{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.3270666298037046,
  "global_step": 2400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.9999895164082156e-05,
      "loss": 1.6349,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999958065720787e-05,
      "loss": 1.6199,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999905648201487e-05,
      "loss": 1.4834,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999832264289934e-05,
      "loss": 1.3882,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999737914601591e-05,
      "loss": 1.3679,
      "step": 25
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999622599927756e-05,
      "loss": 1.2396,
      "step": 30
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999486321235559e-05,
      "loss": 1.321,
      "step": 35
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9993290796679516e-05,
      "loss": 1.2874,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999150876543699e-05,
      "loss": 1.2607,
      "step": 45
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9989517133573694e-05,
      "loss": 1.2454,
      "step": 50
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9987315917793174e-05,
      "loss": 1.2799,
      "step": 55
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998490513655676e-05,
      "loss": 1.2575,
      "step": 60
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.998228481008337e-05,
      "loss": 1.2404,
      "step": 65
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997945496034934e-05,
      "loss": 1.2219,
      "step": 70
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9976415611088267e-05,
      "loss": 1.2241,
      "step": 75
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.997316678779079e-05,
      "loss": 1.1716,
      "step": 80
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.996970851770438e-05,
      "loss": 1.1883,
      "step": 85
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9966040829833115e-05,
      "loss": 1.205,
      "step": 90
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9962163754937426e-05,
      "loss": 1.1246,
      "step": 95
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.995807732553384e-05,
      "loss": 1.1636,
      "step": 100
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9953781575894723e-05,
      "loss": 1.158,
      "step": 105
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9949276542048e-05,
      "loss": 1.1477,
      "step": 110
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9944562261776805e-05,
      "loss": 1.1678,
      "step": 115
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9939638774619216e-05,
      "loss": 1.1501,
      "step": 120
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.99345061218679e-05,
      "loss": 1.1955,
      "step": 125
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9929164346569756e-05,
      "loss": 1.1724,
      "step": 130
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9923613493525576e-05,
      "loss": 1.177,
      "step": 135
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991785360928968e-05,
      "loss": 1.1418,
      "step": 140
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991188474216947e-05,
      "loss": 1.1898,
      "step": 145
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9905706942225094e-05,
      "loss": 1.1479,
      "step": 150
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9899320261268966e-05,
      "loss": 1.1356,
      "step": 155
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.989272475286537e-05,
      "loss": 1.1397,
      "step": 160
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9885920472330004e-05,
      "loss": 1.1215,
      "step": 165
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9878907476729516e-05,
      "loss": 1.167,
      "step": 170
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9871685824881e-05,
      "loss": 1.1219,
      "step": 175
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9864255577351534e-05,
      "loss": 1.0835,
      "step": 180
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.985661679645769e-05,
      "loss": 1.0721,
      "step": 185
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9848769546264915e-05,
      "loss": 1.0692,
      "step": 190
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.9840713892587146e-05,
      "loss": 1.0488,
      "step": 195
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.983244990298609e-05,
      "loss": 1.1285,
      "step": 200
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.982397764677081e-05,
      "loss": 1.0832,
      "step": 205
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.981529719499704e-05,
      "loss": 1.0652,
      "step": 210
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.980640862046663e-05,
      "loss": 1.1043,
      "step": 215
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.979731199772693e-05,
      "loss": 1.112,
      "step": 220
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9788007403070146e-05,
      "loss": 1.1029,
      "step": 225
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.977849491453277e-05,
      "loss": 1.0869,
      "step": 230
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.976877461189481e-05,
      "loss": 1.0843,
      "step": 235
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.975884657667922e-05,
      "loss": 1.0789,
      "step": 240
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.974871089215118e-05,
      "loss": 1.0449,
      "step": 245
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9738367643317405e-05,
      "loss": 1.1053,
      "step": 250
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9727816916925395e-05,
      "loss": 1.0651,
      "step": 255
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.971705880146276e-05,
      "loss": 1.0828,
      "step": 260
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.970609338715646e-05,
      "loss": 1.0932,
      "step": 265
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.969492076597203e-05,
      "loss": 1.0648,
      "step": 270
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.968354103161283e-05,
      "loss": 1.0948,
      "step": 275
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.967195427951926e-05,
      "loss": 1.0721,
      "step": 280
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9660160606867936e-05,
      "loss": 1.124,
      "step": 285
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9648160112570896e-05,
      "loss": 1.0963,
      "step": 290
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.9635952897274773e-05,
      "loss": 1.1078,
      "step": 295
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9623539063359925e-05,
      "loss": 1.1059,
      "step": 300
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.961091871493962e-05,
      "loss": 1.1032,
