{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8617146809895834,
  "eval_steps": 24576,
  "global_step": 84710,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0172526041666666e-05,
      "eval_loss": 4.542403221130371,
      "eval_runtime": 144.6811,
      "eval_samples_per_second": 13.872,
      "eval_steps_per_second": 13.872,
      "step": 1
    },
    {
      "epoch": 5.0862630208333336e-05,
      "grad_norm": 45.55976867675781,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 4.478,
      "step": 5
    },
    {
      "epoch": 0.00010172526041666667,
      "grad_norm": 44.849979400634766,
      "learning_rate": 5.000000000000001e-07,
      "loss": 4.381,
      "step": 10
    },
    {
      "epoch": 0.000152587890625,
      "grad_norm": 47.78006362915039,
      "learning_rate": 7.5e-07,
      "loss": 4.2726,
      "step": 15
    },
    {
      "epoch": 0.00020345052083333334,
      "grad_norm": 49.011878967285156,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 4.4009,
      "step": 20
    },
    {
      "epoch": 0.0002543131510416667,
      "grad_norm": 40.69770050048828,
      "learning_rate": 1.25e-06,
      "loss": 4.4735,
      "step": 25
    },
    {
      "epoch": 0.00030517578125,
      "grad_norm": 34.35673522949219,
      "learning_rate": 1.5e-06,
      "loss": 4.2665,
      "step": 30
    },
    {
      "epoch": 0.0003560384114583333,
      "grad_norm": 52.81883239746094,
      "learning_rate": 1.75e-06,
      "loss": 4.423,
      "step": 35
    },
    {
      "epoch": 0.0004069010416666667,
      "grad_norm": 32.41872024536133,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 4.6442,
      "step": 40
    },
    {
      "epoch": 0.000457763671875,
      "grad_norm": 37.73821258544922,
      "learning_rate": 2.25e-06,
      "loss": 4.2681,
      "step": 45
    },
    {
      "epoch": 0.0005086263020833334,
      "grad_norm": 37.740386962890625,
      "learning_rate": 2.5e-06,
      "loss": 4.6894,
      "step": 50
    },
    {
      "epoch": 0.0005594889322916666,
      "grad_norm": 50.862735748291016,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 4.3243,
      "step": 55
    },
    {
      "epoch": 0.0006103515625,
      "grad_norm": 45.02497863769531,
      "learning_rate": 3e-06,
      "loss": 4.3276,
      "step": 60
    },
    {
      "epoch": 0.0006612141927083334,
      "grad_norm": 28.076095581054688,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 3.8881,
      "step": 65
    },
    {
      "epoch": 0.0007120768229166666,
      "grad_norm": 27.940998077392578,
      "learning_rate": 3.5e-06,
      "loss": 4.2496,
      "step": 70
    },
    {
      "epoch": 0.000762939453125,
      "grad_norm": 26.482099533081055,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 4.349,
      "step": 75
    },
    {
      "epoch": 0.0008138020833333334,
      "grad_norm": 32.36627960205078,
      "learning_rate": 4.000000000000001e-06,
      "loss": 4.423,
      "step": 80
    },
    {
      "epoch": 0.0008646647135416666,
      "grad_norm": 48.71563720703125,
      "learning_rate": 4.25e-06,
      "loss": 4.65,
      "step": 85
    },
    {
      "epoch": 0.00091552734375,
      "grad_norm": 25.62880516052246,
      "learning_rate": 4.5e-06,
      "loss": 4.2165,
      "step": 90
    },
    {
      "epoch": 0.0009663899739583334,
      "grad_norm": 33.9738655090332,
      "learning_rate": 4.75e-06,
      "loss": 4.1882,
      "step": 95
    },
    {
      "epoch": 0.0010172526041666667,
      "grad_norm": 24.544464111328125,
      "learning_rate": 5e-06,
      "loss": 3.7919,
      "step": 100
    },
    {
      "epoch": 0.001068115234375,
      "grad_norm": 38.953922271728516,
      "learning_rate": 4.999999968019047e-06,
      "loss": 4.5549,
      "step": 105
    },
    {
      "epoch": 0.0011189778645833333,
      "grad_norm": 30.3071346282959,
      "learning_rate": 4.999999872076186e-06,
      "loss": 4.3079,
      "step": 110
    },
    {
      "epoch": 0.0011698404947916667,
      "grad_norm": 26.42899513244629,
      "learning_rate": 4.999999712171422e-06,
      "loss": 4.3139,
      "step": 115
    },
    {
      "epoch": 0.001220703125,
      "grad_norm": 18.36640739440918,
      "learning_rate": 4.999999488304758e-06,
      "loss": 4.1012,
      "step": 120
    },
    {
      "epoch": 0.0012715657552083333,
      "grad_norm": 22.997697830200195,
      "learning_rate": 4.999999200476199e-06,
      "loss": 4.0616,
      "step": 125
    },
    {
      "epoch": 0.0013224283854166667,
      "grad_norm": 18.358749389648438,
      "learning_rate": 4.999998848685752e-06,
      "loss": 4.0783,
      "step": 130
    },
    {
      "epoch": 0.001373291015625,
      "grad_norm": 16.22219467163086,
      "learning_rate": 4.999998432933428e-06,
      "loss": 4.14,
      "step": 135
    },
    {
      "epoch": 0.0014241536458333333,
      "grad_norm": 31.525222778320312,
      "learning_rate": 4.999997953219238e-06,
      "loss": 4.268,
      "step": 140
    },
    {
      "epoch": 0.0014750162760416667,
      "grad_norm": 16.66249656677246,
      "learning_rate": 4.999997409543191e-06,
      "loss": 3.7311,
      "step": 145
    },
    {
      "epoch": 0.00152587890625,
      "grad_norm": 19.815828323364258,
      "learning_rate": 4.999996801905304e-06,
      "loss": 4.0011,
      "step": 150
    },
    {
      "epoch": 0.0015767415364583333,
      "grad_norm": 27.2629337310791,
      "learning_rate": 4.9999961303055906e-06,
      "loss": 4.4585,
      "step": 155
    },
    {
      "epoch": 0.0016276041666666667,
      "grad_norm": 20.656837463378906,
      "learning_rate": 4.99999539474407e-06,
      "loss": 3.6024,
      "step": 160
    },
    {
      "epoch": 0.001678466796875,
      "grad_norm": 32.12629699707031,
      "learning_rate": 4.999994595220758e-06,
      "loss": 3.6387,
      "step": 165
    },
    {
      "epoch": 0.0017293294270833333,
      "grad_norm": 17.384239196777344,
      "learning_rate": 4.9999937317356776e-06,
      "loss": 3.7885,
      "step": 170
    },
    {
      "epoch": 0.0017801920572916667,
      "grad_norm": 16.839290618896484,
      "learning_rate": 4.99999280428885e-06,
      "loss": 3.9351,
      "step": 175
    },
    {
      "epoch": 0.0018310546875,
      "grad_norm": 17.16065216064453,
      "learning_rate": 4.9999918128803e-06,
      "loss": 3.829,
      "step": 180
    },
    {
      "epoch": 0.0018819173177083333,
      "grad_norm": 19.670543670654297,
      "learning_rate": 4.999990757510052e-06,
      "loss": 3.9205,
      "step": 185
    },
    {
      "epoch": 0.0019327799479166667,
      "grad_norm": 19.776790618896484,
      "learning_rate": 4.999989638178131e-06,
      "loss": 4.0733,
      "step": 190
    },
    {
      "epoch": 0.001983642578125,
      "grad_norm": 17.121217727661133,
      "learning_rate": 4.99998845488457e-06,
      "loss": 3.9968,
      "step": 195
    },
    {
      "epoch": 0.0020345052083333335,
      "grad_norm": 18.365943908691406,
      "learning_rate": 4.999987207629396e-06,
      "loss": 3.6687,
      "step": 200
    },
    {
      "epoch": 0.0020853678385416665,
      "grad_norm": 17.672025680541992,
      "learning_rate": 4.9999858964126415e-06,
      "loss": 3.9062,
      "step": 205
    },
    {
      "epoch": 0.00213623046875,
      "grad_norm": 14.973541259765625,
      "learning_rate": 4.9999845212343415e-06,
      "loss": 3.4274,
      "step": 210
    },
    {
      "epoch": 0.0021870930989583335,
      "grad_norm": 20.310712814331055,
      "learning_rate": 4.999983082094529e-06,
      "loss": 3.8899,
      "step": 215
    },
    {
      "epoch": 0.0022379557291666665,
      "grad_norm": 30.07281494140625,
      "learning_rate": 4.999981578993242e-06,
      "loss": 4.083,
      "step": 220
    },
    {
      "epoch": 0.002288818359375,
      "grad_norm": 14.967458724975586,
      "learning_rate": 4.999980011930519e-06,
      "loss": 3.8021,
      "step": 225
    },
    {
      "epoch": 0.0023396809895833335,
      "grad_norm": 18.596702575683594,
      "learning_rate": 4.999978380906401e-06,
      "loss": 4.2005,
      "step": 230
    },
    {
      "epoch": 0.0023905436197916665,
      "grad_norm": 14.92667293548584,
      "learning_rate": 4.999976685920927e-06,
      "loss": 3.7878,
      "step": 235
    },
    {
      "epoch": 0.00244140625,
      "grad_norm": 26.372528076171875,
      "learning_rate": 4.999974926974142e-06,
      "loss": 3.906,
      "step": 240
    },
    {
      "epoch": 0.0024922688802083335,
      "grad_norm": 11.505077362060547,
      "learning_rate": 4.9999731040660925e-06,
      "loss": 3.7275,
      "step": 245
    },
    {
      "epoch": 0.0025431315104166665,
      "grad_norm": 19.219104766845703,
      "learning_rate": 4.999971217196824e-06,
      "loss": 3.6951,
      "step": 250
    },
    {
      "epoch": 0.002593994140625,
      "grad_norm": 20.686763763427734,
      "learning_rate": 4.999969266366383e-06,
      "loss": 4.2166,
      "step": 255
    },
    {
      "epoch": 0.0026448567708333335,
      "grad_norm": 14.91057300567627,
      "learning_rate": 4.999967251574821e-06,
      "loss": 4.1096,
      "step": 260
    },
    {
      "epoch": 0.0026957194010416665,
      "grad_norm": 14.4781494140625,
      "learning_rate": 4.99996517282219e-06,
      "loss": 3.9429,
      "step": 265
    },
    {
      "epoch": 0.00274658203125,
      "grad_norm": 16.21483039855957,
      "learning_rate": 4.9999630301085425e-06,
      "loss": 4.5956,
      "step": 270
    },
    {
      "epoch": 0.0027974446614583335,
      "grad_norm": 20.831764221191406,
      "learning_rate": 4.9999608234339336e-06,
      "loss": 4.0729,
      "step": 275
    },
    {
      "epoch": 0.0028483072916666665,
      "grad_norm": 16.851608276367188,
      "learning_rate": 4.999958552798419e-06,
      "loss": 3.9821,
      "step": 280
    },
    {
      "epoch": 0.002899169921875,
      "grad_norm": 16.131776809692383,
      "learning_rate": 4.999956218202058e-06,
      "loss": 3.912,
      "step": 285
    },
    {
      "epoch": 0.0029500325520833335,
      "grad_norm": 22.348773956298828,
      "learning_rate": 4.9999538196449096e-06,
      "loss": 3.7261,
      "step": 290
    },
    {
      "epoch": 0.0030008951822916665,
      "grad_norm": 27.426599502563477,
      "learning_rate": 4.9999513571270355e-06,
      "loss": 3.9633,
      "step": 295
    },
    {
      "epoch": 0.0030517578125,
      "grad_norm": 19.74297523498535,
      "learning_rate": 4.999948830648497e-06,
      "loss": 3.6241,
      "step": 300
    },
    {
      "epoch": 0.0031026204427083335,
      "grad_norm": 26.0491943359375,
      "learning_rate": 4.999946240209362e-06,
      "loss": 3.8093,
      "step": 305
    },
    {
      "epoch": 0.0031534830729166665,
      "grad_norm": 17.24481964111328,
      "learning_rate": 4.999943585809694e-06,
      "loss": 3.9762,
      "step": 310
    },
    {
      "epoch": 0.003204345703125,
      "grad_norm": 16.89327621459961,
      "learning_rate": 4.999940867449562e-06,
      "loss": 3.5947,
      "step": 315
    },
    {
      "epoch": 0.0032552083333333335,
      "grad_norm": 19.533809661865234,
      "learning_rate": 4.999938085129036e-06,
      "loss": 3.5673,
      "step": 320
    },
    {
      "epoch": 0.0033060709635416665,
      "grad_norm": 16.76618766784668,
      "learning_rate": 4.999935238848187e-06,
      "loss": 3.8095,
      "step": 325
    },
    {
      "epoch": 0.00335693359375,
      "grad_norm": 15.267281532287598,
      "learning_rate": 4.999932328607087e-06,
      "loss": 3.8809,
      "step": 330
    },
    {
      "epoch": 0.0034077962239583335,
      "grad_norm": 14.550617218017578,
      "learning_rate": 4.999929354405811e-06,
      "loss": 3.9913,
      "step": 335
    },
    {
      "epoch": 0.0034586588541666665,
      "grad_norm": 22.565776824951172,
      "learning_rate": 4.999926316244434e-06,
      "loss": 4.0828,
      "step": 340
    },
    {
      "epoch": 0.003509521484375,
      "grad_norm": 14.583137512207031,
      "learning_rate": 4.999923214123036e-06,
      "loss": 4.0128,
      "step": 345
    },
    {
      "epoch": 0.0035603841145833335,
      "grad_norm": 14.065851211547852,
      "learning_rate": 4.999920048041694e-06,
      "loss": 3.5976,
      "step": 350
    },
    {
      "epoch": 0.0036112467447916665,
      "grad_norm": 22.18216323852539,
      "learning_rate": 4.999916818000491e-06,
      "loss": 3.8117,
      "step": 355
    },
    {
      "epoch": 0.003662109375,
      "grad_norm": 21.645153045654297,
      "learning_rate": 4.9999135239995076e-06,
      "loss": 3.5906,
      "step": 360
    },
    {
      "epoch": 0.0037129720052083335,
      "grad_norm": 13.195281028747559,
      "learning_rate": 4.9999101660388305e-06,
      "loss": 3.7585,
      "step": 365
    },
    {
      "epoch": 0.0037638346354166665,
      "grad_norm": 10.816890716552734,
      "learning_rate": 4.999906744118545e-06,
      "loss": 3.7736,
      "step": 370
    },
    {
      "epoch": 0.003814697265625,
      "grad_norm": 14.775542259216309,
      "learning_rate": 4.999903258238736e-06,
      "loss": 3.6809,
      "step": 375
    },
    {
      "epoch": 0.0038655598958333335,
      "grad_norm": 22.4163761138916,
      "learning_rate": 4.999899708399496e-06,
      "loss": 3.6198,
      "step": 380
    },
    {
      "epoch": 0.003916422526041667,
      "grad_norm": 20.934917449951172,
      "learning_rate": 4.999896094600914e-06,
      "loss": 3.7003,
      "step": 385
    },
    {
      "epoch": 0.00396728515625,
      "grad_norm": 13.397461891174316,
      "learning_rate": 4.999892416843083e-06,
      "loss": 3.9415,
      "step": 390
    },
    {
      "epoch": 0.004018147786458333,
      "grad_norm": 18.704856872558594,
      "learning_rate": 4.999888675126097e-06,
      "loss": 3.7429,
      "step": 395
    },
    {
      "epoch": 0.004069010416666667,
      "grad_norm": 14.650379180908203,
      "learning_rate": 4.9998848694500526e-06,
      "loss": 3.8455,
      "step": 400
    },
    {
      "epoch": 0.004119873046875,
      "grad_norm": 22.322834014892578,
      "learning_rate": 4.999880999815045e-06,
      "loss": 3.9125,
      "step": 405
    },
    {
      "epoch": 0.004170735677083333,
      "grad_norm": 16.105815887451172,
      "learning_rate": 4.999877066221175e-06,
      "loss": 3.5135,
      "step": 410
    },
    {
      "epoch": 0.004221598307291667,
      "grad_norm": 10.558919906616211,
      "learning_rate": 4.999873068668544e-06,
      "loss": 3.6677,
      "step": 415
    },
    {
      "epoch": 0.0042724609375,
      "grad_norm": 10.016486167907715,
      "learning_rate": 4.999869007157252e-06,
      "loss": 3.5761,
      "step": 420
    },
    {
      "epoch": 0.004323323567708333,
      "grad_norm": 16.01995849609375,
      "learning_rate": 4.999864881687404e-06,
      "loss": 3.9739,
      "step": 425
    },
    {
      "epoch": 0.004374186197916667,
      "grad_norm": 21.719324111938477,
      "learning_rate": 4.999860692259105e-06,
      "loss": 3.9211,
      "step": 430
    },
    {
      "epoch": 0.004425048828125,
      "grad_norm": 42.85234069824219,
      "learning_rate": 4.999856438872463e-06,
      "loss": 3.9438,
      "step": 435
    },
    {
      "epoch": 0.004475911458333333,
      "grad_norm": 13.891044616699219,
      "learning_rate": 4.999852121527588e-06,
      "loss": 3.6916,
      "step": 440
    },
    {
      "epoch": 0.004526774088541667,
      "grad_norm": 18.872058868408203,
      "learning_rate": 4.999847740224587e-06,
      "loss": 4.0558,
      "step": 445
    },
    {
      "epoch": 0.00457763671875,
      "grad_norm": 16.054645538330078,
      "learning_rate": 4.999843294963576e-06,
      "loss": 3.8301,
      "step": 450
    },
    {
      "epoch": 0.004628499348958333,
      "grad_norm": 14.235097885131836,
      "learning_rate": 4.999838785744665e-06,
      "loss": 3.745,
      "step": 455
    },
    {
      "epoch": 0.004679361979166667,
      "grad_norm": 21.37851905822754,
      "learning_rate": 4.999834212567972e-06,
      "loss": 3.8782,
      "step": 460
    },
    {
      "epoch": 0.004730224609375,
      "grad_norm": 17.109268188476562,
      "learning_rate": 4.999829575433613e-06,
      "loss": 3.4868,
      "step": 465
    },
    {
      "epoch": 0.004781087239583333,
      "grad_norm": 16.911109924316406,
      "learning_rate": 4.999824874341708e-06,
      "loss": 3.4223,
      "step": 470
    },
    {
      "epoch": 0.004831949869791667,
      "grad_norm": 9.432239532470703,
      "learning_rate": 4.9998201092923746e-06,
      "loss": 3.3816,
      "step": 475
    },
    {
      "epoch": 0.0048828125,
      "grad_norm": 18.146076202392578,
      "learning_rate": 4.999815280285737e-06,
      "loss": 4.0282,
      "step": 480
    },
    {
      "epoch": 0.004933675130208333,
      "grad_norm": 26.35177993774414,
      "learning_rate": 4.999810387321917e-06,
      "loss": 3.2918,
      "step": 485
    },
    {
      "epoch": 0.004984537760416667,
      "grad_norm": 17.22603988647461,
      "learning_rate": 4.9998054304010425e-06,
      "loss": 3.7617,
      "step": 490
    },
    {
      "epoch": 0.005035400390625,
      "grad_norm": 15.333989143371582,
      "learning_rate": 4.999800409523237e-06,
      "loss": 4.2317,
      "step": 495
    },
    {
      "epoch": 0.005086263020833333,
      "grad_norm": 19.130863189697266,
      "learning_rate": 4.999795324688631e-06,
      "loss": 3.8223,
      "step": 500
    },
    {
      "epoch": 0.005137125651041667,
      "grad_norm": 12.130922317504883,
      "learning_rate": 4.999790175897355e-06,
      "loss": 3.7448,
      "step": 505
    },
    {
      "epoch": 0.00518798828125,
      "grad_norm": 18.195701599121094,
      "learning_rate": 4.999784963149539e-06,
      "loss": 3.8377,
      "step": 510
    },
    {
      "epoch": 0.005238850911458333,
      "grad_norm": 13.157418251037598,
      "learning_rate": 4.999779686445318e-06,
      "loss": 4.2839,
      "step": 515
    },
    {
      "epoch": 0.005289713541666667,
      "grad_norm": 15.427362442016602,
      "learning_rate": 4.999774345784825e-06,
      "loss": 3.6147,
      "step": 520
    },
    {
      "epoch": 0.005340576171875,
      "grad_norm": 18.931623458862305,
      "learning_rate": 4.9997689411681986e-06,
      "loss": 4.246,
      "step": 525
    },
    {
      "epoch": 0.005391438802083333,
      "grad_norm": 16.495134353637695,
      "learning_rate": 4.9997634725955756e-06,
      "loss": 3.811,
      "step": 530
    },
    {
      "epoch": 0.005442301432291667,
      "grad_norm": 17.250377655029297,
      "learning_rate": 4.999757940067098e-06,
      "loss": 3.9962,
      "step": 535
    },
    {
      "epoch": 0.0054931640625,
      "grad_norm": 19.960880279541016,
      "learning_rate": 4.999752343582905e-06,
      "loss": 3.8214,
      "step": 540
    },
    {
      "epoch": 0.005544026692708333,
      "grad_norm": 21.063762664794922,
      "learning_rate": 4.999746683143141e-06,
      "loss": 3.5984,
      "step": 545
    },
    {
      "epoch": 0.005594889322916667,
      "grad_norm": 15.372790336608887,
      "learning_rate": 4.999740958747951e-06,
      "loss": 3.8445,
      "step": 550
    },
    {
      "epoch": 0.005645751953125,
      "grad_norm": 12.668439865112305,
      "learning_rate": 4.9997351703974804e-06,
      "loss": 4.1112,
      "step": 555
    },
    {
      "epoch": 0.005696614583333333,
      "grad_norm": 14.557056427001953,
      "learning_rate": 4.999729318091878e-06,
      "loss": 3.8517,
      "step": 560
    },
    {
      "epoch": 0.005747477213541667,
      "grad_norm": 18.399843215942383,
      "learning_rate": 4.9997234018312945e-06,
      "loss": 3.5826,
      "step": 565
    },
    {
      "epoch": 0.00579833984375,
      "grad_norm": 16.69244956970215,
      "learning_rate": 4.9997174216158795e-06,
      "loss": 3.6708,
      "step": 570
    },
    {
      "epoch": 0.005849202473958333,
      "grad_norm": 20.53217887878418,
      "learning_rate": 4.9997113774457865e-06,
      "loss": 4.0217,
      "step": 575
    },
    {
      "epoch": 0.005900065104166667,
      "grad_norm": 15.122044563293457,
      "learning_rate": 4.999705269321171e-06,
      "loss": 3.8238,
      "step": 580
    },
    {
      "epoch": 0.005950927734375,
      "grad_norm": 9.872886657714844,
      "learning_rate": 4.999699097242189e-06,
      "loss": 3.4889,
      "step": 585
    },
    {
      "epoch": 0.006001790364583333,
      "grad_norm": 14.158080101013184,
      "learning_rate": 4.999692861208997e-06,
      "loss": 3.5411,
      "step": 590
    },
    {
      "epoch": 0.006052652994791667,
      "grad_norm": 15.669339179992676,
      "learning_rate": 4.999686561221756e-06,
      "loss": 3.936,
      "step": 595
    },
    {
      "epoch": 0.006103515625,
      "grad_norm": 11.445816040039062,
      "learning_rate": 4.999680197280628e-06,
      "loss": 3.8597,
      "step": 600
    },
    {
      "epoch": 0.006154378255208333,
      "grad_norm": 16.03506088256836,
      "learning_rate": 4.999673769385773e-06,
      "loss": 3.4291,
      "step": 605
    },
    {
      "epoch": 0.006205240885416667,
      "grad_norm": 13.63383960723877,
      "learning_rate": 4.999667277537358e-06,
      "loss": 3.8566,
      "step": 610
    },
    {
      "epoch": 0.006256103515625,
      "grad_norm": 11.830592155456543,
      "learning_rate": 4.999660721735547e-06,
      "loss": 3.8303,
      "step": 615
    },
    {
      "epoch": 0.006306966145833333,
      "grad_norm": 19.99570655822754,
      "learning_rate": 4.999654101980511e-06,
      "loss": 3.8236,
      "step": 620
    },
    {
      "epoch": 0.006357828776041667,
      "grad_norm": 12.528511047363281,
      "learning_rate": 4.999647418272415e-06,
      "loss": 4.0486,
      "step": 625
    },
    {
      "epoch": 0.00640869140625,
      "grad_norm": 15.760631561279297,
      "learning_rate": 4.999640670611434e-06,
      "loss": 3.4468,
      "step": 630
    },
    {
      "epoch": 0.006459554036458333,
      "grad_norm": 19.84269905090332,
      "learning_rate": 4.999633858997738e-06,
      "loss": 3.8804,
      "step": 635
    },
    {
      "epoch": 0.006510416666666667,
      "grad_norm": 11.239544868469238,
      "learning_rate": 4.999626983431503e-06,
      "loss": 3.5703,
      "step": 640
    },
    {
      "epoch": 0.006561279296875,
      "grad_norm": 13.65317440032959,
      "learning_rate": 4.999620043912904e-06,
      "loss": 3.7155,
      "step": 645
    },
    {
      "epoch": 0.006612141927083333,
      "grad_norm": 12.85306167602539,
      "learning_rate": 4.999613040442118e-06,
      "loss": 3.9372,
      "step": 650
    },
    {
      "epoch": 0.006663004557291667,
      "grad_norm": 16.895217895507812,
      "learning_rate": 4.999605973019325e-06,
      "loss": 4.0348,
      "step": 655
    },
    {
      "epoch": 0.0067138671875,
      "grad_norm": 15.865885734558105,
      "learning_rate": 4.999598841644706e-06,
      "loss": 3.6799,
      "step": 660
    },
    {
      "epoch": 0.006764729817708333,
      "grad_norm": 17.68973731994629,
      "learning_rate": 4.999591646318443e-06,
      "loss": 3.446,
      "step": 665
    },
    {
      "epoch": 0.006815592447916667,
      "grad_norm": 17.351646423339844,
      "learning_rate": 4.99958438704072e-06,
      "loss": 3.9662,
      "step": 670
    },
    {
      "epoch": 0.006866455078125,
      "grad_norm": 12.52562427520752,
      "learning_rate": 4.999577063811723e-06,
      "loss": 3.564,
      "step": 675
    },
    {
      "epoch": 0.006917317708333333,
      "grad_norm": 15.416642189025879,
      "learning_rate": 4.999569676631639e-06,
      "loss": 3.9313,
      "step": 680
    },
    {
      "epoch": 0.006968180338541667,
      "grad_norm": 11.908843994140625,
      "learning_rate": 4.999562225500658e-06,
      "loss": 3.7715,
      "step": 685
    },
    {
      "epoch": 0.00701904296875,
      "grad_norm": 15.781961441040039,
      "learning_rate": 4.999554710418969e-06,
      "loss": 3.467,
      "step": 690
    },
    {
      "epoch": 0.007069905598958333,
      "grad_norm": 17.706050872802734,
      "learning_rate": 4.999547131386766e-06,
      "loss": 3.442,
      "step": 695
    },
    {
      "epoch": 0.007120768229166667,
      "grad_norm": 16.651485443115234,
      "learning_rate": 4.999539488404242e-06,
      "loss": 3.7181,
      "step": 700
    },
    {
      "epoch": 0.007171630859375,
      "grad_norm": 10.602474212646484,
      "learning_rate": 4.9995317814715925e-06,
      "loss": 3.6258,
      "step": 705
    },
    {
      "epoch": 0.007222493489583333,
      "grad_norm": 20.507596969604492,
      "learning_rate": 4.999524010589015e-06,
      "loss": 3.9877,
      "step": 710
    },
    {
      "epoch": 0.007273356119791667,
      "grad_norm": 17.32587432861328,
      "learning_rate": 4.999516175756708e-06,
      "loss": 3.7221,
      "step": 715
    },
    {
      "epoch": 0.00732421875,
      "grad_norm": 11.81552791595459,
      "learning_rate": 4.9995082769748715e-06,
      "loss": 3.4813,
      "step": 720
    },
    {
      "epoch": 0.007375081380208333,
      "grad_norm": 13.437356948852539,
      "learning_rate": 4.9995003142437086e-06,
      "loss": 3.6507,
      "step": 725
    },
    {
      "epoch": 0.007425944010416667,
      "grad_norm": 25.208234786987305,
      "learning_rate": 4.999492287563422e-06,
      "loss": 3.9219,
      "step": 730
    },
    {
      "epoch": 0.007476806640625,
      "grad_norm": 13.51179313659668,
      "learning_rate": 4.999484196934219e-06,
      "loss": 3.5389,
      "step": 735
    },
    {
      "epoch": 0.007527669270833333,
      "grad_norm": 13.544529914855957,
      "learning_rate": 4.999476042356305e-06,
      "loss": 3.597,
      "step": 740
    },
    {
      "epoch": 0.007578531901041667,
      "grad_norm": 10.314176559448242,
      "learning_rate": 4.999467823829888e-06,
      "loss": 3.8758,
      "step": 745
    },
    {
      "epoch": 0.00762939453125,
      "grad_norm": 21.49091148376465,
      "learning_rate": 4.99945954135518e-06,
      "loss": 3.3777,
      "step": 750
    },
    {
      "epoch": 0.007680257161458333,
      "grad_norm": 12.480860710144043,
      "learning_rate": 4.999451194932392e-06,
      "loss": 3.511,
      "step": 755
    },
    {
      "epoch": 0.007731119791666667,
      "grad_norm": 15.50587272644043,
      "learning_rate": 4.999442784561737e-06,
      "loss": 3.6119,
      "step": 760
    },
    {
      "epoch": 0.007781982421875,
      "grad_norm": 17.906938552856445,
      "learning_rate": 4.9994343102434314e-06,
      "loss": 3.8537,
      "step": 765
    },
    {
      "epoch": 0.007832845052083334,
      "grad_norm": 18.00349998474121,
      "learning_rate": 4.9994257719776915e-06,
      "loss": 3.5682,
      "step": 770
    },
    {
      "epoch": 0.007883707682291666,
      "grad_norm": 14.669235229492188,
      "learning_rate": 4.999417169764735e-06,
      "loss": 3.4905,
      "step": 775
    },
    {
      "epoch": 0.0079345703125,
      "grad_norm": 15.826396942138672,
      "learning_rate": 4.999408503604783e-06,
      "loss": 3.9937,
      "step": 780
    },
    {
      "epoch": 0.007985432942708334,
      "grad_norm": 17.248136520385742,
      "learning_rate": 4.999399773498057e-06,
      "loss": 4.14,
      "step": 785
    },
    {
      "epoch": 0.008036295572916666,
      "grad_norm": 16.377283096313477,
      "learning_rate": 4.99939097944478e-06,
      "loss": 4.2206,
      "step": 790
    },
    {
      "epoch": 0.008087158203125,
      "grad_norm": 11.323334693908691,
      "learning_rate": 4.9993821214451774e-06,
      "loss": 4.1021,
      "step": 795
    },
    {
      "epoch": 0.008138020833333334,
      "grad_norm": 14.058307647705078,
      "learning_rate": 4.999373199499476e-06,
      "loss": 3.9362,
      "step": 800
    },
    {
      "epoch": 0.008188883463541666,
      "grad_norm": 17.746889114379883,
      "learning_rate": 4.9993642136079025e-06,
      "loss": 4.0898,
      "step": 805
    },
    {
      "epoch": 0.00823974609375,
      "grad_norm": 17.276458740234375,
      "learning_rate": 4.999355163770688e-06,
      "loss": 3.9971,
      "step": 810
    },
    {
      "epoch": 0.008290608723958334,
      "grad_norm": 20.424314498901367,
      "learning_rate": 4.999346049988065e-06,
      "loss": 3.9069,
      "step": 815
    },
    {
      "epoch": 0.008341471354166666,
      "grad_norm": 21.62326431274414,
      "learning_rate": 4.999336872260266e-06,
      "loss": 3.812,
      "step": 820
    },
    {
      "epoch": 0.008392333984375,
      "grad_norm": 12.865884780883789,
      "learning_rate": 4.999327630587525e-06,
      "loss": 3.7855,
      "step": 825
    },
    {
      "epoch": 0.008443196614583334,
      "grad_norm": 14.172892570495605,
      "learning_rate": 4.999318324970079e-06,
      "loss": 3.8214,
      "step": 830
    },
    {
      "epoch": 0.008494059244791666,
      "grad_norm": 19.895843505859375,
      "learning_rate": 4.999308955408166e-06,
      "loss": 3.5814,
      "step": 835
    },
    {
      "epoch": 0.008544921875,
      "grad_norm": 14.874984741210938,
      "learning_rate": 4.999299521902026e-06,
      "loss": 3.64,
      "step": 840
    },
    {
      "epoch": 0.008595784505208334,
      "grad_norm": 15.312943458557129,
      "learning_rate": 4.9992900244519e-06,
      "loss": 3.7124,
      "step": 845
    },
    {
      "epoch": 0.008646647135416666,
      "grad_norm": 17.663049697875977,
      "learning_rate": 4.999280463058031e-06,
      "loss": 3.3959,
      "step": 850
    },
    {
      "epoch": 0.008697509765625,
      "grad_norm": 19.949892044067383,
      "learning_rate": 4.999270837720663e-06,
      "loss": 3.4157,
      "step": 855
    },
    {
      "epoch": 0.008748372395833334,
      "grad_norm": 14.926891326904297,
      "learning_rate": 4.9992611484400444e-06,
      "loss": 3.4777,
      "step": 860
    },
    {
      "epoch": 0.008799235026041666,
      "grad_norm": 13.239298820495605,
      "learning_rate": 4.999251395216421e-06,
      "loss": 3.6932,
      "step": 865
    },
    {
      "epoch": 0.00885009765625,
      "grad_norm": 16.126522064208984,
      "learning_rate": 4.999241578050044e-06,
      "loss": 3.7468,
      "step": 870
    },
    {
      "epoch": 0.008900960286458334,
      "grad_norm": 26.55894660949707,
      "learning_rate": 4.999231696941162e-06,
      "loss": 3.2945,
      "step": 875
    },
    {
      "epoch": 0.008951822916666666,
      "grad_norm": 16.377870559692383,
      "learning_rate": 4.99922175189003e-06,
      "loss": 3.8092,
      "step": 880
    },
    {
      "epoch": 0.009002685546875,
      "grad_norm": 25.377391815185547,
      "learning_rate": 4.999211742896902e-06,
      "loss": 4.0082,
      "step": 885
    },
    {
      "epoch": 0.009053548177083334,
      "grad_norm": 16.60061264038086,
      "learning_rate": 4.999201669962034e-06,
      "loss": 3.717,
      "step": 890
    },
    {
      "epoch": 0.009104410807291666,
      "grad_norm": 14.67759895324707,
      "learning_rate": 4.999191533085684e-06,
      "loss": 3.5591,
      "step": 895
    },
    {
      "epoch": 0.0091552734375,
      "grad_norm": 10.630406379699707,
      "learning_rate": 4.9991813322681105e-06,
      "loss": 3.6146,
      "step": 900
    },
    {
      "epoch": 0.009206136067708334,
      "grad_norm": 13.745230674743652,
      "learning_rate": 4.999171067509575e-06,
      "loss": 3.5511,
      "step": 905
    },
    {
      "epoch": 0.009256998697916666,
      "grad_norm": 15.126039505004883,
      "learning_rate": 4.99916073881034e-06,
      "loss": 3.6083,
      "step": 910
    },
    {
      "epoch": 0.009307861328125,
      "grad_norm": 18.30549430847168,
      "learning_rate": 4.99915034617067e-06,
      "loss": 3.4733,
      "step": 915
    },
    {
      "epoch": 0.009358723958333334,
      "grad_norm": 12.323296546936035,
      "learning_rate": 4.999139889590832e-06,
      "loss": 3.6615,
      "step": 920
    },
    {
      "epoch": 0.009409586588541666,
      "grad_norm": 19.943538665771484,
      "learning_rate": 4.999129369071091e-06,
      "loss": 4.1944,
      "step": 925
    },
    {
      "epoch": 0.00946044921875,
      "grad_norm": 10.738496780395508,
      "learning_rate": 4.9991187846117175e-06,
      "loss": 3.4904,
      "step": 930
    },
    {
      "epoch": 0.009511311848958334,
      "grad_norm": 11.223962783813477,
      "learning_rate": 4.999108136212982e-06,
      "loss": 3.6925,
      "step": 935
    },
    {
      "epoch": 0.009562174479166666,
      "grad_norm": 13.026283264160156,
      "learning_rate": 4.999097423875158e-06,
      "loss": 4.1644,
      "step": 940
    },
    {
      "epoch": 0.009613037109375,
      "grad_norm": 16.166419982910156,
      "learning_rate": 4.999086647598518e-06,
      "loss": 3.5475,
      "step": 945
    },
    {
      "epoch": 0.009663899739583334,
      "grad_norm": 18.75146484375,
      "learning_rate": 4.999075807383339e-06,
      "loss": 3.4752,
      "step": 950
    },
    {
      "epoch": 0.009714762369791666,
      "grad_norm": 14.348006248474121,
      "learning_rate": 4.999064903229897e-06,
      "loss": 3.5193,
      "step": 955
    },
    {
      "epoch": 0.009765625,
      "grad_norm": 13.05683708190918,
      "learning_rate": 4.9990539351384725e-06,
      "loss": 3.5619,
      "step": 960
    },
    {
      "epoch": 0.009816487630208334,
      "grad_norm": 14.792659759521484,
      "learning_rate": 4.999042903109345e-06,
      "loss": 3.2695,
      "step": 965
    },
    {
      "epoch": 0.009867350260416666,
      "grad_norm": 11.70416259765625,
      "learning_rate": 4.999031807142798e-06,
      "loss": 3.7578,
      "step": 970
    },
    {
      "epoch": 0.009918212890625,
      "grad_norm": 10.735535621643066,
      "learning_rate": 4.999020647239114e-06,
      "loss": 3.8695,
      "step": 975
    },
    {
      "epoch": 0.009969075520833334,
      "grad_norm": 17.912128448486328,
      "learning_rate": 4.999009423398579e-06,
      "loss": 3.5609,
      "step": 980
    },
    {
      "epoch": 0.010019938151041666,
      "grad_norm": 14.438199043273926,
      "learning_rate": 4.99899813562148e-06,
      "loss": 3.7282,
      "step": 985
    },
    {
      "epoch": 0.01007080078125,
      "grad_norm": 15.654008865356445,
      "learning_rate": 4.9989867839081065e-06,
      "loss": 3.8585,
      "step": 990
    },
    {
      "epoch": 0.010121663411458334,
      "grad_norm": 15.130616188049316,
      "learning_rate": 4.998975368258749e-06,
      "loss": 3.4857,
      "step": 995
    },
    {
      "epoch": 0.010172526041666666,
      "grad_norm": 15.098103523254395,
      "learning_rate": 4.998963888673698e-06,
      "loss": 3.7006,
      "step": 1000
    },
    {
      "epoch": 0.010223388671875,
      "grad_norm": 14.421039581298828,
      "learning_rate": 4.998952345153249e-06,
      "loss": 3.7491,
      "step": 1005
    },
    {
      "epoch": 0.010274251302083334,
      "grad_norm": 20.087514877319336,
      "learning_rate": 4.998940737697695e-06,
      "loss": 3.7776,
      "step": 1010
    },
    {
      "epoch": 0.010325113932291666,
      "grad_norm": 18.88400650024414,
      "learning_rate": 4.998929066307336e-06,
      "loss": 3.7481,
      "step": 1015
    },
    {
      "epoch": 0.0103759765625,
      "grad_norm": 21.07455062866211,
      "learning_rate": 4.998917330982469e-06,
      "loss": 3.5294,
      "step": 1020
    },
    {
      "epoch": 0.010426839192708334,
      "grad_norm": 11.424040794372559,
      "learning_rate": 4.998905531723394e-06,
      "loss": 3.9097,
      "step": 1025
    },
    {
      "epoch": 0.010477701822916666,
      "grad_norm": 10.922247886657715,
      "learning_rate": 4.998893668530414e-06,
      "loss": 3.5514,
      "step": 1030
    },
    {
      "epoch": 0.010528564453125,
      "grad_norm": 17.195194244384766,
      "learning_rate": 4.99888174140383e-06,
      "loss": 3.616,
      "step": 1035
    },
    {
      "epoch": 0.010579427083333334,
      "grad_norm": 10.144538879394531,
      "learning_rate": 4.998869750343951e-06,
      "loss": 3.7874,
      "step": 1040
    },
    {
      "epoch": 0.010630289713541666,
      "grad_norm": 11.742879867553711,
      "learning_rate": 4.998857695351081e-06,
      "loss": 3.9271,
      "step": 1045
    },
    {
      "epoch": 0.01068115234375,
      "grad_norm": 13.740591049194336,
      "learning_rate": 4.998845576425529e-06,
      "loss": 3.7697,
      "step": 1050
    },
    {
      "epoch": 0.010732014973958334,
      "grad_norm": 15.526152610778809,
      "learning_rate": 4.998833393567605e-06,
      "loss": 3.3944,
      "step": 1055
    },
    {
      "epoch": 0.010782877604166666,
      "grad_norm": 14.3406982421875,
      "learning_rate": 4.998821146777622e-06,
      "loss": 3.8095,
      "step": 1060
    },
    {
      "epoch": 0.010833740234375,
      "grad_norm": 12.40285873413086,
      "learning_rate": 4.99880883605589e-06,
      "loss": 3.6193,
      "step": 1065
    },
    {
      "epoch": 0.010884602864583334,
      "grad_norm": 20.011245727539062,
      "learning_rate": 4.998796461402729e-06,
      "loss": 3.8485,
      "step": 1070
    },
    {
      "epoch": 0.010935465494791666,
      "grad_norm": 15.710371017456055,
      "learning_rate": 4.998784022818452e-06,
      "loss": 3.8256,
      "step": 1075
    },
    {
      "epoch": 0.010986328125,
      "grad_norm": 18.561986923217773,
      "learning_rate": 4.998771520303376e-06,
      "loss": 3.9108,
      "step": 1080
    },
    {
      "epoch": 0.011037190755208334,
      "grad_norm": 18.281129837036133,
      "learning_rate": 4.998758953857825e-06,
      "loss": 3.4311,
      "step": 1085
    },
    {
      "epoch": 0.011088053385416666,
      "grad_norm": 14.150856018066406,
      "learning_rate": 4.998746323482117e-06,
      "loss": 3.5551,
      "step": 1090
    },
    {
      "epoch": 0.011138916015625,
      "grad_norm": 13.725726127624512,
      "learning_rate": 4.9987336291765784e-06,
      "loss": 3.6229,
      "step": 1095
    },
    {
      "epoch": 0.011189778645833334,
      "grad_norm": 11.532977104187012,
      "learning_rate": 4.998720870941531e-06,
      "loss": 3.5328,
      "step": 1100
    },
    {
      "epoch": 0.011240641276041666,
      "grad_norm": 15.219062805175781,
      "learning_rate": 4.998708048777303e-06,
      "loss": 3.7603,
      "step": 1105
    },
    {
      "epoch": 0.01129150390625,
      "grad_norm": 10.167997360229492,
      "learning_rate": 4.9986951626842215e-06,
      "loss": 4.157,
      "step": 1110
    },
    {
      "epoch": 0.011342366536458334,
      "grad_norm": 25.62779998779297,
      "learning_rate": 4.9986822126626165e-06,
      "loss": 3.539,
      "step": 1115
    },
    {
      "epoch": 0.011393229166666666,
      "grad_norm": 13.080843925476074,
      "learning_rate": 4.998669198712819e-06,
      "loss": 3.9684,
      "step": 1120
    },
    {
      "epoch": 0.011444091796875,
      "grad_norm": 13.87924861907959,
      "learning_rate": 4.998656120835163e-06,
      "loss": 3.6138,
      "step": 1125
    },
    {
      "epoch": 0.011494954427083334,
      "grad_norm": 16.160778045654297,
      "learning_rate": 4.998642979029982e-06,
      "loss": 3.7762,
      "step": 1130
    },
    {
      "epoch": 0.011545817057291666,
      "grad_norm": 12.456393241882324,
      "learning_rate": 4.998629773297613e-06,
      "loss": 3.5095,
      "step": 1135
    },
    {
      "epoch": 0.0115966796875,
      "grad_norm": 18.339937210083008,
      "learning_rate": 4.998616503638393e-06,
      "loss": 3.6277,
      "step": 1140
    },
    {
      "epoch": 0.011647542317708334,
      "grad_norm": 13.26982307434082,
      "learning_rate": 4.998603170052662e-06,
      "loss": 3.9732,
      "step": 1145
    },
    {
      "epoch": 0.011698404947916666,
      "grad_norm": 12.922799110412598,
      "learning_rate": 4.9985897725407616e-06,
      "loss": 3.4633,
      "step": 1150
    },
    {
      "epoch": 0.011749267578125,
      "grad_norm": 19.501089096069336,
      "learning_rate": 4.998576311103033e-06,
      "loss": 3.7619,
      "step": 1155
    },
    {
      "epoch": 0.011800130208333334,
      "grad_norm": 12.718639373779297,
      "learning_rate": 4.998562785739823e-06,
      "loss": 3.7302,
      "step": 1160
    },
    {
      "epoch": 0.011850992838541666,
      "grad_norm": 10.442389488220215,
      "learning_rate": 4.998549196451475e-06,
      "loss": 3.5264,
      "step": 1165
    },
    {
      "epoch": 0.01190185546875,
      "grad_norm": 15.280603408813477,
      "learning_rate": 4.99853554323834e-06,
      "loss": 3.7362,
      "step": 1170
    },
    {
      "epoch": 0.011952718098958334,
      "grad_norm": 14.212106704711914,
      "learning_rate": 4.998521826100764e-06,
      "loss": 3.7874,
      "step": 1175
    },
    {
      "epoch": 0.012003580729166666,
      "grad_norm": 13.269865989685059,
      "learning_rate": 4.998508045039099e-06,
      "loss": 3.6369,
      "step": 1180
    },
    {
      "epoch": 0.012054443359375,
      "grad_norm": 12.6091890335083,
      "learning_rate": 4.998494200053698e-06,
      "loss": 3.6542,
      "step": 1185
    },
    {
      "epoch": 0.012105305989583334,
      "grad_norm": 14.155792236328125,
      "learning_rate": 4.998480291144916e-06,
      "loss": 4.0782,
      "step": 1190
    },
    {
      "epoch": 0.012156168619791666,
      "grad_norm": 19.589990615844727,
      "learning_rate": 4.998466318313108e-06,
      "loss": 3.7343,
      "step": 1195
    },
    {
      "epoch": 0.01220703125,
      "grad_norm": 13.791729927062988,
      "learning_rate": 4.99845228155863e-06,
      "loss": 3.6456,
      "step": 1200
    },
    {
      "epoch": 0.012257893880208334,
      "grad_norm": 13.561478614807129,
      "learning_rate": 4.998438180881844e-06,
      "loss": 4.0166,
      "step": 1205
    },
    {
      "epoch": 0.012308756510416666,
      "grad_norm": 18.43284034729004,
      "learning_rate": 4.998424016283109e-06,
      "loss": 3.6783,
      "step": 1210
    },
    {
      "epoch": 0.012359619140625,
      "grad_norm": 19.545591354370117,
      "learning_rate": 4.9984097877627865e-06,
      "loss": 3.7069,
      "step": 1215
    },
    {
      "epoch": 0.012410481770833334,
      "grad_norm": 16.589338302612305,
      "learning_rate": 4.998395495321243e-06,
      "loss": 3.5455,
      "step": 1220
    },
    {
      "epoch": 0.012461344401041666,
      "grad_norm": 13.449020385742188,
      "learning_rate": 4.998381138958843e-06,
      "loss": 3.6977,
      "step": 1225
    },
    {
      "epoch": 0.01251220703125,
      "grad_norm": 17.311525344848633,
      "learning_rate": 4.9983667186759535e-06,
      "loss": 3.6899,
      "step": 1230
    },
    {
      "epoch": 0.012563069661458334,
      "grad_norm": 15.647948265075684,
      "learning_rate": 4.998352234472944e-06,
      "loss": 3.1426,
      "step": 1235
    },
    {
      "epoch": 0.012613932291666666,
      "grad_norm": 22.255765914916992,
      "learning_rate": 4.998337686350184e-06,
      "loss": 3.66,
      "step": 1240
    },
    {
      "epoch": 0.012664794921875,
      "grad_norm": 15.966592788696289,
      "learning_rate": 4.998323074308047e-06,
      "loss": 3.6801,
      "step": 1245
    },
    {
      "epoch": 0.012715657552083334,
      "grad_norm": 10.439255714416504,
      "learning_rate": 4.998308398346906e-06,
      "loss": 3.6352,
      "step": 1250
    },
    {
      "epoch": 0.012766520182291666,
      "grad_norm": 14.009291648864746,
      "learning_rate": 4.998293658467137e-06,
      "loss": 3.5148,
      "step": 1255
    },
    {
      "epoch": 0.0128173828125,
      "grad_norm": 9.830448150634766,
      "learning_rate": 4.998278854669117e-06,
      "loss": 3.405,
      "step": 1260
    },
    {
      "epoch": 0.012868245442708334,
      "grad_norm": 18.137758255004883,
      "learning_rate": 4.998263986953224e-06,
      "loss": 3.2475,
      "step": 1265
    },
    {
      "epoch": 0.012919108072916666,
      "grad_norm": 13.296918869018555,
      "learning_rate": 4.99824905531984e-06,
      "loss": 3.5983,
      "step": 1270
    },
    {
      "epoch": 0.012969970703125,
      "grad_norm": 17.331226348876953,
      "learning_rate": 4.9982340597693455e-06,
      "loss": 3.4618,
      "step": 1275
    },
    {
      "epoch": 0.013020833333333334,
      "grad_norm": 16.383100509643555,
      "learning_rate": 4.998219000302125e-06,
      "loss": 3.5292,
      "step": 1280
    },
    {
      "epoch": 0.013071695963541666,
      "grad_norm": 10.87190055847168,
      "learning_rate": 4.998203876918564e-06,
      "loss": 3.6293,
      "step": 1285
    },
    {
      "epoch": 0.01312255859375,
      "grad_norm": 12.597033500671387,
      "learning_rate": 4.998188689619048e-06,
      "loss": 3.7316,
      "step": 1290
    },
    {
      "epoch": 0.013173421223958334,
      "grad_norm": 12.681578636169434,
      "learning_rate": 4.998173438403966e-06,
      "loss": 3.478,
      "step": 1295
    },
    {
      "epoch": 0.013224283854166666,
      "grad_norm": 16.530866622924805,
      "learning_rate": 4.99815812327371e-06,
      "loss": 3.7343,
      "step": 1300
    },
    {
      "epoch": 0.013275146484375,
      "grad_norm": 17.380083084106445,
      "learning_rate": 4.99814274422867e-06,
      "loss": 3.5332,
      "step": 1305
    },
    {
      "epoch": 0.013326009114583334,
      "grad_norm": 15.711562156677246,
      "learning_rate": 4.998127301269241e-06,
      "loss": 3.4281,
      "step": 1310
    },
    {
      "epoch": 0.013376871744791666,
      "grad_norm": 13.467732429504395,
      "learning_rate": 4.998111794395816e-06,
      "loss": 3.588,
      "step": 1315
    },
    {
      "epoch": 0.013427734375,
      "grad_norm": 16.989633560180664,
      "learning_rate": 4.998096223608792e-06,
      "loss": 3.304,
      "step": 1320
    },
    {
      "epoch": 0.013478597005208334,
      "grad_norm": 12.060418128967285,
      "learning_rate": 4.998080588908571e-06,
      "loss": 3.121,
      "step": 1325
    },
    {
      "epoch": 0.013529459635416666,
      "grad_norm": 15.241015434265137,
      "learning_rate": 4.9980648902955475e-06,
      "loss": 3.6774,
      "step": 1330
    },
    {
      "epoch": 0.013580322265625,
      "grad_norm": 15.760343551635742,
      "learning_rate": 4.998049127770127e-06,
      "loss": 3.2477,
      "step": 1335
    },
    {
      "epoch": 0.013631184895833334,
      "grad_norm": 15.907750129699707,
      "learning_rate": 4.998033301332712e-06,
      "loss": 3.3698,
      "step": 1340
    },
    {
      "epoch": 0.013682047526041666,
      "grad_norm": 17.1384334564209,
      "learning_rate": 4.9980174109837065e-06,
      "loss": 3.7272,
      "step": 1345
    },
    {
      "epoch": 0.01373291015625,
      "grad_norm": 16.641490936279297,
      "learning_rate": 4.998001456723518e-06,
      "loss": 3.6683,
      "step": 1350
    },
    {
      "epoch": 0.013783772786458334,
      "grad_norm": 8.19157600402832,
      "learning_rate": 4.997985438552554e-06,
      "loss": 3.6485,
      "step": 1355
    },
    {
      "epoch": 0.013834635416666666,
      "grad_norm": 13.124740600585938,
      "learning_rate": 4.997969356471225e-06,
      "loss": 3.6489,
      "step": 1360
    },
    {
      "epoch": 0.013885498046875,
      "grad_norm": 11.489811897277832,
      "learning_rate": 4.997953210479941e-06,
      "loss": 3.9778,
      "step": 1365
    },
    {
      "epoch": 0.013936360677083334,
      "grad_norm": 18.886642456054688,
      "learning_rate": 4.997937000579118e-06,
      "loss": 3.7343,
      "step": 1370
    },
    {
      "epoch": 0.013987223307291666,
      "grad_norm": 9.009153366088867,
      "learning_rate": 4.997920726769168e-06,
      "loss": 3.5347,
      "step": 1375
    },
    {
      "epoch": 0.0140380859375,
      "grad_norm": 14.774568557739258,
      "learning_rate": 4.997904389050508e-06,
      "loss": 3.7011,
      "step": 1380
    },
    {
      "epoch": 0.014088948567708334,
      "grad_norm": 26.565807342529297,
      "learning_rate": 4.997887987423556e-06,
      "loss": 4.026,
      "step": 1385
    },
    {
      "epoch": 0.014139811197916666,
      "grad_norm": 13.156881332397461,
      "learning_rate": 4.997871521888733e-06,
      "loss": 3.645,
      "step": 1390
    },
    {
      "epoch": 0.014190673828125,
      "grad_norm": 13.961007118225098,
      "learning_rate": 4.9978549924464595e-06,
      "loss": 3.638,
      "step": 1395
    },
    {
      "epoch": 0.014241536458333334,
      "grad_norm": 12.008218765258789,
      "learning_rate": 4.997838399097157e-06,
      "loss": 3.7365,
      "step": 1400
    },
    {
      "epoch": 0.014292399088541666,
      "grad_norm": 16.033838272094727,
      "learning_rate": 4.997821741841251e-06,
      "loss": 3.8982,
      "step": 1405
    },
    {
      "epoch": 0.01434326171875,
      "grad_norm": 10.620261192321777,
      "learning_rate": 4.997805020679169e-06,
      "loss": 3.5443,
      "step": 1410
    },
    {
      "epoch": 0.014394124348958334,
      "grad_norm": 13.818375587463379,
      "learning_rate": 4.997788235611336e-06,
      "loss": 3.8064,
      "step": 1415
    },
    {
      "epoch": 0.014444986979166666,
      "grad_norm": 14.587963104248047,
      "learning_rate": 4.997771386638184e-06,
      "loss": 3.599,
      "step": 1420
    },
    {
      "epoch": 0.014495849609375,
      "grad_norm": 11.063785552978516,
      "learning_rate": 4.997754473760143e-06,
      "loss": 3.6159,
      "step": 1425
    },
    {
      "epoch": 0.014546712239583334,
      "grad_norm": 11.75071907043457,
      "learning_rate": 4.997737496977645e-06,
      "loss": 4.4565,
      "step": 1430
    },
    {
      "epoch": 0.014597574869791666,
      "grad_norm": 14.461017608642578,
      "learning_rate": 4.997720456291126e-06,
      "loss": 3.7098,
      "step": 1435
    },
    {
      "epoch": 0.0146484375,
      "grad_norm": 11.124221801757812,
      "learning_rate": 4.997703351701021e-06,
      "loss": 3.5637,
      "step": 1440
    },
    {
      "epoch": 0.014699300130208334,
      "grad_norm": 13.678889274597168,
      "learning_rate": 4.997686183207767e-06,
      "loss": 3.5194,
      "step": 1445
    },
    {
      "epoch": 0.014750162760416666,
      "grad_norm": 11.239283561706543,
      "learning_rate": 4.9976689508118055e-06,
      "loss": 3.6568,
      "step": 1450
    },
    {
      "epoch": 0.014801025390625,
      "grad_norm": 12.112907409667969,
      "learning_rate": 4.997651654513575e-06,
      "loss": 3.6608,
      "step": 1455
    },
    {
      "epoch": 0.014851888020833334,
      "grad_norm": 13.270151138305664,
      "learning_rate": 4.997634294313519e-06,
      "loss": 3.5079,
      "step": 1460
    },
    {
      "epoch": 0.014902750651041666,
      "grad_norm": 13.076834678649902,
      "learning_rate": 4.997616870212082e-06,
      "loss": 3.3748,
      "step": 1465
    },
    {
      "epoch": 0.01495361328125,
      "grad_norm": 10.513318061828613,
      "learning_rate": 4.997599382209709e-06,
      "loss": 3.6342,
      "step": 1470
    },
    {
      "epoch": 0.015004475911458334,
      "grad_norm": 18.05800437927246,
      "learning_rate": 4.997581830306848e-06,
      "loss": 3.6844,
      "step": 1475
    },
    {
      "epoch": 0.015055338541666666,
      "grad_norm": 10.047406196594238,
      "learning_rate": 4.997564214503947e-06,
      "loss": 4.0638,
      "step": 1480
    },
    {
      "epoch": 0.015106201171875,
      "grad_norm": 15.92277717590332,
      "learning_rate": 4.997546534801459e-06,
      "loss": 3.7102,
      "step": 1485
    },
    {
      "epoch": 0.015157063802083334,
      "grad_norm": 15.660829544067383,
      "learning_rate": 4.997528791199834e-06,
      "loss": 4.0566,
      "step": 1490
    },
    {
      "epoch": 0.015207926432291666,
      "grad_norm": 21.788442611694336,
      "learning_rate": 4.997510983699527e-06,
      "loss": 4.1864,
      "step": 1495
    },
    {
      "epoch": 0.0152587890625,
      "grad_norm": 10.556007385253906,
      "learning_rate": 4.997493112300994e-06,
      "loss": 3.621,
      "step": 1500
    },
    {
      "epoch": 0.015309651692708334,
      "grad_norm": 13.028473854064941,
      "learning_rate": 4.99747517700469e-06,
      "loss": 4.406,
      "step": 1505
    },
    {
      "epoch": 0.015360514322916666,
      "grad_norm": 14.941851615905762,
      "learning_rate": 4.997457177811077e-06,
      "loss": 3.1199,
      "step": 1510
    },
    {
      "epoch": 0.015411376953125,
      "grad_norm": 18.102357864379883,
      "learning_rate": 4.997439114720614e-06,
      "loss": 3.7058,
      "step": 1515
    },
    {
      "epoch": 0.015462239583333334,
      "grad_norm": 22.9876651763916,
      "learning_rate": 4.997420987733763e-06,
      "loss": 3.5268,
      "step": 1520
    },
    {
      "epoch": 0.015513102213541666,
      "grad_norm": 11.794015884399414,
      "learning_rate": 4.997402796850989e-06,
      "loss": 3.6392,
      "step": 1525
    },
    {
      "epoch": 0.01556396484375,
      "grad_norm": 11.62551212310791,
      "learning_rate": 4.997384542072755e-06,
      "loss": 3.4675,
      "step": 1530
    },
    {
      "epoch": 0.015614827473958334,
      "grad_norm": 14.436004638671875,
      "learning_rate": 4.99736622339953e-06,
      "loss": 3.657,
      "step": 1535
    },
    {
      "epoch": 0.015665690104166668,
      "grad_norm": 13.236321449279785,
      "learning_rate": 4.997347840831782e-06,
      "loss": 3.2599,
      "step": 1540
    },
    {
      "epoch": 0.015716552734375,
      "grad_norm": 11.505661010742188,
      "learning_rate": 4.997329394369981e-06,
      "loss": 3.6238,
      "step": 1545
    },
    {
      "epoch": 0.015767415364583332,
      "grad_norm": 14.669344902038574,
      "learning_rate": 4.997310884014599e-06,
      "loss": 3.4919,
      "step": 1550
    },
    {
      "epoch": 0.015818277994791668,
      "grad_norm": 9.981671333312988,
      "learning_rate": 4.997292309766111e-06,
      "loss": 3.4323,
      "step": 1555
    },
    {
      "epoch": 0.015869140625,
      "grad_norm": 13.35000228881836,
      "learning_rate": 4.997273671624991e-06,
      "loss": 3.5778,
      "step": 1560
    },
    {
      "epoch": 0.015920003255208332,
      "grad_norm": 13.411084175109863,
      "learning_rate": 4.997254969591716e-06,
      "loss": 3.8477,
      "step": 1565
    },
    {
      "epoch": 0.015970865885416668,
      "grad_norm": 13.634716033935547,
      "learning_rate": 4.997236203666764e-06,
      "loss": 3.4397,
      "step": 1570
    },
    {
      "epoch": 0.016021728515625,
      "grad_norm": 9.219624519348145,
      "learning_rate": 4.997217373850617e-06,
      "loss": 3.5114,
      "step": 1575
    },
    {
      "epoch": 0.016072591145833332,
      "grad_norm": 12.118866920471191,
      "learning_rate": 4.997198480143755e-06,
      "loss": 3.7325,
      "step": 1580
    },
    {
      "epoch": 0.016123453776041668,
      "grad_norm": 12.5546875,
      "learning_rate": 4.99717952254666e-06,
      "loss": 3.6924,
      "step": 1585
    },
    {
      "epoch": 0.01617431640625,
      "grad_norm": 15.333346366882324,
      "learning_rate": 4.99716050105982e-06,
      "loss": 3.4589,
      "step": 1590
    },
    {
      "epoch": 0.016225179036458332,
      "grad_norm": 11.861115455627441,
      "learning_rate": 4.997141415683721e-06,
      "loss": 3.7009,
      "step": 1595
    },
    {
      "epoch": 0.016276041666666668,
      "grad_norm": 15.62850570678711,
      "learning_rate": 4.99712226641885e-06,
      "loss": 4.2749,
      "step": 1600
    },
    {
      "epoch": 0.016326904296875,
      "grad_norm": 16.968338012695312,
      "learning_rate": 4.997103053265698e-06,
      "loss": 3.458,
      "step": 1605
    },
    {
      "epoch": 0.016377766927083332,
      "grad_norm": 15.927022933959961,
      "learning_rate": 4.997083776224757e-06,
      "loss": 3.5739,
      "step": 1610
    },
    {
      "epoch": 0.016428629557291668,
      "grad_norm": 12.973907470703125,
      "learning_rate": 4.997064435296518e-06,
      "loss": 3.5018,
      "step": 1615
    },
    {
      "epoch": 0.0164794921875,
      "grad_norm": 16.26294708251953,
      "learning_rate": 4.997045030481478e-06,
      "loss": 3.8331,
      "step": 1620
    },
    {
      "epoch": 0.016530354817708332,
      "grad_norm": 14.90145492553711,
      "learning_rate": 4.997025561780133e-06,
      "loss": 3.7663,
      "step": 1625
    },
    {
      "epoch": 0.016581217447916668,
      "grad_norm": 10.96834945678711,
      "learning_rate": 4.9970060291929816e-06,
      "loss": 3.2283,
      "step": 1630
    },
    {
      "epoch": 0.016632080078125,
      "grad_norm": 14.805373191833496,
      "learning_rate": 4.996986432720521e-06,
      "loss": 3.5404,
      "step": 1635
    },
    {
      "epoch": 0.016682942708333332,
      "grad_norm": 14.054951667785645,
      "learning_rate": 4.996966772363255e-06,
      "loss": 3.7508,
      "step": 1640
    },
    {
      "epoch": 0.016733805338541668,
      "grad_norm": 18.448801040649414,
      "learning_rate": 4.996947048121686e-06,
      "loss": 3.1579,
      "step": 1645
    },
    {
      "epoch": 0.01678466796875,
      "grad_norm": 11.387920379638672,
      "learning_rate": 4.996927259996319e-06,
      "loss": 3.1547,
      "step": 1650
    },
    {
      "epoch": 0.016835530598958332,
      "grad_norm": 17.80491828918457,
      "learning_rate": 4.99690740798766e-06,
      "loss": 3.5202,
      "step": 1655
    },
    {
      "epoch": 0.016886393229166668,
      "grad_norm": 15.007222175598145,
      "learning_rate": 4.9968874920962165e-06,
      "loss": 3.6235,
      "step": 1660
    },
    {
      "epoch": 0.016937255859375,
      "grad_norm": 10.149649620056152,
      "learning_rate": 4.996867512322499e-06,
      "loss": 3.6799,
      "step": 1665
    },
    {
      "epoch": 0.016988118489583332,
      "grad_norm": 17.945354461669922,
      "learning_rate": 4.996847468667016e-06,
      "loss": 3.9798,
      "step": 1670
    },
    {
      "epoch": 0.017038981119791668,
      "grad_norm": 15.979263305664062,
      "learning_rate": 4.9968273611302845e-06,
      "loss": 3.3488,
      "step": 1675
    },
    {
      "epoch": 0.01708984375,
      "grad_norm": 14.131532669067383,
      "learning_rate": 4.996807189712815e-06,
      "loss": 3.3508,
      "step": 1680
    },
    {
      "epoch": 0.017140706380208332,
      "grad_norm": 19.903356552124023,
      "learning_rate": 4.996786954415127e-06,
      "loss": 3.5976,
      "step": 1685
    },
    {
      "epoch": 0.017191569010416668,
      "grad_norm": 13.6408052444458,
      "learning_rate": 4.996766655237736e-06,
      "loss": 3.4822,
      "step": 1690
    },
    {
      "epoch": 0.017242431640625,
      "grad_norm": 9.797231674194336,
      "learning_rate": 4.9967462921811614e-06,
      "loss": 3.9847,
      "step": 1695
    },
    {
      "epoch": 0.017293294270833332,
      "grad_norm": 13.015440940856934,
      "learning_rate": 4.996725865245926e-06,
      "loss": 3.5052,
      "step": 1700
    },
    {
      "epoch": 0.017344156901041668,
      "grad_norm": 17.70994758605957,
      "learning_rate": 4.99670537443255e-06,
      "loss": 3.9466,
      "step": 1705
    },
    {
      "epoch": 0.01739501953125,
      "grad_norm": 9.877345085144043,
      "learning_rate": 4.996684819741559e-06,
      "loss": 3.6346,
      "step": 1710
    },
    {
      "epoch": 0.017445882161458332,
      "grad_norm": 12.302515983581543,
      "learning_rate": 4.996664201173478e-06,
      "loss": 3.3808,
      "step": 1715
    },
    {
      "epoch": 0.017496744791666668,
      "grad_norm": 12.325230598449707,
      "learning_rate": 4.9966435187288365e-06,
      "loss": 3.8128,
      "step": 1720
    },
    {
      "epoch": 0.017547607421875,
      "grad_norm": 14.4461669921875,
      "learning_rate": 4.996622772408162e-06,
      "loss": 4.3367,
      "step": 1725
    },
    {
      "epoch": 0.017598470052083332,
      "grad_norm": 15.207497596740723,
      "learning_rate": 4.996601962211985e-06,
      "loss": 3.5895,
      "step": 1730
    },
    {
      "epoch": 0.017649332682291668,
      "grad_norm": 10.924785614013672,
      "learning_rate": 4.9965810881408384e-06,
      "loss": 3.3813,
      "step": 1735
    },
    {
      "epoch": 0.0177001953125,
      "grad_norm": 19.209896087646484,
      "learning_rate": 4.996560150195257e-06,
      "loss": 3.4489,
      "step": 1740
    },
    {
      "epoch": 0.017751057942708332,
      "grad_norm": 14.56915283203125,
      "learning_rate": 4.9965391483757765e-06,
      "loss": 3.4132,
      "step": 1745
    },
    {
      "epoch": 0.017801920572916668,
      "grad_norm": 9.958040237426758,
      "learning_rate": 4.996518082682933e-06,
      "loss": 3.5184,
      "step": 1750
    },
    {
      "epoch": 0.017852783203125,
      "grad_norm": 14.741584777832031,
      "learning_rate": 4.9964969531172656e-06,
      "loss": 3.7564,
      "step": 1755
    },
    {
      "epoch": 0.017903645833333332,
      "grad_norm": 12.061718940734863,
      "learning_rate": 4.996475759679316e-06,
      "loss": 3.5494,
      "step": 1760
    },
    {
      "epoch": 0.017954508463541668,
      "grad_norm": 19.455101013183594,
      "learning_rate": 4.9964545023696255e-06,
      "loss": 3.5547,
      "step": 1765
    },
    {
      "epoch": 0.01800537109375,
      "grad_norm": 15.922080993652344,
      "learning_rate": 4.996433181188739e-06,
      "loss": 3.6292,
      "step": 1770
    },
    {
      "epoch": 0.018056233723958332,
      "grad_norm": 10.75054931640625,
      "learning_rate": 4.996411796137201e-06,
      "loss": 3.3277,
      "step": 1775
    },
    {
      "epoch": 0.018107096354166668,
      "grad_norm": 14.0524263381958,
      "learning_rate": 4.996390347215558e-06,
      "loss": 3.671,
      "step": 1780
    },
    {
      "epoch": 0.018157958984375,
      "grad_norm": 17.705394744873047,
      "learning_rate": 4.9963688344243605e-06,
      "loss": 3.4854,
      "step": 1785
    },
    {
      "epoch": 0.018208821614583332,
      "grad_norm": 14.042203903198242,
      "learning_rate": 4.996347257764158e-06,
      "loss": 3.3586,
      "step": 1790
    },
    {
      "epoch": 0.018259684244791668,
      "grad_norm": 15.437379837036133,
      "learning_rate": 4.996325617235502e-06,
      "loss": 3.1949,
      "step": 1795
    },
    {
      "epoch": 0.018310546875,
      "grad_norm": 14.990214347839355,
      "learning_rate": 4.996303912838948e-06,
      "loss": 3.2657,
      "step": 1800
    },
    {
      "epoch": 0.018361409505208332,
      "grad_norm": 20.377416610717773,
      "learning_rate": 4.9962821445750485e-06,
      "loss": 3.5048,
      "step": 1805
    },
    {
      "epoch": 0.018412272135416668,
      "grad_norm": 11.091985702514648,
      "learning_rate": 4.996260312444363e-06,
      "loss": 3.5013,
      "step": 1810
    },
    {
      "epoch": 0.018463134765625,
      "grad_norm": 17.564790725708008,
      "learning_rate": 4.9962384164474495e-06,
      "loss": 3.9346,
      "step": 1815
    },
    {
      "epoch": 0.018513997395833332,
      "grad_norm": 11.695920944213867,
      "learning_rate": 4.996216456584867e-06,
      "loss": 3.7216,
      "step": 1820
    },
    {
      "epoch": 0.018564860026041668,
      "grad_norm": 12.64533519744873,
      "learning_rate": 4.9961944328571785e-06,
      "loss": 3.4994,
      "step": 1825
    },
    {
      "epoch": 0.01861572265625,
      "grad_norm": 11.813600540161133,
      "learning_rate": 4.9961723452649465e-06,
      "loss": 3.3742,
      "step": 1830
    },
    {
      "epoch": 0.018666585286458332,
      "grad_norm": 13.247116088867188,
      "learning_rate": 4.9961501938087375e-06,
      "loss": 3.7691,
      "step": 1835
    },
    {
      "epoch": 0.018717447916666668,
      "grad_norm": 10.45460319519043,
      "learning_rate": 4.996127978489117e-06,
      "loss": 3.6143,
      "step": 1840
    },
    {
      "epoch": 0.018768310546875,
      "grad_norm": 19.198963165283203,
      "learning_rate": 4.996105699306654e-06,
      "loss": 3.4559,
      "step": 1845
    },
    {
      "epoch": 0.018819173177083332,
      "grad_norm": 15.408687591552734,
      "learning_rate": 4.996083356261918e-06,
      "loss": 3.6425,
      "step": 1850
    },
    {
      "epoch": 0.018870035807291668,
      "grad_norm": 16.426103591918945,
      "learning_rate": 4.996060949355481e-06,
      "loss": 3.6617,
      "step": 1855
    },
    {
      "epoch": 0.0189208984375,
      "grad_norm": 9.62375259399414,
      "learning_rate": 4.996038478587916e-06,
      "loss": 4.2817,
      "step": 1860
    },
    {
      "epoch": 0.018971761067708332,
      "grad_norm": 13.927205085754395,
      "learning_rate": 4.9960159439598e-06,
      "loss": 3.1512,
      "step": 1865
    },
    {
      "epoch": 0.019022623697916668,
      "grad_norm": 16.292879104614258,
      "learning_rate": 4.995993345471706e-06,
      "loss": 3.7999,
      "step": 1870
    },
    {
      "epoch": 0.019073486328125,
      "grad_norm": 16.1326904296875,
      "learning_rate": 4.995970683124214e-06,
      "loss": 3.0987,
      "step": 1875
    },
    {
      "epoch": 0.019124348958333332,
      "grad_norm": 17.09796142578125,
      "learning_rate": 4.995947956917904e-06,
      "loss": 3.6616,
      "step": 1880
    },
    {
      "epoch": 0.019175211588541668,
      "grad_norm": 8.162797927856445,
      "learning_rate": 4.995925166853357e-06,
      "loss": 3.5208,
      "step": 1885
    },
    {
      "epoch": 0.01922607421875,
      "grad_norm": 35.92325210571289,
      "learning_rate": 4.995902312931156e-06,
      "loss": 3.819,
      "step": 1890
    },
    {
      "epoch": 0.019276936848958332,
      "grad_norm": 15.114096641540527,
      "learning_rate": 4.995879395151886e-06,
      "loss": 3.5895,
      "step": 1895
    },
    {
      "epoch": 0.019327799479166668,
      "grad_norm": 18.872722625732422,
      "learning_rate": 4.995856413516134e-06,
      "loss": 3.731,
      "step": 1900
    },
    {
      "epoch": 0.019378662109375,
      "grad_norm": 15.703900337219238,
      "learning_rate": 4.9958333680244865e-06,
      "loss": 3.8427,
      "step": 1905
    },
    {
      "epoch": 0.019429524739583332,
      "grad_norm": 14.129450798034668,
      "learning_rate": 4.9958102586775334e-06,
      "loss": 3.6527,
      "step": 1910
    },
    {
      "epoch": 0.019480387369791668,
      "grad_norm": 11.964359283447266,
      "learning_rate": 4.9957870854758675e-06,
      "loss": 3.3432,
      "step": 1915
    },
    {
      "epoch": 0.01953125,
      "grad_norm": 19.071924209594727,
      "learning_rate": 4.99576384842008e-06,
      "loss": 3.7882,
      "step": 1920
    },
    {
      "epoch": 0.019582112630208332,
      "grad_norm": 14.704536437988281,
      "learning_rate": 4.995740547510766e-06,
      "loss": 3.2511,
      "step": 1925
    },
    {
      "epoch": 0.019632975260416668,
      "grad_norm": 20.858272552490234,
      "learning_rate": 4.9957171827485215e-06,
      "loss": 4.266,
      "step": 1930
    },
    {
      "epoch": 0.019683837890625,
      "grad_norm": 12.940155982971191,
      "learning_rate": 4.995693754133944e-06,
      "loss": 3.7349,
      "step": 1935
    },
    {
      "epoch": 0.019734700520833332,
      "grad_norm": 17.16631507873535,
      "learning_rate": 4.995670261667635e-06,
      "loss": 3.6617,
      "step": 1940
    },
    {
      "epoch": 0.019785563151041668,
      "grad_norm": 12.754393577575684,
      "learning_rate": 4.995646705350193e-06,
      "loss": 3.6483,
      "step": 1945
    },
    {
      "epoch": 0.01983642578125,
      "grad_norm": 12.014126777648926,
      "learning_rate": 4.995623085182221e-06,
      "loss": 3.4977,
      "step": 1950
    },
    {
      "epoch": 0.019887288411458332,
      "grad_norm": 16.564720153808594,
      "learning_rate": 4.995599401164325e-06,
      "loss": 3.7256,
      "step": 1955
    },
    {
      "epoch": 0.019938151041666668,
      "grad_norm": 11.473775863647461,
      "learning_rate": 4.995575653297109e-06,
      "loss": 3.416,
      "step": 1960
    },
    {
      "epoch": 0.019989013671875,
      "grad_norm": 14.353362083435059,
      "learning_rate": 4.995551841581181e-06,
      "loss": 3.4641,
      "step": 1965
    },
    {
      "epoch": 0.020039876302083332,
      "grad_norm": 16.31317901611328,
      "learning_rate": 4.9955279660171514e-06,
      "loss": 3.7767,
      "step": 1970
    },
    {
      "epoch": 0.020090738932291668,
      "grad_norm": 8.309894561767578,
      "learning_rate": 4.99550402660563e-06,
      "loss": 3.6317,
      "step": 1975
    },
    {
      "epoch": 0.0201416015625,
      "grad_norm": 18.78270149230957,
      "learning_rate": 4.99548002334723e-06,
      "loss": 3.2542,
      "step": 1980
    },
    {
      "epoch": 0.020192464192708332,
      "grad_norm": 10.488119125366211,
      "learning_rate": 4.9954559562425654e-06,
      "loss": 3.3925,
      "step": 1985
    },
    {
      "epoch": 0.020243326822916668,
      "grad_norm": 19.56174087524414,
      "learning_rate": 4.99543182529225e-06,
      "loss": 3.4966,
      "step": 1990
    },
    {
      "epoch": 0.020294189453125,
      "grad_norm": 11.951705932617188,
      "learning_rate": 4.995407630496905e-06,
      "loss": 3.6632,
      "step": 1995
    },
    {
      "epoch": 0.020345052083333332,
      "grad_norm": 15.856471061706543,
      "learning_rate": 4.995383371857145e-06,
      "loss": 3.3239,
      "step": 2000
    },
    {
      "epoch": 0.020395914713541668,
      "grad_norm": 12.084221839904785,
      "learning_rate": 4.9953590493735945e-06,
      "loss": 3.5615,
      "step": 2005
    },
    {
      "epoch": 0.02044677734375,
      "grad_norm": 27.98715591430664,
      "learning_rate": 4.995334663046874e-06,
      "loss": 3.6218,
      "step": 2010
    },
    {
      "epoch": 0.020497639973958332,
      "grad_norm": 9.576325416564941,
      "learning_rate": 4.995310212877608e-06,
      "loss": 3.4569,
      "step": 2015
    },
    {
      "epoch": 0.020548502604166668,
      "grad_norm": 13.834068298339844,
      "learning_rate": 4.9952856988664205e-06,
      "loss": 3.2584,
      "step": 2020
    },
    {
      "epoch": 0.020599365234375,
      "grad_norm": 17.03768539428711,
      "learning_rate": 4.99526112101394e-06,
      "loss": 3.7883,
      "step": 2025
    },
    {
      "epoch": 0.020650227864583332,
      "grad_norm": 11.945119857788086,
      "learning_rate": 4.995236479320796e-06,
      "loss": 3.3385,
      "step": 2030
    },
    {
      "epoch": 0.020701090494791668,
      "grad_norm": 13.012072563171387,
      "learning_rate": 4.995211773787617e-06,
      "loss": 3.5075,
      "step": 2035
    },
    {
      "epoch": 0.020751953125,
      "grad_norm": 11.229742050170898,
      "learning_rate": 4.995187004415038e-06,
      "loss": 3.1753,
      "step": 2040
    },
    {
      "epoch": 0.020802815755208332,
      "grad_norm": 15.254374504089355,
      "learning_rate": 4.995162171203689e-06,
      "loss": 3.4893,
      "step": 2045
    },
    {
      "epoch": 0.020853678385416668,
      "grad_norm": 12.569649696350098,
      "learning_rate": 4.9951372741542084e-06,
      "loss": 3.7053,
      "step": 2050
    },
    {
      "epoch": 0.020904541015625,
      "grad_norm": 13.927783966064453,
      "learning_rate": 4.995112313267231e-06,
      "loss": 3.43,
      "step": 2055
    },
    {
      "epoch": 0.020955403645833332,
      "grad_norm": 10.135937690734863,
      "learning_rate": 4.995087288543397e-06,
      "loss": 3.3201,
      "step": 2060
    },
    {
      "epoch": 0.021006266276041668,
      "grad_norm": 17.4373722076416,
      "learning_rate": 4.995062199983346e-06,
      "loss": 3.9556,
      "step": 2065
    },
    {
      "epoch": 0.02105712890625,
      "grad_norm": 8.249671936035156,
      "learning_rate": 4.9950370475877204e-06,
      "loss": 3.4842,
      "step": 2070
    },
    {
      "epoch": 0.021107991536458332,
      "grad_norm": 7.946985244750977,
      "learning_rate": 4.995011831357164e-06,
      "loss": 3.7847,
      "step": 2075
    },
    {
      "epoch": 0.021158854166666668,
      "grad_norm": 14.219249725341797,
      "learning_rate": 4.99498655129232e-06,
      "loss": 3.6362,
      "step": 2080
    },
    {
      "epoch": 0.021209716796875,
      "grad_norm": 16.812952041625977,
      "learning_rate": 4.994961207393837e-06,
      "loss": 3.7761,
      "step": 2085
    },
    {
      "epoch": 0.021260579427083332,
      "grad_norm": 12.385375022888184,
      "learning_rate": 4.994935799662363e-06,
      "loss": 3.5996,
      "step": 2090
    },
    {
      "epoch": 0.021311442057291668,
      "grad_norm": 8.916481971740723,
      "learning_rate": 4.994910328098548e-06,
      "loss": 3.6023,
      "step": 2095
    },
    {
      "epoch": 0.0213623046875,
      "grad_norm": 14.721908569335938,
      "learning_rate": 4.994884792703043e-06,
      "loss": 3.3011,
      "step": 2100
    },
    {
      "epoch": 0.021413167317708332,
      "grad_norm": 15.370878219604492,
      "learning_rate": 4.9948591934765025e-06,
      "loss": 3.5518,
      "step": 2105
    },
    {
      "epoch": 0.021464029947916668,
      "grad_norm": 14.788433074951172,
      "learning_rate": 4.994833530419581e-06,
      "loss": 3.5008,
      "step": 2110
    },
    {
      "epoch": 0.021514892578125,
      "grad_norm": 12.971417427062988,
      "learning_rate": 4.994807803532934e-06,
      "loss": 3.5467,
      "step": 2115
    },
    {
      "epoch": 0.021565755208333332,
      "grad_norm": 14.517382621765137,
      "learning_rate": 4.994782012817221e-06,
      "loss": 3.6258,
      "step": 2120
    },
    {
      "epoch": 0.021616617838541668,
      "grad_norm": 10.307328224182129,
      "learning_rate": 4.994756158273102e-06,
      "loss": 4.0953,
      "step": 2125
    },
    {
      "epoch": 0.02166748046875,
      "grad_norm": 15.38490104675293,
      "learning_rate": 4.994730239901238e-06,
      "loss": 4.2029,
      "step": 2130
    },
    {
      "epoch": 0.021718343098958332,
      "grad_norm": 12.546719551086426,
      "learning_rate": 4.994704257702292e-06,
      "loss": 3.4173,
      "step": 2135
    },
    {
      "epoch": 0.021769205729166668,
      "grad_norm": 11.977869033813477,
      "learning_rate": 4.994678211676929e-06,
      "loss": 3.4095,
      "step": 2140
    },
    {
      "epoch": 0.021820068359375,
      "grad_norm": 10.538824081420898,
      "learning_rate": 4.994652101825815e-06,
      "loss": 3.607,
      "step": 2145
    },
    {
      "epoch": 0.021870930989583332,
      "grad_norm": 18.2629337310791,
      "learning_rate": 4.994625928149619e-06,
      "loss": 3.9144,
      "step": 2150
    },
    {
      "epoch": 0.021921793619791668,
      "grad_norm": 12.87021255493164,
      "learning_rate": 4.994599690649009e-06,
      "loss": 3.9643,
      "step": 2155
    },
    {
      "epoch": 0.02197265625,
      "grad_norm": 16.46050453186035,
      "learning_rate": 4.994573389324657e-06,
      "loss": 3.7189,
      "step": 2160
    },
    {
      "epoch": 0.022023518880208332,
      "grad_norm": 14.574606895446777,
      "learning_rate": 4.994547024177236e-06,
      "loss": 3.906,
      "step": 2165
    },
    {
      "epoch": 0.022074381510416668,
      "grad_norm": 10.178772926330566,
      "learning_rate": 4.994520595207422e-06,
      "loss": 3.7685,
      "step": 2170
    },
    {
      "epoch": 0.022125244140625,
      "grad_norm": 16.5206298828125,
      "learning_rate": 4.994494102415889e-06,
      "loss": 3.6191,
      "step": 2175
    },
    {
      "epoch": 0.022176106770833332,
      "grad_norm": 13.858648300170898,
      "learning_rate": 4.9944675458033156e-06,
      "loss": 3.5425,
      "step": 2180
    },
    {
      "epoch": 0.022226969401041668,
      "grad_norm": 15.840536117553711,
      "learning_rate": 4.994440925370382e-06,
      "loss": 3.5171,
      "step": 2185
    },
    {
      "epoch": 0.02227783203125,
      "grad_norm": 13.754470825195312,
      "learning_rate": 4.9944142411177675e-06,
      "loss": 3.3044,
      "step": 2190
    },
    {
      "epoch": 0.022328694661458332,
      "grad_norm": 13.027044296264648,
      "learning_rate": 4.994387493046157e-06,
      "loss": 3.3715,
      "step": 2195
    },
    {
      "epoch": 0.022379557291666668,
      "grad_norm": 10.115904808044434,
      "learning_rate": 4.994360681156233e-06,
      "loss": 3.8424,
      "step": 2200
    },
    {
      "epoch": 0.022430419921875,
      "grad_norm": 16.806591033935547,
      "learning_rate": 4.994333805448682e-06,
      "loss": 4.0637,
      "step": 2205
    },
    {
      "epoch": 0.022481282552083332,
      "grad_norm": 9.8330078125,
      "learning_rate": 4.994306865924192e-06,
      "loss": 3.383,
      "step": 2210
    },
    {
      "epoch": 0.022532145182291668,
      "grad_norm": 13.583646774291992,
      "learning_rate": 4.994279862583453e-06,
      "loss": 3.3756,
      "step": 2215
    },
    {
      "epoch": 0.0225830078125,
      "grad_norm": 12.496264457702637,
      "learning_rate": 4.994252795427153e-06,
      "loss": 3.5715,
      "step": 2220
    },
    {
      "epoch": 0.022633870442708332,
      "grad_norm": 9.020492553710938,
      "learning_rate": 4.994225664455989e-06,
      "loss": 3.3013,
      "step": 2225
    },
    {
      "epoch": 0.022684733072916668,
      "grad_norm": 14.377449989318848,
      "learning_rate": 4.99419846967065e-06,
      "loss": 3.3859,
      "step": 2230
    },
    {
      "epoch": 0.022735595703125,
      "grad_norm": 13.713812828063965,
      "learning_rate": 4.994171211071836e-06,
      "loss": 3.3353,
      "step": 2235
    },
    {
      "epoch": 0.022786458333333332,
      "grad_norm": 18.0495548248291,
      "learning_rate": 4.994143888660242e-06,
      "loss": 3.5454,
      "step": 2240
    },
    {
      "epoch": 0.022837320963541668,
      "grad_norm": 13.879693031311035,
      "learning_rate": 4.994116502436568e-06,
      "loss": 3.9231,
      "step": 2245
    },
    {
      "epoch": 0.02288818359375,
      "grad_norm": 14.874258041381836,
      "learning_rate": 4.994089052401515e-06,
      "loss": 3.114,
      "step": 2250
    },
    {
      "epoch": 0.022939046223958332,
      "grad_norm": 17.684362411499023,
      "learning_rate": 4.994061538555784e-06,
      "loss": 3.5418,
      "step": 2255
    },
    {
      "epoch": 0.022989908854166668,
      "grad_norm": 20.040653228759766,
      "learning_rate": 4.9940339609000796e-06,
      "loss": 4.0738,
      "step": 2260
    },
    {
      "epoch": 0.023040771484375,
      "grad_norm": 8.416106224060059,
      "learning_rate": 4.994006319435108e-06,
      "loss": 3.7515,
      "step": 2265
    },
    {
      "epoch": 0.023091634114583332,
      "grad_norm": 14.56326675415039,
      "learning_rate": 4.9939786141615754e-06,
      "loss": 3.4834,
      "step": 2270
    },
    {
      "epoch": 0.023142496744791668,
      "grad_norm": 9.106837272644043,
      "learning_rate": 4.993950845080191e-06,
      "loss": 3.363,
      "step": 2275
    },
    {
      "epoch": 0.023193359375,
      "grad_norm": 16.284809112548828,
      "learning_rate": 4.993923012191666e-06,
      "loss": 3.4758,
      "step": 2280
    },
    {
      "epoch": 0.023244222005208332,
      "grad_norm": 12.630521774291992,
      "learning_rate": 4.993895115496712e-06,
      "loss": 3.5318,
      "step": 2285
    },
    {
      "epoch": 0.023295084635416668,
      "grad_norm": 11.593666076660156,
      "learning_rate": 4.993867154996042e-06,
      "loss": 3.7887,
      "step": 2290
    },
    {
      "epoch": 0.023345947265625,
      "grad_norm": 14.084927558898926,
      "learning_rate": 4.993839130690372e-06,
      "loss": 3.5525,
      "step": 2295
    },
    {
      "epoch": 0.023396809895833332,
      "grad_norm": 15.547626495361328,
      "learning_rate": 4.993811042580419e-06,
      "loss": 3.3359,
      "step": 2300
    },
    {
      "epoch": 0.023447672526041668,
      "grad_norm": 17.31133460998535,
      "learning_rate": 4.993782890666902e-06,
      "loss": 3.6608,
      "step": 2305
    },
    {
      "epoch": 0.02349853515625,
      "grad_norm": 10.99673080444336,
      "learning_rate": 4.99375467495054e-06,
      "loss": 3.4517,
      "step": 2310
    },
    {
      "epoch": 0.023549397786458332,
      "grad_norm": 10.245965957641602,
      "learning_rate": 4.993726395432056e-06,
      "loss": 3.5961,
      "step": 2315
    },
    {
      "epoch": 0.023600260416666668,
      "grad_norm": 11.301896095275879,
      "learning_rate": 4.993698052112174e-06,
      "loss": 3.2348,
      "step": 2320
    },
    {
      "epoch": 0.023651123046875,
      "grad_norm": 16.186355590820312,
      "learning_rate": 4.993669644991617e-06,
      "loss": 3.6148,
      "step": 2325
    },
    {
      "epoch": 0.023701985677083332,
      "grad_norm": 15.257999420166016,
      "learning_rate": 4.993641174071115e-06,
      "loss": 3.7732,
      "step": 2330
    },
    {
      "epoch": 0.023752848307291668,
      "grad_norm": 17.36763572692871,
      "learning_rate": 4.993612639351393e-06,
      "loss": 3.5761,
      "step": 2335
    },
    {
      "epoch": 0.0238037109375,
      "grad_norm": 12.962672233581543,
      "learning_rate": 4.993584040833183e-06,
      "loss": 3.6535,
      "step": 2340
    },
    {
      "epoch": 0.023854573567708332,
      "grad_norm": 13.28327751159668,
      "learning_rate": 4.993555378517217e-06,
      "loss": 3.5254,
      "step": 2345
    },
    {
      "epoch": 0.023905436197916668,
      "grad_norm": 14.348062515258789,
      "learning_rate": 4.993526652404227e-06,
      "loss": 3.6133,
      "step": 2350
    },
    {
      "epoch": 0.023956298828125,
      "grad_norm": 15.535712242126465,
      "learning_rate": 4.993497862494949e-06,
      "loss": 3.5402,
      "step": 2355
    },
    {
      "epoch": 0.024007161458333332,
      "grad_norm": 10.898641586303711,
      "learning_rate": 4.993469008790119e-06,
      "loss": 3.5233,
      "step": 2360
    },
    {
      "epoch": 0.024058024088541668,
      "grad_norm": 9.476696968078613,
      "learning_rate": 4.993440091290476e-06,
      "loss": 4.3757,
      "step": 2365
    },
    {
      "epoch": 0.02410888671875,
      "grad_norm": 16.282838821411133,
      "learning_rate": 4.993411109996759e-06,
      "loss": 3.8592,
      "step": 2370
    },
    {
      "epoch": 0.024159749348958332,
      "grad_norm": 15.315332412719727,
      "learning_rate": 4.99338206490971e-06,
      "loss": 3.626,
      "step": 2375
    },
    {
      "epoch": 0.024210611979166668,
      "grad_norm": 11.935681343078613,
      "learning_rate": 4.993352956030071e-06,
      "loss": 3.5642,
      "step": 2380
    },
    {
      "epoch": 0.024261474609375,
      "grad_norm": 14.131113052368164,
      "learning_rate": 4.993323783358588e-06,
      "loss": 3.6816,
      "step": 2385
    },
    {
      "epoch": 0.024312337239583332,
      "grad_norm": 17.588123321533203,
      "learning_rate": 4.993294546896007e-06,
      "loss": 3.574,
      "step": 2390
    },
    {
      "epoch": 0.024363199869791668,
      "grad_norm": 11.560687065124512,
      "learning_rate": 4.993265246643076e-06,
      "loss": 3.6406,
      "step": 2395
    },
    {
      "epoch": 0.0244140625,
      "grad_norm": 9.761478424072266,
      "learning_rate": 4.993235882600545e-06,
      "loss": 3.5061,
      "step": 2400
    },
    {
      "epoch": 0.024464925130208332,
      "grad_norm": 13.649985313415527,
      "learning_rate": 4.993206454769165e-06,
      "loss": 3.4024,
      "step": 2405
    },
    {
      "epoch": 0.024515787760416668,
      "grad_norm": 8.24244499206543,
      "learning_rate": 4.993176963149689e-06,
      "loss": 3.6048,
      "step": 2410
    },
    {
      "epoch": 0.024566650390625,
      "grad_norm": 16.1616268157959,
      "learning_rate": 4.99314740774287e-06,
      "loss": 3.5386,
      "step": 2415
    },
    {
      "epoch": 0.024617513020833332,
      "grad_norm": 9.706306457519531,
      "learning_rate": 4.993117788549466e-06,
      "loss": 3.5126,
      "step": 2420
    },
    {
      "epoch": 0.024668375651041668,
      "grad_norm": 12.5220365524292,
      "learning_rate": 4.993088105570235e-06,
      "loss": 4.1297,
      "step": 2425
    },
    {
      "epoch": 0.02471923828125,
      "grad_norm": 12.254471778869629,
      "learning_rate": 4.993058358805935e-06,
      "loss": 3.4553,
      "step": 2430
    },
    {
      "epoch": 0.024770100911458332,
      "grad_norm": 14.104937553405762,
      "learning_rate": 4.993028548257328e-06,
      "loss": 4.2029,
      "step": 2435
    },
    {
      "epoch": 0.024820963541666668,
      "grad_norm": 14.170137405395508,
      "learning_rate": 4.992998673925177e-06,
      "loss": 3.45,
      "step": 2440
    },
    {
      "epoch": 0.024871826171875,
      "grad_norm": 16.678157806396484,
      "learning_rate": 4.9929687358102455e-06,
      "loss": 3.464,
      "step": 2445
    },
    {
      "epoch": 0.024922688802083332,
      "grad_norm": 11.808859825134277,
      "learning_rate": 4.9929387339133e-06,
      "loss": 4.1151,
      "step": 2450
    },
    {
      "epoch": 0.024973551432291668,
      "grad_norm": 13.544400215148926,
      "learning_rate": 4.992908668235107e-06,
      "loss": 3.4802,
      "step": 2455
    },
    {
      "epoch": 0.0250244140625,
      "grad_norm": 10.196426391601562,
      "learning_rate": 4.992878538776438e-06,
      "loss": 3.6122,
      "step": 2460
    },
    {
      "epoch": 0.025075276692708332,
      "grad_norm": 24.040620803833008,
      "learning_rate": 4.992848345538062e-06,
      "loss": 3.7206,
      "step": 2465
    },
    {
      "epoch": 0.025126139322916668,
      "grad_norm": 13.72429370880127,
      "learning_rate": 4.992818088520751e-06,
      "loss": 3.7046,
      "step": 2470
    },
    {
      "epoch": 0.025177001953125,
      "grad_norm": 17.09827995300293,
      "learning_rate": 4.992787767725281e-06,
      "loss": 3.6007,
      "step": 2475
    },
    {
      "epoch": 0.025227864583333332,
      "grad_norm": 17.559282302856445,
      "learning_rate": 4.992757383152427e-06,
      "loss": 3.5262,
      "step": 2480
    },
    {
      "epoch": 0.025278727213541668,
      "grad_norm": 12.092161178588867,
      "learning_rate": 4.992726934802965e-06,
      "loss": 3.4942,
      "step": 2485
    },
    {
      "epoch": 0.02532958984375,
      "grad_norm": 12.694467544555664,
      "learning_rate": 4.992696422677677e-06,
      "loss": 3.6092,
      "step": 2490
    },
    {
      "epoch": 0.025380452473958332,
      "grad_norm": 19.91484832763672,
      "learning_rate": 4.99266584677734e-06,
      "loss": 3.6935,
      "step": 2495
    },
    {
      "epoch": 0.025431315104166668,
      "grad_norm": 12.521865844726562,
      "learning_rate": 4.99263520710274e-06,
      "loss": 3.509,
      "step": 2500
    },
    {
      "epoch": 0.025482177734375,
      "grad_norm": 14.746912002563477,
      "learning_rate": 4.9926045036546576e-06,
      "loss": 3.4715,
      "step": 2505
    },
    {
      "epoch": 0.025533040364583332,
      "grad_norm": 13.788453102111816,
      "learning_rate": 4.99257373643388e-06,
      "loss": 3.7318,
      "step": 2510
    },
    {
      "epoch": 0.025583902994791668,
      "grad_norm": 15.496885299682617,
      "learning_rate": 4.992542905441194e-06,
      "loss": 3.6001,
      "step": 2515
    },
    {
      "epoch": 0.025634765625,
      "grad_norm": 11.678740501403809,
      "learning_rate": 4.992512010677389e-06,
      "loss": 3.3945,
      "step": 2520
    },
    {
      "epoch": 0.025685628255208332,
      "grad_norm": 11.555937767028809,
      "learning_rate": 4.992481052143256e-06,
      "loss": 3.5856,
      "step": 2525
    },
    {
      "epoch": 0.025736490885416668,
      "grad_norm": 11.76779556274414,
      "learning_rate": 4.992450029839584e-06,
      "loss": 3.6207,
      "step": 2530
    },
    {
      "epoch": 0.025787353515625,
      "grad_norm": 11.902135848999023,
      "learning_rate": 4.99241894376717e-06,
      "loss": 3.6504,
      "step": 2535
    },
    {
      "epoch": 0.025838216145833332,
      "grad_norm": 13.56098461151123,
      "learning_rate": 4.992387793926808e-06,
      "loss": 3.561,
      "step": 2540
    },
    {
      "epoch": 0.025889078776041668,
      "grad_norm": 9.4614839553833,
      "learning_rate": 4.9923565803192945e-06,
      "loss": 3.7419,
      "step": 2545
    },
    {
      "epoch": 0.02593994140625,
      "grad_norm": 15.035961151123047,
      "learning_rate": 4.9923253029454295e-06,
      "loss": 3.4019,
      "step": 2550
    },
    {
      "epoch": 0.025990804036458332,
      "grad_norm": 14.106694221496582,
      "learning_rate": 4.992293961806012e-06,
      "loss": 3.6503,
      "step": 2555
    },
    {
      "epoch": 0.026041666666666668,
      "grad_norm": 13.903535842895508,
      "learning_rate": 4.992262556901844e-06,
      "loss": 3.5632,
      "step": 2560
    },
    {
      "epoch": 0.026092529296875,
      "grad_norm": 21.27461814880371,
      "learning_rate": 4.99223108823373e-06,
      "loss": 3.3556,
      "step": 2565
    },
    {
      "epoch": 0.026143391927083332,
      "grad_norm": 19.755544662475586,
      "learning_rate": 4.992199555802473e-06,
      "loss": 3.9516,
      "step": 2570
    },
    {
      "epoch": 0.026194254557291668,
      "grad_norm": 10.459436416625977,
      "learning_rate": 4.992167959608882e-06,
      "loss": 4.184,
      "step": 2575
    },
    {
      "epoch": 0.0262451171875,
      "grad_norm": 15.970342636108398,
      "learning_rate": 4.992136299653763e-06,
      "loss": 3.4005,
      "step": 2580
    },
    {
      "epoch": 0.026295979817708332,
      "grad_norm": 12.037581443786621,
      "learning_rate": 4.992104575937929e-06,
      "loss": 4.3835,
      "step": 2585
    },
    {
      "epoch": 0.026346842447916668,
      "grad_norm": 9.693264961242676,
      "learning_rate": 4.99207278846219e-06,
      "loss": 3.7904,
      "step": 2590
    },
    {
      "epoch": 0.026397705078125,
      "grad_norm": 18.028060913085938,
      "learning_rate": 4.99204093722736e-06,
      "loss": 3.4888,
      "step": 2595
    },
    {
      "epoch": 0.026448567708333332,
      "grad_norm": 14.927176475524902,
      "learning_rate": 4.992009022234252e-06,
      "loss": 3.7141,
      "step": 2600
    },
    {
      "epoch": 0.026499430338541668,
      "grad_norm": 10.663310050964355,
      "learning_rate": 4.991977043483684e-06,
      "loss": 3.3668,
      "step": 2605
    },
    {
      "epoch": 0.02655029296875,
      "grad_norm": 21.105947494506836,
      "learning_rate": 4.991945000976475e-06,
      "loss": 3.7438,
      "step": 2610
    },
    {
      "epoch": 0.026601155598958332,
      "grad_norm": 11.864397048950195,
      "learning_rate": 4.991912894713443e-06,
      "loss": 3.4406,
      "step": 2615
    },
    {
      "epoch": 0.026652018229166668,
      "grad_norm": 16.927186965942383,
      "learning_rate": 4.99188072469541e-06,
      "loss": 3.8746,
      "step": 2620
    },
    {
      "epoch": 0.026702880859375,
      "grad_norm": 10.140044212341309,
      "learning_rate": 4.9918484909232e-06,
      "loss": 3.7149,
      "step": 2625
    },
    {
      "epoch": 0.026753743489583332,
      "grad_norm": 10.26709270477295,
      "learning_rate": 4.991816193397637e-06,
      "loss": 3.4153,
      "step": 2630
    },
    {
      "epoch": 0.026804606119791668,
      "grad_norm": 18.33658790588379,
      "learning_rate": 4.991783832119547e-06,
      "loss": 3.6604,
      "step": 2635
    },
    {
      "epoch": 0.02685546875,
      "grad_norm": 15.353517532348633,
      "learning_rate": 4.991751407089759e-06,
      "loss": 3.4237,
      "step": 2640
    },
    {
      "epoch": 0.026906331380208332,
      "grad_norm": 13.645818710327148,
      "learning_rate": 4.991718918309101e-06,
      "loss": 3.3133,
      "step": 2645
    },
    {
      "epoch": 0.026957194010416668,
      "grad_norm": 8.63716983795166,
      "learning_rate": 4.991686365778405e-06,
      "loss": 3.3966,
      "step": 2650
    },
    {
      "epoch": 0.027008056640625,
      "grad_norm": 12.624420166015625,
      "learning_rate": 4.991653749498504e-06,
      "loss": 3.3418,
      "step": 2655
    },
    {
      "epoch": 0.027058919270833332,
      "grad_norm": 16.91732406616211,
      "learning_rate": 4.991621069470233e-06,
      "loss": 4.1244,
      "step": 2660
    },
    {
      "epoch": 0.027109781901041668,
      "grad_norm": 15.740944862365723,
      "learning_rate": 4.991588325694426e-06,
      "loss": 3.4994,
      "step": 2665
    },
    {
      "epoch": 0.02716064453125,
      "grad_norm": 12.525944709777832,
      "learning_rate": 4.9915555181719235e-06,
      "loss": 3.7722,
      "step": 2670
    },
    {
      "epoch": 0.027211507161458332,
      "grad_norm": 13.604511260986328,
      "learning_rate": 4.991522646903564e-06,
      "loss": 3.4386,
      "step": 2675
    },
    {
      "epoch": 0.027262369791666668,
      "grad_norm": 10.147235870361328,
      "learning_rate": 4.991489711890188e-06,
      "loss": 3.5642,
      "step": 2680
    },
    {
      "epoch": 0.027313232421875,
      "grad_norm": 14.985891342163086,
      "learning_rate": 4.991456713132637e-06,
      "loss": 3.5542,
      "step": 2685
    },
    {
      "epoch": 0.027364095052083332,
      "grad_norm": 9.699203491210938,
      "learning_rate": 4.991423650631758e-06,
      "loss": 3.4597,
      "step": 2690
    },
    {
      "epoch": 0.027414957682291668,
      "grad_norm": 19.256378173828125,
      "learning_rate": 4.991390524388394e-06,
      "loss": 3.9099,
      "step": 2695
    },
    {
      "epoch": 0.0274658203125,
      "grad_norm": 13.979656219482422,
      "learning_rate": 4.991357334403396e-06,
      "loss": 3.4309,
      "step": 2700
    },
    {
      "epoch": 0.027516682942708332,
      "grad_norm": 15.490063667297363,
      "learning_rate": 4.9913240806776095e-06,
      "loss": 3.6231,
      "step": 2705
    },
    {
      "epoch": 0.027567545572916668,
      "grad_norm": 11.414307594299316,
      "learning_rate": 4.991290763211887e-06,
      "loss": 3.5332,
      "step": 2710
    },
    {
      "epoch": 0.027618408203125,
      "grad_norm": 14.588093757629395,
      "learning_rate": 4.991257382007081e-06,
      "loss": 3.8702,
      "step": 2715
    },
    {
      "epoch": 0.027669270833333332,
      "grad_norm": 11.244452476501465,
      "learning_rate": 4.9912239370640455e-06,
      "loss": 4.1972,
      "step": 2720
    },
    {
      "epoch": 0.027720133463541668,
      "grad_norm": 18.345767974853516,
      "learning_rate": 4.991190428383637e-06,
      "loss": 3.4974,
      "step": 2725
    },
    {
      "epoch": 0.02777099609375,
      "grad_norm": 16.70025634765625,
      "learning_rate": 4.99115685596671e-06,
      "loss": 3.4827,
      "step": 2730
    },
    {
      "epoch": 0.027821858723958332,
      "grad_norm": 9.72400951385498,
      "learning_rate": 4.9911232198141266e-06,
      "loss": 3.7941,
      "step": 2735
    },
    {
      "epoch": 0.027872721354166668,
      "grad_norm": 18.471965789794922,
      "learning_rate": 4.991089519926746e-06,
      "loss": 3.3165,
      "step": 2740
    },
    {
      "epoch": 0.027923583984375,
      "grad_norm": 8.832108497619629,
      "learning_rate": 4.9910557563054295e-06,
      "loss": 4.046,
      "step": 2745
    },
    {
      "epoch": 0.027974446614583332,
      "grad_norm": 13.375359535217285,
      "learning_rate": 4.991021928951043e-06,
      "loss": 3.7558,
      "step": 2750
    },
    {
      "epoch": 0.028025309244791668,
      "grad_norm": 14.425735473632812,
      "learning_rate": 4.99098803786445e-06,
      "loss": 3.7207,
      "step": 2755
    },
    {
      "epoch": 0.028076171875,
      "grad_norm": 11.840872764587402,
      "learning_rate": 4.99095408304652e-06,
      "loss": 3.8429,
      "step": 2760
    },
    {
      "epoch": 0.028127034505208332,
      "grad_norm": 11.43602466583252,
      "learning_rate": 4.990920064498119e-06,
      "loss": 3.3477,
      "step": 2765
    },
    {
      "epoch": 0.028177897135416668,
      "grad_norm": 21.40333366394043,
      "learning_rate": 4.9908859822201186e-06,
      "loss": 3.3127,
      "step": 2770
    },
    {
      "epoch": 0.028228759765625,
      "grad_norm": 9.050609588623047,
      "learning_rate": 4.990851836213391e-06,
      "loss": 3.5749,
      "step": 2775
    },
    {
      "epoch": 0.028279622395833332,
      "grad_norm": 18.72174644470215,
      "learning_rate": 4.990817626478809e-06,
      "loss": 3.4886,
      "step": 2780
    },
    {
      "epoch": 0.028330485026041668,
      "grad_norm": 17.724315643310547,
      "learning_rate": 4.990783353017249e-06,
      "loss": 3.6107,
      "step": 2785
    },
    {
      "epoch": 0.02838134765625,
      "grad_norm": 17.255958557128906,
      "learning_rate": 4.990749015829587e-06,
      "loss": 3.4667,
      "step": 2790
    },
    {
      "epoch": 0.028432210286458332,
      "grad_norm": 15.389385223388672,
      "learning_rate": 4.9907146149167025e-06,
      "loss": 3.6194,
      "step": 2795
    },
    {
      "epoch": 0.028483072916666668,
      "grad_norm": 16.21523666381836,
      "learning_rate": 4.990680150279474e-06,
      "loss": 3.7462,
      "step": 2800
    },
    {
      "epoch": 0.028533935546875,
      "grad_norm": 10.564112663269043,
      "learning_rate": 4.990645621918785e-06,
      "loss": 3.5578,
      "step": 2805
    },
    {
      "epoch": 0.028584798177083332,
      "grad_norm": 11.502965927124023,
      "learning_rate": 4.990611029835518e-06,
      "loss": 3.3145,
      "step": 2810
    },
    {
      "epoch": 0.028635660807291668,
      "grad_norm": 11.729395866394043,
      "learning_rate": 4.990576374030558e-06,
      "loss": 3.5022,
      "step": 2815
    },
    {
      "epoch": 0.0286865234375,
      "grad_norm": 17.083139419555664,
      "learning_rate": 4.9905416545047914e-06,
      "loss": 3.1965,
      "step": 2820
    },
    {
      "epoch": 0.028737386067708332,
      "grad_norm": 12.707961082458496,
      "learning_rate": 4.990506871259107e-06,
      "loss": 3.3708,
      "step": 2825
    },
    {
      "epoch": 0.028788248697916668,
      "grad_norm": 12.2274808883667,
      "learning_rate": 4.990472024294395e-06,
      "loss": 3.6543,
      "step": 2830
    },
    {
      "epoch": 0.028839111328125,
      "grad_norm": 10.415914535522461,
      "learning_rate": 4.990437113611546e-06,
      "loss": 3.3149,
      "step": 2835
    },
    {
      "epoch": 0.028889973958333332,
      "grad_norm": 14.803680419921875,
      "learning_rate": 4.990402139211454e-06,
      "loss": 3.6954,
      "step": 2840
    },
    {
      "epoch": 0.028940836588541668,
      "grad_norm": 9.584573745727539,
      "learning_rate": 4.990367101095014e-06,
      "loss": 3.2306,
      "step": 2845
    },
    {
      "epoch": 0.02899169921875,
      "grad_norm": 9.19983196258545,
      "learning_rate": 4.9903319992631215e-06,
      "loss": 3.5329,
      "step": 2850
    },
    {
      "epoch": 0.029042561848958332,
      "grad_norm": 13.915033340454102,
      "learning_rate": 4.990296833716676e-06,
      "loss": 3.5633,
      "step": 2855
    },
    {
      "epoch": 0.029093424479166668,
      "grad_norm": 8.67843246459961,
      "learning_rate": 4.990261604456575e-06,
      "loss": 3.5621,
      "step": 2860
    },
    {
      "epoch": 0.029144287109375,
      "grad_norm": 14.383540153503418,
      "learning_rate": 4.990226311483721e-06,
      "loss": 3.6706,
      "step": 2865
    },
    {
      "epoch": 0.029195149739583332,
      "grad_norm": 15.184710502624512,
      "learning_rate": 4.990190954799018e-06,
      "loss": 3.5497,
      "step": 2870
    },
    {
      "epoch": 0.029246012369791668,
      "grad_norm": 17.983572006225586,
      "learning_rate": 4.990155534403369e-06,
      "loss": 3.2129,
      "step": 2875
    },
    {
      "epoch": 0.029296875,
      "grad_norm": 19.197294235229492,
      "learning_rate": 4.9901200502976825e-06,
      "loss": 3.7564,
      "step": 2880
    },
    {
      "epoch": 0.029347737630208332,
      "grad_norm": 14.85810661315918,
      "learning_rate": 4.990084502482863e-06,
      "loss": 3.4496,
      "step": 2885
    },
    {
      "epoch": 0.029398600260416668,
      "grad_norm": 12.596855163574219,
      "learning_rate": 4.990048890959822e-06,
      "loss": 3.7283,
      "step": 2890
    },
    {
      "epoch": 0.029449462890625,
      "grad_norm": 17.732402801513672,
      "learning_rate": 4.99001321572947e-06,
      "loss": 3.8592,
      "step": 2895
    },
    {
      "epoch": 0.029500325520833332,
      "grad_norm": 12.990941047668457,
      "learning_rate": 4.989977476792721e-06,
      "loss": 3.5972,
      "step": 2900
    },
    {
      "epoch": 0.029551188151041668,
      "grad_norm": 15.059063911437988,
      "learning_rate": 4.989941674150488e-06,
      "loss": 3.6941,
      "step": 2905
    },
    {
      "epoch": 0.02960205078125,
      "grad_norm": 13.130831718444824,
      "learning_rate": 4.989905807803688e-06,
      "loss": 3.5952,
      "step": 2910
    },
    {
      "epoch": 0.029652913411458332,
      "grad_norm": 15.897665977478027,
      "learning_rate": 4.989869877753237e-06,
      "loss": 3.5691,
      "step": 2915
    },
    {
      "epoch": 0.029703776041666668,
      "grad_norm": 11.824736595153809,
      "learning_rate": 4.989833884000056e-06,
      "loss": 3.7098,
      "step": 2920
    },
    {
      "epoch": 0.029754638671875,
      "grad_norm": 14.641777038574219,
      "learning_rate": 4.989797826545065e-06,
      "loss": 3.6694,
      "step": 2925
    },
    {
      "epoch": 0.029805501302083332,
      "grad_norm": 12.433199882507324,
      "learning_rate": 4.989761705389187e-06,
      "loss": 3.5255,
      "step": 2930
    },
    {
      "epoch": 0.029856363932291668,
      "grad_norm": 12.532072067260742,
      "learning_rate": 4.989725520533346e-06,
      "loss": 3.5487,
      "step": 2935
    },
    {
      "epoch": 0.0299072265625,
      "grad_norm": 16.237571716308594,
      "learning_rate": 4.9896892719784675e-06,
      "loss": 3.562,
      "step": 2940
    },
    {
      "epoch": 0.029958089192708332,
      "grad_norm": 15.376089096069336,
      "learning_rate": 4.989652959725479e-06,
      "loss": 3.4396,
      "step": 2945
    },
    {
      "epoch": 0.030008951822916668,
      "grad_norm": 13.294553756713867,
      "learning_rate": 4.98961658377531e-06,
      "loss": 3.8229,
      "step": 2950
    },
    {
      "epoch": 0.030059814453125,
      "grad_norm": 15.016079902648926,
      "learning_rate": 4.98958014412889e-06,
      "loss": 3.4993,
      "step": 2955
    },
    {
      "epoch": 0.030110677083333332,
      "grad_norm": 16.527847290039062,
      "learning_rate": 4.989543640787153e-06,
      "loss": 3.6308,
      "step": 2960
    },
    {
      "epoch": 0.030161539713541668,
      "grad_norm": 15.960955619812012,
      "learning_rate": 4.989507073751032e-06,
      "loss": 3.5736,
      "step": 2965
    },
    {
      "epoch": 0.03021240234375,
      "grad_norm": 19.20722198486328,
      "learning_rate": 4.989470443021462e-06,
      "loss": 3.3009,
      "step": 2970
    },
    {
      "epoch": 0.030263264973958332,
      "grad_norm": 11.201433181762695,
      "learning_rate": 4.989433748599381e-06,
      "loss": 3.6235,
      "step": 2975
    },
    {
      "epoch": 0.030314127604166668,
      "grad_norm": 13.62901782989502,
      "learning_rate": 4.989396990485727e-06,
      "loss": 4.1232,
      "step": 2980
    },
    {
      "epoch": 0.030364990234375,
      "grad_norm": 15.175191879272461,
      "learning_rate": 4.989360168681442e-06,
      "loss": 3.6938,
      "step": 2985
    },
    {
      "epoch": 0.030415852864583332,
      "grad_norm": 12.873221397399902,
      "learning_rate": 4.9893232831874676e-06,
      "loss": 3.7898,
      "step": 2990
    },
    {
      "epoch": 0.030466715494791668,
      "grad_norm": 10.560770034790039,
      "learning_rate": 4.989286334004746e-06,
      "loss": 3.1075,
      "step": 2995
    },
    {
      "epoch": 0.030517578125,
      "grad_norm": 16.364398956298828,
      "learning_rate": 4.9892493211342235e-06,
      "loss": 3.8729,
      "step": 3000
    },
    {
      "epoch": 0.030568440755208332,
      "grad_norm": 14.276408195495605,
      "learning_rate": 4.989212244576848e-06,
      "loss": 3.4278,
      "step": 3005
    },
    {
      "epoch": 0.030619303385416668,
      "grad_norm": 17.391857147216797,
      "learning_rate": 4.989175104333567e-06,
      "loss": 3.5136,
      "step": 3010
    },
    {
      "epoch": 0.030670166015625,
      "grad_norm": 16.542964935302734,
      "learning_rate": 4.98913790040533e-06,
      "loss": 3.7648,
      "step": 3015
    },
    {
      "epoch": 0.030721028645833332,
      "grad_norm": 12.984308242797852,
      "learning_rate": 4.9891006327930905e-06,
      "loss": 3.2495,
      "step": 3020
    },
    {
      "epoch": 0.030771891276041668,
      "grad_norm": 16.2092227935791,
      "learning_rate": 4.989063301497801e-06,
      "loss": 3.2186,
      "step": 3025
    },
    {
      "epoch": 0.03082275390625,
      "grad_norm": 9.466002464294434,
      "learning_rate": 4.989025906520417e-06,
      "loss": 3.5156,
      "step": 3030
    },
    {
      "epoch": 0.030873616536458332,
      "grad_norm": 11.886150360107422,
      "learning_rate": 4.988988447861895e-06,
      "loss": 3.5858,
      "step": 3035
    },
    {
      "epoch": 0.030924479166666668,
      "grad_norm": 10.326323509216309,
      "learning_rate": 4.988950925523194e-06,
      "loss": 3.3724,
      "step": 3040
    },
    {
      "epoch": 0.030975341796875,
      "grad_norm": 15.874045372009277,
      "learning_rate": 4.988913339505274e-06,
      "loss": 3.5452,
      "step": 3045
    },
    {
      "epoch": 0.031026204427083332,
      "grad_norm": 11.754904747009277,
      "learning_rate": 4.988875689809095e-06,
      "loss": 3.4911,
      "step": 3050
    },
    {
      "epoch": 0.031077067057291668,
      "grad_norm": 17.041898727416992,
      "learning_rate": 4.988837976435622e-06,
      "loss": 3.6784,
      "step": 3055
    },
    {
      "epoch": 0.0311279296875,
      "grad_norm": 11.09570026397705,
      "learning_rate": 4.988800199385819e-06,
      "loss": 3.366,
      "step": 3060
    },
    {
      "epoch": 0.031178792317708332,
      "grad_norm": 14.231475830078125,
      "learning_rate": 4.988762358660654e-06,
      "loss": 3.3146,
      "step": 3065
    },
    {
      "epoch": 0.031229654947916668,
      "grad_norm": 11.312032699584961,
      "learning_rate": 4.988724454261092e-06,
      "loss": 3.6764,
      "step": 3070
    },
    {
      "epoch": 0.031280517578125,
      "grad_norm": 15.22202205657959,
      "learning_rate": 4.988686486188105e-06,
      "loss": 3.4084,
      "step": 3075
    },
    {
      "epoch": 0.031331380208333336,
      "grad_norm": 14.88944149017334,
      "learning_rate": 4.988648454442666e-06,
      "loss": 3.1168,
      "step": 3080
    },
    {
      "epoch": 0.031382242838541664,
      "grad_norm": 10.730203628540039,
      "learning_rate": 4.988610359025745e-06,
      "loss": 3.3345,
      "step": 3085
    },
    {
      "epoch": 0.03143310546875,
      "grad_norm": 13.363783836364746,
      "learning_rate": 4.988572199938317e-06,
      "loss": 3.0878,
      "step": 3090
    },
    {
      "epoch": 0.031483968098958336,
      "grad_norm": 21.714414596557617,
      "learning_rate": 4.9885339771813604e-06,
      "loss": 3.6447,
      "step": 3095
    },
    {
      "epoch": 0.031534830729166664,
      "grad_norm": 18.616336822509766,
      "learning_rate": 4.9884956907558515e-06,
      "loss": 3.5968,
      "step": 3100
    },
    {
      "epoch": 0.031585693359375,
      "grad_norm": 12.650989532470703,
      "learning_rate": 4.98845734066277e-06,
      "loss": 3.4597,
      "step": 3105
    },
    {
      "epoch": 0.031636555989583336,
      "grad_norm": 14.19465446472168,
      "learning_rate": 4.988418926903098e-06,
      "loss": 3.399,
      "step": 3110
    },
    {
      "epoch": 0.031687418619791664,
      "grad_norm": 9.14580249786377,
      "learning_rate": 4.9883804494778165e-06,
      "loss": 3.6816,
      "step": 3115
    },
    {
      "epoch": 0.03173828125,
      "grad_norm": 10.480582237243652,
      "learning_rate": 4.988341908387912e-06,
      "loss": 3.128,
      "step": 3120
    },
    {
      "epoch": 0.031789143880208336,
      "grad_norm": 10.952704429626465,
      "learning_rate": 4.988303303634368e-06,
      "loss": 3.4517,
      "step": 3125
    },
    {
      "epoch": 0.031840006510416664,
      "grad_norm": 12.522828102111816,
      "learning_rate": 4.988264635218175e-06,
      "loss": 3.9186,
      "step": 3130
    },
    {
      "epoch": 0.031890869140625,
      "grad_norm": 12.782024383544922,
      "learning_rate": 4.988225903140321e-06,
      "loss": 3.725,
      "step": 3135
    },
    {
      "epoch": 0.031941731770833336,
      "grad_norm": 11.72850227355957,
      "learning_rate": 4.988187107401797e-06,
      "loss": 3.7264,
      "step": 3140
    },
    {
      "epoch": 0.031992594401041664,
      "grad_norm": 12.133509635925293,
      "learning_rate": 4.988148248003595e-06,
      "loss": 3.4472,
      "step": 3145
    },
    {
      "epoch": 0.03204345703125,
      "grad_norm": 12.93433666229248,
      "learning_rate": 4.98810932494671e-06,
      "loss": 3.9443,
      "step": 3150
    },
    {
      "epoch": 0.032094319661458336,
      "grad_norm": 16.135568618774414,
      "learning_rate": 4.988070338232138e-06,
      "loss": 3.4966,
      "step": 3155
    },
    {
      "epoch": 0.032145182291666664,
      "grad_norm": 18.191791534423828,
      "learning_rate": 4.988031287860877e-06,
      "loss": 3.5644,
      "step": 3160
    },
    {
      "epoch": 0.032196044921875,
      "grad_norm": 11.924603462219238,
      "learning_rate": 4.987992173833924e-06,
      "loss": 3.4058,
      "step": 3165
    },
    {
      "epoch": 0.032246907552083336,
      "grad_norm": 12.82744026184082,
      "learning_rate": 4.987952996152281e-06,
      "loss": 3.492,
      "step": 3170
    },
    {
      "epoch": 0.032297770182291664,
      "grad_norm": 8.64533519744873,
      "learning_rate": 4.987913754816951e-06,
      "loss": 3.9099,
      "step": 3175
    },
    {
      "epoch": 0.0323486328125,
      "grad_norm": 16.365373611450195,
      "learning_rate": 4.987874449828937e-06,
      "loss": 3.1102,
      "step": 3180
    },
    {
      "epoch": 0.032399495442708336,
      "grad_norm": 15.942273139953613,
      "learning_rate": 4.987835081189245e-06,
      "loss": 3.7335,
      "step": 3185
    },
    {
      "epoch": 0.032450358072916664,
      "grad_norm": 17.56052589416504,
      "learning_rate": 4.987795648898882e-06,
      "loss": 3.8561,
      "step": 3190
    },
    {
      "epoch": 0.032501220703125,
      "grad_norm": 9.97986125946045,
      "learning_rate": 4.987756152958857e-06,
      "loss": 3.3722,
      "step": 3195
    },
    {
      "epoch": 0.032552083333333336,
      "grad_norm": 13.236599922180176,
      "learning_rate": 4.98771659337018e-06,
      "loss": 3.5341,
      "step": 3200
    },
    {
      "epoch": 0.032602945963541664,
      "grad_norm": 15.567898750305176,
      "learning_rate": 4.987676970133864e-06,
      "loss": 3.6612,
      "step": 3205
    },
    {
      "epoch": 0.03265380859375,
      "grad_norm": 14.741514205932617,
      "learning_rate": 4.987637283250923e-06,
      "loss": 3.7577,
      "step": 3210
    },
    {
      "epoch": 0.032704671223958336,
      "grad_norm": 12.727431297302246,
      "learning_rate": 4.98759753272237e-06,
      "loss": 4.3001,
      "step": 3215
    },
    {
      "epoch": 0.032755533854166664,
      "grad_norm": 11.071520805358887,
      "learning_rate": 4.987557718549225e-06,
      "loss": 3.61,
      "step": 3220
    },
    {
      "epoch": 0.032806396484375,
      "grad_norm": 10.508591651916504,
      "learning_rate": 4.987517840732505e-06,
      "loss": 3.3211,
      "step": 3225
    },
    {
      "epoch": 0.032857259114583336,
      "grad_norm": 12.959603309631348,
      "learning_rate": 4.987477899273232e-06,
      "loss": 3.5145,
      "step": 3230
    },
    {
      "epoch": 0.032908121744791664,
      "grad_norm": 14.553711891174316,
      "learning_rate": 4.987437894172426e-06,
      "loss": 3.9318,
      "step": 3235
    },
    {
      "epoch": 0.032958984375,
      "grad_norm": 14.899251937866211,
      "learning_rate": 4.987397825431109e-06,
      "loss": 3.5583,
      "step": 3240
    },
    {
      "epoch": 0.033009847005208336,
      "grad_norm": 11.648773193359375,
      "learning_rate": 4.98735769305031e-06,
      "loss": 3.2714,
      "step": 3245
    },
    {
      "epoch": 0.033060709635416664,
      "grad_norm": 13.525915145874023,
      "learning_rate": 4.987317497031055e-06,
      "loss": 3.3922,
      "step": 3250
    },
    {
      "epoch": 0.033111572265625,
      "grad_norm": 16.7314453125,
      "learning_rate": 4.987277237374369e-06,
      "loss": 3.6725,
      "step": 3255
    },
    {
      "epoch": 0.033162434895833336,
      "grad_norm": 16.31707191467285,
      "learning_rate": 4.987236914081286e-06,
      "loss": 3.3669,
      "step": 3260
    },
    {
      "epoch": 0.033213297526041664,
      "grad_norm": 11.438323974609375,
      "learning_rate": 4.987196527152835e-06,
      "loss": 3.4222,
      "step": 3265
    },
    {
      "epoch": 0.03326416015625,
      "grad_norm": 14.705175399780273,
      "learning_rate": 4.987156076590051e-06,
      "loss": 3.6076,
      "step": 3270
    },
    {
      "epoch": 0.033315022786458336,
      "grad_norm": 14.206781387329102,
      "learning_rate": 4.987115562393969e-06,
      "loss": 3.4704,
      "step": 3275
    },
    {
      "epoch": 0.033365885416666664,
      "grad_norm": 10.712113380432129,
      "learning_rate": 4.987074984565624e-06,
      "loss": 3.575,
      "step": 3280
    },
    {
      "epoch": 0.033416748046875,
      "grad_norm": 12.62575626373291,
      "learning_rate": 4.987034343106055e-06,
      "loss": 3.2375,
      "step": 3285
    },
    {
      "epoch": 0.033467610677083336,
      "grad_norm": 13.155769348144531,
      "learning_rate": 4.986993638016302e-06,
      "loss": 3.0587,
      "step": 3290
    },
    {
      "epoch": 0.033518473307291664,
      "grad_norm": 15.919364929199219,
      "learning_rate": 4.986952869297407e-06,
      "loss": 3.569,
      "step": 3295
    },
    {
      "epoch": 0.0335693359375,
      "grad_norm": 16.994239807128906,
      "learning_rate": 4.986912036950411e-06,
      "loss": 3.7481,
      "step": 3300
    },
    {
      "epoch": 0.033620198567708336,
      "grad_norm": 13.510692596435547,
      "learning_rate": 4.986871140976361e-06,
      "loss": 3.5751,
      "step": 3305
    },
    {
      "epoch": 0.033671061197916664,
      "grad_norm": 14.510315895080566,
      "learning_rate": 4.986830181376302e-06,
      "loss": 3.2262,
      "step": 3310
    },
    {
      "epoch": 0.033721923828125,
      "grad_norm": 13.225347518920898,
      "learning_rate": 4.986789158151282e-06,
      "loss": 3.3544,
      "step": 3315
    },
    {
      "epoch": 0.033772786458333336,
      "grad_norm": 15.174870491027832,
      "learning_rate": 4.9867480713023506e-06,
      "loss": 3.39,
      "step": 3320
    },
    {
      "epoch": 0.033823649088541664,
      "grad_norm": 13.420331954956055,
      "learning_rate": 4.98670692083056e-06,
      "loss": 3.4115,
      "step": 3325
    },
    {
      "epoch": 0.03387451171875,
      "grad_norm": 14.904714584350586,
      "learning_rate": 4.986665706736962e-06,
      "loss": 3.5199,
      "step": 3330
    },
    {
      "epoch": 0.033925374348958336,
      "grad_norm": 9.973340034484863,
      "learning_rate": 4.986624429022611e-06,
      "loss": 3.5716,
      "step": 3335
    },
    {
      "epoch": 0.033976236979166664,
      "grad_norm": 12.208763122558594,
      "learning_rate": 4.986583087688563e-06,
      "loss": 3.6914,
      "step": 3340
    },
    {
      "epoch": 0.034027099609375,
      "grad_norm": 13.434621810913086,
      "learning_rate": 4.986541682735877e-06,
      "loss": 3.2363,
      "step": 3345
    },
    {
      "epoch": 0.034077962239583336,
      "grad_norm": 17.55096435546875,
      "learning_rate": 4.986500214165611e-06,
      "loss": 3.4762,
      "step": 3350
    },
    {
      "epoch": 0.034128824869791664,
      "grad_norm": 11.070765495300293,
      "learning_rate": 4.986458681978826e-06,
      "loss": 3.6086,
      "step": 3355
    },
    {
      "epoch": 0.0341796875,
      "grad_norm": 12.782002449035645,
      "learning_rate": 4.986417086176586e-06,
      "loss": 3.3807,
      "step": 3360
    },
    {
      "epoch": 0.034230550130208336,
      "grad_norm": 11.93525218963623,
      "learning_rate": 4.9863754267599535e-06,
      "loss": 3.6002,
      "step": 3365
    },
    {
      "epoch": 0.034281412760416664,
      "grad_norm": 12.5736083984375,
      "learning_rate": 4.986333703729995e-06,
      "loss": 3.6075,
      "step": 3370
    },
    {
      "epoch": 0.034332275390625,
      "grad_norm": 14.359274864196777,
      "learning_rate": 4.986291917087778e-06,
      "loss": 3.461,
      "step": 3375
    },
    {
      "epoch": 0.034383138020833336,
      "grad_norm": 12.868420600891113,
      "learning_rate": 4.9862500668343714e-06,
      "loss": 3.9434,
      "step": 3380
    },
    {
      "epoch": 0.034434000651041664,
      "grad_norm": 18.13227653503418,
      "learning_rate": 4.986208152970847e-06,
      "loss": 3.2659,
      "step": 3385
    },
    {
      "epoch": 0.03448486328125,
      "grad_norm": 8.368807792663574,
      "learning_rate": 4.986166175498276e-06,
      "loss": 3.6618,
      "step": 3390
    },
    {
      "epoch": 0.034535725911458336,
      "grad_norm": 11.16234302520752,
      "learning_rate": 4.986124134417732e-06,
      "loss": 3.3186,
      "step": 3395
    },
    {
      "epoch": 0.034586588541666664,
      "grad_norm": 14.434364318847656,
      "learning_rate": 4.986082029730292e-06,
      "loss": 3.8224,
      "step": 3400
    },
    {
      "epoch": 0.034637451171875,
      "grad_norm": 11.538227081298828,
      "learning_rate": 4.9860398614370324e-06,
      "loss": 3.3252,
      "step": 3405
    },
    {
      "epoch": 0.034688313802083336,
      "grad_norm": 15.10848617553711,
      "learning_rate": 4.985997629539032e-06,
      "loss": 3.3582,
      "step": 3410
    },
    {
      "epoch": 0.034739176432291664,
      "grad_norm": 12.740103721618652,
      "learning_rate": 4.985955334037372e-06,
      "loss": 3.672,
      "step": 3415
    },
    {
      "epoch": 0.0347900390625,
      "grad_norm": 15.706042289733887,
      "learning_rate": 4.985912974933134e-06,
      "loss": 3.379,
      "step": 3420
    },
    {
      "epoch": 0.034840901692708336,
      "grad_norm": 21.957704544067383,
      "learning_rate": 4.985870552227401e-06,
      "loss": 3.8408,
      "step": 3425
    },
    {
      "epoch": 0.034891764322916664,
      "grad_norm": 8.741256713867188,
      "learning_rate": 4.9858280659212595e-06,
      "loss": 3.4575,
      "step": 3430
    },
    {
      "epoch": 0.034942626953125,
      "grad_norm": 16.37959098815918,
      "learning_rate": 4.9857855160157965e-06,
      "loss": 3.7038,
      "step": 3435
    },
    {
      "epoch": 0.034993489583333336,
      "grad_norm": 15.630584716796875,
      "learning_rate": 4.9857429025120996e-06,
      "loss": 3.3838,
      "step": 3440
    },
    {
      "epoch": 0.035044352213541664,
      "grad_norm": 13.591779708862305,
      "learning_rate": 4.98570022541126e-06,
      "loss": 3.4661,
      "step": 3445
    },
    {
      "epoch": 0.03509521484375,
      "grad_norm": 14.534747123718262,
      "learning_rate": 4.985657484714369e-06,
      "loss": 3.487,
      "step": 3450
    },
    {
      "epoch": 0.035146077473958336,
      "grad_norm": 10.500582695007324,
      "learning_rate": 4.985614680422521e-06,
      "loss": 3.6731,
      "step": 3455
    },
    {
      "epoch": 0.035196940104166664,
      "grad_norm": 10.115405082702637,
      "learning_rate": 4.9855718125368105e-06,
      "loss": 3.4601,
      "step": 3460
    },
    {
      "epoch": 0.035247802734375,
      "grad_norm": 17.56122398376465,
      "learning_rate": 4.985528881058334e-06,
      "loss": 3.6548,
      "step": 3465
    },
    {
      "epoch": 0.035298665364583336,
      "grad_norm": 9.477646827697754,
      "learning_rate": 4.9854858859881905e-06,
      "loss": 3.3522,
      "step": 3470
    },
    {
      "epoch": 0.035349527994791664,
      "grad_norm": 12.787845611572266,
      "learning_rate": 4.985442827327479e-06,
      "loss": 3.3005,
      "step": 3475
    },
    {
      "epoch": 0.035400390625,
      "grad_norm": 9.624972343444824,
      "learning_rate": 4.985399705077303e-06,
      "loss": 3.6494,
      "step": 3480
    },
    {
      "epoch": 0.035451253255208336,
      "grad_norm": 16.32085609436035,
      "learning_rate": 4.985356519238764e-06,
      "loss": 3.6078,
      "step": 3485
    },
    {
      "epoch": 0.035502115885416664,
      "grad_norm": 14.953262329101562,
      "learning_rate": 4.985313269812968e-06,
      "loss": 3.256,
      "step": 3490
    },
    {
      "epoch": 0.035552978515625,
      "grad_norm": 19.811403274536133,
      "learning_rate": 4.985269956801021e-06,
      "loss": 3.6117,
      "step": 3495
    },
    {
      "epoch": 0.035603841145833336,
      "grad_norm": 15.88729190826416,
      "learning_rate": 4.985226580204031e-06,
      "loss": 3.5138,
      "step": 3500
    },
    {
      "epoch": 0.035654703776041664,
      "grad_norm": 16.489349365234375,
      "learning_rate": 4.9851831400231075e-06,
      "loss": 3.5093,
      "step": 3505
    },
    {
      "epoch": 0.03570556640625,
      "grad_norm": 11.760025024414062,
      "learning_rate": 4.985139636259363e-06,
      "loss": 4.247,
      "step": 3510
    },
    {
      "epoch": 0.035756429036458336,
      "grad_norm": 12.459917068481445,
      "learning_rate": 4.98509606891391e-06,
      "loss": 3.4997,
      "step": 3515
    },
    {
      "epoch": 0.035807291666666664,
      "grad_norm": 10.306233406066895,
      "learning_rate": 4.985052437987863e-06,
      "loss": 3.4573,
      "step": 3520
    },
    {
      "epoch": 0.035858154296875,
      "grad_norm": 16.409082412719727,
      "learning_rate": 4.9850087434823384e-06,
      "loss": 3.4351,
      "step": 3525
    },
    {
      "epoch": 0.035909016927083336,
      "grad_norm": 9.501917839050293,
      "learning_rate": 4.984964985398454e-06,
      "loss": 3.4707,
      "step": 3530
    },
    {
      "epoch": 0.035959879557291664,
      "grad_norm": 13.083230972290039,
      "learning_rate": 4.98492116373733e-06,
      "loss": 3.3439,
      "step": 3535
    },
    {
      "epoch": 0.0360107421875,
      "grad_norm": 11.03459358215332,
      "learning_rate": 4.984877278500087e-06,
      "loss": 3.3296,
      "step": 3540
    },
    {
      "epoch": 0.036061604817708336,
      "grad_norm": 15.221756935119629,
      "learning_rate": 4.984833329687847e-06,
      "loss": 3.5049,
      "step": 3545
    },
    {
      "epoch": 0.036112467447916664,
      "grad_norm": 12.892142295837402,
      "learning_rate": 4.9847893173017345e-06,
      "loss": 3.6549,
      "step": 3550
    },
    {
      "epoch": 0.036163330078125,
      "grad_norm": 13.091279029846191,
      "learning_rate": 4.984745241342877e-06,
      "loss": 3.6493,
      "step": 3555
    },
    {
      "epoch": 0.036214192708333336,
      "grad_norm": 14.84507942199707,
      "learning_rate": 4.984701101812402e-06,
      "loss": 3.5608,
      "step": 3560
    },
    {
      "epoch": 0.036265055338541664,
      "grad_norm": 12.539996147155762,
      "learning_rate": 4.984656898711438e-06,
      "loss": 3.5464,
      "step": 3565
    },
    {
      "epoch": 0.03631591796875,
      "grad_norm": 12.54378604888916,
      "learning_rate": 4.984612632041117e-06,
      "loss": 3.7036,
      "step": 3570
    },
    {
      "epoch": 0.036366780598958336,
      "grad_norm": 15.650618553161621,
      "learning_rate": 4.98456830180257e-06,
      "loss": 3.7464,
      "step": 3575
    },
    {
      "epoch": 0.036417643229166664,
      "grad_norm": 12.167283058166504,
      "learning_rate": 4.984523907996932e-06,
      "loss": 3.6591,
      "step": 3580
    },
    {
      "epoch": 0.036468505859375,
      "grad_norm": 13.527252197265625,
      "learning_rate": 4.984479450625338e-06,
      "loss": 3.5705,
      "step": 3585
    },
    {
      "epoch": 0.036519368489583336,
      "grad_norm": 16.410930633544922,
      "learning_rate": 4.9844349296889275e-06,
      "loss": 3.7471,
      "step": 3590
    },
    {
      "epoch": 0.036570231119791664,
      "grad_norm": 10.030250549316406,
      "learning_rate": 4.984390345188838e-06,
      "loss": 3.841,
      "step": 3595
    },
    {
      "epoch": 0.03662109375,
      "grad_norm": 11.255728721618652,
      "learning_rate": 4.9843456971262095e-06,
      "loss": 3.5435,
      "step": 3600
    },
    {
      "epoch": 0.036671956380208336,
      "grad_norm": 9.756819725036621,
      "learning_rate": 4.984300985502185e-06,
      "loss": 3.3497,
      "step": 3605
    },
    {
      "epoch": 0.036722819010416664,
      "grad_norm": 12.754555702209473,
      "learning_rate": 4.984256210317909e-06,
      "loss": 3.723,
      "step": 3610
    },
    {
      "epoch": 0.036773681640625,
      "grad_norm": 15.393765449523926,
      "learning_rate": 4.984211371574527e-06,
      "loss": 3.99,
      "step": 3615
    },
    {
      "epoch": 0.036824544270833336,
      "grad_norm": 16.75415802001953,
      "learning_rate": 4.984166469273186e-06,
      "loss": 3.1881,
      "step": 3620
    },
    {
      "epoch": 0.036875406901041664,
      "grad_norm": 15.176877975463867,
      "learning_rate": 4.984121503415034e-06,
      "loss": 3.6833,
      "step": 3625
    },
    {
      "epoch": 0.03692626953125,
      "grad_norm": 95.34676361083984,
      "learning_rate": 4.9840764740012225e-06,
      "loss": 3.5113,
      "step": 3630
    },
    {
      "epoch": 0.036977132161458336,
      "grad_norm": 12.578669548034668,
      "learning_rate": 4.984031381032903e-06,
      "loss": 4.0164,
      "step": 3635
    },
    {
      "epoch": 0.037027994791666664,
      "grad_norm": 12.883451461791992,
      "learning_rate": 4.98398622451123e-06,
      "loss": 3.2015,
      "step": 3640
    },
    {
      "epoch": 0.037078857421875,
      "grad_norm": 14.403608322143555,
      "learning_rate": 4.983941004437358e-06,
      "loss": 3.2212,
      "step": 3645
    },
    {
      "epoch": 0.037129720052083336,
      "grad_norm": 14.960567474365234,
      "learning_rate": 4.983895720812444e-06,
      "loss": 3.2845,
      "step": 3650
    },
    {
      "epoch": 0.037180582682291664,
      "grad_norm": 9.842466354370117,
      "learning_rate": 4.9838503736376465e-06,
      "loss": 3.4744,
      "step": 3655
    },
    {
      "epoch": 0.0372314453125,
      "grad_norm": 9.656144142150879,
      "learning_rate": 4.983804962914126e-06,
      "loss": 3.6732,
      "step": 3660
    },
    {
      "epoch": 0.037282307942708336,
      "grad_norm": 15.675665855407715,
      "learning_rate": 4.983759488643045e-06,
      "loss": 3.4493,
      "step": 3665
    },
    {
      "epoch": 0.037333170572916664,
      "grad_norm": 11.249836921691895,
      "learning_rate": 4.983713950825565e-06,
      "loss": 3.6472,
      "step": 3670
    },
    {
      "epoch": 0.037384033203125,
      "grad_norm": 16.177907943725586,
      "learning_rate": 4.983668349462853e-06,
      "loss": 3.3561,
      "step": 3675
    },
    {
      "epoch": 0.037434895833333336,
      "grad_norm": 14.318443298339844,
      "learning_rate": 4.983622684556075e-06,
      "loss": 3.4628,
      "step": 3680
    },
    {
      "epoch": 0.037485758463541664,
      "grad_norm": 14.132331848144531,
      "learning_rate": 4.9835769561064e-06,
      "loss": 3.4866,
      "step": 3685
    },
    {
      "epoch": 0.03753662109375,
      "grad_norm": 12.774168014526367,
      "learning_rate": 4.9835311641149955e-06,
      "loss": 3.2058,
      "step": 3690
    },
    {
      "epoch": 0.037587483723958336,
      "grad_norm": 10.149826049804688,
      "learning_rate": 4.983485308583036e-06,
      "loss": 3.2685,
      "step": 3695
    },
    {
      "epoch": 0.037638346354166664,
      "grad_norm": 12.236494064331055,
      "learning_rate": 4.983439389511693e-06,
      "loss": 3.3043,
      "step": 3700
    },
    {
      "epoch": 0.037689208984375,
      "grad_norm": 13.76052188873291,
      "learning_rate": 4.983393406902142e-06,
      "loss": 3.6717,
      "step": 3705
    },
    {
      "epoch": 0.037740071614583336,
      "grad_norm": 16.32668113708496,
      "learning_rate": 4.983347360755559e-06,
      "loss": 3.6747,
      "step": 3710
    },
    {
      "epoch": 0.037790934244791664,
      "grad_norm": 14.244294166564941,
      "learning_rate": 4.983301251073124e-06,
      "loss": 3.8085,
      "step": 3715
    },
    {
      "epoch": 0.037841796875,
      "grad_norm": 15.147490501403809,
      "learning_rate": 4.983255077856014e-06,
      "loss": 3.2321,
      "step": 3720
    },
    {
      "epoch": 0.037892659505208336,
      "grad_norm": 12.199444770812988,
      "learning_rate": 4.983208841105411e-06,
      "loss": 3.4871,
      "step": 3725
    },
    {
      "epoch": 0.037943522135416664,
      "grad_norm": 18.730581283569336,
      "learning_rate": 4.983162540822498e-06,
      "loss": 4.198,
      "step": 3730
    },
    {
      "epoch": 0.037994384765625,
      "grad_norm": 14.874238967895508,
      "learning_rate": 4.983116177008461e-06,
      "loss": 3.5318,
      "step": 3735
    },
    {
      "epoch": 0.038045247395833336,
      "grad_norm": 10.15190601348877,
      "learning_rate": 4.9830697496644855e-06,
      "loss": 3.3821,
      "step": 3740
    },
    {
      "epoch": 0.038096110026041664,
      "grad_norm": 17.800188064575195,
      "learning_rate": 4.983023258791758e-06,
      "loss": 3.5971,
      "step": 3745
    },
    {
      "epoch": 0.03814697265625,
      "grad_norm": 45.00802993774414,
      "learning_rate": 4.98297670439147e-06,
      "loss": 3.0957,
      "step": 3750
    },
    {
      "epoch": 0.038197835286458336,
      "grad_norm": 11.17233657836914,
      "learning_rate": 4.9829300864648104e-06,
      "loss": 3.6894,
      "step": 3755
    },
    {
      "epoch": 0.038248697916666664,
      "grad_norm": 16.200355529785156,
      "learning_rate": 4.982883405012974e-06,
      "loss": 3.2511,
      "step": 3760
    },
    {
      "epoch": 0.038299560546875,
      "grad_norm": 11.851147651672363,
      "learning_rate": 4.982836660037154e-06,
      "loss": 3.31,
      "step": 3765
    },
    {
      "epoch": 0.038350423177083336,
      "grad_norm": 8.351678848266602,
      "learning_rate": 4.982789851538545e-06,
      "loss": 3.1318,
      "step": 3770
    },
    {
      "epoch": 0.038401285807291664,
      "grad_norm": 12.307997703552246,
      "learning_rate": 4.982742979518348e-06,
      "loss": 3.3283,
      "step": 3775
    },
    {
      "epoch": 0.0384521484375,
      "grad_norm": 11.954642295837402,
      "learning_rate": 4.98269604397776e-06,
      "loss": 3.7551,
      "step": 3780
    },
    {
      "epoch": 0.038503011067708336,
      "grad_norm": 11.246746063232422,
      "learning_rate": 4.982649044917982e-06,
      "loss": 3.5243,
      "step": 3785
    },
    {
      "epoch": 0.038553873697916664,
      "grad_norm": 19.319927215576172,
      "learning_rate": 4.982601982340216e-06,
      "loss": 3.546,
      "step": 3790
    },
    {
      "epoch": 0.038604736328125,
      "grad_norm": 15.717957496643066,
      "learning_rate": 4.982554856245668e-06,
      "loss": 3.9242,
      "step": 3795
    },
    {
      "epoch": 0.038655598958333336,
      "grad_norm": 10.940032005310059,
      "learning_rate": 4.982507666635541e-06,
      "loss": 3.5011,
      "step": 3800
    },
    {
      "epoch": 0.038706461588541664,
      "grad_norm": 17.505874633789062,
      "learning_rate": 4.982460413511045e-06,
      "loss": 3.4926,
      "step": 3805
    },
    {
      "epoch": 0.03875732421875,
      "grad_norm": 13.361002922058105,
      "learning_rate": 4.9824130968733875e-06,
      "loss": 3.431,
      "step": 3810
    },
    {
      "epoch": 0.038808186848958336,
      "grad_norm": 11.022466659545898,
      "learning_rate": 4.982365716723779e-06,
      "loss": 3.4269,
      "step": 3815
    },
    {
      "epoch": 0.038859049479166664,
      "grad_norm": 13.40270709991455,
      "learning_rate": 4.982318273063432e-06,
      "loss": 3.4141,
      "step": 3820
    },
    {
      "epoch": 0.038909912109375,
      "grad_norm": 12.13808536529541,
      "learning_rate": 4.98227076589356e-06,
      "loss": 3.4167,
      "step": 3825
    },
    {
      "epoch": 0.038960774739583336,
      "grad_norm": 13.911450386047363,
      "learning_rate": 4.98222319521538e-06,
      "loss": 3.5935,
      "step": 3830
    },
    {
      "epoch": 0.039011637369791664,
      "grad_norm": 9.00546646118164,
      "learning_rate": 4.982175561030107e-06,
      "loss": 3.4011,
      "step": 3835
    },
    {
      "epoch": 0.0390625,
      "grad_norm": 15.89461612701416,
      "learning_rate": 4.982127863338961e-06,
      "loss": 3.3942,
      "step": 3840
    },
    {
      "epoch": 0.039113362630208336,
      "grad_norm": 15.606123924255371,
      "learning_rate": 4.982080102143161e-06,
      "loss": 3.4825,
      "step": 3845
    },
    {
      "epoch": 0.039164225260416664,
      "grad_norm": 16.339702606201172,
      "learning_rate": 4.982032277443931e-06,
      "loss": 3.5663,
      "step": 3850
    },
    {
      "epoch": 0.039215087890625,
      "grad_norm": 9.590949058532715,
      "learning_rate": 4.981984389242493e-06,
      "loss": 3.5181,
      "step": 3855
    },
    {
      "epoch": 0.039265950520833336,
      "grad_norm": 8.50133228302002,
      "learning_rate": 4.981936437540073e-06,
      "loss": 3.5247,
      "step": 3860
    },
    {
      "epoch": 0.039316813151041664,
      "grad_norm": 15.226415634155273,
      "learning_rate": 4.981888422337897e-06,
      "loss": 3.5207,
      "step": 3865
    },
    {
      "epoch": 0.03936767578125,
      "grad_norm": 15.351889610290527,
      "learning_rate": 4.981840343637194e-06,
      "loss": 3.6821,
      "step": 3870
    },
    {
      "epoch": 0.039418538411458336,
      "grad_norm": 11.84135627746582,
      "learning_rate": 4.981792201439195e-06,
      "loss": 3.4474,
      "step": 3875
    },
    {
      "epoch": 0.039469401041666664,
      "grad_norm": 19.106542587280273,
      "learning_rate": 4.9817439957451295e-06,
      "loss": 3.4201,
      "step": 3880
    },
    {
      "epoch": 0.039520263671875,
      "grad_norm": 14.762589454650879,
      "learning_rate": 4.981695726556233e-06,
      "loss": 3.6159,
      "step": 3885
    },
    {
      "epoch": 0.039571126302083336,
      "grad_norm": 10.755833625793457,
      "learning_rate": 4.98164739387374e-06,
      "loss": 3.6952,
      "step": 3890
    },
    {
      "epoch": 0.039621988932291664,
      "grad_norm": 7.459182262420654,
      "learning_rate": 4.9815989976988856e-06,
      "loss": 3.3008,
      "step": 3895
    },
    {
      "epoch": 0.0396728515625,
      "grad_norm": 16.29890251159668,
      "learning_rate": 4.98155053803291e-06,
      "loss": 3.7614,
      "step": 3900
    },
    {
      "epoch": 0.039723714192708336,
      "grad_norm": 15.238051414489746,
      "learning_rate": 4.981502014877051e-06,
      "loss": 3.5197,
      "step": 3905
    },
    {
      "epoch": 0.039774576822916664,
      "grad_norm": 16.749492645263672,
      "learning_rate": 4.981453428232551e-06,
      "loss": 4.1579,
      "step": 3910
    },
    {
      "epoch": 0.039825439453125,
      "grad_norm": 14.327499389648438,
      "learning_rate": 4.981404778100654e-06,
      "loss": 3.1206,
      "step": 3915
    },
    {
      "epoch": 0.039876302083333336,
      "grad_norm": 18.08525848388672,
      "learning_rate": 4.981356064482604e-06,
      "loss": 3.5667,
      "step": 3920
    },
    {
      "epoch": 0.039927164713541664,
      "grad_norm": 9.953446388244629,
      "learning_rate": 4.981307287379647e-06,
      "loss": 3.1772,
      "step": 3925
    },
    {
      "epoch": 0.03997802734375,
      "grad_norm": 12.199675559997559,
      "learning_rate": 4.9812584467930315e-06,
      "loss": 3.3682,
      "step": 3930
    },
    {
      "epoch": 0.040028889973958336,
      "grad_norm": 14.886872291564941,
      "learning_rate": 4.981209542724006e-06,
      "loss": 3.3351,
      "step": 3935
    },
    {
      "epoch": 0.040079752604166664,
      "grad_norm": 15.33847427368164,
      "learning_rate": 4.981160575173823e-06,
      "loss": 3.2065,
      "step": 3940
    },
    {
      "epoch": 0.040130615234375,
      "grad_norm": 10.76516056060791,
      "learning_rate": 4.981111544143735e-06,
      "loss": 3.6167,
      "step": 3945
    },
    {
      "epoch": 0.040181477864583336,
      "grad_norm": 10.835091590881348,
      "learning_rate": 4.981062449634996e-06,
      "loss": 3.3896,
      "step": 3950
    },
    {
      "epoch": 0.040232340494791664,
      "grad_norm": 8.555797576904297,
      "learning_rate": 4.981013291648861e-06,
      "loss": 3.3618,
      "step": 3955
    },
    {
      "epoch": 0.040283203125,
      "grad_norm": 20.06720733642578,
      "learning_rate": 4.980964070186591e-06,
      "loss": 3.1369,
      "step": 3960
    },
    {
      "epoch": 0.040334065755208336,
      "grad_norm": 10.057026863098145,
      "learning_rate": 4.9809147852494425e-06,
      "loss": 3.4979,
      "step": 3965
    },
    {
      "epoch": 0.040384928385416664,
      "grad_norm": 9.190557479858398,
      "learning_rate": 4.980865436838677e-06,
      "loss": 3.5271,
      "step": 3970
    },
    {
      "epoch": 0.040435791015625,
      "grad_norm": 13.96591854095459,
      "learning_rate": 4.9808160249555585e-06,
      "loss": 3.8185,
      "step": 3975
    },
    {
      "epoch": 0.040486653645833336,
      "grad_norm": 13.018309593200684,
      "learning_rate": 4.980766549601349e-06,
      "loss": 3.3418,
      "step": 3980
    },
    {
      "epoch": 0.040537516276041664,
      "grad_norm": 12.482734680175781,
      "learning_rate": 4.9807170107773155e-06,
      "loss": 3.7106,
      "step": 3985
    },
    {
      "epoch": 0.04058837890625,
      "grad_norm": 19.169965744018555,
      "learning_rate": 4.980667408484725e-06,
      "loss": 3.7609,
      "step": 3990
    },
    {
      "epoch": 0.040639241536458336,
      "grad_norm": 16.576343536376953,
      "learning_rate": 4.980617742724847e-06,
      "loss": 3.564,
      "step": 3995
    },
    {
      "epoch": 0.040690104166666664,
      "grad_norm": 13.268610954284668,
      "learning_rate": 4.980568013498952e-06,
      "loss": 3.5274,
      "step": 4000
    },
    {
      "epoch": 0.040740966796875,
      "grad_norm": 11.195540428161621,
      "learning_rate": 4.980518220808312e-06,
      "loss": 3.6116,
      "step": 4005
    },
    {
      "epoch": 0.040791829427083336,
      "grad_norm": 10.44593620300293,
      "learning_rate": 4.980468364654202e-06,
      "loss": 3.4021,
      "step": 4010
    },
    {
      "epoch": 0.040842692057291664,
      "grad_norm": 49.621131896972656,
      "learning_rate": 4.980418445037897e-06,
      "loss": 3.6859,
      "step": 4015
    },
    {
      "epoch": 0.0408935546875,
      "grad_norm": 14.462821006774902,
      "learning_rate": 4.980368461960673e-06,
      "loss": 3.4476,
      "step": 4020
    },
    {
      "epoch": 0.040944417317708336,
      "grad_norm": 10.35971736907959,
      "learning_rate": 4.98031841542381e-06,
      "loss": 3.6455,
      "step": 4025
    },
    {
      "epoch": 0.040995279947916664,
      "grad_norm": 16.127132415771484,
      "learning_rate": 4.980268305428589e-06,
      "loss": 3.4884,
      "step": 4030
    },
    {
      "epoch": 0.041046142578125,
      "grad_norm": 11.943824768066406,
      "learning_rate": 4.980218131976291e-06,
      "loss": 3.7037,
      "step": 4035
    },
    {
      "epoch": 0.041097005208333336,
      "grad_norm": 13.393162727355957,
      "learning_rate": 4.9801678950682e-06,
      "loss": 3.5249,
      "step": 4040
    },
    {
      "epoch": 0.041147867838541664,
      "grad_norm": 13.681143760681152,
      "learning_rate": 4.9801175947056005e-06,
      "loss": 3.4255,
      "step": 4045
    },
    {
      "epoch": 0.04119873046875,
      "grad_norm": 14.354728698730469,
      "learning_rate": 4.980067230889781e-06,
      "loss": 3.3633,
      "step": 4050
    },
    {
      "epoch": 0.041249593098958336,
      "grad_norm": 15.5615873336792,
      "learning_rate": 4.9800168036220295e-06,
      "loss": 3.5337,
      "step": 4055
    },
    {
      "epoch": 0.041300455729166664,
      "grad_norm": 8.801639556884766,
      "learning_rate": 4.9799663129036354e-06,
      "loss": 3.5424,
      "step": 4060
    },
    {
      "epoch": 0.041351318359375,
      "grad_norm": 11.887249946594238,
      "learning_rate": 4.9799157587358905e-06,
      "loss": 3.595,
      "step": 4065
    },
    {
      "epoch": 0.041402180989583336,
      "grad_norm": 10.93061637878418,
      "learning_rate": 4.979865141120089e-06,
      "loss": 3.2786,
      "step": 4070
    },
    {
      "epoch": 0.041453043619791664,
      "grad_norm": 13.5011625289917,
      "learning_rate": 4.979814460057527e-06,
      "loss": 3.3899,
      "step": 4075
    },
    {
      "epoch": 0.04150390625,
      "grad_norm": 13.714580535888672,
      "learning_rate": 4.979763715549498e-06,
      "loss": 3.3754,
      "step": 4080
    },
    {
      "epoch": 0.041554768880208336,
      "grad_norm": 8.235960960388184,
      "learning_rate": 4.9797129075973025e-06,
      "loss": 3.4213,
      "step": 4085
    },
    {
      "epoch": 0.041605631510416664,
      "grad_norm": 9.007630348205566,
      "learning_rate": 4.979662036202241e-06,
      "loss": 3.5357,
      "step": 4090
    },
    {
      "epoch": 0.041656494140625,
      "grad_norm": 11.23054313659668,
      "learning_rate": 4.979611101365613e-06,
      "loss": 3.7044,
      "step": 4095
    },
    {
      "epoch": 0.041707356770833336,
      "grad_norm": 12.494560241699219,
      "learning_rate": 4.979560103088723e-06,
      "loss": 3.5578,
      "step": 4100
    },
    {
      "epoch": 0.041758219401041664,
      "grad_norm": 15.440023422241211,
      "learning_rate": 4.979509041372876e-06,
      "loss": 3.3444,
      "step": 4105
    },
    {
      "epoch": 0.04180908203125,
      "grad_norm": 14.82787799835205,
      "learning_rate": 4.979457916219378e-06,
      "loss": 3.7682,
      "step": 4110
    },
    {
      "epoch": 0.041859944661458336,
      "grad_norm": 13.421795845031738,
      "learning_rate": 4.979406727629536e-06,
      "loss": 3.673,
      "step": 4115
    },
    {
      "epoch": 0.041910807291666664,
      "grad_norm": 19.039793014526367,
      "learning_rate": 4.979355475604661e-06,
      "loss": 3.5433,
      "step": 4120
    },
    {
      "epoch": 0.041961669921875,
      "grad_norm": 16.655475616455078,
      "learning_rate": 4.979304160146064e-06,
      "loss": 3.2191,
      "step": 4125
    },
    {
      "epoch": 0.042012532552083336,
      "grad_norm": 13.381031036376953,
      "learning_rate": 4.979252781255057e-06,
      "loss": 3.1412,
      "step": 4130
    },
    {
      "epoch": 0.042063395182291664,
      "grad_norm": 11.062989234924316,
      "learning_rate": 4.979201338932956e-06,
      "loss": 3.8812,
      "step": 4135
    },
    {
      "epoch": 0.0421142578125,
      "grad_norm": 11.04766845703125,
      "learning_rate": 4.979149833181076e-06,
      "loss": 3.1701,
      "step": 4140
    },
    {
      "epoch": 0.042165120442708336,
      "grad_norm": 12.870123863220215,
      "learning_rate": 4.979098264000736e-06,
      "loss": 3.4735,
      "step": 4145
    },
    {
      "epoch": 0.042215983072916664,
      "grad_norm": 11.865944862365723,
      "learning_rate": 4.979046631393253e-06,
      "loss": 3.666,
      "step": 4150
    },
    {
      "epoch": 0.042266845703125,
      "grad_norm": 10.82459831237793,
      "learning_rate": 4.97899493535995e-06,
      "loss": 3.6551,
      "step": 4155
    },
    {
      "epoch": 0.042317708333333336,
      "grad_norm": 12.600262641906738,
      "learning_rate": 4.97894317590215e-06,
      "loss": 3.8571,
      "step": 4160
    },
    {
      "epoch": 0.042368570963541664,
      "grad_norm": 9.6876220703125,
      "learning_rate": 4.978891353021176e-06,
      "loss": 3.5783,
      "step": 4165
    },
    {
      "epoch": 0.04241943359375,
      "grad_norm": 13.637721061706543,
      "learning_rate": 4.978839466718354e-06,
      "loss": 3.7054,
      "step": 4170
    },
    {
      "epoch": 0.042470296223958336,
      "grad_norm": 10.315239906311035,
      "learning_rate": 4.978787516995012e-06,
      "loss": 3.442,
      "step": 4175
    },
    {
      "epoch": 0.042521158854166664,
      "grad_norm": 12.303563117980957,
      "learning_rate": 4.9787355038524785e-06,
      "loss": 3.425,
      "step": 4180
    },
    {
      "epoch": 0.042572021484375,
      "grad_norm": 8.549363136291504,
      "learning_rate": 4.978683427292086e-06,
      "loss": 3.3095,
      "step": 4185
    },
    {
      "epoch": 0.042622884114583336,
      "grad_norm": 14.406485557556152,
      "learning_rate": 4.978631287315165e-06,
      "loss": 3.4881,
      "step": 4190
    },
    {
      "epoch": 0.042673746744791664,
      "grad_norm": 13.812122344970703,
      "learning_rate": 4.978579083923049e-06,
      "loss": 3.5369,
      "step": 4195
    },
    {
      "epoch": 0.042724609375,
      "grad_norm": 14.43535041809082,
      "learning_rate": 4.978526817117075e-06,
      "loss": 3.5955,
      "step": 4200
    },
    {
      "epoch": 0.042775472005208336,
      "grad_norm": 14.955559730529785,
      "learning_rate": 4.97847448689858e-06,
      "loss": 3.4783,
      "step": 4205
    },
    {
      "epoch": 0.042826334635416664,
      "grad_norm": 15.662449836730957,
      "learning_rate": 4.978422093268903e-06,
      "loss": 3.0456,
      "step": 4210
    },
    {
      "epoch": 0.042877197265625,
      "grad_norm": 21.45525360107422,
      "learning_rate": 4.978369636229383e-06,
      "loss": 3.3702,
      "step": 4215
    },
    {
      "epoch": 0.042928059895833336,
      "grad_norm": 11.133001327514648,
      "learning_rate": 4.978317115781365e-06,
      "loss": 3.7719,
      "step": 4220
    },
    {
      "epoch": 0.042978922526041664,
      "grad_norm": 14.71216869354248,
      "learning_rate": 4.97826453192619e-06,
      "loss": 3.2268,
      "step": 4225
    },
    {
      "epoch": 0.04302978515625,
      "grad_norm": 8.537089347839355,
      "learning_rate": 4.978211884665205e-06,
      "loss": 3.3107,
      "step": 4230
    },
    {
      "epoch": 0.043080647786458336,
      "grad_norm": 20.554950714111328,
      "learning_rate": 4.978159173999756e-06,
      "loss": 3.9013,
      "step": 4235
    },
    {
      "epoch": 0.043131510416666664,
      "grad_norm": 16.465560913085938,
      "learning_rate": 4.9781063999311914e-06,
      "loss": 3.5264,
      "step": 4240
    },
    {
      "epoch": 0.043182373046875,
      "grad_norm": 12.345917701721191,
      "learning_rate": 4.978053562460863e-06,
      "loss": 3.6222,
      "step": 4245
    },
    {
      "epoch": 0.043233235677083336,
      "grad_norm": 14.286460876464844,
      "learning_rate": 4.978000661590121e-06,
      "loss": 3.55,
      "step": 4250
    },
    {
      "epoch": 0.043284098307291664,
      "grad_norm": 16.970375061035156,
      "learning_rate": 4.97794769732032e-06,
      "loss": 3.7756,
      "step": 4255
    },
    {
      "epoch": 0.0433349609375,
      "grad_norm": 12.45829963684082,
      "learning_rate": 4.977894669652814e-06,
      "loss": 3.5234,
      "step": 4260
    },
    {
      "epoch": 0.043385823567708336,
      "grad_norm": 7.817399501800537,
      "learning_rate": 4.97784157858896e-06,
      "loss": 3.658,
      "step": 4265
    },
    {
      "epoch": 0.043436686197916664,
      "grad_norm": 16.19847869873047,
      "learning_rate": 4.9777884241301165e-06,
      "loss": 3.5021,
      "step": 4270
    },
    {
      "epoch": 0.043487548828125,
      "grad_norm": 8.416207313537598,
      "learning_rate": 4.977735206277644e-06,
      "loss": 3.3046,
      "step": 4275
    },
    {
      "epoch": 0.043538411458333336,
      "grad_norm": 13.35494613647461,
      "learning_rate": 4.977681925032902e-06,
      "loss": 3.3918,
      "step": 4280
    },
    {
      "epoch": 0.043589274088541664,
      "grad_norm": 10.832646369934082,
      "learning_rate": 4.977628580397257e-06,
      "loss": 3.4338,
      "step": 4285
    },
    {
      "epoch": 0.04364013671875,
      "grad_norm": 9.665356636047363,
      "learning_rate": 4.977575172372072e-06,
      "loss": 3.5508,
      "step": 4290
    },
    {
      "epoch": 0.043690999348958336,
      "grad_norm": 11.935736656188965,
      "learning_rate": 4.977521700958712e-06,
      "loss": 3.463,
      "step": 4295
    },
    {
      "epoch": 0.043741861979166664,
      "grad_norm": 16.61982536315918,
      "learning_rate": 4.977468166158548e-06,
      "loss": 3.7634,
      "step": 4300
    },
    {
      "epoch": 0.043792724609375,
      "grad_norm": 14.604238510131836,
      "learning_rate": 4.977414567972948e-06,
      "loss": 3.3686,
      "step": 4305
    },
    {
      "epoch": 0.043843587239583336,
      "grad_norm": 13.716455459594727,
      "learning_rate": 4.977360906403283e-06,
      "loss": 3.374,
      "step": 4310
    },
    {
      "epoch": 0.043894449869791664,
      "grad_norm": 11.446566581726074,
      "learning_rate": 4.977307181450926e-06,
      "loss": 3.3746,
      "step": 4315
    },
    {
      "epoch": 0.0439453125,
      "grad_norm": 9.788915634155273,
      "learning_rate": 4.977253393117253e-06,
      "loss": 3.5568,
      "step": 4320
    },
    {
      "epoch": 0.043996175130208336,
      "grad_norm": 203.79078674316406,
      "learning_rate": 4.977199541403638e-06,
      "loss": 3.7115,
      "step": 4325
    },
    {
      "epoch": 0.044047037760416664,
      "grad_norm": 14.938610076904297,
      "learning_rate": 4.97714562631146e-06,
      "loss": 3.6643,
      "step": 4330
    },
    {
      "epoch": 0.044097900390625,
      "grad_norm": 15.485106468200684,
      "learning_rate": 4.977091647842099e-06,
      "loss": 3.5269,
      "step": 4335
    },
    {
      "epoch": 0.044148763020833336,
      "grad_norm": 12.620315551757812,
      "learning_rate": 4.977037605996936e-06,
      "loss": 4.1156,
      "step": 4340
    },
    {
      "epoch": 0.044199625651041664,
      "grad_norm": 16.161775588989258,
      "learning_rate": 4.976983500777352e-06,
      "loss": 3.845,
      "step": 4345
    },
    {
      "epoch": 0.04425048828125,
      "grad_norm": 14.427170753479004,
      "learning_rate": 4.976929332184732e-06,
      "loss": 3.7344,
      "step": 4350
    },
    {
      "epoch": 0.044301350911458336,
      "grad_norm": 15.274232864379883,
      "learning_rate": 4.976875100220462e-06,
      "loss": 3.3062,
      "step": 4355
    },
    {
      "epoch": 0.044352213541666664,
      "grad_norm": 8.53735637664795,
      "learning_rate": 4.97682080488593e-06,
      "loss": 3.272,
      "step": 4360
    },
    {
      "epoch": 0.044403076171875,
      "grad_norm": 20.58081817626953,
      "learning_rate": 4.9767664461825246e-06,
      "loss": 3.3835,
      "step": 4365
    },
    {
      "epoch": 0.044453938802083336,
      "grad_norm": 8.867008209228516,
      "learning_rate": 4.976712024111637e-06,
      "loss": 3.36,
      "step": 4370
    },
    {
      "epoch": 0.044504801432291664,
      "grad_norm": 13.777779579162598,
      "learning_rate": 4.976657538674659e-06,
      "loss": 3.6504,
      "step": 4375
    },
    {
      "epoch": 0.0445556640625,
      "grad_norm": 13.533966064453125,
      "learning_rate": 4.9766029898729865e-06,
      "loss": 3.6923,
      "step": 4380
    },
    {
      "epoch": 0.044606526692708336,
      "grad_norm": 16.9124698638916,
      "learning_rate": 4.976548377708011e-06,
      "loss": 3.7102,
      "step": 4385
    },
    {
      "epoch": 0.044657389322916664,
      "grad_norm": 14.597479820251465,
      "learning_rate": 4.9764937021811345e-06,
      "loss": 3.6415,
      "step": 4390
    },
    {
      "epoch": 0.044708251953125,
      "grad_norm": 12.073945999145508,
      "learning_rate": 4.976438963293753e-06,
      "loss": 3.5683,
      "step": 4395
    },
    {
      "epoch": 0.044759114583333336,
      "grad_norm": 10.953951835632324,
      "learning_rate": 4.976384161047266e-06,
      "loss": 3.4114,
      "step": 4400
    },
    {
      "epoch": 0.044809977213541664,
      "grad_norm": 11.895225524902344,
      "learning_rate": 4.976329295443079e-06,
      "loss": 3.5981,
      "step": 4405
    },
    {
      "epoch": 0.04486083984375,
      "grad_norm": 13.530261039733887,
      "learning_rate": 4.976274366482593e-06,
      "loss": 3.416,
      "step": 4410
    },
    {
      "epoch": 0.044911702473958336,
      "grad_norm": 12.748217582702637,
      "learning_rate": 4.9762193741672145e-06,
      "loss": 3.6393,
      "step": 4415
    },
    {
      "epoch": 0.044962565104166664,
      "grad_norm": 34.55634307861328,
      "learning_rate": 4.976164318498351e-06,
      "loss": 3.6435,
      "step": 4420
    },
    {
      "epoch": 0.045013427734375,
      "grad_norm": 9.771318435668945,
      "learning_rate": 4.9761091994774095e-06,
      "loss": 3.8975,
      "step": 4425
    },
    {
      "epoch": 0.045064290364583336,
      "grad_norm": 10.606549263000488,
      "learning_rate": 4.976054017105801e-06,
      "loss": 3.8694,
      "step": 4430
    },
    {
      "epoch": 0.045115152994791664,
      "grad_norm": 18.156574249267578,
      "learning_rate": 4.975998771384938e-06,
      "loss": 3.8872,
      "step": 4435
    },
    {
      "epoch": 0.045166015625,
      "grad_norm": 12.91622257232666,
      "learning_rate": 4.9759434623162325e-06,
      "loss": 3.5453,
      "step": 4440
    },
    {
      "epoch": 0.045216878255208336,
      "grad_norm": 12.400276184082031,
      "learning_rate": 4.975888089901101e-06,
      "loss": 3.7277,
      "step": 4445
    },
    {
      "epoch": 0.045267740885416664,
      "grad_norm": 11.56432056427002,
      "learning_rate": 4.97583265414096e-06,
      "loss": 3.62,
      "step": 4450
    },
    {
      "epoch": 0.045318603515625,
      "grad_norm": 12.74065113067627,
      "learning_rate": 4.975777155037226e-06,
      "loss": 3.2927,
      "step": 4455
    },
    {
      "epoch": 0.045369466145833336,
      "grad_norm": 16.921592712402344,
      "learning_rate": 4.975721592591321e-06,
      "loss": 3.4098,
      "step": 4460
    },
    {
      "epoch": 0.045420328776041664,
      "grad_norm": 10.564414024353027,
      "learning_rate": 4.975665966804666e-06,
      "loss": 3.5746,
      "step": 4465
    },
    {
      "epoch": 0.04547119140625,
      "grad_norm": 11.815017700195312,
      "learning_rate": 4.9756102776786845e-06,
      "loss": 3.0929,
      "step": 4470
    },
    {
      "epoch": 0.045522054036458336,
      "grad_norm": 11.405243873596191,
      "learning_rate": 4.9755545252147995e-06,
      "loss": 3.8489,
      "step": 4475
    },
    {
      "epoch": 0.045572916666666664,
      "grad_norm": 14.713571548461914,
      "learning_rate": 4.97549870941444e-06,
      "loss": 3.7178,
      "step": 4480
    },
    {
      "epoch": 0.045623779296875,
      "grad_norm": 11.61874771118164,
      "learning_rate": 4.9754428302790325e-06,
      "loss": 3.3223,
      "step": 4485
    },
    {
      "epoch": 0.045674641927083336,
      "grad_norm": 8.922009468078613,
      "learning_rate": 4.975386887810007e-06,
      "loss": 3.7321,
      "step": 4490
    },
    {
      "epoch": 0.045725504557291664,
      "grad_norm": 8.890037536621094,
      "learning_rate": 4.975330882008794e-06,
      "loss": 3.4178,
      "step": 4495
    },
    {
      "epoch": 0.0457763671875,
      "grad_norm": 11.523167610168457,
      "learning_rate": 4.9752748128768275e-06,
      "loss": 3.8556,
      "step": 4500
    },
    {
      "epoch": 0.045827229817708336,
      "grad_norm": 15.356710433959961,
      "learning_rate": 4.975218680415541e-06,
      "loss": 3.4063,
      "step": 4505
    },
    {
      "epoch": 0.045878092447916664,
      "grad_norm": 12.003191947937012,
      "learning_rate": 4.9751624846263725e-06,
      "loss": 3.727,
      "step": 4510
    },
    {
      "epoch": 0.045928955078125,
      "grad_norm": 11.115232467651367,
      "learning_rate": 4.9751062255107575e-06,
      "loss": 3.3798,
      "step": 4515
    },
    {
      "epoch": 0.045979817708333336,
      "grad_norm": 12.786600112915039,
      "learning_rate": 4.975049903070137e-06,
      "loss": 3.8232,
      "step": 4520
    },
    {
      "epoch": 0.046030680338541664,
      "grad_norm": 9.778971672058105,
      "learning_rate": 4.974993517305952e-06,
      "loss": 3.5046,
      "step": 4525
    },
    {
      "epoch": 0.04608154296875,
      "grad_norm": 11.696513175964355,
      "learning_rate": 4.974937068219643e-06,
      "loss": 3.5063,
      "step": 4530
    },
    {
      "epoch": 0.046132405598958336,
      "grad_norm": 10.245550155639648,
      "learning_rate": 4.974880555812656e-06,
      "loss": 3.2886,
      "step": 4535
    },
    {
      "epoch": 0.046183268229166664,
      "grad_norm": 15.772186279296875,
      "learning_rate": 4.9748239800864375e-06,
      "loss": 3.482,
      "step": 4540
    },
    {
      "epoch": 0.046234130859375,
      "grad_norm": 11.759252548217773,
      "learning_rate": 4.974767341042433e-06,
      "loss": 3.663,
      "step": 4545
    },
    {
      "epoch": 0.046284993489583336,
      "grad_norm": 9.798601150512695,
      "learning_rate": 4.9747106386820934e-06,
      "loss": 3.3534,
      "step": 4550
    },
    {
      "epoch": 0.046335856119791664,
      "grad_norm": 13.721288681030273,
      "learning_rate": 4.9746538730068684e-06,
      "loss": 3.4893,
      "step": 4555
    },
    {
      "epoch": 0.04638671875,
      "grad_norm": 11.46217155456543,
      "learning_rate": 4.974597044018211e-06,
      "loss": 3.3706,
      "step": 4560
    },
    {
      "epoch": 0.046437581380208336,
      "grad_norm": 8.599750518798828,
      "learning_rate": 4.974540151717574e-06,
      "loss": 4.0034,
      "step": 4565
    },
    {
      "epoch": 0.046488444010416664,
      "grad_norm": 15.226737022399902,
      "learning_rate": 4.974483196106415e-06,
      "loss": 3.591,
      "step": 4570
    },
    {
      "epoch": 0.046539306640625,
      "grad_norm": 8.270947456359863,
      "learning_rate": 4.9744261771861894e-06,
      "loss": 3.3649,
      "step": 4575
    },
    {
      "epoch": 0.046590169270833336,
      "grad_norm": 13.826691627502441,
      "learning_rate": 4.974369094958356e-06,
      "loss": 3.7075,
      "step": 4580
    },
    {
      "epoch": 0.046641031901041664,
      "grad_norm": 15.241434097290039,
      "learning_rate": 4.974311949424376e-06,
      "loss": 3.9796,
      "step": 4585
    },
    {
      "epoch": 0.04669189453125,
      "grad_norm": 9.045734405517578,
      "learning_rate": 4.974254740585712e-06,
      "loss": 3.3453,
      "step": 4590
    },
    {
      "epoch": 0.046742757161458336,
      "grad_norm": 12.359314918518066,
      "learning_rate": 4.974197468443826e-06,
      "loss": 3.5554,
      "step": 4595
    },
    {
      "epoch": 0.046793619791666664,
      "grad_norm": 16.416982650756836,
      "learning_rate": 4.974140133000184e-06,
      "loss": 3.4799,
      "step": 4600
    },
    {
      "epoch": 0.046844482421875,
      "grad_norm": 12.722060203552246,
      "learning_rate": 4.974082734256254e-06,
      "loss": 3.8829,
      "step": 4605
    },
    {
      "epoch": 0.046895345052083336,
      "grad_norm": 14.915850639343262,
      "learning_rate": 4.9740252722135035e-06,
      "loss": 3.0652,
      "step": 4610
    },
    {
      "epoch": 0.046946207682291664,
      "grad_norm": 15.305362701416016,
      "learning_rate": 4.973967746873403e-06,
      "loss": 3.2942,
      "step": 4615
    },
    {
      "epoch": 0.0469970703125,
      "grad_norm": 18.63602638244629,
      "learning_rate": 4.973910158237423e-06,
      "loss": 3.1882,
      "step": 4620
    },
    {
      "epoch": 0.047047932942708336,
      "grad_norm": 15.647112846374512,
      "learning_rate": 4.973852506307039e-06,
      "loss": 3.2307,
      "step": 4625
    },
    {
      "epoch": 0.047098795572916664,
      "grad_norm": 13.300541877746582,
      "learning_rate": 4.973794791083725e-06,
      "loss": 3.6601,
      "step": 4630
    },
    {
      "epoch": 0.047149658203125,
      "grad_norm": 12.027398109436035,
      "learning_rate": 4.9737370125689575e-06,
      "loss": 3.5073,
      "step": 4635
    },
    {
      "epoch": 0.047200520833333336,
      "grad_norm": 9.269347190856934,
      "learning_rate": 4.973679170764214e-06,
      "loss": 4.1268,
      "step": 4640
    },
    {
      "epoch": 0.047251383463541664,
      "grad_norm": 9.16490650177002,
      "learning_rate": 4.973621265670976e-06,
      "loss": 3.2929,
      "step": 4645
    },
    {
      "epoch": 0.04730224609375,
      "grad_norm": 11.5504789352417,
      "learning_rate": 4.973563297290724e-06,
      "loss": 3.369,
      "step": 4650
    },
    {
      "epoch": 0.047353108723958336,
      "grad_norm": 15.80301284790039,
      "learning_rate": 4.973505265624942e-06,
      "loss": 3.4352,
      "step": 4655
    },
    {
      "epoch": 0.047403971354166664,
      "grad_norm": 9.65170669555664,
      "learning_rate": 4.9734471706751135e-06,
      "loss": 3.8803,
      "step": 4660
    },
    {
      "epoch": 0.047454833984375,
      "grad_norm": 11.459211349487305,
      "learning_rate": 4.9733890124427255e-06,
      "loss": 3.5879,
      "step": 4665
    },
    {
      "epoch": 0.047505696614583336,
      "grad_norm": 16.0178279876709,
      "learning_rate": 4.973330790929266e-06,
      "loss": 3.6492,
      "step": 4670
    },
    {
      "epoch": 0.047556559244791664,
      "grad_norm": 16.536197662353516,
      "learning_rate": 4.973272506136224e-06,
      "loss": 3.2745,
      "step": 4675
    },
    {
      "epoch": 0.047607421875,
      "grad_norm": 10.868372917175293,
      "learning_rate": 4.973214158065092e-06,
      "loss": 3.4472,
      "step": 4680
    },
    {
      "epoch": 0.047658284505208336,
      "grad_norm": 14.260408401489258,
      "learning_rate": 4.973155746717361e-06,
      "loss": 3.4165,
      "step": 4685
    },
    {
      "epoch": 0.047709147135416664,
      "grad_norm": 10.171429634094238,
      "learning_rate": 4.973097272094527e-06,
      "loss": 3.209,
      "step": 4690
    },
    {
      "epoch": 0.047760009765625,
      "grad_norm": 13.096430778503418,
      "learning_rate": 4.973038734198086e-06,
      "loss": 3.4693,
      "step": 4695
    },
    {
      "epoch": 0.047810872395833336,
      "grad_norm": 15.2460355758667,
      "learning_rate": 4.972980133029534e-06,
      "loss": 3.4677,
      "step": 4700
    },
    {
      "epoch": 0.047861735026041664,
      "grad_norm": 13.545559883117676,
      "learning_rate": 4.9729214685903725e-06,
      "loss": 3.4694,
      "step": 4705
    },
    {
      "epoch": 0.04791259765625,
      "grad_norm": 17.726600646972656,
      "learning_rate": 4.9728627408821e-06,
      "loss": 3.5932,
      "step": 4710
    },
    {
      "epoch": 0.047963460286458336,
      "grad_norm": 13.018586158752441,
      "learning_rate": 4.972803949906222e-06,
      "loss": 3.4267,
      "step": 4715
    },
    {
      "epoch": 0.048014322916666664,
      "grad_norm": 15.820887565612793,
      "learning_rate": 4.9727450956642395e-06,
      "loss": 3.3878,
      "step": 4720
    },
    {
      "epoch": 0.048065185546875,
      "grad_norm": 10.09266471862793,
      "learning_rate": 4.972686178157661e-06,
      "loss": 3.7088,
      "step": 4725
    },
    {
      "epoch": 0.048116048177083336,
      "grad_norm": 14.535094261169434,
      "learning_rate": 4.972627197387993e-06,
      "loss": 3.492,
      "step": 4730
    },
    {
      "epoch": 0.048166910807291664,
      "grad_norm": 16.890783309936523,
      "learning_rate": 4.972568153356744e-06,
      "loss": 3.3656,
      "step": 4735
    },
    {
      "epoch": 0.0482177734375,
      "grad_norm": 8.608535766601562,
      "learning_rate": 4.972509046065423e-06,
      "loss": 3.4594,
      "step": 4740
    },
    {
      "epoch": 0.048268636067708336,
      "grad_norm": 12.446272850036621,
      "learning_rate": 4.9724498755155455e-06,
      "loss": 3.3436,
      "step": 4745
    },
    {
      "epoch": 0.048319498697916664,
      "grad_norm": 11.247904777526855,
      "learning_rate": 4.972390641708625e-06,
      "loss": 3.7488,
      "step": 4750
    },
    {
      "epoch": 0.048370361328125,
      "grad_norm": 10.85303783416748,
      "learning_rate": 4.972331344646175e-06,
      "loss": 3.4678,
      "step": 4755
    },
    {
      "epoch": 0.048421223958333336,
      "grad_norm": 12.21154499053955,
      "learning_rate": 4.972271984329713e-06,
      "loss": 3.4773,
      "step": 4760
    },
    {
      "epoch": 0.048472086588541664,
      "grad_norm": 12.006597518920898,
      "learning_rate": 4.9722125607607595e-06,
      "loss": 3.5087,
      "step": 4765
    },
    {
      "epoch": 0.04852294921875,
      "grad_norm": 9.981345176696777,
      "learning_rate": 4.972153073940833e-06,
      "loss": 3.5338,
      "step": 4770
    },
    {
      "epoch": 0.048573811848958336,
      "grad_norm": 13.28870964050293,
      "learning_rate": 4.972093523871456e-06,
      "loss": 3.2124,
      "step": 4775
    },
    {
      "epoch": 0.048624674479166664,
      "grad_norm": 11.399847984313965,
      "learning_rate": 4.972033910554151e-06,
      "loss": 3.251,
      "step": 4780
    },
    {
      "epoch": 0.048675537109375,
      "grad_norm": 11.87983226776123,
      "learning_rate": 4.971974233990447e-06,
      "loss": 3.2554,
      "step": 4785
    },
    {
      "epoch": 0.048726399739583336,
      "grad_norm": 10.418863296508789,
      "learning_rate": 4.971914494181866e-06,
      "loss": 3.8199,
      "step": 4790
    },
    {
      "epoch": 0.048777262369791664,
      "grad_norm": 13.546000480651855,
      "learning_rate": 4.971854691129939e-06,
      "loss": 3.6464,
      "step": 4795
    },
    {
      "epoch": 0.048828125,
      "grad_norm": 16.547273635864258,
      "learning_rate": 4.9717948248361954e-06,
      "loss": 3.3086,
      "step": 4800
    },
    {
      "epoch": 0.048878987630208336,
      "grad_norm": 10.11819839477539,
      "learning_rate": 4.971734895302168e-06,
      "loss": 3.5591,
      "step": 4805
    },
    {
      "epoch": 0.048929850260416664,
      "grad_norm": 11.45274829864502,
      "learning_rate": 4.971674902529389e-06,
      "loss": 3.666,
      "step": 4810
    },
    {
      "epoch": 0.048980712890625,
      "grad_norm": 12.099120140075684,
      "learning_rate": 4.971614846519393e-06,
      "loss": 3.6187,
      "step": 4815
    },
    {
      "epoch": 0.049031575520833336,
      "grad_norm": 9.340841293334961,
      "learning_rate": 4.971554727273718e-06,
      "loss": 3.4518,
      "step": 4820
    },
    {
      "epoch": 0.049082438151041664,
      "grad_norm": 14.398073196411133,
      "learning_rate": 4.9714945447939e-06,
      "loss": 3.514,
      "step": 4825
    },
    {
      "epoch": 0.04913330078125,
      "grad_norm": 10.84057331085205,
      "learning_rate": 4.97143429908148e-06,
      "loss": 4.0011,
      "step": 4830
    },
    {
      "epoch": 0.049184163411458336,
      "grad_norm": 14.066889762878418,
      "learning_rate": 4.971373990137999e-06,
      "loss": 3.4912,
      "step": 4835
    },
    {
      "epoch": 0.049235026041666664,
      "grad_norm": 13.424138069152832,
      "learning_rate": 4.971313617965001e-06,
      "loss": 3.4337,
      "step": 4840
    },
    {
      "epoch": 0.049285888671875,
      "grad_norm": 15.70202922821045,
      "learning_rate": 4.971253182564029e-06,
      "loss": 3.5523,
      "step": 4845
    },
    {
      "epoch": 0.049336751302083336,
      "grad_norm": 15.583520889282227,
      "learning_rate": 4.971192683936631e-06,
      "loss": 3.548,
      "step": 4850
    },
    {
      "epoch": 0.049387613932291664,
      "grad_norm": 16.691743850708008,
      "learning_rate": 4.9711321220843535e-06,
      "loss": 3.3919,
      "step": 4855
    },
    {
      "epoch": 0.0494384765625,
      "grad_norm": 15.311267852783203,
      "learning_rate": 4.971071497008746e-06,
      "loss": 3.275,
      "step": 4860
    },
    {
      "epoch": 0.049489339192708336,
      "grad_norm": 11.667826652526855,
      "learning_rate": 4.971010808711361e-06,
      "loss": 3.3042,
      "step": 4865
    },
    {
      "epoch": 0.049540201822916664,
      "grad_norm": 12.931231498718262,
      "learning_rate": 4.9709500571937485e-06,
      "loss": 3.6733,
      "step": 4870
    },
    {
      "epoch": 0.049591064453125,
      "grad_norm": 14.819002151489258,
      "learning_rate": 4.970889242457466e-06,
      "loss": 3.3424,
      "step": 4875
    },
    {
      "epoch": 0.049641927083333336,
      "grad_norm": 11.588568687438965,
      "learning_rate": 4.9708283645040675e-06,
      "loss": 3.2318,
      "step": 4880
    },
    {
      "epoch": 0.049692789713541664,
      "grad_norm": 14.38076400756836,
      "learning_rate": 4.970767423335111e-06,
      "loss": 3.5042,
      "step": 4885
    },
    {
      "epoch": 0.04974365234375,
      "grad_norm": 11.927538871765137,
      "learning_rate": 4.970706418952155e-06,
      "loss": 3.4414,
      "step": 4890
    },
    {
      "epoch": 0.049794514973958336,
      "grad_norm": 17.747844696044922,
      "learning_rate": 4.970645351356761e-06,
      "loss": 3.9686,
      "step": 4895
    },
    {
      "epoch": 0.049845377604166664,
      "grad_norm": 13.796122550964355,
      "learning_rate": 4.970584220550492e-06,
      "loss": 4.094,
      "step": 4900
    },
    {
      "epoch": 0.049896240234375,
      "grad_norm": 13.095043182373047,
      "learning_rate": 4.97052302653491e-06,
      "loss": 3.7906,
      "step": 4905
    },
    {
      "epoch": 0.049947102864583336,
      "grad_norm": 12.416662216186523,
      "learning_rate": 4.970461769311583e-06,
      "loss": 3.4955,
      "step": 4910
    },
    {
      "epoch": 0.049997965494791664,
      "grad_norm": 12.235672950744629,
      "learning_rate": 4.970400448882078e-06,
      "loss": 3.0274,
      "step": 4915
    },
    {
      "epoch": 0.050048828125,
      "grad_norm": 12.012406349182129,
      "learning_rate": 4.9703390652479615e-06,
      "loss": 3.3271,
      "step": 4920
    },
    {
      "epoch": 0.050099690755208336,
      "grad_norm": 11.558436393737793,
      "learning_rate": 4.970277618410806e-06,
      "loss": 3.4439,
      "step": 4925
    },
    {
      "epoch": 0.050150553385416664,
      "grad_norm": 10.99045467376709,
      "learning_rate": 4.970216108372184e-06,
      "loss": 3.4762,
      "step": 4930
    },
    {
      "epoch": 0.050201416015625,
      "grad_norm": 14.257964134216309,
      "learning_rate": 4.970154535133667e-06,
      "loss": 4.1728,
      "step": 4935
    },
    {
      "epoch": 0.050252278645833336,
      "grad_norm": 14.24341106414795,
      "learning_rate": 4.970092898696832e-06,
      "loss": 3.3222,
      "step": 4940
    },
    {
      "epoch": 0.050303141276041664,
      "grad_norm": 17.90021324157715,
      "learning_rate": 4.9700311990632565e-06,
      "loss": 3.2803,
      "step": 4945
    },
    {
      "epoch": 0.05035400390625,
      "grad_norm": 13.78532600402832,
      "learning_rate": 4.9699694362345175e-06,
      "loss": 3.6847,
      "step": 4950
    },
    {
      "epoch": 0.050404866536458336,
      "grad_norm": 9.49715518951416,
      "learning_rate": 4.969907610212197e-06,
      "loss": 3.4094,
      "step": 4955
    },
    {
      "epoch": 0.050455729166666664,
      "grad_norm": 16.413652420043945,
      "learning_rate": 4.969845720997874e-06,
      "loss": 3.4125,
      "step": 4960
    },
    {
      "epoch": 0.050506591796875,
      "grad_norm": 11.681251525878906,
      "learning_rate": 4.969783768593135e-06,
      "loss": 3.7634,
      "step": 4965
    },
    {
      "epoch": 0.050557454427083336,
      "grad_norm": 12.140727043151855,
      "learning_rate": 4.969721752999563e-06,
      "loss": 3.3175,
      "step": 4970
    },
    {
      "epoch": 0.050608317057291664,
      "grad_norm": 17.70606231689453,
      "learning_rate": 4.9696596742187455e-06,
      "loss": 3.7512,
      "step": 4975
    },
    {
      "epoch": 0.0506591796875,
      "grad_norm": 14.161558151245117,
      "learning_rate": 4.969597532252271e-06,
      "loss": 3.2145,
      "step": 4980
    },
    {
      "epoch": 0.050710042317708336,
      "grad_norm": 9.752479553222656,
      "learning_rate": 4.969535327101729e-06,
      "loss": 3.3454,
      "step": 4985
    },
    {
      "epoch": 0.050760904947916664,
      "grad_norm": 10.96493911743164,
      "learning_rate": 4.96947305876871e-06,
      "loss": 3.493,
      "step": 4990
    },
    {
      "epoch": 0.050811767578125,
      "grad_norm": 10.895770072937012,
      "learning_rate": 4.969410727254809e-06,
      "loss": 3.1966,
      "step": 4995
    },
    {
      "epoch": 0.050862630208333336,
      "grad_norm": 14.102956771850586,
      "learning_rate": 4.96934833256162e-06,
      "loss": 3.4336,
      "step": 5000
    },
    {
      "epoch": 0.050913492838541664,
      "grad_norm": 10.122415542602539,
      "learning_rate": 4.9692858746907395e-06,
      "loss": 3.275,
      "step": 5005
    },
    {
      "epoch": 0.05096435546875,
      "grad_norm": 16.442476272583008,
      "learning_rate": 4.969223353643764e-06,
      "loss": 3.7156,
      "step": 5010
    },
    {
      "epoch": 0.051015218098958336,
      "grad_norm": 12.107405662536621,
      "learning_rate": 4.969160769422294e-06,
      "loss": 3.645,
      "step": 5015
    },
    {
      "epoch": 0.051066080729166664,
      "grad_norm": 18.26190948486328,
      "learning_rate": 4.969098122027932e-06,
      "loss": 3.5743,
      "step": 5020
    },
    {
      "epoch": 0.051116943359375,
      "grad_norm": 11.404193878173828,
      "learning_rate": 4.96903541146228e-06,
      "loss": 3.2735,
      "step": 5025
    },
    {
      "epoch": 0.051167805989583336,
      "grad_norm": 14.428089141845703,
      "learning_rate": 4.968972637726942e-06,
      "loss": 3.7864,
      "step": 5030
    },
    {
      "epoch": 0.051218668619791664,
      "grad_norm": 10.478737831115723,
      "learning_rate": 4.9689098008235235e-06,
      "loss": 3.8017,
      "step": 5035
    },
    {
      "epoch": 0.05126953125,
      "grad_norm": 15.701272010803223,
      "learning_rate": 4.968846900753634e-06,
      "loss": 3.4754,
      "step": 5040
    },
    {
      "epoch": 0.051320393880208336,
      "grad_norm": 11.944446563720703,
      "learning_rate": 4.968783937518882e-06,
      "loss": 3.5693,
      "step": 5045
    },
    {
      "epoch": 0.051371256510416664,
      "grad_norm": 9.105219841003418,
      "learning_rate": 4.968720911120876e-06,
      "loss": 3.3752,
      "step": 5050
    },
    {
      "epoch": 0.051422119140625,
      "grad_norm": 17.28725814819336,
      "learning_rate": 4.968657821561233e-06,
      "loss": 3.542,
      "step": 5055
    },
    {
      "epoch": 0.051472981770833336,
      "grad_norm": 9.695812225341797,
      "learning_rate": 4.9685946688415635e-06,
      "loss": 3.3903,
      "step": 5060
    },
    {
      "epoch": 0.051523844401041664,
      "grad_norm": 11.877903938293457,
      "learning_rate": 4.968531452963485e-06,
      "loss": 3.8497,
      "step": 5065
    },
    {
      "epoch": 0.05157470703125,
      "grad_norm": 14.348855018615723,
      "learning_rate": 4.968468173928614e-06,
      "loss": 3.2782,
      "step": 5070
    },
    {
      "epoch": 0.051625569661458336,
      "grad_norm": 14.756049156188965,
      "learning_rate": 4.96840483173857e-06,
      "loss": 4.0242,
      "step": 5075
    },
    {
      "epoch": 0.051676432291666664,
      "grad_norm": 14.059425354003906,
      "learning_rate": 4.968341426394974e-06,
      "loss": 3.2894,
      "step": 5080
    },
    {
      "epoch": 0.051727294921875,
      "grad_norm": 11.158661842346191,
      "learning_rate": 4.968277957899446e-06,
      "loss": 3.6611,
      "step": 5085
    },
    {
      "epoch": 0.051778157552083336,
      "grad_norm": 18.206151962280273,
      "learning_rate": 4.968214426253613e-06,
      "loss": 3.4665,
      "step": 5090
    },
    {
      "epoch": 0.051829020182291664,
      "grad_norm": 8.762948989868164,
      "learning_rate": 4.968150831459099e-06,
      "loss": 3.4422,
      "step": 5095
    },
    {
      "epoch": 0.0518798828125,
      "grad_norm": 14.994571685791016,
      "learning_rate": 4.968087173517531e-06,
      "loss": 3.602,
      "step": 5100
    },
    {
      "epoch": 0.051930745442708336,
      "grad_norm": 12.887874603271484,
      "learning_rate": 4.968023452430537e-06,
      "loss": 3.12,
      "step": 5105
    },
    {
      "epoch": 0.051981608072916664,
      "grad_norm": 15.499711990356445,
      "learning_rate": 4.967959668199748e-06,
      "loss": 3.7486,
      "step": 5110
    },
    {
      "epoch": 0.052032470703125,
      "grad_norm": 12.390300750732422,
      "learning_rate": 4.967895820826796e-06,
      "loss": 3.4817,
      "step": 5115
    },
    {
      "epoch": 0.052083333333333336,
      "grad_norm": 13.03585433959961,
      "learning_rate": 4.967831910313314e-06,
      "loss": 3.1236,
      "step": 5120
    },
    {
      "epoch": 0.052134195963541664,
      "grad_norm": 11.844696998596191,
      "learning_rate": 4.967767936660939e-06,
      "loss": 3.4756,
      "step": 5125
    },
    {
      "epoch": 0.05218505859375,
      "grad_norm": 12.35226821899414,
      "learning_rate": 4.967703899871304e-06,
      "loss": 3.481,
      "step": 5130
    },
    {
      "epoch": 0.052235921223958336,
      "grad_norm": 10.799506187438965,
      "learning_rate": 4.967639799946052e-06,
      "loss": 3.7089,
      "step": 5135
    },
    {
      "epoch": 0.052286783854166664,
      "grad_norm": 11.297710418701172,
      "learning_rate": 4.967575636886819e-06,
      "loss": 3.5326,
      "step": 5140
    },
    {
      "epoch": 0.052337646484375,
      "grad_norm": 9.424179077148438,
      "learning_rate": 4.967511410695249e-06,
      "loss": 3.7654,
      "step": 5145
    },
    {
      "epoch": 0.052388509114583336,
      "grad_norm": 11.021271705627441,
      "learning_rate": 4.9674471213729836e-06,
      "loss": 3.278,
      "step": 5150
    },
    {
      "epoch": 0.052439371744791664,
      "grad_norm": 12.35921573638916,
      "learning_rate": 4.9673827689216695e-06,
      "loss": 3.4304,
      "step": 5155
    },
    {
      "epoch": 0.052490234375,
      "grad_norm": 12.476812362670898,
      "learning_rate": 4.967318353342952e-06,
      "loss": 3.3936,
      "step": 5160
    },
    {
      "epoch": 0.052541097005208336,
      "grad_norm": 10.683613777160645,
      "learning_rate": 4.967253874638478e-06,
      "loss": 3.1559,
      "step": 5165
    },
    {
      "epoch": 0.052591959635416664,
      "grad_norm": 13.43266773223877,
      "learning_rate": 4.967189332809899e-06,
      "loss": 3.8937,
      "step": 5170
    },
    {
      "epoch": 0.052642822265625,
      "grad_norm": 17.731548309326172,
      "learning_rate": 4.967124727858867e-06,
      "loss": 3.4165,
      "step": 5175
    },
    {
      "epoch": 0.052693684895833336,
      "grad_norm": 15.984275817871094,
      "learning_rate": 4.967060059787032e-06,
      "loss": 3.6395,
      "step": 5180
    },
    {
      "epoch": 0.052744547526041664,
      "grad_norm": 18.11372184753418,
      "learning_rate": 4.96699532859605e-06,
      "loss": 3.495,
      "step": 5185
    },
    {
      "epoch": 0.05279541015625,
      "grad_norm": 13.443792343139648,
      "learning_rate": 4.9669305342875785e-06,
      "loss": 4.1017,
      "step": 5190
    },
    {
      "epoch": 0.052846272786458336,
      "grad_norm": 16.402788162231445,
      "learning_rate": 4.9668656768632725e-06,
      "loss": 3.4839,
      "step": 5195
    },
    {
      "epoch": 0.052897135416666664,
      "grad_norm": 14.646184921264648,
      "learning_rate": 4.966800756324794e-06,
      "loss": 3.7131,
      "step": 5200
    },
    {
      "epoch": 0.052947998046875,
      "grad_norm": 16.304649353027344,
      "learning_rate": 4.966735772673803e-06,
      "loss": 3.4049,
      "step": 5205
    },
    {
      "epoch": 0.052998860677083336,
      "grad_norm": 17.611019134521484,
      "learning_rate": 4.966670725911962e-06,
      "loss": 3.2992,
      "step": 5210
    },
    {
      "epoch": 0.053049723307291664,
      "grad_norm": 9.94887638092041,
      "learning_rate": 4.966605616040935e-06,
      "loss": 3.498,
      "step": 5215
    },
    {
      "epoch": 0.0531005859375,
      "grad_norm": 14.684574127197266,
      "learning_rate": 4.9665404430623874e-06,
      "loss": 3.6487,
      "step": 5220
    },
    {
      "epoch": 0.053151448567708336,
      "grad_norm": 14.74726676940918,
      "learning_rate": 4.9664752069779875e-06,
      "loss": 3.3533,
      "step": 5225
    },
    {
      "epoch": 0.053202311197916664,
      "grad_norm": 16.72749137878418,
      "learning_rate": 4.966409907789403e-06,
      "loss": 3.203,
      "step": 5230
    },
    {
      "epoch": 0.053253173828125,
      "grad_norm": 12.44540786743164,
      "learning_rate": 4.966344545498307e-06,
      "loss": 3.4401,
      "step": 5235
    },
    {
      "epoch": 0.053304036458333336,
      "grad_norm": 10.099539756774902,
      "learning_rate": 4.96627912010637e-06,
      "loss": 3.5915,
      "step": 5240
    },
    {
      "epoch": 0.053354899088541664,
      "grad_norm": 11.097341537475586,
      "learning_rate": 4.966213631615266e-06,
      "loss": 3.9558,
      "step": 5245
    },
    {
      "epoch": 0.05340576171875,
      "grad_norm": 13.878585815429688,
      "learning_rate": 4.966148080026671e-06,
      "loss": 3.4617,
      "step": 5250
    },
    {
      "epoch": 0.053456624348958336,
      "grad_norm": 10.348923683166504,
      "learning_rate": 4.966082465342263e-06,
      "loss": 3.6141,
      "step": 5255
    },
    {
      "epoch": 0.053507486979166664,
      "grad_norm": 9.475302696228027,
      "learning_rate": 4.966016787563719e-06,
      "loss": 3.2901,
      "step": 5260
    },
    {
      "epoch": 0.053558349609375,
      "grad_norm": 16.38394546508789,
      "learning_rate": 4.965951046692719e-06,
      "loss": 3.1904,
      "step": 5265
    },
    {
      "epoch": 0.053609212239583336,
      "grad_norm": 14.318309783935547,
      "learning_rate": 4.965885242730947e-06,
      "loss": 3.3258,
      "step": 5270
    },
    {
      "epoch": 0.053660074869791664,
      "grad_norm": 8.593515396118164,
      "learning_rate": 4.965819375680085e-06,
      "loss": 3.341,
      "step": 5275
    },
    {
      "epoch": 0.0537109375,
      "grad_norm": 14.667823791503906,
      "learning_rate": 4.9657534455418186e-06,
      "loss": 3.5297,
      "step": 5280
    },
    {
      "epoch": 0.053761800130208336,
      "grad_norm": 15.408576965332031,
      "learning_rate": 4.965687452317836e-06,
      "loss": 3.3244,
      "step": 5285
    },
    {
      "epoch": 0.053812662760416664,
      "grad_norm": 15.13183307647705,
      "learning_rate": 4.9656213960098235e-06,
      "loss": 3.3597,
      "step": 5290
    },
    {
      "epoch": 0.053863525390625,
      "grad_norm": 10.050660133361816,
      "learning_rate": 4.965555276619471e-06,
      "loss": 3.2173,
      "step": 5295
    },
    {
      "epoch": 0.053914388020833336,
      "grad_norm": 11.996289253234863,
      "learning_rate": 4.965489094148473e-06,
      "loss": 3.258,
      "step": 5300
    },
    {
      "epoch": 0.053965250651041664,
      "grad_norm": 13.994873046875,
      "learning_rate": 4.965422848598519e-06,
      "loss": 3.9475,
      "step": 5305
    },
    {
      "epoch": 0.05401611328125,
      "grad_norm": 13.173226356506348,
      "learning_rate": 4.965356539971307e-06,
      "loss": 3.5109,
      "step": 5310
    },
    {
      "epoch": 0.054066975911458336,
      "grad_norm": 11.68813419342041,
      "learning_rate": 4.965290168268532e-06,
      "loss": 3.7385,
      "step": 5315
    },
    {
      "epoch": 0.054117838541666664,
      "grad_norm": 13.431297302246094,
      "learning_rate": 4.965223733491893e-06,
      "loss": 3.4336,
      "step": 5320
    },
    {
      "epoch": 0.054168701171875,
      "grad_norm": 14.174638748168945,
      "learning_rate": 4.965157235643088e-06,
      "loss": 3.2346,
      "step": 5325
    },
    {
      "epoch": 0.054219563802083336,
      "grad_norm": 14.978144645690918,
      "learning_rate": 4.96509067472382e-06,
      "loss": 3.6034,
      "step": 5330
    },
    {
      "epoch": 0.054270426432291664,
      "grad_norm": 11.926872253417969,
      "learning_rate": 4.965024050735792e-06,
      "loss": 3.4996,
      "step": 5335
    },
    {
      "epoch": 0.0543212890625,
      "grad_norm": 9.099247932434082,
      "learning_rate": 4.9649573636807065e-06,
      "loss": 3.1218,
      "step": 5340
    },
    {
      "epoch": 0.054372151692708336,
      "grad_norm": 17.655933380126953,
      "learning_rate": 4.964890613560272e-06,
      "loss": 3.6185,
      "step": 5345
    },
    {
      "epoch": 0.054423014322916664,
      "grad_norm": 10.588809967041016,
      "learning_rate": 4.964823800376195e-06,
      "loss": 3.5396,
      "step": 5350
    },
    {
      "epoch": 0.054473876953125,
      "grad_norm": 12.350024223327637,
      "learning_rate": 4.964756924130186e-06,
      "loss": 3.7472,
      "step": 5355
    },
    {
      "epoch": 0.054524739583333336,
      "grad_norm": 13.325206756591797,
      "learning_rate": 4.964689984823955e-06,
      "loss": 3.1619,
      "step": 5360
    },
    {
      "epoch": 0.054575602213541664,
      "grad_norm": 12.419147491455078,
      "learning_rate": 4.964622982459214e-06,
      "loss": 3.3821,
      "step": 5365
    },
    {
      "epoch": 0.05462646484375,
      "grad_norm": 11.1918306350708,
      "learning_rate": 4.964555917037679e-06,
      "loss": 3.2978,
      "step": 5370
    },
    {
      "epoch": 0.054677327473958336,
      "grad_norm": 8.211668014526367,
      "learning_rate": 4.964488788561066e-06,
      "loss": 3.1058,
      "step": 5375
    },
    {
      "epoch": 0.054728190104166664,
      "grad_norm": 14.706374168395996,
      "learning_rate": 4.96442159703109e-06,
      "loss": 3.5759,
      "step": 5380
    },
    {
      "epoch": 0.054779052734375,
      "grad_norm": 12.363343238830566,
      "learning_rate": 4.964354342449472e-06,
      "loss": 3.3521,
      "step": 5385
    },
    {
      "epoch": 0.054829915364583336,
      "grad_norm": 11.750041007995605,
      "learning_rate": 4.964287024817933e-06,
      "loss": 3.7942,
      "step": 5390
    },
    {
      "epoch": 0.054880777994791664,
      "grad_norm": 15.14163589477539,
      "learning_rate": 4.964219644138194e-06,
      "loss": 3.7889,
      "step": 5395
    },
    {
      "epoch": 0.054931640625,
      "grad_norm": 14.809743881225586,
      "learning_rate": 4.964152200411979e-06,
      "loss": 3.1966,
      "step": 5400
    },
    {
      "epoch": 0.054982503255208336,
      "grad_norm": 10.114827156066895,
      "learning_rate": 4.964084693641014e-06,
      "loss": 3.5248,
      "step": 5405
    },
    {
      "epoch": 0.055033365885416664,
      "grad_norm": 12.720572471618652,
      "learning_rate": 4.964017123827027e-06,
      "loss": 3.1197,
      "step": 5410
    },
    {
      "epoch": 0.055084228515625,
      "grad_norm": 8.508932113647461,
      "learning_rate": 4.963949490971746e-06,
      "loss": 3.8478,
      "step": 5415
    },
    {
      "epoch": 0.055135091145833336,
      "grad_norm": 7.770327091217041,
      "learning_rate": 4.963881795076901e-06,
      "loss": 3.0364,
      "step": 5420
    },
    {
      "epoch": 0.055185953776041664,
      "grad_norm": 13.418745994567871,
      "learning_rate": 4.963814036144223e-06,
      "loss": 3.1178,
      "step": 5425
    },
    {
      "epoch": 0.05523681640625,
      "grad_norm": 17.769771575927734,
      "learning_rate": 4.963746214175448e-06,
      "loss": 3.5998,
      "step": 5430
    },
    {
      "epoch": 0.055287679036458336,
      "grad_norm": 12.966880798339844,
      "learning_rate": 4.96367832917231e-06,
      "loss": 3.2851,
      "step": 5435
    },
    {
      "epoch": 0.055338541666666664,
      "grad_norm": 13.409180641174316,
      "learning_rate": 4.9636103811365464e-06,
      "loss": 3.7049,
      "step": 5440
    },
    {
      "epoch": 0.055389404296875,
      "grad_norm": 13.363730430603027,
      "learning_rate": 4.963542370069895e-06,
      "loss": 3.6024,
      "step": 5445
    },
    {
      "epoch": 0.055440266927083336,
      "grad_norm": 13.596270561218262,
      "learning_rate": 4.963474295974095e-06,
      "loss": 3.5165,
      "step": 5450
    },
    {
      "epoch": 0.055491129557291664,
      "grad_norm": 10.22996711730957,
      "learning_rate": 4.96340615885089e-06,
      "loss": 3.6675,
      "step": 5455
    },
    {
      "epoch": 0.0555419921875,
      "grad_norm": 14.241738319396973,
      "learning_rate": 4.963337958702022e-06,
      "loss": 3.3507,
      "step": 5460
    },
    {
      "epoch": 0.055592854817708336,
      "grad_norm": 9.234009742736816,
      "learning_rate": 4.963269695529236e-06,
      "loss": 3.4202,
      "step": 5465
    },
    {
      "epoch": 0.055643717447916664,
      "grad_norm": 13.222739219665527,
      "learning_rate": 4.963201369334279e-06,
      "loss": 3.7927,
      "step": 5470
    },
    {
      "epoch": 0.055694580078125,
      "grad_norm": 11.704004287719727,
      "learning_rate": 4.963132980118899e-06,
      "loss": 3.3365,
      "step": 5475
    },
    {
      "epoch": 0.055745442708333336,
      "grad_norm": 11.018651008605957,
      "learning_rate": 4.963064527884845e-06,
      "loss": 3.6051,
      "step": 5480
    },
    {
      "epoch": 0.055796305338541664,
      "grad_norm": 14.942146301269531,
      "learning_rate": 4.96299601263387e-06,
      "loss": 3.2443,
      "step": 5485
    },
    {
      "epoch": 0.05584716796875,
      "grad_norm": 9.587372779846191,
      "learning_rate": 4.962927434367724e-06,
      "loss": 3.637,
      "step": 5490
    },
    {
      "epoch": 0.055898030598958336,
      "grad_norm": 9.663505554199219,
      "learning_rate": 4.9628587930881646e-06,
      "loss": 3.5114,
      "step": 5495
    },
    {
      "epoch": 0.055948893229166664,
      "grad_norm": 15.723515510559082,
      "learning_rate": 4.962790088796946e-06,
      "loss": 3.5283,
      "step": 5500
    },
    {
      "epoch": 0.055999755859375,
      "grad_norm": 11.564032554626465,
      "learning_rate": 4.962721321495827e-06,
      "loss": 3.523,
      "step": 5505
    },
    {
      "epoch": 0.056050618489583336,
      "grad_norm": 15.196375846862793,
      "learning_rate": 4.962652491186567e-06,
      "loss": 3.4245,
      "step": 5510
    },
    {
      "epoch": 0.056101481119791664,
      "grad_norm": 11.132039070129395,
      "learning_rate": 4.962583597870927e-06,
      "loss": 3.6306,
      "step": 5515
    },
    {
      "epoch": 0.05615234375,
      "grad_norm": 13.807767868041992,
      "learning_rate": 4.962514641550668e-06,
      "loss": 3.284,
      "step": 5520
    },
    {
      "epoch": 0.056203206380208336,
      "grad_norm": 8.87157154083252,
      "learning_rate": 4.962445622227558e-06,
      "loss": 3.5285,
      "step": 5525
    },
    {
      "epoch": 0.056254069010416664,
      "grad_norm": 16.512109756469727,
      "learning_rate": 4.962376539903359e-06,
      "loss": 3.5056,
      "step": 5530
    },
    {
      "epoch": 0.056304931640625,
      "grad_norm": 10.745555877685547,
      "learning_rate": 4.962307394579839e-06,
      "loss": 3.2763,
      "step": 5535
    },
    {
      "epoch": 0.056355794270833336,
      "grad_norm": 10.458619117736816,
      "learning_rate": 4.9622381862587685e-06,
      "loss": 3.3848,
      "step": 5540
    },
    {
      "epoch": 0.056406656901041664,
      "grad_norm": 10.014710426330566,
      "learning_rate": 4.962168914941919e-06,
      "loss": 3.4247,
      "step": 5545
    },
    {
      "epoch": 0.05645751953125,
      "grad_norm": 15.231337547302246,
      "learning_rate": 4.96209958063106e-06,
      "loss": 3.368,
      "step": 5550
    },
    {
      "epoch": 0.056508382161458336,
      "grad_norm": 14.728731155395508,
      "learning_rate": 4.962030183327967e-06,
      "loss": 3.4259,
      "step": 5555
    },
    {
      "epoch": 0.056559244791666664,
      "grad_norm": 14.27651596069336,
      "learning_rate": 4.961960723034415e-06,
      "loss": 3.3908,
      "step": 5560
    },
    {
      "epoch": 0.056610107421875,
      "grad_norm": 15.396354675292969,
      "learning_rate": 4.961891199752182e-06,
      "loss": 3.6155,
      "step": 5565
    },
    {
      "epoch": 0.056660970052083336,
      "grad_norm": 13.059240341186523,
      "learning_rate": 4.961821613483047e-06,
      "loss": 3.737,
      "step": 5570
    },
    {
      "epoch": 0.056711832682291664,
      "grad_norm": 21.01386833190918,
      "learning_rate": 4.961751964228788e-06,
      "loss": 3.4121,
      "step": 5575
    },
    {
      "epoch": 0.0567626953125,
      "grad_norm": 12.324559211730957,
      "learning_rate": 4.961682251991189e-06,
      "loss": 3.1513,
      "step": 5580
    },
    {
      "epoch": 0.056813557942708336,
      "grad_norm": 11.327527046203613,
      "learning_rate": 4.961612476772033e-06,
      "loss": 3.4368,
      "step": 5585
    },
    {
      "epoch": 0.056864420572916664,
      "grad_norm": 9.986822128295898,
      "learning_rate": 4.961542638573106e-06,
      "loss": 3.5226,
      "step": 5590
    },
    {
      "epoch": 0.056915283203125,
      "grad_norm": 16.936189651489258,
      "learning_rate": 4.961472737396193e-06,
      "loss": 3.698,
      "step": 5595
    },
    {
      "epoch": 0.056966145833333336,
      "grad_norm": 15.606807708740234,
      "learning_rate": 4.9614027732430835e-06,
      "loss": 3.4487,
      "step": 5600
    },
    {
      "epoch": 0.057017008463541664,
      "grad_norm": 11.489161491394043,
      "learning_rate": 4.961332746115568e-06,
      "loss": 3.5364,
      "step": 5605
    },
    {
      "epoch": 0.05706787109375,
      "grad_norm": 13.7288818359375,
      "learning_rate": 4.9612626560154375e-06,
      "loss": 3.6168,
      "step": 5610
    },
    {
      "epoch": 0.057118733723958336,
      "grad_norm": 8.421143531799316,
      "learning_rate": 4.961192502944485e-06,
      "loss": 3.4713,
      "step": 5615
    },
    {
      "epoch": 0.057169596354166664,
      "grad_norm": 17.100997924804688,
      "learning_rate": 4.961122286904506e-06,
      "loss": 3.392,
      "step": 5620
    },
    {
      "epoch": 0.057220458984375,
      "grad_norm": 17.024621963500977,
      "learning_rate": 4.961052007897297e-06,
      "loss": 3.6999,
      "step": 5625
    },
    {
      "epoch": 0.057271321614583336,
      "grad_norm": 8.612135887145996,
      "learning_rate": 4.960981665924655e-06,
      "loss": 3.2406,
      "step": 5630
    },
    {
      "epoch": 0.057322184244791664,
      "grad_norm": 11.135616302490234,
      "learning_rate": 4.9609112609883816e-06,
      "loss": 3.9177,
      "step": 5635
    },
    {
      "epoch": 0.057373046875,
      "grad_norm": 13.029520988464355,
      "learning_rate": 4.960840793090276e-06,
      "loss": 3.4563,
      "step": 5640
    },
    {
      "epoch": 0.057423909505208336,
      "grad_norm": 12.987259864807129,
      "learning_rate": 4.960770262232141e-06,
      "loss": 3.4741,
      "step": 5645
    },
    {
      "epoch": 0.057474772135416664,
      "grad_norm": 8.016860008239746,
      "learning_rate": 4.960699668415784e-06,
      "loss": 3.423,
      "step": 5650
    },
    {
      "epoch": 0.057525634765625,
      "grad_norm": 11.892767906188965,
      "learning_rate": 4.960629011643008e-06,
      "loss": 3.0988,
      "step": 5655
    },
    {
      "epoch": 0.057576497395833336,
      "grad_norm": 7.135873794555664,
      "learning_rate": 4.960558291915622e-06,
      "loss": 3.2616,
      "step": 5660
    },
    {
      "epoch": 0.057627360026041664,
      "grad_norm": 10.335238456726074,
      "learning_rate": 4.960487509235435e-06,
      "loss": 3.5319,
      "step": 5665
    },
    {
      "epoch": 0.05767822265625,
      "grad_norm": 10.540165901184082,
      "learning_rate": 4.96041666360426e-06,
      "loss": 3.3135,
      "step": 5670
    },
    {
      "epoch": 0.057729085286458336,
      "grad_norm": 17.09685516357422,
      "learning_rate": 4.9603457550239065e-06,
      "loss": 3.5976,
      "step": 5675
    },
    {
      "epoch": 0.057779947916666664,
      "grad_norm": 13.739051818847656,
      "learning_rate": 4.96027478349619e-06,
      "loss": 3.5974,
      "step": 5680
    },
    {
      "epoch": 0.057830810546875,
      "grad_norm": 13.99421501159668,
      "learning_rate": 4.960203749022927e-06,
      "loss": 3.2972,
      "step": 5685
    },
    {
      "epoch": 0.057881673177083336,
      "grad_norm": 12.529129028320312,
      "learning_rate": 4.960132651605934e-06,
      "loss": 3.4988,
      "step": 5690
    },
    {
      "epoch": 0.057932535807291664,
      "grad_norm": 13.570207595825195,
      "learning_rate": 4.96006149124703e-06,
      "loss": 3.2342,
      "step": 5695
    },
    {
      "epoch": 0.0579833984375,
      "grad_norm": 8.2381591796875,
      "learning_rate": 4.959990267948035e-06,
      "loss": 3.6295,
      "step": 5700
    },
    {
      "epoch": 0.058034261067708336,
      "grad_norm": 10.554726600646973,
      "learning_rate": 4.959918981710773e-06,
      "loss": 3.6647,
      "step": 5705
    },
    {
      "epoch": 0.058085123697916664,
      "grad_norm": 10.161996841430664,
      "learning_rate": 4.959847632537067e-06,
      "loss": 3.3671,
      "step": 5710
    },
    {
      "epoch": 0.058135986328125,
      "grad_norm": 10.479255676269531,
      "learning_rate": 4.959776220428743e-06,
      "loss": 3.6848,
      "step": 5715
    },
    {
      "epoch": 0.058186848958333336,
      "grad_norm": 13.689743995666504,
      "learning_rate": 4.959704745387626e-06,
      "loss": 3.6923,
      "step": 5720
    },
    {
      "epoch": 0.058237711588541664,
      "grad_norm": 15.075189590454102,
      "learning_rate": 4.9596332074155465e-06,
      "loss": 3.1929,
      "step": 5725
    },
    {
      "epoch": 0.05828857421875,
      "grad_norm": 10.657966613769531,
      "learning_rate": 4.959561606514335e-06,
      "loss": 3.3856,
      "step": 5730
    },
    {
      "epoch": 0.058339436848958336,
      "grad_norm": 17.425458908081055,
      "learning_rate": 4.959489942685822e-06,
      "loss": 3.5663,
      "step": 5735
    },
    {
      "epoch": 0.058390299479166664,
      "grad_norm": 10.390463829040527,
      "learning_rate": 4.959418215931843e-06,
      "loss": 4.1035,
      "step": 5740
    },
    {
      "epoch": 0.058441162109375,
      "grad_norm": 9.901952743530273,
      "learning_rate": 4.959346426254231e-06,
      "loss": 3.8645,
      "step": 5745
    },
    {
      "epoch": 0.058492024739583336,
      "grad_norm": 12.191425323486328,
      "learning_rate": 4.9592745736548235e-06,
      "loss": 3.5104,
      "step": 5750
    },
    {
      "epoch": 0.058542887369791664,
      "grad_norm": 8.743103981018066,
      "learning_rate": 4.959202658135459e-06,
      "loss": 3.4598,
      "step": 5755
    },
    {
      "epoch": 0.05859375,
      "grad_norm": 12.222476959228516,
      "learning_rate": 4.959130679697978e-06,
      "loss": 3.4976,
      "step": 5760
    },
    {
      "epoch": 0.058644612630208336,
      "grad_norm": 12.773666381835938,
      "learning_rate": 4.95905863834422e-06,
      "loss": 3.3259,
      "step": 5765
    },
    {
      "epoch": 0.058695475260416664,
      "grad_norm": 12.655311584472656,
      "learning_rate": 4.958986534076031e-06,
      "loss": 3.6635,
      "step": 5770
    },
    {
      "epoch": 0.058746337890625,
      "grad_norm": 10.425118446350098,
      "learning_rate": 4.9589143668952536e-06,
      "loss": 3.3051,
      "step": 5775
    },
    {
      "epoch": 0.058797200520833336,
      "grad_norm": 12.868205070495605,
      "learning_rate": 4.958842136803735e-06,
      "loss": 3.6249,
      "step": 5780
    },
    {
      "epoch": 0.058848063151041664,
      "grad_norm": 15.800631523132324,
      "learning_rate": 4.958769843803324e-06,
      "loss": 3.5991,
      "step": 5785
    },
    {
      "epoch": 0.05889892578125,
      "grad_norm": 10.733988761901855,
      "learning_rate": 4.958697487895869e-06,
      "loss": 3.6806,
      "step": 5790
    },
    {
      "epoch": 0.058949788411458336,
      "grad_norm": 14.783018112182617,
      "learning_rate": 4.9586250690832214e-06,
      "loss": 3.6191,
      "step": 5795
    },
    {
      "epoch": 0.059000651041666664,
      "grad_norm": 9.55103874206543,
      "learning_rate": 4.958552587367233e-06,
      "loss": 3.2202,
      "step": 5800
    },
    {
      "epoch": 0.059051513671875,
      "grad_norm": 12.465963363647461,
      "learning_rate": 4.958480042749762e-06,
      "loss": 3.0866,
      "step": 5805
    },
    {
      "epoch": 0.059102376302083336,
      "grad_norm": 10.959587097167969,
      "learning_rate": 4.958407435232659e-06,
      "loss": 3.6214,
      "step": 5810
    },
    {
      "epoch": 0.059153238932291664,
      "grad_norm": 20.63323974609375,
      "learning_rate": 4.958334764817786e-06,
      "loss": 4.0361,
      "step": 5815
    },
    {
      "epoch": 0.0592041015625,
      "grad_norm": 7.549029350280762,
      "learning_rate": 4.9582620315070005e-06,
      "loss": 3.3141,
      "step": 5820
    },
    {
      "epoch": 0.059254964192708336,
      "grad_norm": 12.9930419921875,
      "learning_rate": 4.958189235302164e-06,
      "loss": 3.397,
      "step": 5825
    },
    {
      "epoch": 0.059305826822916664,
      "grad_norm": 15.772438049316406,
      "learning_rate": 4.958116376205138e-06,
      "loss": 3.375,
      "step": 5830
    },
    {
      "epoch": 0.059356689453125,
      "grad_norm": 14.877005577087402,
      "learning_rate": 4.9580434542177875e-06,
      "loss": 3.2812,
      "step": 5835
    },
    {
      "epoch": 0.059407552083333336,
      "grad_norm": 14.17616081237793,
      "learning_rate": 4.957970469341977e-06,
      "loss": 3.4907,
      "step": 5840
    },
    {
      "epoch": 0.059458414713541664,
      "grad_norm": 16.345909118652344,
      "learning_rate": 4.957897421579576e-06,
      "loss": 3.3758,
      "step": 5845
    },
    {
      "epoch": 0.05950927734375,
      "grad_norm": 15.003165245056152,
      "learning_rate": 4.957824310932451e-06,
      "loss": 3.0737,
      "step": 5850
    },
    {
      "epoch": 0.059560139973958336,
      "grad_norm": 14.897906303405762,
      "learning_rate": 4.957751137402475e-06,
      "loss": 3.2275,
      "step": 5855
    },
    {
      "epoch": 0.059611002604166664,
      "grad_norm": 13.421442031860352,
      "learning_rate": 4.957677900991516e-06,
      "loss": 3.2209,
      "step": 5860
    },
    {
      "epoch": 0.059661865234375,
      "grad_norm": 9.91036319732666,
      "learning_rate": 4.957604601701453e-06,
      "loss": 3.9609,
      "step": 5865
    },
    {
      "epoch": 0.059712727864583336,
      "grad_norm": 18.146198272705078,
      "learning_rate": 4.957531239534158e-06,
      "loss": 3.4241,
      "step": 5870
    },
    {
      "epoch": 0.059763590494791664,
      "grad_norm": 13.376401901245117,
      "learning_rate": 4.957457814491509e-06,
      "loss": 3.4611,
      "step": 5875
    },
    {
      "epoch": 0.059814453125,
      "grad_norm": 9.991517066955566,
      "learning_rate": 4.957384326575383e-06,
      "loss": 3.1355,
      "step": 5880
    },
    {
      "epoch": 0.059865315755208336,
      "grad_norm": 8.1071138381958,
      "learning_rate": 4.9573107757876625e-06,
      "loss": 3.4358,
      "step": 5885
    },
    {
      "epoch": 0.059916178385416664,
      "grad_norm": 16.746749877929688,
      "learning_rate": 4.9572371621302284e-06,
      "loss": 3.6471,
      "step": 5890
    },
    {
      "epoch": 0.059967041015625,
      "grad_norm": 16.02521514892578,
      "learning_rate": 4.957163485604963e-06,
      "loss": 3.5407,
      "step": 5895
    },
    {
      "epoch": 0.060017903645833336,
      "grad_norm": 8.261488914489746,
      "learning_rate": 4.957089746213753e-06,
      "loss": 3.2416,
      "step": 5900
    },
    {
      "epoch": 0.060068766276041664,
      "grad_norm": 12.552051544189453,
      "learning_rate": 4.957015943958484e-06,
      "loss": 3.3061,
      "step": 5905
    },
    {
      "epoch": 0.06011962890625,
      "grad_norm": 15.044920921325684,
      "learning_rate": 4.956942078841045e-06,
      "loss": 3.8459,
      "step": 5910
    },
    {
      "epoch": 0.060170491536458336,
      "grad_norm": 13.593306541442871,
      "learning_rate": 4.9568681508633246e-06,
      "loss": 3.2017,
      "step": 5915
    },
    {
      "epoch": 0.060221354166666664,
      "grad_norm": 12.529535293579102,
      "learning_rate": 4.956794160027215e-06,
      "loss": 3.7657,
      "step": 5920
    },
    {
      "epoch": 0.060272216796875,
      "grad_norm": 13.670187950134277,
      "learning_rate": 4.9567201063346096e-06,
      "loss": 3.5618,
      "step": 5925
    },
    {
      "epoch": 0.060323079427083336,
      "grad_norm": 14.781356811523438,
      "learning_rate": 4.956645989787402e-06,
      "loss": 3.3705,
      "step": 5930
    },
    {
      "epoch": 0.060373942057291664,
      "grad_norm": 12.808194160461426,
      "learning_rate": 4.95657181038749e-06,
      "loss": 3.5778,
      "step": 5935
    },
    {
      "epoch": 0.0604248046875,
      "grad_norm": 9.7361421585083,
      "learning_rate": 4.956497568136769e-06,
      "loss": 3.9065,
      "step": 5940
    },
    {
      "epoch": 0.060475667317708336,
      "grad_norm": 8.544392585754395,
      "learning_rate": 4.9564232630371414e-06,
      "loss": 3.0755,
      "step": 5945
    },
    {
      "epoch": 0.060526529947916664,
      "grad_norm": 16.483707427978516,
      "learning_rate": 4.956348895090506e-06,
      "loss": 3.2558,
      "step": 5950
    },
    {
      "epoch": 0.060577392578125,
      "grad_norm": 17.052963256835938,
      "learning_rate": 4.956274464298766e-06,
      "loss": 3.6055,
      "step": 5955
    },
    {
      "epoch": 0.060628255208333336,
      "grad_norm": 13.169681549072266,
      "learning_rate": 4.956199970663827e-06,
      "loss": 3.6049,
      "step": 5960
    },
    {
      "epoch": 0.060679117838541664,
      "grad_norm": 11.642385482788086,
      "learning_rate": 4.956125414187594e-06,
      "loss": 3.3229,
      "step": 5965
    },
    {
      "epoch": 0.06072998046875,
      "grad_norm": 9.508110046386719,
      "learning_rate": 4.956050794871974e-06,
      "loss": 3.2438,
      "step": 5970
    },
    {
      "epoch": 0.060780843098958336,
      "grad_norm": 9.767243385314941,
      "learning_rate": 4.955976112718876e-06,
      "loss": 3.623,
      "step": 5975
    },
    {
      "epoch": 0.060831705729166664,
      "grad_norm": 13.47176456451416,
      "learning_rate": 4.955901367730212e-06,
      "loss": 3.5425,
      "step": 5980
    },
    {
      "epoch": 0.060882568359375,
      "grad_norm": 13.689672470092773,
      "learning_rate": 4.9558265599078935e-06,
      "loss": 3.4059,
      "step": 5985
    },
    {
      "epoch": 0.060933430989583336,
      "grad_norm": 14.70511531829834,
      "learning_rate": 4.955751689253834e-06,
      "loss": 3.2993,
      "step": 5990
    },
    {
      "epoch": 0.060984293619791664,
      "grad_norm": 10.678279876708984,
      "learning_rate": 4.955676755769951e-06,
      "loss": 3.8429,
      "step": 5995
    },
    {
      "epoch": 0.06103515625,
      "grad_norm": 9.907837867736816,
      "learning_rate": 4.955601759458158e-06,
      "loss": 3.6905,
      "step": 6000
    },
    {
      "epoch": 0.061086018880208336,
      "grad_norm": 14.26388931274414,
      "learning_rate": 4.955526700320378e-06,
      "loss": 3.8103,
      "step": 6005
    },
    {
      "epoch": 0.061136881510416664,
      "grad_norm": 13.094128608703613,
      "learning_rate": 4.955451578358529e-06,
      "loss": 3.5114,
      "step": 6010
    },
    {
      "epoch": 0.061187744140625,
      "grad_norm": 9.898237228393555,
      "learning_rate": 4.955376393574533e-06,
      "loss": 3.0525,
      "step": 6015
    },
    {
      "epoch": 0.061238606770833336,
      "grad_norm": 20.232635498046875,
      "learning_rate": 4.955301145970314e-06,
      "loss": 3.3842,
      "step": 6020
    },
    {
      "epoch": 0.061289469401041664,
      "grad_norm": 10.714902877807617,
      "learning_rate": 4.955225835547798e-06,
      "loss": 3.2338,
      "step": 6025
    },
    {
      "epoch": 0.06134033203125,
      "grad_norm": 16.868967056274414,
      "learning_rate": 4.95515046230891e-06,
      "loss": 3.3336,
      "step": 6030
    },
    {
      "epoch": 0.061391194661458336,
      "grad_norm": 10.545510292053223,
      "learning_rate": 4.9550750262555795e-06,
      "loss": 3.4626,
      "step": 6035
    },
    {
      "epoch": 0.061442057291666664,
      "grad_norm": 16.030223846435547,
      "learning_rate": 4.9549995273897365e-06,
      "loss": 3.746,
      "step": 6040
    },
    {
      "epoch": 0.061492919921875,
      "grad_norm": 16.585622787475586,
      "learning_rate": 4.954923965713312e-06,
      "loss": 3.5239,
      "step": 6045
    },
    {
      "epoch": 0.061543782552083336,
      "grad_norm": 10.373279571533203,
      "learning_rate": 4.95484834122824e-06,
      "loss": 3.4698,
      "step": 6050
    },
    {
      "epoch": 0.061594645182291664,
      "grad_norm": 11.033610343933105,
      "learning_rate": 4.954772653936455e-06,
      "loss": 3.521,
      "step": 6055
    },
    {
      "epoch": 0.0616455078125,
      "grad_norm": 12.441916465759277,
      "learning_rate": 4.954696903839894e-06,
      "loss": 3.5568,
      "step": 6060
    },
    {
      "epoch": 0.061696370442708336,
      "grad_norm": 12.675573348999023,
      "learning_rate": 4.954621090940495e-06,
      "loss": 3.199,
      "step": 6065
    },
    {
      "epoch": 0.061747233072916664,
      "grad_norm": 11.934941291809082,
      "learning_rate": 4.9545452152401965e-06,
      "loss": 3.5181,
      "step": 6070
    },
    {
      "epoch": 0.061798095703125,
      "grad_norm": 12.990299224853516,
      "learning_rate": 4.95446927674094e-06,
      "loss": 3.1343,
      "step": 6075
    },
    {
      "epoch": 0.061848958333333336,
      "grad_norm": 15.312989234924316,
      "learning_rate": 4.954393275444669e-06,
      "loss": 2.912,
      "step": 6080
    },
    {
      "epoch": 0.061899820963541664,
      "grad_norm": 16.04402732849121,
      "learning_rate": 4.954317211353328e-06,
      "loss": 3.5654,
      "step": 6085
    },
    {
      "epoch": 0.06195068359375,
      "grad_norm": 16.57796287536621,
      "learning_rate": 4.954241084468863e-06,
      "loss": 3.4681,
      "step": 6090
    },
    {
      "epoch": 0.062001546223958336,
      "grad_norm": 12.278860092163086,
      "learning_rate": 4.954164894793222e-06,
      "loss": 3.4626,
      "step": 6095
    },
    {
      "epoch": 0.062052408854166664,
      "grad_norm": 85.06364440917969,
      "learning_rate": 4.954088642328353e-06,
      "loss": 3.5676,
      "step": 6100
    },
    {
      "epoch": 0.062103271484375,
      "grad_norm": 9.45626163482666,
      "learning_rate": 4.954012327076207e-06,
      "loss": 3.7756,
      "step": 6105
    },
    {
      "epoch": 0.062154134114583336,
      "grad_norm": 13.716742515563965,
      "learning_rate": 4.95393594903874e-06,
      "loss": 3.5243,
      "step": 6110
    },
    {
      "epoch": 0.062204996744791664,
      "grad_norm": 19.727203369140625,
      "learning_rate": 4.953859508217901e-06,
      "loss": 3.479,
      "step": 6115
    },
    {
      "epoch": 0.062255859375,
      "grad_norm": 14.221769332885742,
      "learning_rate": 4.953783004615649e-06,
      "loss": 3.3731,
      "step": 6120
    },
    {
      "epoch": 0.062306722005208336,
      "grad_norm": 8.173396110534668,
      "learning_rate": 4.953706438233941e-06,
      "loss": 3.2895,
      "step": 6125
    },
    {
      "epoch": 0.062357584635416664,
      "grad_norm": 11.394742965698242,
      "learning_rate": 4.953629809074734e-06,
      "loss": 3.3431,
      "step": 6130
    },
    {
      "epoch": 0.062408447265625,
      "grad_norm": 7.780871391296387,
      "learning_rate": 4.953553117139991e-06,
      "loss": 3.6523,
      "step": 6135
    },
    {
      "epoch": 0.062459309895833336,
      "grad_norm": 12.438297271728516,
      "learning_rate": 4.953476362431672e-06,
      "loss": 3.4039,
      "step": 6140
    },
    {
      "epoch": 0.06251017252604167,
      "grad_norm": 12.829780578613281,
      "learning_rate": 4.953399544951742e-06,
      "loss": 3.5707,
      "step": 6145
    },
    {
      "epoch": 0.06256103515625,
      "grad_norm": 11.045709609985352,
      "learning_rate": 4.953322664702167e-06,
      "loss": 3.1758,
      "step": 6150
    },
    {
      "epoch": 0.06261189778645833,
      "grad_norm": 13.500697135925293,
      "learning_rate": 4.953245721684913e-06,
      "loss": 3.5634,
      "step": 6155
    },
    {
      "epoch": 0.06266276041666667,
      "grad_norm": 13.87531852722168,
      "learning_rate": 4.953168715901949e-06,
      "loss": 3.0295,
      "step": 6160
    },
    {
      "epoch": 0.062713623046875,
      "grad_norm": 10.557450294494629,
      "learning_rate": 4.953091647355244e-06,
      "loss": 3.6556,
      "step": 6165
    },
    {
      "epoch": 0.06276448567708333,
      "grad_norm": 14.243014335632324,
      "learning_rate": 4.953014516046771e-06,
      "loss": 3.493,
      "step": 6170
    },
    {
      "epoch": 0.06281534830729167,
      "grad_norm": 14.641172409057617,
      "learning_rate": 4.952937321978502e-06,
      "loss": 3.5743,
      "step": 6175
    },
    {
      "epoch": 0.0628662109375,
      "grad_norm": 14.950688362121582,
      "learning_rate": 4.952860065152415e-06,
      "loss": 3.3458,
      "step": 6180
    },
    {
      "epoch": 0.06291707356770833,
      "grad_norm": 10.996521949768066,
      "learning_rate": 4.952782745570483e-06,
      "loss": 3.3334,
      "step": 6185
    },
    {
      "epoch": 0.06296793619791667,
      "grad_norm": 10.796847343444824,
      "learning_rate": 4.952705363234687e-06,
      "loss": 3.6746,
      "step": 6190
    },
    {
      "epoch": 0.063018798828125,
      "grad_norm": 16.92947006225586,
      "learning_rate": 4.952627918147005e-06,
      "loss": 3.3581,
      "step": 6195
    },
    {
      "epoch": 0.06306966145833333,
      "grad_norm": 9.009227752685547,
      "learning_rate": 4.952550410309419e-06,
      "loss": 3.3688,
      "step": 6200
    },
    {
      "epoch": 0.06312052408854167,
      "grad_norm": 9.859138488769531,
      "learning_rate": 4.952472839723912e-06,
      "loss": 3.2627,
      "step": 6205
    },
    {
      "epoch": 0.06317138671875,
      "grad_norm": 16.701183319091797,
      "learning_rate": 4.952395206392469e-06,
      "loss": 3.6775,
      "step": 6210
    },
    {
      "epoch": 0.06322224934895833,
      "grad_norm": 17.42451286315918,
      "learning_rate": 4.952317510317076e-06,
      "loss": 3.5911,
      "step": 6215
    },
    {
      "epoch": 0.06327311197916667,
      "grad_norm": 12.738759994506836,
      "learning_rate": 4.952239751499721e-06,
      "loss": 3.2518,
      "step": 6220
    },
    {
      "epoch": 0.063323974609375,
      "grad_norm": 12.116572380065918,
      "learning_rate": 4.952161929942393e-06,
      "loss": 3.6424,
      "step": 6225
    },
    {
      "epoch": 0.06337483723958333,
      "grad_norm": 15.029314041137695,
      "learning_rate": 4.952084045647083e-06,
      "loss": 3.4677,
      "step": 6230
    },
    {
      "epoch": 0.06342569986979167,
      "grad_norm": 16.844175338745117,
      "learning_rate": 4.952006098615784e-06,
      "loss": 3.0719,
      "step": 6235
    },
    {
      "epoch": 0.0634765625,
      "grad_norm": 16.886533737182617,
      "learning_rate": 4.95192808885049e-06,
      "loss": 3.4529,
      "step": 6240
    },
    {
      "epoch": 0.06352742513020833,
      "grad_norm": 15.081376075744629,
      "learning_rate": 4.9518500163531966e-06,
      "loss": 3.1723,
      "step": 6245
    },
    {
      "epoch": 0.06357828776041667,
      "grad_norm": 9.575623512268066,
      "learning_rate": 4.951771881125903e-06,
      "loss": 3.5933,
      "step": 6250
    },
    {
      "epoch": 0.063629150390625,
      "grad_norm": 9.829601287841797,
      "learning_rate": 4.951693683170606e-06,
      "loss": 3.7078,
      "step": 6255
    },
    {
      "epoch": 0.06368001302083333,
      "grad_norm": 10.027034759521484,
      "learning_rate": 4.951615422489308e-06,
      "loss": 3.3214,
      "step": 6260
    },
    {
      "epoch": 0.06373087565104167,
      "grad_norm": 14.713983535766602,
      "learning_rate": 4.9515370990840095e-06,
      "loss": 3.9297,
      "step": 6265
    },
    {
      "epoch": 0.06378173828125,
      "grad_norm": 10.196232795715332,
      "learning_rate": 4.951458712956716e-06,
      "loss": 3.3136,
      "step": 6270
    },
    {
      "epoch": 0.06383260091145833,
      "grad_norm": 8.880327224731445,
      "learning_rate": 4.9513802641094325e-06,
      "loss": 3.4526,
      "step": 6275
    },
    {
      "epoch": 0.06388346354166667,
      "grad_norm": 10.190267562866211,
      "learning_rate": 4.951301752544165e-06,
      "loss": 3.8351,
      "step": 6280
    },
    {
      "epoch": 0.063934326171875,
      "grad_norm": 13.929616928100586,
      "learning_rate": 4.951223178262924e-06,
      "loss": 3.3708,
      "step": 6285
    },
    {
      "epoch": 0.06398518880208333,
      "grad_norm": 12.750432014465332,
      "learning_rate": 4.951144541267719e-06,
      "loss": 3.617,
      "step": 6290
    },
    {
      "epoch": 0.06403605143229167,
      "grad_norm": 17.89909553527832,
      "learning_rate": 4.951065841560561e-06,
      "loss": 3.6072,
      "step": 6295
    },
    {
      "epoch": 0.0640869140625,
      "grad_norm": 14.314258575439453,
      "learning_rate": 4.950987079143465e-06,
      "loss": 3.4483,
      "step": 6300
    },
    {
      "epoch": 0.06413777669270833,
      "grad_norm": 12.541434288024902,
      "learning_rate": 4.950908254018446e-06,
      "loss": 3.9403,
      "step": 6305
    },
    {
      "epoch": 0.06418863932291667,
      "grad_norm": 10.588662147521973,
      "learning_rate": 4.9508293661875205e-06,
      "loss": 3.3919,
      "step": 6310
    },
    {
      "epoch": 0.064239501953125,
      "grad_norm": 14.57321834564209,
      "learning_rate": 4.950750415652706e-06,
      "loss": 3.6334,
      "step": 6315
    },
    {
      "epoch": 0.06429036458333333,
      "grad_norm": 12.940896034240723,
      "learning_rate": 4.950671402416023e-06,
      "loss": 3.5378,
      "step": 6320
    },
    {
      "epoch": 0.06434122721354167,
      "grad_norm": 14.000555992126465,
      "learning_rate": 4.9505923264794935e-06,
      "loss": 3.7651,
      "step": 6325
    },
    {
      "epoch": 0.06439208984375,
      "grad_norm": 12.165482521057129,
      "learning_rate": 4.95051318784514e-06,
      "loss": 3.3013,
      "step": 6330
    },
    {
      "epoch": 0.06444295247395833,
      "grad_norm": 10.64943790435791,
      "learning_rate": 4.950433986514986e-06,
      "loss": 3.6277,
      "step": 6335
    },
    {
      "epoch": 0.06449381510416667,
      "grad_norm": 9.537559509277344,
      "learning_rate": 4.9503547224910605e-06,
      "loss": 3.3196,
      "step": 6340
    },
    {
      "epoch": 0.064544677734375,
      "grad_norm": 12.809176445007324,
      "learning_rate": 4.9502753957753905e-06,
      "loss": 3.5917,
      "step": 6345
    },
    {
      "epoch": 0.06459554036458333,
      "grad_norm": 9.395776748657227,
      "learning_rate": 4.950196006370005e-06,
      "loss": 3.4648,
      "step": 6350
    },
    {
      "epoch": 0.06464640299479167,
      "grad_norm": 14.865406036376953,
      "learning_rate": 4.950116554276936e-06,
      "loss": 3.5405,
      "step": 6355
    },
    {
      "epoch": 0.064697265625,
      "grad_norm": 13.822293281555176,
      "learning_rate": 4.950037039498215e-06,
      "loss": 3.2149,
      "step": 6360
    },
    {
      "epoch": 0.06474812825520833,
      "grad_norm": 10.368663787841797,
      "learning_rate": 4.949957462035877e-06,
      "loss": 3.2916,
      "step": 6365
    },
    {
      "epoch": 0.06479899088541667,
      "grad_norm": 15.785028457641602,
      "learning_rate": 4.949877821891958e-06,
      "loss": 3.2872,
      "step": 6370
    },
    {
      "epoch": 0.064849853515625,
      "grad_norm": 9.993640899658203,
      "learning_rate": 4.949798119068495e-06,
      "loss": 3.3167,
      "step": 6375
    },
    {
      "epoch": 0.06490071614583333,
      "grad_norm": 14.402359962463379,
      "learning_rate": 4.949718353567529e-06,
      "loss": 3.7203,
      "step": 6380
    },
    {
      "epoch": 0.06495157877604167,
      "grad_norm": 13.888138771057129,
      "learning_rate": 4.9496385253910996e-06,
      "loss": 3.4504,
      "step": 6385
    },
    {
      "epoch": 0.06500244140625,
      "grad_norm": 19.04587745666504,
      "learning_rate": 4.949558634541249e-06,
      "loss": 3.4553,
      "step": 6390
    },
    {
      "epoch": 0.06505330403645833,
      "grad_norm": 17.151155471801758,
      "learning_rate": 4.94947868102002e-06,
      "loss": 3.7058,
      "step": 6395
    },
    {
      "epoch": 0.06510416666666667,
      "grad_norm": 13.129286766052246,
      "learning_rate": 4.949398664829461e-06,
      "loss": 3.5657,
      "step": 6400
    },
    {
      "epoch": 0.065155029296875,
      "grad_norm": 12.586851119995117,
      "learning_rate": 4.949318585971617e-06,
      "loss": 3.2961,
      "step": 6405
    },
    {
      "epoch": 0.06520589192708333,
      "grad_norm": 15.264113426208496,
      "learning_rate": 4.949238444448539e-06,
      "loss": 3.7745,
      "step": 6410
    },
    {
      "epoch": 0.06525675455729167,
      "grad_norm": 12.554951667785645,
      "learning_rate": 4.949158240262274e-06,
      "loss": 3.455,
      "step": 6415
    },
    {
      "epoch": 0.0653076171875,
      "grad_norm": 11.962057113647461,
      "learning_rate": 4.949077973414877e-06,
      "loss": 3.2963,
      "step": 6420
    },
    {
      "epoch": 0.06535847981770833,
      "grad_norm": 12.596719741821289,
      "learning_rate": 4.9489976439084e-06,
      "loss": 3.7278,
      "step": 6425
    },
    {
      "epoch": 0.06540934244791667,
      "grad_norm": 10.47852897644043,
      "learning_rate": 4.948917251744899e-06,
      "loss": 3.6283,
      "step": 6430
    },
    {
      "epoch": 0.065460205078125,
      "grad_norm": 9.309669494628906,
      "learning_rate": 4.9488367969264304e-06,
      "loss": 3.3136,
      "step": 6435
    },
    {
      "epoch": 0.06551106770833333,
      "grad_norm": 13.344404220581055,
      "learning_rate": 4.9487562794550535e-06,
      "loss": 3.3655,
      "step": 6440
    },
    {
      "epoch": 0.06556193033854167,
      "grad_norm": 13.794371604919434,
      "learning_rate": 4.948675699332827e-06,
      "loss": 3.2034,
      "step": 6445
    },
    {
      "epoch": 0.06561279296875,
      "grad_norm": 14.009343147277832,
      "learning_rate": 4.9485950565618134e-06,
      "loss": 3.4136,
      "step": 6450
    },
    {
      "epoch": 0.06566365559895833,
      "grad_norm": 10.496328353881836,
      "learning_rate": 4.9485143511440754e-06,
      "loss": 3.5469,
      "step": 6455
    },
    {
      "epoch": 0.06571451822916667,
      "grad_norm": 18.860055923461914,
      "learning_rate": 4.948433583081678e-06,
      "loss": 4.3312,
      "step": 6460
    },
    {
      "epoch": 0.065765380859375,
      "grad_norm": 12.226606369018555,
      "learning_rate": 4.948352752376689e-06,
      "loss": 3.187,
      "step": 6465
    },
    {
      "epoch": 0.06581624348958333,
      "grad_norm": 8.17588996887207,
      "learning_rate": 4.948271859031173e-06,
      "loss": 3.4183,
      "step": 6470
    },
    {
      "epoch": 0.06586710611979167,
      "grad_norm": 18.25055694580078,
      "learning_rate": 4.948190903047203e-06,
      "loss": 3.9294,
      "step": 6475
    },
    {
      "epoch": 0.06591796875,
      "grad_norm": 15.113214492797852,
      "learning_rate": 4.948109884426849e-06,
      "loss": 3.2971,
      "step": 6480
    },
    {
      "epoch": 0.06596883138020833,
      "grad_norm": 10.636944770812988,
      "learning_rate": 4.9480288031721835e-06,
      "loss": 3.1685,
      "step": 6485
    },
    {
      "epoch": 0.06601969401041667,
      "grad_norm": 12.23661994934082,
      "learning_rate": 4.947947659285281e-06,
      "loss": 3.2994,
      "step": 6490
    },
    {
      "epoch": 0.066070556640625,
      "grad_norm": 14.394771575927734,
      "learning_rate": 4.9478664527682194e-06,
      "loss": 3.5549,
      "step": 6495
    },
    {
      "epoch": 0.06612141927083333,
      "grad_norm": 12.141425132751465,
      "learning_rate": 4.947785183623074e-06,
      "loss": 3.5396,
      "step": 6500
    },
    {
      "epoch": 0.06617228190104167,
      "grad_norm": 11.212850570678711,
      "learning_rate": 4.9477038518519235e-06,
      "loss": 3.3503,
      "step": 6505
    },
    {
      "epoch": 0.06622314453125,
      "grad_norm": 13.835770606994629,
      "learning_rate": 4.947622457456852e-06,
      "loss": 3.1723,
      "step": 6510
    },
    {
      "epoch": 0.06627400716145833,
      "grad_norm": 13.431824684143066,
      "learning_rate": 4.94754100043994e-06,
      "loss": 3.339,
      "step": 6515
    },
    {
      "epoch": 0.06632486979166667,
      "grad_norm": 12.01131820678711,
      "learning_rate": 4.94745948080327e-06,
      "loss": 3.2717,
      "step": 6520
    },
    {
      "epoch": 0.066375732421875,
      "grad_norm": 14.991222381591797,
      "learning_rate": 4.947377898548931e-06,
      "loss": 3.52,
      "step": 6525
    },
    {
      "epoch": 0.06642659505208333,
      "grad_norm": 14.108927726745605,
      "learning_rate": 4.947296253679008e-06,
      "loss": 3.1859,
      "step": 6530
    },
    {
      "epoch": 0.06647745768229167,
      "grad_norm": 9.38899040222168,
      "learning_rate": 4.94721454619559e-06,
      "loss": 3.6626,
      "step": 6535
    },
    {
      "epoch": 0.0665283203125,
      "grad_norm": 12.356621742248535,
      "learning_rate": 4.947132776100768e-06,
      "loss": 3.8692,
      "step": 6540
    },
    {
      "epoch": 0.06657918294270833,
      "grad_norm": 16.474075317382812,
      "learning_rate": 4.947050943396634e-06,
      "loss": 3.0192,
      "step": 6545
    },
    {
      "epoch": 0.06663004557291667,
      "grad_norm": 12.742081642150879,
      "learning_rate": 4.9469690480852824e-06,
      "loss": 3.3703,
      "step": 6550
    },
    {
      "epoch": 0.066680908203125,
      "grad_norm": 13.639100074768066,
      "learning_rate": 4.946887090168807e-06,
      "loss": 3.8247,
      "step": 6555
    },
    {
      "epoch": 0.06673177083333333,
      "grad_norm": 9.607970237731934,
      "learning_rate": 4.946805069649305e-06,
      "loss": 3.4566,
      "step": 6560
    },
    {
      "epoch": 0.06678263346354167,
      "grad_norm": 13.177118301391602,
      "learning_rate": 4.946722986528876e-06,
      "loss": 3.5717,
      "step": 6565
    },
    {
      "epoch": 0.06683349609375,
      "grad_norm": 12.43066120147705,
      "learning_rate": 4.946640840809619e-06,
      "loss": 3.5391,
      "step": 6570
    },
    {
      "epoch": 0.06688435872395833,
      "grad_norm": 13.892637252807617,
      "learning_rate": 4.946558632493636e-06,
      "loss": 3.802,
      "step": 6575
    },
    {
      "epoch": 0.06693522135416667,
      "grad_norm": 13.632702827453613,
      "learning_rate": 4.94647636158303e-06,
      "loss": 3.3285,
      "step": 6580
    },
    {
      "epoch": 0.066986083984375,
      "grad_norm": 12.024981498718262,
      "learning_rate": 4.946394028079907e-06,
      "loss": 3.2208,
      "step": 6585
    },
    {
      "epoch": 0.06703694661458333,
      "grad_norm": 15.032366752624512,
      "learning_rate": 4.946311631986372e-06,
      "loss": 3.374,
      "step": 6590
    },
    {
      "epoch": 0.06708780924479167,
      "grad_norm": 10.841226577758789,
      "learning_rate": 4.946229173304535e-06,
      "loss": 3.2525,
      "step": 6595
    },
    {
      "epoch": 0.067138671875,
      "grad_norm": 10.338508605957031,
      "learning_rate": 4.946146652036502e-06,
      "loss": 3.5958,
      "step": 6600
    },
    {
      "epoch": 0.06718953450520833,
      "grad_norm": 10.029006004333496,
      "learning_rate": 4.9460640681843885e-06,
      "loss": 3.6985,
      "step": 6605
    },
    {
      "epoch": 0.06724039713541667,
      "grad_norm": 9.308262825012207,
      "learning_rate": 4.945981421750305e-06,
      "loss": 3.3633,
      "step": 6610
    },
    {
      "epoch": 0.067291259765625,
      "grad_norm": 8.373737335205078,
      "learning_rate": 4.945898712736366e-06,
      "loss": 3.5412,
      "step": 6615
    },
    {
      "epoch": 0.06734212239583333,
      "grad_norm": 11.022754669189453,
      "learning_rate": 4.94581594114469e-06,
      "loss": 4.3451,
      "step": 6620
    },
    {
      "epoch": 0.06739298502604167,
      "grad_norm": 7.635653972625732,
      "learning_rate": 4.945733106977391e-06,
      "loss": 3.1132,
      "step": 6625
    },
    {
      "epoch": 0.06744384765625,
      "grad_norm": 11.958892822265625,
      "learning_rate": 4.945650210236591e-06,
      "loss": 3.736,
      "step": 6630
    },
    {
      "epoch": 0.06749471028645833,
      "grad_norm": 17.302410125732422,
      "learning_rate": 4.94556725092441e-06,
      "loss": 3.2335,
      "step": 6635
    },
    {
      "epoch": 0.06754557291666667,
      "grad_norm": 10.765392303466797,
      "learning_rate": 4.945484229042971e-06,
      "loss": 3.0814,
      "step": 6640
    },
    {
      "epoch": 0.067596435546875,
      "grad_norm": 10.385746002197266,
      "learning_rate": 4.945401144594397e-06,
      "loss": 3.4643,
      "step": 6645
    },
    {
      "epoch": 0.06764729817708333,
      "grad_norm": 14.88101577758789,
      "learning_rate": 4.945317997580814e-06,
      "loss": 3.7192,
      "step": 6650
    },
    {
      "epoch": 0.06769816080729167,
      "grad_norm": 10.534723281860352,
      "learning_rate": 4.9452347880043505e-06,
      "loss": 3.4338,
      "step": 6655
    },
    {
      "epoch": 0.0677490234375,
      "grad_norm": 10.487231254577637,
      "learning_rate": 4.945151515867134e-06,
      "loss": 3.6227,
      "step": 6660
    },
    {
      "epoch": 0.06779988606770833,
      "grad_norm": 10.452747344970703,
      "learning_rate": 4.9450681811712954e-06,
      "loss": 3.2158,
      "step": 6665
    },
    {
      "epoch": 0.06785074869791667,
      "grad_norm": 12.921323776245117,
      "learning_rate": 4.944984783918968e-06,
      "loss": 3.5547,
      "step": 6670
    },
    {
      "epoch": 0.067901611328125,
      "grad_norm": 12.540404319763184,
      "learning_rate": 4.944901324112283e-06,
      "loss": 3.5022,
      "step": 6675
    },
    {
      "epoch": 0.06795247395833333,
      "grad_norm": 11.059189796447754,
      "learning_rate": 4.9448178017533775e-06,
      "loss": 3.3226,
      "step": 6680
    },
    {
      "epoch": 0.06800333658854167,
      "grad_norm": 19.822628021240234,
      "learning_rate": 4.944734216844388e-06,
      "loss": 3.7347,
      "step": 6685
    },
    {
      "epoch": 0.06805419921875,
      "grad_norm": 12.81627082824707,
      "learning_rate": 4.944650569387453e-06,
      "loss": 3.9766,
      "step": 6690
    },
    {
      "epoch": 0.06810506184895833,
      "grad_norm": 14.65715217590332,
      "learning_rate": 4.944566859384714e-06,
      "loss": 3.2335,
      "step": 6695
    },
    {
      "epoch": 0.06815592447916667,
      "grad_norm": 9.100480079650879,
      "learning_rate": 4.94448308683831e-06,
      "loss": 3.3709,
      "step": 6700
    },
    {
      "epoch": 0.068206787109375,
      "grad_norm": 12.44413948059082,
      "learning_rate": 4.944399251750386e-06,
      "loss": 3.705,
      "step": 6705
    },
    {
      "epoch": 0.06825764973958333,
      "grad_norm": 612.3552856445312,
      "learning_rate": 4.944315354123086e-06,
      "loss": 3.851,
      "step": 6710
    },
    {
      "epoch": 0.06830851236979167,
      "grad_norm": 11.150618553161621,
      "learning_rate": 4.944231393958558e-06,
      "loss": 3.3883,
      "step": 6715
    },
    {
      "epoch": 0.068359375,
      "grad_norm": 15.825736999511719,
      "learning_rate": 4.944147371258948e-06,
      "loss": 3.1653,
      "step": 6720
    },
    {
      "epoch": 0.06841023763020833,
      "grad_norm": 13.351125717163086,
      "learning_rate": 4.944063286026408e-06,
      "loss": 3.6262,
      "step": 6725
    },
    {
      "epoch": 0.06846110026041667,
      "grad_norm": 10.806169509887695,
      "learning_rate": 4.9439791382630875e-06,
      "loss": 3.4346,
      "step": 6730
    },
    {
      "epoch": 0.068511962890625,
      "grad_norm": 14.80085277557373,
      "learning_rate": 4.94389492797114e-06,
      "loss": 4.8293,
      "step": 6735
    },
    {
      "epoch": 0.06856282552083333,
      "grad_norm": 15.626360893249512,
      "learning_rate": 4.94381065515272e-06,
      "loss": 3.5732,
      "step": 6740
    },
    {
      "epoch": 0.06861368815104167,
      "grad_norm": 12.220220565795898,
      "learning_rate": 4.943726319809984e-06,
      "loss": 3.1158,
      "step": 6745
    },
    {
      "epoch": 0.06866455078125,
      "grad_norm": 11.84122371673584,
      "learning_rate": 4.943641921945089e-06,
      "loss": 3.2894,
      "step": 6750
    },
    {
      "epoch": 0.06871541341145833,
      "grad_norm": 11.213668823242188,
      "learning_rate": 4.943557461560195e-06,
      "loss": 3.5569,
      "step": 6755
    },
    {
      "epoch": 0.06876627604166667,
      "grad_norm": 10.223577499389648,
      "learning_rate": 4.943472938657462e-06,
      "loss": 3.4687,
      "step": 6760
    },
    {
      "epoch": 0.068817138671875,
      "grad_norm": 11.476598739624023,
      "learning_rate": 4.943388353239053e-06,
      "loss": 3.3717,
      "step": 6765
    },
    {
      "epoch": 0.06886800130208333,
      "grad_norm": 12.814501762390137,
      "learning_rate": 4.943303705307133e-06,
      "loss": 3.4168,
      "step": 6770
    },
    {
      "epoch": 0.06891886393229167,
      "grad_norm": 15.519283294677734,
      "learning_rate": 4.943218994863866e-06,
      "loss": 3.4865,
      "step": 6775
    },
    {
      "epoch": 0.0689697265625,
      "grad_norm": 10.789101600646973,
      "learning_rate": 4.943134221911421e-06,
      "loss": 3.3488,
      "step": 6780
    },
    {
      "epoch": 0.06902058919270833,
      "grad_norm": 16.44157600402832,
      "learning_rate": 4.943049386451964e-06,
      "loss": 3.7548,
      "step": 6785
    },
    {
      "epoch": 0.06907145182291667,
      "grad_norm": 12.07598876953125,
      "learning_rate": 4.942964488487669e-06,
      "loss": 3.6223,
      "step": 6790
    },
    {
      "epoch": 0.069122314453125,
      "grad_norm": 13.612014770507812,
      "learning_rate": 4.942879528020707e-06,
      "loss": 3.9421,
      "step": 6795
    },
    {
      "epoch": 0.06917317708333333,
      "grad_norm": 10.37803840637207,
      "learning_rate": 4.9427945050532515e-06,
      "loss": 3.6429,
      "step": 6800
    },
    {
      "epoch": 0.06922403971354167,
      "grad_norm": 10.748969078063965,
      "learning_rate": 4.942709419587476e-06,
      "loss": 3.4308,
      "step": 6805
    },
    {
      "epoch": 0.06927490234375,
      "grad_norm": 8.533020973205566,
      "learning_rate": 4.9426242716255605e-06,
      "loss": 3.2213,
      "step": 6810
    },
    {
      "epoch": 0.06932576497395833,
      "grad_norm": 14.239413261413574,
      "learning_rate": 4.942539061169681e-06,
      "loss": 3.5663,
      "step": 6815
    },
    {
      "epoch": 0.06937662760416667,
      "grad_norm": 12.664621353149414,
      "learning_rate": 4.942453788222019e-06,
      "loss": 3.5002,
      "step": 6820
    },
    {
      "epoch": 0.069427490234375,
      "grad_norm": 12.779163360595703,
      "learning_rate": 4.942368452784756e-06,
      "loss": 3.2068,
      "step": 6825
    },
    {
      "epoch": 0.06947835286458333,
      "grad_norm": 16.653684616088867,
      "learning_rate": 4.9422830548600745e-06,
      "loss": 3.3543,
      "step": 6830
    },
    {
      "epoch": 0.06952921549479167,
      "grad_norm": 16.349224090576172,
      "learning_rate": 4.94219759445016e-06,
      "loss": 3.5099,
      "step": 6835
    },
    {
      "epoch": 0.069580078125,
      "grad_norm": 8.445566177368164,
      "learning_rate": 4.942112071557199e-06,
      "loss": 3.793,
      "step": 6840
    },
    {
      "epoch": 0.06963094075520833,
      "grad_norm": 14.703461647033691,
      "learning_rate": 4.94202648618338e-06,
      "loss": 3.5832,
      "step": 6845
    },
    {
      "epoch": 0.06968180338541667,
      "grad_norm": 14.559622764587402,
      "learning_rate": 4.941940838330891e-06,
      "loss": 3.6992,
      "step": 6850
    },
    {
      "epoch": 0.069732666015625,
      "grad_norm": 11.245277404785156,
      "learning_rate": 4.941855128001925e-06,
      "loss": 2.9931,
      "step": 6855
    },
    {
      "epoch": 0.06978352864583333,
      "grad_norm": 17.25091552734375,
      "learning_rate": 4.941769355198675e-06,
      "loss": 3.243,
      "step": 6860
    },
    {
      "epoch": 0.06983439127604167,
      "grad_norm": 11.740355491638184,
      "learning_rate": 4.941683519923335e-06,
      "loss": 3.3695,
      "step": 6865
    },
    {
      "epoch": 0.06988525390625,
      "grad_norm": 12.884564399719238,
      "learning_rate": 4.9415976221781e-06,
      "loss": 3.3783,
      "step": 6870
    },
    {
      "epoch": 0.06993611653645833,
      "grad_norm": 14.360733032226562,
      "learning_rate": 4.9415116619651685e-06,
      "loss": 3.332,
      "step": 6875
    },
    {
      "epoch": 0.06998697916666667,
      "grad_norm": 15.234169960021973,
      "learning_rate": 4.94142563928674e-06,
      "loss": 3.0307,
      "step": 6880
    },
    {
      "epoch": 0.070037841796875,
      "grad_norm": 12.016648292541504,
      "learning_rate": 4.941339554145015e-06,
      "loss": 3.607,
      "step": 6885
    },
    {
      "epoch": 0.07008870442708333,
      "grad_norm": 10.132488250732422,
      "learning_rate": 4.941253406542197e-06,
      "loss": 3.7322,
      "step": 6890
    },
    {
      "epoch": 0.07013956705729167,
      "grad_norm": 11.677488327026367,
      "learning_rate": 4.941167196480489e-06,
      "loss": 3.7035,
      "step": 6895
    },
    {
      "epoch": 0.0701904296875,
      "grad_norm": 8.434168815612793,
      "learning_rate": 4.941080923962096e-06,
      "loss": 3.3046,
      "step": 6900
    },
    {
      "epoch": 0.07024129231770833,
      "grad_norm": 18.53049659729004,
      "learning_rate": 4.940994588989227e-06,
      "loss": 3.9959,
      "step": 6905
    },
    {
      "epoch": 0.07029215494791667,
      "grad_norm": 14.731730461120605,
      "learning_rate": 4.94090819156409e-06,
      "loss": 3.5238,
      "step": 6910
    },
    {
      "epoch": 0.070343017578125,
      "grad_norm": 9.122929573059082,
      "learning_rate": 4.940821731688895e-06,
      "loss": 3.525,
      "step": 6915
    },
    {
      "epoch": 0.07039388020833333,
      "grad_norm": 11.580704689025879,
      "learning_rate": 4.940735209365855e-06,
      "loss": 3.9973,
      "step": 6920
    },
    {
      "epoch": 0.07044474283854167,
      "grad_norm": 13.921919822692871,
      "learning_rate": 4.940648624597183e-06,
      "loss": 3.3295,
      "step": 6925
    },
    {
      "epoch": 0.07049560546875,
      "grad_norm": 11.508909225463867,
      "learning_rate": 4.9405619773850944e-06,
      "loss": 3.3846,
      "step": 6930
    },
    {
      "epoch": 0.07054646809895833,
      "grad_norm": 15.52819538116455,
      "learning_rate": 4.940475267731806e-06,
      "loss": 3.7001,
      "step": 6935
    },
    {
      "epoch": 0.07059733072916667,
      "grad_norm": 12.931517601013184,
      "learning_rate": 4.940388495639537e-06,
      "loss": 3.4707,
      "step": 6940
    },
    {
      "epoch": 0.070648193359375,
      "grad_norm": 16.401697158813477,
      "learning_rate": 4.9403016611105055e-06,
      "loss": 3.5479,
      "step": 6945
    },
    {
      "epoch": 0.07069905598958333,
      "grad_norm": 14.115461349487305,
      "learning_rate": 4.940214764146935e-06,
      "loss": 3.4651,
      "step": 6950
    },
    {
      "epoch": 0.07074991861979167,
      "grad_norm": 13.501302719116211,
      "learning_rate": 4.940127804751048e-06,
      "loss": 3.2613,
      "step": 6955
    },
    {
      "epoch": 0.07080078125,
      "grad_norm": 9.428580284118652,
      "learning_rate": 4.94004078292507e-06,
      "loss": 3.528,
      "step": 6960
    },
    {
      "epoch": 0.07085164388020833,
      "grad_norm": 15.635141372680664,
      "learning_rate": 4.939953698671227e-06,
      "loss": 3.413,
      "step": 6965
    },
    {
      "epoch": 0.07090250651041667,
      "grad_norm": 18.93190574645996,
      "learning_rate": 4.939866551991746e-06,
      "loss": 3.6146,
      "step": 6970
    },
    {
      "epoch": 0.070953369140625,
      "grad_norm": 13.07038688659668,
      "learning_rate": 4.939779342888858e-06,
      "loss": 3.6089,
      "step": 6975
    },
    {
      "epoch": 0.07100423177083333,
      "grad_norm": 11.98829174041748,
      "learning_rate": 4.939692071364794e-06,
      "loss": 3.2366,
      "step": 6980
    },
    {
      "epoch": 0.07105509440104167,
      "grad_norm": 13.00400161743164,
      "learning_rate": 4.939604737421787e-06,
      "loss": 3.6997,
      "step": 6985
    },
    {
      "epoch": 0.07110595703125,
      "grad_norm": 10.635180473327637,
      "learning_rate": 4.9395173410620714e-06,
      "loss": 3.5263,
      "step": 6990
    },
    {
      "epoch": 0.07115681966145833,
      "grad_norm": 8.874784469604492,
      "learning_rate": 4.939429882287881e-06,
      "loss": 3.7435,
      "step": 6995
    },
    {
      "epoch": 0.07120768229166667,
      "grad_norm": 17.223342895507812,
      "learning_rate": 4.939342361101457e-06,
      "loss": 3.248,
      "step": 7000
    },
    {
      "epoch": 0.071258544921875,
      "grad_norm": 15.709144592285156,
      "learning_rate": 4.939254777505037e-06,
      "loss": 3.3038,
      "step": 7005
    },
    {
      "epoch": 0.07130940755208333,
      "grad_norm": 10.639139175415039,
      "learning_rate": 4.93916713150086e-06,
      "loss": 3.433,
      "step": 7010
    },
    {
      "epoch": 0.07136027018229167,
      "grad_norm": 13.915593147277832,
      "learning_rate": 4.9390794230911715e-06,
      "loss": 3.4058,
      "step": 7015
    },
    {
      "epoch": 0.0714111328125,
      "grad_norm": 8.267925262451172,
      "learning_rate": 4.938991652278213e-06,
      "loss": 3.5928,
      "step": 7020
    },
    {
      "epoch": 0.07146199544270833,
      "grad_norm": 16.01654624938965,
      "learning_rate": 4.938903819064232e-06,
      "loss": 3.1889,
      "step": 7025
    },
    {
      "epoch": 0.07151285807291667,
      "grad_norm": 13.366415977478027,
      "learning_rate": 4.938815923451476e-06,
      "loss": 3.3296,
      "step": 7030
    },
    {
      "epoch": 0.071563720703125,
      "grad_norm": 12.922379493713379,
      "learning_rate": 4.9387279654421905e-06,
      "loss": 3.4845,
      "step": 7035
    },
    {
      "epoch": 0.07161458333333333,
      "grad_norm": 10.66179084777832,
      "learning_rate": 4.938639945038629e-06,
      "loss": 3.3688,
      "step": 7040
    },
    {
      "epoch": 0.07166544596354167,
      "grad_norm": 9.114995002746582,
      "learning_rate": 4.938551862243042e-06,
      "loss": 3.2822,
      "step": 7045
    },
    {
      "epoch": 0.07171630859375,
      "grad_norm": 10.946172714233398,
      "learning_rate": 4.9384637170576844e-06,
      "loss": 3.5603,
      "step": 7050
    },
    {
      "epoch": 0.07176717122395833,
      "grad_norm": 11.345202445983887,
      "learning_rate": 4.93837550948481e-06,
      "loss": 3.4276,
      "step": 7055
    },
    {
      "epoch": 0.07181803385416667,
      "grad_norm": 9.533265113830566,
      "learning_rate": 4.938287239526676e-06,
      "loss": 3.1513,
      "step": 7060
    },
    {
      "epoch": 0.071868896484375,
      "grad_norm": 9.234315872192383,
      "learning_rate": 4.938198907185542e-06,
      "loss": 3.3517,
      "step": 7065
    },
    {
      "epoch": 0.07191975911458333,
      "grad_norm": 10.09774398803711,
      "learning_rate": 4.938110512463666e-06,
      "loss": 3.5838,
      "step": 7070
    },
    {
      "epoch": 0.07197062174479167,
      "grad_norm": 14.220784187316895,
      "learning_rate": 4.938022055363311e-06,
      "loss": 3.4819,
      "step": 7075
    },
    {
      "epoch": 0.072021484375,
      "grad_norm": 10.781584739685059,
      "learning_rate": 4.9379335358867384e-06,
      "loss": 3.7672,
      "step": 7080
    },
    {
      "epoch": 0.07207234700520833,
      "grad_norm": 9.52526569366455,
      "learning_rate": 4.937844954036215e-06,
      "loss": 3.3561,
      "step": 7085
    },
    {
      "epoch": 0.07212320963541667,
      "grad_norm": 12.460453033447266,
      "learning_rate": 4.9377563098140065e-06,
      "loss": 3.4052,
      "step": 7090
    },
    {
      "epoch": 0.072174072265625,
      "grad_norm": 14.343060493469238,
      "learning_rate": 4.9376676032223805e-06,
      "loss": 3.214,
      "step": 7095
    },
    {
      "epoch": 0.07222493489583333,
      "grad_norm": 12.104986190795898,
      "learning_rate": 4.937578834263607e-06,
      "loss": 3.472,
      "step": 7100
    },
    {
      "epoch": 0.07227579752604167,
      "grad_norm": 12.275232315063477,
      "learning_rate": 4.9374900029399555e-06,
      "loss": 3.7637,
      "step": 7105
    },
    {
      "epoch": 0.07232666015625,
      "grad_norm": 12.134095191955566,
      "learning_rate": 4.937401109253701e-06,
      "loss": 3.6845,
      "step": 7110
    },
    {
      "epoch": 0.07237752278645833,
      "grad_norm": 15.203348159790039,
      "learning_rate": 4.937312153207117e-06,
      "loss": 3.5803,
      "step": 7115
    },
    {
      "epoch": 0.07242838541666667,
      "grad_norm": 17.58793830871582,
      "learning_rate": 4.937223134802478e-06,
      "loss": 3.1135,
      "step": 7120
    },
    {
      "epoch": 0.072479248046875,
      "grad_norm": 9.813260078430176,
      "learning_rate": 4.937134054042064e-06,
      "loss": 3.487,
      "step": 7125
    },
    {
      "epoch": 0.07253011067708333,
      "grad_norm": 12.799726486206055,
      "learning_rate": 4.9370449109281524e-06,
      "loss": 3.5097,
      "step": 7130
    },
    {
      "epoch": 0.07258097330729167,
      "grad_norm": 10.807867050170898,
      "learning_rate": 4.936955705463025e-06,
      "loss": 3.5704,
      "step": 7135
    },
    {
      "epoch": 0.0726318359375,
      "grad_norm": 13.212722778320312,
      "learning_rate": 4.936866437648963e-06,
      "loss": 3.2637,
      "step": 7140
    },
    {
      "epoch": 0.07268269856770833,
      "grad_norm": 10.858726501464844,
      "learning_rate": 4.936777107488251e-06,
      "loss": 3.3638,
      "step": 7145
    },
    {
      "epoch": 0.07273356119791667,
      "grad_norm": 14.568108558654785,
      "learning_rate": 4.936687714983174e-06,
      "loss": 3.2916,
      "step": 7150
    },
    {
      "epoch": 0.072784423828125,
      "grad_norm": 12.920387268066406,
      "learning_rate": 4.9365982601360194e-06,
      "loss": 3.4835,
      "step": 7155
    },
    {
      "epoch": 0.07283528645833333,
      "grad_norm": 13.888976097106934,
      "learning_rate": 4.9365087429490765e-06,
      "loss": 3.4057,
      "step": 7160
    },
    {
      "epoch": 0.07288614908854167,
      "grad_norm": 8.182246208190918,
      "learning_rate": 4.936419163424634e-06,
      "loss": 3.4935,
      "step": 7165
    },
    {
      "epoch": 0.07293701171875,
      "grad_norm": 9.68957805633545,
      "learning_rate": 4.936329521564986e-06,
      "loss": 3.4725,
      "step": 7170
    },
    {
      "epoch": 0.07298787434895833,
      "grad_norm": 8.523253440856934,
      "learning_rate": 4.936239817372423e-06,
      "loss": 3.3035,
      "step": 7175
    },
    {
      "epoch": 0.07303873697916667,
      "grad_norm": 15.208540916442871,
      "learning_rate": 4.936150050849242e-06,
      "loss": 3.3124,
      "step": 7180
    },
    {
      "epoch": 0.073089599609375,
      "grad_norm": 9.686799049377441,
      "learning_rate": 4.93606022199774e-06,
      "loss": 3.1586,
      "step": 7185
    },
    {
      "epoch": 0.07314046223958333,
      "grad_norm": 17.809568405151367,
      "learning_rate": 4.935970330820215e-06,
      "loss": 3.5217,
      "step": 7190
    },
    {
      "epoch": 0.07319132486979167,
      "grad_norm": 9.103455543518066,
      "learning_rate": 4.935880377318965e-06,
      "loss": 3.3448,
      "step": 7195
    },
    {
      "epoch": 0.0732421875,
      "grad_norm": 17.965932846069336,
      "learning_rate": 4.935790361496295e-06,
      "loss": 3.4703,
      "step": 7200
    },
    {
      "epoch": 0.07329305013020833,
      "grad_norm": 17.913076400756836,
      "learning_rate": 4.935700283354504e-06,
      "loss": 3.7618,
      "step": 7205
    },
    {
      "epoch": 0.07334391276041667,
      "grad_norm": 10.30252456665039,
      "learning_rate": 4.9356101428959e-06,
      "loss": 3.0109,
      "step": 7210
    },
    {
      "epoch": 0.073394775390625,
      "grad_norm": 9.662969589233398,
      "learning_rate": 4.935519940122787e-06,
      "loss": 3.4018,
      "step": 7215
    },
    {
      "epoch": 0.07344563802083333,
      "grad_norm": 15.70295524597168,
      "learning_rate": 4.935429675037474e-06,
      "loss": 3.083,
      "step": 7220
    },
    {
      "epoch": 0.07349650065104167,
      "grad_norm": 8.53459644317627,
      "learning_rate": 4.935339347642269e-06,
      "loss": 3.0109,
      "step": 7225
    },
    {
      "epoch": 0.07354736328125,
      "grad_norm": 15.32571792602539,
      "learning_rate": 4.935248957939486e-06,
      "loss": 3.5678,
      "step": 7230
    },
    {
      "epoch": 0.07359822591145833,
      "grad_norm": 15.416007995605469,
      "learning_rate": 4.935158505931434e-06,
      "loss": 3.2472,
      "step": 7235
    },
    {
      "epoch": 0.07364908854166667,
      "grad_norm": 12.033348083496094,
      "learning_rate": 4.93506799162043e-06,
      "loss": 3.3393,
      "step": 7240
    },
    {
      "epoch": 0.073699951171875,
      "grad_norm": 12.217211723327637,
      "learning_rate": 4.934977415008787e-06,
      "loss": 3.7256,
      "step": 7245
    },
    {
      "epoch": 0.07375081380208333,
      "grad_norm": 13.5538969039917,
      "learning_rate": 4.934886776098825e-06,
      "loss": 3.6471,
      "step": 7250
    },
    {
      "epoch": 0.07380167643229167,
      "grad_norm": 11.107625961303711,
      "learning_rate": 4.934796074892862e-06,
      "loss": 3.4801,
      "step": 7255
    },
    {
      "epoch": 0.0738525390625,
      "grad_norm": 8.796854972839355,
      "learning_rate": 4.934705311393219e-06,
      "loss": 2.999,
      "step": 7260
    },
    {
      "epoch": 0.07390340169270833,
      "grad_norm": 15.173271179199219,
      "learning_rate": 4.934614485602217e-06,
      "loss": 3.2074,
      "step": 7265
    },
    {
      "epoch": 0.07395426432291667,
      "grad_norm": 14.630196571350098,
      "learning_rate": 4.9345235975221804e-06,
      "loss": 3.2888,
      "step": 7270
    },
    {
      "epoch": 0.074005126953125,
      "grad_norm": 10.86033821105957,
      "learning_rate": 4.934432647155435e-06,
      "loss": 3.2936,
      "step": 7275
    },
    {
      "epoch": 0.07405598958333333,
      "grad_norm": 13.982941627502441,
      "learning_rate": 4.934341634504307e-06,
      "loss": 3.7184,
      "step": 7280
    },
    {
      "epoch": 0.07410685221354167,
      "grad_norm": 9.288113594055176,
      "learning_rate": 4.934250559571126e-06,
      "loss": 3.1865,
      "step": 7285
    },
    {
      "epoch": 0.07415771484375,
      "grad_norm": 11.010740280151367,
      "learning_rate": 4.93415942235822e-06,
      "loss": 3.442,
      "step": 7290
    },
    {
      "epoch": 0.07420857747395833,
      "grad_norm": 15.357698440551758,
      "learning_rate": 4.934068222867923e-06,
      "loss": 3.6836,
      "step": 7295
    },
    {
      "epoch": 0.07425944010416667,
      "grad_norm": 14.378852844238281,
      "learning_rate": 4.9339769611025675e-06,
      "loss": 3.4716,
      "step": 7300
    },
    {
      "epoch": 0.074310302734375,
      "grad_norm": 10.144874572753906,
      "learning_rate": 4.933885637064489e-06,
      "loss": 3.7662,
      "step": 7305
    },
    {
      "epoch": 0.07436116536458333,
      "grad_norm": 15.959395408630371,
      "learning_rate": 4.933794250756022e-06,
      "loss": 3.2475,
      "step": 7310
    },
    {
      "epoch": 0.07441202799479167,
      "grad_norm": 15.771014213562012,
      "learning_rate": 4.933702802179506e-06,
      "loss": 3.5408,
      "step": 7315
    },
    {
      "epoch": 0.074462890625,
      "grad_norm": 16.440584182739258,
      "learning_rate": 4.933611291337282e-06,
      "loss": 3.5703,
      "step": 7320
    },
    {
      "epoch": 0.07451375325520833,
      "grad_norm": 13.77549934387207,
      "learning_rate": 4.933519718231689e-06,
      "loss": 3.7564,
      "step": 7325
    },
    {
      "epoch": 0.07456461588541667,
      "grad_norm": 13.803625106811523,
      "learning_rate": 4.9334280828650714e-06,
      "loss": 3.9605,
      "step": 7330
    },
    {
      "epoch": 0.074615478515625,
      "grad_norm": 16.266210556030273,
      "learning_rate": 4.933336385239772e-06,
      "loss": 3.6834,
      "step": 7335
    },
    {
      "epoch": 0.07466634114583333,
      "grad_norm": 12.476020812988281,
      "learning_rate": 4.933244625358139e-06,
      "loss": 3.2462,
      "step": 7340
    },
    {
      "epoch": 0.07471720377604167,
      "grad_norm": 14.898695945739746,
      "learning_rate": 4.9331528032225186e-06,
      "loss": 4.0511,
      "step": 7345
    },
    {
      "epoch": 0.07476806640625,
      "grad_norm": 13.865912437438965,
      "learning_rate": 4.933060918835261e-06,
      "loss": 3.5242,
      "step": 7350
    },
    {
      "epoch": 0.07481892903645833,
      "grad_norm": 15.214755058288574,
      "learning_rate": 4.932968972198715e-06,
      "loss": 3.6132,
      "step": 7355
    },
    {
      "epoch": 0.07486979166666667,
      "grad_norm": 14.871356010437012,
      "learning_rate": 4.932876963315236e-06,
      "loss": 3.3851,
      "step": 7360
    },
    {
      "epoch": 0.074920654296875,
      "grad_norm": 12.997641563415527,
      "learning_rate": 4.932784892187176e-06,
      "loss": 3.1629,
      "step": 7365
    },
    {
      "epoch": 0.07497151692708333,
      "grad_norm": 10.09203052520752,
      "learning_rate": 4.932692758816892e-06,
      "loss": 3.3311,
      "step": 7370
    },
    {
      "epoch": 0.07502237955729167,
      "grad_norm": 8.960355758666992,
      "learning_rate": 4.932600563206739e-06,
      "loss": 3.1715,
      "step": 7375
    },
    {
      "epoch": 0.0750732421875,
      "grad_norm": 16.078197479248047,
      "learning_rate": 4.932508305359078e-06,
      "loss": 3.5975,
      "step": 7380
    },
    {
      "epoch": 0.07512410481770833,
      "grad_norm": 15.18477725982666,
      "learning_rate": 4.9324159852762685e-06,
      "loss": 3.341,
      "step": 7385
    },
    {
      "epoch": 0.07517496744791667,
      "grad_norm": 15.12987995147705,
      "learning_rate": 4.932323602960673e-06,
      "loss": 3.5689,
      "step": 7390
    },
    {
      "epoch": 0.075225830078125,
      "grad_norm": 13.448582649230957,
      "learning_rate": 4.932231158414653e-06,
      "loss": 3.6534,
      "step": 7395
    },
    {
      "epoch": 0.07527669270833333,
      "grad_norm": 9.764447212219238,
      "learning_rate": 4.932138651640577e-06,
      "loss": 3.5477,
      "step": 7400
    },
    {
      "epoch": 0.07532755533854167,
      "grad_norm": 11.549277305603027,
      "learning_rate": 4.932046082640809e-06,
      "loss": 3.0506,
      "step": 7405
    },
    {
      "epoch": 0.07537841796875,
      "grad_norm": 9.322796821594238,
      "learning_rate": 4.9319534514177196e-06,
      "loss": 3.2946,
      "step": 7410
    },
    {
      "epoch": 0.07542928059895833,
      "grad_norm": 10.205277442932129,
      "learning_rate": 4.931860757973676e-06,
      "loss": 3.3928,
      "step": 7415
    },
    {
      "epoch": 0.07548014322916667,
      "grad_norm": 8.97710132598877,
      "learning_rate": 4.931768002311052e-06,
      "loss": 3.5807,
      "step": 7420
    },
    {
      "epoch": 0.075531005859375,
      "grad_norm": 15.988985061645508,
      "learning_rate": 4.931675184432221e-06,
      "loss": 3.5203,
      "step": 7425
    },
    {
      "epoch": 0.07558186848958333,
      "grad_norm": 12.848153114318848,
      "learning_rate": 4.931582304339556e-06,
      "loss": 3.1023,
      "step": 7430
    },
    {
      "epoch": 0.07563273111979167,
      "grad_norm": 14.775106430053711,
      "learning_rate": 4.931489362035434e-06,
      "loss": 3.6487,
      "step": 7435
    },
    {
      "epoch": 0.07568359375,
      "grad_norm": 15.638808250427246,
      "learning_rate": 4.931396357522233e-06,
      "loss": 3.4362,
      "step": 7440
    },
    {
      "epoch": 0.07573445638020833,
      "grad_norm": 17.059715270996094,
      "learning_rate": 4.931303290802333e-06,
      "loss": 4.0533,
      "step": 7445
    },
    {
      "epoch": 0.07578531901041667,
      "grad_norm": 13.491350173950195,
      "learning_rate": 4.931210161878114e-06,
      "loss": 3.1253,
      "step": 7450
    },
    {
      "epoch": 0.075836181640625,
      "grad_norm": 11.535757064819336,
      "learning_rate": 4.93111697075196e-06,
      "loss": 3.4529,
      "step": 7455
    },
    {
      "epoch": 0.07588704427083333,
      "grad_norm": 8.205851554870605,
      "learning_rate": 4.9310237174262535e-06,
      "loss": 3.3011,
      "step": 7460
    },
    {
      "epoch": 0.07593790690104167,
      "grad_norm": 10.262289047241211,
      "learning_rate": 4.930930401903382e-06,
      "loss": 3.2739,
      "step": 7465
    },
    {
      "epoch": 0.07598876953125,
      "grad_norm": 11.151719093322754,
      "learning_rate": 4.930837024185732e-06,
      "loss": 3.6304,
      "step": 7470
    },
    {
      "epoch": 0.07603963216145833,
      "grad_norm": 13.860788345336914,
      "learning_rate": 4.930743584275694e-06,
      "loss": 3.3064,
      "step": 7475
    },
    {
      "epoch": 0.07609049479166667,
      "grad_norm": 12.631816864013672,
      "learning_rate": 4.930650082175656e-06,
      "loss": 3.3838,
      "step": 7480
    },
    {
      "epoch": 0.076141357421875,
      "grad_norm": 15.496430397033691,
      "learning_rate": 4.930556517888013e-06,
      "loss": 3.3085,
      "step": 7485
    },
    {
      "epoch": 0.07619222005208333,
      "grad_norm": 15.001185417175293,
      "learning_rate": 4.930462891415156e-06,
      "loss": 3.1146,
      "step": 7490
    },
    {
      "epoch": 0.07624308268229167,
      "grad_norm": 9.253118515014648,
      "learning_rate": 4.930369202759484e-06,
      "loss": 3.5206,
      "step": 7495
    },
    {
      "epoch": 0.0762939453125,
      "grad_norm": 9.574434280395508,
      "learning_rate": 4.9302754519233905e-06,
      "loss": 3.0834,
      "step": 7500
    },
    {
      "epoch": 0.07634480794270833,
      "grad_norm": 8.028079986572266,
      "learning_rate": 4.9301816389092775e-06,
      "loss": 3.3453,
      "step": 7505
    },
    {
      "epoch": 0.07639567057291667,
      "grad_norm": 15.80912971496582,
      "learning_rate": 4.930087763719541e-06,
      "loss": 3.3046,
      "step": 7510
    },
    {
      "epoch": 0.076446533203125,
      "grad_norm": 12.27437686920166,
      "learning_rate": 4.929993826356586e-06,
      "loss": 3.3498,
      "step": 7515
    },
    {
      "epoch": 0.07649739583333333,
      "grad_norm": 8.595794677734375,
      "learning_rate": 4.9298998268228154e-06,
      "loss": 3.4443,
      "step": 7520
    },
    {
      "epoch": 0.07654825846354167,
      "grad_norm": 14.03144645690918,
      "learning_rate": 4.929805765120633e-06,
      "loss": 3.4373,
      "step": 7525
    },
    {
      "epoch": 0.07659912109375,
      "grad_norm": 16.364065170288086,
      "learning_rate": 4.929711641252446e-06,
      "loss": 3.3485,
      "step": 7530
    },
    {
      "epoch": 0.07664998372395833,
      "grad_norm": 13.637460708618164,
      "learning_rate": 4.929617455220664e-06,
      "loss": 3.1559,
      "step": 7535
    },
    {
      "epoch": 0.07670084635416667,
      "grad_norm": 9.677896499633789,
      "learning_rate": 4.929523207027693e-06,
      "loss": 3.1611,
      "step": 7540
    },
    {
      "epoch": 0.076751708984375,
      "grad_norm": 11.03842830657959,
      "learning_rate": 4.929428896675949e-06,
      "loss": 3.468,
      "step": 7545
    },
    {
      "epoch": 0.07680257161458333,
      "grad_norm": 8.96399974822998,
      "learning_rate": 4.92933452416784e-06,
      "loss": 3.2743,
      "step": 7550
    },
    {
      "epoch": 0.07685343424479167,
      "grad_norm": 14.87559986114502,
      "learning_rate": 4.929240089505785e-06,
      "loss": 3.4153,
      "step": 7555
    },
    {
      "epoch": 0.076904296875,
      "grad_norm": 16.70171356201172,
      "learning_rate": 4.929145592692197e-06,
      "loss": 3.3959,
      "step": 7560
    },
    {
      "epoch": 0.07695515950520833,
      "grad_norm": 10.184401512145996,
      "learning_rate": 4.929051033729495e-06,
      "loss": 3.3099,
      "step": 7565
    },
    {
      "epoch": 0.07700602213541667,
      "grad_norm": 16.423912048339844,
      "learning_rate": 4.928956412620098e-06,
      "loss": 3.5852,
      "step": 7570
    },
    {
      "epoch": 0.077056884765625,
      "grad_norm": 12.25012493133545,
      "learning_rate": 4.928861729366427e-06,
      "loss": 3.3323,
      "step": 7575
    },
    {
      "epoch": 0.07710774739583333,
      "grad_norm": 14.227832794189453,
      "learning_rate": 4.928766983970905e-06,
      "loss": 3.7213,
      "step": 7580
    },
    {
      "epoch": 0.07715861002604167,
      "grad_norm": 9.725828170776367,
      "learning_rate": 4.928672176435955e-06,
      "loss": 3.8787,
      "step": 7585
    },
    {
      "epoch": 0.07720947265625,
      "grad_norm": 13.200632095336914,
      "learning_rate": 4.928577306764003e-06,
      "loss": 3.3488,
      "step": 7590
    },
    {
      "epoch": 0.07726033528645833,
      "grad_norm": 16.116952896118164,
      "learning_rate": 4.928482374957476e-06,
      "loss": 3.5104,
      "step": 7595
    },
    {
      "epoch": 0.07731119791666667,
      "grad_norm": 14.31235408782959,
      "learning_rate": 4.928387381018803e-06,
      "loss": 3.4802,
      "step": 7600
    },
    {
      "epoch": 0.077362060546875,
      "grad_norm": 14.306567192077637,
      "learning_rate": 4.928292324950415e-06,
      "loss": 3.1986,
      "step": 7605
    },
    {
      "epoch": 0.07741292317708333,
      "grad_norm": 10.679328918457031,
      "learning_rate": 4.9281972067547435e-06,
      "loss": 3.4726,
      "step": 7610
    },
    {
      "epoch": 0.07746378580729167,
      "grad_norm": 13.975683212280273,
      "learning_rate": 4.928102026434221e-06,
      "loss": 3.5155,
      "step": 7615
    },
    {
      "epoch": 0.0775146484375,
      "grad_norm": 11.450372695922852,
      "learning_rate": 4.928006783991285e-06,
      "loss": 3.26,
      "step": 7620
    },
    {
      "epoch": 0.07756551106770833,
      "grad_norm": 9.844697952270508,
      "learning_rate": 4.92791147942837e-06,
      "loss": 3.552,
      "step": 7625
    },
    {
      "epoch": 0.07761637369791667,
      "grad_norm": 13.510926246643066,
      "learning_rate": 4.927816112747915e-06,
      "loss": 3.2648,
      "step": 7630
    },
    {
      "epoch": 0.077667236328125,
      "grad_norm": 16.180030822753906,
      "learning_rate": 4.927720683952361e-06,
      "loss": 3.8344,
      "step": 7635
    },
    {
      "epoch": 0.07771809895833333,
      "grad_norm": 10.086830139160156,
      "learning_rate": 4.9276251930441485e-06,
      "loss": 3.5012,
      "step": 7640
    },
    {
      "epoch": 0.07776896158854167,
      "grad_norm": 16.06332778930664,
      "learning_rate": 4.927529640025721e-06,
      "loss": 3.5488,
      "step": 7645
    },
    {
      "epoch": 0.07781982421875,
      "grad_norm": 9.544757843017578,
      "learning_rate": 4.927434024899522e-06,
      "loss": 3.3311,
      "step": 7650
    },
    {
      "epoch": 0.07787068684895833,
      "grad_norm": 10.502907752990723,
      "learning_rate": 4.927338347668e-06,
      "loss": 3.7151,
      "step": 7655
    },
    {
      "epoch": 0.07792154947916667,
      "grad_norm": 21.776485443115234,
      "learning_rate": 4.927242608333601e-06,
      "loss": 3.6689,
      "step": 7660
    },
    {
      "epoch": 0.077972412109375,
      "grad_norm": 9.892070770263672,
      "learning_rate": 4.927146806898776e-06,
      "loss": 3.4572,
      "step": 7665
    },
    {
      "epoch": 0.07802327473958333,
      "grad_norm": 10.099709510803223,
      "learning_rate": 4.927050943365974e-06,
      "loss": 3.1367,
      "step": 7670
    },
    {
      "epoch": 0.07807413736979167,
      "grad_norm": 11.609224319458008,
      "learning_rate": 4.92695501773765e-06,
      "loss": 3.6545,
      "step": 7675
    },
    {
      "epoch": 0.078125,
      "grad_norm": 12.737848281860352,
      "learning_rate": 4.926859030016257e-06,
      "loss": 3.4818,
      "step": 7680
    },
    {
      "epoch": 0.07817586263020833,
      "grad_norm": 11.716086387634277,
      "learning_rate": 4.926762980204251e-06,
      "loss": 3.3399,
      "step": 7685
    },
    {
      "epoch": 0.07822672526041667,
      "grad_norm": 15.350922584533691,
      "learning_rate": 4.926666868304089e-06,
      "loss": 3.2643,
      "step": 7690
    },
    {
      "epoch": 0.078277587890625,
      "grad_norm": 12.929101943969727,
      "learning_rate": 4.9265706943182305e-06,
      "loss": 3.1085,
      "step": 7695
    },
    {
      "epoch": 0.07832845052083333,
      "grad_norm": 11.21131706237793,
      "learning_rate": 4.926474458249137e-06,
      "loss": 3.3425,
      "step": 7700
    },
    {
      "epoch": 0.07837931315104167,
      "grad_norm": 16.155336380004883,
      "learning_rate": 4.9263781600992675e-06,
      "loss": 3.7774,
      "step": 7705
    },
    {
      "epoch": 0.07843017578125,
      "grad_norm": 16.52438735961914,
      "learning_rate": 4.926281799871089e-06,
      "loss": 3.6957,
      "step": 7710
    },
    {
      "epoch": 0.07848103841145833,
      "grad_norm": 15.890446662902832,
      "learning_rate": 4.926185377567065e-06,
      "loss": 3.5257,
      "step": 7715
    },
    {
      "epoch": 0.07853190104166667,
      "grad_norm": 12.618369102478027,
      "learning_rate": 4.926088893189665e-06,
      "loss": 3.5028,
      "step": 7720
    },
    {
      "epoch": 0.078582763671875,
      "grad_norm": 8.554431915283203,
      "learning_rate": 4.925992346741354e-06,
      "loss": 3.0891,
      "step": 7725
    },
    {
      "epoch": 0.07863362630208333,
      "grad_norm": 15.57974624633789,
      "learning_rate": 4.9258957382246045e-06,
      "loss": 3.4668,
      "step": 7730
    },
    {
      "epoch": 0.07868448893229167,
      "grad_norm": 13.180612564086914,
      "learning_rate": 4.925799067641888e-06,
      "loss": 3.6996,
      "step": 7735
    },
    {
      "epoch": 0.0787353515625,
      "grad_norm": 10.547351837158203,
      "learning_rate": 4.9257023349956765e-06,
      "loss": 3.8278,
      "step": 7740
    },
    {
      "epoch": 0.07878621419270833,
      "grad_norm": 18.47353744506836,
      "learning_rate": 4.925605540288445e-06,
      "loss": 3.5644,
      "step": 7745
    },
    {
      "epoch": 0.07883707682291667,
      "grad_norm": 14.555137634277344,
      "learning_rate": 4.925508683522673e-06,
      "loss": 3.3353,
      "step": 7750
    },
    {
      "epoch": 0.078887939453125,
      "grad_norm": 14.03541088104248,
      "learning_rate": 4.925411764700834e-06,
      "loss": 3.3763,
      "step": 7755
    },
    {
      "epoch": 0.07893880208333333,
      "grad_norm": 13.177834510803223,
      "learning_rate": 4.925314783825411e-06,
      "loss": 3.6113,
      "step": 7760
    },
    {
      "epoch": 0.07898966471354167,
      "grad_norm": 12.112380027770996,
      "learning_rate": 4.925217740898884e-06,
      "loss": 3.3866,
      "step": 7765
    },
    {
      "epoch": 0.07904052734375,
      "grad_norm": 12.616769790649414,
      "learning_rate": 4.925120635923736e-06,
      "loss": 3.7202,
      "step": 7770
    },
    {
      "epoch": 0.07909138997395833,
      "grad_norm": 12.70201587677002,
      "learning_rate": 4.925023468902451e-06,
      "loss": 3.316,
      "step": 7775
    },
    {
      "epoch": 0.07914225260416667,
      "grad_norm": 15.856217384338379,
      "learning_rate": 4.924926239837515e-06,
      "loss": 3.36,
      "step": 7780
    },
    {
      "epoch": 0.079193115234375,
      "grad_norm": 12.738685607910156,
      "learning_rate": 4.9248289487314174e-06,
      "loss": 3.3463,
      "step": 7785
    },
    {
      "epoch": 0.07924397786458333,
      "grad_norm": 15.38477897644043,
      "learning_rate": 4.924731595586645e-06,
      "loss": 3.3254,
      "step": 7790
    },
    {
      "epoch": 0.07929484049479167,
      "grad_norm": 12.949041366577148,
      "learning_rate": 4.924634180405689e-06,
      "loss": 3.2682,
      "step": 7795
    },
    {
      "epoch": 0.079345703125,
      "grad_norm": 17.46603775024414,
      "learning_rate": 4.924536703191043e-06,
      "loss": 3.5482,
      "step": 7800
    },
    {
      "epoch": 0.07939656575520833,
      "grad_norm": 15.399724960327148,
      "learning_rate": 4.9244391639451995e-06,
      "loss": 3.5545,
      "step": 7805
    },
    {
      "epoch": 0.07944742838541667,
      "grad_norm": 13.765381813049316,
      "learning_rate": 4.924341562670655e-06,
      "loss": 3.2777,
      "step": 7810
    },
    {
      "epoch": 0.079498291015625,
      "grad_norm": 13.219882011413574,
      "learning_rate": 4.924243899369906e-06,
      "loss": 3.4312,
      "step": 7815
    },
    {
      "epoch": 0.07954915364583333,
      "grad_norm": 15.765204429626465,
      "learning_rate": 4.924146174045451e-06,
      "loss": 3.4005,
      "step": 7820
    },
    {
      "epoch": 0.07960001627604167,
      "grad_norm": 14.722639083862305,
      "learning_rate": 4.924048386699792e-06,
      "loss": 3.4344,
      "step": 7825
    },
    {
      "epoch": 0.07965087890625,
      "grad_norm": 14.24759292602539,
      "learning_rate": 4.923950537335429e-06,
      "loss": 3.5187,
      "step": 7830
    },
    {
      "epoch": 0.07970174153645833,
      "grad_norm": 14.9491548538208,
      "learning_rate": 4.923852625954866e-06,
      "loss": 3.3459,
      "step": 7835
    },
    {
      "epoch": 0.07975260416666667,
      "grad_norm": 10.572219848632812,
      "learning_rate": 4.9237546525606075e-06,
      "loss": 3.6514,
      "step": 7840
    },
    {
      "epoch": 0.079803466796875,
      "grad_norm": 13.39721393585205,
      "learning_rate": 4.923656617155162e-06,
      "loss": 3.4159,
      "step": 7845
    },
    {
      "epoch": 0.07985432942708333,
      "grad_norm": 13.185128211975098,
      "learning_rate": 4.923558519741035e-06,
      "loss": 3.3885,
      "step": 7850
    },
    {
      "epoch": 0.07990519205729167,
      "grad_norm": 10.458941459655762,
      "learning_rate": 4.923460360320738e-06,
      "loss": 3.8603,
      "step": 7855
    },
    {
      "epoch": 0.0799560546875,
      "grad_norm": 16.00554084777832,
      "learning_rate": 4.923362138896782e-06,
      "loss": 3.8007,
      "step": 7860
    },
    {
      "epoch": 0.08000691731770833,
      "grad_norm": 19.089359283447266,
      "learning_rate": 4.923263855471681e-06,
      "loss": 3.4031,
      "step": 7865
    },
    {
      "epoch": 0.08005777994791667,
      "grad_norm": 16.948728561401367,
      "learning_rate": 4.923165510047948e-06,
      "loss": 3.4621,
      "step": 7870
    },
    {
      "epoch": 0.080108642578125,
      "grad_norm": 15.898715019226074,
      "learning_rate": 4.9230671026281e-06,
      "loss": 3.2493,
      "step": 7875
    },
    {
      "epoch": 0.08015950520833333,
      "grad_norm": 15.451108932495117,
      "learning_rate": 4.922968633214654e-06,
      "loss": 3.5485,
      "step": 7880
    },
    {
      "epoch": 0.08021036783854167,
      "grad_norm": 8.44896411895752,
      "learning_rate": 4.922870101810131e-06,
      "loss": 3.6708,
      "step": 7885
    },
    {
      "epoch": 0.08026123046875,
      "grad_norm": 15.81263542175293,
      "learning_rate": 4.92277150841705e-06,
      "loss": 3.0769,
      "step": 7890
    },
    {
      "epoch": 0.08031209309895833,
      "grad_norm": 11.589963912963867,
      "learning_rate": 4.922672853037934e-06,
      "loss": 3.1524,
      "step": 7895
    },
    {
      "epoch": 0.08036295572916667,
      "grad_norm": 11.018848419189453,
      "learning_rate": 4.922574135675308e-06,
      "loss": 3.4289,
      "step": 7900
    },
    {
      "epoch": 0.080413818359375,
      "grad_norm": 14.702198028564453,
      "learning_rate": 4.922475356331696e-06,
      "loss": 3.7443,
      "step": 7905
    },
    {
      "epoch": 0.08046468098958333,
      "grad_norm": 7.636882305145264,
      "learning_rate": 4.922376515009627e-06,
      "loss": 3.2372,
      "step": 7910
    },
    {
      "epoch": 0.08051554361979167,
      "grad_norm": 14.174872398376465,
      "learning_rate": 4.922277611711629e-06,
      "loss": 3.2774,
      "step": 7915
    },
    {
      "epoch": 0.08056640625,
      "grad_norm": 7.712305068969727,
      "learning_rate": 4.922178646440232e-06,
      "loss": 2.9999,
      "step": 7920
    },
    {
      "epoch": 0.08061726888020833,
      "grad_norm": 13.299323081970215,
      "learning_rate": 4.922079619197968e-06,
      "loss": 3.4508,
      "step": 7925
    },
    {
      "epoch": 0.08066813151041667,
      "grad_norm": 12.260127067565918,
      "learning_rate": 4.9219805299873715e-06,
      "loss": 3.5971,
      "step": 7930
    },
    {
      "epoch": 0.080718994140625,
      "grad_norm": 11.661309242248535,
      "learning_rate": 4.9218813788109776e-06,
      "loss": 3.2826,
      "step": 7935
    },
    {
      "epoch": 0.08076985677083333,
      "grad_norm": 14.099932670593262,
      "learning_rate": 4.921782165671322e-06,
      "loss": 3.2441,
      "step": 7940
    },
    {
      "epoch": 0.08082071940104167,
      "grad_norm": 11.960383415222168,
      "learning_rate": 4.9216828905709445e-06,
      "loss": 3.2706,
      "step": 7945
    },
    {
      "epoch": 0.08087158203125,
      "grad_norm": 14.964332580566406,
      "learning_rate": 4.921583553512384e-06,
      "loss": 2.997,
      "step": 7950
    },
    {
      "epoch": 0.08092244466145833,
      "grad_norm": 11.150708198547363,
      "learning_rate": 4.9214841544981826e-06,
      "loss": 3.3165,
      "step": 7955
    },
    {
      "epoch": 0.08097330729166667,
      "grad_norm": 9.034783363342285,
      "learning_rate": 4.9213846935308816e-06,
      "loss": 3.4451,
      "step": 7960
    },
    {
      "epoch": 0.081024169921875,
      "grad_norm": 12.717357635498047,
      "learning_rate": 4.921285170613029e-06,
      "loss": 3.4678,
      "step": 7965
    },
    {
      "epoch": 0.08107503255208333,
      "grad_norm": 15.22363567352295,
      "learning_rate": 4.921185585747168e-06,
      "loss": 3.3836,
      "step": 7970
    },
    {
      "epoch": 0.08112589518229167,
      "grad_norm": 12.619773864746094,
      "learning_rate": 4.9210859389358475e-06,
      "loss": 3.3497,
      "step": 7975
    },
    {
      "epoch": 0.0811767578125,
      "grad_norm": 10.314453125,
      "learning_rate": 4.920986230181618e-06,
      "loss": 3.3937,
      "step": 7980
    },
    {
      "epoch": 0.08122762044270833,
      "grad_norm": 16.396963119506836,
      "learning_rate": 4.920886459487029e-06,
      "loss": 3.2124,
      "step": 7985
    },
    {
      "epoch": 0.08127848307291667,
      "grad_norm": 10.935754776000977,
      "learning_rate": 4.920786626854634e-06,
      "loss": 3.3134,
      "step": 7990
    },
    {
      "epoch": 0.081329345703125,
      "grad_norm": 11.221820831298828,
      "learning_rate": 4.920686732286988e-06,
      "loss": 3.3703,
      "step": 7995
    },
    {
      "epoch": 0.08138020833333333,
      "grad_norm": 15.557806968688965,
      "learning_rate": 4.9205867757866445e-06,
      "loss": 3.2063,
      "step": 8000
    },
    {
      "epoch": 0.08143107096354167,
      "grad_norm": 14.617751121520996,
      "learning_rate": 4.920486757356162e-06,
      "loss": 3.4361,
      "step": 8005
    },
    {
      "epoch": 0.08148193359375,
      "grad_norm": 7.648261070251465,
      "learning_rate": 4.9203866769981e-06,
      "loss": 3.1502,
      "step": 8010
    },
    {
      "epoch": 0.08153279622395833,
      "grad_norm": 13.730716705322266,
      "learning_rate": 4.920286534715018e-06,
      "loss": 3.7119,
      "step": 8015
    },
    {
      "epoch": 0.08158365885416667,
      "grad_norm": 14.1985502243042,
      "learning_rate": 4.9201863305094786e-06,
      "loss": 3.4887,
      "step": 8020
    },
    {
      "epoch": 0.081634521484375,
      "grad_norm": 13.51031494140625,
      "learning_rate": 4.920086064384046e-06,
      "loss": 3.0307,
      "step": 8025
    },
    {
      "epoch": 0.08168538411458333,
      "grad_norm": 11.816555976867676,
      "learning_rate": 4.919985736341286e-06,
      "loss": 3.4726,
      "step": 8030
    },
    {
      "epoch": 0.08173624674479167,
      "grad_norm": 15.351333618164062,
      "learning_rate": 4.919885346383764e-06,
      "loss": 3.6725,
      "step": 8035
    },
    {
      "epoch": 0.081787109375,
      "grad_norm": 17.565916061401367,
      "learning_rate": 4.919784894514048e-06,
      "loss": 3.551,
      "step": 8040
    },
    {
      "epoch": 0.08183797200520833,
      "grad_norm": 11.794846534729004,
      "learning_rate": 4.91968438073471e-06,
      "loss": 3.7001,
      "step": 8045
    },
    {
      "epoch": 0.08188883463541667,
      "grad_norm": 13.469988822937012,
      "learning_rate": 4.919583805048321e-06,
      "loss": 3.2706,
      "step": 8050
    },
    {
      "epoch": 0.081939697265625,
      "grad_norm": 9.399026870727539,
      "learning_rate": 4.919483167457452e-06,
      "loss": 3.476,
      "step": 8055
    },
    {
      "epoch": 0.08199055989583333,
      "grad_norm": 10.455305099487305,
      "learning_rate": 4.919382467964681e-06,
      "loss": 3.78,
      "step": 8060
    },
    {
      "epoch": 0.08204142252604167,
      "grad_norm": 13.494087219238281,
      "learning_rate": 4.919281706572583e-06,
      "loss": 3.5696,
      "step": 8065
    },
    {
      "epoch": 0.08209228515625,
      "grad_norm": 9.601789474487305,
      "learning_rate": 4.919180883283735e-06,
      "loss": 3.4697,
      "step": 8070
    },
    {
      "epoch": 0.08214314778645833,
      "grad_norm": 8.843679428100586,
      "learning_rate": 4.919079998100719e-06,
      "loss": 3.3273,
      "step": 8075
    },
    {
      "epoch": 0.08219401041666667,
      "grad_norm": 17.03862953186035,
      "learning_rate": 4.918979051026113e-06,
      "loss": 3.3333,
      "step": 8080
    },
    {
      "epoch": 0.082244873046875,
      "grad_norm": 9.111615180969238,
      "learning_rate": 4.918878042062503e-06,
      "loss": 3.1043,
      "step": 8085
    },
    {
      "epoch": 0.08229573567708333,
      "grad_norm": 9.032851219177246,
      "learning_rate": 4.918776971212471e-06,
      "loss": 3.1896,
      "step": 8090
    },
    {
      "epoch": 0.08234659830729167,
      "grad_norm": 15.128028869628906,
      "learning_rate": 4.918675838478603e-06,
      "loss": 3.5589,
      "step": 8095
    },
    {
      "epoch": 0.0823974609375,
      "grad_norm": 9.789177894592285,
      "learning_rate": 4.918574643863488e-06,
      "loss": 3.4767,
      "step": 8100
    },
    {
      "epoch": 0.08244832356770833,
      "grad_norm": 8.648837089538574,
      "learning_rate": 4.918473387369713e-06,
      "loss": 3.6547,
      "step": 8105
    },
    {
      "epoch": 0.08249918619791667,
      "grad_norm": 12.925644874572754,
      "learning_rate": 4.91837206899987e-06,
      "loss": 3.4412,
      "step": 8110
    },
    {
      "epoch": 0.082550048828125,
      "grad_norm": 10.411079406738281,
      "learning_rate": 4.918270688756551e-06,
      "loss": 3.501,
      "step": 8115
    },
    {
      "epoch": 0.08260091145833333,
      "grad_norm": 16.97960662841797,
      "learning_rate": 4.918169246642349e-06,
      "loss": 3.9448,
      "step": 8120
    },
    {
      "epoch": 0.08265177408854167,
      "grad_norm": 13.027390480041504,
      "learning_rate": 4.91806774265986e-06,
      "loss": 3.2829,
      "step": 8125
    },
    {
      "epoch": 0.08270263671875,
      "grad_norm": 11.88217544555664,
      "learning_rate": 4.9179661768116815e-06,
      "loss": 3.6066,
      "step": 8130
    },
    {
      "epoch": 0.08275349934895833,
      "grad_norm": 9.799572944641113,
      "learning_rate": 4.9178645491004115e-06,
      "loss": 3.2738,
      "step": 8135
    },
    {
      "epoch": 0.08280436197916667,
      "grad_norm": 12.099909782409668,
      "learning_rate": 4.91776285952865e-06,
      "loss": 3.2961,
      "step": 8140
    },
    {
      "epoch": 0.082855224609375,
      "grad_norm": 11.036294937133789,
      "learning_rate": 4.917661108098999e-06,
      "loss": 3.5535,
      "step": 8145
    },
    {
      "epoch": 0.08290608723958333,
      "grad_norm": 16.11235237121582,
      "learning_rate": 4.9175592948140614e-06,
      "loss": 3.5957,
      "step": 8150
    },
    {
      "epoch": 0.08295694986979167,
      "grad_norm": 12.630952835083008,
      "learning_rate": 4.917457419676443e-06,
      "loss": 3.1787,
      "step": 8155
    },
    {
      "epoch": 0.0830078125,
      "grad_norm": 10.421707153320312,
      "learning_rate": 4.9173554826887485e-06,
      "loss": 3.5609,
      "step": 8160
    },
    {
      "epoch": 0.08305867513020833,
      "grad_norm": 14.865351676940918,
      "learning_rate": 4.917253483853587e-06,
      "loss": 3.4347,
      "step": 8165
    },
    {
      "epoch": 0.08310953776041667,
      "grad_norm": 10.349051475524902,
      "learning_rate": 4.917151423173568e-06,
      "loss": 3.3103,
      "step": 8170
    },
    {
      "epoch": 0.083160400390625,
      "grad_norm": 12.763856887817383,
      "learning_rate": 4.917049300651303e-06,
      "loss": 3.2145,
      "step": 8175
    },
    {
      "epoch": 0.08321126302083333,
      "grad_norm": 13.702238082885742,
      "learning_rate": 4.916947116289405e-06,
      "loss": 3.4389,
      "step": 8180
    },
    {
      "epoch": 0.08326212565104167,
      "grad_norm": 8.125384330749512,
      "learning_rate": 4.916844870090487e-06,
      "loss": 3.1778,
      "step": 8185
    },
    {
      "epoch": 0.08331298828125,
      "grad_norm": 13.532679557800293,
      "learning_rate": 4.916742562057166e-06,
      "loss": 3.5005,
      "step": 8190
    },
    {
      "epoch": 0.08336385091145833,
      "grad_norm": 11.276693344116211,
      "learning_rate": 4.91664019219206e-06,
      "loss": 3.1357,
      "step": 8195
    },
    {
      "epoch": 0.08341471354166667,
      "grad_norm": 10.653390884399414,
      "learning_rate": 4.916537760497787e-06,
      "loss": 2.9572,
      "step": 8200
    },
    {
      "epoch": 0.083465576171875,
      "grad_norm": 19.044883728027344,
      "learning_rate": 4.9164352669769685e-06,
      "loss": 3.4532,
      "step": 8205
    },
    {
      "epoch": 0.08351643880208333,
      "grad_norm": 9.898783683776855,
      "learning_rate": 4.916332711632227e-06,
      "loss": 3.3115,
      "step": 8210
    },
    {
      "epoch": 0.08356730143229167,
      "grad_norm": 7.111556053161621,
      "learning_rate": 4.916230094466185e-06,
      "loss": 3.5531,
      "step": 8215
    },
    {
      "epoch": 0.0836181640625,
      "grad_norm": 15.385550498962402,
      "learning_rate": 4.916127415481469e-06,
      "loss": 3.5578,
      "step": 8220
    },
    {
      "epoch": 0.08366902669270833,
      "grad_norm": 9.553897857666016,
      "learning_rate": 4.916024674680705e-06,
      "loss": 3.4616,
      "step": 8225
    },
    {
      "epoch": 0.08371988932291667,
      "grad_norm": 9.77919864654541,
      "learning_rate": 4.915921872066524e-06,
      "loss": 3.5612,
      "step": 8230
    },
    {
      "epoch": 0.083770751953125,
      "grad_norm": 13.859053611755371,
      "learning_rate": 4.915819007641553e-06,
      "loss": 3.3759,
      "step": 8235
    },
    {
      "epoch": 0.08382161458333333,
      "grad_norm": 13.282938003540039,
      "learning_rate": 4.915716081408426e-06,
      "loss": 3.1999,
      "step": 8240
    },
    {
      "epoch": 0.08387247721354167,
      "grad_norm": 8.742036819458008,
      "learning_rate": 4.9156130933697756e-06,
      "loss": 3.6131,
      "step": 8245
    },
    {
      "epoch": 0.08392333984375,
      "grad_norm": 10.683076858520508,
      "learning_rate": 4.915510043528237e-06,
      "loss": 3.6148,
      "step": 8250
    },
    {
      "epoch": 0.08397420247395833,
      "grad_norm": 13.600173950195312,
      "learning_rate": 4.915406931886446e-06,
      "loss": 3.0754,
      "step": 8255
    },
    {
      "epoch": 0.08402506510416667,
      "grad_norm": 9.838703155517578,
      "learning_rate": 4.915303758447041e-06,
      "loss": 3.422,
      "step": 8260
    },
    {
      "epoch": 0.084075927734375,
      "grad_norm": 17.08401107788086,
      "learning_rate": 4.915200523212662e-06,
      "loss": 3.5272,
      "step": 8265
    },
    {
      "epoch": 0.08412679036458333,
      "grad_norm": 11.889626502990723,
      "learning_rate": 4.91509722618595e-06,
      "loss": 3.244,
      "step": 8270
    },
    {
      "epoch": 0.08417765299479167,
      "grad_norm": 13.750388145446777,
      "learning_rate": 4.914993867369549e-06,
      "loss": 3.0886,
      "step": 8275
    },
    {
      "epoch": 0.084228515625,
      "grad_norm": 12.008491516113281,
      "learning_rate": 4.914890446766101e-06,
      "loss": 3.4445,
      "step": 8280
    },
    {
      "epoch": 0.08427937825520833,
      "grad_norm": 8.624137878417969,
      "learning_rate": 4.914786964378253e-06,
      "loss": 3.1456,
      "step": 8285
    },
    {
      "epoch": 0.08433024088541667,
      "grad_norm": 13.373948097229004,
      "learning_rate": 4.914683420208654e-06,
      "loss": 3.1974,
      "step": 8290
    },
    {
      "epoch": 0.084381103515625,
      "grad_norm": 11.786771774291992,
      "learning_rate": 4.914579814259952e-06,
      "loss": 3.7054,
      "step": 8295
    },
    {
      "epoch": 0.08443196614583333,
      "grad_norm": 12.221219062805176,
      "learning_rate": 4.914476146534797e-06,
      "loss": 3.3905,
      "step": 8300
    },
    {
      "epoch": 0.08448282877604167,
      "grad_norm": 11.495316505432129,
      "learning_rate": 4.914372417035843e-06,
      "loss": 3.9629,
      "step": 8305
    },
    {
      "epoch": 0.08453369140625,
      "grad_norm": 13.219930648803711,
      "learning_rate": 4.914268625765742e-06,
      "loss": 3.359,
      "step": 8310
    },
    {
      "epoch": 0.08458455403645833,
      "grad_norm": 11.233851432800293,
      "learning_rate": 4.9141647727271515e-06,
      "loss": 3.6808,
      "step": 8315
    },
    {
      "epoch": 0.08463541666666667,
      "grad_norm": 10.176496505737305,
      "learning_rate": 4.914060857922727e-06,
      "loss": 3.3135,
      "step": 8320
    },
    {
      "epoch": 0.084686279296875,
      "grad_norm": 10.016979217529297,
      "learning_rate": 4.9139568813551275e-06,
      "loss": 3.8656,
      "step": 8325
    },
    {
      "epoch": 0.08473714192708333,
      "grad_norm": 14.729958534240723,
      "learning_rate": 4.913852843027013e-06,
      "loss": 3.4418,
      "step": 8330
    },
    {
      "epoch": 0.08478800455729167,
      "grad_norm": 16.3724422454834,
      "learning_rate": 4.913748742941046e-06,
      "loss": 3.3672,
      "step": 8335
    },
    {
      "epoch": 0.0848388671875,
      "grad_norm": 9.82769775390625,
      "learning_rate": 4.91364458109989e-06,
      "loss": 3.0876,
      "step": 8340
    },
    {
      "epoch": 0.08488972981770833,
      "grad_norm": 12.202662467956543,
      "learning_rate": 4.913540357506209e-06,
      "loss": 3.3755,
      "step": 8345
    },
    {
      "epoch": 0.08494059244791667,
      "grad_norm": 13.47652530670166,
      "learning_rate": 4.913436072162671e-06,
      "loss": 3.3207,
      "step": 8350
    },
    {
      "epoch": 0.084991455078125,
      "grad_norm": 13.997584342956543,
      "learning_rate": 4.913331725071942e-06,
      "loss": 3.4016,
      "step": 8355
    },
    {
      "epoch": 0.08504231770833333,
      "grad_norm": 10.103306770324707,
      "learning_rate": 4.9132273162366926e-06,
      "loss": 3.9784,
      "step": 8360
    },
    {
      "epoch": 0.08509318033854167,
      "grad_norm": 13.56583023071289,
      "learning_rate": 4.913122845659595e-06,
      "loss": 3.5657,
      "step": 8365
    },
    {
      "epoch": 0.08514404296875,
      "grad_norm": 16.747276306152344,
      "learning_rate": 4.913018313343322e-06,
      "loss": 3.3329,
      "step": 8370
    },
    {
      "epoch": 0.08519490559895833,
      "grad_norm": 15.131979942321777,
      "learning_rate": 4.912913719290546e-06,
      "loss": 3.7669,
      "step": 8375
    },
    {
      "epoch": 0.08524576822916667,
      "grad_norm": 13.406637191772461,
      "learning_rate": 4.912809063503945e-06,
      "loss": 3.8842,
      "step": 8380
    },
    {
      "epoch": 0.085296630859375,
      "grad_norm": 10.718530654907227,
      "learning_rate": 4.912704345986196e-06,
      "loss": 3.3867,
      "step": 8385
    },
    {
      "epoch": 0.08534749348958333,
      "grad_norm": 10.112154006958008,
      "learning_rate": 4.912599566739979e-06,
      "loss": 3.1978,
      "step": 8390
    },
    {
      "epoch": 0.08539835611979167,
      "grad_norm": 9.753875732421875,
      "learning_rate": 4.912494725767972e-06,
      "loss": 3.2426,
      "step": 8395
    },
    {
      "epoch": 0.08544921875,
      "grad_norm": 13.011728286743164,
      "learning_rate": 4.9123898230728616e-06,
      "loss": 3.689,
      "step": 8400
    },
    {
      "epoch": 0.08550008138020833,
      "grad_norm": 10.342684745788574,
      "learning_rate": 4.912284858657328e-06,
      "loss": 3.2583,
      "step": 8405
    },
    {
      "epoch": 0.08555094401041667,
      "grad_norm": 15.096055030822754,
      "learning_rate": 4.9121798325240574e-06,
      "loss": 3.797,
      "step": 8410
    },
    {
      "epoch": 0.085601806640625,
      "grad_norm": 16.6988582611084,
      "learning_rate": 4.912074744675739e-06,
      "loss": 3.7339,
      "step": 8415
    },
    {
      "epoch": 0.08565266927083333,
      "grad_norm": 9.644712448120117,
      "learning_rate": 4.911969595115059e-06,
      "loss": 3.6449,
      "step": 8420
    },
    {
      "epoch": 0.08570353190104167,
      "grad_norm": 9.427045822143555,
      "learning_rate": 4.911864383844709e-06,
      "loss": 3.7292,
      "step": 8425
    },
    {
      "epoch": 0.08575439453125,
      "grad_norm": 12.242361068725586,
      "learning_rate": 4.9117591108673815e-06,
      "loss": 3.6643,
      "step": 8430
    },
    {
      "epoch": 0.08580525716145833,
      "grad_norm": 15.966888427734375,
      "learning_rate": 4.911653776185768e-06,
      "loss": 3.7233,
      "step": 8435
    },
    {
      "epoch": 0.08585611979166667,
      "grad_norm": 13.699934005737305,
      "learning_rate": 4.9115483798025635e-06,
      "loss": 3.2923,
      "step": 8440
    },
    {
      "epoch": 0.085906982421875,
      "grad_norm": 18.62406349182129,
      "learning_rate": 4.911442921720465e-06,
      "loss": 3.3712,
      "step": 8445
    },
    {
      "epoch": 0.08595784505208333,
      "grad_norm": 12.284649848937988,
      "learning_rate": 4.911337401942172e-06,
      "loss": 3.0775,
      "step": 8450
    },
    {
      "epoch": 0.08600870768229167,
      "grad_norm": 15.336729049682617,
      "learning_rate": 4.911231820470383e-06,
      "loss": 3.1454,
      "step": 8455
    },
    {
      "epoch": 0.0860595703125,
      "grad_norm": 12.336068153381348,
      "learning_rate": 4.911126177307799e-06,
      "loss": 3.5428,
      "step": 8460
    },
    {
      "epoch": 0.08611043294270833,
      "grad_norm": 9.04693603515625,
      "learning_rate": 4.911020472457124e-06,
      "loss": 4.0488,
      "step": 8465
    },
    {
      "epoch": 0.08616129557291667,
      "grad_norm": 8.606358528137207,
      "learning_rate": 4.91091470592106e-06,
      "loss": 3.2326,
      "step": 8470
    },
    {
      "epoch": 0.086212158203125,
      "grad_norm": 16.552690505981445,
      "learning_rate": 4.910808877702317e-06,
      "loss": 3.6028,
      "step": 8475
    },
    {
      "epoch": 0.08626302083333333,
      "grad_norm": 10.208168029785156,
      "learning_rate": 4.910702987803599e-06,
      "loss": 3.4335,
      "step": 8480
    },
    {
      "epoch": 0.08631388346354167,
      "grad_norm": 12.139317512512207,
      "learning_rate": 4.910597036227617e-06,
      "loss": 3.2395,
      "step": 8485
    },
    {
      "epoch": 0.08636474609375,
      "grad_norm": 10.915331840515137,
      "learning_rate": 4.91049102297708e-06,
      "loss": 3.4295,
      "step": 8490
    },
    {
      "epoch": 0.08641560872395833,
      "grad_norm": 9.047663688659668,
      "learning_rate": 4.910384948054703e-06,
      "loss": 3.7313,
      "step": 8495
    },
    {
      "epoch": 0.08646647135416667,
      "grad_norm": 11.79930305480957,
      "learning_rate": 4.910278811463197e-06,
      "loss": 3.8923,
      "step": 8500
    },
    {
      "epoch": 0.086517333984375,
      "grad_norm": 9.152587890625,
      "learning_rate": 4.91017261320528e-06,
      "loss": 3.4595,
      "step": 8505
    },
    {
      "epoch": 0.08656819661458333,
      "grad_norm": 10.824846267700195,
      "learning_rate": 4.910066353283668e-06,
      "loss": 3.2886,
      "step": 8510
    },
    {
      "epoch": 0.08661905924479167,
      "grad_norm": 6.95084285736084,
      "learning_rate": 4.909960031701079e-06,
      "loss": 3.3848,
      "step": 8515
    },
    {
      "epoch": 0.086669921875,
      "grad_norm": 16.21605682373047,
      "learning_rate": 4.9098536484602334e-06,
      "loss": 3.2412,
      "step": 8520
    },
    {
      "epoch": 0.08672078450520833,
      "grad_norm": 8.85698127746582,
      "learning_rate": 4.909747203563855e-06,
      "loss": 3.5511,
      "step": 8525
    },
    {
      "epoch": 0.08677164713541667,
      "grad_norm": 16.60267448425293,
      "learning_rate": 4.909640697014664e-06,
      "loss": 3.9667,
      "step": 8530
    },
    {
      "epoch": 0.086822509765625,
      "grad_norm": 14.518190383911133,
      "learning_rate": 4.909534128815387e-06,
      "loss": 3.3415,
      "step": 8535
    },
    {
      "epoch": 0.08687337239583333,
      "grad_norm": 10.557718276977539,
      "learning_rate": 4.909427498968752e-06,
      "loss": 3.4739,
      "step": 8540
    },
    {
      "epoch": 0.08692423502604167,
      "grad_norm": 11.117448806762695,
      "learning_rate": 4.909320807477485e-06,
      "loss": 3.345,
      "step": 8545
    },
    {
      "epoch": 0.08697509765625,
      "grad_norm": 13.413477897644043,
      "learning_rate": 4.9092140543443145e-06,
      "loss": 3.8153,
      "step": 8550
    },
    {
      "epoch": 0.08702596028645833,
      "grad_norm": 11.934691429138184,
      "learning_rate": 4.909107239571975e-06,
      "loss": 3.548,
      "step": 8555
    },
    {
      "epoch": 0.08707682291666667,
      "grad_norm": 10.76050853729248,
      "learning_rate": 4.9090003631631975e-06,
      "loss": 3.3043,
      "step": 8560
    },
    {
      "epoch": 0.087127685546875,
      "grad_norm": 10.168484687805176,
      "learning_rate": 4.9088934251207165e-06,
      "loss": 3.8714,
      "step": 8565
    },
    {
      "epoch": 0.08717854817708333,
      "grad_norm": 13.013615608215332,
      "learning_rate": 4.908786425447269e-06,
      "loss": 3.5501,
      "step": 8570
    },
    {
      "epoch": 0.08722941080729167,
      "grad_norm": 19.23887062072754,
      "learning_rate": 4.908679364145591e-06,
      "loss": 3.5044,
      "step": 8575
    },
    {
      "epoch": 0.0872802734375,
      "grad_norm": 14.924409866333008,
      "learning_rate": 4.908572241218422e-06,
      "loss": 3.2358,
      "step": 8580
    },
    {
      "epoch": 0.08733113606770833,
      "grad_norm": 12.832475662231445,
      "learning_rate": 4.908465056668504e-06,
      "loss": 3.6383,
      "step": 8585
    },
    {
      "epoch": 0.08738199869791667,
      "grad_norm": 15.433663368225098,
      "learning_rate": 4.908357810498578e-06,
      "loss": 3.4119,
      "step": 8590
    },
    {
      "epoch": 0.087432861328125,
      "grad_norm": 8.936971664428711,
      "learning_rate": 4.908250502711388e-06,
      "loss": 3.2385,
      "step": 8595
    },
    {
      "epoch": 0.08748372395833333,
      "grad_norm": 15.772163391113281,
      "learning_rate": 4.9081431333096805e-06,
      "loss": 3.5685,
      "step": 8600
    },
    {
      "epoch": 0.08753458658854167,
      "grad_norm": 11.028748512268066,
      "learning_rate": 4.908035702296201e-06,
      "loss": 3.4128,
      "step": 8605
    },
    {
      "epoch": 0.08758544921875,
      "grad_norm": 12.614755630493164,
      "learning_rate": 4.907928209673699e-06,
      "loss": 3.2888,
      "step": 8610
    },
    {
      "epoch": 0.08763631184895833,
      "grad_norm": 12.599555969238281,
      "learning_rate": 4.907820655444924e-06,
      "loss": 3.1925,
      "step": 8615
    },
    {
      "epoch": 0.08768717447916667,
      "grad_norm": 8.425646781921387,
      "learning_rate": 4.907713039612629e-06,
      "loss": 3.2201,
      "step": 8620
    },
    {
      "epoch": 0.087738037109375,
      "grad_norm": 11.59953784942627,
      "learning_rate": 4.907605362179566e-06,
      "loss": 3.2422,
      "step": 8625
    },
    {
      "epoch": 0.08778889973958333,
      "grad_norm": 7.5655131340026855,
      "learning_rate": 4.907497623148491e-06,
      "loss": 3.5319,
      "step": 8630
    },
    {
      "epoch": 0.08783976236979167,
      "grad_norm": 15.107267379760742,
      "learning_rate": 4.90738982252216e-06,
      "loss": 3.5978,
      "step": 8635
    },
    {
      "epoch": 0.087890625,
      "grad_norm": 15.67873764038086,
      "learning_rate": 4.90728196030333e-06,
      "loss": 3.2055,
      "step": 8640
    },
    {
      "epoch": 0.08794148763020833,
      "grad_norm": 13.261436462402344,
      "learning_rate": 4.907174036494763e-06,
      "loss": 3.0959,
      "step": 8645
    },
    {
      "epoch": 0.08799235026041667,
      "grad_norm": 8.268048286437988,
      "learning_rate": 4.907066051099219e-06,
      "loss": 4.1154,
      "step": 8650
    },
    {
      "epoch": 0.088043212890625,
      "grad_norm": 10.855912208557129,
      "learning_rate": 4.906958004119459e-06,
      "loss": 3.7448,
      "step": 8655
    },
    {
      "epoch": 0.08809407552083333,
      "grad_norm": 11.688394546508789,
      "learning_rate": 4.90684989555825e-06,
      "loss": 3.5907,
      "step": 8660
    },
    {
      "epoch": 0.08814493815104167,
      "grad_norm": 10.814862251281738,
      "learning_rate": 4.906741725418357e-06,
      "loss": 3.0817,
      "step": 8665
    },
    {
      "epoch": 0.08819580078125,
      "grad_norm": 15.317708969116211,
      "learning_rate": 4.906633493702547e-06,
      "loss": 3.3432,
      "step": 8670
    },
    {
      "epoch": 0.08824666341145833,
      "grad_norm": 9.729673385620117,
      "learning_rate": 4.9065252004135896e-06,
      "loss": 3.2816,
      "step": 8675
    },
    {
      "epoch": 0.08829752604166667,
      "grad_norm": 16.496000289916992,
      "learning_rate": 4.906416845554255e-06,
      "loss": 3.511,
      "step": 8680
    },
    {
      "epoch": 0.088348388671875,
      "grad_norm": 16.304975509643555,
      "learning_rate": 4.906308429127317e-06,
      "loss": 3.3911,
      "step": 8685
    },
    {
      "epoch": 0.08839925130208333,
      "grad_norm": 9.595266342163086,
      "learning_rate": 4.906199951135547e-06,
      "loss": 3.5494,
      "step": 8690
    },
    {
      "epoch": 0.08845011393229167,
      "grad_norm": 13.496864318847656,
      "learning_rate": 4.906091411581722e-06,
      "loss": 3.0696,
      "step": 8695
    },
    {
      "epoch": 0.0885009765625,
      "grad_norm": 9.127147674560547,
      "learning_rate": 4.905982810468619e-06,
      "loss": 3.5201,
      "step": 8700
    },
    {
      "epoch": 0.08855183919270833,
      "grad_norm": 17.793336868286133,
      "learning_rate": 4.905874147799015e-06,
      "loss": 3.9616,
      "step": 8705
    },
    {
      "epoch": 0.08860270182291667,
      "grad_norm": 12.619993209838867,
      "learning_rate": 4.905765423575692e-06,
      "loss": 3.6193,
      "step": 8710
    },
    {
      "epoch": 0.088653564453125,
      "grad_norm": 13.0260648727417,
      "learning_rate": 4.90565663780143e-06,
      "loss": 3.1035,
      "step": 8715
    },
    {
      "epoch": 0.08870442708333333,
      "grad_norm": 11.968487739562988,
      "learning_rate": 4.905547790479015e-06,
      "loss": 3.8811,
      "step": 8720
    },
    {
      "epoch": 0.08875528971354167,
      "grad_norm": 14.09224796295166,
      "learning_rate": 4.905438881611228e-06,
      "loss": 3.2689,
      "step": 8725
    },
    {
      "epoch": 0.08880615234375,
      "grad_norm": 15.479565620422363,
      "learning_rate": 4.905329911200858e-06,
      "loss": 3.3083,
      "step": 8730
    },
    {
      "epoch": 0.08885701497395833,
      "grad_norm": 12.190298080444336,
      "learning_rate": 4.905220879250693e-06,
      "loss": 3.0763,
      "step": 8735
    },
    {
      "epoch": 0.08890787760416667,
      "grad_norm": 13.899861335754395,
      "learning_rate": 4.905111785763521e-06,
      "loss": 3.5266,
      "step": 8740
    },
    {
      "epoch": 0.088958740234375,
      "grad_norm": 11.89075756072998,
      "learning_rate": 4.905002630742135e-06,
      "loss": 3.9321,
      "step": 8745
    },
    {
      "epoch": 0.08900960286458333,
      "grad_norm": 13.489803314208984,
      "learning_rate": 4.904893414189326e-06,
      "loss": 3.4766,
      "step": 8750
    },
    {
      "epoch": 0.08906046549479167,
      "grad_norm": 13.605609893798828,
      "learning_rate": 4.904784136107888e-06,
      "loss": 3.1691,
      "step": 8755
    },
    {
      "epoch": 0.089111328125,
      "grad_norm": 10.71879768371582,
      "learning_rate": 4.90467479650062e-06,
      "loss": 3.2934,
      "step": 8760
    },
    {
      "epoch": 0.08916219075520833,
      "grad_norm": 10.595026016235352,
      "learning_rate": 4.9045653953703156e-06,
      "loss": 3.1218,
      "step": 8765
    },
    {
      "epoch": 0.08921305338541667,
      "grad_norm": 11.80540943145752,
      "learning_rate": 4.9044559327197764e-06,
      "loss": 3.2589,
      "step": 8770
    },
    {
      "epoch": 0.089263916015625,
      "grad_norm": 14.13592529296875,
      "learning_rate": 4.9043464085518026e-06,
      "loss": 3.3531,
      "step": 8775
    },
    {
      "epoch": 0.08931477864583333,
      "grad_norm": 12.386794090270996,
      "learning_rate": 4.904236822869195e-06,
      "loss": 3.529,
      "step": 8780
    },
    {
      "epoch": 0.08936564127604167,
      "grad_norm": 15.09712028503418,
      "learning_rate": 4.904127175674758e-06,
      "loss": 3.5851,
      "step": 8785
    },
    {
      "epoch": 0.08941650390625,
      "grad_norm": 22.025758743286133,
      "learning_rate": 4.904017466971297e-06,
      "loss": 3.4247,
      "step": 8790
    },
    {
      "epoch": 0.08946736653645833,
      "grad_norm": 12.594602584838867,
      "learning_rate": 4.9039076967616196e-06,
      "loss": 3.5087,
      "step": 8795
    },
    {
      "epoch": 0.08951822916666667,
      "grad_norm": 12.804073333740234,
      "learning_rate": 4.903797865048533e-06,
      "loss": 3.3348,
      "step": 8800
    },
    {
      "epoch": 0.089569091796875,
      "grad_norm": 10.385623931884766,
      "learning_rate": 4.903687971834848e-06,
      "loss": 3.4176,
      "step": 8805
    },
    {
      "epoch": 0.08961995442708333,
      "grad_norm": 9.570121765136719,
      "learning_rate": 4.903578017123376e-06,
      "loss": 3.3923,
      "step": 8810
    },
    {
      "epoch": 0.08967081705729167,
      "grad_norm": 12.22430419921875,
      "learning_rate": 4.90346800091693e-06,
      "loss": 3.6975,
      "step": 8815
    },
    {
      "epoch": 0.0897216796875,
      "grad_norm": 14.948432922363281,
      "learning_rate": 4.9033579232183256e-06,
      "loss": 3.483,
      "step": 8820
    },
    {
      "epoch": 0.08977254231770833,
      "grad_norm": 14.922759056091309,
      "learning_rate": 4.903247784030377e-06,
      "loss": 3.4358,
      "step": 8825
    },
    {
      "epoch": 0.08982340494791667,
      "grad_norm": 11.016508102416992,
      "learning_rate": 4.903137583355905e-06,
      "loss": 3.4084,
      "step": 8830
    },
    {
      "epoch": 0.089874267578125,
      "grad_norm": 17.078842163085938,
      "learning_rate": 4.903027321197726e-06,
      "loss": 3.5926,
      "step": 8835
    },
    {
      "epoch": 0.08992513020833333,
      "grad_norm": 13.634777069091797,
      "learning_rate": 4.902916997558665e-06,
      "loss": 3.5905,
      "step": 8840
    },
    {
      "epoch": 0.08997599283854167,
      "grad_norm": 8.89342212677002,
      "learning_rate": 4.902806612441539e-06,
      "loss": 3.2723,
      "step": 8845
    },
    {
      "epoch": 0.09002685546875,
      "grad_norm": 14.009095191955566,
      "learning_rate": 4.902696165849178e-06,
      "loss": 3.5151,
      "step": 8850
    },
    {
      "epoch": 0.09007771809895833,
      "grad_norm": 16.494112014770508,
      "learning_rate": 4.902585657784404e-06,
      "loss": 3.229,
      "step": 8855
    },
    {
      "epoch": 0.09012858072916667,
      "grad_norm": 9.716590881347656,
      "learning_rate": 4.902475088250045e-06,
      "loss": 3.4538,
      "step": 8860
    },
    {
      "epoch": 0.090179443359375,
      "grad_norm": 14.583306312561035,
      "learning_rate": 4.90236445724893e-06,
      "loss": 3.6892,
      "step": 8865
    },
    {
      "epoch": 0.09023030598958333,
      "grad_norm": 11.482134819030762,
      "learning_rate": 4.902253764783891e-06,
      "loss": 3.3962,
      "step": 8870
    },
    {
      "epoch": 0.09028116861979167,
      "grad_norm": 14.702147483825684,
      "learning_rate": 4.902143010857758e-06,
      "loss": 3.2805,
      "step": 8875
    },
    {
      "epoch": 0.09033203125,
      "grad_norm": 16.604808807373047,
      "learning_rate": 4.902032195473366e-06,
      "loss": 3.2619,
      "step": 8880
    },
    {
      "epoch": 0.09038289388020833,
      "grad_norm": 12.936920166015625,
      "learning_rate": 4.901921318633549e-06,
      "loss": 3.4328,
      "step": 8885
    },
    {
      "epoch": 0.09043375651041667,
      "grad_norm": 7.750491142272949,
      "learning_rate": 4.901810380341145e-06,
      "loss": 3.7113,
      "step": 8890
    },
    {
      "epoch": 0.090484619140625,
      "grad_norm": 17.840023040771484,
      "learning_rate": 4.901699380598992e-06,
      "loss": 3.1851,
      "step": 8895
    },
    {
      "epoch": 0.09053548177083333,
      "grad_norm": 9.541340827941895,
      "learning_rate": 4.901588319409929e-06,
      "loss": 3.5305,
      "step": 8900
    },
    {
      "epoch": 0.09058634440104167,
      "grad_norm": 15.900449752807617,
      "learning_rate": 4.901477196776798e-06,
      "loss": 3.084,
      "step": 8905
    },
    {
      "epoch": 0.09063720703125,
      "grad_norm": 11.218430519104004,
      "learning_rate": 4.901366012702443e-06,
      "loss": 3.5674,
      "step": 8910
    },
    {
      "epoch": 0.09068806966145833,
      "grad_norm": 12.750350952148438,
      "learning_rate": 4.901254767189707e-06,
      "loss": 3.8215,
      "step": 8915
    },
    {
      "epoch": 0.09073893229166667,
      "grad_norm": 13.617959976196289,
      "learning_rate": 4.901143460241437e-06,
      "loss": 3.3986,
      "step": 8920
    },
    {
      "epoch": 0.090789794921875,
      "grad_norm": 13.11312484741211,
      "learning_rate": 4.90103209186048e-06,
      "loss": 3.453,
      "step": 8925
    },
    {
      "epoch": 0.09084065755208333,
      "grad_norm": 9.750372886657715,
      "learning_rate": 4.9009206620496875e-06,
      "loss": 3.348,
      "step": 8930
    },
    {
      "epoch": 0.09089152018229167,
      "grad_norm": 13.289474487304688,
      "learning_rate": 4.900809170811908e-06,
      "loss": 3.284,
      "step": 8935
    },
    {
      "epoch": 0.0909423828125,
      "grad_norm": 10.578423500061035,
      "learning_rate": 4.900697618149995e-06,
      "loss": 3.8172,
      "step": 8940
    },
    {
      "epoch": 0.09099324544270833,
      "grad_norm": 14.269426345825195,
      "learning_rate": 4.900586004066803e-06,
      "loss": 3.8127,
      "step": 8945
    },
    {
      "epoch": 0.09104410807291667,
      "grad_norm": 11.517721176147461,
      "learning_rate": 4.900474328565186e-06,
      "loss": 3.2798,
      "step": 8950
    },
    {
      "epoch": 0.091094970703125,
      "grad_norm": 12.702051162719727,
      "learning_rate": 4.900362591648003e-06,
      "loss": 3.1521,
      "step": 8955
    },
    {
      "epoch": 0.09114583333333333,
      "grad_norm": 13.737582206726074,
      "learning_rate": 4.900250793318112e-06,
      "loss": 3.3586,
      "step": 8960
    },
    {
      "epoch": 0.09119669596354167,
      "grad_norm": 14.892784118652344,
      "learning_rate": 4.900138933578373e-06,
      "loss": 3.4807,
      "step": 8965
    },
    {
      "epoch": 0.09124755859375,
      "grad_norm": 11.482332229614258,
      "learning_rate": 4.9000270124316495e-06,
      "loss": 3.0206,
      "step": 8970
    },
    {
      "epoch": 0.09129842122395833,
      "grad_norm": 13.81802749633789,
      "learning_rate": 4.899915029880803e-06,
      "loss": 3.3766,
      "step": 8975
    },
    {
      "epoch": 0.09134928385416667,
      "grad_norm": 11.666979789733887,
      "learning_rate": 4.899802985928699e-06,
      "loss": 3.2013,
      "step": 8980
    },
    {
      "epoch": 0.091400146484375,
      "grad_norm": 14.300288200378418,
      "learning_rate": 4.899690880578205e-06,
      "loss": 3.1857,
      "step": 8985
    },
    {
      "epoch": 0.09145100911458333,
      "grad_norm": 9.281865119934082,
      "learning_rate": 4.899578713832188e-06,
      "loss": 3.4702,
      "step": 8990
    },
    {
      "epoch": 0.09150187174479167,
      "grad_norm": 13.91549015045166,
      "learning_rate": 4.899466485693518e-06,
      "loss": 3.9119,
      "step": 8995
    },
    {
      "epoch": 0.091552734375,
      "grad_norm": 8.730186462402344,
      "learning_rate": 4.899354196165068e-06,
      "loss": 3.136,
      "step": 9000
    },
    {
      "epoch": 0.09160359700520833,
      "grad_norm": 9.029413223266602,
      "learning_rate": 4.899241845249708e-06,
      "loss": 3.6015,
      "step": 9005
    },
    {
      "epoch": 0.09165445963541667,
      "grad_norm": 10.518607139587402,
      "learning_rate": 4.899129432950316e-06,
      "loss": 3.5339,
      "step": 9010
    },
    {
      "epoch": 0.091705322265625,
      "grad_norm": 13.978117942810059,
      "learning_rate": 4.899016959269764e-06,
      "loss": 3.6865,
      "step": 9015
    },
    {
      "epoch": 0.09175618489583333,
      "grad_norm": 13.56008243560791,
      "learning_rate": 4.898904424210934e-06,
      "loss": 3.1692,
      "step": 9020
    },
    {
      "epoch": 0.09180704752604167,
      "grad_norm": 11.420557975769043,
      "learning_rate": 4.898791827776701e-06,
      "loss": 3.3184,
      "step": 9025
    },
    {
      "epoch": 0.09185791015625,
      "grad_norm": 11.740954399108887,
      "learning_rate": 4.898679169969949e-06,
      "loss": 3.5797,
      "step": 9030
    },
    {
      "epoch": 0.09190877278645833,
      "grad_norm": 11.412757873535156,
      "learning_rate": 4.898566450793558e-06,
      "loss": 3.4047,
      "step": 9035
    },
    {
      "epoch": 0.09195963541666667,
      "grad_norm": 7.887825965881348,
      "learning_rate": 4.898453670250413e-06,
      "loss": 3.3213,
      "step": 9040
    },
    {
      "epoch": 0.092010498046875,
      "grad_norm": 10.620329856872559,
      "learning_rate": 4.8983408283433995e-06,
      "loss": 3.3956,
      "step": 9045
    },
    {
      "epoch": 0.09206136067708333,
      "grad_norm": 10.935351371765137,
      "learning_rate": 4.898227925075405e-06,
      "loss": 3.1719,
      "step": 9050
    },
    {
      "epoch": 0.09211222330729167,
      "grad_norm": 13.198664665222168,
      "learning_rate": 4.898114960449317e-06,
      "loss": 3.2274,
      "step": 9055
    },
    {
      "epoch": 0.0921630859375,
      "grad_norm": 7.661378383636475,
      "learning_rate": 4.8980019344680255e-06,
      "loss": 3.4533,
      "step": 9060
    },
    {
      "epoch": 0.09221394856770833,
      "grad_norm": 14.63359546661377,
      "learning_rate": 4.897888847134424e-06,
      "loss": 3.2218,
      "step": 9065
    },
    {
      "epoch": 0.09226481119791667,
      "grad_norm": 10.786942481994629,
      "learning_rate": 4.897775698451404e-06,
      "loss": 3.0604,
      "step": 9070
    },
    {
      "epoch": 0.092315673828125,
      "grad_norm": 13.290276527404785,
      "learning_rate": 4.897662488421861e-06,
      "loss": 3.4856,
      "step": 9075
    },
    {
      "epoch": 0.09236653645833333,
      "grad_norm": 13.736598014831543,
      "learning_rate": 4.897549217048692e-06,
      "loss": 3.4483,
      "step": 9080
    },
    {
      "epoch": 0.09241739908854167,
      "grad_norm": 13.418255805969238,
      "learning_rate": 4.897435884334795e-06,
      "loss": 3.3985,
      "step": 9085
    },
    {
      "epoch": 0.09246826171875,
      "grad_norm": 11.04183578491211,
      "learning_rate": 4.897322490283069e-06,
      "loss": 3.746,
      "step": 9090
    },
    {
      "epoch": 0.09251912434895833,
      "grad_norm": 14.375475883483887,
      "learning_rate": 4.897209034896414e-06,
      "loss": 3.3322,
      "step": 9095
    },
    {
      "epoch": 0.09256998697916667,
      "grad_norm": 12.415508270263672,
      "learning_rate": 4.897095518177735e-06,
      "loss": 3.221,
      "step": 9100
    },
    {
      "epoch": 0.092620849609375,
      "grad_norm": 13.276843070983887,
      "learning_rate": 4.896981940129935e-06,
      "loss": 3.1526,
      "step": 9105
    },
    {
      "epoch": 0.09267171223958333,
      "grad_norm": 16.24017333984375,
      "learning_rate": 4.8968683007559204e-06,
      "loss": 3.1622,
      "step": 9110
    },
    {
      "epoch": 0.09272257486979167,
      "grad_norm": 13.676992416381836,
      "learning_rate": 4.8967546000585985e-06,
      "loss": 3.4367,
      "step": 9115
    },
    {
      "epoch": 0.0927734375,
      "grad_norm": 12.518017768859863,
      "learning_rate": 4.896640838040878e-06,
      "loss": 3.149,
      "step": 9120
    },
    {
      "epoch": 0.09282430013020833,
      "grad_norm": 9.173798561096191,
      "learning_rate": 4.89652701470567e-06,
      "loss": 3.4939,
      "step": 9125
    },
    {
      "epoch": 0.09287516276041667,
      "grad_norm": 9.206432342529297,
      "learning_rate": 4.896413130055887e-06,
      "loss": 3.2798,
      "step": 9130
    },
    {
      "epoch": 0.092926025390625,
      "grad_norm": 10.400619506835938,
      "learning_rate": 4.896299184094441e-06,
      "loss": 3.5009,
      "step": 9135
    },
    {
      "epoch": 0.09297688802083333,
      "grad_norm": 14.5686616897583,
      "learning_rate": 4.896185176824249e-06,
      "loss": 3.0607,
      "step": 9140
    },
    {
      "epoch": 0.09302775065104167,
      "grad_norm": 15.614046096801758,
      "learning_rate": 4.8960711082482275e-06,
      "loss": 3.5764,
      "step": 9145
    },
    {
      "epoch": 0.09307861328125,
      "grad_norm": 11.736412048339844,
      "learning_rate": 4.895956978369294e-06,
      "loss": 3.2503,
      "step": 9150
    },
    {
      "epoch": 0.09312947591145833,
      "grad_norm": 8.38943862915039,
      "learning_rate": 4.895842787190369e-06,
      "loss": 3.5593,
      "step": 9155
    },
    {
      "epoch": 0.09318033854166667,
      "grad_norm": 10.80936050415039,
      "learning_rate": 4.895728534714375e-06,
      "loss": 3.7306,
      "step": 9160
    },
    {
      "epoch": 0.093231201171875,
      "grad_norm": 8.9891939163208,
      "learning_rate": 4.895614220944233e-06,
      "loss": 3.5101,
      "step": 9165
    },
    {
      "epoch": 0.09328206380208333,
      "grad_norm": 10.55257511138916,
      "learning_rate": 4.895499845882869e-06,
      "loss": 3.642,
      "step": 9170
    },
    {
      "epoch": 0.09333292643229167,
      "grad_norm": 11.787149429321289,
      "learning_rate": 4.895385409533211e-06,
      "loss": 3.3967,
      "step": 9175
    },
    {
      "epoch": 0.0933837890625,
      "grad_norm": 14.418353080749512,
      "learning_rate": 4.895270911898183e-06,
      "loss": 3.3822,
      "step": 9180
    },
    {
      "epoch": 0.09343465169270833,
      "grad_norm": 10.53858470916748,
      "learning_rate": 4.895156352980718e-06,
      "loss": 3.5411,
      "step": 9185
    },
    {
      "epoch": 0.09348551432291667,
      "grad_norm": 14.017937660217285,
      "learning_rate": 4.895041732783745e-06,
      "loss": 3.2294,
      "step": 9190
    },
    {
      "epoch": 0.093536376953125,
      "grad_norm": 9.317304611206055,
      "learning_rate": 4.8949270513101965e-06,
      "loss": 3.0859,
      "step": 9195
    },
    {
      "epoch": 0.09358723958333333,
      "grad_norm": 14.010085105895996,
      "learning_rate": 4.894812308563007e-06,
      "loss": 3.3879,
      "step": 9200
    },
    {
      "epoch": 0.09363810221354167,
      "grad_norm": 14.181611061096191,
      "learning_rate": 4.8946975045451125e-06,
      "loss": 3.4383,
      "step": 9205
    },
    {
      "epoch": 0.09368896484375,
      "grad_norm": 17.099210739135742,
      "learning_rate": 4.894582639259451e-06,
      "loss": 3.2286,
      "step": 9210
    },
    {
      "epoch": 0.09373982747395833,
      "grad_norm": 15.619363784790039,
      "learning_rate": 4.894467712708959e-06,
      "loss": 3.5737,
      "step": 9215
    },
    {
      "epoch": 0.09379069010416667,
      "grad_norm": 8.870353698730469,
      "learning_rate": 4.8943527248965786e-06,
      "loss": 3.2734,
      "step": 9220
    },
    {
      "epoch": 0.093841552734375,
      "grad_norm": 8.589174270629883,
      "learning_rate": 4.894237675825251e-06,
      "loss": 3.5581,
      "step": 9225
    },
    {
      "epoch": 0.09389241536458333,
      "grad_norm": 16.112144470214844,
      "learning_rate": 4.89412256549792e-06,
      "loss": 3.4103,
      "step": 9230
    },
    {
      "epoch": 0.09394327799479167,
      "grad_norm": 10.410552978515625,
      "learning_rate": 4.89400739391753e-06,
      "loss": 3.4895,
      "step": 9235
    },
    {
      "epoch": 0.093994140625,
      "grad_norm": 13.545750617980957,
      "learning_rate": 4.89389216108703e-06,
      "loss": 3.2598,
      "step": 9240
    },
    {
      "epoch": 0.09404500325520833,
      "grad_norm": 18.490036010742188,
      "learning_rate": 4.893776867009365e-06,
      "loss": 3.3705,
      "step": 9245
    },
    {
      "epoch": 0.09409586588541667,
      "grad_norm": 12.499258041381836,
      "learning_rate": 4.893661511687487e-06,
      "loss": 3.3089,
      "step": 9250
    },
    {
      "epoch": 0.094146728515625,
      "grad_norm": 46.41202163696289,
      "learning_rate": 4.893546095124346e-06,
      "loss": 3.46,
      "step": 9255
    },
    {
      "epoch": 0.09419759114583333,
      "grad_norm": 13.263053894042969,
      "learning_rate": 4.893430617322895e-06,
      "loss": 3.4889,
      "step": 9260
    },
    {
      "epoch": 0.09424845377604167,
      "grad_norm": 7.173794269561768,
      "learning_rate": 4.8933150782860905e-06,
      "loss": 3.4028,
      "step": 9265
    },
    {
      "epoch": 0.09429931640625,
      "grad_norm": 15.863984107971191,
      "learning_rate": 4.893199478016886e-06,
      "loss": 3.1168,
      "step": 9270
    },
    {
      "epoch": 0.09435017903645833,
      "grad_norm": 9.970428466796875,
      "learning_rate": 4.8930838165182405e-06,
      "loss": 3.3857,
      "step": 9275
    },
    {
      "epoch": 0.09440104166666667,
      "grad_norm": 6.910197734832764,
      "learning_rate": 4.892968093793112e-06,
      "loss": 3.6812,
      "step": 9280
    },
    {
      "epoch": 0.094451904296875,
      "grad_norm": 12.154581069946289,
      "learning_rate": 4.892852309844462e-06,
      "loss": 3.459,
      "step": 9285
    },
    {
      "epoch": 0.09450276692708333,
      "grad_norm": 13.089887619018555,
      "learning_rate": 4.892736464675254e-06,
      "loss": 3.75,
      "step": 9290
    },
    {
      "epoch": 0.09455362955729167,
      "grad_norm": 12.283695220947266,
      "learning_rate": 4.89262055828845e-06,
      "loss": 3.5142,
      "step": 9295
    },
    {
      "epoch": 0.0946044921875,
      "grad_norm": 16.34111976623535,
      "learning_rate": 4.892504590687016e-06,
      "loss": 3.4612,
      "step": 9300
    },
    {
      "epoch": 0.09465535481770833,
      "grad_norm": 14.332940101623535,
      "learning_rate": 4.89238856187392e-06,
      "loss": 2.9819,
      "step": 9305
    },
    {
      "epoch": 0.09470621744791667,
      "grad_norm": 9.912870407104492,
      "learning_rate": 4.892272471852128e-06,
      "loss": 3.3375,
      "step": 9310
    },
    {
      "epoch": 0.094757080078125,
      "grad_norm": 18.02761459350586,
      "learning_rate": 4.892156320624613e-06,
      "loss": 3.5184,
      "step": 9315
    },
    {
      "epoch": 0.09480794270833333,
      "grad_norm": 14.656006813049316,
      "learning_rate": 4.892040108194346e-06,
      "loss": 3.7168,
      "step": 9320
    },
    {
      "epoch": 0.09485880533854167,
      "grad_norm": 14.194197654724121,
      "learning_rate": 4.8919238345643e-06,
      "loss": 3.7077,
      "step": 9325
    },
    {
      "epoch": 0.09490966796875,
      "grad_norm": 14.731765747070312,
      "learning_rate": 4.891807499737449e-06,
      "loss": 3.526,
      "step": 9330
    },
    {
      "epoch": 0.09496053059895833,
      "grad_norm": 11.650579452514648,
      "learning_rate": 4.891691103716769e-06,
      "loss": 3.2138,
      "step": 9335
    },
    {
      "epoch": 0.09501139322916667,
      "grad_norm": 8.169177055358887,
      "learning_rate": 4.89157464650524e-06,
      "loss": 3.4209,
      "step": 9340
    },
    {
      "epoch": 0.095062255859375,
      "grad_norm": 13.967549324035645,
      "learning_rate": 4.89145812810584e-06,
      "loss": 3.523,
      "step": 9345
    },
    {
      "epoch": 0.09511311848958333,
      "grad_norm": 15.903494834899902,
      "learning_rate": 4.891341548521552e-06,
      "loss": 3.5998,
      "step": 9350
    },
    {
      "epoch": 0.09516398111979167,
      "grad_norm": 18.27981185913086,
      "learning_rate": 4.8912249077553566e-06,
      "loss": 4.1232,
      "step": 9355
    },
    {
      "epoch": 0.09521484375,
      "grad_norm": 12.90363597869873,
      "learning_rate": 4.8911082058102375e-06,
      "loss": 3.6707,
      "step": 9360
    },
    {
      "epoch": 0.09526570638020833,
      "grad_norm": 15.968925476074219,
      "learning_rate": 4.890991442689184e-06,
      "loss": 3.1965,
      "step": 9365
    },
    {
      "epoch": 0.09531656901041667,
      "grad_norm": 12.801103591918945,
      "learning_rate": 4.890874618395179e-06,
      "loss": 3.2281,
      "step": 9370
    },
    {
      "epoch": 0.095367431640625,
      "grad_norm": 9.642354011535645,
      "learning_rate": 4.890757732931215e-06,
      "loss": 3.6309,
      "step": 9375
    },
    {
      "epoch": 0.09541829427083333,
      "grad_norm": 18.670930862426758,
      "learning_rate": 4.8906407863002805e-06,
      "loss": 3.7841,
      "step": 9380
    },
    {
      "epoch": 0.09546915690104167,
      "grad_norm": 12.024273872375488,
      "learning_rate": 4.8905237785053675e-06,
      "loss": 3.3917,
      "step": 9385
    },
    {
      "epoch": 0.09552001953125,
      "grad_norm": 8.89647102355957,
      "learning_rate": 4.8904067095494714e-06,
      "loss": 3.703,
      "step": 9390
    },
    {
      "epoch": 0.09557088216145833,
      "grad_norm": 11.106717109680176,
      "learning_rate": 4.890289579435585e-06,
      "loss": 3.2822,
      "step": 9395
    },
    {
      "epoch": 0.09562174479166667,
      "grad_norm": 14.37429428100586,
      "learning_rate": 4.8901723881667075e-06,
      "loss": 3.3098,
      "step": 9400
    },
    {
      "epoch": 0.095672607421875,
      "grad_norm": 12.311078071594238,
      "learning_rate": 4.890055135745835e-06,
      "loss": 3.1674,
      "step": 9405
    },
    {
      "epoch": 0.09572347005208333,
      "grad_norm": 13.66702938079834,
      "learning_rate": 4.88993782217597e-06,
      "loss": 3.0287,
      "step": 9410
    },
    {
      "epoch": 0.09577433268229167,
      "grad_norm": 11.145536422729492,
      "learning_rate": 4.889820447460111e-06,
      "loss": 3.3768,
      "step": 9415
    },
    {
      "epoch": 0.0958251953125,
      "grad_norm": 12.304450988769531,
      "learning_rate": 4.889703011601262e-06,
      "loss": 3.5019,
      "step": 9420
    },
    {
      "epoch": 0.09587605794270833,
      "grad_norm": 9.890800476074219,
      "learning_rate": 4.889585514602429e-06,
      "loss": 3.1514,
      "step": 9425
    },
    {
      "epoch": 0.09592692057291667,
      "grad_norm": 9.260703086853027,
      "learning_rate": 4.889467956466616e-06,
      "loss": 3.4432,
      "step": 9430
    },
    {
      "epoch": 0.095977783203125,
      "grad_norm": 13.874801635742188,
      "learning_rate": 4.889350337196832e-06,
      "loss": 3.2378,
      "step": 9435
    },
    {
      "epoch": 0.09602864583333333,
      "grad_norm": 13.591817855834961,
      "learning_rate": 4.889232656796086e-06,
      "loss": 3.4383,
      "step": 9440
    },
    {
      "epoch": 0.09607950846354167,
      "grad_norm": 9.10405445098877,
      "learning_rate": 4.8891149152673875e-06,
      "loss": 3.4487,
      "step": 9445
    },
    {
      "epoch": 0.09613037109375,
      "grad_norm": 8.167961120605469,
      "learning_rate": 4.888997112613752e-06,
      "loss": 3.6359,
      "step": 9450
    },
    {
      "epoch": 0.09618123372395833,
      "grad_norm": 14.467639923095703,
      "learning_rate": 4.888879248838191e-06,
      "loss": 3.1424,
      "step": 9455
    },
    {
      "epoch": 0.09623209635416667,
      "grad_norm": 12.446633338928223,
      "learning_rate": 4.888761323943721e-06,
      "loss": 3.5426,
      "step": 9460
    },
    {
      "epoch": 0.096282958984375,
      "grad_norm": 11.570879936218262,
      "learning_rate": 4.888643337933358e-06,
      "loss": 3.5324,
      "step": 9465
    },
    {
      "epoch": 0.09633382161458333,
      "grad_norm": 13.078673362731934,
      "learning_rate": 4.8885252908101226e-06,
      "loss": 3.3334,
      "step": 9470
    },
    {
      "epoch": 0.09638468424479167,
      "grad_norm": 10.936042785644531,
      "learning_rate": 4.888407182577032e-06,
      "loss": 3.2394,
      "step": 9475
    },
    {
      "epoch": 0.096435546875,
      "grad_norm": 13.004395484924316,
      "learning_rate": 4.888289013237112e-06,
      "loss": 3.3966,
      "step": 9480
    },
    {
      "epoch": 0.09648640950520833,
      "grad_norm": 11.469902992248535,
      "learning_rate": 4.888170782793382e-06,
      "loss": 3.717,
      "step": 9485
    },
    {
      "epoch": 0.09653727213541667,
      "grad_norm": 14.799452781677246,
      "learning_rate": 4.888052491248869e-06,
      "loss": 3.4483,
      "step": 9490
    },
    {
      "epoch": 0.096588134765625,
      "grad_norm": 10.995068550109863,
      "learning_rate": 4.887934138606599e-06,
      "loss": 3.3911,
      "step": 9495
    },
    {
      "epoch": 0.09663899739583333,
      "grad_norm": 13.323307037353516,
      "learning_rate": 4.8878157248696e-06,
      "loss": 3.749,
      "step": 9500
    },
    {
      "epoch": 0.09668986002604167,
      "grad_norm": 20.117530822753906,
      "learning_rate": 4.887697250040901e-06,
      "loss": 3.5721,
      "step": 9505
    },
    {
      "epoch": 0.09674072265625,
      "grad_norm": 10.738547325134277,
      "learning_rate": 4.887578714123536e-06,
      "loss": 3.3881,
      "step": 9510
    },
    {
      "epoch": 0.09679158528645833,
      "grad_norm": 12.129579544067383,
      "learning_rate": 4.887460117120533e-06,
      "loss": 3.5233,
      "step": 9515
    },
    {
      "epoch": 0.09684244791666667,
      "grad_norm": 14.902997970581055,
      "learning_rate": 4.88734145903493e-06,
      "loss": 3.5717,
      "step": 9520
    },
    {
      "epoch": 0.096893310546875,
      "grad_norm": 11.737396240234375,
      "learning_rate": 4.887222739869761e-06,
      "loss": 3.175,
      "step": 9525
    },
    {
      "epoch": 0.09694417317708333,
      "grad_norm": 6.6605730056762695,
      "learning_rate": 4.8871039596280654e-06,
      "loss": 3.4191,
      "step": 9530
    },
    {
      "epoch": 0.09699503580729167,
      "grad_norm": 11.67459487915039,
      "learning_rate": 4.88698511831288e-06,
      "loss": 3.4298,
      "step": 9535
    },
    {
      "epoch": 0.0970458984375,
      "grad_norm": 7.061823844909668,
      "learning_rate": 4.886866215927246e-06,
      "loss": 3.4257,
      "step": 9540
    },
    {
      "epoch": 0.09709676106770833,
      "grad_norm": 14.894327163696289,
      "learning_rate": 4.8867472524742055e-06,
      "loss": 3.4476,
      "step": 9545
    },
    {
      "epoch": 0.09714762369791667,
      "grad_norm": 14.296971321105957,
      "learning_rate": 4.8866282279568024e-06,
      "loss": 3.7858,
      "step": 9550
    },
    {
      "epoch": 0.097198486328125,
      "grad_norm": 12.031661033630371,
      "learning_rate": 4.886509142378082e-06,
      "loss": 3.5772,
      "step": 9555
    },
    {
      "epoch": 0.09724934895833333,
      "grad_norm": 13.860501289367676,
      "learning_rate": 4.88638999574109e-06,
      "loss": 3.259,
      "step": 9560
    },
    {
      "epoch": 0.09730021158854167,
      "grad_norm": 12.739351272583008,
      "learning_rate": 4.886270788048877e-06,
      "loss": 3.358,
      "step": 9565
    },
    {
      "epoch": 0.09735107421875,
      "grad_norm": 8.996005058288574,
      "learning_rate": 4.8861515193044905e-06,
      "loss": 3.0429,
      "step": 9570
    },
    {
      "epoch": 0.09740193684895833,
      "grad_norm": 10.95101261138916,
      "learning_rate": 4.886032189510983e-06,
      "loss": 3.8154,
      "step": 9575
    },
    {
      "epoch": 0.09745279947916667,
      "grad_norm": 13.15895938873291,
      "learning_rate": 4.885912798671408e-06,
      "loss": 3.3369,
      "step": 9580
    },
    {
      "epoch": 0.097503662109375,
      "grad_norm": 13.110836029052734,
      "learning_rate": 4.885793346788819e-06,
      "loss": 3.5893,
      "step": 9585
    },
    {
      "epoch": 0.09755452473958333,
      "grad_norm": 13.61912727355957,
      "learning_rate": 4.885673833866273e-06,
      "loss": 3.3841,
      "step": 9590
    },
    {
      "epoch": 0.09760538736979167,
      "grad_norm": 8.887752532958984,
      "learning_rate": 4.885554259906827e-06,
      "loss": 3.7583,
      "step": 9595
    },
    {
      "epoch": 0.09765625,
      "grad_norm": 14.288346290588379,
      "learning_rate": 4.885434624913541e-06,
      "loss": 3.0627,
      "step": 9600
    },
    {
      "epoch": 0.09770711263020833,
      "grad_norm": 13.13482666015625,
      "learning_rate": 4.8853149288894765e-06,
      "loss": 3.6814,
      "step": 9605
    },
    {
      "epoch": 0.09775797526041667,
      "grad_norm": 10.950799942016602,
      "learning_rate": 4.885195171837694e-06,
      "loss": 3.0417,
      "step": 9610
    },
    {
      "epoch": 0.097808837890625,
      "grad_norm": 15.985264778137207,
      "learning_rate": 4.885075353761258e-06,
      "loss": 3.4947,
      "step": 9615
    },
    {
      "epoch": 0.09785970052083333,
      "grad_norm": 11.092909812927246,
      "learning_rate": 4.884955474663235e-06,
      "loss": 3.6053,
      "step": 9620
    },
    {
      "epoch": 0.09791056315104167,
      "grad_norm": 10.567216873168945,
      "learning_rate": 4.884835534546692e-06,
      "loss": 3.5717,
      "step": 9625
    },
    {
      "epoch": 0.09796142578125,
      "grad_norm": 16.9993896484375,
      "learning_rate": 4.884715533414696e-06,
      "loss": 3.8297,
      "step": 9630
    },
    {
      "epoch": 0.09801228841145833,
      "grad_norm": 9.653956413269043,
      "learning_rate": 4.884595471270319e-06,
      "loss": 3.2011,
      "step": 9635
    },
    {
      "epoch": 0.09806315104166667,
      "grad_norm": 7.467728137969971,
      "learning_rate": 4.884475348116631e-06,
      "loss": 3.3446,
      "step": 9640
    },
    {
      "epoch": 0.098114013671875,
      "grad_norm": 16.218908309936523,
      "learning_rate": 4.884355163956708e-06,
      "loss": 3.4298,
      "step": 9645
    },
    {
      "epoch": 0.09816487630208333,
      "grad_norm": 8.244502067565918,
      "learning_rate": 4.884234918793622e-06,
      "loss": 3.3024,
      "step": 9650
    },
    {
      "epoch": 0.09821573893229167,
      "grad_norm": 13.385653495788574,
      "learning_rate": 4.884114612630451e-06,
      "loss": 3.2548,
      "step": 9655
    },
    {
      "epoch": 0.0982666015625,
      "grad_norm": 12.53419303894043,
      "learning_rate": 4.883994245470274e-06,
      "loss": 3.0654,
      "step": 9660
    },
    {
      "epoch": 0.09831746419270833,
      "grad_norm": 7.7406005859375,
      "learning_rate": 4.883873817316168e-06,
      "loss": 3.7867,
      "step": 9665
    },
    {
      "epoch": 0.09836832682291667,
      "grad_norm": 16.32261085510254,
      "learning_rate": 4.883753328171216e-06,
      "loss": 3.5244,
      "step": 9670
    },
    {
      "epoch": 0.098419189453125,
      "grad_norm": 14.094369888305664,
      "learning_rate": 4.8836327780385e-06,
      "loss": 3.6854,
      "step": 9675
    },
    {
      "epoch": 0.09847005208333333,
      "grad_norm": 12.182512283325195,
      "learning_rate": 4.883512166921104e-06,
      "loss": 3.5938,
      "step": 9680
    },
    {
      "epoch": 0.09852091471354167,
      "grad_norm": 11.643973350524902,
      "learning_rate": 4.883391494822114e-06,
      "loss": 3.3795,
      "step": 9685
    },
    {
      "epoch": 0.09857177734375,
      "grad_norm": 13.109829902648926,
      "learning_rate": 4.883270761744617e-06,
      "loss": 3.4673,
      "step": 9690
    },
    {
      "epoch": 0.09862263997395833,
      "grad_norm": 9.588884353637695,
      "learning_rate": 4.883149967691704e-06,
      "loss": 3.4358,
      "step": 9695
    },
    {
      "epoch": 0.09867350260416667,
      "grad_norm": 16.999698638916016,
      "learning_rate": 4.883029112666463e-06,
      "loss": 3.4918,
      "step": 9700
    },
    {
      "epoch": 0.098724365234375,
      "grad_norm": 14.010063171386719,
      "learning_rate": 4.882908196671987e-06,
      "loss": 3.5411,
      "step": 9705
    },
    {
      "epoch": 0.09877522786458333,
      "grad_norm": 14.857710838317871,
      "learning_rate": 4.88278721971137e-06,
      "loss": 3.4874,
      "step": 9710
    },
    {
      "epoch": 0.09882609049479167,
      "grad_norm": 12.407607078552246,
      "learning_rate": 4.882666181787707e-06,
      "loss": 3.2071,
      "step": 9715
    },
    {
      "epoch": 0.098876953125,
      "grad_norm": 12.295507431030273,
      "learning_rate": 4.882545082904094e-06,
      "loss": 3.3899,
      "step": 9720
    },
    {
      "epoch": 0.09892781575520833,
      "grad_norm": 12.28351879119873,
      "learning_rate": 4.88242392306363e-06,
      "loss": 3.8177,
      "step": 9725
    },
    {
      "epoch": 0.09897867838541667,
      "grad_norm": 8.689821243286133,
      "learning_rate": 4.882302702269415e-06,
      "loss": 3.5002,
      "step": 9730
    },
    {
      "epoch": 0.099029541015625,
      "grad_norm": 10.74190616607666,
      "learning_rate": 4.882181420524548e-06,
      "loss": 3.1903,
      "step": 9735
    },
    {
      "epoch": 0.09908040364583333,
      "grad_norm": 11.707921981811523,
      "learning_rate": 4.882060077832137e-06,
      "loss": 3.2921,
      "step": 9740
    },
    {
      "epoch": 0.09913126627604167,
      "grad_norm": 13.249473571777344,
      "learning_rate": 4.881938674195282e-06,
      "loss": 3.3386,
      "step": 9745
    },
    {
      "epoch": 0.09918212890625,
      "grad_norm": 16.1980037689209,
      "learning_rate": 4.88181720961709e-06,
      "loss": 3.5874,
      "step": 9750
    },
    {
      "epoch": 0.09923299153645833,
      "grad_norm": 14.430301666259766,
      "learning_rate": 4.88169568410067e-06,
      "loss": 3.2323,
      "step": 9755
    },
    {
      "epoch": 0.09928385416666667,
      "grad_norm": 8.47429084777832,
      "learning_rate": 4.881574097649131e-06,
      "loss": 3.2952,
      "step": 9760
    },
    {
      "epoch": 0.099334716796875,
      "grad_norm": 8.88611888885498,
      "learning_rate": 4.881452450265583e-06,
      "loss": 3.3712,
      "step": 9765
    },
    {
      "epoch": 0.09938557942708333,
      "grad_norm": 10.999044418334961,
      "learning_rate": 4.881330741953137e-06,
      "loss": 3.504,
      "step": 9770
    },
    {
      "epoch": 0.09943644205729167,
      "grad_norm": 13.003782272338867,
      "learning_rate": 4.88120897271491e-06,
      "loss": 3.8129,
      "step": 9775
    },
    {
      "epoch": 0.0994873046875,
      "grad_norm": 11.573512077331543,
      "learning_rate": 4.881087142554015e-06,
      "loss": 3.3195,
      "step": 9780
    },
    {
      "epoch": 0.09953816731770833,
      "grad_norm": 16.704687118530273,
      "learning_rate": 4.880965251473571e-06,
      "loss": 3.0435,
      "step": 9785
    },
    {
      "epoch": 0.09958902994791667,
      "grad_norm": 13.198221206665039,
      "learning_rate": 4.8808432994766944e-06,
      "loss": 3.2013,
      "step": 9790
    },
    {
      "epoch": 0.099639892578125,
      "grad_norm": 9.851302146911621,
      "learning_rate": 4.880721286566506e-06,
      "loss": 3.3261,
      "step": 9795
    },
    {
      "epoch": 0.09969075520833333,
      "grad_norm": 14.158812522888184,
      "learning_rate": 4.880599212746128e-06,
      "loss": 3.2733,
      "step": 9800
    },
    {
      "epoch": 0.09974161783854167,
      "grad_norm": 15.338472366333008,
      "learning_rate": 4.880477078018684e-06,
      "loss": 3.3534,
      "step": 9805
    },
    {
      "epoch": 0.09979248046875,
      "grad_norm": 14.599812507629395,
      "learning_rate": 4.8803548823872985e-06,
      "loss": 3.6244,
      "step": 9810
    },
    {
      "epoch": 0.09984334309895833,
      "grad_norm": 11.007782936096191,
      "learning_rate": 4.880232625855096e-06,
      "loss": 3.0961,
      "step": 9815
    },
    {
      "epoch": 0.09989420572916667,
      "grad_norm": 13.516772270202637,
      "learning_rate": 4.880110308425207e-06,
      "loss": 3.3936,
      "step": 9820
    },
    {
      "epoch": 0.099945068359375,
      "grad_norm": 7.3785905838012695,
      "learning_rate": 4.8799879301007596e-06,
      "loss": 3.7464,
      "step": 9825
    },
    {
      "epoch": 0.09999593098958333,
      "grad_norm": 38.93828582763672,
      "learning_rate": 4.879865490884886e-06,
      "loss": 3.5595,
      "step": 9830
    },
    {
      "epoch": 0.10004679361979167,
      "grad_norm": 14.068145751953125,
      "learning_rate": 4.879742990780717e-06,
      "loss": 3.5197,
      "step": 9835
    },
    {
      "epoch": 0.10009765625,
      "grad_norm": 11.864350318908691,
      "learning_rate": 4.879620429791387e-06,
      "loss": 3.601,
      "step": 9840
    },
    {
      "epoch": 0.10014851888020833,
      "grad_norm": 15.89621639251709,
      "learning_rate": 4.879497807920034e-06,
      "loss": 3.2539,
      "step": 9845
    },
    {
      "epoch": 0.10019938151041667,
      "grad_norm": 11.07425308227539,
      "learning_rate": 4.8793751251697925e-06,
      "loss": 3.5963,
      "step": 9850
    },
    {
      "epoch": 0.100250244140625,
      "grad_norm": 10.702958106994629,
      "learning_rate": 4.879252381543803e-06,
      "loss": 3.6547,
      "step": 9855
    },
    {
      "epoch": 0.10030110677083333,
      "grad_norm": 11.077383995056152,
      "learning_rate": 4.879129577045204e-06,
      "loss": 3.3366,
      "step": 9860
    },
    {
      "epoch": 0.10035196940104167,
      "grad_norm": 11.844645500183105,
      "learning_rate": 4.87900671167714e-06,
      "loss": 3.377,
      "step": 9865
    },
    {
      "epoch": 0.10040283203125,
      "grad_norm": 12.209272384643555,
      "learning_rate": 4.8788837854427525e-06,
      "loss": 4.0542,
      "step": 9870
    },
    {
      "epoch": 0.10045369466145833,
      "grad_norm": 12.173541069030762,
      "learning_rate": 4.878760798345188e-06,
      "loss": 3.5212,
      "step": 9875
    },
    {
      "epoch": 0.10050455729166667,
      "grad_norm": 12.681471824645996,
      "learning_rate": 4.878637750387591e-06,
      "loss": 3.8139,
      "step": 9880
    },
    {
      "epoch": 0.100555419921875,
      "grad_norm": 8.942806243896484,
      "learning_rate": 4.878514641573112e-06,
      "loss": 3.5788,
      "step": 9885
    },
    {
      "epoch": 0.10060628255208333,
      "grad_norm": 11.298273086547852,
      "learning_rate": 4.8783914719048995e-06,
      "loss": 3.3029,
      "step": 9890
    },
    {
      "epoch": 0.10065714518229167,
      "grad_norm": 9.567761421203613,
      "learning_rate": 4.8782682413861046e-06,
      "loss": 3.7846,
      "step": 9895
    },
    {
      "epoch": 0.1007080078125,
      "grad_norm": 10.052159309387207,
      "learning_rate": 4.8781449500198804e-06,
      "loss": 3.3519,
      "step": 9900
    },
    {
      "epoch": 0.10075887044270833,
      "grad_norm": 16.80272674560547,
      "learning_rate": 4.878021597809382e-06,
      "loss": 3.7498,
      "step": 9905
    },
    {
      "epoch": 0.10080973307291667,
      "grad_norm": 11.9678316116333,
      "learning_rate": 4.877898184757765e-06,
      "loss": 3.7687,
      "step": 9910
    },
    {
      "epoch": 0.100860595703125,
      "grad_norm": 14.82444953918457,
      "learning_rate": 4.877774710868185e-06,
      "loss": 3.4121,
      "step": 9915
    },
    {
      "epoch": 0.10091145833333333,
      "grad_norm": 12.413556098937988,
      "learning_rate": 4.877651176143804e-06,
      "loss": 3.5701,
      "step": 9920
    },
    {
      "epoch": 0.10096232096354167,
      "grad_norm": 16.419466018676758,
      "learning_rate": 4.877527580587781e-06,
      "loss": 3.5236,
      "step": 9925
    },
    {
      "epoch": 0.10101318359375,
      "grad_norm": 15.198001861572266,
      "learning_rate": 4.877403924203278e-06,
      "loss": 3.4151,
      "step": 9930
    },
    {
      "epoch": 0.10106404622395833,
      "grad_norm": 11.96876335144043,
      "learning_rate": 4.877280206993459e-06,
      "loss": 3.1713,
      "step": 9935
    },
    {
      "epoch": 0.10111490885416667,
      "grad_norm": 12.78167724609375,
      "learning_rate": 4.8771564289614895e-06,
      "loss": 3.3529,
      "step": 9940
    },
    {
      "epoch": 0.101165771484375,
      "grad_norm": 14.186490058898926,
      "learning_rate": 4.877032590110536e-06,
      "loss": 3.8994,
      "step": 9945
    },
    {
      "epoch": 0.10121663411458333,
      "grad_norm": 9.95617389678955,
      "learning_rate": 4.876908690443767e-06,
      "loss": 3.402,
      "step": 9950
    },
    {
      "epoch": 0.10126749674479167,
      "grad_norm": 14.145368576049805,
      "learning_rate": 4.876784729964353e-06,
      "loss": 3.5655,
      "step": 9955
    },
    {
      "epoch": 0.101318359375,
      "grad_norm": 10.90357780456543,
      "learning_rate": 4.876660708675465e-06,
      "loss": 3.4442,
      "step": 9960
    },
    {
      "epoch": 0.10136922200520833,
      "grad_norm": 13.902193069458008,
      "learning_rate": 4.876536626580276e-06,
      "loss": 3.3617,
      "step": 9965
    },
    {
      "epoch": 0.10142008463541667,
      "grad_norm": 8.821534156799316,
      "learning_rate": 4.876412483681961e-06,
      "loss": 3.445,
      "step": 9970
    },
    {
      "epoch": 0.101470947265625,
      "grad_norm": 11.418331146240234,
      "learning_rate": 4.8762882799836955e-06,
      "loss": 3.4656,
      "step": 9975
    },
    {
      "epoch": 0.10152180989583333,
      "grad_norm": 11.632497787475586,
      "learning_rate": 4.876164015488658e-06,
      "loss": 3.2188,
      "step": 9980
    },
    {
      "epoch": 0.10157267252604167,
      "grad_norm": 10.307924270629883,
      "learning_rate": 4.876039690200027e-06,
      "loss": 3.5271,
      "step": 9985
    },
    {
      "epoch": 0.10162353515625,
      "grad_norm": 11.517049789428711,
      "learning_rate": 4.875915304120984e-06,
      "loss": 3.5669,
      "step": 9990
    },
    {
      "epoch": 0.10167439778645833,
      "grad_norm": 9.514555931091309,
      "learning_rate": 4.875790857254711e-06,
      "loss": 3.3767,
      "step": 9995
    },
    {
      "epoch": 0.10172526041666667,
      "grad_norm": 13.482763290405273,
      "learning_rate": 4.875666349604392e-06,
      "loss": 3.5474,
      "step": 10000
    },
    {
      "epoch": 0.101776123046875,
      "grad_norm": 8.935997009277344,
      "learning_rate": 4.875541781173212e-06,
      "loss": 3.4632,
      "step": 10005
    },
    {
      "epoch": 0.10182698567708333,
      "grad_norm": 8.468302726745605,
      "learning_rate": 4.875417151964359e-06,
      "loss": 3.4322,
      "step": 10010
    },
    {
      "epoch": 0.10187784830729167,
      "grad_norm": 12.212846755981445,
      "learning_rate": 4.875292461981022e-06,
      "loss": 4.2413,
      "step": 10015
    },
    {
      "epoch": 0.1019287109375,
      "grad_norm": 11.472529411315918,
      "learning_rate": 4.87516771122639e-06,
      "loss": 3.2805,
      "step": 10020
    },
    {
      "epoch": 0.10197957356770833,
      "grad_norm": 13.230598449707031,
      "learning_rate": 4.875042899703654e-06,
      "loss": 3.4256,
      "step": 10025
    },
    {
      "epoch": 0.10203043619791667,
      "grad_norm": 13.139010429382324,
      "learning_rate": 4.874918027416009e-06,
      "loss": 3.5081,
      "step": 10030
    },
    {
      "epoch": 0.102081298828125,
      "grad_norm": 6.4391703605651855,
      "learning_rate": 4.874793094366649e-06,
      "loss": 3.3638,
      "step": 10035
    },
    {
      "epoch": 0.10213216145833333,
      "grad_norm": 15.451169967651367,
      "learning_rate": 4.8746681005587715e-06,
      "loss": 3.6243,
      "step": 10040
    },
    {
      "epoch": 0.10218302408854167,
      "grad_norm": 11.433987617492676,
      "learning_rate": 4.874543045995572e-06,
      "loss": 3.6848,
      "step": 10045
    },
    {
      "epoch": 0.10223388671875,
      "grad_norm": 8.72944450378418,
      "learning_rate": 4.874417930680253e-06,
      "loss": 3.3524,
      "step": 10050
    },
    {
      "epoch": 0.10228474934895833,
      "grad_norm": 15.81502628326416,
      "learning_rate": 4.874292754616014e-06,
      "loss": 3.3923,
      "step": 10055
    },
    {
      "epoch": 0.10233561197916667,
      "grad_norm": 12.636479377746582,
      "learning_rate": 4.8741675178060565e-06,
      "loss": 3.2582,
      "step": 10060
    },
    {
      "epoch": 0.102386474609375,
      "grad_norm": 10.026837348937988,
      "learning_rate": 4.874042220253586e-06,
      "loss": 3.5174,
      "step": 10065
    },
    {
      "epoch": 0.10243733723958333,
      "grad_norm": 14.83649730682373,
      "learning_rate": 4.8739168619618086e-06,
      "loss": 3.5073,
      "step": 10070
    },
    {
      "epoch": 0.10248819986979167,
      "grad_norm": 15.73272705078125,
      "learning_rate": 4.873791442933931e-06,
      "loss": 3.3724,
      "step": 10075
    },
    {
      "epoch": 0.1025390625,
      "grad_norm": 15.510259628295898,
      "learning_rate": 4.873665963173161e-06,
      "loss": 3.1763,
      "step": 10080
    },
    {
      "epoch": 0.10258992513020833,
      "grad_norm": 8.296263694763184,
      "learning_rate": 4.873540422682711e-06,
      "loss": 3.4604,
      "step": 10085
    },
    {
      "epoch": 0.10264078776041667,
      "grad_norm": 11.954290390014648,
      "learning_rate": 4.873414821465792e-06,
      "loss": 3.4896,
      "step": 10090
    },
    {
      "epoch": 0.102691650390625,
      "grad_norm": 10.191630363464355,
      "learning_rate": 4.873289159525617e-06,
      "loss": 3.5075,
      "step": 10095
    },
    {
      "epoch": 0.10274251302083333,
      "grad_norm": 14.362115859985352,
      "learning_rate": 4.873163436865401e-06,
      "loss": 3.0294,
      "step": 10100
    },
    {
      "epoch": 0.10279337565104167,
      "grad_norm": 10.870323181152344,
      "learning_rate": 4.873037653488361e-06,
      "loss": 3.5388,
      "step": 10105
    },
    {
      "epoch": 0.10284423828125,
      "grad_norm": 14.673710823059082,
      "learning_rate": 4.872911809397715e-06,
      "loss": 3.2948,
      "step": 10110
    },
    {
      "epoch": 0.10289510091145833,
      "grad_norm": 10.174817085266113,
      "learning_rate": 4.872785904596684e-06,
      "loss": 3.831,
      "step": 10115
    },
    {
      "epoch": 0.10294596354166667,
      "grad_norm": 13.638348579406738,
      "learning_rate": 4.8726599390884866e-06,
      "loss": 3.2765,
      "step": 10120
    },
    {
      "epoch": 0.102996826171875,
      "grad_norm": 14.297739028930664,
      "learning_rate": 4.872533912876348e-06,
      "loss": 3.4857,
      "step": 10125
    },
    {
      "epoch": 0.10304768880208333,
      "grad_norm": 8.859086036682129,
      "learning_rate": 4.872407825963491e-06,
      "loss": 3.5332,
      "step": 10130
    },
    {
      "epoch": 0.10309855143229167,
      "grad_norm": 17.016996383666992,
      "learning_rate": 4.872281678353142e-06,
      "loss": 3.858,
      "step": 10135
    },
    {
      "epoch": 0.1031494140625,
      "grad_norm": 7.245146751403809,
      "learning_rate": 4.872155470048529e-06,
      "loss": 3.3026,
      "step": 10140
    },
    {
      "epoch": 0.10320027669270833,
      "grad_norm": 13.683245658874512,
      "learning_rate": 4.87202920105288e-06,
      "loss": 3.5045,
      "step": 10145
    },
    {
      "epoch": 0.10325113932291667,
      "grad_norm": 10.41574478149414,
      "learning_rate": 4.871902871369427e-06,
      "loss": 3.289,
      "step": 10150
    },
    {
      "epoch": 0.103302001953125,
      "grad_norm": 13.628034591674805,
      "learning_rate": 4.871776481001401e-06,
      "loss": 3.6961,
      "step": 10155
    },
    {
      "epoch": 0.10335286458333333,
      "grad_norm": 9.731292724609375,
      "learning_rate": 4.8716500299520356e-06,
      "loss": 3.4743,
      "step": 10160
    },
    {
      "epoch": 0.10340372721354167,
      "grad_norm": 15.618063926696777,
      "learning_rate": 4.871523518224567e-06,
      "loss": 3.0353,
      "step": 10165
    },
    {
      "epoch": 0.10345458984375,
      "grad_norm": 12.420486450195312,
      "learning_rate": 4.87139694582223e-06,
      "loss": 3.5414,
      "step": 10170
    },
    {
      "epoch": 0.10350545247395833,
      "grad_norm": 9.918728828430176,
      "learning_rate": 4.871270312748265e-06,
      "loss": 3.6799,
      "step": 10175
    },
    {
      "epoch": 0.10355631510416667,
      "grad_norm": 10.1067533493042,
      "learning_rate": 4.871143619005911e-06,
      "loss": 3.4345,
      "step": 10180
    },
    {
      "epoch": 0.103607177734375,
      "grad_norm": 6.703737735748291,
      "learning_rate": 4.87101686459841e-06,
      "loss": 3.2298,
      "step": 10185
    },
    {
      "epoch": 0.10365804036458333,
      "grad_norm": 15.222867012023926,
      "learning_rate": 4.8708900495290035e-06,
      "loss": 3.6949,
      "step": 10190
    },
    {
      "epoch": 0.10370890299479167,
      "grad_norm": 12.256555557250977,
      "learning_rate": 4.8707631738009376e-06,
      "loss": 3.4872,
      "step": 10195
    },
    {
      "epoch": 0.103759765625,
      "grad_norm": 8.05500602722168,
      "learning_rate": 4.870636237417458e-06,
      "loss": 3.5136,
      "step": 10200
    },
    {
      "epoch": 0.10381062825520833,
      "grad_norm": 9.914165496826172,
      "learning_rate": 4.870509240381812e-06,
      "loss": 3.5153,
      "step": 10205
    },
    {
      "epoch": 0.10386149088541667,
      "grad_norm": 14.44861125946045,
      "learning_rate": 4.8703821826972495e-06,
      "loss": 3.4795,
      "step": 10210
    },
    {
      "epoch": 0.103912353515625,
      "grad_norm": 12.164118766784668,
      "learning_rate": 4.87025506436702e-06,
      "loss": 3.4607,
      "step": 10215
    },
    {
      "epoch": 0.10396321614583333,
      "grad_norm": 15.556924819946289,
      "learning_rate": 4.8701278853943764e-06,
      "loss": 3.3538,
      "step": 10220
    },
    {
      "epoch": 0.10401407877604167,
      "grad_norm": 14.256190299987793,
      "learning_rate": 4.870000645782573e-06,
      "loss": 3.3421,
      "step": 10225
    },
    {
      "epoch": 0.10406494140625,
      "grad_norm": 13.732766151428223,
      "learning_rate": 4.869873345534865e-06,
      "loss": 3.0725,
      "step": 10230
    },
    {
      "epoch": 0.10411580403645833,
      "grad_norm": 16.038616180419922,
      "learning_rate": 4.869745984654508e-06,
      "loss": 3.212,
      "step": 10235
    },
    {
      "epoch": 0.10416666666666667,
      "grad_norm": 14.277772903442383,
      "learning_rate": 4.8696185631447635e-06,
      "loss": 3.6345,
      "step": 10240
    },
    {
      "epoch": 0.104217529296875,
      "grad_norm": 10.199970245361328,
      "learning_rate": 4.869491081008889e-06,
      "loss": 3.5137,
      "step": 10245
    },
    {
      "epoch": 0.10426839192708333,
      "grad_norm": 12.390953063964844,
      "learning_rate": 4.869363538250146e-06,
      "loss": 3.3681,
      "step": 10250
    },
    {
      "epoch": 0.10431925455729167,
      "grad_norm": 8.647974967956543,
      "learning_rate": 4.869235934871799e-06,
      "loss": 3.2399,
      "step": 10255
    },
    {
      "epoch": 0.1043701171875,
      "grad_norm": 11.923579216003418,
      "learning_rate": 4.869108270877112e-06,
      "loss": 3.6247,
      "step": 10260
    },
    {
      "epoch": 0.10442097981770833,
      "grad_norm": 13.356016159057617,
      "learning_rate": 4.868980546269352e-06,
      "loss": 3.7096,
      "step": 10265
    },
    {
      "epoch": 0.10447184244791667,
      "grad_norm": 8.816129684448242,
      "learning_rate": 4.868852761051787e-06,
      "loss": 3.3696,
      "step": 10270
    },
    {
      "epoch": 0.104522705078125,
      "grad_norm": 11.433305740356445,
      "learning_rate": 4.868724915227684e-06,
      "loss": 3.3735,
      "step": 10275
    },
    {
      "epoch": 0.10457356770833333,
      "grad_norm": 10.5938138961792,
      "learning_rate": 4.868597008800315e-06,
      "loss": 3.5073,
      "step": 10280
    },
    {
      "epoch": 0.10462443033854167,
      "grad_norm": 13.988579750061035,
      "learning_rate": 4.868469041772955e-06,
      "loss": 3.6398,
      "step": 10285
    },
    {
      "epoch": 0.10467529296875,
      "grad_norm": 8.426319122314453,
      "learning_rate": 4.868341014148875e-06,
      "loss": 3.5602,
      "step": 10290
    },
    {
      "epoch": 0.10472615559895833,
      "grad_norm": 12.661907196044922,
      "learning_rate": 4.868212925931351e-06,
      "loss": 3.3872,
      "step": 10295
    },
    {
      "epoch": 0.10477701822916667,
      "grad_norm": 10.294717788696289,
      "learning_rate": 4.868084777123661e-06,
      "loss": 3.5918,
      "step": 10300
    },
    {
      "epoch": 0.104827880859375,
      "grad_norm": 16.74044418334961,
      "learning_rate": 4.867956567729084e-06,
      "loss": 3.4537,
      "step": 10305
    },
    {
      "epoch": 0.10487874348958333,
      "grad_norm": 10.488282203674316,
      "learning_rate": 4.867828297750899e-06,
      "loss": 3.101,
      "step": 10310
    },
    {
      "epoch": 0.10492960611979167,
      "grad_norm": 10.39643669128418,
      "learning_rate": 4.867699967192388e-06,
      "loss": 3.1673,
      "step": 10315
    },
    {
      "epoch": 0.10498046875,
      "grad_norm": 13.151554107666016,
      "learning_rate": 4.867571576056834e-06,
      "loss": 3.6526,
      "step": 10320
    },
    {
      "epoch": 0.10503133138020833,
      "grad_norm": 11.318380355834961,
      "learning_rate": 4.867443124347523e-06,
      "loss": 3.6569,
      "step": 10325
    },
    {
      "epoch": 0.10508219401041667,
      "grad_norm": 11.30242919921875,
      "learning_rate": 4.867314612067741e-06,
      "loss": 3.2106,
      "step": 10330
    },
    {
      "epoch": 0.105133056640625,
      "grad_norm": 8.88015079498291,
      "learning_rate": 4.867186039220775e-06,
      "loss": 3.3928,
      "step": 10335
    },
    {
      "epoch": 0.10518391927083333,
      "grad_norm": 9.858390808105469,
      "learning_rate": 4.867057405809916e-06,
      "loss": 3.3096,
      "step": 10340
    },
    {
      "epoch": 0.10523478190104167,
      "grad_norm": 14.723819732666016,
      "learning_rate": 4.866928711838455e-06,
      "loss": 3.3486,
      "step": 10345
    },
    {
      "epoch": 0.10528564453125,
      "grad_norm": 8.097146034240723,
      "learning_rate": 4.866799957309682e-06,
      "loss": 4.0018,
      "step": 10350
    },
    {
      "epoch": 0.10533650716145833,
      "grad_norm": 14.983758926391602,
      "learning_rate": 4.866671142226895e-06,
      "loss": 3.5232,
      "step": 10355
    },
    {
      "epoch": 0.10538736979166667,
      "grad_norm": 10.44787311553955,
      "learning_rate": 4.866542266593387e-06,
      "loss": 3.6857,
      "step": 10360
    },
    {
      "epoch": 0.105438232421875,
      "grad_norm": 13.872020721435547,
      "learning_rate": 4.8664133304124555e-06,
      "loss": 3.2874,
      "step": 10365
    },
    {
      "epoch": 0.10548909505208333,
      "grad_norm": 11.341287612915039,
      "learning_rate": 4.8662843336874e-06,
      "loss": 3.202,
      "step": 10370
    },
    {
      "epoch": 0.10553995768229167,
      "grad_norm": 9.611438751220703,
      "learning_rate": 4.866155276421522e-06,
      "loss": 3.4759,
      "step": 10375
    },
    {
      "epoch": 0.1055908203125,
      "grad_norm": 12.784802436828613,
      "learning_rate": 4.8660261586181205e-06,
      "loss": 3.3239,
      "step": 10380
    },
    {
      "epoch": 0.10564168294270833,
      "grad_norm": 15.74527645111084,
      "learning_rate": 4.865896980280501e-06,
      "loss": 3.6173,
      "step": 10385
    },
    {
      "epoch": 0.10569254557291667,
      "grad_norm": 9.02971076965332,
      "learning_rate": 4.865767741411969e-06,
      "loss": 3.7378,
      "step": 10390
    },
    {
      "epoch": 0.105743408203125,
      "grad_norm": 8.090533256530762,
      "learning_rate": 4.8656384420158285e-06,
      "loss": 3.2215,
      "step": 10395
    },
    {
      "epoch": 0.10579427083333333,
      "grad_norm": 9.529847145080566,
      "learning_rate": 4.86550908209539e-06,
      "loss": 3.3138,
      "step": 10400
    },
    {
      "epoch": 0.10584513346354167,
      "grad_norm": 11.145707130432129,
      "learning_rate": 4.865379661653963e-06,
      "loss": 2.8438,
      "step": 10405
    },
    {
      "epoch": 0.10589599609375,
      "grad_norm": 9.50683879852295,
      "learning_rate": 4.8652501806948575e-06,
      "loss": 3.4143,
      "step": 10410
    },
    {
      "epoch": 0.10594685872395833,
      "grad_norm": 14.874054908752441,
      "learning_rate": 4.865120639221386e-06,
      "loss": 3.6896,
      "step": 10415
    },
    {
      "epoch": 0.10599772135416667,
      "grad_norm": 15.502473831176758,
      "learning_rate": 4.864991037236864e-06,
      "loss": 3.7117,
      "step": 10420
    },
    {
      "epoch": 0.106048583984375,
      "grad_norm": 17.096446990966797,
      "learning_rate": 4.864861374744607e-06,
      "loss": 3.4389,
      "step": 10425
    },
    {
      "epoch": 0.10609944661458333,
      "grad_norm": 12.414908409118652,
      "learning_rate": 4.8647316517479326e-06,
      "loss": 3.5158,
      "step": 10430
    },
    {
      "epoch": 0.10615030924479167,
      "grad_norm": 13.846083641052246,
      "learning_rate": 4.864601868250159e-06,
      "loss": 3.4447,
      "step": 10435
    },
    {
      "epoch": 0.106201171875,
      "grad_norm": 12.67673110961914,
      "learning_rate": 4.864472024254607e-06,
      "loss": 3.0979,
      "step": 10440
    },
    {
      "epoch": 0.10625203450520833,
      "grad_norm": 13.106204986572266,
      "learning_rate": 4.864342119764599e-06,
      "loss": 3.4044,
      "step": 10445
    },
    {
      "epoch": 0.10630289713541667,
      "grad_norm": 10.791247367858887,
      "learning_rate": 4.864212154783458e-06,
      "loss": 3.4244,
      "step": 10450
    },
    {
      "epoch": 0.106353759765625,
      "grad_norm": 15.417497634887695,
      "learning_rate": 4.86408212931451e-06,
      "loss": 3.8919,
      "step": 10455
    },
    {
      "epoch": 0.10640462239583333,
      "grad_norm": 12.463935852050781,
      "learning_rate": 4.86395204336108e-06,
      "loss": 3.3458,
      "step": 10460
    },
    {
      "epoch": 0.10645548502604167,
      "grad_norm": 11.733417510986328,
      "learning_rate": 4.863821896926498e-06,
      "loss": 3.5288,
      "step": 10465
    },
    {
      "epoch": 0.10650634765625,
      "grad_norm": 14.689102172851562,
      "learning_rate": 4.863691690014093e-06,
      "loss": 3.2967,
      "step": 10470
    },
    {
      "epoch": 0.10655721028645833,
      "grad_norm": 10.164743423461914,
      "learning_rate": 4.863561422627197e-06,
      "loss": 3.2024,
      "step": 10475
    },
    {
      "epoch": 0.10660807291666667,
      "grad_norm": 10.591276168823242,
      "learning_rate": 4.863431094769141e-06,
      "loss": 3.1854,
      "step": 10480
    },
    {
      "epoch": 0.106658935546875,
      "grad_norm": 7.265041351318359,
      "learning_rate": 4.863300706443261e-06,
      "loss": 3.3769,
      "step": 10485
    },
    {
      "epoch": 0.10670979817708333,
      "grad_norm": 9.011372566223145,
      "learning_rate": 4.8631702576528924e-06,
      "loss": 3.3535,
      "step": 10490
    },
    {
      "epoch": 0.10676066080729167,
      "grad_norm": 12.629626274108887,
      "learning_rate": 4.863039748401374e-06,
      "loss": 3.1713,
      "step": 10495
    },
    {
      "epoch": 0.1068115234375,
      "grad_norm": 10.04238224029541,
      "learning_rate": 4.8629091786920425e-06,
      "loss": 3.2648,
      "step": 10500
    },
    {
      "epoch": 0.10686238606770833,
      "grad_norm": 8.861063003540039,
      "learning_rate": 4.862778548528239e-06,
      "loss": 3.4088,
      "step": 10505
    },
    {
      "epoch": 0.10691324869791667,
      "grad_norm": 10.639623641967773,
      "learning_rate": 4.862647857913308e-06,
      "loss": 3.1807,
      "step": 10510
    },
    {
      "epoch": 0.106964111328125,
      "grad_norm": 11.753366470336914,
      "learning_rate": 4.862517106850592e-06,
      "loss": 4.0414,
      "step": 10515
    },
    {
      "epoch": 0.10701497395833333,
      "grad_norm": 6.897514343261719,
      "learning_rate": 4.862386295343435e-06,
      "loss": 3.1811,
      "step": 10520
    },
    {
      "epoch": 0.10706583658854167,
      "grad_norm": 17.118993759155273,
      "learning_rate": 4.862255423395184e-06,
      "loss": 3.549,
      "step": 10525
    },
    {
      "epoch": 0.10711669921875,
      "grad_norm": 12.017741203308105,
      "learning_rate": 4.862124491009188e-06,
      "loss": 3.4445,
      "step": 10530
    },
    {
      "epoch": 0.10716756184895833,
      "grad_norm": 18.230016708374023,
      "learning_rate": 4.861993498188798e-06,
      "loss": 3.381,
      "step": 10535
    },
    {
      "epoch": 0.10721842447916667,
      "grad_norm": 16.504117965698242,
      "learning_rate": 4.861862444937363e-06,
      "loss": 4.0049,
      "step": 10540
    },
    {
      "epoch": 0.107269287109375,
      "grad_norm": 13.888938903808594,
      "learning_rate": 4.861731331258238e-06,
      "loss": 3.5727,
      "step": 10545
    },
    {
      "epoch": 0.10732014973958333,
      "grad_norm": 14.863890647888184,
      "learning_rate": 4.8616001571547764e-06,
      "loss": 3.4779,
      "step": 10550
    },
    {
      "epoch": 0.10737101236979167,
      "grad_norm": 13.943183898925781,
      "learning_rate": 4.8614689226303345e-06,
      "loss": 3.602,
      "step": 10555
    },
    {
      "epoch": 0.107421875,
      "grad_norm": 10.1746826171875,
      "learning_rate": 4.86133762768827e-06,
      "loss": 3.4561,
      "step": 10560
    },
    {
      "epoch": 0.10747273763020833,
      "grad_norm": 12.866034507751465,
      "learning_rate": 4.861206272331941e-06,
      "loss": 3.2867,
      "step": 10565
    },
    {
      "epoch": 0.10752360026041667,
      "grad_norm": 13.17371654510498,
      "learning_rate": 4.86107485656471e-06,
      "loss": 3.4148,
      "step": 10570
    },
    {
      "epoch": 0.107574462890625,
      "grad_norm": 8.4532470703125,
      "learning_rate": 4.860943380389939e-06,
      "loss": 3.1673,
      "step": 10575
    },
    {
      "epoch": 0.10762532552083333,
      "grad_norm": 10.179365158081055,
      "learning_rate": 4.86081184381099e-06,
      "loss": 3.3304,
      "step": 10580
    },
    {
      "epoch": 0.10767618815104167,
      "grad_norm": 10.048255920410156,
      "learning_rate": 4.860680246831231e-06,
      "loss": 3.3539,
      "step": 10585
    },
    {
      "epoch": 0.10772705078125,
      "grad_norm": 14.505502700805664,
      "learning_rate": 4.860548589454026e-06,
      "loss": 3.5862,
      "step": 10590
    },
    {
      "epoch": 0.10777791341145833,
      "grad_norm": 11.778282165527344,
      "learning_rate": 4.860416871682746e-06,
      "loss": 3.3957,
      "step": 10595
    },
    {
      "epoch": 0.10782877604166667,
      "grad_norm": 10.966447830200195,
      "learning_rate": 4.860285093520759e-06,
      "loss": 3.2105,
      "step": 10600
    },
    {
      "epoch": 0.107879638671875,
      "grad_norm": 12.947134017944336,
      "learning_rate": 4.860153254971437e-06,
      "loss": 3.7268,
