825791968617e-05,
      "loss": 0.9223,
      "step": 1605
    },
    {
      "epoch": 0.523747560182173,
      "grad_norm": 2.3135855197906494,
      "learning_rate": 4.8910735577345925e-05,
      "loss": 0.9122,
      "step": 1610
    },
    {
      "epoch": 0.5253741054001301,
      "grad_norm": 2.163776397705078,
      "learning_rate": 4.890318775311471e-05,
      "loss": 0.9288,
      "step": 1615
    },
    {
      "epoch": 0.5270006506180872,
      "grad_norm": 3.1857199668884277,
      "learning_rate": 4.889561445503632e-05,
      "loss": 0.8805,
      "step": 1620
    },
    {
      "epoch": 0.5286271958360442,
      "grad_norm": 2.046293258666992,
      "learning_rate": 4.888801569118169e-05,
      "loss": 0.879,
      "step": 1625
    },
    {
      "epoch": 0.5302537410540012,
      "grad_norm": 2.4297502040863037,
      "learning_rate": 4.888039146964888e-05,
      "loss": 0.8726,
      "step": 1630
    },
    {
      "epoch": 0.5318802862719584,
      "grad_norm": 2.176730155944824,
      "learning_rate": 4.88727417985631e-05,
      "loss": 0.8869,
      "step": 1635
    },
    {
      "epoch": 0.5335068314899154,
      "grad_norm": 2.660156011581421,
      "learning_rate": 4.8865066686076685e-05,
      "loss": 0.8794,
      "step": 1640
    },
    {
      "epoch": 0.5351333767078725,
      "grad_norm": 1.9085681438446045,
      "learning_rate": 4.885736614036907e-05,
      "loss": 0.8842,
      "step": 1645
    },
    {
      "epoch": 0.5367599219258296,
      "grad_norm": 2.0674500465393066,
      "learning_rate": 4.8849640169646784e-05,
      "loss": 0.9047,
      "step": 1650
    },
    {
      "epoch": 0.5383864671437866,
      "grad_norm": 1.7518874406814575,
      "learning_rate": 4.884188878214347e-05,
      "loss": 0.8731,
      "step": 1655
    },
    {
      "epoch": 0.5400130123617437,
      "grad_norm": 1.9546101093292236,
      "learning_rate": 4.883411198611987e-05,
      "loss": 0.8703,
      "step": 1660
    },
    {
      "epoch": 0.5416395575797007,
      "grad_norm": 2.0076913833618164,
      "learning_rate": 4.882630978986377e-05,
      "loss": 0.8457,
      "step": 1665
    },
    {
      "epoch": 0.5432661027976577,
      "grad_norm": 1.994197130203247,
      "learning_rate": 4.881848220169003e-05,
      "loss": 0.8691,
      "step": 1670
    },
    {
      "epoch": 0.5448926480156149,
      "grad_norm": 1.8957130908966064,
      "learning_rate": 4.881062922994061e-05,
      "loss": 0.9,
      "step": 1675
    },
    {
      "epoch": 0.5465191932335719,
      "grad_norm": 2.4126741886138916,
      "learning_rate": 4.8802750882984484e-05,
      "loss": 0.8792,
      "step": 1680
    },
    {
      "epoch": 0.5481457384515289,
      "grad_norm": 2.029853105545044,
      "learning_rate": 4.879484716921767e-05,
      "loss": 0.8551,
      "step": 1685
    },
    {
      "epoch": 0.549772283669486,
      "grad_norm": 2.168409585952759,
      "learning_rate": 4.878691809706324e-05,
      "loss": 0.8499,
      "step": 1690
    },
    {
      "epoch": 0.5513988288874431,
      "grad_norm": 1.8969876766204834,
      "learning_rate": 4.8778963674971276e-05,
      "loss": 0.8941,
      "step": 1695
    },
    {
      "epoch": 0.5530253741054001,
      "grad_norm": 2.291750192642212,
      "learning_rate": 4.877098391141888e-05,
      "loss": 0.8772,
      "step": 1700
    },
    {
      "epoch": 0.5546519193233572,
      "grad_norm": 2.1070289611816406,
      "learning_rate": 4.876297881491015e-05,
      "loss": 0.8585,
      "step": 1705
    },
    {
      "epoch": 0.5562784645413142,
      "grad_norm": 2.3090732097625732,
      "learning_rate": 4.875494839397621e-05,
      "loss": 0.8563,
      "step": 1710
    },
    {
      "epoch": 0.5579050097592713,
      "grad_norm": 2.1373162269592285,
      "learning_rate": 4.874689265717514e-05,
      "loss": 0.8968,
      "step": 1715
    },
    {
      "epoch": 0.5595315549772284,
      "grad_norm": 2.126462459564209,
      "learning_rate": 4.873881161309201e-05,
      "loss": 0.9171,
      "step": 1720
    },
    {
      "epoch": 0.5611581001951854,
      "grad_norm": 2.2820169925689697,
      "learning_rate": 4.8730705270338864e-05,
      "loss": 0.8692,
      "step": 1725
    },
    {
      "epoch": 0.5627846454131424,
      "grad_norm": 2.154158353805542,
      "learning_rate": 4.872257363755471e-05,
      "loss": 0.8497,
      "step": 1730
    },
    {
      "epoch": 0.5644111906310996,
      "grad_norm": 2.639125347137451,
      "learning_rate": 4.871441672340551e-05,
      "loss": 0.8565,
      "step": 1735
    },
    {
      "epoch": 0.5660377358490566,
      "grad_norm": 2.3836843967437744,
      "learning_rate": 4.870623453658416e-05,
      "loss": 0.9039,
      "step": 1740
    },
    {
      "epoch": 0.5676642810670136,
      "grad_norm": 2.24796199798584,
      "learning_rate": 4.869802708581048e-05,
      "loss": 0.8957,
      "step": 1745
    },
    {
      "epoch": 0.5692908262849707,
      "grad_norm": 2.960399627685547,
      "learning_rate": 4.8689794379831244e-05,
      "loss": 0.8952,
      "step": 1750
    },
    {
      "epoch": 0.5709173715029278,
      "grad_norm": 2.3349416255950928,
      "learning_rate": 4.8681536427420106e-05,
      "loss": 0.8962,
      "step": 1755
    },
    {
      "epoch": 0.5725439167208849,
      "grad_norm": 1.852830410003662,
      "learning_rate": 4.867325323737765e-05,
      "loss": 0.8478,
      "step": 1760
    },
    {
      "epoch": 0.5741704619388419,
      "grad_norm": 2.357431650161743,
      "learning_rate": 4.866494481853134e-05,
      "loss": 0.8773,
      "step": 1765
    },
    {
      "epoch": 0.5757970071567989,
      "grad_norm": 1.9144153594970703,
      "learning_rate": 4.865661117973555e-05,
      "loss": 0.8611,
      "step": 1770
    },
    {
      "epoch": 0.5774235523747561,
      "grad_norm": 2.0974292755126953,
      "learning_rate": 4.8648252329871494e-05,
      "loss": 0.8983,
      "step": 1775
    },
    {
      "epoch": 0.5790500975927131,
      "grad_norm": 4.729403972625732,
      "learning_rate": 4.863986827784729e-05,
      "loss": 0.8712,
      "step": 1780
    },
    {
      "epoch": 0.5806766428106701,
      "grad_norm": 2.222569227218628,
      "learning_rate": 4.8631459032597905e-05,
      "loss": 0.9075,
      "step": 1785
    },
    {
      "epoch": 0.5823031880286272,
      "grad_norm": 2.1208598613739014,
      "learning_rate": 4.8623024603085145e-05,
      "loss": 0.8685,
      "step": 1790
    },
    {
      "epoch": 0.5839297332465843,
      "grad_norm": 2.058300733566284,
      "learning_rate": 4.861456499829764e-05,
      "loss": 0.8901,
      "step": 1795
    },
    {
      "epoch": 0.5855562784645413,
      "grad_norm": 1.7815157175064087,
      "learning_rate": 4.86060802272509e-05,
      "loss": 0.9112,
      "step": 1800
    },
    {
      "epoch": 0.5871828236824984,
      "grad_norm": 1.6161009073257446,
      "learning_rate": 4.8597570298987196e-05,
      "loss": 0.882,
      "step": 1805
    },
    {
      "epoch": 0.5888093689004554,
      "grad_norm": 2.2587006092071533,
      "learning_rate": 4.858903522257565e-05,
      "loss": 0.8494,
      "step": 1810
    },
    {
      "epoch": 0.5904359141184125,
      "grad_norm": 1.8459244966506958,
      "learning_rate": 4.858047500711217e-05,
      "loss": 0.8782,
      "step": 1815
    },
    {
      "epoch": 0.5920624593363696,
      "grad_norm": 2.19490122795105,
      "learning_rate": 4.857188966171946e-05,
      "loss": 0.8794,
      "step": 1820
    },
    {
      "epoch": 0.5936890045543266,
      "grad_norm": 2.3178391456604004,
      "learning_rate": 4.856327919554699e-05,
      "loss": 0.8661,
      "step": 1825
    },
    {
      "epoch": 0.5953155497722836,
      "grad_norm": 2.1639411449432373,
      "learning_rate": 4.855464361777102e-05,
      "loss": 0.888,
      "step": 1830
    },
    {
      "epoch": 0.5969420949902408,
      "grad_norm": 1.616613507270813,
      "learning_rate": 4.8545982937594576e-05,
      "loss": 0.8784,
      "step": 1835
    },
    {
      "epoch": 0.5985686402081978,
      "grad_norm": 2.395139217376709,
      "learning_rate": 4.8537297164247405e-05,
      "loss": 0.8677,
      "step": 1840
    },
    {
      "epoch": 0.6001951854261548,
      "grad_norm": 6.791170597076416,
      "learning_rate": 4.8528586306986034e-05,
      "loss": 0.8745,
      "step": 1845
    },
    {
      "epoch": 0.6018217306441119,
      "grad_norm": 3.1358447074890137,
      "learning_rate": 4.8519850375093686e-05,
      "loss": 0.8636,
      "step": 1850
    },
    {
      "epoch": 0.603448275862069,
      "grad_norm": 2.2898097038269043,
      "learning_rate": 4.851108937788034e-05,
      "loss": 0.9009,
      "step": 1855
    },
    {
      "epoch": 0.605074821080026,
      "grad_norm": 2.2613158226013184,
      "learning_rate": 4.850230332468265e-05,
      "loss": 0.8893,
      "step": 1860
    },
    {
      "epoch": 0.6067013662979831,
      "grad_norm": 1.6697711944580078,
      "learning_rate": 4.849349222486402e-05,
      "loss": 0.8751,
      "step": 1865
    },
    {
      "epoch": 0.6083279115159401,
      "grad_norm": 1.8398504257202148,
      "learning_rate": 4.848465608781451e-05,
      "loss": 0.8845,
      "step": 1870
    },
    {
      "epoch": 0.6099544567338973,
      "grad_norm": 2.2207603454589844,
      "learning_rate": 4.8475794922950855e-05,
      "loss": 0.8973,
      "step": 1875
    },
    {
      "epoch": 0.6115810019518543,
      "grad_norm": 1.824102520942688,
      "learning_rate": 4.846690873971651e-05,
      "loss": 0.8989,
      "step": 1880
    },
    {
      "epoch": 0.6132075471698113,
      "grad_norm": 2.416241407394409,
      "learning_rate": 4.8457997547581546e-05,
      "loss": 0.8822,
      "step": 1885
    },
    {
      "epoch": 0.6148340923877684,
      "grad_norm": 1.9753471612930298,
      "learning_rate": 4.84490613560427e-05,
      "loss": 0.8719,
      "step": 1890
    },
    {
      "epoch": 0.6164606376057254,
      "grad_norm": 1.9445384740829468,
      "learning_rate": 4.8440100174623354e-05,
      "loss": 0.865,
      "step": 1895
    },
    {
      "epoch": 0.6180871828236825,
      "grad_norm": 1.8253824710845947,
      "learning_rate": 4.843111401287353e-05,
      "loss": 0.8734,
      "step": 1900
    },
    {
      "epoch": 0.6197137280416396,
      "grad_norm": 1.9183560609817505,
      "learning_rate": 4.842210288036986e-05,
      "loss": 0.8479,
      "step": 1905
    },
    {
      "epoch": 0.6213402732595966,
      "grad_norm": 1.9014906883239746,
      "learning_rate": 4.841306678671558e-05,
      "loss": 0.898,
      "step": 1910
    },
    {
      "epoch": 0.6229668184775536,
      "grad_norm": 2.550158739089966,
      "learning_rate": 4.840400574154056e-05,
      "loss": 0.8809,
      "step": 1915
    },
    {
      "epoch": 0.6245933636955108,
      "grad_norm": 6.833774089813232,
      "learning_rate": 4.8394919754501217e-05,
      "loss": 0.8755,
      "step": 1920
    },
    {
      "epoch": 0.6262199089134678,
      "grad_norm": 4.296611309051514,
      "learning_rate": 4.8385808835280584e-05,
      "loss": 0.8456,
      "step": 1925
    },
    {
      "epoch": 0.6278464541314248,
      "grad_norm": 2.215914487838745,
      "learning_rate": 4.8376672993588245e-05,
      "loss": 0.8969,
      "step": 1930
    },
    {
      "epoch": 0.629472999349382,
      "grad_norm": 2.6802151203155518,
      "learning_rate": 4.8367512239160356e-05,
      "loss": 0.8768,
      "step": 1935
    },
    {
      "epoch": 0.631099544567339,
      "grad_norm": 2.843503475189209,
      "learning_rate": 4.8358326581759615e-05,
      "loss": 0.8593,
      "step": 1940
    },
    {
      "epoch": 0.632726089785296,
      "grad_norm": 2.6657474040985107,
      "learning_rate": 4.834911603117526e-05,
      "loss": 0.8932,
      "step": 1945
    },
    {
      "epoch": 0.6343526350032531,
      "grad_norm": 1.8572765588760376,
      "learning_rate": 4.833988059722306e-05,
      "loss": 0.8921,
      "step": 1950
    },
    {
      "epoch": 0.6359791802212101,
      "grad_norm": 3.170466184616089,
      "learning_rate": 4.833062028974531e-05,
      "loss": 0.8409,
      "step": 1955
    },
    {
      "epoch": 0.6376057254391672,
      "grad_norm": 1.8783475160598755,
      "learning_rate": 4.832133511861079e-05,
      "loss": 0.8651,
      "step": 1960
    },
    {
      "epoch": 0.6392322706571243,
      "grad_norm": 2.4108924865722656,
      "learning_rate": 4.8312025093714806e-05,
      "loss": 0.8651,
      "step": 1965
    },
    {
      "epoch": 0.6408588158750813,
      "grad_norm": 3.5263514518737793,
      "learning_rate": 4.830269022497913e-05,
      "loss": 0.8509,
      "step": 1970
    },
    {
      "epoch": 0.6424853610930383,
      "grad_norm": 2.5810863971710205,
      "learning_rate": 4.829333052235202e-05,
      "loss": 0.8692,
      "step": 1975
    },
    {
      "epoch": 0.6441119063109955,
      "grad_norm": 2.0188722610473633,
      "learning_rate": 4.82839459958082e-05,
      "loss": 0.8697,
      "step": 1980
    },
    {
      "epoch": 0.6457384515289525,
      "grad_norm": 2.043989896774292,
      "learning_rate": 4.8274536655348834e-05,
      "loss": 0.8836,
      "step": 1985
    },
    {
      "epoch": 0.6473649967469096,
      "grad_norm": 2.661475419998169,
      "learning_rate": 4.826510251100155e-05,
      "loss": 0.8619,
      "step": 1990
    },
    {
      "epoch": 0.6489915419648666,
      "grad_norm": 2.208104133605957,
      "learning_rate": 4.825564357282041e-05,
      "loss": 0.8879,
      "step": 1995
    },
    {
      "epoch": 0.6506180871828237,
      "grad_norm": 4.05889368057251,
      "learning_rate": 4.8246159850885883e-05,
      "loss": 0.8645,
      "step": 2000
    },
    {
      "epoch": 0.6522446324007808,
      "grad_norm": 2.1343226432800293,
      "learning_rate": 4.823665135530486e-05,
      "loss": 0.8617,
      "step": 2005
    },
    {
      "epoch": 0.6538711776187378,
      "grad_norm": 2.087005615234375,
      "learning_rate": 4.822711809621063e-05,
      "loss": 0.8902,
      "step": 2010
    },
    {
      "epoch": 0.6554977228366948,
      "grad_norm": 4.40689754486084,
      "learning_rate": 4.821756008376289e-05,
      "loss": 0.9144,
      "step": 2015
    },
    {
      "epoch": 0.657124268054652,
      "grad_norm": 3.3655898571014404,
      "learning_rate": 4.820797732814768e-05,
      "loss": 0.8924,
      "step": 2020
    },
    {
      "epoch": 0.658750813272609,
      "grad_norm": 2.2318570613861084,
      "learning_rate": 4.819836983957744e-05,
      "loss": 0.8448,
      "step": 2025
    },
    {
      "epoch": 0.660377358490566,
      "grad_norm": 2.6342506408691406,
      "learning_rate": 4.818873762829097e-05,
      "loss": 0.8864,
      "step": 2030
    },
    {
      "epoch": 0.6620039037085231,
      "grad_norm": 2.1214475631713867,
      "learning_rate": 4.8179080704553386e-05,
      "loss": 0.8495,
      "step": 2035
    },
    {
      "epoch": 0.6636304489264802,
      "grad_norm": 1.7027548551559448,
      "learning_rate": 4.8169399078656175e-05,
      "loss": 0.8495,
      "step": 2040
    },
    {
      "epoch": 0.6652569941444372,
      "grad_norm": 2.5632596015930176,
      "learning_rate": 4.815969276091713e-05,
      "loss": 0.8696,
      "step": 2045
    },
    {
      "epoch": 0.6668835393623943,
      "grad_norm": 2.2075538635253906,
      "learning_rate": 4.814996176168036e-05,
      "loss": 0.8568,
      "step": 2050
    },
    {
      "epoch": 0.6685100845803513,
      "grad_norm": 2.355635404586792,
      "learning_rate": 4.8140206091316284e-05,
      "loss": 0.8534,
      "step": 2055
    },
    {
      "epoch": 0.6701366297983083,
      "grad_norm": 2.9022514820098877,
      "learning_rate": 4.8130425760221615e-05,
      "loss": 0.8704,
      "step": 2060
    },
    {
      "epoch": 0.6717631750162655,
      "grad_norm": 2.1720938682556152,
      "learning_rate": 4.8120620778819334e-05,
      "loss": 0.8646,
      "step": 2065
    },
    {
      "epoch": 0.6733897202342225,
      "grad_norm": 2.2691755294799805,
      "learning_rate": 4.81107911575587e-05,
      "loss": 0.8392,
      "step": 2070
    },
    {
      "epoch": 0.6750162654521795,
      "grad_norm": 1.8914642333984375,
      "learning_rate": 4.810093690691524e-05,
      "loss": 0.8764,
      "step": 2075
    },
    {
      "epoch": 0.6766428106701367,
      "grad_norm": 1.8982810974121094,
      "learning_rate": 4.80910580373907e-05,
      "loss": 0.8917,
      "step": 2080
    },
    {
      "epoch": 0.6782693558880937,
      "grad_norm": 2.189270496368408,
      "learning_rate": 4.8081154559513106e-05,
      "loss": 0.9098,
      "step": 2085
    },
    {
      "epoch": 0.6798959011060507,
      "grad_norm": 2.0608716011047363,
      "learning_rate": 4.807122648383667e-05,
      "loss": 0.8694,
      "step": 2090
    },
    {
      "epoch": 0.6815224463240078,
      "grad_norm": 3.0705621242523193,
      "learning_rate": 4.8061273820941845e-05,
      "loss": 0.86,
      "step": 2095
    },
    {
      "epoch": 0.6831489915419648,
      "grad_norm": 2.9095020294189453,
      "learning_rate": 4.8051296581435264e-05,
      "loss": 0.876,
      "step": 2100
    },
    {
      "epoch": 0.684775536759922,
      "grad_norm": 1.7684783935546875,
      "learning_rate": 4.804129477594977e-05,
      "loss": 0.8863,
      "step": 2105
    },
    {
      "epoch": 0.686402081977879,
      "grad_norm": 2.1000592708587646,
      "learning_rate": 4.803126841514437e-05,
      "loss": 0.9342,
      "step": 2110
    },
    {
      "epoch": 0.688028627195836,
      "grad_norm": 2.359872579574585,
      "learning_rate": 4.802121750970425e-05,
      "loss": 0.8656,
      "step": 2115
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 2.3630592823028564,
      "learning_rate": 4.8011142070340774e-05,
      "loss": 0.8587,
      "step": 2120
    },
    {
      "epoch": 0.6912817176317502,
      "grad_norm": 2.767029285430908,
      "learning_rate": 4.8001042107791405e-05,
      "loss": 0.8634,
      "step": 2125
    },
    {
      "epoch": 0.6929082628497072,
      "grad_norm": 6.400025367736816,
      "learning_rate": 4.799091763281978e-05,
      "loss": 0.8751,
      "step": 2130
    },
    {
      "epoch": 0.6945348080676643,
      "grad_norm": 3.766014814376831,
      "learning_rate": 4.798076865621564e-05,
      "loss": 0.8715,
      "step": 2135
    },
    {
      "epoch": 0.6961613532856213,
      "grad_norm": 1.9491114616394043,
      "learning_rate": 4.7970595188794846e-05,
      "loss": 0.8969,
      "step": 2140
    },
    {
      "epoch": 0.6977878985035784,
      "grad_norm": 2.2208404541015625,
      "learning_rate": 4.796039724139935e-05,
      "loss": 0.8848,
      "step": 2145
    },
    {
      "epoch": 0.6994144437215355,
      "grad_norm": 2.700895309448242,
      "learning_rate": 4.79501748248972e-05,
      "loss": 0.8695,
      "step": 2150
    },
    {
      "epoch": 0.7010409889394925,
      "grad_norm": 2.6097826957702637,
      "learning_rate": 4.793992795018253e-05,
      "loss": 0.8604,
      "step": 2155
    },
    {
      "epoch": 0.7026675341574495,
      "grad_norm": 1.8880006074905396,
      "learning_rate": 4.792965662817551e-05,
      "loss": 0.885,
      "step": 2160
    },
    {
      "epoch": 0.7042940793754067,
      "grad_norm": 3.3072919845581055,
      "learning_rate": 4.7919360869822394e-05,
      "loss": 0.9164,
      "step": 2165
    },
    {
      "epoch": 0.7059206245933637,
      "grad_norm": 2.4248101711273193,
      "learning_rate": 4.790904068609546e-05,
      "loss": 0.8677,
      "step": 2170
    },
    {
      "epoch": 0.7075471698113207,
      "grad_norm": 2.345674991607666,
      "learning_rate": 4.789869608799304e-05,
      "loss": 0.837,
      "step": 2175
    },
    {
      "epoch": 0.7091737150292778,
      "grad_norm": 1.677289605140686,
      "learning_rate": 4.7888327086539446e-05,
      "loss": 0.8544,
      "step": 2180
    },
    {
      "epoch": 0.7108002602472349,
      "grad_norm": 2.025778293609619,
      "learning_rate": 4.787793369278503e-05,
      "loss": 0.8867,
      "step": 2185
    },
    {
      "epoch": 0.7124268054651919,
      "grad_norm": 2.362762689590454,
      "learning_rate": 4.7867515917806124e-05,
      "loss": 0.8777,
      "step": 2190
    },
    {
      "epoch": 0.714053350683149,
      "grad_norm": 5.319698333740234,
      "learning_rate": 4.785707377270505e-05,
      "loss": 0.858,
      "step": 2195
    },
    {
      "epoch": 0.715679895901106,
      "grad_norm": 4.411806106567383,
      "learning_rate": 4.784660726861009e-05,
      "loss": 0.8466,
      "step": 2200
    },
    {
      "epoch": 0.7173064411190632,
      "grad_norm": 2.4684970378875732,
      "learning_rate": 4.78361164166755e-05,
      "loss": 0.9027,
      "step": 2205
    },
    {
      "epoch": 0.7189329863370202,
      "grad_norm": 2.4765100479125977,
      "learning_rate": 4.7825601228081484e-05,
      "loss": 0.8433,
      "step": 2210
    },
    {
      "epoch": 0.7205595315549772,
      "grad_norm": 1.9810550212860107,
      "learning_rate": 4.781506171403416e-05,
      "loss": 0.8736,
      "step": 2215
    },
    {
      "epoch": 0.7221860767729343,
      "grad_norm": 1.7304781675338745,
      "learning_rate": 4.78044978857656e-05,
      "loss": 0.8681,
      "step": 2220
    },
    {
      "epoch": 0.7238126219908914,
      "grad_norm": 1.7415772676467896,
      "learning_rate": 4.779390975453376e-05,
      "loss": 0.8731,
      "step": 2225
    },
    {
      "epoch": 0.7254391672088484,
      "grad_norm": 1.787624716758728,
      "learning_rate": 4.778329733162251e-05,
      "loss": 0.8768,
      "step": 2230
    },
    {
      "epoch": 0.7270657124268055,
      "grad_norm": 2.4273312091827393,
      "learning_rate": 4.777266062834162e-05,
      "loss": 0.858,
      "step": 2235
    },
    {
      "epoch": 0.7286922576447625,
      "grad_norm": 1.8364176750183105,
      "learning_rate": 4.776199965602671e-05,
      "loss": 0.892,
      "step": 2240
    },
    {
      "epoch": 0.7303188028627196,
      "grad_norm": 1.8039227724075317,
      "learning_rate": 4.7751314426039286e-05,
      "loss": 0.8345,
      "step": 2245
    },
    {
      "epoch": 0.7319453480806767,
      "grad_norm": 1.9627631902694702,
      "learning_rate": 4.7740604949766686e-05,
      "loss": 0.8597,
      "step": 2250
    },
    {
      "epoch": 0.7335718932986337,
      "grad_norm": 1.7530760765075684,
      "learning_rate": 4.77298712386221e-05,
      "loss": 0.885,
      "step": 2255
    },
    {
      "epoch": 0.7351984385165907,
      "grad_norm": 2.443120002746582,
      "learning_rate": 4.771911330404455e-05,
      "loss": 0.864,
      "step": 2260
    },
    {
      "epoch": 0.7368249837345479,
      "grad_norm": 1.866398572921753,
      "learning_rate": 4.7708331157498855e-05,
      "loss": 0.9056,
      "step": 2265
    },
    {
      "epoch": 0.7384515289525049,
      "grad_norm": 5.161078929901123,
      "learning_rate": 4.769752481047566e-05,
      "loss": 0.8454,
      "step": 2270
    },
    {
      "epoch": 0.7400780741704619,
      "grad_norm": 1.9637868404388428,
      "learning_rate": 4.768669427449137e-05,
      "loss": 0.8917,
      "step": 2275
    },
    {
      "epoch": 0.741704619388419,
      "grad_norm": 2.199129819869995,
      "learning_rate": 4.767583956108821e-05,
      "loss": 0.8515,
      "step": 2280
    },
    {
      "epoch": 0.743331164606376,
      "grad_norm": 2.3960113525390625,
      "learning_rate": 4.766496068183413e-05,
      "loss": 0.8803,
      "step": 2285
    },
    {
      "epoch": 0.7449577098243331,
      "grad_norm": 2.398287534713745,
      "learning_rate": 4.765405764832286e-05,
      "loss": 0.8582,
      "step": 2290
    },
    {
      "epoch": 0.7465842550422902,
      "grad_norm": 1.8750267028808594,
      "learning_rate": 4.764313047217386e-05,
      "loss": 0.8609,
      "step": 2295
    },
    {
      "epoch": 0.7482108002602472,
      "grad_norm": 2.4340929985046387,
      "learning_rate": 4.763217916503233e-05,
      "loss": 0.846,
      "step": 2300
    },
    {
      "epoch": 0.7498373454782042,
      "grad_norm": 2.125413656234741,
      "learning_rate": 4.762120373856917e-05,
      "loss": 0.8741,
      "step": 2305
    },
    {
      "epoch": 0.7514638906961614,
      "grad_norm": 2.3385329246520996,
      "learning_rate": 4.761020420448099e-05,
      "loss": 0.8747,
      "step": 2310
    },
    {
      "epoch": 0.7530904359141184,
      "grad_norm": 2.125859260559082,
      "learning_rate": 4.759918057449011e-05,
      "loss": 0.8934,
      "step": 2315
    },
    {
      "epoch": 0.7547169811320755,
      "grad_norm": 3.7632899284362793,
      "learning_rate": 4.758813286034449e-05,
      "loss": 0.9177,
      "step": 2320
    },
    {
      "epoch": 0.7563435263500325,
      "grad_norm": 3.737948417663574,
      "learning_rate": 4.7577061073817806e-05,
      "loss": 0.9371,
      "step": 2325
    },
    {
      "epoch": 0.7579700715679896,
      "grad_norm": 3.86966609954834,
      "learning_rate": 4.7565965226709353e-05,
      "loss": 0.8809,
      "step": 2330
    },
    {
      "epoch": 0.7595966167859467,
      "grad_norm": 1.5923842191696167,
      "learning_rate": 4.7554845330844066e-05,
      "loss": 0.9081,
      "step": 2335
    },
    {
      "epoch": 0.7612231620039037,
      "grad_norm": 2.317854166030884,
      "learning_rate": 4.754370139807254e-05,
      "loss": 0.8979,
      "step": 2340
    },
    {
      "epoch": 0.7628497072218607,
      "grad_norm": 1.9626786708831787,
      "learning_rate": 4.753253344027095e-05,
      "loss": 0.8509,
      "step": 2345
    },
    {
      "epoch": 0.7644762524398179,
      "grad_norm": 2.554448127746582,
      "learning_rate": 4.7521341469341106e-05,
      "loss": 0.861,
      "step": 2350
    },
    {
      "epoch": 0.7661027976577749,
      "grad_norm": 2.060730457305908,
      "learning_rate": 4.7510125497210386e-05,
      "loss": 0.8527,
      "step": 2355
    },
    {
      "epoch": 0.7677293428757319,
      "grad_norm": 2.024789571762085,
      "learning_rate": 4.749888553583175e-05,
      "loss": 0.8477,
      "step": 2360
    },
    {
      "epoch": 0.769355888093689,
      "grad_norm": 1.6188427209854126,
      "learning_rate": 4.7487621597183734e-05,
      "loss": 0.852,
      "step": 2365
    },
    {
      "epoch": 0.7709824333116461,
      "grad_norm": 2.30036997795105,
      "learning_rate": 4.747633369327043e-05,
      "loss": 0.8654,
      "step": 2370
    },
    {
      "epoch": 0.7726089785296031,
      "grad_norm": 2.2033796310424805,
      "learning_rate": 4.746502183612144e-05,
      "loss": 0.8451,
      "step": 2375
    },
    {
      "epoch": 0.7742355237475602,
      "grad_norm": 2.1281795501708984,
      "learning_rate": 4.745368603779193e-05,
      "loss": 0.9064,
      "step": 2380
    },
    {
      "epoch": 0.7758620689655172,
      "grad_norm": 2.251528263092041,
      "learning_rate": 4.7442326310362565e-05,
      "loss": 0.8481,
      "step": 2385
    },
    {
      "epoch": 0.7774886141834743,
      "grad_norm": 2.573554039001465,
      "learning_rate": 4.743094266593951e-05,
      "loss": 0.8699,
      "step": 2390
    },
    {
      "epoch": 0.7791151594014314,
      "grad_norm": 2.4464991092681885,
      "learning_rate": 4.74195351166544e-05,
      "loss": 0.9089,
      "step": 2395
    },
    {
      "epoch": 0.7807417046193884,
      "grad_norm": 2.2439992427825928,
      "learning_rate": 4.74081036746644e-05,
      "loss": 0.8616,
      "step": 2400
    },
    {
      "epoch": 0.7823682498373454,
      "grad_norm": 2.5337162017822266,
      "learning_rate": 4.7396648352152075e-05,
      "loss": 0.8716,
      "step": 2405
    },
    {
      "epoch": 0.7839947950553026,
      "grad_norm": 1.8930824995040894,
      "learning_rate": 4.7385169161325485e-05,
      "loss": 0.8365,
      "step": 2410
    },
    {
      "epoch": 0.7856213402732596,
      "grad_norm": 2.614783763885498,
      "learning_rate": 4.737366611441809e-05,
      "loss": 0.8967,
      "step": 2415
    },
    {
      "epoch": 0.7872478854912166,
      "grad_norm": 1.6895397901535034,
      "learning_rate": 4.736213922368881e-05,
      "loss": 0.8813,
      "step": 2420
    },
    {
      "epoch": 0.7888744307091737,
      "grad_norm": 4.0224223136901855,
      "learning_rate": 4.735058850142196e-05,
      "loss": 0.8602,
      "step": 2425
    },
    {
      "epoch": 0.7905009759271308,
      "grad_norm": 3.099029302597046,
      "learning_rate": 4.733901395992724e-05,
      "loss": 0.8604,
      "step": 2430
    },
    {
      "epoch": 0.7921275211450879,
      "grad_norm": 2.1614017486572266,
      "learning_rate": 4.732741561153974e-05,
      "loss": 0.87,
      "step": 2435
    },
    {
      "epoch": 0.7937540663630449,
      "grad_norm": 5.100318908691406,
      "learning_rate": 4.7315793468619954e-05,
      "loss": 0.8428,
      "step": 2440
    },
    {
      "epoch": 0.7953806115810019,
      "grad_norm": 3.786672830581665,
      "learning_rate": 4.730414754355368e-05,
      "loss": 0.8319,
      "step": 2445
    },
    {
      "epoch": 0.7970071567989591,
      "grad_norm": 2.4570653438568115,
      "learning_rate": 4.729247784875209e-05,
      "loss": 0.8752,
      "step": 2450
    },
    {
      "epoch": 0.7986337020169161,
      "grad_norm": 1.931662917137146,
      "learning_rate": 4.72807843966517e-05,
      "loss": 0.8473,
      "step": 2455
    },
    {
      "epoch": 0.8002602472348731,
      "grad_norm": 9.377140045166016,
      "learning_rate": 4.7269067199714324e-05,
      "loss": 0.8335,
      "step": 2460
    },
    {
      "epoch": 0.8018867924528302,
      "grad_norm": 2.5859813690185547,
      "learning_rate": 4.725732627042707e-05,
      "loss": 0.8586,
      "step": 2465
    },
    {
      "epoch": 0.8035133376707873,
      "grad_norm": 2.093609094619751,
      "learning_rate": 4.724556162130238e-05,
      "loss": 0.8479,
      "step": 2470
    },
    {
      "epoch": 0.8051398828887443,
      "grad_norm": 3.501932144165039,
      "learning_rate": 4.723377326487794e-05,
      "loss": 0.8601,
      "step": 2475
    },
    {
      "epoch": 0.8067664281067014,
      "grad_norm": 2.1503453254699707,
      "learning_rate": 4.7221961213716695e-05,
      "loss": 0.8555,
      "step": 2480
    },
    {
      "epoch": 0.8083929733246584,
      "grad_norm": 2.191429376602173,
      "learning_rate": 4.721012548040688e-05,
      "loss": 0.862,
      "step": 2485
    },
    {
      "epoch": 0.8100195185426154,
      "grad_norm": 2.2335188388824463,
      "learning_rate": 4.7198266077561925e-05,
      "loss": 0.8459,
      "step": 2490
    },
    {
      "epoch": 0.8116460637605726,
      "grad_norm": 2.0258047580718994,
      "learning_rate": 4.7186383017820516e-05,
      "loss": 0.8639,
      "step": 2495
    },
    {
      "epoch": 0.8132726089785296,
      "grad_norm": 3.4457333087921143,
      "learning_rate": 4.717447631384654e-05,
      "loss": 0.8673,
      "step": 2500
    },
    {
      "epoch": 0.8148991541964866,
      "grad_norm": 2.7301559448242188,
      "learning_rate": 4.7162545978329076e-05,
      "loss": 0.8611,
      "step": 2505
    },
    {
      "epoch": 0.8165256994144438,
      "grad_norm": 2.1690988540649414,
      "learning_rate": 4.715059202398239e-05,
      "loss": 0.8665,
      "step": 2510
    },
    {
      "epoch": 0.8181522446324008,
      "grad_norm": 2.5224900245666504,
      "learning_rate": 4.7138614463545924e-05,
      "loss": 0.8405,
      "step": 2515
    },
    {
      "epoch": 0.8197787898503578,
      "grad_norm": 2.2839272022247314,
      "learning_rate": 4.712661330978428e-05,
      "loss": 0.8488,
      "step": 2520
    },
    {
      "epoch": 0.8214053350683149,
      "grad_norm": 3.877641439437866,
      "learning_rate": 4.711458857548719e-05,
      "loss": 0.883,
      "step": 2525
    },
    {
      "epoch": 0.8230318802862719,
      "grad_norm": 2.267813205718994,
      "learning_rate": 4.710254027346953e-05,
      "loss": 0.8822,
      "step": 2530
    },
    {
      "epoch": 0.824658425504229,
      "grad_norm": 1.544561743736267,
      "learning_rate": 4.709046841657129e-05,
      "loss": 0.8479,
      "step": 2535
    },
    {
      "epoch": 0.8262849707221861,
      "grad_norm": 1.6922204494476318,
      "learning_rate": 4.707837301765754e-05,
      "loss": 0.8411,
      "step": 2540
    },
    {
      "epoch": 0.8279115159401431,
      "grad_norm": 1.6176890134811401,
      "learning_rate": 4.706625408961848e-05,
      "loss": 0.8749,
      "step": 2545
    },
    {
      "epoch": 0.8295380611581002,
      "grad_norm": 2.7533674240112305,
      "learning_rate": 4.705411164536935e-05,
      "loss": 0.8706,
      "step": 2550
    },
    {
      "epoch": 0.8311646063760573,
      "grad_norm": 1.8867239952087402,
      "learning_rate": 4.7041945697850466e-05,
      "loss": 0.8325,
      "step": 2555
    },
    {
      "epoch": 0.8327911515940143,
      "grad_norm": 2.392642021179199,
      "learning_rate": 4.7029756260027197e-05,
      "loss": 0.8796,
      "step": 2560
    },
    {
      "epoch": 0.8344176968119714,
      "grad_norm": 1.9180896282196045,
      "learning_rate": 4.701754334488993e-05,
      "loss": 0.8565,
      "step": 2565
    },
    {
      "epoch": 0.8360442420299284,
      "grad_norm": 2.039537191390991,
      "learning_rate": 4.700530696545409e-05,
      "loss": 0.8845,
      "step": 2570
    },
    {
      "epoch": 0.8376707872478855,
      "grad_norm": 1.9221045970916748,
      "learning_rate": 4.699304713476009e-05,
      "loss": 0.8876,
      "step": 2575
    },
    {
      "epoch": 0.8392973324658426,
      "grad_norm": 1.6282926797866821,
      "learning_rate": 4.698076386587335e-05,
      "loss": 0.8514,
      "step": 2580
    },
    {
      "epoch": 0.8409238776837996,
      "grad_norm": 2.0738115310668945,
      "learning_rate": 4.6968457171884274e-05,
      "loss": 0.8403,
      "step": 2585
    },
    {
      "epoch": 0.8425504229017566,
      "grad_norm": 2.1336333751678467,
      "learning_rate": 4.695612706590819e-05,
      "loss": 0.8634,
      "step": 2590
    },
    {
      "epoch": 0.8441769681197138,
      "grad_norm": 1.7246681451797485,
      "learning_rate": 4.6943773561085435e-05,
      "loss": 0.8568,
      "step": 2595
    },
    {
      "epoch": 0.8458035133376708,
      "grad_norm": 1.9486162662506104,
      "learning_rate": 4.693139667058123e-05,
      "loss": 0.8453,
      "step": 2600
    },
    {
      "epoch": 0.8474300585556278,
      "grad_norm": 2.4294521808624268,
      "learning_rate": 4.691899640758576e-05,
      "loss": 0.8708,
      "step": 2605
    },
    {
      "epoch": 0.8490566037735849,
      "grad_norm": 1.9123241901397705,
      "learning_rate": 4.690657278531409e-05,
      "loss": 0.8639,
      "step": 2610
    },
    {
      "epoch": 0.850683148991542,
      "grad_norm": 1.5926252603530884,
      "learning_rate": 4.689412581700618e-05,
      "loss": 0.8444,
      "step": 2615
    },
    {
      "epoch": 0.852309694209499,
      "grad_norm": 2.1733107566833496,
      "learning_rate": 4.6881655515926907e-05,
      "loss": 0.8815,
      "step": 2620
    },
    {
      "epoch": 0.8539362394274561,
      "grad_norm": 1.63432776927948,
      "learning_rate": 4.686916189536594e-05,
      "loss": 0.8626,
      "step": 2625
    },
    {
      "epoch": 0.8555627846454131,
      "grad_norm": 2.2323520183563232,
      "learning_rate": 4.685664496863789e-05,
      "loss": 0.8533,
      "step": 2630
    },
    {
      "epoch": 0.8571893298633702,
      "grad_norm": 1.6251574754714966,
      "learning_rate": 4.6844104749082144e-05,
      "loss": 0.8621,
      "step": 2635
    },
    {
      "epoch": 0.8588158750813273,
      "grad_norm": 1.9397878646850586,
      "learning_rate": 4.683154125006293e-05,
      "loss": 0.876,
      "step": 2640
    },
    {
      "epoch": 0.8604424202992843,
      "grad_norm": 1.4942872524261475,
      "learning_rate": 4.681895448496927e-05,
      "loss": 0.8755,
      "step": 2645
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 2.520001173019409,
      "learning_rate": 4.680634446721501e-05,
      "loss": 0.8728,
      "step": 2650
    },
    {
      "epoch": 0.8636955107351985,
      "grad_norm": 1.6082730293273926,
      "learning_rate": 4.679371121023877e-05,
      "loss": 0.85,
      "step": 2655
    },
    {
      "epoch": 0.8653220559531555,
      "grad_norm": 2.3203847408294678,
      "learning_rate": 4.678105472750391e-05,
      "loss": 0.8446,
      "step": 2660
    },
    {
      "epoch": 0.8669486011711126,
      "grad_norm": 2.4071483612060547,
      "learning_rate": 4.6768375032498577e-05,
      "loss": 0.882,
      "step": 2665
    },
    {
      "epoch": 0.8685751463890696,
      "grad_norm": 2.0908102989196777,
      "learning_rate": 4.675567213873563e-05,
      "loss": 0.9034,
      "step": 2670
    },
    {
      "epoch": 0.8702016916070267,
      "grad_norm": 1.8213086128234863,
      "learning_rate": 4.6742946059752654e-05,
      "loss": 0.8876,
      "step": 2675
    },
    {
      "epoch": 0.8718282368249838,
      "grad_norm": 2.208648681640625,
      "learning_rate": 4.673019680911196e-05,
      "loss": 0.861,
      "step": 2680
    },
    {
      "epoch": 0.8734547820429408,
      "grad_norm": 2.068134307861328,
      "learning_rate": 4.671742440040053e-05,
      "loss": 0.8674,
      "step": 2685
    },
    {
      "epoch": 0.8750813272608978,
      "grad_norm": 3.2821731567382812,
      "learning_rate": 4.670462884723005e-05,
      "loss": 0.834,
      "step": 2690
    },
    {
      "epoch": 0.876707872478855,
      "grad_norm": 1.8073285818099976,
      "learning_rate": 4.6691810163236855e-05,
      "loss": 0.8688,
      "step": 2695
    },
    {
      "epoch": 0.878334417696812,
      "grad_norm": 2.1804659366607666,
      "learning_rate": 4.667896836208192e-05,
      "loss": 0.8713,
      "step": 2700
    },
    {
      "epoch": 0.879960962914769,
      "grad_norm": 2.100541830062866,
      "learning_rate": 4.66661034574509e-05,
      "loss": 0.8517,
      "step": 2705
    },
    {
      "epoch": 0.8815875081327261,
      "grad_norm": 1.5606156587600708,
      "learning_rate": 4.665321546305401e-05,
      "loss": 0.8812,
      "step": 2710
    },
    {
      "epoch": 0.8832140533506831,
      "grad_norm": 1.7539852857589722,
      "learning_rate": 4.664030439262613e-05,
      "loss": 0.8356,
      "step": 2715
    },
    {
      "epoch": 0.8848405985686402,
      "grad_norm": 2.1108126640319824,
      "learning_rate": 4.6627370259926695e-05,
      "loss": 0.8198,
      "step": 2720
    },
    {
      "epoch": 0.8864671437865973,
      "grad_norm": 2.120741605758667,
      "learning_rate": 4.661441307873973e-05,
      "loss": 0.8813,
      "step": 2725
    },
    {
      "epoch": 0.8880936890045543,
      "grad_norm": 1.7935503721237183,
      "learning_rate": 4.660143286287381e-05,
      "loss": 0.8686,
      "step": 2730
    },
    {
      "epoch": 0.8897202342225113,
      "grad_norm": 1.799494981765747,
      "learning_rate": 4.6588429626162095e-05,
      "loss": 0.8386,
      "step": 2735
    },
    {
      "epoch": 0.8913467794404685,
      "grad_norm": 1.954468846321106,
      "learning_rate": 4.657540338246224e-05,
      "loss": 0.8909,
      "step": 2740
    },
    {
      "epoch": 0.8929733246584255,
      "grad_norm": 2.7000679969787598,
      "learning_rate": 4.656235414565644e-05,
      "loss": 0.8463,
      "step": 2745
    },
    {
      "epoch": 0.8945998698763825,
      "grad_norm": 1.9565421342849731,
      "learning_rate": 4.654928192965137e-05,
      "loss": 0.8411,
      "step": 2750
    },
    {
      "epoch": 0.8962264150943396,
      "grad_norm": 2.5602779388427734,
      "learning_rate": 4.653618674837823e-05,
      "loss": 0.8649,
      "step": 2755
    },
    {
      "epoch": 0.8978529603122967,
      "grad_norm": 1.341413974761963,
      "learning_rate": 4.652306861579266e-05,
      "loss": 0.8423,
      "step": 2760
    },
    {
      "epoch": 0.8994795055302537,
      "grad_norm": 1.5747458934783936,
      "learning_rate": 4.6509927545874795e-05,
      "loss": 0.8435,
      "step": 2765
    },
    {
      "epoch": 0.9011060507482108,
      "grad_norm": 2.1025686264038086,
      "learning_rate": 4.6496763552629174e-05,
      "loss": 0.8145,
      "step": 2770
    },
    {
      "epoch": 0.9027325959661678,
      "grad_norm": 2.007817506790161,
      "learning_rate": 4.6483576650084795e-05,
      "loss": 0.9037,
      "step": 2775
    },
    {
      "epoch": 0.904359141184125,
      "grad_norm": 1.7015758752822876,
      "learning_rate": 4.6470366852295057e-05,
      "loss": 0.8736,
      "step": 2780
    },
    {
      "epoch": 0.905985686402082,
      "grad_norm": 1.8775756359100342,
      "learning_rate": 4.645713417333777e-05,
      "loss": 0.8791,
      "step": 2785
    },
    {
      "epoch": 0.907612231620039,
      "grad_norm": 1.6114087104797363,
      "learning_rate": 4.644387862731511e-05,
      "loss": 0.8418,
      "step": 2790
    },
    {
      "epoch": 0.9092387768379961,
      "grad_norm": 2.247753620147705,
      "learning_rate": 4.643060022835365e-05,
      "loss": 0.8993,
      "step": 2795
    },
    {
      "epoch": 0.9108653220559532,
      "grad_norm": 1.7884836196899414,
      "learning_rate": 4.641729899060429e-05,
      "loss": 0.8357,
      "step": 2800
    },
    {
      "epoch": 0.9124918672739102,
      "grad_norm": 8.958685874938965,
      "learning_rate": 4.6403974928242286e-05,
      "loss": 0.8406,
      "step": 2805
    },
    {
      "epoch": 0.9141184124918673,
      "grad_norm": 1.895634412765503,
      "learning_rate": 4.639062805546721e-05,
      "loss": 0.8973,
      "step": 2810
    },
    {
      "epoch": 0.9157449577098243,
      "grad_norm": 1.6189725399017334,
      "learning_rate": 4.6377258386502956e-05,
      "loss": 0.8472,
      "step": 2815
    },
    {
      "epoch": 0.9173715029277814,
      "grad_norm": 6.746943950653076,
      "learning_rate": 4.636386593559769e-05,
      "loss": 0.8247,
      "step": 2820
    },
    {
      "epoch": 0.9189980481457385,
      "grad_norm": 2.1239964962005615,
      "learning_rate": 4.6350450717023886e-05,
      "loss": 0.8693,
      "step": 2825
    },
    {
      "epoch": 0.9206245933636955,
      "grad_norm": 1.823723554611206,
      "learning_rate": 4.6337012745078254e-05,
      "loss": 0.8896,
      "step": 2830
    },
    {
      "epoch": 0.9222511385816525,
      "grad_norm": 1.6684235334396362,
      "learning_rate": 4.6323552034081773e-05,
      "loss": 0.8505,
      "step": 2835
    },
    {
      "epoch": 0.9238776837996097,
      "grad_norm": 1.6994017362594604,
      "learning_rate": 4.631006859837964e-05,
      "loss": 0.8624,
      "step": 2840
    },
    {
      "epoch": 0.9255042290175667,
      "grad_norm": 2.0857439041137695,
      "learning_rate": 4.629656245234127e-05,
      "loss": 0.8559,
      "step": 2845
    },
    {
      "epoch": 0.9271307742355237,
      "grad_norm": 2.246621608734131,
      "learning_rate": 4.628303361036031e-05,
      "loss": 0.8567,
      "step": 2850
    },
    {
      "epoch": 0.9287573194534808,
      "grad_norm": 1.9449462890625,
      "learning_rate": 4.626948208685454e-05,
      "loss": 0.8432,
      "step": 2855
    },
    {
      "epoch": 0.9303838646714379,
      "grad_norm": 2.311030864715576,
      "learning_rate": 4.625590789626597e-05,
      "loss": 0.8541,
      "step": 2860
    },
    {
      "epoch": 0.9320104098893949,
      "grad_norm": 2.238831043243408,
      "learning_rate": 4.6242311053060725e-05,
      "loss": 0.8271,
      "step": 2865
    },
    {
      "epoch": 0.933636955107352,
      "grad_norm": 2.0517778396606445,
      "learning_rate": 4.6228691571729095e-05,
      "loss": 0.8739,
      "step": 2870
    },
    {
      "epoch": 0.935263500325309,
      "grad_norm": 1.7728257179260254,
      "learning_rate": 4.621504946678548e-05,
      "loss": 0.8727,
      "step": 2875
    },
    {
      "epoch": 0.936890045543266,
      "grad_norm": 2.1137027740478516,
      "learning_rate": 4.62013847527684e-05,
      "loss": 0.8521,
      "step": 2880
    },
    {
      "epoch": 0.9385165907612232,
      "grad_norm": 2.272860050201416,
      "learning_rate": 4.6187697444240466e-05,
      "loss": 0.8431,
      "step": 2885
    },
    {
      "epoch": 0.9401431359791802,
      "grad_norm": 1.9327101707458496,
      "learning_rate": 4.6173987555788366e-05,
      "loss": 0.872,
      "step": 2890
    },
    {
      "epoch": 0.9417696811971373,
      "grad_norm": 1.739867925643921,
      "learning_rate": 4.6160255102022865e-05,
      "loss": 0.8599,
      "step": 2895
    },
    {
      "epoch": 0.9433962264150944,
      "grad_norm": 1.8884609937667847,
      "learning_rate": 4.614650009757875e-05,
      "loss": 0.8507,
      "step": 2900
    },
    {
      "epoch": 0.9450227716330514,
      "grad_norm": 2.3399882316589355,
      "learning_rate": 4.613272255711486e-05,
      "loss": 0.8606,
      "step": 2905
    },
    {
      "epoch": 0.9466493168510085,
      "grad_norm": 2.2510006427764893,
      "learning_rate": 4.611892249531405e-05,
      "loss": 0.8765,
      "step": 2910
    },
    {
      "epoch": 0.9482758620689655,
      "grad_norm": 1.770300269126892,
      "learning_rate": 4.610509992688319e-05,
      "loss": 0.8449,
      "step": 2915
    },
    {
      "epoch": 0.9499024072869225,
      "grad_norm": 2.5434508323669434,
      "learning_rate": 4.609125486655308e-05,
      "loss": 0.863,
      "step": 2920
    },
    {
      "epoch": 0.9515289525048797,
      "grad_norm": 1.7364468574523926,
      "learning_rate": 4.607738732907856e-05,
      "loss": 0.8432,
      "step": 2925
    },
    {
      "epoch": 0.9531554977228367,
      "grad_norm": 2.0038013458251953,
      "learning_rate": 4.606349732923837e-05,
      "loss": 0.8606,
      "step": 2930
    },
    {
      "epoch": 0.9547820429407937,
      "grad_norm": 1.9203795194625854,
      "learning_rate": 4.604958488183523e-05,
      "loss": 0.8497,
      "step": 2935
    },
    {
      "epoch": 0.9564085881587508,
      "grad_norm": 1.5324605703353882,
      "learning_rate": 4.603565000169576e-05,
      "loss": 0.8482,
      "step": 2940
    },
    {
      "epoch": 0.9580351333767079,
      "grad_norm": 1.6473455429077148,
      "learning_rate": 4.602169270367048e-05,
      "loss": 0.844,
      "step": 2945
    },
    {
      "epoch": 0.9596616785946649,
      "grad_norm": 1.987318992614746,
      "learning_rate": 4.600771300263381e-05,
      "loss": 0.8641,
      "step": 2950
    },
    {
      "epoch": 0.961288223812622,
      "grad_norm": 1.7535080909729004,
      "learning_rate": 4.5993710913484065e-05,
      "loss": 0.8919,
      "step": 2955
    },
    {
      "epoch": 0.962914769030579,
      "grad_norm": 1.718846082687378,
      "learning_rate": 4.597968645114338e-05,
      "loss": 0.8473,
      "step": 2960
    },
    {
      "epoch": 0.9645413142485361,
      "grad_norm": 1.794784426689148,
      "learning_rate": 4.596563963055775e-05,
      "loss": 0.8786,
      "step": 2965
    },
    {
      "epoch": 0.9661678594664932,
      "grad_norm": 1.9121543169021606,
      "learning_rate": 4.5951570466697034e-05,
      "loss": 0.846,
      "step": 2970
    },
    {
      "epoch": 0.9677944046844502,
      "grad_norm": 1.9233866930007935,
      "learning_rate": 4.593747897455484e-05,
      "loss": 0.8646,
      "step": 2975
    },
    {
      "epoch": 0.9694209499024072,
      "grad_norm": 2.0550642013549805,
      "learning_rate": 4.59233651691486e-05,
      "loss": 0.8518,
      "step": 2980
    },
    {
      "epoch": 0.9710474951203644,
      "grad_norm": 1.9614688158035278,
      "learning_rate": 4.590922906551954e-05,
      "loss": 0.8722,
      "step": 2985
    },
    {
      "epoch": 0.9726740403383214,
      "grad_norm": 1.5085740089416504,
      "learning_rate": 4.5895070678732635e-05,
      "loss": 0.8485,
      "step": 2990
    },
    {
      "epoch": 0.9743005855562785,
      "grad_norm": 2.253807306289673,
      "learning_rate": 4.58808900238766e-05,
      "loss": 0.8831,
      "step": 2995
    },
    {
      "epoch": 0.9759271307742355,
      "grad_norm": 2.406322956085205,
      "learning_rate": 4.586668711606389e-05,
      "loss": 0.8215,
      "step": 3000
    },
    {
      "epoch": 0.9775536759921926,
      "grad_norm": 1.6362812519073486,
      "learning_rate": 4.585246197043068e-05,
      "loss": 0.8557,
      "step": 3005
    },
    {
      "epoch": 0.9791802212101497,
      "grad_norm": 2.0213871002197266,
      "learning_rate": 4.583821460213684e-05,
      "loss": 0.8706,
      "step": 3010
    },
    {
      "epoch": 0.9808067664281067,
      "grad_norm": 2.181262254714966,
      "learning_rate": 4.582394502636591e-05,
      "loss": 0.8551,
      "step": 3015
    },
    {
      "epoch": 0.9824333116460637,
      "grad_norm": 1.651312232017517,
      "learning_rate": 4.580965325832512e-05,
      "loss": 0.8624,
      "step": 3020
    },
    {
      "epoch": 0.9840598568640209,
      "grad_norm": 1.6381468772888184,
      "learning_rate": 4.579533931324533e-05,
      "loss": 0.8456,
      "step": 3025
    },
    {
      "epoch": 0.9856864020819779,
      "grad_norm": 1.8208574056625366,
      "learning_rate": 4.578100320638106e-05,
      "loss": 0.864,
      "step": 3030
    },
    {
      "epoch": 0.9873129472999349,
      "grad_norm": 1.6688610315322876,
      "learning_rate": 4.5766644953010406e-05,
      "loss": 0.8426,
      "step": 3035
    },
    {
      "epoch": 0.988939492517892,
      "grad_norm": 2.131165027618408,
      "learning_rate": 4.5752264568435104e-05,
      "loss": 0.8771,
      "step": 3040
    },
    {
      "epoch": 0.9905660377358491,
      "grad_norm": 1.726283311843872,
      "learning_rate": 4.573786206798046e-05,
      "loss": 0.8573,
      "step": 3045
    },
    {
      "epoch": 0.9921925829538061,
      "grad_norm": 2.4873554706573486,
      "learning_rate": 4.5723437466995364e-05,
      "loss": 0.8448,
      "step": 3050
    },
    {
      "epoch": 0.9938191281717632,
      "grad_norm": 2.203801155090332,
      "learning_rate": 4.570899078085223e-05,
      "loss": 0.8839,
      "step": 3055
    },
    {
      "epoch": 0.9954456733897202,
      "grad_norm": 1.5567647218704224,
      "learning_rate": 4.5694522024947026e-05,
      "loss": 0.8411,
      "step": 3060
    },
    {
      "epoch": 0.9970722186076773,
      "grad_norm": 2.006624221801758,
      "learning_rate": 4.5680031214699245e-05,
      "loss": 0.8666,
      "step": 3065
    },
    {
      "epoch": 0.9986987638256344,
      "grad_norm": 2.026901960372925,
      "learning_rate": 4.5665518365551876e-05,
      "loss": 0.8677,
      "step": 3070
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.8113036588681131,
      "eval_loss": 0.416015625,
      "eval_precision": 0.8161222961369498,
      "eval_recall": 0.8094014323999219,
      "eval_runtime": 1031.3491,
      "eval_samples_per_second": 381.476,
      "eval_steps_per_second": 0.746,
      "step": 3074
    },
    {
      "epoch": 1.0003253090435915,
      "grad_norm": 1.9352390766143799,
      "learning_rate": 4.565098349297139e-05,
      "loss": 0.875,
      "step": 3075
    },
    {
      "epoch": 1.0019518542615484,
      "grad_norm": 1.7438503503799438,
      "learning_rate": 4.5636426612447735e-05,
      "loss": 0.8357,
      "step": 3080
    },
    {
      "epoch": 1.0035783994795056,
      "grad_norm": 1.6971402168273926,
      "learning_rate": 4.5621847739494315e-05,
      "loss": 0.8136,
      "step": 3085
    },
    {
      "epoch": 1.0052049446974627,
      "grad_norm": 1.4777500629425049,
      "learning_rate": 4.560724688964797e-05,
      "loss": 0.8161,
      "step": 3090
    },
    {
      "epoch": 1.0068314899154196,
      "grad_norm": 2.5030698776245117,
      "learning_rate": 4.559262407846896e-05,
      "loss": 0.7944,
      "step": 3095
    },
    {
      "epoch": 1.0084580351333767,
      "grad_norm": 1.8135157823562622,
      "learning_rate": 4.5577979321540956e-05,
      "loss": 0.7664,
      "step": 3100
    },
    {
      "epoch": 1.0100845803513339,
      "grad_norm": 2.1230430603027344,
      "learning_rate": 4.5563312634471e-05,
      "loss": 0.815,
      "step": 3105
    },
    {
      "epoch": 1.0117111255692908,
      "grad_norm": 1.7101008892059326,
      "learning_rate": 4.5548624032889515e-05,
      "loss": 0.8123,
      "step": 3110
    },
    {
      "epoch": 1.013337670787248,
      "grad_norm": 1.7563308477401733,
      "learning_rate": 4.553391353245028e-05,
      "loss": 0.8166,
      "step": 3115
    },
    {
      "epoch": 1.014964216005205,
      "grad_norm": 1.9305217266082764,
      "learning_rate": 4.551918114883042e-05,
      "loss": 0.8292,
      "step": 3120
    },
    {
      "epoch": 1.016590761223162,
      "grad_norm": 1.791092872619629,
      "learning_rate": 4.550442689773036e-05,
      "loss": 0.8163,
      "step": 3125
    },
    {
      "epoch": 1.018217306441119,
      "grad_norm": 1.7695600986480713,
      "learning_rate": 4.5489650794873853e-05,
      "loss": 0.7985,
      "step": 3130
    },
    {
      "epoch": 1.0198438516590762,
      "grad_norm": 2.181884765625,
      "learning_rate": 4.547485285600791e-05,
      "loss": 0.7809,
      "step": 3135
    },
    {
      "epoch": 1.0214703968770331,
      "grad_norm": 1.7219114303588867,
      "learning_rate": 4.546003309690285e-05,
      "loss": 0.8179,
      "step": 3140
    },
    {
      "epoch": 1.0230969420949902,
      "grad_norm": 1.8673876523971558,
      "learning_rate": 4.544519153335222e-05,
      "loss": 0.7763,
      "step": 3145
    },
    {
      "epoch": 1.0247234873129474,
      "grad_norm": 1.8142812252044678,
      "learning_rate": 4.54303281811728e-05,
      "loss": 0.8538,
      "step": 3150
    },
    {
      "epoch": 1.0263500325309043,
      "grad_norm": 1.7296979427337646,
      "learning_rate": 4.541544305620462e-05,
      "loss": 0.8141,
      "step": 3155
    },
    {
      "epoch": 1.0279765777488614,
      "grad_norm": 2.0520095825195312,
      "learning_rate": 4.5400536174310875e-05,
      "loss": 0.8305,
      "step": 3160
    },
    {
      "epoch": 1.0296031229668186,
      "grad_norm": 1.7631771564483643,
      "learning_rate": 4.538560755137798e-05,
      "loss": 0.7926,
      "step": 3165
    },
    {
      "epoch": 1.0312296681847755,
      "grad_norm": 2.381673812866211,
      "learning_rate": 4.537065720331551e-05,
      "loss": 0.7962,
      "step": 3170
    },
    {
      "epoch": 1.0328562134027326,
      "grad_norm": 1.8267555236816406,
      "learning_rate": 4.535568514605617e-05,
      "loss": 0.7908,
      "step": 3175
    },
    {
      "epoch": 1.0344827586206897,
      "grad_norm": 2.032332420349121,
      "learning_rate": 4.5340691395555835e-05,
      "loss": 0.8091,
      "step": 3180
    },
    {
      "epoch": 1.0361093038386466,
      "grad_norm": 1.757824420928955,
      "learning_rate": 4.532567596779348e-05,
      "loss": 0.8042,
      "step": 3185
    },
    {
      "epoch": 1.0377358490566038,
      "grad_norm": 2.0004148483276367,
      "learning_rate": 4.531063887877118e-05,
      "loss": 0.8403,
      "step": 3190
    },
    {
      "epoch": 1.039362394274561,
      "grad_norm": 2.3177530765533447,
      "learning_rate": 4.52955801445141e-05,
      "loss": 0.815,
      "step": 3195
    },
    {
      "epoch": 1.0409889394925178,
      "grad_norm": 1.7317501306533813,
      "learning_rate": 4.528049978107047e-05,
      "loss": 0.8243,
      "step": 3200
    },
    {
      "epoch": 1.042615484710475,
      "grad_norm": 1.8553853034973145,
      "learning_rate": 4.5265397804511575e-05,
      "loss": 0.8064,
      "step": 3205
    },
    {
      "epoch": 1.044242029928432,
      "grad_norm": 1.816109538078308,
      "learning_rate": 4.525027423093174e-05,
      "loss": 0.8166,
      "step": 3210
    },
    {
      "epoch": 1.045868575146389,
      "grad_norm": 1.9716473817825317,
      "learning_rate": 4.523512907644828e-05,
      "loss": 0.8125,
      "step": 3215
    },
    {
      "epoch": 1.047495120364346,
      "grad_norm": 2.2402427196502686,
      "learning_rate": 4.521996235720154e-05,
      "loss": 0.7953,
      "step": 3220
    },
    {
      "epoch": 1.0491216655823032,
      "grad_norm": 2.047614812850952,
      "learning_rate": 4.520477408935482e-05,
      "loss": 0.8102,
      "step": 3225
    },
    {
      "epoch": 1.0507482108002602,
      "grad_norm": 1.8117738962173462,
      "learning_rate": 4.51895642890944e-05,
      "loss": 0.8187,
      "step": 3230
    },
    {
      "epoch": 1.0523747560182173,
      "grad_norm": 1.726954460144043,
      "learning_rate": 4.517433297262951e-05,
      "loss": 0.8269,
      "step": 3235
    },
    {
      "epoch": 1.0540013012361744,
      "grad_norm": 2.0745491981506348,
      "learning_rate": 4.51590801561923e-05,
      "loss": 0.83,
      "step": 3240
    },
    {
      "epoch": 1.0556278464541313,
      "grad_norm": 1.9231599569320679,
      "learning_rate": 4.514380585603783e-05,
      "loss": 0.8036,
      "step": 3245
    },
    {
      "epoch": 1.0572543916720885,
      "grad_norm": 2.1803252696990967,
      "learning_rate": 4.512851008844408e-05,
      "loss": 0.8117,
      "step": 3250
    },
    {
      "epoch": 1.0588809368900456,
      "grad_norm": 1.8261420726776123,
      "learning_rate": 4.5113192869711867e-05,
      "loss": 0.8199,
      "step": 3255
    },
    {
      "epoch": 1.0605074821080025,
      "grad_norm": 1.6798607110977173,
      "learning_rate": 4.509785421616492e-05,
      "loss": 0.8419,
      "step": 3260
    },
    {
      "epoch": 1.0621340273259596,
      "grad_norm": 1.9406812191009521,
      "learning_rate": 4.508249414414975e-05,
      "loss": 0.8361,
      "step": 3265
    },
    {
      "epoch": 1.0637605725439168,
      "grad_norm": 1.4756301641464233,
      "learning_rate": 4.506711267003575e-05,
      "loss": 0.8085,
      "step": 3270
    },
    {
      "epoch": 1.065387117761874,
      "grad_norm": 1.636296272277832,
      "learning_rate": 4.50517098102151e-05,
      "loss": 0.8001,
      "step": 3275
    },
    {
      "epoch": 1.0670136629798308,
      "grad_norm": 3.0141994953155518,
      "learning_rate": 4.503628558110276e-05,
      "loss": 0.79,
      "step": 3280
    },
    {
      "epoch": 1.068640208197788,
      "grad_norm": 1.9245976209640503,
      "learning_rate": 4.502083999913648e-05,
      "loss": 0.8144,
      "step": 3285
    },
    {
      "epoch": 1.070266753415745,
      "grad_norm": 2.0785765647888184,
      "learning_rate": 4.500537308077675e-05,
      "loss": 0.8126,
      "step": 3290
    },
    {
      "epoch": 1.071893298633702,
      "grad_norm": 2.4258196353912354,
      "learning_rate": 4.498988484250681e-05,
      "loss": 0.8266,
      "step": 3295
    },
    {
      "epoch": 1.073519843851659,
      "grad_norm": 1.7474132776260376,
      "learning_rate": 4.4974375300832614e-05,
      "loss": 0.8128,
      "step": 3300
    },
    {
      "epoch": 1.0751463890696162,
      "grad_norm": 1.7980300188064575,
      "learning_rate": 4.4958844472282845e-05,
      "loss": 0.8257,
      "step": 3305
    },
    {
      "epoch": 1.0767729342875731,
      "grad_norm": 1.6487613916397095,
      "learning_rate": 4.494329237340882e-05,
      "loss": 0.7875,
      "step": 3310
    },
    {
      "epoch": 1.0783994795055303,
      "grad_norm": 2.1843864917755127,
      "learning_rate": 4.4927719020784575e-05,
      "loss": 0.8347,
      "step": 3315
    },
    {
      "epoch": 1.0800260247234874,
      "grad_norm": 1.9588794708251953,
      "learning_rate": 4.491212443100677e-05,
      "loss": 0.7839,
      "step": 3320
    },
    {
      "epoch": 1.0816525699414443,
      "grad_norm": 2.564725160598755,
      "learning_rate": 4.48965086206947e-05,
      "loss": 0.8277,
      "step": 3325