      "step": 305
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.959809195785912e-05,
      "loss": 1.0595,
      "step": 310
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.958505889969481e-05,
      "loss": 1.1096,
      "step": 315
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.957181964975329e-05,
      "loss": 1.0589,
      "step": 320
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.955837431907049e-05,
      "loss": 1.0608,
      "step": 325
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.954472302041069e-05,
      "loss": 1.0819,
      "step": 330
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9530865868265605e-05,
      "loss": 1.0759,
      "step": 335
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.951680297885342e-05,
      "loss": 1.0515,
      "step": 340
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.950253447011779e-05,
      "loss": 1.0371,
      "step": 345
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.948806046172691e-05,
      "loss": 1.0619,
      "step": 350
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.947338107507245e-05,
      "loss": 1.0757,
      "step": 355
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.945849643326857e-05,
      "loss": 1.0686,
      "step": 360
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9443406661150874e-05,
      "loss": 1.0809,
      "step": 365
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.942811188527537e-05,
      "loss": 1.0704,
      "step": 370
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.941261223391742e-05,
      "loss": 1.0655,
      "step": 375
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.939690783707063e-05,
      "loss": 1.1182,
      "step": 380
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.938099882644578e-05,
      "loss": 1.081,
      "step": 385
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9364885335469734e-05,
      "loss": 1.0792,
      "step": 390
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.93485674992843e-05,
      "loss": 1.0244,
      "step": 395
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.933204545474511e-05,
      "loss": 1.0531,
      "step": 400
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9315319340420465e-05,
      "loss": 1.048,
      "step": 405
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.929838929659015e-05,
      "loss": 1.04,
      "step": 410
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9281255465244314e-05,
      "loss": 1.0686,
      "step": 415
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.926391799008223e-05,
      "loss": 1.0933,
      "step": 420
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.924637701651111e-05,
      "loss": 1.0313,
      "step": 425
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9228632691644874e-05,
      "loss": 1.0811,
      "step": 430
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.921068516430293e-05,
      "loss": 1.0617,
      "step": 435
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.919253458500892e-05,
      "loss": 1.0482,
      "step": 440
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.9174181105989445e-05,
      "loss": 1.0681,
      "step": 445
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.9155624881172834e-05,
      "loss": 1.0476,
      "step": 450
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.913686606618777e-05,
      "loss": 1.0463,
      "step": 455
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.911790481836208e-05,
      "loss": 1.0213,
      "step": 460
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.909874129672133e-05,
      "loss": 0.9855,
      "step": 465
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.907937566198757e-05,
      "loss": 1.045,
      "step": 470
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.9059808076577914e-05,
      "loss": 1.0442,
      "step": 475
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.904003870460323e-05,
      "loss": 1.0354,
      "step": 480
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.9020067711866735e-05,
      "loss": 1.0331,
      "step": 485
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.899989526586261e-05,
      "loss": 1.0585,
      "step": 490
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.8979521535774636e-05,
      "loss": 1.0223,
      "step": 495
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.895894669247468e-05,
      "loss": 1.0118,
      "step": 500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8938170908521356e-05,
      "loss": 1.0508,
      "step": 505
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8917194358158534e-05,
      "loss": 1.0656,
      "step": 510
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8896017217313886e-05,
      "loss": 1.0655,
      "step": 515
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.887463966359741e-05,
      "loss": 1.0833,
      "step": 520
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8853061876299956e-05,
      "loss": 1.068,
      "step": 525
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8831284036391684e-05,
      "loss": 1.0487,
      "step": 530
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.880930632652058e-05,
      "loss": 1.059,
      "step": 535
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.878712893101092e-05,
      "loss": 1.0408,
      "step": 540
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.876475203586171e-05,
      "loss": 0.9976,
      "step": 545
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.874217582874514e-05,
      "loss": 1.0358,
      "step": 550
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8719400499005e-05,
      "loss": 1.0073,
      "step": 555
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.869642623765509e-05,
      "loss": 1.0742,
      "step": 560
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.867325323737765e-05,
      "loss": 1.0463,
      "step": 565
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.864988169252168e-05,
      "loss": 1.0242,
      "step": 570
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8626311799101375e-05,
      "loss": 1.0773,
      "step": 575
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.860254375479446e-05,
      "loss": 1.0371,
      "step": 580
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8578577758940504e-05,
      "loss": 1.0399,
      "step": 585
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.855441401253928e-05,
      "loss": 1.0411,
      "step": 590
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8530052718249076e-05,
      "loss": 1.0476,
      "step": 595
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.850549408038498e-05,
      "loss": 1.0075,
      "step": 600
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.848073830491717e-05,
      "loss": 1.0045,
      "step": 605