      "step": 10605
    },
    {
      "epoch": 0.10793050130208333,
      "grad_norm": 12.653708457946777,
      "learning_rate": 4.860021356038155e-06,
      "loss": 3.4977,
      "step": 10610
    },
    {
      "epoch": 0.10798136393229167,
      "grad_norm": 15.799676895141602,
      "learning_rate": 4.859889396724284e-06,
      "loss": 3.7318,
      "step": 10615
    },
    {
      "epoch": 0.1080322265625,
      "grad_norm": 10.117684364318848,
      "learning_rate": 4.859757377033204e-06,
      "loss": 3.8512,
      "step": 10620
    },
    {
      "epoch": 0.10808308919270833,
      "grad_norm": 16.469112396240234,
      "learning_rate": 4.85962529696829e-06,
      "loss": 3.6864,
      "step": 10625
    },
    {
      "epoch": 0.10813395182291667,
      "grad_norm": 8.12623119354248,
      "learning_rate": 4.859493156532922e-06,
      "loss": 3.162,
      "step": 10630
    },
    {
      "epoch": 0.108184814453125,
      "grad_norm": 10.442404747009277,
      "learning_rate": 4.859360955730481e-06,
      "loss": 3.3013,
      "step": 10635
    },
    {
      "epoch": 0.10823567708333333,
      "grad_norm": 13.673596382141113,
      "learning_rate": 4.8592286945643485e-06,
      "loss": 3.3599,
      "step": 10640
    },
    {
      "epoch": 0.10828653971354167,
      "grad_norm": 8.366315841674805,
      "learning_rate": 4.859096373037911e-06,
      "loss": 3.2563,
      "step": 10645
    },
    {
      "epoch": 0.10833740234375,
      "grad_norm": 16.115209579467773,
      "learning_rate": 4.8589639911545495e-06,
      "loss": 3.6421,
      "step": 10650
    },
    {
      "epoch": 0.10838826497395833,
      "grad_norm": 12.849785804748535,
      "learning_rate": 4.858831548917655e-06,
      "loss": 3.4333,
      "step": 10655
    },
    {
      "epoch": 0.10843912760416667,
      "grad_norm": 12.12121295928955,
      "learning_rate": 4.858699046330614e-06,
      "loss": 3.8339,
      "step": 10660
    },
    {
      "epoch": 0.108489990234375,
      "grad_norm": 12.373353004455566,
      "learning_rate": 4.858566483396816e-06,
      "loss": 3.3636,
      "step": 10665
    },
    {
      "epoch": 0.10854085286458333,
      "grad_norm": 13.366467475891113,
      "learning_rate": 4.858433860119655e-06,
      "loss": 3.1534,
      "step": 10670
    },
    {
      "epoch": 0.10859171549479167,
      "grad_norm": 11.035595893859863,
      "learning_rate": 4.858301176502522e-06,
      "loss": 3.0366,
      "step": 10675
    },
    {
      "epoch": 0.108642578125,
      "grad_norm": 9.009740829467773,
      "learning_rate": 4.858168432548813e-06,
      "loss": 3.1213,
      "step": 10680
    },
    {
      "epoch": 0.10869344075520833,
      "grad_norm": 13.181190490722656,
      "learning_rate": 4.858035628261924e-06,
      "loss": 3.2362,
      "step": 10685
    },
    {
      "epoch": 0.10874430338541667,
      "grad_norm": 12.755681991577148,
      "learning_rate": 4.85790276364525e-06,
      "loss": 3.493,
      "step": 10690
    },
    {
      "epoch": 0.108795166015625,
      "grad_norm": 16.021320343017578,
      "learning_rate": 4.857769838702195e-06,
      "loss": 3.3145,
      "step": 10695
    },
    {
      "epoch": 0.10884602864583333,
      "grad_norm": 11.281843185424805,
      "learning_rate": 4.857636853436156e-06,
      "loss": 3.4385,
      "step": 10700
    },
    {
      "epoch": 0.10889689127604167,
      "grad_norm": 13.89880657196045,
      "learning_rate": 4.857503807850538e-06,
      "loss": 3.5553,
      "step": 10705
    },
    {
      "epoch": 0.10894775390625,
      "grad_norm": 17.039382934570312,
      "learning_rate": 4.857370701948744e-06,
      "loss": 3.7388,
      "step": 10710
    },
    {
      "epoch": 0.10899861653645833,
      "grad_norm": 8.904913902282715,
      "learning_rate": 4.857237535734179e-06,
      "loss": 3.5905,
      "step": 10715
    },
    {
      "epoch": 0.10904947916666667,
      "grad_norm": 10.64655876159668,
      "learning_rate": 4.8571043092102496e-06,
      "loss": 3.174,
      "step": 10720
    },
    {
      "epoch": 0.109100341796875,
      "grad_norm": 11.749823570251465,
      "learning_rate": 4.856971022380366e-06,
      "loss": 3.295,
      "step": 10725
    },
    {
      "epoch": 0.10915120442708333,
      "grad_norm": 9.72337532043457,
      "learning_rate": 4.856837675247938e-06,
      "loss": 3.3855,
      "step": 10730
    },
    {
      "epoch": 0.10920206705729167,
      "grad_norm": 16.54507827758789,
      "learning_rate": 4.856704267816375e-06,
      "loss": 3.5955,
      "step": 10735
    },
    {
      "epoch": 0.1092529296875,
      "grad_norm": 13.975875854492188,
      "learning_rate": 4.856570800089093e-06,
      "loss": 3.3142,
      "step": 10740
    },
    {
      "epoch": 0.10930379231770833,
      "grad_norm": 15.316740989685059,
      "learning_rate": 4.856437272069506e-06,
      "loss": 3.4958,
      "step": 10745
    },
    {
      "epoch": 0.10935465494791667,
      "grad_norm": 10.500741004943848,
      "learning_rate": 4.856303683761029e-06,
      "loss": 3.2991,
      "step": 10750
    },
    {
      "epoch": 0.109405517578125,
      "grad_norm": 10.17525577545166,
      "learning_rate": 4.8561700351670815e-06,
      "loss": 3.7226,
      "step": 10755
    },
    {
      "epoch": 0.10945638020833333,
      "grad_norm": 12.4932861328125,
      "learning_rate": 4.856036326291082e-06,
      "loss": 3.2116,
      "step": 10760
    },
    {
      "epoch": 0.10950724283854167,
      "grad_norm": 9.614534378051758,
      "learning_rate": 4.855902557136451e-06,
      "loss": 3.351,
      "step": 10765
    },
    {
      "epoch": 0.10955810546875,
      "grad_norm": 14.387618064880371,
      "learning_rate": 4.855768727706613e-06,
      "loss": 3.5085,
      "step": 10770
    },
    {
      "epoch": 0.10960896809895833,
      "grad_norm": 11.664185523986816,
      "learning_rate": 4.855634838004988e-06,
      "loss": 3.2219,
      "step": 10775
    },
    {
      "epoch": 0.10965983072916667,
      "grad_norm": 8.379794120788574,
      "learning_rate": 4.8555008880350055e-06,
      "loss": 3.0816,
      "step": 10780
    },
    {
      "epoch": 0.109710693359375,
      "grad_norm": 14.983808517456055,
      "learning_rate": 4.8553668778000905e-06,
      "loss": 3.0428,
      "step": 10785
    },
    {
      "epoch": 0.10976155598958333,
      "grad_norm": 10.499786376953125,
      "learning_rate": 4.855232807303673e-06,
      "loss": 3.1102,
      "step": 10790
    },
    {
      "epoch": 0.10981241861979167,
      "grad_norm": 15.770118713378906,
      "learning_rate": 4.8550986765491825e-06,
      "loss": 4.0818,
      "step": 10795
    },
    {
      "epoch": 0.10986328125,
      "grad_norm": 13.66915225982666,
      "learning_rate": 4.85496448554005e-06,
      "loss": 3.7093,
      "step": 10800
    },
    {
      "epoch": 0.10991414388020833,
      "grad_norm": 8.571983337402344,
      "learning_rate": 4.85483023427971e-06,
      "loss": 3.0368,
      "step": 10805
    },
    {
      "epoch": 0.10996500651041667,
      "grad_norm": 15.759471893310547,
      "learning_rate": 4.854695922771595e-06,
      "loss": 3.5172,
      "step": 10810
    },
    {
      "epoch": 0.110015869140625,
      "grad_norm": 13.83719539642334,
      "learning_rate": 4.854561551019145e-06,
      "loss": 3.5774,
      "step": 10815
    },
    {
      "epoch": 0.11006673177083333,
      "grad_norm": 11.690563201904297,
      "learning_rate": 4.854427119025794e-06,
      "loss": 3.5709,
      "step": 10820
    },
    {
      "epoch": 0.11011759440104167,
      "grad_norm": 16.460613250732422,
      "learning_rate": 4.854292626794984e-06,
      "loss": 3.0922,
      "step": 10825
    },
    {
      "epoch": 0.11016845703125,
      "grad_norm": 9.609016418457031,
      "learning_rate": 4.854158074330155e-06,
      "loss": 3.6565,
      "step": 10830
    },
    {
      "epoch": 0.11021931966145833,
      "grad_norm": 8.492650985717773,
      "learning_rate": 4.85402346163475e-06,
      "loss": 3.0948,
      "step": 10835
    },
    {
      "epoch": 0.11027018229166667,
      "grad_norm": 8.52239990234375,
      "learning_rate": 4.853888788712213e-06,
      "loss": 3.6836,
      "step": 10840
    },
    {
      "epoch": 0.110321044921875,
      "grad_norm": 14.75137710571289,
      "learning_rate": 4.853754055565988e-06,
      "loss": 3.5731,
      "step": 10845
    },
    {
      "epoch": 0.11037190755208333,
      "grad_norm": 17.106433868408203,
      "learning_rate": 4.853619262199525e-06,
      "loss": 3.3211,
      "step": 10850
    },
    {
      "epoch": 0.11042277018229167,
      "grad_norm": 13.21174430847168,
      "learning_rate": 4.85348440861627e-06,
      "loss": 3.1972,
      "step": 10855
    },
    {
      "epoch": 0.1104736328125,
      "grad_norm": 15.949371337890625,
      "learning_rate": 4.8533494948196746e-06,
      "loss": 3.7361,
      "step": 10860
    },
    {
      "epoch": 0.11052449544270833,
      "grad_norm": 13.830065727233887,
      "learning_rate": 4.8532145208131894e-06,
      "loss": 3.4961,
      "step": 10865
    },
    {
      "epoch": 0.11057535807291667,
      "grad_norm": 14.657466888427734,
      "learning_rate": 4.85307948660027e-06,
      "loss": 3.3326,
      "step": 10870
    },
    {
      "epoch": 0.110626220703125,
      "grad_norm": 18.071182250976562,
      "learning_rate": 4.852944392184369e-06,
      "loss": 3.5664,
      "step": 10875
    },
    {
      "epoch": 0.11067708333333333,
      "grad_norm": 259.91387939453125,
      "learning_rate": 4.852809237568943e-06,
      "loss": 3.6806,
      "step": 10880
    },
    {
      "epoch": 0.11072794596354167,
      "grad_norm": 12.160555839538574,
      "learning_rate": 4.85267402275745e-06,
      "loss": 3.8418,
      "step": 10885
    },
    {
      "epoch": 0.11077880859375,
      "grad_norm": 15.426597595214844,
      "learning_rate": 4.852538747753351e-06,
      "loss": 3.5979,
      "step": 10890
    },
    {
      "epoch": 0.11082967122395833,
      "grad_norm": 13.01453685760498,
      "learning_rate": 4.852403412560105e-06,
      "loss": 3.2011,
      "step": 10895
    },
    {
      "epoch": 0.11088053385416667,
      "grad_norm": 10.577211380004883,
      "learning_rate": 4.852268017181176e-06,
      "loss": 3.727,
      "step": 10900
    },
    {
      "epoch": 0.110931396484375,
      "grad_norm": 14.287973403930664,
      "learning_rate": 4.852132561620026e-06,
      "loss": 3.758,
      "step": 10905
    },
    {
      "epoch": 0.11098225911458333,
      "grad_norm": 11.887785911560059,
      "learning_rate": 4.851997045880123e-06,
      "loss": 2.9819,
      "step": 10910
    },
    {
      "epoch": 0.11103312174479167,
      "grad_norm": 7.388796329498291,
      "learning_rate": 4.851861469964932e-06,
      "loss": 3.0688,
      "step": 10915
    },
    {
      "epoch": 0.111083984375,
      "grad_norm": 11.511301040649414,
      "learning_rate": 4.851725833877924e-06,
      "loss": 3.3384,
      "step": 10920
    },
    {
      "epoch": 0.11113484700520833,
      "grad_norm": 13.27392864227295,
      "learning_rate": 4.851590137622567e-06,
      "loss": 3.0957,
      "step": 10925
    },
    {
      "epoch": 0.11118570963541667,
      "grad_norm": 14.255321502685547,
      "learning_rate": 4.851454381202334e-06,
      "loss": 3.2306,
      "step": 10930
    },
    {
      "epoch": 0.111236572265625,
      "grad_norm": 11.45173168182373,
      "learning_rate": 4.851318564620699e-06,
      "loss": 3.4344,
      "step": 10935
    },
    {
      "epoch": 0.11128743489583333,
      "grad_norm": 12.557133674621582,
      "learning_rate": 4.851182687881134e-06,
      "loss": 3.4574,
      "step": 10940
    },
    {
      "epoch": 0.11133829752604167,
      "grad_norm": 12.226534843444824,
      "learning_rate": 4.851046750987118e-06,
      "loss": 3.8019,
      "step": 10945
    },
    {
      "epoch": 0.11138916015625,
      "grad_norm": 10.195042610168457,
      "learning_rate": 4.850910753942129e-06,
      "loss": 3.0214,
      "step": 10950
    },
    {
      "epoch": 0.11144002278645833,
      "grad_norm": 14.165132522583008,
      "learning_rate": 4.850774696749645e-06,
      "loss": 3.5724,
      "step": 10955
    },
    {
      "epoch": 0.11149088541666667,
      "grad_norm": 13.284117698669434,
      "learning_rate": 4.850638579413147e-06,
      "loss": 4.1911,
      "step": 10960
    },
    {
      "epoch": 0.111541748046875,
      "grad_norm": 16.30782127380371,
      "learning_rate": 4.850502401936119e-06,
      "loss": 3.1399,
      "step": 10965
    },
    {
      "epoch": 0.11159261067708333,
      "grad_norm": 13.27759838104248,
      "learning_rate": 4.850366164322044e-06,
      "loss": 3.4511,
      "step": 10970
    },
    {
      "epoch": 0.11164347330729167,
      "grad_norm": 12.456367492675781,
      "learning_rate": 4.850229866574407e-06,
      "loss": 3.5491,
      "step": 10975
    },
    {
      "epoch": 0.1116943359375,
      "grad_norm": 10.18445873260498,
      "learning_rate": 4.850093508696697e-06,
      "loss": 3.5397,
      "step": 10980
    },
    {
      "epoch": 0.11174519856770833,
      "grad_norm": 14.981690406799316,
      "learning_rate": 4.849957090692401e-06,
      "loss": 3.3078,
      "step": 10985
    },
    {
      "epoch": 0.11179606119791667,
      "grad_norm": 18.036338806152344,
      "learning_rate": 4.84982061256501e-06,
      "loss": 3.2716,
      "step": 10990
    },
    {
      "epoch": 0.111846923828125,
      "grad_norm": 18.29755973815918,
      "learning_rate": 4.849684074318015e-06,
      "loss": 3.6321,
      "step": 10995
    },
    {
      "epoch": 0.11189778645833333,
      "grad_norm": 13.462907791137695,
      "learning_rate": 4.849547475954911e-06,
      "loss": 3.3401,
      "step": 11000
    },
    {
      "epoch": 0.11194864908854167,
      "grad_norm": 9.00394058227539,
      "learning_rate": 4.849410817479191e-06,
      "loss": 3.6376,
      "step": 11005
    },
    {
      "epoch": 0.11199951171875,
      "grad_norm": 7.56473970413208,
      "learning_rate": 4.849274098894352e-06,
      "loss": 3.5239,
      "step": 11010
    },
    {
      "epoch": 0.11205037434895833,
      "grad_norm": 10.673871994018555,
      "learning_rate": 4.849137320203892e-06,
      "loss": 3.4489,
      "step": 11015
    },
    {
      "epoch": 0.11210123697916667,
      "grad_norm": 17.68508529663086,
      "learning_rate": 4.849000481411312e-06,
      "loss": 3.3573,
      "step": 11020
    },
    {
      "epoch": 0.112152099609375,
      "grad_norm": 13.00468921661377,
      "learning_rate": 4.84886358252011e-06,
      "loss": 3.3832,
      "step": 11025
    },
    {
      "epoch": 0.11220296223958333,
      "grad_norm": 8.050721168518066,
      "learning_rate": 4.8487266235337895e-06,
      "loss": 3.2107,
      "step": 11030
    },
    {
      "epoch": 0.11225382486979167,
      "grad_norm": 10.355036735534668,
      "learning_rate": 4.848589604455856e-06,
      "loss": 3.304,
      "step": 11035
    },
    {
      "epoch": 0.1123046875,
      "grad_norm": 19.039920806884766,
      "learning_rate": 4.848452525289814e-06,
      "loss": 3.7786,
      "step": 11040
    },
    {
      "epoch": 0.11235555013020833,
      "grad_norm": 12.501426696777344,
      "learning_rate": 4.8483153860391705e-06,
      "loss": 3.5708,
      "step": 11045
    },
    {
      "epoch": 0.11240641276041667,
      "grad_norm": 15.971506118774414,
      "learning_rate": 4.848178186707435e-06,
      "loss": 3.209,
      "step": 11050
    },
    {
      "epoch": 0.112457275390625,
      "grad_norm": 7.047082424163818,
      "learning_rate": 4.8480409272981165e-06,
      "loss": 3.6421,
      "step": 11055
    },
    {
      "epoch": 0.11250813802083333,
      "grad_norm": 14.781586647033691,
      "learning_rate": 4.847903607814728e-06,
      "loss": 3.4269,
      "step": 11060
    },
    {
      "epoch": 0.11255900065104167,
      "grad_norm": 10.393120765686035,
      "learning_rate": 4.847766228260781e-06,
      "loss": 3.5995,
      "step": 11065
    },
    {
      "epoch": 0.11260986328125,
      "grad_norm": 10.039695739746094,
      "learning_rate": 4.847628788639793e-06,
      "loss": 3.3204,
      "step": 11070
    },
    {
      "epoch": 0.11266072591145833,
      "grad_norm": 13.824012756347656,
      "learning_rate": 4.847491288955279e-06,
      "loss": 3.6698,
      "step": 11075
    },
    {
      "epoch": 0.11271158854166667,
      "grad_norm": 11.334912300109863,
      "learning_rate": 4.847353729210756e-06,
      "loss": 3.5687,
      "step": 11080
    },
    {
      "epoch": 0.112762451171875,
      "grad_norm": 15.515557289123535,
      "learning_rate": 4.847216109409744e-06,
      "loss": 3.761,
      "step": 11085
    },
    {
      "epoch": 0.11281331380208333,
      "grad_norm": 7.701551914215088,
      "learning_rate": 4.847078429555765e-06,
      "loss": 3.6052,
      "step": 11090
    },
    {
      "epoch": 0.11286417643229167,
      "grad_norm": 13.438464164733887,
      "learning_rate": 4.8469406896523405e-06,
      "loss": 3.2666,
      "step": 11095
    },
    {
      "epoch": 0.1129150390625,
      "grad_norm": 11.519120216369629,
      "learning_rate": 4.846802889702994e-06,
      "loss": 3.1568,
      "step": 11100
    },
    {
      "epoch": 0.11296590169270833,
      "grad_norm": 10.409358978271484,
      "learning_rate": 4.8466650297112525e-06,
      "loss": 3.5128,
      "step": 11105
    },
    {
      "epoch": 0.11301676432291667,
      "grad_norm": 10.928455352783203,
      "learning_rate": 4.846527109680642e-06,
      "loss": 3.7998,
      "step": 11110
    },
    {
      "epoch": 0.113067626953125,
      "grad_norm": 13.899455070495605,
      "learning_rate": 4.846389129614692e-06,
      "loss": 3.1142,
      "step": 11115
    },
    {
      "epoch": 0.11311848958333333,
      "grad_norm": 10.756808280944824,
      "learning_rate": 4.846251089516932e-06,
      "loss": 3.3781,
      "step": 11120
    },
    {
      "epoch": 0.11316935221354167,
      "grad_norm": 10.080056190490723,
      "learning_rate": 4.846112989390894e-06,
      "loss": 3.1493,
      "step": 11125
    },
    {
      "epoch": 0.11322021484375,
      "grad_norm": 16.393461227416992,
      "learning_rate": 4.845974829240112e-06,
      "loss": 3.6096,
      "step": 11130
    },
    {
      "epoch": 0.11327107747395833,
      "grad_norm": 13.437868118286133,
      "learning_rate": 4.845836609068119e-06,
      "loss": 3.3507,
      "step": 11135
    },
    {
      "epoch": 0.11332194010416667,
      "grad_norm": 12.06811237335205,
      "learning_rate": 4.8456983288784535e-06,
      "loss": 3.197,
      "step": 11140
    },
    {
      "epoch": 0.113372802734375,
      "grad_norm": 11.709611892700195,
      "learning_rate": 4.845559988674651e-06,
      "loss": 3.5839,
      "step": 11145
    },
    {
      "epoch": 0.11342366536458333,
      "grad_norm": 17.550220489501953,
      "learning_rate": 4.8454215884602525e-06,
      "loss": 3.6834,
      "step": 11150
    },
    {
      "epoch": 0.11347452799479167,
      "grad_norm": 10.27842903137207,
      "learning_rate": 4.845283128238799e-06,
      "loss": 3.5057,
      "step": 11155
    },
    {
      "epoch": 0.113525390625,
      "grad_norm": 16.689462661743164,
      "learning_rate": 4.845144608013832e-06,
      "loss": 3.3252,
      "step": 11160
    },
    {
      "epoch": 0.11357625325520833,
      "grad_norm": 10.314776420593262,
      "learning_rate": 4.845006027788897e-06,
      "loss": 4.4668,
      "step": 11165
    },
    {
      "epoch": 0.11362711588541667,
      "grad_norm": 12.564552307128906,
      "learning_rate": 4.844867387567538e-06,
      "loss": 3.3531,
      "step": 11170
    },
    {
      "epoch": 0.113677978515625,
      "grad_norm": 16.5703182220459,
      "learning_rate": 4.8447286873533025e-06,
      "loss": 3.3511,
      "step": 11175
    },
    {
      "epoch": 0.11372884114583333,
      "grad_norm": 10.776202201843262,
      "learning_rate": 4.84458992714974e-06,
      "loss": 3.5057,
      "step": 11180
    },
    {
      "epoch": 0.11377970377604167,
      "grad_norm": 9.926742553710938,
      "learning_rate": 4.844451106960399e-06,
      "loss": 3.5368,
      "step": 11185
    },
    {
      "epoch": 0.11383056640625,
      "grad_norm": 13.466375350952148,
      "learning_rate": 4.844312226788833e-06,
      "loss": 3.4744,
      "step": 11190
    },
    {
      "epoch": 0.11388142903645833,
      "grad_norm": 13.84046745300293,
      "learning_rate": 4.844173286638593e-06,
      "loss": 3.4421,
      "step": 11195
    },
    {
      "epoch": 0.11393229166666667,
      "grad_norm": 16.326255798339844,
      "learning_rate": 4.8440342865132365e-06,
      "loss": 3.5388,
      "step": 11200
    },
    {
      "epoch": 0.113983154296875,
      "grad_norm": 11.938819885253906,
      "learning_rate": 4.843895226416317e-06,
      "loss": 3.171,
      "step": 11205
    },
    {
      "epoch": 0.11403401692708333,
      "grad_norm": 8.013636589050293,
      "learning_rate": 4.843756106351396e-06,
      "loss": 4.0519,
      "step": 11210
    },
    {
      "epoch": 0.11408487955729167,
      "grad_norm": 12.114385604858398,
      "learning_rate": 4.843616926322029e-06,
      "loss": 3.5346,
      "step": 11215
    },
    {
      "epoch": 0.1141357421875,
      "grad_norm": 14.332465171813965,
      "learning_rate": 4.843477686331778e-06,
      "loss": 3.3585,
      "step": 11220
    },
    {
      "epoch": 0.11418660481770833,
      "grad_norm": 13.207276344299316,
      "learning_rate": 4.8433383863842065e-06,
      "loss": 3.2574,
      "step": 11225
    },
    {
      "epoch": 0.11423746744791667,
      "grad_norm": 13.37678050994873,
      "learning_rate": 4.8431990264828775e-06,
      "loss": 3.0765,
      "step": 11230
    },
    {
      "epoch": 0.114288330078125,
      "grad_norm": 10.288887977600098,
      "learning_rate": 4.843059606631358e-06,
      "loss": 3.4786,
      "step": 11235
    },
    {
      "epoch": 0.11433919270833333,
      "grad_norm": 9.490957260131836,
      "learning_rate": 4.842920126833212e-06,
      "loss": 3.0409,
      "step": 11240
    },
    {
      "epoch": 0.11439005533854167,
      "grad_norm": 12.170294761657715,
      "learning_rate": 4.842780587092011e-06,
      "loss": 3.4692,
      "step": 11245
    },
    {
      "epoch": 0.11444091796875,
      "grad_norm": 8.890890121459961,
      "learning_rate": 4.842640987411323e-06,
      "loss": 3.1144,
      "step": 11250
    },
    {
      "epoch": 0.11449178059895833,
      "grad_norm": 10.649124145507812,
      "learning_rate": 4.842501327794722e-06,
      "loss": 3.3058,
      "step": 11255
    },
    {
      "epoch": 0.11454264322916667,
      "grad_norm": 15.376280784606934,
      "learning_rate": 4.842361608245779e-06,
      "loss": 3.2482,
      "step": 11260
    },
    {
      "epoch": 0.114593505859375,
      "grad_norm": 11.391185760498047,
      "learning_rate": 4.84222182876807e-06,
      "loss": 3.2936,
      "step": 11265
    },
    {
      "epoch": 0.11464436848958333,
      "grad_norm": 8.210448265075684,
      "learning_rate": 4.84208198936517e-06,
      "loss": 3.5134,
      "step": 11270
    },
    {
      "epoch": 0.11469523111979167,
      "grad_norm": 15.108521461486816,
      "learning_rate": 4.841942090040658e-06,
      "loss": 3.4495,
      "step": 11275
    },
    {
      "epoch": 0.11474609375,
      "grad_norm": 10.561015129089355,
      "learning_rate": 4.841802130798112e-06,
      "loss": 3.3369,
      "step": 11280
    },
    {
      "epoch": 0.11479695638020833,
      "grad_norm": 13.979567527770996,
      "learning_rate": 4.841662111641114e-06,
      "loss": 3.5918,
      "step": 11285
    },
    {
      "epoch": 0.11484781901041667,
      "grad_norm": 12.601395606994629,
      "learning_rate": 4.841522032573246e-06,
      "loss": 3.3726,
      "step": 11290
    },
    {
      "epoch": 0.114898681640625,
      "grad_norm": 14.680171966552734,
      "learning_rate": 4.841381893598092e-06,
      "loss": 3.3161,
      "step": 11295
    },
    {
      "epoch": 0.11494954427083333,
      "grad_norm": 11.404961585998535,
      "learning_rate": 4.841241694719236e-06,
      "loss": 3.9943,
      "step": 11300
    },
    {
      "epoch": 0.11500040690104167,
      "grad_norm": 7.569732666015625,
      "learning_rate": 4.841101435940268e-06,
      "loss": 3.2494,
      "step": 11305
    },
    {
      "epoch": 0.11505126953125,
      "grad_norm": 13.507386207580566,
      "learning_rate": 4.840961117264773e-06,
      "loss": 3.2946,
      "step": 11310
    },
    {
      "epoch": 0.11510213216145833,
      "grad_norm": 8.6294584274292,
      "learning_rate": 4.840820738696343e-06,
      "loss": 3.6376,
      "step": 11315
    },
    {
      "epoch": 0.11515299479166667,
      "grad_norm": 12.276251792907715,
      "learning_rate": 4.8406803002385696e-06,
      "loss": 3.5475,
      "step": 11320
    },
    {
      "epoch": 0.115203857421875,
      "grad_norm": 13.95095157623291,
      "learning_rate": 4.8405398018950465e-06,
      "loss": 3.4318,
      "step": 11325
    },
    {
      "epoch": 0.11525472005208333,
      "grad_norm": 8.981563568115234,
      "learning_rate": 4.840399243669366e-06,
      "loss": 3.3984,
      "step": 11330
    },
    {
      "epoch": 0.11530558268229167,
      "grad_norm": 17.332687377929688,
      "learning_rate": 4.840258625565126e-06,
      "loss": 3.4236,
      "step": 11335
    },
    {
      "epoch": 0.1153564453125,
      "grad_norm": 13.317326545715332,
      "learning_rate": 4.840117947585924e-06,
      "loss": 3.7795,
      "step": 11340
    },
    {
      "epoch": 0.11540730794270833,
      "grad_norm": 7.8791656494140625,
      "learning_rate": 4.8399772097353585e-06,
      "loss": 3.316,
      "step": 11345
    },
    {
      "epoch": 0.11545817057291667,
      "grad_norm": 11.29205322265625,
      "learning_rate": 4.839836412017031e-06,
      "loss": 3.5903,
      "step": 11350
    },
    {
      "epoch": 0.115509033203125,
      "grad_norm": 9.628116607666016,
      "learning_rate": 4.839695554434543e-06,
      "loss": 3.9415,
      "step": 11355
    },
    {
      "epoch": 0.11555989583333333,
      "grad_norm": 13.431588172912598,
      "learning_rate": 4.839554636991499e-06,
      "loss": 3.3724,
      "step": 11360
    },
    {
      "epoch": 0.11561075846354167,
      "grad_norm": 16.05211639404297,
      "learning_rate": 4.8394136596915044e-06,
      "loss": 3.6998,
      "step": 11365
    },
    {
      "epoch": 0.11566162109375,
      "grad_norm": 15.686810493469238,
      "learning_rate": 4.839272622538166e-06,
      "loss": 3.3292,
      "step": 11370
    },
    {
      "epoch": 0.11571248372395833,
      "grad_norm": 10.349496841430664,
      "learning_rate": 4.839131525535093e-06,
      "loss": 3.7189,
      "step": 11375
    },
    {
      "epoch": 0.11576334635416667,
      "grad_norm": 12.135592460632324,
      "learning_rate": 4.838990368685892e-06,
      "loss": 4.0048,
      "step": 11380
    },
    {
      "epoch": 0.115814208984375,
      "grad_norm": 16.356788635253906,
      "learning_rate": 4.838849151994178e-06,
      "loss": 3.6877,
      "step": 11385
    },
    {
      "epoch": 0.11586507161458333,
      "grad_norm": 16.019092559814453,
      "learning_rate": 4.838707875463563e-06,
      "loss": 3.4608,
      "step": 11390
    },
    {
      "epoch": 0.11591593424479167,
      "grad_norm": 9.560792922973633,
      "learning_rate": 4.838566539097661e-06,
      "loss": 3.3188,
      "step": 11395
    },
    {
      "epoch": 0.115966796875,
      "grad_norm": 14.858901023864746,
      "learning_rate": 4.838425142900089e-06,
      "loss": 3.589,
      "step": 11400
    },
    {
      "epoch": 0.11601765950520833,
      "grad_norm": 8.294513702392578,
      "learning_rate": 4.8382836868744635e-06,
      "loss": 3.3469,
      "step": 11405
    },
    {
      "epoch": 0.11606852213541667,
      "grad_norm": 10.487101554870605,
      "learning_rate": 4.838142171024404e-06,
      "loss": 3.6045,
      "step": 11410
    },
    {
      "epoch": 0.116119384765625,
      "grad_norm": 12.35394287109375,
      "learning_rate": 4.838000595353531e-06,
      "loss": 3.4698,
      "step": 11415
    },
    {
      "epoch": 0.11617024739583333,
      "grad_norm": 16.841279983520508,
      "learning_rate": 4.8378589598654675e-06,
      "loss": 3.3743,
      "step": 11420
    },
    {
      "epoch": 0.11622111002604167,
      "grad_norm": 8.052149772644043,
      "learning_rate": 4.837717264563837e-06,
      "loss": 3.7649,
      "step": 11425
    },
    {
      "epoch": 0.11627197265625,
      "grad_norm": 8.459037780761719,
      "learning_rate": 4.837575509452264e-06,
      "loss": 3.1424,
      "step": 11430
    },
    {
      "epoch": 0.11632283528645833,
      "grad_norm": 13.29317855834961,
      "learning_rate": 4.837433694534376e-06,
      "loss": 3.2511,
      "step": 11435
    },
    {
      "epoch": 0.11637369791666667,
      "grad_norm": 10.47177791595459,
      "learning_rate": 4.8372918198138e-06,
      "loss": 3.1985,
      "step": 11440
    },
    {
      "epoch": 0.116424560546875,
      "grad_norm": 12.912489891052246,
      "learning_rate": 4.837149885294167e-06,
      "loss": 3.414,
      "step": 11445
    },
    {
      "epoch": 0.11647542317708333,
      "grad_norm": 11.507333755493164,
      "learning_rate": 4.837007890979108e-06,
      "loss": 3.362,
      "step": 11450
    },
    {
      "epoch": 0.11652628580729167,
      "grad_norm": 11.616119384765625,
      "learning_rate": 4.836865836872257e-06,
      "loss": 3.4036,
      "step": 11455
    },
    {
      "epoch": 0.1165771484375,
      "grad_norm": 16.553955078125,
      "learning_rate": 4.8367237229772466e-06,
      "loss": 3.5742,
      "step": 11460
    },
    {
      "epoch": 0.11662801106770833,
      "grad_norm": 10.673224449157715,
      "learning_rate": 4.836581549297715e-06,
      "loss": 3.6775,
      "step": 11465
    },
    {
      "epoch": 0.11667887369791667,
      "grad_norm": 12.565034866333008,
      "learning_rate": 4.836439315837297e-06,
      "loss": 3.8003,
      "step": 11470
    },
    {
      "epoch": 0.116729736328125,
      "grad_norm": 8.820014953613281,
      "learning_rate": 4.8362970225996334e-06,
      "loss": 3.1671,
      "step": 11475
    },
    {
      "epoch": 0.11678059895833333,
      "grad_norm": 13.71123218536377,
      "learning_rate": 4.836154669588363e-06,
      "loss": 3.2272,
      "step": 11480
    },
    {
      "epoch": 0.11683146158854167,
      "grad_norm": 12.12161636352539,
      "learning_rate": 4.8360122568071304e-06,
      "loss": 3.6246,
      "step": 11485
    },
    {
      "epoch": 0.11688232421875,
      "grad_norm": 9.981605529785156,
      "learning_rate": 4.835869784259578e-06,
      "loss": 3.2711,
      "step": 11490
    },
    {
      "epoch": 0.11693318684895833,
      "grad_norm": 8.769081115722656,
      "learning_rate": 4.83572725194935e-06,
      "loss": 3.4075,
      "step": 11495
    },
    {
      "epoch": 0.11698404947916667,
      "grad_norm": 9.689691543579102,
      "learning_rate": 4.835584659880095e-06,
      "loss": 3.2837,
      "step": 11500
    },
    {
      "epoch": 0.117034912109375,
      "grad_norm": 10.834155082702637,
      "learning_rate": 4.835442008055459e-06,
      "loss": 3.1802,
      "step": 11505
    },
    {
      "epoch": 0.11708577473958333,
      "grad_norm": 13.873346328735352,
      "learning_rate": 4.835299296479093e-06,
      "loss": 3.8511,
      "step": 11510
    },
    {
      "epoch": 0.11713663736979167,
      "grad_norm": 10.343167304992676,
      "learning_rate": 4.835156525154648e-06,
      "loss": 3.5516,
      "step": 11515
    },
    {
      "epoch": 0.1171875,
      "grad_norm": 13.721023559570312,
      "learning_rate": 4.8350136940857775e-06,
      "loss": 3.4011,
      "step": 11520
    },
    {
      "epoch": 0.11723836263020833,
      "grad_norm": 15.717689514160156,
      "learning_rate": 4.834870803276134e-06,
      "loss": 3.5618,
      "step": 11525
    },
    {
      "epoch": 0.11728922526041667,
      "grad_norm": 16.386823654174805,
      "learning_rate": 4.834727852729375e-06,
      "loss": 3.3112,
      "step": 11530
    },
    {
      "epoch": 0.117340087890625,
      "grad_norm": 15.418145179748535,
      "learning_rate": 4.834584842449158e-06,
      "loss": 4.0253,
      "step": 11535
    },
    {
      "epoch": 0.11739095052083333,
      "grad_norm": 8.149100303649902,
      "learning_rate": 4.83444177243914e-06,
      "loss": 3.5978,
      "step": 11540
    },
    {
      "epoch": 0.11744181315104167,
      "grad_norm": 8.802492141723633,
      "learning_rate": 4.834298642702983e-06,
      "loss": 3.931,
      "step": 11545
    },
    {
      "epoch": 0.11749267578125,
      "grad_norm": 10.84029483795166,
      "learning_rate": 4.834155453244348e-06,
      "loss": 3.4902,
      "step": 11550
    },
    {
      "epoch": 0.11754353841145833,
      "grad_norm": 13.952017784118652,
      "learning_rate": 4.8340122040669e-06,
      "loss": 3.3633,
      "step": 11555
    },
    {
      "epoch": 0.11759440104166667,
      "grad_norm": 9.742794036865234,
      "learning_rate": 4.833868895174303e-06,
      "loss": 3.46,
      "step": 11560
    },
    {
      "epoch": 0.117645263671875,
      "grad_norm": 9.079483032226562,
      "learning_rate": 4.833725526570223e-06,
      "loss": 3.2567,
      "step": 11565
    },
    {
      "epoch": 0.11769612630208333,
      "grad_norm": 11.4492826461792,
      "learning_rate": 4.833582098258328e-06,
      "loss": 3.332,
      "step": 11570
    },
    {
      "epoch": 0.11774698893229167,
      "grad_norm": 13.898246765136719,
      "learning_rate": 4.833438610242289e-06,
      "loss": 3.6246,
      "step": 11575
    },
    {
      "epoch": 0.1177978515625,
      "grad_norm": 18.6462459564209,
      "learning_rate": 4.833295062525775e-06,
      "loss": 4.2665,
      "step": 11580
    },
    {
      "epoch": 0.11784871419270833,
      "grad_norm": 15.118053436279297,
      "learning_rate": 4.833151455112462e-06,
      "loss": 3.4488,
      "step": 11585
    },
    {
      "epoch": 0.11789957682291667,
      "grad_norm": 12.25283432006836,
      "learning_rate": 4.833007788006021e-06,
      "loss": 3.3655,
      "step": 11590
    },
    {
      "epoch": 0.117950439453125,
      "grad_norm": 12.915300369262695,
      "learning_rate": 4.832864061210128e-06,
      "loss": 3.2749,
      "step": 11595
    },
    {
      "epoch": 0.11800130208333333,
      "grad_norm": 13.077988624572754,
      "learning_rate": 4.832720274728462e-06,
      "loss": 3.7166,
      "step": 11600
    },
    {
      "epoch": 0.11805216471354167,
      "grad_norm": 12.226242065429688,
      "learning_rate": 4.8325764285647e-06,
      "loss": 3.4749,
      "step": 11605
    },
    {
      "epoch": 0.11810302734375,
      "grad_norm": 12.33906078338623,
      "learning_rate": 4.832432522722523e-06,
      "loss": 3.8167,
      "step": 11610
    },
    {
      "epoch": 0.11815388997395833,
      "grad_norm": 18.989513397216797,
      "learning_rate": 4.832288557205612e-06,
      "loss": 3.2801,
      "step": 11615
    },
    {
      "epoch": 0.11820475260416667,
      "grad_norm": 13.292867660522461,
      "learning_rate": 4.832144532017653e-06,
      "loss": 3.4304,
      "step": 11620
    },
    {
      "epoch": 0.118255615234375,
      "grad_norm": 13.21804141998291,
      "learning_rate": 4.832000447162328e-06,
      "loss": 3.1607,
      "step": 11625
    },
    {
      "epoch": 0.11830647786458333,
      "grad_norm": 9.76976203918457,
      "learning_rate": 4.8318563026433244e-06,
      "loss": 3.6529,
      "step": 11630
    },
    {
      "epoch": 0.11835734049479167,
      "grad_norm": 14.536739349365234,
      "learning_rate": 4.831712098464329e-06,
      "loss": 3.216,
      "step": 11635
    },
    {
      "epoch": 0.118408203125,
      "grad_norm": 7.9985880851745605,
      "learning_rate": 4.831567834629033e-06,
      "loss": 3.0674,
      "step": 11640
    },
    {
      "epoch": 0.11845906575520833,
      "grad_norm": 11.221656799316406,
      "learning_rate": 4.831423511141127e-06,
      "loss": 3.3163,
      "step": 11645
    },
    {
      "epoch": 0.11850992838541667,
      "grad_norm": 16.915536880493164,
      "learning_rate": 4.831279128004303e-06,
      "loss": 3.6208,
      "step": 11650
    },
    {
      "epoch": 0.118560791015625,
      "grad_norm": 11.9306058883667,
      "learning_rate": 4.831134685222255e-06,
      "loss": 3.6851,
      "step": 11655
    },
    {
      "epoch": 0.11861165364583333,
      "grad_norm": 10.024066925048828,
      "learning_rate": 4.8309901827986785e-06,
      "loss": 3.4386,
      "step": 11660
    },
    {
      "epoch": 0.11866251627604167,
      "grad_norm": 14.097890853881836,
      "learning_rate": 4.83084562073727e-06,
      "loss": 3.3924,
      "step": 11665
    },
    {
      "epoch": 0.11871337890625,
      "grad_norm": 17.218984603881836,
      "learning_rate": 4.83070099904173e-06,
      "loss": 3.3288,
      "step": 11670
    },
    {
      "epoch": 0.11876424153645833,
      "grad_norm": 9.430810928344727,
      "learning_rate": 4.830556317715757e-06,
      "loss": 3.1723,
      "step": 11675
    },
    {
      "epoch": 0.11881510416666667,
      "grad_norm": 16.22243309020996,
      "learning_rate": 4.830411576763052e-06,
      "loss": 3.2928,
      "step": 11680
    },
    {
      "epoch": 0.118865966796875,
      "grad_norm": 14.999611854553223,
      "learning_rate": 4.83026677618732e-06,
      "loss": 3.5795,
      "step": 11685
    },
    {
      "epoch": 0.11891682942708333,
      "grad_norm": 13.795806884765625,
      "learning_rate": 4.830121915992265e-06,
      "loss": 3.6035,
      "step": 11690
    },
    {
      "epoch": 0.11896769205729167,
      "grad_norm": 11.390329360961914,
      "learning_rate": 4.829976996181593e-06,
      "loss": 3.6227,
      "step": 11695
    },
    {
      "epoch": 0.1190185546875,
      "grad_norm": 13.322566986083984,
      "learning_rate": 4.829832016759012e-06,
      "loss": 3.2774,
      "step": 11700
    },
    {
      "epoch": 0.11906941731770833,
      "grad_norm": 13.909505844116211,
      "learning_rate": 4.829686977728231e-06,
      "loss": 3.3357,
      "step": 11705
    },
    {
      "epoch": 0.11912027994791667,
      "grad_norm": 12.06095027923584,
      "learning_rate": 4.82954187909296e-06,
      "loss": 3.9097,
      "step": 11710
    },
    {
      "epoch": 0.119171142578125,
      "grad_norm": 11.268613815307617,
      "learning_rate": 4.829396720856913e-06,
      "loss": 3.2764,
      "step": 11715
    },
    {
      "epoch": 0.11922200520833333,
      "grad_norm": 8.79672622680664,
      "learning_rate": 4.829251503023803e-06,
      "loss": 3.587,
      "step": 11720
    },
    {
      "epoch": 0.11927286783854167,
      "grad_norm": 15.61196231842041,
      "learning_rate": 4.8291062255973455e-06,
      "loss": 3.2073,
      "step": 11725
    },
    {
      "epoch": 0.11932373046875,
      "grad_norm": 8.96154499053955,
      "learning_rate": 4.828960888581256e-06,
      "loss": 3.4395,
      "step": 11730
    },
    {
      "epoch": 0.11937459309895833,
      "grad_norm": 13.609565734863281,
      "learning_rate": 4.828815491979256e-06,
      "loss": 3.3523,
      "step": 11735
    },
    {
      "epoch": 0.11942545572916667,
      "grad_norm": 8.735452651977539,
      "learning_rate": 4.828670035795063e-06,
      "loss": 3.3582,
      "step": 11740
    },
    {
      "epoch": 0.119476318359375,
      "grad_norm": 13.924631118774414,
      "learning_rate": 4.828524520032399e-06,
      "loss": 3.3311,
      "step": 11745
    },
    {
      "epoch": 0.11952718098958333,
      "grad_norm": 13.989679336547852,
      "learning_rate": 4.828378944694987e-06,
      "loss": 3.1395,
      "step": 11750
    },
    {
      "epoch": 0.11957804361979167,
      "grad_norm": 17.6063289642334,
      "learning_rate": 4.828233309786552e-06,
      "loss": 3.2459,
      "step": 11755
    },
    {
      "epoch": 0.11962890625,
      "grad_norm": 14.077485084533691,
      "learning_rate": 4.828087615310819e-06,
      "loss": 3.5539,
      "step": 11760
    },
    {
      "epoch": 0.11967976888020833,
      "grad_norm": 11.900781631469727,
      "learning_rate": 4.8279418612715165e-06,
      "loss": 3.3846,
      "step": 11765
    },
    {
      "epoch": 0.11973063151041667,
      "grad_norm": 9.877041816711426,
      "learning_rate": 4.8277960476723726e-06,
      "loss": 3.7396,
      "step": 11770
    },
    {
      "epoch": 0.119781494140625,
      "grad_norm": 16.47829818725586,
      "learning_rate": 4.82765017451712e-06,
      "loss": 3.1649,
      "step": 11775
    },
    {
      "epoch": 0.11983235677083333,
      "grad_norm": 12.586395263671875,
      "learning_rate": 4.827504241809488e-06,
      "loss": 3.513,
      "step": 11780
    },
    {
      "epoch": 0.11988321940104167,
      "grad_norm": 12.90040111541748,
      "learning_rate": 4.827358249553213e-06,
      "loss": 3.5154,
      "step": 11785
    },
    {
      "epoch": 0.11993408203125,
      "grad_norm": 7.8959503173828125,
      "learning_rate": 4.8272121977520266e-06,
      "loss": 3.5876,
      "step": 11790
    },
    {
      "epoch": 0.11998494466145833,
      "grad_norm": 14.199788093566895,
      "learning_rate": 4.82706608640967e-06,
      "loss": 3.3843,
      "step": 11795
    },
    {
      "epoch": 0.12003580729166667,
      "grad_norm": 14.042448043823242,
      "learning_rate": 4.826919915529878e-06,
      "loss": 2.9448,
      "step": 11800
    },
    {
      "epoch": 0.120086669921875,
      "grad_norm": 14.99729061126709,
      "learning_rate": 4.826773685116392e-06,
      "loss": 3.3641,
      "step": 11805
    },
    {
      "epoch": 0.12013753255208333,
      "grad_norm": 8.840888023376465,
      "learning_rate": 4.826627395172952e-06,
      "loss": 3.1467,
      "step": 11810
    },
    {
      "epoch": 0.12018839518229167,
      "grad_norm": 8.198177337646484,
      "learning_rate": 4.8264810457033025e-06,
      "loss": 3.5442,
      "step": 11815
    },
    {
      "epoch": 0.1202392578125,
      "grad_norm": 10.527442932128906,
      "learning_rate": 4.826334636711186e-06,
      "loss": 3.431,
      "step": 11820
    },
    {
      "epoch": 0.12029012044270833,
      "grad_norm": 11.18076229095459,
      "learning_rate": 4.82618816820035e-06,
      "loss": 3.2887,
      "step": 11825
    },
    {
      "epoch": 0.12034098307291667,
      "grad_norm": 13.812418937683105,
      "learning_rate": 4.826041640174542e-06,
      "loss": 3.5518,
      "step": 11830
    },
    {
      "epoch": 0.120391845703125,
      "grad_norm": 17.44132423400879,
      "learning_rate": 4.825895052637508e-06,
      "loss": 3.629,
      "step": 11835
    },
    {
      "epoch": 0.12044270833333333,
      "grad_norm": 13.55534553527832,
      "learning_rate": 4.825748405593001e-06,
      "loss": 3.4921,
      "step": 11840
    },
    {
      "epoch": 0.12049357096354167,
      "grad_norm": 10.288969993591309,
      "learning_rate": 4.825601699044773e-06,
      "loss": 3.876,
      "step": 11845
    },
    {
      "epoch": 0.12054443359375,
      "grad_norm": 15.818678855895996,
      "learning_rate": 4.825454932996576e-06,
      "loss": 3.2189,
      "step": 11850
    },
    {
      "epoch": 0.12059529622395833,
      "grad_norm": 8.30019474029541,
      "learning_rate": 4.825308107452166e-06,
      "loss": 3.3629,
      "step": 11855
    },
    {
      "epoch": 0.12064615885416667,
      "grad_norm": 8.714160919189453,
      "learning_rate": 4.825161222415299e-06,
      "loss": 3.2909,
      "step": 11860
    },
    {
      "epoch": 0.120697021484375,
      "grad_norm": 8.198768615722656,
      "learning_rate": 4.825014277889733e-06,
      "loss": 3.5741,
      "step": 11865
    },
    {
      "epoch": 0.12074788411458333,
      "grad_norm": 14.308550834655762,
      "learning_rate": 4.824867273879229e-06,
      "loss": 3.5978,
      "step": 11870
    },
    {
      "epoch": 0.12079874674479167,
      "grad_norm": 11.020129203796387,
      "learning_rate": 4.8247202103875455e-06,
      "loss": 3.3011,
      "step": 11875
    },
    {
      "epoch": 0.120849609375,
      "grad_norm": 12.168553352355957,
      "learning_rate": 4.824573087418447e-06,
      "loss": 3.4922,
      "step": 11880
    },
    {
      "epoch": 0.12090047200520833,
      "grad_norm": 8.048748016357422,
      "learning_rate": 4.824425904975697e-06,
      "loss": 3.7659,
      "step": 11885
    },
    {
      "epoch": 0.12095133463541667,
      "grad_norm": 13.554608345031738,
      "learning_rate": 4.8242786630630615e-06,
      "loss": 3.2647,
      "step": 11890
    },
    {
      "epoch": 0.121002197265625,
      "grad_norm": 14.179122924804688,
      "learning_rate": 4.824131361684308e-06,
      "loss": 3.3172,
      "step": 11895
    },
    {
      "epoch": 0.12105305989583333,
      "grad_norm": 17.352275848388672,
      "learning_rate": 4.823984000843203e-06,
      "loss": 3.9405,
      "step": 11900
    },
    {
      "epoch": 0.12110392252604167,
      "grad_norm": 7.687533378601074,
      "learning_rate": 4.823836580543519e-06,
      "loss": 3.6776,
      "step": 11905
    },
    {
      "epoch": 0.12115478515625,
      "grad_norm": 18.918060302734375,
      "learning_rate": 4.823689100789026e-06,
      "loss": 3.6594,
      "step": 11910
    },
    {
      "epoch": 0.12120564778645833,
      "grad_norm": 16.174633026123047,
      "learning_rate": 4.823541561583499e-06,
      "loss": 3.4054,
      "step": 11915
    },
    {
      "epoch": 0.12125651041666667,
      "grad_norm": 12.678868293762207,
      "learning_rate": 4.8233939629307115e-06,
      "loss": 3.4099,
      "step": 11920
    },
    {
      "epoch": 0.121307373046875,
      "grad_norm": 12.389989852905273,
      "learning_rate": 4.82324630483444e-06,
      "loss": 3.6306,
      "step": 11925
    },
    {
      "epoch": 0.12135823567708333,
      "grad_norm": 8.188562393188477,
      "learning_rate": 4.823098587298463e-06,
      "loss": 3.2383,
      "step": 11930
    },
    {
      "epoch": 0.12140909830729167,
      "grad_norm": 8.033101081848145,
      "learning_rate": 4.8229508103265595e-06,
      "loss": 3.322,
      "step": 11935
    },
    {
      "epoch": 0.1214599609375,
      "grad_norm": 11.021990776062012,
      "learning_rate": 4.822802973922509e-06,
      "loss": 3.704,
      "step": 11940
    },
    {
      "epoch": 0.12151082356770833,
      "grad_norm": 6.139153003692627,
      "learning_rate": 4.822655078090096e-06,
      "loss": 3.2763,
      "step": 11945
    },
    {
      "epoch": 0.12156168619791667,
      "grad_norm": 13.162734985351562,
      "learning_rate": 4.822507122833104e-06,
      "loss": 3.8643,
      "step": 11950
    },
    {
      "epoch": 0.121612548828125,
      "grad_norm": 11.088370323181152,
      "learning_rate": 4.8223591081553154e-06,
      "loss": 3.5046,
      "step": 11955
    },
    {
      "epoch": 0.12166341145833333,
      "grad_norm": 13.00243091583252,
      "learning_rate": 4.822211034060521e-06,
      "loss": 4.2186,
      "step": 11960
    },
    {
      "epoch": 0.12171427408854167,
      "grad_norm": 11.786962509155273,
      "learning_rate": 4.822062900552507e-06,
      "loss": 3.4922,
      "step": 11965
    },
    {
      "epoch": 0.12176513671875,
      "grad_norm": 8.783388137817383,
      "learning_rate": 4.821914707635065e-06,
      "loss": 3.3294,
      "step": 11970
    },
    {
      "epoch": 0.12181599934895833,
      "grad_norm": 9.021117210388184,
      "learning_rate": 4.821766455311986e-06,
      "loss": 3.4608,
      "step": 11975
    },
    {
      "epoch": 0.12186686197916667,
      "grad_norm": 11.563819885253906,
      "learning_rate": 4.821618143587062e-06,
      "loss": 3.4183,
      "step": 11980
    },
    {
      "epoch": 0.121917724609375,
      "grad_norm": 7.746058464050293,
      "learning_rate": 4.821469772464087e-06,
      "loss": 3.0433,
      "step": 11985
    },
    {
      "epoch": 0.12196858723958333,
      "grad_norm": 16.20210838317871,
      "learning_rate": 4.821321341946859e-06,
      "loss": 3.6232,
      "step": 11990
    },
    {
      "epoch": 0.12201944986979167,
      "grad_norm": 14.226766586303711,
      "learning_rate": 4.821172852039175e-06,
      "loss": 3.5745,
      "step": 11995
    },
    {
      "epoch": 0.1220703125,
      "grad_norm": 13.494339942932129,
      "learning_rate": 4.821024302744834e-06,
      "loss": 3.5401,
      "step": 12000
    },
    {
      "epoch": 0.12212117513020833,
      "grad_norm": 9.108319282531738,
      "learning_rate": 4.820875694067635e-06,
      "loss": 3.2929,
      "step": 12005
    },
    {
      "epoch": 0.12217203776041667,
      "grad_norm": 9.119816780090332,
      "learning_rate": 4.820727026011382e-06,
      "loss": 2.9584,
      "step": 12010
    },
    {
      "epoch": 0.122222900390625,
      "grad_norm": 14.945995330810547,
      "learning_rate": 4.820578298579879e-06,
      "loss": 3.4583,
      "step": 12015
    },
    {
      "epoch": 0.12227376302083333,
      "grad_norm": 12.933895111083984,
      "learning_rate": 4.820429511776929e-06,
      "loss": 3.5372,
      "step": 12020
    },
    {
      "epoch": 0.12232462565104167,
      "grad_norm": 14.16519546508789,
      "learning_rate": 4.820280665606341e-06,
      "loss": 3.3085,
      "step": 12025
    },
    {
      "epoch": 0.12237548828125,
      "grad_norm": 14.623350143432617,
      "learning_rate": 4.820131760071921e-06,
      "loss": 2.9811,
      "step": 12030
    },
    {
      "epoch": 0.12242635091145833,
      "grad_norm": 13.574602127075195,
      "learning_rate": 4.8199827951774805e-06,
      "loss": 3.5069,
      "step": 12035
    },
    {
      "epoch": 0.12247721354166667,
      "grad_norm": 15.912371635437012,
      "learning_rate": 4.8198337709268305e-06,
      "loss": 3.6601,
      "step": 12040
    },
    {
      "epoch": 0.122528076171875,
      "grad_norm": 15.12542724609375,
      "learning_rate": 4.819684687323783e-06,
      "loss": 3.3761,
      "step": 12045
    },
    {
      "epoch": 0.12257893880208333,
      "grad_norm": 8.16356372833252,
      "learning_rate": 4.819535544372153e-06,
      "loss": 3.474,
      "step": 12050
    },
    {
      "epoch": 0.12262980143229167,
      "grad_norm": 9.369928359985352,
      "learning_rate": 4.819386342075755e-06,
      "loss": 3.548,
      "step": 12055
    },
    {
      "epoch": 0.1226806640625,
      "grad_norm": 16.48990249633789,
      "learning_rate": 4.8192370804384075e-06,
      "loss": 3.5765,
      "step": 12060
    },
    {
      "epoch": 0.12273152669270833,
      "grad_norm": 14.442497253417969,
      "learning_rate": 4.819087759463929e-06,
      "loss": 3.3908,
      "step": 12065
    },
    {
      "epoch": 0.12278238932291667,
      "grad_norm": 15.705107688903809,
      "learning_rate": 4.81893837915614e-06,
      "loss": 3.6492,
      "step": 12070
    },
    {
      "epoch": 0.122833251953125,
      "grad_norm": 15.82107162475586,
      "learning_rate": 4.818788939518863e-06,
      "loss": 3.2617,
      "step": 12075
    },
    {
      "epoch": 0.12288411458333333,
      "grad_norm": 16.034608840942383,
      "learning_rate": 4.8186394405559186e-06,
      "loss": 3.2683,
      "step": 12080
    },
    {
      "epoch": 0.12293497721354167,
      "grad_norm": 10.579744338989258,
      "learning_rate": 4.818489882271135e-06,
      "loss": 3.2758,
      "step": 12085
    },
    {
      "epoch": 0.12298583984375,
      "grad_norm": 12.461812019348145,
      "learning_rate": 4.818340264668337e-06,
      "loss": 3.2529,
      "step": 12090
    },
    {
      "epoch": 0.12303670247395833,
      "grad_norm": 17.07436752319336,
      "learning_rate": 4.8181905877513535e-06,
      "loss": 3.7906,
      "step": 12095
    },
    {
      "epoch": 0.12308756510416667,
      "grad_norm": 11.529945373535156,
      "learning_rate": 4.818040851524013e-06,
      "loss": 3.5339,
      "step": 12100
    },
    {
      "epoch": 0.123138427734375,
      "grad_norm": 15.79837417602539,
      "learning_rate": 4.817891055990146e-06,
      "loss": 3.4221,
      "step": 12105
    },
    {
      "epoch": 0.12318929036458333,
      "grad_norm": 15.319779396057129,
      "learning_rate": 4.817741201153587e-06,
      "loss": 3.4168,
      "step": 12110
    },
    {
      "epoch": 0.12324015299479167,
      "grad_norm": 13.032570838928223,
      "learning_rate": 4.817591287018168e-06,
      "loss": 3.5855,
      "step": 12115
    },
    {
      "epoch": 0.123291015625,
      "grad_norm": 10.453932762145996,
      "learning_rate": 4.817441313587725e-06,
      "loss": 3.3555,
      "step": 12120
    },
    {
      "epoch": 0.12334187825520833,
      "grad_norm": 15.642955780029297,
      "learning_rate": 4.817291280866096e-06,
      "loss": 3.5448,
      "step": 12125
    },
    {
      "epoch": 0.12339274088541667,
      "grad_norm": 10.044322967529297,
      "learning_rate": 4.8171411888571185e-06,
      "loss": 3.3746,
      "step": 12130
    },
    {
      "epoch": 0.123443603515625,
      "grad_norm": 11.619243621826172,
      "learning_rate": 4.816991037564632e-06,
      "loss": 3.6049,
      "step": 12135
    },
    {
      "epoch": 0.12349446614583333,
      "grad_norm": 9.354294776916504,
      "learning_rate": 4.81684082699248e-06,
      "loss": 3.5606,
      "step": 12140
    },
    {
      "epoch": 0.12354532877604167,
      "grad_norm": 15.51186752319336,
      "learning_rate": 4.816690557144505e-06,
      "loss": 3.6708,
      "step": 12145
    },
    {
      "epoch": 0.12359619140625,
      "grad_norm": 13.874622344970703,
      "learning_rate": 4.816540228024551e-06,
      "loss": 3.5003,
      "step": 12150
    },
    {
      "epoch": 0.12364705403645833,
      "grad_norm": 11.789558410644531,
      "learning_rate": 4.816389839636463e-06,
      "loss": 3.2361,
      "step": 12155
    },
    {
      "epoch": 0.12369791666666667,
      "grad_norm": 14.9029541015625,
      "learning_rate": 4.816239391984091e-06,
      "loss": 3.8518,
      "step": 12160
    },
    {
      "epoch": 0.123748779296875,
      "grad_norm": 16.188743591308594,
      "learning_rate": 4.8160888850712835e-06,
      "loss": 3.4669,
      "step": 12165
    },
    {
      "epoch": 0.12379964192708333,
      "grad_norm": 16.114046096801758,
      "learning_rate": 4.81593831890189e-06,
      "loss": 3.329,
      "step": 12170
    },
    {
      "epoch": 0.12385050455729167,
      "grad_norm": 9.05451774597168,
      "learning_rate": 4.815787693479764e-06,
      "loss": 3.1477,
      "step": 12175
    },
    {
      "epoch": 0.1239013671875,
      "grad_norm": 13.087759971618652,
      "learning_rate": 4.815637008808759e-06,
      "loss": 3.4289,
      "step": 12180
    },
    {
      "epoch": 0.12395222981770833,
      "grad_norm": 11.878509521484375,
      "learning_rate": 4.81548626489273e-06,
      "loss": 3.3253,
      "step": 12185
    },
    {
      "epoch": 0.12400309244791667,
      "grad_norm": 23.86294937133789,
      "learning_rate": 4.815335461735534e-06,
      "loss": 3.7135,
      "step": 12190
    },
    {
      "epoch": 0.124053955078125,
      "grad_norm": 12.333020210266113,
      "learning_rate": 4.815184599341029e-06,
      "loss": 3.5888,
      "step": 12195
    },
    {
      "epoch": 0.12410481770833333,
      "grad_norm": 11.767260551452637,
      "learning_rate": 4.8150336777130736e-06,
      "loss": 3.2373,
      "step": 12200
    },
    {
      "epoch": 0.12415568033854167,
      "grad_norm": 11.746440887451172,
      "learning_rate": 4.8148826968555306e-06,
      "loss": 3.3153,
      "step": 12205
    },
    {
      "epoch": 0.12420654296875,
      "grad_norm": 12.361739158630371,
      "learning_rate": 4.814731656772263e-06,
      "loss": 3.4058,
      "step": 12210
    },
    {
      "epoch": 0.12425740559895833,
      "grad_norm": 10.730027198791504,
      "learning_rate": 4.8145805574671346e-06,
      "loss": 3.4855,
      "step": 12215
    },
    {
      "epoch": 0.12430826822916667,
      "grad_norm": 16.28338623046875,
      "learning_rate": 4.814429398944011e-06,
      "loss": 3.5275,
      "step": 12220
    },
    {
      "epoch": 0.124359130859375,
      "grad_norm": 16.588144302368164,
      "learning_rate": 4.81427818120676e-06,
      "loss": 3.6741,
      "step": 12225
    },
    {
      "epoch": 0.12440999348958333,
      "grad_norm": 15.339471817016602,
      "learning_rate": 4.81412690425925e-06,
      "loss": 3.9718,
      "step": 12230
    },
    {
      "epoch": 0.12446085611979167,
      "grad_norm": 8.622764587402344,
      "learning_rate": 4.8139755681053526e-06,
      "loss": 3.5536,
      "step": 12235
    },
    {
      "epoch": 0.12451171875,
      "grad_norm": 13.108068466186523,
      "learning_rate": 4.813824172748938e-06,
      "loss": 3.283,
      "step": 12240
    },
    {
      "epoch": 0.12456258138020833,
      "grad_norm": 12.999624252319336,
      "learning_rate": 4.8136727181938804e-06,
      "loss": 3.7082,
      "step": 12245
    },
    {
      "epoch": 0.12461344401041667,
      "grad_norm": 9.05246639251709,
      "learning_rate": 4.813521204444055e-06,
      "loss": 3.655,
      "step": 12250
    },
    {
      "epoch": 0.124664306640625,
      "grad_norm": 11.840641021728516,
      "learning_rate": 4.8133696315033375e-06,
      "loss": 3.4432,
      "step": 12255
    },
    {
      "epoch": 0.12471516927083333,
      "grad_norm": 12.203230857849121,
      "learning_rate": 4.813217999375606e-06,
      "loss": 3.1917,
      "step": 12260
    },
    {
      "epoch": 0.12476603190104167,
      "grad_norm": 10.946849822998047,
      "learning_rate": 4.813066308064741e-06,
      "loss": 3.2396,
      "step": 12265
    },
    {
      "epoch": 0.12481689453125,
      "grad_norm": 9.767146110534668,
      "learning_rate": 4.812914557574622e-06,
      "loss": 3.337,
      "step": 12270
    },
    {
      "epoch": 0.12486775716145833,
      "grad_norm": 10.88211727142334,
      "learning_rate": 4.8127627479091336e-06,
      "loss": 3.2582,
      "step": 12275
    },
    {
      "epoch": 0.12491861979166667,
      "grad_norm": 9.982248306274414,
      "learning_rate": 4.812610879072157e-06,
      "loss": 3.2663,
      "step": 12280
    },
    {
      "epoch": 0.124969482421875,
      "grad_norm": 8.785655975341797,
      "learning_rate": 4.8124589510675805e-06,
      "loss": 3.1978,
      "step": 12285
    },
    {
      "epoch": 0.12502034505208334,
      "grad_norm": 13.370884895324707,
      "learning_rate": 4.812306963899289e-06,
      "loss": 3.5219,
      "step": 12290
    },
    {
      "epoch": 0.12507120768229166,
      "grad_norm": 14.19118595123291,
      "learning_rate": 4.812154917571172e-06,
      "loss": 3.1948,
      "step": 12295
    },
    {
      "epoch": 0.1251220703125,
      "grad_norm": 15.058856010437012,
      "learning_rate": 4.81200281208712e-06,
      "loss": 3.4093,
      "step": 12300
    },
    {
      "epoch": 0.12517293294270834,
      "grad_norm": 17.146102905273438,
      "learning_rate": 4.811850647451024e-06,
      "loss": 3.3073,
      "step": 12305
    },
    {
      "epoch": 0.12522379557291666,
      "grad_norm": 10.474042892456055,
      "learning_rate": 4.811698423666777e-06,
      "loss": 3.3589,
      "step": 12310
    },
    {
      "epoch": 0.125274658203125,
      "grad_norm": 12.146330833435059,
      "learning_rate": 4.811546140738273e-06,
      "loss": 3.4111,
      "step": 12315
    },
    {
      "epoch": 0.12532552083333334,
      "grad_norm": 8.893861770629883,
      "learning_rate": 4.811393798669409e-06,
      "loss": 3.4912,
      "step": 12320
    },
    {
      "epoch": 0.12537638346354166,
      "grad_norm": 14.769695281982422,
      "learning_rate": 4.811241397464083e-06,
      "loss": 3.459,
      "step": 12325
    },
    {
      "epoch": 0.12542724609375,
      "grad_norm": 16.10647964477539,
      "learning_rate": 4.811088937126194e-06,
      "loss": 3.3882,
      "step": 12330
    },
    {
      "epoch": 0.12547810872395834,
      "grad_norm": 14.999397277832031,
      "learning_rate": 4.8109364176596416e-06,
      "loss": 3.3708,
      "step": 12335
    },
    {
      "epoch": 0.12552897135416666,
      "grad_norm": 11.39329719543457,
      "learning_rate": 4.810783839068329e-06,
      "loss": 3.5732,
      "step": 12340
    },
    {
      "epoch": 0.125579833984375,
      "grad_norm": 10.63199234008789,
      "learning_rate": 4.81063120135616e-06,
      "loss": 3.5412,
      "step": 12345
    },
    {
      "epoch": 0.12563069661458334,
      "grad_norm": 9.905284881591797,
      "learning_rate": 4.81047850452704e-06,
      "loss": 3.1996,
      "step": 12350
    },
    {
      "epoch": 0.12568155924479166,
      "grad_norm": 15.250255584716797,
      "learning_rate": 4.810325748584873e-06,
      "loss": 3.3639,
      "step": 12355
    },
    {
      "epoch": 0.125732421875,
      "grad_norm": 12.014054298400879,
      "learning_rate": 4.8101729335335716e-06,
      "loss": 3.2804,
      "step": 12360
    },
    {
      "epoch": 0.12578328450520834,
      "grad_norm": 15.268871307373047,
      "learning_rate": 4.810020059377042e-06,
      "loss": 3.4151,
      "step": 12365
    },
    {
      "epoch": 0.12583414713541666,
      "grad_norm": 15.318325996398926,
      "learning_rate": 4.809867126119197e-06,
      "loss": 3.6018,
      "step": 12370
    },
    {
      "epoch": 0.125885009765625,
      "grad_norm": 13.372442245483398,
      "learning_rate": 4.8097141337639485e-06,
      "loss": 3.8433,
      "step": 12375
    },
    {
      "epoch": 0.12593587239583334,
      "grad_norm": 15.433815956115723,
      "learning_rate": 4.809561082315212e-06,
      "loss": 3.2409,
      "step": 12380
    },
    {
      "epoch": 0.12598673502604166,
      "grad_norm": 9.874568939208984,
      "learning_rate": 4.809407971776902e-06,
      "loss": 3.5737,
      "step": 12385
    },
    {
      "epoch": 0.12603759765625,
      "grad_norm": 13.617438316345215,
      "learning_rate": 4.809254802152937e-06,
      "loss": 3.598,
      "step": 12390
    },
    {
      "epoch": 0.12608846028645834,
      "grad_norm": 17.306371688842773,
      "learning_rate": 4.809101573447236e-06,
      "loss": 3.9022,
      "step": 12395
    },
    {
      "epoch": 0.12613932291666666,
      "grad_norm": 12.497845649719238,
      "learning_rate": 4.808948285663717e-06,
      "loss": 3.292,
      "step": 12400
    },
    {
      "epoch": 0.126190185546875,
      "grad_norm": 11.501578330993652,
      "learning_rate": 4.808794938806305e-06,
      "loss": 3.4564,
      "step": 12405
    },
    {
      "epoch": 0.12624104817708334,
      "grad_norm": 15.333441734313965,
      "learning_rate": 4.808641532878921e-06,
      "loss": 3.257,
      "step": 12410
    },
    {
      "epoch": 0.12629191080729166,
      "grad_norm": 9.874608993530273,
      "learning_rate": 4.80848806788549e-06,
      "loss": 3.2448,
      "step": 12415
    },
    {
      "epoch": 0.1263427734375,
      "grad_norm": 9.775317192077637,
      "learning_rate": 4.808334543829939e-06,
      "loss": 3.4979,
      "step": 12420
    },
    {
      "epoch": 0.12639363606770834,
      "grad_norm": 9.998175621032715,
      "learning_rate": 4.808180960716196e-06,
      "loss": 3.1983,
      "step": 12425
    },
    {
      "epoch": 0.12644449869791666,
      "grad_norm": 10.001486778259277,
      "learning_rate": 4.808027318548191e-06,
      "loss": 3.2788,
      "step": 12430
    },
    {
      "epoch": 0.126495361328125,
      "grad_norm": 15.605623245239258,
      "learning_rate": 4.807873617329854e-06,
      "loss": 3.7982,
      "step": 12435
    },
    {
      "epoch": 0.12654622395833334,
      "grad_norm": 16.90433692932129,
      "learning_rate": 4.807719857065117e-06,
      "loss": 3.7252,
      "step": 12440
    },
    {
      "epoch": 0.12659708658854166,
      "grad_norm": 9.86048412322998,
      "learning_rate": 4.807566037757914e-06,
      "loss": 3.3411,
      "step": 12445
    },
    {
      "epoch": 0.12664794921875,
      "grad_norm": 9.026899337768555,
      "learning_rate": 4.807412159412181e-06,
      "loss": 3.1362,
      "step": 12450
    },
    {
      "epoch": 0.12669881184895834,
      "grad_norm": 11.268509864807129,
      "learning_rate": 4.807258222031855e-06,
      "loss": 3.2253,
      "step": 12455
    },
    {
      "epoch": 0.12674967447916666,
      "grad_norm": 14.036666870117188,
      "learning_rate": 4.807104225620875e-06,
      "loss": 3.1731,
      "step": 12460
    },
    {
      "epoch": 0.126800537109375,
      "grad_norm": 9.418083190917969,
      "learning_rate": 4.8069501701831795e-06,
      "loss": 3.1714,
      "step": 12465
    },
    {
      "epoch": 0.12685139973958334,
      "grad_norm": 12.404669761657715,
      "learning_rate": 4.8067960557227114e-06,
      "loss": 3.6047,
      "step": 12470
    },
    {
      "epoch": 0.12690226236979166,
      "grad_norm": 12.391560554504395,
      "learning_rate": 4.806641882243412e-06,
      "loss": 3.4368,
      "step": 12475
    },
    {
      "epoch": 0.126953125,
      "grad_norm": 13.154583930969238,
      "learning_rate": 4.806487649749228e-06,
      "loss": 4.0829,
      "step": 12480
    },
    {
      "epoch": 0.12700398763020834,
      "grad_norm": 13.884321212768555,
      "learning_rate": 4.806333358244103e-06,
      "loss": 3.1453,
      "step": 12485
    },
    {
      "epoch": 0.12705485026041666,
      "grad_norm": 11.85291862487793,
      "learning_rate": 4.806179007731986e-06,
      "loss": 3.4844,
      "step": 12490
    },
    {
      "epoch": 0.127105712890625,
      "grad_norm": 12.22746753692627,
      "learning_rate": 4.806024598216826e-06,
      "loss": 3.7238,
      "step": 12495
    },
    {
      "epoch": 0.12715657552083334,
      "grad_norm": 14.834999084472656,
      "learning_rate": 4.805870129702573e-06,
      "loss": 3.7607,
      "step": 12500
    },
    {
      "epoch": 0.12720743815104166,
      "grad_norm": 18.713239669799805,
      "learning_rate": 4.8057156021931795e-06,
      "loss": 3.5314,
      "step": 12505
    },
    {
      "epoch": 0.12725830078125,
      "grad_norm": 15.553114891052246,
      "learning_rate": 4.8055610156925984e-06,
      "loss": 3.9413,
      "step": 12510
    },
    {
      "epoch": 0.12730916341145834,
      "grad_norm": 12.916542053222656,
      "learning_rate": 4.805406370204785e-06,
      "loss": 3.3368,
      "step": 12515
    },
    {
      "epoch": 0.12736002604166666,
      "grad_norm": 13.694195747375488,
      "learning_rate": 4.805251665733696e-06,
      "loss": 3.2709,
      "step": 12520
    },
    {
      "epoch": 0.127410888671875,
      "grad_norm": 10.602334976196289,
      "learning_rate": 4.805096902283291e-06,
      "loss": 3.3952,
      "step": 12525
    },
    {
      "epoch": 0.12746175130208334,
      "grad_norm": 14.200652122497559,
      "learning_rate": 4.804942079857527e-06,
      "loss": 3.5215,
      "step": 12530
    },
    {
      "epoch": 0.12751261393229166,
      "grad_norm": 7.785111427307129,
      "learning_rate": 4.804787198460366e-06,
      "loss": 3.3147,
      "step": 12535
    },
    {
      "epoch": 0.1275634765625,
      "grad_norm": 12.794201850891113,
      "learning_rate": 4.804632258095772e-06,
      "loss": 3.4677,
      "step": 12540
    },
    {
      "epoch": 0.12761433919270834,
      "grad_norm": 12.663290977478027,
      "learning_rate": 4.804477258767707e-06,
      "loss": 3.6271,
      "step": 12545
    },
    {
      "epoch": 0.12766520182291666,
      "grad_norm": 10.109106063842773,
      "learning_rate": 4.804322200480138e-06,
      "loss": 3.5064,
      "step": 12550
    },
    {
      "epoch": 0.127716064453125,
      "grad_norm": 15.36133861541748,
      "learning_rate": 4.804167083237031e-06,
      "loss": 3.4629,
      "step": 12555
    },
    {
      "epoch": 0.12776692708333334,
      "grad_norm": 12.881776809692383,
      "learning_rate": 4.804011907042356e-06,
      "loss": 3.4292,
      "step": 12560
    },
    {
      "epoch": 0.12781778971354166,
      "grad_norm": 8.816761016845703,
      "learning_rate": 4.8038566719000825e-06,
      "loss": 3.0355,
      "step": 12565
    },
    {
      "epoch": 0.12786865234375,
      "grad_norm": 15.458996772766113,
      "learning_rate": 4.803701377814181e-06,
      "loss": 4.0244,
      "step": 12570
    },
    {
      "epoch": 0.12791951497395834,
      "grad_norm": 9.55286979675293,
      "learning_rate": 4.803546024788628e-06,
      "loss": 3.4125,
      "step": 12575
    },
    {
      "epoch": 0.12797037760416666,
      "grad_norm": 15.313224792480469,
      "learning_rate": 4.803390612827394e-06,
      "loss": 3.4685,
      "step": 12580
    },
    {
      "epoch": 0.128021240234375,
      "grad_norm": 13.189021110534668,
      "learning_rate": 4.803235141934458e-06,
      "loss": 3.3217,
      "step": 12585
    },
    {
      "epoch": 0.12807210286458334,
      "grad_norm": 16.103500366210938,
      "learning_rate": 4.803079612113796e-06,
      "loss": 3.1009,
      "step": 12590
    },
    {
      "epoch": 0.12812296549479166,
      "grad_norm": 16.239477157592773,
      "learning_rate": 4.802924023369388e-06,
      "loss": 3.4193,
      "step": 12595
    },
    {
      "epoch": 0.128173828125,
      "grad_norm": 11.9988374710083,
      "learning_rate": 4.802768375705216e-06,
      "loss": 3.6531,
      "step": 12600
    },
    {
      "epoch": 0.12822469075520834,
      "grad_norm": 11.093536376953125,
      "learning_rate": 4.802612669125261e-06,
      "loss": 3.2401,
      "step": 12605
    },
    {
      "epoch": 0.12827555338541666,
      "grad_norm": 10.944393157958984,
      "learning_rate": 4.8024569036335055e-06,
      "loss": 3.2432,
      "step": 12610
    },
    {
      "epoch": 0.128326416015625,
      "grad_norm": 14.31828498840332,
      "learning_rate": 4.802301079233936e-06,
      "loss": 3.6411,
      "step": 12615
    },
    {
      "epoch": 0.12837727864583334,
      "grad_norm": 9.36100959777832,
      "learning_rate": 4.802145195930539e-06,
      "loss": 3.4192,
      "step": 12620
    },
    {
      "epoch": 0.12842814127604166,
      "grad_norm": 13.963410377502441,
      "learning_rate": 4.801989253727303e-06,
      "loss": 3.1725,
      "step": 12625
    },
    {
      "epoch": 0.12847900390625,
      "grad_norm": 16.763933181762695,
      "learning_rate": 4.801833252628218e-06,
      "loss": 3.3481,
      "step": 12630
    },
    {
      "epoch": 0.12852986653645834,
      "grad_norm": 12.801989555358887,
      "learning_rate": 4.801677192637275e-06,
      "loss": 3.5976,
      "step": 12635
    },
    {
      "epoch": 0.12858072916666666,
      "grad_norm": 81.20991516113281,
      "learning_rate": 4.801521073758466e-06,
      "loss": 3.3127,
      "step": 12640
    },
    {
      "epoch": 0.128631591796875,
      "grad_norm": 13.878232955932617,
      "learning_rate": 4.801364895995786e-06,
      "loss": 3.313,
      "step": 12645
    },
    {
      "epoch": 0.12868245442708334,
      "grad_norm": 10.5735445022583,
      "learning_rate": 4.8012086593532306e-06,
      "loss": 3.298,
      "step": 12650
    },
    {
      "epoch": 0.12873331705729166,
      "grad_norm": 10.607198715209961,
      "learning_rate": 4.8010523638347965e-06,
      "loss": 3.4635,
      "step": 12655
    },
    {
      "epoch": 0.1287841796875,
      "grad_norm": 10.28955078125,
      "learning_rate": 4.800896009444484e-06,
      "loss": 3.5608,
      "step": 12660
    },
    {
      "epoch": 0.12883504231770834,
      "grad_norm": 12.767330169677734,
      "learning_rate": 4.800739596186293e-06,
      "loss": 3.222,
      "step": 12665
    },
    {
      "epoch": 0.12888590494791666,
      "grad_norm": 15.003509521484375,
      "learning_rate": 4.800583124064223e-06,
      "loss": 3.2579,
      "step": 12670
    },
    {
      "epoch": 0.128936767578125,
      "grad_norm": 14.754562377929688,
      "learning_rate": 4.80042659308228e-06,
      "loss": 3.2074,
      "step": 12675
    },
    {
      "epoch": 0.12898763020833334,
      "grad_norm": 15.33542251586914,
      "learning_rate": 4.800270003244467e-06,
      "loss": 3.2303,
      "step": 12680
    },
    {
      "epoch": 0.12903849283854166,
      "grad_norm": 14.25710391998291,
      "learning_rate": 4.800113354554793e-06,
      "loss": 3.8232,
      "step": 12685
    },
    {
      "epoch": 0.12908935546875,
      "grad_norm": 16.291183471679688,
      "learning_rate": 4.799956647017262e-06,
      "loss": 3.3102,
      "step": 12690
    },
    {
      "epoch": 0.12914021809895834,
      "grad_norm": 13.827568054199219,
      "learning_rate": 4.799799880635887e-06,
      "loss": 3.8675,
      "step": 12695
    },
    {
      "epoch": 0.12919108072916666,
      "grad_norm": 13.914209365844727,
      "learning_rate": 4.799643055414677e-06,
      "loss": 3.0558,
      "step": 12700
    },
    {
      "epoch": 0.129241943359375,
      "grad_norm": 10.144859313964844,
      "learning_rate": 4.799486171357644e-06,
      "loss": 3.4235,
      "step": 12705
    },
    {
      "epoch": 0.12929280598958334,
      "grad_norm": 10.29310417175293,
      "learning_rate": 4.799329228468802e-06,
      "loss": 3.4347,
      "step": 12710
    },
    {
      "epoch": 0.12934366861979166,
      "grad_norm": 13.560129165649414,
      "learning_rate": 4.7991722267521665e-06,
      "loss": 3.3288,
      "step": 12715
    },
    {
      "epoch": 0.12939453125,
      "grad_norm": 13.470560073852539,
      "learning_rate": 4.799015166211756e-06,
      "loss": 3.2676,
      "step": 12720
    },
    {
      "epoch": 0.12944539388020834,
      "grad_norm": 12.207547187805176,
      "learning_rate": 4.798858046851587e-06,
      "loss": 3.0864,
      "step": 12725
    },
    {
      "epoch": 0.12949625651041666,
      "grad_norm": 15.827969551086426,
      "learning_rate": 4.798700868675679e-06,
      "loss": 3.3807,
      "step": 12730
    },
    {
      "epoch": 0.129547119140625,
      "grad_norm": 14.415461540222168,
      "learning_rate": 4.798543631688054e-06,
      "loss": 3.0016,
      "step": 12735
    },
    {
      "epoch": 0.12959798177083334,
      "grad_norm": 7.998185157775879,
      "learning_rate": 4.798386335892735e-06,
      "loss": 3.2422,
      "step": 12740
    },
    {
      "epoch": 0.12964884440104166,
      "grad_norm": 7.472536087036133,
      "learning_rate": 4.798228981293747e-06,
      "loss": 3.4627,
      "step": 12745
    },
    {
      "epoch": 0.12969970703125,
      "grad_norm": 7.803865909576416,
      "learning_rate": 4.798071567895115e-06,
      "loss": 3.4333,
      "step": 12750
    },
    {
      "epoch": 0.12975056966145834,
      "grad_norm": 11.206137657165527,
      "learning_rate": 4.797914095700867e-06,
      "loss": 3.2908,
      "step": 12755
    },
    {
      "epoch": 0.12980143229166666,
      "grad_norm": 13.730634689331055,
      "learning_rate": 4.797756564715031e-06,
      "loss": 3.4254,
      "step": 12760
    },
    {
      "epoch": 0.129852294921875,
      "grad_norm": 13.240988731384277,
      "learning_rate": 4.797598974941638e-06,
      "loss": 3.8927,
      "step": 12765
    },
    {
      "epoch": 0.12990315755208334,
      "grad_norm": 14.986083030700684,
      "learning_rate": 4.79744132638472e-06,
      "loss": 2.9281,
      "step": 12770
    },
    {
      "epoch": 0.12995402018229166,
      "grad_norm": 12.508495330810547,
      "learning_rate": 4.79728361904831e-06,
      "loss": 3.2039,
      "step": 12775
    },
    {
      "epoch": 0.1300048828125,
      "grad_norm": 13.931479454040527,
      "learning_rate": 4.797125852936444e-06,
      "loss": 3.3995,
      "step": 12780
    },
    {
      "epoch": 0.13005574544270834,
      "grad_norm": 15.499288558959961,
      "learning_rate": 4.796968028053156e-06,
      "loss": 3.5259,
      "step": 12785
    },
    {
      "epoch": 0.13010660807291666,
      "grad_norm": 15.569916725158691,
      "learning_rate": 4.796810144402486e-06,
      "loss": 3.5487,
      "step": 12790
    },
    {
      "epoch": 0.130157470703125,
      "grad_norm": 11.951336860656738,
      "learning_rate": 4.796652201988474e-06,
      "loss": 3.1815,
      "step": 12795
    },
    {
      "epoch": 0.13020833333333334,
      "grad_norm": 10.30989933013916,
      "learning_rate": 4.796494200815158e-06,
      "loss": 3.4612,
      "step": 12800
    },
    {
      "epoch": 0.13025919596354166,
      "grad_norm": 10.12043571472168,
      "learning_rate": 4.796336140886584e-06,
      "loss": 3.9442,
      "step": 12805
    },
    {
      "epoch": 0.13031005859375,
      "grad_norm": 16.965572357177734,
      "learning_rate": 4.796178022206793e-06,
      "loss": 3.603,
      "step": 12810
    },
    {
      "epoch": 0.13036092122395834,
      "grad_norm": 12.111398696899414,
      "learning_rate": 4.796019844779831e-06,
      "loss": 3.6743,
      "step": 12815
    },
    {
      "epoch": 0.13041178385416666,
      "grad_norm": 14.302387237548828,
      "learning_rate": 4.795861608609747e-06,
      "loss": 3.2853,
      "step": 12820
    },
    {
      "epoch": 0.130462646484375,
      "grad_norm": 12.947281837463379,
      "learning_rate": 4.795703313700587e-06,
      "loss": 3.3062,
      "step": 12825
    },
    {
      "epoch": 0.13051350911458334,
      "grad_norm": 12.033870697021484,
      "learning_rate": 4.795544960056402e-06,
      "loss": 3.1838,
      "step": 12830
    },
    {
      "epoch": 0.13056437174479166,
      "grad_norm": 9.891314506530762,
      "learning_rate": 4.7953865476812435e-06,
      "loss": 3.4533,
      "step": 12835
    },
    {
      "epoch": 0.130615234375,
      "grad_norm": 11.173048973083496,
      "learning_rate": 4.795228076579164e-06,
      "loss": 3.3019,
      "step": 12840
    },
    {
      "epoch": 0.13066609700520834,
      "grad_norm": 58.573089599609375,
      "learning_rate": 4.795069546754219e-06,
      "loss": 3.2771,
      "step": 12845
    },
    {
      "epoch": 0.13071695963541666,
      "grad_norm": 10.534260749816895,
      "learning_rate": 4.794910958210463e-06,
      "loss": 3.5155,
      "step": 12850
    },
    {
      "epoch": 0.130767822265625,
      "grad_norm": 10.294408798217773,
      "learning_rate": 4.7947523109519535e-06,
      "loss": 3.1176,
      "step": 12855
    },
    {
      "epoch": 0.13081868489583334,
      "grad_norm": 8.581644058227539,
      "learning_rate": 4.79459360498275e-06,
      "loss": 3.6988,
      "step": 12860
    },
    {
      "epoch": 0.13086954752604166,
      "grad_norm": 7.670634746551514,
      "learning_rate": 4.794434840306914e-06,
      "loss": 3.4395,
      "step": 12865
    },
    {
      "epoch": 0.13092041015625,
      "grad_norm": 8.824413299560547,
      "learning_rate": 4.794276016928506e-06,
      "loss": 3.7669,
      "step": 12870
    },
    {
      "epoch": 0.13097127278645834,
      "grad_norm": 16.448566436767578,
      "learning_rate": 4.794117134851589e-06,
      "loss": 3.6758,
      "step": 12875
    },
    {
      "epoch": 0.13102213541666666,
      "grad_norm": 17.62189483642578,
      "learning_rate": 4.79395819408023e-06,
      "loss": 3.4432,
      "step": 12880
    },
    {
      "epoch": 0.131072998046875,
      "grad_norm": 13.63729190826416,
      "learning_rate": 4.793799194618495e-06,
      "loss": 3.1668,
      "step": 12885
    },
    {
      "epoch": 0.13112386067708334,
      "grad_norm": 9.173359870910645,
      "learning_rate": 4.79364013647045e-06,
      "loss": 3.28,
      "step": 12890
    },
    {
      "epoch": 0.13117472330729166,
      "grad_norm": 10.53978157043457,
      "learning_rate": 4.793481019640166e-06,
      "loss": 3.2492,
      "step": 12895
    },
    {
      "epoch": 0.1312255859375,
      "grad_norm": 11.081829071044922,
      "learning_rate": 4.793321844131714e-06,
      "loss": 3.3987,
      "step": 12900
    },
    {
      "epoch": 0.13127644856770834,
      "grad_norm": 8.339278221130371,
      "learning_rate": 4.793162609949166e-06,
      "loss": 3.1725,
      "step": 12905
    },
    {
      "epoch": 0.13132731119791666,
      "grad_norm": 9.031716346740723,
      "learning_rate": 4.793003317096596e-06,
      "loss": 3.2547,
      "step": 12910
    },
    {
      "epoch": 0.131378173828125,
      "grad_norm": 12.638387680053711,
      "learning_rate": 4.79284396557808e-06,
      "loss": 3.367,
      "step": 12915
    },
    {
      "epoch": 0.13142903645833334,
      "grad_norm": 12.951153755187988,
      "learning_rate": 4.7926845553976945e-06,
      "loss": 3.2392,
      "step": 12920
    },
    {
      "epoch": 0.13147989908854166,
      "grad_norm": 15.467037200927734,
      "learning_rate": 4.792525086559518e-06,
      "loss": 3.5068,
      "step": 12925
    },
    {
      "epoch": 0.13153076171875,
      "grad_norm": 8.698431015014648,
      "learning_rate": 4.792365559067631e-06,
      "loss": 3.3362,
      "step": 12930
    },
    {
      "epoch": 0.13158162434895834,
      "grad_norm": 14.218523979187012,
      "learning_rate": 4.792205972926114e-06,
      "loss": 3.0097,
      "step": 12935
    },
    {
      "epoch": 0.13163248697916666,
      "grad_norm": 7.461666584014893,
      "learning_rate": 4.792046328139051e-06,
      "loss": 3.3224,
      "step": 12940
    },
    {
      "epoch": 0.131683349609375,
      "grad_norm": 15.739773750305176,
      "learning_rate": 4.791886624710525e-06,
      "loss": 3.8441,
      "step": 12945
    },
    {
      "epoch": 0.13173421223958334,
      "grad_norm": 10.866748809814453,
      "learning_rate": 4.791726862644623e-06,
      "loss": 3.861,
      "step": 12950
    },
    {
      "epoch": 0.13178507486979166,
      "grad_norm": 9.433417320251465,
      "learning_rate": 4.791567041945433e-06,
      "loss": 3.0838,
      "step": 12955
    },
    {
      "epoch": 0.1318359375,
      "grad_norm": 16.90796661376953,
      "learning_rate": 4.791407162617043e-06,
      "loss": 3.7086,
      "step": 12960
    },
    {
      "epoch": 0.13188680013020834,
      "grad_norm": 14.038589477539062,
      "learning_rate": 4.791247224663545e-06,
      "loss": 3.4241,
      "step": 12965
    },
    {
      "epoch": 0.13193766276041666,
      "grad_norm": 8.530431747436523,
      "learning_rate": 4.791087228089029e-06,
      "loss": 3.4557,
      "step": 12970
    },
    {
      "epoch": 0.131988525390625,
      "grad_norm": 16.40488624572754,
      "learning_rate": 4.790927172897589e-06,
      "loss": 3.7428,
      "step": 12975
    },
    {
      "epoch": 0.13203938802083334,
      "grad_norm": 10.48965072631836,
      "learning_rate": 4.790767059093321e-06,
      "loss": 3.8246,
      "step": 12980
    },
    {
      "epoch": 0.13209025065104166,
      "grad_norm": 11.309000968933105,
      "learning_rate": 4.79060688668032e-06,
      "loss": 3.3004,
      "step": 12985
    },
    {
      "epoch": 0.13214111328125,
      "grad_norm": 13.044082641601562,
      "learning_rate": 4.790446655662686e-06,
      "loss": 3.4936,
      "step": 12990
    },
    {
      "epoch": 0.13219197591145834,
      "grad_norm": 14.110028266906738,
      "learning_rate": 4.790286366044516e-06,
      "loss": 2.9904,
      "step": 12995
    },
    {
      "epoch": 0.13224283854166666,
      "grad_norm": 16.38824462890625,
      "learning_rate": 4.790126017829913e-06,
      "loss": 3.1172,
      "step": 13000
    },
    {
      "epoch": 0.132293701171875,
      "grad_norm": 12.276939392089844,
      "learning_rate": 4.789965611022977e-06,
      "loss": 3.588,
      "step": 13005
    },
    {
      "epoch": 0.13234456380208334,
      "grad_norm": 12.605525016784668,
      "learning_rate": 4.7898051456278155e-06,
      "loss": 3.1167,
      "step": 13010
    },
    {
      "epoch": 0.13239542643229166,
      "grad_norm": 10.54214096069336,
      "learning_rate": 4.7896446216485314e-06,
      "loss": 3.496,
      "step": 13015
    },
    {
      "epoch": 0.1324462890625,
      "grad_norm": 11.415157318115234,
      "learning_rate": 4.789484039089232e-06,
      "loss": 3.5018,
      "step": 13020
    },
    {
      "epoch": 0.13249715169270834,
      "grad_norm": 10.655110359191895,
      "learning_rate": 4.789323397954027e-06,
      "loss": 3.5284,
      "step": 13025
    },
    {
      "epoch": 0.13254801432291666,
      "grad_norm": 12.17111873626709,
      "learning_rate": 4.789162698247024e-06,
      "loss": 3.0777,
      "step": 13030
    },
    {
      "epoch": 0.132598876953125,
      "grad_norm": 14.725519180297852,
      "learning_rate": 4.789001939972338e-06,
      "loss": 3.358,
      "step": 13035
    },
    {
      "epoch": 0.13264973958333334,
      "grad_norm": 15.81676959991455,
      "learning_rate": 4.7888411231340785e-06,
      "loss": 3.2394,
      "step": 13040
    },
    {
      "epoch": 0.13270060221354166,
      "grad_norm": 9.43869686126709,
      "learning_rate": 4.788680247736362e-06,
      "loss": 3.3947,
      "step": 13045
    },
    {
      "epoch": 0.13275146484375,
      "grad_norm": 14.853139877319336,
      "learning_rate": 4.788519313783303e-06,
      "loss": 3.0722,
      "step": 13050
    },
    {
      "epoch": 0.13280232747395834,
      "grad_norm": 12.241186141967773,
      "learning_rate": 4.788358321279021e-06,
      "loss": 3.8822,
      "step": 13055
    },
    {
      "epoch": 0.13285319010416666,
      "grad_norm": 8.0377779006958,
      "learning_rate": 4.788197270227633e-06,
      "loss": 3.4792,
      "step": 13060
    },
    {
      "epoch": 0.132904052734375,
      "grad_norm": 13.722543716430664,
      "learning_rate": 4.78803616063326e-06,
      "loss": 3.3762,
      "step": 13065
    },
    {
      "epoch": 0.13295491536458334,
      "grad_norm": 9.478768348693848,
      "learning_rate": 4.787874992500024e-06,
      "loss": 3.1809,
      "step": 13070
    },
    {
      "epoch": 0.13300577799479166,
      "grad_norm": 9.255359649658203,
      "learning_rate": 4.7877137658320496e-06,
      "loss": 3.4478,
      "step": 13075
    },
    {
      "epoch": 0.133056640625,
      "grad_norm": 14.590103149414062,
      "learning_rate": 4.7875524806334605e-06,
      "loss": 3.4376,
      "step": 13080
    },
    {
      "epoch": 0.13310750325520834,
      "grad_norm": 9.905108451843262,
      "learning_rate": 4.787391136908383e-06,
      "loss": 3.4683,
      "step": 13085
    },
    {
      "epoch": 0.13315836588541666,
      "grad_norm": 14.800880432128906,
      "learning_rate": 4.787229734660945e-06,
      "loss": 3.3281,
      "step": 13090
    },
    {
      "epoch": 0.133209228515625,
      "grad_norm": 15.413259506225586,
      "learning_rate": 4.787068273895278e-06,
      "loss": 3.222,
      "step": 13095
    },
    {
      "epoch": 0.13326009114583334,
      "grad_norm": 14.764119148254395,
      "learning_rate": 4.7869067546155105e-06,
      "loss": 3.3957,
      "step": 13100
    },
    {
      "epoch": 0.13331095377604166,
      "grad_norm": 10.476200103759766,
      "learning_rate": 4.786745176825775e-06,
      "loss": 3.2024,
      "step": 13105
    },
    {
      "epoch": 0.13336181640625,
      "grad_norm": 6.172470569610596,
      "learning_rate": 4.786583540530206e-06,
      "loss": 3.0385,
      "step": 13110
    },
    {
      "epoch": 0.13341267903645834,
      "grad_norm": 16.662702560424805,
      "learning_rate": 4.78642184573294e-06,
      "loss": 3.3144,
      "step": 13115
    },
    {
      "epoch": 0.13346354166666666,
      "grad_norm": 14.16555404663086,
      "learning_rate": 4.786260092438113e-06,
      "loss": 3.4183,
      "step": 13120
    },
    {
      "epoch": 0.133514404296875,
      "grad_norm": 13.170214653015137,
      "learning_rate": 4.7860982806498635e-06,
      "loss": 3.2948,
      "step": 13125
    },
    {
      "epoch": 0.13356526692708334,
      "grad_norm": 14.692107200622559,
      "learning_rate": 4.78593641037233e-06,
      "loss": 3.583,
      "step": 13130
    },
    {
      "epoch": 0.13361612955729166,
      "grad_norm": 14.330788612365723,
      "learning_rate": 4.785774481609657e-06,
      "loss": 3.5935,
      "step": 13135
    },
    {
      "epoch": 0.1336669921875,
      "grad_norm": 10.425037384033203,
      "learning_rate": 4.785612494365985e-06,
      "loss": 3.4952,
      "step": 13140
    },
    {
      "epoch": 0.13371785481770834,
      "grad_norm": 9.754368782043457,
      "learning_rate": 4.785450448645459e-06,
      "loss": 3.1403,
      "step": 13145
    },
    {
      "epoch": 0.13376871744791666,
      "grad_norm": 12.570056915283203,
      "learning_rate": 4.785288344452226e-06,
      "loss": 4.0699,
      "step": 13150
    },
    {
      "epoch": 0.133819580078125,
      "grad_norm": 9.117902755737305,
      "learning_rate": 4.78512618179043e-06,
      "loss": 3.8218,
      "step": 13155
    },
    {
      "epoch": 0.13387044270833334,
      "grad_norm": 13.497501373291016,
      "learning_rate": 4.784963960664224e-06,
      "loss": 3.5526,
      "step": 13160
    },
    {
      "epoch": 0.13392130533854166,
      "grad_norm": 12.687576293945312,
      "learning_rate": 4.784801681077757e-06,
      "loss": 3.5295,
      "step": 13165
    },
    {
      "epoch": 0.13397216796875,
      "grad_norm": 9.852676391601562,
      "learning_rate": 4.78463934303518e-06,
      "loss": 3.4353,
      "step": 13170
    },
    {
      "epoch": 0.13402303059895834,
      "grad_norm": 11.296305656433105,
      "learning_rate": 4.7844769465406464e-06,
      "loss": 3.3819,
      "step": 13175
    },
    {
      "epoch": 0.13407389322916666,
      "grad_norm": 9.368551254272461,
      "learning_rate": 4.784314491598312e-06,
      "loss": 3.1461,
      "step": 13180
    },
    {
      "epoch": 0.134124755859375,
      "grad_norm": 9.357016563415527,
      "learning_rate": 4.784151978212333e-06,
      "loss": 3.582,
      "step": 13185
    },
    {
      "epoch": 0.13417561848958334,
      "grad_norm": 11.293736457824707,
      "learning_rate": 4.783989406386867e-06,
      "loss": 3.5203,
      "step": 13190
    },
    {
      "epoch": 0.13422648111979166,
      "grad_norm": 13.790380477905273,
      "learning_rate": 4.783826776126073e-06,
      "loss": 3.3007,
      "step": 13195
    },
    {
      "epoch": 0.13427734375,
      "grad_norm": 14.679352760314941,
      "learning_rate": 4.783664087434112e-06,
      "loss": 3.8487,
      "step": 13200
    },
    {
      "epoch": 0.13432820638020834,
      "grad_norm": 15.090688705444336,
      "learning_rate": 4.783501340315147e-06,
      "loss": 2.9251,
      "step": 13205
    },
    {
      "epoch": 0.13437906901041666,
      "grad_norm": 15.12507152557373,
      "learning_rate": 4.783338534773343e-06,
      "loss": 3.2318,
      "step": 13210
    },
    {
      "epoch": 0.134429931640625,
      "grad_norm": 10.328474044799805,
      "learning_rate": 4.783175670812862e-06,
      "loss": 3.3133,
      "step": 13215
    },
    {
      "epoch": 0.13448079427083334,
      "grad_norm": 9.897954940795898,
      "learning_rate": 4.783012748437873e-06,
      "loss": 3.3901,
      "step": 13220
    },
    {
      "epoch": 0.13453165690104166,
      "grad_norm": 12.634456634521484,
      "learning_rate": 4.782849767652544e-06,
      "loss": 3.4324,
      "step": 13225
    },
    {
      "epoch": 0.13458251953125,
      "grad_norm": 13.496397972106934,
      "learning_rate": 4.782686728461044e-06,
      "loss": 3.4852,
      "step": 13230
    },
    {
      "epoch": 0.13463338216145834,
      "grad_norm": 9.415319442749023,
      "learning_rate": 4.782523630867546e-06,
      "loss": 3.8193,
      "step": 13235
    },
    {
      "epoch": 0.13468424479166666,
      "grad_norm": 14.183247566223145,
      "learning_rate": 4.782360474876222e-06,
      "loss": 3.7154,
      "step": 13240
    },
    {
      "epoch": 0.134735107421875,
      "grad_norm": 9.120186805725098,
      "learning_rate": 4.7821972604912464e-06,
      "loss": 3.0765,
      "step": 13245
    },
    {
      "epoch": 0.13478597005208334,
      "grad_norm": 15.911162376403809,
      "learning_rate": 4.782033987716794e-06,
      "loss": 3.2823,
      "step": 13250
    },
    {
      "epoch": 0.13483683268229166,
      "grad_norm": 12.479145050048828,
      "learning_rate": 4.781870656557044e-06,
      "loss": 3.3999,
      "step": 13255
    },
    {
      "epoch": 0.1348876953125,
      "grad_norm": 10.130742073059082,
      "learning_rate": 4.781707267016174e-06,
      "loss": 3.7453,
      "step": 13260
    },
    {
      "epoch": 0.13493855794270834,
      "grad_norm": 13.885787963867188,
      "learning_rate": 4.781543819098363e-06,
      "loss": 3.348,
      "step": 13265
    },
    {
      "epoch": 0.13498942057291666,
      "grad_norm": 9.897067070007324,
      "learning_rate": 4.781380312807795e-06,
      "loss": 3.6482,
      "step": 13270
    },
    {
      "epoch": 0.135040283203125,
      "grad_norm": 10.115368843078613,
      "learning_rate": 4.781216748148653e-06,
      "loss": 3.5157,
      "step": 13275
    },
    {
      "epoch": 0.13509114583333334,
      "grad_norm": 11.32646369934082,
      "learning_rate": 4.78105312512512e-06,
      "loss": 3.3044,
      "step": 13280
    },
    {
      "epoch": 0.13514200846354166,
      "grad_norm": 12.162534713745117,
      "learning_rate": 4.780889443741384e-06,
      "loss": 3.3909,
      "step": 13285
    },
    {
      "epoch": 0.13519287109375,
      "grad_norm": 14.136061668395996,
      "learning_rate": 4.780725704001633e-06,
      "loss": 3.5301,
      "step": 13290
    },
    {
      "epoch": 0.13524373372395834,
      "grad_norm": 12.348136901855469,
      "learning_rate": 4.780561905910055e-06,
      "loss": 3.979,
      "step": 13295
    },
    {
      "epoch": 0.13529459635416666,
      "grad_norm": 12.501129150390625,
      "learning_rate": 4.780398049470841e-06,
      "loss": 3.6263,
      "step": 13300
    },
    {
      "epoch": 0.135345458984375,
      "grad_norm": 16.70313835144043,
      "learning_rate": 4.780234134688184e-06,
      "loss": 3.584,
      "step": 13305
    },
    {
      "epoch": 0.13539632161458334,
      "grad_norm": 12.98574447631836,
      "learning_rate": 4.780070161566276e-06,
      "loss": 2.9994,
      "step": 13310
    },
    {
      "epoch": 0.13544718424479166,
      "grad_norm": 9.226192474365234,
      "learning_rate": 4.7799061301093144e-06,
      "loss": 3.1362,
      "step": 13315
    },
    {
      "epoch": 0.135498046875,
      "grad_norm": 7.743569850921631,
      "learning_rate": 4.779742040321494e-06,
      "loss": 3.789,
      "step": 13320
    },
    {
      "epoch": 0.13554890950520834,
      "grad_norm": 10.638326644897461,
      "learning_rate": 4.779577892207015e-06,
      "loss": 3.5861,
      "step": 13325
    },
    {
      "epoch": 0.13559977213541666,
      "grad_norm": 10.221707344055176,
      "learning_rate": 4.779413685770075e-06,
      "loss": 3.1357,
      "step": 13330
    },
    {
      "epoch": 0.135650634765625,
      "grad_norm": 12.546625137329102,
      "learning_rate": 4.779249421014876e-06,
      "loss": 3.2787,
      "step": 13335
    },
    {
      "epoch": 0.13570149739583334,
      "grad_norm": 15.528656959533691,
      "learning_rate": 4.779085097945621e-06,
      "loss": 3.4499,
      "step": 13340
    },
    {
      "epoch": 0.13575236002604166,
      "grad_norm": 15.490991592407227,
      "learning_rate": 4.778920716566514e-06,
      "loss": 3.7281,
      "step": 13345
    },
    {
      "epoch": 0.13580322265625,
      "grad_norm": 12.92549991607666,
      "learning_rate": 4.7787562768817605e-06,
      "loss": 3.3561,
      "step": 13350
    },
    {
      "epoch": 0.13585408528645834,
      "grad_norm": 14.30997371673584,
      "learning_rate": 4.778591778895568e-06,
      "loss": 3.4102,
      "step": 13355
    },
    {
      "epoch": 0.13590494791666666,
      "grad_norm": 9.108956336975098,
      "learning_rate": 4.778427222612145e-06,
      "loss": 3.2839,
      "step": 13360
    },
    {
      "epoch": 0.135955810546875,
      "grad_norm": 15.586246490478516,
      "learning_rate": 4.778262608035702e-06,
      "loss": 3.6318,
      "step": 13365
    },
    {
      "epoch": 0.13600667317708334,
      "grad_norm": 12.77320671081543,
      "learning_rate": 4.778097935170449e-06,
      "loss": 3.4799,
      "step": 13370
    },
    {
      "epoch": 0.13605753580729166,
      "grad_norm": 8.657517433166504,
      "learning_rate": 4.777933204020602e-06,
      "loss": 3.2848,
      "step": 13375
    },
    {
      "epoch": 0.1361083984375,
      "grad_norm": 13.57066822052002,
      "learning_rate": 4.777768414590372e-06,
      "loss": 3.0883,
      "step": 13380
    },
    {
      "epoch": 0.13615926106770834,
      "grad_norm": 15.412737846374512,
      "learning_rate": 4.777603566883978e-06,
      "loss": 3.3734,
      "step": 13385
    },
    {
      "epoch": 0.13621012369791666,
      "grad_norm": 7.619217395782471,
      "learning_rate": 4.777438660905637e-06,
      "loss": 3.2357,
      "step": 13390
    },
    {
      "epoch": 0.136260986328125,
      "grad_norm": 8.894469261169434,
      "learning_rate": 4.777273696659567e-06,
      "loss": 3.3946,
      "step": 13395
    },
    {
      "epoch": 0.13631184895833334,
      "grad_norm": 11.534381866455078,
      "learning_rate": 4.7771086741499895e-06,
      "loss": 3.3128,
      "step": 13400
    },
    {
      "epoch": 0.13636271158854166,
      "grad_norm": 9.838509559631348,
      "learning_rate": 4.776943593381126e-06,
      "loss": 3.5254,
      "step": 13405
    },
    {
      "epoch": 0.13641357421875,
      "grad_norm": 12.967144012451172,
      "learning_rate": 4.7767784543572e-06,
      "loss": 3.1434,
      "step": 13410
    },
    {
      "epoch": 0.13646443684895834,
      "grad_norm": 9.366642951965332,
      "learning_rate": 4.776613257082439e-06,
      "loss": 2.9998,
      "step": 13415
    },
    {
      "epoch": 0.13651529947916666,
      "grad_norm": 10.693907737731934,
      "learning_rate": 4.776448001561065e-06,
      "loss": 3.9294,
      "step": 13420
    },
    {
      "epoch": 0.136566162109375,
      "grad_norm": 14.118185997009277,
      "learning_rate": 4.7762826877973095e-06,
      "loss": 3.4184,
      "step": 13425
    },
    {
      "epoch": 0.13661702473958334,
      "grad_norm": 16.79452133178711,
      "learning_rate": 4.776117315795401e-06,
      "loss": 3.4651,
      "step": 13430
    },
    {
      "epoch": 0.13666788736979166,
      "grad_norm": 7.560031890869141,
      "learning_rate": 4.77595188555957e-06,
      "loss": 3.3598,
      "step": 13435
    },
    {
      "epoch": 0.13671875,
      "grad_norm": 8.069879531860352,
      "learning_rate": 4.77578639709405e-06,
      "loss": 3.3903,
      "step": 13440
    },
    {
      "epoch": 0.13676961263020834,
      "grad_norm": 14.968077659606934,
      "learning_rate": 4.775620850403075e-06,
      "loss": 3.1777,
      "step": 13445
    },
    {
      "epoch": 0.13682047526041666,
      "grad_norm": 11.37788200378418,
      "learning_rate": 4.775455245490879e-06,
      "loss": 3.1365,
      "step": 13450
    },
    {
      "epoch": 0.136871337890625,
      "grad_norm": 11.77219295501709,
      "learning_rate": 4.7752895823616995e-06,
      "loss": 3.397,
      "step": 13455
    },
    {
      "epoch": 0.13692220052083334,
      "grad_norm": 11.196371078491211,
      "learning_rate": 4.775123861019776e-06,
      "loss": 3.146,
      "step": 13460
    },
    {
      "epoch": 0.13697306315104166,
      "grad_norm": 19.208072662353516,
      "learning_rate": 4.774958081469348e-06,
      "loss": 3.2435,
      "step": 13465
    },
    {
      "epoch": 0.13702392578125,
      "grad_norm": 13.371184349060059,
      "learning_rate": 4.774792243714656e-06,
      "loss": 3.4511,
      "step": 13470
    },
    {
      "epoch": 0.13707478841145834,
      "grad_norm": 7.495774745941162,
      "learning_rate": 4.774626347759944e-06,
      "loss": 3.174,
      "step": 13475
    },
    {
      "epoch": 0.13712565104166666,
      "grad_norm": 9.855815887451172,
      "learning_rate": 4.774460393609456e-06,
      "loss": 3.4519,
      "step": 13480
    },
    {
      "epoch": 0.137176513671875,
      "grad_norm": 15.752311706542969,
      "learning_rate": 4.774294381267438e-06,
      "loss": 3.3032,
      "step": 13485
    },
    {
      "epoch": 0.13722737630208334,
      "grad_norm": 10.933402061462402,
      "learning_rate": 4.774128310738137e-06,
      "loss": 3.5641,
      "step": 13490
    },
    {
      "epoch": 0.13727823893229166,
      "grad_norm": 10.031787872314453,
      "learning_rate": 4.773962182025803e-06,
      "loss": 3.4982,
      "step": 13495
    },
    {
      "epoch": 0.1373291015625,
      "grad_norm": 10.56356430053711,
      "learning_rate": 4.773795995134685e-06,
      "loss": 3.2439,
      "step": 13500
    },
    {
      "epoch": 0.13737996419270834,
      "grad_norm": 7.9167585372924805,
      "learning_rate": 4.773629750069036e-06,
      "loss": 2.9149,
      "step": 13505
    },
    {
      "epoch": 0.13743082682291666,
      "grad_norm": 11.75800609588623,
      "learning_rate": 4.773463446833108e-06,
      "loss": 3.3936,
      "step": 13510
    },
    {
      "epoch": 0.137481689453125,
      "grad_norm": 10.611553192138672,
      "learning_rate": 4.773297085431156e-06,
      "loss": 3.3874,
      "step": 13515
    },
    {
      "epoch": 0.13753255208333334,
      "grad_norm": 7.607089996337891,
      "learning_rate": 4.773130665867438e-06,
      "loss": 3.1678,
      "step": 13520
    },
    {
      "epoch": 0.13758341471354166,
      "grad_norm": 13.853635787963867,
      "learning_rate": 4.7729641881462106e-06,
      "loss": 3.5818,
      "step": 13525
    },
    {
      "epoch": 0.13763427734375,
      "grad_norm": 14.964798927307129,
      "learning_rate": 4.772797652271732e-06,
      "loss": 3.722,
      "step": 13530
    },
    {
      "epoch": 0.13768513997395834,
      "grad_norm": 13.768731117248535,
      "learning_rate": 4.772631058248266e-06,
      "loss": 4.1596,
      "step": 13535
    },
    {
      "epoch": 0.13773600260416666,
      "grad_norm": 14.737395286560059,
      "learning_rate": 4.772464406080072e-06,
      "loss": 3.3029,
      "step": 13540
    },
    {
      "epoch": 0.137786865234375,
      "grad_norm": 13.914802551269531,
      "learning_rate": 4.772297695771415e-06,
      "loss": 3.4859,
      "step": 13545
    },
    {
      "epoch": 0.13783772786458334,
      "grad_norm": 11.838445663452148,
      "learning_rate": 4.7721309273265605e-06,
      "loss": 3.3564,
      "step": 13550
    },
    {
      "epoch": 0.13788859049479166,
      "grad_norm": 12.323065757751465,
      "learning_rate": 4.771964100749774e-06,
      "loss": 3.3043,
      "step": 13555
    },
    {
      "epoch": 0.137939453125,
      "grad_norm": 13.693683624267578,
      "learning_rate": 4.771797216045325e-06,
      "loss": 3.4427,
      "step": 13560
    },
    {
      "epoch": 0.13799031575520834,
      "grad_norm": 14.236191749572754,
      "learning_rate": 4.771630273217483e-06,
      "loss": 3.3801,
      "step": 13565
    },
    {
      "epoch": 0.13804117838541666,
      "grad_norm": 12.197969436645508,
      "learning_rate": 4.7714632722705175e-06,
      "loss": 3.6923,
      "step": 13570
    },
    {
      "epoch": 0.138092041015625,
      "grad_norm": 12.962698936462402,
      "learning_rate": 4.771296213208704e-06,
      "loss": 3.3643,
      "step": 13575
    },
    {
      "epoch": 0.13814290364583334,
      "grad_norm": 15.476211547851562,
      "learning_rate": 4.7711290960363145e-06,
      "loss": 3.3381,
      "step": 13580
    },
    {
      "epoch": 0.13819376627604166,
      "grad_norm": 12.055511474609375,
      "learning_rate": 4.770961920757626e-06,
      "loss": 3.1077,
      "step": 13585
    },
    {
      "epoch": 0.13824462890625,
      "grad_norm": 11.437824249267578,
      "learning_rate": 4.7707946873769144e-06,
      "loss": 3.4872,
      "step": 13590
    },
    {
      "epoch": 0.13829549153645834,
      "grad_norm": 14.684785842895508,
      "learning_rate": 4.77062739589846e-06,
      "loss": 3.1012,
      "step": 13595
    },
    {
      "epoch": 0.13834635416666666,
      "grad_norm": 11.902538299560547,
      "learning_rate": 4.77046004632654e-06,
      "loss": 3.3424,
      "step": 13600
    },
    {
      "epoch": 0.138397216796875,
      "grad_norm": 13.112920761108398,
      "learning_rate": 4.770292638665439e-06,
      "loss": 3.6353,
      "step": 13605
    },
    {
      "epoch": 0.13844807942708334,
      "grad_norm": 12.018083572387695,
      "learning_rate": 4.7701251729194396e-06,
      "loss": 3.7497,
      "step": 13610
    },
    {
      "epoch": 0.13849894205729166,
      "grad_norm": 12.88236141204834,
      "learning_rate": 4.769957649092825e-06,
      "loss": 3.9322,
      "step": 13615
    },
    {
      "epoch": 0.1385498046875,
      "grad_norm": 16.120954513549805,
      "learning_rate": 4.769790067189882e-06,
      "loss": 3.4474,
      "step": 13620
    },
    {
      "epoch": 0.13860066731770834,
      "grad_norm": 9.990036010742188,
      "learning_rate": 4.769622427214898e-06,
      "loss": 3.1569,
      "step": 13625
    },
    {
      "epoch": 0.13865152994791666,
      "grad_norm": 9.095274925231934,
      "learning_rate": 4.769454729172163e-06,
      "loss": 3.3547,
      "step": 13630
    },
    {
      "epoch": 0.138702392578125,
      "grad_norm": 12.9636812210083,
      "learning_rate": 4.7692869730659655e-06,
      "loss": 3.3197,
      "step": 13635
    },
    {
      "epoch": 0.13875325520833334,
      "grad_norm": 14.65504264831543,
      "learning_rate": 4.769119158900599e-06,
      "loss": 3.4425,
      "step": 13640
    },
    {
      "epoch": 0.13880411783854166,
      "grad_norm": 9.957447052001953,
      "learning_rate": 4.768951286680357e-06,
      "loss": 3.4952,
      "step": 13645
    },
    {
      "epoch": 0.13885498046875,
      "grad_norm": 12.655380249023438,
      "learning_rate": 4.768783356409535e-06,
      "loss": 3.4519,
      "step": 13650
    },
    {
      "epoch": 0.13890584309895834,
      "grad_norm": 14.230171203613281,
      "learning_rate": 4.768615368092427e-06,
      "loss": 3.2948,
      "step": 13655
    },
    {
      "epoch": 0.13895670572916666,
      "grad_norm": 15.258803367614746,
      "learning_rate": 4.768447321733332e-06,
      "loss": 3.5278,
      "step": 13660
    },
    {
      "epoch": 0.139007568359375,
      "grad_norm": 8.109949111938477,
      "learning_rate": 4.7682792173365525e-06,
      "loss": 3.2576,
      "step": 13665
    },
    {
      "epoch": 0.13905843098958334,
      "grad_norm": 12.126956939697266,
      "learning_rate": 4.768111054906384e-06,
      "loss": 3.5308,
      "step": 13670
    },
    {
      "epoch": 0.13910929361979166,
      "grad_norm": 12.557209014892578,
      "learning_rate": 4.767942834447134e-06,
      "loss": 3.405,
      "step": 13675
    },
    {
      "epoch": 0.13916015625,
      "grad_norm": 8.383021354675293,
      "learning_rate": 4.767774555963103e-06,
      "loss": 3.3727,
      "step": 13680
    },
    {
      "epoch": 0.13921101888020834,
      "grad_norm": 9.561806678771973,
      "learning_rate": 4.767606219458598e-06,
      "loss": 3.7015,
      "step": 13685
    },
    {
      "epoch": 0.13926188151041666,
      "grad_norm": 11.462172508239746,
      "learning_rate": 4.767437824937926e-06,
      "loss": 3.6486,
      "step": 13690
    },
    {
      "epoch": 0.139312744140625,
      "grad_norm": 16.29766273498535,
      "learning_rate": 4.767269372405393e-06,
      "loss": 3.0515,
      "step": 13695
    },
    {
      "epoch": 0.13936360677083334,
      "grad_norm": 8.491412162780762,
      "learning_rate": 4.767100861865311e-06,
      "loss": 3.3933,
      "step": 13700
    },
    {
      "epoch": 0.13941446940104166,
      "grad_norm": 8.013467788696289,
      "learning_rate": 4.766932293321992e-06,
      "loss": 3.3615,
      "step": 13705
    },
    {
      "epoch": 0.13946533203125,
      "grad_norm": 8.9276704788208,
      "learning_rate": 4.766763666779747e-06,
      "loss": 3.4722,
      "step": 13710
    },
    {
      "epoch": 0.13951619466145834,
      "grad_norm": 8.328059196472168,
      "learning_rate": 4.76659498224289e-06,
      "loss": 3.6743,
      "step": 13715
    },
    {
      "epoch": 0.13956705729166666,
      "grad_norm": 10.622822761535645,
      "learning_rate": 4.766426239715739e-06,
      "loss": 3.3682,
      "step": 13720
    },
    {
      "epoch": 0.139617919921875,
      "grad_norm": 13.057267189025879,
      "learning_rate": 4.766257439202609e-06,
      "loss": 3.41,
      "step": 13725
    },
    {
      "epoch": 0.13966878255208334,
      "grad_norm": 13.841679573059082,
      "learning_rate": 4.766088580707819e-06,
      "loss": 3.2495,
      "step": 13730
    },
    {
      "epoch": 0.13971964518229166,
      "grad_norm": 16.03135108947754,
      "learning_rate": 4.765919664235691e-06,
      "loss": 3.3903,
      "step": 13735
    },
    {
      "epoch": 0.1397705078125,
      "grad_norm": 8.109480857849121,
      "learning_rate": 4.765750689790545e-06,
      "loss": 3.2099,
      "step": 13740
    },
    {
      "epoch": 0.13982137044270834,
      "grad_norm": 15.432792663574219,
      "learning_rate": 4.765581657376705e-06,
      "loss": 3.2156,
      "step": 13745
    },
    {
      "epoch": 0.13987223307291666,
      "grad_norm": 13.207000732421875,
      "learning_rate": 4.7654125669984945e-06,
      "loss": 3.3425,
      "step": 13750
    },
    {
      "epoch": 0.139923095703125,
      "grad_norm": 10.853780746459961,
      "learning_rate": 4.765243418660241e-06,
      "loss": 3.5909,
      "step": 13755
    },
    {
      "epoch": 0.13997395833333334,
      "grad_norm": 11.465790748596191,
      "learning_rate": 4.765074212366271e-06,
      "loss": 3.7235,
      "step": 13760
    },
    {
      "epoch": 0.14002482096354166,
      "grad_norm": 14.370993614196777,
      "learning_rate": 4.764904948120915e-06,
      "loss": 3.6259,
      "step": 13765
    },
    {
      "epoch": 0.14007568359375,
      "grad_norm": 11.535296440124512,
      "learning_rate": 4.7647356259285025e-06,
      "loss": 3.5197,
      "step": 13770
    },
    {
      "epoch": 0.14012654622395834,
      "grad_norm": 11.556644439697266,
      "learning_rate": 4.764566245793365e-06,
      "loss": 3.5663,
      "step": 13775
    },
    {
      "epoch": 0.14017740885416666,
      "grad_norm": 14.494634628295898,
      "learning_rate": 4.764396807719838e-06,
      "loss": 3.4005,
      "step": 13780
    },
    {
      "epoch": 0.140228271484375,
      "grad_norm": 14.56704330444336,
      "learning_rate": 4.764227311712255e-06,
      "loss": 3.335,
      "step": 13785
    },
    {
      "epoch": 0.14027913411458334,
      "grad_norm": 10.961151123046875,
      "learning_rate": 4.764057757774953e-06,
      "loss": 3.7573,
      "step": 13790
    },
    {
      "epoch": 0.14032999674479166,
      "grad_norm": 8.999013900756836,
      "learning_rate": 4.76388814591227e-06,
      "loss": 3.4393,
      "step": 13795
    },
    {
      "epoch": 0.140380859375,
      "grad_norm": 85.62931823730469,
      "learning_rate": 4.763718476128545e-06,
      "loss": 3.9256,
      "step": 13800
    },
    {
      "epoch": 0.14043172200520834,
      "grad_norm": 11.35169792175293,
      "learning_rate": 4.763548748428119e-06,
      "loss": 3.6844,
      "step": 13805
    },
    {
      "epoch": 0.14048258463541666,
      "grad_norm": 10.133095741271973,
      "learning_rate": 4.763378962815335e-06,
      "loss": 3.3458,
      "step": 13810
    },
    {
      "epoch": 0.140533447265625,
      "grad_norm": 14.392997741699219,
      "learning_rate": 4.763209119294537e-06,
      "loss": 3.4801,
      "step": 13815
    },
    {
      "epoch": 0.14058430989583334,
      "grad_norm": 13.355161666870117,
      "learning_rate": 4.76303921787007e-06,
      "loss": 3.98,
      "step": 13820
    },
    {
      "epoch": 0.14063517252604166,
      "grad_norm": 15.28199291229248,
      "learning_rate": 4.762869258546281e-06,
      "loss": 3.565,
      "step": 13825
    },
    {
      "epoch": 0.14068603515625,
      "grad_norm": 15.826837539672852,
      "learning_rate": 4.762699241327518e-06,
      "loss": 3.9634,
      "step": 13830
    },
    {
      "epoch": 0.14073689778645834,
      "grad_norm": 19.135751724243164,
      "learning_rate": 4.762529166218133e-06,
      "loss": 3.298,
      "step": 13835
    },
    {
      "epoch": 0.14078776041666666,
      "grad_norm": 14.346761703491211,
      "learning_rate": 4.7623590332224735e-06,
      "loss": 3.3069,
      "step": 13840
    },
    {
      "epoch": 0.140838623046875,
      "grad_norm": 18.030282974243164,
      "learning_rate": 4.762188842344896e-06,
      "loss": 3.2781,
      "step": 13845
    },
    {
      "epoch": 0.14088948567708334,
      "grad_norm": 11.988079071044922,
      "learning_rate": 4.762018593589752e-06,
      "loss": 3.7506,
      "step": 13850
    },
    {
      "epoch": 0.14094034830729166,
      "grad_norm": 12.060202598571777,
      "learning_rate": 4.761848286961398e-06,
      "loss": 3.3371,
      "step": 13855
    },
    {
      "epoch": 0.1409912109375,
      "grad_norm": 10.818347930908203,
      "learning_rate": 4.7616779224641925e-06,
      "loss": 3.528,
      "step": 13860
    },
    {
      "epoch": 0.14104207356770834,
      "grad_norm": 14.578571319580078,
      "learning_rate": 4.761507500102493e-06,
      "loss": 3.3091,
      "step": 13865
    },
    {
      "epoch": 0.14109293619791666,
      "grad_norm": 14.782280921936035,
      "learning_rate": 4.761337019880661e-06,
      "loss": 3.382,
      "step": 13870
    },
    {
      "epoch": 0.141143798828125,
      "grad_norm": 16.304611206054688,
      "learning_rate": 4.761166481803057e-06,
      "loss": 3.3789,
      "step": 13875
    },
    {
      "epoch": 0.14119466145833334,
      "grad_norm": 17.154775619506836,
      "learning_rate": 4.760995885874045e-06,
      "loss": 4.1372,
      "step": 13880
    },
    {
      "epoch": 0.14124552408854166,
      "grad_norm": 7.446081161499023,
      "learning_rate": 4.760825232097988e-06,
      "loss": 3.44,
      "step": 13885
    },
    {
      "epoch": 0.14129638671875,
      "grad_norm": 15.864761352539062,
      "learning_rate": 4.760654520479254e-06,
      "loss": 3.7214,
      "step": 13890
    },
    {
      "epoch": 0.14134724934895834,
      "grad_norm": 10.105338096618652,
      "learning_rate": 4.76048375102221e-06,
      "loss": 3.0877,
      "step": 13895
    },
    {
      "epoch": 0.14139811197916666,
      "grad_norm": 11.060812950134277,
      "learning_rate": 4.760312923731224e-06,
      "loss": 3.406,
      "step": 13900
    },
    {
      "epoch": 0.141448974609375,
      "grad_norm": 17.411237716674805,
      "learning_rate": 4.760142038610669e-06,
      "loss": 3.4056,
      "step": 13905
    },
    {
      "epoch": 0.14149983723958334,
      "grad_norm": 12.01961898803711,
      "learning_rate": 4.759971095664915e-06,
      "loss": 3.6575,
      "step": 13910
    },
    {
      "epoch": 0.14155069986979166,
      "grad_norm": 10.522672653198242,
      "learning_rate": 4.7598000948983355e-06,
      "loss": 3.462,
      "step": 13915
    },
    {
      "epoch": 0.1416015625,
      "grad_norm": 14.614590644836426,
      "learning_rate": 4.759629036315307e-06,
      "loss": 3.5353,
      "step": 13920
    },
    {
      "epoch": 0.14165242513020834,
      "grad_norm": 13.122455596923828,
      "learning_rate": 4.759457919920206e-06,
      "loss": 3.2238,
      "step": 13925
    },
    {
      "epoch": 0.14170328776041666,
      "grad_norm": 11.244791984558105,
      "learning_rate": 4.759286745717409e-06,
      "loss": 3.8902,
      "step": 13930
    },
    {
      "epoch": 0.141754150390625,
      "grad_norm": 13.832063674926758,
      "learning_rate": 4.759115513711296e-06,
      "loss": 3.2644,
      "step": 13935
    },
    {
      "epoch": 0.14180501302083334,
      "grad_norm": 18.43947982788086,
      "learning_rate": 4.758944223906248e-06,
      "loss": 3.3131,
      "step": 13940
    },
    {
      "epoch": 0.14185587565104166,
      "grad_norm": 15.581222534179688,
      "learning_rate": 4.758772876306647e-06,
      "loss": 3.2354,
      "step": 13945
    },
    {
      "epoch": 0.14190673828125,
      "grad_norm": 7.942548751831055,
      "learning_rate": 4.758601470916878e-06,
      "loss": 3.2157,
      "step": 13950
    },
    {
      "epoch": 0.14195760091145834,
      "grad_norm": 15.951170921325684,
      "learning_rate": 4.758430007741325e-06,
      "loss": 3.3562,
      "step": 13955
    },
    {
      "epoch": 0.14200846354166666,
      "grad_norm": 14.815132141113281,
      "learning_rate": 4.7582584867843764e-06,
      "loss": 2.939,
      "step": 13960
    },
    {
      "epoch": 0.142059326171875,
      "grad_norm": 12.236970901489258,
      "learning_rate": 4.7580869080504185e-06,
      "loss": 3.2991,
      "step": 13965
    },
    {
      "epoch": 0.14211018880208334,
      "grad_norm": 12.710593223571777,
      "learning_rate": 4.757915271543844e-06,
      "loss": 3.2584,
      "step": 13970
    },
    {
      "epoch": 0.14216105143229166,
      "grad_norm": 12.963603019714355,
      "learning_rate": 4.757743577269042e-06,
      "loss": 3.1463,
      "step": 13975
    },
    {
      "epoch": 0.1422119140625,
      "grad_norm": 10.430787086486816,
      "learning_rate": 4.7575718252304046e-06,
      "loss": 4.2165,
      "step": 13980
    },
    {
      "epoch": 0.14226277669270834,
      "grad_norm": 10.955326080322266,
      "learning_rate": 4.7574000154323274e-06,
      "loss": 3.195,
      "step": 13985
    },
    {
      "epoch": 0.14231363932291666,
      "grad_norm": 8.364349365234375,
      "learning_rate": 4.757228147879207e-06,
      "loss": 3.1683,
      "step": 13990
    },
    {
      "epoch": 0.142364501953125,
      "grad_norm": 14.662646293640137,
      "learning_rate": 4.757056222575438e-06,
      "loss": 3.2215,
      "step": 13995
    },
    {
      "epoch": 0.14241536458333334,
      "grad_norm": 14.448378562927246,
      "learning_rate": 4.756884239525422e-06,
      "loss": 3.3889,
      "step": 14000
    },
    {
      "epoch": 0.14246622721354166,
      "grad_norm": 12.963875770568848,
      "learning_rate": 4.756712198733557e-06,
      "loss": 3.2769,
      "step": 14005
    },
    {
      "epoch": 0.14251708984375,
      "grad_norm": 11.409318923950195,
      "learning_rate": 4.756540100204245e-06,
      "loss": 3.3303,
      "step": 14010
    },
    {
      "epoch": 0.14256795247395834,
      "grad_norm": 9.132806777954102,
      "learning_rate": 4.75636794394189e-06,
      "loss": 3.2503,
      "step": 14015
    },
    {
      "epoch": 0.14261881510416666,
      "grad_norm": 12.652349472045898,
      "learning_rate": 4.756195729950896e-06,
      "loss": 3.2854,
      "step": 14020
    },
    {
      "epoch": 0.142669677734375,
      "grad_norm": 11.568374633789062,
      "learning_rate": 4.756023458235668e-06,
      "loss": 3.2795,
      "step": 14025
    },
    {
      "epoch": 0.14272054036458334,
      "grad_norm": 14.18939208984375,
      "learning_rate": 4.755851128800616e-06,
      "loss": 3.2294,
      "step": 14030
    },
    {
      "epoch": 0.14277140299479166,
      "grad_norm": 12.417285919189453,
      "learning_rate": 4.755678741650146e-06,
      "loss": 3.3423,
      "step": 14035
    },
    {
      "epoch": 0.142822265625,
      "grad_norm": 12.653460502624512,
      "learning_rate": 4.755506296788671e-06,
      "loss": 3.4815,
      "step": 14040
    },
    {
      "epoch": 0.14287312825520834,
      "grad_norm": 10.841574668884277,
      "learning_rate": 4.7553337942206025e-06,
      "loss": 3.3619,
      "step": 14045
    },
    {
      "epoch": 0.14292399088541666,
      "grad_norm": 9.72189712524414,
      "learning_rate": 4.7551612339503524e-06,
      "loss": 3.0766,
      "step": 14050
    },
    {
      "epoch": 0.142974853515625,
      "grad_norm": 10.824090003967285,
      "learning_rate": 4.754988615982336e-06,
      "loss": 3.2932,
      "step": 14055
    },
    {
      "epoch": 0.14302571614583334,
      "grad_norm": 14.650367736816406,
      "learning_rate": 4.7548159403209725e-06,
      "loss": 3.3557,
      "step": 14060
    },
    {
      "epoch": 0.14307657877604166,
      "grad_norm": 13.419211387634277,
      "learning_rate": 4.7546432069706765e-06,
      "loss": 3.135,
      "step": 14065
    },
    {
      "epoch": 0.14312744140625,
      "grad_norm": 10.86098861694336,
      "learning_rate": 4.754470415935868e-06,
      "loss": 3.7383,
      "step": 14070
    },
    {
      "epoch": 0.14317830403645834,
      "grad_norm": 16.1043758392334,
      "learning_rate": 4.754297567220969e-06,
      "loss": 3.5326,
      "step": 14075
    },
    {
      "epoch": 0.14322916666666666,
      "grad_norm": 11.805451393127441,
      "learning_rate": 4.754124660830401e-06,
      "loss": 3.2176,
      "step": 14080
    },
    {
      "epoch": 0.143280029296875,
      "grad_norm": 10.292030334472656,
      "learning_rate": 4.753951696768587e-06,
      "loss": 3.5026,
      "step": 14085
    },
    {
      "epoch": 0.14333089192708334,
      "grad_norm": 9.801261901855469,
      "learning_rate": 4.753778675039954e-06,
      "loss": 3.5177,
      "step": 14090
    },
    {
      "epoch": 0.14338175455729166,
      "grad_norm": 14.867837905883789,
      "learning_rate": 4.753605595648928e-06,
      "loss": 3.1356,
      "step": 14095
    },
    {
      "epoch": 0.1434326171875,
      "grad_norm": 19.2530460357666,
      "learning_rate": 4.753432458599936e-06,
      "loss": 3.4818,
      "step": 14100
    },
    {
      "epoch": 0.14348347981770834,
      "grad_norm": 11.164162635803223,
      "learning_rate": 4.753259263897409e-06,
      "loss": 3.65,
      "step": 14105
    },
    {
      "epoch": 0.14353434244791666,
      "grad_norm": 7.864550590515137,
      "learning_rate": 4.7530860115457785e-06,
      "loss": 3.1865,
      "step": 14110
    },
    {
      "epoch": 0.143585205078125,
      "grad_norm": 7.927250862121582,
      "learning_rate": 4.7529127015494754e-06,
      "loss": 3.2852,
      "step": 14115
    },
    {
      "epoch": 0.14363606770833334,
      "grad_norm": 15.011672973632812,
      "learning_rate": 4.752739333912936e-06,
      "loss": 3.2359,
      "step": 14120
    },
    {
      "epoch": 0.14368693033854166,
      "grad_norm": 10.188263893127441,
      "learning_rate": 4.752565908640594e-06,
      "loss": 3.4599,
      "step": 14125
    },
    {
      "epoch": 0.14373779296875,
      "grad_norm": 13.60891056060791,
      "learning_rate": 4.752392425736888e-06,
      "loss": 3.5074,
      "step": 14130
    },
    {
      "epoch": 0.14378865559895834,
      "grad_norm": 7.990344524383545,
      "learning_rate": 4.752218885206255e-06,
      "loss": 3.1402,
      "step": 14135
    },
    {
      "epoch": 0.14383951822916666,
      "grad_norm": 13.563754081726074,
      "learning_rate": 4.752045287053135e-06,
      "loss": 3.6083,
      "step": 14140
    },
    {
      "epoch": 0.143890380859375,
      "grad_norm": 7.188718795776367,
      "learning_rate": 4.751871631281971e-06,
      "loss": 3.6155,
      "step": 14145
    },
    {
      "epoch": 0.14394124348958334,
      "grad_norm": 11.118795394897461,
      "learning_rate": 4.751697917897204e-06,
      "loss": 3.2539,
      "step": 14150
    },
    {
      "epoch": 0.14399210611979166,
      "grad_norm": 11.786707878112793,
      "learning_rate": 4.7515241469032805e-06,
      "loss": 3.4012,
      "step": 14155
    },
    {
      "epoch": 0.14404296875,
      "grad_norm": 15.172300338745117,
      "learning_rate": 4.751350318304645e-06,
      "loss": 3.4998,
      "step": 14160
    },
    {
      "epoch": 0.14409383138020834,
      "grad_norm": 14.242692947387695,
      "learning_rate": 4.751176432105746e-06,
      "loss": 3.374,
      "step": 14165
    },
    {
      "epoch": 0.14414469401041666,
      "grad_norm": 12.410831451416016,
      "learning_rate": 4.751002488311031e-06,
      "loss": 3.1663,
      "step": 14170
    },
    {
      "epoch": 0.144195556640625,
      "grad_norm": 16.139020919799805,
      "learning_rate": 4.75082848692495e-06,
      "loss": 3.6473,
      "step": 14175
    },
    {
      "epoch": 0.14424641927083334,
      "grad_norm": 9.563576698303223,
      "learning_rate": 4.750654427951957e-06,
      "loss": 3.133,
      "step": 14180
    },
    {
      "epoch": 0.14429728190104166,
      "grad_norm": 9.554606437683105,
      "learning_rate": 4.750480311396503e-06,
      "loss": 3.2635,
      "step": 14185
    },
    {
      "epoch": 0.14434814453125,
      "grad_norm": 16.014537811279297,
      "learning_rate": 4.750306137263044e-06,
      "loss": 3.4213,
      "step": 14190
    },
    {
      "epoch": 0.14439900716145834,
      "grad_norm": 15.562359809875488,
      "learning_rate": 4.750131905556036e-06,
      "loss": 3.2832,
      "step": 14195
    },
    {
      "epoch": 0.14444986979166666,
      "grad_norm": 15.114102363586426,
      "learning_rate": 4.749957616279937e-06,
      "loss": 3.3687,
      "step": 14200
    },
    {
      "epoch": 0.144500732421875,
      "grad_norm": 15.910360336303711,
      "learning_rate": 4.749783269439205e-06,
      "loss": 3.1607,
      "step": 14205
    },
    {
      "epoch": 0.14455159505208334,
      "grad_norm": 8.769811630249023,
      "learning_rate": 4.749608865038301e-06,
      "loss": 3.3552,
      "step": 14210
    },
    {
      "epoch": 0.14460245768229166,
      "grad_norm": 9.91917896270752,
      "learning_rate": 4.749434403081688e-06,
      "loss": 3.2902,
      "step": 14215
    },
    {
      "epoch": 0.1446533203125,
      "grad_norm": 9.88044261932373,
      "learning_rate": 4.749259883573829e-06,
      "loss": 3.4576,
      "step": 14220
    },
    {
      "epoch": 0.14470418294270834,
      "grad_norm": 15.199368476867676,
      "learning_rate": 4.749085306519189e-06,
      "loss": 3.2871,
      "step": 14225
    },
    {
      "epoch": 0.14475504557291666,
      "grad_norm": 15.931928634643555,
      "learning_rate": 4.748910671922234e-06,
      "loss": 3.5535,
      "step": 14230
    },
    {
      "epoch": 0.144805908203125,
      "grad_norm": 11.521714210510254,
      "learning_rate": 4.748735979787433e-06,
      "loss": 3.3337,
      "step": 14235
    },
    {
      "epoch": 0.14485677083333334,
      "grad_norm": 9.132640838623047,
      "learning_rate": 4.7485612301192545e-06,
      "loss": 3.599,
      "step": 14240
    },
    {
      "epoch": 0.14490763346354166,
      "grad_norm": 14.95056438446045,
      "learning_rate": 4.74838642292217e-06,
      "loss": 3.2524,
      "step": 14245
    },
    {
      "epoch": 0.14495849609375,
      "grad_norm": 13.88601303100586,
      "learning_rate": 4.748211558200653e-06,
      "loss": 3.3675,
      "step": 14250
    },
    {
      "epoch": 0.14500935872395834,
      "grad_norm": 11.538806915283203,
      "learning_rate": 4.748036635959174e-06,
      "loss": 3.4054,
      "step": 14255
    },
    {
      "epoch": 0.14506022135416666,
      "grad_norm": 12.001321792602539,
      "learning_rate": 4.747861656202212e-06,
      "loss": 3.481,
      "step": 14260
    },
    {
      "epoch": 0.145111083984375,
      "grad_norm": 13.24527359008789,
      "learning_rate": 4.747686618934242e-06,
      "loss": 2.8579,
      "step": 14265
    },
    {
      "epoch": 0.14516194661458334,
      "grad_norm": 13.960221290588379,
      "learning_rate": 4.747511524159743e-06,
      "loss": 3.6163,
      "step": 14270
    },
    {
      "epoch": 0.14521280924479166,
      "grad_norm": 14.137311935424805,
      "learning_rate": 4.747336371883194e-06,
      "loss": 3.2439,
      "step": 14275
    },
    {
      "epoch": 0.145263671875,
      "grad_norm": 13.414327621459961,
      "learning_rate": 4.747161162109076e-06,
      "loss": 3.6854,
      "step": 14280
    },
    {
      "epoch": 0.14531453450520834,
      "grad_norm": 7.808047294616699,
      "learning_rate": 4.746985894841873e-06,
      "loss": 3.4737,
      "step": 14285
    },
    {
      "epoch": 0.14536539713541666,
      "grad_norm": 7.56891393661499,
      "learning_rate": 4.746810570086069e-06,
      "loss": 3.2434,
      "step": 14290
    },
    {
      "epoch": 0.145416259765625,
      "grad_norm": 15.081562042236328,
      "learning_rate": 4.746635187846148e-06,
      "loss": 3.3065,
      "step": 14295
    },
    {
      "epoch": 0.14546712239583334,
      "grad_norm": 12.434776306152344,
      "learning_rate": 4.746459748126599e-06,
      "loss": 3.2885,
      "step": 14300
    },
    {
      "epoch": 0.14551798502604166,
      "grad_norm": 14.816580772399902,
      "learning_rate": 4.74628425093191e-06,
      "loss": 3.3603,
      "step": 14305
    },
    {
      "epoch": 0.14556884765625,
      "grad_norm": 10.694483757019043,
      "learning_rate": 4.74610869626657e-06,
      "loss": 3.4447,
      "step": 14310
    },
    {
      "epoch": 0.14561971028645834,
      "grad_norm": 11.717911720275879,
      "learning_rate": 4.745933084135071e-06,
      "loss": 3.7955,
      "step": 14315
    },
    {
      "epoch": 0.14567057291666666,
      "grad_norm": 12.842635154724121,
      "learning_rate": 4.745757414541908e-06,
      "loss": 3.63,
      "step": 14320
    },
    {
      "epoch": 0.145721435546875,
      "grad_norm": 13.658503532409668,
      "learning_rate": 4.745581687491573e-06,
      "loss": 3.5957,
      "step": 14325
    },
    {
      "epoch": 0.14577229817708334,
      "grad_norm": 8.604068756103516,
      "learning_rate": 4.745405902988563e-06,
      "loss": 3.3326,
      "step": 14330
    },
    {
      "epoch": 0.14582316080729166,
      "grad_norm": 10.351202964782715,
      "learning_rate": 4.745230061037375e-06,
      "loss": 3.2391,
      "step": 14335
    },
    {
      "epoch": 0.1458740234375,
      "grad_norm": 9.471860885620117,
      "learning_rate": 4.745054161642508e-06,
      "loss": 3.3626,
      "step": 14340
    },
    {
      "epoch": 0.14592488606770834,
      "grad_norm": 10.003438949584961,
      "learning_rate": 4.744878204808463e-06,
      "loss": 3.5786,
      "step": 14345
    },
    {
      "epoch": 0.14597574869791666,
      "grad_norm": 10.829023361206055,
      "learning_rate": 4.744702190539741e-06,
      "loss": 3.7278,
      "step": 14350
    },
    {
      "epoch": 0.146026611328125,
      "grad_norm": 9.213700294494629,
      "learning_rate": 4.744526118840844e-06,
      "loss": 2.9384,
      "step": 14355
    },
    {
      "epoch": 0.14607747395833334,
      "grad_norm": 12.245752334594727,
      "learning_rate": 4.7443499897162794e-06,
      "loss": 3.4541,
      "step": 14360
    },
    {
      "epoch": 0.14612833658854166,
      "grad_norm": 9.579618453979492,
      "learning_rate": 4.744173803170553e-06,
      "loss": 3.2422,
      "step": 14365
    },
    {
      "epoch": 0.14617919921875,
      "grad_norm": 7.907454967498779,
      "learning_rate": 4.743997559208171e-06,
      "loss": 3.2129,
      "step": 14370
    },
    {
      "epoch": 0.14623006184895834,
      "grad_norm": 15.023747444152832,
      "learning_rate": 4.743821257833644e-06,
      "loss": 3.3496,
      "step": 14375
    },
    {
      "epoch": 0.14628092447916666,
      "grad_norm": 8.13438892364502,
      "learning_rate": 4.743644899051481e-06,
      "loss": 3.4805,
      "step": 14380
    },
    {
      "epoch": 0.146331787109375,
      "grad_norm": 10.506806373596191,
      "learning_rate": 4.743468482866196e-06,
      "loss": 3.653,
      "step": 14385
    },
    {
      "epoch": 0.14638264973958334,
      "grad_norm": 13.162545204162598,
      "learning_rate": 4.743292009282301e-06,
      "loss": 3.4791,
      "step": 14390
    },
    {
      "epoch": 0.14643351236979166,
      "grad_norm": 14.953128814697266,
      "learning_rate": 4.743115478304312e-06,
      "loss": 3.3943,
      "step": 14395
    },
    {
      "epoch": 0.146484375,
      "grad_norm": 8.338605880737305,
      "learning_rate": 4.742938889936745e-06,
      "loss": 3.484,
      "step": 14400
    },
    {
      "epoch": 0.14653523763020834,
      "grad_norm": 11.624462127685547,
      "learning_rate": 4.742762244184117e-06,
      "loss": 3.1865,
      "step": 14405
    },
    {
      "epoch": 0.14658610026041666,
      "grad_norm": 9.61160945892334,
      "learning_rate": 4.74258554105095e-06,
      "loss": 3.518,
      "step": 14410
    },
    {
      "epoch": 0.146636962890625,
      "grad_norm": 10.283778190612793,
      "learning_rate": 4.742408780541763e-06,
      "loss": 3.4033,
      "step": 14415
    },
    {
      "epoch": 0.14668782552083334,
      "grad_norm": 14.608953475952148,
      "learning_rate": 4.742231962661079e-06,
      "loss": 3.0345,
      "step": 14420
    },
    {
      "epoch": 0.14673868815104166,
      "grad_norm": 13.46399974822998,
      "learning_rate": 4.742055087413422e-06,
      "loss": 3.3489,
      "step": 14425
    },
    {
      "epoch": 0.14678955078125,
      "grad_norm": 9.469012260437012,
      "learning_rate": 4.741878154803316e-06,
      "loss": 2.9819,
      "step": 14430
    },
    {
      "epoch": 0.14684041341145834,
      "grad_norm": 10.287275314331055,
      "learning_rate": 4.741701164835291e-06,
      "loss": 3.191,
      "step": 14435
    },
    {
      "epoch": 0.14689127604166666,
      "grad_norm": 12.825785636901855,
      "learning_rate": 4.741524117513871e-06,
      "loss": 2.9444,
      "step": 14440
    },
    {
      "epoch": 0.146942138671875,
      "grad_norm": 11.87830924987793,
      "learning_rate": 4.741347012843588e-06,
      "loss": 3.3462,
      "step": 14445
    },
    {
      "epoch": 0.14699300130208334,
      "grad_norm": 9.253486633300781,
      "learning_rate": 4.7411698508289735e-06,
      "loss": 3.7622,
      "step": 14450
    },
    {
      "epoch": 0.14704386393229166,
      "grad_norm": 11.679107666015625,
      "learning_rate": 4.740992631474559e-06,
      "loss": 3.2838,
      "step": 14455
    },
    {
      "epoch": 0.1470947265625,
      "grad_norm": 13.79211711883545,
      "learning_rate": 4.740815354784879e-06,
      "loss": 3.2918,
      "step": 14460
    },
    {
      "epoch": 0.14714558919270834,
      "grad_norm": 14.639359474182129,
      "learning_rate": 4.74063802076447e-06,
      "loss": 3.4211,
      "step": 14465
    },
    {
      "epoch": 0.14719645182291666,
      "grad_norm": 12.899956703186035,
      "learning_rate": 4.7404606294178684e-06,
      "loss": 3.6037,
      "step": 14470
    },
    {
      "epoch": 0.147247314453125,
      "grad_norm": 15.321117401123047,
      "learning_rate": 4.740283180749613e-06,
      "loss": 3.5764,
      "step": 14475
    },
    {
      "epoch": 0.14729817708333334,
      "grad_norm": 12.093023300170898,
      "learning_rate": 4.740105674764243e-06,
      "loss": 3.4525,
      "step": 14480
    },
    {
      "epoch": 0.14734903971354166,
      "grad_norm": 12.851874351501465,
      "learning_rate": 4.739928111466299e-06,
      "loss": 3.7187,
      "step": 14485
    },
    {
      "epoch": 0.14739990234375,
      "grad_norm": 11.185502052307129,
      "learning_rate": 4.739750490860327e-06,
      "loss": 3.6987,
      "step": 14490
    },
    {
      "epoch": 0.14745076497395834,
      "grad_norm": 10.547276496887207,
      "learning_rate": 4.7395728129508686e-06,
      "loss": 3.746,
      "step": 14495
    },
    {
      "epoch": 0.14750162760416666,
      "grad_norm": 14.449246406555176,
      "learning_rate": 4.739395077742471e-06,
      "loss": 3.6518,
      "step": 14500
    },
    {
      "epoch": 0.147552490234375,
      "grad_norm": 14.639830589294434,
      "learning_rate": 4.739217285239681e-06,
      "loss": 4.1111,
      "step": 14505
    },
    {
      "epoch": 0.14760335286458334,
      "grad_norm": 13.559803009033203,
      "learning_rate": 4.739039435447047e-06,
      "loss": 3.5169,
      "step": 14510
    },
    {
      "epoch": 0.14765421549479166,
      "grad_norm": 17.011873245239258,
      "learning_rate": 4.738861528369121e-06,
      "loss": 3.3476,
      "step": 14515
    },
    {
      "epoch": 0.147705078125,
      "grad_norm": 10.405207633972168,
      "learning_rate": 4.7386835640104525e-06,
      "loss": 3.1015,
      "step": 14520
    },
    {
      "epoch": 0.14775594075520834,
      "grad_norm": 15.131099700927734,
      "learning_rate": 4.738505542375595e-06,
      "loss": 3.7291,
      "step": 14525
    },
    {
      "epoch": 0.14780680338541666,
      "grad_norm": 17.359588623046875,
      "learning_rate": 4.738327463469105e-06,
      "loss": 3.3012,
      "step": 14530
    },
    {
      "epoch": 0.147857666015625,
      "grad_norm": 8.4977445602417,
      "learning_rate": 4.738149327295537e-06,
      "loss": 4.0538,
      "step": 14535
    },
    {
      "epoch": 0.14790852864583334,
      "grad_norm": 14.337766647338867,
      "learning_rate": 4.737971133859449e-06,
      "loss": 2.9475,
      "step": 14540
    },
    {
      "epoch": 0.14795939127604166,
      "grad_norm": 6.989287853240967,
      "learning_rate": 4.7377928831654e-06,
      "loss": 3.5471,
      "step": 14545
    },
    {
      "epoch": 0.14801025390625,
      "grad_norm": 8.3816499710083,
      "learning_rate": 4.73761457521795e-06,
      "loss": 3.4316,
      "step": 14550
    },
    {
      "epoch": 0.14806111653645834,
      "grad_norm": 10.17712116241455,
      "learning_rate": 4.7374362100216625e-06,
      "loss": 3.5723,
      "step": 14555
    },
    {
      "epoch": 0.14811197916666666,
      "grad_norm": 13.229191780090332,
      "learning_rate": 4.737257787581099e-06,
      "loss": 3.5133,
      "step": 14560
    },
    {
      "epoch": 0.148162841796875,
      "grad_norm": 13.506834983825684,
      "learning_rate": 4.737079307900826e-06,
      "loss": 3.5203,
      "step": 14565
    },
    {
      "epoch": 0.14821370442708334,
      "grad_norm": 13.098067283630371,
      "learning_rate": 4.736900770985409e-06,
      "loss": 3.1275,
      "step": 14570
    },
    {
      "epoch": 0.14826456705729166,
      "grad_norm": 15.300044059753418,
      "learning_rate": 4.7367221768394155e-06,
      "loss": 3.3876,
      "step": 14575
    },
    {
      "epoch": 0.1483154296875,
      "grad_norm": 11.651641845703125,
      "learning_rate": 4.736543525467415e-06,
      "loss": 3.1862,
      "step": 14580
    },
    {
      "epoch": 0.14836629231770834,
      "grad_norm": 17.858064651489258,
      "learning_rate": 4.736364816873979e-06,
      "loss": 3.2058,
      "step": 14585
    },
    {
      "epoch": 0.14841715494791666,
      "grad_norm": 11.784101486206055,
      "learning_rate": 4.73618605106368e-06,
      "loss": 3.3627,
      "step": 14590
    },
    {
      "epoch": 0.148468017578125,
      "grad_norm": 16.125946044921875,
      "learning_rate": 4.73600722804109e-06,
      "loss": 3.2323,
      "step": 14595
    },
    {
      "epoch": 0.14851888020833334,
      "grad_norm": 17.72896385192871,
      "learning_rate": 4.735828347810785e-06,
      "loss": 3.542,
      "step": 14600
    },
    {
      "epoch": 0.14856974283854166,
      "grad_norm": 13.747530937194824,
      "learning_rate": 4.735649410377342e-06,
      "loss": 3.0246,
      "step": 14605
    },
    {
      "epoch": 0.14862060546875,
      "grad_norm": 10.233755111694336,
      "learning_rate": 4.735470415745339e-06,
      "loss": 3.6899,
      "step": 14610
    },
    {
      "epoch": 0.14867146809895834,
      "grad_norm": 13.699902534484863,
      "learning_rate": 4.735291363919355e-06,
      "loss": 3.1592,
      "step": 14615
    },
    {
      "epoch": 0.14872233072916666,
      "grad_norm": 11.987502098083496,
      "learning_rate": 4.735112254903971e-06,
      "loss": 3.3525,
      "step": 14620
    },
    {
      "epoch": 0.148773193359375,
      "grad_norm": 13.890175819396973,
      "learning_rate": 4.73493308870377e-06,
      "loss": 3.3058,
      "step": 14625
    },
    {
      "epoch": 0.14882405598958334,
      "grad_norm": 14.703634262084961,
      "learning_rate": 4.734753865323336e-06,
      "loss": 3.3691,
      "step": 14630
    },
    {
      "epoch": 0.14887491861979166,
      "grad_norm": 11.793313980102539,
      "learning_rate": 4.734574584767253e-06,
      "loss": 3.1809,
      "step": 14635
    },
    {
      "epoch": 0.14892578125,
      "grad_norm": 13.151676177978516,
      "learning_rate": 4.73439524704011e-06,
      "loss": 3.4036,
      "step": 14640
    },
    {
      "epoch": 0.14897664388020834,
      "grad_norm": 10.763456344604492,
      "learning_rate": 4.734215852146493e-06,
      "loss": 3.2028,
      "step": 14645
    },
    {
      "epoch": 0.14902750651041666,
      "grad_norm": 15.11712646484375,
      "learning_rate": 4.734036400090994e-06,
      "loss": 3.5627,
      "step": 14650
    },
    {
      "epoch": 0.149078369140625,
      "grad_norm": 7.594333171844482,
      "learning_rate": 4.7338568908782036e-06,
      "loss": 3.526,
      "step": 14655
    },
    {
      "epoch": 0.14912923177083334,
      "grad_norm": 16.161596298217773,
      "learning_rate": 4.733677324512713e-06,
      "loss": 3.3709,
      "step": 14660
    },
    {
      "epoch": 0.14918009440104166,
      "grad_norm": 9.044793128967285,
      "learning_rate": 4.733497700999119e-06,
      "loss": 3.2466,
      "step": 14665
    },
    {
      "epoch": 0.14923095703125,
      "grad_norm": 9.249896049499512,
      "learning_rate": 4.733318020342014e-06,
      "loss": 3.7076,
      "step": 14670
    },
    {
      "epoch": 0.14928181966145834,
      "grad_norm": 9.653711318969727,
      "learning_rate": 4.7331382825459985e-06,
      "loss": 3.3792,
      "step": 14675
    },
    {
      "epoch": 0.14933268229166666,
      "grad_norm": 10.778152465820312,
      "learning_rate": 4.732958487615668e-06,
      "loss": 3.2468,
      "step": 14680
    },
    {
      "epoch": 0.149383544921875,
      "grad_norm": 12.605219841003418,
      "learning_rate": 4.7327786355556235e-06,
      "loss": 3.717,
      "step": 14685
    },
    {
      "epoch": 0.14943440755208334,
      "grad_norm": 11.11141300201416,
      "learning_rate": 4.7325987263704685e-06,
      "loss": 3.1906,
      "step": 14690
    },
    {
      "epoch": 0.14948527018229166,
      "grad_norm": 13.047835350036621,
      "learning_rate": 4.732418760064803e-06,
      "loss": 3.5693,
      "step": 14695
    },
    {
      "epoch": 0.1495361328125,
      "grad_norm": 12.675945281982422,
      "learning_rate": 4.7322387366432335e-06,
      "loss": 3.2765,
      "step": 14700
    },
    {
      "epoch": 0.14958699544270834,
      "grad_norm": 11.04984188079834,
      "learning_rate": 4.732058656110364e-06,
      "loss": 3.0329,
      "step": 14705
    },
    {
      "epoch": 0.14963785807291666,
      "grad_norm": 14.509688377380371,
      "learning_rate": 4.7318785184708035e-06,
      "loss": 3.5102,
      "step": 14710
    },
    {
      "epoch": 0.149688720703125,
      "grad_norm": 12.124781608581543,
      "learning_rate": 4.731698323729161e-06,
      "loss": 3.2282,
      "step": 14715
    },
    {
      "epoch": 0.14973958333333334,
      "grad_norm": 10.267120361328125,
      "learning_rate": 4.731518071890045e-06,
      "loss": 3.3788,
      "step": 14720
    },
    {
      "epoch": 0.14979044596354166,
      "grad_norm": 9.034321784973145,
      "learning_rate": 4.731337762958067e-06,
      "loss": 3.4097,
      "step": 14725
    },
    {
      "epoch": 0.14984130859375,
      "grad_norm": 13.264888763427734,
      "learning_rate": 4.731157396937842e-06,
      "loss": 3.5671,
      "step": 14730
    },
    {
      "epoch": 0.14989217122395834,
      "grad_norm": 12.145549774169922,
      "learning_rate": 4.730976973833984e-06,
      "loss": 3.1394,
      "step": 14735
    },
    {
      "epoch": 0.14994303385416666,
      "grad_norm": 10.380547523498535,
      "learning_rate": 4.7307964936511095e-06,
      "loss": 3.4537,
      "step": 14740
    },
    {
      "epoch": 0.149993896484375,
      "grad_norm": 10.687427520751953,
      "learning_rate": 4.730615956393835e-06,
      "loss": 3.2803,
      "step": 14745
    },
    {
      "epoch": 0.15004475911458334,
      "grad_norm": 9.110136985778809,
      "learning_rate": 4.730435362066779e-06,
      "loss": 3.1083,
      "step": 14750
    },
    {
      "epoch": 0.15009562174479166,
      "grad_norm": 27.618864059448242,
      "learning_rate": 4.730254710674564e-06,
      "loss": 3.4778,
      "step": 14755
    },
    {
      "epoch": 0.150146484375,
      "grad_norm": 9.159721374511719,
      "learning_rate": 4.73007400222181e-06,
      "loss": 3.5116,
      "step": 14760
    },
    {
      "epoch": 0.15019734700520834,
      "grad_norm": 13.149566650390625,
      "learning_rate": 4.729893236713142e-06,
      "loss": 3.3386,
      "step": 14765
    },
    {
      "epoch": 0.15024820963541666,
      "grad_norm": 10.829029083251953,
      "learning_rate": 4.729712414153184e-06,
      "loss": 3.862,
      "step": 14770
    },
    {
      "epoch": 0.150299072265625,
      "grad_norm": 11.96367073059082,
      "learning_rate": 4.729531534546563e-06,
      "loss": 3.66,
      "step": 14775
    },
    {
      "epoch": 0.15034993489583334,
      "grad_norm": 14.411954879760742,
      "learning_rate": 4.729350597897905e-06,
      "loss": 3.2965,
      "step": 14780
    },
    {
      "epoch": 0.15040079752604166,
      "grad_norm": 12.512883186340332,
      "learning_rate": 4.729169604211841e-06,
      "loss": 3.2615,
      "step": 14785
    },
    {
      "epoch": 0.15045166015625,
      "grad_norm": 11.264093399047852,
      "learning_rate": 4.728988553493001e-06,
      "loss": 3.1227,
      "step": 14790
    },
    {
      "epoch": 0.15050252278645834,
      "grad_norm": 7.70527458190918,
      "learning_rate": 4.728807445746018e-06,
      "loss": 3.3477,
      "step": 14795
    },
    {
      "epoch": 0.15055338541666666,
      "grad_norm": 14.769386291503906,
      "learning_rate": 4.728626280975523e-06,
      "loss": 3.6628,
      "step": 14800
    },
    {
      "epoch": 0.150604248046875,
      "grad_norm": 13.030516624450684,
      "learning_rate": 4.7284450591861545e-06,
      "loss": 3.377,
      "step": 14805
    },
    {
      "epoch": 0.15065511067708334,
      "grad_norm": 12.844939231872559,
      "learning_rate": 4.728263780382546e-06,
      "loss": 3.8196,
      "step": 14810
    },
    {
      "epoch": 0.15070597330729166,
      "grad_norm": 10.097814559936523,
      "learning_rate": 4.728082444569337e-06,
      "loss": 3.8156,
      "step": 14815
    },
    {
      "epoch": 0.1507568359375,
      "grad_norm": 10.388051986694336,
      "learning_rate": 4.727901051751167e-06,
      "loss": 3.4716,
      "step": 14820
    },
    {
      "epoch": 0.15080769856770834,
      "grad_norm": 13.543767929077148,
      "learning_rate": 4.727719601932678e-06,
      "loss": 3.4144,
      "step": 14825
    },
    {
      "epoch": 0.15085856119791666,
      "grad_norm": 11.783435821533203,
      "learning_rate": 4.727538095118509e-06,
      "loss": 3.2369,
      "step": 14830
    },
    {
      "epoch": 0.150909423828125,
      "grad_norm": 14.941376686096191,
      "learning_rate": 4.727356531313307e-06,
      "loss": 3.5681,
      "step": 14835
    },
    {
      "epoch": 0.15096028645833334,
      "grad_norm": 13.9028959274292,
      "learning_rate": 4.727174910521716e-06,
      "loss": 3.2229,
      "step": 14840
    },
    {
      "epoch": 0.15101114908854166,
      "grad_norm": 13.589739799499512,
      "learning_rate": 4.726993232748382e-06,
      "loss": 3.3788,
      "step": 14845
    },
    {
      "epoch": 0.15106201171875,
      "grad_norm": 11.502520561218262,
      "learning_rate": 4.7268114979979555e-06,
      "loss": 3.4142,
      "step": 14850
    },
    {
      "epoch": 0.15111287434895834,
      "grad_norm": 12.786849975585938,
      "learning_rate": 4.726629706275083e-06,
      "loss": 3.5127,
      "step": 14855
    },
    {
      "epoch": 0.15116373697916666,
      "grad_norm": 7.40986967086792,
      "learning_rate": 4.7264478575844185e-06,
      "loss": 3.127,
      "step": 14860
    },
    {
      "epoch": 0.151214599609375,
      "grad_norm": 10.638337135314941,
      "learning_rate": 4.726265951930612e-06,
      "loss": 3.0998,
      "step": 14865
    },
    {
      "epoch": 0.15126546223958334,
      "grad_norm": 10.695724487304688,
      "learning_rate": 4.72608398931832e-06,
      "loss": 3.5446,
      "step": 14870
    },
    {
      "epoch": 0.15131632486979166,
      "grad_norm": 11.744098663330078,
      "learning_rate": 4.7259019697521955e-06,
      "loss": 2.9158,
      "step": 14875
    },
    {
      "epoch": 0.1513671875,
      "grad_norm": 9.356399536132812,
      "learning_rate": 4.725719893236898e-06,
      "loss": 3.2486,
      "step": 14880
    },
    {
      "epoch": 0.15141805013020834,
      "grad_norm": 12.832469940185547,
      "learning_rate": 4.725537759777084e-06,
      "loss": 3.2114,
      "step": 14885
    },
    {
      "epoch": 0.15146891276041666,
      "grad_norm": 12.14478588104248,
      "learning_rate": 4.725355569377415e-06,
      "loss": 2.9888,
      "step": 14890
    },
    {
      "epoch": 0.151519775390625,
      "grad_norm": 8.157812118530273,
      "learning_rate": 4.72517332204255e-06,
      "loss": 3.3431,
      "step": 14895
    },
    {
      "epoch": 0.15157063802083334,
      "grad_norm": 12.803070068359375,
      "learning_rate": 4.724991017777153e-06,
      "loss": 3.5178,
      "step": 14900
    },
    {
      "epoch": 0.15162150065104166,
      "grad_norm": 8.299994468688965,
      "learning_rate": 4.7248086565858886e-06,
      "loss": 3.5268,
      "step": 14905
    },
    {
      "epoch": 0.15167236328125,
      "grad_norm": 12.789700508117676,
      "learning_rate": 4.724626238473421e-06,
      "loss": 3.2294,
      "step": 14910
    },
    {
      "epoch": 0.15172322591145834,
      "grad_norm": 9.37491226196289,
      "learning_rate": 4.724443763444419e-06,
      "loss": 3.22,
      "step": 14915
    },
    {
      "epoch": 0.15177408854166666,
      "grad_norm": 14.288036346435547,
      "learning_rate": 4.724261231503552e-06,
      "loss": 3.3765,
      "step": 14920
    },
    {
      "epoch": 0.151824951171875,
      "grad_norm": 17.946796417236328,
      "learning_rate": 4.724078642655487e-06,
      "loss": 3.1767,
      "step": 14925
    },
    {
      "epoch": 0.15187581380208334,
      "grad_norm": 9.992767333984375,
      "learning_rate": 4.723895996904897e-06,
      "loss": 4.5134,
      "step": 14930
    },
    {
      "epoch": 0.15192667643229166,
      "grad_norm": 12.649903297424316,
      "learning_rate": 4.7237132942564565e-06,
      "loss": 3.2216,
      "step": 14935
    },
    {
      "epoch": 0.1519775390625,
      "grad_norm": 11.577804565429688,
      "learning_rate": 4.723530534714837e-06,
      "loss": 3.3525,
      "step": 14940
    },
    {
      "epoch": 0.15202840169270834,
      "grad_norm": 11.609172821044922,
      "learning_rate": 4.723347718284716e-06,
      "loss": 3.2994,
      "step": 14945
    },
    {
      "epoch": 0.15207926432291666,
      "grad_norm": 11.450626373291016,
      "learning_rate": 4.723164844970771e-06,
      "loss": 3.1224,
      "step": 14950
    },
    {
      "epoch": 0.152130126953125,
      "grad_norm": 12.139467239379883,
      "learning_rate": 4.722981914777681e-06,
      "loss": 2.981,
      "step": 14955
    },
    {
      "epoch": 0.15218098958333334,
      "grad_norm": 14.975263595581055,
      "learning_rate": 4.722798927710124e-06,
      "loss": 3.2868,
      "step": 14960
    },
    {
      "epoch": 0.15223185221354166,
      "grad_norm": 14.647977828979492,
      "learning_rate": 4.722615883772785e-06,
      "loss": 3.3901,
      "step": 14965
    },
    {
      "epoch": 0.15228271484375,
      "grad_norm": 9.550837516784668,
      "learning_rate": 4.7224327829703444e-06,
      "loss": 3.8862,
      "step": 14970
    },
    {
      "epoch": 0.15233357747395834,
      "grad_norm": 10.410728454589844,
      "learning_rate": 4.7222496253074876e-06,
      "loss": 3.3335,
      "step": 14975
    },
    {
      "epoch": 0.15238444010416666,
      "grad_norm": 9.198328018188477,
      "learning_rate": 4.722066410788902e-06,
      "loss": 3.7736,
      "step": 14980
    },
    {
      "epoch": 0.152435302734375,
      "grad_norm": 15.138815879821777,
      "learning_rate": 4.721883139419273e-06,
      "loss": 3.2838,
      "step": 14985
    },
    {
      "epoch": 0.15248616536458334,
      "grad_norm": 10.338898658752441,
      "learning_rate": 4.721699811203291e-06,
      "loss": 3.3141,
      "step": 14990
    },
    {
      "epoch": 0.15253702799479166,
      "grad_norm": 11.960433959960938,
      "learning_rate": 4.721516426145646e-06,
      "loss": 3.3508,
      "step": 14995
    },
    {
      "epoch": 0.152587890625,
      "grad_norm": 12.279603958129883,
      "learning_rate": 4.7213329842510295e-06,
      "loss": 3.3861,
      "step": 15000
    },
    {
      "epoch": 0.15263875325520834,
      "grad_norm": 11.51431941986084,
      "learning_rate": 4.721149485524135e-06,
      "loss": 3.3131,
      "step": 15005
    },
    {
      "epoch": 0.15268961588541666,
      "grad_norm": 8.839967727661133,
      "learning_rate": 4.720965929969658e-06,
      "loss": 3.2533,
      "step": 15010
    },
    {
      "epoch": 0.152740478515625,
      "grad_norm": 8.14935302734375,
      "learning_rate": 4.720782317592293e-06,
      "loss": 3.6221,
      "step": 15015
    },
    {
      "epoch": 0.15279134114583334,
      "grad_norm": 12.631562232971191,
      "learning_rate": 4.7205986483967396e-06,
      "loss": 3.6434,
      "step": 15020
    },
    {
      "epoch": 0.15284220377604166,
      "grad_norm": 12.91220760345459,
      "learning_rate": 4.720414922387696e-06,
      "loss": 3.1931,
      "step": 15025
    },
    {
      "epoch": 0.15289306640625,
      "grad_norm": 15.801246643066406,
      "learning_rate": 4.720231139569863e-06,
      "loss": 3.4381,
      "step": 15030
    },
    {
      "epoch": 0.15294392903645834,
      "grad_norm": 11.53238296508789,
      "learning_rate": 4.720047299947943e-06,
      "loss": 3.7497,
      "step": 15035
    },
    {
      "epoch": 0.15299479166666666,
      "grad_norm": 12.755319595336914,
      "learning_rate": 4.7198634035266375e-06,
      "loss": 3.2225,
      "step": 15040
    },
    {
      "epoch": 0.153045654296875,
      "grad_norm": 9.745579719543457,
      "learning_rate": 4.719679450310654e-06,
      "loss": 3.5779,
      "step": 15045
    },
    {
      "epoch": 0.15309651692708334,
      "grad_norm": 12.108492851257324,
      "learning_rate": 4.719495440304698e-06,
      "loss": 3.9278,
      "step": 15050
    },
    {
      "epoch": 0.15314737955729166,
      "grad_norm": 13.5195951461792,
      "learning_rate": 4.719311373513477e-06,
      "loss": 3.3074,
      "step": 15055
    },
    {
      "epoch": 0.1531982421875,
      "grad_norm": 14.34796142578125,
      "learning_rate": 4.719127249941701e-06,
      "loss": 3.2327,
      "step": 15060
    },
    {
      "epoch": 0.15324910481770834,
      "grad_norm": 10.067496299743652,
      "learning_rate": 4.718943069594079e-06,
      "loss": 3.8594,
      "step": 15065
    },
    {
      "epoch": 0.15329996744791666,
      "grad_norm": 9.026041984558105,
      "learning_rate": 4.718758832475326e-06,
      "loss": 3.4053,
      "step": 15070
    },
    {
      "epoch": 0.153350830078125,
      "grad_norm": 14.132732391357422,
      "learning_rate": 4.718574538590154e-06,
      "loss": 3.3339,
      "step": 15075
    },
    {
      "epoch": 0.15340169270833334,
      "grad_norm": 13.146659851074219,
      "learning_rate": 4.718390187943278e-06,
      "loss": 4.1499,
      "step": 15080
    },
    {
      "epoch": 0.15345255533854166,
      "grad_norm": 12.891654014587402,
      "learning_rate": 4.7182057805394145e-06,
      "loss": 3.1477,
      "step": 15085
    },
    {
      "epoch": 0.15350341796875,
      "grad_norm": 8.704354286193848,
      "learning_rate": 4.718021316383282e-06,
      "loss": 3.4241,
      "step": 15090
    },
    {
      "epoch": 0.15355428059895834,
      "grad_norm": 8.283037185668945,
      "learning_rate": 4.7178367954796e-06,
      "loss": 3.4241,
      "step": 15095
    },
    {
      "epoch": 0.15360514322916666,
      "grad_norm": 14.645727157592773,
      "learning_rate": 4.7176522178330895e-06,
      "loss": 4.3103,
      "step": 15100
    },
    {
      "epoch": 0.153656005859375,
      "grad_norm": 8.679193496704102,
      "learning_rate": 4.717467583448472e-06,
      "loss": 4.3254,
      "step": 15105
    },
    {
      "epoch": 0.15370686848958334,
      "grad_norm": 22.80760383605957,
      "learning_rate": 4.717282892330472e-06,
      "loss": 3.3115,
      "step": 15110
    },
    {
      "epoch": 0.15375773111979166,
      "grad_norm": 10.638278007507324,
      "learning_rate": 4.717098144483815e-06,
      "loss": 3.4555,
      "step": 15115
    },
    {
      "epoch": 0.15380859375,
      "grad_norm": 16.177385330200195,
      "learning_rate": 4.7169133399132285e-06,
      "loss": 3.2353,
      "step": 15120
    },
    {
      "epoch": 0.15385945638020834,
      "grad_norm": 13.588946342468262,
      "learning_rate": 4.7167284786234385e-06,
      "loss": 3.512,
      "step": 15125
    },
    {
      "epoch": 0.15391031901041666,
      "grad_norm": 14.394001960754395,
      "learning_rate": 4.716543560619175e-06,
      "loss": 3.0573,
      "step": 15130
    },
    {
      "epoch": 0.153961181640625,
      "grad_norm": 9.461201667785645,
      "learning_rate": 4.716358585905172e-06,
      "loss": 3.2514,
      "step": 15135
    },
    {
      "epoch": 0.15401204427083334,
      "grad_norm": 13.264630317687988,
      "learning_rate": 4.716173554486159e-06,
      "loss": 3.2544,
      "step": 15140
    },
    {
      "epoch": 0.15406290690104166,
      "grad_norm": 11.358147621154785,
      "learning_rate": 4.71598846636687e-06,
      "loss": 3.5706,
      "step": 15145
    },
    {
      "epoch": 0.15411376953125,
      "grad_norm": 14.053950309753418,
      "learning_rate": 4.715803321552043e-06,
      "loss": 3.4664,
      "step": 15150
    },
    {
      "epoch": 0.15416463216145834,
      "grad_norm": 14.992704391479492,
      "learning_rate": 4.715618120046412e-06,
      "loss": 3.9273,
      "step": 15155
    },
    {
      "epoch": 0.15421549479166666,
      "grad_norm": 13.31781005859375,
      "learning_rate": 4.715432861854717e-06,
      "loss": 3.5456,
      "step": 15160
    },
    {
      "epoch": 0.154266357421875,
      "grad_norm": 8.325743675231934,
      "learning_rate": 4.715247546981697e-06,
      "loss": 3.3918,
      "step": 15165
    },
    {
      "epoch": 0.15431722005208334,
      "grad_norm": 11.262593269348145,
      "learning_rate": 4.715062175432093e-06,
      "loss": 3.5529,
      "step": 15170
    },
    {
      "epoch": 0.15436808268229166,
      "grad_norm": 10.394039154052734,
      "learning_rate": 4.71487674721065e-06,
      "loss": 3.3751,
      "step": 15175
    },
    {
      "epoch": 0.1544189453125,
      "grad_norm": 16.578327178955078,
      "learning_rate": 4.7146912623221095e-06,
      "loss": 3.1776,
      "step": 15180
    },
    {
      "epoch": 0.15446980794270834,
      "grad_norm": 11.776280403137207,
      "learning_rate": 4.7145057207712175e-06,
      "loss": 3.5786,
      "step": 15185
    },
    {
      "epoch": 0.15452067057291666,
      "grad_norm": 12.497096061706543,
      "learning_rate": 4.714320122562722e-06,
      "loss": 3.0066,
      "step": 15190
    },
    {
      "epoch": 0.154571533203125,
      "grad_norm": 9.880024909973145,
      "learning_rate": 4.714134467701371e-06,
      "loss": 3.3478,
      "step": 15195
    },
    {
      "epoch": 0.15462239583333334,
      "grad_norm": 9.223662376403809,
      "learning_rate": 4.713948756191915e-06,
      "loss": 3.3263,
      "step": 15200
    },
    {
      "epoch": 0.15467325846354166,
      "grad_norm": 7.955392837524414,
      "learning_rate": 4.713762988039105e-06,
      "loss": 3.4025,
      "step": 15205
    },
    {
      "epoch": 0.15472412109375,
      "grad_norm": 10.604644775390625,
      "learning_rate": 4.713577163247692e-06,
      "loss": 2.9697,
      "step": 15210
    },
    {
      "epoch": 0.15477498372395834,
      "grad_norm": 7.452188014984131,
      "learning_rate": 4.713391281822433e-06,
      "loss": 3.6369,
      "step": 15215
    },
    {
      "epoch": 0.15482584635416666,
      "grad_norm": 12.423209190368652,
      "learning_rate": 4.713205343768082e-06,
      "loss": 3.3861,
      "step": 15220
    },
    {
      "epoch": 0.154876708984375,
      "grad_norm": 17.55681610107422,
      "learning_rate": 4.713019349089399e-06,
      "loss": 3.6858,
      "step": 15225
    },
    {
      "epoch": 0.15492757161458334,
      "grad_norm": 14.02679443359375,
      "learning_rate": 4.712833297791138e-06,
      "loss": 3.167,
      "step": 15230
    },
    {
      "epoch": 0.15497843424479166,
      "grad_norm": 11.839001655578613,
      "learning_rate": 4.712647189878063e-06,
      "loss": 2.955,
      "step": 15235
    },
    {
      "epoch": 0.155029296875,
      "grad_norm": 11.20364761352539,
      "learning_rate": 4.712461025354933e-06,
      "loss": 3.0909,
      "step": 15240
    },
    {
      "epoch": 0.15508015950520834,
      "grad_norm": 18.19971466064453,
      "learning_rate": 4.712274804226513e-06,
      "loss": 3.2367,
      "step": 15245
    },
    {
      "epoch": 0.15513102213541666,
      "grad_norm": 7.526338577270508,
      "learning_rate": 4.712088526497566e-06,
      "loss": 3.3709,
      "step": 15250
    },
    {
      "epoch": 0.155181884765625,
      "grad_norm": 15.513593673706055,
      "learning_rate": 4.711902192172858e-06,
      "loss": 3.9162,
      "step": 15255
    },
    {
      "epoch": 0.15523274739583334,
      "grad_norm": 10.80632495880127,
      "learning_rate": 4.7117158012571585e-06,
      "loss": 3.1795,
      "step": 15260
    },
    {
      "epoch": 0.15528361002604166,
      "grad_norm": 13.773353576660156,
      "learning_rate": 4.711529353755233e-06,
      "loss": 3.7235,
      "step": 15265
    },
    {
      "epoch": 0.15533447265625,
      "grad_norm": 9.968823432922363,
      "learning_rate": 4.711342849671853e-06,
      "loss": 3.4381,
      "step": 15270
    },
    {
      "epoch": 0.15538533528645834,
      "grad_norm": 7.348340034484863,
      "learning_rate": 4.711156289011792e-06,
      "loss": 2.9837,
      "step": 15275
    },
    {
      "epoch": 0.15543619791666666,
      "grad_norm": 19.31353187561035,
      "learning_rate": 4.710969671779819e-06,
      "loss": 3.2548,
      "step": 15280
    },
    {
      "epoch": 0.155487060546875,
      "grad_norm": 13.9642915725708,
      "learning_rate": 4.7107829979807124e-06,
      "loss": 3.1202,
      "step": 15285
    },
    {
      "epoch": 0.15553792317708334,
      "grad_norm": 12.23745346069336,
      "learning_rate": 4.710596267619247e-06,
      "loss": 3.3532,
      "step": 15290
    },
    {
      "epoch": 0.15558878580729166,
      "grad_norm": 10.830820083618164,
      "learning_rate": 4.710409480700199e-06,
      "loss": 3.5973,
      "step": 15295
    },
    {
      "epoch": 0.1556396484375,
      "grad_norm": 10.229269981384277,
      "learning_rate": 4.710222637228349e-06,
      "loss": 3.5623,
      "step": 15300
    },
    {
      "epoch": 0.15569051106770834,
      "grad_norm": 11.168558120727539,
      "learning_rate": 4.710035737208477e-06,
      "loss": 3.0141,
      "step": 15305
    },
    {
      "epoch": 0.15574137369791666,
      "grad_norm": 7.080033302307129,
      "learning_rate": 4.709848780645364e-06,
      "loss": 3.4462,
      "step": 15310
    },
    {
      "epoch": 0.155792236328125,
      "grad_norm": 7.330209255218506,
      "learning_rate": 4.709661767543794e-06,
      "loss": 3.2099,
      "step": 15315
    },
    {
      "epoch": 0.15584309895833334,
      "grad_norm": 7.501708030700684,
      "learning_rate": 4.709474697908552e-06,
      "loss": 3.3722,
      "step": 15320
    },
    {
      "epoch": 0.15589396158854166,
      "grad_norm": 9.475503921508789,
      "learning_rate": 4.709287571744423e-06,
      "loss": 3.6584,
      "step": 15325
    },
    {
      "epoch": 0.15594482421875,
      "grad_norm": 14.18191146850586,
      "learning_rate": 4.709100389056195e-06,
      "loss": 3.3899,
      "step": 15330
    },
    {
      "epoch": 0.15599568684895834,
      "grad_norm": 9.33660888671875,
      "learning_rate": 4.708913149848658e-06,
      "loss": 3.455,
      "step": 15335
    },
    {
      "epoch": 0.15604654947916666,
      "grad_norm": 10.5362548828125,
      "learning_rate": 4.7087258541266e-06,
      "loss": 3.5219,
      "step": 15340
    },
    {
      "epoch": 0.156097412109375,
      "grad_norm": 13.641423225402832,
      "learning_rate": 4.7085385018948155e-06,
      "loss": 3.8866,
      "step": 15345
    },
    {
      "epoch": 0.15614827473958334,
      "grad_norm": 14.66014289855957,
      "learning_rate": 4.708351093158097e-06,
      "loss": 3.6575,
      "step": 15350
    },
    {
      "epoch": 0.15619913736979166,
      "grad_norm": 10.71635627746582,
      "learning_rate": 4.708163627921239e-06,
      "loss": 3.5451,
      "step": 15355
    },
    {
      "epoch": 0.15625,
      "grad_norm": 12.844642639160156,
      "learning_rate": 4.7079761061890374e-06,
      "loss": 3.3541,
      "step": 15360
    },
    {
      "epoch": 0.15630086263020834,
      "grad_norm": 9.12993335723877,
      "learning_rate": 4.707788527966291e-06,
      "loss": 3.3004,
      "step": 15365
    },
    {
      "epoch": 0.15635172526041666,
      "grad_norm": 8.735359191894531,
      "learning_rate": 4.707600893257799e-06,
      "loss": 3.6416,
      "step": 15370
    },
    {
      "epoch": 0.156402587890625,
      "grad_norm": 17.718563079833984,
      "learning_rate": 4.707413202068361e-06,
      "loss": 3.4838,
      "step": 15375
    },
    {
      "epoch": 0.15645345052083334,
      "grad_norm": 13.61154556274414,
      "learning_rate": 4.707225454402779e-06,
      "loss": 3.3291,
      "step": 15380
    },
    {
      "epoch": 0.15650431315104166,
      "grad_norm": 10.323596954345703,
      "learning_rate": 4.707037650265857e-06,
      "loss": 3.44,
      "step": 15385
    },
    {
      "epoch": 0.15655517578125,
      "grad_norm": 12.949178695678711,
      "learning_rate": 4.7068497896624014e-06,
      "loss": 3.6229,
      "step": 15390
    },
    {
      "epoch": 0.15660603841145834,
      "grad_norm": 10.059599876403809,
      "learning_rate": 4.706661872597215e-06,
      "loss": 3.1794,
      "step": 15395
    },
    {
      "epoch": 0.15665690104166666,
      "grad_norm": 13.678911209106445,
      "learning_rate": 4.706473899075108e-06,
      "loss": 3.9318,
      "step": 15400
    },
    {
      "epoch": 0.156707763671875,
      "grad_norm": 11.830682754516602,
      "learning_rate": 4.7062858691008906e-06,
      "loss": 3.383,
      "step": 15405
    },
    {
      "epoch": 0.15675862630208334,
      "grad_norm": 17.724088668823242,
      "learning_rate": 4.706097782679371e-06,
      "loss": 3.9705,
      "step": 15410
    },
    {
      "epoch": 0.15680948893229166,
      "grad_norm": 7.41434907913208,
      "learning_rate": 4.7059096398153624e-06,
      "loss": 3.3684,
      "step": 15415
    },
    {
      "epoch": 0.1568603515625,
      "grad_norm": 10.549930572509766,
      "learning_rate": 4.705721440513679e-06,
      "loss": 3.5617,
      "step": 15420
    },
    {
      "epoch": 0.15691121419270834,
      "grad_norm": 13.3803071975708,
      "learning_rate": 4.705533184779135e-06,
      "loss": 3.4956,
      "step": 15425
    },
    {
      "epoch": 0.15696207682291666,
      "grad_norm": 9.431879043579102,
      "learning_rate": 4.705344872616548e-06,
      "loss": 3.3123,
      "step": 15430
    },
    {
      "epoch": 0.157012939453125,
      "grad_norm": 13.48584270477295,
      "learning_rate": 4.705156504030735e-06,
      "loss": 3.5462,
      "step": 15435
    },
    {
      "epoch": 0.15706380208333334,
      "grad_norm": 12.505586624145508,
      "learning_rate": 4.7049680790265145e-06,
      "loss": 3.5568,
      "step": 15440
    },
    {
      "epoch": 0.15711466471354166,
      "grad_norm": 14.382122993469238,
      "learning_rate": 4.704779597608709e-06,
      "loss": 3.2458,
      "step": 15445
    },
    {
      "epoch": 0.15716552734375,
      "grad_norm": 17.052955627441406,
      "learning_rate": 4.70459105978214e-06,
      "loss": 3.3895,
      "step": 15450
    },
    {
      "epoch": 0.15721638997395834,
      "grad_norm": 9.036126136779785,
      "learning_rate": 4.704402465551632e-06,
      "loss": 3.3079,
      "step": 15455
    },
    {
      "epoch": 0.15726725260416666,
      "grad_norm": 10.451210021972656,
      "learning_rate": 4.704213814922008e-06,
      "loss": 3.5565,
      "step": 15460
    },
    {
      "epoch": 0.157318115234375,
      "grad_norm": 9.082598686218262,
      "learning_rate": 4.704025107898097e-06,
      "loss": 3.1954,
      "step": 15465
    },
    {
      "epoch": 0.15736897786458334,
      "grad_norm": 15.115447044372559,
      "learning_rate": 4.703836344484726e-06,
      "loss": 3.3543,
      "step": 15470
    },
    {
      "epoch": 0.15741984049479166,
      "grad_norm": 11.461282730102539,
      "learning_rate": 4.7036475246867245e-06,
      "loss": 3.7555,
      "step": 15475
    },
    {
      "epoch": 0.157470703125,
      "grad_norm": 10.950669288635254,
      "learning_rate": 4.703458648508923e-06,
      "loss": 3.5753,
      "step": 15480
    },
    {
      "epoch": 0.15752156575520834,
      "grad_norm": 10.564921379089355,
      "learning_rate": 4.703269715956154e-06,
      "loss": 3.0771,
      "step": 15485
    },
    {
      "epoch": 0.15757242838541666,
      "grad_norm": 11.57642650604248,
      "learning_rate": 4.703080727033252e-06,
      "loss": 3.1755,
      "step": 15490
    },
    {
      "epoch": 0.157623291015625,
      "grad_norm": 10.133732795715332,
      "learning_rate": 4.702891681745052e-06,
      "loss": 3.6182,
      "step": 15495
    },
    {
      "epoch": 0.15767415364583334,
      "grad_norm": 11.367161750793457,
      "learning_rate": 4.70270258009639e-06,
      "loss": 3.2131,
      "step": 15500
    },
    {
      "epoch": 0.15772501627604166,
      "grad_norm": 10.937525749206543,
      "learning_rate": 4.702513422092106e-06,
      "loss": 3.6436,
      "step": 15505
    },
    {
      "epoch": 0.15777587890625,
      "grad_norm": 7.845228672027588,
      "learning_rate": 4.7023242077370365e-06,
      "loss": 3.2949,
      "step": 15510
    },
    {
      "epoch": 0.15782674153645834,
      "grad_norm": 10.833147048950195,
      "learning_rate": 4.7021349370360246e-06,
      "loss": 3.5691,
      "step": 15515
    },
    {
      "epoch": 0.15787760416666666,
      "grad_norm": 10.04218578338623,
      "learning_rate": 4.701945609993912e-06,
      "loss": 3.6244,
      "step": 15520
    },
    {
      "epoch": 0.157928466796875,
      "grad_norm": 15.010988235473633,
      "learning_rate": 4.701756226615544e-06,
      "loss": 3.5053,
      "step": 15525
    },
    {
      "epoch": 0.15797932942708334,
      "grad_norm": 12.318516731262207,
      "learning_rate": 4.701566786905763e-06,
      "loss": 3.2215,
      "step": 15530
    },
    {
      "epoch": 0.15803019205729166,
      "grad_norm": 12.219952583312988,
      "learning_rate": 4.701377290869419e-06,
      "loss": 3.5105,
      "step": 15535
    },
    {
      "epoch": 0.1580810546875,
      "grad_norm": 15.182580947875977,
      "learning_rate": 4.701187738511358e-06,
      "loss": 3.6047,
      "step": 15540
    },
    {
      "epoch": 0.15813191731770834,
      "grad_norm": 8.09954833984375,
      "learning_rate": 4.700998129836431e-06,
      "loss": 3.2873,
      "step": 15545
    },
    {
      "epoch": 0.15818277994791666,
      "grad_norm": 12.153461456298828,
      "learning_rate": 4.700808464849489e-06,
      "loss": 3.2927,
      "step": 15550
    },
    {
      "epoch": 0.158233642578125,
      "grad_norm": 16.8814754486084,
      "learning_rate": 4.700618743555384e-06,
      "loss": 3.6992,
      "step": 15555
    },
    {
      "epoch": 0.15828450520833334,
      "grad_norm": 8.98434066772461,
      "learning_rate": 4.700428965958968e-06,
      "loss": 3.2662,
      "step": 15560
    },
    {
      "epoch": 0.15833536783854166,
      "grad_norm": 14.116239547729492,
      "learning_rate": 4.700239132065101e-06,
      "loss": 3.6816,
      "step": 15565
    },
    {
      "epoch": 0.15838623046875,
      "grad_norm": 14.018820762634277,
      "learning_rate": 4.700049241878637e-06,
      "loss": 3.4,
      "step": 15570
    },
    {
      "epoch": 0.15843709309895834,
      "grad_norm": 9.490670204162598,
      "learning_rate": 4.699859295404433e-06,
      "loss": 3.3853,
      "step": 15575
    },
    {
      "epoch": 0.15848795572916666,
      "grad_norm": 12.291707992553711,
      "learning_rate": 4.699669292647352e-06,
      "loss": 3.4993,
      "step": 15580
    },
    {
      "epoch": 0.158538818359375,
      "grad_norm": 11.097095489501953,
      "learning_rate": 4.699479233612252e-06,
      "loss": 3.4258,
      "step": 15585
    },
    {
      "epoch": 0.15858968098958334,
      "grad_norm": 16.145360946655273,
      "learning_rate": 4.699289118303998e-06,
      "loss": 3.6462,
      "step": 15590
    },
    {
      "epoch": 0.15864054361979166,
      "grad_norm": 13.295560836791992,
      "learning_rate": 4.699098946727454e-06,
      "loss": 3.7865,
      "step": 15595
    },
    {
      "epoch": 0.15869140625,
      "grad_norm": 7.7286152839660645,
      "learning_rate": 4.6989087188874835e-06,
      "loss": 3.2336,
      "step": 15600
    },
    {
      "epoch": 0.15874226888020834,
      "grad_norm": 8.864896774291992,
      "learning_rate": 4.698718434788955e-06,
      "loss": 3.5222,
      "step": 15605
    },
    {
      "epoch": 0.15879313151041666,
      "grad_norm": 16.60683822631836,
      "learning_rate": 4.698528094436737e-06,
      "loss": 3.5871,
      "step": 15610
    },
    {
      "epoch": 0.158843994140625,
      "grad_norm": 13.247583389282227,
      "learning_rate": 4.698337697835697e-06,
      "loss": 3.3722,
      "step": 15615
    },
    {
      "epoch": 0.15889485677083334,
      "grad_norm": 10.77502155303955,
      "learning_rate": 4.69814724499071e-06,
      "loss": 3.3955,
      "step": 15620
    },
    {
      "epoch": 0.15894571940104166,
      "grad_norm": 12.344950675964355,
      "learning_rate": 4.697956735906646e-06,
      "loss": 3.197,
      "step": 15625
    },
    {
      "epoch": 0.15899658203125,
      "grad_norm": 14.187060356140137,
      "learning_rate": 4.6977661705883805e-06,
      "loss": 3.3216,
      "step": 15630
    },
    {
      "epoch": 0.15904744466145834,
      "grad_norm": 14.362591743469238,
      "learning_rate": 4.697575549040788e-06,
      "loss": 3.4256,
      "step": 15635
    },
    {
      "epoch": 0.15909830729166666,
      "grad_norm": 10.191631317138672,
      "learning_rate": 4.697384871268745e-06,
      "loss": 3.4145,
      "step": 15640
    },
    {
      "epoch": 0.159149169921875,
      "grad_norm": 15.049766540527344,
      "learning_rate": 4.697194137277132e-06,
      "loss": 3.4378,
      "step": 15645
    },
    {
      "epoch": 0.15920003255208334,
      "grad_norm": 16.25983428955078,
      "learning_rate": 4.697003347070828e-06,
      "loss": 3.3361,
      "step": 15650
    },
    {
      "epoch": 0.15925089518229166,
      "grad_norm": 17.122467041015625,
      "learning_rate": 4.696812500654714e-06,
      "loss": 3.2949,
      "step": 15655
    },
    {
      "epoch": 0.1593017578125,
      "grad_norm": 11.814437866210938,
      "learning_rate": 4.696621598033673e-06,
      "loss": 3.1715,
      "step": 15660
    },
    {
      "epoch": 0.15935262044270834,
      "grad_norm": 9.952800750732422,
      "learning_rate": 4.696430639212588e-06,
      "loss": 3.3326,
      "step": 15665
    },
    {
      "epoch": 0.15940348307291666,
      "grad_norm": 13.430285453796387,
      "learning_rate": 4.696239624196346e-06,
      "loss": 3.542,
      "step": 15670
    },
    {
      "epoch": 0.159454345703125,
      "grad_norm": 8.788776397705078,
      "learning_rate": 4.696048552989835e-06,
      "loss": 3.3412,
      "step": 15675
    },
    {
      "epoch": 0.15950520833333334,
      "grad_norm": 11.158987045288086,
      "learning_rate": 4.69585742559794e-06,
      "loss": 3.1511,
      "step": 15680
    },
    {
      "epoch": 0.15955607096354166,
      "grad_norm": 8.167466163635254,
      "learning_rate": 4.695666242025556e-06,
      "loss": 3.7305,
      "step": 15685
    },
    {
      "epoch": 0.15960693359375,
      "grad_norm": 13.274105072021484,
      "learning_rate": 4.69547500227757e-06,
      "loss": 3.3312,
      "step": 15690
    },
    {
      "epoch": 0.15965779622395834,
      "grad_norm": 10.515645027160645,
      "learning_rate": 4.6952837063588766e-06,
      "loss": 3.07,
      "step": 15695
    },
    {
      "epoch": 0.15970865885416666,
      "grad_norm": 12.83730411529541,
      "learning_rate": 4.69509235427437e-06,
      "loss": 3.5047,
      "step": 15700
    },
    {
      "epoch": 0.159759521484375,
      "grad_norm": 12.311369895935059,
      "learning_rate": 4.694900946028946e-06,
      "loss": 3.6497,
      "step": 15705
    },
    {
      "epoch": 0.15981038411458334,
      "grad_norm": 12.914471626281738,
      "learning_rate": 4.694709481627502e-06,
      "loss": 3.2395,
      "step": 15710
    },
    {
      "epoch": 0.15986124674479166,
      "grad_norm": 14.76524543762207,
      "learning_rate": 4.694517961074934e-06,
      "loss": 3.2397,
      "step": 15715
    },
    {
      "epoch": 0.159912109375,
      "grad_norm": 18.405723571777344,
      "learning_rate": 4.694326384376146e-06,
      "loss": 3.3388,
      "step": 15720
    },
    {
      "epoch": 0.15996297200520834,
      "grad_norm": 9.726785659790039,
      "learning_rate": 4.694134751536038e-06,
      "loss": 3.287,
      "step": 15725
    },
    {
      "epoch": 0.16001383463541666,
      "grad_norm": 12.435685157775879,
      "learning_rate": 4.693943062559512e-06,
      "loss": 3.2259,
      "step": 15730
    },
    {
      "epoch": 0.160064697265625,
      "grad_norm": 11.35924243927002,
      "learning_rate": 4.693751317451472e-06,
      "loss": 3.1574,
      "step": 15735
    },
    {
      "epoch": 0.16011555989583334,
      "grad_norm": 10.276074409484863,
      "learning_rate": 4.693559516216825e-06,
      "loss": 3.3365,
      "step": 15740
    },
    {
      "epoch": 0.16016642252604166,
      "grad_norm": 8.467253684997559,
      "learning_rate": 4.693367658860478e-06,
      "loss": 3.1872,
      "step": 15745
    },
    {
      "epoch": 0.16021728515625,
      "grad_norm": 11.348249435424805,
      "learning_rate": 4.693175745387339e-06,
      "loss": 3.3335,
      "step": 15750
    },
    {
      "epoch": 0.16026814778645834,
      "grad_norm": 12.088251113891602,
      "learning_rate": 4.692983775802318e-06,
      "loss": 3.4162,
      "step": 15755
    },
    {
      "epoch": 0.16031901041666666,
      "grad_norm": 10.288962364196777,
      "learning_rate": 4.692791750110327e-06,
      "loss": 3.4439,
      "step": 15760
    },
    {
      "epoch": 0.160369873046875,
      "grad_norm": 15.410057067871094,
      "learning_rate": 4.692599668316279e-06,
      "loss": 3.4407,
      "step": 15765
    },
    {
      "epoch": 0.16042073567708334,
      "grad_norm": 13.116768836975098,
      "learning_rate": 4.692407530425089e-06,
      "loss": 3.4639,
      "step": 15770
    },
    {
      "epoch": 0.16047159830729166,
      "grad_norm": 9.727956771850586,
      "learning_rate": 4.692215336441671e-06,
      "loss": 3.3015,
      "step": 15775
    },
    {
      "epoch": 0.1605224609375,
      "grad_norm": 11.078639030456543,
      "learning_rate": 4.692023086370944e-06,
      "loss": 3.265,
      "step": 15780
    },
    {
      "epoch": 0.16057332356770834,
      "grad_norm": 7.883986473083496,
      "learning_rate": 4.6918307802178255e-06,
      "loss": 3.4122,
      "step": 15785
    },
    {
      "epoch": 0.16062418619791666,
      "grad_norm": 12.443114280700684,
      "learning_rate": 4.6916384179872356e-06,
      "loss": 3.0951,
      "step": 15790
    },
    {
      "epoch": 0.160675048828125,
      "grad_norm": 9.635088920593262,
      "learning_rate": 4.691445999684097e-06,
      "loss": 3.8282,
      "step": 15795
    },
    {
      "epoch": 0.16072591145833334,
      "grad_norm": 8.927238464355469,
      "learning_rate": 4.69125352531333e-06,
      "loss": 3.4864,
      "step": 15800
    },
    {
      "epoch": 0.16077677408854166,
      "grad_norm": 10.879573822021484,
      "learning_rate": 4.6910609948798636e-06,
      "loss": 3.7555,
      "step": 15805
    },
    {
      "epoch": 0.16082763671875,
      "grad_norm": 8.361101150512695,
      "learning_rate": 4.69086840838862e-06,
      "loss": 3.5791,
      "step": 15810
    },
    {
      "epoch": 0.16087849934895834,
      "grad_norm": 10.21220588684082,
      "learning_rate": 4.6906757658445265e-06,
      "loss": 3.2996,
      "step": 15815
    },
    {
      "epoch": 0.16092936197916666,
      "grad_norm": 12.77525520324707,
      "learning_rate": 4.690483067252514e-06,
      "loss": 3.4817,
      "step": 15820
    },
    {
      "epoch": 0.160980224609375,
      "grad_norm": 12.158894538879395,
      "learning_rate": 4.690290312617512e-06,
      "loss": 3.3044,
      "step": 15825
    },
    {
      "epoch": 0.16103108723958334,
      "grad_norm": 15.93702507019043,
      "learning_rate": 4.69009750194445e-06,
      "loss": 3.3049,
      "step": 15830
    },
    {
      "epoch": 0.16108194986979166,
      "grad_norm": 14.75804615020752,
      "learning_rate": 4.6899046352382625e-06,
      "loss": 3.2458,
      "step": 15835
    },
    {
      "epoch": 0.1611328125,
      "grad_norm": 7.818066596984863,
      "learning_rate": 4.689711712503885e-06,
      "loss": 3.3629,
      "step": 15840
    },
    {
      "epoch": 0.16118367513020834,
      "grad_norm": 11.566434860229492,
      "learning_rate": 4.689518733746251e-06,
      "loss": 3.3139,
      "step": 15845
    },
    {
      "epoch": 0.16123453776041666,
      "grad_norm": 11.620870590209961,
      "learning_rate": 4.689325698970301e-06,
      "loss": 3.3736,
      "step": 15850
    },
    {
      "epoch": 0.161285400390625,
      "grad_norm": 9.550261497497559,
      "learning_rate": 4.6891326081809705e-06,
      "loss": 3.1503,
      "step": 15855
    },
    {
      "epoch": 0.16133626302083334,
      "grad_norm": 14.424015045166016,
      "learning_rate": 4.688939461383202e-06,
      "loss": 3.3043,
      "step": 15860
    },
    {
      "epoch": 0.16138712565104166,
      "grad_norm": 14.749124526977539,
      "learning_rate": 4.688746258581936e-06,
      "loss": 3.2469,
      "step": 15865
    },
    {
      "epoch": 0.16143798828125,
      "grad_norm": 14.39970588684082,
      "learning_rate": 4.688552999782114e-06,
      "loss": 3.4534,
      "step": 15870
    },
    {
      "epoch": 0.16148885091145834,
      "grad_norm": 11.76413631439209,
      "learning_rate": 4.6883596849886845e-06,
      "loss": 3.1851,
      "step": 15875
    },
    {
      "epoch": 0.16153971354166666,
      "grad_norm": 14.800198554992676,
      "learning_rate": 4.68816631420659e-06,
      "loss": 3.7561,
      "step": 15880
    },
    {
      "epoch": 0.161590576171875,
      "grad_norm": 13.527456283569336,
      "learning_rate": 4.68797288744078e-06,
      "loss": 3.4993,
      "step": 15885
    },
    {
      "epoch": 0.16164143880208334,
      "grad_norm": 13.89983081817627,
      "learning_rate": 4.6877794046962014e-06,
      "loss": 3.5607,
      "step": 15890
    },
    {
      "epoch": 0.16169230143229166,
      "grad_norm": 12.58095932006836,
      "learning_rate": 4.687585865977806e-06,
      "loss": 3.2699,
      "step": 15895
    },
    {
      "epoch": 0.1617431640625,
      "grad_norm": 14.336568832397461,
      "learning_rate": 4.687392271290544e-06,
      "loss": 3.5323,
      "step": 15900
    },
    {
      "epoch": 0.16179402669270834,
      "grad_norm": 12.0953950881958,
      "learning_rate": 4.6871986206393695e-06,
      "loss": 3.549,
      "step": 15905
    },
    {
      "epoch": 0.16184488932291666,
      "grad_norm": 10.101968765258789,
      "learning_rate": 4.687004914029237e-06,
      "loss": 3.3465,
      "step": 15910
    },
    {
      "epoch": 0.161895751953125,
      "grad_norm": 14.160582542419434,
      "learning_rate": 4.6868111514651025e-06,
      "loss": 3.3592,
      "step": 15915
    },
    {
      "epoch": 0.16194661458333334,
      "grad_norm": 8.53564453125,
      "learning_rate": 4.686617332951922e-06,
      "loss": 3.3656,
      "step": 15920
    },
    {
      "epoch": 0.16199747721354166,
      "grad_norm": 13.714225769042969,
      "learning_rate": 4.686423458494655e-06,
      "loss": 2.912,
      "step": 15925
    },
    {
      "epoch": 0.16204833984375,
      "grad_norm": 13.133865356445312,
      "learning_rate": 4.686229528098263e-06,
      "loss": 3.5427,
      "step": 15930
    },
    {
      "epoch": 0.16209920247395834,
      "grad_norm": 10.839924812316895,
      "learning_rate": 4.686035541767707e-06,
      "loss": 3.1262,
      "step": 15935
    },
    {
      "epoch": 0.16215006510416666,
      "grad_norm": 16.54444694519043,
      "learning_rate": 4.6858414995079495e-06,
      "loss": 3.5354,
      "step": 15940
    },
    {
      "epoch": 0.162200927734375,
      "grad_norm": 14.630023002624512,
      "learning_rate": 4.685647401323955e-06,
      "loss": 3.9413,
      "step": 15945
    },
    {
      "epoch": 0.16225179036458334,
      "grad_norm": 9.191962242126465,
      "learning_rate": 4.68545324722069e-06,
      "loss": 3.011,
      "step": 15950
    },
    {
      "epoch": 0.16230265299479166,
      "grad_norm": 9.845999717712402,
      "learning_rate": 4.685259037203121e-06,
      "loss": 3.433,
      "step": 15955
    },
    {
      "epoch": 0.162353515625,
      "grad_norm": 10.282391548156738,
      "learning_rate": 4.685064771276219e-06,
      "loss": 4.0526,
      "step": 15960
    },
    {
      "epoch": 0.16240437825520834,
      "grad_norm": 11.311758995056152,
      "learning_rate": 4.684870449444951e-06,
      "loss": 3.3342,
      "step": 15965
    },
    {
      "epoch": 0.16245524088541666,
      "grad_norm": 10.48366641998291,
      "learning_rate": 4.684676071714292e-06,
      "loss": 3.3553,
      "step": 15970
    },
    {
      "epoch": 0.162506103515625,
      "grad_norm": 14.828400611877441,
      "learning_rate": 4.684481638089212e-06,
      "loss": 3.3099,
      "step": 15975
    },
    {
      "epoch": 0.16255696614583334,
      "grad_norm": 9.731256484985352,
      "learning_rate": 4.684287148574689e-06,
      "loss": 3.4463,
      "step": 15980
    },
    {
      "epoch": 0.16260782877604166,
      "grad_norm": 11.136251449584961,
      "learning_rate": 4.684092603175696e-06,
      "loss": 3.5473,
      "step": 15985
    },
    {
      "epoch": 0.16265869140625,
      "grad_norm": 13.161229133605957,
      "learning_rate": 4.683898001897211e-06,
      "loss": 3.3273,
      "step": 15990
    },
    {
      "epoch": 0.16270955403645834,
      "grad_norm": 9.826051712036133,
      "learning_rate": 4.683703344744213e-06,
      "loss": 3.3483,
      "step": 15995
    },
    {
      "epoch": 0.16276041666666666,
      "grad_norm": 7.0690460205078125,
      "learning_rate": 4.683508631721684e-06,
      "loss": 3.0126,
      "step": 16000
    },
    {
      "epoch": 0.162811279296875,
      "grad_norm": 13.985572814941406,
      "learning_rate": 4.683313862834603e-06,
      "loss": 3.6359,
      "step": 16005
    },
    {
      "epoch": 0.16286214192708334,
      "grad_norm": 10.578701972961426,
      "learning_rate": 4.683119038087955e-06,
      "loss": 3.2651,
      "step": 16010
    },
    {
      "epoch": 0.16291300455729166,
      "grad_norm": 11.11215591430664,
      "learning_rate": 4.682924157486724e-06,
      "loss": 3.1562,
      "step": 16015
    },
    {
      "epoch": 0.1629638671875,
      "grad_norm": 11.88314151763916,
      "learning_rate": 4.682729221035895e-06,
      "loss": 3.677,
      "step": 16020
    },
    {
      "epoch": 0.16301472981770834,
      "grad_norm": 8.311817169189453,
      "learning_rate": 4.6825342287404564e-06,
      "loss": 3.6961,
      "step": 16025
    },
    {
      "epoch": 0.16306559244791666,
      "grad_norm": 13.549260139465332,
      "learning_rate": 4.682339180605397e-06,
      "loss": 3.7595,
      "step": 16030
    },
    {
      "epoch": 0.163116455078125,
      "grad_norm": 9.020939826965332,
      "learning_rate": 4.682144076635707e-06,
      "loss": 3.2208,
      "step": 16035
    },
    {
      "epoch": 0.16316731770833334,
      "grad_norm": 16.133804321289062,
      "learning_rate": 4.681948916836378e-06,
      "loss": 3.4582,
      "step": 16040
    },
    {
      "epoch": 0.16321818033854166,
      "grad_norm": 9.808793067932129,
      "learning_rate": 4.681753701212404e-06,
      "loss": 3.3957,
      "step": 16045
    },
    {
      "epoch": 0.16326904296875,
      "grad_norm": 13.307278633117676,
      "learning_rate": 4.681558429768777e-06,
      "loss": 3.2849,
      "step": 16050
    },
    {
      "epoch": 0.16331990559895834,
      "grad_norm": 17.242029190063477,
      "learning_rate": 4.681363102510496e-06,
      "loss": 3.2799,
      "step": 16055
    },
    {
      "epoch": 0.16337076822916666,
      "grad_norm": 9.080634117126465,
      "learning_rate": 4.6811677194425566e-06,
      "loss": 2.9969,
      "step": 16060
    },
    {
      "epoch": 0.163421630859375,
      "grad_norm": 8.223971366882324,
      "learning_rate": 4.680972280569958e-06,
      "loss": 3.3398,
      "step": 16065
    },
    {
      "epoch": 0.16347249348958334,
      "grad_norm": 13.515527725219727,
      "learning_rate": 4.680776785897701e-06,
      "loss": 3.1698,
      "step": 16070
    },
    {
      "epoch": 0.16352335611979166,
      "grad_norm": 10.615397453308105,
      "learning_rate": 4.680581235430786e-06,
      "loss": 3.644,
      "step": 16075
    },
    {
      "epoch": 0.16357421875,
      "grad_norm": 13.023798942565918,
      "learning_rate": 4.680385629174218e-06,
      "loss": 3.8601,
      "step": 16080
    },
    {
      "epoch": 0.16362508138020834,
      "grad_norm": 9.510238647460938,
      "learning_rate": 4.680189967133e-06,
      "loss": 3.2448,
      "step": 16085
    },
    {
      "epoch": 0.16367594401041666,
      "grad_norm": 10.635534286499023,
      "learning_rate": 4.6799942493121374e-06,
      "loss": 3.7123,
      "step": 16090
    },
    {
      "epoch": 0.163726806640625,
      "grad_norm": 8.131168365478516,
      "learning_rate": 4.67979847571664e-06,
      "loss": 3.4378,
      "step": 16095
    },
    {
      "epoch": 0.16377766927083334,
      "grad_norm": 9.373008728027344,
      "learning_rate": 4.679602646351515e-06,
      "loss": 3.5492,
      "step": 16100
    },
    {
      "epoch": 0.16382853190104166,
      "grad_norm": 15.981472969055176,
      "learning_rate": 4.6794067612217734e-06,
      "loss": 3.7413,
      "step": 16105
    },
    {
      "epoch": 0.16387939453125,
      "grad_norm": 9.636930465698242,
      "learning_rate": 4.679210820332425e-06,
      "loss": 3.5977,
      "step": 16110
    },
    {
      "epoch": 0.16393025716145834,
      "grad_norm": 13.382928848266602,
      "learning_rate": 4.679014823688485e-06,
      "loss": 3.3873,
      "step": 16115
    },
    {
      "epoch": 0.16398111979166666,
      "grad_norm": 11.43319320678711,
      "learning_rate": 4.678818771294967e-06,
      "loss": 3.4517,
      "step": 16120
    },
    {
      "epoch": 0.164031982421875,
      "grad_norm": 10.521183013916016,
      "learning_rate": 4.678622663156888e-06,
      "loss": 3.4823,
      "step": 16125
    },
    {
      "epoch": 0.16408284505208334,
      "grad_norm": 10.852574348449707,
      "learning_rate": 4.678426499279264e-06,
      "loss": 3.4299,
      "step": 16130
    },
    {
      "epoch": 0.16413370768229166,
      "grad_norm": 13.135746955871582,
      "learning_rate": 4.6782302796671145e-06,
      "loss": 3.4938,
      "step": 16135
    },
    {
      "epoch": 0.1641845703125,
      "grad_norm": 16.104429244995117,
      "learning_rate": 4.678034004325459e-06,
      "loss": 3.4273,
      "step": 16140
    },
    {
      "epoch": 0.16423543294270834,
      "grad_norm": 14.600593566894531,
      "learning_rate": 4.677837673259321e-06,
      "loss": 3.4873,
      "step": 16145
    },
    {
      "epoch": 0.16428629557291666,
      "grad_norm": 10.380234718322754,
      "learning_rate": 4.677641286473722e-06,
      "loss": 3.4697,
      "step": 16150
    },
    {
      "epoch": 0.164337158203125,
      "grad_norm": 13.653000831604004,
      "learning_rate": 4.677444843973685e-06,
      "loss": 3.2568,
      "step": 16155
    },
    {
      "epoch": 0.16438802083333334,
      "grad_norm": 10.491547584533691,
      "learning_rate": 4.67724834576424e-06,
      "loss": 3.7773,
      "step": 16160
    },
    {
      "epoch": 0.16443888346354166,
      "grad_norm": 18.804861068725586,
      "learning_rate": 4.677051791850411e-06,
      "loss": 3.1534,
      "step": 16165
    },
    {
      "epoch": 0.16448974609375,
      "grad_norm": 10.031415939331055,
      "learning_rate": 4.676855182237229e-06,
      "loss": 3.1366,
      "step": 16170
    },
    {
      "epoch": 0.16454060872395834,
      "grad_norm": 11.652718544006348,
      "learning_rate": 4.6766585169297215e-06,
      "loss": 3.302,
      "step": 16175
    },
    {
      "epoch": 0.16459147135416666,
      "grad_norm": 12.849843978881836,
      "learning_rate": 4.6764617959329226e-06,
      "loss": 3.7761,
      "step": 16180
    },
    {
      "epoch": 0.164642333984375,
      "grad_norm": 12.61086368560791,
      "learning_rate": 4.676265019251865e-06,
      "loss": 3.3563,
      "step": 16185
    },
    {
      "epoch": 0.16469319661458334,
      "grad_norm": 10.435957908630371,
      "learning_rate": 4.676068186891582e-06,
      "loss": 3.4891,
      "step": 16190
    },
    {
      "epoch": 0.16474405924479166,
      "grad_norm": 13.85044002532959,
      "learning_rate": 4.675871298857111e-06,
      "loss": 3.0696,
      "step": 16195
    },
    {
      "epoch": 0.164794921875,
      "grad_norm": 9.022653579711914,
      "learning_rate": 4.675674355153488e-06,
      "loss": 3.4813,
      "step": 16200
    },
    {
      "epoch": 0.16484578450520834,
      "grad_norm": 11.870932579040527,
      "learning_rate": 4.675477355785752e-06,
      "loss": 3.7666,
      "step": 16205
    },
    {
      "epoch": 0.16489664713541666,
      "grad_norm": 12.24802303314209,
      "learning_rate": 4.675280300758944e-06,
      "loss": 3.3968,
      "step": 16210
    },
    {
      "epoch": 0.164947509765625,
      "grad_norm": 12.894149780273438,
      "learning_rate": 4.6750831900781055e-06,
      "loss": 3.4708,
      "step": 16215
    },
    {
      "epoch": 0.16499837239583334,
      "grad_norm": 9.178655624389648,
      "learning_rate": 4.674886023748279e-06,
      "loss": 3.3961,
      "step": 16220
    },
    {
      "epoch": 0.16504923502604166,
      "grad_norm": 15.221061706542969,
      "learning_rate": 4.674688801774508e-06,
      "loss": 3.5445,
      "step": 16225
    },
    {
      "epoch": 0.16510009765625,
      "grad_norm": 12.730554580688477,
      "learning_rate": 4.67449152416184e-06,
      "loss": 3.3576,
      "step": 16230
    },
    {
      "epoch": 0.16515096028645834,
      "grad_norm": 10.138298988342285,
      "learning_rate": 4.674294190915321e-06,
      "loss": 3.3352,
      "step": 16235
    },
    {
      "epoch": 0.16520182291666666,
      "grad_norm": 9.869894981384277,
      "learning_rate": 4.674096802040003e-06,
      "loss": 3.3375,
      "step": 16240
    },
    {
      "epoch": 0.165252685546875,
      "grad_norm": 9.15701961517334,
      "learning_rate": 4.673899357540932e-06,
      "loss": 3.2079,
      "step": 16245
    },
    {
      "epoch": 0.16530354817708334,
      "grad_norm": 13.204168319702148,
      "learning_rate": 4.673701857423161e-06,
      "loss": 3.2676,
      "step": 16250
    },
    {
      "epoch": 0.16535441080729166,
      "grad_norm": 15.898265838623047,
      "learning_rate": 4.6735043016917435e-06,
      "loss": 3.3132,
      "step": 16255
    },
    {
      "epoch": 0.1654052734375,
      "grad_norm": 10.086653709411621,
      "learning_rate": 4.673306690351733e-06,
      "loss": 3.2754,
      "step": 16260
    },
    {
      "epoch": 0.16545613606770834,
      "grad_norm": 15.631150245666504,
      "learning_rate": 4.6731090234081865e-06,
      "loss": 3.6231,
      "step": 16265
    },
    {
      "epoch": 0.16550699869791666,
      "grad_norm": 10.62215518951416,
      "learning_rate": 4.672911300866161e-06,
      "loss": 3.2062,
      "step": 16270
    },
    {
      "epoch": 0.165557861328125,
      "grad_norm": 8.654640197753906,
      "learning_rate": 4.672713522730715e-06,
      "loss": 3.2674,
      "step": 16275
    },
    {
      "epoch": 0.16560872395833334,
      "grad_norm": 11.249765396118164,
      "learning_rate": 4.672515689006908e-06,
      "loss": 3.0822,
      "step": 16280
    },
    {
      "epoch": 0.16565958658854166,
      "grad_norm": 14.134060859680176,
      "learning_rate": 4.6723177996998025e-06,
      "loss": 3.58,
      "step": 16285
    },
    {
      "epoch": 0.16571044921875,
      "grad_norm": 16.133581161499023,
      "learning_rate": 4.672119854814461e-06,
      "loss": 3.6814,
      "step": 16290
    },
    {
      "epoch": 0.16576131184895834,
      "grad_norm": 13.389992713928223,
      "learning_rate": 4.671921854355947e-06,
      "loss": 3.4922,
      "step": 16295
    },
    {
      "epoch": 0.16581217447916666,
      "grad_norm": 13.663773536682129,
      "learning_rate": 4.671723798329328e-06,
      "loss": 3.1055,
      "step": 16300
    },
    {
      "epoch": 0.165863037109375,
      "grad_norm": 13.669440269470215,
      "learning_rate": 4.671525686739669e-06,
      "loss": 3.1514,
      "step": 16305
    },
    {
      "epoch": 0.16591389973958334,
      "grad_norm": 12.217931747436523,
      "learning_rate": 4.671327519592042e-06,
      "loss": 3.6507,
      "step": 16310
    },
    {
      "epoch": 0.16596476236979166,
      "grad_norm": 15.413629531860352,
      "learning_rate": 4.6711292968915145e-06,
      "loss": 3.4689,
      "step": 16315
    },
    {
      "epoch": 0.166015625,
      "grad_norm": 11.545214653015137,
      "learning_rate": 4.670931018643158e-06,
      "loss": 3.2486,
      "step": 16320
    },
    {
      "epoch": 0.16606648763020834,
      "grad_norm": 12.391217231750488,
      "learning_rate": 4.670732684852046e-06,
      "loss": 3.3009,
      "step": 16325
    },
    {
      "epoch": 0.16611735026041666,
      "grad_norm": 10.62708854675293,
      "learning_rate": 4.670534295523253e-06,
      "loss": 3.3732,
      "step": 16330
    },
    {
      "epoch": 0.166168212890625,
      "grad_norm": 8.105783462524414,
      "learning_rate": 4.670335850661855e-06,
      "loss": 3.3016,
      "step": 16335
    },
    {
      "epoch": 0.16621907552083334,
      "grad_norm": 7.761030197143555,
      "learning_rate": 4.670137350272927e-06,
      "loss": 3.2436,
      "step": 16340
    },
    {
      "epoch": 0.16626993815104166,
      "grad_norm": 12.776176452636719,
      "learning_rate": 4.669938794361552e-06,
      "loss": 3.3074,
      "step": 16345
    },
    {
      "epoch": 0.16632080078125,
      "grad_norm": 7.695553302764893,
      "learning_rate": 4.669740182932805e-06,
      "loss": 3.0618,
      "step": 16350
    },
    {
      "epoch": 0.16637166341145834,
      "grad_norm": 14.618805885314941,
      "learning_rate": 4.66954151599177e-06,
      "loss": 3.4561,
      "step": 16355
    },
    {
      "epoch": 0.16642252604166666,
      "grad_norm": 14.894763946533203,
      "learning_rate": 4.66934279354353e-06,
      "loss": 3.492,
      "step": 16360
    },
    {
      "epoch": 0.166473388671875,
      "grad_norm": 9.516297340393066,
      "learning_rate": 4.669144015593169e-06,
      "loss": 3.3012,
      "step": 16365
    },
    {
      "epoch": 0.16652425130208334,
      "grad_norm": 15.990934371948242,
      "learning_rate": 4.668945182145773e-06,
      "loss": 3.4032,
      "step": 16370
    },
    {
      "epoch": 0.16657511393229166,
      "grad_norm": 11.829225540161133,
      "learning_rate": 4.668746293206428e-06,
      "loss": 3.57,
      "step": 16375
    },
    {
      "epoch": 0.1666259765625,
      "grad_norm": 9.242964744567871,
      "learning_rate": 4.668547348780222e-06,
      "loss": 3.43,
      "step": 16380
    },
    {
      "epoch": 0.16667683919270834,
      "grad_norm": 15.611289024353027,
      "learning_rate": 4.668348348872248e-06,
      "loss": 3.2432,
      "step": 16385
    },
    {
      "epoch": 0.16672770182291666,
      "grad_norm": 12.005697250366211,
      "learning_rate": 4.668149293487595e-06,
      "loss": 2.9898,
      "step": 16390
    },
    {
      "epoch": 0.166778564453125,
      "grad_norm": 14.01111888885498,
      "learning_rate": 4.6679501826313554e-06,
      "loss": 3.5077,
      "step": 16395
    },
    {
      "epoch": 0.16682942708333334,
      "grad_norm": 17.95404052734375,
      "learning_rate": 4.667751016308624e-06,
      "loss": 3.3168,
      "step": 16400
    },
    {
      "epoch": 0.16688028971354166,
      "grad_norm": 14.67130184173584,
      "learning_rate": 4.6675517945244975e-06,
      "loss": 3.0814,
      "step": 16405
    },
    {
      "epoch": 0.16693115234375,
      "grad_norm": 9.70960521697998,
      "learning_rate": 4.667352517284072e-06,
      "loss": 3.3854,
      "step": 16410
    },
    {
      "epoch": 0.16698201497395834,
      "grad_norm": 10.68616008758545,
      "learning_rate": 4.667153184592446e-06,
      "loss": 3.375,
      "step": 16415
    },
    {
      "epoch": 0.16703287760416666,
      "grad_norm": 12.836880683898926,
      "learning_rate": 4.6669537964547195e-06,
      "loss": 3.2576,
      "step": 16420
    },
    {
      "epoch": 0.167083740234375,
      "grad_norm": 10.322628021240234,
      "learning_rate": 4.666754352875994e-06,
      "loss": 3.5553,
      "step": 16425
    },
    {
      "epoch": 0.16713460286458334,
      "grad_norm": 13.36998176574707,
      "learning_rate": 4.6665548538613715e-06,
      "loss": 3.5801,
      "step": 16430
    },
    {
      "epoch": 0.16718546549479166,
      "grad_norm": 11.6786527633667,
      "learning_rate": 4.666355299415956e-06,
      "loss": 3.1604,
      "step": 16435
    },
    {
      "epoch": 0.167236328125,
      "grad_norm": 15.699369430541992,
      "learning_rate": 4.666155689544855e-06,
      "loss": 3.2281,
      "step": 16440
    },
    {
      "epoch": 0.16728719075520834,
      "grad_norm": 23.027801513671875,
      "learning_rate": 4.6659560242531735e-06,
      "loss": 3.0434,
      "step": 16445
    },
    {
      "epoch": 0.16733805338541666,
      "grad_norm": 10.911322593688965,
      "learning_rate": 4.665756303546021e-06,
      "loss": 3.5173,
      "step": 16450
    },
    {
      "epoch": 0.167388916015625,
      "grad_norm": 9.32667064666748,
      "learning_rate": 4.665556527428506e-06,
      "loss": 3.5175,
      "step": 16455
    },
    {
      "epoch": 0.16743977864583334,
      "grad_norm": 13.066845893859863,
      "learning_rate": 4.66535669590574e-06,
      "loss": 3.7572,
      "step": 16460
    },
    {
      "epoch": 0.16749064127604166,
      "grad_norm": 14.599390029907227,
      "learning_rate": 4.6651568089828384e-06,
      "loss": 3.1729,
      "step": 16465
    },
    {
      "epoch": 0.16754150390625,
      "grad_norm": 10.783397674560547,
      "learning_rate": 4.664956866664912e-06,
      "loss": 3.1904,
      "step": 16470
    },
    {
      "epoch": 0.16759236653645834,
      "grad_norm": 15.606278419494629,
      "learning_rate": 4.664756868957076e-06,
      "loss": 3.3966,
      "step": 16475
    },
    {
      "epoch": 0.16764322916666666,
      "grad_norm": 11.634224891662598,
      "learning_rate": 4.6645568158644496e-06,
      "loss": 3.4924,
      "step": 16480
    },
    {
      "epoch": 0.167694091796875,
      "grad_norm": 8.687528610229492,
      "learning_rate": 4.66435670739215e-06,
      "loss": 3.4183,
      "step": 16485
    },
    {
      "epoch": 0.16774495442708334,
      "grad_norm": 13.002685546875,
      "learning_rate": 4.6641565435452975e-06,
      "loss": 3.3721,
      "step": 16490
    },
    {
      "epoch": 0.16779581705729166,
      "grad_norm": 13.268308639526367,
      "learning_rate": 4.663956324329012e-06,
      "loss": 3.7302,
      "step": 16495
    },
    {
      "epoch": 0.1678466796875,
      "grad_norm": 9.55381965637207,
      "learning_rate": 4.663756049748418e-06,
      "loss": 3.2204,
      "step": 16500
    },
    {
      "epoch": 0.16789754231770834,
      "grad_norm": 13.7435302734375,
      "learning_rate": 4.6635557198086375e-06,
      "loss": 3.6933,
      "step": 16505
    },
    {
      "epoch": 0.16794840494791666,
      "grad_norm": 13.537063598632812,
      "learning_rate": 4.663355334514796e-06,
      "loss": 3.0507,
      "step": 16510
    },
    {
      "epoch": 0.167999267578125,
      "grad_norm": 13.002664566040039,
      "learning_rate": 4.663154893872023e-06,
      "loss": 3.1599,
      "step": 16515
    },
    {
      "epoch": 0.16805013020833334,
      "grad_norm": 8.084352493286133,
      "learning_rate": 4.662954397885443e-06,
      "loss": 3.5544,
      "step": 16520
    },
    {
      "epoch": 0.16810099283854166,
      "grad_norm": 13.704971313476562,
      "learning_rate": 4.662753846560189e-06,
      "loss": 3.5446,
      "step": 16525
    },
    {
      "epoch": 0.16815185546875,
      "grad_norm": 13.047155380249023,
      "learning_rate": 4.662553239901389e-06,
      "loss": 3.6407,
      "step": 16530
    },
    {
      "epoch": 0.16820271809895834,
      "grad_norm": 10.460413932800293,
      "learning_rate": 4.662352577914178e-06,
      "loss": 3.5399,
      "step": 16535
    },
    {
      "epoch": 0.16825358072916666,
      "grad_norm": 12.304898262023926,
      "learning_rate": 4.6621518606036875e-06,
      "loss": 3.3801,
      "step": 16540
    },
    {
      "epoch": 0.168304443359375,
      "grad_norm": 10.110926628112793,
      "learning_rate": 4.661951087975055e-06,
      "loss": 3.2155,
      "step": 16545
    },
    {
      "epoch": 0.16835530598958334,
      "grad_norm": 7.640111446380615,
      "learning_rate": 4.661750260033417e-06,
      "loss": 3.1164,
      "step": 16550
    },
    {
      "epoch": 0.16840616861979166,
      "grad_norm": 8.465399742126465,
      "learning_rate": 4.66154937678391e-06,
      "loss": 3.7351,
      "step": 16555
    },
    {
      "epoch": 0.16845703125,
      "grad_norm": 11.797218322753906,
      "learning_rate": 4.661348438231675e-06,
      "loss": 3.6523,
      "step": 16560
    },
    {
      "epoch": 0.16850789388020834,
      "grad_norm": 9.346522331237793,
      "learning_rate": 4.6611474443818525e-06,
      "loss": 3.34,
      "step": 16565
    },
    {
      "epoch": 0.16855875651041666,
      "grad_norm": 14.899089813232422,
      "learning_rate": 4.660946395239584e-06,
      "loss": 3.5477,
      "step": 16570
    },
    {
      "epoch": 0.168609619140625,
      "grad_norm": 11.063288688659668,
      "learning_rate": 4.660745290810015e-06,
      "loss": 3.333,
      "step": 16575
    },
    {
      "epoch": 0.16866048177083334,
      "grad_norm": 9.561201095581055,
      "learning_rate": 4.66054413109829e-06,
      "loss": 3.8984,
      "step": 16580
    },
    {
      "epoch": 0.16871134440104166,
      "grad_norm": 13.455229759216309,
      "learning_rate": 4.6603429161095556e-06,
      "loss": 3.2126,
      "step": 16585
    },
    {
      "epoch": 0.16876220703125,
      "grad_norm": 10.709653854370117,
      "learning_rate": 4.660141645848959e-06,
      "loss": 3.2361,
      "step": 16590
    },
    {
      "epoch": 0.16881306966145834,
      "grad_norm": 12.909401893615723,
      "learning_rate": 4.659940320321651e-06,
      "loss": 3.3619,
      "step": 16595
    },
    {
      "epoch": 0.16886393229166666,
      "grad_norm": 15.584907531738281,
      "learning_rate": 4.6597389395327816e-06,
      "loss": 3.1329,
      "step": 16600
    },
    {
      "epoch": 0.168914794921875,
      "grad_norm": 14.770988464355469,
      "learning_rate": 4.659537503487503e-06,
      "loss": 3.9151,
      "step": 16605
    },
    {
      "epoch": 0.16896565755208334,
      "grad_norm": 7.884305953979492,
      "learning_rate": 4.6593360121909706e-06,
      "loss": 3.2376,
      "step": 16610
    },
    {
      "epoch": 0.16901652018229166,
      "grad_norm": 14.378878593444824,
      "learning_rate": 4.659134465648338e-06,
      "loss": 3.4527,
      "step": 16615
    },
    {
      "epoch": 0.1690673828125,
      "grad_norm": 9.519623756408691,
      "learning_rate": 4.65893286386476e-06,
      "loss": 3.445,
      "step": 16620
    },
    {
      "epoch": 0.16911824544270834,
      "grad_norm": 9.221073150634766,
      "learning_rate": 4.658731206845398e-06,
      "loss": 3.3477,
      "step": 16625
    },
    {
      "epoch": 0.16916910807291666,
      "grad_norm": 10.170306205749512,
      "learning_rate": 4.658529494595408e-06,
      "loss": 3.3417,
      "step": 16630
    },
    {
      "epoch": 0.169219970703125,
      "grad_norm": 15.229034423828125,
      "learning_rate": 4.6583277271199545e-06,
      "loss": 3.2686,
      "step": 16635
    },
    {
      "epoch": 0.16927083333333334,
      "grad_norm": 12.785039901733398,
      "learning_rate": 4.658125904424197e-06,
      "loss": 3.4329,
      "step": 16640
    },
    {
      "epoch": 0.16932169596354166,
      "grad_norm": 9.296720504760742,
      "learning_rate": 4.6579240265133e-06,
      "loss": 3.3711,
      "step": 16645
    },
    {
      "epoch": 0.16937255859375,
      "grad_norm": 10.980588912963867,
      "learning_rate": 4.657722093392428e-06,
      "loss": 3.3667,
      "step": 16650
    },
    {
      "epoch": 0.16942342122395834,
      "grad_norm": 11.73270034790039,
      "learning_rate": 4.657520105066747e-06,
      "loss": 3.4216,
      "step": 16655
    },
    {
      "epoch": 0.16947428385416666,
      "grad_norm": 18.23244857788086,
      "learning_rate": 4.6573180615414265e-06,
      "loss": 3.6996,
      "step": 16660
    },
    {
      "epoch": 0.169525146484375,
      "grad_norm": 13.225893020629883,
      "learning_rate": 4.657115962821635e-06,
      "loss": 3.5311,
      "step": 16665
    },
    {
      "epoch": 0.16957600911458334,
      "grad_norm": 12.947046279907227,
      "learning_rate": 4.656913808912542e-06,
      "loss": 3.2338,
      "step": 16670
    },
    {
      "epoch": 0.16962687174479166,
      "grad_norm": 16.09568977355957,
      "learning_rate": 4.65671159981932e-06,
      "loss": 3.3158,
      "step": 16675
    },
    {
      "epoch": 0.169677734375,
      "grad_norm": 9.442344665527344,
      "learning_rate": 4.656509335547144e-06,
      "loss": 3.5821,
      "step": 16680
    },
    {
      "epoch": 0.16972859700520834,
      "grad_norm": 8.947610855102539,
      "learning_rate": 4.656307016101187e-06,
      "loss": 3.2935,
      "step": 16685
    },
    {
      "epoch": 0.16977945963541666,
      "grad_norm": 15.548479080200195,
      "learning_rate": 4.656104641486628e-06,
      "loss": 3.6593,
      "step": 16690
    },
    {
      "epoch": 0.169830322265625,
      "grad_norm": 9.9256591796875,
      "learning_rate": 4.655902211708641e-06,
      "loss": 3.4071,
      "step": 16695
    },
    {
      "epoch": 0.16988118489583334,
      "grad_norm": 11.299330711364746,
      "learning_rate": 4.655699726772407e-06,
      "loss": 3.4039,
      "step": 16700
    },
    {
      "epoch": 0.16993204752604166,
      "grad_norm": 10.457588195800781,
      "learning_rate": 4.655497186683107e-06,
      "loss": 3.4151,
      "step": 16705
    },
    {
      "epoch": 0.16998291015625,
      "grad_norm": 9.101759910583496,
      "learning_rate": 4.655294591445921e-06,
      "loss": 3.0531,
      "step": 16710
    },
    {
      "epoch": 0.17003377278645834,
      "grad_norm": 9.755556106567383,
      "learning_rate": 4.6550919410660355e-06,
      "loss": 3.7921,
      "step": 16715
    },
    {
      "epoch": 0.17008463541666666,
      "grad_norm": 6.866489410400391,
      "learning_rate": 4.654889235548633e-06,
      "loss": 3.3934,
      "step": 16720
    },
    {
      "epoch": 0.170135498046875,
      "grad_norm": 11.931754112243652,
      "learning_rate": 4.6546864748989e-06,
      "loss": 3.2464,
      "step": 16725
    },
    {
      "epoch": 0.17018636067708334,
      "grad_norm": 12.620525360107422,
      "learning_rate": 4.654483659122025e-06,
      "loss": 3.1906,
      "step": 16730
    },
    {
      "epoch": 0.17023722330729166,
      "grad_norm": 9.603962898254395,
      "learning_rate": 4.654280788223195e-06,
      "loss": 3.1315,
      "step": 16735
    },
    {
      "epoch": 0.1702880859375,
      "grad_norm": 10.157326698303223,
      "learning_rate": 4.654077862207601e-06,
      "loss": 3.4375,
      "step": 16740
    },
    {
      "epoch": 0.17033894856770834,
      "grad_norm": 6.509634971618652,
      "learning_rate": 4.653874881080437e-06,
      "loss": 3.2071,
      "step": 16745
    },
    {
      "epoch": 0.17038981119791666,
      "grad_norm": 11.346307754516602,
      "learning_rate": 4.653671844846895e-06,
      "loss": 3.4308,
      "step": 16750
    },
    {
      "epoch": 0.170440673828125,
      "grad_norm": 10.316970825195312,
      "learning_rate": 4.653468753512168e-06,
      "loss": 3.5133,
      "step": 16755
    },
    {
      "epoch": 0.17049153645833334,
      "grad_norm": 12.79603099822998,
      "learning_rate": 4.653265607081454e-06,
      "loss": 3.2712,
      "step": 16760
    },
    {
      "epoch": 0.17054239908854166,
      "grad_norm": 12.55677604675293,
      "learning_rate": 4.653062405559951e-06,
      "loss": 3.2636,
      "step": 16765
    },
    {
      "epoch": 0.17059326171875,
      "grad_norm": 15.563427925109863,
      "learning_rate": 4.652859148952855e-06,
      "loss": 3.2278,
      "step": 16770
    },
    {
      "epoch": 0.17064412434895834,
      "grad_norm": 13.07026481628418,
      "learning_rate": 4.652655837265369e-06,
      "loss": 3.3884,
      "step": 16775
    },
    {
      "epoch": 0.17069498697916666,
      "grad_norm": 13.581730842590332,
      "learning_rate": 4.6524524705026925e-06,
      "loss": 3.6175,
      "step": 16780
    },
    {
      "epoch": 0.170745849609375,
      "grad_norm": 8.78980827331543,
      "learning_rate": 4.65224904867003e-06,
      "loss": 3.4371,
      "step": 16785
    },
    {
      "epoch": 0.17079671223958334,
      "grad_norm": 11.52972412109375,
      "learning_rate": 4.652045571772586e-06,
      "loss": 3.2436,
      "step": 16790
    },
    {
      "epoch": 0.17084757486979166,
      "grad_norm": 17.321304321289062,
      "learning_rate": 4.651842039815566e-06,
      "loss": 3.393,
      "step": 16795
    },
    {
      "epoch": 0.1708984375,
      "grad_norm": 12.183976173400879,
      "learning_rate": 4.651638452804178e-06,
      "loss": 3.3443,
      "step": 16800
    },
    {
      "epoch": 0.17094930013020834,
      "grad_norm": 11.435518264770508,
      "learning_rate": 4.6514348107436305e-06,
      "loss": 3.4001,
      "step": 16805
    },
    {
      "epoch": 0.17100016276041666,
      "grad_norm": 15.479721069335938,
      "learning_rate": 4.651231113639132e-06,
      "loss": 3.0751,
      "step": 16810
    },
    {
      "epoch": 0.171051025390625,
      "grad_norm": 12.146183013916016,
      "learning_rate": 4.651027361495896e-06,
      "loss": 3.238,
      "step": 16815
    },
    {
      "epoch": 0.17110188802083334,
      "grad_norm": 12.965581893920898,
      "learning_rate": 4.650823554319135e-06,
      "loss": 3.8735,
      "step": 16820
    },
    {
      "epoch": 0.17115275065104166,
      "grad_norm": 9.760518074035645,
      "learning_rate": 4.650619692114063e-06,
      "loss": 3.2039,
      "step": 16825
    },
    {
      "epoch": 0.17120361328125,
      "grad_norm": 12.903510093688965,
      "learning_rate": 4.650415774885896e-06,
      "loss": 3.5834,
      "step": 16830
    },
    {
      "epoch": 0.17125447591145834,
      "grad_norm": 13.00151538848877,
      "learning_rate": 4.650211802639851e-06,
      "loss": 3.243,
      "step": 16835
    },
    {
      "epoch": 0.17130533854166666,
      "grad_norm": 15.11430549621582,
      "learning_rate": 4.6500077753811465e-06,
      "loss": 3.4178,
      "step": 16840
    },
    {
      "epoch": 0.171356201171875,
      "grad_norm": 10.01813793182373,
      "learning_rate": 4.649803693115003e-06,
      "loss": 3.2601,
      "step": 16845
    },
    {
      "epoch": 0.17140706380208334,
      "grad_norm": 11.622364044189453,
      "learning_rate": 4.649599555846641e-06,
      "loss": 3.6725,
      "step": 16850
    },
    {
      "epoch": 0.17145792643229166,
      "grad_norm": 15.42918872833252,
      "learning_rate": 4.649395363581285e-06,
      "loss": 3.3048,
      "step": 16855
    },
    {
      "epoch": 0.1715087890625,
      "grad_norm": 11.032341957092285,
      "learning_rate": 4.649191116324158e-06,
      "loss": 3.4853,
      "step": 16860
    },
    {
      "epoch": 0.17155965169270834,
      "grad_norm": 12.54845142364502,
      "learning_rate": 4.648986814080485e-06,
      "loss": 3.2918,
      "step": 16865
    },
    {
      "epoch": 0.17161051432291666,
      "grad_norm": 8.605419158935547,
      "learning_rate": 4.648782456855493e-06,
      "loss": 3.4214,
      "step": 16870
    },
    {
      "epoch": 0.171661376953125,
      "grad_norm": 9.82323169708252,
      "learning_rate": 4.648578044654412e-06,
      "loss": 3.3439,
      "step": 16875
    },
    {
      "epoch": 0.17171223958333334,
      "grad_norm": 15.435731887817383,
      "learning_rate": 4.648373577482471e-06,
      "loss": 3.2196,
      "step": 16880
    },
    {
      "epoch": 0.17176310221354166,
      "grad_norm": 9.784239768981934,
      "learning_rate": 4.6481690553449015e-06,
      "loss": 4.0164,
      "step": 16885
    },
    {
      "epoch": 0.17181396484375,
      "grad_norm": 11.91480541229248,
      "learning_rate": 4.647964478246936e-06,
      "loss": 3.2925,
      "step": 16890
    },
    {
      "epoch": 0.17186482747395834,
      "grad_norm": 7.702728271484375,
      "learning_rate": 4.647759846193808e-06,
      "loss": 3.2802,
      "step": 16895
    },
    {
      "epoch": 0.17191569010416666,
      "grad_norm": 7.844374656677246,
      "learning_rate": 4.647555159190753e-06,
      "loss": 3.3919,
      "step": 16900
    },
    {
      "epoch": 0.171966552734375,
      "grad_norm": 9.673126220703125,
      "learning_rate": 4.647350417243009e-06,
      "loss": 3.7028,
      "step": 16905
    },
    {
      "epoch": 0.17201741536458334,
      "grad_norm": 13.436027526855469,
      "learning_rate": 4.647145620355813e-06,
      "loss": 3.0947,
      "step": 16910
    },
    {
      "epoch": 0.17206827799479166,
      "grad_norm": 9.890853881835938,
      "learning_rate": 4.646940768534406e-06,
      "loss": 3.3502,
      "step": 16915
    },
    {
      "epoch": 0.172119140625,
      "grad_norm": 12.191718101501465,
      "learning_rate": 4.6467358617840275e-06,
      "loss": 3.1166,
      "step": 16920
    },
    {
      "epoch": 0.17217000325520834,
      "grad_norm": 10.807548522949219,
      "learning_rate": 4.646530900109921e-06,
      "loss": 3.3743,
      "step": 16925
    },
    {
      "epoch": 0.17222086588541666,
      "grad_norm": 15.573385238647461,
      "learning_rate": 4.646325883517331e-06,
      "loss": 3.4527,
      "step": 16930
    },
    {
      "epoch": 0.172271728515625,
      "grad_norm": 12.284563064575195,
      "learning_rate": 4.646120812011501e-06,
      "loss": 3.1113,
      "step": 16935
    },
    {
      "epoch": 0.17232259114583334,
      "grad_norm": 15.29260540008545,
      "learning_rate": 4.645915685597679e-06,
      "loss": 3.4474,
      "step": 16940
    },
    {
      "epoch": 0.17237345377604166,
      "grad_norm": 12.571720123291016,
      "learning_rate": 4.645710504281113e-06,
      "loss": 3.1863,
      "step": 16945
    },
    {
      "epoch": 0.17242431640625,
      "grad_norm": 14.381819725036621,
      "learning_rate": 4.645505268067052e-06,
      "loss": 3.3417,
      "step": 16950
    },
    {
      "epoch": 0.17247517903645834,
      "grad_norm": 9.561775207519531,
      "learning_rate": 4.645299976960747e-06,
      "loss": 3.1955,
      "step": 16955
    },
    {
      "epoch": 0.17252604166666666,
      "grad_norm": 12.148945808410645,
      "learning_rate": 4.645094630967451e-06,
      "loss": 3.7219,
      "step": 16960
    },
    {
      "epoch": 0.172576904296875,
      "grad_norm": 10.793240547180176,
      "learning_rate": 4.644889230092418e-06,
      "loss": 3.5109,
      "step": 16965
    },
    {
      "epoch": 0.17262776692708334,
      "grad_norm": 13.821535110473633,
      "learning_rate": 4.644683774340902e-06,
      "loss": 3.3895,
      "step": 16970
    },
    {
      "epoch": 0.17267862955729166,
      "grad_norm": 10.016497611999512,
      "learning_rate": 4.64447826371816e-06,
      "loss": 3.317,
      "step": 16975
    },
    {
      "epoch": 0.1727294921875,
      "grad_norm": 12.0796480178833,
      "learning_rate": 4.64427269822945e-06,
      "loss": 3.1043,
      "step": 16980
    },
    {
      "epoch": 0.17278035481770834,
      "grad_norm": 11.921280860900879,
      "learning_rate": 4.644067077880031e-06,
      "loss": 3.4567,
      "step": 16985
    },
    {
      "epoch": 0.17283121744791666,
      "grad_norm": 10.839923858642578,
      "learning_rate": 4.643861402675164e-06,
      "loss": 3.4736,
      "step": 16990
    },
    {
      "epoch": 0.172882080078125,
      "grad_norm": 13.740301132202148,
      "learning_rate": 4.643655672620111e-06,
      "loss": 3.6096,
      "step": 16995
    },
    {
      "epoch": 0.17293294270833334,
      "grad_norm": 14.093925476074219,
      "learning_rate": 4.643449887720136e-06,
      "loss": 3.2615,
      "step": 17000
    },
    {
      "epoch": 0.17298380533854166,
      "grad_norm": 10.022769927978516,
      "learning_rate": 4.643244047980503e-06,
      "loss": 3.0516,
      "step": 17005
    },
    {
      "epoch": 0.17303466796875,
      "grad_norm": 9.807087898254395,
      "learning_rate": 4.64303815340648e-06,
      "loss": 3.6837,
      "step": 17010
    },
    {
      "epoch": 0.17308553059895834,
      "grad_norm": 10.774398803710938,
      "learning_rate": 4.642832204003333e-06,
      "loss": 3.046,
      "step": 17015
    },
    {
      "epoch": 0.17313639322916666,
      "grad_norm": 8.116972923278809,
      "learning_rate": 4.642626199776333e-06,
      "loss": 3.071,
      "step": 17020
    },
    {
      "epoch": 0.173187255859375,
      "grad_norm": 13.305663108825684,
      "learning_rate": 4.642420140730749e-06,
      "loss": 3.8665,
      "step": 17025
    },
    {
      "epoch": 0.17323811848958334,
      "grad_norm": 16.738006591796875,
      "learning_rate": 4.642214026871853e-06,
      "loss": 3.6563,
      "step": 17030
    },
    {
      "epoch": 0.17328898111979166,
      "grad_norm": 10.710490226745605,
      "learning_rate": 4.642007858204919e-06,
      "loss": 3.3579,
      "step": 17035
    },
    {
      "epoch": 0.17333984375,
      "grad_norm": 17.46071434020996,
      "learning_rate": 4.641801634735222e-06,
      "loss": 3.2495,
      "step": 17040
    },
    {
      "epoch": 0.17339070638020834,
      "grad_norm": 8.741815567016602,
      "learning_rate": 4.6415953564680385e-06,
      "loss": 3.0387,
      "step": 17045
    },
    {
      "epoch": 0.17344156901041666,
      "grad_norm": 14.374468803405762,
      "learning_rate": 4.641389023408644e-06,
      "loss": 3.3034,
      "step": 17050
    },
    {
      "epoch": 0.173492431640625,
      "grad_norm": 10.53064250946045,
      "learning_rate": 4.64118263556232e-06,
      "loss": 3.3958,
      "step": 17055
    },
    {
      "epoch": 0.17354329427083334,
      "grad_norm": 16.45563316345215,
      "learning_rate": 4.640976192934345e-06,
      "loss": 3.6537,
      "step": 17060
    },
    {
      "epoch": 0.17359415690104166,
      "grad_norm": 13.669546127319336,
      "learning_rate": 4.6407696955300025e-06,
      "loss": 3.136,
      "step": 17065
    },
    {
      "epoch": 0.17364501953125,
      "grad_norm": 15.326542854309082,
      "learning_rate": 4.640563143354574e-06,
      "loss": 3.4866,
      "step": 17070
    },
    {
      "epoch": 0.17369588216145834,
      "grad_norm": 7.852635860443115,
      "learning_rate": 4.640356536413345e-06,
      "loss": 3.4082,
      "step": 17075
    },
    {
      "epoch": 0.17374674479166666,
      "grad_norm": 15.086888313293457,
      "learning_rate": 4.640149874711601e-06,
      "loss": 3.3045,
      "step": 17080
    },
    {
      "epoch": 0.173797607421875,
      "grad_norm": 13.305411338806152,
      "learning_rate": 4.63994315825463e-06,
      "loss": 2.9298,
      "step": 17085
    },
    {
      "epoch": 0.17384847005208334,
      "grad_norm": 6.963002681732178,
      "learning_rate": 4.639736387047722e-06,
      "loss": 3.5608,
      "step": 17090
    },
    {
      "epoch": 0.17389933268229166,
      "grad_norm": 13.062210083007812,
      "learning_rate": 4.639529561096164e-06,
      "loss": 3.1996,
      "step": 17095
    },
    {
      "epoch": 0.1739501953125,
      "grad_norm": 9.515623092651367,
      "learning_rate": 4.639322680405249e-06,
      "loss": 3.3422,
      "step": 17100
    },
    {
      "epoch": 0.17400105794270834,
      "grad_norm": 11.610623359680176,
      "learning_rate": 4.639115744980272e-06,
      "loss": 3.3988,
      "step": 17105
    },
    {
      "epoch": 0.17405192057291666,
      "grad_norm": 11.479599952697754,
      "learning_rate": 4.6389087548265245e-06,
      "loss": 3.2056,
      "step": 17110
    },
    {
      "epoch": 0.174102783203125,
      "grad_norm": 11.929533004760742,
      "learning_rate": 4.638701709949303e-06,
      "loss": 3.0456,
      "step": 17115
    },
    {
      "epoch": 0.17415364583333334,
      "grad_norm": 12.549202919006348,
      "learning_rate": 4.638494610353907e-06,
      "loss": 3.3503,
      "step": 17120
    },
    {
      "epoch": 0.17420450846354166,
      "grad_norm": 14.711870193481445,
      "learning_rate": 4.638287456045632e-06,
      "loss": 3.4214,
      "step": 17125
    },
    {
      "epoch": 0.17425537109375,
      "grad_norm": 14.64809513092041,
      "learning_rate": 4.638080247029779e-06,
      "loss": 3.4133,
      "step": 17130
    },
    {
      "epoch": 0.17430623372395834,
      "grad_norm": 16.768775939941406,
      "learning_rate": 4.63787298331165e-06,
      "loss": 3.3163,
      "step": 17135
    },
    {
      "epoch": 0.17435709635416666,
      "grad_norm": 13.071782112121582,
      "learning_rate": 4.637665664896547e-06,
      "loss": 3.8572,
      "step": 17140
    },
    {
      "epoch": 0.174407958984375,
      "grad_norm": 11.809013366699219,
      "learning_rate": 4.637458291789776e-06,
      "loss": 3.4033,
      "step": 17145
    },
    {
      "epoch": 0.17445882161458334,
      "grad_norm": 11.942599296569824,
      "learning_rate": 4.63725086399664e-06,
      "loss": 3.3937,
      "step": 17150
    },
    {
      "epoch": 0.17450968424479166,
      "grad_norm": 11.510283470153809,
      "learning_rate": 4.637043381522447e-06,
      "loss": 3.2646,
      "step": 17155
    },
    {
      "epoch": 0.174560546875,
      "grad_norm": 15.842909812927246,
      "learning_rate": 4.636835844372507e-06,
      "loss": 3.2227,
      "step": 17160
    },
    {
      "epoch": 0.17461140950520834,
      "grad_norm": 9.673988342285156,
      "learning_rate": 4.636628252552128e-06,
      "loss": 3.5092,
      "step": 17165
    },
    {
      "epoch": 0.17466227213541666,
      "grad_norm": 12.802469253540039,
      "learning_rate": 4.636420606066621e-06,
      "loss": 3.3495,
      "step": 17170
    },
    {
      "epoch": 0.174713134765625,
      "grad_norm": 9.603645324707031,
      "learning_rate": 4.636212904921299e-06,
      "loss": 3.404,
      "step": 17175
    },
    {
      "epoch": 0.17476399739583334,
      "grad_norm": 11.273427963256836,
      "learning_rate": 4.6360051491214765e-06,
      "loss": 3.2583,
      "step": 17180
    },
    {
      "epoch": 0.17481486002604166,
      "grad_norm": 9.479177474975586,
      "learning_rate": 4.635797338672469e-06,
      "loss": 3.086,
      "step": 17185
    },
    {
      "epoch": 0.17486572265625,
      "grad_norm": 15.164192199707031,
      "learning_rate": 4.635589473579592e-06,
      "loss": 3.7945,
      "step": 17190
    },
    {
      "epoch": 0.17491658528645834,
      "grad_norm": 13.359272003173828,
      "learning_rate": 4.635381553848165e-06,
      "loss": 3.3775,
      "step": 17195
    },
    {
      "epoch": 0.17496744791666666,
      "grad_norm": 7.546788692474365,
      "learning_rate": 4.635173579483507e-06,
      "loss": 3.2542,
      "step": 17200
    },
    {
      "epoch": 0.175018310546875,
      "grad_norm": 13.577391624450684,
      "learning_rate": 4.634965550490939e-06,
      "loss": 3.4331,
      "step": 17205
    },
    {
      "epoch": 0.17506917317708334,
      "grad_norm": 14.945005416870117,
      "learning_rate": 4.6347574668757835e-06,
      "loss": 3.2556,
      "step": 17210
    },
    {
      "epoch": 0.17512003580729166,
      "grad_norm": 13.663776397705078,
      "learning_rate": 4.634549328643364e-06,
      "loss": 3.4574,
      "step": 17215
    },
    {
      "epoch": 0.1751708984375,
      "grad_norm": 8.942784309387207,
      "learning_rate": 4.634341135799007e-06,
      "loss": 3.4193,
      "step": 17220
    },
    {
      "epoch": 0.17522176106770834,
      "grad_norm": 15.326900482177734,
      "learning_rate": 4.634132888348037e-06,
      "loss": 3.6722,
      "step": 17225
    },
    {
      "epoch": 0.17527262369791666,
      "grad_norm": 13.63649845123291,
      "learning_rate": 4.633924586295782e-06,
      "loss": 3.4126,
      "step": 17230
    },
    {
      "epoch": 0.175323486328125,
      "grad_norm": 12.05754280090332,
      "learning_rate": 4.633716229647573e-06,
      "loss": 3.5295,
      "step": 17235
    },
    {
      "epoch": 0.17537434895833334,
      "grad_norm": 14.691134452819824,
      "learning_rate": 4.633507818408741e-06,
      "loss": 3.2017,
      "step": 17240
    },
    {
      "epoch": 0.17542521158854166,
      "grad_norm": 12.102783203125,
      "learning_rate": 4.633299352584616e-06,
      "loss": 3.3013,
      "step": 17245
    },
    {
      "epoch": 0.17547607421875,
      "grad_norm": 8.052931785583496,
      "learning_rate": 4.6330908321805336e-06,
      "loss": 3.4158,
      "step": 17250
    },
    {
      "epoch": 0.17552693684895834,
      "grad_norm": 11.040821075439453,
      "learning_rate": 4.632882257201826e-06,
      "loss": 3.1246,
      "step": 17255
    },
    {
      "epoch": 0.17557779947916666,
      "grad_norm": 11.392463684082031,
      "learning_rate": 4.632673627653833e-06,
      "loss": 3.4227,
      "step": 17260
    },
    {
      "epoch": 0.175628662109375,
      "grad_norm": 14.34813404083252,
      "learning_rate": 4.6324649435418916e-06,
      "loss": 3.2329,
      "step": 17265
    },
    {
      "epoch": 0.17567952473958334,
      "grad_norm": 9.003851890563965,
      "learning_rate": 4.632256204871338e-06,
      "loss": 3.5315,
      "step": 17270
    },
    {
      "epoch": 0.17573038736979166,
      "grad_norm": 11.924999237060547,
      "learning_rate": 4.632047411647516e-06,
      "loss": 3.4919,
      "step": 17275
    },
    {
      "epoch": 0.17578125,
      "grad_norm": 11.398469924926758,
      "learning_rate": 4.6318385638757665e-06,
      "loss": 3.9053,
      "step": 17280
    },
    {
      "epoch": 0.17583211263020834,
      "grad_norm": 10.507608413696289,
      "learning_rate": 4.631629661561432e-06,
      "loss": 3.8949,
      "step": 17285
    },
    {
      "epoch": 0.17588297526041666,
      "grad_norm": 6.944844722747803,
      "learning_rate": 4.6314207047098585e-06,
      "loss": 3.6582,
      "step": 17290
    },
    {
      "epoch": 0.175933837890625,
      "grad_norm": 15.985193252563477,
      "learning_rate": 4.63121169332639e-06,
      "loss": 3.1175,
      "step": 17295
    },
    {
      "epoch": 0.17598470052083334,
      "grad_norm": 14.670384407043457,
      "learning_rate": 4.6310026274163765e-06,
      "loss": 3.547,
      "step": 17300
    },
    {
      "epoch": 0.17603556315104166,
      "grad_norm": 12.856901168823242,
      "learning_rate": 4.630793506985166e-06,
      "loss": 3.3646,
      "step": 17305
    },
    {
      "epoch": 0.17608642578125,
      "grad_norm": 14.06156063079834,
      "learning_rate": 4.6305843320381085e-06,
      "loss": 3.4701,
      "step": 17310
    },
    {
      "epoch": 0.17613728841145834,
      "grad_norm": 18.088773727416992,
      "learning_rate": 4.630375102580557e-06,
      "loss": 3.5199,
      "step": 17315
    },
    {
      "epoch": 0.17618815104166666,
      "grad_norm": 16.56032943725586,
      "learning_rate": 4.630165818617862e-06,
      "loss": 3.3376,
      "step": 17320
    },
    {
      "epoch": 0.176239013671875,
      "grad_norm": 13.883744239807129,
      "learning_rate": 4.62995648015538e-06,
      "loss": 3.3735,
      "step": 17325
    },
    {
      "epoch": 0.17628987630208334,
      "grad_norm": 9.11486530303955,
      "learning_rate": 4.629747087198466e-06,
      "loss": 3.2933,
      "step": 17330
    },
    {
      "epoch": 0.17634073893229166,
      "grad_norm": 12.755553245544434,
      "learning_rate": 4.629537639752477e-06,
      "loss": 3.5813,
      "step": 17335
    },
    {
      "epoch": 0.1763916015625,
      "grad_norm": 9.951680183410645,
      "learning_rate": 4.629328137822774e-06,
      "loss": 3.4514,
      "step": 17340
    },
    {
      "epoch": 0.17644246419270834,
      "grad_norm": 14.455367088317871,
      "learning_rate": 4.629118581414713e-06,
      "loss": 2.6122,
      "step": 17345
    },
    {
      "epoch": 0.17649332682291666,
      "grad_norm": 8.975110054016113,
      "learning_rate": 4.6289089705336595e-06,
      "loss": 3.1559,
      "step": 17350
    },
    {
      "epoch": 0.176544189453125,
      "grad_norm": 17.275592803955078,
      "learning_rate": 4.628699305184974e-06,
      "loss": 3.3324,
      "step": 17355
    },
    {
      "epoch": 0.17659505208333334,
      "grad_norm": 14.314291954040527,
      "learning_rate": 4.628489585374022e-06,
      "loss": 3.4651,
      "step": 17360
    },
    {
      "epoch": 0.17664591471354166,
      "grad_norm": 15.956419944763184,
      "learning_rate": 4.628279811106168e-06,
      "loss": 3.6316,
      "step": 17365
    },
    {
      "epoch": 0.17669677734375,
      "grad_norm": 13.444526672363281,
      "learning_rate": 4.628069982386779e-06,
      "loss": 3.4095,
      "step": 17370
    },
    {
      "epoch": 0.17674763997395834,
      "grad_norm": 14.760951042175293,
      "learning_rate": 4.627860099221224e-06,
      "loss": 3.2458,
      "step": 17375
    },
    {
      "epoch": 0.17679850260416666,
      "grad_norm": 9.031475067138672,
      "learning_rate": 4.627650161614873e-06,
      "loss": 3.1318,
      "step": 17380
    },
    {
      "epoch": 0.176849365234375,
      "grad_norm": 9.814698219299316,
      "learning_rate": 4.627440169573098e-06,
      "loss": 3.8271,
      "step": 17385
    },
    {
      "epoch": 0.17690022786458334,
      "grad_norm": 12.851936340332031,
      "learning_rate": 4.627230123101268e-06,
      "loss": 3.5598,
      "step": 17390
    },
    {
      "epoch": 0.17695109049479166,
      "grad_norm": 44.61574935913086,
      "learning_rate": 4.627020022204761e-06,
      "loss": 3.6756,
      "step": 17395
    },
    {
      "epoch": 0.177001953125,
      "grad_norm": 11.95059871673584,
      "learning_rate": 4.626809866888951e-06,
      "loss": 3.3316,
      "step": 17400
    },
    {
      "epoch": 0.17705281575520834,
      "grad_norm": 16.417972564697266,
      "learning_rate": 4.626599657159216e-06,
      "loss": 3.3807,
      "step": 17405
    },
    {
      "epoch": 0.17710367838541666,
      "grad_norm": 13.880332946777344,
      "learning_rate": 4.6263893930209304e-06,
      "loss": 3.3911,
      "step": 17410
    },
    {
      "epoch": 0.177154541015625,
      "grad_norm": 15.084425926208496,
      "learning_rate": 4.6261790744794765e-06,
      "loss": 3.4722,
      "step": 17415
    },
    {
      "epoch": 0.17720540364583334,
      "grad_norm": 15.803872108459473,
      "learning_rate": 4.625968701540236e-06,
      "loss": 3.1724,
      "step": 17420
    },
    {
      "epoch": 0.17725626627604166,
      "grad_norm": 7.609067440032959,
      "learning_rate": 4.62575827420859e-06,
      "loss": 3.2543,
      "step": 17425
    },
    {
      "epoch": 0.17730712890625,
      "grad_norm": 11.47111701965332,
      "learning_rate": 4.625547792489922e-06,
      "loss": 3.3642,
      "step": 17430
    },
    {
      "epoch": 0.17735799153645834,
      "grad_norm": 9.357483863830566,
      "learning_rate": 4.625337256389618e-06,
      "loss": 3.5617,
      "step": 17435
    },
    {
      "epoch": 0.17740885416666666,
      "grad_norm": 12.891478538513184,
      "learning_rate": 4.625126665913063e-06,
      "loss": 3.2682,
      "step": 17440
    },
    {
      "epoch": 0.177459716796875,
      "grad_norm": 14.151500701904297,
      "learning_rate": 4.6249160210656476e-06,
      "loss": 3.3943,
      "step": 17445
    },
    {
      "epoch": 0.17751057942708334,
      "grad_norm": 11.056961059570312,
      "learning_rate": 4.624705321852758e-06,
      "loss": 3.2693,
      "step": 17450
    },
    {
      "epoch": 0.17756144205729166,
      "grad_norm": 16.99146270751953,
      "learning_rate": 4.624494568279787e-06,
      "loss": 3.5405,
      "step": 17455
    },
    {
      "epoch": 0.1776123046875,
      "grad_norm": 9.272522926330566,
      "learning_rate": 4.624283760352126e-06,
      "loss": 3.0549,
      "step": 17460
    },
    {
      "epoch": 0.17766316731770834,
      "grad_norm": 10.904102325439453,
      "learning_rate": 4.624072898075168e-06,
      "loss": 3.0915,
      "step": 17465
    },
    {
      "epoch": 0.17771402994791666,
      "grad_norm": 15.731610298156738,
      "learning_rate": 4.6238619814543094e-06,
      "loss": 3.3617,
      "step": 17470
    },
    {
      "epoch": 0.177764892578125,
      "grad_norm": 15.29787540435791,
      "learning_rate": 4.623651010494945e-06,
      "loss": 3.4183,
      "step": 17475
    },
    {
      "epoch": 0.17781575520833334,
      "grad_norm": 13.272435188293457,
      "learning_rate": 4.623439985202472e-06,
      "loss": 3.3338,
      "step": 17480
    },
    {
      "epoch": 0.17786661783854166,
      "grad_norm": 10.852397918701172,
      "learning_rate": 4.623228905582292e-06,
      "loss": 3.6102,
      "step": 17485
    },
    {
      "epoch": 0.17791748046875,
      "grad_norm": 10.077887535095215,
      "learning_rate": 4.623017771639803e-06,
      "loss": 3.4454,
      "step": 17490
    },
    {
      "epoch": 0.17796834309895834,
      "grad_norm": 13.571096420288086,
      "learning_rate": 4.622806583380407e-06,
      "loss": 3.3273,
      "step": 17495
    },
    {
      "epoch": 0.17801920572916666,
      "grad_norm": 9.822858810424805,
      "learning_rate": 4.622595340809508e-06,
      "loss": 3.4383,
      "step": 17500
    },
    {
      "epoch": 0.178070068359375,
      "grad_norm": 16.22844123840332,
      "learning_rate": 4.622384043932509e-06,
      "loss": 3.0372,
      "step": 17505
    },
    {
      "epoch": 0.17812093098958334,
      "grad_norm": 7.383034706115723,
      "learning_rate": 4.622172692754819e-06,
      "loss": 3.1181,
      "step": 17510
    },
    {
      "epoch": 0.17817179361979166,
      "grad_norm": 9.609219551086426,
      "learning_rate": 4.621961287281843e-06,
      "loss": 3.3651,
      "step": 17515
    },
    {
      "epoch": 0.17822265625,
      "grad_norm": 9.730359077453613,
      "learning_rate": 4.621749827518991e-06,
      "loss": 3.4125,
      "step": 17520
    },
    {
      "epoch": 0.17827351888020834,
      "grad_norm": 7.237141132354736,
      "learning_rate": 4.621538313471673e-06,
      "loss": 3.2425,
      "step": 17525
    },
    {
      "epoch": 0.17832438151041666,
      "grad_norm": 10.888049125671387,
      "learning_rate": 4.621326745145299e-06,
      "loss": 3.5376,
      "step": 17530
    },
    {
      "epoch": 0.178375244140625,
      "grad_norm": 12.970224380493164,
      "learning_rate": 4.6211151225452835e-06,
      "loss": 3.5308,
      "step": 17535
    },
    {
      "epoch": 0.17842610677083334,
      "grad_norm": 12.810026168823242,
      "learning_rate": 4.62090344567704e-06,
      "loss": 3.1474,
      "step": 17540
    },
    {
      "epoch": 0.17847696940104166,
      "grad_norm": 8.826539993286133,
      "learning_rate": 4.6206917145459855e-06,
      "loss": 3.2885,
      "step": 17545
    },
    {
      "epoch": 0.17852783203125,
      "grad_norm": 10.239234924316406,
      "learning_rate": 4.620479929157535e-06,
      "loss": 3.1573,
      "step": 17550
    },
    {
      "epoch": 0.17857869466145834,
      "grad_norm": 14.187760353088379,
      "learning_rate": 4.620268089517108e-06,
      "loss": 3.2755,
      "step": 17555
    },
    {
      "epoch": 0.17862955729166666,
      "grad_norm": 11.916472434997559,
      "learning_rate": 4.620056195630125e-06,
      "loss": 3.2658,
      "step": 17560
    },
    {
      "epoch": 0.178680419921875,
      "grad_norm": 18.4185791015625,
      "learning_rate": 4.619844247502007e-06,
      "loss": 3.3846,
      "step": 17565
    },
    {
      "epoch": 0.17873128255208334,
      "grad_norm": 10.282975196838379,
      "learning_rate": 4.619632245138176e-06,
      "loss": 3.497,
      "step": 17570
    },
    {
      "epoch": 0.17878214518229166,
      "grad_norm": 15.2913179397583,
      "learning_rate": 4.619420188544057e-06,
      "loss": 3.5962,
      "step": 17575
    },
    {
      "epoch": 0.1788330078125,
      "grad_norm": 8.401836395263672,
      "learning_rate": 4.619208077725075e-06,
      "loss": 3.2448,
      "step": 17580
    },
    {
      "epoch": 0.17888387044270834,
      "grad_norm": 11.083022117614746,
      "learning_rate": 4.6189959126866555e-06,
      "loss": 3.0683,
      "step": 17585
    },
    {
      "epoch": 0.17893473307291666,
      "grad_norm": 11.012330055236816,
      "learning_rate": 4.618783693434229e-06,
      "loss": 3.3922,
      "step": 17590
    },
    {
      "epoch": 0.178985595703125,
      "grad_norm": 10.7522611618042,
      "learning_rate": 4.618571419973222e-06,
      "loss": 3.1457,
      "step": 17595
    },
    {
      "epoch": 0.17903645833333334,
      "grad_norm": 9.602813720703125,
      "learning_rate": 4.6183590923090696e-06,
      "loss": 3.2299,
      "step": 17600
    },
    {
      "epoch": 0.17908732096354166,
      "grad_norm": 8.291756629943848,
      "learning_rate": 4.6181467104472005e-06,
      "loss": 3.0146,
      "step": 17605
    },
    {
      "epoch": 0.17913818359375,
      "grad_norm": 13.860746383666992,
      "learning_rate": 4.61793427439305e-06,
      "loss": 3.1955,
      "step": 17610
    },
    {
      "epoch": 0.17918904622395834,
      "grad_norm": 13.45958423614502,
      "learning_rate": 4.6177217841520535e-06,
      "loss": 2.8876,
      "step": 17615
    },
    {
      "epoch": 0.17923990885416666,
      "grad_norm": 12.669150352478027,
      "learning_rate": 4.617509239729647e-06,
      "loss": 3.6825,
      "step": 17620
    },
    {
      "epoch": 0.179290771484375,
      "grad_norm": 11.674629211425781,
      "learning_rate": 4.61729664113127e-06,
      "loss": 3.4646,
      "step": 17625
    },
    {
      "epoch": 0.17934163411458334,
      "grad_norm": 14.575140953063965,
      "learning_rate": 4.617083988362358e-06,
      "loss": 2.9819,
      "step": 17630
    },
    {
      "epoch": 0.17939249674479166,
      "grad_norm": 13.144431114196777,
      "learning_rate": 4.616871281428355e-06,
      "loss": 3.5436,
      "step": 17635
    },
    {
      "epoch": 0.179443359375,
      "grad_norm": 14.929411888122559,
      "learning_rate": 4.616658520334701e-06,
      "loss": 3.3415,
      "step": 17640
    },
    {
      "epoch": 0.17949422200520834,
      "grad_norm": 13.124006271362305,
      "learning_rate": 4.616445705086842e-06,
      "loss": 3.3565,
      "step": 17645
    },
    {
      "epoch": 0.17954508463541666,
      "grad_norm": 8.569463729858398,
      "learning_rate": 4.616232835690221e-06,
      "loss": 4.0458,
      "step": 17650
    },
    {
      "epoch": 0.179595947265625,
      "grad_norm": 11.597271919250488,
      "learning_rate": 4.616019912150284e-06,
      "loss": 3.5618,
      "step": 17655
    },
    {
      "epoch": 0.17964680989583334,
      "grad_norm": 12.50623607635498,
      "learning_rate": 4.615806934472479e-06,
      "loss": 3.4437,
      "step": 17660
    },
    {
      "epoch": 0.17969767252604166,
      "grad_norm": 13.184673309326172,
      "learning_rate": 4.615593902662256e-06,
      "loss": 3.4312,
      "step": 17665
    },
    {
      "epoch": 0.17974853515625,
      "grad_norm": 10.000298500061035,
      "learning_rate": 4.615380816725063e-06,
      "loss": 3.693,
      "step": 17670
    },
    {
      "epoch": 0.17979939778645834,
      "grad_norm": 15.589776039123535,
      "learning_rate": 4.6151676766663536e-06,
      "loss": 3.4443,
      "step": 17675
    },
    {
      "epoch": 0.17985026041666666,
      "grad_norm": 12.44705867767334,
      "learning_rate": 4.614954482491581e-06,
      "loss": 3.2575,
      "step": 17680
    },
    {
      "epoch": 0.179901123046875,
      "grad_norm": 16.3400821685791,
      "learning_rate": 4.6147412342061995e-06,
      "loss": 3.1107,
      "step": 17685
    },
    {
      "epoch": 0.17995198567708334,
      "grad_norm": 9.2102689743042,
      "learning_rate": 4.614527931815664e-06,
      "loss": 3.2749,
      "step": 17690
    },
    {
      "epoch": 0.18000284830729166,
      "grad_norm": 9.545188903808594,
      "learning_rate": 4.6143145753254335e-06,
      "loss": 3.5115,
      "step": 17695
    },
    {
      "epoch": 0.1800537109375,
      "grad_norm": 13.04957389831543,
      "learning_rate": 4.614101164740965e-06,
      "loss": 3.5562,
      "step": 17700
    },
    {
      "epoch": 0.18010457356770834,
      "grad_norm": 10.874368667602539,
      "learning_rate": 4.613887700067719e-06,
      "loss": 3.3832,
      "step": 17705
    },
    {
      "epoch": 0.18015543619791666,
      "grad_norm": 11.728689193725586,
      "learning_rate": 4.613674181311158e-06,
      "loss": 3.2963,
      "step": 17710
    },
    {
      "epoch": 0.180206298828125,
      "grad_norm": 10.86733341217041,
      "learning_rate": 4.613460608476744e-06,
      "loss": 3.2448,
      "step": 17715
    },
    {
      "epoch": 0.18025716145833334,
      "grad_norm": 13.16111946105957,
      "learning_rate": 4.613246981569941e-06,
      "loss": 3.5583,
      "step": 17720
    },
    {
      "epoch": 0.18030802408854166,
      "grad_norm": 15.626137733459473,
      "learning_rate": 4.6130333005962144e-06,
      "loss": 3.4084,
      "step": 17725
    },
    {
      "epoch": 0.18035888671875,
      "grad_norm": 10.255508422851562,
      "learning_rate": 4.612819565561033e-06,
      "loss": 3.2747,
      "step": 17730
    },
    {
      "epoch": 0.18040974934895834,
      "grad_norm": 10.121077537536621,
      "learning_rate": 4.612605776469863e-06,
      "loss": 3.3058,
      "step": 17735
    },
    {
      "epoch": 0.18046061197916666,
      "grad_norm": 12.807663917541504,
      "learning_rate": 4.612391933328175e-06,
      "loss": 3.1888,
      "step": 17740
    },
    {
      "epoch": 0.180511474609375,
      "grad_norm": 7.487052917480469,
      "learning_rate": 4.61217803614144e-06,
      "loss": 3.2892,
      "step": 17745
    },
    {
      "epoch": 0.18056233723958334,
      "grad_norm": 13.956525802612305,
      "learning_rate": 4.61196408491513e-06,
      "loss": 3.7043,
      "step": 17750
    },
    {
      "epoch": 0.18061319986979166,
      "grad_norm": 9.125615119934082,
      "learning_rate": 4.611750079654721e-06,
      "loss": 3.4562,
      "step": 17755
    },
    {
      "epoch": 0.1806640625,
      "grad_norm": 14.65468692779541,
      "learning_rate": 4.611536020365686e-06,
      "loss": 3.3539,
      "step": 17760
    },
    {
      "epoch": 0.18071492513020834,
      "grad_norm": 10.214896202087402,
      "learning_rate": 4.611321907053502e-06,
      "loss": 3.2788,
      "step": 17765
    },
    {
      "epoch": 0.18076578776041666,
      "grad_norm": 13.62902545928955,
      "learning_rate": 4.611107739723647e-06,
      "loss": 3.3931,
      "step": 17770
    },
    {
      "epoch": 0.180816650390625,
      "grad_norm": 12.189156532287598,
      "learning_rate": 4.610893518381602e-06,
      "loss": 3.4467,
      "step": 17775
    },
    {
      "epoch": 0.18086751302083334,
      "grad_norm": 17.752410888671875,
      "learning_rate": 4.610679243032846e-06,
      "loss": 3.1061,
      "step": 17780
    },
    {
      "epoch": 0.18091837565104166,
      "grad_norm": 9.7039794921875,
      "learning_rate": 4.610464913682863e-06,
      "loss": 3.3899,
      "step": 17785
    },
    {
      "epoch": 0.18096923828125,
      "grad_norm": 18.238666534423828,
      "learning_rate": 4.610250530337134e-06,
      "loss": 3.4644,
      "step": 17790
    },
    {
      "epoch": 0.18102010091145834,
      "grad_norm": 8.860824584960938,
      "learning_rate": 4.6100360930011455e-06,
      "loss": 3.4091,
      "step": 17795
    },
    {
      "epoch": 0.18107096354166666,
      "grad_norm": 10.903741836547852,
      "learning_rate": 4.6098216016803845e-06,
      "loss": 3.276,
      "step": 17800
    },
    {
      "epoch": 0.181121826171875,
      "grad_norm": 13.30379581451416,
      "learning_rate": 4.609607056380337e-06,
      "loss": 3.2626,
      "step": 17805
    },
    {
      "epoch": 0.18117268880208334,
      "grad_norm": 12.02137565612793,
      "learning_rate": 4.609392457106494e-06,
      "loss": 3.6866,
      "step": 17810
    },
    {
      "epoch": 0.18122355143229166,
      "grad_norm": 10.083224296569824,
      "learning_rate": 4.6091778038643445e-06,
      "loss": 3.2943,
      "step": 17815
    },
    {
      "epoch": 0.1812744140625,
      "grad_norm": 13.689177513122559,
      "learning_rate": 4.608963096659381e-06,
      "loss": 3.4036,
      "step": 17820
    },
    {
      "epoch": 0.18132527669270834,
      "grad_norm": 10.178418159484863,
      "learning_rate": 4.608748335497096e-06,
      "loss": 3.3081,
      "step": 17825
    },
    {
      "epoch": 0.18137613932291666,
      "grad_norm": 13.94970417022705,
      "learning_rate": 4.608533520382985e-06,
      "loss": 3.3372,
      "step": 17830
    },
    {
      "epoch": 0.181427001953125,
      "grad_norm": 10.877111434936523,
      "learning_rate": 4.608318651322543e-06,
      "loss": 3.6664,
      "step": 17835
    },
    {
      "epoch": 0.18147786458333334,
      "grad_norm": 11.025904655456543,
      "learning_rate": 4.608103728321269e-06,
      "loss": 3.7866,
      "step": 17840
    },
    {
      "epoch": 0.18152872721354166,
      "grad_norm": 14.711570739746094,
      "learning_rate": 4.6078887513846605e-06,
      "loss": 3.4579,
      "step": 17845
    },
    {
      "epoch": 0.18157958984375,
      "grad_norm": 9.070608139038086,
      "learning_rate": 4.607673720518218e-06,
      "loss": 3.3378,
      "step": 17850
    },
    {
      "epoch": 0.18163045247395834,
      "grad_norm": 10.735955238342285,
      "learning_rate": 4.607458635727443e-06,
      "loss": 3.4461,
      "step": 17855
    },
    {
      "epoch": 0.18168131510416666,
      "grad_norm": 16.661575317382812,
      "learning_rate": 4.607243497017838e-06,
      "loss": 3.3425,
      "step": 17860
    },
    {
      "epoch": 0.181732177734375,
      "grad_norm": 12.659605979919434,
      "learning_rate": 4.607028304394907e-06,
      "loss": 3.9381,
      "step": 17865
    },
    {
      "epoch": 0.18178304036458334,
      "grad_norm": 16.774187088012695,
      "learning_rate": 4.606813057864158e-06,
      "loss": 3.5132,
      "step": 17870
    },
    {
      "epoch": 0.18183390299479166,
      "grad_norm": 6.721248149871826,
      "learning_rate": 4.606597757431095e-06,
      "loss": 3.3477,
      "step": 17875
    },
    {
      "epoch": 0.181884765625,
      "grad_norm": 11.525874137878418,
      "learning_rate": 4.606382403101228e-06,
      "loss": 3.5922,
      "step": 17880
    },
    {
      "epoch": 0.18193562825520834,
      "grad_norm": 13.829561233520508,
      "learning_rate": 4.606166994880067e-06,
      "loss": 3.2248,
      "step": 17885
    },
    {
      "epoch": 0.18198649088541666,
      "grad_norm": 8.705430030822754,
      "learning_rate": 4.605951532773122e-06,
      "loss": 3.2562,
      "step": 17890
    },
    {
      "epoch": 0.182037353515625,
      "grad_norm": 8.117154121398926,
      "learning_rate": 4.605736016785905e-06,
      "loss": 3.2657,
      "step": 17895
    },
    {
      "epoch": 0.18208821614583334,
      "grad_norm": 12.32487964630127,
      "learning_rate": 4.605520446923933e-06,
      "loss": 3.1516,
      "step": 17900
    },
    {
      "epoch": 0.18213907877604166,
      "grad_norm": 14.840004920959473,
      "learning_rate": 4.605304823192719e-06,
      "loss": 3.4167,
      "step": 17905
    },
    {
      "epoch": 0.18218994140625,
      "grad_norm": 14.861167907714844,
      "learning_rate": 4.60508914559778e-06,
      "loss": 3.729,
      "step": 17910
    },
    {
      "epoch": 0.18224080403645834,
      "grad_norm": 12.168790817260742,
      "learning_rate": 4.6048734141446335e-06,
      "loss": 3.2407,
      "step": 17915
    },
    {
      "epoch": 0.18229166666666666,
      "grad_norm": 7.106325149536133,
      "learning_rate": 4.604657628838801e-06,
      "loss": 3.2299,
      "step": 17920
    },
    {
      "epoch": 0.182342529296875,
      "grad_norm": 7.858881950378418,
      "learning_rate": 4.604441789685801e-06,
      "loss": 3.477,
      "step": 17925
    },
    {
      "epoch": 0.18239339192708334,
      "grad_norm": 11.167740821838379,
      "learning_rate": 4.604225896691157e-06,
      "loss": 3.673,
      "step": 17930
    },
    {
      "epoch": 0.18244425455729166,
      "grad_norm": 14.516616821289062,
      "learning_rate": 4.604009949860392e-06,
      "loss": 3.0951,
      "step": 17935
    },
    {
      "epoch": 0.1824951171875,
      "grad_norm": 11.77841567993164,
      "learning_rate": 4.603793949199031e-06,
      "loss": 3.4499,
      "step": 17940
    },
    {
      "epoch": 0.18254597981770834,
      "grad_norm": 12.027185440063477,
      "learning_rate": 4.603577894712601e-06,
      "loss": 3.8324,
      "step": 17945
    },
    {
      "epoch": 0.18259684244791666,
      "grad_norm": 13.794437408447266,
      "learning_rate": 4.603361786406628e-06,
      "loss": 3.6594,
      "step": 17950
    },
    {
      "epoch": 0.182647705078125,
      "grad_norm": 13.343639373779297,
      "learning_rate": 4.603145624286643e-06,
      "loss": 3.3471,
      "step": 17955
    },
    {
      "epoch": 0.18269856770833334,
      "grad_norm": 10.639410018920898,
      "learning_rate": 4.602929408358176e-06,
      "loss": 3.3093,
      "step": 17960
    },
    {
      "epoch": 0.18274943033854166,
      "grad_norm": 15.507131576538086,
      "learning_rate": 4.602713138626758e-06,
      "loss": 3.3536,
      "step": 17965
    },
    {
      "epoch": 0.18280029296875,
      "grad_norm": 13.404929161071777,
      "learning_rate": 4.602496815097923e-06,
      "loss": 3.244,
      "step": 17970
    },
    {
      "epoch": 0.18285115559895834,
      "grad_norm": 13.34079647064209,
      "learning_rate": 4.602280437777205e-06,
      "loss": 3.4758,
      "step": 17975
    },
    {
      "epoch": 0.18290201822916666,
      "grad_norm": 10.848614692687988,
      "learning_rate": 4.602064006670141e-06,
      "loss": 3.2688,
      "step": 17980
    },
    {
      "epoch": 0.182952880859375,
      "grad_norm": 11.669103622436523,
      "learning_rate": 4.601847521782268e-06,
      "loss": 3.51,
      "step": 17985
    },
    {
      "epoch": 0.18300374348958334,
      "grad_norm": 9.459100723266602,
      "learning_rate": 4.601630983119123e-06,
      "loss": 3.3047,
      "step": 17990
    },
    {
      "epoch": 0.18305460611979166,
      "grad_norm": 10.783151626586914,
      "learning_rate": 4.601414390686248e-06,
      "loss": 3.1554,
      "step": 17995
    },
    {
      "epoch": 0.18310546875,
      "grad_norm": 7.823272705078125,
      "learning_rate": 4.601197744489184e-06,
      "loss": 3.2805,
      "step": 18000
    },
    {
      "epoch": 0.18315633138020834,
      "grad_norm": 10.413420677185059,
      "learning_rate": 4.600981044533473e-06,
      "loss": 3.1569,
      "step": 18005
    },
    {
      "epoch": 0.18320719401041666,
      "grad_norm": 18.30168342590332,
      "learning_rate": 4.600764290824661e-06,
      "loss": 3.2039,
      "step": 18010
    },
    {
      "epoch": 0.183258056640625,
      "grad_norm": 13.457409858703613,
      "learning_rate": 4.600547483368292e-06,
      "loss": 3.5678,
      "step": 18015
    },
    {
      "epoch": 0.18330891927083334,
      "grad_norm": 12.043107032775879,
      "learning_rate": 4.600330622169914e-06,
      "loss": 3.2696,
      "step": 18020
    },
    {
      "epoch": 0.18335978190104166,
      "grad_norm": 10.646010398864746,
      "learning_rate": 4.600113707235075e-06,
      "loss": 3.2272,
      "step": 18025
    },
    {
      "epoch": 0.18341064453125,
      "grad_norm": 12.23196792602539,
      "learning_rate": 4.5998967385693235e-06,
      "loss": 3.4313,
      "step": 18030
    },
    {
      "epoch": 0.18346150716145834,
      "grad_norm": 9.132007598876953,
      "learning_rate": 4.599679716178212e-06,
      "loss": 3.5767,
      "step": 18035
    },
    {
      "epoch": 0.18351236979166666,
      "grad_norm": 15.052911758422852,
      "learning_rate": 4.599462640067294e-06,
      "loss": 3.5414,
      "step": 18040
    },
    {
      "epoch": 0.183563232421875,
      "grad_norm": 12.69869327545166,
      "learning_rate": 4.599245510242121e-06,
      "loss": 3.2965,
      "step": 18045
    },
    {
      "epoch": 0.18361409505208334,
      "grad_norm": 10.353547096252441,
      "learning_rate": 4.599028326708248e-06,
      "loss": 2.9671,
      "step": 18050
    },
    {
      "epoch": 0.18366495768229166,
      "grad_norm": 8.513712882995605,
      "learning_rate": 4.598811089471235e-06,
      "loss": 3.4682,
      "step": 18055
    },
    {
      "epoch": 0.1837158203125,
      "grad_norm": 10.49561882019043,
      "learning_rate": 4.598593798536636e-06,
      "loss": 3.1158,
      "step": 18060
    },
    {
      "epoch": 0.18376668294270834,
      "grad_norm": 7.227719306945801,
      "learning_rate": 4.598376453910013e-06,
      "loss": 3.4783,
      "step": 18065
    },
    {
      "epoch": 0.18381754557291666,
      "grad_norm": 8.991704940795898,
      "learning_rate": 4.598159055596926e-06,
      "loss": 3.258,
      "step": 18070
    },
    {
      "epoch": 0.183868408203125,
      "grad_norm": 11.315898895263672,
      "learning_rate": 4.5979416036029366e-06,
      "loss": 3.3756,
      "step": 18075
    },
    {
      "epoch": 0.18391927083333334,
      "grad_norm": 10.222885131835938,
      "learning_rate": 4.597724097933608e-06,
      "loss": 3.6465,
      "step": 18080
    },
    {
      "epoch": 0.18397013346354166,
      "grad_norm": 15.640301704406738,
      "learning_rate": 4.597506538594506e-06,
      "loss": 3.247,
      "step": 18085
    },
    {
      "epoch": 0.18402099609375,
      "grad_norm": 14.560544967651367,
      "learning_rate": 4.597288925591196e-06,
      "loss": 3.6574,
      "step": 18090
    },
    {
      "epoch": 0.18407185872395834,
      "grad_norm": 11.333292007446289,
      "learning_rate": 4.597071258929247e-06,
      "loss": 3.6407,
      "step": 18095
    },
    {
      "epoch": 0.18412272135416666,
      "grad_norm": 16.446786880493164,
      "learning_rate": 4.596853538614226e-06,
      "loss": 3.505,
      "step": 18100
    },
    {
      "epoch": 0.184173583984375,
      "grad_norm": 8.610262870788574,
      "learning_rate": 4.596635764651704e-06,
      "loss": 3.867,
      "step": 18105
    },
    {
      "epoch": 0.18422444661458334,
      "grad_norm": 11.611706733703613,
      "learning_rate": 4.596417937047253e-06,
      "loss": 3.4687,
      "step": 18110
    },
    {
      "epoch": 0.18427530924479166,
      "grad_norm": 9.320916175842285,
      "learning_rate": 4.5962000558064465e-06,
      "loss": 3.6369,
      "step": 18115
    },
    {
      "epoch": 0.184326171875,
      "grad_norm": 10.020218849182129,
      "learning_rate": 4.5959821209348585e-06,
      "loss": 3.2122,
      "step": 18120
    },
    {
      "epoch": 0.18437703450520834,
      "grad_norm": 15.03809642791748,
      "learning_rate": 4.595764132438064e-06,
      "loss": 3.3018,
      "step": 18125
    },
    {
      "epoch": 0.18442789713541666,
      "grad_norm": 13.933128356933594,
      "learning_rate": 4.595546090321642e-06,
      "loss": 3.4161,
      "step": 18130
    },
    {
      "epoch": 0.184478759765625,
      "grad_norm": 15.626864433288574,
      "learning_rate": 4.595327994591169e-06,
      "loss": 3.1919,
      "step": 18135
    },
    {
      "epoch": 0.18452962239583334,
      "grad_norm": 15.305878639221191,
      "learning_rate": 4.595109845252226e-06,
      "loss": 3.7762,
      "step": 18140
    },
    {
      "epoch": 0.18458048502604166,
      "grad_norm": 12.688610076904297,
      "learning_rate": 4.594891642310395e-06,
      "loss": 3.5695,
      "step": 18145
    },
    {
      "epoch": 0.18463134765625,
      "grad_norm": 12.089020729064941,
      "learning_rate": 4.594673385771257e-06,
      "loss": 3.651,
      "step": 18150
    },
    {
      "epoch": 0.18468221028645834,
      "grad_norm": 14.751028060913086,
      "learning_rate": 4.594455075640397e-06,
      "loss": 3.3362,
      "step": 18155
    },
    {
      "epoch": 0.18473307291666666,
      "grad_norm": 14.168632507324219,
      "learning_rate": 4.594236711923401e-06,
      "loss": 3.5103,
      "step": 18160
    },
    {
      "epoch": 0.184783935546875,
      "grad_norm": 11.68898868560791,
      "learning_rate": 4.594018294625855e-06,
      "loss": 3.0502,
      "step": 18165
    },
    {
      "epoch": 0.18483479817708334,
      "grad_norm": 11.675402641296387,
      "learning_rate": 4.593799823753347e-06,
      "loss": 3.2437,
      "step": 18170
    },
    {
      "epoch": 0.18488566080729166,
      "grad_norm": 14.426863670349121,
      "learning_rate": 4.593581299311467e-06,
      "loss": 3.5249,
      "step": 18175
    },
    {
      "epoch": 0.1849365234375,
      "grad_norm": 14.298029899597168,
      "learning_rate": 4.593362721305805e-06,
      "loss": 3.3278,
      "step": 18180
    },
    {
      "epoch": 0.18498738606770834,
      "grad_norm": 13.843632698059082,
      "learning_rate": 4.5931440897419546e-06,
      "loss": 3.3659,
      "step": 18185
    },
    {
      "epoch": 0.18503824869791666,
      "grad_norm": 15.963536262512207,
      "learning_rate": 4.592925404625509e-06,
      "loss": 3.2535,
      "step": 18190
    },
    {
      "epoch": 0.185089111328125,
      "grad_norm": 13.099442481994629,
      "learning_rate": 4.592706665962063e-06,
      "loss": 3.4257,
      "step": 18195
    },
    {
      "epoch": 0.18513997395833334,
      "grad_norm": 10.886616706848145,
      "learning_rate": 4.592487873757212e-06,
      "loss": 3.3174,
      "step": 18200
    },
    {
      "epoch": 0.18519083658854166,
      "grad_norm": 14.132744789123535,
      "learning_rate": 4.592269028016555e-06,
      "loss": 3.5966,
      "step": 18205
    },
    {
      "epoch": 0.18524169921875,
      "grad_norm": 12.172066688537598,
      "learning_rate": 4.5920501287456905e-06,
      "loss": 3.25,
      "step": 18210
    },
    {
      "epoch": 0.18529256184895834,
      "grad_norm": 9.599930763244629,
      "learning_rate": 4.591831175950221e-06,
      "loss": 3.3482,
      "step": 18215
    },
    {
      "epoch": 0.18534342447916666,
      "grad_norm": 14.642792701721191,
      "learning_rate": 4.5916121696357454e-06,
      "loss": 3.195,
      "step": 18220
    },
    {
      "epoch": 0.185394287109375,
      "grad_norm": 17.15027618408203,
      "learning_rate": 4.591393109807868e-06,
      "loss": 3.6213,
      "step": 18225
    },
    {
      "epoch": 0.18544514973958334,
      "grad_norm": 10.881146430969238,
      "learning_rate": 4.591173996472195e-06,
      "loss": 3.054,
      "step": 18230
    },
    {
      "epoch": 0.18549601236979166,
      "grad_norm": 16.243993759155273,
      "learning_rate": 4.5909548296343295e-06,
      "loss": 3.2491,
      "step": 18235
    },
    {
      "epoch": 0.185546875,
      "grad_norm": 12.273772239685059,
      "learning_rate": 4.590735609299881e-06,
      "loss": 3.208,
      "step": 18240
    },
    {
      "epoch": 0.18559773763020834,
      "grad_norm": 12.737349510192871,
      "learning_rate": 4.590516335474458e-06,
      "loss": 3.0281,
      "step": 18245
    },
    {
      "epoch": 0.18564860026041666,
      "grad_norm": 10.880949020385742,
      "learning_rate": 4.590297008163669e-06,
      "loss": 3.3972,
      "step": 18250
    },
    {
      "epoch": 0.185699462890625,
      "grad_norm": 10.571752548217773,
      "learning_rate": 4.590077627373126e-06,
      "loss": 3.5009,
      "step": 18255
    },
    {
      "epoch": 0.18575032552083334,
      "grad_norm": 17.717178344726562,
      "learning_rate": 4.589858193108444e-06,
      "loss": 3.5101,
      "step": 18260
    },
    {
      "epoch": 0.18580118815104166,
      "grad_norm": 12.134099006652832,
      "learning_rate": 4.589638705375234e-06,
      "loss": 3.3066,
      "step": 18265
    },
    {
      "epoch": 0.18585205078125,
      "grad_norm": 15.478253364562988,
      "learning_rate": 4.5894191641791145e-06,
      "loss": 3.329,
      "step": 18270
    },
    {
      "epoch": 0.18590291341145834,
      "grad_norm": 8.206337928771973,
      "learning_rate": 4.5891995695257e-06,
      "loss": 3.2733,
      "step": 18275
    },
    {
      "epoch": 0.18595377604166666,
      "grad_norm": 10.964665412902832,
      "learning_rate": 4.58897992142061e-06,
      "loss": 3.5635,
      "step": 18280
    },
    {
      "epoch": 0.186004638671875,
      "grad_norm": 12.019591331481934,
      "learning_rate": 4.588760219869463e-06,
      "loss": 3.3073,
      "step": 18285
    },
    {
      "epoch": 0.18605550130208334,
      "grad_norm": 12.096267700195312,
      "learning_rate": 4.588540464877882e-06,
      "loss": 3.4862,
      "step": 18290
    },
    {
      "epoch": 0.18610636393229166,
      "grad_norm": 10.249032020568848,
      "learning_rate": 4.588320656451487e-06,
      "loss": 3.6171,
      "step": 18295
    },
    {
      "epoch": 0.1861572265625,
      "grad_norm": 16.42626190185547,
      "learning_rate": 4.588100794595904e-06,
      "loss": 3.5845,
      "step": 18300
    },
    {
      "epoch": 0.18620808919270834,
      "grad_norm": 16.055395126342773,
      "learning_rate": 4.587880879316758e-06,
      "loss": 3.5212,
      "step": 18305
    },
    {
      "epoch": 0.18625895182291666,
      "grad_norm": 14.237022399902344,
      "learning_rate": 4.587660910619672e-06,
      "loss": 3.1991,
      "step": 18310
    },
    {
      "epoch": 0.186309814453125,
      "grad_norm": 7.7949652671813965,
      "learning_rate": 4.5874408885102785e-06,
      "loss": 3.4076,
      "step": 18315
    },
    {
      "epoch": 0.18636067708333334,
      "grad_norm": 8.902517318725586,
      "learning_rate": 4.5872208129942045e-06,
      "loss": 3.2999,
      "step": 18320
    },
    {
      "epoch": 0.18641153971354166,
      "grad_norm": 11.778844833374023,
      "learning_rate": 4.58700068407708e-06,
      "loss": 3.2865,
      "step": 18325
    },
    {
      "epoch": 0.18646240234375,
      "grad_norm": 8.537171363830566,
      "learning_rate": 4.586780501764538e-06,
      "loss": 3.2332,
      "step": 18330
    },
    {
      "epoch": 0.18651326497395834,
      "grad_norm": 15.2643461227417,
      "learning_rate": 4.586560266062211e-06,
      "loss": 3.3204,
      "step": 18335
    },
    {
      "epoch": 0.18656412760416666,
      "grad_norm": 8.688283920288086,
      "learning_rate": 4.586339976975735e-06,
      "loss": 3.3692,
      "step": 18340
    },
    {
      "epoch": 0.186614990234375,
      "grad_norm": 12.671655654907227,
      "learning_rate": 4.586119634510745e-06,
      "loss": 3.3349,
      "step": 18345
    },
    {
      "epoch": 0.18666585286458334,
      "grad_norm": 12.594867706298828,
      "learning_rate": 4.585899238672878e-06,
      "loss": 3.4412,
      "step": 18350
    },
    {
      "epoch": 0.18671671549479166,
      "grad_norm": 16.03032875061035,
      "learning_rate": 4.585678789467774e-06,
      "loss": 3.474,
      "step": 18355
    },
    {
      "epoch": 0.186767578125,
      "grad_norm": 13.385184288024902,
      "learning_rate": 4.585458286901072e-06,
      "loss": 3.2519,
      "step": 18360
    },
    {
      "epoch": 0.18681844075520834,
      "grad_norm": 9.913110733032227,
      "learning_rate": 4.5852377309784146e-06,
      "loss": 3.3587,
      "step": 18365
    },
    {
      "epoch": 0.18686930338541666,
      "grad_norm": 10.111922264099121,
      "learning_rate": 4.585017121705444e-06,
      "loss": 3.1416,
      "step": 18370
    },
    {
      "epoch": 0.186920166015625,
      "grad_norm": 9.840933799743652,
      "learning_rate": 4.584796459087805e-06,
      "loss": 3.5775,
      "step": 18375
    },
    {
      "epoch": 0.18697102864583334,
      "grad_norm": 12.728557586669922,
      "learning_rate": 4.584575743131142e-06,
      "loss": 3.6974,
      "step": 18380
    },
    {
      "epoch": 0.18702189127604166,
      "grad_norm": 12.671290397644043,
      "learning_rate": 4.584354973841103e-06,
      "loss": 3.343,
      "step": 18385
    },
    {
      "epoch": 0.18707275390625,
      "grad_norm": 14.869614601135254,
      "learning_rate": 4.584134151223335e-06,
      "loss": 3.4718,
      "step": 18390
    },
    {
      "epoch": 0.18712361653645834,
      "grad_norm": 11.287585258483887,
      "learning_rate": 4.58391327528349e-06,
      "loss": 3.4128,
      "step": 18395
    },
    {
      "epoch": 0.18717447916666666,
      "grad_norm": 16.449617385864258,
      "learning_rate": 4.5836923460272175e-06,
      "loss": 3.7161,
      "step": 18400
    },
    {
      "epoch": 0.187225341796875,
      "grad_norm": 16.260473251342773,
      "learning_rate": 4.58347136346017e-06,
      "loss": 3.3264,
      "step": 18405
    },
    {
      "epoch": 0.18727620442708334,
      "grad_norm": 13.514466285705566,
      "learning_rate": 4.5832503275880015e-06,
      "loss": 3.294,
      "step": 18410
    },
    {
      "epoch": 0.18732706705729166,
      "grad_norm": 13.021100044250488,
      "learning_rate": 4.583029238416368e-06,
      "loss": 3.3275,
      "step": 18415
    },
    {
      "epoch": 0.1873779296875,
      "grad_norm": 10.005925178527832,
      "learning_rate": 4.582808095950924e-06,
      "loss": 3.308,
      "step": 18420
    },
    {
      "epoch": 0.18742879231770834,
      "grad_norm": 11.48652172088623,
      "learning_rate": 4.5825869001973285e-06,
      "loss": 3.5553,
      "step": 18425
    },
    {
      "epoch": 0.18747965494791666,
      "grad_norm": 15.618109703063965,
      "learning_rate": 4.582365651161242e-06,
      "loss": 3.3904,
      "step": 18430
    },
    {
      "epoch": 0.187530517578125,
      "grad_norm": 14.050233840942383,
      "learning_rate": 4.582144348848323e-06,
      "loss": 3.7672,
      "step": 18435
    },
    {
      "epoch": 0.18758138020833334,
      "grad_norm": 10.15687370300293,
      "learning_rate": 4.581922993264235e-06,
      "loss": 3.2861,
      "step": 18440
    },
    {
      "epoch": 0.18763224283854166,
      "grad_norm": 10.453628540039062,
      "learning_rate": 4.58170158441464e-06,
      "loss": 3.1252,
      "step": 18445
    },
    {
      "epoch": 0.18768310546875,
      "grad_norm": 16.22777557373047,
      "learning_rate": 4.5814801223052035e-06,
      "loss": 3.43,
      "step": 18450
    },
    {
      "epoch": 0.18773396809895834,
      "grad_norm": 9.1096830368042,
      "learning_rate": 4.581258606941592e-06,
      "loss": 3.3274,
      "step": 18455
    },
    {
      "epoch": 0.18778483072916666,
      "grad_norm": 16.2208251953125,
      "learning_rate": 4.581037038329472e-06,
      "loss": 3.8525,
      "step": 18460
    },
    {
      "epoch": 0.187835693359375,
      "grad_norm": 16.626689910888672,
      "learning_rate": 4.580815416474512e-06,
      "loss": 3.2766,
      "step": 18465
    },
    {
      "epoch": 0.18788655598958334,
      "grad_norm": 16.821229934692383,
      "learning_rate": 4.580593741382384e-06,
      "loss": 3.6567,
      "step": 18470
    },
    {
      "epoch": 0.18793741861979166,
      "grad_norm": 11.31299877166748,
      "learning_rate": 4.580372013058757e-06,
      "loss": 3.4805,
      "step": 18475
    },
    {
      "epoch": 0.18798828125,
      "grad_norm": 16.890230178833008,
      "learning_rate": 4.580150231509306e-06,
      "loss": 3.2733,
      "step": 18480
    },
    {
      "epoch": 0.18803914388020834,
      "grad_norm": 8.345052719116211,
      "learning_rate": 4.579928396739704e-06,
      "loss": 3.7072,
      "step": 18485
    },
    {
      "epoch": 0.18809000651041666,
      "grad_norm": 7.271616458892822,
      "learning_rate": 4.579706508755627e-06,
      "loss": 3.4539,
      "step": 18490
    },
    {
      "epoch": 0.188140869140625,
      "grad_norm": 14.4603910446167,
      "learning_rate": 4.579484567562752e-06,
      "loss": 3.8369,
      "step": 18495
    },
    {
      "epoch": 0.18819173177083334,
      "grad_norm": 8.192587852478027,
      "learning_rate": 4.579262573166757e-06,
      "loss": 2.9722,
      "step": 18500
    },
    {
      "epoch": 0.18824259440104166,
      "grad_norm": 8.657429695129395,
      "learning_rate": 4.579040525573323e-06,
      "loss": 3.7435,
      "step": 18505
    },
    {
      "epoch": 0.18829345703125,
      "grad_norm": 11.86208724975586,
      "learning_rate": 4.578818424788129e-06,
      "loss": 3.3249,
      "step": 18510
    },
    {
      "epoch": 0.18834431966145834,
      "grad_norm": 11.226592063903809,
      "learning_rate": 4.578596270816858e-06,
      "loss": 3.4405,
      "step": 18515
    },
    {
      "epoch": 0.18839518229166666,
      "grad_norm": 7.9588727951049805,
      "learning_rate": 4.578374063665195e-06,
      "loss": 3.378,
      "step": 18520
    },
    {
      "epoch": 0.188446044921875,
      "grad_norm": 10.29086971282959,
      "learning_rate": 4.578151803338824e-06,
      "loss": 3.6522,
      "step": 18525
    },
    {
      "epoch": 0.18849690755208334,
      "grad_norm": 12.614090919494629,
      "learning_rate": 4.577929489843431e-06,
      "loss": 3.0822,
      "step": 18530
    },
    {
      "epoch": 0.18854777018229166,
      "grad_norm": 14.052781105041504,
      "learning_rate": 4.577707123184705e-06,
      "loss": 3.3621,
      "step": 18535
    },
    {
      "epoch": 0.1885986328125,
      "grad_norm": 13.460591316223145,
      "learning_rate": 4.577484703368335e-06,
      "loss": 3.5421,
      "step": 18540
    },
    {
      "epoch": 0.18864949544270834,
      "grad_norm": 10.110848426818848,
      "learning_rate": 4.57726223040001e-06,
      "loss": 3.5516,
      "step": 18545
    },
    {
      "epoch": 0.18870035807291666,
      "grad_norm": 11.632022857666016,
      "learning_rate": 4.577039704285424e-06,
      "loss": 3.5707,
      "step": 18550
    },
    {
      "epoch": 0.188751220703125,
      "grad_norm": 8.971404075622559,
      "learning_rate": 4.5768171250302706e-06,
      "loss": 3.4157,
      "step": 18555
    },
    {
      "epoch": 0.18880208333333334,
      "grad_norm": 14.906126976013184,
      "learning_rate": 4.576594492640242e-06,
      "loss": 3.7568,
      "step": 18560
    },
    {
      "epoch": 0.18885294596354166,
      "grad_norm": 12.4305419921875,
      "learning_rate": 4.576371807121036e-06,
      "loss": 3.5414,
      "step": 18565
    },
    {
      "epoch": 0.18890380859375,
      "grad_norm": 13.76634407043457,
      "learning_rate": 4.57614906847835e-06,
      "loss": 3.4253,
      "step": 18570
    },
    {
      "epoch": 0.18895467122395834,
      "grad_norm": 14.682574272155762,
      "learning_rate": 4.5759262767178805e-06,
      "loss": 3.4316,
      "step": 18575
    },
    {
      "epoch": 0.18900553385416666,
      "grad_norm": 12.575339317321777,
      "learning_rate": 4.575703431845331e-06,
      "loss": 3.6921,
      "step": 18580
    },
    {
      "epoch": 0.189056396484375,
      "grad_norm": 7.278709411621094,
      "learning_rate": 4.5754805338664e-06,
      "loss": 3.1521,
      "step": 18585
    },
    {
      "epoch": 0.18910725911458334,
      "grad_norm": 12.746679306030273,
      "learning_rate": 4.575257582786792e-06,
      "loss": 3.2062,
      "step": 18590
    },
    {
      "epoch": 0.18915812174479166,
      "grad_norm": 15.597989082336426,
      "learning_rate": 4.575034578612211e-06,
      "loss": 3.5262,
      "step": 18595
    },
    {
      "epoch": 0.189208984375,
      "grad_norm": 7.950192928314209,
      "learning_rate": 4.574811521348361e-06,
      "loss": 3.6785,
      "step": 18600
    },
    {
      "epoch": 0.18925984700520834,
      "grad_norm": 10.346031188964844,
      "learning_rate": 4.574588411000951e-06,
      "loss": 3.3503,
      "step": 18605
    },
    {
      "epoch": 0.18931070963541666,
      "grad_norm": 11.723124504089355,
      "learning_rate": 4.574365247575688e-06,
      "loss": 3.3964,
      "step": 18610
    },
    {
      "epoch": 0.189361572265625,
      "grad_norm": 8.930464744567871,
      "learning_rate": 4.574142031078282e-06,
      "loss": 3.2352,
      "step": 18615
    },
    {
      "epoch": 0.18941243489583334,
      "grad_norm": 13.8888521194458,
      "learning_rate": 4.573918761514443e-06,
      "loss": 3.315,
      "step": 18620
    },
    {
      "epoch": 0.18946329752604166,
      "grad_norm": 7.327556133270264,
      "learning_rate": 4.573695438889885e-06,
      "loss": 3.1141,
      "step": 18625
    },
    {
      "epoch": 0.18951416015625,
      "grad_norm": 8.570147514343262,
      "learning_rate": 4.57347206321032e-06,
      "loss": 3.1851,
      "step": 18630
    },
    {
      "epoch": 0.18956502278645834,
      "grad_norm": 9.34282112121582,
      "learning_rate": 4.573248634481464e-06,
      "loss": 3.5862,
      "step": 18635
    },
    {
      "epoch": 0.18961588541666666,
      "grad_norm": 12.512747764587402,
      "learning_rate": 4.573025152709033e-06,
      "loss": 3.5737,
      "step": 18640
    },
    {
      "epoch": 0.189666748046875,
      "grad_norm": 12.564814567565918,
      "learning_rate": 4.572801617898746e-06,
      "loss": 3.4582,
      "step": 18645
    },
    {
      "epoch": 0.18971761067708334,
      "grad_norm": 14.594040870666504,
      "learning_rate": 4.572578030056319e-06,
      "loss": 3.0915,
      "step": 18650
    },
    {
      "epoch": 0.18976847330729166,
      "grad_norm": 11.229549407958984,
      "learning_rate": 4.572354389187476e-06,
      "loss": 3.6208,
      "step": 18655
    },
    {
      "epoch": 0.1898193359375,
      "grad_norm": 13.392086029052734,
      "learning_rate": 4.572130695297936e-06,
      "loss": 3.5334,
      "step": 18660
    },
    {
      "epoch": 0.18987019856770834,
      "grad_norm": 9.601693153381348,
      "learning_rate": 4.571906948393424e-06,
      "loss": 3.3134,
      "step": 18665
    },
    {
      "epoch": 0.18992106119791666,
      "grad_norm": 13.067540168762207,
      "learning_rate": 4.571683148479663e-06,
      "loss": 3.0544,
      "step": 18670
    },
    {
      "epoch": 0.189971923828125,
      "grad_norm": 10.653081893920898,
      "learning_rate": 4.57145929556238e-06,
      "loss": 3.7998,
      "step": 18675
    },
    {
      "epoch": 0.19002278645833334,
      "grad_norm": 12.351947784423828,
      "learning_rate": 4.5712353896473025e-06,
      "loss": 3.3205,
      "step": 18680
    },
    {
      "epoch": 0.19007364908854166,
      "grad_norm": 9.99809455871582,
      "learning_rate": 4.571011430740158e-06,
      "loss": 3.577,
      "step": 18685
    },
    {
      "epoch": 0.19012451171875,
      "grad_norm": 11.64851188659668,
      "learning_rate": 4.5707874188466774e-06,
      "loss": 3.3207,
      "step": 18690
    },
    {
      "epoch": 0.19017537434895834,
      "grad_norm": 10.789388656616211,
      "learning_rate": 4.5705633539725915e-06,
      "loss": 3.5073,
      "step": 18695
    },
    {
      "epoch": 0.19022623697916666,
      "grad_norm": 11.41800308227539,
      "learning_rate": 4.570339236123632e-06,
      "loss": 3.4615,
      "step": 18700
    },
    {
      "epoch": 0.190277099609375,
      "grad_norm": 12.321640968322754,
      "learning_rate": 4.5701150653055345e-06,
      "loss": 3.3763,
      "step": 18705
    },
    {
      "epoch": 0.19032796223958334,
      "grad_norm": 14.531020164489746,
      "learning_rate": 4.569890841524034e-06,
      "loss": 3.5657,
      "step": 18710
    },
    {
      "epoch": 0.19037882486979166,
      "grad_norm": 10.604043960571289,
      "learning_rate": 4.569666564784867e-06,
      "loss": 3.4327,
      "step": 18715
    },
    {
      "epoch": 0.1904296875,
      "grad_norm": 9.537259101867676,
      "learning_rate": 4.569442235093771e-06,
      "loss": 3.3424,
      "step": 18720
    },
    {
      "epoch": 0.19048055013020834,
      "grad_norm": 10.841087341308594,
      "learning_rate": 4.569217852456486e-06,
      "loss": 3.4501,
      "step": 18725
    },
    {
      "epoch": 0.19053141276041666,
      "grad_norm": 10.725885391235352,
      "learning_rate": 4.568993416878753e-06,
      "loss": 3.6736,
      "step": 18730
    },
    {
      "epoch": 0.190582275390625,
      "grad_norm": 13.705768585205078,
      "learning_rate": 4.568768928366313e-06,
      "loss": 3.7702,
      "step": 18735
    },
    {
      "epoch": 0.19063313802083334,
      "grad_norm": 8.565932273864746,
      "learning_rate": 4.568544386924911e-06,
      "loss": 3.6213,
      "step": 18740
    },
    {
      "epoch": 0.19068400065104166,
      "grad_norm": 14.157207489013672,
      "learning_rate": 4.568319792560292e-06,
      "loss": 3.4415,
      "step": 18745
    },
    {
      "epoch": 0.19073486328125,
      "grad_norm": 12.56330680847168,
      "learning_rate": 4.5680951452782e-06,
      "loss": 3.1235,
      "step": 18750
    },
    {
      "epoch": 0.19078572591145834,
      "grad_norm": 7.816888809204102,
      "learning_rate": 4.567870445084385e-06,
      "loss": 3.1856,
      "step": 18755
    },
    {
      "epoch": 0.19083658854166666,
      "grad_norm": 13.05587387084961,
      "learning_rate": 4.567645691984594e-06,
      "loss": 3.1723,
      "step": 18760
    },
    {
      "epoch": 0.190887451171875,
      "grad_norm": 13.687058448791504,
      "learning_rate": 4.567420885984578e-06,
      "loss": 3.8625,
      "step": 18765
    },
    {
      "epoch": 0.19093831380208334,
      "grad_norm": 12.832854270935059,
      "learning_rate": 4.567196027090088e-06,
      "loss": 3.1376,
      "step": 18770
    },
    {
      "epoch": 0.19098917643229166,
      "grad_norm": 13.068547248840332,
      "learning_rate": 4.56697111530688e-06,
      "loss": 3.3375,
      "step": 18775
    },
    {
      "epoch": 0.1910400390625,
      "grad_norm": 10.62330150604248,
      "learning_rate": 4.566746150640704e-06,
      "loss": 3.4257,
      "step": 18780
    },
    {
      "epoch": 0.19109090169270834,
      "grad_norm": 13.381546020507812,
      "learning_rate": 4.566521133097318e-06,
      "loss": 3.2436,
      "step": 18785
    },
    {
      "epoch": 0.19114176432291666,
      "grad_norm": 9.422579765319824,
      "learning_rate": 4.566296062682478e-06,
      "loss": 3.4788,
      "step": 18790
    },
    {
      "epoch": 0.191192626953125,
      "grad_norm": 16.37440299987793,
      "learning_rate": 4.566070939401944e-06,
      "loss": 3.5952,
      "step": 18795
    },
    {
      "epoch": 0.19124348958333334,
      "grad_norm": 11.154594421386719,
      "learning_rate": 4.565845763261475e-06,
      "loss": 3.1788,
      "step": 18800
    },
    {
      "epoch": 0.19129435221354166,
      "grad_norm": 13.181290626525879,
      "learning_rate": 4.565620534266831e-06,
      "loss": 3.2884,
      "step": 18805
    },
    {
      "epoch": 0.19134521484375,
      "grad_norm": 10.758809089660645,
      "learning_rate": 4.565395252423775e-06,
      "loss": 3.2639,
      "step": 18810
    },
    {
      "epoch": 0.19139607747395834,
      "grad_norm": 11.981339454650879,
      "learning_rate": 4.5651699177380714e-06,
      "loss": 3.2688,
      "step": 18815
    },
    {
      "epoch": 0.19144694010416666,
      "grad_norm": 9.572751998901367,
      "learning_rate": 4.564944530215486e-06,
      "loss": 3.5481,
      "step": 18820
    },
    {
      "epoch": 0.191497802734375,
      "grad_norm": 9.87020492553711,
      "learning_rate": 4.564719089861783e-06,
      "loss": 3.4662,
      "step": 18825
    },
    {
      "epoch": 0.19154866536458334,
      "grad_norm": 9.3651123046875,
      "learning_rate": 4.564493596682732e-06,
      "loss": 3.2547,
      "step": 18830
    },
    {
      "epoch": 0.19159952799479166,
      "grad_norm": 9.596661567687988,
      "learning_rate": 4.564268050684101e-06,
      "loss": 3.3904,
      "step": 18835
    },
    {
      "epoch": 0.191650390625,
      "grad_norm": 15.922765731811523,
      "learning_rate": 4.564042451871662e-06,
      "loss": 3.2157,
      "step": 18840
    },
    {
      "epoch": 0.19170125325520834,
      "grad_norm": 11.004732131958008,
      "learning_rate": 4.563816800251185e-06,
      "loss": 3.3153,
      "step": 18845
    },
    {
      "epoch": 0.19175211588541666,
      "grad_norm": 8.292093276977539,
      "learning_rate": 4.563591095828446e-06,
      "loss": 3.2079,
      "step": 18850
    },
    {
      "epoch": 0.191802978515625,
      "grad_norm": 8.698326110839844,
      "learning_rate": 4.563365338609216e-06,
      "loss": 3.404,
      "step": 18855
    },
    {
      "epoch": 0.19185384114583334,
      "grad_norm": 11.080509185791016,
      "learning_rate": 4.563139528599274e-06,
      "loss": 3.2749,
      "step": 18860
    },
    {
      "epoch": 0.19190470377604166,
      "grad_norm": 14.920134544372559,
      "learning_rate": 4.562913665804397e-06,
      "loss": 3.455,
      "step": 18865
    },
    {
      "epoch": 0.19195556640625,
      "grad_norm": 10.069123268127441,
      "learning_rate": 4.562687750230361e-06,
      "loss": 3.6829,
      "step": 18870
    },
    {
      "epoch": 0.19200642903645834,
      "grad_norm": 8.710797309875488,
      "learning_rate": 4.562461781882949e-06,
      "loss": 3.3883,
      "step": 18875
    },
    {
      "epoch": 0.19205729166666666,
      "grad_norm": 13.156225204467773,
      "learning_rate": 4.5622357607679415e-06,
      "loss": 3.1545,
      "step": 18880
    },
    {
      "epoch": 0.192108154296875,
      "grad_norm": 12.595132827758789,
      "learning_rate": 4.5620096868911205e-06,
      "loss": 3.4499,
      "step": 18885
    },
    {
      "epoch": 0.19215901692708334,
      "grad_norm": 12.80722713470459,
      "learning_rate": 4.561783560258269e-06,
      "loss": 3.6053,
      "step": 18890
    },
    {
      "epoch": 0.19220987955729166,
      "grad_norm": 10.552962303161621,
      "learning_rate": 4.561557380875175e-06,
      "loss": 3.4,
      "step": 18895
    },
    {
      "epoch": 0.1922607421875,
      "grad_norm": 12.450384140014648,
      "learning_rate": 4.561331148747624e-06,
      "loss": 3.4466,
      "step": 18900
    },
    {
      "epoch": 0.19231160481770834,
      "grad_norm": 9.195076942443848,
      "learning_rate": 4.5611048638814045e-06,
      "loss": 3.0326,
      "step": 18905
    },
    {
      "epoch": 0.19236246744791666,
      "grad_norm": 10.586387634277344,
      "learning_rate": 4.560878526282305e-06,
      "loss": 3.1759,
      "step": 18910
    },
    {
      "epoch": 0.192413330078125,
      "grad_norm": 17.3674259185791,
      "learning_rate": 4.560652135956117e-06,
      "loss": 3.8675,
      "step": 18915
    },
    {
      "epoch": 0.19246419270833334,
      "grad_norm": 8.781184196472168,
      "learning_rate": 4.5604256929086334e-06,
      "loss": 3.292,
      "step": 18920
    },
    {
      "epoch": 0.19251505533854166,
      "grad_norm": 9.68058967590332,
      "learning_rate": 4.560199197145646e-06,
      "loss": 3.3997,
      "step": 18925
    },
    {
      "epoch": 0.19256591796875,
      "grad_norm": 8.510169982910156,
      "learning_rate": 4.559972648672951e-06,
      "loss": 3.4601,
      "step": 18930
    },
    {
      "epoch": 0.19261678059895834,
      "grad_norm": 13.213333129882812,
      "learning_rate": 4.559746047496343e-06,
      "loss": 3.2534,
      "step": 18935
    },
    {
      "epoch": 0.19266764322916666,
      "grad_norm": 13.007706642150879,
      "learning_rate": 4.559519393621621e-06,
      "loss": 3.3954,
      "step": 18940
    },
    {
      "epoch": 0.192718505859375,
      "grad_norm": 14.338438987731934,
      "learning_rate": 4.559292687054584e-06,
      "loss": 3.1865,
      "step": 18945
    },
    {
      "epoch": 0.19276936848958334,
      "grad_norm": 12.187652587890625,
      "learning_rate": 4.559065927801032e-06,
      "loss": 3.03,
      "step": 18950
    },
    {
      "epoch": 0.19282023111979166,
      "grad_norm": 12.161337852478027,
      "learning_rate": 4.558839115866765e-06,
      "loss": 3.426,
      "step": 18955
    },
    {
      "epoch": 0.19287109375,
      "grad_norm": 10.178563117980957,
      "learning_rate": 4.558612251257588e-06,
      "loss": 3.2395,
      "step": 18960
    },
    {
      "epoch": 0.19292195638020834,
      "grad_norm": 9.272007942199707,
      "learning_rate": 4.558385333979304e-06,
      "loss": 3.7845,
      "step": 18965
    },
    {
      "epoch": 0.19297281901041666,
      "grad_norm": 13.764449119567871,
      "learning_rate": 4.55815836403772e-06,
      "loss": 3.1891,
      "step": 18970
    },
    {
      "epoch": 0.193023681640625,
      "grad_norm": 13.366809844970703,
      "learning_rate": 4.557931341438642e-06,
      "loss": 3.2882,
      "step": 18975
    },
    {
      "epoch": 0.19307454427083334,
      "grad_norm": 11.97940731048584,
      "learning_rate": 4.557704266187878e-06,
      "loss": 3.0493,
      "step": 18980
    },
    {
      "epoch": 0.19312540690104166,
      "grad_norm": 15.40910816192627,
      "learning_rate": 4.557477138291238e-06,
      "loss": 3.4534,
      "step": 18985
    },
    {
      "epoch": 0.19317626953125,
      "grad_norm": 9.693882942199707,
      "learning_rate": 4.557249957754533e-06,
      "loss": 3.628,
      "step": 18990
    },
    {
      "epoch": 0.19322713216145834,
      "grad_norm": 9.635483741760254,
      "learning_rate": 4.557022724583575e-06,
      "loss": 3.3948,
      "step": 18995
    },
    {
      "epoch": 0.19327799479166666,
      "grad_norm": 8.52334213256836,
      "learning_rate": 4.55679543878418e-06,
      "loss": 3.5678,
      "step": 19000
    },
    {
      "epoch": 0.193328857421875,
      "grad_norm": 10.489480018615723,
      "learning_rate": 4.5565681003621596e-06,
      "loss": 3.2659,
      "step": 19005
    },
    {
      "epoch": 0.19337972005208334,
      "grad_norm": 10.904191970825195,
      "learning_rate": 4.556340709323333e-06,
      "loss": 3.7266,
      "step": 19010
    },
    {
      "epoch": 0.19343058268229166,
      "grad_norm": 9.743817329406738,
      "learning_rate": 4.556113265673515e-06,
      "loss": 3.8189,
      "step": 19015
    },
    {
      "epoch": 0.1934814453125,
      "grad_norm": 16.735994338989258,
      "learning_rate": 4.555885769418528e-06,
      "loss": 3.6075,
      "step": 19020
    },
    {
      "epoch": 0.19353230794270834,
      "grad_norm": 13.078174591064453,
      "learning_rate": 4.55565822056419e-06,
      "loss": 3.3227,
      "step": 19025
    },
    {
      "epoch": 0.19358317057291666,
      "grad_norm": 6.742351055145264,
      "learning_rate": 4.555430619116324e-06,
      "loss": 3.2009,
      "step": 19030
    },
    {
      "epoch": 0.193634033203125,
      "grad_norm": 11.052045822143555,
      "learning_rate": 4.555202965080752e-06,
      "loss": 3.5503,
      "step": 19035
    },
    {
      "epoch": 0.19368489583333334,
      "grad_norm": 12.830464363098145,
      "learning_rate": 4.554975258463301e-06,
      "loss": 3.4425,
      "step": 19040
    },
    {
      "epoch": 0.19373575846354166,
      "grad_norm": 14.631771087646484,
      "learning_rate": 4.554747499269794e-06,
      "loss": 2.9493,
      "step": 19045
    },
    {
      "epoch": 0.19378662109375,
      "grad_norm": 12.542807579040527,
      "learning_rate": 4.554519687506059e-06,
      "loss": 3.3026,
      "step": 19050
    },
    {
      "epoch": 0.19383748372395834,
      "grad_norm": 12.711750030517578,
      "learning_rate": 4.5542918231779255e-06,
      "loss": 3.3107,
      "step": 19055
    },
    {
      "epoch": 0.19388834635416666,
      "grad_norm": 15.20833969116211,
      "learning_rate": 4.554063906291222e-06,
      "loss": 3.2423,
      "step": 19060
    },
    {
      "epoch": 0.193939208984375,
      "grad_norm": 11.543058395385742,
      "learning_rate": 4.553835936851781e-06,
      "loss": 3.4903,
      "step": 19065
    },
    {
      "epoch": 0.19399007161458334,
      "grad_norm": 12.416997909545898,
      "learning_rate": 4.553607914865434e-06,
      "loss": 3.3484,
      "step": 19070
    },
    {
      "epoch": 0.19404093424479166,
      "grad_norm": 15.233044624328613,
      "learning_rate": 4.553379840338015e-06,
      "loss": 3.187,
      "step": 19075
    },
    {
      "epoch": 0.194091796875,
      "grad_norm": 16.927047729492188,
      "learning_rate": 4.55315171327536e-06,
      "loss": 3.4679,
      "step": 19080
    },
    {
      "epoch": 0.19414265950520834,
      "grad_norm": 13.578765869140625,
      "learning_rate": 4.552923533683305e-06,
      "loss": 2.9851,
      "step": 19085
    },
    {
      "epoch": 0.19419352213541666,
      "grad_norm": 12.253621101379395,
      "learning_rate": 4.552695301567688e-06,
      "loss": 3.3136,
      "step": 19090
    },
    {
      "epoch": 0.194244384765625,
      "grad_norm": 9.944585800170898,
      "learning_rate": 4.5524670169343475e-06,
      "loss": 3.7707,
      "step": 19095
    },
    {
      "epoch": 0.19429524739583334,
      "grad_norm": 13.689692497253418,
      "learning_rate": 4.552238679789126e-06,
      "loss": 3.6948,
      "step": 19100
    },
    {
      "epoch": 0.19434611002604166,
      "grad_norm": 11.918596267700195,
      "learning_rate": 4.5520102901378636e-06,
      "loss": 3.4049,
      "step": 19105
    },
    {
      "epoch": 0.19439697265625,
      "grad_norm": 14.738883972167969,
      "learning_rate": 4.551781847986404e-06,
      "loss": 3.6657,
      "step": 19110
    },
    {
      "epoch": 0.19444783528645834,
      "grad_norm": 8.853859901428223,
      "learning_rate": 4.551553353340593e-06,
      "loss": 3.5498,
      "step": 19115
    },
    {
      "epoch": 0.19449869791666666,
      "grad_norm": 11.349398612976074,
      "learning_rate": 4.551324806206276e-06,
      "loss": 3.5923,
      "step": 19120
    },
    {
      "epoch": 0.194549560546875,
      "grad_norm": 11.168492317199707,
      "learning_rate": 4.551096206589298e-06,
      "loss": 3.4872,
      "step": 19125
    },
    {
      "epoch": 0.19460042317708334,
      "grad_norm": 6.9756574630737305,
      "learning_rate": 4.5508675544955115e-06,
      "loss": 3.0787,
      "step": 19130
    },
    {
      "epoch": 0.19465128580729166,
      "grad_norm": 9.951805114746094,
      "learning_rate": 4.550638849930764e-06,
      "loss": 3.2865,
      "step": 19135
    },
    {
      "epoch": 0.1947021484375,
      "grad_norm": 9.90751838684082,
      "learning_rate": 4.550410092900908e-06,
      "loss": 3.3865,
      "step": 19140
    },
    {
      "epoch": 0.19475301106770834,
      "grad_norm": 12.605509757995605,
      "learning_rate": 4.550181283411796e-06,
      "loss": 3.3004,
      "step": 19145
    },
    {
      "epoch": 0.19480387369791666,
      "grad_norm": 10.202880859375,
      "learning_rate": 4.549952421469281e-06,
      "loss": 3.3373,
      "step": 19150
    },
    {
      "epoch": 0.194854736328125,
      "grad_norm": 9.19309139251709,
      "learning_rate": 4.54972350707922e-06,
      "loss": 2.9459,
      "step": 19155
    },
    {
      "epoch": 0.19490559895833334,
      "grad_norm": 12.980036735534668,
      "learning_rate": 4.549494540247469e-06,
      "loss": 3.4799,
      "step": 19160
    },
    {
      "epoch": 0.19495646158854166,
      "grad_norm": 12.7277250289917,
      "learning_rate": 4.549265520979885e-06,
      "loss": 3.0953,
      "step": 19165
    },
    {
      "epoch": 0.19500732421875,
      "grad_norm": 13.657605171203613,
      "learning_rate": 4.549036449282329e-06,
      "loss": 3.1377,
      "step": 19170
    },
    {
      "epoch": 0.19505818684895834,
      "grad_norm": 15.17473316192627,
      "learning_rate": 4.5488073251606616e-06,
      "loss": 3.1005,
      "step": 19175
    },
    {
      "epoch": 0.19510904947916666,
      "grad_norm": 13.394558906555176,
      "learning_rate": 4.548578148620744e-06,
      "loss": 3.162,
      "step": 19180
    },
    {
      "epoch": 0.195159912109375,
      "grad_norm": 11.801448822021484,
      "learning_rate": 4.54834891966844e-06,
      "loss": 3.1549,
      "step": 19185
    },
    {
      "epoch": 0.19521077473958334,
      "grad_norm": 13.888011932373047,
      "learning_rate": 4.548119638309613e-06,
      "loss": 3.4705,
      "step": 19190
    },
    {
      "epoch": 0.19526163736979166,
      "grad_norm": 15.35958194732666,
      "learning_rate": 4.547890304550132e-06,
      "loss": 3.7615,
      "step": 19195
    },
    {
      "epoch": 0.1953125,
      "grad_norm": 14.944520950317383,
      "learning_rate": 4.547660918395863e-06,
      "loss": 3.6172,
      "step": 19200
    },
    {
      "epoch": 0.19536336263020834,
      "grad_norm": 9.878191947937012,
      "learning_rate": 4.547431479852674e-06,
      "loss": 3.1867,
      "step": 19205
    },
    {
      "epoch": 0.19541422526041666,
      "grad_norm": 11.971900939941406,
      "learning_rate": 4.5472019889264375e-06,
      "loss": 3.1045,
      "step": 19210
    },
    {
      "epoch": 0.195465087890625,
      "grad_norm": 12.584593772888184,
      "learning_rate": 4.546972445623021e-06,
      "loss": 3.3828,
      "step": 19215
    },
    {
      "epoch": 0.19551595052083334,
      "grad_norm": 11.079330444335938,
      "learning_rate": 4.546742849948302e-06,
      "loss": 3.579,
      "step": 19220
    },
    {
      "epoch": 0.19556681315104166,
      "grad_norm": 11.899502754211426,
      "learning_rate": 4.54651320190815e-06,
      "loss": 3.4399,
      "step": 19225
    },
    {
      "epoch": 0.19561767578125,
      "grad_norm": 10.531402587890625,
      "learning_rate": 4.546283501508445e-06,
      "loss": 3.4826,
      "step": 19230
    },
    {
      "epoch": 0.19566853841145834,
      "grad_norm": 13.053157806396484,
      "learning_rate": 4.54605374875506e-06,
      "loss": 3.4088,
      "step": 19235
    },
    {
      "epoch": 0.19571940104166666,
      "grad_norm": 8.246091842651367,
      "learning_rate": 4.545823943653876e-06,
      "loss": 3.4349,
      "step": 19240
    },
    {
      "epoch": 0.195770263671875,
      "grad_norm": 8.850236892700195,
      "learning_rate": 4.545594086210771e-06,
      "loss": 3.022,
      "step": 19245
    },
    {
      "epoch": 0.19582112630208334,
      "grad_norm": 14.258888244628906,
      "learning_rate": 4.545364176431626e-06,
      "loss": 3.1912,
      "step": 19250
    },
    {
      "epoch": 0.19587198893229166,
      "grad_norm": 9.115057945251465,
      "learning_rate": 4.545134214322323e-06,
      "loss": 3.1928,
      "step": 19255
    },
    {
      "epoch": 0.1959228515625,
      "grad_norm": 12.057110786437988,
      "learning_rate": 4.544904199888746e-06,
      "loss": 3.3195,
      "step": 19260
    },
    {
      "epoch": 0.19597371419270834,
      "grad_norm": 8.72459888458252,
      "learning_rate": 4.54467413313678e-06,
      "loss": 3.5811,
      "step": 19265
    },
    {
      "epoch": 0.19602457682291666,
      "grad_norm": 7.331991195678711,
      "learning_rate": 4.544444014072311e-06,
      "loss": 3.191,
      "step": 19270
    },
    {
      "epoch": 0.196075439453125,
      "grad_norm": 13.493756294250488,
      "learning_rate": 4.544213842701226e-06,
      "loss": 2.8969,
      "step": 19275
    },
    {
      "epoch": 0.19612630208333334,
      "grad_norm": 13.320565223693848,
      "learning_rate": 4.543983619029415e-06,
      "loss": 3.3989,
      "step": 19280
    },
    {
      "epoch": 0.19617716471354166,
      "grad_norm": 12.885554313659668,
      "learning_rate": 4.543753343062767e-06,
      "loss": 3.4129,
      "step": 19285
    },
    {
      "epoch": 0.19622802734375,
      "grad_norm": 7.874749183654785,
      "learning_rate": 4.543523014807175e-06,
      "loss": 3.277,
      "step": 19290
    },
    {
      "epoch": 0.19627888997395834,
      "grad_norm": 11.990129470825195,
      "learning_rate": 4.54329263426853e-06,
      "loss": 3.3223,
      "step": 19295
    },
    {
      "epoch": 0.19632975260416666,
      "grad_norm": 14.290430068969727,
      "learning_rate": 4.543062201452728e-06,
      "loss": 3.366,
      "step": 19300
    },
    {
      "epoch": 0.196380615234375,
      "grad_norm": 12.931253433227539,
      "learning_rate": 4.5428317163656635e-06,
      "loss": 3.4963,
      "step": 19305
    },
    {
      "epoch": 0.19643147786458334,
      "grad_norm": 10.713802337646484,
      "learning_rate": 4.542601179013234e-06,
      "loss": 3.2372,
      "step": 19310
    },
    {
      "epoch": 0.19648234049479166,
      "grad_norm": 10.50666332244873,
      "learning_rate": 4.542370589401337e-06,
      "loss": 3.3301,
      "step": 19315
    },
    {
      "epoch": 0.196533203125,
      "grad_norm": 7.86500883102417,
      "learning_rate": 4.542139947535873e-06,
      "loss": 3.2692,
      "step": 19320
    },
    {
      "epoch": 0.19658406575520834,
      "grad_norm": 15.505828857421875,
      "learning_rate": 4.541909253422742e-06,
      "loss": 3.5111,
      "step": 19325
    },
    {
      "epoch": 0.19663492838541666,
      "grad_norm": 15.200455665588379,
      "learning_rate": 4.5416785070678475e-06,
      "loss": 3.5127,
      "step": 19330
    },
    {
      "epoch": 0.196685791015625,
      "grad_norm": 8.061166763305664,
      "learning_rate": 4.541447708477091e-06,
      "loss": 3.558,
      "step": 19335
    },
    {
      "epoch": 0.19673665364583334,
      "grad_norm": 15.958837509155273,
      "learning_rate": 4.5412168576563805e-06,
      "loss": 3.1756,
      "step": 19340
    },
    {
      "epoch": 0.19678751627604166,
      "grad_norm": 7.027966022491455,
      "learning_rate": 4.5409859546116195e-06,
      "loss": 3.2773,
      "step": 19345
    },
    {
      "epoch": 0.19683837890625,
      "grad_norm": 15.966991424560547,
      "learning_rate": 4.540754999348716e-06,
      "loss": 3.443,
      "step": 19350
    },
    {
      "epoch": 0.19688924153645834,
      "grad_norm": 10.5971040725708,
      "learning_rate": 4.540523991873581e-06,
      "loss": 3.3096,
      "step": 19355
    },
    {
      "epoch": 0.19694010416666666,
      "grad_norm": 15.873735427856445,
      "learning_rate": 4.540292932192123e-06,
      "loss": 3.1159,
      "step": 19360
    },
    {
      "epoch": 0.196990966796875,
      "grad_norm": 11.828910827636719,
      "learning_rate": 4.540061820310253e-06,
      "loss": 3.63,
      "step": 19365
    },
    {
      "epoch": 0.19704182942708334,
      "grad_norm": 13.37854290008545,
      "learning_rate": 4.539830656233886e-06,
      "loss": 2.9163,
      "step": 19370
    },
    {
      "epoch": 0.19709269205729166,
      "grad_norm": 12.53745174407959,
      "learning_rate": 4.5395994399689344e-06,
      "loss": 3.7792,
      "step": 19375
    },
    {
      "epoch": 0.1971435546875,
      "grad_norm": 14.388495445251465,
      "learning_rate": 4.539368171521315e-06,
      "loss": 3.4103,
      "step": 19380
    },
    {
      "epoch": 0.19719441731770834,
      "grad_norm": 13.567573547363281,
      "learning_rate": 4.539136850896944e-06,
      "loss": 3.9385,
      "step": 19385
    },
    {
      "epoch": 0.19724527994791666,
      "grad_norm": 11.386924743652344,
      "learning_rate": 4.53890547810174e-06,
      "loss": 3.1541,
      "step": 19390
    },
    {
      "epoch": 0.197296142578125,
      "grad_norm": 16.922727584838867,
      "learning_rate": 4.538674053141623e-06,
      "loss": 3.1228,
      "step": 19395
    },
    {
      "epoch": 0.19734700520833334,
      "grad_norm": 8.739219665527344,
      "learning_rate": 4.538442576022513e-06,
      "loss": 3.4801,
      "step": 19400
    },
    {
      "epoch": 0.19739786783854166,
      "grad_norm": 8.573755264282227,
      "learning_rate": 4.538211046750334e-06,
      "loss": 3.7405,
      "step": 19405
    },
    {
      "epoch": 0.19744873046875,
      "grad_norm": 10.428905487060547,
      "learning_rate": 4.537979465331007e-06,
      "loss": 3.4514,
      "step": 19410
    },
    {
      "epoch": 0.19749959309895834,
      "grad_norm": 10.055191993713379,
      "learning_rate": 4.53774783177046e-06,
      "loss": 3.3846,
      "step": 19415
    },
    {
      "epoch": 0.19755045572916666,
      "grad_norm": 7.410876750946045,
      "learning_rate": 4.537516146074616e-06,
      "loss": 3.3679,
      "step": 19420
    },
    {
      "epoch": 0.197601318359375,
      "grad_norm": 12.150521278381348,
      "learning_rate": 4.5372844082494045e-06,
      "loss": 3.8932,
      "step": 19425
    },
    {
      "epoch": 0.19765218098958334,
      "grad_norm": 13.015496253967285,
      "learning_rate": 4.537052618300756e-06,
      "loss": 3.3727,
      "step": 19430
    },
    {
      "epoch": 0.19770304361979166,
      "grad_norm": 12.705521583557129,
      "learning_rate": 4.536820776234597e-06,
      "loss": 3.5446,
      "step": 19435
    },
    {
      "epoch": 0.19775390625,
      "grad_norm": 13.614885330200195,
      "learning_rate": 4.536588882056863e-06,
      "loss": 3.3355,
      "step": 19440
    },
    {
      "epoch": 0.19780476888020834,
      "grad_norm": 10.389793395996094,
      "learning_rate": 4.536356935773485e-06,
      "loss": 3.4202,
      "step": 19445
    },
    {
      "epoch": 0.19785563151041666,
      "grad_norm": 11.007695198059082,
      "learning_rate": 4.536124937390396e-06,
      "loss": 3.7334,
      "step": 19450
    },
    {
      "epoch": 0.197906494140625,
      "grad_norm": 16.408811569213867,
      "learning_rate": 4.535892886913534e-06,
      "loss": 3.6896,
      "step": 19455
    },
    {
      "epoch": 0.19795735677083334,
      "grad_norm": 11.945623397827148,
      "learning_rate": 4.535660784348836e-06,
      "loss": 3.3185,
      "step": 19460
    },
    {
      "epoch": 0.19800821940104166,
      "grad_norm": 10.001398086547852,
      "learning_rate": 4.535428629702238e-06,
      "loss": 3.0122,
      "step": 19465
    },
    {
      "epoch": 0.19805908203125,
      "grad_norm": 18.04474449157715,
      "learning_rate": 4.535196422979681e-06,
      "loss": 3.6597,
      "step": 19470
    },
    {
      "epoch": 0.19810994466145834,
      "grad_norm": 14.083473205566406,
      "learning_rate": 4.534964164187106e-06,
      "loss": 3.4169,
      "step": 19475
    },
    {
      "epoch": 0.19816080729166666,
      "grad_norm": 14.401663780212402,
      "learning_rate": 4.534731853330455e-06,
      "loss": 3.5108,
      "step": 19480
    },
    {
      "epoch": 0.198211669921875,
      "grad_norm": 12.377889633178711,
      "learning_rate": 4.534499490415672e-06,
      "loss": 3.9204,
      "step": 19485
    },
    {
      "epoch": 0.19826253255208334,
      "grad_norm": 13.024721145629883,
      "learning_rate": 4.534267075448703e-06,
      "loss": 3.6746,
      "step": 19490
    },
    {
      "epoch": 0.19831339518229166,
      "grad_norm": 12.98880386352539,
      "learning_rate": 4.534034608435491e-06,
      "loss": 3.4475,
      "step": 19495
    },
    {
      "epoch": 0.1983642578125,
      "grad_norm": 12.89251708984375,
      "learning_rate": 4.533802089381988e-06,
      "loss": 3.5353,
      "step": 19500
    },
    {
      "epoch": 0.19841512044270834,
      "grad_norm": 11.248384475708008,
      "learning_rate": 4.533569518294139e-06,
      "loss": 3.211,
      "step": 19505
    },
    {
      "epoch": 0.19846598307291666,
      "grad_norm": 11.449090003967285,
      "learning_rate": 4.533336895177896e-06,
      "loss": 3.7413,
      "step": 19510
    },
    {
      "epoch": 0.198516845703125,
      "grad_norm": 9.919585227966309,
      "learning_rate": 4.53310422003921e-06,
      "loss": 3.7445,
      "step": 19515
    },
    {
      "epoch": 0.19856770833333334,
      "grad_norm": 9.99984359741211,
      "learning_rate": 4.532871492884035e-06,
      "loss": 3.2623,
      "step": 19520
    },
    {
      "epoch": 0.19861857096354166,
      "grad_norm": 16.127626419067383,
      "learning_rate": 4.532638713718326e-06,
      "loss": 3.4915,
      "step": 19525
    },
    {
      "epoch": 0.19866943359375,
      "grad_norm": 13.185535430908203,
      "learning_rate": 4.532405882548035e-06,
      "loss": 3.6796,
      "step": 19530
    },
    {
      "epoch": 0.19872029622395834,
      "grad_norm": 9.342668533325195,
      "learning_rate": 4.532172999379123e-06,
      "loss": 3.3614,
      "step": 19535
    },
    {
      "epoch": 0.19877115885416666,
      "grad_norm": 9.016987800598145,
      "learning_rate": 4.531940064217546e-06,
      "loss": 3.218,
      "step": 19540
    },
    {
      "epoch": 0.198822021484375,
      "grad_norm": 14.6638765335083,
      "learning_rate": 4.531707077069264e-06,
      "loss": 3.7645,
      "step": 19545
    },
    {
      "epoch": 0.19887288411458334,
      "grad_norm": 14.099635124206543,
      "learning_rate": 4.531474037940237e-06,
      "loss": 3.2719,
      "step": 19550
    },
    {
      "epoch": 0.19892374674479166,
      "grad_norm": 10.298885345458984,
      "learning_rate": 4.531240946836431e-06,
      "loss": 3.5016,
      "step": 19555
    },
    {
      "epoch": 0.198974609375,
      "grad_norm": 8.187545776367188,
      "learning_rate": 4.531007803763804e-06,
      "loss": 3.3259,
      "step": 19560
    },
    {
      "epoch": 0.19902547200520834,
      "grad_norm": 13.553715705871582,
      "learning_rate": 4.530774608728325e-06,
      "loss": 3.6581,
      "step": 19565
    },
    {
      "epoch": 0.19907633463541666,
      "grad_norm": 11.441377639770508,
      "learning_rate": 4.530541361735959e-06,
      "loss": 3.3082,
      "step": 19570
    },
    {
      "epoch": 0.199127197265625,
      "grad_norm": 13.178167343139648,
      "learning_rate": 4.530308062792673e-06,
      "loss": 3.2573,
      "step": 19575
    },
    {
      "epoch": 0.19917805989583334,
      "grad_norm": 7.674156188964844,
      "learning_rate": 4.530074711904437e-06,
      "loss": 3.5968,
      "step": 19580
    },
    {
      "epoch": 0.19922892252604166,
      "grad_norm": 13.66543197631836,
      "learning_rate": 4.52984130907722e-06,
      "loss": 3.1891,
      "step": 19585
    },
    {
      "epoch": 0.19927978515625,
      "grad_norm": 9.917035102844238,
      "learning_rate": 4.529607854316995e-06,
      "loss": 3.1431,
      "step": 19590
    },
    {
      "epoch": 0.19933064778645834,
      "grad_norm": 7.638389587402344,
      "learning_rate": 4.529374347629734e-06,
      "loss": 3.2769,
      "step": 19595
    },
    {
      "epoch": 0.19938151041666666,
      "grad_norm": 13.556234359741211,
      "learning_rate": 4.5291407890214105e-06,
      "loss": 3.3984,
      "step": 19600
    },
    {
      "epoch": 0.199432373046875,
      "grad_norm": 14.63976764678955,
      "learning_rate": 4.528907178498002e-06,
      "loss": 3.5608,
      "step": 19605
    },
    {
      "epoch": 0.19948323567708334,
      "grad_norm": 9.650500297546387,
      "learning_rate": 4.528673516065484e-06,
      "loss": 3.5258,
      "step": 19610
    },
    {
      "epoch": 0.19953409830729166,
      "grad_norm": 12.210488319396973,
      "learning_rate": 4.528439801729834e-06,
      "loss": 3.3766,
      "step": 19615
    },
    {
      "epoch": 0.1995849609375,
      "grad_norm": 10.5303373336792,
      "learning_rate": 4.528206035497034e-06,
      "loss": 3.4387,
      "step": 19620
    },
    {
      "epoch": 0.19963582356770834,
      "grad_norm": 11.196531295776367,
      "learning_rate": 4.527972217373062e-06,
      "loss": 3.3538,
      "step": 19625
    },
    {
      "epoch": 0.19968668619791666,
      "grad_norm": 10.746190071105957,
      "learning_rate": 4.527738347363902e-06,
      "loss": 3.5098,
      "step": 19630
    },
    {
      "epoch": 0.199737548828125,
      "grad_norm": 11.233555793762207,
      "learning_rate": 4.527504425475536e-06,
      "loss": 3.5044,
      "step": 19635
    },
    {
      "epoch": 0.19978841145833334,
      "grad_norm": 11.152935981750488,
      "learning_rate": 4.527270451713951e-06,
      "loss": 3.2676,
      "step": 19640
    },
    {
      "epoch": 0.19983927408854166,
      "grad_norm": 12.642974853515625,
      "learning_rate": 4.527036426085132e-06,
      "loss": 3.4784,
      "step": 19645
    },
    {
      "epoch": 0.19989013671875,
      "grad_norm": 11.634695053100586,
      "learning_rate": 4.526802348595066e-06,
      "loss": 3.3569,
      "step": 19650
    },
    {
      "epoch": 0.19994099934895834,
      "grad_norm": 16.355712890625,
      "learning_rate": 4.526568219249743e-06,
      "loss": 3.3902,
      "step": 19655
    },
    {
      "epoch": 0.19999186197916666,
      "grad_norm": 13.97595500946045,
      "learning_rate": 4.526334038055152e-06,
      "loss": 3.3811,
      "step": 19660
    },
    {
      "epoch": 0.200042724609375,
      "grad_norm": 13.499397277832031,
      "learning_rate": 4.526099805017285e-06,
      "loss": 3.4869,
      "step": 19665
    },
    {
      "epoch": 0.20009358723958334,
      "grad_norm": 14.013762474060059,
      "learning_rate": 4.525865520142135e-06,
      "loss": 3.0368,
      "step": 19670
    },
    {
      "epoch": 0.20014444986979166,
      "grad_norm": 15.110490798950195,
      "learning_rate": 4.525631183435696e-06,
      "loss": 3.3731,
      "step": 19675
    },
    {
      "epoch": 0.2001953125,
      "grad_norm": 11.767560005187988,
      "learning_rate": 4.525396794903962e-06,
      "loss": 3.3685,
      "step": 19680
    },
    {
      "epoch": 0.20024617513020834,
      "grad_norm": 11.866461753845215,
      "learning_rate": 4.525162354552932e-06,
      "loss": 3.1614,
      "step": 19685
    },
    {
      "epoch": 0.20029703776041666,
      "grad_norm": 11.312857627868652,
      "learning_rate": 4.524927862388604e-06,
      "loss": 3.203,
      "step": 19690
    },
    {
      "epoch": 0.200347900390625,
      "grad_norm": 13.30744457244873,
      "learning_rate": 4.524693318416975e-06,
      "loss": 3.5835,
      "step": 19695
    },
    {
      "epoch": 0.20039876302083334,
      "grad_norm": 12.074060440063477,
      "learning_rate": 4.524458722644049e-06,
      "loss": 3.5643,
      "step": 19700
    },
    {
      "epoch": 0.20044962565104166,
      "grad_norm": 15.581877708435059,
      "learning_rate": 4.524224075075825e-06,
      "loss": 3.3831,
      "step": 19705
    },
    {
      "epoch": 0.20050048828125,
      "grad_norm": 16.677257537841797,
      "learning_rate": 4.523989375718309e-06,
      "loss": 3.0894,
      "step": 19710
    },
    {
      "epoch": 0.20055135091145834,
      "grad_norm": 11.157782554626465,
      "learning_rate": 4.523754624577503e-06,
      "loss": 3.5676,
      "step": 19715
    },
    {
      "epoch": 0.20060221354166666,
      "grad_norm": 14.951972961425781,
      "learning_rate": 4.523519821659416e-06,
      "loss": 3.0777,
      "step": 19720
    },
    {
      "epoch": 0.200653076171875,
      "grad_norm": 12.17221450805664,
      "learning_rate": 4.523284966970053e-06,
      "loss": 3.3629,
      "step": 19725
    },
    {
      "epoch": 0.20070393880208334,
      "grad_norm": 12.007098197937012,
      "learning_rate": 4.523050060515424e-06,
      "loss": 3.7851,
      "step": 19730
    },
    {
      "epoch": 0.20075480143229166,
      "grad_norm": 11.723403930664062,
      "learning_rate": 4.5228151023015396e-06,
      "loss": 3.1963,
      "step": 19735
    },
    {
      "epoch": 0.2008056640625,
      "grad_norm": 10.277305603027344,
      "learning_rate": 4.522580092334409e-06,
      "loss": 3.6475,
      "step": 19740
    },
    {
      "epoch": 0.20085652669270834,
      "grad_norm": 12.673345565795898,
      "learning_rate": 4.522345030620047e-06,
      "loss": 3.1923,
      "step": 19745
    },
    {
      "epoch": 0.20090738932291666,
      "grad_norm": 10.174751281738281,
      "learning_rate": 4.5221099171644664e-06,
      "loss": 3.2029,
      "step": 19750
    },
    {
      "epoch": 0.200958251953125,
      "grad_norm": 13.342915534973145,
      "learning_rate": 4.5218747519736824e-06,
      "loss": 3.1688,
      "step": 19755
    },
    {
      "epoch": 0.20100911458333334,
      "grad_norm": 13.501158714294434,
      "learning_rate": 4.521639535053714e-06,
      "loss": 3.6031,
      "step": 19760
    },
    {
      "epoch": 0.20105997721354166,
      "grad_norm": 13.277938842773438,
      "learning_rate": 4.521404266410575e-06,
      "loss": 3.2503,
      "step": 19765
    },
    {
      "epoch": 0.20111083984375,
      "grad_norm": 14.475388526916504,
      "learning_rate": 4.521168946050288e-06,
      "loss": 3.169,
      "step": 19770
    },
    {
      "epoch": 0.20116170247395834,
      "grad_norm": 11.820828437805176,
      "learning_rate": 4.520933573978873e-06,
      "loss": 3.2875,
      "step": 19775
    },
    {
      "epoch": 0.20121256510416666,
      "grad_norm": 13.294929504394531,
      "learning_rate": 4.52069815020235e-06,
      "loss": 3.4509,
      "step": 19780
    },
    {
      "epoch": 0.201263427734375,
      "grad_norm": 11.186004638671875,
      "learning_rate": 4.520462674726746e-06,
      "loss": 3.3601,
      "step": 19785
    },
    {
      "epoch": 0.20131429036458334,
      "grad_norm": 9.814837455749512,
      "learning_rate": 4.520227147558081e-06,
      "loss": 3.1333,
      "step": 19790
    },
    {
      "epoch": 0.20136515299479166,
      "grad_norm": 14.82311725616455,
      "learning_rate": 4.519991568702385e-06,
      "loss": 3.4463,
      "step": 19795
    },
    {
      "epoch": 0.201416015625,
      "grad_norm": 12.3350248336792,
      "learning_rate": 4.519755938165683e-06,
      "loss": 2.8538,
      "step": 19800
    },
    {
      "epoch": 0.20146687825520834,
      "grad_norm": 9.856627464294434,
      "learning_rate": 4.519520255954004e-06,
      "loss": 3.3451,
      "step": 19805
    },
    {
      "epoch": 0.20151774088541666,
      "grad_norm": 12.37125015258789,
      "learning_rate": 4.519284522073377e-06,
      "loss": 3.5301,
      "step": 19810
    },
    {
      "epoch": 0.201568603515625,
      "grad_norm": 14.167351722717285,
      "learning_rate": 4.519048736529836e-06,
      "loss": 3.2116,
      "step": 19815
    },
    {
      "epoch": 0.20161946614583334,
      "grad_norm": 14.822011947631836,
      "learning_rate": 4.51881289932941e-06,
      "loss": 3.6803,
      "step": 19820
    },
    {
      "epoch": 0.20167032877604166,
      "grad_norm": 8.40614128112793,
      "learning_rate": 4.518577010478135e-06,
      "loss": 3.3569,
      "step": 19825
    },
    {
      "epoch": 0.20172119140625,
      "grad_norm": 7.354465007781982,
      "learning_rate": 4.518341069982046e-06,
      "loss": 3.3537,
      "step": 19830
    },
    {
      "epoch": 0.20177205403645834,
      "grad_norm": 14.864770889282227,
      "learning_rate": 4.518105077847179e-06,
      "loss": 3.6265,
      "step": 19835
    },
    {
      "epoch": 0.20182291666666666,
      "grad_norm": 11.00089168548584,
      "learning_rate": 4.5178690340795725e-06,
      "loss": 3.1597,
      "step": 19840
    },
    {
      "epoch": 0.201873779296875,
      "grad_norm": 10.744074821472168,
      "learning_rate": 4.5176329386852645e-06,
      "loss": 3.4948,
      "step": 19845
    },
    {
      "epoch": 0.20192464192708334,
      "grad_norm": 11.751157760620117,
      "learning_rate": 4.517396791670295e-06,
      "loss": 3.9409,
      "step": 19850
    },
    {
      "epoch": 0.20197550455729166,
      "grad_norm": 9.079497337341309,
      "learning_rate": 4.517160593040708e-06,
      "loss": 3.3642,
      "step": 19855
    },
    {
      "epoch": 0.2020263671875,
      "grad_norm": 13.547245979309082,
      "learning_rate": 4.516924342802545e-06,
      "loss": 3.5323,
      "step": 19860
    },
    {
      "epoch": 0.20207722981770834,
      "grad_norm": 11.432042121887207,
      "learning_rate": 4.516688040961851e-06,
      "loss": 3.0839,
      "step": 19865
    },
    {
      "epoch": 0.20212809244791666,
      "grad_norm": 8.827293395996094,
      "learning_rate": 4.516451687524671e-06,
      "loss": 3.1659,
      "step": 19870
    },
    {
      "epoch": 0.202178955078125,
      "grad_norm": 12.12353229522705,
      "learning_rate": 4.516215282497053e-06,
      "loss": 3.2329,
      "step": 19875
    },
    {
      "epoch": 0.20222981770833334,
      "grad_norm": 14.670421600341797,
      "learning_rate": 4.515978825885045e-06,
      "loss": 3.2416,
      "step": 19880
    },
    {
      "epoch": 0.20228068033854166,
      "grad_norm": 12.11870288848877,
      "learning_rate": 4.515742317694696e-06,
      "loss": 3.5721,
      "step": 19885
    },
    {
      "epoch": 0.20233154296875,
      "grad_norm": 7.884826183319092,
      "learning_rate": 4.515505757932057e-06,
      "loss": 3.4912,
      "step": 19890
    },
    {
      "epoch": 0.20238240559895834,
      "grad_norm": 15.665653228759766,
      "learning_rate": 4.515269146603183e-06,
      "loss": 3.2936,
      "step": 19895
    },
    {
      "epoch": 0.20243326822916666,
      "grad_norm": 13.599945068359375,
      "learning_rate": 4.5150324837141235e-06,
      "loss": 3.1648,
      "step": 19900
    },
    {
      "epoch": 0.202484130859375,
      "grad_norm": 13.613029479980469,
      "learning_rate": 4.514795769270937e-06,
      "loss": 3.1707,
      "step": 19905
    },
    {
      "epoch": 0.20253499348958334,
      "grad_norm": 10.294753074645996,
      "learning_rate": 4.514559003279677e-06,
      "loss": 3.6093,
      "step": 19910
    },
    {
      "epoch": 0.20258585611979166,
      "grad_norm": 9.080965995788574,
      "learning_rate": 4.5143221857464025e-06,
      "loss": 3.4589,
      "step": 19915
    },
    {
      "epoch": 0.20263671875,
      "grad_norm": 11.949151992797852,
      "learning_rate": 4.514085316677173e-06,
      "loss": 3.2098,
      "step": 19920
    },
    {
      "epoch": 0.20268758138020834,
      "grad_norm": 14.945830345153809,
      "learning_rate": 4.513848396078048e-06,
      "loss": 3.5203,
      "step": 19925
    },
    {
      "epoch": 0.20273844401041666,
      "grad_norm": 9.910847663879395,
      "learning_rate": 4.51361142395509e-06,
      "loss": 3.1446,
      "step": 19930
    },
    {
      "epoch": 0.202789306640625,
      "grad_norm": 13.371243476867676,
      "learning_rate": 4.5133744003143595e-06,
      "loss": 3.395,
      "step": 19935
    },
    {
      "epoch": 0.20284016927083334,
      "grad_norm": 16.05323600769043,
      "learning_rate": 4.513137325161923e-06,
      "loss": 3.5278,
      "step": 19940
    },
    {
      "epoch": 0.20289103190104166,
      "grad_norm": 15.017134666442871,
      "learning_rate": 4.5129001985038445e-06,
      "loss": 3.2377,
      "step": 19945
    },
    {
      "epoch": 0.20294189453125,
      "grad_norm": 13.222308158874512,
      "learning_rate": 4.512663020346192e-06,
      "loss": 3.3901,
      "step": 19950
    },
    {
      "epoch": 0.20299275716145834,
      "grad_norm": 9.478425025939941,
      "learning_rate": 4.512425790695033e-06,
      "loss": 3.4694,
      "step": 19955
    },
    {
      "epoch": 0.20304361979166666,
      "grad_norm": 12.377535820007324,
      "learning_rate": 4.5121885095564374e-06,
      "loss": 3.9845,
      "step": 19960
    },
    {
      "epoch": 0.203094482421875,
      "grad_norm": 11.293395042419434,
      "learning_rate": 4.5119511769364755e-06,
      "loss": 3.4415,
      "step": 19965
    },
    {
      "epoch": 0.20314534505208334,
      "grad_norm": 10.15709114074707,
      "learning_rate": 4.511713792841219e-06,
      "loss": 3.5203,
      "step": 19970
    },
    {
      "epoch": 0.20319620768229166,
      "grad_norm": 14.42606258392334,
      "learning_rate": 4.511476357276743e-06,
      "loss": 3.207,
      "step": 19975
    },
    {
      "epoch": 0.2032470703125,
      "grad_norm": 9.816762924194336,
      "learning_rate": 4.511238870249121e-06,
      "loss": 2.8994,
      "step": 19980
    },
    {
      "epoch": 0.20329793294270834,
      "grad_norm": 11.36377239227295,
      "learning_rate": 4.511001331764429e-06,
      "loss": 3.6271,
      "step": 19985
    },
    {
      "epoch": 0.20334879557291666,
      "grad_norm": 12.325090408325195,
      "learning_rate": 4.510763741828744e-06,
      "loss": 3.1638,
      "step": 19990
    },
    {
      "epoch": 0.203399658203125,
      "grad_norm": 9.825456619262695,
      "learning_rate": 4.510526100448146e-06,
      "loss": 3.251,
      "step": 19995
    },
    {
      "epoch": 0.20345052083333334,
      "grad_norm": 8.139120101928711,
      "learning_rate": 4.510288407628714e-06,
      "loss": 3.5742,
      "step": 20000
    },
    {
      "epoch": 0.20350138346354166,
      "grad_norm": 15.358031272888184,
      "learning_rate": 4.51005066337653e-06,
      "loss": 3.4079,
      "step": 20005
    },
    {
      "epoch": 0.20355224609375,
      "grad_norm": 14.130391120910645,
      "learning_rate": 4.509812867697676e-06,
      "loss": 3.3927,
      "step": 20010
    },
    {
      "epoch": 0.20360310872395834,
      "grad_norm": 8.38779067993164,
      "learning_rate": 4.509575020598237e-06,
      "loss": 3.4258,
      "step": 20015
    },
    {
      "epoch": 0.20365397135416666,
      "grad_norm": 16.604267120361328,
      "learning_rate": 4.509337122084297e-06,
      "loss": 3.3513,
      "step": 20020
    },
    {
      "epoch": 0.203704833984375,
      "grad_norm": 11.351149559020996,
      "learning_rate": 4.509099172161941e-06,
      "loss": 3.5092,
      "step": 20025
    },
    {
      "epoch": 0.20375569661458334,
      "grad_norm": 14.427471160888672,
      "learning_rate": 4.508861170837261e-06,
      "loss": 3.1501,
      "step": 20030
    },
    {
      "epoch": 0.20380655924479166,
      "grad_norm": 8.998710632324219,
      "learning_rate": 4.508623118116344e-06,
      "loss": 3.6365,
      "step": 20035
    },
    {
      "epoch": 0.203857421875,
      "grad_norm": 16.93038558959961,
      "learning_rate": 4.5083850140052795e-06,
      "loss": 3.538,
      "step": 20040
    },
    {
      "epoch": 0.20390828450520834,
      "grad_norm": 10.420814514160156,
      "learning_rate": 4.508146858510162e-06,
      "loss": 3.6521,
      "step": 20045
    },
    {
      "epoch": 0.20395914713541666,
      "grad_norm": 13.199938774108887,
      "learning_rate": 4.507908651637082e-06,
      "loss": 3.1762,
      "step": 20050
    },
    {
      "epoch": 0.204010009765625,
      "grad_norm": 8.664349555969238,
      "learning_rate": 4.507670393392135e-06,
      "loss": 2.9778,
      "step": 20055
    },
    {
      "epoch": 0.20406087239583334,
      "grad_norm": 16.507963180541992,
      "learning_rate": 4.5074320837814165e-06,
      "loss": 3.602,
      "step": 20060
    },
    {
      "epoch": 0.20411173502604166,
      "grad_norm": 11.548971176147461,
      "learning_rate": 4.507193722811025e-06,
      "loss": 3.3214,
      "step": 20065
    },
    {
      "epoch": 0.20416259765625,
      "grad_norm": 8.971203804016113,
      "learning_rate": 4.506955310487058e-06,
      "loss": 3.5269,
      "step": 20070
    },
    {
      "epoch": 0.20421346028645834,
      "grad_norm": 15.892366409301758,
      "learning_rate": 4.506716846815614e-06,
      "loss": 3.6735,
      "step": 20075
    },
    {
      "epoch": 0.20426432291666666,
      "grad_norm": 12.524632453918457,
      "learning_rate": 4.506478331802795e-06,
      "loss": 3.0179,
      "step": 20080
    },
    {
      "epoch": 0.204315185546875,
      "grad_norm": 13.37697982788086,
      "learning_rate": 4.506239765454704e-06,
      "loss": 3.1897,
      "step": 20085
    },
    {
      "epoch": 0.20436604817708334,
      "grad_norm": 11.929197311401367,
      "learning_rate": 4.506001147777445e-06,
      "loss": 3.2953,
      "step": 20090
    },
    {
      "epoch": 0.20441691080729166,
      "grad_norm": 11.28937816619873,
      "learning_rate": 4.505762478777121e-06,
      "loss": 3.4106,
      "step": 20095
    },
    {
      "epoch": 0.2044677734375,
      "grad_norm": 15.234105110168457,
      "learning_rate": 4.50552375845984e-06,
      "loss": 3.3616,
      "step": 20100
    },
    {
      "epoch": 0.20451863606770834,
      "grad_norm": 8.608086585998535,
      "learning_rate": 4.505284986831709e-06,
      "loss": 2.7565,
      "step": 20105
    },
    {
      "epoch": 0.20456949869791666,
      "grad_norm": 14.365691184997559,
      "learning_rate": 4.5050461638988366e-06,
      "loss": 3.4174,
      "step": 20110
    },
    {
      "epoch": 0.204620361328125,
      "grad_norm": 14.005748748779297,
      "learning_rate": 4.504807289667333e-06,
      "loss": 3.429,
      "step": 20115
    },
    {
      "epoch": 0.20467122395833334,
      "grad_norm": 14.597360610961914,
      "learning_rate": 4.50456836414331e-06,
      "loss": 3.188,
      "step": 20120
    },
    {
      "epoch": 0.20472208658854166,
      "grad_norm": 10.109228134155273,
      "learning_rate": 4.504329387332882e-06,
      "loss": 3.204,
      "step": 20125
    },
    {
      "epoch": 0.20477294921875,
      "grad_norm": 10.633742332458496,
      "learning_rate": 4.50409035924216e-06,
      "loss": 3.4177,
      "step": 20130
    },
    {
      "epoch": 0.20482381184895834,
      "grad_norm": 14.340601921081543,
      "learning_rate": 4.503851279877262e-06,
      "loss": 3.4076,
      "step": 20135
    },
    {
      "epoch": 0.20487467447916666,
      "grad_norm": 12.750170707702637,
      "learning_rate": 4.503612149244304e-06,
      "loss": 3.9644,
      "step": 20140
    },
    {
      "epoch": 0.204925537109375,
      "grad_norm": 9.519548416137695,
      "learning_rate": 4.503372967349404e-06,
      "loss": 3.4598,
      "step": 20145
    },
    {
      "epoch": 0.20497639973958334,
      "grad_norm": 11.770230293273926,
      "learning_rate": 4.503133734198682e-06,
      "loss": 3.7069,
      "step": 20150
    },
    {
      "epoch": 0.20502726236979166,
      "grad_norm": 7.587390422821045,
      "learning_rate": 4.502894449798258e-06,
      "loss": 4.2052,
      "step": 20155
    },
    {
      "epoch": 0.205078125,
      "grad_norm": 13.156473159790039,
      "learning_rate": 4.502655114154253e-06,
      "loss": 3.4261,
      "step": 20160
    },
    {
      "epoch": 0.20512898763020834,
      "grad_norm": 11.953205108642578,
      "learning_rate": 4.502415727272792e-06,
      "loss": 3.5789,
      "step": 20165
    },
    {
      "epoch": 0.20517985026041666,
      "grad_norm": 12.134018898010254,
      "learning_rate": 4.50217628916e-06,
      "loss": 3.4589,
      "step": 20170
    },
    {
      "epoch": 0.205230712890625,
      "grad_norm": 13.219483375549316,
      "learning_rate": 4.501936799822002e-06,
      "loss": 3.4232,
      "step": 20175
    },
    {
      "epoch": 0.20528157552083334,
      "grad_norm": 14.755402565002441,
      "learning_rate": 4.501697259264926e-06,
      "loss": 3.378,
      "step": 20180
    },
    {
      "epoch": 0.20533243815104166,
      "grad_norm": 14.997588157653809,
      "learning_rate": 4.501457667494899e-06,
      "loss": 3.4037,
      "step": 20185
    },
    {
      "epoch": 0.20538330078125,
      "grad_norm": 10.769617080688477,
      "learning_rate": 4.5012180245180525e-06,
      "loss": 3.4164,
      "step": 20190
    },
    {
      "epoch": 0.20543416341145834,
      "grad_norm": 13.86328125,
      "learning_rate": 4.500978330340517e-06,
      "loss": 3.5354,
      "step": 20195
    },
    {
      "epoch": 0.20548502604166666,
      "grad_norm": 9.748161315917969,
      "learning_rate": 4.500738584968425e-06,
      "loss": 3.3028,
      "step": 20200
    },
    {
      "epoch": 0.205535888671875,
      "grad_norm": 12.368988990783691,
      "learning_rate": 4.500498788407911e-06,
      "loss": 3.7509,
      "step": 20205
    },
    {
      "epoch": 0.20558675130208334,
      "grad_norm": 7.2794318199157715,
      "learning_rate": 4.50025894066511e-06,
      "loss": 3.5151,
      "step": 20210
    },
    {
      "epoch": 0.20563761393229166,
      "grad_norm": 8.730498313903809,
      "learning_rate": 4.500019041746157e-06,
      "loss": 3.2818,
      "step": 20215
    },
    {
      "epoch": 0.2056884765625,
      "grad_norm": 14.868056297302246,
      "learning_rate": 4.499779091657192e-06,
      "loss": 3.0804,
      "step": 20220
    },
    {
      "epoch": 0.20573933919270834,
      "grad_norm": 14.980323791503906,
      "learning_rate": 4.499539090404352e-06,
      "loss": 3.0393,
      "step": 20225
    },
    {
      "epoch": 0.20579020182291666,
      "grad_norm": 13.772435188293457,
      "learning_rate": 4.49929903799378e-06,
      "loss": 3.1725,
      "step": 20230
    },
    {
      "epoch": 0.205841064453125,
      "grad_norm": 16.341049194335938,
      "learning_rate": 4.499058934431613e-06,
      "loss": 3.5846,
      "step": 20235
    },
    {
      "epoch": 0.20589192708333334,
      "grad_norm": 12.222286224365234,
      "learning_rate": 4.498818779723999e-06,
      "loss": 3.5112,
      "step": 20240
    },
    {
      "epoch": 0.20594278971354166,
      "grad_norm": 14.283944129943848,
      "learning_rate": 4.49857857387708e-06,
      "loss": 3.5837,
      "step": 20245
    },
    {
      "epoch": 0.20599365234375,
      "grad_norm": 9.407193183898926,
      "learning_rate": 4.498338316897003e-06,
      "loss": 3.1534,
      "step": 20250
    },
    {
      "epoch": 0.20604451497395834,
      "grad_norm": 10.292035102844238,
      "learning_rate": 4.498098008789913e-06,
      "loss": 2.8781,
      "step": 20255
    },
    {
      "epoch": 0.20609537760416666,
      "grad_norm": 15.310603141784668,
      "learning_rate": 4.497857649561959e-06,
      "loss": 3.4637,
      "step": 20260
    },
    {
      "epoch": 0.206146240234375,
      "grad_norm": 12.086821556091309,
      "learning_rate": 4.4976172392192895e-06,
      "loss": 3.7627,
      "step": 20265
    },
    {
      "epoch": 0.20619710286458334,
      "grad_norm": 9.511617660522461,
      "learning_rate": 4.4973767777680576e-06,
      "loss": 3.2272,
      "step": 20270
    },
    {
      "epoch": 0.20624796549479166,
      "grad_norm": 14.292364120483398,
      "learning_rate": 4.497136265214413e-06,
      "loss": 3.4433,
      "step": 20275
    },
    {
      "epoch": 0.206298828125,
      "grad_norm": 9.073116302490234,
      "learning_rate": 4.496895701564512e-06,
      "loss": 3.5995,
      "step": 20280
    },
    {
      "epoch": 0.20634969075520834,
      "grad_norm": 12.548660278320312,
      "learning_rate": 4.496655086824507e-06,
      "loss": 3.4855,
      "step": 20285
    },
    {
      "epoch": 0.20640055338541666,
      "grad_norm": 12.357610702514648,
      "learning_rate": 4.496414421000555e-06,
      "loss": 3.4397,
      "step": 20290
    },
    {
      "epoch": 0.206451416015625,
      "grad_norm": 9.429367065429688,
      "learning_rate": 4.496173704098814e-06,
      "loss": 2.7644,
      "step": 20295
    },
    {
      "epoch": 0.20650227864583334,
      "grad_norm": 9.462889671325684,
      "learning_rate": 4.495932936125441e-06,
      "loss": 3.4458,
      "step": 20300
    },
    {
      "epoch": 0.20655314127604166,
      "grad_norm": 13.479082107543945,
      "learning_rate": 4.495692117086597e-06,
      "loss": 3.3089,
      "step": 20305
    },
    {
      "epoch": 0.20660400390625,
      "grad_norm": 15.136412620544434,
      "learning_rate": 4.495451246988443e-06,
      "loss": 3.4846,
      "step": 20310
    },
    {
      "epoch": 0.20665486653645834,
      "grad_norm": 15.00229263305664,
      "learning_rate": 4.495210325837143e-06,
      "loss": 3.159,
      "step": 20315
    },
    {
      "epoch": 0.20670572916666666,
      "grad_norm": 14.153552055358887,
      "learning_rate": 4.494969353638859e-06,
      "loss": 3.2891,
      "step": 20320
    },
    {
      "epoch": 0.206756591796875,
      "grad_norm": 10.715276718139648,
      "learning_rate": 4.494728330399757e-06,
      "loss": 3.3983,
      "step": 20325
    },
    {
      "epoch": 0.20680745442708334,
      "grad_norm": 14.318894386291504,
      "learning_rate": 4.494487256126003e-06,
      "loss": 3.3678,
      "step": 20330
    },
    {
      "epoch": 0.20685831705729166,
      "grad_norm": 10.274447441101074,
      "learning_rate": 4.494246130823765e-06,
      "loss": 3.0082,
      "step": 20335
    },
    {
      "epoch": 0.2069091796875,
      "grad_norm": 9.914955139160156,
      "learning_rate": 4.494004954499214e-06,
      "loss": 3.4777,
      "step": 20340
    },
    {
      "epoch": 0.20696004231770834,
      "grad_norm": 9.05743408203125,
      "learning_rate": 4.4937637271585174e-06,
      "loss": 3.186,
      "step": 20345
    },
    {
      "epoch": 0.20701090494791666,
      "grad_norm": 13.139704704284668,
      "learning_rate": 4.49352244880785e-06,
      "loss": 3.2584,
      "step": 20350
    },
    {
      "epoch": 0.207061767578125,
      "grad_norm": 11.612173080444336,
      "learning_rate": 4.493281119453382e-06,
      "loss": 3.6306,
      "step": 20355
    },
    {
      "epoch": 0.20711263020833334,
      "grad_norm": 8.827756881713867,
      "learning_rate": 4.493039739101289e-06,
      "loss": 3.5112,
      "step": 20360
    },
    {
      "epoch": 0.20716349283854166,
      "grad_norm": 13.69195556640625,
      "learning_rate": 4.492798307757747e-06,
      "loss": 3.6277,
      "step": 20365
    },
    {
      "epoch": 0.20721435546875,
      "grad_norm": 7.879222393035889,
      "learning_rate": 4.492556825428932e-06,
      "loss": 3.4635,
      "step": 20370
    },
    {
      "epoch": 0.20726521809895834,
      "grad_norm": 10.974207878112793,
      "learning_rate": 4.492315292121023e-06,
      "loss": 3.5284,
      "step": 20375
    },
    {
      "epoch": 0.20731608072916666,
      "grad_norm": 13.299656867980957,
      "learning_rate": 4.492073707840201e-06,
      "loss": 3.5775,
      "step": 20380
    },
    {
      "epoch": 0.207366943359375,
      "grad_norm": 10.87471866607666,
      "learning_rate": 4.491832072592644e-06,
      "loss": 3.3934,
      "step": 20385
    },
    {
      "epoch": 0.20741780598958334,
      "grad_norm": 10.13965129852295,
      "learning_rate": 4.491590386384536e-06,
      "loss": 3.431,
      "step": 20390
    },
    {
      "epoch": 0.20746866861979166,
      "grad_norm": 16.28556251525879,
      "learning_rate": 4.491348649222059e-06,
      "loss": 3.453,
      "step": 20395
    },
    {
      "epoch": 0.20751953125,
      "grad_norm": 15.164170265197754,
      "learning_rate": 4.4911068611114e-06,
      "loss": 3.3485,
      "step": 20400
    },
    {
      "epoch": 0.20757039388020834,
      "grad_norm": 13.961816787719727,
      "learning_rate": 4.4908650220587435e-06,
      "loss": 3.3264,
      "step": 20405
    },
    {
      "epoch": 0.20762125651041666,
      "grad_norm": 18.67232322692871,
      "learning_rate": 4.490623132070278e-06,
      "loss": 3.4781,
      "step": 20410
    },
    {
      "epoch": 0.207672119140625,
      "grad_norm": 10.25964069366455,
      "learning_rate": 4.49038119115219e-06,
      "loss": 3.2967,
      "step": 20415
    },
    {
      "epoch": 0.20772298177083334,
      "grad_norm": 14.518511772155762,
      "learning_rate": 4.490139199310672e-06,
      "loss": 3.6741,
      "step": 20420
    },
    {
      "epoch": 0.20777384440104166,
      "grad_norm": 12.733075141906738,
      "learning_rate": 4.489897156551914e-06,
      "loss": 3.4169,
      "step": 20425
    },
    {
      "epoch": 0.20782470703125,
      "grad_norm": 8.101183891296387,
      "learning_rate": 4.48965506288211e-06,
      "loss": 3.0928,
      "step": 20430
    },
    {
      "epoch": 0.20787556966145834,
      "grad_norm": 13.758424758911133,
      "learning_rate": 4.489412918307452e-06,
      "loss": 2.9868,
      "step": 20435
    },
    {
      "epoch": 0.20792643229166666,
      "grad_norm": 13.24248218536377,
      "learning_rate": 4.489170722834136e-06,
      "loss": 3.5981,
      "step": 20440
    },
    {
      "epoch": 0.207977294921875,
      "grad_norm": 12.845187187194824,
      "learning_rate": 4.488928476468358e-06,
      "loss": 3.4161,
      "step": 20445
    },
    {
      "epoch": 0.20802815755208334,
      "grad_norm": 16.167795181274414,
      "learning_rate": 4.488686179216317e-06,
      "loss": 3.45,
      "step": 20450
    },
    {
      "epoch": 0.20807902018229166,
      "grad_norm": 10.462541580200195,
      "learning_rate": 4.4884438310842115e-06,
      "loss": 3.3036,
      "step": 20455
    },
    {
      "epoch": 0.2081298828125,
      "grad_norm": 53.6070556640625,
      "learning_rate": 4.488201432078241e-06,
      "loss": 3.3821,
      "step": 20460
    },
    {
      "epoch": 0.20818074544270834,
      "grad_norm": 9.783699035644531,
      "learning_rate": 4.487958982204609e-06,
      "loss": 3.1686,
      "step": 20465
    },
    {
      "epoch": 0.20823160807291666,
      "grad_norm": 10.521998405456543,
      "learning_rate": 4.487716481469517e-06,
      "loss": 3.6972,
      "step": 20470
    },
    {
      "epoch": 0.208282470703125,
      "grad_norm": 15.251840591430664,
      "learning_rate": 4.487473929879171e-06,
      "loss": 3.729,
      "step": 20475
    },
    {
      "epoch": 0.20833333333333334,
      "grad_norm": 7.9105987548828125,
      "learning_rate": 4.487231327439775e-06,
      "loss": 3.7705,
      "step": 20480
    },
    {
      "epoch": 0.20838419596354166,
      "grad_norm": 12.525910377502441,
      "learning_rate": 4.486988674157536e-06,
      "loss": 3.5227,
      "step": 20485
    },
    {
      "epoch": 0.20843505859375,
      "grad_norm": 12.639177322387695,
      "learning_rate": 4.486745970038664e-06,
      "loss": 3.3517,
      "step": 20490
    },
    {
      "epoch": 0.20848592122395834,
      "grad_norm": 16.748825073242188,
      "learning_rate": 4.486503215089366e-06,
      "loss": 3.1295,
      "step": 20495
    },
    {
      "epoch": 0.20853678385416666,
      "grad_norm": 14.206311225891113,
      "learning_rate": 4.486260409315855e-06,
      "loss": 3.3888,
      "step": 20500
    },
    {
      "epoch": 0.208587646484375,
      "grad_norm": 17.936237335205078,
      "learning_rate": 4.486017552724342e-06,
      "loss": 3.6911,
      "step": 20505
    },
    {
      "epoch": 0.20863850911458334,
      "grad_norm": 14.42271900177002,
      "learning_rate": 4.4857746453210404e-06,
      "loss": 3.443,
      "step": 20510
    },
    {
      "epoch": 0.20868937174479166,
      "grad_norm": 14.182038307189941,
      "learning_rate": 4.485531687112165e-06,
      "loss": 3.4927,
      "step": 20515
    },
    {
      "epoch": 0.208740234375,
      "grad_norm": 19.564302444458008,
      "learning_rate": 4.4852886781039325e-06,
      "loss": 3.5512,
      "step": 20520
    },
    {
      "epoch": 0.20879109700520834,
      "grad_norm": 14.08537483215332,
      "learning_rate": 4.485045618302559e-06,
      "loss": 3.6225,
      "step": 20525
    },
    {
      "epoch": 0.20884195963541666,
      "grad_norm": 9.4270658493042,
      "learning_rate": 4.4848025077142645e-06,
      "loss": 3.2626,
      "step": 20530
    },
    {
      "epoch": 0.208892822265625,
      "grad_norm": 16.107717514038086,
      "learning_rate": 4.484559346345268e-06,
      "loss": 3.4001,
      "step": 20535
    },
    {
      "epoch": 0.20894368489583334,
      "grad_norm": 8.756304740905762,
      "learning_rate": 4.484316134201792e-06,
      "loss": 3.0488,
      "step": 20540
    },
    {
      "epoch": 0.20899454752604166,
      "grad_norm": 10.211101531982422,
      "learning_rate": 4.484072871290057e-06,
      "loss": 4.1292,
      "step": 20545
    },
    {
      "epoch": 0.20904541015625,
      "grad_norm": 7.968142986297607,
      "learning_rate": 4.483829557616287e-06,
      "loss": 3.3119,
      "step": 20550
    },
    {
      "epoch": 0.20909627278645834,
      "grad_norm": 9.335982322692871,
      "learning_rate": 4.48358619318671e-06,
      "loss": 3.2166,
      "step": 20555
    },
    {
      "epoch": 0.20914713541666666,
      "grad_norm": 12.940924644470215,
      "learning_rate": 4.483342778007549e-06,
      "loss": 3.1356,
      "step": 20560
    },
    {
      "epoch": 0.209197998046875,
      "grad_norm": 13.300333976745605,
      "learning_rate": 4.483099312085034e-06,
      "loss": 3.3874,
      "step": 20565
    },
    {
      "epoch": 0.20924886067708334,
      "grad_norm": 14.966835021972656,
      "learning_rate": 4.482855795425392e-06,
      "loss": 3.2123,
      "step": 20570
    },
    {
      "epoch": 0.20929972330729166,
      "grad_norm": 13.067811965942383,
      "learning_rate": 4.482612228034856e-06,
      "loss": 3.2944,
      "step": 20575
    },
    {
      "epoch": 0.2093505859375,
      "grad_norm": 15.019890785217285,
      "learning_rate": 4.4823686099196554e-06,
      "loss": 3.38,
      "step": 20580
    },
    {
      "epoch": 0.20940144856770834,
      "grad_norm": 14.11163330078125,
      "learning_rate": 4.482124941086023e-06,
      "loss": 3.553,
      "step": 20585
    },
    {
      "epoch": 0.20945231119791666,
      "grad_norm": 16.385244369506836,
      "learning_rate": 4.481881221540195e-06,
      "loss": 3.4293,
      "step": 20590
    },
    {
      "epoch": 0.209503173828125,
      "grad_norm": 10.880959510803223,
      "learning_rate": 4.481637451288405e-06,
      "loss": 3.1211,
      "step": 20595
    },
    {
      "epoch": 0.20955403645833334,
      "grad_norm": 14.787332534790039,
      "learning_rate": 4.481393630336891e-06,
      "loss": 3.1561,
      "step": 20600
    },
    {
      "epoch": 0.20960489908854166,
      "grad_norm": 11.408947944641113,
      "learning_rate": 4.48114975869189e-06,
      "loss": 3.5343,
      "step": 20605
    },
    {
      "epoch": 0.20965576171875,
      "grad_norm": 15.02966022491455,
      "learning_rate": 4.480905836359643e-06,
      "loss": 3.2872,
      "step": 20610
    },
    {
      "epoch": 0.20970662434895834,
      "grad_norm": 10.301656723022461,
      "learning_rate": 4.480661863346389e-06,
      "loss": 3.6296,
      "step": 20615
    },
    {
      "epoch": 0.20975748697916666,
      "grad_norm": 14.3646879196167,
      "learning_rate": 4.480417839658369e-06,
      "loss": 3.0304,
      "step": 20620
    },
    {
      "epoch": 0.209808349609375,
      "grad_norm": 12.71937084197998,
      "learning_rate": 4.48017376530183e-06,
      "loss": 3.2242,
      "step": 20625
    },
    {
      "epoch": 0.20985921223958334,
      "grad_norm": 12.150714874267578,
      "learning_rate": 4.479929640283015e-06,
      "loss": 3.3238,
      "step": 20630
    },
    {
      "epoch": 0.20991007486979166,
      "grad_norm": 12.3690185546875,
      "learning_rate": 4.479685464608169e-06,
      "loss": 3.4552,
      "step": 20635
    },
    {
      "epoch": 0.2099609375,
      "grad_norm": 8.464226722717285,
      "learning_rate": 4.479441238283539e-06,
      "loss": 3.7959,
      "step": 20640
    },
    {
      "epoch": 0.21001180013020834,
      "grad_norm": 11.325082778930664,
      "learning_rate": 4.479196961315374e-06,
      "loss": 3.5601,
      "step": 20645
    },
    {
      "epoch": 0.21006266276041666,
      "grad_norm": 13.100303649902344,
      "learning_rate": 4.478952633709925e-06,
      "loss": 3.3486,
      "step": 20650
    },
    {
      "epoch": 0.210113525390625,
      "grad_norm": 9.751178741455078,
      "learning_rate": 4.4787082554734406e-06,
      "loss": 3.401,
      "step": 20655
    },
    {
      "epoch": 0.21016438802083334,
      "grad_norm": 8.583620071411133,
      "learning_rate": 4.4784638266121755e-06,
      "loss": 3.4622,
      "step": 20660
    },
    {
      "epoch": 0.21021525065104166,
      "grad_norm": 8.873421669006348,
      "learning_rate": 4.478219347132382e-06,
      "loss": 3.4424,
      "step": 20665
    },
    {
      "epoch": 0.21026611328125,
      "grad_norm": 8.892828941345215,
      "learning_rate": 4.477974817040315e-06,
      "loss": 3.3382,
      "step": 20670
    },
    {
      "epoch": 0.21031697591145834,
      "grad_norm": 8.137938499450684,
      "learning_rate": 4.477730236342231e-06,
      "loss": 3.3886,
      "step": 20675
    },
    {
      "epoch": 0.21036783854166666,
      "grad_norm": 9.50516414642334,
      "learning_rate": 4.477485605044389e-06,
      "loss": 3.4875,
      "step": 20680
    },
    {
      "epoch": 0.210418701171875,
      "grad_norm": 17.410694122314453,
      "learning_rate": 4.477240923153046e-06,
      "loss": 3.3188,
      "step": 20685
    },
    {
      "epoch": 0.21046956380208334,
      "grad_norm": 8.748286247253418,
      "learning_rate": 4.476996190674462e-06,
      "loss": 3.8879,
      "step": 20690
    },
    {
      "epoch": 0.21052042643229166,
      "grad_norm": 12.05586051940918,
      "learning_rate": 4.476751407614899e-06,
      "loss": 3.3138,
      "step": 20695
    },
    {
      "epoch": 0.2105712890625,
      "grad_norm": 9.442726135253906,
      "learning_rate": 4.4765065739806205e-06,
      "loss": 3.0828,
      "step": 20700
    },
    {
      "epoch": 0.21062215169270834,
      "grad_norm": 13.491192817687988,
      "learning_rate": 4.47626168977789e-06,
      "loss": 3.1086,
      "step": 20705
    },
    {
      "epoch": 0.21067301432291666,
      "grad_norm": 11.93213176727295,
      "learning_rate": 4.476016755012971e-06,
      "loss": 3.5522,
      "step": 20710
    },
    {
      "epoch": 0.210723876953125,
      "grad_norm": 14.229875564575195,
      "learning_rate": 4.4757717696921335e-06,
      "loss": 3.5604,
      "step": 20715
    },
    {
      "epoch": 0.21077473958333334,
      "grad_norm": 11.76516056060791,
      "learning_rate": 4.475526733821642e-06,
      "loss": 3.4999,
      "step": 20720
    },
    {
      "epoch": 0.21082560221354166,
      "grad_norm": 10.882475852966309,
      "learning_rate": 4.475281647407768e-06,
      "loss": 3.859,
      "step": 20725
    },
    {
      "epoch": 0.21087646484375,
      "grad_norm": 17.010963439941406,
      "learning_rate": 4.475036510456782e-06,
      "loss": 4.1897,
      "step": 20730
    },
    {
      "epoch": 0.21092732747395834,
      "grad_norm": 11.392085075378418,
      "learning_rate": 4.474791322974954e-06,
      "loss": 3.4268,
      "step": 20735
    },
    {
      "epoch": 0.21097819010416666,
      "grad_norm": 8.899847984313965,
      "learning_rate": 4.4745460849685584e-06,
      "loss": 3.8221,
      "step": 20740
    },
    {
      "epoch": 0.211029052734375,
      "grad_norm": 8.42432975769043,
      "learning_rate": 4.4743007964438686e-06,
      "loss": 3.6357,
      "step": 20745
    },
    {
      "epoch": 0.21107991536458334,
      "grad_norm": 16.582895278930664,
      "learning_rate": 4.474055457407162e-06,
      "loss": 3.4706,
      "step": 20750
    },
    {
      "epoch": 0.21113077799479166,
      "grad_norm": 13.940359115600586,
      "learning_rate": 4.4738100678647135e-06,
      "loss": 3.3978,
      "step": 20755
    },
    {
      "epoch": 0.211181640625,
      "grad_norm": 13.06972599029541,
      "learning_rate": 4.473564627822803e-06,
      "loss": 3.2053,
      "step": 20760
    },
    {
      "epoch": 0.21123250325520834,
      "grad_norm": 11.801219940185547,
      "learning_rate": 4.4733191372877094e-06,
      "loss": 3.0616,
      "step": 20765
    },
    {
      "epoch": 0.21128336588541666,
      "grad_norm": 10.158659934997559,
      "learning_rate": 4.4730735962657125e-06,
      "loss": 3.3077,
      "step": 20770
    },
    {
      "epoch": 0.211334228515625,
      "grad_norm": 15.788928031921387,
      "learning_rate": 4.4728280047630955e-06,
      "loss": 3.3533,
      "step": 20775
    },
    {
      "epoch": 0.21138509114583334,
      "grad_norm": 14.644887924194336,
      "learning_rate": 4.472582362786142e-06,
      "loss": 3.7334,
      "step": 20780
    },
    {
      "epoch": 0.21143595377604166,
      "grad_norm": 17.853012084960938,
      "learning_rate": 4.472336670341136e-06,
      "loss": 3.4043,
      "step": 20785
    },
    {
      "epoch": 0.21148681640625,
      "grad_norm": 14.728127479553223,
      "learning_rate": 4.472090927434364e-06,
      "loss": 3.3049,
      "step": 20790
    },
    {
      "epoch": 0.21153767903645834,
      "grad_norm": 69.46292114257812,
      "learning_rate": 4.4718451340721136e-06,
      "loss": 3.7458,
      "step": 20795
    },
    {
      "epoch": 0.21158854166666666,
      "grad_norm": 16.712196350097656,
      "learning_rate": 4.471599290260672e-06,
      "loss": 3.9221,
      "step": 20800
    },
    {
      "epoch": 0.211639404296875,
      "grad_norm": 15.2913236618042,
      "learning_rate": 4.471353396006331e-06,
      "loss": 3.5172,
      "step": 20805
    },
    {
      "epoch": 0.21169026692708334,
      "grad_norm": 16.197208404541016,
      "learning_rate": 4.4711074513153795e-06,
      "loss": 3.344,
      "step": 20810
    },
    {
      "epoch": 0.21174112955729166,
      "grad_norm": 10.586454391479492,
      "learning_rate": 4.470861456194111e-06,
      "loss": 3.3205,
      "step": 20815
    },
    {
      "epoch": 0.2117919921875,
      "grad_norm": 13.682154655456543,
      "learning_rate": 4.470615410648819e-06,
      "loss": 3.408,
      "step": 20820
    },
    {
      "epoch": 0.21184285481770834,
      "grad_norm": 8.82666015625,
      "learning_rate": 4.4703693146858e-06,
      "loss": 2.9437,
      "step": 20825
    },
    {
      "epoch": 0.21189371744791666,
      "grad_norm": 12.18643569946289,
      "learning_rate": 4.470123168311349e-06,
      "loss": 3.5294,
      "step": 20830
    },
    {
      "epoch": 0.211944580078125,
      "grad_norm": 13.015913009643555,
      "learning_rate": 4.469876971531763e-06,
      "loss": 2.9248,
      "step": 20835
    },
    {
      "epoch": 0.21199544270833334,
      "grad_norm": 13.3177490234375,
      "learning_rate": 4.469630724353342e-06,
      "loss": 3.1084,
      "step": 20840
    },
    {
      "epoch": 0.21204630533854166,
      "grad_norm": 9.608677864074707,
      "learning_rate": 4.469384426782385e-06,
      "loss": 3.246,
      "step": 20845
    },
    {
      "epoch": 0.21209716796875,
      "grad_norm": 13.394899368286133,
      "learning_rate": 4.469138078825195e-06,
      "loss": 3.2107,
      "step": 20850
    },
    {
      "epoch": 0.21214803059895834,
      "grad_norm": 10.666743278503418,
      "learning_rate": 4.468891680488074e-06,
      "loss": 3.6222,
      "step": 20855
    },
    {
      "epoch": 0.21219889322916666,
      "grad_norm": 11.721149444580078,
      "learning_rate": 4.468645231777326e-06,
      "loss": 3.6107,
      "step": 20860
    },
    {
      "epoch": 0.212249755859375,
      "grad_norm": 12.614690780639648,
      "learning_rate": 4.4683987326992565e-06,
      "loss": 3.3962,
      "step": 20865
    },
    {
      "epoch": 0.21230061848958334,
      "grad_norm": 10.944849967956543,
      "learning_rate": 4.468152183260172e-06,
      "loss": 3.2563,
      "step": 20870
    },
    {
      "epoch": 0.21235148111979166,
      "grad_norm": 11.871996879577637,
      "learning_rate": 4.467905583466379e-06,
      "loss": 3.0758,
      "step": 20875
    },
    {
      "epoch": 0.21240234375,
      "grad_norm": 13.066951751708984,
      "learning_rate": 4.46765893332419e-06,
      "loss": 3.5028,
      "step": 20880
    },
    {
      "epoch": 0.21245320638020834,
      "grad_norm": 15.884061813354492,
      "learning_rate": 4.467412232839913e-06,
      "loss": 3.4618,
      "step": 20885
    },
    {
      "epoch": 0.21250406901041666,
      "grad_norm": 9.060402870178223,
      "learning_rate": 4.46716548201986e-06,
      "loss": 3.4159,
      "step": 20890
    },
    {
      "epoch": 0.212554931640625,
      "grad_norm": 9.965991973876953,
      "learning_rate": 4.466918680870344e-06,
      "loss": 3.4609,
      "step": 20895
    },
    {
      "epoch": 0.21260579427083334,
      "grad_norm": 12.760985374450684,
      "learning_rate": 4.46667182939768e-06,
      "loss": 3.7191,
      "step": 20900
    },
    {
      "epoch": 0.21265665690104166,
      "grad_norm": 12.720123291015625,
      "learning_rate": 4.466424927608184e-06,
      "loss": 3.3874,
      "step": 20905
    },
    {
      "epoch": 0.21270751953125,
      "grad_norm": 10.351378440856934,
      "learning_rate": 4.466177975508172e-06,
      "loss": 3.3034,
      "step": 20910
    },
    {
      "epoch": 0.21275838216145834,
      "grad_norm": 9.52064037322998,
      "learning_rate": 4.465930973103963e-06,
      "loss": 3.4826,
      "step": 20915
    },
    {
      "epoch": 0.21280924479166666,
      "grad_norm": 13.159667015075684,
      "learning_rate": 4.465683920401875e-06,
      "loss": 3.4415,
      "step": 20920
    },
    {
      "epoch": 0.212860107421875,
      "grad_norm": 14.236352920532227,
      "learning_rate": 4.465436817408231e-06,
      "loss": 3.261,
      "step": 20925
    },
    {
      "epoch": 0.21291097005208334,
      "grad_norm": 10.393044471740723,
      "learning_rate": 4.465189664129351e-06,
      "loss": 3.2247,
      "step": 20930
    },
    {
      "epoch": 0.21296183268229166,
      "grad_norm": 8.096722602844238,
      "learning_rate": 4.46494246057156e-06,
      "loss": 3.2013,
      "step": 20935
    },
    {
      "epoch": 0.2130126953125,
      "grad_norm": 15.64661693572998,
      "learning_rate": 4.4646952067411805e-06,
      "loss": 3.1279,
      "step": 20940
    },
    {
      "epoch": 0.21306355794270834,
      "grad_norm": 11.443922996520996,
      "learning_rate": 4.4644479026445416e-06,
      "loss": 3.0867,
      "step": 20945
    },
    {
      "epoch": 0.21311442057291666,
      "grad_norm": 8.727100372314453,
      "learning_rate": 4.464200548287968e-06,
      "loss": 3.3102,
      "step": 20950
    },
    {
      "epoch": 0.213165283203125,
      "grad_norm": 16.262468338012695,
      "learning_rate": 4.463953143677788e-06,
      "loss": 3.2918,
      "step": 20955
    },
    {
      "epoch": 0.21321614583333334,
      "grad_norm": 13.321330070495605,
      "learning_rate": 4.463705688820333e-06,
      "loss": 3.2454,
      "step": 20960
    },
    {
      "epoch": 0.21326700846354166,
      "grad_norm": 11.59324836730957,
      "learning_rate": 4.463458183721934e-06,
      "loss": 3.6288,
      "step": 20965
    },
    {
      "epoch": 0.21331787109375,
      "grad_norm": 14.173730850219727,
      "learning_rate": 4.463210628388922e-06,
      "loss": 3.0699,
      "step": 20970
    },
    {
      "epoch": 0.21336873372395834,
      "grad_norm": 15.06574821472168,
      "learning_rate": 4.462963022827632e-06,
      "loss": 3.6585,
      "step": 20975
    },
    {
      "epoch": 0.21341959635416666,
      "grad_norm": 9.580982208251953,
      "learning_rate": 4.462715367044398e-06,
      "loss": 3.5402,
      "step": 20980
    },
    {
      "epoch": 0.213470458984375,
      "grad_norm": 6.7173027992248535,
      "learning_rate": 4.462467661045556e-06,
      "loss": 3.4419,
      "step": 20985
    },
    {
      "epoch": 0.21352132161458334,
      "grad_norm": 9.555313110351562,
      "learning_rate": 4.462219904837445e-06,
      "loss": 3.2101,
      "step": 20990
    },
    {
      "epoch": 0.21357218424479166,
      "grad_norm": 14.696534156799316,
      "learning_rate": 4.461972098426402e-06,
      "loss": 3.6329,
      "step": 20995
    },
    {
      "epoch": 0.213623046875,
      "grad_norm": 13.641729354858398,
      "learning_rate": 4.461724241818769e-06,
      "loss": 3.6916,
      "step": 21000
    },
    {
      "epoch": 0.21367390950520834,
      "grad_norm": 13.780049324035645,
      "learning_rate": 4.461476335020886e-06,
      "loss": 3.3034,
      "step": 21005
    },
    {
      "epoch": 0.21372477213541666,
      "grad_norm": 13.071630477905273,
      "learning_rate": 4.461228378039096e-06,
      "loss": 3.2033,
      "step": 21010
    },
    {
      "epoch": 0.213775634765625,
      "grad_norm": 10.242560386657715,
      "learning_rate": 4.460980370879742e-06,
      "loss": 3.6485,
      "step": 21015
    },
    {
      "epoch": 0.21382649739583334,
      "grad_norm": 13.603046417236328,
      "learning_rate": 4.46073231354917e-06,
      "loss": 3.4695,
      "step": 21020
    },
    {
      "epoch": 0.21387736002604166,
      "grad_norm": 14.24183464050293,
      "learning_rate": 4.460484206053727e-06,
      "loss": 3.5122,
      "step": 21025
    },
    {
      "epoch": 0.21392822265625,
      "grad_norm": 12.189628601074219,
      "learning_rate": 4.46023604839976e-06,
      "loss": 3.4621,
      "step": 21030
    },
    {
      "epoch": 0.21397908528645834,
      "grad_norm": 15.0325288772583,
      "learning_rate": 4.459987840593618e-06,
      "loss": 3.6319,
      "step": 21035
    },
    {
      "epoch": 0.21402994791666666,
      "grad_norm": 15.047171592712402,
      "learning_rate": 4.4597395826416525e-06,
      "loss": 3.1841,
      "step": 21040
    },
    {
      "epoch": 0.214080810546875,
      "grad_norm": 11.110899925231934,
      "learning_rate": 4.459491274550214e-06,
      "loss": 3.261,
      "step": 21045
    },
    {
      "epoch": 0.21413167317708334,
      "grad_norm": 10.35274887084961,
      "learning_rate": 4.459242916325656e-06,
      "loss": 3.469,
      "step": 21050
    },
    {
      "epoch": 0.21418253580729166,
      "grad_norm": 8.398360252380371,
      "learning_rate": 4.4589945079743315e-06,
      "loss": 3.7006,
      "step": 21055
    },
    {
      "epoch": 0.2142333984375,
      "grad_norm": 13.217538833618164,
      "learning_rate": 4.458746049502597e-06,
      "loss": 3.3198,
      "step": 21060
    },
    {
      "epoch": 0.21428426106770834,
      "grad_norm": 12.722177505493164,
      "learning_rate": 4.4584975409168095e-06,
      "loss": 3.3009,
      "step": 21065
    },
    {
      "epoch": 0.21433512369791666,
      "grad_norm": 11.289009094238281,
      "learning_rate": 4.458248982223327e-06,
      "loss": 3.4066,
      "step": 21070
    },
    {
      "epoch": 0.214385986328125,
      "grad_norm": 13.187050819396973,
      "learning_rate": 4.458000373428507e-06,
      "loss": 3.5669,
      "step": 21075
    },
    {
      "epoch": 0.21443684895833334,
      "grad_norm": 13.233553886413574,
      "learning_rate": 4.457751714538713e-06,
      "loss": 3.1861,
      "step": 21080
    },
    {
      "epoch": 0.21448771158854166,
      "grad_norm": 11.534300804138184,
      "learning_rate": 4.457503005560305e-06,
      "loss": 3.4397,
      "step": 21085
    },
    {
      "epoch": 0.21453857421875,
      "grad_norm": 17.411758422851562,
      "learning_rate": 4.457254246499646e-06,
      "loss": 3.6133,
      "step": 21090
    },
    {
      "epoch": 0.21458943684895834,
      "grad_norm": 13.31394100189209,
      "learning_rate": 4.457005437363102e-06,
      "loss": 2.982,
      "step": 21095
    },
    {
      "epoch": 0.21464029947916666,
      "grad_norm": 8.906242370605469,
      "learning_rate": 4.4567565781570374e-06,
      "loss": 4.1482,
      "step": 21100
    },
    {
      "epoch": 0.214691162109375,
      "grad_norm": 13.039563179016113,
      "learning_rate": 4.456507668887819e-06,
      "loss": 3.5613,
      "step": 21105
    },
    {
      "epoch": 0.21474202473958334,
      "grad_norm": 8.964936256408691,
      "learning_rate": 4.456258709561817e-06,
      "loss": 3.5627,
      "step": 21110
    },
    {
      "epoch": 0.21479288736979166,
      "grad_norm": 10.239699363708496,
      "learning_rate": 4.456009700185398e-06,
      "loss": 3.1105,
      "step": 21115
    },
    {
      "epoch": 0.21484375,
      "grad_norm": 12.297407150268555,
      "learning_rate": 4.455760640764935e-06,
      "loss": 3.0417,
      "step": 21120
    },
    {
      "epoch": 0.21489461263020834,
      "grad_norm": 11.456560134887695,
      "learning_rate": 4.4555115313068e-06,
      "loss": 3.0432,
      "step": 21125
    },
    {
      "epoch": 0.21494547526041666,
      "grad_norm": 15.854055404663086,
      "learning_rate": 4.455262371817366e-06,
      "loss": 3.6049,
      "step": 21130
    },
    {
      "epoch": 0.214996337890625,
      "grad_norm": 13.468208312988281,
      "learning_rate": 4.455013162303007e-06,
      "loss": 3.3275,
      "step": 21135
    },
    {
      "epoch": 0.21504720052083334,
      "grad_norm": 15.826327323913574,
      "learning_rate": 4.454763902770099e-06,
      "loss": 3.4567,
      "step": 21140
    },
    {
      "epoch": 0.21509806315104166,
      "grad_norm": 14.552145957946777,
      "learning_rate": 4.454514593225022e-06,
      "loss": 3.3701,
      "step": 21145
    },
    {
      "epoch": 0.21514892578125,
      "grad_norm": 14.960928916931152,
      "learning_rate": 4.454265233674151e-06,
      "loss": 3.3224,
      "step": 21150
    },
    {
      "epoch": 0.21519978841145834,
      "grad_norm": 14.692644119262695,
      "learning_rate": 4.454015824123867e-06,
      "loss": 3.4464,
      "step": 21155
    },
    {
      "epoch": 0.21525065104166666,
      "grad_norm": 14.767197608947754,
      "learning_rate": 4.453766364580552e-06,
      "loss": 3.2771,
      "step": 21160
    },
    {
      "epoch": 0.215301513671875,
      "grad_norm": 12.008683204650879,
      "learning_rate": 4.453516855050588e-06,
      "loss": 3.2921,
      "step": 21165
    },
    {
      "epoch": 0.21535237630208334,
      "grad_norm": 469.43402099609375,
      "learning_rate": 4.453267295540357e-06,
      "loss": 3.4084,
      "step": 21170
    },
    {
      "epoch": 0.21540323893229166,
      "grad_norm": 9.679253578186035,
      "learning_rate": 4.453017686056246e-06,
      "loss": 3.5274,
      "step": 21175
    },
    {
      "epoch": 0.2154541015625,
      "grad_norm": 11.721029281616211,
      "learning_rate": 4.45276802660464e-06,