    },
    {
      "epoch": 1.0832791151594015,
      "grad_norm": 2.135406255722046,
      "learning_rate": 4.4880871606490274e-05,
      "loss": 0.766,
      "step": 3330
    },
    {
      "epoch": 1.0849056603773586,
      "grad_norm": 1.9069466590881348,
      "learning_rate": 4.4865213405058014e-05,
      "loss": 0.8192,
      "step": 3335
    },
    {
      "epoch": 1.0865322055953155,
      "grad_norm": 1.691408395767212,
      "learning_rate": 4.4849534033084996e-05,
      "loss": 0.801,
      "step": 3340
    },
    {
      "epoch": 1.0881587508132726,
      "grad_norm": 1.6413347721099854,
      "learning_rate": 4.4833833507280884e-05,
      "loss": 0.8187,
      "step": 3345
    },
    {
      "epoch": 1.0897852960312298,
      "grad_norm": 1.7701938152313232,
      "learning_rate": 4.481811184437786e-05,
      "loss": 0.8126,
      "step": 3350
    },
    {
      "epoch": 1.0914118412491867,
      "grad_norm": 2.274620294570923,
      "learning_rate": 4.480236906113066e-05,
      "loss": 0.8246,
      "step": 3355
    },
    {
      "epoch": 1.0930383864671438,
      "grad_norm": 1.8796124458312988,
      "learning_rate": 4.478660517431649e-05,
      "loss": 0.8227,
      "step": 3360
    },
    {
      "epoch": 1.094664931685101,
      "grad_norm": 2.04233455657959,
      "learning_rate": 4.477082020073509e-05,
      "loss": 0.8392,
      "step": 3365
    },
    {
      "epoch": 1.0962914769030578,
      "grad_norm": 1.7941282987594604,
      "learning_rate": 4.475501415720864e-05,
      "loss": 0.7835,
      "step": 3370
    },
    {
      "epoch": 1.097918022121015,
      "grad_norm": 2.058685302734375,
      "learning_rate": 4.47391870605818e-05,
      "loss": 0.7946,
      "step": 3375
    },
    {
      "epoch": 1.099544567338972,
      "grad_norm": 1.6518946886062622,
      "learning_rate": 4.4723338927721634e-05,
      "loss": 0.7946,
      "step": 3380
    },
    {
      "epoch": 1.101171112556929,
      "grad_norm": 2.23441481590271,
      "learning_rate": 4.4707469775517664e-05,
      "loss": 0.7921,
      "step": 3385
    },
    {
      "epoch": 1.1027976577748861,
      "grad_norm": 2.3232648372650146,
      "learning_rate": 4.469157962088177e-05,
      "loss": 0.853,
      "step": 3390
    },
    {
      "epoch": 1.1044242029928433,
      "grad_norm": 2.5304653644561768,
      "learning_rate": 4.467566848074825e-05,
      "loss": 0.8243,
      "step": 3395
    },
    {
      "epoch": 1.1060507482108002,
      "grad_norm": 1.7827421426773071,
      "learning_rate": 4.4659736372073744e-05,
      "loss": 0.7872,
      "step": 3400
    },
    {
      "epoch": 1.1076772934287573,
      "grad_norm": 1.954492211341858,
      "learning_rate": 4.4643783311837254e-05,
      "loss": 0.7794,
      "step": 3405
    },
    {
      "epoch": 1.1093038386467144,
      "grad_norm": 2.82947039604187,
      "learning_rate": 4.462780931704011e-05,
      "loss": 0.8311,
      "step": 3410
    },
    {
      "epoch": 1.1109303838646714,
      "grad_norm": 1.5368119478225708,
      "learning_rate": 4.461181440470592e-05,
      "loss": 0.7937,
      "step": 3415
    },
    {
      "epoch": 1.1125569290826285,
      "grad_norm": 1.84377121925354,
      "learning_rate": 4.4595798591880635e-05,
      "loss": 0.7949,
      "step": 3420
    },
    {
      "epoch": 1.1141834743005856,
      "grad_norm": 2.373713970184326,
      "learning_rate": 4.457976189563244e-05,
      "loss": 0.8294,
      "step": 3425
    },
    {
      "epoch": 1.1158100195185425,
      "grad_norm": 3.3402259349823,
      "learning_rate": 4.4563704333051795e-05,
      "loss": 0.8156,
      "step": 3430
    },
    {
      "epoch": 1.1174365647364997,
      "grad_norm": 2.2353696823120117,
      "learning_rate": 4.454762592125139e-05,
      "loss": 0.8355,
      "step": 3435
    },
    {
      "epoch": 1.1190631099544568,
      "grad_norm": 1.695486068725586,
      "learning_rate": 4.453152667736614e-05,
      "loss": 0.8266,
      "step": 3440
    },
    {
      "epoch": 1.1206896551724137,
      "grad_norm": 1.7904291152954102,
      "learning_rate": 4.451540661855315e-05,
      "loss": 0.8075,
      "step": 3445
    },
    {
      "epoch": 1.1223162003903708,
      "grad_norm": 3.15537166595459,
      "learning_rate": 4.449926576199173e-05,
      "loss": 0.8403,
      "step": 3450
    },
    {
      "epoch": 1.123942745608328,
      "grad_norm": 2.273641586303711,
      "learning_rate": 4.4483104124883324e-05,
      "loss": 0.827,
      "step": 3455
    },
    {
      "epoch": 1.1255692908262849,
      "grad_norm": 2.7670817375183105,
      "learning_rate": 4.4466921724451535e-05,
      "loss": 0.8333,
      "step": 3460
    },
    {
      "epoch": 1.127195836044242,
      "grad_norm": 1.8045217990875244,
      "learning_rate": 4.4450718577942114e-05,
      "loss": 0.7822,
      "step": 3465
    },
    {
      "epoch": 1.1288223812621991,
      "grad_norm": 3.31329345703125,
      "learning_rate": 4.443449470262289e-05,
      "loss": 0.8287,
      "step": 3470
    },
    {
      "epoch": 1.130448926480156,
      "grad_norm": 3.028852939605713,
      "learning_rate": 4.44182501157838e-05,
      "loss": 0.8143,
      "step": 3475
    },
    {
      "epoch": 1.1320754716981132,
      "grad_norm": 1.77473783493042,
      "learning_rate": 4.440198483473684e-05,
      "loss": 0.8133,
      "step": 3480
    },
    {
      "epoch": 1.1337020169160703,
      "grad_norm": 2.183030366897583,
      "learning_rate": 4.4385698876816095e-05,
      "loss": 0.7916,
      "step": 3485
    },
    {
      "epoch": 1.1353285621340272,
      "grad_norm": 1.941740870475769,
      "learning_rate": 4.436939225937764e-05,
      "loss": 0.7976,
      "step": 3490
    },
    {
      "epoch": 1.1369551073519844,
      "grad_norm": 1.8414844274520874,
      "learning_rate": 4.4353064999799583e-05,
      "loss": 0.834,
      "step": 3495
    },
    {
      "epoch": 1.1385816525699415,
      "grad_norm": 3.1121630668640137,
      "learning_rate": 4.4336717115482044e-05,
      "loss": 0.7853,
      "step": 3500
    },
    {
      "epoch": 1.1402081977878984,
      "grad_norm": 2.0056183338165283,
      "learning_rate": 4.432034862384712e-05,
      "loss": 0.8098,
      "step": 3505
    },
    {
      "epoch": 1.1418347430058555,
      "grad_norm": 4.22079610824585,
      "learning_rate": 4.4303959542338854e-05,
      "loss": 0.811,
      "step": 3510
    },
    {
      "epoch": 1.1434612882238127,
      "grad_norm": 8.928543090820312,
      "learning_rate": 4.428754988842324e-05,
      "loss": 0.8039,
      "step": 3515
    },
    {
      "epoch": 1.1450878334417696,
      "grad_norm": 2.357715368270874,
      "learning_rate": 4.4271119679588204e-05,
      "loss": 0.8244,
      "step": 3520
    },
    {
      "epoch": 1.1467143786597267,
      "grad_norm": 2.109389305114746,
      "learning_rate": 4.4254668933343566e-05,
      "loss": 0.7981,
      "step": 3525
    },
    {
      "epoch": 1.1483409238776838,
      "grad_norm": 1.7039803266525269,
      "learning_rate": 4.423819766722104e-05,
      "loss": 0.7762,
      "step": 3530
    },
    {
      "epoch": 1.1499674690956407,
      "grad_norm": 2.0489721298217773,
      "learning_rate": 4.4221705898774196e-05,
      "loss": 0.8192,
      "step": 3535
    },
    {
      "epoch": 1.1515940143135979,
      "grad_norm": 1.818631649017334,
      "learning_rate": 4.420519364557848e-05,
      "loss": 0.8002,
      "step": 3540
    },
    {
      "epoch": 1.153220559531555,
      "grad_norm": 1.7001547813415527,
      "learning_rate": 4.418866092523114e-05,
      "loss": 0.8348,
      "step": 3545
    },
    {
      "epoch": 1.1548471047495121,
      "grad_norm": 2.8383827209472656,
      "learning_rate": 4.4172107755351256e-05,
      "loss": 0.8103,
      "step": 3550
    },
    {
      "epoch": 1.156473649967469,
      "grad_norm": 2.1644821166992188,
      "learning_rate": 4.415553415357969e-05,
      "loss": 0.7796,
      "step": 3555
    },
    {
      "epoch": 1.1581001951854262,
      "grad_norm": 3.3231115341186523,
      "learning_rate": 4.4138940137579084e-05,
      "loss": 0.7834,
      "step": 3560
    },
    {
      "epoch": 1.1597267404033833,
      "grad_norm": 1.8891205787658691,
      "learning_rate": 4.412232572503383e-05,
      "loss": 0.8016,
      "step": 3565
    },
    {
      "epoch": 1.1613532856213402,
      "grad_norm": 1.6741869449615479,
      "learning_rate": 4.4105690933650055e-05,
      "loss": 0.8139,
      "step": 3570
    },
    {
      "epoch": 1.1629798308392973,
      "grad_norm": 1.8224472999572754,
      "learning_rate": 4.4089035781155624e-05,
      "loss": 0.805,
      "step": 3575
    },
    {
      "epoch": 1.1646063760572545,
      "grad_norm": 1.8925952911376953,
      "learning_rate": 4.4072360285300076e-05,
      "loss": 0.799,
      "step": 3580
    },
    {
      "epoch": 1.1662329212752114,
      "grad_norm": 1.6234843730926514,
      "learning_rate": 4.405566446385464e-05,
      "loss": 0.8237,
      "step": 3585
    },
    {
      "epoch": 1.1678594664931685,
      "grad_norm": 1.7579517364501953,
      "learning_rate": 4.4038948334612215e-05,
      "loss": 0.7959,
      "step": 3590
    },
    {
      "epoch": 1.1694860117111257,
      "grad_norm": 1.6992871761322021,
      "learning_rate": 4.4022211915387324e-05,
      "loss": 0.7827,
      "step": 3595
    },
    {
      "epoch": 1.1711125569290826,
      "grad_norm": 1.990018606185913,
      "learning_rate": 4.400545522401613e-05,
      "loss": 0.8171,
      "step": 3600
    },
    {
      "epoch": 1.1727391021470397,
      "grad_norm": 1.8367072343826294,
      "learning_rate": 4.3988678278356386e-05,
      "loss": 0.7709,
      "step": 3605
    },
    {
      "epoch": 1.1743656473649968,
      "grad_norm": 1.8837988376617432,
      "learning_rate": 4.3971881096287446e-05,
      "loss": 0.7936,
      "step": 3610
    },
    {
      "epoch": 1.1759921925829537,
      "grad_norm": 2.361079454421997,
      "learning_rate": 4.395506369571022e-05,
      "loss": 0.8119,
      "step": 3615
    },
    {
      "epoch": 1.1776187378009109,
      "grad_norm": 2.039238214492798,
      "learning_rate": 4.3938226094547156e-05,
      "loss": 0.8025,
      "step": 3620
    },
    {
      "epoch": 1.179245283018868,
      "grad_norm": 1.535110354423523,
      "learning_rate": 4.392136831074225e-05,
      "loss": 0.8029,
      "step": 3625
    },
    {
      "epoch": 1.180871828236825,
      "grad_norm": 1.8643568754196167,
      "learning_rate": 4.3904490362260994e-05,
      "loss": 0.8278,
      "step": 3630
    },
    {
      "epoch": 1.182498373454782,
      "grad_norm": 1.7051054239273071,
      "learning_rate": 4.388759226709038e-05,
      "loss": 0.803,
      "step": 3635
    },
    {
      "epoch": 1.1841249186727392,
      "grad_norm": 1.8260678052902222,
      "learning_rate": 4.3870674043238866e-05,
      "loss": 0.8136,
      "step": 3640
    },
    {
      "epoch": 1.185751463890696,
      "grad_norm": 2.171731948852539,
      "learning_rate": 4.3853735708736346e-05,
      "loss": 0.8347,
      "step": 3645
    },
    {
      "epoch": 1.1873780091086532,
      "grad_norm": 1.6170158386230469,
      "learning_rate": 4.3836777281634156e-05,
      "loss": 0.7931,
      "step": 3650
    },
    {
      "epoch": 1.1890045543266103,
      "grad_norm": 2.0424461364746094,
      "learning_rate": 4.381979878000506e-05,
      "loss": 0.8187,
      "step": 3655
    },
    {
      "epoch": 1.1906310995445673,
      "grad_norm": 1.7460720539093018,
      "learning_rate": 4.38028002219432e-05,
      "loss": 0.7913,
      "step": 3660
    },
    {
      "epoch": 1.1922576447625244,
      "grad_norm": 1.7360622882843018,
      "learning_rate": 4.378578162556409e-05,
      "loss": 0.8587,
      "step": 3665
    },
    {
      "epoch": 1.1938841899804815,
      "grad_norm": 1.6812613010406494,
      "learning_rate": 4.376874300900462e-05,
      "loss": 0.7885,
      "step": 3670
    },
    {
      "epoch": 1.1955107351984384,
      "grad_norm": 1.8114211559295654,
      "learning_rate": 4.3751684390422984e-05,
      "loss": 0.8399,
      "step": 3675
    },
    {
      "epoch": 1.1971372804163956,
      "grad_norm": 2.1626474857330322,
      "learning_rate": 4.3734605787998715e-05,
      "loss": 0.8353,
      "step": 3680
    },
    {
      "epoch": 1.1987638256343527,
      "grad_norm": 2.1298441886901855,
      "learning_rate": 4.371750721993264e-05,
      "loss": 0.7979,
      "step": 3685
    },
    {
      "epoch": 1.2003903708523098,
      "grad_norm": 2.7537834644317627,
      "learning_rate": 4.3700388704446856e-05,
      "loss": 0.8168,
      "step": 3690
    },
    {
      "epoch": 1.2020169160702667,
      "grad_norm": 2.0105204582214355,
      "learning_rate": 4.368325025978472e-05,
      "loss": 0.7984,
      "step": 3695
    },
    {
      "epoch": 1.2036434612882239,
      "grad_norm": 1.7766393423080444,
      "learning_rate": 4.3666091904210836e-05,
      "loss": 0.7903,
      "step": 3700
    },
    {
      "epoch": 1.205270006506181,
      "grad_norm": 1.8659816980361938,
      "learning_rate": 4.364891365601102e-05,
      "loss": 0.8235,
      "step": 3705
    },
    {
      "epoch": 1.206896551724138,
      "grad_norm": 1.7383310794830322,
      "learning_rate": 4.3631715533492296e-05,
      "loss": 0.8176,
      "step": 3710
    },
    {
      "epoch": 1.208523096942095,
      "grad_norm": 1.6841621398925781,
      "learning_rate": 4.361449755498285e-05,
      "loss": 0.8118,
      "step": 3715
    },
    {
      "epoch": 1.2101496421600522,
      "grad_norm": 1.9598522186279297,
      "learning_rate": 4.359725973883204e-05,
      "loss": 0.7958,
      "step": 3720
    },
    {
      "epoch": 1.211776187378009,
      "grad_norm": 2.602379560470581,
      "learning_rate": 4.3580002103410375e-05,
      "loss": 0.8183,
      "step": 3725
    },
    {
      "epoch": 1.2134027325959662,
      "grad_norm": 1.8115839958190918,
      "learning_rate": 4.356272466710947e-05,
      "loss": 0.7928,
      "step": 3730
    },
    {
      "epoch": 1.2150292778139233,
      "grad_norm": 1.8908313512802124,
      "learning_rate": 4.354542744834204e-05,
      "loss": 0.8089,
      "step": 3735
    },
    {
      "epoch": 1.2166558230318802,
      "grad_norm": 2.1180243492126465,
      "learning_rate": 4.352811046554191e-05,
      "loss": 0.8151,
      "step": 3740
    },
    {
      "epoch": 1.2182823682498374,
      "grad_norm": 1.7875014543533325,
      "learning_rate": 4.351077373716393e-05,
      "loss": 0.7996,
      "step": 3745
    },
    {
      "epoch": 1.2199089134677945,
      "grad_norm": 2.1548900604248047,
      "learning_rate": 4.349341728168402e-05,
      "loss": 0.7836,
      "step": 3750
    },
    {
      "epoch": 1.2215354586857514,
      "grad_norm": 1.9586331844329834,
      "learning_rate": 4.347604111759911e-05,
      "loss": 0.82,
      "step": 3755
    },
    {
      "epoch": 1.2231620039037086,
      "grad_norm": 1.666062593460083,
      "learning_rate": 4.3458645263427146e-05,
      "loss": 0.8127,
      "step": 3760
    },
    {
      "epoch": 1.2247885491216657,
      "grad_norm": 1.6794251203536987,
      "learning_rate": 4.3441229737707034e-05,
      "loss": 0.8092,
      "step": 3765
    },
    {
      "epoch": 1.2264150943396226,
      "grad_norm": 2.466075897216797,
      "learning_rate": 4.3423794558998675e-05,
      "loss": 0.7803,
      "step": 3770
    },
    {
      "epoch": 1.2280416395575797,
      "grad_norm": 1.6132360696792603,
      "learning_rate": 4.340633974588289e-05,
      "loss": 0.7889,
      "step": 3775
    },
    {
      "epoch": 1.2296681847755369,
      "grad_norm": 1.8046766519546509,
      "learning_rate": 4.338886531696145e-05,
      "loss": 0.8327,
      "step": 3780
    },
    {
      "epoch": 1.2312947299934938,
      "grad_norm": 2.1257879734039307,
      "learning_rate": 4.3371371290856996e-05,
      "loss": 0.8275,
      "step": 3785
    },
    {
      "epoch": 1.232921275211451,
      "grad_norm": 1.8193806409835815,
      "learning_rate": 4.335385768621307e-05,
      "loss": 0.8043,
      "step": 3790
    },
    {
      "epoch": 1.234547820429408,
      "grad_norm": 1.484757661819458,
      "learning_rate": 4.3336324521694104e-05,
      "loss": 0.836,
      "step": 3795
    },
    {
      "epoch": 1.236174365647365,
      "grad_norm": 1.6302132606506348,
      "learning_rate": 4.331877181598534e-05,
      "loss": 0.8157,
      "step": 3800
    },
    {
      "epoch": 1.237800910865322,
      "grad_norm": 3.4269461631774902,
      "learning_rate": 4.330119958779285e-05,
      "loss": 0.8175,
      "step": 3805
    },
    {
      "epoch": 1.2394274560832792,
      "grad_norm": 1.798671841621399,
      "learning_rate": 4.328360785584353e-05,
      "loss": 0.7987,
      "step": 3810
    },
    {
      "epoch": 1.241054001301236,
      "grad_norm": 4.2418413162231445,
      "learning_rate": 4.3265996638885054e-05,
      "loss": 0.8141,
      "step": 3815
    },
    {
      "epoch": 1.2426805465191932,
      "grad_norm": 1.7764673233032227,
      "learning_rate": 4.3248365955685856e-05,
      "loss": 0.8297,
      "step": 3820
    },
    {
      "epoch": 1.2443070917371504,
      "grad_norm": 1.755462408065796,
      "learning_rate": 4.32307158250351e-05,
      "loss": 0.819,
      "step": 3825
    },
    {
      "epoch": 1.2459336369551073,
      "grad_norm": 2.2177834510803223,
      "learning_rate": 4.321304626574271e-05,
      "loss": 0.7973,
      "step": 3830
    },
    {
      "epoch": 1.2475601821730644,
      "grad_norm": 2.8550915718078613,
      "learning_rate": 4.319535729663929e-05,
      "loss": 0.7799,
      "step": 3835
    },
    {
      "epoch": 1.2491867273910215,
      "grad_norm": 1.7983542680740356,
      "learning_rate": 4.317764893657616e-05,
      "loss": 0.7766,
      "step": 3840
    },
    {
      "epoch": 1.2508132726089785,
      "grad_norm": 2.1239964962005615,
      "learning_rate": 4.3159921204425246e-05,
      "loss": 0.8215,
      "step": 3845
    },
    {
      "epoch": 1.2524398178269356,
      "grad_norm": 2.1900477409362793,
      "learning_rate": 4.3142174119079175e-05,
      "loss": 0.8195,
      "step": 3850
    },
    {
      "epoch": 1.2540663630448927,
      "grad_norm": 1.7720361948013306,
      "learning_rate": 4.3124407699451174e-05,
      "loss": 0.8211,
      "step": 3855
    },
    {
      "epoch": 1.2556929082628496,
      "grad_norm": 2.2531039714813232,
      "learning_rate": 4.310662196447509e-05,
      "loss": 0.8173,
      "step": 3860
    },
    {
      "epoch": 1.2573194534808068,
      "grad_norm": 1.7785147428512573,
      "learning_rate": 4.3088816933105336e-05,
      "loss": 0.8015,
      "step": 3865
    },
    {
      "epoch": 1.258945998698764,
      "grad_norm": 1.6724070310592651,
      "learning_rate": 4.3070992624316895e-05,
      "loss": 0.8426,
      "step": 3870
    },
    {
      "epoch": 1.2605725439167208,
      "grad_norm": 1.6628553867340088,
      "learning_rate": 4.305314905710531e-05,
      "loss": 0.8194,
      "step": 3875
    },
    {
      "epoch": 1.262199089134678,
      "grad_norm": 2.1507434844970703,
      "learning_rate": 4.3035286250486616e-05,
      "loss": 0.8368,
      "step": 3880
    },
    {
      "epoch": 1.263825634352635,
      "grad_norm": 1.816701054573059,
      "learning_rate": 4.3017404223497385e-05,
      "loss": 0.8028,
      "step": 3885
    },
    {
      "epoch": 1.265452179570592,
      "grad_norm": 1.667349100112915,
      "learning_rate": 4.299950299519465e-05,
      "loss": 0.835,
      "step": 3890
    },
    {
      "epoch": 1.267078724788549,
      "grad_norm": 1.932996392250061,
      "learning_rate": 4.298158258465592e-05,
      "loss": 0.8239,
      "step": 3895
    },
    {
      "epoch": 1.2687052700065062,
      "grad_norm": 2.326045513153076,
      "learning_rate": 4.296364301097914e-05,
      "loss": 0.7889,
      "step": 3900
    },
    {
      "epoch": 1.2703318152244631,
      "grad_norm": 2.015550374984741,
      "learning_rate": 4.2945684293282685e-05,
      "loss": 0.8212,
      "step": 3905
    },
    {
      "epoch": 1.2719583604424203,
      "grad_norm": 1.7425355911254883,
      "learning_rate": 4.2927706450705305e-05,
      "loss": 0.8129,
      "step": 3910
    },
    {
      "epoch": 1.2735849056603774,
      "grad_norm": 1.7180896997451782,
      "learning_rate": 4.290970950240617e-05,
      "loss": 0.7754,
      "step": 3915
    },
    {
      "epoch": 1.2752114508783343,
      "grad_norm": 1.7110702991485596,
      "learning_rate": 4.28916934675648e-05,
      "loss": 0.8199,
      "step": 3920
    },
    {
      "epoch": 1.2768379960962914,
      "grad_norm": 1.8523958921432495,
      "learning_rate": 4.2873658365381026e-05,
      "loss": 0.8303,
      "step": 3925
    },
    {
      "epoch": 1.2784645413142486,
      "grad_norm": 1.8597928285598755,
      "learning_rate": 4.285560421507504e-05,
      "loss": 0.8032,
      "step": 3930
    },
    {
      "epoch": 1.2800910865322055,
      "grad_norm": 1.6834561824798584,
      "learning_rate": 4.2837531035887305e-05,
      "loss": 0.8038,
      "step": 3935
    },
    {
      "epoch": 1.2817176317501626,
      "grad_norm": 2.0416834354400635,
      "learning_rate": 4.281943884707859e-05,
      "loss": 0.8087,
      "step": 3940
    },
    {
      "epoch": 1.2833441769681198,
      "grad_norm": 1.819843053817749,
      "learning_rate": 4.280132766792989e-05,
      "loss": 0.8153,
      "step": 3945
    },
    {
      "epoch": 1.2849707221860767,
      "grad_norm": 1.82972252368927,
      "learning_rate": 4.2783197517742464e-05,
      "loss": 0.7974,
      "step": 3950
    },
    {
      "epoch": 1.2865972674040338,
      "grad_norm": 1.898067831993103,
      "learning_rate": 4.276504841583778e-05,
      "loss": 0.7882,
      "step": 3955
    },
    {
      "epoch": 1.288223812621991,
      "grad_norm": 2.071033477783203,
      "learning_rate": 4.27468803815575e-05,
      "loss": 0.8017,
      "step": 3960
    },
    {
      "epoch": 1.2898503578399478,
      "grad_norm": 1.8522893190383911,
      "learning_rate": 4.2728693434263476e-05,
      "loss": 0.8115,
      "step": 3965
    },
    {
      "epoch": 1.291476903057905,
      "grad_norm": 1.7110276222229004,
      "learning_rate": 4.2710487593337684e-05,
      "loss": 0.8063,
      "step": 3970
    },
    {
      "epoch": 1.293103448275862,
      "grad_norm": 1.731191873550415,
      "learning_rate": 4.269226287818228e-05,
      "loss": 0.7753,
      "step": 3975
    },
    {
      "epoch": 1.294729993493819,
      "grad_norm": 2.6229982376098633,
      "learning_rate": 4.2674019308219484e-05,
      "loss": 0.8478,
      "step": 3980
    },
    {
      "epoch": 1.2963565387117761,
      "grad_norm": 1.8522100448608398,
      "learning_rate": 4.2655756902891665e-05,
      "loss": 0.8106,
      "step": 3985
    },
    {
      "epoch": 1.2979830839297333,
      "grad_norm": 2.134326457977295,
      "learning_rate": 4.2637475681661214e-05,
      "loss": 0.8006,
      "step": 3990
    },
    {
      "epoch": 1.2996096291476902,
      "grad_norm": 2.10435152053833,
      "learning_rate": 4.261917566401061e-05,
      "loss": 0.857,
      "step": 3995
    },
    {
      "epoch": 1.3012361743656473,
      "grad_norm": 1.500496745109558,
      "learning_rate": 4.260085686944235e-05,
      "loss": 0.8273,
      "step": 4000
    },
    {
      "epoch": 1.3028627195836044,
      "grad_norm": 1.9462052583694458,
      "learning_rate": 4.258251931747893e-05,
      "loss": 0.7947,
      "step": 4005
    },
    {
      "epoch": 1.3044892648015614,
      "grad_norm": 3.217682361602783,
      "learning_rate": 4.256416302766286e-05,
      "loss": 0.7836,
      "step": 4010
    },
    {
      "epoch": 1.3061158100195185,
      "grad_norm": 1.6719838380813599,
      "learning_rate": 4.25457880195566e-05,
      "loss": 0.8,
      "step": 4015
    },
    {
      "epoch": 1.3077423552374756,
      "grad_norm": 2.3355538845062256,
      "learning_rate": 4.2527394312742574e-05,
      "loss": 0.8194,
      "step": 4020
    },
    {
      "epoch": 1.3093689004554325,
      "grad_norm": 2.0860061645507812,
      "learning_rate": 4.250898192682311e-05,
      "loss": 0.8147,
      "step": 4025
    },
    {
      "epoch": 1.3109954456733897,
      "grad_norm": 1.890384316444397,
      "learning_rate": 4.249055088142047e-05,
      "loss": 0.808,
      "step": 4030
    },
    {
      "epoch": 1.3126219908913468,
      "grad_norm": 2.144073009490967,
      "learning_rate": 4.247210119617679e-05,
      "loss": 0.7904,
      "step": 4035
    },
    {
      "epoch": 1.3142485361093037,
      "grad_norm": 2.1934149265289307,
      "learning_rate": 4.245363289075406e-05,
      "loss": 0.8062,
      "step": 4040
    },
    {
      "epoch": 1.3158750813272608,
      "grad_norm": 1.7943332195281982,
      "learning_rate": 4.243514598483412e-05,
      "loss": 0.8102,
      "step": 4045
    },
    {
      "epoch": 1.317501626545218,
      "grad_norm": 4.055437088012695,
      "learning_rate": 4.241664049811864e-05,
      "loss": 0.8231,
      "step": 4050
    },
    {
      "epoch": 1.319128171763175,
      "grad_norm": 2.4395828247070312,
      "learning_rate": 4.23981164503291e-05,
      "loss": 0.7925,
      "step": 4055
    },
    {
      "epoch": 1.320754716981132,
      "grad_norm": 2.233398675918579,
      "learning_rate": 4.237957386120674e-05,
      "loss": 0.8168,
      "step": 4060
    },
    {
      "epoch": 1.3223812621990891,
      "grad_norm": 2.865018606185913,
      "learning_rate": 4.236101275051256e-05,
      "loss": 0.8261,
      "step": 4065
    },
    {
      "epoch": 1.3240078074170463,
      "grad_norm": 2.397876501083374,
      "learning_rate": 4.234243313802732e-05,
      "loss": 0.812,
      "step": 4070
    },
    {
      "epoch": 1.3256343526350032,
      "grad_norm": 4.140904903411865,
      "learning_rate": 4.232383504355147e-05,
      "loss": 0.7997,
      "step": 4075
    },
    {
      "epoch": 1.3272608978529603,
      "grad_norm": 2.3326687812805176,
      "learning_rate": 4.230521848690517e-05,
      "loss": 0.7922,
      "step": 4080
    },
    {
      "epoch": 1.3288874430709174,
      "grad_norm": 2.385756731033325,
      "learning_rate": 4.228658348792828e-05,
      "loss": 0.8036,
      "step": 4085
    },
    {
      "epoch": 1.3305139882888743,
      "grad_norm": 2.274498224258423,
      "learning_rate": 4.2267930066480266e-05,
      "loss": 0.7874,
      "step": 4090
    },
    {
      "epoch": 1.3321405335068315,
      "grad_norm": 3.2354235649108887,
      "learning_rate": 4.224925824244025e-05,
      "loss": 0.8055,
      "step": 4095
    },
    {
      "epoch": 1.3337670787247886,
      "grad_norm": 1.8239346742630005,
      "learning_rate": 4.2230568035706987e-05,
      "loss": 0.8126,
      "step": 4100
    },
    {
      "epoch": 1.3353936239427457,
      "grad_norm": 1.9903501272201538,
      "learning_rate": 4.2211859466198785e-05,
      "loss": 0.7992,
      "step": 4105
    },
    {
      "epoch": 1.3370201691607027,
      "grad_norm": 1.6283726692199707,
      "learning_rate": 4.219313255385354e-05,
      "loss": 0.7975,
      "step": 4110
    },
    {
      "epoch": 1.3386467143786598,
      "grad_norm": 1.444446086883545,
      "learning_rate": 4.217438731862871e-05,
      "loss": 0.7741,
      "step": 4115
    },
    {
      "epoch": 1.340273259596617,
      "grad_norm": 2.0824830532073975,
      "learning_rate": 4.2155623780501236e-05,
      "loss": 0.8521,
      "step": 4120
    },
    {
      "epoch": 1.3418998048145738,
      "grad_norm": 2.0181331634521484,
      "learning_rate": 4.213684195946762e-05,
      "loss": 0.7845,
      "step": 4125
    },
    {
      "epoch": 1.343526350032531,
      "grad_norm": 2.0623695850372314,
      "learning_rate": 4.211804187554381e-05,
      "loss": 0.8115,
      "step": 4130
    },
    {
      "epoch": 1.345152895250488,
      "grad_norm": 1.8838248252868652,
      "learning_rate": 4.2099223548765224e-05,
      "loss": 0.8091,
      "step": 4135
    },
    {
      "epoch": 1.346779440468445,
      "grad_norm": 2.2108914852142334,
      "learning_rate": 4.208038699918674e-05,
      "loss": 0.8139,
      "step": 4140
    },
    {
      "epoch": 1.3484059856864021,
      "grad_norm": 1.8372282981872559,
      "learning_rate": 4.206153224688264e-05,
      "loss": 0.7977,
      "step": 4145
    },
    {
      "epoch": 1.3500325309043593,
      "grad_norm": 2.2820627689361572,
      "learning_rate": 4.2042659311946586e-05,
      "loss": 0.8083,
      "step": 4150
    },
    {
      "epoch": 1.3516590761223162,
      "grad_norm": 1.6137808561325073,
      "learning_rate": 4.202376821449167e-05,
      "loss": 0.8127,
      "step": 4155
    },
    {
      "epoch": 1.3532856213402733,
      "grad_norm": 1.8549315929412842,
      "learning_rate": 4.2004858974650285e-05,
      "loss": 0.8071,
      "step": 4160
    },
    {
      "epoch": 1.3549121665582304,
      "grad_norm": 1.7504771947860718,
      "learning_rate": 4.1985931612574186e-05,
      "loss": 0.8305,
      "step": 4165
    },
    {
      "epoch": 1.3565387117761873,
      "grad_norm": 1.92816162109375,
      "learning_rate": 4.196698614843445e-05,
      "loss": 0.8238,
      "step": 4170
    },
    {
      "epoch": 1.3581652569941445,
      "grad_norm": 4.2875471115112305,
      "learning_rate": 4.194802260242141e-05,
      "loss": 0.8139,
      "step": 4175
    },
    {
      "epoch": 1.3597918022121016,
      "grad_norm": 2.0751023292541504,
      "learning_rate": 4.192904099474472e-05,
      "loss": 0.8061,
      "step": 4180
    },
    {
      "epoch": 1.3614183474300585,
      "grad_norm": 1.8640562295913696,
      "learning_rate": 4.191004134563322e-05,
      "loss": 0.808,
      "step": 4185
    },
    {
      "epoch": 1.3630448926480156,
      "grad_norm": 1.9962856769561768,
      "learning_rate": 4.1891023675335044e-05,
      "loss": 0.8016,
      "step": 4190
    },
    {
      "epoch": 1.3646714378659728,
      "grad_norm": 1.7219542264938354,
      "learning_rate": 4.187198800411748e-05,
      "loss": 0.8172,
      "step": 4195
    },
    {
      "epoch": 1.3662979830839297,
      "grad_norm": 1.8055989742279053,
      "learning_rate": 4.1852934352267017e-05,
      "loss": 0.8162,
      "step": 4200
    },
    {
      "epoch": 1.3679245283018868,
      "grad_norm": 1.8981190919876099,
      "learning_rate": 4.183386274008932e-05,
      "loss": 0.8172,
      "step": 4205
    },
    {
      "epoch": 1.369551073519844,
      "grad_norm": 2.515291690826416,
      "learning_rate": 4.181477318790917e-05,
      "loss": 0.8164,
      "step": 4210
    },
    {
      "epoch": 1.3711776187378009,
      "grad_norm": 1.5880002975463867,
      "learning_rate": 4.1795665716070474e-05,
      "loss": 0.8087,
      "step": 4215
    },
    {
      "epoch": 1.372804163955758,
      "grad_norm": 1.719185709953308,
      "learning_rate": 4.177654034493626e-05,
      "loss": 0.801,
      "step": 4220
    },
    {
      "epoch": 1.3744307091737151,
      "grad_norm": 2.8107213973999023,
      "learning_rate": 4.1757397094888594e-05,
      "loss": 0.7675,
      "step": 4225
    },
    {
      "epoch": 1.376057254391672,
      "grad_norm": 1.931240200996399,
      "learning_rate": 4.173823598632862e-05,
      "loss": 0.824,
      "step": 4230
    },
    {
      "epoch": 1.3776837996096292,
      "grad_norm": 1.6347918510437012,
      "learning_rate": 4.1719057039676515e-05,
      "loss": 0.8057,
      "step": 4235
    },
    {
      "epoch": 1.3793103448275863,
      "grad_norm": 2.2919232845306396,
      "learning_rate": 4.1699860275371435e-05,
      "loss": 0.8068,
      "step": 4240
    },
    {
      "epoch": 1.3809368900455432,
      "grad_norm": 1.8959335088729858,
      "learning_rate": 4.168064571387159e-05,
      "loss": 0.7934,
      "step": 4245
    },
    {
      "epoch": 1.3825634352635003,
      "grad_norm": 1.9993014335632324,
      "learning_rate": 4.166141337565407e-05,
      "loss": 0.8189,
      "step": 4250
    },
    {
      "epoch": 1.3841899804814575,
      "grad_norm": 2.6744425296783447,
      "learning_rate": 4.1642163281214984e-05,
      "loss": 0.8283,
      "step": 4255
    },
    {
      "epoch": 1.3858165256994144,
      "grad_norm": 2.359461545944214,
      "learning_rate": 4.162289545106932e-05,
      "loss": 0.7969,
      "step": 4260
    },
    {
      "epoch": 1.3874430709173715,
      "grad_norm": 2.145277976989746,
      "learning_rate": 4.160360990575099e-05,
      "loss": 0.8002,
      "step": 4265
    },
    {
      "epoch": 1.3890696161353286,
      "grad_norm": 1.8672142028808594,
      "learning_rate": 4.1584306665812787e-05,
      "loss": 0.8007,
      "step": 4270
    },
    {
      "epoch": 1.3906961613532856,
      "grad_norm": 1.8362537622451782,
      "learning_rate": 4.156498575182633e-05,
      "loss": 0.8324,
      "step": 4275
    },
    {
      "epoch": 1.3923227065712427,
      "grad_norm": 2.507525682449341,
      "learning_rate": 4.15456471843821e-05,
      "loss": 0.8186,
      "step": 4280
    },
    {
      "epoch": 1.3939492517891998,
      "grad_norm": 1.6994293928146362,
      "learning_rate": 4.152629098408939e-05,
      "loss": 0.7999,
      "step": 4285
    },
    {
      "epoch": 1.3955757970071567,
      "grad_norm": 1.9243533611297607,
      "learning_rate": 4.1506917171576295e-05,
      "loss": 0.7933,
      "step": 4290
    },
    {
      "epoch": 1.3972023422251139,
      "grad_norm": 1.678754210472107,
      "learning_rate": 4.1487525767489635e-05,
      "loss": 0.7926,
      "step": 4295
    },
    {
      "epoch": 1.398828887443071,
      "grad_norm": 3.0569796562194824,
      "learning_rate": 4.146811679249504e-05,
      "loss": 0.8216,
      "step": 4300
    },
    {
      "epoch": 1.400455432661028,
      "grad_norm": 1.5882444381713867,
      "learning_rate": 4.144869026727681e-05,
      "loss": 0.8101,
      "step": 4305
    },
    {
      "epoch": 1.402081977878985,
      "grad_norm": 2.770890235900879,
      "learning_rate": 4.1429246212537974e-05,
      "loss": 0.7915,
      "step": 4310
    },
    {
      "epoch": 1.4037085230969422,
      "grad_norm": 4.253058433532715,
      "learning_rate": 4.140978464900025e-05,
      "loss": 0.8204,
      "step": 4315
    },
    {
      "epoch": 1.405335068314899,
      "grad_norm": 2.0877606868743896,
      "learning_rate": 4.1390305597404e-05,
      "loss": 0.8163,
      "step": 4320
    },
    {
      "epoch": 1.4069616135328562,
      "grad_norm": 1.743794560432434,
      "learning_rate": 4.137080907850823e-05,
      "loss": 0.8297,
      "step": 4325
    },
    {
      "epoch": 1.4085881587508133,
      "grad_norm": 1.6247589588165283,
      "learning_rate": 4.135129511309056e-05,
      "loss": 0.81,
      "step": 4330
    },
    {
      "epoch": 1.4102147039687702,
      "grad_norm": 5.552340984344482,
      "learning_rate": 4.13317637219472e-05,
      "loss": 0.8039,
      "step": 4335
    },
    {
      "epoch": 1.4118412491867274,
      "grad_norm": 1.9774831533432007,
      "learning_rate": 4.131221492589295e-05,
      "loss": 0.8447,
      "step": 4340
    },
    {
      "epoch": 1.4134677944046845,
      "grad_norm": 1.8310171365737915,
      "learning_rate": 4.129264874576111e-05,
      "loss": 0.7966,
      "step": 4345
    },
    {
      "epoch": 1.4150943396226414,
      "grad_norm": 2.037260055541992,
      "learning_rate": 4.127306520240356e-05,
      "loss": 0.793,
      "step": 4350
    },
    {
      "epoch": 1.4167208848405985,
      "grad_norm": 1.7230867147445679,
      "learning_rate": 4.125346431669065e-05,
      "loss": 0.8316,
      "step": 4355
    },
    {
      "epoch": 1.4183474300585557,
      "grad_norm": 3.2659876346588135,
      "learning_rate": 4.123384610951124e-05,
      "loss": 0.7975,
      "step": 4360
    },
    {
      "epoch": 1.4199739752765126,
      "grad_norm": 1.837768316268921,
      "learning_rate": 4.121421060177263e-05,
      "loss": 0.7825,
      "step": 4365
    },
    {
      "epoch": 1.4216005204944697,
      "grad_norm": 2.2004690170288086,
      "learning_rate": 4.1194557814400545e-05,
      "loss": 0.8181,
      "step": 4370
    },
    {
      "epoch": 1.4232270657124269,
      "grad_norm": 1.841178059577942,
      "learning_rate": 4.1174887768339164e-05,
      "loss": 0.7974,
      "step": 4375
    },
    {
      "epoch": 1.4248536109303838,
      "grad_norm": 2.3017239570617676,
      "learning_rate": 4.115520048455102e-05,
      "loss": 0.8045,
      "step": 4380
    },
    {
      "epoch": 1.426480156148341,
      "grad_norm": 1.8260972499847412,
      "learning_rate": 4.113549598401704e-05,
      "loss": 0.8175,
      "step": 4385
    },
    {
      "epoch": 1.428106701366298,
      "grad_norm": 2.360365867614746,
      "learning_rate": 4.111577428773649e-05,
      "loss": 0.8042,
      "step": 4390
    },
    {
      "epoch": 1.429733246584255,
      "grad_norm": 1.8616596460342407,
      "learning_rate": 4.1096035416726966e-05,
      "loss": 0.7756,
      "step": 4395
    },
    {
      "epoch": 1.431359791802212,
      "grad_norm": 1.970524787902832,
      "learning_rate": 4.107627939202435e-05,
      "loss": 0.8207,
      "step": 4400
    },
    {
      "epoch": 1.4329863370201692,
      "grad_norm": 1.5485093593597412,
      "learning_rate": 4.105650623468284e-05,
      "loss": 0.7761,
      "step": 4405
    },
    {
      "epoch": 1.434612882238126,
      "grad_norm": 1.799946665763855,
      "learning_rate": 4.103671596577486e-05,
      "loss": 0.8384,
      "step": 4410
    },
    {
      "epoch": 1.4362394274560832,
      "grad_norm": 2.041393518447876,
      "learning_rate": 4.101690860639108e-05,
      "loss": 0.8278,
      "step": 4415
    },
    {
      "epoch": 1.4378659726740404,
      "grad_norm": 1.733588457107544,
      "learning_rate": 4.09970841776404e-05,
      "loss": 0.7961,
      "step": 4420
    },
    {
      "epoch": 1.4394925178919973,
      "grad_norm": 2.301828145980835,
      "learning_rate": 4.097724270064988e-05,
      "loss": 0.8102,
      "step": 4425
    },
    {
      "epoch": 1.4411190631099544,
      "grad_norm": 1.7932064533233643,
      "learning_rate": 4.0961357258533774e-05,
      "loss": 0.7817,
      "step": 4430
    },
    {
      "epoch": 1.4427456083279115,
      "grad_norm": 1.8835421800613403,
      "learning_rate": 4.0941485148009765e-05,
      "loss": 0.7893,
      "step": 4435
    },
    {
      "epoch": 1.4443721535458685,
      "grad_norm": 2.5245280265808105,
      "learning_rate": 4.0921596048498315e-05,
      "loss": 0.8149,
      "step": 4440
    },
    {
      "epoch": 1.4459986987638256,
      "grad_norm": 3.859861135482788,
      "learning_rate": 4.090168998119542e-05,
      "loss": 0.814,
      "step": 4445
    },
    {
      "epoch": 1.4476252439817827,
      "grad_norm": 1.7895616292953491,
      "learning_rate": 4.088176696731517e-05,
      "loss": 0.8054,
      "step": 4450
    },
    {
      "epoch": 1.4492517891997396,
      "grad_norm": 2.096266508102417,
      "learning_rate": 4.08618270280897e-05,
      "loss": 0.8064,
      "step": 4455
    },
    {
      "epoch": 1.4508783344176968,
      "grad_norm": 4.13993501663208,
      "learning_rate": 4.084187018476918e-05,
      "loss": 0.7698,
      "step": 4460
    },
    {
      "epoch": 1.452504879635654,
      "grad_norm": 1.7063169479370117,
      "learning_rate": 4.0821896458621814e-05,
      "loss": 0.8107,
      "step": 4465
    },
    {
      "epoch": 1.4541314248536108,
      "grad_norm": 1.8808252811431885,
      "learning_rate": 4.0801905870933764e-05,
      "loss": 0.7758,
      "step": 4470
    },
    {
      "epoch": 1.455757970071568,
      "grad_norm": 2.3222527503967285,
      "learning_rate": 4.07818984430092e-05,
      "loss": 0.771,
      "step": 4475
    },
    {
      "epoch": 1.457384515289525,
      "grad_norm": 1.7792950868606567,
      "learning_rate": 4.076187419617024e-05,
      "loss": 0.8149,
      "step": 4480
    },
    {
      "epoch": 1.459011060507482,
      "grad_norm": 1.9383238554000854,
      "learning_rate": 4.074183315175686e-05,
      "loss": 0.8008,
      "step": 4485
    },
    {
      "epoch": 1.460637605725439,
      "grad_norm": 1.767006516456604,
      "learning_rate": 4.072177533112703e-05,
      "loss": 0.7871,
      "step": 4490
    },
    {
      "epoch": 1.4622641509433962,
      "grad_norm": 2.0811262130737305,
      "learning_rate": 4.0701700755656534e-05,
      "loss": 0.8408,
      "step": 4495
    },
    {
      "epoch": 1.4638906961613534,
      "grad_norm": 1.673139214515686,
      "learning_rate": 4.068160944673903e-05,
      "loss": 0.8747,
      "step": 4500
    },
    {
      "epoch": 1.4655172413793103,
      "grad_norm": 1.6479978561401367,
      "learning_rate": 4.066150142578602e-05,
      "loss": 0.8025,
      "step": 4505
    },
    {
      "epoch": 1.4671437865972674,
      "grad_norm": 5.902811050415039,
      "learning_rate": 4.0641376714226795e-05,
      "loss": 0.8073,
      "step": 4510
    },
    {
      "epoch": 1.4687703318152245,
      "grad_norm": 1.991217851638794,
      "learning_rate": 4.062123533350847e-05,
      "loss": 0.8249,
      "step": 4515
    },
    {
      "epoch": 1.4703968770331814,
      "grad_norm": 2.3157060146331787,
      "learning_rate": 4.060107730509587e-05,
      "loss": 0.8106,
      "step": 4520
    },
    {
      "epoch": 1.4720234222511386,
      "grad_norm": 1.7831164598464966,
      "learning_rate": 4.05809026504716e-05,
      "loss": 0.7792,
      "step": 4525
    },
    {
      "epoch": 1.4736499674690957,
      "grad_norm": 2.218675136566162,
      "learning_rate": 4.0560711391135986e-05,
      "loss": 0.7869,
      "step": 4530
    },
    {
      "epoch": 1.4752765126870526,
      "grad_norm": 1.9791754484176636,
      "learning_rate": 4.0540503548607035e-05,
      "loss": 0.8083,
      "step": 4535
    },
    {
      "epoch": 1.4769030579050098,
      "grad_norm": 1.804709553718567,
      "learning_rate": 4.052027914442043e-05,
      "loss": 0.8167,
      "step": 4540
    },
    {
      "epoch": 1.4785296031229669,
      "grad_norm": 1.798051357269287,
      "learning_rate": 4.050003820012948e-05,
      "loss": 0.8041,
      "step": 4545
    },
    {
      "epoch": 1.480156148340924,
      "grad_norm": 1.6213825941085815,
      "learning_rate": 4.047978073730519e-05,
      "loss": 0.8297,
      "step": 4550
    },
    {
      "epoch": 1.481782693558881,
      "grad_norm": 1.7084782123565674,
      "learning_rate": 4.045950677753611e-05,
      "loss": 0.7996,
      "step": 4555
    },
    {
      "epoch": 1.483409238776838,
      "grad_norm": 2.0501437187194824,
      "learning_rate": 4.043921634242836e-05,
      "loss": 0.7992,
      "step": 4560
    },
    {
      "epoch": 1.4850357839947952,
      "grad_norm": 1.5576157569885254,
      "learning_rate": 4.041890945360567e-05,
      "loss": 0.7909,
      "step": 4565
    },
    {
      "epoch": 1.486662329212752,
      "grad_norm": 1.9696029424667358,
      "learning_rate": 4.039858613270927e-05,
      "loss": 0.8363,
      "step": 4570
    },
    {
      "epoch": 1.4882888744307092,
      "grad_norm": 1.7343369722366333,
      "learning_rate": 4.037824640139791e-05,
      "loss": 0.8187,
      "step": 4575
    },
    {
      "epoch": 1.4899154196486664,
      "grad_norm": 1.8161100149154663,
      "learning_rate": 4.035789028134782e-05,
      "loss": 0.8193,
      "step": 4580
    },
    {
      "epoch": 1.4915419648666233,
      "grad_norm": 1.7704228162765503,
      "learning_rate": 4.033751779425272e-05,
      "loss": 0.7905,
      "step": 4585
    },
    {
      "epoch": 1.4931685100845804,
      "grad_norm": 1.9006760120391846,
      "learning_rate": 4.031712896182376e-05,
      "loss": 0.8017,
      "step": 4590
    },
    {
      "epoch": 1.4947950553025375,
      "grad_norm": 1.5826624631881714,
      "learning_rate": 4.029672380578948e-05,
      "loss": 0.7889,
      "step": 4595
    },
    {
      "epoch": 1.4964216005204944,
      "grad_norm": 1.7704415321350098,
      "learning_rate": 4.0276302347895864e-05,
      "loss": 0.8064,
      "step": 4600
    },
    {
      "epoch": 1.4980481457384516,
      "grad_norm": 1.7217965126037598,
      "learning_rate": 4.025586460990625e-05,
      "loss": 0.8,
      "step": 4605
    },
    {
      "epoch": 1.4996746909564087,
      "grad_norm": 1.4004994630813599,
      "learning_rate": 4.023541061360131e-05,
      "loss": 0.7898,
      "step": 4610
    },
    {
      "epoch": 1.5013012361743656,
      "grad_norm": 1.8781967163085938,
      "learning_rate": 4.021494038077907e-05,
      "loss": 0.7992,
      "step": 4615
    },
    {
      "epoch": 1.5029277813923227,
      "grad_norm": 2.0038421154022217,
      "learning_rate": 4.019445393325483e-05,
      "loss": 0.8055,
      "step": 4620
    },
    {
      "epoch": 1.5045543266102799,
      "grad_norm": 1.9267133474349976,
      "learning_rate": 4.01739512928612e-05,
      "loss": 0.8447,
      "step": 4625
    },
    {
      "epoch": 1.5061808718282368,
      "grad_norm": 1.6246354579925537,
      "learning_rate": 4.0153432481448027e-05,
      "loss": 0.816,
      "step": 4630
    },
    {
      "epoch": 1.507807417046194,
      "grad_norm": 1.8404632806777954,
      "learning_rate": 4.01328975208824e-05,
      "loss": 0.7858,
      "step": 4635
    },
    {
      "epoch": 1.509433962264151,
      "grad_norm": 1.8644263744354248,
      "learning_rate": 4.01123464330486e-05,
      "loss": 0.8302,
      "step": 4640
    },
    {
      "epoch": 1.511060507482108,
      "grad_norm": 1.5745633840560913,
      "learning_rate": 4.009177923984812e-05,
      "loss": 0.7719,
      "step": 4645
    },
    {
      "epoch": 1.512687052700065,
      "grad_norm": 1.6362204551696777,
      "learning_rate": 4.007119596319962e-05,
      "loss": 0.7955,
      "step": 4650
    },
    {
      "epoch": 1.5143135979180222,
      "grad_norm": 1.4978214502334595,
      "learning_rate": 4.005059662503888e-05,
      "loss": 0.8176,
      "step": 4655
    },
    {
      "epoch": 1.5159401431359791,
      "grad_norm": 1.6588901281356812,
      "learning_rate": 4.002998124731879e-05,
      "loss": 0.8118,
      "step": 4660
    },
    {
      "epoch": 1.5175666883539363,
      "grad_norm": 1.4084504842758179,
      "learning_rate": 4.000934985200937e-05,
      "loss": 0.8055,
      "step": 4665
    },
    {
      "epoch": 1.5191932335718934,
      "grad_norm": 1.8851374387741089,
      "learning_rate": 3.998870246109767e-05,
      "loss": 0.7762,
      "step": 4670
    },
    {
      "epoch": 1.5208197787898503,
      "grad_norm": 2.7642505168914795,
      "learning_rate": 3.996803909658782e-05,
      "loss": 0.7927,
      "step": 4675
    },
    {
      "epoch": 1.5224463240078074,
      "grad_norm": 1.984631896018982,
      "learning_rate": 3.994735978050094e-05,
      "loss": 0.8018,
      "step": 4680
    },
    {
      "epoch": 1.5240728692257646,
      "grad_norm": 1.5648013353347778,
      "learning_rate": 3.992666453487518e-05,
      "loss": 0.8135,
      "step": 4685
    },
    {
      "epoch": 1.5256994144437215,
      "grad_norm": 1.6731303930282593,
      "learning_rate": 3.990595338176564e-05,
      "loss": 0.8305,
      "step": 4690
    },
    {
      "epoch": 1.5273259596616786,
      "grad_norm": 2.1696157455444336,
      "learning_rate": 3.988522634324441e-05,
      "loss": 0.7958,
      "step": 4695
    },
    {
      "epoch": 1.5289525048796357,
      "grad_norm": 1.3574761152267456,
      "learning_rate": 3.986448344140047e-05,
      "loss": 0.7748,
      "step": 4700
    },
    {
      "epoch": 1.5305790500975927,
      "grad_norm": 1.553859829902649,
      "learning_rate": 3.984372469833972e-05,
      "loss": 0.794,
      "step": 4705
    },
    {
      "epoch": 1.5322055953155498,
      "grad_norm": 1.8226816654205322,
      "learning_rate": 3.9822950136184946e-05,
      "loss": 0.8153,
      "step": 4710
    },
    {
      "epoch": 1.533832140533507,
      "grad_norm": 1.8266657590866089,
      "learning_rate": 3.9802159777075796e-05,
      "loss": 0.8073,
      "step": 4715
    },
    {
      "epoch": 1.5354586857514638,
      "grad_norm": 1.4437830448150635,
      "learning_rate": 3.978135364316874e-05,
      "loss": 0.775,
      "step": 4720
    },
    {
      "epoch": 1.537085230969421,
      "grad_norm": 2.241546392440796,
      "learning_rate": 3.976053175663707e-05,
      "loss": 0.7779,
      "step": 4725
    },
    {
      "epoch": 1.538711776187378,
      "grad_norm": 1.7656699419021606,
      "learning_rate": 3.973969413967086e-05,
      "loss": 0.8373,
      "step": 4730
    },
    {
      "epoch": 1.540338321405335,
      "grad_norm": 1.451778531074524,
      "learning_rate": 3.971884081447695e-05,
      "loss": 0.8314,
      "step": 4735
    },
    {
      "epoch": 1.5419648666232921,
      "grad_norm": 1.667837381362915,
      "learning_rate": 3.9697971803278924e-05,
      "loss": 0.8282,
      "step": 4740
    },
    {
      "epoch": 1.5435914118412493,
      "grad_norm": 1.6840225458145142,
      "learning_rate": 3.967708712831707e-05,
      "loss": 0.7994,
      "step": 4745
    },
    {
      "epoch": 1.5452179570592062,
      "grad_norm": 1.5227510929107666,
      "learning_rate": 3.9656186811848395e-05,
      "loss": 0.7899,
      "step": 4750
    },
    {
      "epoch": 1.5468445022771633,
      "grad_norm": 1.7087949514389038,
      "learning_rate": 3.963527087614655e-05,
      "loss": 0.8272,
      "step": 4755
    },
    {
      "epoch": 1.5484710474951204,
      "grad_norm": 1.5343523025512695,
      "learning_rate": 3.9614339343501836e-05,
      "loss": 0.796,
      "step": 4760
    },
    {
      "epoch": 1.5500975927130773,
      "grad_norm": 1.9713935852050781,
      "learning_rate": 3.9593392236221176e-05,
      "loss": 0.7851,
      "step": 4765
    },
    {
      "epoch": 1.5517241379310345,
      "grad_norm": 1.7553633451461792,
      "learning_rate": 3.9572429576628114e-05,
      "loss": 0.7687,
      "step": 4770
    },
    {
      "epoch": 1.5533506831489916,
      "grad_norm": 1.773939847946167,
      "learning_rate": 3.955145138706273e-05,
      "loss": 0.8141,
      "step": 4775
    },
    {
      "epoch": 1.5549772283669485,
      "grad_norm": 1.7146141529083252,
      "learning_rate": 3.9530457689881684e-05,
      "loss": 0.7813,
      "step": 4780
    },
    {
      "epoch": 1.5566037735849056,
      "grad_norm": 2.0186431407928467,
      "learning_rate": 3.9509448507458146e-05,
      "loss": 0.7816,
      "step": 4785
    },
    {
      "epoch": 1.5582303188028628,
      "grad_norm": 1.6115589141845703,
      "learning_rate": 3.94884238621818e-05,
      "loss": 0.8232,
      "step": 4790
    },
    {
      "epoch": 1.5598568640208197,
      "grad_norm": 1.784650444984436,
      "learning_rate": 3.94673837764588e-05,
      "loss": 0.8161,
      "step": 4795
    },
    {
      "epoch": 1.5614834092387768,
      "grad_norm": 1.9830811023712158,
      "learning_rate": 3.944632827271176e-05,
      "loss": 0.8087,
      "step": 4800
    },
    {
      "epoch": 1.563109954456734,
      "grad_norm": 1.8103023767471313,
      "learning_rate": 3.942525737337973e-05,
      "loss": 0.7763,
      "step": 4805
    },
    {
      "epoch": 1.5647364996746909,
      "grad_norm": 1.7412750720977783,
      "learning_rate": 3.940417110091816e-05,
      "loss": 0.795,
      "step": 4810
    },
    {
      "epoch": 1.566363044892648,
      "grad_norm": 1.4648451805114746,
      "learning_rate": 3.9383069477798886e-05,
      "loss": 0.8255,
      "step": 4815
    },
    {
      "epoch": 1.5679895901106051,
      "grad_norm": 1.4472988843917847,
      "learning_rate": 3.9361952526510085e-05,
      "loss": 0.8215,
      "step": 4820
    },
    {
      "epoch": 1.569616135328562,
      "grad_norm": 1.5990488529205322,
      "learning_rate": 3.93408202695563e-05,
      "loss": 0.7938,
      "step": 4825
    },
    {
      "epoch": 1.5712426805465192,
      "grad_norm": 2.124126672744751,
      "learning_rate": 3.9319672729458376e-05,
      "loss": 0.8117,
      "step": 4830
    },
    {
      "epoch": 1.5728692257644763,
      "grad_norm": 1.7679290771484375,
      "learning_rate": 3.9298509928753434e-05,
      "loss": 0.7868,
      "step": 4835
    },
    {
      "epoch": 1.5744957709824332,
      "grad_norm": 2.1029303073883057,
      "learning_rate": 3.927733188999486e-05,
      "loss": 0.7573,
      "step": 4840
    },
    {
      "epoch": 1.5761223162003903,
      "grad_norm": 1.73438560962677,
      "learning_rate": 3.9256138635752304e-05,
      "loss": 0.8231,
      "step": 4845
    },
    {
      "epoch": 1.5777488614183475,
      "grad_norm": 1.4345791339874268,
      "learning_rate": 3.92349301886116e-05,
      "loss": 0.7866,
      "step": 4850
    },
    {
      "epoch": 1.5793754066363044,
      "grad_norm": 2.6696362495422363,
      "learning_rate": 3.921370657117478e-05,
      "loss": 0.8242,
      "step": 4855
    },
    {
      "epoch": 1.5810019518542615,
      "grad_norm": 2.566169261932373,
      "learning_rate": 3.9192467806060044e-05,
      "loss": 0.7749,
      "step": 4860
    },
    {
      "epoch": 1.5826284970722186,
      "grad_norm": 1.6439261436462402,
      "learning_rate": 3.917121391590176e-05,
      "loss": 0.8143,
      "step": 4865
    },
    {
      "epoch": 1.5842550422901756,
      "grad_norm": 1.8413790464401245,
      "learning_rate": 3.914994492335038e-05,
      "loss": 0.8113,
      "step": 4870
    },
    {
      "epoch": 1.5858815875081327,
      "grad_norm": 1.6771591901779175,
      "learning_rate": 3.912866085107247e-05,
      "loss": 0.7638,
      "step": 4875
    },
    {
      "epoch": 1.5875081327260898,
      "grad_norm": 1.6582680940628052,
      "learning_rate": 3.910736172175066e-05,
      "loss": 0.8137,
      "step": 4880
    },
    {
      "epoch": 1.5891346779440467,
      "grad_norm": 1.6781288385391235,
      "learning_rate": 3.908604755808363e-05,
      "loss": 0.7764,
      "step": 4885
    },
    {
      "epoch": 1.5907612231620039,
      "grad_norm": 2.101989507675171,
      "learning_rate": 3.9064718382786076e-05,
      "loss": 0.781,
      "step": 4890
    },
    {
      "epoch": 1.592387768379961,
      "grad_norm": 2.3528003692626953,
      "learning_rate": 3.90433742185887e-05,
      "loss": 0.8243,
      "step": 4895
    },
    {
      "epoch": 1.594014313597918,
      "grad_norm": 1.4553966522216797,
      "learning_rate": 3.9022015088238174e-05,
      "loss": 0.7994,
      "step": 4900
    },
    {
      "epoch": 1.595640858815875,
      "grad_norm": 1.6103534698486328,
      "learning_rate": 3.9000641014497124e-05,
      "loss": 0.8042,
      "step": 4905
    },
    {
      "epoch": 1.5972674040338322,
      "grad_norm": 1.993911623954773,
      "learning_rate": 3.897925202014409e-05,
      "loss": 0.8148,
      "step": 4910
    },
    {
      "epoch": 1.598893949251789,
      "grad_norm": 1.6934844255447388,
      "learning_rate": 3.895784812797352e-05,
      "loss": 0.7999,
      "step": 4915
    },
    {
      "epoch": 1.6005204944697464,
      "grad_norm": 1.602703332901001,
      "learning_rate": 3.8936429360795745e-05,
      "loss": 0.8024,
      "step": 4920
    },
    {
      "epoch": 1.6021470396877033,
      "grad_norm": 1.9552648067474365,
      "learning_rate": 3.891499574143693e-05,
      "loss": 0.8171,
      "step": 4925
    },
    {
      "epoch": 1.6037735849056602,
      "grad_norm": 1.9080613851547241,
      "learning_rate": 3.88935472927391e-05,
      "loss": 0.8218,
      "step": 4930
    },
    {
      "epoch": 1.6054001301236176,
      "grad_norm": 1.7507500648498535,
      "learning_rate": 3.887208403756005e-05,
      "loss": 0.8017,
      "step": 4935
    },
    {
      "epoch": 1.6070266753415745,
      "grad_norm": 1.6852693557739258,
      "learning_rate": 3.885060599877337e-05,
      "loss": 0.8115,
      "step": 4940
    },
    {
      "epoch": 1.6086532205595314,
      "grad_norm": 1.4053925275802612,
      "learning_rate": 3.8829113199268403e-05,
      "loss": 0.8327,
      "step": 4945
    },
    {
      "epoch": 1.6102797657774888,
      "grad_norm": 1.6863847970962524,
      "learning_rate": 3.880760566195023e-05,
      "loss": 0.7858,
      "step": 4950
    },
    {
      "epoch": 1.6119063109954457,
      "grad_norm": 1.752721905708313,
      "learning_rate": 3.878608340973962e-05,
      "loss": 0.7895,
      "step": 4955
    },
    {
      "epoch": 1.6135328562134026,
      "grad_norm": 2.294194221496582,
      "learning_rate": 3.876454646557305e-05,
      "loss": 0.7745,
      "step": 4960
    },
    {
      "epoch": 1.61515940143136,
      "grad_norm": 1.9643068313598633,
      "learning_rate": 3.8742994852402637e-05,
      "loss": 0.8077,
      "step": 4965
    },
    {
      "epoch": 1.6167859466493169,
      "grad_norm": 1.7443420886993408,
      "learning_rate": 3.872142859319612e-05,
      "loss": 0.7877,
      "step": 4970
    },
    {
      "epoch": 1.6184124918672738,
      "grad_norm": 2.0942933559417725,
      "learning_rate": 3.869984771093687e-05,
      "loss": 0.8167,
      "step": 4975
    },
    {
      "epoch": 1.6200390370852311,
      "grad_norm": 1.7708196640014648,
      "learning_rate": 3.867825222862383e-05,
      "loss": 0.8007,
      "step": 4980
    },
    {
      "epoch": 1.621665582303188,
      "grad_norm": 2.0296547412872314,
      "learning_rate": 3.8656642169271505e-05,
      "loss": 0.8063,
      "step": 4985
    },
    {
      "epoch": 1.623292127521145,
      "grad_norm": 1.888785481452942,
      "learning_rate": 3.863501755590994e-05,
      "loss": 0.8319,
      "step": 4990
    },
    {
      "epoch": 1.6249186727391023,
      "grad_norm": 1.4504474401474,
      "learning_rate": 3.8613378411584665e-05,
      "loss": 0.8158,
      "step": 4995
    },
    {
      "epoch": 1.6265452179570592,
      "grad_norm": 2.6302969455718994,
      "learning_rate": 3.8591724759356734e-05,
      "loss": 0.81,
      "step": 5000
    },
    {
      "epoch": 1.628171763175016,
      "grad_norm": 1.873052954673767,
      "learning_rate": 3.857005662230264e-05,
      "loss": 0.8089,
      "step": 5005
    },
    {
      "epoch": 1.6297983083929735,
      "grad_norm": 1.664070725440979,
      "learning_rate": 3.854837402351431e-05,
      "loss": 0.8248,
      "step": 5010
    },
    {
      "epoch": 1.6314248536109304,
      "grad_norm": 1.9076143503189087,
      "learning_rate": 3.85266769860991e-05,
      "loss": 0.8255,
      "step": 5015
    },
    {
      "epoch": 1.6330513988288873,
      "grad_norm": 2.2082643508911133,
      "learning_rate": 3.8504965533179724e-05,
      "loss": 0.783,
      "step": 5020
    },
    {
      "epoch": 1.6346779440468446,
      "grad_norm": 1.6590235233306885,
      "learning_rate": 3.84832396878943e-05,
      "loss": 0.7941,
      "step": 5025
    },
    {
      "epoch": 1.6363044892648015,
      "grad_norm": 1.7328729629516602,
      "learning_rate": 3.8461499473396246e-05,
      "loss": 0.8133,
      "step": 5030
    },
    {
      "epoch": 1.6379310344827587,
      "grad_norm": 3.048574209213257,
      "learning_rate": 3.843974491285432e-05,
      "loss": 0.804,
      "step": 5035
    },
    {
      "epoch": 1.6395575797007158,
      "grad_norm": 1.6420484781265259,
      "learning_rate": 3.841797602945254e-05,
      "loss": 0.7681,
      "step": 5040
    },
    {
      "epoch": 1.6411841249186727,
      "grad_norm": 6.178708553314209,
      "learning_rate": 3.839619284639022e-05,
      "loss": 0.7657,
      "step": 5045
    },
    {
      "epoch": 1.6428106701366298,
      "grad_norm": 1.622899055480957,
      "learning_rate": 3.837439538688189e-05,
      "loss": 0.7869,
      "step": 5050
    },
    {
      "epoch": 1.644437215354587,
      "grad_norm": 1.6254470348358154,
      "learning_rate": 3.8352583674157314e-05,
      "loss": 0.8116,
      "step": 5055
    },
    {