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.845578559946923e-05,
      "loss": 1.0368,
      "step": 610
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.8430636173316306e-05,
      "loss": 1.0156,
      "step": 615
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.840529023738348e-05,
      "loss": 1.0334,
      "step": 620
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.837974800424389e-05,
      "loss": 0.9994,
      "step": 625
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.8354009688117026e-05,
      "loss": 1.0467,
      "step": 630
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.8328075504866874e-05,
      "loss": 0.9968,
      "step": 635
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.8301945672000164e-05,
      "loss": 1.041,
      "step": 640
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.8275620408664487e-05,
      "loss": 0.9811,
      "step": 645
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.8249099935646494e-05,
      "loss": 1.0235,
      "step": 650
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.822238447537003e-05,
      "loss": 1.0152,
      "step": 655
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.819547425189429e-05,
      "loss": 1.038,
      "step": 660
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.81683694909119e-05,
      "loss": 1.0387,
      "step": 665
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.814107041974707e-05,
      "loss": 1.019,
      "step": 670
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.811357726735366e-05,
      "loss": 1.0403,
      "step": 675
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.808589026431324e-05,
      "loss": 1.0828,
      "step": 680
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.805800964283322e-05,
      "loss": 1.0128,
      "step": 685
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.802993563674483e-05,
      "loss": 1.0502,
      "step": 690
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.80016684815012e-05,
      "loss": 1.0169,
      "step": 695
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.7973208414175406e-05,
      "loss": 1.0316,
      "step": 700
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.794455567345842e-05,
      "loss": 1.0263,
      "step": 705
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.791571049965714e-05,
      "loss": 1.0801,
      "step": 710
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.7886673134692404e-05,
      "loss": 1.0575,
      "step": 715
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.7857443822096905e-05,
      "loss": 0.9767,
      "step": 720
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.782802280701319e-05,
      "loss": 0.9666,
      "step": 725
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.77984103361916e-05,
      "loss": 0.9983,
      "step": 730
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.776860665798816e-05,
      "loss": 1.0219,
      "step": 735
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.773861202236257e-05,
      "loss": 0.9963,
      "step": 740
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.770842668087602e-05,
      "loss": 1.0219,
      "step": 745
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.767805088668916e-05,
      "loss": 1.0029,
      "step": 750
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.7647484894559936e-05,
      "loss": 1.0077,
      "step": 755
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.7616728960841444e-05,
      "loss": 0.9908,
      "step": 760
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.758578334347981e-05,
      "loss": 1.0092,
      "step": 765
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.7554648302012015e-05,
      "loss": 0.9874,
      "step": 770
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.7523324097563706e-05,
      "loss": 1.0185,
      "step": 775
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.749181099284703e-05,
      "loss": 1.0533,
      "step": 780
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.746010925215839e-05,
      "loss": 1.0083,
      "step": 785
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.74282191413763e-05,
      "loss": 1.0191,
      "step": 790
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.7396140927959045e-05,
      "loss": 0.9786,
      "step": 795
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.7363874880942574e-05,
      "loss": 1.0197,
      "step": 800
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.733142127093813e-05,
      "loss": 1.0471,
      "step": 805
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.7298780370130014e-05,
      "loss": 1.0134,
      "step": 810
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.726595245227336e-05,
      "loss": 0.9801,
      "step": 815
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.723293779269173e-05,
      "loss": 0.9944,
      "step": 820
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7199736668274924e-05,
      "loss": 1.0358,
      "step": 825
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.716634935747655e-05,
      "loss": 0.9907,
      "step": 830
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.713277614031177e-05,
      "loss": 1.0698,
      "step": 835
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.70990172983549e-05,
      "loss": 1.0466,
      "step": 840
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.706507311473707e-05,
      "loss": 0.9489,
      "step": 845
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.703094387414385e-05,
      "loss": 0.9936,
      "step": 850
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.699662986281288e-05,
      "loss": 0.9672,
      "step": 855
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.696213136853141e-05,
      "loss": 1.0164,
      "step": 860
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.6927448680633954e-05,
      "loss": 0.9817,
      "step": 865
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.689258208999983e-05,
      "loss": 1.028,
      "step": 870
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.6857531889050716e-05,
      "loss": 1.0239,
      "step": 875
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.682229837174821e-05,
      "loss": 1.0378,
      "step": 880
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.678688183359135e-05,
      "loss": 0.9908,
      "step": 885
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.675128257161418e-05,
      "loss": 0.9384,
      "step": 890
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.671550088438319e-05,
      "loss": 0.9481,
      "step": 895
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.6679537071994874e-05,
      "loss": 0.9541,
      "step": 900
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.6643391436073165e-05,
      "loss": 0.9787,
      "step": 905
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.660706427976693e-05,