      "loss": 3.6015,
      "step": 21180
    },
    {
      "epoch": 0.21550496419270834,
      "grad_norm": 14.105132102966309,
      "learning_rate": 4.452518317191928e-06,
      "loss": 3.498,
      "step": 21185
    },
    {
      "epoch": 0.21555582682291666,
      "grad_norm": 14.37909984588623,
      "learning_rate": 4.452268557824497e-06,
      "loss": 3.1592,
      "step": 21190
    },
    {
      "epoch": 0.215606689453125,
      "grad_norm": 15.932445526123047,
      "learning_rate": 4.452018748508737e-06,
      "loss": 3.2462,
      "step": 21195
    },
    {
      "epoch": 0.21565755208333334,
      "grad_norm": 14.666515350341797,
      "learning_rate": 4.451768889251041e-06,
      "loss": 3.7043,
      "step": 21200
    },
    {
      "epoch": 0.21570841471354166,
      "grad_norm": 6.903579235076904,
      "learning_rate": 4.4515189800578e-06,
      "loss": 3.0669,
      "step": 21205
    },
    {
      "epoch": 0.21575927734375,
      "grad_norm": 13.199335098266602,
      "learning_rate": 4.451269020935409e-06,
      "loss": 3.1371,
      "step": 21210
    },
    {
      "epoch": 0.21581013997395834,
      "grad_norm": 11.0657320022583,
      "learning_rate": 4.451019011890262e-06,
      "loss": 2.911,
      "step": 21215
    },
    {
      "epoch": 0.21586100260416666,
      "grad_norm": 10.579645156860352,
      "learning_rate": 4.450768952928756e-06,
      "loss": 3.5178,
      "step": 21220
    },
    {
      "epoch": 0.215911865234375,
      "grad_norm": 8.820984840393066,
      "learning_rate": 4.450518844057289e-06,
      "loss": 3.3613,
      "step": 21225
    },
    {
      "epoch": 0.21596272786458334,
      "grad_norm": 16.864948272705078,
      "learning_rate": 4.4502686852822595e-06,
      "loss": 3.5213,
      "step": 21230
    },
    {
      "epoch": 0.21601359049479166,
      "grad_norm": 8.104302406311035,
      "learning_rate": 4.450018476610068e-06,
      "loss": 3.2945,
      "step": 21235
    },
    {
      "epoch": 0.216064453125,
      "grad_norm": 7.796335220336914,
      "learning_rate": 4.449768218047117e-06,
      "loss": 3.93,
      "step": 21240
    },
    {
      "epoch": 0.21611531575520834,
      "grad_norm": 12.010110855102539,
      "learning_rate": 4.449517909599807e-06,
      "loss": 3.0726,
      "step": 21245
    },
    {
      "epoch": 0.21616617838541666,
      "grad_norm": 7.897410869598389,
      "learning_rate": 4.449267551274543e-06,
      "loss": 3.2575,
      "step": 21250
    },
    {
      "epoch": 0.216217041015625,
      "grad_norm": 14.556855201721191,
      "learning_rate": 4.4490171430777315e-06,
      "loss": 3.2561,
      "step": 21255
    },
    {
      "epoch": 0.21626790364583334,
      "grad_norm": 15.189546585083008,
      "learning_rate": 4.448766685015778e-06,
      "loss": 3.2168,
      "step": 21260
    },
    {
      "epoch": 0.21631876627604166,
      "grad_norm": 9.562654495239258,
      "learning_rate": 4.448516177095092e-06,
      "loss": 3.527,
      "step": 21265
    },
    {
      "epoch": 0.21636962890625,
      "grad_norm": 14.520811080932617,
      "learning_rate": 4.448265619322081e-06,
      "loss": 3.447,
      "step": 21270
    },
    {
      "epoch": 0.21642049153645834,
      "grad_norm": 8.628990173339844,
      "learning_rate": 4.448015011703155e-06,
      "loss": 3.8215,
      "step": 21275
    },
    {
      "epoch": 0.21647135416666666,
      "grad_norm": 13.623302459716797,
      "learning_rate": 4.4477643542447265e-06,
      "loss": 3.2572,
      "step": 21280
    },
    {
      "epoch": 0.216522216796875,
      "grad_norm": 14.402774810791016,
      "learning_rate": 4.44751364695321e-06,
      "loss": 3.4455,
      "step": 21285
    },
    {
      "epoch": 0.21657307942708334,
      "grad_norm": 9.292447090148926,
      "learning_rate": 4.447262889835018e-06,
      "loss": 3.407,
      "step": 21290
    },
    {
      "epoch": 0.21662394205729166,
      "grad_norm": 13.544245719909668,
      "learning_rate": 4.447012082896566e-06,
      "loss": 3.1637,
      "step": 21295
    },
    {
      "epoch": 0.2166748046875,
      "grad_norm": 10.58362865447998,
      "learning_rate": 4.446761226144272e-06,
      "loss": 3.0985,
      "step": 21300
    },
    {
      "epoch": 0.21672566731770834,
      "grad_norm": 13.475932121276855,
      "learning_rate": 4.446510319584553e-06,
      "loss": 4.2408,
      "step": 21305
    },
    {
      "epoch": 0.21677652994791666,
      "grad_norm": 13.142467498779297,
      "learning_rate": 4.446259363223829e-06,
      "loss": 3.1587,
      "step": 21310
    },
    {
      "epoch": 0.216827392578125,
      "grad_norm": 16.452293395996094,
      "learning_rate": 4.44600835706852e-06,
      "loss": 3.46,
      "step": 21315
    },
    {
      "epoch": 0.21687825520833334,
      "grad_norm": 13.97862434387207,
      "learning_rate": 4.445757301125049e-06,
      "loss": 3.6838,
      "step": 21320
    },
    {
      "epoch": 0.21692911783854166,
      "grad_norm": 14.464639663696289,
      "learning_rate": 4.445506195399839e-06,
      "loss": 3.5322,
      "step": 21325
    },
    {
      "epoch": 0.21697998046875,
      "grad_norm": 9.015607833862305,
      "learning_rate": 4.445255039899313e-06,
      "loss": 3.4834,
      "step": 21330
    },
    {
      "epoch": 0.21703084309895834,
      "grad_norm": 9.246883392333984,
      "learning_rate": 4.4450038346298985e-06,
      "loss": 3.1686,
      "step": 21335
    },
    {
      "epoch": 0.21708170572916666,
      "grad_norm": 9.191157341003418,
      "learning_rate": 4.4447525795980224e-06,
      "loss": 3.4368,
      "step": 21340
    },
    {
      "epoch": 0.217132568359375,
      "grad_norm": 13.155099868774414,
      "learning_rate": 4.444501274810112e-06,
      "loss": 3.3268,
      "step": 21345
    },
    {
      "epoch": 0.21718343098958334,
      "grad_norm": 11.375064849853516,
      "learning_rate": 4.444249920272598e-06,
      "loss": 3.153,
      "step": 21350
    },
    {
      "epoch": 0.21723429361979166,
      "grad_norm": 10.894318580627441,
      "learning_rate": 4.443998515991909e-06,
      "loss": 3.6735,
      "step": 21355
    },
    {
      "epoch": 0.21728515625,
      "grad_norm": 11.658163070678711,
      "learning_rate": 4.443747061974481e-06,
      "loss": 3.1592,
      "step": 21360
    },
    {
      "epoch": 0.21733601888020834,
      "grad_norm": 15.226777076721191,
      "learning_rate": 4.443495558226744e-06,
      "loss": 3.4328,
      "step": 21365
    },
    {
      "epoch": 0.21738688151041666,
      "grad_norm": 12.534459114074707,
      "learning_rate": 4.4432440047551325e-06,
      "loss": 3.5584,
      "step": 21370
    },
    {
      "epoch": 0.217437744140625,
      "grad_norm": 16.684511184692383,
      "learning_rate": 4.442992401566085e-06,
      "loss": 3.4321,
      "step": 21375
    },
    {
      "epoch": 0.21748860677083334,
      "grad_norm": 10.544679641723633,
      "learning_rate": 4.442740748666038e-06,
      "loss": 3.3937,
      "step": 21380
    },
    {
      "epoch": 0.21753946940104166,
      "grad_norm": 14.9170560836792,
      "learning_rate": 4.442489046061429e-06,
      "loss": 3.4367,
      "step": 21385
    },
    {
      "epoch": 0.21759033203125,
      "grad_norm": 14.46596622467041,
      "learning_rate": 4.442237293758698e-06,
      "loss": 3.539,
      "step": 21390
    },
    {
      "epoch": 0.21764119466145834,
      "grad_norm": 11.867507934570312,
      "learning_rate": 4.4419854917642854e-06,
      "loss": 3.107,
      "step": 21395
    },
    {
      "epoch": 0.21769205729166666,
      "grad_norm": 17.437856674194336,
      "learning_rate": 4.441733640084636e-06,
      "loss": 3.3885,
      "step": 21400
    },
    {
      "epoch": 0.217742919921875,
      "grad_norm": 13.814154624938965,
      "learning_rate": 4.44148173872619e-06,
      "loss": 3.4518,
      "step": 21405
    },
    {
      "epoch": 0.21779378255208334,
      "grad_norm": 14.542771339416504,
      "learning_rate": 4.441229787695395e-06,
      "loss": 3.4028,
      "step": 21410
    },
    {
      "epoch": 0.21784464518229166,
      "grad_norm": 8.311295509338379,
      "learning_rate": 4.440977786998696e-06,
      "loss": 3.2889,
      "step": 21415
    },
    {
      "epoch": 0.2178955078125,
      "grad_norm": 7.098788261413574,
      "learning_rate": 4.44072573664254e-06,
      "loss": 3.1133,
      "step": 21420
    },
    {
      "epoch": 0.21794637044270834,
      "grad_norm": 12.52330493927002,
      "learning_rate": 4.440473636633376e-06,
      "loss": 3.0992,
      "step": 21425
    },
    {
      "epoch": 0.21799723307291666,
      "grad_norm": 11.01602554321289,
      "learning_rate": 4.4402214869776536e-06,
      "loss": 3.4034,
      "step": 21430
    },
    {
      "epoch": 0.218048095703125,
      "grad_norm": 13.289205551147461,
      "learning_rate": 4.439969287681826e-06,
      "loss": 3.6383,
      "step": 21435
    },
    {
      "epoch": 0.21809895833333334,
      "grad_norm": 11.71587085723877,
      "learning_rate": 4.4397170387523425e-06,
      "loss": 3.7132,
      "step": 21440
    },
    {
      "epoch": 0.21814982096354166,
      "grad_norm": 13.405722618103027,
      "learning_rate": 4.439464740195658e-06,
      "loss": 3.3589,
      "step": 21445
    },
    {
      "epoch": 0.21820068359375,
      "grad_norm": 8.185508728027344,
      "learning_rate": 4.439212392018228e-06,
      "loss": 3.276,
      "step": 21450
    },
    {
      "epoch": 0.21825154622395834,
      "grad_norm": 14.30631160736084,
      "learning_rate": 4.438959994226509e-06,
      "loss": 3.8529,
      "step": 21455
    },
    {
      "epoch": 0.21830240885416666,
      "grad_norm": 11.583393096923828,
      "learning_rate": 4.4387075468269595e-06,
      "loss": 3.1893,
      "step": 21460
    },
    {
      "epoch": 0.218353271484375,
      "grad_norm": 11.868514060974121,
      "learning_rate": 4.438455049826035e-06,
      "loss": 3.4185,
      "step": 21465
    },
    {
      "epoch": 0.21840413411458334,
      "grad_norm": 11.683053970336914,
      "learning_rate": 4.438202503230198e-06,
      "loss": 3.144,
      "step": 21470
    },
    {
      "epoch": 0.21845499674479166,
      "grad_norm": 12.934630393981934,
      "learning_rate": 4.437949907045909e-06,
      "loss": 3.142,
      "step": 21475
    },
    {
      "epoch": 0.218505859375,
      "grad_norm": 12.29751968383789,
      "learning_rate": 4.437697261279632e-06,
      "loss": 3.4487,
      "step": 21480
    },
    {
      "epoch": 0.21855672200520834,
      "grad_norm": 15.596671104431152,
      "learning_rate": 4.43744456593783e-06,
      "loss": 3.1818,
      "step": 21485
    },
    {
      "epoch": 0.21860758463541666,
      "grad_norm": 12.191415786743164,
      "learning_rate": 4.4371918210269665e-06,
      "loss": 3.4811,
      "step": 21490
    },
    {
      "epoch": 0.218658447265625,
      "grad_norm": 15.3618803024292,
      "learning_rate": 4.43693902655351e-06,
      "loss": 3.2084,
      "step": 21495
    },
    {
      "epoch": 0.21870930989583334,
      "grad_norm": 15.887072563171387,
      "learning_rate": 4.436686182523928e-06,
      "loss": 3.3075,
      "step": 21500
    },
    {
      "epoch": 0.21876017252604166,
      "grad_norm": 13.241227149963379,
      "learning_rate": 4.436433288944689e-06,
      "loss": 2.9199,
      "step": 21505
    },
    {
      "epoch": 0.21881103515625,
      "grad_norm": 9.028556823730469,
      "learning_rate": 4.436180345822263e-06,
      "loss": 3.3118,
      "step": 21510
    },
    {
      "epoch": 0.21886189778645834,
      "grad_norm": 7.4255452156066895,
      "learning_rate": 4.435927353163122e-06,
      "loss": 3.438,
      "step": 21515
    },
    {
      "epoch": 0.21891276041666666,
      "grad_norm": 8.402618408203125,
      "learning_rate": 4.435674310973739e-06,
      "loss": 3.6758,
      "step": 21520
    },
    {
      "epoch": 0.218963623046875,
      "grad_norm": 11.299454689025879,
      "learning_rate": 4.435421219260587e-06,
      "loss": 3.4711,
      "step": 21525
    },
    {
      "epoch": 0.21901448567708334,
      "grad_norm": 15.088027000427246,
      "learning_rate": 4.4351680780301425e-06,
      "loss": 3.558,
      "step": 21530
    },
    {
      "epoch": 0.21906534830729166,
      "grad_norm": 15.521208763122559,
      "learning_rate": 4.434914887288881e-06,
      "loss": 3.3314,
      "step": 21535
    },
    {
      "epoch": 0.2191162109375,
      "grad_norm": 14.964542388916016,
      "learning_rate": 4.4346616470432795e-06,
      "loss": 3.2707,
      "step": 21540
    },
    {
      "epoch": 0.21916707356770834,
      "grad_norm": 8.358473777770996,
      "learning_rate": 4.43440835729982e-06,
      "loss": 3.2384,
      "step": 21545
    },
    {
      "epoch": 0.21921793619791666,
      "grad_norm": 12.672173500061035,
      "learning_rate": 4.43415501806498e-06,
      "loss": 3.2498,
      "step": 21550
    },
    {
      "epoch": 0.219268798828125,
      "grad_norm": 13.793452262878418,
      "learning_rate": 4.433901629345243e-06,
      "loss": 3.5783,
      "step": 21555
    },
    {
      "epoch": 0.21931966145833334,
      "grad_norm": 14.273289680480957,
      "learning_rate": 4.433648191147091e-06,
      "loss": 3.5216,
      "step": 21560
    },
    {
      "epoch": 0.21937052408854166,
      "grad_norm": 15.174352645874023,
      "learning_rate": 4.433394703477009e-06,
      "loss": 3.5776,
      "step": 21565
    },
    {
      "epoch": 0.21942138671875,
      "grad_norm": 12.471453666687012,
      "learning_rate": 4.43314116634148e-06,
      "loss": 3.6904,
      "step": 21570
    },
    {
      "epoch": 0.21947224934895834,
      "grad_norm": 9.688103675842285,
      "learning_rate": 4.4328875797469944e-06,
      "loss": 3.4065,
      "step": 21575
    },
    {
      "epoch": 0.21952311197916666,
      "grad_norm": 14.465635299682617,
      "learning_rate": 4.432633943700038e-06,
      "loss": 3.5407,
      "step": 21580
    },
    {
      "epoch": 0.219573974609375,
      "grad_norm": 14.079458236694336,
      "learning_rate": 4.432380258207099e-06,
      "loss": 3.0966,
      "step": 21585
    },
    {
      "epoch": 0.21962483723958334,
      "grad_norm": 10.63652229309082,
      "learning_rate": 4.43212652327467e-06,
      "loss": 3.5664,
      "step": 21590
    },
    {
      "epoch": 0.21967569986979166,
      "grad_norm": 13.68020248413086,
      "learning_rate": 4.431872738909242e-06,
      "loss": 3.1991,
      "step": 21595
    },
    {
      "epoch": 0.2197265625,
      "grad_norm": 13.262728691101074,
      "learning_rate": 4.431618905117308e-06,
      "loss": 3.369,
      "step": 21600
    },
    {
      "epoch": 0.21977742513020834,
      "grad_norm": 10.705604553222656,
      "learning_rate": 4.431365021905361e-06,
      "loss": 3.725,
      "step": 21605
    },
    {
      "epoch": 0.21982828776041666,
      "grad_norm": 15.042895317077637,
      "learning_rate": 4.431111089279898e-06,
      "loss": 3.414,
      "step": 21610
    },
    {
      "epoch": 0.219879150390625,
      "grad_norm": 15.064384460449219,
      "learning_rate": 4.430857107247416e-06,
      "loss": 3.2864,
      "step": 21615
    },
    {
      "epoch": 0.21993001302083334,
      "grad_norm": 8.881093978881836,
      "learning_rate": 4.430603075814413e-06,
      "loss": 3.4096,
      "step": 21620
    },
    {
      "epoch": 0.21998087565104166,
      "grad_norm": 14.863802909851074,
      "learning_rate": 4.430348994987387e-06,
      "loss": 3.3993,
      "step": 21625
    },
    {
      "epoch": 0.22003173828125,
      "grad_norm": 12.123024940490723,
      "learning_rate": 4.430094864772841e-06,
      "loss": 3.7743,
      "step": 21630
    },
    {
      "epoch": 0.22008260091145834,
      "grad_norm": 10.595017433166504,
      "learning_rate": 4.4298406851772734e-06,
      "loss": 3.6968,
      "step": 21635
    },
    {
      "epoch": 0.22013346354166666,
      "grad_norm": 10.866302490234375,
      "learning_rate": 4.42958645620719e-06,
      "loss": 3.0981,
      "step": 21640
    },
    {
      "epoch": 0.220184326171875,
      "grad_norm": 13.092309951782227,
      "learning_rate": 4.429332177869094e-06,
      "loss": 3.1676,
      "step": 21645
    },
    {
      "epoch": 0.22023518880208334,
      "grad_norm": 12.232417106628418,
      "learning_rate": 4.429077850169493e-06,
      "loss": 3.4387,
      "step": 21650
    },
    {
      "epoch": 0.22028605143229166,
      "grad_norm": 14.34591293334961,
      "learning_rate": 4.428823473114891e-06,
      "loss": 3.4943,
      "step": 21655
    },
    {
      "epoch": 0.2203369140625,
      "grad_norm": 16.25722312927246,
      "learning_rate": 4.428569046711799e-06,
      "loss": 3.2681,
      "step": 21660
    },
    {
      "epoch": 0.22038777669270834,
      "grad_norm": 8.571758270263672,
      "learning_rate": 4.428314570966724e-06,
      "loss": 3.0704,
      "step": 21665
    },
    {
      "epoch": 0.22043863932291666,
      "grad_norm": 16.985572814941406,
      "learning_rate": 4.428060045886178e-06,
      "loss": 3.6197,
      "step": 21670
    },
    {
      "epoch": 0.220489501953125,
      "grad_norm": 9.714963912963867,
      "learning_rate": 4.427805471476673e-06,
      "loss": 3.2242,
      "step": 21675
    },
    {
      "epoch": 0.22054036458333334,
      "grad_norm": 8.596430778503418,
      "learning_rate": 4.427550847744721e-06,
      "loss": 3.3244,
      "step": 21680
    },
    {
      "epoch": 0.22059122721354166,
      "grad_norm": 12.92138957977295,
      "learning_rate": 4.4272961746968394e-06,
      "loss": 3.5212,
      "step": 21685
    },
    {
      "epoch": 0.22064208984375,
      "grad_norm": 11.397533416748047,
      "learning_rate": 4.427041452339541e-06,
      "loss": 3.32,
      "step": 21690
    },
    {
      "epoch": 0.22069295247395834,
      "grad_norm": 14.433199882507324,
      "learning_rate": 4.426786680679344e-06,
      "loss": 3.5267,
      "step": 21695
    },
    {
      "epoch": 0.22074381510416666,
      "grad_norm": 13.278440475463867,
      "learning_rate": 4.426531859722765e-06,
      "loss": 3.1954,
      "step": 21700
    },
    {
      "epoch": 0.220794677734375,
      "grad_norm": 13.766365051269531,
      "learning_rate": 4.426276989476327e-06,
      "loss": 3.2458,
      "step": 21705
    },
    {
      "epoch": 0.22084554036458334,
      "grad_norm": 12.732465744018555,
      "learning_rate": 4.4260220699465474e-06,
      "loss": 3.4943,
      "step": 21710
    },
    {
      "epoch": 0.22089640299479166,
      "grad_norm": 9.512740135192871,
      "learning_rate": 4.42576710113995e-06,
      "loss": 3.458,
      "step": 21715
    },
    {
      "epoch": 0.220947265625,
      "grad_norm": 9.880742073059082,
      "learning_rate": 4.425512083063058e-06,
      "loss": 3.0285,
      "step": 21720
    },
    {
      "epoch": 0.22099812825520834,
      "grad_norm": 11.119718551635742,
      "learning_rate": 4.425257015722396e-06,
      "loss": 2.956,
      "step": 21725
    },
    {
      "epoch": 0.22104899088541666,
      "grad_norm": 10.98831844329834,
      "learning_rate": 4.4250018991244895e-06,
      "loss": 3.2721,
      "step": 21730
    },
    {
      "epoch": 0.221099853515625,
      "grad_norm": 8.650774002075195,
      "learning_rate": 4.424746733275864e-06,
      "loss": 3.2025,
      "step": 21735
    },
    {
      "epoch": 0.22115071614583334,
      "grad_norm": 14.937943458557129,
      "learning_rate": 4.424491518183052e-06,
      "loss": 3.3219,
      "step": 21740
    },
    {
      "epoch": 0.22120157877604166,
      "grad_norm": 13.311105728149414,
      "learning_rate": 4.424236253852579e-06,
      "loss": 3.4122,
      "step": 21745
    },
    {
      "epoch": 0.22125244140625,
      "grad_norm": 13.529938697814941,
      "learning_rate": 4.423980940290978e-06,
      "loss": 3.0956,
      "step": 21750
    },
    {
      "epoch": 0.22130330403645834,
      "grad_norm": 8.813322067260742,
      "learning_rate": 4.423725577504781e-06,
      "loss": 3.3761,
      "step": 21755
    },
    {
      "epoch": 0.22135416666666666,
      "grad_norm": 12.712224006652832,
      "learning_rate": 4.423470165500521e-06,
      "loss": 2.9878,
      "step": 21760
    },
    {
      "epoch": 0.221405029296875,
      "grad_norm": 23.10104751586914,
      "learning_rate": 4.423214704284733e-06,
      "loss": 3.9616,
      "step": 21765
    },
    {
      "epoch": 0.22145589192708334,
      "grad_norm": 12.10202693939209,
      "learning_rate": 4.422959193863953e-06,
      "loss": 3.2412,
      "step": 21770
    },
    {
      "epoch": 0.22150675455729166,
      "grad_norm": 13.642451286315918,
      "learning_rate": 4.422703634244716e-06,
      "loss": 3.6663,
      "step": 21775
    },
    {
      "epoch": 0.2215576171875,
      "grad_norm": 13.663246154785156,
      "learning_rate": 4.422448025433564e-06,
      "loss": 3.497,
      "step": 21780
    },
    {
      "epoch": 0.22160847981770834,
      "grad_norm": 12.732048988342285,
      "learning_rate": 4.422192367437034e-06,
      "loss": 3.459,
      "step": 21785
    },
    {
      "epoch": 0.22165934244791666,
      "grad_norm": 10.020483016967773,
      "learning_rate": 4.421936660261668e-06,
      "loss": 3.1006,
      "step": 21790
    },
    {
      "epoch": 0.221710205078125,
      "grad_norm": 9.460532188415527,
      "learning_rate": 4.421680903914009e-06,
      "loss": 3.3341,
      "step": 21795
    },
    {
      "epoch": 0.22176106770833334,
      "grad_norm": 7.5237860679626465,
      "learning_rate": 4.421425098400598e-06,
      "loss": 3.2786,
      "step": 21800
    },
    {
      "epoch": 0.22181193033854166,
      "grad_norm": 13.649006843566895,
      "learning_rate": 4.421169243727983e-06,
      "loss": 3.2036,
      "step": 21805
    },
    {
      "epoch": 0.22186279296875,
      "grad_norm": 15.066064834594727,
      "learning_rate": 4.420913339902707e-06,
      "loss": 3.2731,
      "step": 21810
    },
    {
      "epoch": 0.22191365559895834,
      "grad_norm": 9.609551429748535,
      "learning_rate": 4.420657386931319e-06,
      "loss": 3.3248,
      "step": 21815
    },
    {
      "epoch": 0.22196451822916666,
      "grad_norm": 9.863035202026367,
      "learning_rate": 4.420401384820367e-06,
      "loss": 3.4538,
      "step": 21820
    },
    {
      "epoch": 0.222015380859375,
      "grad_norm": 7.887870788574219,
      "learning_rate": 4.4201453335764e-06,
      "loss": 3.3845,
      "step": 21825
    },
    {
      "epoch": 0.22206624348958334,
      "grad_norm": 13.143853187561035,
      "learning_rate": 4.4198892332059705e-06,
      "loss": 3.3623,
      "step": 21830
    },
    {
      "epoch": 0.22211710611979166,
      "grad_norm": 15.148838996887207,
      "learning_rate": 4.41963308371563e-06,
      "loss": 4.2309,
      "step": 21835
    },
    {
      "epoch": 0.22216796875,
      "grad_norm": 15.139830589294434,
      "learning_rate": 4.419376885111932e-06,
      "loss": 3.1964,
      "step": 21840
    },
    {
      "epoch": 0.22221883138020834,
      "grad_norm": 15.096305847167969,
      "learning_rate": 4.419120637401432e-06,
      "loss": 3.2714,
      "step": 21845
    },
    {
      "epoch": 0.22226969401041666,
      "grad_norm": 15.630721092224121,
      "learning_rate": 4.418864340590684e-06,
      "loss": 3.4326,
      "step": 21850
    },
    {
      "epoch": 0.222320556640625,
      "grad_norm": 12.31386661529541,
      "learning_rate": 4.418607994686248e-06,
      "loss": 3.7328,
      "step": 21855
    },
    {
      "epoch": 0.22237141927083334,
      "grad_norm": 7.927239894866943,
      "learning_rate": 4.41835159969468e-06,
      "loss": 3.5338,
      "step": 21860
    },
    {
      "epoch": 0.22242228190104166,
      "grad_norm": 9.581936836242676,
      "learning_rate": 4.418095155622542e-06,
      "loss": 3.3615,
      "step": 21865
    },
    {
      "epoch": 0.22247314453125,
      "grad_norm": 14.914663314819336,
      "learning_rate": 4.4178386624763935e-06,
      "loss": 3.5794,
      "step": 21870
    },
    {
      "epoch": 0.22252400716145834,
      "grad_norm": 9.037361145019531,
      "learning_rate": 4.417582120262798e-06,
      "loss": 3.6717,
      "step": 21875
    },
    {
      "epoch": 0.22257486979166666,
      "grad_norm": 13.858770370483398,
      "learning_rate": 4.4173255289883175e-06,
      "loss": 3.3966,
      "step": 21880
    },
    {
      "epoch": 0.222625732421875,
      "grad_norm": 13.544123649597168,
      "learning_rate": 4.41706888865952e-06,
      "loss": 3.4346,
      "step": 21885
    },
    {
      "epoch": 0.22267659505208334,
      "grad_norm": 14.441933631896973,
      "learning_rate": 4.416812199282966e-06,
      "loss": 3.564,
      "step": 21890
    },
    {
      "epoch": 0.22272745768229166,
      "grad_norm": 8.17879581451416,
      "learning_rate": 4.416555460865228e-06,
      "loss": 3.2117,
      "step": 21895
    },
    {
      "epoch": 0.2227783203125,
      "grad_norm": 16.045257568359375,
      "learning_rate": 4.416298673412874e-06,
      "loss": 3.6617,
      "step": 21900
    },
    {
      "epoch": 0.22282918294270834,
      "grad_norm": 8.84123706817627,
      "learning_rate": 4.416041836932471e-06,
      "loss": 3.3166,
      "step": 21905
    },
    {
      "epoch": 0.22288004557291666,
      "grad_norm": 8.175050735473633,
      "learning_rate": 4.415784951430592e-06,
      "loss": 3.8243,
      "step": 21910
    },
    {
      "epoch": 0.222930908203125,
      "grad_norm": 12.569129943847656,
      "learning_rate": 4.415528016913809e-06,
      "loss": 3.275,
      "step": 21915
    },
    {
      "epoch": 0.22298177083333334,
      "grad_norm": 9.564921379089355,
      "learning_rate": 4.415271033388696e-06,
      "loss": 2.9735,
      "step": 21920
    },
    {
      "epoch": 0.22303263346354166,
      "grad_norm": 15.479812622070312,
      "learning_rate": 4.415014000861828e-06,
      "loss": 3.6172,
      "step": 21925
    },
    {
      "epoch": 0.22308349609375,
      "grad_norm": 8.315549850463867,
      "learning_rate": 4.41475691933978e-06,
      "loss": 3.412,
      "step": 21930
    },
    {
      "epoch": 0.22313435872395834,
      "grad_norm": 9.115899085998535,
      "learning_rate": 4.41449978882913e-06,
      "loss": 3.4299,
      "step": 21935
    },
    {
      "epoch": 0.22318522135416666,
      "grad_norm": 13.794336318969727,
      "learning_rate": 4.4142426093364575e-06,
      "loss": 3.1094,
      "step": 21940
    },
    {
      "epoch": 0.223236083984375,
      "grad_norm": 12.609228134155273,
      "learning_rate": 4.413985380868341e-06,
      "loss": 3.367,
      "step": 21945
    },
    {
      "epoch": 0.22328694661458334,
      "grad_norm": 14.703821182250977,
      "learning_rate": 4.413728103431362e-06,
      "loss": 3.9406,
      "step": 21950
    },
    {
      "epoch": 0.22333780924479166,
      "grad_norm": 15.378171920776367,
      "learning_rate": 4.413470777032104e-06,
      "loss": 2.9513,
      "step": 21955
    },
    {
      "epoch": 0.223388671875,
      "grad_norm": 11.910841941833496,
      "learning_rate": 4.413213401677149e-06,
      "loss": 3.1509,
      "step": 21960
    },
    {
      "epoch": 0.22343953450520834,
      "grad_norm": 14.728949546813965,
      "learning_rate": 4.412955977373082e-06,
      "loss": 3.3036,
      "step": 21965
    },
    {
      "epoch": 0.22349039713541666,
      "grad_norm": 16.039880752563477,
      "learning_rate": 4.412698504126491e-06,
      "loss": 3.3996,
      "step": 21970
    },
    {
      "epoch": 0.223541259765625,
      "grad_norm": 9.042581558227539,
      "learning_rate": 4.412440981943962e-06,
      "loss": 3.3724,
      "step": 21975
    },
    {
      "epoch": 0.22359212239583334,
      "grad_norm": 11.22207260131836,
      "learning_rate": 4.4121834108320824e-06,
      "loss": 3.3767,
      "step": 21980
    },
    {
      "epoch": 0.22364298502604166,
      "grad_norm": 9.639937400817871,
      "learning_rate": 4.411925790797444e-06,
      "loss": 3.3927,
      "step": 21985
    },
    {
      "epoch": 0.22369384765625,
      "grad_norm": 11.378259658813477,
      "learning_rate": 4.4116681218466386e-06,
      "loss": 3.2632,
      "step": 21990
    },
    {
      "epoch": 0.22374471028645834,
      "grad_norm": 12.211450576782227,
      "learning_rate": 4.411410403986257e-06,
      "loss": 3.1877,
      "step": 21995
    },
    {
      "epoch": 0.22379557291666666,
      "grad_norm": 13.47460651397705,
      "learning_rate": 4.411152637222893e-06,
      "loss": 3.4715,
      "step": 22000
    },
    {
      "epoch": 0.223846435546875,
      "grad_norm": 9.644412994384766,
      "learning_rate": 4.410894821563141e-06,
      "loss": 3.2384,
      "step": 22005
    },
    {
      "epoch": 0.22389729817708334,
      "grad_norm": 12.741238594055176,
      "learning_rate": 4.410636957013599e-06,
      "loss": 3.4233,
      "step": 22010
    },
    {
      "epoch": 0.22394816080729166,
      "grad_norm": 18.5694580078125,
      "learning_rate": 4.410379043580863e-06,
      "loss": 3.4207,
      "step": 22015
    },
    {
      "epoch": 0.2239990234375,
      "grad_norm": 13.11888313293457,
      "learning_rate": 4.410121081271532e-06,
      "loss": 3.4742,
      "step": 22020
    },
    {
      "epoch": 0.22404988606770834,
      "grad_norm": 8.950638771057129,
      "learning_rate": 4.4098630700922055e-06,
      "loss": 3.2443,
      "step": 22025
    },
    {
      "epoch": 0.22410074869791666,
      "grad_norm": 9.21252727508545,
      "learning_rate": 4.409605010049486e-06,
      "loss": 3.2816,
      "step": 22030
    },
    {
      "epoch": 0.224151611328125,
      "grad_norm": 11.447957992553711,
      "learning_rate": 4.409346901149973e-06,
      "loss": 3.6363,
      "step": 22035
    },
    {
      "epoch": 0.22420247395833334,
      "grad_norm": 14.11512565612793,
      "learning_rate": 4.409088743400274e-06,
      "loss": 3.0668,
      "step": 22040
    },
    {
      "epoch": 0.22425333658854166,
      "grad_norm": 11.56049919128418,
      "learning_rate": 4.408830536806991e-06,
      "loss": 3.6871,
      "step": 22045
    },
    {
      "epoch": 0.22430419921875,
      "grad_norm": 16.231426239013672,
      "learning_rate": 4.408572281376732e-06,
      "loss": 3.8049,
      "step": 22050
    },
    {
      "epoch": 0.22435506184895834,
      "grad_norm": 8.128877639770508,
      "learning_rate": 4.408313977116103e-06,
      "loss": 3.3654,
      "step": 22055
    },
    {
      "epoch": 0.22440592447916666,
      "grad_norm": 7.407121658325195,
      "learning_rate": 4.408055624031713e-06,
      "loss": 3.2684,
      "step": 22060
    },
    {
      "epoch": 0.224456787109375,
      "grad_norm": 17.4244384765625,
      "learning_rate": 4.407797222130173e-06,
      "loss": 3.0736,
      "step": 22065
    },
    {
      "epoch": 0.22450764973958334,
      "grad_norm": 15.093779563903809,
      "learning_rate": 4.407538771418092e-06,
      "loss": 3.4779,
      "step": 22070
    },
    {
      "epoch": 0.22455851236979166,
      "grad_norm": 10.524264335632324,
      "learning_rate": 4.407280271902084e-06,
      "loss": 3.2996,
      "step": 22075
    },
    {
      "epoch": 0.224609375,
      "grad_norm": 9.79942512512207,
      "learning_rate": 4.407021723588764e-06,
      "loss": 3.4059,
      "step": 22080
    },
    {
      "epoch": 0.22466023763020834,
      "grad_norm": 15.806706428527832,
      "learning_rate": 4.4067631264847436e-06,
      "loss": 3.5924,
      "step": 22085
    },
    {
      "epoch": 0.22471110026041666,
      "grad_norm": 17.772584915161133,
      "learning_rate": 4.406504480596641e-06,
      "loss": 3.6493,
      "step": 22090
    },
    {
      "epoch": 0.224761962890625,
      "grad_norm": 9.457268714904785,
      "learning_rate": 4.406245785931073e-06,
      "loss": 3.3044,
      "step": 22095
    },
    {
      "epoch": 0.22481282552083334,
      "grad_norm": 12.059775352478027,
      "learning_rate": 4.405987042494658e-06,
      "loss": 3.8077,
      "step": 22100
    },
    {
      "epoch": 0.22486368815104166,
      "grad_norm": 11.504251480102539,
      "learning_rate": 4.405728250294017e-06,
      "loss": 3.5037,
      "step": 22105
    },
    {
      "epoch": 0.22491455078125,
      "grad_norm": 9.41455078125,
      "learning_rate": 4.40546940933577e-06,
      "loss": 3.6421,
      "step": 22110
    },
    {
      "epoch": 0.22496541341145834,
      "grad_norm": 10.715293884277344,
      "learning_rate": 4.40521051962654e-06,
      "loss": 3.6148,
      "step": 22115
    },
    {
      "epoch": 0.22501627604166666,
      "grad_norm": 13.229304313659668,
      "learning_rate": 4.40495158117295e-06,
      "loss": 3.4736,
      "step": 22120
    },
    {
      "epoch": 0.225067138671875,
      "grad_norm": 15.428013801574707,
      "learning_rate": 4.404692593981626e-06,
      "loss": 3.6475,
      "step": 22125
    },
    {
      "epoch": 0.22511800130208334,
      "grad_norm": 16.025081634521484,
      "learning_rate": 4.404433558059193e-06,
      "loss": 3.1062,
      "step": 22130
    },
    {
      "epoch": 0.22516886393229166,
      "grad_norm": 16.080984115600586,
      "learning_rate": 4.404174473412279e-06,
      "loss": 3.6567,
      "step": 22135
    },
    {
      "epoch": 0.2252197265625,
      "grad_norm": 8.873684883117676,
      "learning_rate": 4.403915340047512e-06,
      "loss": 3.2109,
      "step": 22140
    },
    {
      "epoch": 0.22527058919270834,
      "grad_norm": 11.24513053894043,
      "learning_rate": 4.403656157971523e-06,
      "loss": 3.7436,
      "step": 22145
    },
    {
      "epoch": 0.22532145182291666,
      "grad_norm": 13.644927024841309,
      "learning_rate": 4.403396927190942e-06,
      "loss": 3.2305,
      "step": 22150
    },
    {
      "epoch": 0.225372314453125,
      "grad_norm": 6.955374717712402,
      "learning_rate": 4.403137647712401e-06,
      "loss": 3.3256,
      "step": 22155
    },
    {
      "epoch": 0.22542317708333334,
      "grad_norm": 12.21574878692627,
      "learning_rate": 4.402878319542536e-06,
      "loss": 3.2371,
      "step": 22160
    },
    {
      "epoch": 0.22547403971354166,
      "grad_norm": 13.770991325378418,
      "learning_rate": 4.402618942687979e-06,
      "loss": 3.147,
      "step": 22165
    },
    {
      "epoch": 0.22552490234375,
      "grad_norm": 11.066407203674316,
      "learning_rate": 4.402359517155368e-06,
      "loss": 3.194,
      "step": 22170
    },
    {
      "epoch": 0.22557576497395834,
      "grad_norm": 9.962047576904297,
      "learning_rate": 4.40210004295134e-06,
      "loss": 3.1988,
      "step": 22175
    },
    {
      "epoch": 0.22562662760416666,
      "grad_norm": 14.844084739685059,
      "learning_rate": 4.401840520082532e-06,
      "loss": 3.2153,
      "step": 22180
    },
    {
      "epoch": 0.225677490234375,
      "grad_norm": 13.172402381896973,
      "learning_rate": 4.401580948555586e-06,
      "loss": 3.3641,
      "step": 22185
    },
    {
      "epoch": 0.22572835286458334,
      "grad_norm": 7.525303363800049,
      "learning_rate": 4.401321328377142e-06,
      "loss": 3.2718,
      "step": 22190
    },
    {
      "epoch": 0.22577921549479166,
      "grad_norm": 11.506841659545898,
      "learning_rate": 4.401061659553843e-06,
      "loss": 2.9344,
      "step": 22195
    },
    {
      "epoch": 0.225830078125,
      "grad_norm": 13.197406768798828,
      "learning_rate": 4.400801942092331e-06,
      "loss": 3.0663,
      "step": 22200
    },
    {
      "epoch": 0.22588094075520834,
      "grad_norm": 8.670125961303711,
      "learning_rate": 4.400542175999253e-06,
      "loss": 3.2565,
      "step": 22205
    },
    {
      "epoch": 0.22593180338541666,
      "grad_norm": 11.305577278137207,
      "learning_rate": 4.400282361281253e-06,
      "loss": 3.3172,
      "step": 22210
    },
    {
      "epoch": 0.225982666015625,
      "grad_norm": 13.321239471435547,
      "learning_rate": 4.400022497944979e-06,
      "loss": 3.5229,
      "step": 22215
    },
    {
      "epoch": 0.22603352864583334,
      "grad_norm": 10.876173973083496,
      "learning_rate": 4.3997625859970805e-06,
      "loss": 3.6165,
      "step": 22220
    },
    {
      "epoch": 0.22608439127604166,
      "grad_norm": 13.838961601257324,
      "learning_rate": 4.399502625444206e-06,
      "loss": 3.4273,
      "step": 22225
    },
    {
      "epoch": 0.22613525390625,
      "grad_norm": 11.505608558654785,
      "learning_rate": 4.399242616293007e-06,
      "loss": 3.6678,
      "step": 22230
    },
    {
      "epoch": 0.22618611653645834,
      "grad_norm": 12.19419002532959,
      "learning_rate": 4.398982558550137e-06,
      "loss": 3.0976,
      "step": 22235
    },
    {
      "epoch": 0.22623697916666666,
      "grad_norm": 9.771162033081055,
      "learning_rate": 4.398722452222247e-06,
      "loss": 3.7784,
      "step": 22240
    },
    {
      "epoch": 0.226287841796875,
      "grad_norm": 11.823652267456055,
      "learning_rate": 4.398462297315993e-06,
      "loss": 3.7503,
      "step": 22245
    },
    {
      "epoch": 0.22633870442708334,
      "grad_norm": 12.599007606506348,
      "learning_rate": 4.398202093838032e-06,
      "loss": 3.747,
      "step": 22250
    },
    {
      "epoch": 0.22638956705729166,
      "grad_norm": 8.704680442810059,
      "learning_rate": 4.3979418417950195e-06,
      "loss": 2.9508,
      "step": 22255
    },
    {
      "epoch": 0.2264404296875,
      "grad_norm": 8.176292419433594,
      "learning_rate": 4.397681541193616e-06,
      "loss": 3.0066,
      "step": 22260
    },
    {
      "epoch": 0.22649129231770834,
      "grad_norm": 12.29770278930664,
      "learning_rate": 4.397421192040478e-06,
      "loss": 3.3324,
      "step": 22265
    },
    {
      "epoch": 0.22654215494791666,
      "grad_norm": 14.355680465698242,
      "learning_rate": 4.3971607943422715e-06,
      "loss": 3.0472,
      "step": 22270
    },
    {
      "epoch": 0.226593017578125,
      "grad_norm": 8.760498046875,
      "learning_rate": 4.396900348105654e-06,
      "loss": 3.1807,
      "step": 22275
    },
    {
      "epoch": 0.22664388020833334,
      "grad_norm": 13.894689559936523,
      "learning_rate": 4.39663985333729e-06,
      "loss": 3.1082,
      "step": 22280
    },
    {
      "epoch": 0.22669474283854166,
      "grad_norm": 8.41318130493164,
      "learning_rate": 4.396379310043847e-06,
      "loss": 3.2702,
      "step": 22285
    },
    {
      "epoch": 0.22674560546875,
      "grad_norm": 10.254307746887207,
      "learning_rate": 4.3961187182319876e-06,
      "loss": 3.4133,
      "step": 22290
    },
    {
      "epoch": 0.22679646809895834,
      "grad_norm": 11.506667137145996,
      "learning_rate": 4.39585807790838e-06,
      "loss": 3.3754,
      "step": 22295
    },
    {
      "epoch": 0.22684733072916666,
      "grad_norm": 14.916718482971191,
      "learning_rate": 4.3955973890796936e-06,
      "loss": 3.4474,
      "step": 22300
    },
    {
      "epoch": 0.226898193359375,
      "grad_norm": 13.130877494812012,
      "learning_rate": 4.3953366517525966e-06,
      "loss": 3.1012,
      "step": 22305
    },
    {
      "epoch": 0.22694905598958334,
      "grad_norm": 16.143596649169922,
      "learning_rate": 4.39507586593376e-06,
      "loss": 3.3571,
      "step": 22310
    },
    {
      "epoch": 0.22699991861979166,
      "grad_norm": 13.136238098144531,
      "learning_rate": 4.394815031629858e-06,
      "loss": 3.8295,
      "step": 22315
    },
    {
      "epoch": 0.22705078125,
      "grad_norm": 10.161462783813477,
      "learning_rate": 4.394554148847562e-06,
      "loss": 3.5558,
      "step": 22320
    },
    {
      "epoch": 0.22710164388020834,
      "grad_norm": 13.627985000610352,
      "learning_rate": 4.394293217593547e-06,
      "loss": 3.3345,
      "step": 22325
    },
    {
      "epoch": 0.22715250651041666,
      "grad_norm": 13.85934829711914,
      "learning_rate": 4.394032237874488e-06,
      "loss": 3.3807,
      "step": 22330
    },
    {
      "epoch": 0.227203369140625,
      "grad_norm": 7.913649559020996,
      "learning_rate": 4.393771209697065e-06,
      "loss": 3.1791,
      "step": 22335
    },
    {
      "epoch": 0.22725423177083334,
      "grad_norm": 12.486056327819824,
      "learning_rate": 4.393510133067954e-06,
      "loss": 3.2871,
      "step": 22340
    },
    {
      "epoch": 0.22730509440104166,
      "grad_norm": 9.422446250915527,
      "learning_rate": 4.393249007993834e-06,
      "loss": 3.2929,
      "step": 22345
    },
    {
      "epoch": 0.22735595703125,
      "grad_norm": 13.53770923614502,
      "learning_rate": 4.392987834481388e-06,
      "loss": 3.753,
      "step": 22350
    },
    {
      "epoch": 0.22740681966145834,
      "grad_norm": 14.302570343017578,
      "learning_rate": 4.3927266125372966e-06,
      "loss": 3.7092,
      "step": 22355
    },
    {
      "epoch": 0.22745768229166666,
      "grad_norm": 9.646662712097168,
      "learning_rate": 4.392465342168244e-06,
      "loss": 3.1584,
      "step": 22360
    },
    {
      "epoch": 0.227508544921875,
      "grad_norm": 12.217989921569824,
      "learning_rate": 4.392204023380913e-06,
      "loss": 3.2846,
      "step": 22365
    },
    {
      "epoch": 0.22755940755208334,
      "grad_norm": 15.172819137573242,
      "learning_rate": 4.391942656181991e-06,
      "loss": 3.232,
      "step": 22370
    },
    {
      "epoch": 0.22761027018229166,
      "grad_norm": 12.54372787475586,
      "learning_rate": 4.391681240578165e-06,
      "loss": 3.7518,
      "step": 22375
    },
    {
      "epoch": 0.2276611328125,
      "grad_norm": 14.048870086669922,
      "learning_rate": 4.391419776576123e-06,
      "loss": 3.4796,
      "step": 22380
    },
    {
      "epoch": 0.22771199544270834,
      "grad_norm": 15.415027618408203,
      "learning_rate": 4.391158264182553e-06,
      "loss": 3.2247,
      "step": 22385
    },
    {
      "epoch": 0.22776285807291666,
      "grad_norm": 8.710091590881348,
      "learning_rate": 4.390896703404148e-06,
      "loss": 3.7823,
      "step": 22390
    },
    {
      "epoch": 0.227813720703125,
      "grad_norm": 8.83072566986084,
      "learning_rate": 4.390635094247599e-06,
      "loss": 3.5134,
      "step": 22395
    },
    {
      "epoch": 0.22786458333333334,
      "grad_norm": 13.684968948364258,
      "learning_rate": 4.3903734367196e-06,
      "loss": 3.1468,
      "step": 22400
    },
    {
      "epoch": 0.22791544596354166,
      "grad_norm": 10.560220718383789,
      "learning_rate": 4.390111730826844e-06,
      "loss": 3.2929,
      "step": 22405
    },
    {
      "epoch": 0.22796630859375,
      "grad_norm": 9.37875747680664,
      "learning_rate": 4.389849976576027e-06,
      "loss": 3.3213,
      "step": 22410
    },
    {
      "epoch": 0.22801717122395834,
      "grad_norm": 12.373004913330078,
      "learning_rate": 4.389588173973847e-06,
      "loss": 3.6029,
      "step": 22415
    },
    {
      "epoch": 0.22806803385416666,
      "grad_norm": 11.633017539978027,
      "learning_rate": 4.389326323027001e-06,
      "loss": 3.4954,
      "step": 22420
    },
    {
      "epoch": 0.228118896484375,
      "grad_norm": 9.542677879333496,
      "learning_rate": 4.389064423742188e-06,
      "loss": 3.1426,
      "step": 22425
    },
    {
      "epoch": 0.22816975911458334,
      "grad_norm": 11.222747802734375,
      "learning_rate": 4.388802476126112e-06,
      "loss": 3.5985,
      "step": 22430
    },
    {
      "epoch": 0.22822062174479166,
      "grad_norm": 11.572148323059082,
      "learning_rate": 4.3885404801854704e-06,
      "loss": 3.5769,
      "step": 22435
    },
    {
      "epoch": 0.228271484375,
      "grad_norm": 9.974691390991211,
      "learning_rate": 4.388278435926968e-06,
      "loss": 4.3666,
      "step": 22440
    },
    {
      "epoch": 0.22832234700520834,
      "grad_norm": 15.239340782165527,
      "learning_rate": 4.3880163433573105e-06,
      "loss": 3.0226,
      "step": 22445
    },
    {
      "epoch": 0.22837320963541666,
      "grad_norm": 14.125506401062012,
      "learning_rate": 4.387754202483201e-06,
      "loss": 3.4135,
      "step": 22450
    },
    {
      "epoch": 0.228424072265625,
      "grad_norm": 12.766975402832031,
      "learning_rate": 4.3874920133113495e-06,
      "loss": 3.422,
      "step": 22455
    },
    {
      "epoch": 0.22847493489583334,
      "grad_norm": 7.391766548156738,
      "learning_rate": 4.3872297758484604e-06,
      "loss": 3.0828,
      "step": 22460
    },
    {
      "epoch": 0.22852579752604166,
      "grad_norm": 13.441563606262207,
      "learning_rate": 4.386967490101246e-06,
      "loss": 3.1743,
      "step": 22465
    },
    {
      "epoch": 0.22857666015625,
      "grad_norm": 13.292790412902832,
      "learning_rate": 4.3867051560764154e-06,
      "loss": 3.1812,
      "step": 22470
    },
    {
      "epoch": 0.22862752278645834,
      "grad_norm": 13.288102149963379,
      "learning_rate": 4.38644277378068e-06,
      "loss": 3.6401,
      "step": 22475
    },
    {
      "epoch": 0.22867838541666666,
      "grad_norm": 13.77042293548584,
      "learning_rate": 4.386180343220754e-06,
      "loss": 3.0652,
      "step": 22480
    },
    {
      "epoch": 0.228729248046875,
      "grad_norm": 12.693780899047852,
      "learning_rate": 4.3859178644033505e-06,
      "loss": 3.6637,
      "step": 22485
    },
    {
      "epoch": 0.22878011067708334,
      "grad_norm": 11.478263854980469,
      "learning_rate": 4.3856553373351865e-06,
      "loss": 3.0771,
      "step": 22490
    },
    {
      "epoch": 0.22883097330729166,
      "grad_norm": 13.068379402160645,
      "learning_rate": 4.385392762022976e-06,
      "loss": 3.3229,
      "step": 22495
    },
    {
      "epoch": 0.2288818359375,
      "grad_norm": 13.19984245300293,
      "learning_rate": 4.38513013847344e-06,
      "loss": 3.1831,
      "step": 22500
    },
    {
      "epoch": 0.22893269856770834,
      "grad_norm": 13.603717803955078,
      "learning_rate": 4.384867466693295e-06,
      "loss": 3.3175,
      "step": 22505
    },
    {
      "epoch": 0.22898356119791666,
      "grad_norm": 11.724016189575195,
      "learning_rate": 4.384604746689264e-06,
      "loss": 3.3074,
      "step": 22510
    },
    {
      "epoch": 0.229034423828125,
      "grad_norm": 12.843661308288574,
      "learning_rate": 4.384341978468067e-06,
      "loss": 3.1878,
      "step": 22515
    },
    {
      "epoch": 0.22908528645833334,
      "grad_norm": 16.34494400024414,
      "learning_rate": 4.384079162036426e-06,
      "loss": 4.0106,
      "step": 22520
    },
    {
      "epoch": 0.22913614908854166,
      "grad_norm": 13.975119590759277,
      "learning_rate": 4.383816297401068e-06,
      "loss": 3.3851,
      "step": 22525
    },
    {
      "epoch": 0.22918701171875,
      "grad_norm": 13.665790557861328,
      "learning_rate": 4.383553384568715e-06,
      "loss": 3.5789,
      "step": 22530
    },
    {
      "epoch": 0.22923787434895834,
      "grad_norm": 11.161511421203613,
      "learning_rate": 4.383290423546095e-06,
      "loss": 3.299,
      "step": 22535
    },
    {
      "epoch": 0.22928873697916666,
      "grad_norm": 15.252664566040039,
      "learning_rate": 4.383027414339936e-06,
      "loss": 3.2013,
      "step": 22540
    },
    {
      "epoch": 0.229339599609375,
      "grad_norm": 13.857137680053711,
      "learning_rate": 4.382764356956968e-06,
      "loss": 3.5467,
      "step": 22545
    },
    {
      "epoch": 0.22939046223958334,
      "grad_norm": 10.059674263000488,
      "learning_rate": 4.38250125140392e-06,
      "loss": 3.4946,
      "step": 22550
    },
    {
      "epoch": 0.22944132486979166,
      "grad_norm": 10.169170379638672,
      "learning_rate": 4.382238097687524e-06,
      "loss": 3.5028,
      "step": 22555
    },
    {
      "epoch": 0.2294921875,
      "grad_norm": 14.963750839233398,
      "learning_rate": 4.381974895814511e-06,
      "loss": 3.2772,
      "step": 22560
    },
    {
      "epoch": 0.22954305013020834,
      "grad_norm": 22.403776168823242,
      "learning_rate": 4.381711645791617e-06,
      "loss": 3.4373,
      "step": 22565
    },
    {
      "epoch": 0.22959391276041666,
      "grad_norm": 14.73482894897461,
      "learning_rate": 4.3814483476255764e-06,
      "loss": 3.1332,
      "step": 22570
    },
    {
      "epoch": 0.229644775390625,
      "grad_norm": 9.600224494934082,
      "learning_rate": 4.381185001323126e-06,
      "loss": 2.9423,
      "step": 22575
    },
    {
      "epoch": 0.22969563802083334,
      "grad_norm": 12.039507865905762,
      "learning_rate": 4.380921606891003e-06,
      "loss": 3.3055,
      "step": 22580
    },
    {
      "epoch": 0.22974650065104166,
      "grad_norm": 13.997042655944824,
      "learning_rate": 4.3806581643359465e-06,
      "loss": 3.2969,
      "step": 22585
    },
    {
      "epoch": 0.22979736328125,
      "grad_norm": 6.661002159118652,
      "learning_rate": 4.380394673664697e-06,
      "loss": 3.3523,
      "step": 22590
    },
    {
      "epoch": 0.22984822591145834,
      "grad_norm": 13.084622383117676,
      "learning_rate": 4.380131134883996e-06,
      "loss": 3.6514,
      "step": 22595
    },
    {
      "epoch": 0.22989908854166666,
      "grad_norm": 8.07689094543457,
      "learning_rate": 4.379867548000585e-06,
      "loss": 4.1248,
      "step": 22600
    },
    {
      "epoch": 0.229949951171875,
      "grad_norm": 12.517134666442871,
      "learning_rate": 4.3796039130212085e-06,
      "loss": 3.6931,
      "step": 22605
    },
    {
      "epoch": 0.23000081380208334,
      "grad_norm": 7.653154373168945,
      "learning_rate": 4.379340229952611e-06,
      "loss": 3.2468,
      "step": 22610
    },
    {
      "epoch": 0.23005167643229166,
      "grad_norm": 12.616921424865723,
      "learning_rate": 4.379076498801539e-06,
      "loss": 3.6261,
      "step": 22615
    },
    {
      "epoch": 0.2301025390625,
      "grad_norm": 11.284850120544434,
      "learning_rate": 4.378812719574741e-06,
      "loss": 3.9268,
      "step": 22620
    },
    {
      "epoch": 0.23015340169270834,
      "grad_norm": 12.80991268157959,
      "learning_rate": 4.378548892278964e-06,
      "loss": 3.0249,
      "step": 22625
    },
    {
      "epoch": 0.23020426432291666,
      "grad_norm": 12.30255126953125,
      "learning_rate": 4.378285016920959e-06,
      "loss": 3.6609,
      "step": 22630
    },
    {
      "epoch": 0.230255126953125,
      "grad_norm": 9.481846809387207,
      "learning_rate": 4.378021093507477e-06,
      "loss": 3.1856,
      "step": 22635
    },
    {
      "epoch": 0.23030598958333334,
      "grad_norm": 8.41667366027832,
      "learning_rate": 4.377757122045271e-06,
      "loss": 3.4197,
      "step": 22640
    },
    {
      "epoch": 0.23035685221354166,
      "grad_norm": 11.499900817871094,
      "learning_rate": 4.377493102541094e-06,
      "loss": 3.1093,
      "step": 22645
    },
    {
      "epoch": 0.23040771484375,
      "grad_norm": 15.506875991821289,
      "learning_rate": 4.3772290350017014e-06,
      "loss": 3.4855,
      "step": 22650
    },
    {
      "epoch": 0.23045857747395834,
      "grad_norm": 7.720246315002441,
      "learning_rate": 4.376964919433848e-06,
      "loss": 3.4042,
      "step": 22655
    },
    {
      "epoch": 0.23050944010416666,
      "grad_norm": 7.877285480499268,
      "learning_rate": 4.376700755844292e-06,
      "loss": 3.2406,
      "step": 22660
    },
    {
      "epoch": 0.230560302734375,
      "grad_norm": 14.253623008728027,
      "learning_rate": 4.376436544239793e-06,
      "loss": 3.3702,
      "step": 22665
    },
    {
      "epoch": 0.23061116536458334,
      "grad_norm": 13.754986763000488,
      "learning_rate": 4.37617228462711e-06,
      "loss": 3.6621,
      "step": 22670
    },
    {
      "epoch": 0.23066202799479166,
      "grad_norm": 13.590864181518555,
      "learning_rate": 4.375907977013002e-06,
      "loss": 3.2626,
      "step": 22675
    },
    {
      "epoch": 0.230712890625,
      "grad_norm": 9.83669376373291,
      "learning_rate": 4.3756436214042344e-06,
      "loss": 3.1272,
      "step": 22680
    },
    {
      "epoch": 0.23076375325520834,
      "grad_norm": 11.492262840270996,
      "learning_rate": 4.375379217807568e-06,
      "loss": 3.3619,
      "step": 22685
    },
    {
      "epoch": 0.23081461588541666,
      "grad_norm": 13.837381362915039,
      "learning_rate": 4.37511476622977e-06,
      "loss": 3.3483,
      "step": 22690
    },
    {
      "epoch": 0.230865478515625,
      "grad_norm": 12.848306655883789,
      "learning_rate": 4.374850266677605e-06,
      "loss": 3.6923,
      "step": 22695
    },
    {
      "epoch": 0.23091634114583334,
      "grad_norm": 10.598101615905762,
      "learning_rate": 4.37458571915784e-06,
      "loss": 3.4643,
      "step": 22700
    },
    {
      "epoch": 0.23096720377604166,
      "grad_norm": 13.353347778320312,
      "learning_rate": 4.374321123677245e-06,
      "loss": 3.3367,
      "step": 22705
    },
    {
      "epoch": 0.23101806640625,
      "grad_norm": 11.848420143127441,
      "learning_rate": 4.374056480242587e-06,
      "loss": 3.8908,
      "step": 22710
    },
    {
      "epoch": 0.23106892903645834,
      "grad_norm": 9.748141288757324,
      "learning_rate": 4.373791788860638e-06,
      "loss": 3.0885,
      "step": 22715
    },
    {
      "epoch": 0.23111979166666666,
      "grad_norm": 11.857436180114746,
      "learning_rate": 4.373527049538171e-06,
      "loss": 3.2744,
      "step": 22720
    },
    {
      "epoch": 0.231170654296875,
      "grad_norm": 12.81821060180664,
      "learning_rate": 4.373262262281958e-06,
      "loss": 3.2552,
      "step": 22725
    },
    {
      "epoch": 0.23122151692708334,
      "grad_norm": 12.025280952453613,
      "learning_rate": 4.372997427098774e-06,
      "loss": 3.5045,
      "step": 22730
    },
    {
      "epoch": 0.23127237955729166,
      "grad_norm": 12.568089485168457,
      "learning_rate": 4.372732543995395e-06,
      "loss": 3.4126,
      "step": 22735
    },
    {
      "epoch": 0.2313232421875,
      "grad_norm": 9.853676795959473,
      "learning_rate": 4.372467612978597e-06,
      "loss": 3.697,
      "step": 22740
    },
    {
      "epoch": 0.23137410481770834,
      "grad_norm": 12.528533935546875,
      "learning_rate": 4.37220263405516e-06,
      "loss": 3.0022,
      "step": 22745
    },
    {
      "epoch": 0.23142496744791666,
      "grad_norm": 9.40113639831543,
      "learning_rate": 4.371937607231862e-06,
      "loss": 3.3832,
      "step": 22750
    },
    {
      "epoch": 0.231475830078125,
      "grad_norm": 16.36553382873535,
      "learning_rate": 4.371672532515484e-06,
      "loss": 3.1619,
      "step": 22755
    },
    {
      "epoch": 0.23152669270833334,
      "grad_norm": 10.763917922973633,
      "learning_rate": 4.371407409912808e-06,
      "loss": 3.4408,
      "step": 22760
    },
    {
      "epoch": 0.23157755533854166,
      "grad_norm": 9.289487838745117,
      "learning_rate": 4.371142239430616e-06,
      "loss": 3.2544,
      "step": 22765
    },
    {
      "epoch": 0.23162841796875,
      "grad_norm": 10.083857536315918,
      "learning_rate": 4.370877021075695e-06,
      "loss": 3.215,
      "step": 22770
    },
    {
      "epoch": 0.23167928059895834,
      "grad_norm": 8.067363739013672,
      "learning_rate": 4.3706117548548276e-06,
      "loss": 3.3603,
      "step": 22775
    },
    {
      "epoch": 0.23173014322916666,
      "grad_norm": 18.238494873046875,
      "learning_rate": 4.3703464407748024e-06,
      "loss": 3.5479,
      "step": 22780
    },
    {
      "epoch": 0.231781005859375,
      "grad_norm": 9.413612365722656,
      "learning_rate": 4.3700810788424065e-06,
      "loss": 3.5223,
      "step": 22785
    },
    {
      "epoch": 0.23183186848958334,
      "grad_norm": 12.918985366821289,
      "learning_rate": 4.36981566906443e-06,
      "loss": 3.4497,
      "step": 22790
    },
    {
      "epoch": 0.23188273111979166,
      "grad_norm": 14.47326374053955,
      "learning_rate": 4.369550211447663e-06,
      "loss": 3.5497,
      "step": 22795
    },
    {
      "epoch": 0.23193359375,
      "grad_norm": 14.251363754272461,
      "learning_rate": 4.369284705998896e-06,
      "loss": 3.8692,
      "step": 22800
    },
    {
      "epoch": 0.23198445638020834,
      "grad_norm": 9.761405944824219,
      "learning_rate": 4.369019152724923e-06,
      "loss": 3.4169,
      "step": 22805
    },
    {
      "epoch": 0.23203531901041666,
      "grad_norm": 16.306804656982422,
      "learning_rate": 4.368753551632539e-06,
      "loss": 3.333,
      "step": 22810
    },
    {
      "epoch": 0.232086181640625,
      "grad_norm": 10.218423843383789,
      "learning_rate": 4.368487902728537e-06,
      "loss": 3.2954,
      "step": 22815
    },
    {
      "epoch": 0.23213704427083334,
      "grad_norm": 14.56457233428955,
      "learning_rate": 4.368222206019716e-06,
      "loss": 3.4159,
      "step": 22820
    },
    {
      "epoch": 0.23218790690104166,
      "grad_norm": 13.245458602905273,
      "learning_rate": 4.367956461512872e-06,
      "loss": 3.6755,
      "step": 22825
    },
    {
      "epoch": 0.23223876953125,
      "grad_norm": 11.180668830871582,
      "learning_rate": 4.367690669214805e-06,
      "loss": 3.4576,
      "step": 22830
    },
    {
      "epoch": 0.23228963216145834,
      "grad_norm": 12.140656471252441,
      "learning_rate": 4.367424829132314e-06,
      "loss": 3.4696,
      "step": 22835
    },
    {
      "epoch": 0.23234049479166666,
      "grad_norm": 15.034218788146973,
      "learning_rate": 4.367158941272203e-06,
      "loss": 3.5265,
      "step": 22840
    },
    {
      "epoch": 0.232391357421875,
      "grad_norm": 10.023531913757324,
      "learning_rate": 4.366893005641272e-06,
      "loss": 3.3705,
      "step": 22845
    },
    {
      "epoch": 0.23244222005208334,
      "grad_norm": 9.87674331665039,
      "learning_rate": 4.3666270222463255e-06,
      "loss": 3.3428,
      "step": 22850
    },
    {
      "epoch": 0.23249308268229166,
      "grad_norm": 8.11019229888916,
      "learning_rate": 4.3663609910941705e-06,
      "loss": 3.0177,
      "step": 22855
    },
    {
      "epoch": 0.2325439453125,
      "grad_norm": 9.974431037902832,
      "learning_rate": 4.3660949121916105e-06,
      "loss": 3.2103,
      "step": 22860
    },
    {
      "epoch": 0.23259480794270834,
      "grad_norm": 8.820601463317871,
      "learning_rate": 4.365828785545456e-06,
      "loss": 3.5385,
      "step": 22865
    },
    {
      "epoch": 0.23264567057291666,
      "grad_norm": 12.899969100952148,
      "learning_rate": 4.365562611162513e-06,
      "loss": 3.2611,
      "step": 22870
    },
    {
      "epoch": 0.232696533203125,
      "grad_norm": 13.530678749084473,
      "learning_rate": 4.365296389049593e-06,
      "loss": 3.4912,
      "step": 22875
    },
    {
      "epoch": 0.23274739583333334,
      "grad_norm": 17.59902000427246,
      "learning_rate": 4.365030119213508e-06,
      "loss": 3.5208,
      "step": 22880
    },
    {
      "epoch": 0.23279825846354166,
      "grad_norm": 10.701493263244629,
      "learning_rate": 4.364763801661068e-06,
      "loss": 3.3438,
      "step": 22885
    },
    {
      "epoch": 0.23284912109375,
      "grad_norm": 13.408337593078613,
      "learning_rate": 4.364497436399089e-06,
      "loss": 2.9106,
      "step": 22890
    },
    {
      "epoch": 0.23289998372395834,
      "grad_norm": 11.21833610534668,
      "learning_rate": 4.364231023434385e-06,
      "loss": 3.7474,
      "step": 22895
    },
    {
      "epoch": 0.23295084635416666,
      "grad_norm": 11.439018249511719,
      "learning_rate": 4.363964562773772e-06,
      "loss": 3.2894,
      "step": 22900
    },
    {
      "epoch": 0.233001708984375,
      "grad_norm": 11.574640274047852,
      "learning_rate": 4.363698054424067e-06,
      "loss": 3.5409,
      "step": 22905
    },
    {
      "epoch": 0.23305257161458334,
      "grad_norm": 11.974193572998047,
      "learning_rate": 4.36343149839209e-06,
      "loss": 3.4101,
      "step": 22910
    },
    {
      "epoch": 0.23310343424479166,
      "grad_norm": 9.374265670776367,
      "learning_rate": 4.363164894684659e-06,
      "loss": 3.3679,
      "step": 22915
    },
    {
      "epoch": 0.233154296875,
      "grad_norm": 14.392888069152832,
      "learning_rate": 4.362898243308596e-06,
      "loss": 3.5128,
      "step": 22920
    },
    {
      "epoch": 0.23320515950520834,
      "grad_norm": 15.125787734985352,
      "learning_rate": 4.362631544270723e-06,
      "loss": 3.3931,
      "step": 22925
    },
    {
      "epoch": 0.23325602213541666,
      "grad_norm": 12.04422378540039,
      "learning_rate": 4.362364797577864e-06,
      "loss": 3.27,
      "step": 22930
    },
    {
      "epoch": 0.233306884765625,
      "grad_norm": 9.42465877532959,
      "learning_rate": 4.362098003236842e-06,
      "loss": 3.1457,
      "step": 22935
    },
    {
      "epoch": 0.23335774739583334,
      "grad_norm": 10.79931354522705,
      "learning_rate": 4.361831161254485e-06,
      "loss": 3.3657,
      "step": 22940
    },
    {
      "epoch": 0.23340861002604166,
      "grad_norm": 7.212630748748779,
      "learning_rate": 4.361564271637618e-06,
      "loss": 3.6657,
      "step": 22945
    },
    {
      "epoch": 0.23345947265625,
      "grad_norm": 10.60206127166748,
      "learning_rate": 4.361297334393071e-06,
      "loss": 3.5798,
      "step": 22950
    },
    {
      "epoch": 0.23351033528645834,
      "grad_norm": 8.320850372314453,
      "learning_rate": 4.361030349527673e-06,
      "loss": 3.4497,
      "step": 22955
    },
    {
      "epoch": 0.23356119791666666,
      "grad_norm": 11.13176441192627,
      "learning_rate": 4.360763317048254e-06,
      "loss": 3.4702,
      "step": 22960
    },
    {
      "epoch": 0.233612060546875,
      "grad_norm": 10.365711212158203,
      "learning_rate": 4.360496236961647e-06,
      "loss": 3.2991,
      "step": 22965
    },
    {
      "epoch": 0.23366292317708334,
      "grad_norm": 14.777729988098145,
      "learning_rate": 4.360229109274685e-06,
      "loss": 3.3027,
      "step": 22970
    },
    {
      "epoch": 0.23371378580729166,
      "grad_norm": 7.491941928863525,
      "learning_rate": 4.359961933994202e-06,
      "loss": 3.2291,
      "step": 22975
    },
    {
      "epoch": 0.2337646484375,
      "grad_norm": 9.205909729003906,
      "learning_rate": 4.359694711127034e-06,
      "loss": 3.5741,
      "step": 22980
    },
    {
      "epoch": 0.23381551106770834,
      "grad_norm": 15.118029594421387,
      "learning_rate": 4.359427440680016e-06,
      "loss": 3.3006,
      "step": 22985
    },
    {
      "epoch": 0.23386637369791666,
      "grad_norm": 8.938360214233398,
      "learning_rate": 4.35916012265999e-06,
      "loss": 3.3448,
      "step": 22990
    },
    {
      "epoch": 0.233917236328125,
      "grad_norm": 10.3394193649292,
      "learning_rate": 4.358892757073792e-06,
      "loss": 3.3876,
      "step": 22995
    },
    {
      "epoch": 0.23396809895833334,
      "grad_norm": 15.624671936035156,
      "learning_rate": 4.358625343928264e-06,
      "loss": 3.6473,
      "step": 23000
    },
    {
      "epoch": 0.23401896158854166,
      "grad_norm": 7.711795806884766,
      "learning_rate": 4.358357883230245e-06,
      "loss": 3.729,
      "step": 23005
    },
    {
      "epoch": 0.23406982421875,
      "grad_norm": 8.540888786315918,
      "learning_rate": 4.358090374986582e-06,
      "loss": 3.3981,
      "step": 23010
    },
    {
      "epoch": 0.23412068684895834,
      "grad_norm": 11.216827392578125,
      "learning_rate": 4.357822819204117e-06,
      "loss": 3.6308,
      "step": 23015
    },
    {
      "epoch": 0.23417154947916666,
      "grad_norm": 11.93607234954834,
      "learning_rate": 4.357555215889695e-06,
      "loss": 3.3311,
      "step": 23020
    },
    {
      "epoch": 0.234222412109375,
      "grad_norm": 14.416646957397461,
      "learning_rate": 4.3572875650501624e-06,
      "loss": 3.3338,
      "step": 23025
    },
    {
      "epoch": 0.23427327473958334,
      "grad_norm": 9.926932334899902,
      "learning_rate": 4.357019866692369e-06,
      "loss": 3.2852,
      "step": 23030
    },
    {
      "epoch": 0.23432413736979166,
      "grad_norm": 13.363643646240234,
      "learning_rate": 4.356752120823162e-06,
      "loss": 3.4826,
      "step": 23035
    },
    {
      "epoch": 0.234375,
      "grad_norm": 10.39242935180664,
      "learning_rate": 4.356484327449392e-06,
      "loss": 3.2185,
      "step": 23040
    },
    {
      "epoch": 0.23442586263020834,
      "grad_norm": 11.942859649658203,
      "learning_rate": 4.356216486577911e-06,
      "loss": 3.3092,
      "step": 23045
    },
    {
      "epoch": 0.23447672526041666,
      "grad_norm": 15.64292049407959,
      "learning_rate": 4.35594859821557e-06,
      "loss": 3.3923,
      "step": 23050
    },
    {
      "epoch": 0.234527587890625,
      "grad_norm": 14.945747375488281,
      "learning_rate": 4.355680662369225e-06,
      "loss": 3.5142,
      "step": 23055
    },
    {
      "epoch": 0.23457845052083334,
      "grad_norm": 12.805582046508789,
      "learning_rate": 4.355412679045729e-06,
      "loss": 3.4128,
      "step": 23060
    },
    {
      "epoch": 0.23462931315104166,
      "grad_norm": 9.039190292358398,
      "learning_rate": 4.355144648251941e-06,
      "loss": 3.5806,
      "step": 23065
    },
    {
      "epoch": 0.23468017578125,
      "grad_norm": 15.245207786560059,
      "learning_rate": 4.354876569994716e-06,
      "loss": 3.7038,
      "step": 23070
    },
    {
      "epoch": 0.23473103841145834,
      "grad_norm": 12.465885162353516,
      "learning_rate": 4.354608444280915e-06,
      "loss": 3.2708,
      "step": 23075
    },
    {
      "epoch": 0.23478190104166666,
      "grad_norm": 8.4971342086792,
      "learning_rate": 4.3543402711173946e-06,
      "loss": 3.0945,
      "step": 23080
    },
    {
      "epoch": 0.234832763671875,
      "grad_norm": 9.483534812927246,
      "learning_rate": 4.354072050511019e-06,
      "loss": 3.514,
      "step": 23085
    },
    {
      "epoch": 0.23488362630208334,
      "grad_norm": 10.665985107421875,
      "learning_rate": 4.353803782468648e-06,
      "loss": 3.7719,
      "step": 23090
    },
    {
      "epoch": 0.23493448893229166,
      "grad_norm": 12.816307067871094,
      "learning_rate": 4.353535466997149e-06,
      "loss": 3.7636,
      "step": 23095
    },
    {
      "epoch": 0.2349853515625,
      "grad_norm": 12.756634712219238,
      "learning_rate": 4.3532671041033825e-06,
      "loss": 3.3117,
      "step": 23100
    },
    {
      "epoch": 0.23503621419270834,
      "grad_norm": 8.485614776611328,
      "learning_rate": 4.352998693794217e-06,
      "loss": 3.102,
      "step": 23105
    },
    {
      "epoch": 0.23508707682291666,
      "grad_norm": 8.25635814666748,
      "learning_rate": 4.352730236076519e-06,
      "loss": 3.0136,
      "step": 23110
    },
    {
      "epoch": 0.235137939453125,
      "grad_norm": 9.216659545898438,
      "learning_rate": 4.352461730957157e-06,
      "loss": 3.7945,
      "step": 23115
    },
    {
      "epoch": 0.23518880208333334,
      "grad_norm": 11.367064476013184,
      "learning_rate": 4.352193178443001e-06,
      "loss": 3.3426,
      "step": 23120
    },
    {
      "epoch": 0.23523966471354166,
      "grad_norm": 12.334614753723145,
      "learning_rate": 4.351924578540922e-06,
      "loss": 3.4398,
      "step": 23125
    },
    {
      "epoch": 0.23529052734375,
      "grad_norm": 11.858649253845215,
      "learning_rate": 4.35165593125779e-06,
      "loss": 3.3179,
      "step": 23130
    },
    {
      "epoch": 0.23534138997395834,
      "grad_norm": 12.8947172164917,
      "learning_rate": 4.351387236600481e-06,
      "loss": 3.2333,
      "step": 23135
    },
    {
      "epoch": 0.23539225260416666,
      "grad_norm": 7.790979862213135,
      "learning_rate": 4.351118494575869e-06,
      "loss": 3.4224,
      "step": 23140
    },
    {
      "epoch": 0.235443115234375,
      "grad_norm": 9.988829612731934,
      "learning_rate": 4.350849705190828e-06,
      "loss": 3.4509,
      "step": 23145
    },
    {
      "epoch": 0.23549397786458334,
      "grad_norm": 8.369173049926758,
      "learning_rate": 4.350580868452237e-06,
      "loss": 3.169,
      "step": 23150
    },
    {
      "epoch": 0.23554484049479166,
      "grad_norm": 12.344426155090332,
      "learning_rate": 4.350311984366972e-06,
      "loss": 3.4007,
      "step": 23155
    },
    {
      "epoch": 0.235595703125,
      "grad_norm": 11.743609428405762,
      "learning_rate": 4.350043052941914e-06,
      "loss": 3.3544,
      "step": 23160
    },
    {
      "epoch": 0.23564656575520834,
      "grad_norm": 11.494827270507812,
      "learning_rate": 4.3497740741839425e-06,
      "loss": 3.342,
      "step": 23165
    },
    {
      "epoch": 0.23569742838541666,
      "grad_norm": 9.599040031433105,
      "learning_rate": 4.349505048099941e-06,
      "loss": 3.4375,
      "step": 23170
    },
    {
      "epoch": 0.235748291015625,
      "grad_norm": 12.405707359313965,
      "learning_rate": 4.34923597469679e-06,
      "loss": 3.9962,
      "step": 23175
    },
    {
      "epoch": 0.23579915364583334,
      "grad_norm": 12.194296836853027,
      "learning_rate": 4.348966853981376e-06,
      "loss": 3.4165,
      "step": 23180
    },
    {
      "epoch": 0.23585001627604166,
      "grad_norm": 9.456016540527344,
      "learning_rate": 4.348697685960582e-06,
      "loss": 3.6132,
      "step": 23185
    },
    {
      "epoch": 0.23590087890625,
      "grad_norm": 12.257650375366211,
      "learning_rate": 4.348428470641297e-06,
      "loss": 3.1384,
      "step": 23190
    },
    {
      "epoch": 0.23595174153645834,
      "grad_norm": 11.710000991821289,
      "learning_rate": 4.348159208030407e-06,
      "loss": 3.6273,
      "step": 23195
    },
    {
      "epoch": 0.23600260416666666,
      "grad_norm": 15.638426780700684,
      "learning_rate": 4.347889898134802e-06,
      "loss": 3.3879,
      "step": 23200
    },
    {
      "epoch": 0.236053466796875,
      "grad_norm": 11.080460548400879,
      "learning_rate": 4.3476205409613715e-06,
      "loss": 3.4117,
      "step": 23205
    },
    {
      "epoch": 0.23610432942708334,
      "grad_norm": 7.962300777435303,
      "learning_rate": 4.347351136517008e-06,
      "loss": 3.6079,
      "step": 23210
    },
    {
      "epoch": 0.23615519205729166,
      "grad_norm": 14.046839714050293,
      "learning_rate": 4.3470816848086025e-06,
      "loss": 3.049,
      "step": 23215
    },
    {
      "epoch": 0.2362060546875,
      "grad_norm": 9.567350387573242,
      "learning_rate": 4.346812185843051e-06,
      "loss": 3.103,
      "step": 23220
    },
    {
      "epoch": 0.23625691731770834,
      "grad_norm": 14.83440113067627,
      "learning_rate": 4.346542639627247e-06,
      "loss": 3.3158,
      "step": 23225
    },
    {
      "epoch": 0.23630777994791666,
      "grad_norm": 14.364097595214844,
      "learning_rate": 4.346273046168088e-06,
      "loss": 3.2224,
      "step": 23230
    },
    {
      "epoch": 0.236358642578125,
      "grad_norm": 9.735870361328125,
      "learning_rate": 4.34600340547247e-06,
      "loss": 3.7907,
      "step": 23235
    },
    {
      "epoch": 0.23640950520833334,
      "grad_norm": 11.21799373626709,
      "learning_rate": 4.3457337175472925e-06,
      "loss": 3.6273,
      "step": 23240
    },
    {
      "epoch": 0.23646036783854166,
      "grad_norm": 13.670818328857422,
      "learning_rate": 4.345463982399456e-06,
      "loss": 3.2087,
      "step": 23245
    },
    {
      "epoch": 0.23651123046875,
      "grad_norm": 10.433359146118164,
      "learning_rate": 4.3451942000358605e-06,
      "loss": 3.4824,
      "step": 23250
    },
    {
      "epoch": 0.23656209309895834,
      "grad_norm": 9.702804565429688,
      "learning_rate": 4.344924370463408e-06,
      "loss": 3.6147,
      "step": 23255
    },
    {
      "epoch": 0.23661295572916666,
      "grad_norm": 8.879476547241211,
      "learning_rate": 4.344654493689004e-06,
      "loss": 3.0391,
      "step": 23260
    },
    {
      "epoch": 0.236663818359375,
      "grad_norm": 7.819838523864746,
      "learning_rate": 4.344384569719551e-06,
      "loss": 3.643,
      "step": 23265
    },
    {
      "epoch": 0.23671468098958334,
      "grad_norm": 10.553860664367676,
      "learning_rate": 4.344114598561957e-06,
      "loss": 3.5373,
      "step": 23270
    },
    {
      "epoch": 0.23676554361979166,
      "grad_norm": 11.761865615844727,
      "learning_rate": 4.3438445802231264e-06,
      "loss": 3.2724,
      "step": 23275
    },
    {
      "epoch": 0.23681640625,
      "grad_norm": 12.256386756896973,
      "learning_rate": 4.34357451470997e-06,
      "loss": 3.3775,
      "step": 23280
    },
    {
      "epoch": 0.23686726888020834,
      "grad_norm": 14.110241889953613,
      "learning_rate": 4.343304402029398e-06,
      "loss": 3.7732,
      "step": 23285
    },
    {
      "epoch": 0.23691813151041666,
      "grad_norm": 8.55147933959961,
      "learning_rate": 4.343034242188318e-06,
      "loss": 3.3465,
      "step": 23290
    },
    {
      "epoch": 0.236968994140625,
      "grad_norm": 12.542463302612305,
      "learning_rate": 4.342764035193645e-06,
      "loss": 3.4555,
      "step": 23295
    },
    {
      "epoch": 0.23701985677083334,
      "grad_norm": 13.53172492980957,
      "learning_rate": 4.3424937810522904e-06,
      "loss": 3.3524,
      "step": 23300
    },
    {
      "epoch": 0.23707071940104166,
      "grad_norm": 12.750872611999512,
      "learning_rate": 4.342223479771169e-06,
      "loss": 3.7538,
      "step": 23305
    },
    {
      "epoch": 0.23712158203125,
      "grad_norm": 9.65660285949707,
      "learning_rate": 4.341953131357196e-06,
      "loss": 3.566,
      "step": 23310
    },
    {
      "epoch": 0.23717244466145834,
      "grad_norm": 15.46808910369873,
      "learning_rate": 4.341682735817289e-06,
      "loss": 3.6563,
      "step": 23315
    },
    {
      "epoch": 0.23722330729166666,
      "grad_norm": 15.83115291595459,
      "learning_rate": 4.341412293158367e-06,
      "loss": 3.3011,
      "step": 23320
    },
    {
      "epoch": 0.237274169921875,
      "grad_norm": 9.020452499389648,
      "learning_rate": 4.341141803387347e-06,
      "loss": 3.2142,
      "step": 23325
    },
    {
      "epoch": 0.23732503255208334,
      "grad_norm": 7.2499470710754395,
      "learning_rate": 4.3408712665111505e-06,
      "loss": 3.2466,
      "step": 23330
    },
    {
      "epoch": 0.23737589518229166,
      "grad_norm": 14.520453453063965,
      "learning_rate": 4.340600682536699e-06,
      "loss": 3.1733,
      "step": 23335
    },
    {
      "epoch": 0.2374267578125,
      "grad_norm": 7.980301856994629,
      "learning_rate": 4.340330051470915e-06,
      "loss": 3.0205,
      "step": 23340
    },
    {
      "epoch": 0.23747762044270834,
      "grad_norm": 7.818747520446777,
      "learning_rate": 4.340059373320723e-06,
      "loss": 3.4661,
      "step": 23345
    },
    {
      "epoch": 0.23752848307291666,
      "grad_norm": 8.310884475708008,
      "learning_rate": 4.339788648093048e-06,
      "loss": 2.7708,
      "step": 23350
    },
    {
      "epoch": 0.237579345703125,
      "grad_norm": 11.503134727478027,
      "learning_rate": 4.339517875794817e-06,
      "loss": 3.2503,
      "step": 23355
    },
    {
      "epoch": 0.23763020833333334,
      "grad_norm": 14.735780715942383,
      "learning_rate": 4.339247056432957e-06,
      "loss": 3.3462,
      "step": 23360
    },
    {
      "epoch": 0.23768107096354166,
      "grad_norm": 14.477924346923828,
      "learning_rate": 4.338976190014397e-06,
      "loss": 3.2735,
      "step": 23365
    },
    {
      "epoch": 0.23773193359375,
      "grad_norm": 14.508803367614746,
      "learning_rate": 4.338705276546067e-06,
      "loss": 3.169,
      "step": 23370
    },
    {
      "epoch": 0.23778279622395834,
      "grad_norm": 10.182372093200684,
      "learning_rate": 4.338434316034897e-06,
      "loss": 3.6902,
      "step": 23375
    },
    {
      "epoch": 0.23783365885416666,
      "grad_norm": 15.233953475952148,
      "learning_rate": 4.338163308487823e-06,
      "loss": 3.7186,
      "step": 23380
    },
    {
      "epoch": 0.237884521484375,
      "grad_norm": 8.143902778625488,
      "learning_rate": 4.337892253911775e-06,
      "loss": 3.2273,
      "step": 23385
    },
    {
      "epoch": 0.23793538411458334,
      "grad_norm": 15.452500343322754,
      "learning_rate": 4.33762115231369e-06,
      "loss": 3.5926,
      "step": 23390
    },
    {
      "epoch": 0.23798624674479166,
      "grad_norm": 14.97335147857666,
      "learning_rate": 4.337350003700502e-06,
      "loss": 3.2986,
      "step": 23395
    },
    {
      "epoch": 0.238037109375,
      "grad_norm": 11.603572845458984,
      "learning_rate": 4.3370788080791516e-06,
      "loss": 3.6042,
      "step": 23400
    },
    {
      "epoch": 0.23808797200520834,
      "grad_norm": 11.123287200927734,
      "learning_rate": 4.336807565456574e-06,
      "loss": 3.5009,
      "step": 23405
    },
    {
      "epoch": 0.23813883463541666,
      "grad_norm": 16.845380783081055,
      "learning_rate": 4.3365362758397115e-06,
      "loss": 3.2503,
      "step": 23410
    },
    {
      "epoch": 0.238189697265625,
      "grad_norm": 10.304829597473145,
      "learning_rate": 4.3362649392355026e-06,
      "loss": 3.478,
      "step": 23415
    },
    {
      "epoch": 0.23824055989583334,
      "grad_norm": 14.178970336914062,
      "learning_rate": 4.335993555650891e-06,
      "loss": 3.2514,
      "step": 23420
    },
    {
      "epoch": 0.23829142252604166,
      "grad_norm": 11.138680458068848,
      "learning_rate": 4.335722125092819e-06,
      "loss": 3.286,
      "step": 23425
    },
    {
      "epoch": 0.23834228515625,
      "grad_norm": 13.36595344543457,
      "learning_rate": 4.335450647568231e-06,
      "loss": 3.4917,
      "step": 23430
    },
    {
      "epoch": 0.23839314778645834,
      "grad_norm": 11.410798072814941,
      "learning_rate": 4.335179123084074e-06,
      "loss": 3.4145,
      "step": 23435
    },
    {
      "epoch": 0.23844401041666666,
      "grad_norm": 11.153116226196289,
      "learning_rate": 4.334907551647295e-06,
      "loss": 3.4745,
      "step": 23440
    },
    {
      "epoch": 0.238494873046875,
      "grad_norm": 10.488436698913574,
      "learning_rate": 4.33463593326484e-06,
      "loss": 3.4593,
      "step": 23445
    },
    {
      "epoch": 0.23854573567708334,
      "grad_norm": 15.080182075500488,
      "learning_rate": 4.3343642679436595e-06,
      "loss": 3.4577,
      "step": 23450
    },
    {
      "epoch": 0.23859659830729166,
      "grad_norm": 13.887222290039062,
      "learning_rate": 4.3340925556907045e-06,
      "loss": 3.0636,
      "step": 23455
    },
    {
      "epoch": 0.2386474609375,
      "grad_norm": 14.262229919433594,
      "learning_rate": 4.333820796512926e-06,
      "loss": 3.1407,
      "step": 23460
    },
    {
      "epoch": 0.23869832356770834,
      "grad_norm": 14.66751480102539,
      "learning_rate": 4.333548990417277e-06,
      "loss": 3.1232,
      "step": 23465
    },
    {
      "epoch": 0.23874918619791666,
      "grad_norm": 7.767457962036133,
      "learning_rate": 4.333277137410712e-06,
      "loss": 3.2363,
      "step": 23470
    },
    {
      "epoch": 0.238800048828125,
      "grad_norm": 13.721875190734863,
      "learning_rate": 4.333005237500186e-06,
      "loss": 3.167,
      "step": 23475
    },
    {
      "epoch": 0.23885091145833334,
      "grad_norm": 7.040370464324951,
      "learning_rate": 4.332733290692655e-06,
      "loss": 3.3249,
      "step": 23480
    },
    {
      "epoch": 0.23890177408854166,
      "grad_norm": 14.299421310424805,
      "learning_rate": 4.332461296995077e-06,
      "loss": 3.3313,
      "step": 23485
    },
    {
      "epoch": 0.23895263671875,
      "grad_norm": 12.305241584777832,
      "learning_rate": 4.3321892564144115e-06,
      "loss": 3.3294,
      "step": 23490
    },
    {
      "epoch": 0.23900349934895834,
      "grad_norm": 15.409032821655273,
      "learning_rate": 4.331917168957619e-06,
      "loss": 3.5023,
      "step": 23495
    },
    {
      "epoch": 0.23905436197916666,
      "grad_norm": 18.326412200927734,
      "learning_rate": 4.3316450346316586e-06,
      "loss": 3.599,
      "step": 23500
    },
    {
      "epoch": 0.239105224609375,
      "grad_norm": 13.42839527130127,
      "learning_rate": 4.331372853443495e-06,
      "loss": 3.3468,
      "step": 23505
    },
    {
      "epoch": 0.23915608723958334,
      "grad_norm": 13.067977905273438,
      "learning_rate": 4.33110062540009e-06,
      "loss": 3.1257,
      "step": 23510
    },
    {
      "epoch": 0.23920694986979166,
      "grad_norm": 10.786623001098633,
      "learning_rate": 4.33082835050841e-06,
      "loss": 3.0433,
      "step": 23515
    },
    {
      "epoch": 0.2392578125,
      "grad_norm": 16.42357635498047,
      "learning_rate": 4.330556028775421e-06,
      "loss": 3.241,
      "step": 23520
    },
    {
      "epoch": 0.23930867513020834,
      "grad_norm": 13.901268005371094,
      "learning_rate": 4.330283660208089e-06,
      "loss": 3.0375,
      "step": 23525
    },
    {
      "epoch": 0.23935953776041666,
      "grad_norm": 13.501899719238281,
      "learning_rate": 4.330011244813384e-06,
      "loss": 2.9452,
      "step": 23530
    },
    {
      "epoch": 0.239410400390625,
      "grad_norm": 12.320079803466797,
      "learning_rate": 4.329738782598276e-06,
      "loss": 3.4211,
      "step": 23535
    },
    {
      "epoch": 0.23946126302083334,
      "grad_norm": 12.208647727966309,
      "learning_rate": 4.329466273569734e-06,
      "loss": 3.389,
      "step": 23540
    },
    {
      "epoch": 0.23951212565104166,
      "grad_norm": 13.780562400817871,
      "learning_rate": 4.329193717734731e-06,
      "loss": 3.313,
      "step": 23545
    },
    {
      "epoch": 0.23956298828125,
      "grad_norm": 9.39169692993164,
      "learning_rate": 4.328921115100239e-06,
      "loss": 3.0223,
      "step": 23550
    },
    {
      "epoch": 0.23961385091145834,
      "grad_norm": 11.84803295135498,
      "learning_rate": 4.3286484656732365e-06,
      "loss": 3.1971,
      "step": 23555
    },
    {
      "epoch": 0.23966471354166666,
      "grad_norm": 8.828704833984375,
      "learning_rate": 4.328375769460694e-06,
      "loss": 3.4065,
      "step": 23560
    },
    {
      "epoch": 0.239715576171875,
      "grad_norm": 12.668590545654297,
      "learning_rate": 4.328103026469592e-06,
      "loss": 3.3995,
      "step": 23565
    },
    {
      "epoch": 0.23976643880208334,
      "grad_norm": 7.958857536315918,
      "learning_rate": 4.327830236706908e-06,
      "loss": 3.1953,
      "step": 23570
    },
    {
      "epoch": 0.23981730143229166,
      "grad_norm": 12.19107723236084,
      "learning_rate": 4.32755740017962e-06,
      "loss": 3.0576,
      "step": 23575
    },
    {
      "epoch": 0.2398681640625,
      "grad_norm": 10.372651100158691,
      "learning_rate": 4.327284516894708e-06,
      "loss": 3.195,
      "step": 23580
    },
    {
      "epoch": 0.23991902669270834,
      "grad_norm": 15.74683952331543,
      "learning_rate": 4.327011586859156e-06,
      "loss": 3.286,
      "step": 23585
    },
    {
      "epoch": 0.23996988932291666,
      "grad_norm": 7.834261417388916,
      "learning_rate": 4.326738610079946e-06,
      "loss": 3.4963,
      "step": 23590
    },
    {
      "epoch": 0.240020751953125,
      "grad_norm": 12.021536827087402,
      "learning_rate": 4.326465586564061e-06,
      "loss": 3.5939,
      "step": 23595
    },
    {
      "epoch": 0.24007161458333334,
      "grad_norm": 11.826726913452148,
      "learning_rate": 4.326192516318486e-06,
      "loss": 3.2587,
      "step": 23600
    },
    {
      "epoch": 0.24012247721354166,
      "grad_norm": 9.564510345458984,
      "learning_rate": 4.325919399350209e-06,
      "loss": 3.1422,
      "step": 23605
    },
    {
      "epoch": 0.24017333984375,
      "grad_norm": 15.419652938842773,
      "learning_rate": 4.3256462356662175e-06,
      "loss": 3.3998,
      "step": 23610
    },
    {
      "epoch": 0.24022420247395834,
      "grad_norm": 7.081390857696533,
      "learning_rate": 4.325373025273499e-06,
      "loss": 3.4022,
      "step": 23615
    },
    {
      "epoch": 0.24027506510416666,
      "grad_norm": 10.660130500793457,
      "learning_rate": 4.325099768179045e-06,
      "loss": 3.2061,
      "step": 23620
    },
    {
      "epoch": 0.240325927734375,
      "grad_norm": 12.876676559448242,
      "learning_rate": 4.324826464389846e-06,
      "loss": 3.2347,
      "step": 23625
    },
    {
      "epoch": 0.24037679036458334,
      "grad_norm": 10.950091361999512,
      "learning_rate": 4.324553113912894e-06,
      "loss": 3.2463,
      "step": 23630
    },
    {
      "epoch": 0.24042765299479166,
      "grad_norm": 13.289194107055664,
      "learning_rate": 4.324279716755183e-06,
      "loss": 3.704,
      "step": 23635
    },
    {
      "epoch": 0.240478515625,
      "grad_norm": 12.992833137512207,
      "learning_rate": 4.324006272923708e-06,
      "loss": 3.1545,
      "step": 23640
    },
    {
      "epoch": 0.24052937825520834,
      "grad_norm": 7.339069843292236,
      "learning_rate": 4.323732782425465e-06,
      "loss": 3.0488,
      "step": 23645
    },
    {
      "epoch": 0.24058024088541666,
      "grad_norm": 11.042210578918457,
      "learning_rate": 4.323459245267451e-06,
      "loss": 3.2029,
      "step": 23650
    },
    {
      "epoch": 0.240631103515625,
      "grad_norm": 13.536554336547852,
      "learning_rate": 4.323185661456665e-06,
      "loss": 3.2709,
      "step": 23655
    },
    {
      "epoch": 0.24068196614583334,
      "grad_norm": 9.641441345214844,
      "learning_rate": 4.322912031000105e-06,
      "loss": 3.6717,
      "step": 23660
    },
    {
      "epoch": 0.24073282877604166,
      "grad_norm": 13.114518165588379,
      "learning_rate": 4.322638353904773e-06,
      "loss": 3.4325,
      "step": 23665
    },
    {
      "epoch": 0.24078369140625,
      "grad_norm": 12.646187782287598,
      "learning_rate": 4.322364630177671e-06,
      "loss": 3.0339,
      "step": 23670
    },
    {
      "epoch": 0.24083455403645834,
      "grad_norm": 10.695713996887207,
      "learning_rate": 4.322090859825802e-06,
      "loss": 3.271,
      "step": 23675
    },
    {
      "epoch": 0.24088541666666666,
      "grad_norm": 8.131204605102539,
      "learning_rate": 4.3218170428561695e-06,
      "loss": 3.7232,
      "step": 23680
    },
    {
      "epoch": 0.240936279296875,
      "grad_norm": 9.798648834228516,
      "learning_rate": 4.3215431792757805e-06,
      "loss": 3.2187,
      "step": 23685
    },
    {
      "epoch": 0.24098714192708334,
      "grad_norm": 9.147968292236328,
      "learning_rate": 4.32126926909164e-06,
      "loss": 3.2248,
      "step": 23690
    },
    {
      "epoch": 0.24103800455729166,
      "grad_norm": 10.552434921264648,
      "learning_rate": 4.320995312310758e-06,
      "loss": 3.5436,
      "step": 23695
    },
    {
      "epoch": 0.2410888671875,
      "grad_norm": 11.929584503173828,
      "learning_rate": 4.320721308940142e-06,
      "loss": 3.6509,
      "step": 23700
    },
    {
      "epoch": 0.24113972981770834,
      "grad_norm": 12.67947006225586,
      "learning_rate": 4.320447258986804e-06,
      "loss": 3.2208,
      "step": 23705
    },
    {
      "epoch": 0.24119059244791666,
      "grad_norm": 10.882842063903809,
      "learning_rate": 4.320173162457754e-06,
      "loss": 3.2851,
      "step": 23710
    },
    {
      "epoch": 0.241241455078125,
      "grad_norm": 9.038286209106445,
      "learning_rate": 4.319899019360003e-06,
      "loss": 3.9344,
      "step": 23715
    },
    {
      "epoch": 0.24129231770833334,
      "grad_norm": 8.482138633728027,
      "learning_rate": 4.319624829700569e-06,
      "loss": 3.2898,
      "step": 23720
    },
    {
      "epoch": 0.24134318033854166,
      "grad_norm": 9.246583938598633,
      "learning_rate": 4.319350593486465e-06,
      "loss": 3.2805,
      "step": 23725
    },
    {
      "epoch": 0.24139404296875,
      "grad_norm": 14.592893600463867,
      "learning_rate": 4.319076310724707e-06,
      "loss": 3.9456,
      "step": 23730
    },
    {
      "epoch": 0.24144490559895834,
      "grad_norm": 13.710060119628906,
      "learning_rate": 4.318801981422312e-06,
      "loss": 3.4525,
      "step": 23735
    },
    {
      "epoch": 0.24149576822916666,
      "grad_norm": 11.734492301940918,
      "learning_rate": 4.3185276055863004e-06,
      "loss": 3.4413,
      "step": 23740
    },
    {
      "epoch": 0.241546630859375,
      "grad_norm": 10.637125015258789,
      "learning_rate": 4.31825318322369e-06,
      "loss": 3.4871,
      "step": 23745
    },
    {
      "epoch": 0.24159749348958334,
      "grad_norm": 11.061915397644043,
      "learning_rate": 4.317978714341504e-06,
      "loss": 3.3846,
      "step": 23750
    },
    {
      "epoch": 0.24164835611979166,
      "grad_norm": 9.801789283752441,
      "learning_rate": 4.317704198946763e-06,
      "loss": 3.3485,
      "step": 23755
    },
    {
      "epoch": 0.24169921875,
      "grad_norm": 13.971748352050781,
      "learning_rate": 4.3174296370464905e-06,
      "loss": 3.1945,
      "step": 23760
    },
    {
      "epoch": 0.24175008138020834,
      "grad_norm": 9.924613952636719,
      "learning_rate": 4.317155028647712e-06,
      "loss": 3.3264,
      "step": 23765
    },
    {
      "epoch": 0.24180094401041666,
      "grad_norm": 14.436348915100098,
      "learning_rate": 4.316880373757453e-06,
      "loss": 3.4252,
      "step": 23770
    },
    {
      "epoch": 0.241851806640625,
      "grad_norm": 16.445823669433594,
      "learning_rate": 4.31660567238274e-06,
      "loss": 3.5997,
      "step": 23775
    },
    {
      "epoch": 0.24190266927083334,
      "grad_norm": 14.193046569824219,
      "learning_rate": 4.316330924530601e-06,
      "loss": 3.13,
      "step": 23780
    },
    {
      "epoch": 0.24195353190104166,
      "grad_norm": 14.578299522399902,
      "learning_rate": 4.316056130208067e-06,
      "loss": 3.4323,
      "step": 23785
    },
    {
      "epoch": 0.24200439453125,
      "grad_norm": 13.892190933227539,
      "learning_rate": 4.315781289422168e-06,
      "loss": 3.5278,
      "step": 23790
    },
    {
      "epoch": 0.24205525716145834,
      "grad_norm": 15.20643424987793,
      "learning_rate": 4.315506402179933e-06,
      "loss": 3.3675,
      "step": 23795
    },
    {
      "epoch": 0.24210611979166666,
      "grad_norm": 11.651402473449707,
      "learning_rate": 4.3152314684883975e-06,
      "loss": 3.95,
      "step": 23800
    },
    {
      "epoch": 0.242156982421875,
      "grad_norm": 12.027578353881836,
      "learning_rate": 4.314956488354596e-06,
      "loss": 3.5839,
      "step": 23805
    },
    {
      "epoch": 0.24220784505208334,
      "grad_norm": 12.57933235168457,
      "learning_rate": 4.314681461785561e-06,
      "loss": 3.2394,
      "step": 23810
    },
    {
      "epoch": 0.24225870768229166,
      "grad_norm": 10.47960376739502,
      "learning_rate": 4.314406388788333e-06,
      "loss": 3.5905,
      "step": 23815
    },
    {
      "epoch": 0.2423095703125,
      "grad_norm": 11.346346855163574,
      "learning_rate": 4.3141312693699465e-06,
      "loss": 3.887,
      "step": 23820
    },
    {
      "epoch": 0.24236043294270834,
      "grad_norm": 12.971914291381836,
      "learning_rate": 4.313856103537442e-06,
      "loss": 3.6812,
      "step": 23825
    },
    {
      "epoch": 0.24241129557291666,
      "grad_norm": 12.668682098388672,
      "learning_rate": 4.313580891297859e-06,
      "loss": 3.2519,
      "step": 23830
    },
    {
      "epoch": 0.242462158203125,
      "grad_norm": 10.159984588623047,
      "learning_rate": 4.3133056326582386e-06,
      "loss": 3.1967,
      "step": 23835
    },
    {
      "epoch": 0.24251302083333334,
      "grad_norm": 15.020711898803711,
      "learning_rate": 4.313030327625623e-06,
      "loss": 3.1332,
      "step": 23840
    },
    {
      "epoch": 0.24256388346354166,
      "grad_norm": 11.418685913085938,
      "learning_rate": 4.312754976207058e-06,
      "loss": 3.2695,
      "step": 23845
    },
    {
      "epoch": 0.24261474609375,
      "grad_norm": 16.103302001953125,
      "learning_rate": 4.3124795784095845e-06,
      "loss": 3.1872,
      "step": 23850
    },
    {
      "epoch": 0.24266560872395834,
      "grad_norm": 8.224347114562988,
      "learning_rate": 4.312204134240252e-06,
      "loss": 3.4198,
      "step": 23855
    },
    {
      "epoch": 0.24271647135416666,
      "grad_norm": 16.922773361206055,
      "learning_rate": 4.311928643706105e-06,
      "loss": 3.6617,
      "step": 23860
    },
    {
      "epoch": 0.242767333984375,
      "grad_norm": 12.889124870300293,
      "learning_rate": 4.311653106814194e-06,
      "loss": 3.2375,
      "step": 23865
    },
    {
      "epoch": 0.24281819661458334,
      "grad_norm": 11.637298583984375,
      "learning_rate": 4.3113775235715675e-06,
      "loss": 3.4952,
      "step": 23870
    },
    {
      "epoch": 0.24286905924479166,
      "grad_norm": 15.415702819824219,
      "learning_rate": 4.3111018939852764e-06,
      "loss": 3.1809,
      "step": 23875
    },
    {
      "epoch": 0.242919921875,
      "grad_norm": 16.010303497314453,
      "learning_rate": 4.3108262180623724e-06,
      "loss": 3.384,
      "step": 23880
    },
    {
      "epoch": 0.24297078450520834,
      "grad_norm": 13.60275650024414,
      "learning_rate": 4.310550495809909e-06,
      "loss": 3.7173,
      "step": 23885
    },
    {
      "epoch": 0.24302164713541666,
      "grad_norm": 15.211081504821777,
      "learning_rate": 4.31027472723494e-06,
      "loss": 3.3849,
      "step": 23890
    },
    {
      "epoch": 0.243072509765625,
      "grad_norm": 15.576075553894043,
      "learning_rate": 4.309998912344522e-06,
      "loss": 3.3945,
      "step": 23895
    },
    {
      "epoch": 0.24312337239583334,
      "grad_norm": 6.691440582275391,
      "learning_rate": 4.30972305114571e-06,
      "loss": 3.64,
      "step": 23900
    },
    {
      "epoch": 0.24317423502604166,
      "grad_norm": 10.863690376281738,
      "learning_rate": 4.309447143645563e-06,
      "loss": 3.5055,
      "step": 23905
    },
    {
      "epoch": 0.24322509765625,
      "grad_norm": 14.350895881652832,
      "learning_rate": 4.30917118985114e-06,
      "loss": 3.259,
      "step": 23910
    },
    {
      "epoch": 0.24327596028645834,
      "grad_norm": 10.835892677307129,
      "learning_rate": 4.3088951897695e-06,
      "loss": 3.3739,
      "step": 23915
    },
    {
      "epoch": 0.24332682291666666,
      "grad_norm": 10.899023056030273,
      "learning_rate": 4.308619143407706e-06,
      "loss": 3.4605,
      "step": 23920
    },
    {
      "epoch": 0.243377685546875,
      "grad_norm": 17.81281280517578,
      "learning_rate": 4.30834305077282e-06,
      "loss": 3.6514,
      "step": 23925
    },
    {
      "epoch": 0.24342854817708334,
      "grad_norm": 10.215516090393066,
      "learning_rate": 4.3080669118719055e-06,
      "loss": 3.2631,
      "step": 23930
    },
    {
      "epoch": 0.24347941080729166,
      "grad_norm": 11.371179580688477,
      "learning_rate": 4.307790726712028e-06,
      "loss": 3.3525,
      "step": 23935
    },
    {
      "epoch": 0.2435302734375,
      "grad_norm": 7.968616962432861,
      "learning_rate": 4.307514495300253e-06,
      "loss": 3.1416,
      "step": 23940
    },
    {
      "epoch": 0.24358113606770834,
      "grad_norm": 11.859014511108398,
      "learning_rate": 4.307238217643648e-06,
      "loss": 3.1803,
      "step": 23945
    },
    {
      "epoch": 0.24363199869791666,
      "grad_norm": 7.970711708068848,
      "learning_rate": 4.306961893749281e-06,
      "loss": 3.1293,
      "step": 23950
    },
    {
      "epoch": 0.243682861328125,
      "grad_norm": 16.512006759643555,
      "learning_rate": 4.306685523624223e-06,
      "loss": 3.3174,
      "step": 23955
    },
    {
      "epoch": 0.24373372395833334,
      "grad_norm": 12.797045707702637,
      "learning_rate": 4.306409107275544e-06,
      "loss": 3.4595,
      "step": 23960
    },
    {
      "epoch": 0.24378458658854166,
      "grad_norm": 15.285560607910156,
      "learning_rate": 4.3061326447103155e-06,
      "loss": 3.0841,
      "step": 23965
    },
    {
      "epoch": 0.24383544921875,
      "grad_norm": 12.97689151763916,
      "learning_rate": 4.305856135935612e-06,
      "loss": 3.112,
      "step": 23970
    },
    {
      "epoch": 0.24388631184895834,
      "grad_norm": 13.685649871826172,
      "learning_rate": 4.3055795809585075e-06,
      "loss": 3.5289,
      "step": 23975
    },
    {
      "epoch": 0.24393717447916666,
      "grad_norm": 12.28869915008545,
      "learning_rate": 4.305302979786077e-06,
      "loss": 3.1983,
      "step": 23980
    },
    {
      "epoch": 0.243988037109375,
      "grad_norm": 7.516073226928711,
      "learning_rate": 4.305026332425398e-06,
      "loss": 3.5407,
      "step": 23985
    },
    {
      "epoch": 0.24403889973958334,
      "grad_norm": 10.212562561035156,
      "learning_rate": 4.304749638883548e-06,
      "loss": 3.4814,
      "step": 23990
    },
    {
      "epoch": 0.24408976236979166,
      "grad_norm": 15.381714820861816,
      "learning_rate": 4.304472899167607e-06,
      "loss": 3.2064,
      "step": 23995
    },
    {
      "epoch": 0.244140625,
      "grad_norm": 12.505240440368652,
      "learning_rate": 4.304196113284653e-06,
      "loss": 3.1553,
      "step": 24000
    },
    {
      "epoch": 0.24419148763020834,
      "grad_norm": 7.604684352874756,
      "learning_rate": 4.30391928124177e-06,
      "loss": 3.1687,
      "step": 24005
    },
    {
      "epoch": 0.24424235026041666,
      "grad_norm": 11.336790084838867,
      "learning_rate": 4.303642403046039e-06,
      "loss": 3.4772,
      "step": 24010
    },
    {
      "epoch": 0.244293212890625,
      "grad_norm": 17.307836532592773,
      "learning_rate": 4.3033654787045456e-06,
      "loss": 3.0816,
      "step": 24015
    },
    {
      "epoch": 0.24434407552083334,
      "grad_norm": 16.10969352722168,
      "learning_rate": 4.303088508224374e-06,
      "loss": 4.3131,
      "step": 24020
    },
    {
      "epoch": 0.24439493815104166,
      "grad_norm": 9.405745506286621,
      "learning_rate": 4.302811491612609e-06,
      "loss": 3.3127,
      "step": 24025
    },
    {
      "epoch": 0.24444580078125,
      "grad_norm": 13.108831405639648,
      "learning_rate": 4.30253442887634e-06,
      "loss": 3.1843,
      "step": 24030
    },
    {
      "epoch": 0.24449666341145834,
      "grad_norm": 9.459420204162598,
      "learning_rate": 4.302257320022655e-06,
      "loss": 3.6039,
      "step": 24035
    },
    {
      "epoch": 0.24454752604166666,
      "grad_norm": 17.34709358215332,
      "learning_rate": 4.301980165058643e-06,
      "loss": 3.4738,
      "step": 24040
    },
    {
      "epoch": 0.244598388671875,
      "grad_norm": 9.753969192504883,
      "learning_rate": 4.301702963991397e-06,
      "loss": 3.2752,
      "step": 24045
    },
    {
      "epoch": 0.24464925130208334,
      "grad_norm": 14.754100799560547,
      "learning_rate": 4.3014257168280074e-06,
      "loss": 2.8529,
      "step": 24050
    },
    {
      "epoch": 0.24470011393229166,
      "grad_norm": 9.242658615112305,
      "learning_rate": 4.3011484235755676e-06,
      "loss": 3.7791,
      "step": 24055
    },
    {
      "epoch": 0.2447509765625,
      "grad_norm": 13.81303882598877,
      "learning_rate": 4.300871084241172e-06,
      "loss": 3.5131,
      "step": 24060
    },
    {
      "epoch": 0.24480183919270834,
      "grad_norm": 14.629620552062988,
      "learning_rate": 4.3005936988319166e-06,
      "loss": 3.393,
      "step": 24065
    },
    {
      "epoch": 0.24485270182291666,
      "grad_norm": 13.942264556884766,
      "learning_rate": 4.3003162673548975e-06,
      "loss": 3.5995,
      "step": 24070
    },
    {
      "epoch": 0.244903564453125,
      "grad_norm": 12.826180458068848,
      "learning_rate": 4.300038789817214e-06,
      "loss": 3.2657,
      "step": 24075
    },
    {
      "epoch": 0.24495442708333334,
      "grad_norm": 14.900425910949707,
      "learning_rate": 4.299761266225965e-06,
      "loss": 3.4057,
      "step": 24080
    },
    {
      "epoch": 0.24500528971354166,
      "grad_norm": 16.51618766784668,
      "learning_rate": 4.299483696588251e-06,
      "loss": 3.4861,
      "step": 24085
    },
    {
      "epoch": 0.24505615234375,
      "grad_norm": 13.200966835021973,
      "learning_rate": 4.299206080911172e-06,
      "loss": 3.4621,
      "step": 24090
    },
    {
      "epoch": 0.24510701497395834,
      "grad_norm": 13.313547134399414,
      "learning_rate": 4.298928419201833e-06,
      "loss": 3.1582,
      "step": 24095
    },
    {
      "epoch": 0.24515787760416666,
      "grad_norm": 12.535401344299316,
      "learning_rate": 4.298650711467336e-06,
      "loss": 3.1646,
      "step": 24100
    },
    {
      "epoch": 0.245208740234375,
      "grad_norm": 15.257453918457031,
      "learning_rate": 4.298372957714787e-06,
      "loss": 3.2669,
      "step": 24105
    },
    {
      "epoch": 0.24525960286458334,
      "grad_norm": 16.784587860107422,
      "learning_rate": 4.298095157951292e-06,
      "loss": 3.5681,
      "step": 24110
    },
    {
      "epoch": 0.24531046549479166,
      "grad_norm": 10.863885879516602,
      "learning_rate": 4.297817312183959e-06,
      "loss": 3.6729,
      "step": 24115
    },
    {
      "epoch": 0.245361328125,
      "grad_norm": 12.638436317443848,
      "learning_rate": 4.2975394204198965e-06,
      "loss": 3.1696,
      "step": 24120
    },
    {
      "epoch": 0.24541219075520834,
      "grad_norm": 7.186621189117432,
      "learning_rate": 4.297261482666214e-06,
      "loss": 3.2565,
      "step": 24125
    },
    {
      "epoch": 0.24546305338541666,
      "grad_norm": 8.10990047454834,
      "learning_rate": 4.296983498930022e-06,
      "loss": 3.5558,
      "step": 24130
    },
    {
      "epoch": 0.245513916015625,
      "grad_norm": 9.806075096130371,
      "learning_rate": 4.296705469218432e-06,
      "loss": 3.3298,
      "step": 24135
    },
    {
      "epoch": 0.24556477864583334,
      "grad_norm": 15.463103294372559,
      "learning_rate": 4.29642739353856e-06,
      "loss": 3.4272,
      "step": 24140
    },
    {
      "epoch": 0.24561564127604166,
      "grad_norm": 15.613325119018555,
      "learning_rate": 4.296149271897519e-06,
      "loss": 3.311,
      "step": 24145
    },
    {
      "epoch": 0.24566650390625,
      "grad_norm": 10.683793067932129,
      "learning_rate": 4.295871104302424e-06,
      "loss": 3.4097,
      "step": 24150
    },
    {
      "epoch": 0.24571736653645834,
      "grad_norm": 8.91572093963623,
      "learning_rate": 4.295592890760392e-06,
      "loss": 3.3294,
      "step": 24155
    },
    {
      "epoch": 0.24576822916666666,
      "grad_norm": 13.804422378540039,
      "learning_rate": 4.295314631278542e-06,
      "loss": 3.3976,
      "step": 24160
    },
    {
      "epoch": 0.245819091796875,
      "grad_norm": 10.138822555541992,
      "learning_rate": 4.295036325863993e-06,
      "loss": 3.3067,
      "step": 24165
    },
    {
      "epoch": 0.24586995442708334,
      "grad_norm": 11.83923053741455,
      "learning_rate": 4.294757974523865e-06,
      "loss": 3.5395,
      "step": 24170
    },
    {
      "epoch": 0.24592081705729166,
      "grad_norm": 12.042837142944336,
      "learning_rate": 4.294479577265279e-06,
      "loss": 3.417,
      "step": 24175
    },
    {
      "epoch": 0.2459716796875,
      "grad_norm": 7.58932638168335,
      "learning_rate": 4.294201134095359e-06,
      "loss": 3.1013,
      "step": 24180
    },
    {
      "epoch": 0.24602254231770834,
      "grad_norm": 14.032426834106445,
      "learning_rate": 4.293922645021229e-06,
      "loss": 3.4314,
      "step": 24185
    },
    {
      "epoch": 0.24607340494791666,
      "grad_norm": 8.114699363708496,
      "learning_rate": 4.293644110050011e-06,
      "loss": 3.2654,
      "step": 24190
    },
    {
      "epoch": 0.246124267578125,
      "grad_norm": 12.10887622833252,
      "learning_rate": 4.293365529188835e-06,
      "loss": 3.2226,
      "step": 24195
    },
    {
      "epoch": 0.24617513020833334,
      "grad_norm": 11.954651832580566,
      "learning_rate": 4.2930869024448276e-06,
      "loss": 3.3103,
      "step": 24200
    },
    {
      "epoch": 0.24622599283854166,
      "grad_norm": 8.925844192504883,
      "learning_rate": 4.292808229825116e-06,
      "loss": 3.4017,
      "step": 24205
    },
    {
      "epoch": 0.24627685546875,
      "grad_norm": 15.956528663635254,
      "learning_rate": 4.292529511336831e-06,
      "loss": 3.2091,
      "step": 24210
    },
    {
      "epoch": 0.24632771809895834,
      "grad_norm": 14.59361457824707,
      "learning_rate": 4.292250746987104e-06,
      "loss": 3.3883,
      "step": 24215
    },
    {
      "epoch": 0.24637858072916666,
      "grad_norm": 15.532940864562988,
      "learning_rate": 4.2919719367830654e-06,
      "loss": 3.2584,
      "step": 24220
    },
    {
      "epoch": 0.246429443359375,
      "grad_norm": 15.537583351135254,
      "learning_rate": 4.29169308073185e-06,
      "loss": 3.459,
      "step": 24225
    },
    {
      "epoch": 0.24648030598958334,
      "grad_norm": 9.399313926696777,
      "learning_rate": 4.291414178840592e-06,
      "loss": 3.0459,
      "step": 24230
    },
    {
      "epoch": 0.24653116861979166,
      "grad_norm": 9.091721534729004,
      "learning_rate": 4.291135231116426e-06,
      "loss": 3.4614,
      "step": 24235
    },
    {
      "epoch": 0.24658203125,
      "grad_norm": 12.210503578186035,
      "learning_rate": 4.29085623756649e-06,
      "loss": 3.6569,
      "step": 24240
    },
    {
      "epoch": 0.24663289388020834,
      "grad_norm": 10.190805435180664,
      "learning_rate": 4.2905771981979225e-06,
      "loss": 3.3828,
      "step": 24245
    },
    {
      "epoch": 0.24668375651041666,
      "grad_norm": 8.21806812286377,
      "learning_rate": 4.290298113017861e-06,
      "loss": 3.4683,
      "step": 24250
    },
    {
      "epoch": 0.246734619140625,
      "grad_norm": 7.7201762199401855,
      "learning_rate": 4.290018982033447e-06,
      "loss": 3.3651,
      "step": 24255
    },
    {
      "epoch": 0.24678548177083334,
      "grad_norm": 14.318398475646973,
      "learning_rate": 4.289739805251821e-06,
      "loss": 3.1525,
      "step": 24260
    },
    {
      "epoch": 0.24683634440104166,
      "grad_norm": 8.810102462768555,
      "learning_rate": 4.2894605826801265e-06,
      "loss": 3.2851,
      "step": 24265
    },
    {
      "epoch": 0.24688720703125,
      "grad_norm": 13.633394241333008,
      "learning_rate": 4.289181314325508e-06,
      "loss": 3.3869,
      "step": 24270
    },
    {
      "epoch": 0.24693806966145834,
      "grad_norm": 17.578227996826172,
      "learning_rate": 4.288902000195108e-06,
      "loss": 3.2742,
      "step": 24275
    },
    {
      "epoch": 0.24698893229166666,
      "grad_norm": 12.528148651123047,
      "learning_rate": 4.288622640296076e-06,
      "loss": 3.379,
      "step": 24280
    },
    {
      "epoch": 0.247039794921875,
      "grad_norm": 12.188680648803711,
      "learning_rate": 4.288343234635557e-06,
      "loss": 3.4696,
      "step": 24285
    },
    {
      "epoch": 0.24709065755208334,
      "grad_norm": 9.514275550842285,
      "learning_rate": 4.2880637832207005e-06,
      "loss": 3.2483,
      "step": 24290
    },
    {
      "epoch": 0.24714152018229166,
      "grad_norm": 13.15481948852539,
      "learning_rate": 4.287784286058655e-06,
      "loss": 3.7612,
      "step": 24295
    },
    {
      "epoch": 0.2471923828125,
      "grad_norm": 13.95080280303955,
      "learning_rate": 4.287504743156574e-06,
      "loss": 3.2929,
      "step": 24300
    },
    {
      "epoch": 0.24724324544270834,
      "grad_norm": 10.395090103149414,
      "learning_rate": 4.287225154521606e-06,
      "loss": 3.3828,
      "step": 24305
    },
    {
      "epoch": 0.24729410807291666,
      "grad_norm": 13.436478614807129,
      "learning_rate": 4.286945520160907e-06,
      "loss": 3.0997,
      "step": 24310
    },
    {
      "epoch": 0.247344970703125,
      "grad_norm": 13.696346282958984,
      "learning_rate": 4.286665840081631e-06,
      "loss": 3.2578,
      "step": 24315
    },
    {
      "epoch": 0.24739583333333334,
      "grad_norm": 14.64127254486084,
      "learning_rate": 4.286386114290932e-06,
      "loss": 3.3801,
      "step": 24320
    },
    {
      "epoch": 0.24744669596354166,
      "grad_norm": 14.282286643981934,
      "learning_rate": 4.286106342795968e-06,
      "loss": 3.4712,
      "step": 24325
    },
    {
      "epoch": 0.24749755859375,
      "grad_norm": 12.41998291015625,
      "learning_rate": 4.285826525603896e-06,
      "loss": 3.613,
      "step": 24330
    },
    {
      "epoch": 0.24754842122395834,
      "grad_norm": 11.224750518798828,
      "learning_rate": 4.285546662721876e-06,
      "loss": 3.2357,
      "step": 24335
    },
    {
      "epoch": 0.24759928385416666,
      "grad_norm": 10.853987693786621,
      "learning_rate": 4.2852667541570685e-06,
      "loss": 3.1096,
      "step": 24340
    },
    {
      "epoch": 0.247650146484375,
      "grad_norm": 9.90202522277832,
      "learning_rate": 4.284986799916633e-06,
      "loss": 3.5553,
      "step": 24345
    },
    {
      "epoch": 0.24770100911458334,
      "grad_norm": 13.316431999206543,
      "learning_rate": 4.284706800007734e-06,
      "loss": 3.4482,
      "step": 24350
    },
    {
      "epoch": 0.24775187174479166,
      "grad_norm": 9.244454383850098,
      "learning_rate": 4.284426754437535e-06,
      "loss": 3.1333,
      "step": 24355
    },
    {
      "epoch": 0.247802734375,
      "grad_norm": 14.12641429901123,
      "learning_rate": 4.2841466632132e-06,
      "loss": 3.245,
      "step": 24360
    },
    {
      "epoch": 0.24785359700520834,
      "grad_norm": 14.708112716674805,
      "learning_rate": 4.283866526341895e-06,
      "loss": 3.0999,
      "step": 24365
    },
    {
      "epoch": 0.24790445963541666,
      "grad_norm": 8.532719612121582,
      "learning_rate": 4.283586343830789e-06,
      "loss": 3.2599,
      "step": 24370
    },
    {
      "epoch": 0.247955322265625,
      "grad_norm": 14.071338653564453,
      "learning_rate": 4.283306115687048e-06,
      "loss": 3.2315,
      "step": 24375
    },
    {
      "epoch": 0.24800618489583334,
      "grad_norm": 10.1383695602417,
      "learning_rate": 4.2830258419178436e-06,
      "loss": 3.3294,
      "step": 24380
    },
    {
      "epoch": 0.24805704752604166,
      "grad_norm": 14.600129127502441,
      "learning_rate": 4.2827455225303445e-06,
      "loss": 3.4252,
      "step": 24385
    },
    {
      "epoch": 0.24810791015625,
      "grad_norm": 12.649153709411621,
      "learning_rate": 4.2824651575317256e-06,
      "loss": 3.4132,
      "step": 24390
    },
    {
      "epoch": 0.24815877278645834,
      "grad_norm": 15.295096397399902,
      "learning_rate": 4.282184746929157e-06,
      "loss": 3.549,
      "step": 24395
    },
    {
      "epoch": 0.24820963541666666,
      "grad_norm": 16.297683715820312,
      "learning_rate": 4.2819042907298135e-06,
      "loss": 3.3736,
      "step": 24400
    },
    {
      "epoch": 0.248260498046875,
      "grad_norm": 14.560355186462402,
      "learning_rate": 4.281623788940872e-06,
      "loss": 3.2375,
      "step": 24405
    },
    {
      "epoch": 0.24831136067708334,
      "grad_norm": 11.617905616760254,
      "learning_rate": 4.281343241569508e-06,
      "loss": 3.1846,
      "step": 24410
    },
    {
      "epoch": 0.24836222330729166,
      "grad_norm": 12.611489295959473,
      "learning_rate": 4.281062648622899e-06,
      "loss": 3.4114,
      "step": 24415
    },
    {
      "epoch": 0.2484130859375,
      "grad_norm": 40.78852844238281,
      "learning_rate": 4.280782010108224e-06,
      "loss": 3.2481,
      "step": 24420
    },
    {
      "epoch": 0.24846394856770834,
      "grad_norm": 14.916101455688477,
      "learning_rate": 4.280501326032665e-06,
      "loss": 3.2941,
      "step": 24425
    },
    {
      "epoch": 0.24851481119791666,
      "grad_norm": 13.82861614227295,
      "learning_rate": 4.280220596403399e-06,
      "loss": 3.34,
      "step": 24430
    },
    {
      "epoch": 0.248565673828125,
      "grad_norm": 12.18250846862793,
      "learning_rate": 4.279939821227614e-06,
      "loss": 3.1754,
      "step": 24435
    },
    {
      "epoch": 0.24861653645833334,
      "grad_norm": 14.767477989196777,
      "learning_rate": 4.279659000512488e-06,
      "loss": 3.298,
      "step": 24440
    },
    {
      "epoch": 0.24866739908854166,
      "grad_norm": 13.104989051818848,
      "learning_rate": 4.27937813426521e-06,
      "loss": 3.068,
      "step": 24445
    },
    {
      "epoch": 0.24871826171875,
      "grad_norm": 7.446710586547852,
      "learning_rate": 4.279097222492963e-06,
      "loss": 2.9791,
      "step": 24450
    },
    {
      "epoch": 0.24876912434895834,
      "grad_norm": 8.23807144165039,
      "learning_rate": 4.278816265202936e-06,
      "loss": 3.1207,
      "step": 24455
    },
    {
      "epoch": 0.24881998697916666,
      "grad_norm": 14.00052547454834,
      "learning_rate": 4.278535262402316e-06,
      "loss": 3.3176,
      "step": 24460
    },
    {
      "epoch": 0.248870849609375,
      "grad_norm": 15.20098876953125,
      "learning_rate": 4.278254214098293e-06,
      "loss": 3.3625,
      "step": 24465
    },
    {
      "epoch": 0.24892171223958334,
      "grad_norm": 13.704755783081055,
      "learning_rate": 4.277973120298058e-06,
      "loss": 3.3232,
      "step": 24470
    },
    {
      "epoch": 0.24897257486979166,
      "grad_norm": 10.962750434875488,
      "learning_rate": 4.277691981008802e-06,
      "loss": 3.4346,
      "step": 24475
    },
    {
      "epoch": 0.2490234375,
      "grad_norm": 12.34006404876709,
      "learning_rate": 4.277410796237717e-06,
      "loss": 3.3349,
      "step": 24480
    },
    {
      "epoch": 0.24907430013020834,
      "grad_norm": 13.989150047302246,
      "learning_rate": 4.277129565991999e-06,
      "loss": 3.5294,
      "step": 24485
    },
    {
      "epoch": 0.24912516276041666,
      "grad_norm": 10.373230934143066,
      "learning_rate": 4.276848290278842e-06,
      "loss": 3.3079,
      "step": 24490
    },
    {
      "epoch": 0.249176025390625,
      "grad_norm": 10.260279655456543,
      "learning_rate": 4.276566969105443e-06,
      "loss": 3.2219,
      "step": 24495
    },
    {
      "epoch": 0.24922688802083334,
      "grad_norm": 9.731127738952637,
      "learning_rate": 4.276285602478999e-06,
      "loss": 3.3173,
      "step": 24500
    },
    {
      "epoch": 0.24927775065104166,
      "grad_norm": 11.35506534576416,
      "learning_rate": 4.276004190406708e-06,
      "loss": 3.2735,
      "step": 24505
    },
    {
      "epoch": 0.24932861328125,
      "grad_norm": 8.43806266784668,
      "learning_rate": 4.2757227328957715e-06,
      "loss": 3.1982,
      "step": 24510
    },
    {
      "epoch": 0.24937947591145834,
      "grad_norm": 9.247403144836426,
      "learning_rate": 4.27544122995339e-06,
      "loss": 3.5729,
      "step": 24515
    },
    {
      "epoch": 0.24943033854166666,
      "grad_norm": 14.659168243408203,
      "learning_rate": 4.275159681586766e-06,
      "loss": 3.3769,
      "step": 24520
    },
    {
      "epoch": 0.249481201171875,
      "grad_norm": 11.563913345336914,
      "learning_rate": 4.274878087803101e-06,
      "loss": 3.3972,
      "step": 24525
    },
    {
      "epoch": 0.24953206380208334,
      "grad_norm": 14.300829887390137,
      "learning_rate": 4.2745964486096015e-06,
      "loss": 2.9669,
      "step": 24530
    },
    {
      "epoch": 0.24958292643229166,
      "grad_norm": 8.588785171508789,
      "learning_rate": 4.274314764013472e-06,
      "loss": 3.1472,
      "step": 24535
    },
    {
      "epoch": 0.2496337890625,
      "grad_norm": 15.398540496826172,
      "learning_rate": 4.274033034021921e-06,
      "loss": 3.4659,
      "step": 24540
    },
    {
      "epoch": 0.24968465169270834,
      "grad_norm": 9.775829315185547,
      "learning_rate": 4.273751258642154e-06,
      "loss": 3.5144,
      "step": 24545
    },
    {
      "epoch": 0.24973551432291666,
      "grad_norm": 10.324780464172363,
      "learning_rate": 4.2734694378813825e-06,
      "loss": 2.9803,
      "step": 24550
    },
    {
      "epoch": 0.249786376953125,
      "grad_norm": 17.24770164489746,
      "learning_rate": 4.273187571746815e-06,
      "loss": 3.6072,
      "step": 24555
    },
    {
      "epoch": 0.24983723958333334,
      "grad_norm": 13.051507949829102,
      "learning_rate": 4.272905660245664e-06,
      "loss": 3.2217,
      "step": 24560
    },
    {
      "epoch": 0.24988810221354166,
      "grad_norm": 15.218391418457031,
      "learning_rate": 4.272623703385141e-06,
      "loss": 3.1289,
      "step": 24565
    },
    {
      "epoch": 0.24993896484375,
      "grad_norm": 11.163311004638672,
      "learning_rate": 4.272341701172462e-06,
      "loss": 3.6372,
      "step": 24570
    },
    {
      "epoch": 0.24998982747395834,
      "grad_norm": 8.303897857666016,
      "learning_rate": 4.27205965361484e-06,
      "loss": 3.2102,
      "step": 24575
    },
    {
      "epoch": 0.25,
      "eval_loss": 3.3746256828308105,
      "eval_runtime": 161.3971,
      "eval_samples_per_second": 12.435,
      "eval_steps_per_second": 12.435,
      "step": 24576
    },
    {
      "epoch": 0.2500406901041667,
      "grad_norm": 16.447832107543945,
      "learning_rate": 4.271777560719494e-06,
      "loss": 3.4576,
      "step": 24580
    },
    {
      "epoch": 0.250091552734375,
      "grad_norm": 14.935396194458008,
      "learning_rate": 4.271495422493636e-06,
      "loss": 2.9597,
      "step": 24585
    },
    {
      "epoch": 0.2501424153645833,
      "grad_norm": 12.183156967163086,
      "learning_rate": 4.271213238944489e-06,
      "loss": 3.1356,
      "step": 24590
    },
    {
      "epoch": 0.2501932779947917,
      "grad_norm": 9.555173873901367,
      "learning_rate": 4.270931010079271e-06,
      "loss": 3.418,
      "step": 24595
    },
    {
      "epoch": 0.250244140625,
      "grad_norm": 13.501853942871094,
      "learning_rate": 4.270648735905203e-06,
      "loss": 3.4476,
      "step": 24600
    },
    {
      "epoch": 0.2502950032552083,
      "grad_norm": 9.53705883026123,
      "learning_rate": 4.270366416429506e-06,
      "loss": 3.3436,
      "step": 24605
    },
    {
      "epoch": 0.2503458658854167,
      "grad_norm": 10.591668128967285,
      "learning_rate": 4.270084051659404e-06,
      "loss": 3.6996,
      "step": 24610
    },
    {
      "epoch": 0.250396728515625,
      "grad_norm": 12.271478652954102,
      "learning_rate": 4.2698016416021215e-06,
      "loss": 3.1625,
      "step": 24615
    },
    {
      "epoch": 0.2504475911458333,
      "grad_norm": 10.937945365905762,
      "learning_rate": 4.269519186264884e-06,
      "loss": 3.0254,
      "step": 24620
    },
    {
      "epoch": 0.2504984537760417,
      "grad_norm": 13.449241638183594,
      "learning_rate": 4.269236685654916e-06,
      "loss": 3.3296,
      "step": 24625
    },
    {
      "epoch": 0.25054931640625,
      "grad_norm": 13.686077117919922,
      "learning_rate": 4.268954139779447e-06,
      "loss": 3.2587,
      "step": 24630
    },
    {
      "epoch": 0.2506001790364583,
      "grad_norm": 10.223738670349121,
      "learning_rate": 4.268671548645707e-06,
      "loss": 3.5311,
      "step": 24635
    },
    {
      "epoch": 0.2506510416666667,
      "grad_norm": 8.637419700622559,
      "learning_rate": 4.268388912260924e-06,
      "loss": 3.1581,
      "step": 24640
    },
    {
      "epoch": 0.250701904296875,
      "grad_norm": 7.338236331939697,
      "learning_rate": 4.268106230632329e-06,
      "loss": 3.0383,
      "step": 24645
    },
    {
      "epoch": 0.2507527669270833,
      "grad_norm": 10.000212669372559,
      "learning_rate": 4.267823503767156e-06,
      "loss": 3.3966,
      "step": 24650
    },
    {
      "epoch": 0.2508036295572917,
      "grad_norm": 14.338170051574707,
      "learning_rate": 4.267540731672638e-06,
      "loss": 3.561,
      "step": 24655
    },
    {
      "epoch": 0.2508544921875,
      "grad_norm": 8.327690124511719,
      "learning_rate": 4.267257914356009e-06,
      "loss": 3.2426,
      "step": 24660
    },
    {
      "epoch": 0.2509053548177083,
      "grad_norm": 12.349832534790039,
      "learning_rate": 4.2669750518245055e-06,
      "loss": 3.118,
      "step": 24665
    },
    {
      "epoch": 0.2509562174479167,
      "grad_norm": 14.336140632629395,
      "learning_rate": 4.266692144085364e-06,
      "loss": 3.5699,
      "step": 24670
    },
    {
      "epoch": 0.251007080078125,
      "grad_norm": 13.961627960205078,
      "learning_rate": 4.266409191145822e-06,
      "loss": 3.4062,
      "step": 24675
    },
    {
      "epoch": 0.2510579427083333,
      "grad_norm": 17.38461685180664,
      "learning_rate": 4.266126193013121e-06,
      "loss": 3.4216,
      "step": 24680
    },
    {
      "epoch": 0.2511088053385417,
      "grad_norm": 14.52137565612793,
      "learning_rate": 4.265843149694499e-06,
      "loss": 3.3358,
      "step": 24685
    },
    {
      "epoch": 0.25115966796875,
      "grad_norm": 9.268922805786133,
      "learning_rate": 4.265560061197198e-06,
      "loss": 3.5275,
      "step": 24690
    },
    {
      "epoch": 0.2512105305989583,
      "grad_norm": 11.482409477233887,
      "learning_rate": 4.265276927528462e-06,
      "loss": 3.5046,
      "step": 24695
    },
    {
      "epoch": 0.2512613932291667,
      "grad_norm": 11.873368263244629,
      "learning_rate": 4.264993748695534e-06,
      "loss": 3.3458,
      "step": 24700
    },
    {
      "epoch": 0.251312255859375,
      "grad_norm": 8.131156921386719,
      "learning_rate": 4.264710524705659e-06,
      "loss": 3.0646,
      "step": 24705
    },
    {
      "epoch": 0.2513631184895833,
      "grad_norm": 7.6707048416137695,
      "learning_rate": 4.264427255566084e-06,
      "loss": 3.1548,
      "step": 24710
    },
    {
      "epoch": 0.2514139811197917,
      "grad_norm": 13.597451210021973,
      "learning_rate": 4.2641439412840555e-06,
      "loss": 3.3094,
      "step": 24715
    },
    {
      "epoch": 0.25146484375,
      "grad_norm": 9.331670761108398,
      "learning_rate": 4.263860581866823e-06,
      "loss": 3.2787,
      "step": 24720
    },
    {
      "epoch": 0.2515157063802083,
      "grad_norm": 12.325285911560059,
      "learning_rate": 4.263577177321635e-06,
      "loss": 3.5232,
      "step": 24725
    },
    {
      "epoch": 0.2515665690104167,
      "grad_norm": 10.015127182006836,
      "learning_rate": 4.2632937276557425e-06,
      "loss": 3.3396,
      "step": 24730
    },
    {
      "epoch": 0.251617431640625,
      "grad_norm": 11.384984970092773,
      "learning_rate": 4.263010232876399e-06,
      "loss": 3.4962,
      "step": 24735
    },
    {
      "epoch": 0.2516682942708333,
      "grad_norm": 15.124960899353027,
      "learning_rate": 4.262726692990856e-06,
      "loss": 3.1269,
      "step": 24740
    },
    {
      "epoch": 0.2517191569010417,
      "grad_norm": 11.73608112335205,
      "learning_rate": 4.262443108006369e-06,
      "loss": 3.5437,
      "step": 24745
    },
    {
      "epoch": 0.25177001953125,
      "grad_norm": 12.246577262878418,
      "learning_rate": 4.262159477930192e-06,
      "loss": 3.264,
      "step": 24750
    },
    {
      "epoch": 0.2518208821614583,
      "grad_norm": 15.663727760314941,
      "learning_rate": 4.261875802769583e-06,
      "loss": 3.2328,
      "step": 24755
    },
    {
      "epoch": 0.2518717447916667,
      "grad_norm": 10.710125923156738,
      "learning_rate": 4.261592082531799e-06,
      "loss": 3.3633,
      "step": 24760
    },
    {
      "epoch": 0.251922607421875,
      "grad_norm": 12.465362548828125,
      "learning_rate": 4.261308317224099e-06,
      "loss": 3.4696,
      "step": 24765
    },
    {
      "epoch": 0.2519734700520833,
      "grad_norm": 14.243422508239746,
      "learning_rate": 4.261024506853744e-06,
      "loss": 3.5099,
      "step": 24770
    },
    {
      "epoch": 0.2520243326822917,
      "grad_norm": 16.999818801879883,
      "learning_rate": 4.260740651427994e-06,
      "loss": 3.5493,
      "step": 24775
    },
    {
      "epoch": 0.2520751953125,
      "grad_norm": 13.684869766235352,
      "learning_rate": 4.260456750954111e-06,
      "loss": 3.6394,
      "step": 24780
    },
    {
      "epoch": 0.2521260579427083,
      "grad_norm": 11.110882759094238,
      "learning_rate": 4.26017280543936e-06,
      "loss": 3.1611,
      "step": 24785
    },
    {
      "epoch": 0.2521769205729167,
      "grad_norm": 9.618308067321777,
      "learning_rate": 4.259888814891004e-06,
      "loss": 3.4842,
      "step": 24790
    },
    {
      "epoch": 0.252227783203125,
      "grad_norm": 14.321953773498535,
      "learning_rate": 4.259604779316311e-06,
      "loss": 3.167,
      "step": 24795
    },
    {
      "epoch": 0.2522786458333333,
      "grad_norm": 7.829379558563232,
      "learning_rate": 4.2593206987225466e-06,
      "loss": 3.2783,
      "step": 24800
    },
    {
      "epoch": 0.2523295084635417,
      "grad_norm": 11.342432022094727,
      "learning_rate": 4.2590365731169784e-06,
      "loss": 3.3707,
      "step": 24805
    },
    {
      "epoch": 0.25238037109375,
      "grad_norm": 12.385947227478027,
      "learning_rate": 4.258752402506877e-06,
      "loss": 3.6793,
      "step": 24810
    },
    {
      "epoch": 0.2524312337239583,
      "grad_norm": 15.225542068481445,
      "learning_rate": 4.258468186899513e-06,
      "loss": 3.1762,
      "step": 24815
    },
    {
      "epoch": 0.2524820963541667,
      "grad_norm": 10.188759803771973,
      "learning_rate": 4.258183926302157e-06,
      "loss": 3.0373,
      "step": 24820
    },
    {
      "epoch": 0.252532958984375,
      "grad_norm": 12.267877578735352,
      "learning_rate": 4.257899620722081e-06,
      "loss": 3.3485,
      "step": 24825
    },
    {
      "epoch": 0.2525838216145833,
      "grad_norm": 8.663679122924805,
      "learning_rate": 4.25761527016656e-06,
      "loss": 3.5647,
      "step": 24830
    },
    {
      "epoch": 0.2526346842447917,
      "grad_norm": 11.342432022094727,
      "learning_rate": 4.257330874642869e-06,
      "loss": 3.8163,
      "step": 24835
    },
    {
      "epoch": 0.252685546875,
      "grad_norm": 14.044656753540039,
      "learning_rate": 4.257046434158284e-06,
      "loss": 3.6804,
      "step": 24840
    },
    {
      "epoch": 0.2527364095052083,
      "grad_norm": 14.019876480102539,
      "learning_rate": 4.256761948720083e-06,
      "loss": 3.2891,
      "step": 24845
    },
    {
      "epoch": 0.2527872721354167,
      "grad_norm": 8.586852073669434,
      "learning_rate": 4.256477418335543e-06,
      "loss": 3.3606,
      "step": 24850
    },
    {
      "epoch": 0.252838134765625,
      "grad_norm": 15.628178596496582,
      "learning_rate": 4.256192843011946e-06,
      "loss": 3.695,
      "step": 24855
    },
    {
      "epoch": 0.2528889973958333,
      "grad_norm": 12.91670036315918,
      "learning_rate": 4.2559082227565696e-06,
      "loss": 3.2321,
      "step": 24860
    },
    {
      "epoch": 0.2529398600260417,
      "grad_norm": 15.387182235717773,
      "learning_rate": 4.255623557576699e-06,
      "loss": 3.7515,
      "step": 24865
    },
    {
      "epoch": 0.25299072265625,
      "grad_norm": 11.862276077270508,
      "learning_rate": 4.255338847479615e-06,
      "loss": 3.1253,
      "step": 24870
    },
    {
      "epoch": 0.2530415852864583,
      "grad_norm": 16.38576889038086,
      "learning_rate": 4.255054092472604e-06,
      "loss": 3.1637,
      "step": 24875
    },
    {
      "epoch": 0.2530924479166667,
      "grad_norm": 8.539254188537598,
      "learning_rate": 4.254769292562948e-06,
      "loss": 3.4981,
      "step": 24880
    },
    {
      "epoch": 0.253143310546875,
      "grad_norm": 7.540289402008057,
      "learning_rate": 4.2544844477579375e-06,
      "loss": 3.4908,
      "step": 24885
    },
    {
      "epoch": 0.2531941731770833,
      "grad_norm": 11.539024353027344,
      "learning_rate": 4.254199558064857e-06,
      "loss": 3.0611,
      "step": 24890
    },
    {
      "epoch": 0.2532450358072917,
      "grad_norm": 9.188528060913086,
      "learning_rate": 4.2539146234909965e-06,
      "loss": 3.7936,
      "step": 24895
    },
    {
      "epoch": 0.2532958984375,
      "grad_norm": 13.862792015075684,
      "learning_rate": 4.2536296440436465e-06,
      "loss": 3.2935,
      "step": 24900
    },
    {
      "epoch": 0.2533467610677083,
      "grad_norm": 15.22266960144043,
      "learning_rate": 4.253344619730098e-06,
      "loss": 3.6186,
      "step": 24905
    },
    {
      "epoch": 0.2533976236979167,
      "grad_norm": 8.168689727783203,
      "learning_rate": 4.2530595505576425e-06,
      "loss": 3.4356,
      "step": 24910
    },
    {
      "epoch": 0.253448486328125,
      "grad_norm": 8.944549560546875,
      "learning_rate": 4.252774436533574e-06,
      "loss": 3.3316,
      "step": 24915
    },
    {
      "epoch": 0.2534993489583333,
      "grad_norm": 12.66011905670166,
      "learning_rate": 4.252489277665188e-06,
      "loss": 3.8579,
      "step": 24920
    },
    {
      "epoch": 0.2535502115885417,
      "grad_norm": 14.050813674926758,
      "learning_rate": 4.252204073959778e-06,
      "loss": 3.3374,
      "step": 24925
    },
    {
      "epoch": 0.25360107421875,
      "grad_norm": 13.361396789550781,
      "learning_rate": 4.251918825424642e-06,
      "loss": 3.3719,
      "step": 24930
    },
    {
      "epoch": 0.2536519368489583,
      "grad_norm": 11.718053817749023,
      "learning_rate": 4.251633532067079e-06,
      "loss": 3.3046,
      "step": 24935
    },
    {
      "epoch": 0.2537027994791667,
      "grad_norm": 12.97679615020752,
      "learning_rate": 4.251348193894386e-06,
      "loss": 3.5764,
      "step": 24940
    },
    {
      "epoch": 0.253753662109375,
      "grad_norm": 9.908891677856445,
      "learning_rate": 4.251062810913867e-06,
      "loss": 3.2527,
      "step": 24945
    },
    {
      "epoch": 0.2538045247395833,
      "grad_norm": 13.16048526763916,
      "learning_rate": 4.250777383132819e-06,
      "loss": 3.912,
      "step": 24950
    },
    {
      "epoch": 0.2538553873697917,
      "grad_norm": 15.14319896697998,
      "learning_rate": 4.250491910558546e-06,
      "loss": 3.3679,
      "step": 24955
    },
    {
      "epoch": 0.25390625,
      "grad_norm": 10.365280151367188,
      "learning_rate": 4.250206393198354e-06,
      "loss": 3.3091,
      "step": 24960
    },
    {
      "epoch": 0.2539571126302083,
      "grad_norm": 12.37360668182373,
      "learning_rate": 4.249920831059546e-06,
      "loss": 3.19,
      "step": 24965
    },
    {
      "epoch": 0.2540079752604167,
      "grad_norm": 10.952878952026367,
      "learning_rate": 4.249635224149428e-06,
      "loss": 3.2674,
      "step": 24970
    },
    {
      "epoch": 0.254058837890625,
      "grad_norm": 13.540122032165527,
      "learning_rate": 4.249349572475307e-06,
      "loss": 3.0913,
      "step": 24975
    },
    {
      "epoch": 0.2541097005208333,
      "grad_norm": 13.925262451171875,
      "learning_rate": 4.249063876044493e-06,
      "loss": 3.2861,
      "step": 24980
    },
    {
      "epoch": 0.2541605631510417,
      "grad_norm": 15.029991149902344,
      "learning_rate": 4.248778134864293e-06,
      "loss": 3.4239,
      "step": 24985
    },
    {
      "epoch": 0.25421142578125,
      "grad_norm": 14.305146217346191,
      "learning_rate": 4.248492348942021e-06,
      "loss": 3.9219,
      "step": 24990
    },
    {
      "epoch": 0.2542622884114583,
      "grad_norm": 8.43241024017334,
      "learning_rate": 4.248206518284984e-06,
      "loss": 3.3433,
      "step": 24995
    },
    {
      "epoch": 0.2543131510416667,
      "grad_norm": 8.676648139953613,
      "learning_rate": 4.247920642900499e-06,
      "loss": 3.4091,
      "step": 25000
    },
    {
      "epoch": 0.254364013671875,
      "grad_norm": 14.321340560913086,
      "learning_rate": 4.247634722795879e-06,
      "loss": 3.2965,
      "step": 25005
    },
    {
      "epoch": 0.2544148763020833,
      "grad_norm": 12.503641128540039,
      "learning_rate": 4.247348757978438e-06,
      "loss": 3.3245,
      "step": 25010
    },
    {
      "epoch": 0.2544657389322917,
      "grad_norm": 10.391291618347168,
      "learning_rate": 4.247062748455494e-06,
      "loss": 3.0437,
      "step": 25015
    },
    {
      "epoch": 0.2545166015625,
      "grad_norm": 9.460679054260254,
      "learning_rate": 4.246776694234364e-06,
      "loss": 3.0883,
      "step": 25020
    },
    {
      "epoch": 0.2545674641927083,
      "grad_norm": 9.043420791625977,
      "learning_rate": 4.2464905953223644e-06,
      "loss": 2.9897,
      "step": 25025
    },
    {
      "epoch": 0.2546183268229167,
      "grad_norm": 15.813292503356934,
      "learning_rate": 4.246204451726819e-06,
      "loss": 3.0896,
      "step": 25030
    },
    {
      "epoch": 0.254669189453125,
      "grad_norm": 10.58598804473877,
      "learning_rate": 4.245918263455045e-06,
      "loss": 3.1905,
      "step": 25035
    },
    {
      "epoch": 0.2547200520833333,
      "grad_norm": 7.66256856918335,
      "learning_rate": 4.2456320305143675e-06,
      "loss": 2.9339,
      "step": 25040
    },
    {
      "epoch": 0.2547709147135417,
      "grad_norm": 13.066410064697266,
      "learning_rate": 4.245345752912107e-06,
      "loss": 3.4187,
      "step": 25045
    },
    {
      "epoch": 0.25482177734375,
      "grad_norm": 10.616400718688965,
      "learning_rate": 4.245059430655589e-06,
      "loss": 3.6924,
      "step": 25050
    },
    {
      "epoch": 0.2548726399739583,
      "grad_norm": 9.606547355651855,
      "learning_rate": 4.24477306375214e-06,
      "loss": 3.4028,
      "step": 25055
    },
    {
      "epoch": 0.2549235026041667,
      "grad_norm": 12.106054306030273,
      "learning_rate": 4.244486652209086e-06,
      "loss": 3.0908,
      "step": 25060
    },
    {
      "epoch": 0.254974365234375,
      "grad_norm": 10.97599983215332,
      "learning_rate": 4.244200196033753e-06,
      "loss": 3.1756,
      "step": 25065
    },
    {
      "epoch": 0.2550252278645833,
      "grad_norm": 15.771632194519043,
      "learning_rate": 4.243913695233472e-06,
      "loss": 3.3499,
      "step": 25070
    },
    {
      "epoch": 0.2550760904947917,
      "grad_norm": 13.067724227905273,
      "learning_rate": 4.243627149815574e-06,
      "loss": 3.4849,
      "step": 25075
    },
    {
      "epoch": 0.255126953125,
      "grad_norm": 15.27235221862793,
      "learning_rate": 4.2433405597873865e-06,
      "loss": 3.3464,
      "step": 25080
    },
    {
      "epoch": 0.2551778157552083,
      "grad_norm": 14.676030158996582,
      "learning_rate": 4.243053925156245e-06,
      "loss": 3.5484,
      "step": 25085
    },
    {
      "epoch": 0.2552286783854167,
      "grad_norm": 14.75976276397705,
      "learning_rate": 4.242767245929482e-06,
      "loss": 3.4655,
      "step": 25090
    },
    {
      "epoch": 0.255279541015625,
      "grad_norm": 10.113024711608887,
      "learning_rate": 4.242480522114432e-06,
      "loss": 3.4032,
      "step": 25095
    },
    {
      "epoch": 0.2553304036458333,
      "grad_norm": 15.981361389160156,
      "learning_rate": 4.242193753718431e-06,
      "loss": 2.9224,
      "step": 25100
    },
    {
      "epoch": 0.2553812662760417,
      "grad_norm": 10.057245254516602,
      "learning_rate": 4.241906940748815e-06,
      "loss": 3.4517,
      "step": 25105
    },
    {
      "epoch": 0.25543212890625,
      "grad_norm": 13.309122085571289,
      "learning_rate": 4.241620083212924e-06,
      "loss": 3.3819,
      "step": 25110
    },
    {
      "epoch": 0.2554829915364583,
      "grad_norm": 13.638635635375977,
      "learning_rate": 4.2413331811180945e-06,
      "loss": 3.2491,
      "step": 25115
    },
    {
      "epoch": 0.2555338541666667,
      "grad_norm": 9.086438179016113,
      "learning_rate": 4.24104623447167e-06,
      "loss": 3.3659,
      "step": 25120
    },
    {
      "epoch": 0.255584716796875,
      "grad_norm": 15.72012710571289,
      "learning_rate": 4.240759243280989e-06,
      "loss": 3.2641,
      "step": 25125
    },
    {
      "epoch": 0.2556355794270833,
      "grad_norm": 8.95547866821289,
      "learning_rate": 4.240472207553395e-06,
      "loss": 2.916,
      "step": 25130
    },
    {
      "epoch": 0.2556864420572917,
      "grad_norm": 10.37775993347168,
      "learning_rate": 4.240185127296233e-06,
      "loss": 3.2412,
      "step": 25135
    },
    {
      "epoch": 0.2557373046875,
      "grad_norm": 12.915329933166504,
      "learning_rate": 4.239898002516848e-06,
      "loss": 3.3121,
      "step": 25140
    },
    {
      "epoch": 0.2557881673177083,
      "grad_norm": 14.557816505432129,
      "learning_rate": 4.239610833222583e-06,
      "loss": 3.604,
      "step": 25145
    },
    {
      "epoch": 0.2558390299479167,
      "grad_norm": 10.346381187438965,
      "learning_rate": 4.239323619420789e-06,
      "loss": 3.3416,
      "step": 25150
    },
    {
      "epoch": 0.255889892578125,
      "grad_norm": 9.316656112670898,
      "learning_rate": 4.239036361118811e-06,
      "loss": 3.5399,
      "step": 25155
    },
    {
      "epoch": 0.2559407552083333,
      "grad_norm": 10.786599159240723,
      "learning_rate": 4.238749058324001e-06,
      "loss": 3.5514,
      "step": 25160
    },
    {
      "epoch": 0.2559916178385417,
      "grad_norm": 12.696056365966797,
      "learning_rate": 4.238461711043708e-06,
      "loss": 3.1802,
      "step": 25165
    },
    {
      "epoch": 0.25604248046875,
      "grad_norm": 12.015348434448242,
      "learning_rate": 4.238174319285283e-06,
      "loss": 3.1336,
      "step": 25170
    },
    {
      "epoch": 0.2560933430989583,
      "grad_norm": 12.305672645568848,
      "learning_rate": 4.237886883056082e-06,
      "loss": 3.1156,
      "step": 25175
    },
    {
      "epoch": 0.2561442057291667,
      "grad_norm": 15.55685043334961,
      "learning_rate": 4.237599402363456e-06,
      "loss": 3.3255,
      "step": 25180
    },
    {
      "epoch": 0.256195068359375,
      "grad_norm": 13.057061195373535,
      "learning_rate": 4.237311877214761e-06,
      "loss": 3.5198,
      "step": 25185
    },
    {
      "epoch": 0.2562459309895833,
      "grad_norm": 13.559088706970215,
      "learning_rate": 4.237024307617354e-06,
      "loss": 3.0956,
      "step": 25190
    },
    {
      "epoch": 0.2562967936197917,
      "grad_norm": 8.187592506408691,
      "learning_rate": 4.236736693578591e-06,
      "loss": 3.4795,
      "step": 25195
    },
    {
      "epoch": 0.25634765625,
      "grad_norm": 10.342399597167969,
      "learning_rate": 4.236449035105832e-06,
      "loss": 3.3141,
      "step": 25200
    },
    {
      "epoch": 0.2563985188802083,
      "grad_norm": 12.40994930267334,
      "learning_rate": 4.236161332206436e-06,
      "loss": 3.7125,
      "step": 25205
    },
    {
      "epoch": 0.2564493815104167,
      "grad_norm": 14.115476608276367,
      "learning_rate": 4.235873584887764e-06,
      "loss": 3.8294,
      "step": 25210
    },
    {
      "epoch": 0.256500244140625,
      "grad_norm": 9.711331367492676,
      "learning_rate": 4.2355857931571765e-06,
      "loss": 2.9399,
      "step": 25215
    },
    {
      "epoch": 0.2565511067708333,
      "grad_norm": 11.328681945800781,
      "learning_rate": 4.235297957022039e-06,
      "loss": 3.3246,
      "step": 25220
    },
    {
      "epoch": 0.2566019694010417,
      "grad_norm": 9.998047828674316,
      "learning_rate": 4.235010076489715e-06,
      "loss": 3.3671,
      "step": 25225
    },
    {
      "epoch": 0.25665283203125,
      "grad_norm": 9.312052726745605,
      "learning_rate": 4.234722151567569e-06,
      "loss": 3.6494,
      "step": 25230
    },
    {
      "epoch": 0.2567036946614583,
      "grad_norm": 14.352344512939453,
      "learning_rate": 4.2344341822629675e-06,
      "loss": 3.1944,
      "step": 25235
    },
    {
      "epoch": 0.2567545572916667,
      "grad_norm": 14.686880111694336,
      "learning_rate": 4.234146168583278e-06,
      "loss": 3.1691,
      "step": 25240
    },
    {
      "epoch": 0.256805419921875,
      "grad_norm": 11.81808853149414,
      "learning_rate": 4.233858110535871e-06,
      "loss": 3.2245,
      "step": 25245
    },
    {
      "epoch": 0.2568562825520833,
      "grad_norm": 11.544474601745605,
      "learning_rate": 4.233570008128115e-06,
      "loss": 3.4465,
      "step": 25250
    },
    {
      "epoch": 0.2569071451822917,
      "grad_norm": 9.362020492553711,
      "learning_rate": 4.233281861367381e-06,
      "loss": 3.4905,
      "step": 25255
    },
    {
      "epoch": 0.2569580078125,
      "grad_norm": 14.878250122070312,
      "learning_rate": 4.232993670261041e-06,
      "loss": 3.437,
      "step": 25260
    },
    {
      "epoch": 0.2570088704427083,
      "grad_norm": 10.850860595703125,
      "learning_rate": 4.23270543481647e-06,
      "loss": 3.1745,
      "step": 25265
    },
    {
      "epoch": 0.2570597330729167,
      "grad_norm": 7.937302112579346,
      "learning_rate": 4.23241715504104e-06,
      "loss": 3.3978,
      "step": 25270
    },
    {
      "epoch": 0.257110595703125,
      "grad_norm": 11.461185455322266,
      "learning_rate": 4.232128830942128e-06,
      "loss": 3.3198,
      "step": 25275
    },
    {
      "epoch": 0.2571614583333333,
      "grad_norm": 12.847282409667969,
      "learning_rate": 4.231840462527111e-06,
      "loss": 3.3679,
      "step": 25280
    },
    {
      "epoch": 0.2572123209635417,
      "grad_norm": 8.424904823303223,
      "learning_rate": 4.231552049803366e-06,
      "loss": 3.529,
      "step": 25285
    },
    {
      "epoch": 0.25726318359375,
      "grad_norm": 8.810689926147461,
      "learning_rate": 4.231263592778271e-06,
      "loss": 3.1633,
      "step": 25290
    },
    {
      "epoch": 0.2573140462239583,
      "grad_norm": 12.490745544433594,
      "learning_rate": 4.230975091459209e-06,
      "loss": 3.3352,
      "step": 25295
    },
    {
      "epoch": 0.2573649088541667,
      "grad_norm": 9.57895278930664,
      "learning_rate": 4.230686545853558e-06,
      "loss": 3.1129,
      "step": 25300
    },
    {
      "epoch": 0.257415771484375,
      "grad_norm": 11.672348022460938,
      "learning_rate": 4.230397955968704e-06,
      "loss": 3.7327,
      "step": 25305
    },
    {
      "epoch": 0.2574666341145833,
      "grad_norm": 13.51906967163086,
      "learning_rate": 4.230109321812027e-06,
      "loss": 3.0903,
      "step": 25310
    },
    {
      "epoch": 0.2575174967447917,
      "grad_norm": 8.938597679138184,
      "learning_rate": 4.2298206433909135e-06,
      "loss": 3.3603,
      "step": 25315
    },
    {
      "epoch": 0.257568359375,
      "grad_norm": 13.414045333862305,
      "learning_rate": 4.229531920712748e-06,
      "loss": 2.9802,
      "step": 25320
    },
    {
      "epoch": 0.2576192220052083,
      "grad_norm": 11.90649127960205,
      "learning_rate": 4.2292431537849194e-06,
      "loss": 3.3947,
      "step": 25325
    },
    {
      "epoch": 0.2576700846354167,
      "grad_norm": 16.518070220947266,
      "learning_rate": 4.228954342614814e-06,
      "loss": 3.4593,
      "step": 25330
    },
    {
      "epoch": 0.257720947265625,
      "grad_norm": 13.547666549682617,
      "learning_rate": 4.2286654872098225e-06,
      "loss": 3.7998,
      "step": 25335
    },
    {
      "epoch": 0.2577718098958333,
      "grad_norm": 10.90760326385498,
      "learning_rate": 4.2283765875773325e-06,
      "loss": 3.3047,
      "step": 25340
    },
    {
      "epoch": 0.2578226725260417,
      "grad_norm": 17.5865535736084,
      "learning_rate": 4.228087643724738e-06,
      "loss": 3.4747,
      "step": 25345
    },
    {
      "epoch": 0.25787353515625,
      "grad_norm": 10.39489459991455,
      "learning_rate": 4.227798655659431e-06,
      "loss": 3.4802,
      "step": 25350
    },
    {
      "epoch": 0.2579243977864583,
      "grad_norm": 12.231903076171875,
      "learning_rate": 4.227509623388805e-06,
      "loss": 3.5976,
      "step": 25355
    },
    {
      "epoch": 0.2579752604166667,
      "grad_norm": 15.524836540222168,
      "learning_rate": 4.2272205469202544e-06,
      "loss": 3.0221,
      "step": 25360
    },
    {
      "epoch": 0.258026123046875,
      "grad_norm": 13.699244499206543,
      "learning_rate": 4.226931426261175e-06,
      "loss": 3.6252,
      "step": 25365
    },
    {
      "epoch": 0.2580769856770833,
      "grad_norm": 10.88588809967041,
      "learning_rate": 4.2266422614189656e-06,
      "loss": 3.4606,
      "step": 25370
    },
    {
      "epoch": 0.2581278483072917,
      "grad_norm": 12.582985877990723,
      "learning_rate": 4.226353052401023e-06,
      "loss": 4.0478,
      "step": 25375
    },
    {
      "epoch": 0.2581787109375,
      "grad_norm": 14.824645042419434,
      "learning_rate": 4.2260637992147455e-06,
      "loss": 3.245,
      "step": 25380
    },
    {
      "epoch": 0.2582295735677083,
      "grad_norm": 8.575550079345703,
      "learning_rate": 4.225774501867536e-06,
      "loss": 3.2326,
      "step": 25385
    },
    {
      "epoch": 0.2582804361979167,
      "grad_norm": 11.453362464904785,
      "learning_rate": 4.225485160366795e-06,
      "loss": 3.3931,
      "step": 25390
    },
    {
      "epoch": 0.258331298828125,
      "grad_norm": 12.148056030273438,
      "learning_rate": 4.2251957747199246e-06,
      "loss": 3.3584,
      "step": 25395
    },
    {
      "epoch": 0.2583821614583333,
      "grad_norm": 13.610539436340332,
      "learning_rate": 4.22490634493433e-06,
      "loss": 3.7102,
      "step": 25400
    },
    {
      "epoch": 0.2584330240885417,
      "grad_norm": 15.177289962768555,
      "learning_rate": 4.224616871017414e-06,
      "loss": 3.6817,
      "step": 25405
    },
    {
      "epoch": 0.25848388671875,
      "grad_norm": 6.891851902008057,
      "learning_rate": 4.224327352976585e-06,
      "loss": 3.2669,
      "step": 25410
    },
    {
      "epoch": 0.2585347493489583,
      "grad_norm": 14.693187713623047,
      "learning_rate": 4.224037790819249e-06,
      "loss": 3.4193,
      "step": 25415
    },
    {
      "epoch": 0.2585856119791667,
      "grad_norm": 12.441463470458984,
      "learning_rate": 4.223748184552815e-06,
      "loss": 3.5708,
      "step": 25420
    },
    {
      "epoch": 0.258636474609375,
      "grad_norm": 12.066529273986816,
      "learning_rate": 4.223458534184693e-06,
      "loss": 3.327,
      "step": 25425
    },
    {
      "epoch": 0.2586873372395833,
      "grad_norm": 13.036898612976074,
      "learning_rate": 4.223168839722293e-06,
      "loss": 3.4348,
      "step": 25430
    },
    {
      "epoch": 0.2587381998697917,
      "grad_norm": 8.6354341506958,
      "learning_rate": 4.222879101173025e-06,
      "loss": 3.3506,
      "step": 25435
    },
    {
      "epoch": 0.2587890625,
      "grad_norm": 15.209229469299316,
      "learning_rate": 4.222589318544305e-06,
      "loss": 3.3671,
      "step": 25440
    },
    {
      "epoch": 0.2588399251302083,
      "grad_norm": 8.996095657348633,
      "learning_rate": 4.222299491843544e-06,
      "loss": 3.3109,
      "step": 25445
    },
    {
      "epoch": 0.2588907877604167,
      "grad_norm": 15.306842803955078,
      "learning_rate": 4.2220096210781604e-06,
      "loss": 3.2306,
      "step": 25450
    },
    {
      "epoch": 0.258941650390625,
      "grad_norm": 10.3480863571167,
      "learning_rate": 4.221719706255568e-06,
      "loss": 3.7634,
      "step": 25455
    },
    {
      "epoch": 0.2589925130208333,
      "grad_norm": 8.632323265075684,
      "learning_rate": 4.221429747383185e-06,
      "loss": 3.5949,
      "step": 25460
    },
    {
      "epoch": 0.2590433756510417,
      "grad_norm": 7.815703392028809,
      "learning_rate": 4.221139744468431e-06,
      "loss": 3.553,
      "step": 25465
    },
    {
      "epoch": 0.25909423828125,
      "grad_norm": 11.302229881286621,
      "learning_rate": 4.220849697518723e-06,
      "loss": 3.4386,
      "step": 25470
    },
    {
      "epoch": 0.2591451009114583,
      "grad_norm": 14.221625328063965,
      "learning_rate": 4.2205596065414845e-06,
      "loss": 3.1683,
      "step": 25475
    },
    {
      "epoch": 0.2591959635416667,
      "grad_norm": 12.27514362335205,
      "learning_rate": 4.220269471544136e-06,
      "loss": 3.2244,
      "step": 25480
    },
    {
      "epoch": 0.259246826171875,
      "grad_norm": 13.549633026123047,
      "learning_rate": 4.2199792925341e-06,
      "loss": 3.4984,
      "step": 25485
    },
    {
      "epoch": 0.2592976888020833,
      "grad_norm": 10.52974796295166,
      "learning_rate": 4.219689069518802e-06,
      "loss": 3.3162,
      "step": 25490
    },
    {
      "epoch": 0.2593485514322917,
      "grad_norm": 10.306483268737793,
      "learning_rate": 4.219398802505668e-06,
      "loss": 2.8758,
      "step": 25495
    },
    {
      "epoch": 0.2593994140625,
      "grad_norm": 13.99836540222168,
      "learning_rate": 4.2191084915021215e-06,
      "loss": 3.3814,
      "step": 25500
    },
    {
      "epoch": 0.2594502766927083,
      "grad_norm": 14.140470504760742,
      "learning_rate": 4.218818136515593e-06,
      "loss": 3.4431,
      "step": 25505
    },
    {
      "epoch": 0.2595011393229167,
      "grad_norm": 9.10169506072998,
      "learning_rate": 4.2185277375535086e-06,
      "loss": 3.4482,
      "step": 25510
    },
    {
      "epoch": 0.259552001953125,
      "grad_norm": 16.0885009765625,
      "learning_rate": 4.218237294623301e-06,
      "loss": 3.5298,
      "step": 25515
    },
    {
      "epoch": 0.2596028645833333,
      "grad_norm": 17.33360481262207,
      "learning_rate": 4.217946807732398e-06,
      "loss": 3.2658,
      "step": 25520
    },
    {
      "epoch": 0.2596537272135417,
      "grad_norm": 11.709734916687012,
      "learning_rate": 4.217656276888233e-06,
      "loss": 3.4651,
      "step": 25525
    },
    {
      "epoch": 0.25970458984375,
      "grad_norm": 8.238667488098145,
      "learning_rate": 4.2173657020982405e-06,
      "loss": 3.2665,
      "step": 25530
    },
    {
      "epoch": 0.2597554524739583,
      "grad_norm": 6.8458099365234375,
      "learning_rate": 4.217075083369853e-06,
      "loss": 3.512,
      "step": 25535
    },
    {
      "epoch": 0.2598063151041667,
      "grad_norm": 15.638907432556152,
      "learning_rate": 4.216784420710506e-06,
      "loss": 3.6318,
      "step": 25540
    },
    {
      "epoch": 0.259857177734375,
      "grad_norm": 9.434146881103516,
      "learning_rate": 4.216493714127638e-06,
      "loss": 3.2256,
      "step": 25545
    },
    {
      "epoch": 0.2599080403645833,
      "grad_norm": 16.002166748046875,
      "learning_rate": 4.216202963628684e-06,
      "loss": 3.6784,
      "step": 25550
    },
    {
      "epoch": 0.2599589029947917,
      "grad_norm": 10.18635368347168,
      "learning_rate": 4.215912169221083e-06,
      "loss": 3.5994,
      "step": 25555
    },
    {
      "epoch": 0.260009765625,
      "grad_norm": 8.981634140014648,
      "learning_rate": 4.215621330912277e-06,
      "loss": 3.4051,
      "step": 25560
    },
    {
      "epoch": 0.2600606282552083,
      "grad_norm": 11.532533645629883,
      "learning_rate": 4.2153304487097065e-06,
      "loss": 3.3125,
      "step": 25565
    },
    {
      "epoch": 0.2601114908854167,
      "grad_norm": 11.587449073791504,
      "learning_rate": 4.215039522620813e-06,
      "loss": 3.2259,
      "step": 25570
    },
    {
      "epoch": 0.260162353515625,
      "grad_norm": 11.612677574157715,
      "learning_rate": 4.214748552653038e-06,
      "loss": 3.4293,
      "step": 25575
    },
    {
      "epoch": 0.2602132161458333,
      "grad_norm": 8.308237075805664,
      "learning_rate": 4.21445753881383e-06,
      "loss": 3.0921,
      "step": 25580
    },
    {
      "epoch": 0.2602640787760417,
      "grad_norm": 12.204816818237305,
      "learning_rate": 4.214166481110631e-06,
      "loss": 3.1295,
      "step": 25585
    },
    {
      "epoch": 0.26031494140625,
      "grad_norm": 12.297460556030273,
      "learning_rate": 4.213875379550889e-06,
      "loss": 3.623,
      "step": 25590
    },
    {
      "epoch": 0.2603658040364583,
      "grad_norm": 9.227754592895508,
      "learning_rate": 4.213584234142052e-06,
      "loss": 3.2325,
      "step": 25595
    },
    {
      "epoch": 0.2604166666666667,
      "grad_norm": 11.207293510437012,
      "learning_rate": 4.213293044891568e-06,
      "loss": 3.1893,
      "step": 25600
    },
    {
      "epoch": 0.260467529296875,
      "grad_norm": 7.930004596710205,
      "learning_rate": 4.213001811806888e-06,
      "loss": 3.0561,
      "step": 25605
    },
    {
      "epoch": 0.2605183919270833,
      "grad_norm": 14.136908531188965,
      "learning_rate": 4.212710534895462e-06,
      "loss": 3.5516,
      "step": 25610
    },
    {
      "epoch": 0.2605692545572917,
      "grad_norm": 14.007734298706055,
      "learning_rate": 4.212419214164744e-06,
      "loss": 3.3025,
      "step": 25615
    },
    {
      "epoch": 0.2606201171875,
      "grad_norm": 14.646519660949707,
      "learning_rate": 4.212127849622186e-06,
      "loss": 3.4518,
      "step": 25620
    },
    {
      "epoch": 0.2606709798177083,
      "grad_norm": 15.032630920410156,
      "learning_rate": 4.211836441275242e-06,
      "loss": 3.4148,
      "step": 25625
    },
    {
      "epoch": 0.2607218424479167,
      "grad_norm": 9.241325378417969,
      "learning_rate": 4.211544989131369e-06,
      "loss": 3.1756,
      "step": 25630
    },
    {
      "epoch": 0.260772705078125,
      "grad_norm": 13.654480934143066,
      "learning_rate": 4.211253493198023e-06,
      "loss": 3.1426,
      "step": 25635
    },
    {
      "epoch": 0.2608235677083333,
      "grad_norm": 10.502730369567871,
      "learning_rate": 4.210961953482662e-06,
      "loss": 3.0288,
      "step": 25640
    },
    {
      "epoch": 0.2608744303385417,
      "grad_norm": 11.17243480682373,
      "learning_rate": 4.210670369992745e-06,
      "loss": 3.2139,
      "step": 25645
    },
    {
      "epoch": 0.26092529296875,
      "grad_norm": 17.18389320373535,
      "learning_rate": 4.210378742735732e-06,
      "loss": 3.17,
      "step": 25650
    },
    {
      "epoch": 0.2609761555989583,
      "grad_norm": 10.495933532714844,
      "learning_rate": 4.210087071719084e-06,
      "loss": 3.1387,
      "step": 25655
    },
    {
      "epoch": 0.2610270182291667,
      "grad_norm": 21.2041015625,
      "learning_rate": 4.209795356950264e-06,
      "loss": 3.7895,
      "step": 25660
    },
    {
      "epoch": 0.261077880859375,
      "grad_norm": 13.002775192260742,
      "learning_rate": 4.209503598436735e-06,
      "loss": 3.4469,
      "step": 25665
    },
    {
      "epoch": 0.2611287434895833,
      "grad_norm": 14.954187393188477,
      "learning_rate": 4.209211796185961e-06,
      "loss": 3.4128,
      "step": 25670
    },
    {
      "epoch": 0.2611796061197917,
      "grad_norm": 12.93187141418457,
      "learning_rate": 4.20891995020541e-06,
      "loss": 3.4071,
      "step": 25675
    },
    {
      "epoch": 0.26123046875,
      "grad_norm": 13.916560173034668,
      "learning_rate": 4.208628060502545e-06,
      "loss": 3.174,
      "step": 25680
    },
    {
      "epoch": 0.2612813313802083,
      "grad_norm": 10.77669620513916,
      "learning_rate": 4.2083361270848374e-06,
      "loss": 3.2852,
      "step": 25685
    },
    {
      "epoch": 0.2613321940104167,
      "grad_norm": 9.5277099609375,
      "learning_rate": 4.2080441499597546e-06,
      "loss": 3.375,
      "step": 25690
    },
    {
      "epoch": 0.261383056640625,
      "grad_norm": 12.276413917541504,
      "learning_rate": 4.207752129134768e-06,
      "loss": 3.5693,
      "step": 25695
    },
    {
      "epoch": 0.2614339192708333,
      "grad_norm": 26.959915161132812,
      "learning_rate": 4.207460064617347e-06,
      "loss": 3.549,
      "step": 25700
    },
    {
      "epoch": 0.2614847819010417,
      "grad_norm": 11.67579460144043,
      "learning_rate": 4.207167956414965e-06,
      "loss": 3.703,
      "step": 25705
    },
    {
      "epoch": 0.26153564453125,
      "grad_norm": 10.390996932983398,
      "learning_rate": 4.206875804535095e-06,
      "loss": 3.3588,
      "step": 25710
    },
    {
      "epoch": 0.2615865071614583,
      "grad_norm": 17.176240921020508,
      "learning_rate": 4.206583608985213e-06,
      "loss": 3.6047,
      "step": 25715
    },
    {
      "epoch": 0.2616373697916667,
      "grad_norm": 8.651700019836426,
      "learning_rate": 4.206291369772793e-06,
      "loss": 3.5504,
      "step": 25720
    },
    {
      "epoch": 0.261688232421875,
      "grad_norm": 12.320103645324707,
      "learning_rate": 4.205999086905313e-06,
      "loss": 3.5346,
      "step": 25725
    },
    {
      "epoch": 0.2617390950520833,
      "grad_norm": 10.914161682128906,
      "learning_rate": 4.205706760390251e-06,
      "loss": 2.9514,
      "step": 25730
    },
    {
      "epoch": 0.2617899576822917,
      "grad_norm": 11.681282997131348,
      "learning_rate": 4.205414390235087e-06,
      "loss": 3.2611,
      "step": 25735
    },
    {
      "epoch": 0.2618408203125,
      "grad_norm": 13.976754188537598,
      "learning_rate": 4.205121976447298e-06,
      "loss": 3.4745,
      "step": 25740
    },
    {
      "epoch": 0.2618916829427083,
      "grad_norm": 11.412381172180176,
      "learning_rate": 4.204829519034368e-06,
      "loss": 2.935,
      "step": 25745
    },
    {
      "epoch": 0.2619425455729167,
      "grad_norm": 16.802989959716797,
      "learning_rate": 4.204537018003779e-06,
      "loss": 3.4396,
      "step": 25750
    },
    {
      "epoch": 0.261993408203125,
      "grad_norm": 13.803888320922852,
      "learning_rate": 4.204244473363015e-06,
      "loss": 3.4686,
      "step": 25755
    },
    {
      "epoch": 0.2620442708333333,
      "grad_norm": 13.405187606811523,
      "learning_rate": 4.203951885119559e-06,
      "loss": 3.2148,
      "step": 25760
    },
    {
      "epoch": 0.2620951334635417,
      "grad_norm": 11.889911651611328,
      "learning_rate": 4.2036592532808986e-06,
      "loss": 3.3572,
      "step": 25765
    },
    {
      "epoch": 0.26214599609375,
      "grad_norm": 16.20751190185547,
      "learning_rate": 4.20336657785452e-06,
      "loss": 3.1729,
      "step": 25770
    },
    {
      "epoch": 0.2621968587239583,
      "grad_norm": 8.080984115600586,
      "learning_rate": 4.203073858847912e-06,
      "loss": 3.1279,
      "step": 25775
    },
    {
      "epoch": 0.2622477213541667,
      "grad_norm": 13.503204345703125,
      "learning_rate": 4.202781096268561e-06,
      "loss": 3.3724,
      "step": 25780
    },
    {
      "epoch": 0.262298583984375,
      "grad_norm": 12.27587890625,
      "learning_rate": 4.20248829012396e-06,
      "loss": 3.5408,
      "step": 25785
    },
    {
      "epoch": 0.2623494466145833,
      "grad_norm": 8.207764625549316,
      "learning_rate": 4.202195440421601e-06,
      "loss": 3.4311,
      "step": 25790
    },
    {
      "epoch": 0.2624003092447917,
      "grad_norm": 7.626378536224365,
      "learning_rate": 4.201902547168974e-06,
      "loss": 3.2892,
      "step": 25795
    },
    {
      "epoch": 0.262451171875,
      "grad_norm": 7.448328495025635,
      "learning_rate": 4.201609610373573e-06,
      "loss": 3.3419,
      "step": 25800
    },
    {
      "epoch": 0.2625020345052083,
      "grad_norm": 9.728148460388184,
      "learning_rate": 4.201316630042895e-06,
      "loss": 3.623,
      "step": 25805
    },
    {
      "epoch": 0.2625528971354167,
      "grad_norm": 13.286189079284668,
      "learning_rate": 4.201023606184433e-06,
      "loss": 3.4971,
      "step": 25810
    },
    {
      "epoch": 0.262603759765625,
      "grad_norm": 9.246475219726562,
      "learning_rate": 4.200730538805687e-06,
      "loss": 3.9024,
      "step": 25815
    },
    {
      "epoch": 0.2626546223958333,
      "grad_norm": 8.17664909362793,
      "learning_rate": 4.200437427914151e-06,
      "loss": 3.3405,
      "step": 25820
    },
    {
      "epoch": 0.2627054850260417,
      "grad_norm": 9.446663856506348,
      "learning_rate": 4.200144273517328e-06,
      "loss": 3.332,
      "step": 25825
    },
    {
      "epoch": 0.26275634765625,
      "grad_norm": 10.07276725769043,
      "learning_rate": 4.1998510756227165e-06,
      "loss": 3.7568,
      "step": 25830
    },
    {
      "epoch": 0.2628072102864583,
      "grad_norm": 8.375320434570312,
      "learning_rate": 4.199557834237818e-06,
      "loss": 3.5829,
      "step": 25835
    },
    {
      "epoch": 0.2628580729166667,
      "grad_norm": 11.211957931518555,
      "learning_rate": 4.1992645493701346e-06,
      "loss": 3.3721,
      "step": 25840
    },
    {
      "epoch": 0.262908935546875,
      "grad_norm": 15.27626895904541,
      "learning_rate": 4.198971221027172e-06,
      "loss": 3.3733,
      "step": 25845
    },
    {
      "epoch": 0.2629597981770833,
      "grad_norm": 18.51883888244629,
      "learning_rate": 4.198677849216432e-06,
      "loss": 3.5998,
      "step": 25850
    },
    {
      "epoch": 0.2630106608072917,
      "grad_norm": 11.279624938964844,
      "learning_rate": 4.198384433945422e-06,
      "loss": 3.3106,
      "step": 25855
    },
    {
      "epoch": 0.2630615234375,
      "grad_norm": 8.512675285339355,
      "learning_rate": 4.198090975221649e-06,
      "loss": 2.9766,
      "step": 25860
    },
    {
      "epoch": 0.2631123860677083,
      "grad_norm": 11.04448413848877,
      "learning_rate": 4.1977974730526206e-06,
      "loss": 3.4403,
      "step": 25865
    },
    {
      "epoch": 0.2631632486979167,
      "grad_norm": 8.66542911529541,
      "learning_rate": 4.197503927445846e-06,
      "loss": 3.1101,
      "step": 25870
    },
    {
      "epoch": 0.263214111328125,
      "grad_norm": 14.436765670776367,
      "learning_rate": 4.197210338408837e-06,
      "loss": 3.4673,
      "step": 25875
    },
    {
      "epoch": 0.2632649739583333,
      "grad_norm": 12.47187614440918,
      "learning_rate": 4.196916705949103e-06,
      "loss": 3.1727,
      "step": 25880
    },
    {
      "epoch": 0.2633158365885417,
      "grad_norm": 14.223555564880371,
      "learning_rate": 4.196623030074158e-06,
      "loss": 3.6004,
      "step": 25885
    },
    {
      "epoch": 0.26336669921875,
      "grad_norm": 12.401254653930664,
      "learning_rate": 4.196329310791514e-06,
      "loss": 3.2072,
      "step": 25890
    },
    {
      "epoch": 0.2634175618489583,
      "grad_norm": 11.804680824279785,
      "learning_rate": 4.196035548108687e-06,
      "loss": 3.5086,
      "step": 25895
    },
    {
      "epoch": 0.2634684244791667,
      "grad_norm": 12.847404479980469,
      "learning_rate": 4.195741742033192e-06,
      "loss": 3.1709,
      "step": 25900
    },
    {
      "epoch": 0.263519287109375,
      "grad_norm": 12.488205909729004,
      "learning_rate": 4.195447892572548e-06,
      "loss": 3.2119,
      "step": 25905
    },
    {
      "epoch": 0.2635701497395833,
      "grad_norm": 11.74796199798584,
      "learning_rate": 4.19515399973427e-06,
      "loss": 3.4599,
      "step": 25910
    },
    {
      "epoch": 0.2636210123697917,
      "grad_norm": 11.54658317565918,
      "learning_rate": 4.1948600635258795e-06,
      "loss": 3.391,
      "step": 25915
    },
    {
      "epoch": 0.263671875,
      "grad_norm": 12.342092514038086,
      "learning_rate": 4.194566083954895e-06,
      "loss": 3.3041,
      "step": 25920
    },
    {
      "epoch": 0.2637227376302083,
      "grad_norm": 11.62911605834961,
      "learning_rate": 4.19427206102884e-06,
      "loss": 3.2533,
      "step": 25925
    },
    {
      "epoch": 0.2637736002604167,
      "grad_norm": 11.094088554382324,
      "learning_rate": 4.193977994755236e-06,
      "loss": 3.1159,
      "step": 25930
    },
    {
      "epoch": 0.263824462890625,
      "grad_norm": 13.771241188049316,
      "learning_rate": 4.193683885141606e-06,
      "loss": 3.7694,
      "step": 25935
    },
    {
      "epoch": 0.2638753255208333,
      "grad_norm": 14.977154731750488,
      "learning_rate": 4.193389732195476e-06,
      "loss": 3.5497,
      "step": 25940
    },
    {
      "epoch": 0.2639261881510417,
      "grad_norm": 11.603812217712402,
      "learning_rate": 4.19309553592437e-06,
      "loss": 3.3818,
      "step": 25945
    },
    {
      "epoch": 0.26397705078125,
      "grad_norm": 13.614372253417969,
      "learning_rate": 4.192801296335816e-06,
      "loss": 3.3199,
      "step": 25950
    },
    {
      "epoch": 0.2640279134114583,
      "grad_norm": 9.210861206054688,
      "learning_rate": 4.192507013437343e-06,
      "loss": 3.6322,
      "step": 25955
    },
    {
      "epoch": 0.2640787760416667,
      "grad_norm": 12.073121070861816,
      "learning_rate": 4.1922126872364774e-06,
      "loss": 3.5082,
      "step": 25960
    },
    {
      "epoch": 0.264129638671875,
      "grad_norm": 10.942256927490234,
      "learning_rate": 4.191918317740753e-06,
      "loss": 3.2529,
      "step": 25965
    },
    {
      "epoch": 0.2641805013020833,
      "grad_norm": 12.275918960571289,
      "learning_rate": 4.1916239049577e-06,
      "loss": 3.3314,
      "step": 25970
    },
    {
      "epoch": 0.2642313639322917,
      "grad_norm": 11.296648979187012,
      "learning_rate": 4.191329448894849e-06,
      "loss": 3.2901,
      "step": 25975
    },
    {
      "epoch": 0.2642822265625,
      "grad_norm": 13.480021476745605,
      "learning_rate": 4.191034949559735e-06,
      "loss": 3.3637,
      "step": 25980
    },
    {
      "epoch": 0.2643330891927083,
      "grad_norm": 12.317750930786133,
      "learning_rate": 4.190740406959893e-06,
      "loss": 3.5085,
      "step": 25985
    },
    {
      "epoch": 0.2643839518229167,
      "grad_norm": 14.043403625488281,
      "learning_rate": 4.190445821102858e-06,
      "loss": 3.1108,
      "step": 25990
    },
    {
      "epoch": 0.264434814453125,
      "grad_norm": 12.085020065307617,
      "learning_rate": 4.190151191996168e-06,
      "loss": 3.3186,
      "step": 25995
    },
    {
      "epoch": 0.2644856770833333,
      "grad_norm": 11.749665260314941,
      "learning_rate": 4.189856519647361e-06,
      "loss": 3.9153,
      "step": 26000
    },
    {
      "epoch": 0.2645365397135417,
      "grad_norm": 12.28537654876709,
      "learning_rate": 4.189561804063974e-06,
      "loss": 3.1333,
      "step": 26005
    },
    {
      "epoch": 0.26458740234375,
      "grad_norm": 10.340080261230469,
      "learning_rate": 4.18926704525355e-06,
      "loss": 3.9485,
      "step": 26010
    },
    {
      "epoch": 0.2646382649739583,
      "grad_norm": 9.838654518127441,
      "learning_rate": 4.188972243223628e-06,
      "loss": 3.3617,
      "step": 26015
    },
    {
      "epoch": 0.2646891276041667,
      "grad_norm": 8.775800704956055,
      "learning_rate": 4.188677397981752e-06,
      "loss": 3.5872,
      "step": 26020
    },
    {
      "epoch": 0.264739990234375,
      "grad_norm": 13.727091789245605,
      "learning_rate": 4.188382509535465e-06,
      "loss": 3.3762,
      "step": 26025
    },
    {
      "epoch": 0.2647908528645833,
      "grad_norm": 7.585343360900879,
      "learning_rate": 4.188087577892311e-06,
      "loss": 3.5484,
      "step": 26030
    },
    {
      "epoch": 0.2648417154947917,
      "grad_norm": 14.522687911987305,
      "learning_rate": 4.187792603059837e-06,
      "loss": 3.2677,
      "step": 26035
    },
    {
      "epoch": 0.264892578125,
      "grad_norm": 11.510234832763672,
      "learning_rate": 4.187497585045589e-06,
      "loss": 3.6247,
      "step": 26040
    },
    {
      "epoch": 0.2649434407552083,
      "grad_norm": 16.558944702148438,
      "learning_rate": 4.1872025238571155e-06,
      "loss": 3.324,
      "step": 26045
    },
    {
      "epoch": 0.2649943033854167,
      "grad_norm": 12.075737953186035,
      "learning_rate": 4.186907419501966e-06,
      "loss": 3.6676,
      "step": 26050
    },
    {
      "epoch": 0.265045166015625,
      "grad_norm": 10.342079162597656,
      "learning_rate": 4.1866122719876896e-06,
      "loss": 3.2716,
      "step": 26055
    },
    {
      "epoch": 0.2650960286458333,
      "grad_norm": 10.918512344360352,
      "learning_rate": 4.186317081321838e-06,
      "loss": 3.4466,
      "step": 26060
    },
    {
      "epoch": 0.2651468912760417,
      "grad_norm": 9.312711715698242,
      "learning_rate": 4.186021847511963e-06,
      "loss": 3.1628,
      "step": 26065
    },
    {
      "epoch": 0.26519775390625,
      "grad_norm": 14.83906078338623,
      "learning_rate": 4.185726570565619e-06,
      "loss": 3.2134,
      "step": 26070
    },
    {
      "epoch": 0.2652486165364583,
      "grad_norm": 16.265422821044922,
      "learning_rate": 4.18543125049036e-06,
      "loss": 3.5201,
      "step": 26075
    },
    {
      "epoch": 0.2652994791666667,
      "grad_norm": 10.572354316711426,
      "learning_rate": 4.185135887293743e-06,
      "loss": 3.1069,
      "step": 26080
    },
    {
      "epoch": 0.265350341796875,
      "grad_norm": 11.070435523986816,
      "learning_rate": 4.1848404809833235e-06,
      "loss": 3.2837,
      "step": 26085
    },
    {
      "epoch": 0.2654012044270833,
      "grad_norm": 11.942509651184082,
      "learning_rate": 4.184545031566659e-06,
      "loss": 3.6628,
      "step": 26090
    },
    {
      "epoch": 0.2654520670572917,
      "grad_norm": 15.380331039428711,
      "learning_rate": 4.18424953905131e-06,
      "loss": 3.5475,
      "step": 26095
    },
    {
      "epoch": 0.2655029296875,
      "grad_norm": 9.15284538269043,
      "learning_rate": 4.183954003444834e-06,
      "loss": 2.9877,
      "step": 26100
    },
    {
      "epoch": 0.2655537923177083,
      "grad_norm": 12.45242691040039,
      "learning_rate": 4.1836584247547955e-06,
      "loss": 3.514,
      "step": 26105
    },
    {
      "epoch": 0.2656046549479167,
      "grad_norm": 11.011215209960938,
      "learning_rate": 4.183362802988754e-06,
      "loss": 3.3444,
      "step": 26110
    },
    {
      "epoch": 0.265655517578125,
      "grad_norm": 9.054574012756348,
      "learning_rate": 4.183067138154275e-06,
      "loss": 3.3986,
      "step": 26115
    },
    {
      "epoch": 0.2657063802083333,
      "grad_norm": 9.417643547058105,
      "learning_rate": 4.182771430258923e-06,
      "loss": 3.2725,
      "step": 26120
    },
    {
      "epoch": 0.2657572428385417,
      "grad_norm": 11.767871856689453,
      "learning_rate": 4.182475679310262e-06,
      "loss": 3.3302,
      "step": 26125
    },
    {
      "epoch": 0.26580810546875,
      "grad_norm": 11.49780559539795,
      "learning_rate": 4.182179885315859e-06,
      "loss": 3.461,
      "step": 26130
    },
    {
      "epoch": 0.2658589680989583,
      "grad_norm": 20.007266998291016,
      "learning_rate": 4.181884048283283e-06,
      "loss": 3.3254,
      "step": 26135
    },
    {
      "epoch": 0.2659098307291667,
      "grad_norm": 15.877753257751465,
      "learning_rate": 4.181588168220102e-06,
      "loss": 3.3056,
      "step": 26140
    },
    {
      "epoch": 0.265960693359375,
      "grad_norm": 10.472766876220703,
      "learning_rate": 4.181292245133887e-06,
      "loss": 3.4033,
      "step": 26145
    },
    {
      "epoch": 0.2660115559895833,
      "grad_norm": 10.49460506439209,
      "learning_rate": 4.180996279032208e-06,
      "loss": 3.3645,
      "step": 26150
    },
    {
      "epoch": 0.2660624186197917,
      "grad_norm": 13.416446685791016,
      "learning_rate": 4.180700269922637e-06,
      "loss": 3.2229,
      "step": 26155
    },
    {
      "epoch": 0.26611328125,
      "grad_norm": 15.531248092651367,
      "learning_rate": 4.180404217812749e-06,
      "loss": 3.1068,
      "step": 26160
    },
    {
      "epoch": 0.2661641438802083,
      "grad_norm": 12.707284927368164,
      "learning_rate": 4.180108122710116e-06,
      "loss": 3.1724,
      "step": 26165
    },
    {
      "epoch": 0.2662150065104167,
      "grad_norm": 8.874678611755371,
      "learning_rate": 4.1798119846223164e-06,
      "loss": 3.7423,
      "step": 26170
    },
    {
      "epoch": 0.266265869140625,
      "grad_norm": 9.22908878326416,
      "learning_rate": 4.1795158035569235e-06,
      "loss": 3.5431,
      "step": 26175
    },
    {
      "epoch": 0.2663167317708333,
      "grad_norm": 12.564714431762695,
      "learning_rate": 4.179219579521519e-06,
      "loss": 3.1369,
      "step": 26180
    },
    {
      "epoch": 0.2663675944010417,
      "grad_norm": 11.059355735778809,
      "learning_rate": 4.178923312523679e-06,
      "loss": 3.4337,
      "step": 26185
    },
    {
      "epoch": 0.26641845703125,
      "grad_norm": 11.744783401489258,
      "learning_rate": 4.178627002570983e-06,
      "loss": 3.4429,
      "step": 26190
    },
    {
      "epoch": 0.2664693196614583,
      "grad_norm": 13.256790161132812,
      "learning_rate": 4.178330649671014e-06,
      "loss": 3.2407,
      "step": 26195
    },
    {
      "epoch": 0.2665201822916667,
      "grad_norm": 12.98029899597168,
      "learning_rate": 4.178034253831353e-06,
      "loss": 3.5527,
      "step": 26200
    },
    {
      "epoch": 0.266571044921875,
      "grad_norm": 13.788947105407715,
      "learning_rate": 4.177737815059582e-06,
      "loss": 3.3113,
      "step": 26205
    },
    {
      "epoch": 0.2666219075520833,
      "grad_norm": 15.371501922607422,
      "learning_rate": 4.177441333363289e-06,
      "loss": 3.4981,
      "step": 26210
    },
    {
      "epoch": 0.2666727701822917,
      "grad_norm": 12.833020210266113,
      "learning_rate": 4.177144808750055e-06,
      "loss": 3.3368,
      "step": 26215
    },
    {
      "epoch": 0.2667236328125,
      "grad_norm": 16.910667419433594,
      "learning_rate": 4.17684824122747e-06,
      "loss": 3.3971,
      "step": 26220
    },
    {
      "epoch": 0.2667744954427083,
      "grad_norm": 14.4556303024292,
      "learning_rate": 4.176551630803119e-06,
      "loss": 3.1049,
      "step": 26225
    },
    {
      "epoch": 0.2668253580729167,
      "grad_norm": 8.58405876159668,
      "learning_rate": 4.176254977484593e-06,
      "loss": 3.2514,
      "step": 26230
    },
    {
      "epoch": 0.266876220703125,
      "grad_norm": 9.659404754638672,
      "learning_rate": 4.175958281279481e-06,
      "loss": 3.3754,
      "step": 26235
    },
    {
      "epoch": 0.2669270833333333,
      "grad_norm": 10.980110168457031,
      "learning_rate": 4.1756615421953725e-06,
      "loss": 3.0858,
      "step": 26240
    },
    {
      "epoch": 0.2669779459635417,
      "grad_norm": 9.385514259338379,
      "learning_rate": 4.175364760239862e-06,
      "loss": 3.2213,
      "step": 26245
    },
    {
      "epoch": 0.26702880859375,
      "grad_norm": 8.674199104309082,
      "learning_rate": 4.1750679354205395e-06,
      "loss": 3.8795,
      "step": 26250
    },
    {
      "epoch": 0.2670796712239583,
      "grad_norm": 15.488489151000977,
      "learning_rate": 4.1747710677450015e-06,
      "loss": 3.4649,
      "step": 26255
    },
    {
      "epoch": 0.2671305338541667,
      "grad_norm": 7.275283336639404,
      "learning_rate": 4.174474157220843e-06,
      "loss": 2.7254,
      "step": 26260
    },
    {
      "epoch": 0.267181396484375,
      "grad_norm": 12.355480194091797,
      "learning_rate": 4.17417720385566e-06,
      "loss": 3.3009,
      "step": 26265
    },
    {
      "epoch": 0.2672322591145833,
      "grad_norm": 14.901166915893555,
      "learning_rate": 4.17388020765705e-06,
      "loss": 3.684,
      "step": 26270
    },
    {
      "epoch": 0.2672831217447917,
      "grad_norm": 15.172636985778809,
      "learning_rate": 4.1735831686326126e-06,
      "loss": 3.2967,
      "step": 26275
    },
    {
      "epoch": 0.267333984375,
      "grad_norm": 13.416927337646484,
      "learning_rate": 4.173286086789945e-06,
      "loss": 3.3368,
      "step": 26280
    },
    {
      "epoch": 0.2673848470052083,
      "grad_norm": 12.028944969177246,
      "learning_rate": 4.17298896213665e-06,
      "loss": 3.3686,
      "step": 26285
    },
    {
      "epoch": 0.2674357096354167,
      "grad_norm": 9.418871879577637,
      "learning_rate": 4.172691794680329e-06,
      "loss": 3.6334,
      "step": 26290
    },
    {
      "epoch": 0.267486572265625,
      "grad_norm": 11.656323432922363,
      "learning_rate": 4.172394584428585e-06,
      "loss": 3.1989,
      "step": 26295
    },
    {
      "epoch": 0.2675374348958333,
      "grad_norm": 8.896805763244629,
      "learning_rate": 4.172097331389022e-06,
      "loss": 3.1311,
      "step": 26300
    },
    {
      "epoch": 0.2675882975260417,
      "grad_norm": 11.2920560836792,
      "learning_rate": 4.171800035569245e-06,
      "loss": 3.4342,
      "step": 26305
    },
    {
      "epoch": 0.26763916015625,
      "grad_norm": 13.756390571594238,
      "learning_rate": 4.1715026969768605e-06,
      "loss": 3.7989,
      "step": 26310
    },
    {
      "epoch": 0.2676900227864583,
      "grad_norm": 10.483635902404785,
      "learning_rate": 4.171205315619476e-06,
      "loss": 3.3567,
      "step": 26315
    },
    {
      "epoch": 0.2677408854166667,
      "grad_norm": 7.852400779724121,
      "learning_rate": 4.170907891504699e-06,
      "loss": 3.1506,
      "step": 26320
    },
    {
      "epoch": 0.267791748046875,
      "grad_norm": 14.112519264221191,
      "learning_rate": 4.170610424640139e-06,
      "loss": 3.3308,
      "step": 26325
    },
    {
      "epoch": 0.2678426106770833,
      "grad_norm": 13.232032775878906,
      "learning_rate": 4.170312915033408e-06,
      "loss": 3.0412,
      "step": 26330
    },
    {
      "epoch": 0.2678934733072917,
      "grad_norm": 8.810139656066895,
      "learning_rate": 4.170015362692117e-06,
      "loss": 3.426,
      "step": 26335
    },
    {
      "epoch": 0.2679443359375,
      "grad_norm": 11.008728981018066,
      "learning_rate": 4.1697177676238786e-06,
      "loss": 3.0953,
      "step": 26340
    },
    {
      "epoch": 0.2679951985677083,
      "grad_norm": 15.006650924682617,
      "learning_rate": 4.169420129836307e-06,
      "loss": 3.2985,
      "step": 26345
    },
    {
      "epoch": 0.2680460611979167,
      "grad_norm": 19.160358428955078,
      "learning_rate": 4.1691224493370175e-06,
      "loss": 3.5361,
      "step": 26350
    },
    {
      "epoch": 0.268096923828125,
      "grad_norm": 12.861884117126465,
      "learning_rate": 4.168824726133625e-06,
      "loss": 3.2308,
      "step": 26355
    },
    {
      "epoch": 0.2681477864583333,
      "grad_norm": 10.119865417480469,
      "learning_rate": 4.1685269602337484e-06,
      "loss": 3.256,
      "step": 26360
    },
    {
      "epoch": 0.2681986490885417,
      "grad_norm": 12.688722610473633,
      "learning_rate": 4.168229151645004e-06,
      "loss": 3.7487,
      "step": 26365
    },
    {
      "epoch": 0.26824951171875,
      "grad_norm": 11.832856178283691,
      "learning_rate": 4.167931300375014e-06,
      "loss": 3.3695,
      "step": 26370
    },
    {
      "epoch": 0.2683003743489583,
      "grad_norm": 16.034059524536133,
      "learning_rate": 4.167633406431395e-06,
      "loss": 3.3954,
      "step": 26375
    },
    {
      "epoch": 0.2683512369791667,
      "grad_norm": 12.99596881866455,
      "learning_rate": 4.167335469821771e-06,
      "loss": 3.0622,
      "step": 26380
    },
    {
      "epoch": 0.268402099609375,
      "grad_norm": 8.741852760314941,
      "learning_rate": 4.167037490553764e-06,
      "loss": 3.5418,
      "step": 26385
    },
    {
      "epoch": 0.2684529622395833,
      "grad_norm": 15.765666007995605,
      "learning_rate": 4.1667394686349984e-06,
      "loss": 4.2059,
      "step": 26390
    },
    {
      "epoch": 0.2685038248697917,
      "grad_norm": 11.599302291870117,
      "learning_rate": 4.166441404073099e-06,
      "loss": 3.1983,
      "step": 26395
    },
    {
      "epoch": 0.2685546875,
      "grad_norm": 14.83521842956543,
      "learning_rate": 4.166143296875691e-06,
      "loss": 2.924,
      "step": 26400
    },
    {
      "epoch": 0.2686055501302083,
      "grad_norm": 13.936452865600586,
      "learning_rate": 4.165845147050401e-06,
      "loss": 3.3603,
      "step": 26405
    },
    {
      "epoch": 0.2686564127604167,
      "grad_norm": 14.793567657470703,
      "learning_rate": 4.165546954604859e-06,
      "loss": 3.4469,
      "step": 26410
    },
    {
      "epoch": 0.268707275390625,
      "grad_norm": 10.850431442260742,
      "learning_rate": 4.165248719546691e-06,
      "loss": 3.3443,
      "step": 26415
    },
    {
      "epoch": 0.2687581380208333,
      "grad_norm": 11.647077560424805,
      "learning_rate": 4.164950441883531e-06,
      "loss": 3.1829,
      "step": 26420
    },
    {
      "epoch": 0.2688090006510417,
      "grad_norm": 14.193302154541016,
      "learning_rate": 4.1646521216230075e-06,
      "loss": 3.4609,
      "step": 26425
    },
    {
      "epoch": 0.26885986328125,
      "grad_norm": 8.489500045776367,
      "learning_rate": 4.164353758772755e-06,
      "loss": 3.1324,
      "step": 26430
    },
    {
      "epoch": 0.2689107259114583,
      "grad_norm": 10.395947456359863,
      "learning_rate": 4.1640553533404046e-06,
      "loss": 3.8962,
      "step": 26435
    },
    {
      "epoch": 0.2689615885416667,
      "grad_norm": 9.547869682312012,
      "learning_rate": 4.163756905333594e-06,
      "loss": 2.9235,
      "step": 26440
    },
    {
      "epoch": 0.269012451171875,
      "grad_norm": 8.927133560180664,
      "learning_rate": 4.1634584147599564e-06,
      "loss": 3.3013,
      "step": 26445
    },
    {
      "epoch": 0.2690633138020833,
      "grad_norm": 9.086165428161621,
      "learning_rate": 4.16315988162713e-06,
      "loss": 3.7275,
      "step": 26450
    },
    {
      "epoch": 0.2691141764322917,
      "grad_norm": 14.933182716369629,
      "learning_rate": 4.162861305942753e-06,
      "loss": 3.4349,
      "step": 26455
    },
    {
      "epoch": 0.2691650390625,
      "grad_norm": 11.188149452209473,
      "learning_rate": 4.162562687714462e-06,
      "loss": 3.49,
      "step": 26460
    },
    {
      "epoch": 0.2692159016927083,
      "grad_norm": 14.859213829040527,
      "learning_rate": 4.1622640269499e-06,
      "loss": 3.5914,
      "step": 26465
    },
    {
      "epoch": 0.2692667643229167,
      "grad_norm": 14.470888137817383,
      "learning_rate": 4.161965323656706e-06,
      "loss": 4.0164,
      "step": 26470
    },
    {
      "epoch": 0.269317626953125,
      "grad_norm": 8.29883861541748,
      "learning_rate": 4.161666577842524e-06,
      "loss": 3.7569,
      "step": 26475
    },
    {
      "epoch": 0.2693684895833333,
      "grad_norm": 16.3638858795166,
      "learning_rate": 4.161367789514995e-06,
      "loss": 3.1554,
      "step": 26480
    },
    {
      "epoch": 0.2694193522135417,
      "grad_norm": 9.231216430664062,
      "learning_rate": 4.161068958681766e-06,
      "loss": 3.3017,
      "step": 26485
    },
    {
      "epoch": 0.26947021484375,
      "grad_norm": 14.40170669555664,
      "learning_rate": 4.160770085350481e-06,
      "loss": 3.4301,
      "step": 26490
    },
    {
      "epoch": 0.2695210774739583,
      "grad_norm": 9.079069137573242,
      "learning_rate": 4.1604711695287876e-06,
      "loss": 3.3494,
      "step": 26495
    },
    {
      "epoch": 0.2695719401041667,
      "grad_norm": 10.551512718200684,
      "learning_rate": 4.160172211224332e-06,
      "loss": 3.287,
      "step": 26500
    },
    {
      "epoch": 0.269622802734375,
      "grad_norm": 12.600727081298828,
      "learning_rate": 4.159873210444765e-06,
      "loss": 3.4167,
      "step": 26505
    },
    {
      "epoch": 0.2696736653645833,
      "grad_norm": 9.81279468536377,
      "learning_rate": 4.159574167197734e-06,
      "loss": 3.3034,
      "step": 26510
    },
    {
      "epoch": 0.2697245279947917,
      "grad_norm": 15.392224311828613,
      "learning_rate": 4.159275081490892e-06,
      "loss": 3.1249,
      "step": 26515
    },
    {
      "epoch": 0.269775390625,
      "grad_norm": 10.038593292236328,
      "learning_rate": 4.158975953331891e-06,
      "loss": 3.2412,
      "step": 26520
    },
    {
      "epoch": 0.2698262532552083,
      "grad_norm": 13.212175369262695,
      "learning_rate": 4.158676782728383e-06,
      "loss": 3.4358,
      "step": 26525
    },
    {
      "epoch": 0.2698771158854167,
      "grad_norm": 10.415067672729492,
      "learning_rate": 4.158377569688022e-06,
      "loss": 3.503,
      "step": 26530
    },
    {
      "epoch": 0.269927978515625,
      "grad_norm": 12.571605682373047,
      "learning_rate": 4.158078314218464e-06,
      "loss": 3.5547,
      "step": 26535
    },
    {
      "epoch": 0.2699788411458333,
      "grad_norm": 9.009819984436035,
      "learning_rate": 4.157779016327366e-06,
      "loss": 3.4935,
      "step": 26540
    },
    {
      "epoch": 0.2700297037760417,
      "grad_norm": 14.811439514160156,
      "learning_rate": 4.1574796760223845e-06,
      "loss": 3.4557,
      "step": 26545
    },
    {
      "epoch": 0.27008056640625,
      "grad_norm": 15.486289978027344,
      "learning_rate": 4.157180293311179e-06,
      "loss": 3.7938,
      "step": 26550
    },
    {
      "epoch": 0.2701314290364583,
      "grad_norm": 15.374136924743652,
      "learning_rate": 4.156880868201407e-06,
      "loss": 3.4213,
      "step": 26555
    },
    {
      "epoch": 0.2701822916666667,
      "grad_norm": 9.167612075805664,
      "learning_rate": 4.156581400700732e-06,
      "loss": 3.3672,
      "step": 26560
    },
    {
      "epoch": 0.270233154296875,
      "grad_norm": 9.546422958374023,
      "learning_rate": 4.156281890816814e-06,
      "loss": 3.5656,
      "step": 26565
    },
    {
      "epoch": 0.2702840169270833,
      "grad_norm": 16.142099380493164,
      "learning_rate": 4.155982338557317e-06,
      "loss": 3.25,
      "step": 26570
    },
    {
      "epoch": 0.2703348795572917,
      "grad_norm": 12.354496955871582,
      "learning_rate": 4.155682743929904e-06,
      "loss": 3.4914,
      "step": 26575
    },
    {
      "epoch": 0.2703857421875,
      "grad_norm": 10.914223670959473,
      "learning_rate": 4.15538310694224e-06,
      "loss": 3.4107,
      "step": 26580
    },
    {
      "epoch": 0.2704366048177083,
      "grad_norm": 12.899299621582031,
      "learning_rate": 4.155083427601992e-06,
      "loss": 3.2644,
      "step": 26585
    },
    {
      "epoch": 0.2704874674479167,
      "grad_norm": 14.942575454711914,
      "learning_rate": 4.154783705916827e-06,
      "loss": 3.1591,
      "step": 26590
    },
    {
      "epoch": 0.270538330078125,
      "grad_norm": 10.189476013183594,
      "learning_rate": 4.1544839418944124e-06,
      "loss": 3.1835,
      "step": 26595
    },
    {
      "epoch": 0.2705891927083333,
      "grad_norm": 13.850764274597168,
      "learning_rate": 4.154184135542419e-06,
      "loss": 3.2307,
      "step": 26600
    },
    {
      "epoch": 0.2706400553385417,
      "grad_norm": 10.06661605834961,
      "learning_rate": 4.1538842868685155e-06,
      "loss": 3.3825,
      "step": 26605
    },
    {
      "epoch": 0.27069091796875,
      "grad_norm": 7.886444568634033,
      "learning_rate": 4.153584395880376e-06,
      "loss": 2.9482,
      "step": 26610
    },
    {
      "epoch": 0.2707417805989583,
      "grad_norm": 13.357390403747559,
      "learning_rate": 4.153284462585672e-06,
      "loss": 3.6921,
      "step": 26615
    },
    {
      "epoch": 0.2707926432291667,
      "grad_norm": 12.01125431060791,
      "learning_rate": 4.152984486992075e-06,
      "loss": 3.6984,
      "step": 26620
    },
    {
      "epoch": 0.270843505859375,
      "grad_norm": 11.155698776245117,
      "learning_rate": 4.152684469107263e-06,
      "loss": 3.1299,
      "step": 26625
    },
    {
      "epoch": 0.2708943684895833,
      "grad_norm": 14.0589017868042,
      "learning_rate": 4.15238440893891e-06,
      "loss": 3.6016,
      "step": 26630
    },
    {
      "epoch": 0.2709452311197917,
      "grad_norm": 15.438796043395996,
      "learning_rate": 4.1520843064946936e-06,
      "loss": 3.5558,
      "step": 26635
    },
    {
      "epoch": 0.27099609375,
      "grad_norm": 11.46037483215332,
      "learning_rate": 4.151784161782292e-06,
      "loss": 3.6484,
      "step": 26640
    },
    {
      "epoch": 0.2710469563802083,
      "grad_norm": 15.121505737304688,
      "learning_rate": 4.1514839748093845e-06,
      "loss": 3.2945,
      "step": 26645
    },
    {
      "epoch": 0.2710978190104167,
      "grad_norm": 13.236750602722168,
      "learning_rate": 4.15118374558365e-06,
      "loss": 3.2456,
      "step": 26650
    },
    {
      "epoch": 0.271148681640625,
      "grad_norm": 16.323974609375,
      "learning_rate": 4.150883474112771e-06,
      "loss": 3.434,
      "step": 26655
    },
    {
      "epoch": 0.2711995442708333,
      "grad_norm": 14.29743766784668,
      "learning_rate": 4.150583160404431e-06,
      "loss": 3.3447,
      "step": 26660
    },
    {
      "epoch": 0.2712504069010417,
      "grad_norm": 14.510125160217285,
      "learning_rate": 4.150282804466311e-06,
      "loss": 3.2886,
      "step": 26665
    },
    {
      "epoch": 0.27130126953125,
      "grad_norm": 9.165349006652832,
      "learning_rate": 4.1499824063060965e-06,
      "loss": 3.3194,
      "step": 26670
    },
    {
      "epoch": 0.2713521321614583,
      "grad_norm": 9.51498031616211,
      "learning_rate": 4.149681965931473e-06,
      "loss": 3.72,
      "step": 26675
    },
    {
      "epoch": 0.2714029947916667,
      "grad_norm": 16.4527645111084,
      "learning_rate": 4.149381483350128e-06,
      "loss": 3.4032,
      "step": 26680
    },
    {
      "epoch": 0.271453857421875,
      "grad_norm": 13.104421615600586,
      "learning_rate": 4.149080958569749e-06,
      "loss": 3.4986,
      "step": 26685
    },
    {
      "epoch": 0.2715047200520833,
      "grad_norm": 12.464369773864746,
      "learning_rate": 4.148780391598023e-06,
      "loss": 3.7014,
      "step": 26690
    },
    {
      "epoch": 0.2715555826822917,
      "grad_norm": 12.509614944458008,
      "learning_rate": 4.148479782442642e-06,
      "loss": 3.1501,
      "step": 26695
    },
    {
      "epoch": 0.2716064453125,
      "grad_norm": 10.114399909973145,
      "learning_rate": 4.148179131111298e-06,
      "loss": 3.1244,
      "step": 26700
    },
    {
      "epoch": 0.2716573079427083,
      "grad_norm": 9.907551765441895,
      "learning_rate": 4.147878437611681e-06,
      "loss": 3.5617,
      "step": 26705
    },
    {
      "epoch": 0.2717081705729167,
      "grad_norm": 9.493293762207031,
      "learning_rate": 4.147577701951483e-06,
      "loss": 3.0425,
      "step": 26710
    },
    {
      "epoch": 0.271759033203125,
      "grad_norm": 12.182016372680664,
      "learning_rate": 4.147276924138402e-06,
      "loss": 3.1682,
      "step": 26715
    },
    {
      "epoch": 0.2718098958333333,
      "grad_norm": 9.14958667755127,
      "learning_rate": 4.14697610418013e-06,
      "loss": 3.8109,
      "step": 26720
    },
    {
      "epoch": 0.2718607584635417,
      "grad_norm": 12.60875415802002,
      "learning_rate": 4.1466752420843656e-06,
      "loss": 3.7398,
      "step": 26725
    },
    {
      "epoch": 0.27191162109375,
      "grad_norm": 13.810464859008789,
      "learning_rate": 4.146374337858805e-06,
      "loss": 3.2147,
      "step": 26730
    },
    {
      "epoch": 0.2719624837239583,
      "grad_norm": 10.453813552856445,
      "learning_rate": 4.146073391511148e-06,
      "loss": 3.7852,
      "step": 26735
    },
    {
      "epoch": 0.2720133463541667,
      "grad_norm": 12.554015159606934,
      "learning_rate": 4.145772403049093e-06,
      "loss": 3.589,
      "step": 26740
    },
    {
      "epoch": 0.272064208984375,
      "grad_norm": 10.385574340820312,
      "learning_rate": 4.145471372480341e-06,
      "loss": 3.2686,
      "step": 26745
    },
    {
      "epoch": 0.2721150716145833,
      "grad_norm": 7.913267612457275,
      "learning_rate": 4.145170299812594e-06,
      "loss": 3.4415,
      "step": 26750
    },
    {
      "epoch": 0.2721659342447917,
      "grad_norm": 9.897327423095703,
      "learning_rate": 4.144869185053555e-06,
      "loss": 3.3393,
      "step": 26755
    },
    {
      "epoch": 0.272216796875,
      "grad_norm": 10.874954223632812,
      "learning_rate": 4.144568028210928e-06,
      "loss": 3.4982,
      "step": 26760
    },
    {
      "epoch": 0.2722676595052083,
      "grad_norm": 15.309311866760254,
      "learning_rate": 4.144266829292417e-06,
      "loss": 3.4094,
      "step": 26765
    },
    {
      "epoch": 0.2723185221354167,
      "grad_norm": 9.066498756408691,
      "learning_rate": 4.1439655883057295e-06,
      "loss": 3.1665,
      "step": 26770
    },
    {
      "epoch": 0.272369384765625,
      "grad_norm": 9.98532772064209,
      "learning_rate": 4.1436643052585715e-06,
      "loss": 3.3225,
      "step": 26775
    },
    {
      "epoch": 0.2724202473958333,
      "grad_norm": 12.59506893157959,
      "learning_rate": 4.143362980158653e-06,
      "loss": 3.7703,
      "step": 26780
    },
    {
      "epoch": 0.2724711100260417,
      "grad_norm": 9.37546157836914,
      "learning_rate": 4.143061613013681e-06,
      "loss": 3.2978,
      "step": 26785
    },
    {
      "epoch": 0.27252197265625,
      "grad_norm": 7.499221324920654,
      "learning_rate": 4.142760203831367e-06,
      "loss": 3.4306,
      "step": 26790
    },
    {
      "epoch": 0.2725728352864583,
      "grad_norm": 10.164257049560547,
      "learning_rate": 4.142458752619423e-06,
      "loss": 3.6316,
      "step": 26795
    },
    {
      "epoch": 0.2726236979166667,
      "grad_norm": 10.297619819641113,
      "learning_rate": 4.142157259385562e-06,
      "loss": 3.3527,
      "step": 26800
    },
    {
      "epoch": 0.272674560546875,
      "grad_norm": 7.55012321472168,
      "learning_rate": 4.1418557241374955e-06,
      "loss": 3.4203,
      "step": 26805
    },
    {
      "epoch": 0.2727254231770833,
      "grad_norm": 18.338563919067383,
      "learning_rate": 4.14155414688294e-06,
      "loss": 2.9126,
      "step": 26810
    },
    {
      "epoch": 0.2727762858072917,
      "grad_norm": 13.562408447265625,
      "learning_rate": 4.141252527629611e-06,
      "loss": 3.3561,
      "step": 26815
    },
    {
      "epoch": 0.2728271484375,
      "grad_norm": 10.19615364074707,
      "learning_rate": 4.140950866385225e-06,
      "loss": 3.0776,
      "step": 26820
    },
    {
      "epoch": 0.2728780110677083,
      "grad_norm": 11.81718635559082,
      "learning_rate": 4.140649163157499e-06,
      "loss": 3.2956,
      "step": 26825
    },
    {
      "epoch": 0.2729288736979167,
      "grad_norm": 12.27664566040039,
      "learning_rate": 4.140347417954154e-06,
      "loss": 3.3226,
      "step": 26830
    },
    {
      "epoch": 0.272979736328125,
      "grad_norm": 15.785837173461914,
      "learning_rate": 4.14004563078291e-06,
      "loss": 3.2568,
      "step": 26835
    },
    {
      "epoch": 0.2730305989583333,
      "grad_norm": 13.030390739440918,
      "learning_rate": 4.1397438016514856e-06,
      "loss": 3.5478,
      "step": 26840
    },
    {
      "epoch": 0.2730814615885417,
      "grad_norm": 10.237311363220215,
      "learning_rate": 4.139441930567604e-06,
      "loss": 3.4174,
      "step": 26845
    },
    {
      "epoch": 0.27313232421875,
      "grad_norm": 14.9881591796875,
      "learning_rate": 4.139140017538992e-06,
      "loss": 4.0547,
      "step": 26850
    },
    {
      "epoch": 0.2731831868489583,
      "grad_norm": 15.239192008972168,
      "learning_rate": 4.13883806257337e-06,
      "loss": 3.1832,
      "step": 26855
    },
    {
      "epoch": 0.2732340494791667,
      "grad_norm": 14.115588188171387,
      "learning_rate": 4.138536065678463e-06,
      "loss": 3.3923,
      "step": 26860
    },
    {
      "epoch": 0.273284912109375,
      "grad_norm": 9.392797470092773,
      "learning_rate": 4.138234026862002e-06,
      "loss": 3.9802,
      "step": 26865
    },
    {
      "epoch": 0.2733357747395833,
      "grad_norm": 9.962571144104004,
      "learning_rate": 4.137931946131709e-06,
      "loss": 3.1848,
      "step": 26870
    },
    {
      "epoch": 0.2733866373697917,
      "grad_norm": 11.43811321258545,
      "learning_rate": 4.137629823495317e-06,
      "loss": 3.4973,
      "step": 26875
    },
    {
      "epoch": 0.2734375,
      "grad_norm": 13.599096298217773,
      "learning_rate": 4.1373276589605535e-06,
      "loss": 3.869,
      "step": 26880
    },
    {
      "epoch": 0.2734883626302083,
      "grad_norm": 14.69780158996582,
      "learning_rate": 4.13702545253515e-06,
      "loss": 3.1746,
      "step": 26885
    },
    {
      "epoch": 0.2735392252604167,
      "grad_norm": 15.275219917297363,
      "learning_rate": 4.1367232042268376e-06,
      "loss": 3.4442,
      "step": 26890
    },
    {
      "epoch": 0.273590087890625,
      "grad_norm": 11.607757568359375,
      "learning_rate": 4.136420914043352e-06,
      "loss": 3.6636,
      "step": 26895
    },
    {
      "epoch": 0.2736409505208333,
      "grad_norm": 10.504234313964844,
      "learning_rate": 4.136118581992423e-06,
      "loss": 3.2825,
      "step": 26900
    },
    {
      "epoch": 0.2736918131510417,
      "grad_norm": 9.813732147216797,
      "learning_rate": 4.13581620808179e-06,
      "loss": 3.2447,
      "step": 26905
    },
    {
      "epoch": 0.27374267578125,
      "grad_norm": 10.727784156799316,
      "learning_rate": 4.135513792319185e-06,
      "loss": 3.4474,
      "step": 26910
    },
    {
      "epoch": 0.2737935384114583,
      "grad_norm": 9.403905868530273,
      "learning_rate": 4.135211334712349e-06,
      "loss": 3.4208,
      "step": 26915
    },
    {
      "epoch": 0.2738444010416667,
      "grad_norm": 9.684918403625488,
      "learning_rate": 4.134908835269018e-06,
      "loss": 3.5309,
      "step": 26920
    },
    {
      "epoch": 0.273895263671875,
      "grad_norm": 10.809853553771973,
      "learning_rate": 4.1346062939969316e-06,
      "loss": 4.2717,
      "step": 26925
    },
    {
      "epoch": 0.2739461263020833,
      "grad_norm": 13.863726615905762,
      "learning_rate": 4.13430371090383e-06,
      "loss": 3.4833,
      "step": 26930
    },
    {
      "epoch": 0.2739969889322917,
      "grad_norm": 15.26383113861084,
      "learning_rate": 4.134001085997457e-06,
      "loss": 3.5963,
      "step": 26935
    },
    {
      "epoch": 0.2740478515625,
      "grad_norm": 15.077373504638672,
      "learning_rate": 4.133698419285552e-06,
      "loss": 3.7273,
      "step": 26940
    },
    {
      "epoch": 0.2740987141927083,
      "grad_norm": 10.0264892578125,
      "learning_rate": 4.13339571077586e-06,
      "loss": 3.2443,
      "step": 26945
    },
    {
      "epoch": 0.2741495768229167,
      "grad_norm": 9.452435493469238,
      "learning_rate": 4.1330929604761275e-06,
      "loss": 3.4279,
      "step": 26950
    },
    {
      "epoch": 0.274200439453125,
      "grad_norm": 15.747602462768555,
      "learning_rate": 4.132790168394098e-06,
      "loss": 3.5983,
      "step": 26955
    },
    {
      "epoch": 0.2742513020833333,
      "grad_norm": 12.317953109741211,
      "learning_rate": 4.1324873345375185e-06,
      "loss": 3.8896,
      "step": 26960
    },
    {
      "epoch": 0.2743021647135417,
      "grad_norm": 11.992025375366211,
      "learning_rate": 4.132184458914138e-06,
      "loss": 3.5027,
      "step": 26965
    },
    {
      "epoch": 0.27435302734375,
      "grad_norm": 14.02786922454834,
      "learning_rate": 4.131881541531705e-06,
      "loss": 4.0145,
      "step": 26970
    },
    {
      "epoch": 0.2744038899739583,
      "grad_norm": 18.570852279663086,
      "learning_rate": 4.131578582397969e-06,
      "loss": 3.5407,
      "step": 26975
    },
    {
      "epoch": 0.2744547526041667,
      "grad_norm": 9.13254165649414,
      "learning_rate": 4.131275581520683e-06,
      "loss": 3.531,
      "step": 26980
    },
    {
      "epoch": 0.274505615234375,
      "grad_norm": 9.889278411865234,
      "learning_rate": 4.130972538907597e-06,
      "loss": 3.6133,
      "step": 26985
    },
    {
      "epoch": 0.2745564778645833,
      "grad_norm": 14.381463050842285,
      "learning_rate": 4.130669454566465e-06,
      "loss": 3.3638,
      "step": 26990
    },
    {
      "epoch": 0.2746073404947917,
      "grad_norm": 14.76690673828125,
      "learning_rate": 4.130366328505041e-06,
      "loss": 3.4389,
      "step": 26995
    },
    {
      "epoch": 0.274658203125,
      "grad_norm": 14.930278778076172,
      "learning_rate": 4.1300631607310824e-06,
      "loss": 3.3233,
      "step": 27000
    },
    {
      "epoch": 0.2747090657552083,
      "grad_norm": 9.765680313110352,
      "learning_rate": 4.129759951252344e-06,
      "loss": 3.1376,
      "step": 27005
    },
    {
      "epoch": 0.2747599283854167,
      "grad_norm": 10.81482219696045,
      "learning_rate": 4.129456700076583e-06,
      "loss": 3.2255,
      "step": 27010
    },
    {
      "epoch": 0.274810791015625,
      "grad_norm": 10.286974906921387,
      "learning_rate": 4.129153407211558e-06,
      "loss": 3.4194,
      "step": 27015
    },
    {
      "epoch": 0.2748616536458333,
      "grad_norm": 13.686568260192871,
      "learning_rate": 4.128850072665029e-06,
      "loss": 3.1024,
      "step": 27020
    },
    {
      "epoch": 0.2749125162760417,
      "grad_norm": 15.8404541015625,
      "learning_rate": 4.128546696444759e-06,
      "loss": 3.2685,
      "step": 27025
    },
    {
      "epoch": 0.27496337890625,
      "grad_norm": 11.484049797058105,
      "learning_rate": 4.128243278558506e-06,
      "loss": 3.8495,
      "step": 27030
    },
    {
      "epoch": 0.2750142415364583,
      "grad_norm": 9.2147798538208,
      "learning_rate": 4.127939819014035e-06,
      "loss": 3.4059,
      "step": 27035
    },
    {
      "epoch": 0.2750651041666667,
      "grad_norm": 11.897221565246582,
      "learning_rate": 4.127636317819109e-06,
      "loss": 3.4944,
      "step": 27040
    },
    {
      "epoch": 0.275115966796875,
      "grad_norm": 14.68928050994873,
      "learning_rate": 4.127332774981495e-06,
      "loss": 3.1315,
      "step": 27045
    },
    {
      "epoch": 0.2751668294270833,
      "grad_norm": 16.452741622924805,
      "learning_rate": 4.127029190508957e-06,
      "loss": 3.4271,
      "step": 27050
    },
    {
      "epoch": 0.2752176920572917,
      "grad_norm": 14.647496223449707,
      "learning_rate": 4.126725564409263e-06,
      "loss": 3.5874,
      "step": 27055
    },
    {
      "epoch": 0.2752685546875,
      "grad_norm": 12.843924522399902,
      "learning_rate": 4.126421896690181e-06,
      "loss": 3.3549,
      "step": 27060
    },
    {
      "epoch": 0.2753194173177083,
      "grad_norm": 12.407615661621094,
      "learning_rate": 4.12611818735948e-06,
      "loss": 3.3919,
      "step": 27065
    },
    {
      "epoch": 0.2753702799479167,
      "grad_norm": 12.004281997680664,
      "learning_rate": 4.125814436424931e-06,
      "loss": 3.2099,
      "step": 27070
    },
    {
      "epoch": 0.275421142578125,
      "grad_norm": 12.618803024291992,
      "learning_rate": 4.1255106438943045e-06,
      "loss": 2.9556,
      "step": 27075
    },
    {
      "epoch": 0.2754720052083333,
      "grad_norm": 14.524580001831055,
      "learning_rate": 4.125206809775374e-06,
      "loss": 3.7834,
      "step": 27080
    },
    {
      "epoch": 0.2755228678385417,
      "grad_norm": 14.32988166809082,
      "learning_rate": 4.124902934075912e-06,
      "loss": 3.8865,
      "step": 27085
    },
    {
      "epoch": 0.27557373046875,
      "grad_norm": 11.782154083251953,
      "learning_rate": 4.124599016803695e-06,
      "loss": 3.2699,
      "step": 27090
    },
    {
      "epoch": 0.2756245930989583,
      "grad_norm": 10.000688552856445,
      "learning_rate": 4.124295057966496e-06,
      "loss": 3.0126,
      "step": 27095
    },
    {
      "epoch": 0.2756754557291667,
      "grad_norm": 11.379739761352539,
      "learning_rate": 4.123991057572092e-06,
      "loss": 3.6174,
      "step": 27100
    },
    {
      "epoch": 0.275726318359375,
      "grad_norm": 13.724488258361816,
      "learning_rate": 4.123687015628263e-06,
      "loss": 3.2716,
      "step": 27105
    },
    {
      "epoch": 0.2757771809895833,
      "grad_norm": 15.961641311645508,
      "learning_rate": 4.123382932142786e-06,
      "loss": 3.2398,
      "step": 27110
    },
    {
      "epoch": 0.2758280436197917,
      "grad_norm": 10.406798362731934,
      "learning_rate": 4.123078807123442e-06,
      "loss": 3.2444,
      "step": 27115
    },
    {
      "epoch": 0.27587890625,
      "grad_norm": 8.292187690734863,
      "learning_rate": 4.12277464057801e-06,
      "loss": 3.3237,
      "step": 27120
    },
    {
      "epoch": 0.2759297688802083,
      "grad_norm": 9.379560470581055,
      "learning_rate": 4.1224704325142754e-06,
      "loss": 3.5323,
      "step": 27125
    },
    {
      "epoch": 0.2759806315104167,
      "grad_norm": 12.531656265258789,
      "learning_rate": 4.122166182940018e-06,
      "loss": 3.2237,
      "step": 27130
    },
    {
      "epoch": 0.276031494140625,
      "grad_norm": 8.860665321350098,
      "learning_rate": 4.1218618918630235e-06,
      "loss": 2.8973,
      "step": 27135
    },
    {
      "epoch": 0.2760823567708333,
      "grad_norm": 18.059009552001953,
      "learning_rate": 4.121557559291078e-06,
      "loss": 3.2918,
      "step": 27140
    },
    {
      "epoch": 0.2761332194010417,
      "grad_norm": 7.78617525100708,
      "learning_rate": 4.121253185231966e-06,
      "loss": 3.0191,
      "step": 27145
    },
    {
      "epoch": 0.27618408203125,
      "grad_norm": 14.027008056640625,
      "learning_rate": 4.120948769693476e-06,
      "loss": 3.0125,
      "step": 27150
    },
    {
      "epoch": 0.2762349446614583,
      "grad_norm": 14.529438972473145,
      "learning_rate": 4.120644312683395e-06,
      "loss": 3.3326,
      "step": 27155
    },
    {
      "epoch": 0.2762858072916667,
      "grad_norm": 10.114949226379395,
      "learning_rate": 4.120339814209514e-06,
      "loss": 3.273,
      "step": 27160
    },
    {
      "epoch": 0.276336669921875,
      "grad_norm": 8.970900535583496,
      "learning_rate": 4.120035274279623e-06,
      "loss": 3.4991,
      "step": 27165
    },
    {
      "epoch": 0.2763875325520833,
      "grad_norm": 10.103755950927734,
      "learning_rate": 4.119730692901513e-06,
      "loss": 3.273,
      "step": 27170
    },
    {
      "epoch": 0.2764383951822917,
      "grad_norm": 11.475733757019043,
      "learning_rate": 4.119426070082977e-06,
      "loss": 3.2216,
      "step": 27175
    },
    {
      "epoch": 0.2764892578125,
      "grad_norm": 14.949260711669922,
      "learning_rate": 4.11912140583181e-06,
      "loss": 3.4254,
      "step": 27180
    },
    {
      "epoch": 0.2765401204427083,
      "grad_norm": 11.60977554321289,
      "learning_rate": 4.118816700155804e-06,
      "loss": 3.5087,
      "step": 27185
    },
    {
      "epoch": 0.2765909830729167,
      "grad_norm": 9.910012245178223,
      "learning_rate": 4.118511953062758e-06,
      "loss": 3.2046,
      "step": 27190
    },
    {
      "epoch": 0.276641845703125,
      "grad_norm": 9.686962127685547,
      "learning_rate": 4.1182071645604656e-06,
      "loss": 3.5146,
      "step": 27195
    },
    {
      "epoch": 0.2766927083333333,
      "grad_norm": 8.127184867858887,
      "learning_rate": 4.1179023346567286e-06,
      "loss": 3.4442,
      "step": 27200
    },
    {
      "epoch": 0.2767435709635417,
      "grad_norm": 10.926115036010742,
      "learning_rate": 4.117597463359342e-06,
      "loss": 3.1616,
      "step": 27205
    },
    {
      "epoch": 0.27679443359375,
      "grad_norm": 11.111457824707031,
      "learning_rate": 4.117292550676108e-06,
      "loss": 3.1595,
      "step": 27210
    },
    {
      "epoch": 0.2768452962239583,
      "grad_norm": 9.984879493713379,
      "learning_rate": 4.116987596614828e-06,
      "loss": 3.5258,
      "step": 27215
    },
    {
      "epoch": 0.2768961588541667,
      "grad_norm": 10.812644958496094,
      "learning_rate": 4.116682601183304e-06,
      "loss": 3.4125,
      "step": 27220
    },
    {
      "epoch": 0.276947021484375,
      "grad_norm": 10.24685287475586,
      "learning_rate": 4.1163775643893374e-06,
      "loss": 3.5382,
      "step": 27225
    },
    {
      "epoch": 0.2769978841145833,
      "grad_norm": 11.69446849822998,
      "learning_rate": 4.116072486240735e-06,
      "loss": 3.1572,
      "step": 27230
    },
    {
      "epoch": 0.2770487467447917,
      "grad_norm": 11.58246898651123,
      "learning_rate": 4.115767366745301e-06,
      "loss": 3.2686,
      "step": 27235
    },
    {
      "epoch": 0.277099609375,
      "grad_norm": 13.597831726074219,
      "learning_rate": 4.1154622059108415e-06,
      "loss": 3.3648,
      "step": 27240
    },
    {
      "epoch": 0.2771504720052083,
      "grad_norm": 13.547455787658691,
      "learning_rate": 4.1151570037451645e-06,
      "loss": 3.2043,
      "step": 27245
    },
    {
      "epoch": 0.2772013346354167,
      "grad_norm": 8.566726684570312,
      "learning_rate": 4.114851760256079e-06,
      "loss": 3.4433,
      "step": 27250
    },
    {
      "epoch": 0.277252197265625,
      "grad_norm": 13.299729347229004,
      "learning_rate": 4.114546475451394e-06,
      "loss": 3.7028,
      "step": 27255
    },
    {
      "epoch": 0.2773030598958333,
      "grad_norm": 13.903881072998047,
      "learning_rate": 4.114241149338919e-06,
      "loss": 3.248,
      "step": 27260
    },
    {
      "epoch": 0.2773539225260417,
      "grad_norm": 9.658771514892578,
      "learning_rate": 4.113935781926468e-06,
      "loss": 3.2362,
      "step": 27265
    },
    {
      "epoch": 0.27740478515625,
      "grad_norm": 14.61761474609375,
      "learning_rate": 4.113630373221852e-06,
      "loss": 3.3815,
      "step": 27270
    },
    {
      "epoch": 0.2774556477864583,
      "grad_norm": 8.722289085388184,
      "learning_rate": 4.113324923232886e-06,
      "loss": 3.3268,
      "step": 27275
    },
    {
      "epoch": 0.2775065104166667,
      "grad_norm": 14.244725227355957,
      "learning_rate": 4.113019431967383e-06,
      "loss": 3.3132,
      "step": 27280
    },
    {
      "epoch": 0.277557373046875,
      "grad_norm": 10.710943222045898,
      "learning_rate": 4.112713899433161e-06,
      "loss": 3.4074,
      "step": 27285
    },
    {
      "epoch": 0.2776082356770833,
      "grad_norm": 16.3253116607666,
      "learning_rate": 4.1124083256380354e-06,
      "loss": 3.2259,
      "step": 27290
    },
    {
      "epoch": 0.2776590983072917,
      "grad_norm": 13.641079902648926,
      "learning_rate": 4.112102710589826e-06,
      "loss": 3.0738,
      "step": 27295
    },
    {
      "epoch": 0.2777099609375,
      "grad_norm": 13.473037719726562,
      "learning_rate": 4.1117970542963504e-06,
      "loss": 3.2011,
      "step": 27300
    },
    {
      "epoch": 0.2777608235677083,
      "grad_norm": 16.01197624206543,
      "learning_rate": 4.111491356765429e-06,
      "loss": 3.3992,
      "step": 27305
    },
    {
      "epoch": 0.2778116861979167,
      "grad_norm": 8.830290794372559,
      "learning_rate": 4.111185618004884e-06,
      "loss": 3.1679,
      "step": 27310
    },
    {
      "epoch": 0.277862548828125,
      "grad_norm": 9.501608848571777,
      "learning_rate": 4.110879838022536e-06,
      "loss": 3.3934,
      "step": 27315
    },
    {
      "epoch": 0.2779134114583333,
      "grad_norm": 10.511950492858887,
      "learning_rate": 4.1105740168262085e-06,
      "loss": 3.1209,
      "step": 27320
    },
    {
      "epoch": 0.2779642740885417,
      "grad_norm": 13.050411224365234,
      "learning_rate": 4.110268154423728e-06,
      "loss": 3.3571,
      "step": 27325
    },
    {
      "epoch": 0.27801513671875,
      "grad_norm": 7.376755714416504,
      "learning_rate": 4.109962250822918e-06,
      "loss": 3.3481,
      "step": 27330
    },
    {
      "epoch": 0.2780659993489583,
      "grad_norm": 13.177421569824219,
      "learning_rate": 4.109656306031606e-06,
      "loss": 3.0446,
      "step": 27335
    },
    {
      "epoch": 0.2781168619791667,
      "grad_norm": 10.911126136779785,
      "learning_rate": 4.109350320057617e-06,
      "loss": 3.1723,
      "step": 27340
    },
    {
      "epoch": 0.278167724609375,
      "grad_norm": 15.355131149291992,
      "learning_rate": 4.109044292908783e-06,
      "loss": 3.5127,
      "step": 27345
    },
    {
      "epoch": 0.2782185872395833,
      "grad_norm": 9.465426445007324,
      "learning_rate": 4.108738224592932e-06,
      "loss": 3.5795,
      "step": 27350
    },
    {
      "epoch": 0.2782694498697917,
      "grad_norm": 11.438790321350098,
      "learning_rate": 4.108432115117894e-06,
      "loss": 3.7179,
      "step": 27355
    },
    {
      "epoch": 0.2783203125,
      "grad_norm": 15.0741605758667,
      "learning_rate": 4.108125964491503e-06,
      "loss": 3.372,
      "step": 27360
    },
    {
      "epoch": 0.2783711751302083,
      "grad_norm": 13.863125801086426,
      "learning_rate": 4.10781977272159e-06,
      "loss": 3.4886,
      "step": 27365
    },
    {
      "epoch": 0.2784220377604167,
      "grad_norm": 9.793328285217285,
      "learning_rate": 4.1075135398159885e-06,
      "loss": 3.2055,
      "step": 27370
    },
    {
      "epoch": 0.278472900390625,
      "grad_norm": 12.542889595031738,
      "learning_rate": 4.107207265782535e-06,
      "loss": 3.18,
      "step": 27375
    },
    {
      "epoch": 0.2785237630208333,
      "grad_norm": 10.684040069580078,
      "learning_rate": 4.106900950629063e-06,
      "loss": 3.6244,
      "step": 27380
    },
    {
      "epoch": 0.2785746256510417,
      "grad_norm": 11.289875984191895,
      "learning_rate": 4.106594594363413e-06,
      "loss": 3.3408,
      "step": 27385
    },
    {
      "epoch": 0.27862548828125,
      "grad_norm": 17.948379516601562,
      "learning_rate": 4.10628819699342e-06,
      "loss": 3.3196,
      "step": 27390
    },
    {
      "epoch": 0.2786763509114583,
      "grad_norm": 12.828132629394531,
      "learning_rate": 4.105981758526924e-06,
      "loss": 3.6498,
      "step": 27395
    },
    {
      "epoch": 0.2787272135416667,
      "grad_norm": 11.889634132385254,
      "learning_rate": 4.105675278971766e-06,
      "loss": 3.6316,
      "step": 27400
    },
    {
      "epoch": 0.278778076171875,
      "grad_norm": 9.463981628417969,
      "learning_rate": 4.105368758335787e-06,
      "loss": 3.2058,
      "step": 27405
    },
    {
      "epoch": 0.2788289388020833,
      "grad_norm": 12.077744483947754,
      "learning_rate": 4.105062196626828e-06,
      "loss": 3.7315,
      "step": 27410
    },
    {
      "epoch": 0.2788798014322917,
      "grad_norm": 17.923763275146484,
      "learning_rate": 4.104755593852733e-06,
      "loss": 3.3367,
      "step": 27415
    },
    {
      "epoch": 0.2789306640625,
      "grad_norm": 15.344260215759277,
      "learning_rate": 4.104448950021347e-06,
      "loss": 3.0758,
      "step": 27420
    },
    {
      "epoch": 0.2789815266927083,
      "grad_norm": 13.663217544555664,
      "learning_rate": 4.104142265140515e-06,
      "loss": 3.0715,
      "step": 27425
    },
    {
      "epoch": 0.2790323893229167,
      "grad_norm": 11.255102157592773,
      "learning_rate": 4.103835539218083e-06,
      "loss": 3.7433,
      "step": 27430
    },
    {
      "epoch": 0.279083251953125,
      "grad_norm": 11.063042640686035,
      "learning_rate": 4.103528772261899e-06,
      "loss": 3.2718,
      "step": 27435
    },
    {
      "epoch": 0.2791341145833333,
      "grad_norm": 15.842066764831543,
      "learning_rate": 4.103221964279811e-06,
      "loss": 3.7132,
      "step": 27440
    },
    {
      "epoch": 0.2791849772135417,
      "grad_norm": 13.824544906616211,
      "learning_rate": 4.10291511527967e-06,
      "loss": 3.6161,
      "step": 27445
    },
    {
      "epoch": 0.27923583984375,
      "grad_norm": 13.159097671508789,
      "learning_rate": 4.102608225269324e-06,
      "loss": 3.4805,
      "step": 27450
    },
    {
      "epoch": 0.2792867024739583,
      "grad_norm": 14.674468994140625,
      "learning_rate": 4.1023012942566285e-06,
      "loss": 3.423,
      "step": 27455
    },
    {
      "epoch": 0.2793375651041667,
      "grad_norm": 8.352295875549316,
      "learning_rate": 4.101994322249433e-06,
      "loss": 3.0751,
      "step": 27460
    },
    {
      "epoch": 0.279388427734375,
      "grad_norm": 10.348424911499023,
      "learning_rate": 4.1016873092555935e-06,
      "loss": 3.8127,
      "step": 27465
    },
    {
      "epoch": 0.2794392903645833,
      "grad_norm": 10.092679977416992,
      "learning_rate": 4.101380255282963e-06,
      "loss": 3.8686,
      "step": 27470
    },
    {
      "epoch": 0.2794901529947917,
      "grad_norm": 9.561445236206055,
      "learning_rate": 4.101073160339398e-06,
      "loss": 3.1852,
      "step": 27475
    },
    {
      "epoch": 0.279541015625,
      "grad_norm": 12.506081581115723,
      "learning_rate": 4.100766024432756e-06,
      "loss": 3.4527,
      "step": 27480
    },
    {
      "epoch": 0.2795918782552083,
      "grad_norm": 11.840296745300293,
      "learning_rate": 4.100458847570895e-06,
      "loss": 3.3029,
      "step": 27485
    },
    {
      "epoch": 0.2796427408854167,
      "grad_norm": 10.97950267791748,
      "learning_rate": 4.100151629761673e-06,
      "loss": 3.3937,
      "step": 27490
    },
    {
      "epoch": 0.279693603515625,
      "grad_norm": 12.873710632324219,
      "learning_rate": 4.099844371012951e-06,
      "loss": 3.4339,
      "step": 27495
    },
    {
      "epoch": 0.2797444661458333,
      "grad_norm": 10.452871322631836,
      "learning_rate": 4.099537071332591e-06,
      "loss": 3.7318,
      "step": 27500
    },
    {
      "epoch": 0.2797953287760417,
      "grad_norm": 10.990396499633789,
      "learning_rate": 4.099229730728453e-06,
      "loss": 3.7192,
      "step": 27505
    },
    {
      "epoch": 0.27984619140625,
      "grad_norm": 14.749526977539062,
      "learning_rate": 4.098922349208402e-06,
      "loss": 3.0498,
      "step": 27510
    },
    {
      "epoch": 0.2798970540364583,
      "grad_norm": 15.220508575439453,
      "learning_rate": 4.098614926780301e-06,
      "loss": 3.1953,
      "step": 27515
    },
    {
      "epoch": 0.2799479166666667,
      "grad_norm": 9.671578407287598,
      "learning_rate": 4.098307463452017e-06,
      "loss": 3.2799,
      "step": 27520
    },
    {
      "epoch": 0.279998779296875,
      "grad_norm": 14.070199966430664,
      "learning_rate": 4.097999959231414e-06,
      "loss": 3.4482,
      "step": 27525
    },
    {
      "epoch": 0.2800496419270833,
      "grad_norm": 11.59659194946289,
      "learning_rate": 4.097692414126363e-06,
      "loss": 3.2175,
      "step": 27530
    },
    {
      "epoch": 0.2801005045572917,
      "grad_norm": 10.20411491394043,
      "learning_rate": 4.097384828144728e-06,
      "loss": 3.68,
      "step": 27535
    },
    {
      "epoch": 0.2801513671875,
      "grad_norm": 9.166175842285156,
      "learning_rate": 4.097077201294383e-06,
      "loss": 3.2877,
      "step": 27540
    },
    {
      "epoch": 0.2802022298177083,
      "grad_norm": 11.604389190673828,
      "learning_rate": 4.096769533583195e-06,
      "loss": 3.4169,
      "step": 27545
    },
    {
      "epoch": 0.2802530924479167,
      "grad_norm": 9.63940715789795,
      "learning_rate": 4.096461825019037e-06,
      "loss": 3.1978,
      "step": 27550
    },
    {
      "epoch": 0.280303955078125,
      "grad_norm": 13.217351913452148,
      "learning_rate": 4.096154075609781e-06,
      "loss": 3.2307,
      "step": 27555
    },
    {
      "epoch": 0.2803548177083333,
      "grad_norm": 8.332501411437988,
      "learning_rate": 4.095846285363302e-06,
      "loss": 3.5514,
      "step": 27560
    },
    {
      "epoch": 0.2804056803385417,
      "grad_norm": 7.972536563873291,
      "learning_rate": 4.0955384542874745e-06,
      "loss": 3.111,
      "step": 27565
    },
    {
      "epoch": 0.28045654296875,
      "grad_norm": 13.895954132080078,
      "learning_rate": 4.0952305823901726e-06,
      "loss": 3.4349,
      "step": 27570
    },
    {
      "epoch": 0.2805074055989583,
      "grad_norm": 16.52075958251953,
      "learning_rate": 4.094922669679275e-06,
      "loss": 3.6657,
      "step": 27575
    },
    {
      "epoch": 0.2805582682291667,
      "grad_norm": 13.944071769714355,
      "learning_rate": 4.09461471616266e-06,
      "loss": 3.4333,
      "step": 27580
    },
    {
      "epoch": 0.280609130859375,
      "grad_norm": 15.831037521362305,
      "learning_rate": 4.0943067218482044e-06,
      "loss": 3.2349,
      "step": 27585
    },
    {
      "epoch": 0.2806599934895833,
      "grad_norm": 10.248038291931152,
      "learning_rate": 4.09399868674379e-06,
      "loss": 3.6537,
      "step": 27590
    },
    {
      "epoch": 0.2807108561197917,
      "grad_norm": 8.459166526794434,
      "learning_rate": 4.0936906108572955e-06,
      "loss": 3.6726,
      "step": 27595
    },
    {
      "epoch": 0.28076171875,
      "grad_norm": 12.822853088378906,
      "learning_rate": 4.0933824941966055e-06,
      "loss": 3.4818,
      "step": 27600
    },
    {
      "epoch": 0.2808125813802083,
      "grad_norm": 13.79715633392334,
      "learning_rate": 4.093074336769601e-06,
      "loss": 3.5654,
      "step": 27605
    },
    {
      "epoch": 0.2808634440104167,
      "grad_norm": 12.120914459228516,
      "learning_rate": 4.092766138584169e-06,
      "loss": 2.9587,
      "step": 27610
    },
    {
      "epoch": 0.280914306640625,
      "grad_norm": 10.529516220092773,
      "learning_rate": 4.092457899648192e-06,
      "loss": 3.1997,
      "step": 27615
    },
    {
      "epoch": 0.2809651692708333,
      "grad_norm": 12.962397575378418,
      "learning_rate": 4.092149619969556e-06,
      "loss": 3.8154,
      "step": 27620
    },
    {
      "epoch": 0.2810160319010417,
      "grad_norm": 11.064664840698242,
      "learning_rate": 4.091841299556151e-06,
      "loss": 3.0046,
      "step": 27625
    },
    {
      "epoch": 0.28106689453125,
      "grad_norm": 14.863337516784668,
      "learning_rate": 4.091532938415863e-06,
      "loss": 3.0863,
      "step": 27630
    },
    {
      "epoch": 0.2811177571614583,
      "grad_norm": 8.072478294372559,
      "learning_rate": 4.091224536556583e-06,
      "loss": 3.5172,
      "step": 27635
    },
    {
      "epoch": 0.2811686197916667,
      "grad_norm": 9.187117576599121,
      "learning_rate": 4.090916093986199e-06,
      "loss": 3.6613,
      "step": 27640
    },
    {
      "epoch": 0.281219482421875,
      "grad_norm": 10.171202659606934,
      "learning_rate": 4.090607610712604e-06,
      "loss": 3.2175,
      "step": 27645
    },
    {
      "epoch": 0.2812703450520833,
      "grad_norm": 15.561723709106445,
      "learning_rate": 4.090299086743691e-06,
      "loss": 3.1775,
      "step": 27650
    },
    {
      "epoch": 0.2813212076822917,
      "grad_norm": 12.204340934753418,
      "learning_rate": 4.089990522087352e-06,
      "loss": 3.1616,
      "step": 27655
    },
    {
      "epoch": 0.2813720703125,
      "grad_norm": 17.98369026184082,
      "learning_rate": 4.089681916751483e-06,
      "loss": 3.4742,
      "step": 27660
    },
    {
      "epoch": 0.2814229329427083,
      "grad_norm": 12.7793607711792,
      "learning_rate": 4.089373270743978e-06,
      "loss": 2.9907,
      "step": 27665
    },
    {
      "epoch": 0.2814737955729167,
      "grad_norm": 17.77541732788086,
      "learning_rate": 4.089064584072735e-06,
      "loss": 3.446,
      "step": 27670
    },
    {
      "epoch": 0.281524658203125,
      "grad_norm": 13.95572280883789,
      "learning_rate": 4.088755856745652e-06,
      "loss": 3.262,
      "step": 27675
    },
    {
      "epoch": 0.2815755208333333,
      "grad_norm": 16.80682373046875,
      "learning_rate": 4.088447088770627e-06,
      "loss": 3.6142,
      "step": 27680
    },
    {
      "epoch": 0.2816263834635417,
      "grad_norm": 14.185280799865723,
      "learning_rate": 4.088138280155559e-06,
      "loss": 3.7258,
      "step": 27685
    },
    {
      "epoch": 0.28167724609375,
      "grad_norm": 10.341408729553223,
      "learning_rate": 4.08782943090835e-06,
      "loss": 3.1562,
      "step": 27690
    },
    {
      "epoch": 0.2817281087239583,
      "grad_norm": 12.043773651123047,
      "learning_rate": 4.087520541036901e-06,
      "loss": 3.1961,
      "step": 27695
    },
    {
      "epoch": 0.2817789713541667,
      "grad_norm": 12.786687850952148,
      "learning_rate": 4.087211610549115e-06,
      "loss": 3.3973,
      "step": 27700
    },
    {
      "epoch": 0.281829833984375,
      "grad_norm": 14.750553131103516,
      "learning_rate": 4.086902639452897e-06,
      "loss": 3.3499,
      "step": 27705
    },
    {
      "epoch": 0.2818806966145833,
      "grad_norm": 7.420515537261963,
      "learning_rate": 4.08659362775615e-06,
      "loss": 3.2789,
      "step": 27710
    },
    {
      "epoch": 0.2819315592447917,
      "grad_norm": 8.589488983154297,
      "learning_rate": 4.086284575466783e-06,
      "loss": 3.4395,
      "step": 27715
    },
    {
      "epoch": 0.281982421875,
      "grad_norm": 14.356674194335938,
      "learning_rate": 4.085975482592699e-06,
      "loss": 3.8196,
      "step": 27720
    },
    {
      "epoch": 0.2820332845052083,
      "grad_norm": 10.523839950561523,
      "learning_rate": 4.085666349141809e-06,
      "loss": 3.5714,
      "step": 27725
    },
    {
      "epoch": 0.2820841471354167,
      "grad_norm": 14.31343936920166,
      "learning_rate": 4.085357175122022e-06,
      "loss": 3.2955,
      "step": 27730
    },
    {
      "epoch": 0.282135009765625,
      "grad_norm": 13.042136192321777,
      "learning_rate": 4.085047960541247e-06,
      "loss": 3.2577,
      "step": 27735
    },
    {
      "epoch": 0.2821858723958333,
      "grad_norm": 13.704607963562012,
      "learning_rate": 4.084738705407395e-06,
      "loss": 3.2682,
      "step": 27740
    },
    {
      "epoch": 0.2822367350260417,
      "grad_norm": 13.314915657043457,
      "learning_rate": 4.084429409728379e-06,
      "loss": 4.1069,
      "step": 27745
    },
    {
      "epoch": 0.28228759765625,
      "grad_norm": 14.6658935546875,
      "learning_rate": 4.084120073512113e-06,
      "loss": 3.1926,
      "step": 27750
    },
    {
      "epoch": 0.2823384602864583,
      "grad_norm": 11.541994094848633,
      "learning_rate": 4.083810696766509e-06,
      "loss": 3.8473,
      "step": 27755
    },
    {
      "epoch": 0.2823893229166667,
      "grad_norm": 9.510300636291504,
      "learning_rate": 4.083501279499485e-06,
      "loss": 3.2756,
      "step": 27760
    },
    {
      "epoch": 0.282440185546875,
      "grad_norm": 11.580392837524414,
      "learning_rate": 4.083191821718956e-06,
      "loss": 3.4178,
      "step": 27765
    },
    {
      "epoch": 0.2824910481770833,
      "grad_norm": 9.93171501159668,
      "learning_rate": 4.082882323432839e-06,
      "loss": 3.6801,
      "step": 27770
    },
    {
      "epoch": 0.2825419108072917,
      "grad_norm": 15.171422958374023,
      "learning_rate": 4.082572784649053e-06,
      "loss": 3.283,
      "step": 27775
    },
    {
      "epoch": 0.2825927734375,
      "grad_norm": 13.232819557189941,
      "learning_rate": 4.082263205375517e-06,
      "loss": 3.6217,
      "step": 27780
    },
    {
      "epoch": 0.2826436360677083,
      "grad_norm": 11.675466537475586,
      "learning_rate": 4.081953585620153e-06,
      "loss": 3.4335,
      "step": 27785
    },
    {
      "epoch": 0.2826944986979167,
      "grad_norm": 8.862447738647461,
      "learning_rate": 4.081643925390881e-06,
      "loss": 3.3755,
      "step": 27790
    },
    {
      "epoch": 0.282745361328125,
      "grad_norm": 10.922281265258789,
      "learning_rate": 4.081334224695623e-06,
      "loss": 3.6146,
      "step": 27795
    },
    {
      "epoch": 0.2827962239583333,
      "grad_norm": 9.969596862792969,
      "learning_rate": 4.081024483542306e-06,
      "loss": 3.6792,
      "step": 27800
    },
    {
      "epoch": 0.2828470865885417,
      "grad_norm": 11.884249687194824,
      "learning_rate": 4.08071470193885e-06,
      "loss": 3.3235,
      "step": 27805
    },
    {
      "epoch": 0.28289794921875,
      "grad_norm": 14.386703491210938,
      "learning_rate": 4.080404879893184e-06,
      "loss": 3.3408,
      "step": 27810
    },
    {
      "epoch": 0.2829488118489583,
      "grad_norm": 15.549338340759277,
      "learning_rate": 4.080095017413234e-06,
      "loss": 3.3848,
      "step": 27815
    },
    {
      "epoch": 0.2829996744791667,
      "grad_norm": 11.88217830657959,
      "learning_rate": 4.079785114506927e-06,
      "loss": 3.8501,
      "step": 27820
    },
    {
      "epoch": 0.283050537109375,
      "grad_norm": 16.217618942260742,
      "learning_rate": 4.079475171182193e-06,
      "loss": 3.5115,
      "step": 27825
    },
    {
      "epoch": 0.2831013997395833,
      "grad_norm": 14.34183406829834,
      "learning_rate": 4.0791651874469605e-06,
      "loss": 3.622,
      "step": 27830
    },
    {
      "epoch": 0.2831522623697917,
      "grad_norm": 14.53591251373291,
      "learning_rate": 4.078855163309161e-06,
      "loss": 3.2665,
      "step": 27835
    },
    {
      "epoch": 0.283203125,
      "grad_norm": 10.941570281982422,
      "learning_rate": 4.078545098776726e-06,
      "loss": 3.858,
      "step": 27840
    },
    {
      "epoch": 0.2832539876302083,
      "grad_norm": 9.548334121704102,
      "learning_rate": 4.07823499385759e-06,
      "loss": 3.6304,
      "step": 27845
    },
    {
      "epoch": 0.2833048502604167,
      "grad_norm": 13.581596374511719,
      "learning_rate": 4.077924848559685e-06,
      "loss": 3.3858,
      "step": 27850
    },
    {
      "epoch": 0.283355712890625,
      "grad_norm": 11.762258529663086,
      "learning_rate": 4.077614662890947e-06,
      "loss": 3.2353,
      "step": 27855
    },
    {
      "epoch": 0.2834065755208333,
      "grad_norm": 15.425300598144531,
      "learning_rate": 4.077304436859311e-06,
      "loss": 3.2407,
      "step": 27860
    },
    {
      "epoch": 0.2834574381510417,
      "grad_norm": 16.47221565246582,
      "learning_rate": 4.0769941704727154e-06,
      "loss": 3.4532,
      "step": 27865
    },
    {
      "epoch": 0.28350830078125,
      "grad_norm": 10.942659378051758,
      "learning_rate": 4.076683863739098e-06,
      "loss": 3.0886,
      "step": 27870
    },
    {
      "epoch": 0.2835591634114583,
      "grad_norm": 15.116689682006836,
      "learning_rate": 4.076373516666398e-06,
      "loss": 3.561,
      "step": 27875
    },
    {
      "epoch": 0.2836100260416667,
      "grad_norm": 12.724804878234863,
      "learning_rate": 4.076063129262554e-06,
      "loss": 3.1554,
      "step": 27880
    },
    {
      "epoch": 0.283660888671875,
      "grad_norm": 15.09601879119873,
      "learning_rate": 4.07575270153551e-06,
      "loss": 3.1598,
      "step": 27885
    },
    {
      "epoch": 0.2837117513020833,
      "grad_norm": 10.501201629638672,
      "learning_rate": 4.0754422334932054e-06,
      "loss": 2.8968,
      "step": 27890
    },
    {
      "epoch": 0.2837626139322917,
      "grad_norm": 13.239876747131348,
      "learning_rate": 4.075131725143585e-06,
      "loss": 3.1013,
      "step": 27895
    },
    {
      "epoch": 0.2838134765625,
      "grad_norm": 9.330605506896973,
      "learning_rate": 4.074821176494592e-06,
      "loss": 3.3436,
      "step": 27900
    },
    {
      "epoch": 0.2838643391927083,
      "grad_norm": 13.973194122314453,
      "learning_rate": 4.074510587554173e-06,
      "loss": 3.3293,
      "step": 27905
    },
    {
      "epoch": 0.2839152018229167,
      "grad_norm": 16.430049896240234,
      "learning_rate": 4.074199958330275e-06,
      "loss": 3.6129,
      "step": 27910
    },
    {
      "epoch": 0.283966064453125,
      "grad_norm": 9.183777809143066,
      "learning_rate": 4.073889288830843e-06,
      "loss": 3.6742,
      "step": 27915
    },
    {
      "epoch": 0.2840169270833333,
      "grad_norm": 9.213640213012695,
      "learning_rate": 4.073578579063827e-06,
      "loss": 3.4222,
      "step": 27920
    },
    {
      "epoch": 0.2840677897135417,
      "grad_norm": 14.945640563964844,
      "learning_rate": 4.073267829037175e-06,
      "loss": 3.3423,
      "step": 27925
    },
    {
      "epoch": 0.28411865234375,
      "grad_norm": 9.454371452331543,
      "learning_rate": 4.07295703875884e-06,
      "loss": 3.3122,
      "step": 27930
    },
    {
      "epoch": 0.2841695149739583,
      "grad_norm": 17.040342330932617,
      "learning_rate": 4.072646208236772e-06,
      "loss": 3.4337,
      "step": 27935
    },
    {
      "epoch": 0.2842203776041667,
      "grad_norm": 15.587324142456055,
      "learning_rate": 4.072335337478922e-06,
      "loss": 3.4444,
      "step": 27940
    },
    {
      "epoch": 0.284271240234375,
      "grad_norm": 85.70834350585938,
      "learning_rate": 4.072024426493246e-06,
      "loss": 3.3656,
      "step": 27945
    },
    {
      "epoch": 0.2843221028645833,
      "grad_norm": 7.995316505432129,
      "learning_rate": 4.071713475287698e-06,
      "loss": 3.1318,
      "step": 27950
    },
    {
      "epoch": 0.2843729654947917,
      "grad_norm": 13.728548049926758,
      "learning_rate": 4.071402483870233e-06,
      "loss": 3.6891,
      "step": 27955
    },
    {
      "epoch": 0.284423828125,
      "grad_norm": 18.6165828704834,
      "learning_rate": 4.071091452248807e-06,
      "loss": 3.1881,
      "step": 27960
    },
    {
      "epoch": 0.2844746907552083,
      "grad_norm": 12.543664932250977,
      "learning_rate": 4.070780380431379e-06,
      "loss": 3.5005,
      "step": 27965
    },
    {
      "epoch": 0.2845255533854167,
      "grad_norm": 15.050885200500488,
      "learning_rate": 4.070469268425909e-06,
      "loss": 3.4116,
      "step": 27970
    },
    {
      "epoch": 0.284576416015625,
      "grad_norm": 10.877278327941895,
      "learning_rate": 4.070158116240353e-06,
      "loss": 3.8081,
      "step": 27975
    },
    {
      "epoch": 0.2846272786458333,
      "grad_norm": 14.296137809753418,
      "learning_rate": 4.069846923882674e-06,
      "loss": 3.4596,
      "step": 27980
    },
    {
      "epoch": 0.2846781412760417,
      "grad_norm": 7.9767560958862305,
      "learning_rate": 4.069535691360835e-06,
      "loss": 3.3673,
      "step": 27985
    },
    {
      "epoch": 0.28472900390625,
      "grad_norm": 11.449370384216309,
      "learning_rate": 4.069224418682796e-06,
      "loss": 2.9643,
      "step": 27990
    },
    {
      "epoch": 0.2847798665364583,
      "grad_norm": 11.578384399414062,
      "learning_rate": 4.068913105856522e-06,
      "loss": 3.1005,
      "step": 27995
    },
    {
      "epoch": 0.2848307291666667,
      "grad_norm": 13.277164459228516,
      "learning_rate": 4.068601752889979e-06,
      "loss": 3.0292,
      "step": 28000
    },
    {
      "epoch": 0.284881591796875,
      "grad_norm": 14.91826343536377,
      "learning_rate": 4.068290359791131e-06,
      "loss": 3.2388,
      "step": 28005
    },
    {
      "epoch": 0.2849324544270833,
      "grad_norm": 28.24471092224121,
      "learning_rate": 4.067978926567947e-06,
      "loss": 3.4851,
      "step": 28010
    },
    {
      "epoch": 0.2849833170572917,
      "grad_norm": 11.40623950958252,
      "learning_rate": 4.0676674532283936e-06,
      "loss": 3.4192,
      "step": 28015
    },
    {
      "epoch": 0.2850341796875,
      "grad_norm": 7.215197563171387,
      "learning_rate": 4.067355939780439e-06,
      "loss": 2.9501,
      "step": 28020
    },
    {
      "epoch": 0.2850850423177083,
      "grad_norm": 13.303850173950195,
      "learning_rate": 4.067044386232055e-06,
      "loss": 3.3592,
      "step": 28025
    },
    {
      "epoch": 0.2851359049479167,
      "grad_norm": 11.901801109313965,
      "learning_rate": 4.066732792591211e-06,
      "loss": 3.1835,
      "step": 28030
    },
    {
      "epoch": 0.285186767578125,
      "grad_norm": 11.704801559448242,
      "learning_rate": 4.066421158865881e-06,
      "loss": 3.851,
      "step": 28035
    },
    {
      "epoch": 0.2852376302083333,
      "grad_norm": 9.540196418762207,
      "learning_rate": 4.066109485064037e-06,
      "loss": 3.4988,
      "step": 28040
    },
    {
      "epoch": 0.2852884928385417,
      "grad_norm": 7.771119117736816,
      "learning_rate": 4.065797771193652e-06,
      "loss": 3.3013,
      "step": 28045
    },
    {
      "epoch": 0.28533935546875,
      "grad_norm": 9.430291175842285,
      "learning_rate": 4.065486017262703e-06,
      "loss": 3.1248,
      "step": 28050
    },
    {
      "epoch": 0.2853902180989583,
      "grad_norm": 10.184014320373535,
      "learning_rate": 4.065174223279165e-06,
      "loss": 3.3792,
      "step": 28055
    },
    {
      "epoch": 0.2854410807291667,
      "grad_norm": 14.535456657409668,
      "learning_rate": 4.0648623892510145e-06,
      "loss": 3.9619,
      "step": 28060
    },
    {
      "epoch": 0.285491943359375,
      "grad_norm": 11.507352828979492,
      "learning_rate": 4.064550515186232e-06,
      "loss": 3.2716,
      "step": 28065
    },
    {
      "epoch": 0.2855428059895833,
      "grad_norm": 10.623822212219238,
      "learning_rate": 4.064238601092795e-06,
      "loss": 3.1316,
      "step": 28070
    },
    {
      "epoch": 0.2855936686197917,
      "grad_norm": 10.219310760498047,
      "learning_rate": 4.063926646978684e-06,
      "loss": 3.2253,
      "step": 28075
    },
    {
      "epoch": 0.28564453125,
      "grad_norm": 11.938286781311035,
      "learning_rate": 4.06361465285188e-06,
      "loss": 3.2599,
      "step": 28080
    },
    {
      "epoch": 0.2856953938802083,
      "grad_norm": 15.574570655822754,
      "learning_rate": 4.063302618720367e-06,
      "loss": 3.5626,
      "step": 28085
    },
    {
      "epoch": 0.2857462565104167,
      "grad_norm": 10.333740234375,
      "learning_rate": 4.062990544592126e-06,
      "loss": 3.8863,
      "step": 28090
    },
    {
      "epoch": 0.285797119140625,
      "grad_norm": 17.025602340698242,
      "learning_rate": 4.062678430475143e-06,
      "loss": 3.6249,
      "step": 28095
    },
    {
      "epoch": 0.2858479817708333,
      "grad_norm": 9.703519821166992,
      "learning_rate": 4.062366276377403e-06,
      "loss": 3.3165,
      "step": 28100
    },
    {
      "epoch": 0.2858988444010417,
      "grad_norm": 14.47970962524414,
      "learning_rate": 4.062054082306891e-06,
      "loss": 3.3594,
      "step": 28105
    },
    {
      "epoch": 0.28594970703125,
      "grad_norm": 15.501633644104004,
      "learning_rate": 4.061741848271597e-06,
      "loss": 3.4847,
      "step": 28110
    },
    {
      "epoch": 0.2860005696614583,
      "grad_norm": 12.506375312805176,
      "learning_rate": 4.061429574279506e-06,
      "loss": 3.8149,
      "step": 28115
    },
    {
      "epoch": 0.2860514322916667,
      "grad_norm": 8.678794860839844,
      "learning_rate": 4.061117260338611e-06,
      "loss": 3.5832,
      "step": 28120
    },
    {
      "epoch": 0.286102294921875,
      "grad_norm": 16.247833251953125,
      "learning_rate": 4.0608049064569e-06,
      "loss": 2.9734,
      "step": 28125
    },
    {
      "epoch": 0.2861531575520833,
      "grad_norm": 12.02649974822998,
      "learning_rate": 4.060492512642366e-06,
      "loss": 3.3525,
      "step": 28130
    },
    {
      "epoch": 0.2862040201822917,
      "grad_norm": 7.934184551239014,
      "learning_rate": 4.060180078903001e-06,
      "loss": 3.7371,
      "step": 28135
    },
    {
      "epoch": 0.2862548828125,
      "grad_norm": 8.472125053405762,
      "learning_rate": 4.059867605246799e-06,
      "loss": 3.0991,
      "step": 28140
    },
    {
      "epoch": 0.2863057454427083,
      "grad_norm": 12.102945327758789,
      "learning_rate": 4.059555091681753e-06,
      "loss": 3.4988,
      "step": 28145
    },
    {
      "epoch": 0.2863566080729167,
      "grad_norm": 12.248966217041016,
      "learning_rate": 4.05924253821586e-06,
      "loss": 3.1905,
      "step": 28150
    },
    {
      "epoch": 0.286407470703125,
      "grad_norm": 10.715188026428223,
      "learning_rate": 4.0589299448571164e-06,
      "loss": 3.6046,
      "step": 28155
    },
    {
      "epoch": 0.2864583333333333,
      "grad_norm": 13.544321060180664,
      "learning_rate": 4.05861731161352e-06,
      "loss": 3.216,
      "step": 28160
    },
    {
      "epoch": 0.2865091959635417,
      "grad_norm": 12.897768020629883,
      "learning_rate": 4.058304638493068e-06,
      "loss": 3.0793,
      "step": 28165
    },
    {
      "epoch": 0.28656005859375,
      "grad_norm": 9.566934585571289,
      "learning_rate": 4.057991925503762e-06,
      "loss": 3.435,
      "step": 28170
    },
    {
      "epoch": 0.2866109212239583,
      "grad_norm": 15.859289169311523,
      "learning_rate": 4.0576791726536015e-06,
      "loss": 3.0893,
      "step": 28175
    },
    {
      "epoch": 0.2866617838541667,
      "grad_norm": 9.841944694519043,
      "learning_rate": 4.057366379950589e-06,
      "loss": 3.0007,
      "step": 28180
    },
    {
      "epoch": 0.286712646484375,
      "grad_norm": 14.592065811157227,
      "learning_rate": 4.057053547402726e-06,
      "loss": 3.0607,
      "step": 28185
    },
    {
      "epoch": 0.2867635091145833,
      "grad_norm": 7.597635746002197,
      "learning_rate": 4.056740675018018e-06,
      "loss": 3.4154,
      "step": 28190
    },
    {
      "epoch": 0.2868143717447917,
      "grad_norm": 11.658868789672852,
      "learning_rate": 4.056427762804468e-06,
      "loss": 3.5037,
      "step": 28195
    },
    {
      "epoch": 0.286865234375,
      "grad_norm": 14.114136695861816,
      "learning_rate": 4.0561148107700836e-06,
      "loss": 3.3819,
      "step": 28200
    },
    {
      "epoch": 0.2869160970052083,
      "grad_norm": 9.104647636413574,
      "learning_rate": 4.05580181892287e-06,
      "loss": 3.049,
      "step": 28205
    },
    {
      "epoch": 0.2869669596354167,
      "grad_norm": 8.600118637084961,
      "learning_rate": 4.055488787270836e-06,
      "loss": 3.2881,
      "step": 28210
    },
    {
      "epoch": 0.287017822265625,
      "grad_norm": 8.562788009643555,
      "learning_rate": 4.055175715821989e-06,
      "loss": 3.3843,
      "step": 28215
    },
    {
      "epoch": 0.2870686848958333,
      "grad_norm": 9.191661834716797,
      "learning_rate": 4.054862604584341e-06,
      "loss": 3.013,
      "step": 28220
    },
    {
      "epoch": 0.2871195475260417,
      "grad_norm": 9.98511028289795,
      "learning_rate": 4.054549453565902e-06,
      "loss": 3.4819,
      "step": 28225
    },
    {
      "epoch": 0.28717041015625,
      "grad_norm": 11.296266555786133,
      "learning_rate": 4.0542362627746825e-06,
      "loss": 3.2387,
      "step": 28230
    },
    {
      "epoch": 0.2872212727864583,
      "grad_norm": 15.573591232299805,
      "learning_rate": 4.053923032218698e-06,
      "loss": 3.583,
      "step": 28235
    },
    {
      "epoch": 0.2872721354166667,
      "grad_norm": 14.96290111541748,
      "learning_rate": 4.05360976190596e-06,
      "loss": 3.3793,
      "step": 28240
    },
    {
      "epoch": 0.287322998046875,
      "grad_norm": 13.247879981994629,
      "learning_rate": 4.053296451844485e-06,
      "loss": 3.342,
      "step": 28245
    },
    {
      "epoch": 0.2873738606770833,
      "grad_norm": 10.612177848815918,
      "learning_rate": 4.052983102042288e-06,
      "loss": 3.2241,
      "step": 28250
    },
    {
      "epoch": 0.2874247233072917,
      "grad_norm": 13.62741756439209,
      "learning_rate": 4.0526697125073876e-06,
      "loss": 3.4002,
      "step": 28255
    },
    {
      "epoch": 0.2874755859375,
      "grad_norm": 13.484682083129883,
      "learning_rate": 4.052356283247799e-06,
      "loss": 3.1952,
      "step": 28260
    },
    {
      "epoch": 0.2875264485677083,
      "grad_norm": 14.832202911376953,
      "learning_rate": 4.052042814271544e-06,
      "loss": 3.1964,
      "step": 28265
    },
    {
      "epoch": 0.2875773111979167,
      "grad_norm": 13.506696701049805,
      "learning_rate": 4.051729305586641e-06,
      "loss": 3.4984,
      "step": 28270
    },
    {
      "epoch": 0.287628173828125,
      "grad_norm": 13.403464317321777,
      "learning_rate": 4.051415757201111e-06,
      "loss": 3.051,
      "step": 28275
    },
    {
      "epoch": 0.2876790364583333,
      "grad_norm": 13.91868782043457,
      "learning_rate": 4.051102169122977e-06,
      "loss": 3.4574,
      "step": 28280
    },
    {
      "epoch": 0.2877298990885417,
      "grad_norm": 12.299845695495605,
      "learning_rate": 4.050788541360262e-06,
      "loss": 3.5358,
      "step": 28285
    },
    {
      "epoch": 0.28778076171875,
      "grad_norm": 13.38621997833252,
      "learning_rate": 4.05047487392099e-06,
      "loss": 3.3492,
      "step": 28290
    },
    {
      "epoch": 0.2878316243489583,
      "grad_norm": 14.945199966430664,
      "learning_rate": 4.050161166813186e-06,
      "loss": 3.8799,
      "step": 28295
    },
    {
      "epoch": 0.2878824869791667,
      "grad_norm": 10.105154991149902,
      "learning_rate": 4.049847420044874e-06,
      "loss": 3.613,
      "step": 28300
    },
    {
      "epoch": 0.287933349609375,
      "grad_norm": 11.531231880187988,
      "learning_rate": 4.049533633624086e-06,
      "loss": 3.4646,
      "step": 28305
    },
    {
      "epoch": 0.2879842122395833,
      "grad_norm": 12.640409469604492,
      "learning_rate": 4.049219807558845e-06,
      "loss": 3.1404,
      "step": 28310
    },
    {
      "epoch": 0.2880350748697917,
      "grad_norm": 14.797945976257324,
      "learning_rate": 4.048905941857183e-06,
      "loss": 3.8047,
      "step": 28315
    },
    {
      "epoch": 0.2880859375,
      "grad_norm": 7.843355655670166,
      "learning_rate": 4.048592036527131e-06,
      "loss": 3.6593,
      "step": 28320
    },
    {
      "epoch": 0.2881368001302083,
      "grad_norm": 13.126662254333496,
      "learning_rate": 4.048278091576716e-06,
      "loss": 3.5298,
      "step": 28325
    },
    {
      "epoch": 0.2881876627604167,
      "grad_norm": 7.7960429191589355,
      "learning_rate": 4.047964107013976e-06,
      "loss": 3.4729,
      "step": 28330
    },
    {
      "epoch": 0.288238525390625,
      "grad_norm": 14.168012619018555,
      "learning_rate": 4.047650082846939e-06,
      "loss": 3.2386,
      "step": 28335
    },
    {
      "epoch": 0.2882893880208333,
      "grad_norm": 12.29859447479248,
      "learning_rate": 4.047336019083642e-06,
      "loss": 3.6913,
      "step": 28340
    },
    {
      "epoch": 0.2883402506510417,
      "grad_norm": 12.323288917541504,
      "learning_rate": 4.04702191573212e-06,
      "loss": 3.2937,
      "step": 28345
    },
    {
      "epoch": 0.28839111328125,
      "grad_norm": 17.101675033569336,
      "learning_rate": 4.046707772800409e-06,
      "loss": 3.4794,
      "step": 28350
    },
    {
      "epoch": 0.2884419759114583,
      "grad_norm": 8.433690071105957,
      "learning_rate": 4.046393590296547e-06,
      "loss": 3.2544,
      "step": 28355
    },
    {
      "epoch": 0.2884928385416667,
      "grad_norm": 11.87043285369873,
      "learning_rate": 4.046079368228571e-06,
      "loss": 3.274,
      "step": 28360
    },
    {
      "epoch": 0.288543701171875,
      "grad_norm": 14.319379806518555,
      "learning_rate": 4.04576510660452e-06,
      "loss": 3.1518,
      "step": 28365
    },
    {
      "epoch": 0.2885945638020833,
      "grad_norm": 16.121383666992188,
      "learning_rate": 4.045450805432436e-06,
      "loss": 3.5974,
      "step": 28370
    },
    {
      "epoch": 0.2886454264322917,
      "grad_norm": 10.960174560546875,
      "learning_rate": 4.045136464720358e-06,
      "loss": 3.4076,
      "step": 28375
    },
    {
      "epoch": 0.2886962890625,
      "grad_norm": 17.95393180847168,
      "learning_rate": 4.044822084476332e-06,
      "loss": 3.2632,
      "step": 28380
    },
    {
      "epoch": 0.2887471516927083,
      "grad_norm": 13.769865989685059,
      "learning_rate": 4.044507664708398e-06,
      "loss": 3.3918,
      "step": 28385
    },
    {
      "epoch": 0.2887980143229167,
      "grad_norm": 7.544826984405518,
      "learning_rate": 4.0441932054246015e-06,
      "loss": 3.4184,
      "step": 28390
    },
    {
      "epoch": 0.288848876953125,
      "grad_norm": 7.78319787979126,
      "learning_rate": 4.043878706632988e-06,
      "loss": 3.4036,
      "step": 28395
    },
    {
      "epoch": 0.2888997395833333,
      "grad_norm": 10.946330070495605,
      "learning_rate": 4.0435641683416035e-06,
      "loss": 3.235,
      "step": 28400
    },
    {
      "epoch": 0.2889506022135417,
      "grad_norm": 15.540562629699707,
      "learning_rate": 4.043249590558496e-06,
      "loss": 3.3337,
      "step": 28405
    },
    {
      "epoch": 0.28900146484375,
      "grad_norm": 11.995061874389648,
      "learning_rate": 4.042934973291713e-06,
      "loss": 3.3849,
      "step": 28410
    },
    {
      "epoch": 0.2890523274739583,
      "grad_norm": 9.146485328674316,
      "learning_rate": 4.042620316549305e-06,
      "loss": 3.2122,
      "step": 28415
    },
    {
      "epoch": 0.2891031901041667,
      "grad_norm": 10.783262252807617,
      "learning_rate": 4.0423056203393204e-06,
      "loss": 3.1673,
      "step": 28420
    },
    {
      "epoch": 0.289154052734375,
      "grad_norm": 14.132058143615723,
      "learning_rate": 4.041990884669814e-06,
      "loss": 3.2879,
      "step": 28425
    },
    {
      "epoch": 0.2892049153645833,
      "grad_norm": 13.759737968444824,
      "learning_rate": 4.041676109548835e-06,
      "loss": 3.6415,
      "step": 28430
    },
    {
      "epoch": 0.2892557779947917,
      "grad_norm": 12.73385238647461,
      "learning_rate": 4.041361294984439e-06,
      "loss": 3.3467,
      "step": 28435
    },
    {
      "epoch": 0.289306640625,
      "grad_norm": 13.97321605682373,
      "learning_rate": 4.04104644098468e-06,
      "loss": 3.6723,
      "step": 28440
    },
    {
      "epoch": 0.2893575032552083,
      "grad_norm": 9.306679725646973,
      "learning_rate": 4.040731547557612e-06,
      "loss": 3.3643,
      "step": 28445
    },
    {
      "epoch": 0.2894083658854167,
      "grad_norm": 15.250093460083008,
      "learning_rate": 4.040416614711293e-06,
      "loss": 3.3897,
      "step": 28450
    },
    {
      "epoch": 0.289459228515625,
      "grad_norm": 12.86093807220459,
      "learning_rate": 4.04010164245378e-06,
      "loss": 3.2651,
      "step": 28455
    },
    {
      "epoch": 0.2895100911458333,
      "grad_norm": 17.20577621459961,
      "learning_rate": 4.039786630793131e-06,
      "loss": 3.1926,
      "step": 28460
    },
    {
      "epoch": 0.2895609537760417,
      "grad_norm": 10.533615112304688,
      "learning_rate": 4.039471579737407e-06,
      "loss": 2.9851,
      "step": 28465
    },
    {
      "epoch": 0.28961181640625,
      "grad_norm": 14.544201850891113,
      "learning_rate": 4.039156489294667e-06,
      "loss": 3.0023,
      "step": 28470
    },
    {
      "epoch": 0.2896626790364583,
      "grad_norm": 11.979805946350098,
      "learning_rate": 4.038841359472973e-06,
      "loss": 3.6672,
      "step": 28475
    },
    {
      "epoch": 0.2897135416666667,
      "grad_norm": 13.206119537353516,
      "learning_rate": 4.038526190280387e-06,
      "loss": 3.3212,
      "step": 28480
    },
    {
      "epoch": 0.289764404296875,
      "grad_norm": 15.055665969848633,
      "learning_rate": 4.038210981724975e-06,
      "loss": 3.6517,
      "step": 28485
    },
    {
      "epoch": 0.2898152669270833,
      "grad_norm": 10.177913665771484,
      "learning_rate": 4.037895733814797e-06,
      "loss": 3.0735,
      "step": 28490
    },
    {
      "epoch": 0.2898661295572917,
      "grad_norm": 12.19504451751709,
      "learning_rate": 4.037580446557923e-06,
      "loss": 3.3941,
      "step": 28495
    },
    {
      "epoch": 0.2899169921875,
      "grad_norm": 9.21094799041748,
      "learning_rate": 4.037265119962417e-06,
      "loss": 2.9818,
      "step": 28500
    },
    {
      "epoch": 0.2899678548177083,
      "grad_norm": 9.886246681213379,
      "learning_rate": 4.036949754036348e-06,
      "loss": 3.5824,
      "step": 28505
    },
    {
      "epoch": 0.2900187174479167,
      "grad_norm": 11.721346855163574,
      "learning_rate": 4.036634348787783e-06,
      "loss": 3.245,
      "step": 28510
    },
    {
      "epoch": 0.290069580078125,
      "grad_norm": 12.692378997802734,
      "learning_rate": 4.036318904224793e-06,
      "loss": 3.122,
      "step": 28515
    },
    {
      "epoch": 0.2901204427083333,
      "grad_norm": 14.110976219177246,
      "learning_rate": 4.036003420355447e-06,
      "loss": 3.3814,
      "step": 28520
    },
    {
      "epoch": 0.2901713053385417,
      "grad_norm": 10.645319938659668,
      "learning_rate": 4.035687897187819e-06,
      "loss": 3.4385,
      "step": 28525
    },
    {
      "epoch": 0.29022216796875,
      "grad_norm": 10.692234992980957,
      "learning_rate": 4.03537233472998e-06,
      "loss": 3.0013,
      "step": 28530
    },
    {
      "epoch": 0.2902730305989583,
      "grad_norm": 11.266454696655273,
      "learning_rate": 4.035056732990003e-06,
      "loss": 3.7197,
      "step": 28535
    },
    {
      "epoch": 0.2903238932291667,
      "grad_norm": 10.607219696044922,
      "learning_rate": 4.034741091975963e-06,
      "loss": 3.2842,
      "step": 28540
    },
    {
      "epoch": 0.290374755859375,
      "grad_norm": 11.192341804504395,
      "learning_rate": 4.034425411695937e-06,
      "loss": 3.2919,
      "step": 28545
    },
    {
      "epoch": 0.2904256184895833,
      "grad_norm": 12.919684410095215,
      "learning_rate": 4.0341096921580005e-06,
      "loss": 3.4981,
      "step": 28550
    },
    {
      "epoch": 0.2904764811197917,
      "grad_norm": 8.975446701049805,
      "learning_rate": 4.03379393337023e-06,
      "loss": 3.1235,
      "step": 28555
    },
    {
      "epoch": 0.29052734375,
      "grad_norm": 9.11768627166748,
      "learning_rate": 4.033478135340706e-06,
      "loss": 3.2628,
      "step": 28560
    },
    {
      "epoch": 0.2905782063802083,
      "grad_norm": 7.787058353424072,
      "learning_rate": 4.033162298077507e-06,
      "loss": 3.5017,
      "step": 28565
    },
    {
      "epoch": 0.2906290690104167,
      "grad_norm": 8.185754776000977,
      "learning_rate": 4.0328464215887145e-06,
      "loss": 3.0679,
      "step": 28570
    },
    {
      "epoch": 0.290679931640625,
      "grad_norm": 15.081027030944824,
      "learning_rate": 4.032530505882409e-06,
      "loss": 3.4288,
      "step": 28575
    },
    {
      "epoch": 0.2907307942708333,
      "grad_norm": 15.316882133483887,
      "learning_rate": 4.032214550966674e-06,
      "loss": 3.5306,
      "step": 28580
    },
    {
      "epoch": 0.2907816569010417,
      "grad_norm": 11.870295524597168,
      "learning_rate": 4.031898556849594e-06,
      "loss": 3.7683,
      "step": 28585
    },
    {
      "epoch": 0.29083251953125,
      "grad_norm": 10.215821266174316,
      "learning_rate": 4.0315825235392516e-06,
      "loss": 3.4146,
      "step": 28590
    },
    {
      "epoch": 0.2908833821614583,
      "grad_norm": 9.163251876831055,
      "learning_rate": 4.0312664510437325e-06,
      "loss": 3.5628,
      "step": 28595
    },
    {
      "epoch": 0.2909342447916667,
      "grad_norm": 11.067456245422363,
      "learning_rate": 4.030950339371125e-06,
      "loss": 3.2844,
      "step": 28600
    },
    {
      "epoch": 0.290985107421875,
      "grad_norm": 15.394378662109375,
      "learning_rate": 4.030634188529517e-06,
      "loss": 3.2655,
      "step": 28605
    },
    {
      "epoch": 0.2910359700520833,
      "grad_norm": 12.825260162353516,
      "learning_rate": 4.030317998526995e-06,
      "loss": 3.3786,
      "step": 28610
    },
    {
      "epoch": 0.2910868326822917,
      "grad_norm": 8.954317092895508,
      "learning_rate": 4.03000176937165e-06,
      "loss": 3.397,
      "step": 28615
    },
    {
      "epoch": 0.2911376953125,
      "grad_norm": 10.459576606750488,
      "learning_rate": 4.029685501071572e-06,
      "loss": 3.3017,
      "step": 28620
    },
    {
      "epoch": 0.2911885579427083,
      "grad_norm": 9.525602340698242,
      "learning_rate": 4.029369193634853e-06,
      "loss": 3.4694,
      "step": 28625
    },
    {
      "epoch": 0.2912394205729167,
      "grad_norm": 20.42050552368164,
      "learning_rate": 4.029052847069585e-06,
      "loss": 3.3009,
      "step": 28630
    },
    {
      "epoch": 0.291290283203125,
      "grad_norm": 12.305190086364746,
      "learning_rate": 4.028736461383864e-06,
      "loss": 3.5505,
      "step": 28635
    },
    {
      "epoch": 0.2913411458333333,
      "grad_norm": 10.575671195983887,
      "learning_rate": 4.028420036585782e-06,
      "loss": 3.5277,
      "step": 28640
    },
    {
      "epoch": 0.2913920084635417,
      "grad_norm": 11.413809776306152,
      "learning_rate": 4.0281035726834364e-06,
      "loss": 3.5491,
      "step": 28645
    },
    {
      "epoch": 0.29144287109375,
      "grad_norm": 15.26525592803955,
      "learning_rate": 4.027787069684922e-06,
      "loss": 3.4472,
      "step": 28650
    },
    {
      "epoch": 0.2914937337239583,
      "grad_norm": 13.700064659118652,
      "learning_rate": 4.0274705275983385e-06,
      "loss": 3.7096,
      "step": 28655
    },
    {
      "epoch": 0.2915445963541667,
      "grad_norm": 11.321172714233398,
      "learning_rate": 4.027153946431783e-06,
      "loss": 3.2004,
      "step": 28660
    },
    {
      "epoch": 0.291595458984375,
      "grad_norm": 11.454513549804688,
      "learning_rate": 4.026837326193357e-06,
      "loss": 3.331,
      "step": 28665
    },
    {
      "epoch": 0.2916463216145833,
      "grad_norm": 8.657747268676758,
      "learning_rate": 4.026520666891158e-06,
      "loss": 3.2785,
      "step": 28670
    },
    {
      "epoch": 0.2916971842447917,
      "grad_norm": 13.227458953857422,
      "learning_rate": 4.026203968533291e-06,
      "loss": 3.4965,
      "step": 28675
    },
    {
      "epoch": 0.291748046875,
      "grad_norm": 13.470231056213379,
      "learning_rate": 4.025887231127856e-06,
      "loss": 3.1858,
      "step": 28680
    },
    {
      "epoch": 0.2917989095052083,
      "grad_norm": 10.020124435424805,
      "learning_rate": 4.025570454682959e-06,
      "loss": 3.2221,
      "step": 28685
    },
    {
      "epoch": 0.2918497721354167,
      "grad_norm": 8.721534729003906,
      "learning_rate": 4.025253639206703e-06,
      "loss": 3.3351,
      "step": 28690
    },
    {
      "epoch": 0.291900634765625,
      "grad_norm": 7.433453559875488,
      "learning_rate": 4.024936784707194e-06,
      "loss": 3.0843,
      "step": 28695
    },
    {
      "epoch": 0.2919514973958333,
      "grad_norm": 7.928054332733154,
      "learning_rate": 4.02461989119254e-06,
      "loss": 3.2303,
      "step": 28700
    },
    {
      "epoch": 0.2920023600260417,
      "grad_norm": 14.633429527282715,
      "learning_rate": 4.0243029586708455e-06,
      "loss": 3.2592,
      "step": 28705
    },
    {
      "epoch": 0.29205322265625,
      "grad_norm": 11.753192901611328,
      "learning_rate": 4.023985987150222e-06,
      "loss": 3.2326,
      "step": 28710
    },
    {
      "epoch": 0.2921040852864583,
      "grad_norm": 18.38849449157715,
      "learning_rate": 4.023668976638778e-06,
      "loss": 3.6247,
      "step": 28715
    },
    {
      "epoch": 0.2921549479166667,
      "grad_norm": 7.707657337188721,
      "learning_rate": 4.023351927144625e-06,
      "loss": 3.3972,
      "step": 28720
    },
    {
      "epoch": 0.292205810546875,
      "grad_norm": 11.268938064575195,
      "learning_rate": 4.023034838675874e-06,
      "loss": 3.0399,
      "step": 28725
    },
    {
      "epoch": 0.2922566731770833,
      "grad_norm": 10.834413528442383,
      "learning_rate": 4.022717711240637e-06,
      "loss": 3.2737,
      "step": 28730
    },
    {
      "epoch": 0.2923075358072917,
      "grad_norm": 9.380146026611328,
      "learning_rate": 4.022400544847028e-06,
      "loss": 3.3967,
      "step": 28735
    },
    {
      "epoch": 0.2923583984375,
      "grad_norm": 9.512231826782227,
      "learning_rate": 4.022083339503164e-06,
      "loss": 2.9652,
      "step": 28740
    },
    {
      "epoch": 0.2924092610677083,
      "grad_norm": 13.55062484741211,
      "learning_rate": 4.021766095217157e-06,
      "loss": 2.9839,
      "step": 28745
    },
    {
      "epoch": 0.2924601236979167,
      "grad_norm": 11.842476844787598,
      "learning_rate": 4.021448811997125e-06,
      "loss": 3.5804,
      "step": 28750
    },
    {
      "epoch": 0.292510986328125,
      "grad_norm": 14.86185073852539,
      "learning_rate": 4.021131489851186e-06,
      "loss": 3.2041,
      "step": 28755
    },
    {
      "epoch": 0.2925618489583333,
      "grad_norm": 10.015793800354004,
      "learning_rate": 4.020814128787458e-06,
      "loss": 3.4947,
      "step": 28760
    },
    {
      "epoch": 0.2926127115885417,
      "grad_norm": 16.945362091064453,
      "learning_rate": 4.020496728814062e-06,
      "loss": 3.4086,
      "step": 28765
    },
    {
      "epoch": 0.29266357421875,
      "grad_norm": 14.16063404083252,
      "learning_rate": 4.020179289939117e-06,
      "loss": 3.505,
      "step": 28770
    },
    {
      "epoch": 0.2927144368489583,
      "grad_norm": 13.34571361541748,
      "learning_rate": 4.019861812170744e-06,
      "loss": 3.2015,
      "step": 28775
    },
    {
      "epoch": 0.2927652994791667,
      "grad_norm": 8.515129089355469,
      "learning_rate": 4.019544295517069e-06,
      "loss": 3.3871,
      "step": 28780
    },
    {
      "epoch": 0.292816162109375,
      "grad_norm": 10.501129150390625,
      "learning_rate": 4.019226739986212e-06,
      "loss": 3.1915,
      "step": 28785
    },
    {
      "epoch": 0.2928670247395833,
      "grad_norm": 15.085137367248535,
      "learning_rate": 4.0189091455862995e-06,
      "loss": 3.245,
      "step": 28790
    },
    {
      "epoch": 0.2929178873697917,
      "grad_norm": 12.195261001586914,
      "learning_rate": 4.018591512325456e-06,
      "loss": 3.7411,
      "step": 28795
    },
    {
      "epoch": 0.29296875,
      "grad_norm": 12.424592018127441,
      "learning_rate": 4.018273840211809e-06,
      "loss": 3.7522,
      "step": 28800
    },
    {
      "epoch": 0.2930196126302083,
      "grad_norm": 10.4650297164917,
      "learning_rate": 4.0179561292534855e-06,
      "loss": 3.5745,
      "step": 28805
    },
    {
      "epoch": 0.2930704752604167,
      "grad_norm": 8.884088516235352,
      "learning_rate": 4.017638379458614e-06,
      "loss": 3.3963,
      "step": 28810
    },
    {
      "epoch": 0.293121337890625,
      "grad_norm": 11.254535675048828,
      "learning_rate": 4.017320590835325e-06,
      "loss": 3.4626,
      "step": 28815
    },
    {
      "epoch": 0.2931722005208333,
      "grad_norm": 10.561436653137207,
      "learning_rate": 4.0170027633917474e-06,
      "loss": 3.747,
      "step": 28820
    },
    {
      "epoch": 0.2932230631510417,
      "grad_norm": 10.550532341003418,
      "learning_rate": 4.016684897136015e-06,
      "loss": 3.5104,
      "step": 28825
    },
    {
      "epoch": 0.29327392578125,
      "grad_norm": 13.52685546875,
      "learning_rate": 4.0163669920762574e-06,
      "loss": 3.0843,
      "step": 28830
    },
    {
      "epoch": 0.2933247884114583,
      "grad_norm": 15.23268985748291,
      "learning_rate": 4.016049048220611e-06,
      "loss": 3.21,
      "step": 28835
    },
    {
      "epoch": 0.2933756510416667,
      "grad_norm": 14.624536514282227,
      "learning_rate": 4.015731065577209e-06,
      "loss": 3.3289,
      "step": 28840
    },
    {
      "epoch": 0.293426513671875,
      "grad_norm": 9.752080917358398,
      "learning_rate": 4.015413044154186e-06,
      "loss": 3.5604,
      "step": 28845
    },
    {
      "epoch": 0.2934773763020833,
      "grad_norm": 13.152545928955078,
      "learning_rate": 4.01509498395968e-06,
      "loss": 3.1857,
      "step": 28850
    },
    {
      "epoch": 0.2935282389322917,
      "grad_norm": 12.336877822875977,
      "learning_rate": 4.014776885001828e-06,
      "loss": 3.0739,
      "step": 28855
    },
    {
      "epoch": 0.2935791015625,
      "grad_norm": 14.324713706970215,
      "learning_rate": 4.014458747288768e-06,
      "loss": 3.1812,
      "step": 28860
    },
    {
      "epoch": 0.2936299641927083,
      "grad_norm": 8.907907485961914,
      "learning_rate": 4.014140570828641e-06,
      "loss": 3.2912,
      "step": 28865
    },
    {
      "epoch": 0.2936808268229167,
      "grad_norm": 14.255101203918457,
      "learning_rate": 4.0138223556295855e-06,
      "loss": 3.4148,
      "step": 28870
    },
    {
      "epoch": 0.293731689453125,
      "grad_norm": 10.45189094543457,
      "learning_rate": 4.013504101699743e-06,
      "loss": 3.2952,
      "step": 28875
    },
    {
      "epoch": 0.2937825520833333,
      "grad_norm": 11.531671524047852,
      "learning_rate": 4.013185809047259e-06,
      "loss": 3.132,
      "step": 28880
    },
    {
      "epoch": 0.2938334147135417,
      "grad_norm": 9.269369125366211,
      "learning_rate": 4.012867477680273e-06,
      "loss": 3.3035,
      "step": 28885
    },
    {
      "epoch": 0.29388427734375,
      "grad_norm": 13.622968673706055,
      "learning_rate": 4.012549107606931e-06,
      "loss": 3.4165,
      "step": 28890
    },
    {
      "epoch": 0.2939351399739583,
      "grad_norm": 12.69628620147705,
      "learning_rate": 4.01223069883538e-06,
      "loss": 3.1884,
      "step": 28895
    },
    {
      "epoch": 0.2939860026041667,
      "grad_norm": 12.34487533569336,
      "learning_rate": 4.0119122513737645e-06,
      "loss": 3.3732,
      "step": 28900
    },
    {
      "epoch": 0.294036865234375,
      "grad_norm": 12.57749080657959,
      "learning_rate": 4.011593765230232e-06,
      "loss": 3.0678,
      "step": 28905
    },
    {
      "epoch": 0.2940877278645833,
      "grad_norm": 9.764747619628906,
      "learning_rate": 4.011275240412932e-06,
      "loss": 3.3889,
      "step": 28910
    },
    {
      "epoch": 0.2941385904947917,
      "grad_norm": 14.425663948059082,
      "learning_rate": 4.010956676930013e-06,
      "loss": 3.7294,
      "step": 28915
    },
    {
      "epoch": 0.294189453125,
      "grad_norm": 10.81959056854248,
      "learning_rate": 4.0106380747896245e-06,
      "loss": 3.803,
      "step": 28920
    },
    {
      "epoch": 0.2942403157552083,
      "grad_norm": 14.788504600524902,
      "learning_rate": 4.01031943399992e-06,
      "loss": 2.9475,
      "step": 28925
    },
    {
      "epoch": 0.2942911783854167,
      "grad_norm": 9.9393949508667,
      "learning_rate": 4.01000075456905e-06,
      "loss": 3.0866,
      "step": 28930
    },
    {
      "epoch": 0.294342041015625,
      "grad_norm": 8.403119087219238,
      "learning_rate": 4.009682036505169e-06,
      "loss": 3.4695,
      "step": 28935
    },
    {
      "epoch": 0.2943929036458333,
      "grad_norm": 9.470560073852539,
      "learning_rate": 4.009363279816431e-06,
      "loss": 3.2326,
      "step": 28940
    },
    {
      "epoch": 0.2944437662760417,
      "grad_norm": 9.08181381225586,
      "learning_rate": 4.009044484510991e-06,
      "loss": 3.6787,
      "step": 28945
    },
    {
      "epoch": 0.29449462890625,
      "grad_norm": 9.185422897338867,
      "learning_rate": 4.0087256505970065e-06,
      "loss": 3.2821,
      "step": 28950
    },
    {
      "epoch": 0.2945454915364583,
      "grad_norm": 16.905046463012695,
      "learning_rate": 4.0084067780826326e-06,
      "loss": 3.5609,
      "step": 28955
    },
    {
      "epoch": 0.2945963541666667,
      "grad_norm": 15.509016036987305,
      "learning_rate": 4.008087866976029e-06,
      "loss": 3.1874,
      "step": 28960
    },
    {
      "epoch": 0.294647216796875,
      "grad_norm": 14.498684883117676,
      "learning_rate": 4.007768917285356e-06,
      "loss": 3.7175,
      "step": 28965
    },
    {
      "epoch": 0.2946980794270833,
      "grad_norm": 11.822076797485352,
      "learning_rate": 4.007449929018772e-06,
      "loss": 3.1827,
      "step": 28970
    },
    {
      "epoch": 0.2947489420572917,
      "grad_norm": 14.711019515991211,
      "learning_rate": 4.007130902184438e-06,
      "loss": 3.1797,
      "step": 28975
    },
    {
      "epoch": 0.2947998046875,
      "grad_norm": 9.385677337646484,
      "learning_rate": 4.006811836790518e-06,
      "loss": 3.3217,
      "step": 28980
    },
    {
      "epoch": 0.2948506673177083,
      "grad_norm": 15.740737915039062,
      "learning_rate": 4.006492732845174e-06,
      "loss": 3.4122,
      "step": 28985
    },
    {
      "epoch": 0.2949015299479167,
      "grad_norm": 15.037389755249023,
      "learning_rate": 4.00617359035657e-06,
      "loss": 3.2279,
      "step": 28990
    },
    {
      "epoch": 0.294952392578125,
      "grad_norm": 880.1937255859375,
      "learning_rate": 4.005854409332872e-06,
      "loss": 3.9876,
      "step": 28995
    },
    {
      "epoch": 0.2950032552083333,
      "grad_norm": 12.129514694213867,
      "learning_rate": 4.0055351897822464e-06,
      "loss": 3.2343,
      "step": 29000
    },
    {
      "epoch": 0.2950541178385417,
      "grad_norm": 12.854998588562012,
      "learning_rate": 4.00521593171286e-06,
      "loss": 3.4128,
      "step": 29005
    },
    {
      "epoch": 0.29510498046875,
      "grad_norm": 12.726030349731445,
      "learning_rate": 4.00489663513288e-06,
      "loss": 3.4242,
      "step": 29010
    },
    {
      "epoch": 0.2951558430989583,
      "grad_norm": 9.434267044067383,
      "learning_rate": 4.004577300050476e-06,
      "loss": 3.6865,
      "step": 29015
    },
    {
      "epoch": 0.2952067057291667,
      "grad_norm": 14.372798919677734,
      "learning_rate": 4.004257926473819e-06,
      "loss": 3.5117,
      "step": 29020
    },
    {
      "epoch": 0.295257568359375,
      "grad_norm": 9.412242889404297,
      "learning_rate": 4.00393851441108e-06,
      "loss": 3.2831,
      "step": 29025
    },
    {
      "epoch": 0.2953084309895833,
      "grad_norm": 17.29656982421875,
      "learning_rate": 4.00361906387043e-06,
      "loss": 3.8513,
      "step": 29030
    },
    {
      "epoch": 0.2953592936197917,
      "grad_norm": 12.378039360046387,
      "learning_rate": 4.003299574860042e-06,
      "loss": 3.2805,
      "step": 29035
    },
    {
      "epoch": 0.29541015625,
      "grad_norm": 9.60687255859375,
      "learning_rate": 4.002980047388092e-06,
      "loss": 3.25,
      "step": 29040
    },
    {
      "epoch": 0.2954610188802083,
      "grad_norm": 8.89541244506836,
      "learning_rate": 4.002660481462753e-06,
      "loss": 3.3592,
      "step": 29045
    },
    {
      "epoch": 0.2955118815104167,
      "grad_norm": 8.48711109161377,
      "learning_rate": 4.002340877092202e-06,
      "loss": 3.995,
      "step": 29050
    },
    {
      "epoch": 0.295562744140625,
      "grad_norm": 16.27423095703125,
      "learning_rate": 4.0020212342846155e-06,
      "loss": 3.4849,
      "step": 29055
    },
    {
      "epoch": 0.2956136067708333,
      "grad_norm": 15.46958065032959,
      "learning_rate": 4.001701553048172e-06,
      "loss": 3.3011,
      "step": 29060
    },
    {
      "epoch": 0.2956644694010417,
      "grad_norm": 10.370869636535645,
      "learning_rate": 4.0013818333910505e-06,
      "loss": 3.3733,
      "step": 29065
    },
    {
      "epoch": 0.29571533203125,
      "grad_norm": 16.184940338134766,
      "learning_rate": 4.00106207532143e-06,
      "loss": 4.0936,
      "step": 29070
    },
    {
      "epoch": 0.2957661946614583,
      "grad_norm": 12.648850440979004,
      "learning_rate": 4.000742278847492e-06,
      "loss": 3.0688,
      "step": 29075
    },
    {
      "epoch": 0.2958170572916667,
      "grad_norm": 8.855330467224121,
      "learning_rate": 4.0004224439774185e-06,
      "loss": 3.2634,
      "step": 29080
    },
    {
      "epoch": 0.295867919921875,
      "grad_norm": 16.087480545043945,
      "learning_rate": 4.000102570719393e-06,
      "loss": 3.2605,
      "step": 29085
    },
    {
      "epoch": 0.2959187825520833,
      "grad_norm": 13.387033462524414,
      "learning_rate": 3.999782659081598e-06,
      "loss": 3.2089,
      "step": 29090
    },
    {
      "epoch": 0.2959696451822917,
      "grad_norm": 12.484994888305664,
      "learning_rate": 3.99946270907222e-06,
      "loss": 3.3006,
      "step": 29095
    },
    {
      "epoch": 0.2960205078125,
      "grad_norm": 14.613950729370117,
      "learning_rate": 3.999142720699444e-06,
      "loss": 3.2682,
      "step": 29100
    },
    {
      "epoch": 0.2960713704427083,
      "grad_norm": 11.924738883972168,
      "learning_rate": 3.998822693971456e-06,
      "loss": 3.3534,
      "step": 29105
    },
    {
      "epoch": 0.2961222330729167,
      "grad_norm": 11.536297798156738,
      "learning_rate": 3.998502628896446e-06,
      "loss": 3.3467,
      "step": 29110
    },
    {
      "epoch": 0.296173095703125,
      "grad_norm": 11.012533187866211,
      "learning_rate": 3.9981825254826e-06,
      "loss": 3.0033,
      "step": 29115
    },
    {
      "epoch": 0.2962239583333333,
      "grad_norm": 8.716772079467773,
      "learning_rate": 3.99786238373811e-06,
      "loss": 3.2317,
      "step": 29120
    },
    {
      "epoch": 0.2962748209635417,
      "grad_norm": 11.559467315673828,
      "learning_rate": 3.997542203671165e-06,
      "loss": 3.5196,
      "step": 29125
    },
    {
      "epoch": 0.29632568359375,
      "grad_norm": 13.896491050720215,
      "learning_rate": 3.997221985289959e-06,
      "loss": 3.5077,
      "step": 29130
    },
    {
      "epoch": 0.2963765462239583,
      "grad_norm": 12.800115585327148,
      "learning_rate": 3.996901728602683e-06,
      "loss": 3.2497,
      "step": 29135
    },
    {
      "epoch": 0.2964274088541667,
      "grad_norm": 16.33082389831543,
      "learning_rate": 3.996581433617531e-06,
      "loss": 3.2995,
      "step": 29140
    },
    {
      "epoch": 0.296478271484375,
      "grad_norm": 11.650870323181152,
      "learning_rate": 3.996261100342697e-06,
      "loss": 3.3726,
      "step": 29145
    },
    {
      "epoch": 0.2965291341145833,
      "grad_norm": 9.717093467712402,
      "learning_rate": 3.995940728786378e-06,
      "loss": 3.4764,
      "step": 29150
    },
    {
      "epoch": 0.2965799967447917,
      "grad_norm": 10.83946704864502,
      "learning_rate": 3.99562031895677e-06,
      "loss": 3.3277,
      "step": 29155
    },
    {
      "epoch": 0.296630859375,
      "grad_norm": 12.137735366821289,
      "learning_rate": 3.9952998708620714e-06,
      "loss": 3.5005,
      "step": 29160
    },
    {
      "epoch": 0.2966817220052083,
      "grad_norm": 9.967763900756836,
      "learning_rate": 3.994979384510479e-06,
      "loss": 3.1285,
      "step": 29165
    },
    {
      "epoch": 0.2967325846354167,
      "grad_norm": 10.213086128234863,
      "learning_rate": 3.994658859910193e-06,
      "loss": 3.1881,
      "step": 29170
    },
    {
      "epoch": 0.296783447265625,
      "grad_norm": 12.435432434082031,
      "learning_rate": 3.9943382970694155e-06,
      "loss": 3.5875,
      "step": 29175
    },
    {
      "epoch": 0.2968343098958333,
      "grad_norm": 9.136934280395508,
      "learning_rate": 3.994017695996346e-06,
      "loss": 3.713,
      "step": 29180
    },
    {
      "epoch": 0.2968851725260417,
      "grad_norm": 13.187419891357422,
      "learning_rate": 3.993697056699188e-06,
      "loss": 3.2372,
      "step": 29185
    },
    {
      "epoch": 0.29693603515625,
      "grad_norm": 7.720186233520508,
      "learning_rate": 3.993376379186145e-06,
      "loss": 3.278,
      "step": 29190
    },
    {
      "epoch": 0.2969868977864583,
      "grad_norm": 9.980789184570312,
      "learning_rate": 3.993055663465422e-06,
      "loss": 3.4021,
      "step": 29195
    },
    {
      "epoch": 0.2970377604166667,
      "grad_norm": 14.341978073120117,
      "learning_rate": 3.9927349095452235e-06,
      "loss": 3.3752,
      "step": 29200
    },
    {
      "epoch": 0.297088623046875,
      "grad_norm": 9.22407341003418,
      "learning_rate": 3.992414117433756e-06,
      "loss": 3.1495,
      "step": 29205
    },
    {
      "epoch": 0.2971394856770833,
      "grad_norm": 14.73730754852295,
      "learning_rate": 3.992093287139226e-06,
      "loss": 3.2453,
      "step": 29210
    },
    {
      "epoch": 0.2971903483072917,
      "grad_norm": 12.117033004760742,
      "learning_rate": 3.991772418669844e-06,
      "loss": 3.467,
      "step": 29215
    },
    {
      "epoch": 0.2972412109375,
      "grad_norm": 12.779738426208496,
      "learning_rate": 3.991451512033818e-06,
      "loss": 3.425,
      "step": 29220
    },
    {
      "epoch": 0.2972920735677083,
      "grad_norm": 12.695642471313477,
      "learning_rate": 3.991130567239359e-06,
      "loss": 3.3072,
      "step": 29225
    },
    {
      "epoch": 0.2973429361979167,
      "grad_norm": 14.683784484863281,
      "learning_rate": 3.990809584294677e-06,
      "loss": 3.7767,
      "step": 29230
    },
    {
      "epoch": 0.297393798828125,
      "grad_norm": 14.795294761657715,
      "learning_rate": 3.990488563207986e-06,
      "loss": 3.3255,
      "step": 29235
    },
    {
      "epoch": 0.2974446614583333,
      "grad_norm": 12.686258316040039,
      "learning_rate": 3.990167503987498e-06,
      "loss": 3.7086,
      "step": 29240
    },
    {
      "epoch": 0.2974955240885417,
      "grad_norm": 8.185772895812988,
      "learning_rate": 3.989846406641428e-06,
      "loss": 3.4043,
      "step": 29245
    },
    {
      "epoch": 0.29754638671875,
      "grad_norm": 10.710786819458008,
      "learning_rate": 3.98952527117799e-06,
      "loss": 3.1477,
      "step": 29250
    },
    {
      "epoch": 0.2975972493489583,
      "grad_norm": 11.759841918945312,
      "learning_rate": 3.989204097605402e-06,
      "loss": 3.4075,
      "step": 29255
    },
    {
      "epoch": 0.2976481119791667,
      "grad_norm": 13.554819107055664,
      "learning_rate": 3.98888288593188e-06,
      "loss": 3.1757,
      "step": 29260
    },
    {
      "epoch": 0.297698974609375,
      "grad_norm": 12.05898666381836,
      "learning_rate": 3.988561636165641e-06,
      "loss": 3.0917,
      "step": 29265
    },
    {
      "epoch": 0.2977498372395833,
      "grad_norm": 11.725159645080566,
      "learning_rate": 3.988240348314907e-06,
      "loss": 3.2702,
      "step": 29270
    },
    {
      "epoch": 0.2978006998697917,
      "grad_norm": 15.247992515563965,
      "learning_rate": 3.987919022387895e-06,
      "loss": 3.6453,
      "step": 29275
    },
    {
      "epoch": 0.2978515625,
      "grad_norm": 14.956148147583008,
      "learning_rate": 3.987597658392828e-06,
      "loss": 3.6442,
      "step": 29280
    },
    {
      "epoch": 0.2979024251302083,
      "grad_norm": 13.976456642150879,
      "learning_rate": 3.987276256337928e-06,
      "loss": 3.8726,
      "step": 29285
    },
    {
      "epoch": 0.2979532877604167,
      "grad_norm": 13.797755241394043,
      "learning_rate": 3.9869548162314176e-06,
      "loss": 3.4543,
      "step": 29290
    },
    {
      "epoch": 0.298004150390625,
      "grad_norm": 9.575790405273438,
      "learning_rate": 3.98663333808152e-06,
      "loss": 3.2062,
      "step": 29295
    },
    {
      "epoch": 0.2980550130208333,
      "grad_norm": 12.933478355407715,
      "learning_rate": 3.9863118218964606e-06,
      "loss": 3.2992,
      "step": 29300
    },
    {
      "epoch": 0.2981058756510417,
      "grad_norm": 11.104291915893555,
      "learning_rate": 3.985990267684466e-06,
      "loss": 2.822,
      "step": 29305
    },
    {
      "epoch": 0.29815673828125,
      "grad_norm": 7.797368049621582,
      "learning_rate": 3.985668675453763e-06,
      "loss": 3.5728,
      "step": 29310
    },
    {
      "epoch": 0.2982076009114583,
      "grad_norm": 13.247109413146973,
      "learning_rate": 3.985347045212578e-06,
      "loss": 3.2314,
      "step": 29315
    },
    {
      "epoch": 0.2982584635416667,
      "grad_norm": 14.105443000793457,
      "learning_rate": 3.985025376969143e-06,
      "loss": 3.4454,
      "step": 29320
    },
    {
      "epoch": 0.298309326171875,
      "grad_norm": 15.554372787475586,
      "learning_rate": 3.984703670731683e-06,
      "loss": 3.7341,
      "step": 29325
    },
    {
      "epoch": 0.2983601888020833,
      "grad_norm": 9.164495468139648,
      "learning_rate": 3.9843819265084334e-06,
      "loss": 3.2357,
      "step": 29330
    },
    {
      "epoch": 0.2984110514322917,
      "grad_norm": 11.580939292907715,
      "learning_rate": 3.984060144307623e-06,
      "loss": 3.3563,
      "step": 29335
    },
    {
      "epoch": 0.2984619140625,
      "grad_norm": 8.120573043823242,
      "learning_rate": 3.983738324137486e-06,
      "loss": 3.5399,
      "step": 29340
    },
    {
      "epoch": 0.2985127766927083,
      "grad_norm": 9.155670166015625,
      "learning_rate": 3.983416466006255e-06,
      "loss": 3.4847,
      "step": 29345
    },
    {
      "epoch": 0.2985636393229167,
      "grad_norm": 17.27814483642578,
      "learning_rate": 3.9830945699221674e-06,
      "loss": 3.273,
      "step": 29350
    },
    {
      "epoch": 0.298614501953125,
      "grad_norm": 15.654536247253418,
      "learning_rate": 3.982772635893456e-06,
      "loss": 3.5948,
      "step": 29355
    },
    {
      "epoch": 0.2986653645833333,
      "grad_norm": 10.65638256072998,
      "learning_rate": 3.9824506639283564e-06,
      "loss": 3.2658,
      "step": 29360
    },
    {
      "epoch": 0.2987162272135417,
      "grad_norm": 10.578728675842285,
      "learning_rate": 3.98212865403511e-06,
      "loss": 3.1781,
      "step": 29365
    },
    {
      "epoch": 0.29876708984375,
      "grad_norm": 7.720913410186768,
      "learning_rate": 3.981806606221952e-06,
      "loss": 3.4242,
      "step": 29370
    },
    {
      "epoch": 0.2988179524739583,
      "grad_norm": 10.320748329162598,
      "learning_rate": 3.981484520497123e-06,
      "loss": 3.3708,
      "step": 29375
    },
    {
      "epoch": 0.2988688151041667,
      "grad_norm": 15.515741348266602,
      "learning_rate": 3.981162396868865e-06,
      "loss": 3.2143,
      "step": 29380
    },
    {
      "epoch": 0.298919677734375,
      "grad_norm": 14.272456169128418,
      "learning_rate": 3.980840235345417e-06,
      "loss": 3.2696,
      "step": 29385
    },
    {
      "epoch": 0.2989705403645833,
      "grad_norm": 10.035221099853516,
      "learning_rate": 3.980518035935023e-06,
      "loss": 3.7437,
      "step": 29390
    },
    {
      "epoch": 0.2990214029947917,
      "grad_norm": 10.790133476257324,
      "learning_rate": 3.980195798645926e-06,
      "loss": 3.1117,
      "step": 29395
    },
    {
      "epoch": 0.299072265625,
      "grad_norm": 17.080608367919922,
      "learning_rate": 3.979873523486371e-06,
      "loss": 3.6969,
      "step": 29400
    },
    {
      "epoch": 0.2991231282552083,
      "grad_norm": 12.20949935913086,
      "learning_rate": 3.979551210464602e-06,
      "loss": 4.0681,
      "step": 29405
    },
    {
      "epoch": 0.2991739908854167,
      "grad_norm": 11.205544471740723,
      "learning_rate": 3.9792288595888655e-06,
      "loss": 3.3577,
      "step": 29410
    },
    {
      "epoch": 0.299224853515625,
      "grad_norm": 14.382440567016602,
      "learning_rate": 3.978906470867409e-06,
      "loss": 3.1179,
      "step": 29415
    },
    {
      "epoch": 0.2992757161458333,
      "grad_norm": 10.04643726348877,
      "learning_rate": 3.978584044308482e-06,
      "loss": 3.174,
      "step": 29420
    },
    {
      "epoch": 0.2993265787760417,
      "grad_norm": 12.206974029541016,
      "learning_rate": 3.978261579920333e-06,
      "loss": 3.5778,
      "step": 29425
    },
    {
      "epoch": 0.29937744140625,
      "grad_norm": 9.708235740661621,
      "learning_rate": 3.977939077711211e-06,
      "loss": 3.1949,
      "step": 29430
    },
    {
      "epoch": 0.2994283040364583,
      "grad_norm": 10.987796783447266,
      "learning_rate": 3.977616537689368e-06,
      "loss": 3.5749,
      "step": 29435
    },
    {
      "epoch": 0.2994791666666667,
      "grad_norm": 8.089970588684082,
      "learning_rate": 3.977293959863057e-06,
      "loss": 3.5678,
      "step": 29440
    },
    {
      "epoch": 0.299530029296875,
      "grad_norm": 14.080187797546387,
      "learning_rate": 3.97697134424053e-06,
      "loss": 3.7299,
      "step": 29445
    },
    {
      "epoch": 0.2995808919270833,
      "grad_norm": 14.45607852935791,
      "learning_rate": 3.97664869083004e-06,
      "loss": 3.6074,
      "step": 29450
    },
    {
      "epoch": 0.2996317545572917,
      "grad_norm": 9.266926765441895,
      "learning_rate": 3.976325999639844e-06,
      "loss": 3.5337,
      "step": 29455
    },
    {
      "epoch": 0.2996826171875,
      "grad_norm": 8.380191802978516,
      "learning_rate": 3.976003270678197e-06,
      "loss": 3.3772,
      "step": 29460
    },
    {
      "epoch": 0.2997334798177083,
      "grad_norm": 9.378839492797852,
      "learning_rate": 3.975680503953358e-06,
      "loss": 3.5624,
      "step": 29465
    },
    {
      "epoch": 0.2997843424479167,
      "grad_norm": 7.663854598999023,
      "learning_rate": 3.97535769947358e-06,
      "loss": 3.4436,
      "step": 29470
    },
    {
      "epoch": 0.299835205078125,
      "grad_norm": 8.099380493164062,
      "learning_rate": 3.9750348572471276e-06,
      "loss": 3.0022,
      "step": 29475
    },
    {
      "epoch": 0.2998860677083333,
      "grad_norm": 10.462038040161133,
      "learning_rate": 3.974711977282256e-06,
      "loss": 3.0147,
      "step": 29480
    },
    {
      "epoch": 0.2999369303385417,
      "grad_norm": 8.75016975402832,
      "learning_rate": 3.974389059587229e-06,
      "loss": 3.4312,
      "step": 29485
    },
    {
      "epoch": 0.29998779296875,
      "grad_norm": 15.210886001586914,
      "learning_rate": 3.9740661041703075e-06,
      "loss": 3.013,
      "step": 29490
    },
    {
      "epoch": 0.3000386555989583,
      "grad_norm": 10.11583423614502,
      "learning_rate": 3.973743111039754e-06,
      "loss": 3.249,
      "step": 29495
    },
    {
      "epoch": 0.3000895182291667,
      "grad_norm": 16.003278732299805,
      "learning_rate": 3.973420080203832e-06,
      "loss": 3.3855,
      "step": 29500
    },
    {
      "epoch": 0.300140380859375,
      "grad_norm": 9.147961616516113,
      "learning_rate": 3.9730970116708065e-06,
      "loss": 3.2473,
      "step": 29505
    },
    {
      "epoch": 0.3001912434895833,
      "grad_norm": 9.03951358795166,
      "learning_rate": 3.9727739054489436e-06,
      "loss": 3.0429,
      "step": 29510
    },
    {
      "epoch": 0.3002421061197917,
      "grad_norm": 12.876129150390625,
      "learning_rate": 3.972450761546509e-06,
      "loss": 3.327,
      "step": 29515
    },
    {
      "epoch": 0.30029296875,
      "grad_norm": 11.136457443237305,
      "learning_rate": 3.972127579971771e-06,
      "loss": 3.2739,
      "step": 29520
    },
    {
      "epoch": 0.3003438313802083,
      "grad_norm": 13.348973274230957,
      "learning_rate": 3.971804360732997e-06,
      "loss": 3.2297,
      "step": 29525
    },
    {
      "epoch": 0.3003946940104167,
      "grad_norm": 10.658791542053223,
      "learning_rate": 3.971481103838459e-06,
      "loss": 3.3988,
      "step": 29530
    },
    {
      "epoch": 0.300445556640625,
      "grad_norm": 12.942093849182129,
      "learning_rate": 3.971157809296424e-06,
      "loss": 3.246,
      "step": 29535
    },
    {
      "epoch": 0.3004964192708333,
      "grad_norm": 8.947063446044922,
      "learning_rate": 3.970834477115167e-06,
      "loss": 3.2981,
      "step": 29540
    },
    {
      "epoch": 0.3005472819010417,
      "grad_norm": 14.3213472366333,
      "learning_rate": 3.970511107302956e-06,
      "loss": 3.0184,
      "step": 29545
    },
    {
      "epoch": 0.30059814453125,
      "grad_norm": 11.473852157592773,
      "learning_rate": 3.970187699868069e-06,
      "loss": 3.3741,
      "step": 29550
    },
    {
      "epoch": 0.3006490071614583,
      "grad_norm": 16.973127365112305,
      "learning_rate": 3.969864254818777e-06,
      "loss": 3.4917,
      "step": 29555
    },
    {
      "epoch": 0.3006998697916667,
      "grad_norm": 10.866402626037598,
      "learning_rate": 3.969540772163357e-06,
      "loss": 3.0928,
      "step": 29560
    },
    {
      "epoch": 0.300750732421875,
      "grad_norm": 12.010110855102539,
      "learning_rate": 3.969217251910084e-06,
      "loss": 3.0374,
      "step": 29565
    },
    {
      "epoch": 0.3008015950520833,
      "grad_norm": 15.12334156036377,
      "learning_rate": 3.968893694067236e-06,
      "loss": 3.4492,
      "step": 29570
    },
    {
      "epoch": 0.3008524576822917,
      "grad_norm": 12.803681373596191,
      "learning_rate": 3.968570098643092e-06,
      "loss": 3.2817,
      "step": 29575
    },
    {
      "epoch": 0.3009033203125,
      "grad_norm": 8.338359832763672,
      "learning_rate": 3.968246465645929e-06,
      "loss": 3.0263,
      "step": 29580
    },
    {
      "epoch": 0.3009541829427083,
      "grad_norm": 15.683090209960938,
      "learning_rate": 3.967922795084029e-06,
      "loss": 3.5276,
      "step": 29585
    },
    {
      "epoch": 0.3010050455729167,
      "grad_norm": 11.639715194702148,
      "learning_rate": 3.967599086965671e-06,
      "loss": 3.413,
      "step": 29590
    },
    {
      "epoch": 0.301055908203125,
      "grad_norm": 12.234172821044922,
      "learning_rate": 3.967275341299139e-06,
      "loss": 2.8704,
      "step": 29595
    },
    {
      "epoch": 0.3011067708333333,
      "grad_norm": 8.11499309539795,
      "learning_rate": 3.9669515580927155e-06,
      "loss": 3.2947,
      "step": 29600
    },
    {
      "epoch": 0.3011576334635417,
      "grad_norm": 9.019293785095215,
      "learning_rate": 3.9666277373546844e-06,
      "loss": 3.1113,
      "step": 29605
    },
    {
      "epoch": 0.30120849609375,
      "grad_norm": 14.087760925292969,
      "learning_rate": 3.966303879093329e-06,
      "loss": 3.0913,
      "step": 29610
    },
    {
      "epoch": 0.3012593587239583,
      "grad_norm": 10.682214736938477,
      "learning_rate": 3.9659799833169374e-06,
      "loss": 3.7827,
      "step": 29615
    },
    {
      "epoch": 0.3013102213541667,
      "grad_norm": 8.64210033416748,
      "learning_rate": 3.9656560500337945e-06,
      "loss": 3.07,
      "step": 29620
    },
    {
      "epoch": 0.301361083984375,
      "grad_norm": 11.499073028564453,
      "learning_rate": 3.96533207925219e-06,
      "loss": 3.3049,
      "step": 29625
    },
    {
      "epoch": 0.3014119466145833,
      "grad_norm": 14.486063003540039,
      "learning_rate": 3.965008070980411e-06,
      "loss": 3.6134,
      "step": 29630
    },
    {
      "epoch": 0.3014628092447917,
      "grad_norm": 11.338687896728516,
      "learning_rate": 3.9646840252267485e-06,
      "loss": 3.2003,
      "step": 29635
    },
    {
      "epoch": 0.301513671875,
      "grad_norm": 9.439972877502441,
      "learning_rate": 3.964359941999491e-06,
      "loss": 3.3831,
      "step": 29640
    },
    {
      "epoch": 0.3015645345052083,
      "grad_norm": 10.964188575744629,
      "learning_rate": 3.964035821306933e-06,
      "loss": 3.5014,
      "step": 29645
    },
    {
      "epoch": 0.3016153971354167,
      "grad_norm": 14.766501426696777,
      "learning_rate": 3.963711663157365e-06,
      "loss": 3.0886,
      "step": 29650
    },
    {
      "epoch": 0.301666259765625,
      "grad_norm": 11.056827545166016,
      "learning_rate": 3.96338746755908e-06,
      "loss": 3.1309,
      "step": 29655
    },
    {
      "epoch": 0.3017171223958333,
      "grad_norm": 13.033686637878418,
      "learning_rate": 3.963063234520374e-06,
      "loss": 3.3055,
      "step": 29660
    },
    {
      "epoch": 0.3017679850260417,
      "grad_norm": 15.789066314697266,
      "learning_rate": 3.962738964049543e-06,
      "loss": 2.9897,
      "step": 29665
    },
    {
      "epoch": 0.30181884765625,
      "grad_norm": 13.167356491088867,
      "learning_rate": 3.962414656154882e-06,
      "loss": 3.3225,
      "step": 29670
    },
    {
      "epoch": 0.3018697102864583,
      "grad_norm": 12.949690818786621,
      "learning_rate": 3.962090310844687e-06,
      "loss": 3.186,
      "step": 29675
    },
    {
      "epoch": 0.3019205729166667,
      "grad_norm": 16.214996337890625,
      "learning_rate": 3.96176592812726e-06,
      "loss": 3.8273,
      "step": 29680
    },
    {
      "epoch": 0.301971435546875,
      "grad_norm": 16.74712371826172,
      "learning_rate": 3.961441508010897e-06,
      "loss": 3.3157,
      "step": 29685
    },
    {
      "epoch": 0.3020222981770833,
      "grad_norm": 12.501587867736816,
      "learning_rate": 3.9611170505039e-06,
      "loss": 2.9707,
      "step": 29690
    },
    {
      "epoch": 0.3020731608072917,
      "grad_norm": 13.367961883544922,
      "learning_rate": 3.960792555614569e-06,
      "loss": 3.5666,
      "step": 29695
    },
    {
      "epoch": 0.3021240234375,
      "grad_norm": 10.78051471710205,
      "learning_rate": 3.960468023351207e-06,
      "loss": 3.4304,
      "step": 29700
    },
    {
      "epoch": 0.3021748860677083,
      "grad_norm": 13.978754997253418,
      "learning_rate": 3.960143453722117e-06,
      "loss": 3.1024,
      "step": 29705
    },
    {
      "epoch": 0.3022257486979167,
      "grad_norm": 13.29320240020752,
      "learning_rate": 3.959818846735602e-06,
      "loss": 3.649,
      "step": 29710
    },
    {
      "epoch": 0.302276611328125,
      "grad_norm": 7.485118389129639,
      "learning_rate": 3.959494202399969e-06,
      "loss": 3.5782,
      "step": 29715
    },
    {
      "epoch": 0.3023274739583333,
      "grad_norm": 11.284222602844238,
      "learning_rate": 3.9591695207235225e-06,
      "loss": 3.4861,
      "step": 29720
    },
    {
      "epoch": 0.3023783365885417,
      "grad_norm": 17.11857032775879,
      "learning_rate": 3.958844801714569e-06,
      "loss": 3.1366,
      "step": 29725
    },
    {
      "epoch": 0.30242919921875,
      "grad_norm": 13.37563705444336,
      "learning_rate": 3.958520045381417e-06,
      "loss": 3.2703,
      "step": 29730
    },
    {
      "epoch": 0.3024800618489583,
      "grad_norm": 8.232620239257812,
      "learning_rate": 3.958195251732376e-06,
      "loss": 3.2489,
      "step": 29735
    },
    {
      "epoch": 0.3025309244791667,
      "grad_norm": 10.953129768371582,
      "learning_rate": 3.957870420775756e-06,
      "loss": 3.5771,
      "step": 29740
    },
    {
      "epoch": 0.302581787109375,
      "grad_norm": 16.61271095275879,
      "learning_rate": 3.9575455525198655e-06,
      "loss": 3.2988,
      "step": 29745
    },
    {
      "epoch": 0.3026326497395833,
      "grad_norm": 9.787528991699219,
      "learning_rate": 3.957220646973018e-06,
      "loss": 3.1724,
      "step": 29750
    },
    {
      "epoch": 0.3026835123697917,
      "grad_norm": 10.20058822631836,
      "learning_rate": 3.956895704143525e-06,
      "loss": 3.4975,
      "step": 29755
    },
    {
      "epoch": 0.302734375,
      "grad_norm": 10.130928039550781,
      "learning_rate": 3.956570724039701e-06,
      "loss": 3.2884,
      "step": 29760
    },
    {
      "epoch": 0.3027852376302083,
      "grad_norm": 10.151378631591797,
      "learning_rate": 3.9562457066698615e-06,
      "loss": 3.2238,
      "step": 29765
    },
    {
      "epoch": 0.3028361002604167,
      "grad_norm": 10.487987518310547,
      "learning_rate": 3.955920652042319e-06,
      "loss": 3.35,
      "step": 29770
    },
    {
      "epoch": 0.302886962890625,
      "grad_norm": 10.246733665466309,
      "learning_rate": 3.955595560165392e-06,
      "loss": 4.0199,
      "step": 29775
    },
    {
      "epoch": 0.3029378255208333,
      "grad_norm": 11.14050006866455,
      "learning_rate": 3.955270431047398e-06,
      "loss": 3.4622,
      "step": 29780
    },
    {
      "epoch": 0.3029886881510417,
      "grad_norm": 15.32699966430664,
      "learning_rate": 3.954945264696656e-06,
      "loss": 3.5683,
      "step": 29785
    },
    {
      "epoch": 0.30303955078125,
      "grad_norm": 9.070558547973633,
      "learning_rate": 3.954620061121482e-06,
      "loss": 3.5751,
      "step": 29790
    },
    {
      "epoch": 0.3030904134114583,
      "grad_norm": 12.999809265136719,
      "learning_rate": 3.9542948203302e-06,
      "loss": 3.3605,
      "step": 29795
    },
    {
      "epoch": 0.3031412760416667,
      "grad_norm": 10.635300636291504,
      "learning_rate": 3.953969542331129e-06,
      "loss": 3.2415,
      "step": 29800
    },
    {
      "epoch": 0.303192138671875,
      "grad_norm": 15.898357391357422,
      "learning_rate": 3.953644227132593e-06,
      "loss": 3.1742,
      "step": 29805
    },
    {
      "epoch": 0.3032430013020833,
      "grad_norm": 11.910710334777832,
      "learning_rate": 3.9533188747429135e-06,
      "loss": 3.3774,
      "step": 29810
    },
    {
      "epoch": 0.3032938639322917,
      "grad_norm": 7.786906719207764,
      "learning_rate": 3.9529934851704155e-06,
      "loss": 3.1679,
      "step": 29815
    },
    {
      "epoch": 0.3033447265625,
      "grad_norm": 16.13606071472168,
      "learning_rate": 3.952668058423422e-06,
      "loss": 3.0755,
      "step": 29820
    },
    {
      "epoch": 0.3033955891927083,
      "grad_norm": 9.732821464538574,
      "learning_rate": 3.952342594510262e-06,
      "loss": 3.0813,
      "step": 29825
    },
    {
      "epoch": 0.3034464518229167,
      "grad_norm": 10.42857837677002,
      "learning_rate": 3.952017093439261e-06,
      "loss": 3.399,
      "step": 29830
    },
    {
      "epoch": 0.303497314453125,
      "grad_norm": 7.4075727462768555,
      "learning_rate": 3.951691555218746e-06,
      "loss": 3.5959,
      "step": 29835
    },
    {
      "epoch": 0.3035481770833333,
      "grad_norm": 8.515162467956543,
      "learning_rate": 3.951365979857047e-06,
      "loss": 3.36,
      "step": 29840
    },
    {
      "epoch": 0.3035990397135417,
      "grad_norm": 11.024301528930664,
      "learning_rate": 3.951040367362493e-06,
      "loss": 3.3992,
      "step": 29845
    },
    {
      "epoch": 0.30364990234375,
      "grad_norm": 12.866596221923828,
      "learning_rate": 3.9507147177434155e-06,
      "loss": 3.2026,
      "step": 29850
    },
    {
      "epoch": 0.3037007649739583,
      "grad_norm": 10.392648696899414,
      "learning_rate": 3.950389031008147e-06,
      "loss": 3.6532,
      "step": 29855
    },
    {
      "epoch": 0.3037516276041667,
      "grad_norm": 11.186022758483887,
      "learning_rate": 3.950063307165018e-06,
      "loss": 3.365,
      "step": 29860
    },
    {
      "epoch": 0.303802490234375,
      "grad_norm": 11.300045013427734,
      "learning_rate": 3.949737546222362e-06,
      "loss": 3.1194,
      "step": 29865
    },
    {
      "epoch": 0.3038533528645833,
      "grad_norm": 15.685585975646973,
      "learning_rate": 3.949411748188515e-06,
      "loss": 3.2783,
      "step": 29870
    },
    {
      "epoch": 0.3039042154947917,
      "grad_norm": 11.391825675964355,
      "learning_rate": 3.949085913071812e-06,
      "loss": 3.4238,
      "step": 29875
    },
    {
      "epoch": 0.303955078125,
      "grad_norm": 10.095256805419922,
      "learning_rate": 3.948760040880591e-06,
      "loss": 3.1203,
      "step": 29880
    },
    {
      "epoch": 0.3040059407552083,
      "grad_norm": 12.93649673461914,
      "learning_rate": 3.948434131623185e-06,
      "loss": 3.2666,
      "step": 29885
    },
    {
      "epoch": 0.3040568033854167,
      "grad_norm": 12.856670379638672,
      "learning_rate": 3.948108185307937e-06,
      "loss": 3.0006,
      "step": 29890
    },
    {
      "epoch": 0.304107666015625,
      "grad_norm": 13.006121635437012,
      "learning_rate": 3.947782201943183e-06,
      "loss": 3.2726,
      "step": 29895
    },
    {
      "epoch": 0.3041585286458333,
      "grad_norm": 8.508830070495605,
      "learning_rate": 3.9474561815372655e-06,
      "loss": 2.7996,
      "step": 29900
    },
    {
      "epoch": 0.3042093912760417,
      "grad_norm": 11.361810684204102,
      "learning_rate": 3.947130124098525e-06,
      "loss": 3.5956,
      "step": 29905
    },
    {
      "epoch": 0.30426025390625,
      "grad_norm": 15.425206184387207,
      "learning_rate": 3.946804029635303e-06,
      "loss": 3.8641,
      "step": 29910
    },
    {
      "epoch": 0.3043111165364583,
      "grad_norm": 10.67442512512207,
      "learning_rate": 3.9464778981559425e-06,
      "loss": 3.4641,
      "step": 29915
    },
    {
      "epoch": 0.3043619791666667,
      "grad_norm": 11.199278831481934,
      "learning_rate": 3.946151729668788e-06,
      "loss": 3.3431,
      "step": 29920
    },
    {
      "epoch": 0.304412841796875,
      "grad_norm": 11.408109664916992,
      "learning_rate": 3.9458255241821835e-06,
      "loss": 3.2934,
      "step": 29925
    },
    {
      "epoch": 0.3044637044270833,
      "grad_norm": 11.57914924621582,
      "learning_rate": 3.9454992817044765e-06,
      "loss": 3.0696,
      "step": 29930
    },
    {
      "epoch": 0.3045145670572917,
      "grad_norm": 8.558871269226074,
      "learning_rate": 3.9451730022440125e-06,
      "loss": 3.0752,
      "step": 29935
    },
    {
      "epoch": 0.3045654296875,
      "grad_norm": 11.476302146911621,
      "learning_rate": 3.944846685809141e-06,
      "loss": 3.289,
      "step": 29940
    },
    {
      "epoch": 0.3046162923177083,
      "grad_norm": 12.977760314941406,
      "learning_rate": 3.944520332408208e-06,
      "loss": 3.3298,
      "step": 29945
    },
    {
      "epoch": 0.3046671549479167,
      "grad_norm": 17.571115493774414,
      "learning_rate": 3.9441939420495655e-06,
      "loss": 3.5423,
      "step": 29950
    },
    {
      "epoch": 0.304718017578125,
      "grad_norm": 11.36111068725586,
      "learning_rate": 3.943867514741563e-06,
      "loss": 3.1054,
      "step": 29955
    },
    {
      "epoch": 0.3047688802083333,
      "grad_norm": 12.079353332519531,
      "learning_rate": 3.943541050492552e-06,
      "loss": 3.0162,
      "step": 29960
    },
    {
      "epoch": 0.3048197428385417,
      "grad_norm": 10.957902908325195,
      "learning_rate": 3.9432145493108864e-06,
      "loss": 3.2614,
      "step": 29965
    },
    {
      "epoch": 0.30487060546875,
      "grad_norm": 13.214820861816406,
      "learning_rate": 3.942888011204918e-06,
      "loss": 3.2373,
      "step": 29970
    },
    {
      "epoch": 0.3049214680989583,
      "grad_norm": 11.408329010009766,
      "learning_rate": 3.942561436183002e-06,
      "loss": 3.0805,
      "step": 29975
    },
    {
      "epoch": 0.3049723307291667,
      "grad_norm": 9.606331825256348,
      "learning_rate": 3.9422348242534935e-06,
      "loss": 3.462,
      "step": 29980
    },
    {
      "epoch": 0.305023193359375,
      "grad_norm": 9.041865348815918,
      "learning_rate": 3.941908175424749e-06,
      "loss": 3.6158,
      "step": 29985
    },
    {
      "epoch": 0.3050740559895833,
      "grad_norm": 12.085333824157715,
      "learning_rate": 3.941581489705125e-06,
      "loss": 3.2969,
      "step": 29990
    },
    {
      "epoch": 0.3051249186197917,
      "grad_norm": 14.818300247192383,
      "learning_rate": 3.941254767102981e-06,
      "loss": 3.5366,
      "step": 29995
    },
    {
      "epoch": 0.30517578125,
      "grad_norm": 12.06135082244873,
      "learning_rate": 3.940928007626675e-06,
      "loss": 3.5027,
      "step": 30000
    },
    {
      "epoch": 0.3052266438802083,
      "grad_norm": 8.41988754272461,
      "learning_rate": 3.940601211284568e-06,
      "loss": 3.5845,
      "step": 30005
    },
    {
      "epoch": 0.3052775065104167,
      "grad_norm": 9.827239990234375,
      "learning_rate": 3.94027437808502e-06,
      "loss": 3.5265,
      "step": 30010
    },
    {
      "epoch": 0.305328369140625,
      "grad_norm": 15.531473159790039,
      "learning_rate": 3.939947508036394e-06,
      "loss": 3.2988,
      "step": 30015
    },
    {
      "epoch": 0.3053792317708333,
      "grad_norm": 16.49957847595215,
      "learning_rate": 3.939620601147051e-06,
      "loss": 3.6751,
      "step": 30020
    },
    {
      "epoch": 0.3054300944010417,
      "grad_norm": 17.067161560058594,
      "learning_rate": 3.939293657425357e-06,
      "loss": 4.2174,
      "step": 30025
    },
    {
      "epoch": 0.30548095703125,
      "grad_norm": 10.203935623168945,
      "learning_rate": 3.9389666768796765e-06,
      "loss": 3.4124,
      "step": 30030
    },
    {
      "epoch": 0.3055318196614583,
      "grad_norm": 8.776480674743652,
      "learning_rate": 3.938639659518374e-06,
      "loss": 3.111,
      "step": 30035
    },
    {
      "epoch": 0.3055826822916667,
      "grad_norm": 16.952741622924805,
      "learning_rate": 3.938312605349817e-06,
      "loss": 3.3653,
      "step": 30040
    },
    {
      "epoch": 0.305633544921875,
      "grad_norm": 13.235602378845215,
      "learning_rate": 3.937985514382372e-06,
      "loss": 3.3832,
      "step": 30045
    },
    {
      "epoch": 0.3056844075520833,
      "grad_norm": 12.165931701660156,
      "learning_rate": 3.93765838662441e-06,
      "loss": 3.6152,
      "step": 30050
    },
    {
      "epoch": 0.3057352701822917,
      "grad_norm": 12.84066104888916,
      "learning_rate": 3.937331222084298e-06,
      "loss": 3.1008,
      "step": 30055
    },
    {
      "epoch": 0.3057861328125,
      "grad_norm": 11.93371868133545,
      "learning_rate": 3.937004020770407e-06,
      "loss": 3.1214,
      "step": 30060
    },
    {
      "epoch": 0.3058369954427083,
      "grad_norm": 10.240893363952637,
      "learning_rate": 3.936676782691109e-06,
      "loss": 3.5683,
      "step": 30065
    },
    {
      "epoch": 0.3058878580729167,
      "grad_norm": 14.546442031860352,
      "learning_rate": 3.9363495078547765e-06,
      "loss": 3.4732,
      "step": 30070
    },
    {
      "epoch": 0.305938720703125,
      "grad_norm": 14.076205253601074,
      "learning_rate": 3.936022196269782e-06,
      "loss": 3.0446,
      "step": 30075
    },
    {
      "epoch": 0.3059895833333333,
      "grad_norm": 8.096552848815918,
      "learning_rate": 3.9356948479445e-06,
      "loss": 3.5257,
      "step": 30080
    },
    {
      "epoch": 0.3060404459635417,
      "grad_norm": 8.671744346618652,
      "learning_rate": 3.935367462887306e-06,
      "loss": 3.302,
      "step": 30085
    },
    {
      "epoch": 0.30609130859375,
      "grad_norm": 13.429875373840332,
      "learning_rate": 3.935040041106576e-06,
      "loss": 3.1311,
      "step": 30090
    },
    {
      "epoch": 0.3061421712239583,
      "grad_norm": 13.481354713439941,
      "learning_rate": 3.934712582610686e-06,
      "loss": 3.5084,
      "step": 30095
    },
    {
      "epoch": 0.3061930338541667,
      "grad_norm": 12.646124839782715,
      "learning_rate": 3.934385087408014e-06,
      "loss": 3.1246,
      "step": 30100
    },
    {
      "epoch": 0.306243896484375,
      "grad_norm": 10.231154441833496,
      "learning_rate": 3.93405755550694e-06,
      "loss": 3.0921,
      "step": 30105
    },
    {
      "epoch": 0.3062947591145833,
      "grad_norm": 10.196928024291992,
      "learning_rate": 3.9337299869158445e-06,
      "loss": 3.5011,
      "step": 30110
    },
    {
      "epoch": 0.3063456217447917,
      "grad_norm": 13.01209545135498,
      "learning_rate": 3.933402381643106e-06,
      "loss": 3.5474,
      "step": 30115
    },
    {
      "epoch": 0.306396484375,
      "grad_norm": 7.661176681518555,
      "learning_rate": 3.933074739697107e-06,
      "loss": 3.8555,
      "step": 30120
    },
    {
      "epoch": 0.3064473470052083,
      "grad_norm": 14.038602828979492,
      "learning_rate": 3.9327470610862305e-06,
      "loss": 3.0677,
      "step": 30125
    },
    {
      "epoch": 0.3064982096354167,
      "grad_norm": 15.984299659729004,
      "learning_rate": 3.932419345818861e-06,
      "loss": 3.0514,
      "step": 30130
    },
    {
      "epoch": 0.306549072265625,
      "grad_norm": 13.001832962036133,
      "learning_rate": 3.932091593903381e-06,
      "loss": 3.3576,
      "step": 30135
    },
    {
      "epoch": 0.3065999348958333,
      "grad_norm": 15.798726081848145,
      "learning_rate": 3.931763805348179e-06,
      "loss": 3.5285,
      "step": 30140
    },
    {
      "epoch": 0.3066507975260417,
      "grad_norm": 12.089749336242676,
      "learning_rate": 3.9314359801616374e-06,
      "loss": 3.4186,
      "step": 30145
    },
    {
      "epoch": 0.30670166015625,
      "grad_norm": 14.937867164611816,
      "learning_rate": 3.931108118352146e-06,
      "loss": 2.9777,
      "step": 30150
    },
    {
      "epoch": 0.3067525227864583,
      "grad_norm": 7.658392429351807,
      "learning_rate": 3.930780219928093e-06,
      "loss": 3.2119,
      "step": 30155
    },
    {
      "epoch": 0.3068033854166667,
      "grad_norm": 13.586400985717773,
      "learning_rate": 3.930452284897867e-06,
      "loss": 3.2893,
      "step": 30160
    },
    {
      "epoch": 0.306854248046875,
      "grad_norm": 14.8994722366333,
      "learning_rate": 3.930124313269858e-06,
      "loss": 3.5276,
      "step": 30165
    },
    {
      "epoch": 0.3069051106770833,
      "grad_norm": 12.488736152648926,
      "learning_rate": 3.929796305052458e-06,
      "loss": 3.8872,
      "step": 30170
    },
    {
      "epoch": 0.3069559733072917,
      "grad_norm": 8.962054252624512,
      "learning_rate": 3.929468260254058e-06,
      "loss": 3.234,
      "step": 30175
    },
    {
      "epoch": 0.3070068359375,
      "grad_norm": 9.514458656311035,
      "learning_rate": 3.9291401788830505e-06,
      "loss": 3.4633,
      "step": 30180
    },
    {
      "epoch": 0.3070576985677083,
      "grad_norm": 12.32824420928955,
      "learning_rate": 3.928812060947832e-06,
      "loss": 3.6234,
      "step": 30185
    },
    {
      "epoch": 0.3071085611979167,
      "grad_norm": 7.72622013092041,
      "learning_rate": 3.928483906456795e-06,
      "loss": 3.188,
      "step": 30190
    },
    {
      "epoch": 0.307159423828125,
      "grad_norm": 16.456998825073242,
      "learning_rate": 3.9281557154183355e-06,
      "loss": 3.5812,
      "step": 30195
    },
    {
      "epoch": 0.3072102864583333,
      "grad_norm": 7.127664089202881,
      "learning_rate": 3.927827487840851e-06,
      "loss": 3.2705,
      "step": 30200
    },
    {
      "epoch": 0.3072611490885417,
      "grad_norm": 13.026443481445312,
      "learning_rate": 3.927499223732738e-06,
      "loss": 3.2092,
      "step": 30205
    },
    {
      "epoch": 0.30731201171875,
      "grad_norm": 12.97545051574707,
      "learning_rate": 3.927170923102396e-06,
      "loss": 3.5209,
      "step": 30210
    },
    {
      "epoch": 0.3073628743489583,
      "grad_norm": 7.901330947875977,
      "learning_rate": 3.926842585958225e-06,
      "loss": 3.4217,
      "step": 30215
    },
    {
      "epoch": 0.3074137369791667,
      "grad_norm": 11.98694896697998,
      "learning_rate": 3.926514212308624e-06,
      "loss": 3.086,
      "step": 30220
    },
    {
      "epoch": 0.307464599609375,
      "grad_norm": 11.968120574951172,
      "learning_rate": 3.926185802161995e-06,
      "loss": 3.5508,
      "step": 30225
    },
    {
      "epoch": 0.3075154622395833,
      "grad_norm": 9.247671127319336,
      "learning_rate": 3.92585735552674e-06,
      "loss": 3.6386,
      "step": 30230
    },
    {
      "epoch": 0.3075663248697917,
      "grad_norm": 9.899497985839844,
      "learning_rate": 3.925528872411263e-06,
      "loss": 3.4517,
      "step": 30235
    },
    {
      "epoch": 0.3076171875,
      "grad_norm": 13.091283798217773,
      "learning_rate": 3.925200352823967e-06,
      "loss": 3.8267,
      "step": 30240
    },
    {
      "epoch": 0.3076680501302083,
      "grad_norm": 12.776516914367676,
      "learning_rate": 3.924871796773259e-06,
      "loss": 3.7947,
      "step": 30245
    },
    {
      "epoch": 0.3077189127604167,
      "grad_norm": 7.171541690826416,
      "learning_rate": 3.924543204267542e-06,
      "loss": 3.1933,
      "step": 30250
    },
    {
      "epoch": 0.307769775390625,
      "grad_norm": 17.34246826171875,
      "learning_rate": 3.924214575315226e-06,
      "loss": 3.6964,
      "step": 30255
    },
    {
      "epoch": 0.3078206380208333,
      "grad_norm": 17.20662498474121,
      "learning_rate": 3.923885909924717e-06,
      "loss": 3.5014,
      "step": 30260
    },
    {
      "epoch": 0.3078715006510417,
      "grad_norm": 10.750012397766113,
      "learning_rate": 3.923557208104426e-06,
      "loss": 3.4796,
      "step": 30265
    },
    {
      "epoch": 0.30792236328125,
      "grad_norm": 14.636932373046875,
      "learning_rate": 3.92322846986276e-06,
      "loss": 2.8819,
      "step": 30270
    },
    {
      "epoch": 0.3079732259114583,
      "grad_norm": 13.901700973510742,
      "learning_rate": 3.922899695208131e-06,
      "loss": 3.3566,
      "step": 30275
    },
    {
      "epoch": 0.3080240885416667,
      "grad_norm": 32.55217742919922,
      "learning_rate": 3.922570884148951e-06,
      "loss": 3.2527,
      "step": 30280
    },
    {
      "epoch": 0.308074951171875,
      "grad_norm": 16.40342903137207,
      "learning_rate": 3.922242036693632e-06,
      "loss": 3.8343,
      "step": 30285
    },
    {
      "epoch": 0.3081258138020833,
      "grad_norm": 12.78229808807373,
      "learning_rate": 3.921913152850587e-06,
      "loss": 3.2929,
      "step": 30290
    },
    {
      "epoch": 0.3081766764322917,
      "grad_norm": 14.18374252319336,
      "learning_rate": 3.921584232628232e-06,
      "loss": 3.3324,
      "step": 30295
    },
    {
      "epoch": 0.3082275390625,
      "grad_norm": 13.564252853393555,
      "learning_rate": 3.92125527603498e-06,
      "loss": 3.1763,
      "step": 30300
    },
    {
      "epoch": 0.3082784016927083,
      "grad_norm": 10.080500602722168,
      "learning_rate": 3.920926283079249e-06,
      "loss": 3.3237,
      "step": 30305
    },
    {
      "epoch": 0.3083292643229167,
      "grad_norm": 15.610957145690918,
      "learning_rate": 3.920597253769457e-06,
      "loss": 3.3335,
      "step": 30310
    },
    {
      "epoch": 0.308380126953125,
      "grad_norm": 11.604490280151367,
      "learning_rate": 3.9202681881140206e-06,
      "loss": 3.3937,
      "step": 30315
    },
    {
      "epoch": 0.3084309895833333,
      "grad_norm": 12.033234596252441,
      "learning_rate": 3.919939086121359e-06,
      "loss": 3.5559,
      "step": 30320
    },
    {
      "epoch": 0.3084818522135417,
      "grad_norm": 10.802011489868164,
      "learning_rate": 3.919609947799892e-06,
      "loss": 3.1012,
      "step": 30325
    },
    {
      "epoch": 0.30853271484375,
      "grad_norm": 12.17712116241455,
      "learning_rate": 3.9192807731580415e-06,
      "loss": 3.3434,
      "step": 30330
    },
    {
      "epoch": 0.3085835774739583,
      "grad_norm": 11.824995994567871,
      "learning_rate": 3.9189515622042285e-06,
      "loss": 3.5515,
      "step": 30335
    },
    {
      "epoch": 0.3086344401041667,
      "grad_norm": 9.751220703125,
      "learning_rate": 3.9186223149468764e-06,
      "loss": 3.212,
      "step": 30340
    },
    {
      "epoch": 0.308685302734375,
      "grad_norm": 12.710297584533691,
      "learning_rate": 3.918293031394408e-06,
      "loss": 3.1981,
      "step": 30345
    },
    {
      "epoch": 0.3087361653645833,
      "grad_norm": 8.323152542114258,
      "learning_rate": 3.917963711555249e-06,
      "loss": 3.2414,
      "step": 30350
    },
    {
      "epoch": 0.3087870279947917,
      "grad_norm": 7.017961025238037,
      "learning_rate": 3.917634355437825e-06,
      "loss": 3.3119,
      "step": 30355
    },
    {
      "epoch": 0.308837890625,
      "grad_norm": 15.37889289855957,
      "learning_rate": 3.917304963050562e-06,
      "loss": 3.4028,
      "step": 30360
    },
    {
      "epoch": 0.3088887532552083,
      "grad_norm": 8.823172569274902,
      "learning_rate": 3.916975534401887e-06,
      "loss": 3.4427,
      "step": 30365
    },
    {
      "epoch": 0.3089396158854167,
      "grad_norm": 11.566137313842773,
      "learning_rate": 3.916646069500228e-06,
      "loss": 3.1481,
      "step": 30370
    },
    {
      "epoch": 0.308990478515625,
      "grad_norm": 13.274648666381836,
      "learning_rate": 3.916316568354017e-06,
      "loss": 3.1607,
      "step": 30375
    },
    {
      "epoch": 0.3090413411458333,
      "grad_norm": 9.833974838256836,
      "learning_rate": 3.915987030971681e-06,
      "loss": 3.4893,
      "step": 30380
    },
    {
      "epoch": 0.3090922037760417,
      "grad_norm": 9.17192554473877,
      "learning_rate": 3.915657457361652e-06,
      "loss": 3.3514,
      "step": 30385
    },
    {
      "epoch": 0.30914306640625,
      "grad_norm": 10.900861740112305,
      "learning_rate": 3.915327847532364e-06,
      "loss": 3.1907,
      "step": 30390
    },
    {
      "epoch": 0.3091939290364583,
      "grad_norm": 10.886656761169434,
      "learning_rate": 3.914998201492248e-06,
      "loss": 3.4113,
      "step": 30395
    },
    {
      "epoch": 0.3092447916666667,
      "grad_norm": 10.764690399169922,
      "learning_rate": 3.914668519249738e-06,
      "loss": 3.2688,
      "step": 30400
    },
    {
      "epoch": 0.309295654296875,
      "grad_norm": 16.22494888305664,
      "learning_rate": 3.9143388008132695e-06,
      "loss": 3.357,
      "step": 30405
    },
    {
      "epoch": 0.3093465169270833,
      "grad_norm": 8.67312240600586,
      "learning_rate": 3.914009046191277e-06,
      "loss": 3.7883,
      "step": 30410
    },
    {
      "epoch": 0.3093973795572917,
      "grad_norm": 9.882831573486328,
      "learning_rate": 3.9136792553922e-06,
      "loss": 3.2242,
      "step": 30415
    },
    {
      "epoch": 0.3094482421875,
      "grad_norm": 12.467557907104492,
      "learning_rate": 3.913349428424473e-06,
      "loss": 3.2608,
      "step": 30420
    },
    {
      "epoch": 0.3094991048177083,
      "grad_norm": 13.472617149353027,
      "learning_rate": 3.913019565296536e-06,
      "loss": 3.4081,
      "step": 30425
    },
    {
      "epoch": 0.3095499674479167,
      "grad_norm": 13.341675758361816,
      "learning_rate": 3.912689666016829e-06,
      "loss": 3.3591,
      "step": 30430
    },
    {
      "epoch": 0.309600830078125,
      "grad_norm": 15.18631649017334,
      "learning_rate": 3.9123597305937915e-06,
      "loss": 3.2089,
      "step": 30435
    },
    {
      "epoch": 0.3096516927083333,
      "grad_norm": 10.999125480651855,
      "learning_rate": 3.912029759035864e-06,
      "loss": 3.2369,
      "step": 30440
    },
    {
      "epoch": 0.3097025553385417,
      "grad_norm": 11.49664306640625,
      "learning_rate": 3.911699751351492e-06,
      "loss": 3.3434,
      "step": 30445
    },
    {
      "epoch": 0.30975341796875,
      "grad_norm": 9.98498249053955,
      "learning_rate": 3.911369707549114e-06,
      "loss": 3.2589,
      "step": 30450
    },
    {
      "epoch": 0.3098042805989583,
      "grad_norm": 10.603560447692871,
      "learning_rate": 3.911039627637179e-06,
      "loss": 2.9951,
      "step": 30455
    },
    {
      "epoch": 0.3098551432291667,
      "grad_norm": 8.391109466552734,
      "learning_rate": 3.910709511624128e-06,
      "loss": 3.558,
      "step": 30460
    },
    {
      "epoch": 0.309906005859375,
      "grad_norm": 15.02904224395752,
      "learning_rate": 3.910379359518409e-06,
      "loss": 4.0405,
      "step": 30465
    },
    {
      "epoch": 0.3099568684895833,
      "grad_norm": 11.91486644744873,
      "learning_rate": 3.910049171328468e-06,
      "loss": 3.6384,
      "step": 30470
    },
    {
      "epoch": 0.3100077311197917,
      "grad_norm": 13.936861991882324,
      "learning_rate": 3.9097189470627535e-06,
      "loss": 3.1747,
      "step": 30475
    },
    {
      "epoch": 0.31005859375,
      "grad_norm": 18.80107307434082,
      "learning_rate": 3.909388686729714e-06,
      "loss": 3.248,
      "step": 30480
    },
    {
      "epoch": 0.3101094563802083,
      "grad_norm": 12.20251750946045,
      "learning_rate": 3.909058390337799e-06,
      "loss": 3.5339,
      "step": 30485
    },
    {
      "epoch": 0.3101603190104167,
      "grad_norm": 12.874309539794922,
      "learning_rate": 3.908728057895459e-06,
      "loss": 3.1153,
      "step": 30490
    },
    {
      "epoch": 0.310211181640625,
      "grad_norm": 12.415948867797852,
      "learning_rate": 3.908397689411146e-06,
      "loss": 3.4896,
      "step": 30495
    },
    {
      "epoch": 0.3102620442708333,
      "grad_norm": 8.233391761779785,
      "learning_rate": 3.908067284893311e-06,
      "loss": 3.1929,
      "step": 30500
    },
    {
      "epoch": 0.3103129069010417,
      "grad_norm": 15.04599380493164,
      "learning_rate": 3.907736844350408e-06,
      "loss": 3.385,
      "step": 30505
    },
    {
      "epoch": 0.31036376953125,
      "grad_norm": 9.692916870117188,
      "learning_rate": 3.907406367790892e-06,
      "loss": 3.16,
      "step": 30510
    },
    {
      "epoch": 0.3104146321614583,
      "grad_norm": 13.403112411499023,
      "learning_rate": 3.907075855223218e-06,
      "loss": 3.398,
      "step": 30515
    },
    {
      "epoch": 0.3104654947916667,
      "grad_norm": 18.633678436279297,
      "learning_rate": 3.9067453066558415e-06,
      "loss": 3.4533,
      "step": 30520
    },
    {
      "epoch": 0.310516357421875,
      "grad_norm": 13.57737922668457,
      "learning_rate": 3.90641472209722e-06,
      "loss": 3.2045,
      "step": 30525
    },
    {
      "epoch": 0.3105672200520833,
      "grad_norm": 14.787510871887207,
      "learning_rate": 3.906084101555812e-06,
      "loss": 3.4004,
      "step": 30530
    },
    {
      "epoch": 0.3106180826822917,
      "grad_norm": 8.80350112915039,
      "learning_rate": 3.9057534450400735e-06,
      "loss": 3.3955,
      "step": 30535
    },
    {
      "epoch": 0.3106689453125,
      "grad_norm": 10.163858413696289,
      "learning_rate": 3.905422752558468e-06,
      "loss": 3.324,
      "step": 30540
    },
    {
      "epoch": 0.3107198079427083,
      "grad_norm": 9.57807731628418,
      "learning_rate": 3.905092024119454e-06,
      "loss": 3.5297,
      "step": 30545
    },
    {
      "epoch": 0.3107706705729167,
      "grad_norm": 9.638726234436035,
      "learning_rate": 3.904761259731493e-06,
      "loss": 3.2324,
      "step": 30550
    },
    {
      "epoch": 0.310821533203125,
      "grad_norm": 11.510026931762695,
      "learning_rate": 3.904430459403049e-06,
      "loss": 3.2802,
      "step": 30555
    },
    {
      "epoch": 0.3108723958333333,
      "grad_norm": 14.225361824035645,
      "learning_rate": 3.904099623142584e-06,
      "loss": 4.1202,
      "step": 30560
    },
    {
      "epoch": 0.3109232584635417,
      "grad_norm": 8.770647048950195,
      "learning_rate": 3.903768750958563e-06,
      "loss": 3.258,
      "step": 30565
    },
    {
      "epoch": 0.31097412109375,
      "grad_norm": 11.526542663574219,
      "learning_rate": 3.903437842859451e-06,
      "loss": 3.1181,
      "step": 30570
    },
    {
      "epoch": 0.3110249837239583,
      "grad_norm": 11.786979675292969,
      "learning_rate": 3.903106898853715e-06,
      "loss": 2.9532,
      "step": 30575
    },
    {
      "epoch": 0.3110758463541667,
      "grad_norm": 10.150199890136719,
      "learning_rate": 3.9027759189498215e-06,
      "loss": 3.2941,
      "step": 30580
    },
    {
      "epoch": 0.311126708984375,
      "grad_norm": 13.700420379638672,
      "learning_rate": 3.902444903156239e-06,
      "loss": 3.4857,
      "step": 30585
    },
    {
      "epoch": 0.3111775716145833,
      "grad_norm": 14.847107887268066,
      "learning_rate": 3.902113851481435e-06,
      "loss": 3.3416,
      "step": 30590
    },
    {
      "epoch": 0.3112284342447917,
      "grad_norm": 15.067854881286621,
      "learning_rate": 3.901782763933881e-06,
      "loss": 3.1923,
      "step": 30595
    },
    {
      "epoch": 0.311279296875,
      "grad_norm": 13.336148262023926,
      "learning_rate": 3.901451640522047e-06,
      "loss": 3.0086,
      "step": 30600
    },
    {
      "epoch": 0.3113301595052083,
      "grad_norm": 9.487920761108398,
      "learning_rate": 3.901120481254405e-06,
      "loss": 3.1972,
      "step": 30605
    },
    {
      "epoch": 0.3113810221354167,
      "grad_norm": 10.81859302520752,
      "learning_rate": 3.900789286139429e-06,
      "loss": 3.1985,
      "step": 30610
    },
    {
      "epoch": 0.311431884765625,
      "grad_norm": 8.982028007507324,
      "learning_rate": 3.900458055185589e-06,
      "loss": 3.1001,
      "step": 30615
    },
    {
      "epoch": 0.3114827473958333,
      "grad_norm": 9.588761329650879,
      "learning_rate": 3.900126788401363e-06,
      "loss": 3.204,
      "step": 30620
    },
    {
      "epoch": 0.3115336100260417,
      "grad_norm": 9.439041137695312,
      "learning_rate": 3.899795485795225e-06,
      "loss": 3.0524,
      "step": 30625
    },
    {
      "epoch": 0.31158447265625,
      "grad_norm": 12.190141677856445,
      "learning_rate": 3.899464147375651e-06,
      "loss": 3.6372,
      "step": 30630
    },
    {
      "epoch": 0.3116353352864583,
      "grad_norm": 13.552200317382812,
      "learning_rate": 3.899132773151118e-06,
      "loss": 3.2865,
      "step": 30635
    },
    {
      "epoch": 0.3116861979166667,
      "grad_norm": 13.921281814575195,
      "learning_rate": 3.898801363130105e-06,
      "loss": 3.3674,
      "step": 30640
    },
    {
      "epoch": 0.311737060546875,
      "grad_norm": 8.771903038024902,
      "learning_rate": 3.898469917321091e-06,
      "loss": 3.2259,
      "step": 30645
    },
    {
      "epoch": 0.3117879231770833,
      "grad_norm": 13.268839836120605,
      "learning_rate": 3.8981384357325545e-06,
      "loss": 3.368,
      "step": 30650
    },
    {
      "epoch": 0.3118387858072917,
      "grad_norm": 13.863824844360352,
      "learning_rate": 3.897806918372978e-06,
      "loss": 3.2181,
      "step": 30655
    },
    {
      "epoch": 0.3118896484375,
      "grad_norm": 15.39890193939209,
      "learning_rate": 3.8974753652508436e-06,
      "loss": 3.374,
      "step": 30660
    },
    {
      "epoch": 0.3119405110677083,
      "grad_norm": 13.512394905090332,
      "learning_rate": 3.897143776374633e-06,
      "loss": 3.0666,
      "step": 30665
    },
    {
      "epoch": 0.3119913736979167,
      "grad_norm": 11.035284042358398,
      "learning_rate": 3.89681215175283e-06,
      "loss": 3.2409,
      "step": 30670
    },
    {
      "epoch": 0.312042236328125,
      "grad_norm": 12.44509506225586,
      "learning_rate": 3.896480491393918e-06,
      "loss": 3.638,
      "step": 30675
    },
    {
      "epoch": 0.3120930989583333,
      "grad_norm": 14.145005226135254,
      "learning_rate": 3.8961487953063845e-06,
      "loss": 3.1806,
      "step": 30680
    },
    {
      "epoch": 0.3121439615885417,
      "grad_norm": 7.920605182647705,
      "learning_rate": 3.895817063498716e-06,
      "loss": 3.5746,
      "step": 30685
    },
    {
      "epoch": 0.31219482421875,
      "grad_norm": 11.769000053405762,
      "learning_rate": 3.895485295979398e-06,
      "loss": 3.2961,
      "step": 30690
    },
    {
      "epoch": 0.3122456868489583,
      "grad_norm": 11.063861846923828,
      "learning_rate": 3.895153492756919e-06,
      "loss": 3.9124,
      "step": 30695
    },
    {
      "epoch": 0.3122965494791667,
      "grad_norm": 14.211889266967773,
      "learning_rate": 3.894821653839769e-06,
      "loss": 3.4248,
      "step": 30700
    },
    {
      "epoch": 0.312347412109375,
      "grad_norm": 12.15110969543457,
      "learning_rate": 3.8944897792364366e-06,
      "loss": 3.1158,
      "step": 30705
    },
    {
      "epoch": 0.3123982747395833,
      "grad_norm": 14.808752059936523,
      "learning_rate": 3.894157868955415e-06,
      "loss": 3.3555,
      "step": 30710
    },
    {
      "epoch": 0.3124491373697917,
      "grad_norm": 14.335022926330566,
      "learning_rate": 3.893825923005195e-06,
      "loss": 3.2379,
      "step": 30715
    },
    {
      "epoch": 0.3125,
      "grad_norm": 14.32910442352295,
      "learning_rate": 3.893493941394268e-06,
      "loss": 3.1627,
      "step": 30720
    },
    {
      "epoch": 0.3125508626302083,
      "grad_norm": 11.805767059326172,
      "learning_rate": 3.893161924131129e-06,
      "loss": 3.5045,
      "step": 30725
    },
    {
      "epoch": 0.3126017252604167,
      "grad_norm": 10.202608108520508,
      "learning_rate": 3.892829871224273e-06,
      "loss": 3.4099,
      "step": 30730
    },
    {
      "epoch": 0.312652587890625,
      "grad_norm": 10.266390800476074,
      "learning_rate": 3.892497782682195e-06,
      "loss": 3.1493,
      "step": 30735
    },
    {
      "epoch": 0.3127034505208333,
      "grad_norm": 12.03805160522461,
      "learning_rate": 3.89216565851339e-06,
      "loss": 3.225,
      "step": 30740
    },
    {
      "epoch": 0.3127543131510417,
      "grad_norm": 11.497048377990723,
      "learning_rate": 3.891833498726359e-06,
      "loss": 2.9702,
      "step": 30745
    },
    {
      "epoch": 0.31280517578125,
      "grad_norm": 11.640868186950684,
      "learning_rate": 3.891501303329596e-06,
      "loss": 3.4517,
      "step": 30750
    },
    {
      "epoch": 0.3128560384114583,
      "grad_norm": 14.435894966125488,
      "learning_rate": 3.891169072331601e-06,
      "loss": 3.4013,
      "step": 30755
    },
    {
      "epoch": 0.3129069010416667,
      "grad_norm": 14.759724617004395,
      "learning_rate": 3.890836805740877e-06,
      "loss": 3.4514,
      "step": 30760
    },
    {
      "epoch": 0.312957763671875,
      "grad_norm": 9.112564086914062,
      "learning_rate": 3.890504503565922e-06,
      "loss": 3.2215,
      "step": 30765
    },
    {
      "epoch": 0.3130086263020833,
      "grad_norm": 12.83095645904541,
      "learning_rate": 3.890172165815239e-06,
      "loss": 3.357,
      "step": 30770
    },
    {
      "epoch": 0.3130594889322917,
      "grad_norm": 11.383550643920898,
      "learning_rate": 3.889839792497331e-06,
      "loss": 3.113,
      "step": 30775
    },
    {
      "epoch": 0.3131103515625,
      "grad_norm": 11.741923332214355,
      "learning_rate": 3.8895073836207015e-06,
      "loss": 3.3635,
      "step": 30780
    },
    {
      "epoch": 0.3131612141927083,
      "grad_norm": 8.802398681640625,
      "learning_rate": 3.889174939193855e-06,
      "loss": 3.3568,
      "step": 30785
    },
    {
      "epoch": 0.3132120768229167,
      "grad_norm": 14.020269393920898,
      "learning_rate": 3.888842459225296e-06,
      "loss": 3.4345,
      "step": 30790
    },
    {
      "epoch": 0.313262939453125,
      "grad_norm": 12.648975372314453,
      "learning_rate": 3.8885099437235325e-06,
      "loss": 3.5363,
      "step": 30795
    },
    {
      "epoch": 0.3133138020833333,
      "grad_norm": 13.394015312194824,
      "learning_rate": 3.888177392697071e-06,
      "loss": 3.3288,
      "step": 30800
    },
    {
      "epoch": 0.3133646647135417,
      "grad_norm": 10.144981384277344,
      "learning_rate": 3.8878448061544204e-06,
      "loss": 3.1019,
      "step": 30805
    },
    {
      "epoch": 0.31341552734375,
      "grad_norm": 11.660987854003906,
      "learning_rate": 3.887512184104088e-06,
      "loss": 3.2372,
      "step": 30810
    },
    {
      "epoch": 0.3134663899739583,
      "grad_norm": 14.93466854095459,
      "learning_rate": 3.8871795265545875e-06,
      "loss": 3.1339,
      "step": 30815
    },
    {
      "epoch": 0.3135172526041667,
      "grad_norm": 13.46787166595459,
      "learning_rate": 3.886846833514426e-06,
      "loss": 3.6266,
      "step": 30820
    },
    {
      "epoch": 0.313568115234375,
      "grad_norm": 14.1391019821167,
      "learning_rate": 3.886514104992117e-06,
      "loss": 3.4762,
      "step": 30825
    },
    {
      "epoch": 0.3136189778645833,
      "grad_norm": 16.144411087036133,
      "learning_rate": 3.886181340996174e-06,
      "loss": 3.6282,
      "step": 30830
    },
    {
      "epoch": 0.3136698404947917,
      "grad_norm": 13.948810577392578,
      "learning_rate": 3.88584854153511e-06,
      "loss": 3.3651,
      "step": 30835
    },
    {
      "epoch": 0.313720703125,
      "grad_norm": 7.992741584777832,
      "learning_rate": 3.885515706617439e-06,
      "loss": 3.6157,
      "step": 30840
    },
    {
      "epoch": 0.3137715657552083,
      "grad_norm": 12.564249992370605,
      "learning_rate": 3.885182836251678e-06,
      "loss": 3.2923,
      "step": 30845
    },
    {
      "epoch": 0.3138224283854167,
      "grad_norm": 13.613372802734375,
      "learning_rate": 3.884849930446341e-06,
      "loss": 3.39,
      "step": 30850
    },
    {
      "epoch": 0.313873291015625,
      "grad_norm": 13.717891693115234,
      "learning_rate": 3.884516989209948e-06,
      "loss": 3.0935,
      "step": 30855
    },
    {
      "epoch": 0.3139241536458333,
      "grad_norm": 14.83971881866455,
      "learning_rate": 3.8841840125510145e-06,
      "loss": 3.3069,
      "step": 30860
    },
    {
      "epoch": 0.3139750162760417,
      "grad_norm": 14.060420036315918,
      "learning_rate": 3.8838510004780615e-06,
      "loss": 3.3003,
      "step": 30865
    },
    {
      "epoch": 0.31402587890625,
      "grad_norm": 12.112072944641113,
      "learning_rate": 3.88351795299961e-06,
      "loss": 3.2948,
      "step": 30870
    },
    {
      "epoch": 0.3140767415364583,
      "grad_norm": 12.657612800598145,
      "learning_rate": 3.883184870124178e-06,
      "loss": 2.8017,
      "step": 30875
    },
    {
      "epoch": 0.3141276041666667,
      "grad_norm": 9.51637077331543,
      "learning_rate": 3.88285175186029e-06,
      "loss": 3.451,
      "step": 30880
    },
    {
      "epoch": 0.314178466796875,
      "grad_norm": 8.902334213256836,
      "learning_rate": 3.882518598216467e-06,
      "loss": 3.4797,
      "step": 30885
    },
    {
      "epoch": 0.3142293294270833,
      "grad_norm": 11.905548095703125,
      "learning_rate": 3.8821854092012345e-06,
      "loss": 3.1486,
      "step": 30890
    },
    {
      "epoch": 0.3142801920572917,
      "grad_norm": 11.69454574584961,
      "learning_rate": 3.8818521848231155e-06,
      "loss": 3.357,
      "step": 30895
    },
    {
      "epoch": 0.3143310546875,
      "grad_norm": 9.603013038635254,
      "learning_rate": 3.881518925090636e-06,
      "loss": 3.7997,
      "step": 30900
    },
    {
      "epoch": 0.3143819173177083,
      "grad_norm": 8.937087059020996,
      "learning_rate": 3.881185630012321e-06,
      "loss": 3.197,
      "step": 30905
    },
    {
      "epoch": 0.3144327799479167,
      "grad_norm": 14.387102127075195,
      "learning_rate": 3.8808522995967e-06,
      "loss": 3.3608,
      "step": 30910
    },
    {
      "epoch": 0.314483642578125,
      "grad_norm": 8.974607467651367,
      "learning_rate": 3.880518933852301e-06,
      "loss": 3.3294,
      "step": 30915
    },
    {
      "epoch": 0.3145345052083333,
      "grad_norm": 16.349624633789062,
      "learning_rate": 3.880185532787651e-06,
      "loss": 3.7891,
      "step": 30920
    },
    {
      "epoch": 0.3145853678385417,
      "grad_norm": 10.930258750915527,
      "learning_rate": 3.879852096411282e-06,
      "loss": 3.6586,
      "step": 30925
    },
    {
      "epoch": 0.31463623046875,
      "grad_norm": 12.886204719543457,
      "learning_rate": 3.879518624731724e-06,
      "loss": 3.3628,
      "step": 30930
    },
    {
      "epoch": 0.3146870930989583,
      "grad_norm": 12.439726829528809,
      "learning_rate": 3.8791851177575085e-06,
      "loss": 3.2021,
      "step": 30935
    },
    {
      "epoch": 0.3147379557291667,
      "grad_norm": 8.862081527709961,
      "learning_rate": 3.878851575497169e-06,
      "loss": 3.4902,
      "step": 30940
    },
    {
      "epoch": 0.314788818359375,
      "grad_norm": 10.594961166381836,
      "learning_rate": 3.878517997959239e-06,
      "loss": 3.1558,
      "step": 30945
    },
    {
      "epoch": 0.3148396809895833,
      "grad_norm": 11.894721984863281,
      "learning_rate": 3.8781843851522525e-06,
      "loss": 3.5946,
      "step": 30950
    },
    {
      "epoch": 0.3148905436197917,
      "grad_norm": 9.851780891418457,
      "learning_rate": 3.877850737084745e-06,
      "loss": 3.2834,
      "step": 30955
    },
    {
      "epoch": 0.31494140625,
      "grad_norm": 14.598539352416992,
      "learning_rate": 3.877517053765253e-06,
      "loss": 3.8892,
      "step": 30960
    },
    {
      "epoch": 0.3149922688802083,
      "grad_norm": 12.510114669799805,
      "learning_rate": 3.877183335202314e-06,
      "loss": 3.0889,
      "step": 30965
    },
    {
      "epoch": 0.3150431315104167,
      "grad_norm": 10.286641120910645,
      "learning_rate": 3.876849581404466e-06,
      "loss": 3.2251,
      "step": 30970
    },
    {
      "epoch": 0.315093994140625,
      "grad_norm": 14.571932792663574,
      "learning_rate": 3.876515792380247e-06,
      "loss": 3.2544,
      "step": 30975
    },
    {
      "epoch": 0.3151448567708333,
      "grad_norm": 8.331310272216797,
      "learning_rate": 3.876181968138199e-06,
      "loss": 3.6741,
      "step": 30980
    },
    {
      "epoch": 0.3151957194010417,
      "grad_norm": 14.296549797058105,
      "learning_rate": 3.8758481086868605e-06,
      "loss": 3.3286,
      "step": 30985
    },
    {
      "epoch": 0.31524658203125,
      "grad_norm": 10.8761568069458,
      "learning_rate": 3.875514214034775e-06,
      "loss": 3.6252,
      "step": 30990
    },
    {
      "epoch": 0.3152974446614583,
      "grad_norm": 11.81722640991211,
      "learning_rate": 3.875180284190483e-06,
      "loss": 3.6989,
      "step": 30995
    },
    {
      "epoch": 0.3153483072916667,
      "grad_norm": 14.4451904296875,
      "learning_rate": 3.874846319162531e-06,
      "loss": 3.2001,
      "step": 31000
    },
    {
      "epoch": 0.315399169921875,
      "grad_norm": 11.83335018157959,
      "learning_rate": 3.87451231895946e-06,
      "loss": 3.3927,
      "step": 31005
    },
    {
      "epoch": 0.3154500325520833,
      "grad_norm": 18.12177276611328,
      "learning_rate": 3.874178283589819e-06,
      "loss": 3.4793,
      "step": 31010
    },
    {
      "epoch": 0.3155008951822917,
      "grad_norm": 13.053730010986328,
      "learning_rate": 3.873844213062151e-06,
      "loss": 3.3342,
      "step": 31015
    },
    {
      "epoch": 0.3155517578125,
      "grad_norm": 11.309110641479492,
      "learning_rate": 3.873510107385004e-06,
      "loss": 3.9116,
      "step": 31020
    },
    {
      "epoch": 0.3156026204427083,
      "grad_norm": 14.312835693359375,
      "learning_rate": 3.873175966566929e-06,
      "loss": 3.5496,
      "step": 31025
    },
    {
      "epoch": 0.3156534830729167,
      "grad_norm": 13.165633201599121,
      "learning_rate": 3.87284179061647e-06,
      "loss": 3.8198,
      "step": 31030
    },
    {
      "epoch": 0.315704345703125,
      "grad_norm": 10.730169296264648,
      "learning_rate": 3.872507579542181e-06,
      "loss": 3.2504,
      "step": 31035
    },
    {
      "epoch": 0.3157552083333333,
      "grad_norm": 14.646590232849121,
      "learning_rate": 3.872173333352609e-06,
      "loss": 3.1502,
      "step": 31040
    },
    {
      "epoch": 0.3158060709635417,
      "grad_norm": 9.55855655670166,
      "learning_rate": 3.871839052056309e-06,
      "loss": 3.1742,
      "step": 31045
    },
    {
      "epoch": 0.31585693359375,
      "grad_norm": 15.399552345275879,
      "learning_rate": 3.871504735661831e-06,
      "loss": 3.1138,
      "step": 31050
    },
    {
      "epoch": 0.3159077962239583,
      "grad_norm": 9.473359107971191,
      "learning_rate": 3.87117038417773e-06,
      "loss": 3.4158,
      "step": 31055
    },
    {
      "epoch": 0.3159586588541667,
      "grad_norm": 13.662787437438965,
      "learning_rate": 3.870835997612559e-06,
      "loss": 2.8579,
      "step": 31060
    },
    {
      "epoch": 0.316009521484375,
      "grad_norm": 15.330162048339844,
      "learning_rate": 3.870501575974875e-06,
      "loss": 3.4082,
      "step": 31065
    },
    {
      "epoch": 0.3160603841145833,
      "grad_norm": 10.36662483215332,
      "learning_rate": 3.870167119273233e-06,
      "loss": 3.5759,
      "step": 31070
    },
    {
      "epoch": 0.3161112467447917,
      "grad_norm": 12.250576972961426,
      "learning_rate": 3.869832627516188e-06,
      "loss": 3.5613,
      "step": 31075
    },
    {
      "epoch": 0.316162109375,
      "grad_norm": 10.067193031311035,
      "learning_rate": 3.8694981007123025e-06,
      "loss": 3.4331,
      "step": 31080
    },
    {
      "epoch": 0.3162129720052083,
      "grad_norm": 10.213749885559082,
      "learning_rate": 3.869163538870131e-06,
      "loss": 3.1883,
      "step": 31085
    },
    {
      "epoch": 0.3162638346354167,
      "grad_norm": 10.69174861907959,
      "learning_rate": 3.868828941998236e-06,
      "loss": 3.4121,
      "step": 31090
    },
    {
      "epoch": 0.316314697265625,
      "grad_norm": 11.320060729980469,
      "learning_rate": 3.8684943101051754e-06,
      "loss": 3.309,
      "step": 31095
    },
    {
      "epoch": 0.3163655598958333,
      "grad_norm": 15.44389820098877,
      "learning_rate": 3.868159643199514e-06,
      "loss": 3.4406,
      "step": 31100
    },
    {
      "epoch": 0.3164164225260417,
      "grad_norm": 12.214240074157715,
      "learning_rate": 3.867824941289812e-06,
      "loss": 3.4774,
      "step": 31105
    },
    {
      "epoch": 0.31646728515625,
      "grad_norm": 10.657689094543457,
      "learning_rate": 3.867490204384632e-06,
      "loss": 3.3771,
      "step": 31110
    },
    {
      "epoch": 0.3165181477864583,
      "grad_norm": 11.51058578491211,
      "learning_rate": 3.8671554324925405e-06,
      "loss": 3.4663,
      "step": 31115
    },
    {
      "epoch": 0.3165690104166667,
      "grad_norm": 9.098102569580078,
      "learning_rate": 3.8668206256221005e-06,
      "loss": 3.5193,
      "step": 31120
    },
    {
      "epoch": 0.316619873046875,
      "grad_norm": 13.172540664672852,
      "learning_rate": 3.866485783781879e-06,
      "loss": 3.3625,
      "step": 31125
    },
    {
      "epoch": 0.3166707356770833,
      "grad_norm": 10.455148696899414,
      "learning_rate": 3.866150906980444e-06,
      "loss": 3.1612,
      "step": 31130
    },
    {
      "epoch": 0.3167215983072917,
      "grad_norm": 12.411478042602539,
      "learning_rate": 3.865815995226359e-06,
      "loss": 3.2446,
      "step": 31135
    },
    {
      "epoch": 0.3167724609375,
      "grad_norm": 11.790022850036621,
      "learning_rate": 3.865481048528198e-06,
      "loss": 3.383,
      "step": 31140
    },
    {
      "epoch": 0.3168233235677083,
      "grad_norm": 9.37701416015625,
      "learning_rate": 3.8651460668945274e-06,
      "loss": 3.336,
      "step": 31145
    },
    {
      "epoch": 0.3168741861979167,
      "grad_norm": 13.848721504211426,
      "learning_rate": 3.864811050333917e-06,
      "loss": 3.6014,
      "step": 31150
    },
    {
      "epoch": 0.316925048828125,
      "grad_norm": 14.876688003540039,
      "learning_rate": 3.864475998854941e-06,
      "loss": 3.5189,
      "step": 31155
    },
    {
      "epoch": 0.3169759114583333,
      "grad_norm": 11.406241416931152,
      "learning_rate": 3.864140912466169e-06,
      "loss": 3.3236,
      "step": 31160
    },
    {
      "epoch": 0.3170267740885417,
      "grad_norm": 15.621129035949707,
      "learning_rate": 3.863805791176176e-06,
      "loss": 3.897,
      "step": 31165
    },
    {
      "epoch": 0.31707763671875,
      "grad_norm": 12.045602798461914,
      "learning_rate": 3.8634706349935344e-06,
      "loss": 3.459,
      "step": 31170
    },
    {
      "epoch": 0.3171284993489583,
      "grad_norm": 7.246892929077148,
      "learning_rate": 3.8631354439268195e-06,
      "loss": 3.3646,
      "step": 31175
    },
    {
      "epoch": 0.3171793619791667,
      "grad_norm": 15.568187713623047,
      "learning_rate": 3.862800217984608e-06,
      "loss": 3.4646,
      "step": 31180
    },
    {
      "epoch": 0.317230224609375,
      "grad_norm": 9.922911643981934,
      "learning_rate": 3.862464957175476e-06,
      "loss": 3.3717,
      "step": 31185
    },
    {
      "epoch": 0.3172810872395833,
      "grad_norm": 12.449997901916504,
      "learning_rate": 3.862129661508001e-06,
      "loss": 3.4387,
      "step": 31190
    },
    {
      "epoch": 0.3173319498697917,
      "grad_norm": 16.325824737548828,
      "learning_rate": 3.8617943309907605e-06,
      "loss": 3.14,
      "step": 31195
    },
    {
      "epoch": 0.3173828125,
      "grad_norm": 15.284937858581543,
      "learning_rate": 3.861458965632335e-06,
      "loss": 3.3294,
      "step": 31200
    },
    {
      "epoch": 0.3174336751302083,
      "grad_norm": 11.433347702026367,
      "learning_rate": 3.861123565441306e-06,
      "loss": 3.334,
      "step": 31205
    },
    {
      "epoch": 0.3174845377604167,
      "grad_norm": 12.585484504699707,
      "learning_rate": 3.860788130426252e-06,
      "loss": 3.3748,
      "step": 31210
    },
    {
      "epoch": 0.317535400390625,
      "grad_norm": 16.614295959472656,
      "learning_rate": 3.860452660595757e-06,
      "loss": 3.1318,
      "step": 31215
    },
    {
      "epoch": 0.3175862630208333,
      "grad_norm": 9.032590866088867,
      "learning_rate": 3.860117155958403e-06,
      "loss": 3.4407,
      "step": 31220
    },
    {
      "epoch": 0.3176371256510417,
      "grad_norm": 6.906585693359375,
      "learning_rate": 3.859781616522773e-06,
      "loss": 3.8926,
      "step": 31225
    },
    {
      "epoch": 0.31768798828125,
      "grad_norm": 15.229521751403809,
      "learning_rate": 3.859446042297453e-06,
      "loss": 3.289,
      "step": 31230
    },
    {
      "epoch": 0.3177388509114583,
      "grad_norm": 8.980449676513672,
      "learning_rate": 3.859110433291029e-06,
      "loss": 3.2042,
      "step": 31235
    },
    {
      "epoch": 0.3177897135416667,
      "grad_norm": 12.09065055847168,
      "learning_rate": 3.858774789512086e-06,
      "loss": 3.2158,
      "step": 31240
    },
    {
      "epoch": 0.317840576171875,
      "grad_norm": 10.273479461669922,
      "learning_rate": 3.858439110969212e-06,
      "loss": 3.4836,
      "step": 31245
    },
    {
      "epoch": 0.3178914388020833,
      "grad_norm": 12.64652156829834,
      "learning_rate": 3.858103397670996e-06,
      "loss": 3.1187,
      "step": 31250
    },
    {
      "epoch": 0.3179423014322917,
      "grad_norm": 14.755744934082031,
      "learning_rate": 3.8577676496260255e-06,
      "loss": 3.2408,
      "step": 31255
    },
    {
      "epoch": 0.3179931640625,
      "grad_norm": 11.015496253967285,
      "learning_rate": 3.8574318668428925e-06,
      "loss": 3.2791,
      "step": 31260
    },
    {
      "epoch": 0.3180440266927083,
      "grad_norm": 11.528286933898926,
      "learning_rate": 3.857096049330186e-06,
      "loss": 3.1186,
      "step": 31265
    },
    {
      "epoch": 0.3180948893229167,
      "grad_norm": 11.770857810974121,
      "learning_rate": 3.8567601970964995e-06,
      "loss": 3.3614,
      "step": 31270
    },
    {
      "epoch": 0.318145751953125,
      "grad_norm": 16.619205474853516,
      "learning_rate": 3.856424310150425e-06,
      "loss": 3.3688,
      "step": 31275
    },
    {
      "epoch": 0.3181966145833333,
      "grad_norm": 14.829237937927246,
      "learning_rate": 3.856088388500555e-06,
      "loss": 3.5595,
      "step": 31280
    },
    {
      "epoch": 0.3182474772135417,
      "grad_norm": 10.562813758850098,
      "learning_rate": 3.855752432155486e-06,
      "loss": 2.9822,
      "step": 31285
    },
    {
      "epoch": 0.31829833984375,
      "grad_norm": 14.399613380432129,
      "learning_rate": 3.855416441123813e-06,
      "loss": 3.4562,
      "step": 31290
    },
    {
      "epoch": 0.3183492024739583,
      "grad_norm": 9.176456451416016,
      "learning_rate": 3.85508041541413e-06,
      "loss": 3.3423,
      "step": 31295
    },
    {
      "epoch": 0.3184000651041667,
      "grad_norm": 11.619348526000977,
      "learning_rate": 3.8547443550350364e-06,
      "loss": 3.552,
      "step": 31300
    },
    {
      "epoch": 0.318450927734375,
      "grad_norm": 11.135879516601562,
      "learning_rate": 3.85440825999513e-06,
      "loss": 3.3815,
      "step": 31305
    },
    {
      "epoch": 0.3185017903645833,
      "grad_norm": 12.544413566589355,
      "learning_rate": 3.8540721303030095e-06,
      "loss": 3.3132,
      "step": 31310
    },
    {
      "epoch": 0.3185526529947917,
      "grad_norm": 12.913429260253906,
      "learning_rate": 3.8537359659672735e-06,
      "loss": 3.1569,
      "step": 31315
    },
    {
      "epoch": 0.318603515625,
      "grad_norm": 11.757790565490723,
      "learning_rate": 3.853399766996524e-06,
      "loss": 3.0326,
      "step": 31320
    },
    {
      "epoch": 0.3186543782552083,
      "grad_norm": 11.700207710266113,
      "learning_rate": 3.853063533399363e-06,
      "loss": 3.2504,
      "step": 31325
    },
    {
      "epoch": 0.3187052408854167,
      "grad_norm": 8.301576614379883,
      "learning_rate": 3.852727265184391e-06,
      "loss": 3.5017,
      "step": 31330
    },
    {
      "epoch": 0.318756103515625,
      "grad_norm": 12.666961669921875,
      "learning_rate": 3.852390962360213e-06,
      "loss": 3.9952,
      "step": 31335
    },
    {
      "epoch": 0.3188069661458333,
      "grad_norm": 14.605937957763672,
      "learning_rate": 3.852054624935433e-06,
      "loss": 3.2524,
      "step": 31340
    },
    {
      "epoch": 0.3188578287760417,
      "grad_norm": 11.621174812316895,
      "learning_rate": 3.851718252918656e-06,
      "loss": 3.6715,
      "step": 31345
    },
    {
      "epoch": 0.31890869140625,
      "grad_norm": 9.262434959411621,
      "learning_rate": 3.8513818463184874e-06,
      "loss": 3.0173,
      "step": 31350
    },
    {
      "epoch": 0.3189595540364583,
      "grad_norm": 13.999554634094238,
      "learning_rate": 3.851045405143535e-06,
      "loss": 3.2864,
      "step": 31355
    },
    {
      "epoch": 0.3190104166666667,
      "grad_norm": 12.603280067443848,
      "learning_rate": 3.850708929402406e-06,
      "loss": 3.3036,
      "step": 31360
    },
    {
      "epoch": 0.319061279296875,
      "grad_norm": 11.629323959350586,
      "learning_rate": 3.8503724191037095e-06,
      "loss": 3.4563,
      "step": 31365
    },
    {
      "epoch": 0.3191121419270833,
      "grad_norm": 10.602560043334961,
      "learning_rate": 3.850035874256054e-06,
      "loss": 3.4163,
      "step": 31370
    },
    {
      "epoch": 0.3191630045572917,
      "grad_norm": 14.061223030090332,
      "learning_rate": 3.849699294868051e-06,
      "loss": 3.0912,
      "step": 31375
    },
    {
      "epoch": 0.3192138671875,
      "grad_norm": 9.001642227172852,
      "learning_rate": 3.849362680948312e-06,
      "loss": 3.3378,
      "step": 31380
    },
    {
      "epoch": 0.3192647298177083,
      "grad_norm": 12.85893726348877,
      "learning_rate": 3.849026032505449e-06,
      "loss": 3.6936,
      "step": 31385
    },
    {
      "epoch": 0.3193155924479167,
      "grad_norm": 15.685237884521484,
      "learning_rate": 3.848689349548073e-06,
      "loss": 3.2416,
      "step": 31390
    },
    {
      "epoch": 0.319366455078125,
      "grad_norm": 11.73633861541748,
      "learning_rate": 3.848352632084801e-06,
      "loss": 3.5264,
      "step": 31395
    },
    {
      "epoch": 0.3194173177083333,
      "grad_norm": 14.574155807495117,
      "learning_rate": 3.8480158801242464e-06,
      "loss": 3.4237,
      "step": 31400
    },
    {
      "epoch": 0.3194681803385417,
      "grad_norm": 9.669441223144531,
      "learning_rate": 3.847679093675024e-06,
      "loss": 3.1957,
      "step": 31405
    },
    {
      "epoch": 0.31951904296875,
      "grad_norm": 17.44133186340332,
      "learning_rate": 3.847342272745752e-06,
      "loss": 3.8577,
      "step": 31410
    },
    {
      "epoch": 0.3195699055989583,
      "grad_norm": 8.022843360900879,
      "learning_rate": 3.847005417345048e-06,
      "loss": 3.0239,
      "step": 31415
    },
    {
      "epoch": 0.3196207682291667,
      "grad_norm": 9.9092435836792,
      "learning_rate": 3.846668527481529e-06,
      "loss": 2.864,
      "step": 31420
    },
    {
      "epoch": 0.319671630859375,
      "grad_norm": 15.318598747253418,
      "learning_rate": 3.846331603163815e-06,
      "loss": 3.4041,
      "step": 31425
    },
    {
      "epoch": 0.3197224934895833,
      "grad_norm": 9.314645767211914,
      "learning_rate": 3.845994644400526e-06,
      "loss": 3.4976,
      "step": 31430
    },
    {
      "epoch": 0.3197733561197917,
      "grad_norm": 14.026188850402832,
      "learning_rate": 3.8456576512002836e-06,
      "loss": 3.3014,
      "step": 31435
    },
    {
      "epoch": 0.31982421875,
      "grad_norm": 15.728029251098633,
      "learning_rate": 3.845320623571708e-06,
      "loss": 2.8761,
      "step": 31440
    },
    {
      "epoch": 0.3198750813802083,
      "grad_norm": 8.416905403137207,
      "learning_rate": 3.8449835615234244e-06,
      "loss": 3.8231,
      "step": 31445
    },
    {
      "epoch": 0.3199259440104167,
      "grad_norm": 10.172744750976562,
      "learning_rate": 3.844646465064055e-06,
      "loss": 3.4616,
      "step": 31450
    },
    {
      "epoch": 0.319976806640625,
      "grad_norm": 12.25190258026123,
      "learning_rate": 3.844309334202223e-06,
      "loss": 3.1986,
      "step": 31455
    },
    {
      "epoch": 0.3200276692708333,
      "grad_norm": 12.856034278869629,
      "learning_rate": 3.843972168946557e-06,
      "loss": 3.1889,
      "step": 31460
    },
    {
      "epoch": 0.3200785319010417,
      "grad_norm": 14.786962509155273,
      "learning_rate": 3.843634969305681e-06,
      "loss": 3.4426,
      "step": 31465
    },
    {
      "epoch": 0.32012939453125,
      "grad_norm": 12.744573593139648,
      "learning_rate": 3.8432977352882225e-06,
      "loss": 3.2549,
      "step": 31470
    },
    {
      "epoch": 0.3201802571614583,
      "grad_norm": 14.236984252929688,
      "learning_rate": 3.842960466902811e-06,
      "loss": 3.6449,
      "step": 31475
    },
    {
      "epoch": 0.3202311197916667,
      "grad_norm": 14.349587440490723,
      "learning_rate": 3.8426231641580725e-06,
      "loss": 3.1954,
      "step": 31480
    },
    {
      "epoch": 0.320281982421875,
      "grad_norm": 13.996590614318848,
      "learning_rate": 3.84228582706264e-06,
      "loss": 3.6826,
      "step": 31485
    },
    {
      "epoch": 0.3203328450520833,
      "grad_norm": 10.456219673156738,
      "learning_rate": 3.841948455625142e-06,
      "loss": 3.6421,
      "step": 31490
    },
    {
      "epoch": 0.3203837076822917,
      "grad_norm": 10.800615310668945,
      "learning_rate": 3.841611049854211e-06,
      "loss": 3.5236,
      "step": 31495
    },
    {
      "epoch": 0.3204345703125,
      "grad_norm": 11.609787940979004,
      "learning_rate": 3.84127360975848e-06,
      "loss": 3.4777,
      "step": 31500
    },
    {
      "epoch": 0.3204854329427083,
      "grad_norm": 6.661224842071533,
      "learning_rate": 3.84093613534658e-06,
      "loss": 3.431,
      "step": 31505
    },
    {
      "epoch": 0.3205362955729167,
      "grad_norm": 14.083808898925781,
      "learning_rate": 3.840598626627149e-06,
      "loss": 3.5221,
      "step": 31510
    },
    {
      "epoch": 0.320587158203125,
      "grad_norm": 12.521286964416504,
      "learning_rate": 3.840261083608818e-06,
      "loss": 3.9192,
      "step": 31515
    },
    {
      "epoch": 0.3206380208333333,
      "grad_norm": 15.08007526397705,
      "learning_rate": 3.839923506300226e-06,
      "loss": 3.5092,
      "step": 31520
    },
    {
      "epoch": 0.3206888834635417,
      "grad_norm": 14.394757270812988,
      "learning_rate": 3.839585894710009e-06,
      "loss": 3.295,
      "step": 31525
    },
    {
      "epoch": 0.32073974609375,
      "grad_norm": 13.734356880187988,
      "learning_rate": 3.839248248846803e-06,
      "loss": 3.5945,
      "step": 31530
    },
    {
      "epoch": 0.3207906087239583,
      "grad_norm": 16.14764976501465,
      "learning_rate": 3.838910568719249e-06,
      "loss": 3.6131,
      "step": 31535
    },
    {
      "epoch": 0.3208414713541667,
      "grad_norm": 14.201319694519043,
      "learning_rate": 3.838572854335985e-06,
      "loss": 3.2369,
      "step": 31540
    },
    {
      "epoch": 0.320892333984375,
      "grad_norm": 13.397239685058594,
      "learning_rate": 3.838235105705651e-06,
      "loss": 3.8006,
      "step": 31545
    },
    {
      "epoch": 0.3209431966145833,
      "grad_norm": 12.374748229980469,
      "learning_rate": 3.837897322836891e-06,
      "loss": 3.2664,
      "step": 31550
    },
    {
      "epoch": 0.3209940592447917,
      "grad_norm": 14.16633129119873,
      "learning_rate": 3.8375595057383445e-06,
      "loss": 3.7635,
      "step": 31555
    },
    {
      "epoch": 0.321044921875,
      "grad_norm": 13.520079612731934,
      "learning_rate": 3.837221654418654e-06,
      "loss": 3.4169,
      "step": 31560
    },
    {
      "epoch": 0.3210957845052083,
      "grad_norm": 14.925244331359863,
      "learning_rate": 3.836883768886465e-06,
      "loss": 3.5138,
      "step": 31565
    },
    {
      "epoch": 0.3211466471354167,
      "grad_norm": 10.590404510498047,
      "learning_rate": 3.836545849150421e-06,
      "loss": 3.2101,
      "step": 31570
    },
    {
      "epoch": 0.321197509765625,
      "grad_norm": 14.485934257507324,
      "learning_rate": 3.836207895219169e-06,
      "loss": 3.1994,
      "step": 31575
    },
    {
      "epoch": 0.3212483723958333,
      "grad_norm": 9.394538879394531,
      "learning_rate": 3.835869907101355e-06,
      "loss": 3.2738,
      "step": 31580
    },
    {
      "epoch": 0.3212992350260417,
      "grad_norm": 9.60177230834961,
      "learning_rate": 3.835531884805626e-06,
      "loss": 3.0774,
      "step": 31585
    },
    {
      "epoch": 0.32135009765625,
      "grad_norm": 17.322458267211914,
      "learning_rate": 3.835193828340629e-06,
      "loss": 3.6144,
      "step": 31590
    },
    {
      "epoch": 0.3214009602864583,
      "grad_norm": 13.787790298461914,
      "learning_rate": 3.834855737715016e-06,
      "loss": 3.3578,
      "step": 31595
    },
    {
      "epoch": 0.3214518229166667,
      "grad_norm": 8.987893104553223,
      "learning_rate": 3.834517612937435e-06,
      "loss": 3.3803,
      "step": 31600
    },
    {
      "epoch": 0.321502685546875,
      "grad_norm": 16.33568572998047,
      "learning_rate": 3.834179454016537e-06,
      "loss": 3.4638,
      "step": 31605
    },
    {
      "epoch": 0.3215535481770833,
      "grad_norm": 11.19406509399414,
      "learning_rate": 3.833841260960973e-06,
      "loss": 3.3542,
      "step": 31610
    },
    {
      "epoch": 0.3216044108072917,
      "grad_norm": 14.263209342956543,
      "learning_rate": 3.833503033779398e-06,
      "loss": 3.1218,
      "step": 31615
    },
    {
      "epoch": 0.3216552734375,
      "grad_norm": 6.953857898712158,
      "learning_rate": 3.833164772480463e-06,
      "loss": 3.3436,
      "step": 31620
    },
    {
      "epoch": 0.3217061360677083,
      "grad_norm": 8.796978950500488,
      "learning_rate": 3.832826477072824e-06,
      "loss": 3.6858,
      "step": 31625
    },
    {
      "epoch": 0.3217569986979167,
      "grad_norm": 12.743858337402344,
      "learning_rate": 3.832488147565135e-06,
      "loss": 3.4351,
      "step": 31630
    },
    {
      "epoch": 0.321807861328125,
      "grad_norm": 10.61341667175293,
      "learning_rate": 3.832149783966053e-06,
      "loss": 3.5864,
      "step": 31635
    },
    {
      "epoch": 0.3218587239583333,
      "grad_norm": 12.128693580627441,
      "learning_rate": 3.831811386284234e-06,
      "loss": 3.103,
      "step": 31640
    },
    {
      "epoch": 0.3219095865885417,
      "grad_norm": 13.619905471801758,
      "learning_rate": 3.831472954528336e-06,
      "loss": 3.085,
      "step": 31645
    },
    {
      "epoch": 0.32196044921875,
      "grad_norm": 8.86436653137207,
      "learning_rate": 3.831134488707019e-06,
      "loss": 3.3038,
      "step": 31650
    },
    {
      "epoch": 0.3220113118489583,
      "grad_norm": 10.96633529663086,
      "learning_rate": 3.830795988828941e-06,
      "loss": 3.3176,
      "step": 31655
    },
    {
      "epoch": 0.3220621744791667,
      "grad_norm": 11.57568073272705,
      "learning_rate": 3.830457454902763e-06,
      "loss": 3.305,
      "step": 31660
    },
    {
      "epoch": 0.322113037109375,
      "grad_norm": 14.083612442016602,
      "learning_rate": 3.830118886937146e-06,
      "loss": 4.0346,
      "step": 31665
    },
    {
      "epoch": 0.3221638997395833,
      "grad_norm": 9.294708251953125,
      "learning_rate": 3.829780284940755e-06,
      "loss": 3.2512,
      "step": 31670
    },
    {
      "epoch": 0.3222147623697917,
      "grad_norm": 19.22852325439453,
      "learning_rate": 3.829441648922247e-06,
      "loss": 3.9869,
      "step": 31675
    },
    {
      "epoch": 0.322265625,
      "grad_norm": 14.847890853881836,
      "learning_rate": 3.8291029788902925e-06,
      "loss": 2.9014,
      "step": 31680
    },
    {
      "epoch": 0.3223164876302083,
      "grad_norm": 15.327040672302246,
      "learning_rate": 3.828764274853552e-06,
      "loss": 3.4145,
      "step": 31685
    },
    {
      "epoch": 0.3223673502604167,
      "grad_norm": 16.938589096069336,
      "learning_rate": 3.828425536820693e-06,
      "loss": 3.2736,
      "step": 31690
    },
    {
      "epoch": 0.322418212890625,
      "grad_norm": 10.017430305480957,
      "learning_rate": 3.828086764800381e-06,
      "loss": 3.7319,
      "step": 31695
    },
    {
      "epoch": 0.3224690755208333,
      "grad_norm": 14.776453018188477,
      "learning_rate": 3.827747958801283e-06,
      "loss": 3.64,
      "step": 31700
    },
    {
      "epoch": 0.3225199381510417,
      "grad_norm": 7.89909553527832,
      "learning_rate": 3.8274091188320695e-06,
      "loss": 3.294,
      "step": 31705
    },
    {
      "epoch": 0.32257080078125,
      "grad_norm": 13.75835132598877,
      "learning_rate": 3.8270702449014085e-06,
      "loss": 3.744,
      "step": 31710
    },
    {
      "epoch": 0.3226216634114583,
      "grad_norm": 10.914870262145996,
      "learning_rate": 3.82673133701797e-06,
      "loss": 3.1938,
      "step": 31715
    },
    {
      "epoch": 0.3226725260416667,
      "grad_norm": 17.479778289794922,
      "learning_rate": 3.826392395190424e-06,
      "loss": 3.3668,
      "step": 31720
    },
    {
      "epoch": 0.322723388671875,
      "grad_norm": 11.222243309020996,
      "learning_rate": 3.826053419427443e-06,
      "loss": 3.3123,
      "step": 31725
    },
    {
      "epoch": 0.3227742513020833,
      "grad_norm": 11.842264175415039,
      "learning_rate": 3.825714409737699e-06,
      "loss": 3.3255,
      "step": 31730
    },
    {
      "epoch": 0.3228251139322917,
      "grad_norm": 7.541260719299316,
      "learning_rate": 3.825375366129867e-06,
      "loss": 3.4824,
      "step": 31735
    },
    {
      "epoch": 0.3228759765625,
      "grad_norm": 11.848604202270508,
      "learning_rate": 3.82503628861262e-06,
      "loss": 3.7925,
      "step": 31740
    },
    {
      "epoch": 0.3229268391927083,
      "grad_norm": 11.389389038085938,
      "learning_rate": 3.824697177194634e-06,
      "loss": 3.0986,
      "step": 31745
    },
    {
      "epoch": 0.3229777018229167,
      "grad_norm": 9.12468147277832,
      "learning_rate": 3.824358031884584e-06,
      "loss": 3.3481,
      "step": 31750
    },
    {
      "epoch": 0.323028564453125,
      "grad_norm": 10.102771759033203,
      "learning_rate": 3.824018852691148e-06,
      "loss": 3.6813,
      "step": 31755
    },
    {
      "epoch": 0.3230794270833333,
      "grad_norm": 9.482272148132324,
      "learning_rate": 3.823679639623004e-06,
      "loss": 3.5337,
      "step": 31760
    },
    {
      "epoch": 0.3231302897135417,
      "grad_norm": 9.679491996765137,
      "learning_rate": 3.823340392688829e-06,
      "loss": 3.4032,
      "step": 31765
    },
    {
      "epoch": 0.32318115234375,
      "grad_norm": 13.405425071716309,
      "learning_rate": 3.823001111897304e-06,
      "loss": 3.4294,
      "step": 31770
    },
    {
      "epoch": 0.3232320149739583,
      "grad_norm": 7.808204174041748,
      "learning_rate": 3.822661797257111e-06,
      "loss": 3.0086,
      "step": 31775
    },
    {