      "epoch": 1.6460637605725439,
      "grad_norm": 1.450502872467041,
      "learning_rate": 3.833075773146142e-05,
      "loss": 0.8245,
      "step": 5060
    },
    {
      "epoch": 1.647690305790501,
      "grad_norm": 1.6301724910736084,
      "learning_rate": 3.8308917582054324e-05,
      "loss": 0.77,
      "step": 5065
    },
    {
      "epoch": 1.6493168510084582,
      "grad_norm": 3.7551519870758057,
      "learning_rate": 3.828706324921128e-05,
      "loss": 0.7974,
      "step": 5070
    },
    {
      "epoch": 1.650943396226415,
      "grad_norm": 2.2062652111053467,
      "learning_rate": 3.826519475622265e-05,
      "loss": 0.8043,
      "step": 5075
    },
    {
      "epoch": 1.6525699414443722,
      "grad_norm": 1.5353881120681763,
      "learning_rate": 3.824331212639388e-05,
      "loss": 0.7968,
      "step": 5080
    },
    {
      "epoch": 1.6541964866623293,
      "grad_norm": 1.8680821657180786,
      "learning_rate": 3.822141538304549e-05,
      "loss": 0.7878,
      "step": 5085
    },
    {
      "epoch": 1.6558230318802862,
      "grad_norm": 1.8406147956848145,
      "learning_rate": 3.8199504549513055e-05,
      "loss": 0.7833,
      "step": 5090
    },
    {
      "epoch": 1.6574495770982434,
      "grad_norm": 1.6851632595062256,
      "learning_rate": 3.817757964914713e-05,
      "loss": 0.8122,
      "step": 5095
    },
    {
      "epoch": 1.6590761223162005,
      "grad_norm": 1.4229167699813843,
      "learning_rate": 3.81556407053133e-05,
      "loss": 0.8089,
      "step": 5100
    },
    {
      "epoch": 1.6607026675341574,
      "grad_norm": 1.7230238914489746,
      "learning_rate": 3.81336877413921e-05,
      "loss": 0.8054,
      "step": 5105
    },
    {
      "epoch": 1.6623292127521145,
      "grad_norm": 1.6220729351043701,
      "learning_rate": 3.811172078077899e-05,
      "loss": 0.7732,
      "step": 5110
    },
    {
      "epoch": 1.6639557579700717,
      "grad_norm": 1.9733960628509521,
      "learning_rate": 3.808973984688439e-05,
      "loss": 0.7928,
      "step": 5115
    },
    {
      "epoch": 1.6655823031880286,
      "grad_norm": 1.9078205823898315,
      "learning_rate": 3.806774496313355e-05,
      "loss": 0.8029,
      "step": 5120
    },
    {
      "epoch": 1.6672088484059857,
      "grad_norm": 2.955739736557007,
      "learning_rate": 3.8045736152966635e-05,
      "loss": 0.7885,
      "step": 5125
    },
    {
      "epoch": 1.6688353936239428,
      "grad_norm": 2.7966153621673584,
      "learning_rate": 3.802371343983865e-05,
      "loss": 0.7869,
      "step": 5130
    },
    {
      "epoch": 1.6704619388418998,
      "grad_norm": 1.9048030376434326,
      "learning_rate": 3.800167684721938e-05,
      "loss": 0.7971,
      "step": 5135
    },
    {
      "epoch": 1.6720884840598569,
      "grad_norm": 1.6827515363693237,
      "learning_rate": 3.797962639859344e-05,
      "loss": 0.7805,
      "step": 5140
    },
    {
      "epoch": 1.673715029277814,
      "grad_norm": 1.5637474060058594,
      "learning_rate": 3.7957562117460187e-05,
      "loss": 0.7994,
      "step": 5145
    },
    {
      "epoch": 1.675341574495771,
      "grad_norm": 1.4919729232788086,
      "learning_rate": 3.7935484027333746e-05,
      "loss": 0.807,
      "step": 5150
    },
    {
      "epoch": 1.676968119713728,
      "grad_norm": 1.6017789840698242,
      "learning_rate": 3.7913392151742924e-05,
      "loss": 0.8152,
      "step": 5155
    },
    {
      "epoch": 1.6785946649316852,
      "grad_norm": 2.5162606239318848,
      "learning_rate": 3.7891286514231225e-05,
      "loss": 0.7919,
      "step": 5160
    },
    {
      "epoch": 1.680221210149642,
      "grad_norm": 1.5875083208084106,
      "learning_rate": 3.786916713835685e-05,
      "loss": 0.8212,
      "step": 5165
    },
    {
      "epoch": 1.6818477553675992,
      "grad_norm": 1.7888031005859375,
      "learning_rate": 3.784703404769263e-05,
      "loss": 0.7905,
      "step": 5170
    },
    {
      "epoch": 1.6834743005855564,
      "grad_norm": 1.5636552572250366,
      "learning_rate": 3.782488726582598e-05,
      "loss": 0.7896,
      "step": 5175
    },
    {
      "epoch": 1.6851008458035133,
      "grad_norm": 2.8158295154571533,
      "learning_rate": 3.780272681635894e-05,
      "loss": 0.7746,
      "step": 5180
    },
    {
      "epoch": 1.6867273910214704,
      "grad_norm": 1.6401851177215576,
      "learning_rate": 3.77805527229081e-05,
      "loss": 0.7874,
      "step": 5185
    },
    {
      "epoch": 1.6883539362394275,
      "grad_norm": 2.0426077842712402,
      "learning_rate": 3.77583650091046e-05,
      "loss": 0.7929,
      "step": 5190
    },
    {
      "epoch": 1.6899804814573844,
      "grad_norm": 1.6837797164916992,
      "learning_rate": 3.7736163698594094e-05,
      "loss": 0.8331,
      "step": 5195
    },
    {
      "epoch": 1.6916070266753416,
      "grad_norm": 1.5901471376419067,
      "learning_rate": 3.771394881503673e-05,
      "loss": 0.7882,
      "step": 5200
    },
    {
      "epoch": 1.6932335718932987,
      "grad_norm": 2.020697593688965,
      "learning_rate": 3.7691720382107084e-05,
      "loss": 0.8056,
      "step": 5205
    },
    {
      "epoch": 1.6948601171112556,
      "grad_norm": 1.6831104755401611,
      "learning_rate": 3.766947842349423e-05,
      "loss": 0.8024,
      "step": 5210
    },
    {
      "epoch": 1.6964866623292127,
      "grad_norm": 1.7317795753479004,
      "learning_rate": 3.764722296290162e-05,
      "loss": 0.8069,
      "step": 5215
    },
    {
      "epoch": 1.6981132075471699,
      "grad_norm": 1.4958361387252808,
      "learning_rate": 3.76249540240471e-05,
      "loss": 0.8255,
      "step": 5220
    },
    {
      "epoch": 1.6997397527651268,
      "grad_norm": 1.7042391300201416,
      "learning_rate": 3.7602671630662886e-05,
      "loss": 0.8126,
      "step": 5225
    },
    {
      "epoch": 1.701366297983084,
      "grad_norm": 1.7775369882583618,
      "learning_rate": 3.7580375806495524e-05,
      "loss": 0.7854,
      "step": 5230
    },
    {
      "epoch": 1.702992843201041,
      "grad_norm": 1.6327508687973022,
      "learning_rate": 3.755806657530589e-05,
      "loss": 0.8067,
      "step": 5235
    },
    {
      "epoch": 1.704619388418998,
      "grad_norm": 1.5772026777267456,
      "learning_rate": 3.753574396086913e-05,
      "loss": 0.7861,
      "step": 5240
    },
    {
      "epoch": 1.706245933636955,
      "grad_norm": 1.642557144165039,
      "learning_rate": 3.751340798697466e-05,
      "loss": 0.7904,
      "step": 5245
    },
    {
      "epoch": 1.7078724788549122,
      "grad_norm": 1.9723272323608398,
      "learning_rate": 3.7491058677426135e-05,
      "loss": 0.8274,
      "step": 5250
    },
    {
      "epoch": 1.7094990240728691,
      "grad_norm": 1.5701051950454712,
      "learning_rate": 3.7468696056041406e-05,
      "loss": 0.8231,
      "step": 5255
    },
    {
      "epoch": 1.7111255692908263,
      "grad_norm": 1.5915799140930176,
      "learning_rate": 3.7446320146652556e-05,
      "loss": 0.8006,
      "step": 5260
    },
    {
      "epoch": 1.7127521145087834,
      "grad_norm": 1.9272072315216064,
      "learning_rate": 3.7423930973105766e-05,
      "loss": 0.7917,
      "step": 5265
    },
    {
      "epoch": 1.7143786597267403,
      "grad_norm": 2.022019386291504,
      "learning_rate": 3.740152855926139e-05,
      "loss": 0.8224,
      "step": 5270
    },
    {
      "epoch": 1.7160052049446974,
      "grad_norm": 1.566528081893921,
      "learning_rate": 3.7379112928993904e-05,
      "loss": 0.7587,
      "step": 5275
    },
    {
      "epoch": 1.7176317501626546,
      "grad_norm": 2.751673936843872,
      "learning_rate": 3.735668410619183e-05,
      "loss": 0.7843,
      "step": 5280
    },
    {
      "epoch": 1.7192582953806115,
      "grad_norm": 2.7237708568573,
      "learning_rate": 3.73342421147578e-05,
      "loss": 0.7954,
      "step": 5285
    },
    {
      "epoch": 1.7208848405985686,
      "grad_norm": 1.8551533222198486,
      "learning_rate": 3.7311786978608415e-05,
      "loss": 0.8091,
      "step": 5290
    },
    {
      "epoch": 1.7225113858165257,
      "grad_norm": 1.67500638961792,
      "learning_rate": 3.7289318721674346e-05,
      "loss": 0.8154,
      "step": 5295
    },
    {
      "epoch": 1.7241379310344827,
      "grad_norm": 1.9620552062988281,
      "learning_rate": 3.726683736790022e-05,
      "loss": 0.7943,
      "step": 5300
    },
    {
      "epoch": 1.7257644762524398,
      "grad_norm": 2.0189971923828125,
      "learning_rate": 3.72443429412446e-05,
      "loss": 0.7814,
      "step": 5305
    },
    {
      "epoch": 1.727391021470397,
      "grad_norm": 1.632895827293396,
      "learning_rate": 3.7221835465680024e-05,
      "loss": 0.7814,
      "step": 5310
    },
    {
      "epoch": 1.7290175666883538,
      "grad_norm": 1.7532051801681519,
      "learning_rate": 3.719931496519291e-05,
      "loss": 0.8273,
      "step": 5315
    },
    {
      "epoch": 1.730644111906311,
      "grad_norm": 1.4986096620559692,
      "learning_rate": 3.717678146378357e-05,
      "loss": 0.7931,
      "step": 5320
    },
    {
      "epoch": 1.732270657124268,
      "grad_norm": 1.7292662858963013,
      "learning_rate": 3.7154234985466155e-05,
      "loss": 0.8032,
      "step": 5325
    },
    {
      "epoch": 1.733897202342225,
      "grad_norm": 2.197148084640503,
      "learning_rate": 3.7131675554268654e-05,
      "loss": 0.7974,
      "step": 5330
    },
    {
      "epoch": 1.7355237475601821,
      "grad_norm": 1.840177297592163,
      "learning_rate": 3.7109103194232856e-05,
      "loss": 0.8385,
      "step": 5335
    },
    {
      "epoch": 1.7371502927781393,
      "grad_norm": 1.728795051574707,
      "learning_rate": 3.7086517929414346e-05,
      "loss": 0.7905,
      "step": 5340
    },
    {
      "epoch": 1.7387768379960962,
      "grad_norm": 3.702511787414551,
      "learning_rate": 3.706391978388245e-05,
      "loss": 0.8194,
      "step": 5345
    },
    {
      "epoch": 1.7404033832140533,
      "grad_norm": 1.6523361206054688,
      "learning_rate": 3.70413087817202e-05,
      "loss": 0.8152,
      "step": 5350
    },
    {
      "epoch": 1.7420299284320104,
      "grad_norm": 1.5317710638046265,
      "learning_rate": 3.701868494702437e-05,
      "loss": 0.7935,
      "step": 5355
    },
    {
      "epoch": 1.7436564736499673,
      "grad_norm": 1.8728350400924683,
      "learning_rate": 3.699604830390537e-05,
      "loss": 0.7902,
      "step": 5360
    },
    {
      "epoch": 1.7452830188679245,
      "grad_norm": 1.674080491065979,
      "learning_rate": 3.69733988764873e-05,
      "loss": 0.7796,
      "step": 5365
    },
    {
      "epoch": 1.7469095640858816,
      "grad_norm": 2.6330764293670654,
      "learning_rate": 3.695073668890785e-05,
      "loss": 0.8109,
      "step": 5370
    },
    {
      "epoch": 1.7485361093038385,
      "grad_norm": 1.6176660060882568,
      "learning_rate": 3.692806176531832e-05,
      "loss": 0.7639,
      "step": 5375
    },
    {
      "epoch": 1.7501626545217959,
      "grad_norm": 1.6828603744506836,
      "learning_rate": 3.690537412988359e-05,
      "loss": 0.8128,
      "step": 5380
    },
    {
      "epoch": 1.7517891997397528,
      "grad_norm": 1.547345757484436,
      "learning_rate": 3.688267380678208e-05,
      "loss": 0.852,
      "step": 5385
    },
    {
      "epoch": 1.7534157449577097,
      "grad_norm": 1.802438735961914,
      "learning_rate": 3.685996082020574e-05,
      "loss": 0.7841,
      "step": 5390
    },
    {
      "epoch": 1.755042290175667,
      "grad_norm": 2.5347976684570312,
      "learning_rate": 3.683723519436e-05,
      "loss": 0.8074,
      "step": 5395
    },
    {
      "epoch": 1.756668835393624,
      "grad_norm": 1.9095966815948486,
      "learning_rate": 3.681449695346376e-05,
      "loss": 0.7791,
      "step": 5400
    },
    {
      "epoch": 1.7582953806115809,
      "grad_norm": 2.8300509452819824,
      "learning_rate": 3.67917461217494e-05,
      "loss": 0.7851,
      "step": 5405
    },
    {
      "epoch": 1.7599219258295382,
      "grad_norm": 2.3183364868164062,
      "learning_rate": 3.676898272346266e-05,
      "loss": 0.768,
      "step": 5410
    },
    {
      "epoch": 1.7615484710474951,
      "grad_norm": 1.687774896621704,
      "learning_rate": 3.674620678286273e-05,
      "loss": 0.7968,
      "step": 5415
    },
    {
      "epoch": 1.763175016265452,
      "grad_norm": 1.9604495763778687,
      "learning_rate": 3.6723418324222126e-05,
      "loss": 0.8386,
      "step": 5420
    },
    {
      "epoch": 1.7648015614834094,
      "grad_norm": 1.6525349617004395,
      "learning_rate": 3.670061737182672e-05,
      "loss": 0.8047,
      "step": 5425
    },
    {
      "epoch": 1.7664281067013663,
      "grad_norm": 1.938249945640564,
      "learning_rate": 3.667780394997569e-05,
      "loss": 0.7749,
      "step": 5430
    },
    {
      "epoch": 1.7680546519193232,
      "grad_norm": 3.016706943511963,
      "learning_rate": 3.6654978082981514e-05,
      "loss": 0.8117,
      "step": 5435
    },
    {
      "epoch": 1.7696811971372806,
      "grad_norm": 1.8364930152893066,
      "learning_rate": 3.663213979516994e-05,
      "loss": 0.8016,
      "step": 5440
    },
    {
      "epoch": 1.7713077423552375,
      "grad_norm": 2.14928936958313,
      "learning_rate": 3.660928911087993e-05,
      "loss": 0.796,
      "step": 5445
    },
    {
      "epoch": 1.7729342875731944,
      "grad_norm": 2.1587629318237305,
      "learning_rate": 3.658642605446367e-05,
      "loss": 0.7777,
      "step": 5450
    },
    {
      "epoch": 1.7745608327911517,
      "grad_norm": 1.730048418045044,
      "learning_rate": 3.6563550650286526e-05,
      "loss": 0.7868,
      "step": 5455
    },
    {
      "epoch": 1.7761873780091086,
      "grad_norm": 1.746994137763977,
      "learning_rate": 3.6540662922727034e-05,
      "loss": 0.7828,
      "step": 5460
    },
    {
      "epoch": 1.7778139232270656,
      "grad_norm": 1.678442358970642,
      "learning_rate": 3.651776289617685e-05,
      "loss": 0.8107,
      "step": 5465
    },
    {
      "epoch": 1.779440468445023,
      "grad_norm": 1.93441641330719,
      "learning_rate": 3.6494850595040745e-05,
      "loss": 0.7975,
      "step": 5470
    },
    {
      "epoch": 1.7810670136629798,
      "grad_norm": 2.052583932876587,
      "learning_rate": 3.647192604373658e-05,
      "loss": 0.8173,
      "step": 5475
    },
    {
      "epoch": 1.7826935588809367,
      "grad_norm": 1.5014948844909668,
      "learning_rate": 3.644898926669524e-05,
      "loss": 0.781,
      "step": 5480
    },
    {
      "epoch": 1.784320104098894,
      "grad_norm": 1.6949294805526733,
      "learning_rate": 3.6426040288360674e-05,
      "loss": 0.8263,
      "step": 5485
    },
    {
      "epoch": 1.785946649316851,
      "grad_norm": 1.567044734954834,
      "learning_rate": 3.640307913318982e-05,
      "loss": 0.8007,
      "step": 5490
    },
    {
      "epoch": 1.7875731945348081,
      "grad_norm": 1.6042593717575073,
      "learning_rate": 3.638010582565257e-05,
      "loss": 0.7956,
      "step": 5495
    },
    {
      "epoch": 1.7891997397527653,
      "grad_norm": 1.9586327075958252,
      "learning_rate": 3.6357120390231825e-05,
      "loss": 0.8122,
      "step": 5500
    },
    {
      "epoch": 1.7908262849707222,
      "grad_norm": 5.827934265136719,
      "learning_rate": 3.6334122851423344e-05,
      "loss": 0.8541,
      "step": 5505
    },
    {
      "epoch": 1.7924528301886793,
      "grad_norm": 1.5693747997283936,
      "learning_rate": 3.6311113233735836e-05,
      "loss": 0.8042,
      "step": 5510
    },
    {
      "epoch": 1.7940793754066364,
      "grad_norm": 1.6549158096313477,
      "learning_rate": 3.6288091561690855e-05,
      "loss": 0.8172,
      "step": 5515
    },
    {
      "epoch": 1.7957059206245933,
      "grad_norm": 1.5341600179672241,
      "learning_rate": 3.626505785982281e-05,
      "loss": 0.8073,
      "step": 5520
    },
    {
      "epoch": 1.7973324658425505,
      "grad_norm": 4.368505954742432,
      "learning_rate": 3.6242012152678925e-05,
      "loss": 0.7982,
      "step": 5525
    },
    {
      "epoch": 1.7989590110605076,
      "grad_norm": 1.8337335586547852,
      "learning_rate": 3.6218954464819224e-05,
      "loss": 0.7842,
      "step": 5530
    },
    {
      "epoch": 1.8005855562784645,
      "grad_norm": 1.8748178482055664,
      "learning_rate": 3.61958848208165e-05,
      "loss": 0.7305,
      "step": 5535
    },
    {
      "epoch": 1.8022121014964216,
      "grad_norm": 1.6605912446975708,
      "learning_rate": 3.6172803245256284e-05,
      "loss": 0.7669,
      "step": 5540
    },
    {
      "epoch": 1.8038386467143788,
      "grad_norm": 1.788952350616455,
      "learning_rate": 3.614970976273681e-05,
      "loss": 0.7843,
      "step": 5545
    },
    {
      "epoch": 1.8054651919323357,
      "grad_norm": 1.8774856328964233,
      "learning_rate": 3.612660439786904e-05,
      "loss": 0.7941,
      "step": 5550
    },
    {
      "epoch": 1.8070917371502928,
      "grad_norm": 1.7823916673660278,
      "learning_rate": 3.6103487175276564e-05,
      "loss": 0.7977,
      "step": 5555
    },
    {
      "epoch": 1.80871828236825,
      "grad_norm": 2.3198554515838623,
      "learning_rate": 3.608035811959561e-05,
      "loss": 0.8222,
      "step": 5560
    },
    {
      "epoch": 1.8103448275862069,
      "grad_norm": 2.542912244796753,
      "learning_rate": 3.6057217255475034e-05,
      "loss": 0.7993,
      "step": 5565
    },
    {
      "epoch": 1.811971372804164,
      "grad_norm": 2.11651349067688,
      "learning_rate": 3.603406460757627e-05,
      "loss": 0.8015,
      "step": 5570
    },
    {
      "epoch": 1.8135979180221211,
      "grad_norm": 2.6846766471862793,
      "learning_rate": 3.601090020057329e-05,
      "loss": 0.81,
      "step": 5575
    },
    {
      "epoch": 1.815224463240078,
      "grad_norm": 1.87996244430542,
      "learning_rate": 3.598772405915264e-05,
      "loss": 0.8084,
      "step": 5580
    },
    {
      "epoch": 1.8168510084580352,
      "grad_norm": 1.4050390720367432,
      "learning_rate": 3.596453620801334e-05,
      "loss": 0.7983,
      "step": 5585
    },
    {
      "epoch": 1.8184775536759923,
      "grad_norm": 1.6851388216018677,
      "learning_rate": 3.594133667186688e-05,
      "loss": 0.7948,
      "step": 5590
    },
    {
      "epoch": 1.8201040988939492,
      "grad_norm": 1.6844091415405273,
      "learning_rate": 3.591812547543725e-05,
      "loss": 0.7935,
      "step": 5595
    },
    {
      "epoch": 1.8217306441119063,
      "grad_norm": 1.704921841621399,
      "learning_rate": 3.58949026434608e-05,
      "loss": 0.7882,
      "step": 5600
    },
    {
      "epoch": 1.8233571893298635,
      "grad_norm": 1.488486409187317,
      "learning_rate": 3.587166820068635e-05,
      "loss": 0.8135,
      "step": 5605
    },
    {
      "epoch": 1.8249837345478204,
      "grad_norm": 1.907758355140686,
      "learning_rate": 3.584842217187503e-05,
      "loss": 0.8007,
      "step": 5610
    },
    {
      "epoch": 1.8266102797657775,
      "grad_norm": 1.9598406553268433,
      "learning_rate": 3.582516458180036e-05,
      "loss": 0.78,
      "step": 5615
    },
    {
      "epoch": 1.8282368249837346,
      "grad_norm": 1.5850437879562378,
      "learning_rate": 3.580189545524818e-05,
      "loss": 0.7971,
      "step": 5620
    },
    {
      "epoch": 1.8298633702016915,
      "grad_norm": 1.4961295127868652,
      "learning_rate": 3.577861481701659e-05,
      "loss": 0.8059,
      "step": 5625
    },
    {
      "epoch": 1.8314899154196487,
      "grad_norm": 1.4719113111495972,
      "learning_rate": 3.575532269191599e-05,
      "loss": 0.7722,
      "step": 5630
    },
    {
      "epoch": 1.8331164606376058,
      "grad_norm": 1.7238733768463135,
      "learning_rate": 3.573201910476902e-05,
      "loss": 0.7766,
      "step": 5635
    },
    {
      "epoch": 1.8347430058555627,
      "grad_norm": 2.2540855407714844,
      "learning_rate": 3.57087040804105e-05,
      "loss": 0.8187,
      "step": 5640
    },
    {
      "epoch": 1.8363695510735198,
      "grad_norm": 1.552124261856079,
      "learning_rate": 3.568537764368751e-05,
      "loss": 0.7887,
      "step": 5645
    },
    {
      "epoch": 1.837996096291477,
      "grad_norm": 1.6686851978302002,
      "learning_rate": 3.566203981945921e-05,
      "loss": 0.8184,
      "step": 5650
    },
    {
      "epoch": 1.8396226415094339,
      "grad_norm": 1.840654730796814,
      "learning_rate": 3.5638690632596956e-05,
      "loss": 0.7635,
      "step": 5655
    },
    {
      "epoch": 1.841249186727391,
      "grad_norm": 1.8260648250579834,
      "learning_rate": 3.561533010798418e-05,
      "loss": 0.766,
      "step": 5660
    },
    {
      "epoch": 1.8428757319453482,
      "grad_norm": 1.6572096347808838,
      "learning_rate": 3.559195827051641e-05,
      "loss": 0.8138,
      "step": 5665
    },
    {
      "epoch": 1.844502277163305,
      "grad_norm": 1.450379490852356,
      "learning_rate": 3.556857514510123e-05,
      "loss": 0.7801,
      "step": 5670
    },
    {
      "epoch": 1.8461288223812622,
      "grad_norm": 1.6766986846923828,
      "learning_rate": 3.554518075665826e-05,
      "loss": 0.7615,
      "step": 5675
    },
    {
      "epoch": 1.8477553675992193,
      "grad_norm": 1.8202241659164429,
      "learning_rate": 3.5521775130119095e-05,
      "loss": 0.8211,
      "step": 5680
    },
    {
      "epoch": 1.8493819128171762,
      "grad_norm": 1.6854424476623535,
      "learning_rate": 3.549835829042735e-05,
      "loss": 0.848,
      "step": 5685
    },
    {
      "epoch": 1.8510084580351334,
      "grad_norm": 1.4499651193618774,
      "learning_rate": 3.547493026253854e-05,
      "loss": 0.788,
      "step": 5690
    },
    {
      "epoch": 1.8526350032530905,
      "grad_norm": 1.5754683017730713,
      "learning_rate": 3.545149107142016e-05,
      "loss": 0.7962,
      "step": 5695
    },
    {
      "epoch": 1.8542615484710474,
      "grad_norm": 1.6627172231674194,
      "learning_rate": 3.542804074205155e-05,
      "loss": 0.8025,
      "step": 5700
    },
    {
      "epoch": 1.8558880936890045,
      "grad_norm": 1.7090286016464233,
      "learning_rate": 3.5404579299423944e-05,
      "loss": 0.7991,
      "step": 5705
    },
    {
      "epoch": 1.8575146389069617,
      "grad_norm": 1.612816572189331,
      "learning_rate": 3.5381106768540426e-05,
      "loss": 0.7932,
      "step": 5710
    },
    {
      "epoch": 1.8591411841249186,
      "grad_norm": 1.7315623760223389,
      "learning_rate": 3.5357623174415886e-05,
      "loss": 0.7914,
      "step": 5715
    },
    {
      "epoch": 1.8607677293428757,
      "grad_norm": 1.6280250549316406,
      "learning_rate": 3.5334128542077004e-05,
      "loss": 0.7666,
      "step": 5720
    },
    {
      "epoch": 1.8623942745608328,
      "grad_norm": 2.159470796585083,
      "learning_rate": 3.531062289656223e-05,
      "loss": 0.826,
      "step": 5725
    },
    {
      "epoch": 1.8640208197787898,
      "grad_norm": 1.5661296844482422,
      "learning_rate": 3.528710626292174e-05,
      "loss": 0.8098,
      "step": 5730
    },
    {
      "epoch": 1.8656473649967469,
      "grad_norm": 1.5078445672988892,
      "learning_rate": 3.5263578666217426e-05,
      "loss": 0.8125,
      "step": 5735
    },
    {
      "epoch": 1.867273910214704,
      "grad_norm": 1.8959999084472656,
      "learning_rate": 3.5240040131522876e-05,
      "loss": 0.7917,
      "step": 5740
    },
    {
      "epoch": 1.868900455432661,
      "grad_norm": 2.221889019012451,
      "learning_rate": 3.5216490683923306e-05,
      "loss": 0.7969,
      "step": 5745
    },
    {
      "epoch": 1.870527000650618,
      "grad_norm": 1.801939845085144,
      "learning_rate": 3.519293034851559e-05,
      "loss": 0.8124,
      "step": 5750
    },
    {
      "epoch": 1.8721535458685752,
      "grad_norm": 1.5591964721679688,
      "learning_rate": 3.51693591504082e-05,
      "loss": 0.8293,
      "step": 5755
    },
    {
      "epoch": 1.873780091086532,
      "grad_norm": 1.4893189668655396,
      "learning_rate": 3.5150494387658796e-05,
      "loss": 0.7798,
      "step": 5760
    },
    {
      "epoch": 1.8754066363044892,
      "grad_norm": 1.6666206121444702,
      "learning_rate": 3.512690370000224e-05,
      "loss": 0.7907,
      "step": 5765
    },
    {
      "epoch": 1.8770331815224464,
      "grad_norm": 1.7753440141677856,
      "learning_rate": 3.5103302220011216e-05,
      "loss": 0.7612,
      "step": 5770
    },
    {
      "epoch": 1.8786597267404033,
      "grad_norm": 1.8589496612548828,
      "learning_rate": 3.5079689972838033e-05,
      "loss": 0.797,
      "step": 5775
    },
    {
      "epoch": 1.8802862719583604,
      "grad_norm": 1.7949668169021606,
      "learning_rate": 3.505606698364648e-05,
      "loss": 0.8046,
      "step": 5780
    },
    {
      "epoch": 1.8819128171763175,
      "grad_norm": 1.6644870042800903,
      "learning_rate": 3.503243327761179e-05,
      "loss": 0.7843,
      "step": 5785
    },
    {
      "epoch": 1.8835393623942744,
      "grad_norm": 1.7566967010498047,
      "learning_rate": 3.500878887992063e-05,
      "loss": 0.792,
      "step": 5790
    },
    {
      "epoch": 1.8851659076122316,
      "grad_norm": 1.666772484779358,
      "learning_rate": 3.4985133815771037e-05,
      "loss": 0.792,
      "step": 5795
    },
    {
      "epoch": 1.8867924528301887,
      "grad_norm": 1.5666133165359497,
      "learning_rate": 3.4961468110372445e-05,
      "loss": 0.7709,
      "step": 5800
    },
    {
      "epoch": 1.8884189980481456,
      "grad_norm": 1.8184480667114258,
      "learning_rate": 3.4937791788945615e-05,
      "loss": 0.7961,
      "step": 5805
    },
    {
      "epoch": 1.8900455432661027,
      "grad_norm": 1.7216947078704834,
      "learning_rate": 3.49141048767226e-05,
      "loss": 0.7997,
      "step": 5810
    },
    {
      "epoch": 1.8916720884840599,
      "grad_norm": 2.7872517108917236,
      "learning_rate": 3.489040739894679e-05,
      "loss": 0.7889,
      "step": 5815
    },
    {
      "epoch": 1.8932986337020168,
      "grad_norm": 1.7658991813659668,
      "learning_rate": 3.486669938087278e-05,
      "loss": 0.8139,
      "step": 5820
    },
    {
      "epoch": 1.8949251789199741,
      "grad_norm": 1.5896073579788208,
      "learning_rate": 3.484298084776644e-05,
      "loss": 0.8105,
      "step": 5825
    },
    {
      "epoch": 1.896551724137931,
      "grad_norm": 1.5800156593322754,
      "learning_rate": 3.4819251824904814e-05,
      "loss": 0.7968,
      "step": 5830
    },
    {
      "epoch": 1.898178269355888,
      "grad_norm": 1.7342997789382935,
      "learning_rate": 3.479551233757616e-05,
      "loss": 0.8007,
      "step": 5835
    },
    {
      "epoch": 1.8998048145738453,
      "grad_norm": 2.0258543491363525,
      "learning_rate": 3.477176241107985e-05,
      "loss": 0.778,
      "step": 5840
    },
    {
      "epoch": 1.9014313597918022,
      "grad_norm": 1.7815632820129395,
      "learning_rate": 3.47480020707264e-05,
      "loss": 0.7997,
      "step": 5845
    },
    {
      "epoch": 1.9030579050097591,
      "grad_norm": 1.4552288055419922,
      "learning_rate": 3.4724231341837446e-05,
      "loss": 0.8031,
      "step": 5850
    },
    {
      "epoch": 1.9046844502277165,
      "grad_norm": 1.822088360786438,
      "learning_rate": 3.470045024974564e-05,
      "loss": 0.8377,
      "step": 5855
    },
    {
      "epoch": 1.9063109954456734,
      "grad_norm": 1.5762214660644531,
      "learning_rate": 3.467665881979473e-05,
      "loss": 0.825,
      "step": 5860
    },
    {
      "epoch": 1.9079375406636303,
      "grad_norm": 1.3905141353607178,
      "learning_rate": 3.4652857077339465e-05,
      "loss": 0.8005,
      "step": 5865
    },
    {
      "epoch": 1.9095640858815877,
      "grad_norm": 2.5645837783813477,
      "learning_rate": 3.4629045047745566e-05,
      "loss": 0.7714,
      "step": 5870
    },
    {
      "epoch": 1.9111906310995446,
      "grad_norm": 1.9059853553771973,
      "learning_rate": 3.460522275638974e-05,
      "loss": 0.7933,
      "step": 5875
    },
    {
      "epoch": 1.9128171763175015,
      "grad_norm": 1.685237169265747,
      "learning_rate": 3.4581390228659634e-05,
      "loss": 0.786,
      "step": 5880
    },
    {
      "epoch": 1.9144437215354588,
      "grad_norm": 1.6190307140350342,
      "learning_rate": 3.455754748995377e-05,
      "loss": 0.8037,
      "step": 5885
    },
    {
      "epoch": 1.9160702667534157,
      "grad_norm": 1.8023563623428345,
      "learning_rate": 3.453369456568159e-05,
      "loss": 0.7781,
      "step": 5890
    },
    {
      "epoch": 1.9176968119713727,
      "grad_norm": 1.6431732177734375,
      "learning_rate": 3.450983148126337e-05,
      "loss": 0.8081,
      "step": 5895
    },
    {
      "epoch": 1.91932335718933,
      "grad_norm": 1.851529836654663,
      "learning_rate": 3.4485958262130215e-05,
      "loss": 0.8115,
      "step": 5900
    },
    {
      "epoch": 1.920949902407287,
      "grad_norm": 1.6920840740203857,
      "learning_rate": 3.446207493372404e-05,
      "loss": 0.7958,
      "step": 5905
    },
    {
      "epoch": 1.9225764476252438,
      "grad_norm": 2.3522324562072754,
      "learning_rate": 3.4438181521497525e-05,
      "loss": 0.7921,
      "step": 5910
    },
    {
      "epoch": 1.9242029928432012,
      "grad_norm": 2.0344767570495605,
      "learning_rate": 3.44142780509141e-05,
      "loss": 0.788,
      "step": 5915
    },
    {
      "epoch": 1.925829538061158,
      "grad_norm": 1.7460458278656006,
      "learning_rate": 3.439036454744791e-05,
      "loss": 0.8313,
      "step": 5920
    },
    {
      "epoch": 1.927456083279115,
      "grad_norm": 1.6961497068405151,
      "learning_rate": 3.43664410365838e-05,
      "loss": 0.7826,
      "step": 5925
    },
    {
      "epoch": 1.9290826284970723,
      "grad_norm": 2.1205198764801025,
      "learning_rate": 3.434250754381728e-05,
      "loss": 0.7765,
      "step": 5930
    },
    {
      "epoch": 1.9307091737150293,
      "grad_norm": 1.5347671508789062,
      "learning_rate": 3.4318564094654484e-05,
      "loss": 0.7735,
      "step": 5935
    },
    {
      "epoch": 1.9323357189329864,
      "grad_norm": 1.7982996702194214,
      "learning_rate": 3.4294610714612176e-05,
      "loss": 0.8435,
      "step": 5940
    },
    {
      "epoch": 1.9339622641509435,
      "grad_norm": 1.7098652124404907,
      "learning_rate": 3.427064742921768e-05,
      "loss": 0.7757,
      "step": 5945
    },
    {
      "epoch": 1.9355888093689004,
      "grad_norm": 1.446061611175537,
      "learning_rate": 3.42466742640089e-05,
      "loss": 0.8333,
      "step": 5950
    },
    {
      "epoch": 1.9372153545868576,
      "grad_norm": 1.6027209758758545,
      "learning_rate": 3.4222691244534253e-05,
      "loss": 0.7904,
      "step": 5955
    },
    {
      "epoch": 1.9388418998048147,
      "grad_norm": 1.6385594606399536,
      "learning_rate": 3.419869839635267e-05,
      "loss": 0.7825,
      "step": 5960
    },
    {
      "epoch": 1.9404684450227716,
      "grad_norm": 1.4547863006591797,
      "learning_rate": 3.417469574503356e-05,
      "loss": 0.7726,
      "step": 5965
    },
    {
      "epoch": 1.9420949902407287,
      "grad_norm": 1.7256838083267212,
      "learning_rate": 3.415068331615674e-05,
      "loss": 0.7763,
      "step": 5970
    },
    {
      "epoch": 1.9437215354586859,
      "grad_norm": 1.7924857139587402,
      "learning_rate": 3.41266611353125e-05,
      "loss": 0.8119,
      "step": 5975
    },
    {
      "epoch": 1.9453480806766428,
      "grad_norm": 1.5046923160552979,
      "learning_rate": 3.4102629228101494e-05,
      "loss": 0.7997,
      "step": 5980
    },
    {
      "epoch": 1.9469746258946,
      "grad_norm": 1.7069694995880127,
      "learning_rate": 3.407858762013474e-05,
      "loss": 0.8143,
      "step": 5985
    },
    {
      "epoch": 1.948601171112557,
      "grad_norm": 2.707512140274048,
      "learning_rate": 3.405453633703361e-05,
      "loss": 0.7946,
      "step": 5990
    },
    {
      "epoch": 1.950227716330514,
      "grad_norm": 1.4022822380065918,
      "learning_rate": 3.4030475404429776e-05,
      "loss": 0.7807,
      "step": 5995
    },
    {
      "epoch": 1.951854261548471,
      "grad_norm": 1.9132970571517944,
      "learning_rate": 3.4006404847965196e-05,
      "loss": 0.7888,
      "step": 6000
    },
    {
      "epoch": 1.9534808067664282,
      "grad_norm": 2.214003324508667,
      "learning_rate": 3.398232469329207e-05,
      "loss": 0.7712,
      "step": 6005
    },
    {
      "epoch": 1.9551073519843851,
      "grad_norm": 3.6037864685058594,
      "learning_rate": 3.395823496607286e-05,
      "loss": 0.8125,
      "step": 6010
    },
    {
      "epoch": 1.9567338972023423,
      "grad_norm": 1.6346657276153564,
      "learning_rate": 3.39341356919802e-05,
      "loss": 0.8014,
      "step": 6015
    },
    {
      "epoch": 1.9583604424202994,
      "grad_norm": 1.836584448814392,
      "learning_rate": 3.39100268966969e-05,
      "loss": 0.792,
      "step": 6020
    },
    {
      "epoch": 1.9599869876382563,
      "grad_norm": 1.9783848524093628,
      "learning_rate": 3.388590860591595e-05,
      "loss": 0.7552,
      "step": 6025
    },
    {
      "epoch": 1.9616135328562134,
      "grad_norm": 1.9278146028518677,
      "learning_rate": 3.3861780845340406e-05,
      "loss": 0.7921,
      "step": 6030
    },
    {
      "epoch": 1.9632400780741706,
      "grad_norm": 2.9828009605407715,
      "learning_rate": 3.383764364068346e-05,
      "loss": 0.7936,
      "step": 6035
    },
    {
      "epoch": 1.9648666232921275,
      "grad_norm": 2.507549524307251,
      "learning_rate": 3.381349701766835e-05,
      "loss": 0.7761,
      "step": 6040
    },
    {
      "epoch": 1.9664931685100846,
      "grad_norm": 2.000004768371582,
      "learning_rate": 3.3789341002028363e-05,
      "loss": 0.8025,
      "step": 6045
    },
    {
      "epoch": 1.9681197137280417,
      "grad_norm": 1.9173941612243652,
      "learning_rate": 3.376517561950677e-05,
      "loss": 0.7907,
      "step": 6050
    },
    {
      "epoch": 1.9697462589459986,
      "grad_norm": 2.1533915996551514,
      "learning_rate": 3.374100089585685e-05,
      "loss": 0.7914,
      "step": 6055
    },
    {
      "epoch": 1.9713728041639558,
      "grad_norm": 1.5206011533737183,
      "learning_rate": 3.371681685684184e-05,
      "loss": 0.7902,
      "step": 6060
    },
    {
      "epoch": 1.972999349381913,
      "grad_norm": 1.5686813592910767,
      "learning_rate": 3.3692623528234865e-05,
      "loss": 0.7862,
      "step": 6065
    },
    {
      "epoch": 1.9746258945998698,
      "grad_norm": 1.5873268842697144,
      "learning_rate": 3.366842093581901e-05,
      "loss": 0.7717,
      "step": 6070
    },
    {
      "epoch": 1.976252439817827,
      "grad_norm": 1.777116060256958,
      "learning_rate": 3.364420910538718e-05,
      "loss": 0.7947,
      "step": 6075
    },
    {
      "epoch": 1.977878985035784,
      "grad_norm": 2.015531301498413,
      "learning_rate": 3.361998806274214e-05,
      "loss": 0.785,
      "step": 6080
    },
    {
      "epoch": 1.979505530253741,
      "grad_norm": 1.6734198331832886,
      "learning_rate": 3.3595757833696505e-05,
      "loss": 0.7841,
      "step": 6085
    },
    {
      "epoch": 1.9811320754716981,
      "grad_norm": 2.073692560195923,
      "learning_rate": 3.3571518444072626e-05,
      "loss": 0.7715,
      "step": 6090
    },
    {
      "epoch": 1.9827586206896552,
      "grad_norm": 1.8034336566925049,
      "learning_rate": 3.354726991970266e-05,
      "loss": 0.8163,
      "step": 6095
    },
    {
      "epoch": 1.9843851659076122,
      "grad_norm": 1.5698885917663574,
      "learning_rate": 3.352301228642847e-05,
      "loss": 0.786,
      "step": 6100
    },
    {
      "epoch": 1.9860117111255693,
      "grad_norm": 1.3378742933273315,
      "learning_rate": 3.349874557010166e-05,
      "loss": 0.7908,
      "step": 6105
    },
    {
      "epoch": 1.9876382563435264,
      "grad_norm": 1.6845659017562866,
      "learning_rate": 3.3474469796583477e-05,
      "loss": 0.8112,
      "step": 6110
    },
    {
      "epoch": 1.9892648015614833,
      "grad_norm": 1.834529995918274,
      "learning_rate": 3.345018499174482e-05,
      "loss": 0.7599,
      "step": 6115
    },
    {
      "epoch": 1.9908913467794405,
      "grad_norm": 1.9633735418319702,
      "learning_rate": 3.3425891181466275e-05,
      "loss": 0.825,
      "step": 6120
    },
    {
      "epoch": 1.9925178919973976,
      "grad_norm": 1.81381094455719,
      "learning_rate": 3.340158839163794e-05,
      "loss": 0.7793,
      "step": 6125
    },
    {
      "epoch": 1.9941444372153545,
      "grad_norm": 1.7863167524337769,
      "learning_rate": 3.337727664815952e-05,
      "loss": 0.7549,
      "step": 6130
    },
    {
      "epoch": 1.9957709824333116,
      "grad_norm": 2.0586788654327393,
      "learning_rate": 3.335295597694029e-05,
      "loss": 0.7701,
      "step": 6135
    },
    {
      "epoch": 1.9973975276512688,
      "grad_norm": 1.8435872793197632,
      "learning_rate": 3.3328626403899e-05,
      "loss": 0.7667,
      "step": 6140
    },
    {
      "epoch": 1.9990240728692257,
      "grad_norm": 1.6310111284255981,
      "learning_rate": 3.33042879549639e-05,
      "loss": 0.7994,
      "step": 6145
    },
    {
      "epoch": 2.0,
      "eval_f1": 0.8204322346195636,
      "eval_loss": 0.4052734375,
      "eval_precision": 0.8203424383778088,
      "eval_recall": 0.8207026062554238,
      "eval_runtime": 1029.4017,
      "eval_samples_per_second": 382.198,
      "eval_steps_per_second": 0.747,
      "step": 6148
    },
    {
      "epoch": 2.000650618087183,
      "grad_norm": 2.080751419067383,
      "learning_rate": 3.3279940656072706e-05,
      "loss": 0.7787,
      "step": 6150
    },
    {
      "epoch": 2.00227716330514,
      "grad_norm": 2.657672643661499,
      "learning_rate": 3.325558453317254e-05,
      "loss": 0.7258,
      "step": 6155
    },
    {
      "epoch": 2.003903708523097,
      "grad_norm": 2.20680570602417,
      "learning_rate": 3.323121961221996e-05,
      "loss": 0.7325,
      "step": 6160
    },
    {
      "epoch": 2.005530253741054,
      "grad_norm": 2.3011794090270996,
      "learning_rate": 3.320684591918089e-05,
      "loss": 0.7722,
      "step": 6165
    },
    {
      "epoch": 2.007156798959011,
      "grad_norm": 1.6290371417999268,
      "learning_rate": 3.3182463480030595e-05,
      "loss": 0.7134,
      "step": 6170
    },
    {
      "epoch": 2.008783344176968,
      "grad_norm": 1.6625546216964722,
      "learning_rate": 3.3158072320753664e-05,
      "loss": 0.737,
      "step": 6175
    },
    {
      "epoch": 2.0104098893949254,
      "grad_norm": 1.832472562789917,
      "learning_rate": 3.3133672467343965e-05,
      "loss": 0.7456,
      "step": 6180
    },
    {
      "epoch": 2.0120364346128823,
      "grad_norm": 1.8868688344955444,
      "learning_rate": 3.310926394580468e-05,
      "loss": 0.7438,
      "step": 6185
    },
    {
      "epoch": 2.013662979830839,
      "grad_norm": 2.2079997062683105,
      "learning_rate": 3.308484678214817e-05,
      "loss": 0.7365,
      "step": 6190
    },
    {
      "epoch": 2.0152895250487965,
      "grad_norm": 2.1085147857666016,
      "learning_rate": 3.306042100239602e-05,
      "loss": 0.7352,
      "step": 6195
    },
    {
      "epoch": 2.0169160702667535,
      "grad_norm": 2.2749311923980713,
      "learning_rate": 3.303598663257904e-05,
      "loss": 0.7529,
      "step": 6200
    },
    {
      "epoch": 2.0185426154847104,
      "grad_norm": 3.618288516998291,
      "learning_rate": 3.3011543698737155e-05,
      "loss": 0.7432,
      "step": 6205
    },
    {
      "epoch": 2.0201691607026677,
      "grad_norm": 2.5536632537841797,
      "learning_rate": 3.29870922269194e-05,
      "loss": 0.7465,
      "step": 6210
    },
    {
      "epoch": 2.0217957059206246,
      "grad_norm": 1.9676142930984497,
      "learning_rate": 3.296263224318397e-05,
      "loss": 0.7615,
      "step": 6215
    },
    {
      "epoch": 2.0234222511385815,
      "grad_norm": 1.650276780128479,
      "learning_rate": 3.2938163773598065e-05,
      "loss": 0.7579,
      "step": 6220
    },
    {
      "epoch": 2.025048796356539,
      "grad_norm": 1.7215265035629272,
      "learning_rate": 3.291368684423796e-05,
      "loss": 0.7333,
      "step": 6225
    },
    {
      "epoch": 2.026675341574496,
      "grad_norm": 1.7647247314453125,
      "learning_rate": 3.288920148118896e-05,
      "loss": 0.7282,
      "step": 6230
    },
    {
      "epoch": 2.0283018867924527,
      "grad_norm": 1.90190851688385,
      "learning_rate": 3.286470771054534e-05,
      "loss": 0.7317,
      "step": 6235
    },
    {
      "epoch": 2.02992843201041,
      "grad_norm": 1.6041595935821533,
      "learning_rate": 3.284020555841033e-05,
      "loss": 0.7003,
      "step": 6240
    },
    {
      "epoch": 2.031554977228367,
      "grad_norm": 1.7647874355316162,
      "learning_rate": 3.2815695050896093e-05,
      "loss": 0.7459,
      "step": 6245
    },
    {
      "epoch": 2.033181522446324,
      "grad_norm": 1.9271812438964844,
      "learning_rate": 3.279117621412372e-05,
      "loss": 0.7299,
      "step": 6250
    },
    {
      "epoch": 2.0348080676642812,
      "grad_norm": 1.810985803604126,
      "learning_rate": 3.2766649074223145e-05,
      "loss": 0.7284,
      "step": 6255
    },
    {
      "epoch": 2.036434612882238,
      "grad_norm": 1.575209379196167,
      "learning_rate": 3.274211365733317e-05,
      "loss": 0.7332,
      "step": 6260
    },
    {
      "epoch": 2.038061158100195,
      "grad_norm": 2.2633728981018066,
      "learning_rate": 3.271756998960143e-05,
      "loss": 0.7421,
      "step": 6265
    },
    {
      "epoch": 2.0396877033181524,
      "grad_norm": 2.159065008163452,
      "learning_rate": 3.2693018097184314e-05,
      "loss": 0.7199,
      "step": 6270
    },
    {
      "epoch": 2.0413142485361093,
      "grad_norm": 1.732470154762268,
      "learning_rate": 3.2668458006247e-05,
      "loss": 0.7401,
      "step": 6275
    },
    {
      "epoch": 2.0429407937540662,
      "grad_norm": 2.0348281860351562,
      "learning_rate": 3.2643889742963434e-05,
      "loss": 0.7274,
      "step": 6280
    },
    {
      "epoch": 2.0445673389720236,
      "grad_norm": 1.893390417098999,
      "learning_rate": 3.261931333351621e-05,
      "loss": 0.7055,
      "step": 6285
    },
    {
      "epoch": 2.0461938841899805,
      "grad_norm": 2.049177646636963,
      "learning_rate": 3.259472880409664e-05,
      "loss": 0.7244,
      "step": 6290
    },
    {
      "epoch": 2.0478204294079374,
      "grad_norm": 1.857323169708252,
      "learning_rate": 3.2570136180904696e-05,
      "loss": 0.7428,
      "step": 6295
    },
    {
      "epoch": 2.0494469746258948,
      "grad_norm": 1.827819585800171,
      "learning_rate": 3.2545535490148955e-05,
      "loss": 0.7575,
      "step": 6300
    },
    {
      "epoch": 2.0510735198438517,
      "grad_norm": 1.724376916885376,
      "learning_rate": 3.2520926758046586e-05,
      "loss": 0.7204,
      "step": 6305
    },
    {
      "epoch": 2.0527000650618086,
      "grad_norm": 1.9833449125289917,
      "learning_rate": 3.249631001082336e-05,
      "loss": 0.7377,
      "step": 6310
    },
    {
      "epoch": 2.054326610279766,
      "grad_norm": 1.9445111751556396,
      "learning_rate": 3.247168527471357e-05,
      "loss": 0.7548,
      "step": 6315
    },
    {
      "epoch": 2.055953155497723,
      "grad_norm": 1.842955470085144,
      "learning_rate": 3.244705257596001e-05,
      "loss": 0.7519,
      "step": 6320
    },
    {
      "epoch": 2.0575797007156797,
      "grad_norm": 1.5295389890670776,
      "learning_rate": 3.242241194081398e-05,
      "loss": 0.7478,
      "step": 6325
    },
    {
      "epoch": 2.059206245933637,
      "grad_norm": 2.5027713775634766,
      "learning_rate": 3.2397763395535244e-05,
      "loss": 0.7513,
      "step": 6330
    },
    {
      "epoch": 2.060832791151594,
      "grad_norm": 1.6601084470748901,
      "learning_rate": 3.2373106966391965e-05,
      "loss": 0.6987,
      "step": 6335
    },
    {
      "epoch": 2.062459336369551,
      "grad_norm": 2.04126238822937,
      "learning_rate": 3.234844267966074e-05,
      "loss": 0.7315,
      "step": 6340
    },
    {
      "epoch": 2.0640858815875083,
      "grad_norm": 1.687272548675537,
      "learning_rate": 3.232377056162652e-05,
      "loss": 0.7296,
      "step": 6345
    },
    {
      "epoch": 2.065712426805465,
      "grad_norm": 1.9358770847320557,
      "learning_rate": 3.229909063858261e-05,
      "loss": 0.7125,
      "step": 6350
    },
    {
      "epoch": 2.067338972023422,
      "grad_norm": 1.9139622449874878,
      "learning_rate": 3.2274402936830626e-05,
      "loss": 0.7454,
      "step": 6355
    },
    {
      "epoch": 2.0689655172413794,
      "grad_norm": 2.3934218883514404,
      "learning_rate": 3.224970748268048e-05,
      "loss": 0.7467,
      "step": 6360
    },
    {
      "epoch": 2.0705920624593364,
      "grad_norm": 2.055274486541748,
      "learning_rate": 3.2225004302450354e-05,
      "loss": 0.7388,
      "step": 6365
    },
    {
      "epoch": 2.0722186076772933,
      "grad_norm": 1.8648474216461182,
      "learning_rate": 3.220029342246665e-05,
      "loss": 0.7241,
      "step": 6370
    },
    {
      "epoch": 2.0738451528952506,
      "grad_norm": 2.0889534950256348,
      "learning_rate": 3.217557486906398e-05,
      "loss": 0.7711,
      "step": 6375
    },
    {
      "epoch": 2.0754716981132075,
      "grad_norm": 1.6840265989303589,
      "learning_rate": 3.215084866858513e-05,
      "loss": 0.7331,
      "step": 6380
    },
    {
      "epoch": 2.0770982433311644,
      "grad_norm": 1.883111596107483,
      "learning_rate": 3.212611484738105e-05,
      "loss": 0.7269,
      "step": 6385
    },
    {
      "epoch": 2.078724788549122,
      "grad_norm": 1.9425476789474487,
      "learning_rate": 3.210137343181078e-05,
      "loss": 0.7518,
      "step": 6390
    },
    {
      "epoch": 2.0803513337670787,
      "grad_norm": 1.6710432767868042,
      "learning_rate": 3.20766244482415e-05,
      "loss": 0.7307,
      "step": 6395
    },
    {
      "epoch": 2.0819778789850356,
      "grad_norm": 1.703465461730957,
      "learning_rate": 3.2051867923048415e-05,
      "loss": 0.7412,
      "step": 6400
    },
    {
      "epoch": 2.083604424202993,
      "grad_norm": 1.7874739170074463,
      "learning_rate": 3.2027103882614774e-05,
      "loss": 0.7208,
      "step": 6405
    },
    {
      "epoch": 2.08523096942095,
      "grad_norm": 3.441638708114624,
      "learning_rate": 3.200233235333186e-05,
      "loss": 0.7439,
      "step": 6410
    },
    {
      "epoch": 2.086857514638907,
      "grad_norm": 2.2441136837005615,
      "learning_rate": 3.197755336159891e-05,
      "loss": 0.721,
      "step": 6415
    },
    {
      "epoch": 2.088484059856864,
      "grad_norm": 2.098034620285034,
      "learning_rate": 3.195276693382313e-05,
      "loss": 0.7125,
      "step": 6420
    },
    {
      "epoch": 2.090110605074821,
      "grad_norm": 2.6204288005828857,
      "learning_rate": 3.192797309641964e-05,
      "loss": 0.7352,
      "step": 6425
    },
    {
      "epoch": 2.091737150292778,
      "grad_norm": 1.990392804145813,
      "learning_rate": 3.190317187581146e-05,
      "loss": 0.7655,
      "step": 6430
    },
    {
      "epoch": 2.0933636955107353,
      "grad_norm": 1.7657074928283691,
      "learning_rate": 3.187836329842947e-05,
      "loss": 0.7603,
      "step": 6435
    },
    {
      "epoch": 2.094990240728692,
      "grad_norm": 1.7322503328323364,
      "learning_rate": 3.185354739071242e-05,
      "loss": 0.7477,
      "step": 6440
    },
    {
      "epoch": 2.096616785946649,
      "grad_norm": 1.8331769704818726,
      "learning_rate": 3.182872417910684e-05,
      "loss": 0.7181,
      "step": 6445
    },
    {
      "epoch": 2.0982433311646065,
      "grad_norm": 1.7432118654251099,
      "learning_rate": 3.1803893690067046e-05,
      "loss": 0.7255,
      "step": 6450
    },
    {
      "epoch": 2.0998698763825634,
      "grad_norm": 1.8132730722427368,
      "learning_rate": 3.177905595005514e-05,
      "loss": 0.7447,
      "step": 6455
    },
    {
      "epoch": 2.1014964216005203,
      "grad_norm": 1.6996136903762817,
      "learning_rate": 3.175421098554091e-05,
      "loss": 0.7374,
      "step": 6460
    },
    {
      "epoch": 2.1031229668184777,
      "grad_norm": 2.2226970195770264,
      "learning_rate": 3.172935882300187e-05,
      "loss": 0.7402,
      "step": 6465
    },
    {
      "epoch": 2.1047495120364346,
      "grad_norm": 2.5579731464385986,
      "learning_rate": 3.1704499488923204e-05,
      "loss": 0.7471,
      "step": 6470
    },
    {
      "epoch": 2.1063760572543915,
      "grad_norm": 1.8817853927612305,
      "learning_rate": 3.167963300979773e-05,
      "loss": 0.7068,
      "step": 6475
    },
    {
      "epoch": 2.108002602472349,
      "grad_norm": 1.7366390228271484,
      "learning_rate": 3.165475941212588e-05,
      "loss": 0.7132,
      "step": 6480
    },
    {
      "epoch": 2.1096291476903057,
      "grad_norm": 1.832213282585144,
      "learning_rate": 3.162987872241568e-05,
      "loss": 0.7295,
      "step": 6485
    },
    {
      "epoch": 2.1112556929082626,
      "grad_norm": 1.9638258218765259,
      "learning_rate": 3.160499096718271e-05,
      "loss": 0.7532,
      "step": 6490
    },
    {
      "epoch": 2.11288223812622,
      "grad_norm": 1.7910124063491821,
      "learning_rate": 3.1580096172950076e-05,
      "loss": 0.718,
      "step": 6495
    },
    {
      "epoch": 2.114508783344177,
      "grad_norm": 1.870341181755066,
      "learning_rate": 3.155519436624839e-05,
      "loss": 0.7315,
      "step": 6500
    },
    {
      "epoch": 2.116135328562134,
      "grad_norm": 1.8709512948989868,
      "learning_rate": 3.1530285573615744e-05,
      "loss": 0.7133,
      "step": 6505
    },
    {
      "epoch": 2.117761873780091,
      "grad_norm": 1.7289118766784668,
      "learning_rate": 3.150536982159766e-05,
      "loss": 0.7378,
      "step": 6510
    },
    {
      "epoch": 2.119388418998048,
      "grad_norm": 1.747144341468811,
      "learning_rate": 3.148044713674708e-05,
      "loss": 0.7342,
      "step": 6515
    },
    {
      "epoch": 2.121014964216005,
      "grad_norm": 1.8033082485198975,
      "learning_rate": 3.145551754562435e-05,
      "loss": 0.7491,
      "step": 6520
    },
    {
      "epoch": 2.1226415094339623,
      "grad_norm": 1.4740681648254395,
      "learning_rate": 3.143058107479716e-05,
      "loss": 0.7274,
      "step": 6525
    },
    {
      "epoch": 2.1242680546519193,
      "grad_norm": 1.6742298603057861,
      "learning_rate": 3.1405637750840536e-05,
      "loss": 0.7452,
      "step": 6530
    },
    {
      "epoch": 2.125894599869876,
      "grad_norm": 2.113196611404419,
      "learning_rate": 3.1380687600336804e-05,
      "loss": 0.7333,
      "step": 6535
    },
    {
      "epoch": 2.1275211450878335,
      "grad_norm": 2.39341402053833,
      "learning_rate": 3.135573064987558e-05,
      "loss": 0.7318,
      "step": 6540
    },
    {
      "epoch": 2.1291476903057904,
      "grad_norm": 1.6623047590255737,
      "learning_rate": 3.1330766926053696e-05,
      "loss": 0.732,
      "step": 6545
    },
    {
      "epoch": 2.130774235523748,
      "grad_norm": 3.4303982257843018,
      "learning_rate": 3.130579645547525e-05,
      "loss": 0.7304,
      "step": 6550
    },
    {
      "epoch": 2.1324007807417047,
      "grad_norm": 1.9972918033599854,
      "learning_rate": 3.128081926475148e-05,
      "loss": 0.7587,
      "step": 6555
    },
    {
      "epoch": 2.1340273259596616,
      "grad_norm": 2.0030083656311035,
      "learning_rate": 3.1255835380500816e-05,
      "loss": 0.7177,
      "step": 6560
    },
    {
      "epoch": 2.1356538711776185,
      "grad_norm": 1.7061758041381836,
      "learning_rate": 3.12308448293488e-05,
      "loss": 0.699,
      "step": 6565
    },
    {
      "epoch": 2.137280416395576,
      "grad_norm": 2.000803232192993,
      "learning_rate": 3.1205847637928126e-05,
      "loss": 0.7215,
      "step": 6570
    },
    {
      "epoch": 2.1389069616135328,
      "grad_norm": 1.9984855651855469,
      "learning_rate": 3.11808438328785e-05,
      "loss": 0.7449,
      "step": 6575
    },
    {
      "epoch": 2.14053350683149,
      "grad_norm": 2.0136473178863525,
      "learning_rate": 3.1155833440846715e-05,
      "loss": 0.7581,
      "step": 6580
    },
    {
      "epoch": 2.142160052049447,
      "grad_norm": 2.2597501277923584,
      "learning_rate": 3.113081648848658e-05,
      "loss": 0.714,
      "step": 6585
    },
    {
      "epoch": 2.143786597267404,
      "grad_norm": 9.528533935546875,
      "learning_rate": 3.110579300245889e-05,
      "loss": 0.7427,
      "step": 6590
    },
    {
      "epoch": 2.145413142485361,
      "grad_norm": 1.6550788879394531,
      "learning_rate": 3.10807630094314e-05,
      "loss": 0.7517,
      "step": 6595
    },
    {
      "epoch": 2.147039687703318,
      "grad_norm": 1.8862191438674927,
      "learning_rate": 3.105572653607881e-05,
      "loss": 0.7572,
      "step": 6600
    },
    {
      "epoch": 2.148666232921275,
      "grad_norm": 1.7123699188232422,
      "learning_rate": 3.1030683609082725e-05,
      "loss": 0.7123,
      "step": 6605
    },
    {
      "epoch": 2.1502927781392325,
      "grad_norm": 1.5585683584213257,
      "learning_rate": 3.10056342551316e-05,
      "loss": 0.7351,
      "step": 6610
    },
    {
      "epoch": 2.1519193233571894,
      "grad_norm": 2.117034912109375,
      "learning_rate": 3.098057850092079e-05,
      "loss": 0.7155,
      "step": 6615
    },
    {
      "epoch": 2.1535458685751463,
      "grad_norm": 2.3031628131866455,
      "learning_rate": 3.095551637315243e-05,
      "loss": 0.7411,
      "step": 6620
    },
    {
      "epoch": 2.1551724137931036,
      "grad_norm": 1.798038363456726,
      "learning_rate": 3.093044789853545e-05,
      "loss": 0.7209,
      "step": 6625
    },
    {
      "epoch": 2.1567989590110606,
      "grad_norm": 1.8484983444213867,
      "learning_rate": 3.090537310378558e-05,
      "loss": 0.7439,
      "step": 6630
    },
    {
      "epoch": 2.1584255042290175,
      "grad_norm": 2.020235300064087,
      "learning_rate": 3.088029201562523e-05,
      "loss": 0.7375,
      "step": 6635
    },
    {
      "epoch": 2.160052049446975,
      "grad_norm": 1.7193057537078857,
      "learning_rate": 3.085520466078357e-05,
      "loss": 0.7456,
      "step": 6640
    },
    {
      "epoch": 2.1616785946649317,
      "grad_norm": 1.806383490562439,
      "learning_rate": 3.083011106599641e-05,
      "loss": 0.7164,
      "step": 6645
    },
    {
      "epoch": 2.1633051398828886,
      "grad_norm": 1.6970350742340088,
      "learning_rate": 3.080501125800623e-05,
      "loss": 0.7252,
      "step": 6650
    },
    {
      "epoch": 2.164931685100846,
      "grad_norm": 1.8691726922988892,
      "learning_rate": 3.0779905263562135e-05,
      "loss": 0.7489,
      "step": 6655
    },
    {
      "epoch": 2.166558230318803,
      "grad_norm": 1.8269325494766235,
      "learning_rate": 3.07547931094198e-05,
      "loss": 0.7649,
      "step": 6660
    },
    {
      "epoch": 2.16818477553676,
      "grad_norm": 1.6941983699798584,
      "learning_rate": 3.072967482234148e-05,
      "loss": 0.7388,
      "step": 6665
    },
    {
      "epoch": 2.169811320754717,
      "grad_norm": 1.8301239013671875,
      "learning_rate": 3.070455042909598e-05,
      "loss": 0.7383,
      "step": 6670
    },
    {
      "epoch": 2.171437865972674,
      "grad_norm": 2.8997089862823486,
      "learning_rate": 3.067941995645858e-05,
      "loss": 0.7449,
      "step": 6675
    },
    {
      "epoch": 2.173064411190631,
      "grad_norm": 1.7533553838729858,
      "learning_rate": 3.0654283431211066e-05,
      "loss": 0.7708,
      "step": 6680
    },
    {
      "epoch": 2.1746909564085883,
      "grad_norm": 1.7293111085891724,
      "learning_rate": 3.062914088014167e-05,
      "loss": 0.7573,
      "step": 6685
    },
    {
      "epoch": 2.1763175016265452,
      "grad_norm": 1.7983448505401611,
      "learning_rate": 3.060399233004502e-05,
      "loss": 0.7403,
      "step": 6690
    },
    {
      "epoch": 2.177944046844502,
      "grad_norm": 2.226148843765259,
      "learning_rate": 3.057883780772217e-05,
      "loss": 0.7125,
      "step": 6695
    },
    {
      "epoch": 2.1795705920624595,
      "grad_norm": 1.9191621541976929,
      "learning_rate": 3.055367733998052e-05,
      "loss": 0.7808,
      "step": 6700
    },
    {
      "epoch": 2.1811971372804164,
      "grad_norm": 1.9563281536102295,
      "learning_rate": 3.0528510953633824e-05,
      "loss": 0.7457,
      "step": 6705
    },
    {
      "epoch": 2.1828236824983733,
      "grad_norm": 1.8121241331100464,
      "learning_rate": 3.0503338675502118e-05,
      "loss": 0.7365,
      "step": 6710
    },
    {
      "epoch": 2.1844502277163307,
      "grad_norm": 2.0334646701812744,
      "learning_rate": 3.0478160532411747e-05,
      "loss": 0.7262,
      "step": 6715
    },
    {
      "epoch": 2.1860767729342876,
      "grad_norm": 2.2769808769226074,
      "learning_rate": 3.0452976551195268e-05,
      "loss": 0.7255,
      "step": 6720
    },
    {
      "epoch": 2.1877033181522445,
      "grad_norm": 1.9302563667297363,
      "learning_rate": 3.0427786758691496e-05,
      "loss": 0.7599,
      "step": 6725
    },
    {
      "epoch": 2.189329863370202,
      "grad_norm": 1.4247409105300903,
      "learning_rate": 3.0402591181745428e-05,
      "loss": 0.7228,
      "step": 6730
    },
    {
      "epoch": 2.1909564085881588,
      "grad_norm": 1.5811034440994263,
      "learning_rate": 3.037738984720821e-05,
      "loss": 0.7201,
      "step": 6735
    },
    {
      "epoch": 2.1925829538061157,
      "grad_norm": 1.704803466796875,
      "learning_rate": 3.035218278193714e-05,
      "loss": 0.7429,
      "step": 6740
    },
    {
      "epoch": 2.194209499024073,
      "grad_norm": 1.7844103574752808,
      "learning_rate": 3.0326970012795626e-05,
      "loss": 0.7436,
      "step": 6745
    },
    {
      "epoch": 2.19583604424203,
      "grad_norm": 2.1869356632232666,
      "learning_rate": 3.030175156665314e-05,
      "loss": 0.75,
      "step": 6750
    },
    {
      "epoch": 2.197462589459987,
      "grad_norm": 1.7514694929122925,
      "learning_rate": 3.027652747038522e-05,
      "loss": 0.7291,
      "step": 6755
    },
    {
      "epoch": 2.199089134677944,
      "grad_norm": 1.5369763374328613,
      "learning_rate": 3.0251297750873408e-05,
      "loss": 0.7379,
      "step": 6760
    },
    {
      "epoch": 2.200715679895901,
      "grad_norm": 1.9312034845352173,
      "learning_rate": 3.022606243500526e-05,
      "loss": 0.7354,
      "step": 6765
    },
    {
      "epoch": 2.202342225113858,
      "grad_norm": 1.956268548965454,
      "learning_rate": 3.0200821549674264e-05,
      "loss": 0.7345,
      "step": 6770
    },
    {
      "epoch": 2.2039687703318154,
      "grad_norm": 1.5713850259780884,
      "learning_rate": 3.0175575121779886e-05,
      "loss": 0.7387,
      "step": 6775
    },
    {
      "epoch": 2.2055953155497723,
      "grad_norm": 1.7191925048828125,