      "loss": 0.9897,
      "step": 910
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.657055590774745e-05,
      "loss": 0.9947,
      "step": 915
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.6533866626205805e-05,
      "loss": 0.9614,
      "step": 920
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.649699674285036e-05,
      "loss": 1.0323,
      "step": 925
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.645994656690417e-05,
      "loss": 1.0345,
      "step": 930
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.642271640910235e-05,
      "loss": 1.0432,
      "step": 935
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.638530658168954e-05,
      "loss": 1.0102,
      "step": 940
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.6347717398417203e-05,
      "loss": 1.0361,
      "step": 945
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.6309949174541096e-05,
      "loss": 0.9818,
      "step": 950
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.627200222681851e-05,
      "loss": 1.0293,
      "step": 955
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.6233876873505694e-05,
      "loss": 1.0125,
      "step": 960
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.619557343435516e-05,
      "loss": 1.0039,
      "step": 965
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.615709223061302e-05,
      "loss": 1.0146,
      "step": 970
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.611843358501624e-05,
      "loss": 1.0352,
      "step": 975
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.6079597821789993e-05,
      "loss": 1.0405,
      "step": 980
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.604058526664491e-05,
      "loss": 0.9834,
      "step": 985
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.600139624677436e-05,
      "loss": 1.0252,
      "step": 990
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.596203109085168e-05,
      "loss": 1.0516,
      "step": 995
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.5922490129027464e-05,
      "loss": 0.9762,
      "step": 1000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.588277369292674e-05,
      "loss": 0.9801,
      "step": 1005
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.5842882115646234e-05,
      "loss": 1.0127,
      "step": 1010
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.580281573175157e-05,
      "loss": 1.0415,
      "step": 1015
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.576257487727442e-05,
      "loss": 0.9974,
      "step": 1020
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.572215988970974e-05,
      "loss": 1.0097,
      "step": 1025
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.568157110801293e-05,
      "loss": 1.0004,
      "step": 1030
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.5640808872596944e-05,
      "loss": 0.9932,
      "step": 1035
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.5599873525329505e-05,
      "loss": 0.9437,
      "step": 1040
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.555876540953019e-05,
      "loss": 1.0004,
      "step": 1045
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.551748486996755e-05,
      "loss": 0.9743,
      "step": 1050
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.547603225285626e-05,
      "loss": 1.0303,
      "step": 1055
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.543440790585417e-05,
      "loss": 1.0225,
      "step": 1060
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.539261217805939e-05,
      "loss": 1.0102,
      "step": 1065
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.535064542000743e-05,
      "loss": 1.0087,
      "step": 1070
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.5308507983668165e-05,
      "loss": 0.952,
      "step": 1075
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.526620022244293e-05,
      "loss": 0.9767,
      "step": 1080
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.522372249116158e-05,
      "loss": 1.0049,
      "step": 1085
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.5181075146079456e-05,
      "loss": 1.0397,
      "step": 1090
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.5138258544874455e-05,
      "loss": 1.0167,
      "step": 1095
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.5095273046643985e-05,
      "loss": 0.9966,
      "step": 1100
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.5052119011901986e-05,
      "loss": 0.983,
      "step": 1105
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.500879680257587e-05,
      "loss": 1.0301,
      "step": 1110
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.4965306782003535e-05,
      "loss": 1.0371,
      "step": 1115
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.492164931493028e-05,
      "loss": 1.0083,
      "step": 1120
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.487782476750575e-05,
      "loss": 0.9728,
      "step": 1125
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.4833833507280884e-05,
      "loss": 0.9892,
      "step": 1130
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.4789675903204805e-05,
      "loss": 0.9847,
      "step": 1135
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.474535232562176e-05,
      "loss": 1.0545,
      "step": 1140
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.470086314626797e-05,
      "loss": 1.0204,
      "step": 1145
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.465620873826856e-05,
      "loss": 0.9648,
      "step": 1150
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.46113894761344e-05,
      "loss": 0.9545,
      "step": 1155
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.456640573575896e-05,
      "loss": 1.0256,
      "step": 1160
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.4521257894415183e-05,
      "loss": 0.99,
      "step": 1165
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.44759463307523e-05,
      "loss": 0.9906,
      "step": 1170
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.443047142479266e-05,
      "loss": 0.9876,
      "step": 1175
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.4384833557928553e-05,
      "loss": 1.0495,
      "step": 1180
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.4339033112918966e-05,
      "loss": 0.9869,
      "step": 1185
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.4293070473886456e-05,
      "loss": 1.0299,
      "step": 1190
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.424694602631385e-05,
      "loss": 1.0073,
      "step": 1195
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.420066015704105e-05,
      "loss": 1.0023,
      "step": 1200
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.41542132542618e-05,
      "loss": 0.9462,
      "step": 1205
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.410760570752037e-05,