      "epoch": 0.3232828776041667,
      "grad_norm": 12.230095863342285,
      "learning_rate": 3.822322448776927e-06,
      "loss": 3.2176,
      "step": 31780
    },
    {
      "epoch": 0.323333740234375,
      "grad_norm": 10.221451759338379,
      "learning_rate": 3.821983066465438e-06,
      "loss": 3.34,
      "step": 31785
    },
    {
      "epoch": 0.3233846028645833,
      "grad_norm": 11.985438346862793,
      "learning_rate": 3.821643650331325e-06,
      "loss": 2.9988,
      "step": 31790
    },
    {
      "epoch": 0.3234354654947917,
      "grad_norm": 18.49512481689453,
      "learning_rate": 3.821304200383273e-06,
      "loss": 3.5417,
      "step": 31795
    },
    {
      "epoch": 0.323486328125,
      "grad_norm": 9.309732437133789,
      "learning_rate": 3.820964716629965e-06,
      "loss": 3.3165,
      "step": 31800
    },
    {
      "epoch": 0.3235371907552083,
      "grad_norm": 10.926666259765625,
      "learning_rate": 3.8206251990800894e-06,
      "loss": 3.4721,
      "step": 31805
    },
    {
      "epoch": 0.3235880533854167,
      "grad_norm": 11.638176918029785,
      "learning_rate": 3.82028564774233e-06,
      "loss": 3.2567,
      "step": 31810
    },
    {
      "epoch": 0.323638916015625,
      "grad_norm": 15.134770393371582,
      "learning_rate": 3.819946062625376e-06,
      "loss": 3.7382,
      "step": 31815
    },
    {
      "epoch": 0.3236897786458333,
      "grad_norm": 8.549654006958008,
      "learning_rate": 3.819606443737914e-06,
      "loss": 3.0723,
      "step": 31820
    },
    {
      "epoch": 0.3237406412760417,
      "grad_norm": 9.45890998840332,
      "learning_rate": 3.819266791088635e-06,
      "loss": 3.608,
      "step": 31825
    },
    {
      "epoch": 0.32379150390625,
      "grad_norm": 10.157885551452637,
      "learning_rate": 3.818927104686226e-06,
      "loss": 3.3638,
      "step": 31830
    },
    {
      "epoch": 0.3238423665364583,
      "grad_norm": 12.008575439453125,
      "learning_rate": 3.818587384539381e-06,
      "loss": 2.9182,
      "step": 31835
    },
    {
      "epoch": 0.3238932291666667,
      "grad_norm": 13.765893936157227,
      "learning_rate": 3.818247630656791e-06,
      "loss": 3.6027,
      "step": 31840
    },
    {
      "epoch": 0.323944091796875,
      "grad_norm": 12.659904479980469,
      "learning_rate": 3.817907843047148e-06,
      "loss": 3.6605,
      "step": 31845
    },
    {
      "epoch": 0.3239949544270833,
      "grad_norm": 12.302201271057129,
      "learning_rate": 3.817568021719143e-06,
      "loss": 3.5059,
      "step": 31850
    },
    {
      "epoch": 0.3240458170572917,
      "grad_norm": 8.679214477539062,
      "learning_rate": 3.817228166681474e-06,
      "loss": 3.3955,
      "step": 31855
    },
    {
      "epoch": 0.3240966796875,
      "grad_norm": 11.400491714477539,
      "learning_rate": 3.816888277942834e-06,
      "loss": 3.4571,
      "step": 31860
    },
    {
      "epoch": 0.3241475423177083,
      "grad_norm": 11.593133926391602,
      "learning_rate": 3.81654835551192e-06,
      "loss": 3.5486,
      "step": 31865
    },
    {
      "epoch": 0.3241984049479167,
      "grad_norm": 16.467374801635742,
      "learning_rate": 3.816208399397428e-06,
      "loss": 3.3701,
      "step": 31870
    },
    {
      "epoch": 0.324249267578125,
      "grad_norm": 16.079572677612305,
      "learning_rate": 3.815868409608056e-06,
      "loss": 3.5682,
      "step": 31875
    },
    {
      "epoch": 0.3243001302083333,
      "grad_norm": 17.888187408447266,
      "learning_rate": 3.815528386152504e-06,
      "loss": 3.4161,
      "step": 31880
    },
    {
      "epoch": 0.3243509928385417,
      "grad_norm": 17.085041046142578,
      "learning_rate": 3.815188329039468e-06,
      "loss": 3.2796,
      "step": 31885
    },
    {
      "epoch": 0.32440185546875,
      "grad_norm": 12.836435317993164,
      "learning_rate": 3.814848238277652e-06,
      "loss": 3.4633,
      "step": 31890
    },
    {
      "epoch": 0.3244527180989583,
      "grad_norm": 18.368680953979492,
      "learning_rate": 3.8145081138757543e-06,
      "loss": 3.8005,
      "step": 31895
    },
    {
      "epoch": 0.3245035807291667,
      "grad_norm": 11.071703910827637,
      "learning_rate": 3.8141679558424786e-06,
      "loss": 3.3738,
      "step": 31900
    },
    {
      "epoch": 0.324554443359375,
      "grad_norm": 13.304140090942383,
      "learning_rate": 3.8138277641865274e-06,
      "loss": 2.9373,
      "step": 31905
    },
    {
      "epoch": 0.3246053059895833,
      "grad_norm": 53.66118621826172,
      "learning_rate": 3.813487538916604e-06,
      "loss": 3.8442,
      "step": 31910
    },
    {
      "epoch": 0.3246561686197917,
      "grad_norm": 10.714417457580566,
      "learning_rate": 3.813147280041414e-06,
      "loss": 3.2688,
      "step": 31915
    },
    {
      "epoch": 0.32470703125,
      "grad_norm": 10.948854446411133,
      "learning_rate": 3.8128069875696603e-06,
      "loss": 3.278,
      "step": 31920
    },
    {
      "epoch": 0.3247578938802083,
      "grad_norm": 8.774290084838867,
      "learning_rate": 3.8124666615100523e-06,
      "loss": 3.4067,
      "step": 31925
    },
    {
      "epoch": 0.3248087565104167,
      "grad_norm": 8.86909294128418,
      "learning_rate": 3.812126301871296e-06,
      "loss": 3.3751,
      "step": 31930
    },
    {
      "epoch": 0.324859619140625,
      "grad_norm": 11.547953605651855,
      "learning_rate": 3.8117859086620988e-06,
      "loss": 3.2366,
      "step": 31935
    },
    {
      "epoch": 0.3249104817708333,
      "grad_norm": 7.339703559875488,
      "learning_rate": 3.8114454818911707e-06,
      "loss": 3.2675,
      "step": 31940
    },
    {
      "epoch": 0.3249613444010417,
      "grad_norm": 13.766386032104492,
      "learning_rate": 3.8111050215672196e-06,
      "loss": 3.3403,
      "step": 31945
    },
    {
      "epoch": 0.32501220703125,
      "grad_norm": 8.798249244689941,
      "learning_rate": 3.810764527698958e-06,
      "loss": 3.3275,
      "step": 31950
    },
    {
      "epoch": 0.3250630696614583,
      "grad_norm": 12.045044898986816,
      "learning_rate": 3.8104240002950963e-06,
      "loss": 3.8319,
      "step": 31955
    },
    {
      "epoch": 0.3251139322916667,
      "grad_norm": 13.24450969696045,
      "learning_rate": 3.810083439364347e-06,
      "loss": 3.4244,
      "step": 31960
    },
    {
      "epoch": 0.325164794921875,
      "grad_norm": 12.002233505249023,
      "learning_rate": 3.809742844915425e-06,
      "loss": 3.5424,
      "step": 31965
    },
    {
      "epoch": 0.3252156575520833,
      "grad_norm": 7.133974552154541,
      "learning_rate": 3.8094022169570412e-06,
      "loss": 3.2509,
      "step": 31970
    },
    {
      "epoch": 0.3252665201822917,
      "grad_norm": 12.657421112060547,
      "learning_rate": 3.8090615554979127e-06,
      "loss": 3.1012,
      "step": 31975
    },
    {
      "epoch": 0.3253173828125,
      "grad_norm": 14.409326553344727,
      "learning_rate": 3.808720860546754e-06,
      "loss": 3.3739,
      "step": 31980
    },
    {
      "epoch": 0.3253682454427083,
      "grad_norm": 11.968620300292969,
      "learning_rate": 3.8083801321122824e-06,
      "loss": 3.1848,
      "step": 31985
    },
    {
      "epoch": 0.3254191080729167,
      "grad_norm": 11.809328079223633,
      "learning_rate": 3.808039370203216e-06,
      "loss": 3.3059,
      "step": 31990
    },
    {
      "epoch": 0.325469970703125,
      "grad_norm": 17.407115936279297,
      "learning_rate": 3.8076985748282725e-06,
      "loss": 3.6285,
      "step": 31995
    },
    {
      "epoch": 0.3255208333333333,
      "grad_norm": 7.9313483238220215,
      "learning_rate": 3.8073577459961697e-06,
      "loss": 3.3734,
      "step": 32000
    },
    {
      "epoch": 0.3255716959635417,
      "grad_norm": 8.173722267150879,
      "learning_rate": 3.8070168837156294e-06,
      "loss": 3.6385,
      "step": 32005
    },
    {
      "epoch": 0.32562255859375,
      "grad_norm": 11.985635757446289,
      "learning_rate": 3.806675987995372e-06,
      "loss": 3.2194,
      "step": 32010
    },
    {
      "epoch": 0.3256734212239583,
      "grad_norm": 10.79389762878418,
      "learning_rate": 3.8063350588441194e-06,
      "loss": 3.2867,
      "step": 32015
    },
    {
      "epoch": 0.3257242838541667,
      "grad_norm": 11.685125350952148,
      "learning_rate": 3.8059940962705934e-06,
      "loss": 3.2666,
      "step": 32020
    },
    {
      "epoch": 0.325775146484375,
      "grad_norm": 11.623699188232422,
      "learning_rate": 3.8056531002835184e-06,
      "loss": 3.4345,
      "step": 32025
    },
    {
      "epoch": 0.3258260091145833,
      "grad_norm": 8.528718948364258,
      "learning_rate": 3.805312070891618e-06,
      "loss": 3.0183,
      "step": 32030
    },
    {
      "epoch": 0.3258768717447917,
      "grad_norm": 8.97779655456543,
      "learning_rate": 3.8049710081036174e-06,
      "loss": 3.2587,
      "step": 32035
    },
    {
      "epoch": 0.325927734375,
      "grad_norm": 8.91556167602539,
      "learning_rate": 3.8046299119282437e-06,
      "loss": 3.3258,
      "step": 32040
    },
    {
      "epoch": 0.3259785970052083,
      "grad_norm": 9.004982948303223,
      "learning_rate": 3.8042887823742224e-06,
      "loss": 3.235,
      "step": 32045
    },
    {
      "epoch": 0.3260294596354167,
      "grad_norm": 12.783254623413086,
      "learning_rate": 3.8039476194502816e-06,
      "loss": 3.56,
      "step": 32050
    },
    {
      "epoch": 0.326080322265625,
      "grad_norm": 8.640666961669922,
      "learning_rate": 3.8036064231651502e-06,
      "loss": 3.0825,
      "step": 32055
    },
    {
      "epoch": 0.3261311848958333,
      "grad_norm": 15.606740951538086,
      "learning_rate": 3.8032651935275583e-06,
      "loss": 3.2792,
      "step": 32060
    },
    {
      "epoch": 0.3261820475260417,
      "grad_norm": 12.018491744995117,
      "learning_rate": 3.802923930546234e-06,
      "loss": 3.4059,
      "step": 32065
    },
    {
      "epoch": 0.32623291015625,
      "grad_norm": 13.65191650390625,
      "learning_rate": 3.802582634229911e-06,
      "loss": 3.4325,
      "step": 32070
    },
    {
      "epoch": 0.3262837727864583,
      "grad_norm": 11.970004081726074,
      "learning_rate": 3.802241304587319e-06,
      "loss": 3.4715,
      "step": 32075
    },
    {
      "epoch": 0.3263346354166667,
      "grad_norm": 11.346258163452148,
      "learning_rate": 3.8018999416271924e-06,
      "loss": 3.4341,
      "step": 32080
    },
    {
      "epoch": 0.326385498046875,
      "grad_norm": 8.91111946105957,
      "learning_rate": 3.8015585453582645e-06,
      "loss": 3.7919,
      "step": 32085
    },
    {
      "epoch": 0.3264363606770833,
      "grad_norm": 10.042716026306152,
      "learning_rate": 3.801217115789269e-06,
      "loss": 2.9715,
      "step": 32090
    },
    {
      "epoch": 0.3264872233072917,
      "grad_norm": 8.29838752746582,
      "learning_rate": 3.8008756529289424e-06,
      "loss": 3.8137,
      "step": 32095
    },
    {
      "epoch": 0.3265380859375,
      "grad_norm": 13.774858474731445,
      "learning_rate": 3.800534156786021e-06,
      "loss": 3.4684,
      "step": 32100
    },
    {
      "epoch": 0.3265889485677083,
      "grad_norm": 10.080721855163574,
      "learning_rate": 3.800192627369241e-06,
      "loss": 3.6454,
      "step": 32105
    },
    {
      "epoch": 0.3266398111979167,
      "grad_norm": 11.651310920715332,
      "learning_rate": 3.799851064687341e-06,
      "loss": 3.1652,
      "step": 32110
    },
    {
      "epoch": 0.326690673828125,
      "grad_norm": 13.935715675354004,
      "learning_rate": 3.799509468749059e-06,
      "loss": 3.2123,
      "step": 32115
    },
    {
      "epoch": 0.3267415364583333,
      "grad_norm": 10.228445053100586,
      "learning_rate": 3.7991678395631366e-06,
      "loss": 3.958,
      "step": 32120
    },
    {
      "epoch": 0.3267923990885417,
      "grad_norm": 9.072430610656738,
      "learning_rate": 3.7988261771383115e-06,
      "loss": 3.7092,
      "step": 32125
    },
    {
      "epoch": 0.32684326171875,
      "grad_norm": 10.914796829223633,
      "learning_rate": 3.7984844814833277e-06,
      "loss": 3.1457,
      "step": 32130
    },
    {
      "epoch": 0.3268941243489583,
      "grad_norm": 13.16501522064209,
      "learning_rate": 3.798142752606926e-06,
      "loss": 2.9826,
      "step": 32135
    },
    {
      "epoch": 0.3269449869791667,
      "grad_norm": 8.828996658325195,
      "learning_rate": 3.797800990517849e-06,
      "loss": 2.97,
      "step": 32140
    },
    {
      "epoch": 0.326995849609375,
      "grad_norm": 11.060222625732422,
      "learning_rate": 3.797459195224842e-06,
      "loss": 3.0853,
      "step": 32145
    },
    {
      "epoch": 0.3270467122395833,
      "grad_norm": 11.040933609008789,
      "learning_rate": 3.7971173667366478e-06,
      "loss": 3.1886,
      "step": 32150
    },
    {
      "epoch": 0.3270975748697917,
      "grad_norm": 8.803873062133789,
      "learning_rate": 3.796775505062014e-06,
      "loss": 3.4695,
      "step": 32155
    },
    {
      "epoch": 0.3271484375,
      "grad_norm": 11.375222206115723,
      "learning_rate": 3.7964336102096867e-06,
      "loss": 3.1505,
      "step": 32160
    },
    {
      "epoch": 0.3271993001302083,
      "grad_norm": 12.583060264587402,
      "learning_rate": 3.7960916821884132e-06,
      "loss": 3.1935,
      "step": 32165
    },
    {
      "epoch": 0.3272501627604167,
      "grad_norm": 9.795934677124023,
      "learning_rate": 3.7957497210069395e-06,
      "loss": 3.2419,
      "step": 32170
    },
    {
      "epoch": 0.327301025390625,
      "grad_norm": 13.176041603088379,
      "learning_rate": 3.7954077266740176e-06,
      "loss": 3.3261,
      "step": 32175
    },
    {
      "epoch": 0.3273518880208333,
      "grad_norm": 10.81923770904541,
      "learning_rate": 3.795065699198395e-06,
      "loss": 3.2257,
      "step": 32180
    },
    {
      "epoch": 0.3274027506510417,
      "grad_norm": 15.314313888549805,
      "learning_rate": 3.794723638588825e-06,
      "loss": 3.5417,
      "step": 32185
    },
    {
      "epoch": 0.32745361328125,
      "grad_norm": 12.650420188903809,
      "learning_rate": 3.7943815448540566e-06,
      "loss": 3.4859,
      "step": 32190
    },
    {
      "epoch": 0.3275044759114583,
      "grad_norm": 14.460564613342285,
      "learning_rate": 3.7940394180028427e-06,
      "loss": 3.2375,
      "step": 32195
    },
    {
      "epoch": 0.3275553385416667,
      "grad_norm": 11.314863204956055,
      "learning_rate": 3.793697258043938e-06,
      "loss": 3.3251,
      "step": 32200
    },
    {
      "epoch": 0.327606201171875,
      "grad_norm": 10.726693153381348,
      "learning_rate": 3.793355064986095e-06,
      "loss": 3.1633,
      "step": 32205
    },
    {
      "epoch": 0.3276570638020833,
      "grad_norm": 14.347328186035156,
      "learning_rate": 3.7930128388380694e-06,
      "loss": 3.3347,
      "step": 32210
    },
    {
      "epoch": 0.3277079264322917,
      "grad_norm": 9.588699340820312,
      "learning_rate": 3.792670579608617e-06,
      "loss": 3.4328,
      "step": 32215
    },
    {
      "epoch": 0.3277587890625,
      "grad_norm": 13.1727294921875,
      "learning_rate": 3.7923282873064937e-06,
      "loss": 3.1322,
      "step": 32220
    },
    {
      "epoch": 0.3278096516927083,
      "grad_norm": 12.08580493927002,
      "learning_rate": 3.7919859619404586e-06,
      "loss": 3.4606,
      "step": 32225
    },
    {
      "epoch": 0.3278605143229167,
      "grad_norm": 16.556385040283203,
      "learning_rate": 3.791643603519268e-06,
      "loss": 3.7167,
      "step": 32230
    },
    {
      "epoch": 0.327911376953125,
      "grad_norm": 9.649636268615723,
      "learning_rate": 3.7913012120516823e-06,
      "loss": 3.774,
      "step": 32235
    },
    {
      "epoch": 0.3279622395833333,
      "grad_norm": 13.704182624816895,
      "learning_rate": 3.7909587875464602e-06,
      "loss": 3.5504,
      "step": 32240
    },
    {
      "epoch": 0.3280131022135417,
      "grad_norm": 7.9661126136779785,
      "learning_rate": 3.7906163300123645e-06,
      "loss": 3.4497,
      "step": 32245
    },
    {
      "epoch": 0.32806396484375,
      "grad_norm": 10.539923667907715,
      "learning_rate": 3.7902738394581563e-06,
      "loss": 3.3076,
      "step": 32250
    },
    {
      "epoch": 0.3281148274739583,
      "grad_norm": 13.280152320861816,
      "learning_rate": 3.7899313158925967e-06,
      "loss": 3.2912,
      "step": 32255
    },
    {
      "epoch": 0.3281656901041667,
      "grad_norm": 16.510393142700195,
      "learning_rate": 3.7895887593244506e-06,
      "loss": 2.9562,
      "step": 32260
    },
    {
      "epoch": 0.328216552734375,
      "grad_norm": 8.455154418945312,
      "learning_rate": 3.7892461697624814e-06,
      "loss": 3.0784,
      "step": 32265
    },
    {
      "epoch": 0.3282674153645833,
      "grad_norm": 16.78911590576172,
      "learning_rate": 3.7889035472154552e-06,
      "loss": 3.3725,
      "step": 32270
    },
    {
      "epoch": 0.3283182779947917,
      "grad_norm": 15.698395729064941,
      "learning_rate": 3.7885608916921366e-06,
      "loss": 3.5096,
      "step": 32275
    },
    {
      "epoch": 0.328369140625,
      "grad_norm": 11.11893367767334,
      "learning_rate": 3.7882182032012936e-06,
      "loss": 3.0646,
      "step": 32280
    },
    {
      "epoch": 0.3284200032552083,
      "grad_norm": 12.249351501464844,
      "learning_rate": 3.7878754817516927e-06,
      "loss": 3.1635,
      "step": 32285
    },
    {
      "epoch": 0.3284708658854167,
      "grad_norm": 11.541364669799805,
      "learning_rate": 3.7875327273521035e-06,
      "loss": 3.1003,
      "step": 32290
    },
    {
      "epoch": 0.328521728515625,
      "grad_norm": 13.217779159545898,
      "learning_rate": 3.787189940011294e-06,
      "loss": 3.5588,
      "step": 32295
    },
    {
      "epoch": 0.3285725911458333,
      "grad_norm": 15.351944923400879,
      "learning_rate": 3.7868471197380353e-06,
      "loss": 3.5397,
      "step": 32300
    },
    {
      "epoch": 0.3286234537760417,
      "grad_norm": 7.318637847900391,
      "learning_rate": 3.786504266541098e-06,
      "loss": 3.4449,
      "step": 32305
    },
    {
      "epoch": 0.32867431640625,
      "grad_norm": 17.64267349243164,
      "learning_rate": 3.7861613804292542e-06,
      "loss": 3.6639,
      "step": 32310
    },
    {
      "epoch": 0.3287251790364583,
      "grad_norm": 10.956125259399414,
      "learning_rate": 3.7858184614112758e-06,
      "loss": 3.1958,
      "step": 32315
    },
    {
      "epoch": 0.3287760416666667,
      "grad_norm": 11.594647407531738,
      "learning_rate": 3.7854755094959374e-06,
      "loss": 3.473,
      "step": 32320
    },
    {
      "epoch": 0.328826904296875,
      "grad_norm": 13.282804489135742,
      "learning_rate": 3.7851325246920117e-06,
      "loss": 3.3825,
      "step": 32325
    },
    {
      "epoch": 0.3288777669270833,
      "grad_norm": 10.976999282836914,
      "learning_rate": 3.7847895070082763e-06,
      "loss": 3.06,
      "step": 32330
    },
    {
      "epoch": 0.3289286295572917,
      "grad_norm": 12.472411155700684,
      "learning_rate": 3.784446456453505e-06,
      "loss": 3.4135,
      "step": 32335
    },
    {
      "epoch": 0.3289794921875,
      "grad_norm": 12.325407981872559,
      "learning_rate": 3.7841033730364764e-06,
      "loss": 3.239,
      "step": 32340
    },
    {
      "epoch": 0.3290303548177083,
      "grad_norm": 7.367898464202881,
      "learning_rate": 3.783760256765967e-06,
      "loss": 3.4794,
      "step": 32345
    },
    {
      "epoch": 0.3290812174479167,
      "grad_norm": 7.337851524353027,
      "learning_rate": 3.7834171076507546e-06,
      "loss": 3.1509,
      "step": 32350
    },
    {
      "epoch": 0.329132080078125,
      "grad_norm": 15.461119651794434,
      "learning_rate": 3.7830739256996213e-06,
      "loss": 2.9971,
      "step": 32355
    },
    {
      "epoch": 0.3291829427083333,
      "grad_norm": 11.367182731628418,
      "learning_rate": 3.7827307109213453e-06,
      "loss": 3.182,
      "step": 32360
    },
    {
      "epoch": 0.3292338053385417,
      "grad_norm": 12.533026695251465,
      "learning_rate": 3.7823874633247075e-06,
      "loss": 3.354,
      "step": 32365
    },
    {
      "epoch": 0.32928466796875,
      "grad_norm": 16.5407657623291,
      "learning_rate": 3.7820441829184907e-06,
      "loss": 3.2569,
      "step": 32370
    },
    {
      "epoch": 0.3293355305989583,
      "grad_norm": 14.216818809509277,
      "learning_rate": 3.781700869711478e-06,
      "loss": 3.6429,
      "step": 32375
    },
    {
      "epoch": 0.3293863932291667,
      "grad_norm": 15.732698440551758,
      "learning_rate": 3.781357523712451e-06,
      "loss": 3.5927,
      "step": 32380
    },
    {
      "epoch": 0.329437255859375,
      "grad_norm": 9.038349151611328,
      "learning_rate": 3.7810141449301976e-06,
      "loss": 3.4536,
      "step": 32385
    },
    {
      "epoch": 0.3294881184895833,
      "grad_norm": 14.39390754699707,
      "learning_rate": 3.780670733373499e-06,
      "loss": 3.2767,
      "step": 32390
    },
    {
      "epoch": 0.3295389811197917,
      "grad_norm": 9.732683181762695,
      "learning_rate": 3.7803272890511437e-06,
      "loss": 3.4086,
      "step": 32395
    },
    {
      "epoch": 0.32958984375,
      "grad_norm": 15.71953296661377,
      "learning_rate": 3.779983811971919e-06,
      "loss": 3.1739,
      "step": 32400
    },
    {
      "epoch": 0.3296407063802083,
      "grad_norm": 14.279335021972656,
      "learning_rate": 3.7796403021446106e-06,
      "loss": 3.3621,
      "step": 32405
    },
    {
      "epoch": 0.3296915690104167,
      "grad_norm": 12.024027824401855,
      "learning_rate": 3.7792967595780095e-06,
      "loss": 3.6476,
      "step": 32410
    },
    {
      "epoch": 0.329742431640625,
      "grad_norm": 10.159197807312012,
      "learning_rate": 3.7789531842809034e-06,
      "loss": 3.4376,
      "step": 32415
    },
    {
      "epoch": 0.3297932942708333,
      "grad_norm": 13.340349197387695,
      "learning_rate": 3.7786095762620836e-06,
      "loss": 3.6374,
      "step": 32420
    },
    {
      "epoch": 0.3298441569010417,
      "grad_norm": 8.4384126663208,
      "learning_rate": 3.7782659355303397e-06,
      "loss": 3.2591,
      "step": 32425
    },
    {
      "epoch": 0.32989501953125,
      "grad_norm": 14.894301414489746,
      "learning_rate": 3.777922262094466e-06,
      "loss": 3.2045,
      "step": 32430
    },
    {
      "epoch": 0.3299458821614583,
      "grad_norm": 9.043319702148438,
      "learning_rate": 3.777578555963254e-06,
      "loss": 3.133,
      "step": 32435
    },
    {
      "epoch": 0.3299967447916667,
      "grad_norm": 7.9150919914245605,
      "learning_rate": 3.777234817145497e-06,
      "loss": 3.0001,
      "step": 32440
    },
    {
      "epoch": 0.330047607421875,
      "grad_norm": 13.222947120666504,
      "learning_rate": 3.77689104564999e-06,
      "loss": 3.2136,
      "step": 32445
    },
    {
      "epoch": 0.3300984700520833,
      "grad_norm": 11.708751678466797,
      "learning_rate": 3.7765472414855294e-06,
      "loss": 3.4354,
      "step": 32450
    },
    {
      "epoch": 0.3301493326822917,
      "grad_norm": 15.85578727722168,
      "learning_rate": 3.7762034046609085e-06,
      "loss": 3.5524,
      "step": 32455
    },
    {
      "epoch": 0.3302001953125,
      "grad_norm": 7.159616470336914,
      "learning_rate": 3.7758595351849276e-06,
      "loss": 3.0285,
      "step": 32460
    },
    {
      "epoch": 0.3302510579427083,
      "grad_norm": 8.465569496154785,
      "learning_rate": 3.7755156330663824e-06,
      "loss": 3.2453,
      "step": 32465
    },
    {
      "epoch": 0.3303019205729167,
      "grad_norm": 6.593746662139893,
      "learning_rate": 3.7751716983140713e-06,
      "loss": 3.0596,
      "step": 32470
    },
    {
      "epoch": 0.330352783203125,
      "grad_norm": 11.99484634399414,
      "learning_rate": 3.7748277309367946e-06,
      "loss": 3.342,
      "step": 32475
    },
    {
      "epoch": 0.3304036458333333,
      "grad_norm": 13.411246299743652,
      "learning_rate": 3.7744837309433534e-06,
      "loss": 3.2973,
      "step": 32480
    },
    {
      "epoch": 0.3304545084635417,
      "grad_norm": 10.368703842163086,
      "learning_rate": 3.774139698342547e-06,
      "loss": 3.322,
      "step": 32485
    },
    {
      "epoch": 0.33050537109375,
      "grad_norm": 11.17487621307373,
      "learning_rate": 3.773795633143179e-06,
      "loss": 2.9814,
      "step": 32490
    },
    {
      "epoch": 0.3305562337239583,
      "grad_norm": 13.313066482543945,
      "learning_rate": 3.7734515353540526e-06,
      "loss": 3.746,
      "step": 32495
    },
    {
      "epoch": 0.3306070963541667,
      "grad_norm": 17.076797485351562,
      "learning_rate": 3.7731074049839695e-06,
      "loss": 3.3898,
      "step": 32500
    },
    {
      "epoch": 0.330657958984375,
      "grad_norm": 9.995706558227539,
      "learning_rate": 3.772763242041735e-06,
      "loss": 3.1957,
      "step": 32505
    },
    {
      "epoch": 0.3307088216145833,
      "grad_norm": 12.41891098022461,
      "learning_rate": 3.7724190465361553e-06,
      "loss": 3.2141,
      "step": 32510
    },
    {
      "epoch": 0.3307596842447917,
      "grad_norm": 9.968422889709473,
      "learning_rate": 3.7720748184760353e-06,
      "loss": 3.4549,
      "step": 32515
    },
    {
      "epoch": 0.330810546875,
      "grad_norm": 9.146455764770508,
      "learning_rate": 3.7717305578701824e-06,
      "loss": 3.6653,
      "step": 32520
    },
    {
      "epoch": 0.3308614095052083,
      "grad_norm": 7.412753105163574,
      "learning_rate": 3.7713862647274054e-06,
      "loss": 3.9118,
      "step": 32525
    },
    {
      "epoch": 0.3309122721354167,
      "grad_norm": 11.762595176696777,
      "learning_rate": 3.7710419390565117e-06,
      "loss": 3.6556,
      "step": 32530
    },
    {
      "epoch": 0.330963134765625,
      "grad_norm": 8.641430854797363,
      "learning_rate": 3.7706975808663117e-06,
      "loss": 3.6632,
      "step": 32535
    },
    {
      "epoch": 0.3310139973958333,
      "grad_norm": 12.438641548156738,
      "learning_rate": 3.770353190165614e-06,
      "loss": 3.4598,
      "step": 32540
    },
    {
      "epoch": 0.3310648600260417,
      "grad_norm": 78.32935333251953,
      "learning_rate": 3.7700087669632325e-06,
      "loss": 3.5554,
      "step": 32545
    },
    {
      "epoch": 0.33111572265625,
      "grad_norm": 7.5797295570373535,
      "learning_rate": 3.769664311267977e-06,
      "loss": 3.04,
      "step": 32550
    },
    {
      "epoch": 0.3311665852864583,
      "grad_norm": 14.218374252319336,
      "learning_rate": 3.769319823088662e-06,
      "loss": 2.6611,
      "step": 32555
    },
    {
      "epoch": 0.3312174479166667,
      "grad_norm": 15.772465705871582,
      "learning_rate": 3.7689753024340987e-06,
      "loss": 3.4001,
      "step": 32560
    },
    {
      "epoch": 0.331268310546875,
      "grad_norm": 12.117385864257812,
      "learning_rate": 3.7686307493131043e-06,
      "loss": 3.4332,
      "step": 32565
    },
    {
      "epoch": 0.3313191731770833,
      "grad_norm": 10.31899356842041,
      "learning_rate": 3.7682861637344914e-06,
      "loss": 3.3407,
      "step": 32570
    },
    {
      "epoch": 0.3313700358072917,
      "grad_norm": 15.768966674804688,
      "learning_rate": 3.7679415457070787e-06,
      "loss": 3.4107,
      "step": 32575
    },
    {
      "epoch": 0.3314208984375,
      "grad_norm": 11.954106330871582,
      "learning_rate": 3.7675968952396824e-06,
      "loss": 3.4454,
      "step": 32580
    },
    {
      "epoch": 0.3314717610677083,
      "grad_norm": 12.425500869750977,
      "learning_rate": 3.7672522123411194e-06,
      "loss": 3.7169,
      "step": 32585
    },
    {
      "epoch": 0.3315226236979167,
      "grad_norm": 15.04654598236084,
      "learning_rate": 3.7669074970202092e-06,
      "loss": 3.389,
      "step": 32590
    },
    {
      "epoch": 0.331573486328125,
      "grad_norm": 12.327473640441895,
      "learning_rate": 3.7665627492857702e-06,
      "loss": 2.9167,
      "step": 32595
    },
    {
      "epoch": 0.3316243489583333,
      "grad_norm": 12.949674606323242,
      "learning_rate": 3.766217969146625e-06,
      "loss": 3.2564,
      "step": 32600
    },
    {
      "epoch": 0.3316752115885417,
      "grad_norm": 10.511869430541992,
      "learning_rate": 3.7658731566115924e-06,
      "loss": 3.2482,
      "step": 32605
    },
    {
      "epoch": 0.33172607421875,
      "grad_norm": 13.890068054199219,
      "learning_rate": 3.7655283116894947e-06,
      "loss": 3.3871,
      "step": 32610
    },
    {
      "epoch": 0.3317769368489583,
      "grad_norm": 15.479687690734863,
      "learning_rate": 3.765183434389156e-06,
      "loss": 3.3261,
      "step": 32615
    },
    {
      "epoch": 0.3318277994791667,
      "grad_norm": 10.50954532623291,
      "learning_rate": 3.764838524719398e-06,
      "loss": 3.1074,
      "step": 32620
    },
    {
      "epoch": 0.331878662109375,
      "grad_norm": 16.40087127685547,
      "learning_rate": 3.764493582689047e-06,
      "loss": 3.1189,
      "step": 32625
    },
    {
      "epoch": 0.3319295247395833,
      "grad_norm": 11.862818717956543,
      "learning_rate": 3.764148608306927e-06,
      "loss": 3.6677,
      "step": 32630
    },
    {
      "epoch": 0.3319803873697917,
      "grad_norm": 15.727245330810547,
      "learning_rate": 3.763803601581864e-06,
      "loss": 3.3587,
      "step": 32635
    },
    {
      "epoch": 0.33203125,
      "grad_norm": 12.317474365234375,
      "learning_rate": 3.763458562522686e-06,
      "loss": 3.4326,
      "step": 32640
    },
    {
      "epoch": 0.3320821126302083,
      "grad_norm": 9.582719802856445,
      "learning_rate": 3.7631134911382196e-06,
      "loss": 4.0108,
      "step": 32645
    },
    {
      "epoch": 0.3321329752604167,
      "grad_norm": 15.248978614807129,
      "learning_rate": 3.7627683874372943e-06,
      "loss": 3.1654,
      "step": 32650
    },
    {
      "epoch": 0.332183837890625,
      "grad_norm": 14.71537971496582,
      "learning_rate": 3.7624232514287396e-06,
      "loss": 3.4238,
      "step": 32655
    },
    {
      "epoch": 0.3322347005208333,
      "grad_norm": 10.918207168579102,
      "learning_rate": 3.7620780831213842e-06,
      "loss": 3.5747,
      "step": 32660
    },
    {
      "epoch": 0.3322855631510417,
      "grad_norm": 12.324572563171387,
      "learning_rate": 3.7617328825240614e-06,
      "loss": 3.325,
      "step": 32665
    },
    {
      "epoch": 0.33233642578125,
      "grad_norm": 8.128783226013184,
      "learning_rate": 3.7613876496456014e-06,
      "loss": 3.8681,
      "step": 32670
    },
    {
      "epoch": 0.3323872884114583,
      "grad_norm": 12.606327056884766,
      "learning_rate": 3.7610423844948367e-06,
      "loss": 3.4388,
      "step": 32675
    },
    {
      "epoch": 0.3324381510416667,
      "grad_norm": 8.71562671661377,
      "learning_rate": 3.7606970870806024e-06,
      "loss": 3.1369,
      "step": 32680
    },
    {
      "epoch": 0.332489013671875,
      "grad_norm": 17.26483917236328,
      "learning_rate": 3.760351757411732e-06,
      "loss": 3.4954,
      "step": 32685
    },
    {
      "epoch": 0.3325398763020833,
      "grad_norm": 14.267714500427246,
      "learning_rate": 3.76000639549706e-06,
      "loss": 3.2069,
      "step": 32690
    },
    {
      "epoch": 0.3325907389322917,
      "grad_norm": 7.968595504760742,
      "learning_rate": 3.7596610013454226e-06,
      "loss": 3.4987,
      "step": 32695
    },
    {
      "epoch": 0.3326416015625,
      "grad_norm": 12.251919746398926,
      "learning_rate": 3.7593155749656583e-06,
      "loss": 3.4392,
      "step": 32700
    },
    {
      "epoch": 0.3326924641927083,
      "grad_norm": 11.02563190460205,
      "learning_rate": 3.7589701163666026e-06,
      "loss": 2.8879,
      "step": 32705
    },
    {
      "epoch": 0.3327433268229167,
      "grad_norm": 10.374553680419922,
      "learning_rate": 3.758624625557096e-06,
      "loss": 3.5271,
      "step": 32710
    },
    {
      "epoch": 0.332794189453125,
      "grad_norm": 11.383188247680664,
      "learning_rate": 3.7582791025459753e-06,
      "loss": 3.1762,
      "step": 32715
    },
    {
      "epoch": 0.3328450520833333,
      "grad_norm": 14.80070972442627,
      "learning_rate": 3.757933547342083e-06,
      "loss": 3.7152,
      "step": 32720
    },
    {
      "epoch": 0.3328959147135417,
      "grad_norm": 10.358442306518555,
      "learning_rate": 3.7575879599542574e-06,
      "loss": 3.2842,
      "step": 32725
    },
    {
      "epoch": 0.33294677734375,
      "grad_norm": 11.465337753295898,
      "learning_rate": 3.7572423403913437e-06,
      "loss": 3.3594,
      "step": 32730
    },
    {
      "epoch": 0.3329976399739583,
      "grad_norm": 16.327484130859375,
      "learning_rate": 3.7568966886621827e-06,
      "loss": 3.2038,
      "step": 32735
    },
    {
      "epoch": 0.3330485026041667,
      "grad_norm": 11.995901107788086,
      "learning_rate": 3.7565510047756165e-06,
      "loss": 3.1667,
      "step": 32740
    },
    {
      "epoch": 0.333099365234375,
      "grad_norm": 11.180763244628906,
      "learning_rate": 3.756205288740492e-06,
      "loss": 3.2824,
      "step": 32745
    },
    {
      "epoch": 0.3331502278645833,
      "grad_norm": 13.001599311828613,
      "learning_rate": 3.755859540565652e-06,
      "loss": 3.4875,
      "step": 32750
    },
    {
      "epoch": 0.3332010904947917,
      "grad_norm": 12.533217430114746,
      "learning_rate": 3.755513760259944e-06,
      "loss": 3.0541,
      "step": 32755
    },
    {
      "epoch": 0.333251953125,
      "grad_norm": 14.233641624450684,
      "learning_rate": 3.755167947832213e-06,
      "loss": 2.9513,
      "step": 32760
    },
    {
      "epoch": 0.3333028157552083,
      "grad_norm": 10.958354949951172,
      "learning_rate": 3.754822103291309e-06,
      "loss": 3.5791,
      "step": 32765
    },
    {
      "epoch": 0.3333536783854167,
      "grad_norm": 8.330187797546387,
      "learning_rate": 3.7544762266460784e-06,
      "loss": 3.4526,
      "step": 32770
    },
    {
      "epoch": 0.333404541015625,
      "grad_norm": 9.813884735107422,
      "learning_rate": 3.754130317905371e-06,
      "loss": 3.0888,
      "step": 32775
    },
    {
      "epoch": 0.3334554036458333,
      "grad_norm": 10.979777336120605,
      "learning_rate": 3.753784377078036e-06,
      "loss": 3.5022,
      "step": 32780
    },
    {
      "epoch": 0.3335062662760417,
      "grad_norm": 14.47551155090332,
      "learning_rate": 3.7534384041729255e-06,
      "loss": 3.3508,
      "step": 32785
    },
    {
      "epoch": 0.33355712890625,
      "grad_norm": 11.598821640014648,
      "learning_rate": 3.75309239919889e-06,
      "loss": 3.2422,
      "step": 32790
    },
    {
      "epoch": 0.3336079915364583,
      "grad_norm": 12.611983299255371,
      "learning_rate": 3.752746362164783e-06,
      "loss": 3.2404,
      "step": 32795
    },
    {
      "epoch": 0.3336588541666667,
      "grad_norm": 13.306591987609863,
      "learning_rate": 3.7524002930794567e-06,
      "loss": 3.3322,
      "step": 32800
    },
    {
      "epoch": 0.333709716796875,
      "grad_norm": 14.114723205566406,
      "learning_rate": 3.752054191951766e-06,
      "loss": 3.4821,
      "step": 32805
    },
    {
      "epoch": 0.3337605794270833,
      "grad_norm": 8.142168045043945,
      "learning_rate": 3.751708058790566e-06,
      "loss": 3.6567,
      "step": 32810
    },
    {
      "epoch": 0.3338114420572917,
      "grad_norm": 16.089536666870117,
      "learning_rate": 3.7513618936047116e-06,
      "loss": 3.0567,
      "step": 32815
    },
    {
      "epoch": 0.3338623046875,
      "grad_norm": 13.037550926208496,
      "learning_rate": 3.7510156964030597e-06,
      "loss": 3.6844,
      "step": 32820
    },
    {
      "epoch": 0.3339131673177083,
      "grad_norm": 13.749690055847168,
      "learning_rate": 3.750669467194469e-06,
      "loss": 3.5359,
      "step": 32825
    },
    {
      "epoch": 0.3339640299479167,
      "grad_norm": 11.21825122833252,
      "learning_rate": 3.7503232059877947e-06,
      "loss": 3.0431,
      "step": 32830
    },
    {
      "epoch": 0.334014892578125,
      "grad_norm": 10.482390403747559,
      "learning_rate": 3.7499769127918977e-06,
      "loss": 3.1787,
      "step": 32835
    },
    {
      "epoch": 0.3340657552083333,
      "grad_norm": 14.753950119018555,
      "learning_rate": 3.7496305876156385e-06,
      "loss": 3.9245,
      "step": 32840
    },
    {
      "epoch": 0.3341166178385417,
      "grad_norm": 15.145987510681152,
      "learning_rate": 3.749284230467876e-06,
      "loss": 3.4858,
      "step": 32845
    },
    {
      "epoch": 0.33416748046875,
      "grad_norm": 9.231215476989746,
      "learning_rate": 3.7489378413574735e-06,
      "loss": 3.7131,
      "step": 32850
    },
    {
      "epoch": 0.3342183430989583,
      "grad_norm": 14.36355972290039,
      "learning_rate": 3.748591420293292e-06,
      "loss": 3.2538,
      "step": 32855
    },
    {
      "epoch": 0.3342692057291667,
      "grad_norm": 11.919439315795898,
      "learning_rate": 3.748244967284195e-06,
      "loss": 3.6638,
      "step": 32860
    },
    {
      "epoch": 0.334320068359375,
      "grad_norm": 10.983847618103027,
      "learning_rate": 3.7478984823390465e-06,
      "loss": 3.1181,
      "step": 32865
    },
    {
      "epoch": 0.3343709309895833,
      "grad_norm": 13.097049713134766,
      "learning_rate": 3.74755196546671e-06,
      "loss": 3.3605,
      "step": 32870
    },
    {
      "epoch": 0.3344217936197917,
      "grad_norm": 20.51844024658203,
      "learning_rate": 3.7472054166760535e-06,
      "loss": 3.0601,
      "step": 32875
    },
    {
      "epoch": 0.33447265625,
      "grad_norm": 12.644318580627441,
      "learning_rate": 3.7468588359759416e-06,
      "loss": 3.4744,
      "step": 32880
    },
    {
      "epoch": 0.3345235188802083,
      "grad_norm": 10.257139205932617,
      "learning_rate": 3.7465122233752415e-06,
      "loss": 3.3923,
      "step": 32885
    },
    {
      "epoch": 0.3345743815104167,
      "grad_norm": 12.080252647399902,
      "learning_rate": 3.7461655788828227e-06,
      "loss": 3.308,
      "step": 32890
    },
    {
      "epoch": 0.334625244140625,
      "grad_norm": 11.510242462158203,
      "learning_rate": 3.745818902507552e-06,
      "loss": 3.2991,
      "step": 32895
    },
    {
      "epoch": 0.3346761067708333,
      "grad_norm": 11.790918350219727,
      "learning_rate": 3.7454721942583007e-06,
      "loss": 3.4248,
      "step": 32900
    },
    {
      "epoch": 0.3347269694010417,
      "grad_norm": 10.164443016052246,
      "learning_rate": 3.745125454143938e-06,
      "loss": 3.2437,
      "step": 32905
    },
    {
      "epoch": 0.33477783203125,
      "grad_norm": 15.713869094848633,
      "learning_rate": 3.7447786821733356e-06,
      "loss": 3.3894,
      "step": 32910
    },
    {
      "epoch": 0.3348286946614583,
      "grad_norm": 10.7219877243042,
      "learning_rate": 3.744431878355366e-06,
      "loss": 3.2601,
      "step": 32915
    },
    {
      "epoch": 0.3348795572916667,
      "grad_norm": 15.252687454223633,
      "learning_rate": 3.7440850426989013e-06,
      "loss": 4.0186,
      "step": 32920
    },
    {
      "epoch": 0.334930419921875,
      "grad_norm": 14.942272186279297,
      "learning_rate": 3.743738175212816e-06,
      "loss": 3.5481,
      "step": 32925
    },
    {
      "epoch": 0.3349812825520833,
      "grad_norm": 14.428038597106934,
      "learning_rate": 3.743391275905984e-06,
      "loss": 3.4527,
      "step": 32930
    },
    {
      "epoch": 0.3350321451822917,
      "grad_norm": 12.873485565185547,
      "learning_rate": 3.743044344787281e-06,
      "loss": 3.0538,
      "step": 32935
    },
    {
      "epoch": 0.3350830078125,
      "grad_norm": 14.081838607788086,
      "learning_rate": 3.7426973818655833e-06,
      "loss": 3.5227,
      "step": 32940
    },
    {
      "epoch": 0.3351338704427083,
      "grad_norm": 10.624444007873535,
      "learning_rate": 3.7423503871497674e-06,
      "loss": 3.285,
      "step": 32945
    },
    {
      "epoch": 0.3351847330729167,
      "grad_norm": 11.455140113830566,
      "learning_rate": 3.7420033606487114e-06,
      "loss": 3.121,
      "step": 32950
    },
    {
      "epoch": 0.335235595703125,
      "grad_norm": 10.438578605651855,
      "learning_rate": 3.741656302371294e-06,
      "loss": 3.1361,
      "step": 32955
    },
    {
      "epoch": 0.3352864583333333,
      "grad_norm": 12.183462142944336,
      "learning_rate": 3.741309212326394e-06,
      "loss": 3.1173,
      "step": 32960
    },
    {
      "epoch": 0.3353373209635417,
      "grad_norm": 11.119400978088379,
      "learning_rate": 3.7409620905228927e-06,
      "loss": 2.9163,
      "step": 32965
    },
    {
      "epoch": 0.33538818359375,
      "grad_norm": 13.267557144165039,
      "learning_rate": 3.7406149369696704e-06,
      "loss": 2.92,
      "step": 32970
    },
    {
      "epoch": 0.3354390462239583,
      "grad_norm": 7.646348476409912,
      "learning_rate": 3.740267751675608e-06,
      "loss": 3.6382,
      "step": 32975
    },
    {
      "epoch": 0.3354899088541667,
      "grad_norm": 10.166210174560547,
      "learning_rate": 3.7399205346495905e-06,
      "loss": 3.2228,
      "step": 32980
    },
    {
      "epoch": 0.335540771484375,
      "grad_norm": 9.681705474853516,
      "learning_rate": 3.739573285900499e-06,
      "loss": 3.2517,
      "step": 32985
    },
    {
      "epoch": 0.3355916341145833,
      "grad_norm": 14.030284881591797,
      "learning_rate": 3.739226005437219e-06,
      "loss": 3.2506,
      "step": 32990
    },
    {
      "epoch": 0.3356424967447917,
      "grad_norm": 13.2785062789917,
      "learning_rate": 3.738878693268636e-06,
      "loss": 3.5483,
      "step": 32995
    },
    {
      "epoch": 0.335693359375,
      "grad_norm": 10.970511436462402,
      "learning_rate": 3.738531349403634e-06,
      "loss": 3.4073,
      "step": 33000
    },
    {
      "epoch": 0.3357442220052083,
      "grad_norm": 14.737278938293457,
      "learning_rate": 3.7381839738511024e-06,
      "loss": 3.2362,
      "step": 33005
    },
    {
      "epoch": 0.3357950846354167,
      "grad_norm": 10.005160331726074,
      "learning_rate": 3.737836566619927e-06,
      "loss": 3.5164,
      "step": 33010
    },
    {
      "epoch": 0.335845947265625,
      "grad_norm": 13.465721130371094,
      "learning_rate": 3.7374891277189957e-06,
      "loss": 3.4245,
      "step": 33015
    },
    {
      "epoch": 0.3358968098958333,
      "grad_norm": 15.173518180847168,
      "learning_rate": 3.7371416571571983e-06,
      "loss": 3.0978,
      "step": 33020
    },
    {
      "epoch": 0.3359476725260417,
      "grad_norm": 14.85016918182373,
      "learning_rate": 3.7367941549434254e-06,
      "loss": 3.4838,
      "step": 33025
    },
    {
      "epoch": 0.33599853515625,
      "grad_norm": 11.801966667175293,
      "learning_rate": 3.7364466210865676e-06,
      "loss": 3.839,
      "step": 33030
    },
    {
      "epoch": 0.3360493977864583,
      "grad_norm": 10.579994201660156,
      "learning_rate": 3.736099055595515e-06,
      "loss": 3.2351,
      "step": 33035
    },
    {
      "epoch": 0.3361002604166667,
      "grad_norm": 12.842473983764648,
      "learning_rate": 3.735751458479162e-06,
      "loss": 3.5836,
      "step": 33040
    },
    {
      "epoch": 0.336151123046875,
      "grad_norm": 12.5321683883667,
      "learning_rate": 3.7354038297464e-06,
      "loss": 3.1939,
      "step": 33045
    },
    {
      "epoch": 0.3362019856770833,
      "grad_norm": 10.775299072265625,
      "learning_rate": 3.7350561694061245e-06,
      "loss": 3.3556,
      "step": 33050
    },
    {
      "epoch": 0.3362528483072917,
      "grad_norm": 11.678428649902344,
      "learning_rate": 3.7347084774672295e-06,
      "loss": 3.8341,
      "step": 33055
    },
    {
      "epoch": 0.3363037109375,
      "grad_norm": 15.017143249511719,
      "learning_rate": 3.734360753938611e-06,
      "loss": 3.647,
      "step": 33060
    },
    {
      "epoch": 0.3363545735677083,
      "grad_norm": 14.71731185913086,
      "learning_rate": 3.7340129988291645e-06,
      "loss": 3.2912,
      "step": 33065
    },
    {
      "epoch": 0.3364054361979167,
      "grad_norm": 12.676302909851074,
      "learning_rate": 3.733665212147789e-06,
      "loss": 3.313,
      "step": 33070
    },
    {
      "epoch": 0.336456298828125,
      "grad_norm": 10.882745742797852,
      "learning_rate": 3.7333173939033807e-06,
      "loss": 2.9863,
      "step": 33075
    },
    {
      "epoch": 0.3365071614583333,
      "grad_norm": 11.048257827758789,
      "learning_rate": 3.732969544104839e-06,
      "loss": 3.0796,
      "step": 33080
    },
    {
      "epoch": 0.3365580240885417,
      "grad_norm": 13.560098648071289,
      "learning_rate": 3.732621662761064e-06,
      "loss": 3.1773,
      "step": 33085
    },
    {
      "epoch": 0.33660888671875,
      "grad_norm": 10.831151962280273,
      "learning_rate": 3.732273749880956e-06,
      "loss": 3.264,
      "step": 33090
    },
    {
      "epoch": 0.3366597493489583,
      "grad_norm": 12.574798583984375,
      "learning_rate": 3.7319258054734166e-06,
      "loss": 3.4186,
      "step": 33095
    },
    {
      "epoch": 0.3367106119791667,
      "grad_norm": 12.322128295898438,
      "learning_rate": 3.7315778295473467e-06,
      "loss": 3.4011,
      "step": 33100
    },
    {
      "epoch": 0.336761474609375,
      "grad_norm": 11.991111755371094,
      "learning_rate": 3.73122982211165e-06,
      "loss": 3.152,
      "step": 33105
    },
    {
      "epoch": 0.3368123372395833,
      "grad_norm": 9.92126178741455,
      "learning_rate": 3.730881783175231e-06,
      "loss": 3.0965,
      "step": 33110
    },
    {
      "epoch": 0.3368631998697917,
      "grad_norm": 7.607785224914551,
      "learning_rate": 3.730533712746992e-06,
      "loss": 3.1248,
      "step": 33115
    },
    {
      "epoch": 0.3369140625,
      "grad_norm": 10.90238094329834,
      "learning_rate": 3.7301856108358407e-06,
      "loss": 3.5398,
      "step": 33120
    },
    {
      "epoch": 0.3369649251302083,
      "grad_norm": 11.631036758422852,
      "learning_rate": 3.729837477450682e-06,
      "loss": 3.4372,
      "step": 33125
    },
    {
      "epoch": 0.3370157877604167,
      "grad_norm": 10.736931800842285,
      "learning_rate": 3.729489312600423e-06,
      "loss": 3.4555,
      "step": 33130
    },
    {
      "epoch": 0.337066650390625,
      "grad_norm": 11.401076316833496,
      "learning_rate": 3.7291411162939705e-06,
      "loss": 3.1215,
      "step": 33135
    },
    {
      "epoch": 0.3371175130208333,
      "grad_norm": 13.46949577331543,
      "learning_rate": 3.728792888540235e-06,
      "loss": 3.5968,
      "step": 33140
    },
    {
      "epoch": 0.3371683756510417,
      "grad_norm": 11.45217227935791,
      "learning_rate": 3.7284446293481235e-06,
      "loss": 3.2885,
      "step": 33145
    },
    {
      "epoch": 0.33721923828125,
      "grad_norm": 13.1810941696167,
      "learning_rate": 3.7280963387265484e-06,
      "loss": 3.298,
      "step": 33150
    },
    {
      "epoch": 0.3372701009114583,
      "grad_norm": 6.902253150939941,
      "learning_rate": 3.727748016684419e-06,
      "loss": 3.6278,
      "step": 33155
    },
    {
      "epoch": 0.3373209635416667,
      "grad_norm": 7.337198734283447,
      "learning_rate": 3.7273996632306473e-06,
      "loss": 3.5548,
      "step": 33160
    },
    {
      "epoch": 0.337371826171875,
      "grad_norm": 9.131155014038086,
      "learning_rate": 3.7270512783741475e-06,
      "loss": 3.3623,
      "step": 33165
    },
    {
      "epoch": 0.3374226888020833,
      "grad_norm": 13.159061431884766,
      "learning_rate": 3.7267028621238305e-06,
      "loss": 3.4383,
      "step": 33170
    },
    {
      "epoch": 0.3374735514322917,
      "grad_norm": 10.10805606842041,
      "learning_rate": 3.7263544144886123e-06,
      "loss": 3.2849,
      "step": 33175
    },
    {
      "epoch": 0.3375244140625,
      "grad_norm": 13.516425132751465,
      "learning_rate": 3.726005935477406e-06,
      "loss": 3.5075,
      "step": 33180
    },
    {
      "epoch": 0.3375752766927083,
      "grad_norm": 16.516651153564453,
      "learning_rate": 3.725657425099129e-06,
      "loss": 3.4631,
      "step": 33185
    },
    {
      "epoch": 0.3376261393229167,
      "grad_norm": 10.670194625854492,
      "learning_rate": 3.7253088833626973e-06,
      "loss": 2.9451,
      "step": 33190
    },
    {
      "epoch": 0.337677001953125,
      "grad_norm": 13.951617240905762,
      "learning_rate": 3.724960310277028e-06,
      "loss": 3.5035,
      "step": 33195
    },
    {
      "epoch": 0.3377278645833333,
      "grad_norm": 9.054759979248047,
      "learning_rate": 3.72461170585104e-06,
      "loss": 3.4928,
      "step": 33200
    },
    {
      "epoch": 0.3377787272135417,
      "grad_norm": 11.50307846069336,
      "learning_rate": 3.724263070093651e-06,
      "loss": 3.1345,
      "step": 33205
    },
    {
      "epoch": 0.33782958984375,
      "grad_norm": 6.693308353424072,
      "learning_rate": 3.723914403013782e-06,
      "loss": 3.3025,
      "step": 33210
    },
    {
      "epoch": 0.3378804524739583,
      "grad_norm": 14.62735652923584,
      "learning_rate": 3.7235657046203537e-06,
      "loss": 3.4618,
      "step": 33215
    },
    {
      "epoch": 0.3379313151041667,
      "grad_norm": 10.572195053100586,
      "learning_rate": 3.7232169749222856e-06,
      "loss": 3.9669,
      "step": 33220
    },
    {
      "epoch": 0.337982177734375,
      "grad_norm": 11.943432807922363,
      "learning_rate": 3.722868213928502e-06,
      "loss": 3.3552,
      "step": 33225
    },
    {
      "epoch": 0.3380330403645833,
      "grad_norm": 9.331600189208984,
      "learning_rate": 3.7225194216479255e-06,
      "loss": 3.3776,
      "step": 33230
    },
    {
      "epoch": 0.3380839029947917,
      "grad_norm": 13.232575416564941,
      "learning_rate": 3.722170598089478e-06,
      "loss": 3.1956,
      "step": 33235
    },
    {
      "epoch": 0.338134765625,
      "grad_norm": 11.602453231811523,
      "learning_rate": 3.7218217432620863e-06,
      "loss": 3.5115,
      "step": 33240
    },
    {
      "epoch": 0.3381856282552083,
      "grad_norm": 9.096488952636719,
      "learning_rate": 3.721472857174675e-06,
      "loss": 3.6,
      "step": 33245
    },
    {
      "epoch": 0.3382364908854167,
      "grad_norm": 8.760087013244629,
      "learning_rate": 3.7211239398361697e-06,
      "loss": 3.2964,
      "step": 33250
    },
    {
      "epoch": 0.338287353515625,
      "grad_norm": 12.91967487335205,
      "learning_rate": 3.720774991255498e-06,
      "loss": 3.1863,
      "step": 33255
    },
    {
      "epoch": 0.3383382161458333,
      "grad_norm": 15.063009262084961,
      "learning_rate": 3.7204260114415875e-06,
      "loss": 3.2829,
      "step": 33260
    },
    {
      "epoch": 0.3383890787760417,
      "grad_norm": 6.675595760345459,
      "learning_rate": 3.7200770004033666e-06,
      "loss": 3.0992,
      "step": 33265
    },
    {
      "epoch": 0.33843994140625,
      "grad_norm": 11.452447891235352,
      "learning_rate": 3.719727958149765e-06,
      "loss": 3.5164,
      "step": 33270
    },
    {
      "epoch": 0.3384908040364583,
      "grad_norm": 9.369376182556152,
      "learning_rate": 3.7193788846897126e-06,
      "loss": 3.4509,
      "step": 33275
    },
    {
      "epoch": 0.3385416666666667,
      "grad_norm": 11.180233001708984,
      "learning_rate": 3.719029780032141e-06,
      "loss": 3.1632,
      "step": 33280
    },
    {
      "epoch": 0.338592529296875,
      "grad_norm": 7.883488655090332,
      "learning_rate": 3.7186806441859806e-06,
      "loss": 3.4346,
      "step": 33285
    },
    {
      "epoch": 0.3386433919270833,
      "grad_norm": 9.978160858154297,
      "learning_rate": 3.7183314771601652e-06,
      "loss": 3.2653,
      "step": 33290
    },
    {
      "epoch": 0.3386942545572917,
      "grad_norm": 15.136893272399902,
      "learning_rate": 3.7179822789636276e-06,
      "loss": 3.1239,
      "step": 33295
    },
    {
      "epoch": 0.3387451171875,
      "grad_norm": 8.271352767944336,
      "learning_rate": 3.7176330496053026e-06,
      "loss": 3.0803,
      "step": 33300
    },
    {
      "epoch": 0.3387959798177083,
      "grad_norm": 14.263900756835938,
      "learning_rate": 3.7172837890941243e-06,
      "loss": 3.3083,
      "step": 33305
    },
    {
      "epoch": 0.3388468424479167,
      "grad_norm": 13.307559967041016,
      "learning_rate": 3.7169344974390286e-06,
      "loss": 3.3729,
      "step": 33310
    },
    {
      "epoch": 0.338897705078125,
      "grad_norm": 11.805386543273926,
      "learning_rate": 3.716585174648952e-06,
      "loss": 3.2258,
      "step": 33315
    },
    {
      "epoch": 0.3389485677083333,
      "grad_norm": 12.272468566894531,
      "learning_rate": 3.716235820732833e-06,
      "loss": 3.3413,
      "step": 33320
    },
    {
      "epoch": 0.3389994303385417,
      "grad_norm": 8.26090145111084,
      "learning_rate": 3.7158864356996076e-06,
      "loss": 3.3389,
      "step": 33325
    },
    {
      "epoch": 0.33905029296875,
      "grad_norm": 11.451239585876465,
      "learning_rate": 3.7155370195582167e-06,
      "loss": 3.2675,
      "step": 33330
    },
    {
      "epoch": 0.3391011555989583,
      "grad_norm": 13.9033203125,
      "learning_rate": 3.715187572317599e-06,
      "loss": 3.3937,
      "step": 33335
    },
    {
      "epoch": 0.3391520182291667,
      "grad_norm": 11.666393280029297,
      "learning_rate": 3.7148380939866956e-06,
      "loss": 3.2749,
      "step": 33340
    },
    {
      "epoch": 0.339202880859375,
      "grad_norm": 7.004508018493652,
      "learning_rate": 3.714488584574448e-06,
      "loss": 3.6018,
      "step": 33345
    },
    {
      "epoch": 0.3392537434895833,
      "grad_norm": 11.859896659851074,
      "learning_rate": 3.7141390440897964e-06,
      "loss": 3.2948,
      "step": 33350
    },
    {
      "epoch": 0.3393046061197917,
      "grad_norm": 13.420574188232422,
      "learning_rate": 3.7137894725416863e-06,
      "loss": 3.476,
      "step": 33355
    },
    {
      "epoch": 0.33935546875,
      "grad_norm": 14.50550651550293,
      "learning_rate": 3.7134398699390597e-06,
      "loss": 3.2388,
      "step": 33360
    },
    {
      "epoch": 0.3394063313802083,
      "grad_norm": 14.22556209564209,
      "learning_rate": 3.7130902362908616e-06,
      "loss": 3.5979,
      "step": 33365
    },
    {
      "epoch": 0.3394571940104167,
      "grad_norm": 8.849228858947754,
      "learning_rate": 3.7127405716060376e-06,
      "loss": 3.4067,
      "step": 33370
    },
    {
      "epoch": 0.339508056640625,
      "grad_norm": 14.5337553024292,
      "learning_rate": 3.7123908758935332e-06,
      "loss": 3.3617,
      "step": 33375
    },
    {
      "epoch": 0.3395589192708333,
      "grad_norm": 11.952080726623535,
      "learning_rate": 3.712041149162296e-06,
      "loss": 3.4079,
      "step": 33380
    },
    {
      "epoch": 0.3396097819010417,
      "grad_norm": 8.4275484085083,
      "learning_rate": 3.7116913914212737e-06,
      "loss": 3.1598,
      "step": 33385
    },
    {
      "epoch": 0.33966064453125,
      "grad_norm": 17.485008239746094,
      "learning_rate": 3.7113416026794135e-06,
      "loss": 3.5281,
      "step": 33390
    },
    {
      "epoch": 0.3397115071614583,
      "grad_norm": 12.192771911621094,
      "learning_rate": 3.710991782945666e-06,
      "loss": 3.6066,
      "step": 33395
    },
    {
      "epoch": 0.3397623697916667,
      "grad_norm": 7.462334632873535,
      "learning_rate": 3.7106419322289807e-06,
      "loss": 3.3792,
      "step": 33400
    },
    {
      "epoch": 0.339813232421875,
      "grad_norm": 11.422493934631348,
      "learning_rate": 3.710292050538308e-06,
      "loss": 3.5719,
      "step": 33405
    },
    {
      "epoch": 0.3398640950520833,
      "grad_norm": 9.912327766418457,
      "learning_rate": 3.7099421378826004e-06,
      "loss": 3.5519,
      "step": 33410
    },
    {
      "epoch": 0.3399149576822917,
      "grad_norm": 11.72535228729248,
      "learning_rate": 3.7095921942708103e-06,
      "loss": 3.3383,
      "step": 33415
    },
    {
      "epoch": 0.3399658203125,
      "grad_norm": 8.569254875183105,
      "learning_rate": 3.7092422197118905e-06,
      "loss": 3.2259,
      "step": 33420
    },
    {
      "epoch": 0.3400166829427083,
      "grad_norm": 13.561548233032227,
      "learning_rate": 3.708892214214795e-06,
      "loss": 3.586,
      "step": 33425
    },
    {
      "epoch": 0.3400675455729167,
      "grad_norm": 16.515724182128906,
      "learning_rate": 3.7085421777884786e-06,
      "loss": 3.584,
      "step": 33430
    },
    {
      "epoch": 0.340118408203125,
      "grad_norm": 12.934800148010254,
      "learning_rate": 3.7081921104418974e-06,
      "loss": 3.8999,
      "step": 33435
    },
    {
      "epoch": 0.3401692708333333,
      "grad_norm": 7.906087398529053,
      "learning_rate": 3.7078420121840063e-06,
      "loss": 3.283,
      "step": 33440
    },
    {
      "epoch": 0.3402201334635417,
      "grad_norm": 15.088556289672852,
      "learning_rate": 3.7074918830237647e-06,
      "loss": 3.113,
      "step": 33445
    },
    {
      "epoch": 0.34027099609375,
      "grad_norm": 9.234477043151855,
      "learning_rate": 3.70714172297013e-06,
      "loss": 3.3398,
      "step": 33450
    },
    {
      "epoch": 0.3403218587239583,
      "grad_norm": 13.867288589477539,
      "learning_rate": 3.706791532032059e-06,
      "loss": 3.2631,
      "step": 33455
    },
    {
      "epoch": 0.3403727213541667,
      "grad_norm": 12.773273468017578,
      "learning_rate": 3.7064413102185137e-06,
      "loss": 3.3552,
      "step": 33460
    },
    {
      "epoch": 0.340423583984375,
      "grad_norm": 8.415233612060547,
      "learning_rate": 3.7060910575384534e-06,
      "loss": 3.3519,
      "step": 33465
    },
    {
      "epoch": 0.3404744466145833,
      "grad_norm": 14.105103492736816,
      "learning_rate": 3.705740774000839e-06,
      "loss": 3.4592,
      "step": 33470
    },
    {
      "epoch": 0.3405253092447917,
      "grad_norm": 7.179952621459961,
      "learning_rate": 3.705390459614633e-06,
      "loss": 3.0682,
      "step": 33475
    },
    {
      "epoch": 0.340576171875,
      "grad_norm": 8.724873542785645,
      "learning_rate": 3.705040114388797e-06,
      "loss": 3.6412,
      "step": 33480
    },
    {
      "epoch": 0.3406270345052083,
      "grad_norm": 13.953536033630371,
      "learning_rate": 3.7046897383322962e-06,
      "loss": 3.4858,
      "step": 33485
    },
    {
      "epoch": 0.3406778971354167,
      "grad_norm": 11.59171199798584,
      "learning_rate": 3.704339331454093e-06,
      "loss": 3.3904,
      "step": 33490
    },
    {
      "epoch": 0.340728759765625,
      "grad_norm": 9.839463233947754,
      "learning_rate": 3.703988893763154e-06,
      "loss": 3.2637,
      "step": 33495
    },
    {
      "epoch": 0.3407796223958333,
      "grad_norm": 12.875106811523438,
      "learning_rate": 3.7036384252684454e-06,
      "loss": 3.5458,
      "step": 33500
    },
    {
      "epoch": 0.3408304850260417,
      "grad_norm": 12.247830390930176,
      "learning_rate": 3.7032879259789315e-06,
      "loss": 3.0541,
      "step": 33505
    },
    {
      "epoch": 0.34088134765625,
      "grad_norm": 13.42300033569336,
      "learning_rate": 3.7029373959035825e-06,
      "loss": 3.296,
      "step": 33510
    },
    {
      "epoch": 0.3409322102864583,
      "grad_norm": 13.851029396057129,
      "learning_rate": 3.702586835051364e-06,
      "loss": 3.9352,
      "step": 33515
    },
    {
      "epoch": 0.3409830729166667,
      "grad_norm": 11.831171989440918,
      "learning_rate": 3.7022362434312475e-06,
      "loss": 3.2998,
      "step": 33520
    },
    {
      "epoch": 0.341033935546875,
      "grad_norm": 11.240053176879883,
      "learning_rate": 3.7018856210522013e-06,
      "loss": 4.1,
      "step": 33525
    },
    {
      "epoch": 0.3410847981770833,
      "grad_norm": 11.248284339904785,
      "learning_rate": 3.7015349679231964e-06,
      "loss": 3.0341,
      "step": 33530
    },
    {
      "epoch": 0.3411356608072917,
      "grad_norm": 12.049551963806152,
      "learning_rate": 3.7011842840532042e-06,
      "loss": 3.5037,
      "step": 33535
    },
    {
      "epoch": 0.3411865234375,
      "grad_norm": 14.031254768371582,
      "learning_rate": 3.7008335694511966e-06,
      "loss": 3.1509,
      "step": 33540
    },
    {
      "epoch": 0.3412373860677083,
      "grad_norm": 12.392552375793457,
      "learning_rate": 3.700482824126147e-06,
      "loss": 3.2808,
      "step": 33545
    },
    {
      "epoch": 0.3412882486979167,
      "grad_norm": 9.247160911560059,
      "learning_rate": 3.7001320480870286e-06,
      "loss": 3.1487,
      "step": 33550
    },
    {
      "epoch": 0.341339111328125,
      "grad_norm": 11.892711639404297,
      "learning_rate": 3.699781241342817e-06,
      "loss": 3.4006,
      "step": 33555
    },
    {
      "epoch": 0.3413899739583333,
      "grad_norm": 8.088289260864258,
      "learning_rate": 3.699430403902485e-06,
      "loss": 3.5574,
      "step": 33560
    },
    {
      "epoch": 0.3414408365885417,
      "grad_norm": 8.417235374450684,
      "learning_rate": 3.699079535775012e-06,
      "loss": 3.5206,
      "step": 33565
    },
    {
      "epoch": 0.34149169921875,
      "grad_norm": 14.191048622131348,
      "learning_rate": 3.6987286369693724e-06,
      "loss": 3.2619,
      "step": 33570
    },
    {
      "epoch": 0.3415425618489583,
      "grad_norm": 12.321990013122559,
      "learning_rate": 3.698377707494546e-06,
      "loss": 3.454,
      "step": 33575
    },
    {
      "epoch": 0.3415934244791667,
      "grad_norm": 13.851666450500488,
      "learning_rate": 3.698026747359509e-06,
      "loss": 3.1697,
      "step": 33580
    },
    {
      "epoch": 0.341644287109375,
      "grad_norm": 14.189324378967285,
      "learning_rate": 3.6976757565732413e-06,
      "loss": 3.4847,
      "step": 33585
    },
    {
      "epoch": 0.3416951497395833,
      "grad_norm": 9.495662689208984,
      "learning_rate": 3.697324735144724e-06,
      "loss": 3.3849,
      "step": 33590
    },
    {
      "epoch": 0.3417460123697917,
      "grad_norm": 12.85490894317627,
      "learning_rate": 3.6969736830829367e-06,
      "loss": 3.681,
      "step": 33595
    },
    {
      "epoch": 0.341796875,
      "grad_norm": 10.4622802734375,
      "learning_rate": 3.6966226003968613e-06,
      "loss": 3.3093,
      "step": 33600
    },
    {
      "epoch": 0.3418477376302083,
      "grad_norm": 13.98997974395752,
      "learning_rate": 3.6962714870954815e-06,
      "loss": 3.1084,
      "step": 33605
    },
    {
      "epoch": 0.3418986002604167,
      "grad_norm": 9.652063369750977,
      "learning_rate": 3.6959203431877777e-06,
      "loss": 3.8275,
      "step": 33610
    },
    {
      "epoch": 0.341949462890625,
      "grad_norm": 9.276095390319824,
      "learning_rate": 3.6955691686827367e-06,
      "loss": 3.2222,
      "step": 33615
    },
    {
      "epoch": 0.3420003255208333,
      "grad_norm": 13.359003067016602,
      "learning_rate": 3.695217963589341e-06,
      "loss": 3.5332,
      "step": 33620
    },
    {
      "epoch": 0.3420511881510417,
      "grad_norm": 9.17261028289795,
      "learning_rate": 3.694866727916576e-06,
      "loss": 3.1112,
      "step": 33625
    },
    {
      "epoch": 0.34210205078125,
      "grad_norm": 13.203807830810547,
      "learning_rate": 3.6945154616734315e-06,
      "loss": 3.4515,
      "step": 33630
    },
    {
      "epoch": 0.3421529134114583,
      "grad_norm": 12.045472145080566,
      "learning_rate": 3.6941641648688895e-06,
      "loss": 3.5353,
      "step": 33635
    },
    {
      "epoch": 0.3422037760416667,
      "grad_norm": 14.34790325164795,
      "learning_rate": 3.6938128375119425e-06,
      "loss": 3.4131,
      "step": 33640
    },
    {
      "epoch": 0.342254638671875,
      "grad_norm": 12.461421966552734,
      "learning_rate": 3.6934614796115764e-06,
      "loss": 3.2753,
      "step": 33645
    },
    {
      "epoch": 0.3423055013020833,
      "grad_norm": 11.347406387329102,
      "learning_rate": 3.69311009117678e-06,
      "loss": 3.6192,
      "step": 33650
    },
    {
      "epoch": 0.3423563639322917,
      "grad_norm": 9.508161544799805,
      "learning_rate": 3.6927586722165466e-06,
      "loss": 3.2605,
      "step": 33655
    },
    {
      "epoch": 0.3424072265625,
      "grad_norm": 11.28418254852295,
      "learning_rate": 3.692407222739864e-06,
      "loss": 3.005,
      "step": 33660
    },
    {
      "epoch": 0.3424580891927083,
      "grad_norm": 8.46851634979248,
      "learning_rate": 3.6920557427557257e-06,
      "loss": 3.5675,
      "step": 33665
    },
    {
      "epoch": 0.3425089518229167,
      "grad_norm": 12.884858131408691,
      "learning_rate": 3.691704232273124e-06,
      "loss": 3.572,
      "step": 33670
    },
    {
      "epoch": 0.342559814453125,
      "grad_norm": 12.269371032714844,
      "learning_rate": 3.691352691301051e-06,
      "loss": 3.5664,
      "step": 33675
    },
    {
      "epoch": 0.3426106770833333,
      "grad_norm": 12.720943450927734,
      "learning_rate": 3.691001119848503e-06,
      "loss": 3.2918,
      "step": 33680
    },
    {
      "epoch": 0.3426615397135417,
      "grad_norm": 8.249712944030762,
      "learning_rate": 3.690649517924473e-06,
      "loss": 3.2787,
      "step": 33685
    },
    {
      "epoch": 0.34271240234375,
      "grad_norm": 11.139851570129395,
      "learning_rate": 3.6902978855379577e-06,
      "loss": 3.1602,
      "step": 33690
    },
    {
      "epoch": 0.3427632649739583,
      "grad_norm": 9.868725776672363,
      "learning_rate": 3.689946222697953e-06,
      "loss": 3.0968,
      "step": 33695
    },
    {
      "epoch": 0.3428141276041667,
      "grad_norm": 19.89038848876953,
      "learning_rate": 3.6895945294134556e-06,
      "loss": 3.4411,
      "step": 33700
    },
    {
      "epoch": 0.342864990234375,
      "grad_norm": 13.837854385375977,
      "learning_rate": 3.689242805693465e-06,
      "loss": 3.3441,
      "step": 33705
    },
    {
      "epoch": 0.3429158528645833,
      "grad_norm": 13.5803804397583,
      "learning_rate": 3.688891051546979e-06,
      "loss": 3.2884,
      "step": 33710
    },
    {
      "epoch": 0.3429667154947917,
      "grad_norm": 15.511903762817383,
      "learning_rate": 3.688539266982996e-06,
      "loss": 3.4876,
      "step": 33715
    },
    {
      "epoch": 0.343017578125,
      "grad_norm": 11.785426139831543,
      "learning_rate": 3.688187452010519e-06,
      "loss": 3.4428,
      "step": 33720
    },
    {
      "epoch": 0.3430684407552083,
      "grad_norm": 14.247519493103027,
      "learning_rate": 3.6878356066385467e-06,
      "loss": 3.3219,
      "step": 33725
    },
    {
      "epoch": 0.3431193033854167,
      "grad_norm": 11.359977722167969,
      "learning_rate": 3.6874837308760823e-06,
      "loss": 3.2945,
      "step": 33730
    },
    {
      "epoch": 0.343170166015625,
      "grad_norm": 17.293371200561523,
      "learning_rate": 3.6871318247321288e-06,
      "loss": 3.3167,
      "step": 33735
    },
    {
      "epoch": 0.3432210286458333,
      "grad_norm": 10.258779525756836,
      "learning_rate": 3.686779888215688e-06,
      "loss": 3.1372,
      "step": 33740
    },
    {
      "epoch": 0.3432718912760417,
      "grad_norm": 11.674057006835938,
      "learning_rate": 3.6864279213357655e-06,
      "loss": 3.477,
      "step": 33745
    },
    {
      "epoch": 0.34332275390625,
      "grad_norm": 15.193865776062012,
      "learning_rate": 3.686075924101366e-06,
      "loss": 3.3141,
      "step": 33750
    },
    {
      "epoch": 0.3433736165364583,
      "grad_norm": 9.285595893859863,
      "learning_rate": 3.685723896521495e-06,
      "loss": 3.4461,
      "step": 33755
    },
    {
      "epoch": 0.3434244791666667,
      "grad_norm": 11.371384620666504,
      "learning_rate": 3.685371838605159e-06,
      "loss": 3.3281,
      "step": 33760
    },
    {
      "epoch": 0.343475341796875,
      "grad_norm": 16.047948837280273,
      "learning_rate": 3.6850197503613656e-06,
      "loss": 3.1408,
      "step": 33765
    },
    {
      "epoch": 0.3435262044270833,
      "grad_norm": 10.679207801818848,
      "learning_rate": 3.6846676317991227e-06,
      "loss": 3.4766,
      "step": 33770
    },
    {
      "epoch": 0.3435770670572917,
      "grad_norm": 11.644343376159668,
      "learning_rate": 3.6843154829274397e-06,
      "loss": 3.4585,
      "step": 33775
    },
    {
      "epoch": 0.3436279296875,
      "grad_norm": 12.160526275634766,
      "learning_rate": 3.6839633037553245e-06,
      "loss": 3.1787,
      "step": 33780
    },
    {
      "epoch": 0.3436787923177083,
      "grad_norm": 13.829774856567383,
      "learning_rate": 3.68361109429179e-06,
      "loss": 3.2787,
      "step": 33785
    },
    {
      "epoch": 0.3437296549479167,
      "grad_norm": 10.009960174560547,
      "learning_rate": 3.6832588545458454e-06,
      "loss": 3.3127,
      "step": 33790
    },
    {
      "epoch": 0.343780517578125,
      "grad_norm": 17.351730346679688,
      "learning_rate": 3.6829065845265043e-06,
      "loss": 3.5596,
      "step": 33795
    },
    {
      "epoch": 0.3438313802083333,
      "grad_norm": 14.011494636535645,
      "learning_rate": 3.682554284242778e-06,
      "loss": 3.2904,
      "step": 33800
    },
    {
      "epoch": 0.3438822428385417,
      "grad_norm": 7.7989959716796875,
      "learning_rate": 3.6822019537036814e-06,
      "loss": 3.4226,
      "step": 33805
    },
    {
      "epoch": 0.34393310546875,
      "grad_norm": 9.405271530151367,
      "learning_rate": 3.681849592918227e-06,
      "loss": 4.0221,
      "step": 33810
    },
    {
      "epoch": 0.3439839680989583,
      "grad_norm": 14.77958869934082,
      "learning_rate": 3.6814972018954313e-06,
      "loss": 3.4452,
      "step": 33815
    },
    {
      "epoch": 0.3440348307291667,
      "grad_norm": 12.135478019714355,
      "learning_rate": 3.68114478064431e-06,
      "loss": 3.0434,
      "step": 33820
    },
    {
      "epoch": 0.344085693359375,
      "grad_norm": 13.508832931518555,
      "learning_rate": 3.680792329173879e-06,
      "loss": 3.3546,
      "step": 33825
    },
    {
      "epoch": 0.3441365559895833,
      "grad_norm": 14.546008110046387,
      "learning_rate": 3.6804398474931563e-06,
      "loss": 2.841,
      "step": 33830
    },
    {
      "epoch": 0.3441874186197917,
      "grad_norm": 10.957643508911133,
      "learning_rate": 3.68008733561116e-06,
      "loss": 3.2414,
      "step": 33835
    },
    {
      "epoch": 0.34423828125,
      "grad_norm": 10.273690223693848,
      "learning_rate": 3.679734793536909e-06,
      "loss": 3.2178,
      "step": 33840
    },
    {
      "epoch": 0.3442891438802083,
      "grad_norm": 15.164803504943848,
      "learning_rate": 3.6793822212794227e-06,
      "loss": 3.2884,
      "step": 33845
    },
    {
      "epoch": 0.3443400065104167,
      "grad_norm": 11.058144569396973,
      "learning_rate": 3.6790296188477228e-06,
      "loss": 3.6806,
      "step": 33850
    },
    {
      "epoch": 0.344390869140625,
      "grad_norm": 7.803112506866455,
      "learning_rate": 3.6786769862508282e-06,
      "loss": 3.3499,
      "step": 33855
    },
    {
      "epoch": 0.3444417317708333,
      "grad_norm": 9.694926261901855,
      "learning_rate": 3.678324323497763e-06,
      "loss": 4.0269,
      "step": 33860
    },
    {
      "epoch": 0.3444925944010417,
      "grad_norm": 15.972514152526855,
      "learning_rate": 3.6779716305975487e-06,
      "loss": 3.8841,
      "step": 33865
    },
    {
      "epoch": 0.34454345703125,
      "grad_norm": 15.146234512329102,
      "learning_rate": 3.67761890755921e-06,
      "loss": 3.0953,
      "step": 33870
    },
    {
      "epoch": 0.3445943196614583,
      "grad_norm": 14.086612701416016,
      "learning_rate": 3.6772661543917703e-06,
      "loss": 3.8149,
      "step": 33875
    },
    {
      "epoch": 0.3446451822916667,
      "grad_norm": 10.04620361328125,
      "learning_rate": 3.6769133711042556e-06,
      "loss": 3.4665,
      "step": 33880
    },
    {
      "epoch": 0.344696044921875,
      "grad_norm": 10.64113998413086,
      "learning_rate": 3.6765605577056907e-06,
      "loss": 3.5521,
      "step": 33885
    },
    {
      "epoch": 0.3447469075520833,
      "grad_norm": 13.816940307617188,
      "learning_rate": 3.6762077142051034e-06,
      "loss": 3.4866,
      "step": 33890
    },
    {
      "epoch": 0.3447977701822917,
      "grad_norm": 8.723909378051758,
      "learning_rate": 3.6758548406115196e-06,
      "loss": 3.1867,
      "step": 33895
    },
    {
      "epoch": 0.3448486328125,
      "grad_norm": 10.090558052062988,
      "learning_rate": 3.6755019369339695e-06,
      "loss": 3.2891,
      "step": 33900
    },
    {
      "epoch": 0.3448994954427083,
      "grad_norm": 9.237555503845215,
      "learning_rate": 3.67514900318148e-06,
      "loss": 3.3041,
      "step": 33905
    },
    {
      "epoch": 0.3449503580729167,
      "grad_norm": 9.231812477111816,
      "learning_rate": 3.674796039363083e-06,
      "loss": 3.2584,
      "step": 33910
    },
    {
      "epoch": 0.345001220703125,
      "grad_norm": 11.18399715423584,
      "learning_rate": 3.674443045487807e-06,
      "loss": 3.2969,
      "step": 33915
    },
    {
      "epoch": 0.3450520833333333,
      "grad_norm": 13.937256813049316,
      "learning_rate": 3.6740900215646836e-06,
      "loss": 3.417,
      "step": 33920
    },
    {
      "epoch": 0.3451029459635417,
      "grad_norm": 13.297497749328613,
      "learning_rate": 3.6737369676027457e-06,
      "loss": 3.1705,
      "step": 33925
    },
    {
      "epoch": 0.34515380859375,
      "grad_norm": 9.702543258666992,
      "learning_rate": 3.6733838836110258e-06,
      "loss": 3.1908,
      "step": 33930
    },
    {
      "epoch": 0.3452046712239583,
      "grad_norm": 9.547173500061035,
      "learning_rate": 3.6730307695985575e-06,
      "loss": 3.3946,
      "step": 33935
    },
    {
      "epoch": 0.3452555338541667,
      "grad_norm": 11.548481941223145,
      "learning_rate": 3.6726776255743744e-06,
      "loss": 3.4355,
      "step": 33940
    },
    {
      "epoch": 0.345306396484375,
      "grad_norm": 16.107242584228516,
      "learning_rate": 3.6723244515475126e-06,
      "loss": 3.2942,
      "step": 33945
    },
    {
      "epoch": 0.3453572591145833,
      "grad_norm": 8.494128227233887,
      "learning_rate": 3.6719712475270075e-06,
      "loss": 3.4592,
      "step": 33950
    },
    {
      "epoch": 0.3454081217447917,
      "grad_norm": 13.408146858215332,
      "learning_rate": 3.6716180135218965e-06,
      "loss": 3.3184,
      "step": 33955
    },
    {
      "epoch": 0.345458984375,
      "grad_norm": 14.734374046325684,
      "learning_rate": 3.6712647495412153e-06,
      "loss": 3.2729,
      "step": 33960
    },
    {
      "epoch": 0.3455098470052083,
      "grad_norm": 18.635473251342773,
      "learning_rate": 3.670911455594004e-06,
      "loss": 3.3193,
      "step": 33965
    },
    {
      "epoch": 0.3455607096354167,
      "grad_norm": 15.248563766479492,
      "learning_rate": 3.6705581316893e-06,
      "loss": 3.3903,
      "step": 33970
    },
    {
      "epoch": 0.345611572265625,
      "grad_norm": 14.444631576538086,
      "learning_rate": 3.670204777836144e-06,
      "loss": 3.2452,
      "step": 33975
    },
    {
      "epoch": 0.3456624348958333,
      "grad_norm": 9.606155395507812,
      "learning_rate": 3.669851394043576e-06,
      "loss": 3.0582,
      "step": 33980
    },
    {
      "epoch": 0.3457132975260417,
      "grad_norm": 18.810192108154297,
      "learning_rate": 3.669497980320637e-06,
      "loss": 2.8645,
      "step": 33985
    },
    {
      "epoch": 0.34576416015625,
      "grad_norm": 8.963079452514648,
      "learning_rate": 3.66914453667637e-06,
      "loss": 3.222,
      "step": 33990
    },
    {
      "epoch": 0.3458150227864583,
      "grad_norm": 10.067646026611328,
      "learning_rate": 3.668791063119817e-06,
      "loss": 3.4052,
      "step": 33995
    },
    {
      "epoch": 0.3458658854166667,
      "grad_norm": 7.015271186828613,
      "learning_rate": 3.6684375596600215e-06,
      "loss": 3.2415,
      "step": 34000
    },
    {
      "epoch": 0.345916748046875,
      "grad_norm": 14.001973152160645,
      "learning_rate": 3.668084026306028e-06,
      "loss": 3.3947,
      "step": 34005
    },
    {
      "epoch": 0.3459676106770833,
      "grad_norm": 11.566573143005371,
      "learning_rate": 3.667730463066882e-06,
      "loss": 3.3749,
      "step": 34010
    },
    {
      "epoch": 0.3460184733072917,
      "grad_norm": 13.14124870300293,
      "learning_rate": 3.667376869951628e-06,
      "loss": 3.2533,
      "step": 34015
    },
    {
      "epoch": 0.3460693359375,
      "grad_norm": 9.880794525146484,
      "learning_rate": 3.667023246969314e-06,
      "loss": 3.0014,
      "step": 34020
    },
    {
      "epoch": 0.3461201985677083,
      "grad_norm": 13.688362121582031,
      "learning_rate": 3.6666695941289866e-06,
      "loss": 3.4169,
      "step": 34025
    },
    {
      "epoch": 0.3461710611979167,
      "grad_norm": 13.356725692749023,
      "learning_rate": 3.6663159114396946e-06,
      "loss": 3.3283,
      "step": 34030
    },
    {
      "epoch": 0.346221923828125,
      "grad_norm": 15.20695686340332,
      "learning_rate": 3.665962198910486e-06,
      "loss": 3.1513,
      "step": 34035
    },
    {
      "epoch": 0.3462727864583333,
      "grad_norm": 16.854326248168945,
      "learning_rate": 3.665608456550411e-06,
      "loss": 3.7669,
      "step": 34040
    },
    {
      "epoch": 0.3463236490885417,
      "grad_norm": 10.738151550292969,
      "learning_rate": 3.66525468436852e-06,
      "loss": 3.5412,
      "step": 34045
    },
    {
      "epoch": 0.34637451171875,
      "grad_norm": 30.94281578063965,
      "learning_rate": 3.664900882373864e-06,
      "loss": 3.4419,
      "step": 34050
    },
    {
      "epoch": 0.3464253743489583,
      "grad_norm": 10.15609359741211,
      "learning_rate": 3.6645470505754954e-06,
      "loss": 3.7038,
      "step": 34055
    },
    {
      "epoch": 0.3464762369791667,
      "grad_norm": 10.925048828125,
      "learning_rate": 3.6641931889824666e-06,
      "loss": 3.0991,
      "step": 34060
    },
    {
      "epoch": 0.346527099609375,
      "grad_norm": 15.451886177062988,
      "learning_rate": 3.6638392976038304e-06,
      "loss": 3.3793,
      "step": 34065
    },
    {
      "epoch": 0.3465779622395833,
      "grad_norm": 10.374808311462402,
      "learning_rate": 3.663485376448642e-06,
      "loss": 3.9683,
      "step": 34070
    },
    {
      "epoch": 0.3466288248697917,
      "grad_norm": 10.30376148223877,
      "learning_rate": 3.6631314255259563e-06,
      "loss": 3.2515,
      "step": 34075
    },
    {
      "epoch": 0.3466796875,
      "grad_norm": 10.25160026550293,
      "learning_rate": 3.6627774448448284e-06,
      "loss": 3.009,
      "step": 34080
    },
    {
      "epoch": 0.3467305501302083,
      "grad_norm": 11.332584381103516,
      "learning_rate": 3.662423434414315e-06,
      "loss": 3.1738,
      "step": 34085
    },
    {
      "epoch": 0.3467814127604167,
      "grad_norm": 10.428305625915527,
      "learning_rate": 3.662069394243474e-06,
      "loss": 3.2948,
      "step": 34090
    },
    {
      "epoch": 0.346832275390625,
      "grad_norm": 11.851875305175781,
      "learning_rate": 3.661715324341363e-06,
      "loss": 3.5217,
      "step": 34095
    },
    {
      "epoch": 0.3468831380208333,
      "grad_norm": 12.3597993850708,
      "learning_rate": 3.6613612247170407e-06,
      "loss": 3.642,
      "step": 34100
    },
    {
      "epoch": 0.3469340006510417,
      "grad_norm": 16.413686752319336,
      "learning_rate": 3.6610070953795662e-06,
      "loss": 3.5137,
      "step": 34105
    },
    {
      "epoch": 0.34698486328125,
      "grad_norm": 8.804444313049316,
      "learning_rate": 3.660652936338001e-06,
      "loss": 3.2862,
      "step": 34110
    },
    {
      "epoch": 0.3470357259114583,
      "grad_norm": 17.43308448791504,
      "learning_rate": 3.6602987476014045e-06,
      "loss": 3.6762,
      "step": 34115
    },
    {
      "epoch": 0.3470865885416667,
      "grad_norm": 15.472068786621094,
      "learning_rate": 3.6599445291788404e-06,
      "loss": 3.5438,
      "step": 34120
    },
    {
      "epoch": 0.347137451171875,
      "grad_norm": 16.01087188720703,
      "learning_rate": 3.659590281079371e-06,
      "loss": 3.5225,
      "step": 34125
    },
    {
      "epoch": 0.3471883138020833,
      "grad_norm": 14.543718338012695,
      "learning_rate": 3.659236003312058e-06,
      "loss": 3.8949,
      "step": 34130
    },
    {
      "epoch": 0.3472391764322917,
      "grad_norm": 13.618134498596191,
      "learning_rate": 3.6588816958859673e-06,
      "loss": 3.3487,
      "step": 34135
    },
    {
      "epoch": 0.3472900390625,
      "grad_norm": 13.667912483215332,
      "learning_rate": 3.6585273588101634e-06,
      "loss": 3.2609,
      "step": 34140
    },
    {
      "epoch": 0.3473409016927083,
      "grad_norm": 12.12238883972168,
      "learning_rate": 3.658172992093711e-06,
      "loss": 3.2387,
      "step": 34145
    },
    {
      "epoch": 0.3473917643229167,
      "grad_norm": 9.653349876403809,
      "learning_rate": 3.6578185957456767e-06,
      "loss": 3.2573,
      "step": 34150
    },
    {
      "epoch": 0.347442626953125,
      "grad_norm": 9.517043113708496,
      "learning_rate": 3.657464169775129e-06,
      "loss": 3.5676,
      "step": 34155
    },
    {
      "epoch": 0.3474934895833333,
      "grad_norm": 12.061582565307617,
      "learning_rate": 3.6571097141911342e-06,
      "loss": 3.4772,
      "step": 34160
    },
    {
      "epoch": 0.3475443522135417,
      "grad_norm": 8.128222465515137,
      "learning_rate": 3.6567552290027626e-06,
      "loss": 4.0439,
      "step": 34165
    },
    {
      "epoch": 0.34759521484375,
      "grad_norm": 82.14265441894531,
      "learning_rate": 3.6564007142190818e-06,
      "loss": 3.4596,
      "step": 34170
    },
    {
      "epoch": 0.3476460774739583,
      "grad_norm": 9.599939346313477,
      "learning_rate": 3.6560461698491627e-06,
      "loss": 3.1819,
      "step": 34175
    },
    {
      "epoch": 0.3476969401041667,
      "grad_norm": 10.134929656982422,
      "learning_rate": 3.6556915959020766e-06,
      "loss": 3.5317,
      "step": 34180
    },
    {
      "epoch": 0.347747802734375,
      "grad_norm": 12.05319881439209,
      "learning_rate": 3.6553369923868953e-06,
      "loss": 3.8354,
      "step": 34185
    },
    {
      "epoch": 0.3477986653645833,
      "grad_norm": 16.20045280456543,
      "learning_rate": 3.6549823593126905e-06,
      "loss": 3.3641,
      "step": 34190
    },
    {
      "epoch": 0.3478495279947917,
      "grad_norm": 13.195075035095215,
      "learning_rate": 3.654627696688536e-06,
      "loss": 3.6895,
      "step": 34195
    },
    {
      "epoch": 0.347900390625,
      "grad_norm": 10.847936630249023,
      "learning_rate": 3.654273004523505e-06,
      "loss": 3.9151,
      "step": 34200
    },
    {
      "epoch": 0.3479512532552083,
      "grad_norm": 11.85213851928711,
      "learning_rate": 3.6539182828266736e-06,
      "loss": 3.8494,
      "step": 34205
    },
    {
      "epoch": 0.3480021158854167,
      "grad_norm": 12.77187442779541,
      "learning_rate": 3.653563531607116e-06,
      "loss": 3.1948,
      "step": 34210
    },
    {
      "epoch": 0.348052978515625,
      "grad_norm": 11.966257095336914,
      "learning_rate": 3.653208750873909e-06,
      "loss": 3.4879,
      "step": 34215
    },
    {
      "epoch": 0.3481038411458333,
      "grad_norm": 10.672539710998535,
      "learning_rate": 3.652853940636129e-06,
      "loss": 3.4875,
      "step": 34220
    },
    {
      "epoch": 0.3481547037760417,
      "grad_norm": 11.228493690490723,
      "learning_rate": 3.6524991009028544e-06,
      "loss": 3.4064,
      "step": 34225
    },
    {
      "epoch": 0.34820556640625,
      "grad_norm": 10.102757453918457,
      "learning_rate": 3.6521442316831645e-06,
      "loss": 3.3489,
      "step": 34230
    },
    {
      "epoch": 0.3482564290364583,
      "grad_norm": 10.402213096618652,
      "learning_rate": 3.651789332986136e-06,
      "loss": 3.0321,
      "step": 34235
    },
    {
      "epoch": 0.3483072916666667,
      "grad_norm": 15.23025894165039,
      "learning_rate": 3.6514344048208518e-06,
      "loss": 3.3532,
      "step": 34240
    },
    {
      "epoch": 0.348358154296875,
      "grad_norm": 126.61017608642578,
      "learning_rate": 3.6510794471963903e-06,
      "loss": 3.3617,
      "step": 34245
    },
    {
      "epoch": 0.3484090169270833,
      "grad_norm": 9.169806480407715,
      "learning_rate": 3.650724460121834e-06,
      "loss": 3.3471,
      "step": 34250
    },
    {
      "epoch": 0.3484598795572917,
      "grad_norm": 16.55471420288086,
      "learning_rate": 3.650369443606265e-06,
      "loss": 3.2522,
      "step": 34255
    },
    {
      "epoch": 0.3485107421875,
      "grad_norm": 13.121299743652344,
      "learning_rate": 3.650014397658767e-06,
      "loss": 3.428,
      "step": 34260
    },
    {
      "epoch": 0.3485616048177083,
      "grad_norm": 11.240519523620605,
      "learning_rate": 3.649659322288423e-06,
      "loss": 3.2809,
      "step": 34265
    },
    {
      "epoch": 0.3486124674479167,
      "grad_norm": 13.24062728881836,
      "learning_rate": 3.6493042175043182e-06,
      "loss": 3.2971,
      "step": 34270
    },
    {
      "epoch": 0.348663330078125,
      "grad_norm": 9.990971565246582,
      "learning_rate": 3.6489490833155365e-06,
      "loss": 3.0945,
      "step": 34275
    },
    {
      "epoch": 0.3487141927083333,
      "grad_norm": 17.911039352416992,
      "learning_rate": 3.6485939197311658e-06,
      "loss": 3.2916,
      "step": 34280
    },
    {
      "epoch": 0.3487650553385417,
      "grad_norm": 8.286005020141602,
      "learning_rate": 3.648238726760291e-06,
      "loss": 3.2104,
      "step": 34285
    },
    {
      "epoch": 0.34881591796875,
      "grad_norm": 11.795724868774414,
      "learning_rate": 3.6478835044120016e-06,
      "loss": 3.0933,
      "step": 34290
    },
    {
      "epoch": 0.3488667805989583,
      "grad_norm": 11.091285705566406,
      "learning_rate": 3.6475282526953838e-06,
      "loss": 3.4352,
      "step": 34295
    },
    {
      "epoch": 0.3489176432291667,
      "grad_norm": 7.997657775878906,
      "learning_rate": 3.6471729716195283e-06,
      "loss": 3.3882,
      "step": 34300
    },
    {
      "epoch": 0.348968505859375,
      "grad_norm": 7.602253437042236,
      "learning_rate": 3.6468176611935237e-06,
      "loss": 3.1502,
      "step": 34305
    },
    {
      "epoch": 0.3490193684895833,
      "grad_norm": 9.574580192565918,
      "learning_rate": 3.6464623214264618e-06,
      "loss": 3.1318,
      "step": 34310
    },
    {
      "epoch": 0.3490702311197917,
      "grad_norm": 14.666267395019531,
      "learning_rate": 3.6461069523274324e-06,
      "loss": 3.381,
      "step": 34315
    },
    {
      "epoch": 0.34912109375,
      "grad_norm": 12.042957305908203,
      "learning_rate": 3.6457515539055284e-06,
      "loss": 3.4057,
      "step": 34320
    },
    {
      "epoch": 0.3491719563802083,
      "grad_norm": 12.966578483581543,
      "learning_rate": 3.6453961261698424e-06,
      "loss": 3.361,
      "step": 34325
    },
    {
      "epoch": 0.3492228190104167,
      "grad_norm": 12.08524227142334,
      "learning_rate": 3.645040669129468e-06,
      "loss": 3.4151,
      "step": 34330
    },
    {
      "epoch": 0.349273681640625,
      "grad_norm": 9.36681079864502,
      "learning_rate": 3.6446851827935e-06,
      "loss": 3.6036,
      "step": 34335
    },
    {
      "epoch": 0.3493245442708333,
      "grad_norm": 11.86538028717041,
      "learning_rate": 3.6443296671710326e-06,
      "loss": 3.5037,
      "step": 34340
    },
    {
      "epoch": 0.3493754069010417,
      "grad_norm": 8.338165283203125,
      "learning_rate": 3.643974122271162e-06,
      "loss": 3.2591,
      "step": 34345
    },
    {
      "epoch": 0.34942626953125,
      "grad_norm": 12.40888786315918,
      "learning_rate": 3.643618548102984e-06,
      "loss": 3.3841,
      "step": 34350
    },
    {
      "epoch": 0.3494771321614583,
      "grad_norm": 15.538311958312988,
      "learning_rate": 3.643262944675598e-06,
      "loss": 3.423,
      "step": 34355
    },
    {
      "epoch": 0.3495279947916667,
      "grad_norm": 13.414130210876465,
      "learning_rate": 3.6429073119980994e-06,
      "loss": 3.4831,
      "step": 34360
    },
    {
      "epoch": 0.349578857421875,
      "grad_norm": 10.631868362426758,
      "learning_rate": 3.6425516500795883e-06,
      "loss": 3.6589,
      "step": 34365
    },
    {
      "epoch": 0.3496297200520833,
      "grad_norm": 10.415438652038574,
      "learning_rate": 3.642195958929164e-06,
      "loss": 3.522,
      "step": 34370
    },
    {
      "epoch": 0.3496805826822917,
      "grad_norm": 11.739583015441895,
      "learning_rate": 3.641840238555927e-06,
      "loss": 3.6878,
      "step": 34375
    },
    {
      "epoch": 0.3497314453125,
      "grad_norm": 7.453786849975586,
      "learning_rate": 3.6414844889689784e-06,
      "loss": 3.1575,
      "step": 34380
    },
    {
      "epoch": 0.3497823079427083,
      "grad_norm": 13.174354553222656,
      "learning_rate": 3.6411287101774194e-06,
      "loss": 3.3389,
      "step": 34385
    },
    {
      "epoch": 0.3498331705729167,
      "grad_norm": 15.388251304626465,
      "learning_rate": 3.640772902190352e-06,
      "loss": 3.2237,
      "step": 34390
    },
    {
      "epoch": 0.349884033203125,
      "grad_norm": 13.908624649047852,
      "learning_rate": 3.6404170650168817e-06,
      "loss": 3.4687,
      "step": 34395
    },
    {
      "epoch": 0.3499348958333333,
      "grad_norm": 13.464629173278809,
      "learning_rate": 3.6400611986661107e-06,
      "loss": 3.3808,
      "step": 34400
    },
    {
      "epoch": 0.3499857584635417,
      "grad_norm": 15.773457527160645,
      "learning_rate": 3.6397053031471435e-06,
      "loss": 3.3681,
      "step": 34405
    },
    {
      "epoch": 0.35003662109375,
      "grad_norm": 10.64908218383789,
      "learning_rate": 3.639349378469087e-06,
      "loss": 3.5906,
      "step": 34410
    },
    {
      "epoch": 0.3500874837239583,
      "grad_norm": 10.5111722946167,
      "learning_rate": 3.6389934246410456e-06,
      "loss": 3.145,
      "step": 34415
    },
    {
      "epoch": 0.3501383463541667,
      "grad_norm": 11.134115219116211,
      "learning_rate": 3.6386374416721286e-06,
      "loss": 2.9944,
      "step": 34420
    },
    {
      "epoch": 0.350189208984375,
      "grad_norm": 10.555818557739258,
      "learning_rate": 3.638281429571442e-06,
      "loss": 3.4728,
      "step": 34425
    },
    {
      "epoch": 0.3502400716145833,
      "grad_norm": 16.187204360961914,
      "learning_rate": 3.637925388348095e-06,
      "loss": 3.3397,
      "step": 34430
    },
    {
      "epoch": 0.3502909342447917,
      "grad_norm": 8.328694343566895,
      "learning_rate": 3.6375693180111965e-06,
      "loss": 3.543,
      "step": 34435
    },
    {
      "epoch": 0.350341796875,
      "grad_norm": 10.651793479919434,
      "learning_rate": 3.637213218569857e-06,
      "loss": 3.9065,
      "step": 34440
    },
    {
      "epoch": 0.3503926595052083,
      "grad_norm": 14.604371070861816,
      "learning_rate": 3.6368570900331866e-06,
      "loss": 3.3349,
      "step": 34445
    },
    {
      "epoch": 0.3504435221354167,
      "grad_norm": 8.624008178710938,
      "learning_rate": 3.6365009324102975e-06,
      "loss": 3.1049,
      "step": 34450
    },
    {
      "epoch": 0.350494384765625,
      "grad_norm": 8.160076141357422,
      "learning_rate": 3.6361447457103004e-06,
      "loss": 3.5228,
      "step": 34455
    },
    {
      "epoch": 0.3505452473958333,
      "grad_norm": 7.852463245391846,
      "learning_rate": 3.63578852994231e-06,
      "loss": 3.3375,
      "step": 34460
    },
    {
      "epoch": 0.3505961100260417,
      "grad_norm": 12.908072471618652,
      "learning_rate": 3.6354322851154397e-06,
      "loss": 3.2781,
      "step": 34465
    },
    {
      "epoch": 0.35064697265625,
      "grad_norm": 13.951038360595703,
      "learning_rate": 3.6350760112388024e-06,
      "loss": 3.3372,
      "step": 34470
    },
    {
      "epoch": 0.3506978352864583,
      "grad_norm": 11.072229385375977,
      "learning_rate": 3.6347197083215147e-06,
      "loss": 3.3124,
      "step": 34475
    },
    {
      "epoch": 0.3507486979166667,
      "grad_norm": 9.052788734436035,
      "learning_rate": 3.634363376372693e-06,
      "loss": 3.3465,
      "step": 34480
    },
    {
      "epoch": 0.350799560546875,
      "grad_norm": 7.021347999572754,
      "learning_rate": 3.634007015401452e-06,
      "loss": 3.2082,
      "step": 34485
    },
    {
      "epoch": 0.3508504231770833,
      "grad_norm": 13.342665672302246,
      "learning_rate": 3.6336506254169113e-06,
      "loss": 3.3017,
      "step": 34490
    },
    {
      "epoch": 0.3509012858072917,
      "grad_norm": 11.334712028503418,
      "learning_rate": 3.6332942064281874e-06,
      "loss": 3.3756,
      "step": 34495
    },
    {
      "epoch": 0.3509521484375,
      "grad_norm": 9.231757164001465,
      "learning_rate": 3.6329377584444004e-06,
      "loss": 3.0124,
      "step": 34500
    },
    {
      "epoch": 0.3510030110677083,
      "grad_norm": 11.232418060302734,
      "learning_rate": 3.6325812814746687e-06,
      "loss": 3.6996,
      "step": 34505
    },
    {
      "epoch": 0.3510538736979167,
      "grad_norm": 14.518206596374512,
      "learning_rate": 3.6322247755281144e-06,
      "loss": 3.1982,
      "step": 34510
    },
    {
      "epoch": 0.351104736328125,
      "grad_norm": 15.891706466674805,
      "learning_rate": 3.6318682406138565e-06,
      "loss": 3.0786,
      "step": 34515
    },
    {
      "epoch": 0.3511555989583333,
      "grad_norm": 8.141637802124023,
      "learning_rate": 3.631511676741018e-06,
      "loss": 3.4413,
      "step": 34520
    },
    {
      "epoch": 0.3512064615885417,
      "grad_norm": 13.049849510192871,
      "learning_rate": 3.6311550839187227e-06,
      "loss": 2.6128,
      "step": 34525
    },
    {
      "epoch": 0.35125732421875,
      "grad_norm": 10.91442584991455,
      "learning_rate": 3.6307984621560914e-06,
      "loss": 3.0481,
      "step": 34530
    },
    {
      "epoch": 0.3513081868489583,
      "grad_norm": 11.624651908874512,
      "learning_rate": 3.6304418114622498e-06,
      "loss": 3.4811,
      "step": 34535
    },
    {
      "epoch": 0.3513590494791667,
      "grad_norm": 11.577051162719727,
      "learning_rate": 3.6300851318463224e-06,
      "loss": 3.4565,
      "step": 34540
    },
    {
      "epoch": 0.351409912109375,
      "grad_norm": 13.190190315246582,
      "learning_rate": 3.629728423317435e-06,
      "loss": 3.0826,
      "step": 34545
    },
    {
      "epoch": 0.3514607747395833,
      "grad_norm": 10.728463172912598,
      "learning_rate": 3.629371685884714e-06,
      "loss": 3.6738,
      "step": 34550
    },
    {
      "epoch": 0.3515116373697917,
      "grad_norm": 9.551852226257324,
      "learning_rate": 3.6290149195572853e-06,
      "loss": 3.2391,
      "step": 34555
    },
    {
      "epoch": 0.3515625,
      "grad_norm": 10.663229942321777,
      "learning_rate": 3.6286581243442776e-06,
      "loss": 3.7829,
      "step": 34560
    },
    {
      "epoch": 0.3516133626302083,
      "grad_norm": 12.385886192321777,
      "learning_rate": 3.6283013002548202e-06,
      "loss": 3.2569,
      "step": 34565
    },
    {
      "epoch": 0.3516642252604167,
      "grad_norm": 9.58331298828125,
      "learning_rate": 3.62794444729804e-06,
      "loss": 3.1043,
      "step": 34570
    },
    {
      "epoch": 0.351715087890625,
      "grad_norm": 10.994060516357422,
      "learning_rate": 3.62758756548307e-06,
      "loss": 3.6247,
      "step": 34575
    },
    {
      "epoch": 0.3517659505208333,
      "grad_norm": 9.025533676147461,
      "learning_rate": 3.6272306548190383e-06,
      "loss": 3.1066,
      "step": 34580
    },
    {
      "epoch": 0.3518168131510417,
      "grad_norm": 13.23122501373291,
      "learning_rate": 3.626873715315078e-06,
      "loss": 3.2877,
      "step": 34585
    },
    {
      "epoch": 0.35186767578125,
      "grad_norm": 14.55508804321289,
      "learning_rate": 3.6265167469803207e-06,
      "loss": 3.6446,
      "step": 34590
    },
    {
      "epoch": 0.3519185384114583,
      "grad_norm": 14.5059232711792,
      "learning_rate": 3.6261597498238993e-06,
      "loss": 3.2737,
      "step": 34595
    },
    {
      "epoch": 0.3519694010416667,
      "grad_norm": 11.445781707763672,
      "learning_rate": 3.6258027238549475e-06,
      "loss": 3.0839,
      "step": 34600
    },
    {
      "epoch": 0.352020263671875,
      "grad_norm": 12.936619758605957,
      "learning_rate": 3.6254456690826005e-06,
      "loss": 3.4727,
      "step": 34605
    },
    {
      "epoch": 0.3520711263020833,
      "grad_norm": 14.624140739440918,
      "learning_rate": 3.6250885855159924e-06,
      "loss": 3.5822,
      "step": 34610
    },
    {
      "epoch": 0.3521219889322917,
      "grad_norm": 11.065165519714355,
      "learning_rate": 3.624731473164259e-06,
      "loss": 3.2801,
      "step": 34615
    },
    {
      "epoch": 0.3521728515625,
      "grad_norm": 13.35992431640625,
      "learning_rate": 3.624374332036539e-06,
      "loss": 3.3101,
      "step": 34620
    },
    {
      "epoch": 0.3522237141927083,
      "grad_norm": 12.415696144104004,
      "learning_rate": 3.6240171621419663e-06,
      "loss": 3.4841,
      "step": 34625
    },
    {
      "epoch": 0.3522745768229167,
      "grad_norm": 11.691892623901367,
      "learning_rate": 3.6236599634896823e-06,
      "loss": 3.3766,
      "step": 34630
    },
    {
      "epoch": 0.352325439453125,
      "grad_norm": 10.133980751037598,
      "learning_rate": 3.6233027360888236e-06,
      "loss": 3.3009,
      "step": 34635
    },
    {
      "epoch": 0.3523763020833333,
      "grad_norm": 11.231637001037598,
      "learning_rate": 3.6229454799485316e-06,
      "loss": 3.2301,
      "step": 34640
    },
    {
      "epoch": 0.3524271647135417,
      "grad_norm": 9.963400840759277,
      "learning_rate": 3.6225881950779447e-06,
      "loss": 3.2479,
      "step": 34645
    },
    {
      "epoch": 0.35247802734375,
      "grad_norm": 14.1806640625,
      "learning_rate": 3.622230881486205e-06,
      "loss": 3.3654,
      "step": 34650
    },
    {
      "epoch": 0.3525288899739583,
      "grad_norm": 20.314727783203125,
      "learning_rate": 3.6218735391824545e-06,
      "loss": 3.3579,
      "step": 34655
    },
    {
      "epoch": 0.3525797526041667,
      "grad_norm": 14.851888656616211,
      "learning_rate": 3.621516168175836e-06,
      "loss": 3.0346,
      "step": 34660
    },
    {
      "epoch": 0.352630615234375,
      "grad_norm": 11.343681335449219,
      "learning_rate": 3.6211587684754914e-06,
      "loss": 3.3709,
      "step": 34665
    },
    {
      "epoch": 0.3526814778645833,
      "grad_norm": 13.536211967468262,
      "learning_rate": 3.6208013400905664e-06,
      "loss": 3.5392,
      "step": 34670
    },
    {
      "epoch": 0.3527323404947917,
      "grad_norm": 10.644675254821777,
      "learning_rate": 3.620443883030203e-06,
      "loss": 3.5664,
      "step": 34675
    },
    {
      "epoch": 0.352783203125,
      "grad_norm": 13.698735237121582,
      "learning_rate": 3.62008639730355e-06,
      "loss": 3.1855,
      "step": 34680
    },
    {
      "epoch": 0.3528340657552083,
      "grad_norm": 11.285604476928711,
      "learning_rate": 3.6197288829197514e-06,
      "loss": 3.2029,
      "step": 34685
    },
    {
      "epoch": 0.3528849283854167,
      "grad_norm": 10.606297492980957,
      "learning_rate": 3.6193713398879547e-06,
      "loss": 3.5998,
      "step": 34690
    },
    {
      "epoch": 0.352935791015625,
      "grad_norm": 8.037817001342773,
      "learning_rate": 3.6190137682173075e-06,
      "loss": 3.5124,
      "step": 34695
    },
    {
      "epoch": 0.3529866536458333,
      "grad_norm": 16.26264762878418,
      "learning_rate": 3.6186561679169586e-06,
      "loss": 3.193,
      "step": 34700
    },
    {
      "epoch": 0.3530375162760417,
      "grad_norm": 8.826468467712402,
      "learning_rate": 3.6182985389960563e-06,
      "loss": 3.5367,
      "step": 34705
    },
    {
      "epoch": 0.35308837890625,
      "grad_norm": 12.868377685546875,
      "learning_rate": 3.617940881463751e-06,
      "loss": 3.3782,
      "step": 34710
    },
    {
      "epoch": 0.3531392415364583,
      "grad_norm": 9.595586776733398,
      "learning_rate": 3.617583195329194e-06,
      "loss": 3.557,
      "step": 34715
    },
    {
      "epoch": 0.3531901041666667,
      "grad_norm": 8.487686157226562,
      "learning_rate": 3.617225480601535e-06,
      "loss": 3.2387,
      "step": 34720
    },
    {
      "epoch": 0.353240966796875,
      "grad_norm": 15.628540992736816,
      "learning_rate": 3.6168677372899273e-06,
      "loss": 3.1546,
      "step": 34725
    },
    {
      "epoch": 0.3532918294270833,
      "grad_norm": 13.827096939086914,
      "learning_rate": 3.6165099654035227e-06,
      "loss": 3.466,
      "step": 34730
    },
    {
      "epoch": 0.3533426920572917,
      "grad_norm": 10.873969078063965,
      "learning_rate": 3.616152164951477e-06,
      "loss": 3.5061,
      "step": 34735
    },
    {
      "epoch": 0.3533935546875,
      "grad_norm": 14.796255111694336,
      "learning_rate": 3.615794335942941e-06,
      "loss": 3.7256,
      "step": 34740
    },
    {
      "epoch": 0.3534444173177083,
      "grad_norm": 7.992857933044434,
      "learning_rate": 3.615436478387072e-06,
      "loss": 3.1234,
      "step": 34745
    },
    {
      "epoch": 0.3534952799479167,
      "grad_norm": 8.480143547058105,
      "learning_rate": 3.6150785922930253e-06,
      "loss": 3.0514,
      "step": 34750
    },
    {
      "epoch": 0.353546142578125,
      "grad_norm": 15.358332633972168,
      "learning_rate": 3.6147206776699567e-06,
      "loss": 3.5486,
      "step": 34755
    },
    {
      "epoch": 0.3535970052083333,
      "grad_norm": 12.503549575805664,
      "learning_rate": 3.6143627345270237e-06,
      "loss": 3.1947,
      "step": 34760
    },
    {
      "epoch": 0.3536478678385417,
      "grad_norm": 14.133999824523926,
      "learning_rate": 3.6140047628733843e-06,
      "loss": 3.4198,
      "step": 34765
    },
    {
      "epoch": 0.35369873046875,
      "grad_norm": 13.71657657623291,
      "learning_rate": 3.613646762718197e-06,
      "loss": 3.4467,
      "step": 34770
    },
    {
      "epoch": 0.3537495930989583,
      "grad_norm": 12.718137741088867,
      "learning_rate": 3.6132887340706224e-06,
      "loss": 3.5934,
      "step": 34775
    },
    {
      "epoch": 0.3538004557291667,
      "grad_norm": 11.0115327835083,
      "learning_rate": 3.6129306769398174e-06,
      "loss": 3.4235,
      "step": 34780
    },
    {
      "epoch": 0.353851318359375,
      "grad_norm": 8.977180480957031,
      "learning_rate": 3.612572591334947e-06,
      "loss": 3.4629,
      "step": 34785
    },
    {
      "epoch": 0.3539021809895833,
      "grad_norm": 11.48985481262207,
      "learning_rate": 3.612214477265169e-06,
      "loss": 3.433,
      "step": 34790
    },
    {
      "epoch": 0.3539530436197917,
      "grad_norm": 9.702241897583008,
      "learning_rate": 3.611856334739648e-06,
      "loss": 3.5835,
      "step": 34795
    },
    {
      "epoch": 0.35400390625,
      "grad_norm": 15.769079208374023,
      "learning_rate": 3.6114981637675455e-06,
      "loss": 3.4196,
      "step": 34800
    },
    {
      "epoch": 0.3540547688802083,
      "grad_norm": 8.449577331542969,
      "learning_rate": 3.6111399643580266e-06,
      "loss": 3.2795,
      "step": 34805
    },
    {
      "epoch": 0.3541056315104167,
      "grad_norm": 14.473113059997559,
      "learning_rate": 3.6107817365202545e-06,
      "loss": 3.4543,
      "step": 34810
    },
    {
      "epoch": 0.354156494140625,
      "grad_norm": 13.105792045593262,
      "learning_rate": 3.6104234802633953e-06,
      "loss": 3.0334,
      "step": 34815
    },
    {
      "epoch": 0.3542073567708333,
      "grad_norm": 13.570204734802246,
      "learning_rate": 3.610065195596615e-06,
      "loss": 3.5513,
      "step": 34820
    },
    {
      "epoch": 0.3542582194010417,
      "grad_norm": 15.032855987548828,
      "learning_rate": 3.6097068825290793e-06,
      "loss": 3.3926,
      "step": 34825
    },
    {
      "epoch": 0.35430908203125,
      "grad_norm": 13.967044830322266,
      "learning_rate": 3.6093485410699557e-06,
      "loss": 3.1879,
      "step": 34830
    },
    {
      "epoch": 0.3543599446614583,
      "grad_norm": 13.504987716674805,
      "learning_rate": 3.608990171228413e-06,
      "loss": 3.6169,
      "step": 34835
    },
    {
      "epoch": 0.3544108072916667,
      "grad_norm": 14.612581253051758,
      "learning_rate": 3.60863177301362e-06,
      "loss": 3.257,
      "step": 34840
    },
    {
      "epoch": 0.354461669921875,
      "grad_norm": 14.804563522338867,
      "learning_rate": 3.608273346434745e-06,
      "loss": 3.4849,
      "step": 34845
    },
    {
      "epoch": 0.3545125325520833,
      "grad_norm": 13.429917335510254,
      "learning_rate": 3.60791489150096e-06,
      "loss": 3.4647,
      "step": 34850
    },
    {
      "epoch": 0.3545633951822917,
      "grad_norm": 11.944478034973145,
      "learning_rate": 3.607556408221435e-06,
      "loss": 3.3186,
      "step": 34855
    },
    {
      "epoch": 0.3546142578125,
      "grad_norm": 8.366698265075684,
      "learning_rate": 3.607197896605341e-06,
      "loss": 3.1635,
      "step": 34860
    },
    {
      "epoch": 0.3546651204427083,
      "grad_norm": 11.39499568939209,
      "learning_rate": 3.606839356661852e-06,
      "loss": 3.3217,
      "step": 34865
    },
    {
      "epoch": 0.3547159830729167,
      "grad_norm": 13.886192321777344,
      "learning_rate": 3.6064807884001404e-06,
      "loss": 3.5587,
      "step": 34870
    },
    {
      "epoch": 0.354766845703125,
      "grad_norm": 16.516136169433594,
      "learning_rate": 3.6061221918293797e-06,
      "loss": 3.5913,
      "step": 34875
    },
    {
      "epoch": 0.3548177083333333,
      "grad_norm": 10.966737747192383,
      "learning_rate": 3.6057635669587448e-06,
      "loss": 3.3713,
      "step": 34880
    },
    {
      "epoch": 0.3548685709635417,
      "grad_norm": 10.386698722839355,
      "learning_rate": 3.6054049137974114e-06,
      "loss": 3.091,
      "step": 34885
    },
    {
      "epoch": 0.35491943359375,
      "grad_norm": 8.499550819396973,
      "learning_rate": 3.605046232354556e-06,
      "loss": 3.4826,
      "step": 34890
    },
    {
      "epoch": 0.3549702962239583,
      "grad_norm": 11.241985321044922,
      "learning_rate": 3.604687522639353e-06,
      "loss": 3.2215,
      "step": 34895
    },
    {
      "epoch": 0.3550211588541667,
      "grad_norm": 12.003620147705078,
      "learning_rate": 3.604328784660983e-06,
      "loss": 3.7037,
      "step": 34900
    },
    {
      "epoch": 0.355072021484375,
      "grad_norm": 10.417730331420898,
      "learning_rate": 3.6039700184286224e-06,
      "loss": 3.714,
      "step": 34905
    },
    {
      "epoch": 0.3551228841145833,
      "grad_norm": 12.439254760742188,
      "learning_rate": 3.60361122395145e-06,
      "loss": 3.3445,
      "step": 34910
    },
    {
      "epoch": 0.3551737467447917,
      "grad_norm": 11.346061706542969,
      "learning_rate": 3.603252401238647e-06,
      "loss": 3.537,
      "step": 34915
    },
    {
      "epoch": 0.355224609375,
      "grad_norm": 12.715902328491211,
      "learning_rate": 3.6028935502993922e-06,
      "loss": 3.2552,
      "step": 34920
    },
    {
      "epoch": 0.3552754720052083,
      "grad_norm": 16.785329818725586,
      "learning_rate": 3.6025346711428677e-06,
      "loss": 3.4534,
      "step": 34925
    },
    {
      "epoch": 0.3553263346354167,
      "grad_norm": 9.780241012573242,
      "learning_rate": 3.6021757637782546e-06,
      "loss": 3.5138,
      "step": 34930
    },
    {
      "epoch": 0.355377197265625,
      "grad_norm": 12.479494094848633,
      "learning_rate": 3.601816828214736e-06,
      "loss": 3.2663,
      "step": 34935
    },
    {
      "epoch": 0.3554280598958333,
      "grad_norm": 13.093775749206543,
      "learning_rate": 3.601457864461495e-06,
      "loss": 3.3687,
      "step": 34940
    },
    {
      "epoch": 0.3554789225260417,
      "grad_norm": 15.283235549926758,
      "learning_rate": 3.6010988725277154e-06,
      "loss": 3.5688,
      "step": 34945
    },
    {
      "epoch": 0.35552978515625,
      "grad_norm": 17.717275619506836,
      "learning_rate": 3.6007398524225824e-06,
      "loss": 3.4344,
      "step": 34950
    },
    {
      "epoch": 0.3555806477864583,
      "grad_norm": 13.317954063415527,
      "learning_rate": 3.600380804155281e-06,
      "loss": 3.5325,
      "step": 34955
    },
    {
      "epoch": 0.3556315104166667,
      "grad_norm": 13.02704906463623,
      "learning_rate": 3.6000217277349976e-06,
      "loss": 3.2244,
      "step": 34960
    },
    {
      "epoch": 0.355682373046875,
      "grad_norm": 8.38286304473877,
      "learning_rate": 3.599662623170919e-06,
      "loss": 3.4214,
      "step": 34965
    },
    {
      "epoch": 0.3557332356770833,
      "grad_norm": 9.83238697052002,
      "learning_rate": 3.5993034904722333e-06,
      "loss": 3.072,
      "step": 34970
    },
    {
      "epoch": 0.3557840983072917,
      "grad_norm": 12.883691787719727,
      "learning_rate": 3.598944329648128e-06,
      "loss": 3.3078,
      "step": 34975
    },
    {
      "epoch": 0.3558349609375,
      "grad_norm": 9.63593578338623,
      "learning_rate": 3.598585140707792e-06,
      "loss": 3.6575,
      "step": 34980
    },
    {
      "epoch": 0.3558858235677083,
      "grad_norm": 14.492904663085938,
      "learning_rate": 3.598225923660416e-06,
      "loss": 3.0748,
      "step": 34985
    },
    {
      "epoch": 0.3559366861979167,
      "grad_norm": 13.749297142028809,
      "learning_rate": 3.5978666785151904e-06,
      "loss": 2.9942,
      "step": 34990
    },
    {
      "epoch": 0.355987548828125,
      "grad_norm": 14.713109970092773,
      "learning_rate": 3.5975074052813064e-06,
      "loss": 3.1924,
      "step": 34995
    },
    {
      "epoch": 0.3560384114583333,
      "grad_norm": 10.480639457702637,
      "learning_rate": 3.597148103967954e-06,
      "loss": 3.5809,
      "step": 35000
    },
    {
      "epoch": 0.3560892740885417,
      "grad_norm": 14.827054977416992,
      "learning_rate": 3.5967887745843284e-06,
      "loss": 5.4383,
      "step": 35005
    },
    {
      "epoch": 0.35614013671875,
      "grad_norm": 15.557093620300293,
      "learning_rate": 3.5964294171396227e-06,
      "loss": 3.301,
      "step": 35010
    },
    {
      "epoch": 0.3561909993489583,
      "grad_norm": 8.71878433227539,
      "learning_rate": 3.5960700316430284e-06,
      "loss": 3.3072,
      "step": 35015
    },
    {
      "epoch": 0.3562418619791667,
      "grad_norm": 8.105778694152832,
      "learning_rate": 3.5957106181037437e-06,
      "loss": 2.8905,
      "step": 35020
    },
    {
      "epoch": 0.356292724609375,
      "grad_norm": 13.239446640014648,
      "learning_rate": 3.595351176530961e-06,
      "loss": 3.4834,
      "step": 35025
    },
    {
      "epoch": 0.3563435872395833,
      "grad_norm": 12.645708084106445,
      "learning_rate": 3.59499170693388e-06,
      "loss": 3.3948,
      "step": 35030
    },
    {
      "epoch": 0.3563944498697917,
      "grad_norm": 11.654731750488281,
      "learning_rate": 3.5946322093216943e-06,
      "loss": 3.103,
      "step": 35035
    },
    {
      "epoch": 0.3564453125,
      "grad_norm": 10.604202270507812,
      "learning_rate": 3.594272683703604e-06,
      "loss": 3.252,
      "step": 35040
    },
    {
      "epoch": 0.3564961751302083,
      "grad_norm": 10.206571578979492,
      "learning_rate": 3.593913130088806e-06,
      "loss": 3.9125,
      "step": 35045
    },
    {
      "epoch": 0.3565470377604167,
      "grad_norm": 15.597684860229492,
      "learning_rate": 3.5935535484864996e-06,
      "loss": 3.2234,
      "step": 35050
    },
    {
      "epoch": 0.356597900390625,
      "grad_norm": 18.20827865600586,
      "learning_rate": 3.5931939389058852e-06,
      "loss": 3.2449,
      "step": 35055
    },
    {
      "epoch": 0.3566487630208333,
      "grad_norm": 14.010954856872559,
      "learning_rate": 3.5928343013561633e-06,
      "loss": 3.3907,
      "step": 35060
    },
    {
      "epoch": 0.3566996256510417,
      "grad_norm": 11.052334785461426,
      "learning_rate": 3.5924746358465335e-06,
      "loss": 3.5313,
      "step": 35065
    },
    {
      "epoch": 0.35675048828125,
      "grad_norm": 12.661726951599121,
      "learning_rate": 3.5921149423862005e-06,
      "loss": 3.582,
      "step": 35070
    },
    {
      "epoch": 0.3568013509114583,
      "grad_norm": 12.107625007629395,
      "learning_rate": 3.591755220984365e-06,
      "loss": 3.2842,
      "step": 35075
    },
    {
      "epoch": 0.3568522135416667,
      "grad_norm": 12.133563041687012,
      "learning_rate": 3.591395471650231e-06,
      "loss": 2.9995,
      "step": 35080
    },
    {
      "epoch": 0.356903076171875,
      "grad_norm": 13.665731430053711,
      "learning_rate": 3.5910356943930026e-06,
      "loss": 3.2549,
      "step": 35085
    },
    {
      "epoch": 0.3569539388020833,
      "grad_norm": 11.094895362854004,
      "learning_rate": 3.5906758892218844e-06,
      "loss": 3.2817,
      "step": 35090
    },
    {
      "epoch": 0.3570048014322917,
      "grad_norm": 11.758183479309082,
      "learning_rate": 3.5903160561460822e-06,
      "loss": 3.3395,
      "step": 35095
    },
    {
      "epoch": 0.3570556640625,
      "grad_norm": 11.913673400878906,
      "learning_rate": 3.5899561951748026e-06,
      "loss": 2.9786,
      "step": 35100
    },
    {
      "epoch": 0.3571065266927083,
      "grad_norm": 10.955697059631348,
      "learning_rate": 3.5895963063172514e-06,
      "loss": 3.8209,
      "step": 35105
    },
    {
      "epoch": 0.3571573893229167,
      "grad_norm": 12.346541404724121,
      "learning_rate": 3.5892363895826367e-06,
      "loss": 3.82,
      "step": 35110
    },
    {
      "epoch": 0.357208251953125,
      "grad_norm": 10.143900871276855,
      "learning_rate": 3.5888764449801676e-06,
      "loss": 3.3187,
      "step": 35115
    },
    {
      "epoch": 0.3572591145833333,
      "grad_norm": 10.030485153198242,
      "learning_rate": 3.588516472519053e-06,
      "loss": 3.4764,
      "step": 35120
    },
    {
      "epoch": 0.3573099772135417,
      "grad_norm": 14.733952522277832,
      "learning_rate": 3.5881564722085025e-06,
      "loss": 3.3969,
      "step": 35125
    },
    {
      "epoch": 0.35736083984375,
      "grad_norm": 14.487006187438965,
      "learning_rate": 3.5877964440577257e-06,
      "loss": 3.3901,
      "step": 35130
    },
    {
      "epoch": 0.3574117024739583,
      "grad_norm": 14.437053680419922,
      "learning_rate": 3.5874363880759357e-06,
      "loss": 3.0773,
      "step": 35135
    },
    {
      "epoch": 0.3574625651041667,
      "grad_norm": 8.986260414123535,
      "learning_rate": 3.5870763042723435e-06,
      "loss": 3.7938,
      "step": 35140
    },
    {
      "epoch": 0.357513427734375,
      "grad_norm": 13.2808256149292,
      "learning_rate": 3.586716192656161e-06,
      "loss": 3.2321,
      "step": 35145
    },
    {
      "epoch": 0.3575642903645833,
      "grad_norm": 11.628425598144531,
      "learning_rate": 3.586356053236603e-06,
      "loss": 3.4993,
      "step": 35150
    },
    {
      "epoch": 0.3576151529947917,
      "grad_norm": 13.323860168457031,
      "learning_rate": 3.5859958860228815e-06,
      "loss": 3.4119,
      "step": 35155
    },
    {
      "epoch": 0.357666015625,
      "grad_norm": 8.600274085998535,
      "learning_rate": 3.585635691024214e-06,
      "loss": 3.1944,
      "step": 35160
    },
    {
      "epoch": 0.3577168782552083,
      "grad_norm": 12.005955696105957,
      "learning_rate": 3.5852754682498147e-06,
      "loss": 3.3114,
      "step": 35165
    },
    {
      "epoch": 0.3577677408854167,
      "grad_norm": 11.952313423156738,
      "learning_rate": 3.5849152177088986e-06,
      "loss": 3.3038,
      "step": 35170
    },
    {
      "epoch": 0.357818603515625,
      "grad_norm": 15.262046813964844,
      "learning_rate": 3.5845549394106853e-06,
      "loss": 3.3386,
      "step": 35175
    },
    {
      "epoch": 0.3578694661458333,
      "grad_norm": 9.86393928527832,
      "learning_rate": 3.5841946333643896e-06,
      "loss": 3.655,
      "step": 35180
    },
    {
      "epoch": 0.3579203287760417,
      "grad_norm": 10.560075759887695,
      "learning_rate": 3.5838342995792324e-06,
      "loss": 3.1984,
      "step": 35185
    },
    {
      "epoch": 0.35797119140625,
      "grad_norm": 10.773649215698242,
      "learning_rate": 3.583473938064431e-06,
      "loss": 3.0001,
      "step": 35190
    },
    {
      "epoch": 0.3580220540364583,
      "grad_norm": 18.054447174072266,
      "learning_rate": 3.5831135488292055e-06,
      "loss": 3.3267,
      "step": 35195
    },
    {
      "epoch": 0.3580729166666667,
      "grad_norm": 14.0558500289917,
      "learning_rate": 3.582753131882777e-06,
      "loss": 3.0825,
      "step": 35200
    },
    {
      "epoch": 0.358123779296875,
      "grad_norm": 13.016155242919922,
      "learning_rate": 3.582392687234366e-06,
      "loss": 3.5745,
      "step": 35205
    },
    {
      "epoch": 0.3581746419270833,
      "grad_norm": 12.913225173950195,
      "learning_rate": 3.5820322148931945e-06,
      "loss": 3.4292,
      "step": 35210
    },
    {
      "epoch": 0.3582255045572917,
      "grad_norm": 9.55213737487793,
      "learning_rate": 3.5816717148684854e-06,
      "loss": 3.482,
      "step": 35215
    },
    {
      "epoch": 0.3582763671875,
      "grad_norm": 16.18498992919922,
      "learning_rate": 3.5813111871694616e-06,
      "loss": 3.4156,
      "step": 35220
    },
    {
      "epoch": 0.3583272298177083,
      "grad_norm": 9.790040969848633,
      "learning_rate": 3.5809506318053478e-06,
      "loss": 3.1728,
      "step": 35225
    },
    {
      "epoch": 0.3583780924479167,
      "grad_norm": 16.061195373535156,
      "learning_rate": 3.5805900487853685e-06,
      "loss": 3.1453,
      "step": 35230
    },
    {
      "epoch": 0.358428955078125,
      "grad_norm": 14.159446716308594,
      "learning_rate": 3.580229438118748e-06,
      "loss": 3.278,
      "step": 35235
    },
    {
      "epoch": 0.3584798177083333,
      "grad_norm": 14.750076293945312,
      "learning_rate": 3.5798687998147146e-06,
      "loss": 3.1484,
      "step": 35240
    },
    {
      "epoch": 0.3585306803385417,
      "grad_norm": 14.17802906036377,
      "learning_rate": 3.579508133882493e-06,
      "loss": 3.1719,
      "step": 35245
    },
    {
      "epoch": 0.35858154296875,
      "grad_norm": 13.584555625915527,
      "learning_rate": 3.579147440331312e-06,
      "loss": 3.2081,
      "step": 35250
    },
    {
      "epoch": 0.3586324055989583,
      "grad_norm": 10.397396087646484,
      "learning_rate": 3.5787867191703995e-06,
      "loss": 3.1229,
      "step": 35255
    },
    {
      "epoch": 0.3586832682291667,
      "grad_norm": 13.019648551940918,
      "learning_rate": 3.5784259704089842e-06,
      "loss": 3.3473,
      "step": 35260
    },
    {
      "epoch": 0.358734130859375,
      "grad_norm": 8.410650253295898,
      "learning_rate": 3.5780651940562966e-06,
      "loss": 3.0992,
      "step": 35265
    },
    {
      "epoch": 0.3587849934895833,
      "grad_norm": 10.615445137023926,
      "learning_rate": 3.577704390121566e-06,
      "loss": 3.5595,
      "step": 35270
    },
    {
      "epoch": 0.3588358561197917,
      "grad_norm": 9.943309783935547,
      "learning_rate": 3.577343558614024e-06,
      "loss": 3.3618,
      "step": 35275
    },
    {
      "epoch": 0.35888671875,
      "grad_norm": 14.800674438476562,
      "learning_rate": 3.5769826995429035e-06,
      "loss": 3.169,
      "step": 35280
    },
    {
      "epoch": 0.3589375813802083,
      "grad_norm": 10.84054183959961,
      "learning_rate": 3.5766218129174346e-06,
      "loss": 3.1432,
      "step": 35285
    },
    {
      "epoch": 0.3589884440104167,
      "grad_norm": 13.17013931274414,
      "learning_rate": 3.576260898746853e-06,
      "loss": 3.1751,
      "step": 35290
    },
    {
      "epoch": 0.359039306640625,
      "grad_norm": 13.68505573272705,
      "learning_rate": 3.57589995704039e-06,
      "loss": 3.4098,
      "step": 35295
    },
    {
      "epoch": 0.3590901692708333,
      "grad_norm": 9.305538177490234,
      "learning_rate": 3.575538987807282e-06,
      "loss": 3.655,
      "step": 35300
    },
    {
      "epoch": 0.3591410319010417,
      "grad_norm": 11.81442928314209,
      "learning_rate": 3.5751779910567647e-06,
      "loss": 3.0263,
      "step": 35305
    },
    {
      "epoch": 0.35919189453125,
      "grad_norm": 12.021374702453613,
      "learning_rate": 3.5748169667980727e-06,
      "loss": 3.1293,
      "step": 35310
    },
    {
      "epoch": 0.3592427571614583,
      "grad_norm": 13.424615859985352,
      "learning_rate": 3.5744559150404434e-06,
      "loss": 3.644,
      "step": 35315
    },
    {
      "epoch": 0.3592936197916667,
      "grad_norm": 17.6203670501709,
      "learning_rate": 3.574094835793115e-06,
      "loss": 3.3415,
      "step": 35320
    },
    {
      "epoch": 0.359344482421875,
      "grad_norm": 10.047724723815918,
      "learning_rate": 3.573733729065324e-06,
      "loss": 3.47,
      "step": 35325
    },
    {
      "epoch": 0.3593953450520833,
      "grad_norm": 14.883319854736328,
      "learning_rate": 3.57337259486631e-06,
      "loss": 3.4704,
      "step": 35330
    },
    {
      "epoch": 0.3594462076822917,
      "grad_norm": 9.46360969543457,
      "learning_rate": 3.573011433205312e-06,
      "loss": 3.1698,
      "step": 35335
    },
    {
      "epoch": 0.3594970703125,
      "grad_norm": 14.531940460205078,
      "learning_rate": 3.5726502440915724e-06,
      "loss": 3.2961,
      "step": 35340
    },
    {
      "epoch": 0.3595479329427083,
      "grad_norm": 10.947972297668457,
      "learning_rate": 3.57228902753433e-06,
      "loss": 3.3002,
      "step": 35345
    },
    {
      "epoch": 0.3595987955729167,
      "grad_norm": 15.049712181091309,
      "learning_rate": 3.5719277835428263e-06,
      "loss": 3.3406,
      "step": 35350
    },
    {
      "epoch": 0.359649658203125,
      "grad_norm": 8.383793830871582,
      "learning_rate": 3.5715665121263048e-06,
      "loss": 3.1684,
      "step": 35355
    },
    {
      "epoch": 0.3597005208333333,
      "grad_norm": 7.295913219451904,
      "learning_rate": 3.5712052132940085e-06,
      "loss": 3.0662,
      "step": 35360
    },
    {
      "epoch": 0.3597513834635417,
      "grad_norm": 11.55411434173584,
      "learning_rate": 3.5708438870551798e-06,
      "loss": 2.963,
      "step": 35365
    },
    {
      "epoch": 0.35980224609375,
      "grad_norm": 9.769346237182617,
      "learning_rate": 3.5704825334190645e-06,
      "loss": 3.092,
      "step": 35370
    },
    {
      "epoch": 0.3598531087239583,
      "grad_norm": 14.446013450622559,
      "learning_rate": 3.5701211523949075e-06,
      "loss": 3.2916,
      "step": 35375
    },
    {
      "epoch": 0.3599039713541667,
      "grad_norm": 15.183908462524414,
      "learning_rate": 3.5697597439919542e-06,
      "loss": 3.4919,
      "step": 35380
    },
    {
      "epoch": 0.359954833984375,
      "grad_norm": 11.906778335571289,
      "learning_rate": 3.5693983082194517e-06,
      "loss": 3.0823,
      "step": 35385
    },
    {
      "epoch": 0.3600056966145833,
      "grad_norm": 10.959227561950684,
      "learning_rate": 3.5690368450866466e-06,
      "loss": 3.2799,
      "step": 35390
    },
    {
      "epoch": 0.3600565592447917,
      "grad_norm": 9.536362648010254,
      "learning_rate": 3.5686753546027876e-06,
      "loss": 3.4638,
      "step": 35395
    },
    {
      "epoch": 0.360107421875,
      "grad_norm": 15.469267845153809,
      "learning_rate": 3.568313836777123e-06,
      "loss": 3.2376,
      "step": 35400
    },
    {
      "epoch": 0.3601582845052083,
      "grad_norm": 7.204216957092285,
      "learning_rate": 3.5679522916189014e-06,
      "loss": 3.501,
      "step": 35405
    },
    {
      "epoch": 0.3602091471354167,
      "grad_norm": 10.316786766052246,
      "learning_rate": 3.567590719137374e-06,
      "loss": 3.2892,
      "step": 35410
    },
    {
      "epoch": 0.360260009765625,
      "grad_norm": 14.605274200439453,
      "learning_rate": 3.5672291193417907e-06,
      "loss": 4.14,
      "step": 35415
    },
    {
      "epoch": 0.3603108723958333,
      "grad_norm": 13.843880653381348,
      "learning_rate": 3.5668674922414043e-06,
      "loss": 3.2044,
      "step": 35420
    },
    {
      "epoch": 0.3603617350260417,
      "grad_norm": 7.237750053405762,
      "learning_rate": 3.5665058378454655e-06,
      "loss": 3.8642,
      "step": 35425
    },
    {
      "epoch": 0.36041259765625,
      "grad_norm": 12.802541732788086,
      "learning_rate": 3.566144156163227e-06,
      "loss": 3.3252,
      "step": 35430
    },
    {
      "epoch": 0.3604634602864583,
      "grad_norm": 7.766964912414551,
      "learning_rate": 3.5657824472039437e-06,
      "loss": 3.6876,
      "step": 35435
    },
    {
      "epoch": 0.3605143229166667,
      "grad_norm": 16.223665237426758,
      "learning_rate": 3.565420710976869e-06,
      "loss": 3.157,
      "step": 35440
    },
    {
      "epoch": 0.360565185546875,
      "grad_norm": 14.279911994934082,
      "learning_rate": 3.5650589474912577e-06,
      "loss": 3.3762,
      "step": 35445
    },
    {
      "epoch": 0.3606160481770833,
      "grad_norm": 9.755158424377441,
      "learning_rate": 3.5646971567563663e-06,
      "loss": 3.6622,
      "step": 35450
    },
    {
      "epoch": 0.3606669108072917,
      "grad_norm": 10.42287540435791,
      "learning_rate": 3.5643353387814493e-06,
      "loss": 3.6163,
      "step": 35455
    },
    {
      "epoch": 0.3607177734375,
      "grad_norm": 13.92138671875,
      "learning_rate": 3.5639734935757664e-06,
      "loss": 3.197,
      "step": 35460
    },
    {
      "epoch": 0.3607686360677083,
      "grad_norm": 11.900999069213867,
      "learning_rate": 3.563611621148573e-06,
      "loss": 3.4892,
      "step": 35465
    },
    {
      "epoch": 0.3608194986979167,
      "grad_norm": 15.42886734008789,
      "learning_rate": 3.563249721509129e-06,
      "loss": 3.3542,
      "step": 35470
    },
    {
      "epoch": 0.360870361328125,
      "grad_norm": 14.31777286529541,
      "learning_rate": 3.562887794666692e-06,
      "loss": 3.5636,
      "step": 35475
    },
    {
      "epoch": 0.3609212239583333,
      "grad_norm": 14.096817016601562,
      "learning_rate": 3.562525840630523e-06,
      "loss": 3.1952,
      "step": 35480
    },
    {
      "epoch": 0.3609720865885417,
      "grad_norm": 12.328357696533203,
      "learning_rate": 3.5621638594098825e-06,
      "loss": 3.3161,
      "step": 35485
    },
    {
      "epoch": 0.36102294921875,
      "grad_norm": 7.772528648376465,
      "learning_rate": 3.5618018510140307e-06,
      "loss": 3.3572,
      "step": 35490
    },
    {
      "epoch": 0.3610738118489583,
      "grad_norm": 16.212244033813477,
      "learning_rate": 3.5614398154522313e-06,
      "loss": 3.2458,
      "step": 35495
    },
    {
      "epoch": 0.3611246744791667,
      "grad_norm": 11.501498222351074,
      "learning_rate": 3.5610777527337447e-06,
      "loss": 2.8993,
      "step": 35500
    },
    {
      "epoch": 0.361175537109375,
      "grad_norm": 10.752714157104492,
      "learning_rate": 3.5607156628678364e-06,
      "loss": 3.2463,
      "step": 35505
    },
    {
      "epoch": 0.3612263997395833,
      "grad_norm": 16.357065200805664,
      "learning_rate": 3.5603535458637684e-06,
      "loss": 3.2976,
      "step": 35510
    },
    {
      "epoch": 0.3612772623697917,
      "grad_norm": 9.25450325012207,
      "learning_rate": 3.5599914017308067e-06,
      "loss": 3.0116,
      "step": 35515
    },
    {
      "epoch": 0.361328125,
      "grad_norm": 12.183612823486328,
      "learning_rate": 3.559629230478216e-06,
      "loss": 3.4034,
      "step": 35520
    },
    {
      "epoch": 0.3613789876302083,
      "grad_norm": 10.711119651794434,
      "learning_rate": 3.559267032115263e-06,
      "loss": 3.1804,
      "step": 35525
    },
    {
      "epoch": 0.3614298502604167,
      "grad_norm": 10.811841011047363,
      "learning_rate": 3.558904806651214e-06,
      "loss": 3.385,
      "step": 35530
    },
    {
      "epoch": 0.361480712890625,
      "grad_norm": 11.532083511352539,
      "learning_rate": 3.558542554095336e-06,
      "loss": 3.3263,
      "step": 35535
    },
    {
      "epoch": 0.3615315755208333,
      "grad_norm": 15.292985916137695,
      "learning_rate": 3.5581802744568984e-06,
      "loss": 3.3674,
      "step": 35540
    },
    {
      "epoch": 0.3615824381510417,
      "grad_norm": 9.049236297607422,
      "learning_rate": 3.5578179677451685e-06,
      "loss": 3.825,
      "step": 35545
    },
    {
      "epoch": 0.36163330078125,
      "grad_norm": 12.779874801635742,
      "learning_rate": 3.5574556339694166e-06,
      "loss": 3.2203,
      "step": 35550
    },
    {
      "epoch": 0.3616841634114583,
      "grad_norm": 7.979409694671631,
      "learning_rate": 3.557093273138914e-06,
      "loss": 3.0655,
      "step": 35555
    },
    {
      "epoch": 0.3617350260416667,
      "grad_norm": 7.819886207580566,
      "learning_rate": 3.5567308852629295e-06,
      "loss": 3.2863,
      "step": 35560
    },
    {
      "epoch": 0.361785888671875,
      "grad_norm": 11.725508689880371,
      "learning_rate": 3.5563684703507366e-06,
      "loss": 3.6002,
      "step": 35565
    },
    {
      "epoch": 0.3618367513020833,
      "grad_norm": 11.572553634643555,
      "learning_rate": 3.556006028411606e-06,
      "loss": 3.2667,
      "step": 35570
    },
    {
      "epoch": 0.3618876139322917,
      "grad_norm": 11.396615028381348,
      "learning_rate": 3.5556435594548123e-06,
      "loss": 3.1082,
      "step": 35575
    },
    {
      "epoch": 0.3619384765625,
      "grad_norm": 13.791967391967773,
      "learning_rate": 3.5552810634896275e-06,
      "loss": 3.001,
      "step": 35580
    },
    {
      "epoch": 0.3619893391927083,
      "grad_norm": 11.551700592041016,
      "learning_rate": 3.5549185405253273e-06,
      "loss": 3.1784,
      "step": 35585
    },
    {
      "epoch": 0.3620402018229167,
      "grad_norm": 11.629484176635742,
      "learning_rate": 3.5545559905711857e-06,
      "loss": 3.2978,
      "step": 35590
    },
    {
      "epoch": 0.362091064453125,
      "grad_norm": 15.747698783874512,
      "learning_rate": 3.5541934136364795e-06,
      "loss": 3.2315,
      "step": 35595
    },
    {
      "epoch": 0.3621419270833333,
      "grad_norm": 11.128524780273438,
      "learning_rate": 3.5538308097304846e-06,
      "loss": 3.089,
      "step": 35600
    },
    {
      "epoch": 0.3621927897135417,
      "grad_norm": 11.763813018798828,
      "learning_rate": 3.5534681788624785e-06,
      "loss": 3.4707,
      "step": 35605
    },
    {
      "epoch": 0.36224365234375,
      "grad_norm": 13.028651237487793,
      "learning_rate": 3.5531055210417387e-06,
      "loss": 3.2072,
      "step": 35610
    },
    {
      "epoch": 0.3622945149739583,
      "grad_norm": 13.94176959991455,
      "learning_rate": 3.5527428362775433e-06,
      "loss": 3.2547,
      "step": 35615
    },
    {
      "epoch": 0.3623453776041667,
      "grad_norm": 8.804715156555176,
      "learning_rate": 3.5523801245791727e-06,
      "loss": 3.14,
      "step": 35620
    },
    {
      "epoch": 0.362396240234375,
      "grad_norm": 8.877320289611816,
      "learning_rate": 3.552017385955905e-06,
      "loss": 3.2741,
      "step": 35625
    },
    {
      "epoch": 0.3624471028645833,
      "grad_norm": 32.209754943847656,
      "learning_rate": 3.551654620417023e-06,
      "loss": 3.4959,
      "step": 35630
    },
    {
      "epoch": 0.3624979654947917,
      "grad_norm": 10.85776424407959,
      "learning_rate": 3.5512918279718054e-06,
      "loss": 3.2148,
      "step": 35635
    },
    {
      "epoch": 0.362548828125,
      "grad_norm": 11.025604248046875,
      "learning_rate": 3.5509290086295373e-06,
      "loss": 3.0602,
      "step": 35640
    },
    {
      "epoch": 0.3625996907552083,
      "grad_norm": 14.648173332214355,
      "learning_rate": 3.5505661623994987e-06,
      "loss": 3.3042,
      "step": 35645
    },
    {
      "epoch": 0.3626505533854167,
      "grad_norm": 8.177518844604492,
      "learning_rate": 3.550203289290974e-06,
      "loss": 3.5925,
      "step": 35650
    },
    {
      "epoch": 0.362701416015625,
      "grad_norm": 17.484180450439453,
      "learning_rate": 3.5498403893132465e-06,
      "loss": 3.1191,
      "step": 35655
    },
    {
      "epoch": 0.3627522786458333,
      "grad_norm": 12.216513633728027,
      "learning_rate": 3.5494774624756025e-06,
      "loss": 3.4687,
      "step": 35660
    },
    {
      "epoch": 0.3628031412760417,
      "grad_norm": 11.965054512023926,
      "learning_rate": 3.5491145087873253e-06,
      "loss": 3.1928,
      "step": 35665
    },
    {
      "epoch": 0.36285400390625,
      "grad_norm": 11.027030944824219,
      "learning_rate": 3.5487515282577027e-06,
      "loss": 3.4374,
      "step": 35670
    },
    {
      "epoch": 0.3629048665364583,
      "grad_norm": 12.054620742797852,
      "learning_rate": 3.5483885208960207e-06,
      "loss": 3.1135,
      "step": 35675
    },
    {
      "epoch": 0.3629557291666667,
      "grad_norm": 8.037942886352539,
      "learning_rate": 3.548025486711567e-06,
      "loss": 2.9982,
      "step": 35680
    },
    {
      "epoch": 0.363006591796875,
      "grad_norm": 9.676033973693848,
      "learning_rate": 3.5476624257136294e-06,
      "loss": 3.3297,
      "step": 35685
    },
    {
      "epoch": 0.3630574544270833,
      "grad_norm": 13.849072456359863,
      "learning_rate": 3.547299337911497e-06,
      "loss": 3.254,
      "step": 35690
    },
    {
      "epoch": 0.3631083170572917,
      "grad_norm": 13.027873992919922,
      "learning_rate": 3.5469362233144594e-06,
      "loss": 3.4734,
      "step": 35695
    },
    {
      "epoch": 0.3631591796875,
      "grad_norm": 13.047901153564453,
      "learning_rate": 3.5465730819318067e-06,
      "loss": 3.2893,
      "step": 35700
    },
    {
      "epoch": 0.3632100423177083,
      "grad_norm": 17.80720329284668,
      "learning_rate": 3.5462099137728294e-06,
      "loss": 3.3718,
      "step": 35705
    },
    {
      "epoch": 0.3632609049479167,
      "grad_norm": 9.708136558532715,
      "learning_rate": 3.5458467188468198e-06,
      "loss": 3.4184,
      "step": 35710
    },
    {
      "epoch": 0.363311767578125,
      "grad_norm": 16.222246170043945,
      "learning_rate": 3.54548349716307e-06,
      "loss": 3.3733,
      "step": 35715
    },
    {
      "epoch": 0.3633626302083333,
      "grad_norm": 12.150138854980469,
      "learning_rate": 3.545120248730872e-06,
      "loss": 3.3269,
      "step": 35720
    },
    {
      "epoch": 0.3634134928385417,
      "grad_norm": 13.112444877624512,
      "learning_rate": 3.5447569735595203e-06,
      "loss": 3.3737,
      "step": 35725
    },
    {
      "epoch": 0.36346435546875,
      "grad_norm": 16.09695816040039,
      "learning_rate": 3.5443936716583093e-06,
      "loss": 3.308,
      "step": 35730
    },
    {
      "epoch": 0.3635152180989583,
      "grad_norm": 14.034189224243164,
      "learning_rate": 3.5440303430365346e-06,
      "loss": 3.4142,
      "step": 35735
    },
    {
      "epoch": 0.3635660807291667,
      "grad_norm": 14.190086364746094,
      "learning_rate": 3.5436669877034903e-06,
      "loss": 3.5965,
      "step": 35740
    },
    {
      "epoch": 0.363616943359375,
      "grad_norm": 11.754796028137207,
      "learning_rate": 3.543303605668473e-06,
      "loss": 3.2113,
      "step": 35745
    },
    {
      "epoch": 0.3636678059895833,
      "grad_norm": 12.327696800231934,
      "learning_rate": 3.542940196940781e-06,
      "loss": 3.6533,
      "step": 35750
    },
    {
      "epoch": 0.3637186686197917,
      "grad_norm": 11.785676956176758,
      "learning_rate": 3.542576761529711e-06,
      "loss": 3.3057,
      "step": 35755
    },
    {
      "epoch": 0.36376953125,
      "grad_norm": 11.127363204956055,
      "learning_rate": 3.5422132994445617e-06,
      "loss": 3.2788,
      "step": 35760
    },
    {
      "epoch": 0.3638203938802083,
      "grad_norm": 14.203107833862305,
      "learning_rate": 3.541849810694632e-06,
      "loss": 3.0148,
      "step": 35765
    },
    {
      "epoch": 0.3638712565104167,
      "grad_norm": 9.956028938293457,
      "learning_rate": 3.541486295289222e-06,
      "loss": 3.6195,
      "step": 35770
    },
    {
      "epoch": 0.363922119140625,
      "grad_norm": 11.492575645446777,
      "learning_rate": 3.5411227532376323e-06,
      "loss": 3.2679,
      "step": 35775
    },
    {
      "epoch": 0.3639729817708333,
      "grad_norm": 7.851870536804199,
      "learning_rate": 3.540759184549163e-06,
      "loss": 3.0741,
      "step": 35780
    },
    {
      "epoch": 0.3640238444010417,
      "grad_norm": 13.882920265197754,
      "learning_rate": 3.540395589233117e-06,
      "loss": 3.476,
      "step": 35785
    },
    {
      "epoch": 0.36407470703125,
      "grad_norm": 7.528973579406738,
      "learning_rate": 3.540031967298796e-06,
      "loss": 3.3073,
      "step": 35790
    },
    {
      "epoch": 0.3641255696614583,
      "grad_norm": 7.640775680541992,
      "learning_rate": 3.539668318755504e-06,
      "loss": 3.2905,
      "step": 35795
    },
    {
      "epoch": 0.3641764322916667,
      "grad_norm": 10.673866271972656,
      "learning_rate": 3.5393046436125456e-06,
      "loss": 3.5734,
      "step": 35800
    },
    {
      "epoch": 0.364227294921875,
      "grad_norm": 13.123295783996582,
      "learning_rate": 3.5389409418792226e-06,
      "loss": 3.4012,
      "step": 35805
    },
    {
      "epoch": 0.3642781575520833,
      "grad_norm": 10.05083179473877,
      "learning_rate": 3.5385772135648428e-06,
      "loss": 3.2255,
      "step": 35810
    },
    {
      "epoch": 0.3643290201822917,
      "grad_norm": 9.180085182189941,
      "learning_rate": 3.5382134586787107e-06,
      "loss": 3.2786,
      "step": 35815
    },
    {
      "epoch": 0.3643798828125,
      "grad_norm": 10.960928916931152,
      "learning_rate": 3.5378496772301337e-06,
      "loss": 4.0712,
      "step": 35820
    },
    {
      "epoch": 0.3644307454427083,
      "grad_norm": 10.767245292663574,
      "learning_rate": 3.537485869228418e-06,
      "loss": 3.1526,
      "step": 35825
    },
    {
      "epoch": 0.3644816080729167,
      "grad_norm": 15.244921684265137,
      "learning_rate": 3.5371220346828725e-06,
      "loss": 3.1579,
      "step": 35830
    },
    {
      "epoch": 0.364532470703125,
      "grad_norm": 11.677315711975098,
      "learning_rate": 3.536758173602806e-06,
      "loss": 3.2464,
      "step": 35835
    },
    {
      "epoch": 0.3645833333333333,
      "grad_norm": 10.261982917785645,
      "learning_rate": 3.5363942859975275e-06,
      "loss": 3.2389,
      "step": 35840
    },
    {
      "epoch": 0.3646341959635417,
      "grad_norm": 10.926580429077148,
      "learning_rate": 3.536030371876346e-06,
      "loss": 3.6832,
      "step": 35845
    },
    {
      "epoch": 0.36468505859375,
      "grad_norm": 16.905780792236328,
      "learning_rate": 3.5356664312485743e-06,
      "loss": 3.5338,
      "step": 35850
    },
    {
      "epoch": 0.3647359212239583,
      "grad_norm": 9.730697631835938,
      "learning_rate": 3.535302464123522e-06,
      "loss": 3.182,
      "step": 35855
    },
    {
      "epoch": 0.3647867838541667,
      "grad_norm": 14.526654243469238,
      "learning_rate": 3.5349384705105006e-06,
      "loss": 3.247,
      "step": 35860
    },
    {
      "epoch": 0.364837646484375,
      "grad_norm": 15.030230522155762,
      "learning_rate": 3.5345744504188246e-06,
      "loss": 3.7499,
      "step": 35865
    },
    {
      "epoch": 0.3648885091145833,
      "grad_norm": 9.691508293151855,
      "learning_rate": 3.534210403857806e-06,
      "loss": 3.5489,
      "step": 35870
    },
    {
      "epoch": 0.3649393717447917,
      "grad_norm": 12.639229774475098,
      "learning_rate": 3.53384633083676e-06,
      "loss": 3.5291,
      "step": 35875
    },
    {
      "epoch": 0.364990234375,
      "grad_norm": 9.587438583374023,
      "learning_rate": 3.5334822313649998e-06,
      "loss": 3.1705,
      "step": 35880
    },
    {
      "epoch": 0.3650410970052083,
      "grad_norm": 13.8088960647583,
      "learning_rate": 3.533118105451842e-06,
      "loss": 3.8654,
      "step": 35885
    },
    {
      "epoch": 0.3650919596354167,
      "grad_norm": 15.760499000549316,
      "learning_rate": 3.532753953106603e-06,
      "loss": 3.3849,
      "step": 35890
    },
    {
      "epoch": 0.365142822265625,
      "grad_norm": 12.67516040802002,
      "learning_rate": 3.5323897743385983e-06,
      "loss": 3.447,
      "step": 35895
    },
    {
      "epoch": 0.3651936848958333,
      "grad_norm": 14.23686408996582,
      "learning_rate": 3.532025569157146e-06,
      "loss": 3.479,
      "step": 35900
    },
    {
      "epoch": 0.3652445475260417,
      "grad_norm": 14.446793556213379,
      "learning_rate": 3.531661337571564e-06,
      "loss": 3.3813,
      "step": 35905
    },
    {
      "epoch": 0.36529541015625,
      "grad_norm": 12.83717155456543,
      "learning_rate": 3.531297079591171e-06,
      "loss": 3.966,
      "step": 35910
    },
    {
      "epoch": 0.3653462727864583,
      "grad_norm": 8.24056625366211,
      "learning_rate": 3.5309327952252875e-06,
      "loss": 3.3361,
      "step": 35915
    },
    {
      "epoch": 0.3653971354166667,
      "grad_norm": 11.2170991897583,
      "learning_rate": 3.530568484483232e-06,
      "loss": 3.5179,
      "step": 35920
    },
    {
      "epoch": 0.365447998046875,
      "grad_norm": 12.390713691711426,
      "learning_rate": 3.5302041473743263e-06,
      "loss": 3.5603,
      "step": 35925
    },
    {
      "epoch": 0.3654988606770833,
      "grad_norm": 7.594927787780762,
      "learning_rate": 3.5298397839078923e-06,
      "loss": 3.2695,
      "step": 35930
    },
    {
      "epoch": 0.3655497233072917,
      "grad_norm": 15.015494346618652,
      "learning_rate": 3.529475394093251e-06,
      "loss": 3.2524,
      "step": 35935
    },
    {
      "epoch": 0.3656005859375,
      "grad_norm": 16.249225616455078,
      "learning_rate": 3.529110977939726e-06,
      "loss": 2.9257,
      "step": 35940
    },
    {
      "epoch": 0.3656514485677083,
      "grad_norm": 11.581992149353027,
      "learning_rate": 3.52874653545664e-06,
      "loss": 3.3796,
      "step": 35945
    },
    {
      "epoch": 0.3657023111979167,
      "grad_norm": 10.994409561157227,
      "learning_rate": 3.528382066653318e-06,
      "loss": 3.5111,
      "step": 35950
    },
    {
      "epoch": 0.365753173828125,
      "grad_norm": 12.003817558288574,
      "learning_rate": 3.528017571539085e-06,
      "loss": 3.6897,
      "step": 35955
    },
    {
      "epoch": 0.3658040364583333,
      "grad_norm": 9.575628280639648,
      "learning_rate": 3.527653050123265e-06,
      "loss": 3.7692,
      "step": 35960
    },
    {
      "epoch": 0.3658548990885417,
      "grad_norm": 9.068621635437012,
      "learning_rate": 3.5272885024151864e-06,
      "loss": 3.353,
      "step": 35965
    },
    {
      "epoch": 0.36590576171875,
      "grad_norm": 17.891878128051758,
      "learning_rate": 3.5269239284241747e-06,
      "loss": 3.2435,
      "step": 35970
    },
    {
      "epoch": 0.3659566243489583,
      "grad_norm": 11.083831787109375,
      "learning_rate": 3.526559328159558e-06,
      "loss": 3.449,
      "step": 35975
    },
    {
      "epoch": 0.3660074869791667,
      "grad_norm": 11.851883888244629,
      "learning_rate": 3.5261947016306634e-06,
      "loss": 3.6004,
      "step": 35980
    },
    {
      "epoch": 0.366058349609375,
      "grad_norm": 9.52242374420166,
      "learning_rate": 3.525830048846821e-06,
      "loss": 3.009,
      "step": 35985
    },
    {
      "epoch": 0.3661092122395833,
      "grad_norm": 10.086869239807129,
      "learning_rate": 3.52546536981736e-06,
      "loss": 3.0381,
      "step": 35990
    },
    {
      "epoch": 0.3661600748697917,
      "grad_norm": 15.746369361877441,
      "learning_rate": 3.5251006645516105e-06,
      "loss": 3.4968,
      "step": 35995
    },
    {
      "epoch": 0.3662109375,
      "grad_norm": 8.71462631225586,
      "learning_rate": 3.5247359330589032e-06,
      "loss": 3.6169,
      "step": 36000
    },
    {
      "epoch": 0.3662618001302083,
      "grad_norm": 16.58160400390625,
      "learning_rate": 3.5243711753485704e-06,
      "loss": 3.5387,
      "step": 36005
    },
    {
      "epoch": 0.3663126627604167,
      "grad_norm": 7.80565071105957,
      "learning_rate": 3.524006391429944e-06,
      "loss": 3.561,
      "step": 36010
    },
    {
      "epoch": 0.366363525390625,
      "grad_norm": 12.95910930633545,
      "learning_rate": 3.523641581312356e-06,
      "loss": 3.1926,
      "step": 36015
    },
    {
      "epoch": 0.3664143880208333,
      "grad_norm": 18.815176010131836,
      "learning_rate": 3.523276745005142e-06,
      "loss": 4.1247,
      "step": 36020
    },
    {
      "epoch": 0.3664652506510417,
      "grad_norm": 10.081045150756836,
      "learning_rate": 3.5229118825176336e-06,
      "loss": 3.4267,
      "step": 36025
    },
    {
      "epoch": 0.36651611328125,
      "grad_norm": 9.349115371704102,
      "learning_rate": 3.5225469938591682e-06,
      "loss": 3.5161,
      "step": 36030
    },
    {
      "epoch": 0.3665669759114583,
      "grad_norm": 12.156893730163574,
      "learning_rate": 3.52218207903908e-06,
      "loss": 3.151,
      "step": 36035
    },
    {
      "epoch": 0.3666178385416667,
      "grad_norm": 11.794998168945312,
      "learning_rate": 3.5218171380667056e-06,
      "loss": 3.6706,
      "step": 36040
    },
    {
      "epoch": 0.366668701171875,
      "grad_norm": 6.907103538513184,
      "learning_rate": 3.5214521709513815e-06,
      "loss": 3.1906,
      "step": 36045
    },
    {
      "epoch": 0.3667195638020833,
      "grad_norm": 8.981483459472656,
      "learning_rate": 3.5210871777024466e-06,
      "loss": 3.3278,
      "step": 36050
    },
    {
      "epoch": 0.3667704264322917,
      "grad_norm": 9.719010353088379,
      "learning_rate": 3.520722158329237e-06,
      "loss": 3.4919,
      "step": 36055
    },
    {
      "epoch": 0.3668212890625,
      "grad_norm": 13.264166831970215,
      "learning_rate": 3.5203571128410945e-06,
      "loss": 3.2168,
      "step": 36060
    },
    {
      "epoch": 0.3668721516927083,
      "grad_norm": 6.929196834564209,
      "learning_rate": 3.519992041247356e-06,
      "loss": 3.3111,
      "step": 36065
    },
    {
      "epoch": 0.3669230143229167,
      "grad_norm": 14.912343978881836,
      "learning_rate": 3.519626943557364e-06,
      "loss": 3.6861,
      "step": 36070
    },
    {
      "epoch": 0.366973876953125,
      "grad_norm": 11.711191177368164,
      "learning_rate": 3.5192618197804566e-06,
      "loss": 3.4719,
      "step": 36075
    },
    {
      "epoch": 0.3670247395833333,
      "grad_norm": 14.388081550598145,
      "learning_rate": 3.5188966699259785e-06,
      "loss": 3.6801,
      "step": 36080
    },
    {
      "epoch": 0.3670756022135417,
      "grad_norm": 13.36806869506836,
      "learning_rate": 3.51853149400327e-06,
      "loss": 3.3277,
      "step": 36085
    },
    {
      "epoch": 0.36712646484375,
      "grad_norm": 9.424427032470703,
      "learning_rate": 3.518166292021675e-06,
      "loss": 3.5005,
      "step": 36090
    },
    {
      "epoch": 0.3671773274739583,
      "grad_norm": 11.621115684509277,
      "learning_rate": 3.517801063990536e-06,
      "loss": 3.4082,
      "step": 36095
    },
    {
      "epoch": 0.3672281901041667,
      "grad_norm": 13.065186500549316,
      "learning_rate": 3.5174358099191986e-06,
      "loss": 3.0164,
      "step": 36100
    },
    {
      "epoch": 0.367279052734375,
      "grad_norm": 14.8209867477417,
      "learning_rate": 3.517070529817007e-06,
      "loss": 3.5625,
      "step": 36105
    },
    {
      "epoch": 0.3673299153645833,
      "grad_norm": 15.603599548339844,
      "learning_rate": 3.5167052236933076e-06,
      "loss": 3.2894,
      "step": 36110
    },
    {
      "epoch": 0.3673807779947917,
      "grad_norm": 13.455853462219238,
      "learning_rate": 3.516339891557445e-06,
      "loss": 3.205,
      "step": 36115
    },
    {
      "epoch": 0.367431640625,
      "grad_norm": 11.697784423828125,
      "learning_rate": 3.5159745334187677e-06,
      "loss": 3.3178,
      "step": 36120
    },
    {
      "epoch": 0.3674825032552083,
      "grad_norm": 14.168938636779785,
      "learning_rate": 3.5156091492866236e-06,
      "loss": 3.5605,
      "step": 36125
    },
    {
      "epoch": 0.3675333658854167,
      "grad_norm": 14.687520980834961,
      "learning_rate": 3.515243739170359e-06,
      "loss": 3.2411,
      "step": 36130
    },
    {
      "epoch": 0.367584228515625,
      "grad_norm": 14.495407104492188,
      "learning_rate": 3.5148783030793244e-06,
      "loss": 3.7752,
      "step": 36135
    },
    {
      "epoch": 0.3676350911458333,
      "grad_norm": 12.858672142028809,
      "learning_rate": 3.514512841022869e-06,
      "loss": 3.1062,
      "step": 36140
    },
    {
      "epoch": 0.3676859537760417,
      "grad_norm": 11.596325874328613,
      "learning_rate": 3.514147353010343e-06,
      "loss": 3.1808,
      "step": 36145
    },
    {
      "epoch": 0.36773681640625,
      "grad_norm": 11.595054626464844,
      "learning_rate": 3.513781839051097e-06,
      "loss": 3.3619,
      "step": 36150
    },
    {
      "epoch": 0.3677876790364583,
      "grad_norm": 8.274419784545898,
      "learning_rate": 3.5134162991544828e-06,
      "loss": 3.0254,
      "step": 36155
    },
    {
      "epoch": 0.3678385416666667,
      "grad_norm": 7.9185099601745605,
      "learning_rate": 3.5130507333298535e-06,
      "loss": 3.4626,
      "step": 36160
    },
    {
      "epoch": 0.367889404296875,
      "grad_norm": 9.295361518859863,
      "learning_rate": 3.5126851415865615e-06,
      "loss": 3.1589,
      "step": 36165
    },
    {
      "epoch": 0.3679402669270833,
      "grad_norm": 9.102407455444336,
      "learning_rate": 3.5123195239339593e-06,
      "loss": 3.3127,
      "step": 36170
    },
    {
      "epoch": 0.3679911295572917,
      "grad_norm": 12.359614372253418,
      "learning_rate": 3.511953880381403e-06,
      "loss": 3.3832,
      "step": 36175
    },
    {
      "epoch": 0.3680419921875,
      "grad_norm": 8.219758987426758,
      "learning_rate": 3.511588210938245e-06,
      "loss": 3.3632,
      "step": 36180
    },
    {
      "epoch": 0.3680928548177083,
      "grad_norm": 16.555068969726562,
      "learning_rate": 3.511222515613844e-06,
      "loss": 3.3802,
      "step": 36185
    },
    {
      "epoch": 0.3681437174479167,
      "grad_norm": 15.864944458007812,
      "learning_rate": 3.5108567944175537e-06,
      "loss": 3.4895,
      "step": 36190
    },
    {
      "epoch": 0.368194580078125,
      "grad_norm": 12.22673511505127,
      "learning_rate": 3.5104910473587315e-06,
      "loss": 3.0151,
      "step": 36195
    },
    {
      "epoch": 0.3682454427083333,
      "grad_norm": 8.683699607849121,
      "learning_rate": 3.5101252744467367e-06,
      "loss": 3.3945,
      "step": 36200
    },
    {
      "epoch": 0.3682963053385417,
      "grad_norm": 8.718663215637207,
      "learning_rate": 3.509759475690926e-06,
      "loss": 3.5381,
      "step": 36205
    },
    {
      "epoch": 0.36834716796875,
      "grad_norm": 10.717947006225586,
      "learning_rate": 3.5093936511006578e-06,
      "loss": 3.2846,
      "step": 36210
    },
    {
      "epoch": 0.3683980305989583,
      "grad_norm": 12.331782341003418,
      "learning_rate": 3.509027800685292e-06,
      "loss": 3.1712,
      "step": 36215
    },
    {
      "epoch": 0.3684488932291667,
      "grad_norm": 11.512397766113281,
      "learning_rate": 3.5086619244541896e-06,
      "loss": 3.4319,
      "step": 36220
    },
    {
      "epoch": 0.368499755859375,
      "grad_norm": 14.478857040405273,
      "learning_rate": 3.5082960224167107e-06,
      "loss": 2.9699,
      "step": 36225
    },
    {
      "epoch": 0.3685506184895833,
      "grad_norm": 12.586868286132812,
      "learning_rate": 3.5079300945822183e-06,
      "loss": 3.2637,
      "step": 36230
    },
    {
      "epoch": 0.3686014811197917,
      "grad_norm": 17.139347076416016,
      "learning_rate": 3.5075641409600712e-06,
      "loss": 3.1607,
      "step": 36235
    },
    {
      "epoch": 0.36865234375,
      "grad_norm": 14.53843879699707,
      "learning_rate": 3.5071981615596364e-06,
      "loss": 3.2149,
      "step": 36240
    },
    {
      "epoch": 0.3687032063802083,
      "grad_norm": 10.349335670471191,
      "learning_rate": 3.506832156390274e-06,
      "loss": 3.5323,
      "step": 36245
    },
    {
      "epoch": 0.3687540690104167,
      "grad_norm": 9.536778450012207,
      "learning_rate": 3.5064661254613497e-06,
      "loss": 3.1457,
      "step": 36250
    },
    {
      "epoch": 0.368804931640625,
      "grad_norm": 8.634055137634277,
      "learning_rate": 3.506100068782229e-06,
      "loss": 3.4729,
      "step": 36255
    },
    {
      "epoch": 0.3688557942708333,
      "grad_norm": 14.773881912231445,
      "learning_rate": 3.505733986362275e-06,
      "loss": 3.271,
      "step": 36260
    },
    {
      "epoch": 0.3689066569010417,
      "grad_norm": 15.675344467163086,
      "learning_rate": 3.505367878210857e-06,
      "loss": 3.2313,
      "step": 36265
    },
    {
      "epoch": 0.36895751953125,
      "grad_norm": 12.131467819213867,
      "learning_rate": 3.5050017443373392e-06,
      "loss": 3.429,
      "step": 36270
    },
    {
      "epoch": 0.3690083821614583,
      "grad_norm": 16.940414428710938,
      "learning_rate": 3.5046355847510897e-06,
      "loss": 3.3015,
      "step": 36275
    },
    {
      "epoch": 0.3690592447916667,
      "grad_norm": 12.1268310546875,
      "learning_rate": 3.504269399461478e-06,
      "loss": 3.0125,
      "step": 36280
    },
    {
      "epoch": 0.369110107421875,
      "grad_norm": 10.741314888000488,
      "learning_rate": 3.5039031884778707e-06,
      "loss": 3.0643,
      "step": 36285
    },
    {
      "epoch": 0.3691609700520833,
      "grad_norm": 17.02005386352539,
      "learning_rate": 3.5035369518096384e-06,
      "loss": 3.4271,
      "step": 36290
    },
    {
      "epoch": 0.3692118326822917,
      "grad_norm": 15.335061073303223,
      "learning_rate": 3.503170689466152e-06,
      "loss": 3.2227,
      "step": 36295
    },
    {
      "epoch": 0.3692626953125,
      "grad_norm": 8.688138008117676,
      "learning_rate": 3.50280440145678e-06,
      "loss": 3.0821,
      "step": 36300
    },
    {
      "epoch": 0.3693135579427083,
      "grad_norm": 11.173579216003418,
      "learning_rate": 3.5024380877908957e-06,
      "loss": 3.1695,
      "step": 36305
    },
    {
      "epoch": 0.3693644205729167,
      "grad_norm": 14.764756202697754,
      "learning_rate": 3.502071748477871e-06,
      "loss": 3.2324,
      "step": 36310
    },
    {
      "epoch": 0.369415283203125,
      "grad_norm": 16.455608367919922,
      "learning_rate": 3.5017053835270777e-06,
      "loss": 3.4476,
      "step": 36315
    },
    {
      "epoch": 0.3694661458333333,
      "grad_norm": 14.511435508728027,
      "learning_rate": 3.501338992947889e-06,
      "loss": 3.5649,
      "step": 36320
    },
    {
      "epoch": 0.3695170084635417,
      "grad_norm": 10.788686752319336,
      "learning_rate": 3.5009725767496806e-06,
      "loss": 3.3017,
      "step": 36325
    },
    {
      "epoch": 0.36956787109375,
      "grad_norm": 14.52601146697998,
      "learning_rate": 3.5006061349418253e-06,
      "loss": 3.6233,
      "step": 36330
    },
    {
      "epoch": 0.3696187337239583,
      "grad_norm": 10.335064888000488,
      "learning_rate": 3.5002396675336993e-06,
      "loss": 3.3017,
      "step": 36335
    },
    {
      "epoch": 0.3696695963541667,
      "grad_norm": 12.67693042755127,
      "learning_rate": 3.4998731745346786e-06,
      "loss": 3.4847,
      "step": 36340
    },
    {
      "epoch": 0.369720458984375,
      "grad_norm": 8.478259086608887,
      "learning_rate": 3.49950665595414e-06,
      "loss": 3.1835,
      "step": 36345
    },
    {
      "epoch": 0.3697713216145833,
      "grad_norm": 13.773451805114746,
      "learning_rate": 3.49914011180146e-06,
      "loss": 3.2737,
      "step": 36350
    },
    {
      "epoch": 0.3698221842447917,
      "grad_norm": 11.555225372314453,
      "learning_rate": 3.498773542086018e-06,
      "loss": 3.3203,
      "step": 36355
    },
    {
      "epoch": 0.369873046875,
      "grad_norm": 15.813162803649902,
      "learning_rate": 3.498406946817191e-06,
      "loss": 3.4779,
      "step": 36360
    },
    {
      "epoch": 0.3699239095052083,
      "grad_norm": 11.773980140686035,
      "learning_rate": 3.498040326004359e-06,
      "loss": 3.0863,
      "step": 36365
    },
    {
      "epoch": 0.3699747721354167,
      "grad_norm": 15.894186019897461,
      "learning_rate": 3.497673679656901e-06,
      "loss": 3.3233,
      "step": 36370
    },
    {
      "epoch": 0.370025634765625,
      "grad_norm": 10.541888236999512,
      "learning_rate": 3.4973070077841997e-06,
      "loss": 3.4923,
      "step": 36375
    },
    {
      "epoch": 0.3700764973958333,
      "grad_norm": 11.18358325958252,
      "learning_rate": 3.496940310395634e-06,
      "loss": 3.1329,
      "step": 36380
    },
    {
      "epoch": 0.3701273600260417,
      "grad_norm": 15.291526794433594,
      "learning_rate": 3.4965735875005873e-06,
      "loss": 3.4248,
      "step": 36385
    },
    {
      "epoch": 0.37017822265625,
      "grad_norm": 11.989048957824707,
      "learning_rate": 3.4962068391084414e-06,
      "loss": 3.3592,
      "step": 36390
    },
    {
      "epoch": 0.3702290852864583,
      "grad_norm": 10.62578010559082,
      "learning_rate": 3.4958400652285795e-06,
      "loss": 3.6871,
      "step": 36395
    },
    {
      "epoch": 0.3702799479166667,
      "grad_norm": 14.085732460021973,
      "learning_rate": 3.4954732658703862e-06,
      "loss": 2.7789,
      "step": 36400
    },
    {
      "epoch": 0.370330810546875,
      "grad_norm": 15.742152214050293,
      "learning_rate": 3.4951064410432446e-06,
      "loss": 3.7796,
      "step": 36405
    },
    {
      "epoch": 0.3703816731770833,
      "grad_norm": 14.037017822265625,
      "learning_rate": 3.494739590756541e-06,
      "loss": 3.049,
      "step": 36410
    },
    {
      "epoch": 0.3704325358072917,
      "grad_norm": 13.601982116699219,
      "learning_rate": 3.4943727150196607e-06,
      "loss": 3.3264,
      "step": 36415
    },
    {
      "epoch": 0.3704833984375,
      "grad_norm": 10.641363143920898,
      "learning_rate": 3.4940058138419903e-06,
      "loss": 3.3848,
      "step": 36420
    },
    {
      "epoch": 0.3705342610677083,
      "grad_norm": 12.306524276733398,
      "learning_rate": 3.4936388872329163e-06,
      "loss": 3.1746,
      "step": 36425
    },
    {
      "epoch": 0.3705851236979167,
      "grad_norm": 9.204092979431152,
      "learning_rate": 3.4932719352018272e-06,
      "loss": 3.4841,
      "step": 36430
    },
    {
      "epoch": 0.370635986328125,
      "grad_norm": 12.298256874084473,
      "learning_rate": 3.4929049577581107e-06,
      "loss": 3.1072,
      "step": 36435
    },
    {
      "epoch": 0.3706868489583333,
      "grad_norm": 10.112834930419922,
      "learning_rate": 3.492537954911157e-06,
      "loss": 3.3517,
      "step": 36440
    },
    {
      "epoch": 0.3707377115885417,
      "grad_norm": 9.618741989135742,
      "learning_rate": 3.492170926670354e-06,
      "loss": 3.3893,
      "step": 36445
    },
    {
      "epoch": 0.37078857421875,
      "grad_norm": 11.383841514587402,
      "learning_rate": 3.491803873045094e-06,
      "loss": 3.5973,
      "step": 36450
    },
    {
      "epoch": 0.3708394368489583,
      "grad_norm": 9.854310035705566,
      "learning_rate": 3.491436794044766e-06,
      "loss": 3.3188,
      "step": 36455
    },
    {
      "epoch": 0.3708902994791667,
      "grad_norm": 6.895596981048584,
      "learning_rate": 3.491069689678764e-06,
      "loss": 3.1569,
      "step": 36460
    },
    {
      "epoch": 0.370941162109375,
      "grad_norm": 7.189676761627197,
      "learning_rate": 3.490702559956478e-06,
      "loss": 3.1208,
      "step": 36465
    },
    {
      "epoch": 0.3709920247395833,
      "grad_norm": 12.943304061889648,
      "learning_rate": 3.4903354048873017e-06,
      "loss": 3.2073,
      "step": 36470
    },
    {
      "epoch": 0.3710428873697917,
      "grad_norm": 13.477482795715332,
      "learning_rate": 3.489968224480629e-06,
      "loss": 3.3182,
      "step": 36475
    },
    {
      "epoch": 0.37109375,
      "grad_norm": 12.34349250793457,
      "learning_rate": 3.489601018745854e-06,
      "loss": 3.531,
      "step": 36480
    },
    {
      "epoch": 0.3711446126302083,
      "grad_norm": 10.274271965026855,
      "learning_rate": 3.489233787692371e-06,
      "loss": 3.4869,
      "step": 36485
    },
    {
      "epoch": 0.3711954752604167,
      "grad_norm": 6.926421642303467,
      "learning_rate": 3.4888665313295767e-06,
      "loss": 3.6718,
      "step": 36490
    },
    {
      "epoch": 0.371246337890625,
      "grad_norm": 11.027121543884277,
      "learning_rate": 3.4884992496668655e-06,
      "loss": 3.0825,
      "step": 36495
    },
    {
      "epoch": 0.3712972005208333,
      "grad_norm": 14.85596752166748,
      "learning_rate": 3.488131942713636e-06,
      "loss": 3.6047,
      "step": 36500
    },
    {
      "epoch": 0.3713480631510417,
      "grad_norm": 16.440946578979492,
      "learning_rate": 3.487764610479285e-06,
      "loss": 3.0812,
      "step": 36505
    },
    {
      "epoch": 0.37139892578125,
      "grad_norm": 13.991331100463867,
      "learning_rate": 3.487397252973211e-06,
      "loss": 3.5671,
      "step": 36510
    },
    {
      "epoch": 0.3714497884114583,
      "grad_norm": 10.95915699005127,
      "learning_rate": 3.487029870204812e-06,
      "loss": 3.5479,
      "step": 36515
    },
    {
      "epoch": 0.3715006510416667,
      "grad_norm": 14.857094764709473,
      "learning_rate": 3.4866624621834877e-06,
      "loss": 3.2309,
      "step": 36520
    },
    {
      "epoch": 0.371551513671875,
      "grad_norm": 16.348299026489258,
      "learning_rate": 3.4862950289186383e-06,
      "loss": 3.8629,
      "step": 36525
    },
    {
      "epoch": 0.3716023763020833,
      "grad_norm": 12.496493339538574,
      "learning_rate": 3.4859275704196644e-06,
      "loss": 3.7878,
      "step": 36530
    },
    {
      "epoch": 0.3716532389322917,
      "grad_norm": 13.966819763183594,
      "learning_rate": 3.4855600866959667e-06,
      "loss": 2.906,
      "step": 36535
    },
    {
      "epoch": 0.3717041015625,
      "grad_norm": 8.030806541442871,
      "learning_rate": 3.485192577756948e-06,
      "loss": 3.4312,
      "step": 36540
    },
    {
      "epoch": 0.3717549641927083,
      "grad_norm": 11.806658744812012,
      "learning_rate": 3.4848250436120113e-06,
      "loss": 3.5966,
      "step": 36545
    },
    {
      "epoch": 0.3718058268229167,
      "grad_norm": 9.586294174194336,
      "learning_rate": 3.4844574842705586e-06,
      "loss": 3.2212,
      "step": 36550
    },
    {
      "epoch": 0.371856689453125,
      "grad_norm": 10.939385414123535,
      "learning_rate": 3.4840898997419955e-06,
      "loss": 3.9693,
      "step": 36555
    },
    {
      "epoch": 0.3719075520833333,
      "grad_norm": 14.730448722839355,
      "learning_rate": 3.4837222900357236e-06,
      "loss": 3.2931,
      "step": 36560
    },
    {
      "epoch": 0.3719584147135417,
      "grad_norm": 8.028823852539062,
      "learning_rate": 3.4833546551611517e-06,
      "loss": 3.288,
      "step": 36565
    },
    {
      "epoch": 0.37200927734375,
      "grad_norm": 8.313758850097656,
      "learning_rate": 3.4829869951276833e-06,
      "loss": 3.3429,
      "step": 36570
    },
    {
      "epoch": 0.3720601399739583,
      "grad_norm": 13.379473686218262,
      "learning_rate": 3.482619309944727e-06,
      "loss": 3.2575,
      "step": 36575
    },
    {
      "epoch": 0.3721110026041667,
      "grad_norm": 13.700751304626465,
      "learning_rate": 3.482251599621687e-06,
      "loss": 3.3155,
      "step": 36580
    },
    {
      "epoch": 0.372161865234375,
      "grad_norm": 7.261434555053711,
      "learning_rate": 3.481883864167973e-06,
      "loss": 3.2011,
      "step": 36585
    },
    {
      "epoch": 0.3722127278645833,
      "grad_norm": 13.945944786071777,
      "learning_rate": 3.4815161035929935e-06,
      "loss": 3.6079,
      "step": 36590
    },
    {
      "epoch": 0.3722635904947917,
      "grad_norm": 15.99990177154541,
      "learning_rate": 3.4811483179061565e-06,
      "loss": 3.5521,
      "step": 36595
    },
    {
      "epoch": 0.372314453125,
      "grad_norm": 10.341609001159668,
      "learning_rate": 3.480780507116873e-06,
      "loss": 3.4476,
      "step": 36600
    },
    {
      "epoch": 0.3723653157552083,
      "grad_norm": 10.77724552154541,
      "learning_rate": 3.480412671234552e-06,
      "loss": 3.3492,
      "step": 36605
    },
    {
      "epoch": 0.3724161783854167,
      "grad_norm": 13.822732925415039,
      "learning_rate": 3.480044810268606e-06,
      "loss": 3.2997,
      "step": 36610
    },
    {
      "epoch": 0.372467041015625,
      "grad_norm": 9.237661361694336,
      "learning_rate": 3.4796769242284457e-06,
      "loss": 3.246,
      "step": 36615
    },
    {
      "epoch": 0.3725179036458333,
      "grad_norm": 14.378121376037598,
      "learning_rate": 3.479309013123483e-06,
      "loss": 3.0785,
      "step": 36620
    },
    {
      "epoch": 0.3725687662760417,
      "grad_norm": 10.455818176269531,
      "learning_rate": 3.4789410769631317e-06,
      "loss": 3.5482,
      "step": 36625
    },
    {
      "epoch": 0.37261962890625,
      "grad_norm": 11.180048942565918,
      "learning_rate": 3.478573115756805e-06,
      "loss": 2.9193,
      "step": 36630
    },
    {
      "epoch": 0.3726704915364583,
      "grad_norm": 13.03493595123291,
      "learning_rate": 3.4782051295139164e-06,
      "loss": 3.1065,
      "step": 36635
    },
    {
      "epoch": 0.3727213541666667,
      "grad_norm": 10.241646766662598,
      "learning_rate": 3.4778371182438825e-06,
      "loss": 3.4099,
      "step": 36640
    },
    {
      "epoch": 0.372772216796875,
      "grad_norm": 12.54874038696289,
      "learning_rate": 3.477469081956117e-06,
      "loss": 3.4081,
      "step": 36645
    },
    {
      "epoch": 0.3728230794270833,
      "grad_norm": 13.122376441955566,
      "learning_rate": 3.4771010206600365e-06,
      "loss": 3.5846,
      "step": 36650
    },
    {
      "epoch": 0.3728739420572917,
      "grad_norm": 15.833231925964355,
      "learning_rate": 3.4767329343650584e-06,
      "loss": 3.4329,
      "step": 36655
    },
    {
      "epoch": 0.3729248046875,
      "grad_norm": 19.701417922973633,
      "learning_rate": 3.4763648230805997e-06,
      "loss": 3.8892,
      "step": 36660
    },
    {
      "epoch": 0.3729756673177083,
      "grad_norm": 11.933082580566406,
      "learning_rate": 3.4759966868160784e-06,
      "loss": 3.3443,
      "step": 36665
    },
    {
      "epoch": 0.3730265299479167,
      "grad_norm": 12.45585823059082,
      "learning_rate": 3.4756285255809137e-06,
      "loss": 3.1525,
      "step": 36670
    },
    {
      "epoch": 0.373077392578125,
      "grad_norm": 13.080124855041504,
      "learning_rate": 3.475260339384523e-06,
      "loss": 3.3393,
      "step": 36675
    },
    {
      "epoch": 0.3731282552083333,
      "grad_norm": 17.313114166259766,
      "learning_rate": 3.474892128236329e-06,
      "loss": 3.6527,
      "step": 36680
    },
    {
      "epoch": 0.3731791178385417,
      "grad_norm": 12.358736991882324,
      "learning_rate": 3.474523892145751e-06,
      "loss": 3.2541,
      "step": 36685
    },
    {
      "epoch": 0.37322998046875,
      "grad_norm": 14.195704460144043,
      "learning_rate": 3.4741556311222093e-06,
      "loss": 3.2213,
      "step": 36690
    },
    {
      "epoch": 0.3732808430989583,
      "grad_norm": 12.779534339904785,
      "learning_rate": 3.473787345175127e-06,
      "loss": 3.409,
      "step": 36695
    },
    {
      "epoch": 0.3733317057291667,
      "grad_norm": 12.559048652648926,
      "learning_rate": 3.4734190343139264e-06,
      "loss": 3.289,
      "step": 36700
    },
    {
      "epoch": 0.373382568359375,
      "grad_norm": 15.576386451721191,
      "learning_rate": 3.4730506985480307e-06,
      "loss": 3.7107,
      "step": 36705
    },
    {
      "epoch": 0.3734334309895833,
      "grad_norm": 8.136879920959473,
      "learning_rate": 3.4726823378868634e-06,
      "loss": 3.1032,
      "step": 36710
    },
    {
      "epoch": 0.3734842936197917,
      "grad_norm": 12.802074432373047,
      "learning_rate": 3.4723139523398487e-06,
      "loss": 3.2926,
      "step": 36715
    },
    {
      "epoch": 0.37353515625,
      "grad_norm": 16.977983474731445,
      "learning_rate": 3.4719455419164122e-06,
      "loss": 3.0812,
      "step": 36720
    },
    {
      "epoch": 0.3735860188802083,
      "grad_norm": 15.70494556427002,
      "learning_rate": 3.4715771066259795e-06,
      "loss": 3.2869,
      "step": 36725
    },
    {
      "epoch": 0.3736368815104167,
      "grad_norm": 9.459453582763672,
      "learning_rate": 3.471208646477977e-06,
      "loss": 3.335,
      "step": 36730
    },
    {
      "epoch": 0.373687744140625,
      "grad_norm": 14.831958770751953,
      "learning_rate": 3.4708401614818315e-06,
      "loss": 3.4771,
      "step": 36735
    },
    {
      "epoch": 0.3737386067708333,
      "grad_norm": 12.741683006286621,
      "learning_rate": 3.47047165164697e-06,
      "loss": 3.2332,
      "step": 36740
    },
    {
      "epoch": 0.3737894694010417,
      "grad_norm": 15.601336479187012,
      "learning_rate": 3.470103116982822e-06,
      "loss": 3.4188,
      "step": 36745
    },
    {
      "epoch": 0.37384033203125,
      "grad_norm": 11.928319931030273,
      "learning_rate": 3.469734557498815e-06,
      "loss": 3.5366,
      "step": 36750
    },
    {
      "epoch": 0.3738911946614583,
      "grad_norm": 13.991730690002441,
      "learning_rate": 3.46936597320438e-06,
      "loss": 3.5389,
      "step": 36755
    },
    {
      "epoch": 0.3739420572916667,
      "grad_norm": 12.347566604614258,
      "learning_rate": 3.4689973641089457e-06,
      "loss": 3.3752,
      "step": 36760
    },
    {
      "epoch": 0.373992919921875,
      "grad_norm": 16.722991943359375,
      "learning_rate": 3.4686287302219433e-06,
      "loss": 3.5947,
      "step": 36765
    },
    {
      "epoch": 0.3740437825520833,
      "grad_norm": 12.147093772888184,
      "learning_rate": 3.4682600715528055e-06,
      "loss": 3.3451,
      "step": 36770
    },
    {
      "epoch": 0.3740946451822917,
      "grad_norm": 11.044589042663574,
      "learning_rate": 3.4678913881109626e-06,
      "loss": 3.1253,
      "step": 36775
    },
    {
      "epoch": 0.3741455078125,
      "grad_norm": 7.72145938873291,
      "learning_rate": 3.4675226799058476e-06,
      "loss": 3.1175,
      "step": 36780
    },
    {
      "epoch": 0.3741963704427083,
      "grad_norm": 9.892139434814453,
      "learning_rate": 3.4671539469468944e-06,
      "loss": 3.5971,
      "step": 36785
    },
    {
      "epoch": 0.3742472330729167,
      "grad_norm": 12.230167388916016,
      "learning_rate": 3.4667851892435367e-06,
      "loss": 3.389,
      "step": 36790
    },
    {
      "epoch": 0.374298095703125,
      "grad_norm": 9.405281066894531,
      "learning_rate": 3.4664164068052094e-06,
      "loss": 2.8224,
      "step": 36795
    },
    {
      "epoch": 0.3743489583333333,
      "grad_norm": 11.115234375,
      "learning_rate": 3.466047599641348e-06,
      "loss": 3.2145,
      "step": 36800
    },
    {
      "epoch": 0.3743998209635417,
      "grad_norm": 10.069756507873535,
      "learning_rate": 3.4656787677613855e-06,
      "loss": 3.0822,
      "step": 36805
    },
    {
      "epoch": 0.37445068359375,
      "grad_norm": 8.526987075805664,
      "learning_rate": 3.465309911174763e-06,
      "loss": 3.2606,
      "step": 36810
    },
    {
      "epoch": 0.3745015462239583,
      "grad_norm": 10.341978073120117,
      "learning_rate": 3.464941029890914e-06,
      "loss": 3.1676,
      "step": 36815
    },
    {
      "epoch": 0.3745524088541667,
      "grad_norm": 10.42945671081543,
      "learning_rate": 3.464572123919278e-06,
      "loss": 3.0921,
      "step": 36820
    },
    {
      "epoch": 0.374603271484375,
      "grad_norm": 15.445566177368164,
      "learning_rate": 3.464203193269293e-06,
      "loss": 3.425,
      "step": 36825
    },
    {
      "epoch": 0.3746541341145833,
      "grad_norm": 8.392379760742188,
      "learning_rate": 3.463834237950397e-06,
      "loss": 3.0214,
      "step": 36830
    },
    {
      "epoch": 0.3747049967447917,
      "grad_norm": 14.550819396972656,
      "learning_rate": 3.463465257972031e-06,
      "loss": 3.2801,
      "step": 36835
    },
    {
      "epoch": 0.374755859375,
      "grad_norm": 11.48135757446289,
      "learning_rate": 3.463096253343635e-06,
      "loss": 3.2577,
      "step": 36840
    },
    {
      "epoch": 0.3748067220052083,
      "grad_norm": 10.536014556884766,
      "learning_rate": 3.462727224074649e-06,
      "loss": 3.4733,
      "step": 36845
    },
    {
      "epoch": 0.3748575846354167,
      "grad_norm": 15.377198219299316,
      "learning_rate": 3.4623581701745156e-06,
      "loss": 3.2647,
      "step": 36850
    },
    {
      "epoch": 0.374908447265625,
      "grad_norm": 9.998665809631348,
      "learning_rate": 3.4619890916526766e-06,
      "loss": 3.7961,
      "step": 36855
    },
    {
      "epoch": 0.3749593098958333,
      "grad_norm": 12.39189624786377,
      "learning_rate": 3.461619988518574e-06,
      "loss": 3.211,
      "step": 36860
    },
    {
      "epoch": 0.3750101725260417,
      "grad_norm": 11.28947639465332,
      "learning_rate": 3.4612508607816533e-06,
      "loss": 3.4307,
      "step": 36865
    },
    {
      "epoch": 0.37506103515625,
      "grad_norm": 13.22697925567627,
      "learning_rate": 3.460881708451356e-06,
      "loss": 3.2993,
      "step": 36870
    },
    {
      "epoch": 0.3751118977864583,
      "grad_norm": 11.20608901977539,
      "learning_rate": 3.460512531537128e-06,
      "loss": 3.534,
      "step": 36875
    },
    {
      "epoch": 0.3751627604166667,
      "grad_norm": 9.633331298828125,
      "learning_rate": 3.4601433300484143e-06,
      "loss": 3.2138,
      "step": 36880
    },
    {
      "epoch": 0.375213623046875,
      "grad_norm": 8.668368339538574,
      "learning_rate": 3.459774103994662e-06,
      "loss": 3.4622,
      "step": 36885
    },
    {
      "epoch": 0.3752644856770833,
      "grad_norm": 9.311735153198242,
      "learning_rate": 3.4594048533853166e-06,
      "loss": 3.3273,
      "step": 36890
    },
    {
      "epoch": 0.3753153483072917,
      "grad_norm": 12.04916000366211,
      "learning_rate": 3.4590355782298245e-06,
      "loss": 3.8702,
      "step": 36895
    },
    {
      "epoch": 0.3753662109375,
      "grad_norm": 10.879121780395508,
      "learning_rate": 3.458666278537635e-06,
      "loss": 3.2429,
      "step": 36900
    },
    {
      "epoch": 0.3754170735677083,
      "grad_norm": 8.562045097351074,
      "learning_rate": 3.4582969543181966e-06,
      "loss": 3.6933,
      "step": 36905
    },
    {
      "epoch": 0.3754679361979167,
      "grad_norm": 13.595769882202148,
      "learning_rate": 3.4579276055809566e-06,
      "loss": 3.4905,
      "step": 36910
    },
    {
      "epoch": 0.375518798828125,
      "grad_norm": 10.669339179992676,
      "learning_rate": 3.4575582323353673e-06,
      "loss": 3.0703,
      "step": 36915
    },
    {
      "epoch": 0.3755696614583333,
      "grad_norm": 12.172415733337402,
      "learning_rate": 3.4571888345908767e-06,
      "loss": 3.2722,
      "step": 36920
    },
    {
      "epoch": 0.3756205240885417,
      "grad_norm": 12.426019668579102,
      "learning_rate": 3.4568194123569365e-06,
      "loss": 2.9787,
      "step": 36925
    },
    {
      "epoch": 0.37567138671875,
      "grad_norm": 17.960912704467773,
      "learning_rate": 3.4564499656429983e-06,
      "loss": 3.682,
      "step": 36930
    },
    {
      "epoch": 0.3757222493489583,
      "grad_norm": 10.33563232421875,
      "learning_rate": 3.456080494458515e-06,
      "loss": 3.0111,
      "step": 36935
    },
    {
      "epoch": 0.3757731119791667,
      "grad_norm": 8.455516815185547,
      "learning_rate": 3.455710998812939e-06,
      "loss": 3.2332,
      "step": 36940
    },
    {
      "epoch": 0.375823974609375,
      "grad_norm": 12.977243423461914,
      "learning_rate": 3.4553414787157234e-06,
      "loss": 3.4772,
      "step": 36945
    },
    {
      "epoch": 0.3758748372395833,
      "grad_norm": 10.120162010192871,
      "learning_rate": 3.454971934176322e-06,
      "loss": 2.9089,
      "step": 36950
    },
    {
      "epoch": 0.3759256998697917,
      "grad_norm": 10.681510925292969,
      "learning_rate": 3.4546023652041903e-06,
      "loss": 3.2806,
      "step": 36955
    },
    {
      "epoch": 0.3759765625,
      "grad_norm": 12.329384803771973,
      "learning_rate": 3.454232771808783e-06,
      "loss": 3.3713,
      "step": 36960
    },
    {
      "epoch": 0.3760274251302083,
      "grad_norm": 13.539546966552734,
      "learning_rate": 3.4538631539995575e-06,
      "loss": 3.4736,
      "step": 36965
    },
    {
      "epoch": 0.3760782877604167,
      "grad_norm": 13.913695335388184,
      "learning_rate": 3.453493511785969e-06,
      "loss": 3.0884,
      "step": 36970
    },
    {
      "epoch": 0.376129150390625,
      "grad_norm": 13.61619758605957,
      "learning_rate": 3.4531238451774746e-06,
      "loss": 3.2207,
      "step": 36975
    },
    {
      "epoch": 0.3761800130208333,
      "grad_norm": 8.48856258392334,
      "learning_rate": 3.4527541541835323e-06,
      "loss": 3.3079,
      "step": 36980
    },
    {
      "epoch": 0.3762308756510417,
      "grad_norm": 14.161815643310547,
      "learning_rate": 3.4523844388136013e-06,
      "loss": 3.7884,
      "step": 36985
    },
    {
      "epoch": 0.37628173828125,
      "grad_norm": 17.034425735473633,
      "learning_rate": 3.4520146990771395e-06,
      "loss": 3.52,
      "step": 36990
    },
    {
      "epoch": 0.3763326009114583,
      "grad_norm": 8.844666481018066,
      "learning_rate": 3.451644934983608e-06,
      "loss": 3.49,
      "step": 36995
    },
    {
      "epoch": 0.3763834635416667,
      "grad_norm": 9.999996185302734,
      "learning_rate": 3.451275146542466e-06,
      "loss": 3.3897,
      "step": 37000
    },
    {
      "epoch": 0.376434326171875,
      "grad_norm": 12.458309173583984,
      "learning_rate": 3.450905333763175e-06,
      "loss": 3.9959,
      "step": 37005
    },
    {
      "epoch": 0.3764851888020833,
      "grad_norm": 12.006963729858398,
      "learning_rate": 3.4505354966551968e-06,
      "loss": 3.4137,
      "step": 37010
    },
    {
      "epoch": 0.3765360514322917,
      "grad_norm": 10.976564407348633,
      "learning_rate": 3.450165635227992e-06,
      "loss": 3.3991,
      "step": 37015
    },
    {
      "epoch": 0.3765869140625,
      "grad_norm": 9.879744529724121,
      "learning_rate": 3.449795749491026e-06,
      "loss": 3.1944,
      "step": 37020
    },
    {
      "epoch": 0.3766377766927083,
      "grad_norm": 8.062317848205566,
      "learning_rate": 3.4494258394537604e-06,
      "loss": 3.489,
      "step": 37025
    },
    {
      "epoch": 0.3766886393229167,
      "grad_norm": 8.62879753112793,
      "learning_rate": 3.44905590512566e-06,
      "loss": 3.1431,
      "step": 37030
    },
    {
      "epoch": 0.376739501953125,
      "grad_norm": 10.014484405517578,
      "learning_rate": 3.448685946516189e-06,
      "loss": 3.2168,
      "step": 37035
    },
    {
      "epoch": 0.3767903645833333,
      "grad_norm": 16.322481155395508,
      "learning_rate": 3.4483159636348127e-06,
      "loss": 3.0759,
      "step": 37040
    },
    {
      "epoch": 0.3768412272135417,
      "grad_norm": 14.804939270019531,
      "learning_rate": 3.4479459564909974e-06,
      "loss": 3.7576,
      "step": 37045
    },
    {
      "epoch": 0.37689208984375,
      "grad_norm": 10.91519546508789,
      "learning_rate": 3.447575925094209e-06,
      "loss": 3.4439,
      "step": 37050
    },
    {
      "epoch": 0.3769429524739583,
      "grad_norm": 10.703877449035645,
      "learning_rate": 3.447205869453916e-06,
      "loss": 4.0554,
      "step": 37055
    },
    {
      "epoch": 0.3769938151041667,
      "grad_norm": 15.431044578552246,
      "learning_rate": 3.4468357895795856e-06,
      "loss": 3.1134,
      "step": 37060
    },
    {
      "epoch": 0.377044677734375,
      "grad_norm": 10.122700691223145,
      "learning_rate": 3.4464656854806843e-06,
      "loss": 3.2573,
      "step": 37065
    },
    {
      "epoch": 0.3770955403645833,
      "grad_norm": 12.273554801940918,
      "learning_rate": 3.4460955571666844e-06,
      "loss": 3.44,
      "step": 37070
    },
    {
      "epoch": 0.3771464029947917,
      "grad_norm": 15.437214851379395,
      "learning_rate": 3.4457254046470533e-06,
      "loss": 3.9162,
      "step": 37075
    },
    {
      "epoch": 0.377197265625,
      "grad_norm": 14.604323387145996,
      "learning_rate": 3.4453552279312615e-06,
      "loss": 3.6954,
      "step": 37080
    },
    {
      "epoch": 0.3772481282552083,
      "grad_norm": 21.16621971130371,
      "learning_rate": 3.444985027028781e-06,
      "loss": 3.8536,
      "step": 37085
    },
    {
      "epoch": 0.3772989908854167,
      "grad_norm": 16.204978942871094,
      "learning_rate": 3.444614801949082e-06,
      "loss": 3.4781,
      "step": 37090
    },
    {
      "epoch": 0.377349853515625,
      "grad_norm": 13.42151927947998,
      "learning_rate": 3.444244552701638e-06,
      "loss": 3.5459,
      "step": 37095
    },
    {
      "epoch": 0.3774007161458333,
      "grad_norm": 8.625651359558105,
      "learning_rate": 3.44387427929592e-06,
      "loss": 3.0503,
      "step": 37100
    },
    {
      "epoch": 0.3774515787760417,
      "grad_norm": 12.579015731811523,
      "learning_rate": 3.4435039817414025e-06,
      "loss": 3.4166,
      "step": 37105
    },
    {
      "epoch": 0.37750244140625,
      "grad_norm": 10.912002563476562,
      "learning_rate": 3.4431336600475595e-06,
      "loss": 2.9924,
      "step": 37110
    },
    {
      "epoch": 0.3775533040364583,
      "grad_norm": 11.0430269241333,
      "learning_rate": 3.442763314223865e-06,
      "loss": 3.471,
      "step": 37115
    },
    {
      "epoch": 0.3776041666666667,
      "grad_norm": 12.075160026550293,
      "learning_rate": 3.442392944279795e-06,
      "loss": 3.1275,
      "step": 37120
    },
    {
      "epoch": 0.377655029296875,
      "grad_norm": 13.223108291625977,
      "learning_rate": 3.442022550224825e-06,
      "loss": 3.1128,
      "step": 37125
    },
    {
      "epoch": 0.3777058919270833,
      "grad_norm": 13.981732368469238,
      "learning_rate": 3.4416521320684305e-06,
      "loss": 3.4978,
      "step": 37130
    },
    {
      "epoch": 0.3777567545572917,
      "grad_norm": 10.219934463500977,
      "learning_rate": 3.441281689820091e-06,
      "loss": 3.1208,
      "step": 37135
    },
    {
      "epoch": 0.3778076171875,
      "grad_norm": 10.667137145996094,
      "learning_rate": 3.440911223489281e-06,
      "loss": 4.0081,
      "step": 37140
    },
    {
      "epoch": 0.3778584798177083,
      "grad_norm": 9.314359664916992,
      "learning_rate": 3.4405407330854806e-06,
      "loss": 3.142,
      "step": 37145
    },
    {
      "epoch": 0.3779093424479167,
      "grad_norm": 8.450724601745605,
      "learning_rate": 3.440170218618169e-06,
      "loss": 3.5646,
      "step": 37150
    },
    {
      "epoch": 0.377960205078125,
      "grad_norm": 12.683512687683105,
      "learning_rate": 3.439799680096825e-06,
      "loss": 3.3085,
      "step": 37155
    },
    {
      "epoch": 0.3780110677083333,
      "grad_norm": 15.177215576171875,
      "learning_rate": 3.4394291175309284e-06,
      "loss": 3.2349,
      "step": 37160
    },
    {
      "epoch": 0.3780619303385417,
      "grad_norm": 10.385830879211426,
      "learning_rate": 3.4390585309299617e-06,
      "loss": 3.7784,
      "step": 37165
    },
    {
      "epoch": 0.37811279296875,
      "grad_norm": 12.019537925720215,
      "learning_rate": 3.438687920303404e-06,
      "loss": 3.3907,
      "step": 37170
    },
    {
      "epoch": 0.3781636555989583,
      "grad_norm": 21.841266632080078,
      "learning_rate": 3.4383172856607383e-06,
      "loss": 3.4745,
      "step": 37175
    },
    {
      "epoch": 0.3782145182291667,
      "grad_norm": 15.879341125488281,
      "learning_rate": 3.437946627011448e-06,
      "loss": 3.584,
      "step": 37180
    },
    {
      "epoch": 0.378265380859375,
      "grad_norm": 14.873907089233398,
      "learning_rate": 3.4375759443650148e-06,
      "loss": 3.4617,
      "step": 37185
    },
    {
      "epoch": 0.3783162434895833,
      "grad_norm": 10.92282772064209,
      "learning_rate": 3.4372052377309244e-06,
      "loss": 3.6786,
      "step": 37190
    },
    {
      "epoch": 0.3783671061197917,
      "grad_norm": 16.87035369873047,
      "learning_rate": 3.4368345071186594e-06,
      "loss": 3.4028,
      "step": 37195
    },
    {
      "epoch": 0.37841796875,
      "grad_norm": 17.945995330810547,
      "learning_rate": 3.436463752537706e-06,
      "loss": 3.4505,
      "step": 37200
    },
    {
      "epoch": 0.3784688313802083,
      "grad_norm": 10.609006881713867,
      "learning_rate": 3.436092973997549e-06,
      "loss": 3.5485,
      "step": 37205
    },
    {
      "epoch": 0.3785196940104167,
      "grad_norm": 8.49173641204834,
      "learning_rate": 3.4357221715076748e-06,
      "loss": 3.4046,
      "step": 37210
    },
    {
      "epoch": 0.378570556640625,
      "grad_norm": 10.780359268188477,
      "learning_rate": 3.435351345077571e-06,
      "loss": 2.913,
      "step": 37215
    },
    {
      "epoch": 0.3786214192708333,
      "grad_norm": 9.702098846435547,
      "learning_rate": 3.4349804947167253e-06,
      "loss": 3.3036,
      "step": 37220
    },
    {
      "epoch": 0.3786722819010417,
      "grad_norm": 12.375340461730957,
      "learning_rate": 3.4346096204346247e-06,
      "loss": 3.3027,
      "step": 37225
    },
    {
      "epoch": 0.37872314453125,
      "grad_norm": 9.697147369384766,
      "learning_rate": 3.434238722240759e-06,
      "loss": 3.4894,
      "step": 37230
    },
    {
      "epoch": 0.3787740071614583,
      "grad_norm": 12.709754943847656,
      "learning_rate": 3.4338678001446156e-06,
      "loss": 3.2872,
      "step": 37235
    },
    {
      "epoch": 0.3788248697916667,
      "grad_norm": 9.060853958129883,
      "learning_rate": 3.433496854155688e-06,
      "loss": 3.6688,
      "step": 37240
    },
    {
      "epoch": 0.378875732421875,
      "grad_norm": 11.765962600708008,
      "learning_rate": 3.433125884283463e-06,
      "loss": 2.81,
      "step": 37245
    },
    {
      "epoch": 0.3789265950520833,
      "grad_norm": 11.612247467041016,
      "learning_rate": 3.432754890537434e-06,
      "loss": 3.3576,
      "step": 37250
    },
    {
      "epoch": 0.3789774576822917,
      "grad_norm": 11.189865112304688,
      "learning_rate": 3.432383872927092e-06,
      "loss": 3.1866,
      "step": 37255
    },
    {
      "epoch": 0.3790283203125,
      "grad_norm": 8.750208854675293,
      "learning_rate": 3.43201283146193e-06,
      "loss": 3.2166,
      "step": 37260
    },
    {
      "epoch": 0.3790791829427083,
      "grad_norm": 16.319190979003906,
      "learning_rate": 3.43164176615144e-06,
      "loss": 3.3616,
      "step": 37265
    },
    {
      "epoch": 0.3791300455729167,
      "grad_norm": 12.739177703857422,
      "learning_rate": 3.431270677005117e-06,
      "loss": 3.4428,
      "step": 37270
    },
    {
      "epoch": 0.379180908203125,
      "grad_norm": 12.700593948364258,
      "learning_rate": 3.430899564032454e-06,
      "loss": 3.8315,
      "step": 37275
    },
    {
      "epoch": 0.3792317708333333,
      "grad_norm": 10.68607234954834,
      "learning_rate": 3.4305284272429463e-06,
      "loss": 3.041,
      "step": 37280
    },
    {
      "epoch": 0.3792826334635417,
      "grad_norm": 10.991312980651855,
      "learning_rate": 3.430157266646089e-06,
      "loss": 3.2114,
      "step": 37285
    },
    {
      "epoch": 0.37933349609375,
      "grad_norm": 11.34199333190918,
      "learning_rate": 3.4297860822513794e-06,
      "loss": 3.1788,
      "step": 37290
    },
    {
      "epoch": 0.3793843587239583,
      "grad_norm": 15.527979850769043,
      "learning_rate": 3.429414874068313e-06,
      "loss": 3.0974,
      "step": 37295
    },
    {
      "epoch": 0.3794352213541667,
      "grad_norm": 15.220682144165039,
      "learning_rate": 3.4290436421063867e-06,
      "loss": 3.6317,
      "step": 37300
    },
    {
      "epoch": 0.379486083984375,
      "grad_norm": 11.738930702209473,
      "learning_rate": 3.4286723863751004e-06,
      "loss": 3.5017,
      "step": 37305
    },
    {
      "epoch": 0.3795369466145833,
      "grad_norm": 11.440299034118652,
      "learning_rate": 3.42830110688395e-06,
      "loss": 3.6732,
      "step": 37310
    },
    {
      "epoch": 0.3795878092447917,
      "grad_norm": 12.166158676147461,
      "learning_rate": 3.4279298036424356e-06,
      "loss": 2.5458,
      "step": 37315
    },
    {
      "epoch": 0.379638671875,
      "grad_norm": 9.943821907043457,
      "learning_rate": 3.4275584766600578e-06,
      "loss": 3.2652,
      "step": 37320
    },
    {
      "epoch": 0.3796895345052083,
      "grad_norm": 9.365017890930176,
      "learning_rate": 3.4271871259463164e-06,
      "loss": 3.2439,
      "step": 37325
    },
    {
      "epoch": 0.3797403971354167,
      "grad_norm": 12.738382339477539,
      "learning_rate": 3.4268157515107113e-06,
      "loss": 3.2796,
      "step": 37330
    },
    {
      "epoch": 0.379791259765625,
      "grad_norm": 11.039345741271973,
      "learning_rate": 3.4264443533627454e-06,
      "loss": 3.5859,
      "step": 37335
    },
    {
      "epoch": 0.3798421223958333,
      "grad_norm": 14.100454330444336,
      "learning_rate": 3.4260729315119207e-06,
      "loss": 3.0601,
      "step": 37340
    },
    {
      "epoch": 0.3798929850260417,
      "grad_norm": 15.386163711547852,
      "learning_rate": 3.4257014859677397e-06,
      "loss": 3.2344,
      "step": 37345
    },
    {
      "epoch": 0.37994384765625,
      "grad_norm": 8.511161804199219,
      "learning_rate": 3.4253300167397042e-06,
      "loss": 3.1838,
      "step": 37350
    },
    {
      "epoch": 0.3799947102864583,
      "grad_norm": 15.614014625549316,
      "learning_rate": 3.4249585238373207e-06,
      "loss": 3.9259,
      "step": 37355
    },
    {
      "epoch": 0.3800455729166667,
      "grad_norm": 13.464808464050293,
      "learning_rate": 3.4245870072700927e-06,
      "loss": 3.2666,
      "step": 37360
    },
    {
      "epoch": 0.380096435546875,
      "grad_norm": 8.618041038513184,
      "learning_rate": 3.4242154670475245e-06,
      "loss": 3.5037,
      "step": 37365
    },
    {
      "epoch": 0.3801472981770833,
      "grad_norm": 16.455469131469727,
      "learning_rate": 3.423843903179123e-06,
      "loss": 3.7024,
      "step": 37370
    },
    {
      "epoch": 0.3801981608072917,
      "grad_norm": 7.997560501098633,
      "learning_rate": 3.4234723156743945e-06,
      "loss": 2.9579,
      "step": 37375
    },
    {
      "epoch": 0.3802490234375,
      "grad_norm": 8.839055061340332,
      "learning_rate": 3.423100704542845e-06,
      "loss": 3.616,
      "step": 37380
    },
    {
      "epoch": 0.3802998860677083,
      "grad_norm": 17.621545791625977,
      "learning_rate": 3.422729069793983e-06,
      "loss": 3.3201,
      "step": 37385
    },
    {
      "epoch": 0.3803507486979167,
      "grad_norm": 8.127358436584473,
      "learning_rate": 3.422357411437317e-06,
      "loss": 3.215,
      "step": 37390
    },
    {
      "epoch": 0.380401611328125,
      "grad_norm": 11.943425178527832,
      "learning_rate": 3.4219857294823546e-06,
      "loss": 3.4149,
      "step": 37395
    },
    {
      "epoch": 0.3804524739583333,
      "grad_norm": 10.405409812927246,
      "learning_rate": 3.421614023938607e-06,
      "loss": 2.9753,
      "step": 37400
    },
    {
      "epoch": 0.3805033365885417,
      "grad_norm": 8.548178672790527,
      "learning_rate": 3.4212422948155816e-06,
      "loss": 3.5656,
      "step": 37405
    },
    {
      "epoch": 0.38055419921875,
      "grad_norm": 12.904085159301758,
      "learning_rate": 3.420870542122791e-06,
      "loss": 3.1404,
      "step": 37410
    },
    {
      "epoch": 0.3806050618489583,
      "grad_norm": 11.220914840698242,
      "learning_rate": 3.420498765869746e-06,
      "loss": 3.2931,
      "step": 37415
    },
    {
      "epoch": 0.3806559244791667,
      "grad_norm": 9.857573509216309,
      "learning_rate": 3.420126966065958e-06,
      "loss": 3.2885,
      "step": 37420
    },
    {
      "epoch": 0.380706787109375,
      "grad_norm": 10.07679557800293,
      "learning_rate": 3.41975514272094e-06,
      "loss": 3.2386,
      "step": 37425
    },
    {
      "epoch": 0.3807576497395833,
      "grad_norm": 14.047000885009766,
      "learning_rate": 3.419383295844204e-06,
      "loss": 3.236,
      "step": 37430
    },
    {
      "epoch": 0.3808085123697917,
      "grad_norm": 9.781455039978027,
      "learning_rate": 3.4190114254452654e-06,
      "loss": 3.4674,
      "step": 37435
    },
    {
      "epoch": 0.380859375,
      "grad_norm": 17.191829681396484,
      "learning_rate": 3.4186395315336364e-06,
      "loss": 3.4247,
      "step": 37440
    },
    {
      "epoch": 0.3809102376302083,
      "grad_norm": 8.731918334960938,
      "learning_rate": 3.4182676141188332e-06,
      "loss": 3.1671,
      "step": 37445
    },
    {
      "epoch": 0.3809611002604167,
      "grad_norm": 12.978816032409668,
      "learning_rate": 3.4178956732103706e-06,
      "loss": 3.4202,
      "step": 37450
    },
    {
      "epoch": 0.381011962890625,
      "grad_norm": 12.776073455810547,
      "learning_rate": 3.4175237088177645e-06,
      "loss": 3.2907,
      "step": 37455
    },
    {
      "epoch": 0.3810628255208333,
      "grad_norm": 17.695144653320312,
      "learning_rate": 3.417151720950532e-06,
      "loss": 3.5898,
      "step": 37460
    },
    {
      "epoch": 0.3811136881510417,
      "grad_norm": 10.285888671875,
      "learning_rate": 3.4167797096181905e-06,
      "loss": 3.3763,
      "step": 37465
    },
    {
      "epoch": 0.38116455078125,
      "grad_norm": 10.863819122314453,
      "learning_rate": 3.4164076748302567e-06,
      "loss": 3.5098,
      "step": 37470
    },
    {
      "epoch": 0.3812154134114583,
      "grad_norm": 11.196011543273926,
      "learning_rate": 3.4160356165962506e-06,
      "loss": 3.2739,
      "step": 37475
    },
    {
      "epoch": 0.3812662760416667,
      "grad_norm": 11.469472885131836,
      "learning_rate": 3.4156635349256895e-06,
      "loss": 2.9974,
      "step": 37480
    },
    {
      "epoch": 0.381317138671875,
      "grad_norm": 10.920792579650879,
      "learning_rate": 3.4152914298280944e-06,
      "loss": 3.9813,
      "step": 37485
    },
    {
      "epoch": 0.3813680013020833,
      "grad_norm": 10.986191749572754,
      "learning_rate": 3.414919301312985e-06,
      "loss": 3.4946,
      "step": 37490
    },
    {
      "epoch": 0.3814188639322917,
      "grad_norm": 17.591659545898438,
      "learning_rate": 3.414547149389882e-06,
      "loss": 3.3492,
      "step": 37495
    },
    {
      "epoch": 0.3814697265625,
      "grad_norm": 9.149885177612305,
      "learning_rate": 3.4141749740683068e-06,
      "loss": 3.3342,
      "step": 37500
    },
    {
      "epoch": 0.3815205891927083,
      "grad_norm": 8.332978248596191,
      "learning_rate": 3.413802775357782e-06,
      "loss": 3.405,
      "step": 37505
    },
    {
      "epoch": 0.3815714518229167,
      "grad_norm": 7.789477825164795,
      "learning_rate": 3.4134305532678293e-06,
      "loss": 3.8016,
      "step": 37510
    },
    {
      "epoch": 0.381622314453125,
      "grad_norm": 9.647196769714355,
      "learning_rate": 3.4130583078079725e-06,
      "loss": 2.9892,
      "step": 37515
    },
    {
      "epoch": 0.3816731770833333,
      "grad_norm": 9.158622741699219,
      "learning_rate": 3.412686038987735e-06,
      "loss": 3.3382,
      "step": 37520
    },
    {
      "epoch": 0.3817240397135417,
      "grad_norm": 8.998952865600586,
      "learning_rate": 3.4123137468166427e-06,
      "loss": 3.0108,
      "step": 37525
    },
    {
      "epoch": 0.38177490234375,
      "grad_norm": 13.929523468017578,
      "learning_rate": 3.4119414313042186e-06,
      "loss": 3.0273,
      "step": 37530
    },
    {
      "epoch": 0.3818257649739583,
      "grad_norm": 10.859432220458984,
      "learning_rate": 3.411569092459989e-06,
      "loss": 3.6175,
      "step": 37535
    },
    {
      "epoch": 0.3818766276041667,
      "grad_norm": 8.946311950683594,
      "learning_rate": 3.4111967302934807e-06,
      "loss": 3.3618,
      "step": 37540
    },
    {
      "epoch": 0.381927490234375,
      "grad_norm": 14.10853385925293,
      "learning_rate": 3.4108243448142196e-06,
      "loss": 3.1185,
      "step": 37545
    },
    {
      "epoch": 0.3819783528645833,
      "grad_norm": 11.832138061523438,
      "learning_rate": 3.410451936031734e-06,
      "loss": 3.5672,
      "step": 37550
    },
    {
      "epoch": 0.3820292154947917,
      "grad_norm": 12.631633758544922,
      "learning_rate": 3.410079503955552e-06,
      "loss": 3.3899,
      "step": 37555
    },
    {
      "epoch": 0.382080078125,
      "grad_norm": 9.619515419006348,
      "learning_rate": 3.4097070485951998e-06,
      "loss": 3.1777,
      "step": 37560
    },
    {
      "epoch": 0.3821309407552083,
      "grad_norm": 11.570428848266602,
      "learning_rate": 3.4093345699602097e-06,
      "loss": 3.4874,
      "step": 37565
    },
    {
      "epoch": 0.3821818033854167,
      "grad_norm": 15.203789710998535,
      "learning_rate": 3.408962068060109e-06,
      "loss": 3.653,
      "step": 37570
    },
    {
      "epoch": 0.382232666015625,
      "grad_norm": 14.011837005615234,
      "learning_rate": 3.4085895429044303e-06,
      "loss": 3.3553,
      "step": 37575
    },
    {
      "epoch": 0.3822835286458333,
      "grad_norm": 15.710648536682129,
      "learning_rate": 3.4082169945027037e-06,
      "loss": 3.8274,
      "step": 37580
    },
    {
      "epoch": 0.3823343912760417,
      "grad_norm": 13.160861015319824,
      "learning_rate": 3.4078444228644596e-06,
      "loss": 3.4303,
      "step": 37585
    },
    {
      "epoch": 0.38238525390625,
      "grad_norm": 12.528409957885742,
      "learning_rate": 3.4074718279992324e-06,
      "loss": 3.1339,
      "step": 37590
    },
    {
      "epoch": 0.3824361165364583,
      "grad_norm": 15.585745811462402,
      "learning_rate": 3.4070992099165523e-06,
      "loss": 3.173,
      "step": 37595
    },
    {
      "epoch": 0.3824869791666667,
      "grad_norm": 14.537068367004395,
      "learning_rate": 3.406726568625955e-06,
      "loss": 3.5707,
      "step": 37600
    },
    {
      "epoch": 0.382537841796875,
      "grad_norm": 14.80423641204834,
      "learning_rate": 3.4063539041369726e-06,
      "loss": 3.3544,
      "step": 37605
    },
    {
      "epoch": 0.3825887044270833,
      "grad_norm": 14.486916542053223,
      "learning_rate": 3.4059812164591407e-06,
      "loss": 3.2886,
      "step": 37610
    },
    {
      "epoch": 0.3826395670572917,
      "grad_norm": 10.272647857666016,
      "learning_rate": 3.405608505601994e-06,
      "loss": 3.2059,
      "step": 37615
    },
    {
      "epoch": 0.3826904296875,
      "grad_norm": 14.196993827819824,
      "learning_rate": 3.4052357715750683e-06,
      "loss": 3.7612,
      "step": 37620
    },
    {
      "epoch": 0.3827412923177083,
      "grad_norm": 11.729535102844238,
      "learning_rate": 3.4048630143878997e-06,
      "loss": 3.5948,
      "step": 37625
    },
    {
      "epoch": 0.3827921549479167,
      "grad_norm": 12.782881736755371,
      "learning_rate": 3.4044902340500263e-06,
      "loss": 3.5323,
      "step": 37630
    },
    {
      "epoch": 0.382843017578125,
      "grad_norm": 13.800341606140137,
      "learning_rate": 3.4041174305709833e-06,
      "loss": 3.7534,
      "step": 37635
    },
    {
      "epoch": 0.3828938802083333,
      "grad_norm": 9.593366622924805,
      "learning_rate": 3.403744603960311e-06,
      "loss": 3.3979,
      "step": 37640
    },
    {
      "epoch": 0.3829447428385417,
      "grad_norm": 13.039984703063965,
      "learning_rate": 3.4033717542275466e-06,
      "loss": 3.4147,
      "step": 37645
    },
    {
      "epoch": 0.38299560546875,
      "grad_norm": 11.21808910369873,
      "learning_rate": 3.402998881382231e-06,
      "loss": 3.0617,
      "step": 37650
    },
    {
      "epoch": 0.3830464680989583,
      "grad_norm": 14.062593460083008,
      "learning_rate": 3.4026259854339023e-06,
      "loss": 3.0885,
      "step": 37655
    },
    {
      "epoch": 0.3830973307291667,
      "grad_norm": 8.29125690460205,
      "learning_rate": 3.4022530663921015e-06,
      "loss": 3.4393,
      "step": 37660
    },
    {
      "epoch": 0.383148193359375,
      "grad_norm": 14.169401168823242,
      "learning_rate": 3.4018801242663706e-06,
      "loss": 3.2393,
      "step": 37665
    },
    {
      "epoch": 0.3831990559895833,
      "grad_norm": 8.234171867370605,
      "learning_rate": 3.40150715906625e-06,
      "loss": 3.3952,
      "step": 37670
    },
    {
      "epoch": 0.3832499186197917,
      "grad_norm": 11.920029640197754,
      "learning_rate": 3.4011341708012826e-06,
      "loss": 3.3447,
      "step": 37675
    },
    {
      "epoch": 0.38330078125,
      "grad_norm": 12.649847984313965,
      "learning_rate": 3.400761159481011e-06,
      "loss": 3.7396,
      "step": 37680
    },
    {
      "epoch": 0.3833516438802083,
      "grad_norm": 12.721214294433594,
      "learning_rate": 3.400388125114979e-06,
      "loss": 3.185,
      "step": 37685
    },
    {
      "epoch": 0.3834025065104167,
      "grad_norm": 13.129537582397461,
      "learning_rate": 3.4000150677127297e-06,
      "loss": 3.4912,
      "step": 37690
    },
    {
      "epoch": 0.383453369140625,
      "grad_norm": 15.017121315002441,
      "learning_rate": 3.399641987283808e-06,
      "loss": 3.5474,
      "step": 37695
    },
    {
      "epoch": 0.3835042317708333,
      "grad_norm": 7.627927303314209,
      "learning_rate": 3.39926888383776e-06,
      "loss": 2.946,
      "step": 37700
    },
    {
      "epoch": 0.3835550944010417,
      "grad_norm": 10.144737243652344,
      "learning_rate": 3.3988957573841306e-06,
      "loss": 3.4099,
      "step": 37705
    },
    {
      "epoch": 0.38360595703125,
      "grad_norm": 15.387688636779785,
      "learning_rate": 3.398522607932466e-06,
      "loss": 3.6402,
      "step": 37710
    },
    {
      "epoch": 0.3836568196614583,
      "grad_norm": 11.039338111877441,
      "learning_rate": 3.3981494354923137e-06,
      "loss": 3.7447,
      "step": 37715
    },
    {
      "epoch": 0.3837076822916667,
      "grad_norm": 11.357863426208496,
      "learning_rate": 3.3977762400732206e-06,
      "loss": 3.3448,
      "step": 37720
    },
    {
      "epoch": 0.383758544921875,
      "grad_norm": 13.65638542175293,
      "learning_rate": 3.397403021684735e-06,
      "loss": 2.9614,
      "step": 37725
    },
    {
      "epoch": 0.3838094075520833,
      "grad_norm": 10.85644245147705,
      "learning_rate": 3.3970297803364066e-06,
      "loss": 3.4445,
      "step": 37730
    },
    {
      "epoch": 0.3838602701822917,
      "grad_norm": 8.943359375,
      "learning_rate": 3.396656516037784e-06,
      "loss": 3.6558,
      "step": 37735
    },
    {
      "epoch": 0.3839111328125,
      "grad_norm": 10.290979385375977,
      "learning_rate": 3.396283228798416e-06,
      "loss": 3.4049,
      "step": 37740
    },
    {
      "epoch": 0.3839619954427083,
      "grad_norm": 14.245318412780762,
      "learning_rate": 3.3959099186278553e-06,
      "loss": 3.4897,
      "step": 37745
    },
    {
      "epoch": 0.3840128580729167,
      "grad_norm": 12.544392585754395,
      "learning_rate": 3.395536585535651e-06,
      "loss": 3.0121,
      "step": 37750
    },
    {
      "epoch": 0.384063720703125,
      "grad_norm": 8.912129402160645,
      "learning_rate": 3.3951632295313552e-06,
      "loss": 2.9532,
      "step": 37755
    },
    {
      "epoch": 0.3841145833333333,
      "grad_norm": 11.526346206665039,
      "learning_rate": 3.3947898506245203e-06,
      "loss": 3.4343,
      "step": 37760
    },
    {
      "epoch": 0.3841654459635417,
      "grad_norm": 8.159797668457031,
      "learning_rate": 3.3944164488246992e-06,
      "loss": 3.1769,
      "step": 37765
    },
    {
      "epoch": 0.38421630859375,
      "grad_norm": 13.55785083770752,
      "learning_rate": 3.394043024141446e-06,
      "loss": 3.3025,
      "step": 37770
    },
    {
      "epoch": 0.3842671712239583,
      "grad_norm": 14.68910026550293,
      "learning_rate": 3.393669576584313e-06,
      "loss": 3.4664,
      "step": 37775
    },
    {
      "epoch": 0.3843180338541667,
      "grad_norm": 16.233421325683594,
      "learning_rate": 3.393296106162856e-06,
      "loss": 3.3517,
      "step": 37780
    },
    {
      "epoch": 0.384368896484375,
      "grad_norm": 10.414290428161621,
      "learning_rate": 3.39292261288663e-06,
      "loss": 3.6864,
      "step": 37785
    },
    {
      "epoch": 0.3844197591145833,
      "grad_norm": 8.684490203857422,
      "learning_rate": 3.3925490967651904e-06,
      "loss": 3.3804,
      "step": 37790
    },
    {
      "epoch": 0.3844706217447917,
      "grad_norm": 14.901695251464844,
      "learning_rate": 3.392175557808094e-06,
      "loss": 3.2678,
      "step": 37795
    },
    {
      "epoch": 0.384521484375,
      "grad_norm": 8.012540817260742,
      "learning_rate": 3.3918019960248976e-06,
      "loss": 3.3453,
      "step": 37800
    },
    {
      "epoch": 0.3845723470052083,
      "grad_norm": 14.347160339355469,
      "learning_rate": 3.3914284114251573e-06,
      "loss": 3.6749,
      "step": 37805
    },
    {
      "epoch": 0.3846232096354167,
      "grad_norm": 11.914584159851074,
      "learning_rate": 3.391054804018434e-06,
      "loss": 3.6538,
      "step": 37810
    },
    {
      "epoch": 0.384674072265625,
      "grad_norm": 12.325974464416504,
      "learning_rate": 3.3906811738142843e-06,
      "loss": 3.7795,
      "step": 37815
    },
    {
      "epoch": 0.3847249348958333,
      "grad_norm": 11.537654876708984,
      "learning_rate": 3.3903075208222667e-06,
      "loss": 3.1268,
      "step": 37820
    },
    {
      "epoch": 0.3847757975260417,
      "grad_norm": 13.890557289123535,
      "learning_rate": 3.3899338450519437e-06,
      "loss": 3.4425,
      "step": 37825
    },
    {
      "epoch": 0.38482666015625,
      "grad_norm": 22.127164840698242,
      "learning_rate": 3.3895601465128736e-06,
      "loss": 3.7079,
      "step": 37830
    },
    {
      "epoch": 0.3848775227864583,
      "grad_norm": 11.531108856201172,
      "learning_rate": 3.3891864252146177e-06,
      "loss": 3.5737,
      "step": 37835
    },
    {
      "epoch": 0.3849283854166667,
      "grad_norm": 14.390570640563965,
      "learning_rate": 3.3888126811667387e-06,
      "loss": 3.6822,
      "step": 37840
    },
    {
      "epoch": 0.384979248046875,
      "grad_norm": 9.560125350952148,
      "learning_rate": 3.3884389143787965e-06,
      "loss": 3.6045,
      "step": 37845
    },
    {
      "epoch": 0.3850301106770833,
      "grad_norm": 15.163629531860352,
      "learning_rate": 3.3880651248603567e-06,
      "loss": 3.3025,
      "step": 37850
    },
    {
      "epoch": 0.3850809733072917,
      "grad_norm": 15.606112480163574,
      "learning_rate": 3.387691312620981e-06,
      "loss": 3.3638,
      "step": 37855
    },
    {
      "epoch": 0.3851318359375,
      "grad_norm": 13.837167739868164,
      "learning_rate": 3.387317477670233e-06,
      "loss": 3.6738,
      "step": 37860
    },
    {
      "epoch": 0.3851826985677083,
      "grad_norm": 9.533172607421875,
      "learning_rate": 3.3869436200176775e-06,
      "loss": 3.3599,
      "step": 37865
    },
    {
      "epoch": 0.3852335611979167,
      "grad_norm": 13.04584789276123,
      "learning_rate": 3.3865697396728795e-06,
      "loss": 3.492,
      "step": 37870
    },
    {
      "epoch": 0.385284423828125,
      "grad_norm": 10.076404571533203,
      "learning_rate": 3.386195836645406e-06,
      "loss": 3.3765,
      "step": 37875
    },
    {
      "epoch": 0.3853352864583333,
      "grad_norm": 12.413851737976074,
      "learning_rate": 3.385821910944821e-06,
      "loss": 3.2374,
      "step": 37880
    },
    {
      "epoch": 0.3853861490885417,
      "grad_norm": 8.019109725952148,
      "learning_rate": 3.385447962580693e-06,
      "loss": 3.2756,
      "step": 37885
    },
    {
      "epoch": 0.38543701171875,
      "grad_norm": 7.4843645095825195,
      "learning_rate": 3.3850739915625885e-06,
      "loss": 3.2692,
      "step": 37890
    },
    {
      "epoch": 0.3854878743489583,
      "grad_norm": 14.0797119140625,
      "learning_rate": 3.3846999979000757e-06,
      "loss": 3.5322,
      "step": 37895
    },
    {
      "epoch": 0.3855387369791667,
      "grad_norm": 11.294448852539062,
      "learning_rate": 3.384325981602723e-06,
      "loss": 3.4381,
      "step": 37900
    },
    {
      "epoch": 0.385589599609375,
      "grad_norm": 17.284252166748047,
      "learning_rate": 3.3839519426801006e-06,
      "loss": 3.3795,
      "step": 37905
    },
    {
      "epoch": 0.3856404622395833,
      "grad_norm": 11.139725685119629,
      "learning_rate": 3.3835778811417757e-06,
      "loss": 3.2291,
      "step": 37910
    },
    {
      "epoch": 0.3856913248697917,
      "grad_norm": 14.060380935668945,
      "learning_rate": 3.3832037969973214e-06,
      "loss": 3.5328,
      "step": 37915
    },
    {
      "epoch": 0.3857421875,
      "grad_norm": 16.855520248413086,
      "learning_rate": 3.3828296902563066e-06,
      "loss": 3.7078,
      "step": 37920
    },
    {
      "epoch": 0.3857930501302083,
      "grad_norm": 9.740677833557129,
      "learning_rate": 3.382455560928304e-06,
      "loss": 3.6188,
      "step": 37925
    },
    {
      "epoch": 0.3858439127604167,
      "grad_norm": 11.402836799621582,
      "learning_rate": 3.3820814090228847e-06,
      "loss": 3.5231,
      "step": 37930
    },
    {
      "epoch": 0.385894775390625,
      "grad_norm": 20.03525733947754,
      "learning_rate": 3.3817072345496214e-06,
      "loss": 3.5392,
      "step": 37935
    },
    {
      "epoch": 0.3859456380208333,
      "grad_norm": 10.353108406066895,
      "learning_rate": 3.3813330375180873e-06,
      "loss": 3.2455,
      "step": 37940
    },
    {
      "epoch": 0.3859965006510417,
      "grad_norm": 14.113561630249023,
      "learning_rate": 3.3809588179378577e-06,
      "loss": 3.0992,
      "step": 37945
    },
    {
      "epoch": 0.38604736328125,
      "grad_norm": 9.764810562133789,
      "learning_rate": 3.380584575818504e-06,
      "loss": 3.2733,
      "step": 37950
    },
    {
      "epoch": 0.3860982259114583,
      "grad_norm": 9.683736801147461,
      "learning_rate": 3.380210311169604e-06,
      "loss": 3.3936,
      "step": 37955
    },
    {
      "epoch": 0.3861490885416667,
      "grad_norm": 7.924754619598389,
      "learning_rate": 3.3798360240007306e-06,
      "loss": 3.3648,
      "step": 37960
    },
    {
      "epoch": 0.386199951171875,
      "grad_norm": 11.664175987243652,
      "learning_rate": 3.3794617143214615e-06,
      "loss": 3.2296,
      "step": 37965
    },
    {
      "epoch": 0.3862508138020833,
      "grad_norm": 8.792074203491211,
      "learning_rate": 3.3790873821413733e-06,
      "loss": 3.8841,
      "step": 37970
    },
    {
      "epoch": 0.3863016764322917,
      "grad_norm": 14.745179176330566,
      "learning_rate": 3.3787130274700417e-06,
      "loss": 3.5543,
      "step": 37975
    },
    {
      "epoch": 0.3863525390625,
      "grad_norm": 10.410104751586914,
      "learning_rate": 3.3783386503170467e-06,
      "loss": 3.6948,
      "step": 37980
    },
    {
      "epoch": 0.3864034016927083,
      "grad_norm": 10.549935340881348,
      "learning_rate": 3.377964250691965e-06,
      "loss": 3.4288,
      "step": 37985
    },
    {
      "epoch": 0.3864542643229167,
      "grad_norm": 10.103965759277344,
      "learning_rate": 3.3775898286043762e-06,
      "loss": 3.451,
      "step": 37990
    },
    {
      "epoch": 0.386505126953125,
      "grad_norm": 12.432733535766602,
      "learning_rate": 3.37721538406386e-06,
      "loss": 3.3372,
      "step": 37995
    },
    {
      "epoch": 0.3865559895833333,
      "grad_norm": 10.565135955810547,
      "learning_rate": 3.3768409170799956e-06,
      "loss": 3.3918,
      "step": 38000
    },
    {
      "epoch": 0.3866068522135417,
      "grad_norm": 10.372321128845215,
      "learning_rate": 3.376466427662364e-06,
      "loss": 3.3803,
      "step": 38005
    },
    {
      "epoch": 0.38665771484375,
      "grad_norm": 11.90112590789795,
      "learning_rate": 3.3760919158205472e-06,
      "loss": 3.3195,
      "step": 38010
    },
    {
      "epoch": 0.3867085774739583,
      "grad_norm": 8.72597885131836,
      "learning_rate": 3.375717381564126e-06,
      "loss": 3.4056,
      "step": 38015
    },
    {
      "epoch": 0.3867594401041667,
      "grad_norm": 12.27939510345459,
      "learning_rate": 3.375342824902684e-06,
      "loss": 3.9223,
      "step": 38020
    },
    {
      "epoch": 0.386810302734375,
      "grad_norm": 13.502927780151367,
      "learning_rate": 3.3749682458458016e-06,
      "loss": 2.9973,
      "step": 38025
    },
    {
      "epoch": 0.3868611653645833,
      "grad_norm": 14.647866249084473,
      "learning_rate": 3.374593644403066e-06,
      "loss": 3.3111,
      "step": 38030
    },
    {
      "epoch": 0.3869120279947917,
      "grad_norm": 8.057086944580078,
      "learning_rate": 3.3742190205840574e-06,
      "loss": 3.7146,
      "step": 38035
    },
    {
      "epoch": 0.386962890625,
      "grad_norm": 14.464299201965332,
      "learning_rate": 3.373844374398364e-06,
      "loss": 3.6537,
      "step": 38040
    },
    {
      "epoch": 0.3870137532552083,
      "grad_norm": 8.134767532348633,
      "learning_rate": 3.373469705855568e-06,
      "loss": 3.5302,
      "step": 38045
    },
    {
      "epoch": 0.3870646158854167,
      "grad_norm": 12.267831802368164,
      "learning_rate": 3.3730950149652574e-06,
      "loss": 3.4322,
      "step": 38050
    },
    {
      "epoch": 0.387115478515625,
      "grad_norm": 14.318890571594238,
      "learning_rate": 3.372720301737018e-06,
      "loss": 3.4095,
      "step": 38055
    },
    {
      "epoch": 0.3871663411458333,
      "grad_norm": 11.825246810913086,
      "learning_rate": 3.3723455661804364e-06,
      "loss": 3.3964,
      "step": 38060
    },
    {
      "epoch": 0.3872172037760417,
      "grad_norm": 10.063560485839844,
      "learning_rate": 3.3719708083050996e-06,
      "loss": 3.2893,
      "step": 38065
    },
    {
      "epoch": 0.38726806640625,
      "grad_norm": 12.584623336791992,
      "learning_rate": 3.3715960281205968e-06,
      "loss": 3.6519,
      "step": 38070
    },
    {
      "epoch": 0.3873189290364583,
      "grad_norm": 9.536698341369629,
      "learning_rate": 3.3712212256365167e-06,
      "loss": 3.3757,
      "step": 38075
    },
    {
      "epoch": 0.3873697916666667,
      "grad_norm": 15.417890548706055,
      "learning_rate": 3.370846400862447e-06,
      "loss": 3.301,
      "step": 38080
    },
    {
      "epoch": 0.387420654296875,
      "grad_norm": 10.475431442260742,
      "learning_rate": 3.3704715538079794e-06,
      "loss": 3.2111,
      "step": 38085
    },
    {
      "epoch": 0.3874715169270833,
      "grad_norm": 2325.23974609375,
      "learning_rate": 3.3700966844827033e-06,
      "loss": 5.4233,
      "step": 38090
    },
    {
      "epoch": 0.3875223795572917,
      "grad_norm": 11.278244018554688,
      "learning_rate": 3.3697217928962095e-06,
      "loss": 3.4104,
      "step": 38095
    },
    {
      "epoch": 0.3875732421875,
      "grad_norm": 12.78177261352539,
      "learning_rate": 3.3693468790580895e-06,
      "loss": 3.5219,
      "step": 38100
    },
    {
      "epoch": 0.3876241048177083,
      "grad_norm": 12.451904296875,
      "learning_rate": 3.3689719429779367e-06,
      "loss": 3.0766,
      "step": 38105
    },
    {
      "epoch": 0.3876749674479167,
      "grad_norm": 13.55923843383789,
      "learning_rate": 3.3685969846653417e-06,
      "loss": 3.3094,
      "step": 38110
    },
    {
      "epoch": 0.387725830078125,
      "grad_norm": 15.302332878112793,
      "learning_rate": 3.3682220041298993e-06,
      "loss": 3.0365,
      "step": 38115
    },
    {
      "epoch": 0.3877766927083333,
      "grad_norm": 8.71083927154541,
      "learning_rate": 3.367847001381202e-06,
      "loss": 3.2149,
      "step": 38120
    },
    {
      "epoch": 0.3878275553385417,
      "grad_norm": 11.540759086608887,
      "learning_rate": 3.367471976428846e-06,
      "loss": 3.2459,
      "step": 38125
    },
    {
      "epoch": 0.38787841796875,
      "grad_norm": 12.014065742492676,
      "learning_rate": 3.367096929282424e-06,
      "loss": 3.3404,
      "step": 38130
    },
    {
      "epoch": 0.3879292805989583,
      "grad_norm": 14.558552742004395,
      "learning_rate": 3.366721859951533e-06,
      "loss": 3.2887,
      "step": 38135
    },
    {
      "epoch": 0.3879801432291667,
      "grad_norm": 10.725698471069336,
      "learning_rate": 3.366346768445769e-06,
      "loss": 3.5935,
      "step": 38140
    },
    {
      "epoch": 0.388031005859375,
      "grad_norm": 10.478405952453613,
      "learning_rate": 3.365971654774728e-06,
      "loss": 3.1538,
      "step": 38145
    },
    {
      "epoch": 0.3880818684895833,
      "grad_norm": 9.388585090637207,
      "learning_rate": 3.365596518948007e-06,
      "loss": 3.4627,
      "step": 38150
    },
    {
      "epoch": 0.3881327311197917,
      "grad_norm": 9.345610618591309,
      "learning_rate": 3.365221360975205e-06,
      "loss": 3.3876,
      "step": 38155
    },
    {
      "epoch": 0.38818359375,
      "grad_norm": 15.602376937866211,
      "learning_rate": 3.3648461808659194e-06,
      "loss": 3.1752,
      "step": 38160
    },
    {
      "epoch": 0.3882344563802083,
      "grad_norm": 8.027948379516602,
      "learning_rate": 3.364470978629749e-06,
      "loss": 3.1229,
      "step": 38165
    },
    {
      "epoch": 0.3882853190104167,
      "grad_norm": 16.314498901367188,
      "learning_rate": 3.364095754276293e-06,
      "loss": 3.73,
      "step": 38170
    },
    {
      "epoch": 0.388336181640625,
      "grad_norm": 8.504803657531738,
      "learning_rate": 3.363720507815153e-06,
      "loss": 3.3806,
      "step": 38175
    },
    {
      "epoch": 0.3883870442708333,
      "grad_norm": 16.09746551513672,
      "learning_rate": 3.363345239255928e-06,
      "loss": 3.6046,
      "step": 38180
    },
    {
      "epoch": 0.3884379069010417,
      "grad_norm": 10.618088722229004,
      "learning_rate": 3.3629699486082197e-06,
      "loss": 3.0786,
      "step": 38185
    },
    {
      "epoch": 0.38848876953125,
      "grad_norm": 10.036343574523926,
      "learning_rate": 3.3625946358816303e-06,
      "loss": 2.9537,
      "step": 38190
    },
    {
      "epoch": 0.3885396321614583,
      "grad_norm": 10.493426322937012,
      "learning_rate": 3.3622193010857606e-06,
      "loss": 3.3826,
      "step": 38195
    },
    {
      "epoch": 0.3885904947916667,
      "grad_norm": 13.192093849182129,
      "learning_rate": 3.3618439442302154e-06,
      "loss": 3.2187,
      "step": 38200
    },
    {
      "epoch": 0.388641357421875,
      "grad_norm": 15.833111763000488,
      "learning_rate": 3.361468565324597e-06,
      "loss": 3.2154,
      "step": 38205
    },
    {
      "epoch": 0.3886922200520833,
      "grad_norm": 12.631217002868652,
      "learning_rate": 3.3610931643785095e-06,
      "loss": 3.4353,
      "step": 38210
    },
    {
      "epoch": 0.3887430826822917,
      "grad_norm": 13.263562202453613,
      "learning_rate": 3.360717741401557e-06,
      "loss": 3.3342,
      "step": 38215
    },
    {
      "epoch": 0.3887939453125,
      "grad_norm": 9.171459197998047,
      "learning_rate": 3.3603422964033456e-06,
      "loss": 3.9951,
      "step": 38220
    },
    {
      "epoch": 0.3888448079427083,
      "grad_norm": 11.791070938110352,
      "learning_rate": 3.35996682939348e-06,
      "loss": 3.0921,
      "step": 38225
    },
    {
      "epoch": 0.3888956705729167,
      "grad_norm": 13.00941276550293,
      "learning_rate": 3.359591340381568e-06,
      "loss": 3.9208,
      "step": 38230
    },
    {
      "epoch": 0.388946533203125,
      "grad_norm": 8.12760066986084,
      "learning_rate": 3.3592158293772135e-06,
      "loss": 3.1266,
      "step": 38235
    },
    {
      "epoch": 0.3889973958333333,
      "grad_norm": 9.810503005981445,
      "learning_rate": 3.3588402963900273e-06,
      "loss": 3.2701,
      "step": 38240
    },
    {
      "epoch": 0.3890482584635417,
      "grad_norm": 9.685894012451172,
      "learning_rate": 3.358464741429615e-06,
      "loss": 3.4047,
      "step": 38245
    },
    {
      "epoch": 0.38909912109375,
      "grad_norm": 9.390931129455566,
      "learning_rate": 3.3580891645055857e-06,
      "loss": 3.1132,
      "step": 38250
    },
    {
      "epoch": 0.3891499837239583,
      "grad_norm": 13.093478202819824,
      "learning_rate": 3.3577135656275485e-06,
      "loss": 3.3899,
      "step": 38255
    },
    {
      "epoch": 0.3892008463541667,
      "grad_norm": 10.473499298095703,
      "learning_rate": 3.3573379448051125e-06,
      "loss": 3.2168,
      "step": 38260
    },
    {
      "epoch": 0.389251708984375,
      "grad_norm": 15.70926570892334,
      "learning_rate": 3.3569623020478896e-06,
      "loss": 3.515,
      "step": 38265
    },
    {
      "epoch": 0.3893025716145833,
      "grad_norm": 14.697281837463379,
      "learning_rate": 3.3565866373654887e-06,
      "loss": 3.3272,
      "step": 38270
    },
    {
      "epoch": 0.3893534342447917,
      "grad_norm": 9.951302528381348,
      "learning_rate": 3.356210950767522e-06,
      "loss": 3.1649,
      "step": 38275
    },
    {
      "epoch": 0.389404296875,
      "grad_norm": 9.453201293945312,
      "learning_rate": 3.3558352422636007e-06,
      "loss": 3.5697,
      "step": 38280
    },
    {
      "epoch": 0.3894551595052083,
      "grad_norm": 14.826644897460938,
      "learning_rate": 3.3554595118633376e-06,
      "loss": 3.1434,
      "step": 38285
    },
    {
      "epoch": 0.3895060221354167,
      "grad_norm": 14.816740036010742,
      "learning_rate": 3.3550837595763457e-06,
      "loss": 3.0265,
      "step": 38290
    },
    {
      "epoch": 0.389556884765625,
      "grad_norm": 9.352396965026855,
      "learning_rate": 3.354707985412239e-06,
      "loss": 3.5541,
      "step": 38295
    },
    {
      "epoch": 0.3896077473958333,
      "grad_norm": 14.234426498413086,
      "learning_rate": 3.3543321893806303e-06,
      "loss": 3.2591,
      "step": 38300
    },
    {
      "epoch": 0.3896586100260417,
      "grad_norm": 13.746682167053223,
      "learning_rate": 3.353956371491136e-06,
      "loss": 3.3658,
      "step": 38305
    },
    {
      "epoch": 0.38970947265625,
      "grad_norm": 13.630725860595703,
      "learning_rate": 3.35358053175337e-06,
      "loss": 3.3378,
      "step": 38310
    },
    {
      "epoch": 0.3897603352864583,
      "grad_norm": 9.117375373840332,
      "learning_rate": 3.353204670176948e-06,
      "loss": 3.5115,
      "step": 38315
    },
    {
      "epoch": 0.3898111979166667,
      "grad_norm": 13.624043464660645,
      "learning_rate": 3.3528287867714864e-06,
      "loss": 3.3602,
      "step": 38320
    },
    {
      "epoch": 0.389862060546875,
      "grad_norm": 8.452689170837402,
      "learning_rate": 3.352452881546603e-06,
      "loss": 3.6228,
      "step": 38325
    },
    {
      "epoch": 0.3899129231770833,
      "grad_norm": 11.07059097290039,
      "learning_rate": 3.3520769545119146e-06,
      "loss": 3.1787,
      "step": 38330
    },
    {
      "epoch": 0.3899637858072917,
      "grad_norm": 14.2334623336792,
      "learning_rate": 3.3517010056770395e-06,
      "loss": 3.2317,
      "step": 38335
    },
    {
      "epoch": 0.3900146484375,
      "grad_norm": 11.949200630187988,
      "learning_rate": 3.3513250350515956e-06,
      "loss": 3.1173,
      "step": 38340
    },
    {
      "epoch": 0.3900655110677083,
      "grad_norm": 40.111392974853516,
      "learning_rate": 3.350949042645203e-06,
      "loss": 3.4921,
      "step": 38345
    },
    {
      "epoch": 0.3901163736979167,
      "grad_norm": 14.802889823913574,
      "learning_rate": 3.35057302846748e-06,
      "loss": 3.3277,
      "step": 38350
    },
    {
      "epoch": 0.390167236328125,
      "grad_norm": 9.277702331542969,
      "learning_rate": 3.3501969925280486e-06,
      "loss": 3.4958,
      "step": 38355
    },
    {
      "epoch": 0.3902180989583333,
      "grad_norm": 15.449729919433594,
      "learning_rate": 3.3498209348365285e-06,
      "loss": 3.4463,
      "step": 38360
    },
    {
      "epoch": 0.3902689615885417,
      "grad_norm": 15.135710716247559,
      "learning_rate": 3.3494448554025404e-06,
      "loss": 3.4774,
      "step": 38365
    },
    {
      "epoch": 0.39031982421875,
      "grad_norm": 9.1742582321167,
      "learning_rate": 3.3490687542357083e-06,
      "loss": 3.5518,
      "step": 38370
    },
    {
      "epoch": 0.3903706868489583,
      "grad_norm": 14.67941665649414,
      "learning_rate": 3.3486926313456524e-06,
      "loss": 3.5348,
      "step": 38375
    },
    {
      "epoch": 0.3904215494791667,
      "grad_norm": 9.221536636352539,
      "learning_rate": 3.348316486741997e-06,
      "loss": 3.2008,
      "step": 38380
    },
    {
      "epoch": 0.390472412109375,
      "grad_norm": 10.60205078125,
      "learning_rate": 3.347940320434365e-06,
      "loss": 3.1944,
      "step": 38385
    },
    {
      "epoch": 0.3905232747395833,
      "grad_norm": 8.800256729125977,
      "learning_rate": 3.3475641324323814e-06,
      "loss": 3.3865,
      "step": 38390
    },
    {
      "epoch": 0.3905741373697917,
      "grad_norm": 7.432771682739258,
      "learning_rate": 3.3471879227456706e-06,
      "loss": 3.5595,
      "step": 38395
    },
    {
      "epoch": 0.390625,
      "grad_norm": 14.499105453491211,
      "learning_rate": 3.346811691383857e-06,
      "loss": 3.3196,
      "step": 38400
    },
    {
      "epoch": 0.3906758626302083,
      "grad_norm": 12.427467346191406,
      "learning_rate": 3.346435438356567e-06,
      "loss": 3.555,
      "step": 38405
    },
    {
      "epoch": 0.3907267252604167,
      "grad_norm": 13.272066116333008,
      "learning_rate": 3.3460591636734274e-06,
      "loss": 3.0785,
      "step": 38410
    },
    {
      "epoch": 0.390777587890625,
      "grad_norm": 10.83601188659668,
      "learning_rate": 3.3456828673440637e-06,
      "loss": 3.5042,
      "step": 38415
    },
    {
      "epoch": 0.3908284505208333,
      "grad_norm": 9.56989860534668,
      "learning_rate": 3.3453065493781057e-06,
      "loss": 3.2477,
      "step": 38420
    },
    {
      "epoch": 0.3908793131510417,
      "grad_norm": 9.048479080200195,
      "learning_rate": 3.344930209785179e-06,
      "loss": 3.6054,
      "step": 38425
    },
    {
      "epoch": 0.39093017578125,
      "grad_norm": 9.921485900878906,
      "learning_rate": 3.3445538485749135e-06,
      "loss": 3.2071,
      "step": 38430
    },
    {
      "epoch": 0.3909810384114583,
      "grad_norm": 13.206910133361816,
      "learning_rate": 3.3441774657569374e-06,
      "loss": 3.3644,
      "step": 38435
    },
    {
      "epoch": 0.3910319010416667,
      "grad_norm": 10.711992263793945,
      "learning_rate": 3.3438010613408816e-06,
      "loss": 3.0718,
      "step": 38440
    },
    {
      "epoch": 0.391082763671875,
      "grad_norm": 8.891669273376465,
      "learning_rate": 3.343424635336375e-06,
      "loss": 3.4491,
      "step": 38445
    },
    {
      "epoch": 0.3911336263020833,
      "grad_norm": 16.935821533203125,
      "learning_rate": 3.343048187753051e-06,
      "loss": 3.6171,
      "step": 38450
    },
    {
      "epoch": 0.3911844889322917,
      "grad_norm": 13.522112846374512,
      "learning_rate": 3.3426717186005365e-06,
      "loss": 3.6182,
      "step": 38455
    },
    {
      "epoch": 0.3912353515625,
      "grad_norm": 12.0574369430542,
      "learning_rate": 3.3422952278884663e-06,
      "loss": 3.3017,
      "step": 38460
    },
    {
      "epoch": 0.3912862141927083,
      "grad_norm": 10.464619636535645,
      "learning_rate": 3.341918715626473e-06,
      "loss": 3.3591,
      "step": 38465
    },
    {
      "epoch": 0.3913370768229167,
      "grad_norm": 9.480717658996582,
      "learning_rate": 3.341542181824188e-06,
      "loss": 3.2459,
      "step": 38470
    },
    {
      "epoch": 0.391387939453125,
      "grad_norm": 8.027229309082031,
      "learning_rate": 3.3411656264912473e-06,
      "loss": 3.4763,
      "step": 38475
    },
    {
      "epoch": 0.3914388020833333,
      "grad_norm": 12.631207466125488,
      "learning_rate": 3.340789049637282e-06,
      "loss": 3.5192,
      "step": 38480
    },
    {
      "epoch": 0.3914896647135417,
      "grad_norm": 8.793301582336426,
      "learning_rate": 3.3404124512719295e-06,
      "loss": 3.0144,
      "step": 38485
    },
    {
      "epoch": 0.39154052734375,
      "grad_norm": 12.1557035446167,
      "learning_rate": 3.3400358314048227e-06,
      "loss": 3.5435,
      "step": 38490
    },
    {
      "epoch": 0.3915913899739583,
      "grad_norm": 14.589640617370605,
      "learning_rate": 3.3396591900455983e-06,
      "loss": 3.5327,
      "step": 38495
    },
    {
      "epoch": 0.3916422526041667,
      "grad_norm": 9.701825141906738,
      "learning_rate": 3.3392825272038925e-06,
      "loss": 2.9686,
      "step": 38500
    },
    {
      "epoch": 0.391693115234375,
      "grad_norm": 15.59730052947998,
      "learning_rate": 3.3389058428893424e-06,
      "loss": 3.3883,
      "step": 38505
    },
    {
      "epoch": 0.3917439778645833,
      "grad_norm": 9.84763240814209,
      "learning_rate": 3.3385291371115847e-06,
      "loss": 3.1096,
      "step": 38510
    },
    {
      "epoch": 0.3917948404947917,
      "grad_norm": 11.552181243896484,
      "learning_rate": 3.3381524098802588e-06,
      "loss": 3.4595,
      "step": 38515
    },
    {
      "epoch": 0.391845703125,
      "grad_norm": 12.711458206176758,
      "learning_rate": 3.3377756612050003e-06,
      "loss": 3.2141,
      "step": 38520
    },
    {
      "epoch": 0.3918965657552083,
      "grad_norm": 11.764037132263184,
      "learning_rate": 3.3373988910954525e-06,
      "loss": 3.6333,
      "step": 38525
    },
    {
      "epoch": 0.3919474283854167,
      "grad_norm": 13.153989791870117,
      "learning_rate": 3.337022099561251e-06,
      "loss": 3.3329,
      "step": 38530
    },
    {
      "epoch": 0.391998291015625,
      "grad_norm": 13.215889930725098,
      "learning_rate": 3.336645286612038e-06,
      "loss": 3.3511,
      "step": 38535
    },
    {
      "epoch": 0.3920491536458333,
      "grad_norm": 10.890106201171875,
      "learning_rate": 3.3362684522574536e-06,
      "loss": 3.1375,
      "step": 38540
    },
    {
      "epoch": 0.3921000162760417,
      "grad_norm": 12.711638450622559,
      "learning_rate": 3.3358915965071387e-06,
      "loss": 3.1453,
      "step": 38545
    },
    {
      "epoch": 0.39215087890625,
      "grad_norm": 13.96977424621582,
      "learning_rate": 3.3355147193707355e-06,
      "loss": 3.4909,
      "step": 38550
    },
    {
      "epoch": 0.3922017415364583,
      "grad_norm": 8.59538745880127,
      "learning_rate": 3.3351378208578864e-06,
      "loss": 3.1961,
      "step": 38555
    },
    {
      "epoch": 0.3922526041666667,
      "grad_norm": 14.170539855957031,
      "learning_rate": 3.334760900978234e-06,
      "loss": 3.1656,
      "step": 38560
    },
    {
      "epoch": 0.392303466796875,
      "grad_norm": 16.233667373657227,
      "learning_rate": 3.3343839597414216e-06,
      "loss": 3.0222,
      "step": 38565
    },
    {
      "epoch": 0.3923543294270833,
      "grad_norm": 13.056116104125977,
      "learning_rate": 3.3340069971570937e-06,
      "loss": 3.082,
      "step": 38570
    },
    {
      "epoch": 0.3924051920572917,
      "grad_norm": 10.929540634155273,
      "learning_rate": 3.333630013234894e-06,
      "loss": 3.2129,
      "step": 38575
    },
    {
      "epoch": 0.3924560546875,
      "grad_norm": 13.191286087036133,
      "learning_rate": 3.3332530079844694e-06,
      "loss": 3.2389,
      "step": 38580
    },
    {
      "epoch": 0.3925069173177083,
      "grad_norm": 12.56567096710205,
      "learning_rate": 3.3328759814154628e-06,
      "loss": 3.3224,
      "step": 38585
    },
    {
      "epoch": 0.3925577799479167,
      "grad_norm": 9.680841445922852,
      "learning_rate": 3.3324989335375226e-06,
      "loss": 3.4063,
      "step": 38590
    },
    {
      "epoch": 0.392608642578125,
      "grad_norm": 14.00575065612793,
      "learning_rate": 3.332121864360294e-06,
      "loss": 3.1624,
      "step": 38595
    },
    {
      "epoch": 0.3926595052083333,
      "grad_norm": 10.88829517364502,
      "learning_rate": 3.331744773893424e-06,
      "loss": 2.9704,
      "step": 38600
    },
    {
      "epoch": 0.3927103678385417,
      "grad_norm": 9.82624626159668,
      "learning_rate": 3.331367662146562e-06,
      "loss": 3.2943,
      "step": 38605
    },
    {
      "epoch": 0.39276123046875,
      "grad_norm": 9.934083938598633,
      "learning_rate": 3.330990529129355e-06,
      "loss": 3.341,
      "step": 38610
    },
    {
      "epoch": 0.3928120930989583,
      "grad_norm": 9.311182022094727,
      "learning_rate": 3.3306133748514523e-06,
      "loss": 3.3888,
      "step": 38615
    },
    {
      "epoch": 0.3928629557291667,
      "grad_norm": 12.890077590942383,
      "learning_rate": 3.3302361993225036e-06,
      "loss": 3.4668,
      "step": 38620
    },
    {
      "epoch": 0.392913818359375,
      "grad_norm": 11.402129173278809,
      "learning_rate": 3.3298590025521577e-06,
      "loss": 3.4034,
      "step": 38625
    },
    {
      "epoch": 0.3929646809895833,
      "grad_norm": 12.979676246643066,
      "learning_rate": 3.329481784550067e-06,
      "loss": 3.4632,
      "step": 38630
    },
    {
      "epoch": 0.3930155436197917,
      "grad_norm": 11.994227409362793,
      "learning_rate": 3.329104545325881e-06,
      "loss": 3.5591,
      "step": 38635
    },
    {
      "epoch": 0.39306640625,
      "grad_norm": 8.088932037353516,
      "learning_rate": 3.3287272848892515e-06,
      "loss": 3.4373,
      "step": 38640
    },
    {
      "epoch": 0.3931172688802083,
      "grad_norm": 9.991019248962402,
      "learning_rate": 3.3283500032498313e-06,
      "loss": 3.1668,
      "step": 38645
    },
    {
      "epoch": 0.3931681315104167,
      "grad_norm": 8.292196273803711,
      "learning_rate": 3.3279727004172724e-06,
      "loss": 3.2774,
      "step": 38650
    },
    {
      "epoch": 0.393218994140625,
      "grad_norm": 14.125115394592285,
      "learning_rate": 3.327595376401228e-06,
      "loss": 3.1344,
      "step": 38655
    },
    {
      "epoch": 0.3932698567708333,
      "grad_norm": 16.854019165039062,
      "learning_rate": 3.327218031211352e-06,
      "loss": 4.2892,
      "step": 38660
    },
    {
      "epoch": 0.3933207194010417,
      "grad_norm": 16.030033111572266,
      "learning_rate": 3.326840664857299e-06,
      "loss": 3.4319,
      "step": 38665
    },
    {
      "epoch": 0.39337158203125,
      "grad_norm": 11.797090530395508,
      "learning_rate": 3.326463277348724e-06,
      "loss": 3.4787,
      "step": 38670
    },
    {
      "epoch": 0.3934224446614583,
      "grad_norm": 15.557062149047852,
      "learning_rate": 3.3260858686952813e-06,
      "loss": 3.2515,
      "step": 38675
    },
    {
      "epoch": 0.3934733072916667,
      "grad_norm": 11.602935791015625,
      "learning_rate": 3.3257084389066273e-06,
      "loss": 3.1548,
      "step": 38680
    },
    {
      "epoch": 0.393524169921875,
      "grad_norm": 11.777176856994629,
      "learning_rate": 3.3253309879924194e-06,
      "loss": 3.424,
      "step": 38685
    },
    {
      "epoch": 0.3935750325520833,
      "grad_norm": 11.584993362426758,
      "learning_rate": 3.3249535159623133e-06,
      "loss": 3.413,
      "step": 38690
    },
    {
      "epoch": 0.3936258951822917,
      "grad_norm": 13.79917049407959,
      "learning_rate": 3.324576022825967e-06,
      "loss": 3.3521,
      "step": 38695
    },
    {
      "epoch": 0.3936767578125,
      "grad_norm": 10.56005573272705,
      "learning_rate": 3.324198508593039e-06,
      "loss": 3.0865,
      "step": 38700
    },
    {
      "epoch": 0.3937276204427083,
      "grad_norm": 17.659931182861328,
      "learning_rate": 3.3238209732731868e-06,
      "loss": 3.2109,
      "step": 38705
    },
    {
      "epoch": 0.3937784830729167,
      "grad_norm": 11.138640403747559,
      "learning_rate": 3.323443416876071e-06,
      "loss": 3.1011,
      "step": 38710
    },
    {
      "epoch": 0.393829345703125,
      "grad_norm": 9.119121551513672,
      "learning_rate": 3.3230658394113504e-06,
      "loss": 3.2088,
      "step": 38715
    },
    {
      "epoch": 0.3938802083333333,
      "grad_norm": 10.155940055847168,
      "learning_rate": 3.3226882408886846e-06,
      "loss": 3.3344,
      "step": 38720
    },
    {
      "epoch": 0.3939310709635417,
      "grad_norm": 11.100317001342773,
      "learning_rate": 3.322310621317736e-06,
      "loss": 3.4379,
      "step": 38725
    },
    {
      "epoch": 0.39398193359375,
      "grad_norm": 11.67934799194336,
      "learning_rate": 3.3219329807081645e-06,
      "loss": 3.1253,
      "step": 38730
    },
    {
      "epoch": 0.3940327962239583,
      "grad_norm": 11.354928970336914,
      "learning_rate": 3.3215553190696336e-06,
      "loss": 2.928,
      "step": 38735
    },
    {
      "epoch": 0.3940836588541667,
      "grad_norm": 9.212844848632812,
      "learning_rate": 3.321177636411803e-06,
      "loss": 3.7323,
      "step": 38740
    },
    {
      "epoch": 0.394134521484375,
      "grad_norm": 14.865660667419434,
      "learning_rate": 3.3207999327443385e-06,
      "loss": 3.1197,
      "step": 38745
    },
    {
      "epoch": 0.3941853841145833,
      "grad_norm": 15.50028133392334,
      "learning_rate": 3.320422208076902e-06,
      "loss": 3.1172,
      "step": 38750
    },
    {
      "epoch": 0.3942362467447917,
      "grad_norm": 10.393682479858398,
      "learning_rate": 3.320044462419157e-06,
      "loss": 3.2642,
      "step": 38755
    },
    {
      "epoch": 0.394287109375,
      "grad_norm": 12.6064453125,
      "learning_rate": 3.31966669578077e-06,
      "loss": 3.602,
      "step": 38760
    },
    {
      "epoch": 0.3943379720052083,
      "grad_norm": 15.318078994750977,
      "learning_rate": 3.319288908171404e-06,
      "loss": 3.5894,
      "step": 38765
    },
    {
      "epoch": 0.3943888346354167,
      "grad_norm": 15.770649909973145,
      "learning_rate": 3.3189110996007257e-06,
      "loss": 3.8763,
      "step": 38770
    },
    {
      "epoch": 0.394439697265625,
      "grad_norm": 8.988329887390137,
      "learning_rate": 3.318533270078401e-06,
      "loss": 3.4983,
      "step": 38775
    },
    {
      "epoch": 0.3944905598958333,
      "grad_norm": 18.160795211791992,
      "learning_rate": 3.3181554196140964e-06,
      "loss": 3.348,
      "step": 38780
    },
    {
      "epoch": 0.3945414225260417,
      "grad_norm": 13.910605430603027,
      "learning_rate": 3.3177775482174796e-06,
      "loss": 3.4063,
      "step": 38785
    },
    {
      "epoch": 0.39459228515625,
      "grad_norm": 14.50535774230957,
      "learning_rate": 3.3173996558982186e-06,
      "loss": 3.5568,
      "step": 38790
    },
    {
      "epoch": 0.3946431477864583,
      "grad_norm": 10.57275676727295,
      "learning_rate": 3.31702174266598e-06,
      "loss": 3.222,
      "step": 38795
    },
    {
      "epoch": 0.3946940104166667,
      "grad_norm": 9.792989730834961,
      "learning_rate": 3.3166438085304347e-06,
      "loss": 3.3537,
      "step": 38800
    },
    {
      "epoch": 0.394744873046875,
      "grad_norm": 15.056622505187988,
      "learning_rate": 3.31626585350125e-06,
      "loss": 3.378,
      "step": 38805
    },
    {
      "epoch": 0.3947957356770833,
      "grad_norm": 10.857216835021973,
      "learning_rate": 3.315887877588098e-06,
      "loss": 3.4722,
      "step": 38810
    },
    {
      "epoch": 0.3948465983072917,
      "grad_norm": 11.502994537353516,
      "learning_rate": 3.3155098808006483e-06,
      "loss": 3.6144,
      "step": 38815
    },
    {
      "epoch": 0.3948974609375,
      "grad_norm": 14.918729782104492,
      "learning_rate": 3.3151318631485707e-06,
      "loss": 3.3524,
      "step": 38820
    },
    {
      "epoch": 0.3949483235677083,
      "grad_norm": 11.880370140075684,
      "learning_rate": 3.314753824641538e-06,
      "loss": 3.5212,
      "step": 38825
    },
    {
      "epoch": 0.3949991861979167,
      "grad_norm": 12.545670509338379,
      "learning_rate": 3.3143757652892216e-06,
      "loss": 3.5019,
      "step": 38830
    },
    {
      "epoch": 0.395050048828125,
      "grad_norm": 13.899107933044434,
      "learning_rate": 3.313997685101294e-06,
      "loss": 3.0761,
      "step": 38835
    },
    {
      "epoch": 0.3951009114583333,
      "grad_norm": 14.468853950500488,
      "learning_rate": 3.3136195840874297e-06,
      "loss": 3.463,
      "step": 38840
    },
    {
      "epoch": 0.3951517740885417,
      "grad_norm": 9.901703834533691,
      "learning_rate": 3.3132414622572995e-06,
      "loss": 3.1865,
      "step": 38845
    },
    {
      "epoch": 0.39520263671875,
      "grad_norm": 14.128046035766602,
      "learning_rate": 3.3128633196205805e-06,
      "loss": 3.4865,
      "step": 38850
    },
    {
      "epoch": 0.3952534993489583,
      "grad_norm": 8.942483901977539,
      "learning_rate": 3.3124851561869465e-06,
      "loss": 3.3604,
      "step": 38855
    },
    {
      "epoch": 0.3953043619791667,
      "grad_norm": 12.734345436096191,
      "learning_rate": 3.3121069719660707e-06,
      "loss": 3.6297,
      "step": 38860
    },
    {
      "epoch": 0.395355224609375,
      "grad_norm": 12.513358116149902,
      "learning_rate": 3.3117287669676317e-06,
      "loss": 3.4542,
      "step": 38865
    },
    {
      "epoch": 0.3954060872395833,
      "grad_norm": 8.752370834350586,
      "learning_rate": 3.311350541201304e-06,
      "loss": 3.3082,
      "step": 38870
    },
    {
      "epoch": 0.3954569498697917,
      "grad_norm": 9.757655143737793,
      "learning_rate": 3.310972294676766e-06,
      "loss": 3.2355,
      "step": 38875
    },
    {
      "epoch": 0.3955078125,
      "grad_norm": 10.152702331542969,
      "learning_rate": 3.3105940274036935e-06,
      "loss": 3.3853,
      "step": 38880
    },
    {
      "epoch": 0.3955586751302083,
      "grad_norm": 8.71654224395752,
      "learning_rate": 3.310215739391765e-06,
      "loss": 3.4702,
      "step": 38885
    },
    {
      "epoch": 0.3956095377604167,
      "grad_norm": 11.565936088562012,
      "learning_rate": 3.3098374306506586e-06,
      "loss": 3.125,
      "step": 38890
    },
    {
      "epoch": 0.395660400390625,
      "grad_norm": 13.354435920715332,
      "learning_rate": 3.3094591011900535e-06,
      "loss": 3.4371,
      "step": 38895
    },
    {
      "epoch": 0.3957112630208333,
      "grad_norm": 8.289600372314453,
      "learning_rate": 3.3090807510196295e-06,
      "loss": 3.2972,
      "step": 38900
    },
    {
      "epoch": 0.3957621256510417,
      "grad_norm": 12.343297958374023,
      "learning_rate": 3.3087023801490663e-06,
      "loss": 3.1386,
      "step": 38905
    },
    {
      "epoch": 0.39581298828125,
      "grad_norm": 16.656522750854492,
      "learning_rate": 3.3083239885880437e-06,
      "loss": 3.5522,
      "step": 38910
    },
    {
      "epoch": 0.3958638509114583,
      "grad_norm": 16.437965393066406,
      "learning_rate": 3.307945576346244e-06,
      "loss": 3.3625,
      "step": 38915
    },
    {
      "epoch": 0.3959147135416667,
      "grad_norm": 12.688868522644043,
      "learning_rate": 3.3075671434333482e-06,
      "loss": 3.1965,
      "step": 38920
    },
    {
      "epoch": 0.395965576171875,
      "grad_norm": 13.657536506652832,
      "learning_rate": 3.307188689859038e-06,
      "loss": 3.2702,
      "step": 38925
    },
    {
      "epoch": 0.3960164388020833,
      "grad_norm": 13.055109024047852,
      "learning_rate": 3.306810215632997e-06,
      "loss": 3.6169,
      "step": 38930
    },
    {
      "epoch": 0.3960673014322917,
      "grad_norm": 18.273826599121094,
      "learning_rate": 3.3064317207649077e-06,
      "loss": 3.5969,
      "step": 38935
    },
    {
      "epoch": 0.3961181640625,
      "grad_norm": 15.294970512390137,
      "learning_rate": 3.3060532052644547e-06,
      "loss": 3.4694,
      "step": 38940
    },
    {
      "epoch": 0.3961690266927083,
      "grad_norm": 12.5772705078125,
      "learning_rate": 3.305674669141321e-06,
      "loss": 3.4484,
      "step": 38945
    },
    {
      "epoch": 0.3962198893229167,
      "grad_norm": 9.832720756530762,
      "learning_rate": 3.3052961124051907e-06,
      "loss": 3.0731,
      "step": 38950
    },
    {
      "epoch": 0.396270751953125,
      "grad_norm": 12.782217979431152,
      "learning_rate": 3.3049175350657514e-06,
      "loss": 3.1778,
      "step": 38955
    },
    {
      "epoch": 0.3963216145833333,
      "grad_norm": 12.149210929870605,
      "learning_rate": 3.3045389371326873e-06,
      "loss": 3.3827,
      "step": 38960
    },
    {
      "epoch": 0.3963724772135417,
      "grad_norm": 12.495834350585938,
      "learning_rate": 3.3041603186156857e-06,
      "loss": 3.4491,
      "step": 38965
    },
    {
      "epoch": 0.39642333984375,
      "grad_norm": 13.705297470092773,
      "learning_rate": 3.303781679524433e-06,
      "loss": 3.9191,
      "step": 38970
    },
    {
      "epoch": 0.3964742024739583,
      "grad_norm": 12.02988052368164,
      "learning_rate": 3.303403019868616e-06,
      "loss": 3.2583,
      "step": 38975
    },
    {
      "epoch": 0.3965250651041667,
      "grad_norm": 12.529438018798828,
      "learning_rate": 3.303024339657924e-06,
      "loss": 3.1721,
      "step": 38980
    },
    {
      "epoch": 0.396575927734375,
      "grad_norm": 15.870917320251465,
      "learning_rate": 3.302645638902043e-06,
      "loss": 3.3295,
      "step": 38985
    },
    {
      "epoch": 0.3966267903645833,
      "grad_norm": 13.028214454650879,
      "learning_rate": 3.3022669176106653e-06,
      "loss": 3.4199,
      "step": 38990
    },
    {
      "epoch": 0.3966776529947917,
      "grad_norm": 10.765739440917969,
      "learning_rate": 3.3018881757934777e-06,
      "loss": 3.3773,
      "step": 38995
    },
    {
      "epoch": 0.396728515625,
      "grad_norm": 9.587928771972656,
      "learning_rate": 3.3015094134601716e-06,
      "loss": 3.2503,
      "step": 39000
    },
    {
      "epoch": 0.3967793782552083,
      "grad_norm": 14.970799446105957,
      "learning_rate": 3.301130630620437e-06,
      "loss": 3.3718,
      "step": 39005
    },
    {
      "epoch": 0.3968302408854167,
      "grad_norm": 11.802124977111816,
      "learning_rate": 3.3007518272839658e-06,
      "loss": 3.3806,
      "step": 39010
    },
    {
      "epoch": 0.396881103515625,
      "grad_norm": 11.562552452087402,
      "learning_rate": 3.3003730034604477e-06,
      "loss": 3.2105,
      "step": 39015
    },
    {
      "epoch": 0.3969319661458333,
      "grad_norm": 13.692420959472656,
      "learning_rate": 3.299994159159576e-06,
      "loss": 3.4629,
      "step": 39020
    },
    {
      "epoch": 0.3969828287760417,
      "grad_norm": 13.56282901763916,
      "learning_rate": 3.299615294391044e-06,
      "loss": 3.5262,
      "step": 39025
    },
    {
      "epoch": 0.39703369140625,
      "grad_norm": 17.018476486206055,
      "learning_rate": 3.2992364091645446e-06,
      "loss": 3.5014,
      "step": 39030
    },
    {
      "epoch": 0.3970845540364583,
      "grad_norm": 12.96693229675293,
      "learning_rate": 3.298857503489771e-06,
      "loss": 3.194,
      "step": 39035
    },
    {
      "epoch": 0.3971354166666667,
      "grad_norm": 13.554923057556152,
      "learning_rate": 3.2984785773764167e-06,
      "loss": 3.2144,
      "step": 39040
    },
    {
      "epoch": 0.397186279296875,
      "grad_norm": 10.257043838500977,
      "learning_rate": 3.2980996308341775e-06,
      "loss": 3.3742,
      "step": 39045
    },
    {
      "epoch": 0.3972371419270833,
      "grad_norm": 9.593994140625,
      "learning_rate": 3.2977206638727487e-06,
      "loss": 3.6419,
      "step": 39050
    },
    {
      "epoch": 0.3972880045572917,
      "grad_norm": 6.843266487121582,
      "learning_rate": 3.2973416765018257e-06,
      "loss": 3.053,
      "step": 39055
    },
    {
      "epoch": 0.3973388671875,
      "grad_norm": 7.151871204376221,
      "learning_rate": 3.2969626687311047e-06,
      "loss": 3.7464,
      "step": 39060
    },
    {
      "epoch": 0.3973897298177083,
      "grad_norm": 9.020061492919922,
      "learning_rate": 3.296583640570283e-06,
      "loss": 3.4251,
      "step": 39065
    },
    {
      "epoch": 0.3974405924479167,
      "grad_norm": 10.482080459594727,
      "learning_rate": 3.2962045920290575e-06,
      "loss": 3.5422,
      "step": 39070
    },
    {
      "epoch": 0.397491455078125,
      "grad_norm": 10.591201782226562,
      "learning_rate": 3.2958255231171267e-06,
      "loss": 3.4915,
      "step": 39075
    },
    {
      "epoch": 0.3975423177083333,
      "grad_norm": 15.000960350036621,
      "learning_rate": 3.2954464338441874e-06,
      "loss": 3.0752,
      "step": 39080
    },
    {
      "epoch": 0.3975931803385417,
      "grad_norm": 9.86584186553955,
      "learning_rate": 3.295067324219941e-06,
      "loss": 3.3665,
      "step": 39085
    },
    {
      "epoch": 0.39764404296875,
      "grad_norm": 15.093664169311523,
      "learning_rate": 3.2946881942540846e-06,
      "loss": 3.1153,
      "step": 39090
    },
    {
      "epoch": 0.3976949055989583,
      "grad_norm": 13.898468971252441,
      "learning_rate": 3.2943090439563195e-06,
      "loss": 3.3152,
      "step": 39095
    },
    {
      "epoch": 0.3977457682291667,
      "grad_norm": 9.757120132446289,
      "learning_rate": 3.2939298733363455e-06,
      "loss": 3.2708,
      "step": 39100
    },
    {
      "epoch": 0.397796630859375,
      "grad_norm": 7.900053977966309,
      "learning_rate": 3.293550682403864e-06,
      "loss": 3.2929,
      "step": 39105
    },
    {
      "epoch": 0.3978474934895833,
      "grad_norm": 11.337242126464844,
      "learning_rate": 3.293171471168577e-06,
      "loss": 3.3876,
      "step": 39110
    },
    {
      "epoch": 0.3978983561197917,
      "grad_norm": 12.0416259765625,
      "learning_rate": 3.2927922396401853e-06,
      "loss": 3.1044,
      "step": 39115
    },
    {
      "epoch": 0.39794921875,
      "grad_norm": 7.893319606781006,
      "learning_rate": 3.2924129878283917e-06,
      "loss": 3.9901,
      "step": 39120
    },
    {
      "epoch": 0.3980000813802083,
      "grad_norm": 15.375621795654297,
      "learning_rate": 3.2920337157429004e-06,
      "loss": 3.1633,
      "step": 39125
    },
    {
      "epoch": 0.3980509440104167,
      "grad_norm": 7.960724830627441,
      "learning_rate": 3.291654423393414e-06,
      "loss": 3.4319,
      "step": 39130
    },
    {
      "epoch": 0.398101806640625,
      "grad_norm": 15.423291206359863,
      "learning_rate": 3.291275110789637e-06,
      "loss": 3.1555,
      "step": 39135
    },
    {
      "epoch": 0.3981526692708333,
      "grad_norm": 14.006745338439941,
      "learning_rate": 3.290895777941273e-06,
      "loss": 3.1331,
      "step": 39140
    },
    {
      "epoch": 0.3982035319010417,
      "grad_norm": 13.815712928771973,
      "learning_rate": 3.2905164248580287e-06,
      "loss": 3.5096,
      "step": 39145
    },
    {
      "epoch": 0.39825439453125,
      "grad_norm": 12.66348934173584,
      "learning_rate": 3.290137051549609e-06,
      "loss": 2.9672,
      "step": 39150
    },
    {
      "epoch": 0.3983052571614583,
      "grad_norm": 8.546638488769531,
      "learning_rate": 3.28975765802572e-06,
      "loss": 3.4663,
      "step": 39155
    },
    {
      "epoch": 0.3983561197916667,
      "grad_norm": 13.081283569335938,
      "learning_rate": 3.2893782442960686e-06,
      "loss": 3.091,
      "step": 39160
    },
    {
      "epoch": 0.398406982421875,
      "grad_norm": 9.0906343460083,
      "learning_rate": 3.2889988103703617e-06,
      "loss": 3.0275,
      "step": 39165
    },
    {
      "epoch": 0.3984578450520833,
      "grad_norm": 9.618579864501953,
      "learning_rate": 3.2886193562583074e-06,
      "loss": 3.4742,
      "step": 39170
    },
    {
      "epoch": 0.3985087076822917,
      "grad_norm": 11.26836109161377,
      "learning_rate": 3.2882398819696137e-06,
      "loss": 3.328,
      "step": 39175
    },
    {
      "epoch": 0.3985595703125,
      "grad_norm": 12.071304321289062,
      "learning_rate": 3.2878603875139896e-06,
      "loss": 3.5502,
      "step": 39180
    },
    {
      "epoch": 0.3986104329427083,
      "grad_norm": 11.873283386230469,
      "learning_rate": 3.2874808729011443e-06,
      "loss": 3.2419,
      "step": 39185
    },
    {
      "epoch": 0.3986612955729167,
      "grad_norm": 12.063312530517578,
      "learning_rate": 3.2871013381407877e-06,
      "loss": 3.1923,
      "step": 39190
    },
    {
      "epoch": 0.398712158203125,
      "grad_norm": 15.63031005859375,
      "learning_rate": 3.286721783242629e-06,
      "loss": 3.3086,
      "step": 39195
    },
    {
      "epoch": 0.3987630208333333,
      "grad_norm": 15.412569046020508,
      "learning_rate": 3.286342208216381e-06,
      "loss": 3.2832,
      "step": 39200
    },
    {
      "epoch": 0.3988138834635417,
      "grad_norm": 10.493462562561035,
      "learning_rate": 3.2859626130717536e-06,
      "loss": 3.7159,
      "step": 39205
    },
    {
      "epoch": 0.39886474609375,
      "grad_norm": 11.053747177124023,
      "learning_rate": 3.2855829978184587e-06,
      "loss": 3.5785,
      "step": 39210
    },
    {
      "epoch": 0.3989156087239583,
      "grad_norm": 12.584538459777832,
      "learning_rate": 3.2852033624662095e-06,
      "loss": 3.4814,
      "step": 39215
    },
    {
      "epoch": 0.3989664713541667,
      "grad_norm": 13.197242736816406,
      "learning_rate": 3.284823707024718e-06,
      "loss": 3.036,
      "step": 39220
    },
    {
      "epoch": 0.399017333984375,
      "grad_norm": 11.060696601867676,
      "learning_rate": 3.284444031503698e-06,
      "loss": 3.4832,
      "step": 39225
    },
    {
      "epoch": 0.3990681966145833,
      "grad_norm": 11.496437072753906,
      "learning_rate": 3.2840643359128643e-06,
      "loss": 3.4278,
      "step": 39230
    },
    {
      "epoch": 0.3991190592447917,
      "grad_norm": 15.989913940429688,
      "learning_rate": 3.283684620261929e-06,
      "loss": 3.6831,
      "step": 39235
    },
    {
      "epoch": 0.399169921875,
      "grad_norm": 12.756202697753906,
      "learning_rate": 3.2833048845606095e-06,
      "loss": 3.4123,
      "step": 39240
    },
    {
      "epoch": 0.3992207845052083,
      "grad_norm": 12.794174194335938,
      "learning_rate": 3.2829251288186197e-06,
      "loss": 3.0151,
      "step": 39245
    },
    {
      "epoch": 0.3992716471354167,
      "grad_norm": 9.944336891174316,
      "learning_rate": 3.282545353045676e-06,
      "loss": 3.6058,
      "step": 39250
    },
    {
      "epoch": 0.399322509765625,
      "grad_norm": 12.509191513061523,
      "learning_rate": 3.2821655572514955e-06,
      "loss": 3.5381,
      "step": 39255
    },
    {
      "epoch": 0.3993733723958333,
      "grad_norm": 7.6840009689331055,
      "learning_rate": 3.2817857414457934e-06,
      "loss": 3.1112,
      "step": 39260
    },
    {
      "epoch": 0.3994242350260417,
      "grad_norm": 12.166910171508789,
      "learning_rate": 3.2814059056382895e-06,
      "loss": 3.7905,
      "step": 39265
    },
    {
      "epoch": 0.39947509765625,
      "grad_norm": 11.381367683410645,
      "learning_rate": 3.2810260498387002e-06,
      "loss": 3.2597,
      "step": 39270
    },
    {
      "epoch": 0.3995259602864583,
      "grad_norm": 9.968490600585938,
      "learning_rate": 3.2806461740567453e-06,
      "loss": 3.4514,
      "step": 39275
    },
    {
      "epoch": 0.3995768229166667,
      "grad_norm": 14.34542179107666,
      "learning_rate": 3.280266278302142e-06,
      "loss": 3.2793,
      "step": 39280
    },
    {
      "epoch": 0.399627685546875,
      "grad_norm": 11.94149398803711,
      "learning_rate": 3.2798863625846116e-06,
      "loss": 3.2478,
      "step": 39285
    },
    {
      "epoch": 0.3996785481770833,
      "grad_norm": 15.4173583984375,
      "learning_rate": 3.279506426913873e-06,
      "loss": 3.3845,
      "step": 39290
    },
    {
      "epoch": 0.3997294108072917,
      "grad_norm": 11.536433219909668,
      "learning_rate": 3.2791264712996486e-06,
      "loss": 3.4988,
      "step": 39295
    },
    {
      "epoch": 0.3997802734375,
      "grad_norm": 15.47469425201416,
      "learning_rate": 3.2787464957516556e-06,
      "loss": 3.614,
      "step": 39300
    },
    {
      "epoch": 0.3998311360677083,
      "grad_norm": 13.07068920135498,
      "learning_rate": 3.2783665002796207e-06,
      "loss": 3.2102,
      "step": 39305
    },
    {
      "epoch": 0.3998819986979167,
      "grad_norm": 11.186180114746094,
      "learning_rate": 3.2779864848932618e-06,
      "loss": 3.178,
      "step": 39310
    },
    {
      "epoch": 0.399932861328125,
      "grad_norm": 14.144645690917969,
      "learning_rate": 3.2776064496023037e-06,
      "loss": 3.5366,
      "step": 39315
    },
    {
      "epoch": 0.3999837239583333,
      "grad_norm": 13.992406845092773,
      "learning_rate": 3.2772263944164683e-06,
      "loss": 3.2436,
      "step": 39320
    },
    {
      "epoch": 0.4000345865885417,
      "grad_norm": 17.365379333496094,
      "learning_rate": 3.2768463193454803e-06,
      "loss": 3.3642,
      "step": 39325
    },
    {
      "epoch": 0.40008544921875,
      "grad_norm": 9.210827827453613,
      "learning_rate": 3.2764662243990632e-06,
      "loss": 3.5888,
      "step": 39330
    },
    {
      "epoch": 0.4001363118489583,
      "grad_norm": 14.269044876098633,
      "learning_rate": 3.276086109586942e-06,
      "loss": 3.056,
      "step": 39335
    },
    {
      "epoch": 0.4001871744791667,
      "grad_norm": 13.896188735961914,
      "learning_rate": 3.2757059749188415e-06,
      "loss": 3.7036,
      "step": 39340
    },
    {
      "epoch": 0.400238037109375,
      "grad_norm": 12.715867042541504,
      "learning_rate": 3.2753258204044873e-06,
      "loss": 3.6218,
      "step": 39345
    },
    {
      "epoch": 0.4002888997395833,
      "grad_norm": 8.956153869628906,
      "learning_rate": 3.2749456460536057e-06,
      "loss": 3.3922,
      "step": 39350
    },
    {
      "epoch": 0.4003397623697917,
      "grad_norm": 14.240581512451172,
      "learning_rate": 3.274565451875924e-06,
      "loss": 3.332,
      "step": 39355
    },
    {
      "epoch": 0.400390625,
      "grad_norm": 15.196061134338379,
      "learning_rate": 3.274185237881169e-06,
      "loss": 3.2174,
      "step": 39360
    },
    {
      "epoch": 0.4004414876302083,
      "grad_norm": 17.01357650756836,
      "learning_rate": 3.273805004079067e-06,
      "loss": 3.5295,
      "step": 39365
    },
    {
      "epoch": 0.4004923502604167,
      "grad_norm": 11.21593189239502,
      "learning_rate": 3.2734247504793483e-06,
      "loss": 3.1045,
      "step": 39370
    },
    {
      "epoch": 0.400543212890625,
      "grad_norm": 9.288373947143555,
      "learning_rate": 3.27304447709174e-06,
      "loss": 3.4651,
      "step": 39375
    },
    {
      "epoch": 0.4005940755208333,
      "grad_norm": 7.687575340270996,
      "learning_rate": 3.272664183925973e-06,
      "loss": 3.4994,
      "step": 39380
    },
    {
      "epoch": 0.4006449381510417,
      "grad_norm": 13.036829948425293,
      "learning_rate": 3.2722838709917753e-06,
      "loss": 3.1886,
      "step": 39385
    },
    {
      "epoch": 0.40069580078125,
      "grad_norm": 8.436564445495605,
      "learning_rate": 3.271903538298878e-06,
      "loss": 3.1957,
      "step": 39390
    },
    {
      "epoch": 0.4007466634114583,
      "grad_norm": 11.310314178466797,
      "learning_rate": 3.271523185857011e-06,
      "loss": 3.4303,
      "step": 39395
    },
    {
      "epoch": 0.4007975260416667,
      "grad_norm": 14.260161399841309,
      "learning_rate": 3.271142813675908e-06,
      "loss": 3.6256,
      "step": 39400
    },
    {
      "epoch": 0.400848388671875,
      "grad_norm": 9.424277305603027,
      "learning_rate": 3.270762421765297e-06,
      "loss": 3.5836,
      "step": 39405
    },
    {
      "epoch": 0.4008992513020833,
      "grad_norm": 6.394208908081055,
      "learning_rate": 3.2703820101349127e-06,
      "loss": 3.3019,
      "step": 39410
    },
    {
      "epoch": 0.4009501139322917,
      "grad_norm": 14.603628158569336,
      "learning_rate": 3.2700015787944865e-06,
      "loss": 2.9622,
      "step": 39415
    },
    {
      "epoch": 0.4010009765625,
      "grad_norm": 10.907510757446289,
      "learning_rate": 3.2696211277537535e-06,
      "loss": 3.8089,
      "step": 39420
    },
    {
      "epoch": 0.4010518391927083,
      "grad_norm": 14.512213706970215,
      "learning_rate": 3.2692406570224468e-06,
      "loss": 3.1383,
      "step": 39425
    },
    {
      "epoch": 0.4011027018229167,
      "grad_norm": 13.748353958129883,
      "learning_rate": 3.2688601666102994e-06,
      "loss": 3.3665,
      "step": 39430
    },
    {
      "epoch": 0.401153564453125,
      "grad_norm": 14.942314147949219,
      "learning_rate": 3.2684796565270464e-06,
      "loss": 3.3018,
      "step": 39435
    },
    {
      "epoch": 0.4012044270833333,
      "grad_norm": 15.140890121459961,
      "learning_rate": 3.2680991267824248e-06,
      "loss": 3.581,
      "step": 39440
    },
    {
      "epoch": 0.4012552897135417,
      "grad_norm": 17.58057975769043,
      "learning_rate": 3.267718577386168e-06,
      "loss": 3.0269,
      "step": 39445
    },
    {
      "epoch": 0.40130615234375,
      "grad_norm": 10.288013458251953,
      "learning_rate": 3.267338008348014e-06,
      "loss": 3.4707,
      "step": 39450
    },
    {
      "epoch": 0.4013570149739583,
      "grad_norm": 15.400893211364746,
      "learning_rate": 3.266957419677699e-06,
      "loss": 3.4214,
      "step": 39455
    },
    {
      "epoch": 0.4014078776041667,
      "grad_norm": 13.282986640930176,
      "learning_rate": 3.2665768113849596e-06,
      "loss": 3.3755,
      "step": 39460
    },
    {
      "epoch": 0.401458740234375,
      "grad_norm": 15.696666717529297,
      "learning_rate": 3.2661961834795346e-06,
      "loss": 3.48,
      "step": 39465
    },
    {
      "epoch": 0.4015096028645833,
      "grad_norm": 9.888688087463379,
      "learning_rate": 3.2658155359711612e-06,
      "loss": 3.4776,
      "step": 39470
    },
    {
      "epoch": 0.4015604654947917,
      "grad_norm": 13.198280334472656,
      "learning_rate": 3.26543486886958e-06,
      "loss": 3.1052,
      "step": 39475
    },
    {
      "epoch": 0.401611328125,
      "grad_norm": 8.186661720275879,
      "learning_rate": 3.265054182184528e-06,
      "loss": 3.5331,
      "step": 39480
    },
    {
      "epoch": 0.4016621907552083,
      "grad_norm": 10.0850191116333,
      "learning_rate": 3.264673475925747e-06,
      "loss": 3.302,
      "step": 39485
    },
    {
      "epoch": 0.4017130533854167,
      "grad_norm": 9.62065601348877,
      "learning_rate": 3.264292750102976e-06,
      "loss": 2.9363,
      "step": 39490
    },
    {
      "epoch": 0.401763916015625,
      "grad_norm": 12.95029354095459,
      "learning_rate": 3.263912004725956e-06,
      "loss": 3.5954,
      "step": 39495
    },
    {
      "epoch": 0.4018147786458333,
      "grad_norm": 14.583548545837402,
      "learning_rate": 3.263531239804428e-06,
      "loss": 3.3027,
      "step": 39500
    },
    {
      "epoch": 0.4018656412760417,
      "grad_norm": 9.015926361083984,
      "learning_rate": 3.263150455348135e-06,
      "loss": 3.0706,
      "step": 39505
    },
    {
      "epoch": 0.40191650390625,
      "grad_norm": 14.476874351501465,
      "learning_rate": 3.2627696513668182e-06,
      "loss": 3.4707,
      "step": 39510
    },
    {
      "epoch": 0.4019673665364583,
      "grad_norm": 9.141477584838867,
      "learning_rate": 3.2623888278702214e-06,
      "loss": 3.8334,
      "step": 39515
    },
    {
      "epoch": 0.4020182291666667,
      "grad_norm": 11.498889923095703,
      "learning_rate": 3.262007984868086e-06,
      "loss": 3.3327,
      "step": 39520
    },
    {
      "epoch": 0.402069091796875,
      "grad_norm": 10.279153823852539,
      "learning_rate": 3.261627122370158e-06,
      "loss": 3.6671,
      "step": 39525
    },
    {
      "epoch": 0.4021199544270833,
      "grad_norm": 6.8709211349487305,
      "learning_rate": 3.26124624038618e-06,
      "loss": 3.0967,
      "step": 39530
    },
    {
      "epoch": 0.4021708170572917,
      "grad_norm": 10.445243835449219,
      "learning_rate": 3.260865338925897e-06,
      "loss": 3.2211,
      "step": 39535
    },
    {
      "epoch": 0.4022216796875,
      "grad_norm": 11.800230979919434,
      "learning_rate": 3.2604844179990547e-06,
      "loss": 3.1659,
      "step": 39540
    },
    {
      "epoch": 0.4022725423177083,
      "grad_norm": 8.571924209594727,
      "learning_rate": 3.2601034776153997e-06,
      "loss": 2.9351,
      "step": 39545
    },
    {
      "epoch": 0.4023234049479167,
      "grad_norm": 10.99386215209961,
      "learning_rate": 3.2597225177846762e-06,
      "loss": 3.0938,
      "step": 39550
    },
    {
      "epoch": 0.402374267578125,
      "grad_norm": 14.340719223022461,
      "learning_rate": 3.259341538516633e-06,
      "loss": 3.4287,
      "step": 39555
    },
    {
      "epoch": 0.4024251302083333,
      "grad_norm": 15.162818908691406,
      "learning_rate": 3.258960539821016e-06,
      "loss": 3.4843,
      "step": 39560
    },
    {
      "epoch": 0.4024759928385417,
      "grad_norm": 10.90530014038086,
      "learning_rate": 3.258579521707574e-06,
      "loss": 3.1177,
      "step": 39565
    },
    {
      "epoch": 0.40252685546875,
      "grad_norm": 15.255401611328125,
      "learning_rate": 3.258198484186054e-06,
      "loss": 3.0077,
      "step": 39570
    },
    {
      "epoch": 0.4025777180989583,
      "grad_norm": 15.258075714111328,
      "learning_rate": 3.257817427266206e-06,
      "loss": 3.1873,
      "step": 39575
    },
    {
      "epoch": 0.4026285807291667,
      "grad_norm": 11.011824607849121,
      "learning_rate": 3.2574363509577794e-06,
      "loss": 3.7077,
      "step": 39580
    },
    {
      "epoch": 0.402679443359375,
      "grad_norm": 12.97339153289795,
      "learning_rate": 3.257055255270522e-06,
      "loss": 3.1655,
      "step": 39585
    },
    {
      "epoch": 0.4027303059895833,
      "grad_norm": 10.90396499633789,
      "learning_rate": 3.2566741402141866e-06,
      "loss": 3.3544,
      "step": 39590
    },
    {
      "epoch": 0.4027811686197917,
      "grad_norm": 16.23613929748535,
      "learning_rate": 3.256293005798522e-06,
      "loss": 3.1684,
      "step": 39595
    },
    {
      "epoch": 0.40283203125,
      "grad_norm": 13.79626750946045,
      "learning_rate": 3.25591185203328e-06,
      "loss": 3.4085,
      "step": 39600
    },
    {
      "epoch": 0.4028828938802083,
      "grad_norm": 15.524015426635742,
      "learning_rate": 3.2555306789282127e-06,
      "loss": 3.5288,
      "step": 39605
    },
    {
      "epoch": 0.4029337565104167,
      "grad_norm": 12.625886917114258,
      "learning_rate": 3.2551494864930716e-06,
      "loss": 3.1274,
      "step": 39610
    },
    {
      "epoch": 0.402984619140625,
      "grad_norm": 10.071491241455078,
      "learning_rate": 3.2547682747376103e-06,
      "loss": 3.1743,
      "step": 39615
    },
    {
      "epoch": 0.4030354817708333,
      "grad_norm": 8.225013732910156,
      "learning_rate": 3.2543870436715818e-06,
      "loss": 3.1853,
      "step": 39620
    },
    {
      "epoch": 0.4030863444010417,
      "grad_norm": 14.447437286376953,
      "learning_rate": 3.2540057933047386e-06,
      "loss": 4.0105,
      "step": 39625
    },
    {
      "epoch": 0.40313720703125,
      "grad_norm": 12.400070190429688,
      "learning_rate": 3.2536245236468373e-06,
      "loss": 3.5334,
      "step": 39630
    },
    {
      "epoch": 0.4031880696614583,
      "grad_norm": 9.871600151062012,
      "learning_rate": 3.2532432347076303e-06,
      "loss": 3.0529,
      "step": 39635
    },
    {
      "epoch": 0.4032389322916667,
      "grad_norm": 10.247000694274902,
      "learning_rate": 3.2528619264968737e-06,
      "loss": 3.4275,
      "step": 39640
    },
    {
      "epoch": 0.403289794921875,
      "grad_norm": 15.366523742675781,
      "learning_rate": 3.2524805990243236e-06,
      "loss": 3.1253,
      "step": 39645
    },
    {
      "epoch": 0.4033406575520833,
      "grad_norm": 8.520768165588379,
      "learning_rate": 3.2520992522997353e-06,
      "loss": 3.5719,
      "step": 39650
    },
    {
      "epoch": 0.4033915201822917,
      "grad_norm": 10.238525390625,
      "learning_rate": 3.251717886332866e-06,
      "loss": 3.2692,
      "step": 39655
    },
    {
      "epoch": 0.4034423828125,
      "grad_norm": 16.172805786132812,
      "learning_rate": 3.251336501133473e-06,
      "loss": 3.3253,
      "step": 39660
    },
    {
      "epoch": 0.4034932454427083,
      "grad_norm": 14.514467239379883,
      "learning_rate": 3.2509550967113133e-06,
      "loss": 3.1995,
      "step": 39665
    },
    {
      "epoch": 0.4035441080729167,
      "grad_norm": 17.597890853881836,
      "learning_rate": 3.2505736730761457e-06,
      "loss": 3.3171,
      "step": 39670
    },
    {
      "epoch": 0.403594970703125,
      "grad_norm": 14.871464729309082,
      "learning_rate": 3.2501922302377285e-06,
      "loss": 3.131,
      "step": 39675
    },
    {
      "epoch": 0.4036458333333333,
      "grad_norm": 14.677927017211914,
      "learning_rate": 3.2498107682058213e-06,
      "loss": 3.705,
      "step": 39680
    },
    {
      "epoch": 0.4036966959635417,
      "grad_norm": 11.570945739746094,
      "learning_rate": 3.2494292869901835e-06,
      "loss": 3.2704,
      "step": 39685
    },
    {
      "epoch": 0.40374755859375,
      "grad_norm": 14.676580429077148,
      "learning_rate": 3.249047786600574e-06,
      "loss": 3.4076,
      "step": 39690
    },
    {
      "epoch": 0.4037984212239583,
      "grad_norm": 8.944365501403809,
      "learning_rate": 3.248666267046755e-06,
      "loss": 3.3502,
      "step": 39695
    },
    {
      "epoch": 0.4038492838541667,
      "grad_norm": 7.9353790283203125,
      "learning_rate": 3.2482847283384866e-06,
      "loss": 3.3113,
      "step": 39700
    },
    {
      "epoch": 0.403900146484375,
      "grad_norm": 16.29496955871582,
      "learning_rate": 3.2479031704855313e-06,
      "loss": 3.3095,
      "step": 39705
    },
    {
      "epoch": 0.4039510091145833,
      "grad_norm": 12.825359344482422,
      "learning_rate": 3.2475215934976506e-06,
      "loss": 3.4118,
      "step": 39710
    },
    {
      "epoch": 0.4040018717447917,
      "grad_norm": 8.6707124710083,
      "learning_rate": 3.2471399973846067e-06,
      "loss": 3.4168,
      "step": 39715
    },
    {
      "epoch": 0.404052734375,
      "grad_norm": 16.991374969482422,
      "learning_rate": 3.2467583821561632e-06,
      "loss": 3.3508,
      "step": 39720
    },
    {
      "epoch": 0.4041035970052083,
      "grad_norm": 10.363889694213867,
      "learning_rate": 3.2463767478220836e-06,
      "loss": 3.5395,
      "step": 39725
    },
    {
      "epoch": 0.4041544596354167,
      "grad_norm": 13.849224090576172,
      "learning_rate": 3.245995094392132e-06,
      "loss": 3.5952,
      "step": 39730
    },
    {
      "epoch": 0.404205322265625,
      "grad_norm": 11.806497573852539,
      "learning_rate": 3.2456134218760732e-06,
      "loss": 3.2266,
      "step": 39735
    },
    {
      "epoch": 0.4042561848958333,
      "grad_norm": 8.696374893188477,
      "learning_rate": 3.2452317302836705e-06,
      "loss": 3.0952,
      "step": 39740
    },
    {
      "epoch": 0.4043070475260417,
      "grad_norm": 8.538705825805664,
      "learning_rate": 3.244850019624691e-06,
      "loss": 3.3628,
      "step": 39745
    },
    {
      "epoch": 0.40435791015625,
      "grad_norm": 11.07223129272461,
      "learning_rate": 3.2444682899089007e-06,
      "loss": 3.0328,
      "step": 39750
    },
    {
      "epoch": 0.4044087727864583,
      "grad_norm": 7.9295125007629395,
      "learning_rate": 3.2440865411460655e-06,
      "loss": 3.476,
      "step": 39755
    },
    {
      "epoch": 0.4044596354166667,
      "grad_norm": 15.09688663482666,
      "learning_rate": 3.2437047733459526e-06,
      "loss": 3.7012,
      "step": 39760
    },
    {
      "epoch": 0.404510498046875,
      "grad_norm": 16.56294822692871,
      "learning_rate": 3.2433229865183296e-06,
      "loss": 3.7268,
      "step": 39765
    },
    {
      "epoch": 0.4045613606770833,
      "grad_norm": 15.655258178710938,
      "learning_rate": 3.2429411806729636e-06,
      "loss": 3.1404,
      "step": 39770
    },
    {
      "epoch": 0.4046122233072917,
      "grad_norm": 8.41606330871582,
      "learning_rate": 3.242559355819624e-06,
      "loss": 3.4145,
      "step": 39775
    },
    {
      "epoch": 0.4046630859375,
      "grad_norm": 13.415966987609863,
      "learning_rate": 3.2421775119680785e-06,
      "loss": 2.911,
      "step": 39780
    },
    {
      "epoch": 0.4047139485677083,
      "grad_norm": 16.03108024597168,
      "learning_rate": 3.2417956491280983e-06,
      "loss": 3.3638,
      "step": 39785
    },
    {
      "epoch": 0.4047648111979167,
      "grad_norm": 9.713606834411621,
      "learning_rate": 3.2414137673094525e-06,
      "loss": 3.5204,
      "step": 39790
    },
    {
      "epoch": 0.404815673828125,
      "grad_norm": 9.479934692382812,
      "learning_rate": 3.2410318665219097e-06,
      "loss": 3.909,
      "step": 39795
    },
    {
      "epoch": 0.4048665364583333,
      "grad_norm": 13.595467567443848,
      "learning_rate": 3.240649946775243e-06,
      "loss": 3.5359,
      "step": 39800
    },
    {
      "epoch": 0.4049173990885417,
      "grad_norm": 8.232876777648926,
      "learning_rate": 3.240268008079223e-06,
      "loss": 3.3631,
      "step": 39805
    },
    {
      "epoch": 0.40496826171875,
      "grad_norm": 14.731582641601562,
      "learning_rate": 3.2398860504436218e-06,
      "loss": 3.2081,
      "step": 39810
    },
    {
      "epoch": 0.4050191243489583,
      "grad_norm": 10.484471321105957,
      "learning_rate": 3.2395040738782113e-06,
      "loss": 3.629,
      "step": 39815
    },
    {
      "epoch": 0.4050699869791667,
      "grad_norm": 8.36217975616455,
      "learning_rate": 3.239122078392764e-06,
      "loss": 3.1228,
      "step": 39820
    },
    {
      "epoch": 0.405120849609375,
      "grad_norm": 11.761137962341309,
      "learning_rate": 3.2387400639970532e-06,
      "loss": 3.4016,
      "step": 39825
    },
    {
      "epoch": 0.4051717122395833,
      "grad_norm": 9.876433372497559,
      "learning_rate": 3.238358030700854e-06,
      "loss": 3.5512,
      "step": 39830
    },
    {
      "epoch": 0.4052225748697917,
      "grad_norm": 7.70481538772583,
      "learning_rate": 3.2379759785139386e-06,
      "loss": 3.2944,
      "step": 39835
    },
    {
      "epoch": 0.4052734375,
      "grad_norm": 12.468230247497559,
      "learning_rate": 3.2375939074460833e-06,
      "loss": 3.0154,
      "step": 39840
    },
    {
      "epoch": 0.4053243001302083,
      "grad_norm": 11.805959701538086,
      "learning_rate": 3.237211817507062e-06,
      "loss": 3.4327,
      "step": 39845
    },
    {
      "epoch": 0.4053751627604167,
      "grad_norm": 7.966749668121338,
      "learning_rate": 3.236829708706652e-06,
      "loss": 3.295,
      "step": 39850
    },
    {
      "epoch": 0.405426025390625,
      "grad_norm": 13.538468360900879,
      "learning_rate": 3.2364475810546287e-06,
      "loss": 3.0303,
      "step": 39855
    },
    {
      "epoch": 0.4054768880208333,
      "grad_norm": 11.909079551696777,
      "learning_rate": 3.236065434560767e-06,
      "loss": 3.2183,
      "step": 39860
    },
    {
      "epoch": 0.4055277506510417,
      "grad_norm": 12.326981544494629,
      "learning_rate": 3.235683269234847e-06,
      "loss": 3.4753,
      "step": 39865
    },
    {
      "epoch": 0.40557861328125,
      "grad_norm": 13.924688339233398,
      "learning_rate": 3.2353010850866445e-06,
      "loss": 3.3513,
      "step": 39870
    },
    {
      "epoch": 0.4056294759114583,
      "grad_norm": 7.7965087890625,
      "learning_rate": 3.234918882125939e-06,
      "loss": 3.1596,
      "step": 39875
    },
    {
      "epoch": 0.4056803385416667,
      "grad_norm": 13.77583122253418,
      "learning_rate": 3.2345366603625065e-06,
      "loss": 3.2544,
      "step": 39880
    },
    {
      "epoch": 0.405731201171875,
      "grad_norm": 8.31397819519043,
      "learning_rate": 3.2341544198061287e-06,
      "loss": 3.0558,
      "step": 39885
    },
    {
      "epoch": 0.4057820638020833,
      "grad_norm": 8.37823486328125,
      "learning_rate": 3.233772160466584e-06,
      "loss": 3.1776,
      "step": 39890
    },
    {
      "epoch": 0.4058329264322917,
      "grad_norm": 8.468817710876465,
      "learning_rate": 3.233389882353652e-06,
      "loss": 3.1737,
      "step": 39895
    },
    {
      "epoch": 0.4058837890625,
      "grad_norm": 12.608991622924805,
      "learning_rate": 3.233007585477114e-06,
      "loss": 3.1122,
      "step": 39900
    },
    {
      "epoch": 0.4059346516927083,
      "grad_norm": 15.017359733581543,
      "learning_rate": 3.2326252698467507e-06,
      "loss": 3.3513,
      "step": 39905
    },
    {
      "epoch": 0.4059855143229167,
      "grad_norm": 14.726430892944336,
      "learning_rate": 3.232242935472343e-06,
      "loss": 3.3759,
      "step": 39910
    },
    {
      "epoch": 0.406036376953125,
      "grad_norm": 11.463208198547363,
      "learning_rate": 3.2318605823636743e-06,
      "loss": 3.0833,
      "step": 39915
    },
    {
      "epoch": 0.4060872395833333,
      "grad_norm": 10.70964241027832,
      "learning_rate": 3.2314782105305254e-06,
      "loss": 3.7387,
      "step": 39920
    },
    {
      "epoch": 0.4061381022135417,
      "grad_norm": 11.310379028320312,
      "learning_rate": 3.2310958199826802e-06,
      "loss": 3.3598,
      "step": 39925
    },
    {
      "epoch": 0.40618896484375,
      "grad_norm": 11.360577583312988,
      "learning_rate": 3.2307134107299213e-06,
      "loss": 3.1917,
      "step": 39930
    },
    {
      "epoch": 0.4062398274739583,
      "grad_norm": 16.626344680786133,
      "learning_rate": 3.2303309827820335e-06,
      "loss": 3.6492,
      "step": 39935
    },
    {
      "epoch": 0.4062906901041667,
      "grad_norm": 12.946182250976562,
      "learning_rate": 3.2299485361488005e-06,
      "loss": 3.2241,
      "step": 39940
    },
    {
      "epoch": 0.406341552734375,
      "grad_norm": 12.617770195007324,
      "learning_rate": 3.2295660708400067e-06,
      "loss": 3.1731,
      "step": 39945
    },
    {
      "epoch": 0.4063924153645833,
      "grad_norm": 13.231428146362305,
      "learning_rate": 3.229183586865438e-06,
      "loss": 3.4243,
      "step": 39950
    },
    {
      "epoch": 0.4064432779947917,
      "grad_norm": 13.319751739501953,
      "learning_rate": 3.2288010842348806e-06,
      "loss": 3.5972,
      "step": 39955
    },
    {
      "epoch": 0.406494140625,
      "grad_norm": 16.305463790893555,
      "learning_rate": 3.22841856295812e-06,
      "loss": 3.3722,
      "step": 39960
    },
    {
      "epoch": 0.4065450032552083,
      "grad_norm": 13.032581329345703,
      "learning_rate": 3.2280360230449425e-06,
      "loss": 3.3068,
      "step": 39965
    },
    {
      "epoch": 0.4065958658854167,
      "grad_norm": 12.078779220581055,
      "learning_rate": 3.227653464505137e-06,
      "loss": 3.1496,
      "step": 39970
    },
    {
      "epoch": 0.406646728515625,
      "grad_norm": 8.394103050231934,
      "learning_rate": 3.2272708873484892e-06,
      "loss": 3.4932,
      "step": 39975
    },
    {
      "epoch": 0.4066975911458333,
      "grad_norm": 15.861308097839355,
      "learning_rate": 3.226888291584789e-06,
      "loss": 3.4615,
      "step": 39980
    },
    {
      "epoch": 0.4067484537760417,
      "grad_norm": 12.765704154968262,
      "learning_rate": 3.2265056772238235e-06,
      "loss": 3.5515,
      "step": 39985
    },
    {
      "epoch": 0.40679931640625,
      "grad_norm": 11.849873542785645,
      "learning_rate": 3.2261230442753828e-06,
      "loss": 3.1232,
      "step": 39990
    },
    {
      "epoch": 0.4068501790364583,
      "grad_norm": 8.603911399841309,
      "learning_rate": 3.225740392749256e-06,
      "loss": 3.1466,
      "step": 39995
    },
    {
      "epoch": 0.4069010416666667,
      "grad_norm": 15.368163108825684,
      "learning_rate": 3.225357722655233e-06,
      "loss": 3.3987,
      "step": 40000
    },
    {
      "epoch": 0.406951904296875,
      "grad_norm": 17.930904388427734,
      "learning_rate": 3.224975034003105e-06,
      "loss": 3.541,
      "step": 40005
    },
    {
      "epoch": 0.4070027669270833,
      "grad_norm": 11.624622344970703,
      "learning_rate": 3.2245923268026623e-06,
      "loss": 3.4463,
      "step": 40010
    },
    {
      "epoch": 0.4070536295572917,
      "grad_norm": 9.872394561767578,
      "learning_rate": 3.2242096010636964e-06,
      "loss": 3.3589,
      "step": 40015
    },
    {
      "epoch": 0.4071044921875,
      "grad_norm": 10.430153846740723,
      "learning_rate": 3.223826856796001e-06,
      "loss": 3.1212,
      "step": 40020
    },
    {
      "epoch": 0.4071553548177083,
      "grad_norm": 16.75178337097168,
      "learning_rate": 3.223444094009365e-06,
      "loss": 3.5168,
      "step": 40025
    },
    {
      "epoch": 0.4072062174479167,
      "grad_norm": 8.424890518188477,
      "learning_rate": 3.2230613127135846e-06,
      "loss": 3.3789,
      "step": 40030
    },
    {
      "epoch": 0.407257080078125,
      "grad_norm": 13.780265808105469,
      "learning_rate": 3.2226785129184514e-06,
      "loss": 3.329,
      "step": 40035
    },
    {
      "epoch": 0.4073079427083333,
      "grad_norm": 7.752483367919922,
      "learning_rate": 3.2222956946337593e-06,
      "loss": 3.1299,
      "step": 40040
    },
    {
      "epoch": 0.4073588053385417,
      "grad_norm": 13.269911766052246,
      "learning_rate": 3.221912857869304e-06,
      "loss": 3.4392,
      "step": 40045
    },
    {
      "epoch": 0.40740966796875,
      "grad_norm": 12.497602462768555,
      "learning_rate": 3.221530002634879e-06,
      "loss": 3.4714,
      "step": 40050
    },
    {
      "epoch": 0.4074605305989583,
      "grad_norm": 19.447315216064453,
      "learning_rate": 3.2211471289402797e-06,
      "loss": 3.0852,
      "step": 40055
    },
    {
      "epoch": 0.4075113932291667,
      "grad_norm": 9.918488502502441,
      "learning_rate": 3.2207642367953025e-06,
      "loss": 3.1742,
      "step": 40060
    },
    {
      "epoch": 0.407562255859375,
      "grad_norm": 9.899670600891113,
      "learning_rate": 3.2203813262097427e-06,
      "loss": 3.4072,
      "step": 40065
    },
    {
      "epoch": 0.4076131184895833,
      "grad_norm": 13.883380889892578,
      "learning_rate": 3.219998397193397e-06,
      "loss": 3.3758,
      "step": 40070
    },
    {
      "epoch": 0.4076639811197917,
      "grad_norm": 16.322729110717773,
      "learning_rate": 3.2196154497560642e-06,
      "loss": 3.201,
      "step": 40075
    },
    {
      "epoch": 0.40771484375,
      "grad_norm": 7.2647600173950195,
      "learning_rate": 3.2192324839075397e-06,
      "loss": 3.3452,
      "step": 40080
    },
    {
      "epoch": 0.4077657063802083,
      "grad_norm": 10.059800148010254,
      "learning_rate": 3.218849499657623e-06,
      "loss": 3.4234,
      "step": 40085
    },
    {
      "epoch": 0.4078165690104167,
      "grad_norm": 14.233746528625488,
      "learning_rate": 3.2184664970161124e-06,
      "loss": 3.5071,
      "step": 40090
    },
    {
      "epoch": 0.407867431640625,
      "grad_norm": 9.250040054321289,
      "learning_rate": 3.2180834759928065e-06,
      "loss": 3.467,
      "step": 40095
    },
    {
      "epoch": 0.4079182942708333,
      "grad_norm": 13.586669921875,
      "learning_rate": 3.217700436597505e-06,
      "loss": 3.0783,
      "step": 40100
    },
    {
      "epoch": 0.4079691569010417,
      "grad_norm": 15.873119354248047,
      "learning_rate": 3.217317378840008e-06,
      "loss": 3.3066,
      "step": 40105
    },
    {
      "epoch": 0.40802001953125,
      "grad_norm": 12.343993186950684,
      "learning_rate": 3.216934302730116e-06,
      "loss": 3.0871,
      "step": 40110
    },
    {
      "epoch": 0.4080708821614583,
      "grad_norm": 12.203359603881836,
      "learning_rate": 3.2165512082776297e-06,
      "loss": 3.1915,
      "step": 40115
    },
    {
      "epoch": 0.4081217447916667,
      "grad_norm": 9.524660110473633,
      "learning_rate": 3.2161680954923504e-06,
      "loss": 3.3897,
      "step": 40120
    },
    {
      "epoch": 0.408172607421875,
      "grad_norm": 11.293547630310059,
      "learning_rate": 3.2157849643840806e-06,
      "loss": 3.573,
      "step": 40125
    },
    {
      "epoch": 0.4082234700520833,
      "grad_norm": 10.416081428527832,
      "learning_rate": 3.215401814962621e-06,
      "loss": 3.3696,
      "step": 40130
    },
    {
      "epoch": 0.4082743326822917,
      "grad_norm": 9.743280410766602,
      "learning_rate": 3.2150186472377767e-06,
      "loss": 3.4491,
      "step": 40135
    },
    {
      "epoch": 0.4083251953125,
      "grad_norm": 9.836259841918945,
      "learning_rate": 3.2146354612193497e-06,
      "loss": 3.2513,
      "step": 40140
    },
    {
      "epoch": 0.4083760579427083,
      "grad_norm": 8.281194686889648,
      "learning_rate": 3.214252256917143e-06,
      "loss": 3.3309,
      "step": 40145
    },
    {
      "epoch": 0.4084269205729167,
      "grad_norm": 12.972834587097168,
      "learning_rate": 3.213869034340963e-06,
      "loss": 3.2908,
      "step": 40150
    },
    {
      "epoch": 0.408477783203125,
      "grad_norm": 9.326537132263184,
      "learning_rate": 3.213485793500612e-06,
      "loss": 3.3215,
      "step": 40155
    },
    {
      "epoch": 0.4085286458333333,
      "grad_norm": 14.250112533569336,
      "learning_rate": 3.213102534405896e-06,
      "loss": 3.3848,
      "step": 40160
    },
    {
      "epoch": 0.4085795084635417,
      "grad_norm": 14.024307250976562,
      "learning_rate": 3.212719257066621e-06,
      "loss": 3.2285,
      "step": 40165
    },
    {
      "epoch": 0.40863037109375,
      "grad_norm": 12.0462007522583,
      "learning_rate": 3.2123359614925925e-06,
      "loss": 3.5627,
      "step": 40170
    },
    {
      "epoch": 0.4086812337239583,
      "grad_norm": 13.510028839111328,
      "learning_rate": 3.2119526476936177e-06,
      "loss": 3.0223,
      "step": 40175
    },
    {
      "epoch": 0.4087320963541667,
      "grad_norm": 11.582918167114258,
      "learning_rate": 3.2115693156795026e-06,
      "loss": 3.3018,
      "step": 40180
    },
    {
      "epoch": 0.408782958984375,
      "grad_norm": 8.617096900939941,
      "learning_rate": 3.2111859654600554e-06,
      "loss": 3.4995,
      "step": 40185
    },
    {
      "epoch": 0.4088338216145833,
      "grad_norm": 16.85012435913086,
      "learning_rate": 3.210802597045084e-06,
      "loss": 3.5344,
      "step": 40190
    },
    {
      "epoch": 0.4088846842447917,
      "grad_norm": 12.56917667388916,
      "learning_rate": 3.2104192104443965e-06,
      "loss": 3.2028,
      "step": 40195
    },
    {
      "epoch": 0.408935546875,
      "grad_norm": 13.219779968261719,
      "learning_rate": 3.210035805667802e-06,
      "loss": 3.5665,
      "step": 40200
    },
    {
      "epoch": 0.4089864095052083,
      "grad_norm": 9.362250328063965,
      "learning_rate": 3.2096523827251096e-06,
      "loss": 3.5273,
      "step": 40205
    },
    {
      "epoch": 0.4090372721354167,
      "grad_norm": 12.411036491394043,
      "learning_rate": 3.2092689416261295e-06,
      "loss": 3.619,
      "step": 40210
    },
    {
      "epoch": 0.409088134765625,
      "grad_norm": 13.231313705444336,
      "learning_rate": 3.2088854823806714e-06,
      "loss": 3.682,
      "step": 40215
    },
    {
      "epoch": 0.4091389973958333,
      "grad_norm": 15.368950843811035,
      "learning_rate": 3.208502004998546e-06,
      "loss": 3.4148,
      "step": 40220
    },
    {
      "epoch": 0.4091898600260417,
      "grad_norm": 8.858798027038574,
      "learning_rate": 3.208118509489565e-06,
      "loss": 3.5826,
      "step": 40225
    },
    {
      "epoch": 0.40924072265625,
      "grad_norm": 12.872635841369629,
      "learning_rate": 3.20773499586354e-06,
      "loss": 3.4896,
      "step": 40230
    },
    {
      "epoch": 0.4092915852864583,
      "grad_norm": 12.837394714355469,
      "learning_rate": 3.2073514641302827e-06,
      "loss": 3.192,
      "step": 40235
    },
    {
      "epoch": 0.4093424479166667,
      "grad_norm": 11.67916202545166,
      "learning_rate": 3.2069679142996055e-06,
      "loss": 3.3476,
      "step": 40240
    },
    {
      "epoch": 0.409393310546875,
      "grad_norm": 9.539045333862305,
      "learning_rate": 3.206584346381323e-06,
      "loss": 3.2381,
      "step": 40245
    },
    {
      "epoch": 0.4094441731770833,
      "grad_norm": 9.679408073425293,
      "learning_rate": 3.206200760385246e-06,
      "loss": 3.7806,
      "step": 40250
    },
    {
      "epoch": 0.4094950358072917,
      "grad_norm": 8.230401992797852,
      "learning_rate": 3.2058171563211905e-06,
      "loss": 3.5604,
      "step": 40255
    },
    {
      "epoch": 0.4095458984375,
      "grad_norm": 13.984374046325684,
      "learning_rate": 3.20543353419897e-06,
      "loss": 3.1788,
      "step": 40260
    },
    {
      "epoch": 0.4095967610677083,
      "grad_norm": 14.798341751098633,
      "learning_rate": 3.2050498940284003e-06,
      "loss": 4.2144,
      "step": 40265
    },
    {
      "epoch": 0.4096476236979167,
      "grad_norm": 10.024030685424805,
      "learning_rate": 3.204666235819296e-06,
      "loss": 3.4174,
      "step": 40270
    },
    {
      "epoch": 0.409698486328125,
      "grad_norm": 19.80939483642578,
      "learning_rate": 3.204282559581473e-06,
      "loss": 3.6143,
      "step": 40275
    },
    {
      "epoch": 0.4097493489583333,
      "grad_norm": 12.339203834533691,
      "learning_rate": 3.2038988653247477e-06,
      "loss": 3.3237,
      "step": 40280
    },
    {
      "epoch": 0.4098002115885417,
      "grad_norm": 13.307379722595215,
      "learning_rate": 3.203515153058937e-06,
      "loss": 3.0846,
      "step": 40285
    },
    {
      "epoch": 0.40985107421875,
      "grad_norm": 13.441474914550781,
      "learning_rate": 3.203131422793857e-06,
      "loss": 2.8081,
      "step": 40290
    },
    {
      "epoch": 0.4099019368489583,
      "grad_norm": 9.035524368286133,
      "learning_rate": 3.2027476745393277e-06,
      "loss": 3.4538,
      "step": 40295
    },
    {
      "epoch": 0.4099527994791667,
      "grad_norm": 9.35716438293457,
      "learning_rate": 3.2023639083051644e-06,
      "loss": 3.2194,
      "step": 40300
    },
    {
      "epoch": 0.410003662109375,
      "grad_norm": 15.0332612991333,
      "learning_rate": 3.2019801241011877e-06,
      "loss": 3.0303,
      "step": 40305
    },
    {
      "epoch": 0.4100545247395833,
      "grad_norm": 12.640445709228516,
      "learning_rate": 3.2015963219372158e-06,
      "loss": 3.6029,
      "step": 40310
    },
    {
      "epoch": 0.4101053873697917,
      "grad_norm": 9.166719436645508,
      "learning_rate": 3.2012125018230677e-06,
      "loss": 2.9824,
      "step": 40315
    },
    {
      "epoch": 0.41015625,
      "grad_norm": 13.075416564941406,
      "learning_rate": 3.200828663768565e-06,
      "loss": 3.0641,
      "step": 40320
    },
    {
      "epoch": 0.4102071126302083,
      "grad_norm": 9.390490531921387,
      "learning_rate": 3.200444807783526e-06,
      "loss": 3.2462,
      "step": 40325
    },
    {
      "epoch": 0.4102579752604167,
      "grad_norm": 13.261961936950684,
      "learning_rate": 3.200060933877773e-06,
      "loss": 3.1896,
      "step": 40330
    },
    {
      "epoch": 0.410308837890625,
      "grad_norm": 9.776216506958008,
      "learning_rate": 3.1996770420611265e-06,
      "loss": 3.715,
      "step": 40335
    },
    {
      "epoch": 0.4103597005208333,
      "grad_norm": 8.422574996948242,
      "learning_rate": 3.1992931323434097e-06,
      "loss": 3.3871,
      "step": 40340
    },
    {
      "epoch": 0.4104105631510417,
      "grad_norm": 10.582122802734375,
      "learning_rate": 3.198909204734443e-06,
      "loss": 3.4996,
      "step": 40345
    },
    {
      "epoch": 0.41046142578125,
      "grad_norm": 11.832892417907715,
      "learning_rate": 3.1985252592440507e-06,
      "loss": 3.3974,
      "step": 40350
    },
    {
      "epoch": 0.4105122884114583,
      "grad_norm": 15.56053352355957,
      "learning_rate": 3.1981412958820547e-06,
      "loss": 3.265,
      "step": 40355
    },
    {
      "epoch": 0.4105631510416667,
      "grad_norm": 17.74793243408203,
      "learning_rate": 3.1977573146582798e-06,
      "loss": 3.6518,
      "step": 40360
    },
    {
      "epoch": 0.410614013671875,
      "grad_norm": 14.441006660461426,
      "learning_rate": 3.197373315582548e-06,
      "loss": 3.4867,
      "step": 40365
    },
    {
      "epoch": 0.4106648763020833,
      "grad_norm": 12.163776397705078,
      "learning_rate": 3.1969892986646866e-06,
      "loss": 3.4167,
      "step": 40370
    },
    {
      "epoch": 0.4107157389322917,
      "grad_norm": 7.191044807434082,
      "learning_rate": 3.1966052639145188e-06,
      "loss": 3.267,
      "step": 40375
    },
    {
      "epoch": 0.4107666015625,
      "grad_norm": 14.789758682250977,
      "learning_rate": 3.19622121134187e-06,
      "loss": 3.2197,
      "step": 40380
    },
    {
      "epoch": 0.4108174641927083,
      "grad_norm": 10.832036018371582,
      "learning_rate": 3.195837140956567e-06,
      "loss": 3.1452,
      "step": 40385
    },
    {
      "epoch": 0.4108683268229167,
      "grad_norm": 12.134127616882324,
      "learning_rate": 3.1954530527684356e-06,
      "loss": 4.023,
      "step": 40390
    },
    {
      "epoch": 0.410919189453125,
      "grad_norm": 15.02275276184082,
      "learning_rate": 3.1950689467873026e-06,
      "loss": 2.8569,
      "step": 40395
    },
    {
      "epoch": 0.4109700520833333,
      "grad_norm": 11.308838844299316,
      "learning_rate": 3.1946848230229955e-06,
      "loss": 3.1812,
      "step": 40400
    },
    {
      "epoch": 0.4110209147135417,
      "grad_norm": 12.479435920715332,
      "learning_rate": 3.1943006814853417e-06,
      "loss": 3.8656,
      "step": 40405
    },
    {
      "epoch": 0.41107177734375,
      "grad_norm": 16.549497604370117,
      "learning_rate": 3.1939165221841696e-06,
      "loss": 3.2209,
      "step": 40410
    },
    {
      "epoch": 0.4111226399739583,
      "grad_norm": 10.488978385925293,
      "learning_rate": 3.193532345129307e-06,
      "loss": 4.3753,
      "step": 40415
    },
    {
      "epoch": 0.4111735026041667,
      "grad_norm": 17.199155807495117,
      "learning_rate": 3.193148150330585e-06,
      "loss": 3.2604,
      "step": 40420
    },
    {
      "epoch": 0.411224365234375,
      "grad_norm": 11.074233055114746,
      "learning_rate": 3.1927639377978314e-06,
      "loss": 3.5254,
      "step": 40425
    },
    {
      "epoch": 0.4112752278645833,
      "grad_norm": 14.462937355041504,
      "learning_rate": 3.1923797075408757e-06,
      "loss": 3.1973,
      "step": 40430
    },
    {
      "epoch": 0.4113260904947917,
      "grad_norm": 14.890127182006836,
      "learning_rate": 3.191995459569551e-06,
      "loss": 3.4347,
      "step": 40435
    },
    {
      "epoch": 0.411376953125,
      "grad_norm": 14.60438346862793,
      "learning_rate": 3.1916111938936857e-06,
      "loss": 3.4392,
      "step": 40440
    },
    {
      "epoch": 0.4114278157552083,
      "grad_norm": 10.742544174194336,
      "learning_rate": 3.191226910523112e-06,
      "loss": 3.4331,
      "step": 40445
    },
    {
      "epoch": 0.4114786783854167,
      "grad_norm": 14.729475021362305,
      "learning_rate": 3.1908426094676616e-06,
      "loss": 3.1607,
      "step": 40450
    },
    {
      "epoch": 0.411529541015625,
      "grad_norm": 7.225653648376465,
      "learning_rate": 3.1904582907371663e-06,
      "loss": 3.2739,
      "step": 40455
    },
    {
      "epoch": 0.4115804036458333,
      "grad_norm": 16.015857696533203,
      "learning_rate": 3.1900739543414592e-06,
      "loss": 3.3931,
      "step": 40460
    },
    {
      "epoch": 0.4116312662760417,
      "grad_norm": 11.942667961120605,
      "learning_rate": 3.189689600290375e-06,
      "loss": 3.2618,
      "step": 40465
    },
    {
      "epoch": 0.41168212890625,
      "grad_norm": 10.380414962768555,
      "learning_rate": 3.1893052285937443e-06,
      "loss": 3.5267,
      "step": 40470
    },
    {
      "epoch": 0.4117329915364583,
      "grad_norm": 15.762372016906738,
      "learning_rate": 3.1889208392614037e-06,
      "loss": 3.4845,
      "step": 40475
    },
    {
      "epoch": 0.4117838541666667,
      "grad_norm": 11.058128356933594,
      "learning_rate": 3.1885364323031853e-06,
      "loss": 3.2676,
      "step": 40480
    },
    {
      "epoch": 0.411834716796875,
      "grad_norm": 12.959691047668457,
      "learning_rate": 3.1881520077289272e-06,
      "loss": 3.8364,
      "step": 40485
    },
    {
      "epoch": 0.4118855794270833,
      "grad_norm": 9.470345497131348,
      "learning_rate": 3.1877675655484625e-06,
      "loss": 3.337,
      "step": 40490
    },
    {
      "epoch": 0.4119364420572917,
      "grad_norm": 10.67660903930664,
      "learning_rate": 3.1873831057716277e-06,
      "loss": 3.1722,
      "step": 40495
    },
    {
      "epoch": 0.4119873046875,
      "grad_norm": 14.760432243347168,
      "learning_rate": 3.1869986284082587e-06,
      "loss": 3.3537,
      "step": 40500
    },
    {
      "epoch": 0.4120381673177083,
      "grad_norm": 11.180314064025879,
      "learning_rate": 3.186614133468193e-06,
      "loss": 3.3887,
      "step": 40505
    },
    {
      "epoch": 0.4120890299479167,
      "grad_norm": 17.601219177246094,
      "learning_rate": 3.186229620961267e-06,
      "loss": 3.3066,
      "step": 40510
    },
    {
      "epoch": 0.412139892578125,
      "grad_norm": 11.670747756958008,
      "learning_rate": 3.185845090897319e-06,
      "loss": 3.4507,
      "step": 40515
    },
    {
      "epoch": 0.4121907552083333,
      "grad_norm": 11.915066719055176,
      "learning_rate": 3.185460543286187e-06,
      "loss": 3.4408,
      "step": 40520
    },
    {
      "epoch": 0.4122416178385417,
      "grad_norm": 10.542513847351074,
      "learning_rate": 3.18507597813771e-06,
      "loss": 3.3083,
      "step": 40525
    },
    {
      "epoch": 0.41229248046875,
      "grad_norm": 9.632444381713867,
      "learning_rate": 3.1846913954617257e-06,
      "loss": 3.736,
      "step": 40530
    },
    {
      "epoch": 0.4123433430989583,
      "grad_norm": 11.756135940551758,
      "learning_rate": 3.184306795268074e-06,
      "loss": 3.3021,
      "step": 40535
    },
    {
      "epoch": 0.4123942057291667,
      "grad_norm": 11.835874557495117,
      "learning_rate": 3.1839221775665964e-06,
      "loss": 3.2396,
      "step": 40540
    },
    {
      "epoch": 0.412445068359375,
      "grad_norm": 16.399227142333984,
      "learning_rate": 3.183537542367131e-06,
      "loss": 3.3441,
      "step": 40545
    },
    {
      "epoch": 0.4124959309895833,
      "grad_norm": 13.708842277526855,
      "learning_rate": 3.18315288967952e-06,
      "loss": 3.224,
      "step": 40550
    },
    {
      "epoch": 0.4125467936197917,
      "grad_norm": 11.074498176574707,
      "learning_rate": 3.1827682195136038e-06,
      "loss": 3.8189,
      "step": 40555
    },
    {
      "epoch": 0.41259765625,
      "grad_norm": 12.426509857177734,
      "learning_rate": 3.182383531879225e-06,
      "loss": 4.0286,
      "step": 40560
    },
    {
      "epoch": 0.4126485188802083,
      "grad_norm": 15.233750343322754,
      "learning_rate": 3.1819988267862253e-06,
      "loss": 3.5985,
      "step": 40565
    },
    {
      "epoch": 0.4126993815104167,
      "grad_norm": 12.714953422546387,
      "learning_rate": 3.181614104244447e-06,
      "loss": 3.2174,
      "step": 40570
    },
    {
      "epoch": 0.412750244140625,
      "grad_norm": 13.589609146118164,
      "learning_rate": 3.181229364263734e-06,
      "loss": 3.2432,
      "step": 40575
    },
    {
      "epoch": 0.4128011067708333,
      "grad_norm": 14.362196922302246,
      "learning_rate": 3.180844606853929e-06,
      "loss": 3.0738,
      "step": 40580
    },
    {
      "epoch": 0.4128519694010417,
      "grad_norm": 14.002069473266602,
      "learning_rate": 3.1804598320248747e-06,
      "loss": 3.1727,
      "step": 40585
    },
    {
      "epoch": 0.41290283203125,
      "grad_norm": 11.42420482635498,
      "learning_rate": 3.180075039786419e-06,
      "loss": 3.5389,
      "step": 40590
    },
    {
      "epoch": 0.4129536946614583,
      "grad_norm": 8.944576263427734,
      "learning_rate": 3.1796902301484033e-06,
      "loss": 3.7406,
      "step": 40595
    },
    {
      "epoch": 0.4130045572916667,
      "grad_norm": 11.496153831481934,
      "learning_rate": 3.1793054031206742e-06,
      "loss": 3.1414,
      "step": 40600
    },
    {
      "epoch": 0.413055419921875,
      "grad_norm": 11.331368446350098,
      "learning_rate": 3.1789205587130777e-06,
      "loss": 3.2902,
      "step": 40605
    },
    {
      "epoch": 0.4131062825520833,
      "grad_norm": 15.670097351074219,
      "learning_rate": 3.17853569693546e-06,
      "loss": 3.5692,
      "step": 40610
    },
    {
      "epoch": 0.4131571451822917,
      "grad_norm": 12.869950294494629,
      "learning_rate": 3.1781508177976667e-06,
      "loss": 3.3101,
      "step": 40615
    },
    {
      "epoch": 0.4132080078125,
      "grad_norm": 17.0570125579834,
      "learning_rate": 3.177765921309546e-06,
      "loss": 3.3191,
      "step": 40620
    },
    {
      "epoch": 0.4132588704427083,
      "grad_norm": 9.5101318359375,
      "learning_rate": 3.177381007480944e-06,
      "loss": 3.1341,
      "step": 40625
    },
    {
      "epoch": 0.4133097330729167,
      "grad_norm": 9.91584300994873,
      "learning_rate": 3.17699607632171e-06,
      "loss": 3.5979,
      "step": 40630
    },
    {
      "epoch": 0.413360595703125,
      "grad_norm": 13.129979133605957,
      "learning_rate": 3.1766111278416916e-06,
      "loss": 3.4869,
      "step": 40635
    },
    {
      "epoch": 0.4134114583333333,
      "grad_norm": 10.525177001953125,
      "learning_rate": 3.1762261620507386e-06,
      "loss": 3.0418,
      "step": 40640
    },
    {
      "epoch": 0.4134623209635417,
      "grad_norm": 10.457929611206055,
      "learning_rate": 3.1758411789586992e-06,
      "loss": 3.219,
      "step": 40645
    },
    {
      "epoch": 0.41351318359375,
      "grad_norm": 9.514257431030273,
      "learning_rate": 3.1754561785754225e-06,
      "loss": 3.1455,
      "step": 40650
    },
    {
      "epoch": 0.4135640462239583,
      "grad_norm": 16.182477951049805,
      "learning_rate": 3.175071160910761e-06,
      "loss": 3.3036,
      "step": 40655
    },
    {
      "epoch": 0.4136149088541667,
      "grad_norm": 15.760019302368164,
      "learning_rate": 3.1746861259745633e-06,
      "loss": 3.3682,
      "step": 40660
    },
    {
      "epoch": 0.413665771484375,
      "grad_norm": 10.646612167358398,
      "learning_rate": 3.1743010737766803e-06,
      "loss": 3.5076,
      "step": 40665
    },
    {
      "epoch": 0.4137166341145833,
      "grad_norm": 12.429669380187988,
      "learning_rate": 3.173916004326965e-06,
      "loss": 2.9539,
      "step": 40670
    },
    {
      "epoch": 0.4137674967447917,
      "grad_norm": 9.560185432434082,
      "learning_rate": 3.173530917635268e-06,
      "loss": 3.1944,
      "step": 40675
    },
    {
      "epoch": 0.413818359375,
      "grad_norm": 10.479032516479492,
      "learning_rate": 3.1731458137114423e-06,
      "loss": 3.4629,
      "step": 40680
    },
    {
      "epoch": 0.4138692220052083,
      "grad_norm": 14.869799613952637,
      "learning_rate": 3.1727606925653416e-06,
      "loss": 3.681,
      "step": 40685
    },
    {
      "epoch": 0.4139200846354167,
      "grad_norm": 14.663705825805664,
      "learning_rate": 3.1723755542068157e-06,
      "loss": 3.5367,
      "step": 40690
    },
    {
      "epoch": 0.413970947265625,
      "grad_norm": 8.11724853515625,
      "learning_rate": 3.171990398645723e-06,
      "loss": 3.2612,
      "step": 40695
    },
    {
      "epoch": 0.4140218098958333,
      "grad_norm": 8.949166297912598,
      "learning_rate": 3.171605225891914e-06,
      "loss": 3.157,
      "step": 40700
    },
    {
      "epoch": 0.4140726725260417,
      "grad_norm": 9.309784889221191,
      "learning_rate": 3.1712200359552445e-06,
      "loss": 2.7949,
      "step": 40705
    },
    {
      "epoch": 0.41412353515625,
      "grad_norm": 9.50880241394043,
      "learning_rate": 3.170834828845569e-06,
      "loss": 3.4913,
      "step": 40710
    },
    {
      "epoch": 0.4141743977864583,
      "grad_norm": 10.189599990844727,
      "learning_rate": 3.1704496045727447e-06,
      "loss": 3.4456,
      "step": 40715
    },
    {
      "epoch": 0.4142252604166667,
      "grad_norm": 16.509906768798828,
      "learning_rate": 3.1700643631466256e-06,
      "loss": 3.2289,
      "step": 40720
    },
    {
      "epoch": 0.414276123046875,
      "grad_norm": 9.536114692687988,
      "learning_rate": 3.1696791045770682e-06,
      "loss": 3.3998,
      "step": 40725
    },
    {
      "epoch": 0.4143269856770833,
      "grad_norm": 14.481447219848633,
      "learning_rate": 3.1692938288739296e-06,
      "loss": 3.6835,
      "step": 40730
    },
    {
      "epoch": 0.4143778483072917,
      "grad_norm": 12.22691535949707,
      "learning_rate": 3.1689085360470674e-06,
      "loss": 3.6889,
      "step": 40735
    },
    {
      "epoch": 0.4144287109375,
      "grad_norm": 16.401357650756836,
      "learning_rate": 3.1685232261063386e-06,
      "loss": 3.0401,
      "step": 40740
    },
    {
      "epoch": 0.4144795735677083,
      "grad_norm": 7.340702056884766,
      "learning_rate": 3.168137899061602e-06,
      "loss": 3.0076,
      "step": 40745
    },
    {
      "epoch": 0.4145304361979167,
      "grad_norm": 9.923087120056152,
      "learning_rate": 3.167752554922716e-06,
      "loss": 3.1689,
      "step": 40750
    },
    {
      "epoch": 0.414581298828125,
      "grad_norm": 15.847943305969238,
      "learning_rate": 3.1673671936995377e-06,
      "loss": 3.2989,
      "step": 40755
    },
    {
      "epoch": 0.4146321614583333,
      "grad_norm": 10.91378116607666,
      "learning_rate": 3.1669818154019287e-06,
      "loss": 3.1673,
      "step": 40760
    },
    {
      "epoch": 0.4146830240885417,
      "grad_norm": 9.859515190124512,
      "learning_rate": 3.166596420039748e-06,
      "loss": 3.3117,
      "step": 40765
    },
    {
      "epoch": 0.41473388671875,
      "grad_norm": 15.599798202514648,
      "learning_rate": 3.1662110076228564e-06,
      "loss": 3.3584,
      "step": 40770
    },
    {
      "epoch": 0.4147847493489583,
      "grad_norm": 14.321131706237793,
      "learning_rate": 3.1658255781611138e-06,
      "loss": 3.8719,
      "step": 40775
    },
    {
      "epoch": 0.4148356119791667,
      "grad_norm": 13.262664794921875,
      "learning_rate": 3.165440131664381e-06,
      "loss": 3.406,
      "step": 40780
    },
    {
      "epoch": 0.414886474609375,
      "grad_norm": 14.643454551696777,
      "learning_rate": 3.1650546681425204e-06,
      "loss": 3.4258,
      "step": 40785
    },
    {
      "epoch": 0.4149373372395833,
      "grad_norm": 9.192747116088867,
      "learning_rate": 3.164669187605395e-06,
      "loss": 3.7027,
      "step": 40790
    },
    {
      "epoch": 0.4149881998697917,
      "grad_norm": 12.051810264587402,
      "learning_rate": 3.1642836900628637e-06,
      "loss": 3.4306,
      "step": 40795
    },
    {
      "epoch": 0.4150390625,
      "grad_norm": 9.823638916015625,
      "learning_rate": 3.1638981755247934e-06,
      "loss": 3.5007,
      "step": 40800
    },
    {
      "epoch": 0.4150899251302083,
      "grad_norm": 11.998472213745117,
      "learning_rate": 3.1635126440010444e-06,
      "loss": 3.477,
      "step": 40805
    },
    {
      "epoch": 0.4151407877604167,
      "grad_norm": 14.473235130310059,
      "learning_rate": 3.1631270955014833e-06,
      "loss": 3.4691,
      "step": 40810
    },
    {
      "epoch": 0.415191650390625,
      "grad_norm": 8.37810230255127,
      "learning_rate": 3.1627415300359717e-06,
      "loss": 3.4367,
      "step": 40815
    },
    {
      "epoch": 0.4152425130208333,
      "grad_norm": 19.481773376464844,
      "learning_rate": 3.162355947614375e-06,
      "loss": 3.3668,
      "step": 40820
    },
    {
      "epoch": 0.4152933756510417,
      "grad_norm": 16.640199661254883,
      "learning_rate": 3.1619703482465586e-06,
      "loss": 3.0726,
      "step": 40825
    },
    {
      "epoch": 0.41534423828125,
      "grad_norm": 12.760043144226074,
      "learning_rate": 3.1615847319423874e-06,
      "loss": 3.6446,
      "step": 40830
    },
    {
      "epoch": 0.4153951009114583,
      "grad_norm": 15.554396629333496,
      "learning_rate": 3.161199098711728e-06,
      "loss": 3.2753,
      "step": 40835
    },
    {
      "epoch": 0.4154459635416667,
      "grad_norm": 9.984983444213867,
      "learning_rate": 3.1608134485644458e-06,
      "loss": 2.8998,
      "step": 40840
    },
    {
      "epoch": 0.415496826171875,
      "grad_norm": 12.402283668518066,
      "learning_rate": 3.160427781510409e-06,
      "loss": 3.2864,
      "step": 40845
    },
    {
      "epoch": 0.4155476888020833,
      "grad_norm": 11.073039054870605,
      "learning_rate": 3.160042097559484e-06,
      "loss": 2.866,
      "step": 40850
    },
    {
      "epoch": 0.4155985514322917,
      "grad_norm": 10.235372543334961,
      "learning_rate": 3.159656396721538e-06,
      "loss": 3.1743,
      "step": 40855
    },
    {
      "epoch": 0.4156494140625,
      "grad_norm": 11.603734970092773,
      "learning_rate": 3.1592706790064387e-06,
      "loss": 3.4963,
      "step": 40860
    },
    {
      "epoch": 0.4157002766927083,
      "grad_norm": 9.706622123718262,
      "learning_rate": 3.1588849444240564e-06,
      "loss": 3.244,
      "step": 40865
    },
    {
      "epoch": 0.4157511393229167,
      "grad_norm": 10.390192985534668,
      "learning_rate": 3.1584991929842586e-06,
      "loss": 3.0856,
      "step": 40870
    },
    {
      "epoch": 0.415802001953125,
      "grad_norm": 12.130228996276855,
      "learning_rate": 3.1581134246969153e-06,
      "loss": 3.2899,
      "step": 40875
    },
    {
      "epoch": 0.4158528645833333,
      "grad_norm": 9.266514778137207,
      "learning_rate": 3.1577276395718957e-06,
      "loss": 3.3391,
      "step": 40880
    },
    {
      "epoch": 0.4159037272135417,
      "grad_norm": 10.627102851867676,
      "learning_rate": 3.157341837619071e-06,
      "loss": 3.7669,
      "step": 40885
    },
    {
      "epoch": 0.41595458984375,
      "grad_norm": 10.077190399169922,
      "learning_rate": 3.15695601884831e-06,
      "loss": 3.2736,
      "step": 40890
    },
    {
      "epoch": 0.4160054524739583,
      "grad_norm": 14.39554500579834,
      "learning_rate": 3.1565701832694856e-06,
      "loss": 3.5625,
      "step": 40895
    },
    {
      "epoch": 0.4160563151041667,
      "grad_norm": 11.000900268554688,
      "learning_rate": 3.1561843308924687e-06,
      "loss": 3.4582,
      "step": 40900
    },
    {
      "epoch": 0.416107177734375,
      "grad_norm": 7.12355375289917,
      "learning_rate": 3.155798461727132e-06,
      "loss": 3.2976,
      "step": 40905
    },
    {
      "epoch": 0.4161580403645833,
      "grad_norm": 7.608798503875732,
      "learning_rate": 3.1554125757833453e-06,
      "loss": 3.3805,
      "step": 40910
    },
    {
      "epoch": 0.4162089029947917,
      "grad_norm": 17.018798828125,
      "learning_rate": 3.155026673070985e-06,
      "loss": 3.5029,
      "step": 40915
    },
    {
      "epoch": 0.416259765625,
      "grad_norm": 10.469268798828125,
      "learning_rate": 3.1546407535999214e-06,
      "loss": 3.0716,
      "step": 40920
    },
    {
      "epoch": 0.4163106282552083,
      "grad_norm": 7.395883083343506,
      "learning_rate": 3.1542548173800293e-06,
      "loss": 3.6238,
      "step": 40925
    },
    {
      "epoch": 0.4163614908854167,
      "grad_norm": 14.05190372467041,
      "learning_rate": 3.1538688644211834e-06,
      "loss": 3.0576,
      "step": 40930
    },
    {
      "epoch": 0.416412353515625,
      "grad_norm": 10.219439506530762,
      "learning_rate": 3.1534828947332573e-06,
      "loss": 3.6053,
      "step": 40935
    },
    {
      "epoch": 0.4164632161458333,
      "grad_norm": 11.207366943359375,
      "learning_rate": 3.1530969083261265e-06,
      "loss": 3.9513,
      "step": 40940
    },
    {
      "epoch": 0.4165140787760417,
      "grad_norm": 10.072528839111328,
      "learning_rate": 3.1527109052096656e-06,
      "loss": 3.1873,
      "step": 40945
    },
    {
      "epoch": 0.41656494140625,
      "grad_norm": 7.561704635620117,
      "learning_rate": 3.152324885393751e-06,
      "loss": 3.3976,
      "step": 40950
    },
    {
      "epoch": 0.4166158040364583,
      "grad_norm": 13.165698051452637,
      "learning_rate": 3.1519388488882586e-06,
      "loss": 3.2867,
      "step": 40955
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 10.214173316955566,
      "learning_rate": 3.1515527957030657e-06,
      "loss": 3.0653,
      "step": 40960
    },
    {
      "epoch": 0.416717529296875,
      "grad_norm": 12.931204795837402,
      "learning_rate": 3.151166725848049e-06,
      "loss": 3.2135,
      "step": 40965
    },
    {
      "epoch": 0.4167683919270833,
      "grad_norm": 8.744141578674316,
      "learning_rate": 3.1507806393330865e-06,
      "loss": 3.33,
      "step": 40970
    },
    {
      "epoch": 0.4168192545572917,
      "grad_norm": 13.387395858764648,
      "learning_rate": 3.1503945361680543e-06,
      "loss": 3.1974,
      "step": 40975
    },
    {
      "epoch": 0.4168701171875,
      "grad_norm": 8.401640892028809,
      "learning_rate": 3.150008416362833e-06,
      "loss": 3.1772,
      "step": 40980
    },
    {
      "epoch": 0.4169209798177083,
      "grad_norm": 13.557087898254395,
      "learning_rate": 3.1496222799273002e-06,
      "loss": 3.2859,
      "step": 40985
    },
    {
      "epoch": 0.4169718424479167,
      "grad_norm": 12.962182998657227,
      "learning_rate": 3.149236126871335e-06,
      "loss": 3.4554,
      "step": 40990
    },
    {
      "epoch": 0.417022705078125,
      "grad_norm": 13.028253555297852,
      "learning_rate": 3.1488499572048177e-06,
      "loss": 3.6858,
      "step": 40995
    },
    {
      "epoch": 0.4170735677083333,
      "grad_norm": 10.180688858032227,
      "learning_rate": 3.1484637709376274e-06,
      "loss": 3.274,
      "step": 41000
    },
    {
      "epoch": 0.4171244303385417,
      "grad_norm": 18.28401756286621,
      "learning_rate": 3.148077568079646e-06,
      "loss": 3.2267,
      "step": 41005
    },
    {
      "epoch": 0.41717529296875,
      "grad_norm": 13.654256820678711,
      "learning_rate": 3.147691348640754e-06,
      "loss": 3.538,
      "step": 41010
    },
    {
      "epoch": 0.4172261555989583,
      "grad_norm": 13.836880683898926,
      "learning_rate": 3.1473051126308314e-06,
      "loss": 3.0475,
      "step": 41015
    },
    {
      "epoch": 0.4172770182291667,
      "grad_norm": 11.592706680297852,
      "learning_rate": 3.1469188600597613e-06,
      "loss": 4.1781,
      "step": 41020
    },
    {
      "epoch": 0.417327880859375,
      "grad_norm": 12.247729301452637,
      "learning_rate": 3.1465325909374255e-06,
      "loss": 3.3115,
      "step": 41025
    },
    {
      "epoch": 0.4173787434895833,
      "grad_norm": 11.236956596374512,
      "learning_rate": 3.1461463052737064e-06,
      "loss": 3.3245,
      "step": 41030
    },
    {
      "epoch": 0.4174296061197917,
      "grad_norm": 12.398927688598633,
      "learning_rate": 3.145760003078488e-06,
      "loss": 3.2676,
      "step": 41035
    },
    {
      "epoch": 0.41748046875,
      "grad_norm": 9.001666069030762,
      "learning_rate": 3.1453736843616515e-06,
      "loss": 3.1629,
      "step": 41040
    },
    {
      "epoch": 0.4175313313802083,
      "grad_norm": 15.180867195129395,
      "learning_rate": 3.144987349133084e-06,
      "loss": 3.2056,
      "step": 41045
    },
    {
      "epoch": 0.4175821940104167,
      "grad_norm": 14.018913269042969,
      "learning_rate": 3.144600997402667e-06,
      "loss": 3.4004,
      "step": 41050
    },
    {
      "epoch": 0.417633056640625,
      "grad_norm": 10.41834545135498,
      "learning_rate": 3.1442146291802866e-06,
      "loss": 3.9112,
      "step": 41055
    },
    {
      "epoch": 0.4176839192708333,
      "grad_norm": 22.930706024169922,
      "learning_rate": 3.143828244475828e-06,
      "loss": 3.8011,
      "step": 41060
    },
    {
      "epoch": 0.4177347819010417,
      "grad_norm": 8.728473663330078,
      "learning_rate": 3.1434418432991755e-06,
      "loss": 3.4481,
      "step": 41065
    },
    {
      "epoch": 0.41778564453125,
      "grad_norm": 13.227070808410645,
      "learning_rate": 3.1430554256602165e-06,
      "loss": 3.3979,
      "step": 41070
    },
    {
      "epoch": 0.4178365071614583,
      "grad_norm": 13.314940452575684,
      "learning_rate": 3.1426689915688367e-06,
      "loss": 3.5534,
      "step": 41075
    },
    {
      "epoch": 0.4178873697916667,
      "grad_norm": 16.44963836669922,
      "learning_rate": 3.142282541034923e-06,
      "loss": 3.0206,
      "step": 41080
    },
    {
      "epoch": 0.417938232421875,
      "grad_norm": 8.901022911071777,
      "learning_rate": 3.141896074068363e-06,
      "loss": 3.0547,
      "step": 41085
    },
    {
      "epoch": 0.4179890950520833,
      "grad_norm": 12.60720157623291,
      "learning_rate": 3.1415095906790436e-06,
      "loss": 3.4113,
      "step": 41090
    },
    {
      "epoch": 0.4180399576822917,
      "grad_norm": 12.682599067687988,
      "learning_rate": 3.141123090876854e-06,
      "loss": 3.7857,
      "step": 41095
    },
    {
      "epoch": 0.4180908203125,
      "grad_norm": 8.320108413696289,
      "learning_rate": 3.1407365746716816e-06,
      "loss": 3.243,
      "step": 41100
    },
    {
      "epoch": 0.4181416829427083,
      "grad_norm": 13.9808931350708,
      "learning_rate": 3.140350042073416e-06,
      "loss": 4.0655,
      "step": 41105
    },
    {
      "epoch": 0.4181925455729167,
      "grad_norm": 9.053630828857422,
      "learning_rate": 3.1399634930919463e-06,
      "loss": 3.3757,
      "step": 41110
    },
    {
      "epoch": 0.418243408203125,
      "grad_norm": 12.047245979309082,
      "learning_rate": 3.1395769277371624e-06,
      "loss": 3.209,
      "step": 41115
    },
    {
      "epoch": 0.4182942708333333,
      "grad_norm": 15.260944366455078,
      "learning_rate": 3.1391903460189543e-06,
      "loss": 3.4581,
      "step": 41120
    },
    {
      "epoch": 0.4183451334635417,
      "grad_norm": 8.774908065795898,
      "learning_rate": 3.138803747947213e-06,
      "loss": 3.1256,
      "step": 41125
    },
    {
      "epoch": 0.41839599609375,
      "grad_norm": 8.514405250549316,
      "learning_rate": 3.138417133531829e-06,
      "loss": 3.3727,
      "step": 41130
    },
    {
      "epoch": 0.4184468587239583,
      "grad_norm": 11.792862892150879,
      "learning_rate": 3.1380305027826933e-06,
      "loss": 3.5753,
      "step": 41135
    },
    {
      "epoch": 0.4184977213541667,
      "grad_norm": 15.699068069458008,
      "learning_rate": 3.1376438557096995e-06,
      "loss": 3.6065,
      "step": 41140
    },
    {
      "epoch": 0.418548583984375,
      "grad_norm": 11.407424926757812,
      "learning_rate": 3.137257192322738e-06,
      "loss": 3.1444,
      "step": 41145
    },
    {
      "epoch": 0.4185994466145833,
      "grad_norm": 10.706039428710938,
      "learning_rate": 3.136870512631703e-06,
      "loss": 3.0158,
      "step": 41150
    },
    {
      "epoch": 0.4186503092447917,
      "grad_norm": 8.059556007385254,
      "learning_rate": 3.136483816646486e-06,
      "loss": 3.3934,
      "step": 41155
    },
    {
      "epoch": 0.418701171875,
      "grad_norm": 14.3755521774292,
      "learning_rate": 3.136097104376982e-06,
      "loss": 3.1492,
      "step": 41160
    },
    {
      "epoch": 0.4187520345052083,
      "grad_norm": 12.719627380371094,
      "learning_rate": 3.1357103758330844e-06,
      "loss": 3.523,
      "step": 41165
    },
    {
      "epoch": 0.4188028971354167,
      "grad_norm": 10.903666496276855,
      "learning_rate": 3.1353236310246878e-06,
      "loss": 3.41,
      "step": 41170
    },
    {
      "epoch": 0.418853759765625,
      "grad_norm": 14.709659576416016,
      "learning_rate": 3.1349368699616857e-06,
      "loss": 3.1697,
      "step": 41175
    },
    {
      "epoch": 0.4189046223958333,
      "grad_norm": 9.065529823303223,
      "learning_rate": 3.1345500926539753e-06,
      "loss": 3.2718,
      "step": 41180
    },
    {
      "epoch": 0.4189554850260417,
      "grad_norm": 14.227614402770996,
      "learning_rate": 3.13416329911145e-06,
      "loss": 3.3014,
      "step": 41185
    },
    {
      "epoch": 0.41900634765625,
      "grad_norm": 12.289112091064453,
      "learning_rate": 3.1337764893440086e-06,
      "loss": 3.155,
      "step": 41190
    },
    {
      "epoch": 0.4190572102864583,
      "grad_norm": 13.675725936889648,
      "learning_rate": 3.1333896633615444e-06,
      "loss": 3.2811,
      "step": 41195
    },
    {
      "epoch": 0.4191080729166667,
      "grad_norm": 15.298088073730469,
      "learning_rate": 3.133002821173957e-06,
      "loss": 3.3435,
      "step": 41200
    },
    {
      "epoch": 0.419158935546875,
      "grad_norm": 10.695393562316895,
      "learning_rate": 3.132615962791142e-06,
      "loss": 3.4197,
      "step": 41205
    },
    {
      "epoch": 0.4192097981770833,
      "grad_norm": 9.276285171508789,
      "learning_rate": 3.132229088222997e-06,
      "loss": 3.8276,
      "step": 41210
    },
    {
      "epoch": 0.4192606608072917,
      "grad_norm": 8.871838569641113,
      "learning_rate": 3.131842197479421e-06,
      "loss": 3.3686,
      "step": 41215
    },
    {
      "epoch": 0.4193115234375,
      "grad_norm": 11.672993659973145,
      "learning_rate": 3.1314552905703126e-06,
      "loss": 3.384,
      "step": 41220
    },
    {
      "epoch": 0.4193623860677083,
      "grad_norm": 9.316916465759277,
      "learning_rate": 3.13106836750557e-06,
      "loss": 3.2806,
      "step": 41225
    },
    {
      "epoch": 0.4194132486979167,
      "grad_norm": 10.380117416381836,
      "learning_rate": 3.1306814282950926e-06,
      "loss": 3.4212,
      "step": 41230
    },
    {
      "epoch": 0.419464111328125,
      "grad_norm": 12.29304313659668,
      "learning_rate": 3.1302944729487804e-06,
      "loss": 3.5514,
      "step": 41235
    },
    {
      "epoch": 0.4195149739583333,
      "grad_norm": 16.73832130432129,
      "learning_rate": 3.1299075014765334e-06,
      "loss": 3.4344,
      "step": 41240
    },
    {
      "epoch": 0.4195658365885417,
      "grad_norm": 15.393379211425781,
      "learning_rate": 3.129520513888253e-06,
      "loss": 3.0878,
      "step": 41245
    },
    {
      "epoch": 0.41961669921875,
      "grad_norm": 8.426596641540527,
      "learning_rate": 3.129133510193838e-06,
      "loss": 3.2919,
      "step": 41250
    },
    {
      "epoch": 0.4196675618489583,
      "grad_norm": 14.54990291595459,
      "learning_rate": 3.128746490403193e-06,
      "loss": 3.2858,
      "step": 41255
    },
    {
      "epoch": 0.4197184244791667,
      "grad_norm": 11.128558158874512,
      "learning_rate": 3.1283594545262163e-06,
      "loss": 3.3531,
      "step": 41260
    },
    {
      "epoch": 0.419769287109375,
      "grad_norm": 7.920388698577881,
      "learning_rate": 3.1279724025728137e-06,
      "loss": 3.1741,
      "step": 41265
    },
    {
      "epoch": 0.4198201497395833,
      "grad_norm": 14.856907844543457,
      "learning_rate": 3.127585334552885e-06,
      "loss": 4.0395,
      "step": 41270
    },
    {
      "epoch": 0.4198710123697917,
      "grad_norm": 12.031919479370117,
      "learning_rate": 3.127198250476334e-06,
      "loss": 3.2902,
      "step": 41275
    },
    {
      "epoch": 0.419921875,
      "grad_norm": 8.964862823486328,
      "learning_rate": 3.1268111503530652e-06,
      "loss": 3.4011,
      "step": 41280
    },
    {
      "epoch": 0.4199727376302083,
      "grad_norm": 15.2599458694458,
      "learning_rate": 3.126424034192981e-06,
      "loss": 3.0933,
      "step": 41285
    },
    {
      "epoch": 0.4200236002604167,
      "grad_norm": 10.90890884399414,
      "learning_rate": 3.1260369020059864e-06,
      "loss": 3.1269,
      "step": 41290
    },
    {
      "epoch": 0.420074462890625,
      "grad_norm": 16.50959587097168,
      "learning_rate": 3.1256497538019875e-06,
      "loss": 3.8762,
      "step": 41295
    },
    {
      "epoch": 0.4201253255208333,
      "grad_norm": 15.000621795654297,
      "learning_rate": 3.125262589590886e-06,
      "loss": 3.6052,
      "step": 41300
    },
    {
      "epoch": 0.4201761881510417,
      "grad_norm": 15.3053560256958,
      "learning_rate": 3.124875409382591e-06,
      "loss": 3.4111,
      "step": 41305
    },
    {
      "epoch": 0.42022705078125,
      "grad_norm": 14.705530166625977,
      "learning_rate": 3.1244882131870056e-06,
      "loss": 3.3694,
      "step": 41310
    },
    {
      "epoch": 0.4202779134114583,
      "grad_norm": 14.04732894897461,
      "learning_rate": 3.124101001014037e-06,
      "loss": 3.4689,
      "step": 41315
    },
    {
      "epoch": 0.4203287760416667,
      "grad_norm": 10.53316879272461,
      "learning_rate": 3.123713772873593e-06,
      "loss": 3.1923,
      "step": 41320
    },
    {
      "epoch": 0.420379638671875,
      "grad_norm": 13.080552101135254,
      "learning_rate": 3.1233265287755793e-06,
      "loss": 2.9996,
      "step": 41325
    },
    {
      "epoch": 0.4204305013020833,
      "grad_norm": 10.696488380432129,
      "learning_rate": 3.1229392687299056e-06,
      "loss": 3.5193,
      "step": 41330
    },
    {
      "epoch": 0.4204813639322917,
      "grad_norm": 14.827099800109863,
      "learning_rate": 3.122551992746477e-06,
      "loss": 2.8518,
      "step": 41335
    },
    {
      "epoch": 0.4205322265625,
      "grad_norm": 12.05504322052002,
      "learning_rate": 3.1221647008352035e-06,
      "loss": 3.7252,
      "step": 41340
    },
    {
      "epoch": 0.4205830891927083,
      "grad_norm": 12.480534553527832,
      "learning_rate": 3.1217773930059936e-06,
      "loss": 3.0831,
      "step": 41345
    },
    {
      "epoch": 0.4206339518229167,
      "grad_norm": 9.91734790802002,
      "learning_rate": 3.1213900692687565e-06,
      "loss": 2.9163,
      "step": 41350
    },
    {
      "epoch": 0.420684814453125,
      "grad_norm": 16.006969451904297,
      "learning_rate": 3.121002729633402e-06,
      "loss": 3.2898,
      "step": 41355
    },
    {
      "epoch": 0.4207356770833333,
      "grad_norm": 9.468695640563965,
      "learning_rate": 3.1206153741098397e-06,
      "loss": 3.2785,
      "step": 41360
    },
    {
      "epoch": 0.4207865397135417,
      "grad_norm": 13.025923728942871,
      "learning_rate": 3.1202280027079797e-06,
      "loss": 3.3407,
      "step": 41365
    },
    {
      "epoch": 0.42083740234375,
      "grad_norm": 8.222331047058105,
      "learning_rate": 3.119840615437734e-06,
      "loss": 3.6244,
      "step": 41370
    },
    {
      "epoch": 0.4208882649739583,
      "grad_norm": 8.104004859924316,
      "learning_rate": 3.1194532123090125e-06,
      "loss": 3.0128,
      "step": 41375
    },
    {
      "epoch": 0.4209391276041667,
      "grad_norm": 15.87883186340332,
      "learning_rate": 3.119065793331728e-06,
      "loss": 3.5668,
      "step": 41380
    },
    {
      "epoch": 0.420989990234375,
      "grad_norm": 9.6399564743042,
      "learning_rate": 3.1186783585157913e-06,
      "loss": 3.4511,
      "step": 41385
    },
    {
      "epoch": 0.4210408528645833,
      "grad_norm": 14.242037773132324,
      "learning_rate": 3.1182909078711154e-06,
      "loss": 3.3657,
      "step": 41390
    },
    {
      "epoch": 0.4210917154947917,
      "grad_norm": 15.923033714294434,
      "learning_rate": 3.1179034414076138e-06,
      "loss": 3.1755,
      "step": 41395
    },
    {
      "epoch": 0.421142578125,
      "grad_norm": 10.9227876663208,
      "learning_rate": 3.117515959135199e-06,
      "loss": 3.2076,
      "step": 41400
    },
    {
      "epoch": 0.4211934407552083,
      "grad_norm": 13.24682331085205,
      "learning_rate": 3.117128461063784e-06,
      "loss": 3.177,
      "step": 41405
    },
    {
      "epoch": 0.4212443033854167,
      "grad_norm": 14.045320510864258,
      "learning_rate": 3.116740947203285e-06,
      "loss": 3.5047,
      "step": 41410
    },
    {
      "epoch": 0.421295166015625,
      "grad_norm": 12.409698486328125,
      "learning_rate": 3.1163534175636134e-06,
      "loss": 3.2689,
      "step": 41415
    },
    {
      "epoch": 0.4213460286458333,
      "grad_norm": 8.52938461303711,
      "learning_rate": 3.1159658721546866e-06,
      "loss": 3.4171,
      "step": 41420
    },
    {
      "epoch": 0.4213968912760417,
      "grad_norm": 7.385495185852051,
      "learning_rate": 3.115578310986419e-06,
      "loss": 3.3524,
      "step": 41425
    },
    {
      "epoch": 0.42144775390625,
      "grad_norm": 17.02286720275879,
      "learning_rate": 3.1151907340687256e-06,
      "loss": 3.314,
      "step": 41430
    },
    {
      "epoch": 0.4214986165364583,
      "grad_norm": 14.409526824951172,
      "learning_rate": 3.114803141411524e-06,
      "loss": 2.9813,
      "step": 41435
    },
    {
      "epoch": 0.4215494791666667,
      "grad_norm": 9.918559074401855,
      "learning_rate": 3.1144155330247292e-06,
      "loss": 3.5645,
      "step": 41440
    },
    {
      "epoch": 0.421600341796875,
      "grad_norm": 14.866680145263672,
      "learning_rate": 3.1140279089182584e-06,
      "loss": 3.3423,
      "step": 41445
    },
    {
      "epoch": 0.4216512044270833,
      "grad_norm": 13.22377872467041,
      "learning_rate": 3.1136402691020294e-06,
      "loss": 3.5402,
      "step": 41450
    },
    {
      "epoch": 0.4217020670572917,
      "grad_norm": 12.47979736328125,
      "learning_rate": 3.11325261358596e-06,
      "loss": 2.97,
      "step": 41455
    },
    {
      "epoch": 0.4217529296875,
      "grad_norm": 8.604597091674805,
      "learning_rate": 3.112864942379967e-06,
      "loss": 3.2102,
      "step": 41460
    },
    {
      "epoch": 0.4218037923177083,
      "grad_norm": 12.656006813049316,
      "learning_rate": 3.112477255493971e-06,
      "loss": 4.0634,
      "step": 41465
    },
    {
      "epoch": 0.4218546549479167,
      "grad_norm": 16.776290893554688,
      "learning_rate": 3.112089552937888e-06,
      "loss": 3.5199,
      "step": 41470
    },
    {
      "epoch": 0.421905517578125,
      "grad_norm": 17.225765228271484,
      "learning_rate": 3.1117018347216398e-06,
      "loss": 3.5333,
      "step": 41475
    },
    {
      "epoch": 0.4219563802083333,
      "grad_norm": 13.472655296325684,
      "learning_rate": 3.1113141008551446e-06,
      "loss": 3.4921,
      "step": 41480
    },
    {
      "epoch": 0.4220072428385417,
      "grad_norm": 12.355141639709473,
      "learning_rate": 3.1109263513483234e-06,
      "loss": 3.3573,
      "step": 41485
    },
    {
      "epoch": 0.42205810546875,
      "grad_norm": 12.142220497131348,
      "learning_rate": 3.1105385862110964e-06,
      "loss": 3.428,
      "step": 41490
    },
    {
      "epoch": 0.4221089680989583,
      "grad_norm": 8.77768611907959,
      "learning_rate": 3.110150805453384e-06,
      "loss": 3.4354,
      "step": 41495
    },
    {
      "epoch": 0.4221598307291667,
      "grad_norm": 19.088306427001953,
      "learning_rate": 3.1097630090851084e-06,
      "loss": 3.127,
      "step": 41500
    },
    {
      "epoch": 0.422210693359375,
      "grad_norm": 13.503829002380371,
      "learning_rate": 3.10937519711619e-06,
      "loss": 3.7707,
      "step": 41505
    },
    {
      "epoch": 0.4222615559895833,
      "grad_norm": 11.96296215057373,
      "learning_rate": 3.1089873695565523e-06,
      "loss": 3.2588,
      "step": 41510
    },
    {
      "epoch": 0.4223124186197917,
      "grad_norm": 13.315999984741211,
      "learning_rate": 3.1085995264161167e-06,
      "loss": 3.3339,
      "step": 41515
    },
    {
      "epoch": 0.42236328125,
      "grad_norm": 14.427928924560547,
      "learning_rate": 3.108211667704807e-06,
      "loss": 3.5766,
      "step": 41520
    },
    {
      "epoch": 0.4224141438802083,
      "grad_norm": 14.34740924835205,
      "learning_rate": 3.107823793432545e-06,
      "loss": 3.1351,
      "step": 41525
    },
    {
      "epoch": 0.4224650065104167,
      "grad_norm": 11.97638988494873,
      "learning_rate": 3.107435903609257e-06,
      "loss": 3.3518,
      "step": 41530
    },
    {
      "epoch": 0.422515869140625,
      "grad_norm": 14.582337379455566,
      "learning_rate": 3.107047998244863e-06,
      "loss": 3.1531,
      "step": 41535
    },
    {
      "epoch": 0.4225667317708333,
      "grad_norm": 12.777178764343262,
      "learning_rate": 3.106660077349292e-06,
      "loss": 3.6059,
      "step": 41540
    },
    {
      "epoch": 0.4226175944010417,
      "grad_norm": 15.88293170928955,
      "learning_rate": 3.106272140932466e-06,
      "loss": 3.6732,
      "step": 41545
    },
    {
      "epoch": 0.42266845703125,
      "grad_norm": 10.44229507446289,
      "learning_rate": 3.1058841890043105e-06,
      "loss": 3.8964,
      "step": 41550
    },
    {
      "epoch": 0.4227193196614583,
      "grad_norm": 11.0337553024292,
      "learning_rate": 3.105496221574752e-06,
      "loss": 3.2802,
      "step": 41555
    },
    {
      "epoch": 0.4227701822916667,
      "grad_norm": 10.296064376831055,
      "learning_rate": 3.105108238653716e-06,
      "loss": 3.2045,
      "step": 41560
    },
    {
      "epoch": 0.422821044921875,
      "grad_norm": 14.51941204071045,
      "learning_rate": 3.1047202402511294e-06,
      "loss": 3.7077,
      "step": 41565
    },
    {
      "epoch": 0.4228719075520833,
      "grad_norm": 12.24843978881836,
      "learning_rate": 3.104332226376919e-06,
      "loss": 3.1883,
      "step": 41570
    },
    {
      "epoch": 0.4229227701822917,
      "grad_norm": 11.264163970947266,
      "learning_rate": 3.103944197041011e-06,
      "loss": 3.4847,
      "step": 41575
    },
    {
      "epoch": 0.4229736328125,
      "grad_norm": 11.658236503601074,
      "learning_rate": 3.103556152253335e-06,
      "loss": 3.2653,
      "step": 41580
    },
    {
      "epoch": 0.4230244954427083,
      "grad_norm": 15.09562873840332,
      "learning_rate": 3.1031680920238165e-06,
      "loss": 3.1765,
      "step": 41585
    },
    {
      "epoch": 0.4230753580729167,
      "grad_norm": 14.389519691467285,
      "learning_rate": 3.1027800163623866e-06,
      "loss": 3.8784,
      "step": 41590
    },
    {
      "epoch": 0.423126220703125,
      "grad_norm": 13.224161148071289,
      "learning_rate": 3.1023919252789724e-06,
      "loss": 3.4739,
      "step": 41595
    },
    {
      "epoch": 0.4231770833333333,
      "grad_norm": 11.407111167907715,
      "learning_rate": 3.102003818783503e-06,
      "loss": 3.22,
      "step": 41600
    },
    {
      "epoch": 0.4232279459635417,
      "grad_norm": 12.572507858276367,
      "learning_rate": 3.1016156968859095e-06,
      "loss": 3.3877,
      "step": 41605
    },
    {
      "epoch": 0.42327880859375,
      "grad_norm": 9.610457420349121,
      "learning_rate": 3.1012275595961204e-06,
      "loss": 3.2691,
      "step": 41610
    },
    {
      "epoch": 0.4233296712239583,
      "grad_norm": 12.857911109924316,
      "learning_rate": 3.1008394069240672e-06,
      "loss": 3.4903,
      "step": 41615
    },
    {
      "epoch": 0.4233805338541667,
      "grad_norm": 10.919528007507324,
      "learning_rate": 3.10045123887968e-06,
      "loss": 3.2011,
      "step": 41620
    },
    {
      "epoch": 0.423431396484375,
      "grad_norm": 16.585697174072266,
      "learning_rate": 3.10006305547289e-06,
      "loss": 3.4325,
      "step": 41625
    },
    {
      "epoch": 0.4234822591145833,
      "grad_norm": 18.34198570251465,
      "learning_rate": 3.0996748567136286e-06,
      "loss": 3.4349,
      "step": 41630
    },
    {
      "epoch": 0.4235331217447917,
      "grad_norm": 16.982179641723633,
      "learning_rate": 3.0992866426118294e-06,
      "loss": 3.3223,
      "step": 41635
    },
    {
      "epoch": 0.423583984375,
      "grad_norm": 14.218949317932129,
      "learning_rate": 3.098898413177422e-06,
      "loss": 3.4303,
      "step": 41640
    },
    {
      "epoch": 0.4236348470052083,
      "grad_norm": 7.693777561187744,
      "learning_rate": 3.098510168420342e-06,
      "loss": 3.1307,
      "step": 41645
    },
    {
      "epoch": 0.4236857096354167,
      "grad_norm": 11.543193817138672,
      "learning_rate": 3.09812190835052e-06,
      "loss": 2.9734,
      "step": 41650
    },
    {
      "epoch": 0.423736572265625,
      "grad_norm": 14.944109916687012,
      "learning_rate": 3.097733632977891e-06,
      "loss": 3.3276,
      "step": 41655
    },
    {
      "epoch": 0.4237874348958333,
      "grad_norm": 10.503029823303223,
      "learning_rate": 3.0973453423123895e-06,
      "loss": 3.3436,
      "step": 41660
    },
    {
      "epoch": 0.4238382975260417,
      "grad_norm": 12.030123710632324,
      "learning_rate": 3.096957036363948e-06,
      "loss": 3.5391,
      "step": 41665
    },
    {
      "epoch": 0.42388916015625,
      "grad_norm": 14.35947036743164,
      "learning_rate": 3.096568715142503e-06,
      "loss": 3.2357,
      "step": 41670
    },
    {
      "epoch": 0.4239400227864583,
      "grad_norm": 18.176090240478516,
      "learning_rate": 3.096180378657988e-06,
      "loss": 3.8218,
      "step": 41675
    },
    {
      "epoch": 0.4239908854166667,
      "grad_norm": 14.452081680297852,
      "learning_rate": 3.09579202692034e-06,
      "loss": 3.3728,
      "step": 41680
    },
    {
      "epoch": 0.424041748046875,
      "grad_norm": 8.979386329650879,
      "learning_rate": 3.0954036599394945e-06,
      "loss": 3.3105,
      "step": 41685
    },
    {
      "epoch": 0.4240926106770833,
      "grad_norm": 9.725926399230957,
      "learning_rate": 3.0950152777253866e-06,
      "loss": 3.2534,
      "step": 41690
    },
    {
      "epoch": 0.4241434733072917,
      "grad_norm": 10.265632629394531,
      "learning_rate": 3.094626880287955e-06,
      "loss": 3.2297,
      "step": 41695
    },
    {
      "epoch": 0.4241943359375,
      "grad_norm": 12.90949821472168,
      "learning_rate": 3.0942384676371346e-06,
      "loss": 3.2365,
      "step": 41700
    },
    {
      "epoch": 0.4242451985677083,
      "grad_norm": 16.56032943725586,
      "learning_rate": 3.093850039782864e-06,
      "loss": 3.3633,
      "step": 41705
    },
    {
      "epoch": 0.4242960611979167,
      "grad_norm": 9.831174850463867,
      "learning_rate": 3.093461596735081e-06,
      "loss": 3.3728,
      "step": 41710
    },
    {
      "epoch": 0.424346923828125,
      "grad_norm": 11.747597694396973,
      "learning_rate": 3.093073138503723e-06,
      "loss": 3.2889,
      "step": 41715
    },
    {
      "epoch": 0.4243977864583333,
      "grad_norm": 15.865684509277344,
      "learning_rate": 3.092684665098731e-06,
      "loss": 3.9162,
      "step": 41720
    },
    {
      "epoch": 0.4244486490885417,
      "grad_norm": 14.615534782409668,
      "learning_rate": 3.0922961765300407e-06,
      "loss": 3.1447,
      "step": 41725
    },
    {
      "epoch": 0.42449951171875,
      "grad_norm": 15.222784996032715,
      "learning_rate": 3.091907672807594e-06,
      "loss": 3.5148,
      "step": 41730
    },
    {
      "epoch": 0.4245503743489583,
      "grad_norm": 25.343584060668945,
      "learning_rate": 3.091519153941329e-06,
      "loss": 3.6415,
      "step": 41735
    },
    {
      "epoch": 0.4246012369791667,
      "grad_norm": 13.214790344238281,
      "learning_rate": 3.0911306199411876e-06,
      "loss": 3.4832,
      "step": 41740
    },
    {
      "epoch": 0.424652099609375,
      "grad_norm": 12.833844184875488,
      "learning_rate": 3.090742070817109e-06,
      "loss": 3.3397,
      "step": 41745
    },
    {
      "epoch": 0.4247029622395833,
      "grad_norm": 12.496012687683105,
      "learning_rate": 3.0903535065790353e-06,
      "loss": 3.5337,
      "step": 41750
    },
    {
      "epoch": 0.4247538248697917,
      "grad_norm": 8.92209243774414,
      "learning_rate": 3.0899649272369054e-06,
      "loss": 3.48,
      "step": 41755
    },
    {
      "epoch": 0.4248046875,
      "grad_norm": 9.917067527770996,
      "learning_rate": 3.0895763328006644e-06,
      "loss": 3.2993,
      "step": 41760
    },
    {
      "epoch": 0.4248555501302083,
      "grad_norm": 8.58231258392334,
      "learning_rate": 3.0891877232802515e-06,
      "loss": 3.6489,
      "step": 41765
    },
    {
      "epoch": 0.4249064127604167,
      "grad_norm": 10.70272445678711,
      "learning_rate": 3.088799098685611e-06,
      "loss": 2.9813,
      "step": 41770
    },
    {
      "epoch": 0.424957275390625,
      "grad_norm": 10.071898460388184,
      "learning_rate": 3.0884104590266844e-06,
      "loss": 3.4546,
      "step": 41775
    },
    {
      "epoch": 0.4250081380208333,
      "grad_norm": 14.641159057617188,
      "learning_rate": 3.0880218043134164e-06,
      "loss": 3.164,
      "step": 41780
    },
    {
      "epoch": 0.4250590006510417,
      "grad_norm": 15.49095344543457,
      "learning_rate": 3.0876331345557498e-06,
      "loss": 3.4119,
      "step": 41785
    },
    {
      "epoch": 0.42510986328125,
      "grad_norm": 9.982519149780273,
      "learning_rate": 3.087244449763629e-06,
      "loss": 3.6413,
      "step": 41790
    },
    {
      "epoch": 0.4251607259114583,
      "grad_norm": 10.193666458129883,
      "learning_rate": 3.086855749946997e-06,
      "loss": 3.2032,
      "step": 41795
    },
    {
      "epoch": 0.4252115885416667,
      "grad_norm": 15.675253868103027,
      "learning_rate": 3.0864670351158012e-06,
      "loss": 3.5083,
      "step": 41800
    },
    {
      "epoch": 0.425262451171875,
      "grad_norm": 11.914649963378906,
      "learning_rate": 3.0860783052799837e-06,
      "loss": 3.4752,
      "step": 41805
    },
    {
      "epoch": 0.4253133138020833,
      "grad_norm": 11.520695686340332,
      "learning_rate": 3.085689560449492e-06,
      "loss": 3.3944,
      "step": 41810
    },
    {
      "epoch": 0.4253641764322917,
      "grad_norm": 12.948492050170898,
      "learning_rate": 3.0853008006342734e-06,
      "loss": 3.4788,
      "step": 41815
    },
    {
      "epoch": 0.4254150390625,
      "grad_norm": 14.214773178100586,
      "learning_rate": 3.0849120258442706e-06,
      "loss": 2.8699,
      "step": 41820
    },
    {
      "epoch": 0.4254659016927083,
      "grad_norm": 13.516385078430176,
      "learning_rate": 3.084523236089434e-06,
      "loss": 3.3412,
      "step": 41825
    },
    {
      "epoch": 0.4255167643229167,
      "grad_norm": 15.694234848022461,
      "learning_rate": 3.084134431379708e-06,
      "loss": 2.9498,
      "step": 41830
    },
    {
      "epoch": 0.425567626953125,
      "grad_norm": 10.564866065979004,
      "learning_rate": 3.083745611725041e-06,
      "loss": 3.2879,
      "step": 41835
    },
    {
      "epoch": 0.4256184895833333,
      "grad_norm": 9.610162734985352,
      "learning_rate": 3.0833567771353806e-06,
      "loss": 3.6513,
      "step": 41840
    },
    {
      "epoch": 0.4256693522135417,
      "grad_norm": 17.4929141998291,
      "learning_rate": 3.082967927620676e-06,
      "loss": 3.3903,
      "step": 41845
    },
    {
      "epoch": 0.42572021484375,
      "grad_norm": 12.404966354370117,
      "learning_rate": 3.082579063190874e-06,
      "loss": 3.2674,
      "step": 41850
    },
    {
      "epoch": 0.4257710774739583,
      "grad_norm": 8.494036674499512,
      "learning_rate": 3.0821901838559263e-06,
      "loss": 3.5712,
      "step": 41855
    },
    {
      "epoch": 0.4258219401041667,
      "grad_norm": 12.155200958251953,
      "learning_rate": 3.0818012896257797e-06,
      "loss": 3.2202,
      "step": 41860
    },
    {
      "epoch": 0.425872802734375,
      "grad_norm": 9.61408805847168,
      "learning_rate": 3.081412380510386e-06,
      "loss": 3.0132,
      "step": 41865
    },
    {
      "epoch": 0.4259236653645833,
      "grad_norm": 10.762458801269531,
      "learning_rate": 3.081023456519693e-06,
      "loss": 3.2979,
      "step": 41870
    },
    {
      "epoch": 0.4259745279947917,
      "grad_norm": 9.656315803527832,
      "learning_rate": 3.0806345176636536e-06,
      "loss": 3.2092,
      "step": 41875
    },
    {
      "epoch": 0.426025390625,
      "grad_norm": 15.16588306427002,
      "learning_rate": 3.080245563952217e-06,
      "loss": 3.336,
      "step": 41880
    },
    {
      "epoch": 0.4260762532552083,
      "grad_norm": 13.997447967529297,
      "learning_rate": 3.079856595395336e-06,
      "loss": 3.2055,
      "step": 41885
    },
    {
      "epoch": 0.4261271158854167,
      "grad_norm": 9.956498146057129,
      "learning_rate": 3.0794676120029605e-06,
      "loss": 3.5072,
      "step": 41890
    },
    {
      "epoch": 0.426177978515625,
      "grad_norm": 11.688020706176758,
      "learning_rate": 3.0790786137850442e-06,
      "loss": 3.2936,
      "step": 41895
    },
    {
      "epoch": 0.4262288411458333,
      "grad_norm": 10.722555160522461,
      "learning_rate": 3.0786896007515383e-06,
      "loss": 3.178,
      "step": 41900
    },
    {
      "epoch": 0.4262797037760417,
      "grad_norm": 12.866437911987305,
      "learning_rate": 3.078300572912397e-06,
      "loss": 3.2504,
      "step": 41905
    },
    {
      "epoch": 0.42633056640625,
      "grad_norm": 11.225545883178711,
      "learning_rate": 3.0779115302775713e-06,
      "loss": 3.1698,
      "step": 41910
    },
    {
      "epoch": 0.4263814290364583,
      "grad_norm": 9.123448371887207,
      "learning_rate": 3.077522472857017e-06,
      "loss": 3.5688,
      "step": 41915
    },
    {
      "epoch": 0.4264322916666667,
      "grad_norm": 17.776391983032227,
      "learning_rate": 3.077133400660687e-06,
      "loss": 3.4547,
      "step": 41920
    },
    {
      "epoch": 0.426483154296875,
      "grad_norm": 7.346116542816162,
      "learning_rate": 3.076744313698535e-06,
      "loss": 3.1451,
      "step": 41925
    },
    {
      "epoch": 0.4265340169270833,
      "grad_norm": 20.024919509887695,
      "learning_rate": 3.076355211980518e-06,
      "loss": 3.3667,
      "step": 41930
    },
    {
      "epoch": 0.4265848795572917,
      "grad_norm": 14.997565269470215,
      "learning_rate": 3.0759660955165887e-06,
      "loss": 3.0529,
      "step": 41935
    },
    {
      "epoch": 0.4266357421875,
      "grad_norm": 13.311877250671387,
      "learning_rate": 3.075576964316703e-06,
      "loss": 3.4168,
      "step": 41940
    },
    {
      "epoch": 0.4266866048177083,
      "grad_norm": 14.687515258789062,
      "learning_rate": 3.0751878183908175e-06,
      "loss": 2.9444,
      "step": 41945
    },
    {
      "epoch": 0.4267374674479167,
      "grad_norm": 15.228010177612305,
      "learning_rate": 3.074798657748888e-06,
      "loss": 3.3023,
      "step": 41950
    },
    {
      "epoch": 0.426788330078125,
      "grad_norm": 15.780599594116211,
      "learning_rate": 3.0744094824008707e-06,
      "loss": 3.3186,
      "step": 41955
    },
    {
      "epoch": 0.4268391927083333,
      "grad_norm": 11.762367248535156,
      "learning_rate": 3.074020292356723e-06,
      "loss": 3.5871,
      "step": 41960
    },
    {
      "epoch": 0.4268900553385417,
      "grad_norm": 9.062051773071289,
      "learning_rate": 3.0736310876264024e-06,
      "loss": 3.0746,
      "step": 41965
    },
    {
      "epoch": 0.42694091796875,
      "grad_norm": 11.719466209411621,
      "learning_rate": 3.073241868219867e-06,
      "loss": 3.362,
      "step": 41970
    },
    {
      "epoch": 0.4269917805989583,
      "grad_norm": 16.24999237060547,
      "learning_rate": 3.072852634147073e-06,
      "loss": 3.2346,
      "step": 41975
    },
    {
      "epoch": 0.4270426432291667,
      "grad_norm": 12.696512222290039,
      "learning_rate": 3.0724633854179814e-06,
      "loss": 3.3545,
      "step": 41980
    },
    {
      "epoch": 0.427093505859375,
      "grad_norm": 9.804299354553223,
      "learning_rate": 3.072074122042549e-06,
      "loss": 2.9765,
      "step": 41985
    },
    {
      "epoch": 0.4271443684895833,
      "grad_norm": 14.234360694885254,
      "learning_rate": 3.071684844030736e-06,
      "loss": 3.2353,
      "step": 41990
    },
    {
      "epoch": 0.4271952311197917,
      "grad_norm": 9.232892990112305,
      "learning_rate": 3.0712955513925017e-06,
      "loss": 3.9413,
      "step": 41995
    },
    {
      "epoch": 0.42724609375,
      "grad_norm": 10.719136238098145,
      "learning_rate": 3.0709062441378062e-06,
      "loss": 3.2093,
      "step": 42000
    },
    {
      "epoch": 0.4272969563802083,
      "grad_norm": 11.004392623901367,
      "learning_rate": 3.0705169222766095e-06,
      "loss": 3.4891,
      "step": 42005
    },
    {
      "epoch": 0.4273478190104167,
      "grad_norm": 8.278154373168945,
      "learning_rate": 3.070127585818873e-06,
      "loss": 3.3757,
      "step": 42010
    },
    {
      "epoch": 0.427398681640625,
      "grad_norm": 13.0377836227417,
      "learning_rate": 3.069738234774557e-06,
      "loss": 3.4463,
      "step": 42015
    },
    {
      "epoch": 0.4274495442708333,
      "grad_norm": 11.712611198425293,
      "learning_rate": 3.069348869153623e-06,
      "loss": 3.3497,
      "step": 42020
    },
    {
      "epoch": 0.4275004069010417,
      "grad_norm": 10.128121376037598,
      "learning_rate": 3.068959488966034e-06,
      "loss": 3.2071,
      "step": 42025
    },
    {
      "epoch": 0.42755126953125,
      "grad_norm": 16.13844108581543,
      "learning_rate": 3.0685700942217506e-06,
      "loss": 3.2666,
      "step": 42030
    },
    {
      "epoch": 0.4276021321614583,
      "grad_norm": 11.657336235046387,
      "learning_rate": 3.068180684930737e-06,
      "loss": 3.5323,
      "step": 42035
    },
    {
      "epoch": 0.4276529947916667,
      "grad_norm": 10.846715927124023,
      "learning_rate": 3.0677912611029544e-06,
      "loss": 3.0523,
      "step": 42040
    },
    {
      "epoch": 0.427703857421875,
      "grad_norm": 15.682669639587402,
      "learning_rate": 3.0674018227483673e-06,
      "loss": 3.2705,
      "step": 42045
    },
    {
      "epoch": 0.4277547200520833,
      "grad_norm": 9.073358535766602,
      "learning_rate": 3.067012369876939e-06,
      "loss": 3.3751,
      "step": 42050
    },
    {
      "epoch": 0.4278055826822917,
      "grad_norm": 10.69783878326416,
      "learning_rate": 3.0666229024986337e-06,
      "loss": 3.6506,
      "step": 42055
    },
    {
      "epoch": 0.4278564453125,
      "grad_norm": 15.259477615356445,
      "learning_rate": 3.0662334206234155e-06,
      "loss": 3.3437,
      "step": 42060
    },
    {
      "epoch": 0.4279073079427083,
      "grad_norm": 10.186830520629883,
      "learning_rate": 3.065843924261249e-06,
      "loss": 3.4981,
      "step": 42065
    },
    {
      "epoch": 0.4279581705729167,
      "grad_norm": 9.871879577636719,
      "learning_rate": 3.065454413422101e-06,
      "loss": 3.3813,
      "step": 42070
    },
    {
      "epoch": 0.428009033203125,
      "grad_norm": 14.064395904541016,
      "learning_rate": 3.0650648881159357e-06,
      "loss": 3.571,
      "step": 42075
    },
    {
      "epoch": 0.4280598958333333,
      "grad_norm": 11.894659996032715,
      "learning_rate": 3.064675348352718e-06,
      "loss": 3.1771,
      "step": 42080
    },
    {
      "epoch": 0.4281107584635417,
      "grad_norm": 9.00032901763916,
      "learning_rate": 3.0642857941424163e-06,
      "loss": 3.429,
      "step": 42085
    },
    {
      "epoch": 0.42816162109375,
      "grad_norm": 14.137030601501465,
      "learning_rate": 3.063896225494996e-06,
      "loss": 2.908,
      "step": 42090
    },
    {
      "epoch": 0.4282124837239583,
      "grad_norm": 11.593270301818848,
      "learning_rate": 3.0635066424204245e-06,
      "loss": 3.2668,
      "step": 42095
    },
    {
      "epoch": 0.4282633463541667,
      "grad_norm": 13.249083518981934,
      "learning_rate": 3.0631170449286695e-06,
      "loss": 3.1671,
      "step": 42100
    },
    {
      "epoch": 0.428314208984375,
      "grad_norm": 13.262222290039062,
      "learning_rate": 3.0627274330296976e-06,
      "loss": 3.1571,
      "step": 42105
    },
    {
      "epoch": 0.4283650716145833,
      "grad_norm": 15.089553833007812,
      "learning_rate": 3.0623378067334785e-06,
      "loss": 3.2122,
      "step": 42110
    },
    {
      "epoch": 0.4284159342447917,
      "grad_norm": 7.551955223083496,
      "learning_rate": 3.0619481660499795e-06,
      "loss": 2.9784,
      "step": 42115
    },
    {
      "epoch": 0.428466796875,
      "grad_norm": 10.404687881469727,
      "learning_rate": 3.06155851098917e-06,
      "loss": 3.5793,
      "step": 42120
    },
    {
      "epoch": 0.4285176595052083,
      "grad_norm": 16.388019561767578,
      "learning_rate": 3.0611688415610187e-06,
      "loss": 3.2954,
      "step": 42125
    },
    {
      "epoch": 0.4285685221354167,
      "grad_norm": 8.546836853027344,
      "learning_rate": 3.060779157775496e-06,
      "loss": 3.3968,
      "step": 42130
    },
    {
      "epoch": 0.428619384765625,
      "grad_norm": 14.341412544250488,
      "learning_rate": 3.060389459642571e-06,
      "loss": 3.6612,
      "step": 42135
    },
    {
      "epoch": 0.4286702473958333,
      "grad_norm": 10.097040176391602,
      "learning_rate": 3.0599997471722153e-06,
      "loss": 3.036,
      "step": 42140
    },
    {
      "epoch": 0.4287211100260417,
      "grad_norm": 11.047060012817383,
      "learning_rate": 3.0596100203743977e-06,
      "loss": 3.4227,
      "step": 42145
    },
    {
      "epoch": 0.42877197265625,
      "grad_norm": 11.28085708618164,
      "learning_rate": 3.059220279259092e-06,
      "loss": 3.1427,
      "step": 42150
    },
    {
      "epoch": 0.4288228352864583,
      "grad_norm": 8.427857398986816,
      "learning_rate": 3.058830523836267e-06,
      "loss": 3.4633,
      "step": 42155
    },
    {
      "epoch": 0.4288736979166667,
      "grad_norm": 12.277863502502441,
      "learning_rate": 3.0584407541158958e-06,
      "loss": 3.2633,
      "step": 42160
    },
    {
      "epoch": 0.428924560546875,
      "grad_norm": 10.43952465057373,
      "learning_rate": 3.05805097010795e-06,
      "loss": 3.2075,
      "step": 42165
    },
    {
      "epoch": 0.4289754231770833,
      "grad_norm": 12.587766647338867,
      "learning_rate": 3.057661171822402e-06,
      "loss": 3.3297,
      "step": 42170
    },
    {
      "epoch": 0.4290262858072917,
      "grad_norm": 13.791691780090332,
      "learning_rate": 3.0572713592692255e-06,
      "loss": 3.4883,
      "step": 42175
    },
    {
      "epoch": 0.4290771484375,
      "grad_norm": 9.079309463500977,
      "learning_rate": 3.0568815324583933e-06,
      "loss": 3.4208,
      "step": 42180
    },
    {
      "epoch": 0.4291280110677083,
      "grad_norm": 13.565601348876953,
      "learning_rate": 3.0564916913998787e-06,
      "loss": 3.6558,
      "step": 42185
    },
    {
      "epoch": 0.4291788736979167,
      "grad_norm": 12.191876411437988,
      "learning_rate": 3.0561018361036575e-06,
      "loss": 3.8942,
      "step": 42190
    },
    {
      "epoch": 0.429229736328125,
      "grad_norm": 10.371522903442383,
      "learning_rate": 3.0557119665797007e-06,
      "loss": 3.0518,
      "step": 42195
    },
    {
      "epoch": 0.4292805989583333,
      "grad_norm": 12.896306037902832,
      "learning_rate": 3.0553220828379867e-06,
      "loss": 3.5945,
      "step": 42200
    },
    {
      "epoch": 0.4293314615885417,
      "grad_norm": 16.03690528869629,
      "learning_rate": 3.0549321848884876e-06,
      "loss": 3.1975,
      "step": 42205
    },
    {
      "epoch": 0.42938232421875,
      "grad_norm": 13.459425926208496,
      "learning_rate": 3.0545422727411807e-06,
      "loss": 3.1506,
      "step": 42210
    },
    {
      "epoch": 0.4294331868489583,
      "grad_norm": 15.138161659240723,
      "learning_rate": 3.0541523464060414e-06,
      "loss": 3.5121,
      "step": 42215
    },
    {
      "epoch": 0.4294840494791667,
      "grad_norm": 15.504406929016113,
      "learning_rate": 3.053762405893045e-06,
      "loss": 3.0507,
      "step": 42220
    },
    {
      "epoch": 0.429534912109375,
      "grad_norm": 14.8836030960083,
      "learning_rate": 3.0533724512121686e-06,
      "loss": 3.2129,
      "step": 42225
    },
    {
      "epoch": 0.4295857747395833,
      "grad_norm": 12.824821472167969,
      "learning_rate": 3.05298248237339e-06,
      "loss": 3.2611,
      "step": 42230
    },
    {
      "epoch": 0.4296366373697917,
      "grad_norm": 14.585503578186035,
      "learning_rate": 3.052592499386685e-06,
      "loss": 3.4047,
      "step": 42235
    },
    {
      "epoch": 0.4296875,
      "grad_norm": 9.65004825592041,
      "learning_rate": 3.0522025022620323e-06,
      "loss": 3.4409,
      "step": 42240
    },
    {
      "epoch": 0.4297383626302083,
      "grad_norm": 10.6661376953125,
      "learning_rate": 3.05181249100941e-06,
      "loss": 3.4552,
      "step": 42245
    },
    {
      "epoch": 0.4297892252604167,
      "grad_norm": 13.470560073852539,
      "learning_rate": 3.0514224656387948e-06,
      "loss": 3.5744,
      "step": 42250
    },
    {
      "epoch": 0.429840087890625,
      "grad_norm": 9.623108863830566,
      "learning_rate": 3.051032426160167e-06,
      "loss": 3.9649,
      "step": 42255
    },
    {
      "epoch": 0.4298909505208333,
      "grad_norm": 15.441165924072266,
      "learning_rate": 3.050642372583505e-06,
      "loss": 3.4889,
      "step": 42260
    },
    {
      "epoch": 0.4299418131510417,
      "grad_norm": 10.242965698242188,
      "learning_rate": 3.050252304918789e-06,
      "loss": 3.4703,
      "step": 42265
    },
    {
      "epoch": 0.42999267578125,
      "grad_norm": 14.811580657958984,
      "learning_rate": 3.049862223175998e-06,
      "loss": 3.6813,
      "step": 42270
    },
    {
      "epoch": 0.4300435384114583,
      "grad_norm": 8.836296081542969,
      "learning_rate": 3.049472127365112e-06,
      "loss": 3.0087,
      "step": 42275
    },
    {
      "epoch": 0.4300944010416667,
      "grad_norm": 8.60825252532959,
      "learning_rate": 3.0490820174961115e-06,
      "loss": 3.1833,
      "step": 42280
    },
    {
      "epoch": 0.430145263671875,
      "grad_norm": 13.509065628051758,
      "learning_rate": 3.048691893578978e-06,
      "loss": 3.6716,
      "step": 42285
    },
    {
      "epoch": 0.4301961263020833,
      "grad_norm": 12.435196876525879,
      "learning_rate": 3.0483017556236926e-06,
      "loss": 3.7522,
      "step": 42290
    },
    {
      "epoch": 0.4302469889322917,
      "grad_norm": 11.747030258178711,
      "learning_rate": 3.0479116036402374e-06,
      "loss": 3.435,
      "step": 42295
    },
    {
      "epoch": 0.4302978515625,
      "grad_norm": 17.41107749938965,
      "learning_rate": 3.0475214376385927e-06,
      "loss": 3.3643,
      "step": 42300
    },
    {
      "epoch": 0.4303487141927083,
      "grad_norm": 15.284570693969727,
      "learning_rate": 3.047131257628742e-06,
      "loss": 3.0763,
      "step": 42305
    },
    {
      "epoch": 0.4303995768229167,
      "grad_norm": 12.56916332244873,
      "learning_rate": 3.0467410636206684e-06,
      "loss": 3.1925,
      "step": 42310
    },
    {
      "epoch": 0.430450439453125,
      "grad_norm": 11.555366516113281,
      "learning_rate": 3.0463508556243534e-06,
      "loss": 3.6569,
      "step": 42315
    },
    {
      "epoch": 0.4305013020833333,
      "grad_norm": 8.468916893005371,
      "learning_rate": 3.0459606336497822e-06,
      "loss": 3.2736,
      "step": 42320
    },
    {
      "epoch": 0.4305521647135417,
      "grad_norm": 15.363468170166016,
      "learning_rate": 3.045570397706936e-06,
      "loss": 3.1511,
      "step": 42325
    },
    {
      "epoch": 0.43060302734375,
      "grad_norm": 14.744574546813965,
      "learning_rate": 3.0451801478058025e-06,
      "loss": 3.0805,
      "step": 42330
    },
    {
      "epoch": 0.4306538899739583,
      "grad_norm": 8.937968254089355,
      "learning_rate": 3.0447898839563628e-06,
      "loss": 3.3218,
      "step": 42335
    },
    {
      "epoch": 0.4307047526041667,
      "grad_norm": 12.4723482131958,
      "learning_rate": 3.0443996061686037e-06,
      "loss": 3.6424,
      "step": 42340
    },
    {
      "epoch": 0.430755615234375,
      "grad_norm": 10.444912910461426,
      "learning_rate": 3.0440093144525097e-06,
      "loss": 3.2465,
      "step": 42345
    },
    {
      "epoch": 0.4308064778645833,
      "grad_norm": 12.38646411895752,
      "learning_rate": 3.0436190088180656e-06,
      "loss": 3.0429,
      "step": 42350
    },
    {
      "epoch": 0.4308573404947917,
      "grad_norm": 14.403653144836426,
      "learning_rate": 3.0432286892752583e-06,
      "loss": 3.3918,
      "step": 42355
    },
    {
      "epoch": 0.430908203125,
      "grad_norm": 9.09655475616455,
      "learning_rate": 3.0428383558340744e-06,
      "loss": 3.1717,
      "step": 42360
    },
    {
      "epoch": 0.4309590657552083,
      "grad_norm": 13.981171607971191,