      "learning_rate": 3.0150323178227473e-05,
      "loss": 0.725,
      "step": 6780
    },
    {
      "epoch": 2.207221860767729,
      "grad_norm": 1.6469066143035889,
      "learning_rate": 3.012506574592825e-05,
      "loss": 0.6947,
      "step": 6785
    },
    {
      "epoch": 2.2088484059856865,
      "grad_norm": 1.6148409843444824,
      "learning_rate": 3.0099802851799285e-05,
      "loss": 0.7261,
      "step": 6790
    },
    {
      "epoch": 2.2104749512036435,
      "grad_norm": 1.8722944259643555,
      "learning_rate": 3.007453452276349e-05,
      "loss": 0.7269,
      "step": 6795
    },
    {
      "epoch": 2.2121014964216004,
      "grad_norm": 1.997087001800537,
      "learning_rate": 3.0049260785749556e-05,
      "loss": 0.785,
      "step": 6800
    },
    {
      "epoch": 2.2137280416395577,
      "grad_norm": 1.833642601966858,
      "learning_rate": 3.0023981667691926e-05,
      "loss": 0.755,
      "step": 6805
    },
    {
      "epoch": 2.2153545868575146,
      "grad_norm": 1.352949619293213,
      "learning_rate": 2.9998697195530796e-05,
      "loss": 0.7142,
      "step": 6810
    },
    {
      "epoch": 2.2169811320754715,
      "grad_norm": 2.393751382827759,
      "learning_rate": 2.997340739621206e-05,
      "loss": 0.724,
      "step": 6815
    },
    {
      "epoch": 2.218607677293429,
      "grad_norm": 1.5960283279418945,
      "learning_rate": 2.994811229668729e-05,
      "loss": 0.716,
      "step": 6820
    },
    {
      "epoch": 2.220234222511386,
      "grad_norm": 1.8506897687911987,
      "learning_rate": 2.9922811923913714e-05,
      "loss": 0.7783,
      "step": 6825
    },
    {
      "epoch": 2.2218607677293427,
      "grad_norm": 1.6366667747497559,
      "learning_rate": 2.9897506304854157e-05,
      "loss": 0.7637,
      "step": 6830
    },
    {
      "epoch": 2.2234873129473,
      "grad_norm": 1.6875860691070557,
      "learning_rate": 2.9872195466477054e-05,
      "loss": 0.7217,
      "step": 6835
    },
    {
      "epoch": 2.225113858165257,
      "grad_norm": 1.6556414365768433,
      "learning_rate": 2.9846879435756415e-05,
      "loss": 0.6959,
      "step": 6840
    },
    {
      "epoch": 2.226740403383214,
      "grad_norm": 2.3893930912017822,
      "learning_rate": 2.9821558239671744e-05,
      "loss": 0.7626,
      "step": 6845
    },
    {
      "epoch": 2.2283669486011712,
      "grad_norm": 2.506157398223877,
      "learning_rate": 2.9796231905208084e-05,
      "loss": 0.7388,
      "step": 6850
    },
    {
      "epoch": 2.229993493819128,
      "grad_norm": 1.7940316200256348,
      "learning_rate": 2.977090045935594e-05,
      "loss": 0.7482,
      "step": 6855
    },
    {
      "epoch": 2.231620039037085,
      "grad_norm": 2.089195728302002,
      "learning_rate": 2.9745563929111274e-05,
      "loss": 0.7346,
      "step": 6860
    },
    {
      "epoch": 2.2332465842550424,
      "grad_norm": 1.5408855676651,
      "learning_rate": 2.9720222341475445e-05,
      "loss": 0.7277,
      "step": 6865
    },
    {
      "epoch": 2.2348731294729993,
      "grad_norm": 2.133648157119751,
      "learning_rate": 2.969487572345523e-05,
      "loss": 0.7337,
      "step": 6870
    },
    {
      "epoch": 2.2364996746909562,
      "grad_norm": 1.9534162282943726,
      "learning_rate": 2.966952410206275e-05,
      "loss": 0.7325,
      "step": 6875
    },
    {
      "epoch": 2.2381262199089136,
      "grad_norm": 1.9022512435913086,
      "learning_rate": 2.9644167504315458e-05,
      "loss": 0.6883,
      "step": 6880
    },
    {
      "epoch": 2.2397527651268705,
      "grad_norm": 1.4984121322631836,
      "learning_rate": 2.9618805957236113e-05,
      "loss": 0.7144,
      "step": 6885
    },
    {
      "epoch": 2.2413793103448274,
      "grad_norm": 1.6198983192443848,
      "learning_rate": 2.9593439487852753e-05,
      "loss": 0.7098,
      "step": 6890
    },
    {
      "epoch": 2.2430058555627848,
      "grad_norm": 1.6259560585021973,
      "learning_rate": 2.956806812319865e-05,
      "loss": 0.733,
      "step": 6895
    },
    {
      "epoch": 2.2446324007807417,
      "grad_norm": 1.8737263679504395,
      "learning_rate": 2.9542691890312312e-05,
      "loss": 0.7134,
      "step": 6900
    },
    {
      "epoch": 2.2462589459986986,
      "grad_norm": 1.9029572010040283,
      "learning_rate": 2.951731081623742e-05,
      "loss": 0.7171,
      "step": 6905
    },
    {
      "epoch": 2.247885491216656,
      "grad_norm": 1.9204849004745483,
      "learning_rate": 2.9491924928022813e-05,
      "loss": 0.7217,
      "step": 6910
    },
    {
      "epoch": 2.249512036434613,
      "grad_norm": 2.126492977142334,
      "learning_rate": 2.946653425272247e-05,
      "loss": 0.7355,
      "step": 6915
    },
    {
      "epoch": 2.2511385816525697,
      "grad_norm": 2.0191283226013184,
      "learning_rate": 2.944113881739547e-05,
      "loss": 0.7401,
      "step": 6920
    },
    {
      "epoch": 2.252765126870527,
      "grad_norm": 1.8509751558303833,
      "learning_rate": 2.9415738649105963e-05,
      "loss": 0.7331,
      "step": 6925
    },
    {
      "epoch": 2.254391672088484,
      "grad_norm": 2.1766791343688965,
      "learning_rate": 2.9390333774923124e-05,
      "loss": 0.7606,
      "step": 6930
    },
    {
      "epoch": 2.2560182173064414,
      "grad_norm": 1.8282063007354736,
      "learning_rate": 2.9364924221921185e-05,
      "loss": 0.7153,
      "step": 6935
    },
    {
      "epoch": 2.2576447625243983,
      "grad_norm": 1.6795986890792847,
      "learning_rate": 2.9339510017179332e-05,
      "loss": 0.7377,
      "step": 6940
    },
    {
      "epoch": 2.259271307742355,
      "grad_norm": 2.486461877822876,
      "learning_rate": 2.9314091187781715e-05,
      "loss": 0.7285,
      "step": 6945
    },
    {
      "epoch": 2.260897852960312,
      "grad_norm": 1.7224416732788086,
      "learning_rate": 2.9288667760817413e-05,
      "loss": 0.719,
      "step": 6950
    },
    {
      "epoch": 2.2625243981782694,
      "grad_norm": 2.0632481575012207,
      "learning_rate": 2.9263239763380412e-05,
      "loss": 0.7114,
      "step": 6955
    },
    {
      "epoch": 2.2641509433962264,
      "grad_norm": 1.746992588043213,
      "learning_rate": 2.9237807222569558e-05,
      "loss": 0.7529,
      "step": 6960
    },
    {
      "epoch": 2.2657774886141837,
      "grad_norm": 1.490057110786438,
      "learning_rate": 2.921237016548854e-05,
      "loss": 0.7395,
      "step": 6965
    },
    {
      "epoch": 2.2674040338321406,
      "grad_norm": 1.6305809020996094,
      "learning_rate": 2.9186928619245872e-05,
      "loss": 0.7334,
      "step": 6970
    },
    {
      "epoch": 2.2690305790500975,
      "grad_norm": 1.6492929458618164,
      "learning_rate": 2.9161482610954842e-05,
      "loss": 0.7464,
      "step": 6975
    },
    {
      "epoch": 2.2706571242680544,
      "grad_norm": 2.410400390625,
      "learning_rate": 2.913603216773349e-05,
      "loss": 0.7275,
      "step": 6980
    },
    {
      "epoch": 2.272283669486012,
      "grad_norm": 2.1574628353118896,
      "learning_rate": 2.9110577316704602e-05,
      "loss": 0.7284,
      "step": 6985
    },
    {
      "epoch": 2.2739102147039687,
      "grad_norm": 1.6114851236343384,
      "learning_rate": 2.9085118084995626e-05,
      "loss": 0.7167,
      "step": 6990
    },
    {
      "epoch": 2.275536759921926,
      "grad_norm": 2.434839963912964,
      "learning_rate": 2.905965449973871e-05,
      "loss": 0.7251,
      "step": 6995
    },
    {
      "epoch": 2.277163305139883,
      "grad_norm": 2.417687177658081,
      "learning_rate": 2.9034186588070637e-05,
      "loss": 0.7317,
      "step": 7000
    },
    {
      "epoch": 2.27878985035784,
      "grad_norm": 2.3264355659484863,
      "learning_rate": 2.900871437713279e-05,
      "loss": 0.7514,
      "step": 7005
    },
    {
      "epoch": 2.280416395575797,
      "grad_norm": 1.8014682531356812,
      "learning_rate": 2.898323789407113e-05,
      "loss": 0.7324,
      "step": 7010
    },
    {
      "epoch": 2.282042940793754,
      "grad_norm": 3.271698236465454,
      "learning_rate": 2.8957757166036193e-05,
      "loss": 0.6987,
      "step": 7015
    },
    {
      "epoch": 2.283669486011711,
      "grad_norm": 1.8290796279907227,
      "learning_rate": 2.893227222018302e-05,
      "loss": 0.7477,
      "step": 7020
    },
    {
      "epoch": 2.2852960312296684,
      "grad_norm": 1.7644013166427612,
      "learning_rate": 2.890678308367115e-05,
      "loss": 0.7395,
      "step": 7025
    },
    {
      "epoch": 2.2869225764476253,
      "grad_norm": 1.9547981023788452,
      "learning_rate": 2.8881289783664594e-05,
      "loss": 0.7115,
      "step": 7030
    },
    {
      "epoch": 2.288549121665582,
      "grad_norm": 1.6306935548782349,
      "learning_rate": 2.8855792347331793e-05,
      "loss": 0.732,
      "step": 7035
    },
    {
      "epoch": 2.290175666883539,
      "grad_norm": 1.7639504671096802,
      "learning_rate": 2.88302908018456e-05,
      "loss": 0.7446,
      "step": 7040
    },
    {
      "epoch": 2.2918022121014965,
      "grad_norm": 1.500830054283142,
      "learning_rate": 2.8804785174383248e-05,
      "loss": 0.7675,
      "step": 7045
    },
    {
      "epoch": 2.2934287573194534,
      "grad_norm": 1.7111210823059082,
      "learning_rate": 2.8779275492126324e-05,
      "loss": 0.7285,
      "step": 7050
    },
    {
      "epoch": 2.2950553025374107,
      "grad_norm": 1.6088038682937622,
      "learning_rate": 2.8753761782260723e-05,
      "loss": 0.734,
      "step": 7055
    },
    {
      "epoch": 2.2966818477553677,
      "grad_norm": 2.570765256881714,
      "learning_rate": 2.872824407197664e-05,
      "loss": 0.7273,
      "step": 7060
    },
    {
      "epoch": 2.2983083929733246,
      "grad_norm": 2.01545786857605,
      "learning_rate": 2.8702722388468546e-05,
      "loss": 0.7055,
      "step": 7065
    },
    {
      "epoch": 2.2999349381912815,
      "grad_norm": 1.9878475666046143,
      "learning_rate": 2.867719675893512e-05,
      "loss": 0.7317,
      "step": 7070
    },
    {
      "epoch": 2.301561483409239,
      "grad_norm": 2.0020320415496826,
      "learning_rate": 2.8651667210579257e-05,
      "loss": 0.7307,
      "step": 7075
    },
    {
      "epoch": 2.3031880286271957,
      "grad_norm": 1.8539988994598389,
      "learning_rate": 2.8626133770608055e-05,
      "loss": 0.7746,
      "step": 7080
    },
    {
      "epoch": 2.304814573845153,
      "grad_norm": 2.5461862087249756,
      "learning_rate": 2.8600596466232715e-05,
      "loss": 0.7215,
      "step": 7085
    },
    {
      "epoch": 2.30644111906311,
      "grad_norm": 2.5611026287078857,
      "learning_rate": 2.8575055324668583e-05,
      "loss": 0.7412,
      "step": 7090
    },
    {
      "epoch": 2.308067664281067,
      "grad_norm": 1.762474536895752,
      "learning_rate": 2.8549510373135092e-05,
      "loss": 0.7532,
      "step": 7095
    },
    {
      "epoch": 2.3096942094990243,
      "grad_norm": 1.908408522605896,
      "learning_rate": 2.852396163885573e-05,
      "loss": 0.7474,
      "step": 7100
    },
    {
      "epoch": 2.311320754716981,
      "grad_norm": 2.2530789375305176,
      "learning_rate": 2.8498409149058008e-05,
      "loss": 0.7206,
      "step": 7105
    },
    {
      "epoch": 2.312947299934938,
      "grad_norm": 1.8963435888290405,
      "learning_rate": 2.8472852930973464e-05,
      "loss": 0.7429,
      "step": 7110
    },
    {
      "epoch": 2.3145738451528954,
      "grad_norm": 1.78444504737854,
      "learning_rate": 2.8447293011837596e-05,
      "loss": 0.7125,
      "step": 7115
    },
    {
      "epoch": 2.3162003903708523,
      "grad_norm": 1.888048768043518,
      "learning_rate": 2.842172941888983e-05,
      "loss": 0.7308,
      "step": 7120
    },
    {
      "epoch": 2.3178269355888093,
      "grad_norm": 1.765034556388855,
      "learning_rate": 2.8396162179373535e-05,
      "loss": 0.7448,
      "step": 7125
    },
    {
      "epoch": 2.3194534808067666,
      "grad_norm": 1.7544866800308228,
      "learning_rate": 2.837059132053595e-05,
      "loss": 0.7505,
      "step": 7130
    },
    {
      "epoch": 2.3210800260247235,
      "grad_norm": 1.7565666437149048,
      "learning_rate": 2.8345016869628175e-05,
      "loss": 0.7497,
      "step": 7135
    },
    {
      "epoch": 2.3227065712426804,
      "grad_norm": 2.0311663150787354,
      "learning_rate": 2.8319438853905135e-05,
      "loss": 0.7261,
      "step": 7140
    },
    {
      "epoch": 2.324333116460638,
      "grad_norm": 1.872420310974121,
      "learning_rate": 2.8293857300625555e-05,
      "loss": 0.7257,
      "step": 7145
    },
    {
      "epoch": 2.3259596616785947,
      "grad_norm": 1.6592035293579102,
      "learning_rate": 2.826827223705194e-05,
      "loss": 0.7862,
      "step": 7150
    },
    {
      "epoch": 2.3275862068965516,
      "grad_norm": 1.7728911638259888,
      "learning_rate": 2.8242683690450518e-05,
      "loss": 0.7524,
      "step": 7155
    },
    {
      "epoch": 2.329212752114509,
      "grad_norm": 2.1312925815582275,
      "learning_rate": 2.8217091688091253e-05,
      "loss": 0.7479,
      "step": 7160
    },
    {
      "epoch": 2.330839297332466,
      "grad_norm": 1.6046788692474365,
      "learning_rate": 2.8191496257247764e-05,
      "loss": 0.7141,
      "step": 7165
    },
    {
      "epoch": 2.3324658425504228,
      "grad_norm": 1.7915352582931519,
      "learning_rate": 2.816589742519733e-05,
      "loss": 0.7531,
      "step": 7170
    },
    {
      "epoch": 2.33409238776838,
      "grad_norm": 2.469411849975586,
      "learning_rate": 2.814029521922088e-05,
      "loss": 0.7187,
      "step": 7175
    },
    {
      "epoch": 2.335718932986337,
      "grad_norm": 2.290977954864502,
      "learning_rate": 2.8114689666602915e-05,
      "loss": 0.7171,
      "step": 7180
    },
    {
      "epoch": 2.337345478204294,
      "grad_norm": 1.7341169118881226,
      "learning_rate": 2.8089080794631512e-05,
      "loss": 0.7209,
      "step": 7185
    },
    {
      "epoch": 2.3389720234222513,
      "grad_norm": 1.7666329145431519,
      "learning_rate": 2.806346863059827e-05,
      "loss": 0.7159,
      "step": 7190
    },
    {
      "epoch": 2.340598568640208,
      "grad_norm": 2.103515625,
      "learning_rate": 2.803785320179832e-05,
      "loss": 0.707,
      "step": 7195
    },
    {
      "epoch": 2.342225113858165,
      "grad_norm": 1.8098862171173096,
      "learning_rate": 2.801223453553025e-05,
      "loss": 0.7415,
      "step": 7200
    },
    {
      "epoch": 2.3438516590761225,
      "grad_norm": 1.7912166118621826,
      "learning_rate": 2.7986612659096113e-05,
      "loss": 0.7063,
      "step": 7205
    },
    {
      "epoch": 2.3454782042940794,
      "grad_norm": 1.7796494960784912,
      "learning_rate": 2.796098759980138e-05,
      "loss": 0.7492,
      "step": 7210
    },
    {
      "epoch": 2.3471047495120363,
      "grad_norm": 1.6385430097579956,
      "learning_rate": 2.7935359384954914e-05,
      "loss": 0.7538,
      "step": 7215
    },
    {
      "epoch": 2.3487312947299936,
      "grad_norm": 1.807198166847229,
      "learning_rate": 2.7909728041868928e-05,
      "loss": 0.739,
      "step": 7220
    },
    {
      "epoch": 2.3503578399479506,
      "grad_norm": 1.8463494777679443,
      "learning_rate": 2.7884093597858996e-05,
      "loss": 0.7293,
      "step": 7225
    },
    {
      "epoch": 2.3519843851659075,
      "grad_norm": 2.101288080215454,
      "learning_rate": 2.7858456080243972e-05,
      "loss": 0.7285,
      "step": 7230
    },
    {
      "epoch": 2.353610930383865,
      "grad_norm": 2.048839569091797,
      "learning_rate": 2.783281551634599e-05,
      "loss": 0.7299,
      "step": 7235
    },
    {
      "epoch": 2.3552374756018217,
      "grad_norm": 1.6900931596755981,
      "learning_rate": 2.7812300890266442e-05,
      "loss": 0.7163,
      "step": 7240
    },
    {
      "epoch": 2.3568640208197786,
      "grad_norm": 2.0370028018951416,
      "learning_rate": 2.778665491192128e-05,
      "loss": 0.7452,
      "step": 7245
    },
    {
      "epoch": 2.358490566037736,
      "grad_norm": 1.7076982259750366,
      "learning_rate": 2.7761005963812337e-05,
      "loss": 0.7263,
      "step": 7250
    },
    {
      "epoch": 2.360117111255693,
      "grad_norm": 1.7953894138336182,
      "learning_rate": 2.7735354073273926e-05,
      "loss": 0.7123,
      "step": 7255
    },
    {
      "epoch": 2.36174365647365,
      "grad_norm": 1.7410740852355957,
      "learning_rate": 2.7709699267643503e-05,
      "loss": 0.7263,
      "step": 7260
    },
    {
      "epoch": 2.363370201691607,
      "grad_norm": 1.8876742124557495,
      "learning_rate": 2.7684041574261636e-05,
      "loss": 0.7234,
      "step": 7265
    },
    {
      "epoch": 2.364996746909564,
      "grad_norm": 2.0356671810150146,
      "learning_rate": 2.7658381020471964e-05,
      "loss": 0.7437,
      "step": 7270
    },
    {
      "epoch": 2.366623292127521,
      "grad_norm": 1.7301623821258545,
      "learning_rate": 2.7632717633621164e-05,
      "loss": 0.7301,
      "step": 7275
    },
    {
      "epoch": 2.3682498373454783,
      "grad_norm": 2.022576332092285,
      "learning_rate": 2.7607051441058958e-05,
      "loss": 0.7959,
      "step": 7280
    },
    {
      "epoch": 2.3698763825634352,
      "grad_norm": 1.5123145580291748,
      "learning_rate": 2.758138247013804e-05,
      "loss": 0.7239,
      "step": 7285
    },
    {
      "epoch": 2.371502927781392,
      "grad_norm": 1.9764145612716675,
      "learning_rate": 2.7555710748214064e-05,
      "loss": 0.7279,
      "step": 7290
    },
    {
      "epoch": 2.3731294729993495,
      "grad_norm": 2.1461610794067383,
      "learning_rate": 2.7530036302645618e-05,
      "loss": 0.7559,
      "step": 7295
    },
    {
      "epoch": 2.3747560182173064,
      "grad_norm": 7.111647129058838,
      "learning_rate": 2.75043591607942e-05,
      "loss": 0.7275,
      "step": 7300
    },
    {
      "epoch": 2.3763825634352633,
      "grad_norm": 2.709261655807495,
      "learning_rate": 2.7478679350024183e-05,
      "loss": 0.7627,
      "step": 7305
    },
    {
      "epoch": 2.3780091086532207,
      "grad_norm": 1.5946375131607056,
      "learning_rate": 2.7452996897702765e-05,
      "loss": 0.7428,
      "step": 7310
    },
    {
      "epoch": 2.3796356538711776,
      "grad_norm": 1.9049484729766846,
      "learning_rate": 2.7427311831199975e-05,
      "loss": 0.7153,
      "step": 7315
    },
    {
      "epoch": 2.3812621990891345,
      "grad_norm": 1.988143801689148,
      "learning_rate": 2.7401624177888636e-05,
      "loss": 0.7211,
      "step": 7320
    },
    {
      "epoch": 2.382888744307092,
      "grad_norm": 1.6142897605895996,
      "learning_rate": 2.7375933965144296e-05,
      "loss": 0.7196,
      "step": 7325
    },
    {
      "epoch": 2.3845152895250488,
      "grad_norm": 1.8634716272354126,
      "learning_rate": 2.7350241220345274e-05,
      "loss": 0.7182,
      "step": 7330
    },
    {
      "epoch": 2.3861418347430057,
      "grad_norm": 2.3685801029205322,
      "learning_rate": 2.732454597087255e-05,
      "loss": 0.7148,
      "step": 7335
    },
    {
      "epoch": 2.387768379960963,
      "grad_norm": 1.7274894714355469,
      "learning_rate": 2.729884824410979e-05,
      "loss": 0.7079,
      "step": 7340
    },
    {
      "epoch": 2.38939492517892,
      "grad_norm": 1.8336807489395142,
      "learning_rate": 2.72731480674433e-05,
      "loss": 0.7323,
      "step": 7345
    },
    {
      "epoch": 2.391021470396877,
      "grad_norm": 1.8585869073867798,
      "learning_rate": 2.724744546826199e-05,
      "loss": 0.7227,
      "step": 7350
    },
    {
      "epoch": 2.392648015614834,
      "grad_norm": 1.6090168952941895,
      "learning_rate": 2.722174047395737e-05,
      "loss": 0.7303,
      "step": 7355
    },
    {
      "epoch": 2.394274560832791,
      "grad_norm": 1.7963923215866089,
      "learning_rate": 2.719603311192347e-05,
      "loss": 0.744,
      "step": 7360
    },
    {
      "epoch": 2.395901106050748,
      "grad_norm": 1.7062233686447144,
      "learning_rate": 2.7170323409556875e-05,
      "loss": 0.7137,
      "step": 7365
    },
    {
      "epoch": 2.3975276512687054,
      "grad_norm": 1.6454732418060303,
      "learning_rate": 2.7144611394256653e-05,
      "loss": 0.7312,
      "step": 7370
    },
    {
      "epoch": 2.3991541964866623,
      "grad_norm": 1.6548445224761963,
      "learning_rate": 2.7118897093424323e-05,
      "loss": 0.7121,
      "step": 7375
    },
    {
      "epoch": 2.4007807417046196,
      "grad_norm": 1.5025266408920288,
      "learning_rate": 2.7093180534463863e-05,
      "loss": 0.7235,
      "step": 7380
    },
    {
      "epoch": 2.4024072869225765,
      "grad_norm": 1.7081060409545898,
      "learning_rate": 2.7067461744781642e-05,
      "loss": 0.7204,
      "step": 7385
    },
    {
      "epoch": 2.4040338321405335,
      "grad_norm": 1.6674851179122925,
      "learning_rate": 2.7041740751786408e-05,
      "loss": 0.7278,
      "step": 7390
    },
    {
      "epoch": 2.4056603773584904,
      "grad_norm": 1.753976583480835,
      "learning_rate": 2.7016017582889273e-05,
      "loss": 0.6939,
      "step": 7395
    },
    {
      "epoch": 2.4072869225764477,
      "grad_norm": 2.3544299602508545,
      "learning_rate": 2.6990292265503646e-05,
      "loss": 0.7316,
      "step": 7400
    },
    {
      "epoch": 2.4089134677944046,
      "grad_norm": 1.823217749595642,
      "learning_rate": 2.6964564827045224e-05,
      "loss": 0.7389,
      "step": 7405
    },
    {
      "epoch": 2.410540013012362,
      "grad_norm": 1.7922483682632446,
      "learning_rate": 2.6938835294931996e-05,
      "loss": 0.6885,
      "step": 7410
    },
    {
      "epoch": 2.412166558230319,
      "grad_norm": 1.7357585430145264,
      "learning_rate": 2.6913103696584148e-05,
      "loss": 0.7186,
      "step": 7415
    },
    {
      "epoch": 2.413793103448276,
      "grad_norm": 1.9426454305648804,
      "learning_rate": 2.6887370059424078e-05,
      "loss": 0.7357,
      "step": 7420
    },
    {
      "epoch": 2.4154196486662327,
      "grad_norm": 1.9222463369369507,
      "learning_rate": 2.6861634410876367e-05,
      "loss": 0.7493,
      "step": 7425
    },
    {
      "epoch": 2.41704619388419,
      "grad_norm": 2.432889938354492,
      "learning_rate": 2.6835896778367738e-05,
      "loss": 0.7432,
      "step": 7430
    },
    {
      "epoch": 2.418672739102147,
      "grad_norm": 1.7843577861785889,
      "learning_rate": 2.6810157189327007e-05,
      "loss": 0.7344,
      "step": 7435
    },
    {
      "epoch": 2.4202992843201043,
      "grad_norm": 1.835652232170105,
      "learning_rate": 2.6784415671185104e-05,
      "loss": 0.7489,
      "step": 7440
    },
    {
      "epoch": 2.4219258295380612,
      "grad_norm": 3.3983778953552246,
      "learning_rate": 2.6758672251375e-05,
      "loss": 0.7292,
      "step": 7445
    },
    {
      "epoch": 2.423552374756018,
      "grad_norm": 1.7670029401779175,
      "learning_rate": 2.6732926957331688e-05,
      "loss": 0.7487,
      "step": 7450
    },
    {
      "epoch": 2.425178919973975,
      "grad_norm": 1.4413880109786987,
      "learning_rate": 2.6707179816492168e-05,
      "loss": 0.7441,
      "step": 7455
    },
    {
      "epoch": 2.4268054651919324,
      "grad_norm": 1.9677038192749023,
      "learning_rate": 2.668143085629541e-05,
      "loss": 0.7308,
      "step": 7460
    },
    {
      "epoch": 2.4284320104098893,
      "grad_norm": 1.7730647325515747,
      "learning_rate": 2.6655680104182313e-05,
      "loss": 0.7147,
      "step": 7465
    },
    {
      "epoch": 2.4300585556278467,
      "grad_norm": 1.8835413455963135,
      "learning_rate": 2.6629927587595688e-05,
      "loss": 0.7174,
      "step": 7470
    },
    {
      "epoch": 2.4316851008458036,
      "grad_norm": 1.521517038345337,
      "learning_rate": 2.6604173333980237e-05,
      "loss": 0.7237,
      "step": 7475
    },
    {
      "epoch": 2.4333116460637605,
      "grad_norm": 2.2355844974517822,
      "learning_rate": 2.65784173707825e-05,
      "loss": 0.7137,
      "step": 7480
    },
    {
      "epoch": 2.4349381912817174,
      "grad_norm": 1.9166332483291626,
      "learning_rate": 2.6552659725450836e-05,
      "loss": 0.719,
      "step": 7485
    },
    {
      "epoch": 2.4365647364996748,
      "grad_norm": 3.5268030166625977,
      "learning_rate": 2.6526900425435425e-05,
      "loss": 0.7321,
      "step": 7490
    },
    {
      "epoch": 2.4381912817176317,
      "grad_norm": 1.6944290399551392,
      "learning_rate": 2.650113949818817e-05,
      "loss": 0.7286,
      "step": 7495
    },
    {
      "epoch": 2.439817826935589,
      "grad_norm": 1.8654463291168213,
      "learning_rate": 2.6475376971162734e-05,
      "loss": 0.7352,
      "step": 7500
    },
    {
      "epoch": 2.441444372153546,
      "grad_norm": 1.7177530527114868,
      "learning_rate": 2.644961287181449e-05,
      "loss": 0.7472,
      "step": 7505
    },
    {
      "epoch": 2.443070917371503,
      "grad_norm": 1.724672794342041,
      "learning_rate": 2.642384722760046e-05,
      "loss": 0.7525,
      "step": 7510
    },
    {
      "epoch": 2.4446974625894597,
      "grad_norm": 1.7894189357757568,
      "learning_rate": 2.6398080065979346e-05,
      "loss": 0.7591,
      "step": 7515
    },
    {
      "epoch": 2.446324007807417,
      "grad_norm": 1.6924527883529663,
      "learning_rate": 2.6372311414411427e-05,
      "loss": 0.7397,
      "step": 7520
    },
    {
      "epoch": 2.447950553025374,
      "grad_norm": 1.7091771364212036,
      "learning_rate": 2.6346541300358613e-05,
      "loss": 0.7323,
      "step": 7525
    },
    {
      "epoch": 2.4495770982433314,
      "grad_norm": 2.6545259952545166,
      "learning_rate": 2.6320769751284335e-05,
      "loss": 0.7104,
      "step": 7530
    },
    {
      "epoch": 2.4512036434612883,
      "grad_norm": 1.6754066944122314,
      "learning_rate": 2.6294996794653576e-05,
      "loss": 0.7345,
      "step": 7535
    },
    {
      "epoch": 2.452830188679245,
      "grad_norm": 1.8702524900436401,
      "learning_rate": 2.6269222457932824e-05,
      "loss": 0.7287,
      "step": 7540
    },
    {
      "epoch": 2.4544567338972025,
      "grad_norm": 1.705138087272644,
      "learning_rate": 2.6243446768590015e-05,
      "loss": 0.7213,
      "step": 7545
    },
    {
      "epoch": 2.4560832791151594,
      "grad_norm": 1.8247156143188477,
      "learning_rate": 2.621766975409453e-05,
      "loss": 0.7461,
      "step": 7550
    },
    {
      "epoch": 2.4577098243331164,
      "grad_norm": 1.8440947532653809,
      "learning_rate": 2.6191891441917195e-05,
      "loss": 0.7163,
      "step": 7555
    },
    {
      "epoch": 2.4593363695510737,
      "grad_norm": 1.5848308801651,
      "learning_rate": 2.616611185953018e-05,
      "loss": 0.7156,
      "step": 7560
    },
    {
      "epoch": 2.4609629147690306,
      "grad_norm": 1.6326448917388916,
      "learning_rate": 2.6140331034407013e-05,
      "loss": 0.7358,
      "step": 7565
    },
    {
      "epoch": 2.4625894599869875,
      "grad_norm": 1.6340683698654175,
      "learning_rate": 2.6114548994022576e-05,
      "loss": 0.7263,
      "step": 7570
    },
    {
      "epoch": 2.464216005204945,
      "grad_norm": 1.7180076837539673,
      "learning_rate": 2.608876576585302e-05,
      "loss": 0.7499,
      "step": 7575
    },
    {
      "epoch": 2.465842550422902,
      "grad_norm": 1.7165942192077637,
      "learning_rate": 2.6062981377375762e-05,
      "loss": 0.7535,
      "step": 7580
    },
    {
      "epoch": 2.4674690956408587,
      "grad_norm": 1.7136976718902588,
      "learning_rate": 2.6037195856069462e-05,
      "loss": 0.7404,
      "step": 7585
    },
    {
      "epoch": 2.469095640858816,
      "grad_norm": 1.7513527870178223,
      "learning_rate": 2.6011409229414003e-05,
      "loss": 0.725,
      "step": 7590
    },
    {
      "epoch": 2.470722186076773,
      "grad_norm": 2.0561563968658447,
      "learning_rate": 2.598562152489042e-05,
      "loss": 0.7324,
      "step": 7595
    },
    {
      "epoch": 2.47234873129473,
      "grad_norm": 2.355729818344116,
      "learning_rate": 2.59598327699809e-05,
      "loss": 0.729,
      "step": 7600
    },
    {
      "epoch": 2.4739752765126872,
      "grad_norm": 1.8387353420257568,
      "learning_rate": 2.5934042992168767e-05,
      "loss": 0.7447,
      "step": 7605
    },
    {
      "epoch": 2.475601821730644,
      "grad_norm": 1.8397676944732666,
      "learning_rate": 2.5908252218938423e-05,
      "loss": 0.7583,
      "step": 7610
    },
    {
      "epoch": 2.477228366948601,
      "grad_norm": 1.6779429912567139,
      "learning_rate": 2.5882460477775326e-05,
      "loss": 0.7119,
      "step": 7615
    },
    {
      "epoch": 2.4788549121665584,
      "grad_norm": 1.7980512380599976,
      "learning_rate": 2.585666779616598e-05,
      "loss": 0.746,
      "step": 7620
    },
    {
      "epoch": 2.4804814573845153,
      "grad_norm": 2.2904257774353027,
      "learning_rate": 2.5830874201597883e-05,
      "loss": 0.7102,
      "step": 7625
    },
    {
      "epoch": 2.482108002602472,
      "grad_norm": 1.5908973217010498,
      "learning_rate": 2.5805079721559494e-05,
      "loss": 0.7294,
      "step": 7630
    },
    {
      "epoch": 2.4837345478204296,
      "grad_norm": 1.8799176216125488,
      "learning_rate": 2.577928438354024e-05,
      "loss": 0.7328,
      "step": 7635
    },
    {
      "epoch": 2.4853610930383865,
      "grad_norm": 1.6054154634475708,
      "learning_rate": 2.5753488215030448e-05,
      "loss": 0.7295,
      "step": 7640
    },
    {
      "epoch": 2.4869876382563434,
      "grad_norm": 1.764978289604187,
      "learning_rate": 2.5727691243521325e-05,
      "loss": 0.7409,
      "step": 7645
    },
    {
      "epoch": 2.4886141834743007,
      "grad_norm": 1.6741787195205688,
      "learning_rate": 2.5701893496504953e-05,
      "loss": 0.744,
      "step": 7650
    },
    {
      "epoch": 2.4902407286922577,
      "grad_norm": 1.697543740272522,
      "learning_rate": 2.567609500147422e-05,
      "loss": 0.7269,
      "step": 7655
    },
    {
      "epoch": 2.4918672739102146,
      "grad_norm": 1.884772539138794,
      "learning_rate": 2.5650295785922817e-05,
      "loss": 0.7128,
      "step": 7660
    },
    {
      "epoch": 2.493493819128172,
      "grad_norm": 1.6686285734176636,
      "learning_rate": 2.562449587734522e-05,
      "loss": 0.7619,
      "step": 7665
    },
    {
      "epoch": 2.495120364346129,
      "grad_norm": 2.351414918899536,
      "learning_rate": 2.5598695303236615e-05,
      "loss": 0.7191,
      "step": 7670
    },
    {
      "epoch": 2.4967469095640857,
      "grad_norm": 1.8593933582305908,
      "learning_rate": 2.557289409109291e-05,
      "loss": 0.7325,
      "step": 7675
    },
    {
      "epoch": 2.498373454782043,
      "grad_norm": 1.7637239694595337,
      "learning_rate": 2.5547092268410703e-05,
      "loss": 0.7151,
      "step": 7680
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0318996906280518,
      "learning_rate": 2.5521289862687237e-05,
      "loss": 0.7238,
      "step": 7685
    },
    {
      "epoch": 2.501626545217957,
      "grad_norm": 1.8583095073699951,
      "learning_rate": 2.5495486901420362e-05,
      "loss": 0.7234,
      "step": 7690
    },
    {
      "epoch": 2.5032530904359143,
      "grad_norm": 1.5586649179458618,
      "learning_rate": 2.546968341210853e-05,
      "loss": 0.7417,
      "step": 7695
    },
    {
      "epoch": 2.504879635653871,
      "grad_norm": 1.831864833831787,
      "learning_rate": 2.5443879422250767e-05,
      "loss": 0.7334,
      "step": 7700
    },
    {
      "epoch": 2.506506180871828,
      "grad_norm": 1.5912280082702637,
      "learning_rate": 2.541807495934662e-05,
      "loss": 0.7356,
      "step": 7705
    },
    {
      "epoch": 2.5081327260897854,
      "grad_norm": 1.8658093214035034,
      "learning_rate": 2.539227005089614e-05,
      "loss": 0.7337,
      "step": 7710
    },
    {
      "epoch": 2.5097592713077423,
      "grad_norm": 1.8022907972335815,
      "learning_rate": 2.536646472439986e-05,
      "loss": 0.7362,
      "step": 7715
    },
    {
      "epoch": 2.5113858165256993,
      "grad_norm": 1.79240083694458,
      "learning_rate": 2.5340659007358742e-05,
      "loss": 0.7518,
      "step": 7720
    },
    {
      "epoch": 2.5130123617436566,
      "grad_norm": 2.8876941204071045,
      "learning_rate": 2.531485292727419e-05,
      "loss": 0.7582,
      "step": 7725
    },
    {
      "epoch": 2.5146389069616135,
      "grad_norm": 1.7138187885284424,
      "learning_rate": 2.5289046511647972e-05,
      "loss": 0.7589,
      "step": 7730
    },
    {
      "epoch": 2.5162654521795704,
      "grad_norm": 1.4874415397644043,
      "learning_rate": 2.5263239787982234e-05,
      "loss": 0.7397,
      "step": 7735
    },
    {
      "epoch": 2.517891997397528,
      "grad_norm": 1.6347683668136597,
      "learning_rate": 2.523743278377943e-05,
      "loss": 0.685,
      "step": 7740
    },
    {
      "epoch": 2.5195185426154847,
      "grad_norm": 1.623404860496521,
      "learning_rate": 2.5211625526542313e-05,
      "loss": 0.7399,
      "step": 7745
    },
    {
      "epoch": 2.5211450878334416,
      "grad_norm": 2.0945160388946533,
      "learning_rate": 2.518581804377394e-05,
      "loss": 0.7458,
      "step": 7750
    },
    {
      "epoch": 2.522771633051399,
      "grad_norm": 2.3906166553497314,
      "learning_rate": 2.516001036297756e-05,
      "loss": 0.7179,
      "step": 7755
    },
    {
      "epoch": 2.524398178269356,
      "grad_norm": 1.6636216640472412,
      "learning_rate": 2.5134202511656658e-05,
      "loss": 0.7228,
      "step": 7760
    },
    {
      "epoch": 2.526024723487313,
      "grad_norm": 1.6783771514892578,
      "learning_rate": 2.5108394517314915e-05,
      "loss": 0.7403,
      "step": 7765
    },
    {
      "epoch": 2.52765126870527,
      "grad_norm": 1.7963955402374268,
      "learning_rate": 2.5082586407456134e-05,
      "loss": 0.7181,
      "step": 7770
    },
    {
      "epoch": 2.529277813923227,
      "grad_norm": 2.11568021774292,
      "learning_rate": 2.5056778209584252e-05,
      "loss": 0.7135,
      "step": 7775
    },
    {
      "epoch": 2.530904359141184,
      "grad_norm": 2.0192832946777344,
      "learning_rate": 2.5030969951203316e-05,
      "loss": 0.7509,
      "step": 7780
    },
    {
      "epoch": 2.5325309043591413,
      "grad_norm": 1.4360262155532837,
      "learning_rate": 2.5005161659817418e-05,
      "loss": 0.7579,
      "step": 7785
    },
    {
      "epoch": 2.534157449577098,
      "grad_norm": 2.1884679794311523,
      "learning_rate": 2.4979353362930685e-05,
      "loss": 0.7375,
      "step": 7790
    },
    {
      "epoch": 2.5357839947950556,
      "grad_norm": 1.7255921363830566,
      "learning_rate": 2.4953545088047263e-05,
      "loss": 0.7177,
      "step": 7795
    },
    {
      "epoch": 2.5374105400130125,
      "grad_norm": 1.8770041465759277,
      "learning_rate": 2.492773686267128e-05,
      "loss": 0.7094,
      "step": 7800
    },
    {
      "epoch": 2.5390370852309694,
      "grad_norm": 1.7790896892547607,
      "learning_rate": 2.4901928714306777e-05,
      "loss": 0.7168,
      "step": 7805
    },
    {
      "epoch": 2.5406636304489263,
      "grad_norm": 1.711310863494873,
      "learning_rate": 2.4876120670457754e-05,
      "loss": 0.7277,
      "step": 7810
    },
    {
      "epoch": 2.5422901756668836,
      "grad_norm": 1.9501806497573853,
      "learning_rate": 2.4850312758628062e-05,
      "loss": 0.7278,
      "step": 7815
    },
    {
      "epoch": 2.5439167208848406,
      "grad_norm": 1.7433342933654785,
      "learning_rate": 2.482450500632145e-05,
      "loss": 0.7233,
      "step": 7820
    },
    {
      "epoch": 2.545543266102798,
      "grad_norm": 1.9563379287719727,
      "learning_rate": 2.479869744104146e-05,
      "loss": 0.7464,
      "step": 7825
    },
    {
      "epoch": 2.547169811320755,
      "grad_norm": 1.655651330947876,
      "learning_rate": 2.477289009029147e-05,
      "loss": 0.7808,
      "step": 7830
    },
    {
      "epoch": 2.5487963565387117,
      "grad_norm": 1.86355459690094,
      "learning_rate": 2.4747082981574593e-05,
      "loss": 0.7108,
      "step": 7835
    },
    {
      "epoch": 2.5504229017566686,
      "grad_norm": 1.6612073183059692,
      "learning_rate": 2.4721276142393714e-05,
      "loss": 0.6986,
      "step": 7840
    },
    {
      "epoch": 2.552049446974626,
      "grad_norm": 1.7691892385482788,
      "learning_rate": 2.469546960025141e-05,
      "loss": 0.7595,
      "step": 7845
    },
    {
      "epoch": 2.553675992192583,
      "grad_norm": 1.760162353515625,
      "learning_rate": 2.4669663382649967e-05,
      "loss": 0.7267,
      "step": 7850
    },
    {
      "epoch": 2.5553025374105403,
      "grad_norm": 1.884227991104126,
      "learning_rate": 2.464385751709128e-05,
      "loss": 0.7314,
      "step": 7855
    },
    {
      "epoch": 2.556929082628497,
      "grad_norm": 1.6329880952835083,
      "learning_rate": 2.4618052031076933e-05,
      "loss": 0.7272,
      "step": 7860
    },
    {
      "epoch": 2.558555627846454,
      "grad_norm": 1.9245282411575317,
      "learning_rate": 2.459224695210804e-05,
      "loss": 0.6923,
      "step": 7865
    },
    {
      "epoch": 2.560182173064411,
      "grad_norm": 1.7136482000350952,
      "learning_rate": 2.4566442307685325e-05,
      "loss": 0.7046,
      "step": 7870
    },
    {
      "epoch": 2.5618087182823683,
      "grad_norm": 1.7598689794540405,
      "learning_rate": 2.4540638125309032e-05,
      "loss": 0.7222,
      "step": 7875
    },
    {
      "epoch": 2.5634352635003252,
      "grad_norm": 1.9097983837127686,
      "learning_rate": 2.4514834432478927e-05,
      "loss": 0.7007,
      "step": 7880
    },
    {
      "epoch": 2.5650618087182826,
      "grad_norm": 1.859815001487732,
      "learning_rate": 2.4489031256694212e-05,
      "loss": 0.7399,
      "step": 7885
    },
    {
      "epoch": 2.5666883539362395,
      "grad_norm": 1.593043565750122,
      "learning_rate": 2.4463228625453607e-05,
      "loss": 0.7125,
      "step": 7890
    },
    {
      "epoch": 2.5683148991541964,
      "grad_norm": 1.563043475151062,
      "learning_rate": 2.4437426566255188e-05,
      "loss": 0.6984,
      "step": 7895
    },
    {
      "epoch": 2.5699414443721533,
      "grad_norm": 1.6478209495544434,
      "learning_rate": 2.4411625106596457e-05,
      "loss": 0.7491,
      "step": 7900
    },
    {
      "epoch": 2.5715679895901107,
      "grad_norm": 1.6165393590927124,
      "learning_rate": 2.4385824273974262e-05,
      "loss": 0.7488,
      "step": 7905
    },
    {
      "epoch": 2.5731945348080676,
      "grad_norm": 1.757622480392456,
      "learning_rate": 2.43600240958848e-05,
      "loss": 0.7273,
      "step": 7910
    },
    {
      "epoch": 2.574821080026025,
      "grad_norm": 1.9844690561294556,
      "learning_rate": 2.433422459982354e-05,
      "loss": 0.7552,
      "step": 7915
    },
    {
      "epoch": 2.576447625243982,
      "grad_norm": 1.7524667978286743,
      "learning_rate": 2.4308425813285255e-05,
      "loss": 0.7179,
      "step": 7920
    },
    {
      "epoch": 2.5780741704619388,
      "grad_norm": 1.6379843950271606,
      "learning_rate": 2.428262776376394e-05,
      "loss": 0.74,
      "step": 7925
    },
    {
      "epoch": 2.5797007156798957,
      "grad_norm": 1.8496534824371338,
      "learning_rate": 2.425683047875282e-05,
      "loss": 0.7194,
      "step": 7930
    },
    {
      "epoch": 2.581327260897853,
      "grad_norm": 1.599732518196106,
      "learning_rate": 2.4231033985744305e-05,
      "loss": 0.7326,
      "step": 7935
    },
    {
      "epoch": 2.58295380611581,
      "grad_norm": 1.7700289487838745,
      "learning_rate": 2.420523831222994e-05,
      "loss": 0.7481,
      "step": 7940
    },
    {
      "epoch": 2.5845803513337673,
      "grad_norm": 1.5851463079452515,
      "learning_rate": 2.4179443485700434e-05,
      "loss": 0.7573,
      "step": 7945
    },
    {
      "epoch": 2.586206896551724,
      "grad_norm": 1.7170052528381348,
      "learning_rate": 2.4153649533645545e-05,
      "loss": 0.7222,
      "step": 7950
    },
    {
      "epoch": 2.587833441769681,
      "grad_norm": 1.7591465711593628,
      "learning_rate": 2.412785648355414e-05,
      "loss": 0.7233,
      "step": 7955
    },
    {
      "epoch": 2.589459986987638,
      "grad_norm": 2.046952724456787,
      "learning_rate": 2.4102064362914108e-05,
      "loss": 0.7176,
      "step": 7960
    },
    {
      "epoch": 2.5910865322055954,
      "grad_norm": 1.6409968137741089,
      "learning_rate": 2.4076273199212352e-05,
      "loss": 0.7139,
      "step": 7965
    },
    {
      "epoch": 2.5927130774235523,
      "grad_norm": 2.4438061714172363,
      "learning_rate": 2.4050483019934737e-05,
      "loss": 0.7431,
      "step": 7970
    },
    {
      "epoch": 2.5943396226415096,
      "grad_norm": 1.9839977025985718,
      "learning_rate": 2.4024693852566124e-05,
      "loss": 0.7305,
      "step": 7975
    },
    {
      "epoch": 2.5959661678594665,
      "grad_norm": 1.5234110355377197,
      "learning_rate": 2.3998905724590237e-05,
      "loss": 0.6977,
      "step": 7980
    },
    {
      "epoch": 2.5975927130774235,
      "grad_norm": 1.705267310142517,
      "learning_rate": 2.3973118663489736e-05,
      "loss": 0.6917,
      "step": 7985
    },
    {
      "epoch": 2.5992192582953804,
      "grad_norm": 1.8564801216125488,
      "learning_rate": 2.3947332696746122e-05,
      "loss": 0.7277,
      "step": 7990
    },
    {
      "epoch": 2.6008458035133377,
      "grad_norm": 2.0818612575531006,
      "learning_rate": 2.3921547851839747e-05,
      "loss": 0.7544,
      "step": 7995
    },
    {
      "epoch": 2.6024723487312946,
      "grad_norm": 1.8174341917037964,
      "learning_rate": 2.3895764156249746e-05,
      "loss": 0.7192,
      "step": 8000
    },
    {
      "epoch": 2.604098893949252,
      "grad_norm": 1.606379508972168,
      "learning_rate": 2.3869981637454053e-05,
      "loss": 0.7222,
      "step": 8005
    },
    {
      "epoch": 2.605725439167209,
      "grad_norm": 1.8314141035079956,
      "learning_rate": 2.3844200322929323e-05,
      "loss": 0.7135,
      "step": 8010
    },
    {
      "epoch": 2.607351984385166,
      "grad_norm": 1.7894541025161743,
      "learning_rate": 2.3818420240150947e-05,
      "loss": 0.7342,
      "step": 8015
    },
    {
      "epoch": 2.6089785296031227,
      "grad_norm": 2.068234443664551,
      "learning_rate": 2.3792641416592994e-05,
      "loss": 0.7033,
      "step": 8020
    },
    {
      "epoch": 2.61060507482108,
      "grad_norm": 1.7998398542404175,
      "learning_rate": 2.376686387972821e-05,
      "loss": 0.7421,
      "step": 8025
    },
    {
      "epoch": 2.612231620039037,
      "grad_norm": 2.042323589324951,
      "learning_rate": 2.3741087657027912e-05,
      "loss": 0.7232,
      "step": 8030
    },
    {
      "epoch": 2.6138581652569943,
      "grad_norm": 1.5981550216674805,
      "learning_rate": 2.3715312775962105e-05,
      "loss": 0.7524,
      "step": 8035
    },
    {
      "epoch": 2.6154847104749512,
      "grad_norm": 1.6464756727218628,
      "learning_rate": 2.3689539263999286e-05,
      "loss": 0.7407,
      "step": 8040
    },
    {
      "epoch": 2.617111255692908,
      "grad_norm": 4.4229021072387695,
      "learning_rate": 2.366376714860654e-05,
      "loss": 0.7646,
      "step": 8045
    },
    {
      "epoch": 2.618737800910865,
      "grad_norm": 1.9497524499893188,
      "learning_rate": 2.3637996457249434e-05,
      "loss": 0.7191,
      "step": 8050
    },
    {
      "epoch": 2.6203643461288224,
      "grad_norm": 2.853574514389038,
      "learning_rate": 2.361222721739205e-05,
      "loss": 0.7229,
      "step": 8055
    },
    {
      "epoch": 2.6219908913467793,
      "grad_norm": 1.6402703523635864,
      "learning_rate": 2.3586459456496877e-05,
      "loss": 0.7226,
      "step": 8060
    },
    {
      "epoch": 2.6236174365647367,
      "grad_norm": 1.9785535335540771,
      "learning_rate": 2.3560693202024877e-05,
      "loss": 0.7016,
      "step": 8065
    },
    {
      "epoch": 2.6252439817826936,
      "grad_norm": 1.7689776420593262,
      "learning_rate": 2.3534928481435388e-05,
      "loss": 0.7107,
      "step": 8070
    },
    {
      "epoch": 2.6268705270006505,
      "grad_norm": 2.1093742847442627,
      "learning_rate": 2.3509165322186084e-05,
      "loss": 0.7173,
      "step": 8075
    },
    {
      "epoch": 2.6284970722186074,
      "grad_norm": 1.9455536603927612,
      "learning_rate": 2.348340375173303e-05,
      "loss": 0.7421,
      "step": 8080
    },
    {
      "epoch": 2.6301236174365648,
      "grad_norm": 1.7876079082489014,
      "learning_rate": 2.3457643797530544e-05,
      "loss": 0.7278,
      "step": 8085
    },
    {
      "epoch": 2.6317501626545217,
      "grad_norm": 1.8814291954040527,
      "learning_rate": 2.343188548703127e-05,
      "loss": 0.7403,
      "step": 8090
    },
    {
      "epoch": 2.633376707872479,
      "grad_norm": 1.7408051490783691,
      "learning_rate": 2.340612884768605e-05,
      "loss": 0.7042,
      "step": 8095
    },
    {
      "epoch": 2.635003253090436,
      "grad_norm": 2.139902353286743,
      "learning_rate": 2.3380373906944004e-05,
      "loss": 0.6996,
      "step": 8100
    },
    {
      "epoch": 2.636629798308393,
      "grad_norm": 1.5826632976531982,
      "learning_rate": 2.3354620692252377e-05,
      "loss": 0.7646,
      "step": 8105
    },
    {
      "epoch": 2.63825634352635,
      "grad_norm": 2.2483596801757812,
      "learning_rate": 2.3328869231056627e-05,
      "loss": 0.742,
      "step": 8110
    },
    {
      "epoch": 2.639882888744307,
      "grad_norm": 1.6104656457901,
      "learning_rate": 2.330311955080031e-05,
      "loss": 0.7283,
      "step": 8115
    },
    {
      "epoch": 2.641509433962264,
      "grad_norm": 1.8288098573684692,
      "learning_rate": 2.3277371678925123e-05,
      "loss": 0.7291,
      "step": 8120
    },
    {
      "epoch": 2.6431359791802214,
      "grad_norm": 1.7580839395523071,
      "learning_rate": 2.3251625642870774e-05,
      "loss": 0.7187,
      "step": 8125
    },
    {
      "epoch": 2.6447625243981783,
      "grad_norm": 1.7375248670578003,
      "learning_rate": 2.3225881470075075e-05,
      "loss": 0.7481,
      "step": 8130
    },
    {
      "epoch": 2.646389069616135,
      "grad_norm": 1.697384238243103,
      "learning_rate": 2.3200139187973818e-05,
      "loss": 0.7052,
      "step": 8135
    },
    {
      "epoch": 2.6480156148340925,
      "grad_norm": 2.009641647338867,
      "learning_rate": 2.3174398824000794e-05,
      "loss": 0.7042,
      "step": 8140
    },
    {
      "epoch": 2.6496421600520494,
      "grad_norm": 1.9816298484802246,
      "learning_rate": 2.3148660405587745e-05,
      "loss": 0.7353,
      "step": 8145
    },
    {
      "epoch": 2.6512687052700064,
      "grad_norm": 1.696813702583313,
      "learning_rate": 2.3122923960164346e-05,
      "loss": 0.723,
      "step": 8150
    },
    {
      "epoch": 2.6528952504879637,
      "grad_norm": 8.96198844909668,
      "learning_rate": 2.309718951515815e-05,
      "loss": 0.7429,
      "step": 8155
    },
    {
      "epoch": 2.6545217957059206,
      "grad_norm": 1.714002013206482,
      "learning_rate": 2.3071457097994607e-05,
      "loss": 0.7447,
      "step": 8160
    },
    {
      "epoch": 2.6561483409238775,
      "grad_norm": 1.8118575811386108,
      "learning_rate": 2.3045726736096978e-05,
      "loss": 0.7204,
      "step": 8165
    },
    {
      "epoch": 2.657774886141835,
      "grad_norm": 7.933690547943115,
      "learning_rate": 2.301999845688635e-05,
      "loss": 0.7226,
      "step": 8170
    },
    {
      "epoch": 2.659401431359792,
      "grad_norm": 2.298539638519287,
      "learning_rate": 2.2994272287781593e-05,
      "loss": 0.7301,
      "step": 8175
    },
    {
      "epoch": 2.6610279765777487,
      "grad_norm": 1.950821042060852,
      "learning_rate": 2.2968548256199325e-05,
      "loss": 0.7278,
      "step": 8180
    },
    {
      "epoch": 2.662654521795706,
      "grad_norm": 2.008021831512451,
      "learning_rate": 2.2942826389553865e-05,
      "loss": 0.7421,
      "step": 8185
    },
    {
      "epoch": 2.664281067013663,
      "grad_norm": 1.7788177728652954,
      "learning_rate": 2.291710671525726e-05,
      "loss": 0.7472,
      "step": 8190
    },
    {
      "epoch": 2.66590761223162,
      "grad_norm": 1.5617021322250366,
      "learning_rate": 2.2891389260719197e-05,
      "loss": 0.7154,
      "step": 8195
    },
    {
      "epoch": 2.6675341574495772,
      "grad_norm": 2.134261131286621,
      "learning_rate": 2.2865674053347e-05,
      "loss": 0.7321,
      "step": 8200
    },
    {
      "epoch": 2.669160702667534,
      "grad_norm": 1.7520164251327515,
      "learning_rate": 2.2839961120545614e-05,
      "loss": 0.7015,
      "step": 8205
    },
    {
      "epoch": 2.6707872478854915,
      "grad_norm": 1.7123262882232666,
      "learning_rate": 2.2814250489717536e-05,
      "loss": 0.7041,
      "step": 8210
    },
    {
      "epoch": 2.6724137931034484,
      "grad_norm": 3.0392608642578125,
      "learning_rate": 2.278854218826284e-05,
      "loss": 0.7253,
      "step": 8215
    },
    {
      "epoch": 2.6740403383214053,
      "grad_norm": 1.6789202690124512,
      "learning_rate": 2.2762836243579073e-05,
      "loss": 0.7321,
      "step": 8220
    },
    {
      "epoch": 2.675666883539362,
      "grad_norm": 1.9678678512573242,
      "learning_rate": 2.2737132683061318e-05,
      "loss": 0.7309,
      "step": 8225
    },
    {
      "epoch": 2.6772934287573196,
      "grad_norm": 2.1092047691345215,
      "learning_rate": 2.2711431534102082e-05,
      "loss": 0.767,
      "step": 8230
    },
    {
      "epoch": 2.6789199739752765,
      "grad_norm": 1.7379661798477173,
      "learning_rate": 2.268573282409133e-05,
      "loss": 0.7189,
      "step": 8235
    },
    {
      "epoch": 2.680546519193234,
      "grad_norm": 1.8567099571228027,
      "learning_rate": 2.26600365804164e-05,
      "loss": 0.7098,
      "step": 8240
    },
    {
      "epoch": 2.6821730644111907,
      "grad_norm": 2.021911382675171,
      "learning_rate": 2.263434283046203e-05,
      "loss": 0.7304,
      "step": 8245
    },
    {
      "epoch": 2.6837996096291477,
      "grad_norm": 1.7543399333953857,
      "learning_rate": 2.260865160161027e-05,
      "loss": 0.7289,
      "step": 8250
    },
    {
      "epoch": 2.6854261548471046,
      "grad_norm": 1.5903064012527466,
      "learning_rate": 2.2582962921240507e-05,
      "loss": 0.7363,
      "step": 8255
    },
    {
      "epoch": 2.687052700065062,
      "grad_norm": 1.5687363147735596,
      "learning_rate": 2.2557276816729404e-05,
      "loss": 0.7124,
      "step": 8260
    },
    {
      "epoch": 2.688679245283019,
      "grad_norm": 1.7256449460983276,
      "learning_rate": 2.253159331545089e-05,
      "loss": 0.7333,
      "step": 8265
    },
    {
      "epoch": 2.690305790500976,
      "grad_norm": 1.9225983619689941,
      "learning_rate": 2.2505912444776076e-05,
      "loss": 0.7302,
      "step": 8270
    },
    {
      "epoch": 2.691932335718933,
      "grad_norm": 1.7383098602294922,
      "learning_rate": 2.248023423207334e-05,
      "loss": 0.7349,
      "step": 8275
    },
    {
      "epoch": 2.69355888093689,
      "grad_norm": 1.5088884830474854,
      "learning_rate": 2.2454558704708165e-05,
      "loss": 0.7425,
      "step": 8280
    },
    {
      "epoch": 2.695185426154847,
      "grad_norm": 2.35827374458313,
      "learning_rate": 2.242888589004321e-05,
      "loss": 0.7031,
      "step": 8285
    },
    {
      "epoch": 2.6968119713728043,
      "grad_norm": 1.793815016746521,
      "learning_rate": 2.240321581543822e-05,
      "loss": 0.7313,
      "step": 8290
    },
    {
      "epoch": 2.698438516590761,
      "grad_norm": 2.1256535053253174,
      "learning_rate": 2.2377548508250043e-05,
      "loss": 0.7174,
      "step": 8295
    },
    {
      "epoch": 2.7000650618087185,
      "grad_norm": 1.643990397453308,
      "learning_rate": 2.2351883995832545e-05,
      "loss": 0.7239,
      "step": 8300
    },
    {
      "epoch": 2.7016916070266754,
      "grad_norm": 1.457271933555603,
      "learning_rate": 2.2326222305536653e-05,
      "loss": 0.7184,
      "step": 8305
    },
    {
      "epoch": 2.7033181522446323,
      "grad_norm": 1.6957201957702637,
      "learning_rate": 2.230056346471025e-05,
      "loss": 0.7152,
      "step": 8310
    },
    {
      "epoch": 2.7049446974625893,
      "grad_norm": 1.8537200689315796,
      "learning_rate": 2.2274907500698212e-05,
      "loss": 0.7005,
      "step": 8315
    },
    {
      "epoch": 2.7065712426805466,
      "grad_norm": 1.7019157409667969,
      "learning_rate": 2.2249254440842324e-05,
      "loss": 0.7497,
      "step": 8320
    },
    {
      "epoch": 2.7081977878985035,
      "grad_norm": 1.6967856884002686,
      "learning_rate": 2.222360431248131e-05,
      "loss": 0.7131,
      "step": 8325
    },
    {
      "epoch": 2.709824333116461,
      "grad_norm": 1.7513232231140137,
      "learning_rate": 2.2197957142950714e-05,
      "loss": 0.6965,
      "step": 8330
    },
    {
      "epoch": 2.711450878334418,
      "grad_norm": 2.3701894283294678,
      "learning_rate": 2.2172312959582983e-05,
      "loss": 0.7165,
      "step": 8335
    },
    {
      "epoch": 2.7130774235523747,
      "grad_norm": 1.6562988758087158,
      "learning_rate": 2.214667178970736e-05,
      "loss": 0.7289,
      "step": 8340
    },
    {
      "epoch": 2.7147039687703316,
      "grad_norm": 1.782637119293213,
      "learning_rate": 2.2121033660649863e-05,
      "loss": 0.7281,
      "step": 8345
    },
    {
      "epoch": 2.716330513988289,
      "grad_norm": 1.559083342552185,
      "learning_rate": 2.20953985997333e-05,
      "loss": 0.7544,
      "step": 8350
    },
    {
      "epoch": 2.717957059206246,
      "grad_norm": 2.210763454437256,
      "learning_rate": 2.2069766634277174e-05,
      "loss": 0.7411,
      "step": 8355
    },
    {
      "epoch": 2.719583604424203,
      "grad_norm": 1.729476809501648,
      "learning_rate": 2.2044137791597723e-05,
      "loss": 0.7414,
      "step": 8360
    },
    {
      "epoch": 2.72121014964216,
      "grad_norm": 1.8460407257080078,
      "learning_rate": 2.2018512099007823e-05,
      "loss": 0.7203,
      "step": 8365
    },
    {
      "epoch": 2.722836694860117,
      "grad_norm": 1.8685365915298462,
      "learning_rate": 2.1992889583817023e-05,
      "loss": 0.7463,
      "step": 8370
    },
    {
      "epoch": 2.724463240078074,
      "grad_norm": 1.7120425701141357,
      "learning_rate": 2.1967270273331464e-05,
      "loss": 0.7575,
      "step": 8375
    },
    {
      "epoch": 2.7260897852960313,
      "grad_norm": 1.839551329612732,
      "learning_rate": 2.194165419485389e-05,
      "loss": 0.735,
      "step": 8380
    },
    {
      "epoch": 2.727716330513988,
      "grad_norm": 1.5357433557510376,
      "learning_rate": 2.191604137568358e-05,
      "loss": 0.7125,
      "step": 8385
    },
    {
      "epoch": 2.7293428757319456,
      "grad_norm": 1.843185544013977,
      "learning_rate": 2.1890431843116373e-05,
      "loss": 0.7247,
      "step": 8390
    },
    {
      "epoch": 2.7309694209499025,
      "grad_norm": 1.520994782447815,
      "learning_rate": 2.186482562444456e-05,
      "loss": 0.731,
      "step": 8395
    },
    {
      "epoch": 2.7325959661678594,
      "grad_norm": 1.8172752857208252,
      "learning_rate": 2.1839222746956936e-05,
      "loss": 0.7369,
      "step": 8400
    },
    {
      "epoch": 2.7342225113858163,
      "grad_norm": 2.0478515625,
      "learning_rate": 2.1813623237938714e-05,
      "loss": 0.7047,
      "step": 8405
    },
    {
      "epoch": 2.7358490566037736,
      "grad_norm": 1.963808298110962,
      "learning_rate": 2.178802712467154e-05,
      "loss": 0.735,
      "step": 8410
    },
    {
      "epoch": 2.7374756018217306,
      "grad_norm": 2.2454097270965576,
      "learning_rate": 2.1762434434433414e-05,
      "loss": 0.7352,
      "step": 8415
    },
    {
      "epoch": 2.739102147039688,
      "grad_norm": 1.9860196113586426,
      "learning_rate": 2.173684519449872e-05,
      "loss": 0.7128,
      "step": 8420
    },
    {
      "epoch": 2.740728692257645,
      "grad_norm": 1.8477674722671509,
      "learning_rate": 2.1711259432138118e-05,
      "loss": 0.7336,
      "step": 8425
    },
    {
      "epoch": 2.7423552374756017,
      "grad_norm": 1.6550543308258057,
      "learning_rate": 2.1685677174618615e-05,
      "loss": 0.7001,
      "step": 8430
    },
    {
      "epoch": 2.7439817826935586,
      "grad_norm": 1.5890171527862549,
      "learning_rate": 2.1660098449203446e-05,
      "loss": 0.7494,
      "step": 8435
    },
    {
      "epoch": 2.745608327911516,
      "grad_norm": 1.7456623315811157,
      "learning_rate": 2.1634523283152107e-05,
      "loss": 0.7574,
      "step": 8440
    },
    {
      "epoch": 2.747234873129473,
      "grad_norm": 1.7022398710250854,
      "learning_rate": 2.160895170372026e-05,
      "loss": 0.7174,
      "step": 8445
    },
    {
      "epoch": 2.7488614183474303,
      "grad_norm": 1.7472623586654663,
      "learning_rate": 2.1583383738159812e-05,
      "loss": 0.7331,
      "step": 8450
    },
    {
      "epoch": 2.750487963565387,
      "grad_norm": 1.8784606456756592,
      "learning_rate": 2.1557819413718743e-05,
      "loss": 0.7302,
      "step": 8455
    },
    {
      "epoch": 2.752114508783344,
      "grad_norm": 1.8083624839782715,
      "learning_rate": 2.1532258757641216e-05,
      "loss": 0.7273,
      "step": 8460
    },
    {
      "epoch": 2.753741054001301,
      "grad_norm": 2.1513571739196777,
      "learning_rate": 2.1506701797167435e-05,
      "loss": 0.7327,
      "step": 8465
    },
    {
      "epoch": 2.7553675992192583,
      "grad_norm": 6.521982192993164,
      "learning_rate": 2.14811485595337e-05,
      "loss": 0.7357,
      "step": 8470
    },
    {
      "epoch": 2.7569941444372152,
      "grad_norm": 1.629664659500122,
      "learning_rate": 2.1455599071972344e-05,
      "loss": 0.7231,
      "step": 8475
    },
    {
      "epoch": 2.7586206896551726,
      "grad_norm": 1.6649502515792847,
      "learning_rate": 2.143005336171166e-05,
      "loss": 0.7108,
      "step": 8480
    },
    {
      "epoch": 2.7602472348731295,
      "grad_norm": 1.8977936506271362,
      "learning_rate": 2.1404511455975977e-05,
      "loss": 0.7402,
      "step": 8485
    },
    {
      "epoch": 2.7618737800910864,
      "grad_norm": 1.6810303926467896,
      "learning_rate": 2.137897338198552e-05,
      "loss": 0.7045,
      "step": 8490
    },
    {
      "epoch": 2.7635003253090433,
      "grad_norm": 1.9456619024276733,
      "learning_rate": 2.1353439166956453e-05,
      "loss": 0.7221,
      "step": 8495
    },
    {
      "epoch": 2.7651268705270007,
      "grad_norm": 2.0193536281585693,
      "learning_rate": 2.1327908838100824e-05,
      "loss": 0.7471,
      "step": 8500
    },
    {
      "epoch": 2.7667534157449576,
      "grad_norm": 1.7074350118637085,
      "learning_rate": 2.1302382422626556e-05,
      "loss": 0.7446,
      "step": 8505
    },
    {
      "epoch": 2.768379960962915,