      "loss": 1.0116,
      "step": 1210
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.4060837907708375e-05,
      "loss": 0.9652,
      "step": 1215
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.401391024706142e-05,
      "loss": 1.0411,
      "step": 1220
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.396682311915586e-05,
      "loss": 0.9691,
      "step": 1225
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.3919576918905495e-05,
      "loss": 0.961,
      "step": 1230
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.387217204255819e-05,
      "loss": 0.9602,
      "step": 1235
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.3824608887692666e-05,
      "loss": 1.0197,
      "step": 1240
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.377688785321507e-05,
      "loss": 0.9982,
      "step": 1245
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.372900933935569e-05,
      "loss": 0.986,
      "step": 1250
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.368097374766556e-05,
      "loss": 0.9744,
      "step": 1255
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.3632781481013105e-05,
      "loss": 1.0078,
      "step": 1260
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.358443294358077e-05,
      "loss": 0.9865,
      "step": 1265
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.35359285408616e-05,
      "loss": 0.976,
      "step": 1270
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.348726867965591e-05,
      "loss": 1.0141,
      "step": 1275
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.343845376806777e-05,
      "loss": 0.952,
      "step": 1280
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.338948421550169e-05,
      "loss": 1.0239,
      "step": 1285
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.334036043265909e-05,
      "loss": 0.994,
      "step": 1290
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.329108283153492e-05,
      "loss": 1.0505,
      "step": 1295
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.3241651825414195e-05,
      "loss": 0.9919,
      "step": 1300
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.31920678288685e-05,
      "loss": 1.0169,
      "step": 1305
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.3142331257752546e-05,
      "loss": 0.9938,
      "step": 1310
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.309244252920064e-05,
      "loss": 0.92,
      "step": 1315
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.304240206162326e-05,
      "loss": 0.9537,
      "step": 1320
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.299221027470345e-05,
      "loss": 0.996,
      "step": 1325
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.294186758939339e-05,
      "loss": 0.9652,
      "step": 1330
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.2891374427910795e-05,
      "loss": 1.0138,
      "step": 1335
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.284073121373544e-05,
      "loss": 0.9267,
      "step": 1340
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.278993837160553e-05,
      "loss": 0.9899,
      "step": 1345
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.273899632751422e-05,
      "loss": 0.963,
      "step": 1350
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.2687905508705974e-05,
      "loss": 1.0066,
      "step": 1355
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.263666634367303e-05,
      "loss": 0.9939,
      "step": 1360
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.258527926215178e-05,
      "loss": 0.9667,
      "step": 1365
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.253374469511917e-05,
      "loss": 0.9999,
      "step": 1370
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.248206307478909e-05,
      "loss": 0.9738,
      "step": 1375
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.243023483460875e-05,
      "loss": 0.9829,
      "step": 1380
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.237826040925503e-05,
      "loss": 1.0107,
      "step": 1385
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.232614023463088e-05,
      "loss": 0.9992,
      "step": 1390
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.227387474786159e-05,
      "loss": 0.9566,
      "step": 1395
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.222146438729119e-05,
      "loss": 0.9722,
      "step": 1400
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.216890959247873e-05,
      "loss": 0.988,
      "step": 1405
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.211621080419463e-05,
      "loss": 1.0103,
      "step": 1410
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.206336846441695e-05,
      "loss": 0.9805,
      "step": 1415
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.201038301632772e-05,
      "loss": 1.0177,
      "step": 1420
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.195725490430917e-05,
      "loss": 0.9938,
      "step": 1425
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.190398457394007e-05,
      "loss": 1.0276,
      "step": 1430
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.1850572471991924e-05,
      "loss": 0.9909,
      "step": 1435
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.1797019046425264e-05,
      "loss": 0.9794,
      "step": 1440
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.1743324746385914e-05,
      "loss": 1.0073,
      "step": 1445
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.1689490022201154e-05,
      "loss": 1.0106,
      "step": 1450
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.163551532537601e-05,
      "loss": 1.0241,
      "step": 1455
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.1581401108589425e-05,
      "loss": 1.0178,
      "step": 1460
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.1527147825690495e-05,
      "loss": 0.9426,
      "step": 1465
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.1472755931694626e-05,
      "loss": 0.9642,
      "step": 1470
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.141822588277976e-05,
      "loss": 0.9678,
      "step": 1475
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.136355813628251e-05,
      "loss": 0.9654,
      "step": 1480
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.130875315069435e-05,
      "loss": 0.9748,
      "step": 1485
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.125381138565775e-05,
      "loss": 0.9321,
      "step": 1490
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.1198733301962346e-05,
      "loss": 0.9885,
      "step": 1495
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.114351936154105e-05,
      "loss": 0.9385,
      "step": 1500
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.108817002746619e-05,
      "loss": 0.9362,
      "step": 1505
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.1032685763945625e-05,