      "learning_rate": 3.0424480085044987e-06,
      "loss": 3.3384,
      "step": 42365
    },
    {
      "epoch": 0.4310099283854167,
      "grad_norm": 11.393648147583008,
      "learning_rate": 3.0420576472965207e-06,
      "loss": 3.3822,
      "step": 42370
    },
    {
      "epoch": 0.431060791015625,
      "grad_norm": 9.865949630737305,
      "learning_rate": 3.0416672722201255e-06,
      "loss": 3.3485,
      "step": 42375
    },
    {
      "epoch": 0.4311116536458333,
      "grad_norm": 13.398794174194336,
      "learning_rate": 3.041276883285302e-06,
      "loss": 3.4622,
      "step": 42380
    },
    {
      "epoch": 0.4311625162760417,
      "grad_norm": 12.496500015258789,
      "learning_rate": 3.0408864805020365e-06,
      "loss": 3.3111,
      "step": 42385
    },
    {
      "epoch": 0.43121337890625,
      "grad_norm": 8.879119873046875,
      "learning_rate": 3.04049606388032e-06,
      "loss": 3.2794,
      "step": 42390
    },
    {
      "epoch": 0.4312642415364583,
      "grad_norm": 11.061419486999512,
      "learning_rate": 3.040105633430139e-06,
      "loss": 3.489,
      "step": 42395
    },
    {
      "epoch": 0.4313151041666667,
      "grad_norm": 12.778908729553223,
      "learning_rate": 3.0397151891614833e-06,
      "loss": 3.0799,
      "step": 42400
    },
    {
      "epoch": 0.431365966796875,
      "grad_norm": 7.622555732727051,
      "learning_rate": 3.0393247310843428e-06,
      "loss": 3.2384,
      "step": 42405
    },
    {
      "epoch": 0.4314168294270833,
      "grad_norm": 13.910135269165039,
      "learning_rate": 3.038934259208707e-06,
      "loss": 3.5974,
      "step": 42410
    },
    {
      "epoch": 0.4314676920572917,
      "grad_norm": 9.1984224319458,
      "learning_rate": 3.0385437735445656e-06,
      "loss": 3.8423,
      "step": 42415
    },
    {
      "epoch": 0.4315185546875,
      "grad_norm": 11.517208099365234,
      "learning_rate": 3.038153274101909e-06,
      "loss": 3.668,
      "step": 42420
    },
    {
      "epoch": 0.4315694173177083,
      "grad_norm": 11.862286567687988,
      "learning_rate": 3.03776276089073e-06,
      "loss": 3.171,
      "step": 42425
    },
    {
      "epoch": 0.4316202799479167,
      "grad_norm": 13.053690910339355,
      "learning_rate": 3.0373722339210165e-06,
      "loss": 3.58,
      "step": 42430
    },
    {
      "epoch": 0.431671142578125,
      "grad_norm": 16.717992782592773,
      "learning_rate": 3.036981693202763e-06,
      "loss": 3.3191,
      "step": 42435
    },
    {
      "epoch": 0.4317220052083333,
      "grad_norm": 15.545528411865234,
      "learning_rate": 3.0365911387459596e-06,
      "loss": 3.0347,
      "step": 42440
    },
    {
      "epoch": 0.4317728678385417,
      "grad_norm": 8.379535675048828,
      "learning_rate": 3.036200570560599e-06,
      "loss": 3.1425,
      "step": 42445
    },
    {
      "epoch": 0.43182373046875,
      "grad_norm": 10.65440845489502,
      "learning_rate": 3.0358099886566738e-06,
      "loss": 3.2935,
      "step": 42450
    },
    {
      "epoch": 0.4318745930989583,
      "grad_norm": 16.401159286499023,
      "learning_rate": 3.035419393044177e-06,
      "loss": 3.4272,
      "step": 42455
    },
    {
      "epoch": 0.4319254557291667,
      "grad_norm": 11.619854927062988,
      "learning_rate": 3.035028783733103e-06,
      "loss": 4.0466,
      "step": 42460
    },
    {
      "epoch": 0.431976318359375,
      "grad_norm": 15.325679779052734,
      "learning_rate": 3.0346381607334434e-06,
      "loss": 3.5697,
      "step": 42465
    },
    {
      "epoch": 0.4320271809895833,
      "grad_norm": 13.767721176147461,
      "learning_rate": 3.034247524055193e-06,
      "loss": 2.9997,
      "step": 42470
    },
    {
      "epoch": 0.4320780436197917,
      "grad_norm": 8.000898361206055,
      "learning_rate": 3.0338568737083474e-06,
      "loss": 3.1968,
      "step": 42475
    },
    {
      "epoch": 0.43212890625,
      "grad_norm": 13.607340812683105,
      "learning_rate": 3.033466209702899e-06,
      "loss": 3.4118,
      "step": 42480
    },
    {
      "epoch": 0.4321797688802083,
      "grad_norm": 14.804567337036133,
      "learning_rate": 3.0330755320488446e-06,
      "loss": 3.3317,
      "step": 42485
    },
    {
      "epoch": 0.4322306315104167,
      "grad_norm": 10.511212348937988,
      "learning_rate": 3.032684840756179e-06,
      "loss": 3.5529,
      "step": 42490
    },
    {
      "epoch": 0.432281494140625,
      "grad_norm": 9.441598892211914,
      "learning_rate": 3.0322941358348985e-06,
      "loss": 3.2548,
      "step": 42495
    },
    {
      "epoch": 0.4323323567708333,
      "grad_norm": 8.868534088134766,
      "learning_rate": 3.031903417294998e-06,
      "loss": 3.397,
      "step": 42500
    },
    {
      "epoch": 0.4323832194010417,
      "grad_norm": 8.431057929992676,
      "learning_rate": 3.0315126851464747e-06,
      "loss": 3.4448,
      "step": 42505
    },
    {
      "epoch": 0.43243408203125,
      "grad_norm": 9.394766807556152,
      "learning_rate": 3.0311219393993258e-06,
      "loss": 3.4651,
      "step": 42510
    },
    {
      "epoch": 0.4324849446614583,
      "grad_norm": 8.319314956665039,
      "learning_rate": 3.0307311800635475e-06,
      "loss": 2.8608,
      "step": 42515
    },
    {
      "epoch": 0.4325358072916667,
      "grad_norm": 14.183341026306152,
      "learning_rate": 3.0303404071491377e-06,
      "loss": 3.1964,
      "step": 42520
    },
    {
      "epoch": 0.432586669921875,
      "grad_norm": 8.481855392456055,
      "learning_rate": 3.0299496206660944e-06,
      "loss": 3.454,
      "step": 42525
    },
    {
      "epoch": 0.4326375325520833,
      "grad_norm": 15.066344261169434,
      "learning_rate": 3.0295588206244165e-06,
      "loss": 3.036,
      "step": 42530
    },
    {
      "epoch": 0.4326883951822917,
      "grad_norm": 8.034235000610352,
      "learning_rate": 3.0291680070341005e-06,
      "loss": 2.9867,
      "step": 42535
    },
    {
      "epoch": 0.4327392578125,
      "grad_norm": 15.006841659545898,
      "learning_rate": 3.0287771799051475e-06,
      "loss": 3.4276,
      "step": 42540
    },
    {
      "epoch": 0.4327901204427083,
      "grad_norm": 7.801065921783447,
      "learning_rate": 3.028386339247555e-06,
      "loss": 3.1465,
      "step": 42545
    },
    {
      "epoch": 0.4328409830729167,
      "grad_norm": 14.672794342041016,
      "learning_rate": 3.0279954850713232e-06,
      "loss": 3.2184,
      "step": 42550
    },
    {
      "epoch": 0.432891845703125,
      "grad_norm": 10.196673393249512,
      "learning_rate": 3.0276046173864526e-06,
      "loss": 3.3737,
      "step": 42555
    },
    {
      "epoch": 0.4329427083333333,
      "grad_norm": 7.468010425567627,
      "learning_rate": 3.0272137362029426e-06,
      "loss": 3.1381,
      "step": 42560
    },
    {
      "epoch": 0.4329935709635417,
      "grad_norm": 14.29434871673584,
      "learning_rate": 3.0268228415307938e-06,
      "loss": 3.0089,
      "step": 42565
    },
    {
      "epoch": 0.43304443359375,
      "grad_norm": 8.408454895019531,
      "learning_rate": 3.026431933380008e-06,
      "loss": 3.0385,
      "step": 42570
    },
    {
      "epoch": 0.4330952962239583,
      "grad_norm": 8.719644546508789,
      "learning_rate": 3.0260410117605853e-06,
      "loss": 3.4599,
      "step": 42575
    },
    {
      "epoch": 0.4331461588541667,
      "grad_norm": 13.945448875427246,
      "learning_rate": 3.0256500766825293e-06,
      "loss": 3.1144,
      "step": 42580
    },
    {
      "epoch": 0.433197021484375,
      "grad_norm": 17.763442993164062,
      "learning_rate": 3.0252591281558393e-06,
      "loss": 3.2801,
      "step": 42585
    },
    {
      "epoch": 0.4332478841145833,
      "grad_norm": 9.635198593139648,
      "learning_rate": 3.02486816619052e-06,
      "loss": 3.1231,
      "step": 42590
    },
    {
      "epoch": 0.4332987467447917,
      "grad_norm": 11.811942100524902,
      "learning_rate": 3.0244771907965727e-06,
      "loss": 3.1192,
      "step": 42595
    },
    {
      "epoch": 0.433349609375,
      "grad_norm": 14.521533966064453,
      "learning_rate": 3.024086201984e-06,
      "loss": 3.1643,
      "step": 42600
    },
    {
      "epoch": 0.4334004720052083,
      "grad_norm": 8.677797317504883,
      "learning_rate": 3.0236951997628074e-06,
      "loss": 3.4529,
      "step": 42605
    },
    {
      "epoch": 0.4334513346354167,
      "grad_norm": 12.783221244812012,
      "learning_rate": 3.023304184142996e-06,
      "loss": 3.2235,
      "step": 42610
    },
    {
      "epoch": 0.433502197265625,
      "grad_norm": 8.39991283416748,
      "learning_rate": 3.022913155134572e-06,
      "loss": 3.0379,
      "step": 42615
    },
    {
      "epoch": 0.4335530598958333,
      "grad_norm": 7.332389831542969,
      "learning_rate": 3.0225221127475384e-06,
      "loss": 3.3022,
      "step": 42620
    },
    {
      "epoch": 0.4336039225260417,
      "grad_norm": 7.816256999969482,
      "learning_rate": 3.0221310569919004e-06,
      "loss": 3.1951,
      "step": 42625
    },
    {
      "epoch": 0.43365478515625,
      "grad_norm": 8.470682144165039,
      "learning_rate": 3.021739987877663e-06,
      "loss": 3.258,
      "step": 42630
    },
    {
      "epoch": 0.4337056477864583,
      "grad_norm": 7.367519855499268,
      "learning_rate": 3.0213489054148327e-06,
      "loss": 3.6178,
      "step": 42635
    },
    {
      "epoch": 0.4337565104166667,
      "grad_norm": 11.68221378326416,
      "learning_rate": 3.020957809613413e-06,
      "loss": 3.4607,
      "step": 42640
    },
    {
      "epoch": 0.433807373046875,
      "grad_norm": 10.065248489379883,
      "learning_rate": 3.0205667004834117e-06,
      "loss": 3.4678,
      "step": 42645
    },
    {
      "epoch": 0.4338582356770833,
      "grad_norm": 9.878422737121582,
      "learning_rate": 3.0201755780348343e-06,
      "loss": 3.3738,
      "step": 42650
    },
    {
      "epoch": 0.4339090983072917,
      "grad_norm": 9.238396644592285,
      "learning_rate": 3.019784442277689e-06,
      "loss": 3.4269,
      "step": 42655
    },
    {
      "epoch": 0.4339599609375,
      "grad_norm": 13.242209434509277,
      "learning_rate": 3.019393293221981e-06,
      "loss": 4.0495,
      "step": 42660
    },
    {
      "epoch": 0.4340108235677083,
      "grad_norm": 9.85168170928955,
      "learning_rate": 3.0190021308777194e-06,
      "loss": 3.3997,
      "step": 42665
    },
    {
      "epoch": 0.4340616861979167,
      "grad_norm": 10.898030281066895,
      "learning_rate": 3.018610955254911e-06,
      "loss": 3.8838,
      "step": 42670
    },
    {
      "epoch": 0.434112548828125,
      "grad_norm": 7.273516654968262,
      "learning_rate": 3.0182197663635636e-06,
      "loss": 3.5127,
      "step": 42675
    },
    {
      "epoch": 0.4341634114583333,
      "grad_norm": 13.67503833770752,
      "learning_rate": 3.0178285642136874e-06,
      "loss": 3.2358,
      "step": 42680
    },
    {
      "epoch": 0.4342142740885417,
      "grad_norm": 11.212492942810059,
      "learning_rate": 3.0174373488152896e-06,
      "loss": 3.4223,
      "step": 42685
    },
    {
      "epoch": 0.43426513671875,
      "grad_norm": 12.348718643188477,
      "learning_rate": 3.0170461201783797e-06,
      "loss": 3.4572,
      "step": 42690
    },
    {
      "epoch": 0.4343159993489583,
      "grad_norm": 12.13748836517334,
      "learning_rate": 3.0166548783129675e-06,
      "loss": 3.4449,
      "step": 42695
    },
    {
      "epoch": 0.4343668619791667,
      "grad_norm": 15.374476432800293,
      "learning_rate": 3.0162636232290632e-06,
      "loss": 3.6061,
      "step": 42700
    },
    {
      "epoch": 0.434417724609375,
      "grad_norm": 11.082989692687988,
      "learning_rate": 3.0158723549366754e-06,
      "loss": 3.1109,
      "step": 42705
    },
    {
      "epoch": 0.4344685872395833,
      "grad_norm": 15.245978355407715,
      "learning_rate": 3.0154810734458168e-06,
      "loss": 3.7396,
      "step": 42710
    },
    {
      "epoch": 0.4345194498697917,
      "grad_norm": 12.285462379455566,
      "learning_rate": 3.0150897787664963e-06,
      "loss": 3.3428,
      "step": 42715
    },
    {
      "epoch": 0.4345703125,
      "grad_norm": 13.291040420532227,
      "learning_rate": 3.014698470908727e-06,
      "loss": 3.1849,
      "step": 42720
    },
    {
      "epoch": 0.4346211751302083,
      "grad_norm": 15.192420959472656,
      "learning_rate": 3.014307149882518e-06,
      "loss": 3.4254,
      "step": 42725
    },
    {
      "epoch": 0.4346720377604167,
      "grad_norm": 11.007890701293945,
      "learning_rate": 3.013915815697883e-06,
      "loss": 3.0592,
      "step": 42730
    },
    {
      "epoch": 0.434722900390625,
      "grad_norm": 16.328813552856445,
      "learning_rate": 3.0135244683648335e-06,
      "loss": 3.4706,
      "step": 42735
    },
    {
      "epoch": 0.4347737630208333,
      "grad_norm": 14.131364822387695,
      "learning_rate": 3.0131331078933825e-06,
      "loss": 3.5952,
      "step": 42740
    },
    {
      "epoch": 0.4348246256510417,
      "grad_norm": 8.084444999694824,
      "learning_rate": 3.0127417342935426e-06,
      "loss": 3.2432,
      "step": 42745
    },
    {
      "epoch": 0.43487548828125,
      "grad_norm": 14.293861389160156,
      "learning_rate": 3.0123503475753274e-06,
      "loss": 3.7382,
      "step": 42750
    },
    {
      "epoch": 0.4349263509114583,
      "grad_norm": 11.47707748413086,
      "learning_rate": 3.0119589477487487e-06,
      "loss": 3.5475,
      "step": 42755
    },
    {
      "epoch": 0.4349772135416667,
      "grad_norm": 8.84982681274414,
      "learning_rate": 3.011567534823823e-06,
      "loss": 3.0028,
      "step": 42760
    },
    {
      "epoch": 0.435028076171875,
      "grad_norm": 10.389779090881348,
      "learning_rate": 3.0111761088105625e-06,
      "loss": 3.1776,
      "step": 42765
    },
    {
      "epoch": 0.4350789388020833,
      "grad_norm": 10.124707221984863,
      "learning_rate": 3.0107846697189825e-06,
      "loss": 3.2998,
      "step": 42770
    },
    {
      "epoch": 0.4351298014322917,
      "grad_norm": 17.50204086303711,
      "learning_rate": 3.010393217559098e-06,
      "loss": 3.3225,
      "step": 42775
    },
    {
      "epoch": 0.4351806640625,
      "grad_norm": 9.7677640914917,
      "learning_rate": 3.010001752340923e-06,
      "loss": 3.9175,
      "step": 42780
    },
    {
      "epoch": 0.4352315266927083,
      "grad_norm": 15.209006309509277,
      "learning_rate": 3.0096102740744747e-06,
      "loss": 3.5655,
      "step": 42785
    },
    {
      "epoch": 0.4352823893229167,
      "grad_norm": 14.229117393493652,
      "learning_rate": 3.0092187827697692e-06,
      "loss": 3.6215,
      "step": 42790
    },
    {
      "epoch": 0.435333251953125,
      "grad_norm": 10.619292259216309,
      "learning_rate": 3.0088272784368203e-06,
      "loss": 3.4509,
      "step": 42795
    },
    {
      "epoch": 0.4353841145833333,
      "grad_norm": 14.222017288208008,
      "learning_rate": 3.008435761085647e-06,
      "loss": 3.321,
      "step": 42800
    },
    {
      "epoch": 0.4354349772135417,
      "grad_norm": 13.797867774963379,
      "learning_rate": 3.008044230726265e-06,
      "loss": 3.5503,
      "step": 42805
    },
    {
      "epoch": 0.43548583984375,
      "grad_norm": 14.982044219970703,
      "learning_rate": 3.0076526873686923e-06,
      "loss": 3.3824,
      "step": 42810
    },
    {
      "epoch": 0.4355367024739583,
      "grad_norm": 13.971278190612793,
      "learning_rate": 3.007261131022946e-06,
      "loss": 3.5267,
      "step": 42815
    },
    {
      "epoch": 0.4355875651041667,
      "grad_norm": 15.691052436828613,
      "learning_rate": 3.0068695616990427e-06,
      "loss": 3.2466,
      "step": 42820
    },
    {
      "epoch": 0.435638427734375,
      "grad_norm": 8.944365501403809,
      "learning_rate": 3.0064779794070033e-06,
      "loss": 3.4923,
      "step": 42825
    },
    {
      "epoch": 0.4356892903645833,
      "grad_norm": 6.83592414855957,
      "learning_rate": 3.006086384156844e-06,
      "loss": 3.2261,
      "step": 42830
    },
    {
      "epoch": 0.4357401529947917,
      "grad_norm": 9.382534980773926,
      "learning_rate": 3.0056947759585843e-06,
      "loss": 3.6441,
      "step": 42835
    },
    {
      "epoch": 0.435791015625,
      "grad_norm": 10.255620956420898,
      "learning_rate": 3.0053031548222437e-06,
      "loss": 2.9907,
      "step": 42840
    },
    {
      "epoch": 0.4358418782552083,
      "grad_norm": 12.222493171691895,
      "learning_rate": 3.004911520757842e-06,
      "loss": 3.3177,
      "step": 42845
    },
    {
      "epoch": 0.4358927408854167,
      "grad_norm": 10.061985969543457,
      "learning_rate": 3.0045198737753987e-06,
      "loss": 3.1202,
      "step": 42850
    },
    {
      "epoch": 0.435943603515625,
      "grad_norm": 15.395977973937988,
      "learning_rate": 3.004128213884934e-06,
      "loss": 3.6209,
      "step": 42855
    },
    {
      "epoch": 0.4359944661458333,
      "grad_norm": 8.896660804748535,
      "learning_rate": 3.003736541096468e-06,
      "loss": 3.1989,
      "step": 42860
    },
    {
      "epoch": 0.4360453287760417,
      "grad_norm": 12.098740577697754,
      "learning_rate": 3.0033448554200228e-06,
      "loss": 3.3545,
      "step": 42865
    },
    {
      "epoch": 0.43609619140625,
      "grad_norm": 9.284294128417969,
      "learning_rate": 3.002953156865618e-06,
      "loss": 2.994,
      "step": 42870
    },
    {
      "epoch": 0.4361470540364583,
      "grad_norm": 13.549686431884766,
      "learning_rate": 3.0025614454432765e-06,
      "loss": 3.2236,
      "step": 42875
    },
    {
      "epoch": 0.4361979166666667,
      "grad_norm": 13.168487548828125,
      "learning_rate": 3.0021697211630195e-06,
      "loss": 3.0275,
      "step": 42880
    },
    {
      "epoch": 0.436248779296875,
      "grad_norm": 9.46147346496582,
      "learning_rate": 3.001777984034869e-06,
      "loss": 3.3652,
      "step": 42885
    },
    {
      "epoch": 0.4362996419270833,
      "grad_norm": 8.875849723815918,
      "learning_rate": 3.0013862340688473e-06,
      "loss": 3.3035,
      "step": 42890
    },
    {
      "epoch": 0.4363505045572917,
      "grad_norm": 13.512296676635742,
      "learning_rate": 3.0009944712749783e-06,
      "loss": 3.5612,
      "step": 42895
    },
    {
      "epoch": 0.4364013671875,
      "grad_norm": 8.877745628356934,
      "learning_rate": 3.000602695663284e-06,
      "loss": 3.4272,
      "step": 42900
    },
    {
      "epoch": 0.4364522298177083,
      "grad_norm": 16.58599853515625,
      "learning_rate": 3.0002109072437883e-06,
      "loss": 3.4123,
      "step": 42905
    },
    {
      "epoch": 0.4365030924479167,
      "grad_norm": 12.493461608886719,
      "learning_rate": 2.9998191060265156e-06,
      "loss": 3.4447,
      "step": 42910
    },
    {
      "epoch": 0.436553955078125,
      "grad_norm": 12.325043678283691,
      "learning_rate": 2.9994272920214896e-06,
      "loss": 3.0123,
      "step": 42915
    },
    {
      "epoch": 0.4366048177083333,
      "grad_norm": 9.454997062683105,
      "learning_rate": 2.9990354652387348e-06,
      "loss": 3.2529,
      "step": 42920
    },
    {
      "epoch": 0.4366556803385417,
      "grad_norm": 14.23027229309082,
      "learning_rate": 2.998643625688275e-06,
      "loss": 3.8447,
      "step": 42925
    },
    {
      "epoch": 0.43670654296875,
      "grad_norm": 12.489675521850586,
      "learning_rate": 2.998251773380137e-06,
      "loss": 2.9214,
      "step": 42930
    },
    {
      "epoch": 0.4367574055989583,
      "grad_norm": 10.686521530151367,
      "learning_rate": 2.9978599083243454e-06,
      "loss": 3.5369,
      "step": 42935
    },
    {
      "epoch": 0.4368082682291667,
      "grad_norm": 12.095287322998047,
      "learning_rate": 2.9974680305309257e-06,
      "loss": 3.582,
      "step": 42940
    },
    {
      "epoch": 0.436859130859375,
      "grad_norm": 13.138361930847168,
      "learning_rate": 2.9970761400099047e-06,
      "loss": 3.0216,
      "step": 42945
    },
    {
      "epoch": 0.4369099934895833,
      "grad_norm": 14.297666549682617,
      "learning_rate": 2.9966842367713082e-06,
      "loss": 3.444,
      "step": 42950
    },
    {
      "epoch": 0.4369608561197917,
      "grad_norm": 15.614699363708496,
      "learning_rate": 2.996292320825164e-06,
      "loss": 3.2882,
      "step": 42955
    },
    {
      "epoch": 0.43701171875,
      "grad_norm": 10.250551223754883,
      "learning_rate": 2.9959003921814973e-06,
      "loss": 2.9559,
      "step": 42960
    },
    {
      "epoch": 0.4370625813802083,
      "grad_norm": 13.618687629699707,
      "learning_rate": 2.9955084508503367e-06,
      "loss": 3.5666,
      "step": 42965
    },
    {
      "epoch": 0.4371134440104167,
      "grad_norm": 8.38098430633545,
      "learning_rate": 2.995116496841711e-06,
      "loss": 3.1937,
      "step": 42970
    },
    {
      "epoch": 0.437164306640625,
      "grad_norm": 18.40715980529785,
      "learning_rate": 2.994724530165646e-06,
      "loss": 3.6116,
      "step": 42975
    },
    {
      "epoch": 0.4372151692708333,
      "grad_norm": 16.524473190307617,
      "learning_rate": 2.9943325508321717e-06,
      "loss": 3.3483,
      "step": 42980
    },
    {
      "epoch": 0.4372660319010417,
      "grad_norm": 13.106752395629883,
      "learning_rate": 2.9939405588513158e-06,
      "loss": 3.5727,
      "step": 42985
    },
    {
      "epoch": 0.43731689453125,
      "grad_norm": 15.565990447998047,
      "learning_rate": 2.9935485542331085e-06,
      "loss": 3.8515,
      "step": 42990
    },
    {
      "epoch": 0.4373677571614583,
      "grad_norm": 7.531777858734131,
      "learning_rate": 2.9931565369875776e-06,
      "loss": 3.044,
      "step": 42995
    },
    {
      "epoch": 0.4374186197916667,
      "grad_norm": 10.903145790100098,
      "learning_rate": 2.992764507124754e-06,
      "loss": 3.4493,
      "step": 43000
    },
    {
      "epoch": 0.437469482421875,
      "grad_norm": 11.029446601867676,
      "learning_rate": 2.992372464654667e-06,
      "loss": 3.3143,
      "step": 43005
    },
    {
      "epoch": 0.4375203450520833,
      "grad_norm": 14.928308486938477,
      "learning_rate": 2.9919804095873473e-06,
      "loss": 3.2214,
      "step": 43010
    },
    {
      "epoch": 0.4375712076822917,
      "grad_norm": 9.891948699951172,
      "learning_rate": 2.9915883419328258e-06,
      "loss": 3.4254,
      "step": 43015
    },
    {
      "epoch": 0.4376220703125,
      "grad_norm": 12.621122360229492,
      "learning_rate": 2.9911962617011325e-06,
      "loss": 3.2019,
      "step": 43020
    },
    {
      "epoch": 0.4376729329427083,
      "grad_norm": 10.894399642944336,
      "learning_rate": 2.9908041689023e-06,
      "loss": 3.2527,
      "step": 43025
    },
    {
      "epoch": 0.4377237955729167,
      "grad_norm": 10.197098731994629,
      "learning_rate": 2.9904120635463583e-06,
      "loss": 3.3771,
      "step": 43030
    },
    {
      "epoch": 0.437774658203125,
      "grad_norm": 15.673666954040527,
      "learning_rate": 2.9900199456433414e-06,
      "loss": 3.1911,
      "step": 43035
    },
    {
      "epoch": 0.4378255208333333,
      "grad_norm": 10.539162635803223,
      "learning_rate": 2.9896278152032785e-06,
      "loss": 3.1443,
      "step": 43040
    },
    {
      "epoch": 0.4378763834635417,
      "grad_norm": 12.590764999389648,
      "learning_rate": 2.989235672236206e-06,
      "loss": 3.3408,
      "step": 43045
    },
    {
      "epoch": 0.43792724609375,
      "grad_norm": 10.05128002166748,
      "learning_rate": 2.988843516752154e-06,
      "loss": 3.5955,
      "step": 43050
    },
    {
      "epoch": 0.4379781087239583,
      "grad_norm": 10.43685531616211,
      "learning_rate": 2.988451348761156e-06,
      "loss": 3.3872,
      "step": 43055
    },
    {
      "epoch": 0.4380289713541667,
      "grad_norm": 15.180144309997559,
      "learning_rate": 2.988059168273247e-06,
      "loss": 3.3636,
      "step": 43060
    },
    {
      "epoch": 0.438079833984375,
      "grad_norm": 11.525684356689453,
      "learning_rate": 2.9876669752984588e-06,
      "loss": 3.4391,
      "step": 43065
    },
    {
      "epoch": 0.4381306966145833,
      "grad_norm": 7.944387435913086,
      "learning_rate": 2.9872747698468267e-06,
      "loss": 3.6003,
      "step": 43070
    },
    {
      "epoch": 0.4381815592447917,
      "grad_norm": 8.787629127502441,
      "learning_rate": 2.9868825519283863e-06,
      "loss": 2.9129,
      "step": 43075
    },
    {
      "epoch": 0.438232421875,
      "grad_norm": 14.393413543701172,
      "learning_rate": 2.9864903215531697e-06,
      "loss": 3.3476,
      "step": 43080
    },
    {
      "epoch": 0.4382832845052083,
      "grad_norm": 15.774198532104492,
      "learning_rate": 2.986098078731215e-06,
      "loss": 3.4395,
      "step": 43085
    },
    {
      "epoch": 0.4383341471354167,
      "grad_norm": 8.885433197021484,
      "learning_rate": 2.9857058234725556e-06,
      "loss": 3.651,
      "step": 43090
    },
    {
      "epoch": 0.438385009765625,
      "grad_norm": 10.791901588439941,
      "learning_rate": 2.985313555787227e-06,
      "loss": 3.1684,
      "step": 43095
    },
    {
      "epoch": 0.4384358723958333,
      "grad_norm": 6.774324893951416,
      "learning_rate": 2.984921275685267e-06,
      "loss": 3.4454,
      "step": 43100
    },
    {
      "epoch": 0.4384867350260417,
      "grad_norm": 9.706006050109863,
      "learning_rate": 2.9845289831767103e-06,
      "loss": 3.289,
      "step": 43105
    },
    {
      "epoch": 0.43853759765625,
      "grad_norm": 10.4579496383667,
      "learning_rate": 2.984136678271596e-06,
      "loss": 3.3626,
      "step": 43110
    },
    {
      "epoch": 0.4385884602864583,
      "grad_norm": 14.540216445922852,
      "learning_rate": 2.983744360979958e-06,
      "loss": 3.5328,
      "step": 43115
    },
    {
      "epoch": 0.4386393229166667,
      "grad_norm": 14.207021713256836,
      "learning_rate": 2.983352031311836e-06,
      "loss": 3.4168,
      "step": 43120
    },
    {
      "epoch": 0.438690185546875,
      "grad_norm": 13.922112464904785,
      "learning_rate": 2.982959689277266e-06,
      "loss": 3.4668,
      "step": 43125
    },
    {
      "epoch": 0.4387410481770833,
      "grad_norm": 7.798769950866699,
      "learning_rate": 2.9825673348862878e-06,
      "loss": 3.6244,
      "step": 43130
    },
    {
      "epoch": 0.4387919108072917,
      "grad_norm": 13.192652702331543,
      "learning_rate": 2.9821749681489383e-06,
      "loss": 3.0737,
      "step": 43135
    },
    {
      "epoch": 0.4388427734375,
      "grad_norm": 13.703279495239258,
      "learning_rate": 2.9817825890752566e-06,
      "loss": 3.1973,
      "step": 43140
    },
    {
      "epoch": 0.4388936360677083,
      "grad_norm": 9.974124908447266,
      "learning_rate": 2.981390197675281e-06,
      "loss": 3.1817,
      "step": 43145
    },
    {
      "epoch": 0.4389444986979167,
      "grad_norm": 9.606637954711914,
      "learning_rate": 2.9809977939590525e-06,
      "loss": 3.3443,
      "step": 43150
    },
    {
      "epoch": 0.438995361328125,
      "grad_norm": 10.829172134399414,
      "learning_rate": 2.9806053779366086e-06,
      "loss": 3.2096,
      "step": 43155
    },
    {
      "epoch": 0.4390462239583333,
      "grad_norm": 8.69129753112793,
      "learning_rate": 2.98021294961799e-06,
      "loss": 3.435,
      "step": 43160
    },
    {
      "epoch": 0.4390970865885417,
      "grad_norm": 10.557202339172363,
      "learning_rate": 2.979820509013237e-06,
      "loss": 3.2752,
      "step": 43165
    },
    {
      "epoch": 0.43914794921875,
      "grad_norm": 16.040302276611328,
      "learning_rate": 2.9794280561323895e-06,
      "loss": 3.3947,
      "step": 43170
    },
    {
      "epoch": 0.4391988118489583,
      "grad_norm": 14.875178337097168,
      "learning_rate": 2.9790355909854897e-06,
      "loss": 3.3824,
      "step": 43175
    },
    {
      "epoch": 0.4392496744791667,
      "grad_norm": 12.481240272521973,
      "learning_rate": 2.9786431135825776e-06,
      "loss": 3.5868,
      "step": 43180
    },
    {
      "epoch": 0.439300537109375,
      "grad_norm": 10.882767677307129,
      "learning_rate": 2.9782506239336946e-06,
      "loss": 3.3987,
      "step": 43185
    },
    {
      "epoch": 0.4393513997395833,
      "grad_norm": 10.94859790802002,
      "learning_rate": 2.977858122048883e-06,
      "loss": 3.5864,
      "step": 43190
    },
    {
      "epoch": 0.4394022623697917,
      "grad_norm": 15.125771522521973,
      "learning_rate": 2.977465607938185e-06,
      "loss": 3.6607,
      "step": 43195
    },
    {
      "epoch": 0.439453125,
      "grad_norm": 10.725627899169922,
      "learning_rate": 2.9770730816116416e-06,
      "loss": 3.6029,
      "step": 43200
    },
    {
      "epoch": 0.4395039876302083,
      "grad_norm": 12.071203231811523,
      "learning_rate": 2.9766805430792976e-06,
      "loss": 3.2604,
      "step": 43205
    },
    {
      "epoch": 0.4395548502604167,
      "grad_norm": 13.576530456542969,
      "learning_rate": 2.976287992351194e-06,
      "loss": 3.9804,
      "step": 43210
    },
    {
      "epoch": 0.439605712890625,
      "grad_norm": 14.199870109558105,
      "learning_rate": 2.975895429437375e-06,
      "loss": 3.7286,
      "step": 43215
    },
    {
      "epoch": 0.4396565755208333,
      "grad_norm": 12.568558692932129,
      "learning_rate": 2.9755028543478854e-06,
      "loss": 3.2988,
      "step": 43220
    },
    {
      "epoch": 0.4397074381510417,
      "grad_norm": 13.958497047424316,
      "learning_rate": 2.975110267092767e-06,
      "loss": 3.5233,
      "step": 43225
    },
    {
      "epoch": 0.43975830078125,
      "grad_norm": 13.36304759979248,
      "learning_rate": 2.974717667682065e-06,
      "loss": 3.6748,
      "step": 43230
    },
    {
      "epoch": 0.4398091634114583,
      "grad_norm": 11.973743438720703,
      "learning_rate": 2.974325056125824e-06,
      "loss": 3.2443,
      "step": 43235
    },
    {
      "epoch": 0.4398600260416667,
      "grad_norm": 10.666085243225098,
      "learning_rate": 2.973932432434089e-06,
      "loss": 3.4837,
      "step": 43240
    },
    {
      "epoch": 0.439910888671875,
      "grad_norm": 14.517359733581543,
      "learning_rate": 2.9735397966169055e-06,
      "loss": 3.4415,
      "step": 43245
    },
    {
      "epoch": 0.4399617513020833,
      "grad_norm": 10.464876174926758,
      "learning_rate": 2.973147148684318e-06,
      "loss": 3.5669,
      "step": 43250
    },
    {
      "epoch": 0.4400126139322917,
      "grad_norm": 11.049302101135254,
      "learning_rate": 2.972754488646373e-06,
      "loss": 3.245,
      "step": 43255
    },
    {
      "epoch": 0.4400634765625,
      "grad_norm": 9.860051155090332,
      "learning_rate": 2.972361816513116e-06,
      "loss": 3.3552,
      "step": 43260
    },
    {
      "epoch": 0.4401143391927083,
      "grad_norm": 10.621298789978027,
      "learning_rate": 2.971969132294595e-06,
      "loss": 3.2837,
      "step": 43265
    },
    {
      "epoch": 0.4401652018229167,
      "grad_norm": 14.433876991271973,
      "learning_rate": 2.9715764360008548e-06,
      "loss": 3.3214,
      "step": 43270
    },
    {
      "epoch": 0.440216064453125,
      "grad_norm": 10.880840301513672,
      "learning_rate": 2.971183727641944e-06,
      "loss": 3.3316,
      "step": 43275
    },
    {
      "epoch": 0.4402669270833333,
      "grad_norm": 13.201940536499023,
      "learning_rate": 2.970791007227908e-06,
      "loss": 3.2394,
      "step": 43280
    },
    {
      "epoch": 0.4403177897135417,
      "grad_norm": 11.534923553466797,
      "learning_rate": 2.9703982747687966e-06,
      "loss": 3.3618,
      "step": 43285
    },
    {
      "epoch": 0.44036865234375,
      "grad_norm": 13.809548377990723,
      "learning_rate": 2.9700055302746562e-06,
      "loss": 3.3902,
      "step": 43290
    },
    {
      "epoch": 0.4404195149739583,
      "grad_norm": 8.454058647155762,
      "learning_rate": 2.969612773755536e-06,
      "loss": 3.1477,
      "step": 43295
    },
    {
      "epoch": 0.4404703776041667,
      "grad_norm": 9.063993453979492,
      "learning_rate": 2.9692200052214842e-06,
      "loss": 3.4497,
      "step": 43300
    },
    {
      "epoch": 0.440521240234375,
      "grad_norm": 11.819409370422363,
      "learning_rate": 2.96882722468255e-06,
      "loss": 2.8236,
      "step": 43305
    },
    {
      "epoch": 0.4405721028645833,
      "grad_norm": 11.365645408630371,
      "learning_rate": 2.968434432148783e-06,
      "loss": 3.6733,
      "step": 43310
    },
    {
      "epoch": 0.4406229654947917,
      "grad_norm": 15.08333969116211,
      "learning_rate": 2.9680416276302303e-06,
      "loss": 3.3216,
      "step": 43315
    },
    {
      "epoch": 0.440673828125,
      "grad_norm": 13.111920356750488,
      "learning_rate": 2.9676488111369456e-06,
      "loss": 3.5055,
      "step": 43320
    },
    {
      "epoch": 0.4407246907552083,
      "grad_norm": 11.382067680358887,
      "learning_rate": 2.967255982678975e-06,
      "loss": 3.5201,
      "step": 43325
    },
    {
      "epoch": 0.4407755533854167,
      "grad_norm": 13.540173530578613,
      "learning_rate": 2.9668631422663722e-06,
      "loss": 3.6338,
      "step": 43330
    },
    {
      "epoch": 0.440826416015625,
      "grad_norm": 8.033857345581055,
      "learning_rate": 2.966470289909186e-06,
      "loss": 3.2214,
      "step": 43335
    },
    {
      "epoch": 0.4408772786458333,
      "grad_norm": 8.666276931762695,
      "learning_rate": 2.9660774256174684e-06,
      "loss": 3.4007,
      "step": 43340
    },
    {
      "epoch": 0.4409281412760417,
      "grad_norm": 8.476204872131348,
      "learning_rate": 2.96568454940127e-06,
      "loss": 3.4108,
      "step": 43345
    },
    {
      "epoch": 0.44097900390625,
      "grad_norm": 15.811210632324219,
      "learning_rate": 2.9652916612706426e-06,
      "loss": 3.3204,
      "step": 43350
    },
    {
      "epoch": 0.4410298665364583,
      "grad_norm": 14.496684074401855,
      "learning_rate": 2.9648987612356385e-06,
      "loss": 3.3657,
      "step": 43355
    },
    {
      "epoch": 0.4410807291666667,
      "grad_norm": 9.801753044128418,
      "learning_rate": 2.9645058493063105e-06,
      "loss": 3.8043,
      "step": 43360
    },
    {
      "epoch": 0.441131591796875,
      "grad_norm": 8.788553237915039,
      "learning_rate": 2.9641129254927092e-06,
      "loss": 3.6548,
      "step": 43365
    },
    {
      "epoch": 0.4411824544270833,
      "grad_norm": 7.200490951538086,
      "learning_rate": 2.9637199898048898e-06,
      "loss": 3.1184,
      "step": 43370
    },
    {
      "epoch": 0.4412333170572917,
      "grad_norm": 10.760024070739746,
      "learning_rate": 2.963327042252904e-06,
      "loss": 3.1018,
      "step": 43375
    },
    {
      "epoch": 0.4412841796875,
      "grad_norm": 16.04799461364746,
      "learning_rate": 2.962934082846805e-06,
      "loss": 3.262,
      "step": 43380
    },
    {
      "epoch": 0.4413350423177083,
      "grad_norm": 11.933404922485352,
      "learning_rate": 2.962541111596648e-06,
      "loss": 3.0244,
      "step": 43385
    },
    {
      "epoch": 0.4413859049479167,
      "grad_norm": 11.039929389953613,
      "learning_rate": 2.962148128512486e-06,
      "loss": 3.1555,
      "step": 43390
    },
    {
      "epoch": 0.441436767578125,
      "grad_norm": 13.250151634216309,
      "learning_rate": 2.9617551336043736e-06,
      "loss": 2.922,
      "step": 43395
    },
    {
      "epoch": 0.4414876302083333,
      "grad_norm": 8.479625701904297,
      "learning_rate": 2.9613621268823654e-06,
      "loss": 3.2628,
      "step": 43400
    },
    {
      "epoch": 0.4415384928385417,
      "grad_norm": 7.887979984283447,
      "learning_rate": 2.960969108356517e-06,
      "loss": 3.4097,
      "step": 43405
    },
    {
      "epoch": 0.44158935546875,
      "grad_norm": 14.438159942626953,
      "learning_rate": 2.960576078036883e-06,
      "loss": 2.8774,
      "step": 43410
    },
    {
      "epoch": 0.4416402180989583,
      "grad_norm": 10.12606430053711,
      "learning_rate": 2.9601830359335193e-06,
      "loss": 3.2402,
      "step": 43415
    },
    {
      "epoch": 0.4416910807291667,
      "grad_norm": 18.690998077392578,
      "learning_rate": 2.9597899820564813e-06,
      "loss": 3.9748,
      "step": 43420
    },
    {
      "epoch": 0.441741943359375,
      "grad_norm": 11.186441421508789,
      "learning_rate": 2.9593969164158265e-06,
      "loss": 3.7492,
      "step": 43425
    },
    {
      "epoch": 0.4417928059895833,
      "grad_norm": 13.399593353271484,
      "learning_rate": 2.959003839021609e-06,
      "loss": 3.3452,
      "step": 43430
    },
    {
      "epoch": 0.4418436686197917,
      "grad_norm": 10.064865112304688,
      "learning_rate": 2.958610749883889e-06,
      "loss": 2.8705,
      "step": 43435
    },
    {
      "epoch": 0.44189453125,
      "grad_norm": 11.90968132019043,
      "learning_rate": 2.95821764901272e-06,
      "loss": 3.2716,
      "step": 43440
    },
    {
      "epoch": 0.4419453938802083,
      "grad_norm": 12.014603614807129,
      "learning_rate": 2.957824536418162e-06,
      "loss": 3.3077,
      "step": 43445
    },
    {
      "epoch": 0.4419962565104167,
      "grad_norm": 7.141475200653076,
      "learning_rate": 2.9574314121102717e-06,
      "loss": 3.0187,
      "step": 43450
    },
    {
      "epoch": 0.442047119140625,
      "grad_norm": 11.256348609924316,
      "learning_rate": 2.957038276099107e-06,
      "loss": 3.0342,
      "step": 43455
    },
    {
      "epoch": 0.4420979817708333,
      "grad_norm": 13.187239646911621,
      "learning_rate": 2.956645128394726e-06,
      "loss": 3.8987,
      "step": 43460
    },
    {
      "epoch": 0.4421488444010417,
      "grad_norm": 11.204666137695312,
      "learning_rate": 2.9562519690071876e-06,
      "loss": 3.4287,
      "step": 43465
    },
    {
      "epoch": 0.44219970703125,
      "grad_norm": 8.557195663452148,
      "learning_rate": 2.9558587979465507e-06,
      "loss": 3.4322,
      "step": 43470
    },
    {
      "epoch": 0.4422505696614583,
      "grad_norm": 13.133142471313477,
      "learning_rate": 2.9554656152228758e-06,
      "loss": 3.3743,
      "step": 43475
    },
    {
      "epoch": 0.4423014322916667,
      "grad_norm": 14.486852645874023,
      "learning_rate": 2.9550724208462187e-06,
      "loss": 3.3238,
      "step": 43480
    },
    {
      "epoch": 0.442352294921875,
      "grad_norm": 14.670029640197754,
      "learning_rate": 2.9546792148266436e-06,
      "loss": 3.0607,
      "step": 43485
    },
    {
      "epoch": 0.4424031575520833,
      "grad_norm": 15.148486137390137,
      "learning_rate": 2.954285997174208e-06,
      "loss": 3.3221,
      "step": 43490
    },
    {
      "epoch": 0.4424540201822917,
      "grad_norm": 9.987661361694336,
      "learning_rate": 2.9538927678989725e-06,
      "loss": 3.327,
      "step": 43495
    },
    {
      "epoch": 0.4425048828125,
      "grad_norm": 14.121315002441406,
      "learning_rate": 2.953499527010999e-06,
      "loss": 3.6905,
      "step": 43500
    },
    {
      "epoch": 0.4425557454427083,
      "grad_norm": 10.06773853302002,
      "learning_rate": 2.9531062745203466e-06,
      "loss": 3.3083,
      "step": 43505
    },
    {
      "epoch": 0.4426066080729167,
      "grad_norm": 15.57811450958252,
      "learning_rate": 2.9527130104370776e-06,
      "loss": 4.0533,
      "step": 43510
    },
    {
      "epoch": 0.442657470703125,
      "grad_norm": 12.95801830291748,
      "learning_rate": 2.952319734771254e-06,
      "loss": 3.5316,
      "step": 43515
    },
    {
      "epoch": 0.4427083333333333,
      "grad_norm": 7.576696395874023,
      "learning_rate": 2.9519264475329374e-06,
      "loss": 3.393,
      "step": 43520
    },
    {
      "epoch": 0.4427591959635417,
      "grad_norm": 12.71930980682373,
      "learning_rate": 2.9515331487321893e-06,
      "loss": 3.3971,
      "step": 43525
    },
    {
      "epoch": 0.44281005859375,
      "grad_norm": 10.263254165649414,
      "learning_rate": 2.951139838379073e-06,
      "loss": 3.5974,
      "step": 43530
    },
    {
      "epoch": 0.4428609212239583,
      "grad_norm": 15.159732818603516,
      "learning_rate": 2.95074651648365e-06,
      "loss": 3.3649,
      "step": 43535
    },
    {
      "epoch": 0.4429117838541667,
      "grad_norm": 15.83638858795166,
      "learning_rate": 2.950353183055985e-06,
      "loss": 3.3533,
      "step": 43540
    },
    {
      "epoch": 0.442962646484375,
      "grad_norm": 13.222535133361816,
      "learning_rate": 2.9499598381061408e-06,
      "loss": 3.4962,
      "step": 43545
    },
    {
      "epoch": 0.4430135091145833,
      "grad_norm": 9.307706832885742,
      "learning_rate": 2.94956648164418e-06,
      "loss": 3.1602,
      "step": 43550
    },
    {
      "epoch": 0.4430643717447917,
      "grad_norm": 15.66658878326416,
      "learning_rate": 2.9491731136801677e-06,
      "loss": 3.4844,
      "step": 43555
    },
    {
      "epoch": 0.443115234375,
      "grad_norm": 10.596369743347168,
      "learning_rate": 2.9487797342241677e-06,
      "loss": 2.9099,
      "step": 43560
    },
    {
      "epoch": 0.4431660970052083,
      "grad_norm": 10.095293998718262,
      "learning_rate": 2.9483863432862447e-06,
      "loss": 3.0744,
      "step": 43565
    },
    {
      "epoch": 0.4432169596354167,
      "grad_norm": 14.364975929260254,
      "learning_rate": 2.947992940876463e-06,
      "loss": 3.2898,
      "step": 43570
    },
    {
      "epoch": 0.443267822265625,
      "grad_norm": 8.403112411499023,
      "learning_rate": 2.9475995270048887e-06,
      "loss": 3.217,
      "step": 43575
    },
    {
      "epoch": 0.4433186848958333,
      "grad_norm": 13.855816841125488,
      "learning_rate": 2.9472061016815866e-06,
      "loss": 3.3371,
      "step": 43580
    },
    {
      "epoch": 0.4433695475260417,
      "grad_norm": 15.479269981384277,
      "learning_rate": 2.946812664916622e-06,
      "loss": 3.5384,
      "step": 43585
    },
    {
      "epoch": 0.44342041015625,
      "grad_norm": 9.135083198547363,
      "learning_rate": 2.9464192167200607e-06,
      "loss": 3.6018,
      "step": 43590
    },
    {
      "epoch": 0.4434712727864583,
      "grad_norm": 13.801974296569824,
      "learning_rate": 2.946025757101971e-06,
      "loss": 3.2447,
      "step": 43595
    },
    {
      "epoch": 0.4435221354166667,
      "grad_norm": 13.949684143066406,
      "learning_rate": 2.945632286072416e-06,
      "loss": 3.3698,
      "step": 43600
    },
    {
      "epoch": 0.443572998046875,
      "grad_norm": 16.448776245117188,
      "learning_rate": 2.945238803641467e-06,
      "loss": 3.3822,
      "step": 43605
    },
    {
      "epoch": 0.4436238606770833,
      "grad_norm": 10.418899536132812,
      "learning_rate": 2.944845309819187e-06,
      "loss": 3.3236,
      "step": 43610
    },
    {
      "epoch": 0.4436747233072917,
      "grad_norm": 8.672096252441406,
      "learning_rate": 2.944451804615646e-06,
      "loss": 3.4957,
      "step": 43615
    },
    {
      "epoch": 0.4437255859375,
      "grad_norm": 11.790623664855957,
      "learning_rate": 2.9440582880409106e-06,
      "loss": 3.5351,
      "step": 43620
    },
    {
      "epoch": 0.4437764485677083,
      "grad_norm": 17.244104385375977,
      "learning_rate": 2.9436647601050494e-06,
      "loss": 3.5512,
      "step": 43625
    },
    {
      "epoch": 0.4438273111979167,
      "grad_norm": 12.104947090148926,
      "learning_rate": 2.94327122081813e-06,
      "loss": 2.9641,
      "step": 43630
    },
    {
      "epoch": 0.443878173828125,
      "grad_norm": 7.7502336502075195,
      "learning_rate": 2.9428776701902223e-06,
      "loss": 3.3353,
      "step": 43635
    },
    {
      "epoch": 0.4439290364583333,
      "grad_norm": 11.081683158874512,
      "learning_rate": 2.942484108231393e-06,
      "loss": 3.1159,
      "step": 43640
    },
    {
      "epoch": 0.4439798990885417,
      "grad_norm": 8.795500755310059,
      "learning_rate": 2.9420905349517143e-06,
      "loss": 3.3727,
      "step": 43645
    },
    {
      "epoch": 0.44403076171875,
      "grad_norm": 9.463774681091309,
      "learning_rate": 2.9416969503612527e-06,
      "loss": 3.4369,
      "step": 43650
    },
    {
      "epoch": 0.4440816243489583,
      "grad_norm": 15.080618858337402,
      "learning_rate": 2.94130335447008e-06,
      "loss": 3.4072,
      "step": 43655
    },
    {
      "epoch": 0.4441324869791667,
      "grad_norm": 15.829692840576172,
      "learning_rate": 2.940909747288266e-06,
      "loss": 3.1416,
      "step": 43660
    },
    {
      "epoch": 0.444183349609375,
      "grad_norm": 14.829861640930176,
      "learning_rate": 2.9405161288258797e-06,
      "loss": 3.2442,
      "step": 43665
    },
    {
      "epoch": 0.4442342122395833,
      "grad_norm": 12.270400047302246,
      "learning_rate": 2.940122499092993e-06,
      "loss": 3.2945,
      "step": 43670
    },
    {
      "epoch": 0.4442850748697917,
      "grad_norm": 9.87391185760498,
      "learning_rate": 2.939728858099676e-06,
      "loss": 3.652,
      "step": 43675
    },
    {
      "epoch": 0.4443359375,
      "grad_norm": 8.332228660583496,
      "learning_rate": 2.9393352058560014e-06,
      "loss": 3.4948,
      "step": 43680
    },
    {
      "epoch": 0.4443868001302083,
      "grad_norm": 11.852291107177734,
      "learning_rate": 2.938941542372039e-06,
      "loss": 3.4039,
      "step": 43685
    },
    {
      "epoch": 0.4444376627604167,
      "grad_norm": 9.319053649902344,
      "learning_rate": 2.9385478676578617e-06,
      "loss": 3.3997,
      "step": 43690
    },
    {
      "epoch": 0.444488525390625,
      "grad_norm": 15.921639442443848,
      "learning_rate": 2.938154181723541e-06,
      "loss": 3.6342,
      "step": 43695
    },
    {
      "epoch": 0.4445393880208333,
      "grad_norm": 14.104935646057129,
      "learning_rate": 2.9377604845791498e-06,
      "loss": 3.7262,
      "step": 43700
    },
    {
      "epoch": 0.4445902506510417,
      "grad_norm": 11.885713577270508,
      "learning_rate": 2.937366776234759e-06,
      "loss": 3.49,
      "step": 43705
    },
    {
      "epoch": 0.44464111328125,
      "grad_norm": 13.739585876464844,
      "learning_rate": 2.9369730567004447e-06,
      "loss": 3.5327,
      "step": 43710
    },
    {
      "epoch": 0.4446919759114583,
      "grad_norm": 14.605637550354004,
      "learning_rate": 2.936579325986277e-06,
      "loss": 3.1572,
      "step": 43715
    },
    {
      "epoch": 0.4447428385416667,
      "grad_norm": 14.697052001953125,
      "learning_rate": 2.936185584102331e-06,
      "loss": 3.511,
      "step": 43720
    },
    {
      "epoch": 0.444793701171875,
      "grad_norm": 10.544771194458008,
      "learning_rate": 2.9357918310586808e-06,
      "loss": 3.4963,
      "step": 43725
    },
    {
      "epoch": 0.4448445638020833,
      "grad_norm": 8.620457649230957,
      "learning_rate": 2.9353980668654e-06,
      "loss": 3.2496,
      "step": 43730
    },
    {
      "epoch": 0.4448954264322917,
      "grad_norm": 15.771838188171387,
      "learning_rate": 2.9350042915325617e-06,
      "loss": 3.4466,
      "step": 43735
    },
    {
      "epoch": 0.4449462890625,
      "grad_norm": 9.760074615478516,
      "learning_rate": 2.9346105050702425e-06,
      "loss": 3.1918,
      "step": 43740
    },
    {
      "epoch": 0.4449971516927083,
      "grad_norm": 13.089041709899902,
      "learning_rate": 2.9342167074885165e-06,
      "loss": 3.3699,
      "step": 43745
    },
    {
      "epoch": 0.4450480143229167,
      "grad_norm": 17.783554077148438,
      "learning_rate": 2.933822898797459e-06,
      "loss": 3.2082,
      "step": 43750
    },
    {
      "epoch": 0.445098876953125,
      "grad_norm": 16.995149612426758,
      "learning_rate": 2.933429079007145e-06,
      "loss": 3.0578,
      "step": 43755
    },
    {
      "epoch": 0.4451497395833333,
      "grad_norm": 9.41482925415039,
      "learning_rate": 2.9330352481276513e-06,
      "loss": 3.4183,
      "step": 43760
    },
    {
      "epoch": 0.4452006022135417,
      "grad_norm": 13.689038276672363,
      "learning_rate": 2.9326414061690535e-06,
      "loss": 3.4102,
      "step": 43765
    },
    {
      "epoch": 0.44525146484375,
      "grad_norm": 17.326047897338867,
      "learning_rate": 2.9322475531414275e-06,
      "loss": 3.6235,
      "step": 43770
    },
    {
      "epoch": 0.4453023274739583,
      "grad_norm": 12.240701675415039,
      "learning_rate": 2.93185368905485e-06,
      "loss": 3.7562,
      "step": 43775
    },
    {
      "epoch": 0.4453531901041667,
      "grad_norm": 13.802271842956543,
      "learning_rate": 2.9314598139193984e-06,
      "loss": 3.4392,
      "step": 43780
    },
    {
      "epoch": 0.445404052734375,
      "grad_norm": 10.350001335144043,
      "learning_rate": 2.9310659277451505e-06,
      "loss": 3.0039,
      "step": 43785
    },
    {
      "epoch": 0.4454549153645833,
      "grad_norm": 15.589079856872559,
      "learning_rate": 2.930672030542182e-06,
      "loss": 3.3543,
      "step": 43790
    },
    {
      "epoch": 0.4455057779947917,
      "grad_norm": 14.357402801513672,
      "learning_rate": 2.9302781223205725e-06,
      "loss": 3.5534,
      "step": 43795
    },
    {
      "epoch": 0.445556640625,
      "grad_norm": 10.52519702911377,
      "learning_rate": 2.9298842030903984e-06,
      "loss": 3.1371,
      "step": 43800
    },
    {
      "epoch": 0.4456075032552083,
      "grad_norm": 14.252992630004883,
      "learning_rate": 2.929490272861739e-06,
      "loss": 3.437,
      "step": 43805
    },
    {
      "epoch": 0.4456583658854167,
      "grad_norm": 9.607766151428223,
      "learning_rate": 2.929096331644673e-06,
      "loss": 3.1274,
      "step": 43810
    },
    {
      "epoch": 0.445709228515625,
      "grad_norm": 14.653151512145996,
      "learning_rate": 2.9287023794492797e-06,
      "loss": 3.2154,
      "step": 43815
    },
    {
      "epoch": 0.4457600911458333,
      "grad_norm": 12.342843055725098,
      "learning_rate": 2.928308416285637e-06,
      "loss": 3.8825,
      "step": 43820
    },
    {
      "epoch": 0.4458109537760417,
      "grad_norm": 15.029866218566895,
      "learning_rate": 2.927914442163825e-06,
      "loss": 3.5363,
      "step": 43825
    },
    {
      "epoch": 0.44586181640625,
      "grad_norm": 9.904187202453613,
      "learning_rate": 2.9275204570939236e-06,
      "loss": 3.4545,
      "step": 43830
    },
    {
      "epoch": 0.4459126790364583,
      "grad_norm": 14.366094589233398,
      "learning_rate": 2.9271264610860124e-06,
      "loss": 3.3427,
      "step": 43835
    },
    {
      "epoch": 0.4459635416666667,
      "grad_norm": 13.560625076293945,
      "learning_rate": 2.926732454150172e-06,
      "loss": 3.8012,
      "step": 43840
    },
    {
      "epoch": 0.446014404296875,
      "grad_norm": 16.008623123168945,
      "learning_rate": 2.926338436296483e-06,
      "loss": 3.2344,
      "step": 43845
    },
    {
      "epoch": 0.4460652669270833,
      "grad_norm": 13.258788108825684,
      "learning_rate": 2.925944407535026e-06,
      "loss": 3.5272,
      "step": 43850
    },
    {
      "epoch": 0.4461161295572917,
      "grad_norm": 10.633359909057617,
      "learning_rate": 2.925550367875883e-06,
      "loss": 3.9106,
      "step": 43855
    },
    {
      "epoch": 0.4461669921875,
      "grad_norm": 13.368167877197266,
      "learning_rate": 2.925156317329133e-06,
      "loss": 3.4759,
      "step": 43860
    },
    {
      "epoch": 0.4462178548177083,
      "grad_norm": 10.804916381835938,
      "learning_rate": 2.9247622559048615e-06,
      "loss": 3.3424,
      "step": 43865
    },
    {
      "epoch": 0.4462687174479167,
      "grad_norm": 10.019235610961914,
      "learning_rate": 2.9243681836131466e-06,
      "loss": 3.8252,
      "step": 43870
    },
    {
      "epoch": 0.446319580078125,
      "grad_norm": 11.168046951293945,
      "learning_rate": 2.9239741004640736e-06,
      "loss": 3.6156,
      "step": 43875
    },
    {
      "epoch": 0.4463704427083333,
      "grad_norm": 13.641862869262695,
      "learning_rate": 2.9235800064677235e-06,
      "loss": 3.3427,
      "step": 43880
    },
    {
      "epoch": 0.4464213053385417,
      "grad_norm": 11.824005126953125,
      "learning_rate": 2.923185901634178e-06,
      "loss": 3.1257,
      "step": 43885
    },
    {
      "epoch": 0.44647216796875,
      "grad_norm": 9.489578247070312,
      "learning_rate": 2.9227917859735233e-06,
      "loss": 3.1383,
      "step": 43890
    },
    {
      "epoch": 0.4465230305989583,
      "grad_norm": 12.910775184631348,
      "learning_rate": 2.9223976594958402e-06,
      "loss": 3.29,
      "step": 43895
    },
    {
      "epoch": 0.4465738932291667,
      "grad_norm": 8.678617477416992,
      "learning_rate": 2.922003522211213e-06,
      "loss": 3.1416,
      "step": 43900
    },
    {
      "epoch": 0.446624755859375,
      "grad_norm": 14.237542152404785,
      "learning_rate": 2.9216093741297257e-06,
      "loss": 3.3661,
      "step": 43905
    },
    {
      "epoch": 0.4466756184895833,
      "grad_norm": 13.427087783813477,
      "learning_rate": 2.9212152152614625e-06,
      "loss": 3.5646,
      "step": 43910
    },
    {
      "epoch": 0.4467264811197917,
      "grad_norm": 17.48465347290039,
      "learning_rate": 2.9208210456165083e-06,
      "loss": 3.2427,
      "step": 43915
    },
    {
      "epoch": 0.44677734375,
      "grad_norm": 10.90257740020752,
      "learning_rate": 2.920426865204947e-06,
      "loss": 3.5392,
      "step": 43920
    },
    {
      "epoch": 0.4468282063802083,
      "grad_norm": 9.019089698791504,
      "learning_rate": 2.920032674036864e-06,
      "loss": 3.1428,
      "step": 43925
    },
    {
      "epoch": 0.4468790690104167,
      "grad_norm": 12.710570335388184,
      "learning_rate": 2.9196384721223448e-06,
      "loss": 3.3098,
      "step": 43930
    },
    {
      "epoch": 0.446929931640625,
      "grad_norm": 10.439959526062012,
      "learning_rate": 2.9192442594714747e-06,
      "loss": 3.0495,
      "step": 43935
    },
    {
      "epoch": 0.4469807942708333,
      "grad_norm": 10.362679481506348,
      "learning_rate": 2.9188500360943393e-06,
      "loss": 3.5898,
      "step": 43940
    },
    {
      "epoch": 0.4470316569010417,
      "grad_norm": 10.04350471496582,
      "learning_rate": 2.918455802001026e-06,
      "loss": 3.1113,
      "step": 43945
    },
    {
      "epoch": 0.44708251953125,
      "grad_norm": 10.389842987060547,
      "learning_rate": 2.9180615572016187e-06,
      "loss": 3.0661,
      "step": 43950
    },
    {
      "epoch": 0.4471333821614583,
      "grad_norm": 7.375126361846924,
      "learning_rate": 2.9176673017062064e-06,
      "loss": 3.5667,
      "step": 43955
    },
    {
      "epoch": 0.4471842447916667,
      "grad_norm": 11.38521957397461,
      "learning_rate": 2.9172730355248747e-06,
      "loss": 3.1038,
      "step": 43960
    },
    {
      "epoch": 0.447235107421875,
      "grad_norm": 11.45709228515625,
      "learning_rate": 2.9168787586677116e-06,
      "loss": 3.347,
      "step": 43965
    },
    {
      "epoch": 0.4472859700520833,
      "grad_norm": 14.73982048034668,
      "learning_rate": 2.916484471144805e-06,
      "loss": 3.2477,
      "step": 43970
    },
    {
      "epoch": 0.4473368326822917,
      "grad_norm": 13.484076499938965,
      "learning_rate": 2.9160901729662404e-06,
      "loss": 3.463,
      "step": 43975
    },
    {
      "epoch": 0.4473876953125,
      "grad_norm": 14.229711532592773,
      "learning_rate": 2.9156958641421085e-06,
      "loss": 3.2316,
      "step": 43980
    },
    {
      "epoch": 0.4474385579427083,
      "grad_norm": 12.872627258300781,
      "learning_rate": 2.9153015446824967e-06,
      "loss": 3.1471,
      "step": 43985
    },
    {
      "epoch": 0.4474894205729167,
      "grad_norm": 11.86507797241211,
      "learning_rate": 2.9149072145974923e-06,
      "loss": 3.5043,
      "step": 43990
    },
    {
      "epoch": 0.447540283203125,
      "grad_norm": 12.148114204406738,
      "learning_rate": 2.9145128738971856e-06,
      "loss": 3.3233,
      "step": 43995
    },
    {
      "epoch": 0.4475911458333333,
      "grad_norm": 16.555404663085938,
      "learning_rate": 2.914118522591665e-06,
      "loss": 3.6835,
      "step": 44000
    },
    {
      "epoch": 0.4476420084635417,
      "grad_norm": 10.267139434814453,
      "learning_rate": 2.9137241606910206e-06,
      "loss": 3.4503,
      "step": 44005
    },
    {
      "epoch": 0.44769287109375,
      "grad_norm": 7.522677898406982,
      "learning_rate": 2.9133297882053413e-06,
      "loss": 3.1649,
      "step": 44010
    },
    {
      "epoch": 0.4477437337239583,
      "grad_norm": 14.890913009643555,
      "learning_rate": 2.912935405144718e-06,
      "loss": 3.2125,
      "step": 44015
    },
    {
      "epoch": 0.4477945963541667,
      "grad_norm": 15.445168495178223,
      "learning_rate": 2.912541011519239e-06,
      "loss": 3.6791,
      "step": 44020
    },
    {
      "epoch": 0.447845458984375,
      "grad_norm": 11.479595184326172,
      "learning_rate": 2.912146607338997e-06,
      "loss": 3.3403,
      "step": 44025
    },
    {
      "epoch": 0.4478963216145833,
      "grad_norm": 9.104013442993164,
      "learning_rate": 2.911752192614081e-06,
      "loss": 3.2455,
      "step": 44030
    },
    {
      "epoch": 0.4479471842447917,
      "grad_norm": 12.49300479888916,
      "learning_rate": 2.9113577673545845e-06,
      "loss": 3.3308,
      "step": 44035
    },
    {
      "epoch": 0.447998046875,
      "grad_norm": 11.121861457824707,
      "learning_rate": 2.9109633315705947e-06,
      "loss": 3.3387,
      "step": 44040
    },
    {
      "epoch": 0.4480489095052083,
      "grad_norm": 14.24190902709961,
      "learning_rate": 2.910568885272207e-06,
      "loss": 3.2661,
      "step": 44045
    },
    {
      "epoch": 0.4480997721354167,
      "grad_norm": 9.73490047454834,
      "learning_rate": 2.910174428469511e-06,
      "loss": 3.3985,
      "step": 44050
    },
    {
      "epoch": 0.448150634765625,
      "grad_norm": 7.493213653564453,
      "learning_rate": 2.9097799611726e-06,
      "loss": 3.3632,
      "step": 44055
    },
    {
      "epoch": 0.4482014973958333,
      "grad_norm": 14.473955154418945,
      "learning_rate": 2.909385483391565e-06,
      "loss": 3.1905,
      "step": 44060
    },
    {
      "epoch": 0.4482523600260417,
      "grad_norm": 13.767448425292969,
      "learning_rate": 2.9089909951365003e-06,
      "loss": 3.6482,
      "step": 44065
    },
    {
      "epoch": 0.44830322265625,
      "grad_norm": 15.65981388092041,
      "learning_rate": 2.9085964964174974e-06,
      "loss": 3.3085,
      "step": 44070
    },
    {
      "epoch": 0.4483540852864583,
      "grad_norm": 11.123335838317871,
      "learning_rate": 2.9082019872446503e-06,
      "loss": 3.3357,
      "step": 44075
    },
    {
      "epoch": 0.4484049479166667,
      "grad_norm": 11.893462181091309,
      "learning_rate": 2.907807467628052e-06,
      "loss": 3.3115,
      "step": 44080
    },
    {
      "epoch": 0.448455810546875,
      "grad_norm": 11.333732604980469,
      "learning_rate": 2.907412937577796e-06,
      "loss": 2.977,
      "step": 44085
    },
    {
      "epoch": 0.4485066731770833,
      "grad_norm": 12.191730499267578,
      "learning_rate": 2.907018397103977e-06,
      "loss": 3.4391,
      "step": 44090
    },
    {
      "epoch": 0.4485575358072917,
      "grad_norm": 10.514629364013672,
      "learning_rate": 2.9066238462166885e-06,
      "loss": 3.3427,
      "step": 44095
    },
    {
      "epoch": 0.4486083984375,
      "grad_norm": 15.09231948852539,
      "learning_rate": 2.906229284926026e-06,
      "loss": 3.558,
      "step": 44100
    },
    {
      "epoch": 0.4486592610677083,
      "grad_norm": 7.452277183532715,
      "learning_rate": 2.9058347132420822e-06,
      "loss": 3.3224,
      "step": 44105
    },
    {
      "epoch": 0.4487101236979167,
      "grad_norm": 11.329075813293457,
      "learning_rate": 2.905440131174955e-06,
      "loss": 3.7236,
      "step": 44110
    },
    {
      "epoch": 0.448760986328125,
      "grad_norm": 9.755703926086426,
      "learning_rate": 2.9050455387347367e-06,
      "loss": 3.1724,
      "step": 44115
    },
    {
      "epoch": 0.4488118489583333,
      "grad_norm": 15.50869083404541,
      "learning_rate": 2.9046509359315252e-06,
      "loss": 3.201,
      "step": 44120
    },
    {
      "epoch": 0.4488627115885417,
      "grad_norm": 15.588220596313477,
      "learning_rate": 2.904256322775415e-06,
      "loss": 3.4182,
      "step": 44125
    },
    {
      "epoch": 0.44891357421875,
      "grad_norm": 6.770971298217773,
      "learning_rate": 2.9038616992765024e-06,
      "loss": 3.1646,
      "step": 44130
    },
    {
      "epoch": 0.4489644368489583,
      "grad_norm": 13.444995880126953,
      "learning_rate": 2.9034670654448842e-06,
      "loss": 3.4471,
      "step": 44135
    },
    {
      "epoch": 0.4490152994791667,
      "grad_norm": 10.779013633728027,
      "learning_rate": 2.9030724212906577e-06,
      "loss": 3.3747,
      "step": 44140
    },
    {
      "epoch": 0.449066162109375,
      "grad_norm": 11.873066902160645,
      "learning_rate": 2.902677766823917e-06,
      "loss": 3.4175,
      "step": 44145
    },
    {
      "epoch": 0.4491170247395833,
      "grad_norm": 16.036754608154297,
      "learning_rate": 2.9022831020547624e-06,
      "loss": 3.3016,
      "step": 44150
    },
    {
      "epoch": 0.4491678873697917,
      "grad_norm": 10.282904624938965,
      "learning_rate": 2.90188842699329e-06,
      "loss": 3.101,
      "step": 44155
    },
    {
      "epoch": 0.44921875,
      "grad_norm": 14.96367073059082,
      "learning_rate": 2.901493741649597e-06,
      "loss": 3.606,
      "step": 44160
    },
    {
      "epoch": 0.4492696126302083,
      "grad_norm": 9.411566734313965,
      "learning_rate": 2.901099046033782e-06,
      "loss": 3.7507,
      "step": 44165
    },
    {
      "epoch": 0.4493204752604167,
      "grad_norm": 9.179765701293945,
      "learning_rate": 2.9007043401559427e-06,
      "loss": 3.7467,
      "step": 44170
    },
    {
      "epoch": 0.449371337890625,
      "grad_norm": 11.785852432250977,
      "learning_rate": 2.900309624026178e-06,
      "loss": 3.5494,
      "step": 44175
    },
    {
      "epoch": 0.4494222005208333,
      "grad_norm": 14.475744247436523,
      "learning_rate": 2.8999148976545868e-06,
      "loss": 3.4811,
      "step": 44180
    },
    {
      "epoch": 0.4494730631510417,
      "grad_norm": 10.330108642578125,
      "learning_rate": 2.899520161051267e-06,
      "loss": 3.1912,
      "step": 44185
    },
    {
      "epoch": 0.44952392578125,
      "grad_norm": 10.2774019241333,
      "learning_rate": 2.8991254142263187e-06,
      "loss": 3.1315,
      "step": 44190
    },
    {
      "epoch": 0.4495747884114583,
      "grad_norm": 8.573261260986328,
      "learning_rate": 2.8987306571898417e-06,
      "loss": 3.4502,
      "step": 44195
    },
    {
      "epoch": 0.4496256510416667,
      "grad_norm": 7.999054908752441,
      "learning_rate": 2.898335889951935e-06,
      "loss": 4.0442,
      "step": 44200
    },
    {
      "epoch": 0.449676513671875,
      "grad_norm": 20.809791564941406,
      "learning_rate": 2.8979411125226997e-06,
      "loss": 3.6222,
      "step": 44205
    },
    {
      "epoch": 0.4497273763020833,
      "grad_norm": 8.303959846496582,
      "learning_rate": 2.8975463249122343e-06,
      "loss": 3.0456,
      "step": 44210
    },
    {
      "epoch": 0.4497782389322917,
      "grad_norm": 12.522747039794922,
      "learning_rate": 2.897151527130641e-06,
      "loss": 3.1759,
      "step": 44215
    },
    {
      "epoch": 0.4498291015625,
      "grad_norm": 15.157700538635254,
      "learning_rate": 2.8967567191880203e-06,
      "loss": 3.3152,
      "step": 44220
    },
    {
      "epoch": 0.4498799641927083,
      "grad_norm": 13.41623306274414,
      "learning_rate": 2.8963619010944722e-06,
      "loss": 3.3587,
      "step": 44225
    },
    {
      "epoch": 0.4499308268229167,
      "grad_norm": 8.851814270019531,
      "learning_rate": 2.895967072860099e-06,
      "loss": 3.3081,
      "step": 44230
    },
    {
      "epoch": 0.449981689453125,
      "grad_norm": 13.496138572692871,
      "learning_rate": 2.895572234495003e-06,
      "loss": 3.3986,
      "step": 44235
    },
    {
      "epoch": 0.4500325520833333,
      "grad_norm": 7.6826629638671875,
      "learning_rate": 2.895177386009284e-06,
      "loss": 3.2947,
      "step": 44240
    },
    {
      "epoch": 0.4500834147135417,
      "grad_norm": 9.343023300170898,
      "learning_rate": 2.8947825274130463e-06,
      "loss": 3.2867,
      "step": 44245
    },
    {
      "epoch": 0.45013427734375,
      "grad_norm": 8.761979103088379,
      "learning_rate": 2.89438765871639e-06,
      "loss": 3.3941,
      "step": 44250
    },
    {
      "epoch": 0.4501851399739583,
      "grad_norm": 8.454582214355469,
      "learning_rate": 2.89399277992942e-06,
      "loss": 3.3526,
      "step": 44255
    },
    {
      "epoch": 0.4502360026041667,
      "grad_norm": 15.36679458618164,
      "learning_rate": 2.893597891062237e-06,
      "loss": 3.0334,
      "step": 44260
    },
    {
      "epoch": 0.450286865234375,
      "grad_norm": 8.259591102600098,
      "learning_rate": 2.8932029921249466e-06,
      "loss": 3.1613,
      "step": 44265
    },
    {
      "epoch": 0.4503377278645833,
      "grad_norm": 11.846033096313477,
      "learning_rate": 2.8928080831276503e-06,
      "loss": 3.1133,
      "step": 44270
    },
    {
      "epoch": 0.4503885904947917,
      "grad_norm": 14.526901245117188,
      "learning_rate": 2.8924131640804515e-06,
      "loss": 3.0744,
      "step": 44275
    },
    {
      "epoch": 0.450439453125,
      "grad_norm": 10.386990547180176,
      "learning_rate": 2.8920182349934556e-06,
      "loss": 3.8038,
      "step": 44280
    },
    {
      "epoch": 0.4504903157552083,
      "grad_norm": 12.09192180633545,
      "learning_rate": 2.8916232958767667e-06,
      "loss": 3.1284,
      "step": 44285
    },
    {
      "epoch": 0.4505411783854167,
      "grad_norm": 7.6954569816589355,
      "learning_rate": 2.8912283467404873e-06,
      "loss": 3.095,
      "step": 44290
    },
    {
      "epoch": 0.450592041015625,
      "grad_norm": 10.720759391784668,
      "learning_rate": 2.8908333875947243e-06,
      "loss": 3.0396,
      "step": 44295
    },
    {
      "epoch": 0.4506429036458333,
      "grad_norm": 8.4306001663208,
      "learning_rate": 2.890438418449581e-06,
      "loss": 3.2896,
      "step": 44300
    },
    {
      "epoch": 0.4506937662760417,
      "grad_norm": 11.261151313781738,
      "learning_rate": 2.890043439315164e-06,
      "loss": 3.3463,
      "step": 44305
    },
    {
      "epoch": 0.45074462890625,
      "grad_norm": 10.319689750671387,
      "learning_rate": 2.889648450201578e-06,
      "loss": 3.2779,
      "step": 44310
    },
    {
      "epoch": 0.4507954915364583,
      "grad_norm": 11.34036636352539,
      "learning_rate": 2.889253451118928e-06,
      "loss": 3.4178,
      "step": 44315
    },
    {
      "epoch": 0.4508463541666667,
      "grad_norm": 8.487945556640625,
      "learning_rate": 2.888858442077322e-06,
      "loss": 3.0309,
      "step": 44320
    },
    {
      "epoch": 0.450897216796875,
      "grad_norm": 10.793403625488281,
      "learning_rate": 2.8884634230868636e-06,
      "loss": 3.4121,
      "step": 44325
    },
    {
      "epoch": 0.4509480794270833,
      "grad_norm": 11.224672317504883,
      "learning_rate": 2.8880683941576615e-06,
      "loss": 3.4405,
      "step": 44330
    },
    {
      "epoch": 0.4509989420572917,
      "grad_norm": 11.25199031829834,
      "learning_rate": 2.8876733552998216e-06,
      "loss": 3.7493,
      "step": 44335
    },
    {
      "epoch": 0.4510498046875,
      "grad_norm": 9.063387870788574,
      "learning_rate": 2.88727830652345e-06,
      "loss": 3.3024,
      "step": 44340
    },
    {
      "epoch": 0.4511006673177083,
      "grad_norm": 10.825544357299805,
      "learning_rate": 2.886883247838655e-06,
      "loss": 3.8387,
      "step": 44345
    },
    {
      "epoch": 0.4511515299479167,
      "grad_norm": 9.08434772491455,
      "learning_rate": 2.8864881792555437e-06,
      "loss": 3.6101,
      "step": 44350
    },
    {
      "epoch": 0.451202392578125,
      "grad_norm": 12.246991157531738,
      "learning_rate": 2.8860931007842248e-06,
      "loss": 3.5041,
      "step": 44355
    },
    {
      "epoch": 0.4512532552083333,
      "grad_norm": 7.686623573303223,
      "learning_rate": 2.885698012434805e-06,
      "loss": 3.4835,
      "step": 44360
    },
    {
      "epoch": 0.4513041178385417,
      "grad_norm": 7.685125827789307,
      "learning_rate": 2.885302914217392e-06,
      "loss": 3.4892,
      "step": 44365
    },
    {
      "epoch": 0.45135498046875,
      "grad_norm": 8.283895492553711,
      "learning_rate": 2.8849078061420964e-06,
      "loss": 3.0764,
      "step": 44370
    },
    {
      "epoch": 0.4514058430989583,
      "grad_norm": 14.107020378112793,
      "learning_rate": 2.8845126882190257e-06,
      "loss": 3.5087,
      "step": 44375
    },
    {
      "epoch": 0.4514567057291667,
      "grad_norm": 14.426298141479492,
      "learning_rate": 2.8841175604582882e-06,
      "loss": 3.343,
      "step": 44380
    },
    {
      "epoch": 0.451507568359375,
      "grad_norm": 9.314127922058105,
      "learning_rate": 2.883722422869995e-06,
      "loss": 3.2457,
      "step": 44385
    },
    {
      "epoch": 0.4515584309895833,
      "grad_norm": 14.12999439239502,
      "learning_rate": 2.883327275464254e-06,
      "loss": 3.0399,
      "step": 44390
    },
    {
      "epoch": 0.4516092936197917,
      "grad_norm": 14.253714561462402,
      "learning_rate": 2.8829321182511756e-06,
      "loss": 3.6243,
      "step": 44395
    },
    {
      "epoch": 0.45166015625,
      "grad_norm": 15.71931266784668,
      "learning_rate": 2.8825369512408697e-06,
      "loss": 3.4012,
      "step": 44400
    },
    {
      "epoch": 0.4517110188802083,
      "grad_norm": 8.17529582977295,
      "learning_rate": 2.8821417744434468e-06,
      "loss": 2.9436,
      "step": 44405
    },
    {
      "epoch": 0.4517618815104167,
      "grad_norm": 9.646404266357422,
      "learning_rate": 2.8817465878690164e-06,
      "loss": 3.4106,
      "step": 44410
    },
    {
      "epoch": 0.451812744140625,
      "grad_norm": 8.477702140808105,
      "learning_rate": 2.881351391527691e-06,
      "loss": 2.9446,
      "step": 44415
    },
    {
      "epoch": 0.4518636067708333,
      "grad_norm": 9.89619255065918,
      "learning_rate": 2.8809561854295798e-06,
      "loss": 3.3598,
      "step": 44420
    },
    {
      "epoch": 0.4519144694010417,
      "grad_norm": 12.336702346801758,
      "learning_rate": 2.8805609695847964e-06,
      "loss": 3.8555,
      "step": 44425
    },
    {
      "epoch": 0.45196533203125,
      "grad_norm": 10.78642463684082,
      "learning_rate": 2.8801657440034493e-06,
      "loss": 3.5572,
      "step": 44430
    },
    {
      "epoch": 0.4520161946614583,
      "grad_norm": 16.197214126586914,
      "learning_rate": 2.879770508695652e-06,
      "loss": 3.1144,
      "step": 44435
    },
    {
      "epoch": 0.4520670572916667,
      "grad_norm": 15.63759708404541,
      "learning_rate": 2.8793752636715168e-06,
      "loss": 3.0399,
      "step": 44440
    },
    {
      "epoch": 0.452117919921875,
      "grad_norm": 9.543905258178711,
      "learning_rate": 2.8789800089411546e-06,
      "loss": 3.7195,
      "step": 44445
    },
    {
      "epoch": 0.4521687825520833,
      "grad_norm": 9.484859466552734,
      "learning_rate": 2.8785847445146792e-06,
      "loss": 3.5228,
      "step": 44450
    },
    {
      "epoch": 0.4522196451822917,
      "grad_norm": 11.978971481323242,
      "learning_rate": 2.878189470402203e-06,
      "loss": 3.3213,
      "step": 44455
    },
    {
      "epoch": 0.4522705078125,
      "grad_norm": 16.76762580871582,
      "learning_rate": 2.8777941866138385e-06,
      "loss": 3.2711,
      "step": 44460
    },
    {
      "epoch": 0.4523213704427083,
      "grad_norm": 17.06298065185547,
      "learning_rate": 2.8773988931596995e-06,
      "loss": 3.1585,
      "step": 44465
    },
    {
      "epoch": 0.4523722330729167,
      "grad_norm": 10.328750610351562,
      "learning_rate": 2.877003590049899e-06,
      "loss": 3.1892,
      "step": 44470
    },
    {
      "epoch": 0.452423095703125,
      "grad_norm": 10.768985748291016,
      "learning_rate": 2.8766082772945514e-06,
      "loss": 3.5259,
      "step": 44475
    },
    {
      "epoch": 0.4524739583333333,
      "grad_norm": 9.1839599609375,
      "learning_rate": 2.8762129549037705e-06,
      "loss": 3.726,
      "step": 44480
    },
    {
      "epoch": 0.4525248209635417,
      "grad_norm": 8.133028984069824,
      "learning_rate": 2.8758176228876688e-06,
      "loss": 3.1541,
      "step": 44485
    },
    {
      "epoch": 0.45257568359375,
      "grad_norm": 11.028250694274902,
      "learning_rate": 2.875422281256364e-06,
      "loss": 3.9127,
      "step": 44490
    },
    {
      "epoch": 0.4526265462239583,
      "grad_norm": 14.899149894714355,
      "learning_rate": 2.8750269300199684e-06,
      "loss": 3.6109,
      "step": 44495
    },
    {
      "epoch": 0.4526774088541667,
      "grad_norm": 13.18986701965332,
      "learning_rate": 2.874631569188598e-06,
      "loss": 3.4452,
      "step": 44500
    },
    {
      "epoch": 0.452728271484375,
      "grad_norm": 9.838666915893555,
      "learning_rate": 2.8742361987723676e-06,
      "loss": 3.1275,
      "step": 44505
    },
    {
      "epoch": 0.4527791341145833,
      "grad_norm": 9.220782279968262,
      "learning_rate": 2.8738408187813925e-06,
      "loss": 3.2301,
      "step": 44510
    },
    {
      "epoch": 0.4528299967447917,
      "grad_norm": 15.787191390991211,
      "learning_rate": 2.8734454292257883e-06,
      "loss": 3.3359,
      "step": 44515
    },
    {
      "epoch": 0.452880859375,
      "grad_norm": 10.859371185302734,
      "learning_rate": 2.8730500301156724e-06,
      "loss": 3.1415,
      "step": 44520
    },
    {
      "epoch": 0.4529317220052083,
      "grad_norm": 16.449567794799805,
      "learning_rate": 2.8726546214611595e-06,
      "loss": 3.4536,
      "step": 44525
    },
    {
      "epoch": 0.4529825846354167,
      "grad_norm": 14.785994529724121,
      "learning_rate": 2.8722592032723663e-06,
      "loss": 3.393,
      "step": 44530
    },
    {
      "epoch": 0.453033447265625,
      "grad_norm": 14.163969039916992,
      "learning_rate": 2.8718637755594093e-06,
      "loss": 3.3349,
      "step": 44535
    },
    {
      "epoch": 0.4530843098958333,
      "grad_norm": 10.803580284118652,
      "learning_rate": 2.8714683383324067e-06,
      "loss": 3.0007,
      "step": 44540
    },
    {
      "epoch": 0.4531351725260417,
      "grad_norm": 8.182945251464844,
      "learning_rate": 2.8710728916014734e-06,
      "loss": 3.4462,
      "step": 44545
    },
    {
      "epoch": 0.45318603515625,
      "grad_norm": 11.769540786743164,
      "learning_rate": 2.8706774353767292e-06,
      "loss": 3.1659,
      "step": 44550
    },
    {
      "epoch": 0.4532368977864583,
      "grad_norm": 12.93024730682373,
      "learning_rate": 2.87028196966829e-06,
      "loss": 3.49,
      "step": 44555
    },
    {
      "epoch": 0.4532877604166667,
      "grad_norm": 12.073081970214844,
      "learning_rate": 2.8698864944862746e-06,
      "loss": 3.4769,
      "step": 44560
    },
    {
      "epoch": 0.453338623046875,
      "grad_norm": 8.057173728942871,
      "learning_rate": 2.8694910098408006e-06,
      "loss": 3.081,
      "step": 44565
    },
    {
      "epoch": 0.4533894856770833,
      "grad_norm": 15.351044654846191,
      "learning_rate": 2.8690955157419874e-06,
      "loss": 3.0802,
      "step": 44570
    },
    {
      "epoch": 0.4534403483072917,
      "grad_norm": 8.478313446044922,
      "learning_rate": 2.8687000121999525e-06,
      "loss": 3.3409,
      "step": 44575
    },
    {
      "epoch": 0.4534912109375,
      "grad_norm": 10.503058433532715,
      "learning_rate": 2.868304499224815e-06,
      "loss": 3.1832,
      "step": 44580
    },
    {
      "epoch": 0.4535420735677083,
      "grad_norm": 14.508081436157227,
      "learning_rate": 2.8679089768266945e-06,
      "loss": 3.077,
      "step": 44585
    },
    {
      "epoch": 0.4535929361979167,
      "grad_norm": 12.231405258178711,
      "learning_rate": 2.86751344501571e-06,
      "loss": 3.0866,
      "step": 44590
    },
    {
      "epoch": 0.453643798828125,
      "grad_norm": 10.745758056640625,
      "learning_rate": 2.8671179038019813e-06,
      "loss": 3.1604,
      "step": 44595
    },
    {
      "epoch": 0.4536946614583333,
      "grad_norm": 13.465664863586426,
      "learning_rate": 2.8667223531956274e-06,
      "loss": 3.2526,
      "step": 44600
    },
    {
      "epoch": 0.4537455240885417,
      "grad_norm": 11.453778266906738,
      "learning_rate": 2.8663267932067697e-06,
      "loss": 3.3491,
      "step": 44605
    },
    {
      "epoch": 0.45379638671875,
      "grad_norm": 11.788174629211426,
      "learning_rate": 2.8659312238455276e-06,
      "loss": 3.5024,
      "step": 44610
    },
    {
      "epoch": 0.4538472493489583,
      "grad_norm": 14.00549602508545,
      "learning_rate": 2.865535645122022e-06,
      "loss": 3.0471,
      "step": 44615
    },
    {
      "epoch": 0.4538981119791667,
      "grad_norm": 10.149195671081543,
      "learning_rate": 2.8651400570463728e-06,
      "loss": 3.3712,
      "step": 44620
    },
    {
      "epoch": 0.453948974609375,
      "grad_norm": 7.4141411781311035,
      "learning_rate": 2.864744459628703e-06,
      "loss": 3.2301,
      "step": 44625
    },
    {
      "epoch": 0.4539998372395833,
      "grad_norm": 11.498921394348145,
      "learning_rate": 2.8643488528791313e-06,
      "loss": 3.1663,
      "step": 44630
    },
    {
      "epoch": 0.4540506998697917,
      "grad_norm": 9.980927467346191,
      "learning_rate": 2.863953236807782e-06,
      "loss": 3.5264,
      "step": 44635
    },
    {
      "epoch": 0.4541015625,
      "grad_norm": 11.899552345275879,
      "learning_rate": 2.8635576114247744e-06,
      "loss": 3.1135,
      "step": 44640
    },
    {
      "epoch": 0.4541524251302083,
      "grad_norm": 12.848044395446777,
      "learning_rate": 2.863161976740232e-06,
      "loss": 3.4724,
      "step": 44645
    },
    {
      "epoch": 0.4542032877604167,
      "grad_norm": 14.635205268859863,
      "learning_rate": 2.862766332764276e-06,
      "loss": 3.5827,
      "step": 44650
    },
    {
      "epoch": 0.454254150390625,
      "grad_norm": 13.011972427368164,
      "learning_rate": 2.86237067950703e-06,
      "loss": 3.2725,
      "step": 44655
    },
    {
      "epoch": 0.4543050130208333,
      "grad_norm": 14.444993019104004,
      "learning_rate": 2.861975016978615e-06,
      "loss": 3.9499,
      "step": 44660
    },
    {
      "epoch": 0.4543558756510417,
      "grad_norm": 12.825224876403809,
      "learning_rate": 2.861579345189156e-06,
      "loss": 3.3978,
      "step": 44665
    },
    {
      "epoch": 0.45440673828125,
      "grad_norm": 11.300124168395996,
      "learning_rate": 2.861183664148775e-06,
      "loss": 3.2149,
      "step": 44670
    },
    {
      "epoch": 0.4544576009114583,
      "grad_norm": 16.177522659301758,
      "learning_rate": 2.860787973867595e-06,
      "loss": 3.1184,
      "step": 44675
    },
    {
      "epoch": 0.4545084635416667,
      "grad_norm": 7.653833866119385,
      "learning_rate": 2.860392274355741e-06,
      "loss": 3.395,
      "step": 44680
    },
    {
      "epoch": 0.454559326171875,
      "grad_norm": 13.640836715698242,
      "learning_rate": 2.8599965656233352e-06,
      "loss": 3.1572,
      "step": 44685
    },
    {
      "epoch": 0.4546101888020833,
      "grad_norm": 8.581427574157715,
      "learning_rate": 2.859600847680503e-06,
      "loss": 3.2961,
      "step": 44690
    },
    {
      "epoch": 0.4546610514322917,
      "grad_norm": 13.232842445373535,
      "learning_rate": 2.8592051205373683e-06,
      "loss": 2.9906,
      "step": 44695
    },
    {
      "epoch": 0.4547119140625,
      "grad_norm": 10.621664047241211,
      "learning_rate": 2.8588093842040564e-06,
      "loss": 3.5622,
      "step": 44700
    },
    {
      "epoch": 0.4547627766927083,
      "grad_norm": 11.978408813476562,
      "learning_rate": 2.8584136386906907e-06,
      "loss": 3.232,
      "step": 44705
    },
    {
      "epoch": 0.4548136393229167,
      "grad_norm": 8.554805755615234,
      "learning_rate": 2.8580178840073976e-06,
      "loss": 3.084,
      "step": 44710
    },
    {
      "epoch": 0.454864501953125,
      "grad_norm": 11.701556205749512,
      "learning_rate": 2.857622120164301e-06,
      "loss": 3.4473,
      "step": 44715
    },
    {
      "epoch": 0.4549153645833333,
      "grad_norm": 11.200251579284668,
      "learning_rate": 2.857226347171528e-06,
      "loss": 3.3229,
      "step": 44720
    },
    {
      "epoch": 0.4549662272135417,
      "grad_norm": 15.928792953491211,
      "learning_rate": 2.856830565039203e-06,
      "loss": 3.4841,
      "step": 44725
    },
    {
      "epoch": 0.45501708984375,
      "grad_norm": 8.158794403076172,
      "learning_rate": 2.8564347737774527e-06,
      "loss": 3.5875,
      "step": 44730
    },
    {
      "epoch": 0.4550679524739583,
      "grad_norm": 14.221781730651855,
      "learning_rate": 2.8560389733964035e-06,
      "loss": 3.7044,
      "step": 44735
    },
    {
      "epoch": 0.4551188151041667,
      "grad_norm": 12.200833320617676,
      "learning_rate": 2.855643163906181e-06,
      "loss": 3.1577,
      "step": 44740
    },
    {
      "epoch": 0.455169677734375,
      "grad_norm": 12.440650939941406,
      "learning_rate": 2.8552473453169126e-06,
      "loss": 3.7456,
      "step": 44745
    },
    {
      "epoch": 0.4552205403645833,
      "grad_norm": 12.854925155639648,
      "learning_rate": 2.854851517638726e-06,
      "loss": 3.5147,
      "step": 44750
    },
    {
      "epoch": 0.4552714029947917,
      "grad_norm": 12.155538558959961,
      "learning_rate": 2.8544556808817466e-06,
      "loss": 3.8193,
      "step": 44755
    },
    {
      "epoch": 0.455322265625,
      "grad_norm": 8.608756065368652,
      "learning_rate": 2.854059835056103e-06,
      "loss": 3.0574,
      "step": 44760
    },
    {
      "epoch": 0.4553731282552083,
      "grad_norm": 12.807647705078125,
      "learning_rate": 2.853663980171922e-06,
      "loss": 3.3311,
      "step": 44765
    },
    {
      "epoch": 0.4554239908854167,
      "grad_norm": 14.807835578918457,
      "learning_rate": 2.8532681162393315e-06,
      "loss": 3.1986,
      "step": 44770
    },
    {
      "epoch": 0.455474853515625,
      "grad_norm": 8.61264705657959,
      "learning_rate": 2.8528722432684608e-06,
      "loss": 3.3935,
      "step": 44775
    },
    {
      "epoch": 0.4555257161458333,
      "grad_norm": 14.185766220092773,
      "learning_rate": 2.8524763612694373e-06,
      "loss": 3.2794,
      "step": 44780
    },
    {
      "epoch": 0.4555765787760417,
      "grad_norm": 13.053300857543945,
      "learning_rate": 2.852080470252389e-06,
      "loss": 3.2844,
      "step": 44785
    },
    {
      "epoch": 0.45562744140625,
      "grad_norm": 6.562133312225342,
      "learning_rate": 2.851684570227446e-06,
      "loss": 3.4165,
      "step": 44790
    },
    {
      "epoch": 0.4556783040364583,
      "grad_norm": 13.429403305053711,
      "learning_rate": 2.8512886612047363e-06,
      "loss": 3.4167,
      "step": 44795
    },
    {
      "epoch": 0.4557291666666667,
      "grad_norm": 14.063451766967773,
      "learning_rate": 2.8508927431943895e-06,
      "loss": 3.5461,
      "step": 44800
    },
    {
      "epoch": 0.455780029296875,
      "grad_norm": 10.274933815002441,
      "learning_rate": 2.850496816206535e-06,
      "loss": 3.4237,
      "step": 44805
    },
    {
      "epoch": 0.4558308919270833,
      "grad_norm": 12.584627151489258,
      "learning_rate": 2.850100880251303e-06,
      "loss": 3.5103,
      "step": 44810
    },
    {
      "epoch": 0.4558817545572917,
      "grad_norm": 8.542360305786133,
      "learning_rate": 2.8497049353388227e-06,
      "loss": 3.4182,
      "step": 44815
    },
    {
      "epoch": 0.4559326171875,
      "grad_norm": 14.560110092163086,
      "learning_rate": 2.849308981479224e-06,
      "loss": 3.2972,
      "step": 44820
    },
    {
      "epoch": 0.4559834798177083,
      "grad_norm": 13.858331680297852,
      "learning_rate": 2.8489130186826386e-06,
      "loss": 3.4506,
      "step": 44825
    },
    {
      "epoch": 0.4560343424479167,
      "grad_norm": 9.340231895446777,
      "learning_rate": 2.848517046959196e-06,
      "loss": 2.8409,
      "step": 44830
    },
    {
      "epoch": 0.456085205078125,
      "grad_norm": 16.869752883911133,
      "learning_rate": 2.848121066319027e-06,
      "loss": 3.3378,
      "step": 44835
    },
    {
      "epoch": 0.4561360677083333,
      "grad_norm": 11.588399887084961,
      "learning_rate": 2.8477250767722634e-06,
      "loss": 3.4205,
      "step": 44840
    },
    {
      "epoch": 0.4561869303385417,
      "grad_norm": 8.269610404968262,
      "learning_rate": 2.8473290783290364e-06,
      "loss": 3.361,
      "step": 44845
    },
    {
      "epoch": 0.45623779296875,
      "grad_norm": 12.79600715637207,
      "learning_rate": 2.846933070999477e-06,
      "loss": 3.2668,
      "step": 44850
    },
    {
      "epoch": 0.4562886555989583,
      "grad_norm": 7.245375633239746,
      "learning_rate": 2.846537054793718e-06,
      "loss": 3.6122,
      "step": 44855
    },
    {
      "epoch": 0.4563395182291667,
      "grad_norm": 7.1210150718688965,
      "learning_rate": 2.8461410297218887e-06,
      "loss": 3.1442,
      "step": 44860
    },
    {
      "epoch": 0.456390380859375,
      "grad_norm": 9.796218872070312,
      "learning_rate": 2.8457449957941243e-06,
      "loss": 3.1287,
      "step": 44865
    },
    {
      "epoch": 0.4564412434895833,
      "grad_norm": 16.342761993408203,
      "learning_rate": 2.845348953020556e-06,
      "loss": 3.0487,
      "step": 44870
    },
    {
      "epoch": 0.4564921061197917,
      "grad_norm": 14.852348327636719,
      "learning_rate": 2.8449529014113162e-06,
      "loss": 3.7072,
      "step": 44875
    },
    {
      "epoch": 0.45654296875,
      "grad_norm": 8.907597541809082,
      "learning_rate": 2.8445568409765396e-06,
      "loss": 3.408,
      "step": 44880
    },
    {
      "epoch": 0.4565938313802083,
      "grad_norm": 16.051666259765625,
      "learning_rate": 2.8441607717263565e-06,
      "loss": 3.314,
      "step": 44885
    },
    {
      "epoch": 0.4566446940104167,
      "grad_norm": 13.219680786132812,
      "learning_rate": 2.8437646936709025e-06,
      "loss": 3.0836,
      "step": 44890
    },
    {
      "epoch": 0.456695556640625,
      "grad_norm": 10.055333137512207,
      "learning_rate": 2.84336860682031e-06,
      "loss": 3.2187,
      "step": 44895
    },
    {
      "epoch": 0.4567464192708333,
      "grad_norm": 11.605323791503906,
      "learning_rate": 2.842972511184712e-06,
      "loss": 3.2613,
      "step": 44900
    },
    {
      "epoch": 0.4567972819010417,
      "grad_norm": 10.680103302001953,
      "learning_rate": 2.842576406774245e-06,
      "loss": 3.2029,
      "step": 44905
    },
    {
      "epoch": 0.45684814453125,
      "grad_norm": 15.260845184326172,
      "learning_rate": 2.8421802935990412e-06,
      "loss": 3.2767,
      "step": 44910
    },
    {
      "epoch": 0.4568990071614583,
      "grad_norm": 10.041279792785645,
      "learning_rate": 2.8417841716692356e-06,
      "loss": 3.2855,
      "step": 44915
    },
    {
      "epoch": 0.4569498697916667,
      "grad_norm": 11.605247497558594,
      "learning_rate": 2.841388040994964e-06,
      "loss": 3.3748,
      "step": 44920
    },
    {
      "epoch": 0.457000732421875,
      "grad_norm": 12.599408149719238,
      "learning_rate": 2.840991901586359e-06,
      "loss": 3.1399,
      "step": 44925
    },
    {
      "epoch": 0.4570515950520833,
      "grad_norm": 8.706480979919434,
      "learning_rate": 2.8405957534535583e-06,
      "loss": 3.1279,
      "step": 44930
    },
    {
      "epoch": 0.4571024576822917,
      "grad_norm": 16.365188598632812,
      "learning_rate": 2.8401995966066947e-06,
      "loss": 3.966,
      "step": 44935
    },
    {
      "epoch": 0.4571533203125,
      "grad_norm": 14.899900436401367,
      "learning_rate": 2.839803431055906e-06,
      "loss": 3.3223,
      "step": 44940
    },
    {
      "epoch": 0.4572041829427083,
      "grad_norm": 13.398468971252441,
      "learning_rate": 2.839407256811326e-06,
      "loss": 2.9851,
      "step": 44945
    },
    {
      "epoch": 0.4572550455729167,
      "grad_norm": 11.414717674255371,
      "learning_rate": 2.839011073883093e-06,
      "loss": 3.5403,
      "step": 44950
    },
    {
      "epoch": 0.457305908203125,
      "grad_norm": 16.504152297973633,
      "learning_rate": 2.8386148822813414e-06,
      "loss": 2.9973,
      "step": 44955
    },
    {
      "epoch": 0.4573567708333333,
      "grad_norm": 13.877124786376953,
      "learning_rate": 2.8382186820162084e-06,
      "loss": 3.3215,
      "step": 44960
    },
    {
      "epoch": 0.4574076334635417,
      "grad_norm": 9.994048118591309,
      "learning_rate": 2.8378224730978305e-06,
      "loss": 3.4388,
      "step": 44965
    },
    {
      "epoch": 0.45745849609375,
      "grad_norm": 9.283815383911133,
      "learning_rate": 2.837426255536345e-06,
      "loss": 2.9213,
      "step": 44970
    },
    {
      "epoch": 0.4575093587239583,
      "grad_norm": 12.28955078125,
      "learning_rate": 2.837030029341888e-06,
      "loss": 3.1653,
      "step": 44975
    },
    {
      "epoch": 0.4575602213541667,
      "grad_norm": 18.627973556518555,
      "learning_rate": 2.836633794524598e-06,
      "loss": 3.3746,
      "step": 44980
    },
    {
      "epoch": 0.457611083984375,
      "grad_norm": 8.607205390930176,
      "learning_rate": 2.8362375510946127e-06,
      "loss": 3.2645,
      "step": 44985
    },
    {
      "epoch": 0.4576619466145833,
      "grad_norm": 9.861372947692871,
      "learning_rate": 2.835841299062068e-06,
      "loss": 3.2815,
      "step": 44990
    },
    {
      "epoch": 0.4577128092447917,
      "grad_norm": 14.02267074584961,
      "learning_rate": 2.8354450384371047e-06,
      "loss": 3.3776,
      "step": 44995
    },
    {
      "epoch": 0.457763671875,
      "grad_norm": 11.054256439208984,
      "learning_rate": 2.835048769229859e-06,
      "loss": 3.2593,
      "step": 45000
    },
    {
      "epoch": 0.4578145345052083,
      "grad_norm": 16.410503387451172,
      "learning_rate": 2.83465249145047e-06,
      "loss": 3.0917,
      "step": 45005
    },
    {
      "epoch": 0.4578653971354167,
      "grad_norm": 11.378438949584961,
      "learning_rate": 2.8342562051090762e-06,
      "loss": 3.4537,
      "step": 45010
    },
    {
      "epoch": 0.457916259765625,
      "grad_norm": 12.916954040527344,
      "learning_rate": 2.833859910215817e-06,
      "loss": 3.3718,
      "step": 45015
    },
    {
      "epoch": 0.4579671223958333,
      "grad_norm": 15.579991340637207,
      "learning_rate": 2.833463606780831e-06,
      "loss": 3.5082,
      "step": 45020
    },
    {
      "epoch": 0.4580179850260417,
      "grad_norm": 15.45205020904541,
      "learning_rate": 2.8330672948142568e-06,
      "loss": 3.216,
      "step": 45025
    },
    {
      "epoch": 0.45806884765625,
      "grad_norm": 12.469983100891113,
      "learning_rate": 2.832670974326236e-06,
      "loss": 3.6482,
      "step": 45030
    },
    {
      "epoch": 0.4581197102864583,
      "grad_norm": 13.32430362701416,
      "learning_rate": 2.8322746453269067e-06,
      "loss": 3.1862,
      "step": 45035
    },
    {
      "epoch": 0.4581705729166667,
      "grad_norm": 13.674798965454102,
      "learning_rate": 2.831878307826409e-06,
      "loss": 3.4776,
      "step": 45040
    },
    {
      "epoch": 0.458221435546875,
      "grad_norm": 9.960290908813477,
      "learning_rate": 2.8314819618348844e-06,
      "loss": 3.5553,
      "step": 45045
    },
    {
      "epoch": 0.4582722981770833,
      "grad_norm": 8.078351020812988,
      "learning_rate": 2.8310856073624715e-06,
      "loss": 3.9305,
      "step": 45050
    },
    {
      "epoch": 0.4583231608072917,
      "grad_norm": 11.871135711669922,
      "learning_rate": 2.8306892444193124e-06,
      "loss": 3.1103,
      "step": 45055
    },
    {
      "epoch": 0.4583740234375,
      "grad_norm": 14.729536056518555,
      "learning_rate": 2.830292873015547e-06,
      "loss": 3.2753,
      "step": 45060
    },
    {
      "epoch": 0.4584248860677083,
      "grad_norm": 11.23469352722168,
      "learning_rate": 2.8298964931613167e-06,
      "loss": 3.1144,
      "step": 45065
    },
    {
      "epoch": 0.4584757486979167,
      "grad_norm": 14.00146770477295,
      "learning_rate": 2.829500104866763e-06,
      "loss": 3.3374,
      "step": 45070
    },
    {
      "epoch": 0.458526611328125,
      "grad_norm": 8.97723388671875,
      "learning_rate": 2.829103708142027e-06,
      "loss": 3.3267,
      "step": 45075
    },
    {
      "epoch": 0.4585774739583333,
      "grad_norm": 11.814627647399902,
      "learning_rate": 2.8287073029972506e-06,
      "loss": 2.9089,
      "step": 45080
    },
    {
      "epoch": 0.4586283365885417,
      "grad_norm": 8.553272247314453,
      "learning_rate": 2.828310889442576e-06,
      "loss": 3.1574,
      "step": 45085
    },
    {
      "epoch": 0.45867919921875,
      "grad_norm": 7.917248725891113,
      "learning_rate": 2.8279144674881452e-06,
      "loss": 3.3913,
      "step": 45090
    },
    {
      "epoch": 0.4587300618489583,
      "grad_norm": 12.840087890625,
      "learning_rate": 2.8275180371441003e-06,
      "loss": 3.2862,
      "step": 45095
    },
    {
      "epoch": 0.4587809244791667,
      "grad_norm": 13.9429349899292,
      "learning_rate": 2.8271215984205847e-06,
      "loss": 3.7799,
      "step": 45100
    },
    {
      "epoch": 0.458831787109375,
      "grad_norm": 13.825484275817871,
      "learning_rate": 2.826725151327739e-06,
      "loss": 3.5117,
      "step": 45105
    },
    {
      "epoch": 0.4588826497395833,
      "grad_norm": 13.990893363952637,
      "learning_rate": 2.826328695875709e-06,
      "loss": 3.4283,
      "step": 45110
    },
    {
      "epoch": 0.4589335123697917,
      "grad_norm": 10.44043254852295,
      "learning_rate": 2.8259322320746363e-06,
      "loss": 3.5508,
      "step": 45115
    },
    {
      "epoch": 0.458984375,
      "grad_norm": 14.889739990234375,
      "learning_rate": 2.825535759934665e-06,
      "loss": 3.2424,
      "step": 45120
    },
    {
      "epoch": 0.4590352376302083,
      "grad_norm": 9.356514930725098,
      "learning_rate": 2.825139279465938e-06,
      "loss": 3.2146,
      "step": 45125
    },
    {
      "epoch": 0.4590861002604167,
      "grad_norm": 14.960698127746582,
      "learning_rate": 2.8247427906785994e-06,
      "loss": 3.2932,
      "step": 45130
    },
    {
      "epoch": 0.459136962890625,
      "grad_norm": 14.232577323913574,
      "learning_rate": 2.824346293582794e-06,
      "loss": 3.4178,
      "step": 45135
    },
    {
      "epoch": 0.4591878255208333,
      "grad_norm": 12.84840202331543,
      "learning_rate": 2.8239497881886655e-06,
      "loss": 3.3215,
      "step": 45140
    },
    {
      "epoch": 0.4592386881510417,
      "grad_norm": 11.659974098205566,
      "learning_rate": 2.823553274506358e-06,
      "loss": 3.2795,
      "step": 45145
    },
    {
      "epoch": 0.45928955078125,
      "grad_norm": 13.640230178833008,
      "learning_rate": 2.823156752546018e-06,
      "loss": 3.3699,
      "step": 45150
    },
    {
      "epoch": 0.4593404134114583,
      "grad_norm": 15.071539878845215,
      "learning_rate": 2.822760222317788e-06,
      "loss": 3.9149,
      "step": 45155
    },
    {
      "epoch": 0.4593912760416667,
      "grad_norm": 14.014788627624512,
      "learning_rate": 2.8223636838318134e-06,
      "loss": 3.1896,
      "step": 45160
    },
    {
      "epoch": 0.459442138671875,
      "grad_norm": 8.82933235168457,
      "learning_rate": 2.8219671370982422e-06,
      "loss": 3.1882,
      "step": 45165
    },
    {
      "epoch": 0.4594930013020833,
      "grad_norm": 11.878766059875488,
      "learning_rate": 2.821570582127216e-06,
      "loss": 3.6614,
      "step": 45170
    },
    {
      "epoch": 0.4595438639322917,
      "grad_norm": 13.7857084274292,
      "learning_rate": 2.8211740189288845e-06,
      "loss": 3.1367,
      "step": 45175
    },
    {
      "epoch": 0.4595947265625,
      "grad_norm": 12.247088432312012,
      "learning_rate": 2.8207774475133915e-06,
      "loss": 3.5135,
      "step": 45180
    },
    {
      "epoch": 0.4596455891927083,
      "grad_norm": 14.643757820129395,
      "learning_rate": 2.820380867890883e-06,
      "loss": 3.532,
      "step": 45185
    },
    {
      "epoch": 0.4596964518229167,
      "grad_norm": 16.834749221801758,
      "learning_rate": 2.819984280071506e-06,
      "loss": 3.4057,
      "step": 45190
    },
    {
      "epoch": 0.459747314453125,
      "grad_norm": 11.419657707214355,
      "learning_rate": 2.8195876840654073e-06,
      "loss": 3.3614,
      "step": 45195
    },
    {
      "epoch": 0.4597981770833333,
      "grad_norm": 15.169285774230957,
      "learning_rate": 2.8191910798827332e-06,
      "loss": 3.3418,
      "step": 45200
    },
    {
      "epoch": 0.4598490397135417,
      "grad_norm": 17.71880531311035,
      "learning_rate": 2.818794467533632e-06,
      "loss": 3.8517,
      "step": 45205
    },
    {
      "epoch": 0.45989990234375,
      "grad_norm": 15.081419944763184,
      "learning_rate": 2.818397847028248e-06,
      "loss": 3.199,
      "step": 45210
    },
    {
      "epoch": 0.4599507649739583,
      "grad_norm": 8.779073715209961,
      "learning_rate": 2.8180012183767327e-06,
      "loss": 3.139,
      "step": 45215
    },
    {
      "epoch": 0.4600016276041667,
      "grad_norm": 13.603039741516113,
      "learning_rate": 2.81760458158923e-06,
      "loss": 2.875,
      "step": 45220
    },
    {
      "epoch": 0.460052490234375,
      "grad_norm": 13.92172622680664,
      "learning_rate": 2.81720793667589e-06,
      "loss": 3.1506,
      "step": 45225
    },
    {
      "epoch": 0.4601033528645833,
      "grad_norm": 11.852885246276855,
      "learning_rate": 2.81681128364686e-06,
      "loss": 3.4814,
      "step": 45230
    },
    {
      "epoch": 0.4601542154947917,
      "grad_norm": 14.622265815734863,
      "learning_rate": 2.8164146225122886e-06,
      "loss": 3.6421,
      "step": 45235
    },
    {
      "epoch": 0.460205078125,
      "grad_norm": 7.7051496505737305,
      "learning_rate": 2.8160179532823233e-06,
      "loss": 3.9309,
      "step": 45240
    },
    {
      "epoch": 0.4602559407552083,
      "grad_norm": 11.519131660461426,
      "learning_rate": 2.8156212759671147e-06,
      "loss": 3.3426,
      "step": 45245
    },
    {
      "epoch": 0.4603068033854167,
      "grad_norm": 6.467058181762695,
      "learning_rate": 2.8152245905768093e-06,
      "loss": 3.3662,
      "step": 45250
    },
    {
      "epoch": 0.460357666015625,
      "grad_norm": 12.460644721984863,
      "learning_rate": 2.8148278971215594e-06,
      "loss": 3.3251,
      "step": 45255
    },
    {
      "epoch": 0.4604085286458333,
      "grad_norm": 12.776589393615723,
      "learning_rate": 2.8144311956115102e-06,
      "loss": 3.2917,
      "step": 45260
    },
    {
      "epoch": 0.4604593912760417,
      "grad_norm": 11.396586418151855,
      "learning_rate": 2.8140344860568143e-06,
      "loss": 3.509,
      "step": 45265
    },
    {
      "epoch": 0.46051025390625,
      "grad_norm": 8.867974281311035,
      "learning_rate": 2.813637768467621e-06,
      "loss": 3.2805,
      "step": 45270
    },
    {
      "epoch": 0.4605611165364583,
      "grad_norm": 12.355083465576172,
      "learning_rate": 2.8132410428540786e-06,
      "loss": 3.0796,
      "step": 45275
    },
    {
      "epoch": 0.4606119791666667,
      "grad_norm": 15.814553260803223,
      "learning_rate": 2.8128443092263396e-06,
      "loss": 3.3675,
      "step": 45280
    },
    {
      "epoch": 0.460662841796875,
      "grad_norm": 8.937992095947266,
      "learning_rate": 2.812447567594553e-06,
      "loss": 3.4759,
      "step": 45285
    },
    {
      "epoch": 0.4607137044270833,
      "grad_norm": 7.906702041625977,
      "learning_rate": 2.812050817968869e-06,
      "loss": 3.2888,
      "step": 45290
    },
    {
      "epoch": 0.4607645670572917,
      "grad_norm": 17.2447452545166,
      "learning_rate": 2.8116540603594383e-06,
      "loss": 3.4325,
      "step": 45295
    },
    {
      "epoch": 0.4608154296875,
      "grad_norm": 13.176146507263184,
      "learning_rate": 2.811257294776413e-06,
      "loss": 3.4879,
      "step": 45300
    },
    {
      "epoch": 0.4608662923177083,
      "grad_norm": 14.100749015808105,
      "learning_rate": 2.8108605212299435e-06,
      "loss": 3.6369,
      "step": 45305
    },
    {
      "epoch": 0.4609171549479167,
      "grad_norm": 8.977651596069336,
      "learning_rate": 2.8104637397301817e-06,
      "loss": 3.4339,
      "step": 45310
    },
    {
      "epoch": 0.460968017578125,
      "grad_norm": 6.882852554321289,
      "learning_rate": 2.8100669502872773e-06,
      "loss": 3.4842,
      "step": 45315
    },
    {
      "epoch": 0.4610188802083333,
      "grad_norm": 12.61129093170166,
      "learning_rate": 2.8096701529113847e-06,
      "loss": 3.0149,
      "step": 45320
    },
    {
      "epoch": 0.4610697428385417,
      "grad_norm": 8.961854934692383,
      "learning_rate": 2.8092733476126538e-06,
      "loss": 3.1113,
      "step": 45325
    },
    {
      "epoch": 0.46112060546875,
      "grad_norm": 11.627150535583496,
      "learning_rate": 2.8088765344012388e-06,
      "loss": 3.3965,
      "step": 45330
    },
    {
      "epoch": 0.4611714680989583,
      "grad_norm": 15.812899589538574,
      "learning_rate": 2.8084797132872897e-06,
      "loss": 3.4123,
      "step": 45335
    },
    {
      "epoch": 0.4612223307291667,
      "grad_norm": 8.41597843170166,
      "learning_rate": 2.8080828842809604e-06,
      "loss": 3.2558,
      "step": 45340
    },
    {
      "epoch": 0.461273193359375,
      "grad_norm": 7.589277744293213,
      "learning_rate": 2.807686047392404e-06,
      "loss": 3.5217,
      "step": 45345
    },
    {
      "epoch": 0.4613240559895833,
      "grad_norm": 11.874876976013184,
      "learning_rate": 2.807289202631773e-06,
      "loss": 3.4869,
      "step": 45350
    },
    {
      "epoch": 0.4613749186197917,
      "grad_norm": 12.37261962890625,
      "learning_rate": 2.8068923500092205e-06,
      "loss": 3.2697,
      "step": 45355
    },
    {
      "epoch": 0.46142578125,
      "grad_norm": 8.338521957397461,
      "learning_rate": 2.8064954895348994e-06,
      "loss": 3.5507,
      "step": 45360
    },
    {
      "epoch": 0.4614766438802083,
      "grad_norm": 7.475709915161133,
      "learning_rate": 2.8060986212189645e-06,
      "loss": 3.2731,
      "step": 45365
    },
    {
      "epoch": 0.4615275065104167,
      "grad_norm": 17.138797760009766,
      "learning_rate": 2.805701745071569e-06,
      "loss": 3.1835,
      "step": 45370
    },
    {
      "epoch": 0.461578369140625,
      "grad_norm": 13.825019836425781,
      "learning_rate": 2.8053048611028664e-06,
      "loss": 3.4295,
      "step": 45375
    },
    {
      "epoch": 0.4616292317708333,
      "grad_norm": 14.243266105651855,
      "learning_rate": 2.8049079693230114e-06,
      "loss": 3.3346,
      "step": 45380
    },
    {
      "epoch": 0.4616800944010417,
      "grad_norm": 12.526808738708496,
      "learning_rate": 2.804511069742159e-06,
      "loss": 3.2525,
      "step": 45385
    },
    {
      "epoch": 0.46173095703125,
      "grad_norm": 12.049077033996582,
      "learning_rate": 2.804114162370462e-06,
      "loss": 3.463,
      "step": 45390
    },
    {
      "epoch": 0.4617818196614583,
      "grad_norm": 7.687005519866943,
      "learning_rate": 2.8037172472180766e-06,
      "loss": 2.9475,
      "step": 45395
    },
    {
      "epoch": 0.4618326822916667,
      "grad_norm": 10.13161563873291,
      "learning_rate": 2.803320324295158e-06,
      "loss": 3.3653,
      "step": 45400
    },
    {
      "epoch": 0.461883544921875,
      "grad_norm": 7.479098796844482,
      "learning_rate": 2.80292339361186e-06,
      "loss": 3.8038,
      "step": 45405
    },
    {
      "epoch": 0.4619344075520833,
      "grad_norm": 13.849470138549805,
      "learning_rate": 2.8025264551783388e-06,
      "loss": 3.239,
      "step": 45410
    },
    {
      "epoch": 0.4619852701822917,
      "grad_norm": 12.716960906982422,
      "learning_rate": 2.80212950900475e-06,
      "loss": 3.4989,
      "step": 45415
    },
    {
      "epoch": 0.4620361328125,
      "grad_norm": 13.745390892028809,
      "learning_rate": 2.80173255510125e-06,
      "loss": 3.3843,
      "step": 45420
    },
    {
      "epoch": 0.4620869954427083,
      "grad_norm": 12.5733003616333,
      "learning_rate": 2.801335593477994e-06,
      "loss": 2.9451,
      "step": 45425
    },
    {
      "epoch": 0.4621378580729167,
      "grad_norm": 9.3145112991333,
      "learning_rate": 2.800938624145137e-06,
      "loss": 3.5294,
      "step": 45430
    },
    {
      "epoch": 0.462188720703125,
      "grad_norm": 12.413433074951172,
      "learning_rate": 2.8005416471128384e-06,
      "loss": 3.2586,
      "step": 45435
    },
    {
      "epoch": 0.4622395833333333,
      "grad_norm": 13.790407180786133,
      "learning_rate": 2.8001446623912523e-06,
      "loss": 3.599,
      "step": 45440
    },
    {
      "epoch": 0.4622904459635417,
      "grad_norm": 10.08675479888916,
      "learning_rate": 2.7997476699905352e-06,
      "loss": 3.5403,
      "step": 45445
    },
    {
      "epoch": 0.46234130859375,
      "grad_norm": 13.204947471618652,
      "learning_rate": 2.7993506699208463e-06,
      "loss": 3.2982,
      "step": 45450
    },
    {
      "epoch": 0.4623921712239583,
      "grad_norm": 15.960038185119629,
      "learning_rate": 2.798953662192341e-06,
      "loss": 3.3589,
      "step": 45455
    },
    {
      "epoch": 0.4624430338541667,
      "grad_norm": 15.051876068115234,
      "learning_rate": 2.7985566468151763e-06,
      "loss": 3.1739,
      "step": 45460
    },
    {
      "epoch": 0.462493896484375,
      "grad_norm": 11.148680686950684,
      "learning_rate": 2.7981596237995113e-06,
      "loss": 3.5542,
      "step": 45465
    },
    {
      "epoch": 0.4625447591145833,
      "grad_norm": 10.029537200927734,
      "learning_rate": 2.797762593155503e-06,
      "loss": 3.2435,
      "step": 45470
    },
    {
      "epoch": 0.4625956217447917,
      "grad_norm": 11.558324813842773,
      "learning_rate": 2.797365554893309e-06,
      "loss": 3.1746,
      "step": 45475
    },
    {
      "epoch": 0.462646484375,
      "grad_norm": 14.826126098632812,
      "learning_rate": 2.796968509023089e-06,
      "loss": 3.0957,
      "step": 45480
    },
    {
      "epoch": 0.4626973470052083,
      "grad_norm": 8.89865779876709,
      "learning_rate": 2.796571455554999e-06,
      "loss": 3.543,
      "step": 45485
    },
    {
      "epoch": 0.4627482096354167,
      "grad_norm": 13.865911483764648,
      "learning_rate": 2.7961743944991985e-06,
      "loss": 3.4724,
      "step": 45490
    },
    {
      "epoch": 0.462799072265625,
      "grad_norm": 12.434823036193848,
      "learning_rate": 2.7957773258658466e-06,
      "loss": 3.517,
      "step": 45495
    },
    {
      "epoch": 0.4628499348958333,
      "grad_norm": 10.956098556518555,
      "learning_rate": 2.7953802496651023e-06,
      "loss": 3.3514,
      "step": 45500
    },
    {
      "epoch": 0.4629007975260417,
      "grad_norm": 11.819670677185059,
      "learning_rate": 2.7949831659071237e-06,
      "loss": 3.0912,
      "step": 45505
    },
    {
      "epoch": 0.46295166015625,
      "grad_norm": 12.680577278137207,
      "learning_rate": 2.794586074602071e-06,
      "loss": 3.3441,
      "step": 45510
    },
    {
      "epoch": 0.4630025227864583,
      "grad_norm": 12.40794563293457,
      "learning_rate": 2.794188975760104e-06,
      "loss": 3.2761,
      "step": 45515
    },
    {
      "epoch": 0.4630533854166667,
      "grad_norm": 10.796905517578125,
      "learning_rate": 2.7937918693913808e-06,
      "loss": 3.3365,
      "step": 45520
    },
    {
      "epoch": 0.463104248046875,
      "grad_norm": 16.14789390563965,
      "learning_rate": 2.7933947555060627e-06,
      "loss": 3.5141,
      "step": 45525
    },
    {
      "epoch": 0.4631551106770833,
      "grad_norm": 12.617619514465332,
      "learning_rate": 2.7929976341143095e-06,
      "loss": 2.8931,
      "step": 45530
    },
    {
      "epoch": 0.4632059733072917,
      "grad_norm": 10.712579727172852,
      "learning_rate": 2.7926005052262804e-06,
      "loss": 3.657,
      "step": 45535
    },
    {
      "epoch": 0.4632568359375,
      "grad_norm": 11.029067039489746,
      "learning_rate": 2.7922033688521377e-06,
      "loss": 4.1094,
      "step": 45540
    },
    {
      "epoch": 0.4633076985677083,
      "grad_norm": 17.27669906616211,
      "learning_rate": 2.7918062250020406e-06,
      "loss": 3.5583,
      "step": 45545
    },
    {
      "epoch": 0.4633585611979167,
      "grad_norm": 17.485286712646484,
      "learning_rate": 2.7914090736861503e-06,
      "loss": 2.9874,
      "step": 45550
    },
    {
      "epoch": 0.463409423828125,
      "grad_norm": 9.873194694519043,
      "learning_rate": 2.7910119149146286e-06,
      "loss": 3.1215,
      "step": 45555
    },
    {
      "epoch": 0.4634602864583333,
      "grad_norm": 12.955085754394531,
      "learning_rate": 2.7906147486976347e-06,
      "loss": 3.2259,
      "step": 45560
    },
    {
      "epoch": 0.4635111490885417,
      "grad_norm": 11.360563278198242,
      "learning_rate": 2.790217575045333e-06,
      "loss": 2.8949,
      "step": 45565
    },
    {
      "epoch": 0.46356201171875,
      "grad_norm": 14.963033676147461,
      "learning_rate": 2.7898203939678827e-06,
      "loss": 3.4932,
      "step": 45570
    },
    {
      "epoch": 0.4636128743489583,
      "grad_norm": 7.978033542633057,
      "learning_rate": 2.789423205475446e-06,
      "loss": 3.2817,
      "step": 45575
    },
    {
      "epoch": 0.4636637369791667,
      "grad_norm": 9.456161499023438,
      "learning_rate": 2.7890260095781852e-06,
      "loss": 3.0742,
      "step": 45580
    },
    {
      "epoch": 0.463714599609375,
      "grad_norm": 11.976831436157227,
      "learning_rate": 2.788628806286263e-06,
      "loss": 3.1118,
      "step": 45585
    },
    {
      "epoch": 0.4637654622395833,
      "grad_norm": 15.199114799499512,
      "learning_rate": 2.7882315956098406e-06,
      "loss": 3.5626,
      "step": 45590
    },
    {
      "epoch": 0.4638163248697917,
      "grad_norm": 12.016143798828125,
      "learning_rate": 2.7878343775590823e-06,
      "loss": 3.4078,
      "step": 45595
    },
    {
      "epoch": 0.4638671875,
      "grad_norm": 11.1074800491333,
      "learning_rate": 2.787437152144148e-06,
      "loss": 3.0236,
      "step": 45600
    },
    {
      "epoch": 0.4639180501302083,
      "grad_norm": 12.313332557678223,
      "learning_rate": 2.7870399193752036e-06,
      "loss": 3.1036,
      "step": 45605
    },
    {
      "epoch": 0.4639689127604167,
      "grad_norm": 8.143770217895508,
      "learning_rate": 2.78664267926241e-06,
      "loss": 3.323,
      "step": 45610
    },
    {
      "epoch": 0.464019775390625,
      "grad_norm": 11.691008567810059,
      "learning_rate": 2.7862454318159323e-06,
      "loss": 3.2382,
      "step": 45615
    },
    {
      "epoch": 0.4640706380208333,
      "grad_norm": 8.609746932983398,
      "learning_rate": 2.7858481770459323e-06,
      "loss": 3.6214,
      "step": 45620
    },
    {
      "epoch": 0.4641215006510417,
      "grad_norm": 15.312708854675293,
      "learning_rate": 2.785450914962575e-06,
      "loss": 3.3398,
      "step": 45625
    },
    {
      "epoch": 0.46417236328125,
      "grad_norm": 13.080608367919922,
      "learning_rate": 2.785053645576023e-06,
      "loss": 3.0225,
      "step": 45630
    },
    {
      "epoch": 0.4642232259114583,
      "grad_norm": 7.98482084274292,
      "learning_rate": 2.7846563688964424e-06,
      "loss": 3.3137,
      "step": 45635
    },
    {
      "epoch": 0.4642740885416667,
      "grad_norm": 14.453807830810547,
      "learning_rate": 2.784259084933994e-06,
      "loss": 3.2648,
      "step": 45640
    },
    {
      "epoch": 0.464324951171875,
      "grad_norm": 8.011619567871094,
      "learning_rate": 2.783861793698846e-06,
      "loss": 3.3836,
      "step": 45645
    },
    {
      "epoch": 0.4643758138020833,
      "grad_norm": 10.647971153259277,
      "learning_rate": 2.7834644952011603e-06,
      "loss": 3.2071,
      "step": 45650
    },
    {
      "epoch": 0.4644266764322917,
      "grad_norm": 10.191885948181152,
      "learning_rate": 2.783067189451103e-06,
      "loss": 3.7233,
      "step": 45655
    },
    {
      "epoch": 0.4644775390625,
      "grad_norm": 9.946394920349121,
      "learning_rate": 2.7826698764588395e-06,
      "loss": 3.0366,
      "step": 45660
    },
    {
      "epoch": 0.4645284016927083,
      "grad_norm": 10.91606616973877,
      "learning_rate": 2.782272556234533e-06,
      "loss": 3.1228,
      "step": 45665
    },
    {
      "epoch": 0.4645792643229167,
      "grad_norm": 10.437875747680664,
      "learning_rate": 2.781875228788351e-06,
      "loss": 3.1384,
      "step": 45670
    },
    {
      "epoch": 0.464630126953125,
      "grad_norm": 9.868165969848633,
      "learning_rate": 2.7814778941304575e-06,
      "loss": 3.3624,
      "step": 45675
    },
    {
      "epoch": 0.4646809895833333,
      "grad_norm": 11.684338569641113,
      "learning_rate": 2.7810805522710193e-06,
      "loss": 3.5561,
      "step": 45680
    },
    {
      "epoch": 0.4647318522135417,
      "grad_norm": 11.288381576538086,
      "learning_rate": 2.7806832032202015e-06,
      "loss": 3.4609,
      "step": 45685
    },
    {
      "epoch": 0.46478271484375,
      "grad_norm": 8.975887298583984,
      "learning_rate": 2.78028584698817e-06,
      "loss": 3.5242,
      "step": 45690
    },
    {
      "epoch": 0.4648335774739583,
      "grad_norm": 13.19123649597168,
      "learning_rate": 2.779888483585092e-06,
      "loss": 3.4243,
      "step": 45695
    },
    {
      "epoch": 0.4648844401041667,
      "grad_norm": 9.616811752319336,
      "learning_rate": 2.7794911130211344e-06,
      "loss": 3.0956,
      "step": 45700
    },
    {
      "epoch": 0.464935302734375,
      "grad_norm": 16.0772762298584,
      "learning_rate": 2.7790937353064612e-06,
      "loss": 3.2974,
      "step": 45705
    },
    {
      "epoch": 0.4649861653645833,
      "grad_norm": 9.483195304870605,
      "learning_rate": 2.7786963504512423e-06,
      "loss": 3.244,
      "step": 45710
    },
    {
      "epoch": 0.4650370279947917,
      "grad_norm": 15.58304214477539,
      "learning_rate": 2.7782989584656424e-06,
      "loss": 3.4533,
      "step": 45715
    },
    {
      "epoch": 0.465087890625,
      "grad_norm": 9.435249328613281,
      "learning_rate": 2.77790155935983e-06,
      "loss": 3.8485,
      "step": 45720
    },
    {
      "epoch": 0.4651387532552083,
      "grad_norm": 12.428313255310059,
      "learning_rate": 2.777504153143972e-06,
      "loss": 3.3872,
      "step": 45725
    },
    {
      "epoch": 0.4651896158854167,
      "grad_norm": 15.34945297241211,
      "learning_rate": 2.7771067398282365e-06,
      "loss": 3.1382,
      "step": 45730
    },
    {
      "epoch": 0.465240478515625,
      "grad_norm": 13.501898765563965,
      "learning_rate": 2.77670931942279e-06,
      "loss": 3.186,
      "step": 45735
    },
    {
      "epoch": 0.4652913411458333,
      "grad_norm": 14.027562141418457,
      "learning_rate": 2.7763118919378015e-06,
      "loss": 3.2039,
      "step": 45740
    },
    {
      "epoch": 0.4653422037760417,
      "grad_norm": 16.236597061157227,
      "learning_rate": 2.7759144573834385e-06,
      "loss": 3.237,
      "step": 45745
    },
    {
      "epoch": 0.46539306640625,
      "grad_norm": 16.635051727294922,
      "learning_rate": 2.7755170157698703e-06,
      "loss": 3.1684,
      "step": 45750
    },
    {
      "epoch": 0.4654439290364583,
      "grad_norm": 9.709321022033691,
      "learning_rate": 2.7751195671072634e-06,
      "loss": 3.7887,
      "step": 45755
    },
    {
      "epoch": 0.4654947916666667,
      "grad_norm": 11.649523735046387,
      "learning_rate": 2.7747221114057882e-06,
      "loss": 3.6312,
      "step": 45760
    },
    {
      "epoch": 0.465545654296875,
      "grad_norm": 7.71863317489624,
      "learning_rate": 2.7743246486756133e-06,
      "loss": 4.0583,
      "step": 45765
    },
    {
      "epoch": 0.4655965169270833,
      "grad_norm": 15.480148315429688,
      "learning_rate": 2.7739271789269064e-06,
      "loss": 3.4472,
      "step": 45770
    },
    {
      "epoch": 0.4656473795572917,
      "grad_norm": 11.121807098388672,
      "learning_rate": 2.7735297021698383e-06,
      "loss": 3.2171,
      "step": 45775
    },
    {
      "epoch": 0.4656982421875,
      "grad_norm": 10.893674850463867,
      "learning_rate": 2.773132218414577e-06,
      "loss": 3.2047,
      "step": 45780
    },
    {
      "epoch": 0.4657491048177083,
      "grad_norm": 10.893601417541504,
      "learning_rate": 2.7727347276712935e-06,
      "loss": 3.7683,
      "step": 45785
    },
    {
      "epoch": 0.4657999674479167,
      "grad_norm": 13.179238319396973,
      "learning_rate": 2.7723372299501554e-06,
      "loss": 3.236,
      "step": 45790
    },
    {
      "epoch": 0.465850830078125,
      "grad_norm": 12.484517097473145,
      "learning_rate": 2.771939725261335e-06,
      "loss": 3.4028,
      "step": 45795
    },
    {
      "epoch": 0.4659016927083333,
      "grad_norm": 10.28607177734375,
      "learning_rate": 2.7715422136150003e-06,
      "loss": 3.4997,
      "step": 45800
    },
    {
      "epoch": 0.4659525553385417,
      "grad_norm": 12.38172721862793,
      "learning_rate": 2.771144695021323e-06,
      "loss": 4.0931,
      "step": 45805
    },
    {
      "epoch": 0.46600341796875,
      "grad_norm": 9.357684135437012,
      "learning_rate": 2.7707471694904726e-06,
      "loss": 3.2736,
      "step": 45810
    },
    {
      "epoch": 0.4660542805989583,
      "grad_norm": 8.547158241271973,
      "learning_rate": 2.7703496370326204e-06,
      "loss": 3.3679,
      "step": 45815
    },
    {
      "epoch": 0.4661051432291667,
      "grad_norm": 12.835251808166504,
      "learning_rate": 2.7699520976579366e-06,
      "loss": 3.553,
      "step": 45820
    },
    {
      "epoch": 0.466156005859375,
      "grad_norm": 8.556459426879883,
      "learning_rate": 2.7695545513765927e-06,
      "loss": 3.4662,
      "step": 45825
    },
    {
      "epoch": 0.4662068684895833,
      "grad_norm": 8.816329002380371,
      "learning_rate": 2.7691569981987594e-06,
      "loss": 3.277,
      "step": 45830
    },
    {
      "epoch": 0.4662577311197917,
      "grad_norm": 12.441618919372559,
      "learning_rate": 2.768759438134607e-06,
      "loss": 3.3367,
      "step": 45835
    },
    {
      "epoch": 0.46630859375,
      "grad_norm": 8.362746238708496,
      "learning_rate": 2.7683618711943093e-06,
      "loss": 3.7603,
      "step": 45840
    },
    {
      "epoch": 0.4663594563802083,
      "grad_norm": 11.711898803710938,
      "learning_rate": 2.767964297388037e-06,
      "loss": 3.4323,
      "step": 45845
    },
    {
      "epoch": 0.4664103190104167,
      "grad_norm": 14.386340141296387,
      "learning_rate": 2.76756671672596e-06,
      "loss": 3.4993,
      "step": 45850
    },
    {
      "epoch": 0.466461181640625,
      "grad_norm": 11.949966430664062,
      "learning_rate": 2.7671691292182534e-06,
      "loss": 3.3782,
      "step": 45855
    },
    {
      "epoch": 0.4665120442708333,
      "grad_norm": 8.975872993469238,
      "learning_rate": 2.7667715348750873e-06,
      "loss": 3.4422,
      "step": 45860
    },
    {
      "epoch": 0.4665629069010417,
      "grad_norm": 17.242460250854492,
      "learning_rate": 2.7663739337066352e-06,
      "loss": 3.2899,
      "step": 45865
    },
    {
      "epoch": 0.46661376953125,
      "grad_norm": 10.30756950378418,
      "learning_rate": 2.7659763257230687e-06,
      "loss": 3.4649,
      "step": 45870
    },
    {
      "epoch": 0.4666646321614583,
      "grad_norm": 13.584542274475098,
      "learning_rate": 2.765578710934561e-06,
      "loss": 3.606,
      "step": 45875
    },
    {
      "epoch": 0.4667154947916667,
      "grad_norm": 8.844463348388672,
      "learning_rate": 2.765181089351286e-06,
      "loss": 3.3156,
      "step": 45880
    },
    {
      "epoch": 0.466766357421875,
      "grad_norm": 13.212042808532715,
      "learning_rate": 2.764783460983414e-06,
      "loss": 3.1015,
      "step": 45885
    },
    {
      "epoch": 0.4668172200520833,
      "grad_norm": 11.864336967468262,
      "learning_rate": 2.7643858258411214e-06,
      "loss": 3.4215,
      "step": 45890
    },
    {
      "epoch": 0.4668680826822917,
      "grad_norm": 13.197895050048828,
      "learning_rate": 2.7639881839345795e-06,
      "loss": 3.1889,
      "step": 45895
    },
    {
      "epoch": 0.4669189453125,
      "grad_norm": 9.860047340393066,
      "learning_rate": 2.7635905352739624e-06,
      "loss": 3.5285,
      "step": 45900
    },
    {
      "epoch": 0.4669698079427083,
      "grad_norm": 10.454028129577637,
      "learning_rate": 2.7631928798694435e-06,
      "loss": 3.3109,
      "step": 45905
    },
    {
      "epoch": 0.4670206705729167,
      "grad_norm": 10.30970287322998,
      "learning_rate": 2.7627952177311978e-06,
      "loss": 3.4374,
      "step": 45910
    },
    {
      "epoch": 0.467071533203125,
      "grad_norm": 11.030384063720703,
      "learning_rate": 2.7623975488693984e-06,
      "loss": 3.2407,
      "step": 45915
    },
    {
      "epoch": 0.4671223958333333,
      "grad_norm": 8.984940528869629,
      "learning_rate": 2.7619998732942205e-06,
      "loss": 3.2256,
      "step": 45920
    },
    {
      "epoch": 0.4671732584635417,
      "grad_norm": 11.720098495483398,
      "learning_rate": 2.761602191015837e-06,
      "loss": 3.124,
      "step": 45925
    },
    {
      "epoch": 0.46722412109375,
      "grad_norm": 12.977888107299805,
      "learning_rate": 2.7612045020444243e-06,
      "loss": 3.0434,
      "step": 45930
    },
    {
      "epoch": 0.4672749837239583,
      "grad_norm": 14.955964088439941,
      "learning_rate": 2.760806806390156e-06,
      "loss": 3.2651,
      "step": 45935
    },
    {
      "epoch": 0.4673258463541667,
      "grad_norm": 10.134566307067871,
      "learning_rate": 2.7604091040632076e-06,
      "loss": 3.2583,
      "step": 45940
    },
    {
      "epoch": 0.467376708984375,
      "grad_norm": 13.452347755432129,
      "learning_rate": 2.7600113950737535e-06,
      "loss": 3.1927,
      "step": 45945
    },
    {
      "epoch": 0.4674275716145833,
      "grad_norm": 12.402140617370605,
      "learning_rate": 2.759613679431969e-06,
      "loss": 3.4382,
      "step": 45950
    },
    {
      "epoch": 0.4674784342447917,
      "grad_norm": 13.043981552124023,
      "learning_rate": 2.7592159571480317e-06,
      "loss": 3.4245,
      "step": 45955
    },
    {
      "epoch": 0.467529296875,
      "grad_norm": 15.960465431213379,
      "learning_rate": 2.758818228232114e-06,
      "loss": 3.0721,
      "step": 45960
    },
    {
      "epoch": 0.4675801595052083,
      "grad_norm": 10.671195030212402,
      "learning_rate": 2.758420492694394e-06,
      "loss": 3.0606,
      "step": 45965
    },
    {
      "epoch": 0.4676310221354167,
      "grad_norm": 12.669294357299805,
      "learning_rate": 2.758022750545047e-06,
      "loss": 3.3118,
      "step": 45970
    },
    {
      "epoch": 0.467681884765625,
      "grad_norm": 8.216130256652832,
      "learning_rate": 2.7576250017942484e-06,
      "loss": 3.3467,
      "step": 45975
    },
    {
      "epoch": 0.4677327473958333,
      "grad_norm": 11.873502731323242,
      "learning_rate": 2.757227246452176e-06,
      "loss": 3.4543,
      "step": 45980
    },
    {
      "epoch": 0.4677836100260417,
      "grad_norm": 12.08121395111084,
      "learning_rate": 2.7568294845290054e-06,
      "loss": 3.1965,
      "step": 45985
    },
    {
      "epoch": 0.46783447265625,
      "grad_norm": 10.506508827209473,
      "learning_rate": 2.756431716034913e-06,
      "loss": 3.118,
      "step": 45990
    },
    {
      "epoch": 0.4678853352864583,
      "grad_norm": 15.263740539550781,
      "learning_rate": 2.7560339409800762e-06,
      "loss": 3.3852,
      "step": 45995
    },
    {
      "epoch": 0.4679361979166667,
      "grad_norm": 11.982575416564941,
      "learning_rate": 2.755636159374672e-06,
      "loss": 3.6118,
      "step": 46000
    },
    {
      "epoch": 0.467987060546875,
      "grad_norm": 10.696041107177734,
      "learning_rate": 2.7552383712288766e-06,
      "loss": 3.3743,
      "step": 46005
    },
    {
      "epoch": 0.4680379231770833,
      "grad_norm": 13.898693084716797,
      "learning_rate": 2.754840576552868e-06,
      "loss": 3.0196,
      "step": 46010
    },
    {
      "epoch": 0.4680887858072917,
      "grad_norm": 8.868748664855957,
      "learning_rate": 2.7544427753568237e-06,
      "loss": 3.2477,
      "step": 46015
    },
    {
      "epoch": 0.4681396484375,
      "grad_norm": 15.456969261169434,
      "learning_rate": 2.754044967650922e-06,
      "loss": 3.7001,
      "step": 46020
    },
    {
      "epoch": 0.4681905110677083,
      "grad_norm": 11.392976760864258,
      "learning_rate": 2.753647153445339e-06,
      "loss": 3.349,
      "step": 46025
    },
    {
      "epoch": 0.4682413736979167,
      "grad_norm": 10.876288414001465,
      "learning_rate": 2.753249332750254e-06,
      "loss": 3.7635,
      "step": 46030
    },
    {
      "epoch": 0.468292236328125,
      "grad_norm": 19.523954391479492,
      "learning_rate": 2.7528515055758457e-06,
      "loss": 3.7524,
      "step": 46035
    },
    {
      "epoch": 0.4683430989583333,
      "grad_norm": 11.984675407409668,
      "learning_rate": 2.7524536719322903e-06,
      "loss": 3.5094,
      "step": 46040
    },
    {
      "epoch": 0.4683939615885417,
      "grad_norm": 14.17333698272705,
      "learning_rate": 2.7520558318297686e-06,
      "loss": 3.3934,
      "step": 46045
    },
    {
      "epoch": 0.46844482421875,
      "grad_norm": 13.487282752990723,
      "learning_rate": 2.7516579852784576e-06,
      "loss": 3.286,
      "step": 46050
    },
    {
      "epoch": 0.4684956868489583,
      "grad_norm": 14.06745433807373,
      "learning_rate": 2.7512601322885363e-06,
      "loss": 3.3532,
      "step": 46055
    },
    {
      "epoch": 0.4685465494791667,
      "grad_norm": 13.35567855834961,
      "learning_rate": 2.750862272870185e-06,
      "loss": 3.1667,
      "step": 46060
    },
    {
      "epoch": 0.468597412109375,
      "grad_norm": 15.46142292022705,
      "learning_rate": 2.7504644070335814e-06,
      "loss": 3.5401,
      "step": 46065
    },
    {
      "epoch": 0.4686482747395833,
      "grad_norm": 8.861741065979004,
      "learning_rate": 2.7500665347889054e-06,
      "loss": 3.4026,
      "step": 46070
    },
    {
      "epoch": 0.4686991373697917,
      "grad_norm": 13.047086715698242,
      "learning_rate": 2.7496686561463363e-06,
      "loss": 3.4934,
      "step": 46075
    },
    {
      "epoch": 0.46875,
      "grad_norm": 17.196575164794922,
      "learning_rate": 2.7492707711160537e-06,
      "loss": 3.9396,
      "step": 46080
    },
    {
      "epoch": 0.4688008626302083,
      "grad_norm": 10.11226749420166,
      "learning_rate": 2.7488728797082377e-06,
      "loss": 3.3455,
      "step": 46085
    },
    {
      "epoch": 0.4688517252604167,
      "grad_norm": 11.650439262390137,
      "learning_rate": 2.7484749819330684e-06,
      "loss": 3.3322,
      "step": 46090
    },
    {
      "epoch": 0.468902587890625,
      "grad_norm": 10.389031410217285,
      "learning_rate": 2.748077077800725e-06,
      "loss": 3.2071,
      "step": 46095
    },
    {
      "epoch": 0.4689534505208333,
      "grad_norm": 10.494956970214844,
      "learning_rate": 2.7476791673213888e-06,
      "loss": 3.2638,
      "step": 46100
    },
    {
      "epoch": 0.4690043131510417,
      "grad_norm": 8.282896995544434,
      "learning_rate": 2.7472812505052394e-06,
      "loss": 3.3806,
      "step": 46105
    },
    {
      "epoch": 0.46905517578125,
      "grad_norm": 14.236871719360352,
      "learning_rate": 2.7468833273624586e-06,
      "loss": 3.4765,
      "step": 46110
    },
    {
      "epoch": 0.4691060384114583,
      "grad_norm": 11.046870231628418,
      "learning_rate": 2.7464853979032253e-06,
      "loss": 3.3718,
      "step": 46115
    },
    {
      "epoch": 0.4691569010416667,
      "grad_norm": 14.94047737121582,
      "learning_rate": 2.746087462137722e-06,
      "loss": 3.4382,
      "step": 46120
    },
    {
      "epoch": 0.469207763671875,
      "grad_norm": 7.440110683441162,
      "learning_rate": 2.7456895200761295e-06,
      "loss": 3.5794,
      "step": 46125
    },
    {
      "epoch": 0.4692586263020833,
      "grad_norm": 11.102751731872559,
      "learning_rate": 2.745291571728629e-06,
      "loss": 3.0727,
      "step": 46130
    },
    {
      "epoch": 0.4693094889322917,
      "grad_norm": 11.534234046936035,
      "learning_rate": 2.7448936171054008e-06,
      "loss": 3.4297,
      "step": 46135
    },
    {
      "epoch": 0.4693603515625,
      "grad_norm": 10.486431121826172,
      "learning_rate": 2.7444956562166287e-06,
      "loss": 3.0962,
      "step": 46140
    },
    {
      "epoch": 0.4694112141927083,
      "grad_norm": 12.308408737182617,
      "learning_rate": 2.7440976890724926e-06,
      "loss": 3.5718,
      "step": 46145
    },
    {
      "epoch": 0.4694620768229167,
      "grad_norm": 14.86275863647461,
      "learning_rate": 2.7436997156831744e-06,
      "loss": 3.3616,
      "step": 46150
    },
    {
      "epoch": 0.469512939453125,
      "grad_norm": 10.989408493041992,
      "learning_rate": 2.743301736058858e-06,
      "loss": 3.4982,
      "step": 46155
    },
    {
      "epoch": 0.4695638020833333,
      "grad_norm": 16.7119140625,
      "learning_rate": 2.7429037502097235e-06,
      "loss": 3.5977,
      "step": 46160
    },
    {
      "epoch": 0.4696146647135417,
      "grad_norm": 10.82308292388916,
      "learning_rate": 2.7425057581459547e-06,
      "loss": 3.2861,
      "step": 46165
    },
    {
      "epoch": 0.46966552734375,
      "grad_norm": 15.950583457946777,
      "learning_rate": 2.7421077598777322e-06,
      "loss": 3.529,
      "step": 46170
    },
    {
      "epoch": 0.4697163899739583,
      "grad_norm": 10.276703834533691,
      "learning_rate": 2.7417097554152417e-06,
      "loss": 3.4298,
      "step": 46175
    },
    {
      "epoch": 0.4697672526041667,
      "grad_norm": 13.284745216369629,
      "learning_rate": 2.741311744768664e-06,
      "loss": 3.1905,
      "step": 46180
    },
    {
      "epoch": 0.469818115234375,
      "grad_norm": 10.398514747619629,
      "learning_rate": 2.7409137279481816e-06,
      "loss": 3.3221,
      "step": 46185
    },
    {
      "epoch": 0.4698689778645833,
      "grad_norm": 16.240459442138672,
      "learning_rate": 2.740515704963979e-06,
      "loss": 3.4604,
      "step": 46190
    },
    {
      "epoch": 0.4699198404947917,
      "grad_norm": 13.618303298950195,
      "learning_rate": 2.7401176758262392e-06,
      "loss": 3.6707,
      "step": 46195
    },
    {
      "epoch": 0.469970703125,
      "grad_norm": 7.832457542419434,
      "learning_rate": 2.739719640545146e-06,
      "loss": 3.0944,
      "step": 46200
    },
    {
      "epoch": 0.4700215657552083,
      "grad_norm": 8.400361061096191,
      "learning_rate": 2.739321599130883e-06,
      "loss": 4.1508,
      "step": 46205
    },
    {
      "epoch": 0.4700724283854167,
      "grad_norm": 7.780868053436279,
      "learning_rate": 2.7389235515936317e-06,
      "loss": 3.2587,
      "step": 46210
    },
    {
      "epoch": 0.470123291015625,
      "grad_norm": 15.038986206054688,
      "learning_rate": 2.73852549794358e-06,
      "loss": 3.0948,
      "step": 46215
    },
    {
      "epoch": 0.4701741536458333,
      "grad_norm": 13.191540718078613,
      "learning_rate": 2.738127438190909e-06,
      "loss": 3.2851,
      "step": 46220
    },
    {
      "epoch": 0.4702250162760417,
      "grad_norm": 12.649255752563477,
      "learning_rate": 2.7377293723458045e-06,
      "loss": 3.2563,
      "step": 46225
    },
    {
      "epoch": 0.47027587890625,
      "grad_norm": 11.454198837280273,
      "learning_rate": 2.73733130041845e-06,
      "loss": 3.1829,
      "step": 46230
    },
    {
      "epoch": 0.4703267415364583,
      "grad_norm": 14.024190902709961,
      "learning_rate": 2.7369332224190298e-06,
      "loss": 2.9479,
      "step": 46235
    },
    {
      "epoch": 0.4703776041666667,
      "grad_norm": 13.10719108581543,
      "learning_rate": 2.7365351383577297e-06,
      "loss": 3.1713,
      "step": 46240
    },
    {
      "epoch": 0.470428466796875,
      "grad_norm": 10.908052444458008,
      "learning_rate": 2.7361370482447346e-06,
      "loss": 2.944,
      "step": 46245
    },
    {
      "epoch": 0.4704793294270833,
      "grad_norm": 13.364755630493164,
      "learning_rate": 2.7357389520902282e-06,
      "loss": 3.4202,
      "step": 46250
    },
    {
      "epoch": 0.4705301920572917,
      "grad_norm": 13.245467185974121,
      "learning_rate": 2.7353408499043976e-06,
      "loss": 3.1811,
      "step": 46255
    },
    {
      "epoch": 0.4705810546875,
      "grad_norm": 10.978673934936523,
      "learning_rate": 2.734942741697427e-06,
      "loss": 3.36,
      "step": 46260
    },
    {
      "epoch": 0.4706319173177083,
      "grad_norm": 9.301791191101074,
      "learning_rate": 2.734544627479501e-06,
      "loss": 3.4495,
      "step": 46265
    },
    {
      "epoch": 0.4706827799479167,
      "grad_norm": 8.850616455078125,
      "learning_rate": 2.7341465072608074e-06,
      "loss": 3.1127,
      "step": 46270
    },
    {
      "epoch": 0.470733642578125,
      "grad_norm": 14.622509002685547,
      "learning_rate": 2.7337483810515303e-06,
      "loss": 3.2434,
      "step": 46275
    },
    {
      "epoch": 0.4707845052083333,
      "grad_norm": 8.17994213104248,
      "learning_rate": 2.7333502488618568e-06,
      "loss": 3.2975,
      "step": 46280
    },
    {
      "epoch": 0.4708353678385417,
      "grad_norm": 11.39811897277832,
      "learning_rate": 2.7329521107019723e-06,
      "loss": 3.5219,
      "step": 46285
    },
    {
      "epoch": 0.47088623046875,
      "grad_norm": 8.171241760253906,
      "learning_rate": 2.7325539665820632e-06,
      "loss": 2.908,
      "step": 46290
    },
    {
      "epoch": 0.4709370930989583,
      "grad_norm": 10.729769706726074,
      "learning_rate": 2.7321558165123157e-06,
      "loss": 2.9641,
      "step": 46295
    },
    {
      "epoch": 0.4709879557291667,
      "grad_norm": 11.02481746673584,
      "learning_rate": 2.7317576605029173e-06,
      "loss": 3.1499,
      "step": 46300
    },
    {
      "epoch": 0.471038818359375,
      "grad_norm": 12.968914031982422,
      "learning_rate": 2.7313594985640533e-06,
      "loss": 3.3203,
      "step": 46305
    },
    {
      "epoch": 0.4710896809895833,
      "grad_norm": 10.956926345825195,
      "learning_rate": 2.7309613307059125e-06,
      "loss": 3.4741,
      "step": 46310
    },
    {
      "epoch": 0.4711405436197917,
      "grad_norm": 13.60365104675293,
      "learning_rate": 2.7305631569386793e-06,
      "loss": 3.5447,
      "step": 46315
    },
    {
      "epoch": 0.47119140625,
      "grad_norm": 9.149913787841797,
      "learning_rate": 2.7301649772725435e-06,
      "loss": 3.5997,
      "step": 46320
    },
    {
      "epoch": 0.4712422688802083,
      "grad_norm": 15.682628631591797,
      "learning_rate": 2.7297667917176906e-06,
      "loss": 3.0661,
      "step": 46325
    },
    {
      "epoch": 0.4712931315104167,
      "grad_norm": 11.843368530273438,
      "learning_rate": 2.72936860028431e-06,
      "loss": 3.6361,
      "step": 46330
    },
    {
      "epoch": 0.471343994140625,
      "grad_norm": 12.912580490112305,
      "learning_rate": 2.728970402982587e-06,
      "loss": 3.3276,
      "step": 46335
    },
    {
      "epoch": 0.4713948567708333,
      "grad_norm": 13.227678298950195,
      "learning_rate": 2.728572199822711e-06,
      "loss": 3.2261,
      "step": 46340
    },
    {
      "epoch": 0.4714457194010417,
      "grad_norm": 12.401229858398438,
      "learning_rate": 2.7281739908148693e-06,
      "loss": 3.0861,
      "step": 46345
    },
    {
      "epoch": 0.47149658203125,
      "grad_norm": 10.131817817687988,
      "learning_rate": 2.7277757759692495e-06,
      "loss": 3.5494,
      "step": 46350
    },
    {
      "epoch": 0.4715474446614583,
      "grad_norm": 11.628375053405762,
      "learning_rate": 2.7273775552960413e-06,
      "loss": 3.3065,
      "step": 46355
    },
    {
      "epoch": 0.4715983072916667,
      "grad_norm": 8.976502418518066,
      "learning_rate": 2.726979328805432e-06,
      "loss": 3.2928,
      "step": 46360
    },
    {
      "epoch": 0.471649169921875,
      "grad_norm": 14.638002395629883,
      "learning_rate": 2.7265810965076102e-06,
      "loss": 3.8144,
      "step": 46365
    },
    {
      "epoch": 0.4717000325520833,
      "grad_norm": 13.228276252746582,
      "learning_rate": 2.7261828584127648e-06,
      "loss": 3.2929,
      "step": 46370
    },
    {
      "epoch": 0.4717508951822917,
      "grad_norm": 9.379998207092285,
      "learning_rate": 2.725784614531085e-06,
      "loss": 3.2328,
      "step": 46375
    },
    {
      "epoch": 0.4718017578125,
      "grad_norm": 7.740593910217285,
      "learning_rate": 2.7253863648727585e-06,
      "loss": 3.829,
      "step": 46380
    },
    {
      "epoch": 0.4718526204427083,
      "grad_norm": 13.069509506225586,
      "learning_rate": 2.7249881094479764e-06,
      "loss": 3.1788,
      "step": 46385
    },
    {
      "epoch": 0.4719034830729167,
      "grad_norm": 16.901323318481445,
      "learning_rate": 2.7245898482669265e-06,
      "loss": 3.1605,
      "step": 46390
    },
    {
      "epoch": 0.471954345703125,
      "grad_norm": 13.64307975769043,
      "learning_rate": 2.724191581339798e-06,
      "loss": 3.7023,
      "step": 46395
    },
    {
      "epoch": 0.4720052083333333,
      "grad_norm": 10.633169174194336,
      "learning_rate": 2.7237933086767814e-06,
      "loss": 3.269,
      "step": 46400
    },
    {
      "epoch": 0.4720560709635417,
      "grad_norm": 13.924692153930664,
      "learning_rate": 2.7233950302880658e-06,
      "loss": 3.2844,
      "step": 46405
    },
    {
      "epoch": 0.47210693359375,
      "grad_norm": 12.190840721130371,
      "learning_rate": 2.722996746183841e-06,
      "loss": 3.726,
      "step": 46410
    },
    {
      "epoch": 0.4721577962239583,
      "grad_norm": 12.154855728149414,
      "learning_rate": 2.7225984563742987e-06,
      "loss": 3.2895,
      "step": 46415
    },
    {
      "epoch": 0.4722086588541667,
      "grad_norm": 10.432464599609375,
      "learning_rate": 2.7222001608696264e-06,
      "loss": 3.3704,
      "step": 46420
    },
    {
      "epoch": 0.472259521484375,
      "grad_norm": 11.872044563293457,
      "learning_rate": 2.7218018596800167e-06,
      "loss": 3.1948,
      "step": 46425
    },
    {
      "epoch": 0.4723103841145833,
      "grad_norm": 10.191203117370605,
      "learning_rate": 2.7214035528156572e-06,
      "loss": 3.1356,
      "step": 46430
    },
    {
      "epoch": 0.4723612467447917,
      "grad_norm": 11.7601318359375,
      "learning_rate": 2.721005240286742e-06,
      "loss": 3.1688,
      "step": 46435
    },
    {
      "epoch": 0.472412109375,
      "grad_norm": 16.691404342651367,
      "learning_rate": 2.7206069221034597e-06,
      "loss": 3.407,
      "step": 46440
    },
    {
      "epoch": 0.4724629720052083,
      "grad_norm": 16.61736488342285,
      "learning_rate": 2.720208598276001e-06,
      "loss": 3.5574,
      "step": 46445
    },
    {
      "epoch": 0.4725138346354167,
      "grad_norm": 11.950624465942383,
      "learning_rate": 2.7198102688145583e-06,
      "loss": 3.7005,
      "step": 46450
    },
    {
      "epoch": 0.472564697265625,
      "grad_norm": 15.381657600402832,
      "learning_rate": 2.7194119337293214e-06,
      "loss": 3.3024,
      "step": 46455
    },
    {
      "epoch": 0.4726155598958333,
      "grad_norm": 8.248703956604004,
      "learning_rate": 2.7190135930304824e-06,
      "loss": 3.1889,
      "step": 46460
    },
    {
      "epoch": 0.4726664225260417,
      "grad_norm": 11.456612586975098,
      "learning_rate": 2.7186152467282325e-06,
      "loss": 3.4829,
      "step": 46465
    },
    {
      "epoch": 0.47271728515625,
      "grad_norm": 12.047392845153809,
      "learning_rate": 2.718216894832763e-06,
      "loss": 3.6619,
      "step": 46470
    },
    {
      "epoch": 0.4727681477864583,
      "grad_norm": 13.88084888458252,
      "learning_rate": 2.7178185373542663e-06,
      "loss": 3.1162,
      "step": 46475
    },
    {
      "epoch": 0.4728190104166667,
      "grad_norm": 10.259420394897461,
      "learning_rate": 2.7174201743029343e-06,
      "loss": 3.4904,
      "step": 46480
    },
    {
      "epoch": 0.472869873046875,
      "grad_norm": 16.462200164794922,
      "learning_rate": 2.717021805688958e-06,
      "loss": 3.3561,
      "step": 46485
    },
    {
      "epoch": 0.4729207356770833,
      "grad_norm": 14.48259162902832,
      "learning_rate": 2.7166234315225307e-06,
      "loss": 3.4951,
      "step": 46490
    },
    {
      "epoch": 0.4729715983072917,
      "grad_norm": 13.49642276763916,
      "learning_rate": 2.716225051813844e-06,
      "loss": 3.3612,
      "step": 46495
    },
    {
      "epoch": 0.4730224609375,
      "grad_norm": 12.33002758026123,
      "learning_rate": 2.715826666573091e-06,
      "loss": 3.3729,
      "step": 46500
    },
    {
      "epoch": 0.4730733235677083,
      "grad_norm": 14.000799179077148,
      "learning_rate": 2.715428275810463e-06,
      "loss": 3.4473,
      "step": 46505
    },
    {
      "epoch": 0.4731241861979167,
      "grad_norm": 7.098884105682373,
      "learning_rate": 2.7150298795361544e-06,
      "loss": 3.2642,
      "step": 46510
    },
    {
      "epoch": 0.473175048828125,
      "grad_norm": 10.298365592956543,
      "learning_rate": 2.7146314777603573e-06,
      "loss": 3.6669,
      "step": 46515
    },
    {
      "epoch": 0.4732259114583333,
      "grad_norm": 8.720600128173828,
      "learning_rate": 2.7142330704932647e-06,
      "loss": 3.2124,
      "step": 46520
    },
    {
      "epoch": 0.4732767740885417,
      "grad_norm": 12.410454750061035,
      "learning_rate": 2.7138346577450692e-06,
      "loss": 3.5289,
      "step": 46525
    },
    {
      "epoch": 0.47332763671875,
      "grad_norm": 10.719010353088379,
      "learning_rate": 2.7134362395259655e-06,
      "loss": 2.9803,
      "step": 46530
    },
    {
      "epoch": 0.4733784993489583,
      "grad_norm": 9.425874710083008,
      "learning_rate": 2.713037815846145e-06,
      "loss": 3.6764,
      "step": 46535
    },
    {
      "epoch": 0.4734293619791667,
      "grad_norm": 8.810243606567383,
      "learning_rate": 2.7126393867158034e-06,
      "loss": 3.5622,
      "step": 46540
    },
    {
      "epoch": 0.473480224609375,
      "grad_norm": 11.177980422973633,
      "learning_rate": 2.7122409521451344e-06,
      "loss": 3.3716,
      "step": 46545
    },
    {
      "epoch": 0.4735310872395833,
      "grad_norm": 12.672170639038086,
      "learning_rate": 2.7118425121443286e-06,
      "loss": 3.3576,
      "step": 46550
    },
    {
      "epoch": 0.4735819498697917,
      "grad_norm": 9.455219268798828,
      "learning_rate": 2.7114440667235842e-06,
      "loss": 3.1201,
      "step": 46555
    },
    {
      "epoch": 0.4736328125,
      "grad_norm": 13.621195793151855,
      "learning_rate": 2.7110456158930924e-06,
      "loss": 3.1756,
      "step": 46560
    },
    {
      "epoch": 0.4736836751302083,
      "grad_norm": 13.35601806640625,
      "learning_rate": 2.7106471596630498e-06,
      "loss": 3.4384,
      "step": 46565
    },
    {
      "epoch": 0.4737345377604167,
      "grad_norm": 12.736628532409668,
      "learning_rate": 2.7102486980436486e-06,
      "loss": 3.0624,
      "step": 46570
    },
    {
      "epoch": 0.473785400390625,
      "grad_norm": 12.159968376159668,
      "learning_rate": 2.7098502310450843e-06,
      "loss": 3.013,
      "step": 46575
    },
    {
      "epoch": 0.4738362630208333,
      "grad_norm": 13.247224807739258,
      "learning_rate": 2.709451758677552e-06,
      "loss": 3.506,
      "step": 46580
    },
    {
      "epoch": 0.4738871256510417,
      "grad_norm": 13.124889373779297,
      "learning_rate": 2.7090532809512455e-06,
      "loss": 2.9816,
      "step": 46585
    },
    {
      "epoch": 0.47393798828125,
      "grad_norm": 12.223262786865234,
      "learning_rate": 2.7086547978763604e-06,
      "loss": 3.3751,
      "step": 46590
    },
    {
      "epoch": 0.4739888509114583,
      "grad_norm": 11.755989074707031,
      "learning_rate": 2.708256309463093e-06,
      "loss": 3.5024,
      "step": 46595
    },
    {
      "epoch": 0.4740397135416667,
      "grad_norm": 8.518555641174316,
      "learning_rate": 2.7078578157216356e-06,
      "loss": 3.2307,
      "step": 46600
    },
    {
      "epoch": 0.474090576171875,
      "grad_norm": 11.394265174865723,
      "learning_rate": 2.7074593166621864e-06,
      "loss": 3.1565,
      "step": 46605
    },
    {
      "epoch": 0.4741414388020833,
      "grad_norm": 12.099921226501465,
      "learning_rate": 2.7070608122949394e-06,
      "loss": 3.2933,
      "step": 46610
    },
    {
      "epoch": 0.4741923014322917,
      "grad_norm": 14.934525489807129,
      "learning_rate": 2.706662302630091e-06,
      "loss": 3.0822,
      "step": 46615
    },
    {
      "epoch": 0.4742431640625,
      "grad_norm": 6.391138553619385,
      "learning_rate": 2.706263787677836e-06,
      "loss": 3.2745,
      "step": 46620
    },
    {
      "epoch": 0.4742940266927083,
      "grad_norm": 10.021805763244629,
      "learning_rate": 2.705865267448371e-06,
      "loss": 3.5125,
      "step": 46625
    },
    {
      "epoch": 0.4743448893229167,
      "grad_norm": 15.852633476257324,
      "learning_rate": 2.7054667419518915e-06,
      "loss": 3.794,
      "step": 46630
    },
    {
      "epoch": 0.474395751953125,
      "grad_norm": 15.297828674316406,
      "learning_rate": 2.705068211198595e-06,
      "loss": 3.2434,
      "step": 46635
    },
    {
      "epoch": 0.4744466145833333,
      "grad_norm": 12.144735336303711,
      "learning_rate": 2.704669675198676e-06,
      "loss": 3.2596,
      "step": 46640
    },
    {
      "epoch": 0.4744974772135417,
      "grad_norm": 10.186820030212402,
      "learning_rate": 2.7042711339623322e-06,
      "loss": 3.4567,
      "step": 46645
    },
    {
      "epoch": 0.47454833984375,
      "grad_norm": 10.479610443115234,
      "learning_rate": 2.7038725874997603e-06,
      "loss": 3.6553,
      "step": 46650
    },
    {
      "epoch": 0.4745992024739583,
      "grad_norm": 8.197158813476562,
      "learning_rate": 2.703474035821156e-06,
      "loss": 3.331,
      "step": 46655
    },
    {
      "epoch": 0.4746500651041667,
      "grad_norm": 13.642806053161621,
      "learning_rate": 2.7030754789367174e-06,
      "loss": 3.8816,
      "step": 46660
    },
    {
      "epoch": 0.474700927734375,
      "grad_norm": 13.100593566894531,
      "learning_rate": 2.7026769168566397e-06,
      "loss": 3.3208,
      "step": 46665
    },
    {
      "epoch": 0.4747517903645833,
      "grad_norm": 13.655210494995117,
      "learning_rate": 2.7022783495911228e-06,
      "loss": 2.952,
      "step": 46670
    },
    {
      "epoch": 0.4748026529947917,
      "grad_norm": 12.086578369140625,
      "learning_rate": 2.7018797771503608e-06,
      "loss": 3.1542,
      "step": 46675
    },
    {
      "epoch": 0.474853515625,
      "grad_norm": 11.18812370300293,
      "learning_rate": 2.7014811995445532e-06,
      "loss": 3.2304,
      "step": 46680
    },
    {
      "epoch": 0.4749043782552083,
      "grad_norm": 9.631187438964844,
      "learning_rate": 2.7010826167838967e-06,
      "loss": 3.2375,
      "step": 46685
    },
    {
      "epoch": 0.4749552408854167,
      "grad_norm": 13.716019630432129,
      "learning_rate": 2.700684028878589e-06,
      "loss": 3.3886,
      "step": 46690
    },
    {
      "epoch": 0.475006103515625,
      "grad_norm": 9.768837928771973,
      "learning_rate": 2.700285435838828e-06,
      "loss": 3.4435,
      "step": 46695
    },
    {
      "epoch": 0.4750569661458333,
      "grad_norm": 14.271130561828613,
      "learning_rate": 2.6998868376748123e-06,
      "loss": 3.3833,
      "step": 46700
    },
    {
      "epoch": 0.4751078287760417,
      "grad_norm": 17.306480407714844,
      "learning_rate": 2.6994882343967376e-06,
      "loss": 3.5262,
      "step": 46705
    },
    {
      "epoch": 0.47515869140625,
      "grad_norm": 11.069417953491211,
      "learning_rate": 2.6990896260148057e-06,
      "loss": 3.0509,
      "step": 46710
    },
    {
      "epoch": 0.4752095540364583,
      "grad_norm": 14.676597595214844,
      "learning_rate": 2.698691012539211e-06,
      "loss": 2.9301,
      "step": 46715
    },
    {
      "epoch": 0.4752604166666667,
      "grad_norm": 9.63838005065918,
      "learning_rate": 2.6982923939801555e-06,
      "loss": 3.1717,
      "step": 46720
    },
    {
      "epoch": 0.475311279296875,
      "grad_norm": 8.0972261428833,
      "learning_rate": 2.6978937703478354e-06,
      "loss": 3.1371,
      "step": 46725
    },
    {
      "epoch": 0.4753621419270833,
      "grad_norm": 10.335068702697754,
      "learning_rate": 2.6974951416524505e-06,
      "loss": 3.2931,
      "step": 46730
    },
    {
      "epoch": 0.4754130045572917,
      "grad_norm": 9.206419944763184,
      "learning_rate": 2.6970965079041985e-06,
      "loss": 3.309,
      "step": 46735
    },
    {
      "epoch": 0.4754638671875,
      "grad_norm": 14.041919708251953,
      "learning_rate": 2.696697869113279e-06,
      "loss": 3.267,
      "step": 46740
    },
    {
      "epoch": 0.4755147298177083,
      "grad_norm": 13.776942253112793,
      "learning_rate": 2.696299225289892e-06,
      "loss": 3.3834,
      "step": 46745
    },
    {
      "epoch": 0.4755655924479167,
      "grad_norm": 16.299365997314453,
      "learning_rate": 2.6959005764442348e-06,
      "loss": 3.2208,
      "step": 46750
    },
    {
      "epoch": 0.475616455078125,
      "grad_norm": 9.975279808044434,
      "learning_rate": 2.6955019225865083e-06,
      "loss": 3.4005,
      "step": 46755
    },
    {
      "epoch": 0.4756673177083333,
      "grad_norm": 17.89759063720703,
      "learning_rate": 2.6951032637269113e-06,
      "loss": 3.0012,
      "step": 46760
    },
    {
      "epoch": 0.4757181803385417,
      "grad_norm": 10.703070640563965,
      "learning_rate": 2.6947045998756445e-06,
      "loss": 3.1897,
      "step": 46765
    },
    {
      "epoch": 0.47576904296875,
      "grad_norm": 11.555328369140625,
      "learning_rate": 2.6943059310429054e-06,
      "loss": 3.7983,
      "step": 46770
    },
    {
      "epoch": 0.4758199055989583,
      "grad_norm": 9.706049919128418,
      "learning_rate": 2.6939072572388956e-06,
      "loss": 3.3247,
      "step": 46775
    },
    {
      "epoch": 0.4758707682291667,
      "grad_norm": 16.109375,
      "learning_rate": 2.6935085784738147e-06,
      "loss": 3.2998,
      "step": 46780
    },
    {
      "epoch": 0.475921630859375,
      "grad_norm": 14.335375785827637,
      "learning_rate": 2.6931098947578626e-06,
      "loss": 3.28,
      "step": 46785
    },
    {
      "epoch": 0.4759724934895833,
      "grad_norm": 14.95987606048584,
      "learning_rate": 2.6927112061012393e-06,
      "loss": 3.2788,
      "step": 46790
    },
    {
      "epoch": 0.4760233561197917,
      "grad_norm": 12.191413879394531,
      "learning_rate": 2.6923125125141457e-06,
      "loss": 3.3106,
      "step": 46795
    },
    {
      "epoch": 0.47607421875,
      "grad_norm": 14.423506736755371,
      "learning_rate": 2.6919138140067826e-06,
      "loss": 3.478,
      "step": 46800
    },
    {
      "epoch": 0.4761250813802083,
      "grad_norm": 14.733509063720703,
      "learning_rate": 2.6915151105893494e-06,
      "loss": 3.2981,
      "step": 46805
    },
    {
      "epoch": 0.4761759440104167,
      "grad_norm": 13.273368835449219,
      "learning_rate": 2.6911164022720474e-06,
      "loss": 3.504,
      "step": 46810
    },
    {
      "epoch": 0.476226806640625,
      "grad_norm": 13.908172607421875,
      "learning_rate": 2.6907176890650784e-06,
      "loss": 3.4394,
      "step": 46815
    },
    {
      "epoch": 0.4762776692708333,
      "grad_norm": 11.612787246704102,
      "learning_rate": 2.6903189709786414e-06,
      "loss": 3.3358,
      "step": 46820
    },
    {
      "epoch": 0.4763285319010417,
      "grad_norm": 10.148160934448242,
      "learning_rate": 2.68992024802294e-06,
      "loss": 3.5282,
      "step": 46825
    },
    {
      "epoch": 0.47637939453125,
      "grad_norm": 12.438175201416016,
      "learning_rate": 2.6895215202081727e-06,
      "loss": 3.3597,
      "step": 46830
    },
    {
      "epoch": 0.4764302571614583,
      "grad_norm": 11.603988647460938,
      "learning_rate": 2.689122787544543e-06,
      "loss": 3.1457,
      "step": 46835
    },
    {
      "epoch": 0.4764811197916667,
      "grad_norm": 15.418655395507812,
      "learning_rate": 2.6887240500422513e-06,
      "loss": 3.7434,
      "step": 46840
    },
    {
      "epoch": 0.476531982421875,
      "grad_norm": 14.989448547363281,
      "learning_rate": 2.6883253077114994e-06,
      "loss": 3.2839,
      "step": 46845
    },
    {
      "epoch": 0.4765828450520833,
      "grad_norm": 13.838479995727539,
      "learning_rate": 2.6879265605624895e-06,
      "loss": 3.3141,
      "step": 46850
    },
    {
      "epoch": 0.4766337076822917,
      "grad_norm": 11.194694519042969,
      "learning_rate": 2.6875278086054225e-06,
      "loss": 3.299,
      "step": 46855
    },
    {
      "epoch": 0.4766845703125,
      "grad_norm": 16.84820556640625,
      "learning_rate": 2.6871290518505017e-06,
      "loss": 3.3779,
      "step": 46860
    },
    {
      "epoch": 0.4767354329427083,
      "grad_norm": 13.52037239074707,
      "learning_rate": 2.686730290307928e-06,
      "loss": 3.0612,
      "step": 46865
    },
    {
      "epoch": 0.4767862955729167,
      "grad_norm": 19.260265350341797,
      "learning_rate": 2.686331523987904e-06,
      "loss": 3.7525,
      "step": 46870
    },
    {
      "epoch": 0.476837158203125,
      "grad_norm": 8.696489334106445,
      "learning_rate": 2.6859327529006326e-06,
      "loss": 3.2166,
      "step": 46875
    },
    {
      "epoch": 0.4768880208333333,
      "grad_norm": 8.214338302612305,
      "learning_rate": 2.685533977056316e-06,
      "loss": 3.0088,
      "step": 46880
    },
    {
      "epoch": 0.4769388834635417,
      "grad_norm": 13.720746040344238,
      "learning_rate": 2.685135196465155e-06,
      "loss": 3.4988,
      "step": 46885
    },
    {
      "epoch": 0.47698974609375,
      "grad_norm": 12.400075912475586,
      "learning_rate": 2.6847364111373557e-06,
      "loss": 2.9571,
      "step": 46890
    },
    {
      "epoch": 0.4770406087239583,
      "grad_norm": 13.83202838897705,
      "learning_rate": 2.6843376210831184e-06,
      "loss": 3.375,
      "step": 46895
    },
    {
      "epoch": 0.4770914713541667,
      "grad_norm": 12.201261520385742,
      "learning_rate": 2.683938826312647e-06,
      "loss": 3.3864,
      "step": 46900
    },
    {
      "epoch": 0.477142333984375,
      "grad_norm": 8.518744468688965,
      "learning_rate": 2.683540026836144e-06,
      "loss": 3.5712,
      "step": 46905
    },
    {
      "epoch": 0.4771931966145833,
      "grad_norm": 8.686223030090332,
      "learning_rate": 2.6831412226638125e-06,
      "loss": 3.9282,
      "step": 46910
    },
    {
      "epoch": 0.4772440592447917,
      "grad_norm": 11.989502906799316,
      "learning_rate": 2.682742413805857e-06,
      "loss": 3.6695,
      "step": 46915
    },
    {
      "epoch": 0.477294921875,
      "grad_norm": 11.652146339416504,
      "learning_rate": 2.6823436002724802e-06,
      "loss": 3.4997,
      "step": 46920
    },
    {
      "epoch": 0.4773457845052083,
      "grad_norm": 10.252928733825684,
      "learning_rate": 2.6819447820738847e-06,
      "loss": 3.2355,
      "step": 46925
    },
    {
      "epoch": 0.4773966471354167,
      "grad_norm": 13.859277725219727,
      "learning_rate": 2.6815459592202765e-06,
      "loss": 3.2546,
      "step": 46930
    },
    {
      "epoch": 0.477447509765625,
      "grad_norm": 15.216449737548828,
      "learning_rate": 2.6811471317218567e-06,
      "loss": 3.3368,
      "step": 46935
    },
    {
      "epoch": 0.4774983723958333,
      "grad_norm": 9.995338439941406,
      "learning_rate": 2.6807482995888306e-06,
      "loss": 3.0775,
      "step": 46940
    },
    {
      "epoch": 0.4775492350260417,
      "grad_norm": 14.418883323669434,
      "learning_rate": 2.680349462831403e-06,
      "loss": 3.3683,
      "step": 46945
    },
    {
      "epoch": 0.47760009765625,
      "grad_norm": 14.491920471191406,
      "learning_rate": 2.679950621459776e-06,
      "loss": 3.5651,
      "step": 46950
    },
    {
      "epoch": 0.4776509602864583,
      "grad_norm": 9.108541488647461,
      "learning_rate": 2.6795517754841562e-06,
      "loss": 3.2236,
      "step": 46955
    },
    {
      "epoch": 0.4777018229166667,
      "grad_norm": 12.77468204498291,
      "learning_rate": 2.6791529249147456e-06,
      "loss": 3.0829,
      "step": 46960
    },
    {
      "epoch": 0.477752685546875,
      "grad_norm": 13.290841102600098,
      "learning_rate": 2.6787540697617505e-06,
      "loss": 3.1515,
      "step": 46965
    },
    {
      "epoch": 0.4778035481770833,
      "grad_norm": 13.552038192749023,
      "learning_rate": 2.678355210035375e-06,
      "loss": 3.3246,
      "step": 46970
    },
    {
      "epoch": 0.4778544108072917,
      "grad_norm": 11.967382431030273,
      "learning_rate": 2.677956345745823e-06,
      "loss": 3.3275,
      "step": 46975
    },
    {
      "epoch": 0.4779052734375,
      "grad_norm": 11.457630157470703,
      "learning_rate": 2.677557476903301e-06,
      "loss": 3.331,
      "step": 46980
    },
    {
      "epoch": 0.4779561360677083,
      "grad_norm": 10.440465927124023,
      "learning_rate": 2.677158603518013e-06,
      "loss": 3.2845,
      "step": 46985
    },
    {
      "epoch": 0.4780069986979167,
      "grad_norm": 10.605171203613281,
      "learning_rate": 2.6767597256001633e-06,
      "loss": 3.3102,
      "step": 46990
    },
    {
      "epoch": 0.478057861328125,
      "grad_norm": 8.721869468688965,
      "learning_rate": 2.6763608431599586e-06,
      "loss": 2.8082,
      "step": 46995
    },
    {
      "epoch": 0.4781087239583333,
      "grad_norm": 9.492325782775879,
      "learning_rate": 2.6759619562076035e-06,
      "loss": 3.2523,
      "step": 47000
    },
    {
      "epoch": 0.4781595865885417,
      "grad_norm": 15.880738258361816,
      "learning_rate": 2.6755630647533034e-06,
      "loss": 3.6552,
      "step": 47005
    },
    {
      "epoch": 0.47821044921875,
      "grad_norm": 12.2164306640625,
      "learning_rate": 2.6751641688072633e-06,
      "loss": 3.652,
      "step": 47010
    },
    {
      "epoch": 0.4782613118489583,
      "grad_norm": 14.72760009765625,
      "learning_rate": 2.6747652683796897e-06,
      "loss": 3.2615,
      "step": 47015
    },
    {
      "epoch": 0.4783121744791667,
      "grad_norm": 11.249184608459473,
      "learning_rate": 2.674366363480789e-06,
      "loss": 3.4039,
      "step": 47020
    },
    {
      "epoch": 0.478363037109375,
      "grad_norm": 21.686203002929688,
      "learning_rate": 2.673967454120765e-06,
      "loss": 3.4197,
      "step": 47025
    },
    {
      "epoch": 0.4784138997395833,
      "grad_norm": 15.506630897521973,
      "learning_rate": 2.6735685403098255e-06,
      "loss": 3.3828,
      "step": 47030
    },
    {
      "epoch": 0.4784647623697917,
      "grad_norm": 14.081663131713867,
      "learning_rate": 2.673169622058176e-06,
      "loss": 3.1585,
      "step": 47035
    },
    {
      "epoch": 0.478515625,
      "grad_norm": 12.858463287353516,
      "learning_rate": 2.6727706993760227e-06,
      "loss": 3.1958,
      "step": 47040
    },
    {
      "epoch": 0.4785664876302083,
      "grad_norm": 10.969271659851074,
      "learning_rate": 2.672371772273572e-06,
      "loss": 3.5695,
      "step": 47045
    },
    {
      "epoch": 0.4786173502604167,
      "grad_norm": 11.234213829040527,
      "learning_rate": 2.6719728407610313e-06,
      "loss": 3.6693,
      "step": 47050
    },
    {
      "epoch": 0.478668212890625,
      "grad_norm": 13.769668579101562,
      "learning_rate": 2.6715739048486044e-06,
      "loss": 3.1101,
      "step": 47055
    },
    {
      "epoch": 0.4787190755208333,
      "grad_norm": 14.294875144958496,
      "learning_rate": 2.671174964546502e-06,
      "loss": 3.248,
      "step": 47060
    },
    {
      "epoch": 0.4787699381510417,
      "grad_norm": 13.281566619873047,
      "learning_rate": 2.6707760198649268e-06,
      "loss": 3.5763,
      "step": 47065
    },
    {
      "epoch": 0.47882080078125,
      "grad_norm": 14.868240356445312,
      "learning_rate": 2.670377070814089e-06,
      "loss": 3.3147,
      "step": 47070
    },
    {
      "epoch": 0.4788716634114583,
      "grad_norm": 9.297042846679688,
      "learning_rate": 2.6699781174041935e-06,
      "loss": 2.9528,
      "step": 47075
    },
    {
      "epoch": 0.4789225260416667,
      "grad_norm": 10.145970344543457,
      "learning_rate": 2.6695791596454477e-06,
      "loss": 3.4101,
      "step": 47080
    },
    {
      "epoch": 0.478973388671875,
      "grad_norm": 8.714816093444824,
      "learning_rate": 2.66918019754806e-06,
      "loss": 3.1934,
      "step": 47085
    },
    {
      "epoch": 0.4790242513020833,
      "grad_norm": 12.525447845458984,
      "learning_rate": 2.668781231122237e-06,
      "loss": 3.3849,
      "step": 47090
    },
    {
      "epoch": 0.4790751139322917,
      "grad_norm": 15.579602241516113,
      "learning_rate": 2.6683822603781856e-06,
      "loss": 3.3557,
      "step": 47095
    },
    {
      "epoch": 0.4791259765625,
      "grad_norm": 12.996421813964844,
      "learning_rate": 2.6679832853261153e-06,
      "loss": 3.3016,
      "step": 47100
    },
    {
      "epoch": 0.4791768391927083,
      "grad_norm": 15.442605018615723,
      "learning_rate": 2.6675843059762307e-06,
      "loss": 3.3641,
      "step": 47105
    },
    {
      "epoch": 0.4792277018229167,
      "grad_norm": 14.138955116271973,
      "learning_rate": 2.667185322338743e-06,
      "loss": 3.3516,
      "step": 47110
    },
    {
      "epoch": 0.479278564453125,
      "grad_norm": 12.985639572143555,
      "learning_rate": 2.6667863344238576e-06,
      "loss": 3.4175,
      "step": 47115
    },
    {
      "epoch": 0.4793294270833333,
      "grad_norm": 13.8716402053833,
      "learning_rate": 2.666387342241783e-06,
      "loss": 3.1962,
      "step": 47120
    },
    {
      "epoch": 0.4793802897135417,
      "grad_norm": 10.476211547851562,
      "learning_rate": 2.6659883458027284e-06,
      "loss": 3.3406,
      "step": 47125
    },
    {
      "epoch": 0.47943115234375,
      "grad_norm": 16.908771514892578,
      "learning_rate": 2.6655893451169013e-06,
      "loss": 3.4647,
      "step": 47130
    },
    {
      "epoch": 0.4794820149739583,
      "grad_norm": 9.55135440826416,
      "learning_rate": 2.6651903401945097e-06,
      "loss": 3.6185,
      "step": 47135
    },
    {
      "epoch": 0.4795328776041667,
      "grad_norm": 13.74426555633545,
      "learning_rate": 2.6647913310457625e-06,
      "loss": 3.5157,
      "step": 47140
    },
    {
      "epoch": 0.479583740234375,
      "grad_norm": 10.291647911071777,
      "learning_rate": 2.6643923176808678e-06,
      "loss": 3.2924,
      "step": 47145
    },
    {
      "epoch": 0.4796346028645833,
      "grad_norm": 10.82001781463623,
      "learning_rate": 2.6639933001100347e-06,
      "loss": 4.0843,
      "step": 47150
    },
    {
      "epoch": 0.4796854654947917,
      "grad_norm": 10.689411163330078,
      "learning_rate": 2.663594278343473e-06,
      "loss": 3.1606,
      "step": 47155
    },
    {
      "epoch": 0.479736328125,
      "grad_norm": 12.379398345947266,
      "learning_rate": 2.663195252391389e-06,
      "loss": 3.0887,
      "step": 47160
    },
    {
      "epoch": 0.4797871907552083,
      "grad_norm": 13.633697509765625,
      "learning_rate": 2.6627962222639945e-06,
      "loss": 3.2125,
      "step": 47165
    },
    {
      "epoch": 0.4798380533854167,
      "grad_norm": 14.551989555358887,
      "learning_rate": 2.662397187971496e-06,
      "loss": 3.0531,
      "step": 47170
    },
    {
      "epoch": 0.479888916015625,
      "grad_norm": 11.615093231201172,
      "learning_rate": 2.6619981495241047e-06,
      "loss": 3.2043,
      "step": 47175
    },
    {
      "epoch": 0.4799397786458333,
      "grad_norm": 8.46528434753418,
      "learning_rate": 2.661599106932029e-06,
      "loss": 3.3052,
      "step": 47180
    },
    {
      "epoch": 0.4799906412760417,
      "grad_norm": 13.535140037536621,
      "learning_rate": 2.661200060205478e-06,
      "loss": 3.2812,
      "step": 47185
    },
    {
      "epoch": 0.48004150390625,
      "grad_norm": 9.564166069030762,
      "learning_rate": 2.6608010093546626e-06,
      "loss": 3.0153,
      "step": 47190
    },
    {
      "epoch": 0.4800923665364583,
      "grad_norm": 13.626891136169434,
      "learning_rate": 2.6604019543897907e-06,
      "loss": 3.2109,
      "step": 47195
    },
    {
      "epoch": 0.4801432291666667,
      "grad_norm": 12.772491455078125,
      "learning_rate": 2.6600028953210733e-06,
      "loss": 3.2223,
      "step": 47200
    },
    {
      "epoch": 0.480194091796875,
      "grad_norm": 9.005960464477539,
      "learning_rate": 2.65960383215872e-06,
      "loss": 3.5045,
      "step": 47205
    },
    {
      "epoch": 0.4802449544270833,
      "grad_norm": 8.408181190490723,
      "learning_rate": 2.6592047649129395e-06,
      "loss": 3.6569,
      "step": 47210
    },
    {
      "epoch": 0.4802958170572917,
      "grad_norm": 13.204599380493164,
      "learning_rate": 2.6588056935939434e-06,
      "loss": 3.4019,
      "step": 47215
    },
    {
      "epoch": 0.4803466796875,
      "grad_norm": 7.732039451599121,
      "learning_rate": 2.658406618211941e-06,
      "loss": 3.2256,
      "step": 47220
    },
    {
      "epoch": 0.4803975423177083,
      "grad_norm": 8.743809700012207,
      "learning_rate": 2.6580075387771432e-06,
      "loss": 3.6135,
      "step": 47225
    },
    {
      "epoch": 0.4804484049479167,
      "grad_norm": 14.294320106506348,
      "learning_rate": 2.65760845529976e-06,
      "loss": 3.4267,
      "step": 47230
    },
    {
      "epoch": 0.480499267578125,
      "grad_norm": 13.660139083862305,
      "learning_rate": 2.6572093677900013e-06,
      "loss": 3.468,
      "step": 47235
    },
    {
      "epoch": 0.4805501302083333,
      "grad_norm": 13.269354820251465,
      "learning_rate": 2.6568102762580784e-06,
      "loss": 3.2078,
      "step": 47240
    },
    {
      "epoch": 0.4806009928385417,
      "grad_norm": 11.7168550491333,
      "learning_rate": 2.6564111807142017e-06,
      "loss": 3.388,
      "step": 47245
    },
    {
      "epoch": 0.48065185546875,
      "grad_norm": 9.902116775512695,
      "learning_rate": 2.6560120811685826e-06,
      "loss": 3.5782,
      "step": 47250
    },
    {
      "epoch": 0.4807027180989583,
      "grad_norm": 8.437605857849121,
      "learning_rate": 2.6556129776314304e-06,
      "loss": 3.279,
      "step": 47255
    },
    {
      "epoch": 0.4807535807291667,
      "grad_norm": 14.715630531311035,
      "learning_rate": 2.6552138701129575e-06,
      "loss": 2.9661,
      "step": 47260
    },
    {
      "epoch": 0.480804443359375,
      "grad_norm": 16.577116012573242,
      "learning_rate": 2.6548147586233747e-06,
      "loss": 3.4551,
      "step": 47265
    },
    {
      "epoch": 0.4808553059895833,
      "grad_norm": 10.214470863342285,
      "learning_rate": 2.6544156431728934e-06,
      "loss": 3.3104,
      "step": 47270
    },
    {
      "epoch": 0.4809061686197917,
      "grad_norm": 13.728621482849121,
      "learning_rate": 2.6540165237717235e-06,
      "loss": 3.2211,
      "step": 47275
    },
    {
      "epoch": 0.48095703125,
      "grad_norm": 10.931455612182617,
      "learning_rate": 2.6536174004300783e-06,
      "loss": 2.9952,
      "step": 47280
    },
    {
      "epoch": 0.4810078938802083,
      "grad_norm": 12.8707275390625,
      "learning_rate": 2.653218273158168e-06,
      "loss": 3.3947,
      "step": 47285
    },
    {
      "epoch": 0.4810587565104167,
      "grad_norm": 11.696076393127441,
      "learning_rate": 2.6528191419662043e-06,
      "loss": 3.3118,
      "step": 47290
    },
    {
      "epoch": 0.481109619140625,
      "grad_norm": 11.292045593261719,
      "learning_rate": 2.652420006864399e-06,
      "loss": 3.5174,
      "step": 47295
    },
    {
      "epoch": 0.4811604817708333,
      "grad_norm": 11.5274658203125,
      "learning_rate": 2.6520208678629637e-06,
      "loss": 3.4944,
      "step": 47300
    },
    {
      "epoch": 0.4812113444010417,
      "grad_norm": 13.709296226501465,
      "learning_rate": 2.651621724972111e-06,
      "loss": 3.5253,
      "step": 47305
    },
    {
      "epoch": 0.48126220703125,
      "grad_norm": 13.923632621765137,
      "learning_rate": 2.651222578202053e-06,
      "loss": 3.2138,
      "step": 47310
    },
    {
      "epoch": 0.4813130696614583,
      "grad_norm": 13.395757675170898,
      "learning_rate": 2.6508234275629996e-06,
      "loss": 3.4306,
      "step": 47315
    },
    {
      "epoch": 0.4813639322916667,
      "grad_norm": 14.269627571105957,
      "learning_rate": 2.6504242730651664e-06,
      "loss": 3.2937,
      "step": 47320
    },
    {
      "epoch": 0.481414794921875,
      "grad_norm": 11.33421516418457,
      "learning_rate": 2.650025114718763e-06,
      "loss": 2.8693,
      "step": 47325
    },
    {
      "epoch": 0.4814656575520833,
      "grad_norm": 12.996232032775879,
      "learning_rate": 2.649625952534001e-06,
      "loss": 3.3136,
      "step": 47330
    },
    {
      "epoch": 0.4815165201822917,
      "grad_norm": 12.771052360534668,
      "learning_rate": 2.6492267865210967e-06,
      "loss": 3.3747,
      "step": 47335
    },
    {
      "epoch": 0.4815673828125,
      "grad_norm": 11.761916160583496,
      "learning_rate": 2.648827616690259e-06,
      "loss": 3.9257,
      "step": 47340
    },
    {
      "epoch": 0.4816182454427083,
      "grad_norm": 7.735947608947754,
      "learning_rate": 2.648428443051703e-06,
      "loss": 3.2851,
      "step": 47345
    },
    {
      "epoch": 0.4816691080729167,
      "grad_norm": 17.554040908813477,
      "learning_rate": 2.64802926561564e-06,
      "loss": 3.3919,
      "step": 47350
    },
    {
      "epoch": 0.481719970703125,
      "grad_norm": 7.973572254180908,
      "learning_rate": 2.6476300843922844e-06,
      "loss": 3.1638,
      "step": 47355
    },
    {
      "epoch": 0.4817708333333333,
      "grad_norm": 9.76625919342041,
      "learning_rate": 2.6472308993918465e-06,
      "loss": 3.3234,
      "step": 47360
    },
    {
      "epoch": 0.4818216959635417,
      "grad_norm": 8.04384708404541,
      "learning_rate": 2.646831710624542e-06,
      "loss": 3.2401,
      "step": 47365
    },
    {
      "epoch": 0.48187255859375,
      "grad_norm": 11.144789695739746,
      "learning_rate": 2.6464325181005824e-06,
      "loss": 3.1819,
      "step": 47370
    },
    {
      "epoch": 0.4819234212239583,
      "grad_norm": 17.865840911865234,
      "learning_rate": 2.646033321830182e-06,
      "loss": 3.6869,
      "step": 47375
    },
    {
      "epoch": 0.4819742838541667,
      "grad_norm": 11.546830177307129,
      "learning_rate": 2.6456341218235525e-06,
      "loss": 3.2895,
      "step": 47380
    },
    {
      "epoch": 0.482025146484375,
      "grad_norm": 17.063457489013672,
      "learning_rate": 2.6452349180909103e-06,
      "loss": 3.4353,
      "step": 47385
    },
    {
      "epoch": 0.4820760091145833,
      "grad_norm": 13.253219604492188,
      "learning_rate": 2.644835710642466e-06,
      "loss": 3.2576,
      "step": 47390
    },
    {
      "epoch": 0.4821268717447917,
      "grad_norm": 9.14925765991211,
      "learning_rate": 2.6444364994884348e-06,
      "loss": 3.4164,
      "step": 47395
    },
    {
      "epoch": 0.482177734375,
      "grad_norm": 11.442546844482422,
      "learning_rate": 2.6440372846390293e-06,
      "loss": 3.2947,
      "step": 47400
    },
    {
      "epoch": 0.4822285970052083,
      "grad_norm": 12.624979019165039,
      "learning_rate": 2.6436380661044648e-06,
      "loss": 4.0104,
      "step": 47405
    },
    {
      "epoch": 0.4822794596354167,
      "grad_norm": 8.066694259643555,
      "learning_rate": 2.643238843894954e-06,
      "loss": 2.85,
      "step": 47410
    },
    {
      "epoch": 0.482330322265625,
      "grad_norm": 9.832465171813965,
      "learning_rate": 2.6428396180207115e-06,
      "loss": 3.0949,
      "step": 47415
    },
    {
      "epoch": 0.4823811848958333,
      "grad_norm": 12.458390235900879,
      "learning_rate": 2.6424403884919515e-06,
      "loss": 3.596,
      "step": 47420
    },
    {
      "epoch": 0.4824320475260417,
      "grad_norm": 10.591827392578125,
      "learning_rate": 2.6420411553188878e-06,
      "loss": 3.4678,
      "step": 47425
    },
    {
      "epoch": 0.48248291015625,
      "grad_norm": 10.282755851745605,
      "learning_rate": 2.641641918511734e-06,
      "loss": 3.4219,
      "step": 47430
    },
    {
      "epoch": 0.4825337727864583,
      "grad_norm": 10.006214141845703,
      "learning_rate": 2.641242678080706e-06,
      "loss": 3.3134,
      "step": 47435
    },
    {
      "epoch": 0.4825846354166667,
      "grad_norm": 14.434442520141602,
      "learning_rate": 2.640843434036018e-06,
      "loss": 3.4595,
      "step": 47440
    },
    {
      "epoch": 0.482635498046875,
      "grad_norm": 8.330063819885254,
      "learning_rate": 2.6404441863878832e-06,
      "loss": 3.4476,
      "step": 47445
    },
    {
      "epoch": 0.4826863606770833,
      "grad_norm": 21.621471405029297,
      "learning_rate": 2.640044935146518e-06,
      "loss": 3.7481,
      "step": 47450
    },
    {
      "epoch": 0.4827372233072917,
      "grad_norm": 15.059220314025879,
      "learning_rate": 2.639645680322136e-06,
      "loss": 3.0038,
      "step": 47455
    },
    {
      "epoch": 0.4827880859375,
      "grad_norm": 13.724431037902832,
      "learning_rate": 2.639246421924952e-06,
      "loss": 3.3143,
      "step": 47460
    },
    {
      "epoch": 0.4828389485677083,
      "grad_norm": 15.076763153076172,
      "learning_rate": 2.6388471599651815e-06,
      "loss": 3.2902,
      "step": 47465
    },
    {
      "epoch": 0.4828898111979167,
      "grad_norm": 6.995392322540283,
      "learning_rate": 2.6384478944530397e-06,
      "loss": 3.4319,
      "step": 47470
    },
    {
      "epoch": 0.482940673828125,
      "grad_norm": 12.359930992126465,
      "learning_rate": 2.638048625398741e-06,
      "loss": 3.4122,
      "step": 47475
    },
    {
      "epoch": 0.4829915364583333,
      "grad_norm": 8.89985179901123,
      "learning_rate": 2.6376493528125014e-06,
      "loss": 3.0402,
      "step": 47480
    },
    {
      "epoch": 0.4830423990885417,
      "grad_norm": 15.408087730407715,
      "learning_rate": 2.6372500767045344e-06,
      "loss": 3.0548,
      "step": 47485
    },
    {
      "epoch": 0.48309326171875,
      "grad_norm": 12.19315242767334,
      "learning_rate": 2.636850797085058e-06,
      "loss": 3.2569,
      "step": 47490
    },
    {
      "epoch": 0.4831441243489583,
      "grad_norm": 14.964292526245117,
      "learning_rate": 2.636451513964286e-06,
      "loss": 3.6578,
      "step": 47495
    },
    {
      "epoch": 0.4831949869791667,
      "grad_norm": 15.417756080627441,
      "learning_rate": 2.636052227352435e-06,
      "loss": 3.4411,
      "step": 47500
    },
    {
      "epoch": 0.483245849609375,
      "grad_norm": 10.532073974609375,
      "learning_rate": 2.635652937259719e-06,
      "loss": 3.3885,
      "step": 47505
    },
    {
      "epoch": 0.4832967122395833,
      "grad_norm": 8.670329093933105,
      "learning_rate": 2.6352536436963548e-06,
      "loss": 3.315,
      "step": 47510
    },
    {
      "epoch": 0.4833475748697917,
      "grad_norm": 9.610486030578613,
      "learning_rate": 2.634854346672559e-06,
      "loss": 3.4857,
      "step": 47515
    },
    {
      "epoch": 0.4833984375,
      "grad_norm": 15.70473861694336,
      "learning_rate": 2.634455046198546e-06,
      "loss": 3.2049,
      "step": 47520
    },
    {
      "epoch": 0.4834493001302083,
      "grad_norm": 12.451311111450195,
      "learning_rate": 2.6340557422845324e-06,
      "loss": 3.6464,
      "step": 47525
    },
    {
      "epoch": 0.4835001627604167,
      "grad_norm": 13.600598335266113,
      "learning_rate": 2.633656434940735e-06,
      "loss": 3.6325,
      "step": 47530
    },
    {
      "epoch": 0.483551025390625,
      "grad_norm": 8.907392501831055,
      "learning_rate": 2.6332571241773693e-06,
      "loss": 3.3728,
      "step": 47535
    },
    {
      "epoch": 0.4836018880208333,
      "grad_norm": 15.609326362609863,
      "learning_rate": 2.6328578100046514e-06,
      "loss": 3.5422,
      "step": 47540
    },
    {
      "epoch": 0.4836527506510417,
      "grad_norm": 16.75021743774414,
      "learning_rate": 2.6324584924327982e-06,
      "loss": 3.0167,
      "step": 47545
    },
    {
      "epoch": 0.48370361328125,
      "grad_norm": 7.8447346687316895,
      "learning_rate": 2.6320591714720256e-06,
      "loss": 3.3293,
      "step": 47550
    },
    {
      "epoch": 0.4837544759114583,
      "grad_norm": 12.660308837890625,
      "learning_rate": 2.631659847132551e-06,
      "loss": 3.2761,
      "step": 47555
    },
    {
      "epoch": 0.4838053385416667,
      "grad_norm": 8.696551322937012,
      "learning_rate": 2.6312605194245894e-06,
      "loss": 3.1498,
      "step": 47560
    },
    {
      "epoch": 0.483856201171875,
      "grad_norm": 15.263006210327148,
      "learning_rate": 2.6308611883583592e-06,
      "loss": 3.5946,
      "step": 47565
    },
    {
      "epoch": 0.4839070638020833,
      "grad_norm": 9.13163948059082,
      "learning_rate": 2.6304618539440763e-06,
      "loss": 3.4685,
      "step": 47570
    },
    {
      "epoch": 0.4839579264322917,
      "grad_norm": 10.969802856445312,
      "learning_rate": 2.630062516191958e-06,
      "loss": 3.3894,
      "step": 47575
    },
    {
      "epoch": 0.4840087890625,
      "grad_norm": 8.642189025878906,
      "learning_rate": 2.6296631751122208e-06,
      "loss": 3.314,
      "step": 47580
    },
    {
      "epoch": 0.4840596516927083,
      "grad_norm": 15.151670455932617,
      "learning_rate": 2.629263830715082e-06,
      "loss": 2.9757,
      "step": 47585
    },
    {
      "epoch": 0.4841105143229167,
      "grad_norm": 11.499610900878906,
      "learning_rate": 2.628864483010759e-06,
      "loss": 3.4151,
      "step": 47590
    },
    {
      "epoch": 0.484161376953125,
      "grad_norm": 13.054522514343262,
      "learning_rate": 2.628465132009469e-06,
      "loss": 3.5306,
      "step": 47595
    },
    {
      "epoch": 0.4842122395833333,
      "grad_norm": 15.709760665893555,
      "learning_rate": 2.628065777721428e-06,
      "loss": 3.5197,
      "step": 47600
    },
    {
      "epoch": 0.4842631022135417,
      "grad_norm": 7.843846797943115,
      "learning_rate": 2.6276664201568553e-06,
      "loss": 3.3044,
      "step": 47605
    },
    {
      "epoch": 0.48431396484375,
      "grad_norm": 13.429792404174805,
      "learning_rate": 2.6272670593259674e-06,
      "loss": 3.3137,
      "step": 47610
    },
    {
      "epoch": 0.4843648274739583,
      "grad_norm": 12.689478874206543,
      "learning_rate": 2.6268676952389815e-06,
      "loss": 3.4891,
      "step": 47615
    },
    {
      "epoch": 0.4844156901041667,
      "grad_norm": 8.793680191040039,
      "learning_rate": 2.6264683279061166e-06,
      "loss": 3.6951,
      "step": 47620
    },
    {
      "epoch": 0.484466552734375,
      "grad_norm": 11.071285247802734,
      "learning_rate": 2.6260689573375885e-06,
      "loss": 3.2139,
      "step": 47625
    },
    {
      "epoch": 0.4845174153645833,
      "grad_norm": 8.690736770629883,
      "learning_rate": 2.6256695835436164e-06,
      "loss": 3.5408,
      "step": 47630
    },
    {
      "epoch": 0.4845682779947917,
      "grad_norm": 13.136496543884277,
      "learning_rate": 2.625270206534418e-06,
      "loss": 3.121,
      "step": 47635
    },
    {
      "epoch": 0.484619140625,
      "grad_norm": 10.826349258422852,
      "learning_rate": 2.6248708263202105e-06,
      "loss": 3.7452,
      "step": 47640
    },
    {
      "epoch": 0.4846700032552083,
      "grad_norm": 14.85303020477295,
      "learning_rate": 2.6244714429112126e-06,
      "loss": 3.6537,
      "step": 47645
    },
    {
      "epoch": 0.4847208658854167,
      "grad_norm": 15.568806648254395,
      "learning_rate": 2.624072056317643e-06,
      "loss": 3.1806,
      "step": 47650
    },
    {
      "epoch": 0.484771728515625,
      "grad_norm": 9.954267501831055,
      "learning_rate": 2.6236726665497186e-06,
      "loss": 3.5269,
      "step": 47655
    },
    {
      "epoch": 0.4848225911458333,
      "grad_norm": 12.706521034240723,
      "learning_rate": 2.6232732736176593e-06,
      "loss": 3.9319,
      "step": 47660
    },
    {
      "epoch": 0.4848734537760417,
      "grad_norm": 11.31127643585205,
      "learning_rate": 2.622873877531681e-06,
      "loss": 3.0772,
      "step": 47665
    },
    {
      "epoch": 0.48492431640625,
      "grad_norm": 14.69467830657959,
      "learning_rate": 2.622474478302005e-06,
      "loss": 3.0709,
      "step": 47670
    },
    {
      "epoch": 0.4849751790364583,
      "grad_norm": 13.351578712463379,
      "learning_rate": 2.622075075938848e-06,
      "loss": 3.3081,
      "step": 47675
    },
    {
      "epoch": 0.4850260416666667,
      "grad_norm": 12.570734024047852,
      "learning_rate": 2.6216756704524294e-06,
      "loss": 3.3593,
      "step": 47680
    },
    {
      "epoch": 0.485076904296875,
      "grad_norm": 14.045585632324219,
      "learning_rate": 2.621276261852967e-06,
      "loss": 3.3143,
      "step": 47685
    },
    {
      "epoch": 0.4851277669270833,
      "grad_norm": 13.33901596069336,
      "learning_rate": 2.6208768501506802e-06,
      "loss": 3.7336,
      "step": 47690
    },
    {
      "epoch": 0.4851786295572917,
      "grad_norm": 11.250321388244629,
      "learning_rate": 2.6204774353557883e-06,
      "loss": 3.3786,
      "step": 47695
    },
    {
      "epoch": 0.4852294921875,
      "grad_norm": 12.41751480102539,
      "learning_rate": 2.62007801747851e-06,
      "loss": 2.9969,
      "step": 47700
    },
    {
      "epoch": 0.4852803548177083,
      "grad_norm": 13.19685173034668,
      "learning_rate": 2.6196785965290635e-06,
      "loss": 3.379,
      "step": 47705
    },
    {
      "epoch": 0.4853312174479167,
      "grad_norm": 9.21330738067627,
      "learning_rate": 2.6192791725176687e-06,
      "loss": 3.3872,
      "step": 47710
    },
    {
      "epoch": 0.485382080078125,
      "grad_norm": 14.119096755981445,
      "learning_rate": 2.618879745454544e-06,
      "loss": 3.1882,
      "step": 47715
    },
    {
      "epoch": 0.4854329427083333,
      "grad_norm": 15.260636329650879,
      "learning_rate": 2.61848031534991e-06,
      "loss": 2.9966,
      "step": 47720
    },
    {
      "epoch": 0.4854838053385417,
      "grad_norm": 10.559808731079102,
      "learning_rate": 2.6180808822139854e-06,
      "loss": 3.0944,
      "step": 47725
    },
    {
      "epoch": 0.48553466796875,
      "grad_norm": 12.172982215881348,
      "learning_rate": 2.617681446056988e-06,
      "loss": 3.019,
      "step": 47730
    },
    {
      "epoch": 0.4855855305989583,
      "grad_norm": 14.875792503356934,
      "learning_rate": 2.6172820068891403e-06,
      "loss": 3.4954,
      "step": 47735
    },
    {
      "epoch": 0.4856363932291667,
      "grad_norm": 12.607369422912598,
      "learning_rate": 2.616882564720659e-06,
      "loss": 3.3797,
      "step": 47740
    },
    {
      "epoch": 0.485687255859375,
      "grad_norm": 14.492063522338867,
      "learning_rate": 2.6164831195617655e-06,
      "loss": 3.7695,
      "step": 47745
    },
    {
      "epoch": 0.4857381184895833,
      "grad_norm": 9.796468734741211,
      "learning_rate": 2.6160836714226796e-06,
      "loss": 3.4647,
      "step": 47750
    },
    {
      "epoch": 0.4857889811197917,
      "grad_norm": 11.6259765625,
      "learning_rate": 2.6156842203136196e-06,
      "loss": 3.1608,
      "step": 47755
    },
    {
      "epoch": 0.48583984375,
      "grad_norm": 8.976288795471191,
      "learning_rate": 2.6152847662448067e-06,
      "loss": 3.7266,
      "step": 47760
    },
    {
      "epoch": 0.4858907063802083,
      "grad_norm": 9.242552757263184,
      "learning_rate": 2.6148853092264607e-06,
      "loss": 3.4968,
      "step": 47765
    },
    {
      "epoch": 0.4859415690104167,
      "grad_norm": 9.405117988586426,
      "learning_rate": 2.6144858492688003e-06,
      "loss": 3.2472,
      "step": 47770
    },
    {
      "epoch": 0.485992431640625,
      "grad_norm": 13.579270362854004,
      "learning_rate": 2.6140863863820475e-06,
      "loss": 3.0759,
      "step": 47775
    },
    {
      "epoch": 0.4860432942708333,
      "grad_norm": 14.4361572265625,
      "learning_rate": 2.6136869205764216e-06,
      "loss": 3.3667,
      "step": 47780
    },
    {
      "epoch": 0.4860941569010417,
      "grad_norm": 13.49770736694336,
      "learning_rate": 2.6132874518621423e-06,
      "loss": 3.4313,
      "step": 47785
    },
    {
      "epoch": 0.48614501953125,
      "grad_norm": 12.228191375732422,
      "learning_rate": 2.61288798024943e-06,
      "loss": 3.0355,
      "step": 47790
    },
    {
      "epoch": 0.4861958821614583,
      "grad_norm": 9.5272216796875,
      "learning_rate": 2.612488505748506e-06,
      "loss": 3.2651,
      "step": 47795
    },
    {
      "epoch": 0.4862467447916667,
      "grad_norm": 12.149260520935059,
      "learning_rate": 2.6120890283695905e-06,
      "loss": 3.6675,
      "step": 47800
    },
    {
      "epoch": 0.486297607421875,
      "grad_norm": 10.929622650146484,
      "learning_rate": 2.6116895481229033e-06,
      "loss": 3.3229,
      "step": 47805
    },
    {
      "epoch": 0.4863484700520833,
      "grad_norm": 15.386078834533691,
      "learning_rate": 2.611290065018666e-06,
      "loss": 3.3882,
      "step": 47810
    },
    {
      "epoch": 0.4863993326822917,
      "grad_norm": 6.863919258117676,
      "learning_rate": 2.6108905790670987e-06,
      "loss": 3.9467,
      "step": 47815
    },
    {
      "epoch": 0.4864501953125,
      "grad_norm": 11.039124488830566,
      "learning_rate": 2.610491090278421e-06,
      "loss": 3.0379,
      "step": 47820
    },
    {
      "epoch": 0.4865010579427083,
      "grad_norm": 12.934539794921875,
      "learning_rate": 2.6100915986628562e-06,
      "loss": 3.4597,
      "step": 47825
    },
    {
      "epoch": 0.4865519205729167,
      "grad_norm": 13.209774017333984,
      "learning_rate": 2.609692104230624e-06,
      "loss": 3.5327,
      "step": 47830
    },
    {
      "epoch": 0.486602783203125,
      "grad_norm": 10.195297241210938,
      "learning_rate": 2.609292606991944e-06,
      "loss": 3.6324,
      "step": 47835
    },
    {
      "epoch": 0.4866536458333333,
      "grad_norm": 14.529973983764648,
      "learning_rate": 2.60889310695704e-06,
      "loss": 3.2091,
      "step": 47840
    },
    {
      "epoch": 0.4867045084635417,
      "grad_norm": 15.05029582977295,
      "learning_rate": 2.6084936041361315e-06,
      "loss": 3.3564,
      "step": 47845
    },
    {
      "epoch": 0.48675537109375,
      "grad_norm": 10.648780822753906,
      "learning_rate": 2.6080940985394393e-06,
      "loss": 2.9378,
      "step": 47850
    },
    {
      "epoch": 0.4868062337239583,
      "grad_norm": 15.958991050720215,
      "learning_rate": 2.6076945901771854e-06,
      "loss": 3.5244,
      "step": 47855
    },
    {
      "epoch": 0.4868570963541667,
      "grad_norm": 9.16628360748291,
      "learning_rate": 2.607295079059591e-06,
      "loss": 3.666,
      "step": 47860
    },
    {
      "epoch": 0.486907958984375,
      "grad_norm": 13.665358543395996,
      "learning_rate": 2.606895565196877e-06,
      "loss": 3.5797,
      "step": 47865
    },
    {
      "epoch": 0.4869588216145833,
      "grad_norm": 13.204471588134766,
      "learning_rate": 2.6064960485992657e-06,
      "loss": 3.0869,
      "step": 47870
    },
    {
      "epoch": 0.4870096842447917,
      "grad_norm": 9.907011032104492,
      "learning_rate": 2.6060965292769784e-06,
      "loss": 3.6301,
      "step": 47875
    },
    {
      "epoch": 0.487060546875,
      "grad_norm": 7.6872053146362305,
      "learning_rate": 2.6056970072402365e-06,
      "loss": 3.0106,
      "step": 47880
    },
    {
      "epoch": 0.4871114095052083,
      "grad_norm": 10.525590896606445,
      "learning_rate": 2.605297482499262e-06,
      "loss": 3.1349,
      "step": 47885
    },
    {
      "epoch": 0.4871622721354167,
      "grad_norm": 13.306690216064453,
      "learning_rate": 2.6048979550642755e-06,
      "loss": 3.7184,
      "step": 47890
    },
    {
      "epoch": 0.487213134765625,
      "grad_norm": 8.08199405670166,
      "learning_rate": 2.6044984249455e-06,
      "loss": 3.631,
      "step": 47895
    },
    {
      "epoch": 0.4872639973958333,
      "grad_norm": 14.965778350830078,
      "learning_rate": 2.6040988921531575e-06,
      "loss": 3.4085,
      "step": 47900
    },
    {
      "epoch": 0.4873148600260417,
      "grad_norm": 11.588021278381348,
      "learning_rate": 2.6036993566974693e-06,
      "loss": 2.9787,
      "step": 47905
    },
    {
      "epoch": 0.48736572265625,
      "grad_norm": 8.804408073425293,
      "learning_rate": 2.6032998185886575e-06,
      "loss": 2.9821,
      "step": 47910
    },
    {
      "epoch": 0.4874165852864583,
      "grad_norm": 14.51441478729248,
      "learning_rate": 2.6029002778369445e-06,
      "loss": 3.24,
      "step": 47915
    },
    {
      "epoch": 0.4874674479166667,
      "grad_norm": 10.842745780944824,
      "learning_rate": 2.6025007344525527e-06,
      "loss": 3.3752,
      "step": 47920
    },
    {
      "epoch": 0.487518310546875,
      "grad_norm": 15.25305461883545,
      "learning_rate": 2.6021011884457025e-06,
      "loss": 3.3172,
      "step": 47925
    },
    {
      "epoch": 0.4875691731770833,
      "grad_norm": 11.409184455871582,
      "learning_rate": 2.6017016398266187e-06,
      "loss": 3.2808,
      "step": 47930
    },
    {
      "epoch": 0.4876200358072917,
      "grad_norm": 13.894691467285156,
      "learning_rate": 2.601302088605523e-06,
      "loss": 3.082,
      "step": 47935
    },
    {
      "epoch": 0.4876708984375,
      "grad_norm": 11.15825080871582,
      "learning_rate": 2.6009025347926357e-06,
      "loss": 3.525,
      "step": 47940
    },
    {
      "epoch": 0.4877217610677083,
      "grad_norm": 12.198467254638672,
      "learning_rate": 2.600502978398183e-06,
      "loss": 3.4651,
      "step": 47945
    },
    {
      "epoch": 0.4877726236979167,
      "grad_norm": 17.750070571899414,
      "learning_rate": 2.6001034194323838e-06,
      "loss": 3.5371,
      "step": 47950
    },
    {
      "epoch": 0.487823486328125,
      "grad_norm": 9.900705337524414,
      "learning_rate": 2.5997038579054636e-06,
      "loss": 3.3005,
      "step": 47955
    },
    {
      "epoch": 0.4878743489583333,
      "grad_norm": 11.648094177246094,
      "learning_rate": 2.5993042938276423e-06,
      "loss": 3.3285,
      "step": 47960
    },
    {
      "epoch": 0.4879252115885417,
      "grad_norm": 17.3449764251709,
      "learning_rate": 2.5989047272091457e-06,
      "loss": 3.7537,
      "step": 47965
    },
    {
      "epoch": 0.48797607421875,
      "grad_norm": 10.940754890441895,
      "learning_rate": 2.598505158060194e-06,
      "loss": 3.6213,
      "step": 47970
    },
    {
      "epoch": 0.4880269368489583,
      "grad_norm": 11.876534461975098,
      "learning_rate": 2.5981055863910115e-06,
      "loss": 3.2027,
      "step": 47975
    },
    {
      "epoch": 0.4880777994791667,
      "grad_norm": 11.180745124816895,
      "learning_rate": 2.597706012211821e-06,
      "loss": 3.1598,
      "step": 47980
    },
    {
      "epoch": 0.488128662109375,
      "grad_norm": 19.479246139526367,
      "learning_rate": 2.597306435532846e-06,
      "loss": 3.4352,
      "step": 47985
    },
    {
      "epoch": 0.4881795247395833,
      "grad_norm": 6.949069976806641,
      "learning_rate": 2.5969068563643076e-06,
      "loss": 2.9822,
      "step": 47990
    },
    {
      "epoch": 0.4882303873697917,
      "grad_norm": 13.0645751953125,
      "learning_rate": 2.596507274716431e-06,
      "loss": 3.5882,
      "step": 47995
    },
    {
      "epoch": 0.48828125,
      "grad_norm": 14.730225563049316,
      "learning_rate": 2.5961076905994386e-06,
      "loss": 3.2193,
      "step": 48000
    },
    {
      "epoch": 0.4883321126302083,
      "grad_norm": 14.940589904785156,
      "learning_rate": 2.5957081040235536e-06,
      "loss": 3.6021,
      "step": 48005
    },
    {
      "epoch": 0.4883829752604167,
      "grad_norm": 9.172715187072754,
      "learning_rate": 2.5953085149989993e-06,
      "loss": 3.2141,
      "step": 48010
    },
    {
      "epoch": 0.488433837890625,
      "grad_norm": 12.160663604736328,
      "learning_rate": 2.5949089235359997e-06,
      "loss": 3.2393,
      "step": 48015
    },
    {
      "epoch": 0.4884847005208333,
      "grad_norm": 9.29576587677002,
      "learning_rate": 2.594509329644777e-06,
      "loss": 3.3738,
      "step": 48020
    },
    {
      "epoch": 0.4885355631510417,
      "grad_norm": 8.26999568939209,
      "learning_rate": 2.5941097333355562e-06,
      "loss": 3.4324,
      "step": 48025
    },
    {
      "epoch": 0.48858642578125,
      "grad_norm": 12.226856231689453,
      "learning_rate": 2.59371013461856e-06,
      "loss": 3.3112,
      "step": 48030
    },
    {
      "epoch": 0.4886372884114583,
      "grad_norm": 12.230896949768066,
      "learning_rate": 2.593310533504012e-06,
      "loss": 3.2613,
      "step": 48035
    },
    {
      "epoch": 0.4886881510416667,
      "grad_norm": 7.127318382263184,
      "learning_rate": 2.592910930002136e-06,
      "loss": 3.5765,
      "step": 48040
    },
    {
      "epoch": 0.488739013671875,
      "grad_norm": 6.903397560119629,
      "learning_rate": 2.5925113241231566e-06,
      "loss": 3.5333,
      "step": 48045
    },
    {
      "epoch": 0.4887898763020833,
      "grad_norm": 14.250541687011719,
      "learning_rate": 2.592111715877297e-06,
      "loss": 3.2821,
      "step": 48050
    },
    {
      "epoch": 0.4888407389322917,
      "grad_norm": 10.005630493164062,
      "learning_rate": 2.5917121052747795e-06,
      "loss": 3.3414,
      "step": 48055
    },
    {
      "epoch": 0.4888916015625,
      "grad_norm": 14.25844955444336,
      "learning_rate": 2.5913124923258308e-06,
      "loss": 3.4396,
      "step": 48060
    },
    {
      "epoch": 0.4889424641927083,
      "grad_norm": 11.528429985046387,
      "learning_rate": 2.5909128770406737e-06,
      "loss": 3.6652,
      "step": 48065
    },
    {
      "epoch": 0.4889933268229167,
      "grad_norm": 11.45598316192627,
      "learning_rate": 2.5905132594295316e-06,
      "loss": 3.5595,
      "step": 48070
    },
    {
      "epoch": 0.489044189453125,
      "grad_norm": 11.216914176940918,
      "learning_rate": 2.5901136395026294e-06,
      "loss": 3.5003,
      "step": 48075
    },
    {
      "epoch": 0.4890950520833333,
      "grad_norm": 16.073488235473633,
      "learning_rate": 2.589714017270191e-06,
      "loss": 3.3486,
      "step": 48080
    },
    {
      "epoch": 0.4891459147135417,
      "grad_norm": 13.13344669342041,
      "learning_rate": 2.589314392742441e-06,
      "loss": 3.7033,
      "step": 48085
    },
    {
      "epoch": 0.48919677734375,
      "grad_norm": 9.148369789123535,
      "learning_rate": 2.588914765929604e-06,
      "loss": 3.5296,
      "step": 48090
    },
    {
      "epoch": 0.4892476399739583,
      "grad_norm": 16.00242042541504,
      "learning_rate": 2.588515136841902e-06,
      "loss": 3.1364,
      "step": 48095
    },
    {
      "epoch": 0.4892985026041667,
      "grad_norm": 16.407093048095703,
      "learning_rate": 2.588115505489563e-06,
      "loss": 3.4385,
      "step": 48100
    },
    {
      "epoch": 0.489349365234375,
      "grad_norm": 13.131834983825684,
      "learning_rate": 2.587715871882808e-06,
      "loss": 3.7845,
      "step": 48105
    },
    {
      "epoch": 0.4894002278645833,
      "grad_norm": 13.327959060668945,
      "learning_rate": 2.5873162360318648e-06,
      "loss": 3.2955,
      "step": 48110
    },
    {
      "epoch": 0.4894510904947917,
      "grad_norm": 9.44006061553955,
      "learning_rate": 2.586916597946956e-06,
      "loss": 3.2827,
      "step": 48115
    },
    {
      "epoch": 0.489501953125,
      "grad_norm": 13.902898788452148,
      "learning_rate": 2.5865169576383055e-06,
      "loss": 3.2474,
      "step": 48120
    },
    {
      "epoch": 0.4895528157552083,
      "grad_norm": 12.939776420593262,
      "learning_rate": 2.5861173151161404e-06,
      "loss": 3.6203,
      "step": 48125
    },
    {
      "epoch": 0.4896036783854167,
      "grad_norm": 15.50821590423584,
      "learning_rate": 2.585717670390684e-06,
      "loss": 3.5654,
      "step": 48130
    },
    {
      "epoch": 0.489654541015625,
      "grad_norm": 13.119051933288574,
      "learning_rate": 2.585318023472161e-06,
      "loss": 3.1919,
      "step": 48135
    },
    {
      "epoch": 0.4897054036458333,
      "grad_norm": 10.716808319091797,
      "learning_rate": 2.5849183743707967e-06,
      "loss": 3.2412,
      "step": 48140
    },
    {
      "epoch": 0.4897562662760417,
      "grad_norm": 12.619991302490234,
      "learning_rate": 2.584518723096816e-06,
      "loss": 3.2517,
      "step": 48145
    },
    {
      "epoch": 0.48980712890625,
      "grad_norm": 13.602775573730469,
      "learning_rate": 2.5841190696604434e-06,
      "loss": 2.9025,
      "step": 48150
    },
    {
      "epoch": 0.4898579915364583,
      "grad_norm": 11.507429122924805,
      "learning_rate": 2.5837194140719057e-06,
      "loss": 3.4202,
      "step": 48155
    },
    {
      "epoch": 0.4899088541666667,
      "grad_norm": 12.635565757751465,
      "learning_rate": 2.5833197563414246e-06,
      "loss": 3.4805,
      "step": 48160
    },
    {
      "epoch": 0.489959716796875,
      "grad_norm": 15.815251350402832,
      "learning_rate": 2.5829200964792287e-06,
      "loss": 3.9183,
      "step": 48165
    },
    {
      "epoch": 0.4900105794270833,
      "grad_norm": 13.597136497497559,
      "learning_rate": 2.5825204344955414e-06,
      "loss": 3.4959,
      "step": 48170
    },
    {
      "epoch": 0.4900614420572917,
      "grad_norm": 10.966267585754395,
      "learning_rate": 2.5821207704005885e-06,
      "loss": 3.4393,
      "step": 48175
    },
    {
      "epoch": 0.4901123046875,
      "grad_norm": 12.54188346862793,
      "learning_rate": 2.5817211042045954e-06,
      "loss": 3.4757,
      "step": 48180
    },
    {
      "epoch": 0.4901631673177083,
      "grad_norm": 8.404973983764648,
      "learning_rate": 2.581321435917786e-06,
      "loss": 3.1954,
      "step": 48185
    },
    {
      "epoch": 0.4902140299479167,
      "grad_norm": 11.254782676696777,
      "learning_rate": 2.5809217655503882e-06,
      "loss": 3.3841,
      "step": 48190
    },
    {
      "epoch": 0.490264892578125,
      "grad_norm": 14.785955429077148,
      "learning_rate": 2.580522093112626e-06,
      "loss": 3.095,
      "step": 48195
    },
    {
      "epoch": 0.4903157552083333,
      "grad_norm": 16.44007110595703,
      "learning_rate": 2.5801224186147243e-06,
      "loss": 3.0962,
      "step": 48200
    },
    {
      "epoch": 0.4903666178385417,
      "grad_norm": 15.266268730163574,
      "learning_rate": 2.5797227420669107e-06,
      "loss": 3.2877,
      "step": 48205
    },
    {
      "epoch": 0.49041748046875,
      "grad_norm": 11.603970527648926,
      "learning_rate": 2.5793230634794087e-06,
      "loss": 3.2051,
      "step": 48210
    },
    {
      "epoch": 0.4904683430989583,
      "grad_norm": 10.646866798400879,
      "learning_rate": 2.578923382862446e-06,
      "loss": 3.1391,
      "step": 48215
    },
    {
      "epoch": 0.4905192057291667,
      "grad_norm": 12.524734497070312,
      "learning_rate": 2.5785237002262463e-06,
      "loss": 3.5244,
      "step": 48220
    },
    {
      "epoch": 0.490570068359375,
      "grad_norm": 9.726661682128906,
      "learning_rate": 2.578124015581036e-06,
      "loss": 3.2464,
      "step": 48225
    },
    {
      "epoch": 0.4906209309895833,
      "grad_norm": 9.554346084594727,
      "learning_rate": 2.5777243289370425e-06,
      "loss": 3.1529,
      "step": 48230
    },
    {
      "epoch": 0.4906717936197917,
      "grad_norm": 15.186952590942383,
      "learning_rate": 2.57732464030449e-06,
      "loss": 3.4499,
      "step": 48235
    },
    {
      "epoch": 0.49072265625,
      "grad_norm": 9.830705642700195,
      "learning_rate": 2.5769249496936044e-06,
      "loss": 3.4578,
      "step": 48240
    },
    {
      "epoch": 0.4907735188802083,
      "grad_norm": 11.41324520111084,
      "learning_rate": 2.5765252571146127e-06,
      "loss": 3.267,
      "step": 48245
    },
    {
      "epoch": 0.4908243815104167,
      "grad_norm": 10.870372772216797,
      "learning_rate": 2.57612556257774e-06,
      "loss": 3.5487,
      "step": 48250
    },
    {
      "epoch": 0.490875244140625,
      "grad_norm": 15.754440307617188,
      "learning_rate": 2.575725866093213e-06,
      "loss": 3.463,
      "step": 48255
    },
    {
      "epoch": 0.4909261067708333,
      "grad_norm": 14.528594017028809,
      "learning_rate": 2.575326167671258e-06,
      "loss": 3.1897,
      "step": 48260
    },
    {
      "epoch": 0.4909769694010417,
      "grad_norm": 10.713245391845703,
      "learning_rate": 2.5749264673221005e-06,
      "loss": 3.5306,
      "step": 48265
    },
    {
      "epoch": 0.49102783203125,
      "grad_norm": 11.782358169555664,
      "learning_rate": 2.5745267650559674e-06,
      "loss": 3.3349,
      "step": 48270
    },
    {
      "epoch": 0.4910786946614583,
      "grad_norm": 14.125024795532227,
      "learning_rate": 2.574127060883084e-06,
      "loss": 3.4315,
      "step": 48275
    },
    {
      "epoch": 0.4911295572916667,
      "grad_norm": 11.369453430175781,
      "learning_rate": 2.573727354813678e-06,
      "loss": 3.0201,
      "step": 48280
    },
    {
      "epoch": 0.491180419921875,
      "grad_norm": 8.513801574707031,
      "learning_rate": 2.5733276468579755e-06,
      "loss": 3.1815,
      "step": 48285
    },
    {
      "epoch": 0.4912312825520833,
      "grad_norm": 10.482697486877441,
      "learning_rate": 2.5729279370262017e-06,
      "loss": 3.2698,
      "step": 48290
    },
    {
      "epoch": 0.4912821451822917,
      "grad_norm": 13.890192031860352,
      "learning_rate": 2.5725282253285837e-06,
      "loss": 3.5445,
      "step": 48295
    },
    {
      "epoch": 0.4913330078125,
      "grad_norm": 13.60200309753418,
      "learning_rate": 2.572128511775349e-06,
      "loss": 3.5674,
      "step": 48300
    },
    {
      "epoch": 0.4913838704427083,
      "grad_norm": 17.395017623901367,
      "learning_rate": 2.5717287963767227e-06,
      "loss": 3.266,
      "step": 48305
    },
    {
      "epoch": 0.4914347330729167,
      "grad_norm": 10.49577522277832,
      "learning_rate": 2.5713290791429334e-06,
      "loss": 3.5678,
      "step": 48310
    },
    {
      "epoch": 0.491485595703125,
      "grad_norm": 9.473282814025879,
      "learning_rate": 2.570929360084205e-06,
      "loss": 3.2889,
      "step": 48315
    },
    {
      "epoch": 0.4915364583333333,
      "grad_norm": 11.267034530639648,
      "learning_rate": 2.5705296392107666e-06,
      "loss": 3.3956,
      "step": 48320
    },
    {
      "epoch": 0.4915873209635417,
      "grad_norm": 15.123723030090332,
      "learning_rate": 2.570129916532844e-06,
      "loss": 3.3584,
      "step": 48325
    },
    {
      "epoch": 0.49163818359375,
      "grad_norm": 13.88943099975586,
      "learning_rate": 2.5697301920606636e-06,
      "loss": 3.3674,
      "step": 48330
    },
    {
      "epoch": 0.4916890462239583,
      "grad_norm": 9.391311645507812,
      "learning_rate": 2.569330465804454e-06,
      "loss": 3.3737,
      "step": 48335
    },
    {
      "epoch": 0.4917399088541667,
      "grad_norm": 12.441389083862305,
      "learning_rate": 2.5689307377744394e-06,
      "loss": 2.9606,
      "step": 48340
    },
    {
      "epoch": 0.491790771484375,
      "grad_norm": 12.627106666564941,
      "learning_rate": 2.568531007980849e-06,
      "loss": 3.5774,
      "step": 48345
    },
    {
      "epoch": 0.4918416341145833,
      "grad_norm": 12.987388610839844,
      "learning_rate": 2.568131276433909e-06,
      "loss": 3.6828,
      "step": 48350
    },
    {
      "epoch": 0.4918924967447917,
      "grad_norm": 7.0246171951293945,
      "learning_rate": 2.5677315431438467e-06,
      "loss": 3.4315,
      "step": 48355
    },
    {
      "epoch": 0.491943359375,
      "grad_norm": 7.368935585021973,
      "learning_rate": 2.5673318081208883e-06,
      "loss": 3.259,
      "step": 48360
    },
    {
      "epoch": 0.4919942220052083,
      "grad_norm": 14.435423851013184,
      "learning_rate": 2.5669320713752618e-06,
      "loss": 3.3111,
      "step": 48365
    },
    {
      "epoch": 0.4920450846354167,
      "grad_norm": 14.978443145751953,
      "learning_rate": 2.566532332917194e-06,
      "loss": 3.1992,
      "step": 48370
    },
    {
      "epoch": 0.492095947265625,
      "grad_norm": 7.657479763031006,
      "learning_rate": 2.5661325927569124e-06,
      "loss": 3.4598,
      "step": 48375
    },
    {
      "epoch": 0.4921468098958333,
      "grad_norm": 11.787341117858887,
      "learning_rate": 2.5657328509046437e-06,
      "loss": 3.3408,
      "step": 48380
    },
    {
      "epoch": 0.4921976725260417,
      "grad_norm": 11.836021423339844,
      "learning_rate": 2.565333107370617e-06,
      "loss": 3.2878,
      "step": 48385
    },
    {
      "epoch": 0.49224853515625,
      "grad_norm": 10.637925148010254,
      "learning_rate": 2.564933362165056e-06,
      "loss": 3.5864,
      "step": 48390
    },
    {
      "epoch": 0.4922993977864583,
      "grad_norm": 10.810693740844727,
      "learning_rate": 2.564533615298192e-06,
      "loss": 3.2705,
      "step": 48395
    },
    {
      "epoch": 0.4923502604166667,
      "grad_norm": 13.188955307006836,
      "learning_rate": 2.5641338667802503e-06,
      "loss": 2.8802,
      "step": 48400
    },
    {
      "epoch": 0.492401123046875,
      "grad_norm": 12.960467338562012,
      "learning_rate": 2.5637341166214585e-06,
      "loss": 3.7479,
      "step": 48405
    },
    {
      "epoch": 0.4924519856770833,
      "grad_norm": 13.326948165893555,
      "learning_rate": 2.5633343648320445e-06,
      "loss": 3.3727,
      "step": 48410
    },
    {
      "epoch": 0.4925028483072917,
      "grad_norm": 10.184150695800781,
      "learning_rate": 2.5629346114222357e-06,
      "loss": 3.1968,
      "step": 48415
    },
    {
      "epoch": 0.4925537109375,
      "grad_norm": 11.197040557861328,
      "learning_rate": 2.5625348564022604e-06,
      "loss": 3.4119,
      "step": 48420
    },
    {
      "epoch": 0.4926045735677083,
      "grad_norm": 17.48925018310547,
      "learning_rate": 2.5621350997823454e-06,
      "loss": 3.5111,
      "step": 48425
    },
    {
      "epoch": 0.4926554361979167,
      "grad_norm": 8.670833587646484,
      "learning_rate": 2.561735341572718e-06,
      "loss": 3.3153,
      "step": 48430
    },
    {
      "epoch": 0.492706298828125,
      "grad_norm": 11.009799003601074,
      "learning_rate": 2.5613355817836073e-06,
      "loss": 3.3655,
      "step": 48435
    },
    {
      "epoch": 0.4927571614583333,
      "grad_norm": 10.905400276184082,
      "learning_rate": 2.5609358204252404e-06,
      "loss": 3.5021,
      "step": 48440
    },
    {
      "epoch": 0.4928080240885417,
      "grad_norm": 16.436967849731445,
      "learning_rate": 2.560536057507844e-06,
      "loss": 3.2442,
      "step": 48445
    },
    {
      "epoch": 0.49285888671875,
      "grad_norm": 9.947446823120117,
      "learning_rate": 2.560136293041648e-06,
      "loss": 3.5158,
      "step": 48450
    },
    {
      "epoch": 0.4929097493489583,
      "grad_norm": 10.083742141723633,
      "learning_rate": 2.5597365270368784e-06,
      "loss": 3.4966,
      "step": 48455
    },
    {
      "epoch": 0.4929606119791667,
      "grad_norm": 10.116064071655273,
      "learning_rate": 2.559336759503764e-06,
      "loss": 3.1833,
      "step": 48460
    },
    {
      "epoch": 0.493011474609375,
      "grad_norm": 13.628241539001465,
      "learning_rate": 2.5589369904525328e-06,
      "loss": 3.4158,
      "step": 48465
    },
    {
      "epoch": 0.4930623372395833,
      "grad_norm": 10.907445907592773,
      "learning_rate": 2.558537219893413e-06,
      "loss": 3.4594,
      "step": 48470
    },
    {
      "epoch": 0.4931131998697917,
      "grad_norm": 10.239285469055176,
      "learning_rate": 2.5581374478366315e-06,
      "loss": 3.1526,
      "step": 48475
    },
    {
      "epoch": 0.4931640625,
      "grad_norm": 8.00221061706543,
      "learning_rate": 2.557737674292419e-06,
      "loss": 3.5418,
      "step": 48480
    },
    {
      "epoch": 0.4932149251302083,
      "grad_norm": 7.948908805847168,
      "learning_rate": 2.5573378992709997e-06,
      "loss": 3.5753,
      "step": 48485
    },
    {
      "epoch": 0.4932657877604167,
      "grad_norm": 14.463628768920898,
      "learning_rate": 2.5569381227826047e-06,
      "loss": 3.1937,
      "step": 48490
    },
    {
      "epoch": 0.493316650390625,
      "grad_norm": 10.312268257141113,
      "learning_rate": 2.556538344837461e-06,
      "loss": 3.0908,
      "step": 48495
    },
    {
      "epoch": 0.4933675130208333,
      "grad_norm": 13.522259712219238,
      "learning_rate": 2.5561385654457977e-06,
      "loss": 3.4039,
      "step": 48500
    },
    {
      "epoch": 0.4934183756510417,
      "grad_norm": 18.827590942382812,
      "learning_rate": 2.5557387846178424e-06,
      "loss": 3.5376,
      "step": 48505
    },
    {
      "epoch": 0.49346923828125,
      "grad_norm": 12.525690078735352,
      "learning_rate": 2.5553390023638232e-06,
      "loss": 3.4584,
      "step": 48510
    },
    {
      "epoch": 0.4935201009114583,
      "grad_norm": 11.270380020141602,
      "learning_rate": 2.5549392186939686e-06,
      "loss": 3.3056,
      "step": 48515
    },
    {
      "epoch": 0.4935709635416667,
      "grad_norm": 13.468146324157715,
      "learning_rate": 2.554539433618507e-06,
      "loss": 3.4103,
      "step": 48520
    },
    {
      "epoch": 0.493621826171875,
      "grad_norm": 12.6211519241333,
      "learning_rate": 2.5541396471476677e-06,
      "loss": 3.5486,
      "step": 48525
    },
    {
      "epoch": 0.4936726888020833,
      "grad_norm": 15.345010757446289,
      "learning_rate": 2.553739859291678e-06,
      "loss": 4.056,
      "step": 48530
    },
    {
      "epoch": 0.4937235514322917,
      "grad_norm": 9.263572692871094,
      "learning_rate": 2.553340070060766e-06,
      "loss": 3.4742,
      "step": 48535
    },
    {
      "epoch": 0.4937744140625,
      "grad_norm": 13.135659217834473,
      "learning_rate": 2.552940279465162e-06,
      "loss": 3.0927,
      "step": 48540
    },
    {
      "epoch": 0.4938252766927083,
      "grad_norm": 9.30755615234375,
      "learning_rate": 2.552540487515093e-06,
      "loss": 3.3484,
      "step": 48545
    },
    {
      "epoch": 0.4938761393229167,
      "grad_norm": 14.75592041015625,
      "learning_rate": 2.552140694220788e-06,
      "loss": 3.3657,
      "step": 48550
    },
    {
      "epoch": 0.493927001953125,
      "grad_norm": 14.432010650634766,
      "learning_rate": 2.5517408995924766e-06,
      "loss": 3.3975,
      "step": 48555
    },
    {
      "epoch": 0.4939778645833333,
      "grad_norm": 9.98413372039795,
      "learning_rate": 2.551341103640385e-06,
      "loss": 3.1357,
      "step": 48560
    },
    {
      "epoch": 0.4940287272135417,
      "grad_norm": 13.80256175994873,
      "learning_rate": 2.5509413063747453e-06,
      "loss": 3.1355,
      "step": 48565
    },
    {
      "epoch": 0.49407958984375,
      "grad_norm": 11.76331615447998,
      "learning_rate": 2.5505415078057827e-06,
      "loss": 3.4596,
      "step": 48570
    },
    {
      "epoch": 0.4941304524739583,
      "grad_norm": 10.180501937866211,
      "learning_rate": 2.5501417079437287e-06,
      "loss": 3.5006,
      "step": 48575
    },
    {
      "epoch": 0.4941813151041667,
      "grad_norm": 7.968472003936768,
      "learning_rate": 2.5497419067988098e-06,
      "loss": 3.1618,
      "step": 48580
    },
    {
      "epoch": 0.494232177734375,
      "grad_norm": 10.584066390991211,
      "learning_rate": 2.5493421043812573e-06,
      "loss": 3.2592,
      "step": 48585
    },
    {
      "epoch": 0.4942830403645833,
      "grad_norm": 16.969036102294922,
      "learning_rate": 2.548942300701298e-06,
      "loss": 3.4475,
      "step": 48590
    },
    {
      "epoch": 0.4943339029947917,
      "grad_norm": 10.872123718261719,
      "learning_rate": 2.548542495769162e-06,
      "loss": 3.6009,
      "step": 48595
    },
    {
      "epoch": 0.494384765625,
      "grad_norm": 11.284872055053711,
      "learning_rate": 2.5481426895950775e-06,
      "loss": 3.1851,
      "step": 48600
    },
    {
      "epoch": 0.4944356282552083,
      "grad_norm": 12.281546592712402,
      "learning_rate": 2.5477428821892734e-06,
      "loss": 3.1344,
      "step": 48605
    },
    {
      "epoch": 0.4944864908854167,
      "grad_norm": 13.539896011352539,
      "learning_rate": 2.5473430735619797e-06,
      "loss": 3.4482,
      "step": 48610
    },
    {
      "epoch": 0.494537353515625,
      "grad_norm": 12.465642929077148,
      "learning_rate": 2.5469432637234238e-06,
      "loss": 3.2899,
      "step": 48615
    },
    {
      "epoch": 0.4945882161458333,
      "grad_norm": 8.407029151916504,
      "learning_rate": 2.546543452683836e-06,
      "loss": 3.3725,
      "step": 48620
    },
    {
      "epoch": 0.4946390787760417,
      "grad_norm": 99.68233489990234,
      "learning_rate": 2.546143640453445e-06,
      "loss": 3.8263,
      "step": 48625
    },
    {
      "epoch": 0.49468994140625,
      "grad_norm": 9.953444480895996,
      "learning_rate": 2.54574382704248e-06,
      "loss": 4.2646,
      "step": 48630
    },
    {
      "epoch": 0.4947408040364583,
      "grad_norm": 12.155915260314941,
      "learning_rate": 2.54534401246117e-06,
      "loss": 3.2373,
      "step": 48635
    },
    {
      "epoch": 0.4947916666666667,
      "grad_norm": 11.434561729431152,
      "learning_rate": 2.544944196719744e-06,
      "loss": 3.3427,
      "step": 48640
    },
    {
      "epoch": 0.494842529296875,
      "grad_norm": 13.522272109985352,
      "learning_rate": 2.5445443798284314e-06,
      "loss": 3.5858,
      "step": 48645
    },
    {
      "epoch": 0.4948933919270833,
      "grad_norm": 10.04487133026123,
      "learning_rate": 2.544144561797462e-06,
      "loss": 3.2289,
      "step": 48650
    },
    {
      "epoch": 0.4949442545572917,
      "grad_norm": 16.41366958618164,
      "learning_rate": 2.543744742637064e-06,
      "loss": 3.3868,
      "step": 48655
    },
    {
      "epoch": 0.4949951171875,
      "grad_norm": 15.050728797912598,
      "learning_rate": 2.5433449223574674e-06,
      "loss": 3.2918,
      "step": 48660
    },
    {
      "epoch": 0.4950459798177083,
      "grad_norm": 13.474639892578125,
      "learning_rate": 2.5429451009689006e-06,
      "loss": 3.3679,
      "step": 48665
    },
    {
      "epoch": 0.4950968424479167,
      "grad_norm": 11.2426118850708,
      "learning_rate": 2.542545278481594e-06,
      "loss": 3.277,
      "step": 48670
    },
    {
      "epoch": 0.495147705078125,
      "grad_norm": 8.831446647644043,
      "learning_rate": 2.5421454549057765e-06,
      "loss": 3.3783,
      "step": 48675
    },
    {
      "epoch": 0.4951985677083333,
      "grad_norm": 9.274211883544922,
      "learning_rate": 2.5417456302516775e-06,
      "loss": 3.1851,
      "step": 48680
    },
    {
      "epoch": 0.4952494303385417,
      "grad_norm": 10.510271072387695,
      "learning_rate": 2.5413458045295263e-06,
      "loss": 3.0859,
      "step": 48685
    },
    {
      "epoch": 0.49530029296875,
      "grad_norm": 8.18698501586914,
      "learning_rate": 2.5409459777495525e-06,
      "loss": 2.9118,
      "step": 48690
    },
    {
      "epoch": 0.4953511555989583,
      "grad_norm": 14.988253593444824,
      "learning_rate": 2.5405461499219857e-06,
      "loss": 3.2874,
      "step": 48695
    },
    {
      "epoch": 0.4954020182291667,
      "grad_norm": 11.119796752929688,
      "learning_rate": 2.540146321057056e-06,
      "loss": 3.4541,
      "step": 48700
    },
    {
      "epoch": 0.495452880859375,
      "grad_norm": 11.139017105102539,
      "learning_rate": 2.5397464911649904e-06,
      "loss": 3.5297,
      "step": 48705
    },
    {
      "epoch": 0.4955037434895833,
      "grad_norm": 12.22435188293457,
      "learning_rate": 2.5393466602560215e-06,
      "loss": 3.1274,
      "step": 48710
    },
    {
      "epoch": 0.4955546061197917,
      "grad_norm": 8.149970054626465,
      "learning_rate": 2.5389468283403772e-06,
      "loss": 2.9917,
      "step": 48715
    },
    {
      "epoch": 0.49560546875,
      "grad_norm": 8.23815631866455,
      "learning_rate": 2.5385469954282876e-06,
      "loss": 3.3093,
      "step": 48720
    },
    {
      "epoch": 0.4956563313802083,
      "grad_norm": 7.144126892089844,
      "learning_rate": 2.5381471615299828e-06,
      "loss": 3.3526,
      "step": 48725
    },
    {
      "epoch": 0.4957071940104167,
      "grad_norm": 8.10865306854248,
      "learning_rate": 2.5377473266556914e-06,
      "loss": 3.5419,
      "step": 48730
    },
    {
      "epoch": 0.495758056640625,
      "grad_norm": 12.645817756652832,
      "learning_rate": 2.5373474908156444e-06,
      "loss": 3.5775,
      "step": 48735
    },
    {
      "epoch": 0.4958089192708333,
      "grad_norm": 14.494380950927734,
      "learning_rate": 2.5369476540200693e-06,
      "loss": 3.3425,
      "step": 48740
    },
    {
      "epoch": 0.4958597819010417,
      "grad_norm": 6.908821105957031,
      "learning_rate": 2.5365478162791986e-06,
      "loss": 2.9617,
      "step": 48745
    },
    {
      "epoch": 0.49591064453125,
      "grad_norm": 16.56468963623047,
      "learning_rate": 2.5361479776032594e-06,
      "loss": 3.5609,
      "step": 48750
    },
    {
      "epoch": 0.4959615071614583,
      "grad_norm": 14.772183418273926,
      "learning_rate": 2.5357481380024836e-06,
      "loss": 3.4348,
      "step": 48755
    },
    {
      "epoch": 0.4960123697916667,
      "grad_norm": 9.944372177124023,
      "learning_rate": 2.5353482974870997e-06,
      "loss": 3.1382,
      "step": 48760
    },
    {
      "epoch": 0.496063232421875,
      "grad_norm": 7.180305004119873,
      "learning_rate": 2.5349484560673388e-06,
      "loss": 3.3043,
      "step": 48765
    },
    {
      "epoch": 0.4961140950520833,
      "grad_norm": 10.013150215148926,
      "learning_rate": 2.5345486137534287e-06,
      "loss": 3.4011,
      "step": 48770
    },
    {
      "epoch": 0.4961649576822917,
      "grad_norm": 10.231666564941406,
      "learning_rate": 2.5341487705556016e-06,
      "loss": 3.2933,
      "step": 48775
    },
    {
      "epoch": 0.4962158203125,
      "grad_norm": 16.945215225219727,
      "learning_rate": 2.533748926484086e-06,
      "loss": 3.5064,
      "step": 48780
    },
    {
      "epoch": 0.4962666829427083,
      "grad_norm": 14.766674041748047,
      "learning_rate": 2.533349081549112e-06,
      "loss": 3.152,
      "step": 48785
    },
    {
      "epoch": 0.4963175455729167,
      "grad_norm": 10.92806339263916,
      "learning_rate": 2.53294923576091e-06,
      "loss": 3.3627,
      "step": 48790
    },
    {
      "epoch": 0.496368408203125,
      "grad_norm": 14.547791481018066,
      "learning_rate": 2.532549389129709e-06,
      "loss": 3.7166,
      "step": 48795
    },
    {
      "epoch": 0.4964192708333333,
      "grad_norm": 13.458353042602539,
      "learning_rate": 2.5321495416657394e-06,
      "loss": 3.249,
      "step": 48800
    },
    {
      "epoch": 0.4964701334635417,
      "grad_norm": 13.802599906921387,
      "learning_rate": 2.531749693379232e-06,
      "loss": 3.2997,
      "step": 48805
    },
    {
      "epoch": 0.49652099609375,
      "grad_norm": 9.486502647399902,
      "learning_rate": 2.531349844280416e-06,
      "loss": 3.3716,
      "step": 48810
    },
    {
      "epoch": 0.4965718587239583,
      "grad_norm": 12.073533058166504,
      "learning_rate": 2.5309499943795217e-06,
      "loss": 3.1021,
      "step": 48815
    },
    {
      "epoch": 0.4966227213541667,
      "grad_norm": 13.449509620666504,
      "learning_rate": 2.530550143686779e-06,
      "loss": 3.4323,
      "step": 48820
    },
    {
      "epoch": 0.496673583984375,
      "grad_norm": 15.299708366394043,
      "learning_rate": 2.530150292212418e-06,
      "loss": 3.4064,
      "step": 48825
    },
    {
      "epoch": 0.4967244466145833,
      "grad_norm": 14.66781234741211,
      "learning_rate": 2.5297504399666702e-06,
      "loss": 3.2232,
      "step": 48830
    },
    {
      "epoch": 0.4967753092447917,
      "grad_norm": 11.898265838623047,
      "learning_rate": 2.529350586959763e-06,
      "loss": 3.2682,
      "step": 48835
    },
    {
      "epoch": 0.496826171875,
      "grad_norm": 10.700338363647461,
      "learning_rate": 2.5289507332019286e-06,
      "loss": 2.9856,
      "step": 48840
    },
    {
      "epoch": 0.4968770345052083,
      "grad_norm": 12.895772933959961,
      "learning_rate": 2.528550878703396e-06,
      "loss": 3.3705,
      "step": 48845
    },
    {
      "epoch": 0.4969278971354167,
      "grad_norm": 12.710967063903809,
      "learning_rate": 2.5281510234743966e-06,
      "loss": 3.3809,
      "step": 48850
    },
    {
      "epoch": 0.496978759765625,
      "grad_norm": 12.692770004272461,
      "learning_rate": 2.5277511675251598e-06,
      "loss": 3.0422,
      "step": 48855
    },
    {
      "epoch": 0.4970296223958333,
      "grad_norm": 14.448827743530273,
      "learning_rate": 2.5273513108659153e-06,
      "loss": 3.4203,
      "step": 48860
    },
    {
      "epoch": 0.4970804850260417,
      "grad_norm": 14.509695053100586,
      "learning_rate": 2.5269514535068947e-06,
      "loss": 3.4038,
      "step": 48865
    },
    {
      "epoch": 0.49713134765625,
      "grad_norm": 9.705320358276367,
      "learning_rate": 2.5265515954583273e-06,
      "loss": 3.1937,
      "step": 48870
    },
    {
      "epoch": 0.4971822102864583,
      "grad_norm": 16.22066307067871,
      "learning_rate": 2.5261517367304434e-06,
      "loss": 3.2324,
      "step": 48875
    },
    {
      "epoch": 0.4972330729166667,
      "grad_norm": 14.455748558044434,
      "learning_rate": 2.525751877333474e-06,
      "loss": 3.1136,
      "step": 48880
    },
    {
      "epoch": 0.497283935546875,
      "grad_norm": 15.725226402282715,
      "learning_rate": 2.5253520172776484e-06,
      "loss": 3.4166,
      "step": 48885
    },
    {
      "epoch": 0.4973347981770833,
      "grad_norm": 11.71474552154541,
      "learning_rate": 2.5249521565731983e-06,
      "loss": 3.1958,
      "step": 48890
    },
    {
      "epoch": 0.4973856608072917,
      "grad_norm": 10.169915199279785,
      "learning_rate": 2.524552295230353e-06,
      "loss": 3.2446,
      "step": 48895
    },
    {
      "epoch": 0.4974365234375,
      "grad_norm": 15.470590591430664,
      "learning_rate": 2.5241524332593425e-06,
      "loss": 3.4225,
      "step": 48900
    },
    {
      "epoch": 0.4974873860677083,
      "grad_norm": 12.102554321289062,
      "learning_rate": 2.523752570670398e-06,
      "loss": 3.1009,
      "step": 48905
    },
    {
      "epoch": 0.4975382486979167,
      "grad_norm": 8.739709854125977,
      "learning_rate": 2.5233527074737493e-06,
      "loss": 3.4734,
      "step": 48910
    },
    {
      "epoch": 0.497589111328125,
      "grad_norm": 12.60647201538086,
      "learning_rate": 2.5229528436796274e-06,
      "loss": 3.4694,
      "step": 48915
    },
    {
      "epoch": 0.4976399739583333,
      "grad_norm": 7.893643856048584,
      "learning_rate": 2.5225529792982623e-06,
      "loss": 3.2898,
      "step": 48920
    },
    {
      "epoch": 0.4976908365885417,
      "grad_norm": 7.657857894897461,
      "learning_rate": 2.522153114339885e-06,
      "loss": 3.2166,
      "step": 48925
    },
    {
      "epoch": 0.49774169921875,
      "grad_norm": 14.99483871459961,
      "learning_rate": 2.5217532488147245e-06,
      "loss": 4.0929,
      "step": 48930
    },
    {
      "epoch": 0.4977925618489583,
      "grad_norm": 13.64297866821289,
      "learning_rate": 2.521353382733013e-06,
      "loss": 3.0839,
      "step": 48935
    },
    {
      "epoch": 0.4978434244791667,
      "grad_norm": 8.626524925231934,
      "learning_rate": 2.5209535161049804e-06,
      "loss": 3.2194,
      "step": 48940
    },
    {
      "epoch": 0.497894287109375,
      "grad_norm": 11.188443183898926,
      "learning_rate": 2.520553648940857e-06,
      "loss": 3.2343,
      "step": 48945
    },
    {
      "epoch": 0.4979451497395833,
      "grad_norm": 17.096988677978516,
      "learning_rate": 2.520153781250873e-06,
      "loss": 3.1586,
      "step": 48950
    },
    {
      "epoch": 0.4979960123697917,
      "grad_norm": 14.983606338500977,
      "learning_rate": 2.51975391304526e-06,
      "loss": 4.0485,
      "step": 48955
    },
    {
      "epoch": 0.498046875,
      "grad_norm": 14.723469734191895,
      "learning_rate": 2.5193540443342474e-06,
      "loss": 3.3887,
      "step": 48960
    },
    {
      "epoch": 0.4980977376302083,
      "grad_norm": 7.029719352722168,
      "learning_rate": 2.518954175128066e-06,
      "loss": 3.2619,
      "step": 48965
    },
    {
      "epoch": 0.4981486002604167,
      "grad_norm": 14.817591667175293,
      "learning_rate": 2.5185543054369467e-06,
      "loss": 3.7279,
      "step": 48970
    },
    {
      "epoch": 0.498199462890625,
      "grad_norm": 9.7324857711792,
      "learning_rate": 2.5181544352711194e-06,
      "loss": 3.1242,
      "step": 48975
    },
    {
      "epoch": 0.4982503255208333,
      "grad_norm": 11.890584945678711,
      "learning_rate": 2.5177545646408158e-06,
      "loss": 3.2638,
      "step": 48980
    },
    {
      "epoch": 0.4983011881510417,
      "grad_norm": 13.200179100036621,
      "learning_rate": 2.517354693556266e-06,
      "loss": 3.1061,
      "step": 48985
    },
    {
      "epoch": 0.49835205078125,
      "grad_norm": 11.83781909942627,
      "learning_rate": 2.5169548220276997e-06,
      "loss": 3.4721,
      "step": 48990
    },
    {
      "epoch": 0.4984029134114583,
      "grad_norm": 15.258880615234375,
      "learning_rate": 2.516554950065349e-06,
      "loss": 3.2813,
      "step": 48995
    },
    {
      "epoch": 0.4984537760416667,
      "grad_norm": 12.589689254760742,
      "learning_rate": 2.5161550776794434e-06,
      "loss": 3.4539,
      "step": 49000
    },
    {
      "epoch": 0.498504638671875,
      "grad_norm": 13.380403518676758,
      "learning_rate": 2.5157552048802136e-06,
      "loss": 3.1147,
      "step": 49005
    },
    {
      "epoch": 0.4985555013020833,
      "grad_norm": 10.888894081115723,
      "learning_rate": 2.515355331677892e-06,
      "loss": 3.4481,
      "step": 49010
    },
    {
      "epoch": 0.4986063639322917,
      "grad_norm": 8.734400749206543,
      "learning_rate": 2.514955458082706e-06,
      "loss": 3.6778,
      "step": 49015
    },
    {
      "epoch": 0.4986572265625,
      "grad_norm": 13.816628456115723,
      "learning_rate": 2.5145555841048896e-06,
      "loss": 3.3654,
      "step": 49020
    },
    {
      "epoch": 0.4987080891927083,
      "grad_norm": 14.505660057067871,
      "learning_rate": 2.5141557097546722e-06,
      "loss": 3.0168,
      "step": 49025
    },
    {
      "epoch": 0.4987589518229167,
      "grad_norm": 14.234870910644531,
      "learning_rate": 2.513755835042283e-06,
      "loss": 3.6649,
      "step": 49030
    },
    {
      "epoch": 0.498809814453125,
      "grad_norm": 17.624862670898438,
      "learning_rate": 2.513355959977955e-06,
      "loss": 3.7414,
      "step": 49035
    },
    {
      "epoch": 0.4988606770833333,
      "grad_norm": 14.029349327087402,
      "learning_rate": 2.5129560845719173e-06,
      "loss": 3.109,
      "step": 49040
    },
    {
      "epoch": 0.4989115397135417,
      "grad_norm": 9.352248191833496,
      "learning_rate": 2.512556208834402e-06,
      "loss": 3.5068,
      "step": 49045
    },
    {
      "epoch": 0.49896240234375,
      "grad_norm": 11.51716423034668,
      "learning_rate": 2.5121563327756393e-06,
      "loss": 3.312,
      "step": 49050
    },
    {
      "epoch": 0.4990132649739583,
      "grad_norm": 12.597025871276855,
      "learning_rate": 2.5117564564058582e-06,
      "loss": 3.4719,
      "step": 49055
    },
    {
      "epoch": 0.4990641276041667,
      "grad_norm": 9.703937530517578,
      "learning_rate": 2.5113565797352923e-06,
      "loss": 2.6813,
      "step": 49060
    },
    {
      "epoch": 0.499114990234375,
      "grad_norm": 11.819278717041016,
      "learning_rate": 2.5109567027741704e-06,
      "loss": 3.2599,
      "step": 49065
    },
    {
      "epoch": 0.4991658528645833,
      "grad_norm": 11.828073501586914,
      "learning_rate": 2.5105568255327246e-06,
      "loss": 3.2094,
      "step": 49070
    },
    {
      "epoch": 0.4992167154947917,
      "grad_norm": 10.16787052154541,
      "learning_rate": 2.510156948021184e-06,
      "loss": 3.8233,
      "step": 49075
    },
    {
      "epoch": 0.499267578125,
      "grad_norm": 10.422018051147461,
      "learning_rate": 2.5097570702497803e-06,
      "loss": 3.1277,
      "step": 49080
    },
    {
      "epoch": 0.4993184407552083,
      "grad_norm": 13.755646705627441,
      "learning_rate": 2.5093571922287444e-06,
      "loss": 4.0196,
      "step": 49085
    },
    {
      "epoch": 0.4993693033854167,
      "grad_norm": 10.642024040222168,
      "learning_rate": 2.5089573139683082e-06,
      "loss": 3.3442,
      "step": 49090
    },
    {
      "epoch": 0.499420166015625,
      "grad_norm": 11.056506156921387,
      "learning_rate": 2.5085574354786997e-06,
      "loss": 3.3985,
      "step": 49095
    },
    {
      "epoch": 0.4994710286458333,
      "grad_norm": 13.28608512878418,
      "learning_rate": 2.508157556770152e-06,
      "loss": 3.1175,
      "step": 49100
    },
    {
      "epoch": 0.4995218912760417,
      "grad_norm": 13.478775024414062,
      "learning_rate": 2.5077576778528945e-06,
      "loss": 3.2193,
      "step": 49105
    },
    {
      "epoch": 0.49957275390625,
      "grad_norm": 14.44498062133789,
      "learning_rate": 2.5073577987371587e-06,
      "loss": 3.6488,
      "step": 49110
    },
    {
      "epoch": 0.4996236165364583,
      "grad_norm": 11.836501121520996,
      "learning_rate": 2.5069579194331763e-06,
      "loss": 3.1957,
      "step": 49115
    },
    {
      "epoch": 0.4996744791666667,
      "grad_norm": 11.179645538330078,
      "learning_rate": 2.506558039951176e-06,
      "loss": 3.1457,
      "step": 49120
    },
    {
      "epoch": 0.499725341796875,
      "grad_norm": 11.630443572998047,
      "learning_rate": 2.5061581603013905e-06,
      "loss": 3.6474,
      "step": 49125
    },
    {
      "epoch": 0.4997762044270833,
      "grad_norm": 14.907173156738281,
      "learning_rate": 2.50575828049405e-06,
      "loss": 3.2522,
      "step": 49130
    },
    {
      "epoch": 0.4998270670572917,
      "grad_norm": 17.425331115722656,
      "learning_rate": 2.505358400539385e-06,
      "loss": 3.0893,
      "step": 49135
    },
    {
      "epoch": 0.4998779296875,
      "grad_norm": 13.06432819366455,
      "learning_rate": 2.504958520447627e-06,
      "loss": 3.0929,
      "step": 49140
    },
    {
      "epoch": 0.4999287923177083,
      "grad_norm": 9.655762672424316,
      "learning_rate": 2.5045586402290055e-06,
      "loss": 3.2749,
      "step": 49145
    },
    {
      "epoch": 0.4999796549479167,
      "grad_norm": 8.963852882385254,
      "learning_rate": 2.504158759893753e-06,
      "loss": 3.7212,
      "step": 49150
    },
    {
      "epoch": 0.5,
      "eval_loss": 3.3498637676239014,
      "eval_runtime": 156.6867,
      "eval_samples_per_second": 12.809,
      "eval_steps_per_second": 12.809,
      "step": 49152
    },
    {
      "epoch": 0.500030517578125,
      "grad_norm": 9.196493148803711,
      "learning_rate": 2.5037588794521e-06,
      "loss": 3.5239,
      "step": 49155
    },
    {
      "epoch": 0.5000813802083334,
      "grad_norm": 8.475089073181152,
      "learning_rate": 2.503358998914276e-06,
      "loss": 3.4656,
      "step": 49160
    },
    {
      "epoch": 0.5001322428385416,
      "grad_norm": 16.432350158691406,
      "learning_rate": 2.502959118290514e-06,
      "loss": 3.1871,
      "step": 49165
    },
    {
      "epoch": 0.50018310546875,
      "grad_norm": 15.631834030151367,
      "learning_rate": 2.5025592375910416e-06,
      "loss": 3.5826,
      "step": 49170
    },
    {
      "epoch": 0.5002339680989584,
      "grad_norm": 15.452563285827637,
      "learning_rate": 2.5021593568260938e-06,
      "loss": 3.5809,
      "step": 49175
    },
    {
      "epoch": 0.5002848307291666,
      "grad_norm": 12.592869758605957,
      "learning_rate": 2.501759476005898e-06,
      "loss": 3.5671,
      "step": 49180
    },
    {
      "epoch": 0.500335693359375,
      "grad_norm": 13.242794036865234,
      "learning_rate": 2.501359595140688e-06,
      "loss": 3.2827,
      "step": 49185
    },
    {
      "epoch": 0.5003865559895834,
      "grad_norm": 13.345170021057129,
      "learning_rate": 2.5009597142406917e-06,
      "loss": 3.2537,
      "step": 49190
    },
    {
      "epoch": 0.5004374186197916,
      "grad_norm": 11.155458450317383,
      "learning_rate": 2.500559833316141e-06,
      "loss": 3.0264,
      "step": 49195
    },
    {
      "epoch": 0.50048828125,
      "grad_norm": 11.095985412597656,
      "learning_rate": 2.5001599523772687e-06,
      "loss": 3.3933,
      "step": 49200
    },
    {
      "epoch": 0.5005391438802084,
      "grad_norm": 10.589454650878906,
      "learning_rate": 2.499760071434303e-06,
      "loss": 3.6001,
      "step": 49205
    },
    {
      "epoch": 0.5005900065104166,
      "grad_norm": 7.216246604919434,
      "learning_rate": 2.4993601904974755e-06,
      "loss": 3.1535,
      "step": 49210
    },
    {
      "epoch": 0.500640869140625,
      "grad_norm": 9.691969871520996,
      "learning_rate": 2.4989603095770184e-06,
      "loss": 3.4616,
      "step": 49215
    },
    {
      "epoch": 0.5006917317708334,
      "grad_norm": 13.116881370544434,
      "learning_rate": 2.4985604286831606e-06,
      "loss": 3.1803,
      "step": 49220
    },
    {
      "epoch": 0.5007425944010416,
      "grad_norm": 7.126953125,
      "learning_rate": 2.4981605478261343e-06,
      "loss": 3.4157,
      "step": 49225
    },
    {
      "epoch": 0.50079345703125,
      "grad_norm": 9.899238586425781,
      "learning_rate": 2.49776066701617e-06,
      "loss": 3.3096,
      "step": 49230
    },
    {
      "epoch": 0.5008443196614584,
      "grad_norm": 11.79799747467041,
      "learning_rate": 2.4973607862634987e-06,
      "loss": 3.4077,
      "step": 49235
    },
    {
      "epoch": 0.5008951822916666,
      "grad_norm": 7.4908246994018555,
      "learning_rate": 2.4969609055783502e-06,
      "loss": 3.031,
      "step": 49240
    },
    {
      "epoch": 0.500946044921875,
      "grad_norm": 15.956377029418945,
      "learning_rate": 2.4965610249709573e-06,
      "loss": 3.8845,
      "step": 49245
    },
    {
      "epoch": 0.5009969075520834,
      "grad_norm": 16.496440887451172,
      "learning_rate": 2.4961611444515495e-06,
      "loss": 3.5862,
      "step": 49250
    },
    {
      "epoch": 0.5010477701822916,
      "grad_norm": 15.384745597839355,
      "learning_rate": 2.495761264030358e-06,
      "loss": 3.2148,
      "step": 49255
    },
    {
      "epoch": 0.5010986328125,
      "grad_norm": 8.837871551513672,
      "learning_rate": 2.4953613837176127e-06,
      "loss": 3.1969,
      "step": 49260
    },
    {
      "epoch": 0.5011494954427084,
      "grad_norm": 9.887128829956055,
      "learning_rate": 2.494961503523546e-06,
      "loss": 3.1775,
      "step": 49265
    },
    {
      "epoch": 0.5012003580729166,
      "grad_norm": 12.647445678710938,
      "learning_rate": 2.494561623458388e-06,
      "loss": 3.177,
      "step": 49270
    },
    {
      "epoch": 0.501251220703125,
      "grad_norm": 9.768755912780762,
      "learning_rate": 2.4941617435323695e-06,
      "loss": 3.2917,
      "step": 49275
    },
    {
      "epoch": 0.5013020833333334,
      "grad_norm": 13.205989837646484,
      "learning_rate": 2.4937618637557207e-06,
      "loss": 3.0655,
      "step": 49280
    },
    {
      "epoch": 0.5013529459635416,
      "grad_norm": 11.150520324707031,
      "learning_rate": 2.4933619841386744e-06,
      "loss": 3.437,
      "step": 49285
    },
    {
      "epoch": 0.50140380859375,
      "grad_norm": 8.906401634216309,
      "learning_rate": 2.4929621046914595e-06,
      "loss": 3.4999,
      "step": 49290
    },
    {
      "epoch": 0.5014546712239584,
      "grad_norm": 9.648660659790039,
      "learning_rate": 2.4925622254243078e-06,
      "loss": 3.5996,
      "step": 49295
    },
    {
      "epoch": 0.5015055338541666,
      "grad_norm": 7.539660453796387,
      "learning_rate": 2.492162346347449e-06,
      "loss": 3.0006,
      "step": 49300
    },
    {
      "epoch": 0.501556396484375,
      "grad_norm": 12.649576187133789,
      "learning_rate": 2.4917624674711154e-06,
      "loss": 3.2997,
      "step": 49305
    },
    {
      "epoch": 0.5016072591145834,
      "grad_norm": 12.197525024414062,
      "learning_rate": 2.4913625888055374e-06,
      "loss": 3.3524,
      "step": 49310
    },
    {
      "epoch": 0.5016581217447916,
      "grad_norm": 13.816476821899414,
      "learning_rate": 2.490962710360944e-06,
      "loss": 3.1124,
      "step": 49315
    },
    {
      "epoch": 0.501708984375,
      "grad_norm": 8.340036392211914,
      "learning_rate": 2.490562832147569e-06,
      "loss": 3.5732,
      "step": 49320
    },
    {
      "epoch": 0.5017598470052084,
      "grad_norm": 7.469123363494873,
      "learning_rate": 2.490162954175642e-06,
      "loss": 3.2823,
      "step": 49325
    },
    {
      "epoch": 0.5018107096354166,
      "grad_norm": 8.699426651000977,
      "learning_rate": 2.4897630764553926e-06,
      "loss": 3.4334,
      "step": 49330
    },
    {
      "epoch": 0.501861572265625,
      "grad_norm": 15.343707084655762,
      "learning_rate": 2.489363198997052e-06,
      "loss": 3.2377,
      "step": 49335
    },
    {
      "epoch": 0.5019124348958334,
      "grad_norm": 9.360434532165527,
      "learning_rate": 2.4889633218108524e-06,
      "loss": 3.5039,
      "step": 49340
    },
    {
      "epoch": 0.5019632975260416,
      "grad_norm": 13.143653869628906,
      "learning_rate": 2.4885634449070234e-06,
      "loss": 3.4251,
      "step": 49345
    },
    {
      "epoch": 0.50201416015625,
      "grad_norm": 13.156907081604004,
      "learning_rate": 2.488163568295796e-06,
      "loss": 3.1726,
      "step": 49350
    },
    {
      "epoch": 0.5020650227864584,
      "grad_norm": 7.87513542175293,
      "learning_rate": 2.4877636919873997e-06,
      "loss": 3.2848,
      "step": 49355
    },
    {
      "epoch": 0.5021158854166666,
      "grad_norm": 9.70824146270752,
      "learning_rate": 2.487363815992068e-06,
      "loss": 3.5863,
      "step": 49360
    },
    {
      "epoch": 0.502166748046875,
      "grad_norm": 10.080727577209473,
      "learning_rate": 2.4869639403200295e-06,
      "loss": 3.5781,
      "step": 49365
    },
    {
      "epoch": 0.5022176106770834,
      "grad_norm": 7.460448265075684,
      "learning_rate": 2.486564064981516e-06,
      "loss": 3.4143,
      "step": 49370
    },
    {
      "epoch": 0.5022684733072916,
      "grad_norm": 11.609356880187988,
      "learning_rate": 2.486164189986757e-06,
      "loss": 3.2316,
      "step": 49375
    },
    {
      "epoch": 0.5023193359375,
      "grad_norm": 9.654780387878418,
      "learning_rate": 2.485764315345984e-06,
      "loss": 3.3201,
      "step": 49380
    },
    {
      "epoch": 0.5023701985677084,
      "grad_norm": 11.209192276000977,
      "learning_rate": 2.4853644410694284e-06,
      "loss": 3.371,
      "step": 49385
    },
    {
      "epoch": 0.5024210611979166,
      "grad_norm": 12.05045223236084,
      "learning_rate": 2.4849645671673193e-06,
      "loss": 3.4816,
      "step": 49390
    },
    {
      "epoch": 0.502471923828125,
      "grad_norm": 11.425850868225098,
      "learning_rate": 2.4845646936498878e-06,
      "loss": 3.9463,
      "step": 49395
    },
    {
      "epoch": 0.5025227864583334,
      "grad_norm": 8.453120231628418,
      "learning_rate": 2.4841648205273663e-06,
      "loss": 3.5582,
      "step": 49400
    },
    {
      "epoch": 0.5025736490885416,
      "grad_norm": 8.841841697692871,
      "learning_rate": 2.483764947809983e-06,
      "loss": 3.4371,
      "step": 49405
    },
    {
      "epoch": 0.50262451171875,
      "grad_norm": 7.723162651062012,
      "learning_rate": 2.4833650755079693e-06,
      "loss": 3.3702,
      "step": 49410
    },
    {
      "epoch": 0.5026753743489584,
      "grad_norm": 10.669869422912598,
      "learning_rate": 2.482965203631558e-06,
      "loss": 3.3359,
      "step": 49415
    },
    {
      "epoch": 0.5027262369791666,
      "grad_norm": 8.969093322753906,
      "learning_rate": 2.4825653321909765e-06,
      "loss": 3.2873,
      "step": 49420
    },
    {
      "epoch": 0.502777099609375,
      "grad_norm": 17.20841407775879,
      "learning_rate": 2.4821654611964575e-06,
      "loss": 3.6631,
      "step": 49425
    },
    {
      "epoch": 0.5028279622395834,
      "grad_norm": 15.726045608520508,
      "learning_rate": 2.48176559065823e-06,
      "loss": 2.9552,
      "step": 49430
    },
    {
      "epoch": 0.5028788248697916,
      "grad_norm": 8.223189353942871,
      "learning_rate": 2.4813657205865265e-06,
      "loss": 3.1761,
      "step": 49435
    },
    {
      "epoch": 0.5029296875,
      "grad_norm": 8.99612808227539,
      "learning_rate": 2.4809658509915766e-06,
      "loss": 3.4041,
      "step": 49440
    },
    {
      "epoch": 0.5029805501302084,
      "grad_norm": 11.7155122756958,
      "learning_rate": 2.4805659818836105e-06,
      "loss": 3.3462,
      "step": 49445
    },
    {
      "epoch": 0.5030314127604166,
      "grad_norm": 16.708206176757812,
      "learning_rate": 2.4801661132728587e-06,
      "loss": 3.405,
      "step": 49450
    },
    {
      "epoch": 0.503082275390625,
      "grad_norm": 7.648775100708008,
      "learning_rate": 2.479766245169553e-06,
      "loss": 3.6026,
      "step": 49455
    },
    {
      "epoch": 0.5031331380208334,
      "grad_norm": 16.03279685974121,
      "learning_rate": 2.479366377583923e-06,
      "loss": 3.4089,
      "step": 49460
    },
    {
      "epoch": 0.5031840006510416,
      "grad_norm": 8.516480445861816,
      "learning_rate": 2.4789665105261995e-06,
      "loss": 3.0617,
      "step": 49465
    },
    {
      "epoch": 0.50323486328125,
      "grad_norm": 15.769489288330078,
      "learning_rate": 2.478566644006612e-06,
      "loss": 3.3682,
      "step": 49470
    },
    {
      "epoch": 0.5032857259114584,
      "grad_norm": 9.968362808227539,
      "learning_rate": 2.4781667780353922e-06,
      "loss": 3.2163,
      "step": 49475
    },
    {
      "epoch": 0.5033365885416666,
      "grad_norm": 10.482285499572754,
      "learning_rate": 2.477766912622771e-06,
      "loss": 3.2009,
      "step": 49480
    },
    {
      "epoch": 0.503387451171875,
      "grad_norm": 13.688324928283691,
      "learning_rate": 2.4773670477789774e-06,
      "loss": 3.3076,
      "step": 49485
    },
    {
      "epoch": 0.5034383138020834,
      "grad_norm": 11.150995254516602,
      "learning_rate": 2.476967183514242e-06,
      "loss": 3.38,
      "step": 49490
    },
    {
      "epoch": 0.5034891764322916,
      "grad_norm": 13.235206604003906,
      "learning_rate": 2.4765673198387963e-06,
      "loss": 3.1943,
      "step": 49495
    },
    {
      "epoch": 0.5035400390625,
      "grad_norm": 12.570960998535156,
      "learning_rate": 2.4761674567628707e-06,
      "loss": 3.5157,
      "step": 49500
    },
    {
      "epoch": 0.5035909016927084,
      "grad_norm": 9.195624351501465,
      "learning_rate": 2.4757675942966937e-06,
      "loss": 3.2672,
      "step": 49505
    },
    {
      "epoch": 0.5036417643229166,
      "grad_norm": 8.501916885375977,
      "learning_rate": 2.4753677324504978e-06,
      "loss": 3.6237,
      "step": 49510
    },
    {
      "epoch": 0.503692626953125,
      "grad_norm": 11.110610961914062,
      "learning_rate": 2.4749678712345134e-06,
      "loss": 3.9753,
      "step": 49515
    },
    {
      "epoch": 0.5037434895833334,
      "grad_norm": 12.429216384887695,
      "learning_rate": 2.4745680106589694e-06,
      "loss": 3.4606,
      "step": 49520
    },
    {
      "epoch": 0.5037943522135416,
      "grad_norm": 12.383891105651855,
      "learning_rate": 2.474168150734096e-06,
      "loss": 3.1723,
      "step": 49525
    },
    {
      "epoch": 0.50384521484375,
      "grad_norm": 14.52224349975586,
      "learning_rate": 2.473768291470126e-06,
      "loss": 3.3007,
      "step": 49530
    },
    {
      "epoch": 0.5038960774739584,
      "grad_norm": 13.140124320983887,
      "learning_rate": 2.473368432877287e-06,
      "loss": 3.3674,
      "step": 49535
    },
    {
      "epoch": 0.5039469401041666,
      "grad_norm": 297.5426330566406,
      "learning_rate": 2.472968574965811e-06,
      "loss": 3.9067,
      "step": 49540
    },
    {
      "epoch": 0.503997802734375,
      "grad_norm": 13.380870819091797,
      "learning_rate": 2.4725687177459266e-06,
      "loss": 3.3042,
      "step": 49545
    },
    {
      "epoch": 0.5040486653645834,
      "grad_norm": 12.111773490905762,
      "learning_rate": 2.4721688612278657e-06,
      "loss": 3.2508,
      "step": 49550
    },
    {
      "epoch": 0.5040995279947916,
      "grad_norm": 8.741081237792969,
      "learning_rate": 2.4717690054218583e-06,
      "loss": 3.265,
      "step": 49555
    },
    {
      "epoch": 0.504150390625,
      "grad_norm": 11.820341110229492,
      "learning_rate": 2.4713691503381333e-06,
      "loss": 3.3668,
      "step": 49560
    },
    {
      "epoch": 0.5042012532552084,
      "grad_norm": 10.02950668334961,
      "learning_rate": 2.470969295986922e-06,
      "loss": 3.0367,
      "step": 49565
    },
    {
      "epoch": 0.5042521158854166,
      "grad_norm": 10.921381950378418,
      "learning_rate": 2.4705694423784548e-06,
      "loss": 3.1557,
      "step": 49570
    },
    {
      "epoch": 0.504302978515625,
      "grad_norm": 9.535286903381348,
      "learning_rate": 2.4701695895229613e-06,
      "loss": 3.1932,
      "step": 49575
    },
    {
      "epoch": 0.5043538411458334,
      "grad_norm": 13.039703369140625,
      "learning_rate": 2.469769737430672e-06,
      "loss": 3.4789,
      "step": 49580
    },
    {
      "epoch": 0.5044047037760416,
      "grad_norm": 9.973648071289062,
      "learning_rate": 2.4693698861118162e-06,
      "loss": 3.3607,
      "step": 49585
    },
    {
      "epoch": 0.50445556640625,
      "grad_norm": 16.38591194152832,
      "learning_rate": 2.4689700355766247e-06,
      "loss": 3.0819,
      "step": 49590
    },
    {
      "epoch": 0.5045064290364584,
      "grad_norm": 15.450492858886719,
      "learning_rate": 2.4685701858353283e-06,
      "loss": 3.5412,
      "step": 49595
    },
    {
      "epoch": 0.5045572916666666,
      "grad_norm": 16.321855545043945,
      "learning_rate": 2.468170336898155e-06,
      "loss": 3.1058,
      "step": 49600
    },
    {
      "epoch": 0.504608154296875,
      "grad_norm": 11.17414379119873,
      "learning_rate": 2.4677704887753367e-06,
      "loss": 3.4829,
      "step": 49605
    },
    {
      "epoch": 0.5046590169270834,
      "grad_norm": 14.2786865234375,
      "learning_rate": 2.4673706414771036e-06,
      "loss": 3.3495,
      "step": 49610
    },
    {
      "epoch": 0.5047098795572916,
      "grad_norm": 10.139547348022461,
      "learning_rate": 2.4669707950136837e-06,
      "loss": 3.4174,
      "step": 49615
    },
    {
      "epoch": 0.5047607421875,
      "grad_norm": 7.030655384063721,
      "learning_rate": 2.466570949395308e-06,
      "loss": 3.5727,
      "step": 49620
    },
    {
      "epoch": 0.5048116048177084,
      "grad_norm": 13.815516471862793,
      "learning_rate": 2.466171104632208e-06,
      "loss": 3.296,
      "step": 49625
    },
    {
      "epoch": 0.5048624674479166,
      "grad_norm": 10.212800025939941,
      "learning_rate": 2.4657712607346115e-06,
      "loss": 3.2526,
      "step": 49630
    },
    {
      "epoch": 0.504913330078125,
      "grad_norm": 13.8051176071167,
      "learning_rate": 2.4653714177127495e-06,
      "loss": 3.8579,
      "step": 49635
    },
    {
      "epoch": 0.5049641927083334,
      "grad_norm": 10.22461986541748,
      "learning_rate": 2.4649715755768502e-06,
      "loss": 3.1761,
      "step": 49640
    },
    {
      "epoch": 0.5050150553385416,
      "grad_norm": 9.084115982055664,
      "learning_rate": 2.4645717343371465e-06,
      "loss": 3.2978,
      "step": 49645
    },
    {
      "epoch": 0.50506591796875,
      "grad_norm": 12.154102325439453,
      "learning_rate": 2.464171894003866e-06,
      "loss": 3.3534,
      "step": 49650
    },
    {
      "epoch": 0.5051167805989584,
      "grad_norm": 16.16970443725586,
      "learning_rate": 2.463772054587239e-06,
      "loss": 3.3953,
      "step": 49655
    },
    {
      "epoch": 0.5051676432291666,
      "grad_norm": 12.5291109085083,
      "learning_rate": 2.463372216097495e-06,
      "loss": 3.4107,
      "step": 49660
    },
    {
      "epoch": 0.505218505859375,
      "grad_norm": 8.317669868469238,
      "learning_rate": 2.462972378544865e-06,
      "loss": 3.3117,
      "step": 49665
    },
    {
      "epoch": 0.5052693684895834,
      "grad_norm": 12.62435245513916,
      "learning_rate": 2.462572541939578e-06,
      "loss": 2.8869,
      "step": 49670
    },
    {
      "epoch": 0.5053202311197916,
      "grad_norm": 9.405922889709473,
      "learning_rate": 2.462172706291863e-06,
      "loss": 3.1812,
      "step": 49675
    },
    {
      "epoch": 0.50537109375,
      "grad_norm": 15.787854194641113,
      "learning_rate": 2.4617728716119502e-06,
      "loss": 3.0783,
      "step": 49680
    },
    {
      "epoch": 0.5054219563802084,
      "grad_norm": 7.9198102951049805,
      "learning_rate": 2.46137303791007e-06,
      "loss": 3.4253,
      "step": 49685
    },
    {
      "epoch": 0.5054728190104166,
      "grad_norm": 13.76949405670166,
      "learning_rate": 2.4609732051964514e-06,
      "loss": 3.4277,
      "step": 49690
    },
    {
      "epoch": 0.505523681640625,
      "grad_norm": 10.015167236328125,
      "learning_rate": 2.460573373481323e-06,
      "loss": 3.4706,
      "step": 49695
    },
    {
      "epoch": 0.5055745442708334,
      "grad_norm": 11.512467384338379,
      "learning_rate": 2.460173542774917e-06,
      "loss": 3.4938,
      "step": 49700
    },
    {
      "epoch": 0.5056254069010416,
      "grad_norm": 10.98910903930664,
      "learning_rate": 2.4597737130874612e-06,
      "loss": 3.3244,
      "step": 49705
    },
    {
      "epoch": 0.50567626953125,
      "grad_norm": 10.6687650680542,
      "learning_rate": 2.4593738844291855e-06,
      "loss": 3.7278,
      "step": 49710
    },
    {
      "epoch": 0.5057271321614584,
      "grad_norm": 8.300183296203613,
      "learning_rate": 2.4589740568103183e-06,
      "loss": 3.469,
      "step": 49715
    },
    {
      "epoch": 0.5057779947916666,
      "grad_norm": 10.843482971191406,
      "learning_rate": 2.458574230241091e-06,
      "loss": 3.217,
      "step": 49720
    },
    {
      "epoch": 0.505828857421875,
      "grad_norm": 14.605001449584961,
      "learning_rate": 2.458174404731732e-06,
      "loss": 3.3045,
      "step": 49725
    },
    {
      "epoch": 0.5058797200520834,
      "grad_norm": 10.476160049438477,
      "learning_rate": 2.4577745802924713e-06,
      "loss": 3.3574,
      "step": 49730
    },
    {
      "epoch": 0.5059305826822916,
      "grad_norm": 9.061910629272461,
      "learning_rate": 2.457374756933537e-06,
      "loss": 3.0889,
      "step": 49735
    },
    {
      "epoch": 0.5059814453125,
      "grad_norm": 13.6966552734375,
      "learning_rate": 2.4569749346651604e-06,
      "loss": 5.5981,
      "step": 49740
    },
    {
      "epoch": 0.5060323079427084,
      "grad_norm": 11.65805435180664,
      "learning_rate": 2.4565751134975693e-06,
      "loss": 3.4213,
      "step": 49745
    },
    {
      "epoch": 0.5060831705729166,
      "grad_norm": 8.059772491455078,
      "learning_rate": 2.4561752934409945e-06,
      "loss": 3.1742,
      "step": 49750
    },
    {
      "epoch": 0.506134033203125,
      "grad_norm": 8.113338470458984,
      "learning_rate": 2.4557754745056635e-06,
      "loss": 3.5701,
      "step": 49755
    },
    {
      "epoch": 0.5061848958333334,
      "grad_norm": 8.002399444580078,
      "learning_rate": 2.455375656701806e-06,
      "loss": 3.16,
      "step": 49760
    },
    {
      "epoch": 0.5062357584635416,
      "grad_norm": 13.288044929504395,
      "learning_rate": 2.4549758400396532e-06,
      "loss": 3.2662,
      "step": 49765
    },
    {
      "epoch": 0.50628662109375,
      "grad_norm": 14.36279296875,
      "learning_rate": 2.4545760245294325e-06,
      "loss": 3.2286,
      "step": 49770
    },
    {
      "epoch": 0.5063374837239584,
      "grad_norm": 13.100310325622559,
      "learning_rate": 2.4541762101813722e-06,
      "loss": 3.6597,
      "step": 49775
    },
    {
      "epoch": 0.5063883463541666,
      "grad_norm": 15.703715324401855,
      "learning_rate": 2.4537763970057043e-06,
      "loss": 3.2327,
      "step": 49780
    },
    {
      "epoch": 0.506439208984375,
      "grad_norm": 13.610279083251953,
      "learning_rate": 2.4533765850126557e-06,
      "loss": 3.4582,
      "step": 49785
    },
    {
      "epoch": 0.5064900716145834,
      "grad_norm": 15.596902847290039,
      "learning_rate": 2.4529767742124565e-06,
      "loss": 3.2701,
      "step": 49790
    },
    {
      "epoch": 0.5065409342447916,
      "grad_norm": 14.749031066894531,
      "learning_rate": 2.4525769646153338e-06,
      "loss": 3.1786,
      "step": 49795
    },
    {
      "epoch": 0.506591796875,
      "grad_norm": 12.403141975402832,
      "learning_rate": 2.45217715623152e-06,
      "loss": 3.2539,
      "step": 49800
    },
    {
      "epoch": 0.5066426595052084,
      "grad_norm": 13.405444145202637,
      "learning_rate": 2.4517773490712417e-06,
      "loss": 3.5059,
      "step": 49805
    },
    {
      "epoch": 0.5066935221354166,
      "grad_norm": 9.99687385559082,
      "learning_rate": 2.4513775431447278e-06,
      "loss": 3.0976,
      "step": 49810
    },
    {
      "epoch": 0.506744384765625,
      "grad_norm": 14.107109069824219,
      "learning_rate": 2.450977738462209e-06,
      "loss": 3.3232,
      "step": 49815
    },
    {
      "epoch": 0.5067952473958334,
      "grad_norm": 11.486820220947266,
      "learning_rate": 2.450577935033913e-06,
      "loss": 3.5945,
      "step": 49820
    },
    {
      "epoch": 0.5068461100260416,
      "grad_norm": 8.772841453552246,
      "learning_rate": 2.450178132870069e-06,
      "loss": 3.1173,
      "step": 49825
    },
    {
      "epoch": 0.50689697265625,
      "grad_norm": 11.41469669342041,
      "learning_rate": 2.449778331980905e-06,
      "loss": 3.1338,
      "step": 49830
    },
    {
      "epoch": 0.5069478352864584,
      "grad_norm": 14.06910228729248,
      "learning_rate": 2.4493785323766504e-06,
      "loss": 3.4256,
      "step": 49835
    },
    {
      "epoch": 0.5069986979166666,
      "grad_norm": 10.561040878295898,
      "learning_rate": 2.448978734067535e-06,
      "loss": 3.2773,
      "step": 49840
    },
    {
      "epoch": 0.507049560546875,
      "grad_norm": 13.252824783325195,
      "learning_rate": 2.4485789370637857e-06,
      "loss": 3.2069,
      "step": 49845
    },
    {
      "epoch": 0.5071004231770834,
      "grad_norm": 15.160521507263184,
      "learning_rate": 2.448179141375632e-06,
      "loss": 2.8845,
      "step": 49850
    },
    {
      "epoch": 0.5071512858072916,
      "grad_norm": 14.707564353942871,
      "learning_rate": 2.4477793470133033e-06,
      "loss": 3.173,
      "step": 49855
    },
    {
      "epoch": 0.5072021484375,
      "grad_norm": 12.910215377807617,
      "learning_rate": 2.4473795539870275e-06,
      "loss": 3.2612,
      "step": 49860
    },
    {
      "epoch": 0.5072530110677084,
      "grad_norm": 12.413802146911621,
      "learning_rate": 2.446979762307034e-06,
      "loss": 3.0549,
      "step": 49865
    },
    {
      "epoch": 0.5073038736979166,
      "grad_norm": 14.505427360534668,
      "learning_rate": 2.446579971983549e-06,
      "loss": 3.0984,
      "step": 49870
    },
    {
      "epoch": 0.507354736328125,
      "grad_norm": 16.360313415527344,
      "learning_rate": 2.446180183026804e-06,
      "loss": 3.2245,
      "step": 49875
    },
    {
      "epoch": 0.5074055989583334,
      "grad_norm": 12.215571403503418,
      "learning_rate": 2.4457803954470264e-06,
      "loss": 3.4219,
      "step": 49880
    },
    {
      "epoch": 0.5074564615885416,
      "grad_norm": 9.959635734558105,
      "learning_rate": 2.4453806092544442e-06,
      "loss": 3.2679,
      "step": 49885
    },
    {
      "epoch": 0.50750732421875,
      "grad_norm": 11.354060173034668,
      "learning_rate": 2.444980824459285e-06,
      "loss": 3.3806,
      "step": 49890
    },
    {
      "epoch": 0.5075581868489584,
      "grad_norm": 12.338563919067383,
      "learning_rate": 2.44458104107178e-06,
      "loss": 3.0743,
      "step": 49895
    },
    {
      "epoch": 0.5076090494791666,
      "grad_norm": 12.6554594039917,
      "learning_rate": 2.4441812591021553e-06,
      "loss": 3.8411,
      "step": 49900
    },
    {
      "epoch": 0.507659912109375,
      "grad_norm": 14.3016939163208,
      "learning_rate": 2.4437814785606394e-06,
      "loss": 3.4705,
      "step": 49905
    },
    {
      "epoch": 0.5077107747395834,
      "grad_norm": 10.152817726135254,
      "learning_rate": 2.443381699457462e-06,
      "loss": 3.1855,
      "step": 49910
    },
    {
      "epoch": 0.5077616373697916,
      "grad_norm": 14.350810050964355,
      "learning_rate": 2.44298192180285e-06,
      "loss": 3.6002,
      "step": 49915
    },
    {
      "epoch": 0.5078125,
      "grad_norm": 13.546785354614258,
      "learning_rate": 2.4425821456070326e-06,
      "loss": 3.6468,
      "step": 49920
    },
    {
      "epoch": 0.5078633626302084,
      "grad_norm": 7.947472095489502,
      "learning_rate": 2.442182370880236e-06,
      "loss": 3.3876,
      "step": 49925
    },
    {
      "epoch": 0.5079142252604166,
      "grad_norm": 9.699104309082031,
      "learning_rate": 2.4417825976326908e-06,
      "loss": 3.4474,
      "step": 49930
    },
    {
      "epoch": 0.507965087890625,
      "grad_norm": 11.072478294372559,
      "learning_rate": 2.441382825874624e-06,
      "loss": 3.6583,
      "step": 49935
    },
    {
      "epoch": 0.5080159505208334,
      "grad_norm": 11.983139038085938,
      "learning_rate": 2.4409830556162635e-06,
      "loss": 3.1169,
      "step": 49940
    },
    {
      "epoch": 0.5080668131510416,
      "grad_norm": 8.136286735534668,
      "learning_rate": 2.440583286867837e-06,
      "loss": 3.1332,
      "step": 49945
    },
    {
      "epoch": 0.50811767578125,
      "grad_norm": 13.545182228088379,
      "learning_rate": 2.440183519639573e-06,
      "loss": 3.0072,
      "step": 49950
    },
    {
      "epoch": 0.5081685384114584,
      "grad_norm": 16.493642807006836,
      "learning_rate": 2.439783753941701e-06,
      "loss": 3.238,
      "step": 49955
    },
    {
      "epoch": 0.5082194010416666,
      "grad_norm": 16.028322219848633,
      "learning_rate": 2.4393839897844463e-06,
      "loss": 3.2214,
      "step": 49960
    },
    {
      "epoch": 0.508270263671875,
      "grad_norm": 13.38894271850586,
      "learning_rate": 2.438984227178037e-06,
      "loss": 3.4751,
      "step": 49965
    },
    {
      "epoch": 0.5083211263020834,
      "grad_norm": 9.154212951660156,
      "learning_rate": 2.4385844661327034e-06,
      "loss": 3.4803,
      "step": 49970
    },
    {
      "epoch": 0.5083719889322916,
      "grad_norm": 12.187919616699219,
      "learning_rate": 2.438184706658671e-06,
      "loss": 3.3993,
      "step": 49975
    },
    {
      "epoch": 0.5084228515625,
      "grad_norm": 8.677459716796875,
      "learning_rate": 2.4377849487661686e-06,
      "loss": 3.4102,
      "step": 49980
    },
    {
      "epoch": 0.5084737141927084,
      "grad_norm": 15.56594181060791,
      "learning_rate": 2.437385192465423e-06,
      "loss": 3.432,
      "step": 49985
    },
    {
      "epoch": 0.5085245768229166,
      "grad_norm": 15.693167686462402,
      "learning_rate": 2.436985437766662e-06,
      "loss": 3.3972,
      "step": 49990
    },
    {
      "epoch": 0.508575439453125,
      "grad_norm": 17.262100219726562,
      "learning_rate": 2.4365856846801148e-06,
      "loss": 3.3424,
      "step": 49995
    },
    {
      "epoch": 0.5086263020833334,
      "grad_norm": 14.669502258300781,
      "learning_rate": 2.4361859332160063e-06,
      "loss": 3.4352,
      "step": 50000
    },
    {
      "epoch": 0.5086771647135416,
      "grad_norm": 13.700075149536133,
      "learning_rate": 2.4357861833845665e-06,
      "loss": 3.1324,
      "step": 50005
    },
    {
      "epoch": 0.50872802734375,
      "grad_norm": 13.19578742980957,
      "learning_rate": 2.4353864351960225e-06,
      "loss": 3.5918,
      "step": 50010
    },
    {
      "epoch": 0.5087788899739584,
      "grad_norm": 8.245698928833008,
      "learning_rate": 2.434986688660601e-06,
      "loss": 3.2106,
      "step": 50015
    },
    {
      "epoch": 0.5088297526041666,
      "grad_norm": 10.24449348449707,
      "learning_rate": 2.4345869437885285e-06,
      "loss": 3.3516,
      "step": 50020
    },
    {
      "epoch": 0.508880615234375,
      "grad_norm": 8.788253784179688,
      "learning_rate": 2.434187200590035e-06,
      "loss": 3.3597,
      "step": 50025
    },
    {
      "epoch": 0.5089314778645834,
      "grad_norm": 9.52047061920166,
      "learning_rate": 2.433787459075346e-06,
      "loss": 3.2025,
      "step": 50030
    },
    {
      "epoch": 0.5089823404947916,
      "grad_norm": 9.388339042663574,
      "learning_rate": 2.433387719254689e-06,
      "loss": 3.3501,
      "step": 50035
    },
    {
      "epoch": 0.509033203125,
      "grad_norm": 20.20309829711914,
      "learning_rate": 2.432987981138291e-06,
      "loss": 3.6578,
      "step": 50040
    },
    {
      "epoch": 0.5090840657552084,
      "grad_norm": 13.675063133239746,
      "learning_rate": 2.43258824473638e-06,
      "loss": 3.4585,
      "step": 50045
    },
    {
      "epoch": 0.5091349283854166,
      "grad_norm": 12.403952598571777,
      "learning_rate": 2.4321885100591833e-06,
      "loss": 3.2217,
      "step": 50050
    },
    {
      "epoch": 0.509185791015625,
      "grad_norm": 9.077116966247559,
      "learning_rate": 2.431788777116927e-06,
      "loss": 3.248,
      "step": 50055
    },
    {
      "epoch": 0.5092366536458334,
      "grad_norm": 10.033145904541016,
      "learning_rate": 2.431389045919838e-06,
      "loss": 3.2762,
      "step": 50060
    },
    {
      "epoch": 0.5092875162760416,
      "grad_norm": 9.409383773803711,
      "learning_rate": 2.430989316478145e-06,
      "loss": 3.191,
      "step": 50065
    },
    {
      "epoch": 0.50933837890625,
      "grad_norm": 9.033283233642578,
      "learning_rate": 2.4305895888020737e-06,
      "loss": 3.4433,
      "step": 50070
    },
    {
      "epoch": 0.5093892415364584,
      "grad_norm": 16.920303344726562,
      "learning_rate": 2.4301898629018516e-06,
      "loss": 3.1023,
      "step": 50075
    },
    {
      "epoch": 0.5094401041666666,
      "grad_norm": 13.073542594909668,
      "learning_rate": 2.4297901387877042e-06,
      "loss": 3.2759,
      "step": 50080
    },
    {
      "epoch": 0.509490966796875,
      "grad_norm": 7.634866237640381,
      "learning_rate": 2.4293904164698606e-06,
      "loss": 3.5395,
      "step": 50085
    },
    {
      "epoch": 0.5095418294270834,
      "grad_norm": 10.092267990112305,
      "learning_rate": 2.4289906959585463e-06,
      "loss": 3.4359,
      "step": 50090
    },
    {
      "epoch": 0.5095926920572916,
      "grad_norm": 12.581851959228516,
      "learning_rate": 2.428590977263987e-06,
      "loss": 3.5123,
      "step": 50095
    },
    {
      "epoch": 0.5096435546875,
      "grad_norm": 13.736969947814941,
      "learning_rate": 2.428191260396412e-06,
      "loss": 3.0601,
      "step": 50100
    },
    {
      "epoch": 0.5096944173177084,
      "grad_norm": 8.191214561462402,
      "learning_rate": 2.427791545366046e-06,
      "loss": 3.0694,
      "step": 50105
    },
    {
      "epoch": 0.5097452799479166,
      "grad_norm": 11.425572395324707,
      "learning_rate": 2.427391832183117e-06,
      "loss": 3.3201,
      "step": 50110
    },
    {
      "epoch": 0.509796142578125,
      "grad_norm": 6.909358024597168,
      "learning_rate": 2.4269921208578496e-06,
      "loss": 2.9903,
      "step": 50115
    },
    {
      "epoch": 0.5098470052083334,
      "grad_norm": 13.281767845153809,
      "learning_rate": 2.426592411400472e-06,
      "loss": 3.3728,
      "step": 50120
    },
    {
      "epoch": 0.5098978678385416,
      "grad_norm": 9.757806777954102,
      "learning_rate": 2.4261927038212104e-06,
      "loss": 3.9518,
      "step": 50125
    },
    {
      "epoch": 0.50994873046875,
      "grad_norm": 12.427955627441406,
      "learning_rate": 2.425792998130291e-06,
      "loss": 3.4458,
      "step": 50130
    },
    {
      "epoch": 0.5099995930989584,
      "grad_norm": 9.13679027557373,
      "learning_rate": 2.425393294337939e-06,
      "loss": 3.3046,
      "step": 50135
    },
    {
      "epoch": 0.5100504557291666,
      "grad_norm": 11.100677490234375,
      "learning_rate": 2.4249935924543833e-06,
      "loss": 3.5699,
      "step": 50140
    },
    {
      "epoch": 0.510101318359375,
      "grad_norm": 16.57285499572754,
      "learning_rate": 2.4245938924898484e-06,
      "loss": 3.4786,
      "step": 50145
    },
    {
      "epoch": 0.5101521809895834,
      "grad_norm": 13.360235214233398,
      "learning_rate": 2.424194194454561e-06,
      "loss": 3.0517,
      "step": 50150
    },
    {
      "epoch": 0.5102030436197916,
      "grad_norm": 12.516863822937012,
      "learning_rate": 2.4237944983587465e-06,
      "loss": 3.709,
      "step": 50155
    },
    {
      "epoch": 0.51025390625,
      "grad_norm": 12.673425674438477,
      "learning_rate": 2.423394804212632e-06,
      "loss": 3.3519,
      "step": 50160
    },
    {
      "epoch": 0.5103047688802084,
      "grad_norm": 16.423259735107422,
      "learning_rate": 2.422995112026444e-06,
      "loss": 3.4322,
      "step": 50165
    },
    {
      "epoch": 0.5103556315104166,
      "grad_norm": 15.741813659667969,
      "learning_rate": 2.422595421810407e-06,
      "loss": 3.4376,
      "step": 50170
    },
    {
      "epoch": 0.510406494140625,
      "grad_norm": 9.000436782836914,
      "learning_rate": 2.4221957335747474e-06,
      "loss": 3.5496,
      "step": 50175
    },
    {
      "epoch": 0.5104573567708334,
      "grad_norm": 11.855289459228516,
      "learning_rate": 2.4217960473296923e-06,
      "loss": 3.5276,
      "step": 50180
    },
    {
      "epoch": 0.5105082194010416,
      "grad_norm": 10.971668243408203,
      "learning_rate": 2.4213963630854667e-06,
      "loss": 3.2312,
      "step": 50185
    },
    {
      "epoch": 0.51055908203125,
      "grad_norm": 15.227359771728516,
      "learning_rate": 2.420996680852296e-06,
      "loss": 3.5039,
      "step": 50190
    },
    {
      "epoch": 0.5106099446614584,
      "grad_norm": 7.392257213592529,
      "learning_rate": 2.420597000640407e-06,
      "loss": 3.7844,
      "step": 50195
    },
    {
      "epoch": 0.5106608072916666,
      "grad_norm": 10.824780464172363,
      "learning_rate": 2.4201973224600252e-06,
      "loss": 3.2045,
      "step": 50200
    },
    {
      "epoch": 0.510711669921875,
      "grad_norm": 15.995607376098633,
      "learning_rate": 2.419797646321376e-06,
      "loss": 3.509,
      "step": 50205
    },
    {
      "epoch": 0.5107625325520834,
      "grad_norm": 10.681689262390137,
      "learning_rate": 2.419397972234684e-06,
      "loss": 3.3356,
      "step": 50210
    },
    {
      "epoch": 0.5108133951822916,
      "grad_norm": 13.102217674255371,
      "learning_rate": 2.4189983002101762e-06,
      "loss": 3.5439,
      "step": 50215
    },
    {
      "epoch": 0.5108642578125,
      "grad_norm": 8.509812355041504,
      "learning_rate": 2.418598630258078e-06,
      "loss": 3.1117,
      "step": 50220
    },
    {
      "epoch": 0.5109151204427084,
      "grad_norm": 14.604217529296875,
      "learning_rate": 2.4181989623886147e-06,
      "loss": 3.1855,
      "step": 50225
    },
    {
      "epoch": 0.5109659830729166,
      "grad_norm": 10.331098556518555,
      "learning_rate": 2.4177992966120105e-06,
      "loss": 3.0951,
      "step": 50230
    },
    {
      "epoch": 0.511016845703125,
      "grad_norm": 12.546608924865723,
      "learning_rate": 2.4173996329384933e-06,
      "loss": 3.2482,
      "step": 50235
    },
    {
      "epoch": 0.5110677083333334,
      "grad_norm": 12.26623249053955,
      "learning_rate": 2.416999971378286e-06,
      "loss": 3.5527,
      "step": 50240
    },
    {
      "epoch": 0.5111185709635416,
      "grad_norm": 13.634391784667969,
      "learning_rate": 2.4166003119416154e-06,
      "loss": 3.1796,
      "step": 50245
    },
    {
      "epoch": 0.51116943359375,
      "grad_norm": 12.512577056884766,
      "learning_rate": 2.4162006546387047e-06,
      "loss": 3.0403,
      "step": 50250
    },
    {
      "epoch": 0.5112202962239584,
      "grad_norm": 11.685742378234863,
      "learning_rate": 2.4158009994797816e-06,
      "loss": 3.6144,
      "step": 50255
    },
    {
      "epoch": 0.5112711588541666,
      "grad_norm": 7.155736446380615,
      "learning_rate": 2.4154013464750697e-06,
      "loss": 3.35,
      "step": 50260
    },
    {
      "epoch": 0.511322021484375,
      "grad_norm": 14.116155624389648,
      "learning_rate": 2.415001695634795e-06,
      "loss": 3.2603,
      "step": 50265
    },
    {
      "epoch": 0.5113728841145834,
      "grad_norm": 16.665386199951172,
      "learning_rate": 2.4146020469691803e-06,
      "loss": 3.3092,
      "step": 50270
    },
    {
      "epoch": 0.5114237467447916,
      "grad_norm": 8.797182083129883,
      "learning_rate": 2.4142024004884525e-06,
      "loss": 3.2307,
      "step": 50275
    },
    {
      "epoch": 0.511474609375,
      "grad_norm": 10.283164024353027,
      "learning_rate": 2.4138027562028368e-06,
      "loss": 2.9887,
      "step": 50280
    },
    {
      "epoch": 0.5115254720052084,
      "grad_norm": 16.19517707824707,
      "learning_rate": 2.4134031141225565e-06,
      "loss": 3.5166,
      "step": 50285
    },
    {
      "epoch": 0.5115763346354166,
      "grad_norm": 16.54936408996582,
      "learning_rate": 2.4130034742578365e-06,
      "loss": 3.17,
      "step": 50290
    },
    {
      "epoch": 0.511627197265625,
      "grad_norm": 17.40081024169922,
      "learning_rate": 2.412603836618903e-06,
      "loss": 3.3569,
      "step": 50295
    },
    {
      "epoch": 0.5116780598958334,
      "grad_norm": 8.195357322692871,
      "learning_rate": 2.4122042012159793e-06,
      "loss": 3.4386,
      "step": 50300
    },
    {
      "epoch": 0.5117289225260416,
      "grad_norm": 10.756447792053223,
      "learning_rate": 2.4118045680592893e-06,
      "loss": 3.425,
      "step": 50305
    },
    {
      "epoch": 0.51177978515625,
      "grad_norm": 10.567279815673828,
      "learning_rate": 2.41140493715906e-06,
      "loss": 3.1411,
      "step": 50310
    },
    {
      "epoch": 0.5118306477864584,
      "grad_norm": 10.493538856506348,
      "learning_rate": 2.4110053085255137e-06,
      "loss": 3.7064,
      "step": 50315
    },
    {
      "epoch": 0.5118815104166666,
      "grad_norm": 12.735881805419922,
      "learning_rate": 2.410605682168876e-06,
      "loss": 3.4582,
      "step": 50320
    },
    {
      "epoch": 0.511932373046875,
      "grad_norm": 14.975763320922852,
      "learning_rate": 2.4102060580993696e-06,
      "loss": 2.8556,
      "step": 50325
    },
    {
      "epoch": 0.5119832356770834,
      "grad_norm": 14.091262817382812,
      "learning_rate": 2.4098064363272205e-06,
      "loss": 3.2849,
      "step": 50330
    },
    {
      "epoch": 0.5120340983072916,
      "grad_norm": 15.417219161987305,
      "learning_rate": 2.409406816862653e-06,
      "loss": 3.2253,
      "step": 50335
    },
    {
      "epoch": 0.5120849609375,
      "grad_norm": 14.360461235046387,
      "learning_rate": 2.40900719971589e-06,
      "loss": 3.3464,
      "step": 50340
    },
    {
      "epoch": 0.5121358235677084,
      "grad_norm": 10.996293067932129,
      "learning_rate": 2.408607584897156e-06,
      "loss": 3.5622,
      "step": 50345
    },
    {
      "epoch": 0.5121866861979166,
      "grad_norm": 8.267298698425293,
      "learning_rate": 2.4082079724166756e-06,
      "loss": 3.1208,
      "step": 50350
    },
    {
      "epoch": 0.512237548828125,
      "grad_norm": 8.640914916992188,
      "learning_rate": 2.4078083622846726e-06,
      "loss": 2.9342,
      "step": 50355
    },
    {
      "epoch": 0.5122884114583334,
      "grad_norm": 12.42250919342041,
      "learning_rate": 2.407408754511371e-06,
      "loss": 3.2516,
      "step": 50360
    },
    {
      "epoch": 0.5123392740885416,
      "grad_norm": 9.798332214355469,
      "learning_rate": 2.4070091491069934e-06,
      "loss": 3.4086,
      "step": 50365
    },
    {
      "epoch": 0.51239013671875,
      "grad_norm": 15.376659393310547,
      "learning_rate": 2.4066095460817654e-06,
      "loss": 3.5751,
      "step": 50370
    },
    {
      "epoch": 0.5124409993489584,
      "grad_norm": 9.240889549255371,
      "learning_rate": 2.4062099454459107e-06,
      "loss": 3.3049,
      "step": 50375
    },
    {
      "epoch": 0.5124918619791666,
      "grad_norm": 15.17997932434082,
      "learning_rate": 2.4058103472096515e-06,
      "loss": 3.3706,
      "step": 50380
    },
    {
      "epoch": 0.512542724609375,
      "grad_norm": 11.492697715759277,
      "learning_rate": 2.4054107513832117e-06,
      "loss": 3.4086,
      "step": 50385
    },
    {
      "epoch": 0.5125935872395834,
      "grad_norm": 9.998835563659668,
      "learning_rate": 2.4050111579768166e-06,
      "loss": 2.9284,
      "step": 50390
    },
    {
      "epoch": 0.5126444498697916,
      "grad_norm": 11.688558578491211,
      "learning_rate": 2.4046115670006877e-06,
      "loss": 3.3235,
      "step": 50395
    },
    {
      "epoch": 0.5126953125,
      "grad_norm": 11.273649215698242,
      "learning_rate": 2.4042119784650493e-06,
      "loss": 3.3909,
      "step": 50400
    },
    {
      "epoch": 0.5127461751302084,
      "grad_norm": 10.354995727539062,
      "learning_rate": 2.403812392380125e-06,
      "loss": 3.4836,
      "step": 50405
    },
    {
      "epoch": 0.5127970377604166,
      "grad_norm": 9.114692687988281,
      "learning_rate": 2.4034128087561383e-06,
      "loss": 3.4699,
      "step": 50410
    },
    {
      "epoch": 0.512847900390625,
      "grad_norm": 11.493762016296387,
      "learning_rate": 2.403013227603311e-06,
      "loss": 3.1074,
      "step": 50415
    },
    {
      "epoch": 0.5128987630208334,
      "grad_norm": 691.8250122070312,
      "learning_rate": 2.4026136489318676e-06,
      "loss": 3.792,
      "step": 50420
    },
    {
      "epoch": 0.5129496256510416,
      "grad_norm": 13.626561164855957,
      "learning_rate": 2.4022140727520313e-06,
      "loss": 3.4223,
      "step": 50425
    },
    {
      "epoch": 0.51300048828125,
      "grad_norm": 6.908864498138428,
      "learning_rate": 2.4018144990740245e-06,
      "loss": 3.213,
      "step": 50430
    },
    {
      "epoch": 0.5130513509114584,
      "grad_norm": 13.467784881591797,
      "learning_rate": 2.401414927908071e-06,
      "loss": 3.7017,
      "step": 50435
    },
    {
      "epoch": 0.5131022135416666,
      "grad_norm": 15.351990699768066,
      "learning_rate": 2.4010153592643923e-06,
      "loss": 3.4501,
      "step": 50440
    },
    {
      "epoch": 0.513153076171875,
      "grad_norm": 12.327286720275879,
      "learning_rate": 2.400615793153212e-06,
      "loss": 3.5289,
      "step": 50445
    },
    {
      "epoch": 0.5132039388020834,
      "grad_norm": 14.21029281616211,
      "learning_rate": 2.4002162295847545e-06,
      "loss": 3.0235,
      "step": 50450
    },
    {
      "epoch": 0.5132548014322916,
      "grad_norm": 12.524280548095703,
      "learning_rate": 2.3998166685692402e-06,
      "loss": 3.6627,
      "step": 50455
    },
    {
      "epoch": 0.5133056640625,
      "grad_norm": 12.760085105895996,
      "learning_rate": 2.399417110116892e-06,
      "loss": 3.4636,
      "step": 50460
    },
    {
      "epoch": 0.5133565266927084,
      "grad_norm": 10.160909652709961,
      "learning_rate": 2.3990175542379347e-06,
      "loss": 3.5232,
      "step": 50465
    },
    {
      "epoch": 0.5134073893229166,
      "grad_norm": 17.44170570373535,
      "learning_rate": 2.3986180009425887e-06,
      "loss": 3.5578,
      "step": 50470
    },
    {
      "epoch": 0.513458251953125,
      "grad_norm": 10.31970500946045,
      "learning_rate": 2.3982184502410774e-06,
      "loss": 3.331,
      "step": 50475
    },
    {
      "epoch": 0.5135091145833334,
      "grad_norm": 10.512755393981934,
      "learning_rate": 2.397818902143622e-06,
      "loss": 3.3922,
      "step": 50480
    },
    {
      "epoch": 0.5135599772135416,
      "grad_norm": 9.803839683532715,
      "learning_rate": 2.397419356660446e-06,
      "loss": 3.3884,
      "step": 50485
    },
    {
      "epoch": 0.51361083984375,
      "grad_norm": 13.15937328338623,
      "learning_rate": 2.3970198138017724e-06,
      "loss": 3.4839,
      "step": 50490
    },
    {
      "epoch": 0.5136617024739584,
      "grad_norm": 9.590703964233398,
      "learning_rate": 2.396620273577821e-06,
      "loss": 3.0772,
      "step": 50495
    },
    {
      "epoch": 0.5137125651041666,
      "grad_norm": 14.244230270385742,
      "learning_rate": 2.3962207359988166e-06,
      "loss": 3.1122,
      "step": 50500
    },
    {
      "epoch": 0.513763427734375,
      "grad_norm": 12.112451553344727,
      "learning_rate": 2.39582120107498e-06,
      "loss": 3.1967,
      "step": 50505
    },
    {
      "epoch": 0.5138142903645834,
      "grad_norm": 14.854276657104492,
      "learning_rate": 2.3954216688165328e-06,
      "loss": 3.6053,
      "step": 50510
    },
    {
      "epoch": 0.5138651529947916,
      "grad_norm": 8.958395004272461,
      "learning_rate": 2.395022139233697e-06,
      "loss": 3.2142,
      "step": 50515
    },
    {
      "epoch": 0.513916015625,
      "grad_norm": 8.370209693908691,
      "learning_rate": 2.3946226123366962e-06,
      "loss": 3.4084,
      "step": 50520
    },
    {
      "epoch": 0.5139668782552084,
      "grad_norm": 14.15273380279541,
      "learning_rate": 2.39422308813575e-06,
      "loss": 3.4198,
      "step": 50525
    },
    {
      "epoch": 0.5140177408854166,
      "grad_norm": 10.030532836914062,
      "learning_rate": 2.3938235666410815e-06,
      "loss": 3.345,
      "step": 50530
    },
    {
      "epoch": 0.514068603515625,
      "grad_norm": 12.817328453063965,
      "learning_rate": 2.3934240478629113e-06,
      "loss": 3.5023,
      "step": 50535
    },
    {
      "epoch": 0.5141194661458334,
      "grad_norm": 8.550262451171875,
      "learning_rate": 2.3930245318114617e-06,
      "loss": 3.4911,
      "step": 50540
    },
    {
      "epoch": 0.5141703287760416,
      "grad_norm": 8.891758918762207,
      "learning_rate": 2.3926250184969547e-06,
      "loss": 3.4614,
      "step": 50545
    },
    {
      "epoch": 0.51422119140625,
      "grad_norm": 16.2994384765625,
      "learning_rate": 2.3922255079296106e-06,
      "loss": 3.341,
      "step": 50550
    },
    {
      "epoch": 0.5142720540364584,
      "grad_norm": 13.644462585449219,
      "learning_rate": 2.391826000119651e-06,
      "loss": 3.1707,
      "step": 50555
    },
    {
      "epoch": 0.5143229166666666,
      "grad_norm": 11.300737380981445,
      "learning_rate": 2.391426495077298e-06,
      "loss": 3.1572,
      "step": 50560
    },
    {
      "epoch": 0.514373779296875,
      "grad_norm": 13.053349494934082,
      "learning_rate": 2.3910269928127723e-06,
      "loss": 3.1483,
      "step": 50565
    },
    {
      "epoch": 0.5144246419270834,
      "grad_norm": 10.000075340270996,
      "learning_rate": 2.3906274933362955e-06,
      "loss": 3.4378,
      "step": 50570
    },
    {
      "epoch": 0.5144755045572916,
      "grad_norm": 15.421857833862305,
      "learning_rate": 2.3902279966580873e-06,
      "loss": 3.5272,
      "step": 50575
    },
    {
      "epoch": 0.5145263671875,
      "grad_norm": 7.433443546295166,
      "learning_rate": 2.389828502788371e-06,
      "loss": 3.3488,
      "step": 50580
    },
    {
      "epoch": 0.5145772298177084,
      "grad_norm": 11.181934356689453,
      "learning_rate": 2.3894290117373657e-06,
      "loss": 3.2142,
      "step": 50585
    },
    {
      "epoch": 0.5146280924479166,
      "grad_norm": 13.431914329528809,
      "learning_rate": 2.389029523515293e-06,
      "loss": 3.3927,
      "step": 50590
    },
    {
      "epoch": 0.514678955078125,
      "grad_norm": 13.219193458557129,
      "learning_rate": 2.388630038132374e-06,
      "loss": 3.3451,
      "step": 50595
    },
    {
      "epoch": 0.5147298177083334,
      "grad_norm": 15.109130859375,
      "learning_rate": 2.388230555598829e-06,
      "loss": 3.4154,
      "step": 50600
    },
    {
      "epoch": 0.5147806803385416,
      "grad_norm": 13.607009887695312,
      "learning_rate": 2.387831075924879e-06,
      "loss": 3.0114,
      "step": 50605
    },
    {
      "epoch": 0.51483154296875,
      "grad_norm": 13.028376579284668,
      "learning_rate": 2.3874315991207434e-06,
      "loss": 3.5488,
      "step": 50610
    },
    {
      "epoch": 0.5148824055989584,
      "grad_norm": 11.688109397888184,
      "learning_rate": 2.387032125196644e-06,
      "loss": 3.1004,
      "step": 50615
    },
    {
      "epoch": 0.5149332682291666,
      "grad_norm": 10.281679153442383,
      "learning_rate": 2.3866326541628017e-06,
      "loss": 3.3787,
      "step": 50620
    },
    {
      "epoch": 0.514984130859375,
      "grad_norm": 13.672207832336426,
      "learning_rate": 2.3862331860294353e-06,
      "loss": 3.0825,
      "step": 50625
    },
    {
      "epoch": 0.5150349934895834,
      "grad_norm": 13.863525390625,
      "learning_rate": 2.3858337208067654e-06,
      "loss": 3.4848,
      "step": 50630
    },
    {
      "epoch": 0.5150858561197916,
      "grad_norm": 15.515698432922363,
      "learning_rate": 2.3854342585050138e-06,
      "loss": 3.2334,
      "step": 50635
    },
    {
      "epoch": 0.51513671875,
      "grad_norm": 11.521782875061035,
      "learning_rate": 2.385034799134399e-06,
      "loss": 3.2403,
      "step": 50640
    },
    {
      "epoch": 0.5151875813802084,
      "grad_norm": 14.413559913635254,
      "learning_rate": 2.3846353427051423e-06,
      "loss": 3.2055,
      "step": 50645
    },
    {
      "epoch": 0.5152384440104166,
      "grad_norm": 11.58481502532959,
      "learning_rate": 2.3842358892274614e-06,
      "loss": 3.4076,
      "step": 50650
    },
    {
      "epoch": 0.515289306640625,
      "grad_norm": 9.105278015136719,
      "learning_rate": 2.383836438711579e-06,
      "loss": 3.4334,
      "step": 50655
    },
    {
      "epoch": 0.5153401692708334,
      "grad_norm": 13.805219650268555,
      "learning_rate": 2.383436991167714e-06,
      "loss": 3.4611,
      "step": 50660
    },
    {
      "epoch": 0.5153910319010416,
      "grad_norm": 14.29794979095459,
      "learning_rate": 2.3830375466060855e-06,
      "loss": 3.3283,
      "step": 50665
    },
    {
      "epoch": 0.51544189453125,
      "grad_norm": 14.205374717712402,
      "learning_rate": 2.3826381050369128e-06,
      "loss": 3.3684,
      "step": 50670
    },
    {
      "epoch": 0.5154927571614584,
      "grad_norm": 10.655013084411621,
      "learning_rate": 2.382238666470418e-06,
      "loss": 3.6278,
      "step": 50675
    },
    {
      "epoch": 0.5155436197916666,
      "grad_norm": 10.02477741241455,
      "learning_rate": 2.3818392309168183e-06,
      "loss": 3.0367,
      "step": 50680
    },
    {
      "epoch": 0.515594482421875,
      "grad_norm": 8.622747421264648,
      "learning_rate": 2.381439798386333e-06,
      "loss": 3.2649,
      "step": 50685
    },
    {
      "epoch": 0.5156453450520834,
      "grad_norm": 16.067399978637695,
      "learning_rate": 2.3810403688891835e-06,
      "loss": 3.3215,
      "step": 50690
    },
    {
      "epoch": 0.5156962076822916,
      "grad_norm": 8.392431259155273,
      "learning_rate": 2.3806409424355877e-06,
      "loss": 3.3279,
      "step": 50695
    },
    {
      "epoch": 0.5157470703125,
      "grad_norm": 14.65240478515625,
      "learning_rate": 2.380241519035766e-06,
      "loss": 3.3661,
      "step": 50700
    },
    {
      "epoch": 0.5157979329427084,
      "grad_norm": 15.060720443725586,
      "learning_rate": 2.3798420986999343e-06,
      "loss": 3.2802,
      "step": 50705
    },
    {
      "epoch": 0.5158487955729166,
      "grad_norm": 8.779844284057617,
      "learning_rate": 2.379442681438316e-06,
      "loss": 3.4106,
      "step": 50710
    },
    {
      "epoch": 0.515899658203125,
      "grad_norm": 13.472846031188965,
      "learning_rate": 2.3790432672611275e-06,
      "loss": 3.6712,
      "step": 50715
    },
    {
      "epoch": 0.5159505208333334,
      "grad_norm": 12.658249855041504,
      "learning_rate": 2.3786438561785887e-06,
      "loss": 3.3399,
      "step": 50720
    },
    {
      "epoch": 0.5160013834635416,
      "grad_norm": 15.144787788391113,
      "learning_rate": 2.3782444482009173e-06,
      "loss": 3.2874,
      "step": 50725
    },
    {
      "epoch": 0.51605224609375,
      "grad_norm": 11.097823143005371,
      "learning_rate": 2.3778450433383333e-06,
      "loss": 3.4698,
      "step": 50730
    },
    {
      "epoch": 0.5161031087239584,
      "grad_norm": 12.127666473388672,
      "learning_rate": 2.377445641601055e-06,
      "loss": 3.1847,
      "step": 50735
    },
    {
      "epoch": 0.5161539713541666,
      "grad_norm": 8.602930068969727,
      "learning_rate": 2.377046242999301e-06,
      "loss": 3.3704,
      "step": 50740
    },
    {
      "epoch": 0.516204833984375,
      "grad_norm": 9.570191383361816,
      "learning_rate": 2.3766468475432887e-06,
      "loss": 3.4845,
      "step": 50745
    },
    {
      "epoch": 0.5162556966145834,
      "grad_norm": 12.587255477905273,
      "learning_rate": 2.376247455243239e-06,
      "loss": 3.1955,
      "step": 50750
    },
    {
      "epoch": 0.5163065592447916,
      "grad_norm": 9.452386856079102,
      "learning_rate": 2.3758480661093683e-06,
      "loss": 3.3975,
      "step": 50755
    },
    {
      "epoch": 0.516357421875,
      "grad_norm": 10.14846134185791,
      "learning_rate": 2.375448680151896e-06,
      "loss": 3.3591,
      "step": 50760
    },
    {
      "epoch": 0.5164082845052084,
      "grad_norm": 9.86368179321289,
      "learning_rate": 2.3750492973810383e-06,
      "loss": 3.1278,
      "step": 50765
    },
    {
      "epoch": 0.5164591471354166,
      "grad_norm": 10.385574340820312,
      "learning_rate": 2.3746499178070155e-06,
      "loss": 3.4764,
      "step": 50770
    },
    {
      "epoch": 0.516510009765625,
      "grad_norm": 12.00722599029541,
      "learning_rate": 2.3742505414400454e-06,
      "loss": 3.3049,
      "step": 50775
    },
    {
      "epoch": 0.5165608723958334,
      "grad_norm": 13.417591094970703,
      "learning_rate": 2.373851168290344e-06,
      "loss": 3.1833,
      "step": 50780
    },
    {
      "epoch": 0.5166117350260416,
      "grad_norm": 15.702399253845215,
      "learning_rate": 2.373451798368131e-06,
      "loss": 3.1977,
      "step": 50785
    },
    {
      "epoch": 0.51666259765625,
      "grad_norm": 8.82960033416748,
      "learning_rate": 2.3730524316836247e-06,
      "loss": 3.4057,
      "step": 50790
    },
    {
      "epoch": 0.5167134602864584,
      "grad_norm": 11.624604225158691,
      "learning_rate": 2.372653068247041e-06,
      "loss": 3.5341,
      "step": 50795
    },
    {
      "epoch": 0.5167643229166666,
      "grad_norm": 14.807909965515137,
      "learning_rate": 2.372253708068598e-06,
      "loss": 3.719,
      "step": 50800
    },
    {
      "epoch": 0.516815185546875,
      "grad_norm": 20.163330078125,
      "learning_rate": 2.371854351158515e-06,
      "loss": 3.3702,
      "step": 50805
    },
    {
      "epoch": 0.5168660481770834,
      "grad_norm": 15.910099029541016,
      "learning_rate": 2.371454997527007e-06,
      "loss": 3.7183,
      "step": 50810
    },
    {
      "epoch": 0.5169169108072916,
      "grad_norm": 14.499089241027832,
      "learning_rate": 2.3710556471842934e-06,
      "loss": 3.5245,
      "step": 50815
    },
    {
      "epoch": 0.5169677734375,
      "grad_norm": 9.212532043457031,
      "learning_rate": 2.370656300140589e-06,
      "loss": 3.3738,
      "step": 50820
    },
    {
      "epoch": 0.5170186360677084,
      "grad_norm": 14.878519058227539,
      "learning_rate": 2.370256956406113e-06,
      "loss": 3.2764,
      "step": 50825
    },
    {
      "epoch": 0.5170694986979166,
      "grad_norm": 13.573134422302246,
      "learning_rate": 2.369857615991083e-06,
      "loss": 3.3369,
      "step": 50830
    },
    {
      "epoch": 0.517120361328125,
      "grad_norm": 9.683137893676758,
      "learning_rate": 2.3694582789057145e-06,
      "loss": 3.1725,
      "step": 50835
    },
    {
      "epoch": 0.5171712239583334,
      "grad_norm": 12.827288627624512,
      "learning_rate": 2.369058945160224e-06,
      "loss": 3.366,
      "step": 50840
    },
    {
      "epoch": 0.5172220865885416,
      "grad_norm": 13.126194953918457,
      "learning_rate": 2.3686596147648306e-06,
      "loss": 3.2969,
      "step": 50845
    },
    {
      "epoch": 0.51727294921875,
      "grad_norm": 17.425718307495117,
      "learning_rate": 2.3682602877297497e-06,
      "loss": 3.1351,
      "step": 50850
    },
    {
      "epoch": 0.5173238118489584,
      "grad_norm": 12.274630546569824,
      "learning_rate": 2.367860964065198e-06,
      "loss": 3.4009,
      "step": 50855
    },
    {
      "epoch": 0.5173746744791666,
      "grad_norm": 13.878703117370605,
      "learning_rate": 2.367461643781391e-06,
      "loss": 3.0991,
      "step": 50860
    },
    {
      "epoch": 0.517425537109375,
      "grad_norm": 13.84969425201416,
      "learning_rate": 2.367062326888548e-06,
      "loss": 3.1251,
      "step": 50865
    },
    {
      "epoch": 0.5174763997395834,
      "grad_norm": 7.590649604797363,
      "learning_rate": 2.3666630133968833e-06,
      "loss": 3.6181,
      "step": 50870
    },
    {
      "epoch": 0.5175272623697916,
      "grad_norm": 8.43957805633545,
      "learning_rate": 2.366263703316614e-06,
      "loss": 3.6765,
      "step": 50875
    },
    {
      "epoch": 0.517578125,
      "grad_norm": 13.262419700622559,
      "learning_rate": 2.3658643966579554e-06,
      "loss": 3.394,
      "step": 50880
    },
    {
      "epoch": 0.5176289876302084,
      "grad_norm": 13.628838539123535,
      "learning_rate": 2.365465093431125e-06,
      "loss": 3.6081,
      "step": 50885
    },
    {
      "epoch": 0.5176798502604166,
      "grad_norm": 16.391429901123047,
      "learning_rate": 2.3650657936463386e-06,
      "loss": 3.781,
      "step": 50890
    },
    {
      "epoch": 0.517730712890625,
      "grad_norm": 16.025196075439453,
      "learning_rate": 2.364666497313811e-06,
      "loss": 3.2152,
      "step": 50895
    },
    {
      "epoch": 0.5177815755208334,
      "grad_norm": 9.452840805053711,
      "learning_rate": 2.3642672044437597e-06,
      "loss": 3.3014,
      "step": 50900
    },
    {
      "epoch": 0.5178324381510416,
      "grad_norm": 12.161417007446289,
      "learning_rate": 2.3638679150464e-06,
      "loss": 3.2071,
      "step": 50905
    },
    {
      "epoch": 0.51788330078125,
      "grad_norm": 10.940399169921875,
      "learning_rate": 2.363468629131947e-06,
      "loss": 3.0984,
      "step": 50910
    },
    {
      "epoch": 0.5179341634114584,
      "grad_norm": 9.861112594604492,
      "learning_rate": 2.3630693467106164e-06,
      "loss": 3.2047,
      "step": 50915
    },
    {
      "epoch": 0.5179850260416666,
      "grad_norm": 9.728883743286133,
      "learning_rate": 2.3626700677926253e-06,
      "loss": 3.5047,
      "step": 50920
    },
    {
      "epoch": 0.518035888671875,
      "grad_norm": 14.600275993347168,
      "learning_rate": 2.362270792388187e-06,
      "loss": 3.7304,
      "step": 50925
    },
    {
      "epoch": 0.5180867513020834,
      "grad_norm": 12.862422943115234,
      "learning_rate": 2.3618715205075187e-06,
      "loss": 3.7425,
      "step": 50930
    },
    {
      "epoch": 0.5181376139322916,
      "grad_norm": 10.666069030761719,
      "learning_rate": 2.3614722521608332e-06,
      "loss": 3.2768,
      "step": 50935
    },
    {
      "epoch": 0.5181884765625,
      "grad_norm": 12.862142562866211,
      "learning_rate": 2.3610729873583484e-06,
      "loss": 3.1389,
      "step": 50940
    },
    {
      "epoch": 0.5182393391927084,
      "grad_norm": 13.986405372619629,
      "learning_rate": 2.3606737261102785e-06,
      "loss": 3.0806,
      "step": 50945
    },
    {
      "epoch": 0.5182902018229166,
      "grad_norm": 11.19984245300293,
      "learning_rate": 2.360274468426838e-06,
      "loss": 3.6976,
      "step": 50950
    },
    {
      "epoch": 0.518341064453125,
      "grad_norm": 12.949975967407227,
      "learning_rate": 2.3598752143182414e-06,
      "loss": 3.2316,
      "step": 50955
    },
    {
      "epoch": 0.5183919270833334,
      "grad_norm": 14.107664108276367,
      "learning_rate": 2.359475963794705e-06,
      "loss": 3.7034,
      "step": 50960
    },
    {
      "epoch": 0.5184427897135416,
      "grad_norm": 16.346643447875977,
      "learning_rate": 2.359076716866443e-06,
      "loss": 3.275,
      "step": 50965
    },
    {
      "epoch": 0.51849365234375,
      "grad_norm": 10.664952278137207,
      "learning_rate": 2.3586774735436697e-06,
      "loss": 3.1661,
      "step": 50970
    },
    {
      "epoch": 0.5185445149739584,
      "grad_norm": 9.160752296447754,
      "learning_rate": 2.358278233836599e-06,
      "loss": 3.2862,
      "step": 50975
    },
    {
      "epoch": 0.5185953776041666,
      "grad_norm": 15.120304107666016,
      "learning_rate": 2.3578789977554465e-06,
      "loss": 3.3029,
      "step": 50980
    },
    {
      "epoch": 0.518646240234375,
      "grad_norm": 12.193244934082031,
      "learning_rate": 2.3574797653104266e-06,
      "loss": 2.9656,
      "step": 50985
    },
    {
      "epoch": 0.5186971028645834,
      "grad_norm": 10.405268669128418,
      "learning_rate": 2.3570805365117523e-06,
      "loss": 3.1787,
      "step": 50990
    },
    {
      "epoch": 0.5187479654947916,
      "grad_norm": 13.421646118164062,
      "learning_rate": 2.3566813113696387e-06,
      "loss": 3.4537,
      "step": 50995
    },
    {
      "epoch": 0.518798828125,
      "grad_norm": 9.784440994262695,
      "learning_rate": 2.3562820898943007e-06,
      "loss": 3.3648,
      "step": 51000
    },
    {
      "epoch": 0.5188496907552084,
      "grad_norm": 9.541216850280762,
      "learning_rate": 2.3558828720959505e-06,
      "loss": 3.6462,
      "step": 51005
    },
    {
      "epoch": 0.5189005533854166,
      "grad_norm": 13.812661170959473,
      "learning_rate": 2.355483657984803e-06,
      "loss": 3.1048,
      "step": 51010
    },
    {
      "epoch": 0.518951416015625,
      "grad_norm": 12.758337020874023,
      "learning_rate": 2.3550844475710714e-06,
      "loss": 3.4573,
      "step": 51015
    },
    {
      "epoch": 0.5190022786458334,
      "grad_norm": 8.859757423400879,
      "learning_rate": 2.3546852408649713e-06,
      "loss": 3.2942,
      "step": 51020
    },
    {
      "epoch": 0.5190531412760416,
      "grad_norm": 13.27428913116455,
      "learning_rate": 2.354286037876715e-06,
      "loss": 3.3289,
      "step": 51025
    },
    {
      "epoch": 0.51910400390625,
      "grad_norm": 7.497504711151123,
      "learning_rate": 2.3538868386165133e-06,
      "loss": 3.3419,
      "step": 51030
    },
    {
      "epoch": 0.5191548665364584,
      "grad_norm": 12.467268943786621,
      "learning_rate": 2.353487643094584e-06,
      "loss": 3.3565,
      "step": 51035
    },
    {
      "epoch": 0.5192057291666666,
      "grad_norm": 15.2796630859375,
      "learning_rate": 2.3530884513211387e-06,
      "loss": 3.4037,
      "step": 51040
    },
    {
      "epoch": 0.519256591796875,
      "grad_norm": 13.033889770507812,
      "learning_rate": 2.352689263306391e-06,
      "loss": 3.5713,
      "step": 51045
    },
    {
      "epoch": 0.5193074544270834,
      "grad_norm": 15.384819984436035,
      "learning_rate": 2.352290079060552e-06,
      "loss": 3.6361,
      "step": 51050
    },
    {
      "epoch": 0.5193583170572916,
      "grad_norm": 12.511043548583984,
      "learning_rate": 2.351890898593837e-06,
      "loss": 3.4419,
      "step": 51055
    },
    {
      "epoch": 0.5194091796875,
      "grad_norm": 12.416461944580078,
      "learning_rate": 2.351491721916459e-06,
      "loss": 4.2458,
      "step": 51060
    },
    {
      "epoch": 0.5194600423177084,
      "grad_norm": 12.38286304473877,
      "learning_rate": 2.3510925490386296e-06,
      "loss": 3.8934,
      "step": 51065
    },
    {
      "epoch": 0.5195109049479166,
      "grad_norm": 9.572614669799805,
      "learning_rate": 2.3506933799705613e-06,
      "loss": 3.1409,
      "step": 51070
    },
    {
      "epoch": 0.519561767578125,
      "grad_norm": 11.439574241638184,
      "learning_rate": 2.3502942147224684e-06,
      "loss": 3.4687,
      "step": 51075
    },
    {
      "epoch": 0.5196126302083334,
      "grad_norm": 14.704670906066895,
      "learning_rate": 2.3498950533045624e-06,
      "loss": 3.1802,
      "step": 51080
    },
    {
      "epoch": 0.5196634928385416,
      "grad_norm": 40.19412612915039,
      "learning_rate": 2.349495895727055e-06,
      "loss": 3.3865,
      "step": 51085
    },
    {
      "epoch": 0.51971435546875,
      "grad_norm": 11.445558547973633,
      "learning_rate": 2.349096742000161e-06,
      "loss": 3.2296,
      "step": 51090
    },
    {
      "epoch": 0.5197652180989584,
      "grad_norm": 13.202363014221191,
      "learning_rate": 2.34869759213409e-06,
      "loss": 3.9636,
      "step": 51095
    },
    {
      "epoch": 0.5198160807291666,
      "grad_norm": 8.832242965698242,
      "learning_rate": 2.3482984461390562e-06,
      "loss": 3.1758,
      "step": 51100
    },
    {
      "epoch": 0.519866943359375,
      "grad_norm": 9.222075462341309,
      "learning_rate": 2.347899304025269e-06,
      "loss": 3.2653,
      "step": 51105
    },
    {
      "epoch": 0.5199178059895834,
      "grad_norm": 12.401978492736816,
      "learning_rate": 2.3475001658029432e-06,
      "loss": 3.3345,
      "step": 51110
    },
    {
      "epoch": 0.5199686686197916,
      "grad_norm": 7.745174884796143,
      "learning_rate": 2.34710103148229e-06,
      "loss": 3.5809,
      "step": 51115
    },
    {
      "epoch": 0.52001953125,
      "grad_norm": 13.621798515319824,
      "learning_rate": 2.3467019010735203e-06,
      "loss": 3.3218,
      "step": 51120
    },
    {
      "epoch": 0.5200703938802084,
      "grad_norm": 14.074609756469727,
      "learning_rate": 2.346302774586845e-06,
      "loss": 3.9371,
      "step": 51125
    },
    {
      "epoch": 0.5201212565104166,
      "grad_norm": 11.33251953125,
      "learning_rate": 2.345903652032478e-06,
      "loss": 3.6171,
      "step": 51130
    },
    {
      "epoch": 0.520172119140625,
      "grad_norm": 15.287870407104492,
      "learning_rate": 2.3455045334206294e-06,
      "loss": 3.5255,
      "step": 51135
    },
    {
      "epoch": 0.5202229817708334,
      "grad_norm": 7.853886604309082,
      "learning_rate": 2.345105418761511e-06,
      "loss": 3.231,
      "step": 51140
    },
    {
      "epoch": 0.5202738444010416,
      "grad_norm": 9.591255187988281,
      "learning_rate": 2.3447063080653327e-06,
      "loss": 3.2378,
      "step": 51145
    },
    {
      "epoch": 0.52032470703125,
      "grad_norm": 13.270482063293457,
      "learning_rate": 2.344307201342307e-06,
      "loss": 3.4815,
      "step": 51150
    },
    {
      "epoch": 0.5203755696614584,
      "grad_norm": 19.019012451171875,
      "learning_rate": 2.3439080986026454e-06,
      "loss": 3.2569,
      "step": 51155
    },
    {
      "epoch": 0.5204264322916666,
      "grad_norm": 15.56859302520752,
      "learning_rate": 2.3435089998565576e-06,
      "loss": 3.5787,
      "step": 51160
    },
    {
      "epoch": 0.520477294921875,
      "grad_norm": 11.9397554397583,
      "learning_rate": 2.3431099051142545e-06,
      "loss": 3.2714,
      "step": 51165
    },
    {
      "epoch": 0.5205281575520834,
      "grad_norm": 14.755770683288574,
      "learning_rate": 2.342710814385948e-06,
      "loss": 3.5029,
      "step": 51170
    },
    {
      "epoch": 0.5205790201822916,
      "grad_norm": 9.2692232131958,
      "learning_rate": 2.342311727681848e-06,
      "loss": 3.3179,
      "step": 51175
    },
    {
      "epoch": 0.5206298828125,
      "grad_norm": 11.290838241577148,
      "learning_rate": 2.3419126450121642e-06,
      "loss": 3.2469,
      "step": 51180
    },
    {
      "epoch": 0.5206807454427084,
      "grad_norm": 9.631206512451172,
      "learning_rate": 2.3415135663871083e-06,
      "loss": 3.3658,
      "step": 51185
    },
    {
      "epoch": 0.5207316080729166,
      "grad_norm": 13.162626266479492,
      "learning_rate": 2.341114491816891e-06,
      "loss": 3.1031,
      "step": 51190
    },
    {
      "epoch": 0.520782470703125,
      "grad_norm": 15.770508766174316,
      "learning_rate": 2.340715421311721e-06,
      "loss": 3.6408,
      "step": 51195
    },
    {
      "epoch": 0.5208333333333334,
      "grad_norm": 14.972073554992676,
      "learning_rate": 2.3403163548818093e-06,
      "loss": 3.098,
      "step": 51200
    },
    {
      "epoch": 0.5208841959635416,
      "grad_norm": 12.583430290222168,
      "learning_rate": 2.339917292537366e-06,
      "loss": 3.1795,
      "step": 51205
    },
    {
      "epoch": 0.52093505859375,
      "grad_norm": 14.268254280090332,
      "learning_rate": 2.339518234288601e-06,
      "loss": 3.473,
      "step": 51210
    },
    {
      "epoch": 0.5209859212239584,
      "grad_norm": 13.213074684143066,
      "learning_rate": 2.3391191801457243e-06,
      "loss": 3.4323,
      "step": 51215
    },
    {
      "epoch": 0.5210367838541666,
      "grad_norm": 14.242136001586914,
      "learning_rate": 2.338720130118944e-06,
      "loss": 3.3706,
      "step": 51220
    },
    {
      "epoch": 0.521087646484375,
      "grad_norm": 10.817700386047363,
      "learning_rate": 2.3383210842184713e-06,
      "loss": 3.2105,
      "step": 51225
    },
    {
      "epoch": 0.5211385091145834,
      "grad_norm": 12.685539245605469,
      "learning_rate": 2.3379220424545164e-06,
      "loss": 3.3286,
      "step": 51230
    },
    {
      "epoch": 0.5211893717447916,
      "grad_norm": 12.610810279846191,
      "learning_rate": 2.3375230048372866e-06,
      "loss": 3.3509,
      "step": 51235
    },
    {
      "epoch": 0.521240234375,
      "grad_norm": 12.87966537475586,
      "learning_rate": 2.3371239713769918e-06,
      "loss": 3.0719,
      "step": 51240
    },
    {
      "epoch": 0.5212910970052084,
      "grad_norm": 15.29201602935791,
      "learning_rate": 2.336724942083843e-06,
      "loss": 3.5089,
      "step": 51245
    },
    {
      "epoch": 0.5213419596354166,
      "grad_norm": 6.654240608215332,
      "learning_rate": 2.3363259169680472e-06,
      "loss": 2.8616,
      "step": 51250
    },
    {
      "epoch": 0.521392822265625,
      "grad_norm": 13.208778381347656,
      "learning_rate": 2.3359268960398145e-06,
      "loss": 3.2127,
      "step": 51255
    },
    {
      "epoch": 0.5214436848958334,
      "grad_norm": 12.543509483337402,
      "learning_rate": 2.335527879309352e-06,
      "loss": 3.5661,
      "step": 51260
    },
    {
      "epoch": 0.5214945475260416,
      "grad_norm": 11.997392654418945,
      "learning_rate": 2.3351288667868705e-06,
      "loss": 3.1548,
      "step": 51265
    },
    {
      "epoch": 0.52154541015625,
      "grad_norm": 9.279683113098145,
      "learning_rate": 2.334729858482578e-06,
      "loss": 3.0211,
      "step": 51270
    },
    {
      "epoch": 0.5215962727864584,
      "grad_norm": 12.218412399291992,
      "learning_rate": 2.3343308544066823e-06,
      "loss": 3.2288,
      "step": 51275
    },
    {
      "epoch": 0.5216471354166666,
      "grad_norm": 11.581144332885742,
      "learning_rate": 2.3339318545693924e-06,
      "loss": 3.3161,
      "step": 51280
    },
    {
      "epoch": 0.521697998046875,
      "grad_norm": 13.967557907104492,
      "learning_rate": 2.3335328589809177e-06,
      "loss": 3.2799,
      "step": 51285
    },
    {
      "epoch": 0.5217488606770834,
      "grad_norm": 10.300307273864746,
      "learning_rate": 2.3331338676514648e-06,
      "loss": 3.5126,
      "step": 51290
    },
    {
      "epoch": 0.5217997233072916,
      "grad_norm": 11.853108406066895,
      "learning_rate": 2.3327348805912417e-06,
      "loss": 3.2095,
      "step": 51295
    },
    {
      "epoch": 0.5218505859375,
      "grad_norm": 13.32164192199707,
      "learning_rate": 2.332335897810458e-06,
      "loss": 3.4154,
      "step": 51300
    },
    {
      "epoch": 0.5219014485677084,
      "grad_norm": 17.006755828857422,
      "learning_rate": 2.3319369193193203e-06,
      "loss": 3.0868,
      "step": 51305
    },
    {
      "epoch": 0.5219523111979166,
      "grad_norm": 11.63281536102295,
      "learning_rate": 2.331537945128037e-06,
      "loss": 3.0452,
      "step": 51310
    },
    {
      "epoch": 0.522003173828125,
      "grad_norm": 8.596928596496582,
      "learning_rate": 2.3311389752468144e-06,
      "loss": 3.5104,
      "step": 51315
    },
    {
      "epoch": 0.5220540364583334,
      "grad_norm": 13.741214752197266,
      "learning_rate": 2.330740009685862e-06,
      "loss": 3.1809,
      "step": 51320
    },
    {
      "epoch": 0.5221048990885416,
      "grad_norm": 9.251409530639648,
      "learning_rate": 2.3303410484553863e-06,
      "loss": 3.7666,
      "step": 51325
    },
    {
      "epoch": 0.52215576171875,
      "grad_norm": 13.186622619628906,
      "learning_rate": 2.3299420915655942e-06,
      "loss": 3.5376,
      "step": 51330
    },
    {
      "epoch": 0.5222066243489584,
      "grad_norm": 10.63139533996582,
      "learning_rate": 2.329543139026693e-06,
      "loss": 3.2363,
      "step": 51335
    },
    {
      "epoch": 0.5222574869791666,
      "grad_norm": 15.087735176086426,
      "learning_rate": 2.3291441908488908e-06,
      "loss": 2.8915,
      "step": 51340
    },
    {
      "epoch": 0.522308349609375,
      "grad_norm": 8.385112762451172,
      "learning_rate": 2.328745247042394e-06,
      "loss": 3.5765,
      "step": 51345
    },
    {
      "epoch": 0.5223592122395834,
      "grad_norm": 7.616689205169678,
      "learning_rate": 2.3283463076174097e-06,
      "loss": 3.5431,
      "step": 51350
    },
    {
      "epoch": 0.5224100748697916,
      "grad_norm": 8.684670448303223,
      "learning_rate": 2.3279473725841437e-06,
      "loss": 3.1274,
      "step": 51355
    },
    {
      "epoch": 0.5224609375,
      "grad_norm": 14.156767845153809,
      "learning_rate": 2.3275484419528045e-06,
      "loss": 3.6883,
      "step": 51360
    },
    {
      "epoch": 0.5225118001302084,
      "grad_norm": 17.684885025024414,
      "learning_rate": 2.3271495157335965e-06,
      "loss": 3.2219,
      "step": 51365
    },
    {
      "epoch": 0.5225626627604166,
      "grad_norm": 11.786378860473633,
      "learning_rate": 2.3267505939367284e-06,
      "loss": 3.0865,
      "step": 51370
    },
    {
      "epoch": 0.522613525390625,
      "grad_norm": 11.846403121948242,
      "learning_rate": 2.3263516765724036e-06,
      "loss": 3.993,
      "step": 51375
    },
    {
      "epoch": 0.5226643880208334,
      "grad_norm": 12.272603988647461,
      "learning_rate": 2.325952763650831e-06,
      "loss": 3.5374,
      "step": 51380
    },
    {
      "epoch": 0.5227152506510416,
      "grad_norm": 10.249964714050293,
      "learning_rate": 2.3255538551822157e-06,
      "loss": 3.1231,
      "step": 51385
    },
    {
      "epoch": 0.52276611328125,
      "grad_norm": 12.157092094421387,
      "learning_rate": 2.325154951176763e-06,
      "loss": 3.5138,
      "step": 51390
    },
    {
      "epoch": 0.5228169759114584,
      "grad_norm": 11.733100891113281,
      "learning_rate": 2.32475605164468e-06,
      "loss": 3.5915,
      "step": 51395
    },
    {
      "epoch": 0.5228678385416666,
      "grad_norm": 9.639185905456543,
      "learning_rate": 2.3243571565961724e-06,
      "loss": 3.3849,
      "step": 51400
    },
    {
      "epoch": 0.522918701171875,
      "grad_norm": 13.651537895202637,
      "learning_rate": 2.3239582660414446e-06,
      "loss": 3.294,
      "step": 51405
    },
    {
      "epoch": 0.5229695638020834,
      "grad_norm": 8.989803314208984,
      "learning_rate": 2.3235593799907022e-06,
      "loss": 3.4397,
      "step": 51410
    },
    {
      "epoch": 0.5230204264322916,
      "grad_norm": 9.389153480529785,
      "learning_rate": 2.3231604984541525e-06,
      "loss": 3.2164,
      "step": 51415
    },
    {
      "epoch": 0.5230712890625,
      "grad_norm": 13.706945419311523,
      "learning_rate": 2.322761621441999e-06,
      "loss": 3.1938,
      "step": 51420
    },
    {
      "epoch": 0.5231221516927084,
      "grad_norm": 14.718073844909668,
      "learning_rate": 2.3223627489644478e-06,
      "loss": 3.3858,
      "step": 51425
    },
    {
      "epoch": 0.5231730143229166,
      "grad_norm": 14.35042667388916,
      "learning_rate": 2.321963881031703e-06,
      "loss": 3.2756,
      "step": 51430
    },
    {
      "epoch": 0.523223876953125,
      "grad_norm": 7.722215175628662,
      "learning_rate": 2.32156501765397e-06,
      "loss": 3.3564,
      "step": 51435
    },
    {
      "epoch": 0.5232747395833334,
      "grad_norm": 13.248672485351562,
      "learning_rate": 2.3211661588414546e-06,
      "loss": 3.3001,
      "step": 51440
    },
    {
      "epoch": 0.5233256022135416,
      "grad_norm": 9.243656158447266,
      "learning_rate": 2.32076730460436e-06,
      "loss": 3.3604,
      "step": 51445
    },
    {
      "epoch": 0.52337646484375,
      "grad_norm": 8.182433128356934,
      "learning_rate": 2.320368454952891e-06,
      "loss": 3.0259,
      "step": 51450
    },
    {
      "epoch": 0.5234273274739584,
      "grad_norm": 14.732831001281738,
      "learning_rate": 2.3199696098972535e-06,
      "loss": 3.588,
      "step": 51455
    },
    {
      "epoch": 0.5234781901041666,
      "grad_norm": 11.189901351928711,
      "learning_rate": 2.31957076944765e-06,
      "loss": 3.3701,
      "step": 51460
    },
    {
      "epoch": 0.523529052734375,
      "grad_norm": 10.825516700744629,
      "learning_rate": 2.3191719336142864e-06,
      "loss": 3.2404,
      "step": 51465
    },
    {
      "epoch": 0.5235799153645834,
      "grad_norm": 14.079922676086426,
      "learning_rate": 2.318773102407365e-06,
      "loss": 3.1259,
      "step": 51470
    },
    {
      "epoch": 0.5236307779947916,
      "grad_norm": 15.275527000427246,
      "learning_rate": 2.318374275837092e-06,
      "loss": 3.4218,
      "step": 51475
    },
    {
      "epoch": 0.523681640625,
      "grad_norm": 13.454855918884277,
      "learning_rate": 2.3179754539136693e-06,
      "loss": 3.2168,
      "step": 51480
    },
    {
      "epoch": 0.5237325032552084,
      "grad_norm": 10.511768341064453,
      "learning_rate": 2.317576636647301e-06,
      "loss": 3.367,
      "step": 51485
    },
    {
      "epoch": 0.5237833658854166,
      "grad_norm": 12.318464279174805,
      "learning_rate": 2.317177824048192e-06,
      "loss": 2.9529,
      "step": 51490
    },
    {
      "epoch": 0.523834228515625,
      "grad_norm": 9.753463745117188,
      "learning_rate": 2.316779016126545e-06,
      "loss": 3.1118,
      "step": 51495
    },
    {
      "epoch": 0.5238850911458334,
      "grad_norm": 12.015084266662598,
      "learning_rate": 2.3163802128925637e-06,
      "loss": 3.4823,
      "step": 51500
    },
    {
      "epoch": 0.5239359537760416,
      "grad_norm": 10.5300874710083,
      "learning_rate": 2.3159814143564506e-06,
      "loss": 3.3846,
      "step": 51505
    },
    {
      "epoch": 0.52398681640625,
      "grad_norm": 9.35047435760498,
      "learning_rate": 2.315582620528409e-06,
      "loss": 3.1737,
      "step": 51510
    },
    {
      "epoch": 0.5240376790364584,
      "grad_norm": 10.02412223815918,
      "learning_rate": 2.3151838314186437e-06,
      "loss": 3.1713,
      "step": 51515
    },
    {
      "epoch": 0.5240885416666666,
      "grad_norm": 14.379279136657715,
      "learning_rate": 2.3147850470373558e-06,
      "loss": 3.3371,
      "step": 51520
    },
    {
      "epoch": 0.524139404296875,
      "grad_norm": 9.510961532592773,
      "learning_rate": 2.3143862673947475e-06,
      "loss": 3.3166,
      "step": 51525
    },
    {
      "epoch": 0.5241902669270834,
      "grad_norm": 12.390170097351074,
      "learning_rate": 2.3139874925010235e-06,
      "loss": 3.0833,
      "step": 51530
    },
    {
      "epoch": 0.5242411295572916,
      "grad_norm": 16.991884231567383,
      "learning_rate": 2.313588722366385e-06,
      "loss": 3.4757,
      "step": 51535
    },
    {
      "epoch": 0.5242919921875,
      "grad_norm": 10.350554466247559,
      "learning_rate": 2.313189957001036e-06,
      "loss": 3.18,
      "step": 51540
    },
    {
      "epoch": 0.5243428548177084,
      "grad_norm": 11.769200325012207,
      "learning_rate": 2.312791196415176e-06,
      "loss": 3.317,
      "step": 51545
    },
    {
      "epoch": 0.5243937174479166,
      "grad_norm": 13.403620719909668,
      "learning_rate": 2.3123924406190098e-06,
      "loss": 5.084,
      "step": 51550
    },
    {
      "epoch": 0.524444580078125,
      "grad_norm": 15.726815223693848,
      "learning_rate": 2.311993689622739e-06,
      "loss": 3.2429,
      "step": 51555
    },
    {
      "epoch": 0.5244954427083334,
      "grad_norm": 11.416741371154785,
      "learning_rate": 2.311594943436564e-06,
      "loss": 3.0246,
      "step": 51560
    },
    {
      "epoch": 0.5245463053385416,
      "grad_norm": 13.953176498413086,
      "learning_rate": 2.311196202070688e-06,
      "loss": 3.0309,
      "step": 51565
    },
    {
      "epoch": 0.52459716796875,
      "grad_norm": 11.300572395324707,
      "learning_rate": 2.3107974655353125e-06,
      "loss": 3.1699,
      "step": 51570
    },
    {
      "epoch": 0.5246480305989584,
      "grad_norm": 13.902636528015137,
      "learning_rate": 2.3103987338406394e-06,
      "loss": 3.2971,
      "step": 51575
    },
    {
      "epoch": 0.5246988932291666,
      "grad_norm": 14.147994041442871,
      "learning_rate": 2.310000006996869e-06,
      "loss": 2.9933,
      "step": 51580
    },
    {
      "epoch": 0.524749755859375,
      "grad_norm": 10.730506896972656,
      "learning_rate": 2.309601285014204e-06,
      "loss": 3.0943,
      "step": 51585
    },
    {
      "epoch": 0.5248006184895834,
      "grad_norm": 15.618927001953125,
      "learning_rate": 2.3092025679028447e-06,
      "loss": 3.2672,
      "step": 51590
    },
    {
      "epoch": 0.5248514811197916,
      "grad_norm": 11.174661636352539,
      "learning_rate": 2.3088038556729934e-06,
      "loss": 3.5472,
      "step": 51595
    },
    {
      "epoch": 0.52490234375,
      "grad_norm": 14.61530590057373,
      "learning_rate": 2.3084051483348485e-06,
      "loss": 3.1849,
      "step": 51600
    },
    {
      "epoch": 0.5249532063802084,
      "grad_norm": 7.89102029800415,
      "learning_rate": 2.3080064458986136e-06,
      "loss": 3.3216,
      "step": 51605
    },
    {
      "epoch": 0.5250040690104166,
      "grad_norm": 12.069526672363281,
      "learning_rate": 2.3076077483744884e-06,
      "loss": 3.2755,
      "step": 51610
    },
    {
      "epoch": 0.525054931640625,
      "grad_norm": 12.264208793640137,
      "learning_rate": 2.307209055772673e-06,
      "loss": 3.3825,
      "step": 51615
    },
    {
      "epoch": 0.5251057942708334,
      "grad_norm": 18.144428253173828,
      "learning_rate": 2.3068103681033672e-06,
      "loss": 3.5544,
      "step": 51620
    },
    {
      "epoch": 0.5251566569010416,
      "grad_norm": 12.813551902770996,
      "learning_rate": 2.306411685376774e-06,
      "loss": 3.4971,
      "step": 51625
    },
    {
      "epoch": 0.52520751953125,
      "grad_norm": 10.434419631958008,
      "learning_rate": 2.3060130076030907e-06,
      "loss": 3.1101,
      "step": 51630
    },
    {
      "epoch": 0.5252583821614584,
      "grad_norm": 10.288335800170898,
      "learning_rate": 2.30561433479252e-06,
      "loss": 3.1028,
      "step": 51635
    },
    {
      "epoch": 0.5253092447916666,
      "grad_norm": 9.808342933654785,
      "learning_rate": 2.305215666955258e-06,
      "loss": 3.2641,
      "step": 51640
    },
    {
      "epoch": 0.525360107421875,
      "grad_norm": 11.223031044006348,
      "learning_rate": 2.3048170041015095e-06,
      "loss": 3.0565,
      "step": 51645
    },
    {
      "epoch": 0.5254109700520834,
      "grad_norm": 13.923452377319336,
      "learning_rate": 2.3044183462414706e-06,
      "loss": 3.2248,
      "step": 51650
    },
    {
      "epoch": 0.5254618326822916,
      "grad_norm": 9.677695274353027,
      "learning_rate": 2.304019693385343e-06,
      "loss": 3.2461,
      "step": 51655
    },
    {
      "epoch": 0.5255126953125,
      "grad_norm": 14.185688018798828,
      "learning_rate": 2.3036210455433236e-06,
      "loss": 3.252,
      "step": 51660
    },
    {
      "epoch": 0.5255635579427084,
      "grad_norm": 9.480224609375,
      "learning_rate": 2.3032224027256137e-06,
      "loss": 3.7707,
      "step": 51665
    },
    {
      "epoch": 0.5256144205729166,
      "grad_norm": 14.351865768432617,
      "learning_rate": 2.3028237649424127e-06,
      "loss": 3.4189,
      "step": 51670
    },
    {
      "epoch": 0.525665283203125,
      "grad_norm": 7.74800443649292,
      "learning_rate": 2.3024251322039175e-06,
      "loss": 3.4073,
      "step": 51675
    },
    {
      "epoch": 0.5257161458333334,
      "grad_norm": 11.025547981262207,
      "learning_rate": 2.3020265045203295e-06,
      "loss": 3.0491,
      "step": 51680
    },
    {
      "epoch": 0.5257670084635416,
      "grad_norm": 9.585078239440918,
      "learning_rate": 2.301627881901847e-06,
      "loss": 3.041,
      "step": 51685
    },
    {
      "epoch": 0.52581787109375,
      "grad_norm": 12.840312957763672,
      "learning_rate": 2.301229264358667e-06,
      "loss": 3.3416,
      "step": 51690
    },
    {
      "epoch": 0.5258687337239584,
      "grad_norm": 13.87100887298584,
      "learning_rate": 2.300830651900989e-06,
      "loss": 3.3369,
      "step": 51695
    },
    {
      "epoch": 0.5259195963541666,
      "grad_norm": 7.698983669281006,
      "learning_rate": 2.3004320445390126e-06,
      "loss": 3.2423,
      "step": 51700
    },
    {
      "epoch": 0.525970458984375,
      "grad_norm": 12.41983413696289,
      "learning_rate": 2.3000334422829344e-06,
      "loss": 3.0699,
      "step": 51705
    },
    {
      "epoch": 0.5260213216145834,
      "grad_norm": 10.423108100891113,
      "learning_rate": 2.299634845142954e-06,
      "loss": 3.4923,
      "step": 51710
    },
    {
      "epoch": 0.5260721842447916,
      "grad_norm": 9.7667236328125,
      "learning_rate": 2.2992362531292674e-06,
      "loss": 3.4751,
      "step": 51715
    },
    {
      "epoch": 0.526123046875,
      "grad_norm": 7.892971992492676,
      "learning_rate": 2.2988376662520744e-06,
      "loss": 3.2019,
      "step": 51720
    },
    {
      "epoch": 0.5261739095052084,
      "grad_norm": 10.178163528442383,
      "learning_rate": 2.298439084521572e-06,
      "loss": 3.52,
      "step": 51725
    },
    {
      "epoch": 0.5262247721354166,
      "grad_norm": 12.569657325744629,
      "learning_rate": 2.2980405079479577e-06,
      "loss": 2.9527,
      "step": 51730
    },
    {
      "epoch": 0.526275634765625,
      "grad_norm": 12.244410514831543,
      "learning_rate": 2.297641936541429e-06,
      "loss": 3.2829,
      "step": 51735
    },
    {
      "epoch": 0.5263264973958334,
      "grad_norm": 8.741074562072754,
      "learning_rate": 2.2972433703121835e-06,
      "loss": 3.2828,
      "step": 51740
    },
    {
      "epoch": 0.5263773600260416,
      "grad_norm": 12.224329948425293,
      "learning_rate": 2.2968448092704186e-06,
      "loss": 3.366,
      "step": 51745
    },
    {
      "epoch": 0.52642822265625,
      "grad_norm": 15.625667572021484,
      "learning_rate": 2.2964462534263313e-06,
      "loss": 3.4129,
      "step": 51750
    },
    {
      "epoch": 0.5264790852864584,
      "grad_norm": 14.998575210571289,
      "learning_rate": 2.296047702790117e-06,
      "loss": 3.1494,
      "step": 51755
    },
    {
      "epoch": 0.5265299479166666,
      "grad_norm": 11.03321647644043,
      "learning_rate": 2.295649157371975e-06,
      "loss": 3.4301,
      "step": 51760
    },
    {
      "epoch": 0.526580810546875,
      "grad_norm": 12.48939037322998,
      "learning_rate": 2.295250617182101e-06,
      "loss": 3.3599,
      "step": 51765
    },
    {
      "epoch": 0.5266316731770834,
      "grad_norm": 10.886321067810059,
      "learning_rate": 2.2948520822306906e-06,
      "loss": 3.5091,
      "step": 51770
    },
    {
      "epoch": 0.5266825358072916,
      "grad_norm": 19.278493881225586,
      "learning_rate": 2.294453552527941e-06,
      "loss": 3.2808,
      "step": 51775
    },
    {
      "epoch": 0.5267333984375,
      "grad_norm": 11.889963150024414,
      "learning_rate": 2.29405502808405e-06,
      "loss": 3.1522,
      "step": 51780
    },
    {
      "epoch": 0.5267842610677084,
      "grad_norm": 13.487797737121582,
      "learning_rate": 2.293656508909211e-06,
      "loss": 3.5732,
      "step": 51785
    },
    {
      "epoch": 0.5268351236979166,
      "grad_norm": 13.162525177001953,
      "learning_rate": 2.2932579950136215e-06,
      "loss": 3.6082,
      "step": 51790
    },
    {
      "epoch": 0.526885986328125,
      "grad_norm": 9.454289436340332,
      "learning_rate": 2.292859486407477e-06,
      "loss": 3.153,
      "step": 51795
    },
    {
      "epoch": 0.5269368489583334,
      "grad_norm": 9.60610294342041,
      "learning_rate": 2.2924609831009745e-06,
      "loss": 3.205,
      "step": 51800
    },
    {
      "epoch": 0.5269877115885416,
      "grad_norm": 8.241484642028809,
      "learning_rate": 2.2920624851043077e-06,
      "loss": 3.2567,
      "step": 51805
    },
    {
      "epoch": 0.52703857421875,
      "grad_norm": 9.547820091247559,
      "learning_rate": 2.2916639924276727e-06,
      "loss": 3.6178,
      "step": 51810
    },
    {
      "epoch": 0.5270894368489584,
      "grad_norm": 13.794641494750977,
      "learning_rate": 2.2912655050812657e-06,
      "loss": 3.4859,
      "step": 51815
    },
    {
      "epoch": 0.5271402994791666,
      "grad_norm": 16.587139129638672,
      "learning_rate": 2.290867023075281e-06,
      "loss": 3.2263,
      "step": 51820
    },
    {
      "epoch": 0.527191162109375,
      "grad_norm": 16.068571090698242,
      "learning_rate": 2.2904685464199144e-06,
      "loss": 3.5605,
      "step": 51825
    },
    {
      "epoch": 0.5272420247395834,
      "grad_norm": 13.112360000610352,
      "learning_rate": 2.2900700751253595e-06,
      "loss": 3.6708,
      "step": 51830
    },
    {
      "epoch": 0.5272928873697916,
      "grad_norm": 10.92347526550293,
      "learning_rate": 2.2896716092018125e-06,
      "loss": 3.278,
      "step": 51835
    },
    {
      "epoch": 0.52734375,
      "grad_norm": 11.687875747680664,
      "learning_rate": 2.289273148659468e-06,
      "loss": 3.218,
      "step": 51840
    },
    {
      "epoch": 0.5273946126302084,
      "grad_norm": 12.249234199523926,
      "learning_rate": 2.2888746935085194e-06,
      "loss": 3.1509,
      "step": 51845
    },
    {
      "epoch": 0.5274454752604166,
      "grad_norm": 11.752399444580078,
      "learning_rate": 2.2884762437591614e-06,
      "loss": 3.3611,
      "step": 51850
    },
    {
      "epoch": 0.527496337890625,
      "grad_norm": 11.20430850982666,
      "learning_rate": 2.2880777994215893e-06,
      "loss": 3.5416,
      "step": 51855
    },
    {
      "epoch": 0.5275472005208334,
      "grad_norm": 10.635947227478027,
      "learning_rate": 2.287679360505996e-06,
      "loss": 3.4996,
      "step": 51860
    },
    {
      "epoch": 0.5275980631510416,
      "grad_norm": 12.224534034729004,
      "learning_rate": 2.2872809270225756e-06,
      "loss": 3.6393,
      "step": 51865
    },
    {
      "epoch": 0.52764892578125,
      "grad_norm": 9.653603553771973,
      "learning_rate": 2.2868824989815235e-06,
      "loss": 3.3679,
      "step": 51870
    },
    {
      "epoch": 0.5276997884114584,
      "grad_norm": 15.11462688446045,
      "learning_rate": 2.286484076393031e-06,
      "loss": 3.3414,
      "step": 51875
    },
    {
      "epoch": 0.5277506510416666,
      "grad_norm": 9.089104652404785,
      "learning_rate": 2.2860856592672937e-06,
      "loss": 3.5182,
      "step": 51880
    },
    {
      "epoch": 0.527801513671875,
      "grad_norm": 11.646150588989258,
      "learning_rate": 2.285687247614503e-06,
      "loss": 3.7024,
      "step": 51885
    },
    {
      "epoch": 0.5278523763020834,
      "grad_norm": 11.647123336791992,
      "learning_rate": 2.285288841444854e-06,
      "loss": 2.8685,
      "step": 51890
    },
    {
      "epoch": 0.5279032389322916,
      "grad_norm": 10.295969009399414,
      "learning_rate": 2.2848904407685392e-06,
      "loss": 3.3256,
      "step": 51895
    },
    {
      "epoch": 0.5279541015625,
      "grad_norm": 13.555588722229004,
      "learning_rate": 2.2844920455957515e-06,
      "loss": 3.3036,
      "step": 51900
    },
    {
      "epoch": 0.5280049641927084,
      "grad_norm": 16.191238403320312,
      "learning_rate": 2.2840936559366827e-06,
      "loss": 3.0817,
      "step": 51905
    },
    {
      "epoch": 0.5280558268229166,
      "grad_norm": 12.715621948242188,
      "learning_rate": 2.283695271801528e-06,
      "loss": 3.6958,
      "step": 51910
    },
    {
      "epoch": 0.528106689453125,
      "grad_norm": 15.330379486083984,
      "learning_rate": 2.2832968932004777e-06,
      "loss": 3.2108,
      "step": 51915
    },
    {
      "epoch": 0.5281575520833334,
      "grad_norm": 13.766459465026855,
      "learning_rate": 2.2828985201437257e-06,
      "loss": 3.4376,
      "step": 51920
    },
    {
      "epoch": 0.5282084147135416,
      "grad_norm": 11.895289421081543,
      "learning_rate": 2.2825001526414626e-06,
      "loss": 2.946,
      "step": 51925
    },
    {
      "epoch": 0.52825927734375,
      "grad_norm": 12.957642555236816,
      "learning_rate": 2.282101790703882e-06,
      "loss": 3.0694,
      "step": 51930
    },
    {
      "epoch": 0.5283101399739584,
      "grad_norm": 12.726977348327637,
      "learning_rate": 2.281703434341176e-06,
      "loss": 3.1,
      "step": 51935
    },
    {
      "epoch": 0.5283610026041666,
      "grad_norm": 10.04681396484375,
      "learning_rate": 2.281305083563535e-06,
      "loss": 3.5245,
      "step": 51940
    },
    {
      "epoch": 0.528411865234375,
      "grad_norm": 9.157989501953125,
      "learning_rate": 2.2809067383811514e-06,
      "loss": 3.4185,
      "step": 51945
    },
    {
      "epoch": 0.5284627278645834,
      "grad_norm": 11.363675117492676,
      "learning_rate": 2.2805083988042178e-06,
      "loss": 2.9623,
      "step": 51950
    },
    {
      "epoch": 0.5285135904947916,
      "grad_norm": 18.07217025756836,
      "learning_rate": 2.280110064842925e-06,
      "loss": 3.286,
      "step": 51955
    },
    {
      "epoch": 0.528564453125,
      "grad_norm": 11.635565757751465,
      "learning_rate": 2.279711736507463e-06,
      "loss": 3.6186,
      "step": 51960
    },
    {
      "epoch": 0.5286153157552084,
      "grad_norm": 14.110377311706543,
      "learning_rate": 2.2793134138080243e-06,
      "loss": 2.961,
      "step": 51965
    },
    {
      "epoch": 0.5286661783854166,
      "grad_norm": 9.997685432434082,
      "learning_rate": 2.2789150967548e-06,
      "loss": 3.1046,
      "step": 51970
    },
    {
      "epoch": 0.528717041015625,
      "grad_norm": 10.754714012145996,
      "learning_rate": 2.2785167853579808e-06,
      "loss": 3.6072,
      "step": 51975
    },
    {
      "epoch": 0.5287679036458334,
      "grad_norm": 9.805929183959961,
      "learning_rate": 2.2781184796277557e-06,
      "loss": 3.3923,
      "step": 51980
    },
    {
      "epoch": 0.5288187662760416,
      "grad_norm": 9.262873649597168,
      "learning_rate": 2.2777201795743187e-06,
      "loss": 3.3945,
      "step": 51985
    },
    {
      "epoch": 0.52886962890625,
      "grad_norm": 9.637592315673828,
      "learning_rate": 2.277321885207857e-06,
      "loss": 3.1716,
      "step": 51990
    },
    {
      "epoch": 0.5289204915364584,
      "grad_norm": 7.525077819824219,
      "learning_rate": 2.2769235965385632e-06,
      "loss": 3.2199,
      "step": 51995
    },
    {
      "epoch": 0.5289713541666666,
      "grad_norm": 11.829056739807129,
      "learning_rate": 2.276525313576625e-06,
      "loss": 3.1419,
      "step": 52000
    },
    {
      "epoch": 0.529022216796875,
      "grad_norm": 10.279789924621582,
      "learning_rate": 2.2761270363322338e-06,
      "loss": 2.9866,
      "step": 52005
    },
    {
      "epoch": 0.5290730794270834,
      "grad_norm": 17.86949920654297,
      "learning_rate": 2.2757287648155803e-06,
      "loss": 3.3813,
      "step": 52010
    },
    {
      "epoch": 0.5291239420572916,
      "grad_norm": 8.782398223876953,
      "learning_rate": 2.2753304990368525e-06,
      "loss": 3.3566,
      "step": 52015
    },
    {
      "epoch": 0.5291748046875,
      "grad_norm": 10.083893775939941,
      "learning_rate": 2.27493223900624e-06,
      "loss": 3.37,
      "step": 52020
    },
    {
      "epoch": 0.5292256673177084,
      "grad_norm": 12.253583908081055,
      "learning_rate": 2.274533984733934e-06,
      "loss": 3.3056,
      "step": 52025
    },
    {
      "epoch": 0.5292765299479166,
      "grad_norm": 10.968459129333496,
      "learning_rate": 2.274135736230122e-06,
      "loss": 3.1398,
      "step": 52030
    },
    {
      "epoch": 0.529327392578125,
      "grad_norm": 13.678936004638672,
      "learning_rate": 2.2737374935049946e-06,
      "loss": 3.0078,
      "step": 52035
    },
    {
      "epoch": 0.5293782552083334,
      "grad_norm": 14.763940811157227,
      "learning_rate": 2.2733392565687377e-06,
      "loss": 3.3282,
      "step": 52040
    },
    {
      "epoch": 0.5294291178385416,
      "grad_norm": 8.088820457458496,
      "learning_rate": 2.2729410254315437e-06,
      "loss": 3.1962,
      "step": 52045
    },
    {
      "epoch": 0.52947998046875,
      "grad_norm": 13.720199584960938,
      "learning_rate": 2.2725428001036003e-06,
      "loss": 3.3752,
      "step": 52050
    },
    {
      "epoch": 0.5295308430989584,
      "grad_norm": 13.48266315460205,
      "learning_rate": 2.272144580595094e-06,
      "loss": 3.7975,
      "step": 52055
    },
    {
      "epoch": 0.5295817057291666,
      "grad_norm": 11.64840316772461,
      "learning_rate": 2.271746366916215e-06,
      "loss": 3.4526,
      "step": 52060
    },
    {
      "epoch": 0.529632568359375,
      "grad_norm": 18.696725845336914,
      "learning_rate": 2.2713481590771516e-06,
      "loss": 3.4287,
      "step": 52065
    },
    {
      "epoch": 0.5296834309895834,
      "grad_norm": 14.861400604248047,
      "learning_rate": 2.270949957088091e-06,
      "loss": 3.342,
      "step": 52070
    },
    {
      "epoch": 0.5297342936197916,
      "grad_norm": 7.6815876960754395,
      "learning_rate": 2.270551760959221e-06,
      "loss": 3.5648,
      "step": 52075
    },
    {
      "epoch": 0.52978515625,
      "grad_norm": 9.49208927154541,
      "learning_rate": 2.270153570700731e-06,
      "loss": 3.0591,
      "step": 52080
    },
    {
      "epoch": 0.5298360188802084,
      "grad_norm": 11.423577308654785,
      "learning_rate": 2.269755386322807e-06,
      "loss": 3.3461,
      "step": 52085
    },
    {
      "epoch": 0.5298868815104166,
      "grad_norm": 8.663924217224121,
      "learning_rate": 2.2693572078356367e-06,
      "loss": 3.2918,
      "step": 52090
    },
    {
      "epoch": 0.529937744140625,
      "grad_norm": 17.198579788208008,
      "learning_rate": 2.2689590352494063e-06,
      "loss": 2.9642,
      "step": 52095
    },
    {
      "epoch": 0.5299886067708334,
      "grad_norm": 7.101471424102783,
      "learning_rate": 2.268560868574306e-06,
      "loss": 3.2453,
      "step": 52100
    },
    {
      "epoch": 0.5300394694010416,
      "grad_norm": 12.237752914428711,
      "learning_rate": 2.2681627078205205e-06,
      "loss": 3.2128,
      "step": 52105
    },
    {
      "epoch": 0.53009033203125,
      "grad_norm": 9.680429458618164,
      "learning_rate": 2.2677645529982375e-06,
      "loss": 3.3987,
      "step": 52110
    },
    {
      "epoch": 0.5301411946614584,
      "grad_norm": 10.857959747314453,
      "learning_rate": 2.2673664041176425e-06,
      "loss": 3.1378,
      "step": 52115
    },
    {
      "epoch": 0.5301920572916666,
      "grad_norm": 10.731316566467285,
      "learning_rate": 2.266968261188923e-06,
      "loss": 3.4948,
      "step": 52120
    },
    {
      "epoch": 0.530242919921875,
      "grad_norm": 14.895269393920898,
      "learning_rate": 2.2665701242222666e-06,
      "loss": 3.3257,
      "step": 52125
    },
    {
      "epoch": 0.5302937825520834,
      "grad_norm": 17.698509216308594,
      "learning_rate": 2.266171993227857e-06,
      "loss": 3.9778,
      "step": 52130
    },
    {
      "epoch": 0.5303446451822916,
      "grad_norm": 12.754210472106934,
      "learning_rate": 2.2657738682158812e-06,
      "loss": 3.0264,
      "step": 52135
    },
    {
      "epoch": 0.5303955078125,
      "grad_norm": 12.280755996704102,
      "learning_rate": 2.265375749196527e-06,
      "loss": 3.2183,
      "step": 52140
    },
    {
      "epoch": 0.5304463704427084,
      "grad_norm": 9.895425796508789,
      "learning_rate": 2.2649776361799777e-06,
      "loss": 3.253,
      "step": 52145
    },
    {
      "epoch": 0.5304972330729166,
      "grad_norm": 14.172664642333984,
      "learning_rate": 2.26457952917642e-06,
      "loss": 3.5023,
      "step": 52150
    },
    {
      "epoch": 0.530548095703125,
      "grad_norm": 17.2047119140625,
      "learning_rate": 2.264181428196039e-06,
      "loss": 2.8805,
      "step": 52155
    },
    {
      "epoch": 0.5305989583333334,
      "grad_norm": 9.435441970825195,
      "learning_rate": 2.2637833332490203e-06,
      "loss": 3.1764,
      "step": 52160
    },
    {
      "epoch": 0.5306498209635416,
      "grad_norm": 12.391535758972168,
      "learning_rate": 2.2633852443455497e-06,
      "loss": 3.2382,
      "step": 52165
    },
    {
      "epoch": 0.53070068359375,
      "grad_norm": 13.898744583129883,
      "learning_rate": 2.2629871614958104e-06,
      "loss": 3.1006,
      "step": 52170
    },
    {
      "epoch": 0.5307515462239584,
      "grad_norm": 12.971320152282715,
      "learning_rate": 2.262589084709989e-06,
      "loss": 3.5297,
      "step": 52175
    },
    {
      "epoch": 0.5308024088541666,
      "grad_norm": 8.470605850219727,
      "learning_rate": 2.26219101399827e-06,
      "loss": 3.3854,
      "step": 52180
    },
    {
      "epoch": 0.530853271484375,
      "grad_norm": 10.319451332092285,
      "learning_rate": 2.261792949370837e-06,
      "loss": 3.3321,
      "step": 52185
    },
    {
      "epoch": 0.5309041341145834,
      "grad_norm": 8.604473114013672,
      "learning_rate": 2.2613948908378748e-06,
      "loss": 3.3155,
      "step": 52190
    },
    {
      "epoch": 0.5309549967447916,
      "grad_norm": 9.907164573669434,
      "learning_rate": 2.260996838409568e-06,
      "loss": 3.2708,
      "step": 52195
    },
    {
      "epoch": 0.531005859375,
      "grad_norm": 9.253422737121582,
      "learning_rate": 2.2605987920961004e-06,
      "loss": 3.3554,
      "step": 52200
    },
    {
      "epoch": 0.5310567220052084,
      "grad_norm": 11.168255805969238,
      "learning_rate": 2.260200751907657e-06,
      "loss": 3.3023,
      "step": 52205
    },
    {
      "epoch": 0.5311075846354166,
      "grad_norm": 6.434413433074951,
      "learning_rate": 2.2598027178544186e-06,
      "loss": 3.239,
      "step": 52210
    },
    {
      "epoch": 0.531158447265625,
      "grad_norm": 10.66589641571045,
      "learning_rate": 2.259404689946572e-06,
      "loss": 3.4369,
      "step": 52215
    },
    {
      "epoch": 0.5312093098958334,
      "grad_norm": 11.483353614807129,
      "learning_rate": 2.259006668194299e-06,
      "loss": 3.2524,
      "step": 52220
    },
    {
      "epoch": 0.5312601725260416,
      "grad_norm": 14.248106002807617,
      "learning_rate": 2.2586086526077837e-06,
      "loss": 3.3969,
      "step": 52225
    },
    {
      "epoch": 0.53131103515625,
      "grad_norm": 16.453853607177734,
      "learning_rate": 2.258210643197208e-06,
      "loss": 2.9747,
      "step": 52230
    },
    {
      "epoch": 0.5313618977864584,
      "grad_norm": 11.427420616149902,
      "learning_rate": 2.2578126399727564e-06,
      "loss": 3.4672,
      "step": 52235
    },
    {
      "epoch": 0.5314127604166666,
      "grad_norm": 12.872052192687988,
      "learning_rate": 2.257414642944611e-06,
      "loss": 3.1296,
      "step": 52240
    },
    {
      "epoch": 0.531463623046875,
      "grad_norm": 17.460342407226562,
      "learning_rate": 2.2570166521229545e-06,
      "loss": 3.1822,
      "step": 52245
    },
    {
      "epoch": 0.5315144856770834,
      "grad_norm": 15.031963348388672,
      "learning_rate": 2.256618667517968e-06,
      "loss": 3.5697,
      "step": 52250
    },
    {
      "epoch": 0.5315653483072916,
      "grad_norm": 18.319963455200195,
      "learning_rate": 2.2562206891398367e-06,
      "loss": 3.3926,
      "step": 52255
    },
    {
      "epoch": 0.5316162109375,
      "grad_norm": 15.066197395324707,
      "learning_rate": 2.2558227169987407e-06,
      "loss": 3.341,
      "step": 52260
    },
    {
      "epoch": 0.5316670735677084,
      "grad_norm": 9.521345138549805,
      "learning_rate": 2.255424751104862e-06,
      "loss": 3.0247,
      "step": 52265
    },
    {
      "epoch": 0.5317179361979166,
      "grad_norm": 16.426836013793945,
      "learning_rate": 2.2550267914683845e-06,
      "loss": 3.0748,
      "step": 52270
    },
    {
      "epoch": 0.531768798828125,
      "grad_norm": 7.882839679718018,
      "learning_rate": 2.2546288380994876e-06,
      "loss": 3.1708,
      "step": 52275
    },
    {
      "epoch": 0.5318196614583334,
      "grad_norm": 17.430219650268555,
      "learning_rate": 2.2542308910083546e-06,
      "loss": 3.5405,
      "step": 52280
    },
    {
      "epoch": 0.5318705240885416,
      "grad_norm": 12.507336616516113,
      "learning_rate": 2.2538329502051647e-06,
      "loss": 3.4098,
      "step": 52285
    },
    {
      "epoch": 0.53192138671875,
      "grad_norm": 16.876689910888672,
      "learning_rate": 2.253435015700101e-06,
      "loss": 3.5465,
      "step": 52290
    },
    {
      "epoch": 0.5319722493489584,
      "grad_norm": 13.726040840148926,
      "learning_rate": 2.253037087503345e-06,
      "loss": 3.726,
      "step": 52295
    },
    {
      "epoch": 0.5320231119791666,
      "grad_norm": 9.711737632751465,
      "learning_rate": 2.252639165625075e-06,
      "loss": 3.4336,
      "step": 52300
    },
    {
      "epoch": 0.532073974609375,
      "grad_norm": 14.651636123657227,
      "learning_rate": 2.2522412500754733e-06,
      "loss": 3.1465,
      "step": 52305
    },
    {
      "epoch": 0.5321248372395834,
      "grad_norm": 8.425837516784668,
      "learning_rate": 2.251843340864722e-06,
      "loss": 3.1489,
      "step": 52310
    },
    {
      "epoch": 0.5321756998697916,
      "grad_norm": 16.434213638305664,
      "learning_rate": 2.251445438002999e-06,
      "loss": 3.6863,
      "step": 52315
    },
    {
      "epoch": 0.5322265625,
      "grad_norm": 12.590494155883789,
      "learning_rate": 2.251047541500486e-06,
      "loss": 3.3995,
      "step": 52320
    },
    {
      "epoch": 0.5322774251302084,
      "grad_norm": 17.571537017822266,
      "learning_rate": 2.250649651367362e-06,
      "loss": 3.4516,
      "step": 52325
    },
    {
      "epoch": 0.5323282877604166,
      "grad_norm": 13.119302749633789,
      "learning_rate": 2.2502517676138074e-06,
      "loss": 3.4861,
      "step": 52330
    },
    {
      "epoch": 0.532379150390625,
      "grad_norm": 13.304254531860352,
      "learning_rate": 2.2498538902500035e-06,
      "loss": 3.2404,
      "step": 52335
    },
    {
      "epoch": 0.5324300130208334,
      "grad_norm": 13.901344299316406,
      "learning_rate": 2.2494560192861273e-06,
      "loss": 3.4191,
      "step": 52340
    },
    {
      "epoch": 0.5324808756510416,
      "grad_norm": 9.741520881652832,
      "learning_rate": 2.2490581547323593e-06,
      "loss": 3.2748,
      "step": 52345
    },
    {
      "epoch": 0.53253173828125,
      "grad_norm": 9.868925094604492,
      "learning_rate": 2.24866029659888e-06,
      "loss": 3.1799,
      "step": 52350
    },
    {
      "epoch": 0.5325826009114584,
      "grad_norm": 9.409064292907715,
      "learning_rate": 2.248262444895867e-06,
      "loss": 3.519,
      "step": 52355
    },
    {
      "epoch": 0.5326334635416666,
      "grad_norm": 13.42024040222168,
      "learning_rate": 2.2478645996334986e-06,
      "loss": 3.3728,
      "step": 52360
    },
    {
      "epoch": 0.532684326171875,
      "grad_norm": 12.795633316040039,
      "learning_rate": 2.247466760821956e-06,
      "loss": 3.6022,
      "step": 52365
    },
    {
      "epoch": 0.5327351888020834,
      "grad_norm": 15.062945365905762,
      "learning_rate": 2.247068928471416e-06,
      "loss": 3.4308,
      "step": 52370
    },
    {
      "epoch": 0.5327860514322916,
      "grad_norm": 14.683167457580566,
      "learning_rate": 2.2466711025920583e-06,
      "loss": 3.3802,
      "step": 52375
    },
    {
      "epoch": 0.5328369140625,
      "grad_norm": 8.117255210876465,
      "learning_rate": 2.246273283194059e-06,
      "loss": 3.3244,
      "step": 52380
    },
    {
      "epoch": 0.5328877766927084,
      "grad_norm": 9.679817199707031,
      "learning_rate": 2.2458754702875983e-06,
      "loss": 3.496,
      "step": 52385
    },
    {
      "epoch": 0.5329386393229166,
      "grad_norm": 15.520284652709961,
      "learning_rate": 2.245477663882854e-06,
      "loss": 3.8227,
      "step": 52390
    },
    {
      "epoch": 0.532989501953125,
      "grad_norm": 9.276087760925293,
      "learning_rate": 2.2450798639900026e-06,
      "loss": 3.3125,
      "step": 52395
    },
    {
      "epoch": 0.5330403645833334,
      "grad_norm": 8.678460121154785,
      "learning_rate": 2.244682070619222e-06,
      "loss": 3.4275,
      "step": 52400
    },
    {
      "epoch": 0.5330912272135416,
      "grad_norm": 12.090228080749512,
      "learning_rate": 2.244284283780691e-06,
      "loss": 3.719,
      "step": 52405
    },
    {
      "epoch": 0.53314208984375,
      "grad_norm": 15.500432968139648,
      "learning_rate": 2.243886503484586e-06,
      "loss": 3.1714,
      "step": 52410
    },
    {
      "epoch": 0.5331929524739584,
      "grad_norm": 14.30467414855957,
      "learning_rate": 2.243488729741084e-06,
      "loss": 3.1552,
      "step": 52415
    },
    {
      "epoch": 0.5332438151041666,
      "grad_norm": 15.131646156311035,
      "learning_rate": 2.243090962560361e-06,
      "loss": 3.0603,
      "step": 52420
    },
    {
      "epoch": 0.533294677734375,
      "grad_norm": 12.336730003356934,
      "learning_rate": 2.242693201952596e-06,
      "loss": 3.3295,
      "step": 52425
    },
    {
      "epoch": 0.5333455403645834,
      "grad_norm": 9.866353988647461,
      "learning_rate": 2.2422954479279646e-06,
      "loss": 2.873,
      "step": 52430
    },
    {
      "epoch": 0.5333964029947916,
      "grad_norm": 10.846232414245605,
      "learning_rate": 2.241897700496643e-06,
      "loss": 3.5189,
      "step": 52435
    },
    {
      "epoch": 0.533447265625,
      "grad_norm": 11.261388778686523,
      "learning_rate": 2.2414999596688066e-06,
      "loss": 3.1236,
      "step": 52440
    },
    {
      "epoch": 0.5334981282552084,
      "grad_norm": 14.13904857635498,
      "learning_rate": 2.241102225454633e-06,
      "loss": 3.3259,
      "step": 52445
    },
    {
      "epoch": 0.5335489908854166,
      "grad_norm": 8.11526870727539,
      "learning_rate": 2.240704497864298e-06,
      "loss": 3.1958,
      "step": 52450
    },
    {
      "epoch": 0.533599853515625,
      "grad_norm": 14.283095359802246,
      "learning_rate": 2.2403067769079765e-06,
      "loss": 3.111,
      "step": 52455
    },
    {
      "epoch": 0.5336507161458334,
      "grad_norm": 9.867772102355957,
      "learning_rate": 2.239909062595844e-06,
      "loss": 3.2556,
      "step": 52460
    },
    {
      "epoch": 0.5337015787760416,
      "grad_norm": 15.871087074279785,
      "learning_rate": 2.239511354938078e-06,
      "loss": 2.8123,
      "step": 52465
    },
    {
      "epoch": 0.53375244140625,
      "grad_norm": 12.044535636901855,
      "learning_rate": 2.2391136539448514e-06,
      "loss": 3.3877,
      "step": 52470
    },
    {
      "epoch": 0.5338033040364584,
      "grad_norm": 9.333099365234375,
      "learning_rate": 2.2387159596263396e-06,
      "loss": 3.6512,
      "step": 52475
    },
    {
      "epoch": 0.5338541666666666,
      "grad_norm": 12.61330795288086,
      "learning_rate": 2.238318271992719e-06,
      "loss": 3.3015,
      "step": 52480
    },
    {
      "epoch": 0.533905029296875,
      "grad_norm": 8.70833969116211,
      "learning_rate": 2.2379205910541627e-06,
      "loss": 3.2452,
      "step": 52485
    },
    {
      "epoch": 0.5339558919270834,
      "grad_norm": 9.98299503326416,
      "learning_rate": 2.237522916820847e-06,
      "loss": 3.3227,
      "step": 52490
    },
    {
      "epoch": 0.5340067545572916,
      "grad_norm": 15.434283256530762,
      "learning_rate": 2.237125249302944e-06,
      "loss": 3.4599,
      "step": 52495
    },
    {
      "epoch": 0.5340576171875,
      "grad_norm": 16.889253616333008,
      "learning_rate": 2.2367275885106294e-06,
      "loss": 3.2198,
      "step": 52500
    },
    {
      "epoch": 0.5341084798177084,
      "grad_norm": 9.48607063293457,
      "learning_rate": 2.236329934454078e-06,
      "loss": 3.1513,
      "step": 52505
    },
    {
      "epoch": 0.5341593424479166,
      "grad_norm": 16.795494079589844,
      "learning_rate": 2.235932287143462e-06,
      "loss": 3.3567,
      "step": 52510
    },
    {
      "epoch": 0.534210205078125,
      "grad_norm": 14.146149635314941,
      "learning_rate": 2.235534646588955e-06,
      "loss": 3.3844,
      "step": 52515
    },
    {
      "epoch": 0.5342610677083334,
      "grad_norm": 12.229581832885742,
      "learning_rate": 2.235137012800733e-06,
      "loss": 3.6554,
      "step": 52520
    },
    {
      "epoch": 0.5343119303385416,
      "grad_norm": 11.907014846801758,
      "learning_rate": 2.2347393857889667e-06,
      "loss": 3.3006,
      "step": 52525
    },
    {
      "epoch": 0.53436279296875,
      "grad_norm": 13.531723022460938,
      "learning_rate": 2.234341765563831e-06,
      "loss": 3.2998,
      "step": 52530
    },
    {
      "epoch": 0.5344136555989584,
      "grad_norm": 15.982670783996582,
      "learning_rate": 2.2339441521354977e-06,
      "loss": 3.3834,
      "step": 52535
    },
    {
      "epoch": 0.5344645182291666,
      "grad_norm": 15.747007369995117,
      "learning_rate": 2.23354654551414e-06,
      "loss": 3.299,
      "step": 52540
    },
    {
      "epoch": 0.534515380859375,
      "grad_norm": 11.762879371643066,
      "learning_rate": 2.233148945709932e-06,
      "loss": 3.095,
      "step": 52545
    },
    {
      "epoch": 0.5345662434895834,
      "grad_norm": 13.616257667541504,
      "learning_rate": 2.232751352733044e-06,
      "loss": 3.2841,
      "step": 52550
    },
    {
      "epoch": 0.5346171061197916,
      "grad_norm": 15.09197998046875,
      "learning_rate": 2.232353766593649e-06,
      "loss": 3.4964,
      "step": 52555
    },
    {
      "epoch": 0.53466796875,
      "grad_norm": 9.675505638122559,
      "learning_rate": 2.2319561873019198e-06,
      "loss": 3.1382,
      "step": 52560
    },
    {
      "epoch": 0.5347188313802084,
      "grad_norm": 13.677727699279785,
      "learning_rate": 2.2315586148680283e-06,
      "loss": 3.1489,
      "step": 52565
    },
    {
      "epoch": 0.5347696940104166,
      "grad_norm": 8.051255226135254,
      "learning_rate": 2.2311610493021455e-06,
      "loss": 3.1315,
      "step": 52570
    },
    {
      "epoch": 0.534820556640625,
      "grad_norm": 13.022601127624512,
      "learning_rate": 2.230763490614444e-06,
      "loss": 3.163,
      "step": 52575
    },
    {
      "epoch": 0.5348714192708334,
      "grad_norm": 14.174161911010742,
      "learning_rate": 2.230365938815095e-06,
      "loss": 3.1091,
      "step": 52580
    },
    {
      "epoch": 0.5349222819010416,
      "grad_norm": 8.802834510803223,
      "learning_rate": 2.229968393914269e-06,
      "loss": 3.4788,
      "step": 52585
    },
    {
      "epoch": 0.53497314453125,
      "grad_norm": 7.659459590911865,
      "learning_rate": 2.2295708559221373e-06,
      "loss": 3.6628,
      "step": 52590
    },
    {
      "epoch": 0.5350240071614584,
      "grad_norm": 10.533792495727539,
      "learning_rate": 2.2291733248488723e-06,
      "loss": 3.253,
      "step": 52595
    },
    {
      "epoch": 0.5350748697916666,
      "grad_norm": 12.855596542358398,
      "learning_rate": 2.228775800704643e-06,
      "loss": 3.3276,
      "step": 52600
    },
    {
      "epoch": 0.535125732421875,
      "grad_norm": 12.619338989257812,
      "learning_rate": 2.228378283499621e-06,
      "loss": 3.0476,
      "step": 52605
    },
    {
      "epoch": 0.5351765950520834,
      "grad_norm": 10.090911865234375,
      "learning_rate": 2.2279807732439753e-06,
      "loss": 3.5128,
      "step": 52610
    },
    {
      "epoch": 0.5352274576822916,
      "grad_norm": 9.089913368225098,
      "learning_rate": 2.227583269947877e-06,
      "loss": 3.2851,
      "step": 52615
    },
    {
      "epoch": 0.5352783203125,
      "grad_norm": 17.8261775970459,
      "learning_rate": 2.2271857736214973e-06,
      "loss": 4.0462,
      "step": 52620
    },
    {
      "epoch": 0.5353291829427084,
      "grad_norm": 11.043100357055664,
      "learning_rate": 2.2267882842750044e-06,
      "loss": 3.5272,
      "step": 52625
    },
    {
      "epoch": 0.5353800455729166,
      "grad_norm": 10.309991836547852,
      "learning_rate": 2.226390801918568e-06,
      "loss": 3.1993,
      "step": 52630
    },
    {
      "epoch": 0.535430908203125,
      "grad_norm": 8.734068870544434,
      "learning_rate": 2.2259933265623584e-06,
      "loss": 3.4426,
      "step": 52635
    },
    {
      "epoch": 0.5354817708333334,
      "grad_norm": 13.374344825744629,
      "learning_rate": 2.2255958582165447e-06,
      "loss": 3.252,
      "step": 52640
    },
    {
      "epoch": 0.5355326334635416,
      "grad_norm": 13.33709716796875,
      "learning_rate": 2.2251983968912968e-06,
      "loss": 3.2941,
      "step": 52645
    },
    {
      "epoch": 0.53558349609375,
      "grad_norm": 13.283416748046875,
      "learning_rate": 2.224800942596781e-06,
      "loss": 3.4063,
      "step": 52650
    },
    {
      "epoch": 0.5356343587239584,
      "grad_norm": 12.268871307373047,
      "learning_rate": 2.2244034953431685e-06,
      "loss": 3.1474,
      "step": 52655
    },
    {
      "epoch": 0.5356852213541666,
      "grad_norm": 17.27332305908203,
      "learning_rate": 2.2240060551406276e-06,
      "loss": 3.7438,
      "step": 52660
    },
    {
      "epoch": 0.535736083984375,
      "grad_norm": 8.489108085632324,
      "learning_rate": 2.223608621999326e-06,
      "loss": 3.0981,
      "step": 52665
    },
    {
      "epoch": 0.5357869466145834,
      "grad_norm": 14.006270408630371,
      "learning_rate": 2.2232111959294325e-06,
      "loss": 3.0433,
      "step": 52670
    },
    {
      "epoch": 0.5358378092447916,
      "grad_norm": 10.502427101135254,
      "learning_rate": 2.2228137769411156e-06,
      "loss": 3.2598,
      "step": 52675
    },
    {
      "epoch": 0.535888671875,
      "grad_norm": 13.823921203613281,
      "learning_rate": 2.2224163650445417e-06,
      "loss": 3.4957,
      "step": 52680
    },
    {
      "epoch": 0.5359395345052084,
      "grad_norm": 12.201239585876465,
      "learning_rate": 2.2220189602498788e-06,
      "loss": 3.3212,
      "step": 52685
    },
    {
      "epoch": 0.5359903971354166,
      "grad_norm": 15.84398078918457,
      "learning_rate": 2.221621562567296e-06,
      "loss": 3.3958,
      "step": 52690
    },
    {
      "epoch": 0.536041259765625,
      "grad_norm": 10.147021293640137,
      "learning_rate": 2.2212241720069592e-06,
      "loss": 3.2461,
      "step": 52695
    },
    {
      "epoch": 0.5360921223958334,
      "grad_norm": 9.781103134155273,
      "learning_rate": 2.220826788579036e-06,
      "loss": 3.3178,
      "step": 52700
    },
    {
      "epoch": 0.5361429850260416,
      "grad_norm": 8.797540664672852,
      "learning_rate": 2.2204294122936925e-06,
      "loss": 3.4735,
      "step": 52705
    },
    {
      "epoch": 0.53619384765625,
      "grad_norm": 12.501797676086426,
      "learning_rate": 2.220032043161097e-06,
      "loss": 3.4536,
      "step": 52710
    },
    {
      "epoch": 0.5362447102864584,
      "grad_norm": 11.115304946899414,
      "learning_rate": 2.219634681191415e-06,
      "loss": 3.3875,
      "step": 52715
    },
    {
      "epoch": 0.5362955729166666,
      "grad_norm": 10.252839088439941,
      "learning_rate": 2.219237326394814e-06,
      "loss": 3.3677,
      "step": 52720
    },
    {
      "epoch": 0.536346435546875,
      "grad_norm": 11.421342849731445,
      "learning_rate": 2.218839978781458e-06,
      "loss": 3.1049,
      "step": 52725
    },
    {
      "epoch": 0.5363972981770834,
      "grad_norm": 12.959375381469727,
      "learning_rate": 2.2184426383615155e-06,
      "loss": 3.1831,
      "step": 52730
    },
    {
      "epoch": 0.5364481608072916,
      "grad_norm": 14.955923080444336,
      "learning_rate": 2.2180453051451515e-06,
      "loss": 3.4734,
      "step": 52735
    },
    {
      "epoch": 0.5364990234375,
      "grad_norm": 9.893755912780762,
      "learning_rate": 2.217647979142531e-06,
      "loss": 3.146,
      "step": 52740
    },
    {
      "epoch": 0.5365498860677084,
      "grad_norm": 10.388473510742188,
      "learning_rate": 2.21725066036382e-06,
      "loss": 3.3632,
      "step": 52745
    },
    {
      "epoch": 0.5366007486979166,
      "grad_norm": 9.242728233337402,
      "learning_rate": 2.216853348819184e-06,
      "loss": 3.7097,
      "step": 52750
    },
    {
      "epoch": 0.536651611328125,
      "grad_norm": 13.42584228515625,
      "learning_rate": 2.2164560445187883e-06,
      "loss": 3.2877,
      "step": 52755
    },
    {
      "epoch": 0.5367024739583334,
      "grad_norm": 12.122518539428711,
      "learning_rate": 2.2160587474727962e-06,
      "loss": 3.1327,
      "step": 52760
    },
    {
      "epoch": 0.5367533365885416,
      "grad_norm": 10.502169609069824,
      "learning_rate": 2.2156614576913755e-06,
      "loss": 3.4177,
      "step": 52765
    },
    {
      "epoch": 0.53680419921875,
      "grad_norm": 12.630999565124512,
      "learning_rate": 2.2152641751846883e-06,
      "loss": 3.1985,
      "step": 52770
    },
    {
      "epoch": 0.5368550618489584,
      "grad_norm": 13.926748275756836,
      "learning_rate": 2.2148668999629e-06,
      "loss": 3.281,
      "step": 52775
    },
    {
      "epoch": 0.5369059244791666,
      "grad_norm": 10.87210464477539,
      "learning_rate": 2.2144696320361735e-06,
      "loss": 3.2454,
      "step": 52780
    },
    {
      "epoch": 0.536956787109375,
      "grad_norm": 14.919941902160645,
      "learning_rate": 2.2140723714146747e-06,
      "loss": 3.4325,
      "step": 52785
    },
    {
      "epoch": 0.5370076497395834,
      "grad_norm": 14.381443977355957,
      "learning_rate": 2.2136751181085666e-06,
      "loss": 3.3376,
      "step": 52790
    },
    {
      "epoch": 0.5370585123697916,
      "grad_norm": 13.070241928100586,
      "learning_rate": 2.213277872128012e-06,
      "loss": 3.3856,
      "step": 52795
    },
    {
      "epoch": 0.537109375,
      "grad_norm": 9.268564224243164,
      "learning_rate": 2.2128806334831747e-06,
      "loss": 3.7792,
      "step": 52800
    },
    {
      "epoch": 0.5371602376302084,
      "grad_norm": 15.848664283752441,
      "learning_rate": 2.2124834021842196e-06,
      "loss": 3.5488,
      "step": 52805
    },
    {
      "epoch": 0.5372111002604166,
      "grad_norm": 10.323749542236328,
      "learning_rate": 2.2120861782413083e-06,
      "loss": 3.3862,
      "step": 52810
    },
    {
      "epoch": 0.537261962890625,
      "grad_norm": 9.335864067077637,
      "learning_rate": 2.2116889616646037e-06,
      "loss": 3.2816,
      "step": 52815
    },
    {
      "epoch": 0.5373128255208334,
      "grad_norm": 12.387421607971191,
      "learning_rate": 2.211291752464268e-06,
      "loss": 3.4637,
      "step": 52820
    },
    {
      "epoch": 0.5373636881510416,
      "grad_norm": 15.235649108886719,
      "learning_rate": 2.2108945506504648e-06,
      "loss": 3.7854,
      "step": 52825
    },
    {
      "epoch": 0.53741455078125,
      "grad_norm": 12.311189651489258,
      "learning_rate": 2.2104973562333567e-06,
      "loss": 3.4655,
      "step": 52830
    },
    {
      "epoch": 0.5374654134114584,
      "grad_norm": 15.438931465148926,
      "learning_rate": 2.2101001692231046e-06,
      "loss": 3.1788,
      "step": 52835
    },
    {
      "epoch": 0.5375162760416666,
      "grad_norm": 13.147708892822266,
      "learning_rate": 2.20970298962987e-06,
      "loss": 3.123,
      "step": 52840
    },
    {
      "epoch": 0.537567138671875,
      "grad_norm": 12.911852836608887,
      "learning_rate": 2.209305817463817e-06,
      "loss": 3.2834,
      "step": 52845
    },
    {
      "epoch": 0.5376180013020834,
      "grad_norm": 15.99446964263916,
      "learning_rate": 2.2089086527351054e-06,
      "loss": 3.5062,
      "step": 52850
    },
    {
      "epoch": 0.5376688639322916,
      "grad_norm": 12.867097854614258,
      "learning_rate": 2.208511495453896e-06,
      "loss": 3.0001,
      "step": 52855
    },
    {
      "epoch": 0.5377197265625,
      "grad_norm": 10.72048568725586,
      "learning_rate": 2.2081143456303517e-06,
      "loss": 3.6095,
      "step": 52860
    },
    {
      "epoch": 0.5377705891927084,
      "grad_norm": 9.181864738464355,
      "learning_rate": 2.2077172032746334e-06,
      "loss": 3.3329,
      "step": 52865
    },
    {
      "epoch": 0.5378214518229166,
      "grad_norm": 14.825196266174316,
      "learning_rate": 2.2073200683969e-06,
      "loss": 3.4629,
      "step": 52870
    },
    {
      "epoch": 0.537872314453125,
      "grad_norm": 8.038079261779785,
      "learning_rate": 2.2069229410073134e-06,
      "loss": 3.4778,
      "step": 52875
    },
    {
      "epoch": 0.5379231770833334,
      "grad_norm": 12.720928192138672,
      "learning_rate": 2.2065258211160345e-06,
      "loss": 3.1995,
      "step": 52880
    },
    {
      "epoch": 0.5379740397135416,
      "grad_norm": 13.668231010437012,
      "learning_rate": 2.2061287087332225e-06,
      "loss": 3.2563,
      "step": 52885
    },
    {
      "epoch": 0.53802490234375,
      "grad_norm": 13.181472778320312,
      "learning_rate": 2.2057316038690383e-06,
      "loss": 3.4019,
      "step": 52890
    },
    {
      "epoch": 0.5380757649739584,
      "grad_norm": 12.291973114013672,
      "learning_rate": 2.205334506533641e-06,
      "loss": 3.1949,
      "step": 52895
    },
    {
      "epoch": 0.5381266276041666,
      "grad_norm": 11.615266799926758,
      "learning_rate": 2.2049374167371898e-06,
      "loss": 3.1898,
      "step": 52900
    },
    {
      "epoch": 0.538177490234375,
      "grad_norm": 14.333727836608887,
      "learning_rate": 2.204540334489846e-06,
      "loss": 3.577,
      "step": 52905
    },
    {
      "epoch": 0.5382283528645834,
      "grad_norm": 12.227395057678223,
      "learning_rate": 2.204143259801767e-06,
      "loss": 3.314,
      "step": 52910
    },
    {
      "epoch": 0.5382792154947916,
      "grad_norm": 13.9808931350708,
      "learning_rate": 2.203746192683112e-06,
      "loss": 3.4432,
      "step": 52915
    },
    {
      "epoch": 0.538330078125,
      "grad_norm": 12.037018775939941,
      "learning_rate": 2.203349133144042e-06,
      "loss": 3.4993,
      "step": 52920
    },
    {
      "epoch": 0.5383809407552084,
      "grad_norm": 12.9117431640625,
      "learning_rate": 2.2029520811947134e-06,
      "loss": 3.4597,
      "step": 52925
    },
    {
      "epoch": 0.5384318033854166,
      "grad_norm": 10.380311965942383,
      "learning_rate": 2.2025550368452853e-06,
      "loss": 3.2176,
      "step": 52930
    },
    {
      "epoch": 0.538482666015625,
      "grad_norm": 16.426626205444336,
      "learning_rate": 2.202158000105916e-06,
      "loss": 3.2336,
      "step": 52935
    },
    {
      "epoch": 0.5385335286458334,
      "grad_norm": 9.988062858581543,
      "learning_rate": 2.2017609709867636e-06,
      "loss": 3.0777,
      "step": 52940
    },
    {
      "epoch": 0.5385843912760416,