      "grad_norm": 1.6121134757995605,
      "learning_rate": 2.1276859947737356e-05,
      "loss": 0.7501,
      "step": 8510
    },
    {
      "epoch": 2.770006506180872,
      "grad_norm": 1.8085863590240479,
      "learning_rate": 2.1251341440632783e-05,
      "loss": 0.7112,
      "step": 8515
    },
    {
      "epoch": 2.7716330513988288,
      "grad_norm": 1.7475433349609375,
      "learning_rate": 2.122582692850813e-05,
      "loss": 0.753,
      "step": 8520
    },
    {
      "epoch": 2.7732595966167857,
      "grad_norm": 1.9420067071914673,
      "learning_rate": 2.120031643855446e-05,
      "loss": 0.7381,
      "step": 8525
    },
    {
      "epoch": 2.774886141834743,
      "grad_norm": 1.8206048011779785,
      "learning_rate": 2.117480999795853e-05,
      "loss": 0.7378,
      "step": 8530
    },
    {
      "epoch": 2.7765126870527,
      "grad_norm": 1.6537567377090454,
      "learning_rate": 2.114930763390279e-05,
      "loss": 0.7196,
      "step": 8535
    },
    {
      "epoch": 2.7781392322706573,
      "grad_norm": 1.623516321182251,
      "learning_rate": 2.1123809373565342e-05,
      "loss": 0.7136,
      "step": 8540
    },
    {
      "epoch": 2.779765777488614,
      "grad_norm": 2.062002182006836,
      "learning_rate": 2.1098315244119922e-05,
      "loss": 0.7179,
      "step": 8545
    },
    {
      "epoch": 2.781392322706571,
      "grad_norm": 1.7511444091796875,
      "learning_rate": 2.1072825272735848e-05,
      "loss": 0.7189,
      "step": 8550
    },
    {
      "epoch": 2.7830188679245285,
      "grad_norm": 1.8162951469421387,
      "learning_rate": 2.104733948657803e-05,
      "loss": 0.714,
      "step": 8555
    },
    {
      "epoch": 2.7846454131424854,
      "grad_norm": 1.6305124759674072,
      "learning_rate": 2.102185791280689e-05,
      "loss": 0.7332,
      "step": 8560
    },
    {
      "epoch": 2.7862719583604423,
      "grad_norm": 1.8210047483444214,
      "learning_rate": 2.09963805785784e-05,
      "loss": 0.7362,
      "step": 8565
    },
    {
      "epoch": 2.7878985035783996,
      "grad_norm": 1.965907096862793,
      "learning_rate": 2.0970907511043964e-05,
      "loss": 0.7378,
      "step": 8570
    },
    {
      "epoch": 2.7895250487963565,
      "grad_norm": 1.8362945318222046,
      "learning_rate": 2.0945438737350476e-05,
      "loss": 0.7399,
      "step": 8575
    },
    {
      "epoch": 2.7911515940143135,
      "grad_norm": 1.7062568664550781,
      "learning_rate": 2.0919974284640237e-05,
      "loss": 0.7436,
      "step": 8580
    },
    {
      "epoch": 2.792778139232271,
      "grad_norm": 1.911207914352417,
      "learning_rate": 2.0894514180050964e-05,
      "loss": 0.7419,
      "step": 8585
    },
    {
      "epoch": 2.7944046844502277,
      "grad_norm": 1.878778338432312,
      "learning_rate": 2.0869058450715696e-05,
      "loss": 0.7517,
      "step": 8590
    },
    {
      "epoch": 2.7960312296681846,
      "grad_norm": 1.5588253736495972,
      "learning_rate": 2.0843607123762875e-05,
      "loss": 0.7775,
      "step": 8595
    },
    {
      "epoch": 2.797657774886142,
      "grad_norm": 1.6011550426483154,
      "learning_rate": 2.081816022631618e-05,
      "loss": 0.7463,
      "step": 8600
    },
    {
      "epoch": 2.799284320104099,
      "grad_norm": 1.6946686506271362,
      "learning_rate": 2.079271778549462e-05,
      "loss": 0.7052,
      "step": 8605
    },
    {
      "epoch": 2.800910865322056,
      "grad_norm": 1.7881314754486084,
      "learning_rate": 2.0767279828412442e-05,
      "loss": 0.7566,
      "step": 8610
    },
    {
      "epoch": 2.802537410540013,
      "grad_norm": 1.9608092308044434,
      "learning_rate": 2.0741846382179102e-05,
      "loss": 0.7481,
      "step": 8615
    },
    {
      "epoch": 2.80416395575797,
      "grad_norm": 1.8728355169296265,
      "learning_rate": 2.0716417473899268e-05,
      "loss": 0.7619,
      "step": 8620
    },
    {
      "epoch": 2.805790500975927,
      "grad_norm": 1.7436429262161255,
      "learning_rate": 2.069099313067275e-05,
      "loss": 0.6805,
      "step": 8625
    },
    {
      "epoch": 2.8074170461938843,
      "grad_norm": 2.551387071609497,
      "learning_rate": 2.0665573379594516e-05,
      "loss": 0.7302,
      "step": 8630
    },
    {
      "epoch": 2.8090435914118412,
      "grad_norm": 1.6762300729751587,
      "learning_rate": 2.0640158247754614e-05,
      "loss": 0.7121,
      "step": 8635
    },
    {
      "epoch": 2.810670136629798,
      "grad_norm": 1.5577950477600098,
      "learning_rate": 2.06147477622382e-05,
      "loss": 0.7586,
      "step": 8640
    },
    {
      "epoch": 2.8122966818477555,
      "grad_norm": 1.7521088123321533,
      "learning_rate": 2.0589341950125444e-05,
      "loss": 0.7032,
      "step": 8645
    },
    {
      "epoch": 2.8139232270657124,
      "grad_norm": 1.9056010246276855,
      "learning_rate": 2.056394083849158e-05,
      "loss": 0.7133,
      "step": 8650
    },
    {
      "epoch": 2.8155497722836698,
      "grad_norm": 1.8002731800079346,
      "learning_rate": 2.0538544454406776e-05,
      "loss": 0.7489,
      "step": 8655
    },
    {
      "epoch": 2.8171763175016267,
      "grad_norm": 1.6665791273117065,
      "learning_rate": 2.051315282493622e-05,
      "loss": 0.7431,
      "step": 8660
    },
    {
      "epoch": 2.8188028627195836,
      "grad_norm": 1.627205729484558,
      "learning_rate": 2.048776597713998e-05,
      "loss": 0.7262,
      "step": 8665
    },
    {
      "epoch": 2.8204294079375405,
      "grad_norm": 2.089062452316284,
      "learning_rate": 2.0462383938073078e-05,
      "loss": 0.7246,
      "step": 8670
    },
    {
      "epoch": 2.822055953155498,
      "grad_norm": 1.521073341369629,
      "learning_rate": 2.0437006734785365e-05,
      "loss": 0.7207,
      "step": 8675
    },
    {
      "epoch": 2.8236824983734548,
      "grad_norm": 2.04318904876709,
      "learning_rate": 2.0411634394321578e-05,
      "loss": 0.7386,
      "step": 8680
    },
    {
      "epoch": 2.825309043591412,
      "grad_norm": 1.7933502197265625,
      "learning_rate": 2.0386266943721235e-05,
      "loss": 0.7441,
      "step": 8685
    },
    {
      "epoch": 2.826935588809369,
      "grad_norm": 1.615450143814087,
      "learning_rate": 2.0360904410018676e-05,
      "loss": 0.7017,
      "step": 8690
    },
    {
      "epoch": 2.828562134027326,
      "grad_norm": 2.150987386703491,
      "learning_rate": 2.033554682024298e-05,
      "loss": 0.7638,
      "step": 8695
    },
    {
      "epoch": 2.830188679245283,
      "grad_norm": 2.078624963760376,
      "learning_rate": 2.031019420141797e-05,
      "loss": 0.7263,
      "step": 8700
    },
    {
      "epoch": 2.83181522446324,
      "grad_norm": 1.6702829599380493,
      "learning_rate": 2.028484658056216e-05,
      "loss": 0.7257,
      "step": 8705
    },
    {
      "epoch": 2.833441769681197,
      "grad_norm": 1.7530449628829956,
      "learning_rate": 2.025950398468875e-05,
      "loss": 0.7051,
      "step": 8710
    },
    {
      "epoch": 2.8350683148991545,
      "grad_norm": 2.530587673187256,
      "learning_rate": 2.0234166440805562e-05,
      "loss": 0.7341,
      "step": 8715
    },
    {
      "epoch": 2.8366948601171114,
      "grad_norm": 1.8676536083221436,
      "learning_rate": 2.020883397591507e-05,
      "loss": 0.7295,
      "step": 8720
    },
    {
      "epoch": 2.8383214053350683,
      "grad_norm": 1.7714043855667114,
      "learning_rate": 2.0183506617014293e-05,
      "loss": 0.7247,
      "step": 8725
    },
    {
      "epoch": 2.839947950553025,
      "grad_norm": 1.6985597610473633,
      "learning_rate": 2.015818439109485e-05,
      "loss": 0.6978,
      "step": 8730
    },
    {
      "epoch": 2.8415744957709825,
      "grad_norm": 1.7291275262832642,
      "learning_rate": 2.0132867325142858e-05,
      "loss": 0.7155,
      "step": 8735
    },
    {
      "epoch": 2.8432010409889394,
      "grad_norm": 1.6741697788238525,
      "learning_rate": 2.010755544613895e-05,
      "loss": 0.7253,
      "step": 8740
    },
    {
      "epoch": 2.844827586206897,
      "grad_norm": 1.7950025796890259,
      "learning_rate": 2.008224878105824e-05,
      "loss": 0.7178,
      "step": 8745
    },
    {
      "epoch": 2.8464541314248537,
      "grad_norm": 1.9313833713531494,
      "learning_rate": 2.005694735687025e-05,
      "loss": 0.7128,
      "step": 8750
    },
    {
      "epoch": 2.8480806766428106,
      "grad_norm": 1.7579280138015747,
      "learning_rate": 2.0031651200538963e-05,
      "loss": 0.6925,
      "step": 8755
    },
    {
      "epoch": 2.8497072218607675,
      "grad_norm": 1.711421251296997,
      "learning_rate": 2.000636033902271e-05,
      "loss": 0.7479,
      "step": 8760
    },
    {
      "epoch": 2.851333767078725,
      "grad_norm": 1.752677321434021,
      "learning_rate": 1.998107479927421e-05,
      "loss": 0.7246,
      "step": 8765
    },
    {
      "epoch": 2.852960312296682,
      "grad_norm": 2.1234564781188965,
      "learning_rate": 1.995579460824048e-05,
      "loss": 0.7231,
      "step": 8770
    },
    {
      "epoch": 2.854586857514639,
      "grad_norm": 1.6792376041412354,
      "learning_rate": 1.9930519792862873e-05,
      "loss": 0.6929,
      "step": 8775
    },
    {
      "epoch": 2.856213402732596,
      "grad_norm": 1.910710096359253,
      "learning_rate": 1.9905250380076965e-05,
      "loss": 0.7176,
      "step": 8780
    },
    {
      "epoch": 2.857839947950553,
      "grad_norm": 1.9245860576629639,
      "learning_rate": 1.9879986396812623e-05,
      "loss": 0.7285,
      "step": 8785
    },
    {
      "epoch": 2.85946649316851,
      "grad_norm": 1.8107945919036865,
      "learning_rate": 1.985472786999389e-05,
      "loss": 0.7178,
      "step": 8790
    },
    {
      "epoch": 2.8610930383864672,
      "grad_norm": 1.5665243864059448,
      "learning_rate": 1.982947482653903e-05,
      "loss": 0.7098,
      "step": 8795
    },
    {
      "epoch": 2.862719583604424,
      "grad_norm": 1.9117822647094727,
      "learning_rate": 1.980422729336042e-05,
      "loss": 0.7184,
      "step": 8800
    },
    {
      "epoch": 2.8643461288223815,
      "grad_norm": 2.126077651977539,
      "learning_rate": 1.977898529736462e-05,
      "loss": 0.7414,
      "step": 8805
    },
    {
      "epoch": 2.8659726740403384,
      "grad_norm": 1.8685696125030518,
      "learning_rate": 1.9753748865452226e-05,
      "loss": 0.7074,
      "step": 8810
    },
    {
      "epoch": 2.8675992192582953,
      "grad_norm": 1.9692431688308716,
      "learning_rate": 1.9728518024517957e-05,
      "loss": 0.6982,
      "step": 8815
    },
    {
      "epoch": 2.869225764476252,
      "grad_norm": 1.534635066986084,
      "learning_rate": 1.9703292801450536e-05,
      "loss": 0.7153,
      "step": 8820
    },
    {
      "epoch": 2.8708523096942096,
      "grad_norm": 1.5757933855056763,
      "learning_rate": 1.9678073223132737e-05,
      "loss": 0.7457,
      "step": 8825
    },
    {
      "epoch": 2.8724788549121665,
      "grad_norm": 1.720250129699707,
      "learning_rate": 1.9652859316441267e-05,
      "loss": 0.7032,
      "step": 8830
    },
    {
      "epoch": 2.874105400130124,
      "grad_norm": 1.6449079513549805,
      "learning_rate": 1.9627651108246848e-05,
      "loss": 0.7061,
      "step": 8835
    },
    {
      "epoch": 2.8757319453480807,
      "grad_norm": 1.6099705696105957,
      "learning_rate": 1.9602448625414077e-05,
      "loss": 0.693,
      "step": 8840
    },
    {
      "epoch": 2.8773584905660377,
      "grad_norm": 1.9806228876113892,
      "learning_rate": 1.9577251894801488e-05,
      "loss": 0.7428,
      "step": 8845
    },
    {
      "epoch": 2.8789850357839946,
      "grad_norm": 1.7147135734558105,
      "learning_rate": 1.9552060943261456e-05,
      "loss": 0.7634,
      "step": 8850
    },
    {
      "epoch": 2.880611581001952,
      "grad_norm": 1.7568204402923584,
      "learning_rate": 1.9526875797640226e-05,
      "loss": 0.7062,
      "step": 8855
    },
    {
      "epoch": 2.882238126219909,
      "grad_norm": 1.8958820104599,
      "learning_rate": 1.950169648477782e-05,
      "loss": 0.7187,
      "step": 8860
    },
    {
      "epoch": 2.883864671437866,
      "grad_norm": 1.7266991138458252,
      "learning_rate": 1.947652303150808e-05,
      "loss": 0.7518,
      "step": 8865
    },
    {
      "epoch": 2.885491216655823,
      "grad_norm": 1.6969014406204224,
      "learning_rate": 1.945135546465857e-05,
      "loss": 0.7407,
      "step": 8870
    },
    {
      "epoch": 2.88711776187378,
      "grad_norm": 1.9077585935592651,
      "learning_rate": 1.942619381105061e-05,
      "loss": 0.7169,
      "step": 8875
    },
    {
      "epoch": 2.888744307091737,
      "grad_norm": 1.9687460660934448,
      "learning_rate": 1.9401038097499208e-05,
      "loss": 0.731,
      "step": 8880
    },
    {
      "epoch": 2.8903708523096943,
      "grad_norm": 2.1020829677581787,
      "learning_rate": 1.9375888350813026e-05,
      "loss": 0.7104,
      "step": 8885
    },
    {
      "epoch": 2.891997397527651,
      "grad_norm": 2.6251018047332764,
      "learning_rate": 1.9350744597794405e-05,
      "loss": 0.6999,
      "step": 8890
    },
    {
      "epoch": 2.8936239427456085,
      "grad_norm": 2.1141345500946045,
      "learning_rate": 1.9325606865239243e-05,
      "loss": 0.7493,
      "step": 8895
    },
    {
      "epoch": 2.8952504879635654,
      "grad_norm": 2.002952814102173,
      "learning_rate": 1.9300475179937077e-05,
      "loss": 0.714,
      "step": 8900
    },
    {
      "epoch": 2.8968770331815223,
      "grad_norm": 2.4000797271728516,
      "learning_rate": 1.9275349568670957e-05,
      "loss": 0.7104,
      "step": 8905
    },
    {
      "epoch": 2.8985035783994793,
      "grad_norm": 1.9128988981246948,
      "learning_rate": 1.9250230058217496e-05,
      "loss": 0.7081,
      "step": 8910
    },
    {
      "epoch": 2.9001301236174366,
      "grad_norm": 1.6900957822799683,
      "learning_rate": 1.9225116675346776e-05,
      "loss": 0.7381,
      "step": 8915
    },
    {
      "epoch": 2.9017566688353935,
      "grad_norm": 1.6366512775421143,
      "learning_rate": 1.920000944682237e-05,
      "loss": 0.6969,
      "step": 8920
    },
    {
      "epoch": 2.903383214053351,
      "grad_norm": 1.9840657711029053,
      "learning_rate": 1.9174908399401266e-05,
      "loss": 0.7298,
      "step": 8925
    },
    {
      "epoch": 2.905009759271308,
      "grad_norm": 1.681493878364563,
      "learning_rate": 1.9149813559833897e-05,
      "loss": 0.7284,
      "step": 8930
    },
    {
      "epoch": 2.9066363044892647,
      "grad_norm": 1.9548931121826172,
      "learning_rate": 1.912472495486405e-05,
      "loss": 0.7197,
      "step": 8935
    },
    {
      "epoch": 2.9082628497072216,
      "grad_norm": 1.92655611038208,
      "learning_rate": 1.9099642611228896e-05,
      "loss": 0.7528,
      "step": 8940
    },
    {
      "epoch": 2.909889394925179,
      "grad_norm": 1.547681212425232,
      "learning_rate": 1.907456655565891e-05,
      "loss": 0.7207,
      "step": 8945
    },
    {
      "epoch": 2.911515940143136,
      "grad_norm": 1.506739616394043,
      "learning_rate": 1.9049496814877893e-05,
      "loss": 0.6939,
      "step": 8950
    },
    {
      "epoch": 2.913142485361093,
      "grad_norm": 1.6843504905700684,
      "learning_rate": 1.9024433415602872e-05,
      "loss": 0.7461,
      "step": 8955
    },
    {
      "epoch": 2.91476903057905,
      "grad_norm": 1.856924295425415,
      "learning_rate": 1.899937638454416e-05,
      "loss": 0.7348,
      "step": 8960
    },
    {
      "epoch": 2.916395575797007,
      "grad_norm": 2.313263177871704,
      "learning_rate": 1.8974325748405258e-05,
      "loss": 0.7271,
      "step": 8965
    },
    {
      "epoch": 2.918022121014964,
      "grad_norm": 1.6064339876174927,
      "learning_rate": 1.8949281533882864e-05,
      "loss": 0.7281,
      "step": 8970
    },
    {
      "epoch": 2.9196486662329213,
      "grad_norm": 2.2871596813201904,
      "learning_rate": 1.8924243767666823e-05,
      "loss": 0.745,
      "step": 8975
    },
    {
      "epoch": 2.921275211450878,
      "grad_norm": 1.5119794607162476,
      "learning_rate": 1.8899212476440125e-05,
      "loss": 0.7236,
      "step": 8980
    },
    {
      "epoch": 2.9229017566688356,
      "grad_norm": 1.7199755907058716,
      "learning_rate": 1.887418768687883e-05,
      "loss": 0.7452,
      "step": 8985
    },
    {
      "epoch": 2.9245283018867925,
      "grad_norm": 1.876117467880249,
      "learning_rate": 1.8849169425652095e-05,
      "loss": 0.7237,
      "step": 8990
    },
    {
      "epoch": 2.9261548471047494,
      "grad_norm": 1.9358798265457153,
      "learning_rate": 1.8824157719422112e-05,
      "loss": 0.7311,
      "step": 8995
    },
    {
      "epoch": 2.9277813923227067,
      "grad_norm": 1.8423490524291992,
      "learning_rate": 1.8799152594844093e-05,
      "loss": 0.7247,
      "step": 9000
    },
    {
      "epoch": 2.9294079375406636,
      "grad_norm": 3.251728057861328,
      "learning_rate": 1.8774154078566207e-05,
      "loss": 0.7415,
      "step": 9005
    },
    {
      "epoch": 2.9310344827586206,
      "grad_norm": 1.8368016481399536,
      "learning_rate": 1.8749162197229626e-05,
      "loss": 0.7178,
      "step": 9010
    },
    {
      "epoch": 2.932661027976578,
      "grad_norm": 1.7899824380874634,
      "learning_rate": 1.872417697746843e-05,
      "loss": 0.7046,
      "step": 9015
    },
    {
      "epoch": 2.934287573194535,
      "grad_norm": 1.997498869895935,
      "learning_rate": 1.8699198445909572e-05,
      "loss": 0.7239,
      "step": 9020
    },
    {
      "epoch": 2.9359141184124917,
      "grad_norm": 1.5661916732788086,
      "learning_rate": 1.8674226629172925e-05,
      "loss": 0.7231,
      "step": 9025
    },
    {
      "epoch": 2.937540663630449,
      "grad_norm": 1.6990458965301514,
      "learning_rate": 1.8649261553871176e-05,
      "loss": 0.7206,
      "step": 9030
    },
    {
      "epoch": 2.939167208848406,
      "grad_norm": 1.7623580694198608,
      "learning_rate": 1.8624303246609847e-05,
      "loss": 0.7218,
      "step": 9035
    },
    {
      "epoch": 2.940793754066363,
      "grad_norm": 1.6274583339691162,
      "learning_rate": 1.8599351733987205e-05,
      "loss": 0.7362,
      "step": 9040
    },
    {
      "epoch": 2.9424202992843203,
      "grad_norm": 2.1584455966949463,
      "learning_rate": 1.8574407042594348e-05,
      "loss": 0.7171,
      "step": 9045
    },
    {
      "epoch": 2.944046844502277,
      "grad_norm": 1.5133984088897705,
      "learning_rate": 1.854946919901503e-05,
      "loss": 0.7399,
      "step": 9050
    },
    {
      "epoch": 2.945673389720234,
      "grad_norm": 1.7558419704437256,
      "learning_rate": 1.8524538229825757e-05,
      "loss": 0.7382,
      "step": 9055
    },
    {
      "epoch": 2.9472999349381914,
      "grad_norm": 1.7353185415267944,
      "learning_rate": 1.8499614161595685e-05,
      "loss": 0.7105,
      "step": 9060
    },
    {
      "epoch": 2.9489264801561483,
      "grad_norm": 2.38606595993042,
      "learning_rate": 1.8474697020886636e-05,
      "loss": 0.7605,
      "step": 9065
    },
    {
      "epoch": 2.9505530253741052,
      "grad_norm": 1.7944279909133911,
      "learning_rate": 1.8449786834253015e-05,
      "loss": 0.715,
      "step": 9070
    },
    {
      "epoch": 2.9521795705920626,
      "grad_norm": 1.5300099849700928,
      "learning_rate": 1.8424883628241857e-05,
      "loss": 0.7015,
      "step": 9075
    },
    {
      "epoch": 2.9538061158100195,
      "grad_norm": 1.6979626417160034,
      "learning_rate": 1.8399987429392722e-05,
      "loss": 0.7271,
      "step": 9080
    },
    {
      "epoch": 2.9554326610279764,
      "grad_norm": 1.8683507442474365,
      "learning_rate": 1.837509826423773e-05,
      "loss": 0.687,
      "step": 9085
    },
    {
      "epoch": 2.9570592062459338,
      "grad_norm": 1.8588496446609497,
      "learning_rate": 1.8350216159301483e-05,
      "loss": 0.7267,
      "step": 9090
    },
    {
      "epoch": 2.9586857514638907,
      "grad_norm": 1.57637619972229,
      "learning_rate": 1.8325341141101087e-05,
      "loss": 0.707,
      "step": 9095
    },
    {
      "epoch": 2.960312296681848,
      "grad_norm": 1.714156985282898,
      "learning_rate": 1.830047323614606e-05,
      "loss": 0.7032,
      "step": 9100
    },
    {
      "epoch": 2.961938841899805,
      "grad_norm": 1.7165101766586304,
      "learning_rate": 1.827561247093836e-05,
      "loss": 0.6934,
      "step": 9105
    },
    {
      "epoch": 2.963565387117762,
      "grad_norm": 1.688585877418518,
      "learning_rate": 1.8250758871972335e-05,
      "loss": 0.7076,
      "step": 9110
    },
    {
      "epoch": 2.9651919323357188,
      "grad_norm": 1.927809476852417,
      "learning_rate": 1.82259124657347e-05,
      "loss": 0.7257,
      "step": 9115
    },
    {
      "epoch": 2.966818477553676,
      "grad_norm": 1.6875810623168945,
      "learning_rate": 1.8201073278704492e-05,
      "loss": 0.7341,
      "step": 9120
    },
    {
      "epoch": 2.968445022771633,
      "grad_norm": 1.7902840375900269,
      "learning_rate": 1.8176241337353073e-05,
      "loss": 0.7162,
      "step": 9125
    },
    {
      "epoch": 2.9700715679895904,
      "grad_norm": 1.5761710405349731,
      "learning_rate": 1.815141666814405e-05,
      "loss": 0.7042,
      "step": 9130
    },
    {
      "epoch": 2.9716981132075473,
      "grad_norm": 1.8276005983352661,
      "learning_rate": 1.812659929753332e-05,
      "loss": 0.7388,
      "step": 9135
    },
    {
      "epoch": 2.973324658425504,
      "grad_norm": 1.7913482189178467,
      "learning_rate": 1.810178925196897e-05,
      "loss": 0.7465,
      "step": 9140
    },
    {
      "epoch": 2.974951203643461,
      "grad_norm": 1.6558361053466797,
      "learning_rate": 1.80769865578913e-05,
      "loss": 0.7269,
      "step": 9145
    },
    {
      "epoch": 2.9765777488614185,
      "grad_norm": 1.754176378250122,
      "learning_rate": 1.805219124173278e-05,
      "loss": 0.7365,
      "step": 9150
    },
    {
      "epoch": 2.9782042940793754,
      "grad_norm": 1.735854148864746,
      "learning_rate": 1.802740332991799e-05,
      "loss": 0.7312,
      "step": 9155
    },
    {
      "epoch": 2.9798308392973327,
      "grad_norm": 1.854060173034668,
      "learning_rate": 1.800262284886365e-05,
      "loss": 0.6995,
      "step": 9160
    },
    {
      "epoch": 2.9814573845152896,
      "grad_norm": 1.8062974214553833,
      "learning_rate": 1.797784982497853e-05,
      "loss": 0.7316,
      "step": 9165
    },
    {
      "epoch": 2.9830839297332465,
      "grad_norm": 1.7298572063446045,
      "learning_rate": 1.7953084284663486e-05,
      "loss": 0.7445,
      "step": 9170
    },
    {
      "epoch": 2.9847104749512035,
      "grad_norm": 1.7473677396774292,
      "learning_rate": 1.7928326254311363e-05,
      "loss": 0.7474,
      "step": 9175
    },
    {
      "epoch": 2.986337020169161,
      "grad_norm": 1.807447910308838,
      "learning_rate": 1.7903575760307044e-05,
      "loss": 0.6988,
      "step": 9180
    },
    {
      "epoch": 2.9879635653871177,
      "grad_norm": 2.072762966156006,
      "learning_rate": 1.787883282902734e-05,
      "loss": 0.7396,
      "step": 9185
    },
    {
      "epoch": 2.989590110605075,
      "grad_norm": 1.9413923025131226,
      "learning_rate": 1.7854097486841044e-05,
      "loss": 0.759,
      "step": 9190
    },
    {
      "epoch": 2.991216655823032,
      "grad_norm": 1.696505069732666,
      "learning_rate": 1.782936976010881e-05,
      "loss": 0.7233,
      "step": 9195
    },
    {
      "epoch": 2.992843201040989,
      "grad_norm": 1.5802931785583496,
      "learning_rate": 1.7804649675183223e-05,
      "loss": 0.7215,
      "step": 9200
    },
    {
      "epoch": 2.994469746258946,
      "grad_norm": 1.557307243347168,
      "learning_rate": 1.7779937258408685e-05,
      "loss": 0.7219,
      "step": 9205
    },
    {
      "epoch": 2.996096291476903,
      "grad_norm": 1.7299686670303345,
      "learning_rate": 1.7755232536121477e-05,
      "loss": 0.6935,
      "step": 9210
    },
    {
      "epoch": 2.99772283669486,
      "grad_norm": 1.9171065092086792,
      "learning_rate": 1.7730535534649614e-05,
      "loss": 0.7665,
      "step": 9215
    },
    {
      "epoch": 2.9993493819128174,
      "grad_norm": 1.767626166343689,
      "learning_rate": 1.7705846280312948e-05,
      "loss": 0.687,
      "step": 9220
    },
    {
      "epoch": 3.0,
      "eval_f1": 0.824066158814795,
      "eval_loss": 0.403564453125,
      "eval_precision": 0.8248737220377281,
      "eval_recall": 0.8234550337582588,
      "eval_runtime": 1028.979,
      "eval_samples_per_second": 382.355,
      "eval_steps_per_second": 0.747,
      "step": 9222
    },
    {
      "epoch": 3.0009759271307743,
      "grad_norm": 1.4708726406097412,
      "learning_rate": 1.768116479942303e-05,
      "loss": 0.6647,
      "step": 9225
    },
    {
      "epoch": 3.0026024723487312,
      "grad_norm": 1.9929927587509155,
      "learning_rate": 1.7656491118283135e-05,
      "loss": 0.6565,
      "step": 9230
    },
    {
      "epoch": 3.004229017566688,
      "grad_norm": 1.7582905292510986,
      "learning_rate": 1.7631825263188246e-05,
      "loss": 0.6723,
      "step": 9235
    },
    {
      "epoch": 3.0058555627846455,
      "grad_norm": 1.9024964570999146,
      "learning_rate": 1.760716726042499e-05,
      "loss": 0.6528,
      "step": 9240
    },
    {
      "epoch": 3.0074821080026024,
      "grad_norm": 1.9164223670959473,
      "learning_rate": 1.7582517136271616e-05,
      "loss": 0.647,
      "step": 9245
    },
    {
      "epoch": 3.0091086532205593,
      "grad_norm": 1.8600401878356934,
      "learning_rate": 1.7557874916997996e-05,
      "loss": 0.6601,
      "step": 9250
    },
    {
      "epoch": 3.0107351984385167,
      "grad_norm": 1.8095484972000122,
      "learning_rate": 1.7533240628865567e-05,
      "loss": 0.6627,
      "step": 9255
    },
    {
      "epoch": 3.0123617436564736,
      "grad_norm": 2.208045244216919,
      "learning_rate": 1.7508614298127322e-05,
      "loss": 0.6739,
      "step": 9260
    },
    {
      "epoch": 3.0139882888744305,
      "grad_norm": 2.595229387283325,
      "learning_rate": 1.7483995951027767e-05,
      "loss": 0.6648,
      "step": 9265
    },
    {
      "epoch": 3.015614834092388,
      "grad_norm": 1.7185879945755005,
      "learning_rate": 1.7459385613802903e-05,
      "loss": 0.6742,
      "step": 9270
    },
    {
      "epoch": 3.0172413793103448,
      "grad_norm": 1.649437665939331,
      "learning_rate": 1.743478331268018e-05,
      "loss": 0.6808,
      "step": 9275
    },
    {
      "epoch": 3.018867924528302,
      "grad_norm": 2.5946407318115234,
      "learning_rate": 1.7410189073878513e-05,
      "loss": 0.6392,
      "step": 9280
    },
    {
      "epoch": 3.020494469746259,
      "grad_norm": 1.7240699529647827,
      "learning_rate": 1.7385602923608192e-05,
      "loss": 0.6396,
      "step": 9285
    },
    {
      "epoch": 3.022121014964216,
      "grad_norm": 2.009359359741211,
      "learning_rate": 1.736102488807092e-05,
      "loss": 0.6712,
      "step": 9290
    },
    {
      "epoch": 3.0237475601821733,
      "grad_norm": 1.9779503345489502,
      "learning_rate": 1.7336454993459726e-05,
      "loss": 0.6742,
      "step": 9295
    },
    {
      "epoch": 3.02537410540013,
      "grad_norm": 1.8295644521713257,
      "learning_rate": 1.7311893265958974e-05,
      "loss": 0.6896,
      "step": 9300
    },
    {
      "epoch": 3.027000650618087,
      "grad_norm": 1.862802267074585,
      "learning_rate": 1.7287339731744336e-05,
      "loss": 0.646,
      "step": 9305
    },
    {
      "epoch": 3.0286271958360445,
      "grad_norm": 2.442687749862671,
      "learning_rate": 1.7262794416982716e-05,
      "loss": 0.673,
      "step": 9310
    },
    {
      "epoch": 3.0302537410540014,
      "grad_norm": 1.9544475078582764,
      "learning_rate": 1.72382573478323e-05,
      "loss": 0.6643,
      "step": 9315
    },
    {
      "epoch": 3.0318802862719583,
      "grad_norm": 2.301948070526123,
      "learning_rate": 1.721372855044246e-05,
      "loss": 0.6631,
      "step": 9320
    },
    {
      "epoch": 3.0335068314899156,
      "grad_norm": 1.9081635475158691,
      "learning_rate": 1.7189208050953765e-05,
      "loss": 0.6859,
      "step": 9325
    },
    {
      "epoch": 3.0351333767078725,
      "grad_norm": 1.8731703758239746,
      "learning_rate": 1.7164695875497928e-05,
      "loss": 0.6686,
      "step": 9330
    },
    {
      "epoch": 3.0367599219258294,
      "grad_norm": 2.1195106506347656,
      "learning_rate": 1.714019205019782e-05,
      "loss": 0.6526,
      "step": 9335
    },
    {
      "epoch": 3.038386467143787,
      "grad_norm": 2.022167444229126,
      "learning_rate": 1.711569660116737e-05,
      "loss": 0.664,
      "step": 9340
    },
    {
      "epoch": 3.0400130123617437,
      "grad_norm": 1.7911338806152344,
      "learning_rate": 1.709120955451162e-05,
      "loss": 0.6774,
      "step": 9345
    },
    {
      "epoch": 3.0416395575797006,
      "grad_norm": 2.0143144130706787,
      "learning_rate": 1.706673093632663e-05,
      "loss": 0.691,
      "step": 9350
    },
    {
      "epoch": 3.043266102797658,
      "grad_norm": 1.735304594039917,
      "learning_rate": 1.70422607726995e-05,
      "loss": 0.6552,
      "step": 9355
    },
    {
      "epoch": 3.044892648015615,
      "grad_norm": 1.8288079500198364,
      "learning_rate": 1.7017799089708293e-05,
      "loss": 0.6673,
      "step": 9360
    },
    {
      "epoch": 3.046519193233572,
      "grad_norm": 2.4522781372070312,
      "learning_rate": 1.699334591342207e-05,
      "loss": 0.6906,
      "step": 9365
    },
    {
      "epoch": 3.048145738451529,
      "grad_norm": 1.8423187732696533,
      "learning_rate": 1.696890126990079e-05,
      "loss": 0.6622,
      "step": 9370
    },
    {
      "epoch": 3.049772283669486,
      "grad_norm": 2.284074544906616,
      "learning_rate": 1.694446518519534e-05,
      "loss": 0.7015,
      "step": 9375
    },
    {
      "epoch": 3.051398828887443,
      "grad_norm": 1.8895907402038574,
      "learning_rate": 1.692003768534747e-05,
      "loss": 0.6525,
      "step": 9380
    },
    {
      "epoch": 3.0530253741054003,
      "grad_norm": 2.285318613052368,
      "learning_rate": 1.689561879638982e-05,
      "loss": 0.6873,
      "step": 9385
    },
    {
      "epoch": 3.0546519193233572,
      "grad_norm": 1.9741708040237427,
      "learning_rate": 1.687120854434579e-05,
      "loss": 0.6642,
      "step": 9390
    },
    {
      "epoch": 3.056278464541314,
      "grad_norm": 1.918621301651001,
      "learning_rate": 1.684680695522964e-05,
      "loss": 0.6556,
      "step": 9395
    },
    {
      "epoch": 3.0579050097592715,
      "grad_norm": 1.9934749603271484,
      "learning_rate": 1.682241405504634e-05,
      "loss": 0.6549,
      "step": 9400
    },
    {
      "epoch": 3.0595315549772284,
      "grad_norm": 1.994615912437439,
      "learning_rate": 1.679802986979165e-05,
      "loss": 0.6503,
      "step": 9405
    },
    {
      "epoch": 3.0611581001951853,
      "grad_norm": 1.8361835479736328,
      "learning_rate": 1.6773654425452007e-05,
      "loss": 0.6581,
      "step": 9410
    },
    {
      "epoch": 3.0627846454131427,
      "grad_norm": 1.9985431432724,
      "learning_rate": 1.6749287748004567e-05,
      "loss": 0.6606,
      "step": 9415
    },
    {
      "epoch": 3.0644111906310996,
      "grad_norm": 1.8159586191177368,
      "learning_rate": 1.6724929863417094e-05,
      "loss": 0.6215,
      "step": 9420
    },
    {
      "epoch": 3.0660377358490565,
      "grad_norm": 1.887814998626709,
      "learning_rate": 1.670058079764802e-05,
      "loss": 0.6597,
      "step": 9425
    },
    {
      "epoch": 3.067664281067014,
      "grad_norm": 2.240401029586792,
      "learning_rate": 1.6676240576646387e-05,
      "loss": 0.6489,
      "step": 9430
    },
    {
      "epoch": 3.0692908262849707,
      "grad_norm": 1.9252859354019165,
      "learning_rate": 1.665190922635177e-05,
      "loss": 0.6683,
      "step": 9435
    },
    {
      "epoch": 3.0709173715029277,
      "grad_norm": 1.790296196937561,
      "learning_rate": 1.662758677269432e-05,
      "loss": 0.6634,
      "step": 9440
    },
    {
      "epoch": 3.072543916720885,
      "grad_norm": 1.9673635959625244,
      "learning_rate": 1.66032732415947e-05,
      "loss": 0.6765,
      "step": 9445
    },
    {
      "epoch": 3.074170461938842,
      "grad_norm": 1.8681625127792358,
      "learning_rate": 1.658382885836926e-05,
      "loss": 0.6526,
      "step": 9450
    },
    {
      "epoch": 3.075797007156799,
      "grad_norm": 1.5958857536315918,
      "learning_rate": 1.655953145316344e-05,
      "loss": 0.6675,
      "step": 9455
    },
    {
      "epoch": 3.077423552374756,
      "grad_norm": 2.2521891593933105,
      "learning_rate": 1.6535243043042624e-05,
      "loss": 0.6506,
      "step": 9460
    },
    {
      "epoch": 3.079050097592713,
      "grad_norm": 2.3123247623443604,
      "learning_rate": 1.651096365389121e-05,
      "loss": 0.63,
      "step": 9465
    },
    {
      "epoch": 3.08067664281067,
      "grad_norm": 1.9048078060150146,
      "learning_rate": 1.6486693311583946e-05,
      "loss": 0.6817,
      "step": 9470
    },
    {
      "epoch": 3.0823031880286273,
      "grad_norm": 2.893087387084961,
      "learning_rate": 1.6462432041985988e-05,
      "loss": 0.6824,
      "step": 9475
    },
    {
      "epoch": 3.0839297332465843,
      "grad_norm": 1.932302713394165,
      "learning_rate": 1.6438179870952762e-05,
      "loss": 0.6573,
      "step": 9480
    },
    {
      "epoch": 3.085556278464541,
      "grad_norm": 1.8314918279647827,
      "learning_rate": 1.641393682433005e-05,
      "loss": 0.6671,
      "step": 9485
    },
    {
      "epoch": 3.0871828236824985,
      "grad_norm": 1.8099032640457153,
      "learning_rate": 1.6389702927953876e-05,
      "loss": 0.6589,
      "step": 9490
    },
    {
      "epoch": 3.0888093689004554,
      "grad_norm": 2.079960584640503,
      "learning_rate": 1.6365478207650548e-05,
      "loss": 0.657,
      "step": 9495
    },
    {
      "epoch": 3.0904359141184123,
      "grad_norm": 2.1277928352355957,
      "learning_rate": 1.634126268923655e-05,
      "loss": 0.6561,
      "step": 9500
    },
    {
      "epoch": 3.0920624593363697,
      "grad_norm": 2.0239083766937256,
      "learning_rate": 1.6317056398518603e-05,
      "loss": 0.6121,
      "step": 9505
    },
    {
      "epoch": 3.0936890045543266,
      "grad_norm": 2.638597249984741,
      "learning_rate": 1.6292859361293554e-05,
      "loss": 0.6291,
      "step": 9510
    },
    {
      "epoch": 3.0953155497722835,
      "grad_norm": 2.395009994506836,
      "learning_rate": 1.626867160334843e-05,
      "loss": 0.6799,
      "step": 9515
    },
    {
      "epoch": 3.096942094990241,
      "grad_norm": 2.195422410964966,
      "learning_rate": 1.624449315046032e-05,
      "loss": 0.6687,
      "step": 9520
    },
    {
      "epoch": 3.0985686402081978,
      "grad_norm": 2.0896871089935303,
      "learning_rate": 1.622032402839645e-05,
      "loss": 0.642,
      "step": 9525
    },
    {
      "epoch": 3.1001951854261547,
      "grad_norm": 1.827916145324707,
      "learning_rate": 1.6196164262914064e-05,
      "loss": 0.6713,
      "step": 9530
    },
    {
      "epoch": 3.101821730644112,
      "grad_norm": 1.917366623878479,
      "learning_rate": 1.617201387976045e-05,
      "loss": 0.673,
      "step": 9535
    },
    {
      "epoch": 3.103448275862069,
      "grad_norm": 1.6420366764068604,
      "learning_rate": 1.6147872904672887e-05,
      "loss": 0.6736,
      "step": 9540
    },
    {
      "epoch": 3.105074821080026,
      "grad_norm": 1.9722702503204346,
      "learning_rate": 1.612374136337864e-05,
      "loss": 0.6429,
      "step": 9545
    },
    {
      "epoch": 3.106701366297983,
      "grad_norm": 2.036595582962036,
      "learning_rate": 1.609961928159491e-05,
      "loss": 0.6805,
      "step": 9550
    },
    {
      "epoch": 3.10832791151594,
      "grad_norm": 2.1223628520965576,
      "learning_rate": 1.6075506685028825e-05,
      "loss": 0.6756,
      "step": 9555
    },
    {
      "epoch": 3.109954456733897,
      "grad_norm": 1.976130485534668,
      "learning_rate": 1.6051403599377405e-05,
      "loss": 0.6489,
      "step": 9560
    },
    {
      "epoch": 3.1115810019518544,
      "grad_norm": 2.3929290771484375,
      "learning_rate": 1.6027310050327522e-05,
      "loss": 0.6823,
      "step": 9565
    },
    {
      "epoch": 3.1132075471698113,
      "grad_norm": 1.8878233432769775,
      "learning_rate": 1.6003226063555905e-05,
      "loss": 0.6805,
      "step": 9570
    },
    {
      "epoch": 3.114834092387768,
      "grad_norm": 2.0991029739379883,
      "learning_rate": 1.5979151664729062e-05,
      "loss": 0.6892,
      "step": 9575
    },
    {
      "epoch": 3.1164606376057256,
      "grad_norm": 1.9038327932357788,
      "learning_rate": 1.5955086879503316e-05,
      "loss": 0.6709,
      "step": 9580
    },
    {
      "epoch": 3.1180871828236825,
      "grad_norm": 1.7656376361846924,
      "learning_rate": 1.5931031733524727e-05,
      "loss": 0.6424,
      "step": 9585
    },
    {
      "epoch": 3.1197137280416394,
      "grad_norm": 1.930082082748413,
      "learning_rate": 1.5906986252429087e-05,
      "loss": 0.6529,
      "step": 9590
    },
    {
      "epoch": 3.1213402732595967,
      "grad_norm": 2.5808334350585938,
      "learning_rate": 1.5882950461841872e-05,
      "loss": 0.6669,
      "step": 9595
    },
    {
      "epoch": 3.1229668184775536,
      "grad_norm": 2.0705926418304443,
      "learning_rate": 1.585892438737827e-05,
      "loss": 0.6482,
      "step": 9600
    },
    {
      "epoch": 3.1245933636955106,
      "grad_norm": 2.081015110015869,
      "learning_rate": 1.5834908054643073e-05,
      "loss": 0.6803,
      "step": 9605
    },
    {
      "epoch": 3.126219908913468,
      "grad_norm": 1.792731523513794,
      "learning_rate": 1.581090148923071e-05,
      "loss": 0.6707,
      "step": 9610
    },
    {
      "epoch": 3.127846454131425,
      "grad_norm": 2.385056972503662,
      "learning_rate": 1.5786904716725196e-05,
      "loss": 0.6334,
      "step": 9615
    },
    {
      "epoch": 3.1294729993493817,
      "grad_norm": 2.067519187927246,
      "learning_rate": 1.576291776270013e-05,
      "loss": 0.6579,
      "step": 9620
    },
    {
      "epoch": 3.131099544567339,
      "grad_norm": 1.966447353363037,
      "learning_rate": 1.5738940652718594e-05,
      "loss": 0.6396,
      "step": 9625
    },
    {
      "epoch": 3.132726089785296,
      "grad_norm": 2.1981186866760254,
      "learning_rate": 1.5714973412333257e-05,
      "loss": 0.6833,
      "step": 9630
    },
    {
      "epoch": 3.134352635003253,
      "grad_norm": 1.8088887929916382,
      "learning_rate": 1.56910160670862e-05,
      "loss": 0.6726,
      "step": 9635
    },
    {
      "epoch": 3.1359791802212102,
      "grad_norm": 1.9311386346817017,
      "learning_rate": 1.5667068642508996e-05,
      "loss": 0.6712,
      "step": 9640
    },
    {
      "epoch": 3.137605725439167,
      "grad_norm": 1.7471827268600464,
      "learning_rate": 1.5643131164122626e-05,
      "loss": 0.6524,
      "step": 9645
    },
    {
      "epoch": 3.139232270657124,
      "grad_norm": 1.9465330839157104,
      "learning_rate": 1.5619203657437503e-05,
      "loss": 0.6627,
      "step": 9650
    },
    {
      "epoch": 3.1408588158750814,
      "grad_norm": 2.1639490127563477,
      "learning_rate": 1.5595286147953364e-05,
      "loss": 0.656,
      "step": 9655
    },
    {
      "epoch": 3.1424853610930383,
      "grad_norm": 2.0224697589874268,
      "learning_rate": 1.5571378661159337e-05,
      "loss": 0.6441,
      "step": 9660
    },
    {
      "epoch": 3.1441119063109952,
      "grad_norm": 1.8008503913879395,
      "learning_rate": 1.5547481222533846e-05,
      "loss": 0.6109,
      "step": 9665
    },
    {
      "epoch": 3.1457384515289526,
      "grad_norm": 1.9340412616729736,
      "learning_rate": 1.552359385754461e-05,
      "loss": 0.7001,
      "step": 9670
    },
    {
      "epoch": 3.1473649967469095,
      "grad_norm": 2.107783794403076,
      "learning_rate": 1.549971659164861e-05,
      "loss": 0.6766,
      "step": 9675
    },
    {
      "epoch": 3.1489915419648664,
      "grad_norm": 1.9612401723861694,
      "learning_rate": 1.5475849450292085e-05,
      "loss": 0.6526,
      "step": 9680
    },
    {
      "epoch": 3.1506180871828238,
      "grad_norm": 1.7919425964355469,
      "learning_rate": 1.5451992458910442e-05,
      "loss": 0.659,
      "step": 9685
    },
    {
      "epoch": 3.1522446324007807,
      "grad_norm": 2.4329445362091064,
      "learning_rate": 1.542814564292831e-05,
      "loss": 0.6392,
      "step": 9690
    },
    {
      "epoch": 3.153871177618738,
      "grad_norm": 2.0440330505371094,
      "learning_rate": 1.540430902775946e-05,
      "loss": 0.6085,
      "step": 9695
    },
    {
      "epoch": 3.155497722836695,
      "grad_norm": 2.4745633602142334,
      "learning_rate": 1.5380482638806794e-05,
      "loss": 0.6549,
      "step": 9700
    },
    {
      "epoch": 3.157124268054652,
      "grad_norm": 2.1841065883636475,
      "learning_rate": 1.5356666501462314e-05,
      "loss": 0.6245,
      "step": 9705
    },
    {
      "epoch": 3.1587508132726088,
      "grad_norm": 4.957603454589844,
      "learning_rate": 1.533286064110709e-05,
      "loss": 0.6846,
      "step": 9710
    },
    {
      "epoch": 3.160377358490566,
      "grad_norm": 2.1394810676574707,
      "learning_rate": 1.5309065083111255e-05,
      "loss": 0.641,
      "step": 9715
    },
    {
      "epoch": 3.162003903708523,
      "grad_norm": 1.9957211017608643,
      "learning_rate": 1.5285279852833944e-05,
      "loss": 0.6611,
      "step": 9720
    },
    {
      "epoch": 3.1636304489264804,
      "grad_norm": 2.25938081741333,
      "learning_rate": 1.5261504975623306e-05,
      "loss": 0.6455,
      "step": 9725
    },
    {
      "epoch": 3.1652569941444373,
      "grad_norm": 2.0558090209960938,
      "learning_rate": 1.5237740476816436e-05,
      "loss": 0.6612,
      "step": 9730
    },
    {
      "epoch": 3.166883539362394,
      "grad_norm": 1.9984946250915527,
      "learning_rate": 1.5213986381739393e-05,
      "loss": 0.6593,
      "step": 9735
    },
    {
      "epoch": 3.168510084580351,
      "grad_norm": 2.157356023788452,
      "learning_rate": 1.519024271570712e-05,
      "loss": 0.6646,
      "step": 9740
    },
    {
      "epoch": 3.1701366297983085,
      "grad_norm": 1.8496346473693848,
      "learning_rate": 1.5166509504023473e-05,
      "loss": 0.6537,
      "step": 9745
    },
    {
      "epoch": 3.1717631750162654,
      "grad_norm": 2.2275230884552,
      "learning_rate": 1.5142786771981146e-05,
      "loss": 0.6445,
      "step": 9750
    },
    {
      "epoch": 3.1733897202342227,
      "grad_norm": 1.8542896509170532,
      "learning_rate": 1.5119074544861678e-05,
      "loss": 0.6578,
      "step": 9755
    },
    {
      "epoch": 3.1750162654521796,
      "grad_norm": 2.0999529361724854,
      "learning_rate": 1.5095372847935396e-05,
      "loss": 0.6538,
      "step": 9760
    },
    {
      "epoch": 3.1766428106701365,
      "grad_norm": 1.8991894721984863,
      "learning_rate": 1.5071681706461438e-05,
      "loss": 0.6359,
      "step": 9765
    },
    {
      "epoch": 3.178269355888094,
      "grad_norm": 2.260857105255127,
      "learning_rate": 1.5048001145687646e-05,
      "loss": 0.6646,
      "step": 9770
    },
    {
      "epoch": 3.179895901106051,
      "grad_norm": 1.817297339439392,
      "learning_rate": 1.5024331190850637e-05,
      "loss": 0.6469,
      "step": 9775
    },
    {
      "epoch": 3.1815224463240077,
      "grad_norm": 1.993095874786377,
      "learning_rate": 1.5000671867175678e-05,
      "loss": 0.6479,
      "step": 9780
    },
    {
      "epoch": 3.183148991541965,
      "grad_norm": 2.3499114513397217,
      "learning_rate": 1.4977023199876743e-05,
      "loss": 0.7044,
      "step": 9785
    },
    {
      "epoch": 3.184775536759922,
      "grad_norm": 1.8569183349609375,
      "learning_rate": 1.4953385214156423e-05,
      "loss": 0.6655,
      "step": 9790
    },
    {
      "epoch": 3.186402081977879,
      "grad_norm": 1.9735565185546875,
      "learning_rate": 1.4929757935205951e-05,
      "loss": 0.6926,
      "step": 9795
    },
    {
      "epoch": 3.1880286271958362,
      "grad_norm": 2.0960886478424072,
      "learning_rate": 1.490614138820512e-05,
      "loss": 0.6899,
      "step": 9800
    },
    {
      "epoch": 3.189655172413793,
      "grad_norm": 1.7670766115188599,
      "learning_rate": 1.4882535598322311e-05,
      "loss": 0.6324,
      "step": 9805
    },
    {
      "epoch": 3.19128171763175,
      "grad_norm": 1.9267206192016602,
      "learning_rate": 1.4858940590714427e-05,
      "loss": 0.6515,
      "step": 9810
    },
    {
      "epoch": 3.1929082628497074,
      "grad_norm": 1.7379379272460938,
      "learning_rate": 1.4835356390526888e-05,
      "loss": 0.6452,
      "step": 9815
    },
    {
      "epoch": 3.1945348080676643,
      "grad_norm": 2.3884451389312744,
      "learning_rate": 1.481178302289359e-05,
      "loss": 0.6574,
      "step": 9820
    },
    {
      "epoch": 3.1961613532856212,
      "grad_norm": 2.4783401489257812,
      "learning_rate": 1.478822051293689e-05,
      "loss": 0.6291,
      "step": 9825
    },
    {
      "epoch": 3.1977878985035786,
      "grad_norm": 2.2705469131469727,
      "learning_rate": 1.476466888576758e-05,
      "loss": 0.6549,
      "step": 9830
    },
    {
      "epoch": 3.1994144437215355,
      "grad_norm": 1.7577061653137207,
      "learning_rate": 1.4741128166484824e-05,
      "loss": 0.6498,
      "step": 9835
    },
    {
      "epoch": 3.2010409889394924,
      "grad_norm": 2.099090576171875,
      "learning_rate": 1.4717598380176212e-05,
      "loss": 0.6521,
      "step": 9840
    },
    {
      "epoch": 3.2026675341574498,
      "grad_norm": 2.067274570465088,
      "learning_rate": 1.4694079551917629e-05,
      "loss": 0.6709,
      "step": 9845
    },
    {
      "epoch": 3.2042940793754067,
      "grad_norm": 2.2046360969543457,
      "learning_rate": 1.4670571706773318e-05,
      "loss": 0.6753,
      "step": 9850
    },
    {
      "epoch": 3.2059206245933636,
      "grad_norm": 2.7054951190948486,
      "learning_rate": 1.4647074869795802e-05,
      "loss": 0.6428,
      "step": 9855
    },
    {
      "epoch": 3.207547169811321,
      "grad_norm": 1.866786241531372,
      "learning_rate": 1.462358906602589e-05,
      "loss": 0.6797,
      "step": 9860
    },
    {
      "epoch": 3.209173715029278,
      "grad_norm": 2.2875099182128906,
      "learning_rate": 1.4600114320492594e-05,
      "loss": 0.67,
      "step": 9865
    },
    {
      "epoch": 3.2108002602472347,
      "grad_norm": 2.176388740539551,
      "learning_rate": 1.4576650658213191e-05,
      "loss": 0.6426,
      "step": 9870
    },
    {
      "epoch": 3.212426805465192,
      "grad_norm": 1.9909074306488037,
      "learning_rate": 1.4553198104193094e-05,
      "loss": 0.6714,
      "step": 9875
    },
    {
      "epoch": 3.214053350683149,
      "grad_norm": 2.1360368728637695,
      "learning_rate": 1.452975668342594e-05,
      "loss": 0.6961,
      "step": 9880
    },
    {
      "epoch": 3.215679895901106,
      "grad_norm": 1.810542106628418,
      "learning_rate": 1.450632642089344e-05,
      "loss": 0.6336,
      "step": 9885
    },
    {
      "epoch": 3.2173064411190633,
      "grad_norm": 1.868118405342102,
      "learning_rate": 1.448290734156546e-05,
      "loss": 0.6469,
      "step": 9890
    },
    {
      "epoch": 3.21893298633702,
      "grad_norm": 2.1547558307647705,
      "learning_rate": 1.445949947039991e-05,
      "loss": 0.6085,
      "step": 9895
    },
    {
      "epoch": 3.220559531554977,
      "grad_norm": 2.0586001873016357,
      "learning_rate": 1.443610283234279e-05,
      "loss": 0.6509,
      "step": 9900
    },
    {
      "epoch": 3.2221860767729344,
      "grad_norm": 2.112361431121826,
      "learning_rate": 1.4412717452328084e-05,
      "loss": 0.6525,
      "step": 9905
    },
    {
      "epoch": 3.2238126219908914,
      "grad_norm": 2.0411717891693115,
      "learning_rate": 1.4389343355277852e-05,
      "loss": 0.6475,
      "step": 9910
    },
    {
      "epoch": 3.2254391672088483,
      "grad_norm": 2.1603314876556396,
      "learning_rate": 1.4365980566102044e-05,
      "loss": 0.6718,
      "step": 9915
    },
    {
      "epoch": 3.2270657124268056,
      "grad_norm": 1.6672358512878418,
      "learning_rate": 1.4342629109698627e-05,
      "loss": 0.6203,
      "step": 9920
    },
    {
      "epoch": 3.2286922576447625,
      "grad_norm": 2.003319263458252,
      "learning_rate": 1.431928901095344e-05,
      "loss": 0.6448,
      "step": 9925
    },
    {
      "epoch": 3.2303188028627194,
      "grad_norm": 1.9274591207504272,
      "learning_rate": 1.4295960294740263e-05,
      "loss": 0.6595,
      "step": 9930
    },
    {
      "epoch": 3.231945348080677,
      "grad_norm": 2.407855749130249,
      "learning_rate": 1.4272642985920704e-05,
      "loss": 0.6695,
      "step": 9935
    },
    {
      "epoch": 3.2335718932986337,
      "grad_norm": 2.0356640815734863,
      "learning_rate": 1.4249337109344241e-05,
      "loss": 0.6617,
      "step": 9940
    },
    {
      "epoch": 3.2351984385165906,
      "grad_norm": 1.9890923500061035,
      "learning_rate": 1.4226042689848163e-05,
      "loss": 0.6702,
      "step": 9945
    },
    {
      "epoch": 3.236824983734548,
      "grad_norm": 2.0976686477661133,
      "learning_rate": 1.4202759752257555e-05,
      "loss": 0.6487,
      "step": 9950
    },
    {
      "epoch": 3.238451528952505,
      "grad_norm": 2.0105624198913574,
      "learning_rate": 1.4179488321385243e-05,
      "loss": 0.6666,
      "step": 9955
    },
    {
      "epoch": 3.240078074170462,
      "grad_norm": 1.6774218082427979,
      "learning_rate": 1.415622842203182e-05,
      "loss": 0.6585,
      "step": 9960
    },
    {
      "epoch": 3.241704619388419,
      "grad_norm": 2.181556463241577,
      "learning_rate": 1.4132980078985553e-05,
      "loss": 0.6487,
      "step": 9965
    },
    {
      "epoch": 3.243331164606376,
      "grad_norm": 2.0243122577667236,
      "learning_rate": 1.4109743317022434e-05,
      "loss": 0.6852,
      "step": 9970
    },
    {
      "epoch": 3.244957709824333,
      "grad_norm": 2.3831090927124023,
      "learning_rate": 1.4086518160906084e-05,
      "loss": 0.6428,
      "step": 9975
    },
    {
      "epoch": 3.2465842550422903,
      "grad_norm": 2.026235818862915,
      "learning_rate": 1.4063304635387773e-05,
      "loss": 0.6523,
      "step": 9980
    },
    {
      "epoch": 3.248210800260247,
      "grad_norm": 2.0918619632720947,
      "learning_rate": 1.4040102765206375e-05,
      "loss": 0.6746,
      "step": 9985
    },
    {
      "epoch": 3.249837345478204,
      "grad_norm": 1.862228274345398,
      "learning_rate": 1.4016912575088318e-05,
      "loss": 0.6756,
      "step": 9990
    },
    {
      "epoch": 3.2514638906961615,
      "grad_norm": 2.001894235610962,
      "learning_rate": 1.3993734089747617e-05,
      "loss": 0.6531,
      "step": 9995
    },
    {
      "epoch": 3.2530904359141184,
      "grad_norm": 1.8046789169311523,
      "learning_rate": 1.3970567333885786e-05,
      "loss": 0.6547,
      "step": 10000
    },
    {
      "epoch": 3.2547169811320753,
      "grad_norm": 1.9411813020706177,
      "learning_rate": 1.3947412332191855e-05,
      "loss": 0.6805,
      "step": 10005
    },
    {
      "epoch": 3.2563435263500327,
      "grad_norm": 1.9151054620742798,
      "learning_rate": 1.392426910934232e-05,
      "loss": 0.6648,
      "step": 10010
    },
    {
      "epoch": 3.2579700715679896,
      "grad_norm": 1.8788055181503296,
      "learning_rate": 1.3901137690001137e-05,
      "loss": 0.6224,
      "step": 10015
    },
    {
      "epoch": 3.2595966167859465,
      "grad_norm": 2.2758357524871826,
      "learning_rate": 1.3878018098819657e-05,
      "loss": 0.6579,
      "step": 10020
    },
    {
      "epoch": 3.261223162003904,
      "grad_norm": 2.455655813217163,
      "learning_rate": 1.3854910360436657e-05,
      "loss": 0.6604,
      "step": 10025
    },
    {
      "epoch": 3.2628497072218607,
      "grad_norm": 2.1581814289093018,
      "learning_rate": 1.383181449947825e-05,
      "loss": 0.688,
      "step": 10030
    },
    {
      "epoch": 3.2644762524398176,
      "grad_norm": 2.1598403453826904,
      "learning_rate": 1.3808730540557913e-05,
      "loss": 0.6521,
      "step": 10035
    },
    {
      "epoch": 3.266102797657775,
      "grad_norm": 2.160384178161621,
      "learning_rate": 1.3785658508276436e-05,
      "loss": 0.6739,
      "step": 10040
    },
    {
      "epoch": 3.267729342875732,
      "grad_norm": 1.7373710870742798,
      "learning_rate": 1.3762598427221906e-05,
      "loss": 0.674,
      "step": 10045
    },
    {
      "epoch": 3.269355888093689,
      "grad_norm": 2.675269365310669,
      "learning_rate": 1.3739550321969647e-05,
      "loss": 0.6825,
      "step": 10050
    },
    {
      "epoch": 3.270982433311646,
      "grad_norm": 1.9858858585357666,
      "learning_rate": 1.3716514217082249e-05,
      "loss": 0.6811,
      "step": 10055
    },
    {
      "epoch": 3.272608978529603,
      "grad_norm": 2.0622477531433105,
      "learning_rate": 1.3693490137109485e-05,
      "loss": 0.633,
      "step": 10060
    },
    {
      "epoch": 3.27423552374756,
      "grad_norm": 2.0745816230773926,
      "learning_rate": 1.3670478106588341e-05,
      "loss": 0.6432,
      "step": 10065
    },
    {
      "epoch": 3.2758620689655173,
      "grad_norm": 1.9760076999664307,
      "learning_rate": 1.364747815004295e-05,
      "loss": 0.6758,
      "step": 10070
    },
    {
      "epoch": 3.2774886141834743,
      "grad_norm": 1.8305747509002686,
      "learning_rate": 1.3624490291984582e-05,
      "loss": 0.6453,
      "step": 10075
    },
    {
      "epoch": 3.279115159401431,
      "grad_norm": 1.9638245105743408,
      "learning_rate": 1.3601514556911596e-05,
      "loss": 0.6833,
      "step": 10080
    },
    {
      "epoch": 3.2807417046193885,
      "grad_norm": 1.8853123188018799,
      "learning_rate": 1.3578550969309459e-05,
      "loss": 0.6502,
      "step": 10085
    },
    {
      "epoch": 3.2823682498373454,
      "grad_norm": 2.171973466873169,
      "learning_rate": 1.3555599553650658e-05,
      "loss": 0.6604,
      "step": 10090
    },
    {
      "epoch": 3.2839947950553023,
      "grad_norm": 2.070957899093628,
      "learning_rate": 1.3532660334394742e-05,
      "loss": 0.6766,
      "step": 10095
    },
    {
      "epoch": 3.2856213402732597,
      "grad_norm": 1.9634778499603271,
      "learning_rate": 1.3509733335988245e-05,
      "loss": 0.6532,
      "step": 10100
    },
    {
      "epoch": 3.2872478854912166,
      "grad_norm": 2.1095433235168457,
      "learning_rate": 1.3486818582864678e-05,
      "loss": 0.6566,
      "step": 10105
    },
    {
      "epoch": 3.288874430709174,
      "grad_norm": 2.1379847526550293,
      "learning_rate": 1.3463916099444518e-05,
      "loss": 0.6971,
      "step": 10110
    },
    {
      "epoch": 3.290500975927131,
      "grad_norm": 2.246108293533325,
      "learning_rate": 1.344102591013513e-05,
      "loss": 0.6653,
      "step": 10115
    },
    {
      "epoch": 3.2921275211450878,
      "grad_norm": 2.186302661895752,
      "learning_rate": 1.3418148039330822e-05,
      "loss": 0.6718,
      "step": 10120
    },
    {
      "epoch": 3.2937540663630447,
      "grad_norm": 2.1060941219329834,
      "learning_rate": 1.339528251141273e-05,
      "loss": 0.6681,
      "step": 10125
    },
    {
      "epoch": 3.295380611581002,
      "grad_norm": 2.1231400966644287,
      "learning_rate": 1.3372429350748866e-05,
      "loss": 0.6763,
      "step": 10130
    },
    {
      "epoch": 3.297007156798959,
      "grad_norm": 2.142677068710327,
      "learning_rate": 1.3349588581694058e-05,
      "loss": 0.6504,
      "step": 10135
    },
    {
      "epoch": 3.2986337020169163,
      "grad_norm": 2.07317852973938,
      "learning_rate": 1.332676022858993e-05,
      "loss": 0.645,
      "step": 10140
    },
    {
      "epoch": 3.300260247234873,
      "grad_norm": 2.129087448120117,
      "learning_rate": 1.3303944315764848e-05,
      "loss": 0.6666,
      "step": 10145
    },
    {
      "epoch": 3.30188679245283,
      "grad_norm": 1.9616256952285767,
      "learning_rate": 1.3281140867533962e-05,
      "loss": 0.6404,
      "step": 10150
    },
    {
      "epoch": 3.303513337670787,
      "grad_norm": 1.7806838750839233,
      "learning_rate": 1.3258349908199098e-05,
      "loss": 0.6373,
      "step": 10155
    },
    {
      "epoch": 3.3051398828887444,
      "grad_norm": 1.8721908330917358,
      "learning_rate": 1.3235571462048795e-05,
      "loss": 0.653,
      "step": 10160
    },
    {
      "epoch": 3.3067664281067013,
      "grad_norm": 1.8101102113723755,
      "learning_rate": 1.321280555335826e-05,
      "loss": 0.6488,
      "step": 10165
    },
    {
      "epoch": 3.3083929733246586,
      "grad_norm": 1.9928457736968994,
      "learning_rate": 1.3190052206389337e-05,
      "loss": 0.631,
      "step": 10170
    },
    {
      "epoch": 3.3100195185426156,
      "grad_norm": 1.8210963010787964,
      "learning_rate": 1.3167311445390456e-05,
      "loss": 0.6227,
      "step": 10175
    },
    {
      "epoch": 3.3116460637605725,
      "grad_norm": 1.9028152227401733,
      "learning_rate": 1.314458329459668e-05,
      "loss": 0.6644,
      "step": 10180
    },
    {
      "epoch": 3.3132726089785294,
      "grad_norm": 2.116238832473755,
      "learning_rate": 1.3121867778229588e-05,
      "loss": 0.653,
      "step": 10185
    },
    {
      "epoch": 3.3148991541964867,
      "grad_norm": 1.8992266654968262,
      "learning_rate": 1.309916492049732e-05,
      "loss": 0.6675,
      "step": 10190
    },
    {
      "epoch": 3.3165256994144436,
      "grad_norm": 1.8556034564971924,
      "learning_rate": 1.3076474745594524e-05,
      "loss": 0.6604,
      "step": 10195
    },
    {
      "epoch": 3.318152244632401,
      "grad_norm": 1.9362324476242065,
      "learning_rate": 1.3053797277702339e-05,
      "loss": 0.6423,
      "step": 10200
    },
    {
      "epoch": 3.319778789850358,
      "grad_norm": 1.9838426113128662,
      "learning_rate": 1.3031132540988331e-05,
      "loss": 0.6687,
      "step": 10205
    },
    {
      "epoch": 3.321405335068315,
      "grad_norm": 2.070091485977173,
      "learning_rate": 1.3008480559606534e-05,
      "loss": 0.676,
      "step": 10210
    },
    {
      "epoch": 3.3230318802862717,
      "grad_norm": 1.8255711793899536,
      "learning_rate": 1.2985841357697359e-05,
      "loss": 0.6741,
      "step": 10215
    },
    {
      "epoch": 3.324658425504229,
      "grad_norm": 2.1347503662109375,
      "learning_rate": 1.2963214959387632e-05,
      "loss": 0.681,
      "step": 10220
    },
    {
      "epoch": 3.326284970722186,
      "grad_norm": 1.940697193145752,
      "learning_rate": 1.2940601388790475e-05,
      "loss": 0.6698,
      "step": 10225
    },
    {