      "loss": 0.9764,
      "step": 1510
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.097706703631886e-05,
      "loss": 0.961,
      "step": 1515
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.092131431105312e-05,
      "loss": 0.9818,
      "step": 1520
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.086542805573945e-05,
      "loss": 0.9855,
      "step": 1525
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.080940873908881e-05,
      "loss": 0.9618,
      "step": 1530
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.07532568309281e-05,
      "loss": 1.0102,
      "step": 1535
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.069697280219628e-05,
      "loss": 0.9773,
      "step": 1540
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.0640557124940376e-05,
      "loss": 0.9687,
      "step": 1545
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.058401027231152e-05,
      "loss": 0.9818,
      "step": 1550
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.052733271856103e-05,
      "loss": 0.9911,
      "step": 1555
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.0470524939036355e-05,
      "loss": 0.9694,
      "step": 1560
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.0413587410177155e-05,
      "loss": 0.9374,
      "step": 1565
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.035652060951128e-05,
      "loss": 0.956,
      "step": 1570
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.0299325015650774e-05,
      "loss": 0.9491,
      "step": 1575
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.024200110828783e-05,
      "loss": 0.9569,
      "step": 1580
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.018454936819082e-05,
      "loss": 0.9627,
      "step": 1585
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.012697027720018e-05,
      "loss": 0.9922,
      "step": 1590
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.0069264318224506e-05,
      "loss": 0.9703,
      "step": 1595
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.0011431975236337e-05,
      "loss": 0.9436,
      "step": 1600
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.995347373326822e-05,
      "loss": 0.961,
      "step": 1605
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.989539007840861e-05,
      "loss": 0.9247,
      "step": 1610
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.983718149779775e-05,
      "loss": 0.9537,
      "step": 1615
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.9778848479623656e-05,
      "loss": 0.9415,
      "step": 1620
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.972039151311795e-05,
      "loss": 0.9954,
      "step": 1625
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.966181108855183e-05,
      "loss": 0.9451,
      "step": 1630
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.960310769723189e-05,
      "loss": 0.9977,
      "step": 1635
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.9544281831496034e-05,
      "loss": 0.9806,
      "step": 1640
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.9485333984709374e-05,
      "loss": 0.9851,
      "step": 1645
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.942626465126001e-05,
      "loss": 1.0089,
      "step": 1650
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.9367074326555e-05,
      "loss": 0.9562,
      "step": 1655
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.930776350701609e-05,
      "loss": 0.9892,
      "step": 1660
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.92483326900756e-05,
      "loss": 0.9875,
      "step": 1665
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.91887823741723e-05,
      "loss": 1.0012,
      "step": 1670
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.9129113058747136e-05,
      "loss": 0.97,
      "step": 1675
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.9069325244239095e-05,
      "loss": 0.9793,
      "step": 1680
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.900941943208103e-05,
      "loss": 0.9864,
      "step": 1685
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.894939612469539e-05,
      "loss": 0.9596,
      "step": 1690
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.888925582549006e-05,
      "loss": 0.9679,
      "step": 1695
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.882899903885412e-05,
      "loss": 0.9395,
      "step": 1700
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.876862627015361e-05,
      "loss": 0.9774,
      "step": 1705
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.87081380257273e-05,
      "loss": 0.9686,
      "step": 1710
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.864753481288244e-05,
      "loss": 0.953,
      "step": 1715
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.8586817139890515e-05,
      "loss": 0.9719,
      "step": 1720
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.852598551598294e-05,
      "loss": 0.9748,
      "step": 1725
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.8465040451346874e-05,
      "loss": 0.9741,
      "step": 1730
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.8403982457120836e-05,
      "loss": 0.9747,
      "step": 1735
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.834281204539051e-05,
      "loss": 0.9791,
      "step": 1740
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.828152972918438e-05,
      "loss": 0.9704,
      "step": 1745
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.82201360224695e-05,
      "loss": 0.9556,
      "step": 1750
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.815863144014711e-05,
      "loss": 0.9552,
      "step": 1755
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.809701649804834e-05,
      "loss": 0.9594,
      "step": 1760
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.8035291712929926e-05,
      "loss": 0.959,
      "step": 1765
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.797345760246982e-05,
      "loss": 1.0254,
      "step": 1770
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.791151468526289e-05,
      "loss": 0.9925,
      "step": 1775
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.784946348081654e-05,
      "loss": 0.9516,
      "step": 1780
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.7787304509546365e-05,
      "loss": 0.954,
      "step": 1785
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.7725038292771774e-05,
      "loss": 0.9465,
      "step": 1790
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.766266535271167e-05,
      "loss": 0.9792,
      "step": 1795
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.760018621248e-05,
      "loss": 0.964,
      "step": 1800
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.75376013960814e-05,
      "loss": 0.9419,
      "step": 1805
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.747491142840681e-05,