      "grad_norm": 11.464739799499512,
      "learning_rate": 2.2013639494979865e-06,
      "loss": 3.2367,
      "step": 52945
    },
    {
      "epoch": 0.53863525390625,
      "grad_norm": 8.486089706420898,
      "learning_rate": 2.2009669356497414e-06,
      "loss": 3.3867,
      "step": 52950
    },
    {
      "epoch": 0.5386861165364584,
      "grad_norm": 13.624707221984863,
      "learning_rate": 2.2005699294521867e-06,
      "loss": 3.5823,
      "step": 52955
    },
    {
      "epoch": 0.5387369791666666,
      "grad_norm": 13.097014427185059,
      "learning_rate": 2.2001729309154795e-06,
      "loss": 3.7487,
      "step": 52960
    },
    {
      "epoch": 0.538787841796875,
      "grad_norm": 9.218652725219727,
      "learning_rate": 2.1997759400497764e-06,
      "loss": 3.0382,
      "step": 52965
    },
    {
      "epoch": 0.5388387044270834,
      "grad_norm": 11.32375431060791,
      "learning_rate": 2.199378956865235e-06,
      "loss": 3.7389,
      "step": 52970
    },
    {
      "epoch": 0.5388895670572916,
      "grad_norm": 11.567964553833008,
      "learning_rate": 2.1989819813720116e-06,
      "loss": 3.7298,
      "step": 52975
    },
    {
      "epoch": 0.5389404296875,
      "grad_norm": 13.934860229492188,
      "learning_rate": 2.198585013580263e-06,
      "loss": 3.3685,
      "step": 52980
    },
    {
      "epoch": 0.5389912923177084,
      "grad_norm": 13.693037033081055,
      "learning_rate": 2.1981880535001456e-06,
      "loss": 3.9366,
      "step": 52985
    },
    {
      "epoch": 0.5390421549479166,
      "grad_norm": 14.608330726623535,
      "learning_rate": 2.1977911011418147e-06,
      "loss": 3.314,
      "step": 52990
    },
    {
      "epoch": 0.539093017578125,
      "grad_norm": 9.940957069396973,
      "learning_rate": 2.197394156515427e-06,
      "loss": 3.2194,
      "step": 52995
    },
    {
      "epoch": 0.5391438802083334,
      "grad_norm": 8.905831336975098,
      "learning_rate": 2.1969972196311387e-06,
      "loss": 3.8342,
      "step": 53000
    },
    {
      "epoch": 0.5391947428385416,
      "grad_norm": 11.434901237487793,
      "learning_rate": 2.1966002904991043e-06,
      "loss": 3.4717,
      "step": 53005
    },
    {
      "epoch": 0.53924560546875,
      "grad_norm": 9.313929557800293,
      "learning_rate": 2.196203369129479e-06,
      "loss": 3.592,
      "step": 53010
    },
    {
      "epoch": 0.5392964680989584,
      "grad_norm": 14.059566497802734,
      "learning_rate": 2.1958064555324195e-06,
      "loss": 3.802,
      "step": 53015
    },
    {
      "epoch": 0.5393473307291666,
      "grad_norm": 9.69733715057373,
      "learning_rate": 2.195409549718079e-06,
      "loss": 2.9602,
      "step": 53020
    },
    {
      "epoch": 0.539398193359375,
      "grad_norm": 15.862642288208008,
      "learning_rate": 2.195012651696614e-06,
      "loss": 4.232,
      "step": 53025
    },
    {
      "epoch": 0.5394490559895834,
      "grad_norm": 13.767962455749512,
      "learning_rate": 2.194615761478176e-06,
      "loss": 3.3887,
      "step": 53030
    },
    {
      "epoch": 0.5394999186197916,
      "grad_norm": 13.546216011047363,
      "learning_rate": 2.194218879072923e-06,
      "loss": 3.4194,
      "step": 53035
    },
    {
      "epoch": 0.53955078125,
      "grad_norm": 16.30230712890625,
      "learning_rate": 2.193822004491007e-06,
      "loss": 2.9932,
      "step": 53040
    },
    {
      "epoch": 0.5396016438802084,
      "grad_norm": 9.344871520996094,
      "learning_rate": 2.1934251377425833e-06,
      "loss": 3.5214,
      "step": 53045
    },
    {
      "epoch": 0.5396525065104166,
      "grad_norm": 12.343306541442871,
      "learning_rate": 2.193028278837803e-06,
      "loss": 3.0739,
      "step": 53050
    },
    {
      "epoch": 0.539703369140625,
      "grad_norm": 7.4595723152160645,
      "learning_rate": 2.1926314277868226e-06,
      "loss": 3.3064,
      "step": 53055
    },
    {
      "epoch": 0.5397542317708334,
      "grad_norm": 10.087668418884277,
      "learning_rate": 2.1922345845997947e-06,
      "loss": 3.1303,
      "step": 53060
    },
    {
      "epoch": 0.5398050944010416,
      "grad_norm": 9.357985496520996,
      "learning_rate": 2.191837749286871e-06,
      "loss": 3.156,
      "step": 53065
    },
    {
      "epoch": 0.53985595703125,
      "grad_norm": 9.338754653930664,
      "learning_rate": 2.1914409218582057e-06,
      "loss": 3.8574,
      "step": 53070
    },
    {
      "epoch": 0.5399068196614584,
      "grad_norm": 10.841154098510742,
      "learning_rate": 2.191044102323952e-06,
      "loss": 3.1361,
      "step": 53075
    },
    {
      "epoch": 0.5399576822916666,
      "grad_norm": 10.66102409362793,
      "learning_rate": 2.190647290694261e-06,
      "loss": 3.5059,
      "step": 53080
    },
    {
      "epoch": 0.540008544921875,
      "grad_norm": 9.158910751342773,
      "learning_rate": 2.1902504869792855e-06,
      "loss": 3.075,
      "step": 53085
    },
    {
      "epoch": 0.5400594075520834,
      "grad_norm": 15.92582893371582,
      "learning_rate": 2.1898536911891787e-06,
      "loss": 3.3651,
      "step": 53090
    },
    {
      "epoch": 0.5401102701822916,
      "grad_norm": 8.675463676452637,
      "learning_rate": 2.1894569033340914e-06,
      "loss": 3.6126,
      "step": 53095
    },
    {
      "epoch": 0.5401611328125,
      "grad_norm": 12.356219291687012,
      "learning_rate": 2.1890601234241757e-06,
      "loss": 3.4649,
      "step": 53100
    },
    {
      "epoch": 0.5402119954427084,
      "grad_norm": 7.340629577636719,
      "learning_rate": 2.1886633514695824e-06,
      "loss": 3.1227,
      "step": 53105
    },
    {
      "epoch": 0.5402628580729166,
      "grad_norm": 13.032557487487793,
      "learning_rate": 2.1882665874804637e-06,
      "loss": 3.4904,
      "step": 53110
    },
    {
      "epoch": 0.540313720703125,
      "grad_norm": 10.5289306640625,
      "learning_rate": 2.187869831466971e-06,
      "loss": 3.3025,
      "step": 53115
    },
    {
      "epoch": 0.5403645833333334,
      "grad_norm": 13.882126808166504,
      "learning_rate": 2.1874730834392545e-06,
      "loss": 3.3192,
      "step": 53120
    },
    {
      "epoch": 0.5404154459635416,
      "grad_norm": 13.612321853637695,
      "learning_rate": 2.1870763434074644e-06,
      "loss": 3.4219,
      "step": 53125
    },
    {
      "epoch": 0.54046630859375,
      "grad_norm": 8.61611557006836,
      "learning_rate": 2.1866796113817527e-06,
      "loss": 3.246,
      "step": 53130
    },
    {
      "epoch": 0.5405171712239584,
      "grad_norm": 10.256523132324219,
      "learning_rate": 2.1862828873722684e-06,
      "loss": 3.3299,
      "step": 53135
    },
    {
      "epoch": 0.5405680338541666,
      "grad_norm": 13.90639877319336,
      "learning_rate": 2.185886171389163e-06,
      "loss": 3.5323,
      "step": 53140
    },
    {
      "epoch": 0.540618896484375,
      "grad_norm": 12.011414527893066,
      "learning_rate": 2.185489463442584e-06,
      "loss": 2.9904,
      "step": 53145
    },
    {
      "epoch": 0.5406697591145834,
      "grad_norm": 14.602082252502441,
      "learning_rate": 2.185092763542683e-06,
      "loss": 3.3221,
      "step": 53150
    },
    {
      "epoch": 0.5407206217447916,
      "grad_norm": 11.019375801086426,
      "learning_rate": 2.18469607169961e-06,
      "loss": 3.2125,
      "step": 53155
    },
    {
      "epoch": 0.540771484375,
      "grad_norm": 9.490230560302734,
      "learning_rate": 2.184299387923512e-06,
      "loss": 3.3152,
      "step": 53160
    },
    {
      "epoch": 0.5408223470052084,
      "grad_norm": 15.053577423095703,
      "learning_rate": 2.18390271222454e-06,
      "loss": 3.3199,
      "step": 53165
    },
    {
      "epoch": 0.5408732096354166,
      "grad_norm": 9.94269847869873,
      "learning_rate": 2.183506044612842e-06,
      "loss": 3.1042,
      "step": 53170
    },
    {
      "epoch": 0.540924072265625,
      "grad_norm": 14.025388717651367,
      "learning_rate": 2.183109385098567e-06,
      "loss": 3.3606,
      "step": 53175
    },
    {
      "epoch": 0.5409749348958334,
      "grad_norm": 10.789249420166016,
      "learning_rate": 2.1827127336918622e-06,
      "loss": 3.2626,
      "step": 53180
    },
    {
      "epoch": 0.5410257975260416,
      "grad_norm": 15.582239151000977,
      "learning_rate": 2.182316090402878e-06,
      "loss": 3.3256,
      "step": 53185
    },
    {
      "epoch": 0.54107666015625,
      "grad_norm": 11.45635986328125,
      "learning_rate": 2.181919455241762e-06,
      "loss": 3.6184,
      "step": 53190
    },
    {
      "epoch": 0.5411275227864584,
      "grad_norm": 9.036690711975098,
      "learning_rate": 2.1815228282186603e-06,
      "loss": 3.4225,
      "step": 53195
    },
    {
      "epoch": 0.5411783854166666,
      "grad_norm": 11.719807624816895,
      "learning_rate": 2.181126209343721e-06,
      "loss": 3.565,
      "step": 53200
    },
    {
      "epoch": 0.541229248046875,
      "grad_norm": 12.068906784057617,
      "learning_rate": 2.1807295986270934e-06,
      "loss": 3.1241,
      "step": 53205
    },
    {
      "epoch": 0.5412801106770834,
      "grad_norm": 12.981789588928223,
      "learning_rate": 2.1803329960789227e-06,
      "loss": 3.2856,
      "step": 53210
    },
    {
      "epoch": 0.5413309733072916,
      "grad_norm": 15.0866060256958,
      "learning_rate": 2.1799364017093574e-06,
      "loss": 3.6841,
      "step": 53215
    },
    {
      "epoch": 0.5413818359375,
      "grad_norm": 9.662513732910156,
      "learning_rate": 2.1795398155285415e-06,
      "loss": 3.1087,
      "step": 53220
    },
    {
      "epoch": 0.5414326985677084,
      "grad_norm": 10.991047859191895,
      "learning_rate": 2.1791432375466247e-06,
      "loss": 3.3624,
      "step": 53225
    },
    {
      "epoch": 0.5414835611979166,
      "grad_norm": 10.34939956665039,
      "learning_rate": 2.1787466677737523e-06,
      "loss": 3.3209,
      "step": 53230
    },
    {
      "epoch": 0.541534423828125,
      "grad_norm": 13.844328880310059,
      "learning_rate": 2.17835010622007e-06,
      "loss": 3.1816,
      "step": 53235
    },
    {
      "epoch": 0.5415852864583334,
      "grad_norm": 12.248176574707031,
      "learning_rate": 2.177953552895723e-06,
      "loss": 3.3928,
      "step": 53240
    },
    {
      "epoch": 0.5416361490885416,
      "grad_norm": 12.704273223876953,
      "learning_rate": 2.1775570078108597e-06,
      "loss": 3.3041,
      "step": 53245
    },
    {
      "epoch": 0.54168701171875,
      "grad_norm": 7.139502048492432,
      "learning_rate": 2.177160470975623e-06,
      "loss": 3.214,
      "step": 53250
    },
    {
      "epoch": 0.5417378743489584,
      "grad_norm": 16.03692054748535,
      "learning_rate": 2.1767639424001586e-06,
      "loss": 3.3918,
      "step": 53255
    },
    {
      "epoch": 0.5417887369791666,
      "grad_norm": 14.646202087402344,
      "learning_rate": 2.1763674220946134e-06,
      "loss": 3.6005,
      "step": 53260
    },
    {
      "epoch": 0.541839599609375,
      "grad_norm": 10.91369342803955,
      "learning_rate": 2.1759709100691303e-06,
      "loss": 3.0894,
      "step": 53265
    },
    {
      "epoch": 0.5418904622395834,
      "grad_norm": 18.816396713256836,
      "learning_rate": 2.1755744063338556e-06,
      "loss": 3.5276,
      "step": 53270
    },
    {
      "epoch": 0.5419413248697916,
      "grad_norm": 13.356977462768555,
      "learning_rate": 2.175177910898931e-06,
      "loss": 3.3633,
      "step": 53275
    },
    {
      "epoch": 0.5419921875,
      "grad_norm": 14.471107482910156,
      "learning_rate": 2.1747814237745036e-06,
      "loss": 3.6021,
      "step": 53280
    },
    {
      "epoch": 0.5420430501302084,
      "grad_norm": 8.016794204711914,
      "learning_rate": 2.174384944970717e-06,
      "loss": 3.8809,
      "step": 53285
    },
    {
      "epoch": 0.5420939127604166,
      "grad_norm": 9.090747833251953,
      "learning_rate": 2.1739884744977134e-06,
      "loss": 3.44,
      "step": 53290
    },
    {
      "epoch": 0.542144775390625,
      "grad_norm": 14.494226455688477,
      "learning_rate": 2.1735920123656375e-06,
      "loss": 3.4437,
      "step": 53295
    },
    {
      "epoch": 0.5421956380208334,
      "grad_norm": 10.311285018920898,
      "learning_rate": 2.173195558584633e-06,
      "loss": 3.5714,
      "step": 53300
    },
    {
      "epoch": 0.5422465006510416,
      "grad_norm": 10.30700969696045,
      "learning_rate": 2.172799113164842e-06,
      "loss": 3.3391,
      "step": 53305
    },
    {
      "epoch": 0.54229736328125,
      "grad_norm": 15.927813529968262,
      "learning_rate": 2.1724026761164096e-06,
      "loss": 3.0528,
      "step": 53310
    },
    {
      "epoch": 0.5423482259114584,
      "grad_norm": 9.194511413574219,
      "learning_rate": 2.172006247449476e-06,
      "loss": 3.4069,
      "step": 53315
    },
    {
      "epoch": 0.5423990885416666,
      "grad_norm": 11.598966598510742,
      "learning_rate": 2.1716098271741846e-06,
      "loss": 3.8443,
      "step": 53320
    },
    {
      "epoch": 0.542449951171875,
      "grad_norm": 14.813419342041016,
      "learning_rate": 2.1712134153006786e-06,
      "loss": 3.066,
      "step": 53325
    },
    {
      "epoch": 0.5425008138020834,
      "grad_norm": 11.64278793334961,
      "learning_rate": 2.1708170118390992e-06,
      "loss": 3.6079,
      "step": 53330
    },
    {
      "epoch": 0.5425516764322916,
      "grad_norm": 12.371272087097168,
      "learning_rate": 2.170420616799588e-06,
      "loss": 3.0563,
      "step": 53335
    },
    {
      "epoch": 0.5426025390625,
      "grad_norm": 13.104665756225586,
      "learning_rate": 2.1700242301922875e-06,
      "loss": 3.3068,
      "step": 53340
    },
    {
      "epoch": 0.5426534016927084,
      "grad_norm": 8.993667602539062,
      "learning_rate": 2.169627852027339e-06,
      "loss": 3.176,
      "step": 53345
    },
    {
      "epoch": 0.5427042643229166,
      "grad_norm": 10.01164436340332,
      "learning_rate": 2.169231482314883e-06,
      "loss": 3.2864,
      "step": 53350
    },
    {
      "epoch": 0.542755126953125,
      "grad_norm": 15.4075927734375,
      "learning_rate": 2.168835121065062e-06,
      "loss": 3.3562,
      "step": 53355
    },
    {
      "epoch": 0.5428059895833334,
      "grad_norm": 11.776107788085938,
      "learning_rate": 2.168438768288016e-06,
      "loss": 3.7907,
      "step": 53360
    },
    {
      "epoch": 0.5428568522135416,
      "grad_norm": 13.539200782775879,
      "learning_rate": 2.1680424239938853e-06,
      "loss": 3.2214,
      "step": 53365
    },
    {
      "epoch": 0.54290771484375,
      "grad_norm": 14.385930061340332,
      "learning_rate": 2.16764608819281e-06,
      "loss": 3.2643,
      "step": 53370
    },
    {
      "epoch": 0.5429585774739584,
      "grad_norm": 15.621609687805176,
      "learning_rate": 2.1672497608949312e-06,
      "loss": 3.3567,
      "step": 53375
    },
    {
      "epoch": 0.5430094401041666,
      "grad_norm": 12.618790626525879,
      "learning_rate": 2.1668534421103886e-06,
      "loss": 3.2828,
      "step": 53380
    },
    {
      "epoch": 0.543060302734375,
      "grad_norm": 19.181137084960938,
      "learning_rate": 2.1664571318493216e-06,
      "loss": 3.3321,
      "step": 53385
    },
    {
      "epoch": 0.5431111653645834,
      "grad_norm": 12.526358604431152,
      "learning_rate": 2.1660608301218693e-06,
      "loss": 3.0963,
      "step": 53390
    },
    {
      "epoch": 0.5431620279947916,
      "grad_norm": 7.121951103210449,
      "learning_rate": 2.165664536938172e-06,
      "loss": 3.5169,
      "step": 53395
    },
    {
      "epoch": 0.543212890625,
      "grad_norm": 8.991941452026367,
      "learning_rate": 2.1652682523083686e-06,
      "loss": 3.3319,
      "step": 53400
    },
    {
      "epoch": 0.5432637532552084,
      "grad_norm": 13.199124336242676,
      "learning_rate": 2.164871976242597e-06,
      "loss": 3.5861,
      "step": 53405
    },
    {
      "epoch": 0.5433146158854166,
      "grad_norm": 14.945271492004395,
      "learning_rate": 2.164475708750996e-06,
      "loss": 3.3048,
      "step": 53410
    },
    {
      "epoch": 0.543365478515625,
      "grad_norm": 9.011767387390137,
      "learning_rate": 2.164079449843706e-06,
      "loss": 3.6937,
      "step": 53415
    },
    {
      "epoch": 0.5434163411458334,
      "grad_norm": 12.211606979370117,
      "learning_rate": 2.163683199530862e-06,
      "loss": 3.1385,
      "step": 53420
    },
    {
      "epoch": 0.5434672037760416,
      "grad_norm": 9.78544807434082,
      "learning_rate": 2.163286957822605e-06,
      "loss": 3.3107,
      "step": 53425
    },
    {
      "epoch": 0.54351806640625,
      "grad_norm": 12.810589790344238,
      "learning_rate": 2.1628907247290704e-06,
      "loss": 2.9619,
      "step": 53430
    },
    {
      "epoch": 0.5435689290364584,
      "grad_norm": 20.296228408813477,
      "learning_rate": 2.162494500260397e-06,
      "loss": 3.5441,
      "step": 53435
    },
    {
      "epoch": 0.5436197916666666,
      "grad_norm": 9.777166366577148,
      "learning_rate": 2.1620982844267222e-06,
      "loss": 3.3543,
      "step": 53440
    },
    {
      "epoch": 0.543670654296875,
      "grad_norm": 12.75098705291748,
      "learning_rate": 2.161702077238182e-06,
      "loss": 3.4497,
      "step": 53445
    },
    {
      "epoch": 0.5437215169270834,
      "grad_norm": 10.42978286743164,
      "learning_rate": 2.161305878704914e-06,
      "loss": 3.2527,
      "step": 53450
    },
    {
      "epoch": 0.5437723795572916,
      "grad_norm": 11.12440013885498,
      "learning_rate": 2.1609096888370557e-06,
      "loss": 3.281,
      "step": 53455
    },
    {
      "epoch": 0.5438232421875,
      "grad_norm": 12.265634536743164,
      "learning_rate": 2.160513507644742e-06,
      "loss": 3.3271,
      "step": 53460
    },
    {
      "epoch": 0.5438741048177084,
      "grad_norm": 13.81308364868164,
      "learning_rate": 2.1601173351381093e-06,
      "loss": 3.4146,
      "step": 53465
    },
    {
      "epoch": 0.5439249674479166,
      "grad_norm": 10.630414009094238,
      "learning_rate": 2.159721171327295e-06,
      "loss": 3.4962,
      "step": 53470
    },
    {
      "epoch": 0.543975830078125,
      "grad_norm": 11.793816566467285,
      "learning_rate": 2.1593250162224333e-06,
      "loss": 3.708,
      "step": 53475
    },
    {
      "epoch": 0.5440266927083334,
      "grad_norm": 15.098947525024414,
      "learning_rate": 2.158928869833661e-06,
      "loss": 3.7219,
      "step": 53480
    },
    {
      "epoch": 0.5440775553385416,
      "grad_norm": 9.953935623168945,
      "learning_rate": 2.158532732171111e-06,
      "loss": 3.079,
      "step": 53485
    },
    {
      "epoch": 0.54412841796875,
      "grad_norm": 11.695734977722168,
      "learning_rate": 2.1581366032449217e-06,
      "loss": 3.1605,
      "step": 53490
    },
    {
      "epoch": 0.5441792805989584,
      "grad_norm": 13.40351390838623,
      "learning_rate": 2.157740483065226e-06,
      "loss": 3.398,
      "step": 53495
    },
    {
      "epoch": 0.5442301432291666,
      "grad_norm": 7.162520885467529,
      "learning_rate": 2.157344371642159e-06,
      "loss": 3.6939,
      "step": 53500
    },
    {
      "epoch": 0.544281005859375,
      "grad_norm": 17.510847091674805,
      "learning_rate": 2.1569482689858547e-06,
      "loss": 3.5553,
      "step": 53505
    },
    {
      "epoch": 0.5443318684895834,
      "grad_norm": 10.181023597717285,
      "learning_rate": 2.1565521751064476e-06,
      "loss": 3.1076,
      "step": 53510
    },
    {
      "epoch": 0.5443827311197916,
      "grad_norm": 13.848320007324219,
      "learning_rate": 2.1561560900140726e-06,
      "loss": 3.4043,
      "step": 53515
    },
    {
      "epoch": 0.54443359375,
      "grad_norm": 12.950430870056152,
      "learning_rate": 2.1557600137188615e-06,
      "loss": 3.055,
      "step": 53520
    },
    {
      "epoch": 0.5444844563802084,
      "grad_norm": 12.266128540039062,
      "learning_rate": 2.155363946230949e-06,
      "loss": 3.3766,
      "step": 53525
    },
    {
      "epoch": 0.5445353190104166,
      "grad_norm": 10.952404022216797,
      "learning_rate": 2.1549678875604688e-06,
      "loss": 3.4065,
      "step": 53530
    },
    {
      "epoch": 0.544586181640625,
      "grad_norm": 13.889093399047852,
      "learning_rate": 2.154571837717553e-06,
      "loss": 3.7138,
      "step": 53535
    },
    {
      "epoch": 0.5446370442708334,
      "grad_norm": 14.20893383026123,
      "learning_rate": 2.1541757967123356e-06,
      "loss": 3.283,
      "step": 53540
    },
    {
      "epoch": 0.5446879069010416,
      "grad_norm": 13.899380683898926,
      "learning_rate": 2.1537797645549467e-06,
      "loss": 3.0658,
      "step": 53545
    },
    {
      "epoch": 0.54473876953125,
      "grad_norm": 13.060921669006348,
      "learning_rate": 2.153383741255522e-06,
      "loss": 3.6988,
      "step": 53550
    },
    {
      "epoch": 0.5447896321614584,
      "grad_norm": 10.030902862548828,
      "learning_rate": 2.152987726824192e-06,
      "loss": 3.0602,
      "step": 53555
    },
    {
      "epoch": 0.5448404947916666,
      "grad_norm": 16.82111358642578,
      "learning_rate": 2.1525917212710877e-06,
      "loss": 3.9291,
      "step": 53560
    },
    {
      "epoch": 0.544891357421875,
      "grad_norm": 10.103955268859863,
      "learning_rate": 2.1521957246063427e-06,
      "loss": 3.28,
      "step": 53565
    },
    {
      "epoch": 0.5449422200520834,
      "grad_norm": 9.024051666259766,
      "learning_rate": 2.151799736840088e-06,
      "loss": 3.4298,
      "step": 53570
    },
    {
      "epoch": 0.5449930826822916,
      "grad_norm": 9.407695770263672,
      "learning_rate": 2.151403757982454e-06,
      "loss": 3.4351,
      "step": 53575
    },
    {
      "epoch": 0.5450439453125,
      "grad_norm": 8.37774658203125,
      "learning_rate": 2.151007788043572e-06,
      "loss": 3.2921,
      "step": 53580
    },
    {
      "epoch": 0.5450948079427084,
      "grad_norm": 15.555733680725098,
      "learning_rate": 2.1506118270335734e-06,
      "loss": 3.1459,
      "step": 53585
    },
    {
      "epoch": 0.5451456705729166,
      "grad_norm": 15.700858116149902,
      "learning_rate": 2.1502158749625885e-06,
      "loss": 3.2357,
      "step": 53590
    },
    {
      "epoch": 0.545196533203125,
      "grad_norm": 9.189980506896973,
      "learning_rate": 2.1498199318407477e-06,
      "loss": 4.0753,
      "step": 53595
    },
    {
      "epoch": 0.5452473958333334,
      "grad_norm": 15.427556037902832,
      "learning_rate": 2.14942399767818e-06,
      "loss": 3.0574,
      "step": 53600
    },
    {
      "epoch": 0.5452982584635416,
      "grad_norm": 15.837459564208984,
      "learning_rate": 2.1490280724850164e-06,
      "loss": 3.5831,
      "step": 53605
    },
    {
      "epoch": 0.54534912109375,
      "grad_norm": 13.636536598205566,
      "learning_rate": 2.1486321562713875e-06,
      "loss": 3.0997,
      "step": 53610
    },
    {
      "epoch": 0.5453999837239584,
      "grad_norm": 12.569819450378418,
      "learning_rate": 2.148236249047421e-06,
      "loss": 2.9296,
      "step": 53615
    },
    {
      "epoch": 0.5454508463541666,
      "grad_norm": 7.152266025543213,
      "learning_rate": 2.1478403508232458e-06,
      "loss": 2.8501,
      "step": 53620
    },
    {
      "epoch": 0.545501708984375,
      "grad_norm": 9.717489242553711,
      "learning_rate": 2.1474444616089928e-06,
      "loss": 3.5302,
      "step": 53625
    },
    {
      "epoch": 0.5455525716145834,
      "grad_norm": 9.396700859069824,
      "learning_rate": 2.1470485814147894e-06,
      "loss": 3.2379,
      "step": 53630
    },
    {
      "epoch": 0.5456034342447916,
      "grad_norm": 16.059377670288086,
      "learning_rate": 2.146652710250765e-06,
      "loss": 3.1214,
      "step": 53635
    },
    {
      "epoch": 0.545654296875,
      "grad_norm": 15.743471145629883,
      "learning_rate": 2.1462568481270455e-06,
      "loss": 3.4218,
      "step": 53640
    },
    {
      "epoch": 0.5457051595052084,
      "grad_norm": 11.109064102172852,
      "learning_rate": 2.145860995053762e-06,
      "loss": 3.4594,
      "step": 53645
    },
    {
      "epoch": 0.5457560221354166,
      "grad_norm": 11.835540771484375,
      "learning_rate": 2.1454651510410412e-06,
      "loss": 3.3928,
      "step": 53650
    },
    {
      "epoch": 0.545806884765625,
      "grad_norm": 9.042723655700684,
      "learning_rate": 2.1450693160990093e-06,
      "loss": 3.4778,
      "step": 53655
    },
    {
      "epoch": 0.5458577473958334,
      "grad_norm": 14.188231468200684,
      "learning_rate": 2.144673490237796e-06,
      "loss": 3.1003,
      "step": 53660
    },
    {
      "epoch": 0.5459086100260416,
      "grad_norm": 8.52377986907959,
      "learning_rate": 2.1442776734675268e-06,
      "loss": 3.7347,
      "step": 53665
    },
    {
      "epoch": 0.54595947265625,
      "grad_norm": 12.667838096618652,
      "learning_rate": 2.14388186579833e-06,
      "loss": 3.6695,
      "step": 53670
    },
    {
      "epoch": 0.5460103352864584,
      "grad_norm": 11.661608695983887,
      "learning_rate": 2.1434860672403298e-06,
      "loss": 3.3877,
      "step": 53675
    },
    {
      "epoch": 0.5460611979166666,
      "grad_norm": 12.163783073425293,
      "learning_rate": 2.1430902778036543e-06,
      "loss": 3.2375,
      "step": 53680
    },
    {
      "epoch": 0.546112060546875,
      "grad_norm": 13.89341926574707,
      "learning_rate": 2.1426944974984303e-06,
      "loss": 2.9204,
      "step": 53685
    },
    {
      "epoch": 0.5461629231770834,
      "grad_norm": 9.899314880371094,
      "learning_rate": 2.142298726334782e-06,
      "loss": 3.5734,
      "step": 53690
    },
    {
      "epoch": 0.5462137858072916,
      "grad_norm": 10.043965339660645,
      "learning_rate": 2.141902964322836e-06,
      "loss": 3.4135,
      "step": 53695
    },
    {
      "epoch": 0.5462646484375,
      "grad_norm": 14.102566719055176,
      "learning_rate": 2.1415072114727187e-06,
      "loss": 3.5811,
      "step": 53700
    },
    {
      "epoch": 0.5463155110677084,
      "grad_norm": 9.399636268615723,
      "learning_rate": 2.1411114677945537e-06,
      "loss": 3.2885,
      "step": 53705
    },
    {
      "epoch": 0.5463663736979166,
      "grad_norm": 9.890459060668945,
      "learning_rate": 2.140715733298467e-06,
      "loss": 3.4513,
      "step": 53710
    },
    {
      "epoch": 0.546417236328125,
      "grad_norm": 9.972284317016602,
      "learning_rate": 2.1403200079945826e-06,
      "loss": 3.2842,
      "step": 53715
    },
    {
      "epoch": 0.5464680989583334,
      "grad_norm": 17.944011688232422,
      "learning_rate": 2.1399242918930257e-06,
      "loss": 3.1879,
      "step": 53720
    },
    {
      "epoch": 0.5465189615885416,
      "grad_norm": 10.539843559265137,
      "learning_rate": 2.139528585003921e-06,
      "loss": 3.2273,
      "step": 53725
    },
    {
      "epoch": 0.54656982421875,
      "grad_norm": 10.64560317993164,
      "learning_rate": 2.139132887337392e-06,
      "loss": 2.9197,
      "step": 53730
    },
    {
      "epoch": 0.5466206868489584,
      "grad_norm": 10.019573211669922,
      "learning_rate": 2.1387371989035614e-06,
      "loss": 3.4138,
      "step": 53735
    },
    {
      "epoch": 0.5466715494791666,
      "grad_norm": 6.66783332824707,
      "learning_rate": 2.1383415197125553e-06,
      "loss": 3.1166,
      "step": 53740
    },
    {
      "epoch": 0.546722412109375,
      "grad_norm": 14.750569343566895,
      "learning_rate": 2.1379458497744952e-06,
      "loss": 3.0363,
      "step": 53745
    },
    {
      "epoch": 0.5467732747395834,
      "grad_norm": 10.087311744689941,
      "learning_rate": 2.137550189099504e-06,
      "loss": 2.8389,
      "step": 53750
    },
    {
      "epoch": 0.5468241373697916,
      "grad_norm": 14.355230331420898,
      "learning_rate": 2.1371545376977063e-06,
      "loss": 3.4085,
      "step": 53755
    },
    {
      "epoch": 0.546875,
      "grad_norm": 10.355266571044922,
      "learning_rate": 2.1367588955792233e-06,
      "loss": 2.9538,
      "step": 53760
    },
    {
      "epoch": 0.5469258626302084,
      "grad_norm": 11.927539825439453,
      "learning_rate": 2.1363632627541786e-06,
      "loss": 3.5174,
      "step": 53765
    },
    {
      "epoch": 0.5469767252604166,
      "grad_norm": 10.795658111572266,
      "learning_rate": 2.1359676392326926e-06,
      "loss": 3.3532,
      "step": 53770
    },
    {
      "epoch": 0.547027587890625,
      "grad_norm": 7.102908611297607,
      "learning_rate": 2.1355720250248884e-06,
      "loss": 3.3258,
      "step": 53775
    },
    {
      "epoch": 0.5470784505208334,
      "grad_norm": 13.276028633117676,
      "learning_rate": 2.135176420140888e-06,
      "loss": 3.2748,
      "step": 53780
    },
    {
      "epoch": 0.5471293131510416,
      "grad_norm": 12.033732414245605,
      "learning_rate": 2.1347808245908124e-06,
      "loss": 3.4438,
      "step": 53785
    },
    {
      "epoch": 0.54718017578125,
      "grad_norm": 9.38341999053955,
      "learning_rate": 2.1343852383847824e-06,
      "loss": 3.5425,
      "step": 53790
    },
    {
      "epoch": 0.5472310384114584,
      "grad_norm": 8.150219917297363,
      "learning_rate": 2.1339896615329196e-06,
      "loss": 3.1475,
      "step": 53795
    },
    {
      "epoch": 0.5472819010416666,
      "grad_norm": 37.85032653808594,
      "learning_rate": 2.1335940940453452e-06,
      "loss": 3.4975,
      "step": 53800
    },
    {
      "epoch": 0.547332763671875,
      "grad_norm": 10.433250427246094,
      "learning_rate": 2.1331985359321784e-06,
      "loss": 3.3001,
      "step": 53805
    },
    {
      "epoch": 0.5473836263020834,
      "grad_norm": 12.532647132873535,
      "learning_rate": 2.1328029872035396e-06,
      "loss": 3.1557,
      "step": 53810
    },
    {
      "epoch": 0.5474344889322916,
      "grad_norm": 10.558504104614258,
      "learning_rate": 2.1324074478695506e-06,
      "loss": 3.2694,
      "step": 53815
    },
    {
      "epoch": 0.5474853515625,
      "grad_norm": 16.129058837890625,
      "learning_rate": 2.132011917940329e-06,
      "loss": 3.8683,
      "step": 53820
    },
    {
      "epoch": 0.5475362141927084,
      "grad_norm": 10.495721817016602,
      "learning_rate": 2.131616397425996e-06,
      "loss": 3.4622,
      "step": 53825
    },
    {
      "epoch": 0.5475870768229166,
      "grad_norm": 11.091641426086426,
      "learning_rate": 2.131220886336669e-06,
      "loss": 3.2935,
      "step": 53830
    },
    {
      "epoch": 0.547637939453125,
      "grad_norm": 9.288651466369629,
      "learning_rate": 2.130825384682469e-06,
      "loss": 3.5211,
      "step": 53835
    },
    {
      "epoch": 0.5476888020833334,
      "grad_norm": 10.28441047668457,
      "learning_rate": 2.1304298924735147e-06,
      "loss": 3.0458,
      "step": 53840
    },
    {
      "epoch": 0.5477396647135416,
      "grad_norm": 11.29616928100586,
      "learning_rate": 2.1300344097199222e-06,
      "loss": 3.6167,
      "step": 53845
    },
    {
      "epoch": 0.54779052734375,
      "grad_norm": 9.837610244750977,
      "learning_rate": 2.1296389364318124e-06,
      "loss": 3.6602,
      "step": 53850
    },
    {
      "epoch": 0.5478413899739584,
      "grad_norm": 11.58377456665039,
      "learning_rate": 2.1292434726193035e-06,
      "loss": 3.5914,
      "step": 53855
    },
    {
      "epoch": 0.5478922526041666,
      "grad_norm": 13.809196472167969,
      "learning_rate": 2.128848018292512e-06,
      "loss": 3.2624,
      "step": 53860
    },
    {
      "epoch": 0.547943115234375,
      "grad_norm": 11.803674697875977,
      "learning_rate": 2.1284525734615553e-06,
      "loss": 3.3709,
      "step": 53865
    },
    {
      "epoch": 0.5479939778645834,
      "grad_norm": 9.340080261230469,
      "learning_rate": 2.1280571381365525e-06,
      "loss": 3.6047,
      "step": 53870
    },
    {
      "epoch": 0.5480448404947916,
      "grad_norm": 16.43390655517578,
      "learning_rate": 2.1276617123276192e-06,
      "loss": 4.1752,
      "step": 53875
    },
    {
      "epoch": 0.548095703125,
      "grad_norm": 12.21694564819336,
      "learning_rate": 2.127266296044873e-06,
      "loss": 3.217,
      "step": 53880
    },
    {
      "epoch": 0.5481465657552084,
      "grad_norm": 11.829559326171875,
      "learning_rate": 2.1268708892984293e-06,
      "loss": 3.5265,
      "step": 53885
    },
    {
      "epoch": 0.5481974283854166,
      "grad_norm": 8.994977951049805,
      "learning_rate": 2.126475492098406e-06,
      "loss": 2.9463,
      "step": 53890
    },
    {
      "epoch": 0.548248291015625,
      "grad_norm": 9.751565933227539,
      "learning_rate": 2.1260801044549195e-06,
      "loss": 3.4813,
      "step": 53895
    },
    {
      "epoch": 0.5482991536458334,
      "grad_norm": 10.996219635009766,
      "learning_rate": 2.1256847263780843e-06,
      "loss": 3.1345,
      "step": 53900
    },
    {
      "epoch": 0.5483500162760416,
      "grad_norm": 13.151937484741211,
      "learning_rate": 2.1252893578780158e-06,
      "loss": 3.6332,
      "step": 53905
    },
    {
      "epoch": 0.54840087890625,
      "grad_norm": 12.586631774902344,
      "learning_rate": 2.1248939989648313e-06,
      "loss": 3.2873,
      "step": 53910
    },
    {
      "epoch": 0.5484517415364584,
      "grad_norm": 15.285003662109375,
      "learning_rate": 2.1244986496486445e-06,
      "loss": 3.5958,
      "step": 53915
    },
    {
      "epoch": 0.5485026041666666,
      "grad_norm": 13.51598072052002,
      "learning_rate": 2.1241033099395715e-06,
      "loss": 3.9591,
      "step": 53920
    },
    {
      "epoch": 0.548553466796875,
      "grad_norm": 13.146869659423828,
      "learning_rate": 2.1237079798477253e-06,
      "loss": 3.6961,
      "step": 53925
    },
    {
      "epoch": 0.5486043294270834,
      "grad_norm": 12.482857704162598,
      "learning_rate": 2.1233126593832214e-06,
      "loss": 3.3824,
      "step": 53930
    },
    {
      "epoch": 0.5486551920572916,
      "grad_norm": 11.18795108795166,
      "learning_rate": 2.1229173485561744e-06,
      "loss": 3.0001,
      "step": 53935
    },
    {
      "epoch": 0.5487060546875,
      "grad_norm": 8.530762672424316,
      "learning_rate": 2.122522047376696e-06,
      "loss": 3.4197,
      "step": 53940
    },
    {
      "epoch": 0.5487569173177084,
      "grad_norm": 14.566081047058105,
      "learning_rate": 2.1221267558549037e-06,
      "loss": 3.2567,
      "step": 53945
    },
    {
      "epoch": 0.5488077799479166,
      "grad_norm": 6.921778202056885,
      "learning_rate": 2.1217314740009076e-06,
      "loss": 3.4137,
      "step": 53950
    },
    {
      "epoch": 0.548858642578125,
      "grad_norm": 10.239953994750977,
      "learning_rate": 2.1213362018248227e-06,
      "loss": 3.3854,
      "step": 53955
    },
    {
      "epoch": 0.5489095052083334,
      "grad_norm": 12.461688041687012,
      "learning_rate": 2.1209409393367607e-06,
      "loss": 3.7388,
      "step": 53960
    },
    {
      "epoch": 0.5489603678385416,
      "grad_norm": 13.59146499633789,
      "learning_rate": 2.120545686546835e-06,
      "loss": 3.5482,
      "step": 53965
    },
    {
      "epoch": 0.54901123046875,
      "grad_norm": 12.854676246643066,
      "learning_rate": 2.1201504434651587e-06,
      "loss": 3.5777,
      "step": 53970
    },
    {
      "epoch": 0.5490620930989584,
      "grad_norm": 11.088582992553711,
      "learning_rate": 2.1197552101018425e-06,
      "loss": 3.1657,
      "step": 53975
    },
    {
      "epoch": 0.5491129557291666,
      "grad_norm": 12.971543312072754,
      "learning_rate": 2.119359986466999e-06,
      "loss": 3.5671,
      "step": 53980
    },
    {
      "epoch": 0.549163818359375,
      "grad_norm": 7.972337245941162,
      "learning_rate": 2.1189647725707406e-06,
      "loss": 3.4042,
      "step": 53985
    },
    {
      "epoch": 0.5492146809895834,
      "grad_norm": 9.62437915802002,
      "learning_rate": 2.118569568423178e-06,
      "loss": 3.5359,
      "step": 53990
    },
    {
      "epoch": 0.5492655436197916,
      "grad_norm": 9.917858123779297,
      "learning_rate": 2.1181743740344233e-06,
      "loss": 3.604,
      "step": 53995
    },
    {
      "epoch": 0.54931640625,
      "grad_norm": 13.30362319946289,
      "learning_rate": 2.117779189414585e-06,
      "loss": 3.9732,
      "step": 54000
    },
    {
      "epoch": 0.5493672688802084,
      "grad_norm": 12.902823448181152,
      "learning_rate": 2.117384014573777e-06,
      "loss": 3.0754,
      "step": 54005
    },
    {
      "epoch": 0.5494181315104166,
      "grad_norm": 9.599853515625,
      "learning_rate": 2.116988849522108e-06,
      "loss": 3.4358,
      "step": 54010
    },
    {
      "epoch": 0.549468994140625,
      "grad_norm": 13.125391960144043,
      "learning_rate": 2.116593694269688e-06,
      "loss": 3.2702,
      "step": 54015
    },
    {
      "epoch": 0.5495198567708334,
      "grad_norm": 9.132649421691895,
      "learning_rate": 2.1161985488266275e-06,
      "loss": 3.3415,
      "step": 54020
    },
    {
      "epoch": 0.5495707194010416,
      "grad_norm": 9.473170280456543,
      "learning_rate": 2.115803413203037e-06,
      "loss": 3.3436,
      "step": 54025
    },
    {
      "epoch": 0.54962158203125,
      "grad_norm": 11.078560829162598,
      "learning_rate": 2.1154082874090244e-06,
      "loss": 3.0247,
      "step": 54030
    },
    {
      "epoch": 0.5496724446614584,
      "grad_norm": 12.930471420288086,
      "learning_rate": 2.115013171454699e-06,
      "loss": 3.2621,
      "step": 54035
    },
    {
      "epoch": 0.5497233072916666,
      "grad_norm": 15.67178726196289,
      "learning_rate": 2.1146180653501717e-06,
      "loss": 3.6102,
      "step": 54040
    },
    {
      "epoch": 0.549774169921875,
      "grad_norm": 11.1826810836792,
      "learning_rate": 2.1142229691055494e-06,
      "loss": 3.6965,
      "step": 54045
    },
    {
      "epoch": 0.5498250325520834,
      "grad_norm": 8.767780303955078,
      "learning_rate": 2.1138278827309413e-06,
      "loss": 3.1587,
      "step": 54050
    },
    {
      "epoch": 0.5498758951822916,
      "grad_norm": 11.69979476928711,
      "learning_rate": 2.113432806236454e-06,
      "loss": 3.5907,
      "step": 54055
    },
    {
      "epoch": 0.5499267578125,
      "grad_norm": 11.563873291015625,
      "learning_rate": 2.1130377396321976e-06,
      "loss": 3.079,
      "step": 54060
    },
    {
      "epoch": 0.5499776204427084,
      "grad_norm": 9.545299530029297,
      "learning_rate": 2.1126426829282794e-06,
      "loss": 3.4689,
      "step": 54065
    },
    {
      "epoch": 0.5500284830729166,
      "grad_norm": 14.429900169372559,
      "learning_rate": 2.1122476361348056e-06,
      "loss": 3.2776,
      "step": 54070
    },
    {
      "epoch": 0.550079345703125,
      "grad_norm": 12.01415729522705,
      "learning_rate": 2.1118525992618837e-06,
      "loss": 3.2068,
      "step": 54075
    },
    {
      "epoch": 0.5501302083333334,
      "grad_norm": 9.425143241882324,
      "learning_rate": 2.1114575723196222e-06,
      "loss": 3.2896,
      "step": 54080
    },
    {
      "epoch": 0.5501810709635416,
      "grad_norm": 12.544426918029785,
      "learning_rate": 2.1110625553181263e-06,
      "loss": 3.1276,
      "step": 54085
    },
    {
      "epoch": 0.55023193359375,
      "grad_norm": 11.73938274383545,
      "learning_rate": 2.110667548267503e-06,
      "loss": 3.239,
      "step": 54090
    },
    {
      "epoch": 0.5502827962239584,
      "grad_norm": 7.796879291534424,
      "learning_rate": 2.1102725511778566e-06,
      "loss": 3.7686,
      "step": 54095
    },
    {
      "epoch": 0.5503336588541666,
      "grad_norm": 15.59985065460205,
      "learning_rate": 2.109877564059296e-06,
      "loss": 3.0617,
      "step": 54100
    },
    {
      "epoch": 0.550384521484375,
      "grad_norm": 18.140830993652344,
      "learning_rate": 2.1094825869219255e-06,
      "loss": 3.4484,
      "step": 54105
    },
    {
      "epoch": 0.5504353841145834,
      "grad_norm": 11.294111251831055,
      "learning_rate": 2.1090876197758508e-06,
      "loss": 3.4144,
      "step": 54110
    },
    {
      "epoch": 0.5504862467447916,
      "grad_norm": 8.844625473022461,
      "learning_rate": 2.108692662631175e-06,
      "loss": 3.117,
      "step": 54115
    },
    {
      "epoch": 0.550537109375,
      "grad_norm": 11.545036315917969,
      "learning_rate": 2.108297715498006e-06,
      "loss": 3.8671,
      "step": 54120
    },
    {
      "epoch": 0.5505879720052084,
      "grad_norm": 10.198552131652832,
      "learning_rate": 2.1079027783864472e-06,
      "loss": 2.9302,
      "step": 54125
    },
    {
      "epoch": 0.5506388346354166,
      "grad_norm": 12.223042488098145,
      "learning_rate": 2.1075078513066026e-06,
      "loss": 3.0021,
      "step": 54130
    },
    {
      "epoch": 0.550689697265625,
      "grad_norm": 10.66480827331543,
      "learning_rate": 2.1071129342685758e-06,
      "loss": 3.2338,
      "step": 54135
    },
    {
      "epoch": 0.5507405598958334,
      "grad_norm": 14.180126190185547,
      "learning_rate": 2.1067180272824723e-06,
      "loss": 2.97,
      "step": 54140
    },
    {
      "epoch": 0.5507914225260416,
      "grad_norm": 10.506875991821289,
      "learning_rate": 2.106323130358395e-06,
      "loss": 3.6545,
      "step": 54145
    },
    {
      "epoch": 0.55084228515625,
      "grad_norm": 9.22285270690918,
      "learning_rate": 2.105928243506446e-06,
      "loss": 3.3539,
      "step": 54150
    },
    {
      "epoch": 0.5508931477864584,
      "grad_norm": 11.771363258361816,
      "learning_rate": 2.1055333667367307e-06,
      "loss": 2.9404,
      "step": 54155
    },
    {
      "epoch": 0.5509440104166666,
      "grad_norm": 9.148426055908203,
      "learning_rate": 2.1051385000593506e-06,
      "loss": 3.7044,
      "step": 54160
    },
    {
      "epoch": 0.550994873046875,
      "grad_norm": 10.212970733642578,
      "learning_rate": 2.1047436434844084e-06,
      "loss": 3.4446,
      "step": 54165
    },
    {
      "epoch": 0.5510457356770834,
      "grad_norm": 13.897493362426758,
      "learning_rate": 2.1043487970220057e-06,
      "loss": 2.7659,
      "step": 54170
    },
    {
      "epoch": 0.5510965983072916,
      "grad_norm": 9.90019702911377,
      "learning_rate": 2.103953960682245e-06,
      "loss": 3.2357,
      "step": 54175
    },
    {
      "epoch": 0.5511474609375,
      "grad_norm": 12.732852935791016,
      "learning_rate": 2.10355913447523e-06,
      "loss": 3.6353,
      "step": 54180
    },
    {
      "epoch": 0.5511983235677084,
      "grad_norm": 14.46264934539795,
      "learning_rate": 2.1031643184110596e-06,
      "loss": 3.103,
      "step": 54185
    },
    {
      "epoch": 0.5512491861979166,
      "grad_norm": 11.0048246383667,
      "learning_rate": 2.102769512499835e-06,
      "loss": 3.1272,
      "step": 54190
    },
    {
      "epoch": 0.551300048828125,
      "grad_norm": 16.377975463867188,
      "learning_rate": 2.10237471675166e-06,
      "loss": 3.5995,
      "step": 54195
    },
    {
      "epoch": 0.5513509114583334,
      "grad_norm": 14.817015647888184,
      "learning_rate": 2.1019799311766327e-06,
      "loss": 3.1704,
      "step": 54200
    },
    {
      "epoch": 0.5514017740885416,
      "grad_norm": 8.748543739318848,
      "learning_rate": 2.101585155784855e-06,
      "loss": 3.4014,
      "step": 54205
    },
    {
      "epoch": 0.55145263671875,
      "grad_norm": 11.921297073364258,
      "learning_rate": 2.1011903905864263e-06,
      "loss": 3.2395,
      "step": 54210
    },
    {
      "epoch": 0.5515034993489584,
      "grad_norm": 12.566230773925781,
      "learning_rate": 2.1007956355914465e-06,
      "loss": 3.5617,
      "step": 54215
    },
    {
      "epoch": 0.5515543619791666,
      "grad_norm": 12.688462257385254,
      "learning_rate": 2.1004008908100166e-06,
      "loss": 3.3699,
      "step": 54220
    },
    {
      "epoch": 0.551605224609375,
      "grad_norm": 14.409541130065918,
      "learning_rate": 2.100006156252235e-06,
      "loss": 3.4616,
      "step": 54225
    },
    {
      "epoch": 0.5516560872395834,
      "grad_norm": 9.698654174804688,
      "learning_rate": 2.0996114319282006e-06,
      "loss": 3.3425,
      "step": 54230
    },
    {
      "epoch": 0.5517069498697916,
      "grad_norm": 8.120442390441895,
      "learning_rate": 2.099216717848013e-06,
      "loss": 3.4025,
      "step": 54235
    },
    {
      "epoch": 0.5517578125,
      "grad_norm": 11.480592727661133,
      "learning_rate": 2.098822014021771e-06,
      "loss": 3.2517,
      "step": 54240
    },
    {
      "epoch": 0.5518086751302084,
      "grad_norm": 18.86686897277832,
      "learning_rate": 2.0984273204595722e-06,
      "loss": 4.1665,
      "step": 54245
    },
    {
      "epoch": 0.5518595377604166,
      "grad_norm": 14.03760814666748,
      "learning_rate": 2.0980326371715155e-06,
      "loss": 3.5493,
      "step": 54250
    },
    {
      "epoch": 0.551910400390625,
      "grad_norm": 7.9557929039001465,
      "learning_rate": 2.097637964167699e-06,
      "loss": 3.3811,
      "step": 54255
    },
    {
      "epoch": 0.5519612630208334,
      "grad_norm": 13.905202865600586,
      "learning_rate": 2.0972433014582193e-06,
      "loss": 3.3876,
      "step": 54260
    },
    {
      "epoch": 0.5520121256510416,
      "grad_norm": 10.3096284866333,
      "learning_rate": 2.0968486490531735e-06,
      "loss": 3.1963,
      "step": 54265
    },
    {
      "epoch": 0.55206298828125,
      "grad_norm": 10.08290958404541,
      "learning_rate": 2.0964540069626606e-06,
      "loss": 3.4092,
      "step": 54270
    },
    {
      "epoch": 0.5521138509114584,
      "grad_norm": 14.121380805969238,
      "learning_rate": 2.0960593751967758e-06,
      "loss": 4.521,
      "step": 54275
    },
    {
      "epoch": 0.5521647135416666,
      "grad_norm": 9.761064529418945,
      "learning_rate": 2.095664753765617e-06,
      "loss": 3.3897,
      "step": 54280
    },
    {
      "epoch": 0.552215576171875,
      "grad_norm": 10.380366325378418,
      "learning_rate": 2.095270142679278e-06,
      "loss": 3.1439,
      "step": 54285
    },
    {
      "epoch": 0.5522664388020834,
      "grad_norm": 9.43635368347168,
      "learning_rate": 2.0948755419478576e-06,
      "loss": 3.3955,
      "step": 54290
    },
    {
      "epoch": 0.5523173014322916,
      "grad_norm": 14.51638412475586,
      "learning_rate": 2.094480951581451e-06,
      "loss": 3.3379,
      "step": 54295
    },
    {
      "epoch": 0.5523681640625,
      "grad_norm": 14.440349578857422,
      "learning_rate": 2.0940863715901518e-06,
      "loss": 3.3932,
      "step": 54300
    },
    {
      "epoch": 0.5524190266927084,
      "grad_norm": 11.273665428161621,
      "learning_rate": 2.093691801984057e-06,
      "loss": 3.5603,
      "step": 54305
    },
    {
      "epoch": 0.5524698893229166,
      "grad_norm": 18.805217742919922,
      "learning_rate": 2.0932972427732613e-06,
      "loss": 3.1834,
      "step": 54310
    },
    {
      "epoch": 0.552520751953125,
      "grad_norm": 11.681893348693848,
      "learning_rate": 2.0929026939678595e-06,
      "loss": 3.1483,
      "step": 54315
    },
    {
      "epoch": 0.5525716145833334,
      "grad_norm": 12.73479175567627,
      "learning_rate": 2.0925081555779457e-06,
      "loss": 3.036,
      "step": 54320
    },
    {
      "epoch": 0.5526224772135416,
      "grad_norm": 16.403100967407227,
      "learning_rate": 2.0921136276136134e-06,
      "loss": 3.5831,
      "step": 54325
    },
    {
      "epoch": 0.55267333984375,
      "grad_norm": 7.362949371337891,
      "learning_rate": 2.0917191100849577e-06,
      "loss": 3.2312,
      "step": 54330
    },
    {
      "epoch": 0.5527242024739584,
      "grad_norm": 8.387533187866211,
      "learning_rate": 2.0913246030020725e-06,
      "loss": 3.2315,
      "step": 54335
    },
    {
      "epoch": 0.5527750651041666,
      "grad_norm": 15.326143264770508,
      "learning_rate": 2.090930106375049e-06,
      "loss": 3.3588,
      "step": 54340
    },
    {
      "epoch": 0.552825927734375,
      "grad_norm": 11.431089401245117,
      "learning_rate": 2.0905356202139827e-06,
      "loss": 3.2824,
      "step": 54345
    },
    {
      "epoch": 0.5528767903645834,
      "grad_norm": 8.371926307678223,
      "learning_rate": 2.090141144528966e-06,
      "loss": 3.5238,
      "step": 54350
    },
    {
      "epoch": 0.5529276529947916,
      "grad_norm": 18.97978973388672,
      "learning_rate": 2.08974667933009e-06,
      "loss": 4.286,
      "step": 54355
    },
    {
      "epoch": 0.552978515625,
      "grad_norm": 11.31583309173584,
      "learning_rate": 2.089352224627448e-06,
      "loss": 3.5481,
      "step": 54360
    },
    {
      "epoch": 0.5530293782552084,
      "grad_norm": 7.666788578033447,
      "learning_rate": 2.088957780431133e-06,
      "loss": 3.1631,
      "step": 54365
    },
    {
      "epoch": 0.5530802408854166,
      "grad_norm": 7.157726287841797,
      "learning_rate": 2.0885633467512352e-06,
      "loss": 3.4058,
      "step": 54370
    },
    {
      "epoch": 0.553131103515625,
      "grad_norm": 9.22585678100586,
      "learning_rate": 2.088168923597847e-06,
      "loss": 3.6942,
      "step": 54375
    },
    {
      "epoch": 0.5531819661458334,
      "grad_norm": 15.494122505187988,
      "learning_rate": 2.0877745109810584e-06,
      "loss": 3.2869,
      "step": 54380
    },
    {
      "epoch": 0.5532328287760416,
      "grad_norm": 8.315948486328125,
      "learning_rate": 2.0873801089109615e-06,
      "loss": 3.2937,
      "step": 54385
    },
    {
      "epoch": 0.55328369140625,
      "grad_norm": 14.754796028137207,
      "learning_rate": 2.0869857173976475e-06,
      "loss": 3.1085,
      "step": 54390
    },
    {
      "epoch": 0.5533345540364584,
      "grad_norm": 16.696014404296875,
      "learning_rate": 2.086591336451205e-06,
      "loss": 2.9874,
      "step": 54395
    },
    {
      "epoch": 0.5533854166666666,
      "grad_norm": 9.401805877685547,
      "learning_rate": 2.0861969660817254e-06,
      "loss": 3.4339,
      "step": 54400
    },
    {
      "epoch": 0.553436279296875,
      "grad_norm": 10.695304870605469,
      "learning_rate": 2.085802606299298e-06,
      "loss": 3.4864,
      "step": 54405
    },
    {
      "epoch": 0.5534871419270834,
      "grad_norm": 8.707365989685059,
      "learning_rate": 2.0854082571140138e-06,
      "loss": 3.4632,
      "step": 54410
    },
    {
      "epoch": 0.5535380045572916,
      "grad_norm": 7.770190238952637,
      "learning_rate": 2.0850139185359608e-06,
      "loss": 3.3616,
      "step": 54415
    },
    {
      "epoch": 0.5535888671875,
      "grad_norm": 15.296331405639648,
      "learning_rate": 2.0846195905752273e-06,
      "loss": 3.1424,
      "step": 54420
    },
    {
      "epoch": 0.5536397298177084,
      "grad_norm": 14.825443267822266,
      "learning_rate": 2.0842252732419045e-06,
      "loss": 3.6531,
      "step": 54425
    },
    {
      "epoch": 0.5536905924479166,
      "grad_norm": 13.824195861816406,
      "learning_rate": 2.083830966546079e-06,
      "loss": 3.5766,
      "step": 54430
    },
    {
      "epoch": 0.553741455078125,
      "grad_norm": 9.839577674865723,
      "learning_rate": 2.0834366704978386e-06,
      "loss": 3.9144,
      "step": 54435
    },
    {
      "epoch": 0.5537923177083334,
      "grad_norm": 11.540263175964355,
      "learning_rate": 2.083042385107274e-06,
      "loss": 3.3083,
      "step": 54440
    },
    {
      "epoch": 0.5538431803385416,
      "grad_norm": 9.633585929870605,
      "learning_rate": 2.08264811038447e-06,
      "loss": 3.3286,
      "step": 54445
    },
    {
      "epoch": 0.55389404296875,
      "grad_norm": 9.796147346496582,
      "learning_rate": 2.082253846339516e-06,
      "loss": 3.2135,
      "step": 54450
    },
    {
      "epoch": 0.5539449055989584,
      "grad_norm": 9.02984619140625,
      "learning_rate": 2.081859592982497e-06,
      "loss": 3.2556,
      "step": 54455
    },
    {
      "epoch": 0.5539957682291666,
      "grad_norm": 14.30068302154541,
      "learning_rate": 2.081465350323502e-06,
      "loss": 3.1485,
      "step": 54460
    },
    {
      "epoch": 0.554046630859375,
      "grad_norm": 8.195463180541992,
      "learning_rate": 2.0810711183726173e-06,
      "loss": 3.653,
      "step": 54465
    },
    {
      "epoch": 0.5540974934895834,
      "grad_norm": 16.552175521850586,
      "learning_rate": 2.0806768971399283e-06,
      "loss": 3.0628,
      "step": 54470
    },
    {
      "epoch": 0.5541483561197916,
      "grad_norm": 15.321861267089844,
      "learning_rate": 2.0802826866355207e-06,
      "loss": 3.3406,
      "step": 54475
    },
    {
      "epoch": 0.55419921875,
      "grad_norm": 11.74885368347168,
      "learning_rate": 2.0798884868694824e-06,
      "loss": 2.9268,
      "step": 54480
    },
    {
      "epoch": 0.5542500813802084,
      "grad_norm": 14.062115669250488,
      "learning_rate": 2.079494297851897e-06,
      "loss": 3.3394,
      "step": 54485
    },
    {
      "epoch": 0.5543009440104166,
      "grad_norm": 9.796475410461426,
      "learning_rate": 2.079100119592851e-06,
      "loss": 3.3353,
      "step": 54490
    },
    {
      "epoch": 0.554351806640625,
      "grad_norm": 12.919037818908691,
      "learning_rate": 2.078705952102427e-06,
      "loss": 3.256,
      "step": 54495
    },
    {
      "epoch": 0.5544026692708334,
      "grad_norm": 9.963675498962402,
      "learning_rate": 2.0783117953907123e-06,
      "loss": 3.8251,
      "step": 54500
    },
    {
      "epoch": 0.5544535319010416,
      "grad_norm": 12.951997756958008,
      "learning_rate": 2.077917649467791e-06,
      "loss": 3.1518,
      "step": 54505
    },
    {
      "epoch": 0.55450439453125,
      "grad_norm": 14.596063613891602,
      "learning_rate": 2.0775235143437463e-06,
      "loss": 3.0804,
      "step": 54510
    },
    {
      "epoch": 0.5545552571614584,
      "grad_norm": 9.925080299377441,
      "learning_rate": 2.0771293900286616e-06,
      "loss": 3.5082,
      "step": 54515
    },
    {
      "epoch": 0.5546061197916666,
      "grad_norm": 8.489387512207031,
      "learning_rate": 2.076735276532622e-06,
      "loss": 3.8833,
      "step": 54520
    },
    {
      "epoch": 0.554656982421875,
      "grad_norm": 17.317609786987305,
      "learning_rate": 2.0763411738657097e-06,
      "loss": 3.565,
      "step": 54525
    },
    {
      "epoch": 0.5547078450520834,
      "grad_norm": 9.189920425415039,
      "learning_rate": 2.0759470820380074e-06,
      "loss": 3.6864,
      "step": 54530
    },
    {
      "epoch": 0.5547587076822916,
      "grad_norm": 15.213544845581055,
      "learning_rate": 2.0755530010596e-06,
      "loss": 3.2918,
      "step": 54535
    },
    {
      "epoch": 0.5548095703125,
      "grad_norm": 11.895380020141602,
      "learning_rate": 2.0751589309405673e-06,
      "loss": 3.0242,
      "step": 54540
    },
    {
      "epoch": 0.5548604329427084,
      "grad_norm": 10.334179878234863,
      "learning_rate": 2.0747648716909936e-06,
      "loss": 3.477,
      "step": 54545
    },
    {
      "epoch": 0.5549112955729166,
      "grad_norm": 11.6627779006958,
      "learning_rate": 2.074370823320958e-06,
      "loss": 3.2691,
      "step": 54550
    },
    {
      "epoch": 0.554962158203125,
      "grad_norm": 12.616467475891113,
      "learning_rate": 2.073976785840546e-06,
      "loss": 3.4009,
      "step": 54555
    },
    {
      "epoch": 0.5550130208333334,
      "grad_norm": 9.53610897064209,
      "learning_rate": 2.0735827592598356e-06,
      "loss": 3.8242,
      "step": 54560
    },
    {
      "epoch": 0.5550638834635416,
      "grad_norm": 13.089378356933594,
      "learning_rate": 2.0731887435889105e-06,
      "loss": 3.2991,
      "step": 54565
    },
    {
      "epoch": 0.55511474609375,
      "grad_norm": 12.170549392700195,
      "learning_rate": 2.072794738837849e-06,
      "loss": 3.2447,
      "step": 54570
    },
    {
      "epoch": 0.5551656087239584,
      "grad_norm": 9.022259712219238,
      "learning_rate": 2.0724007450167335e-06,
      "loss": 3.2896,
      "step": 54575
    },
    {
      "epoch": 0.5552164713541666,
      "grad_norm": 14.788315773010254,
      "learning_rate": 2.072006762135644e-06,
      "loss": 3.1942,
      "step": 54580
    },
    {
      "epoch": 0.555267333984375,
      "grad_norm": 12.025224685668945,
      "learning_rate": 2.0716127902046595e-06,
      "loss": 3.3759,
      "step": 54585
    },
    {
      "epoch": 0.5553181966145834,
      "grad_norm": 13.401224136352539,
      "learning_rate": 2.0712188292338596e-06,
      "loss": 3.3078,
      "step": 54590
    },
    {
      "epoch": 0.5553690592447916,
      "grad_norm": 10.349852561950684,
      "learning_rate": 2.0708248792333257e-06,
      "loss": 3.5264,
      "step": 54595
    },
    {
      "epoch": 0.555419921875,
      "grad_norm": 9.907820701599121,
      "learning_rate": 2.0704309402131344e-06,
      "loss": 3.2487,
      "step": 54600
    },
    {
      "epoch": 0.5554707845052084,
      "grad_norm": 12.225205421447754,
      "learning_rate": 2.0700370121833666e-06,
      "loss": 3.2967,
      "step": 54605
    },
    {
      "epoch": 0.5555216471354166,
      "grad_norm": 12.409201622009277,
      "learning_rate": 2.069643095154099e-06,
      "loss": 3.1588,
      "step": 54610
    },
    {
      "epoch": 0.555572509765625,
      "grad_norm": 7.892660140991211,
      "learning_rate": 2.069249189135411e-06,
      "loss": 3.132,
      "step": 54615
    },
    {
      "epoch": 0.5556233723958334,
      "grad_norm": 17.70599937438965,
      "learning_rate": 2.068855294137381e-06,
      "loss": 3.6297,
      "step": 54620
    },
    {
      "epoch": 0.5556742350260416,
      "grad_norm": 12.921568870544434,
      "learning_rate": 2.0684614101700852e-06,
      "loss": 3.8524,
      "step": 54625
    },
    {
      "epoch": 0.55572509765625,
      "grad_norm": 10.770087242126465,
      "learning_rate": 2.0680675372436025e-06,
      "loss": 3.2557,
      "step": 54630
    },
    {
      "epoch": 0.5557759602864584,
      "grad_norm": 11.178862571716309,
      "learning_rate": 2.06767367536801e-06,
      "loss": 3.6751,
      "step": 54635
    },
    {
      "epoch": 0.5558268229166666,
      "grad_norm": 12.682517051696777,
      "learning_rate": 2.0672798245533833e-06,
      "loss": 3.5117,
      "step": 54640
    },
    {
      "epoch": 0.555877685546875,
      "grad_norm": 10.491020202636719,
      "learning_rate": 2.0668859848097995e-06,
      "loss": 3.1276,
      "step": 54645
    },
    {
      "epoch": 0.5559285481770834,
      "grad_norm": 14.331825256347656,
      "learning_rate": 2.0664921561473363e-06,
      "loss": 3.2829,
      "step": 54650
    },
    {
      "epoch": 0.5559794108072916,
      "grad_norm": 10.737212181091309,
      "learning_rate": 2.0660983385760676e-06,
      "loss": 3.1904,
      "step": 54655
    },
    {
      "epoch": 0.5560302734375,
      "grad_norm": 14.648859977722168,
      "learning_rate": 2.0657045321060704e-06,
      "loss": 2.9052,
      "step": 54660
    },
    {
      "epoch": 0.5560811360677084,
      "grad_norm": 15.264992713928223,
      "learning_rate": 2.0653107367474193e-06,
      "loss": 3.3429,
      "step": 54665
    },
    {
      "epoch": 0.5561319986979166,
      "grad_norm": 15.288416862487793,
      "learning_rate": 2.0649169525101904e-06,
      "loss": 3.1437,
      "step": 54670
    },
    {
      "epoch": 0.556182861328125,
      "grad_norm": 10.921524047851562,
      "learning_rate": 2.0645231794044584e-06,
      "loss": 3.3432,
      "step": 54675
    },
    {
      "epoch": 0.5562337239583334,
      "grad_norm": 15.968602180480957,
      "learning_rate": 2.064129417440297e-06,
      "loss": 3.2523,
      "step": 54680
    },
    {
      "epoch": 0.5562845865885416,
      "grad_norm": 12.748848915100098,
      "learning_rate": 2.0637356666277813e-06,
      "loss": 3.4739,
      "step": 54685
    },
    {
      "epoch": 0.55633544921875,
      "grad_norm": 9.147810935974121,
      "learning_rate": 2.0633419269769856e-06,
      "loss": 3.4638,
      "step": 54690
    },
    {
      "epoch": 0.5563863118489584,
      "grad_norm": 13.364506721496582,
      "learning_rate": 2.062948198497983e-06,
      "loss": 3.3763,
      "step": 54695
    },
    {
      "epoch": 0.5564371744791666,
      "grad_norm": 8.214322090148926,
      "learning_rate": 2.0625544812008475e-06,
      "loss": 3.2868,
      "step": 54700
    },
    {
      "epoch": 0.556488037109375,
      "grad_norm": 11.215323448181152,
      "learning_rate": 2.06216077509565e-06,
      "loss": 3.552,
      "step": 54705
    },
    {
      "epoch": 0.5565388997395834,
      "grad_norm": 10.461332321166992,
      "learning_rate": 2.0617670801924674e-06,
      "loss": 3.3382,
      "step": 54710
    },
    {
      "epoch": 0.5565897623697916,
      "grad_norm": 11.168867111206055,
      "learning_rate": 2.0613733965013693e-06,
      "loss": 3.1396,
      "step": 54715
    },
    {
      "epoch": 0.556640625,
      "grad_norm": 12.699355125427246,
      "learning_rate": 2.06097972403243e-06,
      "loss": 3.6432,
      "step": 54720
    },
    {
      "epoch": 0.5566914876302084,
      "grad_norm": 9.216696739196777,
      "learning_rate": 2.0605860627957187e-06,
      "loss": 3.6198,
      "step": 54725
    },
    {
      "epoch": 0.5567423502604166,
      "grad_norm": 10.07630443572998,
      "learning_rate": 2.0601924128013095e-06,
      "loss": 3.3492,
      "step": 54730
    },
    {
      "epoch": 0.556793212890625,
      "grad_norm": 13.220158576965332,
      "learning_rate": 2.059798774059274e-06,
      "loss": 3.1742,
      "step": 54735
    },
    {
      "epoch": 0.5568440755208334,
      "grad_norm": 15.62512493133545,
      "learning_rate": 2.059405146579681e-06,
      "loss": 3.2923,
      "step": 54740
    },
    {
      "epoch": 0.5568949381510416,
      "grad_norm": 12.246667861938477,
      "learning_rate": 2.0590115303726037e-06,
      "loss": 3.3294,
      "step": 54745
    },
    {
      "epoch": 0.55694580078125,
      "grad_norm": 7.680668830871582,
      "learning_rate": 2.0586179254481124e-06,
      "loss": 2.9745,
      "step": 54750
    },
    {
      "epoch": 0.5569966634114584,
      "grad_norm": 10.86223030090332,
      "learning_rate": 2.0582243318162767e-06,
      "loss": 3.0465,
      "step": 54755
    },
    {
      "epoch": 0.5570475260416666,
      "grad_norm": 14.095746994018555,
      "learning_rate": 2.0578307494871656e-06,
      "loss": 3.6199,
      "step": 54760
    },
    {
      "epoch": 0.557098388671875,
      "grad_norm": 16.408039093017578,
      "learning_rate": 2.057437178470851e-06,
      "loss": 3.1182,
      "step": 54765
    },
    {
      "epoch": 0.5571492513020834,
      "grad_norm": 11.257684707641602,
      "learning_rate": 2.0570436187774013e-06,
      "loss": 3.4447,
      "step": 54770
    },
    {
      "epoch": 0.5572001139322916,
      "grad_norm": 17.170454025268555,
      "learning_rate": 2.056650070416886e-06,
      "loss": 3.0475,
      "step": 54775
    },
    {
      "epoch": 0.5572509765625,
      "grad_norm": 8.76863956451416,
      "learning_rate": 2.056256533399372e-06,
      "loss": 3.2926,
      "step": 54780
    },
    {
      "epoch": 0.5573018391927084,
      "grad_norm": 16.362281799316406,
      "learning_rate": 2.0558630077349305e-06,
      "loss": 3.7876,
      "step": 54785
    },
    {
      "epoch": 0.5573527018229166,
      "grad_norm": 9.774134635925293,
      "learning_rate": 2.055469493433629e-06,
      "loss": 3.3734,
      "step": 54790
    },
    {
      "epoch": 0.557403564453125,
      "grad_norm": 13.329634666442871,
      "learning_rate": 2.0550759905055346e-06,
      "loss": 3.4319,
      "step": 54795
    },
    {
      "epoch": 0.5574544270833334,
      "grad_norm": 7.764184474945068,
      "learning_rate": 2.054682498960715e-06,
      "loss": 3.158,
      "step": 54800
    },
    {
      "epoch": 0.5575052897135416,
      "grad_norm": 12.539365768432617,
      "learning_rate": 2.0542890188092398e-06,
      "loss": 3.1955,
      "step": 54805
    },
    {
      "epoch": 0.55755615234375,
      "grad_norm": 17.243762969970703,
      "learning_rate": 2.053895550061173e-06,
      "loss": 3.5431,
      "step": 54810
    },
    {
      "epoch": 0.5576070149739584,
      "grad_norm": 12.289092063903809,
      "learning_rate": 2.0535020927265835e-06,
      "loss": 2.9041,
      "step": 54815
    },
    {
      "epoch": 0.5576578776041666,
      "grad_norm": 13.950140953063965,
      "learning_rate": 2.053108646815536e-06,
      "loss": 3.3025,
      "step": 54820
    },
    {
      "epoch": 0.557708740234375,
      "grad_norm": 11.045097351074219,
      "learning_rate": 2.052715212338099e-06,
      "loss": 3.3789,
      "step": 54825
    },
    {
      "epoch": 0.5577596028645834,
      "grad_norm": 12.279302597045898,
      "learning_rate": 2.0523217893043372e-06,
      "loss": 3.5081,
      "step": 54830
    },
    {
      "epoch": 0.5578104654947916,
      "grad_norm": 11.898847579956055,
      "learning_rate": 2.0519283777243154e-06,
      "loss": 3.2755,
      "step": 54835
    },
    {
      "epoch": 0.557861328125,
      "grad_norm": 10.195027351379395,
      "learning_rate": 2.0515349776081005e-06,
      "loss": 3.508,
      "step": 54840
    },
    {
      "epoch": 0.5579121907552084,
      "grad_norm": 12.330281257629395,
      "learning_rate": 2.0511415889657576e-06,
      "loss": 3.2828,
      "step": 54845
    },
    {
      "epoch": 0.5579630533854166,
      "grad_norm": 8.676578521728516,
      "learning_rate": 2.0507482118073497e-06,
      "loss": 3.3831,
      "step": 54850
    },
    {
      "epoch": 0.558013916015625,
      "grad_norm": 6.987087726593018,
      "learning_rate": 2.050354846142942e-06,
      "loss": 3.5374,
      "step": 54855
    },
    {
      "epoch": 0.5580647786458334,
      "grad_norm": 11.584981918334961,
      "learning_rate": 2.0499614919825998e-06,
      "loss": 3.3075,
      "step": 54860
    },
    {
      "epoch": 0.5581156412760416,
      "grad_norm": 9.410149574279785,
      "learning_rate": 2.0495681493363862e-06,
      "loss": 3.3489,
      "step": 54865
    },
    {
      "epoch": 0.55816650390625,
      "grad_norm": 12.8016996383667,
      "learning_rate": 2.049174818214365e-06,
      "loss": 3.0027,
      "step": 54870
    },
    {
      "epoch": 0.5582173665364584,
      "grad_norm": 11.745713233947754,
      "learning_rate": 2.048781498626598e-06,
      "loss": 3.2402,
      "step": 54875
    },
    {
      "epoch": 0.5582682291666666,
      "grad_norm": 8.826766967773438,
      "learning_rate": 2.0483881905831503e-06,
      "loss": 3.4843,
      "step": 54880
    },
    {
      "epoch": 0.558319091796875,
      "grad_norm": 13.848268508911133,
      "learning_rate": 2.047994894094084e-06,
      "loss": 3.2401,
      "step": 54885
    },
    {
      "epoch": 0.5583699544270834,
      "grad_norm": 15.071582794189453,
      "learning_rate": 2.0476016091694607e-06,
      "loss": 3.133,
      "step": 54890
    },
    {
      "epoch": 0.5584208170572916,
      "grad_norm": 16.107677459716797,
      "learning_rate": 2.0472083358193425e-06,
      "loss": 3.4667,
      "step": 54895
    },
    {
      "epoch": 0.5584716796875,
      "grad_norm": 13.1598539352417,
      "learning_rate": 2.0468150740537924e-06,
      "loss": 3.3724,
      "step": 54900
    },
    {
      "epoch": 0.5585225423177084,
      "grad_norm": 14.7920503616333,
      "learning_rate": 2.0464218238828714e-06,
      "loss": 3.6829,
      "step": 54905
    },
    {
      "epoch": 0.5585734049479166,
      "grad_norm": 13.697108268737793,
      "learning_rate": 2.04602858531664e-06,
      "loss": 3.2593,
      "step": 54910
    },
    {
      "epoch": 0.558624267578125,
      "grad_norm": 10.984542846679688,
      "learning_rate": 2.0456353583651594e-06,
      "loss": 3.4175,
      "step": 54915
    },
    {
      "epoch": 0.5586751302083334,
      "grad_norm": 11.960285186767578,
      "learning_rate": 2.0452421430384913e-06,
      "loss": 3.1616,
      "step": 54920
    },
    {
      "epoch": 0.5587259928385416,
      "grad_norm": 7.636691093444824,
      "learning_rate": 2.044848939346695e-06,
      "loss": 3.4549,
      "step": 54925
    },
    {
      "epoch": 0.55877685546875,
      "grad_norm": 11.358219146728516,
      "learning_rate": 2.0444557472998295e-06,
      "loss": 3.4507,
      "step": 54930
    },
    {
      "epoch": 0.5588277180989584,
      "grad_norm": 8.738129615783691,
      "learning_rate": 2.044062566907957e-06,
      "loss": 3.0457,
      "step": 54935
    },
    {
      "epoch": 0.5588785807291666,
      "grad_norm": 10.17060661315918,
      "learning_rate": 2.0436693981811356e-06,
      "loss": 3.1116,
      "step": 54940
    },
    {
      "epoch": 0.558929443359375,
      "grad_norm": 10.410772323608398,
      "learning_rate": 2.0432762411294246e-06,
      "loss": 3.4821,
      "step": 54945
    },
    {
      "epoch": 0.5589803059895834,
      "grad_norm": 12.336195945739746,
      "learning_rate": 2.042883095762882e-06,
      "loss": 3.1465,
      "step": 54950
    },
    {
      "epoch": 0.5590311686197916,
      "grad_norm": 12.679303169250488,
      "learning_rate": 2.0424899620915674e-06,
      "loss": 3.4706,
      "step": 54955
    },
    {
      "epoch": 0.55908203125,
      "grad_norm": 9.3101224899292,
      "learning_rate": 2.042096840125539e-06,
      "loss": 2.8378,
      "step": 54960
    },
    {
      "epoch": 0.5591328938802084,
      "grad_norm": 12.529419898986816,
      "learning_rate": 2.041703729874854e-06,
      "loss": 3.4076,
      "step": 54965
    },
    {
      "epoch": 0.5591837565104166,
      "grad_norm": 48.43590545654297,
      "learning_rate": 2.04131063134957e-06,
      "loss": 3.2729,
      "step": 54970
    },
    {
      "epoch": 0.559234619140625,
      "grad_norm": 15.862727165222168,
      "learning_rate": 2.0409175445597455e-06,
      "loss": 3.2223,
      "step": 54975
    },
    {
      "epoch": 0.5592854817708334,
      "grad_norm": 14.386452674865723,
      "learning_rate": 2.040524469515437e-06,
      "loss": 3.4878,
      "step": 54980
    },
    {
      "epoch": 0.5593363444010416,
      "grad_norm": 9.835855484008789,
      "learning_rate": 2.0401314062267007e-06,
      "loss": 3.6074,
      "step": 54985
    },
    {
      "epoch": 0.55938720703125,
      "grad_norm": 12.619324684143066,
      "learning_rate": 2.039738354703593e-06,
      "loss": 3.1502,
      "step": 54990
    },
    {
      "epoch": 0.5594380696614584,
      "grad_norm": 7.411263942718506,
      "learning_rate": 2.0393453149561706e-06,
      "loss": 3.6218,
      "step": 54995
    },
    {
      "epoch": 0.5594889322916666,
      "grad_norm": 7.863607883453369,
      "learning_rate": 2.03895228699449e-06,
      "loss": 3.1499,
      "step": 55000
    },
    {
      "epoch": 0.559539794921875,
      "grad_norm": 14.606014251708984,
      "learning_rate": 2.038559270828605e-06,
      "loss": 3.4186,
      "step": 55005
    },
    {
      "epoch": 0.5595906575520834,
      "grad_norm": 15.155403137207031,
      "learning_rate": 2.038166266468571e-06,
      "loss": 3.3658,
      "step": 55010
    },
    {
      "epoch": 0.5596415201822916,
      "grad_norm": 11.197957038879395,
      "learning_rate": 2.0377732739244447e-06,
      "loss": 2.9844,
      "step": 55015
    },
    {
      "epoch": 0.5596923828125,
      "grad_norm": 14.66739273071289,
      "learning_rate": 2.0373802932062798e-06,
      "loss": 3.361,
      "step": 55020
    },
    {
      "epoch": 0.5597432454427084,
      "grad_norm": 13.307490348815918,
      "learning_rate": 2.0369873243241293e-06,
      "loss": 3.3566,
      "step": 55025
    },
    {
      "epoch": 0.5597941080729166,
      "grad_norm": 9.766070365905762,
      "learning_rate": 2.036594367288049e-06,
      "loss": 3.1445,
      "step": 55030
    },
    {
      "epoch": 0.559844970703125,
      "grad_norm": 8.896641731262207,
      "learning_rate": 2.0362014221080924e-06,
      "loss": 3.0715,
      "step": 55035
    },
    {
      "epoch": 0.5598958333333334,
      "grad_norm": 10.176886558532715,
      "learning_rate": 2.035808488794312e-06,
      "loss": 3.1849,
      "step": 55040
    },
    {
      "epoch": 0.5599466959635416,
      "grad_norm": 12.1671724319458,
      "learning_rate": 2.0354155673567612e-06,
      "loss": 3.9336,
      "step": 55045
    },
    {
      "epoch": 0.55999755859375,
      "grad_norm": 15.339162826538086,
      "learning_rate": 2.0350226578054935e-06,
      "loss": 3.4378,
      "step": 55050
    },
    {
      "epoch": 0.5600484212239584,
      "grad_norm": 11.908868789672852,
      "learning_rate": 2.0346297601505605e-06,
      "loss": 3.4508,
      "step": 55055
    },
    {
      "epoch": 0.5600992838541666,
      "grad_norm": 13.39874267578125,
      "learning_rate": 2.0342368744020155e-06,
      "loss": 3.5067,
      "step": 55060
    },
    {
      "epoch": 0.560150146484375,
      "grad_norm": 12.38049602508545,
      "learning_rate": 2.033844000569908e-06,
      "loss": 3.3802,
      "step": 55065
    },
    {
      "epoch": 0.5602010091145834,
      "grad_norm": 9.690685272216797,
      "learning_rate": 2.033451138664293e-06,
      "loss": 3.0584,
      "step": 55070
    },
    {
      "epoch": 0.5602518717447916,
      "grad_norm": 8.989315032958984,
      "learning_rate": 2.0330582886952196e-06,
      "loss": 3.4424,
      "step": 55075
    },
    {
      "epoch": 0.560302734375,
      "grad_norm": 17.36121940612793,
      "learning_rate": 2.0326654506727393e-06,
      "loss": 3.2294,
      "step": 55080
    },
    {
      "epoch": 0.5603535970052084,
      "grad_norm": 15.613003730773926,
      "learning_rate": 2.0322726246069015e-06,
      "loss": 3.4868,
      "step": 55085
    },
    {
      "epoch": 0.5604044596354166,
      "grad_norm": 13.500045776367188,
      "learning_rate": 2.031879810507759e-06,
      "loss": 3.1505,
      "step": 55090
    },
    {
      "epoch": 0.560455322265625,
      "grad_norm": 9.693497657775879,
      "learning_rate": 2.0314870083853604e-06,
      "loss": 3.2669,
      "step": 55095
    },
    {
      "epoch": 0.5605061848958334,
      "grad_norm": 8.206704139709473,
      "learning_rate": 2.031094218249756e-06,
      "loss": 2.9043,
      "step": 55100
    },
    {
      "epoch": 0.5605570475260416,
      "grad_norm": 12.259492874145508,
      "learning_rate": 2.0307014401109935e-06,
      "loss": 3.3699,
      "step": 55105
    },
    {
      "epoch": 0.56060791015625,
      "grad_norm": 10.723469734191895,
      "learning_rate": 2.0303086739791246e-06,
      "loss": 3.4771,
      "step": 55110
    },
    {
      "epoch": 0.5606587727864584,
      "grad_norm": 15.417384147644043,
      "learning_rate": 2.029915919864197e-06,
      "loss": 3.1352,
      "step": 55115
    },
    {
      "epoch": 0.5607096354166666,
      "grad_norm": 8.692520141601562,
      "learning_rate": 2.029523177776258e-06,
      "loss": 3.1695,
      "step": 55120
    },
    {
      "epoch": 0.560760498046875,
      "grad_norm": 11.361414909362793,
      "learning_rate": 2.0291304477253577e-06,
      "loss": 3.0225,
      "step": 55125
    },
    {
      "epoch": 0.5608113606770834,
      "grad_norm": 8.461770057678223,
      "learning_rate": 2.028737729721544e-06,
      "loss": 3.4996,
      "step": 55130
    },
    {
      "epoch": 0.5608622233072916,
      "grad_norm": 14.262664794921875,
      "learning_rate": 2.028345023774863e-06,
      "loss": 3.5087,
      "step": 55135
    },
    {
      "epoch": 0.5609130859375,
      "grad_norm": 10.160379409790039,
      "learning_rate": 2.027952329895362e-06,
      "loss": 3.847,
      "step": 55140
    },
    {
      "epoch": 0.5609639485677084,
      "grad_norm": 18.16651725769043,
      "learning_rate": 2.0275596480930902e-06,
      "loss": 3.2377,
      "step": 55145
    },
    {
      "epoch": 0.5610148111979166,
      "grad_norm": 15.301115036010742,
      "learning_rate": 2.0271669783780923e-06,
      "loss": 3.1241,
      "step": 55150
    },
    {
      "epoch": 0.561065673828125,
      "grad_norm": 10.699058532714844,
      "learning_rate": 2.0267743207604156e-06,
      "loss": 3.7011,
      "step": 55155
    },
    {
      "epoch": 0.5611165364583334,
      "grad_norm": 7.996247291564941,
      "learning_rate": 2.0263816752501046e-06,
      "loss": 3.0694,
      "step": 55160
    },
    {
      "epoch": 0.5611673990885416,
      "grad_norm": 10.596744537353516,
      "learning_rate": 2.025989041857207e-06,
      "loss": 3.2339,
      "step": 55165
    },
    {
      "epoch": 0.56121826171875,
      "grad_norm": 11.930685997009277,
      "learning_rate": 2.0255964205917676e-06,
      "loss": 3.6971,
      "step": 55170
    },
    {
      "epoch": 0.5612691243489584,
      "grad_norm": 10.017671585083008,
      "learning_rate": 2.025203811463831e-06,
      "loss": 3.1608,
      "step": 55175
    },
    {
      "epoch": 0.5613199869791666,
      "grad_norm": 14.618124008178711,
      "learning_rate": 2.024811214483442e-06,
      "loss": 3.3134,
      "step": 55180
    },
    {
      "epoch": 0.561370849609375,
      "grad_norm": 11.243496894836426,
      "learning_rate": 2.0244186296606456e-06,
      "loss": 3.3153,
      "step": 55185
    },
    {
      "epoch": 0.5614217122395834,
      "grad_norm": 14.103216171264648,
      "learning_rate": 2.0240260570054866e-06,
      "loss": 3.324,
      "step": 55190
    },
    {
      "epoch": 0.5614725748697916,
      "grad_norm": 10.062374114990234,
      "learning_rate": 2.0236334965280073e-06,
      "loss": 3.4501,
      "step": 55195
    },
    {
      "epoch": 0.5615234375,
      "grad_norm": 11.451557159423828,
      "learning_rate": 2.023240948238252e-06,
      "loss": 3.3229,
      "step": 55200
    },
    {
      "epoch": 0.5615743001302084,
      "grad_norm": 11.037567138671875,
      "learning_rate": 2.022848412146265e-06,
      "loss": 3.2838,
      "step": 55205
    },
    {
      "epoch": 0.5616251627604166,
      "grad_norm": 9.876566886901855,
      "learning_rate": 2.0224558882620874e-06,
      "loss": 3.4612,
      "step": 55210
    },
    {
      "epoch": 0.561676025390625,
      "grad_norm": 8.376761436462402,
      "learning_rate": 2.0220633765957634e-06,
      "loss": 3.2515,
      "step": 55215
    },
    {
      "epoch": 0.5617268880208334,
      "grad_norm": 12.039939880371094,
      "learning_rate": 2.0216708771573336e-06,
      "loss": 3.21,
      "step": 55220
    },
    {
      "epoch": 0.5617777506510416,
      "grad_norm": 9.83841323852539,
      "learning_rate": 2.0212783899568418e-06,
      "loss": 3.0499,
      "step": 55225
    },
    {
      "epoch": 0.56182861328125,
      "grad_norm": 13.492292404174805,
      "learning_rate": 2.020885915004329e-06,
      "loss": 3.6028,
      "step": 55230
    },
    {
      "epoch": 0.5618794759114584,
      "grad_norm": 9.677485466003418,
      "learning_rate": 2.0204934523098357e-06,
      "loss": 3.1137,
      "step": 55235
    },
    {
      "epoch": 0.5619303385416666,
      "grad_norm": 14.501381874084473,
      "learning_rate": 2.020101001883404e-06,
      "loss": 3.2436,
      "step": 55240
    },
    {
      "epoch": 0.561981201171875,
      "grad_norm": 8.117530822753906,
      "learning_rate": 2.0197085637350757e-06,
      "loss": 3.0716,
      "step": 55245
    },
    {
      "epoch": 0.5620320638020834,
      "grad_norm": 16.134164810180664,
      "learning_rate": 2.0193161378748883e-06,
      "loss": 3.0136,
      "step": 55250
    },
    {
      "epoch": 0.5620829264322916,
      "grad_norm": 7.026928901672363,
      "learning_rate": 2.018923724312884e-06,
      "loss": 3.3786,
      "step": 55255
    },
    {
      "epoch": 0.5621337890625,
      "grad_norm": 12.431653022766113,
      "learning_rate": 2.018531323059103e-06,
      "loss": 3.3799,
      "step": 55260
    },
    {
      "epoch": 0.5621846516927084,
      "grad_norm": 12.698046684265137,
      "learning_rate": 2.018138934123583e-06,
      "loss": 3.7676,
      "step": 55265
    },
    {
      "epoch": 0.5622355143229166,
      "grad_norm": 10.076068878173828,
      "learning_rate": 2.0177465575163652e-06,
      "loss": 3.2673,
      "step": 55270
    },
    {
      "epoch": 0.562286376953125,
      "grad_norm": 16.043195724487305,
      "learning_rate": 2.017354193247486e-06,
      "loss": 3.6135,
      "step": 55275
    },
    {
      "epoch": 0.5623372395833334,
      "grad_norm": 12.170609474182129,
      "learning_rate": 2.0169618413269864e-06,
      "loss": 3.4841,
      "step": 55280
    },
    {
      "epoch": 0.5623881022135416,
      "grad_norm": 15.926358222961426,
      "learning_rate": 2.016569501764904e-06,
      "loss": 3.3622,
      "step": 55285
    },
    {
      "epoch": 0.56243896484375,
      "grad_norm": 9.513419151306152,
      "learning_rate": 2.0161771745712753e-06,
      "loss": 3.1698,
      "step": 55290
    },
    {
      "epoch": 0.5624898274739584,
      "grad_norm": 11.159460067749023,
      "learning_rate": 2.0157848597561388e-06,
      "loss": 3.432,
      "step": 55295
    },
    {
      "epoch": 0.5625406901041666,
      "grad_norm": 14.436287879943848,
      "learning_rate": 2.015392557329533e-06,
      "loss": 3.3272,
      "step": 55300
    },
    {
      "epoch": 0.562591552734375,
      "grad_norm": 11.197450637817383,
      "learning_rate": 2.015000267301493e-06,
      "loss": 3.2097,
      "step": 55305
    },
    {
      "epoch": 0.5626424153645834,
      "grad_norm": 9.489493370056152,
      "learning_rate": 2.014607989682057e-06,
      "loss": 3.3393,
      "step": 55310
    },
    {
      "epoch": 0.5626932779947916,
      "grad_norm": 9.723332405090332,
      "learning_rate": 2.014215724481259e-06,
      "loss": 3.5027,
      "step": 55315
    },
    {
      "epoch": 0.562744140625,
      "grad_norm": 8.030701637268066,
      "learning_rate": 2.0138234717091375e-06,
      "loss": 3.4648,
      "step": 55320
    },
    {
      "epoch": 0.5627950032552084,
      "grad_norm": 11.252326011657715,
      "learning_rate": 2.0134312313757277e-06,
      "loss": 3.2391,
      "step": 55325
    },
    {
      "epoch": 0.5628458658854166,
      "grad_norm": 9.977880477905273,
      "learning_rate": 2.013039003491063e-06,
      "loss": 3.6818,
      "step": 55330
    },
    {
      "epoch": 0.562896728515625,
      "grad_norm": 12.451231002807617,
      "learning_rate": 2.012646788065181e-06,
      "loss": 3.3159,
      "step": 55335
    },
    {
      "epoch": 0.5629475911458334,
      "grad_norm": 14.259267807006836,
      "learning_rate": 2.0122545851081154e-06,
      "loss": 3.3659,
      "step": 55340
    },
    {
      "epoch": 0.5629984537760416,
      "grad_norm": 8.799775123596191,
      "learning_rate": 2.011862394629901e-06,
      "loss": 3.3504,
      "step": 55345
    },
    {
      "epoch": 0.56304931640625,
      "grad_norm": 13.567157745361328,
      "learning_rate": 2.011470216640571e-06,
      "loss": 3.4156,
      "step": 55350
    },
    {
      "epoch": 0.5631001790364584,
      "grad_norm": 8.953998565673828,
      "learning_rate": 2.0110780511501593e-06,
      "loss": 3.1224,
      "step": 55355
    },
    {
      "epoch": 0.5631510416666666,
      "grad_norm": 13.455957412719727,
      "learning_rate": 2.010685898168701e-06,
      "loss": 3.1641,
      "step": 55360
    },
    {
      "epoch": 0.563201904296875,
      "grad_norm": 10.784248352050781,
      "learning_rate": 2.0102937577062274e-06,
      "loss": 3.3754,
      "step": 55365
    },
    {
      "epoch": 0.5632527669270834,
      "grad_norm": 15.13234806060791,
      "learning_rate": 2.009901629772771e-06,
      "loss": 3.2727,
      "step": 55370
    },
    {
      "epoch": 0.5633036295572916,
      "grad_norm": 12.457405090332031,
      "learning_rate": 2.0095095143783666e-06,
      "loss": 3.4566,
      "step": 55375
    },
    {
      "epoch": 0.5633544921875,
      "grad_norm": 13.739633560180664,
      "learning_rate": 2.0091174115330443e-06,
      "loss": 3.4025,
      "step": 55380
    },
    {
      "epoch": 0.5634053548177084,
      "grad_norm": 15.087024688720703,
      "learning_rate": 2.008725321246838e-06,
      "loss": 3.0332,
      "step": 55385
    },
    {
      "epoch": 0.5634562174479166,
      "grad_norm": 8.560545921325684,
      "learning_rate": 2.008333243529776e-06,
      "loss": 3.7152,
      "step": 55390
    },
    {
      "epoch": 0.563507080078125,
      "grad_norm": 13.509528160095215,
      "learning_rate": 2.007941178391892e-06,
      "loss": 3.2793,
      "step": 55395
    },
    {
      "epoch": 0.5635579427083334,
      "grad_norm": 19.74806785583496,
      "learning_rate": 2.007549125843217e-06,
      "loss": 3.2927,
      "step": 55400
    },
    {
      "epoch": 0.5636088053385416,
      "grad_norm": 13.209090232849121,
      "learning_rate": 2.0071570858937804e-06,
      "loss": 3.1358,
      "step": 55405
    },
    {
      "epoch": 0.56365966796875,
      "grad_norm": 11.439593315124512,
      "learning_rate": 2.0067650585536124e-06,
      "loss": 3.3403,
      "step": 55410
    },
    {
      "epoch": 0.5637105305989584,
      "grad_norm": 15.847519874572754,
      "learning_rate": 2.0063730438327445e-06,
      "loss": 3.0433,
      "step": 55415
    },
    {
      "epoch": 0.5637613932291666,
      "grad_norm": 16.50002670288086,
      "learning_rate": 2.0059810417412047e-06,
      "loss": 3.5169,
      "step": 55420
    },
    {
      "epoch": 0.563812255859375,
      "grad_norm": 15.249661445617676,
      "learning_rate": 2.0055890522890225e-06,
      "loss": 3.4845,
      "step": 55425
    },
    {
      "epoch": 0.5638631184895834,
      "grad_norm": 13.227401733398438,
      "learning_rate": 2.0051970754862282e-06,
      "loss": 3.2742,
      "step": 55430
    },
    {
      "epoch": 0.5639139811197916,
      "grad_norm": 62.592098236083984,
      "learning_rate": 2.0048051113428487e-06,
      "loss": 4.0365,
      "step": 55435
    },
    {
      "epoch": 0.56396484375,
      "grad_norm": 10.982181549072266,
      "learning_rate": 2.0044131598689137e-06,
      "loss": 3.4353,
      "step": 55440
    },
    {
      "epoch": 0.5640157063802084,
      "grad_norm": 9.214542388916016,
      "learning_rate": 2.0040212210744495e-06,
      "loss": 3.6739,
      "step": 55445
    },
    {
      "epoch": 0.5640665690104166,
      "grad_norm": 13.622384071350098,
      "learning_rate": 2.0036292949694855e-06,
      "loss": 3.0846,
      "step": 55450
    },
    {
      "epoch": 0.564117431640625,
      "grad_norm": 19.547794342041016,
      "learning_rate": 2.0032373815640487e-06,
      "loss": 3.2805,
      "step": 55455
    },
    {
      "epoch": 0.5641682942708334,
      "grad_norm": 9.698598861694336,
      "learning_rate": 2.002845480868165e-06,
      "loss": 3.0059,
      "step": 55460
    },
    {
      "epoch": 0.5642191569010416,
      "grad_norm": 12.346816062927246,
      "learning_rate": 2.002453592891862e-06,
      "loss": 2.9825,
      "step": 55465
    },
    {
      "epoch": 0.56427001953125,
      "grad_norm": 12.297981262207031,
      "learning_rate": 2.002061717645166e-06,
      "loss": 3.9478,
      "step": 55470
    },
    {
      "epoch": 0.5643208821614584,
      "grad_norm": 12.369157791137695,
      "learning_rate": 2.001669855138103e-06,
      "loss": 3.379,
      "step": 55475
    },
    {
      "epoch": 0.5643717447916666,
      "grad_norm": 13.843544006347656,
      "learning_rate": 2.001278005380699e-06,
      "loss": 3.3144,
      "step": 55480
    },
    {
      "epoch": 0.564422607421875,
      "grad_norm": 11.357953071594238,
      "learning_rate": 2.000886168382978e-06,
      "loss": 3.1551,
      "step": 55485
    },
    {
      "epoch": 0.5644734700520834,
      "grad_norm": 11.626884460449219,
      "learning_rate": 2.0004943441549668e-06,
      "loss": 3.3464,
      "step": 55490
    },
    {
      "epoch": 0.5645243326822916,
      "grad_norm": 12.29926872253418,
      "learning_rate": 2.0001025327066893e-06,
      "loss": 3.8347,
      "step": 55495
    },
    {
      "epoch": 0.5645751953125,
      "grad_norm": 7.423537731170654,
      "learning_rate": 1.99971073404817e-06,
      "loss": 3.3202,
      "step": 55500
    },
    {
      "epoch": 0.5646260579427084,
      "grad_norm": 14.192968368530273,
      "learning_rate": 1.9993189481894328e-06,
      "loss": 3.5049,
      "step": 55505
    },
    {
      "epoch": 0.5646769205729166,
      "grad_norm": 11.486786842346191,
      "learning_rate": 1.9989271751405016e-06,
      "loss": 3.5,
      "step": 55510
    },
    {
      "epoch": 0.564727783203125,
      "grad_norm": 11.493895530700684,
      "learning_rate": 1.998535414911401e-06,
      "loss": 3.218,
      "step": 55515
    },
    {
      "epoch": 0.5647786458333334,
      "grad_norm": 14.027342796325684,
      "learning_rate": 1.9981436675121514e-06,
      "loss": 3.3089,
      "step": 55520
    },
    {
      "epoch": 0.5648295084635416,
      "grad_norm": 14.80307674407959,
      "learning_rate": 1.9977519329527777e-06,
      "loss": 3.0861,
      "step": 55525
    },
    {
      "epoch": 0.56488037109375,
      "grad_norm": 10.044238090515137,
      "learning_rate": 1.997360211243302e-06,
      "loss": 3.6657,
      "step": 55530
    },
    {
      "epoch": 0.5649312337239584,
      "grad_norm": 12.431305885314941,
      "learning_rate": 1.9969685023937462e-06,
      "loss": 3.2477,
      "step": 55535
    },
    {
      "epoch": 0.5649820963541666,
      "grad_norm": 8.12856388092041,
      "learning_rate": 1.996576806414131e-06,
      "loss": 3.1989,
      "step": 55540
    },
    {
      "epoch": 0.565032958984375,
      "grad_norm": 10.21527099609375,
      "learning_rate": 1.99618512331448e-06,
      "loss": 3.4905,
      "step": 55545
    },
    {
      "epoch": 0.5650838216145834,
      "grad_norm": 11.607525825500488,
      "learning_rate": 1.995793453104813e-06,
      "loss": 3.9561,
      "step": 55550
    },
    {
      "epoch": 0.5651346842447916,
      "grad_norm": 9.88409423828125,
      "learning_rate": 1.995401795795151e-06,
      "loss": 3.1461,
      "step": 55555
    },
    {
      "epoch": 0.565185546875,
      "grad_norm": 12.363259315490723,
      "learning_rate": 1.995010151395514e-06,
      "loss": 3.8604,
      "step": 55560
    },
    {
      "epoch": 0.5652364095052084,
      "grad_norm": 15.324316024780273,
      "learning_rate": 1.994618519915923e-06,
      "loss": 3.7273,
      "step": 55565
    },
    {
      "epoch": 0.5652872721354166,
      "grad_norm": 10.770191192626953,
      "learning_rate": 1.994226901366398e-06,
      "loss": 3.4032,
      "step": 55570
    },
    {
      "epoch": 0.565338134765625,
      "grad_norm": 13.22590160369873,
      "learning_rate": 1.9938352957569566e-06,
      "loss": 3.3282,
      "step": 55575
    },
    {
      "epoch": 0.5653889973958334,
      "grad_norm": 13.094693183898926,
      "learning_rate": 1.9934437030976194e-06,
      "loss": 3.7382,
      "step": 55580
    },
    {
      "epoch": 0.5654398600260416,
      "grad_norm": 9.559659004211426,
      "learning_rate": 1.993052123398406e-06,
      "loss": 3.203,
      "step": 55585
    },
    {
      "epoch": 0.56549072265625,
      "grad_norm": 11.719879150390625,
      "learning_rate": 1.9926605566693327e-06,
      "loss": 3.6259,
      "step": 55590
    },
    {
      "epoch": 0.5655415852864584,
      "grad_norm": 13.236265182495117,
      "learning_rate": 1.99226900292042e-06,
      "loss": 3.1326,
      "step": 55595
    },
    {
      "epoch": 0.5655924479166666,
      "grad_norm": 10.034022331237793,
      "learning_rate": 1.991877462161683e-06,
      "loss": 3.3176,
      "step": 55600
    },
    {
      "epoch": 0.565643310546875,
      "grad_norm": 12.410802841186523,
      "learning_rate": 1.9914859344031414e-06,
      "loss": 3.0259,
      "step": 55605
    },
    {
      "epoch": 0.5656941731770834,
      "grad_norm": 11.027775764465332,
      "learning_rate": 1.991094419654812e-06,
      "loss": 3.165,
      "step": 55610
    },
    {
      "epoch": 0.5657450358072916,
      "grad_norm": 11.108048439025879,
      "learning_rate": 1.9907029179267105e-06,
      "loss": 3.4322,
      "step": 55615
    },
    {
      "epoch": 0.5657958984375,
      "grad_norm": 14.826831817626953,
      "learning_rate": 1.990311429228855e-06,
      "loss": 3.4251,
      "step": 55620
    },
    {
      "epoch": 0.5658467610677084,
      "grad_norm": 9.523362159729004,
      "learning_rate": 1.9899199535712603e-06,
      "loss": 3.2679,
      "step": 55625
    },
    {
      "epoch": 0.5658976236979166,
      "grad_norm": 13.137377738952637,
      "learning_rate": 1.989528490963943e-06,
      "loss": 3.3582,
      "step": 55630
    },
    {
      "epoch": 0.565948486328125,
      "grad_norm": 14.75580883026123,
      "learning_rate": 1.9891370414169174e-06,
      "loss": 3.4774,
      "step": 55635
    },
    {
      "epoch": 0.5659993489583334,
      "grad_norm": 11.692719459533691,
      "learning_rate": 1.9887456049401997e-06,
      "loss": 3.224,
      "step": 55640
    },
    {
      "epoch": 0.5660502115885416,
      "grad_norm": 10.250654220581055,
      "learning_rate": 1.9883541815438058e-06,
      "loss": 3.1661,
      "step": 55645
    },
    {
      "epoch": 0.56610107421875,
      "grad_norm": 7.867267608642578,
      "learning_rate": 1.987962771237748e-06,
      "loss": 3.4142,
      "step": 55650
    },
    {
      "epoch": 0.5661519368489584,
      "grad_norm": 17.736726760864258,
      "learning_rate": 1.9875713740320407e-06,
      "loss": 3.1155,
      "step": 55655
    },
    {
      "epoch": 0.5662027994791666,
      "grad_norm": 6.9726386070251465,
      "learning_rate": 1.9871799899366994e-06,
      "loss": 3.2602,
      "step": 55660
    },
    {
      "epoch": 0.566253662109375,
      "grad_norm": 8.041133880615234,
      "learning_rate": 1.9867886189617362e-06,
      "loss": 2.9021,
      "step": 55665
    },
    {
      "epoch": 0.5663045247395834,
      "grad_norm": 10.1356201171875,
      "learning_rate": 1.986397261117165e-06,
      "loss": 3.5967,
      "step": 55670
    },
    {
      "epoch": 0.5663553873697916,
      "grad_norm": 10.222390174865723,
      "learning_rate": 1.9860059164129967e-06,
      "loss": 3.4714,
      "step": 55675
    },
    {
      "epoch": 0.56640625,
      "grad_norm": 11.000347137451172,
      "learning_rate": 1.9856145848592464e-06,
      "loss": 3.6656,
      "step": 55680
    },
    {
      "epoch": 0.5664571126302084,
      "grad_norm": 12.125871658325195,
      "learning_rate": 1.985223266465925e-06,
      "loss": 3.3548,
      "step": 55685
    },
    {
      "epoch": 0.5665079752604166,
      "grad_norm": 12.343113899230957,
      "learning_rate": 1.984831961243044e-06,
      "loss": 3.4821,
      "step": 55690
    },
    {
      "epoch": 0.566558837890625,
      "grad_norm": 13.80804443359375,
      "learning_rate": 1.9844406692006147e-06,
      "loss": 3.9151,
      "step": 55695
    },
    {
      "epoch": 0.5666097005208334,
      "grad_norm": 10.661865234375,
      "learning_rate": 1.9840493903486496e-06,
      "loss": 3.3434,
      "step": 55700
    },
    {
      "epoch": 0.5666605631510416,
      "grad_norm": 7.703371524810791,
      "learning_rate": 1.983658124697158e-06,
      "loss": 3.1871,
      "step": 55705
    },
    {
      "epoch": 0.56671142578125,
      "grad_norm": 14.116416931152344,
      "learning_rate": 1.9832668722561506e-06,
      "loss": 3.2558,
      "step": 55710
    },
    {
      "epoch": 0.5667622884114584,
      "grad_norm": 11.84914493560791,
      "learning_rate": 1.9828756330356386e-06,
      "loss": 2.9877,
      "step": 55715
    },
    {
      "epoch": 0.5668131510416666,
      "grad_norm": 11.780055046081543,
      "learning_rate": 1.9824844070456305e-06,
      "loss": 3.5846,
      "step": 55720
    },
    {
      "epoch": 0.566864013671875,
      "grad_norm": 12.862578392028809,
      "learning_rate": 1.9820931942961364e-06,
      "loss": 3.0623,
      "step": 55725
    },
    {
      "epoch": 0.5669148763020834,
      "grad_norm": 13.045098304748535,
      "learning_rate": 1.9817019947971646e-06,
      "loss": 3.0973,
      "step": 55730
    },
    {
      "epoch": 0.5669657389322916,
      "grad_norm": 16.13922882080078,
      "learning_rate": 1.9813108085587248e-06,
      "loss": 3.186,
      "step": 55735
    },
    {
      "epoch": 0.5670166015625,
      "grad_norm": 13.725215911865234,
      "learning_rate": 1.9809196355908255e-06,
      "loss": 3.4581,
      "step": 55740
    },
    {
      "epoch": 0.5670674641927084,
      "grad_norm": 16.00213050842285,
      "learning_rate": 1.9805284759034737e-06,
      "loss": 3.1418,
      "step": 55745
    },
    {
      "epoch": 0.5671183268229166,
      "grad_norm": 10.054238319396973,
      "learning_rate": 1.9801373295066774e-06,
      "loss": 3.1284,
      "step": 55750
    },
    {
      "epoch": 0.567169189453125,
      "grad_norm": 14.31942367553711,
      "learning_rate": 1.9797461964104452e-06,
      "loss": 3.4611,
      "step": 55755
    },
    {
      "epoch": 0.5672200520833334,
      "grad_norm": 8.36902141571045,
      "learning_rate": 1.9793550766247827e-06,
      "loss": 3.2506,
      "step": 55760
    },
    {
      "epoch": 0.5672709147135416,
      "grad_norm": 11.869385719299316,
      "learning_rate": 1.978963970159698e-06,
      "loss": 3.1739,
      "step": 55765
    },
    {
      "epoch": 0.56732177734375,
      "grad_norm": 11.562711715698242,
      "learning_rate": 1.9785728770251948e-06,
      "loss": 3.2535,
      "step": 55770
    },
    {
      "epoch": 0.5673726399739584,
      "grad_norm": 15.536983489990234,
      "learning_rate": 1.9781817972312822e-06,
      "loss": 3.5746,
      "step": 55775
    },
    {
      "epoch": 0.5674235026041666,
      "grad_norm": 14.286041259765625,
      "learning_rate": 1.9777907307879647e-06,
      "loss": 3.2235,
      "step": 55780
    },
    {
      "epoch": 0.567474365234375,
      "grad_norm": 11.456912994384766,
      "learning_rate": 1.977399677705247e-06,
      "loss": 3.449,
      "step": 55785
    },
    {
      "epoch": 0.5675252278645834,
      "grad_norm": 15.62401294708252,
      "learning_rate": 1.977008637993135e-06,
      "loss": 3.2953,
      "step": 55790
    },
    {
      "epoch": 0.5675760904947916,
      "grad_norm": 7.052567958831787,
      "learning_rate": 1.976617611661633e-06,
      "loss": 3.1743,
      "step": 55795
    },
    {
      "epoch": 0.567626953125,
      "grad_norm": 8.342056274414062,
      "learning_rate": 1.976226598720746e-06,
      "loss": 3.3018,
      "step": 55800
    },
    {
      "epoch": 0.5676778157552084,
      "grad_norm": 16.19434928894043,
      "learning_rate": 1.9758355991804763e-06,
      "loss": 3.4816,
      "step": 55805
    },
    {
      "epoch": 0.5677286783854166,
      "grad_norm": 8.652542114257812,
      "learning_rate": 1.9754446130508285e-06,
      "loss": 3.4816,
      "step": 55810
    },
    {
      "epoch": 0.567779541015625,
      "grad_norm": 16.950336456298828,
      "learning_rate": 1.975053640341807e-06,
      "loss": 3.2507,
      "step": 55815
    },
    {
      "epoch": 0.5678304036458334,
      "grad_norm": 16.509111404418945,
      "learning_rate": 1.9746626810634127e-06,
      "loss": 3.8118,
      "step": 55820
    },
    {
      "epoch": 0.5678812662760416,
      "grad_norm": 11.515554428100586,
      "learning_rate": 1.974271735225649e-06,
      "loss": 3.4826,
      "step": 55825
    },
    {
      "epoch": 0.56793212890625,
      "grad_norm": 15.886150360107422,
      "learning_rate": 1.9738808028385197e-06,
      "loss": 3.0344,
      "step": 55830
    },
    {
      "epoch": 0.5679829915364584,
      "grad_norm": 9.29141902923584,
      "learning_rate": 1.9734898839120246e-06,
      "loss": 3.3004,
      "step": 55835
    },
    {
      "epoch": 0.5680338541666666,
      "grad_norm": 13.30876636505127,
      "learning_rate": 1.9730989784561663e-06,
      "loss": 3.1008,
      "step": 55840
    },
    {
      "epoch": 0.568084716796875,
      "grad_norm": 13.878016471862793,
      "learning_rate": 1.972708086480945e-06,
      "loss": 3.4856,
      "step": 55845
    },
    {
      "epoch": 0.5681355794270834,
      "grad_norm": 12.876764297485352,
      "learning_rate": 1.9723172079963627e-06,
      "loss": 3.0248,
      "step": 55850
    },
    {
      "epoch": 0.5681864420572916,
      "grad_norm": 10.323982238769531,
      "learning_rate": 1.97192634301242e-06,
      "loss": 3.6409,
      "step": 55855
    },
    {
      "epoch": 0.5682373046875,
      "grad_norm": 15.221038818359375,
      "learning_rate": 1.9715354915391163e-06,
      "loss": 3.2885,
      "step": 55860
    },
    {
      "epoch": 0.5682881673177084,
      "grad_norm": 11.07483959197998,
      "learning_rate": 1.971144653586451e-06,
      "loss": 3.5815,
      "step": 55865
    },
    {
      "epoch": 0.5683390299479166,
      "grad_norm": 14.276740074157715,
      "learning_rate": 1.970753829164426e-06,
      "loss": 2.9981,
      "step": 55870
    },
    {
      "epoch": 0.568389892578125,
      "grad_norm": 12.67236328125,
      "learning_rate": 1.970363018283038e-06,
      "loss": 3.3977,
      "step": 55875
    },
    {
      "epoch": 0.5684407552083334,
      "grad_norm": 9.223763465881348,
      "learning_rate": 1.969972220952287e-06,
      "loss": 3.5081,
      "step": 55880
    },
    {
      "epoch": 0.5684916178385416,
      "grad_norm": 12.528619766235352,
      "learning_rate": 1.96958143718217e-06,
      "loss": 3.328,
      "step": 55885
    },
    {
      "epoch": 0.56854248046875,
      "grad_norm": 13.376630783081055,
      "learning_rate": 1.9691906669826873e-06,
      "loss": 3.7906,
      "step": 55890
    },
    {
      "epoch": 0.5685933430989584,
      "grad_norm": 13.842061042785645,
      "learning_rate": 1.9687999103638354e-06,
      "loss": 3.4686,
      "step": 55895
    },
    {
      "epoch": 0.5686442057291666,
      "grad_norm": 12.943464279174805,
      "learning_rate": 1.9684091673356114e-06,
      "loss": 3.1778,
      "step": 55900
    },
    {
      "epoch": 0.568695068359375,
      "grad_norm": 12.769246101379395,
      "learning_rate": 1.968018437908013e-06,
      "loss": 3.3822,
      "step": 55905
    },
    {
      "epoch": 0.5687459309895834,
      "grad_norm": 8.439935684204102,
      "learning_rate": 1.9676277220910374e-06,
      "loss": 3.3736,
      "step": 55910
    },
    {
      "epoch": 0.5687967936197916,
      "grad_norm": 8.644288063049316,
      "learning_rate": 1.9672370198946795e-06,
      "loss": 3.2799,
      "step": 55915
    },
    {
      "epoch": 0.56884765625,
      "grad_norm": 9.586736679077148,
      "learning_rate": 1.966846331328936e-06,
      "loss": 2.8663,
      "step": 55920
    },
    {
      "epoch": 0.5688985188802084,
      "grad_norm": 14.994538307189941,
      "learning_rate": 1.966455656403804e-06,
      "loss": 3.4941,
      "step": 55925
    },
    {
      "epoch": 0.5689493815104166,
      "grad_norm": 10.81418514251709,
      "learning_rate": 1.9660649951292766e-06,
      "loss": 3.6868,
      "step": 55930
    },
    {
      "epoch": 0.569000244140625,
      "grad_norm": 13.596345901489258,
      "learning_rate": 1.9656743475153504e-06,
      "loss": 4.1722,
      "step": 55935
    },
    {
      "epoch": 0.5690511067708334,
      "grad_norm": 14.05886173248291,
      "learning_rate": 1.965283713572018e-06,
      "loss": 3.0352,
      "step": 55940
    },
    {
      "epoch": 0.5691019694010416,
      "grad_norm": 135.2580108642578,
      "learning_rate": 1.964893093309276e-06,
      "loss": 4.1592,
      "step": 55945
    },
    {
      "epoch": 0.56915283203125,
      "grad_norm": 11.427495002746582,
      "learning_rate": 1.9645024867371176e-06,
      "loss": 3.0614,
      "step": 55950
    },
    {
      "epoch": 0.5692036946614584,
      "grad_norm": 8.255717277526855,
      "learning_rate": 1.9641118938655366e-06,
      "loss": 3.7999,
      "step": 55955
    },
    {
      "epoch": 0.5692545572916666,
      "grad_norm": 12.217011451721191,
      "learning_rate": 1.9637213147045243e-06,
      "loss": 3.2522,
      "step": 55960
    },
    {
      "epoch": 0.569305419921875,
      "grad_norm": 14.885222434997559,
      "learning_rate": 1.963330749264076e-06,
      "loss": 3.0997,
      "step": 55965
    },
    {
      "epoch": 0.5693562825520834,
      "grad_norm": 16.567792892456055,
      "learning_rate": 1.9629401975541835e-06,
      "loss": 3.2213,
      "step": 55970
    },
    {
      "epoch": 0.5694071451822916,
      "grad_norm": 12.30616569519043,
      "learning_rate": 1.9625496595848382e-06,
      "loss": 3.568,
      "step": 55975
    },
    {
      "epoch": 0.5694580078125,
      "grad_norm": 11.718070983886719,
      "learning_rate": 1.9621591353660323e-06,
      "loss": 3.0996,
      "step": 55980
    },
    {
      "epoch": 0.5695088704427084,
      "grad_norm": 10.797274589538574,
      "learning_rate": 1.961768624907758e-06,
      "loss": 3.0879,
      "step": 55985
    },
    {
      "epoch": 0.5695597330729166,
      "grad_norm": 9.172492980957031,
      "learning_rate": 1.961378128220006e-06,
      "loss": 3.91,
      "step": 55990
    },
    {
      "epoch": 0.569610595703125,
      "grad_norm": 10.033970832824707,
      "learning_rate": 1.960987645312767e-06,
      "loss": 3.1516,
      "step": 55995
    },
    {
      "epoch": 0.5696614583333334,
      "grad_norm": 14.512290954589844,
      "learning_rate": 1.9605971761960304e-06,
      "loss": 3.1986,
      "step": 56000
    },
    {
      "epoch": 0.5697123209635416,
      "grad_norm": 19.83005714416504,
      "learning_rate": 1.9602067208797874e-06,
      "loss": 3.3876,
      "step": 56005
    },
    {
      "epoch": 0.56976318359375,
      "grad_norm": 10.733776092529297,
      "learning_rate": 1.9598162793740283e-06,
      "loss": 3.4806,
      "step": 56010
    },
    {
      "epoch": 0.5698140462239584,
      "grad_norm": 14.168524742126465,
      "learning_rate": 1.9594258516887404e-06,
      "loss": 3.7829,
      "step": 56015
    },
    {
      "epoch": 0.5698649088541666,
      "grad_norm": 13.03543758392334,
      "learning_rate": 1.959035437833915e-06,
      "loss": 3.2386,
      "step": 56020
    },
    {
      "epoch": 0.569915771484375,
      "grad_norm": 14.626461029052734,
      "learning_rate": 1.95864503781954e-06,
      "loss": 3.1362,
      "step": 56025
    },
    {
      "epoch": 0.5699666341145834,
      "grad_norm": 7.632985591888428,
      "learning_rate": 1.9582546516556025e-06,
      "loss": 3.2632,
      "step": 56030
    },
    {
      "epoch": 0.5700174967447916,
      "grad_norm": 18.721630096435547,
      "learning_rate": 1.957864279352091e-06,
      "loss": 3.1491,
      "step": 56035
    },
    {
      "epoch": 0.570068359375,
      "grad_norm": 12.163301467895508,
      "learning_rate": 1.9574739209189945e-06,
      "loss": 3.4391,
      "step": 56040
    },
    {
      "epoch": 0.5701192220052084,
      "grad_norm": 9.6724214553833,
      "learning_rate": 1.9570835763662986e-06,
      "loss": 3.2104,
      "step": 56045
    },
    {
      "epoch": 0.5701700846354166,
      "grad_norm": 12.953228950500488,
      "learning_rate": 1.9566932457039913e-06,
      "loss": 3.1921,
      "step": 56050
    },
    {
      "epoch": 0.570220947265625,
      "grad_norm": 17.458709716796875,
      "learning_rate": 1.956302928942057e-06,
      "loss": 3.563,
      "step": 56055
    },
    {
      "epoch": 0.5702718098958334,
      "grad_norm": 10.399765968322754,
      "learning_rate": 1.9559126260904848e-06,
      "loss": 3.6487,
      "step": 56060
    },
    {
      "epoch": 0.5703226725260416,
      "grad_norm": 17.193531036376953,
      "learning_rate": 1.955522337159259e-06,
      "loss": 3.3863,
      "step": 56065
    },
    {
      "epoch": 0.57037353515625,
      "grad_norm": 7.868402481079102,
      "learning_rate": 1.955132062158365e-06,
      "loss": 3.3566,
      "step": 56070
    },
    {
      "epoch": 0.5704243977864584,
      "grad_norm": 14.431535720825195,
      "learning_rate": 1.9547418010977874e-06,
      "loss": 3.1291,
      "step": 56075
    },
    {
      "epoch": 0.5704752604166666,
      "grad_norm": 10.154860496520996,
      "learning_rate": 1.9543515539875123e-06,
      "loss": 3.4842,
      "step": 56080
    },
    {
      "epoch": 0.570526123046875,
      "grad_norm": 12.6719970703125,
      "learning_rate": 1.9539613208375234e-06,
      "loss": 3.5905,
      "step": 56085
    },
    {
      "epoch": 0.5705769856770834,
      "grad_norm": 9.200603485107422,
      "learning_rate": 1.953571101657805e-06,
      "loss": 3.9783,
      "step": 56090
    },
    {
      "epoch": 0.5706278483072916,
      "grad_norm": 10.441458702087402,
      "learning_rate": 1.9531808964583388e-06,
      "loss": 3.2192,
      "step": 56095
    },
    {
      "epoch": 0.5706787109375,
      "grad_norm": 10.30545425415039,
      "learning_rate": 1.952790705249111e-06,
      "loss": 3.6149,
      "step": 56100
    },
    {
      "epoch": 0.5707295735677084,
      "grad_norm": 14.64566421508789,
      "learning_rate": 1.952400528040103e-06,
      "loss": 3.5059,
      "step": 56105
    },
    {
      "epoch": 0.5707804361979166,
      "grad_norm": 10.314763069152832,
      "learning_rate": 1.9520103648412973e-06,
      "loss": 3.3058,
      "step": 56110
    },
    {
      "epoch": 0.570831298828125,
      "grad_norm": 12.125476837158203,
      "learning_rate": 1.9516202156626777e-06,
      "loss": 3.2173,
      "step": 56115
    },
    {
      "epoch": 0.5708821614583334,
      "grad_norm": 15.160707473754883,
      "learning_rate": 1.9512300805142243e-06,
      "loss": 3.1716,
      "step": 56120
    },
    {
      "epoch": 0.5709330240885416,
      "grad_norm": 7.939380168914795,
      "learning_rate": 1.9508399594059196e-06,
      "loss": 3.5609,
      "step": 56125
    },
    {
      "epoch": 0.57098388671875,
      "grad_norm": 12.273107528686523,
      "learning_rate": 1.9504498523477434e-06,
      "loss": 3.3506,
      "step": 56130
    },
    {
      "epoch": 0.5710347493489584,
      "grad_norm": 14.89732837677002,
      "learning_rate": 1.9500597593496777e-06,
      "loss": 3.7025,
      "step": 56135
    },
    {
      "epoch": 0.5710856119791666,
      "grad_norm": 11.694847106933594,
      "learning_rate": 1.9496696804217037e-06,
      "loss": 3.2454,
      "step": 56140
    },
    {
      "epoch": 0.571136474609375,
      "grad_norm": 8.561016082763672,
      "learning_rate": 1.9492796155737996e-06,
      "loss": 3.0889,
      "step": 56145
    },
    {
      "epoch": 0.5711873372395834,
      "grad_norm": 15.127306938171387,
      "learning_rate": 1.9488895648159455e-06,
      "loss": 3.1905,
      "step": 56150
    },
    {
      "epoch": 0.5712381998697916,
      "grad_norm": 9.506645202636719,
      "learning_rate": 1.9484995281581225e-06,
      "loss": 3.2974,
      "step": 56155
    },
    {
      "epoch": 0.5712890625,
      "grad_norm": 11.450814247131348,
      "learning_rate": 1.9481095056103077e-06,
      "loss": 3.4543,
      "step": 56160
    },
    {
      "epoch": 0.5713399251302084,
      "grad_norm": 15.476284980773926,
      "learning_rate": 1.9477194971824813e-06,
      "loss": 3.536,
      "step": 56165
    },
    {
      "epoch": 0.5713907877604166,
      "grad_norm": 12.819311141967773,
      "learning_rate": 1.9473295028846192e-06,
      "loss": 3.0641,
      "step": 56170
    },
    {
      "epoch": 0.571441650390625,
      "grad_norm": 13.76785945892334,
      "learning_rate": 1.9469395227267013e-06,
      "loss": 3.399,
      "step": 56175
    },
    {
      "epoch": 0.5714925130208334,
      "grad_norm": 15.641027450561523,
      "learning_rate": 1.9465495567187054e-06,
      "loss": 3.4786,
      "step": 56180
    },
    {
      "epoch": 0.5715433756510416,
      "grad_norm": 12.041227340698242,
      "learning_rate": 1.9461596048706072e-06,
      "loss": 3.3719,
      "step": 56185
    },
    {
      "epoch": 0.57159423828125,
      "grad_norm": 10.25402545928955,
      "learning_rate": 1.945769667192384e-06,
      "loss": 3.2183,
      "step": 56190
    },
    {
      "epoch": 0.5716451009114584,
      "grad_norm": 10.055646896362305,
      "learning_rate": 1.9453797436940126e-06,
      "loss": 3.6629,
      "step": 56195
    },
    {
      "epoch": 0.5716959635416666,
      "grad_norm": 14.09347915649414,
      "learning_rate": 1.9449898343854696e-06,
      "loss": 3.3642,
      "step": 56200
    },
    {
      "epoch": 0.571746826171875,
      "grad_norm": 10.3724365234375,
      "learning_rate": 1.944599939276729e-06,
      "loss": 3.6581,
      "step": 56205
    },
    {
      "epoch": 0.5717976888020834,
      "grad_norm": 10.909431457519531,
      "learning_rate": 1.9442100583777685e-06,
      "loss": 3.2881,
      "step": 56210
    },
    {
      "epoch": 0.5718485514322916,
      "grad_norm": 14.096423149108887,
      "learning_rate": 1.943820191698561e-06,
      "loss": 3.788,
      "step": 56215
    },
    {
      "epoch": 0.5718994140625,
      "grad_norm": 16.65086555480957,
      "learning_rate": 1.943430339249083e-06,
      "loss": 3.8615,
      "step": 56220
    },
    {
      "epoch": 0.5719502766927084,
      "grad_norm": 11.530363082885742,
      "learning_rate": 1.943040501039307e-06,
      "loss": 3.282,
      "step": 56225
    },
    {
      "epoch": 0.5720011393229166,
      "grad_norm": 7.3054680824279785,
      "learning_rate": 1.9426506770792083e-06,
      "loss": 2.9748,
      "step": 56230
    },
    {
      "epoch": 0.572052001953125,
      "grad_norm": 13.849099159240723,
      "learning_rate": 1.94226086737876e-06,
      "loss": 2.9996,
      "step": 56235
    },
    {
      "epoch": 0.5721028645833334,
      "grad_norm": 17.799570083618164,
      "learning_rate": 1.941871071947935e-06,
      "loss": 3.4202,
      "step": 56240
    },
    {
      "epoch": 0.5721537272135416,
      "grad_norm": 15.279867172241211,
      "learning_rate": 1.941481290796706e-06,
      "loss": 3.335,
      "step": 56245
    },
    {
      "epoch": 0.57220458984375,
      "grad_norm": 17.308347702026367,
      "learning_rate": 1.941091523935046e-06,
      "loss": 3.4459,
      "step": 56250
    },
    {
      "epoch": 0.5722554524739584,
      "grad_norm": 7.215347766876221,
      "learning_rate": 1.9407017713729275e-06,
      "loss": 3.1686,
      "step": 56255
    },
    {
      "epoch": 0.5723063151041666,
      "grad_norm": 13.225153923034668,
      "learning_rate": 1.940312033120321e-06,
      "loss": 3.8314,
      "step": 56260
    },
    {
      "epoch": 0.572357177734375,
      "grad_norm": 14.776530265808105,
      "learning_rate": 1.939922309187198e-06,
      "loss": 3.7128,
      "step": 56265
    },
    {
      "epoch": 0.5724080403645834,
      "grad_norm": 9.53740119934082,
      "learning_rate": 1.939532599583531e-06,
      "loss": 3.0098,
      "step": 56270
    },
    {
      "epoch": 0.5724589029947916,
      "grad_norm": 10.148475646972656,
      "learning_rate": 1.939142904319289e-06,
      "loss": 3.2187,
      "step": 56275
    },
    {
      "epoch": 0.572509765625,
      "grad_norm": 14.827524185180664,
      "learning_rate": 1.9387532234044436e-06,
      "loss": 3.4579,
      "step": 56280
    },
    {
      "epoch": 0.5725606282552084,
      "grad_norm": 12.451027870178223,
      "learning_rate": 1.9383635568489624e-06,
      "loss": 3.3303,
      "step": 56285
    },
    {
      "epoch": 0.5726114908854166,
      "grad_norm": 11.020612716674805,
      "learning_rate": 1.937973904662818e-06,
      "loss": 3.4904,
      "step": 56290
    },
    {
      "epoch": 0.572662353515625,
      "grad_norm": 15.075807571411133,
      "learning_rate": 1.9375842668559774e-06,
      "loss": 3.0847,
      "step": 56295
    },
    {
      "epoch": 0.5727132161458334,
      "grad_norm": 14.280423164367676,
      "learning_rate": 1.9371946434384103e-06,
      "loss": 3.183,
      "step": 56300
    },
    {
      "epoch": 0.5727640787760416,
      "grad_norm": 9.974885940551758,
      "learning_rate": 1.9368050344200835e-06,
      "loss": 3.4728,
      "step": 56305
    },
    {
      "epoch": 0.57281494140625,
      "grad_norm": 11.663572311401367,
      "learning_rate": 1.9364154398109684e-06,
      "loss": 3.5697,
      "step": 56310
    },
    {
      "epoch": 0.5728658040364584,
      "grad_norm": 7.344516277313232,
      "learning_rate": 1.93602585962103e-06,
      "loss": 3.3228,
      "step": 56315
    },
    {
      "epoch": 0.5729166666666666,
      "grad_norm": 14.893961906433105,
      "learning_rate": 1.9356362938602352e-06,
      "loss": 3.9728,
      "step": 56320
    },
    {
      "epoch": 0.572967529296875,
      "grad_norm": 8.951798439025879,
      "learning_rate": 1.9352467425385534e-06,
      "loss": 3.2949,
      "step": 56325
    },
    {
      "epoch": 0.5730183919270834,
      "grad_norm": 8.843398094177246,
      "learning_rate": 1.934857205665949e-06,
      "loss": 3.1706,
      "step": 56330
    },
    {
      "epoch": 0.5730692545572916,
      "grad_norm": 8.021584510803223,
      "learning_rate": 1.93446768325239e-06,
      "loss": 3.1792,
      "step": 56335
    },
    {
      "epoch": 0.5731201171875,
      "grad_norm": 15.649279594421387,
      "learning_rate": 1.93407817530784e-06,
      "loss": 3.6829,
      "step": 56340
    },
    {
      "epoch": 0.5731709798177084,
      "grad_norm": 10.430551528930664,
      "learning_rate": 1.933688681842267e-06,
      "loss": 3.3624,
      "step": 56345
    },
    {
      "epoch": 0.5732218424479166,
      "grad_norm": 10.106148719787598,
      "learning_rate": 1.9332992028656343e-06,
      "loss": 3.4437,
      "step": 56350
    },
    {
      "epoch": 0.573272705078125,
      "grad_norm": 15.482913970947266,
      "learning_rate": 1.9329097383879073e-06,
      "loss": 3.6239,
      "step": 56355
    },
    {
      "epoch": 0.5733235677083334,
      "grad_norm": 8.75622272491455,
      "learning_rate": 1.9325202884190497e-06,
      "loss": 2.8706,
      "step": 56360
    },
    {
      "epoch": 0.5733744303385416,
      "grad_norm": 15.633439064025879,
      "learning_rate": 1.932130852969027e-06,
      "loss": 3.7574,
      "step": 56365
    },
    {
      "epoch": 0.57342529296875,
      "grad_norm": 10.85609245300293,
      "learning_rate": 1.931741432047801e-06,
      "loss": 3.5615,
      "step": 56370
    },
    {
      "epoch": 0.5734761555989584,
      "grad_norm": 8.461934089660645,
      "learning_rate": 1.9313520256653365e-06,
      "loss": 3.0443,
      "step": 56375
    },
    {
      "epoch": 0.5735270182291666,
      "grad_norm": 15.088605880737305,
      "learning_rate": 1.9309626338315946e-06,
      "loss": 3.663,
      "step": 56380
    },
    {
      "epoch": 0.573577880859375,
      "grad_norm": 14.905712127685547,
      "learning_rate": 1.9305732565565394e-06,
      "loss": 3.1529,
      "step": 56385
    },
    {
      "epoch": 0.5736287434895834,
      "grad_norm": 12.533961296081543,
      "learning_rate": 1.9301838938501327e-06,
      "loss": 3.118,
      "step": 56390
    },
    {
      "epoch": 0.5736796061197916,
      "grad_norm": 15.248723983764648,
      "learning_rate": 1.929794545722336e-06,
      "loss": 3.8423,
      "step": 56395
    },
    {
      "epoch": 0.57373046875,
      "grad_norm": 14.34217357635498,
      "learning_rate": 1.92940521218311e-06,
      "loss": 3.1248,
      "step": 56400
    },
    {
      "epoch": 0.5737813313802084,
      "grad_norm": 15.595789909362793,
      "learning_rate": 1.929015893242416e-06,
      "loss": 2.9282,
      "step": 56405
    },
    {
      "epoch": 0.5738321940104166,
      "grad_norm": 15.70177936553955,
      "learning_rate": 1.928626588910217e-06,
      "loss": 3.1065,
      "step": 56410
    },
    {
      "epoch": 0.573883056640625,
      "grad_norm": 8.46838092803955,
      "learning_rate": 1.928237299196469e-06,
      "loss": 3.2156,
      "step": 56415
    },
    {
      "epoch": 0.5739339192708334,
      "grad_norm": 12.688814163208008,
      "learning_rate": 1.927848024111136e-06,
      "loss": 3.8477,
      "step": 56420
    },
    {
      "epoch": 0.5739847819010416,
      "grad_norm": 13.845072746276855,
      "learning_rate": 1.9274587636641753e-06,
      "loss": 3.2232,
      "step": 56425
    },
    {
      "epoch": 0.57403564453125,
      "grad_norm": 10.488356590270996,
      "learning_rate": 1.927069517865546e-06,
      "loss": 3.4556,
      "step": 56430
    },
    {
      "epoch": 0.5740865071614584,
      "grad_norm": 16.517683029174805,
      "learning_rate": 1.926680286725207e-06,
      "loss": 3.3017,
      "step": 56435
    },
    {
      "epoch": 0.5741373697916666,
      "grad_norm": 9.872268676757812,
      "learning_rate": 1.926291070253118e-06,
      "loss": 3.2667,
      "step": 56440
    },
    {
      "epoch": 0.574188232421875,
      "grad_norm": 16.947826385498047,
      "learning_rate": 1.925901868459235e-06,
      "loss": 3.4922,
      "step": 56445
    },
    {
      "epoch": 0.5742390950520834,
      "grad_norm": 10.253779411315918,
      "learning_rate": 1.925512681353518e-06,
      "loss": 3.1615,
      "step": 56450
    },
    {
      "epoch": 0.5742899576822916,
      "grad_norm": 11.655627250671387,
      "learning_rate": 1.9251235089459212e-06,
      "loss": 3.8761,
      "step": 56455
    },
    {
      "epoch": 0.5743408203125,
      "grad_norm": 12.3087158203125,
      "learning_rate": 1.9247343512464037e-06,
      "loss": 3.5111,
      "step": 56460
    },
    {
      "epoch": 0.5743916829427084,
      "grad_norm": 8.526979446411133,
      "learning_rate": 1.9243452082649223e-06,
      "loss": 3.3005,
      "step": 56465
    },
    {
      "epoch": 0.5744425455729166,
      "grad_norm": 14.355345726013184,
      "learning_rate": 1.9239560800114316e-06,
      "loss": 2.8323,
      "step": 56470
    },
    {
      "epoch": 0.574493408203125,
      "grad_norm": 14.259971618652344,
      "learning_rate": 1.9235669664958873e-06,
      "loss": 3.7159,
      "step": 56475
    },
    {
      "epoch": 0.5745442708333334,
      "grad_norm": 9.394563674926758,
      "learning_rate": 1.923177867728247e-06,
      "loss": 3.1292,
      "step": 56480
    },
    {
      "epoch": 0.5745951334635416,
      "grad_norm": 14.726558685302734,
      "learning_rate": 1.922788783718463e-06,
      "loss": 3.0501,
      "step": 56485
    },
    {
      "epoch": 0.57464599609375,
      "grad_norm": 13.66686725616455,
      "learning_rate": 1.922399714476493e-06,
      "loss": 3.2185,
      "step": 56490
    },
    {
      "epoch": 0.5746968587239584,
      "grad_norm": 12.804473876953125,
      "learning_rate": 1.922010660012287e-06,
      "loss": 3.6264,
      "step": 56495
    },
    {
      "epoch": 0.5747477213541666,
      "grad_norm": 7.820730686187744,
      "learning_rate": 1.9216216203358024e-06,
      "loss": 3.4058,
      "step": 56500
    },
    {
      "epoch": 0.574798583984375,
      "grad_norm": 11.180335998535156,
      "learning_rate": 1.921232595456992e-06,
      "loss": 2.8898,
      "step": 56505
    },
    {
      "epoch": 0.5748494466145834,
      "grad_norm": 14.343398094177246,
      "learning_rate": 1.9208435853858078e-06,
      "loss": 3.0695,
      "step": 56510
    },
    {
      "epoch": 0.5749003092447916,
      "grad_norm": 16.27067756652832,
      "learning_rate": 1.920454590132203e-06,
      "loss": 3.4296,
      "step": 56515
    },
    {
      "epoch": 0.574951171875,
      "grad_norm": 14.813397407531738,
      "learning_rate": 1.920065609706131e-06,
      "loss": 3.0937,
      "step": 56520
    },
    {
      "epoch": 0.5750020345052084,
      "grad_norm": 9.876043319702148,
      "learning_rate": 1.9196766441175423e-06,
      "loss": 3.4687,
      "step": 56525
    },
    {
      "epoch": 0.5750528971354166,
      "grad_norm": 10.726964950561523,
      "learning_rate": 1.919287693376389e-06,
      "loss": 3.1954,
      "step": 56530
    },
    {
      "epoch": 0.575103759765625,
      "grad_norm": 11.442978858947754,
      "learning_rate": 1.9188987574926234e-06,
      "loss": 3.281,
      "step": 56535
    },
    {
      "epoch": 0.5751546223958334,
      "grad_norm": 13.518041610717773,
      "learning_rate": 1.9185098364761943e-06,
      "loss": 3.3092,
      "step": 56540
    },
    {
      "epoch": 0.5752054850260416,
      "grad_norm": 11.82832145690918,
      "learning_rate": 1.918120930337054e-06,
      "loss": 3.6624,
      "step": 56545
    },
    {
      "epoch": 0.57525634765625,
      "grad_norm": 15.283867835998535,
      "learning_rate": 1.91773203908515e-06,
      "loss": 3.3876,
      "step": 56550
    },
    {
      "epoch": 0.5753072102864584,
      "grad_norm": 10.985995292663574,
      "learning_rate": 1.917343162730436e-06,
      "loss": 3.4584,
      "step": 56555
    },
    {
      "epoch": 0.5753580729166666,
      "grad_norm": 12.337750434875488,
      "learning_rate": 1.916954301282858e-06,
      "loss": 3.4907,
      "step": 56560
    },
    {
      "epoch": 0.575408935546875,
      "grad_norm": 13.062437057495117,
      "learning_rate": 1.9165654547523667e-06,
      "loss": 3.5146,
      "step": 56565
    },
    {
      "epoch": 0.5754597981770834,
      "grad_norm": 12.77077579498291,
      "learning_rate": 1.916176623148909e-06,
      "loss": 3.7536,
      "step": 56570
    },
    {
      "epoch": 0.5755106608072916,
      "grad_norm": 11.67747974395752,
      "learning_rate": 1.915787806482434e-06,
      "loss": 3.3842,
      "step": 56575
    },
    {
      "epoch": 0.5755615234375,
      "grad_norm": 12.839545249938965,
      "learning_rate": 1.9153990047628905e-06,
      "loss": 3.2524,
      "step": 56580
    },
    {
      "epoch": 0.5756123860677084,
      "grad_norm": 8.737242698669434,
      "learning_rate": 1.9150102180002243e-06,
      "loss": 3.2188,
      "step": 56585
    },
    {
      "epoch": 0.5756632486979166,
      "grad_norm": 10.488981246948242,
      "learning_rate": 1.9146214462043826e-06,
      "loss": 3.7352,
      "step": 56590
    },
    {
      "epoch": 0.575714111328125,
      "grad_norm": 10.820968627929688,
      "learning_rate": 1.914232689385313e-06,
      "loss": 3.4999,
      "step": 56595
    },
    {
      "epoch": 0.5757649739583334,
      "grad_norm": 12.983179092407227,
      "learning_rate": 1.9138439475529614e-06,
      "loss": 3.1547,
      "step": 56600
    },
    {
      "epoch": 0.5758158365885416,
      "grad_norm": 8.708410263061523,
      "learning_rate": 1.9134552207172723e-06,
      "loss": 3.0745,
      "step": 56605
    },
    {
      "epoch": 0.57586669921875,
      "grad_norm": 11.97068977355957,
      "learning_rate": 1.913066508888194e-06,
      "loss": 3.314,
      "step": 56610
    },
    {
      "epoch": 0.5759175618489584,
      "grad_norm": 11.596902847290039,
      "learning_rate": 1.9126778120756687e-06,
      "loss": 3.3621,
      "step": 56615
    },
    {
      "epoch": 0.5759684244791666,
      "grad_norm": 14.041169166564941,
      "learning_rate": 1.9122891302896434e-06,
      "loss": 3.4463,
      "step": 56620
    },
    {
      "epoch": 0.576019287109375,
      "grad_norm": 10.68602466583252,
      "learning_rate": 1.9119004635400605e-06,
      "loss": 3.302,
      "step": 56625
    },
    {
      "epoch": 0.5760701497395834,
      "grad_norm": 13.853493690490723,
      "learning_rate": 1.911511811836865e-06,
      "loss": 3.4612,
      "step": 56630
    },
    {
      "epoch": 0.5761210123697916,
      "grad_norm": 16.029159545898438,
      "learning_rate": 1.911123175190001e-06,
      "loss": 3.5562,
      "step": 56635
    },
    {
      "epoch": 0.576171875,
      "grad_norm": 8.260101318359375,
      "learning_rate": 1.9107345536094106e-06,
      "loss": 3.0825,
      "step": 56640
    },
    {
      "epoch": 0.5762227376302084,
      "grad_norm": 9.413946151733398,
      "learning_rate": 1.9103459471050364e-06,
      "loss": 3.0764,
      "step": 56645
    },
    {
      "epoch": 0.5762736002604166,
      "grad_norm": 10.229106903076172,
      "learning_rate": 1.909957355686822e-06,
      "loss": 3.8249,
      "step": 56650
    },
    {
      "epoch": 0.576324462890625,
      "grad_norm": 16.245317459106445,
      "learning_rate": 1.9095687793647085e-06,
      "loss": 3.0807,
      "step": 56655
    },
    {
      "epoch": 0.5763753255208334,
      "grad_norm": 16.102489471435547,
      "learning_rate": 1.9091802181486385e-06,
      "loss": 3.6931,
      "step": 56660
    },
    {
      "epoch": 0.5764261881510416,
      "grad_norm": 10.535594940185547,
      "learning_rate": 1.908791672048551e-06,
      "loss": 2.9715,
      "step": 56665
    },
    {
      "epoch": 0.57647705078125,
      "grad_norm": 9.063885688781738,
      "learning_rate": 1.90840314107439e-06,
      "loss": 3.2165,
      "step": 56670
    },
    {
      "epoch": 0.5765279134114584,
      "grad_norm": 14.134736061096191,
      "learning_rate": 1.908014625236094e-06,
      "loss": 3.3692,
      "step": 56675
    },
    {
      "epoch": 0.5765787760416666,
      "grad_norm": 11.198437690734863,
      "learning_rate": 1.9076261245436034e-06,
      "loss": 3.6974,
      "step": 56680
    },
    {
      "epoch": 0.576629638671875,
      "grad_norm": 11.785255432128906,
      "learning_rate": 1.9072376390068576e-06,
      "loss": 3.3939,
      "step": 56685
    },
    {
      "epoch": 0.5766805013020834,
      "grad_norm": 8.374520301818848,
      "learning_rate": 1.906849168635797e-06,
      "loss": 3.2321,
      "step": 56690
    },
    {
      "epoch": 0.5767313639322916,
      "grad_norm": 9.535998344421387,
      "learning_rate": 1.9064607134403595e-06,
      "loss": 3.0934,
      "step": 56695
    },
    {
      "epoch": 0.5767822265625,
      "grad_norm": 13.945857048034668,
      "learning_rate": 1.9060722734304835e-06,
      "loss": 3.2331,
      "step": 56700
    },
    {
      "epoch": 0.5768330891927084,
      "grad_norm": 7.848199844360352,
      "learning_rate": 1.9056838486161081e-06,
      "loss": 3.391,
      "step": 56705
    },
    {
      "epoch": 0.5768839518229166,
      "grad_norm": 9.019949913024902,
      "learning_rate": 1.905295439007171e-06,
      "loss": 3.2544,
      "step": 56710
    },
    {
      "epoch": 0.576934814453125,
      "grad_norm": 8.967193603515625,
      "learning_rate": 1.9049070446136092e-06,
      "loss": 3.4716,
      "step": 56715
    },
    {
      "epoch": 0.5769856770833334,
      "grad_norm": 9.631693840026855,
      "learning_rate": 1.9045186654453578e-06,
      "loss": 2.999,
      "step": 56720
    },
    {
      "epoch": 0.5770365397135416,
      "grad_norm": 14.838756561279297,
      "learning_rate": 1.904130301512357e-06,
      "loss": 3.102,
      "step": 56725
    },
    {
      "epoch": 0.57708740234375,
      "grad_norm": 9.814398765563965,
      "learning_rate": 1.9037419528245406e-06,
      "loss": 3.1221,
      "step": 56730
    },
    {
      "epoch": 0.5771382649739584,
      "grad_norm": 10.964478492736816,
      "learning_rate": 1.9033536193918455e-06,
      "loss": 3.5277,
      "step": 56735
    },
    {
      "epoch": 0.5771891276041666,
      "grad_norm": 12.982027053833008,
      "learning_rate": 1.9029653012242059e-06,
      "loss": 3.4367,
      "step": 56740
    },
    {
      "epoch": 0.577239990234375,
      "grad_norm": 14.798869132995605,
      "learning_rate": 1.902576998331558e-06,
      "loss": 3.433,
      "step": 56745
    },
    {
      "epoch": 0.5772908528645834,
      "grad_norm": 13.125004768371582,
      "learning_rate": 1.902188710723836e-06,
      "loss": 3.7781,
      "step": 56750
    },
    {
      "epoch": 0.5773417154947916,
      "grad_norm": 9.113572120666504,
      "learning_rate": 1.9018004384109739e-06,
      "loss": 3.17,
      "step": 56755
    },
    {
      "epoch": 0.577392578125,
      "grad_norm": 13.796917915344238,
      "learning_rate": 1.9014121814029054e-06,
      "loss": 3.983,
      "step": 56760
    },
    {
      "epoch": 0.5774434407552084,
      "grad_norm": 10.384062767028809,
      "learning_rate": 1.9010239397095654e-06,
      "loss": 3.5382,
      "step": 56765
    },
    {
      "epoch": 0.5774943033854166,
      "grad_norm": 12.019784927368164,
      "learning_rate": 1.9006357133408854e-06,
      "loss": 3.4912,
      "step": 56770
    },
    {
      "epoch": 0.577545166015625,
      "grad_norm": 11.359028816223145,
      "learning_rate": 1.900247502306799e-06,
      "loss": 3.1923,
      "step": 56775
    },
    {
      "epoch": 0.5775960286458334,
      "grad_norm": 7.998235702514648,
      "learning_rate": 1.8998593066172371e-06,
      "loss": 3.2979,
      "step": 56780
    },
    {
      "epoch": 0.5776468912760416,
      "grad_norm": 10.577157020568848,
      "learning_rate": 1.899471126282133e-06,
      "loss": 3.2425,
      "step": 56785
    },
    {
      "epoch": 0.57769775390625,
      "grad_norm": 14.997776985168457,
      "learning_rate": 1.8990829613114187e-06,
      "loss": 3.3457,
      "step": 56790
    },
    {
      "epoch": 0.5777486165364584,
      "grad_norm": 11.988247871398926,
      "learning_rate": 1.8986948117150231e-06,
      "loss": 3.2524,
      "step": 56795
    },
    {
      "epoch": 0.5777994791666666,
      "grad_norm": 11.470266342163086,
      "learning_rate": 1.8983066775028789e-06,
      "loss": 3.3788,
      "step": 56800
    },
    {
      "epoch": 0.577850341796875,
      "grad_norm": 15.55947208404541,
      "learning_rate": 1.8979185586849158e-06,
      "loss": 3.3545,
      "step": 56805
    },
    {
      "epoch": 0.5779012044270834,
      "grad_norm": 12.32003402709961,
      "learning_rate": 1.8975304552710635e-06,
      "loss": 3.0713,
      "step": 56810
    },
    {
      "epoch": 0.5779520670572916,
      "grad_norm": 51.92470932006836,
      "learning_rate": 1.8971423672712514e-06,
      "loss": 3.3542,
      "step": 56815
    },
    {
      "epoch": 0.5780029296875,
      "grad_norm": 10.97673511505127,
      "learning_rate": 1.8967542946954096e-06,
      "loss": 3.5778,
      "step": 56820
    },
    {
      "epoch": 0.5780537923177084,
      "grad_norm": 13.67583179473877,
      "learning_rate": 1.896366237553466e-06,
      "loss": 3.6961,
      "step": 56825
    },
    {
      "epoch": 0.5781046549479166,
      "grad_norm": 10.543595314025879,
      "learning_rate": 1.895978195855349e-06,
      "loss": 3.6835,
      "step": 56830
    },
    {
      "epoch": 0.578155517578125,
      "grad_norm": 10.750131607055664,
      "learning_rate": 1.8955901696109864e-06,
      "loss": 3.2903,
      "step": 56835
    },
    {
      "epoch": 0.5782063802083334,
      "grad_norm": 11.75191879272461,
      "learning_rate": 1.8952021588303062e-06,
      "loss": 3.6327,
      "step": 56840
    },
    {
      "epoch": 0.5782572428385416,
      "grad_norm": 6.788650989532471,
      "learning_rate": 1.8948141635232358e-06,
      "loss": 3.3047,
      "step": 56845
    },
    {
      "epoch": 0.57830810546875,
      "grad_norm": 11.891761779785156,
      "learning_rate": 1.8944261836997013e-06,
      "loss": 3.1144,
      "step": 56850
    },
    {
      "epoch": 0.5783589680989584,
      "grad_norm": 13.748327255249023,
      "learning_rate": 1.894038219369629e-06,
      "loss": 2.9591,
      "step": 56855
    },
    {
      "epoch": 0.5784098307291666,
      "grad_norm": 14.535821914672852,
      "learning_rate": 1.893650270542946e-06,
      "loss": 3.0379,
      "step": 56860
    },
    {
      "epoch": 0.578460693359375,
      "grad_norm": 11.83864688873291,
      "learning_rate": 1.8932623372295766e-06,
      "loss": 2.9729,
      "step": 56865
    },
    {
      "epoch": 0.5785115559895834,
      "grad_norm": 11.101030349731445,
      "learning_rate": 1.8928744194394472e-06,
      "loss": 3.4374,
      "step": 56870
    },
    {
      "epoch": 0.5785624186197916,
      "grad_norm": 9.511262893676758,
      "learning_rate": 1.8924865171824803e-06,
      "loss": 3.2521,
      "step": 56875
    },
    {
      "epoch": 0.57861328125,
      "grad_norm": 26.6956787109375,
      "learning_rate": 1.8920986304686032e-06,
      "loss": 3.4854,
      "step": 56880
    },
    {
      "epoch": 0.5786641438802084,
      "grad_norm": 9.508420944213867,
      "learning_rate": 1.891710759307738e-06,
      "loss": 3.2311,
      "step": 56885
    },
    {
      "epoch": 0.5787150065104166,
      "grad_norm": 9.723649978637695,
      "learning_rate": 1.8913229037098096e-06,
      "loss": 3.2118,
      "step": 56890
    },
    {
      "epoch": 0.578765869140625,
      "grad_norm": 14.52519702911377,
      "learning_rate": 1.8909350636847391e-06,
      "loss": 3.654,
      "step": 56895
    },
    {
      "epoch": 0.5788167317708334,
      "grad_norm": 11.78041934967041,
      "learning_rate": 1.8905472392424514e-06,
      "loss": 3.1973,
      "step": 56900
    },
    {
      "epoch": 0.5788675944010416,
      "grad_norm": 8.476494789123535,
      "learning_rate": 1.8901594303928686e-06,
      "loss": 3.4497,
      "step": 56905
    },
    {
      "epoch": 0.57891845703125,
      "grad_norm": 14.352686882019043,
      "learning_rate": 1.8897716371459112e-06,
      "loss": 3.5338,
      "step": 56910
    },
    {
      "epoch": 0.5789693196614584,
      "grad_norm": 8.161144256591797,
      "learning_rate": 1.889383859511502e-06,
      "loss": 3.333,
      "step": 56915
    },
    {
      "epoch": 0.5790201822916666,
      "grad_norm": 15.801276206970215,
      "learning_rate": 1.888996097499563e-06,
      "loss": 3.31,
      "step": 56920
    },
    {
      "epoch": 0.579071044921875,
      "grad_norm": 11.01284122467041,
      "learning_rate": 1.8886083511200135e-06,
      "loss": 3.6326,
      "step": 56925
    },
    {
      "epoch": 0.5791219075520834,
      "grad_norm": 8.402137756347656,
      "learning_rate": 1.8882206203827738e-06,
      "loss": 3.0483,
      "step": 56930
    },
    {
      "epoch": 0.5791727701822916,
      "grad_norm": 16.430620193481445,
      "learning_rate": 1.8878329052977657e-06,
      "loss": 3.2311,
      "step": 56935
    },
    {
      "epoch": 0.5792236328125,
      "grad_norm": 14.661127090454102,
      "learning_rate": 1.8874452058749066e-06,
      "loss": 3.5519,
      "step": 56940
    },
    {
      "epoch": 0.5792744954427084,
      "grad_norm": 9.709014892578125,
      "learning_rate": 1.8870575221241178e-06,
      "loss": 3.7096,
      "step": 56945
    },
    {
      "epoch": 0.5793253580729166,
      "grad_norm": 11.967632293701172,
      "learning_rate": 1.8866698540553155e-06,
      "loss": 3.5083,
      "step": 56950
    },
    {
      "epoch": 0.579376220703125,
      "grad_norm": 14.899423599243164,
      "learning_rate": 1.8862822016784204e-06,
      "loss": 3.1922,
      "step": 56955
    },
    {
      "epoch": 0.5794270833333334,
      "grad_norm": 10.922184944152832,
      "learning_rate": 1.8858945650033503e-06,
      "loss": 3.6517,
      "step": 56960
    },
    {
      "epoch": 0.5794779459635416,
      "grad_norm": 12.147903442382812,
      "learning_rate": 1.8855069440400214e-06,
      "loss": 3.2416,
      "step": 56965
    },
    {
      "epoch": 0.57952880859375,
      "grad_norm": 12.775498390197754,
      "learning_rate": 1.8851193387983512e-06,
      "loss": 3.2935,
      "step": 56970
    },
    {
      "epoch": 0.5795796712239584,
      "grad_norm": 9.618669509887695,
      "learning_rate": 1.8847317492882582e-06,
      "loss": 3.8583,
      "step": 56975
    },
    {
      "epoch": 0.5796305338541666,
      "grad_norm": 11.85013484954834,
      "learning_rate": 1.884344175519657e-06,
      "loss": 3.1993,
      "step": 56980
    },
    {
      "epoch": 0.579681396484375,
      "grad_norm": 11.201725959777832,
      "learning_rate": 1.8839566175024648e-06,
      "loss": 3.2567,
      "step": 56985
    },
    {
      "epoch": 0.5797322591145834,
      "grad_norm": 8.808295249938965,
      "learning_rate": 1.8835690752465952e-06,
      "loss": 3.3521,
      "step": 56990
    },
    {
      "epoch": 0.5797831217447916,
      "grad_norm": 11.396921157836914,
      "learning_rate": 1.8831815487619654e-06,
      "loss": 3.1966,
      "step": 56995
    },
    {
      "epoch": 0.579833984375,
      "grad_norm": 13.508956909179688,
      "learning_rate": 1.88279403805849e-06,
      "loss": 3.545,
      "step": 57000
    },
    {
      "epoch": 0.5798848470052084,
      "grad_norm": 17.08087158203125,
      "learning_rate": 1.8824065431460817e-06,
      "loss": 3.224,
      "step": 57005
    },
    {
      "epoch": 0.5799357096354166,
      "grad_norm": 9.624190330505371,
      "learning_rate": 1.8820190640346562e-06,
      "loss": 3.4903,
      "step": 57010
    },
    {
      "epoch": 0.579986572265625,
      "grad_norm": 13.430679321289062,
      "learning_rate": 1.8816316007341268e-06,
      "loss": 3.0292,
      "step": 57015
    },
    {
      "epoch": 0.5800374348958334,
      "grad_norm": 9.54922103881836,
      "learning_rate": 1.881244153254406e-06,
      "loss": 3.5934,
      "step": 57020
    },
    {
      "epoch": 0.5800882975260416,
      "grad_norm": 9.288108825683594,
      "learning_rate": 1.8808567216054063e-06,
      "loss": 3.1301,
      "step": 57025
    },
    {
      "epoch": 0.58013916015625,
      "grad_norm": 11.815570831298828,
      "learning_rate": 1.880469305797041e-06,
      "loss": 3.2464,
      "step": 57030
    },
    {
      "epoch": 0.5801900227864584,
      "grad_norm": 9.707662582397461,
      "learning_rate": 1.8800819058392226e-06,
      "loss": 3.1669,
      "step": 57035
    },
    {
      "epoch": 0.5802408854166666,
      "grad_norm": 10.359423637390137,
      "learning_rate": 1.8796945217418608e-06,
      "loss": 3.3614,
      "step": 57040
    },
    {
      "epoch": 0.580291748046875,
      "grad_norm": 10.065942764282227,
      "learning_rate": 1.8793071535148672e-06,
      "loss": 3.1885,
      "step": 57045
    },
    {
      "epoch": 0.5803426106770834,
      "grad_norm": 11.11904239654541,
      "learning_rate": 1.878919801168154e-06,
      "loss": 3.1507,
      "step": 57050
    },
    {
      "epoch": 0.5803934733072916,
      "grad_norm": 15.753347396850586,
      "learning_rate": 1.8785324647116302e-06,
      "loss": 3.1128,
      "step": 57055
    },
    {
      "epoch": 0.5804443359375,
      "grad_norm": 20.453039169311523,
      "learning_rate": 1.8781451441552062e-06,
      "loss": 3.387,
      "step": 57060
    },
    {
      "epoch": 0.5804951985677084,
      "grad_norm": 16.42840576171875,
      "learning_rate": 1.8777578395087903e-06,
      "loss": 3.292,
      "step": 57065
    },
    {
      "epoch": 0.5805460611979166,
      "grad_norm": 15.925436973571777,
      "learning_rate": 1.877370550782293e-06,
      "loss": 3.1068,
      "step": 57070
    },
    {
      "epoch": 0.580596923828125,
      "grad_norm": 15.755450248718262,
      "learning_rate": 1.8769832779856234e-06,
      "loss": 3.462,
      "step": 57075
    },
    {
      "epoch": 0.5806477864583334,
      "grad_norm": 10.200871467590332,
      "learning_rate": 1.8765960211286885e-06,
      "loss": 3.6675,
      "step": 57080
    },
    {
      "epoch": 0.5806986490885416,
      "grad_norm": 9.75241756439209,
      "learning_rate": 1.8762087802213958e-06,
      "loss": 3.2287,
      "step": 57085
    },
    {
      "epoch": 0.58074951171875,
      "grad_norm": 14.130892753601074,
      "learning_rate": 1.8758215552736548e-06,
      "loss": 3.7924,
      "step": 57090
    },
    {
      "epoch": 0.5808003743489584,
      "grad_norm": 8.34168529510498,
      "learning_rate": 1.875434346295371e-06,
      "loss": 3.3221,
      "step": 57095
    },
    {
      "epoch": 0.5808512369791666,
      "grad_norm": 10.972088813781738,
      "learning_rate": 1.8750471532964504e-06,
      "loss": 3.3444,
      "step": 57100
    },
    {
      "epoch": 0.580902099609375,
      "grad_norm": 10.499613761901855,
      "learning_rate": 1.8746599762868016e-06,
      "loss": 3.2314,
      "step": 57105
    },
    {
      "epoch": 0.5809529622395834,
      "grad_norm": 18.19532585144043,
      "learning_rate": 1.8742728152763284e-06,
      "loss": 3.5773,
      "step": 57110
    },
    {
      "epoch": 0.5810038248697916,
      "grad_norm": 10.94426155090332,
      "learning_rate": 1.8738856702749379e-06,
      "loss": 3.2399,
      "step": 57115
    },
    {
      "epoch": 0.5810546875,
      "grad_norm": 10.755924224853516,
      "learning_rate": 1.8734985412925328e-06,
      "loss": 3.3212,
      "step": 57120
    },
    {
      "epoch": 0.5811055501302084,
      "grad_norm": 10.946020126342773,
      "learning_rate": 1.8731114283390195e-06,
      "loss": 3.8101,
      "step": 57125
    },
    {
      "epoch": 0.5811564127604166,
      "grad_norm": 7.830922603607178,
      "learning_rate": 1.872724331424302e-06,
      "loss": 3.2401,
      "step": 57130
    },
    {
      "epoch": 0.581207275390625,
      "grad_norm": 9.367474555969238,
      "learning_rate": 1.8723372505582837e-06,
      "loss": 3.4203,
      "step": 57135
    },
    {
      "epoch": 0.5812581380208334,
      "grad_norm": 12.73338794708252,
      "learning_rate": 1.8719501857508672e-06,
      "loss": 3.2724,
      "step": 57140
    },
    {
      "epoch": 0.5813090006510416,
      "grad_norm": 10.9208345413208,
      "learning_rate": 1.8715631370119574e-06,
      "loss": 3.503,
      "step": 57145
    },
    {
      "epoch": 0.58135986328125,
      "grad_norm": 15.710588455200195,
      "learning_rate": 1.8711761043514557e-06,
      "loss": 3.4525,
      "step": 57150
    },
    {
      "epoch": 0.5814107259114584,
      "grad_norm": 12.172246932983398,
      "learning_rate": 1.8707890877792644e-06,
      "loss": 3.1944,
      "step": 57155
    },
    {
      "epoch": 0.5814615885416666,
      "grad_norm": 15.046089172363281,
      "learning_rate": 1.8704020873052842e-06,
      "loss": 3.1853,
      "step": 57160
    },
    {
      "epoch": 0.581512451171875,
      "grad_norm": 9.864462852478027,
      "learning_rate": 1.8700151029394182e-06,
      "loss": 4.1923,
      "step": 57165
    },
    {
      "epoch": 0.5815633138020834,
      "grad_norm": 7.275952339172363,
      "learning_rate": 1.8696281346915669e-06,
      "loss": 3.5304,
      "step": 57170
    },
    {
      "epoch": 0.5816141764322916,
      "grad_norm": 9.70826244354248,
      "learning_rate": 1.8692411825716294e-06,
      "loss": 3.5269,
      "step": 57175
    },
    {
      "epoch": 0.5816650390625,
      "grad_norm": 12.5178804397583,
      "learning_rate": 1.8688542465895066e-06,
      "loss": 3.5193,
      "step": 57180
    },
    {
      "epoch": 0.5817159016927084,
      "grad_norm": 13.794378280639648,
      "learning_rate": 1.8684673267550984e-06,
      "loss": 3.3048,
      "step": 57185
    },
    {
      "epoch": 0.5817667643229166,
      "grad_norm": 12.371304512023926,
      "learning_rate": 1.868080423078305e-06,
      "loss": 3.3683,
      "step": 57190
    },
    {
      "epoch": 0.581817626953125,
      "grad_norm": 16.101579666137695,
      "learning_rate": 1.8676935355690228e-06,
      "loss": 3.2473,
      "step": 57195
    },
    {
      "epoch": 0.5818684895833334,
      "grad_norm": 12.373351097106934,
      "learning_rate": 1.8673066642371524e-06,
      "loss": 2.9094,
      "step": 57200
    },
    {
      "epoch": 0.5819193522135416,
      "grad_norm": 10.716297149658203,
      "learning_rate": 1.8669198090925915e-06,
      "loss": 3.1835,
      "step": 57205
    },
    {
      "epoch": 0.58197021484375,
      "grad_norm": 12.337839126586914,
      "learning_rate": 1.8665329701452367e-06,
      "loss": 3.3965,
      "step": 57210
    },
    {
      "epoch": 0.5820210774739584,
      "grad_norm": 7.9632978439331055,
      "learning_rate": 1.8661461474049853e-06,
      "loss": 3.2287,
      "step": 57215
    },
    {
      "epoch": 0.5820719401041666,
      "grad_norm": 13.325944900512695,
      "learning_rate": 1.8657593408817351e-06,
      "loss": 3.543,
      "step": 57220
    },
    {
      "epoch": 0.582122802734375,
      "grad_norm": 8.320006370544434,
      "learning_rate": 1.8653725505853814e-06,
      "loss": 3.4453,
      "step": 57225
    },
    {
      "epoch": 0.5821736653645834,
      "grad_norm": 21.190378189086914,
      "learning_rate": 1.8649857765258213e-06,
      "loss": 3.3959,
      "step": 57230
    },
    {
      "epoch": 0.5822245279947916,
      "grad_norm": 11.889701843261719,
      "learning_rate": 1.8645990187129486e-06,
      "loss": 3.0289,
      "step": 57235
    },
    {
      "epoch": 0.582275390625,
      "grad_norm": 9.687341690063477,
      "learning_rate": 1.8642122771566598e-06,
      "loss": 3.5415,
      "step": 57240
    },
    {
      "epoch": 0.5823262532552084,
      "grad_norm": 9.341906547546387,
      "learning_rate": 1.8638255518668496e-06,
      "loss": 2.9307,
      "step": 57245
    },
    {
      "epoch": 0.5823771158854166,
      "grad_norm": 12.794500350952148,
      "learning_rate": 1.8634388428534116e-06,
      "loss": 3.4308,
      "step": 57250
    },
    {
      "epoch": 0.582427978515625,
      "grad_norm": 13.471244812011719,
      "learning_rate": 1.8630521501262393e-06,
      "loss": 2.9125,
      "step": 57255
    },
    {
      "epoch": 0.5824788411458334,
      "grad_norm": 11.451852798461914,
      "learning_rate": 1.8626654736952279e-06,
      "loss": 3.3624,
      "step": 57260
    },
    {
      "epoch": 0.5825297037760416,
      "grad_norm": 9.186808586120605,
      "learning_rate": 1.8622788135702686e-06,
      "loss": 3.2651,
      "step": 57265
    },
    {
      "epoch": 0.58258056640625,
      "grad_norm": 7.854951858520508,
      "learning_rate": 1.8618921697612552e-06,
      "loss": 2.9606,
      "step": 57270
    },
    {
      "epoch": 0.5826314290364584,
      "grad_norm": 8.688505172729492,
      "learning_rate": 1.8615055422780787e-06,
      "loss": 3.2603,
      "step": 57275
    },
    {
      "epoch": 0.5826822916666666,
      "grad_norm": 15.499544143676758,
      "learning_rate": 1.8611189311306316e-06,
      "loss": 3.137,
      "step": 57280
    },
    {
      "epoch": 0.582733154296875,
      "grad_norm": 13.060683250427246,
      "learning_rate": 1.860732336328806e-06,
      "loss": 3.1386,
      "step": 57285
    },
    {
      "epoch": 0.5827840169270834,
      "grad_norm": 9.663434028625488,
      "learning_rate": 1.860345757882491e-06,
      "loss": 3.2919,
      "step": 57290
    },
    {
      "epoch": 0.5828348795572916,
      "grad_norm": 10.981467247009277,
      "learning_rate": 1.8599591958015787e-06,
      "loss": 3.4003,
      "step": 57295
    },
    {
      "epoch": 0.5828857421875,
      "grad_norm": 10.95694351196289,
      "learning_rate": 1.8595726500959591e-06,
      "loss": 3.4265,
      "step": 57300
    },
    {
      "epoch": 0.5829366048177084,
      "grad_norm": 13.341103553771973,
      "learning_rate": 1.859186120775521e-06,
      "loss": 3.1631,
      "step": 57305
    },
    {
      "epoch": 0.5829874674479166,
      "grad_norm": 10.463839530944824,
      "learning_rate": 1.8587996078501535e-06,
      "loss": 3.4325,
      "step": 57310
    },
    {
      "epoch": 0.583038330078125,
      "grad_norm": 11.585979461669922,
      "learning_rate": 1.858413111329747e-06,
      "loss": 3.0571,
      "step": 57315
    },
    {
      "epoch": 0.5830891927083334,
      "grad_norm": 12.85406494140625,
      "learning_rate": 1.8580266312241888e-06,
      "loss": 2.9965,
      "step": 57320
    },
    {
      "epoch": 0.5831400553385416,
      "grad_norm": 12.769957542419434,
      "learning_rate": 1.8576401675433675e-06,
      "loss": 3.3309,
      "step": 57325
    },
    {
      "epoch": 0.58319091796875,
      "grad_norm": 11.573023796081543,
      "learning_rate": 1.8572537202971686e-06,
      "loss": 3.1622,
      "step": 57330
    },
    {
      "epoch": 0.5832417805989584,
      "grad_norm": 13.39962100982666,
      "learning_rate": 1.8568672894954823e-06,
      "loss": 3.3914,
      "step": 57335
    },
    {
      "epoch": 0.5832926432291666,
      "grad_norm": 10.790485382080078,
      "learning_rate": 1.8564808751481938e-06,
      "loss": 3.508,
      "step": 57340
    },
    {
      "epoch": 0.583343505859375,
      "grad_norm": 17.249465942382812,
      "learning_rate": 1.8560944772651898e-06,
      "loss": 3.2018,
      "step": 57345
    },
    {
      "epoch": 0.5833943684895834,
      "grad_norm": 12.020594596862793,
      "learning_rate": 1.855708095856355e-06,
      "loss": 3.1479,
      "step": 57350
    },
    {
      "epoch": 0.5834452311197916,
      "grad_norm": 15.753055572509766,
      "learning_rate": 1.8553217309315763e-06,
      "loss": 3.2538,
      "step": 57355
    },
    {
      "epoch": 0.58349609375,
      "grad_norm": 13.768061637878418,
      "learning_rate": 1.854935382500739e-06,
      "loss": 3.3499,
      "step": 57360
    },
    {
      "epoch": 0.5835469563802084,
      "grad_norm": 16.716020584106445,
      "learning_rate": 1.8545490505737266e-06,
      "loss": 3.6358,
      "step": 57365
    },
    {
      "epoch": 0.5835978190104166,
      "grad_norm": 15.221572875976562,
      "learning_rate": 1.854162735160423e-06,
      "loss": 3.1542,
      "step": 57370
    },
    {
      "epoch": 0.583648681640625,
      "grad_norm": 10.232190132141113,
      "learning_rate": 1.8537764362707139e-06,
      "loss": 3.1022,
      "step": 57375
    },
    {
      "epoch": 0.5836995442708334,
      "grad_norm": 10.174397468566895,
      "learning_rate": 1.853390153914481e-06,
      "loss": 3.1219,
      "step": 57380
    },
    {
      "epoch": 0.5837504069010416,
      "grad_norm": 8.714889526367188,
      "learning_rate": 1.8530038881016082e-06,
      "loss": 3.5887,
      "step": 57385
    },
    {
      "epoch": 0.58380126953125,
      "grad_norm": 10.971238136291504,
      "learning_rate": 1.8526176388419767e-06,
      "loss": 3.0598,
      "step": 57390
    },
    {
      "epoch": 0.5838521321614584,
      "grad_norm": 7.573197364807129,
      "learning_rate": 1.8522314061454699e-06,
      "loss": 3.1088,
      "step": 57395
    },
    {
      "epoch": 0.5839029947916666,
      "grad_norm": 11.627606391906738,
      "learning_rate": 1.8518451900219695e-06,
      "loss": 3.3229,
      "step": 57400
    },
    {
      "epoch": 0.583953857421875,
      "grad_norm": 13.206156730651855,
      "learning_rate": 1.8514589904813551e-06,
      "loss": 2.9272,
      "step": 57405
    },
    {
      "epoch": 0.5840047200520834,
      "grad_norm": 14.942124366760254,
      "learning_rate": 1.8510728075335093e-06,
      "loss": 3.0356,
      "step": 57410
    },
    {
      "epoch": 0.5840555826822916,
      "grad_norm": 7.587666988372803,
      "learning_rate": 1.8506866411883128e-06,
      "loss": 3.02,
      "step": 57415
    },
    {
      "epoch": 0.5841064453125,
      "grad_norm": 7.650280475616455,
      "learning_rate": 1.8503004914556439e-06,
      "loss": 3.3738,
      "step": 57420
    },
    {
      "epoch": 0.5841573079427084,
      "grad_norm": 8.10651683807373,
      "learning_rate": 1.8499143583453822e-06,
      "loss": 3.0114,
      "step": 57425
    },
    {
      "epoch": 0.5842081705729166,
      "grad_norm": 9.125300407409668,
      "learning_rate": 1.8495282418674088e-06,
      "loss": 3.2324,
      "step": 57430
    },
    {
      "epoch": 0.584259033203125,
      "grad_norm": 11.46408748626709,
      "learning_rate": 1.8491421420316009e-06,
      "loss": 3.0857,
      "step": 57435
    },
    {
      "epoch": 0.5843098958333334,
      "grad_norm": 11.1484956741333,
      "learning_rate": 1.848756058847837e-06,
      "loss": 3.1192,
      "step": 57440
    },
    {
      "epoch": 0.5843607584635416,
      "grad_norm": 10.936185836791992,
      "learning_rate": 1.848369992325994e-06,
      "loss": 3.6811,
      "step": 57445
    },
    {
      "epoch": 0.58441162109375,
      "grad_norm": 16.094820022583008,
      "learning_rate": 1.8479839424759511e-06,
      "loss": 3.5156,
      "step": 57450
    },
    {
      "epoch": 0.5844624837239584,
      "grad_norm": 10.838696479797363,
      "learning_rate": 1.847597909307585e-06,
      "loss": 3.5173,
      "step": 57455
    },
    {
      "epoch": 0.5845133463541666,
      "grad_norm": 13.69144058227539,
      "learning_rate": 1.8472118928307712e-06,
      "loss": 3.2802,
      "step": 57460
    },
    {
      "epoch": 0.584564208984375,
      "grad_norm": 19.055984497070312,
      "learning_rate": 1.8468258930553856e-06,
      "loss": 3.3391,
      "step": 57465
    },
    {
      "epoch": 0.5846150716145834,
      "grad_norm": 14.424437522888184,
      "learning_rate": 1.846439909991306e-06,
      "loss": 3.3721,
      "step": 57470
    },
    {
      "epoch": 0.5846659342447916,
      "grad_norm": 12.521187782287598,
      "learning_rate": 1.846053943648406e-06,
      "loss": 3.2019,
      "step": 57475
    },
    {
      "epoch": 0.584716796875,
      "grad_norm": 16.133342742919922,
      "learning_rate": 1.8456679940365612e-06,
      "loss": 3.5034,
      "step": 57480
    },
    {
      "epoch": 0.5847676595052084,
      "grad_norm": 13.612138748168945,
      "learning_rate": 1.845282061165644e-06,
      "loss": 3.3122,
      "step": 57485
    },
    {
      "epoch": 0.5848185221354166,
      "grad_norm": 11.59680461883545,
      "learning_rate": 1.8448961450455318e-06,
      "loss": 3.8087,
      "step": 57490
    },
    {
      "epoch": 0.584869384765625,
      "grad_norm": 14.264731407165527,
      "learning_rate": 1.844510245686096e-06,
      "loss": 3.065,
      "step": 57495
    },
    {
      "epoch": 0.5849202473958334,
      "grad_norm": 11.807320594787598,
      "learning_rate": 1.8441243630972093e-06,
      "loss": 3.3935,
      "step": 57500
    },
    {
      "epoch": 0.5849711100260416,
      "grad_norm": 19.176963806152344,
      "learning_rate": 1.8437384972887468e-06,
      "loss": 3.2933,
      "step": 57505
    },
    {
      "epoch": 0.58502197265625,
      "grad_norm": 11.13174819946289,
      "learning_rate": 1.8433526482705787e-06,
      "loss": 3.4016,
      "step": 57510
    },
    {
      "epoch": 0.5850728352864584,
      "grad_norm": 8.262331008911133,
      "learning_rate": 1.8429668160525778e-06,
      "loss": 3.1822,
      "step": 57515
    },
    {
      "epoch": 0.5851236979166666,
      "grad_norm": 12.877121925354004,
      "learning_rate": 1.8425810006446144e-06,
      "loss": 3.4957,
      "step": 57520
    },
    {
      "epoch": 0.585174560546875,
      "grad_norm": 18.290891647338867,
      "learning_rate": 1.8421952020565603e-06,
      "loss": 3.334,
      "step": 57525
    },
    {
      "epoch": 0.5852254231770834,
      "grad_norm": 14.739322662353516,
      "learning_rate": 1.841809420298287e-06,
      "loss": 3.3009,
      "step": 57530
    },
    {
      "epoch": 0.5852762858072916,
      "grad_norm": 10.935076713562012,
      "learning_rate": 1.841423655379663e-06,
      "loss": 3.2133,
      "step": 57535
    },
    {
      "epoch": 0.5853271484375,
      "grad_norm": 14.294183731079102,
      "learning_rate": 1.8410379073105583e-06,
      "loss": 3.6313,
      "step": 57540
    },
    {
      "epoch": 0.5853780110677084,
      "grad_norm": 14.829392433166504,
      "learning_rate": 1.8406521761008433e-06,
      "loss": 3.2714,
      "step": 57545
    },
    {
      "epoch": 0.5854288736979166,
      "grad_norm": 14.30656909942627,
      "learning_rate": 1.8402664617603857e-06,
      "loss": 3.5128,
      "step": 57550
    },
    {
      "epoch": 0.585479736328125,
      "grad_norm": 8.534845352172852,
      "learning_rate": 1.8398807642990548e-06,
      "loss": 3.4336,
      "step": 57555
    },
    {
      "epoch": 0.5855305989583334,
      "grad_norm": 9.407914161682129,
      "learning_rate": 1.8394950837267172e-06,
      "loss": 3.5546,
      "step": 57560
    },
    {
      "epoch": 0.5855814615885416,
      "grad_norm": 13.708868026733398,
      "learning_rate": 1.8391094200532417e-06,
      "loss": 3.1824,
      "step": 57565
    },
    {
      "epoch": 0.58563232421875,
      "grad_norm": 9.58594799041748,
      "learning_rate": 1.8387237732884954e-06,
      "loss": 3.2787,
      "step": 57570
    },
    {
      "epoch": 0.5856831868489584,
      "grad_norm": 12.1710844039917,
      "learning_rate": 1.8383381434423448e-06,
      "loss": 3.3865,
      "step": 57575
    },
    {
      "epoch": 0.5857340494791666,
      "grad_norm": 14.10545825958252,
      "learning_rate": 1.8379525305246547e-06,
      "loss": 3.26,
      "step": 57580
    },
    {
      "epoch": 0.585784912109375,
      "grad_norm": 8.860973358154297,
      "learning_rate": 1.8375669345452935e-06,
      "loss": 3.3786,
      "step": 57585
    },
    {
      "epoch": 0.5858357747395834,
      "grad_norm": 15.432395935058594,
      "learning_rate": 1.8371813555141246e-06,
      "loss": 3.4559,
      "step": 57590
    },
    {
      "epoch": 0.5858866373697916,
      "grad_norm": 11.764700889587402,
      "learning_rate": 1.8367957934410132e-06,
      "loss": 3.223,
      "step": 57595
    },
    {
      "epoch": 0.5859375,
      "grad_norm": 10.494485855102539,
      "learning_rate": 1.8364102483358253e-06,
      "loss": 3.4047,
      "step": 57600
    },
    {
      "epoch": 0.5859883626302084,
      "grad_norm": 7.719424247741699,
      "learning_rate": 1.8360247202084236e-06,
      "loss": 3.2869,
      "step": 57605
    },
    {
      "epoch": 0.5860392252604166,
      "grad_norm": 11.962519645690918,
      "learning_rate": 1.8356392090686722e-06,
      "loss": 3.3633,
      "step": 57610
    },
    {
      "epoch": 0.586090087890625,
      "grad_norm": 12.523601531982422,
      "learning_rate": 1.8352537149264332e-06,
      "loss": 3.1762,
      "step": 57615
    },
    {
      "epoch": 0.5861409505208334,
      "grad_norm": 14.770298957824707,
      "learning_rate": 1.834868237791571e-06,
      "loss": 3.7998,
      "step": 57620
    },
    {
      "epoch": 0.5861918131510416,
      "grad_norm": 13.773069381713867,
      "learning_rate": 1.8344827776739478e-06,
      "loss": 3.3806,
      "step": 57625
    },
    {
      "epoch": 0.58624267578125,
      "grad_norm": 7.39944314956665,
      "learning_rate": 1.8340973345834244e-06,
      "loss": 3.0848,
      "step": 57630
    },
    {
      "epoch": 0.5862935384114584,
      "grad_norm": 11.790326118469238,
      "learning_rate": 1.8337119085298622e-06,
      "loss": 3.1497,
      "step": 57635
    },
    {
      "epoch": 0.5863444010416666,
      "grad_norm": 16.520463943481445,
      "learning_rate": 1.8333264995231237e-06,
      "loss": 3.4205,
      "step": 57640
    },
    {
      "epoch": 0.586395263671875,
      "grad_norm": 10.090576171875,
      "learning_rate": 1.832941107573069e-06,
      "loss": 3.6189,
      "step": 57645
    },
    {
      "epoch": 0.5864461263020834,
      "grad_norm": 10.001154899597168,
      "learning_rate": 1.8325557326895576e-06,
      "loss": 3.2779,
      "step": 57650
    },
    {
      "epoch": 0.5864969889322916,
      "grad_norm": 10.244816780090332,
      "learning_rate": 1.8321703748824488e-06,
      "loss": 3.094,
      "step": 57655
    },
    {
      "epoch": 0.5865478515625,
      "grad_norm": 9.945894241333008,
      "learning_rate": 1.8317850341616038e-06,
      "loss": 3.3676,
      "step": 57660
    },
    {
      "epoch": 0.5865987141927084,
      "grad_norm": 13.106073379516602,
      "learning_rate": 1.83139971053688e-06,
      "loss": 3.7474,
      "step": 57665
    },
    {
      "epoch": 0.5866495768229166,
      "grad_norm": 10.941320419311523,
      "learning_rate": 1.8310144040181365e-06,
      "loss": 3.1596,
      "step": 57670
    },
    {
      "epoch": 0.586700439453125,
      "grad_norm": 11.458958625793457,
      "learning_rate": 1.83062911461523e-06,
      "loss": 3.3828,
      "step": 57675
    },
    {
      "epoch": 0.5867513020833334,
      "grad_norm": 16.086597442626953,
      "learning_rate": 1.8302438423380192e-06,
      "loss": 3.5009,
      "step": 57680
    },
    {
      "epoch": 0.5868021647135416,
      "grad_norm": 10.989362716674805,
      "learning_rate": 1.8298585871963614e-06,
      "loss": 3.2763,
      "step": 57685
    },
    {
      "epoch": 0.58685302734375,
      "grad_norm": 13.481184005737305,
      "learning_rate": 1.8294733492001122e-06,
      "loss": 3.7339,
      "step": 57690
    },
    {
      "epoch": 0.5869038899739584,
      "grad_norm": 14.661421775817871,
      "learning_rate": 1.829088128359129e-06,
      "loss": 3.2107,
      "step": 57695
    },
    {
      "epoch": 0.5869547526041666,
      "grad_norm": 10.293462753295898,
      "learning_rate": 1.828702924683267e-06,
      "loss": 3.1729,
      "step": 57700
    },
    {
      "epoch": 0.587005615234375,
      "grad_norm": 12.358012199401855,
      "learning_rate": 1.8283177381823813e-06,
      "loss": 3.8046,
      "step": 57705
    },
    {
      "epoch": 0.5870564778645834,
      "grad_norm": 10.661593437194824,
      "learning_rate": 1.8279325688663263e-06,
      "loss": 3.9206,
      "step": 57710
    },
    {
      "epoch": 0.5871073404947916,
      "grad_norm": 10.541525840759277,
      "learning_rate": 1.8275474167449586e-06,
      "loss": 3.9404,
      "step": 57715
    },
    {
      "epoch": 0.587158203125,
      "grad_norm": 7.5321526527404785,
      "learning_rate": 1.8271622818281305e-06,
      "loss": 3.0435,
      "step": 57720
    },
    {
      "epoch": 0.5872090657552084,
      "grad_norm": 9.740371704101562,
      "learning_rate": 1.8267771641256959e-06,
      "loss": 3.4587,
      "step": 57725
    },
    {
      "epoch": 0.5872599283854166,
      "grad_norm": 12.067994117736816,
      "learning_rate": 1.8263920636475075e-06,
      "loss": 3.479,
      "step": 57730
    },
    {
      "epoch": 0.587310791015625,
      "grad_norm": 9.233492851257324,
      "learning_rate": 1.8260069804034186e-06,
      "loss": 3.1598,
      "step": 57735
    },
    {
      "epoch": 0.5873616536458334,
      "grad_norm": 12.713920593261719,
      "learning_rate": 1.825621914403282e-06,
      "loss": 3.2037,
      "step": 57740
    },
    {
      "epoch": 0.5874125162760416,
      "grad_norm": 14.358120918273926,
      "learning_rate": 1.8252368656569483e-06,
      "loss": 3.373,
      "step": 57745
    },
    {
      "epoch": 0.58746337890625,
      "grad_norm": 8.86528491973877,
      "learning_rate": 1.8248518341742688e-06,
      "loss": 3.1788,
      "step": 57750
    },
    {
      "epoch": 0.5875142415364584,
      "grad_norm": 11.562697410583496,
      "learning_rate": 1.8244668199650967e-06,
      "loss": 3.4024,
      "step": 57755
    },
    {
      "epoch": 0.5875651041666666,
      "grad_norm": 15.956350326538086,
      "learning_rate": 1.8240818230392797e-06,
      "loss": 3.8836,
      "step": 57760
    },
    {
      "epoch": 0.587615966796875,
      "grad_norm": 11.165114402770996,
      "learning_rate": 1.8236968434066701e-06,
      "loss": 3.3067,
      "step": 57765
    },
    {
      "epoch": 0.5876668294270834,
      "grad_norm": 9.314546585083008,
      "learning_rate": 1.823311881077115e-06,
      "loss": 3.6705,
      "step": 57770
    },
    {
      "epoch": 0.5877176920572916,
      "grad_norm": 9.98790168762207,
      "learning_rate": 1.822926936060466e-06,
      "loss": 3.1524,
      "step": 57775
    },
    {
      "epoch": 0.5877685546875,
      "grad_norm": 12.551997184753418,
      "learning_rate": 1.8225420083665713e-06,
      "loss": 3.7721,
      "step": 57780
    },
    {
      "epoch": 0.5878194173177084,
      "grad_norm": 14.371289253234863,
      "learning_rate": 1.8221570980052767e-06,
      "loss": 3.3989,
      "step": 57785
    },
    {
      "epoch": 0.5878702799479166,
      "grad_norm": 12.440781593322754,
      "learning_rate": 1.8217722049864339e-06,
      "loss": 3.0281,
      "step": 57790
    },
    {
      "epoch": 0.587921142578125,
      "grad_norm": 12.106391906738281,
      "learning_rate": 1.821387329319888e-06,
      "loss": 3.229,
      "step": 57795
    },
    {
      "epoch": 0.5879720052083334,
      "grad_norm": 13.718605041503906,
      "learning_rate": 1.8210024710154871e-06,
      "loss": 2.9594,
      "step": 57800
    },
    {
      "epoch": 0.5880228678385416,
      "grad_norm": 12.633403778076172,
      "learning_rate": 1.8206176300830757e-06,
      "loss": 3.2798,
      "step": 57805
    },
    {
      "epoch": 0.58807373046875,
      "grad_norm": 11.287311553955078,
      "learning_rate": 1.820232806532502e-06,
      "loss": 3.5343,
      "step": 57810
    },
    {
      "epoch": 0.5881245930989584,
      "grad_norm": 16.78968620300293,
      "learning_rate": 1.8198480003736114e-06,
      "loss": 3.252,
      "step": 57815
    },
    {
      "epoch": 0.5881754557291666,
      "grad_norm": 15.007670402526855,
      "learning_rate": 1.8194632116162476e-06,
      "loss": 4.1343,
      "step": 57820
    },
    {
      "epoch": 0.588226318359375,
      "grad_norm": 15.56428337097168,
      "learning_rate": 1.819078440270256e-06,
      "loss": 3.3209,
      "step": 57825
    },
    {
      "epoch": 0.5882771809895834,
      "grad_norm": 9.358656883239746,
      "learning_rate": 1.8186936863454819e-06,
      "loss": 3.3396,
      "step": 57830
    },
    {
      "epoch": 0.5883280436197916,
      "grad_norm": 11.954632759094238,
      "learning_rate": 1.818308949851768e-06,
      "loss": 3.9432,
      "step": 57835
    },
    {
      "epoch": 0.58837890625,
      "grad_norm": 10.74793815612793,
      "learning_rate": 1.8179242307989585e-06,
      "loss": 3.2117,
      "step": 57840
    },
    {
      "epoch": 0.5884297688802084,
      "grad_norm": 15.294000625610352,
      "learning_rate": 1.817539529196895e-06,
      "loss": 3.9292,
      "step": 57845
    },
    {
      "epoch": 0.5884806315104166,
      "grad_norm": 15.180642127990723,
      "learning_rate": 1.8171548450554213e-06,
      "loss": 3.2566,
      "step": 57850
    },
    {
      "epoch": 0.588531494140625,
      "grad_norm": 16.156967163085938,
      "learning_rate": 1.8167701783843795e-06,
      "loss": 3.7234,
      "step": 57855
    },
    {
      "epoch": 0.5885823567708334,
      "grad_norm": 15.802820205688477,
      "learning_rate": 1.8163855291936104e-06,
      "loss": 3.2445,
      "step": 57860
    },
    {
      "epoch": 0.5886332194010416,
      "grad_norm": 11.530492782592773,
      "learning_rate": 1.816000897492955e-06,
      "loss": 3.3812,
      "step": 57865
    },
    {
      "epoch": 0.58868408203125,
      "grad_norm": 10.436324119567871,
      "learning_rate": 1.8156162832922553e-06,
      "loss": 3.1804,
      "step": 57870
    },
    {
      "epoch": 0.5887349446614584,
      "grad_norm": 10.296968460083008,
      "learning_rate": 1.8152316866013504e-06,
      "loss": 3.2323,
      "step": 57875
    },
    {
      "epoch": 0.5887858072916666,
      "grad_norm": 8.679858207702637,
      "learning_rate": 1.81484710743008e-06,
      "loss": 2.959,
      "step": 57880
    },
    {
      "epoch": 0.588836669921875,
      "grad_norm": 10.867081642150879,
      "learning_rate": 1.8144625457882848e-06,
      "loss": 3.1545,
      "step": 57885
    },
    {
      "epoch": 0.5888875325520834,
      "grad_norm": 10.677846908569336,
      "learning_rate": 1.8140780016858022e-06,
      "loss": 3.3374,
      "step": 57890
    },
    {
      "epoch": 0.5889383951822916,
      "grad_norm": 7.242464065551758,
      "learning_rate": 1.8136934751324726e-06,
      "loss": 3.5794,
      "step": 57895
    },
    {
      "epoch": 0.5889892578125,
      "grad_norm": 11.086483001708984,
      "learning_rate": 1.813308966138131e-06,
      "loss": 3.9205,
      "step": 57900
    },
    {
      "epoch": 0.5890401204427084,
      "grad_norm": 15.184897422790527,
      "learning_rate": 1.8129244747126176e-06,
      "loss": 3.2889,
      "step": 57905
    },
    {
      "epoch": 0.5890909830729166,
      "grad_norm": 14.707979202270508,
      "learning_rate": 1.812540000865769e-06,
      "loss": 2.9598,
      "step": 57910
    },
    {
      "epoch": 0.589141845703125,
      "grad_norm": 14.677459716796875,
      "learning_rate": 1.8121555446074212e-06,
      "loss": 3.4606,
      "step": 57915
    },
    {
      "epoch": 0.5891927083333334,
      "grad_norm": 10.218809127807617,
      "learning_rate": 1.8117711059474102e-06,
      "loss": 3.0858,
      "step": 57920
    },
    {
      "epoch": 0.5892435709635416,
      "grad_norm": 12.145784378051758,
      "learning_rate": 1.8113866848955733e-06,
      "loss": 3.6013,
      "step": 57925
    },
    {
      "epoch": 0.58929443359375,
      "grad_norm": 9.029804229736328,
      "learning_rate": 1.8110022814617446e-06,
      "loss": 2.7677,
      "step": 57930
    },
    {
      "epoch": 0.5893452962239584,
      "grad_norm": 15.142792701721191,
      "learning_rate": 1.810617895655759e-06,
      "loss": 3.2275,
      "step": 57935
    },
    {
      "epoch": 0.5893961588541666,
      "grad_norm": 14.707941055297852,
      "learning_rate": 1.8102335274874503e-06,
      "loss": 3.5628,
      "step": 57940
    },
    {
      "epoch": 0.589447021484375,
      "grad_norm": 13.977078437805176,
      "learning_rate": 1.8098491769666549e-06,
      "loss": 3.6672,
      "step": 57945
    },
    {
      "epoch": 0.5894978841145834,
      "grad_norm": 11.963973045349121,
      "learning_rate": 1.8094648441032037e-06,
      "loss": 3.4277,
      "step": 57950
    },
    {
      "epoch": 0.5895487467447916,
      "grad_norm": 13.8197021484375,
      "learning_rate": 1.8090805289069316e-06,
      "loss": 3.0472,
      "step": 57955
    },
    {
      "epoch": 0.589599609375,
      "grad_norm": 8.149839401245117,
      "learning_rate": 1.8086962313876693e-06,
      "loss": 3.2089,
      "step": 57960
    },
    {
      "epoch": 0.5896504720052084,
      "grad_norm": 15.027643203735352,
      "learning_rate": 1.8083119515552506e-06,
      "loss": 3.4859,
      "step": 57965
    },
    {
      "epoch": 0.5897013346354166,
      "grad_norm": 10.01371955871582,
      "learning_rate": 1.807927689419507e-06,
      "loss": 3.4033,
      "step": 57970
    },
    {
      "epoch": 0.589752197265625,
      "grad_norm": 16.41242218017578,
      "learning_rate": 1.8075434449902693e-06,
      "loss": 3.3512,
      "step": 57975
    },
    {
      "epoch": 0.5898030598958334,
      "grad_norm": 7.639359474182129,
      "learning_rate": 1.8071592182773676e-06,
      "loss": 3.275,
      "step": 57980
    },
    {
      "epoch": 0.5898539225260416,
      "grad_norm": 17.695276260375977,
      "learning_rate": 1.8067750092906345e-06,
      "loss": 3.1948,
      "step": 57985
    },
    {
      "epoch": 0.58990478515625,
      "grad_norm": 16.86260223388672,
      "learning_rate": 1.8063908180398975e-06,
      "loss": 3.304,
      "step": 57990
    },
    {
      "epoch": 0.5899556477864584,
      "grad_norm": 11.927443504333496,
      "learning_rate": 1.8060066445349866e-06,
      "loss": 3.3504,
      "step": 57995
    },
    {
      "epoch": 0.5900065104166666,
      "grad_norm": 14.542708396911621,
      "learning_rate": 1.8056224887857323e-06,
      "loss": 3.6553,
      "step": 58000
    },
    {
      "epoch": 0.590057373046875,
      "grad_norm": 16.340160369873047,
      "learning_rate": 1.8052383508019616e-06,
      "loss": 2.9317,
      "step": 58005
    },
    {
      "epoch": 0.5901082356770834,
      "grad_norm": 15.192581176757812,
      "learning_rate": 1.8048542305935036e-06,
      "loss": 3.1567,
      "step": 58010
    },
    {
      "epoch": 0.5901590983072916,
      "grad_norm": 12.118131637573242,
      "learning_rate": 1.804470128170184e-06,
      "loss": 3.0627,
      "step": 58015
    },
    {
      "epoch": 0.5902099609375,
      "grad_norm": 12.99780559539795,
      "learning_rate": 1.804086043541832e-06,
      "loss": 3.3608,
      "step": 58020
    },
    {
      "epoch": 0.5902608235677084,
      "grad_norm": 9.27899169921875,
      "learning_rate": 1.8037019767182742e-06,
      "loss": 3.183,
      "step": 58025
    },
    {
      "epoch": 0.5903116861979166,
      "grad_norm": 10.601043701171875,
      "learning_rate": 1.8033179277093358e-06,
      "loss": 2.9006,
      "step": 58030
    },
    {
      "epoch": 0.590362548828125,
      "grad_norm": 10.075181007385254,
      "learning_rate": 1.8029338965248424e-06,
      "loss": 3.3884,
      "step": 58035
    },
    {
      "epoch": 0.5904134114583334,
      "grad_norm": 10.405427932739258,
      "learning_rate": 1.8025498831746212e-06,
      "loss": 2.99,
      "step": 58040
    },
    {
      "epoch": 0.5904642740885416,
      "grad_norm": 13.567709922790527,
      "learning_rate": 1.8021658876684955e-06,
      "loss": 3.2124,
      "step": 58045
    },
    {
      "epoch": 0.59051513671875,
      "grad_norm": 14.217094421386719,
      "learning_rate": 1.8017819100162904e-06,
      "loss": 3.4022,
      "step": 58050
    },
    {
      "epoch": 0.5905659993489584,
      "grad_norm": 12.847613334655762,
      "learning_rate": 1.801397950227829e-06,
      "loss": 3.3816,
      "step": 58055
    },
    {
      "epoch": 0.5906168619791666,
      "grad_norm": 8.856525421142578,
      "learning_rate": 1.8010140083129355e-06,
      "loss": 3.4207,
      "step": 58060
    },
    {
      "epoch": 0.590667724609375,
      "grad_norm": 9.142597198486328,
      "learning_rate": 1.8006300842814337e-06,
      "loss": 3.2907,
      "step": 58065
    },
    {
      "epoch": 0.5907185872395834,
      "grad_norm": 12.949610710144043,
      "learning_rate": 1.8002461781431447e-06,
      "loss": 3.5367,
      "step": 58070
    },
    {
      "epoch": 0.5907694498697916,
      "grad_norm": 11.678544044494629,
      "learning_rate": 1.799862289907891e-06,
      "loss": 3.0713,
      "step": 58075
    },
    {
      "epoch": 0.5908203125,
      "grad_norm": 9.619078636169434,
      "learning_rate": 1.799478419585496e-06,
      "loss": 3.4691,
      "step": 58080
    },
    {
      "epoch": 0.5908711751302084,
      "grad_norm": 12.252608299255371,
      "learning_rate": 1.7990945671857784e-06,
      "loss": 3.4869,
      "step": 58085
    },
    {
      "epoch": 0.5909220377604166,
      "grad_norm": 11.362506866455078,
      "learning_rate": 1.7987107327185605e-06,
      "loss": 3.1936,
      "step": 58090
    },
    {
      "epoch": 0.590972900390625,
      "grad_norm": 14.540589332580566,
      "learning_rate": 1.7983269161936622e-06,
      "loss": 3.2106,
      "step": 58095
    },
    {
      "epoch": 0.5910237630208334,
      "grad_norm": 9.324612617492676,
      "learning_rate": 1.797943117620904e-06,
      "loss": 3.0983,
      "step": 58100
    },
    {
      "epoch": 0.5910746256510416,
      "grad_norm": 10.476067543029785,
      "learning_rate": 1.7975593370101044e-06,
      "loss": 3.0299,
      "step": 58105
    },
    {
      "epoch": 0.59112548828125,
      "grad_norm": 11.967348098754883,
      "learning_rate": 1.797175574371082e-06,
      "loss": 3.1494,
      "step": 58110
    },
    {
      "epoch": 0.5911763509114584,
      "grad_norm": 10.870805740356445,
      "learning_rate": 1.7967918297136572e-06,
      "loss": 3.3598,
      "step": 58115
    },
    {
      "epoch": 0.5912272135416666,
      "grad_norm": 14.618064880371094,
      "learning_rate": 1.796408103047646e-06,
      "loss": 3.1634,
      "step": 58120
    },
    {
      "epoch": 0.591278076171875,
      "grad_norm": 13.103531837463379,
      "learning_rate": 1.796024394382867e-06,
      "loss": 3.3876,
      "step": 58125
    },
    {
      "epoch": 0.5913289388020834,
      "grad_norm": 9.35886287689209,
      "learning_rate": 1.7956407037291365e-06,
      "loss": 3.4578,
      "step": 58130
    },
    {
      "epoch": 0.5913798014322916,
      "grad_norm": 7.379724979400635,
      "learning_rate": 1.795257031096272e-06,
      "loss": 3.3323,
      "step": 58135
    },
    {
      "epoch": 0.5914306640625,
      "grad_norm": 9.186336517333984,
      "learning_rate": 1.7948733764940895e-06,
      "loss": 3.2934,
      "step": 58140
    },
    {
      "epoch": 0.5914815266927084,
      "grad_norm": 13.264702796936035,
      "learning_rate": 1.7944897399324046e-06,
      "loss": 3.083,
      "step": 58145
    },
    {
      "epoch": 0.5915323893229166,
      "grad_norm": 18.051586151123047,
      "learning_rate": 1.7941061214210315e-06,
      "loss": 3.3851,
      "step": 58150
    },
    {
      "epoch": 0.591583251953125,
      "grad_norm": 13.872904777526855,
      "learning_rate": 1.7937225209697873e-06,
      "loss": 2.9398,
      "step": 58155
    },
    {
      "epoch": 0.5916341145833334,
      "grad_norm": 13.26317024230957,
      "learning_rate": 1.7933389385884842e-06,
      "loss": 3.2565,
      "step": 58160
    },
    {
      "epoch": 0.5916849772135416,
      "grad_norm": 16.90981101989746,
      "learning_rate": 1.7929553742869376e-06,
      "loss": 3.3974,
      "step": 58165
    },
    {
      "epoch": 0.59173583984375,
      "grad_norm": 15.267029762268066,
      "learning_rate": 1.7925718280749587e-06,
      "loss": 3.343,
      "step": 58170
    },
    {
      "epoch": 0.5917867024739584,
      "grad_norm": 11.082977294921875,
      "learning_rate": 1.792188299962363e-06,
      "loss": 2.9795,
      "step": 58175
    },
    {
      "epoch": 0.5918375651041666,
      "grad_norm": 9.177314758300781,
      "learning_rate": 1.7918047899589623e-06,
      "loss": 3.4705,
      "step": 58180
    },
    {
      "epoch": 0.591888427734375,
      "grad_norm": 14.937149047851562,
      "learning_rate": 1.7914212980745667e-06,
      "loss": 3.7697,
      "step": 58185
    },
    {
      "epoch": 0.5919392903645834,
      "grad_norm": 10.546945571899414,
      "learning_rate": 1.7910378243189901e-06,
      "loss": 3.4776,
      "step": 58190
    },
    {
      "epoch": 0.5919901529947916,
      "grad_norm": 7.730606555938721,
      "learning_rate": 1.7906543687020433e-06,
      "loss": 3.4102,
      "step": 58195
    },
    {
      "epoch": 0.592041015625,
      "grad_norm": 12.9700345993042,
      "learning_rate": 1.7902709312335356e-06,
      "loss": 3.1313,
      "step": 58200
    },
    {
      "epoch": 0.5920918782552084,
      "grad_norm": 8.473440170288086,
      "learning_rate": 1.7898875119232775e-06,
      "loss": 3.4309,
      "step": 58205
    },
    {
      "epoch": 0.5921427408854166,
      "grad_norm": 11.529534339904785,
      "learning_rate": 1.78950411078108e-06,
      "loss": 3.8647,
      "step": 58210
    },
    {
      "epoch": 0.592193603515625,
      "grad_norm": 15.359539031982422,
      "learning_rate": 1.7891207278167512e-06,
      "loss": 3.3153,
      "step": 58215
    },
    {
      "epoch": 0.5922444661458334,
      "grad_norm": 16.164695739746094,
      "learning_rate": 1.7887373630401007e-06,
      "loss": 3.2186,
      "step": 58220
    },
    {
      "epoch": 0.5922953287760416,
      "grad_norm": 7.963503360748291,
      "learning_rate": 1.7883540164609347e-06,
      "loss": 3.54,
      "step": 58225
    },
    {
      "epoch": 0.59234619140625,
      "grad_norm": 13.482510566711426,
      "learning_rate": 1.787970688089063e-06,
      "loss": 3.3673,
      "step": 58230
    },
    {
      "epoch": 0.5923970540364584,
      "grad_norm": 12.424481391906738,
      "learning_rate": 1.7875873779342933e-06,
      "loss": 3.7461,
      "step": 58235
    },
    {
      "epoch": 0.5924479166666666,
      "grad_norm": 21.41114616394043,
      "learning_rate": 1.7872040860064311e-06,
      "loss": 3.9624,
      "step": 58240
    },
    {
      "epoch": 0.592498779296875,
      "grad_norm": 14.386809349060059,
      "learning_rate": 1.7868208123152827e-06,
      "loss": 3.4405,
      "step": 58245
    },
    {
      "epoch": 0.5925496419270834,
      "grad_norm": 12.011004447937012,
      "learning_rate": 1.7864375568706555e-06,
      "loss": 3.0951,
      "step": 58250
    },
    {
      "epoch": 0.5926005045572916,
      "grad_norm": 12.4212646484375,
      "learning_rate": 1.7860543196823548e-06,
      "loss": 3.3755,
      "step": 58255
    },
    {
      "epoch": 0.5926513671875,
      "grad_norm": 13.907829284667969,
      "learning_rate": 1.7856711007601846e-06,
      "loss": 3.2868,
      "step": 58260
    },
    {
      "epoch": 0.5927022298177084,
      "grad_norm": 11.430810928344727,
      "learning_rate": 1.7852879001139493e-06,
      "loss": 3.2822,
      "step": 58265
    },
    {
      "epoch": 0.5927530924479166,
      "grad_norm": 11.203187942504883,
      "learning_rate": 1.7849047177534546e-06,
      "loss": 3.7246,
      "step": 58270
    },
    {
      "epoch": 0.592803955078125,
      "grad_norm": 11.078207015991211,
      "learning_rate": 1.784521553688503e-06,
      "loss": 3.2157,
      "step": 58275
    },
    {
      "epoch": 0.5928548177083334,
      "grad_norm": 14.518149375915527,
      "learning_rate": 1.784138407928897e-06,
      "loss": 3.5838,
      "step": 58280
    },
    {
      "epoch": 0.5929056803385416,
      "grad_norm": 13.477997779846191,
      "learning_rate": 1.7837552804844416e-06,
      "loss": 3.2663,
      "step": 58285
    },
    {
      "epoch": 0.59295654296875,
      "grad_norm": 14.6279296875,
      "learning_rate": 1.7833721713649368e-06,
      "loss": 3.341,
      "step": 58290
    },
    {
      "epoch": 0.5930074055989584,
      "grad_norm": 16.762426376342773,
      "learning_rate": 1.782989080580186e-06,
      "loss": 3.3536,
      "step": 58295
    },
    {
      "epoch": 0.5930582682291666,
      "grad_norm": 9.366729736328125,
      "learning_rate": 1.7826060081399882e-06,
      "loss": 3.3015,
      "step": 58300
    },
    {
      "epoch": 0.593109130859375,
      "grad_norm": 12.638140678405762,
      "learning_rate": 1.782222954054146e-06,
      "loss": 3.3911,
      "step": 58305
    },
    {
      "epoch": 0.5931599934895834,
      "grad_norm": 12.419198989868164,
      "learning_rate": 1.7818399183324602e-06,
      "loss": 3.2025,
      "step": 58310
    },
    {
      "epoch": 0.5932108561197916,
      "grad_norm": 13.538446426391602,
      "learning_rate": 1.7814569009847293e-06,
      "loss": 3.3975,
      "step": 58315
    },
    {
      "epoch": 0.59326171875,
      "grad_norm": 9.20753288269043,
      "learning_rate": 1.781073902020753e-06,
      "loss": 3.3517,
      "step": 58320
    },
    {
      "epoch": 0.5933125813802084,
      "grad_norm": 12.255483627319336,
      "learning_rate": 1.780690921450331e-06,
      "loss": 3.1009,
      "step": 58325
    },
    {
      "epoch": 0.5933634440104166,
      "grad_norm": 12.996417999267578,
      "learning_rate": 1.7803079592832613e-06,
      "loss": 3.3092,
      "step": 58330
    },
    {
      "epoch": 0.593414306640625,
      "grad_norm": 10.153072357177734,
      "learning_rate": 1.7799250155293421e-06,
      "loss": 3.3059,
      "step": 58335
    },
    {
      "epoch": 0.5934651692708334,
      "grad_norm": 9.475560188293457,
      "learning_rate": 1.7795420901983695e-06,
      "loss": 3.3699,
      "step": 58340
    },
    {
      "epoch": 0.5935160319010416,
      "grad_norm": 13.278281211853027,
      "learning_rate": 1.7791591833001426e-06,
      "loss": 3.1001,
      "step": 58345
    },
    {
      "epoch": 0.59356689453125,
      "grad_norm": 11.277542114257812,
      "learning_rate": 1.7787762948444575e-06,
      "loss": 3.0136,
      "step": 58350
    },
    {
      "epoch": 0.5936177571614584,
      "grad_norm": 9.570847511291504,
      "learning_rate": 1.7783934248411097e-06,
      "loss": 3.2451,
      "step": 58355
    },
    {
      "epoch": 0.5936686197916666,
      "grad_norm": 14.030230522155762,
      "learning_rate": 1.7780105732998942e-06,
      "loss": 3.2859,
      "step": 58360
    },
    {
      "epoch": 0.593719482421875,
      "grad_norm": 7.317997932434082,
      "learning_rate": 1.7776277402306085e-06,
      "loss": 3.2868,
      "step": 58365
    },
    {
      "epoch": 0.5937703450520834,
      "grad_norm": 12.152804374694824,
      "learning_rate": 1.777244925643045e-06,
      "loss": 3.1848,
      "step": 58370
    },
    {
      "epoch": 0.5938212076822916,
      "grad_norm": 7.249147891998291,
      "learning_rate": 1.7768621295469985e-06,
      "loss": 3.2292,
      "step": 58375
    },
    {
      "epoch": 0.5938720703125,
      "grad_norm": 12.727730751037598,
      "learning_rate": 1.7764793519522639e-06,
      "loss": 3.2557,
      "step": 58380
    },
    {
      "epoch": 0.5939229329427084,
      "grad_norm": 10.969575881958008,
      "learning_rate": 1.776096592868633e-06,
      "loss": 3.5441,
      "step": 58385
    },
    {
      "epoch": 0.5939737955729166,
      "grad_norm": 8.714305877685547,
      "learning_rate": 1.7757138523058999e-06,
      "loss": 3.072,
      "step": 58390
    },
    {
      "epoch": 0.594024658203125,
      "grad_norm": 12.414077758789062,
      "learning_rate": 1.775331130273854e-06,
      "loss": 3.6681,
      "step": 58395
    },
    {
      "epoch": 0.5940755208333334,
      "grad_norm": 13.802921295166016,
      "learning_rate": 1.7749484267822917e-06,
      "loss": 3.3932,
      "step": 58400
    },
    {
      "epoch": 0.5941263834635416,
      "grad_norm": 12.4661226272583,
      "learning_rate": 1.774565741841001e-06,
      "loss": 3.008,
      "step": 58405
    },
    {
      "epoch": 0.59417724609375,
      "grad_norm": 8.011397361755371,
      "learning_rate": 1.774183075459775e-06,
      "loss": 3.5852,
      "step": 58410
    },
    {
      "epoch": 0.5942281087239584,
      "grad_norm": 9.829503059387207,
      "learning_rate": 1.7738004276484014e-06,
      "loss": 3.2902,
      "step": 58415
    },
    {
      "epoch": 0.5942789713541666,
      "grad_norm": 8.826567649841309,
      "learning_rate": 1.7734177984166722e-06,
      "loss": 3.3655,
      "step": 58420
    },
    {
      "epoch": 0.594329833984375,
      "grad_norm": 14.86225414276123,
      "learning_rate": 1.7730351877743773e-06,
      "loss": 3.5861,
      "step": 58425
    },
    {
      "epoch": 0.5943806966145834,
      "grad_norm": 18.58818244934082,
      "learning_rate": 1.772652595731304e-06,
      "loss": 3.9893,
      "step": 58430
    },
    {
      "epoch": 0.5944315592447916,
      "grad_norm": 16.581823348999023,
      "learning_rate": 1.7722700222972413e-06,
      "loss": 3.4196,
      "step": 58435
    },
    {
      "epoch": 0.594482421875,
      "grad_norm": 7.3273115158081055,
      "learning_rate": 1.7718874674819786e-06,
      "loss": 3.3065,
      "step": 58440
    },
    {
      "epoch": 0.5945332845052084,
      "grad_norm": 11.707185745239258,
      "learning_rate": 1.771504931295302e-06,
      "loss": 3.5165,
      "step": 58445
    },
    {
      "epoch": 0.5945841471354166,
      "grad_norm": 17.30613899230957,
      "learning_rate": 1.7711224137469996e-06,
      "loss": 3.2353,
      "step": 58450
    },
    {
      "epoch": 0.594635009765625,
      "grad_norm": 13.378973960876465,
      "learning_rate": 1.7707399148468563e-06,
      "loss": 3.0759,
      "step": 58455
    },
    {
      "epoch": 0.5946858723958334,
      "grad_norm": 7.2831621170043945,
      "learning_rate": 1.7703574346046598e-06,
      "loss": 3.0799,
      "step": 58460
    },
    {
      "epoch": 0.5947367350260416,
      "grad_norm": 10.333426475524902,
      "learning_rate": 1.7699749730301962e-06,
      "loss": 3.1861,
      "step": 58465
    },
    {
      "epoch": 0.59478759765625,
      "grad_norm": 6.475285530090332,
      "learning_rate": 1.7695925301332495e-06,
      "loss": 3.4412,
      "step": 58470
    },
    {
      "epoch": 0.5948384602864584,
      "grad_norm": 15.218671798706055,
      "learning_rate": 1.769210105923604e-06,
      "loss": 3.0268,
      "step": 58475
    },
    {
      "epoch": 0.5948893229166666,
      "grad_norm": 11.751965522766113,
      "learning_rate": 1.7688277004110458e-06,
      "loss": 2.8304,
      "step": 58480
    },
    {
      "epoch": 0.594940185546875,
      "grad_norm": 9.582796096801758,
      "learning_rate": 1.768445313605357e-06,
      "loss": 3.4982,
      "step": 58485
    },
    {
      "epoch": 0.5949910481770834,
      "grad_norm": 14.962919235229492,
      "learning_rate": 1.768062945516321e-06,
      "loss": 3.5315,
      "step": 58490
    },
    {
      "epoch": 0.5950419108072916,
      "grad_norm": 10.981086730957031,
      "learning_rate": 1.7676805961537218e-06,
      "loss": 3.3454,
      "step": 58495
    },
    {
      "epoch": 0.5950927734375,
      "grad_norm": 11.686980247497559,
      "learning_rate": 1.7672982655273406e-06,
      "loss": 3.2746,
      "step": 58500
    },
    {
      "epoch": 0.5951436360677084,
      "grad_norm": 11.312081336975098,
      "learning_rate": 1.7669159536469605e-06,
      "loss": 4.051,
      "step": 58505
    },
    {
      "epoch": 0.5951944986979166,
      "grad_norm": 14.038021087646484,
      "learning_rate": 1.766533660522361e-06,
      "loss": 3.3311,
      "step": 58510
    },
    {
      "epoch": 0.595245361328125,
      "grad_norm": 16.62759780883789,
      "learning_rate": 1.766151386163324e-06,
      "loss": 3.4934,
      "step": 58515
    },
    {
      "epoch": 0.5952962239583334,
      "grad_norm": 11.089447021484375,
      "learning_rate": 1.7657691305796303e-06,
      "loss": 3.3098,
      "step": 58520
    },
    {
      "epoch": 0.5953470865885416,
      "grad_norm": 9.269929885864258,
      "learning_rate": 1.7653868937810593e-06,
      "loss": 3.1054,
      "step": 58525
    },
    {
      "epoch": 0.59539794921875,
      "grad_norm": 14.380538940429688,
      "learning_rate": 1.7650046757773899e-06,
      "loss": 2.8936,
      "step": 58530
    },
    {
      "epoch": 0.5954488118489584,
      "grad_norm": 12.806015968322754,
      "learning_rate": 1.7646224765784024e-06,
      "loss": 3.3982,
      "step": 58535
    },
    {
      "epoch": 0.5954996744791666,
      "grad_norm": 9.387150764465332,
      "learning_rate": 1.764240296193874e-06,
      "loss": 3.5869,
      "step": 58540
    },
    {
      "epoch": 0.595550537109375,
      "grad_norm": 10.369173049926758,
      "learning_rate": 1.7638581346335837e-06,
      "loss": 3.205,
      "step": 58545
    },
    {
      "epoch": 0.5956013997395834,
      "grad_norm": 13.991856575012207,
      "learning_rate": 1.7634759919073074e-06,
      "loss": 3.215,
      "step": 58550
    },
    {
      "epoch": 0.5956522623697916,
      "grad_norm": 10.374763488769531,
      "learning_rate": 1.7630938680248243e-06,
      "loss": 3.3059,
      "step": 58555
    },
    {
      "epoch": 0.595703125,
      "grad_norm": 12.562834739685059,
      "learning_rate": 1.76271176299591e-06,
      "loss": 3.0095,
      "step": 58560
    },
    {
      "epoch": 0.5957539876302084,
      "grad_norm": 11.885716438293457,
      "learning_rate": 1.7623296768303406e-06,
      "loss": 3.4797,
      "step": 58565
    },
    {
      "epoch": 0.5958048502604166,
      "grad_norm": 8.960821151733398,
      "learning_rate": 1.7619476095378907e-06,
      "loss": 3.5235,
      "step": 58570
    },
    {
      "epoch": 0.595855712890625,
      "grad_norm": 11.510443687438965,
      "learning_rate": 1.7615655611283366e-06,
      "loss": 3.3977,
      "step": 58575
    },
    {
      "epoch": 0.5959065755208334,
      "grad_norm": 15.014580726623535,
      "learning_rate": 1.7611835316114534e-06,
      "loss": 3.2362,
      "step": 58580
    },
    {
      "epoch": 0.5959574381510416,
      "grad_norm": 10.226090431213379,
      "learning_rate": 1.760801520997013e-06,
      "loss": 3.1782,
      "step": 58585
    },
    {
      "epoch": 0.59600830078125,
      "grad_norm": 11.057794570922852,
      "learning_rate": 1.7604195292947915e-06,
      "loss": 3.224,
      "step": 58590
    },
    {
      "epoch": 0.5960591634114584,
      "grad_norm": 7.222981929779053,
      "learning_rate": 1.760037556514561e-06,
      "loss": 3.656,
      "step": 58595
    },
    {
      "epoch": 0.5961100260416666,
      "grad_norm": 9.445796966552734,
      "learning_rate": 1.7596556026660944e-06,
      "loss": 3.1746,
      "step": 58600
    },
    {
      "epoch": 0.596160888671875,
      "grad_norm": 14.64661979675293,
      "learning_rate": 1.7592736677591627e-06,
      "loss": 3.1428,
      "step": 58605
    },
    {
      "epoch": 0.5962117513020834,
      "grad_norm": 11.342286109924316,
      "learning_rate": 1.7588917518035398e-06,
      "loss": 3.1805,
      "step": 58610
    },
    {
      "epoch": 0.5962626139322916,
      "grad_norm": 14.933340072631836,
      "learning_rate": 1.7585098548089952e-06,
      "loss": 3.0741,
      "step": 58615
    },
    {
      "epoch": 0.5963134765625,
      "grad_norm": 15.432071685791016,
      "learning_rate": 1.7581279767853008e-06,
      "loss": 3.2039,
      "step": 58620
    },
    {
      "epoch": 0.5963643391927084,
      "grad_norm": 9.950495719909668,
      "learning_rate": 1.7577461177422251e-06,
      "loss": 3.5878,
      "step": 58625
    },
    {
      "epoch": 0.5964152018229166,
      "grad_norm": 8.13646411895752,
      "learning_rate": 1.7573642776895397e-06,
      "loss": 3.3971,
      "step": 58630
    },
    {
      "epoch": 0.596466064453125,
      "grad_norm": 12.200660705566406,
      "learning_rate": 1.7569824566370136e-06,
      "loss": 3.2983,
      "step": 58635
    },
    {
      "epoch": 0.5965169270833334,
      "grad_norm": 14.329954147338867,
      "learning_rate": 1.7566006545944148e-06,
      "loss": 3.2166,
      "step": 58640
    },
    {
      "epoch": 0.5965677897135416,
      "grad_norm": 10.474040031433105,
      "learning_rate": 1.7562188715715117e-06,
      "loss": 3.3249,
      "step": 58645
    },
    {
      "epoch": 0.59661865234375,
      "grad_norm": 8.579286575317383,
      "learning_rate": 1.7558371075780733e-06,
      "loss": 3.5359,
      "step": 58650
    },
    {
      "epoch": 0.5966695149739584,
      "grad_norm": 10.491554260253906,
      "learning_rate": 1.7554553626238656e-06,
      "loss": 3.0442,
      "step": 58655
    },
    {
      "epoch": 0.5967203776041666,
      "grad_norm": 16.097248077392578,
      "learning_rate": 1.7550736367186566e-06,
      "loss": 3.2724,
      "step": 58660
    },
    {
      "epoch": 0.596771240234375,
      "grad_norm": 11.073954582214355,
      "learning_rate": 1.754691929872211e-06,
      "loss": 3.2875,
      "step": 58665
    },
    {
      "epoch": 0.5968221028645834,
      "grad_norm": 16.58736228942871,
      "learning_rate": 1.7543102420942961e-06,
      "loss": 3.5692,
      "step": 58670
    },
    {
      "epoch": 0.5968729654947916,
      "grad_norm": 13.786785125732422,
      "learning_rate": 1.7539285733946776e-06,
      "loss": 3.7272,
      "step": 58675
    },
    {
      "epoch": 0.596923828125,
      "grad_norm": 8.079156875610352,
      "learning_rate": 1.7535469237831185e-06,
      "loss": 4.6149,
      "step": 58680
    },
    {
      "epoch": 0.5969746907552084,
      "grad_norm": 14.249360084533691,
      "learning_rate": 1.7531652932693848e-06,
      "loss": 3.1452,
      "step": 58685
    },
    {
      "epoch": 0.5970255533854166,
      "grad_norm": 11.673576354980469,
      "learning_rate": 1.752783681863241e-06,
      "loss": 3.3092,
      "step": 58690
    },
    {
      "epoch": 0.597076416015625,
      "grad_norm": 13.817118644714355,
      "learning_rate": 1.752402089574449e-06,
      "loss": 3.1523,
      "step": 58695
    },
    {
      "epoch": 0.5971272786458334,
      "grad_norm": 15.398924827575684,
      "learning_rate": 1.7520205164127719e-06,
      "loss": 3.3453,
      "step": 58700
    },
    {
      "epoch": 0.5971781412760416,
      "grad_norm": 12.843722343444824,
      "learning_rate": 1.7516389623879733e-06,
      "loss": 3.78,
      "step": 58705
    },
    {
      "epoch": 0.59722900390625,
      "grad_norm": 10.914924621582031,
      "learning_rate": 1.7512574275098138e-06,
      "loss": 3.0658,
      "step": 58710
    },
    {
      "epoch": 0.5972798665364584,
      "grad_norm": 13.276107788085938,
      "learning_rate": 1.7508759117880564e-06,
      "loss": 3.3682,
      "step": 58715
    },
    {
      "epoch": 0.5973307291666666,
      "grad_norm": 7.376313209533691,
      "learning_rate": 1.75049441523246e-06,
      "loss": 3.1761,
      "step": 58720
    },
    {
      "epoch": 0.597381591796875,
      "grad_norm": 12.459508895874023,
      "learning_rate": 1.750112937852787e-06,
      "loss": 3.7262,
      "step": 58725
    },
    {
      "epoch": 0.5974324544270834,
      "grad_norm": 6.936720848083496,
      "learning_rate": 1.749731479658797e-06,
      "loss": 3.2802,
      "step": 58730
    },
    {
      "epoch": 0.5974833170572916,
      "grad_norm": 13.150242805480957,
      "learning_rate": 1.7493500406602492e-06,
      "loss": 3.2631,
      "step": 58735
    },
    {
      "epoch": 0.5975341796875,
      "grad_norm": 8.150609970092773,
      "learning_rate": 1.7489686208669022e-06,
      "loss": 3.1906,
      "step": 58740
    },
    {
      "epoch": 0.5975850423177084,
      "grad_norm": 12.01266860961914,
      "learning_rate": 1.7485872202885152e-06,
      "loss": 3.2785,
      "step": 58745
    },
    {
      "epoch": 0.5976359049479166,
      "grad_norm": 9.509927749633789,
      "learning_rate": 1.7482058389348466e-06,
      "loss": 3.0776,
      "step": 58750
    },
    {
      "epoch": 0.597686767578125,
      "grad_norm": 10.995317459106445,
      "learning_rate": 1.7478244768156532e-06,
      "loss": 3.038,
      "step": 58755
    },
    {
      "epoch": 0.5977376302083334,
      "grad_norm": 15.866677284240723,
      "learning_rate": 1.7474431339406916e-06,
      "loss": 3.3283,
      "step": 58760
    },
    {
      "epoch": 0.5977884928385416,
      "grad_norm": 10.797828674316406,
      "learning_rate": 1.7470618103197201e-06,
      "loss": 3.4482,
      "step": 58765
    },
    {
      "epoch": 0.59783935546875,
      "grad_norm": 10.114940643310547,
      "learning_rate": 1.7466805059624931e-06,
      "loss": 3.336,
      "step": 58770
    },
    {
      "epoch": 0.5978902180989584,
      "grad_norm": 10.19825553894043,
      "learning_rate": 1.746299220878766e-06,
      "loss": 3.0855,
      "step": 58775
    },
    {
      "epoch": 0.5979410807291666,
      "grad_norm": 15.008298873901367,
      "learning_rate": 1.7459179550782962e-06,
      "loss": 3.244,
      "step": 58780
    },
    {
      "epoch": 0.597991943359375,
      "grad_norm": 13.538183212280273,
      "learning_rate": 1.745536708570836e-06,
      "loss": 3.6771,
      "step": 58785
    },
    {
      "epoch": 0.5980428059895834,
      "grad_norm": 11.229158401489258,
      "learning_rate": 1.7451554813661408e-06,
      "loss": 3.2748,
      "step": 58790
    },
    {
      "epoch": 0.5980936686197916,
      "grad_norm": 12.117464065551758,
      "learning_rate": 1.7447742734739628e-06,
      "loss": 3.3054,
      "step": 58795
    },
    {
      "epoch": 0.59814453125,
      "grad_norm": 8.106527328491211,
      "learning_rate": 1.7443930849040564e-06,
      "loss": 3.099,
      "step": 58800
    },
    {
      "epoch": 0.5981953938802084,
      "grad_norm": 17.27057456970215,
      "learning_rate": 1.7440119156661745e-06,
      "loss": 3.2644,
      "step": 58805
    },
    {
      "epoch": 0.5982462565104166,
      "grad_norm": 15.627812385559082,
      "learning_rate": 1.743630765770068e-06,
      "loss": 3.1907,
      "step": 58810
    },
    {
      "epoch": 0.598297119140625,
      "grad_norm": 14.384893417358398,
      "learning_rate": 1.7432496352254885e-06,
      "loss": 3.2499,
      "step": 58815
    },
    {
      "epoch": 0.5983479817708334,
      "grad_norm": 15.350054740905762,
      "learning_rate": 1.7428685240421888e-06,
      "loss": 2.977,
      "step": 58820
    },
    {
      "epoch": 0.5983988444010416,
      "grad_norm": 12.908646583557129,
      "learning_rate": 1.742487432229918e-06,
      "loss": 3.2672,
      "step": 58825
    },
    {
      "epoch": 0.59844970703125,
      "grad_norm": 16.558061599731445,
      "learning_rate": 1.742106359798427e-06,
      "loss": 3.231,
      "step": 58830
    },
    {
      "epoch": 0.5985005696614584,
      "grad_norm": 11.36466121673584,
      "learning_rate": 1.7417253067574642e-06,
      "loss": 3.5004,
      "step": 58835
    },
    {
      "epoch": 0.5985514322916666,
      "grad_norm": 16.1599178314209,
      "learning_rate": 1.7413442731167802e-06,
      "loss": 3.2495,
      "step": 58840
    },
    {
      "epoch": 0.598602294921875,
      "grad_norm": 14.043704986572266,
      "learning_rate": 1.7409632588861236e-06,
      "loss": 3.5283,
      "step": 58845
    },
    {
      "epoch": 0.5986531575520834,
      "grad_norm": 15.785425186157227,
      "learning_rate": 1.7405822640752419e-06,
      "loss": 3.4809,
      "step": 58850
    },
    {
      "epoch": 0.5987040201822916,
      "grad_norm": 16.065847396850586,
      "learning_rate": 1.7402012886938819e-06,
      "loss": 3.4675,
      "step": 58855
    },
    {
      "epoch": 0.5987548828125,
      "grad_norm": 13.364585876464844,
      "learning_rate": 1.739820332751793e-06,
      "loss": 3.2538,
      "step": 58860
    },
    {
      "epoch": 0.5988057454427084,
      "grad_norm": 11.64784049987793,
      "learning_rate": 1.739439396258721e-06,
      "loss": 3.1564,
      "step": 58865
    },
    {
      "epoch": 0.5988566080729166,
      "grad_norm": 10.144237518310547,
      "learning_rate": 1.7390584792244102e-06,
      "loss": 3.3792,
      "step": 58870
    },
    {
      "epoch": 0.598907470703125,
      "grad_norm": 14.06118392944336,
      "learning_rate": 1.7386775816586088e-06,
      "loss": 3.2614,
      "step": 58875
    },
    {
      "epoch": 0.5989583333333334,
      "grad_norm": 10.689396858215332,
      "learning_rate": 1.738296703571062e-06,
      "loss": 3.1,
      "step": 58880
    },
    {
      "epoch": 0.5990091959635416,
      "grad_norm": 14.174532890319824,
      "learning_rate": 1.7379158449715123e-06,
      "loss": 3.6983,
      "step": 58885
    },
    {
      "epoch": 0.59906005859375,
      "grad_norm": 15.426204681396484,
      "learning_rate": 1.737535005869705e-06,
      "loss": 3.5526,
      "step": 58890
    },
    {
      "epoch": 0.5991109212239584,
      "grad_norm": 14.208455085754395,
      "learning_rate": 1.7371541862753843e-06,
      "loss": 3.8607,
      "step": 58895
    },
    {
      "epoch": 0.5991617838541666,
      "grad_norm": 11.174375534057617,
      "learning_rate": 1.7367733861982928e-06,
      "loss": 3.4787,
      "step": 58900
    },
    {
      "epoch": 0.599212646484375,
      "grad_norm": 11.884801864624023,
      "learning_rate": 1.736392605648174e-06,
      "loss": 3.0598,
      "step": 58905
    },
    {
      "epoch": 0.5992635091145834,
      "grad_norm": 6.930401802062988,
      "learning_rate": 1.736011844634768e-06,
      "loss": 2.9284,
      "step": 58910
    },
    {
      "epoch": 0.5993143717447916,
      "grad_norm": 10.723294258117676,
      "learning_rate": 1.7356311031678186e-06,
      "loss": 3.403,
      "step": 58915
    },
    {
      "epoch": 0.599365234375,
      "grad_norm": 12.299421310424805,
      "learning_rate": 1.7352503812570667e-06,
      "loss": 4.1806,
      "step": 58920
    },
    {
      "epoch": 0.5994160970052084,
      "grad_norm": 10.311161994934082,
      "learning_rate": 1.7348696789122522e-06,
      "loss": 2.729,
      "step": 58925
    },
    {
      "epoch": 0.5994669596354166,
      "grad_norm": 7.91664981842041,
      "learning_rate": 1.7344889961431149e-06,
      "loss": 3.3468,
      "step": 58930
    },
    {
      "epoch": 0.599517822265625,
      "grad_norm": 13.23727798461914,
      "learning_rate": 1.734108332959396e-06,
      "loss": 3.2103,
      "step": 58935
    },
    {
      "epoch": 0.5995686848958334,
      "grad_norm": 13.324912071228027,
      "learning_rate": 1.7337276893708338e-06,
      "loss": 3.4046,
      "step": 58940
    },
    {
      "epoch": 0.5996195475260416,
      "grad_norm": 14.138757705688477,
      "learning_rate": 1.7333470653871673e-06,
      "loss": 2.9858,
      "step": 58945
    },
    {
      "epoch": 0.59967041015625,
      "grad_norm": 9.890541076660156,
      "learning_rate": 1.7329664610181335e-06,
      "loss": 3.2769,
      "step": 58950
    },
    {
      "epoch": 0.5997212727864584,
      "grad_norm": 15.262154579162598,
      "learning_rate": 1.7325858762734715e-06,
      "loss": 3.7433,
      "step": 58955
    },
    {
      "epoch": 0.5997721354166666,
      "grad_norm": 13.031238555908203,
      "learning_rate": 1.7322053111629183e-06,
      "loss": 3.5493,
      "step": 58960
    },
    {
      "epoch": 0.599822998046875,
      "grad_norm": 12.975473403930664,
      "learning_rate": 1.7318247656962092e-06,
      "loss": 3.1998,
      "step": 58965
    },
    {
      "epoch": 0.5998738606770834,
      "grad_norm": 15.250924110412598,
      "learning_rate": 1.7314442398830822e-06,
      "loss": 3.7708,
      "step": 58970
    },
    {
      "epoch": 0.5999247233072916,
      "grad_norm": 16.053945541381836,
      "learning_rate": 1.7310637337332725e-06,
      "loss": 3.3696,
      "step": 58975
    },
    {
      "epoch": 0.5999755859375,
      "grad_norm": 13.52525806427002,
      "learning_rate": 1.7306832472565143e-06,
      "loss": 3.1525,
      "step": 58980
    },
    {
      "epoch": 0.6000264485677084,
      "grad_norm": 11.047719955444336,
      "learning_rate": 1.7303027804625426e-06,
      "loss": 2.9954,
      "step": 58985
    },
    {
      "epoch": 0.6000773111979166,
      "grad_norm": 8.911846160888672,
      "learning_rate": 1.7299223333610926e-06,
      "loss": 3.5218,
      "step": 58990
    },
    {
      "epoch": 0.600128173828125,
      "grad_norm": 11.748200416564941,
      "learning_rate": 1.7295419059618967e-06,
      "loss": 3.2972,
      "step": 58995
    },
    {
      "epoch": 0.6001790364583334,
      "grad_norm": 8.499863624572754,
      "learning_rate": 1.729161498274689e-06,
      "loss": 3.4798,
      "step": 59000
    },
    {
      "epoch": 0.6002298990885416,
      "grad_norm": 8.420709609985352,
      "learning_rate": 1.7287811103092006e-06,
      "loss": 3.5517,
      "step": 59005
    },
    {
      "epoch": 0.60028076171875,
      "grad_norm": 16.08116340637207,
      "learning_rate": 1.7284007420751653e-06,
      "loss": 3.7896,
      "step": 59010
    },
    {
      "epoch": 0.6003316243489584,
      "grad_norm": 9.745408058166504,
      "learning_rate": 1.7280203935823148e-06,
      "loss": 3.6144,
      "step": 59015
    },
    {
      "epoch": 0.6003824869791666,
      "grad_norm": 7.625930309295654,
      "learning_rate": 1.7276400648403791e-06,
      "loss": 3.4174,
      "step": 59020
    },
    {
      "epoch": 0.600433349609375,
      "grad_norm": 10.471306800842285,
      "learning_rate": 1.7272597558590883e-06,
      "loss": 3.1803,
      "step": 59025
    },
    {
      "epoch": 0.6004842122395834,
      "grad_norm": 12.285701751708984,
      "learning_rate": 1.7268794666481742e-06,
      "loss": 3.3564,
      "step": 59030
    },
    {
      "epoch": 0.6005350748697916,
      "grad_norm": 11.852290153503418,
      "learning_rate": 1.7264991972173665e-06,
      "loss": 3.1323,
      "step": 59035
    },
    {
      "epoch": 0.6005859375,
      "grad_norm": 9.487098693847656,
      "learning_rate": 1.7261189475763929e-06,
      "loss": 3.249,
      "step": 59040
    },
    {
      "epoch": 0.6006368001302084,
      "grad_norm": 8.560872077941895,
      "learning_rate": 1.7257387177349815e-06,
      "loss": 3.7592,
      "step": 59045
    },
    {
      "epoch": 0.6006876627604166,
      "grad_norm": 13.85609245300293,
      "learning_rate": 1.725358507702863e-06,
      "loss": 3.0362,
      "step": 59050
    },
    {
      "epoch": 0.600738525390625,
      "grad_norm": 11.601663589477539,
      "learning_rate": 1.724978317489763e-06,
      "loss": 3.2711,
      "step": 59055
    },
    {
      "epoch": 0.6007893880208334,
      "grad_norm": 10.587480545043945,
      "learning_rate": 1.7245981471054096e-06,
      "loss": 3.4282,
      "step": 59060
    },
    {
      "epoch": 0.6008402506510416,
      "grad_norm": 14.014335632324219,
      "learning_rate": 1.7242179965595273e-06,
      "loss": 3.5206,
      "step": 59065
    },
    {
      "epoch": 0.60089111328125,
      "grad_norm": 13.849995613098145,
      "learning_rate": 1.723837865861845e-06,
      "loss": 3.8606,
      "step": 59070
    },
    {
      "epoch": 0.6009419759114584,
      "grad_norm": 12.70814037322998,
      "learning_rate": 1.7234577550220869e-06,
      "loss": 3.443,
      "step": 59075
    },
    {
      "epoch": 0.6009928385416666,
      "grad_norm": 16.227697372436523,
      "learning_rate": 1.723077664049977e-06,
      "loss": 3.3012,
      "step": 59080
    },
    {
      "epoch": 0.601043701171875,
      "grad_norm": 14.024761199951172,
      "learning_rate": 1.7226975929552414e-06,
      "loss": 3.3245,
      "step": 59085
    },
    {
      "epoch": 0.6010945638020834,
      "grad_norm": 8.826449394226074,
      "learning_rate": 1.7223175417476046e-06,
      "loss": 3.0356,
      "step": 59090
    },
    {
      "epoch": 0.6011454264322916,
      "grad_norm": 14.546638488769531,
      "learning_rate": 1.7219375104367883e-06,
      "loss": 3.1104,
      "step": 59095
    },
    {
      "epoch": 0.6011962890625,
      "grad_norm": 9.69225788116455,
      "learning_rate": 1.721557499032516e-06,
      "loss": 3.8635,
      "step": 59100
    },
    {
      "epoch": 0.6012471516927084,
      "grad_norm": 10.164959907531738,
      "learning_rate": 1.7211775075445115e-06,
      "loss": 3.1768,
      "step": 59105
    },
    {
      "epoch": 0.6012980143229166,
      "grad_norm": 18.523700714111328,
      "learning_rate": 1.7207975359824953e-06,
      "loss": 3.2138,
      "step": 59110
    },
    {
      "epoch": 0.601348876953125,
      "grad_norm": 10.2969331741333,
      "learning_rate": 1.7204175843561902e-06,
      "loss": 3.3226,
      "step": 59115
    },
    {
      "epoch": 0.6013997395833334,
      "grad_norm": 9.768226623535156,
      "learning_rate": 1.720037652675315e-06,
      "loss": 3.0602,
      "step": 59120
    },
    {
      "epoch": 0.6014506022135416,
      "grad_norm": 10.433574676513672,
      "learning_rate": 1.7196577409495924e-06,
      "loss": 3.1022,
      "step": 59125
    },
    {
      "epoch": 0.60150146484375,
      "grad_norm": 11.451906204223633,
      "learning_rate": 1.719277849188742e-06,
      "loss": 3.4766,
      "step": 59130
    },
    {
      "epoch": 0.6015523274739584,
      "grad_norm": 9.445000648498535,
      "learning_rate": 1.7188979774024825e-06,
      "loss": 3.3239,
      "step": 59135
    },
    {
      "epoch": 0.6016031901041666,
      "grad_norm": 12.806665420532227,
      "learning_rate": 1.7185181256005323e-06,
      "loss": 3.0114,
      "step": 59140
    },
    {
      "epoch": 0.601654052734375,
      "grad_norm": 16.7288761138916,
      "learning_rate": 1.7181382937926117e-06,
      "loss": 3.1627,
      "step": 59145
    },
    {
      "epoch": 0.6017049153645834,
      "grad_norm": 14.747552871704102,
      "learning_rate": 1.7177584819884368e-06,
      "loss": 3.3099,
      "step": 59150
    },
    {
      "epoch": 0.6017557779947916,
      "grad_norm": 15.10758113861084,
      "learning_rate": 1.7173786901977268e-06,
      "loss": 3.398,
      "step": 59155
    },
    {
      "epoch": 0.601806640625,
      "grad_norm": 16.61138916015625,
      "learning_rate": 1.716998918430196e-06,
      "loss": 2.9033,
      "step": 59160
    },
    {
      "epoch": 0.6018575032552084,
      "grad_norm": 9.121318817138672,
      "learning_rate": 1.7166191666955628e-06,
      "loss": 3.1971,
      "step": 59165
    },
    {
      "epoch": 0.6019083658854166,
      "grad_norm": 13.609682083129883,
      "learning_rate": 1.7162394350035433e-06,
      "loss": 3.3779,
      "step": 59170
    },
    {
      "epoch": 0.601959228515625,
      "grad_norm": 11.46147632598877,
      "learning_rate": 1.7158597233638505e-06,
      "loss": 3.5063,
      "step": 59175
    },
    {
      "epoch": 0.6020100911458334,
      "grad_norm": 13.441067695617676,
      "learning_rate": 1.7154800317862024e-06,
      "loss": 3.4049,
      "step": 59180
    },
    {
      "epoch": 0.6020609537760416,
      "grad_norm": 10.99342155456543,
      "learning_rate": 1.7151003602803113e-06,
      "loss": 3.6103,
      "step": 59185
    },
    {
      "epoch": 0.60211181640625,
      "grad_norm": 11.134747505187988,
      "learning_rate": 1.7147207088558914e-06,
      "loss": 3.2871,
      "step": 59190
    },
    {
      "epoch": 0.6021626790364584,
      "grad_norm": 17.46708869934082,
      "learning_rate": 1.7143410775226554e-06,
      "loss": 3.166,
      "step": 59195
    },
    {
      "epoch": 0.6022135416666666,
      "grad_norm": 12.266764640808105,
      "learning_rate": 1.7139614662903174e-06,
      "loss": 3.3268,
      "step": 59200
    },
    {
      "epoch": 0.602264404296875,
      "grad_norm": 13.35457992553711,
      "learning_rate": 1.7135818751685892e-06,
      "loss": 3.6107,
      "step": 59205
    },
    {
      "epoch": 0.6023152669270834,
      "grad_norm": 12.261884689331055,
      "learning_rate": 1.713202304167182e-06,
      "loss": 3.4055,
      "step": 59210
    },
    {
      "epoch": 0.6023661295572916,
      "grad_norm": 10.646519660949707,
      "learning_rate": 1.7128227532958068e-06,
      "loss": 3.4347,
      "step": 59215
    },
    {
      "epoch": 0.6024169921875,
      "grad_norm": 12.019163131713867,
      "learning_rate": 1.7124432225641758e-06,
      "loss": 3.1606,
      "step": 59220
    },
    {
      "epoch": 0.6024678548177084,
      "grad_norm": 12.81616497039795,
      "learning_rate": 1.7120637119819978e-06,
      "loss": 3.1076,
      "step": 59225
    },
    {
      "epoch": 0.6025187174479166,
      "grad_norm": 8.086276054382324,
      "learning_rate": 1.7116842215589834e-06,
      "loss": 3.4989,
      "step": 59230
    },
    {
      "epoch": 0.602569580078125,
      "grad_norm": 7.3298258781433105,
      "learning_rate": 1.7113047513048404e-06,
      "loss": 3.2682,
      "step": 59235
    },
    {
      "epoch": 0.6026204427083334,
      "grad_norm": 12.518247604370117,
      "learning_rate": 1.7109253012292789e-06,
      "loss": 3.5294,
      "step": 59240
    },
    {
      "epoch": 0.6026713053385416,
      "grad_norm": 11.548209190368652,
      "learning_rate": 1.710545871342007e-06,
      "loss": 3.6109,
      "step": 59245
    },
    {
      "epoch": 0.60272216796875,
      "grad_norm": 14.335244178771973,
      "learning_rate": 1.7101664616527314e-06,
      "loss": 3.3422,
      "step": 59250
    },
    {
      "epoch": 0.6027730305989584,
      "grad_norm": 10.248363494873047,
      "learning_rate": 1.7097870721711593e-06,
      "loss": 3.6576,
      "step": 59255
    },
    {
      "epoch": 0.6028238932291666,
      "grad_norm": 11.230707168579102,
      "learning_rate": 1.7094077029069985e-06,
      "loss": 3.2768,
      "step": 59260
    },
    {
      "epoch": 0.602874755859375,
      "grad_norm": 10.547666549682617,
      "learning_rate": 1.7090283538699538e-06,
      "loss": 3.4327,
      "step": 59265
    },
    {
      "epoch": 0.6029256184895834,
      "grad_norm": 7.769001483917236,
      "learning_rate": 1.7086490250697308e-06,
      "loss": 3.6755,
      "step": 59270
    },
    {
      "epoch": 0.6029764811197916,
      "grad_norm": 14.495406150817871,
      "learning_rate": 1.7082697165160356e-06,
      "loss": 2.927,
      "step": 59275
    },
    {
      "epoch": 0.60302734375,
      "grad_norm": 14.664689064025879,
      "learning_rate": 1.7078904282185716e-06,
      "loss": 3.6418,
      "step": 59280
    },
    {
      "epoch": 0.6030782063802084,
      "grad_norm": 11.937549591064453,
      "learning_rate": 1.7075111601870441e-06,
      "loss": 3.2874,
      "step": 59285
    },
    {
      "epoch": 0.6031290690104166,
      "grad_norm": 12.20949935913086,
      "learning_rate": 1.7071319124311544e-06,
      "loss": 3.0881,
      "step": 59290
    },
    {
      "epoch": 0.603179931640625,
      "grad_norm": 13.66655158996582,
      "learning_rate": 1.7067526849606076e-06,
      "loss": 3.1871,
      "step": 59295
    },
    {
      "epoch": 0.6032307942708334,
      "grad_norm": 9.82215404510498,
      "learning_rate": 1.7063734777851054e-06,
      "loss": 3.4314,
      "step": 59300
    },
    {
      "epoch": 0.6032816569010416,
      "grad_norm": 12.262903213500977,
      "learning_rate": 1.7059942909143495e-06,
      "loss": 3.1973,
      "step": 59305
    },
    {
      "epoch": 0.60333251953125,
      "grad_norm": 16.095964431762695,
      "learning_rate": 1.7056151243580407e-06,
      "loss": 3.4884,
      "step": 59310
    },
    {
      "epoch": 0.6033833821614584,
      "grad_norm": 12.30743408203125,
      "learning_rate": 1.7052359781258819e-06,
      "loss": 3.2251,
      "step": 59315
    },
    {
      "epoch": 0.6034342447916666,
      "grad_norm": 12.364432334899902,
      "learning_rate": 1.7048568522275717e-06,
      "loss": 3.1654,
      "step": 59320
    },
    {
      "epoch": 0.603485107421875,
      "grad_norm": 12.293103218078613,
      "learning_rate": 1.7044777466728107e-06,
      "loss": 3.495,
      "step": 59325
    },
    {
      "epoch": 0.6035359700520834,
      "grad_norm": 10.636536598205566,
      "learning_rate": 1.7040986614712968e-06,
      "loss": 3.2208,
      "step": 59330
    },
    {
      "epoch": 0.6035868326822916,
      "grad_norm": 7.765628337860107,
      "learning_rate": 1.7037195966327313e-06,
      "loss": 3.3043,
      "step": 59335
    },
    {
      "epoch": 0.6036376953125,
      "grad_norm": 14.77886962890625,
      "learning_rate": 1.7033405521668106e-06,
      "loss": 3.9433,
      "step": 59340
    },
    {
      "epoch": 0.6036885579427084,
      "grad_norm": 12.097742080688477,
      "learning_rate": 1.702961528083234e-06,
      "loss": 3.5081,
      "step": 59345
    },
    {
      "epoch": 0.6037394205729166,
      "grad_norm": 12.712528228759766,
      "learning_rate": 1.7025825243916963e-06,
      "loss": 3.5559,
      "step": 59350
    },
    {
      "epoch": 0.603790283203125,
      "grad_norm": 12.2130765914917,
      "learning_rate": 1.7022035411018964e-06,
      "loss": 3.4545,
      "step": 59355
    },
    {
      "epoch": 0.6038411458333334,
      "grad_norm": 10.991181373596191,
      "learning_rate": 1.7018245782235305e-06,
      "loss": 3.619,
      "step": 59360
    },
    {
      "epoch": 0.6038920084635416,
      "grad_norm": 9.631376266479492,
      "learning_rate": 1.7014456357662923e-06,
      "loss": 3.213,
      "step": 59365
    },
    {
      "epoch": 0.60394287109375,
      "grad_norm": 13.050620079040527,
      "learning_rate": 1.7010667137398784e-06,
      "loss": 3.0377,
      "step": 59370
    },
    {
      "epoch": 0.6039937337239584,
      "grad_norm": 15.846744537353516,
      "learning_rate": 1.7006878121539844e-06,
      "loss": 3.1607,
      "step": 59375
    },
    {
      "epoch": 0.6040445963541666,
      "grad_norm": 13.608674049377441,
      "learning_rate": 1.7003089310183024e-06,
      "loss": 3.54,
      "step": 59380
    },
    {
      "epoch": 0.604095458984375,
      "grad_norm": 13.009920120239258,
      "learning_rate": 1.6999300703425264e-06,
      "loss": 3.2227,
      "step": 59385
    },
    {
      "epoch": 0.6041463216145834,
      "grad_norm": 13.079833030700684,
      "learning_rate": 1.6995512301363512e-06,
      "loss": 3.22,
      "step": 59390
    },
    {
      "epoch": 0.6041971842447916,
      "grad_norm": 12.880718231201172,
      "learning_rate": 1.6991724104094672e-06,
      "loss": 3.2554,
      "step": 59395
    },
    {
      "epoch": 0.604248046875,
      "grad_norm": 9.71114444732666,
      "learning_rate": 1.698793611171568e-06,
      "loss": 3.1012,
      "step": 59400
    },
    {
      "epoch": 0.6042989095052084,
      "grad_norm": 13.014288902282715,
      "learning_rate": 1.6984148324323433e-06,
      "loss": 3.286,
      "step": 59405
    },
    {
      "epoch": 0.6043497721354166,
      "grad_norm": 7.333260536193848,
      "learning_rate": 1.6980360742014857e-06,
      "loss": 3.668,
      "step": 59410
    },
    {
      "epoch": 0.604400634765625,
      "grad_norm": 10.733609199523926,
      "learning_rate": 1.6976573364886854e-06,
      "loss": 3.2209,
      "step": 59415
    },
    {
      "epoch": 0.6044514973958334,
      "grad_norm": 8.100669860839844,
      "learning_rate": 1.6972786193036313e-06,
      "loss": 3.5353,
      "step": 59420
    },
    {
      "epoch": 0.6045023600260416,
      "grad_norm": 12.514671325683594,
      "learning_rate": 1.6968999226560132e-06,
      "loss": 3.2706,
      "step": 59425
    },
    {
      "epoch": 0.60455322265625,
      "grad_norm": 11.186192512512207,
      "learning_rate": 1.6965212465555214e-06,
      "loss": 3.4426,
      "step": 59430
    },
    {
      "epoch": 0.6046040852864584,
      "grad_norm": 13.176190376281738,
      "learning_rate": 1.6961425910118423e-06,
      "loss": 3.6039,
      "step": 59435
    },
    {
      "epoch": 0.6046549479166666,
      "grad_norm": 13.289362907409668,
      "learning_rate": 1.695763956034665e-06,
      "loss": 3.3651,
      "step": 59440
    },
    {
      "epoch": 0.604705810546875,
      "grad_norm": 12.703752517700195,
      "learning_rate": 1.6953853416336753e-06,
      "loss": 3.1128,
      "step": 59445
    },
    {
      "epoch": 0.6047566731770834,
      "grad_norm": 15.119664192199707,
      "learning_rate": 1.695006747818561e-06,
      "loss": 3.4101,
      "step": 59450
    },
    {
      "epoch": 0.6048075358072916,
      "grad_norm": 8.316143035888672,
      "learning_rate": 1.6946281745990096e-06,
      "loss": 3.0702,
      "step": 59455
    },
    {
      "epoch": 0.6048583984375,
      "grad_norm": 11.337885856628418,
      "learning_rate": 1.6942496219847037e-06,
      "loss": 3.1695,
      "step": 59460
    },
    {
      "epoch": 0.6049092610677084,
      "grad_norm": 15.670044898986816,
      "learning_rate": 1.693871089985331e-06,
      "loss": 3.3852,
      "step": 59465
    },
    {
      "epoch": 0.6049601236979166,
      "grad_norm": 14.897687911987305,
      "learning_rate": 1.693492578610576e-06,
      "loss": 3.5343,
      "step": 59470
    },
    {
      "epoch": 0.605010986328125,
      "grad_norm": 13.250404357910156,
      "learning_rate": 1.693114087870122e-06,
      "loss": 3.0307,
      "step": 59475
    },
    {
      "epoch": 0.6050618489583334,
      "grad_norm": 13.780698776245117,
      "learning_rate": 1.6927356177736516e-06,
      "loss": 3.7522,
      "step": 59480
    },
    {
      "epoch": 0.6051127115885416,
      "grad_norm": 12.571311950683594,
      "learning_rate": 1.6923571683308497e-06,
      "loss": 3.1342,
      "step": 59485
    },
    {
      "epoch": 0.60516357421875,
      "grad_norm": 15.983752250671387,
      "learning_rate": 1.6919787395513993e-06,
      "loss": 3.1635,
      "step": 59490
    },
    {
      "epoch": 0.6052144368489584,
      "grad_norm": 10.591636657714844,
      "learning_rate": 1.6916003314449804e-06,
      "loss": 3.5172,
      "step": 59495
    },
    {
      "epoch": 0.6052652994791666,
      "grad_norm": 15.415030479431152,
      "learning_rate": 1.6912219440212751e-06,
      "loss": 3.251,
      "step": 59500
    },
    {
      "epoch": 0.605316162109375,
      "grad_norm": 16.442806243896484,
      "learning_rate": 1.6908435772899656e-06,
      "loss": 3.2447,
      "step": 59505
    },
    {
      "epoch": 0.6053670247395834,
      "grad_norm": 12.691445350646973,
      "learning_rate": 1.6904652312607306e-06,
      "loss": 2.9315,
      "step": 59510
    },
    {
      "epoch": 0.6054178873697916,
      "grad_norm": 10.97580623626709,
      "learning_rate": 1.6900869059432518e-06,
      "loss": 3.3459,
      "step": 59515
    },
    {
      "epoch": 0.60546875,
      "grad_norm": 7.406706809997559,
      "learning_rate": 1.689708601347206e-06,
      "loss": 3.1818,
      "step": 59520
    },
    {
      "epoch": 0.6055196126302084,
      "grad_norm": 13.156481742858887,
      "learning_rate": 1.6893303174822746e-06,
      "loss": 3.1799,
      "step": 59525
    },
    {
      "epoch": 0.6055704752604166,
      "grad_norm": 15.111701011657715,
      "learning_rate": 1.688952054358135e-06,
      "loss": 3.2728,
      "step": 59530
    },
    {
      "epoch": 0.605621337890625,
      "grad_norm": 14.210783004760742,
      "learning_rate": 1.688573811984464e-06,
      "loss": 3.6025,
      "step": 59535
    },
    {
      "epoch": 0.6056722005208334,
      "grad_norm": 12.524040222167969,
      "learning_rate": 1.68819559037094e-06,
      "loss": 3.5204,
      "step": 59540
    },
    {
      "epoch": 0.6057230631510416,
      "grad_norm": 12.803829193115234,
      "learning_rate": 1.68781738952724e-06,
      "loss": 3.8352,
      "step": 59545
    },
    {
      "epoch": 0.60577392578125,
      "grad_norm": 12.3107328414917,
      "learning_rate": 1.6874392094630392e-06,
      "loss": 3.5257,
      "step": 59550
    },
    {
      "epoch": 0.6058247884114584,
      "grad_norm": 10.546566009521484,
      "learning_rate": 1.687061050188013e-06,
      "loss": 3.1221,
      "step": 59555
    },
    {
      "epoch": 0.6058756510416666,
      "grad_norm": 9.176051139831543,
      "learning_rate": 1.6866829117118383e-06,
      "loss": 3.1705,
      "step": 59560
    },
    {
      "epoch": 0.605926513671875,
      "grad_norm": 10.537932395935059,
      "learning_rate": 1.6863047940441879e-06,
      "loss": 3.1709,
      "step": 59565
    },
    {
      "epoch": 0.6059773763020834,
      "grad_norm": 10.292647361755371,
      "learning_rate": 1.6859266971947369e-06,
      "loss": 3.133,
      "step": 59570
    },
    {
      "epoch": 0.6060282389322916,
      "grad_norm": 10.263216018676758,
      "learning_rate": 1.6855486211731575e-06,
      "loss": 3.3067,
      "step": 59575
    },
    {
      "epoch": 0.6060791015625,
      "grad_norm": 8.906332015991211,
      "learning_rate": 1.685170565989124e-06,
      "loss": 3.4641,
      "step": 59580
    },
    {
      "epoch": 0.6061299641927084,
      "grad_norm": 11.056873321533203,
      "learning_rate": 1.684792531652309e-06,
      "loss": 3.4226,
      "step": 59585
    },
    {
      "epoch": 0.6061808268229166,
      "grad_norm": 12.264915466308594,
      "learning_rate": 1.6844145181723833e-06,
      "loss": 3.3957,
      "step": 59590
    },
    {
      "epoch": 0.606231689453125,
      "grad_norm": 12.649701118469238,
      "learning_rate": 1.6840365255590186e-06,
      "loss": 3.1921,
      "step": 59595
    },
    {
      "epoch": 0.6062825520833334,
      "grad_norm": 13.983546257019043,
      "learning_rate": 1.6836585538218868e-06,
      "loss": 3.0141,
      "step": 59600
    },
    {
      "epoch": 0.6063334147135416,
      "grad_norm": 8.407326698303223,
      "learning_rate": 1.6832806029706568e-06,
      "loss": 3.5029,
      "step": 59605
    },
    {
      "epoch": 0.60638427734375,
      "grad_norm": 14.59919548034668,
      "learning_rate": 1.6829026730149993e-06,
      "loss": 3.2456,
      "step": 59610
    },
    {
      "epoch": 0.6064351399739584,
      "grad_norm": 6.537646293640137,
      "learning_rate": 1.6825247639645828e-06,
      "loss": 3.3567,
      "step": 59615
    },
    {
      "epoch": 0.6064860026041666,
      "grad_norm": 8.798364639282227,
      "learning_rate": 1.6821468758290765e-06,
      "loss": 3.3308,
      "step": 59620
    },
    {
      "epoch": 0.606536865234375,
      "grad_norm": 13.769247055053711,
      "learning_rate": 1.6817690086181494e-06,
      "loss": 3.1139,
      "step": 59625
    },
    {
      "epoch": 0.6065877278645834,
      "grad_norm": 14.075472831726074,
      "learning_rate": 1.6813911623414674e-06,
      "loss": 3.2553,
      "step": 59630
    },
    {
      "epoch": 0.6066385904947916,
      "grad_norm": 16.200557708740234,
      "learning_rate": 1.6810133370086985e-06,
      "loss": 3.3608,
      "step": 59635
    },
    {
      "epoch": 0.606689453125,
      "grad_norm": 8.004645347595215,
      "learning_rate": 1.6806355326295094e-06,
      "loss": 2.9995,
      "step": 59640
    },
    {
      "epoch": 0.6067403157552084,
      "grad_norm": 14.007003784179688,
      "learning_rate": 1.6802577492135664e-06,
      "loss": 3.5005,
      "step": 59645
    },
    {
      "epoch": 0.6067911783854166,
      "grad_norm": 13.227628707885742,
      "learning_rate": 1.679879986770534e-06,
      "loss": 3.1238,
      "step": 59650
    },
    {
      "epoch": 0.606842041015625,
      "grad_norm": 16.293874740600586,
      "learning_rate": 1.6795022453100773e-06,
      "loss": 3.0458,
      "step": 59655
    },
    {
      "epoch": 0.6068929036458334,
      "grad_norm": 10.60593032836914,
      "learning_rate": 1.6791245248418624e-06,
      "loss": 3.5369,
      "step": 59660
    },
    {
      "epoch": 0.6069437662760416,
      "grad_norm": 12.795389175415039,
      "learning_rate": 1.6787468253755516e-06,
      "loss": 3.1973,
      "step": 59665
    },
    {
      "epoch": 0.60699462890625,
      "grad_norm": 7.493161201477051,
      "learning_rate": 1.6783691469208078e-06,
      "loss": 3.3322,
      "step": 59670
    },
    {
      "epoch": 0.6070454915364584,
      "grad_norm": 13.887099266052246,
      "learning_rate": 1.6779914894872955e-06,
      "loss": 3.2289,
      "step": 59675
    },
    {
      "epoch": 0.6070963541666666,
      "grad_norm": 13.77320384979248,
      "learning_rate": 1.6776138530846758e-06,
      "loss": 3.4333,
      "step": 59680
    },
    {
      "epoch": 0.607147216796875,
      "grad_norm": 6.699323654174805,
      "learning_rate": 1.6772362377226115e-06,
      "loss": 2.8962,
      "step": 59685
    },
    {
      "epoch": 0.6071980794270834,
      "grad_norm": 16.819921493530273,
      "learning_rate": 1.6768586434107616e-06,
      "loss": 3.3299,
      "step": 59690
    },
    {
      "epoch": 0.6072489420572916,
      "grad_norm": 7.822608470916748,
      "learning_rate": 1.6764810701587891e-06,
      "loss": 3.3881,
      "step": 59695
    },
    {
      "epoch": 0.6072998046875,
      "grad_norm": 13.360980987548828,
      "learning_rate": 1.6761035179763536e-06,
      "loss": 3.5471,
      "step": 59700
    },
    {
      "epoch": 0.6073506673177084,
      "grad_norm": 8.949075698852539,
      "learning_rate": 1.6757259868731134e-06,
      "loss": 3.1237,
      "step": 59705
    },
    {
      "epoch": 0.6074015299479166,
      "grad_norm": 13.684633255004883,
      "learning_rate": 1.6753484768587285e-06,
      "loss": 3.2771,
      "step": 59710
    },
    {
      "epoch": 0.607452392578125,
      "grad_norm": 9.120360374450684,
      "learning_rate": 1.6749709879428584e-06,
      "loss": 3.329,
      "step": 59715
    },
    {
      "epoch": 0.6075032552083334,
      "grad_norm": 11.726250648498535,
      "learning_rate": 1.6745935201351587e-06,
      "loss": 3.0295,
      "step": 59720
    },
    {
      "epoch": 0.6075541178385416,
      "grad_norm": 7.57747220993042,
      "learning_rate": 1.6742160734452894e-06,
      "loss": 3.4097,
      "step": 59725
    },
    {
      "epoch": 0.60760498046875,
      "grad_norm": 7.077104568481445,
      "learning_rate": 1.6738386478829049e-06,
      "loss": 3.1283,
      "step": 59730
    },
    {
      "epoch": 0.6076558430989584,
      "grad_norm": 12.774635314941406,
      "learning_rate": 1.6734612434576636e-06,
      "loss": 3.305,
      "step": 59735
    },
    {
      "epoch": 0.6077067057291666,
      "grad_norm": 8.077064514160156,
      "learning_rate": 1.6730838601792203e-06,
      "loss": 3.3254,
      "step": 59740
    },
    {
      "epoch": 0.607757568359375,
      "grad_norm": 9.806227684020996,
      "learning_rate": 1.6727064980572306e-06,
      "loss": 2.9541,
      "step": 59745
    },
    {
      "epoch": 0.6078084309895834,
      "grad_norm": 14.675581932067871,
      "learning_rate": 1.672329157101348e-06,
      "loss": 3.3303,
      "step": 59750
    },
    {
      "epoch": 0.6078592936197916,
      "grad_norm": 9.640061378479004,
      "learning_rate": 1.671951837321229e-06,
      "loss": 3.2153,
      "step": 59755
    },
    {
      "epoch": 0.60791015625,
      "grad_norm": 14.155195236206055,
      "learning_rate": 1.6715745387265255e-06,
      "loss": 3.1934,
      "step": 59760
    },
    {
      "epoch": 0.6079610188802084,
      "grad_norm": 13.437745094299316,
      "learning_rate": 1.6711972613268907e-06,
      "loss": 3.4795,
      "step": 59765
    },
    {
      "epoch": 0.6080118815104166,
      "grad_norm": 13.996333122253418,
      "learning_rate": 1.6708200051319783e-06,
      "loss": 3.5947,
      "step": 59770
    },
    {
      "epoch": 0.608062744140625,
      "grad_norm": 11.20358943939209,
      "learning_rate": 1.6704427701514392e-06,
      "loss": 3.2206,
      "step": 59775
    },
    {
      "epoch": 0.6081136067708334,
      "grad_norm": 9.134904861450195,
      "learning_rate": 1.6700655563949256e-06,
      "loss": 3.3026,
      "step": 59780
    },
    {
      "epoch": 0.6081644694010416,
      "grad_norm": 12.813766479492188,
      "learning_rate": 1.6696883638720864e-06,
      "loss": 3.1282,
      "step": 59785
    },
    {
      "epoch": 0.60821533203125,
      "grad_norm": 11.599610328674316,
      "learning_rate": 1.6693111925925753e-06,
      "loss": 3.6992,
      "step": 59790
    },
    {
      "epoch": 0.6082661946614584,
      "grad_norm": 11.954835891723633,
      "learning_rate": 1.66893404256604e-06,
      "loss": 3.4668,
      "step": 59795
    },
    {
      "epoch": 0.6083170572916666,
      "grad_norm": 12.370587348937988,
      "learning_rate": 1.6685569138021308e-06,
      "loss": 3.5618,
      "step": 59800
    },
    {
      "epoch": 0.608367919921875,
      "grad_norm": 9.70938777923584,
      "learning_rate": 1.6681798063104943e-06,
      "loss": 3.1108,
      "step": 59805
    },
    {
      "epoch": 0.6084187825520834,
      "grad_norm": 9.000341415405273,
      "learning_rate": 1.6678027201007813e-06,
      "loss": 3.4236,
      "step": 59810
    },
    {
      "epoch": 0.6084696451822916,
      "grad_norm": 11.784370422363281,
      "learning_rate": 1.667425655182639e-06,
      "loss": 3.5309,
      "step": 59815
    },
    {
      "epoch": 0.6085205078125,
      "grad_norm": 11.995101928710938,
      "learning_rate": 1.6670486115657131e-06,
      "loss": 2.7516,
      "step": 59820
    },
    {
      "epoch": 0.6085713704427084,
      "grad_norm": 12.232709884643555,
      "learning_rate": 1.6666715892596508e-06,
      "loss": 3.4496,
      "step": 59825
    },
    {
      "epoch": 0.6086222330729166,
      "grad_norm": 11.079641342163086,
      "learning_rate": 1.6662945882740993e-06,
      "loss": 3.1395,
      "step": 59830
    },
    {
      "epoch": 0.608673095703125,
      "grad_norm": 12.308189392089844,
      "learning_rate": 1.6659176086187029e-06,
      "loss": 3.4337,
      "step": 59835
    },
    {
      "epoch": 0.6087239583333334,
      "grad_norm": 11.949128150939941,
      "learning_rate": 1.6655406503031068e-06,
      "loss": 3.4379,
      "step": 59840
    },
    {
      "epoch": 0.6087748209635416,
      "grad_norm": 13.886360168457031,
      "learning_rate": 1.6651637133369547e-06,
      "loss": 3.0902,
      "step": 59845
    },
    {
      "epoch": 0.60882568359375,
      "grad_norm": 7.020474433898926,
      "learning_rate": 1.6647867977298915e-06,
      "loss": 3.1472,
      "step": 59850
    },
    {
      "epoch": 0.6088765462239584,
      "grad_norm": 13.490387916564941,
      "learning_rate": 1.6644099034915607e-06,
      "loss": 3.1536,
      "step": 59855
    },
    {
      "epoch": 0.6089274088541666,
      "grad_norm": 12.121207237243652,
      "learning_rate": 1.6640330306316033e-06,
      "loss": 3.1876,
      "step": 59860
    },
    {
      "epoch": 0.608978271484375,
      "grad_norm": 8.730725288391113,
      "learning_rate": 1.6636561791596634e-06,
      "loss": 3.4631,
      "step": 59865
    },
    {
      "epoch": 0.6090291341145834,
      "grad_norm": 10.086065292358398,
      "learning_rate": 1.6632793490853822e-06,
      "loss": 3.2787,
      "step": 59870
    },
    {
      "epoch": 0.6090799967447916,
      "grad_norm": 13.390612602233887,
      "learning_rate": 1.6629025404184e-06,
      "loss": 3.4593,
      "step": 59875
    },
    {
      "epoch": 0.609130859375,
      "grad_norm": 9.88414192199707,
      "learning_rate": 1.6625257531683579e-06,
      "loss": 3.0767,
      "step": 59880
    },
    {
      "epoch": 0.6091817220052084,
      "grad_norm": 8.802709579467773,
      "learning_rate": 1.6621489873448964e-06,
      "loss": 3.3471,
      "step": 59885
    },
    {
      "epoch": 0.6092325846354166,
      "grad_norm": 11.346463203430176,
      "learning_rate": 1.6617722429576542e-06,
      "loss": 3.3545,
      "step": 59890
    },
    {
      "epoch": 0.609283447265625,
      "grad_norm": 12.869057655334473,
      "learning_rate": 1.6613955200162706e-06,
      "loss": 3.1323,
      "step": 59895
    },
    {
      "epoch": 0.6093343098958334,
      "grad_norm": 10.93761157989502,
      "learning_rate": 1.6610188185303834e-06,
      "loss": 3.3741,
      "step": 59900
    },
    {
      "epoch": 0.6093851725260416,
      "grad_norm": 7.27028751373291,
      "learning_rate": 1.6606421385096311e-06,
      "loss": 3.0519,
      "step": 59905
    },
    {
      "epoch": 0.60943603515625,
      "grad_norm": 11.998244285583496,
      "learning_rate": 1.6602654799636512e-06,
      "loss": 3.3283,
      "step": 59910
    },
    {
      "epoch": 0.6094868977864584,
      "grad_norm": 13.576044082641602,
      "learning_rate": 1.6598888429020792e-06,
      "loss": 3.683,
      "step": 59915
    },
    {
      "epoch": 0.6095377604166666,
      "grad_norm": 13.00714111328125,
      "learning_rate": 1.6595122273345521e-06,
      "loss": 3.1331,
      "step": 59920
    },
    {
      "epoch": 0.609588623046875,
      "grad_norm": 10.230525970458984,
      "learning_rate": 1.659135633270706e-06,
      "loss": 3.209,
      "step": 59925
    },
    {
      "epoch": 0.6096394856770834,
      "grad_norm": 8.578835487365723,
      "learning_rate": 1.6587590607201754e-06,
      "loss": 3.1594,
      "step": 59930
    },
    {
      "epoch": 0.6096903483072916,
      "grad_norm": 7.010080814361572,
      "learning_rate": 1.6583825096925948e-06,
      "loss": 3.7019,
      "step": 59935
    },
    {
      "epoch": 0.6097412109375,
      "grad_norm": 9.297857284545898,
      "learning_rate": 1.6580059801975968e-06,
      "loss": 3.7343,
      "step": 59940
    },
    {
      "epoch": 0.6097920735677084,
      "grad_norm": 14.203421592712402,
      "learning_rate": 1.657629472244818e-06,
      "loss": 3.1271,
      "step": 59945
    },
    {
      "epoch": 0.6098429361979166,
      "grad_norm": 7.7891998291015625,
      "learning_rate": 1.6572529858438888e-06,
      "loss": 2.9507,
      "step": 59950
    },
    {
      "epoch": 0.609893798828125,
      "grad_norm": 9.288692474365234,
      "learning_rate": 1.6568765210044419e-06,
      "loss": 3.1269,
      "step": 59955
    },
    {
      "epoch": 0.6099446614583334,
      "grad_norm": 23.345260620117188,
      "learning_rate": 1.65650007773611e-06,
      "loss": 3.6945,
      "step": 59960
    },
    {
      "epoch": 0.6099955240885416,
      "grad_norm": 12.625548362731934,
      "learning_rate": 1.6561236560485233e-06,
      "loss": 3.3782,
      "step": 59965
    },
    {
      "epoch": 0.61004638671875,
      "grad_norm": 9.033101081848145,
      "learning_rate": 1.6557472559513136e-06,
      "loss": 3.4724,
      "step": 59970
    },
    {
      "epoch": 0.6100972493489584,
      "grad_norm": 14.025511741638184,
      "learning_rate": 1.6553708774541087e-06,
      "loss": 3.849,
      "step": 59975
    },
    {
      "epoch": 0.6101481119791666,
      "grad_norm": 12.046526908874512,
      "learning_rate": 1.6549945205665408e-06,
      "loss": 3.3072,
      "step": 59980
    },
    {
      "epoch": 0.610198974609375,
      "grad_norm": 10.726386070251465,
      "learning_rate": 1.654618185298238e-06,
      "loss": 3.3468,
      "step": 59985
    },
    {
      "epoch": 0.6102498372395834,
      "grad_norm": 11.036408424377441,
      "learning_rate": 1.6542418716588283e-06,
      "loss": 3.3872,
      "step": 59990
    },
    {
      "epoch": 0.6103006998697916,
      "grad_norm": 16.50437355041504,
      "learning_rate": 1.653865579657939e-06,
      "loss": 3.1342,
      "step": 59995
    },
    {
      "epoch": 0.6103515625,
      "grad_norm": 9.462359428405762,
      "learning_rate": 1.6534893093051995e-06,
      "loss": 3.4811,
      "step": 60000
    },
    {
      "epoch": 0.6104024251302084,
      "grad_norm": 11.049467086791992,
      "learning_rate": 1.653113060610235e-06,
      "loss": 3.3115,
      "step": 60005
    },
    {
      "epoch": 0.6104532877604166,
      "grad_norm": 13.344293594360352,
      "learning_rate": 1.6527368335826724e-06,
      "loss": 3.6994,
      "step": 60010
    },
    {
      "epoch": 0.610504150390625,
      "grad_norm": 12.55429458618164,
      "learning_rate": 1.6523606282321361e-06,
      "loss": 3.4279,
      "step": 60015
    },
    {
      "epoch": 0.6105550130208334,
      "grad_norm": 14.450660705566406,
      "learning_rate": 1.6519844445682525e-06,
      "loss": 3.2444,
      "step": 60020
    },
    {
      "epoch": 0.6106058756510416,
      "grad_norm": 12.68309497833252,
      "learning_rate": 1.651608282600647e-06,
      "loss": 3.4776,
      "step": 60025
    },
    {
      "epoch": 0.61065673828125,
      "grad_norm": 15.468415260314941,
      "learning_rate": 1.6512321423389416e-06,
      "loss": 3.2383,
      "step": 60030
    },
    {
      "epoch": 0.6107076009114584,
      "grad_norm": 9.671670913696289,
      "learning_rate": 1.6508560237927601e-06,
      "loss": 3.4282,
      "step": 60035
    },
    {
      "epoch": 0.6107584635416666,
      "grad_norm": 11.182158470153809,
      "learning_rate": 1.650479926971727e-06,
      "loss": 3.5716,
      "step": 60040
    },
    {
      "epoch": 0.610809326171875,
      "grad_norm": 10.060457229614258,
      "learning_rate": 1.6501038518854634e-06,
      "loss": 3.1253,
      "step": 60045
    },
    {
      "epoch": 0.6108601888020834,
      "grad_norm": 11.251843452453613,
      "learning_rate": 1.6497277985435903e-06,
      "loss": 3.1022,
      "step": 60050
    },
    {
      "epoch": 0.6109110514322916,
      "grad_norm": 12.302145957946777,
      "learning_rate": 1.6493517669557314e-06,
      "loss": 3.2956,
      "step": 60055
    },
    {
      "epoch": 0.6109619140625,
      "grad_norm": 9.519197463989258,
      "learning_rate": 1.648975757131505e-06,
      "loss": 3.225,
      "step": 60060
    },
    {
      "epoch": 0.6110127766927084,
      "grad_norm": 13.209206581115723,
      "learning_rate": 1.6485997690805328e-06,
      "loss": 3.0118,
      "step": 60065
    },
    {
      "epoch": 0.6110636393229166,
      "grad_norm": 12.686041831970215,
      "learning_rate": 1.6482238028124328e-06,
      "loss": 3.8128,
      "step": 60070
    },
    {
      "epoch": 0.611114501953125,
      "grad_norm": 14.71212100982666,
      "learning_rate": 1.6478478583368255e-06,
      "loss": 3.4088,
      "step": 60075
    },
    {
      "epoch": 0.6111653645833334,
      "grad_norm": 8.9414644241333,
      "learning_rate": 1.647471935663329e-06,
      "loss": 3.508,
      "step": 60080
    },
    {
      "epoch": 0.6112162272135416,
      "grad_norm": 11.188956260681152,
      "learning_rate": 1.6470960348015607e-06,
      "loss": 3.1362,
      "step": 60085
    },
    {
      "epoch": 0.61126708984375,
      "grad_norm": 11.71552848815918,
      "learning_rate": 1.6467201557611377e-06,
      "loss": 3.0803,
      "step": 60090
    },
    {
      "epoch": 0.6113179524739584,
      "grad_norm": 15.764275550842285,
      "learning_rate": 1.6463442985516776e-06,
      "loss": 3.8125,
      "step": 60095
    },
    {
      "epoch": 0.6113688151041666,
      "grad_norm": 9.769700050354004,
      "learning_rate": 1.6459684631827965e-06,
      "loss": 3.257,
      "step": 60100
    },
    {
      "epoch": 0.611419677734375,
      "grad_norm": 9.204610824584961,
      "learning_rate": 1.6455926496641101e-06,
      "loss": 3.2443,
      "step": 60105
    },
    {
      "epoch": 0.6114705403645834,
      "grad_norm": 10.023176193237305,
      "learning_rate": 1.6452168580052318e-06,
      "loss": 3.2031,
      "step": 60110
    },
    {
      "epoch": 0.6115214029947916,
      "grad_norm": 16.631004333496094,
      "learning_rate": 1.6448410882157794e-06,
      "loss": 3.0563,
      "step": 60115
    },
    {
      "epoch": 0.611572265625,
      "grad_norm": 17.92494010925293,
      "learning_rate": 1.6444653403053642e-06,
      "loss": 3.433,
      "step": 60120
    },
    {
      "epoch": 0.6116231282552084,
      "grad_norm": 11.909153938293457,
      "learning_rate": 1.6440896142836011e-06,
      "loss": 3.4044,
      "step": 60125
    },
    {
      "epoch": 0.6116739908854166,
      "grad_norm": 11.371234893798828,
      "learning_rate": 1.6437139101601014e-06,
      "loss": 3.6105,
      "step": 60130
    },
    {
      "epoch": 0.611724853515625,
      "grad_norm": 10.851405143737793,
      "learning_rate": 1.6433382279444787e-06,
      "loss": 3.6112,
      "step": 60135
    },
    {
      "epoch": 0.6117757161458334,
      "grad_norm": 13.315210342407227,
      "learning_rate": 1.6429625676463448e-06,
      "loss": 3.4852,
      "step": 60140
    },
    {
      "epoch": 0.6118265787760416,
      "grad_norm": 10.062915802001953,
      "learning_rate": 1.6425869292753101e-06,
      "loss": 3.215,
      "step": 60145
    },
    {
      "epoch": 0.61187744140625,
      "grad_norm": 15.468650817871094,
      "learning_rate": 1.642211312840985e-06,
      "loss": 3.4406,
      "step": 60150
    },
    {
      "epoch": 0.6119283040364584,
      "grad_norm": 8.585348129272461,
      "learning_rate": 1.6418357183529816e-06,
      "loss": 3.0856,
      "step": 60155
    },
    {
      "epoch": 0.6119791666666666,
      "grad_norm": 12.80462646484375,
      "learning_rate": 1.6414601458209067e-06,
      "loss": 3.8671,
      "step": 60160
    },
    {
      "epoch": 0.612030029296875,
      "grad_norm": 10.727526664733887,
      "learning_rate": 1.6410845952543703e-06,
      "loss": 3.5206,
      "step": 60165
    },
    {
      "epoch": 0.6120808919270834,
      "grad_norm": 13.365429878234863,
      "learning_rate": 1.640709066662982e-06,
      "loss": 4.123,
      "step": 60170
    },
    {
      "epoch": 0.6121317545572916,
      "grad_norm": 9.198284149169922,
      "learning_rate": 1.640333560056348e-06,
      "loss": 3.0953,
      "step": 60175
    },
    {
      "epoch": 0.6121826171875,
      "grad_norm": 10.258539199829102,
      "learning_rate": 1.639958075444077e-06,
      "loss": 3.3011,
      "step": 60180
    },
    {
      "epoch": 0.6122334798177084,
      "grad_norm": 10.84228229522705,
      "learning_rate": 1.6395826128357733e-06,
      "loss": 3.4246,
      "step": 60185
    },
    {
      "epoch": 0.6122843424479166,
      "grad_norm": 12.045772552490234,
      "learning_rate": 1.639207172241045e-06,
      "loss": 3.2353,
      "step": 60190
    },
    {
      "epoch": 0.612335205078125,
      "grad_norm": 13.82982063293457,
      "learning_rate": 1.6388317536694978e-06,
      "loss": 3.5885,
      "step": 60195
    },
    {
      "epoch": 0.6123860677083334,
      "grad_norm": 12.485939025878906,
      "learning_rate": 1.6384563571307356e-06,
      "loss": 3.0635,
      "step": 60200
    },
    {
      "epoch": 0.6124369303385416,
      "grad_norm": 8.347084045410156,
      "learning_rate": 1.6380809826343628e-06,
      "loss": 3.2999,
      "step": 60205
    },
    {
      "epoch": 0.61248779296875,
      "grad_norm": 8.247750282287598,
      "learning_rate": 1.6377056301899847e-06,
      "loss": 3.24,
      "step": 60210
    },
    {
      "epoch": 0.6125386555989584,
      "grad_norm": 8.546690940856934,
      "learning_rate": 1.6373302998072032e-06,
      "loss": 3.3027,
      "step": 60215
    },
    {
      "epoch": 0.6125895182291666,
      "grad_norm": 10.810529708862305,
      "learning_rate": 1.6369549914956217e-06,
      "loss": 3.2058,
      "step": 60220
    },
    {
      "epoch": 0.612640380859375,
      "grad_norm": 11.5653657913208,
      "learning_rate": 1.6365797052648417e-06,
      "loss": 3.596,
      "step": 60225
    },
    {
      "epoch": 0.6126912434895834,
      "grad_norm": 11.945663452148438,
      "learning_rate": 1.6362044411244654e-06,
      "loss": 2.9793,
      "step": 60230
    },
    {
      "epoch": 0.6127421061197916,
      "grad_norm": 11.178726196289062,
      "learning_rate": 1.6358291990840947e-06,
      "loss": 3.3166,
      "step": 60235
    },
    {
      "epoch": 0.61279296875,
      "grad_norm": 11.82205867767334,
      "learning_rate": 1.635453979153328e-06,
      "loss": 3.1893,
      "step": 60240
    },
    {
      "epoch": 0.6128438313802084,
      "grad_norm": 12.103434562683105,
      "learning_rate": 1.6350787813417662e-06,
      "loss": 3.2575,
      "step": 60245
    },
    {
      "epoch": 0.6128946940104166,
      "grad_norm": 10.711541175842285,
      "learning_rate": 1.6347036056590093e-06,
      "loss": 3.4353,
      "step": 60250
    },
    {
      "epoch": 0.612945556640625,
      "grad_norm": 12.123007774353027,
      "learning_rate": 1.6343284521146557e-06,
      "loss": 3.0564,
      "step": 60255
    },
    {
      "epoch": 0.6129964192708334,
      "grad_norm": 12.458571434020996,
      "learning_rate": 1.6339533207183028e-06,
      "loss": 3.5443,
      "step": 60260
    },
    {
      "epoch": 0.6130472819010416,
      "grad_norm": 12.762242317199707,
      "learning_rate": 1.6335782114795494e-06,
      "loss": 3.1803,
      "step": 60265
    },
    {
      "epoch": 0.61309814453125,
      "grad_norm": 13.09855842590332,
      "learning_rate": 1.6332031244079925e-06,
      "loss": 3.6583,
      "step": 60270
    },
    {
      "epoch": 0.6131490071614584,
      "grad_norm": 8.688170433044434,
      "learning_rate": 1.632828059513228e-06,
      "loss": 3.4282,
      "step": 60275
    },
    {
      "epoch": 0.6131998697916666,
      "grad_norm": 10.610824584960938,
      "learning_rate": 1.6324530168048508e-06,
      "loss": 3.2545,
      "step": 60280
    },
    {
      "epoch": 0.613250732421875,
      "grad_norm": 13.22155475616455,
      "learning_rate": 1.6320779962924593e-06,
      "loss": 3.0439,
      "step": 60285
    },
    {
      "epoch": 0.6133015950520834,
      "grad_norm": 9.175676345825195,
      "learning_rate": 1.6317029979856458e-06,
      "loss": 3.5291,
      "step": 60290
    },
    {
      "epoch": 0.6133524576822916,
      "grad_norm": 10.860220909118652,
      "learning_rate": 1.6313280218940056e-06,
      "loss": 3.6229,
      "step": 60295
    },
    {
      "epoch": 0.6134033203125,
      "grad_norm": 14.858694076538086,
      "learning_rate": 1.6309530680271312e-06,
      "loss": 3.1163,
      "step": 60300
    },
    {
      "epoch": 0.6134541829427084,
      "grad_norm": 8.63109302520752,
      "learning_rate": 1.6305781363946172e-06,
      "loss": 2.9914,
      "step": 60305
    },
    {
      "epoch": 0.6135050455729166,
      "grad_norm": 9.039706230163574,
      "learning_rate": 1.630203227006056e-06,
      "loss": 3.4852,
      "step": 60310
    },
    {
      "epoch": 0.613555908203125,
      "grad_norm": 13.092826843261719,
      "learning_rate": 1.6298283398710383e-06,
      "loss": 3.4819,
      "step": 60315
    },
    {
      "epoch": 0.6136067708333334,
      "grad_norm": 10.986658096313477,
      "learning_rate": 1.6294534749991559e-06,
      "loss": 3.3921,
      "step": 60320
    },
    {
      "epoch": 0.6136576334635416,
      "grad_norm": 12.920578002929688,
      "learning_rate": 1.6290786324000007e-06,
      "loss": 3.2766,
      "step": 60325
    },
    {
      "epoch": 0.61370849609375,
      "grad_norm": 12.116365432739258,
      "learning_rate": 1.6287038120831621e-06,
      "loss": 3.1988,
      "step": 60330
    },
    {
      "epoch": 0.6137593587239584,
      "grad_norm": 8.024656295776367,
      "learning_rate": 1.6283290140582305e-06,
      "loss": 3.0811,
      "step": 60335
    },
    {
      "epoch": 0.6138102213541666,
      "grad_norm": 16.094438552856445,
      "learning_rate": 1.6279542383347935e-06,
      "loss": 3.1802,
      "step": 60340
    },
    {
      "epoch": 0.613861083984375,
      "grad_norm": 9.648401260375977,
      "learning_rate": 1.6275794849224411e-06,
      "loss": 3.434,
      "step": 60345
    },
    {
      "epoch": 0.6139119466145834,
      "grad_norm": 10.579446792602539,
      "learning_rate": 1.627204753830761e-06,
      "loss": 3.2385,
      "step": 60350
    },
    {
      "epoch": 0.6139628092447916,
      "grad_norm": 11.142621040344238,
      "learning_rate": 1.6268300450693397e-06,
      "loss": 3.5142,
      "step": 60355
    },
    {
      "epoch": 0.614013671875,
      "grad_norm": 8.502201080322266,
      "learning_rate": 1.6264553586477651e-06,
      "loss": 3.7343,
      "step": 60360
    },
    {
      "epoch": 0.6140645345052084,
      "grad_norm": 12.512798309326172,
      "learning_rate": 1.6260806945756235e-06,
      "loss": 4.0635,
      "step": 60365
    },
    {
      "epoch": 0.6141153971354166,
      "grad_norm": 8.338419914245605,
      "learning_rate": 1.6257060528625e-06,
      "loss": 3.2704,
      "step": 60370
    },
    {
      "epoch": 0.614166259765625,
      "grad_norm": 13.648222923278809,
      "learning_rate": 1.6253314335179792e-06,
      "loss": 3.3203,
      "step": 60375
    },
    {
      "epoch": 0.6142171223958334,
      "grad_norm": 10.438228607177734,
      "learning_rate": 1.6249568365516474e-06,
      "loss": 3.4015,
      "step": 60380
    },
    {
      "epoch": 0.6142679850260416,
      "grad_norm": 12.06532096862793,
      "learning_rate": 1.6245822619730871e-06,
      "loss": 3.1786,
      "step": 60385
    },
    {
      "epoch": 0.61431884765625,
      "grad_norm": 9.721664428710938,
      "learning_rate": 1.6242077097918824e-06,
      "loss": 3.1364,
      "step": 60390
    },
    {
      "epoch": 0.6143697102864584,
      "grad_norm": 13.711265563964844,
      "learning_rate": 1.6238331800176144e-06,
      "loss": 3.8421,
      "step": 60395
    },
    {
      "epoch": 0.6144205729166666,
      "grad_norm": 12.896185874938965,
      "learning_rate": 1.6234586726598685e-06,
      "loss": 3.2728,
      "step": 60400
    },
    {
      "epoch": 0.614471435546875,
      "grad_norm": 6.6597418785095215,
      "learning_rate": 1.6230841877282242e-06,
      "loss": 3.3412,
      "step": 60405
    },
    {
      "epoch": 0.6145222981770834,
      "grad_norm": 10.515288352966309,
      "learning_rate": 1.6227097252322636e-06,
      "loss": 3.2781,
      "step": 60410
    },
    {
      "epoch": 0.6145731608072916,
      "grad_norm": 12.3347749710083,
      "learning_rate": 1.6223352851815654e-06,
      "loss": 3.3966,
      "step": 60415
    },
    {
      "epoch": 0.6146240234375,
      "grad_norm": 9.619572639465332,
      "learning_rate": 1.6219608675857118e-06,
      "loss": 3.4378,
      "step": 60420
    },
    {
      "epoch": 0.6146748860677084,
      "grad_norm": 11.642086029052734,
      "learning_rate": 1.6215864724542819e-06,
      "loss": 3.2912,
      "step": 60425
    },
    {
      "epoch": 0.6147257486979166,
      "grad_norm": 19.23407554626465,
      "learning_rate": 1.6212120997968528e-06,
      "loss": 3.6397,
      "step": 60430
    },
    {
      "epoch": 0.614776611328125,
      "grad_norm": 7.65911865234375,
      "learning_rate": 1.6208377496230038e-06,
      "loss": 3.3781,
      "step": 60435
    },
    {
      "epoch": 0.6148274739583334,
      "grad_norm": 17.415040969848633,
      "learning_rate": 1.620463421942313e-06,
      "loss": 3.3809,
      "step": 60440
    },
    {
      "epoch": 0.6148783365885416,
      "grad_norm": 9.991722106933594,
      "learning_rate": 1.6200891167643573e-06,
      "loss": 3.0369,
      "step": 60445
    },
    {
      "epoch": 0.61492919921875,
      "grad_norm": 10.868343353271484,
      "learning_rate": 1.6197148340987123e-06,
      "loss": 3.4453,
      "step": 60450
    },
    {
      "epoch": 0.6149800618489584,
      "grad_norm": 11.543356895446777,
      "learning_rate": 1.6193405739549555e-06,
      "loss": 3.2694,
      "step": 60455
    },
    {
      "epoch": 0.6150309244791666,
      "grad_norm": 15.196809768676758,
      "learning_rate": 1.6189663363426612e-06,
      "loss": 3.1897,
      "step": 60460
    },
    {
      "epoch": 0.615081787109375,
      "grad_norm": 7.814451217651367,
      "learning_rate": 1.6185921212714046e-06,
      "loss": 3.5805,
      "step": 60465
    },
    {
      "epoch": 0.6151326497395834,
      "grad_norm": 14.216017723083496,
      "learning_rate": 1.6182179287507582e-06,
      "loss": 3.2458,
      "step": 60470
    },
    {
      "epoch": 0.6151835123697916,
      "grad_norm": 13.001861572265625,
      "learning_rate": 1.6178437587902984e-06,
      "loss": 3.0432,
      "step": 60475
    },
    {
      "epoch": 0.615234375,
      "grad_norm": 8.819528579711914,
      "learning_rate": 1.6174696113995969e-06,
      "loss": 3.4943,
      "step": 60480
    },
    {
      "epoch": 0.6152852376302084,
      "grad_norm": 9.95112133026123,
      "learning_rate": 1.617095486588226e-06,
      "loss": 3.1944,
      "step": 60485
    },
    {
      "epoch": 0.6153361002604166,
      "grad_norm": 8.025426864624023,
      "learning_rate": 1.616721384365757e-06,
      "loss": 2.825,
      "step": 60490
    },
    {
      "epoch": 0.615386962890625,
      "grad_norm": 14.088895797729492,
      "learning_rate": 1.6163473047417632e-06,
      "loss": 3.2174,
      "step": 60495
    },
    {
      "epoch": 0.6154378255208334,
      "grad_norm": 11.933631896972656,
      "learning_rate": 1.615973247725814e-06,
      "loss": 3.9773,
      "step": 60500
    },
    {
      "epoch": 0.6154886881510416,
      "grad_norm": 11.40555191040039,
      "learning_rate": 1.6155992133274795e-06,
      "loss": 3.2435,
      "step": 60505
    },
    {
      "epoch": 0.61553955078125,
      "grad_norm": 10.640403747558594,
      "learning_rate": 1.6152252015563292e-06,
      "loss": 3.3319,
      "step": 60510
    },
    {
      "epoch": 0.6155904134114584,
      "grad_norm": 8.765336990356445,
      "learning_rate": 1.6148512124219324e-06,
      "loss": 3.2578,
      "step": 60515
    },
    {
      "epoch": 0.6156412760416666,
      "grad_norm": 9.339484214782715,
      "learning_rate": 1.6144772459338582e-06,
      "loss": 3.4517,
      "step": 60520
    },
    {
      "epoch": 0.615692138671875,
      "grad_norm": 13.520297050476074,
      "learning_rate": 1.6141033021016732e-06,
      "loss": 3.8159,
      "step": 60525
    },
    {
      "epoch": 0.6157430013020834,
      "grad_norm": 8.561175346374512,
      "learning_rate": 1.6137293809349448e-06,
      "loss": 3.4511,
      "step": 60530
    },
    {
      "epoch": 0.6157938639322916,
      "grad_norm": 13.042895317077637,
      "learning_rate": 1.6133554824432413e-06,
      "loss": 2.9227,
      "step": 60535
    },
    {
      "epoch": 0.6158447265625,
      "grad_norm": 16.622398376464844,
      "learning_rate": 1.6129816066361267e-06,
      "loss": 3.5857,
      "step": 60540
    },
    {
      "epoch": 0.6158955891927084,
      "grad_norm": 13.52224349975586,
      "learning_rate": 1.612607753523167e-06,
      "loss": 3.1323,
      "step": 60545
    },
    {
      "epoch": 0.6159464518229166,
      "grad_norm": 12.300966262817383,
      "learning_rate": 1.6122339231139286e-06,
      "loss": 3.8293,
      "step": 60550
    },
    {
      "epoch": 0.615997314453125,
      "grad_norm": 11.415060043334961,
      "learning_rate": 1.611860115417975e-06,
      "loss": 3.1111,
      "step": 60555
    },
    {
      "epoch": 0.6160481770833334,
      "grad_norm": 8.763971328735352,
      "learning_rate": 1.6114863304448699e-06,
      "loss": 3.3285,
      "step": 60560
    },
    {
      "epoch": 0.6160990397135416,
      "grad_norm": 8.566849708557129,
      "learning_rate": 1.6111125682041752e-06,
      "loss": 3.5791,
      "step": 60565
    },
    {
      "epoch": 0.61614990234375,
      "grad_norm": 11.191141128540039,
      "learning_rate": 1.6107388287054558e-06,
      "loss": 2.918,
      "step": 60570
    },
    {
      "epoch": 0.6162007649739584,
      "grad_norm": 6.991872310638428,
      "learning_rate": 1.6103651119582721e-06,
      "loss": 3.2489,
      "step": 60575
    },
    {
      "epoch": 0.6162516276041666,
      "grad_norm": 11.497028350830078,
      "learning_rate": 1.609991417972187e-06,
      "loss": 3.2849,
      "step": 60580
    },
    {
      "epoch": 0.616302490234375,
      "grad_norm": 12.913334846496582,
      "learning_rate": 1.6096177467567594e-06,
      "loss": 3.4357,
      "step": 60585
    },
    {
      "epoch": 0.6163533528645834,
      "grad_norm": 13.08057975769043,
      "learning_rate": 1.6092440983215514e-06,
      "loss": 3.703,
      "step": 60590
    },
    {
      "epoch": 0.6164042154947916,
      "grad_norm": 8.695859909057617,
      "learning_rate": 1.6088704726761222e-06,
      "loss": 3.5304,
      "step": 60595
    },
    {
      "epoch": 0.616455078125,
      "grad_norm": 9.729327201843262,
      "learning_rate": 1.6084968698300304e-06,
      "loss": 3.4571,
      "step": 60600
    },
    {
      "epoch": 0.6165059407552084,
      "grad_norm": 13.946184158325195,
      "learning_rate": 1.6081232897928345e-06,
      "loss": 3.5135,
      "step": 60605
    },
    {
      "epoch": 0.6165568033854166,
      "grad_norm": 14.043551445007324,
      "learning_rate": 1.6077497325740938e-06,
      "loss": 3.1448,
      "step": 60610
    },
    {
      "epoch": 0.616607666015625,
      "grad_norm": 7.511965751647949,
      "learning_rate": 1.607376198183364e-06,
      "loss": 3.381,
      "step": 60615
    },
    {
      "epoch": 0.6166585286458334,
      "grad_norm": 12.47742748260498,
      "learning_rate": 1.607002686630203e-06,
      "loss": 3.7181,
      "step": 60620
    },
    {
      "epoch": 0.6167093912760416,
      "grad_norm": 8.424934387207031,
      "learning_rate": 1.6066291979241658e-06,
      "loss": 3.6728,
      "step": 60625
    },
    {
      "epoch": 0.61676025390625,
      "grad_norm": 15.62203311920166,
      "learning_rate": 1.6062557320748092e-06,
      "loss": 3.1998,
      "step": 60630
    },
    {
      "epoch": 0.6168111165364584,
      "grad_norm": 12.289361000061035,
      "learning_rate": 1.6058822890916887e-06,
      "loss": 3.3507,
      "step": 60635
    },
    {
      "epoch": 0.6168619791666666,
      "grad_norm": 10.98160171508789,
      "learning_rate": 1.6055088689843566e-06,
      "loss": 3.4108,
      "step": 60640
    },
    {
      "epoch": 0.616912841796875,
      "grad_norm": 12.248900413513184,
      "learning_rate": 1.6051354717623685e-06,
      "loss": 3.5467,
      "step": 60645
    },
    {
      "epoch": 0.6169637044270834,
      "grad_norm": 11.305987358093262,
      "learning_rate": 1.6047620974352779e-06,
      "loss": 2.991,
      "step": 60650
    },
    {
      "epoch": 0.6170145670572916,
      "grad_norm": 11.323430061340332,
      "learning_rate": 1.6043887460126361e-06,
      "loss": 3.7423,
      "step": 60655
    },
    {
      "epoch": 0.6170654296875,
      "grad_norm": 9.055760383605957,
      "learning_rate": 1.6040154175039958e-06,
      "loss": 2.8189,
      "step": 60660
    },
    {
      "epoch": 0.6171162923177084,
      "grad_norm": 14.46419620513916,
      "learning_rate": 1.6036421119189096e-06,
      "loss": 3.2592,
      "step": 60665
    },
    {
      "epoch": 0.6171671549479166,
      "grad_norm": 10.725924491882324,
      "learning_rate": 1.6032688292669268e-06,
      "loss": 3.2105,
      "step": 60670
    },
    {
      "epoch": 0.617218017578125,
      "grad_norm": 9.541210174560547,
      "learning_rate": 1.6028955695575993e-06,
      "loss": 3.4469,
      "step": 60675
    },
    {
      "epoch": 0.6172688802083334,
      "grad_norm": 9.244621276855469,
      "learning_rate": 1.6025223328004747e-06,
      "loss": 3.2099,
      "step": 60680
    },
    {
      "epoch": 0.6173197428385416,
      "grad_norm": 18.29025650024414,
      "learning_rate": 1.602149119005104e-06,
      "loss": 4.1906,
      "step": 60685
    },
    {
      "epoch": 0.61737060546875,
      "grad_norm": 12.4437894821167,
      "learning_rate": 1.6017759281810361e-06,
      "loss": 3.2861,
      "step": 60690
    },
    {
      "epoch": 0.6174214680989584,
      "grad_norm": 13.091300010681152,
      "learning_rate": 1.6014027603378174e-06,
      "loss": 3.1759,
      "step": 60695
    },
    {
      "epoch": 0.6174723307291666,
      "grad_norm": 10.310749053955078,
      "learning_rate": 1.6010296154849958e-06,
      "loss": 3.1665,
      "step": 60700
    },
    {
      "epoch": 0.617523193359375,
      "grad_norm": 12.083423614501953,
      "learning_rate": 1.6006564936321193e-06,
      "loss": 3.1931,
      "step": 60705
    },
    {
      "epoch": 0.6175740559895834,
      "grad_norm": 13.541406631469727,
      "learning_rate": 1.6002833947887326e-06,
      "loss": 3.6198,
      "step": 60710
    },
    {
      "epoch": 0.6176249186197916,
      "grad_norm": 7.324504375457764,
      "learning_rate": 1.5999103189643827e-06,
      "loss": 3.2634,
      "step": 60715
    },
    {
      "epoch": 0.61767578125,
      "grad_norm": 12.332198143005371,
      "learning_rate": 1.599537266168612e-06,
      "loss": 3.1276,
      "step": 60720
    },
    {
      "epoch": 0.6177266438802084,
      "grad_norm": 8.473389625549316,
      "learning_rate": 1.5991642364109689e-06,
      "loss": 3.4168,
      "step": 60725
    },
    {
      "epoch": 0.6177775065104166,
      "grad_norm": 17.08176612854004,
      "learning_rate": 1.5987912297009944e-06,
      "loss": 3.3005,
      "step": 60730
    },
    {
      "epoch": 0.617828369140625,
      "grad_norm": 17.63117027282715,
      "learning_rate": 1.5984182460482336e-06,
      "loss": 3.0557,
      "step": 60735
    },
    {
      "epoch": 0.6178792317708334,
      "grad_norm": 17.080352783203125,
      "learning_rate": 1.598045285462227e-06,
      "loss": 3.3817,
      "step": 60740
    },
    {
      "epoch": 0.6179300944010416,
      "grad_norm": 15.415255546569824,
      "learning_rate": 1.5976723479525185e-06,
      "loss": 3.4583,
      "step": 60745
    },
    {
      "epoch": 0.61798095703125,
      "grad_norm": 13.57775592803955,
      "learning_rate": 1.5972994335286499e-06,
      "loss": 3.5549,
      "step": 60750
    },
    {
      "epoch": 0.6180318196614584,
      "grad_norm": 9.898155212402344,
      "learning_rate": 1.59692654220016e-06,
      "loss": 3.1014,
      "step": 60755
    },
    {
      "epoch": 0.6180826822916666,
      "grad_norm": 10.535991668701172,
      "learning_rate": 1.5965536739765914e-06,
      "loss": 3.6444,
      "step": 60760
    },
    {
      "epoch": 0.618133544921875,
      "grad_norm": 11.3397216796875,
      "learning_rate": 1.5961808288674828e-06,
      "loss": 3.128,
      "step": 60765
    },
    {
      "epoch": 0.6181844075520834,
      "grad_norm": 11.788433074951172,
      "learning_rate": 1.5958080068823733e-06,
      "loss": 3.14,
      "step": 60770
    },
    {
      "epoch": 0.6182352701822916,
      "grad_norm": 11.378423690795898,
      "learning_rate": 1.5954352080308011e-06,
      "loss": 3.5258,
      "step": 60775
    },
    {
      "epoch": 0.6182861328125,
      "grad_norm": 16.231063842773438,
      "learning_rate": 1.5950624323223056e-06,
      "loss": 3.1571,
      "step": 60780
    },
    {
      "epoch": 0.6183369954427084,
      "grad_norm": 10.170050621032715,
      "learning_rate": 1.594689679766423e-06,
      "loss": 3.3047,
      "step": 60785
    },
    {
      "epoch": 0.6183878580729166,
      "grad_norm": 10.96178913116455,
      "learning_rate": 1.5943169503726907e-06,
      "loss": 3.3667,
      "step": 60790
    },
    {
      "epoch": 0.618438720703125,
      "grad_norm": 12.257305145263672,
      "learning_rate": 1.5939442441506434e-06,
      "loss": 3.4634,
      "step": 60795
    },
    {
      "epoch": 0.6184895833333334,
      "grad_norm": 10.836377143859863,
      "learning_rate": 1.5935715611098184e-06,
      "loss": 3.0399,
      "step": 60800
    },
    {
      "epoch": 0.6185404459635416,
      "grad_norm": 7.719210147857666,
      "learning_rate": 1.593198901259751e-06,
      "loss": 3.208,
      "step": 60805
    },
    {
      "epoch": 0.61859130859375,
      "grad_norm": 11.608940124511719,
      "learning_rate": 1.592826264609974e-06,
      "loss": 3.5864,
      "step": 60810
    },
    {
      "epoch": 0.6186421712239584,
      "grad_norm": 12.18756103515625,
      "learning_rate": 1.5924536511700216e-06,
      "loss": 3.1128,
      "step": 60815
    },
    {
      "epoch": 0.6186930338541666,
      "grad_norm": 14.461941719055176,
      "learning_rate": 1.5920810609494283e-06,
      "loss": 3.4594,
      "step": 60820
    },
    {
      "epoch": 0.618743896484375,
      "grad_norm": 12.70378303527832,
      "learning_rate": 1.5917084939577255e-06,
      "loss": 3.5058,
      "step": 60825
    },
    {
      "epoch": 0.6187947591145834,
      "grad_norm": 10.086445808410645,
      "learning_rate": 1.5913359502044459e-06,
      "loss": 3.3497,
      "step": 60830
    },
    {
      "epoch": 0.6188456217447916,
      "grad_norm": 13.142423629760742,
      "learning_rate": 1.59096342969912e-06,
      "loss": 3.1421,
      "step": 60835
    },
    {
      "epoch": 0.618896484375,
      "grad_norm": 11.093779563903809,
      "learning_rate": 1.5905909324512792e-06,
      "loss": 3.5211,
      "step": 60840
    },
    {
      "epoch": 0.6189473470052084,
      "grad_norm": 13.340781211853027,
      "learning_rate": 1.5902184584704549e-06,
      "loss": 3.4532,
      "step": 60845
    },
    {
      "epoch": 0.6189982096354166,
      "grad_norm": 11.851021766662598,
      "learning_rate": 1.5898460077661743e-06,
      "loss": 3.3527,
      "step": 60850
    },
    {
      "epoch": 0.619049072265625,
      "grad_norm": 10.32907485961914,
      "learning_rate": 1.5894735803479688e-06,
      "loss": 3.1641,
      "step": 60855
    },
    {
      "epoch": 0.6190999348958334,
      "grad_norm": 13.037763595581055,
      "learning_rate": 1.589101176225366e-06,
      "loss": 3.2318,
      "step": 60860
    },
    {
      "epoch": 0.6191507975260416,
      "grad_norm": 14.834087371826172,
      "learning_rate": 1.5887287954078935e-06,
      "loss": 3.5543,
      "step": 60865
    },
    {
      "epoch": 0.61920166015625,
      "grad_norm": 14.938041687011719,
      "learning_rate": 1.5883564379050782e-06,
      "loss": 3.3169,
      "step": 60870
    },
    {
      "epoch": 0.6192525227864584,
      "grad_norm": 8.798954963684082,
      "learning_rate": 1.5879841037264476e-06,
      "loss": 3.2308,
      "step": 60875
    },
    {
      "epoch": 0.6193033854166666,
      "grad_norm": 11.63089370727539,
      "learning_rate": 1.5876117928815282e-06,
      "loss": 3.1783,
      "step": 60880
    },
    {
      "epoch": 0.619354248046875,
      "grad_norm": 13.736751556396484,
      "learning_rate": 1.5872395053798446e-06,
      "loss": 3.2713,
      "step": 60885
    },
    {
      "epoch": 0.6194051106770834,
      "grad_norm": 9.929327964782715,
      "learning_rate": 1.5868672412309212e-06,
      "loss": 5.0198,
      "step": 60890
    },
    {
      "epoch": 0.6194559733072916,
      "grad_norm": 10.958555221557617,
      "learning_rate": 1.5864950004442836e-06,
      "loss": 3.6732,
      "step": 60895
    },
    {
      "epoch": 0.6195068359375,
      "grad_norm": 13.88362979888916,
      "learning_rate": 1.5861227830294548e-06,
      "loss": 3.5338,
      "step": 60900
    },
    {
      "epoch": 0.6195576985677084,
      "grad_norm": 16.07509422302246,
      "learning_rate": 1.5857505889959585e-06,
      "loss": 3.236,
      "step": 60905
    },
    {
      "epoch": 0.6196085611979166,
      "grad_norm": 9.3330717086792,
      "learning_rate": 1.5853784183533156e-06,
      "loss": 3.0099,
      "step": 60910
    },
    {
      "epoch": 0.619659423828125,
      "grad_norm": 12.54224967956543,
      "learning_rate": 1.5850062711110497e-06,
      "loss": 3.1137,
      "step": 60915
    },
    {
      "epoch": 0.6197102864583334,
      "grad_norm": 10.101669311523438,
      "learning_rate": 1.584634147278682e-06,
      "loss": 3.0886,
      "step": 60920
    },
    {
      "epoch": 0.6197611490885416,
      "grad_norm": 8.978915214538574,
      "learning_rate": 1.5842620468657316e-06,
      "loss": 3.1704,
      "step": 60925
    },
    {
      "epoch": 0.61981201171875,
      "grad_norm": 17.443098068237305,
      "learning_rate": 1.5838899698817197e-06,
      "loss": 3.2783,
      "step": 60930
    },
    {
      "epoch": 0.6198628743489584,
      "grad_norm": 8.883197784423828,
      "learning_rate": 1.583517916336167e-06,
      "loss": 3.483,
      "step": 60935
    },
    {
      "epoch": 0.6199137369791666,
      "grad_norm": 13.600460052490234,
      "learning_rate": 1.58314588623859e-06,
      "loss": 3.2715,
      "step": 60940
    },
    {
      "epoch": 0.619964599609375,
      "grad_norm": 12.828145027160645,
      "learning_rate": 1.582773879598508e-06,
      "loss": 3.33,
      "step": 60945
    },
    {
      "epoch": 0.6200154622395834,
      "grad_norm": 10.351181983947754,
      "learning_rate": 1.58240189642544e-06,
      "loss": 3.2824,
      "step": 60950
    },
    {
      "epoch": 0.6200663248697916,
      "grad_norm": 11.74328327178955,
      "learning_rate": 1.5820299367289015e-06,
      "loss": 2.9625,
      "step": 60955
    },
    {
      "epoch": 0.6201171875,
      "grad_norm": 8.31732177734375,
      "learning_rate": 1.5816580005184101e-06,
      "loss": 4.0897,
      "step": 60960
    },
    {
      "epoch": 0.6201680501302084,
      "grad_norm": 10.640729904174805,
      "learning_rate": 1.5812860878034796e-06,
      "loss": 3.3504,
      "step": 60965
    },
    {
      "epoch": 0.6202189127604166,
      "grad_norm": 12.188005447387695,
      "learning_rate": 1.5809141985936278e-06,
      "loss": 2.9415,
      "step": 60970
    },
    {
      "epoch": 0.620269775390625,
      "grad_norm": 14.315089225769043,
      "learning_rate": 1.5805423328983688e-06,
      "loss": 3.2181,
      "step": 60975
    },
    {
      "epoch": 0.6203206380208334,
      "grad_norm": 13.620156288146973,
      "learning_rate": 1.5801704907272155e-06,
      "loss": 3.3906,
      "step": 60980
    },
    {
      "epoch": 0.6203715006510416,
      "grad_norm": 13.081335067749023,
      "learning_rate": 1.5797986720896819e-06,
      "loss": 3.2793,
      "step": 60985
    },
    {
      "epoch": 0.62042236328125,
      "grad_norm": 14.394508361816406,
      "learning_rate": 1.5794268769952819e-06,
      "loss": 3.253,
      "step": 60990
    },
    {
      "epoch": 0.6204732259114584,
      "grad_norm": 11.010272979736328,
      "learning_rate": 1.579055105453527e-06,
      "loss": 3.2714,
      "step": 60995
    },
    {
      "epoch": 0.6205240885416666,
      "grad_norm": 17.308883666992188,
      "learning_rate": 1.578683357473929e-06,
      "loss": 3.7297,
      "step": 61000
    },
    {
      "epoch": 0.620574951171875,
      "grad_norm": 16.127193450927734,
      "learning_rate": 1.5783116330659982e-06,
      "loss": 4.0288,
      "step": 61005
    },
    {
      "epoch": 0.6206258138020834,
      "grad_norm": 10.189144134521484,
      "learning_rate": 1.5779399322392461e-06,
      "loss": 3.0836,
      "step": 61010
    },
    {
      "epoch": 0.6206766764322916,
      "grad_norm": 12.883061408996582,
      "learning_rate": 1.5775682550031828e-06,
      "loss": 3.2443,
      "step": 61015
    },
    {
      "epoch": 0.6207275390625,
      "grad_norm": 11.590263366699219,
      "learning_rate": 1.5771966013673169e-06,
      "loss": 3.6211,
      "step": 61020
    },
    {
      "epoch": 0.6207784016927084,
      "grad_norm": 12.362700462341309,
      "learning_rate": 1.5768249713411562e-06,
      "loss": 3.5006,
      "step": 61025
    },
    {
      "epoch": 0.6208292643229166,
      "grad_norm": 13.575629234313965,
      "learning_rate": 1.5764533649342107e-06,
      "loss": 3.2972,
      "step": 61030
    },
    {
      "epoch": 0.620880126953125,
      "grad_norm": 10.85159683227539,
      "learning_rate": 1.576081782155987e-06,
      "loss": 2.9801,
      "step": 61035
    },
    {
      "epoch": 0.6209309895833334,
      "grad_norm": 8.414695739746094,
      "learning_rate": 1.5757102230159915e-06,
      "loss": 3.2545,
      "step": 61040
    },
    {
      "epoch": 0.6209818522135416,
      "grad_norm": 81.48870849609375,
      "learning_rate": 1.5753386875237309e-06,
      "loss": 3.462,
      "step": 61045
    },
    {
      "epoch": 0.62103271484375,
      "grad_norm": 10.37614917755127,
      "learning_rate": 1.5749671756887117e-06,
      "loss": 3.5636,
      "step": 61050
    },
    {
      "epoch": 0.6210835774739584,
      "grad_norm": 14.355762481689453,
      "learning_rate": 1.5745956875204372e-06,
      "loss": 3.9059,
      "step": 61055
    },
    {
      "epoch": 0.6211344401041666,
      "grad_norm": 12.319256782531738,
      "learning_rate": 1.5742242230284122e-06,
      "loss": 3.194,
      "step": 61060
    },
    {
      "epoch": 0.621185302734375,
      "grad_norm": 7.4418134689331055,
      "learning_rate": 1.5738527822221421e-06,
      "loss": 3.1786,
      "step": 61065
    },
    {
      "epoch": 0.6212361653645834,
      "grad_norm": 14.95451545715332,
      "learning_rate": 1.573481365111129e-06,
      "loss": 3.5189,
      "step": 61070
    },
    {
      "epoch": 0.6212870279947916,
      "grad_norm": 13.600105285644531,
      "learning_rate": 1.5731099717048753e-06,
      "loss": 3.0263,
      "step": 61075
    },
    {
      "epoch": 0.621337890625,
      "grad_norm": 12.841672897338867,
      "learning_rate": 1.5727386020128826e-06,
      "loss": 3.1362,
      "step": 61080
    },
    {
      "epoch": 0.6213887532552084,
      "grad_norm": 8.517480850219727,
      "learning_rate": 1.5723672560446537e-06,
      "loss": 3.2489,
      "step": 61085
    },
    {
      "epoch": 0.6214396158854166,
      "grad_norm": 11.729193687438965,
      "learning_rate": 1.571995933809689e-06,
      "loss": 3.0913,
      "step": 61090
    },
    {
      "epoch": 0.621490478515625,
      "grad_norm": 7.707211971282959,
      "learning_rate": 1.571624635317488e-06,
      "loss": 3.1272,
      "step": 61095
    },
    {
      "epoch": 0.6215413411458334,
      "grad_norm": 13.470673561096191,
      "learning_rate": 1.57125336057755e-06,
      "loss": 3.1542,
      "step": 61100
    },
    {
      "epoch": 0.6215922037760416,
      "grad_norm": 11.42409610748291,
      "learning_rate": 1.5708821095993756e-06,
      "loss": 3.6853,
      "step": 61105
    },
    {
      "epoch": 0.62164306640625,
      "grad_norm": 15.191719055175781,
      "learning_rate": 1.570510882392462e-06,
      "loss": 3.4534,
      "step": 61110
    },
    {
      "epoch": 0.6216939290364584,
      "grad_norm": 14.386995315551758,
      "learning_rate": 1.5701396789663074e-06,
      "loss": 3.2562,
      "step": 61115
    },
    {
      "epoch": 0.6217447916666666,
      "grad_norm": 15.53824234008789,
      "learning_rate": 1.5697684993304074e-06,
      "loss": 3.5162,
      "step": 61120
    },
    {
      "epoch": 0.621795654296875,
      "grad_norm": 12.227032661437988,
      "learning_rate": 1.5693973434942608e-06,
      "loss": 3.472,
      "step": 61125
    },
    {
      "epoch": 0.6218465169270834,
      "grad_norm": 14.022648811340332,
      "learning_rate": 1.5690262114673632e-06,
      "loss": 3.3509,
      "step": 61130
    },
    {
      "epoch": 0.6218973795572916,
      "grad_norm": 15.221781730651855,
      "learning_rate": 1.568655103259208e-06,
      "loss": 3.6707,
      "step": 61135
    },
    {
      "epoch": 0.6219482421875,
      "grad_norm": 16.67391014099121,
      "learning_rate": 1.568284018879292e-06,
      "loss": 3.6313,
      "step": 61140
    },
    {
      "epoch": 0.6219991048177084,
      "grad_norm": 10.213263511657715,
      "learning_rate": 1.5679129583371086e-06,
      "loss": 3.448,
      "step": 61145
    },
    {
      "epoch": 0.6220499674479166,
      "grad_norm": 12.233593940734863,
      "learning_rate": 1.567541921642151e-06,
      "loss": 2.8903,
      "step": 61150
    },
    {
      "epoch": 0.622100830078125,
      "grad_norm": 14.135449409484863,
      "learning_rate": 1.567170908803912e-06,
      "loss": 3.3162,
      "step": 61155
    },
    {
      "epoch": 0.6221516927083334,
      "grad_norm": 12.742118835449219,
      "learning_rate": 1.566799919831885e-06,
      "loss": 3.0368,
      "step": 61160
    },
    {
      "epoch": 0.6222025553385416,
      "grad_norm": 13.452081680297852,
      "learning_rate": 1.5664289547355604e-06,
      "loss": 3.3572,
      "step": 61165
    },
    {
      "epoch": 0.62225341796875,
      "grad_norm": 14.116361618041992,
      "learning_rate": 1.5660580135244303e-06,
      "loss": 3.6128,
      "step": 61170
    },
    {
      "epoch": 0.6223042805989584,
      "grad_norm": 12.684968948364258,
      "learning_rate": 1.5656870962079827e-06,
      "loss": 3.5763,
      "step": 61175
    },
    {
      "epoch": 0.6223551432291666,
      "grad_norm": 14.246038436889648,
      "learning_rate": 1.5653162027957108e-06,
      "loss": 2.9694,
      "step": 61180
    },
    {
      "epoch": 0.622406005859375,
      "grad_norm": 18.166240692138672,
      "learning_rate": 1.5649453332971022e-06,
      "loss": 3.5386,
      "step": 61185
    },
    {
      "epoch": 0.6224568684895834,
      "grad_norm": 9.938048362731934,
      "learning_rate": 1.564574487721646e-06,
      "loss": 3.2802,
      "step": 61190
    },
    {
      "epoch": 0.6225077311197916,
      "grad_norm": 9.44188117980957,
      "learning_rate": 1.5642036660788286e-06,
      "loss": 3.5396,
      "step": 61195
    },
    {
      "epoch": 0.62255859375,
      "grad_norm": 15.443852424621582,
      "learning_rate": 1.5638328683781396e-06,
      "loss": 3.5264,
      "step": 61200
    },
    {
      "epoch": 0.6226094563802084,
      "grad_norm": 11.711338996887207,
      "learning_rate": 1.5634620946290646e-06,
      "loss": 3.3401,
      "step": 61205
    },
    {
      "epoch": 0.6226603190104166,
      "grad_norm": 18.48138427734375,
      "learning_rate": 1.56309134484109e-06,
      "loss": 3.2839,
      "step": 61210
    },
    {
      "epoch": 0.622711181640625,
      "grad_norm": 14.25977897644043,
      "learning_rate": 1.5627206190237004e-06,
      "loss": 3.3824,
      "step": 61215
    },
    {
      "epoch": 0.6227620442708334,
      "grad_norm": 10.02924919128418,
      "learning_rate": 1.5623499171863826e-06,
      "loss": 3.4369,
      "step": 61220
    },
    {
      "epoch": 0.6228129069010416,
      "grad_norm": 12.48977279663086,
      "learning_rate": 1.5619792393386196e-06,
      "loss": 3.9375,
      "step": 61225
    },
    {
      "epoch": 0.62286376953125,
      "grad_norm": 9.716508865356445,
      "learning_rate": 1.561608585489896e-06,
      "loss": 3.3135,
      "step": 61230
    },
    {
      "epoch": 0.6229146321614584,
      "grad_norm": 10.432272911071777,
      "learning_rate": 1.5612379556496932e-06,
      "loss": 3.3268,
      "step": 61235
    },
    {
      "epoch": 0.6229654947916666,
      "grad_norm": 12.925854682922363,
      "learning_rate": 1.560867349827495e-06,
      "loss": 3.3902,
      "step": 61240
    },
    {
      "epoch": 0.623016357421875,
      "grad_norm": 12.435482025146484,
      "learning_rate": 1.560496768032784e-06,
      "loss": 3.2661,
      "step": 61245
    },
    {
      "epoch": 0.6230672200520834,
      "grad_norm": 8.915007591247559,
      "learning_rate": 1.5601262102750391e-06,
      "loss": 3.4743,
      "step": 61250
    },
    {
      "epoch": 0.6231180826822916,
      "grad_norm": 8.147088050842285,
      "learning_rate": 1.5597556765637434e-06,
      "loss": 3.2753,
      "step": 61255
    },
    {
      "epoch": 0.6231689453125,
      "grad_norm": 12.007842063903809,
      "learning_rate": 1.559385166908376e-06,
      "loss": 3.073,
      "step": 61260
    },
    {
      "epoch": 0.6232198079427084,
      "grad_norm": 10.465039253234863,
      "learning_rate": 1.5590146813184156e-06,
      "loss": 3.2077,
      "step": 61265
    },
    {
      "epoch": 0.6232706705729166,
      "grad_norm": 10.362761497497559,
      "learning_rate": 1.558644219803341e-06,
      "loss": 3.4545,
      "step": 61270
    },
    {
      "epoch": 0.623321533203125,
      "grad_norm": 11.860074043273926,
      "learning_rate": 1.5582737823726322e-06,
      "loss": 3.2515,
      "step": 61275
    },
    {
      "epoch": 0.6233723958333334,
      "grad_norm": 8.5953950881958,
      "learning_rate": 1.5579033690357647e-06,
      "loss": 3.0689,
      "step": 61280
    },
    {
      "epoch": 0.6234232584635416,
      "grad_norm": 15.707984924316406,
      "learning_rate": 1.557532979802217e-06,
      "loss": 3.8959,
      "step": 61285
    },
    {
      "epoch": 0.62347412109375,
      "grad_norm": 15.859342575073242,
      "learning_rate": 1.5571626146814634e-06,
      "loss": 3.5281,
      "step": 61290
    },
    {
      "epoch": 0.6235249837239584,
      "grad_norm": 15.086223602294922,
      "learning_rate": 1.556792273682981e-06,
      "loss": 3.5218,
      "step": 61295
    },
    {
      "epoch": 0.6235758463541666,
      "grad_norm": 10.326348304748535,
      "learning_rate": 1.556421956816246e-06,
      "loss": 3.0658,
      "step": 61300
    },
    {
      "epoch": 0.623626708984375,
      "grad_norm": 16.39034652709961,
      "learning_rate": 1.5560516640907304e-06,
      "loss": 3.3312,
      "step": 61305
    },
    {
      "epoch": 0.6236775716145834,
      "grad_norm": 10.670428276062012,
      "learning_rate": 1.5556813955159089e-06,
      "loss": 3.3061,
      "step": 61310
    },
    {
      "epoch": 0.6237284342447916,
      "grad_norm": 10.530497550964355,
      "learning_rate": 1.5553111511012565e-06,
      "loss": 3.2695,
      "step": 61315
    },
    {
      "epoch": 0.623779296875,
      "grad_norm": 10.719157218933105,
      "learning_rate": 1.5549409308562433e-06,
      "loss": 3.2204,
      "step": 61320
    },
    {
      "epoch": 0.6238301595052084,
      "grad_norm": 9.842597007751465,
      "learning_rate": 1.5545707347903432e-06,
      "loss": 3.0986,
      "step": 61325
    },
    {
      "epoch": 0.6238810221354166,
      "grad_norm": 12.339625358581543,
      "learning_rate": 1.5542005629130253e-06,
      "loss": 3.1297,
      "step": 61330
    },
    {
      "epoch": 0.623931884765625,
      "grad_norm": 14.002964973449707,
      "learning_rate": 1.553830415233763e-06,
      "loss": 3.5682,
      "step": 61335
    },
    {
      "epoch": 0.6239827473958334,
      "grad_norm": 10.0540771484375,
      "learning_rate": 1.553460291762025e-06,
      "loss": 3.4962,
      "step": 61340
    },
    {
      "epoch": 0.6240336100260416,
      "grad_norm": 14.774270057678223,
      "learning_rate": 1.5530901925072806e-06,
      "loss": 3.4707,
      "step": 61345
    },
    {
      "epoch": 0.62408447265625,
      "grad_norm": 7.532827854156494,
      "learning_rate": 1.5527201174789999e-06,
      "loss": 3.2857,
      "step": 61350
    },
    {
      "epoch": 0.6241353352864584,
      "grad_norm": 16.660783767700195,
      "learning_rate": 1.5523500666866503e-06,
      "loss": 3.2688,
      "step": 61355
    },
    {
      "epoch": 0.6241861979166666,
      "grad_norm": 7.445517539978027,
      "learning_rate": 1.5519800401397e-06,
      "loss": 3.1697,
      "step": 61360
    },
    {
      "epoch": 0.624237060546875,
      "grad_norm": 8.375680923461914,
      "learning_rate": 1.5516100378476146e-06,
      "loss": 3.0433,
      "step": 61365
    },
    {
      "epoch": 0.6242879231770834,
      "grad_norm": 14.068017959594727,
      "learning_rate": 1.551240059819862e-06,
      "loss": 3.3664,
      "step": 61370
    },
    {
      "epoch": 0.6243387858072916,
      "grad_norm": 16.048603057861328,
      "learning_rate": 1.550870106065908e-06,
      "loss": 3.3437,
      "step": 61375
    },
    {
      "epoch": 0.6243896484375,
      "grad_norm": 12.818937301635742,
      "learning_rate": 1.5505001765952175e-06,
      "loss": 3.2911,
      "step": 61380
    },
    {
      "epoch": 0.6244405110677084,
      "grad_norm": 10.751648902893066,
      "learning_rate": 1.550130271417254e-06,
      "loss": 3.1494,
      "step": 61385
    },
    {
      "epoch": 0.6244913736979166,
      "grad_norm": 8.600159645080566,
      "learning_rate": 1.5497603905414832e-06,
      "loss": 3.7665,
      "step": 61390
    },
    {
      "epoch": 0.624542236328125,
      "grad_norm": 9.184415817260742,
      "learning_rate": 1.5493905339773674e-06,
      "loss": 3.1591,
      "step": 61395
    },
    {
      "epoch": 0.6245930989583334,
      "grad_norm": 11.268553733825684,
      "learning_rate": 1.54902070173437e-06,
      "loss": 3.1523,
      "step": 61400
    },
    {
      "epoch": 0.6246439615885416,
      "grad_norm": 8.070582389831543,
      "learning_rate": 1.5486508938219514e-06,
      "loss": 3.2748,
      "step": 61405
    },
    {
      "epoch": 0.62469482421875,
      "grad_norm": 13.604533195495605,
      "learning_rate": 1.5482811102495746e-06,
      "loss": 3.5694,
      "step": 61410
    },
    {
      "epoch": 0.6247456868489584,
      "grad_norm": 14.422013282775879,
      "learning_rate": 1.5479113510267008e-06,
      "loss": 3.2941,
      "step": 61415
    },
    {
      "epoch": 0.6247965494791666,
      "grad_norm": 12.898037910461426,
      "learning_rate": 1.5475416161627885e-06,
      "loss": 3.9211,
      "step": 61420
    },
    {
      "epoch": 0.624847412109375,
      "grad_norm": 11.304436683654785,
      "learning_rate": 1.5471719056672979e-06,
      "loss": 3.0899,
      "step": 61425
    },
    {
      "epoch": 0.6248982747395834,
      "grad_norm": 13.482966423034668,
      "learning_rate": 1.546802219549689e-06,
      "loss": 2.9373,
      "step": 61430
    },
    {
      "epoch": 0.6249491373697916,
      "grad_norm": 8.581000328063965,
      "learning_rate": 1.5464325578194194e-06,
      "loss": 3.0325,
      "step": 61435
    },
    {
      "epoch": 0.625,
      "grad_norm": 8.412178039550781,
      "learning_rate": 1.5460629204859458e-06,
      "loss": 3.5235,
      "step": 61440
    },
    {
      "epoch": 0.6250508626302084,
      "grad_norm": 12.63332748413086,
      "learning_rate": 1.5456933075587278e-06,
      "loss": 3.7609,
      "step": 61445
    },
    {
      "epoch": 0.6251017252604166,
      "grad_norm": 10.920650482177734,
      "learning_rate": 1.5453237190472191e-06,
      "loss": 3.3654,
      "step": 61450
    },
    {
      "epoch": 0.625152587890625,
      "grad_norm": 11.684036254882812,
      "learning_rate": 1.544954154960878e-06,
      "loss": 3.0327,
      "step": 61455
    },
    {
      "epoch": 0.6252034505208334,
      "grad_norm": 180.89102172851562,
      "learning_rate": 1.544584615309157e-06,
      "loss": 4.0488,
      "step": 61460
    },
    {
      "epoch": 0.6252543131510416,
      "grad_norm": 9.9097318649292,
      "learning_rate": 1.5442151001015126e-06,
      "loss": 3.1162,
      "step": 61465
    },
    {
      "epoch": 0.62530517578125,
      "grad_norm": 13.314180374145508,
      "learning_rate": 1.5438456093473992e-06,
      "loss": 3.2771,
      "step": 61470
    },
    {
      "epoch": 0.6253560384114584,
      "grad_norm": 6.927476406097412,
      "learning_rate": 1.5434761430562687e-06,
      "loss": 3.0689,
      "step": 61475
    },
    {
      "epoch": 0.6254069010416666,
      "grad_norm": 9.061094284057617,
      "learning_rate": 1.5431067012375736e-06,
      "loss": 3.4548,
      "step": 61480
    },
    {
      "epoch": 0.625457763671875,
      "grad_norm": 9.196162223815918,
      "learning_rate": 1.5427372839007676e-06,
      "loss": 3.1121,
      "step": 61485
    },
    {
      "epoch": 0.6255086263020834,
      "grad_norm": 10.728259086608887,
      "learning_rate": 1.5423678910553014e-06,
      "loss": 3.279,
      "step": 61490
    },
    {
      "epoch": 0.6255594889322916,
      "grad_norm": 14.14704704284668,
      "learning_rate": 1.5419985227106254e-06,
      "loss": 3.7505,
      "step": 61495
    },
    {
      "epoch": 0.6256103515625,
      "grad_norm": 13.941288948059082,
      "learning_rate": 1.5416291788761895e-06,
      "loss": 3.2193,
      "step": 61500
    },
    {
      "epoch": 0.6256612141927084,
      "grad_norm": 10.623527526855469,
      "learning_rate": 1.5412598595614453e-06,
      "loss": 3.2586,
      "step": 61505
    },
    {
      "epoch": 0.6257120768229166,
      "grad_norm": 12.52288818359375,
      "learning_rate": 1.5408905647758394e-06,
      "loss": 3.1298,
      "step": 61510
    },
    {
      "epoch": 0.625762939453125,
      "grad_norm": 8.349310874938965,
      "learning_rate": 1.5405212945288218e-06,
      "loss": 3.1393,
      "step": 61515
    },
    {
      "epoch": 0.6258138020833334,
      "grad_norm": 9.99730110168457,
      "learning_rate": 1.5401520488298382e-06,
      "loss": 3.5043,
      "step": 61520
    },
    {
      "epoch": 0.6258646647135416,
      "grad_norm": 9.792665481567383,
      "learning_rate": 1.5397828276883379e-06,
      "loss": 3.4502,
      "step": 61525
    },
    {
      "epoch": 0.62591552734375,
      "grad_norm": 15.315463066101074,
      "learning_rate": 1.5394136311137663e-06,
      "loss": 3.7809,
      "step": 61530
    },
    {
      "epoch": 0.6259663899739584,
      "grad_norm": 9.428129196166992,
      "learning_rate": 1.5390444591155688e-06,
      "loss": 3.388,
      "step": 61535
    },
    {
      "epoch": 0.6260172526041666,
      "grad_norm": 18.923006057739258,
      "learning_rate": 1.538675311703191e-06,
      "loss": 3.0328,
      "step": 61540
    },
    {
      "epoch": 0.626068115234375,
      "grad_norm": 8.522552490234375,
      "learning_rate": 1.5383061888860784e-06,
      "loss": 3.3867,
      "step": 61545
    },
    {
      "epoch": 0.6261189778645834,
      "grad_norm": 14.090705871582031,
      "learning_rate": 1.5379370906736735e-06,
      "loss": 3.239,
      "step": 61550
    },
    {
      "epoch": 0.6261698404947916,
      "grad_norm": 11.4149808883667,
      "learning_rate": 1.5375680170754192e-06,
      "loss": 3.5858,
      "step": 61555
    },
    {
      "epoch": 0.626220703125,
      "grad_norm": 14.478057861328125,
      "learning_rate": 1.5371989681007604e-06,
      "loss": 3.4361,
      "step": 61560
    },
    {
      "epoch": 0.6262715657552084,
      "grad_norm": 14.236669540405273,
      "learning_rate": 1.5368299437591371e-06,
      "loss": 3.6861,
      "step": 61565
    },
    {
      "epoch": 0.6263224283854166,
      "grad_norm": 12.363887786865234,
      "learning_rate": 1.5364609440599924e-06,
      "loss": 3.1573,
      "step": 61570
    },
    {
      "epoch": 0.626373291015625,
      "grad_norm": 12.806648254394531,
      "learning_rate": 1.5360919690127646e-06,
      "loss": 3.0565,
      "step": 61575
    },
    {
      "epoch": 0.6264241536458334,
      "grad_norm": 10.38432502746582,
      "learning_rate": 1.5357230186268959e-06,
      "loss": 3.375,
      "step": 61580
    },
    {
      "epoch": 0.6264750162760416,
      "grad_norm": 13.700526237487793,
      "learning_rate": 1.5353540929118257e-06,
      "loss": 3.3333,
      "step": 61585
    },
    {
      "epoch": 0.62652587890625,
      "grad_norm": 12.815266609191895,
      "learning_rate": 1.534985191876992e-06,
      "loss": 3.2802,
      "step": 61590
    },
    {
      "epoch": 0.6265767415364584,
      "grad_norm": 15.226441383361816,
      "learning_rate": 1.5346163155318328e-06,
      "loss": 3.2804,
      "step": 61595
    },
    {
      "epoch": 0.6266276041666666,
      "grad_norm": 20.004838943481445,
      "learning_rate": 1.5342474638857874e-06,
      "loss": 3.2972,
      "step": 61600
    },
    {
      "epoch": 0.626678466796875,
      "grad_norm": 16.466358184814453,
      "learning_rate": 1.5338786369482911e-06,
      "loss": 3.2316,
      "step": 61605
    },
    {
      "epoch": 0.6267293294270834,
      "grad_norm": 11.442941665649414,
      "learning_rate": 1.5335098347287813e-06,
      "loss": 3.0558,
      "step": 61610
    },
    {
      "epoch": 0.6267801920572916,
      "grad_norm": 9.389209747314453,
      "learning_rate": 1.5331410572366927e-06,
      "loss": 3.4597,
      "step": 61615
    },
    {
      "epoch": 0.6268310546875,
      "grad_norm": 15.952877044677734,
      "learning_rate": 1.5327723044814612e-06,
      "loss": 3.5438,
      "step": 61620
    },
    {
      "epoch": 0.6268819173177084,
      "grad_norm": 15.153554916381836,
      "learning_rate": 1.5324035764725216e-06,
      "loss": 3.6297,
      "step": 61625
    },
    {
      "epoch": 0.6269327799479166,
      "grad_norm": 10.831584930419922,
      "learning_rate": 1.5320348732193058e-06,
      "loss": 3.3283,
      "step": 61630
    },
    {
      "epoch": 0.626983642578125,
      "grad_norm": 10.518292427062988,
      "learning_rate": 1.5316661947312495e-06,
      "loss": 3.2956,
      "step": 61635
    },
    {
      "epoch": 0.6270345052083334,
      "grad_norm": 8.727978706359863,
      "learning_rate": 1.5312975410177838e-06,
      "loss": 4.0958,
      "step": 61640
    },
    {
      "epoch": 0.6270853678385416,
      "grad_norm": 13.804841995239258,
      "learning_rate": 1.5309289120883414e-06,
      "loss": 3.0639,
      "step": 61645
    },
    {
      "epoch": 0.62713623046875,
      "grad_norm": 13.853920936584473,
      "learning_rate": 1.5305603079523519e-06,
      "loss": 3.2488,
      "step": 61650
    },
    {
      "epoch": 0.6271870930989584,
      "grad_norm": 13.531047821044922,
      "learning_rate": 1.5301917286192478e-06,
      "loss": 3.1688,
      "step": 61655
    },
    {
      "epoch": 0.6272379557291666,
      "grad_norm": 13.21961498260498,
      "learning_rate": 1.529823174098459e-06,
      "loss": 3.1852,
      "step": 61660
    },
    {
      "epoch": 0.627288818359375,
      "grad_norm": 13.482869148254395,
      "learning_rate": 1.5294546443994139e-06,
      "loss": 3.5621,
      "step": 61665
    },
    {
      "epoch": 0.6273396809895834,
      "grad_norm": 13.146780014038086,
      "learning_rate": 1.529086139531541e-06,
      "loss": 3.6089,
      "step": 61670
    },
    {
      "epoch": 0.6273905436197916,
      "grad_norm": 8.145610809326172,
      "learning_rate": 1.5287176595042701e-06,
      "loss": 3.1591,
      "step": 61675
    },
    {
      "epoch": 0.62744140625,
      "grad_norm": 11.181539535522461,
      "learning_rate": 1.528349204327027e-06,
      "loss": 3.2037,
      "step": 61680
    },
    {
      "epoch": 0.6274922688802084,
      "grad_norm": 8.060432434082031,
      "learning_rate": 1.52798077400924e-06,
      "loss": 3.4265,
      "step": 61685
    },
    {
      "epoch": 0.6275431315104166,
      "grad_norm": 8.361469268798828,
      "learning_rate": 1.5276123685603334e-06,
      "loss": 3.7025,
      "step": 61690
    },
    {
      "epoch": 0.627593994140625,
      "grad_norm": 10.280891418457031,
      "learning_rate": 1.5272439879897345e-06,
      "loss": 3.2371,
      "step": 61695
    },
    {
      "epoch": 0.6276448567708334,
      "grad_norm": 8.784655570983887,
      "learning_rate": 1.5268756323068678e-06,
      "loss": 3.1614,
      "step": 61700
    },
    {
      "epoch": 0.6276957194010416,
      "grad_norm": 13.677227020263672,
      "learning_rate": 1.5265073015211574e-06,
      "loss": 3.5891,
      "step": 61705
    },
    {
      "epoch": 0.62774658203125,
      "grad_norm": 10.98469352722168,
      "learning_rate": 1.5261389956420258e-06,
      "loss": 3.4069,
      "step": 61710
    },
    {
      "epoch": 0.6277974446614584,
      "grad_norm": 13.143728256225586,
      "learning_rate": 1.5257707146788985e-06,
      "loss": 3.427,
      "step": 61715
    },
    {
      "epoch": 0.6278483072916666,
      "grad_norm": 8.646453857421875,
      "learning_rate": 1.5254024586411962e-06,
      "loss": 3.2642,
      "step": 61720
    },
    {
      "epoch": 0.627899169921875,
      "grad_norm": 8.369274139404297,
      "learning_rate": 1.52503422753834e-06,
      "loss": 3.2025,
      "step": 61725
    },
    {
      "epoch": 0.6279500325520834,
      "grad_norm": 12.068510055541992,
      "learning_rate": 1.5246660213797537e-06,
      "loss": 3.5439,
      "step": 61730
    },
    {
      "epoch": 0.6280008951822916,
      "grad_norm": 9.908411979675293,
      "learning_rate": 1.5242978401748549e-06,
      "loss": 3.1221,
      "step": 61735
    },
    {
      "epoch": 0.6280517578125,
      "grad_norm": 13.45252513885498,
      "learning_rate": 1.5239296839330656e-06,
      "loss": 3.3488,
      "step": 61740
    },
    {
      "epoch": 0.6281026204427084,
      "grad_norm": 12.97950553894043,
      "learning_rate": 1.5235615526638026e-06,
      "loss": 3.1984,
      "step": 61745
    },
    {
      "epoch": 0.6281534830729166,
      "grad_norm": 14.618078231811523,
      "learning_rate": 1.5231934463764863e-06,
      "loss": 3.3464,
      "step": 61750
    },
    {
      "epoch": 0.628204345703125,
      "grad_norm": 16.660869598388672,
      "learning_rate": 1.5228253650805353e-06,
      "loss": 3.3367,
      "step": 61755
    },
    {
      "epoch": 0.6282552083333334,
      "grad_norm": 9.507746696472168,
      "learning_rate": 1.5224573087853645e-06,
      "loss": 3.1135,
      "step": 61760
    },
    {
      "epoch": 0.6283060709635416,
      "grad_norm": 15.523812294006348,
      "learning_rate": 1.5220892775003915e-06,
      "loss": 3.5686,
      "step": 61765
    },
    {
      "epoch": 0.62835693359375,
      "grad_norm": 15.808150291442871,
      "learning_rate": 1.5217212712350338e-06,
      "loss": 3.2818,
      "step": 61770
    },
    {
      "epoch": 0.6284077962239584,
      "grad_norm": 12.82918643951416,
      "learning_rate": 1.5213532899987044e-06,
      "loss": 3.3855,
      "step": 61775
    },
    {
      "epoch": 0.6284586588541666,
      "grad_norm": 8.895464897155762,
      "learning_rate": 1.52098533380082e-06,
      "loss": 3.0504,
      "step": 61780
    },
    {
      "epoch": 0.628509521484375,
      "grad_norm": 8.048547744750977,
      "learning_rate": 1.5206174026507925e-06,
      "loss": 3.0557,
      "step": 61785
    },
    {
      "epoch": 0.6285603841145834,
      "grad_norm": 12.804614067077637,
      "learning_rate": 1.5202494965580378e-06,
      "loss": 3.738,
      "step": 61790
    },
    {
      "epoch": 0.6286112467447916,
      "grad_norm": 14.638296127319336,
      "learning_rate": 1.5198816155319667e-06,
      "loss": 3.0572,
      "step": 61795
    },
    {
      "epoch": 0.628662109375,
      "grad_norm": 12.839740753173828,
      "learning_rate": 1.519513759581993e-06,
      "loss": 3.1828,
      "step": 61800
    },
    {
      "epoch": 0.6287129720052084,
      "grad_norm": 11.563626289367676,
      "learning_rate": 1.5191459287175266e-06,
      "loss": 3.7413,
      "step": 61805
    },
    {
      "epoch": 0.6287638346354166,
      "grad_norm": 12.180401802062988,
      "learning_rate": 1.5187781229479796e-06,
      "loss": 3.4383,
      "step": 61810
    },
    {
      "epoch": 0.628814697265625,
      "grad_norm": 9.479646682739258,
      "learning_rate": 1.5184103422827618e-06,
      "loss": 3.2573,
      "step": 61815
    },
    {
      "epoch": 0.6288655598958334,
      "grad_norm": 13.48900032043457,
      "learning_rate": 1.5180425867312825e-06,
      "loss": 3.4651,
      "step": 61820
    },
    {
      "epoch": 0.6289164225260416,
      "grad_norm": 12.358738899230957,
      "learning_rate": 1.5176748563029504e-06,
      "loss": 3.4829,
      "step": 61825
    },
    {
      "epoch": 0.62896728515625,
      "grad_norm": 10.56752872467041,
      "learning_rate": 1.5173071510071753e-06,
      "loss": 3.4634,
      "step": 61830
    },
    {
      "epoch": 0.6290181477864584,
      "grad_norm": 10.471076965332031,
      "learning_rate": 1.516939470853363e-06,
      "loss": 2.9963,
      "step": 61835
    },
    {
      "epoch": 0.6290690104166666,
      "grad_norm": 13.142584800720215,
      "learning_rate": 1.5165718158509212e-06,
      "loss": 3.2467,
      "step": 61840
    },
    {
      "epoch": 0.629119873046875,
      "grad_norm": 10.433077812194824,
      "learning_rate": 1.5162041860092574e-06,
      "loss": 3.4712,
      "step": 61845
    },
    {
      "epoch": 0.6291707356770834,
      "grad_norm": 11.209397315979004,
      "learning_rate": 1.5158365813377758e-06,
      "loss": 3.2425,
      "step": 61850
    },
    {
      "epoch": 0.6292215983072916,
      "grad_norm": 10.223564147949219,
      "learning_rate": 1.5154690018458823e-06,
      "loss": 3.2572,
      "step": 61855
    },
    {
      "epoch": 0.6292724609375,
      "grad_norm": 18.375215530395508,
      "learning_rate": 1.5151014475429798e-06,
      "loss": 3.4334,
      "step": 61860
    },
    {
      "epoch": 0.6293233235677084,
      "grad_norm": 10.725737571716309,
      "learning_rate": 1.5147339184384741e-06,
      "loss": 3.3111,
      "step": 61865
    },
    {
      "epoch": 0.6293741861979166,
      "grad_norm": 15.086240768432617,
      "learning_rate": 1.514366414541768e-06,
      "loss": 3.3796,
      "step": 61870
    },
    {
      "epoch": 0.629425048828125,
      "grad_norm": 13.658660888671875,
      "learning_rate": 1.5139989358622633e-06,
      "loss": 3.2555,
      "step": 61875
    },
    {
      "epoch": 0.6294759114583334,
      "grad_norm": 11.644039154052734,
      "learning_rate": 1.5136314824093616e-06,
      "loss": 3.7394,
      "step": 61880
    },
    {
      "epoch": 0.6295267740885416,
      "grad_norm": 9.223989486694336,
      "learning_rate": 1.5132640541924654e-06,
      "loss": 3.5052,
      "step": 61885
    },
    {
      "epoch": 0.62957763671875,
      "grad_norm": 9.818203926086426,
      "learning_rate": 1.5128966512209741e-06,
      "loss": 3.4412,
      "step": 61890
    },
    {
      "epoch": 0.6296284993489584,
      "grad_norm": 9.734807014465332,
      "learning_rate": 1.5125292735042884e-06,
      "loss": 3.2482,
      "step": 61895
    },
    {
      "epoch": 0.6296793619791666,
      "grad_norm": 11.073369026184082,
      "learning_rate": 1.5121619210518062e-06,
      "loss": 3.3511,
      "step": 61900
    },
    {
      "epoch": 0.629730224609375,
      "grad_norm": 11.540410995483398,
      "learning_rate": 1.5117945938729279e-06,
      "loss": 3.3419,
      "step": 61905
    },
    {
      "epoch": 0.6297810872395834,
      "grad_norm": 10.681864738464355,
      "learning_rate": 1.5114272919770506e-06,
      "loss": 3.4348,
      "step": 61910
    },
    {
      "epoch": 0.6298319498697916,
      "grad_norm": 14.877259254455566,
      "learning_rate": 1.5110600153735715e-06,
      "loss": 3.6201,
      "step": 61915
    },
    {
      "epoch": 0.6298828125,
      "grad_norm": 12.818219184875488,
      "learning_rate": 1.510692764071887e-06,
      "loss": 2.9859,
      "step": 61920
    },
    {
      "epoch": 0.6299336751302084,
      "grad_norm": 15.593382835388184,
      "learning_rate": 1.5103255380813953e-06,
      "loss": 2.9853,
      "step": 61925
    },
    {
      "epoch": 0.6299845377604166,
      "grad_norm": 11.005517959594727,
      "learning_rate": 1.509958337411489e-06,
      "loss": 3.2757,
      "step": 61930
    },
    {
      "epoch": 0.630035400390625,
      "grad_norm": 17.521547317504883,
      "learning_rate": 1.5095911620715633e-06,
      "loss": 3.0168,
      "step": 61935
    },
    {
      "epoch": 0.6300862630208334,
      "grad_norm": 13.582290649414062,
      "learning_rate": 1.509224012071014e-06,
      "loss": 3.0386,
      "step": 61940
    },
    {
      "epoch": 0.6301371256510416,
      "grad_norm": 13.638851165771484,
      "learning_rate": 1.5088568874192333e-06,
      "loss": 3.4342,
      "step": 61945
    },
    {
      "epoch": 0.63018798828125,
      "grad_norm": 11.255894660949707,
      "learning_rate": 1.5084897881256142e-06,
      "loss": 3.3469,
      "step": 61950
    },
    {
      "epoch": 0.6302388509114584,
      "grad_norm": 11.319867134094238,
      "learning_rate": 1.5081227141995483e-06,
      "loss": 3.3897,
      "step": 61955
    },
    {
      "epoch": 0.6302897135416666,
      "grad_norm": 10.453210830688477,
      "learning_rate": 1.5077556656504285e-06,
      "loss": 3.5842,
      "step": 61960
    },
    {
      "epoch": 0.630340576171875,
      "grad_norm": 11.986652374267578,
      "learning_rate": 1.5073886424876443e-06,
      "loss": 3.0917,
      "step": 61965
    },
    {
      "epoch": 0.6303914388020834,
      "grad_norm": 13.140625953674316,
      "learning_rate": 1.5070216447205871e-06,
      "loss": 3.1504,
      "step": 61970
    },
    {
      "epoch": 0.6304423014322916,
      "grad_norm": 11.39574146270752,
      "learning_rate": 1.5066546723586448e-06,
      "loss": 3.3914,
      "step": 61975
    },
    {
      "epoch": 0.6304931640625,
      "grad_norm": 9.115983963012695,
      "learning_rate": 1.5062877254112077e-06,
      "loss": 3.6619,
      "step": 61980
    },
    {
      "epoch": 0.6305440266927084,
      "grad_norm": 10.484349250793457,
      "learning_rate": 1.5059208038876639e-06,
      "loss": 3.2941,
      "step": 61985
    },
    {
      "epoch": 0.6305948893229166,
      "grad_norm": 11.486682891845703,
      "learning_rate": 1.5055539077974004e-06,
      "loss": 3.3043,
      "step": 61990
    },
    {
      "epoch": 0.630645751953125,
      "grad_norm": 10.346829414367676,
      "learning_rate": 1.505187037149804e-06,
      "loss": 3.28,
      "step": 61995
    },
    {
      "epoch": 0.6306966145833334,
      "grad_norm": 9.820465087890625,
      "learning_rate": 1.5048201919542623e-06,
      "loss": 3.5778,
      "step": 62000
    },
    {
      "epoch": 0.6307474772135416,
      "grad_norm": 12.007710456848145,
      "learning_rate": 1.50445337222016e-06,
      "loss": 3.1602,
      "step": 62005
    },
    {
      "epoch": 0.63079833984375,
      "grad_norm": 8.000428199768066,
      "learning_rate": 1.5040865779568822e-06,
      "loss": 3.3569,
      "step": 62010
    },
    {
      "epoch": 0.6308492024739584,
      "grad_norm": 8.897919654846191,
      "learning_rate": 1.5037198091738125e-06,
      "loss": 3.1817,
      "step": 62015
    },
    {
      "epoch": 0.6309000651041666,
      "grad_norm": 15.00073528289795,
      "learning_rate": 1.5033530658803357e-06,
      "loss": 3.3084,
      "step": 62020
    },
    {
      "epoch": 0.630950927734375,
      "grad_norm": 12.301580429077148,
      "learning_rate": 1.502986348085835e-06,
      "loss": 3.1554,
      "step": 62025
    },
    {
      "epoch": 0.6310017903645834,
      "grad_norm": 12.632869720458984,
      "learning_rate": 1.5026196557996913e-06,
      "loss": 3.0949,
      "step": 62030
    },
    {
      "epoch": 0.6310526529947916,
      "grad_norm": 16.767946243286133,
      "learning_rate": 1.502252989031288e-06,
      "loss": 3.4847,
      "step": 62035
    },
    {
      "epoch": 0.631103515625,
      "grad_norm": 13.190328598022461,
      "learning_rate": 1.501886347790006e-06,
      "loss": 3.3136,
      "step": 62040
    },
    {
      "epoch": 0.6311543782552084,
      "grad_norm": 10.962267875671387,
      "learning_rate": 1.5015197320852244e-06,
      "loss": 3.6856,
      "step": 62045
    },
    {
      "epoch": 0.6312052408854166,
      "grad_norm": 8.856995582580566,
      "learning_rate": 1.5011531419263231e-06,
      "loss": 3.206,
      "step": 62050
    },
    {
      "epoch": 0.631256103515625,
      "grad_norm": 14.306882858276367,
      "learning_rate": 1.5007865773226831e-06,
      "loss": 3.2216,
      "step": 62055
    },
    {
      "epoch": 0.6313069661458334,
      "grad_norm": 9.836094856262207,
      "learning_rate": 1.5004200382836814e-06,
      "loss": 3.1486,
      "step": 62060
    },
    {
      "epoch": 0.6313578287760416,
      "grad_norm": 10.92822551727295,
      "learning_rate": 1.5000535248186965e-06,
      "loss": 3.336,
      "step": 62065
    },
    {
      "epoch": 0.63140869140625,
      "grad_norm": 16.33973503112793,
      "learning_rate": 1.499687036937104e-06,
      "loss": 3.3093,
      "step": 62070
    },
    {
      "epoch": 0.6314595540364584,
      "grad_norm": 10.504921913146973,
      "learning_rate": 1.4993205746482825e-06,
      "loss": 2.9332,
      "step": 62075
    },
    {
      "epoch": 0.6315104166666666,
      "grad_norm": 8.555520057678223,
      "learning_rate": 1.498954137961607e-06,
      "loss": 3.3795,
      "step": 62080
    },
    {
      "epoch": 0.631561279296875,
      "grad_norm": 17.40530776977539,
      "learning_rate": 1.4985877268864525e-06,
      "loss": 3.3138,
      "step": 62085
    },
    {
      "epoch": 0.6316121419270834,
      "grad_norm": 11.477150917053223,
      "learning_rate": 1.4982213414321928e-06,
      "loss": 3.4017,
      "step": 62090
    },
    {
      "epoch": 0.6316630045572916,
      "grad_norm": 16.35191535949707,
      "learning_rate": 1.497854981608203e-06,
      "loss": 3.5194,
      "step": 62095
    },
    {
      "epoch": 0.6317138671875,
      "grad_norm": 13.10997486114502,
      "learning_rate": 1.4974886474238568e-06,
      "loss": 3.3317,
      "step": 62100
    },
    {
      "epoch": 0.6317647298177084,
      "grad_norm": 9.912026405334473,
      "learning_rate": 1.4971223388885252e-06,
      "loss": 3.213,
      "step": 62105
    },
    {
      "epoch": 0.6318155924479166,
      "grad_norm": 8.097939491271973,
      "learning_rate": 1.4967560560115807e-06,
      "loss": 3.5305,
      "step": 62110
    },
    {
      "epoch": 0.631866455078125,
      "grad_norm": 12.29067611694336,
      "learning_rate": 1.4963897988023952e-06,
      "loss": 3.3737,
      "step": 62115
    },
    {
      "epoch": 0.6319173177083334,
      "grad_norm": 9.925728797912598,
      "learning_rate": 1.4960235672703388e-06,
      "loss": 2.929,
      "step": 62120
    },
    {
      "epoch": 0.6319681803385416,
      "grad_norm": 13.718255996704102,
      "learning_rate": 1.4956573614247804e-06,
      "loss": 3.0586,
      "step": 62125
    },
    {
      "epoch": 0.63201904296875,
      "grad_norm": 12.421064376831055,
      "learning_rate": 1.495291181275092e-06,
      "loss": 3.0733,
      "step": 62130
    },
    {
      "epoch": 0.6320699055989584,
      "grad_norm": 13.674421310424805,
      "learning_rate": 1.4949250268306397e-06,
      "loss": 3.7793,
      "step": 62135
    },
    {
      "epoch": 0.6321207682291666,
      "grad_norm": 10.796311378479004,
      "learning_rate": 1.494558898100793e-06,
      "loss": 3.4028,
      "step": 62140
    },
    {
      "epoch": 0.632171630859375,
      "grad_norm": 9.48852252960205,
      "learning_rate": 1.4941927950949175e-06,
      "loss": 3.5249,
      "step": 62145
    },
    {
      "epoch": 0.6322224934895834,
      "grad_norm": 10.705741882324219,
      "learning_rate": 1.4938267178223814e-06,
      "loss": 3.1707,
      "step": 62150
    },
    {
      "epoch": 0.6322733561197916,
      "grad_norm": 13.889814376831055,
      "learning_rate": 1.4934606662925508e-06,
      "loss": 3.3301,
      "step": 62155
    },
    {
      "epoch": 0.63232421875,
      "grad_norm": 15.42905044555664,
      "learning_rate": 1.49309464051479e-06,
      "loss": 3.2366,
      "step": 62160
    },
    {
      "epoch": 0.6323750813802084,
      "grad_norm": 12.578567504882812,
      "learning_rate": 1.4927286404984637e-06,
      "loss": 3.4207,
      "step": 62165
    },
    {
      "epoch": 0.6324259440104166,
      "grad_norm": 11.376194953918457,
      "learning_rate": 1.4923626662529372e-06,
      "loss": 3.1807,
      "step": 62170
    },
    {
      "epoch": 0.632476806640625,
      "grad_norm": 13.187199592590332,
      "learning_rate": 1.4919967177875726e-06,
      "loss": 3.7483,
      "step": 62175
    },
    {
      "epoch": 0.6325276692708334,
      "grad_norm": 15.951757431030273,
      "learning_rate": 1.4916307951117336e-06,
      "loss": 3.2735,
      "step": 62180
    },
    {
      "epoch": 0.6325785319010416,
      "grad_norm": 15.15351676940918,
      "learning_rate": 1.4912648982347805e-06,
      "loss": 3.0444,
      "step": 62185
    },
    {
      "epoch": 0.63262939453125,
      "grad_norm": 12.16561508178711,
      "learning_rate": 1.4908990271660766e-06,
      "loss": 3.69,
      "step": 62190
    },
    {
      "epoch": 0.6326802571614584,
      "grad_norm": 9.109246253967285,
      "learning_rate": 1.4905331819149827e-06,
      "loss": 3.342,
      "step": 62195
    },
    {
      "epoch": 0.6327311197916666,
      "grad_norm": 16.777318954467773,
      "learning_rate": 1.490167362490857e-06,
      "loss": 3.448,
      "step": 62200
    },
    {
      "epoch": 0.632781982421875,
      "grad_norm": 11.811336517333984,
      "learning_rate": 1.4898015689030598e-06,
      "loss": 3.1386,
      "step": 62205
    },
    {
      "epoch": 0.6328328450520834,
      "grad_norm": 7.091428279876709,
      "learning_rate": 1.4894358011609511e-06,
      "loss": 3.0787,
      "step": 62210
    },
    {
      "epoch": 0.6328837076822916,
      "grad_norm": 9.410467147827148,
      "learning_rate": 1.4890700592738872e-06,
      "loss": 3.7048,
      "step": 62215
    },
    {
      "epoch": 0.6329345703125,
      "grad_norm": 15.245011329650879,
      "learning_rate": 1.4887043432512258e-06,
      "loss": 3.2783,
      "step": 62220
    },
    {
      "epoch": 0.6329854329427084,
      "grad_norm": 11.711705207824707,
      "learning_rate": 1.4883386531023251e-06,
      "loss": 3.6483,
      "step": 62225
    },
    {
      "epoch": 0.6330362955729166,
      "grad_norm": 9.617206573486328,
      "learning_rate": 1.4879729888365395e-06,
      "loss": 3.2444,
      "step": 62230
    },
    {
      "epoch": 0.633087158203125,
      "grad_norm": 17.462865829467773,
      "learning_rate": 1.4876073504632257e-06,
      "loss": 3.5213,
      "step": 62235
    },
    {
      "epoch": 0.6331380208333334,
      "grad_norm": 13.961889266967773,
      "learning_rate": 1.4872417379917365e-06,
      "loss": 3.2226,
      "step": 62240
    },
    {
      "epoch": 0.6331888834635416,
      "grad_norm": 9.276631355285645,
      "learning_rate": 1.4868761514314287e-06,
      "loss": 3.3173,
      "step": 62245
    },
    {
      "epoch": 0.63323974609375,
      "grad_norm": 15.21993350982666,
      "learning_rate": 1.4865105907916539e-06,
      "loss": 3.2775,
      "step": 62250
    },
    {
      "epoch": 0.6332906087239584,
      "grad_norm": 12.131950378417969,
      "learning_rate": 1.4861450560817662e-06,
      "loss": 3.5868,
      "step": 62255
    },
    {
      "epoch": 0.6333414713541666,
      "grad_norm": 13.67236042022705,
      "learning_rate": 1.4857795473111155e-06,
      "loss": 3.2668,
      "step": 62260
    },
    {
      "epoch": 0.633392333984375,
      "grad_norm": 13.7980318069458,
      "learning_rate": 1.4854140644890554e-06,
      "loss": 3.7161,
      "step": 62265
    },
    {
      "epoch": 0.6334431966145834,
      "grad_norm": 10.002974510192871,
      "learning_rate": 1.4850486076249365e-06,
      "loss": 3.2517,
      "step": 62270
    },
    {
      "epoch": 0.6334940592447916,
      "grad_norm": 17.057403564453125,
      "learning_rate": 1.4846831767281082e-06,
      "loss": 3.239,
      "step": 62275
    },
    {
      "epoch": 0.633544921875,
      "grad_norm": 13.9342622756958,
      "learning_rate": 1.4843177718079194e-06,
      "loss": 3.1106,
      "step": 62280
    },
    {
      "epoch": 0.6335957845052084,
      "grad_norm": 9.62298583984375,
      "learning_rate": 1.4839523928737204e-06,
      "loss": 2.8991,
      "step": 62285
    },
    {
      "epoch": 0.6336466471354166,
      "grad_norm": 12.052846908569336,
      "learning_rate": 1.4835870399348584e-06,
      "loss": 3.4887,
      "step": 62290
    },
    {
      "epoch": 0.633697509765625,
      "grad_norm": 8.109212875366211,
      "learning_rate": 1.4832217130006816e-06,
      "loss": 3.2346,
      "step": 62295
    },
    {
      "epoch": 0.6337483723958334,
      "grad_norm": 13.58726692199707,
      "learning_rate": 1.482856412080535e-06,
      "loss": 3.0532,
      "step": 62300
    },
    {
      "epoch": 0.6337992350260416,
      "grad_norm": 10.953584671020508,
      "learning_rate": 1.482491137183767e-06,
      "loss": 3.2216,
      "step": 62305
    },
    {
      "epoch": 0.63385009765625,
      "grad_norm": 14.744987487792969,
      "learning_rate": 1.482125888319722e-06,
      "loss": 3.1238,
      "step": 62310
    },
    {
      "epoch": 0.6339009602864584,
      "grad_norm": 16.00750160217285,
      "learning_rate": 1.4817606654977446e-06,
      "loss": 3.6184,
      "step": 62315
    },
    {
      "epoch": 0.6339518229166666,
      "grad_norm": 11.541016578674316,
      "learning_rate": 1.4813954687271788e-06,
      "loss": 3.4754,
      "step": 62320
    },
    {
      "epoch": 0.634002685546875,
      "grad_norm": 14.393070220947266,
      "learning_rate": 1.4810302980173696e-06,
      "loss": 3.2028,
      "step": 62325
    },
    {
      "epoch": 0.6340535481770834,
      "grad_norm": 7.236337184906006,
      "learning_rate": 1.480665153377658e-06,
      "loss": 3.0415,
      "step": 62330
    },
    {
      "epoch": 0.6341044108072916,
      "grad_norm": 11.734339714050293,
      "learning_rate": 1.4803000348173863e-06,
      "loss": 3.3178,
      "step": 62335
    },
    {
      "epoch": 0.6341552734375,
      "grad_norm": 12.241242408752441,
      "learning_rate": 1.4799349423458979e-06,
      "loss": 3.2239,
      "step": 62340
    },
    {
      "epoch": 0.6342061360677084,
      "grad_norm": 10.239745140075684,
      "learning_rate": 1.4795698759725313e-06,
      "loss": 3.3292,
      "step": 62345
    },
    {
      "epoch": 0.6342569986979166,
      "grad_norm": 10.468929290771484,
      "learning_rate": 1.4792048357066282e-06,
      "loss": 3.1149,
      "step": 62350
    },
    {
      "epoch": 0.634307861328125,
      "grad_norm": 10.030734062194824,
      "learning_rate": 1.4788398215575265e-06,
      "loss": 3.2583,
      "step": 62355
    },
    {
      "epoch": 0.6343587239583334,
      "grad_norm": 8.35238265991211,
      "learning_rate": 1.4784748335345663e-06,
      "loss": 3.4598,
      "step": 62360
    },
    {
      "epoch": 0.6344095865885416,
      "grad_norm": 12.03402042388916,
      "learning_rate": 1.478109871647086e-06,
      "loss": 3.1848,
      "step": 62365
    },
    {
      "epoch": 0.63446044921875,
      "grad_norm": 12.839322090148926,
      "learning_rate": 1.4777449359044217e-06,
      "loss": 3.3202,
      "step": 62370
    },
    {
      "epoch": 0.6345113118489584,
      "grad_norm": 11.045984268188477,
      "learning_rate": 1.4773800263159105e-06,
      "loss": 3.1569,
      "step": 62375
    },
    {
      "epoch": 0.6345621744791666,
      "grad_norm": 9.864595413208008,
      "learning_rate": 1.4770151428908903e-06,
      "loss": 3.1199,
      "step": 62380
    },
    {
      "epoch": 0.634613037109375,
      "grad_norm": 9.273886680603027,
      "learning_rate": 1.4766502856386941e-06,
      "loss": 3.3802,
      "step": 62385
    },
    {
      "epoch": 0.6346638997395834,
      "grad_norm": 14.827187538146973,
      "learning_rate": 1.4762854545686587e-06,
      "loss": 3.652,
      "step": 62390
    },
    {
      "epoch": 0.6347147623697916,
      "grad_norm": 11.817782402038574,
      "learning_rate": 1.4759206496901158e-06,
      "loss": 3.09,
      "step": 62395
    },
    {
      "epoch": 0.634765625,
      "grad_norm": 12.936819076538086,
      "learning_rate": 1.4755558710124012e-06,
      "loss": 3.312,
      "step": 62400
    },
    {
      "epoch": 0.6348164876302084,
      "grad_norm": 8.248379707336426,
      "learning_rate": 1.4751911185448476e-06,
      "loss": 3.3594,
      "step": 62405
    },
    {
      "epoch": 0.6348673502604166,
      "grad_norm": 17.511751174926758,
      "learning_rate": 1.4748263922967854e-06,
      "loss": 3.2554,
      "step": 62410
    },
    {
      "epoch": 0.634918212890625,
      "grad_norm": 14.318390846252441,
      "learning_rate": 1.4744616922775468e-06,
      "loss": 3.3874,
      "step": 62415
    },
    {
      "epoch": 0.6349690755208334,
      "grad_norm": 16.752254486083984,
      "learning_rate": 1.4740970184964631e-06,
      "loss": 3.4619,
      "step": 62420
    },
    {
      "epoch": 0.6350199381510416,
      "grad_norm": 12.39830493927002,
      "learning_rate": 1.4737323709628644e-06,
      "loss": 3.1976,
      "step": 62425
    },
    {
      "epoch": 0.63507080078125,
      "grad_norm": 17.892972946166992,
      "learning_rate": 1.4733677496860789e-06,
      "loss": 3.1798,
      "step": 62430
    },
    {
      "epoch": 0.6351216634114584,
      "grad_norm": 10.058704376220703,
      "learning_rate": 1.4730031546754365e-06,
      "loss": 3.2515,
      "step": 62435
    },
    {
      "epoch": 0.6351725260416666,
      "grad_norm": 7.284869194030762,
      "learning_rate": 1.4726385859402652e-06,
      "loss": 3.6694,
      "step": 62440
    },
    {
      "epoch": 0.635223388671875,
      "grad_norm": 14.861515045166016,
      "learning_rate": 1.4722740434898921e-06,
      "loss": 3.1816,
      "step": 62445
    },
    {
      "epoch": 0.6352742513020834,
      "grad_norm": 13.92569351196289,
      "learning_rate": 1.4719095273336437e-06,
      "loss": 3.2837,
      "step": 62450
    },
    {
      "epoch": 0.6353251139322916,
      "grad_norm": 15.121036529541016,
      "learning_rate": 1.4715450374808469e-06,
      "loss": 3.4223,
      "step": 62455
    },
    {
      "epoch": 0.6353759765625,
      "grad_norm": 10.16884994506836,
      "learning_rate": 1.4711805739408261e-06,
      "loss": 3.3786,
      "step": 62460
    },
    {
      "epoch": 0.6354268391927084,
      "grad_norm": 7.572779178619385,
      "learning_rate": 1.470816136722907e-06,
      "loss": 3.2864,
      "step": 62465
    },
    {
      "epoch": 0.6354777018229166,
      "grad_norm": 14.944366455078125,
      "learning_rate": 1.4704517258364123e-06,
      "loss": 3.4903,
      "step": 62470
    },
    {
      "epoch": 0.635528564453125,
      "grad_norm": 10.901159286499023,
      "learning_rate": 1.4700873412906664e-06,
      "loss": 3.2218,
      "step": 62475
    },
    {
      "epoch": 0.6355794270833334,
      "grad_norm": 14.795011520385742,
      "learning_rate": 1.4697229830949926e-06,
      "loss": 3.0696,
      "step": 62480
    },
    {
      "epoch": 0.6356302897135416,
      "grad_norm": 11.672063827514648,
      "learning_rate": 1.4693586512587113e-06,
      "loss": 3.3608,
      "step": 62485
    },
    {
      "epoch": 0.63568115234375,
      "grad_norm": 11.651619911193848,
      "learning_rate": 1.4689943457911444e-06,
      "loss": 2.9439,
      "step": 62490
    },
    {
      "epoch": 0.6357320149739584,
      "grad_norm": 7.681040287017822,
      "learning_rate": 1.4686300667016135e-06,
      "loss": 3.2055,
      "step": 62495
    },
    {
      "epoch": 0.6357828776041666,
      "grad_norm": 7.397478103637695,
      "learning_rate": 1.4682658139994372e-06,
      "loss": 4.086,
      "step": 62500
    },
    {
      "epoch": 0.635833740234375,
      "grad_norm": 10.454808235168457,
      "learning_rate": 1.4679015876939363e-06,
      "loss": 3.1367,
      "step": 62505
    },
    {
      "epoch": 0.6358846028645834,
      "grad_norm": 15.2985200881958,
      "learning_rate": 1.4675373877944277e-06,
      "loss": 3.6327,
      "step": 62510
    },
    {
      "epoch": 0.6359354654947916,
      "grad_norm": 12.302929878234863,
      "learning_rate": 1.4671732143102308e-06,
      "loss": 3.391,
      "step": 62515
    },
    {
      "epoch": 0.635986328125,
      "grad_norm": 20.90006446838379,
      "learning_rate": 1.4668090672506625e-06,
      "loss": 3.3075,
      "step": 62520
    },
    {
      "epoch": 0.6360371907552084,
      "grad_norm": 14.097554206848145,
      "learning_rate": 1.4664449466250386e-06,
      "loss": 3.181,
      "step": 62525
    },
    {
      "epoch": 0.6360880533854166,
      "grad_norm": 12.83116626739502,
      "learning_rate": 1.4660808524426764e-06,
      "loss": 3.224,
      "step": 62530
    },
    {
      "epoch": 0.636138916015625,
      "grad_norm": 12.021821022033691,
      "learning_rate": 1.4657167847128908e-06,
      "loss": 3.0381,
      "step": 62535
    },
    {
      "epoch": 0.6361897786458334,
      "grad_norm": 9.282414436340332,
      "learning_rate": 1.465352743444996e-06,
      "loss": 2.8872,
      "step": 62540
    },
    {
      "epoch": 0.6362406412760416,
      "grad_norm": 13.512335777282715,
      "learning_rate": 1.464988728648305e-06,
      "loss": 3.324,
      "step": 62545
    },
    {
      "epoch": 0.63629150390625,
      "grad_norm": 8.797607421875,
      "learning_rate": 1.4646247403321334e-06,
      "loss": 3.1153,
      "step": 62550
    },
    {
      "epoch": 0.6363423665364584,
      "grad_norm": 8.574508666992188,
      "learning_rate": 1.4642607785057917e-06,
      "loss": 3.4983,
      "step": 62555
    },
    {
      "epoch": 0.6363932291666666,
      "grad_norm": 9.874283790588379,
      "learning_rate": 1.4638968431785933e-06,
      "loss": 3.7679,
      "step": 62560
    },
    {
      "epoch": 0.636444091796875,
      "grad_norm": 13.190855979919434,
      "learning_rate": 1.4635329343598468e-06,
      "loss": 3.4084,
      "step": 62565
    },
    {
      "epoch": 0.6364949544270834,
      "grad_norm": 9.361830711364746,
      "learning_rate": 1.463169052058866e-06,
      "loss": 3.0899,
      "step": 62570
    },
    {
      "epoch": 0.6365458170572916,
      "grad_norm": 14.177390098571777,
      "learning_rate": 1.4628051962849588e-06,
      "loss": 3.3523,
      "step": 62575
    },
    {
      "epoch": 0.6365966796875,
      "grad_norm": 12.679844856262207,
      "learning_rate": 1.4624413670474355e-06,
      "loss": 3.4776,
      "step": 62580
    },
    {
      "epoch": 0.6366475423177084,
      "grad_norm": 14.361136436462402,
      "learning_rate": 1.4620775643556025e-06,
      "loss": 3.2976,
      "step": 62585
    },
    {
      "epoch": 0.6366984049479166,
      "grad_norm": 10.473655700683594,
      "learning_rate": 1.46171378821877e-06,
      "loss": 3.6659,
      "step": 62590
    },
    {
      "epoch": 0.636749267578125,
      "grad_norm": 8.326841354370117,
      "learning_rate": 1.4613500386462443e-06,
      "loss": 4.0207,
      "step": 62595
    },
    {
      "epoch": 0.6368001302083334,
      "grad_norm": 14.48426342010498,
      "learning_rate": 1.4609863156473315e-06,
      "loss": 2.9788,
      "step": 62600
    },
    {
      "epoch": 0.6368509928385416,
      "grad_norm": 15.910515785217285,
      "learning_rate": 1.4606226192313366e-06,
      "loss": 3.2576,
      "step": 62605
    },
    {
      "epoch": 0.63690185546875,
      "grad_norm": 12.90970230102539,
      "learning_rate": 1.4602589494075667e-06,
      "loss": 3.4674,
      "step": 62610
    },
    {
      "epoch": 0.6369527180989584,
      "grad_norm": 14.598609924316406,
      "learning_rate": 1.459895306185325e-06,
      "loss": 3.3849,
      "step": 62615
    },
    {
      "epoch": 0.6370035807291666,
      "grad_norm": 11.820744514465332,
      "learning_rate": 1.4595316895739148e-06,
      "loss": 3.2215,
      "step": 62620
    },
    {
      "epoch": 0.637054443359375,
      "grad_norm": 13.531217575073242,
      "learning_rate": 1.4591680995826396e-06,
      "loss": 2.81,
      "step": 62625
    },
    {
      "epoch": 0.6371053059895834,
      "grad_norm": 13.425281524658203,
      "learning_rate": 1.4588045362208025e-06,
      "loss": 3.4903,
      "step": 62630
    },
    {
      "epoch": 0.6371561686197916,
      "grad_norm": 13.802509307861328,
      "learning_rate": 1.458440999497705e-06,
      "loss": 3.2356,
      "step": 62635
    },
    {
      "epoch": 0.63720703125,
      "grad_norm": 11.911323547363281,
      "learning_rate": 1.4580774894226466e-06,
      "loss": 3.3578,
      "step": 62640
    },
    {
      "epoch": 0.6372578938802084,
      "grad_norm": 12.430571556091309,
      "learning_rate": 1.4577140060049293e-06,
      "loss": 3.3196,
      "step": 62645
    },
    {
      "epoch": 0.6373087565104166,
      "grad_norm": 9.14650821685791,
      "learning_rate": 1.4573505492538514e-06,
      "loss": 3.2798,
      "step": 62650
    },
    {
      "epoch": 0.637359619140625,
      "grad_norm": 15.641694068908691,
      "learning_rate": 1.4569871191787137e-06,
      "loss": 3.869,
      "step": 62655
    },
    {
      "epoch": 0.6374104817708334,
      "grad_norm": 8.274557113647461,
      "learning_rate": 1.4566237157888122e-06,
      "loss": 3.3859,
      "step": 62660
    },
    {
      "epoch": 0.6374613444010416,
      "grad_norm": 10.886077880859375,
      "learning_rate": 1.4562603390934465e-06,
      "loss": 3.1028,
      "step": 62665
    },
    {
      "epoch": 0.63751220703125,
      "grad_norm": 13.126774787902832,
      "learning_rate": 1.4558969891019126e-06,
      "loss": 3.1486,
      "step": 62670
    },
    {
      "epoch": 0.6375630696614584,
      "grad_norm": 16.401315689086914,
      "learning_rate": 1.4555336658235059e-06,
      "loss": 3.8196,
      "step": 62675
    },
    {
      "epoch": 0.6376139322916666,
      "grad_norm": 7.959024429321289,
      "learning_rate": 1.455170369267523e-06,
      "loss": 3.3189,
      "step": 62680
    },
    {
      "epoch": 0.637664794921875,
      "grad_norm": 7.147455215454102,
      "learning_rate": 1.454807099443259e-06,
      "loss": 3.5084,
      "step": 62685
    },
    {
      "epoch": 0.6377156575520834,
      "grad_norm": 13.789383888244629,
      "learning_rate": 1.454443856360008e-06,
      "loss": 3.4256,
      "step": 62690
    },
    {
      "epoch": 0.6377665201822916,
      "grad_norm": 13.630207061767578,
      "learning_rate": 1.454080640027063e-06,
      "loss": 3.6908,
      "step": 62695
    },
    {
      "epoch": 0.6378173828125,
      "grad_norm": 14.849143028259277,
      "learning_rate": 1.453717450453716e-06,
      "loss": 3.2571,
      "step": 62700
    },
    {
      "epoch": 0.6378682454427084,
      "grad_norm": 8.7376127243042,
      "learning_rate": 1.45335428764926e-06,
      "loss": 3.1046,
      "step": 62705
    },
    {
      "epoch": 0.6379191080729166,
      "grad_norm": 7.985586166381836,
      "learning_rate": 1.4529911516229875e-06,
      "loss": 3.0926,
      "step": 62710
    },
    {
      "epoch": 0.637969970703125,
      "grad_norm": 7.017002582550049,
      "learning_rate": 1.452628042384187e-06,
      "loss": 3.2291,
      "step": 62715
    },
    {
      "epoch": 0.6380208333333334,
      "grad_norm": 14.071551322937012,
      "learning_rate": 1.4522649599421513e-06,
      "loss": 3.2681,
      "step": 62720
    },
    {
      "epoch": 0.6380716959635416,
      "grad_norm": 7.781976699829102,
      "learning_rate": 1.4519019043061666e-06,
      "loss": 3.4062,
      "step": 62725
    },
    {
      "epoch": 0.63812255859375,
      "grad_norm": 9.860779762268066,
      "learning_rate": 1.4515388754855247e-06,
      "loss": 2.7846,
      "step": 62730
    },
    {
      "epoch": 0.6381734212239584,
      "grad_norm": 8.453152656555176,
      "learning_rate": 1.451175873489511e-06,
      "loss": 3.3673,
      "step": 62735
    },
    {
      "epoch": 0.6382242838541666,
      "grad_norm": 12.840231895446777,
      "learning_rate": 1.450812898327415e-06,
      "loss": 3.158,
      "step": 62740
    },
    {
      "epoch": 0.638275146484375,
      "grad_norm": 15.269471168518066,
      "learning_rate": 1.4504499500085223e-06,
      "loss": 3.1418,
      "step": 62745
    },
    {
      "epoch": 0.6383260091145834,
      "grad_norm": 10.599361419677734,
      "learning_rate": 1.4500870285421179e-06,
      "loss": 3.9911,
      "step": 62750
    },
    {
      "epoch": 0.6383768717447916,
      "grad_norm": 7.777949333190918,
      "learning_rate": 1.4497241339374879e-06,
      "loss": 3.1717,
      "step": 62755
    },
    {
      "epoch": 0.638427734375,
      "grad_norm": 11.030150413513184,
      "learning_rate": 1.449361266203918e-06,
      "loss": 3.3687,
      "step": 62760
    },
    {
      "epoch": 0.6384785970052084,
      "grad_norm": 8.590954780578613,
      "learning_rate": 1.4489984253506915e-06,
      "loss": 2.9457,
      "step": 62765
    },
    {
      "epoch": 0.6385294596354166,
      "grad_norm": 12.137688636779785,
      "learning_rate": 1.4486356113870908e-06,
      "loss": 3.5217,
      "step": 62770
    },
    {
      "epoch": 0.638580322265625,
      "grad_norm": 13.380548477172852,
      "learning_rate": 1.4482728243223975e-06,
      "loss": 3.3363,
      "step": 62775
    },
    {
      "epoch": 0.6386311848958334,
      "grad_norm": 13.236571311950684,
      "learning_rate": 1.4479100641658952e-06,
      "loss": 3.1293,
      "step": 62780
    },
    {
      "epoch": 0.6386820475260416,
      "grad_norm": 12.542268753051758,
      "learning_rate": 1.4475473309268652e-06,
      "loss": 3.6334,
      "step": 62785
    },
    {
      "epoch": 0.63873291015625,
      "grad_norm": 16.519969940185547,
      "learning_rate": 1.4471846246145876e-06,
      "loss": 3.5497,
      "step": 62790
    },
    {
      "epoch": 0.6387837727864584,
      "grad_norm": 14.062509536743164,
      "learning_rate": 1.4468219452383408e-06,
      "loss": 3.1123,
      "step": 62795
    },
    {
      "epoch": 0.6388346354166666,
      "grad_norm": 10.798090934753418,
      "learning_rate": 1.4464592928074055e-06,
      "loss": 3.3044,
      "step": 62800
    },
    {
      "epoch": 0.638885498046875,
      "grad_norm": 16.738853454589844,
      "learning_rate": 1.446096667331059e-06,
      "loss": 3.1467,
      "step": 62805
    },
    {
      "epoch": 0.6389363606770834,
      "grad_norm": 10.667718887329102,
      "learning_rate": 1.4457340688185789e-06,
      "loss": 3.2462,
      "step": 62810
    },
    {
      "epoch": 0.6389872233072916,
      "grad_norm": 11.143059730529785,
      "learning_rate": 1.4453714972792443e-06,
      "loss": 3.7527,
      "step": 62815
    },
    {
      "epoch": 0.6390380859375,
      "grad_norm": 15.1730375289917,
      "learning_rate": 1.4450089527223298e-06,
      "loss": 3.5277,
      "step": 62820
    },
    {
      "epoch": 0.6390889485677084,
      "grad_norm": 10.914835929870605,
      "learning_rate": 1.444646435157111e-06,
      "loss": 3.1531,
      "step": 62825
    },
    {
      "epoch": 0.6391398111979166,
      "grad_norm": 7.6904144287109375,
      "learning_rate": 1.4442839445928622e-06,
      "loss": 3.3885,
      "step": 62830
    },
    {
      "epoch": 0.639190673828125,
      "grad_norm": 16.357162475585938,
      "learning_rate": 1.4439214810388582e-06,
      "loss": 3.2952,
      "step": 62835
    },
    {
      "epoch": 0.6392415364583334,
      "grad_norm": 14.014001846313477,
      "learning_rate": 1.443559044504374e-06,
      "loss": 3.3613,
      "step": 62840
    },
    {
      "epoch": 0.6392923990885416,
      "grad_norm": 14.548418045043945,
      "learning_rate": 1.4431966349986814e-06,
      "loss": 3.969,
      "step": 62845
    },
    {
      "epoch": 0.63934326171875,
      "grad_norm": 12.191168785095215,
      "learning_rate": 1.4428342525310513e-06,
      "loss": 3.3481,
      "step": 62850
    },
    {
      "epoch": 0.6393941243489584,
      "grad_norm": 11.43539047241211,
      "learning_rate": 1.4424718971107572e-06,
      "loss": 3.4122,
      "step": 62855
    },
    {
      "epoch": 0.6394449869791666,
      "grad_norm": 13.015048027038574,
      "learning_rate": 1.4421095687470677e-06,
      "loss": 3.1428,
      "step": 62860
    },
    {
      "epoch": 0.639495849609375,
      "grad_norm": 12.808711051940918,
      "learning_rate": 1.4417472674492556e-06,
      "loss": 3.4874,
      "step": 62865
    },
    {
      "epoch": 0.6395467122395834,
      "grad_norm": 11.39872932434082,
      "learning_rate": 1.4413849932265872e-06,
      "loss": 3.9441,
      "step": 62870
    },
    {
      "epoch": 0.6395975748697916,
      "grad_norm": 9.288200378417969,
      "learning_rate": 1.4410227460883345e-06,
      "loss": 3.1887,
      "step": 62875
    },
    {
      "epoch": 0.6396484375,
      "grad_norm": 20.272361755371094,
      "learning_rate": 1.4406605260437626e-06,
      "loss": 3.3801,
      "step": 62880
    },
    {
      "epoch": 0.6396993001302084,
      "grad_norm": 9.169677734375,
      "learning_rate": 1.440298333102141e-06,
      "loss": 3.3967,
      "step": 62885
    },
    {
      "epoch": 0.6397501627604166,
      "grad_norm": 10.119462013244629,
      "learning_rate": 1.4399361672727342e-06,
      "loss": 3.4271,
      "step": 62890
    },
    {
      "epoch": 0.639801025390625,
      "grad_norm": 10.71320629119873,
      "learning_rate": 1.4395740285648109e-06,
      "loss": 3.3772,
      "step": 62895
    },
    {
      "epoch": 0.6398518880208334,
      "grad_norm": 12.625038146972656,
      "learning_rate": 1.439211916987634e-06,
      "loss": 3.303,
      "step": 62900
    },
    {
      "epoch": 0.6399027506510416,
      "grad_norm": 17.40066909790039,
      "learning_rate": 1.4388498325504679e-06,
      "loss": 3.3495,
      "step": 62905
    },
    {
      "epoch": 0.63995361328125,
      "grad_norm": 12.901152610778809,
      "learning_rate": 1.4384877752625776e-06,
      "loss": 3.5543,
      "step": 62910
    },
    {
      "epoch": 0.6400044759114584,
      "grad_norm": 13.61598014831543,
      "learning_rate": 1.438125745133227e-06,
      "loss": 3.5211,
      "step": 62915
    },
    {
      "epoch": 0.6400553385416666,
      "grad_norm": 12.479560852050781,
      "learning_rate": 1.4377637421716773e-06,
      "loss": 3.2052,
      "step": 62920
    },
    {
      "epoch": 0.640106201171875,
      "grad_norm": 13.060637474060059,
      "learning_rate": 1.4374017663871898e-06,
      "loss": 3.3348,
      "step": 62925
    },
    {
      "epoch": 0.6401570638020834,
      "grad_norm": 11.65615463256836,
      "learning_rate": 1.437039817789027e-06,
      "loss": 3.2979,
      "step": 62930
    },
    {
      "epoch": 0.6402079264322916,
      "grad_norm": 11.959769248962402,
      "learning_rate": 1.4366778963864476e-06,
      "loss": 3.5626,
      "step": 62935
    },
    {
      "epoch": 0.6402587890625,
      "grad_norm": 11.341804504394531,
      "learning_rate": 1.436316002188713e-06,
      "loss": 3.4241,
      "step": 62940
    },
    {
      "epoch": 0.6403096516927084,
      "grad_norm": 10.488006591796875,
      "learning_rate": 1.4359541352050805e-06,
      "loss": 3.2683,
      "step": 62945
    },
    {
      "epoch": 0.6403605143229166,
      "grad_norm": 15.101091384887695,
      "learning_rate": 1.4355922954448098e-06,
      "loss": 3.1145,
      "step": 62950
    },
    {
      "epoch": 0.640411376953125,
      "grad_norm": 10.951150894165039,
      "learning_rate": 1.4352304829171587e-06,
      "loss": 3.1962,
      "step": 62955
    },
    {
      "epoch": 0.6404622395833334,
      "grad_norm": 12.33436107635498,
      "learning_rate": 1.4348686976313818e-06,
      "loss": 3.5271,
      "step": 62960
    },
    {
      "epoch": 0.6405131022135416,
      "grad_norm": 11.635231971740723,
      "learning_rate": 1.4345069395967371e-06,
      "loss": 3.8374,
      "step": 62965
    },
    {
      "epoch": 0.64056396484375,
      "grad_norm": 11.820862770080566,
      "learning_rate": 1.4341452088224805e-06,
      "loss": 3.3047,
      "step": 62970
    },
    {
      "epoch": 0.6406148274739584,
      "grad_norm": 9.315909385681152,
      "learning_rate": 1.4337835053178662e-06,
      "loss": 3.003,
      "step": 62975
    },
    {
      "epoch": 0.6406656901041666,
      "grad_norm": 7.3973612785339355,
      "learning_rate": 1.4334218290921483e-06,
      "loss": 3.3779,
      "step": 62980
    },
    {
      "epoch": 0.640716552734375,
      "grad_norm": 9.989119529724121,
      "learning_rate": 1.4330601801545791e-06,
      "loss": 3.3835,
      "step": 62985
    },
    {
      "epoch": 0.6407674153645834,
      "grad_norm": 11.238359451293945,
      "learning_rate": 1.4326985585144121e-06,
      "loss": 2.8967,
      "step": 62990
    },
    {
      "epoch": 0.6408182779947916,
      "grad_norm": 12.609733581542969,
      "learning_rate": 1.432336964180901e-06,
      "loss": 3.661,
      "step": 62995
    },
    {
      "epoch": 0.640869140625,
      "grad_norm": 9.334088325500488,
      "learning_rate": 1.4319753971632948e-06,
      "loss": 3.31,
      "step": 63000
    },
    {
      "epoch": 0.6409200032552084,
      "grad_norm": 15.808244705200195,
      "learning_rate": 1.4316138574708446e-06,
      "loss": 3.6069,
      "step": 63005
    },
    {
      "epoch": 0.6409708658854166,
      "grad_norm": 10.603769302368164,
      "learning_rate": 1.4312523451128013e-06,
      "loss": 3.3133,
      "step": 63010
    },
    {
      "epoch": 0.641021728515625,
      "grad_norm": 10.078536033630371,
      "learning_rate": 1.4308908600984128e-06,
      "loss": 3.278,
      "step": 63015
    },
    {
      "epoch": 0.6410725911458334,
      "grad_norm": 7.120779514312744,
      "learning_rate": 1.4305294024369278e-06,
      "loss": 3.7759,
      "step": 63020
    },
    {
      "epoch": 0.6411234537760416,
      "grad_norm": 13.637188911437988,
      "learning_rate": 1.4301679721375958e-06,
      "loss": 3.2534,
      "step": 63025
    },
    {
      "epoch": 0.64117431640625,
      "grad_norm": 15.969144821166992,
      "learning_rate": 1.4298065692096625e-06,
      "loss": 3.2703,
      "step": 63030
    },
    {
      "epoch": 0.6412251790364584,
      "grad_norm": 14.17587947845459,
      "learning_rate": 1.4294451936623738e-06,
      "loss": 3.5892,
      "step": 63035
    },
    {
      "epoch": 0.6412760416666666,
      "grad_norm": 14.112336158752441,
      "learning_rate": 1.4290838455049754e-06,
      "loss": 3.3573,
      "step": 63040
    },
    {
      "epoch": 0.641326904296875,
      "grad_norm": 9.915677070617676,
      "learning_rate": 1.4287225247467145e-06,
      "loss": 3.4702,
      "step": 63045
    },
    {
      "epoch": 0.6413777669270834,
      "grad_norm": 10.888671875,
      "learning_rate": 1.4283612313968337e-06,
      "loss": 3.2467,
      "step": 63050
    },
    {
      "epoch": 0.6414286295572916,
      "grad_norm": 12.389484405517578,
      "learning_rate": 1.4279999654645771e-06,
      "loss": 3.2647,
      "step": 63055
    },
    {
      "epoch": 0.6414794921875,
      "grad_norm": 11.430819511413574,
      "learning_rate": 1.427638726959186e-06,
      "loss": 3.3085,
      "step": 63060
    },
    {
      "epoch": 0.6415303548177084,
      "grad_norm": 13.8035888671875,
      "learning_rate": 1.4272775158899038e-06,
      "loss": 3.635,
      "step": 63065
    },
    {
      "epoch": 0.6415812174479166,
      "grad_norm": 10.649247169494629,
      "learning_rate": 1.4269163322659734e-06,
      "loss": 3.379,
      "step": 63070
    },
    {
      "epoch": 0.641632080078125,
      "grad_norm": 9.171544075012207,
      "learning_rate": 1.426555176096634e-06,
      "loss": 3.2024,
      "step": 63075
    },
    {
      "epoch": 0.6416829427083334,
      "grad_norm": 8.184263229370117,
      "learning_rate": 1.4261940473911252e-06,
      "loss": 3.0368,
      "step": 63080
    },
    {
      "epoch": 0.6417338053385416,
      "grad_norm": 10.136472702026367,
      "learning_rate": 1.4258329461586879e-06,
      "loss": 3.0165,
      "step": 63085
    },
    {
      "epoch": 0.64178466796875,
      "grad_norm": 17.416135787963867,
      "learning_rate": 1.4254718724085591e-06,
      "loss": 3.9392,
      "step": 63090
    },
    {
      "epoch": 0.6418355305989584,
      "grad_norm": 12.152560234069824,
      "learning_rate": 1.425110826149979e-06,
      "loss": 3.1957,
      "step": 63095
    },
    {
      "epoch": 0.6418863932291666,
      "grad_norm": 223.0073699951172,
      "learning_rate": 1.4247498073921829e-06,
      "loss": 3.7556,
      "step": 63100
    },
    {
      "epoch": 0.641937255859375,
      "grad_norm": 9.165891647338867,
      "learning_rate": 1.4243888161444087e-06,
      "loss": 2.9984,
      "step": 63105
    },
    {
      "epoch": 0.6419881184895834,
      "grad_norm": 10.189545631408691,
      "learning_rate": 1.4240278524158916e-06,
      "loss": 3.3366,
      "step": 63110
    },
    {
      "epoch": 0.6420389811197916,
      "grad_norm": 8.695596694946289,
      "learning_rate": 1.4236669162158662e-06,
      "loss": 3.4958,
      "step": 63115
    },
    {
      "epoch": 0.64208984375,
      "grad_norm": 13.895856857299805,
      "learning_rate": 1.423306007553567e-06,
      "loss": 3.8767,
      "step": 63120
    },
    {
      "epoch": 0.6421407063802084,
      "grad_norm": 14.967219352722168,
      "learning_rate": 1.4229451264382298e-06,
      "loss": 3.4166,
      "step": 63125
    },
    {
      "epoch": 0.6421915690104166,
      "grad_norm": 10.004438400268555,
      "learning_rate": 1.4225842728790866e-06,
      "loss": 3.3435,
      "step": 63130
    },
    {
      "epoch": 0.642242431640625,
      "grad_norm": 15.684747695922852,
      "learning_rate": 1.422223446885368e-06,
      "loss": 3.3918,
      "step": 63135
    },
    {
      "epoch": 0.6422932942708334,
      "grad_norm": 13.166836738586426,
      "learning_rate": 1.421862648466308e-06,
      "loss": 3.2962,
      "step": 63140
    },
    {
      "epoch": 0.6423441569010416,
      "grad_norm": 12.41301441192627,
      "learning_rate": 1.4215018776311359e-06,
      "loss": 3.0552,
      "step": 63145
    },
    {
      "epoch": 0.64239501953125,
      "grad_norm": 13.288503646850586,
      "learning_rate": 1.4211411343890835e-06,
      "loss": 3.0028,
      "step": 63150
    },
    {
      "epoch": 0.6424458821614584,
      "grad_norm": 7.562656879425049,
      "learning_rate": 1.4207804187493784e-06,
      "loss": 3.4307,
      "step": 63155
    },
    {
      "epoch": 0.6424967447916666,
      "grad_norm": 9.266556739807129,
      "learning_rate": 1.4204197307212513e-06,
      "loss": 3.4913,
      "step": 63160
    },
    {
      "epoch": 0.642547607421875,
      "grad_norm": 14.564390182495117,
      "learning_rate": 1.42005907031393e-06,
      "loss": 2.9874,
      "step": 63165
    },
    {
      "epoch": 0.6425984700520834,
      "grad_norm": 12.410235404968262,
      "learning_rate": 1.4196984375366401e-06,
      "loss": 3.2584,
      "step": 63170
    },
    {
      "epoch": 0.6426493326822916,
      "grad_norm": 11.363146781921387,
      "learning_rate": 1.4193378323986096e-06,
      "loss": 3.5505,
      "step": 63175
    },
    {
      "epoch": 0.6427001953125,
      "grad_norm": 14.288817405700684,
      "learning_rate": 1.4189772549090651e-06,
      "loss": 3.0511,
      "step": 63180
    },
    {
      "epoch": 0.6427510579427084,
      "grad_norm": 16.43682098388672,
      "learning_rate": 1.418616705077232e-06,
      "loss": 3.503,
      "step": 63185
    },
    {
      "epoch": 0.6428019205729166,
      "grad_norm": 9.374382972717285,
      "learning_rate": 1.4182561829123332e-06,
      "loss": 3.7444,
      "step": 63190
    },
    {
      "epoch": 0.642852783203125,
      "grad_norm": 7.787618160247803,
      "learning_rate": 1.4178956884235932e-06,
      "loss": 3.3986,
      "step": 63195
    },
    {
      "epoch": 0.6429036458333334,
      "grad_norm": 8.526206970214844,
      "learning_rate": 1.4175352216202364e-06,
      "loss": 3.1476,
      "step": 63200
    },
    {
      "epoch": 0.6429545084635416,
      "grad_norm": 13.395658493041992,
      "learning_rate": 1.4171747825114846e-06,
      "loss": 3.3565,
      "step": 63205
    },
    {
      "epoch": 0.64300537109375,
      "grad_norm": 10.183991432189941,
      "learning_rate": 1.4168143711065585e-06,
      "loss": 3.6162,
      "step": 63210
    },
    {
      "epoch": 0.6430562337239584,
      "grad_norm": 12.942578315734863,
      "learning_rate": 1.4164539874146806e-06,
      "loss": 3.4341,
      "step": 63215
    },
    {
      "epoch": 0.6431070963541666,
      "grad_norm": 14.537596702575684,
      "learning_rate": 1.4160936314450698e-06,
      "loss": 3.3841,
      "step": 63220
    },
    {
      "epoch": 0.643157958984375,
      "grad_norm": 11.549388885498047,
      "learning_rate": 1.4157333032069476e-06,
      "loss": 3.1836,
      "step": 63225
    },
    {
      "epoch": 0.6432088216145834,
      "grad_norm": 16.362728118896484,
      "learning_rate": 1.4153730027095308e-06,
      "loss": 3.3485,
      "step": 63230
    },
    {
      "epoch": 0.6432596842447916,
      "grad_norm": 9.83456802368164,
      "learning_rate": 1.4150127299620398e-06,
      "loss": 3.5327,
      "step": 63235
    },
    {
      "epoch": 0.643310546875,
      "grad_norm": 12.874783515930176,
      "learning_rate": 1.4146524849736906e-06,
      "loss": 3.2493,
      "step": 63240
    },
    {
      "epoch": 0.6433614095052084,
      "grad_norm": 11.884406089782715,
      "learning_rate": 1.4142922677536994e-06,
      "loss": 2.9893,
      "step": 63245
    },
    {
      "epoch": 0.6434122721354166,
      "grad_norm": 7.375535011291504,
      "learning_rate": 1.4139320783112832e-06,
      "loss": 3.9192,
      "step": 63250
    },
    {
      "epoch": 0.643463134765625,
      "grad_norm": 13.060131072998047,
      "learning_rate": 1.4135719166556582e-06,
      "loss": 3.1543,
      "step": 63255
    },
    {
      "epoch": 0.6435139973958334,
      "grad_norm": 11.191900253295898,
      "learning_rate": 1.4132117827960386e-06,
      "loss": 3.5979,
      "step": 63260
    },
    {
      "epoch": 0.6435648600260416,
      "grad_norm": 14.281481742858887,
      "learning_rate": 1.4128516767416378e-06,
      "loss": 3.3245,
      "step": 63265
    },
    {
      "epoch": 0.64361572265625,
      "grad_norm": 12.863600730895996,
      "learning_rate": 1.412491598501668e-06,
      "loss": 3.3724,
      "step": 63270
    },
    {
      "epoch": 0.6436665852864584,
      "grad_norm": 19.69635581970215,
      "learning_rate": 1.412131548085343e-06,
      "loss": 3.481,
      "step": 63275
    },
    {
      "epoch": 0.6437174479166666,
      "grad_norm": 10.664237022399902,
      "learning_rate": 1.4117715255018754e-06,
      "loss": 3.3954,
      "step": 63280
    },
    {
      "epoch": 0.643768310546875,
      "grad_norm": 10.354385375976562,
      "learning_rate": 1.4114115307604756e-06,
      "loss": 3.1319,
      "step": 63285
    },
    {
      "epoch": 0.6438191731770834,
      "grad_norm": 11.787322998046875,
      "learning_rate": 1.4110515638703525e-06,
      "loss": 3.2131,
      "step": 63290
    },
    {
      "epoch": 0.6438700358072916,
      "grad_norm": 9.325935363769531,
      "learning_rate": 1.4106916248407181e-06,
      "loss": 3.0139,
      "step": 63295
    },
    {
      "epoch": 0.6439208984375,
      "grad_norm": 9.582632064819336,
      "learning_rate": 1.4103317136807795e-06,
      "loss": 3.5877,
      "step": 63300
    },
    {
      "epoch": 0.6439717610677084,
      "grad_norm": 8.88012981414795,
      "learning_rate": 1.4099718303997457e-06,
      "loss": 3.0274,
      "step": 63305
    },
    {
      "epoch": 0.6440226236979166,
      "grad_norm": 14.654858589172363,
      "learning_rate": 1.409611975006825e-06,
      "loss": 3.2661,
      "step": 63310
    },
    {
      "epoch": 0.644073486328125,
      "grad_norm": 10.071319580078125,
      "learning_rate": 1.4092521475112237e-06,
      "loss": 3.4717,
      "step": 63315
    },
    {
      "epoch": 0.6441243489583334,
      "grad_norm": 10.545663833618164,
      "learning_rate": 1.4088923479221472e-06,
      "loss": 2.9251,
      "step": 63320
    },
    {
      "epoch": 0.6441752115885416,
      "grad_norm": 10.301362991333008,
      "learning_rate": 1.4085325762488e-06,
      "loss": 3.5179,
      "step": 63325
    },
    {
      "epoch": 0.64422607421875,
      "grad_norm": 15.441143989562988,
      "learning_rate": 1.4081728325003896e-06,
      "loss": 3.718,
      "step": 63330
    },
    {
      "epoch": 0.6442769368489584,
      "grad_norm": 8.218626976013184,
      "learning_rate": 1.407813116686119e-06,
      "loss": 3.2409,
      "step": 63335
    },
    {
      "epoch": 0.6443277994791666,
      "grad_norm": 12.70429515838623,
      "learning_rate": 1.4074534288151906e-06,
      "loss": 3.4641,
      "step": 63340
    },
    {
      "epoch": 0.644378662109375,
      "grad_norm": 10.418517112731934,
      "learning_rate": 1.4070937688968062e-06,
      "loss": 3.0988,
      "step": 63345
    },
    {
      "epoch": 0.6444295247395834,
      "grad_norm": 16.651914596557617,
      "learning_rate": 1.4067341369401688e-06,
      "loss": 3.7684,
      "step": 63350
    },
    {
      "epoch": 0.6444803873697916,
      "grad_norm": 15.939814567565918,
      "learning_rate": 1.4063745329544803e-06,
      "loss": 3.2531,
      "step": 63355
    },
    {
      "epoch": 0.64453125,
      "grad_norm": 13.09463119506836,
      "learning_rate": 1.4060149569489405e-06,
      "loss": 3.3047,
      "step": 63360
    },
    {
      "epoch": 0.6445821126302084,
      "grad_norm": 11.609050750732422,
      "learning_rate": 1.4056554089327474e-06,
      "loss": 3.5869,
      "step": 63365
    },
    {
      "epoch": 0.6446329752604166,
      "grad_norm": 9.671092987060547,
      "learning_rate": 1.4052958889151025e-06,
      "loss": 3.2309,
      "step": 63370
    },
    {
      "epoch": 0.644683837890625,
      "grad_norm": 13.178396224975586,
      "learning_rate": 1.4049363969052016e-06,
      "loss": 3.9849,
      "step": 63375
    },
    {
      "epoch": 0.6447347005208334,
      "grad_norm": 14.983819007873535,
      "learning_rate": 1.4045769329122444e-06,
      "loss": 3.6724,
      "step": 63380
    },
    {
      "epoch": 0.6447855631510416,
      "grad_norm": 8.321818351745605,
      "learning_rate": 1.4042174969454258e-06,
      "loss": 3.1741,
      "step": 63385
    },
    {
      "epoch": 0.64483642578125,
      "grad_norm": 14.03845500946045,
      "learning_rate": 1.4038580890139436e-06,
      "loss": 3.4564,
      "step": 63390
    },
    {
      "epoch": 0.6448872884114584,
      "grad_norm": 16.891016006469727,
      "learning_rate": 1.4034987091269928e-06,
      "loss": 3.4601,
      "step": 63395
    },
    {
      "epoch": 0.6449381510416666,
      "grad_norm": 14.487154960632324,
      "learning_rate": 1.4031393572937662e-06,
      "loss": 3.8728,
      "step": 63400
    },
    {
      "epoch": 0.644989013671875,
      "grad_norm": 9.721774101257324,
      "learning_rate": 1.4027800335234596e-06,
      "loss": 3.5094,
      "step": 63405
    },
    {
      "epoch": 0.6450398763020834,
      "grad_norm": 12.373363494873047,
      "learning_rate": 1.4024207378252664e-06,
      "loss": 3.3832,
      "step": 63410
    },
    {
      "epoch": 0.6450907389322916,
      "grad_norm": 12.942902565002441,
      "learning_rate": 1.4020614702083785e-06,
      "loss": 3.5001,
      "step": 63415
    },
    {
      "epoch": 0.6451416015625,
      "grad_norm": 9.149224281311035,
      "learning_rate": 1.401702230681987e-06,
      "loss": 3.3465,
      "step": 63420
    },
    {
      "epoch": 0.6451924641927084,
      "grad_norm": 7.111023426055908,
      "learning_rate": 1.4013430192552844e-06,
      "loss": 3.4166,
      "step": 63425
    },
    {
      "epoch": 0.6452433268229166,
      "grad_norm": 12.62641716003418,
      "learning_rate": 1.400983835937459e-06,
      "loss": 3.365,
      "step": 63430
    },
    {
      "epoch": 0.645294189453125,
      "grad_norm": 15.175331115722656,
      "learning_rate": 1.4006246807377027e-06,
      "loss": 3.4815,
      "step": 63435
    },
    {
      "epoch": 0.6453450520833334,
      "grad_norm": 10.579191207885742,
      "learning_rate": 1.4002655536652028e-06,
      "loss": 3.4229,
      "step": 63440
    },
    {
      "epoch": 0.6453959147135416,
      "grad_norm": 11.43730640411377,
      "learning_rate": 1.3999064547291485e-06,
      "loss": 3.5058,
      "step": 63445
    },
    {
      "epoch": 0.64544677734375,
      "grad_norm": 11.279643058776855,
      "learning_rate": 1.3995473839387274e-06,
      "loss": 3.1227,
      "step": 63450
    },
    {
      "epoch": 0.6454976399739584,
      "grad_norm": 9.88045883178711,
      "learning_rate": 1.3991883413031243e-06,
      "loss": 3.4518,
      "step": 63455
    },
    {
      "epoch": 0.6455485026041666,
      "grad_norm": 12.620511054992676,
      "learning_rate": 1.3988293268315267e-06,
      "loss": 3.3055,
      "step": 63460
    },
    {
      "epoch": 0.645599365234375,
      "grad_norm": 10.935976028442383,
      "learning_rate": 1.3984703405331204e-06,
      "loss": 3.2843,
      "step": 63465
    },
    {
      "epoch": 0.6456502278645834,
      "grad_norm": 12.42562198638916,
      "learning_rate": 1.3981113824170896e-06,
      "loss": 3.1433,
      "step": 63470
    },
    {
      "epoch": 0.6457010904947916,
      "grad_norm": 8.379152297973633,
      "learning_rate": 1.3977524524926177e-06,
      "loss": 3.2988,
      "step": 63475
    },
    {
      "epoch": 0.645751953125,
      "grad_norm": 10.612476348876953,
      "learning_rate": 1.3973935507688866e-06,
      "loss": 2.9124,
      "step": 63480
    },
    {
      "epoch": 0.6458028157552084,
      "grad_norm": 13.055747032165527,
      "learning_rate": 1.3970346772550819e-06,
      "loss": 3.6181,
      "step": 63485
    },
    {
      "epoch": 0.6458536783854166,
      "grad_norm": 15.127704620361328,
      "learning_rate": 1.3966758319603835e-06,
      "loss": 3.6793,
      "step": 63490
    },
    {
      "epoch": 0.645904541015625,
      "grad_norm": 14.916153907775879,
      "learning_rate": 1.3963170148939719e-06,
      "loss": 3.3365,
      "step": 63495
    },
    {
      "epoch": 0.6459554036458334,
      "grad_norm": 13.672470092773438,
      "learning_rate": 1.395958226065028e-06,
      "loss": 3.5576,
      "step": 63500
    },
    {
      "epoch": 0.6460062662760416,
      "grad_norm": 10.827059745788574,
      "learning_rate": 1.3955994654827303e-06,
      "loss": 3.5075,
      "step": 63505
    },
    {
      "epoch": 0.64605712890625,
      "grad_norm": 12.494633674621582,
      "learning_rate": 1.3952407331562594e-06,
      "loss": 3.3687,
      "step": 63510
    },
    {
      "epoch": 0.6461079915364584,
      "grad_norm": 7.683717727661133,
      "learning_rate": 1.394882029094792e-06,
      "loss": 3.5266,
      "step": 63515
    },
    {
      "epoch": 0.6461588541666666,
      "grad_norm": 10.356765747070312,
      "learning_rate": 1.394523353307507e-06,