      "epoch": 3.3279115159401433,
      "grad_norm": 2.2285683155059814,
      "learning_rate": 1.2918000670005429e-05,
      "loss": 0.6906,
      "step": 10230
    },
    {
      "epoch": 3.3295380611581002,
      "grad_norm": 1.9986547231674194,
      "learning_rate": 1.2895412827118252e-05,
      "loss": 0.6535,
      "step": 10235
    },
    {
      "epoch": 3.331164606376057,
      "grad_norm": 1.889677882194519,
      "learning_rate": 1.2872837884201028e-05,
      "loss": 0.6673,
      "step": 10240
    },
    {
      "epoch": 3.332791151594014,
      "grad_norm": 1.8856425285339355,
      "learning_rate": 1.2850275865312089e-05,
      "loss": 0.6598,
      "step": 10245
    },
    {
      "epoch": 3.3344176968119714,
      "grad_norm": 2.092761278152466,
      "learning_rate": 1.282772679449597e-05,
      "loss": 0.6555,
      "step": 10250
    },
    {
      "epoch": 3.3360442420299283,
      "grad_norm": 1.8956481218338013,
      "learning_rate": 1.2805190695783442e-05,
      "loss": 0.6418,
      "step": 10255
    },
    {
      "epoch": 3.3376707872478857,
      "grad_norm": 1.9665729999542236,
      "learning_rate": 1.2782667593191403e-05,
      "loss": 0.6459,
      "step": 10260
    },
    {
      "epoch": 3.3392973324658426,
      "grad_norm": 1.6512507200241089,
      "learning_rate": 1.276015751072297e-05,
      "loss": 0.6267,
      "step": 10265
    },
    {
      "epoch": 3.3409238776837995,
      "grad_norm": 2.1978135108947754,
      "learning_rate": 1.2737660472367314e-05,
      "loss": 0.6704,
      "step": 10270
    },
    {
      "epoch": 3.342550422901757,
      "grad_norm": 1.9107602834701538,
      "learning_rate": 1.2715176502099755e-05,
      "loss": 0.6351,
      "step": 10275
    },
    {
      "epoch": 3.3441769681197138,
      "grad_norm": 2.22953462600708,
      "learning_rate": 1.2692705623881651e-05,
      "loss": 0.6736,
      "step": 10280
    },
    {
      "epoch": 3.3458035133376707,
      "grad_norm": 1.9832690954208374,
      "learning_rate": 1.2670247861660438e-05,
      "loss": 0.7022,
      "step": 10285
    },
    {
      "epoch": 3.347430058555628,
      "grad_norm": 1.9249918460845947,
      "learning_rate": 1.264780323936954e-05,
      "loss": 0.6531,
      "step": 10290
    },
    {
      "epoch": 3.349056603773585,
      "grad_norm": 1.793448567390442,
      "learning_rate": 1.2625371780928428e-05,
      "loss": 0.663,
      "step": 10295
    },
    {
      "epoch": 3.350683148991542,
      "grad_norm": 1.7719804048538208,
      "learning_rate": 1.2602953510242487e-05,
      "loss": 0.6372,
      "step": 10300
    },
    {
      "epoch": 3.352309694209499,
      "grad_norm": 1.9875365495681763,
      "learning_rate": 1.2580548451203095e-05,
      "loss": 0.6202,
      "step": 10305
    },
    {
      "epoch": 3.353936239427456,
      "grad_norm": 1.892668604850769,
      "learning_rate": 1.2558156627687507e-05,
      "loss": 0.6498,
      "step": 10310
    },
    {
      "epoch": 3.355562784645413,
      "grad_norm": 1.960500955581665,
      "learning_rate": 1.2535778063558917e-05,
      "loss": 0.6427,
      "step": 10315
    },
    {
      "epoch": 3.3571893298633704,
      "grad_norm": 2.0554163455963135,
      "learning_rate": 1.251341278266635e-05,
      "loss": 0.6934,
      "step": 10320
    },
    {
      "epoch": 3.3588158750813273,
      "grad_norm": 1.8425451517105103,
      "learning_rate": 1.2491060808844696e-05,
      "loss": 0.6604,
      "step": 10325
    },
    {
      "epoch": 3.360442420299284,
      "grad_norm": 2.2402796745300293,
      "learning_rate": 1.2468722165914662e-05,
      "loss": 0.666,
      "step": 10330
    },
    {
      "epoch": 3.3620689655172415,
      "grad_norm": 2.013751745223999,
      "learning_rate": 1.2446396877682756e-05,
      "loss": 0.6825,
      "step": 10335
    },
    {
      "epoch": 3.3636955107351985,
      "grad_norm": 2.137054204940796,
      "learning_rate": 1.2424084967941222e-05,
      "loss": 0.6573,
      "step": 10340
    },
    {
      "epoch": 3.3653220559531554,
      "grad_norm": 1.8506207466125488,
      "learning_rate": 1.2401786460468087e-05,
      "loss": 0.6427,
      "step": 10345
    },
    {
      "epoch": 3.3669486011711127,
      "grad_norm": 2.129305124282837,
      "learning_rate": 1.2379501379027059e-05,
      "loss": 0.6433,
      "step": 10350
    },
    {
      "epoch": 3.3685751463890696,
      "grad_norm": 2.1928210258483887,
      "learning_rate": 1.235722974736756e-05,
      "loss": 0.6897,
      "step": 10355
    },
    {
      "epoch": 3.3702016916070265,
      "grad_norm": 1.9839317798614502,
      "learning_rate": 1.2334971589224675e-05,
      "loss": 0.6797,
      "step": 10360
    },
    {
      "epoch": 3.371828236824984,
      "grad_norm": 2.428084135055542,
      "learning_rate": 1.2312726928319138e-05,
      "loss": 0.6598,
      "step": 10365
    },
    {
      "epoch": 3.373454782042941,
      "grad_norm": 2.3940176963806152,
      "learning_rate": 1.2290495788357267e-05,
      "loss": 0.6366,
      "step": 10370
    },
    {
      "epoch": 3.3750813272608977,
      "grad_norm": 2.1621506214141846,
      "learning_rate": 1.2268278193031008e-05,
      "loss": 0.6703,
      "step": 10375
    },
    {
      "epoch": 3.376707872478855,
      "grad_norm": 1.8342362642288208,
      "learning_rate": 1.224607416601786e-05,
      "loss": 0.6521,
      "step": 10380
    },
    {
      "epoch": 3.378334417696812,
      "grad_norm": 1.9424413442611694,
      "learning_rate": 1.2223883730980843e-05,
      "loss": 0.6687,
      "step": 10385
    },
    {
      "epoch": 3.379960962914769,
      "grad_norm": 2.1477859020233154,
      "learning_rate": 1.2201706911568515e-05,
      "loss": 0.6543,
      "step": 10390
    },
    {
      "epoch": 3.3815875081327262,
      "grad_norm": 1.9607585668563843,
      "learning_rate": 1.2179543731414919e-05,
      "loss": 0.6492,
      "step": 10395
    },
    {
      "epoch": 3.383214053350683,
      "grad_norm": 2.1383254528045654,
      "learning_rate": 1.215739421413957e-05,
      "loss": 0.6451,
      "step": 10400
    },
    {
      "epoch": 3.38484059856864,
      "grad_norm": 2.1169769763946533,
      "learning_rate": 1.2135258383347392e-05,
      "loss": 0.646,
      "step": 10405
    },
    {
      "epoch": 3.3864671437865974,
      "grad_norm": 1.802030086517334,
      "learning_rate": 1.211313626262876e-05,
      "loss": 0.6289,
      "step": 10410
    },
    {
      "epoch": 3.3880936890045543,
      "grad_norm": 2.1696512699127197,
      "learning_rate": 1.2091027875559408e-05,
      "loss": 0.6714,
      "step": 10415
    },
    {
      "epoch": 3.3897202342225112,
      "grad_norm": 1.7991942167282104,
      "learning_rate": 1.2068933245700454e-05,
      "loss": 0.6352,
      "step": 10420
    },
    {
      "epoch": 3.3913467794404686,
      "grad_norm": 2.010716199874878,
      "learning_rate": 1.204685239659835e-05,
      "loss": 0.6629,
      "step": 10425
    },
    {
      "epoch": 3.3929733246584255,
      "grad_norm": 1.9059813022613525,
      "learning_rate": 1.2024785351784868e-05,
      "loss": 0.6728,
      "step": 10430
    },
    {
      "epoch": 3.3945998698763824,
      "grad_norm": 1.8924747705459595,
      "learning_rate": 1.2002732134777039e-05,
      "loss": 0.6636,
      "step": 10435
    },
    {
      "epoch": 3.3962264150943398,
      "grad_norm": 1.7315809726715088,
      "learning_rate": 1.1980692769077207e-05,
      "loss": 0.6646,
      "step": 10440
    },
    {
      "epoch": 3.3978529603122967,
      "grad_norm": 1.937632441520691,
      "learning_rate": 1.1958667278172897e-05,
      "loss": 0.6582,
      "step": 10445
    },
    {
      "epoch": 3.3994795055302536,
      "grad_norm": 1.9362993240356445,
      "learning_rate": 1.1936655685536896e-05,
      "loss": 0.692,
      "step": 10450
    },
    {
      "epoch": 3.401106050748211,
      "grad_norm": 2.0336852073669434,
      "learning_rate": 1.1914658014627156e-05,
      "loss": 0.6641,
      "step": 10455
    },
    {
      "epoch": 3.402732595966168,
      "grad_norm": 2.393552303314209,
      "learning_rate": 1.1892674288886807e-05,
      "loss": 0.6588,
      "step": 10460
    },
    {
      "epoch": 3.4043591411841247,
      "grad_norm": 1.6933391094207764,
      "learning_rate": 1.1870704531744093e-05,
      "loss": 0.6424,
      "step": 10465
    },
    {
      "epoch": 3.405985686402082,
      "grad_norm": 2.002310037612915,
      "learning_rate": 1.18487487666124e-05,
      "loss": 0.6804,
      "step": 10470
    },
    {
      "epoch": 3.407612231620039,
      "grad_norm": 2.073310375213623,
      "learning_rate": 1.182680701689017e-05,
      "loss": 0.6614,
      "step": 10475
    },
    {
      "epoch": 3.409238776837996,
      "grad_norm": 1.9920084476470947,
      "learning_rate": 1.1804879305960942e-05,
      "loss": 0.647,
      "step": 10480
    },
    {
      "epoch": 3.4108653220559533,
      "grad_norm": 2.400181531906128,
      "learning_rate": 1.1782965657193277e-05,
      "loss": 0.644,
      "step": 10485
    },
    {
      "epoch": 3.41249186727391,
      "grad_norm": 2.0325639247894287,
      "learning_rate": 1.1761066093940758e-05,
      "loss": 0.6813,
      "step": 10490
    },
    {
      "epoch": 3.414118412491867,
      "grad_norm": 2.365372657775879,
      "learning_rate": 1.1739180639541938e-05,
      "loss": 0.6757,
      "step": 10495
    },
    {
      "epoch": 3.4157449577098244,
      "grad_norm": 1.978880763053894,
      "learning_rate": 1.1717309317320365e-05,
      "loss": 0.6654,
      "step": 10500
    },
    {
      "epoch": 3.4173715029277814,
      "grad_norm": 1.876266360282898,
      "learning_rate": 1.1695452150584484e-05,
      "loss": 0.6799,
      "step": 10505
    },
    {
      "epoch": 3.4189980481457383,
      "grad_norm": 2.210228204727173,
      "learning_rate": 1.1673609162627697e-05,
      "loss": 0.666,
      "step": 10510
    },
    {
      "epoch": 3.4206245933636956,
      "grad_norm": 2.094750165939331,
      "learning_rate": 1.1651780376728269e-05,
      "loss": 0.6802,
      "step": 10515
    },
    {
      "epoch": 3.4222511385816525,
      "grad_norm": 2.0837318897247314,
      "learning_rate": 1.1629965816149343e-05,
      "loss": 0.6626,
      "step": 10520
    },
    {
      "epoch": 3.4238776837996094,
      "grad_norm": 1.8837127685546875,
      "learning_rate": 1.1608165504138904e-05,
      "loss": 0.6722,
      "step": 10525
    },
    {
      "epoch": 3.425504229017567,
      "grad_norm": 1.968827486038208,
      "learning_rate": 1.1586379463929727e-05,
      "loss": 0.6562,
      "step": 10530
    },
    {
      "epoch": 3.4271307742355237,
      "grad_norm": 2.7125394344329834,
      "learning_rate": 1.1564607718739418e-05,
      "loss": 0.6736,
      "step": 10535
    },
    {
      "epoch": 3.4287573194534806,
      "grad_norm": 1.7573403120040894,
      "learning_rate": 1.1542850291770301e-05,
      "loss": 0.6493,
      "step": 10540
    },
    {
      "epoch": 3.430383864671438,
      "grad_norm": 1.8931772708892822,
      "learning_rate": 1.1521107206209478e-05,
      "loss": 0.6374,
      "step": 10545
    },
    {
      "epoch": 3.432010409889395,
      "grad_norm": 2.2354066371917725,
      "learning_rate": 1.1499378485228757e-05,
      "loss": 0.6826,
      "step": 10550
    },
    {
      "epoch": 3.4336369551073522,
      "grad_norm": 2.4254794120788574,
      "learning_rate": 1.1477664151984646e-05,
      "loss": 0.6592,
      "step": 10555
    },
    {
      "epoch": 3.435263500325309,
      "grad_norm": 2.335663318634033,
      "learning_rate": 1.1455964229618287e-05,
      "loss": 0.6697,
      "step": 10560
    },
    {
      "epoch": 3.436890045543266,
      "grad_norm": 1.8259519338607788,
      "learning_rate": 1.1434278741255508e-05,
      "loss": 0.64,
      "step": 10565
    },
    {
      "epoch": 3.438516590761223,
      "grad_norm": 1.8743382692337036,
      "learning_rate": 1.1412607710006717e-05,
      "loss": 0.6743,
      "step": 10570
    },
    {
      "epoch": 3.4401431359791803,
      "grad_norm": 2.0380537509918213,
      "learning_rate": 1.1390951158966936e-05,
      "loss": 0.6458,
      "step": 10575
    },
    {
      "epoch": 3.441769681197137,
      "grad_norm": 2.0234997272491455,
      "learning_rate": 1.1369309111215756e-05,
      "loss": 0.6314,
      "step": 10580
    },
    {
      "epoch": 3.4433962264150946,
      "grad_norm": 1.9040913581848145,
      "learning_rate": 1.1347681589817311e-05,
      "loss": 0.6265,
      "step": 10585
    },
    {
      "epoch": 3.4450227716330515,
      "grad_norm": 2.380798578262329,
      "learning_rate": 1.1326068617820235e-05,
      "loss": 0.6506,
      "step": 10590
    },
    {
      "epoch": 3.4466493168510084,
      "grad_norm": 1.9014941453933716,
      "learning_rate": 1.1304470218257684e-05,
      "loss": 0.6692,
      "step": 10595
    },
    {
      "epoch": 3.4482758620689653,
      "grad_norm": 2.294818878173828,
      "learning_rate": 1.1282886414147261e-05,
      "loss": 0.6765,
      "step": 10600
    },
    {
      "epoch": 3.4499024072869227,
      "grad_norm": 2.3812739849090576,
      "learning_rate": 1.1261317228491035e-05,
      "loss": 0.6485,
      "step": 10605
    },
    {
      "epoch": 3.4515289525048796,
      "grad_norm": 2.180023431777954,
      "learning_rate": 1.1239762684275465e-05,
      "loss": 0.6834,
      "step": 10610
    },
    {
      "epoch": 3.453155497722837,
      "grad_norm": 2.2398617267608643,
      "learning_rate": 1.121822280447146e-05,
      "loss": 0.6561,
      "step": 10615
    },
    {
      "epoch": 3.454782042940794,
      "grad_norm": 1.9909019470214844,
      "learning_rate": 1.119669761203424e-05,
      "loss": 0.6424,
      "step": 10620
    },
    {
      "epoch": 3.4564085881587507,
      "grad_norm": 1.8090051412582397,
      "learning_rate": 1.1175187129903423e-05,
      "loss": 0.6614,
      "step": 10625
    },
    {
      "epoch": 3.4580351333767076,
      "grad_norm": 1.932723045349121,
      "learning_rate": 1.1153691381002904e-05,
      "loss": 0.6504,
      "step": 10630
    },
    {
      "epoch": 3.459661678594665,
      "grad_norm": 2.2373321056365967,
      "learning_rate": 1.1132210388240923e-05,
      "loss": 0.6272,
      "step": 10635
    },
    {
      "epoch": 3.461288223812622,
      "grad_norm": 1.8688981533050537,
      "learning_rate": 1.1110744174509952e-05,
      "loss": 0.6321,
      "step": 10640
    },
    {
      "epoch": 3.4629147690305793,
      "grad_norm": 2.219717502593994,
      "learning_rate": 1.1089292762686732e-05,
      "loss": 0.6321,
      "step": 10645
    },
    {
      "epoch": 3.464541314248536,
      "grad_norm": 2.1592214107513428,
      "learning_rate": 1.1067856175632257e-05,
      "loss": 0.6473,
      "step": 10650
    },
    {
      "epoch": 3.466167859466493,
      "grad_norm": 2.246952533721924,
      "learning_rate": 1.1046434436191669e-05,
      "loss": 0.6827,
      "step": 10655
    },
    {
      "epoch": 3.46779440468445,
      "grad_norm": 1.9593287706375122,
      "learning_rate": 1.1025027567194331e-05,
      "loss": 0.6698,
      "step": 10660
    },
    {
      "epoch": 3.4694209499024073,
      "grad_norm": 1.896320104598999,
      "learning_rate": 1.100363559145372e-05,
      "loss": 0.6467,
      "step": 10665
    },
    {
      "epoch": 3.4710474951203643,
      "grad_norm": 1.9459210634231567,
      "learning_rate": 1.0982258531767484e-05,
      "loss": 0.67,
      "step": 10670
    },
    {
      "epoch": 3.4726740403383216,
      "grad_norm": 2.0499391555786133,
      "learning_rate": 1.096089641091732e-05,
      "loss": 0.6427,
      "step": 10675
    },
    {
      "epoch": 3.4743005855562785,
      "grad_norm": 1.976870059967041,
      "learning_rate": 1.0939549251669079e-05,
      "loss": 0.6851,
      "step": 10680
    },
    {
      "epoch": 3.4759271307742354,
      "grad_norm": 2.0898680686950684,
      "learning_rate": 1.0918217076772591e-05,
      "loss": 0.6911,
      "step": 10685
    },
    {
      "epoch": 3.4775536759921923,
      "grad_norm": 2.5856997966766357,
      "learning_rate": 1.0896899908961775e-05,
      "loss": 0.6432,
      "step": 10690
    },
    {
      "epoch": 3.4791802212101497,
      "grad_norm": 2.0004806518554688,
      "learning_rate": 1.087559777095451e-05,
      "loss": 0.6553,
      "step": 10695
    },
    {
      "epoch": 3.4808067664281066,
      "grad_norm": 1.8624881505966187,
      "learning_rate": 1.0854310685452704e-05,
      "loss": 0.6234,
      "step": 10700
    },
    {
      "epoch": 3.482433311646064,
      "grad_norm": 3.014589548110962,
      "learning_rate": 1.0833038675142176e-05,
      "loss": 0.6503,
      "step": 10705
    },
    {
      "epoch": 3.484059856864021,
      "grad_norm": 2.341999053955078,
      "learning_rate": 1.0811781762692716e-05,
      "loss": 0.6504,
      "step": 10710
    },
    {
      "epoch": 3.4856864020819778,
      "grad_norm": 2.056342840194702,
      "learning_rate": 1.079053997075801e-05,
      "loss": 0.6369,
      "step": 10715
    },
    {
      "epoch": 3.487312947299935,
      "grad_norm": 2.085636854171753,
      "learning_rate": 1.076931332197564e-05,
      "loss": 0.6712,
      "step": 10720
    },
    {
      "epoch": 3.488939492517892,
      "grad_norm": 1.8923035860061646,
      "learning_rate": 1.0748101838967026e-05,
      "loss": 0.6449,
      "step": 10725
    },
    {
      "epoch": 3.490566037735849,
      "grad_norm": 1.9910145998001099,
      "learning_rate": 1.0726905544337454e-05,
      "loss": 0.6528,
      "step": 10730
    },
    {
      "epoch": 3.4921925829538063,
      "grad_norm": 2.0261192321777344,
      "learning_rate": 1.0705724460675994e-05,
      "loss": 0.6638,
      "step": 10735
    },
    {
      "epoch": 3.493819128171763,
      "grad_norm": 1.9802298545837402,
      "learning_rate": 1.0684558610555534e-05,
      "loss": 0.6565,
      "step": 10740
    },
    {
      "epoch": 3.49544567338972,
      "grad_norm": 2.180168390274048,
      "learning_rate": 1.0663408016532708e-05,
      "loss": 0.6994,
      "step": 10745
    },
    {
      "epoch": 3.4970722186076775,
      "grad_norm": 1.9455138444900513,
      "learning_rate": 1.0642272701147915e-05,
      "loss": 0.6534,
      "step": 10750
    },
    {
      "epoch": 3.4986987638256344,
      "grad_norm": 2.1108453273773193,
      "learning_rate": 1.0621152686925234e-05,
      "loss": 0.6493,
      "step": 10755
    },
    {
      "epoch": 3.5003253090435913,
      "grad_norm": 1.8773350715637207,
      "learning_rate": 1.0600047996372474e-05,
      "loss": 0.6524,
      "step": 10760
    },
    {
      "epoch": 3.5019518542615486,
      "grad_norm": 2.0078532695770264,
      "learning_rate": 1.0578958651981089e-05,
      "loss": 0.6526,
      "step": 10765
    },
    {
      "epoch": 3.5035783994795056,
      "grad_norm": 1.8449904918670654,
      "learning_rate": 1.0557884676226184e-05,
      "loss": 0.665,
      "step": 10770
    },
    {
      "epoch": 3.5052049446974625,
      "grad_norm": 2.0787408351898193,
      "learning_rate": 1.0536826091566498e-05,
      "loss": 0.6343,
      "step": 10775
    },
    {
      "epoch": 3.5068314899154194,
      "grad_norm": 2.1818220615386963,
      "learning_rate": 1.0515782920444354e-05,
      "loss": 0.6438,
      "step": 10780
    },
    {
      "epoch": 3.5084580351333767,
      "grad_norm": 1.9410792589187622,
      "learning_rate": 1.0494755185285666e-05,
      "loss": 0.6586,
      "step": 10785
    },
    {
      "epoch": 3.5100845803513336,
      "grad_norm": 1.8695874214172363,
      "learning_rate": 1.0473742908499863e-05,
      "loss": 0.6635,
      "step": 10790
    },
    {
      "epoch": 3.511711125569291,
      "grad_norm": 2.154265880584717,
      "learning_rate": 1.045274611247994e-05,
      "loss": 0.6409,
      "step": 10795
    },
    {
      "epoch": 3.513337670787248,
      "grad_norm": 2.1310086250305176,
      "learning_rate": 1.043176481960236e-05,
      "loss": 0.643,
      "step": 10800
    },
    {
      "epoch": 3.514964216005205,
      "grad_norm": 1.9660890102386475,
      "learning_rate": 1.0410799052227089e-05,
      "loss": 0.6604,
      "step": 10805
    },
    {
      "epoch": 3.516590761223162,
      "grad_norm": 2.0470850467681885,
      "learning_rate": 1.0389848832697536e-05,
      "loss": 0.6494,
      "step": 10810
    },
    {
      "epoch": 3.518217306441119,
      "grad_norm": 1.8898303508758545,
      "learning_rate": 1.0368914183340552e-05,
      "loss": 0.6563,
      "step": 10815
    },
    {
      "epoch": 3.519843851659076,
      "grad_norm": 1.815321445465088,
      "learning_rate": 1.034799512646637e-05,
      "loss": 0.6487,
      "step": 10820
    },
    {
      "epoch": 3.5214703968770333,
      "grad_norm": 2.03470778465271,
      "learning_rate": 1.0327091684368639e-05,
      "loss": 0.6853,
      "step": 10825
    },
    {
      "epoch": 3.5230969420949902,
      "grad_norm": 1.9658763408660889,
      "learning_rate": 1.030620387932433e-05,
      "loss": 0.6784,
      "step": 10830
    },
    {
      "epoch": 3.524723487312947,
      "grad_norm": 2.371723175048828,
      "learning_rate": 1.0285331733593778e-05,
      "loss": 0.6527,
      "step": 10835
    },
    {
      "epoch": 3.5263500325309045,
      "grad_norm": 2.1440389156341553,
      "learning_rate": 1.0264475269420621e-05,
      "loss": 0.6546,
      "step": 10840
    },
    {
      "epoch": 3.5279765777488614,
      "grad_norm": 2.305694103240967,
      "learning_rate": 1.0243634509031794e-05,
      "loss": 0.6663,
      "step": 10845
    },
    {
      "epoch": 3.5296031229668183,
      "grad_norm": 2.1012611389160156,
      "learning_rate": 1.0222809474637471e-05,
      "loss": 0.6327,
      "step": 10850
    },
    {
      "epoch": 3.5312296681847757,
      "grad_norm": 2.1110281944274902,
      "learning_rate": 1.0202000188431097e-05,
      "loss": 0.6849,
      "step": 10855
    },
    {
      "epoch": 3.5328562134027326,
      "grad_norm": 2.1224873065948486,
      "learning_rate": 1.0181206672589305e-05,
      "loss": 0.6643,
      "step": 10860
    },
    {
      "epoch": 3.5344827586206895,
      "grad_norm": 1.970705509185791,
      "learning_rate": 1.016458322947035e-05,
      "loss": 0.6545,
      "step": 10865
    },
    {
      "epoch": 3.536109303838647,
      "grad_norm": 2.3876793384552,
      "learning_rate": 1.0143818156116323e-05,
      "loss": 0.6431,
      "step": 10870
    },
    {
      "epoch": 3.5377358490566038,
      "grad_norm": 1.9727391004562378,
      "learning_rate": 1.0123068915132e-05,
      "loss": 0.6667,
      "step": 10875
    },
    {
      "epoch": 3.5393623942745607,
      "grad_norm": 2.115534543991089,
      "learning_rate": 1.0102335528630061e-05,
      "loss": 0.6587,
      "step": 10880
    },
    {
      "epoch": 3.540988939492518,
      "grad_norm": 1.8757723569869995,
      "learning_rate": 1.008161801870625e-05,
      "loss": 0.6372,
      "step": 10885
    },
    {
      "epoch": 3.542615484710475,
      "grad_norm": 2.2348427772521973,
      "learning_rate": 1.0060916407439413e-05,
      "loss": 0.6413,
      "step": 10890
    },
    {
      "epoch": 3.544242029928432,
      "grad_norm": 1.9503233432769775,
      "learning_rate": 1.0040230716891449e-05,
      "loss": 0.6424,
      "step": 10895
    },
    {
      "epoch": 3.545868575146389,
      "grad_norm": 2.2776994705200195,
      "learning_rate": 1.0019560969107302e-05,
      "loss": 0.6806,
      "step": 10900
    },
    {
      "epoch": 3.547495120364346,
      "grad_norm": 1.9576505422592163,
      "learning_rate": 9.99890718611489e-06,
      "loss": 0.6223,
      "step": 10905
    },
    {
      "epoch": 3.5491216655823035,
      "grad_norm": 2.091238498687744,
      "learning_rate": 9.978269389925157e-06,
      "loss": 0.6282,
      "step": 10910
    },
    {
      "epoch": 3.5507482108002604,
      "grad_norm": 2.4088802337646484,
      "learning_rate": 9.957647602531977e-06,
      "loss": 0.6532,
      "step": 10915
    },
    {
      "epoch": 3.5523747560182173,
      "grad_norm": 2.684156894683838,
      "learning_rate": 9.937041845912188e-06,
      "loss": 0.6525,
      "step": 10920
    },
    {
      "epoch": 3.554001301236174,
      "grad_norm": 2.216970205307007,
      "learning_rate": 9.91645214202553e-06,
      "loss": 0.6473,
      "step": 10925
    },
    {
      "epoch": 3.5556278464541315,
      "grad_norm": 1.7419428825378418,
      "learning_rate": 9.895878512814647e-06,
      "loss": 0.6489,
      "step": 10930
    },
    {
      "epoch": 3.5572543916720885,
      "grad_norm": 2.063995838165283,
      "learning_rate": 9.875320980205046e-06,
      "loss": 0.6622,
      "step": 10935
    },
    {
      "epoch": 3.558880936890046,
      "grad_norm": 1.8147531747817993,
      "learning_rate": 9.854779566105068e-06,
      "loss": 0.6332,
      "step": 10940
    },
    {
      "epoch": 3.5605074821080027,
      "grad_norm": 1.9879847764968872,
      "learning_rate": 9.834254292405901e-06,
      "loss": 0.6661,
      "step": 10945
    },
    {
      "epoch": 3.5621340273259596,
      "grad_norm": 1.8721420764923096,
      "learning_rate": 9.813745180981502e-06,
      "loss": 0.6182,
      "step": 10950
    },
    {
      "epoch": 3.5637605725439165,
      "grad_norm": 1.889029622077942,
      "learning_rate": 9.793252253688626e-06,
      "loss": 0.6687,
      "step": 10955
    },
    {
      "epoch": 3.565387117761874,
      "grad_norm": 2.983214855194092,
      "learning_rate": 9.772775532366774e-06,
      "loss": 0.6681,
      "step": 10960
    },
    {
      "epoch": 3.567013662979831,
      "grad_norm": 1.9701480865478516,
      "learning_rate": 9.75231503883819e-06,
      "loss": 0.6637,
      "step": 10965
    },
    {
      "epoch": 3.568640208197788,
      "grad_norm": 2.113018274307251,
      "learning_rate": 9.731870794907789e-06,
      "loss": 0.6616,
      "step": 10970
    },
    {
      "epoch": 3.570266753415745,
      "grad_norm": 1.971614956855774,
      "learning_rate": 9.711442822363209e-06,
      "loss": 0.6871,
      "step": 10975
    },
    {
      "epoch": 3.571893298633702,
      "grad_norm": 1.9272350072860718,
      "learning_rate": 9.691031142974707e-06,
      "loss": 0.6643,
      "step": 10980
    },
    {
      "epoch": 3.573519843851659,
      "grad_norm": 1.8993514776229858,
      "learning_rate": 9.670635778495213e-06,
      "loss": 0.6594,
      "step": 10985
    },
    {
      "epoch": 3.5751463890696162,
      "grad_norm": 1.8925209045410156,
      "learning_rate": 9.65025675066025e-06,
      "loss": 0.654,
      "step": 10990
    },
    {
      "epoch": 3.576772934287573,
      "grad_norm": 1.8996151685714722,
      "learning_rate": 9.629894081187943e-06,
      "loss": 0.6517,
      "step": 10995
    },
    {
      "epoch": 3.5783994795055305,
      "grad_norm": 2.024343252182007,
      "learning_rate": 9.609547791778964e-06,
      "loss": 0.6838,
      "step": 11000
    },
    {
      "epoch": 3.5800260247234874,
      "grad_norm": 1.8595373630523682,
      "learning_rate": 9.589217904116554e-06,
      "loss": 0.6708,
      "step": 11005
    },
    {
      "epoch": 3.5816525699414443,
      "grad_norm": 1.7943567037582397,
      "learning_rate": 9.568904439866444e-06,
      "loss": 0.6382,
      "step": 11010
    },
    {
      "epoch": 3.5832791151594012,
      "grad_norm": 2.03316593170166,
      "learning_rate": 9.548607420676902e-06,
      "loss": 0.6587,
      "step": 11015
    },
    {
      "epoch": 3.5849056603773586,
      "grad_norm": 1.9646462202072144,
      "learning_rate": 9.528326868178616e-06,
      "loss": 0.6355,
      "step": 11020
    },
    {
      "epoch": 3.5865322055953155,
      "grad_norm": 1.768291711807251,
      "learning_rate": 9.508062803984796e-06,
      "loss": 0.683,
      "step": 11025
    },
    {
      "epoch": 3.588158750813273,
      "grad_norm": 1.892801284790039,
      "learning_rate": 9.487815249691012e-06,
      "loss": 0.636,
      "step": 11030
    },
    {
      "epoch": 3.5897852960312298,
      "grad_norm": 1.9367650747299194,
      "learning_rate": 9.467584226875292e-06,
      "loss": 0.672,
      "step": 11035
    },
    {
      "epoch": 3.5914118412491867,
      "grad_norm": 2.049873113632202,
      "learning_rate": 9.447369757098002e-06,
      "loss": 0.6736,
      "step": 11040
    },
    {
      "epoch": 3.5930383864671436,
      "grad_norm": 2.523022413253784,
      "learning_rate": 9.427171861901903e-06,
      "loss": 0.6591,
      "step": 11045
    },
    {
      "epoch": 3.594664931685101,
      "grad_norm": 2.1238725185394287,
      "learning_rate": 9.406990562812068e-06,
      "loss": 0.6867,
      "step": 11050
    },
    {
      "epoch": 3.596291476903058,
      "grad_norm": 2.2790896892547607,
      "learning_rate": 9.386825881335889e-06,
      "loss": 0.6294,
      "step": 11055
    },
    {
      "epoch": 3.597918022121015,
      "grad_norm": 1.9417608976364136,
      "learning_rate": 9.366677838963078e-06,
      "loss": 0.6331,
      "step": 11060
    },
    {
      "epoch": 3.599544567338972,
      "grad_norm": 1.9915907382965088,
      "learning_rate": 9.34654645716556e-06,
      "loss": 0.6175,
      "step": 11065
    },
    {
      "epoch": 3.601171112556929,
      "grad_norm": 2.109654188156128,
      "learning_rate": 9.32643175739756e-06,
      "loss": 0.6211,
      "step": 11070
    },
    {
      "epoch": 3.602797657774886,
      "grad_norm": 4.143826961517334,
      "learning_rate": 9.306333761095476e-06,
      "loss": 0.6321,
      "step": 11075
    },
    {
      "epoch": 3.6044242029928433,
      "grad_norm": 1.9590835571289062,
      "learning_rate": 9.286252489677944e-06,
      "loss": 0.652,
      "step": 11080
    },
    {
      "epoch": 3.6060507482108,
      "grad_norm": 2.243879556655884,
      "learning_rate": 9.266187964545744e-06,
      "loss": 0.6638,
      "step": 11085
    },
    {
      "epoch": 3.6076772934287575,
      "grad_norm": 1.9153443574905396,
      "learning_rate": 9.246140207081833e-06,
      "loss": 0.6694,
      "step": 11090
    },
    {
      "epoch": 3.6093038386467144,
      "grad_norm": 1.8952282667160034,
      "learning_rate": 9.226109238651293e-06,
      "loss": 0.6449,
      "step": 11095
    },
    {
      "epoch": 3.6109303838646714,
      "grad_norm": 1.9768298864364624,
      "learning_rate": 9.206095080601319e-06,
      "loss": 0.6292,
      "step": 11100
    },
    {
      "epoch": 3.6125569290826283,
      "grad_norm": 2.424776077270508,
      "learning_rate": 9.18609775426116e-06,
      "loss": 0.6764,
      "step": 11105
    },
    {
      "epoch": 3.6141834743005856,
      "grad_norm": 1.9179407358169556,
      "learning_rate": 9.16611728094218e-06,
      "loss": 0.6874,
      "step": 11110
    },
    {
      "epoch": 3.6158100195185425,
      "grad_norm": 1.6501758098602295,
      "learning_rate": 9.146153681937725e-06,
      "loss": 0.6326,
      "step": 11115
    },
    {
      "epoch": 3.6174365647365,
      "grad_norm": 2.2321882247924805,
      "learning_rate": 9.126206978523202e-06,
      "loss": 0.6673,
      "step": 11120
    },
    {
      "epoch": 3.619063109954457,
      "grad_norm": 1.9093389511108398,
      "learning_rate": 9.106277191955992e-06,
      "loss": 0.6701,
      "step": 11125
    },
    {
      "epoch": 3.6206896551724137,
      "grad_norm": 2.682499885559082,
      "learning_rate": 9.086364343475461e-06,
      "loss": 0.6784,
      "step": 11130
    },
    {
      "epoch": 3.6223162003903706,
      "grad_norm": 1.8484448194503784,
      "learning_rate": 9.0664684543029e-06,
      "loss": 0.6666,
      "step": 11135
    },
    {
      "epoch": 3.623942745608328,
      "grad_norm": 1.914831280708313,
      "learning_rate": 9.04658954564156e-06,
      "loss": 0.6698,
      "step": 11140
    },
    {
      "epoch": 3.625569290826285,
      "grad_norm": 1.9206961393356323,
      "learning_rate": 9.026727638676554e-06,
      "loss": 0.6756,
      "step": 11145
    },
    {
      "epoch": 3.6271958360442422,
      "grad_norm": 1.8589897155761719,
      "learning_rate": 9.006882754574914e-06,
      "loss": 0.6601,
      "step": 11150
    },
    {
      "epoch": 3.628822381262199,
      "grad_norm": 2.0659871101379395,
      "learning_rate": 8.98705491448551e-06,
      "loss": 0.6258,
      "step": 11155
    },
    {
      "epoch": 3.630448926480156,
      "grad_norm": 2.1568715572357178,
      "learning_rate": 8.967244139539064e-06,
      "loss": 0.6161,
      "step": 11160
    },
    {
      "epoch": 3.632075471698113,
      "grad_norm": 2.5119800567626953,
      "learning_rate": 8.947450450848086e-06,
      "loss": 0.663,
      "step": 11165
    },
    {
      "epoch": 3.6337020169160703,
      "grad_norm": 2.0948991775512695,
      "learning_rate": 8.927673869506905e-06,
      "loss": 0.6529,
      "step": 11170
    },
    {
      "epoch": 3.635328562134027,
      "grad_norm": 1.8586777448654175,
      "learning_rate": 8.907914416591595e-06,
      "loss": 0.668,
      "step": 11175
    },
    {
      "epoch": 3.6369551073519846,
      "grad_norm": 2.1093571186065674,
      "learning_rate": 8.888172113159989e-06,
      "loss": 0.6499,
      "step": 11180
    },
    {
      "epoch": 3.6385816525699415,
      "grad_norm": 2.1958131790161133,
      "learning_rate": 8.868446980251647e-06,
      "loss": 0.6409,
      "step": 11185
    },
    {
      "epoch": 3.6402081977878984,
      "grad_norm": 2.053478479385376,
      "learning_rate": 8.848739038887822e-06,
      "loss": 0.668,
      "step": 11190
    },
    {
      "epoch": 3.6418347430058553,
      "grad_norm": 1.9138593673706055,
      "learning_rate": 8.829048310071456e-06,
      "loss": 0.6492,
      "step": 11195
    },
    {
      "epoch": 3.6434612882238127,
      "grad_norm": 2.214141607284546,
      "learning_rate": 8.809374814787124e-06,
      "loss": 0.6749,
      "step": 11200
    },
    {
      "epoch": 3.6450878334417696,
      "grad_norm": 2.0137126445770264,
      "learning_rate": 8.789718574001068e-06,
      "loss": 0.6634,
      "step": 11205
    },
    {
      "epoch": 3.646714378659727,
      "grad_norm": 2.1497995853424072,
      "learning_rate": 8.770079608661108e-06,
      "loss": 0.6469,
      "step": 11210
    },
    {
      "epoch": 3.648340923877684,
      "grad_norm": 1.8830994367599487,
      "learning_rate": 8.750457939696677e-06,
      "loss": 0.6608,
      "step": 11215
    },
    {
      "epoch": 3.6499674690956407,
      "grad_norm": 1.8222702741622925,
      "learning_rate": 8.730853588018772e-06,
      "loss": 0.626,
      "step": 11220
    },
    {
      "epoch": 3.6515940143135976,
      "grad_norm": 2.0502288341522217,
      "learning_rate": 8.711266574519935e-06,
      "loss": 0.6704,
      "step": 11225
    },
    {
      "epoch": 3.653220559531555,
      "grad_norm": 2.0919625759124756,
      "learning_rate": 8.691696920074214e-06,
      "loss": 0.625,
      "step": 11230
    },
    {
      "epoch": 3.654847104749512,
      "grad_norm": 2.2549009323120117,
      "learning_rate": 8.67214464553718e-06,
      "loss": 0.6254,
      "step": 11235
    },
    {
      "epoch": 3.6564736499674693,
      "grad_norm": 2.0247724056243896,
      "learning_rate": 8.652609771745862e-06,
      "loss": 0.6393,
      "step": 11240
    },
    {
      "epoch": 3.658100195185426,
      "grad_norm": 1.92485773563385,
      "learning_rate": 8.633092319518763e-06,
      "loss": 0.642,
      "step": 11245
    },
    {
      "epoch": 3.659726740403383,
      "grad_norm": 2.0102007389068604,
      "learning_rate": 8.613592309655804e-06,
      "loss": 0.6835,
      "step": 11250
    },
    {
      "epoch": 3.6613532856213404,
      "grad_norm": 2.189568519592285,
      "learning_rate": 8.594109762938343e-06,
      "loss": 0.6853,
      "step": 11255
    },
    {
      "epoch": 3.6629798308392973,
      "grad_norm": 1.7303372621536255,
      "learning_rate": 8.574644700129087e-06,
      "loss": 0.6129,
      "step": 11260
    },
    {
      "epoch": 3.6646063760572543,
      "grad_norm": 2.110283136367798,
      "learning_rate": 8.555197141972149e-06,
      "loss": 0.6555,
      "step": 11265
    },
    {
      "epoch": 3.6662329212752116,
      "grad_norm": 2.2367451190948486,
      "learning_rate": 8.535767109192955e-06,
      "loss": 0.6312,
      "step": 11270
    },
    {
      "epoch": 3.6678594664931685,
      "grad_norm": 2.0181496143341064,
      "learning_rate": 8.51635462249828e-06,
      "loss": 0.6359,
      "step": 11275
    },
    {
      "epoch": 3.6694860117111254,
      "grad_norm": 2.297574281692505,
      "learning_rate": 8.496959702576187e-06,
      "loss": 0.6564,
      "step": 11280
    },
    {
      "epoch": 3.671112556929083,
      "grad_norm": 2.1040380001068115,
      "learning_rate": 8.477582370096029e-06,
      "loss": 0.6876,
      "step": 11285
    },
    {
      "epoch": 3.6727391021470397,
      "grad_norm": 2.0794670581817627,
      "learning_rate": 8.458222645708394e-06,
      "loss": 0.6591,
      "step": 11290
    },
    {
      "epoch": 3.6743656473649966,
      "grad_norm": 2.1442930698394775,
      "learning_rate": 8.438880550045134e-06,
      "loss": 0.6566,
      "step": 11295
    },
    {
      "epoch": 3.675992192582954,
      "grad_norm": 2.2995944023132324,
      "learning_rate": 8.419556103719279e-06,
      "loss": 0.6501,
      "step": 11300
    },
    {
      "epoch": 3.677618737800911,
      "grad_norm": 1.8101574182510376,
      "learning_rate": 8.40024932732508e-06,
      "loss": 0.6404,
      "step": 11305
    },
    {
      "epoch": 3.6792452830188678,
      "grad_norm": 2.0527381896972656,
      "learning_rate": 8.380960241437947e-06,
      "loss": 0.6646,
      "step": 11310
    },
    {
      "epoch": 3.680871828236825,
      "grad_norm": 1.9029465913772583,
      "learning_rate": 8.361688866614442e-06,
      "loss": 0.6484,
      "step": 11315
    },
    {
      "epoch": 3.682498373454782,
      "grad_norm": 2.3560564517974854,
      "learning_rate": 8.342435223392232e-06,
      "loss": 0.6355,
      "step": 11320
    },
    {
      "epoch": 3.684124918672739,
      "grad_norm": 2.5555644035339355,
      "learning_rate": 8.323199332290108e-06,
      "loss": 0.6552,
      "step": 11325
    },
    {
      "epoch": 3.6857514638906963,
      "grad_norm": 1.7349854707717896,
      "learning_rate": 8.303981213807947e-06,
      "loss": 0.6521,
      "step": 11330
    },
    {
      "epoch": 3.687378009108653,
      "grad_norm": 2.144853353500366,
      "learning_rate": 8.284780888426657e-06,
      "loss": 0.6434,
      "step": 11335
    },
    {
      "epoch": 3.68900455432661,
      "grad_norm": 2.3442630767822266,
      "learning_rate": 8.265598376608211e-06,
      "loss": 0.6548,
      "step": 11340
    },
    {
      "epoch": 3.6906310995445675,
      "grad_norm": 2.33579683303833,
      "learning_rate": 8.246433698795586e-06,
      "loss": 0.623,
      "step": 11345
    },
    {
      "epoch": 3.6922576447625244,
      "grad_norm": 2.119555711746216,
      "learning_rate": 8.227286875412766e-06,
      "loss": 0.6491,
      "step": 11350
    },
    {
      "epoch": 3.6938841899804813,
      "grad_norm": 2.2910735607147217,
      "learning_rate": 8.208157926864677e-06,
      "loss": 0.6694,
      "step": 11355
    },
    {
      "epoch": 3.6955107351984386,
      "grad_norm": 1.9945921897888184,
      "learning_rate": 8.189046873537237e-06,
      "loss": 0.6241,
      "step": 11360
    },
    {
      "epoch": 3.6971372804163956,
      "grad_norm": 2.0588414669036865,
      "learning_rate": 8.169953735797251e-06,
      "loss": 0.6487,
      "step": 11365
    },
    {
      "epoch": 3.6987638256343525,
      "grad_norm": 2.145359754562378,
      "learning_rate": 8.150878533992458e-06,
      "loss": 0.649,
      "step": 11370
    },
    {
      "epoch": 3.70039037085231,
      "grad_norm": 1.9804723262786865,
      "learning_rate": 8.131821288451486e-06,
      "loss": 0.6523,
      "step": 11375
    },
    {
      "epoch": 3.7020169160702667,
      "grad_norm": 2.157130002975464,
      "learning_rate": 8.112782019483813e-06,
      "loss": 0.6437,
      "step": 11380
    },
    {
      "epoch": 3.703643461288224,
      "grad_norm": 1.9447418451309204,
      "learning_rate": 8.093760747379756e-06,
      "loss": 0.7042,
      "step": 11385
    },
    {
      "epoch": 3.705270006506181,
      "grad_norm": 2.0443172454833984,
      "learning_rate": 8.074757492410471e-06,
      "loss": 0.6875,
      "step": 11390
    },
    {
      "epoch": 3.706896551724138,
      "grad_norm": 2.231553554534912,
      "learning_rate": 8.055772274827885e-06,
      "loss": 0.6733,
      "step": 11395
    },
    {
      "epoch": 3.708523096942095,
      "grad_norm": 2.4282796382904053,
      "learning_rate": 8.036805114864736e-06,
      "loss": 0.6496,
      "step": 11400
    },
    {
      "epoch": 3.710149642160052,
      "grad_norm": 2.104163646697998,
      "learning_rate": 8.017856032734484e-06,
      "loss": 0.6515,
      "step": 11405
    },
    {
      "epoch": 3.711776187378009,
      "grad_norm": 1.9769706726074219,
      "learning_rate": 7.998925048631362e-06,
      "loss": 0.6343,
      "step": 11410
    },
    {
      "epoch": 3.7134027325959664,
      "grad_norm": 1.9851341247558594,
      "learning_rate": 7.980012182730273e-06,
      "loss": 0.6458,
      "step": 11415
    },
    {
      "epoch": 3.7150292778139233,
      "grad_norm": 2.1585636138916016,
      "learning_rate": 7.96111745518685e-06,
      "loss": 0.6622,
      "step": 11420
    },
    {
      "epoch": 3.7166558230318802,
      "grad_norm": 1.7822608947753906,
      "learning_rate": 7.942240886137364e-06,
      "loss": 0.6674,
      "step": 11425
    },
    {
      "epoch": 3.718282368249837,
      "grad_norm": 1.9461740255355835,
      "learning_rate": 7.923382495698758e-06,
      "loss": 0.6429,
      "step": 11430
    },
    {
      "epoch": 3.7199089134677945,
      "grad_norm": 2.038227081298828,
      "learning_rate": 7.904542303968585e-06,
      "loss": 0.6472,
      "step": 11435
    },
    {
      "epoch": 3.7215354586857514,
      "grad_norm": 1.929215669631958,
      "learning_rate": 7.88572033102501e-06,
      "loss": 0.6791,
      "step": 11440
    },
    {
      "epoch": 3.7231620039037088,
      "grad_norm": 2.0168604850769043,
      "learning_rate": 7.866916596926784e-06,
      "loss": 0.6516,
      "step": 11445
    },
    {
      "epoch": 3.7247885491216657,
      "grad_norm": 2.7941908836364746,
      "learning_rate": 7.848131121713234e-06,
      "loss": 0.6328,
      "step": 11450
    },
    {
      "epoch": 3.7264150943396226,
      "grad_norm": 2.256675958633423,
      "learning_rate": 7.829363925404195e-06,
      "loss": 0.6523,
      "step": 11455
    },
    {
      "epoch": 3.7280416395575795,
      "grad_norm": 1.9694584608078003,
      "learning_rate": 7.810615028000045e-06,
      "loss": 0.6609,
      "step": 11460
    },
    {
      "epoch": 3.729668184775537,
      "grad_norm": 2.0209426879882812,
      "learning_rate": 7.791884449481668e-06,
      "loss": 0.6536,
      "step": 11465
    },
    {
      "epoch": 3.7312947299934938,
      "grad_norm": 2.078065872192383,
      "learning_rate": 7.773172209810397e-06,
      "loss": 0.6988,
      "step": 11470
    },
    {
      "epoch": 3.732921275211451,
      "grad_norm": 2.086122989654541,
      "learning_rate": 7.754478328928047e-06,
      "loss": 0.6599,
      "step": 11475
    },
    {
      "epoch": 3.734547820429408,
      "grad_norm": 2.021228313446045,
      "learning_rate": 7.735802826756856e-06,
      "loss": 0.6182,
      "step": 11480
    },
    {
      "epoch": 3.736174365647365,
      "grad_norm": 2.1194801330566406,
      "learning_rate": 7.717145723199489e-06,
      "loss": 0.6752,
      "step": 11485
    },
    {
      "epoch": 3.737800910865322,
      "grad_norm": 2.1797292232513428,
      "learning_rate": 7.698507038138978e-06,
      "loss": 0.6815,
      "step": 11490
    },
    {
      "epoch": 3.739427456083279,
      "grad_norm": 2.020460844039917,
      "learning_rate": 7.679886791438754e-06,
      "loss": 0.6521,
      "step": 11495
    },
    {
      "epoch": 3.741054001301236,
      "grad_norm": 2.0722012519836426,
      "learning_rate": 7.661285002942572e-06,
      "loss": 0.6365,
      "step": 11500
    },
    {
      "epoch": 3.7426805465191935,
      "grad_norm": 1.9416598081588745,
      "learning_rate": 7.642701692474535e-06,
      "loss": 0.6474,
      "step": 11505
    },
    {
      "epoch": 3.7443070917371504,
      "grad_norm": 2.0658600330352783,
      "learning_rate": 7.624136879839053e-06,
      "loss": 0.6549,
      "step": 11510
    },
    {
      "epoch": 3.7459336369551073,
      "grad_norm": 2.0475194454193115,
      "learning_rate": 7.605590584820818e-06,
      "loss": 0.6587,
      "step": 11515
    },
    {
      "epoch": 3.747560182173064,
      "grad_norm": 1.8753339052200317,
      "learning_rate": 7.5870628271847765e-06,
      "loss": 0.67,
      "step": 11520
    },
    {
      "epoch": 3.7491867273910215,
      "grad_norm": 2.400254726409912,
      "learning_rate": 7.568553626676145e-06,
      "loss": 0.6502,
      "step": 11525
    },
    {
      "epoch": 3.7508132726089785,
      "grad_norm": 1.9283435344696045,
      "learning_rate": 7.550063003020333e-06,
      "loss": 0.6634,
      "step": 11530
    },
    {
      "epoch": 3.752439817826936,
      "grad_norm": 2.002530574798584,
      "learning_rate": 7.531590975922975e-06,
      "loss": 0.6555,
      "step": 11535
    },
    {
      "epoch": 3.7540663630448927,
      "grad_norm": 2.7513580322265625,
      "learning_rate": 7.5131375650698835e-06,
      "loss": 0.629,
      "step": 11540
    },
    {
      "epoch": 3.7556929082628496,
      "grad_norm": 1.9480247497558594,
      "learning_rate": 7.494702790127031e-06,
      "loss": 0.6704,
      "step": 11545
    },
    {
      "epoch": 3.7573194534808065,
      "grad_norm": 1.8581825494766235,
      "learning_rate": 7.476286670740518e-06,
      "loss": 0.6269,
      "step": 11550
    },
    {
      "epoch": 3.758945998698764,
      "grad_norm": 1.9145394563674927,
      "learning_rate": 7.4578892265365825e-06,
      "loss": 0.6637,
      "step": 11555
    },
    {
      "epoch": 3.760572543916721,
      "grad_norm": 1.9342228174209595,
      "learning_rate": 7.439510477121536e-06,
      "loss": 0.6881,
      "step": 11560
    },
    {
      "epoch": 3.762199089134678,
      "grad_norm": 2.0574638843536377,
      "learning_rate": 7.421150442081787e-06,
      "loss": 0.6495,
      "step": 11565
    },
    {
      "epoch": 3.763825634352635,
      "grad_norm": 1.9603220224380493,
      "learning_rate": 7.402809140983799e-06,
      "loss": 0.6534,
      "step": 11570
    },
    {
      "epoch": 3.765452179570592,
      "grad_norm": 2.2657594680786133,
      "learning_rate": 7.384486593374068e-06,
      "loss": 0.6883,
      "step": 11575
    },
    {
      "epoch": 3.767078724788549,
      "grad_norm": 2.123866558074951,
      "learning_rate": 7.366182818779088e-06,
      "loss": 0.6349,
      "step": 11580
    },
    {
      "epoch": 3.7687052700065062,
      "grad_norm": 2.215325117111206,
      "learning_rate": 7.347897836705378e-06,
      "loss": 0.6603,
      "step": 11585
    },
    {
      "epoch": 3.770331815224463,
      "grad_norm": 2.326101064682007,
      "learning_rate": 7.329631666639392e-06,
      "loss": 0.6544,
      "step": 11590
    },
    {
      "epoch": 3.7719583604424205,
      "grad_norm": 2.234374523162842,
      "learning_rate": 7.311384328047563e-06,
      "loss": 0.6667,
      "step": 11595
    },
    {
      "epoch": 3.7735849056603774,
      "grad_norm": 2.0708627700805664,
      "learning_rate": 7.2931558403762535e-06,
      "loss": 0.6392,
      "step": 11600
    },
    {
      "epoch": 3.7752114508783343,
      "grad_norm": 2.1877968311309814,
      "learning_rate": 7.274946223051723e-06,
      "loss": 0.6175,
      "step": 11605
    },
    {
      "epoch": 3.7768379960962912,
      "grad_norm": 1.891137957572937,
      "learning_rate": 7.256755495480141e-06,
      "loss": 0.6699,
      "step": 11610
    },
    {
      "epoch": 3.7784645413142486,
      "grad_norm": 1.8815498352050781,
      "learning_rate": 7.23858367704752e-06,
      "loss": 0.6501,
      "step": 11615
    },
    {
      "epoch": 3.7800910865322055,
      "grad_norm": 2.035649299621582,
      "learning_rate": 7.220430787119742e-06,
      "loss": 0.6496,
      "step": 11620
    },
    {
      "epoch": 3.781717631750163,
      "grad_norm": 2.20412278175354,
      "learning_rate": 7.2022968450425e-06,
      "loss": 0.6358,
      "step": 11625
    },
    {
      "epoch": 3.7833441769681198,
      "grad_norm": 2.2816989421844482,
      "learning_rate": 7.184181870141307e-06,
      "loss": 0.5931,
      "step": 11630
    },
    {
      "epoch": 3.7849707221860767,
      "grad_norm": 2.1493120193481445,
      "learning_rate": 7.16608588172146e-06,
      "loss": 0.6476,
      "step": 11635
    },
    {
      "epoch": 3.7865972674040336,
      "grad_norm": 2.1440515518188477,
      "learning_rate": 7.148008899068029e-06,
      "loss": 0.6531,
      "step": 11640
    },
    {
      "epoch": 3.788223812621991,
      "grad_norm": 2.1033313274383545,
      "learning_rate": 7.129950941445801e-06,
      "loss": 0.6207,
      "step": 11645
    },
    {
      "epoch": 3.789850357839948,
      "grad_norm": 2.0437424182891846,
      "learning_rate": 7.1119120280993295e-06,
      "loss": 0.6411,
      "step": 11650
    },
    {
      "epoch": 3.791476903057905,
      "grad_norm": 2.332766056060791,
      "learning_rate": 7.093892178252831e-06,
      "loss": 0.6495,
      "step": 11655
    },
    {
      "epoch": 3.793103448275862,
      "grad_norm": 2.138261556625366,
      "learning_rate": 7.0758914111102335e-06,
      "loss": 0.6373,
      "step": 11660
    },
    {
      "epoch": 3.794729993493819,
      "grad_norm": 2.130322217941284,
      "learning_rate": 7.057909745855121e-06,
      "loss": 0.6225,
      "step": 11665
    },
    {
      "epoch": 3.796356538711776,
      "grad_norm": 1.8375707864761353,
      "learning_rate": 7.039947201650726e-06,
      "loss": 0.6488,
      "step": 11670
    },
    {
      "epoch": 3.7979830839297333,
      "grad_norm": 2.5891764163970947,
      "learning_rate": 7.022003797639884e-06,
      "loss": 0.6388,
      "step": 11675
    },
    {
      "epoch": 3.79960962914769,
      "grad_norm": 1.9251084327697754,
      "learning_rate": 7.004079552945062e-06,
      "loss": 0.6282,
      "step": 11680
    },
    {
      "epoch": 3.8012361743656475,
      "grad_norm": 2.1112570762634277,
      "learning_rate": 6.98617448666827e-06,
      "loss": 0.6383,
      "step": 11685
    },
    {
      "epoch": 3.8028627195836044,
      "grad_norm": 2.921325922012329,
      "learning_rate": 6.968288617891116e-06,
      "loss": 0.6436,
      "step": 11690
    },
    {
      "epoch": 3.8044892648015614,
      "grad_norm": 2.328986883163452,
      "learning_rate": 6.950421965674736e-06,
      "loss": 0.655,
      "step": 11695
    },
    {
      "epoch": 3.8061158100195187,
      "grad_norm": 1.931818962097168,
      "learning_rate": 6.932574549059789e-06,
      "loss": 0.6615,
      "step": 11700
    },
    {
      "epoch": 3.8077423552374756,
      "grad_norm": 2.4075441360473633,
      "learning_rate": 6.91474638706642e-06,
      "loss": 0.6549,
      "step": 11705
    },
    {
      "epoch": 3.8093689004554325,
      "grad_norm": 2.0539324283599854,
      "learning_rate": 6.8969374986942784e-06,
      "loss": 0.6631,
      "step": 11710
    },
    {
      "epoch": 3.81099544567339,
      "grad_norm": 2.220444440841675,
      "learning_rate": 6.879147902922442e-06,
      "loss": 0.6443,
      "step": 11715
    },
    {
      "epoch": 3.812621990891347,
      "grad_norm": 2.3892598152160645,
      "learning_rate": 6.861377618709466e-06,
      "loss": 0.6351,
      "step": 11720
    },
    {
      "epoch": 3.8142485361093037,
      "grad_norm": 2.2880091667175293,
      "learning_rate": 6.843626664993286e-06,
      "loss": 0.6462,
      "step": 11725
    },
    {
      "epoch": 3.815875081327261,
      "grad_norm": 2.3915302753448486,
      "learning_rate": 6.825895060691273e-06,
      "loss": 0.6386,
      "step": 11730
    },
    {
      "epoch": 3.817501626545218,
      "grad_norm": 2.0178608894348145,
      "learning_rate": 6.808182824700163e-06,
      "loss": 0.6214,
      "step": 11735
    },
    {
      "epoch": 3.819128171763175,
      "grad_norm": 1.9248898029327393,
      "learning_rate": 6.790489975896033e-06,
      "loss": 0.6505,
      "step": 11740
    },
    {
      "epoch": 3.8207547169811322,
      "grad_norm": 1.9928648471832275,
      "learning_rate": 6.7728165331343355e-06,
      "loss": 0.6256,
      "step": 11745
    },
    {
      "epoch": 3.822381262199089,
      "grad_norm": 2.3208022117614746,
      "learning_rate": 6.755162515249799e-06,
      "loss": 0.6679,
      "step": 11750
    },
    {
      "epoch": 3.824007807417046,
      "grad_norm": 2.1799709796905518,
      "learning_rate": 6.737527941056482e-06,
      "loss": 0.6561,
      "step": 11755
    },
    {
      "epoch": 3.8256343526350034,
      "grad_norm": 1.859642744064331,
      "learning_rate": 6.719912829347716e-06,
      "loss": 0.6632,
      "step": 11760
    },
    {
      "epoch": 3.8272608978529603,
      "grad_norm": 1.8891128301620483,
      "learning_rate": 6.702317198896094e-06,
      "loss": 0.6263,
      "step": 11765
    },
    {
      "epoch": 3.828887443070917,
      "grad_norm": 1.9287997484207153,
      "learning_rate": 6.6847410684534235e-06,
      "loss": 0.6463,
      "step": 11770
    },
    {
      "epoch": 3.8305139882888746,
      "grad_norm": 1.865962266921997,
      "learning_rate": 6.667184456750763e-06,
      "loss": 0.6368,
      "step": 11775
    },
    {
      "epoch": 3.8321405335068315,
      "grad_norm": 2.176776647567749,
      "learning_rate": 6.649647382498345e-06,
      "loss": 0.6621,
      "step": 11780
    },
    {
      "epoch": 3.8337670787247884,
      "grad_norm": 2.3224313259124756,
      "learning_rate": 6.632129864385603e-06,
      "loss": 0.6716,
      "step": 11785
    },
    {
      "epoch": 3.8353936239427457,
      "grad_norm": 2.210333824157715,
      "learning_rate": 6.6146319210810935e-06,
      "loss": 0.6798,
      "step": 11790
    },
    {
      "epoch": 3.8370201691607027,
      "grad_norm": 2.4533655643463135,
      "learning_rate": 6.59715357123257e-06,
      "loss": 0.6685,
      "step": 11795
    },
    {
      "epoch": 3.8386467143786596,
      "grad_norm": 2.1346421241760254,
      "learning_rate": 6.579694833466843e-06,
      "loss": 0.6522,
      "step": 11800
    },
    {
      "epoch": 3.840273259596617,
      "grad_norm": 1.7396992444992065,
      "learning_rate": 6.5622557263898685e-06,
      "loss": 0.6218,
      "step": 11805
    },
    {
      "epoch": 3.841899804814574,
      "grad_norm": 2.317690134048462,
      "learning_rate": 6.5448362685866485e-06,
      "loss": 0.6804,
      "step": 11810
    },
    {
      "epoch": 3.8435263500325307,
      "grad_norm": 2.0960299968719482,
      "learning_rate": 6.527436478621271e-06,
      "loss": 0.692,
      "step": 11815
    },
    {
      "epoch": 3.845152895250488,
      "grad_norm": 2.0712311267852783,
      "learning_rate": 6.510056375036841e-06,
      "loss": 0.6534,
      "step": 11820
    },
    {
      "epoch": 3.846779440468445,
      "grad_norm": 2.082179069519043,
      "learning_rate": 6.4926959763555035e-06,
      "loss": 0.6433,
      "step": 11825
    },
    {
      "epoch": 3.8484059856864024,
      "grad_norm": 2.23954439163208,
      "learning_rate": 6.47535530107839e-06,
      "loss": 0.653,
      "step": 11830
    },
    {
      "epoch": 3.8500325309043593,
      "grad_norm": 1.9695318937301636,
      "learning_rate": 6.458034367685628e-06,
      "loss": 0.6232,
      "step": 11835
    },
    {
      "epoch": 3.851659076122316,
      "grad_norm": 2.3786518573760986,
      "learning_rate": 6.440733194636281e-06,
      "loss": 0.6841,
      "step": 11840
    },
    {
      "epoch": 3.853285621340273,
      "grad_norm": 2.1393187046051025,
      "learning_rate": 6.423451800368382e-06,
      "loss": 0.6422,
      "step": 11845
    },
    {
      "epoch": 3.8549121665582304,
      "grad_norm": 1.9716554880142212,
      "learning_rate": 6.406190203298859e-06,
      "loss": 0.6578,
      "step": 11850
    },
    {
      "epoch": 3.8565387117761873,
      "grad_norm": 2.2387914657592773,
      "learning_rate": 6.388948421823563e-06,
      "loss": 0.6485,
      "step": 11855
    },
    {
      "epoch": 3.8581652569941447,
      "grad_norm": 2.0851616859436035,
      "learning_rate": 6.3717264743172134e-06,
      "loss": 0.6537,
      "step": 11860
    },
    {
      "epoch": 3.8597918022121016,
      "grad_norm": 2.284559965133667,
      "learning_rate": 6.3545243791334055e-06,
      "loss": 0.6397,
      "step": 11865
    },
    {
      "epoch": 3.8614183474300585,
      "grad_norm": 2.1620171070098877,
      "learning_rate": 6.337342154604573e-06,
      "loss": 0.6347,
      "step": 11870
    },
    {
      "epoch": 3.8630448926480154,
      "grad_norm": 2.347256898880005,
      "learning_rate": 6.320179819041958e-06,
      "loss": 0.6278,
      "step": 11875
    },
    {
      "epoch": 3.864671437865973,
      "grad_norm": 2.4992942810058594,
      "learning_rate": 6.303037390735634e-06,
      "loss": 0.6876,
      "step": 11880
    },
    {
      "epoch": 3.8662979830839297,
      "grad_norm": 2.0782930850982666,
      "learning_rate": 6.2859148879544305e-06,
      "loss": 0.6422,
      "step": 11885
    },
    {
      "epoch": 3.867924528301887,
      "grad_norm": 1.9557080268859863,
      "learning_rate": 6.268812328945961e-06,
      "loss": 0.6436,
      "step": 11890
    },
    {
      "epoch": 3.869551073519844,
      "grad_norm": 2.2186222076416016,
      "learning_rate": 6.2517297319365814e-06,
      "loss": 0.6487,
      "step": 11895
    },
    {
      "epoch": 3.871177618737801,
      "grad_norm": 2.5874240398406982,
      "learning_rate": 6.234667115131382e-06,
      "loss": 0.6418,
      "step": 11900
    },
    {
      "epoch": 3.8728041639557578,
      "grad_norm": 2.3835012912750244,
      "learning_rate": 6.217624496714134e-06,
      "loss": 0.696,
      "step": 11905
    },
    {
      "epoch": 3.874430709173715,
      "grad_norm": 2.054152488708496,
      "learning_rate": 6.200601894847324e-06,
      "loss": 0.681,
      "step": 11910
    },
    {
      "epoch": 3.876057254391672,
      "grad_norm": 1.9579323530197144,
      "learning_rate": 6.183599327672088e-06,
      "loss": 0.653,
      "step": 11915
    },
    {
      "epoch": 3.8776837996096294,
      "grad_norm": 2.3043148517608643,
      "learning_rate": 6.166616813308221e-06,
      "loss": 0.6709,
      "step": 11920
    },
    {
      "epoch": 3.8793103448275863,
      "grad_norm": 2.012111186981201,
      "learning_rate": 6.149654369854146e-06,
      "loss": 0.6519,
      "step": 11925
    },
    {
      "epoch": 3.880936890045543,
      "grad_norm": 1.9760338068008423,
      "learning_rate": 6.132712015386902e-06,
      "loss": 0.667,
      "step": 11930
    },
    {
      "epoch": 3.8825634352635,
      "grad_norm": 1.9617925882339478,
      "learning_rate": 6.115789767962099e-06,
      "loss": 0.6333,
      "step": 11935
    },
    {
      "epoch": 3.8841899804814575,
      "grad_norm": 1.8881852626800537,
      "learning_rate": 6.098887645613943e-06,
      "loss": 0.6315,
      "step": 11940
    },
    {
      "epoch": 3.8858165256994144,
      "grad_norm": 2.1504695415496826,
      "learning_rate": 6.08200566635517e-06,
      "loss": 0.6492,
      "step": 11945
    },
    {
      "epoch": 3.8874430709173717,
      "grad_norm": 2.1913535594940186,