      "loss": 0.9818,
      "step": 1810
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.741211683522904e-05,
      "loss": 0.9153,
      "step": 1815
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.734921814319841e-05,
      "loss": 0.955,
      "step": 1820
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.728621587983828e-05,
      "loss": 0.9467,
      "step": 1825
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.722311057354067e-05,
      "loss": 0.9816,
      "step": 1830
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.715990275356178e-05,
      "loss": 0.9727,
      "step": 1835
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7096592950017617e-05,
      "loss": 0.9822,
      "step": 1840
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.703318169387947e-05,
      "loss": 0.942,
      "step": 1845
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.696966951696952e-05,
      "loss": 0.9306,
      "step": 1850
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.690605695195637e-05,
      "loss": 0.9654,
      "step": 1855
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.684234453235054e-05,
      "loss": 0.9592,
      "step": 1860
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.677853279250003e-05,
      "loss": 0.988,
      "step": 1865
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.671462226758583e-05,
      "loss": 0.9462,
      "step": 1870
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.665061349361742e-05,
      "loss": 0.9685,
      "step": 1875
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.658650700742828e-05,
      "loss": 0.9772,
      "step": 1880
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.6522303346671404e-05,
      "loss": 0.9482,
      "step": 1885
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.645800304981477e-05,
      "loss": 1.0069,
      "step": 1890
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.639360665613683e-05,
      "loss": 0.9003,
      "step": 1895
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.632911470572197e-05,
      "loss": 0.9279,
      "step": 1900
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.626452773945603e-05,
      "loss": 0.9237,
      "step": 1905
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.619984629902172e-05,
      "loss": 1.0086,
      "step": 1910
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.613507092689409e-05,
      "loss": 0.9625,
      "step": 1915
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.607020216633599e-05,
      "loss": 0.9297,
      "step": 1920
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.60052405613935e-05,
      "loss": 0.9718,
      "step": 1925
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.594018665689139e-05,
      "loss": 0.9512,
      "step": 1930
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.5875040998428513e-05,
      "loss": 0.9923,
      "step": 1935
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.5809804132373253e-05,
      "loss": 0.9518,
      "step": 1940
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.574447660585897e-05,
      "loss": 0.9324,
      "step": 1945
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.5679058966779344e-05,
      "loss": 0.9724,
      "step": 1950
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.561355176378384e-05,
      "loss": 0.978,
      "step": 1955
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.554795554627307e-05,
      "loss": 0.9893,
      "step": 1960
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.548227086439422e-05,
      "loss": 0.967,
      "step": 1965
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.541649826903639e-05,
      "loss": 0.9529,
      "step": 1970
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.535063831182602e-05,
      "loss": 0.9506,
      "step": 1975
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.528469154512224e-05,
      "loss": 0.9525,
      "step": 1980
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.521865852201223e-05,
      "loss": 0.9258,
      "step": 1985
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.5152539796306596e-05,
      "loss": 0.9417,
      "step": 1990
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.508633592253472e-05,
      "loss": 0.9465,
      "step": 1995
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.502004745594011e-05,
      "loss": 0.9494,
      "step": 2000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.4953674952475755e-05,
      "loss": 0.9709,
      "step": 2005
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.488721896879943e-05,
      "loss": 0.9581,
      "step": 2010
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.4820680062269074e-05,
      "loss": 0.974,
      "step": 2015
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.4754058790938046e-05,
      "loss": 0.9768,
      "step": 2020
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.468735571355055e-05,
      "loss": 0.982,
      "step": 2025
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.4620571389536825e-05,
      "loss": 0.984,
      "step": 2030
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.455370637900856e-05,
      "loss": 0.9604,
      "step": 2035
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.448676124275414e-05,
      "loss": 0.897,
      "step": 2040
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.4419736542233925e-05,
      "loss": 0.9968,
      "step": 2045
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.4352632839575616e-05,
      "loss": 0.9479,
      "step": 2050
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.428545069756946e-05,
      "loss": 0.9724,
      "step": 2055
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.42181906796636e-05,
      "loss": 0.9493,
      "step": 2060
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.415085334995927e-05,
      "loss": 0.9348,
      "step": 2065
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.408343927320613e-05,
      "loss": 0.9702,
      "step": 2070
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.401594901479753e-05,
      "loss": 0.9089,
      "step": 2075
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.394838314076572e-05,
      "loss": 0.9606,
      "step": 2080
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.3880742217777115e-05,
      "loss": 0.9743,
      "step": 2085
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.381302681312759e-05,
      "loss": 0.9469,
      "step": 2090
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.374523749473767e-05,
      "loss": 0.949,
      "step": 2095
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.367737483114779e-05,
      "loss": 0.9567,
      "step": 2100
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.360943939151351e-05,
      "loss": 0.9718,
      "step": 2105
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.354143174560078e-05,