      "loss": 3.0642,
      "step": 63520
    },
    {
      "epoch": 0.646209716796875,
      "grad_norm": 9.716863632202148,
      "learning_rate": 1.3941647058035796e-06,
      "loss": 3.4504,
      "step": 63525
    },
    {
      "epoch": 0.6462605794270834,
      "grad_norm": 14.333846092224121,
      "learning_rate": 1.3938060865921855e-06,
      "loss": 3.1182,
      "step": 63530
    },
    {
      "epoch": 0.6463114420572916,
      "grad_norm": 12.775233268737793,
      "learning_rate": 1.3934474956825e-06,
      "loss": 3.0615,
      "step": 63535
    },
    {
      "epoch": 0.6463623046875,
      "grad_norm": 11.847692489624023,
      "learning_rate": 1.3930889330836994e-06,
      "loss": 3.3087,
      "step": 63540
    },
    {
      "epoch": 0.6464131673177084,
      "grad_norm": 13.588580131530762,
      "learning_rate": 1.392730398804956e-06,
      "loss": 3.6494,
      "step": 63545
    },
    {
      "epoch": 0.6464640299479166,
      "grad_norm": 9.224186897277832,
      "learning_rate": 1.3923718928554428e-06,
      "loss": 3.1022,
      "step": 63550
    },
    {
      "epoch": 0.646514892578125,
      "grad_norm": 8.812080383300781,
      "learning_rate": 1.392013415244331e-06,
      "loss": 3.3728,
      "step": 63555
    },
    {
      "epoch": 0.6465657552083334,
      "grad_norm": 15.281486511230469,
      "learning_rate": 1.3916549659807938e-06,
      "loss": 3.4459,
      "step": 63560
    },
    {
      "epoch": 0.6466166178385416,
      "grad_norm": 12.427238464355469,
      "learning_rate": 1.3912965450740023e-06,
      "loss": 3.3156,
      "step": 63565
    },
    {
      "epoch": 0.64666748046875,
      "grad_norm": 15.172652244567871,
      "learning_rate": 1.3909381525331261e-06,
      "loss": 3.3442,
      "step": 63570
    },
    {
      "epoch": 0.6467183430989584,
      "grad_norm": 14.54151725769043,
      "learning_rate": 1.3905797883673332e-06,
      "loss": 3.1957,
      "step": 63575
    },
    {
      "epoch": 0.6467692057291666,
      "grad_norm": 14.314568519592285,
      "learning_rate": 1.3902214525857944e-06,
      "loss": 3.444,
      "step": 63580
    },
    {
      "epoch": 0.646820068359375,
      "grad_norm": 11.108024597167969,
      "learning_rate": 1.3898631451976757e-06,
      "loss": 3.1377,
      "step": 63585
    },
    {
      "epoch": 0.6468709309895834,
      "grad_norm": 13.669203758239746,
      "learning_rate": 1.3895048662121467e-06,
      "loss": 3.4922,
      "step": 63590
    },
    {
      "epoch": 0.6469217936197916,
      "grad_norm": 11.712994575500488,
      "learning_rate": 1.389146615638371e-06,
      "loss": 3.1179,
      "step": 63595
    },
    {
      "epoch": 0.64697265625,
      "grad_norm": 12.378658294677734,
      "learning_rate": 1.3887883934855173e-06,
      "loss": 3.9974,
      "step": 63600
    },
    {
      "epoch": 0.6470235188802084,
      "grad_norm": 10.653694152832031,
      "learning_rate": 1.3884301997627486e-06,
      "loss": 3.0061,
      "step": 63605
    },
    {
      "epoch": 0.6470743815104166,
      "grad_norm": 9.582548141479492,
      "learning_rate": 1.3880720344792293e-06,
      "loss": 3.8848,
      "step": 63610
    },
    {
      "epoch": 0.647125244140625,
      "grad_norm": 12.967094421386719,
      "learning_rate": 1.3877138976441228e-06,
      "loss": 3.3185,
      "step": 63615
    },
    {
      "epoch": 0.6471761067708334,
      "grad_norm": 11.003528594970703,
      "learning_rate": 1.3873557892665942e-06,
      "loss": 3.3306,
      "step": 63620
    },
    {
      "epoch": 0.6472269694010416,
      "grad_norm": 15.506131172180176,
      "learning_rate": 1.386997709355804e-06,
      "loss": 2.9387,
      "step": 63625
    },
    {
      "epoch": 0.64727783203125,
      "grad_norm": 9.115920066833496,
      "learning_rate": 1.3866396579209124e-06,
      "loss": 3.1316,
      "step": 63630
    },
    {
      "epoch": 0.6473286946614584,
      "grad_norm": 13.750835418701172,
      "learning_rate": 1.3862816349710812e-06,
      "loss": 3.2268,
      "step": 63635
    },
    {
      "epoch": 0.6473795572916666,
      "grad_norm": 14.21649169921875,
      "learning_rate": 1.3859236405154714e-06,
      "loss": 3.177,
      "step": 63640
    },
    {
      "epoch": 0.647430419921875,
      "grad_norm": 13.725008964538574,
      "learning_rate": 1.3855656745632412e-06,
      "loss": 3.3347,
      "step": 63645
    },
    {
      "epoch": 0.6474812825520834,
      "grad_norm": 11.765166282653809,
      "learning_rate": 1.3852077371235481e-06,
      "loss": 3.3822,
      "step": 63650
    },
    {
      "epoch": 0.6475321451822916,
      "grad_norm": 14.484723091125488,
      "learning_rate": 1.3848498282055518e-06,
      "loss": 3.5759,
      "step": 63655
    },
    {
      "epoch": 0.6475830078125,
      "grad_norm": 13.23550796508789,
      "learning_rate": 1.3844919478184072e-06,
      "loss": 3.4329,
      "step": 63660
    },
    {
      "epoch": 0.6476338704427084,
      "grad_norm": 13.231417655944824,
      "learning_rate": 1.384134095971273e-06,
      "loss": 3.3019,
      "step": 63665
    },
    {
      "epoch": 0.6476847330729166,
      "grad_norm": 8.883193016052246,
      "learning_rate": 1.3837762726733024e-06,
      "loss": 3.5572,
      "step": 63670
    },
    {
      "epoch": 0.647735595703125,
      "grad_norm": 15.052298545837402,
      "learning_rate": 1.383418477933652e-06,
      "loss": 3.5487,
      "step": 63675
    },
    {
      "epoch": 0.6477864583333334,
      "grad_norm": 13.665106773376465,
      "learning_rate": 1.3830607117614752e-06,
      "loss": 3.5822,
      "step": 63680
    },
    {
      "epoch": 0.6478373209635416,
      "grad_norm": 10.003432273864746,
      "learning_rate": 1.3827029741659248e-06,
      "loss": 3.4682,
      "step": 63685
    },
    {
      "epoch": 0.64788818359375,
      "grad_norm": 12.446817398071289,
      "learning_rate": 1.3823452651561536e-06,
      "loss": 3.4898,
      "step": 63690
    },
    {
      "epoch": 0.6479390462239584,
      "grad_norm": 8.958677291870117,
      "learning_rate": 1.381987584741315e-06,
      "loss": 3.4242,
      "step": 63695
    },
    {
      "epoch": 0.6479899088541666,
      "grad_norm": 9.530407905578613,
      "learning_rate": 1.381629932930559e-06,
      "loss": 3.4295,
      "step": 63700
    },
    {
      "epoch": 0.648040771484375,
      "grad_norm": 10.85719108581543,
      "learning_rate": 1.3812723097330353e-06,
      "loss": 3.2851,
      "step": 63705
    },
    {
      "epoch": 0.6480916341145834,
      "grad_norm": 10.73904037475586,
      "learning_rate": 1.380914715157895e-06,
      "loss": 3.4357,
      "step": 63710
    },
    {
      "epoch": 0.6481424967447916,
      "grad_norm": 12.596543312072754,
      "learning_rate": 1.3805571492142858e-06,
      "loss": 3.1528,
      "step": 63715
    },
    {
      "epoch": 0.648193359375,
      "grad_norm": 14.61168384552002,
      "learning_rate": 1.3801996119113576e-06,
      "loss": 3.0919,
      "step": 63720
    },
    {
      "epoch": 0.6482442220052084,
      "grad_norm": 10.773356437683105,
      "learning_rate": 1.3798421032582557e-06,
      "loss": 3.2884,
      "step": 63725
    },
    {
      "epoch": 0.6482950846354166,
      "grad_norm": 7.623642921447754,
      "learning_rate": 1.3794846232641291e-06,
      "loss": 3.6828,
      "step": 63730
    },
    {
      "epoch": 0.648345947265625,
      "grad_norm": 14.701728820800781,
      "learning_rate": 1.3791271719381227e-06,
      "loss": 3.4508,
      "step": 63735
    },
    {
      "epoch": 0.6483968098958334,
      "grad_norm": 14.30395221710205,
      "learning_rate": 1.3787697492893815e-06,
      "loss": 3.3275,
      "step": 63740
    },
    {
      "epoch": 0.6484476725260416,
      "grad_norm": 13.148125648498535,
      "learning_rate": 1.37841235532705e-06,
      "loss": 3.4249,
      "step": 63745
    },
    {
      "epoch": 0.64849853515625,
      "grad_norm": 17.876314163208008,
      "learning_rate": 1.3780549900602736e-06,
      "loss": 3.1441,
      "step": 63750
    },
    {
      "epoch": 0.6485493977864584,
      "grad_norm": 12.513371467590332,
      "learning_rate": 1.3776976534981945e-06,
      "loss": 3.1713,
      "step": 63755
    },
    {
      "epoch": 0.6486002604166666,
      "grad_norm": 10.822637557983398,
      "learning_rate": 1.3773403456499545e-06,
      "loss": 3.5461,
      "step": 63760
    },
    {
      "epoch": 0.648651123046875,
      "grad_norm": 12.970320701599121,
      "learning_rate": 1.3769830665246953e-06,
      "loss": 3.0942,
      "step": 63765
    },
    {
      "epoch": 0.6487019856770834,
      "grad_norm": 10.76432991027832,
      "learning_rate": 1.3766258161315577e-06,
      "loss": 2.9625,
      "step": 63770
    },
    {
      "epoch": 0.6487528483072916,
      "grad_norm": 8.161712646484375,
      "learning_rate": 1.3762685944796833e-06,
      "loss": 3.476,
      "step": 63775
    },
    {
      "epoch": 0.6488037109375,
      "grad_norm": 11.488871574401855,
      "learning_rate": 1.375911401578211e-06,
      "loss": 3.3814,
      "step": 63780
    },
    {
      "epoch": 0.6488545735677084,
      "grad_norm": 6.997217655181885,
      "learning_rate": 1.3755542374362776e-06,
      "loss": 3.1818,
      "step": 63785
    },
    {
      "epoch": 0.6489054361979166,
      "grad_norm": 14.117745399475098,
      "learning_rate": 1.375197102063023e-06,
      "loss": 3.3206,
      "step": 63790
    },
    {
      "epoch": 0.648956298828125,
      "grad_norm": 9.993836402893066,
      "learning_rate": 1.3748399954675845e-06,
      "loss": 3.5025,
      "step": 63795
    },
    {
      "epoch": 0.6490071614583334,
      "grad_norm": 11.988035202026367,
      "learning_rate": 1.374482917659097e-06,
      "loss": 3.3537,
      "step": 63800
    },
    {
      "epoch": 0.6490580240885416,
      "grad_norm": 10.683988571166992,
      "learning_rate": 1.3741258686466986e-06,
      "loss": 3.2564,
      "step": 63805
    },
    {
      "epoch": 0.64910888671875,
      "grad_norm": 13.883316040039062,
      "learning_rate": 1.3737688484395228e-06,
      "loss": 3.3515,
      "step": 63810
    },
    {
      "epoch": 0.6491597493489584,
      "grad_norm": 11.055418968200684,
      "learning_rate": 1.373411857046703e-06,
      "loss": 3.1519,
      "step": 63815
    },
    {
      "epoch": 0.6492106119791666,
      "grad_norm": 13.652414321899414,
      "learning_rate": 1.3730548944773739e-06,
      "loss": 3.3909,
      "step": 63820
    },
    {
      "epoch": 0.649261474609375,
      "grad_norm": 13.452717781066895,
      "learning_rate": 1.3726979607406693e-06,
      "loss": 3.365,
      "step": 63825
    },
    {
      "epoch": 0.6493123372395834,
      "grad_norm": 10.121183395385742,
      "learning_rate": 1.3723410558457202e-06,
      "loss": 3.2002,
      "step": 63830
    },
    {
      "epoch": 0.6493631998697916,
      "grad_norm": 13.718680381774902,
      "learning_rate": 1.3719841798016574e-06,
      "loss": 3.0125,
      "step": 63835
    },
    {
      "epoch": 0.6494140625,
      "grad_norm": 11.836580276489258,
      "learning_rate": 1.3716273326176117e-06,
      "loss": 3.7457,
      "step": 63840
    },
    {
      "epoch": 0.6494649251302084,
      "grad_norm": 9.903429985046387,
      "learning_rate": 1.3712705143027127e-06,
      "loss": 3.2571,
      "step": 63845
    },
    {
      "epoch": 0.6495157877604166,
      "grad_norm": 14.779627799987793,
      "learning_rate": 1.370913724866091e-06,
      "loss": 3.1684,
      "step": 63850
    },
    {
      "epoch": 0.649566650390625,
      "grad_norm": 7.905979156494141,
      "learning_rate": 1.3705569643168746e-06,
      "loss": 3.6488,
      "step": 63855
    },
    {
      "epoch": 0.6496175130208334,
      "grad_norm": 15.816743850708008,
      "learning_rate": 1.3702002326641893e-06,
      "loss": 3.2197,
      "step": 63860
    },
    {
      "epoch": 0.6496683756510416,
      "grad_norm": 10.404696464538574,
      "learning_rate": 1.3698435299171641e-06,
      "loss": 3.0273,
      "step": 63865
    },
    {
      "epoch": 0.64971923828125,
      "grad_norm": 13.25161361694336,
      "learning_rate": 1.3694868560849235e-06,
      "loss": 3.3951,
      "step": 63870
    },
    {
      "epoch": 0.6497701009114584,
      "grad_norm": 10.754583358764648,
      "learning_rate": 1.3691302111765945e-06,
      "loss": 3.5149,
      "step": 63875
    },
    {
      "epoch": 0.6498209635416666,
      "grad_norm": 10.602897644042969,
      "learning_rate": 1.3687735952013e-06,
      "loss": 2.9521,
      "step": 63880
    },
    {
      "epoch": 0.649871826171875,
      "grad_norm": 11.862292289733887,
      "learning_rate": 1.3684170081681663e-06,
      "loss": 3.2957,
      "step": 63885
    },
    {
      "epoch": 0.6499226888020834,
      "grad_norm": 8.14902400970459,
      "learning_rate": 1.3680604500863147e-06,
      "loss": 3.073,
      "step": 63890
    },
    {
      "epoch": 0.6499735514322916,
      "grad_norm": 9.84872817993164,
      "learning_rate": 1.3677039209648676e-06,
      "loss": 3.3011,
      "step": 63895
    },
    {
      "epoch": 0.6500244140625,
      "grad_norm": 10.788445472717285,
      "learning_rate": 1.3673474208129472e-06,
      "loss": 3.2758,
      "step": 63900
    },
    {
      "epoch": 0.6500752766927084,
      "grad_norm": 10.837233543395996,
      "learning_rate": 1.3669909496396756e-06,
      "loss": 3.5907,
      "step": 63905
    },
    {
      "epoch": 0.6501261393229166,
      "grad_norm": 7.292312145233154,
      "learning_rate": 1.366634507454172e-06,
      "loss": 3.4772,
      "step": 63910
    },
    {
      "epoch": 0.650177001953125,
      "grad_norm": 12.889204978942871,
      "learning_rate": 1.3662780942655549e-06,
      "loss": 3.3667,
      "step": 63915
    },
    {
      "epoch": 0.6502278645833334,
      "grad_norm": 11.986324310302734,
      "learning_rate": 1.3659217100829447e-06,
      "loss": 3.0141,
      "step": 63920
    },
    {
      "epoch": 0.6502787272135416,
      "grad_norm": 11.964778900146484,
      "learning_rate": 1.3655653549154582e-06,
      "loss": 2.9396,
      "step": 63925
    },
    {
      "epoch": 0.65032958984375,
      "grad_norm": 17.04004669189453,
      "learning_rate": 1.365209028772214e-06,
      "loss": 4.0778,
      "step": 63930
    },
    {
      "epoch": 0.6503804524739584,
      "grad_norm": 10.48777961730957,
      "learning_rate": 1.364852731662327e-06,
      "loss": 3.3832,
      "step": 63935
    },
    {
      "epoch": 0.6504313151041666,
      "grad_norm": 12.528505325317383,
      "learning_rate": 1.3644964635949147e-06,
      "loss": 3.5456,
      "step": 63940
    },
    {
      "epoch": 0.650482177734375,
      "grad_norm": 16.32647705078125,
      "learning_rate": 1.3641402245790902e-06,
      "loss": 3.1487,
      "step": 63945
    },
    {
      "epoch": 0.6505330403645834,
      "grad_norm": 13.028712272644043,
      "learning_rate": 1.3637840146239698e-06,
      "loss": 3.0298,
      "step": 63950
    },
    {
      "epoch": 0.6505839029947916,
      "grad_norm": 13.009231567382812,
      "learning_rate": 1.363427833738665e-06,
      "loss": 3.388,
      "step": 63955
    },
    {
      "epoch": 0.650634765625,
      "grad_norm": 11.50768756866455,
      "learning_rate": 1.3630716819322908e-06,
      "loss": 3.3371,
      "step": 63960
    },
    {
      "epoch": 0.6506856282552084,
      "grad_norm": 10.840215682983398,
      "learning_rate": 1.3627155592139579e-06,
      "loss": 3.0109,
      "step": 63965
    },
    {
      "epoch": 0.6507364908854166,
      "grad_norm": 8.058030128479004,
      "learning_rate": 1.3623594655927775e-06,
      "loss": 3.0849,
      "step": 63970
    },
    {
      "epoch": 0.650787353515625,
      "grad_norm": 12.560381889343262,
      "learning_rate": 1.36200340107786e-06,
      "loss": 3.0619,
      "step": 63975
    },
    {
      "epoch": 0.6508382161458334,
      "grad_norm": 11.920454025268555,
      "learning_rate": 1.3616473656783167e-06,
      "loss": 3.2486,
      "step": 63980
    },
    {
      "epoch": 0.6508890787760416,
      "grad_norm": 14.832955360412598,
      "learning_rate": 1.3612913594032555e-06,
      "loss": 3.0761,
      "step": 63985
    },
    {
      "epoch": 0.65093994140625,
      "grad_norm": 11.608518600463867,
      "learning_rate": 1.3609353822617855e-06,
      "loss": 3.1821,
      "step": 63990
    },
    {
      "epoch": 0.6509908040364584,
      "grad_norm": 9.848051071166992,
      "learning_rate": 1.3605794342630127e-06,
      "loss": 3.1642,
      "step": 63995
    },
    {
      "epoch": 0.6510416666666666,
      "grad_norm": 12.472249984741211,
      "learning_rate": 1.3602235154160448e-06,
      "loss": 3.4782,
      "step": 64000
    },
    {
      "epoch": 0.651092529296875,
      "grad_norm": 12.055642127990723,
      "learning_rate": 1.3598676257299892e-06,
      "loss": 3.3443,
      "step": 64005
    },
    {
      "epoch": 0.6511433919270834,
      "grad_norm": 11.859967231750488,
      "learning_rate": 1.3595117652139492e-06,
      "loss": 3.3312,
      "step": 64010
    },
    {
      "epoch": 0.6511942545572916,
      "grad_norm": 15.819765090942383,
      "learning_rate": 1.3591559338770312e-06,
      "loss": 3.5675,
      "step": 64015
    },
    {
      "epoch": 0.6512451171875,
      "grad_norm": 10.581780433654785,
      "learning_rate": 1.3588001317283384e-06,
      "loss": 3.2153,
      "step": 64020
    },
    {
      "epoch": 0.6512959798177084,
      "grad_norm": 12.730134963989258,
      "learning_rate": 1.3584443587769727e-06,
      "loss": 3.4063,
      "step": 64025
    },
    {
      "epoch": 0.6513468424479166,
      "grad_norm": 15.363131523132324,
      "learning_rate": 1.3580886150320374e-06,
      "loss": 3.0632,
      "step": 64030
    },
    {
      "epoch": 0.651397705078125,
      "grad_norm": 14.5416898727417,
      "learning_rate": 1.3577329005026351e-06,
      "loss": 3.3405,
      "step": 64035
    },
    {
      "epoch": 0.6514485677083334,
      "grad_norm": 15.410543441772461,
      "learning_rate": 1.3573772151978662e-06,
      "loss": 3.359,
      "step": 64040
    },
    {
      "epoch": 0.6514994303385416,
      "grad_norm": 12.796439170837402,
      "learning_rate": 1.3570215591268303e-06,
      "loss": 3.0914,
      "step": 64045
    },
    {
      "epoch": 0.65155029296875,
      "grad_norm": 10.722402572631836,
      "learning_rate": 1.3566659322986259e-06,
      "loss": 3.5775,
      "step": 64050
    },
    {
      "epoch": 0.6516011555989584,
      "grad_norm": 12.038302421569824,
      "learning_rate": 1.3563103347223523e-06,
      "loss": 3.0707,
      "step": 64055
    },
    {
      "epoch": 0.6516520182291666,
      "grad_norm": 14.053557395935059,
      "learning_rate": 1.3559547664071089e-06,
      "loss": 3.3114,
      "step": 64060
    },
    {
      "epoch": 0.651702880859375,
      "grad_norm": 11.177879333496094,
      "learning_rate": 1.3555992273619916e-06,
      "loss": 3.267,
      "step": 64065
    },
    {
      "epoch": 0.6517537434895834,
      "grad_norm": 12.664907455444336,
      "learning_rate": 1.3552437175960959e-06,
      "loss": 3.7841,
      "step": 64070
    },
    {
      "epoch": 0.6518046061197916,
      "grad_norm": 11.753214836120605,
      "learning_rate": 1.3548882371185195e-06,
      "loss": 3.5764,
      "step": 64075
    },
    {
      "epoch": 0.65185546875,
      "grad_norm": 10.01693344116211,
      "learning_rate": 1.354532785938355e-06,
      "loss": 3.4242,
      "step": 64080
    },
    {
      "epoch": 0.6519063313802084,
      "grad_norm": 10.688921928405762,
      "learning_rate": 1.3541773640646982e-06,
      "loss": 3.4629,
      "step": 64085
    },
    {
      "epoch": 0.6519571940104166,
      "grad_norm": 16.974910736083984,
      "learning_rate": 1.3538219715066414e-06,
      "loss": 3.2687,
      "step": 64090
    },
    {
      "epoch": 0.652008056640625,
      "grad_norm": 7.56576681137085,
      "learning_rate": 1.353466608273279e-06,
      "loss": 3.6427,
      "step": 64095
    },
    {
      "epoch": 0.6520589192708334,
      "grad_norm": 9.493888854980469,
      "learning_rate": 1.3531112743737002e-06,
      "loss": 3.6024,
      "step": 64100
    },
    {
      "epoch": 0.6521097819010416,
      "grad_norm": 8.88723373413086,
      "learning_rate": 1.3527559698169978e-06,
      "loss": 3.1188,
      "step": 64105
    },
    {
      "epoch": 0.65216064453125,
      "grad_norm": 9.895440101623535,
      "learning_rate": 1.352400694612263e-06,
      "loss": 3.0877,
      "step": 64110
    },
    {
      "epoch": 0.6522115071614584,
      "grad_norm": 8.68016529083252,
      "learning_rate": 1.3520454487685846e-06,
      "loss": 3.6815,
      "step": 64115
    },
    {
      "epoch": 0.6522623697916666,
      "grad_norm": 9.342655181884766,
      "learning_rate": 1.3516902322950508e-06,
      "loss": 3.3618,
      "step": 64120
    },
    {
      "epoch": 0.652313232421875,
      "grad_norm": 7.545049667358398,
      "learning_rate": 1.3513350452007498e-06,
      "loss": 3.3348,
      "step": 64125
    },
    {
      "epoch": 0.6523640950520834,
      "grad_norm": 12.7225341796875,
      "learning_rate": 1.3509798874947694e-06,
      "loss": 3.3068,
      "step": 64130
    },
    {
      "epoch": 0.6524149576822916,
      "grad_norm": 9.57697582244873,
      "learning_rate": 1.3506247591861969e-06,
      "loss": 3.4641,
      "step": 64135
    },
    {
      "epoch": 0.6524658203125,
      "grad_norm": 12.992835998535156,
      "learning_rate": 1.3502696602841176e-06,
      "loss": 3.3959,
      "step": 64140
    },
    {
      "epoch": 0.6525166829427084,
      "grad_norm": 9.261260986328125,
      "learning_rate": 1.3499145907976153e-06,
      "loss": 3.369,
      "step": 64145
    },
    {
      "epoch": 0.6525675455729166,
      "grad_norm": 14.930026054382324,
      "learning_rate": 1.3495595507357766e-06,
      "loss": 3.5845,
      "step": 64150
    },
    {
      "epoch": 0.652618408203125,
      "grad_norm": 7.567656993865967,
      "learning_rate": 1.3492045401076836e-06,
      "loss": 3.0222,
      "step": 64155
    },
    {
      "epoch": 0.6526692708333334,
      "grad_norm": 7.853155136108398,
      "learning_rate": 1.3488495589224205e-06,
      "loss": 3.2456,
      "step": 64160
    },
    {
      "epoch": 0.6527201334635416,
      "grad_norm": 10.342120170593262,
      "learning_rate": 1.3484946071890676e-06,
      "loss": 3.434,
      "step": 64165
    },
    {
      "epoch": 0.65277099609375,
      "grad_norm": 10.163643836975098,
      "learning_rate": 1.3481396849167078e-06,
      "loss": 3.0487,
      "step": 64170
    },
    {
      "epoch": 0.6528218587239584,
      "grad_norm": 11.490717887878418,
      "learning_rate": 1.3477847921144216e-06,
      "loss": 3.6839,
      "step": 64175
    },
    {
      "epoch": 0.6528727213541666,
      "grad_norm": 12.72451400756836,
      "learning_rate": 1.3474299287912873e-06,
      "loss": 3.5324,
      "step": 64180
    },
    {
      "epoch": 0.652923583984375,
      "grad_norm": 11.382487297058105,
      "learning_rate": 1.347075094956385e-06,
      "loss": 3.1754,
      "step": 64185
    },
    {
      "epoch": 0.6529744466145834,
      "grad_norm": 8.600659370422363,
      "learning_rate": 1.346720290618794e-06,
      "loss": 3.6986,
      "step": 64190
    },
    {
      "epoch": 0.6530253092447916,
      "grad_norm": 9.42199420928955,
      "learning_rate": 1.346365515787591e-06,
      "loss": 3.273,
      "step": 64195
    },
    {
      "epoch": 0.653076171875,
      "grad_norm": 11.763266563415527,
      "learning_rate": 1.3460107704718522e-06,
      "loss": 3.3125,
      "step": 64200
    },
    {
      "epoch": 0.6531270345052084,
      "grad_norm": 10.710270881652832,
      "learning_rate": 1.3456560546806553e-06,
      "loss": 3.0234,
      "step": 64205
    },
    {
      "epoch": 0.6531778971354166,
      "grad_norm": 8.050313949584961,
      "learning_rate": 1.3453013684230732e-06,
      "loss": 3.2202,
      "step": 64210
    },
    {
      "epoch": 0.653228759765625,
      "grad_norm": 10.442748069763184,
      "learning_rate": 1.3449467117081832e-06,
      "loss": 3.2625,
      "step": 64215
    },
    {
      "epoch": 0.6532796223958334,
      "grad_norm": 12.160253524780273,
      "learning_rate": 1.3445920845450567e-06,
      "loss": 3.2401,
      "step": 64220
    },
    {
      "epoch": 0.6533304850260416,
      "grad_norm": 7.322057247161865,
      "learning_rate": 1.3442374869427689e-06,
      "loss": 3.2435,
      "step": 64225
    },
    {
      "epoch": 0.65338134765625,
      "grad_norm": 13.100785255432129,
      "learning_rate": 1.343882918910391e-06,
      "loss": 3.2074,
      "step": 64230
    },
    {
      "epoch": 0.6534322102864584,
      "grad_norm": 11.584278106689453,
      "learning_rate": 1.3435283804569937e-06,
      "loss": 3.5782,
      "step": 64235
    },
    {
      "epoch": 0.6534830729166666,
      "grad_norm": 15.957377433776855,
      "learning_rate": 1.3431738715916488e-06,
      "loss": 2.9493,
      "step": 64240
    },
    {
      "epoch": 0.653533935546875,
      "grad_norm": 9.237360954284668,
      "learning_rate": 1.3428193923234267e-06,
      "loss": 3.064,
      "step": 64245
    },
    {
      "epoch": 0.6535847981770834,
      "grad_norm": 11.011441230773926,
      "learning_rate": 1.3424649426613967e-06,
      "loss": 3.4114,
      "step": 64250
    },
    {
      "epoch": 0.6536356608072916,
      "grad_norm": 11.149901390075684,
      "learning_rate": 1.3421105226146265e-06,
      "loss": 3.2155,
      "step": 64255
    },
    {
      "epoch": 0.6536865234375,
      "grad_norm": 10.68295955657959,
      "learning_rate": 1.3417561321921822e-06,
      "loss": 3.8012,
      "step": 64260
    },
    {
      "epoch": 0.6537373860677084,
      "grad_norm": 8.677666664123535,
      "learning_rate": 1.3414017714031347e-06,
      "loss": 3.3083,
      "step": 64265
    },
    {
      "epoch": 0.6537882486979166,
      "grad_norm": 7.5353569984436035,
      "learning_rate": 1.3410474402565482e-06,
      "loss": 3.1559,
      "step": 64270
    },
    {
      "epoch": 0.653839111328125,
      "grad_norm": 12.037880897521973,
      "learning_rate": 1.340693138761488e-06,
      "loss": 4.1406,
      "step": 64275
    },
    {
      "epoch": 0.6538899739583334,
      "grad_norm": 10.325486183166504,
      "learning_rate": 1.3403388669270183e-06,
      "loss": 3.5111,
      "step": 64280
    },
    {
      "epoch": 0.6539408365885416,
      "grad_norm": 9.48999309539795,
      "learning_rate": 1.3399846247622039e-06,
      "loss": 3.2343,
      "step": 64285
    },
    {
      "epoch": 0.65399169921875,
      "grad_norm": 9.181351661682129,
      "learning_rate": 1.3396304122761089e-06,
      "loss": 3.7118,
      "step": 64290
    },
    {
      "epoch": 0.6540425618489584,
      "grad_norm": 11.652165412902832,
      "learning_rate": 1.3392762294777938e-06,
      "loss": 3.2166,
      "step": 64295
    },
    {
      "epoch": 0.6540934244791666,
      "grad_norm": 8.631184577941895,
      "learning_rate": 1.3389220763763222e-06,
      "loss": 3.0276,
      "step": 64300
    },
    {
      "epoch": 0.654144287109375,
      "grad_norm": 12.19041633605957,
      "learning_rate": 1.3385679529807544e-06,
      "loss": 3.0959,
      "step": 64305
    },
    {
      "epoch": 0.6541951497395834,
      "grad_norm": 18.410381317138672,
      "learning_rate": 1.3382138593001486e-06,
      "loss": 3.5766,
      "step": 64310
    },
    {
      "epoch": 0.6542460123697916,
      "grad_norm": 16.024188995361328,
      "learning_rate": 1.3378597953435662e-06,
      "loss": 3.4945,
      "step": 64315
    },
    {
      "epoch": 0.654296875,
      "grad_norm": 11.737725257873535,
      "learning_rate": 1.3375057611200664e-06,
      "loss": 3.3308,
      "step": 64320
    },
    {
      "epoch": 0.6543477376302084,
      "grad_norm": 11.525810241699219,
      "learning_rate": 1.3371517566387062e-06,
      "loss": 3.2458,
      "step": 64325
    },
    {
      "epoch": 0.6543986002604166,
      "grad_norm": 13.881775856018066,
      "learning_rate": 1.3367977819085425e-06,
      "loss": 3.534,
      "step": 64330
    },
    {
      "epoch": 0.654449462890625,
      "grad_norm": 13.49406623840332,
      "learning_rate": 1.3364438369386313e-06,
      "loss": 3.2864,
      "step": 64335
    },
    {
      "epoch": 0.6545003255208334,
      "grad_norm": 15.493911743164062,
      "learning_rate": 1.3360899217380285e-06,
      "loss": 3.4563,
      "step": 64340
    },
    {
      "epoch": 0.6545511881510416,
      "grad_norm": 9.129322052001953,
      "learning_rate": 1.33573603631579e-06,
      "loss": 3.5747,
      "step": 64345
    },
    {
      "epoch": 0.65460205078125,
      "grad_norm": 15.619573593139648,
      "learning_rate": 1.335382180680969e-06,
      "loss": 3.0429,
      "step": 64350
    },
    {
      "epoch": 0.6546529134114584,
      "grad_norm": 10.902886390686035,
      "learning_rate": 1.3350283548426183e-06,
      "loss": 3.3387,
      "step": 64355
    },
    {
      "epoch": 0.6547037760416666,
      "grad_norm": 10.480283737182617,
      "learning_rate": 1.3346745588097918e-06,
      "loss": 3.66,
      "step": 64360
    },
    {
      "epoch": 0.654754638671875,
      "grad_norm": 15.958941459655762,
      "learning_rate": 1.3343207925915393e-06,
      "loss": 3.7825,
      "step": 64365
    },
    {
      "epoch": 0.6548055013020834,
      "grad_norm": 10.528629302978516,
      "learning_rate": 1.3339670561969142e-06,
      "loss": 3.2601,
      "step": 64370
    },
    {
      "epoch": 0.6548563639322916,
      "grad_norm": 11.030975341796875,
      "learning_rate": 1.3336133496349646e-06,
      "loss": 3.3519,
      "step": 64375
    },
    {
      "epoch": 0.6549072265625,
      "grad_norm": 11.710150718688965,
      "learning_rate": 1.3332596729147418e-06,
      "loss": 3.0692,
      "step": 64380
    },
    {
      "epoch": 0.6549580891927084,
      "grad_norm": 7.4101881980896,
      "learning_rate": 1.3329060260452938e-06,
      "loss": 3.3557,
      "step": 64385
    },
    {
      "epoch": 0.6550089518229166,
      "grad_norm": 10.991748809814453,
      "learning_rate": 1.3325524090356667e-06,
      "loss": 3.5383,
      "step": 64390
    },
    {
      "epoch": 0.655059814453125,
      "grad_norm": 15.911041259765625,
      "learning_rate": 1.3321988218949115e-06,
      "loss": 3.0189,
      "step": 64395
    },
    {
      "epoch": 0.6551106770833334,
      "grad_norm": 9.134468078613281,
      "learning_rate": 1.3318452646320723e-06,
      "loss": 3.0937,
      "step": 64400
    },
    {
      "epoch": 0.6551615397135416,
      "grad_norm": 16.44624900817871,
      "learning_rate": 1.3314917372561954e-06,
      "loss": 3.6389,
      "step": 64405
    },
    {
      "epoch": 0.65521240234375,
      "grad_norm": 12.228934288024902,
      "learning_rate": 1.3311382397763245e-06,
      "loss": 3.4447,
      "step": 64410
    },
    {
      "epoch": 0.6552632649739584,
      "grad_norm": 8.263579368591309,
      "learning_rate": 1.3307847722015043e-06,
      "loss": 3.2181,
      "step": 64415
    },
    {
      "epoch": 0.6553141276041666,
      "grad_norm": 14.126418113708496,
      "learning_rate": 1.3304313345407799e-06,
      "loss": 3.3393,
      "step": 64420
    },
    {
      "epoch": 0.655364990234375,
      "grad_norm": 7.036675453186035,
      "learning_rate": 1.3300779268031928e-06,
      "loss": 3.5008,
      "step": 64425
    },
    {
      "epoch": 0.6554158528645834,
      "grad_norm": 12.67174243927002,
      "learning_rate": 1.329724548997783e-06,
      "loss": 3.9012,
      "step": 64430
    },
    {
      "epoch": 0.6554667154947916,
      "grad_norm": 13.009026527404785,
      "learning_rate": 1.3293712011335946e-06,
      "loss": 3.0976,
      "step": 64435
    },
    {
      "epoch": 0.655517578125,
      "grad_norm": 9.975815773010254,
      "learning_rate": 1.3290178832196654e-06,
      "loss": 3.5631,
      "step": 64440
    },
    {
      "epoch": 0.6555684407552084,
      "grad_norm": 16.536361694335938,
      "learning_rate": 1.3286645952650373e-06,
      "loss": 3.4593,
      "step": 64445
    },
    {
      "epoch": 0.6556193033854166,
      "grad_norm": 9.64079475402832,
      "learning_rate": 1.328311337278747e-06,
      "loss": 3.235,
      "step": 64450
    },
    {
      "epoch": 0.655670166015625,
      "grad_norm": 11.996866226196289,
      "learning_rate": 1.327958109269834e-06,
      "loss": 3.7884,
      "step": 64455
    },
    {
      "epoch": 0.6557210286458334,
      "grad_norm": 14.872767448425293,
      "learning_rate": 1.3276049112473355e-06,
      "loss": 3.3653,
      "step": 64460
    },
    {
      "epoch": 0.6557718912760416,
      "grad_norm": 10.64519214630127,
      "learning_rate": 1.327251743220286e-06,
      "loss": 2.9918,
      "step": 64465
    },
    {
      "epoch": 0.65582275390625,
      "grad_norm": 12.196529388427734,
      "learning_rate": 1.3268986051977228e-06,
      "loss": 3.1684,
      "step": 64470
    },
    {
      "epoch": 0.6558736165364584,
      "grad_norm": 11.566841125488281,
      "learning_rate": 1.3265454971886818e-06,
      "loss": 3.5357,
      "step": 64475
    },
    {
      "epoch": 0.6559244791666666,
      "grad_norm": 11.732039451599121,
      "learning_rate": 1.3261924192021961e-06,
      "loss": 3.3351,
      "step": 64480
    },
    {
      "epoch": 0.655975341796875,
      "grad_norm": 11.234846115112305,
      "learning_rate": 1.3258393712472981e-06,
      "loss": 3.1402,
      "step": 64485
    },
    {
      "epoch": 0.6560262044270834,
      "grad_norm": 13.978314399719238,
      "learning_rate": 1.3254863533330226e-06,
      "loss": 3.3941,
      "step": 64490
    },
    {
      "epoch": 0.6560770670572916,
      "grad_norm": 8.966712951660156,
      "learning_rate": 1.325133365468399e-06,
      "loss": 3.417,
      "step": 64495
    },
    {
      "epoch": 0.6561279296875,
      "grad_norm": 10.272655487060547,
      "learning_rate": 1.324780407662461e-06,
      "loss": 3.2492,
      "step": 64500
    },
    {
      "epoch": 0.6561787923177084,
      "grad_norm": 16.021495819091797,
      "learning_rate": 1.3244274799242368e-06,
      "loss": 3.504,
      "step": 64505
    },
    {
      "epoch": 0.6562296549479166,
      "grad_norm": 13.629439353942871,
      "learning_rate": 1.324074582262758e-06,
      "loss": 3.47,
      "step": 64510
    },
    {
      "epoch": 0.656280517578125,
      "grad_norm": 12.70556354522705,
      "learning_rate": 1.323721714687052e-06,
      "loss": 3.1075,
      "step": 64515
    },
    {
      "epoch": 0.6563313802083334,
      "grad_norm": 10.02698040008545,
      "learning_rate": 1.3233688772061462e-06,
      "loss": 3.6611,
      "step": 64520
    },
    {
      "epoch": 0.6563822428385416,
      "grad_norm": 12.080476760864258,
      "learning_rate": 1.3230160698290689e-06,
      "loss": 3.3706,
      "step": 64525
    },
    {
      "epoch": 0.65643310546875,
      "grad_norm": 10.975835800170898,
      "learning_rate": 1.3226632925648474e-06,
      "loss": 3.0203,
      "step": 64530
    },
    {
      "epoch": 0.6564839680989584,
      "grad_norm": 12.529753684997559,
      "learning_rate": 1.3223105454225067e-06,
      "loss": 3.1196,
      "step": 64535
    },
    {
      "epoch": 0.6565348307291666,
      "grad_norm": 13.119550704956055,
      "learning_rate": 1.3219578284110711e-06,
      "loss": 3.1326,
      "step": 64540
    },
    {
      "epoch": 0.656585693359375,
      "grad_norm": 12.159619331359863,
      "learning_rate": 1.3216051415395645e-06,
      "loss": 3.1074,
      "step": 64545
    },
    {
      "epoch": 0.6566365559895834,
      "grad_norm": 17.395584106445312,
      "learning_rate": 1.321252484817011e-06,
      "loss": 3.4433,
      "step": 64550
    },
    {
      "epoch": 0.6566874186197916,
      "grad_norm": 7.426807880401611,
      "learning_rate": 1.3208998582524343e-06,
      "loss": 3.3032,
      "step": 64555
    },
    {
      "epoch": 0.65673828125,
      "grad_norm": 10.322322845458984,
      "learning_rate": 1.3205472618548554e-06,
      "loss": 3.3263,
      "step": 64560
    },
    {
      "epoch": 0.6567891438802084,
      "grad_norm": 14.927852630615234,
      "learning_rate": 1.320194695633294e-06,
      "loss": 3.6157,
      "step": 64565
    },
    {
      "epoch": 0.6568400065104166,
      "grad_norm": 12.767248153686523,
      "learning_rate": 1.3198421595967718e-06,
      "loss": 3.1727,
      "step": 64570
    },
    {
      "epoch": 0.656890869140625,
      "grad_norm": 15.87508487701416,
      "learning_rate": 1.319489653754309e-06,
      "loss": 3.2067,
      "step": 64575
    },
    {
      "epoch": 0.6569417317708334,
      "grad_norm": 12.932256698608398,
      "learning_rate": 1.3191371781149237e-06,
      "loss": 3.6383,
      "step": 64580
    },
    {
      "epoch": 0.6569925944010416,
      "grad_norm": 10.85139274597168,
      "learning_rate": 1.318784732687633e-06,
      "loss": 3.5736,
      "step": 64585
    },
    {
      "epoch": 0.65704345703125,
      "grad_norm": 12.298839569091797,
      "learning_rate": 1.3184323174814558e-06,
      "loss": 3.2548,
      "step": 64590
    },
    {
      "epoch": 0.6570943196614584,
      "grad_norm": 17.16855812072754,
      "learning_rate": 1.3180799325054067e-06,
      "loss": 3.4335,
      "step": 64595
    },
    {
      "epoch": 0.6571451822916666,
      "grad_norm": 13.399876594543457,
      "learning_rate": 1.3177275777685025e-06,
      "loss": 3.471,
      "step": 64600
    },
    {
      "epoch": 0.657196044921875,
      "grad_norm": 10.207135200500488,
      "learning_rate": 1.317375253279759e-06,
      "loss": 3.0937,
      "step": 64605
    },
    {
      "epoch": 0.6572469075520834,
      "grad_norm": 12.139558792114258,
      "learning_rate": 1.3170229590481892e-06,
      "loss": 3.4177,
      "step": 64610
    },
    {
      "epoch": 0.6572977701822916,
      "grad_norm": 12.48451042175293,
      "learning_rate": 1.3166706950828067e-06,
      "loss": 3.1147,
      "step": 64615
    },
    {
      "epoch": 0.6573486328125,
      "grad_norm": 12.349983215332031,
      "learning_rate": 1.3163184613926228e-06,
      "loss": 3.0056,
      "step": 64620
    },
    {
      "epoch": 0.6573994954427084,
      "grad_norm": 12.045706748962402,
      "learning_rate": 1.3159662579866505e-06,
      "loss": 3.3908,
      "step": 64625
    },
    {
      "epoch": 0.6574503580729166,
      "grad_norm": 10.621950149536133,
      "learning_rate": 1.315614084873902e-06,
      "loss": 3.4618,
      "step": 64630
    },
    {
      "epoch": 0.657501220703125,
      "grad_norm": 10.255273818969727,
      "learning_rate": 1.3152619420633864e-06,
      "loss": 3.2101,
      "step": 64635
    },
    {
      "epoch": 0.6575520833333334,
      "grad_norm": 17.442726135253906,
      "learning_rate": 1.3149098295641122e-06,
      "loss": 3.5172,
      "step": 64640
    },
    {
      "epoch": 0.6576029459635416,
      "grad_norm": 12.190653800964355,
      "learning_rate": 1.3145577473850902e-06,
      "loss": 3.4842,
      "step": 64645
    },
    {
      "epoch": 0.65765380859375,
      "grad_norm": 20.169771194458008,
      "learning_rate": 1.3142056955353262e-06,
      "loss": 3.3943,
      "step": 64650
    },
    {
      "epoch": 0.6577046712239584,
      "grad_norm": 9.705902099609375,
      "learning_rate": 1.3138536740238289e-06,
      "loss": 3.431,
      "step": 64655
    },
    {
      "epoch": 0.6577555338541666,
      "grad_norm": 15.668383598327637,
      "learning_rate": 1.3135016828596037e-06,
      "loss": 3.6356,
      "step": 64660
    },
    {
      "epoch": 0.657806396484375,
      "grad_norm": 12.712372779846191,
      "learning_rate": 1.3131497220516575e-06,
      "loss": 3.3855,
      "step": 64665
    },
    {
      "epoch": 0.6578572591145834,
      "grad_norm": 273.25811767578125,
      "learning_rate": 1.3127977916089948e-06,
      "loss": 3.7067,
      "step": 64670
    },
    {
      "epoch": 0.6579081217447916,
      "grad_norm": 11.827073097229004,
      "learning_rate": 1.312445891540618e-06,
      "loss": 3.6846,
      "step": 64675
    },
    {
      "epoch": 0.657958984375,
      "grad_norm": 14.099276542663574,
      "learning_rate": 1.3120940218555314e-06,
      "loss": 3.2874,
      "step": 64680
    },
    {
      "epoch": 0.6580098470052084,
      "grad_norm": 12.601127624511719,
      "learning_rate": 1.3117421825627388e-06,
      "loss": 3.2402,
      "step": 64685
    },
    {
      "epoch": 0.6580607096354166,
      "grad_norm": 9.837937355041504,
      "learning_rate": 1.3113903736712408e-06,
      "loss": 3.4378,
      "step": 64690
    },
    {
      "epoch": 0.658111572265625,
      "grad_norm": 17.775718688964844,
      "learning_rate": 1.3110385951900373e-06,
      "loss": 3.5461,
      "step": 64695
    },
    {
      "epoch": 0.6581624348958334,
      "grad_norm": 15.552910804748535,
      "learning_rate": 1.3106868471281305e-06,
      "loss": 3.7383,
      "step": 64700
    },
    {
      "epoch": 0.6582132975260416,
      "grad_norm": 11.443589210510254,
      "learning_rate": 1.310335129494518e-06,
      "loss": 3.5422,
      "step": 64705
    },
    {
      "epoch": 0.65826416015625,
      "grad_norm": 12.87744140625,
      "learning_rate": 1.3099834422981999e-06,
      "loss": 3.3954,
      "step": 64710
    },
    {
      "epoch": 0.6583150227864584,
      "grad_norm": 10.184366226196289,
      "learning_rate": 1.3096317855481727e-06,
      "loss": 3.0972,
      "step": 64715
    },
    {
      "epoch": 0.6583658854166666,
      "grad_norm": 13.616209030151367,
      "learning_rate": 1.3092801592534348e-06,
      "loss": 3.3293,
      "step": 64720
    },
    {
      "epoch": 0.658416748046875,
      "grad_norm": 12.936397552490234,
      "learning_rate": 1.3089285634229814e-06,
      "loss": 3.0804,
      "step": 64725
    },
    {
      "epoch": 0.6584676106770834,
      "grad_norm": 12.256488800048828,
      "learning_rate": 1.308576998065809e-06,
      "loss": 3.1282,
      "step": 64730
    },
    {
      "epoch": 0.6585184733072916,
      "grad_norm": 12.041582107543945,
      "learning_rate": 1.3082254631909106e-06,
      "loss": 3.386,
      "step": 64735
    },
    {
      "epoch": 0.6585693359375,
      "grad_norm": 14.85916519165039,
      "learning_rate": 1.3078739588072823e-06,
      "loss": 3.5176,
      "step": 64740
    },
    {
      "epoch": 0.6586201985677084,
      "grad_norm": 10.276881217956543,
      "learning_rate": 1.3075224849239162e-06,
      "loss": 3.3579,
      "step": 64745
    },
    {
      "epoch": 0.6586710611979166,
      "grad_norm": 14.437859535217285,
      "learning_rate": 1.3071710415498038e-06,
      "loss": 3.2329,
      "step": 64750
    },
    {
      "epoch": 0.658721923828125,
      "grad_norm": 10.961657524108887,
      "learning_rate": 1.3068196286939374e-06,
      "loss": 3.4126,
      "step": 64755
    },
    {
      "epoch": 0.6587727864583334,
      "grad_norm": 8.900579452514648,
      "learning_rate": 1.3064682463653091e-06,
      "loss": 3.3339,
      "step": 64760
    },
    {
      "epoch": 0.6588236490885416,
      "grad_norm": 14.60119915008545,
      "learning_rate": 1.3061168945729078e-06,
      "loss": 3.8722,
      "step": 64765
    },
    {
      "epoch": 0.65887451171875,
      "grad_norm": 13.560766220092773,
      "learning_rate": 1.3057655733257227e-06,
      "loss": 3.4208,
      "step": 64770
    },
    {
      "epoch": 0.6589253743489584,
      "grad_norm": 13.868523597717285,
      "learning_rate": 1.3054142826327414e-06,
      "loss": 3.4038,
      "step": 64775
    },
    {
      "epoch": 0.6589762369791666,
      "grad_norm": 9.550907135009766,
      "learning_rate": 1.3050630225029526e-06,
      "loss": 3.323,
      "step": 64780
    },
    {
      "epoch": 0.659027099609375,
      "grad_norm": 13.122942924499512,
      "learning_rate": 1.304711792945344e-06,
      "loss": 3.3643,
      "step": 64785
    },
    {
      "epoch": 0.6590779622395834,
      "grad_norm": 9.805845260620117,
      "learning_rate": 1.3043605939689e-06,
      "loss": 3.303,
      "step": 64790
    },
    {
      "epoch": 0.6591288248697916,
      "grad_norm": 10.915084838867188,
      "learning_rate": 1.3040094255826078e-06,
      "loss": 3.6199,
      "step": 64795
    },
    {
      "epoch": 0.6591796875,
      "grad_norm": 14.074462890625,
      "learning_rate": 1.3036582877954506e-06,
      "loss": 3.2834,
      "step": 64800
    },
    {
      "epoch": 0.6592305501302084,
      "grad_norm": 8.09543228149414,
      "learning_rate": 1.3033071806164122e-06,
      "loss": 3.2665,
      "step": 64805
    },
    {
      "epoch": 0.6592814127604166,
      "grad_norm": 6.727164268493652,
      "learning_rate": 1.3029561040544755e-06,
      "loss": 2.809,
      "step": 64810
    },
    {
      "epoch": 0.659332275390625,
      "grad_norm": 11.766030311584473,
      "learning_rate": 1.3026050581186242e-06,
      "loss": 3.7183,
      "step": 64815
    },
    {
      "epoch": 0.6593831380208334,
      "grad_norm": 12.66413402557373,
      "learning_rate": 1.3022540428178386e-06,
      "loss": 3.4647,
      "step": 64820
    },
    {
      "epoch": 0.6594340006510416,
      "grad_norm": 12.70841121673584,
      "learning_rate": 1.3019030581610998e-06,
      "loss": 3.4055,
      "step": 64825
    },
    {
      "epoch": 0.65948486328125,
      "grad_norm": 11.378216743469238,
      "learning_rate": 1.3015521041573858e-06,
      "loss": 3.2415,
      "step": 64830
    },
    {
      "epoch": 0.6595357259114584,
      "grad_norm": 9.645951271057129,
      "learning_rate": 1.3012011808156775e-06,
      "loss": 3.1352,
      "step": 64835
    },
    {
      "epoch": 0.6595865885416666,
      "grad_norm": 8.79885482788086,
      "learning_rate": 1.3008502881449536e-06,
      "loss": 3.1783,
      "step": 64840
    },
    {
      "epoch": 0.659637451171875,
      "grad_norm": 12.367588996887207,
      "learning_rate": 1.3004994261541908e-06,
      "loss": 3.4423,
      "step": 64845
    },
    {
      "epoch": 0.6596883138020834,
      "grad_norm": 19.567867279052734,
      "learning_rate": 1.3001485948523652e-06,
      "loss": 2.8524,
      "step": 64850
    },
    {
      "epoch": 0.6597391764322916,
      "grad_norm": 12.848206520080566,
      "learning_rate": 1.2997977942484541e-06,
      "loss": 3.3016,
      "step": 64855
    },
    {
      "epoch": 0.6597900390625,
      "grad_norm": 14.344475746154785,
      "learning_rate": 1.2994470243514311e-06,
      "loss": 3.3602,
      "step": 64860
    },
    {
      "epoch": 0.6598409016927084,
      "grad_norm": 16.25145149230957,
      "learning_rate": 1.2990962851702724e-06,
      "loss": 3.3631,
      "step": 64865
    },
    {
      "epoch": 0.6598917643229166,
      "grad_norm": 9.507192611694336,
      "learning_rate": 1.2987455767139496e-06,
      "loss": 3.2929,
      "step": 64870
    },
    {
      "epoch": 0.659942626953125,
      "grad_norm": 9.96194076538086,
      "learning_rate": 1.2983948989914375e-06,
      "loss": 3.2014,
      "step": 64875
    },
    {
      "epoch": 0.6599934895833334,
      "grad_norm": 14.200433731079102,
      "learning_rate": 1.298044252011706e-06,
      "loss": 3.3339,
      "step": 64880
    },
    {
      "epoch": 0.6600443522135416,
      "grad_norm": 11.347167015075684,
      "learning_rate": 1.2976936357837278e-06,
      "loss": 3.0248,
      "step": 64885
    },
    {
      "epoch": 0.66009521484375,
      "grad_norm": 12.991911888122559,
      "learning_rate": 1.2973430503164735e-06,
      "loss": 3.1184,
      "step": 64890
    },
    {
      "epoch": 0.6601460774739584,
      "grad_norm": 9.217988967895508,
      "learning_rate": 1.2969924956189122e-06,
      "loss": 3.4694,
      "step": 64895
    },
    {
      "epoch": 0.6601969401041666,
      "grad_norm": 15.204910278320312,
      "learning_rate": 1.296641971700013e-06,
      "loss": 3.0725,
      "step": 64900
    },
    {
      "epoch": 0.660247802734375,
      "grad_norm": 7.2170538902282715,
      "learning_rate": 1.2962914785687426e-06,
      "loss": 3.5342,
      "step": 64905
    },
    {
      "epoch": 0.6602986653645834,
      "grad_norm": 12.763792037963867,
      "learning_rate": 1.2959410162340694e-06,
      "loss": 3.3776,
      "step": 64910
    },
    {
      "epoch": 0.6603495279947916,
      "grad_norm": 10.381855010986328,
      "learning_rate": 1.295590584704961e-06,
      "loss": 3.1652,
      "step": 64915
    },
    {
      "epoch": 0.660400390625,
      "grad_norm": 14.777809143066406,
      "learning_rate": 1.295240183990382e-06,
      "loss": 3.2149,
      "step": 64920
    },
    {
      "epoch": 0.6604512532552084,
      "grad_norm": 17.232894897460938,
      "learning_rate": 1.294889814099296e-06,
      "loss": 3.5293,
      "step": 64925
    },
    {
      "epoch": 0.6605021158854166,
      "grad_norm": 15.255964279174805,
      "learning_rate": 1.2945394750406699e-06,
      "loss": 4.0128,
      "step": 64930
    },
    {
      "epoch": 0.660552978515625,
      "grad_norm": 17.207504272460938,
      "learning_rate": 1.2941891668234646e-06,
      "loss": 3.5108,
      "step": 64935
    },
    {
      "epoch": 0.6606038411458334,
      "grad_norm": 18.810527801513672,
      "learning_rate": 1.2938388894566445e-06,
      "loss": 3.5506,
      "step": 64940
    },
    {
      "epoch": 0.6606547037760416,
      "grad_norm": 17.15147590637207,
      "learning_rate": 1.2934886429491695e-06,
      "loss": 3.7975,
      "step": 64945
    },
    {
      "epoch": 0.66070556640625,
      "grad_norm": 14.673933982849121,
      "learning_rate": 1.2931384273100025e-06,
      "loss": 3.3426,
      "step": 64950
    },
    {
      "epoch": 0.6607564290364584,
      "grad_norm": 9.279642105102539,
      "learning_rate": 1.292788242548103e-06,
      "loss": 3.5146,
      "step": 64955
    },
    {
      "epoch": 0.6608072916666666,
      "grad_norm": 14.439388275146484,
      "learning_rate": 1.2924380886724292e-06,
      "loss": 3.5971,
      "step": 64960
    },
    {
      "epoch": 0.660858154296875,
      "grad_norm": 16.13722801208496,
      "learning_rate": 1.2920879656919405e-06,
      "loss": 3.2019,
      "step": 64965
    },
    {
      "epoch": 0.6609090169270834,
      "grad_norm": 13.01333999633789,
      "learning_rate": 1.2917378736155961e-06,
      "loss": 3.0868,
      "step": 64970
    },
    {
      "epoch": 0.6609598795572916,
      "grad_norm": 11.99376392364502,
      "learning_rate": 1.2913878124523515e-06,
      "loss": 3.2505,
      "step": 64975
    },
    {
      "epoch": 0.6610107421875,
      "grad_norm": 11.174510955810547,
      "learning_rate": 1.2910377822111624e-06,
      "loss": 3.5321,
      "step": 64980
    },
    {
      "epoch": 0.6610616048177084,
      "grad_norm": 11.970725059509277,
      "learning_rate": 1.2906877829009861e-06,
      "loss": 3.9141,
      "step": 64985
    },
    {
      "epoch": 0.6611124674479166,
      "grad_norm": 8.983936309814453,
      "learning_rate": 1.2903378145307755e-06,
      "loss": 3.2809,
      "step": 64990
    },
    {
      "epoch": 0.661163330078125,
      "grad_norm": 15.522686004638672,
      "learning_rate": 1.2899878771094857e-06,
      "loss": 3.2497,
      "step": 64995
    },
    {
      "epoch": 0.6612141927083334,
      "grad_norm": 10.832334518432617,
      "learning_rate": 1.2896379706460684e-06,
      "loss": 3.2215,
      "step": 65000
    },
    {
      "epoch": 0.6612650553385416,
      "grad_norm": 13.5785493850708,
      "learning_rate": 1.289288095149478e-06,
      "loss": 3.0875,
      "step": 65005
    },
    {
      "epoch": 0.66131591796875,
      "grad_norm": 14.808880805969238,
      "learning_rate": 1.2889382506286646e-06,
      "loss": 3.4202,
      "step": 65010
    },
    {
      "epoch": 0.6613667805989584,
      "grad_norm": 40.9049072265625,
      "learning_rate": 1.2885884370925777e-06,
      "loss": 3.6887,
      "step": 65015
    },
    {
      "epoch": 0.6614176432291666,
      "grad_norm": 8.365484237670898,
      "learning_rate": 1.2882386545501691e-06,
      "loss": 3.122,
      "step": 65020
    },
    {
      "epoch": 0.661468505859375,
      "grad_norm": 8.404397010803223,
      "learning_rate": 1.2878889030103879e-06,
      "loss": 3.0866,
      "step": 65025
    },
    {
      "epoch": 0.6615193684895834,
      "grad_norm": 8.957504272460938,
      "learning_rate": 1.2875391824821818e-06,
      "loss": 3.0308,
      "step": 65030
    },
    {
      "epoch": 0.6615702311197916,
      "grad_norm": 7.428070068359375,
      "learning_rate": 1.2871894929744971e-06,
      "loss": 3.4467,
      "step": 65035
    },
    {
      "epoch": 0.66162109375,
      "grad_norm": 10.792874336242676,
      "learning_rate": 1.2868398344962823e-06,
      "loss": 3.0817,
      "step": 65040
    },
    {
      "epoch": 0.6616719563802084,
      "grad_norm": 11.444332122802734,
      "learning_rate": 1.2864902070564834e-06,
      "loss": 3.0255,
      "step": 65045
    },
    {
      "epoch": 0.6617228190104166,
      "grad_norm": 8.817506790161133,
      "learning_rate": 1.286140610664045e-06,
      "loss": 3.2208,
      "step": 65050
    },
    {
      "epoch": 0.661773681640625,
      "grad_norm": 9.036532402038574,
      "learning_rate": 1.2857910453279115e-06,
      "loss": 3.2843,
      "step": 65055
    },
    {
      "epoch": 0.6618245442708334,
      "grad_norm": 14.336651802062988,
      "learning_rate": 1.2854415110570254e-06,
      "loss": 3.3392,
      "step": 65060
    },
    {
      "epoch": 0.6618754069010416,
      "grad_norm": 11.58194351196289,
      "learning_rate": 1.2850920078603302e-06,
      "loss": 3.5359,
      "step": 65065
    },
    {
      "epoch": 0.66192626953125,
      "grad_norm": 12.473958015441895,
      "learning_rate": 1.2847425357467688e-06,
      "loss": 3.5369,
      "step": 65070
    },
    {
      "epoch": 0.6619771321614584,
      "grad_norm": 18.386960983276367,
      "learning_rate": 1.284393094725282e-06,
      "loss": 3.4017,
      "step": 65075
    },
    {
      "epoch": 0.6620279947916666,
      "grad_norm": 11.408589363098145,
      "learning_rate": 1.2840436848048083e-06,
      "loss": 3.0546,
      "step": 65080
    },
    {
      "epoch": 0.662078857421875,
      "grad_norm": 15.049735069274902,
      "learning_rate": 1.28369430599429e-06,
      "loss": 3.1304,
      "step": 65085
    },
    {
      "epoch": 0.6621297200520834,
      "grad_norm": 11.646748542785645,
      "learning_rate": 1.2833449583026636e-06,
      "loss": 3.7416,
      "step": 65090
    },
    {
      "epoch": 0.6621805826822916,
      "grad_norm": 11.964768409729004,
      "learning_rate": 1.282995641738868e-06,
      "loss": 3.428,
      "step": 65095
    },
    {
      "epoch": 0.6622314453125,
      "grad_norm": 13.79321002960205,
      "learning_rate": 1.2826463563118414e-06,
      "loss": 3.1163,
      "step": 65100
    },
    {
      "epoch": 0.6622823079427084,
      "grad_norm": 12.12221622467041,
      "learning_rate": 1.282297102030519e-06,
      "loss": 3.9972,
      "step": 65105
    },
    {
      "epoch": 0.6623331705729166,
      "grad_norm": 7.613211154937744,
      "learning_rate": 1.281947878903837e-06,
      "loss": 3.2019,
      "step": 65110
    },
    {
      "epoch": 0.662384033203125,
      "grad_norm": 10.106460571289062,
      "learning_rate": 1.2815986869407288e-06,
      "loss": 3.0862,
      "step": 65115
    },
    {
      "epoch": 0.6624348958333334,
      "grad_norm": 7.456476211547852,
      "learning_rate": 1.281249526150129e-06,
      "loss": 3.1201,
      "step": 65120
    },
    {
      "epoch": 0.6624857584635416,
      "grad_norm": 13.168668746948242,
      "learning_rate": 1.2809003965409727e-06,
      "loss": 3.8123,
      "step": 65125
    },
    {
      "epoch": 0.66253662109375,
      "grad_norm": 15.872102737426758,
      "learning_rate": 1.2805512981221904e-06,
      "loss": 3.1302,
      "step": 65130
    },
    {
      "epoch": 0.6625874837239584,
      "grad_norm": 8.025379180908203,
      "learning_rate": 1.2802022309027134e-06,
      "loss": 2.9626,
      "step": 65135
    },
    {
      "epoch": 0.6626383463541666,
      "grad_norm": 10.427035331726074,
      "learning_rate": 1.2798531948914739e-06,
      "loss": 3.0251,
      "step": 65140
    },
    {
      "epoch": 0.662689208984375,
      "grad_norm": 13.071147918701172,
      "learning_rate": 1.2795041900974003e-06,
      "loss": 3.1159,
      "step": 65145
    },
    {
      "epoch": 0.6627400716145834,
      "grad_norm": 17.302522659301758,
      "learning_rate": 1.2791552165294241e-06,
      "loss": 3.2413,
      "step": 65150
    },
    {
      "epoch": 0.6627909342447916,
      "grad_norm": 13.62521743774414,
      "learning_rate": 1.2788062741964708e-06,
      "loss": 3.1752,
      "step": 65155
    },
    {
      "epoch": 0.662841796875,
      "grad_norm": 14.582704544067383,
      "learning_rate": 1.2784573631074708e-06,
      "loss": 3.2239,
      "step": 65160
    },
    {
      "epoch": 0.6628926595052084,
      "grad_norm": 9.311365127563477,
      "learning_rate": 1.2781084832713494e-06,
      "loss": 3.2253,
      "step": 65165
    },
    {
      "epoch": 0.6629435221354166,
      "grad_norm": 9.742881774902344,
      "learning_rate": 1.2777596346970323e-06,
      "loss": 3.5226,
      "step": 65170
    },
    {
      "epoch": 0.662994384765625,
      "grad_norm": 9.145849227905273,
      "learning_rate": 1.2774108173934447e-06,
      "loss": 3.7364,
      "step": 65175
    },
    {
      "epoch": 0.6630452473958334,
      "grad_norm": 10.365301132202148,
      "learning_rate": 1.2770620313695127e-06,
      "loss": 3.0558,
      "step": 65180
    },
    {
      "epoch": 0.6630961100260416,
      "grad_norm": 13.321115493774414,
      "learning_rate": 1.2767132766341588e-06,
      "loss": 3.2569,
      "step": 65185
    },
    {
      "epoch": 0.66314697265625,
      "grad_norm": 12.305617332458496,
      "learning_rate": 1.2763645531963048e-06,
      "loss": 3.1887,
      "step": 65190
    },
    {
      "epoch": 0.6631978352864584,
      "grad_norm": 15.394637107849121,
      "learning_rate": 1.2760158610648738e-06,
      "loss": 2.7999,
      "step": 65195
    },
    {
      "epoch": 0.6632486979166666,
      "grad_norm": 13.64035701751709,
      "learning_rate": 1.2756672002487877e-06,
      "loss": 3.4076,
      "step": 65200
    },
    {
      "epoch": 0.663299560546875,
      "grad_norm": 7.51201057434082,
      "learning_rate": 1.275318570756966e-06,
      "loss": 3.6468,
      "step": 65205
    },
    {
      "epoch": 0.6633504231770834,
      "grad_norm": 10.302927017211914,
      "learning_rate": 1.2749699725983278e-06,
      "loss": 3.2673,
      "step": 65210
    },
    {
      "epoch": 0.6634012858072916,
      "grad_norm": 8.08244514465332,
      "learning_rate": 1.2746214057817929e-06,
      "loss": 3.6082,
      "step": 65215
    },
    {
      "epoch": 0.6634521484375,
      "grad_norm": 12.379706382751465,
      "learning_rate": 1.2742728703162782e-06,
      "loss": 3.1576,
      "step": 65220
    },
    {
      "epoch": 0.6635030110677084,
      "grad_norm": 9.567678451538086,
      "learning_rate": 1.2739243662107025e-06,
      "loss": 3.1563,
      "step": 65225
    },
    {
      "epoch": 0.6635538736979166,
      "grad_norm": 10.015485763549805,
      "learning_rate": 1.2735758934739807e-06,
      "loss": 3.268,
      "step": 65230
    },
    {
      "epoch": 0.663604736328125,
      "grad_norm": 6.774351119995117,
      "learning_rate": 1.2732274521150296e-06,
      "loss": 3.3882,
      "step": 65235
    },
    {
      "epoch": 0.6636555989583334,
      "grad_norm": 7.824482440948486,
      "learning_rate": 1.2728790421427633e-06,
      "loss": 3.1738,
      "step": 65240
    },
    {
      "epoch": 0.6637064615885416,
      "grad_norm": 13.696547508239746,
      "learning_rate": 1.272530663566095e-06,
      "loss": 3.2887,
      "step": 65245
    },
    {
      "epoch": 0.66375732421875,
      "grad_norm": 11.98934268951416,
      "learning_rate": 1.2721823163939384e-06,
      "loss": 3.3336,
      "step": 65250
    },
    {
      "epoch": 0.6638081868489584,
      "grad_norm": 9.976804733276367,
      "learning_rate": 1.2718340006352075e-06,
      "loss": 3.6427,
      "step": 65255
    },
    {
      "epoch": 0.6638590494791666,
      "grad_norm": 11.145037651062012,
      "learning_rate": 1.2714857162988126e-06,
      "loss": 3.7012,
      "step": 65260
    },
    {
      "epoch": 0.663909912109375,
      "grad_norm": 11.77367877960205,
      "learning_rate": 1.2711374633936643e-06,
      "loss": 3.2743,
      "step": 65265
    },
    {
      "epoch": 0.6639607747395834,
      "grad_norm": 10.72303581237793,
      "learning_rate": 1.2707892419286721e-06,
      "loss": 3.3152,
      "step": 65270
    },
    {
      "epoch": 0.6640116373697916,
      "grad_norm": 13.609925270080566,
      "learning_rate": 1.2704410519127452e-06,
      "loss": 3.6111,
      "step": 65275
    },
    {
      "epoch": 0.6640625,
      "grad_norm": 11.802302360534668,
      "learning_rate": 1.270092893354794e-06,
      "loss": 4.0092,
      "step": 65280
    },
    {
      "epoch": 0.6641133626302084,
      "grad_norm": 9.450394630432129,
      "learning_rate": 1.2697447662637236e-06,
      "loss": 3.3385,
      "step": 65285
    },
    {
      "epoch": 0.6641642252604166,
      "grad_norm": 13.917665481567383,
      "learning_rate": 1.2693966706484425e-06,
      "loss": 3.2738,
      "step": 65290
    },
    {
      "epoch": 0.664215087890625,
      "grad_norm": 12.290202140808105,
      "learning_rate": 1.2690486065178562e-06,
      "loss": 3.377,
      "step": 65295
    },
    {
      "epoch": 0.6642659505208334,
      "grad_norm": 15.638297080993652,
      "learning_rate": 1.2687005738808684e-06,
      "loss": 3.3482,
      "step": 65300
    },
    {
      "epoch": 0.6643168131510416,
      "grad_norm": 18.216962814331055,
      "learning_rate": 1.2683525727463844e-06,
      "loss": 3.9138,
      "step": 65305
    },
    {
      "epoch": 0.66436767578125,
      "grad_norm": 9.221813201904297,
      "learning_rate": 1.268004603123309e-06,
      "loss": 3.3517,
      "step": 65310
    },
    {
      "epoch": 0.6644185384114584,
      "grad_norm": 13.317717552185059,
      "learning_rate": 1.2676566650205441e-06,
      "loss": 3.3212,
      "step": 65315
    },
    {
      "epoch": 0.6644694010416666,
      "grad_norm": 10.615367889404297,
      "learning_rate": 1.2673087584469909e-06,
      "loss": 3.8712,
      "step": 65320
    },
    {
      "epoch": 0.664520263671875,
      "grad_norm": 7.501321792602539,
      "learning_rate": 1.2669608834115492e-06,
      "loss": 3.2691,
      "step": 65325
    },
    {
      "epoch": 0.6645711263020834,
      "grad_norm": 16.984865188598633,
      "learning_rate": 1.266613039923123e-06,
      "loss": 3.2706,
      "step": 65330
    },
    {
      "epoch": 0.6646219889322916,
      "grad_norm": 9.495741844177246,
      "learning_rate": 1.2662652279906095e-06,
      "loss": 3.3808,
      "step": 65335
    },
    {
      "epoch": 0.6646728515625,
      "grad_norm": 12.779135704040527,
      "learning_rate": 1.265917447622908e-06,
      "loss": 3.5763,
      "step": 65340
    },
    {
      "epoch": 0.6647237141927084,
      "grad_norm": 12.259757995605469,
      "learning_rate": 1.2655696988289151e-06,
      "loss": 3.3813,
      "step": 65345
    },
    {
      "epoch": 0.6647745768229166,
      "grad_norm": 7.823262691497803,
      "learning_rate": 1.2652219816175287e-06,
      "loss": 3.591,
      "step": 65350
    },
    {
      "epoch": 0.664825439453125,
      "grad_norm": 12.372004508972168,
      "learning_rate": 1.2648742959976468e-06,
      "loss": 3.7868,
      "step": 65355
    },
    {
      "epoch": 0.6648763020833334,
      "grad_norm": 17.824193954467773,
      "learning_rate": 1.2645266419781626e-06,
      "loss": 3.3528,
      "step": 65360
    },
    {
      "epoch": 0.6649271647135416,
      "grad_norm": 13.596588134765625,
      "learning_rate": 1.2641790195679707e-06,
      "loss": 3.5585,
      "step": 65365
    },
    {
      "epoch": 0.66497802734375,
      "grad_norm": 10.611842155456543,
      "learning_rate": 1.2638314287759668e-06,
      "loss": 3.0753,
      "step": 65370
    },
    {
      "epoch": 0.6650288899739584,
      "grad_norm": 9.485777854919434,
      "learning_rate": 1.2634838696110414e-06,
      "loss": 3.1376,
      "step": 65375
    },
    {
      "epoch": 0.6650797526041666,
      "grad_norm": 10.109831809997559,
      "learning_rate": 1.2631363420820886e-06,
      "loss": 3.8677,
      "step": 65380
    },
    {
      "epoch": 0.665130615234375,
      "grad_norm": 14.614781379699707,
      "learning_rate": 1.2627888461979999e-06,
      "loss": 4.1296,
      "step": 65385
    },
    {
      "epoch": 0.6651814778645834,
      "grad_norm": 11.567127227783203,
      "learning_rate": 1.2624413819676657e-06,
      "loss": 3.334,
      "step": 65390
    },
    {
      "epoch": 0.6652323404947916,
      "grad_norm": 8.257408142089844,
      "learning_rate": 1.262093949399975e-06,
      "loss": 3.4889,
      "step": 65395
    },
    {
      "epoch": 0.665283203125,
      "grad_norm": 8.923105239868164,
      "learning_rate": 1.2617465485038163e-06,
      "loss": 4.0292,
      "step": 65400
    },
    {
      "epoch": 0.6653340657552084,
      "grad_norm": 8.343690872192383,
      "learning_rate": 1.2613991792880787e-06,
      "loss": 3.1172,
      "step": 65405
    },
    {
      "epoch": 0.6653849283854166,
      "grad_norm": 12.005006790161133,
      "learning_rate": 1.2610518417616507e-06,
      "loss": 3.5156,
      "step": 65410
    },
    {
      "epoch": 0.665435791015625,
      "grad_norm": 11.469767570495605,
      "learning_rate": 1.2607045359334175e-06,
      "loss": 3.4081,
      "step": 65415
    },
    {
      "epoch": 0.6654866536458334,
      "grad_norm": 10.728708267211914,
      "learning_rate": 1.2603572618122638e-06,
      "loss": 3.1886,
      "step": 65420
    },
    {
      "epoch": 0.6655375162760416,
      "grad_norm": 15.159509658813477,
      "learning_rate": 1.2600100194070769e-06,
      "loss": 3.2632,
      "step": 65425
    },
    {
      "epoch": 0.66558837890625,
      "grad_norm": 12.759864807128906,
      "learning_rate": 1.259662808726739e-06,
      "loss": 3.1454,
      "step": 65430
    },
    {
      "epoch": 0.6656392415364584,
      "grad_norm": 13.23811149597168,
      "learning_rate": 1.259315629780135e-06,
      "loss": 3.3728,
      "step": 65435
    },
    {
      "epoch": 0.6656901041666666,
      "grad_norm": 14.851045608520508,
      "learning_rate": 1.2589684825761456e-06,
      "loss": 3.0316,
      "step": 65440
    },
    {
      "epoch": 0.665740966796875,
      "grad_norm": 14.239526748657227,
      "learning_rate": 1.2586213671236541e-06,
      "loss": 3.3417,
      "step": 65445
    },
    {
      "epoch": 0.6657918294270834,
      "grad_norm": 12.545265197753906,
      "learning_rate": 1.258274283431541e-06,
      "loss": 3.4793,
      "step": 65450
    },
    {
      "epoch": 0.6658426920572916,
      "grad_norm": 16.657880783081055,
      "learning_rate": 1.2579272315086848e-06,
      "loss": 3.3026,
      "step": 65455
    },
    {
      "epoch": 0.6658935546875,
      "grad_norm": 12.7116117477417,
      "learning_rate": 1.2575802113639662e-06,
      "loss": 3.5199,
      "step": 65460
    },
    {
      "epoch": 0.6659444173177084,
      "grad_norm": 12.080648422241211,
      "learning_rate": 1.257233223006264e-06,
      "loss": 3.0086,
      "step": 65465
    },
    {
      "epoch": 0.6659952799479166,
      "grad_norm": 14.472729682922363,
      "learning_rate": 1.2568862664444557e-06,
      "loss": 3.0486,
      "step": 65470
    },
    {
      "epoch": 0.666046142578125,
      "grad_norm": 8.622594833374023,
      "learning_rate": 1.2565393416874164e-06,
      "loss": 3.0956,
      "step": 65475
    },
    {
      "epoch": 0.6660970052083334,
      "grad_norm": 10.788368225097656,
      "learning_rate": 1.2561924487440232e-06,
      "loss": 3.295,
      "step": 65480
    },
    {
      "epoch": 0.6661478678385416,
      "grad_norm": 15.795666694641113,
      "learning_rate": 1.2558455876231526e-06,
      "loss": 3.013,
      "step": 65485
    },
    {
      "epoch": 0.66619873046875,
      "grad_norm": 11.166401863098145,
      "learning_rate": 1.2554987583336779e-06,
      "loss": 3.1137,
      "step": 65490
    },
    {
      "epoch": 0.6662495930989584,
      "grad_norm": 9.57898998260498,
      "learning_rate": 1.2551519608844712e-06,
      "loss": 3.2857,
      "step": 65495
    },
    {
      "epoch": 0.6663004557291666,
      "grad_norm": 8.748499870300293,
      "learning_rate": 1.2548051952844075e-06,
      "loss": 3.0462,
      "step": 65500
    },
    {
      "epoch": 0.666351318359375,
      "grad_norm": 14.830123901367188,
      "learning_rate": 1.2544584615423569e-06,
      "loss": 3.1002,
      "step": 65505
    },
    {
      "epoch": 0.6664021809895834,
      "grad_norm": 14.823974609375,
      "learning_rate": 1.254111759667192e-06,
      "loss": 3.3103,
      "step": 65510
    },
    {
      "epoch": 0.6664530436197916,
      "grad_norm": 15.658598899841309,
      "learning_rate": 1.253765089667782e-06,
      "loss": 3.2588,
      "step": 65515
    },
    {
      "epoch": 0.66650390625,
      "grad_norm": 14.136911392211914,
      "learning_rate": 1.2534184515529974e-06,
      "loss": 3.39,
      "step": 65520
    },
    {
      "epoch": 0.6665547688802084,
      "grad_norm": 11.759278297424316,
      "learning_rate": 1.253071845331706e-06,
      "loss": 3.3365,
      "step": 65525
    },
    {
      "epoch": 0.6666056315104166,
      "grad_norm": 13.362621307373047,
      "learning_rate": 1.2527252710127756e-06,
      "loss": 3.4476,
      "step": 65530
    },
    {
      "epoch": 0.666656494140625,
      "grad_norm": 10.812849044799805,
      "learning_rate": 1.252378728605073e-06,
      "loss": 3.5454,
      "step": 65535
    },
    {
      "epoch": 0.6667073567708334,
      "grad_norm": 18.48908233642578,
      "learning_rate": 1.252032218117466e-06,
      "loss": 3.3882,
      "step": 65540
    },
    {
      "epoch": 0.6667582194010416,
      "grad_norm": 15.132722854614258,
      "learning_rate": 1.2516857395588188e-06,
      "loss": 3.2699,
      "step": 65545
    },
    {
      "epoch": 0.66680908203125,
      "grad_norm": 9.398686408996582,
      "learning_rate": 1.2513392929379965e-06,
      "loss": 3.2346,
      "step": 65550
    },
    {
      "epoch": 0.6668599446614584,
      "grad_norm": 9.966465950012207,
      "learning_rate": 1.2509928782638609e-06,
      "loss": 3.632,
      "step": 65555
    },
    {
      "epoch": 0.6669108072916666,
      "grad_norm": 19.779407501220703,
      "learning_rate": 1.2506464955452768e-06,
      "loss": 2.7112,
      "step": 65560
    },
    {
      "epoch": 0.666961669921875,
      "grad_norm": 10.571062088012695,
      "learning_rate": 1.2503001447911068e-06,
      "loss": 3.2902,
      "step": 65565
    },
    {
      "epoch": 0.6670125325520834,
      "grad_norm": 11.481815338134766,
      "learning_rate": 1.2499538260102105e-06,
      "loss": 4.1773,
      "step": 65570
    },
    {
      "epoch": 0.6670633951822916,
      "grad_norm": 10.567340850830078,
      "learning_rate": 1.2496075392114501e-06,
      "loss": 3.2373,
      "step": 65575
    },
    {
      "epoch": 0.6671142578125,
      "grad_norm": 14.8667631149292,
      "learning_rate": 1.2492612844036845e-06,
      "loss": 3.4899,
      "step": 65580
    },
    {
      "epoch": 0.6671651204427084,
      "grad_norm": 17.505634307861328,
      "learning_rate": 1.2489150615957715e-06,
      "loss": 3.4972,
      "step": 65585
    },
    {
      "epoch": 0.6672159830729166,
      "grad_norm": 10.369640350341797,
      "learning_rate": 1.24856887079657e-06,
      "loss": 3.08,
      "step": 65590
    },
    {
      "epoch": 0.667266845703125,
      "grad_norm": 9.872705459594727,
      "learning_rate": 1.2482227120149387e-06,
      "loss": 2.9665,
      "step": 65595
    },
    {
      "epoch": 0.6673177083333334,
      "grad_norm": 14.500247955322266,
      "learning_rate": 1.247876585259732e-06,
      "loss": 3.1538,
      "step": 65600
    },
    {
      "epoch": 0.6673685709635416,
      "grad_norm": 16.241443634033203,
      "learning_rate": 1.2475304905398067e-06,
      "loss": 3.0915,
      "step": 65605
    },
    {
      "epoch": 0.66741943359375,
      "grad_norm": 9.634785652160645,
      "learning_rate": 1.2471844278640156e-06,
      "loss": 3.2526,
      "step": 65610
    },
    {
      "epoch": 0.6674702962239584,
      "grad_norm": 15.240821838378906,
      "learning_rate": 1.246838397241214e-06,
      "loss": 3.0817,
      "step": 65615
    },
    {
      "epoch": 0.6675211588541666,
      "grad_norm": 8.462591171264648,
      "learning_rate": 1.246492398680256e-06,
      "loss": 3.1551,
      "step": 65620
    },
    {
      "epoch": 0.667572021484375,
      "grad_norm": 15.741432189941406,
      "learning_rate": 1.246146432189993e-06,
      "loss": 3.1494,
      "step": 65625
    },
    {
      "epoch": 0.6676228841145834,
      "grad_norm": 13.4065523147583,
      "learning_rate": 1.2458004977792752e-06,
      "loss": 3.5921,
      "step": 65630
    },
    {
      "epoch": 0.6676737467447916,
      "grad_norm": 12.74771785736084,
      "learning_rate": 1.2454545954569545e-06,
      "loss": 3.255,
      "step": 65635
    },
    {
      "epoch": 0.667724609375,
      "grad_norm": 12.701573371887207,
      "learning_rate": 1.2451087252318813e-06,
      "loss": 3.1132,
      "step": 65640
    },
    {
      "epoch": 0.6677754720052084,
      "grad_norm": 17.424406051635742,
      "learning_rate": 1.244762887112904e-06,
      "loss": 3.5127,
      "step": 65645
    },
    {
      "epoch": 0.6678263346354166,
      "grad_norm": 19.323753356933594,
      "learning_rate": 1.2444170811088701e-06,
      "loss": 2.8202,
      "step": 65650
    },
    {
      "epoch": 0.667877197265625,
      "grad_norm": 12.669976234436035,
      "learning_rate": 1.2440713072286285e-06,
      "loss": 3.2664,
      "step": 65655
    },
    {
      "epoch": 0.6679280598958334,
      "grad_norm": 11.021958351135254,
      "learning_rate": 1.2437255654810234e-06,
      "loss": 3.447,
      "step": 65660
    },
    {
      "epoch": 0.6679789225260416,
      "grad_norm": 15.243814468383789,
      "learning_rate": 1.2433798558749032e-06,
      "loss": 3.3269,
      "step": 65665
    },
    {
      "epoch": 0.66802978515625,
      "grad_norm": 12.763647079467773,
      "learning_rate": 1.2430341784191105e-06,
      "loss": 3.2928,
      "step": 65670
    },
    {
      "epoch": 0.6680806477864584,
      "grad_norm": 9.308406829833984,
      "learning_rate": 1.2426885331224913e-06,
      "loss": 3.4037,
      "step": 65675
    },
    {
      "epoch": 0.6681315104166666,
      "grad_norm": 14.168586730957031,
      "learning_rate": 1.2423429199938882e-06,
      "loss": 3.1222,
      "step": 65680
    },
    {
      "epoch": 0.668182373046875,
      "grad_norm": 12.823803901672363,
      "learning_rate": 1.2419973390421421e-06,
      "loss": 3.471,
      "step": 65685
    },
    {
      "epoch": 0.6682332356770834,
      "grad_norm": 9.29395580291748,
      "learning_rate": 1.2416517902760963e-06,
      "loss": 3.3921,
      "step": 65690
    },
    {
      "epoch": 0.6682840983072916,
      "grad_norm": 9.270599365234375,
      "learning_rate": 1.241306273704592e-06,
      "loss": 3.2971,
      "step": 65695
    },
    {
      "epoch": 0.6683349609375,
      "grad_norm": 10.670681953430176,
      "learning_rate": 1.2409607893364684e-06,
      "loss": 3.2714,
      "step": 65700
    },
    {
      "epoch": 0.6683858235677084,
      "grad_norm": 10.184364318847656,
      "learning_rate": 1.2406153371805636e-06,
      "loss": 3.5247,
      "step": 65705
    },
    {
      "epoch": 0.6684366861979166,
      "grad_norm": 15.072261810302734,
      "learning_rate": 1.2402699172457175e-06,
      "loss": 3.3025,
      "step": 65710
    },
    {
      "epoch": 0.668487548828125,
      "grad_norm": 14.222228050231934,
      "learning_rate": 1.2399245295407663e-06,
      "loss": 3.4413,
      "step": 65715
    },
    {
      "epoch": 0.6685384114583334,
      "grad_norm": 11.479125022888184,
      "learning_rate": 1.2395791740745482e-06,
      "loss": 3.1769,
      "step": 65720
    },
    {
      "epoch": 0.6685892740885416,
      "grad_norm": 11.094748497009277,
      "learning_rate": 1.2392338508558975e-06,
      "loss": 3.2729,
      "step": 65725
    },
    {
      "epoch": 0.66864013671875,
      "grad_norm": 13.094012260437012,
      "learning_rate": 1.2388885598936507e-06,
      "loss": 3.2583,
      "step": 65730
    },
    {
      "epoch": 0.6686909993489584,
      "grad_norm": 9.674267768859863,
      "learning_rate": 1.2385433011966412e-06,
      "loss": 3.5974,
      "step": 65735
    },
    {
      "epoch": 0.6687418619791666,
      "grad_norm": 11.32293701171875,
      "learning_rate": 1.238198074773701e-06,
      "loss": 3.2655,
      "step": 65740
    },
    {
      "epoch": 0.668792724609375,
      "grad_norm": 14.59813117980957,
      "learning_rate": 1.2378528806336643e-06,
      "loss": 3.2294,
      "step": 65745
    },
    {
      "epoch": 0.6688435872395834,
      "grad_norm": 13.594170570373535,
      "learning_rate": 1.2375077187853635e-06,
      "loss": 3.0215,
      "step": 65750
    },
    {
      "epoch": 0.6688944498697916,
      "grad_norm": 17.96681785583496,
      "learning_rate": 1.2371625892376284e-06,
      "loss": 3.6447,
      "step": 65755
    },
    {
      "epoch": 0.6689453125,
      "grad_norm": 14.56413459777832,
      "learning_rate": 1.2368174919992892e-06,
      "loss": 3.4442,
      "step": 65760
    },
    {
      "epoch": 0.6689961751302084,
      "grad_norm": 16.94887351989746,
      "learning_rate": 1.2364724270791738e-06,
      "loss": 3.2471,
      "step": 65765
    },
    {
      "epoch": 0.6690470377604166,
      "grad_norm": 7.452788829803467,
      "learning_rate": 1.2361273944861124e-06,
      "loss": 3.2806,
      "step": 65770
    },
    {
      "epoch": 0.669097900390625,
      "grad_norm": 13.136384010314941,
      "learning_rate": 1.2357823942289326e-06,
      "loss": 3.2754,
      "step": 65775
    },
    {
      "epoch": 0.6691487630208334,
      "grad_norm": 10.442337036132812,
      "learning_rate": 1.2354374263164599e-06,
      "loss": 3.3363,
      "step": 65780
    },
    {
      "epoch": 0.6691996256510416,
      "grad_norm": 10.758193016052246,
      "learning_rate": 1.2350924907575216e-06,
      "loss": 3.4805,
      "step": 65785
    },
    {
      "epoch": 0.66925048828125,
      "grad_norm": 7.44234037399292,
      "learning_rate": 1.2347475875609414e-06,
      "loss": 3.1424,
      "step": 65790
    },
    {
      "epoch": 0.6693013509114584,
      "grad_norm": 14.84299087524414,
      "learning_rate": 1.2344027167355449e-06,
      "loss": 3.5229,
      "step": 65795
    },
    {
      "epoch": 0.6693522135416666,
      "grad_norm": 14.938111305236816,
      "learning_rate": 1.2340578782901542e-06,
      "loss": 3.1912,
      "step": 65800
    },
    {
      "epoch": 0.669403076171875,
      "grad_norm": 8.341020584106445,
      "learning_rate": 1.2337130722335936e-06,
      "loss": 3.4417,
      "step": 65805
    },
    {
      "epoch": 0.6694539388020834,
      "grad_norm": 11.463062286376953,
      "learning_rate": 1.2333682985746836e-06,
      "loss": 3.202,
      "step": 65810
    },
    {
      "epoch": 0.6695048014322916,
      "grad_norm": 15.396414756774902,
      "learning_rate": 1.233023557322245e-06,
      "loss": 3.2796,
      "step": 65815
    },
    {
      "epoch": 0.6695556640625,
      "grad_norm": 8.394917488098145,
      "learning_rate": 1.2326788484850979e-06,
      "loss": 3.2157,
      "step": 65820
    },
    {
      "epoch": 0.6696065266927084,
      "grad_norm": 10.8167085647583,
      "learning_rate": 1.2323341720720634e-06,
      "loss": 3.7603,
      "step": 65825
    },
    {
      "epoch": 0.6696573893229166,
      "grad_norm": 16.594776153564453,
      "learning_rate": 1.2319895280919583e-06,
      "loss": 3.1545,
      "step": 65830
    },
    {
      "epoch": 0.669708251953125,
      "grad_norm": 12.708965301513672,
      "learning_rate": 1.2316449165536007e-06,
      "loss": 3.4222,
      "step": 65835
    },
    {
      "epoch": 0.6697591145833334,
      "grad_norm": 10.709924697875977,
      "learning_rate": 1.2313003374658063e-06,
      "loss": 3.4012,
      "step": 65840
    },
    {
      "epoch": 0.6698099772135416,
      "grad_norm": 11.355735778808594,
      "learning_rate": 1.2309557908373923e-06,
      "loss": 3.2987,
      "step": 65845
    },
    {
      "epoch": 0.66986083984375,
      "grad_norm": 13.788281440734863,
      "learning_rate": 1.2306112766771744e-06,
      "loss": 2.961,
      "step": 65850
    },
    {
      "epoch": 0.6699117024739584,
      "grad_norm": 13.849200248718262,
      "learning_rate": 1.2302667949939662e-06,
      "loss": 3.3111,
      "step": 65855
    },
    {
      "epoch": 0.6699625651041666,
      "grad_norm": 13.513203620910645,
      "learning_rate": 1.2299223457965798e-06,
      "loss": 3.6833,
      "step": 65860
    },
    {
      "epoch": 0.670013427734375,
      "grad_norm": 10.663004875183105,
      "learning_rate": 1.2295779290938304e-06,
      "loss": 3.1343,
      "step": 65865
    },
    {
      "epoch": 0.6700642903645834,
      "grad_norm": 16.419111251831055,
      "learning_rate": 1.2292335448945275e-06,
      "loss": 3.4315,
      "step": 65870
    },
    {
      "epoch": 0.6701151529947916,
      "grad_norm": 10.535104751586914,
      "learning_rate": 1.2288891932074831e-06,
      "loss": 3.0774,
      "step": 65875
    },
    {
      "epoch": 0.670166015625,
      "grad_norm": 10.039164543151855,
      "learning_rate": 1.2285448740415085e-06,
      "loss": 3.3215,
      "step": 65880
    },
    {
      "epoch": 0.6702168782552084,
      "grad_norm": 15.77993392944336,
      "learning_rate": 1.2282005874054118e-06,
      "loss": 3.4125,
      "step": 65885
    },
    {
      "epoch": 0.6702677408854166,
      "grad_norm": 11.289288520812988,
      "learning_rate": 1.2278563333080017e-06,
      "loss": 3.9813,
      "step": 65890
    },
    {
      "epoch": 0.670318603515625,
      "grad_norm": 9.770861625671387,
      "learning_rate": 1.2275121117580848e-06,
      "loss": 3.2592,
      "step": 65895
    },
    {
      "epoch": 0.6703694661458334,
      "grad_norm": 10.595755577087402,
      "learning_rate": 1.227167922764469e-06,
      "loss": 3.3272,
      "step": 65900
    },
    {
      "epoch": 0.6704203287760416,
      "grad_norm": 13.553621292114258,
      "learning_rate": 1.226823766335961e-06,
      "loss": 3.142,
      "step": 65905
    },
    {
      "epoch": 0.67047119140625,
      "grad_norm": 10.56498908996582,
      "learning_rate": 1.226479642481365e-06,
      "loss": 3.2443,
      "step": 65910
    },
    {
      "epoch": 0.6705220540364584,
      "grad_norm": 9.785940170288086,
      "learning_rate": 1.226135551209485e-06,
      "loss": 3.1709,
      "step": 65915
    },
    {
      "epoch": 0.6705729166666666,
      "grad_norm": 18.74580192565918,
      "learning_rate": 1.2257914925291258e-06,
      "loss": 3.6727,
      "step": 65920
    },
    {
      "epoch": 0.670623779296875,
      "grad_norm": 14.1307373046875,
      "learning_rate": 1.2254474664490882e-06,
      "loss": 3.3455,
      "step": 65925
    },
    {
      "epoch": 0.6706746419270834,
      "grad_norm": 13.096781730651855,
      "learning_rate": 1.2251034729781761e-06,
      "loss": 3.0302,
      "step": 65930
    },
    {
      "epoch": 0.6707255045572916,
      "grad_norm": 8.60185432434082,
      "learning_rate": 1.224759512125188e-06,
      "loss": 3.2431,
      "step": 65935
    },
    {
      "epoch": 0.6707763671875,
      "grad_norm": 13.76044750213623,
      "learning_rate": 1.2244155838989272e-06,
      "loss": 3.061,
      "step": 65940
    },
    {
      "epoch": 0.6708272298177084,
      "grad_norm": 14.920628547668457,
      "learning_rate": 1.2240716883081902e-06,
      "loss": 3.2226,
      "step": 65945
    },
    {
      "epoch": 0.6708780924479166,
      "grad_norm": 14.226794242858887,
      "learning_rate": 1.2237278253617774e-06,
      "loss": 3.6226,
      "step": 65950
    },
    {
      "epoch": 0.670928955078125,
      "grad_norm": 8.770042419433594,
      "learning_rate": 1.2233839950684848e-06,
      "loss": 3.4317,
      "step": 65955
    },
    {
      "epoch": 0.6709798177083334,
      "grad_norm": 11.76567554473877,
      "learning_rate": 1.223040197437111e-06,
      "loss": 3.2916,
      "step": 65960
    },
    {
      "epoch": 0.6710306803385416,
      "grad_norm": 16.681499481201172,
      "learning_rate": 1.2226964324764513e-06,
      "loss": 3.2829,
      "step": 65965
    },
    {
      "epoch": 0.67108154296875,
      "grad_norm": 14.340017318725586,
      "learning_rate": 1.2223527001952992e-06,
      "loss": 3.1534,
      "step": 65970
    },
    {
      "epoch": 0.6711324055989584,
      "grad_norm": 18.81924057006836,
      "learning_rate": 1.2220090006024506e-06,
      "loss": 3.2541,
      "step": 65975
    },
    {
      "epoch": 0.6711832682291666,
      "grad_norm": 13.129728317260742,
      "learning_rate": 1.2216653337067e-06,
      "loss": 3.2745,
      "step": 65980
    },
    {
      "epoch": 0.671234130859375,
      "grad_norm": 9.782035827636719,
      "learning_rate": 1.2213216995168386e-06,
      "loss": 3.3167,
      "step": 65985
    },
    {
      "epoch": 0.6712849934895834,
      "grad_norm": 16.59641456604004,
      "learning_rate": 1.2209780980416577e-06,
      "loss": 3.2621,
      "step": 65990
    },
    {
      "epoch": 0.6713358561197916,
      "grad_norm": 15.288775444030762,
      "learning_rate": 1.22063452928995e-06,
      "loss": 3.3434,
      "step": 65995
    },
    {
      "epoch": 0.67138671875,
      "grad_norm": 9.345989227294922,
      "learning_rate": 1.2202909932705034e-06,
      "loss": 3.2314,
      "step": 66000
    },
    {
      "epoch": 0.6714375813802084,
      "grad_norm": 14.896824836730957,
      "learning_rate": 1.2199474899921095e-06,
      "loss": 3.7613,
      "step": 66005
    },
    {
      "epoch": 0.6714884440104166,
      "grad_norm": 12.153127670288086,
      "learning_rate": 1.2196040194635547e-06,
      "loss": 2.748,
      "step": 66010
    },
    {
      "epoch": 0.671539306640625,
      "grad_norm": 9.977337837219238,
      "learning_rate": 1.2192605816936285e-06,
      "loss": 3.1902,
      "step": 66015
    },
    {
      "epoch": 0.6715901692708334,
      "grad_norm": 13.28049087524414,
      "learning_rate": 1.2189171766911168e-06,
      "loss": 3.3504,
      "step": 66020
    },
    {
      "epoch": 0.6716410319010416,
      "grad_norm": 13.772388458251953,
      "learning_rate": 1.2185738044648044e-06,
      "loss": 3.5055,
      "step": 66025
    },
    {
      "epoch": 0.67169189453125,
      "grad_norm": 16.815696716308594,
      "learning_rate": 1.2182304650234778e-06,
      "loss": 3.1191,
      "step": 66030
    },
    {
      "epoch": 0.6717427571614584,
      "grad_norm": 10.708442687988281,
      "learning_rate": 1.2178871583759213e-06,
      "loss": 3.3703,
      "step": 66035
    },
    {
      "epoch": 0.6717936197916666,
      "grad_norm": 12.045191764831543,
      "learning_rate": 1.2175438845309185e-06,
      "loss": 3.4816,
      "step": 66040
    },
    {
      "epoch": 0.671844482421875,
      "grad_norm": 13.626344680786133,
      "learning_rate": 1.2172006434972514e-06,
      "loss": 3.7009,
      "step": 66045
    },
    {
      "epoch": 0.6718953450520834,
      "grad_norm": 11.73372745513916,
      "learning_rate": 1.2168574352837006e-06,
      "loss": 2.9432,
      "step": 66050
    },
    {
      "epoch": 0.6719462076822916,
      "grad_norm": 12.726306915283203,
      "learning_rate": 1.216514259899048e-06,
      "loss": 2.8776,
      "step": 66055
    },
    {
      "epoch": 0.6719970703125,
      "grad_norm": 16.720542907714844,
      "learning_rate": 1.2161711173520754e-06,
      "loss": 3.4371,
      "step": 66060
    },
    {
      "epoch": 0.6720479329427084,
      "grad_norm": 13.193948745727539,
      "learning_rate": 1.215828007651559e-06,
      "loss": 3.7822,
      "step": 66065
    },
    {
      "epoch": 0.6720987955729166,
      "grad_norm": 9.035006523132324,
      "learning_rate": 1.2154849308062799e-06,
      "loss": 3.2921,
      "step": 66070
    },
    {
      "epoch": 0.672149658203125,
      "grad_norm": 17.360515594482422,
      "learning_rate": 1.2151418868250142e-06,
      "loss": 3.8383,
      "step": 66075
    },
    {
      "epoch": 0.6722005208333334,
      "grad_norm": 14.382808685302734,
      "learning_rate": 1.214798875716538e-06,
      "loss": 3.5088,
      "step": 66080
    },
    {
      "epoch": 0.6722513834635416,
      "grad_norm": 8.316353797912598,
      "learning_rate": 1.2144558974896278e-06,
      "loss": 2.9866,
      "step": 66085
    },
    {
      "epoch": 0.67230224609375,
      "grad_norm": 12.267139434814453,
      "learning_rate": 1.2141129521530598e-06,
      "loss": 3.2495,
      "step": 66090
    },
    {
      "epoch": 0.6723531087239584,
      "grad_norm": 13.435066223144531,
      "learning_rate": 1.2137700397156069e-06,
      "loss": 3.3119,
      "step": 66095
    },
    {
      "epoch": 0.6724039713541666,
      "grad_norm": 12.881227493286133,
      "learning_rate": 1.213427160186042e-06,
      "loss": 3.1359,
      "step": 66100
    },
    {
      "epoch": 0.672454833984375,
      "grad_norm": 11.653464317321777,
      "learning_rate": 1.2130843135731379e-06,
      "loss": 3.1652,
      "step": 66105
    },
    {
      "epoch": 0.6725056966145834,
      "grad_norm": 16.216772079467773,
      "learning_rate": 1.212741499885668e-06,
      "loss": 3.8288,
      "step": 66110
    },
    {
      "epoch": 0.6725565592447916,
      "grad_norm": 13.487979888916016,
      "learning_rate": 1.2123987191324016e-06,
      "loss": 3.4653,
      "step": 66115
    },
    {
      "epoch": 0.672607421875,
      "grad_norm": 10.186954498291016,
      "learning_rate": 1.2120559713221087e-06,
      "loss": 3.3901,
      "step": 66120
    },
    {
      "epoch": 0.6726582845052084,
      "grad_norm": 13.468384742736816,
      "learning_rate": 1.2117132564635572e-06,
      "loss": 3.4884,
      "step": 66125
    },
    {
      "epoch": 0.6727091471354166,
      "grad_norm": 8.447620391845703,
      "learning_rate": 1.2113705745655173e-06,
      "loss": 3.0469,
      "step": 66130
    },
    {
      "epoch": 0.672760009765625,
      "grad_norm": 14.793883323669434,
      "learning_rate": 1.2110279256367563e-06,
      "loss": 3.0333,
      "step": 66135
    },
    {
      "epoch": 0.6728108723958334,
      "grad_norm": 11.138039588928223,
      "learning_rate": 1.2106853096860404e-06,
      "loss": 3.3258,
      "step": 66140
    },
    {
      "epoch": 0.6728617350260416,
      "grad_norm": 31.979942321777344,
      "learning_rate": 1.2103427267221344e-06,
      "loss": 3.856,
      "step": 66145
    },
    {
      "epoch": 0.67291259765625,
      "grad_norm": 15.37503433227539,
      "learning_rate": 1.210000176753805e-06,
      "loss": 3.3685,
      "step": 66150
    },
    {
      "epoch": 0.6729634602864584,
      "grad_norm": 12.416085243225098,
      "learning_rate": 1.2096576597898145e-06,
      "loss": 3.3378,
      "step": 66155
    },
    {
      "epoch": 0.6730143229166666,
      "grad_norm": 15.238469123840332,
      "learning_rate": 1.2093151758389276e-06,
      "loss": 3.1085,
      "step": 66160
    },
    {
      "epoch": 0.673065185546875,
      "grad_norm": 9.534880638122559,
      "learning_rate": 1.2089727249099052e-06,
      "loss": 3.3571,
      "step": 66165
    },
    {
      "epoch": 0.6731160481770834,
      "grad_norm": 9.172733306884766,
      "learning_rate": 1.2086303070115104e-06,
      "loss": 3.7219,
      "step": 66170
    },
    {
      "epoch": 0.6731669108072916,
      "grad_norm": 12.223248481750488,
      "learning_rate": 1.2082879221525032e-06,
      "loss": 3.1753,
      "step": 66175
    },
    {
      "epoch": 0.6732177734375,
      "grad_norm": 13.02455997467041,
      "learning_rate": 1.2079455703416425e-06,
      "loss": 3.2062,
      "step": 66180
    },
    {
      "epoch": 0.6732686360677084,
      "grad_norm": 13.005733489990234,
      "learning_rate": 1.2076032515876881e-06,
      "loss": 3.0596,
      "step": 66185
    },
    {
      "epoch": 0.6733194986979166,
      "grad_norm": 12.467793464660645,
      "learning_rate": 1.207260965899399e-06,
      "loss": 3.3055,
      "step": 66190
    },
    {
      "epoch": 0.673370361328125,
      "grad_norm": 12.573270797729492,
      "learning_rate": 1.2069187132855317e-06,
      "loss": 3.7388,
      "step": 66195
    },
    {
      "epoch": 0.6734212239583334,
      "grad_norm": 13.337157249450684,
      "learning_rate": 1.2065764937548419e-06,
      "loss": 3.2291,
      "step": 66200
    },
    {
      "epoch": 0.6734720865885416,
      "grad_norm": 9.928634643554688,
      "learning_rate": 1.2062343073160867e-06,
      "loss": 3.0714,
      "step": 66205
    },
    {
      "epoch": 0.67352294921875,
      "grad_norm": 10.728508949279785,
      "learning_rate": 1.2058921539780192e-06,
      "loss": 4.0894,
      "step": 66210
    },
    {
      "epoch": 0.6735738118489584,
      "grad_norm": 12.563323020935059,
      "learning_rate": 1.205550033749395e-06,
      "loss": 3.1733,
      "step": 66215
    },
    {
      "epoch": 0.6736246744791666,
      "grad_norm": 9.485886573791504,
      "learning_rate": 1.2052079466389658e-06,
      "loss": 3.3199,
      "step": 66220
    },
    {
      "epoch": 0.673675537109375,
      "grad_norm": 8.653766632080078,
      "learning_rate": 1.204865892655485e-06,
      "loss": 4.2234,
      "step": 66225
    },
    {
      "epoch": 0.6737263997395834,
      "grad_norm": 14.788777351379395,
      "learning_rate": 1.2045238718077035e-06,
      "loss": 3.2298,
      "step": 66230
    },
    {
      "epoch": 0.6737772623697916,
      "grad_norm": 12.110572814941406,
      "learning_rate": 1.2041818841043707e-06,
      "loss": 3.0991,
      "step": 66235
    },
    {
      "epoch": 0.673828125,
      "grad_norm": 7.55382776260376,
      "learning_rate": 1.203839929554237e-06,
      "loss": 3.5065,
      "step": 66240
    },
    {
      "epoch": 0.6738789876302084,
      "grad_norm": 13.87610912322998,
      "learning_rate": 1.2034980081660527e-06,
      "loss": 3.0659,
      "step": 66245
    },
    {
      "epoch": 0.6739298502604166,
      "grad_norm": 13.413716316223145,
      "learning_rate": 1.2031561199485644e-06,
      "loss": 3.2979,
      "step": 66250
    },
    {
      "epoch": 0.673980712890625,
      "grad_norm": 8.117220878601074,
      "learning_rate": 1.2028142649105196e-06,
      "loss": 3.3326,
      "step": 66255
    },
    {
      "epoch": 0.6740315755208334,
      "grad_norm": 14.149643898010254,
      "learning_rate": 1.2024724430606624e-06,
      "loss": 2.9656,
      "step": 66260
    },
    {
      "epoch": 0.6740824381510416,
      "grad_norm": 12.142016410827637,
      "learning_rate": 1.202130654407742e-06,
      "loss": 2.9977,
      "step": 66265
    },
    {
      "epoch": 0.67413330078125,
      "grad_norm": 8.97005844116211,
      "learning_rate": 1.201788898960501e-06,
      "loss": 3.6632,
      "step": 66270
    },
    {
      "epoch": 0.6741841634114584,
      "grad_norm": 11.278656959533691,
      "learning_rate": 1.2014471767276827e-06,
      "loss": 3.1504,
      "step": 66275
    },
    {
      "epoch": 0.6742350260416666,
      "grad_norm": 11.183878898620605,
      "learning_rate": 1.2011054877180314e-06,
      "loss": 3.4126,
      "step": 66280
    },
    {
      "epoch": 0.674285888671875,
      "grad_norm": 13.596809387207031,
      "learning_rate": 1.2007638319402877e-06,
      "loss": 3.3145,
      "step": 66285
    },
    {
      "epoch": 0.6743367513020834,
      "grad_norm": 12.224353790283203,
      "learning_rate": 1.200422209403194e-06,
      "loss": 3.4264,
      "step": 66290
    },
    {
      "epoch": 0.6743876139322916,
      "grad_norm": 12.031774520874023,
      "learning_rate": 1.200080620115489e-06,
      "loss": 3.3783,
      "step": 66295
    },
    {
      "epoch": 0.6744384765625,
      "grad_norm": 11.176274299621582,
      "learning_rate": 1.1997390640859143e-06,
      "loss": 3.3811,
      "step": 66300
    },
    {
      "epoch": 0.6744893391927084,
      "grad_norm": 10.890926361083984,
      "learning_rate": 1.1993975413232076e-06,
      "loss": 3.2045,
      "step": 66305
    },
    {
      "epoch": 0.6745402018229166,
      "grad_norm": 11.679183006286621,
      "learning_rate": 1.1990560518361056e-06,
      "loss": 3.306,
      "step": 66310
    },
    {
      "epoch": 0.674591064453125,
      "grad_norm": 7.901872158050537,
      "learning_rate": 1.1987145956333458e-06,
      "loss": 3.4559,
      "step": 66315
    },
    {
      "epoch": 0.6746419270833334,
      "grad_norm": 12.083430290222168,
      "learning_rate": 1.198373172723666e-06,
      "loss": 2.959,
      "step": 66320
    },
    {
      "epoch": 0.6746927897135416,
      "grad_norm": 15.67698860168457,
      "learning_rate": 1.1980317831157998e-06,
      "loss": 3.7072,
      "step": 66325
    },
    {
      "epoch": 0.67474365234375,
      "grad_norm": 8.207538604736328,
      "learning_rate": 1.1976904268184816e-06,
      "loss": 3.5721,
      "step": 66330
    },
    {
      "epoch": 0.6747945149739584,
      "grad_norm": 7.630964279174805,
      "learning_rate": 1.1973491038404444e-06,
      "loss": 3.4813,
      "step": 66335
    },
    {
      "epoch": 0.6748453776041666,
      "grad_norm": 20.761432647705078,
      "learning_rate": 1.1970078141904218e-06,
      "loss": 3.6052,
      "step": 66340
    },
    {
      "epoch": 0.674896240234375,
      "grad_norm": 7.65514612197876,
      "learning_rate": 1.1966665578771463e-06,
      "loss": 3.295,
      "step": 66345
    },
    {
      "epoch": 0.6749471028645834,
      "grad_norm": 11.542274475097656,
      "learning_rate": 1.196325334909348e-06,
      "loss": 3.3361,
      "step": 66350
    },
    {
      "epoch": 0.6749979654947916,
      "grad_norm": 16.69321060180664,
      "learning_rate": 1.195984145295756e-06,
      "loss": 2.9858,
      "step": 66355
    },
    {
      "epoch": 0.675048828125,
      "grad_norm": 16.257966995239258,
      "learning_rate": 1.1956429890451016e-06,
      "loss": 3.1568,
      "step": 66360
    },
    {
      "epoch": 0.6750996907552084,
      "grad_norm": 13.692949295043945,
      "learning_rate": 1.1953018661661112e-06,
      "loss": 2.965,
      "step": 66365
    },
    {
      "epoch": 0.6751505533854166,
      "grad_norm": 15.775273323059082,
      "learning_rate": 1.194960776667513e-06,
      "loss": 3.5861,
      "step": 66370
    },
    {
      "epoch": 0.675201416015625,
      "grad_norm": 15.816407203674316,
      "learning_rate": 1.1946197205580354e-06,
      "loss": 3.4412,
      "step": 66375
    },
    {
      "epoch": 0.6752522786458334,
      "grad_norm": 8.514810562133789,
      "learning_rate": 1.1942786978464024e-06,
      "loss": 3.2944,
      "step": 66380
    },
    {
      "epoch": 0.6753031412760416,
      "grad_norm": 12.756802558898926,
      "learning_rate": 1.1939377085413397e-06,
      "loss": 3.3257,
      "step": 66385
    },
    {
      "epoch": 0.67535400390625,
      "grad_norm": 13.349703788757324,
      "learning_rate": 1.1935967526515701e-06,
      "loss": 3.1229,
      "step": 66390
    },
    {
      "epoch": 0.6754048665364584,
      "grad_norm": 11.398979187011719,
      "learning_rate": 1.1932558301858182e-06,
      "loss": 3.1223,
      "step": 66395
    },
    {
      "epoch": 0.6754557291666666,
      "grad_norm": 11.720551490783691,
      "learning_rate": 1.1929149411528068e-06,
      "loss": 3.0941,
      "step": 66400
    },
    {
      "epoch": 0.675506591796875,
      "grad_norm": 14.701722145080566,
      "learning_rate": 1.1925740855612569e-06,
      "loss": 3.6169,
      "step": 66405
    },
    {
      "epoch": 0.6755574544270834,
      "grad_norm": 15.986696243286133,
      "learning_rate": 1.1922332634198883e-06,
      "loss": 3.8688,
      "step": 66410
    },
    {
      "epoch": 0.6756083170572916,
      "grad_norm": 11.06942081451416,
      "learning_rate": 1.1918924747374214e-06,
      "loss": 3.0135,
      "step": 66415
    },
    {
      "epoch": 0.6756591796875,
      "grad_norm": 10.790997505187988,
      "learning_rate": 1.1915517195225768e-06,
      "loss": 3.078,
      "step": 66420
    },
    {
      "epoch": 0.6757100423177084,
      "grad_norm": 9.799771308898926,
      "learning_rate": 1.191210997784071e-06,
      "loss": 3.1423,
      "step": 66425
    },
    {
      "epoch": 0.6757609049479166,
      "grad_norm": 11.02479076385498,
      "learning_rate": 1.1908703095306207e-06,
      "loss": 3.755,
      "step": 66430
    },
    {
      "epoch": 0.675811767578125,
      "grad_norm": 13.578268051147461,
      "learning_rate": 1.1905296547709445e-06,
      "loss": 3.5172,
      "step": 66435
    },
    {
      "epoch": 0.6758626302083334,
      "grad_norm": 9.204440116882324,
      "learning_rate": 1.1901890335137553e-06,
      "loss": 3.8079,
      "step": 66440
    },
    {
      "epoch": 0.6759134928385416,
      "grad_norm": 14.687267303466797,
      "learning_rate": 1.1898484457677703e-06,
      "loss": 3.5726,
      "step": 66445
    },
    {
      "epoch": 0.67596435546875,
      "grad_norm": 15.788227081298828,
      "learning_rate": 1.1895078915417016e-06,
      "loss": 2.9446,
      "step": 66450
    },
    {
      "epoch": 0.6760152180989584,
      "grad_norm": 10.923992156982422,
      "learning_rate": 1.1891673708442636e-06,
      "loss": 3.3139,
      "step": 66455
    },
    {
      "epoch": 0.6760660807291666,
      "grad_norm": 12.538166999816895,
      "learning_rate": 1.1888268836841677e-06,
      "loss": 3.6844,
      "step": 66460
    },
    {
      "epoch": 0.676116943359375,
      "grad_norm": 15.767918586730957,
      "learning_rate": 1.1884864300701244e-06,
      "loss": 3.818,
      "step": 66465
    },
    {
      "epoch": 0.6761678059895834,
      "grad_norm": 14.704933166503906,
      "learning_rate": 1.1881460100108447e-06,
      "loss": 3.2788,
      "step": 66470
    },
    {
      "epoch": 0.6762186686197916,
      "grad_norm": 8.814567565917969,
      "learning_rate": 1.1878056235150395e-06,
      "loss": 3.4925,
      "step": 66475
    },
    {
      "epoch": 0.67626953125,
      "grad_norm": 13.270614624023438,
      "learning_rate": 1.1874652705914163e-06,
      "loss": 3.2426,
      "step": 66480
    },
    {
      "epoch": 0.6763203938802084,
      "grad_norm": 9.116535186767578,
      "learning_rate": 1.1871249512486821e-06,
      "loss": 3.1969,
      "step": 66485
    },
    {
      "epoch": 0.6763712565104166,
      "grad_norm": 15.07793140411377,
      "learning_rate": 1.1867846654955458e-06,
      "loss": 3.1105,
      "step": 66490
    },
    {
      "epoch": 0.676422119140625,
      "grad_norm": 13.533629417419434,
      "learning_rate": 1.1864444133407119e-06,
      "loss": 3.048,
      "step": 66495
    },
    {
      "epoch": 0.6764729817708334,
      "grad_norm": 11.392133712768555,
      "learning_rate": 1.186104194792887e-06,
      "loss": 3.2305,
      "step": 66500
    },
    {
      "epoch": 0.6765238444010416,
      "grad_norm": 9.979619026184082,
      "learning_rate": 1.1857640098607739e-06,
      "loss": 2.9928,
      "step": 66505
    },
    {
      "epoch": 0.67657470703125,
      "grad_norm": 7.802496433258057,
      "learning_rate": 1.185423858553078e-06,
      "loss": 3.1146,
      "step": 66510
    },
    {
      "epoch": 0.6766255696614584,
      "grad_norm": 15.561924934387207,
      "learning_rate": 1.1850837408785013e-06,
      "loss": 3.2807,
      "step": 66515
    },
    {
      "epoch": 0.6766764322916666,
      "grad_norm": 14.786020278930664,
      "learning_rate": 1.1847436568457444e-06,
      "loss": 3.3316,
      "step": 66520
    },
    {
      "epoch": 0.676727294921875,
      "grad_norm": 15.622934341430664,
      "learning_rate": 1.1844036064635091e-06,
      "loss": 3.1424,
      "step": 66525
    },
    {
      "epoch": 0.6767781575520834,
      "grad_norm": 15.41285228729248,
      "learning_rate": 1.1840635897404967e-06,
      "loss": 3.4605,
      "step": 66530
    },
    {
      "epoch": 0.6768290201822916,
      "grad_norm": 16.166288375854492,
      "learning_rate": 1.1837236066854058e-06,
      "loss": 3.2866,
      "step": 66535
    },
    {
      "epoch": 0.6768798828125,
      "grad_norm": 11.620336532592773,
      "learning_rate": 1.1833836573069341e-06,
      "loss": 3.0053,
      "step": 66540
    },
    {
      "epoch": 0.6769307454427084,
      "grad_norm": 21.284406661987305,
      "learning_rate": 1.1830437416137786e-06,
      "loss": 3.4245,
      "step": 66545
    },
    {
      "epoch": 0.6769816080729166,
      "grad_norm": 12.208685874938965,
      "learning_rate": 1.1827038596146368e-06,
      "loss": 3.4698,
      "step": 66550
    },
    {
      "epoch": 0.677032470703125,
      "grad_norm": 13.61749267578125,
      "learning_rate": 1.1823640113182052e-06,
      "loss": 3.2414,
      "step": 66555
    },
    {
      "epoch": 0.6770833333333334,
      "grad_norm": 6.914754390716553,
      "learning_rate": 1.1820241967331775e-06,
      "loss": 3.3038,
      "step": 66560
    },
    {
      "epoch": 0.6771341959635416,
      "grad_norm": 14.918646812438965,
      "learning_rate": 1.1816844158682492e-06,
      "loss": 3.3787,
      "step": 66565
    },
    {
      "epoch": 0.67718505859375,
      "grad_norm": 16.732187271118164,
      "learning_rate": 1.1813446687321118e-06,
      "loss": 3.1693,
      "step": 66570
    },
    {
      "epoch": 0.6772359212239584,
      "grad_norm": 7.794482231140137,
      "learning_rate": 1.181004955333459e-06,
      "loss": 3.1221,
      "step": 66575
    },
    {
      "epoch": 0.6772867838541666,
      "grad_norm": 15.513784408569336,
      "learning_rate": 1.1806652756809811e-06,
      "loss": 3.147,
      "step": 66580
    },
    {
      "epoch": 0.677337646484375,
      "grad_norm": 13.813277244567871,
      "learning_rate": 1.1803256297833704e-06,
      "loss": 3.3604,
      "step": 66585
    },
    {
      "epoch": 0.6773885091145834,
      "grad_norm": 11.876188278198242,
      "learning_rate": 1.1799860176493159e-06,
      "loss": 3.2065,
      "step": 66590
    },
    {
      "epoch": 0.6774393717447916,
      "grad_norm": 11.655075073242188,
      "learning_rate": 1.179646439287505e-06,
      "loss": 3.0779,
      "step": 66595
    },
    {
      "epoch": 0.677490234375,
      "grad_norm": 13.3148193359375,
      "learning_rate": 1.179306894706627e-06,
      "loss": 3.1405,
      "step": 66600
    },
    {
      "epoch": 0.6775410970052084,
      "grad_norm": 14.856046676635742,
      "learning_rate": 1.17896738391537e-06,
      "loss": 3.5807,
      "step": 66605
    },
    {
      "epoch": 0.6775919596354166,
      "grad_norm": 11.962355613708496,
      "learning_rate": 1.1786279069224193e-06,
      "loss": 3.4211,
      "step": 66610
    },
    {
      "epoch": 0.677642822265625,
      "grad_norm": 13.368785858154297,
      "learning_rate": 1.1782884637364606e-06,
      "loss": 3.3247,
      "step": 66615
    },
    {
      "epoch": 0.6776936848958334,
      "grad_norm": 9.692715644836426,
      "learning_rate": 1.1779490543661773e-06,
      "loss": 3.3903,
      "step": 66620
    },
    {
      "epoch": 0.6777445475260416,
      "grad_norm": 12.040973663330078,
      "learning_rate": 1.177609678820254e-06,
      "loss": 3.5649,
      "step": 66625
    },
    {
      "epoch": 0.67779541015625,
      "grad_norm": 7.997753143310547,
      "learning_rate": 1.1772703371073746e-06,
      "loss": 3.2475,
      "step": 66630
    },
    {
      "epoch": 0.6778462727864584,
      "grad_norm": 11.54784870147705,
      "learning_rate": 1.1769310292362202e-06,
      "loss": 3.3519,
      "step": 66635
    },
    {
      "epoch": 0.6778971354166666,
      "grad_norm": 10.893509864807129,
      "learning_rate": 1.1765917552154707e-06,
      "loss": 3.7834,
      "step": 66640
    },
    {
      "epoch": 0.677947998046875,
      "grad_norm": 17.195344924926758,
      "learning_rate": 1.1762525150538085e-06,
      "loss": 3.4025,
      "step": 66645
    },
    {
      "epoch": 0.6779988606770834,
      "grad_norm": 12.270585060119629,
      "learning_rate": 1.175913308759911e-06,
      "loss": 3.132,
      "step": 66650
    },
    {
      "epoch": 0.6780497233072916,
      "grad_norm": 15.355175971984863,
      "learning_rate": 1.1755741363424575e-06,
      "loss": 4.0733,
      "step": 66655
    },
    {
      "epoch": 0.6781005859375,
      "grad_norm": 11.925270080566406,
      "learning_rate": 1.1752349978101269e-06,
      "loss": 3.2936,
      "step": 66660
    },
    {
      "epoch": 0.6781514485677084,
      "grad_norm": 9.525686264038086,
      "learning_rate": 1.174895893171595e-06,
      "loss": 3.6208,
      "step": 66665
    },
    {
      "epoch": 0.6782023111979166,
      "grad_norm": 8.701624870300293,
      "learning_rate": 1.1745568224355376e-06,
      "loss": 3.2159,
      "step": 66670
    },
    {
      "epoch": 0.678253173828125,
      "grad_norm": 11.974249839782715,
      "learning_rate": 1.1742177856106285e-06,
      "loss": 3.3779,
      "step": 66675
    },
    {
      "epoch": 0.6783040364583334,
      "grad_norm": 9.864870071411133,
      "learning_rate": 1.173878782705543e-06,
      "loss": 3.2289,
      "step": 66680
    },
    {
      "epoch": 0.6783548990885416,
      "grad_norm": 12.554764747619629,
      "learning_rate": 1.1735398137289559e-06,
      "loss": 3.0567,
      "step": 66685
    },
    {
      "epoch": 0.67840576171875,
      "grad_norm": 9.354292869567871,
      "learning_rate": 1.173200878689538e-06,
      "loss": 3.3045,
      "step": 66690
    },
    {
      "epoch": 0.6784566243489584,
      "grad_norm": 13.207695960998535,
      "learning_rate": 1.1728619775959602e-06,
      "loss": 3.0783,
      "step": 66695
    },
    {
      "epoch": 0.6785074869791666,
      "grad_norm": 9.954910278320312,
      "learning_rate": 1.172523110456895e-06,
      "loss": 3.2347,
      "step": 66700
    },
    {
      "epoch": 0.678558349609375,
      "grad_norm": 11.203584671020508,
      "learning_rate": 1.1721842772810108e-06,
      "loss": 3.8233,
      "step": 66705
    },
    {
      "epoch": 0.6786092122395834,
      "grad_norm": 13.040945053100586,
      "learning_rate": 1.1718454780769779e-06,
      "loss": 3.5498,
      "step": 66710
    },
    {
      "epoch": 0.6786600748697916,
      "grad_norm": 8.814359664916992,
      "learning_rate": 1.1715067128534625e-06,
      "loss": 3.3726,
      "step": 66715
    },
    {
      "epoch": 0.6787109375,
      "grad_norm": 7.560615062713623,
      "learning_rate": 1.171167981619134e-06,
      "loss": 3.3403,
      "step": 66720
    },
    {
      "epoch": 0.6787618001302084,
      "grad_norm": 10.767315864562988,
      "learning_rate": 1.1708292843826569e-06,
      "loss": 3.4577,
      "step": 66725
    },
    {
      "epoch": 0.6788126627604166,
      "grad_norm": 14.322761535644531,
      "learning_rate": 1.1704906211526985e-06,
      "loss": 3.1533,
      "step": 66730
    },
    {
      "epoch": 0.678863525390625,
      "grad_norm": 9.371920585632324,
      "learning_rate": 1.1701519919379211e-06,
      "loss": 3.5488,
      "step": 66735
    },
    {
      "epoch": 0.6789143880208334,
      "grad_norm": 16.8697452545166,
      "learning_rate": 1.169813396746991e-06,
      "loss": 3.1652,
      "step": 66740
    },
    {
      "epoch": 0.6789652506510416,
      "grad_norm": 8.889442443847656,
      "learning_rate": 1.1694748355885698e-06,
      "loss": 3.3137,
      "step": 66745
    },
    {
      "epoch": 0.67901611328125,
      "grad_norm": 15.476335525512695,
      "learning_rate": 1.169136308471319e-06,
      "loss": 3.4765,
      "step": 66750
    },
    {
      "epoch": 0.6790669759114584,
      "grad_norm": 8.582023620605469,
      "learning_rate": 1.1687978154038998e-06,
      "loss": 3.2237,
      "step": 66755
    },
    {
      "epoch": 0.6791178385416666,
      "grad_norm": 8.053441047668457,
      "learning_rate": 1.1684593563949739e-06,
      "loss": 2.9329,
      "step": 66760
    },
    {
      "epoch": 0.679168701171875,
      "grad_norm": 10.91099739074707,
      "learning_rate": 1.1681209314532e-06,
      "loss": 3.3324,
      "step": 66765
    },
    {
      "epoch": 0.6792195638020834,
      "grad_norm": 16.067745208740234,
      "learning_rate": 1.1677825405872353e-06,
      "loss": 3.4072,
      "step": 66770
    },
    {
      "epoch": 0.6792704264322916,
      "grad_norm": 8.088774681091309,
      "learning_rate": 1.1674441838057396e-06,
      "loss": 3.0362,
      "step": 66775
    },
    {
      "epoch": 0.6793212890625,
      "grad_norm": 10.353334426879883,
      "learning_rate": 1.1671058611173676e-06,
      "loss": 3.1913,
      "step": 66780
    },
    {
      "epoch": 0.6793721516927084,
      "grad_norm": 9.857560157775879,
      "learning_rate": 1.1667675725307772e-06,
      "loss": 3.5657,
      "step": 66785
    },
    {
      "epoch": 0.6794230143229166,
      "grad_norm": 10.71377944946289,
      "learning_rate": 1.1664293180546216e-06,
      "loss": 3.1443,
      "step": 66790
    },
    {
      "epoch": 0.679473876953125,
      "grad_norm": 12.475671768188477,
      "learning_rate": 1.1660910976975566e-06,
      "loss": 3.2034,
      "step": 66795
    },
    {
      "epoch": 0.6795247395833334,
      "grad_norm": 12.8106050491333,
      "learning_rate": 1.1657529114682344e-06,
      "loss": 3.064,
      "step": 66800
    },
    {
      "epoch": 0.6795756022135416,
      "grad_norm": 12.289118766784668,
      "learning_rate": 1.1654147593753073e-06,
      "loss": 3.429,
      "step": 66805
    },
    {
      "epoch": 0.67962646484375,
      "grad_norm": 13.987550735473633,
      "learning_rate": 1.165076641427427e-06,
      "loss": 3.2674,
      "step": 66810
    },
    {
      "epoch": 0.6796773274739584,
      "grad_norm": 12.636734008789062,
      "learning_rate": 1.1647385576332455e-06,
      "loss": 3.2572,
      "step": 66815
    },
    {
      "epoch": 0.6797281901041666,
      "grad_norm": 9.804733276367188,
      "learning_rate": 1.1644005080014111e-06,
      "loss": 3.2244,
      "step": 66820
    },
    {
      "epoch": 0.679779052734375,
      "grad_norm": 11.967543601989746,
      "learning_rate": 1.1640624925405736e-06,
      "loss": 3.1633,
      "step": 66825
    },
    {
      "epoch": 0.6798299153645834,
      "grad_norm": 13.213789939880371,
      "learning_rate": 1.1637245112593793e-06,
      "loss": 3.2825,
      "step": 66830
    },
    {
      "epoch": 0.6798807779947916,
      "grad_norm": 14.687386512756348,
      "learning_rate": 1.1633865641664768e-06,
      "loss": 3.4654,
      "step": 66835
    },
    {
      "epoch": 0.679931640625,
      "grad_norm": 9.610382080078125,
      "learning_rate": 1.1630486512705132e-06,
      "loss": 2.8534,
      "step": 66840
    },
    {
      "epoch": 0.6799825032552084,
      "grad_norm": 12.395421981811523,
      "learning_rate": 1.1627107725801329e-06,
      "loss": 3.5905,
      "step": 66845
    },
    {
      "epoch": 0.6800333658854166,
      "grad_norm": 13.9788818359375,
      "learning_rate": 1.1623729281039798e-06,
      "loss": 3.6113,
      "step": 66850
    },
    {
      "epoch": 0.680084228515625,
      "grad_norm": 11.481358528137207,
      "learning_rate": 1.162035117850699e-06,
      "loss": 3.5347,
      "step": 66855
    },
    {
      "epoch": 0.6801350911458334,
      "grad_norm": 8.43794059753418,
      "learning_rate": 1.1616973418289319e-06,
      "loss": 3.1554,
      "step": 66860
    },
    {
      "epoch": 0.6801859537760416,
      "grad_norm": 9.162663459777832,
      "learning_rate": 1.1613596000473208e-06,
      "loss": 3.4241,
      "step": 66865
    },
    {
      "epoch": 0.68023681640625,
      "grad_norm": 17.194950103759766,
      "learning_rate": 1.161021892514508e-06,
      "loss": 3.2849,
      "step": 66870
    },
    {
      "epoch": 0.6802876790364584,
      "grad_norm": 12.317100524902344,
      "learning_rate": 1.1606842192391326e-06,
      "loss": 3.4843,
      "step": 66875
    },
    {
      "epoch": 0.6803385416666666,
      "grad_norm": 8.26466178894043,
      "learning_rate": 1.1603465802298334e-06,
      "loss": 3.1841,
      "step": 66880
    },
    {
      "epoch": 0.680389404296875,
      "grad_norm": 8.200136184692383,
      "learning_rate": 1.160008975495249e-06,
      "loss": 3.2368,
      "step": 66885
    },
    {
      "epoch": 0.6804402669270834,
      "grad_norm": 15.516468048095703,
      "learning_rate": 1.1596714050440186e-06,
      "loss": 3.2557,
      "step": 66890
    },
    {
      "epoch": 0.6804911295572916,
      "grad_norm": 15.397497177124023,
      "learning_rate": 1.1593338688847775e-06,
      "loss": 3.3697,
      "step": 66895
    },
    {
      "epoch": 0.6805419921875,
      "grad_norm": 11.23614501953125,
      "learning_rate": 1.1589963670261618e-06,
      "loss": 3.1352,
      "step": 66900
    },
    {
      "epoch": 0.6805928548177084,
      "grad_norm": 9.06709098815918,
      "learning_rate": 1.158658899476805e-06,
      "loss": 2.9887,
      "step": 66905
    },
    {
      "epoch": 0.6806437174479166,
      "grad_norm": 8.044218063354492,
      "learning_rate": 1.1583214662453424e-06,
      "loss": 3.2688,
      "step": 66910
    },
    {
      "epoch": 0.680694580078125,
      "grad_norm": 13.742280006408691,
      "learning_rate": 1.157984067340408e-06,
      "loss": 3.3677,
      "step": 66915
    },
    {
      "epoch": 0.6807454427083334,
      "grad_norm": 7.34171724319458,
      "learning_rate": 1.157646702770633e-06,
      "loss": 3.0535,
      "step": 66920
    },
    {
      "epoch": 0.6807963053385416,
      "grad_norm": 10.71430492401123,
      "learning_rate": 1.1573093725446484e-06,
      "loss": 3.3923,
      "step": 66925
    },
    {
      "epoch": 0.68084716796875,
      "grad_norm": 8.38379192352295,
      "learning_rate": 1.156972076671086e-06,
      "loss": 3.4441,
      "step": 66930
    },
    {
      "epoch": 0.6808980305989584,
      "grad_norm": 13.166692733764648,
      "learning_rate": 1.1566348151585735e-06,
      "loss": 3.2246,
      "step": 66935
    },
    {
      "epoch": 0.6809488932291666,
      "grad_norm": 12.907691955566406,
      "learning_rate": 1.156297588015742e-06,
      "loss": 3.3417,
      "step": 66940
    },
    {
      "epoch": 0.680999755859375,
      "grad_norm": 12.744908332824707,
      "learning_rate": 1.1559603952512175e-06,
      "loss": 3.2215,
      "step": 66945
    },
    {
      "epoch": 0.6810506184895834,
      "grad_norm": 15.622496604919434,
      "learning_rate": 1.1556232368736283e-06,
      "loss": 3.1477,
      "step": 66950
    },
    {
      "epoch": 0.6811014811197916,
      "grad_norm": 11.325515747070312,
      "learning_rate": 1.1552861128915998e-06,
      "loss": 3.243,
      "step": 66955
    },
    {
      "epoch": 0.68115234375,
      "grad_norm": 14.483476638793945,
      "learning_rate": 1.1549490233137569e-06,
      "loss": 3.4576,
      "step": 66960
    },
    {
      "epoch": 0.6812032063802084,
      "grad_norm": 12.413887977600098,
      "learning_rate": 1.154611968148724e-06,
      "loss": 2.9391,
      "step": 66965
    },
    {
      "epoch": 0.6812540690104166,
      "grad_norm": 12.793086051940918,
      "learning_rate": 1.1542749474051263e-06,
      "loss": 2.8667,
      "step": 66970
    },
    {
      "epoch": 0.681304931640625,
      "grad_norm": 8.617049217224121,
      "learning_rate": 1.1539379610915848e-06,
      "loss": 3.3601,
      "step": 66975
    },
    {
      "epoch": 0.6813557942708334,
      "grad_norm": 8.009932518005371,
      "learning_rate": 1.1536010092167208e-06,
      "loss": 3.2637,
      "step": 66980
    },
    {
      "epoch": 0.6814066569010416,
      "grad_norm": 9.720005989074707,
      "learning_rate": 1.1532640917891568e-06,
      "loss": 3.5085,
      "step": 66985
    },
    {
      "epoch": 0.68145751953125,
      "grad_norm": 8.442168235778809,
      "learning_rate": 1.152927208817511e-06,
      "loss": 2.8223,
      "step": 66990
    },
    {
      "epoch": 0.6815083821614584,
      "grad_norm": 9.75130558013916,
      "learning_rate": 1.1525903603104039e-06,
      "loss": 3.4401,
      "step": 66995
    },
    {
      "epoch": 0.6815592447916666,
      "grad_norm": 11.65346622467041,
      "learning_rate": 1.1522535462764519e-06,
      "loss": 3.1385,
      "step": 67000
    },
    {
      "epoch": 0.681610107421875,
      "grad_norm": 12.073299407958984,
      "learning_rate": 1.1519167667242748e-06,
      "loss": 3.1029,
      "step": 67005
    },
    {
      "epoch": 0.6816609700520834,
      "grad_norm": 10.075000762939453,
      "learning_rate": 1.1515800216624875e-06,
      "loss": 3.1759,
      "step": 67010
    },
    {
      "epoch": 0.6817118326822916,
      "grad_norm": 9.205047607421875,
      "learning_rate": 1.151243311099705e-06,
      "loss": 3.2337,
      "step": 67015
    },
    {
      "epoch": 0.6817626953125,
      "grad_norm": 11.416351318359375,
      "learning_rate": 1.1509066350445424e-06,
      "loss": 3.4893,
      "step": 67020
    },
    {
      "epoch": 0.6818135579427084,
      "grad_norm": 15.491963386535645,
      "learning_rate": 1.150569993505615e-06,
      "loss": 3.3777,
      "step": 67025
    },
    {
      "epoch": 0.6818644205729166,
      "grad_norm": 8.515227317810059,
      "learning_rate": 1.1502333864915343e-06,
      "loss": 3.179,
      "step": 67030
    },
    {
      "epoch": 0.681915283203125,
      "grad_norm": 9.592573165893555,
      "learning_rate": 1.1498968140109118e-06,
      "loss": 3.169,
      "step": 67035
    },
    {
      "epoch": 0.6819661458333334,
      "grad_norm": 10.023240089416504,
      "learning_rate": 1.149560276072359e-06,
      "loss": 3.1083,
      "step": 67040
    },
    {
      "epoch": 0.6820170084635416,
      "grad_norm": 10.103043556213379,
      "learning_rate": 1.1492237726844877e-06,
      "loss": 3.141,
      "step": 67045
    },
    {
      "epoch": 0.68206787109375,
      "grad_norm": 9.0494384765625,
      "learning_rate": 1.1488873038559055e-06,
      "loss": 3.1736,
      "step": 67050
    },
    {
      "epoch": 0.6821187337239584,
      "grad_norm": 14.495260238647461,
      "learning_rate": 1.1485508695952211e-06,
      "loss": 3.2705,
      "step": 67055
    },
    {
      "epoch": 0.6821695963541666,
      "grad_norm": 9.811225891113281,
      "learning_rate": 1.1482144699110427e-06,
      "loss": 3.1328,
      "step": 67060
    },
    {
      "epoch": 0.682220458984375,
      "grad_norm": 8.978487014770508,
      "learning_rate": 1.1478781048119762e-06,
      "loss": 3.7037,
      "step": 67065
    },
    {
      "epoch": 0.6822713216145834,
      "grad_norm": 10.287515640258789,
      "learning_rate": 1.1475417743066286e-06,
      "loss": 3.5078,
      "step": 67070
    },
    {
      "epoch": 0.6823221842447916,
      "grad_norm": 12.768781661987305,
      "learning_rate": 1.1472054784036033e-06,
      "loss": 3.256,
      "step": 67075
    },
    {
      "epoch": 0.682373046875,
      "grad_norm": 7.16929292678833,
      "learning_rate": 1.146869217111506e-06,
      "loss": 2.967,
      "step": 67080
    },
    {
      "epoch": 0.6824239095052084,
      "grad_norm": 11.414715766906738,
      "learning_rate": 1.146532990438939e-06,
      "loss": 3.2199,
      "step": 67085
    },
    {
      "epoch": 0.6824747721354166,
      "grad_norm": 10.97852611541748,
      "learning_rate": 1.1461967983945039e-06,
      "loss": 3.3135,
      "step": 67090
    },
    {
      "epoch": 0.682525634765625,
      "grad_norm": 10.633146286010742,
      "learning_rate": 1.1458606409868026e-06,
      "loss": 3.244,
      "step": 67095
    },
    {
      "epoch": 0.6825764973958334,
      "grad_norm": 15.645685195922852,
      "learning_rate": 1.1455245182244371e-06,
      "loss": 3.2323,
      "step": 67100
    },
    {
      "epoch": 0.6826273600260416,
      "grad_norm": 11.883807182312012,
      "learning_rate": 1.1451884301160054e-06,
      "loss": 3.4085,
      "step": 67105
    },
    {
      "epoch": 0.68267822265625,
      "grad_norm": 12.355230331420898,
      "learning_rate": 1.1448523766701067e-06,
      "loss": 3.1958,
      "step": 67110
    },
    {
      "epoch": 0.6827290852864584,
      "grad_norm": 7.595726013183594,
      "learning_rate": 1.1445163578953381e-06,
      "loss": 3.1504,
      "step": 67115
    },
    {
      "epoch": 0.6827799479166666,
      "grad_norm": 8.888771057128906,
      "learning_rate": 1.1441803738002969e-06,
      "loss": 3.1938,
      "step": 67120
    },
    {
      "epoch": 0.682830810546875,
      "grad_norm": 10.272032737731934,
      "learning_rate": 1.1438444243935804e-06,
      "loss": 2.8385,
      "step": 67125
    },
    {
      "epoch": 0.6828816731770834,
      "grad_norm": 9.087508201599121,
      "learning_rate": 1.1435085096837828e-06,
      "loss": 3.0904,
      "step": 67130
    },
    {
      "epoch": 0.6829325358072916,
      "grad_norm": 7.05070686340332,
      "learning_rate": 1.1431726296794975e-06,
      "loss": 2.9817,
      "step": 67135
    },
    {
      "epoch": 0.6829833984375,
      "grad_norm": 14.386911392211914,
      "learning_rate": 1.1428367843893202e-06,
      "loss": 3.3527,
      "step": 67140
    },
    {
      "epoch": 0.6830342610677084,
      "grad_norm": 13.281949996948242,
      "learning_rate": 1.142500973821841e-06,
      "loss": 3.1643,
      "step": 67145
    },
    {
      "epoch": 0.6830851236979166,
      "grad_norm": 11.068631172180176,
      "learning_rate": 1.1421651979856523e-06,
      "loss": 3.206,
      "step": 67150
    },
    {
      "epoch": 0.683135986328125,
      "grad_norm": 12.887954711914062,
      "learning_rate": 1.1418294568893464e-06,
      "loss": 3.3041,
      "step": 67155
    },
    {
      "epoch": 0.6831868489583334,
      "grad_norm": 8.447509765625,
      "learning_rate": 1.141493750541512e-06,
      "loss": 3.5031,
      "step": 67160
    },
    {
      "epoch": 0.6832377115885416,
      "grad_norm": 9.116706848144531,
      "learning_rate": 1.1411580789507377e-06,
      "loss": 2.9863,
      "step": 67165
    },
    {
      "epoch": 0.68328857421875,
      "grad_norm": 12.063055038452148,
      "learning_rate": 1.1408224421256102e-06,
      "loss": 3.2959,
      "step": 67170
    },
    {
      "epoch": 0.6833394368489584,
      "grad_norm": 11.058523178100586,
      "learning_rate": 1.1404868400747202e-06,
      "loss": 3.2343,
      "step": 67175
    },
    {
      "epoch": 0.6833902994791666,
      "grad_norm": 12.993006706237793,
      "learning_rate": 1.1401512728066519e-06,
      "loss": 2.8858,
      "step": 67180
    },
    {
      "epoch": 0.683441162109375,
      "grad_norm": 10.988066673278809,
      "learning_rate": 1.1398157403299911e-06,
      "loss": 3.4298,
      "step": 67185
    },
    {
      "epoch": 0.6834920247395834,
      "grad_norm": 11.185561180114746,
      "learning_rate": 1.1394802426533216e-06,
      "loss": 3.2862,
      "step": 67190
    },
    {
      "epoch": 0.6835428873697916,
      "grad_norm": 15.70999813079834,
      "learning_rate": 1.1391447797852272e-06,
      "loss": 3.257,
      "step": 67195
    },
    {
      "epoch": 0.68359375,
      "grad_norm": 9.060686111450195,
      "learning_rate": 1.1388093517342922e-06,
      "loss": 3.3663,
      "step": 67200
    },
    {
      "epoch": 0.6836446126302084,
      "grad_norm": 16.518936157226562,
      "learning_rate": 1.1384739585090973e-06,
      "loss": 3.9777,
      "step": 67205
    },
    {
      "epoch": 0.6836954752604166,
      "grad_norm": 14.970965385437012,
      "learning_rate": 1.1381386001182223e-06,
      "loss": 3.4142,
      "step": 67210
    },
    {
      "epoch": 0.683746337890625,
      "grad_norm": 12.52270793914795,
      "learning_rate": 1.1378032765702496e-06,
      "loss": 3.4041,
      "step": 67215
    },
    {
      "epoch": 0.6837972005208334,
      "grad_norm": 9.38988971710205,
      "learning_rate": 1.137467987873756e-06,
      "loss": 3.3901,
      "step": 67220
    },
    {
      "epoch": 0.6838480631510416,
      "grad_norm": 15.922767639160156,
      "learning_rate": 1.1371327340373222e-06,
      "loss": 3.5588,
      "step": 67225
    },
    {
      "epoch": 0.68389892578125,
      "grad_norm": 10.0673189163208,
      "learning_rate": 1.1367975150695232e-06,
      "loss": 3.4771,
      "step": 67230
    },
    {
      "epoch": 0.6839497884114584,
      "grad_norm": 14.637181282043457,
      "learning_rate": 1.1364623309789378e-06,
      "loss": 3.7495,
      "step": 67235
    },
    {
      "epoch": 0.6840006510416666,
      "grad_norm": 9.442575454711914,
      "learning_rate": 1.1361271817741403e-06,
      "loss": 3.1345,
      "step": 67240
    },
    {
      "epoch": 0.684051513671875,
      "grad_norm": 10.584227561950684,
      "learning_rate": 1.135792067463705e-06,
      "loss": 3.081,
      "step": 67245
    },
    {
      "epoch": 0.6841023763020834,
      "grad_norm": 8.191669464111328,
      "learning_rate": 1.1354569880562057e-06,
      "loss": 3.3178,
      "step": 67250
    },
    {
      "epoch": 0.6841532389322916,
      "grad_norm": 14.132744789123535,
      "learning_rate": 1.1351219435602172e-06,
      "loss": 3.2586,
      "step": 67255
    },
    {
      "epoch": 0.6842041015625,
      "grad_norm": 17.84623908996582,
      "learning_rate": 1.1347869339843104e-06,
      "loss": 3.186,
      "step": 67260
    },
    {
      "epoch": 0.6842549641927084,
      "grad_norm": 11.495306968688965,
      "learning_rate": 1.1344519593370552e-06,
      "loss": 3.2165,
      "step": 67265
    },
    {
      "epoch": 0.6843058268229166,
      "grad_norm": 11.648442268371582,
      "learning_rate": 1.1341170196270238e-06,
      "loss": 3.3547,
      "step": 67270
    },
    {
      "epoch": 0.684356689453125,
      "grad_norm": 17.59107780456543,
      "learning_rate": 1.133782114862784e-06,
      "loss": 3.57,
      "step": 67275
    },
    {
      "epoch": 0.6844075520833334,
      "grad_norm": 10.777287483215332,
      "learning_rate": 1.1334472450529057e-06,
      "loss": 2.8832,
      "step": 67280
    },
    {
      "epoch": 0.6844584147135416,
      "grad_norm": 9.979857444763184,
      "learning_rate": 1.133112410205955e-06,
      "loss": 3.4884,
      "step": 67285
    },
    {
      "epoch": 0.68450927734375,
      "grad_norm": 11.093391418457031,
      "learning_rate": 1.1327776103304998e-06,
      "loss": 3.3343,
      "step": 67290
    },
    {
      "epoch": 0.6845601399739584,
      "grad_norm": 11.67663860321045,
      "learning_rate": 1.1324428454351058e-06,
      "loss": 3.2795,
      "step": 67295
    },
    {
      "epoch": 0.6846110026041666,
      "grad_norm": 13.728826522827148,
      "learning_rate": 1.1321081155283364e-06,
      "loss": 3.9356,
      "step": 67300
    },
    {
      "epoch": 0.684661865234375,
      "grad_norm": 17.770524978637695,
      "learning_rate": 1.1317734206187563e-06,
      "loss": 3.3391,
      "step": 67305
    },
    {
      "epoch": 0.6847127278645834,
      "grad_norm": 12.959101676940918,
      "learning_rate": 1.1314387607149304e-06,
      "loss": 3.6963,
      "step": 67310
    },
    {
      "epoch": 0.6847635904947916,
      "grad_norm": 15.099030494689941,
      "learning_rate": 1.1311041358254191e-06,
      "loss": 3.5781,
      "step": 67315
    },
    {
      "epoch": 0.684814453125,
      "grad_norm": 14.204859733581543,
      "learning_rate": 1.1307695459587841e-06,
      "loss": 3.0682,
      "step": 67320
    },
    {
      "epoch": 0.6848653157552084,
      "grad_norm": 11.338966369628906,
      "learning_rate": 1.130434991123584e-06,
      "loss": 3.0671,
      "step": 67325
    },
    {
      "epoch": 0.6849161783854166,
      "grad_norm": 10.393125534057617,
      "learning_rate": 1.1301004713283817e-06,
      "loss": 3.4269,
      "step": 67330
    },
    {
      "epoch": 0.684967041015625,
      "grad_norm": 12.963751792907715,
      "learning_rate": 1.1297659865817342e-06,
      "loss": 3.4185,
      "step": 67335
    },
    {
      "epoch": 0.6850179036458334,
      "grad_norm": 14.359334945678711,
      "learning_rate": 1.1294315368921993e-06,
      "loss": 3.2264,
      "step": 67340
    },
    {
      "epoch": 0.6850687662760416,
      "grad_norm": 11.044459342956543,
      "learning_rate": 1.1290971222683325e-06,
      "loss": 2.9845,
      "step": 67345
    },
    {
      "epoch": 0.68511962890625,
      "grad_norm": 10.04295539855957,
      "learning_rate": 1.128762742718691e-06,
      "loss": 3.3286,
      "step": 67350
    },
    {
      "epoch": 0.6851704915364584,
      "grad_norm": 11.318278312683105,
      "learning_rate": 1.1284283982518307e-06,
      "loss": 3.357,
      "step": 67355
    },
    {
      "epoch": 0.6852213541666666,
      "grad_norm": 13.759521484375,
      "learning_rate": 1.1280940888763036e-06,
      "loss": 3.1177,
      "step": 67360
    },
    {
      "epoch": 0.685272216796875,
      "grad_norm": 8.070262908935547,
      "learning_rate": 1.127759814600665e-06,
      "loss": 3.4054,
      "step": 67365
    },
    {
      "epoch": 0.6853230794270834,
      "grad_norm": 15.429027557373047,
      "learning_rate": 1.1274255754334665e-06,
      "loss": 3.0435,
      "step": 67370
    },
    {
      "epoch": 0.6853739420572916,
      "grad_norm": 13.732335090637207,
      "learning_rate": 1.1270913713832584e-06,
      "loss": 3.3527,
      "step": 67375
    },
    {
      "epoch": 0.6854248046875,
      "grad_norm": 12.337641716003418,
      "learning_rate": 1.126757202458592e-06,
      "loss": 3.2012,
      "step": 67380
    },
    {
      "epoch": 0.6854756673177084,
      "grad_norm": 8.89428997039795,
      "learning_rate": 1.126423068668018e-06,
      "loss": 2.9396,
      "step": 67385
    },
    {
      "epoch": 0.6855265299479166,
      "grad_norm": 8.475600242614746,
      "learning_rate": 1.1260889700200843e-06,
      "loss": 3.3154,
      "step": 67390
    },
    {
      "epoch": 0.685577392578125,
      "grad_norm": 14.109234809875488,
      "learning_rate": 1.1257549065233387e-06,
      "loss": 3.4301,
      "step": 67395
    },
    {
      "epoch": 0.6856282552083334,
      "grad_norm": 13.120919227600098,
      "learning_rate": 1.1254208781863271e-06,
      "loss": 3.1728,
      "step": 67400
    },
    {
      "epoch": 0.6856791178385416,
      "grad_norm": 12.971762657165527,
      "learning_rate": 1.1250868850175965e-06,
      "loss": 3.0323,
      "step": 67405
    },
    {
      "epoch": 0.68572998046875,
      "grad_norm": 10.945058822631836,
      "learning_rate": 1.1247529270256929e-06,
      "loss": 3.1961,
      "step": 67410
    },
    {
      "epoch": 0.6857808430989584,
      "grad_norm": 10.669898986816406,
      "learning_rate": 1.1244190042191597e-06,
      "loss": 3.4704,
      "step": 67415
    },
    {
      "epoch": 0.6858317057291666,
      "grad_norm": 9.132181167602539,
      "learning_rate": 1.1240851166065395e-06,
      "loss": 3.5586,
      "step": 67420
    },
    {
      "epoch": 0.685882568359375,
      "grad_norm": 18.08685874938965,
      "learning_rate": 1.1237512641963763e-06,
      "loss": 3.5384,
      "step": 67425
    },
    {
      "epoch": 0.6859334309895834,
      "grad_norm": 15.136528015136719,
      "learning_rate": 1.1234174469972099e-06,
      "loss": 3.0564,
      "step": 67430
    },
    {
      "epoch": 0.6859842936197916,
      "grad_norm": 11.03237247467041,
      "learning_rate": 1.1230836650175825e-06,
      "loss": 2.9931,
      "step": 67435
    },
    {
      "epoch": 0.68603515625,
      "grad_norm": 7.184826374053955,
      "learning_rate": 1.1227499182660326e-06,
      "loss": 3.3431,
      "step": 67440
    },
    {
      "epoch": 0.6860860188802084,
      "grad_norm": 17.306102752685547,
      "learning_rate": 1.1224162067511003e-06,
      "loss": 3.5036,
      "step": 67445
    },
    {
      "epoch": 0.6861368815104166,
      "grad_norm": 8.99876880645752,
      "learning_rate": 1.1220825304813228e-06,
      "loss": 3.4019,
      "step": 67450
    },
    {
      "epoch": 0.686187744140625,
      "grad_norm": 8.364696502685547,
      "learning_rate": 1.1217488894652362e-06,
      "loss": 3.2598,
      "step": 67455
    },
    {
      "epoch": 0.6862386067708334,
      "grad_norm": 12.502495765686035,
      "learning_rate": 1.121415283711378e-06,
      "loss": 3.6549,
      "step": 67460
    },
    {
      "epoch": 0.6862894694010416,
      "grad_norm": 11.477373123168945,
      "learning_rate": 1.1210817132282834e-06,
      "loss": 3.2676,
      "step": 67465
    },
    {
      "epoch": 0.68634033203125,
      "grad_norm": 7.801740646362305,
      "learning_rate": 1.1207481780244866e-06,
      "loss": 3.5728,
      "step": 67470
    },
    {
      "epoch": 0.6863911946614584,
      "grad_norm": 12.995709419250488,
      "learning_rate": 1.12041467810852e-06,
      "loss": 3.4484,
      "step": 67475
    },
    {
      "epoch": 0.6864420572916666,
      "grad_norm": 26.030847549438477,
      "learning_rate": 1.1200812134889163e-06,
      "loss": 3.6054,
      "step": 67480
    },
    {
      "epoch": 0.686492919921875,
      "grad_norm": 10.021547317504883,
      "learning_rate": 1.119747784174209e-06,
      "loss": 3.3571,
      "step": 67485
    },
    {
      "epoch": 0.6865437825520834,
      "grad_norm": 22.86213493347168,
      "learning_rate": 1.1194143901729274e-06,
      "loss": 3.0059,
      "step": 67490
    },
    {
      "epoch": 0.6865946451822916,
      "grad_norm": 11.521246910095215,
      "learning_rate": 1.1190810314936007e-06,
      "loss": 3.2185,
      "step": 67495
    },
    {
      "epoch": 0.6866455078125,
      "grad_norm": 10.095511436462402,
      "learning_rate": 1.118747708144759e-06,
      "loss": 3.5701,
      "step": 67500
    },
    {
      "epoch": 0.6866963704427084,
      "grad_norm": 14.873655319213867,
      "learning_rate": 1.1184144201349291e-06,
      "loss": 3.2541,
      "step": 67505
    },
    {
      "epoch": 0.6867472330729166,
      "grad_norm": 12.867834091186523,
      "learning_rate": 1.1180811674726396e-06,
      "loss": 3.4219,
      "step": 67510
    },
    {
      "epoch": 0.686798095703125,
      "grad_norm": 6.689752101898193,
      "learning_rate": 1.117747950166415e-06,
      "loss": 3.397,
      "step": 67515
    },
    {
      "epoch": 0.6868489583333334,
      "grad_norm": 21.152128219604492,
      "learning_rate": 1.1174147682247827e-06,
      "loss": 3.078,
      "step": 67520
    },
    {
      "epoch": 0.6868998209635416,
      "grad_norm": 10.34592342376709,
      "learning_rate": 1.1170816216562655e-06,
      "loss": 3.0048,
      "step": 67525
    },
    {
      "epoch": 0.68695068359375,
      "grad_norm": 15.649857521057129,
      "learning_rate": 1.1167485104693865e-06,
      "loss": 3.2808,
      "step": 67530
    },
    {
      "epoch": 0.6870015462239584,
      "grad_norm": 13.62820053100586,
      "learning_rate": 1.1164154346726688e-06,
      "loss": 3.4623,
      "step": 67535
    },
    {
      "epoch": 0.6870524088541666,
      "grad_norm": 11.384383201599121,
      "learning_rate": 1.1160823942746353e-06,
      "loss": 3.2355,
      "step": 67540
    },
    {
      "epoch": 0.687103271484375,
      "grad_norm": 15.764509201049805,
      "learning_rate": 1.1157493892838057e-06,
      "loss": 3.4105,
      "step": 67545
    },
    {
      "epoch": 0.6871541341145834,
      "grad_norm": 14.197961807250977,
      "learning_rate": 1.1154164197086988e-06,
      "loss": 3.0341,
      "step": 67550
    },
    {
      "epoch": 0.6872049967447916,
      "grad_norm": 9.72084903717041,
      "learning_rate": 1.1150834855578358e-06,
      "loss": 3.1286,
      "step": 67555
    },
    {
      "epoch": 0.687255859375,
      "grad_norm": 9.875367164611816,
      "learning_rate": 1.1147505868397324e-06,
      "loss": 3.2019,
      "step": 67560
    },
    {
      "epoch": 0.6873067220052084,
      "grad_norm": 13.429791450500488,
      "learning_rate": 1.1144177235629081e-06,
      "loss": 3.5066,
      "step": 67565
    },
    {
      "epoch": 0.6873575846354166,
      "grad_norm": 12.365313529968262,
      "learning_rate": 1.1140848957358768e-06,
      "loss": 3.6088,
      "step": 67570
    },
    {
      "epoch": 0.687408447265625,
      "grad_norm": 23.173267364501953,
      "learning_rate": 1.1137521033671558e-06,
      "loss": 3.3458,
      "step": 67575
    },
    {
      "epoch": 0.6874593098958334,
      "grad_norm": 10.073822975158691,
      "learning_rate": 1.1134193464652588e-06,
      "loss": 3.4049,
      "step": 67580
    },
    {
      "epoch": 0.6875101725260416,
      "grad_norm": 9.316292762756348,
      "learning_rate": 1.1130866250386982e-06,
      "loss": 3.2323,
      "step": 67585
    },
    {
      "epoch": 0.68756103515625,
      "grad_norm": 11.111005783081055,
      "learning_rate": 1.1127539390959877e-06,
      "loss": 3.4849,
      "step": 67590
    },
    {
      "epoch": 0.6876118977864584,
      "grad_norm": 12.63139533996582,
      "learning_rate": 1.1124212886456395e-06,
      "loss": 3.5015,
      "step": 67595
    },
    {
      "epoch": 0.6876627604166666,
      "grad_norm": 15.805822372436523,
      "learning_rate": 1.1120886736961639e-06,
      "loss": 3.8825,
      "step": 67600
    },
    {
      "epoch": 0.687713623046875,
      "grad_norm": 14.851016998291016,
      "learning_rate": 1.1117560942560706e-06,
      "loss": 3.474,
      "step": 67605
    },
    {
      "epoch": 0.6877644856770834,
      "grad_norm": 11.629377365112305,
      "learning_rate": 1.1114235503338677e-06,
      "loss": 3.4285,
      "step": 67610
    },
    {
      "epoch": 0.6878153483072916,
      "grad_norm": 8.709927558898926,
      "learning_rate": 1.111091041938064e-06,
      "loss": 3.2639,
      "step": 67615
    },
    {
      "epoch": 0.6878662109375,
      "grad_norm": 14.179695129394531,
      "learning_rate": 1.1107585690771676e-06,
      "loss": 3.3194,
      "step": 67620
    },
    {
      "epoch": 0.6879170735677084,
      "grad_norm": 13.241247177124023,
      "learning_rate": 1.1104261317596842e-06,
      "loss": 3.141,
      "step": 67625
    },
    {
      "epoch": 0.6879679361979166,
      "grad_norm": 12.496594429016113,
      "learning_rate": 1.1100937299941179e-06,
      "loss": 3.4404,
      "step": 67630
    },
    {
      "epoch": 0.688018798828125,
      "grad_norm": 18.70077896118164,
      "learning_rate": 1.109761363788974e-06,
      "loss": 3.5438,
      "step": 67635
    },
    {
      "epoch": 0.6880696614583334,
      "grad_norm": 15.268617630004883,
      "learning_rate": 1.1094290331527572e-06,
      "loss": 3.7203,
      "step": 67640
    },
    {
      "epoch": 0.6881205240885416,
      "grad_norm": 10.46624755859375,
      "learning_rate": 1.1090967380939676e-06,
      "loss": 3.9093,
      "step": 67645
    },
    {
      "epoch": 0.68817138671875,
      "grad_norm": 11.635927200317383,
      "learning_rate": 1.1087644786211095e-06,
      "loss": 3.1409,
      "step": 67650
    },
    {
      "epoch": 0.6882222493489584,
      "grad_norm": 9.18773365020752,
      "learning_rate": 1.1084322547426827e-06,
      "loss": 3.7527,
      "step": 67655
    },
    {
      "epoch": 0.6882731119791666,
      "grad_norm": 9.737253189086914,
      "learning_rate": 1.1081000664671857e-06,
      "loss": 2.983,
      "step": 67660
    },
    {
      "epoch": 0.688323974609375,
      "grad_norm": 15.006853103637695,
      "learning_rate": 1.1077679138031183e-06,
      "loss": 2.9761,
      "step": 67665
    },
    {
      "epoch": 0.6883748372395834,
      "grad_norm": 8.402088165283203,
      "learning_rate": 1.1074357967589799e-06,
      "loss": 3.4267,
      "step": 67670
    },
    {
      "epoch": 0.6884256998697916,
      "grad_norm": 17.684112548828125,
      "learning_rate": 1.1071037153432669e-06,
      "loss": 3.2649,
      "step": 67675
    },
    {
      "epoch": 0.6884765625,
      "grad_norm": 12.325630187988281,
      "learning_rate": 1.1067716695644748e-06,
      "loss": 3.302,
      "step": 67680
    },
    {
      "epoch": 0.6885274251302084,
      "grad_norm": 12.473956108093262,
      "learning_rate": 1.1064396594310986e-06,
      "loss": 3.3447,
      "step": 67685
    },
    {
      "epoch": 0.6885782877604166,
      "grad_norm": 12.692052841186523,
      "learning_rate": 1.1061076849516333e-06,
      "loss": 3.272,
      "step": 67690
    },
    {
      "epoch": 0.688629150390625,
      "grad_norm": 9.642125129699707,
      "learning_rate": 1.1057757461345734e-06,
      "loss": 3.0806,
      "step": 67695
    },
    {
      "epoch": 0.6886800130208334,
      "grad_norm": 8.635783195495605,
      "learning_rate": 1.1054438429884106e-06,
      "loss": 3.6573,
      "step": 67700
    },
    {
      "epoch": 0.6887308756510416,
      "grad_norm": 10.294721603393555,
      "learning_rate": 1.1051119755216354e-06,
      "loss": 3.2534,
      "step": 67705
    },
    {
      "epoch": 0.68878173828125,
      "grad_norm": 10.347164154052734,
      "learning_rate": 1.104780143742741e-06,
      "loss": 3.7229,
      "step": 67710
    },
    {
      "epoch": 0.6888326009114584,
      "grad_norm": 15.795777320861816,
      "learning_rate": 1.104448347660215e-06,
      "loss": 3.2436,
      "step": 67715
    },
    {
      "epoch": 0.6888834635416666,
      "grad_norm": 10.483985900878906,
      "learning_rate": 1.104116587282548e-06,
      "loss": 3.6812,
      "step": 67720
    },
    {
      "epoch": 0.688934326171875,
      "grad_norm": 11.902549743652344,
      "learning_rate": 1.1037848626182263e-06,
      "loss": 3.4797,
      "step": 67725
    },
    {
      "epoch": 0.6889851888020834,
      "grad_norm": 12.284526824951172,
      "learning_rate": 1.1034531736757387e-06,
      "loss": 3.4157,
      "step": 67730
    },
    {
      "epoch": 0.6890360514322916,
      "grad_norm": 15.990017890930176,
      "learning_rate": 1.103121520463571e-06,
      "loss": 2.9745,
      "step": 67735
    },
    {
      "epoch": 0.6890869140625,
      "grad_norm": 9.489580154418945,
      "learning_rate": 1.1027899029902068e-06,
      "loss": 3.2829,
      "step": 67740
    },
    {
      "epoch": 0.6891377766927084,
      "grad_norm": 7.084834575653076,
      "learning_rate": 1.1024583212641321e-06,
      "loss": 3.1939,
      "step": 67745
    },
    {
      "epoch": 0.6891886393229166,
      "grad_norm": 9.82347297668457,
      "learning_rate": 1.1021267752938305e-06,
      "loss": 3.1355,
      "step": 67750
    },
    {
      "epoch": 0.689239501953125,
      "grad_norm": 7.685935974121094,
      "learning_rate": 1.1017952650877844e-06,
      "loss": 3.4428,
      "step": 67755
    },
    {
      "epoch": 0.6892903645833334,
      "grad_norm": 16.429901123046875,
      "learning_rate": 1.1014637906544739e-06,
      "loss": 3.5422,
      "step": 67760
    },
    {
      "epoch": 0.6893412272135416,
      "grad_norm": 7.084587574005127,
      "learning_rate": 1.101132352002382e-06,
      "loss": 3.3334,
      "step": 67765
    },
    {
      "epoch": 0.68939208984375,
      "grad_norm": 14.835680961608887,
      "learning_rate": 1.1008009491399862e-06,
      "loss": 3.6113,
      "step": 67770
    },
    {
      "epoch": 0.6894429524739584,
      "grad_norm": 12.166670799255371,
      "learning_rate": 1.1004695820757674e-06,
      "loss": 3.2544,
      "step": 67775
    },
    {
      "epoch": 0.6894938151041666,
      "grad_norm": 10.962254524230957,
      "learning_rate": 1.1001382508182018e-06,
      "loss": 3.7049,
      "step": 67780
    },
    {
      "epoch": 0.689544677734375,
      "grad_norm": 15.7714204788208,
      "learning_rate": 1.0998069553757681e-06,
      "loss": 3.4172,
      "step": 67785
    },
    {
      "epoch": 0.6895955403645834,
      "grad_norm": 12.194293022155762,
      "learning_rate": 1.0994756957569405e-06,
      "loss": 3.6877,
      "step": 67790
    },
    {
      "epoch": 0.6896464029947916,
      "grad_norm": 7.582254409790039,
      "learning_rate": 1.0991444719701967e-06,
      "loss": 3.1827,
      "step": 67795
    },
    {
      "epoch": 0.689697265625,
      "grad_norm": 8.773179054260254,
      "learning_rate": 1.098813284024008e-06,
      "loss": 3.3778,
      "step": 67800
    },
    {
      "epoch": 0.6897481282552084,
      "grad_norm": 11.618633270263672,
      "learning_rate": 1.0984821319268504e-06,
      "loss": 3.5551,
      "step": 67805
    },
    {
      "epoch": 0.6897989908854166,
      "grad_norm": 16.125980377197266,
      "learning_rate": 1.0981510156871955e-06,
      "loss": 3.01,
      "step": 67810
    },
    {
      "epoch": 0.689849853515625,
      "grad_norm": 9.524605751037598,
      "learning_rate": 1.0978199353135135e-06,
      "loss": 3.9478,
      "step": 67815
    },
    {
      "epoch": 0.6899007161458334,
      "grad_norm": 15.00719165802002,
      "learning_rate": 1.097488890814276e-06,
      "loss": 3.5804,
      "step": 67820
    },
    {
      "epoch": 0.6899515787760416,
      "grad_norm": 10.659303665161133,
      "learning_rate": 1.0971578821979539e-06,
      "loss": 3.23,
      "step": 67825
    },
    {
      "epoch": 0.69000244140625,
      "grad_norm": 14.772455215454102,
      "learning_rate": 1.0968269094730147e-06,
      "loss": 3.5165,
      "step": 67830
    },
    {
      "epoch": 0.6900533040364584,
      "grad_norm": 10.8951416015625,
      "learning_rate": 1.0964959726479266e-06,
      "loss": 3.4207,
      "step": 67835
    },
    {
      "epoch": 0.6901041666666666,
      "grad_norm": 12.686936378479004,
      "learning_rate": 1.0961650717311553e-06,
      "loss": 3.3497,
      "step": 67840
    },
    {
      "epoch": 0.690155029296875,
      "grad_norm": 14.472989082336426,
      "learning_rate": 1.095834206731168e-06,
      "loss": 3.2775,
      "step": 67845
    },
    {
      "epoch": 0.6902058919270834,
      "grad_norm": 8.697223663330078,
      "learning_rate": 1.0955033776564306e-06,
      "loss": 3.8941,
      "step": 67850
    },
    {
      "epoch": 0.6902567545572916,
      "grad_norm": 11.928094863891602,
      "learning_rate": 1.0951725845154053e-06,
      "loss": 3.0851,
      "step": 67855
    },
    {
      "epoch": 0.6903076171875,
      "grad_norm": 8.547250747680664,
      "learning_rate": 1.094841827316558e-06,
      "loss": 3.2359,
      "step": 67860
    },
    {
      "epoch": 0.6903584798177084,
      "grad_norm": 8.260056495666504,
      "learning_rate": 1.094511106068349e-06,
      "loss": 3.1335,
      "step": 67865
    },
    {
      "epoch": 0.6904093424479166,
      "grad_norm": 14.270990371704102,
      "learning_rate": 1.0941804207792391e-06,
      "loss": 3.4316,
      "step": 67870
    },
    {
      "epoch": 0.690460205078125,
      "grad_norm": 14.395874977111816,
      "learning_rate": 1.09384977145769e-06,
      "loss": 3.2012,
      "step": 67875
    },
    {
      "epoch": 0.6905110677083334,
      "grad_norm": 12.590879440307617,
      "learning_rate": 1.0935191581121624e-06,
      "loss": 3.4119,
      "step": 67880
    },
    {
      "epoch": 0.6905619303385416,
      "grad_norm": 17.32014274597168,
      "learning_rate": 1.0931885807511137e-06,
      "loss": 3.5376,
      "step": 67885
    },
    {
      "epoch": 0.69061279296875,
      "grad_norm": 14.185200691223145,
      "learning_rate": 1.0928580393830015e-06,
      "loss": 3.2657,
      "step": 67890
    },
    {
      "epoch": 0.6906636555989584,
      "grad_norm": 10.420732498168945,
      "learning_rate": 1.092527534016282e-06,
      "loss": 3.3757,
      "step": 67895
    },
    {
      "epoch": 0.6907145182291666,
      "grad_norm": 11.136979103088379,
      "learning_rate": 1.092197064659412e-06,
      "loss": 3.5773,
      "step": 67900
    },
    {
      "epoch": 0.690765380859375,
      "grad_norm": 12.3397216796875,
      "learning_rate": 1.0918666313208472e-06,
      "loss": 3.8071,
      "step": 67905
    },
    {
      "epoch": 0.6908162434895834,
      "grad_norm": 10.15494441986084,
      "learning_rate": 1.091536234009041e-06,
      "loss": 3.1771,
      "step": 67910
    },
    {
      "epoch": 0.6908671061197916,
      "grad_norm": 16.179298400878906,
      "learning_rate": 1.0912058727324456e-06,
      "loss": 3.3966,
      "step": 67915
    },
    {
      "epoch": 0.69091796875,
      "grad_norm": 12.150471687316895,
      "learning_rate": 1.0908755474995147e-06,
      "loss": 3.1623,
      "step": 67920
    },
    {
      "epoch": 0.6909688313802084,
      "grad_norm": 12.015385627746582,
      "learning_rate": 1.090545258318698e-06,
      "loss": 3.1398,
      "step": 67925
    },
    {
      "epoch": 0.6910196940104166,
      "grad_norm": 10.088885307312012,
      "learning_rate": 1.090215005198448e-06,
      "loss": 3.345,
      "step": 67930
    },
    {
      "epoch": 0.691070556640625,
      "grad_norm": 14.526505470275879,
      "learning_rate": 1.0898847881472118e-06,
      "loss": 3.4345,
      "step": 67935
    },
    {
      "epoch": 0.6911214192708334,
      "grad_norm": 13.822447776794434,
      "learning_rate": 1.08955460717344e-06,
      "loss": 3.8992,
      "step": 67940
    },
    {
      "epoch": 0.6911722819010416,
      "grad_norm": 16.220075607299805,
      "learning_rate": 1.0892244622855785e-06,
      "loss": 3.3151,
      "step": 67945
    },
    {
      "epoch": 0.69122314453125,
      "grad_norm": 8.144336700439453,
      "learning_rate": 1.0888943534920745e-06,
      "loss": 3.2789,
      "step": 67950
    },
    {
      "epoch": 0.6912740071614584,
      "grad_norm": 8.377969741821289,
      "learning_rate": 1.0885642808013752e-06,
      "loss": 3.1394,
      "step": 67955
    },
    {
      "epoch": 0.6913248697916666,
      "grad_norm": 16.388376235961914,
      "learning_rate": 1.088234244221924e-06,
      "loss": 3.2968,
      "step": 67960
    },
    {
      "epoch": 0.691375732421875,
      "grad_norm": 13.155487060546875,
      "learning_rate": 1.0879042437621651e-06,
      "loss": 3.5906,
      "step": 67965
    },
    {
      "epoch": 0.6914265950520834,
      "grad_norm": 8.747903823852539,
      "learning_rate": 1.0875742794305404e-06,
      "loss": 3.177,
      "step": 67970
    },
    {
      "epoch": 0.6914774576822916,
      "grad_norm": 14.255596160888672,
      "learning_rate": 1.087244351235493e-06,
      "loss": 3.1131,
      "step": 67975
    },
    {
      "epoch": 0.6915283203125,
      "grad_norm": 7.685469150543213,
      "learning_rate": 1.086914459185465e-06,
      "loss": 3.2055,
      "step": 67980
    },
    {
      "epoch": 0.6915791829427084,
      "grad_norm": 12.629805564880371,
      "learning_rate": 1.0865846032888957e-06,
      "loss": 3.4163,
      "step": 67985
    },
    {
      "epoch": 0.6916300455729166,
      "grad_norm": 14.303412437438965,
      "learning_rate": 1.0862547835542235e-06,
      "loss": 2.9933,
      "step": 67990
    },
    {
      "epoch": 0.691680908203125,
      "grad_norm": 8.761443138122559,
      "learning_rate": 1.0859249999898882e-06,
      "loss": 3.3332,
      "step": 67995
    },
    {
      "epoch": 0.6917317708333334,
      "grad_norm": 9.855063438415527,
      "learning_rate": 1.0855952526043256e-06,
      "loss": 3.4041,
      "step": 68000
    },
    {
      "epoch": 0.6917826334635416,
      "grad_norm": 14.298595428466797,
      "learning_rate": 1.0852655414059745e-06,
      "loss": 3.5639,
      "step": 68005
    },
    {
      "epoch": 0.69183349609375,
      "grad_norm": 13.105066299438477,
      "learning_rate": 1.084935866403268e-06,
      "loss": 3.2815,
      "step": 68010
    },
    {
      "epoch": 0.6918843587239584,
      "grad_norm": 14.746922492980957,
      "learning_rate": 1.0846062276046428e-06,
      "loss": 3.5784,
      "step": 68015
    },
    {
      "epoch": 0.6919352213541666,
      "grad_norm": 21.065969467163086,
      "learning_rate": 1.0842766250185322e-06,
      "loss": 3.4011,
      "step": 68020
    },
    {
      "epoch": 0.691986083984375,
      "grad_norm": 10.775930404663086,
      "learning_rate": 1.0839470586533673e-06,
      "loss": 3.44,
      "step": 68025
    },
    {
      "epoch": 0.6920369466145834,
      "grad_norm": 13.672934532165527,
      "learning_rate": 1.0836175285175815e-06,
      "loss": 3.3936,
      "step": 68030
    },
    {
      "epoch": 0.6920878092447916,
      "grad_norm": 14.887701988220215,
      "learning_rate": 1.0832880346196062e-06,
      "loss": 3.3906,
      "step": 68035
    },
    {
      "epoch": 0.692138671875,
      "grad_norm": 10.036215782165527,
      "learning_rate": 1.0829585769678708e-06,
      "loss": 3.3281,
      "step": 68040
    },
    {
      "epoch": 0.6921895345052084,
      "grad_norm": 13.498652458190918,
      "learning_rate": 1.0826291555708034e-06,
      "loss": 3.4331,
      "step": 68045
    },
    {
      "epoch": 0.6922403971354166,
      "grad_norm": 10.369086265563965,
      "learning_rate": 1.0822997704368344e-06,
      "loss": 3.4975,
      "step": 68050
    },
    {
      "epoch": 0.692291259765625,
      "grad_norm": 11.20858097076416,
      "learning_rate": 1.0819704215743884e-06,
      "loss": 3.0873,
      "step": 68055
    },
    {
      "epoch": 0.6923421223958334,
      "grad_norm": 14.309041976928711,
      "learning_rate": 1.081641108991894e-06,
      "loss": 3.2642,
      "step": 68060
    },
    {
      "epoch": 0.6923929850260416,
      "grad_norm": 14.366168022155762,
      "learning_rate": 1.0813118326977748e-06,
      "loss": 3.1822,
      "step": 68065
    },
    {
      "epoch": 0.69244384765625,
      "grad_norm": 11.820574760437012,
      "learning_rate": 1.080982592700457e-06,
      "loss": 3.0526,
      "step": 68070
    },
    {
      "epoch": 0.6924947102864584,
      "grad_norm": 9.015646934509277,
      "learning_rate": 1.080653389008363e-06,
      "loss": 3.0813,
      "step": 68075
    },
    {
      "epoch": 0.6925455729166666,
      "grad_norm": 13.694621086120605,
      "learning_rate": 1.080324221629915e-06,
      "loss": 3.3853,
      "step": 68080
    },
    {
      "epoch": 0.692596435546875,
      "grad_norm": 8.71569538116455,
      "learning_rate": 1.0799950905735353e-06,
      "loss": 3.5342,
      "step": 68085
    },
    {
      "epoch": 0.6926472981770834,
      "grad_norm": 7.991396903991699,
      "learning_rate": 1.0796659958476456e-06,
      "loss": 3.111,
      "step": 68090
    },
    {
      "epoch": 0.6926981608072916,
      "grad_norm": 13.465011596679688,
      "learning_rate": 1.0793369374606646e-06,
      "loss": 3.4025,
      "step": 68095
    },
    {
      "epoch": 0.6927490234375,
      "grad_norm": 9.13414192199707,
      "learning_rate": 1.0790079154210111e-06,
      "loss": 3.3709,
      "step": 68100
    },
    {
      "epoch": 0.6927998860677084,
      "grad_norm": 13.184441566467285,
      "learning_rate": 1.0786789297371019e-06,
      "loss": 3.3656,
      "step": 68105
    },
    {
      "epoch": 0.6928507486979166,
      "grad_norm": 11.598027229309082,
      "learning_rate": 1.078349980417357e-06,
      "loss": 3.2299,
      "step": 68110
    },
    {
      "epoch": 0.692901611328125,
      "grad_norm": 10.605900764465332,
      "learning_rate": 1.0780210674701907e-06,
      "loss": 3.7082,
      "step": 68115
    },
    {
      "epoch": 0.6929524739583334,
      "grad_norm": 15.235445976257324,
      "learning_rate": 1.0776921909040187e-06,
      "loss": 3.0915,
      "step": 68120
    },
    {
      "epoch": 0.6930033365885416,
      "grad_norm": 9.415688514709473,
      "learning_rate": 1.0773633507272537e-06,
      "loss": 3.338,
      "step": 68125
    },
    {
      "epoch": 0.69305419921875,
      "grad_norm": 15.426976203918457,
      "learning_rate": 1.0770345469483103e-06,
      "loss": 3.2185,
      "step": 68130
    },
    {
      "epoch": 0.6931050618489584,
      "grad_norm": 13.035916328430176,
      "learning_rate": 1.0767057795756018e-06,
      "loss": 3.4561,
      "step": 68135
    },
    {
      "epoch": 0.6931559244791666,
      "grad_norm": 10.376809120178223,
      "learning_rate": 1.0763770486175374e-06,
      "loss": 3.3602,
      "step": 68140
    },
    {
      "epoch": 0.693206787109375,
      "grad_norm": 16.455488204956055,
      "learning_rate": 1.0760483540825294e-06,
      "loss": 3.6002,
      "step": 68145
    },
    {
      "epoch": 0.6932576497395834,
      "grad_norm": 15.267796516418457,
      "learning_rate": 1.0757196959789872e-06,
      "loss": 3.2368,
      "step": 68150
    },
    {
      "epoch": 0.6933085123697916,
      "grad_norm": 14.585723876953125,
      "learning_rate": 1.075391074315318e-06,
      "loss": 3.7951,
      "step": 68155
    },
    {
      "epoch": 0.693359375,
      "grad_norm": 11.693926811218262,
      "learning_rate": 1.0750624890999307e-06,
      "loss": 3.6183,
      "step": 68160
    },
    {
      "epoch": 0.6934102376302084,
      "grad_norm": 8.625460624694824,
      "learning_rate": 1.0747339403412325e-06,
      "loss": 3.4812,
      "step": 68165
    },
    {
      "epoch": 0.6934611002604166,
      "grad_norm": 12.401426315307617,
      "learning_rate": 1.0744054280476285e-06,
      "loss": 3.2696,
      "step": 68170
    },
    {
      "epoch": 0.693511962890625,
      "grad_norm": 10.516034126281738,
      "learning_rate": 1.0740769522275237e-06,
      "loss": 3.0283,
      "step": 68175
    },
    {
      "epoch": 0.6935628255208334,
      "grad_norm": 12.46440315246582,
      "learning_rate": 1.0737485128893215e-06,
      "loss": 3.2971,
      "step": 68180
    },
    {
      "epoch": 0.6936136881510416,
      "grad_norm": 9.502420425415039,
      "learning_rate": 1.0734201100414255e-06,
      "loss": 3.39,
      "step": 68185
    },
    {
      "epoch": 0.69366455078125,
      "grad_norm": 13.025237083435059,
      "learning_rate": 1.0730917436922387e-06,
      "loss": 3.1769,
      "step": 68190
    },
    {
      "epoch": 0.6937154134114584,
      "grad_norm": 12.869500160217285,
      "learning_rate": 1.0727634138501613e-06,
      "loss": 3.6214,
      "step": 68195
    },
    {
      "epoch": 0.6937662760416666,
      "grad_norm": 14.211162567138672,
      "learning_rate": 1.0724351205235928e-06,
      "loss": 3.3619,
      "step": 68200
    },
    {
      "epoch": 0.693817138671875,
      "grad_norm": 7.965907096862793,
      "learning_rate": 1.0721068637209342e-06,
      "loss": 3.0924,
      "step": 68205
    },
    {
      "epoch": 0.6938680013020834,
      "grad_norm": 14.146011352539062,
      "learning_rate": 1.0717786434505823e-06,
      "loss": 3.3728,
      "step": 68210
    },
    {
      "epoch": 0.6939188639322916,
      "grad_norm": 9.873130798339844,
      "learning_rate": 1.0714504597209362e-06,
      "loss": 3.3597,
      "step": 68215
    },
    {
      "epoch": 0.6939697265625,
      "grad_norm": 9.39063835144043,
      "learning_rate": 1.0711223125403906e-06,
      "loss": 3.2263,
      "step": 68220
    },
    {
      "epoch": 0.6940205891927084,
      "grad_norm": 7.686598777770996,
      "learning_rate": 1.0707942019173426e-06,
      "loss": 3.4356,
      "step": 68225
    },
    {
      "epoch": 0.6940714518229166,
      "grad_norm": 13.04306411743164,
      "learning_rate": 1.0704661278601863e-06,
      "loss": 3.3608,
      "step": 68230
    },
    {
      "epoch": 0.694122314453125,
      "grad_norm": 8.329973220825195,
      "learning_rate": 1.0701380903773145e-06,
      "loss": 3.3499,
      "step": 68235
    },
    {
      "epoch": 0.6941731770833334,
      "grad_norm": 11.92003059387207,
      "learning_rate": 1.0698100894771207e-06,
      "loss": 3.3101,
      "step": 68240
    },
    {
      "epoch": 0.6942240397135416,
      "grad_norm": 12.852375030517578,
      "learning_rate": 1.0694821251679977e-06,
      "loss": 3.0116,
      "step": 68245
    },
    {
      "epoch": 0.69427490234375,
      "grad_norm": 11.638413429260254,
      "learning_rate": 1.0691541974583355e-06,
      "loss": 3.2034,
      "step": 68250
    },
    {
      "epoch": 0.6943257649739584,
      "grad_norm": 10.772770881652832,
      "learning_rate": 1.068826306356523e-06,
      "loss": 3.1456,
      "step": 68255
    },
    {
      "epoch": 0.6943766276041666,
      "grad_norm": 15.013161659240723,
      "learning_rate": 1.0684984518709502e-06,
      "loss": 3.8208,
      "step": 68260
    },
    {
      "epoch": 0.694427490234375,
      "grad_norm": 10.826665878295898,
      "learning_rate": 1.0681706340100062e-06,
      "loss": 3.2775,
      "step": 68265
    },
    {
      "epoch": 0.6944783528645834,
      "grad_norm": 12.998494148254395,
      "learning_rate": 1.067842852782077e-06,
      "loss": 3.4062,
      "step": 68270
    },
    {
      "epoch": 0.6945292154947916,
      "grad_norm": 12.212747573852539,
      "learning_rate": 1.0675151081955483e-06,
      "loss": 3.2866,
      "step": 68275
    },
    {
      "epoch": 0.694580078125,
      "grad_norm": 13.134560585021973,
      "learning_rate": 1.0671874002588066e-06,
      "loss": 3.5512,
      "step": 68280
    },
    {
      "epoch": 0.6946309407552084,
      "grad_norm": 7.740173816680908,
      "learning_rate": 1.066859728980235e-06,
      "loss": 3.2582,
      "step": 68285
    },
    {
      "epoch": 0.6946818033854166,
      "grad_norm": 12.876514434814453,
      "learning_rate": 1.0665320943682184e-06,
      "loss": 3.2665,
      "step": 68290
    },
    {
      "epoch": 0.694732666015625,
      "grad_norm": 14.70106315612793,
      "learning_rate": 1.0662044964311375e-06,
      "loss": 3.4344,
      "step": 68295
    },
    {
      "epoch": 0.6947835286458334,
      "grad_norm": 8.250565528869629,
      "learning_rate": 1.0658769351773754e-06,
      "loss": 3.3056,
      "step": 68300
    },
    {
      "epoch": 0.6948343912760416,
      "grad_norm": 9.84837818145752,
      "learning_rate": 1.0655494106153121e-06,
      "loss": 3.7646,
      "step": 68305
    },
    {
      "epoch": 0.69488525390625,
      "grad_norm": 10.122499465942383,
      "learning_rate": 1.0652219227533264e-06,
      "loss": 3.2689,
      "step": 68310
    },
    {
      "epoch": 0.6949361165364584,
      "grad_norm": 15.315901756286621,
      "learning_rate": 1.0648944715997977e-06,
      "loss": 2.8652,
      "step": 68315
    },
    {
      "epoch": 0.6949869791666666,
      "grad_norm": 7.8511061668396,
      "learning_rate": 1.0645670571631044e-06,
      "loss": 3.0047,
      "step": 68320
    },
    {
      "epoch": 0.695037841796875,
      "grad_norm": 14.779093742370605,
      "learning_rate": 1.0642396794516233e-06,
      "loss": 3.5727,
      "step": 68325
    },
    {
      "epoch": 0.6950887044270834,
      "grad_norm": 9.470014572143555,
      "learning_rate": 1.063912338473729e-06,
      "loss": 3.0945,
      "step": 68330
    },
    {
      "epoch": 0.6951395670572916,
      "grad_norm": 17.567630767822266,
      "learning_rate": 1.0635850342377969e-06,
      "loss": 3.3204,
      "step": 68335
    },
    {
      "epoch": 0.6951904296875,
      "grad_norm": 12.101985931396484,
      "learning_rate": 1.0632577667522006e-06,
      "loss": 3.0812,
      "step": 68340
    },
    {
      "epoch": 0.6952412923177084,
      "grad_norm": 18.240774154663086,
      "learning_rate": 1.062930536025315e-06,
      "loss": 3.9153,
      "step": 68345
    },
    {
      "epoch": 0.6952921549479166,
      "grad_norm": 7.594537258148193,
      "learning_rate": 1.0626033420655102e-06,
      "loss": 3.3416,
      "step": 68350
    },
    {
      "epoch": 0.695343017578125,
      "grad_norm": 14.703880310058594,
      "learning_rate": 1.0622761848811588e-06,
      "loss": 3.5369,
      "step": 68355
    },
    {
      "epoch": 0.6953938802083334,
      "grad_norm": 10.368515968322754,
      "learning_rate": 1.0619490644806307e-06,
      "loss": 3.2543,
      "step": 68360
    },
    {
      "epoch": 0.6954447428385416,
      "grad_norm": 14.770052909851074,
      "learning_rate": 1.0616219808722938e-06,
      "loss": 3.1537,
      "step": 68365
    },
    {
      "epoch": 0.69549560546875,
      "grad_norm": 13.274304389953613,
      "learning_rate": 1.0612949340645178e-06,
      "loss": 3.3308,
      "step": 68370
    },
    {
      "epoch": 0.6955464680989584,
      "grad_norm": 12.72620677947998,
      "learning_rate": 1.0609679240656707e-06,
      "loss": 3.2832,
      "step": 68375
    },
    {
      "epoch": 0.6955973307291666,
      "grad_norm": 9.127725601196289,
      "learning_rate": 1.0606409508841181e-06,
      "loss": 3.7104,
      "step": 68380
    },
    {
      "epoch": 0.695648193359375,
      "grad_norm": 9.640445709228516,
      "learning_rate": 1.0603140145282259e-06,
      "loss": 3.2348,
      "step": 68385
    },
    {
      "epoch": 0.6956990559895834,
      "grad_norm": 10.805159568786621,
      "learning_rate": 1.059987115006357e-06,
      "loss": 3.5362,
      "step": 68390
    },
    {
      "epoch": 0.6957499186197916,
      "grad_norm": 11.10505485534668,
      "learning_rate": 1.059660252326877e-06,
      "loss": 3.2466,
      "step": 68395
    },
    {
      "epoch": 0.69580078125,
      "grad_norm": 12.379951477050781,
      "learning_rate": 1.0593334264981486e-06,
      "loss": 3.3931,
      "step": 68400
    },
    {
      "epoch": 0.6958516438802084,
      "grad_norm": 7.506567001342773,
      "learning_rate": 1.0590066375285329e-06,
      "loss": 3.5404,
      "step": 68405
    },
    {
      "epoch": 0.6959025065104166,
      "grad_norm": 11.790244102478027,
      "learning_rate": 1.0586798854263901e-06,
      "loss": 3.3525,
      "step": 68410
    },
    {
      "epoch": 0.695953369140625,
      "grad_norm": 16.516735076904297,
      "learning_rate": 1.0583531702000809e-06,
      "loss": 3.5066,
      "step": 68415
    },
    {
      "epoch": 0.6960042317708334,
      "grad_norm": 10.455042839050293,
      "learning_rate": 1.058026491857965e-06,
      "loss": 3.1206,
      "step": 68420
    },
    {
      "epoch": 0.6960550944010416,
      "grad_norm": 12.830158233642578,
      "learning_rate": 1.0576998504083996e-06,
      "loss": 4.1053,
      "step": 68425
    },
    {
      "epoch": 0.69610595703125,
      "grad_norm": 13.322319030761719,
      "learning_rate": 1.0573732458597404e-06,
      "loss": 3.073,
      "step": 68430
    },
    {
      "epoch": 0.6961568196614584,
      "grad_norm": 13.685236930847168,
      "learning_rate": 1.0570466782203462e-06,
      "loss": 3.3646,
      "step": 68435
    },
    {
      "epoch": 0.6962076822916666,
      "grad_norm": 16.166658401489258,
      "learning_rate": 1.0567201474985694e-06,
      "loss": 3.5602,
      "step": 68440
    },
    {
      "epoch": 0.696258544921875,
      "grad_norm": 11.911019325256348,
      "learning_rate": 1.0563936537027656e-06,
      "loss": 3.1587,
      "step": 68445
    },
    {
      "epoch": 0.6963094075520834,
      "grad_norm": 10.115235328674316,
      "learning_rate": 1.056067196841289e-06,
      "loss": 3.7451,
      "step": 68450
    },
    {
      "epoch": 0.6963602701822916,
      "grad_norm": 11.368029594421387,
      "learning_rate": 1.0557407769224909e-06,
      "loss": 3.4379,
      "step": 68455
    },
    {
      "epoch": 0.6964111328125,
      "grad_norm": 13.152690887451172,
      "learning_rate": 1.0554143939547227e-06,
      "loss": 3.2729,
      "step": 68460
    },
    {
      "epoch": 0.6964619954427084,
      "grad_norm": 10.852267265319824,
      "learning_rate": 1.0550880479463338e-06,
      "loss": 2.9517,
      "step": 68465
    },
    {
      "epoch": 0.6965128580729166,
      "grad_norm": 12.223347663879395,
      "learning_rate": 1.0547617389056747e-06,
      "loss": 3.5249,
      "step": 68470
    },
    {
      "epoch": 0.696563720703125,
      "grad_norm": 11.630764961242676,
      "learning_rate": 1.054435466841095e-06,
      "loss": 3.2771,
      "step": 68475
    },
    {
      "epoch": 0.6966145833333334,
      "grad_norm": 16.688596725463867,
      "learning_rate": 1.054109231760941e-06,
      "loss": 3.6573,
      "step": 68480
    },
    {
      "epoch": 0.6966654459635416,
      "grad_norm": 12.443121910095215,
      "learning_rate": 1.053783033673559e-06,
      "loss": 3.397,
      "step": 68485
    },
    {
      "epoch": 0.69671630859375,
      "grad_norm": 10.7487154006958,
      "learning_rate": 1.0534568725872962e-06,
      "loss": 3.4047,
      "step": 68490
    },
    {
      "epoch": 0.6967671712239584,
      "grad_norm": 11.965703010559082,
      "learning_rate": 1.0531307485104956e-06,
      "loss": 2.9612,
      "step": 68495
    },
    {
      "epoch": 0.6968180338541666,
      "grad_norm": 10.180152893066406,
      "learning_rate": 1.0528046614515024e-06,
      "loss": 3.8558,
      "step": 68500
    },
    {
      "epoch": 0.696868896484375,
      "grad_norm": 9.827269554138184,
      "learning_rate": 1.0524786114186582e-06,
      "loss": 3.1974,
      "step": 68505
    },
    {
      "epoch": 0.6969197591145834,
      "grad_norm": 14.167684555053711,
      "learning_rate": 1.0521525984203065e-06,
      "loss": 3.4702,
      "step": 68510
    },
    {
      "epoch": 0.6969706217447916,
      "grad_norm": 8.219744682312012,
      "learning_rate": 1.0518266224647874e-06,
      "loss": 3.3173,
      "step": 68515
    },
    {
      "epoch": 0.697021484375,
      "grad_norm": 82.138916015625,
      "learning_rate": 1.05150068356044e-06,
      "loss": 3.7151,
      "step": 68520
    },
    {
      "epoch": 0.6970723470052084,
      "grad_norm": 9.34282398223877,
      "learning_rate": 1.0511747817156042e-06,
      "loss": 3.5471,
      "step": 68525
    },
    {
      "epoch": 0.6971232096354166,
      "grad_norm": 9.755684852600098,
      "learning_rate": 1.0508489169386194e-06,
      "loss": 3.4135,
      "step": 68530
    },
    {
      "epoch": 0.697174072265625,
      "grad_norm": 12.70797061920166,
      "learning_rate": 1.0505230892378212e-06,
      "loss": 3.2697,
      "step": 68535
    },
    {
      "epoch": 0.6972249348958334,
      "grad_norm": 14.959371566772461,
      "learning_rate": 1.0501972986215456e-06,
      "loss": 3.1872,
      "step": 68540
    },
    {
      "epoch": 0.6972757975260416,
      "grad_norm": 15.707871437072754,
      "learning_rate": 1.0498715450981293e-06,
      "loss": 3.5537,
      "step": 68545
    },
    {
      "epoch": 0.69732666015625,
      "grad_norm": 13.715645790100098,
      "learning_rate": 1.0495458286759046e-06,
      "loss": 3.2138,
      "step": 68550
    },
    {
      "epoch": 0.6973775227864584,
      "grad_norm": 16.517990112304688,
      "learning_rate": 1.0492201493632073e-06,
      "loss": 3.4732,
      "step": 68555
    },
    {
      "epoch": 0.6974283854166666,
      "grad_norm": 7.940740585327148,
      "learning_rate": 1.0488945071683678e-06,
      "loss": 3.1687,
      "step": 68560
    },
    {
      "epoch": 0.697479248046875,
      "grad_norm": 9.776067733764648,
      "learning_rate": 1.0485689020997191e-06,
      "loss": 3.1857,
      "step": 68565
    },
    {
      "epoch": 0.6975301106770834,
      "grad_norm": 11.337080001831055,
      "learning_rate": 1.0482433341655901e-06,
      "loss": 3.1281,
      "step": 68570
    },
    {
      "epoch": 0.6975809733072916,
      "grad_norm": 13.244839668273926,
      "learning_rate": 1.0479178033743124e-06,
      "loss": 3.1413,
      "step": 68575
    },
    {
      "epoch": 0.6976318359375,
      "grad_norm": 9.634005546569824,
      "learning_rate": 1.0475923097342123e-06,
      "loss": 3.4767,
      "step": 68580
    },
    {
      "epoch": 0.6976826985677084,
      "grad_norm": 14.75212287902832,
      "learning_rate": 1.0472668532536197e-06,
      "loss": 3.2871,
      "step": 68585
    },
    {
      "epoch": 0.6977335611979166,
      "grad_norm": 7.295383453369141,
      "learning_rate": 1.0469414339408604e-06,
      "loss": 3.4713,
      "step": 68590
    },
    {
      "epoch": 0.697784423828125,
      "grad_norm": 13.375802993774414,
      "learning_rate": 1.0466160518042592e-06,
      "loss": 3.0931,
      "step": 68595
    },
    {
      "epoch": 0.6978352864583334,
      "grad_norm": 8.959383010864258,
      "learning_rate": 1.0462907068521416e-06,
      "loss": 3.2586,
      "step": 68600
    },
    {
      "epoch": 0.6978861490885416,
      "grad_norm": 13.755732536315918,
      "learning_rate": 1.0459653990928329e-06,
      "loss": 3.4389,
      "step": 68605
    },
    {
      "epoch": 0.69793701171875,
      "grad_norm": 15.684408187866211,
      "learning_rate": 1.045640128534655e-06,
      "loss": 3.4938,
      "step": 68610
    },
    {
      "epoch": 0.6979878743489584,
      "grad_norm": 14.712481498718262,
      "learning_rate": 1.0453148951859291e-06,
      "loss": 3.1968,
      "step": 68615
    },
    {
      "epoch": 0.6980387369791666,
      "grad_norm": 13.377435684204102,
      "learning_rate": 1.0449896990549762e-06,
      "loss": 3.0805,
      "step": 68620
    },
    {
      "epoch": 0.698089599609375,
      "grad_norm": 10.302496910095215,
      "learning_rate": 1.0446645401501172e-06,
      "loss": 3.2776,
      "step": 68625
    },
    {
      "epoch": 0.6981404622395834,
      "grad_norm": 13.840866088867188,
      "learning_rate": 1.0443394184796715e-06,
      "loss": 3.2508,
      "step": 68630
    },
    {
      "epoch": 0.6981913248697916,
      "grad_norm": 12.550299644470215,
      "learning_rate": 1.044014334051956e-06,
      "loss": 3.4385,
      "step": 68635
    },
    {
      "epoch": 0.6982421875,
      "grad_norm": 15.650467872619629,
      "learning_rate": 1.0436892868752895e-06,
      "loss": 3.4972,
      "step": 68640
    },
    {
      "epoch": 0.6982930501302084,
      "grad_norm": 14.249421119689941,
      "learning_rate": 1.0433642769579873e-06,
      "loss": 3.4271,
      "step": 68645
    },
    {
      "epoch": 0.6983439127604166,
      "grad_norm": 15.138565063476562,
      "learning_rate": 1.0430393043083645e-06,
      "loss": 3.2808,
      "step": 68650
    },
    {
      "epoch": 0.698394775390625,
      "grad_norm": 15.36414909362793,
      "learning_rate": 1.0427143689347352e-06,
      "loss": 3.1296,
      "step": 68655
    },
    {
      "epoch": 0.6984456380208334,
      "grad_norm": 10.230435371398926,
      "learning_rate": 1.0423894708454144e-06,
      "loss": 3.1101,
      "step": 68660
    },
    {
      "epoch": 0.6984965006510416,
      "grad_norm": 18.308584213256836,
      "learning_rate": 1.0420646100487136e-06,
      "loss": 3.034,
      "step": 68665
    },
    {
      "epoch": 0.69854736328125,
      "grad_norm": 15.090258598327637,
      "learning_rate": 1.0417397865529442e-06,
      "loss": 3.7638,
      "step": 68670
    },
    {
      "epoch": 0.6985982259114584,
      "grad_norm": 9.507784843444824,
      "learning_rate": 1.0414150003664155e-06,
      "loss": 3.5592,
      "step": 68675
    },
    {
      "epoch": 0.6986490885416666,
      "grad_norm": 9.35131549835205,
      "learning_rate": 1.0410902514974385e-06,
      "loss": 3.5117,
      "step": 68680
    },
    {
      "epoch": 0.698699951171875,
      "grad_norm": 14.29448413848877,
      "learning_rate": 1.0407655399543227e-06,
      "loss": 3.3596,
      "step": 68685
    },
    {
      "epoch": 0.6987508138020834,
      "grad_norm": 14.195074081420898,
      "learning_rate": 1.0404408657453743e-06,
      "loss": 3.1969,
      "step": 68690
    },
    {
      "epoch": 0.6988016764322916,
      "grad_norm": 12.963168144226074,
      "learning_rate": 1.0401162288788998e-06,
      "loss": 3.2805,
      "step": 68695
    },
    {
      "epoch": 0.6988525390625,
      "grad_norm": 13.396121978759766,
      "learning_rate": 1.0397916293632062e-06,
      "loss": 2.917,
      "step": 68700
    },
    {
      "epoch": 0.6989034016927084,
      "grad_norm": 10.836840629577637,
      "learning_rate": 1.039467067206597e-06,
      "loss": 3.477,
      "step": 68705
    },
    {
      "epoch": 0.6989542643229166,
      "grad_norm": 15.258602142333984,
      "learning_rate": 1.0391425424173777e-06,
      "loss": 3.7361,
      "step": 68710
    },
    {
      "epoch": 0.699005126953125,
      "grad_norm": 12.586681365966797,
      "learning_rate": 1.038818055003849e-06,
      "loss": 3.3715,
      "step": 68715
    },
    {
      "epoch": 0.6990559895833334,
      "grad_norm": 13.802001953125,
      "learning_rate": 1.0384936049743149e-06,
      "loss": 3.7519,
      "step": 68720
    },
    {
      "epoch": 0.6991068522135416,
      "grad_norm": 14.002902030944824,
      "learning_rate": 1.0381691923370748e-06,
      "loss": 3.2382,
      "step": 68725
    },
    {
      "epoch": 0.69915771484375,
      "grad_norm": 13.671459197998047,
      "learning_rate": 1.0378448171004294e-06,
      "loss": 3.2718,
      "step": 68730
    },
    {
      "epoch": 0.6992085774739584,
      "grad_norm": 12.762028694152832,
      "learning_rate": 1.0375204792726786e-06,
      "loss": 3.0765,
      "step": 68735
    },
    {
      "epoch": 0.6992594401041666,
      "grad_norm": 10.430758476257324,
      "learning_rate": 1.0371961788621198e-06,
      "loss": 3.5619,
      "step": 68740
    },
    {
      "epoch": 0.699310302734375,
      "grad_norm": 8.087532997131348,
      "learning_rate": 1.03687191587705e-06,
      "loss": 3.4343,
      "step": 68745
    },
    {
      "epoch": 0.6993611653645834,
      "grad_norm": 11.718721389770508,
      "learning_rate": 1.0365476903257644e-06,
      "loss": 3.2718,
      "step": 68750
    },
    {
      "epoch": 0.6994120279947916,
      "grad_norm": 14.21630859375,
      "learning_rate": 1.0362235022165595e-06,
      "loss": 2.986,
      "step": 68755
    },
    {
      "epoch": 0.699462890625,
      "grad_norm": 12.95578670501709,
      "learning_rate": 1.03589935155773e-06,
      "loss": 3.206,
      "step": 68760
    },
    {
      "epoch": 0.6995137532552084,
      "grad_norm": 15.721181869506836,
      "learning_rate": 1.0355752383575687e-06,
      "loss": 3.1838,
      "step": 68765
    },
    {
      "epoch": 0.6995646158854166,
      "grad_norm": 13.355118751525879,
      "learning_rate": 1.035251162624367e-06,
      "loss": 3.5071,
      "step": 68770
    },
    {
      "epoch": 0.699615478515625,
      "grad_norm": 15.441234588623047,
      "learning_rate": 1.0349271243664177e-06,
      "loss": 3.1547,
      "step": 68775
    },
    {
      "epoch": 0.6996663411458334,
      "grad_norm": 10.825979232788086,
      "learning_rate": 1.0346031235920102e-06,
      "loss": 3.4281,
      "step": 68780
    },
    {
      "epoch": 0.6997172037760416,
      "grad_norm": 6.40767240524292,
      "learning_rate": 1.034279160309435e-06,
      "loss": 3.1522,
      "step": 68785
    },
    {
      "epoch": 0.69976806640625,
      "grad_norm": 10.204715728759766,
      "learning_rate": 1.0339552345269794e-06,
      "loss": 3.6849,
      "step": 68790
    },
    {
      "epoch": 0.6998189290364584,
      "grad_norm": 12.19060230255127,
      "learning_rate": 1.0336313462529324e-06,
      "loss": 3.5668,
      "step": 68795
    },
    {
      "epoch": 0.6998697916666666,
      "grad_norm": 10.142702102661133,
      "learning_rate": 1.0333074954955798e-06,
      "loss": 2.9964,
      "step": 68800
    },
    {
      "epoch": 0.699920654296875,
      "grad_norm": 16.380889892578125,
      "learning_rate": 1.0329836822632064e-06,
      "loss": 2.9444,
      "step": 68805
    },
    {
      "epoch": 0.6999715169270834,
      "grad_norm": 9.951184272766113,
      "learning_rate": 1.032659906564098e-06,
      "loss": 3.5085,
      "step": 68810
    },
    {
      "epoch": 0.7000223795572916,
      "grad_norm": 8.64416217803955,
      "learning_rate": 1.0323361684065385e-06,
      "loss": 3.1798,
      "step": 68815
    },
    {
      "epoch": 0.7000732421875,
      "grad_norm": 10.92931842803955,
      "learning_rate": 1.0320124677988108e-06,
      "loss": 3.269,
      "step": 68820
    },
    {
      "epoch": 0.7001241048177084,
      "grad_norm": 14.081944465637207,
      "learning_rate": 1.031688804749195e-06,
      "loss": 3.5434,
      "step": 68825
    },
    {
      "epoch": 0.7001749674479166,
      "grad_norm": 14.113259315490723,
      "learning_rate": 1.031365179265974e-06,
      "loss": 3.5396,
      "step": 68830
    },
    {
      "epoch": 0.700225830078125,
      "grad_norm": 14.627796173095703,
      "learning_rate": 1.0310415913574258e-06,
      "loss": 3.2589,
      "step": 68835
    },
    {
      "epoch": 0.7002766927083334,
      "grad_norm": 14.891929626464844,
      "learning_rate": 1.0307180410318315e-06,
      "loss": 3.3151,
      "step": 68840
    },
    {
      "epoch": 0.7003275553385416,
      "grad_norm": 11.644354820251465,
      "learning_rate": 1.0303945282974667e-06,
      "loss": 3.0651,
      "step": 68845
    },
    {
      "epoch": 0.70037841796875,
      "grad_norm": 11.94490909576416,
      "learning_rate": 1.0300710531626107e-06,
      "loss": 3.3381,
      "step": 68850
    },
    {
      "epoch": 0.7004292805989584,
      "grad_norm": 12.369614601135254,
      "learning_rate": 1.0297476156355382e-06,
      "loss": 3.4216,
      "step": 68855
    },
    {
      "epoch": 0.7004801432291666,
      "grad_norm": 11.587510108947754,
      "learning_rate": 1.0294242157245235e-06,
      "loss": 3.3817,
      "step": 68860
    },
    {
      "epoch": 0.700531005859375,
      "grad_norm": 8.7716703414917,
      "learning_rate": 1.029100853437842e-06,
      "loss": 3.3383,
      "step": 68865
    },
    {
      "epoch": 0.7005818684895834,
      "grad_norm": 11.64136791229248,
      "learning_rate": 1.028777528783767e-06,
      "loss": 3.2306,
      "step": 68870
    },
    {
      "epoch": 0.7006327311197916,
      "grad_norm": 14.60140323638916,
      "learning_rate": 1.0284542417705703e-06,
      "loss": 3.3414,
      "step": 68875
    },
    {
      "epoch": 0.70068359375,
      "grad_norm": 13.89550495147705,
      "learning_rate": 1.0281309924065224e-06,
      "loss": 2.9955,
      "step": 68880
    },
    {
      "epoch": 0.7007344563802084,
      "grad_norm": 8.002796173095703,
      "learning_rate": 1.027807780699894e-06,
      "loss": 3.3122,
      "step": 68885
    },
    {
      "epoch": 0.7007853190104166,
      "grad_norm": 13.310931205749512,
      "learning_rate": 1.0274846066589556e-06,
      "loss": 3.1744,
      "step": 68890
    },
    {
      "epoch": 0.700836181640625,
      "grad_norm": 15.553311347961426,
      "learning_rate": 1.0271614702919745e-06,
      "loss": 3.3743,
      "step": 68895
    },
    {
      "epoch": 0.7008870442708334,
      "grad_norm": 9.992782592773438,
      "learning_rate": 1.0268383716072178e-06,
      "loss": 3.4188,
      "step": 68900
    },
    {
      "epoch": 0.7009379069010416,
      "grad_norm": 9.381199836730957,
      "learning_rate": 1.0265153106129517e-06,
      "loss": 3.4018,
      "step": 68905
    },
    {
      "epoch": 0.70098876953125,
      "grad_norm": 12.410082817077637,
      "learning_rate": 1.0261922873174418e-06,
      "loss": 3.4402,
      "step": 68910
    },
    {
      "epoch": 0.7010396321614584,
      "grad_norm": 13.554468154907227,
      "learning_rate": 1.025869301728954e-06,
      "loss": 3.4538,
      "step": 68915
    },
    {
      "epoch": 0.7010904947916666,
      "grad_norm": 14.527251243591309,
      "learning_rate": 1.0255463538557505e-06,
      "loss": 3.553,
      "step": 68920
    },
    {
      "epoch": 0.701141357421875,
      "grad_norm": 12.845770835876465,
      "learning_rate": 1.0252234437060932e-06,
      "loss": 3.4095,
      "step": 68925
    },
    {
      "epoch": 0.7011922200520834,
      "grad_norm": 9.762109756469727,
      "learning_rate": 1.0249005712882457e-06,
      "loss": 3.2408,
      "step": 68930
    },
    {
      "epoch": 0.7012430826822916,
      "grad_norm": 10.933417320251465,
      "learning_rate": 1.0245777366104662e-06,
      "loss": 3.2591,
      "step": 68935
    },
    {
      "epoch": 0.7012939453125,
      "grad_norm": 14.173288345336914,
      "learning_rate": 1.0242549396810158e-06,
      "loss": 3.1852,
      "step": 68940
    },
    {
      "epoch": 0.7013448079427084,
      "grad_norm": 15.338982582092285,
      "learning_rate": 1.0239321805081536e-06,
      "loss": 3.0217,
      "step": 68945
    },
    {
      "epoch": 0.7013956705729166,
      "grad_norm": 9.645343780517578,
      "learning_rate": 1.0236094591001372e-06,
      "loss": 3.2361,
      "step": 68950
    },
    {
      "epoch": 0.701446533203125,
      "grad_norm": 9.594852447509766,
      "learning_rate": 1.0232867754652227e-06,
      "loss": 3.2452,
      "step": 68955
    },
    {
      "epoch": 0.7014973958333334,
      "grad_norm": 16.04738998413086,
      "learning_rate": 1.0229641296116647e-06,
      "loss": 3.3969,
      "step": 68960
    },
    {
      "epoch": 0.7015482584635416,
      "grad_norm": 10.130403518676758,
      "learning_rate": 1.0226415215477198e-06,
      "loss": 3.3939,
      "step": 68965
    },
    {
      "epoch": 0.70159912109375,
      "grad_norm": 10.138354301452637,
      "learning_rate": 1.022318951281642e-06,
      "loss": 3.1279,
      "step": 68970
    },
    {
      "epoch": 0.7016499837239584,
      "grad_norm": 12.420388221740723,
      "learning_rate": 1.0219964188216838e-06,
      "loss": 3.7869,
      "step": 68975
    },
    {
      "epoch": 0.7017008463541666,
      "grad_norm": 16.27092742919922,
      "learning_rate": 1.021673924176096e-06,
      "loss": 3.1377,
      "step": 68980
    },
    {
      "epoch": 0.701751708984375,
      "grad_norm": 10.564739227294922,
      "learning_rate": 1.021351467353131e-06,
      "loss": 3.4434,
      "step": 68985
    },
    {
      "epoch": 0.7018025716145834,
      "grad_norm": 13.05958366394043,
      "learning_rate": 1.0210290483610378e-06,
      "loss": 3.3063,
      "step": 68990
    },
    {
      "epoch": 0.7018534342447916,
      "grad_norm": 15.520635604858398,
      "learning_rate": 1.0207066672080664e-06,
      "loss": 2.9001,
      "step": 68995
    },
    {
      "epoch": 0.701904296875,
      "grad_norm": 8.159523963928223,
      "learning_rate": 1.0203843239024636e-06,
      "loss": 3.065,
      "step": 69000
    },
    {
      "epoch": 0.7019551595052084,
      "grad_norm": 14.460269927978516,
      "learning_rate": 1.020062018452478e-06,
      "loss": 3.3251,
      "step": 69005
    },
    {
      "epoch": 0.7020060221354166,
      "grad_norm": 7.9953837394714355,
      "learning_rate": 1.019739750866355e-06,
      "loss": 2.926,
      "step": 69010
    },
    {
      "epoch": 0.702056884765625,
      "grad_norm": 8.935430526733398,
      "learning_rate": 1.0194175211523385e-06,
      "loss": 3.1843,
      "step": 69015
    },
    {
      "epoch": 0.7021077473958334,
      "grad_norm": 13.662001609802246,
      "learning_rate": 1.019095329318674e-06,
      "loss": 3.2766,
      "step": 69020
    },
    {
      "epoch": 0.7021586100260416,
      "grad_norm": 12.309272766113281,
      "learning_rate": 1.0187731753736051e-06,
      "loss": 3.2095,
      "step": 69025
    },
    {
      "epoch": 0.70220947265625,
      "grad_norm": 9.266861915588379,
      "learning_rate": 1.0184510593253738e-06,
      "loss": 3.5939,
      "step": 69030
    },
    {
      "epoch": 0.7022603352864584,
      "grad_norm": 13.035009384155273,
      "learning_rate": 1.01812898118222e-06,
      "loss": 3.036,
      "step": 69035
    },
    {
      "epoch": 0.7023111979166666,
      "grad_norm": 12.960883140563965,
      "learning_rate": 1.0178069409523848e-06,
      "loss": 3.0416,
      "step": 69040
    },
    {
      "epoch": 0.702362060546875,
      "grad_norm": 15.219749450683594,
      "learning_rate": 1.0174849386441086e-06,
      "loss": 3.0765,
      "step": 69045
    },
    {
      "epoch": 0.7024129231770834,
      "grad_norm": 8.982955932617188,
      "learning_rate": 1.0171629742656286e-06,
      "loss": 3.3042,
      "step": 69050
    },
    {
      "epoch": 0.7024637858072916,
      "grad_norm": 13.161221504211426,
      "learning_rate": 1.016841047825182e-06,
      "loss": 3.5382,
      "step": 69055
    },
    {
      "epoch": 0.7025146484375,
      "grad_norm": 9.551596641540527,
      "learning_rate": 1.0165191593310061e-06,
      "loss": 3.3944,
      "step": 69060
    },
    {
      "epoch": 0.7025655110677084,
      "grad_norm": 11.355164527893066,
      "learning_rate": 1.0161973087913355e-06,
      "loss": 3.2315,
      "step": 69065
    },
    {
      "epoch": 0.7026163736979166,
      "grad_norm": 12.059857368469238,
      "learning_rate": 1.0158754962144058e-06,
      "loss": 3.3775,
      "step": 69070
    },
    {
      "epoch": 0.702667236328125,
      "grad_norm": 15.176041603088379,
      "learning_rate": 1.0155537216084486e-06,
      "loss": 3.1982,
      "step": 69075
    },
    {
      "epoch": 0.7027180989583334,
      "grad_norm": 12.852165222167969,
      "learning_rate": 1.0152319849816985e-06,
      "loss": 2.9204,
      "step": 69080
    },
    {
      "epoch": 0.7027689615885416,
      "grad_norm": 11.793715476989746,
      "learning_rate": 1.0149102863423862e-06,
      "loss": 3.3584,
      "step": 69085
    },
    {
      "epoch": 0.70281982421875,
      "grad_norm": 12.076817512512207,
      "learning_rate": 1.0145886256987417e-06,
      "loss": 3.6016,
      "step": 69090
    },
    {
      "epoch": 0.7028706868489584,
      "grad_norm": 7.835428237915039,
      "learning_rate": 1.0142670030589946e-06,
      "loss": 3.2869,
      "step": 69095
    },
    {
      "epoch": 0.7029215494791666,
      "grad_norm": 12.05679988861084,
      "learning_rate": 1.0139454184313752e-06,
      "loss": 3.2197,
      "step": 69100
    },
    {
      "epoch": 0.702972412109375,
      "grad_norm": 11.922722816467285,
      "learning_rate": 1.0136238718241101e-06,
      "loss": 3.2324,
      "step": 69105
    },
    {
      "epoch": 0.7030232747395834,
      "grad_norm": 10.238421440124512,
      "learning_rate": 1.0133023632454258e-06,
      "loss": 3.2494,
      "step": 69110
    },
    {
      "epoch": 0.7030741373697916,
      "grad_norm": 13.973362922668457,
      "learning_rate": 1.0129808927035473e-06,
      "loss": 3.4301,
      "step": 69115
    },
    {
      "epoch": 0.703125,
      "grad_norm": 8.110556602478027,
      "learning_rate": 1.0126594602067004e-06,
      "loss": 3.516,
      "step": 69120
    },
    {
      "epoch": 0.7031758626302084,
      "grad_norm": 13.717449188232422,
      "learning_rate": 1.0123380657631093e-06,
      "loss": 3.4103,
      "step": 69125
    },
    {
      "epoch": 0.7032267252604166,
      "grad_norm": 13.071977615356445,
      "learning_rate": 1.0120167093809954e-06,
      "loss": 3.3773,
      "step": 69130
    },
    {
      "epoch": 0.703277587890625,
      "grad_norm": 10.020981788635254,
      "learning_rate": 1.011695391068582e-06,
      "loss": 3.3164,
      "step": 69135
    },
    {
      "epoch": 0.7033284505208334,
      "grad_norm": 13.737312316894531,
      "learning_rate": 1.0113741108340893e-06,
      "loss": 3.2001,
      "step": 69140
    },
    {
      "epoch": 0.7033793131510416,
      "grad_norm": 13.198023796081543,
      "learning_rate": 1.0110528686857366e-06,
      "loss": 3.4193,
      "step": 69145
    },
    {
      "epoch": 0.70343017578125,
      "grad_norm": 12.423659324645996,
      "learning_rate": 1.010731664631743e-06,
      "loss": 3.2851,
      "step": 69150
    },
    {
      "epoch": 0.7034810384114584,
      "grad_norm": 14.201518058776855,
      "learning_rate": 1.0104104986803276e-06,
      "loss": 3.6451,
      "step": 69155
    },
    {
      "epoch": 0.7035319010416666,
      "grad_norm": 16.92333221435547,
      "learning_rate": 1.0100893708397066e-06,
      "loss": 3.459,
      "step": 69160
    },
    {
      "epoch": 0.703582763671875,
      "grad_norm": 9.523395538330078,
      "learning_rate": 1.0097682811180962e-06,
      "loss": 3.3106,
      "step": 69165
    },
    {
      "epoch": 0.7036336263020834,
      "grad_norm": 9.247840881347656,
      "learning_rate": 1.009447229523709e-06,
      "loss": 3.4684,
      "step": 69170
    },
    {
      "epoch": 0.7036844889322916,
      "grad_norm": 13.878296852111816,
      "learning_rate": 1.009126216064763e-06,
      "loss": 3.444,
      "step": 69175
    },
    {
      "epoch": 0.7037353515625,
      "grad_norm": 12.755209922790527,
      "learning_rate": 1.0088052407494695e-06,
      "loss": 3.4961,
      "step": 69180
    },
    {
      "epoch": 0.7037862141927084,
      "grad_norm": 13.315839767456055,
      "learning_rate": 1.0084843035860404e-06,
      "loss": 3.4761,
      "step": 69185
    },
    {
      "epoch": 0.7038370768229166,
      "grad_norm": 8.987351417541504,
      "learning_rate": 1.0081634045826858e-06,
      "loss": 3.6049,
      "step": 69190
    },
    {
      "epoch": 0.703887939453125,
      "grad_norm": 18.58706283569336,
      "learning_rate": 1.0078425437476169e-06,
      "loss": 3.4532,
      "step": 69195
    },
    {
      "epoch": 0.7039388020833334,
      "grad_norm": 12.25035285949707,
      "learning_rate": 1.007521721089044e-06,
      "loss": 3.6837,
      "step": 69200
    },
    {
      "epoch": 0.7039896647135416,
      "grad_norm": 18.257944107055664,
      "learning_rate": 1.0072009366151739e-06,
      "loss": 3.4599,
      "step": 69205
    },
    {
      "epoch": 0.70404052734375,
      "grad_norm": 7.997075080871582,
      "learning_rate": 1.006880190334213e-06,
      "loss": 3.3349,
      "step": 69210
    },
    {
      "epoch": 0.7040913899739584,
      "grad_norm": 8.410901069641113,
      "learning_rate": 1.0065594822543695e-06,
      "loss": 3.317,
      "step": 69215
    },
    {
      "epoch": 0.7041422526041666,
      "grad_norm": 13.914037704467773,
      "learning_rate": 1.0062388123838467e-06,
      "loss": 2.7173,
      "step": 69220
    },
    {
      "epoch": 0.704193115234375,
      "grad_norm": 13.227640151977539,
      "learning_rate": 1.0059181807308502e-06,
      "loss": 3.4732,
      "step": 69225
    },
    {
      "epoch": 0.7042439778645834,
      "grad_norm": 10.687392234802246,
      "learning_rate": 1.0055975873035836e-06,
      "loss": 3.3346,
      "step": 69230
    },
    {
      "epoch": 0.7042948404947916,
      "grad_norm": 11.866856575012207,
      "learning_rate": 1.0052770321102484e-06,
      "loss": 3.2271,
      "step": 69235
    },
    {
      "epoch": 0.704345703125,
      "grad_norm": 12.58163833618164,
      "learning_rate": 1.0049565151590461e-06,
      "loss": 3.2716,
      "step": 69240
    },
    {
      "epoch": 0.7043965657552084,
      "grad_norm": 10.621543884277344,
      "learning_rate": 1.0046360364581762e-06,
      "loss": 2.9137,
      "step": 69245
    },
    {
      "epoch": 0.7044474283854166,
      "grad_norm": 14.219128608703613,
      "learning_rate": 1.0043155960158388e-06,
      "loss": 3.88,
      "step": 69250
    },
    {
      "epoch": 0.704498291015625,
      "grad_norm": 10.410592079162598,
      "learning_rate": 1.003995193840233e-06,
      "loss": 3.3739,
      "step": 69255
    },
    {
      "epoch": 0.7045491536458334,
      "grad_norm": 8.266712188720703,
      "learning_rate": 1.003674829939556e-06,
      "loss": 3.6565,
      "step": 69260
    },
    {
      "epoch": 0.7046000162760416,
      "grad_norm": 10.390896797180176,
      "learning_rate": 1.0033545043220027e-06,
      "loss": 3.2506,
      "step": 69265
    },
    {
      "epoch": 0.70465087890625,
      "grad_norm": 8.9793119430542,
      "learning_rate": 1.0030342169957707e-06,
      "loss": 3.063,
      "step": 69270
    },
    {
      "epoch": 0.7047017415364584,
      "grad_norm": 14.336295127868652,
      "learning_rate": 1.0027139679690526e-06,
      "loss": 3.5864,
      "step": 69275
    },
    {
      "epoch": 0.7047526041666666,
      "grad_norm": 8.655488967895508,
      "learning_rate": 1.0023937572500433e-06,
      "loss": 3.4331,
      "step": 69280
    },
    {
      "epoch": 0.704803466796875,
      "grad_norm": 10.18228530883789,
      "learning_rate": 1.002073584846934e-06,
      "loss": 3.2225,
      "step": 69285
    },
    {
      "epoch": 0.7048543294270834,
      "grad_norm": 14.573174476623535,
      "learning_rate": 1.001753450767918e-06,
      "loss": 3.39,
      "step": 69290
    },
    {
      "epoch": 0.7049051920572916,
      "grad_norm": 9.717402458190918,
      "learning_rate": 1.0014333550211848e-06,
      "loss": 3.3884,
      "step": 69295
    },
    {
      "epoch": 0.7049560546875,
      "grad_norm": 7.359352111816406,
      "learning_rate": 1.001113297614923e-06,
      "loss": 3.6591,
      "step": 69300
    },
    {
      "epoch": 0.7050069173177084,
      "grad_norm": 7.139311790466309,
      "learning_rate": 1.000793278557322e-06,
      "loss": 3.2576,
      "step": 69305
    },
    {
      "epoch": 0.7050577799479166,
      "grad_norm": 11.167351722717285,
      "learning_rate": 1.000473297856571e-06,
      "loss": 3.2675,
      "step": 69310
    },
    {
      "epoch": 0.705108642578125,
      "grad_norm": 10.71337890625,
      "learning_rate": 1.000153355520855e-06,
      "loss": 3.4261,
      "step": 69315
    },
    {
      "epoch": 0.7051595052083334,
      "grad_norm": 14.224665641784668,
      "learning_rate": 9.99833451558359e-07,
      "loss": 3.3663,
      "step": 69320
    },
    {
      "epoch": 0.7052103678385416,
      "grad_norm": 10.977479934692383,
      "learning_rate": 9.995135859772687e-07,
      "loss": 3.832,
      "step": 69325
    },
    {
      "epoch": 0.70526123046875,
      "grad_norm": 10.125252723693848,
      "learning_rate": 9.991937587857684e-07,
      "loss": 3.582,
      "step": 69330
    },
    {
      "epoch": 0.7053120930989584,
      "grad_norm": 11.237396240234375,
      "learning_rate": 9.9887396999204e-07,
      "loss": 3.3685,
      "step": 69335
    },
    {
      "epoch": 0.7053629557291666,
      "grad_norm": 7.409149169921875,
      "learning_rate": 9.985542196042643e-07,
      "loss": 3.3208,
      "step": 69340
    },
    {
      "epoch": 0.705413818359375,
      "grad_norm": 11.912979125976562,
      "learning_rate": 9.982345076306244e-07,
      "loss": 3.4061,
      "step": 69345
    },
    {
      "epoch": 0.7054646809895834,
      "grad_norm": 16.61011505126953,
      "learning_rate": 9.979148340792974e-07,
      "loss": 3.2365,
      "step": 69350
    },
    {
      "epoch": 0.7055155436197916,
      "grad_norm": 12.74564266204834,
      "learning_rate": 9.975951989584643e-07,
      "loss": 3.5396,
      "step": 69355
    },
    {
      "epoch": 0.70556640625,
      "grad_norm": 11.513494491577148,
      "learning_rate": 9.972756022763009e-07,
      "loss": 3.2285,
      "step": 69360
    },
    {
      "epoch": 0.7056172688802084,
      "grad_norm": 12.513010025024414,
      "learning_rate": 9.96956044040986e-07,
      "loss": 3.0373,
      "step": 69365
    },
    {
      "epoch": 0.7056681315104166,
      "grad_norm": 14.658462524414062,
      "learning_rate": 9.966365242606946e-07,
      "loss": 3.196,
      "step": 69370
    },
    {
      "epoch": 0.705718994140625,
      "grad_norm": 21.01740264892578,
      "learning_rate": 9.963170429436003e-07,
      "loss": 3.4951,
      "step": 69375
    },
    {
      "epoch": 0.7057698567708334,
      "grad_norm": 12.703271865844727,
      "learning_rate": 9.959976000978779e-07,
      "loss": 3.1626,
      "step": 69380
    },
    {
      "epoch": 0.7058207194010416,
      "grad_norm": 13.965160369873047,
      "learning_rate": 9.956781957317016e-07,
      "loss": 3.6084,
      "step": 69385
    },
    {
      "epoch": 0.70587158203125,
      "grad_norm": 16.33390998840332,
      "learning_rate": 9.953588298532422e-07,
      "loss": 3.2092,
      "step": 69390
    },
    {
      "epoch": 0.7059224446614584,
      "grad_norm": 12.628070831298828,
      "learning_rate": 9.950395024706703e-07,
      "loss": 3.4365,
      "step": 69395
    },
    {
      "epoch": 0.7059733072916666,
      "grad_norm": 9.108073234558105,
      "learning_rate": 9.94720213592155e-07,
      "loss": 3.4463,
      "step": 69400
    },
    {
      "epoch": 0.706024169921875,
      "grad_norm": 9.910865783691406,
      "learning_rate": 9.944009632258664e-07,
      "loss": 3.2692,
      "step": 69405
    },
    {
      "epoch": 0.7060750325520834,
      "grad_norm": 7.650475978851318,
      "learning_rate": 9.940817513799734e-07,
      "loss": 3.1788,
      "step": 69410
    },
    {
      "epoch": 0.7061258951822916,
      "grad_norm": 14.100626945495605,
      "learning_rate": 9.937625780626405e-07,
      "loss": 3.3328,
      "step": 69415
    },
    {
      "epoch": 0.7061767578125,
      "grad_norm": 11.57579231262207,
      "learning_rate": 9.934434432820362e-07,
      "loss": 3.272,
      "step": 69420
    },
    {
      "epoch": 0.7062276204427084,
      "grad_norm": 14.377008438110352,
      "learning_rate": 9.931243470463242e-07,
      "loss": 3.5567,
      "step": 69425
    },
    {
      "epoch": 0.7062784830729166,
      "grad_norm": 13.055729866027832,
      "learning_rate": 9.928052893636677e-07,
      "loss": 3.2766,
      "step": 69430
    },
    {
      "epoch": 0.706329345703125,
      "grad_norm": 10.804378509521484,
      "learning_rate": 9.924862702422309e-07,
      "loss": 3.175,
      "step": 69435
    },
    {
      "epoch": 0.7063802083333334,
      "grad_norm": 11.880969047546387,
      "learning_rate": 9.921672896901762e-07,
      "loss": 3.4717,
      "step": 69440
    },
    {
      "epoch": 0.7064310709635416,
      "grad_norm": 38.35500717163086,
      "learning_rate": 9.91848347715664e-07,
      "loss": 3.4817,
      "step": 69445
    },
    {
      "epoch": 0.70648193359375,
      "grad_norm": 6.873586177825928,
      "learning_rate": 9.915294443268544e-07,
      "loss": 3.4381,
      "step": 69450
    },
    {
      "epoch": 0.7065327962239584,
      "grad_norm": 11.292330741882324,
      "learning_rate": 9.912105795319054e-07,
      "loss": 3.6071,
      "step": 69455
    },
    {
      "epoch": 0.7065836588541666,
      "grad_norm": 15.964972496032715,
      "learning_rate": 9.908917533389764e-07,
      "loss": 3.2474,
      "step": 69460
    },
    {
      "epoch": 0.706634521484375,
      "grad_norm": 13.744715690612793,
      "learning_rate": 9.905729657562248e-07,
      "loss": 3.0413,
      "step": 69465
    },
    {
      "epoch": 0.7066853841145834,
      "grad_norm": 15.283783912658691,
      "learning_rate": 9.902542167918064e-07,
      "loss": 3.271,
      "step": 69470
    },
    {
      "epoch": 0.7067362467447916,
      "grad_norm": 10.285155296325684,
      "learning_rate": 9.899355064538749e-07,
      "loss": 3.3835,
      "step": 69475
    },
    {
      "epoch": 0.706787109375,
      "grad_norm": 10.142790794372559,
      "learning_rate": 9.896168347505854e-07,
      "loss": 3.5299,
      "step": 69480
    },
    {
      "epoch": 0.7068379720052084,
      "grad_norm": 14.190443992614746,
      "learning_rate": 9.892982016900923e-07,
      "loss": 2.9823,
      "step": 69485
    },
    {
      "epoch": 0.7068888346354166,
      "grad_norm": 13.098793029785156,
      "learning_rate": 9.889796072805466e-07,
      "loss": 3.1256,
      "step": 69490
    },
    {
      "epoch": 0.706939697265625,
      "grad_norm": 12.860541343688965,
      "learning_rate": 9.886610515300986e-07,
      "loss": 3.9647,
      "step": 69495
    },
    {
      "epoch": 0.7069905598958334,
      "grad_norm": 10.856886863708496,
      "learning_rate": 9.883425344469004e-07,
      "loss": 2.7901,
      "step": 69500
    },
    {
      "epoch": 0.7070414225260416,
      "grad_norm": 17.13384437561035,
      "learning_rate": 9.880240560390992e-07,
      "loss": 3.3969,
      "step": 69505
    },
    {
      "epoch": 0.70709228515625,
      "grad_norm": 11.889996528625488,
      "learning_rate": 9.877056163148448e-07,
      "loss": 3.2367,
      "step": 69510
    },
    {
      "epoch": 0.7071431477864584,
      "grad_norm": 10.63394832611084,
      "learning_rate": 9.87387215282283e-07,
      "loss": 3.5426,
      "step": 69515
    },
    {
      "epoch": 0.7071940104166666,
      "grad_norm": 9.318733215332031,
      "learning_rate": 9.870688529495618e-07,
      "loss": 3.6947,
      "step": 69520
    },
    {
      "epoch": 0.707244873046875,
      "grad_norm": 13.429479598999023,
      "learning_rate": 9.867505293248253e-07,
      "loss": 3.1797,
      "step": 69525
    },
    {
      "epoch": 0.7072957356770834,
      "grad_norm": 10.64814281463623,
      "learning_rate": 9.864322444162166e-07,
      "loss": 2.9945,
      "step": 69530
    },
    {
      "epoch": 0.7073465983072916,
      "grad_norm": 11.657544136047363,
      "learning_rate": 9.861139982318807e-07,
      "loss": 3.2833,
      "step": 69535
    },
    {
      "epoch": 0.7073974609375,
      "grad_norm": 13.675775527954102,
      "learning_rate": 9.857957907799601e-07,
      "loss": 3.5087,
      "step": 69540
    },
    {
      "epoch": 0.7074483235677084,
      "grad_norm": 15.808775901794434,
      "learning_rate": 9.85477622068595e-07,
      "loss": 3.444,
      "step": 69545
    },
    {
      "epoch": 0.7074991861979166,
      "grad_norm": 10.257986068725586,
      "learning_rate": 9.851594921059252e-07,
      "loss": 3.2853,
      "step": 69550
    },
    {
      "epoch": 0.707550048828125,
      "grad_norm": 12.067086219787598,
      "learning_rate": 9.848414009000916e-07,
      "loss": 3.6043,
      "step": 69555
    },
    {
      "epoch": 0.7076009114583334,
      "grad_norm": 13.973238945007324,
      "learning_rate": 9.845233484592308e-07,
      "loss": 3.3014,
      "step": 69560
    },
    {
      "epoch": 0.7076517740885416,
      "grad_norm": 11.880187034606934,
      "learning_rate": 9.84205334791482e-07,
      "loss": 3.3167,
      "step": 69565
    },
    {
      "epoch": 0.70770263671875,
      "grad_norm": 12.605144500732422,
      "learning_rate": 9.838873599049795e-07,
      "loss": 3.3419,
      "step": 69570
    },
    {
      "epoch": 0.7077534993489584,
      "grad_norm": 14.770031929016113,
      "learning_rate": 9.835694238078601e-07,
      "loss": 3.1096,
      "step": 69575
    },
    {
      "epoch": 0.7078043619791666,
      "grad_norm": 13.251535415649414,
      "learning_rate": 9.83251526508258e-07,
      "loss": 3.1371,
      "step": 69580
    },
    {
      "epoch": 0.707855224609375,
      "grad_norm": 11.815899848937988,
      "learning_rate": 9.829336680143048e-07,
      "loss": 3.3847,
      "step": 69585
    },
    {
      "epoch": 0.7079060872395834,
      "grad_norm": 9.919445037841797,
      "learning_rate": 9.826158483341342e-07,
      "loss": 3.4533,
      "step": 69590
    },
    {
      "epoch": 0.7079569498697916,
      "grad_norm": 12.395959854125977,
      "learning_rate": 9.822980674758783e-07,
      "loss": 3.1149,
      "step": 69595
    },
    {
      "epoch": 0.7080078125,
      "grad_norm": 14.6797456741333,
      "learning_rate": 9.819803254476668e-07,
      "loss": 3.2177,
      "step": 69600
    },
    {
      "epoch": 0.7080586751302084,
      "grad_norm": 12.67190170288086,
      "learning_rate": 9.816626222576289e-07,
      "loss": 3.352,
      "step": 69605
    },
    {
      "epoch": 0.7081095377604166,
      "grad_norm": 9.85501480102539,
      "learning_rate": 9.81344957913892e-07,
      "loss": 3.3222,
      "step": 69610
    },
    {
      "epoch": 0.708160400390625,
      "grad_norm": 16.397167205810547,
      "learning_rate": 9.810273324245842e-07,
      "loss": 3.7158,
      "step": 69615
    },
    {
      "epoch": 0.7082112630208334,
      "grad_norm": 11.681159973144531,
      "learning_rate": 9.807097457978331e-07,
      "loss": 3.6375,
      "step": 69620
    },
    {
      "epoch": 0.7082621256510416,
      "grad_norm": 9.919473648071289,
      "learning_rate": 9.803921980417622e-07,
      "loss": 3.0976,
      "step": 69625
    },
    {
      "epoch": 0.70831298828125,
      "grad_norm": 11.035103797912598,
      "learning_rate": 9.800746891644975e-07,
      "loss": 3.1882,
      "step": 69630
    },
    {
      "epoch": 0.7083638509114584,
      "grad_norm": 12.774720191955566,
      "learning_rate": 9.797572191741608e-07,
      "loss": 3.1872,
      "step": 69635
    },
    {
      "epoch": 0.7084147135416666,
      "grad_norm": 10.840656280517578,
      "learning_rate": 9.794397880788761e-07,
      "loss": 3.3356,
      "step": 69640
    },
    {
      "epoch": 0.708465576171875,
      "grad_norm": 13.177209854125977,
      "learning_rate": 9.791223958867631e-07,
      "loss": 3.4211,
      "step": 69645
    },
    {
      "epoch": 0.7085164388020834,
      "grad_norm": 9.744634628295898,
      "learning_rate": 9.788050426059442e-07,
      "loss": 3.279,
      "step": 69650
    },
    {
      "epoch": 0.7085673014322916,
      "grad_norm": 7.26648473739624,
      "learning_rate": 9.784877282445376e-07,
      "loss": 3.1761,
      "step": 69655
    },
    {
      "epoch": 0.7086181640625,
      "grad_norm": 10.062418937683105,
      "learning_rate": 9.78170452810661e-07,
      "loss": 3.3257,
      "step": 69660
    },
    {
      "epoch": 0.7086690266927084,
      "grad_norm": 11.900903701782227,
      "learning_rate": 9.778532163124324e-07,
      "loss": 3.3456,
      "step": 69665
    },
    {
      "epoch": 0.7087198893229166,
      "grad_norm": 10.382468223571777,
      "learning_rate": 9.775360187579697e-07,
      "loss": 3.0791,
      "step": 69670
    },
    {
      "epoch": 0.708770751953125,
      "grad_norm": 7.0931396484375,
      "learning_rate": 9.772188601553869e-07,
      "loss": 3.2358,
      "step": 69675
    },
    {
      "epoch": 0.7088216145833334,
      "grad_norm": 12.681374549865723,
      "learning_rate": 9.769017405127987e-07,
      "loss": 3.0573,
      "step": 69680
    },
    {
      "epoch": 0.7088724772135416,
      "grad_norm": 16.667367935180664,
      "learning_rate": 9.765846598383177e-07,
      "loss": 3.2417,
      "step": 69685
    },
    {
      "epoch": 0.70892333984375,
      "grad_norm": 11.11923599243164,
      "learning_rate": 9.762676181400568e-07,
      "loss": 3.3146,
      "step": 69690
    },
    {
      "epoch": 0.7089742024739584,
      "grad_norm": 8.557740211486816,
      "learning_rate": 9.759506154261289e-07,
      "loss": 3.7042,
      "step": 69695
    },
    {
      "epoch": 0.7090250651041666,
      "grad_norm": 10.00163745880127,
      "learning_rate": 9.756336517046431e-07,
      "loss": 2.9392,
      "step": 69700
    },
    {
      "epoch": 0.709075927734375,
      "grad_norm": 15.909320831298828,
      "learning_rate": 9.75316726983708e-07,
      "loss": 3.6704,
      "step": 69705
    },
    {
      "epoch": 0.7091267903645834,
      "grad_norm": 12.77523422241211,
      "learning_rate": 9.749998412714342e-07,
      "loss": 3.9356,
      "step": 69710
    },
    {
      "epoch": 0.7091776529947916,
      "grad_norm": 12.011052131652832,
      "learning_rate": 9.74682994575927e-07,
      "loss": 3.4001,
      "step": 69715
    },
    {
      "epoch": 0.709228515625,
      "grad_norm": 15.927411079406738,
      "learning_rate": 9.743661869052937e-07,
      "loss": 3.2104,
      "step": 69720
    },
    {
      "epoch": 0.7092793782552084,
      "grad_norm": 11.16408634185791,
      "learning_rate": 9.74049418267641e-07,
      "loss": 3.0609,
      "step": 69725
    },
    {
      "epoch": 0.7093302408854166,
      "grad_norm": 12.752314567565918,
      "learning_rate": 9.73732688671072e-07,
      "loss": 3.6774,
      "step": 69730
    },
    {
      "epoch": 0.709381103515625,
      "grad_norm": 11.089025497436523,
      "learning_rate": 9.734159981236906e-07,
      "loss": 3.4763,
      "step": 69735
    },
    {
      "epoch": 0.7094319661458334,
      "grad_norm": 10.589859008789062,
      "learning_rate": 9.730993466335984e-07,
      "loss": 3.0983,
      "step": 69740
    },
    {
      "epoch": 0.7094828287760416,
      "grad_norm": 9.92780876159668,
      "learning_rate": 9.727827342088972e-07,
      "loss": 3.2294,
      "step": 69745
    },
    {
      "epoch": 0.70953369140625,
      "grad_norm": 11.25098705291748,
      "learning_rate": 9.724661608576885e-07,
      "loss": 3.7487,
      "step": 69750
    },
    {
      "epoch": 0.7095845540364584,
      "grad_norm": 13.248087882995605,
      "learning_rate": 9.721496265880713e-07,
      "loss": 3.1595,
      "step": 69755
    },
    {
      "epoch": 0.7096354166666666,
      "grad_norm": 10.853324890136719,
      "learning_rate": 9.718331314081426e-07,
      "loss": 3.504,
      "step": 69760
    },
    {
      "epoch": 0.709686279296875,
      "grad_norm": 9.86822509765625,
      "learning_rate": 9.715166753260022e-07,
      "loss": 3.2758,
      "step": 69765
    },
    {
      "epoch": 0.7097371419270834,
      "grad_norm": 13.17049503326416,
      "learning_rate": 9.712002583497443e-07,
      "loss": 3.2826,
      "step": 69770
    },
    {
      "epoch": 0.7097880045572916,
      "grad_norm": 9.153229713439941,
      "learning_rate": 9.708838804874663e-07,
      "loss": 3.3263,
      "step": 69775
    },
    {
      "epoch": 0.7098388671875,
      "grad_norm": 9.449225425720215,
      "learning_rate": 9.70567541747261e-07,
      "loss": 3.1877,
      "step": 69780
    },
    {
      "epoch": 0.7098897298177084,
      "grad_norm": 10.048635482788086,
      "learning_rate": 9.702512421372234e-07,
      "loss": 3.1857,
      "step": 69785
    },
    {
      "epoch": 0.7099405924479166,
      "grad_norm": 13.345157623291016,
      "learning_rate": 9.699349816654443e-07,
      "loss": 3.3308,
      "step": 69790
    },
    {
      "epoch": 0.709991455078125,
      "grad_norm": 12.826769828796387,
      "learning_rate": 9.69618760340017e-07,
      "loss": 3.4765,
      "step": 69795
    },
    {
      "epoch": 0.7100423177083334,
      "grad_norm": 8.920169830322266,
      "learning_rate": 9.693025781690299e-07,
      "loss": 3.4218,
      "step": 69800
    },
    {
      "epoch": 0.7100931803385416,
      "grad_norm": 10.686864852905273,
      "learning_rate": 9.689864351605746e-07,
      "loss": 3.2166,
      "step": 69805
    },
    {
      "epoch": 0.71014404296875,
      "grad_norm": 8.101327896118164,
      "learning_rate": 9.686703313227386e-07,
      "loss": 3.4575,
      "step": 69810
    },
    {
      "epoch": 0.7101949055989584,
      "grad_norm": 9.720623970031738,
      "learning_rate": 9.683542666636082e-07,
      "loss": 3.548,
      "step": 69815
    },
    {
      "epoch": 0.7102457682291666,
      "grad_norm": 12.997328758239746,
      "learning_rate": 9.680382411912706e-07,
      "loss": 3.2979,
      "step": 69820
    },
    {
      "epoch": 0.710296630859375,
      "grad_norm": 9.228484153747559,
      "learning_rate": 9.677222549138127e-07,
      "loss": 3.4373,
      "step": 69825
    },
    {
      "epoch": 0.7103474934895834,
      "grad_norm": 11.505775451660156,
      "learning_rate": 9.674063078393175e-07,
      "loss": 3.5473,
      "step": 69830
    },
    {
      "epoch": 0.7103983561197916,
      "grad_norm": 15.89929485321045,
      "learning_rate": 9.67090399975868e-07,
      "loss": 3.222,
      "step": 69835
    },
    {
      "epoch": 0.71044921875,
      "grad_norm": 7.241600036621094,
      "learning_rate": 9.66774531331548e-07,
      "loss": 3.1403,
      "step": 69840
    },
    {
      "epoch": 0.7105000813802084,
      "grad_norm": 10.587393760681152,
      "learning_rate": 9.664587019144373e-07,
      "loss": 3.4444,
      "step": 69845
    },
    {
      "epoch": 0.7105509440104166,
      "grad_norm": 14.452857971191406,
      "learning_rate": 9.66142911732618e-07,
      "loss": 2.9991,
      "step": 69850
    },
    {
      "epoch": 0.710601806640625,
      "grad_norm": 13.695948600769043,
      "learning_rate": 9.65827160794168e-07,
      "loss": 2.9567,
      "step": 69855
    },
    {
      "epoch": 0.7106526692708334,
      "grad_norm": 11.315937042236328,
      "learning_rate": 9.655114491071673e-07,
      "loss": 3.2219,
      "step": 69860
    },
    {
      "epoch": 0.7107035319010416,
      "grad_norm": 11.866053581237793,
      "learning_rate": 9.651957766796927e-07,
      "loss": 3.5155,
      "step": 69865
    },
    {
      "epoch": 0.71075439453125,
      "grad_norm": 12.22433853149414,
      "learning_rate": 9.64880143519819e-07,
      "loss": 3.4572,
      "step": 69870
    },
    {
      "epoch": 0.7108052571614584,
      "grad_norm": 11.939126014709473,
      "learning_rate": 9.645645496356234e-07,
      "loss": 3.1203,
      "step": 69875
    },
    {
      "epoch": 0.7108561197916666,
      "grad_norm": 12.543699264526367,
      "learning_rate": 9.642489950351804e-07,
      "loss": 3.3598,
      "step": 69880
    },
    {
      "epoch": 0.710906982421875,
      "grad_norm": 8.45138168334961,
      "learning_rate": 9.63933479726563e-07,
      "loss": 3.305,
      "step": 69885
    },
    {
      "epoch": 0.7109578450520834,
      "grad_norm": 10.792311668395996,
      "learning_rate": 9.636180037178436e-07,
      "loss": 3.1086,
      "step": 69890
    },
    {
      "epoch": 0.7110087076822916,
      "grad_norm": 12.166705131530762,
      "learning_rate": 9.633025670170923e-07,
      "loss": 3.1846,
      "step": 69895
    },
    {
      "epoch": 0.7110595703125,
      "grad_norm": 10.266594886779785,
      "learning_rate": 9.62987169632381e-07,
      "loss": 3.5435,
      "step": 69900
    },
    {
      "epoch": 0.7111104329427084,
      "grad_norm": 15.00068187713623,
      "learning_rate": 9.62671811571779e-07,
      "loss": 3.2666,
      "step": 69905
    },
    {
      "epoch": 0.7111612955729166,
      "grad_norm": 17.427892684936523,
      "learning_rate": 9.623564928433537e-07,
      "loss": 3.2168,
      "step": 69910
    },
    {
      "epoch": 0.711212158203125,
      "grad_norm": 11.1777925491333,
      "learning_rate": 9.620412134551739e-07,
      "loss": 3.665,
      "step": 69915
    },
    {
      "epoch": 0.7112630208333334,
      "grad_norm": 13.45982551574707,
      "learning_rate": 9.61725973415305e-07,
      "loss": 3.2419,
      "step": 69920
    },
    {
      "epoch": 0.7113138834635416,
      "grad_norm": 12.354000091552734,
      "learning_rate": 9.614107727318116e-07,
      "loss": 3.1457,
      "step": 69925
    },
    {
      "epoch": 0.71136474609375,
      "grad_norm": 14.382411003112793,
      "learning_rate": 9.610956114127592e-07,
      "loss": 3.7577,
      "step": 69930
    },
    {
      "epoch": 0.7114156087239584,
      "grad_norm": 13.039135932922363,
      "learning_rate": 9.607804894662114e-07,
      "loss": 3.369,
      "step": 69935
    },
    {
      "epoch": 0.7114664713541666,
      "grad_norm": 7.945051193237305,
      "learning_rate": 9.6046540690023e-07,
      "loss": 3.5365,
      "step": 69940
    },
    {
      "epoch": 0.711517333984375,
      "grad_norm": 11.75335693359375,
      "learning_rate": 9.601503637228763e-07,
      "loss": 3.4087,
      "step": 69945
    },
    {
      "epoch": 0.7115681966145834,
      "grad_norm": 6.930155277252197,
      "learning_rate": 9.59835359942209e-07,
      "loss": 3.2005,
      "step": 69950
    },
    {
      "epoch": 0.7116190592447916,
      "grad_norm": 14.273049354553223,
      "learning_rate": 9.595203955662905e-07,
      "loss": 3.0292,
      "step": 69955
    },
    {
      "epoch": 0.711669921875,
      "grad_norm": 10.134856224060059,
      "learning_rate": 9.592054706031772e-07,
      "loss": 3.3678,
      "step": 69960
    },
    {
      "epoch": 0.7117207845052084,
      "grad_norm": 13.185558319091797,
      "learning_rate": 9.588905850609274e-07,
      "loss": 3.3033,
      "step": 69965
    },
    {
      "epoch": 0.7117716471354166,
      "grad_norm": 18.926721572875977,
      "learning_rate": 9.585757389475953e-07,
      "loss": 3.2266,
      "step": 69970
    },
    {
      "epoch": 0.711822509765625,
      "grad_norm": 14.387677192687988,
      "learning_rate": 9.582609322712377e-07,
      "loss": 3.305,
      "step": 69975
    },
    {
      "epoch": 0.7118733723958334,
      "grad_norm": 14.691122055053711,
      "learning_rate": 9.579461650399097e-07,
      "loss": 3.3729,
      "step": 69980
    },
    {
      "epoch": 0.7119242350260416,
      "grad_norm": 15.056170463562012,
      "learning_rate": 9.576314372616635e-07,
      "loss": 3.0395,
      "step": 69985
    },
    {
      "epoch": 0.71197509765625,
      "grad_norm": 17.34141731262207,
      "learning_rate": 9.573167489445507e-07,
      "loss": 3.7642,
      "step": 69990
    },
    {
      "epoch": 0.7120259602864584,
      "grad_norm": 10.424257278442383,
      "learning_rate": 9.570021000966237e-07,
      "loss": 3.2243,
      "step": 69995
    },
    {
      "epoch": 0.7120768229166666,
      "grad_norm": 9.840085983276367,
      "learning_rate": 9.566874907259316e-07,
      "loss": 2.7915,
      "step": 70000
    },
    {
      "epoch": 0.712127685546875,
      "grad_norm": 9.245528221130371,
      "learning_rate": 9.563729208405253e-07,
      "loss": 3.2404,
      "step": 70005
    },
    {
      "epoch": 0.7121785481770834,
      "grad_norm": 14.306405067443848,
      "learning_rate": 9.560583904484509e-07,
      "loss": 3.3555,
      "step": 70010
    },
    {
      "epoch": 0.7122294108072916,
      "grad_norm": 10.101147651672363,
      "learning_rate": 9.557438995577573e-07,
      "loss": 3.3163,
      "step": 70015
    },
    {
      "epoch": 0.7122802734375,
      "grad_norm": 16.2298526763916,
      "learning_rate": 9.554294481764902e-07,
      "loss": 3.2587,
      "step": 70020
    },
    {
      "epoch": 0.7123311360677084,
      "grad_norm": 16.712909698486328,
      "learning_rate": 9.551150363126937e-07,
      "loss": 3.5174,
      "step": 70025
    },
    {
      "epoch": 0.7123819986979166,
      "grad_norm": 12.494762420654297,
      "learning_rate": 9.548006639744129e-07,
      "loss": 3.494,
      "step": 70030
    },
    {
      "epoch": 0.712432861328125,
      "grad_norm": 8.947538375854492,
      "learning_rate": 9.544863311696914e-07,
      "loss": 3.1756,
      "step": 70035
    },
    {
      "epoch": 0.7124837239583334,
      "grad_norm": 12.291443824768066,
      "learning_rate": 9.541720379065711e-07,
      "loss": 3.183,
      "step": 70040
    },
    {
      "epoch": 0.7125345865885416,
      "grad_norm": 15.154228210449219,
      "learning_rate": 9.538577841930918e-07,
      "loss": 3.4816,
      "step": 70045
    },
    {
      "epoch": 0.71258544921875,
      "grad_norm": 12.65209674835205,
      "learning_rate": 9.535435700372955e-07,
      "loss": 3.6757,
      "step": 70050
    },
    {
      "epoch": 0.7126363118489584,
      "grad_norm": 12.21906566619873,
      "learning_rate": 9.532293954472197e-07,
      "loss": 3.2487,
      "step": 70055
    },
    {
      "epoch": 0.7126871744791666,
      "grad_norm": 8.432168960571289,
      "learning_rate": 9.529152604309039e-07,
      "loss": 2.9157,
      "step": 70060
    },
    {
      "epoch": 0.712738037109375,
      "grad_norm": 10.848390579223633,
      "learning_rate": 9.526011649963837e-07,
      "loss": 3.3407,
      "step": 70065
    },
    {
      "epoch": 0.7127888997395834,
      "grad_norm": 18.4654541015625,
      "learning_rate": 9.522871091516966e-07,
      "loss": 3.6414,
      "step": 70070
    },
    {
      "epoch": 0.7128397623697916,
      "grad_norm": 12.46338176727295,
      "learning_rate": 9.51973092904877e-07,
      "loss": 3.0989,
      "step": 70075
    },
    {
      "epoch": 0.712890625,
      "grad_norm": 12.28748893737793,
      "learning_rate": 9.516591162639582e-07,
      "loss": 3.4607,
      "step": 70080
    },
    {
      "epoch": 0.7129414876302084,
      "grad_norm": 8.627691268920898,
      "learning_rate": 9.513451792369738e-07,
      "loss": 3.8247,
      "step": 70085
    },
    {
      "epoch": 0.7129923502604166,
      "grad_norm": 9.558513641357422,
      "learning_rate": 9.51031281831957e-07,
      "loss": 3.6145,
      "step": 70090
    },
    {
      "epoch": 0.713043212890625,
      "grad_norm": 9.67013931274414,
      "learning_rate": 9.507174240569375e-07,
      "loss": 3.7408,
      "step": 70095
    },
    {
      "epoch": 0.7130940755208334,
      "grad_norm": 14.165182113647461,
      "learning_rate": 9.504036059199453e-07,
      "loss": 3.1084,
      "step": 70100
    },
    {
      "epoch": 0.7131449381510416,
      "grad_norm": 14.601203918457031,
      "learning_rate": 9.500898274290083e-07,
      "loss": 3.3601,
      "step": 70105
    },
    {
      "epoch": 0.71319580078125,
      "grad_norm": 8.760289192199707,
      "learning_rate": 9.49776088592157e-07,
      "loss": 3.2361,
      "step": 70110
    },
    {
      "epoch": 0.7132466634114584,
      "grad_norm": 12.00662899017334,
      "learning_rate": 9.494623894174171e-07,
      "loss": 3.3172,
      "step": 70115
    },
    {
      "epoch": 0.7132975260416666,
      "grad_norm": 15.779747009277344,
      "learning_rate": 9.491487299128138e-07,
      "loss": 3.1276,
      "step": 70120
    },
    {
      "epoch": 0.713348388671875,
      "grad_norm": 9.786253929138184,
      "learning_rate": 9.488351100863732e-07,
      "loss": 3.1352,
      "step": 70125
    },
    {
      "epoch": 0.7133992513020834,
      "grad_norm": 10.486870765686035,
      "learning_rate": 9.485215299461181e-07,
      "loss": 3.3165,
      "step": 70130
    },
    {
      "epoch": 0.7134501139322916,
      "grad_norm": 10.869139671325684,
      "learning_rate": 9.482079895000726e-07,
      "loss": 3.1894,
      "step": 70135
    },
    {
      "epoch": 0.7135009765625,
      "grad_norm": 13.020895957946777,
      "learning_rate": 9.478944887562569e-07,
      "loss": 3.2775,
      "step": 70140
    },
    {
      "epoch": 0.7135518391927084,
      "grad_norm": 10.164044380187988,
      "learning_rate": 9.475810277226941e-07,
      "loss": 3.8045,
      "step": 70145
    },
    {
      "epoch": 0.7136027018229166,
      "grad_norm": 13.305707931518555,
      "learning_rate": 9.472676064074024e-07,
      "loss": 3.0316,
      "step": 70150
    },
    {
      "epoch": 0.713653564453125,
      "grad_norm": 15.2430419921875,
      "learning_rate": 9.469542248183999e-07,
      "loss": 3.5371,
      "step": 70155
    },
    {
      "epoch": 0.7137044270833334,
      "grad_norm": 7.959345817565918,
      "learning_rate": 9.466408829637058e-07,
      "loss": 3.4444,
      "step": 70160
    },
    {
      "epoch": 0.7137552897135416,
      "grad_norm": 9.482425689697266,
      "learning_rate": 9.463275808513373e-07,
      "loss": 3.3864,
      "step": 70165
    },
    {
      "epoch": 0.71380615234375,
      "grad_norm": 10.571792602539062,
      "learning_rate": 9.460143184893094e-07,
      "loss": 2.9948,
      "step": 70170
    },
    {
      "epoch": 0.7138570149739584,
      "grad_norm": 11.90029525756836,
      "learning_rate": 9.45701095885637e-07,
      "loss": 3.1864,
      "step": 70175
    },
    {
      "epoch": 0.7139078776041666,
      "grad_norm": 10.637242317199707,
      "learning_rate": 9.453879130483323e-07,
      "loss": 3.3903,
      "step": 70180
    },
    {
      "epoch": 0.713958740234375,
      "grad_norm": 15.166751861572266,
      "learning_rate": 9.450747699854099e-07,
      "loss": 3.9749,
      "step": 70185
    },
    {
      "epoch": 0.7140096028645834,
      "grad_norm": 8.722638130187988,
      "learning_rate": 9.447616667048814e-07,
      "loss": 3.7228,
      "step": 70190
    },
    {
      "epoch": 0.7140604654947916,
      "grad_norm": 7.907155990600586,
      "learning_rate": 9.444486032147573e-07,
      "loss": 3.1887,
      "step": 70195
    },
    {
      "epoch": 0.714111328125,
      "grad_norm": 12.05744457244873,
      "learning_rate": 9.441355795230459e-07,
      "loss": 3.4451,
      "step": 70200
    },
    {
      "epoch": 0.7141621907552084,
      "grad_norm": 11.853819847106934,
      "learning_rate": 9.438225956377578e-07,
      "loss": 3.0237,
      "step": 70205
    },
    {
      "epoch": 0.7142130533854166,
      "grad_norm": 11.00097370147705,
      "learning_rate": 9.43509651566899e-07,
      "loss": 3.4609,
      "step": 70210
    },
    {
      "epoch": 0.714263916015625,
      "grad_norm": 17.25116539001465,
      "learning_rate": 9.43196747318477e-07,
      "loss": 3.8878,
      "step": 70215
    },
    {
      "epoch": 0.7143147786458334,
      "grad_norm": 11.077966690063477,
      "learning_rate": 9.428838829004978e-07,
      "loss": 3.3917,
      "step": 70220
    },
    {
      "epoch": 0.7143656412760416,
      "grad_norm": 11.697901725769043,
      "learning_rate": 9.425710583209655e-07,
      "loss": 3.4042,
      "step": 70225
    },
    {
      "epoch": 0.71441650390625,
      "grad_norm": 10.441009521484375,
      "learning_rate": 9.422582735878835e-07,
      "loss": 3.0199,
      "step": 70230
    },
    {
      "epoch": 0.7144673665364584,
      "grad_norm": 9.411457061767578,
      "learning_rate": 9.419455287092533e-07,
      "loss": 3.3712,
      "step": 70235
    },
    {
      "epoch": 0.7145182291666666,
      "grad_norm": 13.511369705200195,
      "learning_rate": 9.416328236930777e-07,
      "loss": 3.3586,
      "step": 70240
    },
    {
      "epoch": 0.714569091796875,
      "grad_norm": 13.114068984985352,
      "learning_rate": 9.413201585473577e-07,
      "loss": 3.1095,
      "step": 70245
    },
    {
      "epoch": 0.7146199544270834,
      "grad_norm": 9.315244674682617,
      "learning_rate": 9.41007533280092e-07,
      "loss": 3.2895,
      "step": 70250
    },
    {
      "epoch": 0.7146708170572916,
      "grad_norm": 11.047579765319824,
      "learning_rate": 9.40694947899278e-07,
      "loss": 3.4116,
      "step": 70255
    },
    {
      "epoch": 0.7147216796875,
      "grad_norm": 14.086073875427246,
      "learning_rate": 9.403824024129143e-07,
      "loss": 3.2052,
      "step": 70260
    },
    {
      "epoch": 0.7147725423177084,
      "grad_norm": 15.68326187133789,
      "learning_rate": 9.400698968289981e-07,
      "loss": 3.2755,
      "step": 70265
    },
    {
      "epoch": 0.7148234049479166,
      "grad_norm": 14.477516174316406,
      "learning_rate": 9.397574311555238e-07,
      "loss": 3.0787,
      "step": 70270
    },
    {
      "epoch": 0.714874267578125,
      "grad_norm": 11.304795265197754,
      "learning_rate": 9.39445005400485e-07,
      "loss": 3.401,
      "step": 70275
    },
    {
      "epoch": 0.7149251302083334,
      "grad_norm": 14.684220314025879,
      "learning_rate": 9.391326195718767e-07,
      "loss": 3.3938,
      "step": 70280
    },
    {
      "epoch": 0.7149759928385416,
      "grad_norm": 8.723044395446777,
      "learning_rate": 9.388202736776894e-07,
      "loss": 3.2873,
      "step": 70285
    },
    {
      "epoch": 0.71502685546875,
      "grad_norm": 10.842280387878418,
      "learning_rate": 9.385079677259165e-07,
      "loss": 3.4228,
      "step": 70290
    },
    {
      "epoch": 0.7150777180989584,
      "grad_norm": 14.38766860961914,
      "learning_rate": 9.381957017245461e-07,
      "loss": 3.2794,
      "step": 70295
    },
    {
      "epoch": 0.7151285807291666,
      "grad_norm": 14.24864387512207,
      "learning_rate": 9.378834756815691e-07,
      "loss": 3.4253,
      "step": 70300
    },
    {
      "epoch": 0.715179443359375,
      "grad_norm": 11.902130126953125,
      "learning_rate": 9.375712896049736e-07,
      "loss": 3.4232,
      "step": 70305
    },
    {
      "epoch": 0.7152303059895834,
      "grad_norm": 15.008284568786621,
      "learning_rate": 9.37259143502745e-07,
      "loss": 3.3055,
      "step": 70310
    },
    {
      "epoch": 0.7152811686197916,
      "grad_norm": 15.369800567626953,
      "learning_rate": 9.36947037382871e-07,
      "loss": 3.1573,
      "step": 70315
    },
    {
      "epoch": 0.71533203125,
      "grad_norm": 15.493510246276855,
      "learning_rate": 9.366349712533374e-07,
      "loss": 3.076,
      "step": 70320
    },
    {
      "epoch": 0.7153828938802084,
      "grad_norm": 9.27544116973877,
      "learning_rate": 9.363229451221273e-07,
      "loss": 3.9509,
      "step": 70325
    },
    {
      "epoch": 0.7154337565104166,
      "grad_norm": 14.310425758361816,
      "learning_rate": 9.360109589972236e-07,
      "loss": 3.301,
      "step": 70330
    },
    {
      "epoch": 0.715484619140625,
      "grad_norm": 14.848944664001465,
      "learning_rate": 9.356990128866092e-07,
      "loss": 3.0776,
      "step": 70335
    },
    {
      "epoch": 0.7155354817708334,
      "grad_norm": 11.941277503967285,
      "learning_rate": 9.353871067982642e-07,
      "loss": 3.4649,
      "step": 70340
    },
    {
      "epoch": 0.7155863444010416,
      "grad_norm": 14.305868148803711,
      "learning_rate": 9.3507524074017e-07,
      "loss": 3.9877,
      "step": 70345
    },
    {
      "epoch": 0.71563720703125,
      "grad_norm": 15.948720932006836,
      "learning_rate": 9.347634147203041e-07,
      "loss": 2.9864,
      "step": 70350
    },
    {
      "epoch": 0.7156880696614584,
      "grad_norm": 11.72199821472168,
      "learning_rate": 9.34451628746646e-07,
      "loss": 3.4454,
      "step": 70355
    },
    {
      "epoch": 0.7157389322916666,
      "grad_norm": 13.969541549682617,
      "learning_rate": 9.34139882827172e-07,
      "loss": 3.5413,
      "step": 70360
    },
    {
      "epoch": 0.715789794921875,
      "grad_norm": 8.047698974609375,
      "learning_rate": 9.338281769698567e-07,
      "loss": 3.3582,
      "step": 70365
    },
    {
      "epoch": 0.7158406575520834,
      "grad_norm": 12.490907669067383,
      "learning_rate": 9.335165111826766e-07,
      "loss": 3.4477,
      "step": 70370
    },
    {
      "epoch": 0.7158915201822916,
      "grad_norm": 11.708551406860352,
      "learning_rate": 9.332048854736061e-07,
      "loss": 3.2479,
      "step": 70375
    },
    {
      "epoch": 0.7159423828125,
      "grad_norm": 14.371440887451172,
      "learning_rate": 9.328932998506171e-07,
      "loss": 3.163,
      "step": 70380
    },
    {
      "epoch": 0.7159932454427084,
      "grad_norm": 11.833093643188477,
      "learning_rate": 9.325817543216817e-07,
      "loss": 3.6719,
      "step": 70385
    },
    {
      "epoch": 0.7160441080729166,
      "grad_norm": 12.058566093444824,
      "learning_rate": 9.322702488947699e-07,
      "loss": 3.4186,
      "step": 70390
    },
    {
      "epoch": 0.716094970703125,
      "grad_norm": 11.052045822143555,
      "learning_rate": 9.31958783577852e-07,
      "loss": 3.2209,
      "step": 70395
    },
    {
      "epoch": 0.7161458333333334,
      "grad_norm": 14.906253814697266,
      "learning_rate": 9.316473583788976e-07,
      "loss": 3.222,
      "step": 70400
    },
    {
      "epoch": 0.7161966959635416,
      "grad_norm": 8.230878829956055,
      "learning_rate": 9.313359733058733e-07,
      "loss": 3.0487,
      "step": 70405
    },
    {
      "epoch": 0.71624755859375,
      "grad_norm": 12.070849418640137,
      "learning_rate": 9.310246283667471e-07,
      "loss": 3.447,
      "step": 70410
    },
    {
      "epoch": 0.7162984212239584,
      "grad_norm": 10.283710479736328,
      "learning_rate": 9.307133235694829e-07,
      "loss": 3.5644,
      "step": 70415
    },
    {
      "epoch": 0.7163492838541666,
      "grad_norm": 10.542003631591797,
      "learning_rate": 9.304020589220472e-07,
      "loss": 3.3768,
      "step": 70420
    },
    {
      "epoch": 0.716400146484375,
      "grad_norm": 8.863672256469727,
      "learning_rate": 9.300908344324023e-07,
      "loss": 3.4833,
      "step": 70425
    },
    {
      "epoch": 0.7164510091145834,
      "grad_norm": 8.094013214111328,
      "learning_rate": 9.297796501085115e-07,
      "loss": 3.1542,
      "step": 70430
    },
    {
      "epoch": 0.7165018717447916,
      "grad_norm": 14.92911148071289,
      "learning_rate": 9.294685059583366e-07,
      "loss": 3.3276,
      "step": 70435
    },
    {
      "epoch": 0.716552734375,
      "grad_norm": 11.061050415039062,
      "learning_rate": 9.29157401989837e-07,
      "loss": 3.2169,
      "step": 70440
    },
    {
      "epoch": 0.7166035970052084,
      "grad_norm": 7.26305627822876,
      "learning_rate": 9.288463382109728e-07,
      "loss": 3.5735,
      "step": 70445
    },
    {
      "epoch": 0.7166544596354166,
      "grad_norm": 12.462404251098633,
      "learning_rate": 9.285353146297032e-07,
      "loss": 3.4802,
      "step": 70450
    },
    {
      "epoch": 0.716705322265625,
      "grad_norm": 10.22716236114502,
      "learning_rate": 9.282243312539854e-07,
      "loss": 3.0933,
      "step": 70455
    },
    {
      "epoch": 0.7167561848958334,
      "grad_norm": 12.187172889709473,
      "learning_rate": 9.279133880917754e-07,
      "loss": 2.952,
      "step": 70460
    },
    {
      "epoch": 0.7168070475260416,
      "grad_norm": 10.628868103027344,
      "learning_rate": 9.276024851510279e-07,
      "loss": 3.1059,
      "step": 70465
    },
    {
      "epoch": 0.71685791015625,
      "grad_norm": 17.583450317382812,
      "learning_rate": 9.272916224396983e-07,
      "loss": 3.4054,
      "step": 70470
    },
    {
      "epoch": 0.7169087727864584,
      "grad_norm": 9.600113868713379,
      "learning_rate": 9.269807999657404e-07,
      "loss": 3.4388,
      "step": 70475
    },
    {
      "epoch": 0.7169596354166666,
      "grad_norm": 12.207353591918945,
      "learning_rate": 9.26670017737106e-07,
      "loss": 3.2619,
      "step": 70480
    },
    {
      "epoch": 0.717010498046875,
      "grad_norm": 9.641359329223633,
      "learning_rate": 9.263592757617454e-07,
      "loss": 3.4046,
      "step": 70485
    },
    {
      "epoch": 0.7170613606770834,
      "grad_norm": 8.686779975891113,
      "learning_rate": 9.260485740476105e-07,
      "loss": 3.4551,
      "step": 70490
    },
    {
      "epoch": 0.7171122233072916,
      "grad_norm": 12.74880599975586,
      "learning_rate": 9.257379126026489e-07,
      "loss": 3.298,
      "step": 70495
    },
    {
      "epoch": 0.7171630859375,
      "grad_norm": 10.921481132507324,
      "learning_rate": 9.254272914348097e-07,
      "loss": 3.3536,
      "step": 70500
    },
    {
      "epoch": 0.7172139485677084,
      "grad_norm": 15.362860679626465,
      "learning_rate": 9.251167105520411e-07,
      "loss": 3.4412,
      "step": 70505
    },
    {
      "epoch": 0.7172648111979166,
      "grad_norm": 8.349514961242676,
      "learning_rate": 9.248061699622879e-07,
      "loss": 3.3569,
      "step": 70510
    },
    {
      "epoch": 0.717315673828125,
      "grad_norm": 10.782577514648438,
      "learning_rate": 9.244956696734958e-07,
      "loss": 3.4511,
      "step": 70515
    },
    {
      "epoch": 0.7173665364583334,
      "grad_norm": 8.114956855773926,
      "learning_rate": 9.241852096936074e-07,
      "loss": 3.1312,
      "step": 70520
    },
    {
      "epoch": 0.7174173990885416,
      "grad_norm": 8.276371002197266,
      "learning_rate": 9.23874790030567e-07,
      "loss": 3.3773,
      "step": 70525
    },
    {
      "epoch": 0.71746826171875,
      "grad_norm": 12.949410438537598,
      "learning_rate": 9.235644106923172e-07,
      "loss": 3.3625,
      "step": 70530
    },
    {
      "epoch": 0.7175191243489584,
      "grad_norm": 7.578267574310303,
      "learning_rate": 9.232540716867985e-07,
      "loss": 3.3437,
      "step": 70535
    },
    {
      "epoch": 0.7175699869791666,
      "grad_norm": 15.299179077148438,
      "learning_rate": 9.229437730219499e-07,
      "loss": 2.824,
      "step": 70540
    },
    {
      "epoch": 0.717620849609375,
      "grad_norm": 10.61105728149414,
      "learning_rate": 9.226335147057116e-07,
      "loss": 3.0508,
      "step": 70545
    },
    {
      "epoch": 0.7176717122395834,
      "grad_norm": 10.850451469421387,
      "learning_rate": 9.223232967460202e-07,
      "loss": 3.3547,
      "step": 70550
    },
    {
      "epoch": 0.7177225748697916,
      "grad_norm": 11.77407169342041,
      "learning_rate": 9.220131191508144e-07,
      "loss": 3.3201,
      "step": 70555
    },
    {
      "epoch": 0.7177734375,
      "grad_norm": 13.269412994384766,
      "learning_rate": 9.217029819280279e-07,
      "loss": 3.5001,
      "step": 70560
    },
    {
      "epoch": 0.7178243001302084,
      "grad_norm": 8.466712951660156,
      "learning_rate": 9.213928850855974e-07,
      "loss": 3.3259,
      "step": 70565
    },
    {
      "epoch": 0.7178751627604166,
      "grad_norm": 11.160103797912598,
      "learning_rate": 9.210828286314546e-07,
      "loss": 3.5799,
      "step": 70570
    },
    {
      "epoch": 0.717926025390625,
      "grad_norm": 8.970287322998047,
      "learning_rate": 9.207728125735346e-07,
      "loss": 3.4847,
      "step": 70575
    },
    {
      "epoch": 0.7179768880208334,
      "grad_norm": 13.498138427734375,
      "learning_rate": 9.204628369197671e-07,
      "loss": 3.1112,
      "step": 70580
    },
    {
      "epoch": 0.7180277506510416,
      "grad_norm": 9.72287368774414,
      "learning_rate": 9.201529016780841e-07,
      "loss": 3.7581,
      "step": 70585
    },
    {
      "epoch": 0.71807861328125,
      "grad_norm": 10.230852127075195,
      "learning_rate": 9.198430068564149e-07,
      "loss": 3.4474,
      "step": 70590
    },
    {
      "epoch": 0.7181294759114584,
      "grad_norm": 12.656306266784668,
      "learning_rate": 9.195331524626866e-07,
      "loss": 3.4458,
      "step": 70595
    },
    {
      "epoch": 0.7181803385416666,
      "grad_norm": 16.775386810302734,
      "learning_rate": 9.192233385048282e-07,
      "loss": 3.7704,
      "step": 70600
    },
    {
      "epoch": 0.718231201171875,
      "grad_norm": 11.617079734802246,
      "learning_rate": 9.189135649907668e-07,
      "loss": 3.426,
      "step": 70605
    },
    {
      "epoch": 0.7182820638020834,
      "grad_norm": 6.835216522216797,
      "learning_rate": 9.186038319284274e-07,
      "loss": 3.1763,
      "step": 70610
    },
    {
      "epoch": 0.7183329264322916,
      "grad_norm": 10.226903915405273,
      "learning_rate": 9.182941393257331e-07,
      "loss": 3.8316,
      "step": 70615
    },
    {
      "epoch": 0.7183837890625,
      "grad_norm": 10.749029159545898,
      "learning_rate": 9.179844871906088e-07,
      "loss": 2.974,
      "step": 70620
    },
    {
      "epoch": 0.7184346516927084,
      "grad_norm": 7.657063961029053,
      "learning_rate": 9.176748755309761e-07,
      "loss": 3.255,
      "step": 70625
    },
    {
      "epoch": 0.7184855143229166,
      "grad_norm": 15.790289878845215,
      "learning_rate": 9.173653043547573e-07,
      "loss": 3.3691,
      "step": 70630
    },
    {
      "epoch": 0.718536376953125,
      "grad_norm": 8.744561195373535,
      "learning_rate": 9.17055773669871e-07,
      "loss": 3.3656,
      "step": 70635
    },
    {
      "epoch": 0.7185872395833334,
      "grad_norm": 9.560819625854492,
      "learning_rate": 9.167462834842386e-07,
      "loss": 3.0058,
      "step": 70640
    },
    {
      "epoch": 0.7186381022135416,
      "grad_norm": 16.007478713989258,
      "learning_rate": 9.164368338057772e-07,
      "loss": 3.3507,
      "step": 70645
    },
    {
      "epoch": 0.71868896484375,
      "grad_norm": 10.624584197998047,
      "learning_rate": 9.161274246424034e-07,
      "loss": 3.6953,
      "step": 70650
    },
    {
      "epoch": 0.7187398274739584,
      "grad_norm": 15.273704528808594,
      "learning_rate": 9.158180560020339e-07,
      "loss": 3.5073,
      "step": 70655
    },
    {
      "epoch": 0.7187906901041666,
      "grad_norm": 13.692089080810547,
      "learning_rate": 9.155087278925847e-07,
      "loss": 3.5895,
      "step": 70660
    },
    {
      "epoch": 0.718841552734375,
      "grad_norm": 11.459607124328613,
      "learning_rate": 9.151994403219691e-07,
      "loss": 3.5057,
      "step": 70665
    },
    {
      "epoch": 0.7188924153645834,
      "grad_norm": 14.022200584411621,
      "learning_rate": 9.148901932981002e-07,
      "loss": 4.0547,
      "step": 70670
    },
    {
      "epoch": 0.7189432779947916,
      "grad_norm": 10.06373405456543,
      "learning_rate": 9.145809868288893e-07,
      "loss": 3.3864,
      "step": 70675
    },
    {
      "epoch": 0.718994140625,
      "grad_norm": 13.52616024017334,
      "learning_rate": 9.142718209222481e-07,
      "loss": 3.4282,
      "step": 70680
    },
    {
      "epoch": 0.7190450032552084,
      "grad_norm": 13.549678802490234,
      "learning_rate": 9.139626955860872e-07,
      "loss": 3.1799,
      "step": 70685
    },
    {
      "epoch": 0.7190958658854166,
      "grad_norm": 11.853752136230469,
      "learning_rate": 9.136536108283148e-07,
      "loss": 3.3303,
      "step": 70690
    },
    {
      "epoch": 0.719146728515625,
      "grad_norm": 10.186666488647461,
      "learning_rate": 9.13344566656838e-07,
      "loss": 3.5477,
      "step": 70695
    },
    {
      "epoch": 0.7191975911458334,
      "grad_norm": 12.17104434967041,
      "learning_rate": 9.130355630795651e-07,
      "loss": 3.476,
      "step": 70700
    },
    {
      "epoch": 0.7192484537760416,
      "grad_norm": 9.603554725646973,
      "learning_rate": 9.127266001044005e-07,
      "loss": 3.5696,
      "step": 70705
    },
    {
      "epoch": 0.71929931640625,
      "grad_norm": 11.729199409484863,
      "learning_rate": 9.124176777392496e-07,
      "loss": 2.8525,
      "step": 70710
    },
    {
      "epoch": 0.7193501790364584,
      "grad_norm": 12.96235179901123,
      "learning_rate": 9.121087959920169e-07,
      "loss": 3.3426,
      "step": 70715
    },
    {
      "epoch": 0.7194010416666666,
      "grad_norm": 11.475220680236816,
      "learning_rate": 9.117999548706041e-07,
      "loss": 3.2288,
      "step": 70720
    },
    {
      "epoch": 0.719451904296875,
      "grad_norm": 8.891611099243164,
      "learning_rate": 9.114911543829124e-07,
      "loss": 3.2626,
      "step": 70725
    },
    {
      "epoch": 0.7195027669270834,
      "grad_norm": 15.098640441894531,
      "learning_rate": 9.111823945368429e-07,
      "loss": 3.3544,
      "step": 70730
    },
    {
      "epoch": 0.7195536295572916,
      "grad_norm": 10.538352966308594,
      "learning_rate": 9.108736753402961e-07,
      "loss": 3.3497,
      "step": 70735
    },
    {
      "epoch": 0.7196044921875,
      "grad_norm": 10.955540657043457,
      "learning_rate": 9.105649968011698e-07,
      "loss": 3.3903,
      "step": 70740
    },
    {
      "epoch": 0.7196553548177084,
      "grad_norm": 14.484964370727539,
      "learning_rate": 9.102563589273611e-07,
      "loss": 3.3029,
      "step": 70745
    },
    {
      "epoch": 0.7197062174479166,
      "grad_norm": 9.111550331115723,
      "learning_rate": 9.09947761726766e-07,
      "loss": 3.1651,
      "step": 70750
    },
    {
      "epoch": 0.719757080078125,
      "grad_norm": 16.72936248779297,
      "learning_rate": 9.096392052072803e-07,
      "loss": 3.2327,
      "step": 70755
    },
    {
      "epoch": 0.7198079427083334,
      "grad_norm": 9.58584976196289,
      "learning_rate": 9.093306893767995e-07,
      "loss": 3.1563,
      "step": 70760
    },
    {
      "epoch": 0.7198588053385416,
      "grad_norm": 8.176050186157227,
      "learning_rate": 9.090222142432162e-07,
      "loss": 3.5646,
      "step": 70765
    },
    {
      "epoch": 0.71990966796875,
      "grad_norm": 14.241776466369629,
      "learning_rate": 9.087137798144213e-07,
      "loss": 3.4337,
      "step": 70770
    },
    {
      "epoch": 0.7199605305989584,
      "grad_norm": 12.04594612121582,
      "learning_rate": 9.084053860983083e-07,
      "loss": 3.3614,
      "step": 70775
    },
    {
      "epoch": 0.7200113932291666,
      "grad_norm": 8.9103364944458,
      "learning_rate": 9.080970331027653e-07,
      "loss": 3.1215,
      "step": 70780
    },
    {
      "epoch": 0.720062255859375,
      "grad_norm": 13.713004112243652,
      "learning_rate": 9.077887208356831e-07,
      "loss": 3.6188,
      "step": 70785
    },
    {
      "epoch": 0.7201131184895834,
      "grad_norm": 13.118133544921875,
      "learning_rate": 9.074804493049483e-07,
      "loss": 3.3569,
      "step": 70790
    },
    {
      "epoch": 0.7201639811197916,
      "grad_norm": 9.286763191223145,
      "learning_rate": 9.071722185184495e-07,
      "loss": 3.3846,
      "step": 70795
    },
    {
      "epoch": 0.72021484375,
      "grad_norm": 10.336640357971191,
      "learning_rate": 9.068640284840721e-07,
      "loss": 3.2238,
      "step": 70800
    },
    {
      "epoch": 0.7202657063802084,
      "grad_norm": 13.235795021057129,
      "learning_rate": 9.065558792096998e-07,
      "loss": 3.329,
      "step": 70805
    },
    {
      "epoch": 0.7203165690104166,
      "grad_norm": 17.055892944335938,
      "learning_rate": 9.06247770703218e-07,
      "loss": 3.4045,
      "step": 70810
    },
    {
      "epoch": 0.720367431640625,
      "grad_norm": 13.522333145141602,
      "learning_rate": 9.059397029725101e-07,
      "loss": 3.5872,
      "step": 70815
    },
    {
      "epoch": 0.7204182942708334,
      "grad_norm": 12.300740242004395,
      "learning_rate": 9.056316760254569e-07,
      "loss": 3.1478,
      "step": 70820
    },
    {
      "epoch": 0.7204691569010416,
      "grad_norm": 11.573324203491211,
      "learning_rate": 9.053236898699383e-07,
      "loss": 3.2798,
      "step": 70825
    },
    {
      "epoch": 0.72052001953125,
      "grad_norm": 11.538972854614258,
      "learning_rate": 9.050157445138363e-07,
      "loss": 3.284,
      "step": 70830
    },
    {
      "epoch": 0.7205708821614584,
      "grad_norm": 11.63953971862793,
      "learning_rate": 9.047078399650274e-07,
      "loss": 3.1113,
      "step": 70835
    },
    {
      "epoch": 0.7206217447916666,
      "grad_norm": 13.707157135009766,
      "learning_rate": 9.043999762313913e-07,
      "loss": 3.1204,
      "step": 70840
    },
    {
      "epoch": 0.720672607421875,
      "grad_norm": 13.982678413391113,
      "learning_rate": 9.040921533208025e-07,
      "loss": 3.3069,
      "step": 70845
    },
    {
      "epoch": 0.7207234700520834,
      "grad_norm": 11.639885902404785,
      "learning_rate": 9.037843712411387e-07,
      "loss": 3.2747,
      "step": 70850
    },
    {
      "epoch": 0.7207743326822916,
      "grad_norm": 12.32170295715332,
      "learning_rate": 9.034766300002734e-07,
      "loss": 3.5874,
      "step": 70855
    },
    {
      "epoch": 0.7208251953125,
      "grad_norm": 12.784284591674805,
      "learning_rate": 9.031689296060791e-07,
      "loss": 3.4228,
      "step": 70860
    },
    {
      "epoch": 0.7208760579427084,
      "grad_norm": 14.382755279541016,
      "learning_rate": 9.028612700664294e-07,
      "loss": 2.9878,
      "step": 70865
    },
    {
      "epoch": 0.7209269205729166,
      "grad_norm": 16.06764793395996,
      "learning_rate": 9.025536513891963e-07,
      "loss": 3.5907,
      "step": 70870
    },
    {
      "epoch": 0.720977783203125,
      "grad_norm": 13.783523559570312,
      "learning_rate": 9.022460735822497e-07,
      "loss": 3.1087,
      "step": 70875
    },
    {
      "epoch": 0.7210286458333334,
      "grad_norm": 7.396638870239258,
      "learning_rate": 9.019385366534575e-07,
      "loss": 3.3846,
      "step": 70880
    },
    {
      "epoch": 0.7210795084635416,
      "grad_norm": 9.226949691772461,
      "learning_rate": 9.01631040610689e-07,
      "loss": 3.1187,
      "step": 70885
    },
    {
      "epoch": 0.72113037109375,
      "grad_norm": 9.044224739074707,
      "learning_rate": 9.013235854618124e-07,
      "loss": 3.422,
      "step": 70890
    },
    {
      "epoch": 0.7211812337239584,
      "grad_norm": 10.633951187133789,
      "learning_rate": 9.010161712146931e-07,
      "loss": 3.2628,
      "step": 70895
    },
    {
      "epoch": 0.7212320963541666,
      "grad_norm": 9.115630149841309,
      "learning_rate": 9.00708797877195e-07,
      "loss": 3.3975,
      "step": 70900
    },
    {
      "epoch": 0.721282958984375,
      "grad_norm": 15.742044448852539,
      "learning_rate": 9.004014654571841e-07,
      "loss": 3.5155,
      "step": 70905
    },
    {
      "epoch": 0.7213338216145834,
      "grad_norm": 11.0806245803833,
      "learning_rate": 9.000941739625219e-07,
      "loss": 3.4679,
      "step": 70910
    },
    {
      "epoch": 0.7213846842447916,
      "grad_norm": 12.643223762512207,
      "learning_rate": 8.997869234010717e-07,
      "loss": 3.729,
      "step": 70915
    },
    {
      "epoch": 0.721435546875,
      "grad_norm": 11.004573822021484,
      "learning_rate": 8.994797137806929e-07,
      "loss": 3.3299,
      "step": 70920
    },
    {
      "epoch": 0.7214864095052084,
      "grad_norm": 16.230148315429688,
      "learning_rate": 8.991725451092473e-07,
      "loss": 3.6769,
      "step": 70925
    },
    {
      "epoch": 0.7215372721354166,
      "grad_norm": 11.277342796325684,
      "learning_rate": 8.988654173945924e-07,
      "loss": 3.6651,
      "step": 70930
    },
    {
      "epoch": 0.721588134765625,
      "grad_norm": 10.552614212036133,
      "learning_rate": 8.985583306445855e-07,
      "loss": 3.4175,
      "step": 70935
    },
    {
      "epoch": 0.7216389973958334,
      "grad_norm": 9.672837257385254,
      "learning_rate": 8.982512848670844e-07,
      "loss": 3.3217,
      "step": 70940
    },
    {
      "epoch": 0.7216898600260416,
      "grad_norm": 16.051069259643555,
      "learning_rate": 8.97944280069945e-07,
      "loss": 3.6637,
      "step": 70945
    },
    {
      "epoch": 0.72174072265625,
      "grad_norm": 13.610672950744629,
      "learning_rate": 8.976373162610219e-07,
      "loss": 3.4613,
      "step": 70950
    },
    {
      "epoch": 0.7217915852864584,
      "grad_norm": 11.690978050231934,
      "learning_rate": 8.973303934481681e-07,
      "loss": 3.5566,
      "step": 70955
    },
    {
      "epoch": 0.7218424479166666,
      "grad_norm": 15.141119003295898,
      "learning_rate": 8.970235116392353e-07,
      "loss": 3.5221,
      "step": 70960
    },
    {
      "epoch": 0.721893310546875,
      "grad_norm": 13.764825820922852,
      "learning_rate": 8.96716670842076e-07,
      "loss": 3.721,
      "step": 70965
    },
    {
      "epoch": 0.7219441731770834,
      "grad_norm": 11.916561126708984,
      "learning_rate": 8.964098710645416e-07,
      "loss": 3.7082,
      "step": 70970
    },
    {
      "epoch": 0.7219950358072916,
      "grad_norm": 11.584885597229004,
      "learning_rate": 8.961031123144806e-07,
      "loss": 3.3698,
      "step": 70975
    },
    {
      "epoch": 0.7220458984375,
      "grad_norm": 11.479866981506348,
      "learning_rate": 8.957963945997403e-07,
      "loss": 3.2962,
      "step": 70980
    },
    {
      "epoch": 0.7220967610677084,
      "grad_norm": 13.004188537597656,
      "learning_rate": 8.954897179281701e-07,
      "loss": 3.2781,
      "step": 70985
    },
    {
      "epoch": 0.7221476236979166,
      "grad_norm": 11.68873405456543,
      "learning_rate": 8.951830823076141e-07,
      "loss": 3.2864,
      "step": 70990
    },
    {
      "epoch": 0.722198486328125,
      "grad_norm": 13.482848167419434,
      "learning_rate": 8.948764877459188e-07,
      "loss": 3.3523,
      "step": 70995
    },
    {
      "epoch": 0.7222493489583334,
      "grad_norm": 13.140082359313965,
      "learning_rate": 8.945699342509287e-07,
      "loss": 3.0924,
      "step": 71000
    },
    {
      "epoch": 0.7223002115885416,
      "grad_norm": 11.259969711303711,
      "learning_rate": 8.942634218304863e-07,
      "loss": 3.3819,
      "step": 71005
    },
    {
      "epoch": 0.72235107421875,
      "grad_norm": 11.4698486328125,
      "learning_rate": 8.939569504924339e-07,
      "loss": 3.2888,
      "step": 71010
    },
    {
      "epoch": 0.7224019368489584,
      "grad_norm": 11.579285621643066,
      "learning_rate": 8.936505202446102e-07,
      "loss": 3.438,
      "step": 71015
    },
    {
      "epoch": 0.7224527994791666,
      "grad_norm": 15.198299407958984,
      "learning_rate": 8.933441310948593e-07,
      "loss": 3.554,
      "step": 71020
    },
    {
      "epoch": 0.722503662109375,
      "grad_norm": 15.81182861328125,
      "learning_rate": 8.930377830510175e-07,
      "loss": 3.6142,
      "step": 71025
    },
    {
      "epoch": 0.7225545247395834,
      "grad_norm": 9.641409873962402,
      "learning_rate": 8.927314761209236e-07,
      "loss": 3.4682,
      "step": 71030
    },
    {
      "epoch": 0.7226053873697916,
      "grad_norm": 15.797674179077148,
      "learning_rate": 8.924252103124128e-07,
      "loss": 3.1733,
      "step": 71035
    },
    {
      "epoch": 0.72265625,
      "grad_norm": 11.661063194274902,
      "learning_rate": 8.92118985633322e-07,
      "loss": 3.2641,
      "step": 71040
    },
    {
      "epoch": 0.7227071126302084,
      "grad_norm": 8.455240249633789,
      "learning_rate": 8.918128020914868e-07,
      "loss": 3.2229,
      "step": 71045
    },
    {
      "epoch": 0.7227579752604166,
      "grad_norm": 11.377432823181152,
      "learning_rate": 8.915066596947397e-07,
      "loss": 3.899,
      "step": 71050
    },
    {
      "epoch": 0.722808837890625,
      "grad_norm": 9.175016403198242,
      "learning_rate": 8.912005584509129e-07,
      "loss": 3.3356,
      "step": 71055
    },
    {
      "epoch": 0.7228597005208334,
      "grad_norm": 10.886006355285645,
      "learning_rate": 8.908944983678388e-07,
      "loss": 3.4519,
      "step": 71060
    },
    {
      "epoch": 0.7229105631510416,
      "grad_norm": 13.152046203613281,
      "learning_rate": 8.905884794533473e-07,
      "loss": 3.2553,
      "step": 71065
    },
    {
      "epoch": 0.72296142578125,
      "grad_norm": 7.559975624084473,
      "learning_rate": 8.902825017152686e-07,
      "loss": 3.901,
      "step": 71070
    },
    {
      "epoch": 0.7230122884114584,
      "grad_norm": 18.686050415039062,
      "learning_rate": 8.899765651614297e-07,
      "loss": 3.6899,
      "step": 71075
    },
    {
      "epoch": 0.7230631510416666,
      "grad_norm": 15.401456832885742,
      "learning_rate": 8.896706697996599e-07,
      "loss": 3.7532,
      "step": 71080
    },
    {
      "epoch": 0.723114013671875,
      "grad_norm": 15.929218292236328,
      "learning_rate": 8.89364815637784e-07,
      "loss": 3.376,
      "step": 71085
    },
    {
      "epoch": 0.7231648763020834,
      "grad_norm": 8.313657760620117,
      "learning_rate": 8.890590026836269e-07,
      "loss": 3.1655,
      "step": 71090
    },
    {
      "epoch": 0.7232157389322916,
      "grad_norm": 8.160225868225098,
      "learning_rate": 8.887532309450131e-07,
      "loss": 3.1708,
      "step": 71095
    },
    {
      "epoch": 0.7232666015625,
      "grad_norm": 13.99388599395752,
      "learning_rate": 8.884475004297671e-07,
      "loss": 3.6068,
      "step": 71100
    },
    {
      "epoch": 0.7233174641927084,
      "grad_norm": 10.984503746032715,
      "learning_rate": 8.881418111457096e-07,
      "loss": 3.312,
      "step": 71105
    },
    {
      "epoch": 0.7233683268229166,
      "grad_norm": 14.216525077819824,
      "learning_rate": 8.878361631006613e-07,
      "loss": 3.6005,
      "step": 71110
    },
    {
      "epoch": 0.723419189453125,
      "grad_norm": 9.797622680664062,
      "learning_rate": 8.87530556302443e-07,
      "loss": 3.1769,
      "step": 71115
    },
    {
      "epoch": 0.7234700520833334,
      "grad_norm": 14.703811645507812,
      "learning_rate": 8.872249907588729e-07,
      "loss": 3.3396,
      "step": 71120
    },
    {
      "epoch": 0.7235209147135416,
      "grad_norm": 8.875903129577637,
      "learning_rate": 8.869194664777697e-07,
      "loss": 3.1241,
      "step": 71125
    },
    {
      "epoch": 0.72357177734375,
      "grad_norm": 10.04088306427002,
      "learning_rate": 8.866139834669487e-07,
      "loss": 3.2426,
      "step": 71130
    },
    {
      "epoch": 0.7236226399739584,
      "grad_norm": 15.39554214477539,
      "learning_rate": 8.863085417342274e-07,
      "loss": 3.4198,
      "step": 71135
    },
    {
      "epoch": 0.7236735026041666,
      "grad_norm": 15.000868797302246,
      "learning_rate": 8.860031412874198e-07,
      "loss": 3.1849,
      "step": 71140
    },
    {
      "epoch": 0.723724365234375,
      "grad_norm": 9.475614547729492,
      "learning_rate": 8.856977821343382e-07,
      "loss": 3.2855,
      "step": 71145
    },
    {
      "epoch": 0.7237752278645834,
      "grad_norm": 13.789505958557129,
      "learning_rate": 8.853924642827963e-07,
      "loss": 3.2375,
      "step": 71150
    },
    {
      "epoch": 0.7238260904947916,
      "grad_norm": 16.8479061126709,
      "learning_rate": 8.850871877406061e-07,
      "loss": 3.2356,
      "step": 71155
    },
    {
      "epoch": 0.723876953125,
      "grad_norm": 13.068145751953125,
      "learning_rate": 8.847819525155779e-07,
      "loss": 3.2793,
      "step": 71160
    },
    {
      "epoch": 0.7239278157552084,
      "grad_norm": 9.874674797058105,
      "learning_rate": 8.844767586155203e-07,
      "loss": 3.0541,
      "step": 71165
    },
    {
      "epoch": 0.7239786783854166,
      "grad_norm": 12.130133628845215,
      "learning_rate": 8.841716060482403e-07,
      "loss": 3.6288,
      "step": 71170
    },
    {
      "epoch": 0.724029541015625,
      "grad_norm": 9.09591293334961,
      "learning_rate": 8.838664948215484e-07,
      "loss": 3.3846,
      "step": 71175
    },
    {
      "epoch": 0.7240804036458334,
      "grad_norm": 11.738776206970215,
      "learning_rate": 8.835614249432491e-07,
      "loss": 3.2677,
      "step": 71180
    },
    {
      "epoch": 0.7241312662760416,
      "grad_norm": 13.365427017211914,
      "learning_rate": 8.832563964211477e-07,
      "loss": 3.0518,
      "step": 71185
    },
    {
      "epoch": 0.72418212890625,
      "grad_norm": 12.114652633666992,
      "learning_rate": 8.829514092630473e-07,
      "loss": 3.125,
      "step": 71190
    },
    {
      "epoch": 0.7242329915364584,
      "grad_norm": 11.502225875854492,
      "learning_rate": 8.826464634767518e-07,
      "loss": 3.5321,
      "step": 71195
    },
    {
      "epoch": 0.7242838541666666,
      "grad_norm": 12.675405502319336,
      "learning_rate": 8.82341559070064e-07,
      "loss": 3.5959,
      "step": 71200
    },
    {
      "epoch": 0.724334716796875,
      "grad_norm": 9.8812837600708,
      "learning_rate": 8.820366960507834e-07,
      "loss": 3.4001,
      "step": 71205
    },
    {
      "epoch": 0.7243855794270834,
      "grad_norm": 7.821342945098877,
      "learning_rate": 8.817318744267109e-07,
      "loss": 3.0264,
      "step": 71210
    },
    {
      "epoch": 0.7244364420572916,
      "grad_norm": 14.114405632019043,
      "learning_rate": 8.81427094205645e-07,
      "loss": 3.5391,
      "step": 71215
    },
    {
      "epoch": 0.7244873046875,
      "grad_norm": 10.161025047302246,
      "learning_rate": 8.811223553953824e-07,
      "loss": 3.1369,
      "step": 71220
    },
    {
      "epoch": 0.7245381673177084,
      "grad_norm": 11.592391014099121,
      "learning_rate": 8.808176580037206e-07,
      "loss": 3.4189,
      "step": 71225
    },
    {
      "epoch": 0.7245890299479166,
      "grad_norm": 10.845025062561035,
      "learning_rate": 8.805130020384562e-07,
      "loss": 3.2114,
      "step": 71230
    },
    {
      "epoch": 0.724639892578125,
      "grad_norm": 12.971354484558105,
      "learning_rate": 8.802083875073828e-07,
      "loss": 3.2945,
      "step": 71235
    },
    {
      "epoch": 0.7246907552083334,
      "grad_norm": 15.216636657714844,
      "learning_rate": 8.799038144182937e-07,
      "loss": 3.3843,
      "step": 71240
    },
    {
      "epoch": 0.7247416178385416,
      "grad_norm": 9.582465171813965,
      "learning_rate": 8.795992827789807e-07,
      "loss": 3.0785,
      "step": 71245
    },
    {
      "epoch": 0.72479248046875,
      "grad_norm": 12.177659034729004,
      "learning_rate": 8.792947925972359e-07,
      "loss": 3.331,
      "step": 71250
    },
    {
      "epoch": 0.7248433430989584,
      "grad_norm": 7.878180503845215,
      "learning_rate": 8.789903438808506e-07,
      "loss": 3.693,
      "step": 71255
    },
    {
      "epoch": 0.7248942057291666,
      "grad_norm": 14.96727180480957,
      "learning_rate": 8.786859366376132e-07,
      "loss": 3.4858,
      "step": 71260
    },
    {
      "epoch": 0.724945068359375,
      "grad_norm": 14.713326454162598,
      "learning_rate": 8.78381570875311e-07,
      "loss": 3.2536,
      "step": 71265
    },
    {
      "epoch": 0.7249959309895834,
      "grad_norm": 16.95229148864746,
      "learning_rate": 8.780772466017326e-07,
      "loss": 3.234,
      "step": 71270
    },
    {
      "epoch": 0.7250467936197916,
      "grad_norm": 14.073349952697754,
      "learning_rate": 8.777729638246627e-07,
      "loss": 3.2735,
      "step": 71275
    },
    {
      "epoch": 0.72509765625,
      "grad_norm": 14.336599349975586,
      "learning_rate": 8.774687225518877e-07,
      "loss": 3.3153,
      "step": 71280
    },
    {
      "epoch": 0.7251485188802084,
      "grad_norm": 10.15825080871582,
      "learning_rate": 8.771645227911902e-07,
      "loss": 3.2011,
      "step": 71285
    },
    {
      "epoch": 0.7251993815104166,
      "grad_norm": 14.921940803527832,
      "learning_rate": 8.768603645503546e-07,
      "loss": 3.5299,
      "step": 71290
    },
    {
      "epoch": 0.725250244140625,
      "grad_norm": 9.78284740447998,
      "learning_rate": 8.765562478371617e-07,
      "loss": 3.2637,
      "step": 71295
    },
    {
      "epoch": 0.7253011067708334,
      "grad_norm": 14.756291389465332,
      "learning_rate": 8.762521726593918e-07,
      "loss": 3.0072,
      "step": 71300
    },
    {
      "epoch": 0.7253519694010416,
      "grad_norm": 15.028759956359863,
      "learning_rate": 8.759481390248248e-07,
      "loss": 3.5329,
      "step": 71305
    },
    {
      "epoch": 0.72540283203125,
      "grad_norm": 9.425098419189453,
      "learning_rate": 8.756441469412408e-07,
      "loss": 3.5755,
      "step": 71310
    },
    {
      "epoch": 0.7254536946614584,
      "grad_norm": 16.844207763671875,
      "learning_rate": 8.753401964164162e-07,
      "loss": 3.503,
      "step": 71315
    },
    {
      "epoch": 0.7255045572916666,
      "grad_norm": 15.11011791229248,
      "learning_rate": 8.75036287458127e-07,
      "loss": 3.2334,
      "step": 71320
    },
    {
      "epoch": 0.725555419921875,
      "grad_norm": 13.918288230895996,
      "learning_rate": 8.747324200741489e-07,
      "loss": 3.2216,
      "step": 71325
    },
    {
      "epoch": 0.7256062825520834,
      "grad_norm": 7.063535690307617,
      "learning_rate": 8.744285942722575e-07,
      "loss": 3.2819,
      "step": 71330
    },
    {
      "epoch": 0.7256571451822916,
      "grad_norm": 7.297150135040283,
      "learning_rate": 8.741248100602253e-07,
      "loss": 3.2245,
      "step": 71335
    },
    {
      "epoch": 0.7257080078125,
      "grad_norm": 12.161453247070312,
      "learning_rate": 8.738210674458236e-07,
      "loss": 3.5233,
      "step": 71340
    },
    {
      "epoch": 0.7257588704427084,
      "grad_norm": 17.07378578186035,
      "learning_rate": 8.735173664368251e-07,
      "loss": 3.1754,
      "step": 71345
    },
    {
      "epoch": 0.7258097330729166,
      "grad_norm": 12.775858879089355,
      "learning_rate": 8.732137070409987e-07,
      "loss": 3.5189,
      "step": 71350
    },
    {
      "epoch": 0.725860595703125,
      "grad_norm": 14.163113594055176,
      "learning_rate": 8.729100892661149e-07,
      "loss": 3.1414,
      "step": 71355
    },
    {
      "epoch": 0.7259114583333334,
      "grad_norm": 13.765534400939941,
      "learning_rate": 8.726065131199399e-07,
      "loss": 3.4463,
      "step": 71360
    },
    {
      "epoch": 0.7259623209635416,
      "grad_norm": 13.300477027893066,
      "learning_rate": 8.723029786102421e-07,
      "loss": 3.039,
      "step": 71365
    },
    {
      "epoch": 0.72601318359375,
      "grad_norm": 12.640270233154297,
      "learning_rate": 8.71999485744787e-07,
      "loss": 3.3477,
      "step": 71370
    },
    {
      "epoch": 0.7260640462239584,
      "grad_norm": 7.233295917510986,
      "learning_rate": 8.716960345313382e-07,
      "loss": 2.9813,
      "step": 71375
    },
    {
      "epoch": 0.7261149088541666,
      "grad_norm": 14.029919624328613,
      "learning_rate": 8.713926249776608e-07,
      "loss": 3.4008,
      "step": 71380
    },
    {
      "epoch": 0.726165771484375,
      "grad_norm": 12.687051773071289,
      "learning_rate": 8.710892570915175e-07,
      "loss": 3.2873,
      "step": 71385
    },
    {
      "epoch": 0.7262166341145834,
      "grad_norm": 9.008588790893555,
      "learning_rate": 8.707859308806696e-07,
      "loss": 3.3346,
      "step": 71390
    },
    {
      "epoch": 0.7262674967447916,
      "grad_norm": 12.562435150146484,
      "learning_rate": 8.704826463528768e-07,
      "loss": 3.3961,
      "step": 71395
    },
    {
      "epoch": 0.726318359375,
      "grad_norm": 10.56306266784668,
      "learning_rate": 8.701794035158998e-07,
      "loss": 3.9296,
      "step": 71400
    },
    {
      "epoch": 0.7263692220052084,
      "grad_norm": 11.310155868530273,
      "learning_rate": 8.698762023774957e-07,
      "loss": 3.3878,
      "step": 71405
    },
    {
      "epoch": 0.7264200846354166,
      "grad_norm": 9.817281723022461,
      "learning_rate": 8.695730429454236e-07,
      "loss": 3.1968,
      "step": 71410
    },
    {
      "epoch": 0.726470947265625,
      "grad_norm": 15.564590454101562,
      "learning_rate": 8.692699252274378e-07,
      "loss": 3.1042,
      "step": 71415
    },
    {
      "epoch": 0.7265218098958334,
      "grad_norm": 10.80466365814209,
      "learning_rate": 8.689668492312952e-07,
      "loss": 3.2071,
      "step": 71420
    },
    {
      "epoch": 0.7265726725260416,
      "grad_norm": 9.413741111755371,
      "learning_rate": 8.686638149647492e-07,
      "loss": 3.1955,
      "step": 71425
    },
    {
      "epoch": 0.72662353515625,
      "grad_norm": 11.282853126525879,
      "learning_rate": 8.68360822435552e-07,
      "loss": 3.3632,
      "step": 71430
    },
    {
      "epoch": 0.7266743977864584,
      "grad_norm": 11.697951316833496,
      "learning_rate": 8.680578716514565e-07,
      "loss": 3.3989,
      "step": 71435
    },
    {
      "epoch": 0.7267252604166666,
      "grad_norm": 7.446695804595947,
      "learning_rate": 8.677549626202142e-07,
      "loss": 3.1363,
      "step": 71440
    },
    {
      "epoch": 0.726776123046875,
      "grad_norm": 15.161650657653809,
      "learning_rate": 8.674520953495743e-07,
      "loss": 3.4819,
      "step": 71445
    },
    {
      "epoch": 0.7268269856770834,
      "grad_norm": 13.455677032470703,
      "learning_rate": 8.671492698472858e-07,
      "loss": 3.4561,
      "step": 71450
    },
    {
      "epoch": 0.7268778483072916,
      "grad_norm": 19.11655616760254,
      "learning_rate": 8.668464861210951e-07,
      "loss": 3.5794,
      "step": 71455
    },
    {
      "epoch": 0.7269287109375,
      "grad_norm": 9.71003246307373,
      "learning_rate": 8.665437441787503e-07,
      "loss": 3.253,
      "step": 71460
    },
    {
      "epoch": 0.7269795735677084,
      "grad_norm": 9.952095031738281,
      "learning_rate": 8.662410440279973e-07,
      "loss": 3.2738,
      "step": 71465
    },
    {
      "epoch": 0.7270304361979166,
      "grad_norm": 13.357952117919922,
      "learning_rate": 8.659383856765799e-07,
      "loss": 3.4893,
      "step": 71470
    },
    {
      "epoch": 0.727081298828125,
      "grad_norm": 12.198160171508789,
      "learning_rate": 8.656357691322407e-07,
      "loss": 3.3286,
      "step": 71475
    },
    {
      "epoch": 0.7271321614583334,
      "grad_norm": 9.439297676086426,
      "learning_rate": 8.65333194402723e-07,
      "loss": 3.3716,
      "step": 71480
    },
    {
      "epoch": 0.7271830240885416,
      "grad_norm": 10.160725593566895,
      "learning_rate": 8.65030661495769e-07,
      "loss": 3.2346,
      "step": 71485
    },
    {
      "epoch": 0.72723388671875,
      "grad_norm": 11.871495246887207,
      "learning_rate": 8.647281704191171e-07,
      "loss": 3.5573,
      "step": 71490
    },
    {
      "epoch": 0.7272847493489584,
      "grad_norm": 14.323995590209961,
      "learning_rate": 8.644257211805082e-07,
      "loss": 3.2742,
      "step": 71495
    },
    {
      "epoch": 0.7273356119791666,
      "grad_norm": 8.815825462341309,
      "learning_rate": 8.641233137876795e-07,
      "loss": 3.4674,
      "step": 71500
    },
    {
      "epoch": 0.727386474609375,
      "grad_norm": 14.749484062194824,
      "learning_rate": 8.638209482483673e-07,
      "loss": 3.6575,
      "step": 71505
    },
    {
      "epoch": 0.7274373372395834,
      "grad_norm": 13.403284072875977,
      "learning_rate": 8.635186245703081e-07,
      "loss": 3.3399,
      "step": 71510
    },
    {
      "epoch": 0.7274881998697916,
      "grad_norm": 14.631975173950195,
      "learning_rate": 8.632163427612381e-07,
      "loss": 3.4657,
      "step": 71515
    },
    {
      "epoch": 0.7275390625,
      "grad_norm": 9.881767272949219,
      "learning_rate": 8.629141028288899e-07,
      "loss": 3.2145,
      "step": 71520
    },
    {
      "epoch": 0.7275899251302084,
      "grad_norm": 9.652484893798828,
      "learning_rate": 8.626119047809963e-07,
      "loss": 3.3607,
      "step": 71525
    },
    {
      "epoch": 0.7276407877604166,
      "grad_norm": 10.012579917907715,
      "learning_rate": 8.623097486252885e-07,
      "loss": 3.7364,
      "step": 71530
    },
    {
      "epoch": 0.727691650390625,
      "grad_norm": 12.135066986083984,
      "learning_rate": 8.620076343694972e-07,
      "loss": 3.7268,
      "step": 71535
    },
    {
      "epoch": 0.7277425130208334,
      "grad_norm": 13.072932243347168,
      "learning_rate": 8.617055620213533e-07,
      "loss": 3.3365,
      "step": 71540
    },
    {
      "epoch": 0.7277933756510416,
      "grad_norm": 9.491469383239746,
      "learning_rate": 8.614035315885844e-07,
      "loss": 3.3821,
      "step": 71545
    },
    {
      "epoch": 0.72784423828125,
      "grad_norm": 11.99251937866211,
      "learning_rate": 8.611015430789169e-07,
      "loss": 3.1887,
      "step": 71550
    },
    {
      "epoch": 0.7278951009114584,
      "grad_norm": 11.372830390930176,
      "learning_rate": 8.607995965000787e-07,
      "loss": 3.2789,
      "step": 71555
    },
    {
      "epoch": 0.7279459635416666,
      "grad_norm": 13.800250053405762,
      "learning_rate": 8.604976918597934e-07,
      "loss": 3.6385,
      "step": 71560
    },
    {
      "epoch": 0.727996826171875,
      "grad_norm": 13.713117599487305,
      "learning_rate": 8.60195829165787e-07,
      "loss": 3.3507,
      "step": 71565
    },
    {
      "epoch": 0.7280476888020834,
      "grad_norm": 9.791834831237793,
      "learning_rate": 8.598940084257807e-07,
      "loss": 3.1859,
      "step": 71570
    },
    {
      "epoch": 0.7280985514322916,
      "grad_norm": 7.772392272949219,
      "learning_rate": 8.595922296474984e-07,
      "loss": 3.2658,
      "step": 71575
    },
    {
      "epoch": 0.7281494140625,
      "grad_norm": 8.682499885559082,
      "learning_rate": 8.5929049283866e-07,
      "loss": 3.3533,
      "step": 71580
    },
    {
      "epoch": 0.7282002766927084,
      "grad_norm": 16.108346939086914,
      "learning_rate": 8.589887980069847e-07,
      "loss": 3.4971,
      "step": 71585
    },
    {
      "epoch": 0.7282511393229166,
      "grad_norm": 7.182333469390869,
      "learning_rate": 8.586871451601919e-07,
      "loss": 3.5022,
      "step": 71590
    },
    {
      "epoch": 0.728302001953125,
      "grad_norm": 13.270331382751465,
      "learning_rate": 8.583855343060004e-07,
      "loss": 3.4255,
      "step": 71595
    },
    {
      "epoch": 0.7283528645833334,
      "grad_norm": 7.81695556640625,
      "learning_rate": 8.580839654521258e-07,
      "loss": 3.1044,
      "step": 71600
    },
    {
      "epoch": 0.7284037272135416,
      "grad_norm": 10.429226875305176,
      "learning_rate": 8.57782438606283e-07,
      "loss": 3.5674,
      "step": 71605
    },
    {
      "epoch": 0.72845458984375,
      "grad_norm": 10.270829200744629,
      "learning_rate": 8.574809537761883e-07,
      "loss": 3.2177,
      "step": 71610
    },
    {
      "epoch": 0.7285054524739584,
      "grad_norm": 11.659355163574219,
      "learning_rate": 8.57179510969553e-07,
      "loss": 3.4913,
      "step": 71615
    },
    {
      "epoch": 0.7285563151041666,
      "grad_norm": 8.569073677062988,
      "learning_rate": 8.568781101940913e-07,
      "loss": 3.8532,
      "step": 71620
    },
    {
      "epoch": 0.728607177734375,
      "grad_norm": 9.819869995117188,
      "learning_rate": 8.565767514575129e-07,
      "loss": 3.2251,
      "step": 71625
    },
    {
      "epoch": 0.7286580403645834,
      "grad_norm": 7.252403736114502,
      "learning_rate": 8.562754347675298e-07,
      "loss": 3.6167,
      "step": 71630
    },
    {
      "epoch": 0.7287089029947916,
      "grad_norm": 15.263439178466797,
      "learning_rate": 8.559741601318492e-07,
      "loss": 3.5082,
      "step": 71635
    },
    {
      "epoch": 0.728759765625,
      "grad_norm": 16.039209365844727,
      "learning_rate": 8.556729275581807e-07,
      "loss": 3.1962,
      "step": 71640
    },
    {
      "epoch": 0.7288106282552084,
      "grad_norm": 10.13549518585205,
      "learning_rate": 8.5537173705423e-07,
      "loss": 3.6743,
      "step": 71645
    },
    {
      "epoch": 0.7288614908854166,
      "grad_norm": 7.909780979156494,
      "learning_rate": 8.550705886277042e-07,
      "loss": 3.1544,
      "step": 71650
    },
    {
      "epoch": 0.728912353515625,
      "grad_norm": 12.192684173583984,
      "learning_rate": 8.547694822863076e-07,
      "loss": 3.4837,
      "step": 71655
    },
    {
      "epoch": 0.7289632161458334,
      "grad_norm": 13.933347702026367,
      "learning_rate": 8.54468418037743e-07,
      "loss": 3.395,
      "step": 71660
    },
    {
      "epoch": 0.7290140787760416,
      "grad_norm": 10.228572845458984,
      "learning_rate": 8.541673958897137e-07,
      "loss": 3.3333,
      "step": 71665
    },
    {
      "epoch": 0.72906494140625,
      "grad_norm": 12.458081245422363,
      "learning_rate": 8.538664158499227e-07,
      "loss": 3.3682,
      "step": 71670
    },
    {
      "epoch": 0.7291158040364584,
      "grad_norm": 14.956607818603516,
      "learning_rate": 8.535654779260688e-07,
      "loss": 3.5155,
      "step": 71675
    },
    {
      "epoch": 0.7291666666666666,
      "grad_norm": 7.825222492218018,
      "learning_rate": 8.532645821258523e-07,
      "loss": 3.1764,
      "step": 71680
    },
    {
      "epoch": 0.729217529296875,
      "grad_norm": 12.992206573486328,
      "learning_rate": 8.529637284569703e-07,
      "loss": 3.2733,
      "step": 71685
    },
    {
      "epoch": 0.7292683919270834,
      "grad_norm": 12.961661338806152,
      "learning_rate": 8.526629169271206e-07,
      "loss": 3.5984,
      "step": 71690
    },
    {
      "epoch": 0.7293192545572916,
      "grad_norm": 15.560072898864746,
      "learning_rate": 8.523621475440008e-07,
      "loss": 3.3413,
      "step": 71695
    },
    {
      "epoch": 0.7293701171875,
      "grad_norm": 17.057737350463867,
      "learning_rate": 8.520614203153041e-07,
      "loss": 3.41,
      "step": 71700
    },
    {
      "epoch": 0.7294209798177084,
      "grad_norm": 13.707793235778809,
      "learning_rate": 8.517607352487258e-07,
      "loss": 3.1453,
      "step": 71705
    },
    {
      "epoch": 0.7294718424479166,
      "grad_norm": 11.256245613098145,
      "learning_rate": 8.514600923519589e-07,
      "loss": 2.9445,
      "step": 71710
    },
    {
      "epoch": 0.729522705078125,
      "grad_norm": 11.20262622833252,
      "learning_rate": 8.511594916326937e-07,
      "loss": 3.1541,
      "step": 71715
    },
    {
      "epoch": 0.7295735677083334,
      "grad_norm": 9.666428565979004,
      "learning_rate": 8.508589330986222e-07,
      "loss": 3.1237,
      "step": 71720
    },
    {
      "epoch": 0.7296244303385416,
      "grad_norm": 11.764225006103516,
      "learning_rate": 8.505584167574348e-07,
      "loss": 3.1072,
      "step": 71725
    },
    {
      "epoch": 0.72967529296875,
      "grad_norm": 9.681159019470215,
      "learning_rate": 8.502579426168195e-07,
      "loss": 3.0048,
      "step": 71730
    },
    {
      "epoch": 0.7297261555989584,
      "grad_norm": 10.529214859008789,
      "learning_rate": 8.499575106844634e-07,
      "loss": 3.3713,
      "step": 71735
    },
    {
      "epoch": 0.7297770182291666,
      "grad_norm": 13.212870597839355,
      "learning_rate": 8.496571209680527e-07,
      "loss": 3.0802,
      "step": 71740
    },
    {
      "epoch": 0.729827880859375,
      "grad_norm": 11.128890037536621,
      "learning_rate": 8.493567734752733e-07,
      "loss": 3.7974,
      "step": 71745
    },
    {
      "epoch": 0.7298787434895834,
      "grad_norm": 13.671854972839355,
      "learning_rate": 8.490564682138105e-07,
      "loss": 3.146,
      "step": 71750
    },
    {
      "epoch": 0.7299296061197916,
      "grad_norm": 13.513593673706055,
      "learning_rate": 8.487562051913464e-07,
      "loss": 2.9435,
      "step": 71755
    },
    {
      "epoch": 0.72998046875,
      "grad_norm": 12.44311809539795,
      "learning_rate": 8.484559844155629e-07,
      "loss": 3.6105,
      "step": 71760
    },
    {
      "epoch": 0.7300313313802084,
      "grad_norm": 13.10235595703125,
      "learning_rate": 8.481558058941422e-07,
      "loss": 3.5066,
      "step": 71765
    },
    {
      "epoch": 0.7300821940104166,
      "grad_norm": 13.457281112670898,
      "learning_rate": 8.478556696347629e-07,
      "loss": 3.5466,
      "step": 71770
    },
    {
      "epoch": 0.730133056640625,
      "grad_norm": 11.804722785949707,
      "learning_rate": 8.475555756451056e-07,
      "loss": 3.2333,
      "step": 71775
    },
    {
      "epoch": 0.7301839192708334,
      "grad_norm": 12.426148414611816,
      "learning_rate": 8.472555239328464e-07,
      "loss": 3.5555,
      "step": 71780
    },
    {
      "epoch": 0.7302347819010416,
      "grad_norm": 13.05529499053955,
      "learning_rate": 8.469555145056635e-07,
      "loss": 3.9392,
      "step": 71785
    },
    {
      "epoch": 0.73028564453125,
      "grad_norm": 15.0496826171875,
      "learning_rate": 8.466555473712312e-07,
      "loss": 3.7523,
      "step": 71790
    },
    {
      "epoch": 0.7303365071614584,
      "grad_norm": 9.555891036987305,
      "learning_rate": 8.463556225372249e-07,
      "loss": 3.2072,
      "step": 71795
    },
    {
      "epoch": 0.7303873697916666,
      "grad_norm": 12.876121520996094,
      "learning_rate": 8.46055740011319e-07,
      "loss": 3.5651,
      "step": 71800
    },
    {
      "epoch": 0.730438232421875,
      "grad_norm": 11.001485824584961,
      "learning_rate": 8.457558998011848e-07,
      "loss": 3.5513,
      "step": 71805
    },
    {
      "epoch": 0.7304890950520834,
      "grad_norm": 12.766059875488281,
      "learning_rate": 8.454561019144938e-07,
      "loss": 3.26,
      "step": 71810
    },
    {
      "epoch": 0.7305399576822916,
      "grad_norm": 11.093799591064453,
      "learning_rate": 8.451563463589155e-07,
      "loss": 3.4896,
      "step": 71815
    },
    {
      "epoch": 0.7305908203125,
      "grad_norm": 12.005446434020996,
      "learning_rate": 8.448566331421198e-07,
      "loss": 3.3718,
      "step": 71820
    },
    {
      "epoch": 0.7306416829427084,
      "grad_norm": 11.861042022705078,
      "learning_rate": 8.445569622717756e-07,
      "loss": 3.2817,
      "step": 71825
    },
    {
      "epoch": 0.7306925455729166,
      "grad_norm": 9.181353569030762,
      "learning_rate": 8.442573337555493e-07,
      "loss": 3.6946,
      "step": 71830
    },
    {
      "epoch": 0.730743408203125,
      "grad_norm": 12.56906795501709,
      "learning_rate": 8.439577476011057e-07,
      "loss": 3.1878,
      "step": 71835
    },
    {
      "epoch": 0.7307942708333334,
      "grad_norm": 11.338751792907715,
      "learning_rate": 8.436582038161114e-07,
      "loss": 3.6575,
      "step": 71840
    },
    {
      "epoch": 0.7308451334635416,
      "grad_norm": 13.186248779296875,
      "learning_rate": 8.433587024082284e-07,
      "loss": 3.2892,
      "step": 71845
    },
    {
      "epoch": 0.73089599609375,
      "grad_norm": 10.832112312316895,
      "learning_rate": 8.430592433851214e-07,
      "loss": 3.4635,
      "step": 71850
    },
    {
      "epoch": 0.7309468587239584,
      "grad_norm": 6.655515193939209,
      "learning_rate": 8.427598267544501e-07,
      "loss": 3.3429,
      "step": 71855
    },
    {
      "epoch": 0.7309977213541666,
      "grad_norm": 7.147447109222412,
      "learning_rate": 8.424604525238766e-07,
      "loss": 3.354,
      "step": 71860
    },
    {
      "epoch": 0.731048583984375,
      "grad_norm": 9.368706703186035,
      "learning_rate": 8.421611207010597e-07,
      "loss": 3.0716,
      "step": 71865
    },
    {
      "epoch": 0.7310994466145834,
      "grad_norm": 9.368498802185059,
      "learning_rate": 8.418618312936567e-07,
      "loss": 3.0367,
      "step": 71870
    },
    {
      "epoch": 0.7311503092447916,
      "grad_norm": 9.8778076171875,
      "learning_rate": 8.415625843093253e-07,
      "loss": 3.3414,
      "step": 71875
    },
    {
      "epoch": 0.731201171875,
      "grad_norm": 12.284628868103027,
      "learning_rate": 8.412633797557231e-07,
      "loss": 3.1554,
      "step": 71880
    },
    {
      "epoch": 0.7312520345052084,
      "grad_norm": 10.680723190307617,
      "learning_rate": 8.409642176405045e-07,
      "loss": 3.9743,
      "step": 71885
    },
    {
      "epoch": 0.7313028971354166,
      "grad_norm": 10.758345603942871,
      "learning_rate": 8.406650979713221e-07,
      "loss": 3.3799,
      "step": 71890
    },
    {
      "epoch": 0.731353759765625,
      "grad_norm": 13.093633651733398,
      "learning_rate": 8.403660207558306e-07,
      "loss": 3.1075,
      "step": 71895
    },
    {
      "epoch": 0.7314046223958334,
      "grad_norm": 7.281319618225098,
      "learning_rate": 8.400669860016805e-07,
      "loss": 3.5233,
      "step": 71900
    },
    {
      "epoch": 0.7314554850260416,
      "grad_norm": 12.522171974182129,
      "learning_rate": 8.397679937165237e-07,
      "loss": 3.2413,
      "step": 71905
    },
    {
      "epoch": 0.73150634765625,
      "grad_norm": 12.937942504882812,
      "learning_rate": 8.394690439080086e-07,
      "loss": 3.2693,
      "step": 71910
    },
    {
      "epoch": 0.7315572102864584,
      "grad_norm": 12.733464241027832,
      "learning_rate": 8.391701365837851e-07,
      "loss": 3.3613,
      "step": 71915
    },
    {
      "epoch": 0.7316080729166666,
      "grad_norm": 10.784283638000488,
      "learning_rate": 8.388712717515002e-07,
      "loss": 3.2883,
      "step": 71920
    },
    {
      "epoch": 0.731658935546875,
      "grad_norm": 9.458428382873535,
      "learning_rate": 8.385724494187991e-07,
      "loss": 3.6336,
      "step": 71925
    },
    {
      "epoch": 0.7317097981770834,
      "grad_norm": 14.847657203674316,
      "learning_rate": 8.382736695933283e-07,
      "loss": 3.5436,
      "step": 71930
    },
    {
      "epoch": 0.7317606608072916,
      "grad_norm": 15.701521873474121,
      "learning_rate": 8.379749322827324e-07,
      "loss": 3.4739,
      "step": 71935
    },
    {
      "epoch": 0.7318115234375,
      "grad_norm": 13.327818870544434,
      "learning_rate": 8.37676237494654e-07,
      "loss": 3.1551,
      "step": 71940
    },
    {
      "epoch": 0.7318623860677084,
      "grad_norm": 11.132681846618652,
      "learning_rate": 8.373775852367349e-07,
      "loss": 3.1077,
      "step": 71945
    },
    {
      "epoch": 0.7319132486979166,
      "grad_norm": 13.124969482421875,
      "learning_rate": 8.370789755166148e-07,
      "loss": 3.0136,
      "step": 71950
    },
    {
      "epoch": 0.731964111328125,
      "grad_norm": 11.906679153442383,
      "learning_rate": 8.367804083419361e-07,
      "loss": 3.4839,
      "step": 71955
    },
    {
      "epoch": 0.7320149739583334,
      "grad_norm": 8.344532012939453,
      "learning_rate": 8.36481883720337e-07,
      "loss": 3.3098,
      "step": 71960
    },
    {
      "epoch": 0.7320658365885416,
      "grad_norm": 11.388338088989258,
      "learning_rate": 8.361834016594544e-07,
      "loss": 3.3811,
      "step": 71965
    },
    {
      "epoch": 0.73211669921875,
      "grad_norm": 15.55689811706543,
      "learning_rate": 8.358849621669244e-07,
      "loss": 3.1579,
      "step": 71970
    },
    {
      "epoch": 0.7321675618489584,
      "grad_norm": 8.578241348266602,
      "learning_rate": 8.355865652503828e-07,
      "loss": 3.6504,
      "step": 71975
    },
    {
      "epoch": 0.7322184244791666,
      "grad_norm": 15.57926082611084,
      "learning_rate": 8.352882109174656e-07,
      "loss": 3.5903,
      "step": 71980
    },
    {
      "epoch": 0.732269287109375,
      "grad_norm": 13.08041763305664,
      "learning_rate": 8.349898991758041e-07,
      "loss": 3.5386,
      "step": 71985
    },
    {
      "epoch": 0.7323201497395834,
      "grad_norm": 12.157262802124023,
      "learning_rate": 8.346916300330321e-07,
      "loss": 3.1902,
      "step": 71990
    },
    {
      "epoch": 0.7323710123697916,
      "grad_norm": 10.394250869750977,
      "learning_rate": 8.343934034967802e-07,
      "loss": 3.6059,
      "step": 71995
    },
    {
      "epoch": 0.732421875,
      "grad_norm": 7.417933940887451,
      "learning_rate": 8.340952195746771e-07,
      "loss": 3.1564,
      "step": 72000
    },
    {
      "epoch": 0.7324727376302084,
      "grad_norm": 10.798310279846191,
      "learning_rate": 8.337970782743532e-07,
      "loss": 3.1306,
      "step": 72005
    },
    {
      "epoch": 0.7325236002604166,
      "grad_norm": 15.50532054901123,
      "learning_rate": 8.33498979603437e-07,
      "loss": 3.2922,
      "step": 72010
    },
    {
      "epoch": 0.732574462890625,
      "grad_norm": 14.92574405670166,
      "learning_rate": 8.332009235695543e-07,
      "loss": 3.1616,
      "step": 72015
    },
    {
      "epoch": 0.7326253255208334,
      "grad_norm": 11.129034042358398,
      "learning_rate": 8.32902910180331e-07,
      "loss": 3.29,
      "step": 72020
    },
    {
      "epoch": 0.7326761881510416,
      "grad_norm": 11.893033981323242,
      "learning_rate": 8.326049394433905e-07,
      "loss": 3.1642,
      "step": 72025
    },
    {
      "epoch": 0.73272705078125,
      "grad_norm": 18.53277587890625,
      "learning_rate": 8.323070113663575e-07,
      "loss": 3.7038,
      "step": 72030
    },
    {
      "epoch": 0.7327779134114584,
      "grad_norm": 14.153130531311035,
      "learning_rate": 8.320091259568552e-07,
      "loss": 3.4444,
      "step": 72035
    },
    {
      "epoch": 0.7328287760416666,
      "grad_norm": 13.978302001953125,
      "learning_rate": 8.317112832225044e-07,
      "loss": 3.357,
      "step": 72040
    },
    {
      "epoch": 0.732879638671875,
      "grad_norm": 12.95738410949707,
      "learning_rate": 8.314134831709236e-07,
      "loss": 3.1264,
      "step": 72045
    },
    {
      "epoch": 0.7329305013020834,
      "grad_norm": 14.97579574584961,
      "learning_rate": 8.311157258097346e-07,
      "loss": 3.1641,
      "step": 72050
    },
    {
      "epoch": 0.7329813639322916,
      "grad_norm": 16.014860153198242,
      "learning_rate": 8.308180111465531e-07,
      "loss": 3.3314,
      "step": 72055
    },
    {
      "epoch": 0.7330322265625,
      "grad_norm": 16.28436279296875,
      "learning_rate": 8.305203391889979e-07,
      "loss": 3.7881,
      "step": 72060
    },
    {
      "epoch": 0.7330830891927084,
      "grad_norm": 16.736268997192383,
      "learning_rate": 8.302227099446836e-07,
      "loss": 3.3879,
      "step": 72065
    },
    {
      "epoch": 0.7331339518229166,
      "grad_norm": 17.092361450195312,
      "learning_rate": 8.299251234212261e-07,
      "loss": 3.4993,
      "step": 72070
    },
    {
      "epoch": 0.733184814453125,
      "grad_norm": 12.747779846191406,
      "learning_rate": 8.296275796262385e-07,
      "loss": 2.9937,
      "step": 72075
    },
    {
      "epoch": 0.7332356770833334,
      "grad_norm": 13.478667259216309,
      "learning_rate": 8.293300785673325e-07,
      "loss": 3.4399,
      "step": 72080
    },
    {
      "epoch": 0.7332865397135416,
      "grad_norm": 23.306598663330078,
      "learning_rate": 8.290326202521207e-07,
      "loss": 3.3257,
      "step": 72085
    },
    {
      "epoch": 0.73333740234375,
      "grad_norm": 8.487061500549316,
      "learning_rate": 8.287352046882139e-07,
      "loss": 3.4002,
      "step": 72090
    },
    {
      "epoch": 0.7333882649739584,
      "grad_norm": 15.717615127563477,
      "learning_rate": 8.284378318832206e-07,
      "loss": 3.2684,
      "step": 72095
    },
    {
      "epoch": 0.7334391276041666,
      "grad_norm": 14.708165168762207,
      "learning_rate": 8.281405018447486e-07,
      "loss": 3.3015,
      "step": 72100
    },
    {
      "epoch": 0.733489990234375,
      "grad_norm": 8.821043968200684,
      "learning_rate": 8.278432145804052e-07,
      "loss": 3.8564,
      "step": 72105
    },
    {
      "epoch": 0.7335408528645834,
      "grad_norm": 10.754634857177734,
      "learning_rate": 8.275459700977981e-07,
      "loss": 3.4867,
      "step": 72110
    },
    {
      "epoch": 0.7335917154947916,
      "grad_norm": 12.012271881103516,
      "learning_rate": 8.272487684045305e-07,
      "loss": 3.1794,
      "step": 72115
    },
    {
      "epoch": 0.733642578125,
      "grad_norm": 12.912185668945312,
      "learning_rate": 8.269516095082061e-07,
      "loss": 3.1557,
      "step": 72120
    },
    {
      "epoch": 0.7336934407552084,
      "grad_norm": 8.939395904541016,
      "learning_rate": 8.266544934164292e-07,
      "loss": 3.2299,
      "step": 72125
    },
    {
      "epoch": 0.7337443033854166,
      "grad_norm": 11.69654655456543,
      "learning_rate": 8.263574201367994e-07,
      "loss": 3.4519,
      "step": 72130
    },
    {
      "epoch": 0.733795166015625,
      "grad_norm": 13.756976127624512,
      "learning_rate": 8.260603896769193e-07,
      "loss": 3.3767,
      "step": 72135
    },
    {
      "epoch": 0.7338460286458334,
      "grad_norm": 16.499744415283203,
      "learning_rate": 8.257634020443863e-07,
      "loss": 3.3217,
      "step": 72140
    },
    {
      "epoch": 0.7338968912760416,
      "grad_norm": 12.474640846252441,
      "learning_rate": 8.254664572468008e-07,
      "loss": 3.1985,
      "step": 72145
    },
    {
      "epoch": 0.73394775390625,
      "grad_norm": 8.013901710510254,
      "learning_rate": 8.251695552917591e-07,
      "loss": 3.7307,
      "step": 72150
    },
    {
      "epoch": 0.7339986165364584,
      "grad_norm": 13.143088340759277,
      "learning_rate": 8.248726961868564e-07,
      "loss": 3.4681,
      "step": 72155
    },
    {
      "epoch": 0.7340494791666666,
      "grad_norm": 8.778640747070312,
      "learning_rate": 8.245758799396889e-07,
      "loss": 3.1804,
      "step": 72160
    },
    {
      "epoch": 0.734100341796875,
      "grad_norm": 15.068650245666504,
      "learning_rate": 8.242791065578509e-07,
      "loss": 3.3836,
      "step": 72165
    },
    {
      "epoch": 0.7341512044270834,
      "grad_norm": 11.814011573791504,
      "learning_rate": 8.239823760489352e-07,
      "loss": 3.3396,
      "step": 72170
    },
    {
      "epoch": 0.7342020670572916,
      "grad_norm": 14.065567970275879,
      "learning_rate": 8.236856884205327e-07,
      "loss": 3.2675,
      "step": 72175
    },
    {
      "epoch": 0.7342529296875,
      "grad_norm": 12.572928428649902,
      "learning_rate": 8.233890436802338e-07,
      "loss": 3.2514,
      "step": 72180
    },
    {
      "epoch": 0.7343037923177084,
      "grad_norm": 10.292560577392578,
      "learning_rate": 8.230924418356292e-07,
      "loss": 3.0932,
      "step": 72185
    },
    {
      "epoch": 0.7343546549479166,
      "grad_norm": 9.869354248046875,
      "learning_rate": 8.227958828943075e-07,
      "loss": 3.3961,
      "step": 72190
    },
    {
      "epoch": 0.734405517578125,
      "grad_norm": 10.191967964172363,
      "learning_rate": 8.224993668638548e-07,
      "loss": 3.1959,
      "step": 72195
    },
    {
      "epoch": 0.7344563802083334,
      "grad_norm": 14.69416618347168,
      "learning_rate": 8.222028937518592e-07,
      "loss": 3.1104,
      "step": 72200
    },
    {
      "epoch": 0.7345072428385416,
      "grad_norm": 8.91283893585205,
      "learning_rate": 8.219064635659047e-07,
      "loss": 4.0492,
      "step": 72205
    },
    {
      "epoch": 0.73455810546875,
      "grad_norm": 10.318886756896973,
      "learning_rate": 8.216100763135748e-07,
      "loss": 3.5003,
      "step": 72210
    },
    {
      "epoch": 0.7346089680989584,
      "grad_norm": 11.990934371948242,
      "learning_rate": 8.213137320024534e-07,
      "loss": 3.1464,
      "step": 72215
    },
    {
      "epoch": 0.7346598307291666,
      "grad_norm": 8.987807273864746,
      "learning_rate": 8.210174306401228e-07,
      "loss": 3.09,
      "step": 72220
    },
    {
      "epoch": 0.734710693359375,
      "grad_norm": 10.0052490234375,
      "learning_rate": 8.207211722341635e-07,
      "loss": 3.7299,
      "step": 72225
    },
    {
      "epoch": 0.7347615559895834,
      "grad_norm": 14.059890747070312,
      "learning_rate": 8.204249567921549e-07,
      "loss": 3.0111,
      "step": 72230
    },
    {
      "epoch": 0.7348124186197916,
      "grad_norm": 8.948488235473633,
      "learning_rate": 8.201287843216749e-07,
      "loss": 3.5322,
      "step": 72235
    },
    {
      "epoch": 0.73486328125,
      "grad_norm": 10.490068435668945,
      "learning_rate": 8.198326548303018e-07,
      "loss": 3.2814,
      "step": 72240
    },
    {
      "epoch": 0.7349141438802084,
      "grad_norm": 12.267974853515625,
      "learning_rate": 8.19536568325613e-07,
      "loss": 3.1633,
      "step": 72245
    },
    {
      "epoch": 0.7349650065104166,
      "grad_norm": 7.18000602722168,
      "learning_rate": 8.192405248151825e-07,
      "loss": 3.5377,
      "step": 72250
    },
    {
      "epoch": 0.735015869140625,
      "grad_norm": 8.979900360107422,
      "learning_rate": 8.189445243065838e-07,
      "loss": 3.1741,
      "step": 72255
    },
    {
      "epoch": 0.7350667317708334,
      "grad_norm": 10.791680335998535,
      "learning_rate": 8.186485668073913e-07,
      "loss": 3.4301,
      "step": 72260
    },
    {
      "epoch": 0.7351175944010416,
      "grad_norm": 8.889440536499023,
      "learning_rate": 8.183526523251775e-07,
      "loss": 3.4613,
      "step": 72265
    },
    {
      "epoch": 0.73516845703125,
      "grad_norm": 9.243463516235352,
      "learning_rate": 8.180567808675125e-07,
      "loss": 3.2286,
      "step": 72270
    },
    {
      "epoch": 0.7352193196614584,
      "grad_norm": 12.702774047851562,
      "learning_rate": 8.17760952441965e-07,
      "loss": 3.4112,
      "step": 72275
    },
    {
      "epoch": 0.7352701822916666,
      "grad_norm": 13.84238052368164,
      "learning_rate": 8.174651670561059e-07,
      "loss": 3.1961,
      "step": 72280
    },
    {
      "epoch": 0.735321044921875,
      "grad_norm": 15.002708435058594,
      "learning_rate": 8.171694247175008e-07,
      "loss": 3.5209,
      "step": 72285
    },
    {
      "epoch": 0.7353719075520834,
      "grad_norm": 12.038010597229004,
      "learning_rate": 8.168737254337169e-07,
      "loss": 3.341,
      "step": 72290
    },
    {
      "epoch": 0.7354227701822916,
      "grad_norm": 9.858930587768555,
      "learning_rate": 8.16578069212321e-07,
      "loss": 3.2706,
      "step": 72295
    },
    {
      "epoch": 0.7354736328125,
      "grad_norm": 9.680066108703613,
      "learning_rate": 8.16282456060876e-07,
      "loss": 3.295,
      "step": 72300
    },
    {
      "epoch": 0.7355244954427084,
      "grad_norm": 9.628165245056152,
      "learning_rate": 8.159868859869452e-07,
      "loss": 3.629,
      "step": 72305
    },
    {
      "epoch": 0.7355753580729166,
      "grad_norm": 13.200835227966309,
      "learning_rate": 8.1569135899809e-07,
      "loss": 3.5584,
      "step": 72310
    },
    {
      "epoch": 0.735626220703125,
      "grad_norm": 13.335905075073242,
      "learning_rate": 8.15395875101872e-07,
      "loss": 3.2494,
      "step": 72315
    },
    {
      "epoch": 0.7356770833333334,
      "grad_norm": 16.441917419433594,
      "learning_rate": 8.151004343058524e-07,
      "loss": 3.6302,
      "step": 72320
    },
    {
      "epoch": 0.7357279459635416,
      "grad_norm": 10.481017112731934,
      "learning_rate": 8.148050366175886e-07,
      "loss": 3.4223,
      "step": 72325
    },
    {
      "epoch": 0.73577880859375,
      "grad_norm": 10.352675437927246,
      "learning_rate": 8.145096820446378e-07,
      "loss": 3.4068,
      "step": 72330
    },
    {
      "epoch": 0.7358296712239584,
      "grad_norm": 7.219904899597168,
      "learning_rate": 8.14214370594558e-07,
      "loss": 2.8696,
      "step": 72335
    },
    {
      "epoch": 0.7358805338541666,
      "grad_norm": 9.463799476623535,
      "learning_rate": 8.139191022749033e-07,
      "loss": 3.4689,
      "step": 72340
    },
    {
      "epoch": 0.735931396484375,
      "grad_norm": 10.011601448059082,
      "learning_rate": 8.136238770932292e-07,
      "loss": 3.0355,
      "step": 72345
    },
    {
      "epoch": 0.7359822591145834,
      "grad_norm": 7.603102684020996,
      "learning_rate": 8.133286950570879e-07,
      "loss": 2.8421,
      "step": 72350
    },
    {
      "epoch": 0.7360331217447916,
      "grad_norm": 9.22468376159668,
      "learning_rate": 8.130335561740329e-07,
      "loss": 3.179,
      "step": 72355
    },
    {
      "epoch": 0.736083984375,
      "grad_norm": 15.12529182434082,
      "learning_rate": 8.127384604516147e-07,
      "loss": 3.5254,
      "step": 72360
    },
    {
      "epoch": 0.7361348470052084,
      "grad_norm": 11.003714561462402,
      "learning_rate": 8.124434078973819e-07,
      "loss": 3.2567,
      "step": 72365
    },
    {
      "epoch": 0.7361857096354166,
      "grad_norm": 8.871549606323242,
      "learning_rate": 8.121483985188849e-07,
      "loss": 3.9288,
      "step": 72370
    },
    {
      "epoch": 0.736236572265625,
      "grad_norm": 11.090631484985352,
      "learning_rate": 8.118534323236716e-07,
      "loss": 3.4408,
      "step": 72375
    },
    {
      "epoch": 0.7362874348958334,
      "grad_norm": 8.611220359802246,
      "learning_rate": 8.115585093192882e-07,
      "loss": 3.488,
      "step": 72380
    },
    {
      "epoch": 0.7363382975260416,
      "grad_norm": 9.688155174255371,
      "learning_rate": 8.112636295132795e-07,
      "loss": 3.7627,
      "step": 72385
    },
    {
      "epoch": 0.73638916015625,
      "grad_norm": 10.717297554016113,
      "learning_rate": 8.10968792913191e-07,
      "loss": 3.582,
      "step": 72390
    },
    {
      "epoch": 0.7364400227864584,
      "grad_norm": 12.765704154968262,
      "learning_rate": 8.10673999526565e-07,
      "loss": 3.3121,
      "step": 72395
    },
    {
      "epoch": 0.7364908854166666,
      "grad_norm": 8.391258239746094,
      "learning_rate": 8.10379249360945e-07,
      "loss": 3.4052,
      "step": 72400
    },
    {
      "epoch": 0.736541748046875,
      "grad_norm": 14.208325386047363,
      "learning_rate": 8.100845424238707e-07,
      "loss": 3.1379,
      "step": 72405
    },
    {
      "epoch": 0.7365926106770834,
      "grad_norm": 12.104424476623535,
      "learning_rate": 8.097898787228836e-07,
      "loss": 3.3069,
      "step": 72410
    },
    {
      "epoch": 0.7366434733072916,
      "grad_norm": 8.169574737548828,
      "learning_rate": 8.094952582655213e-07,
      "loss": 3.6452,
      "step": 72415
    },
    {
      "epoch": 0.7366943359375,
      "grad_norm": 11.388225555419922,
      "learning_rate": 8.092006810593225e-07,
      "loss": 3.5901,
      "step": 72420
    },
    {
      "epoch": 0.7367451985677084,
      "grad_norm": 7.172380447387695,
      "learning_rate": 8.089061471118228e-07,
      "loss": 3.3948,
      "step": 72425
    },
    {
      "epoch": 0.7367960611979166,
      "grad_norm": 11.014195442199707,
      "learning_rate": 8.086116564305596e-07,
      "loss": 3.3273,
      "step": 72430
    },
    {
      "epoch": 0.736846923828125,
      "grad_norm": 7.348326683044434,
      "learning_rate": 8.083172090230659e-07,
      "loss": 3.1885,
      "step": 72435
    },
    {
      "epoch": 0.7368977864583334,
      "grad_norm": 8.079544067382812,
      "learning_rate": 8.08022804896875e-07,
      "loss": 3.4375,
      "step": 72440
    },
    {
      "epoch": 0.7369486490885416,
      "grad_norm": 13.889323234558105,
      "learning_rate": 8.077284440595193e-07,
      "loss": 3.825,
      "step": 72445
    },
    {
      "epoch": 0.73699951171875,
      "grad_norm": 11.928169250488281,
      "learning_rate": 8.074341265185312e-07,
      "loss": 3.3244,
      "step": 72450
    },
    {
      "epoch": 0.7370503743489584,
      "grad_norm": 7.652375221252441,
      "learning_rate": 8.071398522814397e-07,
      "loss": 3.1061,
      "step": 72455
    },
    {
      "epoch": 0.7371012369791666,
      "grad_norm": 11.007584571838379,
      "learning_rate": 8.068456213557738e-07,
      "loss": 2.9223,
      "step": 72460
    },
    {
      "epoch": 0.737152099609375,
      "grad_norm": 14.736150741577148,
      "learning_rate": 8.065514337490607e-07,
      "loss": 3.195,
      "step": 72465
    },
    {
      "epoch": 0.7372029622395834,
      "grad_norm": 14.029659271240234,
      "learning_rate": 8.062572894688275e-07,
      "loss": 3.6088,
      "step": 72470
    },
    {
      "epoch": 0.7372538248697916,
      "grad_norm": 17.324636459350586,
      "learning_rate": 8.059631885226013e-07,
      "loss": 3.2988,
      "step": 72475
    },
    {
      "epoch": 0.7373046875,
      "grad_norm": 14.843786239624023,
      "learning_rate": 8.056691309179043e-07,
      "loss": 3.1665,
      "step": 72480
    },
    {
      "epoch": 0.7373555501302084,
      "grad_norm": 13.8319673538208,
      "learning_rate": 8.053751166622617e-07,
      "loss": 3.1889,
      "step": 72485
    },
    {
      "epoch": 0.7374064127604166,
      "grad_norm": 15.034490585327148,
      "learning_rate": 8.050811457631955e-07,
      "loss": 3.6928,
      "step": 72490
    },
    {
      "epoch": 0.737457275390625,
      "grad_norm": 15.734111785888672,
      "learning_rate": 8.047872182282251e-07,
      "loss": 3.6073,
      "step": 72495
    },
    {
      "epoch": 0.7375081380208334,
      "grad_norm": 12.015183448791504,
      "learning_rate": 8.044933340648722e-07,
      "loss": 3.451,
      "step": 72500
    },
    {
      "epoch": 0.7375590006510416,
      "grad_norm": 8.523397445678711,
      "learning_rate": 8.041994932806562e-07,
      "loss": 3.6919,
      "step": 72505
    },
    {
      "epoch": 0.73760986328125,
      "grad_norm": 9.58163070678711,
      "learning_rate": 8.039056958830943e-07,
      "loss": 3.1693,
      "step": 72510
    },
    {
      "epoch": 0.7376607259114584,
      "grad_norm": 9.87668228149414,
      "learning_rate": 8.03611941879703e-07,
      "loss": 3.3357,
      "step": 72515
    },
    {
      "epoch": 0.7377115885416666,
      "grad_norm": 14.215899467468262,
      "learning_rate": 8.033182312779978e-07,
      "loss": 3.0186,
      "step": 72520
    },
    {
      "epoch": 0.737762451171875,
      "grad_norm": 11.756176948547363,
      "learning_rate": 8.030245640854928e-07,
      "loss": 3.0071,
      "step": 72525
    },
    {
      "epoch": 0.7378133138020834,
      "grad_norm": 10.167065620422363,
      "learning_rate": 8.027309403097034e-07,
      "loss": 2.918,
      "step": 72530
    },
    {
      "epoch": 0.7378641764322916,
      "grad_norm": 9.327035903930664,
      "learning_rate": 8.024373599581403e-07,
      "loss": 3.4352,
      "step": 72535
    },
    {
      "epoch": 0.7379150390625,
      "grad_norm": 13.697539329528809,
      "learning_rate": 8.021438230383141e-07,
      "loss": 3.3313,
      "step": 72540
    },
    {
      "epoch": 0.7379659016927084,
      "grad_norm": 14.816646575927734,
      "learning_rate": 8.018503295577368e-07,
      "loss": 3.4092,
      "step": 72545
    },
    {
      "epoch": 0.7380167643229166,
      "grad_norm": 10.502012252807617,
      "learning_rate": 8.015568795239153e-07,
      "loss": 3.0441,
      "step": 72550
    },
    {
      "epoch": 0.738067626953125,
      "grad_norm": 13.62344741821289,
      "learning_rate": 8.012634729443597e-07,
      "loss": 3.1531,
      "step": 72555
    },
    {
      "epoch": 0.7381184895833334,
      "grad_norm": 8.448206901550293,
      "learning_rate": 8.009701098265743e-07,
      "loss": 3.3725,
      "step": 72560
    },
    {
      "epoch": 0.7381693522135416,
      "grad_norm": 17.432044982910156,
      "learning_rate": 8.006767901780669e-07,
      "loss": 3.2959,
      "step": 72565
    },
    {
      "epoch": 0.73822021484375,
      "grad_norm": 16.10090446472168,
      "learning_rate": 8.003835140063402e-07,
      "loss": 3.6261,
      "step": 72570
    },
    {
      "epoch": 0.7382710774739584,
      "grad_norm": 8.387552261352539,
      "learning_rate": 8.000902813188982e-07,
      "loss": 3.3503,
      "step": 72575
    },
    {
      "epoch": 0.7383219401041666,
      "grad_norm": 15.424768447875977,
      "learning_rate": 7.997970921232446e-07,
      "loss": 4.2835,
      "step": 72580
    },
    {
      "epoch": 0.738372802734375,
      "grad_norm": 10.04584789276123,
      "learning_rate": 7.995039464268791e-07,
      "loss": 3.8604,
      "step": 72585
    },
    {
      "epoch": 0.7384236653645834,
      "grad_norm": 9.393284797668457,
      "learning_rate": 7.992108442373023e-07,
      "loss": 3.1383,
      "step": 72590
    },
    {
      "epoch": 0.7384745279947916,
      "grad_norm": 9.053309440612793,
      "learning_rate": 7.98917785562012e-07,
      "loss": 3.0691,
      "step": 72595
    },
    {
      "epoch": 0.738525390625,
      "grad_norm": 15.204339981079102,
      "learning_rate": 7.986247704085068e-07,
      "loss": 3.2515,
      "step": 72600
    },
    {
      "epoch": 0.7385762532552084,
      "grad_norm": 11.147443771362305,
      "learning_rate": 7.983317987842848e-07,
      "loss": 3.0584,
      "step": 72605
    },
    {
      "epoch": 0.7386271158854166,
      "grad_norm": 10.606264114379883,
      "learning_rate": 7.9803887069684e-07,
      "loss": 3.3795,
      "step": 72610
    },
    {
      "epoch": 0.738677978515625,
      "grad_norm": 7.328312397003174,
      "learning_rate": 7.977459861536666e-07,
      "loss": 3.285,
      "step": 72615
    },
    {
      "epoch": 0.7387288411458334,
      "grad_norm": 10.47547435760498,
      "learning_rate": 7.974531451622594e-07,
      "loss": 3.553,
      "step": 72620
    },
    {
      "epoch": 0.7387797037760416,
      "grad_norm": 16.50904083251953,
      "learning_rate": 7.971603477301093e-07,
      "loss": 3.1711,
      "step": 72625
    },
    {
      "epoch": 0.73883056640625,
      "grad_norm": 13.123628616333008,
      "learning_rate": 7.968675938647088e-07,
      "loss": 3.1417,
      "step": 72630
    },
    {
      "epoch": 0.7388814290364584,
      "grad_norm": 9.2986478805542,
      "learning_rate": 7.965748835735465e-07,
      "loss": 3.3235,
      "step": 72635
    },
    {
      "epoch": 0.7389322916666666,
      "grad_norm": 15.497564315795898,
      "learning_rate": 7.962822168641127e-07,
      "loss": 3.937,
      "step": 72640
    },
    {
      "epoch": 0.738983154296875,
      "grad_norm": 12.958601951599121,
      "learning_rate": 7.959895937438946e-07,
      "loss": 3.7955,
      "step": 72645
    },
    {
      "epoch": 0.7390340169270834,
      "grad_norm": 13.965435028076172,
      "learning_rate": 7.956970142203782e-07,
      "loss": 3.2878,
      "step": 72650
    },
    {
      "epoch": 0.7390848795572916,
      "grad_norm": 12.274550437927246,
      "learning_rate": 7.954044783010498e-07,
      "loss": 3.1043,
      "step": 72655
    },
    {
      "epoch": 0.7391357421875,
      "grad_norm": 8.859460830688477,
      "learning_rate": 7.951119859933942e-07,
      "loss": 3.5147,
      "step": 72660
    },
    {
      "epoch": 0.7391866048177084,
      "grad_norm": 10.588842391967773,
      "learning_rate": 7.948195373048947e-07,
      "loss": 3.6557,
      "step": 72665
    },
    {
      "epoch": 0.7392374674479166,
      "grad_norm": 11.107576370239258,
      "learning_rate": 7.94527132243032e-07,
      "loss": 3.2825,
      "step": 72670
    },
    {
      "epoch": 0.739288330078125,
      "grad_norm": 11.815327644348145,
      "learning_rate": 7.942347708152895e-07,
      "loss": 2.9843,
      "step": 72675
    },
    {
      "epoch": 0.7393391927083334,
      "grad_norm": 8.988689422607422,
      "learning_rate": 7.939424530291451e-07,
      "loss": 3.3577,
      "step": 72680
    },
    {
      "epoch": 0.7393900553385416,
      "grad_norm": 12.207701683044434,
      "learning_rate": 7.936501788920798e-07,
      "loss": 3.3142,
      "step": 72685
    },
    {
      "epoch": 0.73944091796875,
      "grad_norm": 11.993644714355469,
      "learning_rate": 7.933579484115689e-07,
      "loss": 3.1865,
      "step": 72690
    },
    {
      "epoch": 0.7394917805989584,
      "grad_norm": 10.642743110656738,
      "learning_rate": 7.930657615950915e-07,
      "loss": 3.1436,
      "step": 72695
    },
    {
      "epoch": 0.7395426432291666,
      "grad_norm": 9.111330032348633,
      "learning_rate": 7.92773618450122e-07,
      "loss": 3.0421,
      "step": 72700
    },
    {
      "epoch": 0.739593505859375,
      "grad_norm": 7.885223388671875,
      "learning_rate": 7.924815189841339e-07,
      "loss": 3.1854,
      "step": 72705
    },
    {
      "epoch": 0.7396443684895834,
      "grad_norm": 17.917495727539062,
      "learning_rate": 7.921894632046015e-07,
      "loss": 3.8181,
      "step": 72710
    },
    {
      "epoch": 0.7396952311197916,
      "grad_norm": 11.11347770690918,
      "learning_rate": 7.918974511189975e-07,
      "loss": 2.9804,
      "step": 72715
    },
    {
      "epoch": 0.73974609375,
      "grad_norm": 7.590899467468262,
      "learning_rate": 7.916054827347925e-07,
      "loss": 3.3131,
      "step": 72720
    },
    {
      "epoch": 0.7397969563802084,
      "grad_norm": 16.299772262573242,
      "learning_rate": 7.913135580594556e-07,
      "loss": 3.2908,
      "step": 72725
    },
    {
      "epoch": 0.7398478190104166,
      "grad_norm": 11.786675453186035,
      "learning_rate": 7.910216771004561e-07,
      "loss": 3.1259,
      "step": 72730
    },
    {
      "epoch": 0.739898681640625,
      "grad_norm": 10.466172218322754,
      "learning_rate": 7.907298398652627e-07,
      "loss": 3.3856,
      "step": 72735
    },
    {
      "epoch": 0.7399495442708334,
      "grad_norm": 12.573579788208008,
      "learning_rate": 7.904380463613414e-07,
      "loss": 3.5018,
      "step": 72740
    },
    {
      "epoch": 0.7400004069010416,
      "grad_norm": 12.838423728942871,
      "learning_rate": 7.901462965961573e-07,
      "loss": 3.0551,
      "step": 72745
    },
    {
      "epoch": 0.74005126953125,
      "grad_norm": 8.679061889648438,
      "learning_rate": 7.898545905771743e-07,
      "loss": 3.61,
      "step": 72750
    },
    {
      "epoch": 0.7401021321614584,
      "grad_norm": 7.51031494140625,
      "learning_rate": 7.895629283118561e-07,
      "loss": 3.1428,
      "step": 72755
    },
    {
      "epoch": 0.7401529947916666,
      "grad_norm": 12.5309419631958,
      "learning_rate": 7.892713098076659e-07,
      "loss": 3.3681,
      "step": 72760
    },
    {
      "epoch": 0.740203857421875,
      "grad_norm": 13.049712181091309,
      "learning_rate": 7.889797350720637e-07,
      "loss": 2.9711,
      "step": 72765
    },
    {
      "epoch": 0.7402547200520834,
      "grad_norm": 11.52535343170166,
      "learning_rate": 7.886882041125085e-07,
      "loss": 3.2727,
      "step": 72770
    },
    {
      "epoch": 0.7403055826822916,
      "grad_norm": 11.455509185791016,
      "learning_rate": 7.883967169364609e-07,
      "loss": 3.365,
      "step": 72775
    },
    {
      "epoch": 0.7403564453125,
      "grad_norm": 10.567713737487793,
      "learning_rate": 7.881052735513769e-07,
      "loss": 3.1708,
      "step": 72780
    },
    {
      "epoch": 0.7404073079427084,
      "grad_norm": 8.276542663574219,
      "learning_rate": 7.878138739647134e-07,
      "loss": 3.346,
      "step": 72785
    },
    {
      "epoch": 0.7404581705729166,
      "grad_norm": 8.877224922180176,
      "learning_rate": 7.875225181839272e-07,
      "loss": 3.1973,
      "step": 72790
    },
    {
      "epoch": 0.740509033203125,
      "grad_norm": 12.39207649230957,
      "learning_rate": 7.872312062164714e-07,
      "loss": 3.3151,
      "step": 72795
    },
    {
      "epoch": 0.7405598958333334,
      "grad_norm": 10.976158142089844,
      "learning_rate": 7.869399380697992e-07,
      "loss": 2.8954,
      "step": 72800
    },
    {
      "epoch": 0.7406107584635416,
      "grad_norm": 17.814804077148438,
      "learning_rate": 7.866487137513618e-07,
      "loss": 3.5182,
      "step": 72805
    },
    {
      "epoch": 0.74066162109375,
      "grad_norm": 10.832623481750488,
      "learning_rate": 7.86357533268611e-07,
      "loss": 3.3605,
      "step": 72810
    },
    {
      "epoch": 0.7407124837239584,
      "grad_norm": 15.117820739746094,
      "learning_rate": 7.860663966289972e-07,
      "loss": 3.0494,
      "step": 72815
    },
    {
      "epoch": 0.7407633463541666,
      "grad_norm": 10.470831871032715,
      "learning_rate": 7.857753038399684e-07,
      "loss": 3.5723,
      "step": 72820
    },
    {
      "epoch": 0.740814208984375,
      "grad_norm": 12.327982902526855,
      "learning_rate": 7.854842549089717e-07,
      "loss": 3.2256,
      "step": 72825
    },
    {
      "epoch": 0.7408650716145834,
      "grad_norm": 13.123862266540527,
      "learning_rate": 7.851932498434544e-07,
      "loss": 3.1989,
      "step": 72830
    },
    {
      "epoch": 0.7409159342447916,
      "grad_norm": 12.228321075439453,
      "learning_rate": 7.849022886508606e-07,
      "loss": 3.4185,
      "step": 72835
    },
    {
      "epoch": 0.740966796875,
      "grad_norm": 11.303017616271973,
      "learning_rate": 7.846113713386361e-07,
      "loss": 3.5949,
      "step": 72840
    },
    {
      "epoch": 0.7410176595052084,
      "grad_norm": 6.818903923034668,
      "learning_rate": 7.843204979142221e-07,
      "loss": 3.4595,
      "step": 72845
    },
    {
      "epoch": 0.7410685221354166,
      "grad_norm": 12.604376792907715,
      "learning_rate": 7.840296683850626e-07,
      "loss": 3.0496,
      "step": 72850
    },
    {
      "epoch": 0.741119384765625,
      "grad_norm": 16.697912216186523,
      "learning_rate": 7.83738882758597e-07,
      "loss": 3.5606,
      "step": 72855
    },
    {
      "epoch": 0.7411702473958334,
      "grad_norm": 12.759716033935547,
      "learning_rate": 7.834481410422649e-07,
      "loss": 3.3353,
      "step": 72860
    },
    {
      "epoch": 0.7412211100260416,
      "grad_norm": 13.470650672912598,
      "learning_rate": 7.831574432435049e-07,
      "loss": 3.6976,
      "step": 72865
    },
    {
      "epoch": 0.74127197265625,
      "grad_norm": 7.027815341949463,
      "learning_rate": 7.828667893697556e-07,
      "loss": 3.1388,
      "step": 72870
    },
    {
      "epoch": 0.7413228352864584,
      "grad_norm": 13.945786476135254,
      "learning_rate": 7.825761794284526e-07,
      "loss": 3.4302,
      "step": 72875
    },
    {
      "epoch": 0.7413736979166666,
      "grad_norm": 9.513721466064453,
      "learning_rate": 7.822856134270301e-07,
      "loss": 3.1297,
      "step": 72880
    },
    {
      "epoch": 0.741424560546875,
      "grad_norm": 12.166504859924316,
      "learning_rate": 7.81995091372923e-07,
      "loss": 3.8204,
      "step": 72885
    },
    {
      "epoch": 0.7414754231770834,
      "grad_norm": 14.738011360168457,
      "learning_rate": 7.81704613273565e-07,
      "loss": 3.5365,
      "step": 72890
    },
    {
      "epoch": 0.7415262858072916,
      "grad_norm": 9.779434204101562,
      "learning_rate": 7.814141791363875e-07,
      "loss": 3.3975,
      "step": 72895
    },
    {
      "epoch": 0.7415771484375,
      "grad_norm": 13.886914253234863,
      "learning_rate": 7.811237889688198e-07,
      "loss": 3.2775,
      "step": 72900
    },
    {
      "epoch": 0.7416280110677084,
      "grad_norm": 10.23849868774414,
      "learning_rate": 7.808334427782934e-07,
      "loss": 3.5117,
      "step": 72905
    },
    {
      "epoch": 0.7416788736979166,
      "grad_norm": 16.51127815246582,
      "learning_rate": 7.805431405722352e-07,
      "loss": 3.3629,
      "step": 72910
    },
    {
      "epoch": 0.741729736328125,
      "grad_norm": 12.006612777709961,
      "learning_rate": 7.802528823580738e-07,
      "loss": 3.2699,
      "step": 72915
    },
    {
      "epoch": 0.7417805989583334,
      "grad_norm": 14.054769515991211,
      "learning_rate": 7.799626681432341e-07,
      "loss": 3.0308,
      "step": 72920
    },
    {
      "epoch": 0.7418314615885416,
      "grad_norm": 9.321966171264648,
      "learning_rate": 7.796724979351428e-07,
      "loss": 3.0909,
      "step": 72925
    },
    {
      "epoch": 0.74188232421875,
      "grad_norm": 9.320379257202148,
      "learning_rate": 7.793823717412224e-07,
      "loss": 3.6585,
      "step": 72930
    },
    {
      "epoch": 0.7419331868489584,
      "grad_norm": 7.190203666687012,
      "learning_rate": 7.79092289568896e-07,
      "loss": 3.2446,
      "step": 72935
    },
    {
      "epoch": 0.7419840494791666,
      "grad_norm": 9.347601890563965,
      "learning_rate": 7.78802251425585e-07,
      "loss": 3.3853,
      "step": 72940
    },
    {
      "epoch": 0.742034912109375,
      "grad_norm": 11.347188949584961,
      "learning_rate": 7.785122573187115e-07,
      "loss": 3.6131,
      "step": 72945
    },
    {
      "epoch": 0.7420857747395834,
      "grad_norm": 14.256086349487305,
      "learning_rate": 7.782223072556938e-07,
      "loss": 3.1202,
      "step": 72950
    },
    {
      "epoch": 0.7421366373697916,
      "grad_norm": 14.243017196655273,
      "learning_rate": 7.779324012439502e-07,
      "loss": 3.2519,
      "step": 72955
    },
    {
      "epoch": 0.7421875,
      "grad_norm": 7.689399242401123,
      "learning_rate": 7.776425392908971e-07,
      "loss": 3.5394,
      "step": 72960
    },
    {
      "epoch": 0.7422383626302084,
      "grad_norm": 13.22787094116211,
      "learning_rate": 7.773527214039514e-07,
      "loss": 3.5155,
      "step": 72965
    },
    {
      "epoch": 0.7422892252604166,
      "grad_norm": 9.220317840576172,
      "learning_rate": 7.770629475905289e-07,
      "loss": 3.3707,
      "step": 72970
    },
    {
      "epoch": 0.742340087890625,
      "grad_norm": 9.710553169250488,
      "learning_rate": 7.767732178580417e-07,
      "loss": 3.2424,
      "step": 72975
    },
    {
      "epoch": 0.7423909505208334,
      "grad_norm": 14.730201721191406,
      "learning_rate": 7.764835322139039e-07,
      "loss": 3.2154,
      "step": 72980
    },
    {
      "epoch": 0.7424418131510416,
      "grad_norm": 11.835850715637207,
      "learning_rate": 7.761938906655266e-07,
      "loss": 3.3134,
      "step": 72985
    },
    {
      "epoch": 0.74249267578125,
      "grad_norm": 9.529492378234863,
      "learning_rate": 7.759042932203193e-07,
      "loss": 3.0571,
      "step": 72990
    },
    {
      "epoch": 0.7425435384114584,
      "grad_norm": 6.461589813232422,
      "learning_rate": 7.756147398856917e-07,
      "loss": 3.9506,
      "step": 72995
    },
    {
      "epoch": 0.7425944010416666,
      "grad_norm": 16.31830406188965,
      "learning_rate": 7.753252306690532e-07,
      "loss": 3.6828,
      "step": 73000
    },
    {
      "epoch": 0.742645263671875,
      "grad_norm": 8.438053131103516,
      "learning_rate": 7.750357655778098e-07,
      "loss": 3.401,
      "step": 73005
    },
    {
      "epoch": 0.7426961263020834,
      "grad_norm": 9.632655143737793,
      "learning_rate": 7.747463446193676e-07,
      "loss": 3.2164,
      "step": 73010
    },
    {
      "epoch": 0.7427469889322916,
      "grad_norm": 12.210907936096191,
      "learning_rate": 7.744569678011296e-07,
      "loss": 3.3304,
      "step": 73015
    },
    {
      "epoch": 0.7427978515625,
      "grad_norm": 14.016220092773438,
      "learning_rate": 7.741676351305027e-07,
      "loss": 3.6702,
      "step": 73020
    },
    {
      "epoch": 0.7428487141927084,
      "grad_norm": 13.280328750610352,
      "learning_rate": 7.73878346614888e-07,
      "loss": 3.2677,
      "step": 73025
    },
    {
      "epoch": 0.7428995768229166,
      "grad_norm": 8.581869125366211,
      "learning_rate": 7.735891022616864e-07,
      "loss": 3.1247,
      "step": 73030
    },
    {
      "epoch": 0.742950439453125,
      "grad_norm": 12.272854804992676,
      "learning_rate": 7.732999020782978e-07,
      "loss": 3.4134,
      "step": 73035
    },
    {
      "epoch": 0.7430013020833334,
      "grad_norm": 11.38424015045166,
      "learning_rate": 7.730107460721217e-07,
      "loss": 3.4051,
      "step": 73040
    },
    {
      "epoch": 0.7430521647135416,
      "grad_norm": 9.790849685668945,
      "learning_rate": 7.727216342505573e-07,
      "loss": 3.2793,
      "step": 73045
    },
    {
      "epoch": 0.74310302734375,
      "grad_norm": 10.613922119140625,
      "learning_rate": 7.724325666210003e-07,
      "loss": 3.0087,
      "step": 73050
    },
    {
      "epoch": 0.7431538899739584,
      "grad_norm": 7.071836948394775,
      "learning_rate": 7.721435431908458e-07,
      "loss": 3.3846,
      "step": 73055
    },
    {
      "epoch": 0.7432047526041666,
      "grad_norm": 16.532873153686523,
      "learning_rate": 7.718545639674901e-07,
      "loss": 3.2195,
      "step": 73060
    },
    {
      "epoch": 0.743255615234375,
      "grad_norm": 12.714094161987305,
      "learning_rate": 7.715656289583251e-07,
      "loss": 3.2959,
      "step": 73065
    },
    {
      "epoch": 0.7433064778645834,
      "grad_norm": 11.682530403137207,
      "learning_rate": 7.712767381707437e-07,
      "loss": 3.2982,
      "step": 73070
    },
    {
      "epoch": 0.7433573404947916,
      "grad_norm": 9.41639232635498,
      "learning_rate": 7.70987891612138e-07,
      "loss": 3.4517,
      "step": 73075
    },
    {
      "epoch": 0.743408203125,
      "grad_norm": 7.8489227294921875,
      "learning_rate": 7.706990892898972e-07,
      "loss": 3.0582,
      "step": 73080
    },
    {
      "epoch": 0.7434590657552084,
      "grad_norm": 13.319563865661621,
      "learning_rate": 7.704103312114103e-07,
      "loss": 3.3114,
      "step": 73085
    },
    {
      "epoch": 0.7435099283854166,
      "grad_norm": 11.45560359954834,
      "learning_rate": 7.701216173840644e-07,
      "loss": 3.0357,
      "step": 73090
    },
    {
      "epoch": 0.743560791015625,
      "grad_norm": 9.883466720581055,
      "learning_rate": 7.698329478152469e-07,
      "loss": 3.46,
      "step": 73095
    },
    {
      "epoch": 0.7436116536458334,
      "grad_norm": 7.949326515197754,
      "learning_rate": 7.695443225123441e-07,
      "loss": 3.1761,
      "step": 73100
    },
    {
      "epoch": 0.7436625162760416,
      "grad_norm": 8.533098220825195,
      "learning_rate": 7.692557414827395e-07,
      "loss": 3.2972,
      "step": 73105
    },
    {
      "epoch": 0.74371337890625,
      "grad_norm": 17.932315826416016,
      "learning_rate": 7.689672047338159e-07,
      "loss": 3.9118,
      "step": 73110
    },
    {
      "epoch": 0.7437642415364584,
      "grad_norm": 10.43192195892334,
      "learning_rate": 7.686787122729569e-07,
      "loss": 3.4507,
      "step": 73115
    },
    {
      "epoch": 0.7438151041666666,
      "grad_norm": 15.594086647033691,
      "learning_rate": 7.683902641075419e-07,
      "loss": 3.2069,
      "step": 73120
    },
    {
      "epoch": 0.743865966796875,
      "grad_norm": 13.488203048706055,
      "learning_rate": 7.681018602449522e-07,
      "loss": 2.9345,
      "step": 73125
    },
    {
      "epoch": 0.7439168294270834,
      "grad_norm": 15.50025463104248,
      "learning_rate": 7.678135006925652e-07,
      "loss": 3.2471,
      "step": 73130
    },
    {
      "epoch": 0.7439676920572916,
      "grad_norm": 12.578755378723145,
      "learning_rate": 7.6752518545776e-07,
      "loss": 3.4652,
      "step": 73135
    },
    {
      "epoch": 0.7440185546875,
      "grad_norm": 16.930402755737305,
      "learning_rate": 7.672369145479122e-07,
      "loss": 3.2195,
      "step": 73140
    },
    {
      "epoch": 0.7440694173177084,
      "grad_norm": 8.504794120788574,
      "learning_rate": 7.669486879703966e-07,
      "loss": 3.0849,
      "step": 73145
    },
    {
      "epoch": 0.7441202799479166,
      "grad_norm": 12.299077987670898,
      "learning_rate": 7.66660505732588e-07,
      "loss": 3.3146,
      "step": 73150
    },
    {
      "epoch": 0.744171142578125,
      "grad_norm": 11.325913429260254,
      "learning_rate": 7.663723678418602e-07,
      "loss": 3.1139,
      "step": 73155
    },
    {
      "epoch": 0.7442220052083334,
      "grad_norm": 16.42136573791504,
      "learning_rate": 7.660842743055846e-07,
      "loss": 3.4013,
      "step": 73160
    },
    {
      "epoch": 0.7442728678385416,
      "grad_norm": 9.951093673706055,
      "learning_rate": 7.65796225131131e-07,
      "loss": 3.8557,
      "step": 73165
    },
    {
      "epoch": 0.74432373046875,
      "grad_norm": 6.883090019226074,
      "learning_rate": 7.655082203258699e-07,
      "loss": 3.0357,
      "step": 73170
    },
    {
      "epoch": 0.7443745930989584,
      "grad_norm": 13.703225135803223,
      "learning_rate": 7.652202598971706e-07,
      "loss": 3.4455,
      "step": 73175
    },
    {
      "epoch": 0.7444254557291666,
      "grad_norm": 11.733901977539062,
      "learning_rate": 7.649323438523997e-07,
      "loss": 3.3681,
      "step": 73180
    },
    {
      "epoch": 0.744476318359375,
      "grad_norm": 14.60044002532959,
      "learning_rate": 7.64644472198923e-07,
      "loss": 3.135,
      "step": 73185
    },
    {
      "epoch": 0.7445271809895834,
      "grad_norm": 15.660284996032715,
      "learning_rate": 7.643566449441067e-07,
      "loss": 3.5143,
      "step": 73190
    },
    {
      "epoch": 0.7445780436197916,
      "grad_norm": 12.892966270446777,
      "learning_rate": 7.640688620953135e-07,
      "loss": 3.4056,
      "step": 73195
    },
    {
      "epoch": 0.74462890625,
      "grad_norm": 9.747441291809082,
      "learning_rate": 7.637811236599079e-07,
      "loss": 2.9433,
      "step": 73200
    },
    {
      "epoch": 0.7446797688802084,
      "grad_norm": 9.57300090789795,
      "learning_rate": 7.634934296452501e-07,
      "loss": 3.1353,
      "step": 73205
    },
    {
      "epoch": 0.7447306315104166,
      "grad_norm": 13.052162170410156,
      "learning_rate": 7.632057800587017e-07,
      "loss": 3.2355,
      "step": 73210
    },
    {
      "epoch": 0.744781494140625,
      "grad_norm": 7.5890889167785645,
      "learning_rate": 7.62918174907622e-07,
      "loss": 3.483,
      "step": 73215
    },
    {
      "epoch": 0.7448323567708334,
      "grad_norm": 13.81509780883789,
      "learning_rate": 7.626306141993684e-07,
      "loss": 3.4252,
      "step": 73220
    },
    {
      "epoch": 0.7448832194010416,
      "grad_norm": 12.88431167602539,
      "learning_rate": 7.623430979412982e-07,
      "loss": 3.5395,
      "step": 73225
    },
    {
      "epoch": 0.74493408203125,
      "grad_norm": 15.229802131652832,
      "learning_rate": 7.620556261407694e-07,
      "loss": 3.2204,
      "step": 73230
    },
    {
      "epoch": 0.7449849446614584,
      "grad_norm": 10.262614250183105,
      "learning_rate": 7.617681988051351e-07,
      "loss": 3.4247,
      "step": 73235
    },
    {
      "epoch": 0.7450358072916666,
      "grad_norm": 8.711079597473145,
      "learning_rate": 7.614808159417494e-07,
      "loss": 3.5177,
      "step": 73240
    },
    {
      "epoch": 0.745086669921875,
      "grad_norm": 11.199665069580078,
      "learning_rate": 7.611934775579644e-07,
      "loss": 3.3516,
      "step": 73245
    },
    {
      "epoch": 0.7451375325520834,
      "grad_norm": 8.549784660339355,
      "learning_rate": 7.609061836611317e-07,
      "loss": 3.5392,
      "step": 73250
    },
    {
      "epoch": 0.7451883951822916,
      "grad_norm": 15.495569229125977,
      "learning_rate": 7.606189342586034e-07,
      "loss": 3.5441,
      "step": 73255
    },
    {
      "epoch": 0.7452392578125,
      "grad_norm": 12.841856956481934,
      "learning_rate": 7.60331729357727e-07,
      "loss": 3.2102,
      "step": 73260
    },
    {
      "epoch": 0.7452901204427084,
      "grad_norm": 17.090049743652344,
      "learning_rate": 7.600445689658503e-07,
      "loss": 3.2777,
      "step": 73265
    },
    {
      "epoch": 0.7453409830729166,
      "grad_norm": 14.31063461303711,
      "learning_rate": 7.597574530903218e-07,
      "loss": 3.5972,
      "step": 73270
    },
    {
      "epoch": 0.745391845703125,
      "grad_norm": 8.161468505859375,
      "learning_rate": 7.594703817384855e-07,
      "loss": 3.2841,
      "step": 73275
    },
    {
      "epoch": 0.7454427083333334,
      "grad_norm": 8.875266075134277,
      "learning_rate": 7.591833549176869e-07,
      "loss": 3.4786,
      "step": 73280
    },
    {
      "epoch": 0.7454935709635416,
      "grad_norm": 15.609048843383789,
      "learning_rate": 7.588963726352702e-07,
      "loss": 3.1179,
      "step": 73285
    },
    {
      "epoch": 0.74554443359375,
      "grad_norm": 12.657333374023438,
      "learning_rate": 7.586094348985773e-07,
      "loss": 3.2278,
      "step": 73290
    },
    {
      "epoch": 0.7455952962239584,
      "grad_norm": 10.095488548278809,
      "learning_rate": 7.583225417149492e-07,
      "loss": 3.1679,
      "step": 73295
    },
    {
      "epoch": 0.7456461588541666,
      "grad_norm": 12.093870162963867,
      "learning_rate": 7.580356930917254e-07,
      "loss": 3.4705,
      "step": 73300
    },
    {
      "epoch": 0.745697021484375,
      "grad_norm": 11.52402114868164,
      "learning_rate": 7.577488890362456e-07,
      "loss": 3.2479,
      "step": 73305
    },
    {
      "epoch": 0.7457478841145834,
      "grad_norm": 12.448471069335938,
      "learning_rate": 7.574621295558479e-07,
      "loss": 3.5321,
      "step": 73310
    },
    {
      "epoch": 0.7457987467447916,
      "grad_norm": 13.753480911254883,
      "learning_rate": 7.571754146578689e-07,
      "loss": 3.161,
      "step": 73315
    },
    {
      "epoch": 0.745849609375,
      "grad_norm": 9.864426612854004,
      "learning_rate": 7.568887443496429e-07,
      "loss": 3.6129,
      "step": 73320
    },
    {
      "epoch": 0.7459004720052084,
      "grad_norm": 11.091754913330078,
      "learning_rate": 7.56602118638505e-07,
      "loss": 3.4282,
      "step": 73325
    },
    {
      "epoch": 0.7459513346354166,
      "grad_norm": 9.379804611206055,
      "learning_rate": 7.563155375317898e-07,
      "loss": 3.1408,
      "step": 73330
    },
    {
      "epoch": 0.746002197265625,
      "grad_norm": 7.995944023132324,
      "learning_rate": 7.560290010368281e-07,
      "loss": 3.2759,
      "step": 73335
    },
    {
      "epoch": 0.7460530598958334,
      "grad_norm": 11.444838523864746,
      "learning_rate": 7.557425091609505e-07,
      "loss": 3.4338,
      "step": 73340
    },
    {
      "epoch": 0.7461039225260416,
      "grad_norm": 16.494482040405273,
      "learning_rate": 7.554560619114879e-07,
      "loss": 3.5272,
      "step": 73345
    },
    {
      "epoch": 0.74615478515625,
      "grad_norm": 9.846036911010742,
      "learning_rate": 7.551696592957677e-07,
      "loss": 3.4849,
      "step": 73350
    },
    {
      "epoch": 0.7462056477864584,
      "grad_norm": 9.586414337158203,
      "learning_rate": 7.54883301321119e-07,
      "loss": 3.5259,
      "step": 73355
    },
    {
      "epoch": 0.7462565104166666,
      "grad_norm": 9.466014862060547,
      "learning_rate": 7.545969879948667e-07,
      "loss": 3.3932,
      "step": 73360
    },
    {
      "epoch": 0.746307373046875,
      "grad_norm": 11.308871269226074,
      "learning_rate": 7.543107193243373e-07,
      "loss": 3.8975,
      "step": 73365
    },
    {
      "epoch": 0.7463582356770834,
      "grad_norm": 15.013957023620605,
      "learning_rate": 7.540244953168546e-07,
      "loss": 3.5655,
      "step": 73370
    },
    {
      "epoch": 0.7464090983072916,
      "grad_norm": 8.602811813354492,
      "learning_rate": 7.537383159797404e-07,
      "loss": 3.1833,
      "step": 73375
    },
    {
      "epoch": 0.7464599609375,
      "grad_norm": 10.58324909210205,
      "learning_rate": 7.534521813203177e-07,
      "loss": 3.2111,
      "step": 73380
    },
    {
      "epoch": 0.7465108235677084,
      "grad_norm": 9.13379192352295,
      "learning_rate": 7.531660913459077e-07,
      "loss": 3.2012,
      "step": 73385
    },
    {
      "epoch": 0.7465616861979166,
      "grad_norm": 14.996956825256348,
      "learning_rate": 7.528800460638292e-07,
      "loss": 2.994,
      "step": 73390
    },
    {
      "epoch": 0.746612548828125,
      "grad_norm": 13.060670852661133,
      "learning_rate": 7.525940454813996e-07,
      "loss": 3.3944,
      "step": 73395
    },
    {
      "epoch": 0.7466634114583334,
      "grad_norm": 13.515923500061035,
      "learning_rate": 7.523080896059381e-07,
      "loss": 3.6887,
      "step": 73400
    },
    {
      "epoch": 0.7467142740885416,
      "grad_norm": 12.800006866455078,
      "learning_rate": 7.52022178444759e-07,
      "loss": 3.1361,
      "step": 73405
    },
    {
      "epoch": 0.74676513671875,
      "grad_norm": 11.275392532348633,
      "learning_rate": 7.51736312005179e-07,
      "loss": 3.5173,
      "step": 73410
    },
    {
      "epoch": 0.7468159993489584,
      "grad_norm": 13.225747108459473,
      "learning_rate": 7.514504902945105e-07,
      "loss": 3.4082,
      "step": 73415
    },
    {
      "epoch": 0.7468668619791666,
      "grad_norm": 13.548460960388184,
      "learning_rate": 7.511647133200673e-07,
      "loss": 3.2836,
      "step": 73420
    },
    {
      "epoch": 0.746917724609375,
      "grad_norm": 14.141571998596191,
      "learning_rate": 7.508789810891603e-07,
      "loss": 3.8894,
      "step": 73425
    },
    {
      "epoch": 0.7469685872395834,
      "grad_norm": 11.883014678955078,
      "learning_rate": 7.505932936090996e-07,
      "loss": 3.2863,
      "step": 73430
    },
    {
      "epoch": 0.7470194498697916,
      "grad_norm": 13.795418739318848,
      "learning_rate": 7.503076508871945e-07,
      "loss": 3.3979,
      "step": 73435
    },
    {
      "epoch": 0.7470703125,
      "grad_norm": 15.13593864440918,
      "learning_rate": 7.500220529307542e-07,
      "loss": 3.3507,
      "step": 73440
    },
    {
      "epoch": 0.7471211751302084,
      "grad_norm": 7.632009983062744,
      "learning_rate": 7.497364997470852e-07,
      "loss": 3.2965,
      "step": 73445
    },
    {
      "epoch": 0.7471720377604166,
      "grad_norm": 10.563254356384277,
      "learning_rate": 7.494509913434928e-07,
      "loss": 3.2197,
      "step": 73450
    },
    {
      "epoch": 0.747222900390625,
      "grad_norm": 11.43966293334961,
      "learning_rate": 7.491655277272808e-07,
      "loss": 3.508,
      "step": 73455
    },
    {
      "epoch": 0.7472737630208334,
      "grad_norm": 9.749238967895508,
      "learning_rate": 7.48880108905754e-07,
      "loss": 3.1205,
      "step": 73460
    },
    {
      "epoch": 0.7473246256510416,
      "grad_norm": 14.747815132141113,
      "learning_rate": 7.485947348862155e-07,
      "loss": 3.606,
      "step": 73465
    },
    {
      "epoch": 0.74737548828125,
      "grad_norm": 9.424822807312012,
      "learning_rate": 7.483094056759646e-07,
      "loss": 3.0415,
      "step": 73470
    },
    {
      "epoch": 0.7474263509114584,
      "grad_norm": 10.191957473754883,
      "learning_rate": 7.480241212823033e-07,
      "loss": 3.3108,
      "step": 73475
    },
    {
      "epoch": 0.7474772135416666,
      "grad_norm": 12.010482788085938,
      "learning_rate": 7.477388817125286e-07,
      "loss": 3.7169,
      "step": 73480
    },
    {
      "epoch": 0.747528076171875,
      "grad_norm": 11.904157638549805,
      "learning_rate": 7.474536869739402e-07,
      "loss": 3.1726,
      "step": 73485
    },
    {
      "epoch": 0.7475789388020834,
      "grad_norm": 12.716765403747559,
      "learning_rate": 7.471685370738329e-07,
      "loss": 3.7189,
      "step": 73490
    },
    {
      "epoch": 0.7476298014322916,
      "grad_norm": 7.9622578620910645,
      "learning_rate": 7.468834320195042e-07,
      "loss": 3.3852,
      "step": 73495
    },
    {
      "epoch": 0.7476806640625,
      "grad_norm": 9.606403350830078,
      "learning_rate": 7.465983718182471e-07,
      "loss": 3.2574,
      "step": 73500
    },
    {
      "epoch": 0.7477315266927084,
      "grad_norm": 13.480175971984863,
      "learning_rate": 7.463133564773544e-07,
      "loss": 3.423,
      "step": 73505
    },
    {
      "epoch": 0.7477823893229166,
      "grad_norm": 11.57226276397705,
      "learning_rate": 7.460283860041187e-07,
      "loss": 3.2411,
      "step": 73510
    },
    {
      "epoch": 0.747833251953125,
      "grad_norm": 13.545355796813965,
      "learning_rate": 7.457434604058319e-07,
      "loss": 3.6054,
      "step": 73515
    },
    {
      "epoch": 0.7478841145833334,
      "grad_norm": 12.120404243469238,
      "learning_rate": 7.454585796897829e-07,
      "loss": 3.1793,
      "step": 73520
    },
    {
      "epoch": 0.7479349772135416,
      "grad_norm": 9.005751609802246,
      "learning_rate": 7.4517374386326e-07,
      "loss": 3.4021,
      "step": 73525
    },
    {
      "epoch": 0.74798583984375,
      "grad_norm": 18.179729461669922,
      "learning_rate": 7.448889529335504e-07,
      "loss": 3.595,
      "step": 73530
    },
    {
      "epoch": 0.7480367024739584,
      "grad_norm": 12.268835067749023,
      "learning_rate": 7.446042069079407e-07,
      "loss": 3.1909,
      "step": 73535
    },
    {
      "epoch": 0.7480875651041666,
      "grad_norm": 9.217876434326172,
      "learning_rate": 7.443195057937172e-07,
      "loss": 3.4967,
      "step": 73540
    },
    {
      "epoch": 0.748138427734375,
      "grad_norm": 13.504807472229004,
      "learning_rate": 7.440348495981631e-07,
      "loss": 3.445,
      "step": 73545
    },
    {
      "epoch": 0.7481892903645834,
      "grad_norm": 10.301555633544922,
      "learning_rate": 7.437502383285603e-07,
      "loss": 3.5344,
      "step": 73550
    },
    {
      "epoch": 0.7482401529947916,
      "grad_norm": 15.819905281066895,
      "learning_rate": 7.434656719921921e-07,
      "loss": 3.5132,
      "step": 73555
    },
    {
      "epoch": 0.748291015625,
      "grad_norm": 13.971081733703613,
      "learning_rate": 7.431811505963376e-07,
      "loss": 3.3255,
      "step": 73560
    },
    {
      "epoch": 0.7483418782552084,
      "grad_norm": 9.28151798248291,
      "learning_rate": 7.428966741482771e-07,
      "loss": 3.2326,
      "step": 73565
    },
    {
      "epoch": 0.7483927408854166,
      "grad_norm": 14.84122371673584,
      "learning_rate": 7.426122426552893e-07,
      "loss": 3.566,
      "step": 73570
    },
    {
      "epoch": 0.748443603515625,
      "grad_norm": 13.488147735595703,
      "learning_rate": 7.423278561246508e-07,
      "loss": 3.5507,
      "step": 73575
    },
    {
      "epoch": 0.7484944661458334,
      "grad_norm": 17.054447174072266,
      "learning_rate": 7.420435145636376e-07,
      "loss": 3.4997,
      "step": 73580
    },
    {
      "epoch": 0.7485453287760416,
      "grad_norm": 14.33040714263916,
      "learning_rate": 7.417592179795233e-07,
      "loss": 3.1852,
      "step": 73585
    },
    {
      "epoch": 0.74859619140625,
      "grad_norm": 13.27672004699707,
      "learning_rate": 7.414749663795828e-07,
      "loss": 3.2606,
      "step": 73590
    },
    {
      "epoch": 0.7486470540364584,
      "grad_norm": 14.843589782714844,
      "learning_rate": 7.411907597710894e-07,
      "loss": 3.3071,
      "step": 73595
    },
    {
      "epoch": 0.7486979166666666,
      "grad_norm": 11.863862037658691,
      "learning_rate": 7.409065981613134e-07,
      "loss": 2.9452,
      "step": 73600
    },
    {
      "epoch": 0.748748779296875,
      "grad_norm": 12.077499389648438,
      "learning_rate": 7.406224815575245e-07,
      "loss": 3.3808,
      "step": 73605
    },
    {
      "epoch": 0.7487996419270834,
      "grad_norm": 15.173077583312988,
      "learning_rate": 7.403384099669931e-07,
      "loss": 3.3957,
      "step": 73610
    },
    {
      "epoch": 0.7488505045572916,
      "grad_norm": 14.847683906555176,
      "learning_rate": 7.400543833969853e-07,
      "loss": 3.4315,
      "step": 73615
    },
    {
      "epoch": 0.7489013671875,
      "grad_norm": 8.362418174743652,
      "learning_rate": 7.397704018547703e-07,
      "loss": 3.2964,
      "step": 73620
    },
    {
      "epoch": 0.7489522298177084,
      "grad_norm": 13.943264961242676,
      "learning_rate": 7.394864653476111e-07,
      "loss": 3.3526,
      "step": 73625
    },
    {
      "epoch": 0.7490030924479166,
      "grad_norm": 7.470756530761719,
      "learning_rate": 7.392025738827743e-07,
      "loss": 3.4958,
      "step": 73630
    },
    {
      "epoch": 0.749053955078125,
      "grad_norm": 9.26888656616211,
      "learning_rate": 7.389187274675216e-07,
      "loss": 3.5424,
      "step": 73635
    },
    {
      "epoch": 0.7491048177083334,
      "grad_norm": 12.975512504577637,
      "learning_rate": 7.386349261091167e-07,
      "loss": 3.2929,
      "step": 73640
    },
    {
      "epoch": 0.7491556803385416,
      "grad_norm": 14.786529541015625,
      "learning_rate": 7.383511698148185e-07,
      "loss": 3.2,
      "step": 73645
    },
    {
      "epoch": 0.74920654296875,
      "grad_norm": 12.205375671386719,
      "learning_rate": 7.380674585918892e-07,
      "loss": 3.3362,
      "step": 73650
    },
    {
      "epoch": 0.7492574055989584,
      "grad_norm": 13.627872467041016,
      "learning_rate": 7.377837924475864e-07,
      "loss": 3.5033,
      "step": 73655
    },
    {
      "epoch": 0.7493082682291666,
      "grad_norm": 13.841710090637207,
      "learning_rate": 7.375001713891667e-07,
      "loss": 3.6456,
      "step": 73660
    },
    {
      "epoch": 0.749359130859375,
      "grad_norm": 8.06103801727295,
      "learning_rate": 7.372165954238874e-07,
      "loss": 3.5724,
      "step": 73665
    },
    {
      "epoch": 0.7494099934895834,
      "grad_norm": 12.296360969543457,
      "learning_rate": 7.369330645590044e-07,
      "loss": 3.2789,
      "step": 73670
    },
    {
      "epoch": 0.7494608561197916,
      "grad_norm": 13.1085786819458,
      "learning_rate": 7.36649578801771e-07,
      "loss": 3.1036,
      "step": 73675
    },
    {
      "epoch": 0.74951171875,
      "grad_norm": 16.37004280090332,
      "learning_rate": 7.363661381594395e-07,
      "loss": 3.3069,
      "step": 73680
    },
    {
      "epoch": 0.7495625813802084,
      "grad_norm": 12.110971450805664,
      "learning_rate": 7.360827426392631e-07,
      "loss": 3.4275,
      "step": 73685
    },
    {
      "epoch": 0.7496134440104166,
      "grad_norm": 16.157642364501953,
      "learning_rate": 7.357993922484907e-07,
      "loss": 3.5252,
      "step": 73690
    },
    {
      "epoch": 0.749664306640625,
      "grad_norm": 13.888794898986816,
      "learning_rate": 7.355160869943737e-07,
      "loss": 3.1457,
      "step": 73695
    },
    {
      "epoch": 0.7497151692708334,
      "grad_norm": 15.48414421081543,
      "learning_rate": 7.352328268841585e-07,
      "loss": 3.3848,
      "step": 73700
    },
    {
      "epoch": 0.7497660319010416,
      "grad_norm": 7.177224636077881,
      "learning_rate": 7.34949611925094e-07,
      "loss": 3.3622,
      "step": 73705
    },
    {
      "epoch": 0.74981689453125,
      "grad_norm": 11.376160621643066,
      "learning_rate": 7.346664421244251e-07,
      "loss": 3.2707,
      "step": 73710
    },
    {
      "epoch": 0.7498677571614584,
      "grad_norm": 14.925772666931152,
      "learning_rate": 7.343833174893961e-07,
      "loss": 3.0642,
      "step": 73715
    },
    {
      "epoch": 0.7499186197916666,
      "grad_norm": 10.566452026367188,
      "learning_rate": 7.341002380272516e-07,
      "loss": 3.0275,
      "step": 73720
    },
    {
      "epoch": 0.749969482421875,
      "grad_norm": 11.44359302520752,
      "learning_rate": 7.338172037452345e-07,
      "loss": 3.669,
      "step": 73725
    },
    {
      "epoch": 0.75,
      "eval_loss": 3.3423798084259033,
      "eval_runtime": 159.4795,
      "eval_samples_per_second": 12.585,
      "eval_steps_per_second": 12.585,
      "step": 73728
    },
    {
      "epoch": 0.7500203450520834,
      "grad_norm": 14.166296005249023,
      "learning_rate": 7.335342146505859e-07,
      "loss": 3.3664,
      "step": 73730
    },
    {
      "epoch": 0.7500712076822916,
      "grad_norm": 10.570517539978027,
      "learning_rate": 7.332512707505457e-07,
      "loss": 3.4589,
      "step": 73735
    },
    {
      "epoch": 0.7501220703125,
      "grad_norm": 14.14139175415039,
      "learning_rate": 7.32968372052352e-07,
      "loss": 3.563,
      "step": 73740
    },
    {
      "epoch": 0.7501729329427084,
      "grad_norm": 15.443755149841309,
      "learning_rate": 7.326855185632436e-07,
      "loss": 3.2833,
      "step": 73745
    },
    {
      "epoch": 0.7502237955729166,
      "grad_norm": 8.380440711975098,
      "learning_rate": 7.324027102904582e-07,
      "loss": 3.0448,
      "step": 73750
    },
    {
      "epoch": 0.750274658203125,
      "grad_norm": 12.28552532196045,
      "learning_rate": 7.321199472412299e-07,
      "loss": 3.5558,
      "step": 73755
    },
    {
      "epoch": 0.7503255208333334,
      "grad_norm": 10.05579662322998,
      "learning_rate": 7.318372294227944e-07,
      "loss": 3.5021,
      "step": 73760
    },
    {
      "epoch": 0.7503763834635416,
      "grad_norm": 10.5604248046875,
      "learning_rate": 7.31554556842384e-07,
      "loss": 3.4582,
      "step": 73765
    },
    {
      "epoch": 0.75042724609375,
      "grad_norm": 13.349933624267578,
      "learning_rate": 7.312719295072307e-07,
      "loss": 3.1165,
      "step": 73770
    },
    {
      "epoch": 0.7504781087239584,
      "grad_norm": 17.509366989135742,
      "learning_rate": 7.309893474245655e-07,
      "loss": 3.3821,
      "step": 73775
    },
    {
      "epoch": 0.7505289713541666,
      "grad_norm": 8.016525268554688,
      "learning_rate": 7.307068106016196e-07,
      "loss": 3.3826,
      "step": 73780
    },
    {
      "epoch": 0.750579833984375,
      "grad_norm": 10.776978492736816,
      "learning_rate": 7.304243190456203e-07,
      "loss": 2.9895,
      "step": 73785
    },
    {
      "epoch": 0.7506306966145834,
      "grad_norm": 10.558320045471191,
      "learning_rate": 7.301418727637957e-07,
      "loss": 3.0355,
      "step": 73790
    },
    {
      "epoch": 0.7506815592447916,
      "grad_norm": 10.417439460754395,
      "learning_rate": 7.298594717633701e-07,
      "loss": 3.3279,
      "step": 73795
    },
    {
      "epoch": 0.750732421875,
      "grad_norm": 15.79559326171875,
      "learning_rate": 7.295771160515719e-07,
      "loss": 3.3566,
      "step": 73800
    },
    {
      "epoch": 0.7507832845052084,
      "grad_norm": 10.064557075500488,
      "learning_rate": 7.292948056356236e-07,
      "loss": 3.5683,
      "step": 73805
    },
    {
      "epoch": 0.7508341471354166,
      "grad_norm": 13.68875789642334,
      "learning_rate": 7.290125405227478e-07,
      "loss": 3.172,
      "step": 73810
    },
    {
      "epoch": 0.750885009765625,
      "grad_norm": 10.799705505371094,
      "learning_rate": 7.287303207201657e-07,
      "loss": 3.433,
      "step": 73815
    },
    {
      "epoch": 0.7509358723958334,
      "grad_norm": 16.007572174072266,
      "learning_rate": 7.284481462350984e-07,
      "loss": 3.3508,
      "step": 73820
    },
    {
      "epoch": 0.7509867350260416,
      "grad_norm": 12.162935256958008,
      "learning_rate": 7.281660170747665e-07,
      "loss": 3.274,
      "step": 73825
    },
    {
      "epoch": 0.75103759765625,
      "grad_norm": 12.926222801208496,
      "learning_rate": 7.278839332463869e-07,
      "loss": 3.2844,
      "step": 73830
    },
    {
      "epoch": 0.7510884602864584,
      "grad_norm": 9.41971492767334,
      "learning_rate": 7.276018947571759e-07,
      "loss": 2.9304,
      "step": 73835
    },
    {
      "epoch": 0.7511393229166666,
      "grad_norm": 15.09054946899414,
      "learning_rate": 7.273199016143512e-07,
      "loss": 3.4695,
      "step": 73840
    },
    {
      "epoch": 0.751190185546875,
      "grad_norm": 10.432167053222656,
      "learning_rate": 7.270379538251259e-07,
      "loss": 3.6547,
      "step": 73845
    },
    {
      "epoch": 0.7512410481770834,
      "grad_norm": 10.546064376831055,
      "learning_rate": 7.267560513967151e-07,
      "loss": 3.3185,
      "step": 73850
    },
    {
      "epoch": 0.7512919108072916,
      "grad_norm": 10.503418922424316,
      "learning_rate": 7.264741943363296e-07,
      "loss": 3.227,
      "step": 73855
    },
    {
      "epoch": 0.7513427734375,
      "grad_norm": 16.155641555786133,
      "learning_rate": 7.261923826511821e-07,
      "loss": 3.07,
      "step": 73860
    },
    {
      "epoch": 0.7513936360677084,
      "grad_norm": 10.841660499572754,
      "learning_rate": 7.259106163484825e-07,
      "loss": 3.5704,
      "step": 73865
    },
    {
      "epoch": 0.7514444986979166,
      "grad_norm": 15.498570442199707,
      "learning_rate": 7.256288954354382e-07,
      "loss": 3.3501,
      "step": 73870
    },
    {
      "epoch": 0.751495361328125,
      "grad_norm": 10.249595642089844,
      "learning_rate": 7.253472199192579e-07,
      "loss": 3.7279,
      "step": 73875
    },
    {
      "epoch": 0.7515462239583334,
      "grad_norm": 11.629800796508789,
      "learning_rate": 7.250655898071493e-07,
      "loss": 3.2096,
      "step": 73880
    },
    {
      "epoch": 0.7515970865885416,
      "grad_norm": 11.37413501739502,
      "learning_rate": 7.247840051063169e-07,
      "loss": 3.292,
      "step": 73885
    },
    {
      "epoch": 0.75164794921875,
      "grad_norm": 8.198013305664062,
      "learning_rate": 7.245024658239641e-07,
      "loss": 3.1457,
      "step": 73890
    },
    {
      "epoch": 0.7516988118489584,
      "grad_norm": 11.581289291381836,
      "learning_rate": 7.242209719672958e-07,
      "loss": 3.0306,
      "step": 73895
    },
    {
      "epoch": 0.7517496744791666,
      "grad_norm": 14.45716381072998,
      "learning_rate": 7.23939523543512e-07,
      "loss": 3.0793,
      "step": 73900
    },
    {
      "epoch": 0.751800537109375,
      "grad_norm": 13.524023056030273,
      "learning_rate": 7.236581205598154e-07,
      "loss": 3.2285,
      "step": 73905
    },
    {
      "epoch": 0.7518513997395834,
      "grad_norm": 10.072444915771484,
      "learning_rate": 7.23376763023404e-07,
      "loss": 3.1458,
      "step": 73910
    },
    {
      "epoch": 0.7519022623697916,
      "grad_norm": 10.554051399230957,
      "learning_rate": 7.230954509414778e-07,
      "loss": 3.4933,
      "step": 73915
    },
    {
      "epoch": 0.751953125,
      "grad_norm": 10.31640911102295,
      "learning_rate": 7.228141843212333e-07,
      "loss": 3.2726,
      "step": 73920
    },
    {
      "epoch": 0.7520039876302084,
      "grad_norm": 14.159457206726074,
      "learning_rate": 7.225329631698658e-07,
      "loss": 2.9675,
      "step": 73925
    },
    {
      "epoch": 0.7520548502604166,
      "grad_norm": 14.815949440002441,
      "learning_rate": 7.222517874945712e-07,
      "loss": 3.5891,
      "step": 73930
    },
    {
      "epoch": 0.752105712890625,
      "grad_norm": 12.3479585647583,
      "learning_rate": 7.21970657302544e-07,
      "loss": 3.1654,
      "step": 73935
    },
    {
      "epoch": 0.7521565755208334,
      "grad_norm": 10.186405181884766,
      "learning_rate": 7.216895726009762e-07,
      "loss": 3.558,
      "step": 73940
    },
    {
      "epoch": 0.7522074381510416,
      "grad_norm": 12.260489463806152,
      "learning_rate": 7.214085333970591e-07,
      "loss": 3.6643,
      "step": 73945
    },
    {
      "epoch": 0.75225830078125,
      "grad_norm": 12.001791954040527,
      "learning_rate": 7.211275396979814e-07,
      "loss": 3.4012,
      "step": 73950
    },
    {
      "epoch": 0.7523091634114584,
      "grad_norm": 13.754630088806152,
      "learning_rate": 7.208465915109359e-07,
      "loss": 3.5265,
      "step": 73955
    },
    {
      "epoch": 0.7523600260416666,
      "grad_norm": 9.70177936553955,
      "learning_rate": 7.205656888431081e-07,
      "loss": 3.2286,
      "step": 73960
    },
    {
      "epoch": 0.752410888671875,
      "grad_norm": 9.411014556884766,
      "learning_rate": 7.20284831701685e-07,
      "loss": 3.4721,
      "step": 73965
    },
    {
      "epoch": 0.7524617513020834,
      "grad_norm": 11.240584373474121,
      "learning_rate": 7.200040200938532e-07,
      "loss": 3.3887,
      "step": 73970
    },
    {
      "epoch": 0.7525126139322916,
      "grad_norm": 10.180984497070312,
      "learning_rate": 7.197232540267962e-07,
      "loss": 3.3888,
      "step": 73975
    },
    {
      "epoch": 0.7525634765625,
      "grad_norm": 10.874776840209961,
      "learning_rate": 7.194425335076982e-07,
      "loss": 3.2853,
      "step": 73980
    },
    {
      "epoch": 0.7526143391927084,
      "grad_norm": 7.493716716766357,
      "learning_rate": 7.191618585437405e-07,
      "loss": 3.2489,
      "step": 73985
    },
    {
      "epoch": 0.7526652018229166,
      "grad_norm": 11.833789825439453,
      "learning_rate": 7.188812291421049e-07,
      "loss": 3.3848,
      "step": 73990
    },
    {
      "epoch": 0.752716064453125,
      "grad_norm": 12.204221725463867,
      "learning_rate": 7.186006453099712e-07,
      "loss": 3.3684,
      "step": 73995
    },
    {
      "epoch": 0.7527669270833334,
      "grad_norm": 10.87446117401123,
      "learning_rate": 7.18320107054517e-07,
      "loss": 3.0759,
      "step": 74000
    },
    {
      "epoch": 0.7528177897135416,
      "grad_norm": 9.297377586364746,
      "learning_rate": 7.180396143829205e-07,
      "loss": 3.5556,
      "step": 74005
    },
    {
      "epoch": 0.75286865234375,
      "grad_norm": 13.512225151062012,
      "learning_rate": 7.177591673023587e-07,
      "loss": 3.3424,
      "step": 74010
    },
    {
      "epoch": 0.7529195149739584,
      "grad_norm": 14.0761079788208,
      "learning_rate": 7.174787658200064e-07,
      "loss": 3.2936,
      "step": 74015
    },
    {
      "epoch": 0.7529703776041666,
      "grad_norm": 12.869799613952637,
      "learning_rate": 7.171984099430371e-07,
      "loss": 3.2506,
      "step": 74020
    },
    {
      "epoch": 0.753021240234375,
      "grad_norm": 9.71020793914795,
      "learning_rate": 7.169180996786232e-07,
      "loss": 3.2979,
      "step": 74025
    },
    {
      "epoch": 0.7530721028645834,
      "grad_norm": 7.054031848907471,
      "learning_rate": 7.166378350339373e-07,
      "loss": 2.9761,
      "step": 74030
    },
    {
      "epoch": 0.7531229654947916,
      "grad_norm": 13.224422454833984,
      "learning_rate": 7.163576160161503e-07,
      "loss": 3.2398,
      "step": 74035
    },
    {
      "epoch": 0.753173828125,
      "grad_norm": 10.546746253967285,
      "learning_rate": 7.16077442632431e-07,
      "loss": 3.0981,
      "step": 74040
    },
    {
      "epoch": 0.7532246907552084,
      "grad_norm": 12.613469123840332,
      "learning_rate": 7.157973148899466e-07,
      "loss": 3.3591,
      "step": 74045
    },
    {
      "epoch": 0.7532755533854166,
      "grad_norm": 13.701570510864258,
      "learning_rate": 7.155172327958659e-07,
      "loss": 3.3919,
      "step": 74050
    },
    {
      "epoch": 0.753326416015625,
      "grad_norm": 7.327104091644287,
      "learning_rate": 7.152371963573529e-07,
      "loss": 3.4793,
      "step": 74055
    },
    {
      "epoch": 0.7533772786458334,
      "grad_norm": 10.980306625366211,
      "learning_rate": 7.149572055815735e-07,
      "loss": 3.3033,
      "step": 74060
    },
    {
      "epoch": 0.7534281412760416,
      "grad_norm": 13.810663223266602,
      "learning_rate": 7.146772604756913e-07,
      "loss": 3.6748,
      "step": 74065
    },
    {
      "epoch": 0.75347900390625,
      "grad_norm": 11.452163696289062,
      "learning_rate": 7.143973610468685e-07,
      "loss": 3.6406,
      "step": 74070
    },
    {
      "epoch": 0.7535298665364584,
      "grad_norm": 13.183918952941895,
      "learning_rate": 7.141175073022658e-07,
      "loss": 3.7848,
      "step": 74075
    },
    {
      "epoch": 0.7535807291666666,
      "grad_norm": 7.297231197357178,
      "learning_rate": 7.138376992490425e-07,
      "loss": 3.8056,
      "step": 74080
    },
    {
      "epoch": 0.753631591796875,
      "grad_norm": 12.576817512512207,
      "learning_rate": 7.135579368943587e-07,
      "loss": 3.5714,
      "step": 74085
    },
    {
      "epoch": 0.7536824544270834,
      "grad_norm": 11.886563301086426,
      "learning_rate": 7.132782202453722e-07,
      "loss": 3.1065,
      "step": 74090
    },
    {
      "epoch": 0.7537333170572916,
      "grad_norm": 10.314664840698242,
      "learning_rate": 7.129985493092389e-07,
      "loss": 3.461,
      "step": 74095
    },
    {
      "epoch": 0.7537841796875,
      "grad_norm": 7.800176620483398,
      "learning_rate": 7.127189240931135e-07,
      "loss": 3.4499,
      "step": 74100
    },
    {
      "epoch": 0.7538350423177084,
      "grad_norm": 11.081335067749023,
      "learning_rate": 7.124393446041505e-07,
      "loss": 3.1305,
      "step": 74105
    },
    {
      "epoch": 0.7538859049479166,
      "grad_norm": 14.861065864562988,
      "learning_rate": 7.121598108495042e-07,