      "learning_rate": 6.065143848177066e-06,
      "loss": 0.6612,
      "step": 11950
    },
    {
      "epoch": 3.8890696161353286,
      "grad_norm": 1.8416792154312134,
      "learning_rate": 6.048302209049428e-06,
      "loss": 0.6473,
      "step": 11955
    },
    {
      "epoch": 3.8906961613532856,
      "grad_norm": 2.969268798828125,
      "learning_rate": 6.03148076692055e-06,
      "loss": 0.6281,
      "step": 11960
    },
    {
      "epoch": 3.8923227065712425,
      "grad_norm": 2.2109107971191406,
      "learning_rate": 6.014679539717186e-06,
      "loss": 0.6618,
      "step": 11965
    },
    {
      "epoch": 3.8939492517892,
      "grad_norm": 4.5289082527160645,
      "learning_rate": 5.997898545344571e-06,
      "loss": 0.6841,
      "step": 11970
    },
    {
      "epoch": 3.8955757970071567,
      "grad_norm": 1.9927226305007935,
      "learning_rate": 5.981137801686354e-06,
      "loss": 0.652,
      "step": 11975
    },
    {
      "epoch": 3.897202342225114,
      "grad_norm": 2.0749526023864746,
      "learning_rate": 5.9643973266046145e-06,
      "loss": 0.6597,
      "step": 11980
    },
    {
      "epoch": 3.898828887443071,
      "grad_norm": 2.044415235519409,
      "learning_rate": 5.947677137939839e-06,
      "loss": 0.673,
      "step": 11985
    },
    {
      "epoch": 3.900455432661028,
      "grad_norm": 1.8921427726745605,
      "learning_rate": 5.930977253510886e-06,
      "loss": 0.6698,
      "step": 11990
    },
    {
      "epoch": 3.902081977878985,
      "grad_norm": 1.9163424968719482,
      "learning_rate": 5.914297691114967e-06,
      "loss": 0.6368,
      "step": 11995
    },
    {
      "epoch": 3.903708523096942,
      "grad_norm": 2.6371891498565674,
      "learning_rate": 5.897638468527653e-06,
      "loss": 0.6321,
      "step": 12000
    },
    {
      "epoch": 3.905335068314899,
      "grad_norm": 2.0630850791931152,
      "learning_rate": 5.880999603502832e-06,
      "loss": 0.6687,
      "step": 12005
    },
    {
      "epoch": 3.9069616135328564,
      "grad_norm": 1.8964064121246338,
      "learning_rate": 5.864381113772685e-06,
      "loss": 0.6321,
      "step": 12010
    },
    {
      "epoch": 3.9085881587508133,
      "grad_norm": 1.7861818075180054,
      "learning_rate": 5.847783017047698e-06,
      "loss": 0.641,
      "step": 12015
    },
    {
      "epoch": 3.9102147039687702,
      "grad_norm": 1.9545518159866333,
      "learning_rate": 5.831205331016612e-06,
      "loss": 0.6499,
      "step": 12020
    },
    {
      "epoch": 3.911841249186727,
      "grad_norm": 2.174670934677124,
      "learning_rate": 5.814648073346429e-06,
      "loss": 0.6666,
      "step": 12025
    },
    {
      "epoch": 3.9134677944046845,
      "grad_norm": 2.0867092609405518,
      "learning_rate": 5.798111261682357e-06,
      "loss": 0.6656,
      "step": 12030
    },
    {
      "epoch": 3.9150943396226414,
      "grad_norm": 2.1856985092163086,
      "learning_rate": 5.781594913647842e-06,
      "loss": 0.6219,
      "step": 12035
    },
    {
      "epoch": 3.9167208848405988,
      "grad_norm": 2.310839891433716,
      "learning_rate": 5.765099046844491e-06,
      "loss": 0.6511,
      "step": 12040
    },
    {
      "epoch": 3.9183474300585557,
      "grad_norm": 2.0312695503234863,
      "learning_rate": 5.7486236788521135e-06,
      "loss": 0.6478,
      "step": 12045
    },
    {
      "epoch": 3.9199739752765126,
      "grad_norm": 1.90248441696167,
      "learning_rate": 5.7321688272286596e-06,
      "loss": 0.6157,
      "step": 12050
    },
    {
      "epoch": 3.9216005204944695,
      "grad_norm": 1.5778822898864746,
      "learning_rate": 5.715734509510218e-06,
      "loss": 0.6536,
      "step": 12055
    },
    {
      "epoch": 3.923227065712427,
      "grad_norm": 1.9187769889831543,
      "learning_rate": 5.699320743210984e-06,
      "loss": 0.6295,
      "step": 12060
    },
    {
      "epoch": 3.9248536109303838,
      "grad_norm": 2.114119052886963,
      "learning_rate": 5.682927545823272e-06,
      "loss": 0.6452,
      "step": 12065
    },
    {
      "epoch": 3.926480156148341,
      "grad_norm": 1.953281283378601,
      "learning_rate": 5.666554934817447e-06,
      "loss": 0.6088,
      "step": 12070
    },
    {
      "epoch": 3.928106701366298,
      "grad_norm": 1.7641047239303589,
      "learning_rate": 5.650202927641959e-06,
      "loss": 0.6072,
      "step": 12075
    },
    {
      "epoch": 3.929733246584255,
      "grad_norm": 2.1239614486694336,
      "learning_rate": 5.633871541723295e-06,
      "loss": 0.6562,
      "step": 12080
    },
    {
      "epoch": 3.931359791802212,
      "grad_norm": 1.825926423072815,
      "learning_rate": 5.61756079446597e-06,
      "loss": 0.6362,
      "step": 12085
    },
    {
      "epoch": 3.932986337020169,
      "grad_norm": 2.1758463382720947,
      "learning_rate": 5.601270703252481e-06,
      "loss": 0.6146,
      "step": 12090
    },
    {
      "epoch": 3.934612882238126,
      "grad_norm": 3.43782377243042,
      "learning_rate": 5.585001285443345e-06,
      "loss": 0.6418,
      "step": 12095
    },
    {
      "epoch": 3.9362394274560835,
      "grad_norm": 2.19016170501709,
      "learning_rate": 5.5687525583770135e-06,
      "loss": 0.6554,
      "step": 12100
    },
    {
      "epoch": 3.9378659726740404,
      "grad_norm": 2.0171186923980713,
      "learning_rate": 5.55252453936991e-06,
      "loss": 0.6284,
      "step": 12105
    },
    {
      "epoch": 3.9394925178919973,
      "grad_norm": 2.5034661293029785,
      "learning_rate": 5.536317245716391e-06,
      "loss": 0.6363,
      "step": 12110
    },
    {
      "epoch": 3.941119063109954,
      "grad_norm": 2.0403332710266113,
      "learning_rate": 5.520130694688716e-06,
      "loss": 0.6521,
      "step": 12115
    },
    {
      "epoch": 3.9427456083279115,
      "grad_norm": 2.5809988975524902,
      "learning_rate": 5.503964903537037e-06,
      "loss": 0.671,
      "step": 12120
    },
    {
      "epoch": 3.9443721535458685,
      "grad_norm": 2.1316304206848145,
      "learning_rate": 5.487819889489393e-06,
      "loss": 0.673,
      "step": 12125
    },
    {
      "epoch": 3.945998698763826,
      "grad_norm": 2.0063838958740234,
      "learning_rate": 5.471695669751664e-06,
      "loss": 0.6632,
      "step": 12130
    },
    {
      "epoch": 3.9476252439817827,
      "grad_norm": 2.0849430561065674,
      "learning_rate": 5.455592261507589e-06,
      "loss": 0.6336,
      "step": 12135
    },
    {
      "epoch": 3.9492517891997396,
      "grad_norm": 1.7159404754638672,
      "learning_rate": 5.439509681918703e-06,
      "loss": 0.6588,
      "step": 12140
    },
    {
      "epoch": 3.9508783344176965,
      "grad_norm": 2.0891189575195312,
      "learning_rate": 5.42665862639774e-06,
      "loss": 0.6209,
      "step": 12145
    },
    {
      "epoch": 3.952504879635654,
      "grad_norm": 1.9608136415481567,
      "learning_rate": 5.410613581564464e-06,
      "loss": 0.631,
      "step": 12150
    },
    {
      "epoch": 3.954131424853611,
      "grad_norm": 2.2602360248565674,
      "learning_rate": 5.394589413320589e-06,
      "loss": 0.6455,
      "step": 12155
    },
    {
      "epoch": 3.955757970071568,
      "grad_norm": 2.0772197246551514,
      "learning_rate": 5.378586138743203e-06,
      "loss": 0.6371,
      "step": 12160
    },
    {
      "epoch": 3.957384515289525,
      "grad_norm": 2.095057725906372,
      "learning_rate": 5.3626037748871565e-06,
      "loss": 0.6596,
      "step": 12165
    },
    {
      "epoch": 3.959011060507482,
      "grad_norm": 2.1261723041534424,
      "learning_rate": 5.346642338784985e-06,
      "loss": 0.6655,
      "step": 12170
    },
    {
      "epoch": 3.9606376057254393,
      "grad_norm": 2.161123037338257,
      "learning_rate": 5.330701847446962e-06,
      "loss": 0.6528,
      "step": 12175
    },
    {
      "epoch": 3.9622641509433962,
      "grad_norm": 2.0051002502441406,
      "learning_rate": 5.314782317860998e-06,
      "loss": 0.6183,
      "step": 12180
    },
    {
      "epoch": 3.963890696161353,
      "grad_norm": 3.760082721710205,
      "learning_rate": 5.29888376699269e-06,
      "loss": 0.6418,
      "step": 12185
    },
    {
      "epoch": 3.9655172413793105,
      "grad_norm": 2.0798699855804443,
      "learning_rate": 5.2830062117852654e-06,
      "loss": 0.6404,
      "step": 12190
    },
    {
      "epoch": 3.9671437865972674,
      "grad_norm": 1.942769169807434,
      "learning_rate": 5.267149669159588e-06,
      "loss": 0.6642,
      "step": 12195
    },
    {
      "epoch": 3.9687703318152243,
      "grad_norm": 3.208273410797119,
      "learning_rate": 5.251314156014109e-06,
      "loss": 0.6344,
      "step": 12200
    },
    {
      "epoch": 3.9703968770331817,
      "grad_norm": 1.9215033054351807,
      "learning_rate": 5.235499689224885e-06,
      "loss": 0.644,
      "step": 12205
    },
    {
      "epoch": 3.9720234222511386,
      "grad_norm": 2.114232301712036,
      "learning_rate": 5.219706285645545e-06,
      "loss": 0.6734,
      "step": 12210
    },
    {
      "epoch": 3.9736499674690955,
      "grad_norm": 1.9279701709747314,
      "learning_rate": 5.203933962107266e-06,
      "loss": 0.6735,
      "step": 12215
    },
    {
      "epoch": 3.975276512687053,
      "grad_norm": 1.9819085597991943,
      "learning_rate": 5.1881827354187454e-06,
      "loss": 0.6419,
      "step": 12220
    },
    {
      "epoch": 3.9769030579050098,
      "grad_norm": 1.8507076501846313,
      "learning_rate": 5.172452622366228e-06,
      "loss": 0.6434,
      "step": 12225
    },
    {
      "epoch": 3.9785296031229667,
      "grad_norm": 1.9687306880950928,
      "learning_rate": 5.156743639713421e-06,
      "loss": 0.6469,
      "step": 12230
    },
    {
      "epoch": 3.980156148340924,
      "grad_norm": 2.129936933517456,
      "learning_rate": 5.141055804201541e-06,
      "loss": 0.6424,
      "step": 12235
    },
    {
      "epoch": 3.981782693558881,
      "grad_norm": 2.0583765506744385,
      "learning_rate": 5.12538913254926e-06,
      "loss": 0.6132,
      "step": 12240
    },
    {
      "epoch": 3.983409238776838,
      "grad_norm": 2.0079078674316406,
      "learning_rate": 5.109743641452699e-06,
      "loss": 0.6286,
      "step": 12245
    },
    {
      "epoch": 3.985035783994795,
      "grad_norm": 2.323115348815918,
      "learning_rate": 5.094119347585391e-06,
      "loss": 0.6553,
      "step": 12250
    },
    {
      "epoch": 3.986662329212752,
      "grad_norm": 2.1607253551483154,
      "learning_rate": 5.078516267598299e-06,
      "loss": 0.6405,
      "step": 12255
    },
    {
      "epoch": 3.988288874430709,
      "grad_norm": 2.571053981781006,
      "learning_rate": 5.062934418119761e-06,
      "loss": 0.6373,
      "step": 12260
    },
    {
      "epoch": 3.9899154196486664,
      "grad_norm": 2.4664855003356934,
      "learning_rate": 5.047373815755496e-06,
      "loss": 0.6689,
      "step": 12265
    },
    {
      "epoch": 3.9915419648666233,
      "grad_norm": 2.033982038497925,
      "learning_rate": 5.03183447708859e-06,
      "loss": 0.644,
      "step": 12270
    },
    {
      "epoch": 3.9931685100845806,
      "grad_norm": 2.154475212097168,
      "learning_rate": 5.016316418679454e-06,
      "loss": 0.6635,
      "step": 12275
    },
    {
      "epoch": 3.9947950553025375,
      "grad_norm": 3.9255807399749756,
      "learning_rate": 5.000819657065833e-06,
      "loss": 0.661,
      "step": 12280
    },
    {
      "epoch": 3.9964216005204944,
      "grad_norm": 1.9987826347351074,
      "learning_rate": 4.985344208762757e-06,
      "loss": 0.642,
      "step": 12285
    },
    {
      "epoch": 3.9980481457384514,
      "grad_norm": 1.9555718898773193,
      "learning_rate": 4.9698900902625666e-06,
      "loss": 0.6492,
      "step": 12290
    },
    {
      "epoch": 3.9996746909564087,
      "grad_norm": 2.102719783782959,
      "learning_rate": 4.954457318034841e-06,
      "loss": 0.6777,
      "step": 12295
    },
    {
      "epoch": 4.0,
      "eval_f1": 0.8249050225524125,
      "eval_loss": 0.41357421875,
      "eval_precision": 0.825624581488519,
      "eval_recall": 0.8243496742948847,
      "eval_runtime": 1029.9022,
      "eval_samples_per_second": 382.012,
      "eval_steps_per_second": 0.747,
      "step": 12296
    },
    {
      "epoch": 4.001301236174366,
      "grad_norm": 1.9310343265533447,
      "learning_rate": 4.939045908526441e-06,
      "loss": 0.6276,
      "step": 12300
    },
    {
      "epoch": 4.002927781392323,
      "grad_norm": 1.874909520149231,
      "learning_rate": 4.92365587816144e-06,
      "loss": 0.6199,
      "step": 12305
    },
    {
      "epoch": 4.00455432661028,
      "grad_norm": 1.908257246017456,
      "learning_rate": 4.908287243341147e-06,
      "loss": 0.575,
      "step": 12310
    },
    {
      "epoch": 4.006180871828237,
      "grad_norm": 2.0421009063720703,
      "learning_rate": 4.892940020444043e-06,
      "loss": 0.6105,
      "step": 12315
    },
    {
      "epoch": 4.007807417046194,
      "grad_norm": 1.8622931241989136,
      "learning_rate": 4.877614225825816e-06,
      "loss": 0.62,
      "step": 12320
    },
    {
      "epoch": 4.009433962264151,
      "grad_norm": 2.134507179260254,
      "learning_rate": 4.862309875819299e-06,
      "loss": 0.5821,
      "step": 12325
    },
    {
      "epoch": 4.011060507482108,
      "grad_norm": 2.4682130813598633,
      "learning_rate": 4.8470269867344764e-06,
      "loss": 0.616,
      "step": 12330
    },
    {
      "epoch": 4.012687052700065,
      "grad_norm": 2.3341264724731445,
      "learning_rate": 4.831765574858471e-06,
      "loss": 0.6122,
      "step": 12335
    },
    {
      "epoch": 4.014313597918022,
      "grad_norm": 2.212038278579712,
      "learning_rate": 4.816525656455512e-06,
      "loss": 0.6072,
      "step": 12340
    },
    {
      "epoch": 4.015940143135979,
      "grad_norm": 2.357851505279541,
      "learning_rate": 4.801307247766912e-06,
      "loss": 0.5993,
      "step": 12345
    },
    {
      "epoch": 4.017566688353936,
      "grad_norm": 2.0386996269226074,
      "learning_rate": 4.786110365011079e-06,
      "loss": 0.6069,
      "step": 12350
    },
    {
      "epoch": 4.019193233571893,
      "grad_norm": 2.1316967010498047,
      "learning_rate": 4.770935024383458e-06,
      "loss": 0.6128,
      "step": 12355
    },
    {
      "epoch": 4.020819778789851,
      "grad_norm": 2.2749853134155273,
      "learning_rate": 4.755781242056556e-06,
      "loss": 0.6066,
      "step": 12360
    },
    {
      "epoch": 4.022446324007808,
      "grad_norm": 2.4930074214935303,
      "learning_rate": 4.740649034179898e-06,
      "loss": 0.6093,
      "step": 12365
    },
    {
      "epoch": 4.024072869225765,
      "grad_norm": 2.158831834793091,
      "learning_rate": 4.7255384168800235e-06,
      "loss": 0.6197,
      "step": 12370
    },
    {
      "epoch": 4.0256994144437215,
      "grad_norm": 2.402033567428589,
      "learning_rate": 4.7104494062604445e-06,
      "loss": 0.6083,
      "step": 12375
    },
    {
      "epoch": 4.027325959661678,
      "grad_norm": 3.0678963661193848,
      "learning_rate": 4.695382018401673e-06,
      "loss": 0.5798,
      "step": 12380
    },
    {
      "epoch": 4.028952504879635,
      "grad_norm": 2.1048078536987305,
      "learning_rate": 4.680336269361146e-06,
      "loss": 0.6079,
      "step": 12385
    },
    {
      "epoch": 4.030579050097593,
      "grad_norm": 2.014091730117798,
      "learning_rate": 4.665312175173267e-06,
      "loss": 0.6157,
      "step": 12390
    },
    {
      "epoch": 4.03220559531555,
      "grad_norm": 2.081606864929199,
      "learning_rate": 4.650309751849349e-06,
      "loss": 0.6043,
      "step": 12395
    },
    {
      "epoch": 4.033832140533507,
      "grad_norm": 2.3270719051361084,
      "learning_rate": 4.635329015377621e-06,
      "loss": 0.6135,
      "step": 12400
    },
    {
      "epoch": 4.035458685751464,
      "grad_norm": 2.3290863037109375,
      "learning_rate": 4.620369981723174e-06,
      "loss": 0.6371,
      "step": 12405
    },
    {
      "epoch": 4.037085230969421,
      "grad_norm": 2.0165207386016846,
      "learning_rate": 4.605432666828002e-06,
      "loss": 0.621,
      "step": 12410
    },
    {
      "epoch": 4.038711776187378,
      "grad_norm": 1.9335479736328125,
      "learning_rate": 4.590517086610938e-06,
      "loss": 0.588,
      "step": 12415
    },
    {
      "epoch": 4.040338321405335,
      "grad_norm": 1.9899652004241943,
      "learning_rate": 4.575623256967646e-06,
      "loss": 0.6159,
      "step": 12420
    },
    {
      "epoch": 4.041964866623292,
      "grad_norm": 2.0712404251098633,
      "learning_rate": 4.560751193770619e-06,
      "loss": 0.6164,
      "step": 12425
    },
    {
      "epoch": 4.043591411841249,
      "grad_norm": 1.8574680089950562,
      "learning_rate": 4.545900912869156e-06,
      "loss": 0.6229,
      "step": 12430
    },
    {
      "epoch": 4.045217957059206,
      "grad_norm": 1.8692642450332642,
      "learning_rate": 4.531072430089339e-06,
      "loss": 0.5659,
      "step": 12435
    },
    {
      "epoch": 4.046844502277163,
      "grad_norm": 2.001011371612549,
      "learning_rate": 4.516265761234012e-06,
      "loss": 0.6209,
      "step": 12440
    },
    {
      "epoch": 4.04847104749512,
      "grad_norm": 2.2855708599090576,
      "learning_rate": 4.501480922082787e-06,
      "loss": 0.6031,
      "step": 12445
    },
    {
      "epoch": 4.050097592713078,
      "grad_norm": 2.4015791416168213,
      "learning_rate": 4.486717928391993e-06,
      "loss": 0.6152,
      "step": 12450
    },
    {
      "epoch": 4.051724137931035,
      "grad_norm": 1.9134855270385742,
      "learning_rate": 4.471976795894692e-06,
      "loss": 0.5889,
      "step": 12455
    },
    {
      "epoch": 4.053350683148992,
      "grad_norm": 2.1408133506774902,
      "learning_rate": 4.457257540300647e-06,
      "loss": 0.6092,
      "step": 12460
    },
    {
      "epoch": 4.0549772283669485,
      "grad_norm": 2.091780185699463,
      "learning_rate": 4.442560177296307e-06,
      "loss": 0.6578,
      "step": 12465
    },
    {
      "epoch": 4.056603773584905,
      "grad_norm": 2.202208995819092,
      "learning_rate": 4.427884722544776e-06,
      "loss": 0.6078,
      "step": 12470
    },
    {
      "epoch": 4.058230318802862,
      "grad_norm": 1.9250588417053223,
      "learning_rate": 4.413231191685838e-06,
      "loss": 0.583,
      "step": 12475
    },
    {
      "epoch": 4.05985686402082,
      "grad_norm": 2.2674145698547363,
      "learning_rate": 4.398599600335876e-06,
      "loss": 0.6082,
      "step": 12480
    },
    {
      "epoch": 4.061483409238777,
      "grad_norm": 2.1977386474609375,
      "learning_rate": 4.383989964087923e-06,
      "loss": 0.5921,
      "step": 12485
    },
    {
      "epoch": 4.063109954456734,
      "grad_norm": 2.112382650375366,
      "learning_rate": 4.369402298511599e-06,
      "loss": 0.5699,
      "step": 12490
    },
    {
      "epoch": 4.064736499674691,
      "grad_norm": 2.3108229637145996,
      "learning_rate": 4.354836619153124e-06,
      "loss": 0.6073,
      "step": 12495
    },
    {
      "epoch": 4.066363044892648,
      "grad_norm": 2.145864725112915,
      "learning_rate": 4.3402929415352625e-06,
      "loss": 0.6028,
      "step": 12500
    },
    {
      "epoch": 4.067989590110605,
      "grad_norm": 2.464498996734619,
      "learning_rate": 4.325771281157356e-06,
      "loss": 0.633,
      "step": 12505
    },
    {
      "epoch": 4.0696161353285625,
      "grad_norm": 2.521463394165039,
      "learning_rate": 4.311271653495261e-06,
      "loss": 0.6024,
      "step": 12510
    },
    {
      "epoch": 4.071242680546519,
      "grad_norm": 2.3448150157928467,
      "learning_rate": 4.296794074001376e-06,
      "loss": 0.5965,
      "step": 12515
    },
    {
      "epoch": 4.072869225764476,
      "grad_norm": 2.0570671558380127,
      "learning_rate": 4.282338558104573e-06,
      "loss": 0.5832,
      "step": 12520
    },
    {
      "epoch": 4.074495770982433,
      "grad_norm": 2.1091291904449463,
      "learning_rate": 4.267905121210253e-06,
      "loss": 0.6082,
      "step": 12525
    },
    {
      "epoch": 4.07612231620039,
      "grad_norm": 2.690284252166748,
      "learning_rate": 4.2534937787002405e-06,
      "loss": 0.5833,
      "step": 12530
    },
    {
      "epoch": 4.077748861418347,
      "grad_norm": 2.233856439590454,
      "learning_rate": 4.239104545932854e-06,
      "loss": 0.6048,
      "step": 12535
    },
    {
      "epoch": 4.079375406636305,
      "grad_norm": 2.2243893146514893,
      "learning_rate": 4.224737438242815e-06,
      "loss": 0.5929,
      "step": 12540
    },
    {
      "epoch": 4.081001951854262,
      "grad_norm": 2.082343816757202,
      "learning_rate": 4.210392470941288e-06,
      "loss": 0.581,
      "step": 12545
    },
    {
      "epoch": 4.082628497072219,
      "grad_norm": 2.1762783527374268,
      "learning_rate": 4.196069659315846e-06,
      "loss": 0.5979,
      "step": 12550
    },
    {
      "epoch": 4.0842550422901756,
      "grad_norm": 2.3144097328186035,
      "learning_rate": 4.181769018630422e-06,
      "loss": 0.6143,
      "step": 12555
    },
    {
      "epoch": 4.0858815875081325,
      "grad_norm": 2.2833683490753174,
      "learning_rate": 4.167490564125362e-06,
      "loss": 0.5876,
      "step": 12560
    },
    {
      "epoch": 4.087508132726089,
      "grad_norm": 2.1938295364379883,
      "learning_rate": 4.153234311017332e-06,
      "loss": 0.6488,
      "step": 12565
    },
    {
      "epoch": 4.089134677944047,
      "grad_norm": 2.2685635089874268,
      "learning_rate": 4.13900027449936e-06,
      "loss": 0.6016,
      "step": 12570
    },
    {
      "epoch": 4.090761223162004,
      "grad_norm": 2.142134666442871,
      "learning_rate": 4.124788469740784e-06,
      "loss": 0.6206,
      "step": 12575
    },
    {
      "epoch": 4.092387768379961,
      "grad_norm": 2.7002527713775635,
      "learning_rate": 4.110598911887259e-06,
      "loss": 0.6071,
      "step": 12580
    },
    {
      "epoch": 4.094014313597918,
      "grad_norm": 2.357795000076294,
      "learning_rate": 4.096431616060717e-06,
      "loss": 0.609,
      "step": 12585
    },
    {
      "epoch": 4.095640858815875,
      "grad_norm": 2.264958620071411,
      "learning_rate": 4.082286597359395e-06,
      "loss": 0.6337,
      "step": 12590
    },
    {
      "epoch": 4.097267404033832,
      "grad_norm": 2.031507730484009,
      "learning_rate": 4.0681638708577495e-06,
      "loss": 0.5885,
      "step": 12595
    },
    {
      "epoch": 4.0988939492517895,
      "grad_norm": 2.763190507888794,
      "learning_rate": 4.054063451606518e-06,
      "loss": 0.5873,
      "step": 12600
    },
    {
      "epoch": 4.100520494469746,
      "grad_norm": 2.8204002380371094,
      "learning_rate": 4.039985354632633e-06,
      "loss": 0.6147,
      "step": 12605
    },
    {
      "epoch": 4.102147039687703,
      "grad_norm": 2.5044710636138916,
      "learning_rate": 4.025929594939262e-06,
      "loss": 0.6034,
      "step": 12610
    },
    {
      "epoch": 4.10377358490566,
      "grad_norm": 2.4716875553131104,
      "learning_rate": 4.0118961875057485e-06,
      "loss": 0.6319,
      "step": 12615
    },
    {
      "epoch": 4.105400130123617,
      "grad_norm": 1.859110951423645,
      "learning_rate": 3.997885147287628e-06,
      "loss": 0.5791,
      "step": 12620
    },
    {
      "epoch": 4.107026675341574,
      "grad_norm": 1.9830174446105957,
      "learning_rate": 3.983896489216596e-06,
      "loss": 0.5906,
      "step": 12625
    },
    {
      "epoch": 4.108653220559532,
      "grad_norm": 2.4182615280151367,
      "learning_rate": 3.969930228200497e-06,
      "loss": 0.6238,
      "step": 12630
    },
    {
      "epoch": 4.110279765777489,
      "grad_norm": 2.034052610397339,
      "learning_rate": 3.955986379123297e-06,
      "loss": 0.6097,
      "step": 12635
    },
    {
      "epoch": 4.111906310995446,
      "grad_norm": 2.3397281169891357,
      "learning_rate": 3.94206495684509e-06,
      "loss": 0.6108,
      "step": 12640
    },
    {
      "epoch": 4.113532856213403,
      "grad_norm": 2.510648727416992,
      "learning_rate": 3.928165976202058e-06,
      "loss": 0.6007,
      "step": 12645
    },
    {
      "epoch": 4.1151594014313595,
      "grad_norm": 2.819530963897705,
      "learning_rate": 3.914289452006478e-06,
      "loss": 0.6258,
      "step": 12650
    },
    {
      "epoch": 4.116785946649317,
      "grad_norm": 2.195157766342163,
      "learning_rate": 3.900435399046684e-06,
      "loss": 0.5999,
      "step": 12655
    },
    {
      "epoch": 4.118412491867274,
      "grad_norm": 2.176823616027832,
      "learning_rate": 3.88660383208708e-06,
      "loss": 0.6322,
      "step": 12660
    },
    {
      "epoch": 4.120039037085231,
      "grad_norm": 2.2609336376190186,
      "learning_rate": 3.872794765868079e-06,
      "loss": 0.6082,
      "step": 12665
    },
    {
      "epoch": 4.121665582303188,
      "grad_norm": 2.0738308429718018,
      "learning_rate": 3.859008215106141e-06,
      "loss": 0.589,
      "step": 12670
    },
    {
      "epoch": 4.123292127521145,
      "grad_norm": 1.9815113544464111,
      "learning_rate": 3.845244194493711e-06,
      "loss": 0.6007,
      "step": 12675
    },
    {
      "epoch": 4.124918672739102,
      "grad_norm": 2.2510576248168945,
      "learning_rate": 3.831502718699237e-06,
      "loss": 0.6518,
      "step": 12680
    },
    {
      "epoch": 4.126545217957059,
      "grad_norm": 2.5228381156921387,
      "learning_rate": 3.817783802367137e-06,
      "loss": 0.5937,
      "step": 12685
    },
    {
      "epoch": 4.1281717631750166,
      "grad_norm": 2.113708734512329,
      "learning_rate": 3.8040874601177862e-06,
      "loss": 0.6185,
      "step": 12690
    },
    {
      "epoch": 4.1297983083929735,
      "grad_norm": 2.3992342948913574,
      "learning_rate": 3.790413706547505e-06,
      "loss": 0.5963,
      "step": 12695
    },
    {
      "epoch": 4.13142485361093,
      "grad_norm": 2.4300765991210938,
      "learning_rate": 3.7767625562285304e-06,
      "loss": 0.5974,
      "step": 12700
    },
    {
      "epoch": 4.133051398828887,
      "grad_norm": 2.2891955375671387,
      "learning_rate": 3.763134023709031e-06,
      "loss": 0.5951,
      "step": 12705
    },
    {
      "epoch": 4.134677944046844,
      "grad_norm": 2.182823896408081,
      "learning_rate": 3.7495281235130465e-06,
      "loss": 0.5987,
      "step": 12710
    },
    {
      "epoch": 4.136304489264802,
      "grad_norm": 2.214000940322876,
      "learning_rate": 3.7359448701405147e-06,
      "loss": 0.5987,
      "step": 12715
    },
    {
      "epoch": 4.137931034482759,
      "grad_norm": 2.0059731006622314,
      "learning_rate": 3.7223842780672367e-06,
      "loss": 0.6031,
      "step": 12720
    },
    {
      "epoch": 4.139557579700716,
      "grad_norm": 2.5928211212158203,
      "learning_rate": 3.7088463617448637e-06,
      "loss": 0.6273,
      "step": 12725
    },
    {
      "epoch": 4.141184124918673,
      "grad_norm": 2.49120831489563,
      "learning_rate": 3.6953311356008657e-06,
      "loss": 0.6219,
      "step": 12730
    },
    {
      "epoch": 4.14281067013663,
      "grad_norm": 2.263577938079834,
      "learning_rate": 3.6818386140385575e-06,
      "loss": 0.5967,
      "step": 12735
    },
    {
      "epoch": 4.1444372153545865,
      "grad_norm": 2.0892696380615234,
      "learning_rate": 3.6683688114370283e-06,
      "loss": 0.5919,
      "step": 12740
    },
    {
      "epoch": 4.146063760572544,
      "grad_norm": 2.027967929840088,
      "learning_rate": 3.6549217421511795e-06,
      "loss": 0.5978,
      "step": 12745
    },
    {
      "epoch": 4.147690305790501,
      "grad_norm": 2.1417746543884277,
      "learning_rate": 3.6414974205116746e-06,
      "loss": 0.6295,
      "step": 12750
    },
    {
      "epoch": 4.149316851008458,
      "grad_norm": 2.0953240394592285,
      "learning_rate": 3.6280958608249456e-06,
      "loss": 0.6059,
      "step": 12755
    },
    {
      "epoch": 4.150943396226415,
      "grad_norm": 2.216292381286621,
      "learning_rate": 3.614717077373145e-06,
      "loss": 0.6357,
      "step": 12760
    },
    {
      "epoch": 4.152569941444372,
      "grad_norm": 2.3620471954345703,
      "learning_rate": 3.601361084414176e-06,
      "loss": 0.6012,
      "step": 12765
    },
    {
      "epoch": 4.154196486662329,
      "grad_norm": 2.3432223796844482,
      "learning_rate": 3.588027896181631e-06,
      "loss": 0.6326,
      "step": 12770
    },
    {
      "epoch": 4.155823031880287,
      "grad_norm": 2.8889687061309814,
      "learning_rate": 3.5747175268848188e-06,
      "loss": 0.6229,
      "step": 12775
    },
    {
      "epoch": 4.157449577098244,
      "grad_norm": 2.173004627227783,
      "learning_rate": 3.5614299907087227e-06,
      "loss": 0.5868,
      "step": 12780
    },
    {
      "epoch": 4.1590761223162005,
      "grad_norm": 2.1605019569396973,
      "learning_rate": 3.5481653018139995e-06,
      "loss": 0.6088,
      "step": 12785
    },
    {
      "epoch": 4.160702667534157,
      "grad_norm": 2.0582311153411865,
      "learning_rate": 3.534923474336932e-06,
      "loss": 0.5967,
      "step": 12790
    },
    {
      "epoch": 4.162329212752114,
      "grad_norm": 2.2507846355438232,
      "learning_rate": 3.521704522389477e-06,
      "loss": 0.6231,
      "step": 12795
    },
    {
      "epoch": 4.163955757970071,
      "grad_norm": 2.285806894302368,
      "learning_rate": 3.508508460059179e-06,
      "loss": 0.6049,
      "step": 12800
    },
    {
      "epoch": 4.165582303188029,
      "grad_norm": 2.1297316551208496,
      "learning_rate": 3.4953353014092057e-06,
      "loss": 0.5776,
      "step": 12805
    },
    {
      "epoch": 4.167208848405986,
      "grad_norm": 3.20120906829834,
      "learning_rate": 3.4821850604783186e-06,
      "loss": 0.622,
      "step": 12810
    },
    {
      "epoch": 4.168835393623943,
      "grad_norm": 2.3360631465911865,
      "learning_rate": 3.469057751280852e-06,
      "loss": 0.6112,
      "step": 12815
    },
    {
      "epoch": 4.1704619388419,
      "grad_norm": 2.469560146331787,
      "learning_rate": 3.45595338780669e-06,
      "loss": 0.6148,
      "step": 12820
    },
    {
      "epoch": 4.172088484059857,
      "grad_norm": 2.289069890975952,
      "learning_rate": 3.4428719840212814e-06,
      "loss": 0.6116,
      "step": 12825
    },
    {
      "epoch": 4.173715029277814,
      "grad_norm": 2.0820260047912598,
      "learning_rate": 3.4298135538656017e-06,
      "loss": 0.5922,
      "step": 12830
    },
    {
      "epoch": 4.175341574495771,
      "grad_norm": 2.3774619102478027,
      "learning_rate": 3.4167781112561275e-06,
      "loss": 0.6348,
      "step": 12835
    },
    {
      "epoch": 4.176968119713728,
      "grad_norm": 2.2148120403289795,
      "learning_rate": 3.40376567008486e-06,
      "loss": 0.6031,
      "step": 12840
    },
    {
      "epoch": 4.178594664931685,
      "grad_norm": 2.243387222290039,
      "learning_rate": 3.3907762442192735e-06,
      "loss": 0.5964,
      "step": 12845
    },
    {
      "epoch": 4.180221210149642,
      "grad_norm": 2.0168538093566895,
      "learning_rate": 3.3778098475023277e-06,
      "loss": 0.6048,
      "step": 12850
    },
    {
      "epoch": 4.181847755367599,
      "grad_norm": 1.9758124351501465,
      "learning_rate": 3.3648664937524165e-06,
      "loss": 0.5786,
      "step": 12855
    },
    {
      "epoch": 4.183474300585556,
      "grad_norm": 2.3366825580596924,
      "learning_rate": 3.351946196763403e-06,
      "loss": 0.6021,
      "step": 12860
    },
    {
      "epoch": 4.185100845803514,
      "grad_norm": 2.2335920333862305,
      "learning_rate": 3.3390489703045593e-06,
      "loss": 0.5811,
      "step": 12865
    },
    {
      "epoch": 4.186727391021471,
      "grad_norm": 2.1503963470458984,
      "learning_rate": 3.326174828120576e-06,
      "loss": 0.6013,
      "step": 12870
    },
    {
      "epoch": 4.1883539362394275,
      "grad_norm": 2.215834140777588,
      "learning_rate": 3.3133237839315546e-06,
      "loss": 0.5913,
      "step": 12875
    },
    {
      "epoch": 4.189980481457384,
      "grad_norm": 2.486844778060913,
      "learning_rate": 3.3004958514329692e-06,
      "loss": 0.6051,
      "step": 12880
    },
    {
      "epoch": 4.191607026675341,
      "grad_norm": 2.2572789192199707,
      "learning_rate": 3.2876910442956573e-06,
      "loss": 0.5964,
      "step": 12885
    },
    {
      "epoch": 4.193233571893298,
      "grad_norm": 2.700684070587158,
      "learning_rate": 3.27490937616583e-06,
      "loss": 0.6274,
      "step": 12890
    },
    {
      "epoch": 4.194860117111256,
      "grad_norm": 2.216101884841919,
      "learning_rate": 3.262150860665017e-06,
      "loss": 0.5871,
      "step": 12895
    },
    {
      "epoch": 4.196486662329213,
      "grad_norm": 2.355828285217285,
      "learning_rate": 3.2494155113901e-06,
      "loss": 0.6068,
      "step": 12900
    },
    {
      "epoch": 4.19811320754717,
      "grad_norm": 2.384627342224121,
      "learning_rate": 3.2367033419132388e-06,
      "loss": 0.619,
      "step": 12905
    },
    {
      "epoch": 4.199739752765127,
      "grad_norm": 2.2327868938446045,
      "learning_rate": 3.224014365781933e-06,
      "loss": 0.6276,
      "step": 12910
    },
    {
      "epoch": 4.201366297983084,
      "grad_norm": 2.0765182971954346,
      "learning_rate": 3.2113485965189205e-06,
      "loss": 0.6078,
      "step": 12915
    },
    {
      "epoch": 4.202992843201041,
      "grad_norm": 2.2915468215942383,
      "learning_rate": 3.1987060476222435e-06,
      "loss": 0.6005,
      "step": 12920
    },
    {
      "epoch": 4.204619388418998,
      "grad_norm": 2.486490249633789,
      "learning_rate": 3.1860867325651717e-06,
      "loss": 0.6157,
      "step": 12925
    },
    {
      "epoch": 4.206245933636955,
      "grad_norm": 2.343235731124878,
      "learning_rate": 3.173490664796233e-06,
      "loss": 0.5871,
      "step": 12930
    },
    {
      "epoch": 4.207872478854912,
      "grad_norm": 2.1548774242401123,
      "learning_rate": 3.160917857739165e-06,
      "loss": 0.5915,
      "step": 12935
    },
    {
      "epoch": 4.209499024072869,
      "grad_norm": 2.2372446060180664,
      "learning_rate": 3.1483683247929275e-06,
      "loss": 0.5895,
      "step": 12940
    },
    {
      "epoch": 4.211125569290826,
      "grad_norm": 2.7394518852233887,
      "learning_rate": 3.1358420793316744e-06,
      "loss": 0.6046,
      "step": 12945
    },
    {
      "epoch": 4.212752114508783,
      "grad_norm": 2.1555449962615967,
      "learning_rate": 3.1233391347047476e-06,
      "loss": 0.5843,
      "step": 12950
    },
    {
      "epoch": 4.214378659726741,
      "grad_norm": 6.938313961029053,
      "learning_rate": 3.110859504236635e-06,
      "loss": 0.6005,
      "step": 12955
    },
    {
      "epoch": 4.216005204944698,
      "grad_norm": 2.1898958683013916,
      "learning_rate": 3.0984032012270043e-06,
      "loss": 0.6032,
      "step": 12960
    },
    {
      "epoch": 4.217631750162655,
      "grad_norm": 2.1506869792938232,
      "learning_rate": 3.085970238950653e-06,
      "loss": 0.5854,
      "step": 12965
    },
    {
      "epoch": 4.2192582953806115,
      "grad_norm": 2.2408323287963867,
      "learning_rate": 3.0735606306574875e-06,
      "loss": 0.6201,
      "step": 12970
    },
    {
      "epoch": 4.220884840598568,
      "grad_norm": 2.4407107830047607,
      "learning_rate": 3.0611743895725686e-06,
      "loss": 0.5857,
      "step": 12975
    },
    {
      "epoch": 4.222511385816525,
      "grad_norm": 2.288879871368408,
      "learning_rate": 3.048811528896006e-06,
      "loss": 0.6074,
      "step": 12980
    },
    {
      "epoch": 4.224137931034483,
      "grad_norm": 2.333875894546509,
      "learning_rate": 3.036472061803025e-06,
      "loss": 0.6236,
      "step": 12985
    },
    {
      "epoch": 4.22576447625244,
      "grad_norm": 2.0758614540100098,
      "learning_rate": 3.024156001443901e-06,
      "loss": 0.5939,
      "step": 12990
    },
    {
      "epoch": 4.227391021470397,
      "grad_norm": 17.28084945678711,
      "learning_rate": 3.011863360943984e-06,
      "loss": 0.5874,
      "step": 12995
    },
    {
      "epoch": 4.229017566688354,
      "grad_norm": 2.7625977993011475,
      "learning_rate": 2.99959415340364e-06,
      "loss": 0.6448,
      "step": 13000
    },
    {
      "epoch": 4.230644111906311,
      "grad_norm": 2.0970189571380615,
      "learning_rate": 2.987348391898284e-06,
      "loss": 0.5858,
      "step": 13005
    },
    {
      "epoch": 4.232270657124268,
      "grad_norm": 2.935084104537964,
      "learning_rate": 2.9751260894783362e-06,
      "loss": 0.6077,
      "step": 13010
    },
    {
      "epoch": 4.233897202342225,
      "grad_norm": 2.0353829860687256,
      "learning_rate": 2.962927259169221e-06,
      "loss": 0.581,
      "step": 13015
    },
    {
      "epoch": 4.235523747560182,
      "grad_norm": 2.1475942134857178,
      "learning_rate": 2.9507519139713364e-06,
      "loss": 0.578,
      "step": 13020
    },
    {
      "epoch": 4.237150292778139,
      "grad_norm": 2.0711092948913574,
      "learning_rate": 2.9386000668600698e-06,
      "loss": 0.5895,
      "step": 13025
    },
    {
      "epoch": 4.238776837996096,
      "grad_norm": 2.343844413757324,
      "learning_rate": 2.926471730785743e-06,
      "loss": 0.6123,
      "step": 13030
    },
    {
      "epoch": 4.240403383214053,
      "grad_norm": 2.2693004608154297,
      "learning_rate": 2.9143669186736445e-06,
      "loss": 0.5798,
      "step": 13035
    },
    {
      "epoch": 4.24202992843201,
      "grad_norm": 2.457479238510132,
      "learning_rate": 2.9022856434239796e-06,
      "loss": 0.6087,
      "step": 13040
    },
    {
      "epoch": 4.243656473649968,
      "grad_norm": 2.129124164581299,
      "learning_rate": 2.8902279179118837e-06,
      "loss": 0.6415,
      "step": 13045
    },
    {
      "epoch": 4.245283018867925,
      "grad_norm": 2.264616012573242,
      "learning_rate": 2.878193754987374e-06,
      "loss": 0.5968,
      "step": 13050
    },
    {
      "epoch": 4.246909564085882,
      "grad_norm": 2.3036763668060303,
      "learning_rate": 2.866183167475378e-06,
      "loss": 0.5908,
      "step": 13055
    },
    {
      "epoch": 4.2485361093038385,
      "grad_norm": 2.0123422145843506,
      "learning_rate": 2.8541961681756795e-06,
      "loss": 0.6054,
      "step": 13060
    },
    {
      "epoch": 4.250162654521795,
      "grad_norm": 2.1743271350860596,
      "learning_rate": 2.8422327698629405e-06,
      "loss": 0.6075,
      "step": 13065
    },
    {
      "epoch": 4.251789199739752,
      "grad_norm": 2.558833122253418,
      "learning_rate": 2.8302929852866644e-06,
      "loss": 0.5922,
      "step": 13070
    },
    {
      "epoch": 4.25341574495771,
      "grad_norm": 2.240840435028076,
      "learning_rate": 2.818376827171193e-06,
      "loss": 0.6121,
      "step": 13075
    },
    {
      "epoch": 4.255042290175667,
      "grad_norm": 2.0290110111236572,
      "learning_rate": 2.8064843082156787e-06,
      "loss": 0.5783,
      "step": 13080
    },
    {
      "epoch": 4.256668835393624,
      "grad_norm": 2.0330371856689453,
      "learning_rate": 2.794615441094095e-06,
      "loss": 0.5582,
      "step": 13085
    },
    {
      "epoch": 4.258295380611581,
      "grad_norm": 2.271972894668579,
      "learning_rate": 2.782770238455193e-06,
      "loss": 0.6019,
      "step": 13090
    },
    {
      "epoch": 4.259921925829538,
      "grad_norm": 3.2765743732452393,
      "learning_rate": 2.770948712922522e-06,
      "loss": 0.59,
      "step": 13095
    },
    {
      "epoch": 4.261548471047496,
      "grad_norm": 2.7162623405456543,
      "learning_rate": 2.759150877094388e-06,
      "loss": 0.5984,
      "step": 13100
    },
    {
      "epoch": 4.2631750162654525,
      "grad_norm": 2.14172625541687,
      "learning_rate": 2.747376743543853e-06,
      "loss": 0.5702,
      "step": 13105
    },
    {
      "epoch": 4.264801561483409,
      "grad_norm": 2.3000221252441406,
      "learning_rate": 2.735626324818727e-06,
      "loss": 0.5908,
      "step": 13110
    },
    {
      "epoch": 4.266428106701366,
      "grad_norm": 2.5657973289489746,
      "learning_rate": 2.7238996334415262e-06,
      "loss": 0.6329,
      "step": 13115
    },
    {
      "epoch": 4.268054651919323,
      "grad_norm": 2.254990816116333,
      "learning_rate": 2.712196681909507e-06,
      "loss": 0.6069,
      "step": 13120
    },
    {
      "epoch": 4.26968119713728,
      "grad_norm": 2.045351505279541,
      "learning_rate": 2.7005174826946004e-06,
      "loss": 0.5857,
      "step": 13125
    },
    {
      "epoch": 4.271307742355237,
      "grad_norm": 2.3086118698120117,
      "learning_rate": 2.688862048243443e-06,
      "loss": 0.6122,
      "step": 13130
    },
    {
      "epoch": 4.272934287573195,
      "grad_norm": 2.1348495483398438,
      "learning_rate": 2.6772303909773366e-06,
      "loss": 0.5837,
      "step": 13135
    },
    {
      "epoch": 4.274560832791152,
      "grad_norm": 2.0408337116241455,
      "learning_rate": 2.6656225232922492e-06,
      "loss": 0.6114,
      "step": 13140
    },
    {
      "epoch": 4.276187378009109,
      "grad_norm": 2.258314371109009,
      "learning_rate": 2.6540384575587885e-06,
      "loss": 0.5926,
      "step": 13145
    },
    {
      "epoch": 4.2778139232270656,
      "grad_norm": 2.1348114013671875,
      "learning_rate": 2.642478206122201e-06,
      "loss": 0.6269,
      "step": 13150
    },
    {
      "epoch": 4.2794404684450225,
      "grad_norm": 2.749288558959961,
      "learning_rate": 2.6309417813023513e-06,
      "loss": 0.5978,
      "step": 13155
    },
    {
      "epoch": 4.28106701366298,
      "grad_norm": 2.077808380126953,
      "learning_rate": 2.619429195393713e-06,
      "loss": 0.6157,
      "step": 13160
    },
    {
      "epoch": 4.282693558880937,
      "grad_norm": 2.2650487422943115,
      "learning_rate": 2.607940460665359e-06,
      "loss": 0.6376,
      "step": 13165
    },
    {
      "epoch": 4.284320104098894,
      "grad_norm": 2.7509374618530273,
      "learning_rate": 2.5964755893609414e-06,
      "loss": 0.6037,
      "step": 13170
    },
    {
      "epoch": 4.285946649316851,
      "grad_norm": 3.157712697982788,
      "learning_rate": 2.585034593698668e-06,
      "loss": 0.6096,
      "step": 13175
    },
    {
      "epoch": 4.287573194534808,
      "grad_norm": 2.2161614894866943,
      "learning_rate": 2.5736174858713267e-06,
      "loss": 0.6009,
      "step": 13180
    },
    {
      "epoch": 4.289199739752765,
      "grad_norm": 2.189051389694214,
      "learning_rate": 2.5622242780462243e-06,
      "loss": 0.6001,
      "step": 13185
    },
    {
      "epoch": 4.290826284970722,
      "grad_norm": 2.3055875301361084,
      "learning_rate": 2.5508549823652114e-06,
      "loss": 0.6343,
      "step": 13190
    },
    {
      "epoch": 4.2924528301886795,
      "grad_norm": 2.214721918106079,
      "learning_rate": 2.5395096109446488e-06,
      "loss": 0.6182,
      "step": 13195
    },
    {
      "epoch": 4.294079375406636,
      "grad_norm": 2.2350032329559326,
      "learning_rate": 2.528188175875412e-06,
      "loss": 0.6032,
      "step": 13200
    },
    {
      "epoch": 4.295705920624593,
      "grad_norm": 2.064831018447876,
      "learning_rate": 2.516890689222845e-06,
      "loss": 0.6392,
      "step": 13205
    },
    {
      "epoch": 4.29733246584255,
      "grad_norm": 2.1273586750030518,
      "learning_rate": 2.5056171630267937e-06,
      "loss": 0.6279,
      "step": 13210
    },
    {
      "epoch": 4.298959011060507,
      "grad_norm": 2.605733633041382,
      "learning_rate": 2.4943676093015513e-06,
      "loss": 0.6033,
      "step": 13215
    },
    {
      "epoch": 4.300585556278465,
      "grad_norm": 2.7585792541503906,
      "learning_rate": 2.483142040035874e-06,
      "loss": 0.6128,
      "step": 13220
    },
    {
      "epoch": 4.302212101496422,
      "grad_norm": 2.709352731704712,
      "learning_rate": 2.471940467192957e-06,
      "loss": 0.5904,
      "step": 13225
    },
    {
      "epoch": 4.303838646714379,
      "grad_norm": 2.4799320697784424,
      "learning_rate": 2.4607629027104147e-06,
      "loss": 0.6199,
      "step": 13230
    },
    {
      "epoch": 4.305465191932336,
      "grad_norm": 2.359222412109375,
      "learning_rate": 2.449609358500288e-06,
      "loss": 0.5858,
      "step": 13235
    },
    {
      "epoch": 4.307091737150293,
      "grad_norm": 2.0881259441375732,
      "learning_rate": 2.4384798464490016e-06,
      "loss": 0.5912,
      "step": 13240
    },
    {
      "epoch": 4.3087182823682495,
      "grad_norm": 1.9966338872909546,
      "learning_rate": 2.427374378417388e-06,
      "loss": 0.592,
      "step": 13245
    },
    {
      "epoch": 4.310344827586207,
      "grad_norm": 2.474705457687378,
      "learning_rate": 2.416292966240641e-06,
      "loss": 0.6016,
      "step": 13250
    },
    {
      "epoch": 4.311971372804164,
      "grad_norm": 3.0268406867980957,
      "learning_rate": 2.405235621728322e-06,
      "loss": 0.6061,
      "step": 13255
    },
    {
      "epoch": 4.313597918022121,
      "grad_norm": 4.327364444732666,
      "learning_rate": 2.394202356664349e-06,
      "loss": 0.6119,
      "step": 13260
    },
    {
      "epoch": 4.315224463240078,
      "grad_norm": 2.194871187210083,
      "learning_rate": 2.383193182806978e-06,
      "loss": 0.6134,
      "step": 13265
    },
    {
      "epoch": 4.316851008458035,
      "grad_norm": 2.07330322265625,
      "learning_rate": 2.3722081118887767e-06,
      "loss": 0.6216,
      "step": 13270
    },
    {
      "epoch": 4.318477553675992,
      "grad_norm": 2.8549962043762207,
      "learning_rate": 2.3612471556166442e-06,
      "loss": 0.6119,
      "step": 13275
    },
    {
      "epoch": 4.32010409889395,
      "grad_norm": 2.333425283432007,
      "learning_rate": 2.3503103256717673e-06,
      "loss": 0.6029,
      "step": 13280
    },
    {
      "epoch": 4.3217306441119065,
      "grad_norm": 2.603220224380493,
      "learning_rate": 2.3393976337096334e-06,
      "loss": 0.6238,
      "step": 13285
    },
    {
      "epoch": 4.3233571893298635,
      "grad_norm": 2.699075222015381,
      "learning_rate": 2.328509091359984e-06,
      "loss": 0.6103,
      "step": 13290
    },
    {
      "epoch": 4.32498373454782,
      "grad_norm": 2.9168403148651123,
      "learning_rate": 2.3176447102268602e-06,
      "loss": 0.5861,
      "step": 13295
    },
    {
      "epoch": 4.326610279765777,
      "grad_norm": 2.3375420570373535,
      "learning_rate": 2.3068045018885153e-06,
      "loss": 0.623,
      "step": 13300
    },
    {
      "epoch": 4.328236824983734,
      "grad_norm": 2.315999746322632,
      "learning_rate": 2.2959884778974735e-06,
      "loss": 0.5896,
      "step": 13305
    },
    {
      "epoch": 4.329863370201692,
      "grad_norm": 2.135993242263794,
      "learning_rate": 2.285196649780455e-06,
      "loss": 0.6104,
      "step": 13310
    },
    {
      "epoch": 4.331489915419649,
      "grad_norm": 2.2192249298095703,
      "learning_rate": 2.2744290290384247e-06,
      "loss": 0.607,
      "step": 13315
    },
    {
      "epoch": 4.333116460637606,
      "grad_norm": 2.0429885387420654,
      "learning_rate": 2.2636856271465194e-06,
      "loss": 0.6061,
      "step": 13320
    },
    {
      "epoch": 4.334743005855563,
      "grad_norm": 2.8340260982513428,
      "learning_rate": 2.252966455554101e-06,
      "loss": 0.6189,
      "step": 13325
    },
    {
      "epoch": 4.33636955107352,
      "grad_norm": 2.315480947494507,
      "learning_rate": 2.242271525684672e-06,
      "loss": 0.5966,
      "step": 13330
    },
    {
      "epoch": 4.3379960962914765,
      "grad_norm": 3.0812127590179443,
      "learning_rate": 2.2316008489359304e-06,
      "loss": 0.63,
      "step": 13335
    },
    {
      "epoch": 4.339622641509434,
      "grad_norm": 2.2564315795898438,
      "learning_rate": 2.2209544366797066e-06,
      "loss": 0.5944,
      "step": 13340
    },
    {
      "epoch": 4.341249186727391,
      "grad_norm": 2.032088279724121,
      "learning_rate": 2.2103323002619857e-06,
      "loss": 0.5801,
      "step": 13345
    },
    {
      "epoch": 4.342875731945348,
      "grad_norm": 2.143538236618042,
      "learning_rate": 2.199734451002869e-06,
      "loss": 0.6006,
      "step": 13350
    },
    {
      "epoch": 4.344502277163305,
      "grad_norm": 2.0386180877685547,
      "learning_rate": 2.189160900196585e-06,
      "loss": 0.6095,
      "step": 13355
    },
    {
      "epoch": 4.346128822381262,
      "grad_norm": 2.3832762241363525,
      "learning_rate": 2.178611659111468e-06,
      "loss": 0.5936,
      "step": 13360
    },
    {
      "epoch": 4.347755367599219,
      "grad_norm": 2.4224894046783447,
      "learning_rate": 2.1680867389899355e-06,
      "loss": 0.6062,
      "step": 13365
    },
    {
      "epoch": 4.349381912817177,
      "grad_norm": 2.3058247566223145,
      "learning_rate": 2.1575861510485017e-06,
      "loss": 0.6307,
      "step": 13370
    },
    {
      "epoch": 4.351008458035134,
      "grad_norm": 2.045710563659668,
      "learning_rate": 2.147109906477726e-06,
      "loss": 0.6001,
      "step": 13375
    },
    {
      "epoch": 4.3526350032530905,
      "grad_norm": 2.2343456745147705,
      "learning_rate": 2.136658016442253e-06,
      "loss": 0.6363,
      "step": 13380
    },
    {
      "epoch": 4.354261548471047,
      "grad_norm": 2.8338370323181152,
      "learning_rate": 2.126230492080744e-06,
      "loss": 0.624,
      "step": 13385
    },
    {
      "epoch": 4.355888093689004,
      "grad_norm": 2.3945822715759277,
      "learning_rate": 2.1158273445059135e-06,
      "loss": 0.6151,
      "step": 13390
    },
    {
      "epoch": 4.357514638906961,
      "grad_norm": 2.234694004058838,
      "learning_rate": 2.1054485848044952e-06,
      "loss": 0.6072,
      "step": 13395
    },
    {
      "epoch": 4.359141184124919,
      "grad_norm": 2.2890405654907227,
      "learning_rate": 2.095094224037228e-06,
      "loss": 0.6339,
      "step": 13400
    },
    {
      "epoch": 4.360767729342876,
      "grad_norm": 2.266394853591919,
      "learning_rate": 2.0847642732388457e-06,
      "loss": 0.597,
      "step": 13405
    },
    {
      "epoch": 4.362394274560833,
      "grad_norm": 2.2408626079559326,
      "learning_rate": 2.0744587434180757e-06,
      "loss": 0.6224,
      "step": 13410
    },
    {
      "epoch": 4.36402081977879,
      "grad_norm": 2.1054515838623047,
      "learning_rate": 2.0641776455576105e-06,
      "loss": 0.6094,
      "step": 13415
    },
    {
      "epoch": 4.365647364996747,
      "grad_norm": 2.2183995246887207,
      "learning_rate": 2.0539209906141167e-06,
      "loss": 0.5976,
      "step": 13420
    },
    {
      "epoch": 4.367273910214704,
      "grad_norm": 2.342390775680542,
      "learning_rate": 2.0436887895182e-06,
      "loss": 0.5971,
      "step": 13425
    },
    {
      "epoch": 4.368900455432661,
      "grad_norm": 2.0689449310302734,
      "learning_rate": 2.0334810531744213e-06,
      "loss": 0.5908,
      "step": 13430
    },
    {
      "epoch": 4.370527000650618,
      "grad_norm": 2.089670181274414,
      "learning_rate": 2.0232977924612457e-06,
      "loss": 0.5898,
      "step": 13435
    },
    {
      "epoch": 4.372153545868575,
      "grad_norm": 2.30362606048584,
      "learning_rate": 2.0131390182310805e-06,
      "loss": 0.606,
      "step": 13440
    },
    {
      "epoch": 4.373780091086532,
      "grad_norm": 2.311826467514038,
      "learning_rate": 2.003004741310216e-06,
      "loss": 0.6044,
      "step": 13445
    },
    {
      "epoch": 4.375406636304489,
      "grad_norm": 2.338933229446411,
      "learning_rate": 1.992894972498846e-06,
      "loss": 0.6068,
      "step": 13450
    },
    {
      "epoch": 4.377033181522446,
      "grad_norm": 2.32281756401062,
      "learning_rate": 1.982809722571047e-06,
      "loss": 0.6066,
      "step": 13455
    },
    {
      "epoch": 4.378659726740404,
      "grad_norm": 2.3082902431488037,
      "learning_rate": 1.972749002274765e-06,
      "loss": 0.6101,
      "step": 13460
    },
    {
      "epoch": 4.380286271958361,
      "grad_norm": 2.083157777786255,
      "learning_rate": 1.9627128223317942e-06,
      "loss": 0.6151,
      "step": 13465
    },
    {
      "epoch": 4.3819128171763175,
      "grad_norm": 2.093111753463745,
      "learning_rate": 1.952701193437792e-06,
      "loss": 0.6313,
      "step": 13470
    },
    {
      "epoch": 4.383539362394274,
      "grad_norm": 2.210050582885742,
      "learning_rate": 1.942714126262238e-06,
      "loss": 0.6016,
      "step": 13475
    },
    {
      "epoch": 4.385165907612231,
      "grad_norm": 2.0351722240448,
      "learning_rate": 1.93275163144844e-06,
      "loss": 0.6093,
      "step": 13480
    },
    {
      "epoch": 4.386792452830189,
      "grad_norm": 2.6686933040618896,
      "learning_rate": 1.9228137196135254e-06,
      "loss": 0.6177,
      "step": 13485
    },
    {
      "epoch": 4.388418998048146,
      "grad_norm": 2.5283005237579346,
      "learning_rate": 1.912900401348422e-06,
      "loss": 0.6184,
      "step": 13490
    },
    {
      "epoch": 4.390045543266103,
      "grad_norm": 1.9730428457260132,
      "learning_rate": 1.9030116872178316e-06,
      "loss": 0.5938,
      "step": 13495
    },
    {
      "epoch": 4.39167208848406,
      "grad_norm": 2.1344470977783203,
      "learning_rate": 1.8931475877602579e-06,
      "loss": 0.6306,
      "step": 13500
    },
    {
      "epoch": 4.393298633702017,
      "grad_norm": 1.9898133277893066,
      "learning_rate": 1.8833081134879637e-06,
      "loss": 0.611,
      "step": 13505
    },
    {
      "epoch": 4.394925178919974,
      "grad_norm": 2.810833215713501,
      "learning_rate": 1.8734932748869588e-06,
      "loss": 0.5925,
      "step": 13510
    },
    {
      "epoch": 4.396551724137931,
      "grad_norm": 2.492091655731201,
      "learning_rate": 1.8637030824170121e-06,
      "loss": 0.5901,
      "step": 13515
    },
    {
      "epoch": 4.398178269355888,
      "grad_norm": 2.828150749206543,
      "learning_rate": 1.8539375465116193e-06,
      "loss": 0.6431,
      "step": 13520
    },
    {
      "epoch": 4.399804814573845,
      "grad_norm": 2.4096462726593018,
      "learning_rate": 1.8441966775780112e-06,
      "loss": 0.5792,
      "step": 13525
    },
    {
      "epoch": 4.401431359791802,
      "grad_norm": 2.353559732437134,
      "learning_rate": 1.834480485997106e-06,
      "loss": 0.6104,
      "step": 13530
    },
    {
      "epoch": 4.403057905009759,
      "grad_norm": 1.9907680749893188,
      "learning_rate": 1.8247889821235543e-06,
      "loss": 0.5964,
      "step": 13535
    },
    {
      "epoch": 4.404684450227716,
      "grad_norm": 2.5092782974243164,
      "learning_rate": 1.815122176285669e-06,
      "loss": 0.6118,
      "step": 13540
    },
    {
      "epoch": 4.406310995445674,
      "grad_norm": 2.1358978748321533,
      "learning_rate": 1.8054800787854569e-06,
      "loss": 0.5925,
      "step": 13545
    },
    {
      "epoch": 4.407937540663631,
      "grad_norm": 2.424093008041382,
      "learning_rate": 1.7958626998985928e-06,
      "loss": 0.6362,
      "step": 13550
    },
    {
      "epoch": 4.409564085881588,
      "grad_norm": 2.92620849609375,
      "learning_rate": 1.7862700498744085e-06,
      "loss": 0.5897,
      "step": 13555
    },
    {
      "epoch": 4.411190631099545,
      "grad_norm": 2.4644806385040283,
      "learning_rate": 1.7767021389358706e-06,
      "loss": 0.6257,
      "step": 13560
    },
    {
      "epoch": 4.4128171763175015,
      "grad_norm": 2.018040657043457,
      "learning_rate": 1.767158977279601e-06,
      "loss": 0.6118,
      "step": 13565
    },
    {
      "epoch": 4.414443721535458,
      "grad_norm": 2.133514642715454,
      "learning_rate": 1.7576405750758224e-06,
      "loss": 0.5955,
      "step": 13570
    },
    {
      "epoch": 4.416070266753415,
      "grad_norm": 2.5312135219573975,
      "learning_rate": 1.74814694246839e-06,
      "loss": 0.5924,
      "step": 13575
    },
    {
      "epoch": 4.417696811971373,
      "grad_norm": 2.0397260189056396,
      "learning_rate": 1.7386780895747578e-06,
      "loss": 0.6171,
      "step": 13580
    },
    {
      "epoch": 4.41932335718933,
      "grad_norm": 1.9202146530151367,
      "learning_rate": 1.729234026485968e-06,
      "loss": 0.5873,
      "step": 13585
    },
    {
      "epoch": 4.420949902407287,
      "grad_norm": 3.7298386096954346,
      "learning_rate": 1.7198147632666416e-06,
      "loss": 0.587,
      "step": 13590
    },
    {
      "epoch": 4.422576447625244,
      "grad_norm": 2.167330265045166,
      "learning_rate": 1.7104203099549827e-06,
      "loss": 0.5749,
      "step": 13595
    },
    {
      "epoch": 4.424202992843201,
      "grad_norm": 2.2834017276763916,
      "learning_rate": 1.701050676562735e-06,
      "loss": 0.5965,
      "step": 13600
    },
    {
      "epoch": 4.4258295380611585,
      "grad_norm": 2.037825107574463,
      "learning_rate": 1.691705873075211e-06,
      "loss": 0.5955,
      "step": 13605
    },
    {
      "epoch": 4.427456083279115,
      "grad_norm": 2.448396682739258,
      "learning_rate": 1.6823859094512507e-06,
      "loss": 0.624,
      "step": 13610
    },
    {
      "epoch": 4.429082628497072,
      "grad_norm": 2.062052011489868,
      "learning_rate": 1.6730907956232306e-06,
      "loss": 0.5848,
      "step": 13615
    },
    {
      "epoch": 4.430709173715029,
      "grad_norm": 2.03035306930542,
      "learning_rate": 1.6638205414970298e-06,
      "loss": 0.5875,
      "step": 13620
    },
    {
      "epoch": 4.432335718932986,
      "grad_norm": 2.0660645961761475,
      "learning_rate": 1.654575156952054e-06,
      "loss": 0.6194,
      "step": 13625
    },
    {
      "epoch": 4.433962264150943,
      "grad_norm": 2.2230019569396973,
      "learning_rate": 1.6453546518411855e-06,
      "loss": 0.601,
      "step": 13630
    },
    {
      "epoch": 4.4355888093689,
      "grad_norm": 2.1817357540130615,
      "learning_rate": 1.636159035990803e-06,
      "loss": 0.613,
      "step": 13635
    },
    {
      "epoch": 4.437215354586858,
      "grad_norm": 2.207064390182495,
      "learning_rate": 1.6269883192007618e-06,
      "loss": 0.5996,
      "step": 13640
    },
    {
      "epoch": 4.438841899804815,
      "grad_norm": 2.2545690536499023,
      "learning_rate": 1.6178425112443774e-06,
      "loss": 0.6121,
      "step": 13645
    },
    {
      "epoch": 4.440468445022772,
      "grad_norm": 2.3414602279663086,
      "learning_rate": 1.608721621868428e-06,
      "loss": 0.6295,
      "step": 13650
    },
    {
      "epoch": 4.4420949902407285,
      "grad_norm": 2.3920414447784424,
      "learning_rate": 1.5996256607931193e-06,
      "loss": 0.5941,
      "step": 13655
    },
    {
      "epoch": 4.443721535458685,
      "grad_norm": 2.2873120307922363,
      "learning_rate": 1.5905546377121077e-06,
      "loss": 0.6179,
      "step": 13660
    },
    {
      "epoch": 4.445348080676643,
      "grad_norm": 2.398427963256836,
      "learning_rate": 1.583315781100353e-06,
      "loss": 0.5979,
      "step": 13665
    },
    {
      "epoch": 4.4469746258946,