      "loss": 0.9626,
      "step": 2110
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.3473352463781105e-05,
      "loss": 0.9346,
      "step": 2115
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.340520211702681e-05,
      "loss": 0.9208,
      "step": 2120
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.333698127690623e-05,
      "loss": 0.9856,
      "step": 2125
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.326869051557891e-05,
      "loss": 0.9049,
      "step": 2130
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.320033040579082e-05,
      "loss": 0.9222,
      "step": 2135
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.3131901520869565e-05,
      "loss": 0.9648,
      "step": 2140
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.306340443471951e-05,
      "loss": 0.9538,
      "step": 2145
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.299483972181708e-05,
      "loss": 0.9314,
      "step": 2150
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.292620795720583e-05,
      "loss": 0.9576,
      "step": 2155
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.285750971649167e-05,
      "loss": 0.9499,
      "step": 2160
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.278874557583807e-05,
      "loss": 0.9568,
      "step": 2165
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.271991611196117e-05,
      "loss": 0.9642,
      "step": 2170
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.265102190212497e-05,
      "loss": 0.9526,
      "step": 2175
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.258206352413648e-05,
      "loss": 0.933,
      "step": 2180
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.2513041556340887e-05,
      "loss": 0.9683,
      "step": 2185
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.244395657761671e-05,
      "loss": 0.9155,
      "step": 2190
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.2374809167370924e-05,
      "loss": 0.9262,
      "step": 2195
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.230559990553409e-05,
      "loss": 0.9778,
      "step": 2200
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.2236329372555544e-05,
      "loss": 0.9577,
      "step": 2205
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.2166998149398465e-05,
      "loss": 0.9286,
      "step": 2210
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.209760681753505e-05,
      "loss": 0.9634,
      "step": 2215
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.2028155958941615e-05,
      "loss": 0.9467,
      "step": 2220
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.195864615609373e-05,
      "loss": 0.9543,
      "step": 2225
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.1889077991961304e-05,
      "loss": 0.9914,
      "step": 2230
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.181945205000373e-05,
      "loss": 0.9309,
      "step": 2235
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.1749768914164955e-05,
      "loss": 0.9299,
      "step": 2240
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.168002916886864e-05,
      "loss": 0.9462,
      "step": 2245
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.1610233399013194e-05,
      "loss": 0.948,
      "step": 2250
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.15403821899669e-05,
      "loss": 0.9335,
      "step": 2255
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.147047612756302e-05,
      "loss": 0.952,
      "step": 2260
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.140051579809484e-05,
      "loss": 0.9532,
      "step": 2265
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.133050178831079e-05,
      "loss": 0.9853,
      "step": 2270
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.12604346854095e-05,
      "loss": 0.8925,
      "step": 2275
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.119031507703491e-05,
      "loss": 0.942,
      "step": 2280
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.112014355127129e-05,
      "loss": 0.9132,
      "step": 2285
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.104992069663835e-05,
      "loss": 0.9335,
      "step": 2290
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.0979647102086273e-05,
      "loss": 0.9403,
      "step": 2295
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.090932335699081e-05,
      "loss": 0.9246,
      "step": 2300
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.083895005114831e-05,
      "loss": 0.912,
      "step": 2305
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.076852777477079e-05,
      "loss": 0.9334,
      "step": 2310
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.069805711848096e-05,
      "loss": 0.933,
      "step": 2315
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.062753867330729e-05,
      "loss": 0.9348,
      "step": 2320
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.055697303067905e-05,
      "loss": 0.9997,
      "step": 2325
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.048636078242137e-05,
      "loss": 0.9196,
      "step": 2330
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.0415702520750235e-05,
      "loss": 0.9735,
      "step": 2335
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.0344998838267525e-05,
      "loss": 0.9395,
      "step": 2340
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.0274250327956093e-05,
      "loss": 0.9455,
      "step": 2345
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.020345758317474e-05,
      "loss": 0.968,
      "step": 2350
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.0132621197653245e-05,
      "loss": 0.9403,
      "step": 2355
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.0061741765487418e-05,
      "loss": 0.9267,
      "step": 2360
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.9990819881134073e-05,
      "loss": 0.9734,
      "step": 2365
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.9919856139406093e-05,
      "loss": 0.949,
      "step": 2370
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.9848851135467386e-05,
      "loss": 0.9464,
      "step": 2375
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.977780546482794e-05,
      "loss": 0.9709,
      "step": 2380
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.9706719723338795e-05,
      "loss": 0.9202,
      "step": 2385
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.9635594507187074e-05,
      "loss": 0.9625,
      "step": 2390
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.956443041289096e-05,
      "loss": 0.9307,
      "step": 2395
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.9493228037294702e-05,
      "loss": 0.907,
      "step": 2400
    }
  ],
  "max_steps": 5424,
  "num_train_epochs": 3,
  "total_flos": 7.799084377807258e+17,
  "trial_name": null,
  "trial_params": null
}