      "loss": 3.4695,
      "step": 74110
    },
    {
      "epoch": 0.753936767578125,
      "grad_norm": 8.500097274780273,
      "learning_rate": 7.118803228363255e-07,
      "loss": 3.2133,
      "step": 74115
    },
    {
      "epoch": 0.7539876302083334,
      "grad_norm": 10.504951477050781,
      "learning_rate": 7.116008805717639e-07,
      "loss": 3.1917,
      "step": 74120
    },
    {
      "epoch": 0.7540384928385416,
      "grad_norm": 9.952469825744629,
      "learning_rate": 7.113214840629706e-07,
      "loss": 3.6916,
      "step": 74125
    },
    {
      "epoch": 0.75408935546875,
      "grad_norm": 16.40432357788086,
      "learning_rate": 7.110421333170925e-07,
      "loss": 3.5647,
      "step": 74130
    },
    {
      "epoch": 0.7541402180989584,
      "grad_norm": 17.41621208190918,
      "learning_rate": 7.107628283412782e-07,
      "loss": 3.1238,
      "step": 74135
    },
    {
      "epoch": 0.7541910807291666,
      "grad_norm": 9.299099922180176,
      "learning_rate": 7.10483569142672e-07,
      "loss": 3.1899,
      "step": 74140
    },
    {
      "epoch": 0.754241943359375,
      "grad_norm": 9.889762878417969,
      "learning_rate": 7.102043557284205e-07,
      "loss": 3.2099,
      "step": 74145
    },
    {
      "epoch": 0.7542928059895834,
      "grad_norm": 8.289566040039062,
      "learning_rate": 7.09925188105666e-07,
      "loss": 3.2947,
      "step": 74150
    },
    {
      "epoch": 0.7543436686197916,
      "grad_norm": 15.631021499633789,
      "learning_rate": 7.096460662815508e-07,
      "loss": 3.3074,
      "step": 74155
    },
    {
      "epoch": 0.75439453125,
      "grad_norm": 8.1798095703125,
      "learning_rate": 7.093669902632164e-07,
      "loss": 3.825,
      "step": 74160
    },
    {
      "epoch": 0.7544453938802084,
      "grad_norm": 14.07030200958252,
      "learning_rate": 7.09087960057804e-07,
      "loss": 3.3914,
      "step": 74165
    },
    {
      "epoch": 0.7544962565104166,
      "grad_norm": 14.111145973205566,
      "learning_rate": 7.088089756724517e-07,
      "loss": 3.5105,
      "step": 74170
    },
    {
      "epoch": 0.754547119140625,
      "grad_norm": 14.813468933105469,
      "learning_rate": 7.085300371142962e-07,
      "loss": 3.2479,
      "step": 74175
    },
    {
      "epoch": 0.7545979817708334,
      "grad_norm": 13.250635147094727,
      "learning_rate": 7.08251144390476e-07,
      "loss": 2.969,
      "step": 74180
    },
    {
      "epoch": 0.7546488444010416,
      "grad_norm": 12.24089241027832,
      "learning_rate": 7.079722975081249e-07,
      "loss": 3.4867,
      "step": 74185
    },
    {
      "epoch": 0.75469970703125,
      "grad_norm": 8.176153182983398,
      "learning_rate": 7.076934964743784e-07,
      "loss": 2.9599,
      "step": 74190
    },
    {
      "epoch": 0.7547505696614584,
      "grad_norm": 8.18547534942627,
      "learning_rate": 7.074147412963684e-07,
      "loss": 3.4086,
      "step": 74195
    },
    {
      "epoch": 0.7548014322916666,
      "grad_norm": 10.30460262298584,
      "learning_rate": 7.071360319812282e-07,
      "loss": 3.382,
      "step": 74200
    },
    {
      "epoch": 0.754852294921875,
      "grad_norm": 10.885030746459961,
      "learning_rate": 7.068573685360875e-07,
      "loss": 3.1091,
      "step": 74205
    },
    {
      "epoch": 0.7549031575520834,
      "grad_norm": 17.012645721435547,
      "learning_rate": 7.065787509680749e-07,
      "loss": 2.9846,
      "step": 74210
    },
    {
      "epoch": 0.7549540201822916,
      "grad_norm": 10.468649864196777,
      "learning_rate": 7.063001792843202e-07,
      "loss": 3.1097,
      "step": 74215
    },
    {
      "epoch": 0.7550048828125,
      "grad_norm": 7.468170166015625,
      "learning_rate": 7.060216534919512e-07,
      "loss": 3.4211,
      "step": 74220
    },
    {
      "epoch": 0.7550557454427084,
      "grad_norm": 14.28119945526123,
      "learning_rate": 7.057431735980927e-07,
      "loss": 3.601,
      "step": 74225
    },
    {
      "epoch": 0.7551066080729166,
      "grad_norm": 7.192823886871338,
      "learning_rate": 7.054647396098699e-07,
      "loss": 3.1028,
      "step": 74230
    },
    {
      "epoch": 0.755157470703125,
      "grad_norm": 15.882479667663574,
      "learning_rate": 7.051863515344057e-07,
      "loss": 3.4561,
      "step": 74235
    },
    {
      "epoch": 0.7552083333333334,
      "grad_norm": 10.127427101135254,
      "learning_rate": 7.049080093788233e-07,
      "loss": 3.4995,
      "step": 74240
    },
    {
      "epoch": 0.7552591959635416,
      "grad_norm": 12.84577751159668,
      "learning_rate": 7.046297131502447e-07,
      "loss": 3.7194,
      "step": 74245
    },
    {
      "epoch": 0.75531005859375,
      "grad_norm": 15.184442520141602,
      "learning_rate": 7.043514628557885e-07,
      "loss": 3.2178,
      "step": 74250
    },
    {
      "epoch": 0.7553609212239584,
      "grad_norm": 7.603909492492676,
      "learning_rate": 7.040732585025753e-07,
      "loss": 3.4013,
      "step": 74255
    },
    {
      "epoch": 0.7554117838541666,
      "grad_norm": 10.720600128173828,
      "learning_rate": 7.037951000977214e-07,
      "loss": 3.4708,
      "step": 74260
    },
    {
      "epoch": 0.755462646484375,
      "grad_norm": 8.314825057983398,
      "learning_rate": 7.035169876483447e-07,
      "loss": 3.5037,
      "step": 74265
    },
    {
      "epoch": 0.7555135091145834,
      "grad_norm": 10.812227249145508,
      "learning_rate": 7.032389211615595e-07,
      "loss": 3.3262,
      "step": 74270
    },
    {
      "epoch": 0.7555643717447916,
      "grad_norm": 9.320564270019531,
      "learning_rate": 7.029609006444812e-07,
      "loss": 3.554,
      "step": 74275
    },
    {
      "epoch": 0.755615234375,
      "grad_norm": 15.716340065002441,
      "learning_rate": 7.026829261042226e-07,
      "loss": 3.1336,
      "step": 74280
    },
    {
      "epoch": 0.7556660970052084,
      "grad_norm": 9.545758247375488,
      "learning_rate": 7.024049975478944e-07,
      "loss": 3.3385,
      "step": 74285
    },
    {
      "epoch": 0.7557169596354166,
      "grad_norm": 17.533021926879883,
      "learning_rate": 7.021271149826081e-07,
      "loss": 3.3169,
      "step": 74290
    },
    {
      "epoch": 0.755767822265625,
      "grad_norm": 13.390016555786133,
      "learning_rate": 7.018492784154746e-07,
      "loss": 3.7757,
      "step": 74295
    },
    {
      "epoch": 0.7558186848958334,
      "grad_norm": 8.253308296203613,
      "learning_rate": 7.015714878536006e-07,
      "loss": 3.3973,
      "step": 74300
    },
    {
      "epoch": 0.7558695475260416,
      "grad_norm": 15.03423023223877,
      "learning_rate": 7.012937433040942e-07,
      "loss": 3.3413,
      "step": 74305
    },
    {
      "epoch": 0.75592041015625,
      "grad_norm": 12.773712158203125,
      "learning_rate": 7.010160447740602e-07,
      "loss": 3.1559,
      "step": 74310
    },
    {
      "epoch": 0.7559712727864584,
      "grad_norm": 9.488603591918945,
      "learning_rate": 7.00738392270604e-07,
      "loss": 3.4735,
      "step": 74315
    },
    {
      "epoch": 0.7560221354166666,
      "grad_norm": 8.902047157287598,
      "learning_rate": 7.004607858008305e-07,
      "loss": 3.3458,
      "step": 74320
    },
    {
      "epoch": 0.756072998046875,
      "grad_norm": 12.052464485168457,
      "learning_rate": 7.001832253718415e-07,
      "loss": 3.789,
      "step": 74325
    },
    {
      "epoch": 0.7561238606770834,
      "grad_norm": 11.685837745666504,
      "learning_rate": 6.99905710990737e-07,
      "loss": 3.4724,
      "step": 74330
    },
    {
      "epoch": 0.7561747233072916,
      "grad_norm": 8.351921081542969,
      "learning_rate": 6.996282426646192e-07,
      "loss": 3.1278,
      "step": 74335
    },
    {
      "epoch": 0.7562255859375,
      "grad_norm": 12.978894233703613,
      "learning_rate": 6.993508204005853e-07,
      "loss": 3.1652,
      "step": 74340
    },
    {
      "epoch": 0.7562764485677084,
      "grad_norm": 11.092260360717773,
      "learning_rate": 6.990734442057337e-07,
      "loss": 3.0752,
      "step": 74345
    },
    {
      "epoch": 0.7563273111979166,
      "grad_norm": 9.3390531539917,
      "learning_rate": 6.987961140871619e-07,
      "loss": 3.3607,
      "step": 74350
    },
    {
      "epoch": 0.756378173828125,
      "grad_norm": 10.838726043701172,
      "learning_rate": 6.985188300519646e-07,
      "loss": 3.5749,
      "step": 74355
    },
    {
      "epoch": 0.7564290364583334,
      "grad_norm": 12.331197738647461,
      "learning_rate": 6.982415921072361e-07,
      "loss": 3.3488,
      "step": 74360
    },
    {
      "epoch": 0.7564798990885416,
      "grad_norm": 9.617509841918945,
      "learning_rate": 6.979644002600688e-07,
      "loss": 3.4219,
      "step": 74365
    },
    {
      "epoch": 0.75653076171875,
      "grad_norm": 14.07290267944336,
      "learning_rate": 6.976872545175551e-07,
      "loss": 3.3708,
      "step": 74370
    },
    {
      "epoch": 0.7565816243489584,
      "grad_norm": 11.84440803527832,
      "learning_rate": 6.974101548867865e-07,
      "loss": 3.142,
      "step": 74375
    },
    {
      "epoch": 0.7566324869791666,
      "grad_norm": 10.46172046661377,
      "learning_rate": 6.971331013748516e-07,
      "loss": 3.2734,
      "step": 74380
    },
    {
      "epoch": 0.756683349609375,
      "grad_norm": 9.306661605834961,
      "learning_rate": 6.968560939888386e-07,
      "loss": 3.2955,
      "step": 74385
    },
    {
      "epoch": 0.7567342122395834,
      "grad_norm": 9.832356452941895,
      "learning_rate": 6.965791327358357e-07,
      "loss": 3.3325,
      "step": 74390
    },
    {
      "epoch": 0.7567850748697916,
      "grad_norm": 8.555065155029297,
      "learning_rate": 6.963022176229273e-07,
      "loss": 3.4083,
      "step": 74395
    },
    {
      "epoch": 0.7568359375,
      "grad_norm": 13.239713668823242,
      "learning_rate": 6.960253486572e-07,
      "loss": 3.0292,
      "step": 74400
    },
    {
      "epoch": 0.7568868001302084,
      "grad_norm": 9.089715003967285,
      "learning_rate": 6.957485258457358e-07,
      "loss": 3.4664,
      "step": 74405
    },
    {
      "epoch": 0.7569376627604166,
      "grad_norm": 14.721821784973145,
      "learning_rate": 6.954717491956189e-07,
      "loss": 3.4906,
      "step": 74410
    },
    {
      "epoch": 0.756988525390625,
      "grad_norm": 12.5114107131958,
      "learning_rate": 6.951950187139287e-07,
      "loss": 3.1409,
      "step": 74415
    },
    {
      "epoch": 0.7570393880208334,
      "grad_norm": 11.109739303588867,
      "learning_rate": 6.949183344077468e-07,
      "loss": 3.4038,
      "step": 74420
    },
    {
      "epoch": 0.7570902506510416,
      "grad_norm": 8.996869087219238,
      "learning_rate": 6.946416962841507e-07,
      "loss": 3.247,
      "step": 74425
    },
    {
      "epoch": 0.75714111328125,
      "grad_norm": 14.235173225402832,
      "learning_rate": 6.943651043502194e-07,
      "loss": 3.4906,
      "step": 74430
    },
    {
      "epoch": 0.7571919759114584,
      "grad_norm": 7.575634002685547,
      "learning_rate": 6.940885586130295e-07,
      "loss": 3.1856,
      "step": 74435
    },
    {
      "epoch": 0.7572428385416666,
      "grad_norm": 16.175704956054688,
      "learning_rate": 6.938120590796546e-07,
      "loss": 3.3288,
      "step": 74440
    },
    {
      "epoch": 0.757293701171875,
      "grad_norm": 14.900594711303711,
      "learning_rate": 6.935356057571702e-07,
      "loss": 3.4744,
      "step": 74445
    },
    {
      "epoch": 0.7573445638020834,
      "grad_norm": 14.816061973571777,
      "learning_rate": 6.932591986526499e-07,
      "loss": 3.6081,
      "step": 74450
    },
    {
      "epoch": 0.7573954264322916,
      "grad_norm": 11.891322135925293,
      "learning_rate": 6.929828377731649e-07,
      "loss": 3.2964,
      "step": 74455
    },
    {
      "epoch": 0.7574462890625,
      "grad_norm": 14.090703964233398,
      "learning_rate": 6.927065231257846e-07,
      "loss": 3.371,
      "step": 74460
    },
    {
      "epoch": 0.7574971516927084,
      "grad_norm": 12.144478797912598,
      "learning_rate": 6.924302547175805e-07,
      "loss": 3.3383,
      "step": 74465
    },
    {
      "epoch": 0.7575480143229166,
      "grad_norm": 12.723793983459473,
      "learning_rate": 6.921540325556189e-07,
      "loss": 3.2933,
      "step": 74470
    },
    {
      "epoch": 0.757598876953125,
      "grad_norm": 15.725260734558105,
      "learning_rate": 6.918778566469689e-07,
      "loss": 3.362,
      "step": 74475
    },
    {
      "epoch": 0.7576497395833334,
      "grad_norm": 11.462191581726074,
      "learning_rate": 6.916017269986946e-07,
      "loss": 3.4661,
      "step": 74480
    },
    {
      "epoch": 0.7577006022135416,
      "grad_norm": 11.932019233703613,
      "learning_rate": 6.913256436178623e-07,
      "loss": 3.2483,
      "step": 74485
    },
    {
      "epoch": 0.75775146484375,
      "grad_norm": 11.867484092712402,
      "learning_rate": 6.910496065115344e-07,
      "loss": 4.0898,
      "step": 74490
    },
    {
      "epoch": 0.7578023274739584,
      "grad_norm": 12.38962459564209,
      "learning_rate": 6.907736156867731e-07,
      "loss": 2.8809,
      "step": 74495
    },
    {
      "epoch": 0.7578531901041666,
      "grad_norm": 10.69157600402832,
      "learning_rate": 6.904976711506397e-07,
      "loss": 3.4891,
      "step": 74500
    },
    {
      "epoch": 0.757904052734375,
      "grad_norm": 10.238336563110352,
      "learning_rate": 6.902217729101954e-07,
      "loss": 3.4282,
      "step": 74505
    },
    {
      "epoch": 0.7579549153645834,
      "grad_norm": 16.624544143676758,
      "learning_rate": 6.899459209724982e-07,
      "loss": 3.5674,
      "step": 74510
    },
    {
      "epoch": 0.7580057779947916,
      "grad_norm": 15.39177417755127,
      "learning_rate": 6.896701153446056e-07,
      "loss": 3.376,
      "step": 74515
    },
    {
      "epoch": 0.758056640625,
      "grad_norm": 13.53531551361084,
      "learning_rate": 6.893943560335731e-07,
      "loss": 3.1336,
      "step": 74520
    },
    {
      "epoch": 0.7581075032552084,
      "grad_norm": 15.663147926330566,
      "learning_rate": 6.891186430464567e-07,
      "loss": 3.2657,
      "step": 74525
    },
    {
      "epoch": 0.7581583658854166,
      "grad_norm": 16.453989028930664,
      "learning_rate": 6.888429763903115e-07,
      "loss": 3.0325,
      "step": 74530
    },
    {
      "epoch": 0.758209228515625,
      "grad_norm": 7.302393913269043,
      "learning_rate": 6.885673560721895e-07,
      "loss": 3.2141,
      "step": 74535
    },
    {
      "epoch": 0.7582600911458334,
      "grad_norm": 16.061988830566406,
      "learning_rate": 6.882917820991417e-07,
      "loss": 3.0839,
      "step": 74540
    },
    {
      "epoch": 0.7583109537760416,
      "grad_norm": 10.075870513916016,
      "learning_rate": 6.880162544782196e-07,
      "loss": 3.4304,
      "step": 74545
    },
    {
      "epoch": 0.75836181640625,
      "grad_norm": 13.291833877563477,
      "learning_rate": 6.877407732164718e-07,
      "loss": 3.4443,
      "step": 74550
    },
    {
      "epoch": 0.7584126790364584,
      "grad_norm": 11.27462387084961,
      "learning_rate": 6.874653383209464e-07,
      "loss": 3.2295,
      "step": 74555
    },
    {
      "epoch": 0.7584635416666666,
      "grad_norm": 14.254870414733887,
      "learning_rate": 6.871899497986916e-07,
      "loss": 3.1262,
      "step": 74560
    },
    {
      "epoch": 0.758514404296875,
      "grad_norm": 7.6913251876831055,
      "learning_rate": 6.869146076567524e-07,
      "loss": 3.2504,
      "step": 74565
    },
    {
      "epoch": 0.7585652669270834,
      "grad_norm": 10.593586921691895,
      "learning_rate": 6.866393119021725e-07,
      "loss": 3.1265,
      "step": 74570
    },
    {
      "epoch": 0.7586161295572916,
      "grad_norm": 14.967764854431152,
      "learning_rate": 6.863640625419957e-07,
      "loss": 3.665,
      "step": 74575
    },
    {
      "epoch": 0.7586669921875,
      "grad_norm": 15.087532997131348,
      "learning_rate": 6.860888595832654e-07,
      "loss": 3.13,
      "step": 74580
    },
    {
      "epoch": 0.7587178548177084,
      "grad_norm": 13.718179702758789,
      "learning_rate": 6.858137030330217e-07,
      "loss": 3.2086,
      "step": 74585
    },
    {
      "epoch": 0.7587687174479166,
      "grad_norm": 11.125592231750488,
      "learning_rate": 6.855385928983043e-07,
      "loss": 3.0559,
      "step": 74590
    },
    {
      "epoch": 0.758819580078125,
      "grad_norm": 7.489245414733887,
      "learning_rate": 6.852635291861512e-07,
      "loss": 3.4474,
      "step": 74595
    },
    {
      "epoch": 0.7588704427083334,
      "grad_norm": 12.32341480255127,
      "learning_rate": 6.849885119036007e-07,
      "loss": 3.1301,
      "step": 74600
    },
    {
      "epoch": 0.7589213053385416,
      "grad_norm": 12.1867036819458,
      "learning_rate": 6.847135410576897e-07,
      "loss": 3.5196,
      "step": 74605
    },
    {
      "epoch": 0.75897216796875,
      "grad_norm": 11.529767990112305,
      "learning_rate": 6.844386166554523e-07,
      "loss": 3.4061,
      "step": 74610
    },
    {
      "epoch": 0.7590230305989584,
      "grad_norm": 11.689623832702637,
      "learning_rate": 6.841637387039219e-07,
      "loss": 3.3633,
      "step": 74615
    },
    {
      "epoch": 0.7590738932291666,
      "grad_norm": 13.131250381469727,
      "learning_rate": 6.838889072101324e-07,
      "loss": 3.399,
      "step": 74620
    },
    {
      "epoch": 0.759124755859375,
      "grad_norm": 7.971841812133789,
      "learning_rate": 6.836141221811144e-07,
      "loss": 3.0893,
      "step": 74625
    },
    {
      "epoch": 0.7591756184895834,
      "grad_norm": 11.301935195922852,
      "learning_rate": 6.833393836238989e-07,
      "loss": 3.3561,
      "step": 74630
    },
    {
      "epoch": 0.7592264811197916,
      "grad_norm": 15.302087783813477,
      "learning_rate": 6.830646915455141e-07,
      "loss": 3.3727,
      "step": 74635
    },
    {
      "epoch": 0.75927734375,
      "grad_norm": 9.979668617248535,
      "learning_rate": 6.827900459529893e-07,
      "loss": 3.4487,
      "step": 74640
    },
    {
      "epoch": 0.7593282063802084,
      "grad_norm": 7.658801555633545,
      "learning_rate": 6.825154468533504e-07,
      "loss": 3.0497,
      "step": 74645
    },
    {
      "epoch": 0.7593790690104166,
      "grad_norm": 15.597314834594727,
      "learning_rate": 6.822408942536221e-07,
      "loss": 3.431,
      "step": 74650
    },
    {
      "epoch": 0.759429931640625,
      "grad_norm": 12.66695499420166,
      "learning_rate": 6.819663881608299e-07,
      "loss": 3.4224,
      "step": 74655
    },
    {
      "epoch": 0.7594807942708334,
      "grad_norm": 19.707422256469727,
      "learning_rate": 6.816919285819973e-07,
      "loss": 3.1408,
      "step": 74660
    },
    {
      "epoch": 0.7595316569010416,
      "grad_norm": 12.803831100463867,
      "learning_rate": 6.81417515524146e-07,
      "loss": 3.2538,
      "step": 74665
    },
    {
      "epoch": 0.75958251953125,
      "grad_norm": 12.708759307861328,
      "learning_rate": 6.811431489942954e-07,
      "loss": 3.741,
      "step": 74670
    },
    {
      "epoch": 0.7596333821614584,
      "grad_norm": 14.123689651489258,
      "learning_rate": 6.808688289994672e-07,
      "loss": 2.9288,
      "step": 74675
    },
    {
      "epoch": 0.7596842447916666,
      "grad_norm": 9.11545181274414,
      "learning_rate": 6.805945555466781e-07,
      "loss": 3.0954,
      "step": 74680
    },
    {
      "epoch": 0.759735107421875,
      "grad_norm": 17.013774871826172,
      "learning_rate": 6.803203286429466e-07,
      "loss": 4.0014,
      "step": 74685
    },
    {
      "epoch": 0.7597859700520834,
      "grad_norm": 14.704337120056152,
      "learning_rate": 6.800461482952875e-07,
      "loss": 3.7858,
      "step": 74690
    },
    {
      "epoch": 0.7598368326822916,
      "grad_norm": 9.8838472366333,
      "learning_rate": 6.79772014510717e-07,
      "loss": 3.0953,
      "step": 74695
    },
    {
      "epoch": 0.7598876953125,
      "grad_norm": 8.987663269042969,
      "learning_rate": 6.794979272962482e-07,
      "loss": 3.3941,
      "step": 74700
    },
    {
      "epoch": 0.7599385579427084,
      "grad_norm": 11.245233535766602,
      "learning_rate": 6.792238866588927e-07,
      "loss": 3.1513,
      "step": 74705
    },
    {
      "epoch": 0.7599894205729166,
      "grad_norm": 7.750741481781006,
      "learning_rate": 6.789498926056626e-07,
      "loss": 3.2866,
      "step": 74710
    },
    {
      "epoch": 0.760040283203125,
      "grad_norm": 10.813319206237793,
      "learning_rate": 6.786759451435684e-07,
      "loss": 3.3133,
      "step": 74715
    },
    {
      "epoch": 0.7600911458333334,
      "grad_norm": 14.955268859863281,
      "learning_rate": 6.784020442796185e-07,
      "loss": 3.7137,
      "step": 74720
    },
    {
      "epoch": 0.7601420084635416,
      "grad_norm": 11.77611255645752,
      "learning_rate": 6.781281900208197e-07,
      "loss": 3.3437,
      "step": 74725
    },
    {
      "epoch": 0.76019287109375,
      "grad_norm": 14.14409065246582,
      "learning_rate": 6.778543823741796e-07,
      "loss": 3.2431,
      "step": 74730
    },
    {
      "epoch": 0.7602437337239584,
      "grad_norm": 13.484630584716797,
      "learning_rate": 6.775806213467037e-07,
      "loss": 3.4794,
      "step": 74735
    },
    {
      "epoch": 0.7602945963541666,
      "grad_norm": 7.718141555786133,
      "learning_rate": 6.773069069453958e-07,
      "loss": 3.475,
      "step": 74740
    },
    {
      "epoch": 0.760345458984375,
      "grad_norm": 13.957528114318848,
      "learning_rate": 6.770332391772578e-07,
      "loss": 3.2199,
      "step": 74745
    },
    {
      "epoch": 0.7603963216145834,
      "grad_norm": 10.419391632080078,
      "learning_rate": 6.767596180492931e-07,
      "loss": 3.2068,
      "step": 74750
    },
    {
      "epoch": 0.7604471842447916,
      "grad_norm": 11.013030052185059,
      "learning_rate": 6.764860435685005e-07,
      "loss": 3.5959,
      "step": 74755
    },
    {
      "epoch": 0.760498046875,
      "grad_norm": 9.451438903808594,
      "learning_rate": 6.762125157418812e-07,
      "loss": 3.5107,
      "step": 74760
    },
    {
      "epoch": 0.7605489095052084,
      "grad_norm": 10.277724266052246,
      "learning_rate": 6.759390345764316e-07,
      "loss": 3.1288,
      "step": 74765
    },
    {
      "epoch": 0.7605997721354166,
      "grad_norm": 14.881711959838867,
      "learning_rate": 6.756656000791503e-07,
      "loss": 3.2657,
      "step": 74770
    },
    {
      "epoch": 0.760650634765625,
      "grad_norm": 7.927708625793457,
      "learning_rate": 6.75392212257032e-07,
      "loss": 3.5849,
      "step": 74775
    },
    {
      "epoch": 0.7607014973958334,
      "grad_norm": 8.746414184570312,
      "learning_rate": 6.751188711170708e-07,
      "loss": 3.3846,
      "step": 74780
    },
    {
      "epoch": 0.7607523600260416,
      "grad_norm": 10.257974624633789,
      "learning_rate": 6.748455766662609e-07,
      "loss": 2.8426,
      "step": 74785
    },
    {
      "epoch": 0.76080322265625,
      "grad_norm": 14.12350082397461,
      "learning_rate": 6.745723289115949e-07,
      "loss": 3.0598,
      "step": 74790
    },
    {
      "epoch": 0.7608540852864584,
      "grad_norm": 12.958858489990234,
      "learning_rate": 6.742991278600633e-07,
      "loss": 3.4875,
      "step": 74795
    },
    {
      "epoch": 0.7609049479166666,
      "grad_norm": 9.312466621398926,
      "learning_rate": 6.740259735186555e-07,
      "loss": 3.0938,
      "step": 74800
    },
    {
      "epoch": 0.760955810546875,
      "grad_norm": 13.80396842956543,
      "learning_rate": 6.737528658943599e-07,
      "loss": 3.4107,
      "step": 74805
    },
    {
      "epoch": 0.7610066731770834,
      "grad_norm": 11.66608715057373,
      "learning_rate": 6.734798049941644e-07,
      "loss": 3.3786,
      "step": 74810
    },
    {
      "epoch": 0.7610575358072916,
      "grad_norm": 14.635140419006348,
      "learning_rate": 6.732067908250555e-07,
      "loss": 3.4482,
      "step": 74815
    },
    {
      "epoch": 0.7611083984375,
      "grad_norm": 8.224568367004395,
      "learning_rate": 6.729338233940183e-07,
      "loss": 3.3317,
      "step": 74820
    },
    {
      "epoch": 0.7611592610677084,
      "grad_norm": 8.623149871826172,
      "learning_rate": 6.726609027080352e-07,
      "loss": 3.3037,
      "step": 74825
    },
    {
      "epoch": 0.7612101236979166,
      "grad_norm": 10.660038948059082,
      "learning_rate": 6.723880287740905e-07,
      "loss": 3.0179,
      "step": 74830
    },
    {
      "epoch": 0.761260986328125,
      "grad_norm": 9.211432456970215,
      "learning_rate": 6.721152015991644e-07,
      "loss": 3.4817,
      "step": 74835
    },
    {
      "epoch": 0.7613118489583334,
      "grad_norm": 11.59577751159668,
      "learning_rate": 6.718424211902375e-07,
      "loss": 3.3023,
      "step": 74840
    },
    {
      "epoch": 0.7613627115885416,
      "grad_norm": 15.24312686920166,
      "learning_rate": 6.715696875542896e-07,
      "loss": 3.388,
      "step": 74845
    },
    {
      "epoch": 0.76141357421875,
      "grad_norm": 12.55749225616455,
      "learning_rate": 6.712970006982977e-07,
      "loss": 3.2893,
      "step": 74850
    },
    {
      "epoch": 0.7614644368489584,
      "grad_norm": 7.277038097381592,
      "learning_rate": 6.710243606292388e-07,
      "loss": 3.3295,
      "step": 74855
    },
    {
      "epoch": 0.7615152994791666,
      "grad_norm": 7.668971538543701,
      "learning_rate": 6.707517673540867e-07,
      "loss": 3.1016,
      "step": 74860
    },
    {
      "epoch": 0.761566162109375,
      "grad_norm": 8.580836296081543,
      "learning_rate": 6.704792208798183e-07,
      "loss": 3.5713,
      "step": 74865
    },
    {
      "epoch": 0.7616170247395834,
      "grad_norm": 7.763233661651611,
      "learning_rate": 6.702067212134056e-07,
      "loss": 3.6885,
      "step": 74870
    },
    {
      "epoch": 0.7616678873697916,
      "grad_norm": 10.616766929626465,
      "learning_rate": 6.699342683618201e-07,
      "loss": 3.5457,
      "step": 74875
    },
    {
      "epoch": 0.76171875,
      "grad_norm": 14.713553428649902,
      "learning_rate": 6.69661862332032e-07,
      "loss": 3.1444,
      "step": 74880
    },
    {
      "epoch": 0.7617696126302084,
      "grad_norm": 12.911894798278809,
      "learning_rate": 6.69389503131011e-07,
      "loss": 3.1633,
      "step": 74885
    },
    {
      "epoch": 0.7618204752604166,
      "grad_norm": 14.415287017822266,
      "learning_rate": 6.691171907657268e-07,
      "loss": 3.0614,
      "step": 74890
    },
    {
      "epoch": 0.761871337890625,
      "grad_norm": 12.509305953979492,
      "learning_rate": 6.68844925243145e-07,
      "loss": 3.2278,
      "step": 74895
    },
    {
      "epoch": 0.7619222005208334,
      "grad_norm": 10.296442985534668,
      "learning_rate": 6.685727065702311e-07,
      "loss": 3.3172,
      "step": 74900
    },
    {
      "epoch": 0.7619730631510416,
      "grad_norm": 9.007887840270996,
      "learning_rate": 6.683005347539512e-07,
      "loss": 3.371,
      "step": 74905
    },
    {
      "epoch": 0.76202392578125,
      "grad_norm": 12.702661514282227,
      "learning_rate": 6.68028409801267e-07,
      "loss": 3.3113,
      "step": 74910
    },
    {
      "epoch": 0.7620747884114584,
      "grad_norm": 13.30461311340332,
      "learning_rate": 6.677563317191429e-07,
      "loss": 3.2866,
      "step": 74915
    },
    {
      "epoch": 0.7621256510416666,
      "grad_norm": 14.909944534301758,
      "learning_rate": 6.674843005145377e-07,
      "loss": 3.3629,
      "step": 74920
    },
    {
      "epoch": 0.762176513671875,
      "grad_norm": 13.380775451660156,
      "learning_rate": 6.672123161944131e-07,
      "loss": 3.2381,
      "step": 74925
    },
    {
      "epoch": 0.7622273763020834,
      "grad_norm": 8.74083423614502,
      "learning_rate": 6.66940378765727e-07,
      "loss": 3.2115,
      "step": 74930
    },
    {
      "epoch": 0.7622782389322916,
      "grad_norm": 8.978840827941895,
      "learning_rate": 6.66668488235436e-07,
      "loss": 3.4606,
      "step": 74935
    },
    {
      "epoch": 0.7623291015625,
      "grad_norm": 11.639257431030273,
      "learning_rate": 6.663966446104972e-07,
      "loss": 3.3473,
      "step": 74940
    },
    {
      "epoch": 0.7623799641927084,
      "grad_norm": 15.828412055969238,
      "learning_rate": 6.661248478978666e-07,
      "loss": 3.4771,
      "step": 74945
    },
    {
      "epoch": 0.7624308268229166,
      "grad_norm": 11.355568885803223,
      "learning_rate": 6.658530981044969e-07,
      "loss": 3.4224,
      "step": 74950
    },
    {
      "epoch": 0.762481689453125,
      "grad_norm": 13.490345001220703,
      "learning_rate": 6.655813952373402e-07,
      "loss": 3.1535,
      "step": 74955
    },
    {
      "epoch": 0.7625325520833334,
      "grad_norm": 8.671171188354492,
      "learning_rate": 6.653097393033495e-07,
      "loss": 3.1378,
      "step": 74960
    },
    {
      "epoch": 0.7625834147135416,
      "grad_norm": 10.636831283569336,
      "learning_rate": 6.650381303094733e-07,
      "loss": 3.2926,
      "step": 74965
    },
    {
      "epoch": 0.76263427734375,
      "grad_norm": 12.167415618896484,
      "learning_rate": 6.647665682626625e-07,
      "loss": 3.218,
      "step": 74970
    },
    {
      "epoch": 0.7626851399739584,
      "grad_norm": 7.13845157623291,
      "learning_rate": 6.644950531698635e-07,
      "loss": 3.1291,
      "step": 74975
    },
    {
      "epoch": 0.7627360026041666,
      "grad_norm": 11.226144790649414,
      "learning_rate": 6.642235850380241e-07,
      "loss": 3.3579,
      "step": 74980
    },
    {
      "epoch": 0.762786865234375,
      "grad_norm": 10.787151336669922,
      "learning_rate": 6.63952163874089e-07,
      "loss": 3.4915,
      "step": 74985
    },
    {
      "epoch": 0.7628377278645834,
      "grad_norm": 12.106456756591797,
      "learning_rate": 6.63680789685002e-07,
      "loss": 3.3662,
      "step": 74990
    },
    {
      "epoch": 0.7628885904947916,
      "grad_norm": 15.901619911193848,
      "learning_rate": 6.634094624777068e-07,
      "loss": 3.5041,
      "step": 74995
    },
    {
      "epoch": 0.762939453125,
      "grad_norm": 12.083741188049316,
      "learning_rate": 6.631381822591457e-07,
      "loss": 3.4699,
      "step": 75000
    },
    {
      "epoch": 0.7629903157552084,
      "grad_norm": 10.314224243164062,
      "learning_rate": 6.62866949036259e-07,
      "loss": 2.9315,
      "step": 75005
    },
    {
      "epoch": 0.7630411783854166,
      "grad_norm": 8.981534957885742,
      "learning_rate": 6.62595762815986e-07,
      "loss": 3.1706,
      "step": 75010
    },
    {
      "epoch": 0.763092041015625,
      "grad_norm": 8.06904125213623,
      "learning_rate": 6.623246236052633e-07,
      "loss": 3.1097,
      "step": 75015
    },
    {
      "epoch": 0.7631429036458334,
      "grad_norm": 13.305237770080566,
      "learning_rate": 6.62053531411031e-07,
      "loss": 3.1732,
      "step": 75020
    },
    {
      "epoch": 0.7631937662760416,
      "grad_norm": 11.285202026367188,
      "learning_rate": 6.617824862402236e-07,
      "loss": 3.4915,
      "step": 75025
    },
    {
      "epoch": 0.76324462890625,
      "grad_norm": 16.577655792236328,
      "learning_rate": 6.615114880997756e-07,
      "loss": 3.1498,
      "step": 75030
    },
    {
      "epoch": 0.7632954915364584,
      "grad_norm": 8.940192222595215,
      "learning_rate": 6.612405369966193e-07,
      "loss": 3.4379,
      "step": 75035
    },
    {
      "epoch": 0.7633463541666666,
      "grad_norm": 13.985937118530273,
      "learning_rate": 6.609696329376883e-07,
      "loss": 3.3368,
      "step": 75040
    },
    {
      "epoch": 0.763397216796875,
      "grad_norm": 13.95054817199707,
      "learning_rate": 6.606987759299142e-07,
      "loss": 3.176,
      "step": 75045
    },
    {
      "epoch": 0.7634480794270834,
      "grad_norm": 10.873615264892578,
      "learning_rate": 6.604279659802251e-07,
      "loss": 3.1594,
      "step": 75050
    },
    {
      "epoch": 0.7634989420572916,
      "grad_norm": 11.51984691619873,
      "learning_rate": 6.601572030955511e-07,
      "loss": 3.1424,
      "step": 75055
    },
    {
      "epoch": 0.7635498046875,
      "grad_norm": 8.51795482635498,
      "learning_rate": 6.598864872828192e-07,
      "loss": 3.2808,
      "step": 75060
    },
    {
      "epoch": 0.7636006673177084,
      "grad_norm": 8.907057762145996,
      "learning_rate": 6.596158185489543e-07,
      "loss": 3.1381,
      "step": 75065
    },
    {
      "epoch": 0.7636515299479166,
      "grad_norm": 9.872109413146973,
      "learning_rate": 6.593451969008827e-07,
      "loss": 3.3564,
      "step": 75070
    },
    {
      "epoch": 0.763702392578125,
      "grad_norm": 9.058270454406738,
      "learning_rate": 6.590746223455288e-07,
      "loss": 3.3251,
      "step": 75075
    },
    {
      "epoch": 0.7637532552083334,
      "grad_norm": 9.94036865234375,
      "learning_rate": 6.588040948898142e-07,
      "loss": 3.3208,
      "step": 75080
    },
    {
      "epoch": 0.7638041178385416,
      "grad_norm": 15.931431770324707,
      "learning_rate": 6.585336145406604e-07,
      "loss": 3.4207,
      "step": 75085
    },
    {
      "epoch": 0.76385498046875,
      "grad_norm": 16.417871475219727,
      "learning_rate": 6.582631813049867e-07,
      "loss": 3.483,
      "step": 75090
    },
    {
      "epoch": 0.7639058430989584,
      "grad_norm": 8.511138916015625,
      "learning_rate": 6.579927951897134e-07,
      "loss": 3.4558,
      "step": 75095
    },
    {
      "epoch": 0.7639567057291666,
      "grad_norm": 9.964131355285645,
      "learning_rate": 6.577224562017582e-07,
      "loss": 3.2744,
      "step": 75100
    },
    {
      "epoch": 0.764007568359375,
      "grad_norm": 9.064841270446777,
      "learning_rate": 6.574521643480372e-07,
      "loss": 3.5231,
      "step": 75105
    },
    {
      "epoch": 0.7640584309895834,
      "grad_norm": 15.779176712036133,
      "learning_rate": 6.571819196354654e-07,
      "loss": 3.0945,
      "step": 75110
    },
    {
      "epoch": 0.7641092936197916,
      "grad_norm": 13.981021881103516,
      "learning_rate": 6.569117220709581e-07,
      "loss": 3.1934,
      "step": 75115
    },
    {
      "epoch": 0.76416015625,
      "grad_norm": 8.831730842590332,
      "learning_rate": 6.566415716614267e-07,
      "loss": 3.2304,
      "step": 75120
    },
    {
      "epoch": 0.7642110188802084,
      "grad_norm": 19.293806076049805,
      "learning_rate": 6.563714684137845e-07,
      "loss": 3.0182,
      "step": 75125
    },
    {
      "epoch": 0.7642618815104166,
      "grad_norm": 14.205883979797363,
      "learning_rate": 6.561014123349404e-07,
      "loss": 2.9684,
      "step": 75130
    },
    {
      "epoch": 0.764312744140625,
      "grad_norm": 6.975743770599365,
      "learning_rate": 6.558314034318056e-07,
      "loss": 3.0413,
      "step": 75135
    },
    {
      "epoch": 0.7643636067708334,
      "grad_norm": 14.651495933532715,
      "learning_rate": 6.555614417112871e-07,
      "loss": 3.4957,
      "step": 75140
    },
    {
      "epoch": 0.7644144694010416,
      "grad_norm": 11.398849487304688,
      "learning_rate": 6.552915271802915e-07,
      "loss": 3.2468,
      "step": 75145
    },
    {
      "epoch": 0.76446533203125,
      "grad_norm": 13.041194915771484,
      "learning_rate": 6.550216598457246e-07,
      "loss": 3.2752,
      "step": 75150
    },
    {
      "epoch": 0.7645161946614584,
      "grad_norm": 9.893031120300293,
      "learning_rate": 6.547518397144919e-07,
      "loss": 3.0951,
      "step": 75155
    },
    {
      "epoch": 0.7645670572916666,
      "grad_norm": 8.85152816772461,
      "learning_rate": 6.544820667934962e-07,
      "loss": 3.0603,
      "step": 75160
    },
    {
      "epoch": 0.764617919921875,
      "grad_norm": 13.413690567016602,
      "learning_rate": 6.542123410896386e-07,
      "loss": 3.5851,
      "step": 75165
    },
    {
      "epoch": 0.7646687825520834,
      "grad_norm": 10.097662925720215,
      "learning_rate": 6.539426626098208e-07,
      "loss": 3.5162,
      "step": 75170
    },
    {
      "epoch": 0.7647196451822916,
      "grad_norm": 10.521745681762695,
      "learning_rate": 6.536730313609433e-07,
      "loss": 3.6233,
      "step": 75175
    },
    {
      "epoch": 0.7647705078125,
      "grad_norm": 9.089445114135742,
      "learning_rate": 6.534034473499031e-07,
      "loss": 3.0321,
      "step": 75180
    },
    {
      "epoch": 0.7648213704427084,
      "grad_norm": 8.658125877380371,
      "learning_rate": 6.531339105835979e-07,
      "loss": 3.5217,
      "step": 75185
    },
    {
      "epoch": 0.7648722330729166,
      "grad_norm": 11.721631050109863,
      "learning_rate": 6.528644210689242e-07,
      "loss": 3.087,
      "step": 75190
    },
    {
      "epoch": 0.764923095703125,
      "grad_norm": 8.90911865234375,
      "learning_rate": 6.525949788127759e-07,
      "loss": 3.2395,
      "step": 75195
    },
    {
      "epoch": 0.7649739583333334,
      "grad_norm": 10.19819450378418,
      "learning_rate": 6.52325583822048e-07,
      "loss": 3.8694,
      "step": 75200
    },
    {
      "epoch": 0.7650248209635416,
      "grad_norm": 9.232595443725586,
      "learning_rate": 6.520562361036311e-07,
      "loss": 3.3014,
      "step": 75205
    },
    {
      "epoch": 0.76507568359375,
      "grad_norm": 11.723776817321777,
      "learning_rate": 6.517869356644182e-07,
      "loss": 3.49,
      "step": 75210
    },
    {
      "epoch": 0.7651265462239584,
      "grad_norm": 11.485301971435547,
      "learning_rate": 6.515176825112987e-07,
      "loss": 3.7529,
      "step": 75215
    },
    {
      "epoch": 0.7651774088541666,
      "grad_norm": 12.483345031738281,
      "learning_rate": 6.512484766511601e-07,
      "loss": 3.5844,
      "step": 75220
    },
    {
      "epoch": 0.765228271484375,
      "grad_norm": 14.69423770904541,
      "learning_rate": 6.509793180908913e-07,
      "loss": 3.4327,
      "step": 75225
    },
    {
      "epoch": 0.7652791341145834,
      "grad_norm": 15.218012809753418,
      "learning_rate": 6.507102068373788e-07,
      "loss": 3.1909,
      "step": 75230
    },
    {
      "epoch": 0.7653299967447916,
      "grad_norm": 13.670184135437012,
      "learning_rate": 6.504411428975078e-07,
      "loss": 3.6414,
      "step": 75235
    },
    {
      "epoch": 0.765380859375,
      "grad_norm": 14.741324424743652,
      "learning_rate": 6.501721262781607e-07,
      "loss": 3.3462,
      "step": 75240
    },
    {
      "epoch": 0.7654317220052084,
      "grad_norm": 8.339617729187012,
      "learning_rate": 6.499031569862221e-07,
      "loss": 3.1363,
      "step": 75245
    },
    {
      "epoch": 0.7654825846354166,
      "grad_norm": 9.616954803466797,
      "learning_rate": 6.49634235028572e-07,
      "loss": 3.2762,
      "step": 75250
    },
    {
      "epoch": 0.765533447265625,
      "grad_norm": 12.235074043273926,
      "learning_rate": 6.49365360412092e-07,
      "loss": 3.6203,
      "step": 75255
    },
    {
      "epoch": 0.7655843098958334,
      "grad_norm": 15.989127159118652,
      "learning_rate": 6.490965331436599e-07,
      "loss": 3.6928,
      "step": 75260
    },
    {
      "epoch": 0.7656351725260416,
      "grad_norm": 8.670943260192871,
      "learning_rate": 6.488277532301551e-07,
      "loss": 3.0947,
      "step": 75265
    },
    {
      "epoch": 0.76568603515625,
      "grad_norm": 12.120124816894531,
      "learning_rate": 6.485590206784534e-07,
      "loss": 3.474,
      "step": 75270
    },
    {
      "epoch": 0.7657368977864584,
      "grad_norm": 16.00341033935547,
      "learning_rate": 6.482903354954298e-07,
      "loss": 3.313,
      "step": 75275
    },
    {
      "epoch": 0.7657877604166666,
      "grad_norm": 15.028971672058105,
      "learning_rate": 6.480216976879589e-07,
      "loss": 3.2195,
      "step": 75280
    },
    {
      "epoch": 0.765838623046875,
      "grad_norm": 15.711023330688477,
      "learning_rate": 6.477531072629145e-07,
      "loss": 3.1773,
      "step": 75285
    },
    {
      "epoch": 0.7658894856770834,
      "grad_norm": 7.689911365509033,
      "learning_rate": 6.47484564227168e-07,
      "loss": 3.4065,
      "step": 75290
    },
    {
      "epoch": 0.7659403483072916,
      "grad_norm": 12.732672691345215,
      "learning_rate": 6.472160685875897e-07,
      "loss": 3.3059,
      "step": 75295
    },
    {
      "epoch": 0.7659912109375,
      "grad_norm": 11.199082374572754,
      "learning_rate": 6.469476203510486e-07,
      "loss": 3.0007,
      "step": 75300
    },
    {
      "epoch": 0.7660420735677084,
      "grad_norm": 10.820151329040527,
      "learning_rate": 6.466792195244131e-07,
      "loss": 3.4755,
      "step": 75305
    },
    {
      "epoch": 0.7660929361979166,
      "grad_norm": 11.690393447875977,
      "learning_rate": 6.464108661145516e-07,
      "loss": 3.3837,
      "step": 75310
    },
    {
      "epoch": 0.766143798828125,
      "grad_norm": 8.426977157592773,
      "learning_rate": 6.461425601283283e-07,
      "loss": 3.1037,
      "step": 75315
    },
    {
      "epoch": 0.7661946614583334,
      "grad_norm": 7.525014400482178,
      "learning_rate": 6.458743015726079e-07,
      "loss": 3.142,
      "step": 75320
    },
    {
      "epoch": 0.7662455240885416,
      "grad_norm": 14.435430526733398,
      "learning_rate": 6.456060904542538e-07,
      "loss": 3.1921,
      "step": 75325
    },
    {
      "epoch": 0.76629638671875,
      "grad_norm": 12.716520309448242,
      "learning_rate": 6.453379267801291e-07,
      "loss": 3.3251,
      "step": 75330
    },
    {
      "epoch": 0.7663472493489584,
      "grad_norm": 12.661298751831055,
      "learning_rate": 6.45069810557093e-07,
      "loss": 3.1351,
      "step": 75335
    },
    {
      "epoch": 0.7663981119791666,
      "grad_norm": 16.14558219909668,
      "learning_rate": 6.448017417920072e-07,
      "loss": 3.1468,
      "step": 75340
    },
    {
      "epoch": 0.766448974609375,
      "grad_norm": 8.328774452209473,
      "learning_rate": 6.44533720491729e-07,
      "loss": 3.0908,
      "step": 75345
    },
    {
      "epoch": 0.7664998372395834,
      "grad_norm": 13.309703826904297,
      "learning_rate": 6.44265746663115e-07,
      "loss": 3.279,
      "step": 75350
    },
    {
      "epoch": 0.7665506998697916,
      "grad_norm": 11.739712715148926,
      "learning_rate": 6.439978203130218e-07,
      "loss": 3.774,
      "step": 75355
    },
    {
      "epoch": 0.7666015625,
      "grad_norm": 11.170668601989746,
      "learning_rate": 6.437299414483056e-07,
      "loss": 3.0828,
      "step": 75360
    },
    {
      "epoch": 0.7666524251302084,
      "grad_norm": 8.549630165100098,
      "learning_rate": 6.434621100758184e-07,
      "loss": 3.4049,
      "step": 75365
    },
    {
      "epoch": 0.7667032877604166,
      "grad_norm": 16.483396530151367,
      "learning_rate": 6.431943262024135e-07,
      "loss": 3.0375,
      "step": 75370
    },
    {
      "epoch": 0.766754150390625,
      "grad_norm": 18.708301544189453,
      "learning_rate": 6.429265898349407e-07,
      "loss": 3.7378,
      "step": 75375
    },
    {
      "epoch": 0.7668050130208334,
      "grad_norm": 8.167903900146484,
      "learning_rate": 6.426589009802509e-07,
      "loss": 3.1887,
      "step": 75380
    },
    {
      "epoch": 0.7668558756510416,
      "grad_norm": 11.62201976776123,
      "learning_rate": 6.423912596451937e-07,
      "loss": 3.66,
      "step": 75385
    },
    {
      "epoch": 0.76690673828125,
      "grad_norm": 13.677452087402344,
      "learning_rate": 6.421236658366159e-07,
      "loss": 3.2397,
      "step": 75390
    },
    {
      "epoch": 0.7669576009114584,
      "grad_norm": 16.496095657348633,
      "learning_rate": 6.418561195613629e-07,
      "loss": 3.3751,
      "step": 75395
    },
    {
      "epoch": 0.7670084635416666,
      "grad_norm": 14.67243480682373,
      "learning_rate": 6.415886208262814e-07,
      "loss": 3.3761,
      "step": 75400
    },
    {
      "epoch": 0.767059326171875,
      "grad_norm": 11.018709182739258,
      "learning_rate": 6.413211696382137e-07,
      "loss": 3.3401,
      "step": 75405
    },
    {
      "epoch": 0.7671101888020834,
      "grad_norm": 15.72050666809082,
      "learning_rate": 6.410537660040042e-07,
      "loss": 3.1229,
      "step": 75410
    },
    {
      "epoch": 0.7671610514322916,
      "grad_norm": 8.102333068847656,
      "learning_rate": 6.407864099304928e-07,
      "loss": 3.4936,
      "step": 75415
    },
    {
      "epoch": 0.7672119140625,
      "grad_norm": 14.650758743286133,
      "learning_rate": 6.40519101424521e-07,
      "loss": 3.7129,
      "step": 75420
    },
    {
      "epoch": 0.7672627766927084,
      "grad_norm": 15.552038192749023,
      "learning_rate": 6.402518404929273e-07,
      "loss": 3.7393,
      "step": 75425
    },
    {
      "epoch": 0.7673136393229166,
      "grad_norm": 14.366085052490234,
      "learning_rate": 6.399846271425486e-07,
      "loss": 3.3615,
      "step": 75430
    },
    {
      "epoch": 0.767364501953125,
      "grad_norm": 7.687265872955322,
      "learning_rate": 6.397174613802226e-07,
      "loss": 3.0453,
      "step": 75435
    },
    {
      "epoch": 0.7674153645833334,
      "grad_norm": 13.427799224853516,
      "learning_rate": 6.394503432127849e-07,
      "loss": 3.2276,
      "step": 75440
    },
    {
      "epoch": 0.7674662272135416,
      "grad_norm": 16.668745040893555,
      "learning_rate": 6.391832726470693e-07,
      "loss": 2.894,
      "step": 75445
    },
    {
      "epoch": 0.76751708984375,
      "grad_norm": 10.8818941116333,
      "learning_rate": 6.389162496899079e-07,
      "loss": 3.3841,
      "step": 75450
    },
    {
      "epoch": 0.7675679524739584,
      "grad_norm": 12.166550636291504,
      "learning_rate": 6.386492743481337e-07,
      "loss": 3.178,
      "step": 75455
    },
    {
      "epoch": 0.7676188151041666,
      "grad_norm": 14.593305587768555,
      "learning_rate": 6.38382346628576e-07,
      "loss": 3.5089,
      "step": 75460
    },
    {
      "epoch": 0.767669677734375,
      "grad_norm": 9.180642127990723,
      "learning_rate": 6.381154665380654e-07,
      "loss": 3.5605,
      "step": 75465
    },
    {
      "epoch": 0.7677205403645834,
      "grad_norm": 11.039069175720215,
      "learning_rate": 6.378486340834286e-07,
      "loss": 3.3149,
      "step": 75470
    },
    {
      "epoch": 0.7677714029947916,
      "grad_norm": 12.083093643188477,
      "learning_rate": 6.375818492714938e-07,
      "loss": 3.237,
      "step": 75475
    },
    {
      "epoch": 0.767822265625,
      "grad_norm": 10.207786560058594,
      "learning_rate": 6.373151121090853e-07,
      "loss": 3.4434,
      "step": 75480
    },
    {
      "epoch": 0.7678731282552084,
      "grad_norm": 13.316230773925781,
      "learning_rate": 6.370484226030291e-07,
      "loss": 3.2465,
      "step": 75485
    },
    {
      "epoch": 0.7679239908854166,
      "grad_norm": 16.586711883544922,
      "learning_rate": 6.367817807601462e-07,
      "loss": 3.2579,
      "step": 75490
    },
    {
      "epoch": 0.767974853515625,
      "grad_norm": 14.756383895874023,
      "learning_rate": 6.36515186587261e-07,
      "loss": 3.8092,
      "step": 75495
    },
    {
      "epoch": 0.7680257161458334,
      "grad_norm": 12.652300834655762,
      "learning_rate": 6.362486400911927e-07,
      "loss": 3.3194,
      "step": 75500
    },
    {
      "epoch": 0.7680765787760416,
      "grad_norm": 12.614615440368652,
      "learning_rate": 6.359821412787606e-07,
      "loss": 3.0277,
      "step": 75505
    },
    {
      "epoch": 0.76812744140625,
      "grad_norm": 9.340450286865234,
      "learning_rate": 6.357156901567837e-07,
      "loss": 3.5647,
      "step": 75510
    },
    {
      "epoch": 0.7681783040364584,
      "grad_norm": 10.129216194152832,
      "learning_rate": 6.354492867320794e-07,
      "loss": 3.2587,
      "step": 75515
    },
    {
      "epoch": 0.7682291666666666,
      "grad_norm": 16.417810440063477,
      "learning_rate": 6.351829310114635e-07,
      "loss": 3.3869,
      "step": 75520
    },
    {
      "epoch": 0.768280029296875,
      "grad_norm": 11.393208503723145,
      "learning_rate": 6.349166230017501e-07,
      "loss": 3.363,
      "step": 75525
    },
    {
      "epoch": 0.7683308919270834,
      "grad_norm": 9.716226577758789,
      "learning_rate": 6.346503627097522e-07,
      "loss": 3.0523,
      "step": 75530
    },
    {
      "epoch": 0.7683817545572916,
      "grad_norm": 13.312603950500488,
      "learning_rate": 6.343841501422826e-07,
      "loss": 3.4017,
      "step": 75535
    },
    {
      "epoch": 0.7684326171875,
      "grad_norm": 9.42683219909668,
      "learning_rate": 6.34117985306153e-07,
      "loss": 3.1427,
      "step": 75540
    },
    {
      "epoch": 0.7684834798177084,
      "grad_norm": 11.141545295715332,
      "learning_rate": 6.338518682081718e-07,
      "loss": 3.4514,
      "step": 75545
    },
    {
      "epoch": 0.7685343424479166,
      "grad_norm": 16.647235870361328,
      "learning_rate": 6.335857988551489e-07,
      "loss": 3.3977,
      "step": 75550
    },
    {
      "epoch": 0.768585205078125,
      "grad_norm": 10.697653770446777,
      "learning_rate": 6.33319777253891e-07,
      "loss": 3.4897,
      "step": 75555
    },
    {
      "epoch": 0.7686360677083334,
      "grad_norm": 10.337068557739258,
      "learning_rate": 6.330538034112035e-07,
      "loss": 3.5939,
      "step": 75560
    },
    {
      "epoch": 0.7686869303385416,
      "grad_norm": 12.20175552368164,
      "learning_rate": 6.327878773338919e-07,
      "loss": 3.5823,
      "step": 75565
    },
    {
      "epoch": 0.76873779296875,
      "grad_norm": 15.400045394897461,
      "learning_rate": 6.325219990287604e-07,
      "loss": 3.8783,
      "step": 75570
    },
    {
      "epoch": 0.7687886555989584,
      "grad_norm": 13.569960594177246,
      "learning_rate": 6.322561685026112e-07,
      "loss": 3.0406,
      "step": 75575
    },
    {
      "epoch": 0.7688395182291666,
      "grad_norm": 14.378377914428711,
      "learning_rate": 6.319903857622451e-07,
      "loss": 2.861,
      "step": 75580
    },
    {
      "epoch": 0.768890380859375,
      "grad_norm": 13.627618789672852,
      "learning_rate": 6.317246508144614e-07,
      "loss": 3.1944,
      "step": 75585
    },
    {
      "epoch": 0.7689412434895834,
      "grad_norm": 8.510849952697754,
      "learning_rate": 6.314589636660598e-07,
      "loss": 3.1475,
      "step": 75590
    },
    {
      "epoch": 0.7689921061197916,
      "grad_norm": 15.343955993652344,
      "learning_rate": 6.311933243238383e-07,
      "loss": 3.0304,
      "step": 75595
    },
    {
      "epoch": 0.76904296875,
      "grad_norm": 15.795459747314453,
      "learning_rate": 6.309277327945929e-07,
      "loss": 3.2039,
      "step": 75600
    },
    {
      "epoch": 0.7690938313802084,
      "grad_norm": 14.032368659973145,
      "learning_rate": 6.306621890851175e-07,
      "loss": 3.5281,
      "step": 75605
    },
    {
      "epoch": 0.7691446940104166,
      "grad_norm": 13.588349342346191,
      "learning_rate": 6.303966932022076e-07,
      "loss": 2.9674,
      "step": 75610
    },
    {
      "epoch": 0.769195556640625,
      "grad_norm": 10.120952606201172,
      "learning_rate": 6.301312451526545e-07,
      "loss": 3.4561,
      "step": 75615
    },
    {
      "epoch": 0.7692464192708334,
      "grad_norm": 10.869763374328613,
      "learning_rate": 6.298658449432513e-07,
      "loss": 3.1571,
      "step": 75620
    },
    {
      "epoch": 0.7692972819010416,
      "grad_norm": 12.92966365814209,
      "learning_rate": 6.296004925807861e-07,
      "loss": 3.2906,
      "step": 75625
    },
    {
      "epoch": 0.76934814453125,
      "grad_norm": 13.265793800354004,
      "learning_rate": 6.293351880720497e-07,
      "loss": 3.5597,
      "step": 75630
    },
    {
      "epoch": 0.7693990071614584,
      "grad_norm": 9.10334587097168,
      "learning_rate": 6.290699314238294e-07,
      "loss": 3.1268,
      "step": 75635
    },
    {
      "epoch": 0.7694498697916666,
      "grad_norm": 8.346096992492676,
      "learning_rate": 6.288047226429098e-07,
      "loss": 3.198,
      "step": 75640
    },
    {
      "epoch": 0.769500732421875,
      "grad_norm": 9.279213905334473,
      "learning_rate": 6.285395617360793e-07,
      "loss": 3.5123,
      "step": 75645
    },
    {
      "epoch": 0.7695515950520834,
      "grad_norm": 9.381355285644531,
      "learning_rate": 6.282744487101203e-07,
      "loss": 3.2821,
      "step": 75650
    },
    {
      "epoch": 0.7696024576822916,
      "grad_norm": 15.720799446105957,
      "learning_rate": 6.280093835718163e-07,
      "loss": 3.9323,
      "step": 75655
    },
    {
      "epoch": 0.7696533203125,
      "grad_norm": 10.18575668334961,
      "learning_rate": 6.277443663279476e-07,
      "loss": 3.3029,
      "step": 75660
    },
    {
      "epoch": 0.7697041829427084,
      "grad_norm": 9.984713554382324,
      "learning_rate": 6.274793969852954e-07,
      "loss": 3.5297,
      "step": 75665
    },
    {
      "epoch": 0.7697550455729166,
      "grad_norm": 8.69368839263916,
      "learning_rate": 6.272144755506398e-07,
      "loss": 3.4563,
      "step": 75670
    },
    {
      "epoch": 0.769805908203125,
      "grad_norm": 14.534234046936035,
      "learning_rate": 6.26949602030758e-07,
      "loss": 3.1021,
      "step": 75675
    },
    {
      "epoch": 0.7698567708333334,
      "grad_norm": 8.975381851196289,
      "learning_rate": 6.26684776432426e-07,
      "loss": 3.4122,
      "step": 75680
    },
    {
      "epoch": 0.7699076334635416,
      "grad_norm": 13.761950492858887,
      "learning_rate": 6.264199987624206e-07,
      "loss": 3.6279,
      "step": 75685
    },
    {
      "epoch": 0.76995849609375,
      "grad_norm": 14.036504745483398,
      "learning_rate": 6.261552690275147e-07,
      "loss": 3.4036,
      "step": 75690
    },
    {
      "epoch": 0.7700093587239584,
      "grad_norm": 10.96495532989502,
      "learning_rate": 6.258905872344828e-07,
      "loss": 3.5949,
      "step": 75695
    },
    {
      "epoch": 0.7700602213541666,
      "grad_norm": 14.842350959777832,
      "learning_rate": 6.256259533900954e-07,
      "loss": 3.301,
      "step": 75700
    },
    {
      "epoch": 0.770111083984375,
      "grad_norm": 7.5997490882873535,
      "learning_rate": 6.253613675011244e-07,
      "loss": 3.4585,
      "step": 75705
    },
    {
      "epoch": 0.7701619466145834,
      "grad_norm": 11.933259963989258,
      "learning_rate": 6.250968295743384e-07,
      "loss": 3.3573,
      "step": 75710
    },
    {
      "epoch": 0.7702128092447916,
      "grad_norm": 15.504426956176758,
      "learning_rate": 6.248323396165048e-07,
      "loss": 3.3533,
      "step": 75715
    },
    {
      "epoch": 0.770263671875,
      "grad_norm": 10.376163482666016,
      "learning_rate": 6.245678976343916e-07,
      "loss": 3.15,
      "step": 75720
    },
    {
      "epoch": 0.7703145345052084,
      "grad_norm": 13.387959480285645,
      "learning_rate": 6.243035036347647e-07,
      "loss": 3.9129,
      "step": 75725
    },
    {
      "epoch": 0.7703653971354166,
      "grad_norm": 8.265640258789062,
      "learning_rate": 6.240391576243881e-07,
      "loss": 3.6994,
      "step": 75730
    },
    {
      "epoch": 0.770416259765625,
      "grad_norm": 15.54826831817627,
      "learning_rate": 6.237748596100246e-07,
      "loss": 2.9979,
      "step": 75735
    },
    {
      "epoch": 0.7704671223958334,
      "grad_norm": 13.252427101135254,
      "learning_rate": 6.235106095984369e-07,
      "loss": 3.3592,
      "step": 75740
    },
    {
      "epoch": 0.7705179850260416,
      "grad_norm": 9.107203483581543,
      "learning_rate": 6.232464075963851e-07,
      "loss": 3.2647,
      "step": 75745
    },
    {
      "epoch": 0.77056884765625,
      "grad_norm": 11.003279685974121,
      "learning_rate": 6.229822536106298e-07,
      "loss": 3.1104,
      "step": 75750
    },
    {
      "epoch": 0.7706197102864584,
      "grad_norm": 11.99866008758545,
      "learning_rate": 6.227181476479278e-07,
      "loss": 3.5451,
      "step": 75755
    },
    {
      "epoch": 0.7706705729166666,
      "grad_norm": 15.338647842407227,
      "learning_rate": 6.224540897150374e-07,
      "loss": 3.2154,
      "step": 75760
    },
    {
      "epoch": 0.770721435546875,
      "grad_norm": 10.090010643005371,
      "learning_rate": 6.221900798187147e-07,
      "loss": 3.501,
      "step": 75765
    },
    {
      "epoch": 0.7707722981770834,
      "grad_norm": 8.421314239501953,
      "learning_rate": 6.219261179657126e-07,
      "loss": 2.9642,
      "step": 75770
    },
    {
      "epoch": 0.7708231608072916,
      "grad_norm": 13.950006484985352,
      "learning_rate": 6.216622041627857e-07,
      "loss": 3.2954,
      "step": 75775
    },
    {
      "epoch": 0.7708740234375,
      "grad_norm": 10.523149490356445,
      "learning_rate": 6.213983384166868e-07,
      "loss": 3.0715,
      "step": 75780
    },
    {
      "epoch": 0.7709248860677084,
      "grad_norm": 11.545313835144043,
      "learning_rate": 6.211345207341663e-07,
      "loss": 3.3117,
      "step": 75785
    },
    {
      "epoch": 0.7709757486979166,
      "grad_norm": 12.871099472045898,
      "learning_rate": 6.208707511219736e-07,
      "loss": 3.4587,
      "step": 75790
    },
    {
      "epoch": 0.771026611328125,
      "grad_norm": 13.466527938842773,
      "learning_rate": 6.20607029586856e-07,
      "loss": 3.2627,
      "step": 75795
    },
    {
      "epoch": 0.7710774739583334,
      "grad_norm": 10.577140808105469,
      "learning_rate": 6.203433561355634e-07,
      "loss": 3.4507,
      "step": 75800
    },
    {
      "epoch": 0.7711283365885416,
      "grad_norm": 12.235979080200195,
      "learning_rate": 6.200797307748405e-07,
      "loss": 3.419,
      "step": 75805
    },
    {
      "epoch": 0.77117919921875,
      "grad_norm": 8.995306015014648,
      "learning_rate": 6.198161535114322e-07,
      "loss": 3.3275,
      "step": 75810
    },
    {
      "epoch": 0.7712300618489584,
      "grad_norm": 17.704317092895508,
      "learning_rate": 6.195526243520814e-07,
      "loss": 3.3854,
      "step": 75815
    },
    {
      "epoch": 0.7712809244791666,
      "grad_norm": 8.587143898010254,
      "learning_rate": 6.192891433035306e-07,
      "loss": 3.3279,
      "step": 75820
    },
    {
      "epoch": 0.771331787109375,
      "grad_norm": 15.102418899536133,
      "learning_rate": 6.190257103725222e-07,
      "loss": 3.452,
      "step": 75825
    },
    {
      "epoch": 0.7713826497395834,
      "grad_norm": 7.975759506225586,
      "learning_rate": 6.187623255657945e-07,
      "loss": 3.6413,
      "step": 75830
    },
    {
      "epoch": 0.7714335123697916,
      "grad_norm": 10.76710033416748,
      "learning_rate": 6.184989888900874e-07,
      "loss": 3.334,
      "step": 75835
    },
    {
      "epoch": 0.771484375,
      "grad_norm": 9.975716590881348,
      "learning_rate": 6.182357003521377e-07,
      "loss": 3.4803,
      "step": 75840
    },
    {
      "epoch": 0.7715352376302084,
      "grad_norm": 13.248037338256836,
      "learning_rate": 6.179724599586809e-07,
      "loss": 4.0471,
      "step": 75845
    },
    {
      "epoch": 0.7715861002604166,
      "grad_norm": 14.565338134765625,
      "learning_rate": 6.177092677164528e-07,
      "loss": 3.2401,
      "step": 75850
    },
    {
      "epoch": 0.771636962890625,
      "grad_norm": 8.690911293029785,
      "learning_rate": 6.174461236321874e-07,
      "loss": 3.2437,
      "step": 75855
    },
    {
      "epoch": 0.7716878255208334,
      "grad_norm": 10.3700532913208,
      "learning_rate": 6.17183027712617e-07,
      "loss": 3.4599,
      "step": 75860
    },
    {
      "epoch": 0.7717386881510416,
      "grad_norm": 8.442960739135742,
      "learning_rate": 6.169199799644723e-07,
      "loss": 3.5185,
      "step": 75865
    },
    {
      "epoch": 0.77178955078125,
      "grad_norm": 13.842940330505371,
      "learning_rate": 6.166569803944827e-07,
      "loss": 3.309,
      "step": 75870
    },
    {
      "epoch": 0.7718404134114584,
      "grad_norm": 12.644615173339844,
      "learning_rate": 6.163940290093781e-07,
      "loss": 3.4853,
      "step": 75875
    },
    {
      "epoch": 0.7718912760416666,
      "grad_norm": 10.915739059448242,
      "learning_rate": 6.161311258158866e-07,
      "loss": 3.8723,
      "step": 75880
    },
    {
      "epoch": 0.771942138671875,
      "grad_norm": 11.840435981750488,
      "learning_rate": 6.158682708207336e-07,
      "loss": 3.248,
      "step": 75885
    },
    {
      "epoch": 0.7719930013020834,
      "grad_norm": 14.093424797058105,
      "learning_rate": 6.156054640306435e-07,
      "loss": 3.3579,
      "step": 75890
    },
    {
      "epoch": 0.7720438639322916,
      "grad_norm": 11.389480590820312,
      "learning_rate": 6.15342705452342e-07,
      "loss": 3.1137,
      "step": 75895
    },
    {
      "epoch": 0.7720947265625,
      "grad_norm": 10.092771530151367,
      "learning_rate": 6.150799950925496e-07,
      "loss": 3.3353,
      "step": 75900
    },
    {
      "epoch": 0.7721455891927084,
      "grad_norm": 11.442878723144531,
      "learning_rate": 6.148173329579896e-07,
      "loss": 3.0741,
      "step": 75905
    },
    {
      "epoch": 0.7721964518229166,
      "grad_norm": 9.948512077331543,
      "learning_rate": 6.145547190553808e-07,
      "loss": 3.5393,
      "step": 75910
    },
    {
      "epoch": 0.772247314453125,
      "grad_norm": 9.788599967956543,
      "learning_rate": 6.142921533914432e-07,
      "loss": 3.1825,
      "step": 75915
    },
    {
      "epoch": 0.7722981770833334,
      "grad_norm": 126.83859252929688,
      "learning_rate": 6.140296359728937e-07,
      "loss": 3.2213,
      "step": 75920
    },
    {
      "epoch": 0.7723490397135416,
      "grad_norm": 12.804304122924805,
      "learning_rate": 6.137671668064482e-07,
      "loss": 3.1213,
      "step": 75925
    },
    {
      "epoch": 0.77239990234375,
      "grad_norm": 13.880815505981445,
      "learning_rate": 6.135047458988228e-07,
      "loss": 3.341,
      "step": 75930
    },
    {
      "epoch": 0.7724507649739584,
      "grad_norm": 9.59664249420166,
      "learning_rate": 6.13242373256732e-07,
      "loss": 3.3479,
      "step": 75935
    },
    {
      "epoch": 0.7725016276041666,
      "grad_norm": 13.731152534484863,
      "learning_rate": 6.129800488868878e-07,
      "loss": 3.5091,
      "step": 75940
    },
    {
      "epoch": 0.772552490234375,
      "grad_norm": 13.55499267578125,
      "learning_rate": 6.127177727960013e-07,
      "loss": 3.1433,
      "step": 75945
    },
    {
      "epoch": 0.7726033528645834,
      "grad_norm": 8.930282592773438,
      "learning_rate": 6.124555449907832e-07,
      "loss": 3.3842,
      "step": 75950
    },
    {
      "epoch": 0.7726542154947916,
      "grad_norm": 10.433351516723633,
      "learning_rate": 6.12193365477943e-07,
      "loss": 3.6894,
      "step": 75955
    },
    {
      "epoch": 0.772705078125,
      "grad_norm": 14.040569305419922,
      "learning_rate": 6.119312342641883e-07,
      "loss": 3.3541,
      "step": 75960
    },
    {
      "epoch": 0.7727559407552084,
      "grad_norm": 11.76894760131836,
      "learning_rate": 6.116691513562248e-07,
      "loss": 3.1397,
      "step": 75965
    },
    {
      "epoch": 0.7728068033854166,
      "grad_norm": 12.630956649780273,
      "learning_rate": 6.114071167607591e-07,
      "loss": 3.4149,
      "step": 75970
    },
    {
      "epoch": 0.772857666015625,
      "grad_norm": 13.352283477783203,
      "learning_rate": 6.111451304844939e-07,
      "loss": 3.1505,
      "step": 75975
    },
    {
      "epoch": 0.7729085286458334,
      "grad_norm": 10.95576286315918,
      "learning_rate": 6.108831925341339e-07,
      "loss": 3.2773,
      "step": 75980
    },
    {
      "epoch": 0.7729593912760416,
      "grad_norm": 12.064718246459961,
      "learning_rate": 6.106213029163785e-07,
      "loss": 3.3229,
      "step": 75985
    },
    {
      "epoch": 0.77301025390625,
      "grad_norm": 15.52635669708252,
      "learning_rate": 6.103594616379299e-07,
      "loss": 3.1987,
      "step": 75990
    },
    {
      "epoch": 0.7730611165364584,
      "grad_norm": 14.002089500427246,
      "learning_rate": 6.100976687054869e-07,
      "loss": 3.2175,
      "step": 75995
    },
    {
      "epoch": 0.7731119791666666,
      "grad_norm": 13.10692024230957,
      "learning_rate": 6.098359241257462e-07,
      "loss": 3.4373,
      "step": 76000
    },
    {
      "epoch": 0.773162841796875,
      "grad_norm": 7.944686412811279,
      "learning_rate": 6.095742279054053e-07,
      "loss": 3.4441,
      "step": 76005
    },
    {
      "epoch": 0.7732137044270834,
      "grad_norm": 18.644197463989258,
      "learning_rate": 6.093125800511607e-07,
      "loss": 3.3481,
      "step": 76010
    },
    {
      "epoch": 0.7732645670572916,
      "grad_norm": 12.620013236999512,
      "learning_rate": 6.090509805697053e-07,
      "loss": 3.4923,
      "step": 76015
    },
    {
      "epoch": 0.7733154296875,
      "grad_norm": 14.259735107421875,
      "learning_rate": 6.087894294677326e-07,
      "loss": 3.0934,
      "step": 76020
    },
    {
      "epoch": 0.7733662923177084,
      "grad_norm": 17.620956420898438,
      "learning_rate": 6.085279267519334e-07,
      "loss": 3.453,
      "step": 76025
    },
    {
      "epoch": 0.7734171549479166,
      "grad_norm": 14.865498542785645,
      "learning_rate": 6.082664724289986e-07,
      "loss": 3.2248,
      "step": 76030
    },
    {
      "epoch": 0.773468017578125,
      "grad_norm": 10.771533966064453,
      "learning_rate": 6.080050665056186e-07,
      "loss": 3.1181,
      "step": 76035
    },
    {
      "epoch": 0.7735188802083334,
      "grad_norm": 9.232945442199707,
      "learning_rate": 6.077437089884797e-07,
      "loss": 3.438,
      "step": 76040
    },
    {
      "epoch": 0.7735697428385416,
      "grad_norm": 11.017297744750977,
      "learning_rate": 6.074823998842702e-07,
      "loss": 3.1982,
      "step": 76045
    },
    {
      "epoch": 0.77362060546875,
      "grad_norm": 15.718427658081055,
      "learning_rate": 6.072211391996752e-07,
      "loss": 3.5306,
      "step": 76050
    },
    {
      "epoch": 0.7736714680989584,
      "grad_norm": 97.13336181640625,
      "learning_rate": 6.069599269413776e-07,
      "loss": 3.251,
      "step": 76055
    },
    {
      "epoch": 0.7737223307291666,
      "grad_norm": 13.607686042785645,
      "learning_rate": 6.066987631160617e-07,
      "loss": 3.0443,
      "step": 76060
    },
    {
      "epoch": 0.773773193359375,
      "grad_norm": 15.883814811706543,
      "learning_rate": 6.064376477304102e-07,
      "loss": 3.4098,
      "step": 76065
    },
    {
      "epoch": 0.7738240559895834,
      "grad_norm": 8.217681884765625,
      "learning_rate": 6.061765807911022e-07,
      "loss": 3.145,
      "step": 76070
    },
    {
      "epoch": 0.7738749186197916,
      "grad_norm": 12.272388458251953,
      "learning_rate": 6.059155623048177e-07,
      "loss": 3.1563,
      "step": 76075
    },
    {
      "epoch": 0.77392578125,
      "grad_norm": 9.546619415283203,
      "learning_rate": 6.056545922782339e-07,
      "loss": 2.8569,
      "step": 76080
    },
    {
      "epoch": 0.7739766438802084,
      "grad_norm": 7.779810428619385,
      "learning_rate": 6.053936707180283e-07,
      "loss": 3.3145,
      "step": 76085
    },
    {
      "epoch": 0.7740275065104166,
      "grad_norm": 18.373014450073242,
      "learning_rate": 6.051327976308774e-07,
      "loss": 3.1823,
      "step": 76090
    },
    {
      "epoch": 0.774078369140625,
      "grad_norm": 18.18805694580078,
      "learning_rate": 6.048719730234548e-07,
      "loss": 3.2019,
      "step": 76095
    },
    {
      "epoch": 0.7741292317708334,
      "grad_norm": 10.839600563049316,
      "learning_rate": 6.046111969024327e-07,
      "loss": 3.4768,
      "step": 76100
    },
    {
      "epoch": 0.7741800944010416,
      "grad_norm": 12.216818809509277,
      "learning_rate": 6.043504692744837e-07,
      "loss": 3.3204,
      "step": 76105
    },
    {
      "epoch": 0.77423095703125,
      "grad_norm": 13.3114595413208,
      "learning_rate": 6.040897901462797e-07,
      "loss": 3.1056,
      "step": 76110
    },
    {
      "epoch": 0.7742818196614584,
      "grad_norm": 12.069828033447266,
      "learning_rate": 6.038291595244889e-07,
      "loss": 3.0938,
      "step": 76115
    },
    {
      "epoch": 0.7743326822916666,
      "grad_norm": 8.250632286071777,
      "learning_rate": 6.035685774157787e-07,
      "loss": 3.424,
      "step": 76120
    },
    {
      "epoch": 0.774383544921875,
      "grad_norm": 10.238628387451172,
      "learning_rate": 6.033080438268179e-07,
      "loss": 3.5923,
      "step": 76125
    },
    {
      "epoch": 0.7744344075520834,
      "grad_norm": 13.416050910949707,
      "learning_rate": 6.030475587642703e-07,
      "loss": 3.323,
      "step": 76130
    },
    {
      "epoch": 0.7744852701822916,
      "grad_norm": 13.140199661254883,
      "learning_rate": 6.027871222348014e-07,
      "loss": 3.6046,
      "step": 76135
    },
    {
      "epoch": 0.7745361328125,
      "grad_norm": 10.973596572875977,
      "learning_rate": 6.025267342450753e-07,
      "loss": 3.5131,
      "step": 76140
    },
    {
      "epoch": 0.7745869954427084,
      "grad_norm": 13.73587703704834,
      "learning_rate": 6.022663948017524e-07,
      "loss": 2.9926,
      "step": 76145
    },
    {
      "epoch": 0.7746378580729166,
      "grad_norm": 10.315624237060547,
      "learning_rate": 6.020061039114944e-07,
      "loss": 3.1911,
      "step": 76150
    },
    {
      "epoch": 0.774688720703125,
      "grad_norm": 14.70566177368164,
      "learning_rate": 6.017458615809593e-07,
      "loss": 3.2069,
      "step": 76155
    },
    {
      "epoch": 0.7747395833333334,
      "grad_norm": 12.014630317687988,
      "learning_rate": 6.014856678168069e-07,
      "loss": 3.2442,
      "step": 76160
    },
    {
      "epoch": 0.7747904459635416,
      "grad_norm": 13.165812492370605,
      "learning_rate": 6.012255226256938e-07,
      "loss": 3.3926,
      "step": 76165
    },
    {
      "epoch": 0.77484130859375,
      "grad_norm": 7.364107131958008,
      "learning_rate": 6.009654260142761e-07,
      "loss": 3.0502,
      "step": 76170
    },
    {
      "epoch": 0.7748921712239584,
      "grad_norm": 11.134225845336914,
      "learning_rate": 6.007053779892069e-07,
      "loss": 3.406,
      "step": 76175
    },
    {
      "epoch": 0.7749430338541666,
      "grad_norm": 8.265276908874512,
      "learning_rate": 6.004453785571413e-07,
      "loss": 3.2829,
      "step": 76180
    },
    {
      "epoch": 0.774993896484375,
      "grad_norm": 11.279505729675293,
      "learning_rate": 6.001854277247296e-07,
      "loss": 3.7411,
      "step": 76185
    },
    {
      "epoch": 0.7750447591145834,
      "grad_norm": 13.352738380432129,
      "learning_rate": 5.999255254986242e-07,
      "loss": 3.3503,
      "step": 76190
    },
    {
      "epoch": 0.7750956217447916,
      "grad_norm": 8.39072322845459,
      "learning_rate": 5.996656718854733e-07,
      "loss": 3.1168,
      "step": 76195
    },
    {
      "epoch": 0.775146484375,
      "grad_norm": 13.08597469329834,
      "learning_rate": 5.994058668919261e-07,
      "loss": 3.3923,
      "step": 76200
    },
    {
      "epoch": 0.7751973470052084,
      "grad_norm": 13.441854476928711,
      "learning_rate": 5.991461105246296e-07,
      "loss": 3.1712,
      "step": 76205
    },
    {
      "epoch": 0.7752482096354166,
      "grad_norm": 10.724113464355469,
      "learning_rate": 5.988864027902286e-07,
      "loss": 3.559,
      "step": 76210
    },
    {
      "epoch": 0.775299072265625,
      "grad_norm": 13.422377586364746,
      "learning_rate": 5.986267436953683e-07,
      "loss": 2.9751,
      "step": 76215
    },
    {
      "epoch": 0.7753499348958334,
      "grad_norm": 10.692633628845215,
      "learning_rate": 5.983671332466926e-07,
      "loss": 3.2711,
      "step": 76220
    },
    {
      "epoch": 0.7754007975260416,
      "grad_norm": 12.448526382446289,
      "learning_rate": 5.981075714508433e-07,
      "loss": 3.1235,
      "step": 76225
    },
    {
      "epoch": 0.77545166015625,
      "grad_norm": 11.251077651977539,
      "learning_rate": 5.978480583144602e-07,
      "loss": 4.0709,
      "step": 76230
    },
    {
      "epoch": 0.7755025227864584,
      "grad_norm": 10.245332717895508,
      "learning_rate": 5.975885938441844e-07,
      "loss": 3.1104,
      "step": 76235
    },
    {
      "epoch": 0.7755533854166666,
      "grad_norm": 8.04434871673584,
      "learning_rate": 5.973291780466528e-07,
      "loss": 3.4971,
      "step": 76240
    },
    {
      "epoch": 0.775604248046875,
      "grad_norm": 14.250527381896973,
      "learning_rate": 5.970698109285039e-07,
      "loss": 3.1624,
      "step": 76245
    },
    {
      "epoch": 0.7756551106770834,
      "grad_norm": 9.206746101379395,
      "learning_rate": 5.968104924963722e-07,
      "loss": 3.4041,
      "step": 76250
    },
    {
      "epoch": 0.7757059733072916,
      "grad_norm": 8.508057594299316,
      "learning_rate": 5.965512227568937e-07,
      "loss": 3.6937,
      "step": 76255
    },
    {
      "epoch": 0.7757568359375,
      "grad_norm": 10.889847755432129,
      "learning_rate": 5.962920017167001e-07,
      "loss": 3.5504,
      "step": 76260
    },
    {
      "epoch": 0.7758076985677084,
      "grad_norm": 12.170618057250977,
      "learning_rate": 5.960328293824253e-07,
      "loss": 3.1038,
      "step": 76265
    },
    {
      "epoch": 0.7758585611979166,
      "grad_norm": 16.20078468322754,
      "learning_rate": 5.957737057606983e-07,
      "loss": 3.0198,
      "step": 76270
    },
    {
      "epoch": 0.775909423828125,
      "grad_norm": 14.83310604095459,
      "learning_rate": 5.955146308581505e-07,
      "loss": 3.0933,
      "step": 76275
    },
    {
      "epoch": 0.7759602864583334,
      "grad_norm": 10.469974517822266,
      "learning_rate": 5.952556046814096e-07,
      "loss": 3.3732,
      "step": 76280
    },
    {
      "epoch": 0.7760111490885416,
      "grad_norm": 9.828927993774414,
      "learning_rate": 5.949966272371018e-07,
      "loss": 2.8567,
      "step": 76285
    },
    {
      "epoch": 0.77606201171875,
      "grad_norm": 7.971617221832275,
      "learning_rate": 5.947376985318537e-07,
      "loss": 3.6611,
      "step": 76290
    },
    {
      "epoch": 0.7761128743489584,
      "grad_norm": 11.64389419555664,
      "learning_rate": 5.944788185722907e-07,
      "loss": 3.2097,
      "step": 76295
    },
    {
      "epoch": 0.7761637369791666,
      "grad_norm": 12.310096740722656,
      "learning_rate": 5.942199873650356e-07,
      "loss": 3.4852,
      "step": 76300
    },
    {
      "epoch": 0.776214599609375,
      "grad_norm": 14.617456436157227,
      "learning_rate": 5.939612049167104e-07,
      "loss": 3.422,
      "step": 76305
    },
    {
      "epoch": 0.7762654622395834,
      "grad_norm": 12.969735145568848,
      "learning_rate": 5.937024712339351e-07,
      "loss": 3.469,
      "step": 76310
    },
    {
      "epoch": 0.7763163248697916,
      "grad_norm": 13.735860824584961,
      "learning_rate": 5.934437863233303e-07,
      "loss": 3.2358,
      "step": 76315
    },
    {
      "epoch": 0.7763671875,
      "grad_norm": 8.250831604003906,
      "learning_rate": 5.931851501915151e-07,
      "loss": 2.8565,
      "step": 76320
    },
    {
      "epoch": 0.7764180501302084,
      "grad_norm": 14.287775993347168,
      "learning_rate": 5.929265628451051e-07,
      "loss": 3.3142,
      "step": 76325
    },
    {
      "epoch": 0.7764689127604166,
      "grad_norm": 13.908318519592285,
      "learning_rate": 5.926680242907176e-07,
      "loss": 3.3686,
      "step": 76330
    },
    {
      "epoch": 0.776519775390625,
      "grad_norm": 18.167110443115234,
      "learning_rate": 5.924095345349667e-07,
      "loss": 3.3309,
      "step": 76335
    },
    {
      "epoch": 0.7765706380208334,
      "grad_norm": 7.273578643798828,
      "learning_rate": 5.921510935844652e-07,
      "loss": 4.0699,
      "step": 76340
    },
    {
      "epoch": 0.7766215006510416,
      "grad_norm": 11.433211326599121,
      "learning_rate": 5.918927014458253e-07,
      "loss": 2.9545,
      "step": 76345
    },
    {
      "epoch": 0.77667236328125,
      "grad_norm": 13.524624824523926,
      "learning_rate": 5.916343581256596e-07,
      "loss": 3.3202,
      "step": 76350
    },
    {
      "epoch": 0.7767232259114584,
      "grad_norm": 13.013619422912598,
      "learning_rate": 5.91376063630576e-07,
      "loss": 2.9796,
      "step": 76355
    },
    {
      "epoch": 0.7767740885416666,
      "grad_norm": 10.851083755493164,
      "learning_rate": 5.911178179671836e-07,
      "loss": 3.8777,
      "step": 76360
    },
    {
      "epoch": 0.776824951171875,
      "grad_norm": 9.591619491577148,
      "learning_rate": 5.908596211420889e-07,
      "loss": 3.6392,
      "step": 76365
    },
    {
      "epoch": 0.7768758138020834,
      "grad_norm": 11.787693977355957,
      "learning_rate": 5.906014731618981e-07,
      "loss": 2.8807,
      "step": 76370
    },
    {
      "epoch": 0.7769266764322916,
      "grad_norm": 16.933032989501953,
      "learning_rate": 5.903433740332168e-07,
      "loss": 3.5137,
      "step": 76375
    },
    {
      "epoch": 0.7769775390625,
      "grad_norm": 12.437687873840332,
      "learning_rate": 5.900853237626475e-07,
      "loss": 3.4988,
      "step": 76380
    },
    {
      "epoch": 0.7770284016927084,
      "grad_norm": 15.326409339904785,
      "learning_rate": 5.898273223567918e-07,
      "loss": 3.1273,
      "step": 76385
    },
    {
      "epoch": 0.7770792643229166,
      "grad_norm": 14.829540252685547,
      "learning_rate": 5.895693698222521e-07,
      "loss": 3.4237,
      "step": 76390
    },
    {
      "epoch": 0.777130126953125,
      "grad_norm": 11.511395454406738,
      "learning_rate": 5.893114661656266e-07,
      "loss": 3.0452,
      "step": 76395
    },
    {
      "epoch": 0.7771809895833334,
      "grad_norm": 12.296662330627441,
      "learning_rate": 5.89053611393515e-07,
      "loss": 3.1364,
      "step": 76400
    },
    {
      "epoch": 0.7772318522135416,
      "grad_norm": 11.894816398620605,
      "learning_rate": 5.887958055125132e-07,
      "loss": 3.1094,
      "step": 76405
    },
    {
      "epoch": 0.77728271484375,
      "grad_norm": 13.634196281433105,
      "learning_rate": 5.88538048529218e-07,
      "loss": 3.3801,
      "step": 76410
    },
    {
      "epoch": 0.7773335774739584,
      "grad_norm": 17.220705032348633,
      "learning_rate": 5.882803404502235e-07,
      "loss": 3.2957,
      "step": 76415
    },
    {
      "epoch": 0.7773844401041666,
      "grad_norm": 8.81678581237793,
      "learning_rate": 5.880226812821232e-07,
      "loss": 3.3365,
      "step": 76420
    },
    {
      "epoch": 0.777435302734375,
      "grad_norm": 9.809067726135254,
      "learning_rate": 5.877650710315103e-07,
      "loss": 3.4136,
      "step": 76425
    },
    {
      "epoch": 0.7774861653645834,
      "grad_norm": 10.364507675170898,
      "learning_rate": 5.875075097049745e-07,
      "loss": 3.304,
      "step": 76430
    },
    {
      "epoch": 0.7775370279947916,
      "grad_norm": 14.615416526794434,
      "learning_rate": 5.872499973091059e-07,
      "loss": 3.645,
      "step": 76435
    },
    {
      "epoch": 0.777587890625,
      "grad_norm": 14.780035018920898,
      "learning_rate": 5.86992533850492e-07,
      "loss": 3.2824,
      "step": 76440
    },
    {
      "epoch": 0.7776387532552084,
      "grad_norm": 14.5195894241333,
      "learning_rate": 5.867351193357207e-07,
      "loss": 3.2958,
      "step": 76445
    },
    {
      "epoch": 0.7776896158854166,
      "grad_norm": 11.425013542175293,
      "learning_rate": 5.864777537713787e-07,
      "loss": 3.3065,
      "step": 76450
    },
    {
      "epoch": 0.777740478515625,
      "grad_norm": 11.649991989135742,
      "learning_rate": 5.862204371640498e-07,
      "loss": 3.3888,
      "step": 76455
    },
    {
      "epoch": 0.7777913411458334,
      "grad_norm": 9.892515182495117,
      "learning_rate": 5.859631695203167e-07,
      "loss": 2.9988,
      "step": 76460
    },
    {
      "epoch": 0.7778422037760416,
      "grad_norm": 7.605737686157227,
      "learning_rate": 5.857059508467628e-07,
      "loss": 3.2106,
      "step": 76465
    },
    {
      "epoch": 0.77789306640625,
      "grad_norm": 14.123431205749512,
      "learning_rate": 5.854487811499676e-07,
      "loss": 3.3977,
      "step": 76470
    },
    {
      "epoch": 0.7779439290364584,
      "grad_norm": 15.220905303955078,
      "learning_rate": 5.851916604365123e-07,
      "loss": 2.9927,
      "step": 76475
    },
    {
      "epoch": 0.7779947916666666,
      "grad_norm": 11.82323169708252,
      "learning_rate": 5.84934588712974e-07,
      "loss": 2.9505,
      "step": 76480
    },
    {
      "epoch": 0.778045654296875,
      "grad_norm": 13.638838768005371,
      "learning_rate": 5.84677565985931e-07,
      "loss": 3.25,
      "step": 76485
    },
    {
      "epoch": 0.7780965169270834,
      "grad_norm": 11.528968811035156,
      "learning_rate": 5.844205922619584e-07,
      "loss": 2.7836,
      "step": 76490
    },
    {
      "epoch": 0.7781473795572916,
      "grad_norm": 8.256325721740723,
      "learning_rate": 5.841636675476303e-07,
      "loss": 3.6025,
      "step": 76495
    },
    {
      "epoch": 0.7781982421875,
      "grad_norm": 16.759963989257812,
      "learning_rate": 5.839067918495206e-07,
      "loss": 3.2637,
      "step": 76500
    },
    {
      "epoch": 0.7782491048177084,
      "grad_norm": 11.771158218383789,
      "learning_rate": 5.836499651742023e-07,
      "loss": 3.3356,
      "step": 76505
    },
    {
      "epoch": 0.7782999674479166,
      "grad_norm": 12.421799659729004,
      "learning_rate": 5.833931875282451e-07,
      "loss": 3.6195,
      "step": 76510
    },
    {
      "epoch": 0.778350830078125,
      "grad_norm": 13.912510871887207,
      "learning_rate": 5.831364589182184e-07,
      "loss": 3.1761,
      "step": 76515
    },
    {
      "epoch": 0.7784016927083334,
      "grad_norm": 14.278749465942383,
      "learning_rate": 5.828797793506918e-07,
      "loss": 3.1841,
      "step": 76520
    },
    {
      "epoch": 0.7784525553385416,
      "grad_norm": 10.667387962341309,
      "learning_rate": 5.826231488322309e-07,
      "loss": 3.1582,
      "step": 76525
    },
    {
      "epoch": 0.77850341796875,
      "grad_norm": 15.149872779846191,
      "learning_rate": 5.823665673694029e-07,
      "loss": 3.5367,
      "step": 76530
    },
    {
      "epoch": 0.7785542805989584,
      "grad_norm": 8.389341354370117,
      "learning_rate": 5.821100349687711e-07,
      "loss": 3.115,
      "step": 76535
    },
    {
      "epoch": 0.7786051432291666,
      "grad_norm": 10.314140319824219,
      "learning_rate": 5.818535516369003e-07,
      "loss": 3.3898,
      "step": 76540
    },
    {
      "epoch": 0.778656005859375,
      "grad_norm": 13.677864074707031,
      "learning_rate": 5.815971173803515e-07,
      "loss": 2.963,
      "step": 76545
    },
    {
      "epoch": 0.7787068684895834,
      "grad_norm": 8.13903522491455,
      "learning_rate": 5.813407322056852e-07,
      "loss": 3.3911,
      "step": 76550
    },
    {
      "epoch": 0.7787577311197916,
      "grad_norm": 10.670424461364746,
      "learning_rate": 5.810843961194615e-07,
      "loss": 3.0771,
      "step": 76555
    },
    {
      "epoch": 0.77880859375,
      "grad_norm": 11.17371654510498,
      "learning_rate": 5.808281091282392e-07,
      "loss": 3.5922,
      "step": 76560
    },
    {
      "epoch": 0.7788594563802084,
      "grad_norm": 14.148518562316895,
      "learning_rate": 5.805718712385749e-07,
      "loss": 2.9514,
      "step": 76565
    },
    {
      "epoch": 0.7789103190104166,
      "grad_norm": 13.83517074584961,
      "learning_rate": 5.803156824570238e-07,
      "loss": 3.2211,
      "step": 76570
    },
    {
      "epoch": 0.778961181640625,
      "grad_norm": 8.224130630493164,
      "learning_rate": 5.800595427901407e-07,
      "loss": 3.1588,
      "step": 76575
    },
    {
      "epoch": 0.7790120442708334,
      "grad_norm": 17.40386962890625,
      "learning_rate": 5.798034522444801e-07,
      "loss": 3.3977,
      "step": 76580
    },
    {
      "epoch": 0.7790629069010416,
      "grad_norm": 11.589737892150879,
      "learning_rate": 5.795474108265927e-07,
      "loss": 3.9817,
      "step": 76585
    },
    {
      "epoch": 0.77911376953125,
      "grad_norm": 11.125046730041504,
      "learning_rate": 5.792914185430299e-07,
      "loss": 3.5473,
      "step": 76590
    },
    {
      "epoch": 0.7791646321614584,
      "grad_norm": 12.34698486328125,
      "learning_rate": 5.7903547540034e-07,
      "loss": 3.3408,
      "step": 76595
    },
    {
      "epoch": 0.7792154947916666,
      "grad_norm": 13.01732063293457,
      "learning_rate": 5.787795814050722e-07,
      "loss": 3.4159,
      "step": 76600
    },
    {
      "epoch": 0.779266357421875,
      "grad_norm": 12.777318954467773,
      "learning_rate": 5.785237365637742e-07,
      "loss": 3.358,
      "step": 76605
    },
    {
      "epoch": 0.7793172200520834,
      "grad_norm": 13.600346565246582,
      "learning_rate": 5.782679408829911e-07,
      "loss": 3.0258,
      "step": 76610
    },
    {
      "epoch": 0.7793680826822916,
      "grad_norm": 12.758298873901367,
      "learning_rate": 5.780121943692662e-07,
      "loss": 3.2326,
      "step": 76615
    },
    {
      "epoch": 0.7794189453125,
      "grad_norm": 11.570501327514648,
      "learning_rate": 5.777564970291447e-07,
      "loss": 3.0778,
      "step": 76620
    },
    {
      "epoch": 0.7794698079427084,
      "grad_norm": 15.946113586425781,
      "learning_rate": 5.775008488691666e-07,
      "loss": 3.1848,
      "step": 76625
    },
    {
      "epoch": 0.7795206705729166,
      "grad_norm": 11.616817474365234,
      "learning_rate": 5.772452498958739e-07,
      "loss": 3.3857,
      "step": 76630
    },
    {
      "epoch": 0.779571533203125,
      "grad_norm": 10.456148147583008,
      "learning_rate": 5.769897001158065e-07,
      "loss": 3.1544,
      "step": 76635
    },
    {
      "epoch": 0.7796223958333334,
      "grad_norm": 9.678732872009277,
      "learning_rate": 5.767341995355016e-07,
      "loss": 3.4938,
      "step": 76640
    },
    {
      "epoch": 0.7796732584635416,
      "grad_norm": 9.791245460510254,
      "learning_rate": 5.764787481614964e-07,
      "loss": 3.1334,
      "step": 76645
    },
    {
      "epoch": 0.77972412109375,
      "grad_norm": 12.438119888305664,
      "learning_rate": 5.762233460003258e-07,
      "loss": 3.3304,
      "step": 76650
    },
    {
      "epoch": 0.7797749837239584,
      "grad_norm": 8.740230560302734,
      "learning_rate": 5.759679930585249e-07,
      "loss": 3.2382,
      "step": 76655
    },
    {
      "epoch": 0.7798258463541666,
      "grad_norm": 10.377050399780273,
      "learning_rate": 5.757126893426274e-07,
      "loss": 3.1279,
      "step": 76660
    },
    {
      "epoch": 0.779876708984375,
      "grad_norm": 10.924396514892578,
      "learning_rate": 5.754574348591649e-07,
      "loss": 3.5955,
      "step": 76665
    },
    {
      "epoch": 0.7799275716145834,
      "grad_norm": 12.242827415466309,
      "learning_rate": 5.75202229614667e-07,
      "loss": 3.7679,
      "step": 76670
    },
    {
      "epoch": 0.7799784342447916,
      "grad_norm": 11.291284561157227,
      "learning_rate": 5.749470736156646e-07,
      "loss": 3.1324,
      "step": 76675
    },
    {
      "epoch": 0.780029296875,
      "grad_norm": 13.19884204864502,
      "learning_rate": 5.74691966868684e-07,
      "loss": 3.6457,
      "step": 76680
    },
    {
      "epoch": 0.7800801595052084,
      "grad_norm": 12.717314720153809,
      "learning_rate": 5.744369093802541e-07,
      "loss": 3.0554,
      "step": 76685
    },
    {
      "epoch": 0.7801310221354166,
      "grad_norm": 11.942824363708496,
      "learning_rate": 5.741819011568986e-07,
      "loss": 3.5642,
      "step": 76690
    },
    {
      "epoch": 0.780181884765625,
      "grad_norm": 9.46130657196045,
      "learning_rate": 5.739269422051436e-07,
      "loss": 3.3421,
      "step": 76695
    },
    {
      "epoch": 0.7802327473958334,
      "grad_norm": 17.4813232421875,
      "learning_rate": 5.736720325315109e-07,
      "loss": 3.6149,
      "step": 76700
    },
    {
      "epoch": 0.7802836100260416,
      "grad_norm": 15.691143035888672,
      "learning_rate": 5.734171721425222e-07,
      "loss": 3.1462,
      "step": 76705
    },
    {
      "epoch": 0.78033447265625,
      "grad_norm": 9.280314445495605,
      "learning_rate": 5.731623610446985e-07,
      "loss": 3.3634,
      "step": 76710
    },
    {
      "epoch": 0.7803853352864584,
      "grad_norm": 13.122340202331543,
      "learning_rate": 5.729075992445599e-07,
      "loss": 3.4787,
      "step": 76715
    },
    {
      "epoch": 0.7804361979166666,
      "grad_norm": 9.506455421447754,
      "learning_rate": 5.726528867486233e-07,
      "loss": 3.5734,
      "step": 76720
    },
    {
      "epoch": 0.780487060546875,
      "grad_norm": 15.977293968200684,
      "learning_rate": 5.723982235634052e-07,
      "loss": 3.136,
      "step": 76725
    },
    {
      "epoch": 0.7805379231770834,
      "grad_norm": 12.063434600830078,
      "learning_rate": 5.721436096954217e-07,
      "loss": 3.4242,
      "step": 76730
    },
    {
      "epoch": 0.7805887858072916,
      "grad_norm": 8.83809757232666,
      "learning_rate": 5.718890451511877e-07,
      "loss": 3.318,
      "step": 76735
    },
    {
      "epoch": 0.7806396484375,
      "grad_norm": 10.073627471923828,
      "learning_rate": 5.716345299372153e-07,
      "loss": 2.8855,
      "step": 76740
    },
    {
      "epoch": 0.7806905110677084,
      "grad_norm": 14.36465072631836,
      "learning_rate": 5.713800640600159e-07,
      "loss": 3.5889,
      "step": 76745
    },
    {
      "epoch": 0.7807413736979166,
      "grad_norm": 8.055367469787598,
      "learning_rate": 5.711256475261013e-07,
      "loss": 3.4521,
      "step": 76750
    },
    {
      "epoch": 0.780792236328125,
      "grad_norm": 9.556485176086426,
      "learning_rate": 5.708712803419786e-07,
      "loss": 3.294,
      "step": 76755
    },
    {
      "epoch": 0.7808430989583334,
      "grad_norm": 11.192523956298828,
      "learning_rate": 5.70616962514158e-07,
      "loss": 3.4719,
      "step": 76760
    },
    {
      "epoch": 0.7808939615885416,
      "grad_norm": 13.200043678283691,
      "learning_rate": 5.703626940491442e-07,
      "loss": 3.3841,
      "step": 76765
    },
    {
      "epoch": 0.78094482421875,
      "grad_norm": 16.064577102661133,
      "learning_rate": 5.701084749534444e-07,
      "loss": 3.2305,
      "step": 76770
    },
    {
      "epoch": 0.7809956868489584,
      "grad_norm": 10.967706680297852,
      "learning_rate": 5.698543052335617e-07,
      "loss": 3.1965,
      "step": 76775
    },
    {
      "epoch": 0.7810465494791666,
      "grad_norm": 15.476178169250488,
      "learning_rate": 5.696001848959984e-07,
      "loss": 3.598,
      "step": 76780
    },
    {
      "epoch": 0.781097412109375,
      "grad_norm": 11.553808212280273,
      "learning_rate": 5.693461139472569e-07,
      "loss": 3.3123,
      "step": 76785
    },
    {
      "epoch": 0.7811482747395834,
      "grad_norm": 11.69206428527832,
      "learning_rate": 5.690920923938381e-07,
      "loss": 3.0832,
      "step": 76790
    },
    {
      "epoch": 0.7811991373697916,
      "grad_norm": 12.78307056427002,
      "learning_rate": 5.688381202422405e-07,
      "loss": 3.4091,
      "step": 76795
    },
    {
      "epoch": 0.78125,
      "grad_norm": 13.889488220214844,
      "learning_rate": 5.685841974989617e-07,
      "loss": 2.9902,
      "step": 76800
    },
    {
      "epoch": 0.7813008626302084,
      "grad_norm": 14.213640213012695,
      "learning_rate": 5.683303241704979e-07,
      "loss": 3.1667,
      "step": 76805
    },
    {
      "epoch": 0.7813517252604166,
      "grad_norm": 15.656432151794434,
      "learning_rate": 5.680765002633449e-07,
      "loss": 3.0544,
      "step": 76810
    },
    {
      "epoch": 0.781402587890625,
      "grad_norm": 14.070866584777832,
      "learning_rate": 5.678227257839971e-07,
      "loss": 3.3398,
      "step": 76815
    },
    {
      "epoch": 0.7814534505208334,
      "grad_norm": 12.139039993286133,
      "learning_rate": 5.675690007389464e-07,
      "loss": 3.5697,
      "step": 76820
    },
    {
      "epoch": 0.7815043131510416,
      "grad_norm": 10.680137634277344,
      "learning_rate": 5.673153251346858e-07,
      "loss": 3.1589,
      "step": 76825
    },
    {
      "epoch": 0.78155517578125,
      "grad_norm": 13.58749771118164,
      "learning_rate": 5.670616989777042e-07,
      "loss": 2.7444,
      "step": 76830
    },
    {
      "epoch": 0.7816060384114584,
      "grad_norm": 13.907711029052734,
      "learning_rate": 5.668081222744901e-07,
      "loss": 3.3063,
      "step": 76835
    },
    {
      "epoch": 0.7816569010416666,
      "grad_norm": 12.027887344360352,
      "learning_rate": 5.665545950315321e-07,
      "loss": 3.5625,
      "step": 76840
    },
    {
      "epoch": 0.781707763671875,
      "grad_norm": 14.913036346435547,
      "learning_rate": 5.66301117255317e-07,
      "loss": 2.915,
      "step": 76845
    },
    {
      "epoch": 0.7817586263020834,
      "grad_norm": 13.611757278442383,
      "learning_rate": 5.660476889523298e-07,
      "loss": 3.4866,
      "step": 76850
    },
    {
      "epoch": 0.7818094889322916,
      "grad_norm": 7.673834323883057,
      "learning_rate": 5.657943101290539e-07,
      "loss": 3.2079,
      "step": 76855
    },
    {
      "epoch": 0.7818603515625,
      "grad_norm": 13.895394325256348,
      "learning_rate": 5.655409807919706e-07,
      "loss": 3.1756,
      "step": 76860
    },
    {
      "epoch": 0.7819112141927084,
      "grad_norm": 14.912735939025879,
      "learning_rate": 5.652877009475646e-07,
      "loss": 3.8363,
      "step": 76865
    },
    {
      "epoch": 0.7819620768229166,
      "grad_norm": 9.267807006835938,
      "learning_rate": 5.650344706023139e-07,
      "loss": 3.3846,
      "step": 76870
    },
    {
      "epoch": 0.782012939453125,
      "grad_norm": 13.284967422485352,
      "learning_rate": 5.647812897626976e-07,
      "loss": 3.519,
      "step": 76875
    },
    {
      "epoch": 0.7820638020833334,
      "grad_norm": 16.61823081970215,
      "learning_rate": 5.645281584351927e-07,
      "loss": 3.207,
      "step": 76880
    },
    {
      "epoch": 0.7821146647135416,
      "grad_norm": 7.993548393249512,
      "learning_rate": 5.642750766262761e-07,
      "loss": 3.4191,
      "step": 76885
    },
    {
      "epoch": 0.78216552734375,
      "grad_norm": 13.926109313964844,
      "learning_rate": 5.640220443424235e-07,
      "loss": 3.6736,
      "step": 76890
    },
    {
      "epoch": 0.7822163899739584,
      "grad_norm": 9.467702865600586,
      "learning_rate": 5.637690615901082e-07,
      "loss": 3.6149,
      "step": 76895
    },
    {
      "epoch": 0.7822672526041666,
      "grad_norm": 8.834341049194336,
      "learning_rate": 5.635161283758017e-07,
      "loss": 3.1105,
      "step": 76900
    },
    {
      "epoch": 0.782318115234375,
      "grad_norm": 9.07247543334961,
      "learning_rate": 5.632632447059766e-07,
      "loss": 3.4945,
      "step": 76905
    },
    {
      "epoch": 0.7823689778645834,
      "grad_norm": 10.527985572814941,
      "learning_rate": 5.630104105871017e-07,
      "loss": 3.7146,
      "step": 76910
    },
    {
      "epoch": 0.7824198404947916,
      "grad_norm": 15.648237228393555,
      "learning_rate": 5.627576260256465e-07,
      "loss": 3.4944,
      "step": 76915
    },
    {
      "epoch": 0.782470703125,
      "grad_norm": 12.348536491394043,
      "learning_rate": 5.625048910280789e-07,
      "loss": 3.4807,
      "step": 76920
    },
    {
      "epoch": 0.7825215657552084,
      "grad_norm": 11.248263359069824,
      "learning_rate": 5.622522056008642e-07,
      "loss": 3.3978,
      "step": 76925
    },
    {
      "epoch": 0.7825724283854166,
      "grad_norm": 7.9914960861206055,
      "learning_rate": 5.619995697504679e-07,
      "loss": 3.1907,
      "step": 76930
    },
    {
      "epoch": 0.782623291015625,
      "grad_norm": 15.79336929321289,
      "learning_rate": 5.617469834833522e-07,
      "loss": 3.4765,
      "step": 76935
    },
    {
      "epoch": 0.7826741536458334,
      "grad_norm": 13.746410369873047,
      "learning_rate": 5.614944468059808e-07,
      "loss": 3.3628,
      "step": 76940
    },
    {
      "epoch": 0.7827250162760416,
      "grad_norm": 12.827302932739258,
      "learning_rate": 5.612419597248151e-07,
      "loss": 3.005,
      "step": 76945
    },
    {
      "epoch": 0.78277587890625,
      "grad_norm": 11.03933048248291,
      "learning_rate": 5.609895222463141e-07,
      "loss": 3.5766,
      "step": 76950
    },
    {
      "epoch": 0.7828267415364584,
      "grad_norm": 11.479743003845215,
      "learning_rate": 5.60737134376936e-07,
      "loss": 3.5491,
      "step": 76955
    },
    {
      "epoch": 0.7828776041666666,
      "grad_norm": 11.110942840576172,
      "learning_rate": 5.604847961231396e-07,
      "loss": 3.5511,
      "step": 76960
    },
    {
      "epoch": 0.782928466796875,
      "grad_norm": 7.810293674468994,
      "learning_rate": 5.602325074913792e-07,
      "loss": 2.9781,
      "step": 76965
    },
    {
      "epoch": 0.7829793294270834,
      "grad_norm": 10.472963333129883,
      "learning_rate": 5.599802684881109e-07,
      "loss": 3.6425,
      "step": 76970
    },
    {
      "epoch": 0.7830301920572916,
      "grad_norm": 10.024048805236816,
      "learning_rate": 5.597280791197868e-07,
      "loss": 3.826,
      "step": 76975
    },
    {
      "epoch": 0.7830810546875,
      "grad_norm": 12.897725105285645,
      "learning_rate": 5.594759393928608e-07,
      "loss": 3.1493,
      "step": 76980
    },
    {
      "epoch": 0.7831319173177084,
      "grad_norm": 14.503350257873535,
      "learning_rate": 5.592238493137828e-07,
      "loss": 3.5033,
      "step": 76985
    },
    {
      "epoch": 0.7831827799479166,
      "grad_norm": 11.452110290527344,
      "learning_rate": 5.58971808889002e-07,
      "loss": 3.4164,
      "step": 76990
    },
    {
      "epoch": 0.783233642578125,
      "grad_norm": 12.399145126342773,
      "learning_rate": 5.587198181249673e-07,
      "loss": 3.3973,
      "step": 76995
    },
    {
      "epoch": 0.7832845052083334,
      "grad_norm": 13.125426292419434,
      "learning_rate": 5.584678770281269e-07,
      "loss": 3.4428,
      "step": 77000
    },
    {
      "epoch": 0.7833353678385416,
      "grad_norm": 11.738371849060059,
      "learning_rate": 5.582159856049252e-07,
      "loss": 3.0178,
      "step": 77005
    },
    {
      "epoch": 0.78338623046875,
      "grad_norm": 13.799607276916504,
      "learning_rate": 5.579641438618069e-07,
      "loss": 3.4622,
      "step": 77010
    },
    {
      "epoch": 0.7834370930989584,
      "grad_norm": 14.546651840209961,
      "learning_rate": 5.577123518052154e-07,
      "loss": 3.3319,
      "step": 77015
    },
    {
      "epoch": 0.7834879557291666,
      "grad_norm": 11.298008918762207,
      "learning_rate": 5.574606094415941e-07,
      "loss": 3.4026,
      "step": 77020
    },
    {
      "epoch": 0.783538818359375,
      "grad_norm": 7.891462326049805,
      "learning_rate": 5.572089167773823e-07,
      "loss": 3.1837,
      "step": 77025
    },
    {
      "epoch": 0.7835896809895834,
      "grad_norm": 18.3826847076416,
      "learning_rate": 5.569572738190193e-07,
      "loss": 3.2527,
      "step": 77030
    },
    {
      "epoch": 0.7836405436197916,
      "grad_norm": 14.749083518981934,
      "learning_rate": 5.567056805729448e-07,
      "loss": 3.0726,
      "step": 77035
    },
    {
      "epoch": 0.78369140625,
      "grad_norm": 12.866004943847656,
      "learning_rate": 5.564541370455939e-07,
      "loss": 3.6303,
      "step": 77040
    },
    {
      "epoch": 0.7837422688802084,
      "grad_norm": 12.914320945739746,
      "learning_rate": 5.562026432434039e-07,
      "loss": 3.2979,
      "step": 77045
    },
    {
      "epoch": 0.7837931315104166,
      "grad_norm": 12.6251802444458,
      "learning_rate": 5.559511991728081e-07,
      "loss": 3.2706,
      "step": 77050
    },
    {
      "epoch": 0.783843994140625,
      "grad_norm": 12.312646865844727,
      "learning_rate": 5.556998048402407e-07,
      "loss": 3.1125,
      "step": 77055
    },
    {
      "epoch": 0.7838948567708334,
      "grad_norm": 9.286629676818848,
      "learning_rate": 5.554484602521326e-07,
      "loss": 4.0788,
      "step": 77060
    },
    {
      "epoch": 0.7839457194010416,
      "grad_norm": 10.3337984085083,
      "learning_rate": 5.551971654149144e-07,
      "loss": 3.2471,
      "step": 77065
    },
    {
      "epoch": 0.78399658203125,
      "grad_norm": 11.835550308227539,
      "learning_rate": 5.549459203350157e-07,
      "loss": 3.0746,
      "step": 77070
    },
    {
      "epoch": 0.7840474446614584,
      "grad_norm": 15.912817001342773,
      "learning_rate": 5.546947250188653e-07,
      "loss": 3.9986,
      "step": 77075
    },
    {
      "epoch": 0.7840983072916666,
      "grad_norm": 13.086148262023926,
      "learning_rate": 5.544435794728892e-07,
      "loss": 3.4196,
      "step": 77080
    },
    {
      "epoch": 0.784149169921875,
      "grad_norm": 11.389708518981934,
      "learning_rate": 5.541924837035131e-07,
      "loss": 3.461,
      "step": 77085
    },
    {
      "epoch": 0.7842000325520834,
      "grad_norm": 15.930747032165527,
      "learning_rate": 5.539414377171601e-07,
      "loss": 3.5088,
      "step": 77090
    },
    {
      "epoch": 0.7842508951822916,
      "grad_norm": 10.358973503112793,
      "learning_rate": 5.536904415202546e-07,
      "loss": 3.6609,
      "step": 77095
    },
    {
      "epoch": 0.7843017578125,
      "grad_norm": 10.959451675415039,
      "learning_rate": 5.534394951192182e-07,
      "loss": 3.2099,
      "step": 77100
    },
    {
      "epoch": 0.7843526204427084,
      "grad_norm": 13.007122993469238,
      "learning_rate": 5.53188598520471e-07,
      "loss": 3.0855,
      "step": 77105
    },
    {
      "epoch": 0.7844034830729166,
      "grad_norm": 13.349512100219727,
      "learning_rate": 5.529377517304313e-07,
      "loss": 3.2852,
      "step": 77110
    },
    {
      "epoch": 0.784454345703125,
      "grad_norm": 77.22537994384766,
      "learning_rate": 5.526869547555186e-07,
      "loss": 3.686,
      "step": 77115
    },
    {
      "epoch": 0.7845052083333334,
      "grad_norm": 9.13428020477295,
      "learning_rate": 5.524362076021478e-07,
      "loss": 3.4315,
      "step": 77120
    },
    {
      "epoch": 0.7845560709635416,
      "grad_norm": 15.25651741027832,
      "learning_rate": 5.521855102767348e-07,
      "loss": 3.1172,
      "step": 77125
    },
    {
      "epoch": 0.78460693359375,
      "grad_norm": 15.090691566467285,
      "learning_rate": 5.51934862785695e-07,
      "loss": 3.2714,
      "step": 77130
    },
    {
      "epoch": 0.7846577962239584,
      "grad_norm": 13.206595420837402,
      "learning_rate": 5.516842651354395e-07,
      "loss": 3.1315,
      "step": 77135
    },
    {
      "epoch": 0.7847086588541666,
      "grad_norm": 12.150652885437012,
      "learning_rate": 5.514337173323805e-07,
      "loss": 3.4068,
      "step": 77140
    },
    {
      "epoch": 0.784759521484375,
      "grad_norm": 12.016641616821289,
      "learning_rate": 5.511832193829272e-07,
      "loss": 3.3858,
      "step": 77145
    },
    {
      "epoch": 0.7848103841145834,
      "grad_norm": 13.401910781860352,
      "learning_rate": 5.509327712934895e-07,
      "loss": 2.8194,
      "step": 77150
    },
    {
      "epoch": 0.7848612467447916,
      "grad_norm": 8.491545677185059,
      "learning_rate": 5.506823730704755e-07,
      "loss": 3.0501,
      "step": 77155
    },
    {
      "epoch": 0.784912109375,
      "grad_norm": 11.794517517089844,
      "learning_rate": 5.50432024720291e-07,
      "loss": 3.6858,
      "step": 77160
    },
    {
      "epoch": 0.7849629720052084,
      "grad_norm": 15.679458618164062,
      "learning_rate": 5.501817262493403e-07,
      "loss": 2.8706,
      "step": 77165
    },
    {
      "epoch": 0.7850138346354166,
      "grad_norm": 10.70201301574707,
      "learning_rate": 5.499314776640277e-07,
      "loss": 3.8139,
      "step": 77170
    },
    {
      "epoch": 0.785064697265625,
      "grad_norm": 14.19828987121582,
      "learning_rate": 5.49681278970757e-07,
      "loss": 3.2887,
      "step": 77175
    },
    {
      "epoch": 0.7851155598958334,
      "grad_norm": 10.544234275817871,
      "learning_rate": 5.494311301759287e-07,
      "loss": 3.0972,
      "step": 77180
    },
    {
      "epoch": 0.7851664225260416,
      "grad_norm": 10.095036506652832,
      "learning_rate": 5.491810312859417e-07,
      "loss": 3.2643,
      "step": 77185
    },
    {
      "epoch": 0.78521728515625,
      "grad_norm": 15.456887245178223,
      "learning_rate": 5.489309823071962e-07,
      "loss": 3.5483,
      "step": 77190
    },
    {
      "epoch": 0.7852681477864584,
      "grad_norm": 11.434060096740723,
      "learning_rate": 5.486809832460882e-07,
      "loss": 3.1546,
      "step": 77195
    },
    {
      "epoch": 0.7853190104166666,
      "grad_norm": 11.786846160888672,
      "learning_rate": 5.484310341090157e-07,
      "loss": 3.4915,
      "step": 77200
    },
    {
      "epoch": 0.785369873046875,
      "grad_norm": 11.120028495788574,
      "learning_rate": 5.481811349023719e-07,
      "loss": 3.5105,
      "step": 77205
    },
    {
      "epoch": 0.7854207356770834,
      "grad_norm": 13.742429733276367,
      "learning_rate": 5.479312856325519e-07,
      "loss": 3.2009,
      "step": 77210
    },
    {
      "epoch": 0.7854715983072916,
      "grad_norm": 12.266082763671875,
      "learning_rate": 5.476814863059474e-07,
      "loss": 3.7984,
      "step": 77215
    },
    {
      "epoch": 0.7855224609375,
      "grad_norm": 12.933247566223145,
      "learning_rate": 5.474317369289483e-07,
      "loss": 3.2753,
      "step": 77220
    },
    {
      "epoch": 0.7855733235677084,
      "grad_norm": 11.963645935058594,
      "learning_rate": 5.471820375079453e-07,
      "loss": 3.3267,
      "step": 77225
    },
    {
      "epoch": 0.7856241861979166,
      "grad_norm": 11.38316822052002,
      "learning_rate": 5.46932388049328e-07,
      "loss": 3.4104,
      "step": 77230
    },
    {
      "epoch": 0.785675048828125,
      "grad_norm": 16.832744598388672,
      "learning_rate": 5.466827885594828e-07,
      "loss": 3.7793,
      "step": 77235
    },
    {
      "epoch": 0.7857259114583334,
      "grad_norm": 9.768202781677246,
      "learning_rate": 5.464332390447943e-07,
      "loss": 3.1119,
      "step": 77240
    },
    {
      "epoch": 0.7857767740885416,
      "grad_norm": 11.314229011535645,
      "learning_rate": 5.461837395116492e-07,
      "loss": 3.2299,
      "step": 77245
    },
    {
      "epoch": 0.78582763671875,
      "grad_norm": 12.870085716247559,
      "learning_rate": 5.459342899664293e-07,
      "loss": 3.2766,
      "step": 77250
    },
    {
      "epoch": 0.7858784993489584,
      "grad_norm": 7.7160515785217285,
      "learning_rate": 5.456848904155185e-07,
      "loss": 3.1813,
      "step": 77255
    },
    {
      "epoch": 0.7859293619791666,
      "grad_norm": 10.903667449951172,
      "learning_rate": 5.454355408652954e-07,
      "loss": 3.1384,
      "step": 77260
    },
    {
      "epoch": 0.785980224609375,
      "grad_norm": 8.76274585723877,
      "learning_rate": 5.451862413221415e-07,
      "loss": 3.6507,
      "step": 77265
    },
    {
      "epoch": 0.7860310872395834,
      "grad_norm": 14.959699630737305,
      "learning_rate": 5.449369917924344e-07,
      "loss": 3.6169,
      "step": 77270
    },
    {
      "epoch": 0.7860819498697916,
      "grad_norm": 8.773665428161621,
      "learning_rate": 5.446877922825502e-07,
      "loss": 3.1942,
      "step": 77275
    },
    {
      "epoch": 0.7861328125,
      "grad_norm": 11.461071014404297,
      "learning_rate": 5.444386427988655e-07,
      "loss": 3.3724,
      "step": 77280
    },
    {
      "epoch": 0.7861836751302084,
      "grad_norm": 11.221567153930664,
      "learning_rate": 5.441895433477551e-07,
      "loss": 3.1541,
      "step": 77285
    },
    {
      "epoch": 0.7862345377604166,
      "grad_norm": 13.72826862335205,
      "learning_rate": 5.439404939355916e-07,
      "loss": 3.2305,
      "step": 77290
    },
    {
      "epoch": 0.786285400390625,
      "grad_norm": 7.363616943359375,
      "learning_rate": 5.436914945687472e-07,
      "loss": 3.0984,
      "step": 77295
    },
    {
      "epoch": 0.7863362630208334,
      "grad_norm": 17.01660919189453,
      "learning_rate": 5.434425452535913e-07,
      "loss": 3.1953,
      "step": 77300
    },
    {
      "epoch": 0.7863871256510416,
      "grad_norm": 12.926891326904297,
      "learning_rate": 5.431936459964943e-07,
      "loss": 3.2446,
      "step": 77305
    },
    {
      "epoch": 0.78643798828125,
      "grad_norm": 11.168578147888184,
      "learning_rate": 5.429447968038245e-07,
      "loss": 3.0408,
      "step": 77310
    },
    {
      "epoch": 0.7864888509114584,
      "grad_norm": 8.66266918182373,
      "learning_rate": 5.426959976819476e-07,
      "loss": 3.617,
      "step": 77315
    },
    {
      "epoch": 0.7865397135416666,
      "grad_norm": 11.905282974243164,
      "learning_rate": 5.424472486372304e-07,
      "loss": 3.3476,
      "step": 77320
    },
    {
      "epoch": 0.786590576171875,
      "grad_norm": 12.501431465148926,
      "learning_rate": 5.421985496760357e-07,
      "loss": 3.3132,
      "step": 77325
    },
    {
      "epoch": 0.7866414388020834,
      "grad_norm": 9.089613914489746,
      "learning_rate": 5.419499008047277e-07,
      "loss": 3.2856,
      "step": 77330
    },
    {
      "epoch": 0.7866923014322916,
      "grad_norm": 10.10815715789795,
      "learning_rate": 5.417013020296669e-07,
      "loss": 3.7829,
      "step": 77335
    },
    {
      "epoch": 0.7867431640625,
      "grad_norm": 14.02984619140625,
      "learning_rate": 5.414527533572145e-07,
      "loss": 3.0122,
      "step": 77340
    },
    {
      "epoch": 0.7867940266927084,
      "grad_norm": 14.067699432373047,
      "learning_rate": 5.412042547937294e-07,
      "loss": 3.0384,
      "step": 77345
    },
    {
      "epoch": 0.7868448893229166,
      "grad_norm": 13.12199592590332,
      "learning_rate": 5.409558063455683e-07,
      "loss": 3.3337,
      "step": 77350
    },
    {
      "epoch": 0.786895751953125,
      "grad_norm": 11.288161277770996,
      "learning_rate": 5.407074080190886e-07,
      "loss": 3.4466,
      "step": 77355
    },
    {
      "epoch": 0.7869466145833334,
      "grad_norm": 13.173538208007812,
      "learning_rate": 5.404590598206461e-07,
      "loss": 3.131,
      "step": 77360
    },
    {
      "epoch": 0.7869974772135416,
      "grad_norm": 8.635846138000488,
      "learning_rate": 5.402107617565941e-07,
      "loss": 3.53,
      "step": 77365
    },
    {
      "epoch": 0.78704833984375,
      "grad_norm": 12.688966751098633,
      "learning_rate": 5.399625138332854e-07,
      "loss": 3.4413,
      "step": 77370
    },
    {
      "epoch": 0.7870992024739584,
      "grad_norm": 11.470646858215332,
      "learning_rate": 5.397143160570703e-07,
      "loss": 3.4997,
      "step": 77375
    },
    {
      "epoch": 0.7871500651041666,
      "grad_norm": 13.815547943115234,
      "learning_rate": 5.394661684343e-07,
      "loss": 3.3771,
      "step": 77380
    },
    {
      "epoch": 0.787200927734375,
      "grad_norm": 16.58146095275879,
      "learning_rate": 5.392180709713235e-07,
      "loss": 3.2503,
      "step": 77385
    },
    {
      "epoch": 0.7872517903645834,
      "grad_norm": 16.185945510864258,
      "learning_rate": 5.389700236744883e-07,
      "loss": 3.7747,
      "step": 77390
    },
    {
      "epoch": 0.7873026529947916,
      "grad_norm": 7.731017589569092,
      "learning_rate": 5.387220265501392e-07,
      "loss": 3.1052,
      "step": 77395
    },
    {
      "epoch": 0.787353515625,
      "grad_norm": 15.839229583740234,
      "learning_rate": 5.384740796046228e-07,
      "loss": 3.1966,
      "step": 77400
    },
    {
      "epoch": 0.7874043782552084,
      "grad_norm": 10.629376411437988,
      "learning_rate": 5.382261828442817e-07,
      "loss": 3.1147,
      "step": 77405
    },
    {
      "epoch": 0.7874552408854166,
      "grad_norm": 17.61418342590332,
      "learning_rate": 5.379783362754587e-07,
      "loss": 3.2562,
      "step": 77410
    },
    {
      "epoch": 0.787506103515625,
      "grad_norm": 10.025396347045898,
      "learning_rate": 5.377305399044954e-07,
      "loss": 3.0456,
      "step": 77415
    },
    {
      "epoch": 0.7875569661458334,
      "grad_norm": 9.159224510192871,
      "learning_rate": 5.374827937377314e-07,
      "loss": 3.2772,
      "step": 77420
    },
    {
      "epoch": 0.7876078287760416,
      "grad_norm": 8.789568901062012,
      "learning_rate": 5.37235097781505e-07,
      "loss": 2.9066,
      "step": 77425
    },
    {
      "epoch": 0.78765869140625,
      "grad_norm": 7.427765369415283,
      "learning_rate": 5.369874520421525e-07,
      "loss": 3.4291,
      "step": 77430
    },
    {
      "epoch": 0.7877095540364584,
      "grad_norm": 9.716547012329102,
      "learning_rate": 5.367398565260107e-07,
      "loss": 3.0209,
      "step": 77435
    },
    {
      "epoch": 0.7877604166666666,
      "grad_norm": 9.71487045288086,
      "learning_rate": 5.364923112394152e-07,
      "loss": 3.5569,
      "step": 77440
    },
    {
      "epoch": 0.787811279296875,
      "grad_norm": 12.40367317199707,
      "learning_rate": 5.362448161886985e-07,
      "loss": 3.2814,
      "step": 77445
    },
    {
      "epoch": 0.7878621419270834,
      "grad_norm": 13.2744140625,
      "learning_rate": 5.359973713801919e-07,
      "loss": 3.308,
      "step": 77450
    },
    {
      "epoch": 0.7879130045572916,
      "grad_norm": 10.100786209106445,
      "learning_rate": 5.357499768202276e-07,
      "loss": 3.1497,
      "step": 77455
    },
    {
      "epoch": 0.7879638671875,
      "grad_norm": 8.46898078918457,
      "learning_rate": 5.355026325151341e-07,
      "loss": 3.32,
      "step": 77460
    },
    {
      "epoch": 0.7880147298177084,
      "grad_norm": 7.277879238128662,
      "learning_rate": 5.352553384712408e-07,
      "loss": 3.2946,
      "step": 77465
    },
    {
      "epoch": 0.7880655924479166,
      "grad_norm": 15.188435554504395,
      "learning_rate": 5.350080946948732e-07,
      "loss": 3.06,
      "step": 77470
    },
    {
      "epoch": 0.788116455078125,
      "grad_norm": 13.343172073364258,
      "learning_rate": 5.347609011923585e-07,
      "loss": 2.9495,
      "step": 77475
    },
    {
      "epoch": 0.7881673177083334,
      "grad_norm": 11.459025382995605,
      "learning_rate": 5.345137579700202e-07,
      "loss": 3.5003,
      "step": 77480
    },
    {
      "epoch": 0.7882181803385416,
      "grad_norm": 9.683830261230469,
      "learning_rate": 5.34266665034181e-07,
      "loss": 3.2973,
      "step": 77485
    },
    {
      "epoch": 0.78826904296875,
      "grad_norm": 8.150592803955078,
      "learning_rate": 5.340196223911631e-07,
      "loss": 3.2292,
      "step": 77490
    },
    {
      "epoch": 0.7883199055989584,
      "grad_norm": 8.487361907958984,
      "learning_rate": 5.337726300472882e-07,
      "loss": 3.2224,
      "step": 77495
    },
    {
      "epoch": 0.7883707682291666,
      "grad_norm": 11.931181907653809,
      "learning_rate": 5.33525688008874e-07,
      "loss": 3.3767,
      "step": 77500
    },
    {
      "epoch": 0.788421630859375,
      "grad_norm": 15.323692321777344,
      "learning_rate": 5.332787962822386e-07,
      "loss": 3.1439,
      "step": 77505
    },
    {
      "epoch": 0.7884724934895834,
      "grad_norm": 13.419575691223145,
      "learning_rate": 5.330319548736989e-07,
      "loss": 3.4456,
      "step": 77510
    },
    {
      "epoch": 0.7885233561197916,
      "grad_norm": 10.741533279418945,
      "learning_rate": 5.327851637895715e-07,
      "loss": 2.965,
      "step": 77515
    },
    {
      "epoch": 0.78857421875,
      "grad_norm": 10.303194999694824,
      "learning_rate": 5.325384230361689e-07,
      "loss": 3.5112,
      "step": 77520
    },
    {
      "epoch": 0.7886250813802084,
      "grad_norm": 12.528802871704102,
      "learning_rate": 5.322917326198038e-07,
      "loss": 3.5571,
      "step": 77525
    },
    {
      "epoch": 0.7886759440104166,
      "grad_norm": 9.583382606506348,
      "learning_rate": 5.320450925467893e-07,
      "loss": 3.1225,
      "step": 77530
    },
    {
      "epoch": 0.788726806640625,
      "grad_norm": 12.141776084899902,
      "learning_rate": 5.317985028234338e-07,
      "loss": 2.9998,
      "step": 77535
    },
    {
      "epoch": 0.7887776692708334,
      "grad_norm": 15.204798698425293,
      "learning_rate": 5.315519634560476e-07,
      "loss": 3.3831,
      "step": 77540
    },
    {
      "epoch": 0.7888285319010416,
      "grad_norm": 8.962828636169434,
      "learning_rate": 5.313054744509374e-07,
      "loss": 3.173,
      "step": 77545
    },
    {
      "epoch": 0.78887939453125,
      "grad_norm": 12.641799926757812,
      "learning_rate": 5.310590358144105e-07,
      "loss": 3.7621,
      "step": 77550
    },
    {
      "epoch": 0.7889302571614584,
      "grad_norm": 15.058570861816406,
      "learning_rate": 5.308126475527717e-07,
      "loss": 3.1571,
      "step": 77555
    },
    {
      "epoch": 0.7889811197916666,
      "grad_norm": 10.706228256225586,
      "learning_rate": 5.305663096723237e-07,
      "loss": 3.0751,
      "step": 77560
    },
    {
      "epoch": 0.789031982421875,
      "grad_norm": 9.943954467773438,
      "learning_rate": 5.303200221793697e-07,
      "loss": 3.2576,
      "step": 77565
    },
    {
      "epoch": 0.7890828450520834,
      "grad_norm": 10.609087944030762,
      "learning_rate": 5.300737850802121e-07,
      "loss": 3.4295,
      "step": 77570
    },
    {
      "epoch": 0.7891337076822916,
      "grad_norm": 12.822108268737793,
      "learning_rate": 5.298275983811496e-07,
      "loss": 3.605,
      "step": 77575
    },
    {
      "epoch": 0.7891845703125,
      "grad_norm": 11.368073463439941,
      "learning_rate": 5.295814620884812e-07,
      "loss": 3.2867,
      "step": 77580
    },
    {
      "epoch": 0.7892354329427084,
      "grad_norm": 13.873319625854492,
      "learning_rate": 5.293353762085033e-07,
      "loss": 3.5609,
      "step": 77585
    },
    {
      "epoch": 0.7892862955729166,
      "grad_norm": 11.019356727600098,
      "learning_rate": 5.290893407475125e-07,
      "loss": 3.4088,
      "step": 77590
    },
    {
      "epoch": 0.789337158203125,
      "grad_norm": 11.461319923400879,
      "learning_rate": 5.288433557118044e-07,
      "loss": 3.2105,
      "step": 77595
    },
    {
      "epoch": 0.7893880208333334,
      "grad_norm": 14.308027267456055,
      "learning_rate": 5.285974211076714e-07,
      "loss": 3.3516,
      "step": 77600
    },
    {
      "epoch": 0.7894388834635416,
      "grad_norm": 14.984569549560547,
      "learning_rate": 5.283515369414069e-07,
      "loss": 3.3032,
      "step": 77605
    },
    {
      "epoch": 0.78948974609375,
      "grad_norm": 9.603241920471191,
      "learning_rate": 5.281057032193008e-07,
      "loss": 3.3644,
      "step": 77610
    },
    {
      "epoch": 0.7895406087239584,
      "grad_norm": 12.334107398986816,
      "learning_rate": 5.27859919947642e-07,
      "loss": 3.74,
      "step": 77615
    },
    {
      "epoch": 0.7895914713541666,
      "grad_norm": 12.757811546325684,
      "learning_rate": 5.2761418713272e-07,
      "loss": 3.3776,
      "step": 77620
    },
    {
      "epoch": 0.789642333984375,
      "grad_norm": 15.814942359924316,
      "learning_rate": 5.273685047808222e-07,
      "loss": 3.3104,
      "step": 77625
    },
    {
      "epoch": 0.7896931966145834,
      "grad_norm": 13.55041790008545,
      "learning_rate": 5.271228728982333e-07,
      "loss": 3.6321,
      "step": 77630
    },
    {
      "epoch": 0.7897440592447916,
      "grad_norm": 12.658421516418457,
      "learning_rate": 5.268772914912384e-07,
      "loss": 3.3185,
      "step": 77635
    },
    {
      "epoch": 0.789794921875,
      "grad_norm": 14.816579818725586,
      "learning_rate": 5.266317605661189e-07,
      "loss": 3.0483,
      "step": 77640
    },
    {
      "epoch": 0.7898457845052084,
      "grad_norm": 9.840069770812988,
      "learning_rate": 5.263862801291594e-07,
      "loss": 3.4473,
      "step": 77645
    },
    {
      "epoch": 0.7898966471354166,
      "grad_norm": 16.5360050201416,
      "learning_rate": 5.261408501866391e-07,
      "loss": 3.3267,
      "step": 77650
    },
    {
      "epoch": 0.789947509765625,
      "grad_norm": 11.0403470993042,
      "learning_rate": 5.258954707448375e-07,
      "loss": 3.53,
      "step": 77655
    },
    {
      "epoch": 0.7899983723958334,
      "grad_norm": 7.623027324676514,
      "learning_rate": 5.256501418100316e-07,
      "loss": 3.5695,
      "step": 77660
    },
    {
      "epoch": 0.7900492350260416,
      "grad_norm": 10.7011137008667,
      "learning_rate": 5.254048633884987e-07,
      "loss": 3.437,
      "step": 77665
    },
    {
      "epoch": 0.79010009765625,
      "grad_norm": 9.61572551727295,
      "learning_rate": 5.251596354865148e-07,
      "loss": 3.0816,
      "step": 77670
    },
    {
      "epoch": 0.7901509602864584,
      "grad_norm": 13.822772979736328,
      "learning_rate": 5.249144581103541e-07,
      "loss": 3.3933,
      "step": 77675
    },
    {
      "epoch": 0.7902018229166666,
      "grad_norm": 9.52763843536377,
      "learning_rate": 5.246693312662878e-07,
      "loss": 3.4029,
      "step": 77680
    },
    {
      "epoch": 0.790252685546875,
      "grad_norm": 8.417424201965332,
      "learning_rate": 5.24424254960589e-07,
      "loss": 3.1074,
      "step": 77685
    },
    {
      "epoch": 0.7903035481770834,
      "grad_norm": 13.455845832824707,
      "learning_rate": 5.241792291995271e-07,
      "loss": 3.5337,
      "step": 77690
    },
    {
      "epoch": 0.7903544108072916,
      "grad_norm": 23.03131103515625,
      "learning_rate": 5.239342539893719e-07,
      "loss": 3.5476,
      "step": 77695
    },
    {
      "epoch": 0.7904052734375,
      "grad_norm": 12.35528564453125,
      "learning_rate": 5.236893293363895e-07,
      "loss": 3.417,
      "step": 77700
    },
    {
      "epoch": 0.7904561360677084,
      "grad_norm": 11.345623970031738,
      "learning_rate": 5.234444552468482e-07,
      "loss": 3.1162,
      "step": 77705
    },
    {
      "epoch": 0.7905069986979166,
      "grad_norm": 16.01922035217285,
      "learning_rate": 5.231996317270116e-07,
      "loss": 3.4665,
      "step": 77710
    },
    {
      "epoch": 0.790557861328125,
      "grad_norm": 7.456033229827881,
      "learning_rate": 5.229548587831434e-07,
      "loss": 3.3849,
      "step": 77715
    },
    {
      "epoch": 0.7906087239583334,
      "grad_norm": 13.59242057800293,
      "learning_rate": 5.227101364215065e-07,
      "loss": 3.0704,
      "step": 77720
    },
    {
      "epoch": 0.7906595865885416,
      "grad_norm": 15.409017562866211,
      "learning_rate": 5.224654646483629e-07,
      "loss": 3.2839,
      "step": 77725
    },
    {
      "epoch": 0.79071044921875,
      "grad_norm": 13.718083381652832,
      "learning_rate": 5.222208434699716e-07,
      "loss": 3.6831,
      "step": 77730
    },
    {
      "epoch": 0.7907613118489584,
      "grad_norm": 14.573699951171875,
      "learning_rate": 5.219762728925903e-07,
      "loss": 3.0561,
      "step": 77735
    },
    {
      "epoch": 0.7908121744791666,
      "grad_norm": 10.57568359375,
      "learning_rate": 5.217317529224783e-07,
      "loss": 3.0412,
      "step": 77740
    },
    {
      "epoch": 0.790863037109375,
      "grad_norm": 13.119362831115723,
      "learning_rate": 5.214872835658896e-07,
      "loss": 3.3877,
      "step": 77745
    },
    {
      "epoch": 0.7909138997395834,
      "grad_norm": 7.986304759979248,
      "learning_rate": 5.212428648290804e-07,
      "loss": 2.9841,
      "step": 77750
    },
    {
      "epoch": 0.7909647623697916,
      "grad_norm": 9.003944396972656,
      "learning_rate": 5.209984967183029e-07,
      "loss": 3.1749,
      "step": 77755
    },
    {
      "epoch": 0.791015625,
      "grad_norm": 14.619218826293945,
      "learning_rate": 5.207541792398105e-07,
      "loss": 3.3182,
      "step": 77760
    },
    {
      "epoch": 0.7910664876302084,
      "grad_norm": 14.97256088256836,
      "learning_rate": 5.205099123998533e-07,
      "loss": 3.3165,
      "step": 77765
    },
    {
      "epoch": 0.7911173502604166,
      "grad_norm": 15.13031005859375,
      "learning_rate": 5.202656962046801e-07,
      "loss": 3.5063,
      "step": 77770
    },
    {
      "epoch": 0.791168212890625,
      "grad_norm": 17.268144607543945,
      "learning_rate": 5.2002153066054e-07,
      "loss": 3.4668,
      "step": 77775
    },
    {
      "epoch": 0.7912190755208334,
      "grad_norm": 10.193543434143066,
      "learning_rate": 5.197774157736801e-07,
      "loss": 3.3866,
      "step": 77780
    },
    {
      "epoch": 0.7912699381510416,
      "grad_norm": 9.963932991027832,
      "learning_rate": 5.195333515503456e-07,
      "loss": 2.7359,
      "step": 77785
    },
    {
      "epoch": 0.79132080078125,
      "grad_norm": 13.159423828125,
      "learning_rate": 5.192893379967812e-07,
      "loss": 3.2852,
      "step": 77790
    },
    {
      "epoch": 0.7913716634114584,
      "grad_norm": 7.917318344116211,
      "learning_rate": 5.190453751192281e-07,
      "loss": 3.5465,
      "step": 77795
    },
    {
      "epoch": 0.7914225260416666,
      "grad_norm": 43.193885803222656,
      "learning_rate": 5.188014629239308e-07,
      "loss": 3.7018,
      "step": 77800
    },
    {
      "epoch": 0.791473388671875,
      "grad_norm": 13.045291900634766,
      "learning_rate": 5.185576014171287e-07,
      "loss": 3.1991,
      "step": 77805
    },
    {
      "epoch": 0.7915242513020834,
      "grad_norm": 10.419424057006836,
      "learning_rate": 5.183137906050598e-07,
      "loss": 3.6339,
      "step": 77810
    },
    {
      "epoch": 0.7915751139322916,
      "grad_norm": 11.010937690734863,
      "learning_rate": 5.180700304939637e-07,
      "loss": 3.3555,
      "step": 77815
    },
    {
      "epoch": 0.7916259765625,
      "grad_norm": 10.980416297912598,
      "learning_rate": 5.178263210900752e-07,
      "loss": 3.0911,
      "step": 77820
    },
    {
      "epoch": 0.7916768391927084,
      "grad_norm": 7.204528331756592,
      "learning_rate": 5.175826623996313e-07,
      "loss": 3.272,
      "step": 77825
    },
    {
      "epoch": 0.7917277018229166,
      "grad_norm": 19.492891311645508,
      "learning_rate": 5.173390544288645e-07,
      "loss": 3.3251,
      "step": 77830
    },
    {
      "epoch": 0.791778564453125,
      "grad_norm": 10.955602645874023,
      "learning_rate": 5.170954971840087e-07,
      "loss": 3.5388,
      "step": 77835
    },
    {
      "epoch": 0.7918294270833334,
      "grad_norm": 12.813833236694336,
      "learning_rate": 5.168519906712946e-07,
      "loss": 3.3107,
      "step": 77840
    },
    {
      "epoch": 0.7918802897135416,
      "grad_norm": 9.102631568908691,
      "learning_rate": 5.166085348969516e-07,
      "loss": 3.3827,
      "step": 77845
    },
    {
      "epoch": 0.79193115234375,
      "grad_norm": 20.36287498474121,
      "learning_rate": 5.16365129867209e-07,
      "loss": 3.4709,
      "step": 77850
    },
    {
      "epoch": 0.7919820149739584,
      "grad_norm": 9.53981876373291,
      "learning_rate": 5.16121775588295e-07,
      "loss": 3.2573,
      "step": 77855
    },
    {
      "epoch": 0.7920328776041666,
      "grad_norm": 10.992317199707031,
      "learning_rate": 5.158784720664353e-07,
      "loss": 3.138,
      "step": 77860
    },
    {
      "epoch": 0.792083740234375,
      "grad_norm": 10.903949737548828,
      "learning_rate": 5.156352193078545e-07,
      "loss": 3.539,
      "step": 77865
    },
    {
      "epoch": 0.7921346028645834,
      "grad_norm": 8.236926078796387,
      "learning_rate": 5.153920173187757e-07,
      "loss": 3.426,
      "step": 77870
    },
    {
      "epoch": 0.7921854654947916,
      "grad_norm": 8.5499267578125,
      "learning_rate": 5.151488661054216e-07,
      "loss": 2.9827,
      "step": 77875
    },
    {
      "epoch": 0.792236328125,
      "grad_norm": 14.410651206970215,
      "learning_rate": 5.149057656740139e-07,
      "loss": 4.5449,
      "step": 77880
    },
    {
      "epoch": 0.7922871907552084,
      "grad_norm": 14.640835762023926,
      "learning_rate": 5.146627160307718e-07,
      "loss": 3.1035,
      "step": 77885
    },
    {
      "epoch": 0.7923380533854166,
      "grad_norm": 9.395439147949219,
      "learning_rate": 5.144197171819129e-07,
      "loss": 3.1243,
      "step": 77890
    },
    {
      "epoch": 0.792388916015625,
      "grad_norm": 9.009712219238281,
      "learning_rate": 5.141767691336558e-07,
      "loss": 3.7954,
      "step": 77895
    },
    {
      "epoch": 0.7924397786458334,
      "grad_norm": 12.002073287963867,
      "learning_rate": 5.139338718922143e-07,
      "loss": 3.1543,
      "step": 77900
    },
    {
      "epoch": 0.7924906412760416,
      "grad_norm": 9.392443656921387,
      "learning_rate": 5.136910254638042e-07,
      "loss": 3.1261,
      "step": 77905
    },
    {
      "epoch": 0.79254150390625,
      "grad_norm": 9.431241035461426,
      "learning_rate": 5.13448229854639e-07,
      "loss": 3.3929,
      "step": 77910
    },
    {
      "epoch": 0.7925923665364584,
      "grad_norm": 9.476242065429688,
      "learning_rate": 5.132054850709298e-07,
      "loss": 2.8419,
      "step": 77915
    },
    {
      "epoch": 0.7926432291666666,
      "grad_norm": 11.033794403076172,
      "learning_rate": 5.129627911188876e-07,
      "loss": 3.1382,
      "step": 77920
    },
    {
      "epoch": 0.792694091796875,
      "grad_norm": 10.795284271240234,
      "learning_rate": 5.127201480047206e-07,
      "loss": 3.0693,
      "step": 77925
    },
    {
      "epoch": 0.7927449544270834,
      "grad_norm": 14.76865291595459,
      "learning_rate": 5.124775557346376e-07,
      "loss": 3.5296,
      "step": 77930
    },
    {
      "epoch": 0.7927958170572916,
      "grad_norm": 8.955913543701172,
      "learning_rate": 5.122350143148458e-07,
      "loss": 3.3339,
      "step": 77935
    },
    {
      "epoch": 0.7928466796875,
      "grad_norm": 13.093903541564941,
      "learning_rate": 5.119925237515502e-07,
      "loss": 3.5774,
      "step": 77940
    },
    {
      "epoch": 0.7928975423177084,
      "grad_norm": 15.690264701843262,
      "learning_rate": 5.117500840509537e-07,
      "loss": 3.4012,
      "step": 77945
    },
    {
      "epoch": 0.7929484049479166,
      "grad_norm": 10.985313415527344,
      "learning_rate": 5.115076952192605e-07,
      "loss": 3.2087,
      "step": 77950
    },
    {
      "epoch": 0.792999267578125,
      "grad_norm": 8.62189769744873,
      "learning_rate": 5.112653572626719e-07,
      "loss": 3.5361,
      "step": 77955
    },
    {
      "epoch": 0.7930501302083334,
      "grad_norm": 14.569122314453125,
      "learning_rate": 5.110230701873878e-07,
      "loss": 3.0355,
      "step": 77960
    },
    {
      "epoch": 0.7931009928385416,
      "grad_norm": 9.006644248962402,
      "learning_rate": 5.107808339996065e-07,
      "loss": 3.3467,
      "step": 77965
    },
    {
      "epoch": 0.79315185546875,
      "grad_norm": 8.4939603805542,
      "learning_rate": 5.105386487055267e-07,
      "loss": 3.8445,
      "step": 77970
    },
    {
      "epoch": 0.7932027180989584,
      "grad_norm": 10.302860260009766,
      "learning_rate": 5.102965143113431e-07,
      "loss": 3.6555,
      "step": 77975
    },
    {
      "epoch": 0.7932535807291666,
      "grad_norm": 8.247651100158691,
      "learning_rate": 5.100544308232525e-07,
      "loss": 3.6142,
      "step": 77980
    },
    {
      "epoch": 0.793304443359375,
      "grad_norm": 10.880356788635254,
      "learning_rate": 5.098123982474468e-07,
      "loss": 3.0245,
      "step": 77985
    },
    {
      "epoch": 0.7933553059895834,
      "grad_norm": 10.192167282104492,
      "learning_rate": 5.095704165901199e-07,
      "loss": 2.9427,
      "step": 77990
    },
    {
      "epoch": 0.7934061686197916,
      "grad_norm": 11.401342391967773,
      "learning_rate": 5.093284858574621e-07,
      "loss": 3.198,
      "step": 77995
    },
    {
      "epoch": 0.79345703125,
      "grad_norm": 55.87403106689453,
      "learning_rate": 5.090866060556626e-07,
      "loss": 3.3743,
      "step": 78000
    },
    {
      "epoch": 0.7935078938802084,
      "grad_norm": 13.474732398986816,
      "learning_rate": 5.088447771909102e-07,
      "loss": 3.3572,
      "step": 78005
    },
    {
      "epoch": 0.7935587565104166,
      "grad_norm": 13.210291862487793,
      "learning_rate": 5.086029992693931e-07,
      "loss": 3.3775,
      "step": 78010
    },
    {
      "epoch": 0.793609619140625,
      "grad_norm": 14.540139198303223,
      "learning_rate": 5.083612722972961e-07,
      "loss": 3.5608,
      "step": 78015
    },
    {
      "epoch": 0.7936604817708334,
      "grad_norm": 9.909323692321777,
      "learning_rate": 5.081195962808033e-07,
      "loss": 3.5092,
      "step": 78020
    },
    {
      "epoch": 0.7937113444010416,
      "grad_norm": 9.023208618164062,
      "learning_rate": 5.078779712260989e-07,
      "loss": 3.3079,
      "step": 78025
    },
    {
      "epoch": 0.79376220703125,
      "grad_norm": 10.259096145629883,
      "learning_rate": 5.076363971393641e-07,
      "loss": 3.3976,
      "step": 78030
    },
    {
      "epoch": 0.7938130696614584,
      "grad_norm": 8.69201374053955,
      "learning_rate": 5.073948740267804e-07,
      "loss": 3.624,
      "step": 78035
    },
    {
      "epoch": 0.7938639322916666,
      "grad_norm": 13.281329154968262,
      "learning_rate": 5.071534018945259e-07,
      "loss": 3.7112,
      "step": 78040
    },
    {
      "epoch": 0.793914794921875,
      "grad_norm": 9.79742431640625,
      "learning_rate": 5.069119807487799e-07,
      "loss": 3.6791,
      "step": 78045
    },
    {
      "epoch": 0.7939656575520834,
      "grad_norm": 12.21177864074707,
      "learning_rate": 5.066706105957183e-07,
      "loss": 3.5715,
      "step": 78050
    },
    {
      "epoch": 0.7940165201822916,
      "grad_norm": 8.74547290802002,
      "learning_rate": 5.064292914415161e-07,
      "loss": 3.4031,
      "step": 78055
    },
    {
      "epoch": 0.7940673828125,
      "grad_norm": 12.38512134552002,
      "learning_rate": 5.06188023292348e-07,
      "loss": 3.2038,
      "step": 78060
    },
    {
      "epoch": 0.7941182454427084,
      "grad_norm": 16.570751190185547,
      "learning_rate": 5.059468061543874e-07,
      "loss": 3.3855,
      "step": 78065
    },
    {
      "epoch": 0.7941691080729166,
      "grad_norm": 10.972708702087402,
      "learning_rate": 5.057056400338051e-07,
      "loss": 3.2059,
      "step": 78070
    },
    {
      "epoch": 0.794219970703125,
      "grad_norm": 12.961288452148438,
      "learning_rate": 5.054645249367712e-07,
      "loss": 3.5962,
      "step": 78075
    },
    {
      "epoch": 0.7942708333333334,
      "grad_norm": 10.799997329711914,
      "learning_rate": 5.052234608694543e-07,
      "loss": 3.6069,
      "step": 78080
    },
    {
      "epoch": 0.7943216959635416,
      "grad_norm": 13.07223129272461,
      "learning_rate": 5.049824478380219e-07,
      "loss": 3.2428,
      "step": 78085
    },
    {
      "epoch": 0.79437255859375,
      "grad_norm": 7.48736047744751,
      "learning_rate": 5.047414858486416e-07,
      "loss": 3.0928,
      "step": 78090
    },
    {
      "epoch": 0.7944234212239584,
      "grad_norm": 14.10659408569336,
      "learning_rate": 5.045005749074769e-07,
      "loss": 3.5605,
      "step": 78095
    },
    {
      "epoch": 0.7944742838541666,
      "grad_norm": 12.803142547607422,
      "learning_rate": 5.042597150206927e-07,
      "loss": 3.2086,
      "step": 78100
    },
    {
      "epoch": 0.794525146484375,
      "grad_norm": 15.745060920715332,
      "learning_rate": 5.040189061944498e-07,
      "loss": 3.5553,
      "step": 78105
    },
    {
      "epoch": 0.7945760091145834,
      "grad_norm": 14.81272029876709,
      "learning_rate": 5.037781484349108e-07,
      "loss": 3.2929,
      "step": 78110
    },
    {
      "epoch": 0.7946268717447916,
      "grad_norm": 15.14911937713623,
      "learning_rate": 5.035374417482339e-07,
      "loss": 3.7916,
      "step": 78115
    },
    {
      "epoch": 0.794677734375,
      "grad_norm": 12.357599258422852,
      "learning_rate": 5.032967861405794e-07,
      "loss": 3.2061,
      "step": 78120
    },
    {
      "epoch": 0.7947285970052084,
      "grad_norm": 9.263826370239258,
      "learning_rate": 5.030561816181029e-07,
      "loss": 3.4975,
      "step": 78125
    },
    {
      "epoch": 0.7947794596354166,
      "grad_norm": 7.138040542602539,
      "learning_rate": 5.028156281869603e-07,
      "loss": 3.1083,
      "step": 78130
    },
    {
      "epoch": 0.794830322265625,
      "grad_norm": 10.713909149169922,
      "learning_rate": 5.025751258533062e-07,
      "loss": 3.2477,
      "step": 78135
    },
    {
      "epoch": 0.7948811848958334,
      "grad_norm": 11.131681442260742,
      "learning_rate": 5.023346746232952e-07,
      "loss": 3.186,
      "step": 78140
    },
    {
      "epoch": 0.7949320475260416,
      "grad_norm": 12.119609832763672,
      "learning_rate": 5.020942745030777e-07,
      "loss": 2.9858,
      "step": 78145
    },
    {
      "epoch": 0.79498291015625,
      "grad_norm": 13.22558879852295,
      "learning_rate": 5.018539254988047e-07,
      "loss": 2.9334,
      "step": 78150
    },
    {
      "epoch": 0.7950337727864584,
      "grad_norm": 10.023087501525879,
      "learning_rate": 5.016136276166247e-07,
      "loss": 3.3949,
      "step": 78155
    },
    {
      "epoch": 0.7950846354166666,
      "grad_norm": 16.232559204101562,
      "learning_rate": 5.013733808626864e-07,
      "loss": 3.1979,
      "step": 78160
    },
    {
      "epoch": 0.795135498046875,
      "grad_norm": 7.1333699226379395,
      "learning_rate": 5.011331852431369e-07,
      "loss": 3.2484,
      "step": 78165
    },
    {
      "epoch": 0.7951863606770834,
      "grad_norm": 12.75872802734375,
      "learning_rate": 5.008930407641215e-07,
      "loss": 3.6665,
      "step": 78170
    },
    {
      "epoch": 0.7952372233072916,
      "grad_norm": 12.93013858795166,
      "learning_rate": 5.006529474317828e-07,
      "loss": 3.1155,
      "step": 78175
    },
    {
      "epoch": 0.7952880859375,
      "grad_norm": 12.088790893554688,
      "learning_rate": 5.004129052522655e-07,
      "loss": 3.1999,
      "step": 78180
    },
    {
      "epoch": 0.7953389485677084,
      "grad_norm": 12.312433242797852,
      "learning_rate": 5.001729142317091e-07,
      "loss": 3.3969,
      "step": 78185
    },
    {
      "epoch": 0.7953898111979166,
      "grad_norm": 13.875852584838867,
      "learning_rate": 4.999329743762548e-07,
      "loss": 3.4245,
      "step": 78190
    },
    {
      "epoch": 0.795440673828125,
      "grad_norm": 12.507823944091797,
      "learning_rate": 4.996930856920421e-07,
      "loss": 3.3563,
      "step": 78195
    },
    {
      "epoch": 0.7954915364583334,
      "grad_norm": 12.309163093566895,
      "learning_rate": 4.994532481852074e-07,
      "loss": 3.3727,
      "step": 78200
    },
    {
      "epoch": 0.7955423990885416,
      "grad_norm": 8.417670249938965,
      "learning_rate": 4.992134618618874e-07,
      "loss": 3.0717,
      "step": 78205
    },
    {
      "epoch": 0.79559326171875,
      "grad_norm": 12.883235931396484,
      "learning_rate": 4.989737267282157e-07,
      "loss": 3.1688,
      "step": 78210
    },
    {
      "epoch": 0.7956441243489584,
      "grad_norm": 17.78324317932129,
      "learning_rate": 4.987340427903273e-07,
      "loss": 3.2281,
      "step": 78215
    },
    {
      "epoch": 0.7956949869791666,
      "grad_norm": 7.117746353149414,
      "learning_rate": 4.984944100543545e-07,
      "loss": 3.1447,
      "step": 78220
    },
    {
      "epoch": 0.795745849609375,
      "grad_norm": 9.414732933044434,
      "learning_rate": 4.98254828526428e-07,
      "loss": 3.3095,
      "step": 78225
    },
    {
      "epoch": 0.7957967122395834,
      "grad_norm": 15.074447631835938,
      "learning_rate": 4.980152982126765e-07,
      "loss": 3.9253,
      "step": 78230
    },
    {
      "epoch": 0.7958475748697916,
      "grad_norm": 8.129987716674805,
      "learning_rate": 4.977758191192297e-07,
      "loss": 3.6704,
      "step": 78235
    },
    {
      "epoch": 0.7958984375,
      "grad_norm": 9.61725902557373,
      "learning_rate": 4.975363912522136e-07,
      "loss": 3.3034,
      "step": 78240
    },
    {
      "epoch": 0.7959493001302084,
      "grad_norm": 15.898126602172852,
      "learning_rate": 4.972970146177544e-07,
      "loss": 3.1708,
      "step": 78245
    },
    {
      "epoch": 0.7960001627604166,
      "grad_norm": 27.77703094482422,
      "learning_rate": 4.970576892219761e-07,
      "loss": 3.2187,
      "step": 78250
    },
    {
      "epoch": 0.796051025390625,
      "grad_norm": 11.904648780822754,
      "learning_rate": 4.968184150710029e-07,
      "loss": 3.2447,
      "step": 78255
    },
    {
      "epoch": 0.7961018880208334,
      "grad_norm": 16.175411224365234,
      "learning_rate": 4.965791921709548e-07,
      "loss": 3.4115,
      "step": 78260
    },
    {
      "epoch": 0.7961527506510416,
      "grad_norm": 13.698991775512695,
      "learning_rate": 4.96340020527954e-07,
      "loss": 3.9765,
      "step": 78265
    },
    {
      "epoch": 0.79620361328125,
      "grad_norm": 10.724797248840332,
      "learning_rate": 4.961009001481179e-07,
      "loss": 3.6566,
      "step": 78270
    },
    {
      "epoch": 0.7962544759114584,
      "grad_norm": 16.67429542541504,
      "learning_rate": 4.958618310375662e-07,
      "loss": 3.0561,
      "step": 78275
    },
    {
      "epoch": 0.7963053385416666,
      "grad_norm": 12.067134857177734,
      "learning_rate": 4.956228132024146e-07,
      "loss": 3.2328,
      "step": 78280
    },
    {
      "epoch": 0.796356201171875,
      "grad_norm": 12.0044527053833,
      "learning_rate": 4.953838466487773e-07,
      "loss": 3.1284,
      "step": 78285
    },
    {
      "epoch": 0.7964070638020834,
      "grad_norm": 13.68885326385498,
      "learning_rate": 4.951449313827694e-07,
      "loss": 3.272,
      "step": 78290
    },
    {
      "epoch": 0.7964579264322916,
      "grad_norm": 10.359649658203125,
      "learning_rate": 4.949060674105038e-07,
      "loss": 3.1571,
      "step": 78295
    },
    {
      "epoch": 0.7965087890625,
      "grad_norm": 7.889382362365723,
      "learning_rate": 4.946672547380913e-07,
      "loss": 3.0599,
      "step": 78300
    },
    {
      "epoch": 0.7965596516927084,
      "grad_norm": 24.841642379760742,
      "learning_rate": 4.944284933716409e-07,
      "loss": 3.5161,
      "step": 78305
    },
    {
      "epoch": 0.7966105143229166,
      "grad_norm": 12.071474075317383,
      "learning_rate": 4.941897833172629e-07,
      "loss": 3.3784,
      "step": 78310
    },
    {
      "epoch": 0.796661376953125,
      "grad_norm": 12.433374404907227,
      "learning_rate": 4.939511245810632e-07,
      "loss": 3.0686,
      "step": 78315
    },
    {
      "epoch": 0.7967122395833334,
      "grad_norm": 13.961923599243164,
      "learning_rate": 4.93712517169149e-07,
      "loss": 3.3446,
      "step": 78320
    },
    {
      "epoch": 0.7967631022135416,
      "grad_norm": 12.11087417602539,
      "learning_rate": 4.934739610876241e-07,
      "loss": 3.0639,
      "step": 78325
    },
    {
      "epoch": 0.79681396484375,
      "grad_norm": 10.080228805541992,
      "learning_rate": 4.932354563425928e-07,
      "loss": 3.3742,
      "step": 78330
    },
    {
      "epoch": 0.7968648274739584,
      "grad_norm": 9.731584548950195,
      "learning_rate": 4.929970029401568e-07,
      "loss": 3.0383,
      "step": 78335
    },
    {
      "epoch": 0.7969156901041666,
      "grad_norm": 15.536563873291016,
      "learning_rate": 4.92758600886416e-07,
      "loss": 3.1817,
      "step": 78340
    },
    {
      "epoch": 0.796966552734375,
      "grad_norm": 10.678982734680176,
      "learning_rate": 4.925202501874707e-07,
      "loss": 3.4005,
      "step": 78345
    },
    {
      "epoch": 0.7970174153645834,
      "grad_norm": 14.26845932006836,
      "learning_rate": 4.922819508494198e-07,
      "loss": 3.3219,
      "step": 78350
    },
    {
      "epoch": 0.7970682779947916,
      "grad_norm": 8.872315406799316,
      "learning_rate": 4.920437028783589e-07,
      "loss": 3.5714,
      "step": 78355
    },
    {
      "epoch": 0.797119140625,
      "grad_norm": 10.538403511047363,
      "learning_rate": 4.918055062803842e-07,
      "loss": 3.1552,
      "step": 78360
    },
    {
      "epoch": 0.7971700032552084,
      "grad_norm": 11.638603210449219,
      "learning_rate": 4.91567361061589e-07,
      "loss": 3.2873,
      "step": 78365
    },
    {
      "epoch": 0.7972208658854166,
      "grad_norm": 9.17963695526123,
      "learning_rate": 4.913292672280667e-07,
      "loss": 3.3738,
      "step": 78370
    },
    {
      "epoch": 0.797271728515625,
      "grad_norm": 14.535566329956055,
      "learning_rate": 4.910912247859099e-07,
      "loss": 3.2076,
      "step": 78375
    },
    {
      "epoch": 0.7973225911458334,
      "grad_norm": 11.816314697265625,
      "learning_rate": 4.908532337412075e-07,
      "loss": 3.362,
      "step": 78380
    },
    {
      "epoch": 0.7973734537760416,
      "grad_norm": 12.55820083618164,
      "learning_rate": 4.906152941000486e-07,
      "loss": 3.3663,
      "step": 78385
    },
    {
      "epoch": 0.79742431640625,
      "grad_norm": 13.613922119140625,
      "learning_rate": 4.903774058685215e-07,
      "loss": 3.0585,
      "step": 78390
    },
    {
      "epoch": 0.7974751790364584,
      "grad_norm": 12.756303787231445,
      "learning_rate": 4.901395690527116e-07,
      "loss": 3.701,
      "step": 78395
    },
    {
      "epoch": 0.7975260416666666,
      "grad_norm": 13.124650955200195,
      "learning_rate": 4.899017836587047e-07,
      "loss": 3.4298,
      "step": 78400
    },
    {
      "epoch": 0.797576904296875,
      "grad_norm": 13.512313842773438,
      "learning_rate": 4.896640496925845e-07,
      "loss": 3.626,
      "step": 78405
    },
    {
      "epoch": 0.7976277669270834,
      "grad_norm": 9.460736274719238,
      "learning_rate": 4.894263671604332e-07,
      "loss": 3.1125,
      "step": 78410
    },
    {
      "epoch": 0.7976786295572916,
      "grad_norm": 9.055636405944824,
      "learning_rate": 4.891887360683311e-07,
      "loss": 3.2618,
      "step": 78415
    },
    {
      "epoch": 0.7977294921875,
      "grad_norm": 9.254910469055176,
      "learning_rate": 4.889511564223586e-07,
      "loss": 3.7696,
      "step": 78420
    },
    {
      "epoch": 0.7977803548177084,
      "grad_norm": 11.433979034423828,
      "learning_rate": 4.887136282285948e-07,
      "loss": 3.3128,
      "step": 78425
    },
    {
      "epoch": 0.7978312174479166,
      "grad_norm": 8.865793228149414,
      "learning_rate": 4.884761514931161e-07,
      "loss": 3.1384,
      "step": 78430
    },
    {
      "epoch": 0.797882080078125,
      "grad_norm": 8.5194730758667,
      "learning_rate": 4.882387262219981e-07,
      "loss": 3.3785,
      "step": 78435
    },
    {
      "epoch": 0.7979329427083334,
      "grad_norm": 13.52230167388916,
      "learning_rate": 4.880013524213151e-07,
      "loss": 3.1476,
      "step": 78440
    },
    {
      "epoch": 0.7979838053385416,
      "grad_norm": 9.095830917358398,
      "learning_rate": 4.877640300971404e-07,
      "loss": 3.5677,
      "step": 78445
    },
    {
      "epoch": 0.79803466796875,
      "grad_norm": 13.468499183654785,
      "learning_rate": 4.875267592555466e-07,
      "loss": 3.3346,
      "step": 78450
    },
    {
      "epoch": 0.7980855305989584,
      "grad_norm": 14.660035133361816,
      "learning_rate": 4.87289539902604e-07,
      "loss": 3.6494,
      "step": 78455
    },
    {
      "epoch": 0.7981363932291666,
      "grad_norm": 10.19507122039795,
      "learning_rate": 4.870523720443804e-07,
      "loss": 3.3321,
      "step": 78460
    },
    {
      "epoch": 0.798187255859375,
      "grad_norm": 8.802572250366211,
      "learning_rate": 4.868152556869457e-07,
      "loss": 3.1674,
      "step": 78465
    },
    {
      "epoch": 0.7982381184895834,
      "grad_norm": 9.512373924255371,
      "learning_rate": 4.865781908363646e-07,
      "loss": 3.1545,
      "step": 78470
    },
    {
      "epoch": 0.7982889811197916,
      "grad_norm": 15.197171211242676,
      "learning_rate": 4.86341177498704e-07,
      "loss": 3.3482,
      "step": 78475
    },
    {
      "epoch": 0.79833984375,
      "grad_norm": 13.10971736907959,
      "learning_rate": 4.861042156800264e-07,
      "loss": 3.3086,
      "step": 78480
    },
    {
      "epoch": 0.7983907063802084,
      "grad_norm": 16.294416427612305,
      "learning_rate": 4.858673053863957e-07,
      "loss": 3.6134,
      "step": 78485
    },
    {
      "epoch": 0.7984415690104166,
      "grad_norm": 11.440515518188477,
      "learning_rate": 4.856304466238729e-07,
      "loss": 3.3362,
      "step": 78490
    },
    {
      "epoch": 0.798492431640625,
      "grad_norm": 13.755946159362793,
      "learning_rate": 4.853936393985168e-07,
      "loss": 3.273,
      "step": 78495
    },
    {
      "epoch": 0.7985432942708334,
      "grad_norm": 15.543915748596191,
      "learning_rate": 4.85156883716387e-07,
      "loss": 2.8206,
      "step": 78500
    },
    {
      "epoch": 0.7985941569010416,
      "grad_norm": 12.707036972045898,
      "learning_rate": 4.849201795835412e-07,
      "loss": 3.0078,
      "step": 78505
    },
    {
      "epoch": 0.79864501953125,
      "grad_norm": 14.186240196228027,
      "learning_rate": 4.846835270060352e-07,
      "loss": 3.2222,
      "step": 78510
    },
    {
      "epoch": 0.7986958821614584,
      "grad_norm": 10.725924491882324,
      "learning_rate": 4.844469259899226e-07,
      "loss": 3.4706,
      "step": 78515
    },
    {
      "epoch": 0.7987467447916666,
      "grad_norm": 17.00894546508789,
      "learning_rate": 4.842103765412587e-07,
      "loss": 3.7273,
      "step": 78520
    },
    {
      "epoch": 0.798797607421875,
      "grad_norm": 13.25726318359375,
      "learning_rate": 4.839738786660935e-07,
      "loss": 3.49,
      "step": 78525
    },
    {
      "epoch": 0.7988484700520834,
      "grad_norm": 8.638480186462402,
      "learning_rate": 4.837374323704797e-07,
      "loss": 3.297,
      "step": 78530
    },
    {
      "epoch": 0.7988993326822916,
      "grad_norm": 11.479582786560059,
      "learning_rate": 4.835010376604649e-07,
      "loss": 3.3887,
      "step": 78535
    },
    {
      "epoch": 0.7989501953125,
      "grad_norm": 11.228867530822754,
      "learning_rate": 4.832646945420988e-07,
      "loss": 3.4827,
      "step": 78540
    },
    {
      "epoch": 0.7990010579427084,
      "grad_norm": 10.890009880065918,
      "learning_rate": 4.830284030214275e-07,
      "loss": 3.2497,
      "step": 78545
    },
    {
      "epoch": 0.7990519205729166,
      "grad_norm": 14.480691909790039,
      "learning_rate": 4.827921631044957e-07,
      "loss": 3.2666,
      "step": 78550
    },
    {
      "epoch": 0.799102783203125,
      "grad_norm": 13.990694046020508,
      "learning_rate": 4.825559747973482e-07,
      "loss": 3.3072,
      "step": 78555
    },
    {
      "epoch": 0.7991536458333334,
      "grad_norm": 12.1658296585083,
      "learning_rate": 4.823198381060287e-07,
      "loss": 3.1358,
      "step": 78560
    },
    {
      "epoch": 0.7992045084635416,
      "grad_norm": 12.146638870239258,
      "learning_rate": 4.82083753036578e-07,
      "loss": 3.6579,
      "step": 78565
    },
    {
      "epoch": 0.79925537109375,
      "grad_norm": 14.979445457458496,
      "learning_rate": 4.818477195950355e-07,
      "loss": 3.78,
      "step": 78570
    },
    {
      "epoch": 0.7993062337239584,
      "grad_norm": 15.680461883544922,
      "learning_rate": 4.816117377874408e-07,
      "loss": 3.2552,
      "step": 78575
    },
    {
      "epoch": 0.7993570963541666,
      "grad_norm": 7.587637424468994,
      "learning_rate": 4.813758076198316e-07,
      "loss": 3.3191,
      "step": 78580
    },
    {
      "epoch": 0.799407958984375,
      "grad_norm": 13.030993461608887,
      "learning_rate": 4.811399290982443e-07,
      "loss": 3.2942,
      "step": 78585
    },
    {
      "epoch": 0.7994588216145834,
      "grad_norm": 15.706942558288574,
      "learning_rate": 4.809041022287125e-07,
      "loss": 3.0769,
      "step": 78590
    },
    {
      "epoch": 0.7995096842447916,
      "grad_norm": 13.303389549255371,
      "learning_rate": 4.806683270172715e-07,
      "loss": 3.2164,
      "step": 78595
    },
    {
      "epoch": 0.799560546875,
      "grad_norm": 11.963903427124023,
      "learning_rate": 4.804326034699522e-07,
      "loss": 3.1576,
      "step": 78600
    },
    {
      "epoch": 0.7996114095052084,
      "grad_norm": 7.553828716278076,
      "learning_rate": 4.801969315927865e-07,
      "loss": 3.2115,
      "step": 78605
    },
    {
      "epoch": 0.7996622721354166,
      "grad_norm": 14.163137435913086,
      "learning_rate": 4.799613113918028e-07,
      "loss": 2.9159,
      "step": 78610
    },
    {
      "epoch": 0.799713134765625,
      "grad_norm": 12.92434310913086,
      "learning_rate": 4.79725742873031e-07,
      "loss": 2.9415,
      "step": 78615
    },
    {
      "epoch": 0.7997639973958334,
      "grad_norm": 13.198997497558594,
      "learning_rate": 4.794902260424972e-07,
      "loss": 3.4047,
      "step": 78620
    },
    {
      "epoch": 0.7998148600260416,
      "grad_norm": 10.787629127502441,
      "learning_rate": 4.792547609062262e-07,
      "loss": 3.234,
      "step": 78625
    },
    {
      "epoch": 0.79986572265625,
      "grad_norm": 15.940741539001465,
      "learning_rate": 4.790193474702434e-07,
      "loss": 3.1683,
      "step": 78630
    },
    {
      "epoch": 0.7999165852864584,
      "grad_norm": 11.246376991271973,
      "learning_rate": 4.787839857405721e-07,
      "loss": 3.4445,
      "step": 78635
    },
    {
      "epoch": 0.7999674479166666,
      "grad_norm": 13.741132736206055,
      "learning_rate": 4.785486757232336e-07,
      "loss": 3.4073,
      "step": 78640
    },
    {
      "epoch": 0.800018310546875,
      "grad_norm": 7.921855449676514,
      "learning_rate": 4.783134174242479e-07,
      "loss": 3.7101,
      "step": 78645
    },
    {
      "epoch": 0.8000691731770834,
      "grad_norm": 16.312631607055664,
      "learning_rate": 4.780782108496337e-07,
      "loss": 3.6811,
      "step": 78650
    },
    {
      "epoch": 0.8001200358072916,
      "grad_norm": 12.837047576904297,
      "learning_rate": 4.77843056005409e-07,
      "loss": 3.3388,
      "step": 78655
    },
    {
      "epoch": 0.8001708984375,
      "grad_norm": 7.334871292114258,
      "learning_rate": 4.776079528975913e-07,
      "loss": 3.1737,
      "step": 78660
    },
    {
      "epoch": 0.8002217610677084,
      "grad_norm": 11.887138366699219,
      "learning_rate": 4.773729015321949e-07,
      "loss": 3.0676,
      "step": 78665
    },
    {
      "epoch": 0.8002726236979166,
      "grad_norm": 11.066459655761719,
      "learning_rate": 4.771379019152326e-07,
      "loss": 3.1375,
      "step": 78670
    },
    {
      "epoch": 0.800323486328125,
      "grad_norm": 9.975939750671387,
      "learning_rate": 4.769029540527181e-07,
      "loss": 3.3405,
      "step": 78675
    },
    {
      "epoch": 0.8003743489583334,
      "grad_norm": 15.235580444335938,
      "learning_rate": 4.766680579506613e-07,
      "loss": 3.0982,
      "step": 78680
    },
    {
      "epoch": 0.8004252115885416,
      "grad_norm": 9.530317306518555,
      "learning_rate": 4.7643321361507286e-07,
      "loss": 3.6487,
      "step": 78685
    },
    {
      "epoch": 0.80047607421875,
      "grad_norm": 6.792818546295166,
      "learning_rate": 4.761984210519616e-07,
      "loss": 3.8031,
      "step": 78690
    },
    {
      "epoch": 0.8005269368489584,
      "grad_norm": 12.819168090820312,
      "learning_rate": 4.7596368026733396e-07,
      "loss": 3.2232,
      "step": 78695
    },
    {
      "epoch": 0.8005777994791666,
      "grad_norm": 14.127707481384277,
      "learning_rate": 4.7572899126719615e-07,
      "loss": 4.0349,
      "step": 78700
    },
    {
      "epoch": 0.800628662109375,
      "grad_norm": 16.820049285888672,
      "learning_rate": 4.7549435405755075e-07,
      "loss": 3.4131,
      "step": 78705
    },
    {
      "epoch": 0.8006795247395834,
      "grad_norm": 9.99593734741211,
      "learning_rate": 4.7525976864440354e-07,
      "loss": 3.117,
      "step": 78710
    },
    {
      "epoch": 0.8007303873697916,
      "grad_norm": 13.380139350891113,
      "learning_rate": 4.750252350337556e-07,
      "loss": 3.4492,
      "step": 78715
    },
    {
      "epoch": 0.80078125,
      "grad_norm": 13.98039722442627,
      "learning_rate": 4.7479075323160677e-07,
      "loss": 3.0948,
      "step": 78720
    },
    {
      "epoch": 0.8008321126302084,
      "grad_norm": 14.347100257873535,
      "learning_rate": 4.7455632324395596e-07,
      "loss": 3.5968,
      "step": 78725
    },
    {
      "epoch": 0.8008829752604166,
      "grad_norm": 14.859460830688477,
      "learning_rate": 4.7432194507680163e-07,
      "loss": 3.1246,
      "step": 78730
    },
    {
      "epoch": 0.800933837890625,
      "grad_norm": 18.239768981933594,
      "learning_rate": 4.740876187361407e-07,
      "loss": 3.5149,
      "step": 78735
    },
    {
      "epoch": 0.8009847005208334,
      "grad_norm": 10.691304206848145,
      "learning_rate": 4.738533442279677e-07,
      "loss": 3.1573,
      "step": 78740
    },
    {
      "epoch": 0.8010355631510416,
      "grad_norm": 14.137451171875,
      "learning_rate": 4.7361912155827636e-07,
      "loss": 3.9191,
      "step": 78745
    },
    {
      "epoch": 0.80108642578125,
      "grad_norm": 15.178362846374512,
      "learning_rate": 4.733849507330598e-07,
      "loss": 3.3544,
      "step": 78750
    },
    {
      "epoch": 0.8011372884114584,
      "grad_norm": 8.599349975585938,
      "learning_rate": 4.731508317583086e-07,
      "loss": 3.3475,
      "step": 78755
    },
    {
      "epoch": 0.8011881510416666,
      "grad_norm": 16.44230842590332,
      "learning_rate": 4.729167646400135e-07,
      "loss": 3.3797,
      "step": 78760
    },
    {
      "epoch": 0.801239013671875,
      "grad_norm": 14.500465393066406,
      "learning_rate": 4.7268274938416183e-07,
      "loss": 3.4236,
      "step": 78765
    },
    {
      "epoch": 0.8012898763020834,
      "grad_norm": 8.624430656433105,
      "learning_rate": 4.7244878599674196e-07,
      "loss": 3.342,
      "step": 78770
    },
    {
      "epoch": 0.8013407389322916,
      "grad_norm": 10.258706092834473,
      "learning_rate": 4.722148744837393e-07,
      "loss": 3.2887,
      "step": 78775
    },
    {
      "epoch": 0.8013916015625,
      "grad_norm": 7.703892707824707,
      "learning_rate": 4.719810148511381e-07,
      "loss": 3.1236,
      "step": 78780
    },
    {
      "epoch": 0.8014424641927084,
      "grad_norm": 12.779461860656738,
      "learning_rate": 4.717472071049217e-07,
      "loss": 3.0516,
      "step": 78785
    },
    {
      "epoch": 0.8014933268229166,
      "grad_norm": 12.617457389831543,
      "learning_rate": 4.71513451251073e-07,
      "loss": 3.6554,
      "step": 78790
    },
    {
      "epoch": 0.801544189453125,
      "grad_norm": 12.944184303283691,
      "learning_rate": 4.7127974729557183e-07,
      "loss": 3.4866,
      "step": 78795
    },
    {
      "epoch": 0.8015950520833334,
      "grad_norm": 8.754607200622559,
      "learning_rate": 4.710460952443968e-07,
      "loss": 3.1968,
      "step": 78800
    },
    {
      "epoch": 0.8016459147135416,
      "grad_norm": 10.901610374450684,
      "learning_rate": 4.708124951035273e-07,
      "loss": 3.5131,
      "step": 78805
    },
    {
      "epoch": 0.80169677734375,
      "grad_norm": 14.882295608520508,
      "learning_rate": 4.7057894687893847e-07,
      "loss": 3.4044,
      "step": 78810
    },
    {
      "epoch": 0.8017476399739584,
      "grad_norm": 10.60777759552002,
      "learning_rate": 4.703454505766067e-07,
      "loss": 3.1125,
      "step": 78815
    },
    {
      "epoch": 0.8017985026041666,
      "grad_norm": 9.532958984375,
      "learning_rate": 4.70112006202505e-07,
      "loss": 3.3332,
      "step": 78820
    },
    {
      "epoch": 0.801849365234375,
      "grad_norm": 10.766741752624512,
      "learning_rate": 4.6987861376260713e-07,
      "loss": 3.0745,
      "step": 78825
    },
    {
      "epoch": 0.8019002278645834,
      "grad_norm": 12.011417388916016,
      "learning_rate": 4.696452732628837e-07,
      "loss": 3.0788,
      "step": 78830
    },
    {
      "epoch": 0.8019510904947916,
      "grad_norm": 11.187000274658203,
      "learning_rate": 4.69411984709304e-07,
      "loss": 3.2608,
      "step": 78835
    },
    {
      "epoch": 0.802001953125,
      "grad_norm": 16.041290283203125,
      "learning_rate": 4.691787481078375e-07,
      "loss": 3.0582,
      "step": 78840
    },
    {
      "epoch": 0.8020528157552084,
      "grad_norm": 7.748333930969238,
      "learning_rate": 4.6894556346445185e-07,
      "loss": 3.5502,
      "step": 78845
    },
    {
      "epoch": 0.8021036783854166,
      "grad_norm": 15.040388107299805,
      "learning_rate": 4.6871243078511266e-07,
      "loss": 3.8123,
      "step": 78850
    },
    {
      "epoch": 0.802154541015625,
      "grad_norm": 14.229958534240723,
      "learning_rate": 4.684793500757845e-07,
      "loss": 3.4445,
      "step": 78855
    },
    {
      "epoch": 0.8022054036458334,
      "grad_norm": 10.111949920654297,
      "learning_rate": 4.682463213424293e-07,
      "loss": 3.1247,
      "step": 78860
    },
    {
      "epoch": 0.8022562662760416,
      "grad_norm": 13.16586685180664,
      "learning_rate": 4.6801334459101176e-07,
      "loss": 2.9999,
      "step": 78865
    },
    {
      "epoch": 0.80230712890625,
      "grad_norm": 10.070747375488281,
      "learning_rate": 4.67780419827491e-07,
      "loss": 3.4165,
      "step": 78870
    },
    {
      "epoch": 0.8023579915364584,
      "grad_norm": 16.717920303344727,
      "learning_rate": 4.675475470578267e-07,
      "loss": 3.3036,
      "step": 78875
    },
    {
      "epoch": 0.8024088541666666,
      "grad_norm": 15.924241065979004,
      "learning_rate": 4.6731472628797597e-07,
      "loss": 3.1925,
      "step": 78880
    },
    {
      "epoch": 0.802459716796875,
      "grad_norm": 12.359611511230469,
      "learning_rate": 4.6708195752389605e-07,
      "loss": 3.4127,
      "step": 78885
    },
    {
      "epoch": 0.8025105794270834,
      "grad_norm": 15.967865943908691,
      "learning_rate": 4.66849240771543e-07,
      "loss": 3.1058,
      "step": 78890
    },
    {
      "epoch": 0.8025614420572916,
      "grad_norm": 11.010781288146973,
      "learning_rate": 4.666165760368696e-07,
      "loss": 3.3866,
      "step": 78895
    },
    {
      "epoch": 0.8026123046875,
      "grad_norm": 14.635432243347168,
      "learning_rate": 4.663839633258299e-07,
      "loss": 3.3891,
      "step": 78900
    },
    {
      "epoch": 0.8026631673177084,
      "grad_norm": 15.67871379852295,
      "learning_rate": 4.6615140264437447e-07,
      "loss": 3.2746,
      "step": 78905
    },
    {
      "epoch": 0.8027140299479166,
      "grad_norm": 9.079788208007812,
      "learning_rate": 4.6591889399845267e-07,
      "loss": 3.31,
      "step": 78910
    },
    {
      "epoch": 0.802764892578125,
      "grad_norm": 11.785703659057617,
      "learning_rate": 4.65686437394014e-07,
      "loss": 3.3784,
      "step": 78915
    },
    {
      "epoch": 0.8028157552083334,
      "grad_norm": 13.959105491638184,
      "learning_rate": 4.654540328370061e-07,
      "loss": 3.1514,
      "step": 78920
    },
    {
      "epoch": 0.8028666178385416,
      "grad_norm": 10.363253593444824,
      "learning_rate": 4.652216803333745e-07,
      "loss": 3.2574,
      "step": 78925
    },
    {
      "epoch": 0.80291748046875,
      "grad_norm": 11.194170951843262,
      "learning_rate": 4.6498937988906395e-07,
      "loss": 3.0131,
      "step": 78930
    },
    {
      "epoch": 0.8029683430989584,
      "grad_norm": 9.353020668029785,
      "learning_rate": 4.6475713151001686e-07,
      "loss": 3.3285,
      "step": 78935
    },
    {
      "epoch": 0.8030192057291666,
      "grad_norm": 13.347886085510254,
      "learning_rate": 4.6452493520217664e-07,
      "loss": 3.42,
      "step": 78940
    },
    {
      "epoch": 0.803070068359375,
      "grad_norm": 14.261474609375,
      "learning_rate": 4.6429279097148394e-07,
      "loss": 3.1498,
      "step": 78945
    },
    {
      "epoch": 0.8031209309895834,
      "grad_norm": 13.230145454406738,
      "learning_rate": 4.6406069882387743e-07,
      "loss": 3.3409,
      "step": 78950
    },
    {
      "epoch": 0.8031717936197916,
      "grad_norm": 8.211542129516602,
      "learning_rate": 4.6382865876529516e-07,
      "loss": 3.0068,
      "step": 78955
    },
    {
      "epoch": 0.80322265625,
      "grad_norm": 11.118182182312012,
      "learning_rate": 4.6359667080167427e-07,
      "loss": 3.0683,
      "step": 78960
    },
    {
      "epoch": 0.8032735188802084,
      "grad_norm": 13.080451965332031,
      "learning_rate": 4.633647349389495e-07,
      "loss": 3.6222,
      "step": 78965
    },
    {
      "epoch": 0.8033243815104166,
      "grad_norm": 16.388444900512695,
      "learning_rate": 4.6313285118305574e-07,
      "loss": 3.1887,
      "step": 78970
    },
    {
      "epoch": 0.803375244140625,
      "grad_norm": 17.810380935668945,
      "learning_rate": 4.6290101953992473e-07,
      "loss": 3.3287,
      "step": 78975
    },
    {
      "epoch": 0.8034261067708334,
      "grad_norm": 8.733531951904297,
      "learning_rate": 4.6266924001548863e-07,
      "loss": 3.4616,
      "step": 78980
    },
    {
      "epoch": 0.8034769694010416,
      "grad_norm": 9.307930946350098,
      "learning_rate": 4.624375126156774e-07,
      "loss": 3.1054,
      "step": 78985
    },
    {
      "epoch": 0.80352783203125,
      "grad_norm": 11.856342315673828,
      "learning_rate": 4.6220583734641863e-07,
      "loss": 3.2163,
      "step": 78990
    },
    {
      "epoch": 0.8035786946614584,
      "grad_norm": 9.549277305603027,
      "learning_rate": 4.6197421421364075e-07,
      "loss": 2.9922,
      "step": 78995
    },
    {
      "epoch": 0.8036295572916666,
      "grad_norm": 13.3350830078125,
      "learning_rate": 4.6174264322326995e-07,
      "loss": 2.7833,
      "step": 79000
    },
    {
      "epoch": 0.803680419921875,
      "grad_norm": 16.208858489990234,
      "learning_rate": 4.6151112438123064e-07,
      "loss": 3.294,
      "step": 79005
    },
    {
      "epoch": 0.8037312825520834,
      "grad_norm": 12.73587703704834,
      "learning_rate": 4.612796576934453e-07,
      "loss": 3.2964,
      "step": 79010
    },
    {
      "epoch": 0.8037821451822916,
      "grad_norm": 9.596652030944824,
      "learning_rate": 4.6104824316583666e-07,
      "loss": 3.4701,
      "step": 79015
    },
    {
      "epoch": 0.8038330078125,
      "grad_norm": 14.039467811584473,
      "learning_rate": 4.6081688080432616e-07,
      "loss": 3.1576,
      "step": 79020
    },
    {
      "epoch": 0.8038838704427084,
      "grad_norm": 15.404362678527832,
      "learning_rate": 4.605855706148324e-07,
      "loss": 3.3063,
      "step": 79025
    },
    {
      "epoch": 0.8039347330729166,
      "grad_norm": 10.67270565032959,
      "learning_rate": 4.603543126032728e-07,
      "loss": 3.6506,
      "step": 79030
    },
    {
      "epoch": 0.803985595703125,
      "grad_norm": 7.46541166305542,
      "learning_rate": 4.6012310677556545e-07,
      "loss": 3.2762,
      "step": 79035
    },
    {
      "epoch": 0.8040364583333334,
      "grad_norm": 10.64223575592041,
      "learning_rate": 4.598919531376239e-07,
      "loss": 3.0405,
      "step": 79040
    },
    {
      "epoch": 0.8040873209635416,
      "grad_norm": 14.383641242980957,
      "learning_rate": 4.5966085169536407e-07,
      "loss": 3.1108,
      "step": 79045
    },
    {
      "epoch": 0.80413818359375,
      "grad_norm": 12.580687522888184,
      "learning_rate": 4.59429802454697e-07,
      "loss": 3.1669,
      "step": 79050
    },
    {
      "epoch": 0.8041890462239584,
      "grad_norm": 11.952764511108398,
      "learning_rate": 4.5919880542153546e-07,
      "loss": 3.3763,
      "step": 79055
    },
    {
      "epoch": 0.8042399088541666,
      "grad_norm": 12.009092330932617,
      "learning_rate": 4.589678606017889e-07,
      "loss": 3.3642,
      "step": 79060
    },
    {
      "epoch": 0.804290771484375,
      "grad_norm": 15.879464149475098,
      "learning_rate": 4.587369680013651e-07,
      "loss": 3.4183,
      "step": 79065
    },
    {
      "epoch": 0.8043416341145834,
      "grad_norm": 10.191134452819824,
      "learning_rate": 4.585061276261721e-07,
      "loss": 3.0245,
      "step": 79070
    },
    {
      "epoch": 0.8043924967447916,
      "grad_norm": 16.796077728271484,
      "learning_rate": 4.5827533948211657e-07,
      "loss": 3.6091,
      "step": 79075
    },
    {
      "epoch": 0.804443359375,
      "grad_norm": 15.804084777832031,
      "learning_rate": 4.580446035751024e-07,
      "loss": 3.3889,
      "step": 79080
    },
    {
      "epoch": 0.8044942220052084,
      "grad_norm": 13.486974716186523,
      "learning_rate": 4.578139199110326e-07,
      "loss": 3.3648,
      "step": 79085
    },
    {
      "epoch": 0.8045450846354166,
      "grad_norm": 12.026664733886719,
      "learning_rate": 4.5758328849581006e-07,
      "loss": 3.1254,
      "step": 79090
    },
    {
      "epoch": 0.804595947265625,
      "grad_norm": 12.640941619873047,
      "learning_rate": 4.573527093353347e-07,
      "loss": 3.3413,
      "step": 79095
    },
    {
      "epoch": 0.8046468098958334,
      "grad_norm": 7.197507381439209,
      "learning_rate": 4.571221824355065e-07,
      "loss": 3.3461,
      "step": 79100
    },
    {
      "epoch": 0.8046976725260416,
      "grad_norm": 15.275164604187012,
      "learning_rate": 4.5689170780222227e-07,
      "loss": 3.4683,
      "step": 79105
    },
    {
      "epoch": 0.80474853515625,
      "grad_norm": 11.376517295837402,
      "learning_rate": 4.5666128544138047e-07,
      "loss": 3.2734,
      "step": 79110
    },
    {
      "epoch": 0.8047993977864584,
      "grad_norm": 11.782517433166504,
      "learning_rate": 4.56430915358875e-07,
      "loss": 3.1015,
      "step": 79115
    },
    {
      "epoch": 0.8048502604166666,
      "grad_norm": 13.285874366760254,
      "learning_rate": 4.562005975605996e-07,
      "loss": 3.0926,
      "step": 79120
    },
    {
      "epoch": 0.804901123046875,
      "grad_norm": 12.198227882385254,
      "learning_rate": 4.559703320524475e-07,
      "loss": 3.2659,
      "step": 79125
    },
    {
      "epoch": 0.8049519856770834,
      "grad_norm": 8.998311042785645,
      "learning_rate": 4.557401188403107e-07,
      "loss": 3.3936,
      "step": 79130
    },
    {
      "epoch": 0.8050028483072916,
      "grad_norm": 13.996612548828125,
      "learning_rate": 4.555099579300784e-07,
      "loss": 3.2987,
      "step": 79135
    },
    {
      "epoch": 0.8050537109375,
      "grad_norm": 9.745990753173828,
      "learning_rate": 4.552798493276392e-07,
      "loss": 3.1222,
      "step": 79140
    },
    {
      "epoch": 0.8051045735677084,
      "grad_norm": 10.147890090942383,
      "learning_rate": 4.5504979303887974e-07,
      "loss": 3.3142,
      "step": 79145
    },
    {
      "epoch": 0.8051554361979166,
      "grad_norm": 10.35171127319336,
      "learning_rate": 4.5481978906968675e-07,
      "loss": 3.2781,
      "step": 79150
    },
    {
      "epoch": 0.805206298828125,
      "grad_norm": 14.873088836669922,
      "learning_rate": 4.545898374259453e-07,
      "loss": 3.2424,
      "step": 79155
    },
    {
      "epoch": 0.8052571614583334,
      "grad_norm": 13.0719575881958,
      "learning_rate": 4.543599381135377e-07,
      "loss": 3.7585,
      "step": 79160
    },
    {
      "epoch": 0.8053080240885416,
      "grad_norm": 14.930803298950195,
      "learning_rate": 4.541300911383459e-07,
      "loss": 3.1064,
      "step": 79165
    },
    {
      "epoch": 0.80535888671875,
      "grad_norm": 10.585026741027832,
      "learning_rate": 4.5390029650625086e-07,
      "loss": 3.4128,
      "step": 79170
    },
    {
      "epoch": 0.8054097493489584,
      "grad_norm": 15.125510215759277,
      "learning_rate": 4.5367055422313205e-07,
      "loss": 3.1273,
      "step": 79175
    },
    {
      "epoch": 0.8054606119791666,
      "grad_norm": 12.9608793258667,
      "learning_rate": 4.5344086429486675e-07,
      "loss": 3.4222,
      "step": 79180
    },
    {
      "epoch": 0.805511474609375,
      "grad_norm": 12.554879188537598,
      "learning_rate": 4.5321122672733255e-07,
      "loss": 3.5969,
      "step": 79185
    },
    {
      "epoch": 0.8055623372395834,
      "grad_norm": 15.822430610656738,
      "learning_rate": 4.5298164152640376e-07,
      "loss": 3.5563,
      "step": 79190
    },
    {
      "epoch": 0.8056131998697916,
      "grad_norm": 13.774492263793945,
      "learning_rate": 4.527521086979539e-07,
      "loss": 3.1992,
      "step": 79195
    },
    {
      "epoch": 0.8056640625,
      "grad_norm": 12.54838752746582,
      "learning_rate": 4.525226282478559e-07,
      "loss": 3.1414,
      "step": 79200
    },
    {
      "epoch": 0.8057149251302084,
      "grad_norm": 13.793286323547363,
      "learning_rate": 4.522932001819821e-07,
      "loss": 2.9914,
      "step": 79205
    },
    {
      "epoch": 0.8057657877604166,
      "grad_norm": 12.090859413146973,
      "learning_rate": 4.5206382450620117e-07,
      "loss": 3.3911,
      "step": 79210
    },
    {
      "epoch": 0.805816650390625,
      "grad_norm": 16.567838668823242,
      "learning_rate": 4.518345012263822e-07,
      "loss": 3.4067,
      "step": 79215
    },
    {
      "epoch": 0.8058675130208334,
      "grad_norm": 10.558396339416504,
      "learning_rate": 4.516052303483912e-07,
      "loss": 3.3806,
      "step": 79220
    },
    {
      "epoch": 0.8059183756510416,
      "grad_norm": 11.566436767578125,
      "learning_rate": 4.5137601187809464e-07,
      "loss": 3.1802,
      "step": 79225
    },
    {
      "epoch": 0.80596923828125,
      "grad_norm": 14.10775375366211,
      "learning_rate": 4.511468458213583e-07,
      "loss": 3.9436,
      "step": 79230
    },
    {
      "epoch": 0.8060201009114584,
      "grad_norm": 9.806373596191406,
      "learning_rate": 4.509177321840441e-07,
      "loss": 3.2196,
      "step": 79235
    },
    {
      "epoch": 0.8060709635416666,
      "grad_norm": 14.403755187988281,
      "learning_rate": 4.506886709720132e-07,
      "loss": 3.3581,
      "step": 79240
    },
    {
      "epoch": 0.806121826171875,
      "grad_norm": 13.174206733703613,
      "learning_rate": 4.504596621911278e-07,
      "loss": 3.0128,
      "step": 79245
    },
    {
      "epoch": 0.8061726888020834,
      "grad_norm": 9.067151069641113,
      "learning_rate": 4.502307058472455e-07,
      "loss": 3.4482,
      "step": 79250
    },
    {
      "epoch": 0.8062235514322916,
      "grad_norm": 10.918560981750488,
      "learning_rate": 4.5000180194622517e-07,
      "loss": 3.2611,
      "step": 79255
    },
    {
      "epoch": 0.8062744140625,
      "grad_norm": 15.179645538330078,
      "learning_rate": 4.497729504939224e-07,
      "loss": 3.4583,
      "step": 79260
    },
    {
      "epoch": 0.8063252766927084,
      "grad_norm": 14.72915267944336,
      "learning_rate": 4.4954415149619345e-07,
      "loss": 3.4114,
      "step": 79265
    },
    {
      "epoch": 0.8063761393229166,
      "grad_norm": 11.220216751098633,
      "learning_rate": 4.49315404958891e-07,
      "loss": 3.2721,
      "step": 79270
    },
    {
      "epoch": 0.806427001953125,
      "grad_norm": 10.747183799743652,
      "learning_rate": 4.490867108878677e-07,
      "loss": 3.2829,
      "step": 79275
    },
    {
      "epoch": 0.8064778645833334,
      "grad_norm": 11.868279457092285,
      "learning_rate": 4.4885806928897434e-07,
      "loss": 3.5199,
      "step": 79280
    },
    {
      "epoch": 0.8065287272135416,
      "grad_norm": 12.011693954467773,
      "learning_rate": 4.486294801680621e-07,
      "loss": 3.5379,
      "step": 79285
    },
    {
      "epoch": 0.80657958984375,
      "grad_norm": 11.23404312133789,
      "learning_rate": 4.48400943530978e-07,
      "loss": 3.3113,
      "step": 79290
    },
    {
      "epoch": 0.8066304524739584,
      "grad_norm": 13.91883659362793,
      "learning_rate": 4.481724593835693e-07,
      "loss": 3.2927,
      "step": 79295
    },
    {
      "epoch": 0.8066813151041666,
      "grad_norm": 9.497418403625488,
      "learning_rate": 4.479440277316821e-07,
      "loss": 3.4381,
      "step": 79300
    },
    {
      "epoch": 0.806732177734375,
      "grad_norm": 13.011350631713867,
      "learning_rate": 4.4771564858116014e-07,
      "loss": 3.3248,
      "step": 79305
    },
    {
      "epoch": 0.8067830403645834,
      "grad_norm": 12.247965812683105,
      "learning_rate": 4.4748732193784755e-07,
      "loss": 2.9754,
      "step": 79310
    },
    {
      "epoch": 0.8068339029947916,
      "grad_norm": 11.67395305633545,
      "learning_rate": 4.4725904780758443e-07,
      "loss": 3.6511,
      "step": 79315
    },
    {
      "epoch": 0.806884765625,
      "grad_norm": 13.142789840698242,
      "learning_rate": 4.470308261962131e-07,
      "loss": 3.4572,
      "step": 79320
    },
    {
      "epoch": 0.8069356282552084,
      "grad_norm": 8.493226051330566,
      "learning_rate": 4.4680265710957044e-07,
      "loss": 3.5911,
      "step": 79325
    },
    {
      "epoch": 0.8069864908854166,
      "grad_norm": 16.766324996948242,
      "learning_rate": 4.46574540553496e-07,
      "loss": 3.3472,
      "step": 79330
    },
    {
      "epoch": 0.807037353515625,
      "grad_norm": 9.673091888427734,
      "learning_rate": 4.4634647653382435e-07,
      "loss": 3.7766,
      "step": 79335
    },
    {
      "epoch": 0.8070882161458334,
      "grad_norm": 10.032527923583984,
      "learning_rate": 4.4611846505639214e-07,
      "loss": 3.3584,
      "step": 79340
    },
    {
      "epoch": 0.8071390787760416,
      "grad_norm": 10.048256874084473,
      "learning_rate": 4.4589050612703214e-07,
      "loss": 3.5576,
      "step": 79345
    },
    {
      "epoch": 0.80718994140625,
      "grad_norm": 12.669438362121582,
      "learning_rate": 4.4566259975157624e-07,
      "loss": 3.1749,
      "step": 79350
    },
    {
      "epoch": 0.8072408040364584,
      "grad_norm": 13.59167194366455,
      "learning_rate": 4.4543474593585546e-07,
      "loss": 3.3033,
      "step": 79355
    },
    {
      "epoch": 0.8072916666666666,
      "grad_norm": 11.786405563354492,
      "learning_rate": 4.4520694468570047e-07,
      "loss": 3.2057,
      "step": 79360
    },
    {
      "epoch": 0.807342529296875,
      "grad_norm": 8.25817584991455,
      "learning_rate": 4.4497919600693884e-07,
      "loss": 3.1303,
      "step": 79365
    },
    {
      "epoch": 0.8073933919270834,
      "grad_norm": 11.9024019241333,
      "learning_rate": 4.4475149990539727e-07,
      "loss": 3.1778,
      "step": 79370
    },
    {
      "epoch": 0.8074442545572916,
      "grad_norm": 14.694074630737305,
      "learning_rate": 4.4452385638690085e-07,
      "loss": 3.5578,
      "step": 79375
    },
    {
      "epoch": 0.8074951171875,
      "grad_norm": 14.93714714050293,
      "learning_rate": 4.4429626545727437e-07,
      "loss": 3.5228,
      "step": 79380
    },
    {
      "epoch": 0.8075459798177084,
      "grad_norm": 12.39831829071045,
      "learning_rate": 4.440687271223412e-07,
      "loss": 3.2921,
      "step": 79385
    },
    {
      "epoch": 0.8075968424479166,
      "grad_norm": 11.452814102172852,
      "learning_rate": 4.43841241387922e-07,
      "loss": 3.7657,
      "step": 79390
    },
    {
      "epoch": 0.807647705078125,
      "grad_norm": 11.564638137817383,
      "learning_rate": 4.436138082598376e-07,
      "loss": 3.2557,
      "step": 79395
    },
    {
      "epoch": 0.8076985677083334,
      "grad_norm": 13.71728801727295,
      "learning_rate": 4.433864277439068e-07,
      "loss": 3.3731,
      "step": 79400
    },
    {
      "epoch": 0.8077494303385416,
      "grad_norm": 12.129942893981934,
      "learning_rate": 4.431590998459459e-07,
      "loss": 3.3616,
      "step": 79405
    },
    {
      "epoch": 0.80780029296875,
      "grad_norm": 10.588050842285156,
      "learning_rate": 4.42931824571772e-07,
      "loss": 3.3958,
      "step": 79410
    },
    {
      "epoch": 0.8078511555989584,
      "grad_norm": 132.06982421875,
      "learning_rate": 4.427046019272002e-07,
      "loss": 3.5826,
      "step": 79415
    },
    {
      "epoch": 0.8079020182291666,
      "grad_norm": 15.277337074279785,
      "learning_rate": 4.424774319180439e-07,
      "loss": 3.3452,
      "step": 79420
    },
    {
      "epoch": 0.807952880859375,
      "grad_norm": 12.683645248413086,
      "learning_rate": 4.4225031455011445e-07,
      "loss": 3.1069,
      "step": 79425
    },
    {
      "epoch": 0.8080037434895834,
      "grad_norm": 13.64078426361084,
      "learning_rate": 4.420232498292226e-07,
      "loss": 3.1959,
      "step": 79430
    },
    {
      "epoch": 0.8080546061197916,
      "grad_norm": 12.579249382019043,
      "learning_rate": 4.417962377611781e-07,
      "loss": 3.429,
      "step": 79435
    },
    {
      "epoch": 0.80810546875,
      "grad_norm": 15.0173978805542,
      "learning_rate": 4.4156927835178944e-07,
      "loss": 3.1877,
      "step": 79440
    },
    {
      "epoch": 0.8081563313802084,
      "grad_norm": 11.585958480834961,
      "learning_rate": 4.413423716068632e-07,
      "loss": 3.4828,
      "step": 79445
    },
    {
      "epoch": 0.8082071940104166,
      "grad_norm": 11.081127166748047,
      "learning_rate": 4.411155175322035e-07,
      "loss": 3.5287,
      "step": 79450
    },
    {
      "epoch": 0.808258056640625,
      "grad_norm": 11.477083206176758,
      "learning_rate": 4.408887161336162e-07,
      "loss": 3.6381,
      "step": 79455
    },
    {
      "epoch": 0.8083089192708334,
      "grad_norm": 10.804614067077637,
      "learning_rate": 4.4066196741690217e-07,
      "loss": 3.1416,
      "step": 79460
    },
    {
      "epoch": 0.8083597819010416,
      "grad_norm": 14.424979209899902,
      "learning_rate": 4.404352713878643e-07,
      "loss": 3.9914,
      "step": 79465
    },
    {
      "epoch": 0.80841064453125,
      "grad_norm": 10.901899337768555,
      "learning_rate": 4.40208628052301e-07,
      "loss": 3.081,
      "step": 79470
    },
    {
      "epoch": 0.8084615071614584,
      "grad_norm": 8.379731178283691,
      "learning_rate": 4.399820374160127e-07,
      "loss": 3.6199,
      "step": 79475
    },
    {
      "epoch": 0.8085123697916666,
      "grad_norm": 9.010814666748047,
      "learning_rate": 4.397554994847955e-07,
      "loss": 3.3494,
      "step": 79480
    },
    {
      "epoch": 0.808563232421875,
      "grad_norm": 12.562276840209961,
      "learning_rate": 4.3952901426444426e-07,
      "loss": 3.9035,
      "step": 79485
    },
    {
      "epoch": 0.8086140950520834,
      "grad_norm": 13.099066734313965,
      "learning_rate": 4.3930258176075604e-07,
      "loss": 3.8992,
      "step": 79490
    },
    {
      "epoch": 0.8086649576822916,
      "grad_norm": 13.581748962402344,
      "learning_rate": 4.390762019795228e-07,
      "loss": 3.4234,
      "step": 79495
    },
    {
      "epoch": 0.8087158203125,
      "grad_norm": 13.964309692382812,
      "learning_rate": 4.388498749265366e-07,
      "loss": 2.767,
      "step": 79500
    },
    {
      "epoch": 0.8087666829427084,
      "grad_norm": 12.141261100769043,
      "learning_rate": 4.3862360060758724e-07,
      "loss": 3.391,
      "step": 79505
    },
    {
      "epoch": 0.8088175455729166,
      "grad_norm": 15.656390190124512,
      "learning_rate": 4.383973790284643e-07,
      "loss": 3.2007,
      "step": 79510
    },
    {
      "epoch": 0.808868408203125,
      "grad_norm": 14.146927833557129,
      "learning_rate": 4.381712101949562e-07,
      "loss": 3.2904,
      "step": 79515
    },
    {
      "epoch": 0.8089192708333334,
      "grad_norm": 8.907366752624512,
      "learning_rate": 4.3794509411284937e-07,
      "loss": 3.2788,
      "step": 79520
    },
    {
      "epoch": 0.8089701334635416,
      "grad_norm": 11.003146171569824,
      "learning_rate": 4.377190307879275e-07,
      "loss": 3.6502,
      "step": 79525
    },
    {
      "epoch": 0.80902099609375,
      "grad_norm": 13.753011703491211,
      "learning_rate": 4.374930202259764e-07,
      "loss": 3.4284,
      "step": 79530
    },
    {
      "epoch": 0.8090718587239584,
      "grad_norm": 11.551970481872559,
      "learning_rate": 4.3726706243277685e-07,
      "loss": 3.1446,
      "step": 79535
    },
    {
      "epoch": 0.8091227213541666,
      "grad_norm": 9.97728157043457,
      "learning_rate": 4.370411574141112e-07,
      "loss": 3.4146,
      "step": 79540
    },
    {
      "epoch": 0.809173583984375,
      "grad_norm": 13.558149337768555,
      "learning_rate": 4.368153051757579e-07,
      "loss": 3.4341,
      "step": 79545
    },
    {
      "epoch": 0.8092244466145834,
      "grad_norm": 9.130220413208008,
      "learning_rate": 4.3658950572349645e-07,
      "loss": 3.4231,
      "step": 79550
    },
    {
      "epoch": 0.8092753092447916,
      "grad_norm": 6.932345390319824,
      "learning_rate": 4.3636375906310343e-07,
      "loss": 3.5729,
      "step": 79555
    },
    {
      "epoch": 0.809326171875,
      "grad_norm": 15.159601211547852,
      "learning_rate": 4.361380652003541e-07,
      "loss": 3.4203,
      "step": 79560
    },
    {
      "epoch": 0.8093770345052084,
      "grad_norm": 8.320847511291504,
      "learning_rate": 4.3591242414102315e-07,
      "loss": 3.2303,
      "step": 79565
    },
    {
      "epoch": 0.8094278971354166,
      "grad_norm": 11.370145797729492,
      "learning_rate": 4.356868358908839e-07,
      "loss": 3.3368,
      "step": 79570
    },
    {
      "epoch": 0.809478759765625,
      "grad_norm": 7.661952495574951,
      "learning_rate": 4.3546130045570787e-07,
      "loss": 3.3427,
      "step": 79575
    },
    {
      "epoch": 0.8095296223958334,
      "grad_norm": 9.848603248596191,
      "learning_rate": 4.352358178412647e-07,
      "loss": 3.3152,
      "step": 79580
    },
    {
      "epoch": 0.8095804850260416,
      "grad_norm": 10.856739044189453,
      "learning_rate": 4.3501038805332407e-07,
      "loss": 3.5946,
      "step": 79585
    },
    {
      "epoch": 0.80963134765625,
      "grad_norm": 9.38187313079834,
      "learning_rate": 4.347850110976526e-07,
      "loss": 3.2375,
      "step": 79590
    },
    {
      "epoch": 0.8096822102864584,
      "grad_norm": 8.472463607788086,
      "learning_rate": 4.3455968698001783e-07,
      "loss": 3.7288,
      "step": 79595
    },
    {
      "epoch": 0.8097330729166666,
      "grad_norm": 7.300169944763184,
      "learning_rate": 4.3433441570618325e-07,
      "loss": 3.8467,
      "step": 79600
    },
    {
      "epoch": 0.809783935546875,
      "grad_norm": 14.652825355529785,
      "learning_rate": 4.3410919728191365e-07,
      "loss": 3.8647,
      "step": 79605
    },
    {
      "epoch": 0.8098347981770834,
      "grad_norm": 11.837065696716309,
      "learning_rate": 4.3388403171297053e-07,
      "loss": 3.1634,
      "step": 79610
    },
    {
      "epoch": 0.8098856608072916,
      "grad_norm": 12.23351001739502,
      "learning_rate": 4.3365891900511436e-07,
      "loss": 3.4731,
      "step": 79615
    },
    {
      "epoch": 0.8099365234375,
      "grad_norm": 9.871082305908203,
      "learning_rate": 4.3343385916410465e-07,
      "loss": 3.2557,
      "step": 79620
    },
    {
      "epoch": 0.8099873860677084,
      "grad_norm": 12.954231262207031,
      "learning_rate": 4.3320885219570063e-07,
      "loss": 3.3369,
      "step": 79625
    },
    {
      "epoch": 0.8100382486979166,
      "grad_norm": 18.884347915649414,
      "learning_rate": 4.329838981056583e-07,
      "loss": 3.1643,
      "step": 79630
    },
    {
      "epoch": 0.810089111328125,
      "grad_norm": 14.413152694702148,
      "learning_rate": 4.327589968997328e-07,
      "loss": 3.3079,
      "step": 79635
    },
    {
      "epoch": 0.8101399739583334,
      "grad_norm": 7.7837934494018555,
      "learning_rate": 4.325341485836773e-07,
      "loss": 3.4568,
      "step": 79640
    },
    {
      "epoch": 0.8101908365885416,
      "grad_norm": 10.19963550567627,
      "learning_rate": 4.323093531632466e-07,
      "loss": 3.3346,
      "step": 79645
    },
    {
      "epoch": 0.81024169921875,
      "grad_norm": 11.307597160339355,
      "learning_rate": 4.3208461064419114e-07,
      "loss": 3.4442,
      "step": 79650
    },
    {
      "epoch": 0.8102925618489584,
      "grad_norm": 18.645917892456055,
      "learning_rate": 4.3185992103226077e-07,
      "loss": 3.3448,
      "step": 79655
    },
    {
      "epoch": 0.8103434244791666,
      "grad_norm": 12.473796844482422,
      "learning_rate": 4.316352843332036e-07,
      "loss": 3.2527,
      "step": 79660
    },
    {
      "epoch": 0.810394287109375,
      "grad_norm": 11.506953239440918,
      "learning_rate": 4.3141070055276705e-07,
      "loss": 3.6653,
      "step": 79665
    },
    {
      "epoch": 0.8104451497395834,
      "grad_norm": 8.191093444824219,
      "learning_rate": 4.311861696966985e-07,
      "loss": 3.0089,
      "step": 79670
    },
    {
      "epoch": 0.8104960123697916,
      "grad_norm": 15.774288177490234,
      "learning_rate": 4.309616917707404e-07,
      "loss": 3.8248,
      "step": 79675
    },
    {
      "epoch": 0.810546875,
      "grad_norm": 19.564104080200195,
      "learning_rate": 4.307372667806381e-07,
      "loss": 3.934,
      "step": 79680
    },
    {
      "epoch": 0.8105977376302084,
      "grad_norm": 14.84902286529541,
      "learning_rate": 4.305128947321319e-07,
      "loss": 2.8391,
      "step": 79685
    },
    {
      "epoch": 0.8106486002604166,
      "grad_norm": 12.76487922668457,
      "learning_rate": 4.3028857563096246e-07,
      "loss": 3.3434,
      "step": 79690
    },
    {
      "epoch": 0.810699462890625,
      "grad_norm": 16.582035064697266,
      "learning_rate": 4.3006430948286907e-07,
      "loss": 3.4959,
      "step": 79695
    },
    {
      "epoch": 0.8107503255208334,
      "grad_norm": 11.676108360290527,
      "learning_rate": 4.298400962935903e-07,
      "loss": 3.3459,
      "step": 79700
    },
    {
      "epoch": 0.8108011881510416,
      "grad_norm": 12.260275840759277,
      "learning_rate": 4.2961593606886206e-07,
      "loss": 3.239,
      "step": 79705
    },
    {
      "epoch": 0.81085205078125,
      "grad_norm": 8.587767601013184,
      "learning_rate": 4.2939182881441945e-07,
      "loss": 3.0153,
      "step": 79710
    },
    {
      "epoch": 0.8109029134114584,
      "grad_norm": 10.28516674041748,
      "learning_rate": 4.291677745359954e-07,
      "loss": 3.3732,
      "step": 79715
    },
    {
      "epoch": 0.8109537760416666,
      "grad_norm": 13.696601867675781,
      "learning_rate": 4.2894377323932286e-07,
      "loss": 3.3326,
      "step": 79720
    },
    {
      "epoch": 0.811004638671875,
      "grad_norm": 10.845861434936523,
      "learning_rate": 4.2871982493013385e-07,
      "loss": 3.3103,
      "step": 79725
    },
    {
      "epoch": 0.8110555013020834,
      "grad_norm": 13.977005958557129,
      "learning_rate": 4.284959296141569e-07,
      "loss": 3.6096,
      "step": 79730
    },
    {
      "epoch": 0.8111063639322916,
      "grad_norm": 12.174431800842285,
      "learning_rate": 4.282720872971202e-07,
      "loss": 3.646,
      "step": 79735
    },
    {
      "epoch": 0.8111572265625,
      "grad_norm": 9.921180725097656,
      "learning_rate": 4.2804829798475165e-07,
      "loss": 3.6864,
      "step": 79740
    },
    {
      "epoch": 0.8112080891927084,
      "grad_norm": 14.429951667785645,
      "learning_rate": 4.278245616827756e-07,
      "loss": 3.0137,
      "step": 79745
    },
    {
      "epoch": 0.8112589518229166,
      "grad_norm": 9.26318645477295,
      "learning_rate": 4.276008783969177e-07,
      "loss": 3.5189,
      "step": 79750
    },
    {
      "epoch": 0.811309814453125,
      "grad_norm": 8.710182189941406,
      "learning_rate": 4.2737724813289945e-07,
      "loss": 3.1991,
      "step": 79755
    },
    {
      "epoch": 0.8113606770833334,
      "grad_norm": 11.180933952331543,
      "learning_rate": 4.2715367089644354e-07,
      "loss": 3.5738,
      "step": 79760
    },
    {
      "epoch": 0.8114115397135416,
      "grad_norm": 15.223068237304688,
      "learning_rate": 4.2693014669326986e-07,
      "loss": 3.4119,
      "step": 79765
    },
    {
      "epoch": 0.81146240234375,
      "grad_norm": 7.4405999183654785,
      "learning_rate": 4.267066755290963e-07,
      "loss": 3.0198,
      "step": 79770
    },
    {
      "epoch": 0.8115132649739584,
      "grad_norm": 17.034854888916016,
      "learning_rate": 4.2648325740964086e-07,
      "loss": 3.0463,
      "step": 79775
    },
    {
      "epoch": 0.8115641276041666,
      "grad_norm": 11.081181526184082,
      "learning_rate": 4.2625989234062084e-07,
      "loss": 3.3451,
      "step": 79780
    },
    {
      "epoch": 0.811614990234375,
      "grad_norm": 11.616988182067871,
      "learning_rate": 4.260365803277494e-07,
      "loss": 3.4363,
      "step": 79785
    },
    {
      "epoch": 0.8116658528645834,
      "grad_norm": 13.727701187133789,
      "learning_rate": 4.2581332137674014e-07,
      "loss": 3.273,
      "step": 79790
    },
    {
      "epoch": 0.8117167154947916,
      "grad_norm": 10.448540687561035,
      "learning_rate": 4.2559011549330535e-07,
      "loss": 3.1385,
      "step": 79795
    },
    {
      "epoch": 0.811767578125,
      "grad_norm": 9.588068962097168,
      "learning_rate": 4.2536696268315663e-07,
      "loss": 3.5569,
      "step": 79800
    },
    {
      "epoch": 0.8118184407552084,
      "grad_norm": 8.408053398132324,
      "learning_rate": 4.251438629520022e-07,
      "loss": 2.8845,
      "step": 79805
    },
    {
      "epoch": 0.8118693033854166,
      "grad_norm": 13.213644981384277,
      "learning_rate": 4.249208163055496e-07,
      "loss": 3.2393,
      "step": 79810
    },
    {
      "epoch": 0.811920166015625,
      "grad_norm": 15.220480918884277,
      "learning_rate": 4.2469782274950687e-07,
      "loss": 3.8526,
      "step": 79815
    },
    {
      "epoch": 0.8119710286458334,
      "grad_norm": 16.236431121826172,
      "learning_rate": 4.244748822895775e-07,
      "loss": 3.4945,
      "step": 79820
    },
    {
      "epoch": 0.8120218912760416,
      "grad_norm": 8.124288558959961,
      "learning_rate": 4.242519949314672e-07,
      "loss": 3.219,
      "step": 79825
    },
    {
      "epoch": 0.81207275390625,
      "grad_norm": 12.191054344177246,
      "learning_rate": 4.240291606808769e-07,
      "loss": 3.271,
      "step": 79830
    },
    {
      "epoch": 0.8121236165364584,
      "grad_norm": 12.073549270629883,
      "learning_rate": 4.2380637954350904e-07,
      "loss": 3.9508,
      "step": 79835
    },
    {
      "epoch": 0.8121744791666666,
      "grad_norm": 17.343311309814453,
      "learning_rate": 4.2358365152506296e-07,
      "loss": 3.6859,
      "step": 79840
    },
    {
      "epoch": 0.812225341796875,
      "grad_norm": 9.703289985656738,
      "learning_rate": 4.233609766312363e-07,
      "loss": 3.0531,
      "step": 79845
    },
    {
      "epoch": 0.8122762044270834,
      "grad_norm": 14.241864204406738,
      "learning_rate": 4.2313835486772705e-07,
      "loss": 3.7319,
      "step": 79850
    },
    {
      "epoch": 0.8123270670572916,
      "grad_norm": 10.567691802978516,
      "learning_rate": 4.229157862402311e-07,
      "loss": 3.0844,
      "step": 79855
    },
    {
      "epoch": 0.8123779296875,
      "grad_norm": 10.341302871704102,
      "learning_rate": 4.2269327075444233e-07,
      "loss": 3.596,
      "step": 79860
    },
    {
      "epoch": 0.8124287923177084,
      "grad_norm": 12.9301118850708,
      "learning_rate": 4.224708084160542e-07,
      "loss": 3.2738,
      "step": 79865
    },
    {
      "epoch": 0.8124796549479166,
      "grad_norm": 15.44719123840332,
      "learning_rate": 4.2224839923075716e-07,
      "loss": 3.6493,
      "step": 79870
    },
    {
      "epoch": 0.812530517578125,
      "grad_norm": 17.016420364379883,
      "learning_rate": 4.220260432042422e-07,
      "loss": 3.4857,
      "step": 79875
    },
    {
      "epoch": 0.8125813802083334,
      "grad_norm": 9.550246238708496,
      "learning_rate": 4.218037403421993e-07,
      "loss": 3.4953,
      "step": 79880
    },
    {
      "epoch": 0.8126322428385416,
      "grad_norm": 12.058992385864258,
      "learning_rate": 4.2158149065031436e-07,
      "loss": 3.3917,
      "step": 79885
    },
    {
      "epoch": 0.81268310546875,
      "grad_norm": 19.84065055847168,
      "learning_rate": 4.2135929413427504e-07,
      "loss": 3.7188,
      "step": 79890
    },
    {
      "epoch": 0.8127339680989584,
      "grad_norm": 10.726000785827637,
      "learning_rate": 4.211371507997655e-07,
      "loss": 3.2962,
      "step": 79895
    },
    {
      "epoch": 0.8127848307291666,
      "grad_norm": 18.25348663330078,
      "learning_rate": 4.209150606524684e-07,
      "loss": 3.3035,
      "step": 79900
    },
    {
      "epoch": 0.812835693359375,
      "grad_norm": 14.618754386901855,
      "learning_rate": 4.206930236980669e-07,
      "loss": 3.2014,
      "step": 79905
    },
    {
      "epoch": 0.8128865559895834,
      "grad_norm": 14.573039054870605,
      "learning_rate": 4.204710399422421e-07,
      "loss": 3.3201,
      "step": 79910
    },
    {
      "epoch": 0.8129374186197916,
      "grad_norm": 16.528791427612305,
      "learning_rate": 4.202491093906727e-07,
      "loss": 3.5633,
      "step": 79915
    },
    {
      "epoch": 0.81298828125,
      "grad_norm": 9.822667121887207,
      "learning_rate": 4.20027232049037e-07,
      "loss": 3.1876,
      "step": 79920
    },
    {
      "epoch": 0.8130391438802084,
      "grad_norm": 19.91170310974121,
      "learning_rate": 4.19805407923011e-07,
      "loss": 3.3017,
      "step": 79925
    },
    {
      "epoch": 0.8130900065104166,
      "grad_norm": 11.826841354370117,
      "learning_rate": 4.195836370182704e-07,
      "loss": 3.24,
      "step": 79930
    },
    {
      "epoch": 0.813140869140625,
      "grad_norm": 15.935693740844727,
      "learning_rate": 4.193619193404902e-07,
      "loss": 3.2237,
      "step": 79935
    },
    {
      "epoch": 0.8131917317708334,
      "grad_norm": 10.841400146484375,
      "learning_rate": 4.191402548953419e-07,
      "loss": 2.9862,
      "step": 79940
    },
    {
      "epoch": 0.8132425944010416,
      "grad_norm": 14.67307186126709,
      "learning_rate": 4.189186436884965e-07,
      "loss": 3.4626,
      "step": 79945
    },
    {
      "epoch": 0.81329345703125,
      "grad_norm": 12.722312927246094,
      "learning_rate": 4.1869708572562425e-07,
      "loss": 3.3784,
      "step": 79950
    },
    {
      "epoch": 0.8133443196614584,
      "grad_norm": 14.460338592529297,
      "learning_rate": 4.1847558101239414e-07,
      "loss": 3.571,
      "step": 79955
    },
    {
      "epoch": 0.8133951822916666,
      "grad_norm": 6.846340179443359,
      "learning_rate": 4.1825412955447314e-07,
      "loss": 3.1522,
      "step": 79960
    },
    {
      "epoch": 0.813446044921875,
      "grad_norm": 13.961633682250977,
      "learning_rate": 4.1803273135752636e-07,
      "loss": 3.4373,
      "step": 79965
    },
    {
      "epoch": 0.8134969075520834,
      "grad_norm": 10.45997428894043,
      "learning_rate": 4.17811386427219e-07,
      "loss": 3.4645,
      "step": 79970
    },
    {
      "epoch": 0.8135477701822916,
      "grad_norm": 14.717330932617188,
      "learning_rate": 4.1759009476921327e-07,
      "loss": 3.3381,
      "step": 79975
    },
    {
      "epoch": 0.8135986328125,
      "grad_norm": 13.302783966064453,
      "learning_rate": 4.173688563891715e-07,
      "loss": 3.1185,
      "step": 79980
    },
    {
      "epoch": 0.8136494954427084,
      "grad_norm": 10.639537811279297,
      "learning_rate": 4.171476712927544e-07,
      "loss": 3.1504,
      "step": 79985
    },
    {
      "epoch": 0.8137003580729166,
      "grad_norm": 12.81106185913086,
      "learning_rate": 4.1692653948562035e-07,
      "loss": 3.3925,
      "step": 79990
    },
    {
      "epoch": 0.813751220703125,
      "grad_norm": 12.533110618591309,
      "learning_rate": 4.1670546097342723e-07,
      "loss": 3.5072,
      "step": 79995
    },
    {
      "epoch": 0.8138020833333334,
      "grad_norm": 8.067879676818848,
      "learning_rate": 4.164844357618303e-07,
      "loss": 3.2179,
      "step": 80000
    },
    {
      "epoch": 0.8138529459635416,
      "grad_norm": 15.618395805358887,
      "learning_rate": 4.1626346385648553e-07,
      "loss": 3.1992,
      "step": 80005
    },
    {
      "epoch": 0.81390380859375,
      "grad_norm": 9.165993690490723,
      "learning_rate": 4.160425452630465e-07,
      "loss": 3.168,
      "step": 80010
    },
    {
      "epoch": 0.8139546712239584,
      "grad_norm": 6.94345760345459,
      "learning_rate": 4.1582167998716505e-07,
      "loss": 2.9606,
      "step": 80015
    },
    {
      "epoch": 0.8140055338541666,
      "grad_norm": 12.507560729980469,
      "learning_rate": 4.156008680344914e-07,
      "loss": 3.3534,
      "step": 80020
    },
    {
      "epoch": 0.814056396484375,
      "grad_norm": 17.14674949645996,
      "learning_rate": 4.1538010941067575e-07,
      "loss": 3.5355,
      "step": 80025
    },
    {
      "epoch": 0.8141072591145834,
      "grad_norm": 12.378942489624023,
      "learning_rate": 4.151594041213655e-07,
      "loss": 3.1759,
      "step": 80030
    },
    {
      "epoch": 0.8141581217447916,
      "grad_norm": 8.899665832519531,
      "learning_rate": 4.149387521722084e-07,
      "loss": 3.3154,
      "step": 80035
    },
    {
      "epoch": 0.814208984375,
      "grad_norm": 13.174806594848633,
      "learning_rate": 4.1471815356884855e-07,
      "loss": 3.2039,
      "step": 80040
    },
    {
      "epoch": 0.8142598470052084,
      "grad_norm": 10.049018859863281,
      "learning_rate": 4.144976083169311e-07,
      "loss": 3.5286,
      "step": 80045
    },
    {
      "epoch": 0.8143107096354166,
      "grad_norm": 10.214993476867676,
      "learning_rate": 4.1427711642209795e-07,
      "loss": 3.1437,
      "step": 80050
    },
    {
      "epoch": 0.814361572265625,
      "grad_norm": 14.105937004089355,
      "learning_rate": 4.140566778899896e-07,
      "loss": 3.1775,
      "step": 80055
    },
    {
      "epoch": 0.8144124348958334,
      "grad_norm": 12.205314636230469,
      "learning_rate": 4.138362927262471e-07,
      "loss": 3.3314,
      "step": 80060
    },
    {
      "epoch": 0.8144632975260416,
      "grad_norm": 9.415129661560059,
      "learning_rate": 4.13615960936509e-07,
      "loss": 3.3394,
      "step": 80065
    },
    {
      "epoch": 0.81451416015625,
      "grad_norm": 14.456554412841797,
      "learning_rate": 4.133956825264118e-07,
      "loss": 3.2737,
      "step": 80070
    },
    {
      "epoch": 0.8145650227864584,
      "grad_norm": 13.304035186767578,
      "learning_rate": 4.1317545750159114e-07,
      "loss": 3.7365,
      "step": 80075
    },
    {
      "epoch": 0.8146158854166666,
      "grad_norm": 11.47937297821045,
      "learning_rate": 4.1295528586768245e-07,
      "loss": 3.5942,
      "step": 80080
    },
    {
      "epoch": 0.814666748046875,
      "grad_norm": 11.948128700256348,
      "learning_rate": 4.1273516763031736e-07,
      "loss": 3.3358,
      "step": 80085
    },
    {
      "epoch": 0.8147176106770834,
      "grad_norm": 14.301944732666016,
      "learning_rate": 4.125151027951288e-07,
      "loss": 3.3043,
      "step": 80090
    },
    {
      "epoch": 0.8147684733072916,
      "grad_norm": 10.94615364074707,
      "learning_rate": 4.122950913677459e-07,
      "loss": 3.3121,
      "step": 80095
    },
    {
      "epoch": 0.8148193359375,
      "grad_norm": 9.940673828125,
      "learning_rate": 4.1207513335379893e-07,
      "loss": 3.0426,
      "step": 80100
    },
    {
      "epoch": 0.8148701985677084,
      "grad_norm": 11.303471565246582,
      "learning_rate": 4.118552287589142e-07,
      "loss": 3.0874,
      "step": 80105
    },
    {
      "epoch": 0.8149210611979166,
      "grad_norm": 15.60290813446045,
      "learning_rate": 4.1163537758871914e-07,
      "loss": 3.4599,
      "step": 80110
    },
    {
      "epoch": 0.814971923828125,
      "grad_norm": 9.669926643371582,
      "learning_rate": 4.114155798488373e-07,
      "loss": 3.395,
      "step": 80115
    },
    {
      "epoch": 0.8150227864583334,
      "grad_norm": 12.754854202270508,
      "learning_rate": 4.1119583554489334e-07,
      "loss": 3.4149,
      "step": 80120
    },
    {
      "epoch": 0.8150736490885416,
      "grad_norm": 8.297149658203125,
      "learning_rate": 4.109761446825089e-07,
      "loss": 3.0322,
      "step": 80125
    },
    {
      "epoch": 0.81512451171875,
      "grad_norm": 8.285845756530762,
      "learning_rate": 4.107565072673042e-07,
      "loss": 3.4565,
      "step": 80130
    },
    {
      "epoch": 0.8151753743489584,
      "grad_norm": 12.29454231262207,
      "learning_rate": 4.105369233048989e-07,
      "loss": 3.6493,
      "step": 80135
    },
    {
      "epoch": 0.8152262369791666,
      "grad_norm": 12.848197937011719,
      "learning_rate": 4.103173928009116e-07,
      "loss": 3.1917,
      "step": 80140
    },
    {
      "epoch": 0.815277099609375,
      "grad_norm": 14.257293701171875,
      "learning_rate": 4.1009791576095884e-07,
      "loss": 3.3103,
      "step": 80145
    },
    {
      "epoch": 0.8153279622395834,
      "grad_norm": 7.8373847007751465,
      "learning_rate": 4.098784921906554e-07,
      "loss": 3.8208,
      "step": 80150
    },
    {
      "epoch": 0.8153788248697916,
      "grad_norm": 14.487406730651855,
      "learning_rate": 4.096591220956145e-07,
      "loss": 3.3984,
      "step": 80155
    },
    {
      "epoch": 0.8154296875,
      "grad_norm": 16.87422752380371,
      "learning_rate": 4.094398054814497e-07,
      "loss": 3.8078,
      "step": 80160
    },
    {
      "epoch": 0.8154805501302084,
      "grad_norm": 13.352436065673828,
      "learning_rate": 4.0922054235377264e-07,
      "loss": 3.6581,
      "step": 80165
    },
    {
      "epoch": 0.8155314127604166,
      "grad_norm": 8.779556274414062,
      "learning_rate": 4.090013327181916e-07,
      "loss": 3.0287,
      "step": 80170
    },
    {
      "epoch": 0.815582275390625,
      "grad_norm": 12.838428497314453,
      "learning_rate": 4.0878217658031656e-07,
      "loss": 3.0189,
      "step": 80175
    },
    {
      "epoch": 0.8156331380208334,
      "grad_norm": 13.048858642578125,
      "learning_rate": 4.085630739457538e-07,
      "loss": 3.0629,
      "step": 80180
    },
    {
      "epoch": 0.8156840006510416,
      "grad_norm": 12.09554386138916,
      "learning_rate": 4.083440248201087e-07,
      "loss": 3.3387,
      "step": 80185
    },
    {
      "epoch": 0.81573486328125,
      "grad_norm": 9.71069049835205,
      "learning_rate": 4.0812502920898585e-07,
      "loss": 3.4874,
      "step": 80190
    },
    {
      "epoch": 0.8157857259114584,
      "grad_norm": 11.654038429260254,
      "learning_rate": 4.079060871179888e-07,
      "loss": 3.252,
      "step": 80195
    },
    {
      "epoch": 0.8158365885416666,
      "grad_norm": 10.531092643737793,
      "learning_rate": 4.0768719855271866e-07,
      "loss": 3.2504,
      "step": 80200
    },
    {
      "epoch": 0.815887451171875,
      "grad_norm": 13.387083053588867,
      "learning_rate": 4.074683635187757e-07,
      "loss": 3.0339,
      "step": 80205
    },
    {
      "epoch": 0.8159383138020834,
      "grad_norm": 10.929668426513672,
      "learning_rate": 4.0724958202175786e-07,
      "loss": 3.2993,
      "step": 80210
    },
    {
      "epoch": 0.8159891764322916,
      "grad_norm": 13.87834358215332,
      "learning_rate": 4.070308540672635e-07,
      "loss": 3.1624,
      "step": 80215
    },
    {
      "epoch": 0.8160400390625,
      "grad_norm": 9.980636596679688,
      "learning_rate": 4.0681217966088956e-07,
      "loss": 3.4164,
      "step": 80220
    },
    {
      "epoch": 0.8160909016927084,
      "grad_norm": 15.008100509643555,
      "learning_rate": 4.0659355880822955e-07,
      "loss": 3.0908,
      "step": 80225
    },
    {
      "epoch": 0.8161417643229166,
      "grad_norm": 8.804240226745605,
      "learning_rate": 4.0637499151487655e-07,
      "loss": 3.4293,
      "step": 80230
    },
    {
      "epoch": 0.816192626953125,
      "grad_norm": 10.993620872497559,
      "learning_rate": 4.0615647778642383e-07,
      "loss": 3.4002,
      "step": 80235
    },
    {
      "epoch": 0.8162434895833334,
      "grad_norm": 11.344161987304688,
      "learning_rate": 4.059380176284608e-07,
      "loss": 3.3614,
      "step": 80240
    },
    {
      "epoch": 0.8162943522135416,
      "grad_norm": 12.82261848449707,
      "learning_rate": 4.0571961104657773e-07,
      "loss": 3.1997,
      "step": 80245
    },
    {
      "epoch": 0.81634521484375,
      "grad_norm": 16.50777816772461,
      "learning_rate": 4.055012580463613e-07,
      "loss": 3.7641,
      "step": 80250
    },
    {
      "epoch": 0.8163960774739584,
      "grad_norm": 8.51815414428711,
      "learning_rate": 4.052829586333995e-07,
      "loss": 3.4647,
      "step": 80255
    },
    {
      "epoch": 0.8164469401041666,
      "grad_norm": 15.575868606567383,
      "learning_rate": 4.050647128132762e-07,
      "loss": 3.4649,
      "step": 80260
    },
    {
      "epoch": 0.816497802734375,
      "grad_norm": 7.848937034606934,
      "learning_rate": 4.0484652059157554e-07,
      "loss": 3.2815,
      "step": 80265
    },
    {
      "epoch": 0.8165486653645834,
      "grad_norm": 15.685480117797852,
      "learning_rate": 4.0462838197388057e-07,
      "loss": 3.3155,
      "step": 80270
    },
    {
      "epoch": 0.8165995279947916,
      "grad_norm": 14.504178047180176,
      "learning_rate": 4.044102969657718e-07,
      "loss": 3.155,
      "step": 80275
    },
    {
      "epoch": 0.816650390625,
      "grad_norm": 14.9198637008667,
      "learning_rate": 4.0419226557282893e-07,
      "loss": 3.355,
      "step": 80280
    },
    {
      "epoch": 0.8167012532552084,
      "grad_norm": 11.11740779876709,
      "learning_rate": 4.0397428780062975e-07,
      "loss": 3.098,
      "step": 80285
    },
    {
      "epoch": 0.8167521158854166,
      "grad_norm": 10.391766548156738,
      "learning_rate": 4.0375636365475144e-07,
      "loss": 3.2861,
      "step": 80290
    },
    {
      "epoch": 0.816802978515625,
      "grad_norm": 11.091429710388184,
      "learning_rate": 4.0353849314077016e-07,
      "loss": 3.2244,
      "step": 80295
    },
    {
      "epoch": 0.8168538411458334,
      "grad_norm": 12.310245513916016,
      "learning_rate": 4.0332067626425967e-07,
      "loss": 3.8048,
      "step": 80300
    },
    {
      "epoch": 0.8169047037760416,
      "grad_norm": 7.162967681884766,
      "learning_rate": 4.031029130307923e-07,
      "loss": 3.0874,
      "step": 80305
    },
    {
      "epoch": 0.81695556640625,
      "grad_norm": 13.751084327697754,
      "learning_rate": 4.0288520344594045e-07,
      "loss": 3.2759,
      "step": 80310
    },
    {
      "epoch": 0.8170064290364584,
      "grad_norm": 10.215845108032227,
      "learning_rate": 4.0266754751527305e-07,
      "loss": 2.9698,
      "step": 80315
    },
    {
      "epoch": 0.8170572916666666,
      "grad_norm": 10.31053638458252,
      "learning_rate": 4.0244994524435976e-07,
      "loss": 2.9662,
      "step": 80320
    },
    {
      "epoch": 0.817108154296875,
      "grad_norm": 8.255980491638184,
      "learning_rate": 4.0223239663876674e-07,
      "loss": 3.9405,
      "step": 80325
    },
    {
      "epoch": 0.8171590169270834,
      "grad_norm": 11.991215705871582,
      "learning_rate": 4.0201490170406144e-07,
      "loss": 3.0397,
      "step": 80330
    },
    {
      "epoch": 0.8172098795572916,
      "grad_norm": 14.226103782653809,
      "learning_rate": 4.017974604458075e-07,
      "loss": 3.1728,
      "step": 80335
    },
    {
      "epoch": 0.8172607421875,
      "grad_norm": 9.138517379760742,
      "learning_rate": 4.0158007286956765e-07,
      "loss": 2.9813,
      "step": 80340
    },
    {
      "epoch": 0.8173116048177084,
      "grad_norm": 9.908602714538574,
      "learning_rate": 4.013627389809044e-07,
      "loss": 3.0246,
      "step": 80345
    },
    {
      "epoch": 0.8173624674479166,
      "grad_norm": 16.149280548095703,
      "learning_rate": 4.011454587853783e-07,
      "loss": 3.2871,
      "step": 80350
    },
    {
      "epoch": 0.817413330078125,
      "grad_norm": 13.814882278442383,
      "learning_rate": 4.0092823228854857e-07,
      "loss": 3.459,
      "step": 80355
    },
    {
      "epoch": 0.8174641927083334,
      "grad_norm": 9.240151405334473,
      "learning_rate": 4.007110594959715e-07,
      "loss": 3.6859,
      "step": 80360
    },
    {
      "epoch": 0.8175150553385416,
      "grad_norm": 14.212241172790527,
      "learning_rate": 4.0049394041320523e-07,
      "loss": 3.6496,
      "step": 80365
    },
    {
      "epoch": 0.81756591796875,
      "grad_norm": 15.584053993225098,
      "learning_rate": 4.0027687504580333e-07,
      "loss": 3.3167,
      "step": 80370
    },
    {
      "epoch": 0.8176167805989584,
      "grad_norm": 9.203264236450195,
      "learning_rate": 4.000598633993202e-07,
      "loss": 3.4071,
      "step": 80375
    },
    {
      "epoch": 0.8176676432291666,
      "grad_norm": 11.013337135314941,
      "learning_rate": 3.9984290547930737e-07,
      "loss": 3.8117,
      "step": 80380
    },
    {
      "epoch": 0.817718505859375,
      "grad_norm": 8.93172550201416,
      "learning_rate": 3.9962600129131643e-07,
      "loss": 3.1961,
      "step": 80385
    },
    {
      "epoch": 0.8177693684895834,
      "grad_norm": 14.087814331054688,
      "learning_rate": 3.994091508408965e-07,
      "loss": 3.1039,
      "step": 80390
    },
    {
      "epoch": 0.8178202311197916,
      "grad_norm": 14.619377136230469,
      "learning_rate": 3.9919235413359486e-07,
      "loss": 3.6241,
      "step": 80395
    },
    {
      "epoch": 0.81787109375,
      "grad_norm": 14.753565788269043,
      "learning_rate": 3.9897561117495905e-07,
      "loss": 3.3395,
      "step": 80400
    },
    {
      "epoch": 0.8179219563802084,
      "grad_norm": 18.728551864624023,
      "learning_rate": 3.9875892197053486e-07,
      "loss": 3.2673,
      "step": 80405
    },
    {
      "epoch": 0.8179728190104166,
      "grad_norm": 11.525137901306152,
      "learning_rate": 3.9854228652586537e-07,
      "loss": 3.6199,
      "step": 80410
    },
    {
      "epoch": 0.818023681640625,
      "grad_norm": 10.524356842041016,
      "learning_rate": 3.983257048464928e-07,
      "loss": 3.6003,
      "step": 80415
    },
    {
      "epoch": 0.8180745442708334,
      "grad_norm": 16.000789642333984,
      "learning_rate": 3.9810917693795917e-07,
      "loss": 3.3061,
      "step": 80420
    },
    {
      "epoch": 0.8181254069010416,
      "grad_norm": 13.808717727661133,
      "learning_rate": 3.978927028058044e-07,
      "loss": 3.1418,
      "step": 80425
    },
    {
      "epoch": 0.81817626953125,
      "grad_norm": 14.330815315246582,
      "learning_rate": 3.976762824555666e-07,
      "loss": 3.0111,
      "step": 80430
    },
    {
      "epoch": 0.8182271321614584,
      "grad_norm": 15.074409484863281,
      "learning_rate": 3.97459915892783e-07,
      "loss": 3.6294,
      "step": 80435
    },
    {
      "epoch": 0.8182779947916666,
      "grad_norm": 11.594634056091309,
      "learning_rate": 3.972436031229884e-07,
      "loss": 3.4432,
      "step": 80440
    },
    {
      "epoch": 0.818328857421875,
      "grad_norm": 11.404607772827148,
      "learning_rate": 3.970273441517178e-07,
      "loss": 3.7119,
      "step": 80445
    },
    {
      "epoch": 0.8183797200520834,
      "grad_norm": 10.453607559204102,
      "learning_rate": 3.968111389845047e-07,
      "loss": 3.321,
      "step": 80450
    },
    {
      "epoch": 0.8184305826822916,
      "grad_norm": 12.711167335510254,
      "learning_rate": 3.9659498762688016e-07,
      "loss": 3.3319,
      "step": 80455
    },
    {
      "epoch": 0.8184814453125,
      "grad_norm": 11.754729270935059,
      "learning_rate": 3.9637889008437375e-07,
      "loss": 3.0155,
      "step": 80460
    },
    {
      "epoch": 0.8185323079427084,
      "grad_norm": 10.359018325805664,
      "learning_rate": 3.961628463625153e-07,
      "loss": 3.2291,
      "step": 80465
    },
    {
      "epoch": 0.8185831705729166,
      "grad_norm": 12.879046440124512,
      "learning_rate": 3.9594685646683117e-07,
      "loss": 3.2591,
      "step": 80470
    },
    {
      "epoch": 0.818634033203125,
      "grad_norm": 11.944828033447266,
      "learning_rate": 3.957309204028481e-07,
      "loss": 3.0912,
      "step": 80475
    },
    {
      "epoch": 0.8186848958333334,
      "grad_norm": 11.149519920349121,
      "learning_rate": 3.955150381760911e-07,
      "loss": 3.2856,
      "step": 80480
    },
    {
      "epoch": 0.8187357584635416,
      "grad_norm": 9.422290802001953,
      "learning_rate": 3.95299209792083e-07,
      "loss": 3.38,
      "step": 80485
    },
    {
      "epoch": 0.81878662109375,
      "grad_norm": 10.456183433532715,
      "learning_rate": 3.95083435256346e-07,
      "loss": 3.3319,
      "step": 80490
    },
    {
      "epoch": 0.8188374837239584,
      "grad_norm": 13.086915969848633,
      "learning_rate": 3.948677145743995e-07,
      "loss": 3.7862,
      "step": 80495
    },
    {
      "epoch": 0.8188883463541666,
      "grad_norm": 10.651822090148926,
      "learning_rate": 3.9465204775176373e-07,
      "loss": 3.6872,
      "step": 80500
    },
    {
      "epoch": 0.818939208984375,
      "grad_norm": 9.868151664733887,
      "learning_rate": 3.944364347939569e-07,
      "loss": 3.0285,
      "step": 80505
    },
    {
      "epoch": 0.8189900716145834,
      "grad_norm": 11.998753547668457,
      "learning_rate": 3.94220875706495e-07,
      "loss": 3.2442,
      "step": 80510
    },
    {
      "epoch": 0.8190409342447916,
      "grad_norm": 11.459403991699219,
      "learning_rate": 3.940053704948921e-07,
      "loss": 3.3359,
      "step": 80515
    },
    {
      "epoch": 0.819091796875,
      "grad_norm": 14.966922760009766,
      "learning_rate": 3.9378991916466295e-07,
      "loss": 3.3292,
      "step": 80520
    },
    {
      "epoch": 0.8191426595052084,
      "grad_norm": 10.536090850830078,
      "learning_rate": 3.9357452172131937e-07,
      "loss": 2.9818,
      "step": 80525
    },
    {
      "epoch": 0.8191935221354166,
      "grad_norm": 8.744247436523438,
      "learning_rate": 3.9335917817037274e-07,
      "loss": 3.3249,
      "step": 80530
    },
    {
      "epoch": 0.819244384765625,
      "grad_norm": 11.743062973022461,
      "learning_rate": 3.9314388851733136e-07,
      "loss": 3.2735,
      "step": 80535
    },
    {
      "epoch": 0.8192952473958334,
      "grad_norm": 13.098162651062012,
      "learning_rate": 3.929286527677051e-07,
      "loss": 3.361,
      "step": 80540
    },
    {
      "epoch": 0.8193461100260416,
      "grad_norm": 15.775094985961914,
      "learning_rate": 3.9271347092699975e-07,
      "loss": 3.0978,
      "step": 80545
    },
    {
      "epoch": 0.81939697265625,
      "grad_norm": 10.93581485748291,
      "learning_rate": 3.924983430007204e-07,
      "loss": 3.2549,
      "step": 80550
    },
    {
      "epoch": 0.8194478352864584,
      "grad_norm": 7.797806739807129,
      "learning_rate": 3.922832689943712e-07,
      "loss": 3.5114,
      "step": 80555
    },
    {
      "epoch": 0.8194986979166666,
      "grad_norm": 14.676947593688965,
      "learning_rate": 3.9206824891345556e-07,
      "loss": 3.2431,
      "step": 80560
    },
    {
      "epoch": 0.819549560546875,
      "grad_norm": 13.802629470825195,
      "learning_rate": 3.918532827634741e-07,
      "loss": 3.2136,
      "step": 80565
    },
    {
      "epoch": 0.8196004231770834,
      "grad_norm": 16.6387939453125,
      "learning_rate": 3.9163837054992634e-07,
      "loss": 3.5163,
      "step": 80570
    },
    {
      "epoch": 0.8196512858072916,
      "grad_norm": 14.772869110107422,
      "learning_rate": 3.914235122783108e-07,
      "loss": 3.9268,
      "step": 80575
    },
    {
      "epoch": 0.8197021484375,
      "grad_norm": 9.712909698486328,
      "learning_rate": 3.9120870795412575e-07,
      "loss": 3.4206,
      "step": 80580
    },
    {
      "epoch": 0.8197530110677084,
      "grad_norm": 7.7602691650390625,
      "learning_rate": 3.9099395758286586e-07,
      "loss": 3.282,
      "step": 80585
    },
    {
      "epoch": 0.8198038736979166,
      "grad_norm": 12.946903228759766,
      "learning_rate": 3.907792611700253e-07,
      "loss": 3.226,
      "step": 80590
    },
    {
      "epoch": 0.819854736328125,
      "grad_norm": 15.746535301208496,
      "learning_rate": 3.9056461872109794e-07,
      "loss": 3.3232,
      "step": 80595
    },
    {
      "epoch": 0.8199055989583334,
      "grad_norm": 11.906881332397461,
      "learning_rate": 3.903500302415741e-07,
      "loss": 3.2913,
      "step": 80600
    },
    {
      "epoch": 0.8199564615885416,
      "grad_norm": 10.969463348388672,
      "learning_rate": 3.901354957369455e-07,
      "loss": 3.8135,
      "step": 80605
    },
    {
      "epoch": 0.82000732421875,
      "grad_norm": 10.680028915405273,
      "learning_rate": 3.8992101521269947e-07,
      "loss": 3.3671,
      "step": 80610
    },
    {
      "epoch": 0.8200581868489584,
      "grad_norm": 14.710232734680176,
      "learning_rate": 3.8970658867432454e-07,
      "loss": 3.5764,
      "step": 80615
    },
    {
      "epoch": 0.8201090494791666,
      "grad_norm": 9.277506828308105,
      "learning_rate": 3.8949221612730653e-07,
      "loss": 3.781,
      "step": 80620
    },
    {
      "epoch": 0.820159912109375,
      "grad_norm": 9.739691734313965,
      "learning_rate": 3.892778975771294e-07,
      "loss": 3.6042,
      "step": 80625
    },
    {
      "epoch": 0.8202107747395834,
      "grad_norm": 10.505010604858398,
      "learning_rate": 3.8906363302927686e-07,
      "loss": 3.2156,
      "step": 80630
    },
    {
      "epoch": 0.8202616373697916,
      "grad_norm": 9.160713195800781,
      "learning_rate": 3.888494224892314e-07,
      "loss": 3.11,
      "step": 80635
    },
    {
      "epoch": 0.8203125,
      "grad_norm": 10.293058395385742,
      "learning_rate": 3.886352659624734e-07,
      "loss": 3.5203,
      "step": 80640
    },
    {
      "epoch": 0.8203633626302084,
      "grad_norm": 8.177082061767578,
      "learning_rate": 3.8842116345448126e-07,
      "loss": 3.4225,
      "step": 80645
    },
    {
      "epoch": 0.8204142252604166,
      "grad_norm": 16.60538101196289,
      "learning_rate": 3.882071149707328e-07,
      "loss": 3.0571,
      "step": 80650
    },
    {
      "epoch": 0.820465087890625,
      "grad_norm": 9.453953742980957,
      "learning_rate": 3.8799312051670477e-07,
      "loss": 3.1622,
      "step": 80655
    },
    {
      "epoch": 0.8205159505208334,
      "grad_norm": 11.494791984558105,
      "learning_rate": 3.877791800978728e-07,
      "loss": 3.3414,
      "step": 80660
    },
    {
      "epoch": 0.8205668131510416,
      "grad_norm": 12.629426002502441,
      "learning_rate": 3.875652937197091e-07,
      "loss": 3.4787,
      "step": 80665
    },
    {
      "epoch": 0.82061767578125,
      "grad_norm": 13.683670997619629,
      "learning_rate": 3.873514613876872e-07,
      "loss": 3.5224,
      "step": 80670
    },
    {
      "epoch": 0.8206685384114584,
      "grad_norm": 12.38894271850586,
      "learning_rate": 3.871376831072776e-07,
      "loss": 3.3603,
      "step": 80675
    },
    {
      "epoch": 0.8207194010416666,
      "grad_norm": 10.823607444763184,
      "learning_rate": 3.869239588839488e-07,
      "loss": 3.3912,
      "step": 80680
    },
    {
      "epoch": 0.820770263671875,
      "grad_norm": 12.35561466217041,
      "learning_rate": 3.8671028872317e-07,
      "loss": 3.4972,
      "step": 80685
    },
    {
      "epoch": 0.8208211263020834,
      "grad_norm": 10.105222702026367,
      "learning_rate": 3.8649667263040784e-07,
      "loss": 3.5993,
      "step": 80690
    },
    {
      "epoch": 0.8208719889322916,
      "grad_norm": 15.339810371398926,
      "learning_rate": 3.862831106111273e-07,
      "loss": 3.2245,
      "step": 80695
    },
    {
      "epoch": 0.8209228515625,
      "grad_norm": 10.440420150756836,
      "learning_rate": 3.8606960267079244e-07,
      "loss": 3.1923,
      "step": 80700
    },
    {
      "epoch": 0.8209737141927084,
      "grad_norm": 15.502533912658691,
      "learning_rate": 3.858561488148646e-07,
      "loss": 3.3615,
      "step": 80705
    },
    {
      "epoch": 0.8210245768229166,
      "grad_norm": 17.10973358154297,
      "learning_rate": 3.856427490488074e-07,
      "loss": 3.0419,
      "step": 80710
    },
    {
      "epoch": 0.821075439453125,
      "grad_norm": 7.994612216949463,
      "learning_rate": 3.8542940337807923e-07,
      "loss": 3.5715,
      "step": 80715
    },
    {
      "epoch": 0.8211263020833334,
      "grad_norm": 13.796808242797852,
      "learning_rate": 3.852161118081385e-07,
      "loss": 3.35,
      "step": 80720
    },
    {
      "epoch": 0.8211771647135416,
      "grad_norm": 14.857696533203125,
      "learning_rate": 3.8500287434444173e-07,
      "loss": 3.1631,
      "step": 80725
    },
    {
      "epoch": 0.82122802734375,
      "grad_norm": 10.685078620910645,
      "learning_rate": 3.847896909924451e-07,
      "loss": 3.1853,
      "step": 80730
    },
    {
      "epoch": 0.8212788899739584,
      "grad_norm": 7.9912285804748535,
      "learning_rate": 3.8457656175760335e-07,
      "loss": 3.4768,
      "step": 80735
    },
    {
      "epoch": 0.8213297526041666,
      "grad_norm": 11.736037254333496,
      "learning_rate": 3.843634866453691e-07,
      "loss": 3.1898,
      "step": 80740
    },
    {
      "epoch": 0.821380615234375,
      "grad_norm": 9.419626235961914,
      "learning_rate": 3.8415046566119276e-07,
      "loss": 3.0104,
      "step": 80745
    },
    {
      "epoch": 0.8214314778645834,
      "grad_norm": 12.664139747619629,
      "learning_rate": 3.83937498810526e-07,
      "loss": 3.0973,
      "step": 80750
    },
    {
      "epoch": 0.8214823404947916,
      "grad_norm": 17.065969467163086,
      "learning_rate": 3.8372458609881593e-07,
      "loss": 3.1934,
      "step": 80755
    },
    {
      "epoch": 0.821533203125,
      "grad_norm": 11.615386962890625,
      "learning_rate": 3.835117275315109e-07,
      "loss": 3.2291,
      "step": 80760
    },
    {
      "epoch": 0.8215840657552084,
      "grad_norm": 8.41677188873291,
      "learning_rate": 3.8329892311405744e-07,
      "loss": 3.2941,
      "step": 80765
    },
    {
      "epoch": 0.8216349283854166,
      "grad_norm": 9.951393127441406,
      "learning_rate": 3.830861728518989e-07,
      "loss": 3.3313,
      "step": 80770
    },
    {
      "epoch": 0.821685791015625,
      "grad_norm": 13.11707878112793,
      "learning_rate": 3.828734767504791e-07,
      "loss": 3.159,
      "step": 80775
    },
    {
      "epoch": 0.8217366536458334,
      "grad_norm": 12.716404914855957,
      "learning_rate": 3.826608348152391e-07,
      "loss": 3.4763,
      "step": 80780
    },
    {
      "epoch": 0.8217875162760416,
      "grad_norm": 10.998289108276367,
      "learning_rate": 3.824482470516197e-07,
      "loss": 2.9195,
      "step": 80785
    },
    {
      "epoch": 0.82183837890625,
      "grad_norm": 9.000075340270996,
      "learning_rate": 3.8223571346506034e-07,
      "loss": 3.261,
      "step": 80790
    },
    {
      "epoch": 0.8218892415364584,
      "grad_norm": 10.649809837341309,
      "learning_rate": 3.820232340609986e-07,
      "loss": 3.6669,
      "step": 80795
    },
    {
      "epoch": 0.8219401041666666,
      "grad_norm": 16.897676467895508,
      "learning_rate": 3.8181080884486965e-07,
      "loss": 3.5245,
      "step": 80800
    },
    {
      "epoch": 0.821990966796875,
      "grad_norm": 15.520227432250977,
      "learning_rate": 3.8159843782210966e-07,
      "loss": 3.2351,
      "step": 80805
    },
    {
      "epoch": 0.8220418294270834,
      "grad_norm": 12.184919357299805,
      "learning_rate": 3.813861209981512e-07,
      "loss": 3.4723,
      "step": 80810
    },
    {
      "epoch": 0.8220926920572916,
      "grad_norm": 9.80728530883789,
      "learning_rate": 3.8117385837842693e-07,
      "loss": 3.0831,
      "step": 80815
    },
    {
      "epoch": 0.8221435546875,
      "grad_norm": 9.381410598754883,
      "learning_rate": 3.8096164996836686e-07,
      "loss": 3.2771,
      "step": 80820
    },
    {
      "epoch": 0.8221944173177084,
      "grad_norm": 12.89774227142334,
      "learning_rate": 3.807494957734012e-07,
      "loss": 3.1174,
      "step": 80825
    },
    {
      "epoch": 0.8222452799479166,
      "grad_norm": 8.189101219177246,
      "learning_rate": 3.805373957989575e-07,
      "loss": 3.1917,
      "step": 80830
    },
    {
      "epoch": 0.822296142578125,
      "grad_norm": 18.067596435546875,
      "learning_rate": 3.803253500504614e-07,
      "loss": 3.3046,
      "step": 80835
    },
    {
      "epoch": 0.8223470052083334,
      "grad_norm": 13.102996826171875,
      "learning_rate": 3.801133585333389e-07,
      "loss": 3.4559,
      "step": 80840
    },
    {
      "epoch": 0.8223978678385416,
      "grad_norm": 12.395398139953613,
      "learning_rate": 3.7990142125301417e-07,
      "loss": 3.0941,
      "step": 80845
    },
    {
      "epoch": 0.82244873046875,
      "grad_norm": 13.059562683105469,
      "learning_rate": 3.7968953821490904e-07,
      "loss": 3.542,
      "step": 80850
    },
    {
      "epoch": 0.8224995930989584,
      "grad_norm": 13.447836875915527,
      "learning_rate": 3.7947770942444425e-07,
      "loss": 3.4232,
      "step": 80855
    },
    {
      "epoch": 0.8225504557291666,
      "grad_norm": 13.841771125793457,
      "learning_rate": 3.792659348870392e-07,
      "loss": 3.1639,
      "step": 80860
    },
    {
      "epoch": 0.822601318359375,
      "grad_norm": 9.623292922973633,
      "learning_rate": 3.790542146081133e-07,
      "loss": 2.8842,
      "step": 80865
    },
    {
      "epoch": 0.8226521809895834,
      "grad_norm": 16.11116600036621,
      "learning_rate": 3.7884254859308273e-07,
      "loss": 3.4362,
      "step": 80870
    },
    {
      "epoch": 0.8227030436197916,
      "grad_norm": 9.370688438415527,
      "learning_rate": 3.78630936847362e-07,
      "loss": 3.2373,
      "step": 80875
    },
    {
      "epoch": 0.82275390625,
      "grad_norm": 17.88738250732422,
      "learning_rate": 3.784193793763666e-07,
      "loss": 3.5616,
      "step": 80880
    },
    {
      "epoch": 0.8228047688802084,
      "grad_norm": 13.496973991394043,
      "learning_rate": 3.782078761855079e-07,
      "loss": 3.4115,
      "step": 80885
    },
    {
      "epoch": 0.8228556315104166,
      "grad_norm": 9.512796401977539,
      "learning_rate": 3.7799642728019837e-07,
      "loss": 3.0619,
      "step": 80890
    },
    {
      "epoch": 0.822906494140625,
      "grad_norm": 10.060030937194824,
      "learning_rate": 3.7778503266584665e-07,
      "loss": 3.4031,
      "step": 80895
    },
    {
      "epoch": 0.8229573567708334,
      "grad_norm": 13.565192222595215,
      "learning_rate": 3.775736923478626e-07,
      "loss": 3.1807,
      "step": 80900
    },
    {
      "epoch": 0.8230082194010416,
      "grad_norm": 13.113893508911133,
      "learning_rate": 3.773624063316525e-07,
      "loss": 3.4714,
      "step": 80905
    },
    {
      "epoch": 0.82305908203125,
      "grad_norm": 14.605281829833984,
      "learning_rate": 3.771511746226214e-07,
      "loss": 3.3566,
      "step": 80910
    },
    {
      "epoch": 0.8231099446614584,
      "grad_norm": 8.636752128601074,
      "learning_rate": 3.7693999722617445e-07,
      "loss": 3.2208,
      "step": 80915
    },
    {
      "epoch": 0.8231608072916666,
      "grad_norm": 9.907641410827637,
      "learning_rate": 3.7672887414771514e-07,
      "loss": 3.4945,
      "step": 80920
    },
    {
      "epoch": 0.823211669921875,
      "grad_norm": 13.11463451385498,
      "learning_rate": 3.765178053926441e-07,
      "loss": 3.6002,
      "step": 80925
    },
    {
      "epoch": 0.8232625325520834,
      "grad_norm": 14.705850601196289,
      "learning_rate": 3.7630679096636175e-07,
      "loss": 3.4234,
      "step": 80930
    },
    {
      "epoch": 0.8233133951822916,
      "grad_norm": 9.67563533782959,
      "learning_rate": 3.760958308742663e-07,
      "loss": 3.1482,
      "step": 80935
    },
    {
      "epoch": 0.8233642578125,
      "grad_norm": 13.453156471252441,
      "learning_rate": 3.758849251217553e-07,
      "loss": 3.4172,
      "step": 80940
    },
    {
      "epoch": 0.8234151204427084,
      "grad_norm": 13.665534973144531,
      "learning_rate": 3.756740737142259e-07,
      "loss": 3.4066,
      "step": 80945
    },
    {
      "epoch": 0.8234659830729166,
      "grad_norm": 17.34937858581543,
      "learning_rate": 3.7546327665707175e-07,
      "loss": 3.3283,
      "step": 80950
    },
    {
      "epoch": 0.823516845703125,
      "grad_norm": 11.575714111328125,
      "learning_rate": 3.7525253395568536e-07,
      "loss": 3.3866,
      "step": 80955
    },
    {
      "epoch": 0.8235677083333334,
      "grad_norm": 8.186203956604004,
      "learning_rate": 3.7504184561546004e-07,
      "loss": 3.1128,
      "step": 80960
    },
    {
      "epoch": 0.8236185709635416,
      "grad_norm": 11.50184440612793,
      "learning_rate": 3.748312116417846e-07,
      "loss": 3.1947,
      "step": 80965
    },
    {
      "epoch": 0.82366943359375,
      "grad_norm": 9.66376781463623,
      "learning_rate": 3.7462063204004917e-07,
      "loss": 3.4334,
      "step": 80970
    },
    {
      "epoch": 0.8237202962239584,
      "grad_norm": 15.963284492492676,
      "learning_rate": 3.744101068156417e-07,
      "loss": 3.4715,
      "step": 80975
    },
    {
      "epoch": 0.8237711588541666,
      "grad_norm": 8.475472450256348,
      "learning_rate": 3.741996359739475e-07,
      "loss": 3.3516,
      "step": 80980
    },
    {
      "epoch": 0.823822021484375,
      "grad_norm": 14.08765983581543,
      "learning_rate": 3.7398921952035203e-07,
      "loss": 3.1434,
      "step": 80985
    },
    {
      "epoch": 0.8238728841145834,
      "grad_norm": 10.457538604736328,
      "learning_rate": 3.737788574602377e-07,
      "loss": 3.2368,
      "step": 80990
    },
    {
      "epoch": 0.8239237467447916,
      "grad_norm": 9.279627799987793,
      "learning_rate": 3.7356854979898745e-07,
      "loss": 3.2903,
      "step": 80995
    },
    {
      "epoch": 0.823974609375,
      "grad_norm": 8.412238121032715,
      "learning_rate": 3.733582965419824e-07,
      "loss": 3.2409,
      "step": 81000
    },
    {
      "epoch": 0.8240254720052084,
      "grad_norm": 12.581768989562988,
      "learning_rate": 3.731480976946011e-07,
      "loss": 3.2264,
      "step": 81005
    },
    {
      "epoch": 0.8240763346354166,
      "grad_norm": 9.827341079711914,
      "learning_rate": 3.7293795326222127e-07,
      "loss": 3.1994,
      "step": 81010
    },
    {
      "epoch": 0.824127197265625,
      "grad_norm": 10.232128143310547,
      "learning_rate": 3.727278632502196e-07,
      "loss": 3.208,
      "step": 81015
    },
    {
      "epoch": 0.8241780598958334,
      "grad_norm": 27.526899337768555,
      "learning_rate": 3.725178276639721e-07,
      "loss": 3.2577,
      "step": 81020
    },
    {
      "epoch": 0.8242289225260416,
      "grad_norm": 12.687592506408691,
      "learning_rate": 3.7230784650885144e-07,
      "loss": 3.5804,
      "step": 81025
    },
    {
      "epoch": 0.82427978515625,
      "grad_norm": 17.577566146850586,
      "learning_rate": 3.7209791979022967e-07,
      "loss": 3.3725,
      "step": 81030
    },
    {
      "epoch": 0.8243306477864584,
      "grad_norm": 8.21080493927002,
      "learning_rate": 3.7188804751347893e-07,
      "loss": 3.6424,
      "step": 81035
    },
    {
      "epoch": 0.8243815104166666,
      "grad_norm": 13.241853713989258,
      "learning_rate": 3.716782296839677e-07,
      "loss": 3.5274,
      "step": 81040
    },
    {
      "epoch": 0.824432373046875,
      "grad_norm": 9.96329116821289,
      "learning_rate": 3.714684663070647e-07,
      "loss": 3.1033,
      "step": 81045
    },
    {
      "epoch": 0.8244832356770834,
      "grad_norm": 9.271282196044922,
      "learning_rate": 3.71258757388136e-07,
      "loss": 3.7882,
      "step": 81050
    },
    {
      "epoch": 0.8245340983072916,
      "grad_norm": 13.86997127532959,
      "learning_rate": 3.71049102932548e-07,
      "loss": 3.495,
      "step": 81055
    },
    {
      "epoch": 0.8245849609375,
      "grad_norm": 8.317121505737305,
      "learning_rate": 3.708395029456643e-07,
      "loss": 3.3285,
      "step": 81060
    },
    {
      "epoch": 0.8246358235677084,
      "grad_norm": 15.157517433166504,
      "learning_rate": 3.7062995743284647e-07,
      "loss": 3.3673,
      "step": 81065
    },
    {
      "epoch": 0.8246866861979166,
      "grad_norm": 9.469864845275879,
      "learning_rate": 3.704204663994565e-07,
      "loss": 3.3481,
      "step": 81070
    },
    {
      "epoch": 0.824737548828125,
      "grad_norm": 11.107138633728027,
      "learning_rate": 3.702110298508546e-07,
      "loss": 3.301,
      "step": 81075
    },
    {
      "epoch": 0.8247884114583334,
      "grad_norm": 11.196867942810059,
      "learning_rate": 3.70001647792399e-07,
      "loss": 3.3571,
      "step": 81080
    },
    {
      "epoch": 0.8248392740885416,
      "grad_norm": 17.41468620300293,
      "learning_rate": 3.697923202294457e-07,
      "loss": 3.4212,
      "step": 81085
    },
    {
      "epoch": 0.82489013671875,
      "grad_norm": 14.888216018676758,
      "learning_rate": 3.6958304716735115e-07,
      "loss": 3.0231,
      "step": 81090
    },
    {
      "epoch": 0.8249409993489584,
      "grad_norm": 15.072734832763672,
      "learning_rate": 3.6937382861146927e-07,
      "loss": 3.4232,
      "step": 81095
    },
    {
      "epoch": 0.8249918619791666,
      "grad_norm": 17.388551712036133,
      "learning_rate": 3.6916466456715344e-07,
      "loss": 3.3821,
      "step": 81100
    },
    {
      "epoch": 0.825042724609375,
      "grad_norm": 12.327610969543457,
      "learning_rate": 3.6895555503975386e-07,
      "loss": 3.184,
      "step": 81105
    },
    {
      "epoch": 0.8250935872395834,
      "grad_norm": 15.96401596069336,
      "learning_rate": 3.6874650003462186e-07,
      "loss": 3.3028,
      "step": 81110
    },
    {
      "epoch": 0.8251444498697916,
      "grad_norm": 13.771220207214355,
      "learning_rate": 3.685374995571056e-07,
      "loss": 3.6253,
      "step": 81115
    },
    {
      "epoch": 0.8251953125,
      "grad_norm": 7.984499931335449,
      "learning_rate": 3.6832855361255163e-07,
      "loss": 3.4847,
      "step": 81120
    },
    {
      "epoch": 0.8252461751302084,
      "grad_norm": 14.82887077331543,
      "learning_rate": 3.681196622063063e-07,
      "loss": 3.193,
      "step": 81125
    },
    {
      "epoch": 0.8252970377604166,
      "grad_norm": 9.447796821594238,
      "learning_rate": 3.6791082534371495e-07,
      "loss": 3.6481,
      "step": 81130
    },
    {
      "epoch": 0.825347900390625,
      "grad_norm": 9.615379333496094,
      "learning_rate": 3.6770204303011953e-07,
      "loss": 3.0771,
      "step": 81135
    },
    {
      "epoch": 0.8253987630208334,
      "grad_norm": 10.152069091796875,
      "learning_rate": 3.674933152708618e-07,
      "loss": 3.3728,
      "step": 81140
    },
    {
      "epoch": 0.8254496256510416,
      "grad_norm": 7.463944911956787,
      "learning_rate": 3.672846420712817e-07,
      "loss": 3.2594,
      "step": 81145
    },
    {
      "epoch": 0.82550048828125,
      "grad_norm": 20.239410400390625,
      "learning_rate": 3.670760234367185e-07,
      "loss": 3.9383,
      "step": 81150
    },
    {
      "epoch": 0.8255513509114584,
      "grad_norm": 16.32979393005371,
      "learning_rate": 3.6686745937251053e-07,
      "loss": 3.2126,
      "step": 81155
    },
    {
      "epoch": 0.8256022135416666,
      "grad_norm": 10.634185791015625,
      "learning_rate": 3.6665894988399204e-07,
      "loss": 3.1509,
      "step": 81160
    },
    {
      "epoch": 0.825653076171875,
      "grad_norm": 8.721284866333008,
      "learning_rate": 3.664504949764994e-07,
      "loss": 3.1079,
      "step": 81165
    },
    {
      "epoch": 0.8257039388020834,
      "grad_norm": 7.4027276039123535,
      "learning_rate": 3.6624209465536464e-07,
      "loss": 3.3191,
      "step": 81170
    },
    {
      "epoch": 0.8257548014322916,
      "grad_norm": 8.177949905395508,
      "learning_rate": 3.660337489259205e-07,
      "loss": 3.2868,
      "step": 81175
    },
    {
      "epoch": 0.8258056640625,
      "grad_norm": 13.622292518615723,
      "learning_rate": 3.658254577934967e-07,
      "loss": 2.9148,
      "step": 81180
    },
    {
      "epoch": 0.8258565266927084,
      "grad_norm": 15.886625289916992,
      "learning_rate": 3.656172212634232e-07,
      "loss": 3.1127,
      "step": 81185
    },
    {
      "epoch": 0.8259073893229166,
      "grad_norm": 12.290112495422363,
      "learning_rate": 3.6540903934102727e-07,
      "loss": 3.1256,
      "step": 81190
    },
    {
      "epoch": 0.825958251953125,
      "grad_norm": 15.316094398498535,
      "learning_rate": 3.652009120316344e-07,
      "loss": 3.6454,
      "step": 81195
    },
    {
      "epoch": 0.8260091145833334,
      "grad_norm": 13.505681037902832,
      "learning_rate": 3.649928393405702e-07,
      "loss": 3.288,
      "step": 81200
    },
    {
      "epoch": 0.8260599772135416,
      "grad_norm": 10.823572158813477,
      "learning_rate": 3.647848212731589e-07,
      "loss": 3.7845,
      "step": 81205
    },
    {
      "epoch": 0.82611083984375,
      "grad_norm": 9.712890625,
      "learning_rate": 3.645768578347217e-07,
      "loss": 3.23,
      "step": 81210
    },
    {
      "epoch": 0.8261617024739584,
      "grad_norm": 12.620218276977539,
      "learning_rate": 3.643689490305796e-07,
      "loss": 3.3941,
      "step": 81215
    },
    {
      "epoch": 0.8262125651041666,
      "grad_norm": 14.910798072814941,
      "learning_rate": 3.6416109486605095e-07,
      "loss": 3.096,
      "step": 81220
    },
    {
      "epoch": 0.826263427734375,
      "grad_norm": 9.891182899475098,
      "learning_rate": 3.6395329534645444e-07,
      "loss": 3.4,
      "step": 81225
    },
    {
      "epoch": 0.8263142903645834,
      "grad_norm": 8.116037368774414,
      "learning_rate": 3.637455504771073e-07,
      "loss": 3.2759,
      "step": 81230
    },
    {
      "epoch": 0.8263651529947916,
      "grad_norm": 10.094236373901367,
      "learning_rate": 3.635378602633238e-07,
      "loss": 3.9303,
      "step": 81235
    },
    {
      "epoch": 0.826416015625,
      "grad_norm": 14.345175743103027,
      "learning_rate": 3.633302247104173e-07,
      "loss": 3.1749,
      "step": 81240
    },
    {
      "epoch": 0.8264668782552084,
      "grad_norm": 17.376415252685547,
      "learning_rate": 3.631226438237012e-07,
      "loss": 3.2547,
      "step": 81245
    },
    {
      "epoch": 0.8265177408854166,
      "grad_norm": 15.243380546569824,
      "learning_rate": 3.6291511760848496e-07,
      "loss": 3.6868,
      "step": 81250
    },
    {
      "epoch": 0.826568603515625,
      "grad_norm": 11.07956600189209,
      "learning_rate": 3.6270764607007904e-07,
      "loss": 3.4299,
      "step": 81255
    },
    {
      "epoch": 0.8266194661458334,
      "grad_norm": 10.591047286987305,
      "learning_rate": 3.625002292137919e-07,
      "loss": 3.2476,
      "step": 81260
    },
    {
      "epoch": 0.8266703287760416,
      "grad_norm": 19.205686569213867,
      "learning_rate": 3.6229286704492967e-07,
      "loss": 3.316,
      "step": 81265
    },
    {
      "epoch": 0.82672119140625,
      "grad_norm": 9.71511173248291,
      "learning_rate": 3.620855595687978e-07,
      "loss": 3.5159,
      "step": 81270
    },
    {
      "epoch": 0.8267720540364584,
      "grad_norm": 9.56340217590332,
      "learning_rate": 3.6187830679069976e-07,
      "loss": 3.3286,
      "step": 81275
    },
    {
      "epoch": 0.8268229166666666,
      "grad_norm": 16.750654220581055,
      "learning_rate": 3.616711087159383e-07,
      "loss": 3.3161,
      "step": 81280
    },
    {
      "epoch": 0.826873779296875,
      "grad_norm": 7.3144001960754395,
      "learning_rate": 3.61463965349815e-07,
      "loss": 3.5281,
      "step": 81285
    },
    {
      "epoch": 0.8269246419270834,
      "grad_norm": 9.85859203338623,
      "learning_rate": 3.612568766976296e-07,
      "loss": 3.8354,
      "step": 81290
    },
    {
      "epoch": 0.8269755045572916,
      "grad_norm": 10.094393730163574,
      "learning_rate": 3.610498427646794e-07,
      "loss": 3.3545,
      "step": 81295
    },
    {
      "epoch": 0.8270263671875,
      "grad_norm": 9.70475959777832,
      "learning_rate": 3.6084286355626274e-07,
      "loss": 3.2521,
      "step": 81300
    },
    {
      "epoch": 0.8270772298177084,
      "grad_norm": 15.505373001098633,
      "learning_rate": 3.606359390776734e-07,
      "loss": 3.2219,
      "step": 81305
    },
    {
      "epoch": 0.8271280924479166,
      "grad_norm": 7.791288375854492,
      "learning_rate": 3.6042906933420745e-07,
      "loss": 2.8279,
      "step": 81310
    },
    {
      "epoch": 0.827178955078125,
      "grad_norm": 14.234278678894043,
      "learning_rate": 3.6022225433115595e-07,
      "loss": 3.6315,
      "step": 81315
    },
    {
      "epoch": 0.8272298177083334,
      "grad_norm": 7.739132404327393,
      "learning_rate": 3.600154940738113e-07,
      "loss": 3.2901,
      "step": 81320
    },
    {
      "epoch": 0.8272806803385416,
      "grad_norm": 13.4205904006958,
      "learning_rate": 3.598087885674631e-07,
      "loss": 3.1889,
      "step": 81325
    },
    {
      "epoch": 0.82733154296875,
      "grad_norm": 13.76759147644043,
      "learning_rate": 3.5960213781739917e-07,
      "loss": 3.3293,
      "step": 81330
    },
    {
      "epoch": 0.8273824055989584,
      "grad_norm": 10.153120040893555,
      "learning_rate": 3.5939554182890716e-07,
      "loss": 3.208,
      "step": 81335
    },
    {
      "epoch": 0.8274332682291666,
      "grad_norm": 10.668922424316406,
      "learning_rate": 3.5918900060727376e-07,
      "loss": 3.1456,
      "step": 81340
    },
    {
      "epoch": 0.827484130859375,
      "grad_norm": 10.668487548828125,
      "learning_rate": 3.5898251415778187e-07,
      "loss": 3.4589,
      "step": 81345
    },
    {
      "epoch": 0.8275349934895834,
      "grad_norm": 13.005240440368652,
      "learning_rate": 3.587760824857145e-07,
      "loss": 2.8548,
      "step": 81350
    },
    {
      "epoch": 0.8275858561197916,
      "grad_norm": 13.068918228149414,
      "learning_rate": 3.5856970559635355e-07,
      "loss": 3.222,
      "step": 81355
    },
    {
      "epoch": 0.82763671875,
      "grad_norm": 9.316664695739746,
      "learning_rate": 3.5836338349497983e-07,
      "loss": 3.4893,
      "step": 81360
    },
    {
      "epoch": 0.8276875813802084,
      "grad_norm": 7.974001884460449,
      "learning_rate": 3.58157116186871e-07,
      "loss": 3.3725,
      "step": 81365
    },
    {
      "epoch": 0.8277384440104166,
      "grad_norm": 11.986618995666504,
      "learning_rate": 3.579509036773043e-07,
      "loss": 3.3193,
      "step": 81370
    },
    {
      "epoch": 0.827789306640625,
      "grad_norm": 14.938953399658203,
      "learning_rate": 3.577447459715566e-07,
      "loss": 3.2821,
      "step": 81375
    },
    {
      "epoch": 0.8278401692708334,
      "grad_norm": 16.224943161010742,
      "learning_rate": 3.575386430749012e-07,
      "loss": 3.3977,
      "step": 81380
    },
    {
      "epoch": 0.8278910319010416,
      "grad_norm": 11.836395263671875,
      "learning_rate": 3.5733259499261236e-07,
      "loss": 3.4964,
      "step": 81385
    },
    {
      "epoch": 0.82794189453125,
      "grad_norm": 8.097487449645996,
      "learning_rate": 3.5712660172996074e-07,
      "loss": 3.3076,
      "step": 81390
    },
    {
      "epoch": 0.8279927571614584,
      "grad_norm": 11.223922729492188,
      "learning_rate": 3.569206632922176e-07,
      "loss": 3.2277,
      "step": 81395
    },
    {
      "epoch": 0.8280436197916666,
      "grad_norm": 11.929049491882324,
      "learning_rate": 3.567147796846512e-07,
      "loss": 3.4273,
      "step": 81400
    },
    {
      "epoch": 0.828094482421875,
      "grad_norm": 8.872072219848633,
      "learning_rate": 3.5650895091252856e-07,
      "loss": 3.1472,
      "step": 81405
    },
    {
      "epoch": 0.8281453450520834,
      "grad_norm": 14.635814666748047,
      "learning_rate": 3.563031769811165e-07,
      "loss": 3.5105,
      "step": 81410
    },
    {
      "epoch": 0.8281962076822916,
      "grad_norm": 12.586482048034668,
      "learning_rate": 3.5609745789567996e-07,
      "loss": 3.1589,
      "step": 81415
    },
    {
      "epoch": 0.8282470703125,
      "grad_norm": 15.408994674682617,
      "learning_rate": 3.5589179366148195e-07,
      "loss": 3.3757,
      "step": 81420
    },
    {
      "epoch": 0.8282979329427084,
      "grad_norm": 8.373656272888184,
      "learning_rate": 3.5568618428378415e-07,
      "loss": 3.2097,
      "step": 81425
    },
    {
      "epoch": 0.8283487955729166,
      "grad_norm": 12.243661880493164,
      "learning_rate": 3.554806297678465e-07,
      "loss": 3.3773,
      "step": 81430
    },
    {
      "epoch": 0.828399658203125,
      "grad_norm": 15.660725593566895,
      "learning_rate": 3.552751301189286e-07,
      "loss": 3.6567,
      "step": 81435
    },
    {
      "epoch": 0.8284505208333334,
      "grad_norm": 15.802828788757324,
      "learning_rate": 3.5506968534228886e-07,
      "loss": 3.3132,
      "step": 81440
    },
    {
      "epoch": 0.8285013834635416,
      "grad_norm": 12.42456340789795,
      "learning_rate": 3.548642954431822e-07,
      "loss": 3.5158,
      "step": 81445
    },
    {
      "epoch": 0.82855224609375,
      "grad_norm": 14.236919403076172,
      "learning_rate": 3.546589604268644e-07,
      "loss": 3.9095,
      "step": 81450
    },
    {
      "epoch": 0.8286031087239584,
      "grad_norm": 11.435016632080078,
      "learning_rate": 3.544536802985887e-07,
      "loss": 3.2436,
      "step": 81455
    },
    {
      "epoch": 0.8286539713541666,
      "grad_norm": 10.88861083984375,
      "learning_rate": 3.542484550636066e-07,
      "loss": 3.4391,
      "step": 81460
    },
    {
      "epoch": 0.828704833984375,
      "grad_norm": 9.29944896697998,
      "learning_rate": 3.540432847271694e-07,
      "loss": 3.2777,
      "step": 81465
    },
    {
      "epoch": 0.8287556966145834,
      "grad_norm": 13.211068153381348,
      "learning_rate": 3.5383816929452614e-07,
      "loss": 3.3127,
      "step": 81470
    },
    {
      "epoch": 0.8288065592447916,
      "grad_norm": 11.630187034606934,
      "learning_rate": 3.5363310877092494e-07,
      "loss": 3.1737,
      "step": 81475
    },
    {
      "epoch": 0.828857421875,
      "grad_norm": 13.085993766784668,
      "learning_rate": 3.5342810316161183e-07,
      "loss": 3.4868,
      "step": 81480
    },
    {
      "epoch": 0.8289082845052084,
      "grad_norm": 8.661873817443848,
      "learning_rate": 3.5322315247183074e-07,
      "loss": 3.5383,
      "step": 81485
    },
    {
      "epoch": 0.8289591471354166,
      "grad_norm": 12.713909149169922,
      "learning_rate": 3.530182567068277e-07,
      "loss": 3.2469,
      "step": 81490
    },
    {
      "epoch": 0.829010009765625,
      "grad_norm": 12.60568618774414,
      "learning_rate": 3.528134158718438e-07,
      "loss": 3.2449,
      "step": 81495
    },
    {
      "epoch": 0.8290608723958334,
      "grad_norm": 15.496371269226074,
      "learning_rate": 3.526086299721193e-07,
      "loss": 3.4259,
      "step": 81500
    },
    {
      "epoch": 0.8291117350260416,
      "grad_norm": 8.93329906463623,
      "learning_rate": 3.524038990128939e-07,
      "loss": 3.0466,
      "step": 81505
    },
    {
      "epoch": 0.82916259765625,
      "grad_norm": 16.786598205566406,
      "learning_rate": 3.5219922299940536e-07,
      "loss": 3.6377,
      "step": 81510
    },
    {
      "epoch": 0.8292134602864584,
      "grad_norm": 11.137445449829102,
      "learning_rate": 3.5199460193689137e-07,
      "loss": 2.959,
      "step": 81515
    },
    {
      "epoch": 0.8292643229166666,
      "grad_norm": 17.52461814880371,
      "learning_rate": 3.517900358305862e-07,
      "loss": 3.3231,
      "step": 81520
    },
    {
      "epoch": 0.829315185546875,
      "grad_norm": 10.99956226348877,
      "learning_rate": 3.515855246857233e-07,
      "loss": 2.9984,
      "step": 81525
    },
    {
      "epoch": 0.8293660481770834,
      "grad_norm": 11.812445640563965,
      "learning_rate": 3.5138106850753613e-07,
      "loss": 3.4031,
      "step": 81530
    },
    {
      "epoch": 0.8294169108072916,
      "grad_norm": 12.35213565826416,
      "learning_rate": 3.5117666730125455e-07,
      "loss": 3.3872,
      "step": 81535
    },
    {
      "epoch": 0.8294677734375,
      "grad_norm": 14.233818054199219,
      "learning_rate": 3.5097232107210916e-07,
      "loss": 3.4455,
      "step": 81540
    },
    {
      "epoch": 0.8295186360677084,
      "grad_norm": 8.840458869934082,
      "learning_rate": 3.507680298253269e-07,
      "loss": 3.3539,
      "step": 81545
    },
    {
      "epoch": 0.8295694986979166,
      "grad_norm": 12.002241134643555,
      "learning_rate": 3.5056379356613573e-07,
      "loss": 3.0824,
      "step": 81550
    },
    {
      "epoch": 0.829620361328125,
      "grad_norm": 13.244787216186523,
      "learning_rate": 3.5035961229976045e-07,
      "loss": 3.6522,
      "step": 81555
    },
    {
      "epoch": 0.8296712239583334,
      "grad_norm": 9.030220985412598,
      "learning_rate": 3.5015548603142433e-07,
      "loss": 3.3114,
      "step": 81560
    },
    {
      "epoch": 0.8297220865885416,
      "grad_norm": 8.869832992553711,
      "learning_rate": 3.499514147663505e-07,
      "loss": 3.1669,
      "step": 81565
    },
    {
      "epoch": 0.82977294921875,
      "grad_norm": 12.157954216003418,
      "learning_rate": 3.4974739850976074e-07,
      "loss": 3.5634,
      "step": 81570
    },
    {
      "epoch": 0.8298238118489584,
      "grad_norm": 14.200581550598145,
      "learning_rate": 3.495434372668743e-07,
      "loss": 3.2098,
      "step": 81575
    },
    {
      "epoch": 0.8298746744791666,
      "grad_norm": 9.806853294372559,
      "learning_rate": 3.493395310429085e-07,
      "loss": 3.3559,
      "step": 81580
    },
    {
      "epoch": 0.829925537109375,
      "grad_norm": 12.339767456054688,
      "learning_rate": 3.491356798430817e-07,
      "loss": 3.3705,
      "step": 81585
    },
    {
      "epoch": 0.8299763997395834,
      "grad_norm": 15.809764862060547,
      "learning_rate": 3.4893188367260826e-07,
      "loss": 3.6926,
      "step": 81590
    },
    {
      "epoch": 0.8300272623697916,
      "grad_norm": 14.739584922790527,
      "learning_rate": 3.487281425367031e-07,
      "loss": 3.2144,
      "step": 81595
    },
    {
      "epoch": 0.830078125,
      "grad_norm": 6.578415393829346,
      "learning_rate": 3.485244564405779e-07,
      "loss": 3.2765,
      "step": 81600
    },
    {
      "epoch": 0.8301289876302084,
      "grad_norm": 7.773172378540039,
      "learning_rate": 3.4832082538944543e-07,
      "loss": 3.7121,
      "step": 81605
    },
    {
      "epoch": 0.8301798502604166,
      "grad_norm": 13.908849716186523,
      "learning_rate": 3.481172493885143e-07,
      "loss": 3.075,
      "step": 81610
    },
    {
      "epoch": 0.830230712890625,
      "grad_norm": 13.709275245666504,
      "learning_rate": 3.4791372844299285e-07,
      "loss": 3.3412,
      "step": 81615
    },
    {
      "epoch": 0.8302815755208334,
      "grad_norm": 10.29182243347168,
      "learning_rate": 3.477102625580886e-07,
      "loss": 3.5131,
      "step": 81620
    },
    {
      "epoch": 0.8303324381510416,
      "grad_norm": 9.028854370117188,
      "learning_rate": 3.4750685173900764e-07,
      "loss": 3.4103,
      "step": 81625
    },
    {
      "epoch": 0.83038330078125,
      "grad_norm": 16.737489700317383,
      "learning_rate": 3.473034959909538e-07,
      "loss": 3.5828,
      "step": 81630
    },
    {
      "epoch": 0.8304341634114584,
      "grad_norm": 10.40755558013916,
      "learning_rate": 3.471001953191294e-07,
      "loss": 3.1235,
      "step": 81635
    },
    {
      "epoch": 0.8304850260416666,
      "grad_norm": 9.349138259887695,
      "learning_rate": 3.4689694972873554e-07,
      "loss": 3.6267,
      "step": 81640
    },
    {
      "epoch": 0.830535888671875,
      "grad_norm": 15.301726341247559,
      "learning_rate": 3.466937592249739e-07,
      "loss": 3.4729,
      "step": 81645
    },
    {
      "epoch": 0.8305867513020834,
      "grad_norm": 12.60606575012207,
      "learning_rate": 3.4649062381304195e-07,
      "loss": 3.5103,
      "step": 81650
    },
    {
      "epoch": 0.8306376139322916,
      "grad_norm": 6.902255535125732,
      "learning_rate": 3.4628754349813636e-07,
      "loss": 3.4327,
      "step": 81655
    },
    {
      "epoch": 0.8306884765625,
      "grad_norm": 12.60377311706543,
      "learning_rate": 3.460845182854539e-07,
      "loss": 3.351,
      "step": 81660
    },
    {
      "epoch": 0.8307393391927084,
      "grad_norm": 8.240616798400879,
      "learning_rate": 3.4588154818018834e-07,
      "loss": 3.3058,
      "step": 81665
    },
    {
      "epoch": 0.8307902018229166,
      "grad_norm": 11.642782211303711,
      "learning_rate": 3.4567863318753307e-07,
      "loss": 3.171,
      "step": 81670
    },
    {
      "epoch": 0.830841064453125,
      "grad_norm": 10.349153518676758,
      "learning_rate": 3.4547577331267873e-07,
      "loss": 3.2889,
      "step": 81675
    },
    {
      "epoch": 0.8308919270833334,
      "grad_norm": 7.751547336578369,
      "learning_rate": 3.4527296856081664e-07,
      "loss": 3.5357,
      "step": 81680
    },
    {
      "epoch": 0.8309427897135416,
      "grad_norm": 11.443355560302734,
      "learning_rate": 3.4507021893713523e-07,
      "loss": 3.2525,
      "step": 81685
    },
    {
      "epoch": 0.83099365234375,
      "grad_norm": 9.293235778808594,
      "learning_rate": 3.4486752444682053e-07,
      "loss": 3.2354,
      "step": 81690
    },
    {
      "epoch": 0.8310445149739584,
      "grad_norm": 13.506486892700195,
      "learning_rate": 3.446648850950596e-07,
      "loss": 3.6791,
      "step": 81695
    },
    {
      "epoch": 0.8310953776041666,
      "grad_norm": 8.528145790100098,
      "learning_rate": 3.4446230088703734e-07,
      "loss": 3.48,
      "step": 81700
    },
    {
      "epoch": 0.831146240234375,
      "grad_norm": 7.487433433532715,
      "learning_rate": 3.442597718279361e-07,
      "loss": 3.3777,
      "step": 81705
    },
    {
      "epoch": 0.8311971028645834,
      "grad_norm": 10.32982349395752,
      "learning_rate": 3.440572979229376e-07,
      "loss": 3.3119,
      "step": 81710
    },
    {
      "epoch": 0.8312479654947916,
      "grad_norm": 10.598235130310059,
      "learning_rate": 3.4385487917722144e-07,
      "loss": 3.519,
      "step": 81715
    },
    {
      "epoch": 0.831298828125,
      "grad_norm": 13.667901992797852,
      "learning_rate": 3.4365251559596746e-07,
      "loss": 3.3822,
      "step": 81720
    },
    {
      "epoch": 0.8313496907552084,
      "grad_norm": 16.000396728515625,
      "learning_rate": 3.434502071843532e-07,
      "loss": 3.3401,
      "step": 81725
    },
    {
      "epoch": 0.8314005533854166,
      "grad_norm": 11.901004791259766,
      "learning_rate": 3.4324795394755415e-07,
      "loss": 3.1247,
      "step": 81730
    },
    {
      "epoch": 0.831451416015625,
      "grad_norm": 10.683159828186035,
      "learning_rate": 3.4304575589074435e-07,
      "loss": 3.8819,
      "step": 81735
    },
    {
      "epoch": 0.8315022786458334,
      "grad_norm": 9.5502290725708,
      "learning_rate": 3.428436130190982e-07,
      "loss": 3.1455,
      "step": 81740
    },
    {
      "epoch": 0.8315531412760416,
      "grad_norm": 13.432830810546875,
      "learning_rate": 3.4264152533778655e-07,
      "loss": 3.2658,
      "step": 81745
    },
    {
      "epoch": 0.83160400390625,
      "grad_norm": 15.859672546386719,
      "learning_rate": 3.4243949285198004e-07,
      "loss": 3.0781,
      "step": 81750
    },
    {
      "epoch": 0.8316548665364584,
      "grad_norm": 12.149288177490234,
      "learning_rate": 3.4223751556684807e-07,
      "loss": 3.6371,
      "step": 81755
    },
    {
      "epoch": 0.8317057291666666,
      "grad_norm": 11.400646209716797,
      "learning_rate": 3.4203559348755826e-07,
      "loss": 3.6175,
      "step": 81760
    },
    {
      "epoch": 0.831756591796875,
      "grad_norm": 12.680248260498047,
      "learning_rate": 3.418337266192759e-07,
      "loss": 3.3646,
      "step": 81765
    },
    {
      "epoch": 0.8318074544270834,
      "grad_norm": 13.315074920654297,
      "learning_rate": 3.416319149671657e-07,
      "loss": 3.2721,
      "step": 81770
    },
    {
      "epoch": 0.8318583170572916,
      "grad_norm": 8.332613945007324,
      "learning_rate": 3.4143015853639134e-07,
      "loss": 3.8566,
      "step": 81775
    },
    {
      "epoch": 0.8319091796875,
      "grad_norm": 10.389419555664062,
      "learning_rate": 3.412284573321153e-07,
      "loss": 3.3517,
      "step": 81780
    },
    {
      "epoch": 0.8319600423177084,
      "grad_norm": 10.17501449584961,
      "learning_rate": 3.410268113594975e-07,
      "loss": 3.6708,
      "step": 81785
    },
    {
      "epoch": 0.8320109049479166,
      "grad_norm": 14.361525535583496,
      "learning_rate": 3.4082522062369645e-07,
      "loss": 3.9283,
      "step": 81790
    },
    {
      "epoch": 0.832061767578125,
      "grad_norm": 9.244938850402832,
      "learning_rate": 3.406236851298703e-07,
      "loss": 3.2875,
      "step": 81795
    },
    {
      "epoch": 0.8321126302083334,
      "grad_norm": 17.720565795898438,
      "learning_rate": 3.40422204883176e-07,
      "loss": 3.5046,
      "step": 81800
    },
    {
      "epoch": 0.8321634928385416,
      "grad_norm": 9.65694808959961,
      "learning_rate": 3.4022077988876746e-07,
      "loss": 3.2818,
      "step": 81805
    },
    {
      "epoch": 0.83221435546875,
      "grad_norm": 15.131839752197266,
      "learning_rate": 3.400194101517981e-07,
      "loss": 3.471,
      "step": 81810
    },
    {
      "epoch": 0.8322652180989584,
      "grad_norm": 15.426616668701172,
      "learning_rate": 3.3981809567742054e-07,
      "loss": 3.2494,
      "step": 81815
    },
    {
      "epoch": 0.8323160807291666,
      "grad_norm": 12.537252426147461,
      "learning_rate": 3.3961683647078444e-07,
      "loss": 3.2033,
      "step": 81820
    },
    {
      "epoch": 0.832366943359375,
      "grad_norm": 18.45199203491211,
      "learning_rate": 3.394156325370401e-07,
      "loss": 3.3196,
      "step": 81825
    },
    {
      "epoch": 0.8324178059895834,
      "grad_norm": 10.200485229492188,
      "learning_rate": 3.392144838813341e-07,
      "loss": 3.448,
      "step": 81830
    },
    {
      "epoch": 0.8324686686197916,
      "grad_norm": 10.197514533996582,
      "learning_rate": 3.3901339050881415e-07,
      "loss": 3.6044,
      "step": 81835
    },
    {
      "epoch": 0.83251953125,
      "grad_norm": 10.978058815002441,
      "learning_rate": 3.38812352424624e-07,
      "loss": 3.9155,
      "step": 81840
    },
    {
      "epoch": 0.8325703938802084,
      "grad_norm": 20.063888549804688,
      "learning_rate": 3.3861136963390715e-07,
      "loss": 3.902,
      "step": 81845
    },
    {
      "epoch": 0.8326212565104166,
      "grad_norm": 13.886176109313965,
      "learning_rate": 3.384104421418063e-07,
      "loss": 3.1731,
      "step": 81850
    },
    {
      "epoch": 0.832672119140625,
      "grad_norm": 8.191770553588867,
      "learning_rate": 3.382095699534624e-07,
      "loss": 3.0654,
      "step": 81855
    },
    {
      "epoch": 0.8327229817708334,
      "grad_norm": 12.459845542907715,
      "learning_rate": 3.380087530740142e-07,
      "loss": 4.0921,
      "step": 81860
    },
    {
      "epoch": 0.8327738444010416,
      "grad_norm": 12.545857429504395,
      "learning_rate": 3.378079915085988e-07,
      "loss": 3.1966,
      "step": 81865
    },
    {
      "epoch": 0.83282470703125,
      "grad_norm": 16.804519653320312,
      "learning_rate": 3.376072852623544e-07,
      "loss": 3.0937,
      "step": 81870
    },
    {
      "epoch": 0.8328755696614584,
      "grad_norm": 13.775388717651367,
      "learning_rate": 3.374066343404142e-07,
      "loss": 2.9983,
      "step": 81875
    },
    {
      "epoch": 0.8329264322916666,
      "grad_norm": 17.29413604736328,
      "learning_rate": 3.3720603874791343e-07,
      "loss": 3.4973,
      "step": 81880
    },
    {
      "epoch": 0.832977294921875,
      "grad_norm": 9.611454963684082,
      "learning_rate": 3.370054984899829e-07,
      "loss": 3.333,
      "step": 81885
    },
    {
      "epoch": 0.8330281575520834,
      "grad_norm": 13.825835227966309,
      "learning_rate": 3.368050135717546e-07,
      "loss": 3.1104,
      "step": 81890
    },
    {
      "epoch": 0.8330790201822916,
      "grad_norm": 10.248576164245605,
      "learning_rate": 3.366045839983573e-07,
      "loss": 3.4458,
      "step": 81895
    },
    {
      "epoch": 0.8331298828125,
      "grad_norm": 9.065658569335938,
      "learning_rate": 3.3640420977491846e-07,
      "loss": 2.9167,
      "step": 81900
    },
    {
      "epoch": 0.8331807454427084,
      "grad_norm": 10.88503646850586,
      "learning_rate": 3.3620389090656484e-07,
      "loss": 3.2034,
      "step": 81905
    },
    {
      "epoch": 0.8332316080729166,
      "grad_norm": 9.16987133026123,
      "learning_rate": 3.360036273984224e-07,
      "loss": 3.2866,
      "step": 81910
    },
    {
      "epoch": 0.833282470703125,
      "grad_norm": 17.341602325439453,
      "learning_rate": 3.3580341925561447e-07,
      "loss": 3.5577,
      "step": 81915
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 12.902583122253418,
      "learning_rate": 3.3560326648326285e-07,
      "loss": 3.5347,
      "step": 81920
    },
    {
      "epoch": 0.8333841959635416,
      "grad_norm": 8.747199058532715,
      "learning_rate": 3.354031690864881e-07,
      "loss": 3.2654,
      "step": 81925
    },
    {
      "epoch": 0.83343505859375,
      "grad_norm": 13.0546875,
      "learning_rate": 3.3520312707041035e-07,
      "loss": 3.6446,
      "step": 81930
    },
    {
      "epoch": 0.8334859212239584,
      "grad_norm": 8.177398681640625,
      "learning_rate": 3.35003140440148e-07,
      "loss": 3.5208,
      "step": 81935
    },
    {
      "epoch": 0.8335367838541666,
      "grad_norm": 10.24907398223877,
      "learning_rate": 3.348032092008166e-07,
      "loss": 3.251,
      "step": 81940
    },
    {
      "epoch": 0.833587646484375,
      "grad_norm": 11.564193725585938,
      "learning_rate": 3.3460333335753214e-07,
      "loss": 3.0913,
      "step": 81945
    },
    {
      "epoch": 0.8336385091145834,
      "grad_norm": 12.961506843566895,
      "learning_rate": 3.34403512915408e-07,
      "loss": 3.3601,
      "step": 81950
    },
    {
      "epoch": 0.8336893717447916,
      "grad_norm": 9.499430656433105,
      "learning_rate": 3.3420374787955706e-07,
      "loss": 3.0469,
      "step": 81955
    },
    {
      "epoch": 0.833740234375,
      "grad_norm": 7.4130988121032715,
      "learning_rate": 3.3400403825508936e-07,
      "loss": 3.1784,
      "step": 81960
    },
    {
      "epoch": 0.8337910970052084,
      "grad_norm": 12.141097068786621,
      "learning_rate": 3.338043840471153e-07,
      "loss": 3.6559,
      "step": 81965
    },
    {
      "epoch": 0.8338419596354166,
      "grad_norm": 15.333015441894531,
      "learning_rate": 3.3360478526074296e-07,
      "loss": 3.5068,
      "step": 81970
    },
    {
      "epoch": 0.833892822265625,
      "grad_norm": 11.44824504852295,
      "learning_rate": 3.3340524190107774e-07,
      "loss": 3.1621,
      "step": 81975
    },
    {
      "epoch": 0.8339436848958334,
      "grad_norm": 10.404541969299316,
      "learning_rate": 3.3320575397322614e-07,
      "loss": 2.9386,
      "step": 81980
    },
    {
      "epoch": 0.8339945475260416,
      "grad_norm": 9.791139602661133,
      "learning_rate": 3.3300632148229234e-07,
      "loss": 3.7692,
      "step": 81985
    },
    {
      "epoch": 0.83404541015625,
      "grad_norm": 15.392584800720215,
      "learning_rate": 3.328069444333781e-07,
      "loss": 3.3095,
      "step": 81990
    },
    {
      "epoch": 0.8340962727864584,
      "grad_norm": 10.879815101623535,
      "learning_rate": 3.3260762283158446e-07,
      "loss": 3.1776,
      "step": 81995
    },
    {
      "epoch": 0.8341471354166666,
      "grad_norm": 14.615254402160645,
      "learning_rate": 3.324083566820105e-07,
      "loss": 3.5583,
      "step": 82000
    },
    {
      "epoch": 0.834197998046875,
      "grad_norm": 11.408792495727539,
      "learning_rate": 3.32209145989755e-07,
      "loss": 3.1966,
      "step": 82005
    },
    {
      "epoch": 0.8342488606770834,
      "grad_norm": 11.398858070373535,
      "learning_rate": 3.3200999075991533e-07,
      "loss": 3.0101,
      "step": 82010
    },
    {
      "epoch": 0.8342997233072916,
      "grad_norm": 7.233194828033447,
      "learning_rate": 3.318108909975859e-07,
      "loss": 3.0912,
      "step": 82015
    },
    {
      "epoch": 0.8343505859375,
      "grad_norm": 10.705859184265137,
      "learning_rate": 3.316118467078605e-07,
      "loss": 3.4198,
      "step": 82020
    },
    {
      "epoch": 0.8344014485677084,
      "grad_norm": 12.724525451660156,
      "learning_rate": 3.314128578958328e-07,
      "loss": 3.3804,
      "step": 82025
    },
    {
      "epoch": 0.8344523111979166,
      "grad_norm": 11.468843460083008,
      "learning_rate": 3.312139245665924e-07,
      "loss": 3.0083,
      "step": 82030
    },
    {
      "epoch": 0.834503173828125,
      "grad_norm": 16.455656051635742,
      "learning_rate": 3.310150467252302e-07,
      "loss": 3.3906,
      "step": 82035
    },
    {
      "epoch": 0.8345540364583334,
      "grad_norm": 10.620047569274902,
      "learning_rate": 3.308162243768334e-07,
      "loss": 3.2541,
      "step": 82040
    },
    {
      "epoch": 0.8346048990885416,
      "grad_norm": 11.70545768737793,
      "learning_rate": 3.3061745752648976e-07,
      "loss": 3.1841,
      "step": 82045
    },
    {
      "epoch": 0.83465576171875,
      "grad_norm": 9.349608421325684,
      "learning_rate": 3.3041874617928434e-07,
      "loss": 3.1505,
      "step": 82050
    },
    {
      "epoch": 0.8347066243489584,
      "grad_norm": 8.12730598449707,
      "learning_rate": 3.302200903403008e-07,
      "loss": 3.3881,
      "step": 82055
    },
    {
      "epoch": 0.8347574869791666,
      "grad_norm": 12.16651725769043,
      "learning_rate": 3.300214900146215e-07,
      "loss": 3.0127,
      "step": 82060
    },
    {
      "epoch": 0.834808349609375,
      "grad_norm": 12.703459739685059,
      "learning_rate": 3.29822945207329e-07,
      "loss": 3.3549,
      "step": 82065
    },
    {
      "epoch": 0.8348592122395834,
      "grad_norm": 15.172588348388672,
      "learning_rate": 3.296244559235018e-07,
      "loss": 3.3399,
      "step": 82070
    },
    {
      "epoch": 0.8349100748697916,
      "grad_norm": 8.584330558776855,
      "learning_rate": 3.29426022168218e-07,
      "loss": 3.672,
      "step": 82075
    },
    {
      "epoch": 0.8349609375,
      "grad_norm": 11.948514938354492,
      "learning_rate": 3.2922764394655526e-07,
      "loss": 3.4415,
      "step": 82080
    },
    {
      "epoch": 0.8350118001302084,
      "grad_norm": 9.167913436889648,
      "learning_rate": 3.290293212635884e-07,
      "loss": 3.4968,
      "step": 82085
    },
    {
      "epoch": 0.8350626627604166,
      "grad_norm": 10.437028884887695,
      "learning_rate": 3.288310541243925e-07,
      "loss": 3.5139,
      "step": 82090
    },
    {
      "epoch": 0.835113525390625,
      "grad_norm": 10.218053817749023,
      "learning_rate": 3.286328425340385e-07,
      "loss": 3.1128,
      "step": 82095
    },
    {
      "epoch": 0.8351643880208334,
      "grad_norm": 9.678133964538574,
      "learning_rate": 3.284346864975993e-07,
      "loss": 3.3178,
      "step": 82100
    },
    {
      "epoch": 0.8352152506510416,
      "grad_norm": 12.035988807678223,
      "learning_rate": 3.282365860201434e-07,
      "loss": 3.5213,
      "step": 82105
    },
    {
      "epoch": 0.83526611328125,
      "grad_norm": 8.763233184814453,
      "learning_rate": 3.280385411067402e-07,
      "loss": 3.4057,
      "step": 82110
    },
    {
      "epoch": 0.8353169759114584,
      "grad_norm": 15.29818058013916,
      "learning_rate": 3.2784055176245553e-07,
      "loss": 3.1658,
      "step": 82115
    },
    {
      "epoch": 0.8353678385416666,
      "grad_norm": 11.715181350708008,
      "learning_rate": 3.276426179923561e-07,
      "loss": 3.4415,
      "step": 82120
    },
    {
      "epoch": 0.835418701171875,
      "grad_norm": 13.8848295211792,
      "learning_rate": 3.274447398015054e-07,
      "loss": 3.4918,
      "step": 82125
    },
    {
      "epoch": 0.8354695638020834,
      "grad_norm": 14.780017852783203,
      "learning_rate": 3.272469171949655e-07,
      "loss": 3.0801,
      "step": 82130
    },
    {
      "epoch": 0.8355204264322916,
      "grad_norm": 15.15134334564209,
      "learning_rate": 3.2704915017779835e-07,
      "loss": 3.405,
      "step": 82135
    },
    {
      "epoch": 0.8355712890625,
      "grad_norm": 16.6488094329834,
      "learning_rate": 3.268514387550642e-07,
      "loss": 3.2162,
      "step": 82140
    },
    {
      "epoch": 0.8356221516927084,
      "grad_norm": 8.617156982421875,
      "learning_rate": 3.266537829318206e-07,
      "loss": 3.4536,
      "step": 82145
    },
    {
      "epoch": 0.8356730143229166,
      "grad_norm": 14.403000831604004,
      "learning_rate": 3.2645618271312434e-07,
      "loss": 2.9412,
      "step": 82150
    },
    {
      "epoch": 0.835723876953125,
      "grad_norm": 8.742593765258789,
      "learning_rate": 3.2625863810403226e-07,
      "loss": 3.2624,
      "step": 82155
    },
    {
      "epoch": 0.8357747395833334,
      "grad_norm": 14.44861888885498,
      "learning_rate": 3.260611491095969e-07,
      "loss": 3.5661,
      "step": 82160
    },
    {
      "epoch": 0.8358256022135416,
      "grad_norm": 11.183387756347656,
      "learning_rate": 3.258637157348726e-07,
      "loss": 3.425,
      "step": 82165
    },
    {
      "epoch": 0.83587646484375,
      "grad_norm": 14.68026351928711,
      "learning_rate": 3.25666337984909e-07,
      "loss": 3.0987,
      "step": 82170
    },
    {
      "epoch": 0.8359273274739584,
      "grad_norm": 9.858868598937988,
      "learning_rate": 3.254690158647572e-07,
      "loss": 3.5059,
      "step": 82175
    },
    {
      "epoch": 0.8359781901041666,
      "grad_norm": 7.8813629150390625,
      "learning_rate": 3.2527174937946546e-07,
      "loss": 3.2584,
      "step": 82180
    },
    {
      "epoch": 0.836029052734375,
      "grad_norm": 10.828450202941895,
      "learning_rate": 3.2507453853408004e-07,
      "loss": 3.7391,
      "step": 82185
    },
    {
      "epoch": 0.8360799153645834,
      "grad_norm": 7.974329471588135,
      "learning_rate": 3.248773833336469e-07,
      "loss": 3.5249,
      "step": 82190
    },
    {
      "epoch": 0.8361307779947916,
      "grad_norm": 15.77200984954834,
      "learning_rate": 3.246802837832108e-07,
      "loss": 3.2638,
      "step": 82195
    },
    {
      "epoch": 0.836181640625,
      "grad_norm": 13.883968353271484,
      "learning_rate": 3.244832398878142e-07,
      "loss": 3.8104,
      "step": 82200
    },
    {
      "epoch": 0.8362325032552084,
      "grad_norm": 8.318225860595703,
      "learning_rate": 3.2428625165249834e-07,
      "loss": 3.2504,
      "step": 82205
    },
    {
      "epoch": 0.8362833658854166,
      "grad_norm": 11.81405258178711,
      "learning_rate": 3.2408931908230217e-07,
      "loss": 3.3791,
      "step": 82210
    },
    {
      "epoch": 0.836334228515625,
      "grad_norm": 9.700438499450684,
      "learning_rate": 3.2389244218226533e-07,
      "loss": 3.2398,
      "step": 82215
    },
    {
      "epoch": 0.8363850911458334,
      "grad_norm": 14.125505447387695,
      "learning_rate": 3.236956209574249e-07,
      "loss": 3.2274,
      "step": 82220
    },
    {
      "epoch": 0.8364359537760416,
      "grad_norm": 10.842809677124023,
      "learning_rate": 3.2349885541281626e-07,
      "loss": 3.0955,
      "step": 82225
    },
    {
      "epoch": 0.83648681640625,
      "grad_norm": 10.028032302856445,
      "learning_rate": 3.2330214555347297e-07,
      "loss": 3.1531,
      "step": 82230
    },
    {
      "epoch": 0.8365376790364584,
      "grad_norm": 10.2706298828125,
      "learning_rate": 3.231054913844289e-07,
      "loss": 3.6072,
      "step": 82235
    },
    {
      "epoch": 0.8365885416666666,
      "grad_norm": 12.325729370117188,
      "learning_rate": 3.2290889291071434e-07,
      "loss": 3.5087,
      "step": 82240
    },
    {
      "epoch": 0.836639404296875,
      "grad_norm": 11.384655952453613,
      "learning_rate": 3.227123501373597e-07,
      "loss": 3.5067,
      "step": 82245
    },
    {
      "epoch": 0.8366902669270834,
      "grad_norm": 7.226029396057129,
      "learning_rate": 3.225158630693939e-07,
      "loss": 3.604,
      "step": 82250
    },
    {
      "epoch": 0.8367411295572916,
      "grad_norm": 7.987247943878174,
      "learning_rate": 3.223194317118439e-07,
      "loss": 3.1628,
      "step": 82255
    },
    {
      "epoch": 0.8367919921875,
      "grad_norm": 9.990367889404297,
      "learning_rate": 3.221230560697344e-07,
      "loss": 3.2523,
      "step": 82260
    },
    {
      "epoch": 0.8368428548177084,
      "grad_norm": 11.045588493347168,
      "learning_rate": 3.2192673614809015e-07,
      "loss": 3.1911,
      "step": 82265
    },
    {
      "epoch": 0.8368937174479166,
      "grad_norm": 9.236793518066406,
      "learning_rate": 3.217304719519349e-07,
      "loss": 3.6594,
      "step": 82270
    },
    {
      "epoch": 0.836944580078125,
      "grad_norm": 7.034839153289795,
      "learning_rate": 3.215342634862889e-07,
      "loss": 3.1625,
      "step": 82275
    },
    {
      "epoch": 0.8369954427083334,
      "grad_norm": 12.31274700164795,
      "learning_rate": 3.2133811075617275e-07,
      "loss": 3.5175,
      "step": 82280
    },
    {
      "epoch": 0.8370463053385416,
      "grad_norm": 15.286792755126953,
      "learning_rate": 3.2114201376660407e-07,
      "loss": 3.272,
      "step": 82285
    },
    {
      "epoch": 0.83709716796875,
      "grad_norm": 8.54918384552002,
      "learning_rate": 3.2094597252260024e-07,
      "loss": 3.3357,
      "step": 82290
    },
    {
      "epoch": 0.8371480305989584,
      "grad_norm": 13.110342025756836,
      "learning_rate": 3.20749987029178e-07,
      "loss": 3.361,
      "step": 82295
    },
    {
      "epoch": 0.8371988932291666,
      "grad_norm": 15.439437866210938,
      "learning_rate": 3.205540572913507e-07,
      "loss": 3.4519,
      "step": 82300
    },
    {
      "epoch": 0.837249755859375,
      "grad_norm": 8.734394073486328,
      "learning_rate": 3.203581833141306e-07,
      "loss": 3.4643,
      "step": 82305
    },
    {
      "epoch": 0.8373006184895834,
      "grad_norm": 15.3256196975708,
      "learning_rate": 3.2016236510253067e-07,
      "loss": 3.2861,
      "step": 82310
    },
    {
      "epoch": 0.8373514811197916,
      "grad_norm": 12.538545608520508,
      "learning_rate": 3.1996660266155913e-07,
      "loss": 3.1708,
      "step": 82315
    },
    {
      "epoch": 0.83740234375,
      "grad_norm": 13.305954933166504,
      "learning_rate": 3.197708959962259e-07,
      "loss": 3.5401,
      "step": 82320
    },
    {
      "epoch": 0.8374532063802084,
      "grad_norm": 9.767401695251465,
      "learning_rate": 3.1957524511153694e-07,
      "loss": 3.1989,
      "step": 82325
    },
    {
      "epoch": 0.8375040690104166,
      "grad_norm": 12.346323013305664,
      "learning_rate": 3.19379650012499e-07,
      "loss": 3.5382,
      "step": 82330
    },
    {
      "epoch": 0.837554931640625,
      "grad_norm": 7.60872745513916,
      "learning_rate": 3.1918411070411605e-07,
      "loss": 4.027,
      "step": 82335
    },
    {
      "epoch": 0.8376057942708334,
      "grad_norm": 12.669442176818848,
      "learning_rate": 3.1898862719139003e-07,
      "loss": 3.7596,
      "step": 82340
    },
    {
      "epoch": 0.8376566569010416,
      "grad_norm": 8.977251052856445,
      "learning_rate": 3.187931994793231e-07,
      "loss": 3.2094,
      "step": 82345
    },
    {
      "epoch": 0.83770751953125,
      "grad_norm": 10.287897109985352,
      "learning_rate": 3.185978275729157e-07,
      "loss": 3.2141,
      "step": 82350
    },
    {
      "epoch": 0.8377583821614584,
      "grad_norm": 13.242105484008789,
      "learning_rate": 3.184025114771658e-07,
      "loss": 3.4793,
      "step": 82355
    },
    {
      "epoch": 0.8378092447916666,
      "grad_norm": 13.227606773376465,
      "learning_rate": 3.182072511970702e-07,
      "loss": 3.2601,
      "step": 82360
    },
    {
      "epoch": 0.837860107421875,
      "grad_norm": 7.591468811035156,
      "learning_rate": 3.180120467376252e-07,
      "loss": 3.4581,
      "step": 82365
    },
    {
      "epoch": 0.8379109700520834,
      "grad_norm": 12.946093559265137,
      "learning_rate": 3.178168981038243e-07,
      "loss": 3.2214,
      "step": 82370
    },
    {
      "epoch": 0.8379618326822916,
      "grad_norm": 11.933862686157227,
      "learning_rate": 3.176218053006616e-07,
      "loss": 3.2006,
      "step": 82375
    },
    {
      "epoch": 0.8380126953125,
      "grad_norm": 8.173795700073242,
      "learning_rate": 3.17426768333127e-07,
      "loss": 3.1671,
      "step": 82380
    },
    {
      "epoch": 0.8380635579427084,
      "grad_norm": 10.22460651397705,
      "learning_rate": 3.1723178720621167e-07,
      "loss": 2.9694,
      "step": 82385
    },
    {
      "epoch": 0.8381144205729166,
      "grad_norm": 10.599739074707031,
      "learning_rate": 3.1703686192490395e-07,
      "loss": 3.0268,
      "step": 82390
    },
    {
      "epoch": 0.838165283203125,
      "grad_norm": 13.632969856262207,
      "learning_rate": 3.1684199249418983e-07,
      "loss": 3.4566,
      "step": 82395
    },
    {
      "epoch": 0.8382161458333334,
      "grad_norm": 10.012470245361328,
      "learning_rate": 3.1664717891905616e-07,
      "loss": 3.1663,
      "step": 82400
    },
    {
      "epoch": 0.8382670084635416,
      "grad_norm": 17.0555419921875,
      "learning_rate": 3.1645242120448725e-07,
      "loss": 3.4753,
      "step": 82405
    },
    {
      "epoch": 0.83831787109375,
      "grad_norm": 13.731001853942871,
      "learning_rate": 3.1625771935546557e-07,
      "loss": 3.2863,
      "step": 82410
    },
    {
      "epoch": 0.8383687337239584,
      "grad_norm": 9.11953067779541,
      "learning_rate": 3.1606307337697235e-07,
      "loss": 3.403,
      "step": 82415
    },
    {
      "epoch": 0.8384195963541666,
      "grad_norm": 10.7665433883667,
      "learning_rate": 3.1586848327398724e-07,
      "loss": 3.2105,
      "step": 82420
    },
    {
      "epoch": 0.838470458984375,
      "grad_norm": 14.06084156036377,
      "learning_rate": 3.156739490514901e-07,
      "loss": 3.635,
      "step": 82425
    },
    {
      "epoch": 0.8385213216145834,
      "grad_norm": 13.289153099060059,
      "learning_rate": 3.1547947071445736e-07,
      "loss": 3.1906,
      "step": 82430
    },
    {
      "epoch": 0.8385721842447916,
      "grad_norm": 16.027151107788086,
      "learning_rate": 3.152850482678638e-07,
      "loss": 3.6236,
      "step": 82435
    },
    {
      "epoch": 0.838623046875,
      "grad_norm": 11.879664421081543,
      "learning_rate": 3.1509068171668514e-07,
      "loss": 3.5951,
      "step": 82440
    },
    {
      "epoch": 0.8386739095052084,
      "grad_norm": 14.067399978637695,
      "learning_rate": 3.148963710658928e-07,
      "loss": 3.3605,
      "step": 82445
    },
    {
      "epoch": 0.8387247721354166,
      "grad_norm": 8.084772109985352,
      "learning_rate": 3.1470211632045944e-07,
      "loss": 3.5825,
      "step": 82450
    },
    {
      "epoch": 0.838775634765625,
      "grad_norm": 7.3616766929626465,
      "learning_rate": 3.145079174853541e-07,
      "loss": 3.0249,
      "step": 82455
    },
    {
      "epoch": 0.8388264973958334,
      "grad_norm": 15.187037467956543,
      "learning_rate": 3.143137745655461e-07,
      "loss": 3.3028,
      "step": 82460
    },
    {
      "epoch": 0.8388773600260416,
      "grad_norm": 13.803275108337402,
      "learning_rate": 3.141196875660021e-07,
      "loss": 3.3048,
      "step": 82465
    },
    {
      "epoch": 0.83892822265625,
      "grad_norm": 15.948885917663574,
      "learning_rate": 3.1392565649168735e-07,
      "loss": 3.3256,
      "step": 82470
    },
    {
      "epoch": 0.8389790852864584,
      "grad_norm": 8.560640335083008,
      "learning_rate": 3.137316813475663e-07,
      "loss": 3.5241,
      "step": 82475
    },
    {
      "epoch": 0.8390299479166666,
      "grad_norm": 9.086495399475098,
      "learning_rate": 3.1353776213860267e-07,
      "loss": 3.7053,
      "step": 82480
    },
    {
      "epoch": 0.839080810546875,
      "grad_norm": 9.937238693237305,
      "learning_rate": 3.1334389886975707e-07,
      "loss": 3.1616,
      "step": 82485
    },
    {
      "epoch": 0.8391316731770834,
      "grad_norm": 15.993388175964355,
      "learning_rate": 3.131500915459898e-07,
      "loss": 3.5369,
      "step": 82490
    },
    {
      "epoch": 0.8391825358072916,
      "grad_norm": 16.628915786743164,
      "learning_rate": 3.129563401722582e-07,
      "loss": 3.3434,
      "step": 82495
    },
    {
      "epoch": 0.8392333984375,
      "grad_norm": 7.18716287612915,
      "learning_rate": 3.1276264475352036e-07,
      "loss": 3.1887,
      "step": 82500
    },
    {
      "epoch": 0.8392842610677084,
      "grad_norm": 12.53756046295166,
      "learning_rate": 3.125690052947325e-07,
      "loss": 3.2412,
      "step": 82505
    },
    {
      "epoch": 0.8393351236979166,
      "grad_norm": 11.504670143127441,
      "learning_rate": 3.1237542180084817e-07,
      "loss": 3.1316,
      "step": 82510
    },
    {
      "epoch": 0.839385986328125,
      "grad_norm": 9.89035415649414,
      "learning_rate": 3.121818942768193e-07,
      "loss": 3.5028,
      "step": 82515
    },
    {
      "epoch": 0.8394368489583334,
      "grad_norm": 9.171295166015625,
      "learning_rate": 3.1198842272759905e-07,
      "loss": 3.3716,
      "step": 82520
    },
    {
      "epoch": 0.8394877115885416,
      "grad_norm": 12.68211841583252,
      "learning_rate": 3.117950071581355e-07,
      "loss": 3.3988,
      "step": 82525
    },
    {
      "epoch": 0.83953857421875,
      "grad_norm": 7.9228739738464355,
      "learning_rate": 3.1160164757337826e-07,
      "loss": 3.2639,
      "step": 82530
    },
    {
      "epoch": 0.8395894368489584,
      "grad_norm": 16.38813591003418,
      "learning_rate": 3.114083439782745e-07,
      "loss": 3.4395,
      "step": 82535
    },
    {
      "epoch": 0.8396402994791666,
      "grad_norm": 7.744754314422607,
      "learning_rate": 3.1121509637776966e-07,
      "loss": 3.4227,
      "step": 82540
    },
    {
      "epoch": 0.839691162109375,
      "grad_norm": 11.593255043029785,
      "learning_rate": 3.1102190477680763e-07,
      "loss": 3.1681,
      "step": 82545
    },
    {
      "epoch": 0.8397420247395834,
      "grad_norm": 9.948272705078125,
      "learning_rate": 3.1082876918033053e-07,
      "loss": 3.2039,
      "step": 82550
    },
    {
      "epoch": 0.8397928873697916,
      "grad_norm": 14.423352241516113,
      "learning_rate": 3.106356895932813e-07,
      "loss": 3.664,
      "step": 82555
    },
    {
      "epoch": 0.83984375,
      "grad_norm": 9.781453132629395,
      "learning_rate": 3.104426660205989e-07,
      "loss": 3.2269,
      "step": 82560
    },
    {
      "epoch": 0.8398946126302084,
      "grad_norm": 12.1110200881958,
      "learning_rate": 3.1024969846722235e-07,
      "loss": 3.458,
      "step": 82565
    },
    {
      "epoch": 0.8399454752604166,
      "grad_norm": 14.258819580078125,
      "learning_rate": 3.1005678693808745e-07,
      "loss": 3.3119,
      "step": 82570
    },
    {
      "epoch": 0.839996337890625,
      "grad_norm": 9.788232803344727,
      "learning_rate": 3.098639314381305e-07,
      "loss": 3.3446,
      "step": 82575
    },
    {
      "epoch": 0.8400472005208334,
      "grad_norm": 12.888335227966309,
      "learning_rate": 3.0967113197228643e-07,
      "loss": 3.2827,
      "step": 82580
    },
    {
      "epoch": 0.8400980631510416,
      "grad_norm": 16.238554000854492,
      "learning_rate": 3.0947838854548717e-07,
      "loss": 3.9365,
      "step": 82585
    },
    {
      "epoch": 0.84014892578125,
      "grad_norm": 14.492205619812012,
      "learning_rate": 3.092857011626635e-07,
      "loss": 3.6415,
      "step": 82590
    },
    {
      "epoch": 0.8401997884114584,
      "grad_norm": 9.4275541305542,
      "learning_rate": 3.090930698287467e-07,
      "loss": 3.8095,
      "step": 82595
    },
    {
      "epoch": 0.8402506510416666,
      "grad_norm": 12.679570198059082,
      "learning_rate": 3.089004945486637e-07,
      "loss": 3.2866,
      "step": 82600
    },
    {
      "epoch": 0.840301513671875,
      "grad_norm": 12.28504467010498,
      "learning_rate": 3.087079753273428e-07,
      "loss": 3.4753,
      "step": 82605
    },
    {
      "epoch": 0.8403523763020834,
      "grad_norm": 7.86800479888916,
      "learning_rate": 3.085155121697084e-07,
      "loss": 3.365,
      "step": 82610
    },
    {
      "epoch": 0.8404032389322916,
      "grad_norm": 7.812675952911377,
      "learning_rate": 3.0832310508068564e-07,
      "loss": 3.4058,
      "step": 82615
    },
    {
      "epoch": 0.8404541015625,
      "grad_norm": 14.575812339782715,
      "learning_rate": 3.0813075406519705e-07,
      "loss": 3.2456,
      "step": 82620
    },
    {
      "epoch": 0.8405049641927084,
      "grad_norm": 12.685293197631836,
      "learning_rate": 3.079384591281628e-07,
      "loss": 3.446,
      "step": 82625
    },
    {
      "epoch": 0.8405558268229166,
      "grad_norm": 9.814604759216309,
      "learning_rate": 3.077462202745035e-07,
      "loss": 3.2492,
      "step": 82630
    },
    {
      "epoch": 0.840606689453125,
      "grad_norm": 12.434163093566895,
      "learning_rate": 3.075540375091379e-07,
      "loss": 3.3287,
      "step": 82635
    },
    {
      "epoch": 0.8406575520833334,
      "grad_norm": 15.110447883605957,
      "learning_rate": 3.0736191083698297e-07,
      "loss": 3.2399,
      "step": 82640
    },
    {
      "epoch": 0.8407084147135416,
      "grad_norm": 16.645370483398438,
      "learning_rate": 3.0716984026295303e-07,
      "loss": 3.5208,
      "step": 82645
    },
    {
      "epoch": 0.84075927734375,
      "grad_norm": 12.585023880004883,
      "learning_rate": 3.069778257919637e-07,
      "loss": 3.171,
      "step": 82650
    },
    {
      "epoch": 0.8408101399739584,
      "grad_norm": 8.575360298156738,
      "learning_rate": 3.067858674289262e-07,
      "loss": 3.3535,
      "step": 82655
    },
    {
      "epoch": 0.8408610026041666,
      "grad_norm": 7.247581481933594,
      "learning_rate": 3.0659396517875305e-07,
      "loss": 3.3298,
      "step": 82660
    },
    {
      "epoch": 0.840911865234375,
      "grad_norm": 9.335526466369629,
      "learning_rate": 3.0640211904635287e-07,
      "loss": 3.0408,
      "step": 82665
    },
    {
      "epoch": 0.8409627278645834,
      "grad_norm": 9.049152374267578,
      "learning_rate": 3.06210329036635e-07,
      "loss": 3.067,
      "step": 82670
    },
    {
      "epoch": 0.8410135904947916,
      "grad_norm": 10.849244117736816,
      "learning_rate": 3.060185951545061e-07,
      "loss": 3.3345,
      "step": 82675
    },
    {
      "epoch": 0.841064453125,
      "grad_norm": 12.523625373840332,
      "learning_rate": 3.058269174048706e-07,
      "loss": 3.2685,
      "step": 82680
    },
    {
      "epoch": 0.8411153157552084,
      "grad_norm": 14.451581001281738,
      "learning_rate": 3.056352957926334e-07,
      "loss": 3.2992,
      "step": 82685
    },
    {
      "epoch": 0.8411661783854166,
      "grad_norm": 14.779975891113281,
      "learning_rate": 3.054437303226976e-07,
      "loss": 3.5794,
      "step": 82690
    },
    {
      "epoch": 0.841217041015625,
      "grad_norm": 13.709236145019531,
      "learning_rate": 3.052522209999639e-07,
      "loss": 3.1511,
      "step": 82695
    },
    {
      "epoch": 0.8412679036458334,
      "grad_norm": 7.6075358390808105,
      "learning_rate": 3.0506076782933185e-07,
      "loss": 3.5767,
      "step": 82700
    },
    {
      "epoch": 0.8413187662760416,
      "grad_norm": 12.561664581298828,
      "learning_rate": 3.0486937081569855e-07,
      "loss": 3.5576,
      "step": 82705
    },
    {
      "epoch": 0.84136962890625,
      "grad_norm": 12.769579887390137,
      "learning_rate": 3.0467802996396346e-07,
      "loss": 3.3144,
      "step": 82710
    },
    {
      "epoch": 0.8414204915364584,
      "grad_norm": 13.259459495544434,
      "learning_rate": 3.044867452790204e-07,
      "loss": 3.2626,
      "step": 82715
    },
    {
      "epoch": 0.8414713541666666,
      "grad_norm": 14.529406547546387,
      "learning_rate": 3.042955167657635e-07,
      "loss": 3.5059,
      "step": 82720
    },
    {
      "epoch": 0.841522216796875,
      "grad_norm": 10.720898628234863,
      "learning_rate": 3.0410434442908504e-07,
      "loss": 3.2629,
      "step": 82725
    },
    {
      "epoch": 0.8415730794270834,
      "grad_norm": 7.874776363372803,
      "learning_rate": 3.039132282738766e-07,
      "loss": 3.2861,
      "step": 82730
    },
    {
      "epoch": 0.8416239420572916,
      "grad_norm": 11.8858642578125,
      "learning_rate": 3.037221683050279e-07,
      "loss": 3.2717,
      "step": 82735
    },
    {
      "epoch": 0.8416748046875,
      "grad_norm": 13.591537475585938,
      "learning_rate": 3.0353116452742635e-07,
      "loss": 3.1717,
      "step": 82740
    },
    {
      "epoch": 0.8417256673177084,
      "grad_norm": 9.644363403320312,
      "learning_rate": 3.033402169459601e-07,
      "loss": 2.9345,
      "step": 82745
    },
    {
      "epoch": 0.8417765299479166,
      "grad_norm": 19.541770935058594,
      "learning_rate": 3.031493255655138e-07,
      "loss": 3.3933,
      "step": 82750
    },
    {
      "epoch": 0.841827392578125,
      "grad_norm": 10.669909477233887,
      "learning_rate": 3.029584903909705e-07,
      "loss": 3.119,
      "step": 82755
    },
    {
      "epoch": 0.8418782552083334,
      "grad_norm": 14.149001121520996,
      "learning_rate": 3.0276771142721373e-07,
      "loss": 3.4355,
      "step": 82760
    },
    {
      "epoch": 0.8419291178385416,
      "grad_norm": 11.106964111328125,
      "learning_rate": 3.025769886791247e-07,
      "loss": 3.2454,
      "step": 82765
    },
    {
      "epoch": 0.84197998046875,
      "grad_norm": 14.55089282989502,
      "learning_rate": 3.023863221515824e-07,
      "loss": 3.1274,
      "step": 82770
    },
    {
      "epoch": 0.8420308430989584,
      "grad_norm": 11.762307167053223,
      "learning_rate": 3.0219571184946536e-07,
      "loss": 3.5242,
      "step": 82775
    },
    {
      "epoch": 0.8420817057291666,
      "grad_norm": 8.462861061096191,
      "learning_rate": 3.020051577776495e-07,
      "loss": 3.2378,
      "step": 82780
    },
    {
      "epoch": 0.842132568359375,
      "grad_norm": 7.629974365234375,
      "learning_rate": 3.018146599410107e-07,
      "loss": 3.2623,
      "step": 82785
    },
    {
      "epoch": 0.8421834309895834,
      "grad_norm": 12.82613468170166,
      "learning_rate": 3.0162421834442316e-07,
      "loss": 3.6228,
      "step": 82790
    },
    {
      "epoch": 0.8422342936197916,
      "grad_norm": 12.538959503173828,
      "learning_rate": 3.014338329927591e-07,
      "loss": 3.0027,
      "step": 82795
    },
    {
      "epoch": 0.84228515625,
      "grad_norm": 15.235918998718262,
      "learning_rate": 3.0124350389088894e-07,
      "loss": 3.3562,
      "step": 82800
    },
    {
      "epoch": 0.8423360188802084,
      "grad_norm": 14.518644332885742,
      "learning_rate": 3.0105323104368285e-07,
      "loss": 3.6012,
      "step": 82805
    },
    {
      "epoch": 0.8423868815104166,
      "grad_norm": 8.101212501525879,
      "learning_rate": 3.0086301445600816e-07,
      "loss": 3.3002,
      "step": 82810
    },
    {
      "epoch": 0.842437744140625,
      "grad_norm": 7.306506633758545,
      "learning_rate": 3.006728541327325e-07,
      "loss": 3.3596,
      "step": 82815
    },
    {
      "epoch": 0.8424886067708334,
      "grad_norm": 11.812324523925781,
      "learning_rate": 3.0048275007872e-07,
      "loss": 3.365,
      "step": 82820
    },
    {
      "epoch": 0.8425394694010416,
      "grad_norm": 12.635832786560059,
      "learning_rate": 3.002927022988358e-07,
      "loss": 3.4852,
      "step": 82825
    },
    {
      "epoch": 0.84259033203125,
      "grad_norm": 14.342005729675293,
      "learning_rate": 3.001027107979412e-07,
      "loss": 3.2962,
      "step": 82830
    },
    {
      "epoch": 0.8426411946614584,
      "grad_norm": 10.07796859741211,
      "learning_rate": 2.9991277558089673e-07,
      "loss": 3.4087,
      "step": 82835
    },
    {
      "epoch": 0.8426920572916666,
      "grad_norm": 13.13611888885498,
      "learning_rate": 2.9972289665256256e-07,
      "loss": 3.1798,
      "step": 82840
    },
    {
      "epoch": 0.842742919921875,
      "grad_norm": 12.30323314666748,
      "learning_rate": 2.995330740177971e-07,
      "loss": 2.9818,
      "step": 82845
    },
    {
      "epoch": 0.8427937825520834,
      "grad_norm": 13.685778617858887,
      "learning_rate": 2.993433076814561e-07,
      "loss": 3.4816,
      "step": 82850
    },
    {
      "epoch": 0.8428446451822916,
      "grad_norm": 10.468883514404297,
      "learning_rate": 2.991535976483947e-07,
      "loss": 3.3311,
      "step": 82855
    },
    {
      "epoch": 0.8428955078125,
      "grad_norm": 12.333459854125977,
      "learning_rate": 2.989639439234668e-07,
      "loss": 3.3499,
      "step": 82860
    },
    {
      "epoch": 0.8429463704427084,
      "grad_norm": 12.511208534240723,
      "learning_rate": 2.9877434651152506e-07,
      "loss": 3.5325,
      "step": 82865
    },
    {
      "epoch": 0.8429972330729166,
      "grad_norm": 11.679605484008789,
      "learning_rate": 2.985848054174201e-07,
      "loss": 3.553,
      "step": 82870
    },
    {
      "epoch": 0.843048095703125,
      "grad_norm": 8.571442604064941,
      "learning_rate": 2.983953206460005e-07,
      "loss": 3.3191,
      "step": 82875
    },
    {
      "epoch": 0.8430989583333334,
      "grad_norm": 9.685628890991211,
      "learning_rate": 2.982058922021153e-07,
      "loss": 3.2059,
      "step": 82880
    },
    {
      "epoch": 0.8431498209635416,
      "grad_norm": 12.202717781066895,
      "learning_rate": 2.980165200906102e-07,
      "loss": 3.248,
      "step": 82885
    },
    {
      "epoch": 0.84320068359375,
      "grad_norm": 12.813880920410156,
      "learning_rate": 2.978272043163308e-07,
      "loss": 3.4659,
      "step": 82890
    },
    {
      "epoch": 0.8432515462239584,
      "grad_norm": 13.448853492736816,
      "learning_rate": 2.976379448841199e-07,
      "loss": 3.2986,
      "step": 82895
    },
    {
      "epoch": 0.8433024088541666,
      "grad_norm": 14.669266700744629,
      "learning_rate": 2.974487417988206e-07,
      "loss": 3.674,
      "step": 82900
    },
    {
      "epoch": 0.843353271484375,
      "grad_norm": 8.112249374389648,
      "learning_rate": 2.9725959506527317e-07,
      "loss": 3.2246,
      "step": 82905
    },
    {
      "epoch": 0.8434041341145834,
      "grad_norm": 14.91462230682373,
      "learning_rate": 2.970705046883166e-07,
      "loss": 3.377,
      "step": 82910
    },
    {
      "epoch": 0.8434549967447916,
      "grad_norm": 14.564872741699219,
      "learning_rate": 2.968814706727888e-07,
      "loss": 4.0069,
      "step": 82915
    },
    {
      "epoch": 0.843505859375,
      "grad_norm": 10.619250297546387,
      "learning_rate": 2.966924930235268e-07,
      "loss": 3.4485,
      "step": 82920
    },
    {
      "epoch": 0.8435567220052084,
      "grad_norm": 8.611906051635742,
      "learning_rate": 2.965035717453654e-07,
      "loss": 3.3972,
      "step": 82925
    },
    {
      "epoch": 0.8436075846354166,
      "grad_norm": 8.428669929504395,
      "learning_rate": 2.9631470684313713e-07,
      "loss": 3.6412,
      "step": 82930
    },
    {
      "epoch": 0.843658447265625,
      "grad_norm": 11.347962379455566,
      "learning_rate": 2.961258983216755e-07,
      "loss": 3.332,
      "step": 82935
    },
    {
      "epoch": 0.8437093098958334,
      "grad_norm": 10.147064208984375,
      "learning_rate": 2.959371461858096e-07,
      "loss": 3.2402,
      "step": 82940
    },
    {
      "epoch": 0.8437601725260416,
      "grad_norm": 14.61273193359375,
      "learning_rate": 2.9574845044037013e-07,
      "loss": 3.3754,
      "step": 82945
    },
    {
      "epoch": 0.84381103515625,
      "grad_norm": 10.914712905883789,
      "learning_rate": 2.955598110901833e-07,
      "loss": 3.3198,
      "step": 82950
    },
    {
      "epoch": 0.8438618977864584,
      "grad_norm": 8.582925796508789,
      "learning_rate": 2.9537122814007695e-07,
      "loss": 3.4709,
      "step": 82955
    },
    {
      "epoch": 0.8439127604166666,
      "grad_norm": 14.4943208694458,
      "learning_rate": 2.9518270159487524e-07,
      "loss": 3.2319,
      "step": 82960
    },
    {
      "epoch": 0.843963623046875,
      "grad_norm": 12.431747436523438,
      "learning_rate": 2.9499423145940067e-07,
      "loss": 3.3942,
      "step": 82965
    },
    {
      "epoch": 0.8440144856770834,
      "grad_norm": 15.191832542419434,
      "learning_rate": 2.9480581773847636e-07,
      "loss": 3.448,
      "step": 82970
    },
    {
      "epoch": 0.8440653483072916,
      "grad_norm": 11.43300724029541,
      "learning_rate": 2.9461746043692286e-07,
      "loss": 3.5234,
      "step": 82975
    },
    {
      "epoch": 0.8441162109375,
      "grad_norm": 11.807929992675781,
      "learning_rate": 2.9442915955955885e-07,
      "loss": 3.2456,
      "step": 82980
    },
    {
      "epoch": 0.8441670735677084,
      "grad_norm": 11.288966178894043,
      "learning_rate": 2.942409151112022e-07,
      "loss": 3.5194,
      "step": 82985
    },
    {
      "epoch": 0.8442179361979166,
      "grad_norm": 16.285200119018555,
      "learning_rate": 2.940527270966684e-07,
      "loss": 3.1871,
      "step": 82990
    },
    {
      "epoch": 0.844268798828125,
      "grad_norm": 16.906225204467773,
      "learning_rate": 2.9386459552077254e-07,
      "loss": 3.2949,
      "step": 82995
    },
    {
      "epoch": 0.8443196614583334,
      "grad_norm": 11.90079116821289,
      "learning_rate": 2.9367652038832876e-07,
      "loss": 3.2382,
      "step": 83000
    },
    {
      "epoch": 0.8443705240885416,
      "grad_norm": 17.509187698364258,
      "learning_rate": 2.9348850170414805e-07,
      "loss": 3.4055,
      "step": 83005
    },
    {
      "epoch": 0.84442138671875,
      "grad_norm": 12.756876945495605,
      "learning_rate": 2.9330053947304065e-07,
      "loss": 3.3672,
      "step": 83010
    },
    {
      "epoch": 0.8444722493489584,
      "grad_norm": 13.759655952453613,
      "learning_rate": 2.9311263369981606e-07,
      "loss": 3.3343,
      "step": 83015
    },
    {
      "epoch": 0.8445231119791666,
      "grad_norm": 11.43967342376709,
      "learning_rate": 2.9292478438928183e-07,
      "loss": 3.0752,
      "step": 83020
    },
    {
      "epoch": 0.844573974609375,
      "grad_norm": 9.05441951751709,
      "learning_rate": 2.9273699154624356e-07,
      "loss": 3.0076,
      "step": 83025
    },
    {
      "epoch": 0.8446248372395834,
      "grad_norm": 14.203214645385742,
      "learning_rate": 2.925492551755066e-07,
      "loss": 3.6563,
      "step": 83030
    },
    {
      "epoch": 0.8446756998697916,
      "grad_norm": 14.050749778747559,
      "learning_rate": 2.92361575281874e-07,
      "loss": 3.8481,
      "step": 83035
    },
    {
      "epoch": 0.8447265625,
      "grad_norm": 8.36119270324707,
      "learning_rate": 2.9217395187014635e-07,
      "loss": 3.4725,
      "step": 83040
    },
    {
      "epoch": 0.8447774251302084,
      "grad_norm": 14.630526542663574,
      "learning_rate": 2.919863849451249e-07,
      "loss": 3.7519,
      "step": 83045
    },
    {
      "epoch": 0.8448282877604166,
      "grad_norm": 13.809406280517578,
      "learning_rate": 2.9179887451160906e-07,
      "loss": 2.903,
      "step": 83050
    },
    {
      "epoch": 0.844879150390625,
      "grad_norm": 12.206207275390625,
      "learning_rate": 2.9161142057439556e-07,
      "loss": 3.3143,
      "step": 83055
    },
    {
      "epoch": 0.8449300130208334,
      "grad_norm": 10.303620338439941,
      "learning_rate": 2.9142402313828033e-07,
      "loss": 3.3072,
      "step": 83060
    },
    {
      "epoch": 0.8449808756510416,
      "grad_norm": 11.616304397583008,
      "learning_rate": 2.912366822080576e-07,
      "loss": 3.3516,
      "step": 83065
    },
    {
      "epoch": 0.84503173828125,
      "grad_norm": 11.960399627685547,
      "learning_rate": 2.910493977885209e-07,
      "loss": 3.0442,
      "step": 83070
    },
    {
      "epoch": 0.8450826009114584,
      "grad_norm": 8.009824752807617,
      "learning_rate": 2.9086216988446215e-07,
      "loss": 3.3228,
      "step": 83075
    },
    {
      "epoch": 0.8451334635416666,
      "grad_norm": 17.964433670043945,
      "learning_rate": 2.9067499850067124e-07,
      "loss": 3.5728,
      "step": 83080
    },
    {
      "epoch": 0.845184326171875,
      "grad_norm": 9.637298583984375,
      "learning_rate": 2.9048788364193615e-07,
      "loss": 3.3517,
      "step": 83085
    },
    {
      "epoch": 0.8452351888020834,
      "grad_norm": 16.583690643310547,
      "learning_rate": 2.9030082531304546e-07,
      "loss": 3.2254,
      "step": 83090
    },
    {
      "epoch": 0.8452860514322916,
      "grad_norm": 8.117792129516602,
      "learning_rate": 2.901138235187839e-07,
      "loss": 3.1893,
      "step": 83095
    },
    {
      "epoch": 0.8453369140625,
      "grad_norm": 9.406477928161621,
      "learning_rate": 2.8992687826393696e-07,
      "loss": 3.3288,
      "step": 83100
    },
    {
      "epoch": 0.8453877766927084,
      "grad_norm": 11.14752197265625,
      "learning_rate": 2.897399895532863e-07,
      "loss": 3.5397,
      "step": 83105
    },
    {
      "epoch": 0.8454386393229166,
      "grad_norm": 8.81459903717041,
      "learning_rate": 2.895531573916147e-07,
      "loss": 3.3393,
      "step": 83110
    },
    {
      "epoch": 0.845489501953125,
      "grad_norm": 14.247279167175293,
      "learning_rate": 2.893663817837017e-07,
      "loss": 3.4098,
      "step": 83115
    },
    {
      "epoch": 0.8455403645833334,
      "grad_norm": 7.800309658050537,
      "learning_rate": 2.891796627343252e-07,
      "loss": 3.3938,
      "step": 83120
    },
    {
      "epoch": 0.8455912272135416,
      "grad_norm": 14.212621688842773,
      "learning_rate": 2.8899300024826326e-07,
      "loss": 3.0912,
      "step": 83125
    },
    {
      "epoch": 0.84564208984375,
      "grad_norm": 12.19606876373291,
      "learning_rate": 2.8880639433029155e-07,
      "loss": 3.509,
      "step": 83130
    },
    {
      "epoch": 0.8456929524739584,
      "grad_norm": 13.947073936462402,
      "learning_rate": 2.886198449851843e-07,
      "loss": 2.942,
      "step": 83135
    },
    {
      "epoch": 0.8457438151041666,
      "grad_norm": 13.300604820251465,
      "learning_rate": 2.884333522177135e-07,
      "loss": 3.3743,
      "step": 83140
    },
    {
      "epoch": 0.845794677734375,
      "grad_norm": 10.905712127685547,
      "learning_rate": 2.8824691603265205e-07,
      "loss": 5.0671,
      "step": 83145
    },
    {
      "epoch": 0.8458455403645834,
      "grad_norm": 13.009864807128906,
      "learning_rate": 2.880605364347683e-07,
      "loss": 3.2609,
      "step": 83150
    },
    {
      "epoch": 0.8458964029947916,
      "grad_norm": 13.033794403076172,
      "learning_rate": 2.8787421342883174e-07,
      "loss": 3.1639,
      "step": 83155
    },
    {
      "epoch": 0.845947265625,
      "grad_norm": 15.466184616088867,
      "learning_rate": 2.876879470196087e-07,
      "loss": 3.0835,
      "step": 83160
    },
    {
      "epoch": 0.8459981282552084,
      "grad_norm": 10.184174537658691,
      "learning_rate": 2.8750173721186576e-07,
      "loss": 3.2319,
      "step": 83165
    },
    {
      "epoch": 0.8460489908854166,
      "grad_norm": 12.413190841674805,
      "learning_rate": 2.873155840103664e-07,
      "loss": 3.7967,
      "step": 83170
    },
    {
      "epoch": 0.846099853515625,
      "grad_norm": 12.316962242126465,
      "learning_rate": 2.8712948741987296e-07,
      "loss": 3.5576,
      "step": 83175
    },
    {
      "epoch": 0.8461507161458334,
      "grad_norm": 13.641777992248535,
      "learning_rate": 2.869434474451471e-07,
      "loss": 3.7668,
      "step": 83180
    },
    {
      "epoch": 0.8462015787760416,
      "grad_norm": 15.345667839050293,
      "learning_rate": 2.86757464090949e-07,
      "loss": 3.6061,
      "step": 83185
    },
    {
      "epoch": 0.84625244140625,
      "grad_norm": 8.986593246459961,
      "learning_rate": 2.8657153736203646e-07,
      "loss": 3.1737,
      "step": 83190
    },
    {
      "epoch": 0.8463033040364584,
      "grad_norm": 8.112719535827637,
      "learning_rate": 2.8638566726316595e-07,
      "loss": 3.4327,
      "step": 83195
    },
    {
      "epoch": 0.8463541666666666,
      "grad_norm": 6.413035869598389,
      "learning_rate": 2.861998537990937e-07,
      "loss": 3.2719,
      "step": 83200
    },
    {
      "epoch": 0.846405029296875,
      "grad_norm": 16.672090530395508,
      "learning_rate": 2.860140969745739e-07,
      "loss": 3.1371,
      "step": 83205
    },
    {
      "epoch": 0.8464558919270834,
      "grad_norm": 7.789152145385742,
      "learning_rate": 2.858283967943587e-07,
      "loss": 3.3255,
      "step": 83210
    },
    {
      "epoch": 0.8465067545572916,
      "grad_norm": 10.188570976257324,
      "learning_rate": 2.8564275326319917e-07,
      "loss": 3.5444,
      "step": 83215
    },
    {
      "epoch": 0.8465576171875,
      "grad_norm": 8.872726440429688,
      "learning_rate": 2.8545716638584437e-07,
      "loss": 3.2315,
      "step": 83220
    },
    {
      "epoch": 0.8466084798177084,
      "grad_norm": 10.287165641784668,
      "learning_rate": 2.8527163616704296e-07,
      "loss": 3.4485,
      "step": 83225
    },
    {
      "epoch": 0.8466593424479166,
      "grad_norm": 11.121027946472168,
      "learning_rate": 2.8508616261154224e-07,
      "loss": 3.1061,
      "step": 83230
    },
    {
      "epoch": 0.846710205078125,
      "grad_norm": 9.999377250671387,
      "learning_rate": 2.849007457240868e-07,
      "loss": 3.4964,
      "step": 83235
    },
    {
      "epoch": 0.8467610677083334,
      "grad_norm": 10.139492988586426,
      "learning_rate": 2.847153855094212e-07,
      "loss": 3.1607,
      "step": 83240
    },
    {
      "epoch": 0.8468119303385416,
      "grad_norm": 10.248892784118652,
      "learning_rate": 2.8453008197228736e-07,
      "loss": 2.9952,
      "step": 83245
    },
    {
      "epoch": 0.84686279296875,
      "grad_norm": 13.99163818359375,
      "learning_rate": 2.843448351174255e-07,
      "loss": 2.7576,
      "step": 83250
    },
    {
      "epoch": 0.8469136555989584,
      "grad_norm": 13.969670295715332,
      "learning_rate": 2.841596449495762e-07,
      "loss": 3.3207,
      "step": 83255
    },
    {
      "epoch": 0.8469645182291666,
      "grad_norm": 12.393842697143555,
      "learning_rate": 2.839745114734777e-07,
      "loss": 3.2999,
      "step": 83260
    },
    {
      "epoch": 0.847015380859375,
      "grad_norm": 10.31478214263916,
      "learning_rate": 2.8378943469386577e-07,
      "loss": 3.0767,
      "step": 83265
    },
    {
      "epoch": 0.8470662434895834,
      "grad_norm": 14.564029693603516,
      "learning_rate": 2.8360441461547595e-07,
      "loss": 3.316,
      "step": 83270
    },
    {
      "epoch": 0.8471171061197916,
      "grad_norm": 9.800872802734375,
      "learning_rate": 2.8341945124304133e-07,
      "loss": 3.4844,
      "step": 83275
    },
    {
      "epoch": 0.84716796875,
      "grad_norm": 15.539414405822754,
      "learning_rate": 2.832345445812948e-07,
      "loss": 3.5376,
      "step": 83280
    },
    {
      "epoch": 0.8472188313802084,
      "grad_norm": 40.019187927246094,
      "learning_rate": 2.8304969463496754e-07,
      "loss": 3.3367,
      "step": 83285
    },
    {
      "epoch": 0.8472696940104166,
      "grad_norm": 9.034164428710938,
      "learning_rate": 2.828649014087881e-07,
      "loss": 3.3207,
      "step": 83290
    },
    {
      "epoch": 0.847320556640625,
      "grad_norm": 7.02910041809082,
      "learning_rate": 2.826801649074845e-07,
      "loss": 3.1843,
      "step": 83295
    },
    {
      "epoch": 0.8473714192708334,
      "grad_norm": 14.586396217346191,
      "learning_rate": 2.8249548513578364e-07,
      "loss": 3.238,
      "step": 83300
    },
    {
      "epoch": 0.8474222819010416,
      "grad_norm": 8.197888374328613,
      "learning_rate": 2.823108620984097e-07,
      "loss": 3.5281,
      "step": 83305
    },
    {
      "epoch": 0.84747314453125,
      "grad_norm": 9.780033111572266,
      "learning_rate": 2.82126295800087e-07,
      "loss": 3.3656,
      "step": 83310
    },
    {
      "epoch": 0.8475240071614584,
      "grad_norm": 14.232123374938965,
      "learning_rate": 2.81941786245537e-07,
      "loss": 3.7687,
      "step": 83315
    },
    {
      "epoch": 0.8475748697916666,
      "grad_norm": 12.66684341430664,
      "learning_rate": 2.817573334394813e-07,
      "loss": 3.492,
      "step": 83320
    },
    {
      "epoch": 0.847625732421875,
      "grad_norm": 9.312463760375977,
      "learning_rate": 2.8157293738663826e-07,
      "loss": 3.2143,
      "step": 83325
    },
    {
      "epoch": 0.8476765950520834,
      "grad_norm": 9.992286682128906,
      "learning_rate": 2.8138859809172467e-07,
      "loss": 3.3887,
      "step": 83330
    },
    {
      "epoch": 0.8477274576822916,
      "grad_norm": 19.722196578979492,
      "learning_rate": 2.812043155594593e-07,
      "loss": 3.2264,
      "step": 83335
    },
    {
      "epoch": 0.8477783203125,
      "grad_norm": 10.095829963684082,
      "learning_rate": 2.810200897945553e-07,
      "loss": 3.257,
      "step": 83340
    },
    {
      "epoch": 0.8478291829427084,
      "grad_norm": 14.524836540222168,
      "learning_rate": 2.808359208017264e-07,
      "loss": 3.3492,
      "step": 83345
    },
    {
      "epoch": 0.8478800455729166,
      "grad_norm": 10.989709854125977,
      "learning_rate": 2.8065180858568406e-07,
      "loss": 3.4612,
      "step": 83350
    },
    {
      "epoch": 0.847930908203125,
      "grad_norm": 13.342979431152344,
      "learning_rate": 2.804677531511391e-07,
      "loss": 3.2874,
      "step": 83355
    },
    {
      "epoch": 0.8479817708333334,
      "grad_norm": 15.804058074951172,
      "learning_rate": 2.8028375450280117e-07,
      "loss": 3.1215,
      "step": 83360
    },
    {
      "epoch": 0.8480326334635416,
      "grad_norm": 9.274304389953613,
      "learning_rate": 2.8009981264537715e-07,
      "loss": 3.5346,
      "step": 83365
    },
    {
      "epoch": 0.84808349609375,
      "grad_norm": 10.4225492477417,
      "learning_rate": 2.799159275835728e-07,
      "loss": 3.4584,
      "step": 83370
    },
    {
      "epoch": 0.8481343587239584,
      "grad_norm": 9.473432540893555,
      "learning_rate": 2.7973209932209367e-07,
      "loss": 3.2326,
      "step": 83375
    },
    {
      "epoch": 0.8481852213541666,
      "grad_norm": 13.394980430603027,
      "learning_rate": 2.7954832786564216e-07,
      "loss": 3.5046,
      "step": 83380
    },
    {
      "epoch": 0.848236083984375,
      "grad_norm": 12.589401245117188,
      "learning_rate": 2.793646132189207e-07,
      "loss": 3.2171,
      "step": 83385
    },
    {
      "epoch": 0.8482869466145834,
      "grad_norm": 14.52203369140625,
      "learning_rate": 2.7918095538662873e-07,
      "loss": 3.263,
      "step": 83390
    },
    {
      "epoch": 0.8483378092447916,
      "grad_norm": 7.75312614440918,
      "learning_rate": 2.7899735437346644e-07,
      "loss": 3.1735,
      "step": 83395
    },
    {
      "epoch": 0.848388671875,
      "grad_norm": 12.05152416229248,
      "learning_rate": 2.788138101841301e-07,
      "loss": 3.3721,
      "step": 83400
    },
    {
      "epoch": 0.8484395345052084,
      "grad_norm": 15.198909759521484,
      "learning_rate": 2.786303228233156e-07,
      "loss": 3.2584,
      "step": 83405
    },
    {
      "epoch": 0.8484903971354166,
      "grad_norm": 12.124019622802734,
      "learning_rate": 2.784468922957176e-07,
      "loss": 3.3842,
      "step": 83410
    },
    {
      "epoch": 0.848541259765625,
      "grad_norm": 11.108325004577637,
      "learning_rate": 2.782635186060298e-07,
      "loss": 3.3118,
      "step": 83415
    },
    {
      "epoch": 0.8485921223958334,
      "grad_norm": 8.87848949432373,
      "learning_rate": 2.780802017589435e-07,
      "loss": 3.3771,
      "step": 83420
    },
    {
      "epoch": 0.8486429850260416,
      "grad_norm": 8.800379753112793,
      "learning_rate": 2.7789694175914787e-07,
      "loss": 3.3291,
      "step": 83425
    },
    {
      "epoch": 0.84869384765625,
      "grad_norm": 10.400461196899414,
      "learning_rate": 2.777137386113329e-07,
      "loss": 3.203,
      "step": 83430
    },
    {
      "epoch": 0.8487447102864584,
      "grad_norm": 13.805158615112305,
      "learning_rate": 2.775305923201846e-07,
      "loss": 3.357,
      "step": 83435
    },
    {
      "epoch": 0.8487955729166666,
      "grad_norm": 12.790278434753418,
      "learning_rate": 2.7734750289038994e-07,
      "loss": 3.2673,
      "step": 83440
    },
    {
      "epoch": 0.848846435546875,
      "grad_norm": 6.896450996398926,
      "learning_rate": 2.7716447032663215e-07,
      "loss": 3.4602,
      "step": 83445
    },
    {
      "epoch": 0.8488972981770834,
      "grad_norm": 13.737564086914062,
      "learning_rate": 2.769814946335947e-07,
      "loss": 3.5267,
      "step": 83450
    },
    {
      "epoch": 0.8489481608072916,
      "grad_norm": 8.333404541015625,
      "learning_rate": 2.7679857581595913e-07,
      "loss": 3.6422,
      "step": 83455
    },
    {
      "epoch": 0.8489990234375,
      "grad_norm": 20.389053344726562,
      "learning_rate": 2.7661571387840425e-07,
      "loss": 3.3173,
      "step": 83460
    },
    {
      "epoch": 0.8490498860677084,
      "grad_norm": 14.098817825317383,
      "learning_rate": 2.764329088256096e-07,
      "loss": 3.1836,
      "step": 83465
    },
    {
      "epoch": 0.8491007486979166,
      "grad_norm": 10.861899375915527,
      "learning_rate": 2.7625016066225223e-07,
      "loss": 3.4802,
      "step": 83470
    },
    {
      "epoch": 0.849151611328125,
      "grad_norm": 13.729740142822266,
      "learning_rate": 2.7606746939300766e-07,
      "loss": 3.28,
      "step": 83475
    },
    {
      "epoch": 0.8492024739583334,
      "grad_norm": 8.617941856384277,
      "learning_rate": 2.758848350225496e-07,
      "loss": 3.288,
      "step": 83480
    },
    {
      "epoch": 0.8492533365885416,
      "grad_norm": 8.380363464355469,
      "learning_rate": 2.7570225755554985e-07,
      "loss": 3.4407,
      "step": 83485
    },
    {
      "epoch": 0.84930419921875,
      "grad_norm": 8.086506843566895,
      "learning_rate": 2.7551973699668166e-07,
      "loss": 3.5749,
      "step": 83490
    },
    {
      "epoch": 0.8493550618489584,
      "grad_norm": 11.313240051269531,
      "learning_rate": 2.753372733506135e-07,
      "loss": 3.5966,
      "step": 83495
    },
    {
      "epoch": 0.8494059244791666,
      "grad_norm": 11.246660232543945,
      "learning_rate": 2.7515486662201414e-07,
      "loss": 3.0105,
      "step": 83500
    },
    {
      "epoch": 0.849456787109375,
      "grad_norm": 10.222160339355469,
      "learning_rate": 2.749725168155498e-07,
      "loss": 3.5749,
      "step": 83505
    },
    {
      "epoch": 0.8495076497395834,
      "grad_norm": 11.909274101257324,
      "learning_rate": 2.74790223935886e-07,
      "loss": 3.535,
      "step": 83510
    },
    {
      "epoch": 0.8495585123697916,
      "grad_norm": 18.18555450439453,
      "learning_rate": 2.746079879876876e-07,
      "loss": 3.3431,
      "step": 83515
    },
    {
      "epoch": 0.849609375,
      "grad_norm": 8.408985137939453,
      "learning_rate": 2.744258089756155e-07,
      "loss": 3.1151,
      "step": 83520
    },
    {
      "epoch": 0.8496602376302084,
      "grad_norm": 12.356221199035645,
      "learning_rate": 2.742436869043324e-07,
      "loss": 3.3734,
      "step": 83525
    },
    {
      "epoch": 0.8497111002604166,
      "grad_norm": 14.428342819213867,
      "learning_rate": 2.7406162177849693e-07,
      "loss": 3.3936,
      "step": 83530
    },
    {
      "epoch": 0.849761962890625,
      "grad_norm": 7.772458076477051,
      "learning_rate": 2.738796136027666e-07,
      "loss": 3.1878,
      "step": 83535
    },
    {
      "epoch": 0.8498128255208334,
      "grad_norm": 10.488436698913574,
      "learning_rate": 2.7369766238179884e-07,
      "loss": 3.2562,
      "step": 83540
    },
    {
      "epoch": 0.8498636881510416,
      "grad_norm": 10.883633613586426,
      "learning_rate": 2.735157681202491e-07,
      "loss": 3.2546,
      "step": 83545
    },
    {
      "epoch": 0.84991455078125,
      "grad_norm": 11.330202102661133,
      "learning_rate": 2.733339308227709e-07,
      "loss": 3.3776,
      "step": 83550
    },
    {
      "epoch": 0.8499654134114584,
      "grad_norm": 10.628450393676758,
      "learning_rate": 2.73152150494016e-07,
      "loss": 3.3326,
      "step": 83555
    },
    {
      "epoch": 0.8500162760416666,
      "grad_norm": 11.719513893127441,
      "learning_rate": 2.729704271386352e-07,
      "loss": 3.2302,
      "step": 83560
    },
    {
      "epoch": 0.850067138671875,
      "grad_norm": 13.615124702453613,
      "learning_rate": 2.727887607612784e-07,
      "loss": 3.3346,
      "step": 83565
    },
    {
      "epoch": 0.8501180013020834,
      "grad_norm": 14.910679817199707,
      "learning_rate": 2.726071513665934e-07,
      "loss": 3.3568,
      "step": 83570
    },
    {
      "epoch": 0.8501688639322916,
      "grad_norm": 12.558643341064453,
      "learning_rate": 2.724255989592264e-07,
      "loss": 3.2009,
      "step": 83575
    },
    {
      "epoch": 0.8502197265625,
      "grad_norm": 18.387767791748047,
      "learning_rate": 2.7224410354382227e-07,
      "loss": 3.0721,
      "step": 83580
    },
    {
      "epoch": 0.8502705891927084,
      "grad_norm": 14.17619514465332,
      "learning_rate": 2.720626651250249e-07,
      "loss": 2.9736,
      "step": 83585
    },
    {
      "epoch": 0.8503214518229166,
      "grad_norm": 9.583733558654785,
      "learning_rate": 2.718812837074761e-07,
      "loss": 3.5239,
      "step": 83590
    },
    {
      "epoch": 0.850372314453125,
      "grad_norm": 7.749961853027344,
      "learning_rate": 2.7169995929581666e-07,
      "loss": 3.2825,
      "step": 83595
    },
    {
      "epoch": 0.8504231770833334,
      "grad_norm": 7.960093021392822,
      "learning_rate": 2.715186918946852e-07,
      "loss": 3.2317,
      "step": 83600
    },
    {
      "epoch": 0.8504740397135416,
      "grad_norm": 12.65415096282959,
      "learning_rate": 2.713374815087202e-07,
      "loss": 3.2379,
      "step": 83605
    },
    {
      "epoch": 0.85052490234375,
      "grad_norm": 11.380934715270996,
      "learning_rate": 2.711563281425578e-07,
      "loss": 3.315,
      "step": 83610
    },
    {
      "epoch": 0.8505757649739584,
      "grad_norm": 10.991366386413574,
      "learning_rate": 2.7097523180083156e-07,
      "loss": 3.0801,
      "step": 83615
    },
    {
      "epoch": 0.8506266276041666,
      "grad_norm": 14.657071113586426,
      "learning_rate": 2.70794192488176e-07,
      "loss": 3.34,
      "step": 83620
    },
    {
      "epoch": 0.850677490234375,
      "grad_norm": 8.607914924621582,
      "learning_rate": 2.706132102092229e-07,
      "loss": 3.1656,
      "step": 83625
    },
    {
      "epoch": 0.8507283528645834,
      "grad_norm": 14.747588157653809,
      "learning_rate": 2.704322849686025e-07,
      "loss": 3.0968,
      "step": 83630
    },
    {
      "epoch": 0.8507792154947916,
      "grad_norm": 12.232775688171387,
      "learning_rate": 2.702514167709433e-07,
      "loss": 3.1602,
      "step": 83635
    },
    {
      "epoch": 0.850830078125,
      "grad_norm": 8.508930206298828,
      "learning_rate": 2.7007060562087293e-07,
      "loss": 3.322,
      "step": 83640
    },
    {
      "epoch": 0.8508809407552084,
      "grad_norm": 9.501651763916016,
      "learning_rate": 2.6988985152301805e-07,
      "loss": 3.1027,
      "step": 83645
    },
    {
      "epoch": 0.8509318033854166,
      "grad_norm": 15.47403621673584,
      "learning_rate": 2.6970915448200267e-07,
      "loss": 3.0541,
      "step": 83650
    },
    {
      "epoch": 0.850982666015625,
      "grad_norm": 12.417950630187988,
      "learning_rate": 2.6952851450244953e-07,
      "loss": 3.3576,
      "step": 83655
    },
    {
      "epoch": 0.8510335286458334,
      "grad_norm": 12.438616752624512,
      "learning_rate": 2.6934793158898105e-07,
      "loss": 3.5725,
      "step": 83660
    },
    {
      "epoch": 0.8510843912760416,
      "grad_norm": 13.789573669433594,
      "learning_rate": 2.691674057462165e-07,
      "loss": 3.4432,
      "step": 83665
    },
    {
      "epoch": 0.85113525390625,
      "grad_norm": 10.428175926208496,
      "learning_rate": 2.6898693697877593e-07,
      "loss": 3.5053,
      "step": 83670
    },
    {
      "epoch": 0.8511861165364584,
      "grad_norm": 9.609519004821777,
      "learning_rate": 2.6880652529127497e-07,
      "loss": 3.2183,
      "step": 83675
    },
    {
      "epoch": 0.8512369791666666,
      "grad_norm": 14.939981460571289,
      "learning_rate": 2.6862617068833086e-07,
      "loss": 3.2356,
      "step": 83680
    },
    {
      "epoch": 0.851287841796875,
      "grad_norm": 15.604842185974121,
      "learning_rate": 2.6844587317455704e-07,
      "loss": 3.3283,
      "step": 83685
    },
    {
      "epoch": 0.8513387044270834,
      "grad_norm": 8.777827262878418,
      "learning_rate": 2.682656327545663e-07,
      "loss": 3.263,
      "step": 83690
    },
    {
      "epoch": 0.8513895670572916,
      "grad_norm": 13.913224220275879,
      "learning_rate": 2.6808544943297044e-07,
      "loss": 3.1008,
      "step": 83695
    },
    {
      "epoch": 0.8514404296875,
      "grad_norm": 11.493755340576172,
      "learning_rate": 2.679053232143797e-07,
      "loss": 3.2256,
      "step": 83700
    },
    {
      "epoch": 0.8514912923177084,
      "grad_norm": 15.015609741210938,
      "learning_rate": 2.6772525410340225e-07,
      "loss": 3.3691,
      "step": 83705
    },
    {
      "epoch": 0.8515421549479166,
      "grad_norm": 13.246819496154785,
      "learning_rate": 2.6754524210464513e-07,
      "loss": 3.3471,
      "step": 83710
    },
    {
      "epoch": 0.851593017578125,
      "grad_norm": 14.715752601623535,
      "learning_rate": 2.673652872227131e-07,
      "loss": 3.4959,
      "step": 83715
    },
    {
      "epoch": 0.8516438802083334,
      "grad_norm": 11.939118385314941,
      "learning_rate": 2.6718538946221116e-07,
      "loss": 3.5762,
      "step": 83720
    },
    {
      "epoch": 0.8516947428385416,
      "grad_norm": 11.789021492004395,
      "learning_rate": 2.670055488277423e-07,
      "loss": 3.1403,
      "step": 83725
    },
    {
      "epoch": 0.85174560546875,
      "grad_norm": 18.092769622802734,
      "learning_rate": 2.6682576532390673e-07,
      "loss": 3.8434,
      "step": 83730
    },
    {
      "epoch": 0.8517964680989584,
      "grad_norm": 10.054718971252441,
      "learning_rate": 2.666460389553052e-07,
      "loss": 3.3368,
      "step": 83735
    },
    {
      "epoch": 0.8518473307291666,
      "grad_norm": 11.558505058288574,
      "learning_rate": 2.6646636972653546e-07,
      "loss": 3.333,
      "step": 83740
    },
    {
      "epoch": 0.851898193359375,
      "grad_norm": 7.389967918395996,
      "learning_rate": 2.6628675764219346e-07,
      "loss": 3.3784,
      "step": 83745
    },
    {
      "epoch": 0.8519490559895834,
      "grad_norm": 14.015900611877441,
      "learning_rate": 2.6610720270687563e-07,
      "loss": 3.8593,
      "step": 83750
    },
    {
      "epoch": 0.8519999186197916,
      "grad_norm": 12.08255672454834,
      "learning_rate": 2.6592770492517574e-07,
      "loss": 3.3945,
      "step": 83755
    },
    {
      "epoch": 0.85205078125,
      "grad_norm": 8.277338981628418,
      "learning_rate": 2.657482643016862e-07,
      "loss": 3.1407,
      "step": 83760
    },
    {
      "epoch": 0.8521016438802084,
      "grad_norm": 16.168481826782227,
      "learning_rate": 2.655688808409976e-07,
      "loss": 3.4876,
      "step": 83765
    },
    {
      "epoch": 0.8521525065104166,
      "grad_norm": 13.869182586669922,
      "learning_rate": 2.6538955454769896e-07,
      "loss": 3.3737,
      "step": 83770
    },
    {
      "epoch": 0.852203369140625,
      "grad_norm": 7.083217144012451,
      "learning_rate": 2.6521028542637913e-07,
      "loss": 3.1822,
      "step": 83775
    },
    {
      "epoch": 0.8522542317708334,
      "grad_norm": 12.952024459838867,
      "learning_rate": 2.65031073481625e-07,
      "loss": 3.4927,
      "step": 83780
    },
    {
      "epoch": 0.8523050944010416,
      "grad_norm": 11.621052742004395,
      "learning_rate": 2.6485191871802095e-07,
      "loss": 3.1839,
      "step": 83785
    },
    {
      "epoch": 0.85235595703125,
      "grad_norm": 11.824081420898438,
      "learning_rate": 2.646728211401503e-07,
      "loss": 3.0807,
      "step": 83790
    },
    {
      "epoch": 0.8524068196614584,
      "grad_norm": 13.243668556213379,
      "learning_rate": 2.644937807525957e-07,
      "loss": 3.215,
      "step": 83795
    },
    {
      "epoch": 0.8524576822916666,
      "grad_norm": 12.352242469787598,
      "learning_rate": 2.6431479755993833e-07,
      "loss": 3.1282,
      "step": 83800
    },
    {
      "epoch": 0.852508544921875,
      "grad_norm": 9.239322662353516,
      "learning_rate": 2.6413587156675693e-07,
      "loss": 3.2456,
      "step": 83805
    },
    {
      "epoch": 0.8525594075520834,
      "grad_norm": 11.568145751953125,
      "learning_rate": 2.6395700277762894e-07,
      "loss": 3.4627,
      "step": 83810
    },
    {
      "epoch": 0.8526102701822916,
      "grad_norm": 13.08260440826416,
      "learning_rate": 2.6377819119713125e-07,
      "loss": 3.4344,
      "step": 83815
    },
    {
      "epoch": 0.8526611328125,
      "grad_norm": 12.213603019714355,
      "learning_rate": 2.635994368298381e-07,
      "loss": 3.6741,
      "step": 83820
    },
    {
      "epoch": 0.8527119954427084,
      "grad_norm": 12.419811248779297,
      "learning_rate": 2.6342073968032346e-07,
      "loss": 3.2684,
      "step": 83825
    },
    {
      "epoch": 0.8527628580729166,
      "grad_norm": 6.682579040527344,
      "learning_rate": 2.632420997531596e-07,
      "loss": 3.0465,
      "step": 83830
    },
    {
      "epoch": 0.852813720703125,
      "grad_norm": 11.076421737670898,
      "learning_rate": 2.630635170529161e-07,
      "loss": 3.2169,
      "step": 83835
    },
    {
      "epoch": 0.8528645833333334,
      "grad_norm": 10.50160026550293,
      "learning_rate": 2.628849915841625e-07,
      "loss": 3.3344,
      "step": 83840
    },
    {
      "epoch": 0.8529154459635416,
      "grad_norm": 12.040009498596191,
      "learning_rate": 2.6270652335146556e-07,
      "loss": 3.4984,
      "step": 83845
    },
    {
      "epoch": 0.85296630859375,
      "grad_norm": 13.960234642028809,
      "learning_rate": 2.62528112359392e-07,
      "loss": 3.0605,
      "step": 83850
    },
    {
      "epoch": 0.8530171712239584,
      "grad_norm": 9.576654434204102,
      "learning_rate": 2.6234975861250703e-07,
      "loss": 3.4375,
      "step": 83855
    },
    {
      "epoch": 0.8530680338541666,
      "grad_norm": 9.949191093444824,
      "learning_rate": 2.6217146211537286e-07,
      "loss": 3.7202,
      "step": 83860
    },
    {
      "epoch": 0.853118896484375,
      "grad_norm": 10.487893104553223,
      "learning_rate": 2.6199322287255105e-07,
      "loss": 3.2605,
      "step": 83865
    },
    {
      "epoch": 0.8531697591145834,
      "grad_norm": 10.049798965454102,
      "learning_rate": 2.6181504088860247e-07,
      "loss": 3.539,
      "step": 83870
    },
    {
      "epoch": 0.8532206217447916,
      "grad_norm": 13.798386573791504,
      "learning_rate": 2.616369161680851e-07,
      "loss": 3.3636,
      "step": 83875
    },
    {
      "epoch": 0.853271484375,
      "grad_norm": 12.284562110900879,
      "learning_rate": 2.6145884871555725e-07,
      "loss": 3.4647,
      "step": 83880
    },
    {
      "epoch": 0.8533223470052084,
      "grad_norm": 14.121574401855469,
      "learning_rate": 2.612808385355739e-07,
      "loss": 3.2157,
      "step": 83885
    },
    {
      "epoch": 0.8533732096354166,
      "grad_norm": 12.580548286437988,
      "learning_rate": 2.611028856326897e-07,
      "loss": 3.4041,
      "step": 83890
    },
    {
      "epoch": 0.853424072265625,
      "grad_norm": 8.253116607666016,
      "learning_rate": 2.609249900114577e-07,
      "loss": 3.7125,
      "step": 83895
    },
    {
      "epoch": 0.8534749348958334,
      "grad_norm": 15.323606491088867,
      "learning_rate": 2.607471516764287e-07,
      "loss": 3.099,
      "step": 83900
    },
    {
      "epoch": 0.8535257975260416,
      "grad_norm": 12.911646842956543,
      "learning_rate": 2.6056937063215297e-07,
      "loss": 3.4895,
      "step": 83905
    },
    {
      "epoch": 0.85357666015625,
      "grad_norm": 13.993356704711914,
      "learning_rate": 2.6039164688317966e-07,
      "loss": 3.4062,
      "step": 83910
    },
    {
      "epoch": 0.8536275227864584,
      "grad_norm": 12.202286720275879,
      "learning_rate": 2.602139804340553e-07,
      "loss": 3.5569,
      "step": 83915
    },
    {
      "epoch": 0.8536783854166666,
      "grad_norm": 21.464672088623047,
      "learning_rate": 2.6003637128932476e-07,
      "loss": 3.174,
      "step": 83920
    },
    {
      "epoch": 0.853729248046875,
      "grad_norm": 12.035834312438965,
      "learning_rate": 2.5985881945353315e-07,
      "loss": 3.4139,
      "step": 83925
    },
    {
      "epoch": 0.8537801106770834,
      "grad_norm": 8.957596778869629,
      "learning_rate": 2.596813249312222e-07,
      "loss": 3.1844,
      "step": 83930
    },
    {
      "epoch": 0.8538309733072916,
      "grad_norm": 8.745210647583008,
      "learning_rate": 2.5950388772693403e-07,
      "loss": 3.3869,
      "step": 83935
    },
    {
      "epoch": 0.8538818359375,
      "grad_norm": 14.097525596618652,
      "learning_rate": 2.5932650784520755e-07,
      "loss": 3.2003,
      "step": 83940
    },
    {
      "epoch": 0.8539326985677084,
      "grad_norm": 11.521891593933105,
      "learning_rate": 2.591491852905814e-07,
      "loss": 3.1345,
      "step": 83945
    },
    {
      "epoch": 0.8539835611979166,
      "grad_norm": 11.581704139709473,
      "learning_rate": 2.5897192006759193e-07,
      "loss": 3.2405,
      "step": 83950
    },
    {
      "epoch": 0.854034423828125,
      "grad_norm": 8.957106590270996,
      "learning_rate": 2.587947121807752e-07,
      "loss": 3.6836,
      "step": 83955
    },
    {
      "epoch": 0.8540852864583334,
      "grad_norm": 9.746102333068848,
      "learning_rate": 2.5861756163466423e-07,
      "loss": 3.3639,
      "step": 83960
    },
    {
      "epoch": 0.8541361490885416,
      "grad_norm": 15.997204780578613,
      "learning_rate": 2.584404684337918e-07,
      "loss": 3.0967,
      "step": 83965
    },
    {
      "epoch": 0.85418701171875,
      "grad_norm": 14.250624656677246,
      "learning_rate": 2.5826343258268883e-07,
      "loss": 3.0947,
      "step": 83970
    },
    {
      "epoch": 0.8542378743489584,
      "grad_norm": 13.187626838684082,
      "learning_rate": 2.5808645408588406e-07,
      "loss": 3.3481,
      "step": 83975
    },
    {
      "epoch": 0.8542887369791666,
      "grad_norm": 7.7368645668029785,
      "learning_rate": 2.5790953294790616e-07,
      "loss": 2.9572,
      "step": 83980
    },
    {
      "epoch": 0.854339599609375,
      "grad_norm": 9.206607818603516,
      "learning_rate": 2.577326691732818e-07,
      "loss": 3.1199,
      "step": 83985
    },
    {
      "epoch": 0.8543904622395834,
      "grad_norm": 14.619028091430664,
      "learning_rate": 2.575558627665356e-07,
      "loss": 3.4817,
      "step": 83990
    },
    {
      "epoch": 0.8544413248697916,
      "grad_norm": 11.724611282348633,
      "learning_rate": 2.57379113732191e-07,
      "loss": 3.3051,
      "step": 83995
    },
    {
      "epoch": 0.8544921875,
      "grad_norm": 17.9774227142334,
      "learning_rate": 2.5720242207476987e-07,
      "loss": 3.3657,
      "step": 84000
    },
    {
      "epoch": 0.8545430501302084,
      "grad_norm": 14.093949317932129,
      "learning_rate": 2.570257877987931e-07,
      "loss": 3.3961,
      "step": 84005
    },
    {
      "epoch": 0.8545939127604166,
      "grad_norm": 7.811964988708496,
      "learning_rate": 2.5684921090878056e-07,
      "loss": 3.2081,
      "step": 84010
    },
    {
      "epoch": 0.854644775390625,
      "grad_norm": 9.574098587036133,
      "learning_rate": 2.566726914092488e-07,
      "loss": 3.3077,
      "step": 84015
    },
    {
      "epoch": 0.8546956380208334,
      "grad_norm": 10.139022827148438,
      "learning_rate": 2.5649622930471463e-07,
      "loss": 3.5255,
      "step": 84020
    },
    {
      "epoch": 0.8547465006510416,
      "grad_norm": 15.366785049438477,
      "learning_rate": 2.5631982459969314e-07,
      "loss": 3.4556,
      "step": 84025
    },
    {
      "epoch": 0.85479736328125,
      "grad_norm": 9.718144416809082,
      "learning_rate": 2.561434772986962e-07,
      "loss": 3.3304,
      "step": 84030
    },
    {
      "epoch": 0.8548482259114584,
      "grad_norm": 14.18233585357666,
      "learning_rate": 2.5596718740623694e-07,
      "loss": 3.207,
      "step": 84035
    },
    {
      "epoch": 0.8548990885416666,
      "grad_norm": 10.997235298156738,
      "learning_rate": 2.557909549268253e-07,
      "loss": 3.3991,
      "step": 84040
    },
    {
      "epoch": 0.854949951171875,
      "grad_norm": 7.477236270904541,
      "learning_rate": 2.5561477986497053e-07,
      "loss": 3.3484,
      "step": 84045
    },
    {
      "epoch": 0.8550008138020834,
      "grad_norm": 10.705188751220703,
      "learning_rate": 2.554386622251795e-07,
      "loss": 3.2581,
      "step": 84050
    },
    {
      "epoch": 0.8550516764322916,
      "grad_norm": 13.796873092651367,
      "learning_rate": 2.552626020119578e-07,
      "loss": 3.6276,
      "step": 84055
    },
    {
      "epoch": 0.8551025390625,
      "grad_norm": 12.384038925170898,
      "learning_rate": 2.5508659922981046e-07,
      "loss": 3.3599,
      "step": 84060
    },
    {
      "epoch": 0.8551534016927084,
      "grad_norm": 14.427067756652832,
      "learning_rate": 2.549106538832408e-07,
      "loss": 3.4612,
      "step": 84065
    },
    {
      "epoch": 0.8552042643229166,
      "grad_norm": 11.042256355285645,
      "learning_rate": 2.547347659767499e-07,
      "loss": 3.1627,
      "step": 84070
    },
    {
      "epoch": 0.855255126953125,
      "grad_norm": 9.276700019836426,
      "learning_rate": 2.545589355148373e-07,
      "loss": 3.139,
      "step": 84075
    },
    {
      "epoch": 0.8553059895833334,
      "grad_norm": 10.519864082336426,
      "learning_rate": 2.5438316250200267e-07,
      "loss": 3.8058,
      "step": 84080
    },
    {
      "epoch": 0.8553568522135416,
      "grad_norm": 15.285600662231445,
      "learning_rate": 2.5420744694274216e-07,
      "loss": 3.1438,
      "step": 84085
    },
    {
      "epoch": 0.85540771484375,
      "grad_norm": 15.675111770629883,
      "learning_rate": 2.540317888415522e-07,
      "loss": 3.4444,
      "step": 84090
    },
    {
      "epoch": 0.8554585774739584,
      "grad_norm": 44.94129180908203,
      "learning_rate": 2.5385618820292613e-07,
      "loss": 3.3156,
      "step": 84095
    },
    {
      "epoch": 0.8555094401041666,
      "grad_norm": 16.871110916137695,
      "learning_rate": 2.536806450313575e-07,
      "loss": 3.1729,
      "step": 84100
    },
    {
      "epoch": 0.855560302734375,
      "grad_norm": 10.780709266662598,
      "learning_rate": 2.5350515933133643e-07,
      "loss": 3.9411,
      "step": 84105
    },
    {
      "epoch": 0.8556111653645834,
      "grad_norm": 14.381935119628906,
      "learning_rate": 2.533297311073538e-07,
      "loss": 3.2177,
      "step": 84110
    },
    {
      "epoch": 0.8556620279947916,
      "grad_norm": 9.522446632385254,
      "learning_rate": 2.531543603638978e-07,
      "loss": 3.5338,
      "step": 84115
    },
    {
      "epoch": 0.855712890625,
      "grad_norm": 13.749747276306152,
      "learning_rate": 2.5297904710545493e-07,
      "loss": 3.1861,
      "step": 84120
    },
    {
      "epoch": 0.8557637532552084,
      "grad_norm": 9.4033784866333,
      "learning_rate": 2.5280379133651055e-07,
      "loss": 3.4675,
      "step": 84125
    },
    {
      "epoch": 0.8558146158854166,
      "grad_norm": 14.979850769042969,
      "learning_rate": 2.52628593061548e-07,
      "loss": 3.5098,
      "step": 84130
    },
    {
      "epoch": 0.855865478515625,
      "grad_norm": 12.472001075744629,
      "learning_rate": 2.5245345228505024e-07,
      "loss": 3.4768,
      "step": 84135
    },
    {
      "epoch": 0.8559163411458334,
      "grad_norm": 14.229353904724121,
      "learning_rate": 2.5227836901149857e-07,
      "loss": 3.4148,
      "step": 84140
    },
    {
      "epoch": 0.8559672037760416,
      "grad_norm": 9.634638786315918,
      "learning_rate": 2.5210334324537204e-07,
      "loss": 3.5641,
      "step": 84145
    },
    {
      "epoch": 0.85601806640625,
      "grad_norm": 12.365317344665527,
      "learning_rate": 2.519283749911483e-07,
      "loss": 3.316,
      "step": 84150
    },
    {
      "epoch": 0.8560689290364584,
      "grad_norm": 11.534818649291992,
      "learning_rate": 2.5175346425330455e-07,
      "loss": 3.3737,
      "step": 84155
    },
    {
      "epoch": 0.8561197916666666,
      "grad_norm": 8.3718843460083,
      "learning_rate": 2.5157861103631524e-07,
      "loss": 3.1727,
      "step": 84160
    },
    {
      "epoch": 0.856170654296875,
      "grad_norm": 9.538040161132812,
      "learning_rate": 2.5140381534465465e-07,
      "loss": 3.3974,
      "step": 84165
    },
    {
      "epoch": 0.8562215169270834,
      "grad_norm": 14.558609962463379,
      "learning_rate": 2.512290771827941e-07,
      "loss": 3.2332,
      "step": 84170
    },
    {
      "epoch": 0.8562723795572916,
      "grad_norm": 7.958302021026611,
      "learning_rate": 2.510543965552048e-07,
      "loss": 3.1759,
      "step": 84175
    },
    {
      "epoch": 0.8563232421875,
      "grad_norm": 16.645124435424805,
      "learning_rate": 2.5087977346635596e-07,
      "loss": 3.3306,
      "step": 84180
    },
    {
      "epoch": 0.8563741048177084,
      "grad_norm": 8.358590126037598,
      "learning_rate": 2.507052079207145e-07,
      "loss": 3.3807,
      "step": 84185
    },
    {
      "epoch": 0.8564249674479166,
      "grad_norm": 10.768516540527344,
      "learning_rate": 2.505306999227472e-07,
      "loss": 3.2633,
      "step": 84190
    },
    {
      "epoch": 0.856475830078125,
      "grad_norm": 12.022039413452148,
      "learning_rate": 2.5035624947691907e-07,
      "loss": 3.7514,
      "step": 84195
    },
    {
      "epoch": 0.8565266927083334,
      "grad_norm": 15.358153343200684,
      "learning_rate": 2.5018185658769315e-07,
      "loss": 3.2351,
      "step": 84200
    },
    {
      "epoch": 0.8565775553385416,
      "grad_norm": 10.9464750289917,
      "learning_rate": 2.500075212595307e-07,
      "loss": 3.4272,
      "step": 84205
    },
    {
      "epoch": 0.85662841796875,
      "grad_norm": 14.145750045776367,
      "learning_rate": 2.498332434968928e-07,
      "loss": 3.379,
      "step": 84210
    },
    {
      "epoch": 0.8566792805989584,
      "grad_norm": 10.068707466125488,
      "learning_rate": 2.4965902330423785e-07,
      "loss": 3.0267,
      "step": 84215
    },
    {
      "epoch": 0.8567301432291666,
      "grad_norm": 17.326412200927734,
      "learning_rate": 2.4948486068602367e-07,
      "loss": 3.482,
      "step": 84220
    },
    {
      "epoch": 0.856781005859375,
      "grad_norm": 14.676739692687988,
      "learning_rate": 2.493107556467056e-07,
      "loss": 3.2957,
      "step": 84225
    },
    {
      "epoch": 0.8568318684895834,
      "grad_norm": 10.21787166595459,
      "learning_rate": 2.4913670819073863e-07,
      "loss": 2.9701,
      "step": 84230
    },
    {
      "epoch": 0.8568827311197916,
      "grad_norm": 10.94861888885498,
      "learning_rate": 2.489627183225757e-07,
      "loss": 3.6475,
      "step": 84235
    },
    {
      "epoch": 0.85693359375,
      "grad_norm": 14.115302085876465,
      "learning_rate": 2.4878878604666726e-07,
      "loss": 3.4225,
      "step": 84240
    },
    {
      "epoch": 0.8569844563802084,
      "grad_norm": 8.742629051208496,
      "learning_rate": 2.4861491136746436e-07,
      "loss": 3.5919,
      "step": 84245
    },
    {
      "epoch": 0.8570353190104166,
      "grad_norm": 15.903121948242188,
      "learning_rate": 2.484410942894158e-07,
      "loss": 3.4988,
      "step": 84250
    },
    {
      "epoch": 0.857086181640625,
      "grad_norm": 11.232239723205566,
      "learning_rate": 2.4826733481696815e-07,
      "loss": 3.3979,
      "step": 84255
    },
    {
      "epoch": 0.8571370442708334,
      "grad_norm": 12.156452178955078,
      "learning_rate": 2.480936329545666e-07,
      "loss": 3.6659,
      "step": 84260
    },
    {
      "epoch": 0.8571879069010416,
      "grad_norm": 8.747230529785156,
      "learning_rate": 2.479199887066555e-07,
      "loss": 3.8577,
      "step": 84265
    },
    {
      "epoch": 0.85723876953125,
      "grad_norm": 13.07475471496582,
      "learning_rate": 2.477464020776782e-07,
      "loss": 3.8486,
      "step": 84270
    },
    {
      "epoch": 0.8572896321614584,
      "grad_norm": 12.68109130859375,
      "learning_rate": 2.4757287307207555e-07,
      "loss": 3.4064,
      "step": 84275
    },
    {
      "epoch": 0.8573404947916666,
      "grad_norm": 15.027631759643555,
      "learning_rate": 2.473994016942868e-07,
      "loss": 3.2732,
      "step": 84280
    },
    {
      "epoch": 0.857391357421875,
      "grad_norm": 10.339942932128906,
      "learning_rate": 2.4722598794875014e-07,
      "loss": 3.049,
      "step": 84285
    },
    {
      "epoch": 0.8574422200520834,
      "grad_norm": 13.480106353759766,
      "learning_rate": 2.470526318399027e-07,
      "loss": 3.4281,
      "step": 84290
    },
    {
      "epoch": 0.8574930826822916,
      "grad_norm": 13.331635475158691,
      "learning_rate": 2.4687933337218e-07,
      "loss": 3.5205,
      "step": 84295
    },
    {
      "epoch": 0.8575439453125,
      "grad_norm": 10.807036399841309,
      "learning_rate": 2.467060925500156e-07,
      "loss": 3.108,
      "step": 84300
    },
    {
      "epoch": 0.8575948079427084,
      "grad_norm": 10.52746295928955,
      "learning_rate": 2.4653290937784127e-07,
      "loss": 3.3692,
      "step": 84305
    },
    {
      "epoch": 0.8576456705729166,
      "grad_norm": 11.065303802490234,
      "learning_rate": 2.463597838600887e-07,
      "loss": 3.4338,
      "step": 84310
    },
    {
      "epoch": 0.857696533203125,
      "grad_norm": 14.478635787963867,
      "learning_rate": 2.4618671600118633e-07,
      "loss": 3.6296,
      "step": 84315
    },
    {
      "epoch": 0.8577473958333334,
      "grad_norm": 11.326279640197754,
      "learning_rate": 2.460137058055628e-07,
      "loss": 3.1279,
      "step": 84320
    },
    {
      "epoch": 0.8577982584635416,
      "grad_norm": 11.119710922241211,
      "learning_rate": 2.4584075327764466e-07,
      "loss": 3.1166,
      "step": 84325
    },
    {
      "epoch": 0.85784912109375,
      "grad_norm": 10.756710052490234,
      "learning_rate": 2.456678584218566e-07,
      "loss": 3.1323,
      "step": 84330
    },
    {
      "epoch": 0.8578999837239584,
      "grad_norm": 9.484225273132324,
      "learning_rate": 2.4549502124262217e-07,
      "loss": 3.3545,
      "step": 84335
    },
    {
      "epoch": 0.8579508463541666,
      "grad_norm": 12.128405570983887,
      "learning_rate": 2.4532224174436266e-07,
      "loss": 3.3032,
      "step": 84340
    },
    {
      "epoch": 0.858001708984375,
      "grad_norm": 11.937442779541016,
      "learning_rate": 2.4514951993149933e-07,
      "loss": 3.6189,
      "step": 84345
    },
    {
      "epoch": 0.8580525716145834,
      "grad_norm": 8.71423053741455,
      "learning_rate": 2.4497685580845143e-07,
      "loss": 3.3089,
      "step": 84350
    },
    {
      "epoch": 0.8581034342447916,
      "grad_norm": 11.5895414352417,
      "learning_rate": 2.4480424937963596e-07,
      "loss": 3.1087,
      "step": 84355
    },
    {
      "epoch": 0.858154296875,
      "grad_norm": 16.375810623168945,
      "learning_rate": 2.446317006494692e-07,
      "loss": 3.3446,
      "step": 84360
    },
    {
      "epoch": 0.8582051595052084,
      "grad_norm": 7.415759086608887,
      "learning_rate": 2.444592096223658e-07,
      "loss": 3.6263,
      "step": 84365
    },
    {
      "epoch": 0.8582560221354166,
      "grad_norm": 9.510431289672852,
      "learning_rate": 2.442867763027387e-07,
      "loss": 3.3406,
      "step": 84370
    },
    {
      "epoch": 0.858306884765625,
      "grad_norm": 15.061565399169922,
      "learning_rate": 2.4411440069500023e-07,
      "loss": 3.4792,
      "step": 84375
    },
    {
      "epoch": 0.8583577473958334,
      "grad_norm": 17.945556640625,
      "learning_rate": 2.4394208280355965e-07,
      "loss": 3.63,
      "step": 84380
    },
    {
      "epoch": 0.8584086100260416,
      "grad_norm": 14.503132820129395,
      "learning_rate": 2.437698226328267e-07,
      "loss": 3.1014,
      "step": 84385
    },
    {
      "epoch": 0.85845947265625,
      "grad_norm": 11.974645614624023,
      "learning_rate": 2.435976201872078e-07,
      "loss": 3.0912,
      "step": 84390
    },
    {
      "epoch": 0.8585103352864584,
      "grad_norm": 11.591052055358887,
      "learning_rate": 2.4342547547110877e-07,
      "loss": 3.1476,
      "step": 84395
    },
    {
      "epoch": 0.8585611979166666,
      "grad_norm": 11.908485412597656,
      "learning_rate": 2.4325338848893375e-07,
      "loss": 3.0585,
      "step": 84400
    },
    {
      "epoch": 0.858612060546875,
      "grad_norm": 13.596232414245605,
      "learning_rate": 2.430813592450865e-07,
      "loss": 3.4634,
      "step": 84405
    },
    {
      "epoch": 0.8586629231770834,
      "grad_norm": 8.452485084533691,
      "learning_rate": 2.4290938774396783e-07,
      "loss": 3.201,
      "step": 84410
    },
    {
      "epoch": 0.8587137858072916,
      "grad_norm": 8.984891891479492,
      "learning_rate": 2.427374739899771e-07,
      "loss": 3.2724,
      "step": 84415
    },
    {
      "epoch": 0.8587646484375,
      "grad_norm": 9.335774421691895,
      "learning_rate": 2.425656179875127e-07,
      "loss": 3.6574,
      "step": 84420
    },
    {
      "epoch": 0.8588155110677084,
      "grad_norm": 13.307247161865234,
      "learning_rate": 2.4239381974097254e-07,
      "loss": 3.5403,
      "step": 84425
    },
    {
      "epoch": 0.8588663736979166,
      "grad_norm": 7.800132751464844,
      "learning_rate": 2.4222207925475165e-07,
      "loss": 3.3321,
      "step": 84430
    },
    {
      "epoch": 0.858917236328125,
      "grad_norm": 16.67194366455078,
      "learning_rate": 2.4205039653324296e-07,
      "loss": 3.0226,
      "step": 84435
    },
    {
      "epoch": 0.8589680989583334,
      "grad_norm": 13.30135726928711,
      "learning_rate": 2.4187877158084017e-07,
      "loss": 3.4109,
      "step": 84440
    },
    {
      "epoch": 0.8590189615885416,
      "grad_norm": 12.091148376464844,
      "learning_rate": 2.4170720440193337e-07,
      "loss": 3.1486,
      "step": 84445
    },
    {
      "epoch": 0.85906982421875,
      "grad_norm": 12.820271492004395,
      "learning_rate": 2.415356950009129e-07,
      "loss": 3.3555,
      "step": 84450
    },
    {
      "epoch": 0.8591206868489584,
      "grad_norm": 9.892502784729004,
      "learning_rate": 2.4136424338216583e-07,
      "loss": 3.3226,
      "step": 84455
    },
    {
      "epoch": 0.8591715494791666,
      "grad_norm": 10.605794906616211,
      "learning_rate": 2.4119284955007983e-07,
      "loss": 3.2693,
      "step": 84460
    },
    {
      "epoch": 0.859222412109375,
      "grad_norm": 15.961316108703613,
      "learning_rate": 2.4102151350903933e-07,
      "loss": 3.9426,
      "step": 84465
    },
    {
      "epoch": 0.8592732747395834,
      "grad_norm": 8.051385879516602,
      "learning_rate": 2.408502352634276e-07,
      "loss": 3.3205,
      "step": 84470
    },
    {
      "epoch": 0.8593241373697916,
      "grad_norm": 10.661064147949219,
      "learning_rate": 2.406790148176269e-07,
      "loss": 3.4702,
      "step": 84475
    },
    {
      "epoch": 0.859375,
      "grad_norm": 13.357624053955078,
      "learning_rate": 2.4050785217601876e-07,
      "loss": 3.3406,
      "step": 84480
    },
    {
      "epoch": 0.8594258626302084,
      "grad_norm": 12.120694160461426,
      "learning_rate": 2.403367473429816e-07,
      "loss": 3.1657,
      "step": 84485
    },
    {
      "epoch": 0.8594767252604166,
      "grad_norm": 10.15761947631836,
      "learning_rate": 2.40165700322893e-07,
      "loss": 3.0561,
      "step": 84490
    },
    {
      "epoch": 0.859527587890625,
      "grad_norm": 10.528076171875,
      "learning_rate": 2.3999471112012924e-07,
      "loss": 3.1222,
      "step": 84495
    },
    {
      "epoch": 0.8595784505208334,
      "grad_norm": 14.525897979736328,
      "learning_rate": 2.3982377973906487e-07,
      "loss": 2.9979,
      "step": 84500
    },
    {
      "epoch": 0.8596293131510416,
      "grad_norm": 9.03371524810791,
      "learning_rate": 2.396529061840738e-07,
      "loss": 3.4179,
      "step": 84505
    },
    {
      "epoch": 0.85968017578125,
      "grad_norm": 13.414257049560547,
      "learning_rate": 2.3948209045952684e-07,
      "loss": 3.2412,
      "step": 84510
    },
    {
      "epoch": 0.8597310384114584,
      "grad_norm": 7.893692493438721,
      "learning_rate": 2.393113325697954e-07,
      "loss": 3.0921,
      "step": 84515
    },
    {
      "epoch": 0.8597819010416666,
      "grad_norm": 16.511638641357422,
      "learning_rate": 2.3914063251924774e-07,
      "loss": 3.3977,
      "step": 84520
    },
    {
      "epoch": 0.859832763671875,
      "grad_norm": 8.734496116638184,
      "learning_rate": 2.3896999031225054e-07,
      "loss": 3.5417,
      "step": 84525
    },
    {
      "epoch": 0.8598836263020834,
      "grad_norm": 8.743608474731445,
      "learning_rate": 2.387994059531701e-07,
      "loss": 4.0634,
      "step": 84530
    },
    {
      "epoch": 0.8599344889322916,
      "grad_norm": 14.12103271484375,
      "learning_rate": 2.3862887944637146e-07,
      "loss": 3.3827,
      "step": 84535
    },
    {
      "epoch": 0.8599853515625,
      "grad_norm": 11.939756393432617,
      "learning_rate": 2.3845841079621704e-07,
      "loss": 3.5587,
      "step": 84540
    },
    {
      "epoch": 0.8600362141927084,
      "grad_norm": 10.308426856994629,
      "learning_rate": 2.382880000070681e-07,
      "loss": 3.3251,
      "step": 84545
    },
    {
      "epoch": 0.8600870768229166,
      "grad_norm": 17.604177474975586,
      "learning_rate": 2.3811764708328357e-07,
      "loss": 3.7324,
      "step": 84550
    },
    {
      "epoch": 0.860137939453125,
      "grad_norm": 11.988261222839355,
      "learning_rate": 2.3794735202922387e-07,
      "loss": 3.3324,
      "step": 84555
    },
    {
      "epoch": 0.8601888020833334,
      "grad_norm": 13.497099876403809,
      "learning_rate": 2.37777114849245e-07,
      "loss": 3.1458,
      "step": 84560
    },
    {
      "epoch": 0.8602396647135416,
      "grad_norm": 10.483180046081543,
      "learning_rate": 2.376069355477026e-07,
      "loss": 3.0151,
      "step": 84565
    },
    {
      "epoch": 0.86029052734375,
      "grad_norm": 14.707589149475098,
      "learning_rate": 2.3743681412894986e-07,
      "loss": 3.2507,
      "step": 84570
    },
    {
      "epoch": 0.8603413899739584,
      "grad_norm": 12.384051322937012,
      "learning_rate": 2.3726675059734e-07,
      "loss": 3.5616,
      "step": 84575
    },
    {
      "epoch": 0.8603922526041666,
      "grad_norm": 13.253273963928223,
      "learning_rate": 2.3709674495722422e-07,
      "loss": 3.5315,
      "step": 84580
    },
    {
      "epoch": 0.860443115234375,
      "grad_norm": 11.346092224121094,
      "learning_rate": 2.3692679721295215e-07,
      "loss": 3.4492,
      "step": 84585
    },
    {
      "epoch": 0.8604939778645834,
      "grad_norm": 11.59700870513916,
      "learning_rate": 2.3675690736887108e-07,
      "loss": 3.4211,
      "step": 84590
    },
    {
      "epoch": 0.8605448404947916,
      "grad_norm": 12.86052131652832,
      "learning_rate": 2.365870754293287e-07,
      "loss": 3.1789,
      "step": 84595
    },
    {
      "epoch": 0.860595703125,
      "grad_norm": 8.454562187194824,
      "learning_rate": 2.364173013986687e-07,
      "loss": 3.6257,
      "step": 84600
    },
    {
      "epoch": 0.8606465657552084,
      "grad_norm": 14.786294937133789,
      "learning_rate": 2.362475852812357e-07,
      "loss": 3.5912,
      "step": 84605
    },
    {
      "epoch": 0.8606974283854166,
      "grad_norm": 14.837119102478027,
      "learning_rate": 2.3607792708137205e-07,
      "loss": 2.9005,
      "step": 84610
    },
    {
      "epoch": 0.860748291015625,
      "grad_norm": 8.102725982666016,
      "learning_rate": 2.3590832680341818e-07,
      "loss": 3.4289,
      "step": 84615
    },
    {
      "epoch": 0.8607991536458334,
      "grad_norm": 16.29010772705078,
      "learning_rate": 2.3573878445171312e-07,
      "loss": 3.1857,
      "step": 84620
    },
    {
      "epoch": 0.8608500162760416,
      "grad_norm": 8.154257774353027,
      "learning_rate": 2.3556930003059392e-07,
      "loss": 3.3696,
      "step": 84625
    },
    {
      "epoch": 0.86090087890625,
      "grad_norm": 11.515044212341309,
      "learning_rate": 2.353998735443974e-07,
      "loss": 3.8192,
      "step": 84630
    },
    {
      "epoch": 0.8609517415364584,
      "grad_norm": 12.323447227478027,
      "learning_rate": 2.3523050499745904e-07,
      "loss": 3.7585,
      "step": 84635
    },
    {
      "epoch": 0.8610026041666666,
      "grad_norm": 14.341964721679688,
      "learning_rate": 2.350611943941114e-07,
      "loss": 4.0131,
      "step": 84640
    },
    {
      "epoch": 0.861053466796875,
      "grad_norm": 11.782180786132812,
      "learning_rate": 2.348919417386855e-07,
      "loss": 3.4359,
      "step": 84645
    },
    {
      "epoch": 0.8611043294270834,
      "grad_norm": 12.30263900756836,
      "learning_rate": 2.3472274703551317e-07,
      "loss": 3.4441,
      "step": 84650
    },
    {
      "epoch": 0.8611551920572916,
      "grad_norm": 11.4652099609375,
      "learning_rate": 2.3455361028892176e-07,
      "loss": 3.3107,
      "step": 84655
    },
    {
      "epoch": 0.8612060546875,
      "grad_norm": 8.915072441101074,
      "learning_rate": 2.3438453150323975e-07,
      "loss": 3.1885,
      "step": 84660
    },
    {
      "epoch": 0.8612569173177084,
      "grad_norm": 12.996953964233398,
      "learning_rate": 2.342155106827923e-07,
      "loss": 3.14,
      "step": 84665
    },
    {
      "epoch": 0.8613077799479166,
      "grad_norm": 12.54077434539795,
      "learning_rate": 2.3404654783190401e-07,
      "loss": 3.5618,
      "step": 84670
    },
    {
      "epoch": 0.861358642578125,
      "grad_norm": 10.259812355041504,
      "learning_rate": 2.3387764295489806e-07,
      "loss": 3.8982,
      "step": 84675
    },
    {
      "epoch": 0.8614095052083334,
      "grad_norm": 11.128485679626465,
      "learning_rate": 2.3370879605609464e-07,
      "loss": 3.0753,
      "step": 84680
    },
    {
      "epoch": 0.8614603678385416,
      "grad_norm": 12.832159996032715,
      "learning_rate": 2.3354000713981474e-07,
      "loss": 2.9776,
      "step": 84685
    },
    {
      "epoch": 0.86151123046875,
      "grad_norm": 13.322426795959473,
      "learning_rate": 2.333712762103771e-07,
      "loss": 3.1568,
      "step": 84690
    },
    {
      "epoch": 0.8615620930989584,
      "grad_norm": 12.448477745056152,
      "learning_rate": 2.33202603272098e-07,
      "loss": 3.298,
      "step": 84695
    },
    {
      "epoch": 0.8616129557291666,
      "grad_norm": 9.18695068359375,
      "learning_rate": 2.3303398832929259e-07,
      "loss": 3.3789,
      "step": 84700
    },
    {
      "epoch": 0.861663818359375,
      "grad_norm": 9.873669624328613,
      "learning_rate": 2.328654313862752e-07,
      "loss": 3.3122,
      "step": 84705
    },
    {
      "epoch": 0.8617146809895834,
      "grad_norm": 13.608736991882324,
      "learning_rate": 2.326969324473588e-07,
      "loss": 3.4122,
      "step": 84710
    }
  ],
  "logging_steps": 5,
  "max_steps": 98304,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 394,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.288882461717299e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}