      "grad_norm": 2.2126176357269287,
      "learning_rate": 1.5742896707522242e-06,
      "loss": 0.6173,
      "step": 13670
    },
    {
      "epoch": 4.448601171112557,
      "grad_norm": 2.4789159297943115,
      "learning_rate": 1.5652885253991944e-06,
      "loss": 0.6519,
      "step": 13675
    },
    {
      "epoch": 4.450227716330514,
      "grad_norm": 2.2594988346099854,
      "learning_rate": 1.5563123546338572e-06,
      "loss": 0.5882,
      "step": 13680
    },
    {
      "epoch": 4.451854261548471,
      "grad_norm": 2.307455062866211,
      "learning_rate": 1.5473611680222045e-06,
      "loss": 0.6314,
      "step": 13685
    },
    {
      "epoch": 4.453480806766428,
      "grad_norm": 2.2615737915039062,
      "learning_rate": 1.5384349751035948e-06,
      "loss": 0.6094,
      "step": 13690
    },
    {
      "epoch": 4.455107351984386,
      "grad_norm": 2.9631905555725098,
      "learning_rate": 1.5295337853907604e-06,
      "loss": 0.589,
      "step": 13695
    },
    {
      "epoch": 4.4567338972023425,
      "grad_norm": 2.044833183288574,
      "learning_rate": 1.5206576083697687e-06,
      "loss": 0.5841,
      "step": 13700
    },
    {
      "epoch": 4.458360442420299,
      "grad_norm": 2.2695491313934326,
      "learning_rate": 1.5118064535000614e-06,
      "loss": 0.6207,
      "step": 13705
    },
    {
      "epoch": 4.459986987638256,
      "grad_norm": 1.9073365926742554,
      "learning_rate": 1.502980330214379e-06,
      "loss": 0.5814,
      "step": 13710
    },
    {
      "epoch": 4.461613532856213,
      "grad_norm": 2.171907424926758,
      "learning_rate": 1.4941792479188171e-06,
      "loss": 0.6072,
      "step": 13715
    },
    {
      "epoch": 4.46324007807417,
      "grad_norm": 2.340442657470703,
      "learning_rate": 1.4854032159927562e-06,
      "loss": 0.6231,
      "step": 13720
    },
    {
      "epoch": 4.464866623292128,
      "grad_norm": 2.614452600479126,
      "learning_rate": 1.4766522437889035e-06,
      "loss": 0.5633,
      "step": 13725
    },
    {
      "epoch": 4.466493168510085,
      "grad_norm": 2.452821969985962,
      "learning_rate": 1.4679263406332467e-06,
      "loss": 0.5978,
      "step": 13730
    },
    {
      "epoch": 4.468119713728042,
      "grad_norm": 2.4186081886291504,
      "learning_rate": 1.4592255158250605e-06,
      "loss": 0.5868,
      "step": 13735
    },
    {
      "epoch": 4.469746258945999,
      "grad_norm": 2.2780089378356934,
      "learning_rate": 1.450549778636895e-06,
      "loss": 0.6151,
      "step": 13740
    },
    {
      "epoch": 4.4713728041639555,
      "grad_norm": 2.662311315536499,
      "learning_rate": 1.4418991383145675e-06,
      "loss": 0.6099,
      "step": 13745
    },
    {
      "epoch": 4.4729993493819125,
      "grad_norm": 2.089160919189453,
      "learning_rate": 1.43327360407714e-06,
      "loss": 0.6126,
      "step": 13750
    },
    {
      "epoch": 4.47462589459987,
      "grad_norm": 2.2200684547424316,
      "learning_rate": 1.424673185116926e-06,
      "loss": 0.6217,
      "step": 13755
    },
    {
      "epoch": 4.476252439817827,
      "grad_norm": 2.120103359222412,
      "learning_rate": 1.416097890599466e-06,
      "loss": 0.6153,
      "step": 13760
    },
    {
      "epoch": 4.477878985035784,
      "grad_norm": 2.2231285572052,
      "learning_rate": 1.4075477296635359e-06,
      "loss": 0.6043,
      "step": 13765
    },
    {
      "epoch": 4.479505530253741,
      "grad_norm": 2.0046818256378174,
      "learning_rate": 1.3990227114211191e-06,
      "loss": 0.5882,
      "step": 13770
    },
    {
      "epoch": 4.481132075471698,
      "grad_norm": 2.097630023956299,
      "learning_rate": 1.3905228449574066e-06,
      "loss": 0.5992,
      "step": 13775
    },
    {
      "epoch": 4.482758620689655,
      "grad_norm": 2.1202268600463867,
      "learning_rate": 1.3820481393307855e-06,
      "loss": 0.6002,
      "step": 13780
    },
    {
      "epoch": 4.484385165907613,
      "grad_norm": 2.250000476837158,
      "learning_rate": 1.3735986035728232e-06,
      "loss": 0.5961,
      "step": 13785
    },
    {
      "epoch": 4.4860117111255695,
      "grad_norm": 2.1143972873687744,
      "learning_rate": 1.365174246688275e-06,
      "loss": 0.5937,
      "step": 13790
    },
    {
      "epoch": 4.487638256343526,
      "grad_norm": 2.0777087211608887,
      "learning_rate": 1.356775077655048e-06,
      "loss": 0.5981,
      "step": 13795
    },
    {
      "epoch": 4.489264801561483,
      "grad_norm": 2.2611429691314697,
      "learning_rate": 1.3484011054242157e-06,
      "loss": 0.6002,
      "step": 13800
    },
    {
      "epoch": 4.49089134677944,
      "grad_norm": 2.1241838932037354,
      "learning_rate": 1.3400523389199976e-06,
      "loss": 0.5941,
      "step": 13805
    },
    {
      "epoch": 4.492517891997397,
      "grad_norm": 2.054589033126831,
      "learning_rate": 1.3317287870397572e-06,
      "loss": 0.6068,
      "step": 13810
    },
    {
      "epoch": 4.494144437215355,
      "grad_norm": 2.5084924697875977,
      "learning_rate": 1.3234304586539737e-06,
      "loss": 0.622,
      "step": 13815
    },
    {
      "epoch": 4.495770982433312,
      "grad_norm": 2.368989944458008,
      "learning_rate": 1.3151573626062535e-06,
      "loss": 0.6081,
      "step": 13820
    },
    {
      "epoch": 4.497397527651269,
      "grad_norm": 2.2150580883026123,
      "learning_rate": 1.3069095077133108e-06,
      "loss": 0.5901,
      "step": 13825
    },
    {
      "epoch": 4.499024072869226,
      "grad_norm": 2.227565050125122,
      "learning_rate": 1.298686902764959e-06,
      "loss": 0.5816,
      "step": 13830
    },
    {
      "epoch": 4.500650618087183,
      "grad_norm": 2.3818137645721436,
      "learning_rate": 1.2904895565241076e-06,
      "loss": 0.6138,
      "step": 13835
    },
    {
      "epoch": 4.5022771633051395,
      "grad_norm": 2.314545154571533,
      "learning_rate": 1.2823174777267439e-06,
      "loss": 0.6164,
      "step": 13840
    },
    {
      "epoch": 4.503903708523097,
      "grad_norm": 2.640153408050537,
      "learning_rate": 1.2741706750819232e-06,
      "loss": 0.6004,
      "step": 13845
    },
    {
      "epoch": 4.505530253741054,
      "grad_norm": 1.9388872385025024,
      "learning_rate": 1.266049157271773e-06,
      "loss": 0.5973,
      "step": 13850
    },
    {
      "epoch": 4.507156798959011,
      "grad_norm": 2.119119882583618,
      "learning_rate": 1.2579529329514645e-06,
      "loss": 0.6035,
      "step": 13855
    },
    {
      "epoch": 4.508783344176968,
      "grad_norm": 2.359867572784424,
      "learning_rate": 1.2498820107492204e-06,
      "loss": 0.63,
      "step": 13860
    },
    {
      "epoch": 4.510409889394925,
      "grad_norm": 2.2548766136169434,
      "learning_rate": 1.2418363992662997e-06,
      "loss": 0.5911,
      "step": 13865
    },
    {
      "epoch": 4.512036434612883,
      "grad_norm": 2.283270835876465,
      "learning_rate": 1.233816107076985e-06,
      "loss": 0.5932,
      "step": 13870
    },
    {
      "epoch": 4.51366297983084,
      "grad_norm": 2.1963112354278564,
      "learning_rate": 1.2258211427285671e-06,
      "loss": 0.5878,
      "step": 13875
    },
    {
      "epoch": 4.5152895250487965,
      "grad_norm": 2.4247801303863525,
      "learning_rate": 1.2178515147413665e-06,
      "loss": 0.6076,
      "step": 13880
    },
    {
      "epoch": 4.5169160702667535,
      "grad_norm": 2.2438313961029053,
      "learning_rate": 1.2099072316086757e-06,
      "loss": 0.6221,
      "step": 13885
    },
    {
      "epoch": 4.51854261548471,
      "grad_norm": 2.403510570526123,
      "learning_rate": 1.2019883017967943e-06,
      "loss": 0.6022,
      "step": 13890
    },
    {
      "epoch": 4.520169160702667,
      "grad_norm": 2.0666253566741943,
      "learning_rate": 1.194094733745002e-06,
      "loss": 0.5915,
      "step": 13895
    },
    {
      "epoch": 4.521795705920624,
      "grad_norm": 2.325934410095215,
      "learning_rate": 1.1862265358655505e-06,
      "loss": 0.5797,
      "step": 13900
    },
    {
      "epoch": 4.523422251138582,
      "grad_norm": 2.2772769927978516,
      "learning_rate": 1.1783837165436406e-06,
      "loss": 0.5991,
      "step": 13905
    },
    {
      "epoch": 4.525048796356539,
      "grad_norm": 2.4774351119995117,
      "learning_rate": 1.170566284137442e-06,
      "loss": 0.634,
      "step": 13910
    },
    {
      "epoch": 4.526675341574496,
      "grad_norm": 4.598024845123291,
      "learning_rate": 1.1627742469780684e-06,
      "loss": 0.6306,
      "step": 13915
    },
    {
      "epoch": 4.528301886792453,
      "grad_norm": 2.2267725467681885,
      "learning_rate": 1.1550076133695604e-06,
      "loss": 0.5909,
      "step": 13920
    },
    {
      "epoch": 4.52992843201041,
      "grad_norm": 2.3685433864593506,
      "learning_rate": 1.1472663915888888e-06,
      "loss": 0.5712,
      "step": 13925
    },
    {
      "epoch": 4.531554977228367,
      "grad_norm": 2.711865186691284,
      "learning_rate": 1.1395505898859487e-06,
      "loss": 0.6308,
      "step": 13930
    },
    {
      "epoch": 4.533181522446324,
      "grad_norm": 2.2160632610321045,
      "learning_rate": 1.1318602164835434e-06,
      "loss": 0.6059,
      "step": 13935
    },
    {
      "epoch": 4.534808067664281,
      "grad_norm": 1.9569048881530762,
      "learning_rate": 1.1241952795773697e-06,
      "loss": 0.6007,
      "step": 13940
    },
    {
      "epoch": 4.536434612882238,
      "grad_norm": 2.2012217044830322,
      "learning_rate": 1.1165557873360267e-06,
      "loss": 0.5861,
      "step": 13945
    },
    {
      "epoch": 4.538061158100195,
      "grad_norm": 2.1481223106384277,
      "learning_rate": 1.108941747900985e-06,
      "loss": 0.6111,
      "step": 13950
    },
    {
      "epoch": 4.539687703318152,
      "grad_norm": 2.3771395683288574,
      "learning_rate": 1.1013531693865985e-06,
      "loss": 0.5952,
      "step": 13955
    },
    {
      "epoch": 4.541314248536109,
      "grad_norm": 2.468219041824341,
      "learning_rate": 1.0937900598800872e-06,
      "loss": 0.5997,
      "step": 13960
    },
    {
      "epoch": 4.542940793754067,
      "grad_norm": 2.475027561187744,
      "learning_rate": 1.0862524274415282e-06,
      "loss": 0.6236,
      "step": 13965
    },
    {
      "epoch": 4.544567338972024,
      "grad_norm": 2.3337695598602295,
      "learning_rate": 1.0787402801038405e-06,
      "loss": 0.6248,
      "step": 13970
    },
    {
      "epoch": 4.5461938841899805,
      "grad_norm": 2.1060009002685547,
      "learning_rate": 1.071253625872795e-06,
      "loss": 0.5921,
      "step": 13975
    },
    {
      "epoch": 4.547820429407937,
      "grad_norm": 2.5564815998077393,
      "learning_rate": 1.0637924727269822e-06,
      "loss": 0.5999,
      "step": 13980
    },
    {
      "epoch": 4.549446974625894,
      "grad_norm": 2.3401384353637695,
      "learning_rate": 1.0563568286178216e-06,
      "loss": 0.6002,
      "step": 13985
    },
    {
      "epoch": 4.551073519843852,
      "grad_norm": 2.1616568565368652,
      "learning_rate": 1.0489467014695526e-06,
      "loss": 0.6227,
      "step": 13990
    },
    {
      "epoch": 4.552700065061809,
      "grad_norm": 2.3356876373291016,
      "learning_rate": 1.0415620991792135e-06,
      "loss": 0.6052,
      "step": 13995
    },
    {
      "epoch": 4.554326610279766,
      "grad_norm": 2.068387269973755,
      "learning_rate": 1.0342030296166428e-06,
      "loss": 0.6063,
      "step": 14000
    },
    {
      "epoch": 4.555953155497723,
      "grad_norm": 2.4070541858673096,
      "learning_rate": 1.0268695006244695e-06,
      "loss": 0.6329,
      "step": 14005
    },
    {
      "epoch": 4.55757970071568,
      "grad_norm": 2.427407741546631,
      "learning_rate": 1.0195615200180974e-06,
      "loss": 0.6172,
      "step": 14010
    },
    {
      "epoch": 4.559206245933637,
      "grad_norm": 2.068614959716797,
      "learning_rate": 1.0122790955857192e-06,
      "loss": 0.5901,
      "step": 14015
    },
    {
      "epoch": 4.560832791151594,
      "grad_norm": 2.4775562286376953,
      "learning_rate": 1.0050222350882682e-06,
      "loss": 0.6524,
      "step": 14020
    },
    {
      "epoch": 4.562459336369551,
      "grad_norm": 2.4976789951324463,
      "learning_rate": 9.97790946259461e-07,
      "loss": 0.5967,
      "step": 14025
    },
    {
      "epoch": 4.564085881587508,
      "grad_norm": 2.4459400177001953,
      "learning_rate": 9.905852368057383e-07,
      "loss": 0.6231,
      "step": 14030
    },
    {
      "epoch": 4.565712426805465,
      "grad_norm": 2.338926076889038,
      "learning_rate": 9.834051144062994e-07,
      "loss": 0.5923,
      "step": 14035
    },
    {
      "epoch": 4.567338972023422,
      "grad_norm": 2.1870245933532715,
      "learning_rate": 9.762505867130594e-07,
      "loss": 0.6371,
      "step": 14040
    },
    {
      "epoch": 4.568965517241379,
      "grad_norm": 2.2032628059387207,
      "learning_rate": 9.691216613506692e-07,
      "loss": 0.5831,
      "step": 14045
    },
    {
      "epoch": 4.570592062459337,
      "grad_norm": 2.1163077354431152,
      "learning_rate": 9.620183459164878e-07,
      "loss": 0.6174,
      "step": 14050
    },
    {
      "epoch": 4.572218607677294,
      "grad_norm": 2.1451101303100586,
      "learning_rate": 9.549406479805818e-07,
      "loss": 0.5736,
      "step": 14055
    },
    {
      "epoch": 4.573845152895251,
      "grad_norm": 2.404646158218384,
      "learning_rate": 9.478885750857285e-07,
      "loss": 0.5963,
      "step": 14060
    },
    {
      "epoch": 4.5754716981132075,
      "grad_norm": 3.9412124156951904,
      "learning_rate": 9.408621347473751e-07,
      "loss": 0.6251,
      "step": 14065
    },
    {
      "epoch": 4.577098243331164,
      "grad_norm": 2.258427381515503,
      "learning_rate": 9.338613344536701e-07,
      "loss": 0.5948,
      "step": 14070
    },
    {
      "epoch": 4.578724788549121,
      "grad_norm": 2.4941704273223877,
      "learning_rate": 9.26886181665429e-07,
      "loss": 0.6027,
      "step": 14075
    },
    {
      "epoch": 4.580351333767078,
      "grad_norm": 2.2300431728363037,
      "learning_rate": 9.199366838161389e-07,
      "loss": 0.6096,
      "step": 14080
    },
    {
      "epoch": 4.581977878985036,
      "grad_norm": 2.5056817531585693,
      "learning_rate": 9.130128483119366e-07,
      "loss": 0.6497,
      "step": 14085
    },
    {
      "epoch": 4.583604424202993,
      "grad_norm": 2.3211681842803955,
      "learning_rate": 9.06114682531628e-07,
      "loss": 0.5903,
      "step": 14090
    },
    {
      "epoch": 4.58523096942095,
      "grad_norm": 2.2497894763946533,
      "learning_rate": 8.992421938266438e-07,
      "loss": 0.6077,
      "step": 14095
    },
    {
      "epoch": 4.586857514638907,
      "grad_norm": 2.4197893142700195,
      "learning_rate": 8.923953895210612e-07,
      "loss": 0.6446,
      "step": 14100
    },
    {
      "epoch": 4.588484059856864,
      "grad_norm": 2.1274402141571045,
      "learning_rate": 8.855742769115799e-07,
      "loss": 0.6106,
      "step": 14105
    },
    {
      "epoch": 4.5901106050748215,
      "grad_norm": 2.331447124481201,
      "learning_rate": 8.787788632675293e-07,
      "loss": 0.6173,
      "step": 14110
    },
    {
      "epoch": 4.591737150292778,
      "grad_norm": 2.1276087760925293,
      "learning_rate": 8.720091558308357e-07,
      "loss": 0.5834,
      "step": 14115
    },
    {
      "epoch": 4.593363695510735,
      "grad_norm": 2.276212453842163,
      "learning_rate": 8.652651618160424e-07,
      "loss": 0.561,
      "step": 14120
    },
    {
      "epoch": 4.594990240728692,
      "grad_norm": 2.1035265922546387,
      "learning_rate": 8.58546888410286e-07,
      "loss": 0.5967,
      "step": 14125
    },
    {
      "epoch": 4.596616785946649,
      "grad_norm": 2.2376317977905273,
      "learning_rate": 8.51854342773295e-07,
      "loss": 0.5995,
      "step": 14130
    },
    {
      "epoch": 4.598243331164606,
      "grad_norm": 2.7903850078582764,
      "learning_rate": 8.451875320373698e-07,
      "loss": 0.6026,
      "step": 14135
    },
    {
      "epoch": 4.599869876382563,
      "grad_norm": 2.252089262008667,
      "learning_rate": 8.385464633074019e-07,
      "loss": 0.6072,
      "step": 14140
    },
    {
      "epoch": 4.601496421600521,
      "grad_norm": 2.2325665950775146,
      "learning_rate": 8.319311436608301e-07,
      "loss": 0.6099,
      "step": 14145
    },
    {
      "epoch": 4.603122966818478,
      "grad_norm": 2.1295957565307617,
      "learning_rate": 8.25341580147665e-07,
      "loss": 0.5926,
      "step": 14150
    },
    {
      "epoch": 4.604749512036435,
      "grad_norm": 2.0349044799804688,
      "learning_rate": 8.187777797904639e-07,
      "loss": 0.6037,
      "step": 14155
    },
    {
      "epoch": 4.6063760572543915,
      "grad_norm": 2.2015886306762695,
      "learning_rate": 8.122397495843343e-07,
      "loss": 0.6203,
      "step": 14160
    },
    {
      "epoch": 4.608002602472348,
      "grad_norm": 2.2389609813690186,
      "learning_rate": 8.057274964969108e-07,
      "loss": 0.6152,
      "step": 14165
    },
    {
      "epoch": 4.609629147690306,
      "grad_norm": 2.4615068435668945,
      "learning_rate": 7.992410274683615e-07,
      "loss": 0.5913,
      "step": 14170
    },
    {
      "epoch": 4.611255692908263,
      "grad_norm": 2.589155912399292,
      "learning_rate": 7.927803494113761e-07,
      "loss": 0.5855,
      "step": 14175
    },
    {
      "epoch": 4.61288223812622,
      "grad_norm": 2.780627965927124,
      "learning_rate": 7.863454692111583e-07,
      "loss": 0.6109,
      "step": 14180
    },
    {
      "epoch": 4.614508783344177,
      "grad_norm": 2.3839402198791504,
      "learning_rate": 7.799363937254195e-07,
      "loss": 0.6059,
      "step": 14185
    },
    {
      "epoch": 4.616135328562134,
      "grad_norm": 2.354104518890381,
      "learning_rate": 7.735531297843713e-07,
      "loss": 0.6084,
      "step": 14190
    },
    {
      "epoch": 4.617761873780091,
      "grad_norm": 2.4791300296783447,
      "learning_rate": 7.671956841907218e-07,
      "loss": 0.597,
      "step": 14195
    },
    {
      "epoch": 4.6193884189980485,
      "grad_norm": 2.4160773754119873,
      "learning_rate": 7.60864063719649e-07,
      "loss": 0.6093,
      "step": 14200
    },
    {
      "epoch": 4.621014964216005,
      "grad_norm": 2.403437376022339,
      "learning_rate": 7.545582751188274e-07,
      "loss": 0.5975,
      "step": 14205
    },
    {
      "epoch": 4.622641509433962,
      "grad_norm": 2.3289949893951416,
      "learning_rate": 7.482783251083869e-07,
      "loss": 0.5875,
      "step": 14210
    },
    {
      "epoch": 4.624268054651919,
      "grad_norm": 2.0737967491149902,
      "learning_rate": 7.420242203809325e-07,
      "loss": 0.5932,
      "step": 14215
    },
    {
      "epoch": 4.625894599869876,
      "grad_norm": 2.1003692150115967,
      "learning_rate": 7.357959676015214e-07,
      "loss": 0.5966,
      "step": 14220
    },
    {
      "epoch": 4.627521145087833,
      "grad_norm": 3.097031831741333,
      "learning_rate": 7.295935734076609e-07,
      "loss": 0.6009,
      "step": 14225
    },
    {
      "epoch": 4.629147690305791,
      "grad_norm": 2.105490207672119,
      "learning_rate": 7.234170444092942e-07,
      "loss": 0.5875,
      "step": 14230
    },
    {
      "epoch": 4.630774235523748,
      "grad_norm": 2.537712335586548,
      "learning_rate": 7.172663871888113e-07,
      "loss": 0.6117,
      "step": 14235
    },
    {
      "epoch": 4.632400780741705,
      "grad_norm": 2.247049570083618,
      "learning_rate": 7.11141608301022e-07,
      "loss": 0.6105,
      "step": 14240
    },
    {
      "epoch": 4.634027325959662,
      "grad_norm": 2.3542373180389404,
      "learning_rate": 7.050427142731547e-07,
      "loss": 0.6041,
      "step": 14245
    },
    {
      "epoch": 4.6356538711776185,
      "grad_norm": 2.3531222343444824,
      "learning_rate": 6.989697116048633e-07,
      "loss": 0.6171,
      "step": 14250
    },
    {
      "epoch": 4.637280416395575,
      "grad_norm": 2.226870536804199,
      "learning_rate": 6.929226067682037e-07,
      "loss": 0.5884,
      "step": 14255
    },
    {
      "epoch": 4.638906961613533,
      "grad_norm": 3.1148924827575684,
      "learning_rate": 6.86901406207624e-07,
      "loss": 0.6381,
      "step": 14260
    },
    {
      "epoch": 4.64053350683149,
      "grad_norm": 1.998335599899292,
      "learning_rate": 6.809061163399827e-07,
      "loss": 0.5953,
      "step": 14265
    },
    {
      "epoch": 4.642160052049447,
      "grad_norm": 2.1844255924224854,
      "learning_rate": 6.749367435545024e-07,
      "loss": 0.5795,
      "step": 14270
    },
    {
      "epoch": 4.643786597267404,
      "grad_norm": 2.270549774169922,
      "learning_rate": 6.689932942128108e-07,
      "loss": 0.5996,
      "step": 14275
    },
    {
      "epoch": 4.645413142485361,
      "grad_norm": 2.049288511276245,
      "learning_rate": 6.630757746488886e-07,
      "loss": 0.6136,
      "step": 14280
    },
    {
      "epoch": 4.647039687703318,
      "grad_norm": 2.2355549335479736,
      "learning_rate": 6.571841911690968e-07,
      "loss": 0.6488,
      "step": 14285
    },
    {
      "epoch": 4.648666232921276,
      "grad_norm": 3.0274786949157715,
      "learning_rate": 6.513185500521463e-07,
      "loss": 0.5846,
      "step": 14290
    },
    {
      "epoch": 4.6502927781392325,
      "grad_norm": 2.0298526287078857,
      "learning_rate": 6.454788575491061e-07,
      "loss": 0.5942,
      "step": 14295
    },
    {
      "epoch": 4.651919323357189,
      "grad_norm": 2.31168794631958,
      "learning_rate": 6.396651198833897e-07,
      "loss": 0.5852,
      "step": 14300
    },
    {
      "epoch": 4.653545868575146,
      "grad_norm": 2.6897263526916504,
      "learning_rate": 6.338773432507494e-07,
      "loss": 0.6254,
      "step": 14305
    },
    {
      "epoch": 4.655172413793103,
      "grad_norm": 2.1427431106567383,
      "learning_rate": 6.281155338192762e-07,
      "loss": 0.5588,
      "step": 14310
    },
    {
      "epoch": 4.656798959011061,
      "grad_norm": 4.211636543273926,
      "learning_rate": 6.223796977293777e-07,
      "loss": 0.5825,
      "step": 14315
    },
    {
      "epoch": 4.658425504229018,
      "grad_norm": 2.302633047103882,
      "learning_rate": 6.166698410937949e-07,
      "loss": 0.6081,
      "step": 14320
    },
    {
      "epoch": 4.660052049446975,
      "grad_norm": 2.098952054977417,
      "learning_rate": 6.109859699975684e-07,
      "loss": 0.5957,
      "step": 14325
    },
    {
      "epoch": 4.661678594664932,
      "grad_norm": 2.4573733806610107,
      "learning_rate": 6.053280904980557e-07,
      "loss": 0.5685,
      "step": 14330
    },
    {
      "epoch": 4.663305139882889,
      "grad_norm": 2.3456218242645264,
      "learning_rate": 5.996962086249058e-07,
      "loss": 0.6169,
      "step": 14335
    },
    {
      "epoch": 4.6649316851008455,
      "grad_norm": 2.059295415878296,
      "learning_rate": 5.940903303800705e-07,
      "loss": 0.5839,
      "step": 14340
    },
    {
      "epoch": 4.6665582303188025,
      "grad_norm": 2.1461198329925537,
      "learning_rate": 5.885104617377873e-07,
      "loss": 0.6013,
      "step": 14345
    },
    {
      "epoch": 4.66818477553676,
      "grad_norm": 2.2912466526031494,
      "learning_rate": 5.829566086445721e-07,
      "loss": 0.63,
      "step": 14350
    },
    {
      "epoch": 4.669811320754717,
      "grad_norm": 2.241825819015503,
      "learning_rate": 5.774287770192149e-07,
      "loss": 0.596,
      "step": 14355
    },
    {
      "epoch": 4.671437865972674,
      "grad_norm": 1.978074550628662,
      "learning_rate": 5.719269727527843e-07,
      "loss": 0.5816,
      "step": 14360
    },
    {
      "epoch": 4.673064411190631,
      "grad_norm": 2.453749895095825,
      "learning_rate": 5.664512017085926e-07,
      "loss": 0.6131,
      "step": 14365
    },
    {
      "epoch": 4.674690956408588,
      "grad_norm": 2.2468841075897217,
      "learning_rate": 5.610014697222249e-07,
      "loss": 0.6201,
      "step": 14370
    },
    {
      "epoch": 4.676317501626546,
      "grad_norm": 2.2596335411071777,
      "learning_rate": 5.555777826015129e-07,
      "loss": 0.6454,
      "step": 14375
    },
    {
      "epoch": 4.677944046844503,
      "grad_norm": 2.49043607711792,
      "learning_rate": 5.501801461265304e-07,
      "loss": 0.5676,
      "step": 14380
    },
    {
      "epoch": 4.6795705920624595,
      "grad_norm": 2.2760651111602783,
      "learning_rate": 5.448085660495816e-07,
      "loss": 0.5935,
      "step": 14385
    },
    {
      "epoch": 4.681197137280416,
      "grad_norm": 1.9554202556610107,
      "learning_rate": 5.394630480952178e-07,
      "loss": 0.5756,
      "step": 14390
    },
    {
      "epoch": 4.682823682498373,
      "grad_norm": 2.103641986846924,
      "learning_rate": 5.341435979601988e-07,
      "loss": 0.5702,
      "step": 14395
    },
    {
      "epoch": 4.68445022771633,
      "grad_norm": 5.318946361541748,
      "learning_rate": 5.288502213135149e-07,
      "loss": 0.6155,
      "step": 14400
    },
    {
      "epoch": 4.686076772934287,
      "grad_norm": 2.606043815612793,
      "learning_rate": 5.235829237963646e-07,
      "loss": 0.6106,
      "step": 14405
    },
    {
      "epoch": 4.687703318152245,
      "grad_norm": 2.7256171703338623,
      "learning_rate": 5.183417110221606e-07,
      "loss": 0.6194,
      "step": 14410
    },
    {
      "epoch": 4.689329863370202,
      "grad_norm": 2.1221957206726074,
      "learning_rate": 5.131265885765041e-07,
      "loss": 0.6355,
      "step": 14415
    },
    {
      "epoch": 4.690956408588159,
      "grad_norm": 2.3277482986450195,
      "learning_rate": 5.07937562017205e-07,
      "loss": 0.5896,
      "step": 14420
    },
    {
      "epoch": 4.692582953806116,
      "grad_norm": 2.2037417888641357,
      "learning_rate": 5.027746368742536e-07,
      "loss": 0.6172,
      "step": 14425
    },
    {
      "epoch": 4.694209499024073,
      "grad_norm": 2.131485939025879,
      "learning_rate": 4.976378186498293e-07,
      "loss": 0.6077,
      "step": 14430
    },
    {
      "epoch": 4.69583604424203,
      "grad_norm": 2.2293078899383545,
      "learning_rate": 4.925271128182807e-07,
      "loss": 0.5721,
      "step": 14435
    },
    {
      "epoch": 4.697462589459987,
      "grad_norm": 2.614671468734741,
      "learning_rate": 4.874425248261428e-07,
      "loss": 0.6042,
      "step": 14440
    },
    {
      "epoch": 4.699089134677944,
      "grad_norm": 2.4178030490875244,
      "learning_rate": 4.823840600921003e-07,
      "loss": 0.5984,
      "step": 14445
    },
    {
      "epoch": 4.700715679895901,
      "grad_norm": 2.6082773208618164,
      "learning_rate": 4.773517240070108e-07,
      "loss": 0.5976,
      "step": 14450
    },
    {
      "epoch": 4.702342225113858,
      "grad_norm": 2.058941602706909,
      "learning_rate": 4.7234552193387846e-07,
      "loss": 0.6062,
      "step": 14455
    },
    {
      "epoch": 4.703968770331815,
      "grad_norm": 2.080263614654541,
      "learning_rate": 4.67365459207858e-07,
      "loss": 0.5993,
      "step": 14460
    },
    {
      "epoch": 4.705595315549772,
      "grad_norm": 2.656578779220581,
      "learning_rate": 4.624115411362512e-07,
      "loss": 0.6043,
      "step": 14465
    },
    {
      "epoch": 4.70722186076773,
      "grad_norm": 2.144035577774048,
      "learning_rate": 4.5748377299849045e-07,
      "loss": 0.613,
      "step": 14470
    },
    {
      "epoch": 4.7088484059856865,
      "grad_norm": 2.266422748565674,
      "learning_rate": 4.525821600461472e-07,
      "loss": 0.5928,
      "step": 14475
    },
    {
      "epoch": 4.7104749512036435,
      "grad_norm": 2.155153512954712,
      "learning_rate": 4.477067075029123e-07,
      "loss": 0.6138,
      "step": 14480
    },
    {
      "epoch": 4.7121014964216,
      "grad_norm": 2.283352851867676,
      "learning_rate": 4.428574205646047e-07,
      "loss": 0.593,
      "step": 14485
    },
    {
      "epoch": 4.713728041639557,
      "grad_norm": 2.3388490676879883,
      "learning_rate": 4.3803430439915137e-07,
      "loss": 0.5753,
      "step": 14490
    },
    {
      "epoch": 4.715354586857515,
      "grad_norm": 1.8969876766204834,
      "learning_rate": 4.332373641465909e-07,
      "loss": 0.5931,
      "step": 14495
    },
    {
      "epoch": 4.716981132075472,
      "grad_norm": 2.4692556858062744,
      "learning_rate": 4.284666049190644e-07,
      "loss": 0.5988,
      "step": 14500
    },
    {
      "epoch": 4.718607677293429,
      "grad_norm": 2.4041595458984375,
      "learning_rate": 4.2372203180081893e-07,
      "loss": 0.6155,
      "step": 14505
    },
    {
      "epoch": 4.720234222511386,
      "grad_norm": 2.1963233947753906,
      "learning_rate": 4.1900364984818754e-07,
      "loss": 0.6022,
      "step": 14510
    },
    {
      "epoch": 4.721860767729343,
      "grad_norm": 2.479945421218872,
      "learning_rate": 4.143114640895951e-07,
      "loss": 0.613,
      "step": 14515
    },
    {
      "epoch": 4.7234873129473,
      "grad_norm": 2.1995420455932617,
      "learning_rate": 4.0964547952554443e-07,
      "loss": 0.5877,
      "step": 14520
    },
    {
      "epoch": 4.7251138581652565,
      "grad_norm": 2.2108662128448486,
      "learning_rate": 4.05005701128619e-07,
      "loss": 0.6426,
      "step": 14525
    },
    {
      "epoch": 4.726740403383214,
      "grad_norm": 1.9551113843917847,
      "learning_rate": 4.0039213384347187e-07,
      "loss": 0.6173,
      "step": 14530
    },
    {
      "epoch": 4.728366948601171,
      "grad_norm": 2.2449254989624023,
      "learning_rate": 3.958047825868283e-07,
      "loss": 0.5923,
      "step": 14535
    },
    {
      "epoch": 4.729993493819128,
      "grad_norm": 2.076014518737793,
      "learning_rate": 3.912436522474666e-07,
      "loss": 0.6095,
      "step": 14540
    },
    {
      "epoch": 4.731620039037085,
      "grad_norm": 2.1760308742523193,
      "learning_rate": 3.867087476862291e-07,
      "loss": 0.6125,
      "step": 14545
    },
    {
      "epoch": 4.733246584255042,
      "grad_norm": 2.2484352588653564,
      "learning_rate": 3.822000737360026e-07,
      "loss": 0.5837,
      "step": 14550
    },
    {
      "epoch": 4.734873129473,
      "grad_norm": 2.1155903339385986,
      "learning_rate": 3.777176352017242e-07,
      "loss": 0.6028,
      "step": 14555
    },
    {
      "epoch": 4.736499674690957,
      "grad_norm": 2.0362231731414795,
      "learning_rate": 3.7326143686036706e-07,
      "loss": 0.6188,
      "step": 14560
    },
    {
      "epoch": 4.738126219908914,
      "grad_norm": 2.6431407928466797,
      "learning_rate": 3.6883148346094356e-07,
      "loss": 0.6163,
      "step": 14565
    },
    {
      "epoch": 4.7397527651268705,
      "grad_norm": 2.1116814613342285,
      "learning_rate": 3.644277797244966e-07,
      "loss": 0.6308,
      "step": 14570
    },
    {
      "epoch": 4.741379310344827,
      "grad_norm": 3.778200149536133,
      "learning_rate": 3.600503303440972e-07,
      "loss": 0.6118,
      "step": 14575
    },
    {
      "epoch": 4.743005855562784,
      "grad_norm": 2.127812147140503,
      "learning_rate": 3.556991399848275e-07,
      "loss": 0.5945,
      "step": 14580
    },
    {
      "epoch": 4.744632400780741,
      "grad_norm": 2.4827187061309814,
      "learning_rate": 3.5137421328379493e-07,
      "loss": 0.5995,
      "step": 14585
    },
    {
      "epoch": 4.746258945998699,
      "grad_norm": 2.650378465652466,
      "learning_rate": 3.4707555485011533e-07,
      "loss": 0.6254,
      "step": 14590
    },
    {
      "epoch": 4.747885491216656,
      "grad_norm": 2.0619797706604004,
      "learning_rate": 3.4280316926490196e-07,
      "loss": 0.5765,
      "step": 14595
    },
    {
      "epoch": 4.749512036434613,
      "grad_norm": 2.2239131927490234,
      "learning_rate": 3.385570610812794e-07,
      "loss": 0.6025,
      "step": 14600
    },
    {
      "epoch": 4.75113858165257,
      "grad_norm": 2.138410806655884,
      "learning_rate": 3.3433723482436676e-07,
      "loss": 0.5962,
      "step": 14605
    },
    {
      "epoch": 4.752765126870527,
      "grad_norm": 2.1161324977874756,
      "learning_rate": 3.3014369499126675e-07,
      "loss": 0.5921,
      "step": 14610
    },
    {
      "epoch": 4.7543916720884845,
      "grad_norm": 2.4537761211395264,
      "learning_rate": 3.259764460510767e-07,
      "loss": 0.5996,
      "step": 14615
    },
    {
      "epoch": 4.756018217306441,
      "grad_norm": 2.0539655685424805,
      "learning_rate": 3.218354924448719e-07,
      "loss": 0.5926,
      "step": 14620
    },
    {
      "epoch": 4.757644762524398,
      "grad_norm": 1.9618831872940063,
      "learning_rate": 3.177208385857028e-07,
      "loss": 0.5867,
      "step": 14625
    },
    {
      "epoch": 4.759271307742355,
      "grad_norm": 3.0551681518554688,
      "learning_rate": 3.1363248885859506e-07,
      "loss": 0.5933,
      "step": 14630
    },
    {
      "epoch": 4.760897852960312,
      "grad_norm": 2.0586209297180176,
      "learning_rate": 3.0957044762054133e-07,
      "loss": 0.6,
      "step": 14635
    },
    {
      "epoch": 4.762524398178269,
      "grad_norm": 2.6311097145080566,
      "learning_rate": 3.055347192004954e-07,
      "loss": 0.628,
      "step": 14640
    },
    {
      "epoch": 4.764150943396227,
      "grad_norm": 2.4645638465881348,
      "learning_rate": 3.0152530789936963e-07,
      "loss": 0.5861,
      "step": 14645
    },
    {
      "epoch": 4.765777488614184,
      "grad_norm": 2.747060775756836,
      "learning_rate": 2.9754221799003503e-07,
      "loss": 0.6072,
      "step": 14650
    },
    {
      "epoch": 4.767404033832141,
      "grad_norm": 2.1690328121185303,
      "learning_rate": 2.9358545371729883e-07,
      "loss": 0.5802,
      "step": 14655
    },
    {
      "epoch": 4.7690305790500975,
      "grad_norm": 2.304227828979492,
      "learning_rate": 2.8965501929792695e-07,
      "loss": 0.6088,
      "step": 14660
    },
    {
      "epoch": 4.770657124268054,
      "grad_norm": 2.5094473361968994,
      "learning_rate": 2.857509189206187e-07,
      "loss": 0.6202,
      "step": 14665
    },
    {
      "epoch": 4.772283669486011,
      "grad_norm": 2.518124580383301,
      "learning_rate": 2.818731567460098e-07,
      "loss": 0.6064,
      "step": 14670
    },
    {
      "epoch": 4.773910214703969,
      "grad_norm": 2.467226266860962,
      "learning_rate": 2.7802173690666676e-07,
      "loss": 0.5852,
      "step": 14675
    },
    {
      "epoch": 4.775536759921926,
      "grad_norm": 2.188849925994873,
      "learning_rate": 2.741966635070842e-07,
      "loss": 0.5984,
      "step": 14680
    },
    {
      "epoch": 4.777163305139883,
      "grad_norm": 2.122866153717041,
      "learning_rate": 2.7039794062367616e-07,
      "loss": 0.5949,
      "step": 14685
    },
    {
      "epoch": 4.77878985035784,
      "grad_norm": 2.236489772796631,
      "learning_rate": 2.6662557230477667e-07,
      "loss": 0.5542,
      "step": 14690
    },
    {
      "epoch": 4.780416395575797,
      "grad_norm": 2.2290842533111572,
      "learning_rate": 2.6287956257063374e-07,
      "loss": 0.5766,
      "step": 14695
    },
    {
      "epoch": 4.782042940793754,
      "grad_norm": 2.478623151779175,
      "learning_rate": 2.5915991541340667e-07,
      "loss": 0.612,
      "step": 14700
    },
    {
      "epoch": 4.7836694860117115,
      "grad_norm": 2.169400691986084,
      "learning_rate": 2.5546663479715236e-07,
      "loss": 0.6124,
      "step": 14705
    },
    {
      "epoch": 4.785296031229668,
      "grad_norm": 2.0891544818878174,
      "learning_rate": 2.5179972465784186e-07,
      "loss": 0.6029,
      "step": 14710
    },
    {
      "epoch": 4.786922576447625,
      "grad_norm": 2.379319667816162,
      "learning_rate": 2.481591889033269e-07,
      "loss": 0.579,
      "step": 14715
    },
    {
      "epoch": 4.788549121665582,
      "grad_norm": 1.8330259323120117,
      "learning_rate": 2.4454503141336513e-07,
      "loss": 0.5935,
      "step": 14720
    },
    {
      "epoch": 4.790175666883539,
      "grad_norm": 2.2244858741760254,
      "learning_rate": 2.409572560395951e-07,
      "loss": 0.635,
      "step": 14725
    },
    {
      "epoch": 4.791802212101496,
      "grad_norm": 4.858855724334717,
      "learning_rate": 2.3739586660554148e-07,
      "loss": 0.5882,
      "step": 14730
    },
    {
      "epoch": 4.793428757319454,
      "grad_norm": 1.9892836809158325,
      "learning_rate": 2.338608669066128e-07,
      "loss": 0.5708,
      "step": 14735
    },
    {
      "epoch": 4.795055302537411,
      "grad_norm": 2.171619176864624,
      "learning_rate": 2.3035226071008997e-07,
      "loss": 0.6061,
      "step": 14740
    },
    {
      "epoch": 4.796681847755368,
      "grad_norm": 2.2100746631622314,
      "learning_rate": 2.2687005175512642e-07,
      "loss": 0.5932,
      "step": 14745
    },
    {
      "epoch": 4.798308392973325,
      "grad_norm": 2.541048765182495,
      "learning_rate": 2.2341424375274256e-07,
      "loss": 0.5977,
      "step": 14750
    },
    {
      "epoch": 4.7999349381912815,
      "grad_norm": 2.1693179607391357,
      "learning_rate": 2.1998484038582567e-07,
      "loss": 0.5917,
      "step": 14755
    },
    {
      "epoch": 4.801561483409239,
      "grad_norm": 2.8264858722686768,
      "learning_rate": 2.165818453091245e-07,
      "loss": 0.6237,
      "step": 14760
    },
    {
      "epoch": 4.803188028627196,
      "grad_norm": 3.8728349208831787,
      "learning_rate": 2.1320526214924086e-07,
      "loss": 0.5953,
      "step": 14765
    },
    {
      "epoch": 4.804814573845153,
      "grad_norm": 2.0309038162231445,
      "learning_rate": 2.098550945046268e-07,
      "loss": 0.6002,
      "step": 14770
    },
    {
      "epoch": 4.80644111906311,
      "grad_norm": 2.2680821418762207,
      "learning_rate": 2.0653134594559586e-07,
      "loss": 0.5958,
      "step": 14775
    },
    {
      "epoch": 4.808067664281067,
      "grad_norm": 2.4042727947235107,
      "learning_rate": 2.0323402001428682e-07,
      "loss": 0.6018,
      "step": 14780
    },
    {
      "epoch": 4.809694209499024,
      "grad_norm": 2.323235511779785,
      "learning_rate": 1.999631202246971e-07,
      "loss": 0.6215,
      "step": 14785
    },
    {
      "epoch": 4.811320754716981,
      "grad_norm": 10.879158020019531,
      "learning_rate": 1.9671865006265223e-07,
      "loss": 0.6268,
      "step": 14790
    },
    {
      "epoch": 4.8129472999349385,
      "grad_norm": 2.1472039222717285,
      "learning_rate": 1.935006129858169e-07,
      "loss": 0.6221,
      "step": 14795
    },
    {
      "epoch": 4.814573845152895,
      "grad_norm": 2.1684577465057373,
      "learning_rate": 1.9030901242367837e-07,
      "loss": 0.5702,
      "step": 14800
    },
    {
      "epoch": 4.816200390370852,
      "grad_norm": 2.1772913932800293,
      "learning_rate": 1.8714385177756032e-07,
      "loss": 0.592,
      "step": 14805
    },
    {
      "epoch": 4.817826935588809,
      "grad_norm": 2.1808815002441406,
      "learning_rate": 1.8400513442059786e-07,
      "loss": 0.5781,
      "step": 14810
    },
    {
      "epoch": 4.819453480806766,
      "grad_norm": 2.3493869304656982,
      "learning_rate": 1.8089286369775415e-07,
      "loss": 0.6637,
      "step": 14815
    },
    {
      "epoch": 4.821080026024724,
      "grad_norm": 2.1450183391571045,
      "learning_rate": 1.7780704292580107e-07,
      "loss": 0.5833,
      "step": 14820
    },
    {
      "epoch": 4.822706571242681,
      "grad_norm": 2.0887880325317383,
      "learning_rate": 1.7474767539333302e-07,
      "loss": 0.6156,
      "step": 14825
    },
    {
      "epoch": 4.824333116460638,
      "grad_norm": 2.41434383392334,
      "learning_rate": 1.717147643607392e-07,
      "loss": 0.603,
      "step": 14830
    },
    {
      "epoch": 4.825959661678595,
      "grad_norm": 2.320181369781494,
      "learning_rate": 1.687083130602257e-07,
      "loss": 0.6061,
      "step": 14835
    },
    {
      "epoch": 4.827586206896552,
      "grad_norm": 2.6415679454803467,
      "learning_rate": 1.6572832469579357e-07,
      "loss": 0.6161,
      "step": 14840
    },
    {
      "epoch": 4.8292127521145085,
      "grad_norm": 2.5681235790252686,
      "learning_rate": 1.6277480244324127e-07,
      "loss": 0.5878,
      "step": 14845
    },
    {
      "epoch": 4.830839297332465,
      "grad_norm": 2.113208055496216,
      "learning_rate": 1.5984774945017044e-07,
      "loss": 0.59,
      "step": 14850
    },
    {
      "epoch": 4.832465842550423,
      "grad_norm": 2.5020604133605957,
      "learning_rate": 1.5694716883596083e-07,
      "loss": 0.6092,
      "step": 14855
    },
    {
      "epoch": 4.83409238776838,
      "grad_norm": 2.2251505851745605,
      "learning_rate": 1.540730636917953e-07,
      "loss": 0.6021,
      "step": 14860
    },
    {
      "epoch": 4.835718932986337,
      "grad_norm": 2.315005302429199,
      "learning_rate": 1.512254370806293e-07,
      "loss": 0.6022,
      "step": 14865
    },
    {
      "epoch": 4.837345478204294,
      "grad_norm": 2.2444543838500977,
      "learning_rate": 1.4840429203720752e-07,
      "loss": 0.607,
      "step": 14870
    },
    {
      "epoch": 4.838972023422251,
      "grad_norm": 2.1719343662261963,
      "learning_rate": 1.4560963156804997e-07,
      "loss": 0.6111,
      "step": 14875
    },
    {
      "epoch": 4.840598568640209,
      "grad_norm": 2.1927409172058105,
      "learning_rate": 1.4284145865144928e-07,
      "loss": 0.6029,
      "step": 14880
    },
    {
      "epoch": 4.842225113858166,
      "grad_norm": 1.9452162981033325,
      "learning_rate": 1.4009977623747617e-07,
      "loss": 0.5633,
      "step": 14885
    },
    {
      "epoch": 4.8438516590761225,
      "grad_norm": 2.175806999206543,
      "learning_rate": 1.3738458724796288e-07,
      "loss": 0.6247,
      "step": 14890
    },
    {
      "epoch": 4.845478204294079,
      "grad_norm": 1.9696511030197144,
      "learning_rate": 1.3469589457651422e-07,
      "loss": 0.5686,
      "step": 14895
    },
    {
      "epoch": 4.847104749512036,
      "grad_norm": 2.063143253326416,
      "learning_rate": 1.3203370108849644e-07,
      "loss": 0.5636,
      "step": 14900
    },
    {
      "epoch": 4.848731294729993,
      "grad_norm": 2.154536724090576,
      "learning_rate": 1.2939800962103176e-07,
      "loss": 0.5839,
      "step": 14905
    },
    {
      "epoch": 4.85035783994795,
      "grad_norm": 1.9221045970916748,
      "learning_rate": 1.2678882298299833e-07,
      "loss": 0.5907,
      "step": 14910
    },
    {
      "epoch": 4.851984385165908,
      "grad_norm": 2.1464858055114746,
      "learning_rate": 1.2420614395503294e-07,
      "loss": 0.6062,
      "step": 14915
    },
    {
      "epoch": 4.853610930383865,
      "grad_norm": 2.47495174407959,
      "learning_rate": 1.2164997528952004e-07,
      "loss": 0.5889,
      "step": 14920
    },
    {
      "epoch": 4.855237475601822,
      "grad_norm": 2.2816359996795654,
      "learning_rate": 1.1912031971059168e-07,
      "loss": 0.6131,
      "step": 14925
    },
    {
      "epoch": 4.856864020819779,
      "grad_norm": 2.2610902786254883,
      "learning_rate": 1.1661717991412746e-07,
      "loss": 0.5917,
      "step": 14930
    },
    {
      "epoch": 4.8584905660377355,
      "grad_norm": 2.1258704662323,
      "learning_rate": 1.1414055856774075e-07,
      "loss": 0.6128,
      "step": 14935
    },
    {
      "epoch": 4.860117111255693,
      "grad_norm": 2.082951784133911,
      "learning_rate": 1.1169045831079805e-07,
      "loss": 0.6119,
      "step": 14940
    },
    {
      "epoch": 4.86174365647365,
      "grad_norm": 2.3723201751708984,
      "learning_rate": 1.0926688175438571e-07,
      "loss": 0.5465,
      "step": 14945
    },
    {
      "epoch": 4.863370201691607,
      "grad_norm": 2.3645992279052734,
      "learning_rate": 1.0686983148133489e-07,
      "loss": 0.5899,
      "step": 14950
    },
    {
      "epoch": 4.864996746909564,
      "grad_norm": 2.1711809635162354,
      "learning_rate": 1.0449931004620495e-07,
      "loss": 0.5932,
      "step": 14955
    },
    {
      "epoch": 4.866623292127521,
      "grad_norm": 2.588343381881714,
      "learning_rate": 1.0215531997528338e-07,
      "loss": 0.6134,
      "step": 14960
    },
    {
      "epoch": 4.868249837345478,
      "grad_norm": 2.0764243602752686,
      "learning_rate": 9.983786376657755e-08,
      "loss": 0.5932,
      "step": 14965
    },
    {
      "epoch": 4.869876382563435,
      "grad_norm": 2.267519235610962,
      "learning_rate": 9.754694388982854e-08,
      "loss": 0.6049,
      "step": 14970
    },
    {
      "epoch": 4.871502927781393,
      "grad_norm": 2.4809203147888184,
      "learning_rate": 9.52825627864834e-08,
      "loss": 0.6334,
      "step": 14975
    },
    {
      "epoch": 4.8731294729993495,
      "grad_norm": 2.4457435607910156,
      "learning_rate": 9.304472286971733e-08,
      "loss": 0.5986,
      "step": 14980
    },
    {
      "epoch": 4.874756018217306,
      "grad_norm": 11.295981407165527,
      "learning_rate": 9.083342652441706e-08,
      "loss": 0.5956,
      "step": 14985
    },
    {
      "epoch": 4.876382563435263,
      "grad_norm": 1.9970314502716064,
      "learning_rate": 8.864867610718363e-08,
      "loss": 0.6116,
      "step": 14990
    },
    {
      "epoch": 4.87800910865322,
      "grad_norm": 4.389446258544922,
      "learning_rate": 8.649047394632126e-08,
      "loss": 0.6008,
      "step": 14995
    },
    {
      "epoch": 4.879635653871178,
      "grad_norm": 2.124135732650757,
      "learning_rate": 8.435882234184844e-08,
      "loss": 0.5881,
      "step": 15000
    },
    {
      "epoch": 4.881262199089135,
      "grad_norm": 2.077362298965454,
      "learning_rate": 8.225372356548689e-08,
      "loss": 0.5928,
      "step": 15005
    },
    {
      "epoch": 4.882888744307092,
      "grad_norm": 2.21614408493042,
      "learning_rate": 8.017517986065593e-08,
      "loss": 0.6198,
      "step": 15010
    },
    {
      "epoch": 4.884515289525049,
      "grad_norm": 2.278759479522705,
      "learning_rate": 7.812319344248365e-08,
      "loss": 0.6096,
      "step": 15015
    },
    {
      "epoch": 4.886141834743006,
      "grad_norm": 2.567445993423462,
      "learning_rate": 7.609776649778466e-08,
      "loss": 0.6011,
      "step": 15020
    },
    {
      "epoch": 4.887768379960963,
      "grad_norm": 2.1544313430786133,
      "learning_rate": 7.409890118508234e-08,
      "loss": 0.6016,
      "step": 15025
    },
    {
      "epoch": 4.8893949251789195,
      "grad_norm": 1.9982365369796753,
      "learning_rate": 7.212659963458101e-08,
      "loss": 0.6215,
      "step": 15030
    },
    {
      "epoch": 4.891021470396877,
      "grad_norm": 2.2209675312042236,
      "learning_rate": 7.018086394817991e-08,
      "loss": 0.577,
      "step": 15035
    },
    {
      "epoch": 4.892648015614834,
      "grad_norm": 2.1577842235565186,
      "learning_rate": 6.826169619947032e-08,
      "loss": 0.5865,
      "step": 15040
    },
    {
      "epoch": 4.894274560832791,
      "grad_norm": 1.9539618492126465,
      "learning_rate": 6.63690984337273e-08,
      "loss": 0.5889,
      "step": 15045
    },
    {
      "epoch": 4.895901106050748,
      "grad_norm": 2.0748941898345947,
      "learning_rate": 6.45030726679069e-08,
      "loss": 0.5853,
      "step": 15050
    },
    {
      "epoch": 4.897527651268705,
      "grad_norm": 2.1897246837615967,
      "learning_rate": 6.266362089065448e-08,
      "loss": 0.6013,
      "step": 15055
    },
    {
      "epoch": 4.899154196486663,
      "grad_norm": 4.679499626159668,
      "learning_rate": 6.085074506228528e-08,
      "loss": 0.615,
      "step": 15060
    },
    {
      "epoch": 4.90078074170462,
      "grad_norm": 2.0859599113464355,
      "learning_rate": 5.90644471147983e-08,
      "loss": 0.5736,
      "step": 15065
    },
    {
      "epoch": 4.9024072869225765,
      "grad_norm": 2.305415153503418,
      "learning_rate": 5.730472895187355e-08,
      "loss": 0.6139,
      "step": 15070
    },
    {
      "epoch": 4.9040338321405335,
      "grad_norm": 2.4055914878845215,
      "learning_rate": 5.557159244885257e-08,
      "loss": 0.6172,
      "step": 15075
    },
    {
      "epoch": 4.90566037735849,
      "grad_norm": 4.010295867919922,
      "learning_rate": 5.386503945275789e-08,
      "loss": 0.615,
      "step": 15080
    },
    {
      "epoch": 4.907286922576447,
      "grad_norm": 1.898116946220398,
      "learning_rate": 5.2185071782276385e-08,
      "loss": 0.5578,
      "step": 15085
    },
    {
      "epoch": 4.908913467794405,
      "grad_norm": 2.1695468425750732,
      "learning_rate": 5.053169122776757e-08,
      "loss": 0.6188,
      "step": 15090
    },
    {
      "epoch": 4.910540013012362,
      "grad_norm": 2.376075506210327,
      "learning_rate": 4.8904899551255326e-08,
      "loss": 0.608,
      "step": 15095
    },
    {
      "epoch": 4.912166558230319,
      "grad_norm": 2.4588913917541504,
      "learning_rate": 4.730469848642505e-08,
      "loss": 0.5981,
      "step": 15100
    },
    {
      "epoch": 4.913793103448276,
      "grad_norm": 2.42128849029541,
      "learning_rate": 4.573108973862095e-08,
      "loss": 0.6244,
      "step": 15105
    },
    {
      "epoch": 4.915419648666233,
      "grad_norm": 2.012868642807007,
      "learning_rate": 4.418407498485988e-08,
      "loss": 0.585,
      "step": 15110
    },
    {
      "epoch": 4.91704619388419,
      "grad_norm": 2.3483657836914062,
      "learning_rate": 4.2663655873806385e-08,
      "loss": 0.6012,
      "step": 15115
    },
    {
      "epoch": 4.918672739102147,
      "grad_norm": 2.9775454998016357,
      "learning_rate": 4.116983402578656e-08,
      "loss": 0.5937,
      "step": 15120
    },
    {
      "epoch": 4.920299284320104,
      "grad_norm": 2.1648337841033936,
      "learning_rate": 3.9702611032776946e-08,
      "loss": 0.6045,
      "step": 15125
    },
    {
      "epoch": 4.921925829538061,
      "grad_norm": 2.8775291442871094,
      "learning_rate": 3.826198845841289e-08,
      "loss": 0.5842,
      "step": 15130
    },
    {
      "epoch": 4.923552374756018,
      "grad_norm": 2.1625959873199463,
      "learning_rate": 3.684796783798017e-08,
      "loss": 0.6006,
      "step": 15135
    },
    {
      "epoch": 4.925178919973975,
      "grad_norm": 2.0495049953460693,
      "learning_rate": 3.546055067840948e-08,
      "loss": 0.5935,
      "step": 15140
    },
    {
      "epoch": 4.926805465191932,
      "grad_norm": 2.2851064205169678,
      "learning_rate": 3.409973845829029e-08,
      "loss": 0.6197,
      "step": 15145
    },
    {
      "epoch": 4.92843201040989,
      "grad_norm": 2.15488338470459,
      "learning_rate": 3.2765532627845874e-08,
      "loss": 0.6117,
      "step": 15150
    },
    {
      "epoch": 4.930058555627847,
      "grad_norm": 1.9095077514648438,
      "learning_rate": 3.145793460895552e-08,
      "loss": 0.6169,
      "step": 15155
    },
    {
      "epoch": 4.931685100845804,
      "grad_norm": 2.2632343769073486,
      "learning_rate": 3.017694579514063e-08,
      "loss": 0.638,
      "step": 15160
    },
    {
      "epoch": 4.9333116460637605,
      "grad_norm": 3.9604759216308594,
      "learning_rate": 2.8922567551556424e-08,
      "loss": 0.6118,
      "step": 15165
    },
    {
      "epoch": 4.934938191281717,
      "grad_norm": 2.236368417739868,
      "learning_rate": 2.7694801215011333e-08,
      "loss": 0.6115,
      "step": 15170
    },
    {
      "epoch": 4.936564736499674,
      "grad_norm": 2.2910823822021484,
      "learning_rate": 2.6493648093942058e-08,
      "loss": 0.6125,
      "step": 15175
    },
    {
      "epoch": 4.938191281717632,
      "grad_norm": 2.1260986328125,
      "learning_rate": 2.531910946843574e-08,
      "loss": 0.605,
      "step": 15180
    },
    {
      "epoch": 4.939817826935589,
      "grad_norm": 2.2453088760375977,
      "learning_rate": 2.4171186590202233e-08,
      "loss": 0.6114,
      "step": 15185
    },
    {
      "epoch": 4.941444372153546,
      "grad_norm": 2.0122714042663574,
      "learning_rate": 2.3049880682593504e-08,
      "loss": 0.6324,
      "step": 15190
    },
    {
      "epoch": 4.943070917371503,
      "grad_norm": 2.417142629623413,
      "learning_rate": 2.1955192940600887e-08,
      "loss": 0.6188,
      "step": 15195
    },
    {
      "epoch": 4.94469746258946,
      "grad_norm": 2.1205084323883057,
      "learning_rate": 2.0887124530841183e-08,
      "loss": 0.609,
      "step": 15200
    },
    {
      "epoch": 4.9463240078074175,
      "grad_norm": 2.5445995330810547,
      "learning_rate": 1.9845676591559446e-08,
      "loss": 0.6005,
      "step": 15205
    },
    {
      "epoch": 4.9479505530253745,
      "grad_norm": 2.3689987659454346,
      "learning_rate": 1.8830850232645636e-08,
      "loss": 0.6141,
      "step": 15210
    },
    {
      "epoch": 4.949577098243331,
      "grad_norm": 2.399352788925171,
      "learning_rate": 1.7842646535601305e-08,
      "loss": 0.6107,
      "step": 15215
    },
    {
      "epoch": 4.951203643461288,
      "grad_norm": 1.9902371168136597,
      "learning_rate": 1.688106655356736e-08,
      "loss": 0.5823,
      "step": 15220
    },
    {
      "epoch": 4.952830188679245,
      "grad_norm": 2.2933974266052246,
      "learning_rate": 1.5946111311310186e-08,
      "loss": 0.5969,
      "step": 15225
    },
    {
      "epoch": 4.954456733897202,
      "grad_norm": 2.3006935119628906,
      "learning_rate": 1.5037781805218863e-08,
      "loss": 0.6092,
      "step": 15230
    },
    {
      "epoch": 4.956083279115159,
      "grad_norm": 1.9131032228469849,
      "learning_rate": 1.4156079003307953e-08,
      "loss": 0.5738,
      "step": 15235
    },
    {
      "epoch": 4.957709824333117,
      "grad_norm": 2.205113172531128,
      "learning_rate": 1.330100384521471e-08,
      "loss": 0.6286,
      "step": 15240
    },
    {
      "epoch": 4.959336369551074,
      "grad_norm": 2.259881019592285,
      "learning_rate": 1.247255724220464e-08,
      "loss": 0.5911,
      "step": 15245
    },
    {
      "epoch": 4.960962914769031,
      "grad_norm": 3.1508960723876953,
      "learning_rate": 1.1670740077157626e-08,
      "loss": 0.6078,
      "step": 15250
    },
    {
      "epoch": 4.9625894599869875,
      "grad_norm": 2.3667948246002197,
      "learning_rate": 1.0895553204579024e-08,
      "loss": 0.6068,
      "step": 15255
    },
    {
      "epoch": 4.964216005204944,
      "grad_norm": 2.3552937507629395,
      "learning_rate": 1.0146997450591333e-08,
      "loss": 0.6335,
      "step": 15260
    },
    {
      "epoch": 4.965842550422902,
      "grad_norm": 2.3416175842285156,
      "learning_rate": 9.42507361293976e-09,
      "loss": 0.6005,
      "step": 15265
    },
    {
      "epoch": 4.967469095640859,
      "grad_norm": 2.283876419067383,
      "learning_rate": 8.729782460981106e-09,
      "loss": 0.5736,
      "step": 15270
    },
    {
      "epoch": 4.969095640858816,
      "grad_norm": 2.2261712551116943,
      "learning_rate": 8.061124735697645e-09,
      "loss": 0.5841,
      "step": 15275
    },
    {
      "epoch": 4.970722186076773,
      "grad_norm": 2.4273598194122314,
      "learning_rate": 7.4191011496832495e-09,
      "loss": 0.6333,
      "step": 15280
    },
    {
      "epoch": 4.97234873129473,
      "grad_norm": 2.3165061473846436,
      "learning_rate": 6.803712387146166e-09,
      "loss": 0.6144,
      "step": 15285
    },
    {
      "epoch": 4.973975276512687,
      "grad_norm": 2.3885746002197266,
      "learning_rate": 6.2149591039145635e-09,
      "loss": 0.6086,
      "step": 15290
    },
    {
      "epoch": 4.975601821730644,
      "grad_norm": 2.53422212600708,
      "learning_rate": 5.652841927425434e-09,
      "loss": 0.5898,
      "step": 15295
    },
    {
      "epoch": 4.9772283669486015,
      "grad_norm": 2.120922327041626,
      "learning_rate": 5.117361456735692e-09,
      "loss": 0.6188,
      "step": 15300
    },
    {
      "epoch": 4.978854912166558,
      "grad_norm": 1.9496095180511475,
      "learning_rate": 4.7081558927991594e-09,
      "loss": 0.5899,
      "step": 15305
    },
    {
      "epoch": 4.980481457384515,
      "grad_norm": 2.3398711681365967,
      "learning_rate": 4.220622911546568e-09,
      "loss": 0.59,
      "step": 15310
    },
    {
      "epoch": 4.982108002602472,
      "grad_norm": 2.346818685531616,
      "learning_rate": 3.759728162422427e-09,
      "loss": 0.5875,
      "step": 15315
    },
    {
      "epoch": 4.983734547820429,
      "grad_norm": 2.2996296882629395,
      "learning_rate": 3.3254721366032805e-09,
      "loss": 0.5574,
      "step": 15320
    },
    {
      "epoch": 4.985361093038387,
      "grad_norm": 2.248450756072998,
      "learning_rate": 2.9178552968800454e-09,
      "loss": 0.5798,
      "step": 15325
    },
    {
      "epoch": 4.986987638256344,
      "grad_norm": 2.0372941493988037,
      "learning_rate": 2.536878077655236e-09,
      "loss": 0.5893,
      "step": 15330
    },
    {
      "epoch": 4.988614183474301,
      "grad_norm": 2.27799654006958,
      "learning_rate": 2.1825408849401873e-09,
      "loss": 0.6261,
      "step": 15335
    },
    {
      "epoch": 4.990240728692258,
      "grad_norm": 2.2353765964508057,
      "learning_rate": 1.8548440963522818e-09,
      "loss": 0.5939,
      "step": 15340
    },
    {
      "epoch": 4.991867273910215,
      "grad_norm": 1.94557523727417,
      "learning_rate": 1.5537880611260491e-09,
      "loss": 0.6064,
      "step": 15345
    },
    {
      "epoch": 4.9934938191281715,
      "grad_norm": 2.407500743865967,
      "learning_rate": 1.2793731000937393e-09,
      "loss": 0.6106,
      "step": 15350
    },
    {
      "epoch": 4.995120364346128,
      "grad_norm": 2.1297216415405273,
      "learning_rate": 1.0315995057075256e-09,
      "loss": 0.6218,
      "step": 15355
    },
    {
      "epoch": 4.996746909564086,
      "grad_norm": 2.2248427867889404,
      "learning_rate": 8.104675420173014e-10,
      "loss": 0.6253,
      "step": 15360
    },
    {
      "epoch": 4.998373454782043,
      "grad_norm": 2.113090753555298,
      "learning_rate": 6.159774446901079e-10,
      "loss": 0.607,
      "step": 15365
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.143172025680542,
      "learning_rate": 4.481294209907061e-10,
      "loss": 0.5905,
      "step": 15370
    },
    {
      "epoch": 5.0,
      "eval_f1": 0.8242242626847417,
      "eval_loss": 0.43017578125,
      "eval_precision": 0.8247019556006198,
      "eval_recall": 0.8238432612482821,
      "eval_runtime": 1050.5502,
      "eval_samples_per_second": 374.504,
      "eval_steps_per_second": 0.732,
      "step": 15370
    },
    {
      "epoch": 5.0,
      "step": 15370,
      "total_flos": 1.4665425515846828e+19,
      "train_loss": 0.7460857761162166,
      "train_runtime": 176390.3873,
      "train_samples_per_second": 89.219,
      "train_steps_per_second": 0.087
    }
  ],
  "logging_steps": 5,
  "max_steps": 15370,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 5.0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.4665425515846828e+19,
  "train_batch_size": 512,
  "trial_name": null,
  "trial_params": null
}